{
  "best_metric": 0.44262972474098206,
  "best_model_checkpoint": "models/vit-pretraining-2024_03_10/checkpoint-456841",
  "epoch": 200.0,
  "eval_steps": 500,
  "global_step": 488600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 1.1302417516708374,
      "learning_rate": 1.9187474416700776e-09,
      "loss": 1.2056,
      "step": 10
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1404988765716553,
      "learning_rate": 3.837494883340155e-09,
      "loss": 1.2043,
      "step": 20
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1415787935256958,
      "learning_rate": 5.7562423250102325e-09,
      "loss": 1.2043,
      "step": 30
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1234122514724731,
      "learning_rate": 7.67498976668031e-09,
      "loss": 1.2052,
      "step": 40
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1420000791549683,
      "learning_rate": 9.593737208350387e-09,
      "loss": 1.2047,
      "step": 50
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1481159925460815,
      "learning_rate": 1.1512484650020465e-08,
      "loss": 1.2037,
      "step": 60
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1564319133758545,
      "learning_rate": 1.3431232091690543e-08,
      "loss": 1.2034,
      "step": 70
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1357426643371582,
      "learning_rate": 1.534997953336062e-08,
      "loss": 1.2036,
      "step": 80
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1320079565048218,
      "learning_rate": 1.72687269750307e-08,
      "loss": 1.2023,
      "step": 90
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1257023811340332,
      "learning_rate": 1.9187474416700774e-08,
      "loss": 1.2027,
      "step": 100
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1172627210617065,
      "learning_rate": 2.1106221858370852e-08,
      "loss": 1.2017,
      "step": 110
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1365904808044434,
      "learning_rate": 2.302496930004093e-08,
      "loss": 1.2007,
      "step": 120
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1286369562149048,
      "learning_rate": 2.4943716741711008e-08,
      "loss": 1.2008,
      "step": 130
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1155977249145508,
      "learning_rate": 2.6862464183381086e-08,
      "loss": 1.1992,
      "step": 140
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.105649709701538,
      "learning_rate": 2.8781211625051164e-08,
      "loss": 1.1991,
      "step": 150
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1238553524017334,
      "learning_rate": 3.069995906672124e-08,
      "loss": 1.1987,
      "step": 160
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1107207536697388,
      "learning_rate": 3.261870650839132e-08,
      "loss": 1.1978,
      "step": 170
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1147136688232422,
      "learning_rate": 3.45374539500614e-08,
      "loss": 1.1964,
      "step": 180
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1190924644470215,
      "learning_rate": 3.6456201391731476e-08,
      "loss": 1.1952,
      "step": 190
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0940288305282593,
      "learning_rate": 3.837494883340155e-08,
      "loss": 1.1938,
      "step": 200
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0981765985488892,
      "learning_rate": 4.029369627507163e-08,
      "loss": 1.1937,
      "step": 210
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0918140411376953,
      "learning_rate": 4.2212443716741704e-08,
      "loss": 1.1925,
      "step": 220
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0844347476959229,
      "learning_rate": 4.413119115841179e-08,
      "loss": 1.1908,
      "step": 230
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0715129375457764,
      "learning_rate": 4.604993860008186e-08,
      "loss": 1.1901,
      "step": 240
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0669738054275513,
      "learning_rate": 4.7968686041751945e-08,
      "loss": 1.1887,
      "step": 250
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0745245218276978,
      "learning_rate": 4.9887433483422016e-08,
      "loss": 1.1873,
      "step": 260
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0692330598831177,
      "learning_rate": 5.1806180925092094e-08,
      "loss": 1.1862,
      "step": 270
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0457313060760498,
      "learning_rate": 5.372492836676217e-08,
      "loss": 1.1847,
      "step": 280
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0527009963989258,
      "learning_rate": 5.564367580843225e-08,
      "loss": 1.1829,
      "step": 290
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0516666173934937,
      "learning_rate": 5.756242325010233e-08,
      "loss": 1.1823,
      "step": 300
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0446990728378296,
      "learning_rate": 5.9481170691772407e-08,
      "loss": 1.1807,
      "step": 310
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0352306365966797,
      "learning_rate": 6.139991813344248e-08,
      "loss": 1.1783,
      "step": 320
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.031517744064331,
      "learning_rate": 6.331866557511257e-08,
      "loss": 1.1776,
      "step": 330
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0264582633972168,
      "learning_rate": 6.523741301678264e-08,
      "loss": 1.175,
      "step": 340
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0038774013519287,
      "learning_rate": 6.715616045845273e-08,
      "loss": 1.1744,
      "step": 350
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0127763748168945,
      "learning_rate": 6.90749079001228e-08,
      "loss": 1.1726,
      "step": 360
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9882121086120605,
      "learning_rate": 7.099365534179287e-08,
      "loss": 1.17,
      "step": 370
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.980266809463501,
      "learning_rate": 7.291240278346295e-08,
      "loss": 1.1676,
      "step": 380
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9715328812599182,
      "learning_rate": 7.483115022513302e-08,
      "loss": 1.1678,
      "step": 390
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9726504683494568,
      "learning_rate": 7.67498976668031e-08,
      "loss": 1.1655,
      "step": 400
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9702500104904175,
      "learning_rate": 7.86686451084732e-08,
      "loss": 1.1636,
      "step": 410
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9513685703277588,
      "learning_rate": 8.058739255014327e-08,
      "loss": 1.1612,
      "step": 420
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9495577812194824,
      "learning_rate": 8.250613999181334e-08,
      "loss": 1.1592,
      "step": 430
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9544944763183594,
      "learning_rate": 8.442488743348341e-08,
      "loss": 1.1579,
      "step": 440
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9458625316619873,
      "learning_rate": 8.634363487515349e-08,
      "loss": 1.1555,
      "step": 450
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9089348316192627,
      "learning_rate": 8.826238231682358e-08,
      "loss": 1.1541,
      "step": 460
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9211780428886414,
      "learning_rate": 9.018112975849365e-08,
      "loss": 1.1516,
      "step": 470
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8971461057662964,
      "learning_rate": 9.209987720016372e-08,
      "loss": 1.1488,
      "step": 480
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.876408040523529,
      "learning_rate": 9.40186246418338e-08,
      "loss": 1.1475,
      "step": 490
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8645665049552917,
      "learning_rate": 9.593737208350389e-08,
      "loss": 1.1454,
      "step": 500
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8706442713737488,
      "learning_rate": 9.785611952517396e-08,
      "loss": 1.1435,
      "step": 510
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8777413964271545,
      "learning_rate": 9.977486696684403e-08,
      "loss": 1.1411,
      "step": 520
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8405871391296387,
      "learning_rate": 1.016936144085141e-07,
      "loss": 1.14,
      "step": 530
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8351224064826965,
      "learning_rate": 1.0361236185018419e-07,
      "loss": 1.1374,
      "step": 540
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8274722695350647,
      "learning_rate": 1.0553110929185427e-07,
      "loss": 1.1349,
      "step": 550
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8196274638175964,
      "learning_rate": 1.0744985673352434e-07,
      "loss": 1.1332,
      "step": 560
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8109217882156372,
      "learning_rate": 1.0936860417519442e-07,
      "loss": 1.1306,
      "step": 570
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8090499043464661,
      "learning_rate": 1.112873516168645e-07,
      "loss": 1.1285,
      "step": 580
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.778523325920105,
      "learning_rate": 1.1320609905853459e-07,
      "loss": 1.1273,
      "step": 590
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7805694341659546,
      "learning_rate": 1.1512484650020466e-07,
      "loss": 1.1248,
      "step": 600
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7815826535224915,
      "learning_rate": 1.1704359394187473e-07,
      "loss": 1.1228,
      "step": 610
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.749366044998169,
      "learning_rate": 1.1896234138354481e-07,
      "loss": 1.1208,
      "step": 620
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7423666715621948,
      "learning_rate": 1.208810888252149e-07,
      "loss": 1.1199,
      "step": 630
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7208989262580872,
      "learning_rate": 1.2279983626688497e-07,
      "loss": 1.116,
      "step": 640
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7175284624099731,
      "learning_rate": 1.2471858370855504e-07,
      "loss": 1.1148,
      "step": 650
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7036271691322327,
      "learning_rate": 1.2663733115022514e-07,
      "loss": 1.1126,
      "step": 660
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6913628578186035,
      "learning_rate": 1.285560785918952e-07,
      "loss": 1.1106,
      "step": 670
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7200226783752441,
      "learning_rate": 1.3047482603356528e-07,
      "loss": 1.11,
      "step": 680
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6800700426101685,
      "learning_rate": 1.3239357347523535e-07,
      "loss": 1.1068,
      "step": 690
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6605786681175232,
      "learning_rate": 1.3431232091690545e-07,
      "loss": 1.1053,
      "step": 700
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6495028734207153,
      "learning_rate": 1.3623106835857552e-07,
      "loss": 1.1028,
      "step": 710
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6407223343849182,
      "learning_rate": 1.381498158002456e-07,
      "loss": 1.1017,
      "step": 720
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6504897475242615,
      "learning_rate": 1.4006856324191567e-07,
      "loss": 1.099,
      "step": 730
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6524698734283447,
      "learning_rate": 1.4198731068358574e-07,
      "loss": 1.0973,
      "step": 740
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6401053071022034,
      "learning_rate": 1.4390605812525583e-07,
      "loss": 1.0954,
      "step": 750
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6002606749534607,
      "learning_rate": 1.458248055669259e-07,
      "loss": 1.0938,
      "step": 760
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6438450217247009,
      "learning_rate": 1.4774355300859598e-07,
      "loss": 1.093,
      "step": 770
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5883312821388245,
      "learning_rate": 1.4966230045026605e-07,
      "loss": 1.0898,
      "step": 780
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.576004147529602,
      "learning_rate": 1.5158104789193612e-07,
      "loss": 1.0887,
      "step": 790
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.586531400680542,
      "learning_rate": 1.534997953336062e-07,
      "loss": 1.0866,
      "step": 800
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5479790568351746,
      "learning_rate": 1.554185427752763e-07,
      "loss": 1.0866,
      "step": 810
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.558303952217102,
      "learning_rate": 1.573372902169464e-07,
      "loss": 1.0839,
      "step": 820
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5292941927909851,
      "learning_rate": 1.5925603765861646e-07,
      "loss": 1.0833,
      "step": 830
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.540973424911499,
      "learning_rate": 1.6117478510028653e-07,
      "loss": 1.0805,
      "step": 840
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.524495005607605,
      "learning_rate": 1.630935325419566e-07,
      "loss": 1.0792,
      "step": 850
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5079398155212402,
      "learning_rate": 1.6501227998362667e-07,
      "loss": 1.0789,
      "step": 860
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5069434642791748,
      "learning_rate": 1.6693102742529674e-07,
      "loss": 1.075,
      "step": 870
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.49611520767211914,
      "learning_rate": 1.6884977486696682e-07,
      "loss": 1.0748,
      "step": 880
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4824225902557373,
      "learning_rate": 1.707685223086369e-07,
      "loss": 1.073,
      "step": 890
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4720127582550049,
      "learning_rate": 1.7268726975030699e-07,
      "loss": 1.0723,
      "step": 900
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.48235777020454407,
      "learning_rate": 1.7460601719197708e-07,
      "loss": 1.0698,
      "step": 910
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.45931223034858704,
      "learning_rate": 1.7652476463364715e-07,
      "loss": 1.0702,
      "step": 920
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.46550649404525757,
      "learning_rate": 1.7844351207531723e-07,
      "loss": 1.0673,
      "step": 930
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.45142972469329834,
      "learning_rate": 1.803622595169873e-07,
      "loss": 1.0657,
      "step": 940
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4417128264904022,
      "learning_rate": 1.8228100695865737e-07,
      "loss": 1.0659,
      "step": 950
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4317210018634796,
      "learning_rate": 1.8419975440032744e-07,
      "loss": 1.065,
      "step": 960
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4183664917945862,
      "learning_rate": 1.861185018419975e-07,
      "loss": 1.0624,
      "step": 970
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4219318628311157,
      "learning_rate": 1.880372492836676e-07,
      "loss": 1.0603,
      "step": 980
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.403277724981308,
      "learning_rate": 1.8995599672533768e-07,
      "loss": 1.0615,
      "step": 990
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4062301516532898,
      "learning_rate": 1.9187474416700778e-07,
      "loss": 1.0592,
      "step": 1000
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4151647984981537,
      "learning_rate": 1.9379349160867785e-07,
      "loss": 1.0592,
      "step": 1010
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4017297029495239,
      "learning_rate": 1.9571223905034792e-07,
      "loss": 1.056,
      "step": 1020
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38589805364608765,
      "learning_rate": 1.97630986492018e-07,
      "loss": 1.0557,
      "step": 1030
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39735037088394165,
      "learning_rate": 1.9954973393368807e-07,
      "loss": 1.0561,
      "step": 1040
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37628719210624695,
      "learning_rate": 2.0146848137535814e-07,
      "loss": 1.0542,
      "step": 1050
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38165953755378723,
      "learning_rate": 2.033872288170282e-07,
      "loss": 1.0525,
      "step": 1060
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3810892105102539,
      "learning_rate": 2.053059762586983e-07,
      "loss": 1.0504,
      "step": 1070
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3886578381061554,
      "learning_rate": 2.0722472370036838e-07,
      "loss": 1.0501,
      "step": 1080
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36734455823898315,
      "learning_rate": 2.0914347114203848e-07,
      "loss": 1.0505,
      "step": 1090
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3741951286792755,
      "learning_rate": 2.1106221858370855e-07,
      "loss": 1.0501,
      "step": 1100
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3644857108592987,
      "learning_rate": 2.1298096602537862e-07,
      "loss": 1.0493,
      "step": 1110
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37163782119750977,
      "learning_rate": 2.148997134670487e-07,
      "loss": 1.0474,
      "step": 1120
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34015795588493347,
      "learning_rate": 2.1681846090871876e-07,
      "loss": 1.0448,
      "step": 1130
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.387784481048584,
      "learning_rate": 2.1873720835038883e-07,
      "loss": 1.0466,
      "step": 1140
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3708529770374298,
      "learning_rate": 2.2065595579205893e-07,
      "loss": 1.046,
      "step": 1150
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.357849657535553,
      "learning_rate": 2.22574703233729e-07,
      "loss": 1.0447,
      "step": 1160
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32534539699554443,
      "learning_rate": 2.244934506753991e-07,
      "loss": 1.0441,
      "step": 1170
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32729390263557434,
      "learning_rate": 2.2641219811706917e-07,
      "loss": 1.0434,
      "step": 1180
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3092939853668213,
      "learning_rate": 2.2833094555873924e-07,
      "loss": 1.0408,
      "step": 1190
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.315017968416214,
      "learning_rate": 2.3024969300040931e-07,
      "loss": 1.041,
      "step": 1200
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.32678693532943726,
      "learning_rate": 2.3216844044207939e-07,
      "loss": 1.0396,
      "step": 1210
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3045768737792969,
      "learning_rate": 2.3408718788374946e-07,
      "loss": 1.038,
      "step": 1220
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3085119128227234,
      "learning_rate": 2.3600593532541955e-07,
      "loss": 1.0378,
      "step": 1230
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.30969786643981934,
      "learning_rate": 2.3792468276708963e-07,
      "loss": 1.0371,
      "step": 1240
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.313570499420166,
      "learning_rate": 2.398434302087597e-07,
      "loss": 1.0366,
      "step": 1250
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2910727262496948,
      "learning_rate": 2.417621776504298e-07,
      "loss": 1.0361,
      "step": 1260
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3049977421760559,
      "learning_rate": 2.4368092509209984e-07,
      "loss": 1.0373,
      "step": 1270
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3079485595226288,
      "learning_rate": 2.4559967253376994e-07,
      "loss": 1.0357,
      "step": 1280
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2986745238304138,
      "learning_rate": 2.4751841997544e-07,
      "loss": 1.0337,
      "step": 1290
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2821179926395416,
      "learning_rate": 2.494371674171101e-07,
      "loss": 1.0329,
      "step": 1300
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.27660101652145386,
      "learning_rate": 2.513559148587801e-07,
      "loss": 1.0332,
      "step": 1310
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3824383318424225,
      "learning_rate": 2.532746623004503e-07,
      "loss": 1.0318,
      "step": 1320
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2650841474533081,
      "learning_rate": 2.551934097421203e-07,
      "loss": 1.0317,
      "step": 1330
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.269889235496521,
      "learning_rate": 2.571121571837904e-07,
      "loss": 1.0305,
      "step": 1340
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2623484134674072,
      "learning_rate": 2.5903090462546046e-07,
      "loss": 1.0314,
      "step": 1350
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.28557583689689636,
      "learning_rate": 2.6094965206713056e-07,
      "loss": 1.0304,
      "step": 1360
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3150051534175873,
      "learning_rate": 2.628683995088006e-07,
      "loss": 1.029,
      "step": 1370
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2666436433792114,
      "learning_rate": 2.647871469504707e-07,
      "loss": 1.0265,
      "step": 1380
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30076223611831665,
      "learning_rate": 2.6670589439214075e-07,
      "loss": 1.028,
      "step": 1390
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.25557559728622437,
      "learning_rate": 2.686246418338109e-07,
      "loss": 1.0267,
      "step": 1400
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.28760895133018494,
      "learning_rate": 2.7054338927548095e-07,
      "loss": 1.0267,
      "step": 1410
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.24308475852012634,
      "learning_rate": 2.7246213671715104e-07,
      "loss": 1.0259,
      "step": 1420
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24371612071990967,
      "learning_rate": 2.743808841588211e-07,
      "loss": 1.0268,
      "step": 1430
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.25367575883865356,
      "learning_rate": 2.762996316004912e-07,
      "loss": 1.0248,
      "step": 1440
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24026906490325928,
      "learning_rate": 2.7821837904216123e-07,
      "loss": 1.0245,
      "step": 1450
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.23836664855480194,
      "learning_rate": 2.8013712648383133e-07,
      "loss": 1.0239,
      "step": 1460
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.23544681072235107,
      "learning_rate": 2.820558739255014e-07,
      "loss": 1.0244,
      "step": 1470
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.23720157146453857,
      "learning_rate": 2.8397462136717147e-07,
      "loss": 1.0227,
      "step": 1480
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.25152388215065,
      "learning_rate": 2.8589336880884157e-07,
      "loss": 1.0235,
      "step": 1490
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.22095340490341187,
      "learning_rate": 2.8781211625051167e-07,
      "loss": 1.0227,
      "step": 1500
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2538137435913086,
      "learning_rate": 2.897308636921817e-07,
      "loss": 1.0234,
      "step": 1510
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.21703729033470154,
      "learning_rate": 2.916496111338518e-07,
      "loss": 1.0237,
      "step": 1520
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2238396257162094,
      "learning_rate": 2.9356835857552186e-07,
      "loss": 1.0225,
      "step": 1530
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23417343199253082,
      "learning_rate": 2.9548710601719195e-07,
      "loss": 1.0204,
      "step": 1540
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.21133969724178314,
      "learning_rate": 2.97405853458862e-07,
      "loss": 1.0218,
      "step": 1550
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21478645503520966,
      "learning_rate": 2.993246009005321e-07,
      "loss": 1.0192,
      "step": 1560
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.24249108135700226,
      "learning_rate": 3.0124334834220214e-07,
      "loss": 1.019,
      "step": 1570
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20616930723190308,
      "learning_rate": 3.0316209578387224e-07,
      "loss": 1.0204,
      "step": 1580
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.22133825719356537,
      "learning_rate": 3.0508084322554234e-07,
      "loss": 1.018,
      "step": 1590
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20094144344329834,
      "learning_rate": 3.069995906672124e-07,
      "loss": 1.0188,
      "step": 1600
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20146296918392181,
      "learning_rate": 3.089183381088825e-07,
      "loss": 1.0174,
      "step": 1610
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.220841646194458,
      "learning_rate": 3.108370855505526e-07,
      "loss": 1.0165,
      "step": 1620
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19479529559612274,
      "learning_rate": 3.127558329922227e-07,
      "loss": 1.0153,
      "step": 1630
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21818408370018005,
      "learning_rate": 3.146745804338928e-07,
      "loss": 1.0177,
      "step": 1640
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1967555284500122,
      "learning_rate": 3.165933278755628e-07,
      "loss": 1.0165,
      "step": 1650
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.18899010121822357,
      "learning_rate": 3.185120753172329e-07,
      "loss": 1.0164,
      "step": 1660
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1831546276807785,
      "learning_rate": 3.2043082275890296e-07,
      "loss": 1.015,
      "step": 1670
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.18965131044387817,
      "learning_rate": 3.2234957020057306e-07,
      "loss": 1.0155,
      "step": 1680
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20051342248916626,
      "learning_rate": 3.242683176422431e-07,
      "loss": 1.0156,
      "step": 1690
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20127572119235992,
      "learning_rate": 3.261870650839132e-07,
      "loss": 1.014,
      "step": 1700
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.18799249827861786,
      "learning_rate": 3.2810581252558325e-07,
      "loss": 1.0152,
      "step": 1710
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.19969061017036438,
      "learning_rate": 3.3002455996725335e-07,
      "loss": 1.0124,
      "step": 1720
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19544294476509094,
      "learning_rate": 3.319433074089234e-07,
      "loss": 1.0154,
      "step": 1730
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.18440210819244385,
      "learning_rate": 3.338620548505935e-07,
      "loss": 1.0141,
      "step": 1740
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20036162436008453,
      "learning_rate": 3.3578080229226353e-07,
      "loss": 1.0117,
      "step": 1750
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1669704020023346,
      "learning_rate": 3.3769954973393363e-07,
      "loss": 1.0112,
      "step": 1760
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20973049104213715,
      "learning_rate": 3.3961829717560373e-07,
      "loss": 1.0133,
      "step": 1770
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2180255949497223,
      "learning_rate": 3.415370446172738e-07,
      "loss": 1.0094,
      "step": 1780
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22467105090618134,
      "learning_rate": 3.434557920589439e-07,
      "loss": 1.0111,
      "step": 1790
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.19380712509155273,
      "learning_rate": 3.4537453950061397e-07,
      "loss": 1.0118,
      "step": 1800
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.19561414420604706,
      "learning_rate": 3.4729328694228407e-07,
      "loss": 1.0107,
      "step": 1810
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.17537786066532135,
      "learning_rate": 3.4921203438395417e-07,
      "loss": 1.0107,
      "step": 1820
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.16829745471477509,
      "learning_rate": 3.511307818256242e-07,
      "loss": 1.011,
      "step": 1830
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18437370657920837,
      "learning_rate": 3.530495292672943e-07,
      "loss": 1.0099,
      "step": 1840
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22416964173316956,
      "learning_rate": 3.5496827670896435e-07,
      "loss": 1.0118,
      "step": 1850
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.16199608147144318,
      "learning_rate": 3.5688702415063445e-07,
      "loss": 1.0097,
      "step": 1860
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1759682148694992,
      "learning_rate": 3.588057715923045e-07,
      "loss": 1.0122,
      "step": 1870
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.17157995700836182,
      "learning_rate": 3.607245190339746e-07,
      "loss": 1.0095,
      "step": 1880
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1641765534877777,
      "learning_rate": 3.6264326647564464e-07,
      "loss": 1.0092,
      "step": 1890
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.15396302938461304,
      "learning_rate": 3.6456201391731474e-07,
      "loss": 1.0103,
      "step": 1900
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.15870316326618195,
      "learning_rate": 3.664807613589848e-07,
      "loss": 1.0078,
      "step": 1910
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.15351416170597076,
      "learning_rate": 3.683995088006549e-07,
      "loss": 1.0076,
      "step": 1920
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.15729747712612152,
      "learning_rate": 3.70318256242325e-07,
      "loss": 1.0079,
      "step": 1930
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.14869193732738495,
      "learning_rate": 3.72237003683995e-07,
      "loss": 1.0081,
      "step": 1940
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.14780552685260773,
      "learning_rate": 3.741557511256651e-07,
      "loss": 1.0073,
      "step": 1950
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.15063492953777313,
      "learning_rate": 3.760744985673352e-07,
      "loss": 1.0082,
      "step": 1960
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16691546142101288,
      "learning_rate": 3.779932460090053e-07,
      "loss": 1.0077,
      "step": 1970
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.15350006520748138,
      "learning_rate": 3.7991199345067536e-07,
      "loss": 1.0083,
      "step": 1980
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2022244930267334,
      "learning_rate": 3.8183074089234546e-07,
      "loss": 1.0082,
      "step": 1990
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.17419302463531494,
      "learning_rate": 3.8374948833401556e-07,
      "loss": 1.0085,
      "step": 2000
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1644999086856842,
      "learning_rate": 3.856682357756856e-07,
      "loss": 1.0059,
      "step": 2010
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.15001410245895386,
      "learning_rate": 3.875869832173557e-07,
      "loss": 1.0051,
      "step": 2020
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.15016770362854004,
      "learning_rate": 3.8950573065902575e-07,
      "loss": 1.0058,
      "step": 2030
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.14068953692913055,
      "learning_rate": 3.9142447810069584e-07,
      "loss": 1.0068,
      "step": 2040
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.13636018335819244,
      "learning_rate": 3.933432255423659e-07,
      "loss": 1.0065,
      "step": 2050
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.13372232019901276,
      "learning_rate": 3.95261972984036e-07,
      "loss": 1.0063,
      "step": 2060
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.12867872416973114,
      "learning_rate": 3.9718072042570603e-07,
      "loss": 1.0047,
      "step": 2070
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.13062798976898193,
      "learning_rate": 3.9909946786737613e-07,
      "loss": 1.0047,
      "step": 2080
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.13903233408927917,
      "learning_rate": 4.010182153090462e-07,
      "loss": 1.0044,
      "step": 2090
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.14890648424625397,
      "learning_rate": 4.0293696275071627e-07,
      "loss": 1.0066,
      "step": 2100
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.17362573742866516,
      "learning_rate": 4.0485571019238637e-07,
      "loss": 1.0041,
      "step": 2110
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.12294167280197144,
      "learning_rate": 4.067744576340564e-07,
      "loss": 1.004,
      "step": 2120
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1597672402858734,
      "learning_rate": 4.0869320507572657e-07,
      "loss": 1.0019,
      "step": 2130
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.14325286448001862,
      "learning_rate": 4.106119525173966e-07,
      "loss": 1.0052,
      "step": 2140
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.11829974502325058,
      "learning_rate": 4.125306999590667e-07,
      "loss": 1.0041,
      "step": 2150
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1991194486618042,
      "learning_rate": 4.1444944740073675e-07,
      "loss": 1.0041,
      "step": 2160
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23001176118850708,
      "learning_rate": 4.1636819484240685e-07,
      "loss": 1.0019,
      "step": 2170
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.14314135909080505,
      "learning_rate": 4.1828694228407695e-07,
      "loss": 1.0029,
      "step": 2180
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1164059191942215,
      "learning_rate": 4.20205689725747e-07,
      "loss": 1.0037,
      "step": 2190
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.11646254360675812,
      "learning_rate": 4.221244371674171e-07,
      "loss": 1.003,
      "step": 2200
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.12384123355150223,
      "learning_rate": 4.2404318460908714e-07,
      "loss": 1.0044,
      "step": 2210
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.11661216616630554,
      "learning_rate": 4.2596193205075724e-07,
      "loss": 1.0027,
      "step": 2220
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.14190039038658142,
      "learning_rate": 4.278806794924273e-07,
      "loss": 1.0021,
      "step": 2230
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.13870534300804138,
      "learning_rate": 4.297994269340974e-07,
      "loss": 1.0026,
      "step": 2240
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1475309580564499,
      "learning_rate": 4.317181743757674e-07,
      "loss": 1.003,
      "step": 2250
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18177026510238647,
      "learning_rate": 4.336369218174375e-07,
      "loss": 1.0027,
      "step": 2260
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.11788962036371231,
      "learning_rate": 4.3555566925910757e-07,
      "loss": 1.0024,
      "step": 2270
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.11188003420829773,
      "learning_rate": 4.3747441670077766e-07,
      "loss": 1.0047,
      "step": 2280
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20192362368106842,
      "learning_rate": 4.3939316414244776e-07,
      "loss": 1.002,
      "step": 2290
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.13641443848609924,
      "learning_rate": 4.4131191158411786e-07,
      "loss": 1.0001,
      "step": 2300
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.12127813696861267,
      "learning_rate": 4.4323065902578796e-07,
      "loss": 1.0007,
      "step": 2310
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1320171058177948,
      "learning_rate": 4.45149406467458e-07,
      "loss": 1.0022,
      "step": 2320
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.13225774466991425,
      "learning_rate": 4.470681539091281e-07,
      "loss": 1.0,
      "step": 2330
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.15130658447742462,
      "learning_rate": 4.489869013507982e-07,
      "loss": 1.0006,
      "step": 2340
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.12090421468019485,
      "learning_rate": 4.5090564879246824e-07,
      "loss": 1.0013,
      "step": 2350
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.09757827967405319,
      "learning_rate": 4.5282439623413834e-07,
      "loss": 1.0013,
      "step": 2360
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.09331031143665314,
      "learning_rate": 4.547431436758084e-07,
      "loss": 1.0013,
      "step": 2370
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.10475800931453705,
      "learning_rate": 4.566618911174785e-07,
      "loss": 1.0012,
      "step": 2380
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.12264197319746017,
      "learning_rate": 4.5858063855914853e-07,
      "loss": 1.002,
      "step": 2390
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.10976061224937439,
      "learning_rate": 4.6049938600081863e-07,
      "loss": 1.0013,
      "step": 2400
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.10739260166883469,
      "learning_rate": 4.6241813344248867e-07,
      "loss": 1.0018,
      "step": 2410
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.11093743145465851,
      "learning_rate": 4.6433688088415877e-07,
      "loss": 1.0016,
      "step": 2420
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1267680823802948,
      "learning_rate": 4.662556283258288e-07,
      "loss": 1.0011,
      "step": 2430
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.11313279718160629,
      "learning_rate": 4.681743757674989e-07,
      "loss": 1.0002,
      "step": 2440
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.0000314712524414,
      "eval_runtime": 56.9437,
      "eval_samples_per_second": 60.569,
      "eval_steps_per_second": 7.586,
      "step": 2443
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1478390097618103,
      "learning_rate": 4.7009312320916896e-07,
      "loss": 1.0021,
      "step": 2450
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.11842042952775955,
      "learning_rate": 4.720118706508391e-07,
      "loss": 1.001,
      "step": 2460
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.11305660754442215,
      "learning_rate": 4.739306180925092e-07,
      "loss": 1.0021,
      "step": 2470
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.11627020686864853,
      "learning_rate": 4.7584936553417925e-07,
      "loss": 1.0008,
      "step": 2480
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.08868634700775146,
      "learning_rate": 4.777681129758494e-07,
      "loss": 1.0,
      "step": 2490
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1494980901479721,
      "learning_rate": 4.796868604175194e-07,
      "loss": 1.0001,
      "step": 2500
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.10967830568552017,
      "learning_rate": 4.816056078591894e-07,
      "loss": 0.9996,
      "step": 2510
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1095232293009758,
      "learning_rate": 4.835243553008596e-07,
      "loss": 1.0,
      "step": 2520
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.12186755985021591,
      "learning_rate": 4.854431027425296e-07,
      "loss": 0.9992,
      "step": 2530
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1073576807975769,
      "learning_rate": 4.873618501841997e-07,
      "loss": 0.9992,
      "step": 2540
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.12696845829486847,
      "learning_rate": 4.892805976258698e-07,
      "loss": 0.9989,
      "step": 2550
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.08960144966840744,
      "learning_rate": 4.911993450675399e-07,
      "loss": 0.999,
      "step": 2560
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.09792343527078629,
      "learning_rate": 4.931180925092099e-07,
      "loss": 0.9998,
      "step": 2570
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.11528971791267395,
      "learning_rate": 4.9503683995088e-07,
      "loss": 0.9977,
      "step": 2580
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1314706653356552,
      "learning_rate": 4.969555873925501e-07,
      "loss": 0.9985,
      "step": 2590
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.0869363397359848,
      "learning_rate": 4.988743348342202e-07,
      "loss": 0.9988,
      "step": 2600
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.08023642748594284,
      "learning_rate": 5.007930822758902e-07,
      "loss": 0.9979,
      "step": 2610
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.11450603604316711,
      "learning_rate": 5.027118297175603e-07,
      "loss": 1.0012,
      "step": 2620
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.11469870805740356,
      "learning_rate": 5.046305771592304e-07,
      "loss": 0.9985,
      "step": 2630
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.0978398248553276,
      "learning_rate": 5.065493246009006e-07,
      "loss": 0.9979,
      "step": 2640
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.10484980046749115,
      "learning_rate": 5.084680720425706e-07,
      "loss": 0.9981,
      "step": 2650
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.11064465343952179,
      "learning_rate": 5.103868194842406e-07,
      "loss": 0.9984,
      "step": 2660
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.07005872577428818,
      "learning_rate": 5.123055669259107e-07,
      "loss": 0.9991,
      "step": 2670
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.07808951288461685,
      "learning_rate": 5.142243143675808e-07,
      "loss": 0.9985,
      "step": 2680
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.06807954609394073,
      "learning_rate": 5.161430618092509e-07,
      "loss": 0.9988,
      "step": 2690
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.09980972111225128,
      "learning_rate": 5.180618092509209e-07,
      "loss": 0.999,
      "step": 2700
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.21692952513694763,
      "learning_rate": 5.199805566925911e-07,
      "loss": 0.9995,
      "step": 2710
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.07546713948249817,
      "learning_rate": 5.218993041342611e-07,
      "loss": 0.9979,
      "step": 2720
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1015053540468216,
      "learning_rate": 5.238180515759312e-07,
      "loss": 0.9954,
      "step": 2730
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.09382616728544235,
      "learning_rate": 5.257367990176012e-07,
      "loss": 0.9979,
      "step": 2740
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.13619408011436462,
      "learning_rate": 5.276555464592714e-07,
      "loss": 0.9972,
      "step": 2750
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.135032519698143,
      "learning_rate": 5.295742939009414e-07,
      "loss": 0.9971,
      "step": 2760
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.09556297212839127,
      "learning_rate": 5.314930413426115e-07,
      "loss": 0.9967,
      "step": 2770
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.14096589386463165,
      "learning_rate": 5.334117887842815e-07,
      "loss": 0.9977,
      "step": 2780
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.09116736799478531,
      "learning_rate": 5.353305362259517e-07,
      "loss": 0.9972,
      "step": 2790
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.09037278592586517,
      "learning_rate": 5.372492836676218e-07,
      "loss": 0.9969,
      "step": 2800
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.11145661026239395,
      "learning_rate": 5.391680311092918e-07,
      "loss": 0.999,
      "step": 2810
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.09276577830314636,
      "learning_rate": 5.410867785509619e-07,
      "loss": 0.9966,
      "step": 2820
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.06607125699520111,
      "learning_rate": 5.430055259926319e-07,
      "loss": 0.9973,
      "step": 2830
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.09073203057050705,
      "learning_rate": 5.449242734343021e-07,
      "loss": 0.9983,
      "step": 2840
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.08952809870243073,
      "learning_rate": 5.468430208759721e-07,
      "loss": 0.9946,
      "step": 2850
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.08847542107105255,
      "learning_rate": 5.487617683176422e-07,
      "loss": 0.9986,
      "step": 2860
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.10774630308151245,
      "learning_rate": 5.506805157593123e-07,
      "loss": 0.996,
      "step": 2870
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.09260059893131256,
      "learning_rate": 5.525992632009824e-07,
      "loss": 0.9959,
      "step": 2880
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.0652688592672348,
      "learning_rate": 5.545180106426524e-07,
      "loss": 0.9958,
      "step": 2890
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.06935996562242508,
      "learning_rate": 5.564367580843225e-07,
      "loss": 0.9952,
      "step": 2900
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.09041889011859894,
      "learning_rate": 5.583555055259926e-07,
      "loss": 0.9961,
      "step": 2910
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1049184501171112,
      "learning_rate": 5.602742529676627e-07,
      "loss": 0.9943,
      "step": 2920
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.07971151173114777,
      "learning_rate": 5.621930004093327e-07,
      "loss": 0.9964,
      "step": 2930
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.18742699921131134,
      "learning_rate": 5.641117478510028e-07,
      "loss": 0.9964,
      "step": 2940
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.07923754304647446,
      "learning_rate": 5.660304952926729e-07,
      "loss": 0.9976,
      "step": 2950
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.12884142994880676,
      "learning_rate": 5.679492427343429e-07,
      "loss": 0.9975,
      "step": 2960
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.12948425114154816,
      "learning_rate": 5.698679901760131e-07,
      "loss": 0.9977,
      "step": 2970
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.08524078875780106,
      "learning_rate": 5.717867376176831e-07,
      "loss": 0.9943,
      "step": 2980
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.08120322972536087,
      "learning_rate": 5.737054850593532e-07,
      "loss": 0.9969,
      "step": 2990
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.07015325129032135,
      "learning_rate": 5.756242325010233e-07,
      "loss": 0.9967,
      "step": 3000
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.07578184455633163,
      "learning_rate": 5.775429799426934e-07,
      "loss": 0.9955,
      "step": 3010
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.09199531376361847,
      "learning_rate": 5.794617273843634e-07,
      "loss": 0.9968,
      "step": 3020
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.18446844816207886,
      "learning_rate": 5.813804748260335e-07,
      "loss": 0.9971,
      "step": 3030
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.0693512111902237,
      "learning_rate": 5.832992222677036e-07,
      "loss": 0.993,
      "step": 3040
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.09444622695446014,
      "learning_rate": 5.852179697093737e-07,
      "loss": 0.9989,
      "step": 3050
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.08889685571193695,
      "learning_rate": 5.871367171510437e-07,
      "loss": 0.9963,
      "step": 3060
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.10986029356718063,
      "learning_rate": 5.890554645927139e-07,
      "loss": 0.9949,
      "step": 3070
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.2120119333267212,
      "learning_rate": 5.909742120343839e-07,
      "loss": 0.9972,
      "step": 3080
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.11835231631994247,
      "learning_rate": 5.92892959476054e-07,
      "loss": 0.9972,
      "step": 3090
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.08677642792463303,
      "learning_rate": 5.94811706917724e-07,
      "loss": 0.9964,
      "step": 3100
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.07567739486694336,
      "learning_rate": 5.967304543593942e-07,
      "loss": 0.996,
      "step": 3110
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.06509307771921158,
      "learning_rate": 5.986492018010642e-07,
      "loss": 0.9948,
      "step": 3120
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.07024620473384857,
      "learning_rate": 6.005679492427342e-07,
      "loss": 0.9964,
      "step": 3130
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.09849105775356293,
      "learning_rate": 6.024866966844043e-07,
      "loss": 0.9966,
      "step": 3140
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1289616972208023,
      "learning_rate": 6.044054441260744e-07,
      "loss": 0.9925,
      "step": 3150
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.08987656980752945,
      "learning_rate": 6.063241915677445e-07,
      "loss": 0.9954,
      "step": 3160
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.11029987037181854,
      "learning_rate": 6.082429390094145e-07,
      "loss": 0.9946,
      "step": 3170
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.06290663033723831,
      "learning_rate": 6.101616864510847e-07,
      "loss": 0.995,
      "step": 3180
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.0778784528374672,
      "learning_rate": 6.120804338927547e-07,
      "loss": 0.995,
      "step": 3190
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.08488310128450394,
      "learning_rate": 6.139991813344248e-07,
      "loss": 0.9947,
      "step": 3200
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.11601858586072922,
      "learning_rate": 6.159179287760948e-07,
      "loss": 0.9942,
      "step": 3210
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.103058360517025,
      "learning_rate": 6.17836676217765e-07,
      "loss": 0.9935,
      "step": 3220
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.11795709282159805,
      "learning_rate": 6.197554236594351e-07,
      "loss": 0.9928,
      "step": 3230
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.07079046964645386,
      "learning_rate": 6.216741711011052e-07,
      "loss": 0.996,
      "step": 3240
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.10498754680156708,
      "learning_rate": 6.235929185427753e-07,
      "loss": 0.9953,
      "step": 3250
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.059088800102472305,
      "learning_rate": 6.255116659844454e-07,
      "loss": 0.9962,
      "step": 3260
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.09829726070165634,
      "learning_rate": 6.274304134261154e-07,
      "loss": 0.9944,
      "step": 3270
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.0682409256696701,
      "learning_rate": 6.293491608677855e-07,
      "loss": 0.9953,
      "step": 3280
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.08364296704530716,
      "learning_rate": 6.312679083094556e-07,
      "loss": 0.9951,
      "step": 3290
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.08954983204603195,
      "learning_rate": 6.331866557511256e-07,
      "loss": 0.9948,
      "step": 3300
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.06235593929886818,
      "learning_rate": 6.351054031927957e-07,
      "loss": 0.9943,
      "step": 3310
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.08588958531618118,
      "learning_rate": 6.370241506344658e-07,
      "loss": 0.9951,
      "step": 3320
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.09659326076507568,
      "learning_rate": 6.389428980761359e-07,
      "loss": 0.994,
      "step": 3330
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.05402198061347008,
      "learning_rate": 6.408616455178059e-07,
      "loss": 0.9964,
      "step": 3340
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.09634421020746231,
      "learning_rate": 6.42780392959476e-07,
      "loss": 0.9945,
      "step": 3350
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.17377465963363647,
      "learning_rate": 6.446991404011461e-07,
      "loss": 0.9956,
      "step": 3360
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.08749832957983017,
      "learning_rate": 6.466178878428162e-07,
      "loss": 0.9953,
      "step": 3370
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.069999560713768,
      "learning_rate": 6.485366352844862e-07,
      "loss": 0.9929,
      "step": 3380
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.08185983449220657,
      "learning_rate": 6.504553827261564e-07,
      "loss": 0.9933,
      "step": 3390
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.11885799467563629,
      "learning_rate": 6.523741301678264e-07,
      "loss": 0.9974,
      "step": 3400
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.07838065922260284,
      "learning_rate": 6.542928776094965e-07,
      "loss": 0.995,
      "step": 3410
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.12712670862674713,
      "learning_rate": 6.562116250511665e-07,
      "loss": 0.9936,
      "step": 3420
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.15563039481639862,
      "learning_rate": 6.581303724928366e-07,
      "loss": 0.9939,
      "step": 3430
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.11231633275747299,
      "learning_rate": 6.600491199345067e-07,
      "loss": 0.9972,
      "step": 3440
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.07946788519620895,
      "learning_rate": 6.619678673761767e-07,
      "loss": 0.9948,
      "step": 3450
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.1245964840054512,
      "learning_rate": 6.638866148178468e-07,
      "loss": 0.9935,
      "step": 3460
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.13221850991249084,
      "learning_rate": 6.658053622595169e-07,
      "loss": 0.9954,
      "step": 3470
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.04926653578877449,
      "learning_rate": 6.67724109701187e-07,
      "loss": 0.9936,
      "step": 3480
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.06623417884111404,
      "learning_rate": 6.69642857142857e-07,
      "loss": 0.9943,
      "step": 3490
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.10707608610391617,
      "learning_rate": 6.715616045845271e-07,
      "loss": 0.9944,
      "step": 3500
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.11462438106536865,
      "learning_rate": 6.734803520261972e-07,
      "loss": 0.9931,
      "step": 3510
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.10198796540498734,
      "learning_rate": 6.753990994678673e-07,
      "loss": 0.9939,
      "step": 3520
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.10477564483880997,
      "learning_rate": 6.773178469095373e-07,
      "loss": 0.9924,
      "step": 3530
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.08203921467065811,
      "learning_rate": 6.792365943512075e-07,
      "loss": 0.991,
      "step": 3540
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.0628809779882431,
      "learning_rate": 6.811553417928775e-07,
      "loss": 0.9925,
      "step": 3550
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.11531064659357071,
      "learning_rate": 6.830740892345476e-07,
      "loss": 0.9936,
      "step": 3560
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.1015753224492073,
      "learning_rate": 6.849928366762178e-07,
      "loss": 0.9936,
      "step": 3570
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.11710981279611588,
      "learning_rate": 6.869115841178879e-07,
      "loss": 0.9951,
      "step": 3580
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.06816910207271576,
      "learning_rate": 6.888303315595579e-07,
      "loss": 0.9954,
      "step": 3590
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.10914129763841629,
      "learning_rate": 6.907490790012279e-07,
      "loss": 0.9914,
      "step": 3600
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.10395929217338562,
      "learning_rate": 6.926678264428981e-07,
      "loss": 0.996,
      "step": 3610
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.14113903045654297,
      "learning_rate": 6.945865738845681e-07,
      "loss": 0.9952,
      "step": 3620
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.08126106858253479,
      "learning_rate": 6.965053213262382e-07,
      "loss": 0.9915,
      "step": 3630
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.08699488639831543,
      "learning_rate": 6.984240687679083e-07,
      "loss": 0.9929,
      "step": 3640
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.08064918220043182,
      "learning_rate": 7.003428162095784e-07,
      "loss": 0.9951,
      "step": 3650
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.04864778742194176,
      "learning_rate": 7.022615636512484e-07,
      "loss": 0.9957,
      "step": 3660
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.09859640896320343,
      "learning_rate": 7.041803110929185e-07,
      "loss": 0.9929,
      "step": 3670
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.09181062877178192,
      "learning_rate": 7.060990585345886e-07,
      "loss": 0.9919,
      "step": 3680
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.15889447927474976,
      "learning_rate": 7.080178059762587e-07,
      "loss": 0.993,
      "step": 3690
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.09621824324131012,
      "learning_rate": 7.099365534179287e-07,
      "loss": 0.9905,
      "step": 3700
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.08575550466775894,
      "learning_rate": 7.118553008595988e-07,
      "loss": 0.9939,
      "step": 3710
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.11606623977422714,
      "learning_rate": 7.137740483012689e-07,
      "loss": 0.9921,
      "step": 3720
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.12592573463916779,
      "learning_rate": 7.15692795742939e-07,
      "loss": 0.9933,
      "step": 3730
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.11156260967254639,
      "learning_rate": 7.17611543184609e-07,
      "loss": 0.9939,
      "step": 3740
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.16888302564620972,
      "learning_rate": 7.195302906262791e-07,
      "loss": 0.9927,
      "step": 3750
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.05811610072851181,
      "learning_rate": 7.214490380679492e-07,
      "loss": 0.9971,
      "step": 3760
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.0648038238286972,
      "learning_rate": 7.233677855096192e-07,
      "loss": 0.9942,
      "step": 3770
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.05916750431060791,
      "learning_rate": 7.252865329512893e-07,
      "loss": 0.9928,
      "step": 3780
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.0577201247215271,
      "learning_rate": 7.272052803929594e-07,
      "loss": 0.9937,
      "step": 3790
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.14994658529758453,
      "learning_rate": 7.291240278346295e-07,
      "loss": 0.9934,
      "step": 3800
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.1092350035905838,
      "learning_rate": 7.310427752762995e-07,
      "loss": 0.9914,
      "step": 3810
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.06681845337152481,
      "learning_rate": 7.329615227179696e-07,
      "loss": 0.9926,
      "step": 3820
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.09951276332139969,
      "learning_rate": 7.348802701596397e-07,
      "loss": 0.9941,
      "step": 3830
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.10435248166322708,
      "learning_rate": 7.367990176013098e-07,
      "loss": 0.9911,
      "step": 3840
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.09248413145542145,
      "learning_rate": 7.387177650429798e-07,
      "loss": 0.9928,
      "step": 3850
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.0719660297036171,
      "learning_rate": 7.4063651248465e-07,
      "loss": 0.9938,
      "step": 3860
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.07644528150558472,
      "learning_rate": 7.4255525992632e-07,
      "loss": 0.9905,
      "step": 3870
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.07239459455013275,
      "learning_rate": 7.4447400736799e-07,
      "loss": 0.9927,
      "step": 3880
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.09696169942617416,
      "learning_rate": 7.463927548096601e-07,
      "loss": 0.9922,
      "step": 3890
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.134473517537117,
      "learning_rate": 7.483115022513302e-07,
      "loss": 0.9935,
      "step": 3900
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.04703366756439209,
      "learning_rate": 7.502302496930004e-07,
      "loss": 0.9923,
      "step": 3910
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.10332293808460236,
      "learning_rate": 7.521489971346704e-07,
      "loss": 0.9947,
      "step": 3920
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.0960303246974945,
      "learning_rate": 7.540677445763406e-07,
      "loss": 0.9933,
      "step": 3930
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.09719212353229523,
      "learning_rate": 7.559864920180106e-07,
      "loss": 0.9916,
      "step": 3940
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.11049799621105194,
      "learning_rate": 7.579052394596807e-07,
      "loss": 0.9923,
      "step": 3950
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.07870466262102127,
      "learning_rate": 7.598239869013507e-07,
      "loss": 0.9907,
      "step": 3960
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.12369512021541595,
      "learning_rate": 7.617427343430209e-07,
      "loss": 0.9881,
      "step": 3970
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.11466123163700104,
      "learning_rate": 7.636614817846909e-07,
      "loss": 0.9944,
      "step": 3980
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.07550898939371109,
      "learning_rate": 7.65580229226361e-07,
      "loss": 0.9913,
      "step": 3990
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.1001652255654335,
      "learning_rate": 7.674989766680311e-07,
      "loss": 0.9902,
      "step": 4000
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.10350809246301651,
      "learning_rate": 7.694177241097012e-07,
      "loss": 0.993,
      "step": 4010
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.12139185518026352,
      "learning_rate": 7.713364715513712e-07,
      "loss": 0.9912,
      "step": 4020
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.13503433763980865,
      "learning_rate": 7.732552189930413e-07,
      "loss": 0.9919,
      "step": 4030
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.10910269618034363,
      "learning_rate": 7.751739664347114e-07,
      "loss": 0.9907,
      "step": 4040
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.12657350301742554,
      "learning_rate": 7.770927138763814e-07,
      "loss": 0.9888,
      "step": 4050
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.05786283686757088,
      "learning_rate": 7.790114613180515e-07,
      "loss": 0.993,
      "step": 4060
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.07674966752529144,
      "learning_rate": 7.809302087597215e-07,
      "loss": 0.9914,
      "step": 4070
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.12078571319580078,
      "learning_rate": 7.828489562013917e-07,
      "loss": 0.9914,
      "step": 4080
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.1899256408214569,
      "learning_rate": 7.847677036430617e-07,
      "loss": 0.9899,
      "step": 4090
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.07815311849117279,
      "learning_rate": 7.866864510847318e-07,
      "loss": 0.9882,
      "step": 4100
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.09783992916345596,
      "learning_rate": 7.886051985264019e-07,
      "loss": 0.9928,
      "step": 4110
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.09239166229963303,
      "learning_rate": 7.90523945968072e-07,
      "loss": 0.991,
      "step": 4120
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.1123340055346489,
      "learning_rate": 7.92442693409742e-07,
      "loss": 0.9894,
      "step": 4130
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.11851730942726135,
      "learning_rate": 7.943614408514121e-07,
      "loss": 0.9902,
      "step": 4140
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.07506045699119568,
      "learning_rate": 7.962801882930822e-07,
      "loss": 0.994,
      "step": 4150
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.0907919704914093,
      "learning_rate": 7.981989357347523e-07,
      "loss": 0.9907,
      "step": 4160
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.15261855721473694,
      "learning_rate": 8.001176831764223e-07,
      "loss": 0.9907,
      "step": 4170
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.0910738855600357,
      "learning_rate": 8.020364306180924e-07,
      "loss": 0.9908,
      "step": 4180
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.11516007035970688,
      "learning_rate": 8.039551780597625e-07,
      "loss": 0.9903,
      "step": 4190
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.07784128934144974,
      "learning_rate": 8.058739255014325e-07,
      "loss": 0.9889,
      "step": 4200
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.06351088732481003,
      "learning_rate": 8.077926729431026e-07,
      "loss": 0.9848,
      "step": 4210
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.08192408829927444,
      "learning_rate": 8.097114203847727e-07,
      "loss": 0.9896,
      "step": 4220
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.11244099587202072,
      "learning_rate": 8.116301678264428e-07,
      "loss": 0.9911,
      "step": 4230
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.21149280667304993,
      "learning_rate": 8.135489152681128e-07,
      "loss": 0.9899,
      "step": 4240
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.1275319904088974,
      "learning_rate": 8.154676627097831e-07,
      "loss": 0.9854,
      "step": 4250
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.10734979808330536,
      "learning_rate": 8.173864101514531e-07,
      "loss": 0.9876,
      "step": 4260
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.14969807863235474,
      "learning_rate": 8.193051575931232e-07,
      "loss": 0.9935,
      "step": 4270
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.09933243691921234,
      "learning_rate": 8.212239050347932e-07,
      "loss": 0.9898,
      "step": 4280
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.0968184843659401,
      "learning_rate": 8.231426524764634e-07,
      "loss": 0.9905,
      "step": 4290
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.13625262677669525,
      "learning_rate": 8.250613999181334e-07,
      "loss": 0.9891,
      "step": 4300
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.0966690182685852,
      "learning_rate": 8.269801473598035e-07,
      "loss": 0.9918,
      "step": 4310
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.1047445759177208,
      "learning_rate": 8.288988948014735e-07,
      "loss": 0.9906,
      "step": 4320
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.16115440428256989,
      "learning_rate": 8.308176422431437e-07,
      "loss": 0.9903,
      "step": 4330
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.08664358407258987,
      "learning_rate": 8.327363896848137e-07,
      "loss": 0.9874,
      "step": 4340
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.112075075507164,
      "learning_rate": 8.346551371264837e-07,
      "loss": 0.9913,
      "step": 4350
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.12064921110868454,
      "learning_rate": 8.365738845681539e-07,
      "loss": 0.9887,
      "step": 4360
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.17597676813602448,
      "learning_rate": 8.384926320098239e-07,
      "loss": 0.9891,
      "step": 4370
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.09760839492082596,
      "learning_rate": 8.40411379451494e-07,
      "loss": 0.9916,
      "step": 4380
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.1253148317337036,
      "learning_rate": 8.42330126893164e-07,
      "loss": 0.989,
      "step": 4390
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.20593871176242828,
      "learning_rate": 8.442488743348342e-07,
      "loss": 0.9833,
      "step": 4400
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.15127518773078918,
      "learning_rate": 8.461676217765042e-07,
      "loss": 0.9885,
      "step": 4410
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.11975551396608353,
      "learning_rate": 8.480863692181743e-07,
      "loss": 0.9936,
      "step": 4420
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.07349167764186859,
      "learning_rate": 8.500051166598443e-07,
      "loss": 0.9865,
      "step": 4430
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.13162539899349213,
      "learning_rate": 8.519238641015145e-07,
      "loss": 0.9849,
      "step": 4440
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.16806578636169434,
      "learning_rate": 8.538426115431845e-07,
      "loss": 0.9868,
      "step": 4450
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.17965558171272278,
      "learning_rate": 8.557613589848546e-07,
      "loss": 0.9841,
      "step": 4460
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.13365592062473297,
      "learning_rate": 8.576801064265247e-07,
      "loss": 0.9838,
      "step": 4470
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.17037853598594666,
      "learning_rate": 8.595988538681948e-07,
      "loss": 0.9874,
      "step": 4480
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.20422886312007904,
      "learning_rate": 8.615176013098648e-07,
      "loss": 0.9879,
      "step": 4490
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.13434863090515137,
      "learning_rate": 8.634363487515348e-07,
      "loss": 0.9837,
      "step": 4500
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.17359626293182373,
      "learning_rate": 8.65355096193205e-07,
      "loss": 0.99,
      "step": 4510
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.13509529829025269,
      "learning_rate": 8.67273843634875e-07,
      "loss": 0.9836,
      "step": 4520
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.14618609845638275,
      "learning_rate": 8.691925910765451e-07,
      "loss": 0.9857,
      "step": 4530
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.11848451942205429,
      "learning_rate": 8.711113385182151e-07,
      "loss": 0.9843,
      "step": 4540
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.10362444818019867,
      "learning_rate": 8.730300859598853e-07,
      "loss": 0.9905,
      "step": 4550
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.142228901386261,
      "learning_rate": 8.749488334015553e-07,
      "loss": 0.9848,
      "step": 4560
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.17158319056034088,
      "learning_rate": 8.768675808432254e-07,
      "loss": 0.9859,
      "step": 4570
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.08372943103313446,
      "learning_rate": 8.787863282848955e-07,
      "loss": 0.9836,
      "step": 4580
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.14360231161117554,
      "learning_rate": 8.807050757265657e-07,
      "loss": 0.9887,
      "step": 4590
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.20449841022491455,
      "learning_rate": 8.826238231682357e-07,
      "loss": 0.9829,
      "step": 4600
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.14089231193065643,
      "learning_rate": 8.845425706099059e-07,
      "loss": 0.9833,
      "step": 4610
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.09355634450912476,
      "learning_rate": 8.864613180515759e-07,
      "loss": 0.985,
      "step": 4620
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.20468340814113617,
      "learning_rate": 8.88380065493246e-07,
      "loss": 0.9896,
      "step": 4630
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.158100888133049,
      "learning_rate": 8.90298812934916e-07,
      "loss": 0.9828,
      "step": 4640
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.19927608966827393,
      "learning_rate": 8.922175603765862e-07,
      "loss": 0.9853,
      "step": 4650
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.16803690791130066,
      "learning_rate": 8.941363078182562e-07,
      "loss": 0.9853,
      "step": 4660
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.1815691888332367,
      "learning_rate": 8.960550552599262e-07,
      "loss": 0.9853,
      "step": 4670
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.30703645944595337,
      "learning_rate": 8.979738027015964e-07,
      "loss": 0.9819,
      "step": 4680
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.2100110948085785,
      "learning_rate": 8.998925501432664e-07,
      "loss": 0.9808,
      "step": 4690
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.13945052027702332,
      "learning_rate": 9.018112975849365e-07,
      "loss": 0.9824,
      "step": 4700
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.2608013451099396,
      "learning_rate": 9.037300450266065e-07,
      "loss": 0.9828,
      "step": 4710
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.18841801583766937,
      "learning_rate": 9.056487924682767e-07,
      "loss": 0.9808,
      "step": 4720
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.14650943875312805,
      "learning_rate": 9.075675399099467e-07,
      "loss": 0.9791,
      "step": 4730
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.21047791838645935,
      "learning_rate": 9.094862873516168e-07,
      "loss": 0.9838,
      "step": 4740
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.17936518788337708,
      "learning_rate": 9.114050347932868e-07,
      "loss": 0.9772,
      "step": 4750
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.20952719449996948,
      "learning_rate": 9.13323782234957e-07,
      "loss": 0.9835,
      "step": 4760
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.1624646782875061,
      "learning_rate": 9.15242529676627e-07,
      "loss": 0.9741,
      "step": 4770
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.26253876090049744,
      "learning_rate": 9.171612771182971e-07,
      "loss": 0.9839,
      "step": 4780
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.20139701664447784,
      "learning_rate": 9.190800245599671e-07,
      "loss": 0.983,
      "step": 4790
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.302976131439209,
      "learning_rate": 9.209987720016373e-07,
      "loss": 0.9834,
      "step": 4800
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.11667444556951523,
      "learning_rate": 9.229175194433073e-07,
      "loss": 0.9817,
      "step": 4810
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.19623394310474396,
      "learning_rate": 9.248362668849773e-07,
      "loss": 0.9814,
      "step": 4820
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.23934948444366455,
      "learning_rate": 9.267550143266475e-07,
      "loss": 0.9807,
      "step": 4830
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.1704292893409729,
      "learning_rate": 9.286737617683175e-07,
      "loss": 0.9756,
      "step": 4840
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.40305328369140625,
      "learning_rate": 9.305925092099876e-07,
      "loss": 0.9724,
      "step": 4850
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.1523485779762268,
      "learning_rate": 9.325112566516576e-07,
      "loss": 0.9849,
      "step": 4860
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.2198527604341507,
      "learning_rate": 9.344300040933278e-07,
      "loss": 0.9794,
      "step": 4870
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.2999301850795746,
      "learning_rate": 9.363487515349978e-07,
      "loss": 0.9832,
      "step": 4880
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.9752696752548218,
      "eval_runtime": 54.7743,
      "eval_samples_per_second": 62.967,
      "eval_steps_per_second": 7.887,
      "step": 4886
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.29508334398269653,
      "learning_rate": 9.382674989766679e-07,
      "loss": 0.9748,
      "step": 4890
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.34914645552635193,
      "learning_rate": 9.401862464183379e-07,
      "loss": 0.9722,
      "step": 4900
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.32686036825180054,
      "learning_rate": 9.421049938600081e-07,
      "loss": 0.9733,
      "step": 4910
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.29212257266044617,
      "learning_rate": 9.440237413016782e-07,
      "loss": 0.9846,
      "step": 4920
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.22156397998332977,
      "learning_rate": 9.459424887433484e-07,
      "loss": 0.9735,
      "step": 4930
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.3204995393753052,
      "learning_rate": 9.478612361850184e-07,
      "loss": 0.9677,
      "step": 4940
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.26309141516685486,
      "learning_rate": 9.497799836266885e-07,
      "loss": 0.9798,
      "step": 4950
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.3917821943759918,
      "learning_rate": 9.516987310683585e-07,
      "loss": 0.9735,
      "step": 4960
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.5680138468742371,
      "learning_rate": 9.536174785100287e-07,
      "loss": 0.9814,
      "step": 4970
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.5519505739212036,
      "learning_rate": 9.555362259516987e-07,
      "loss": 0.9761,
      "step": 4980
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.22931717336177826,
      "learning_rate": 9.574549733933689e-07,
      "loss": 0.9656,
      "step": 4990
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.512885570526123,
      "learning_rate": 9.593737208350388e-07,
      "loss": 0.9566,
      "step": 5000
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.2888445258140564,
      "learning_rate": 9.61292468276709e-07,
      "loss": 0.9714,
      "step": 5010
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.3276253938674927,
      "learning_rate": 9.632112157183789e-07,
      "loss": 0.9725,
      "step": 5020
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.33691251277923584,
      "learning_rate": 9.65129963160049e-07,
      "loss": 0.9731,
      "step": 5030
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.2185465544462204,
      "learning_rate": 9.670487106017192e-07,
      "loss": 0.969,
      "step": 5040
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.2188493311405182,
      "learning_rate": 9.689674580433891e-07,
      "loss": 0.9659,
      "step": 5050
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.3027728497982025,
      "learning_rate": 9.708862054850593e-07,
      "loss": 0.9629,
      "step": 5060
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.5067316889762878,
      "learning_rate": 9.728049529267294e-07,
      "loss": 0.9601,
      "step": 5070
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.3434458374977112,
      "learning_rate": 9.747237003683994e-07,
      "loss": 0.9539,
      "step": 5080
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.45559942722320557,
      "learning_rate": 9.766424478100695e-07,
      "loss": 0.9708,
      "step": 5090
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8189027309417725,
      "learning_rate": 9.785611952517397e-07,
      "loss": 0.956,
      "step": 5100
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.361594021320343,
      "learning_rate": 9.804799426934096e-07,
      "loss": 0.9622,
      "step": 5110
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.28939497470855713,
      "learning_rate": 9.823986901350798e-07,
      "loss": 0.9601,
      "step": 5120
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.47779732942581177,
      "learning_rate": 9.843174375767497e-07,
      "loss": 0.9525,
      "step": 5130
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.4787381589412689,
      "learning_rate": 9.862361850184198e-07,
      "loss": 0.9629,
      "step": 5140
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.30678215622901917,
      "learning_rate": 9.8815493246009e-07,
      "loss": 0.9614,
      "step": 5150
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.34074023365974426,
      "learning_rate": 9.9007367990176e-07,
      "loss": 0.979,
      "step": 5160
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.44440481066703796,
      "learning_rate": 9.9199242734343e-07,
      "loss": 0.9569,
      "step": 5170
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.7005957365036011,
      "learning_rate": 9.939111747851002e-07,
      "loss": 0.9642,
      "step": 5180
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.6492758989334106,
      "learning_rate": 9.958299222267702e-07,
      "loss": 0.9679,
      "step": 5190
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.4642639756202698,
      "learning_rate": 9.977486696684403e-07,
      "loss": 0.9571,
      "step": 5200
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.34881022572517395,
      "learning_rate": 9.996674171101105e-07,
      "loss": 0.9679,
      "step": 5210
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.4925949275493622,
      "learning_rate": 1.0015861645517804e-06,
      "loss": 0.9613,
      "step": 5220
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.33499762415885925,
      "learning_rate": 1.0035049119934506e-06,
      "loss": 0.9672,
      "step": 5230
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.45999637246131897,
      "learning_rate": 1.0054236594351205e-06,
      "loss": 0.9594,
      "step": 5240
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.3242577612400055,
      "learning_rate": 1.0073424068767907e-06,
      "loss": 0.9525,
      "step": 5250
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.4646061956882477,
      "learning_rate": 1.0092611543184608e-06,
      "loss": 0.9647,
      "step": 5260
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.45758000016212463,
      "learning_rate": 1.011179901760131e-06,
      "loss": 0.958,
      "step": 5270
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.40084317326545715,
      "learning_rate": 1.0130986492018011e-06,
      "loss": 0.9628,
      "step": 5280
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.21909278631210327,
      "learning_rate": 1.015017396643471e-06,
      "loss": 0.9694,
      "step": 5290
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.30396828055381775,
      "learning_rate": 1.0169361440851412e-06,
      "loss": 0.9583,
      "step": 5300
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.6841109395027161,
      "learning_rate": 1.0188548915268113e-06,
      "loss": 0.9514,
      "step": 5310
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.7204678058624268,
      "learning_rate": 1.0207736389684813e-06,
      "loss": 0.9423,
      "step": 5320
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8534573912620544,
      "learning_rate": 1.0226923864101514e-06,
      "loss": 0.9577,
      "step": 5330
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.2831140458583832,
      "learning_rate": 1.0246111338518214e-06,
      "loss": 0.9577,
      "step": 5340
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.46303126215934753,
      "learning_rate": 1.0265298812934915e-06,
      "loss": 0.9595,
      "step": 5350
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.26637256145477295,
      "learning_rate": 1.0284486287351617e-06,
      "loss": 0.954,
      "step": 5360
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.43608155846595764,
      "learning_rate": 1.0303673761768316e-06,
      "loss": 0.9577,
      "step": 5370
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.5589589476585388,
      "learning_rate": 1.0322861236185018e-06,
      "loss": 0.9556,
      "step": 5380
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.4756903648376465,
      "learning_rate": 1.034204871060172e-06,
      "loss": 0.9505,
      "step": 5390
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.692567765712738,
      "learning_rate": 1.0361236185018419e-06,
      "loss": 0.9694,
      "step": 5400
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.3091309368610382,
      "learning_rate": 1.038042365943512e-06,
      "loss": 0.9564,
      "step": 5410
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.4447256326675415,
      "learning_rate": 1.0399611133851822e-06,
      "loss": 0.965,
      "step": 5420
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.36336687207221985,
      "learning_rate": 1.041879860826852e-06,
      "loss": 0.9593,
      "step": 5430
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8403950929641724,
      "learning_rate": 1.0437986082685223e-06,
      "loss": 0.9526,
      "step": 5440
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9146859645843506,
      "learning_rate": 1.0457173557101922e-06,
      "loss": 0.949,
      "step": 5450
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.39466536045074463,
      "learning_rate": 1.0476361031518623e-06,
      "loss": 0.9566,
      "step": 5460
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.4282681941986084,
      "learning_rate": 1.0495548505935325e-06,
      "loss": 0.9515,
      "step": 5470
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.421712875366211,
      "learning_rate": 1.0514735980352024e-06,
      "loss": 0.9486,
      "step": 5480
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.6227210760116577,
      "learning_rate": 1.0533923454768726e-06,
      "loss": 0.9612,
      "step": 5490
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.4698435068130493,
      "learning_rate": 1.0553110929185427e-06,
      "loss": 0.9573,
      "step": 5500
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.3191728889942169,
      "learning_rate": 1.0572298403602127e-06,
      "loss": 0.951,
      "step": 5510
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.35744228959083557,
      "learning_rate": 1.0591485878018828e-06,
      "loss": 0.9588,
      "step": 5520
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.5788488984107971,
      "learning_rate": 1.061067335243553e-06,
      "loss": 0.9396,
      "step": 5530
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.3945732116699219,
      "learning_rate": 1.062986082685223e-06,
      "loss": 0.9516,
      "step": 5540
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.42959731817245483,
      "learning_rate": 1.064904830126893e-06,
      "loss": 0.948,
      "step": 5550
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.7700499296188354,
      "learning_rate": 1.066823577568563e-06,
      "loss": 0.9518,
      "step": 5560
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.37502408027648926,
      "learning_rate": 1.0687423250102332e-06,
      "loss": 0.9574,
      "step": 5570
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.6261919736862183,
      "learning_rate": 1.0706610724519033e-06,
      "loss": 0.9698,
      "step": 5580
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.869029700756073,
      "learning_rate": 1.0725798198935732e-06,
      "loss": 0.9504,
      "step": 5590
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.5572662949562073,
      "learning_rate": 1.0744985673352436e-06,
      "loss": 0.9487,
      "step": 5600
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.5272384285926819,
      "learning_rate": 1.0764173147769135e-06,
      "loss": 0.9594,
      "step": 5610
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.49363797903060913,
      "learning_rate": 1.0783360622185837e-06,
      "loss": 0.9588,
      "step": 5620
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.8330487608909607,
      "learning_rate": 1.0802548096602538e-06,
      "loss": 0.9622,
      "step": 5630
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.7178264260292053,
      "learning_rate": 1.0821735571019238e-06,
      "loss": 0.9512,
      "step": 5640
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.851844072341919,
      "learning_rate": 1.084092304543594e-06,
      "loss": 0.9495,
      "step": 5650
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.45666220784187317,
      "learning_rate": 1.0860110519852639e-06,
      "loss": 0.9358,
      "step": 5660
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.44568219780921936,
      "learning_rate": 1.087929799426934e-06,
      "loss": 0.9584,
      "step": 5670
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.6994748115539551,
      "learning_rate": 1.0898485468686042e-06,
      "loss": 0.9523,
      "step": 5680
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4179335832595825,
      "learning_rate": 1.0917672943102741e-06,
      "loss": 0.9512,
      "step": 5690
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.0123053789138794,
      "learning_rate": 1.0936860417519443e-06,
      "loss": 0.9711,
      "step": 5700
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.28707826137542725,
      "learning_rate": 1.0956047891936144e-06,
      "loss": 0.9408,
      "step": 5710
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.5791871547698975,
      "learning_rate": 1.0975235366352844e-06,
      "loss": 0.959,
      "step": 5720
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.888912558555603,
      "learning_rate": 1.0994422840769545e-06,
      "loss": 0.947,
      "step": 5730
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.8278868198394775,
      "learning_rate": 1.1013610315186247e-06,
      "loss": 0.9459,
      "step": 5740
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.5235679745674133,
      "learning_rate": 1.1032797789602946e-06,
      "loss": 0.9436,
      "step": 5750
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.4255140721797943,
      "learning_rate": 1.1051985264019647e-06,
      "loss": 0.9378,
      "step": 5760
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.516711950302124,
      "learning_rate": 1.1071172738436347e-06,
      "loss": 0.9425,
      "step": 5770
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.7255879044532776,
      "learning_rate": 1.1090360212853048e-06,
      "loss": 0.9529,
      "step": 5780
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.44585269689559937,
      "learning_rate": 1.110954768726975e-06,
      "loss": 0.9512,
      "step": 5790
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.4202837646007538,
      "learning_rate": 1.112873516168645e-06,
      "loss": 0.9532,
      "step": 5800
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.5856006145477295,
      "learning_rate": 1.114792263610315e-06,
      "loss": 0.9477,
      "step": 5810
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.39887529611587524,
      "learning_rate": 1.1167110110519852e-06,
      "loss": 0.9586,
      "step": 5820
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.3354381322860718,
      "learning_rate": 1.1186297584936552e-06,
      "loss": 0.9528,
      "step": 5830
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.954674243927002,
      "learning_rate": 1.1205485059353253e-06,
      "loss": 0.9441,
      "step": 5840
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.263171911239624,
      "learning_rate": 1.1224672533769955e-06,
      "loss": 0.9464,
      "step": 5850
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.5748897194862366,
      "learning_rate": 1.1243860008186654e-06,
      "loss": 0.9571,
      "step": 5860
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.5867511034011841,
      "learning_rate": 1.1263047482603356e-06,
      "loss": 0.944,
      "step": 5870
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.5909374952316284,
      "learning_rate": 1.1282234957020055e-06,
      "loss": 0.9391,
      "step": 5880
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.3255351781845093,
      "learning_rate": 1.1301422431436757e-06,
      "loss": 0.9559,
      "step": 5890
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.5518824458122253,
      "learning_rate": 1.1320609905853458e-06,
      "loss": 0.9459,
      "step": 5900
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.4718870222568512,
      "learning_rate": 1.1339797380270157e-06,
      "loss": 0.9437,
      "step": 5910
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.4974406659603119,
      "learning_rate": 1.1358984854686859e-06,
      "loss": 0.9425,
      "step": 5920
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.4668625593185425,
      "learning_rate": 1.137817232910356e-06,
      "loss": 0.9423,
      "step": 5930
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.29373523592948914,
      "learning_rate": 1.1397359803520262e-06,
      "loss": 0.9484,
      "step": 5940
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9781979918479919,
      "learning_rate": 1.1416547277936961e-06,
      "loss": 0.9497,
      "step": 5950
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.5401967167854309,
      "learning_rate": 1.1435734752353663e-06,
      "loss": 0.9437,
      "step": 5960
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.43299242854118347,
      "learning_rate": 1.1454922226770364e-06,
      "loss": 0.9451,
      "step": 5970
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.48214974999427795,
      "learning_rate": 1.1474109701187064e-06,
      "loss": 0.9506,
      "step": 5980
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.0033848285675049,
      "learning_rate": 1.1493297175603765e-06,
      "loss": 0.9493,
      "step": 5990
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.7182579636573792,
      "learning_rate": 1.1512484650020467e-06,
      "loss": 0.9389,
      "step": 6000
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.46473750472068787,
      "learning_rate": 1.1531672124437166e-06,
      "loss": 0.9407,
      "step": 6010
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.7013241052627563,
      "learning_rate": 1.1550859598853868e-06,
      "loss": 0.9428,
      "step": 6020
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.4730825424194336,
      "learning_rate": 1.157004707327057e-06,
      "loss": 0.944,
      "step": 6030
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.37993475794792175,
      "learning_rate": 1.1589234547687269e-06,
      "loss": 0.9598,
      "step": 6040
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.7115862965583801,
      "learning_rate": 1.160842202210397e-06,
      "loss": 0.952,
      "step": 6050
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.47615423798561096,
      "learning_rate": 1.162760949652067e-06,
      "loss": 0.9427,
      "step": 6060
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.56107097864151,
      "learning_rate": 1.164679697093737e-06,
      "loss": 0.9405,
      "step": 6070
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.6176039576530457,
      "learning_rate": 1.1665984445354072e-06,
      "loss": 0.9205,
      "step": 6080
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.7666685581207275,
      "learning_rate": 1.1685171919770772e-06,
      "loss": 0.9443,
      "step": 6090
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.7009537220001221,
      "learning_rate": 1.1704359394187473e-06,
      "loss": 0.9354,
      "step": 6100
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.4642140865325928,
      "learning_rate": 1.1723546868604175e-06,
      "loss": 0.9514,
      "step": 6110
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.5413902997970581,
      "learning_rate": 1.1742734343020874e-06,
      "loss": 0.9397,
      "step": 6120
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.6469710469245911,
      "learning_rate": 1.1761921817437576e-06,
      "loss": 0.9528,
      "step": 6130
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.792822003364563,
      "learning_rate": 1.1781109291854277e-06,
      "loss": 0.9464,
      "step": 6140
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.5836959481239319,
      "learning_rate": 1.1800296766270977e-06,
      "loss": 0.9378,
      "step": 6150
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.3840739130973816,
      "learning_rate": 1.1819484240687678e-06,
      "loss": 0.9525,
      "step": 6160
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.578011155128479,
      "learning_rate": 1.1838671715104378e-06,
      "loss": 0.9163,
      "step": 6170
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.45682215690612793,
      "learning_rate": 1.185785918952108e-06,
      "loss": 0.9448,
      "step": 6180
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.0815078020095825,
      "learning_rate": 1.187704666393778e-06,
      "loss": 0.9547,
      "step": 6190
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.5811464190483093,
      "learning_rate": 1.189623413835448e-06,
      "loss": 0.9507,
      "step": 6200
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.3441387712955475,
      "learning_rate": 1.1915421612771181e-06,
      "loss": 0.9429,
      "step": 6210
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.5118454694747925,
      "learning_rate": 1.1934609087187883e-06,
      "loss": 0.9372,
      "step": 6220
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.241866111755371,
      "learning_rate": 1.1953796561604582e-06,
      "loss": 0.9525,
      "step": 6230
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.4872983396053314,
      "learning_rate": 1.1972984036021284e-06,
      "loss": 0.943,
      "step": 6240
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.4079504609107971,
      "learning_rate": 1.1992171510437985e-06,
      "loss": 0.9482,
      "step": 6250
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.3479340672492981,
      "learning_rate": 1.2011358984854685e-06,
      "loss": 0.9407,
      "step": 6260
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.408169150352478,
      "learning_rate": 1.2030546459271386e-06,
      "loss": 0.9312,
      "step": 6270
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.1818212270736694,
      "learning_rate": 1.2049733933688086e-06,
      "loss": 0.9449,
      "step": 6280
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.766089141368866,
      "learning_rate": 1.2068921408104787e-06,
      "loss": 0.9378,
      "step": 6290
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.5216675400733948,
      "learning_rate": 1.2088108882521489e-06,
      "loss": 0.9344,
      "step": 6300
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.30144086480140686,
      "learning_rate": 1.2107296356938188e-06,
      "loss": 0.9561,
      "step": 6310
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.6995535492897034,
      "learning_rate": 1.212648383135489e-06,
      "loss": 0.9471,
      "step": 6320
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.2091960459947586,
      "learning_rate": 1.2145671305771591e-06,
      "loss": 0.953,
      "step": 6330
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.7725284695625305,
      "learning_rate": 1.216485878018829e-06,
      "loss": 0.9374,
      "step": 6340
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.640099287033081,
      "learning_rate": 1.2184046254604992e-06,
      "loss": 0.9479,
      "step": 6350
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9802278876304626,
      "learning_rate": 1.2203233729021694e-06,
      "loss": 0.9454,
      "step": 6360
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8196399211883545,
      "learning_rate": 1.2222421203438393e-06,
      "loss": 0.946,
      "step": 6370
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.6724802255630493,
      "learning_rate": 1.2241608677855094e-06,
      "loss": 0.9403,
      "step": 6380
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3374896049499512,
      "learning_rate": 1.2260796152271794e-06,
      "loss": 0.9348,
      "step": 6390
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.5573810935020447,
      "learning_rate": 1.2279983626688495e-06,
      "loss": 0.946,
      "step": 6400
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2149543762207031,
      "learning_rate": 1.2299171101105197e-06,
      "loss": 0.9569,
      "step": 6410
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9224502444267273,
      "learning_rate": 1.2318358575521896e-06,
      "loss": 0.9598,
      "step": 6420
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.6007453203201294,
      "learning_rate": 1.2337546049938598e-06,
      "loss": 0.9373,
      "step": 6430
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.0703836679458618,
      "learning_rate": 1.23567335243553e-06,
      "loss": 0.9406,
      "step": 6440
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.6683436632156372,
      "learning_rate": 1.2375920998772003e-06,
      "loss": 0.9254,
      "step": 6450
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.5607090592384338,
      "learning_rate": 1.2395108473188702e-06,
      "loss": 0.9477,
      "step": 6460
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.5609524250030518,
      "learning_rate": 1.2414295947605404e-06,
      "loss": 0.9278,
      "step": 6470
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.6091378331184387,
      "learning_rate": 1.2433483422022103e-06,
      "loss": 0.9378,
      "step": 6480
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.4764232635498047,
      "learning_rate": 1.2452670896438805e-06,
      "loss": 0.9459,
      "step": 6490
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.4543370306491852,
      "learning_rate": 1.2471858370855506e-06,
      "loss": 0.9421,
      "step": 6500
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.42853641510009766,
      "learning_rate": 1.2491045845272206e-06,
      "loss": 0.9304,
      "step": 6510
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.31137704849243164,
      "learning_rate": 1.2510233319688907e-06,
      "loss": 0.9327,
      "step": 6520
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.5099424719810486,
      "learning_rate": 1.2529420794105609e-06,
      "loss": 0.9468,
      "step": 6530
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.5699959993362427,
      "learning_rate": 1.2548608268522308e-06,
      "loss": 0.9371,
      "step": 6540
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.4368289113044739,
      "learning_rate": 1.256779574293901e-06,
      "loss": 0.9448,
      "step": 6550
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.8702236413955688,
      "learning_rate": 1.258698321735571e-06,
      "loss": 0.9336,
      "step": 6560
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.8469472527503967,
      "learning_rate": 1.260617069177241e-06,
      "loss": 0.9343,
      "step": 6570
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.7925068736076355,
      "learning_rate": 1.2625358166189112e-06,
      "loss": 0.9366,
      "step": 6580
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.5858622789382935,
      "learning_rate": 1.2644545640605811e-06,
      "loss": 0.9274,
      "step": 6590
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.47339630126953125,
      "learning_rate": 1.2663733115022513e-06,
      "loss": 0.9415,
      "step": 6600
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.7344330549240112,
      "learning_rate": 1.2682920589439214e-06,
      "loss": 0.9509,
      "step": 6610
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.7974898815155029,
      "learning_rate": 1.2702108063855914e-06,
      "loss": 0.9517,
      "step": 6620
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.2578082084655762,
      "learning_rate": 1.2721295538272615e-06,
      "loss": 0.9257,
      "step": 6630
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.6431311368942261,
      "learning_rate": 1.2740483012689317e-06,
      "loss": 0.9345,
      "step": 6640
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.7124678492546082,
      "learning_rate": 1.2759670487106016e-06,
      "loss": 0.9269,
      "step": 6650
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.33415839076042175,
      "learning_rate": 1.2778857961522718e-06,
      "loss": 0.9342,
      "step": 6660
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.7094359397888184,
      "learning_rate": 1.279804543593942e-06,
      "loss": 0.9494,
      "step": 6670
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9511408805847168,
      "learning_rate": 1.2817232910356119e-06,
      "loss": 0.9359,
      "step": 6680
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.2325348854064941,
      "learning_rate": 1.283642038477282e-06,
      "loss": 0.9222,
      "step": 6690
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9152774810791016,
      "learning_rate": 1.285560785918952e-06,
      "loss": 0.9368,
      "step": 6700
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9937361478805542,
      "learning_rate": 1.287479533360622e-06,
      "loss": 0.9311,
      "step": 6710
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.5547901391983032,
      "learning_rate": 1.2893982808022922e-06,
      "loss": 0.9366,
      "step": 6720
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.7273530960083008,
      "learning_rate": 1.2913170282439622e-06,
      "loss": 0.9366,
      "step": 6730
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.2415789365768433,
      "learning_rate": 1.2932357756856323e-06,
      "loss": 0.91,
      "step": 6740
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.2771589756011963,
      "learning_rate": 1.2951545231273025e-06,
      "loss": 0.9461,
      "step": 6750
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.8793092370033264,
      "learning_rate": 1.2970732705689724e-06,
      "loss": 0.9517,
      "step": 6760
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0529162883758545,
      "learning_rate": 1.2989920180106426e-06,
      "loss": 0.9306,
      "step": 6770
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.7216672897338867,
      "learning_rate": 1.3009107654523127e-06,
      "loss": 0.935,
      "step": 6780
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.8175162076950073,
      "learning_rate": 1.3028295128939827e-06,
      "loss": 0.9495,
      "step": 6790
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.2285363674163818,
      "learning_rate": 1.3047482603356528e-06,
      "loss": 0.9551,
      "step": 6800
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.8173148036003113,
      "learning_rate": 1.3066670077773228e-06,
      "loss": 0.9298,
      "step": 6810
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.3430655300617218,
      "learning_rate": 1.308585755218993e-06,
      "loss": 0.9295,
      "step": 6820
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.5921844840049744,
      "learning_rate": 1.310504502660663e-06,
      "loss": 0.9177,
      "step": 6830
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9468276500701904,
      "learning_rate": 1.312423250102333e-06,
      "loss": 0.9337,
      "step": 6840
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.5005115270614624,
      "learning_rate": 1.3143419975440031e-06,
      "loss": 0.9237,
      "step": 6850
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.1414697170257568,
      "learning_rate": 1.3162607449856733e-06,
      "loss": 0.9402,
      "step": 6860
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.5211734771728516,
      "learning_rate": 1.3181794924273432e-06,
      "loss": 0.9514,
      "step": 6870
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.0055410861968994,
      "learning_rate": 1.3200982398690134e-06,
      "loss": 0.9345,
      "step": 6880
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.8020129799842834,
      "learning_rate": 1.3220169873106833e-06,
      "loss": 0.9379,
      "step": 6890
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.5040076375007629,
      "learning_rate": 1.3239357347523535e-06,
      "loss": 0.9308,
      "step": 6900
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.42365941405296326,
      "learning_rate": 1.3258544821940236e-06,
      "loss": 0.9477,
      "step": 6910
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.8183250427246094,
      "learning_rate": 1.3277732296356936e-06,
      "loss": 0.9383,
      "step": 6920
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.548681378364563,
      "learning_rate": 1.3296919770773637e-06,
      "loss": 0.9358,
      "step": 6930
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.4857513904571533,
      "learning_rate": 1.3316107245190339e-06,
      "loss": 0.9241,
      "step": 6940
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.6472055315971375,
      "learning_rate": 1.3335294719607038e-06,
      "loss": 0.9151,
      "step": 6950
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.7625853419303894,
      "learning_rate": 1.335448219402374e-06,
      "loss": 0.9356,
      "step": 6960
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.7316339612007141,
      "learning_rate": 1.3373669668440441e-06,
      "loss": 0.9476,
      "step": 6970
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.6763690710067749,
      "learning_rate": 1.339285714285714e-06,
      "loss": 0.929,
      "step": 6980
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.5594536662101746,
      "learning_rate": 1.3412044617273842e-06,
      "loss": 0.9376,
      "step": 6990
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.5051941871643066,
      "learning_rate": 1.3431232091690541e-06,
      "loss": 0.9228,
      "step": 7000
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.6497828960418701,
      "learning_rate": 1.3450419566107243e-06,
      "loss": 0.925,
      "step": 7010
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.7407678961753845,
      "learning_rate": 1.3469607040523944e-06,
      "loss": 0.9521,
      "step": 7020
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5873762369155884,
      "learning_rate": 1.3488794514940644e-06,
      "loss": 0.937,
      "step": 7030
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8801110982894897,
      "learning_rate": 1.3507981989357345e-06,
      "loss": 0.945,
      "step": 7040
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.5163893699645996,
      "learning_rate": 1.3527169463774047e-06,
      "loss": 0.9323,
      "step": 7050
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9585764408111572,
      "learning_rate": 1.3546356938190746e-06,
      "loss": 0.9416,
      "step": 7060
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.7154420614242554,
      "learning_rate": 1.3565544412607448e-06,
      "loss": 0.9364,
      "step": 7070
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.38925448060035706,
      "learning_rate": 1.358473188702415e-06,
      "loss": 0.9284,
      "step": 7080
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9571309089660645,
      "learning_rate": 1.3603919361440849e-06,
      "loss": 0.9371,
      "step": 7090
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.0068739652633667,
      "learning_rate": 1.362310683585755e-06,
      "loss": 0.923,
      "step": 7100
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.4589304029941559,
      "learning_rate": 1.364229431027425e-06,
      "loss": 0.9392,
      "step": 7110
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9613423347473145,
      "learning_rate": 1.366148178469095e-06,
      "loss": 0.9356,
      "step": 7120
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.6306473016738892,
      "learning_rate": 1.3680669259107655e-06,
      "loss": 0.9521,
      "step": 7130
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.8076304197311401,
      "learning_rate": 1.3699856733524356e-06,
      "loss": 0.9326,
      "step": 7140
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.29986846446990967,
      "learning_rate": 1.3719044207941056e-06,
      "loss": 0.9328,
      "step": 7150
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.5027610659599304,
      "learning_rate": 1.3738231682357757e-06,
      "loss": 0.9375,
      "step": 7160
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.7516102194786072,
      "learning_rate": 1.3757419156774459e-06,
      "loss": 0.9367,
      "step": 7170
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.1813656091690063,
      "learning_rate": 1.3776606631191158e-06,
      "loss": 0.9291,
      "step": 7180
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.8535031080245972,
      "learning_rate": 1.379579410560786e-06,
      "loss": 0.9257,
      "step": 7190
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4171490669250488,
      "learning_rate": 1.3814981580024559e-06,
      "loss": 0.9251,
      "step": 7200
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.607743501663208,
      "learning_rate": 1.383416905444126e-06,
      "loss": 0.9215,
      "step": 7210
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0631513595581055,
      "learning_rate": 1.3853356528857962e-06,
      "loss": 0.9292,
      "step": 7220
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.4893511235713959,
      "learning_rate": 1.3872544003274661e-06,
      "loss": 0.9246,
      "step": 7230
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.437801331281662,
      "learning_rate": 1.3891731477691363e-06,
      "loss": 0.9285,
      "step": 7240
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.8945329785346985,
      "learning_rate": 1.3910918952108064e-06,
      "loss": 0.9249,
      "step": 7250
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.5025817155838013,
      "learning_rate": 1.3930106426524764e-06,
      "loss": 0.9131,
      "step": 7260
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.4224840998649597,
      "learning_rate": 1.3949293900941465e-06,
      "loss": 0.9343,
      "step": 7270
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9892147779464722,
      "learning_rate": 1.3968481375358167e-06,
      "loss": 0.9388,
      "step": 7280
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.8519554138183594,
      "learning_rate": 1.3987668849774866e-06,
      "loss": 0.9205,
      "step": 7290
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.7914392352104187,
      "learning_rate": 1.4006856324191568e-06,
      "loss": 0.9491,
      "step": 7300
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.2933136224746704,
      "learning_rate": 1.4026043798608267e-06,
      "loss": 0.9336,
      "step": 7310
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.4553003311157227,
      "learning_rate": 1.4045231273024968e-06,
      "loss": 0.9246,
      "step": 7320
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.9304126501083374,
      "eval_runtime": 54.2837,
      "eval_samples_per_second": 63.537,
      "eval_steps_per_second": 7.958,
      "step": 7329
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.4941023588180542,
      "learning_rate": 1.406441874744167e-06,
      "loss": 0.9362,
      "step": 7330
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8567590713500977,
      "learning_rate": 1.408360622185837e-06,
      "loss": 0.9243,
      "step": 7340
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7519862055778503,
      "learning_rate": 1.410279369627507e-06,
      "loss": 0.938,
      "step": 7350
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.5692330002784729,
      "learning_rate": 1.4121981170691772e-06,
      "loss": 0.9295,
      "step": 7360
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.6891592741012573,
      "learning_rate": 1.4141168645108472e-06,
      "loss": 0.9304,
      "step": 7370
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.562261939048767,
      "learning_rate": 1.4160356119525173e-06,
      "loss": 0.9204,
      "step": 7380
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.5060169100761414,
      "learning_rate": 1.4179543593941875e-06,
      "loss": 0.9335,
      "step": 7390
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0711064338684082,
      "learning_rate": 1.4198731068358574e-06,
      "loss": 0.9205,
      "step": 7400
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.49203965067863464,
      "learning_rate": 1.4217918542775276e-06,
      "loss": 0.9324,
      "step": 7410
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6391666531562805,
      "learning_rate": 1.4237106017191975e-06,
      "loss": 0.9289,
      "step": 7420
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6151255965232849,
      "learning_rate": 1.4256293491608677e-06,
      "loss": 0.9416,
      "step": 7430
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.7915700078010559,
      "learning_rate": 1.4275480966025378e-06,
      "loss": 0.9295,
      "step": 7440
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.7198977470397949,
      "learning_rate": 1.4294668440442077e-06,
      "loss": 0.9216,
      "step": 7450
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.1339706182479858,
      "learning_rate": 1.431385591485878e-06,
      "loss": 0.9301,
      "step": 7460
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.9740000367164612,
      "learning_rate": 1.433304338927548e-06,
      "loss": 0.9224,
      "step": 7470
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.3749115467071533,
      "learning_rate": 1.435223086369218e-06,
      "loss": 0.9401,
      "step": 7480
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9076142311096191,
      "learning_rate": 1.4371418338108881e-06,
      "loss": 0.93,
      "step": 7490
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.7934820055961609,
      "learning_rate": 1.4390605812525583e-06,
      "loss": 0.9299,
      "step": 7500
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9414240121841431,
      "learning_rate": 1.4409793286942282e-06,
      "loss": 0.9323,
      "step": 7510
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.8602343797683716,
      "learning_rate": 1.4428980761358984e-06,
      "loss": 0.9228,
      "step": 7520
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.8451366424560547,
      "learning_rate": 1.4448168235775683e-06,
      "loss": 0.9371,
      "step": 7530
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.6954772472381592,
      "learning_rate": 1.4467355710192385e-06,
      "loss": 0.9263,
      "step": 7540
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.7507630586624146,
      "learning_rate": 1.4486543184609086e-06,
      "loss": 0.941,
      "step": 7550
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.8232126235961914,
      "learning_rate": 1.4505730659025786e-06,
      "loss": 0.9215,
      "step": 7560
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.6999447345733643,
      "learning_rate": 1.4524918133442487e-06,
      "loss": 0.914,
      "step": 7570
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.6535377502441406,
      "learning_rate": 1.4544105607859189e-06,
      "loss": 0.9293,
      "step": 7580
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.7268761992454529,
      "learning_rate": 1.4563293082275888e-06,
      "loss": 0.9105,
      "step": 7590
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.1322938203811646,
      "learning_rate": 1.458248055669259e-06,
      "loss": 0.9096,
      "step": 7600
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.231655240058899,
      "learning_rate": 1.460166803110929e-06,
      "loss": 0.9218,
      "step": 7610
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.9337775707244873,
      "learning_rate": 1.462085550552599e-06,
      "loss": 0.9217,
      "step": 7620
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.1947689056396484,
      "learning_rate": 1.4640042979942692e-06,
      "loss": 0.9184,
      "step": 7630
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.650382399559021,
      "learning_rate": 1.4659230454359391e-06,
      "loss": 0.9176,
      "step": 7640
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.1038070917129517,
      "learning_rate": 1.4678417928776093e-06,
      "loss": 0.9146,
      "step": 7650
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.7645888328552246,
      "learning_rate": 1.4697605403192794e-06,
      "loss": 0.9302,
      "step": 7660
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.5786889791488647,
      "learning_rate": 1.4716792877609494e-06,
      "loss": 0.9239,
      "step": 7670
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.6292659044265747,
      "learning_rate": 1.4735980352026195e-06,
      "loss": 0.9351,
      "step": 7680
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.5686979293823242,
      "learning_rate": 1.4755167826442897e-06,
      "loss": 0.9316,
      "step": 7690
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9709435701370239,
      "learning_rate": 1.4774355300859596e-06,
      "loss": 0.9151,
      "step": 7700
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.8054671883583069,
      "learning_rate": 1.4793542775276298e-06,
      "loss": 0.9168,
      "step": 7710
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.4217844009399414,
      "learning_rate": 1.4812730249693e-06,
      "loss": 0.9239,
      "step": 7720
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.6483398675918579,
      "learning_rate": 1.4831917724109699e-06,
      "loss": 0.9181,
      "step": 7730
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.8579164743423462,
      "learning_rate": 1.48511051985264e-06,
      "loss": 0.909,
      "step": 7740
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.3692634105682373,
      "learning_rate": 1.48702926729431e-06,
      "loss": 0.9071,
      "step": 7750
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.339156985282898,
      "learning_rate": 1.48894801473598e-06,
      "loss": 0.8958,
      "step": 7760
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.5062222480773926,
      "learning_rate": 1.4908667621776502e-06,
      "loss": 0.9128,
      "step": 7770
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.8722869157791138,
      "learning_rate": 1.4927855096193202e-06,
      "loss": 0.9317,
      "step": 7780
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.3963524103164673,
      "learning_rate": 1.4947042570609903e-06,
      "loss": 0.9201,
      "step": 7790
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.8112188577651978,
      "learning_rate": 1.4966230045026605e-06,
      "loss": 0.912,
      "step": 7800
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.9279875159263611,
      "learning_rate": 1.4985417519443306e-06,
      "loss": 0.9146,
      "step": 7810
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0172861814498901,
      "learning_rate": 1.5004604993860008e-06,
      "loss": 0.92,
      "step": 7820
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.6324945092201233,
      "learning_rate": 1.502379246827671e-06,
      "loss": 0.9007,
      "step": 7830
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.1090272665023804,
      "learning_rate": 1.5042979942693409e-06,
      "loss": 0.9137,
      "step": 7840
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.7935272455215454,
      "learning_rate": 1.506216741711011e-06,
      "loss": 0.9086,
      "step": 7850
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.6961172819137573,
      "learning_rate": 1.5081354891526812e-06,
      "loss": 0.9199,
      "step": 7860
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.052855134010315,
      "learning_rate": 1.5100542365943511e-06,
      "loss": 0.8969,
      "step": 7870
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.9742554426193237,
      "learning_rate": 1.5119729840360213e-06,
      "loss": 0.9103,
      "step": 7880
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.7613158822059631,
      "learning_rate": 1.5138917314776914e-06,
      "loss": 0.926,
      "step": 7890
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.1401052474975586,
      "learning_rate": 1.5158104789193614e-06,
      "loss": 0.9075,
      "step": 7900
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.6504545211791992,
      "learning_rate": 1.5177292263610315e-06,
      "loss": 0.9132,
      "step": 7910
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.9928746819496155,
      "learning_rate": 1.5196479738027015e-06,
      "loss": 0.9056,
      "step": 7920
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.7847699522972107,
      "learning_rate": 1.5215667212443716e-06,
      "loss": 0.9145,
      "step": 7930
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.6485918164253235,
      "learning_rate": 1.5234854686860418e-06,
      "loss": 0.9043,
      "step": 7940
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.9584844708442688,
      "learning_rate": 1.5254042161277117e-06,
      "loss": 0.9029,
      "step": 7950
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.7609745860099792,
      "learning_rate": 1.5273229635693818e-06,
      "loss": 0.9194,
      "step": 7960
    },
    {
      "epoch": 3.26,
      "grad_norm": 3.524991512298584,
      "learning_rate": 1.529241711011052e-06,
      "loss": 0.9123,
      "step": 7970
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.8698152899742126,
      "learning_rate": 1.531160458452722e-06,
      "loss": 0.9083,
      "step": 7980
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.9221356511116028,
      "learning_rate": 1.533079205894392e-06,
      "loss": 0.9214,
      "step": 7990
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.2344526052474976,
      "learning_rate": 1.5349979533360622e-06,
      "loss": 0.91,
      "step": 8000
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.7726818323135376,
      "learning_rate": 1.5369167007777322e-06,
      "loss": 0.9214,
      "step": 8010
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.1102505922317505,
      "learning_rate": 1.5388354482194023e-06,
      "loss": 0.9126,
      "step": 8020
    },
    {
      "epoch": 3.29,
      "grad_norm": 9.880892753601074,
      "learning_rate": 1.5407541956610723e-06,
      "loss": 0.912,
      "step": 8030
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0284404754638672,
      "learning_rate": 1.5426729431027424e-06,
      "loss": 0.9077,
      "step": 8040
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.9507349133491516,
      "learning_rate": 1.5445916905444126e-06,
      "loss": 0.9132,
      "step": 8050
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.7096238136291504,
      "learning_rate": 1.5465104379860825e-06,
      "loss": 0.9189,
      "step": 8060
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.5568134784698486,
      "learning_rate": 1.5484291854277527e-06,
      "loss": 0.9057,
      "step": 8070
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.7714537978172302,
      "learning_rate": 1.5503479328694228e-06,
      "loss": 0.9025,
      "step": 8080
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.2309315204620361,
      "learning_rate": 1.5522666803110927e-06,
      "loss": 0.9032,
      "step": 8090
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.3283387422561646,
      "learning_rate": 1.554185427752763e-06,
      "loss": 0.9014,
      "step": 8100
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.9314442276954651,
      "learning_rate": 1.556104175194433e-06,
      "loss": 0.9099,
      "step": 8110
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.291503667831421,
      "learning_rate": 1.558022922636103e-06,
      "loss": 0.9086,
      "step": 8120
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.554129958152771,
      "learning_rate": 1.5599416700777731e-06,
      "loss": 0.9014,
      "step": 8130
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.4969301223754883,
      "learning_rate": 1.561860417519443e-06,
      "loss": 0.9,
      "step": 8140
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.6042664051055908,
      "learning_rate": 1.5637791649611132e-06,
      "loss": 0.9311,
      "step": 8150
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.9990862607955933,
      "learning_rate": 1.5656979124027834e-06,
      "loss": 0.9109,
      "step": 8160
    },
    {
      "epoch": 3.34,
      "grad_norm": 2.3835573196411133,
      "learning_rate": 1.5676166598444533e-06,
      "loss": 0.9131,
      "step": 8170
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0823179483413696,
      "learning_rate": 1.5695354072861235e-06,
      "loss": 0.9052,
      "step": 8180
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.6567211747169495,
      "learning_rate": 1.5714541547277936e-06,
      "loss": 0.9141,
      "step": 8190
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.7015533447265625,
      "learning_rate": 1.5733729021694636e-06,
      "loss": 0.9061,
      "step": 8200
    },
    {
      "epoch": 3.36,
      "grad_norm": 2.0049850940704346,
      "learning_rate": 1.5752916496111337e-06,
      "loss": 0.8914,
      "step": 8210
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.5795691013336182,
      "learning_rate": 1.5772103970528039e-06,
      "loss": 0.8876,
      "step": 8220
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.9794216156005859,
      "learning_rate": 1.5791291444944738e-06,
      "loss": 0.9074,
      "step": 8230
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.127440333366394,
      "learning_rate": 1.581047891936144e-06,
      "loss": 0.9167,
      "step": 8240
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.273620843887329,
      "learning_rate": 1.5829666393778139e-06,
      "loss": 0.9072,
      "step": 8250
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.6870872378349304,
      "learning_rate": 1.584885386819484e-06,
      "loss": 0.9055,
      "step": 8260
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.2166780233383179,
      "learning_rate": 1.5868041342611542e-06,
      "loss": 0.9189,
      "step": 8270
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.621080994606018,
      "learning_rate": 1.5887228817028241e-06,
      "loss": 0.8839,
      "step": 8280
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.9377428889274597,
      "learning_rate": 1.5906416291444943e-06,
      "loss": 0.9111,
      "step": 8290
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.2558926343917847,
      "learning_rate": 1.5925603765861644e-06,
      "loss": 0.9154,
      "step": 8300
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.9110545516014099,
      "learning_rate": 1.5944791240278344e-06,
      "loss": 0.9156,
      "step": 8310
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9553611278533936,
      "learning_rate": 1.5963978714695045e-06,
      "loss": 0.9116,
      "step": 8320
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.5956575870513916,
      "learning_rate": 1.5983166189111747e-06,
      "loss": 0.9165,
      "step": 8330
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.8375097513198853,
      "learning_rate": 1.6002353663528446e-06,
      "loss": 0.8925,
      "step": 8340
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.8722730875015259,
      "learning_rate": 1.6021541137945148e-06,
      "loss": 0.9078,
      "step": 8350
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.3309659957885742,
      "learning_rate": 1.6040728612361847e-06,
      "loss": 0.9121,
      "step": 8360
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.8088542222976685,
      "learning_rate": 1.6059916086778549e-06,
      "loss": 0.8998,
      "step": 8370
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.9746458530426025,
      "learning_rate": 1.607910356119525e-06,
      "loss": 0.9074,
      "step": 8380
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.4853399991989136,
      "learning_rate": 1.609829103561195e-06,
      "loss": 0.894,
      "step": 8390
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0129954814910889,
      "learning_rate": 1.611747851002865e-06,
      "loss": 0.9068,
      "step": 8400
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.4728926420211792,
      "learning_rate": 1.6136665984445352e-06,
      "loss": 0.9043,
      "step": 8410
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.382433295249939,
      "learning_rate": 1.6155853458862052e-06,
      "loss": 0.8982,
      "step": 8420
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.9626283049583435,
      "learning_rate": 1.6175040933278753e-06,
      "loss": 0.896,
      "step": 8430
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.776641845703125,
      "learning_rate": 1.6194228407695455e-06,
      "loss": 0.8957,
      "step": 8440
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.7894164323806763,
      "learning_rate": 1.6213415882112154e-06,
      "loss": 0.889,
      "step": 8450
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.2059746980667114,
      "learning_rate": 1.6232603356528856e-06,
      "loss": 0.8955,
      "step": 8460
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.1420741081237793,
      "learning_rate": 1.6251790830945555e-06,
      "loss": 0.8985,
      "step": 8470
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.240789532661438,
      "learning_rate": 1.6270978305362257e-06,
      "loss": 0.8937,
      "step": 8480
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.0763700008392334,
      "learning_rate": 1.629016577977896e-06,
      "loss": 0.8981,
      "step": 8490
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.1322758197784424,
      "learning_rate": 1.6309353254195662e-06,
      "loss": 0.9082,
      "step": 8500
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.4919174909591675,
      "learning_rate": 1.6328540728612361e-06,
      "loss": 0.8772,
      "step": 8510
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.528772234916687,
      "learning_rate": 1.6347728203029063e-06,
      "loss": 0.9039,
      "step": 8520
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.2946752309799194,
      "learning_rate": 1.6366915677445762e-06,
      "loss": 0.917,
      "step": 8530
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.757124662399292,
      "learning_rate": 1.6386103151862464e-06,
      "loss": 0.8826,
      "step": 8540
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.6338216066360474,
      "learning_rate": 1.6405290626279165e-06,
      "loss": 0.9086,
      "step": 8550
    },
    {
      "epoch": 3.5,
      "grad_norm": 2.1975772380828857,
      "learning_rate": 1.6424478100695864e-06,
      "loss": 0.8917,
      "step": 8560
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.3774125576019287,
      "learning_rate": 1.6443665575112566e-06,
      "loss": 0.8863,
      "step": 8570
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0993012189865112,
      "learning_rate": 1.6462853049529267e-06,
      "loss": 0.9217,
      "step": 8580
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.3264656066894531,
      "learning_rate": 1.6482040523945967e-06,
      "loss": 0.918,
      "step": 8590
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.168540120124817,
      "learning_rate": 1.6501227998362668e-06,
      "loss": 0.8939,
      "step": 8600
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.010085105895996,
      "learning_rate": 1.652041547277937e-06,
      "loss": 0.8843,
      "step": 8610
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.6029551029205322,
      "learning_rate": 1.653960294719607e-06,
      "loss": 0.8861,
      "step": 8620
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.0494368076324463,
      "learning_rate": 1.655879042161277e-06,
      "loss": 0.8987,
      "step": 8630
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.8529078364372253,
      "learning_rate": 1.657797789602947e-06,
      "loss": 0.8962,
      "step": 8640
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.4450767040252686,
      "learning_rate": 1.6597165370446172e-06,
      "loss": 0.8958,
      "step": 8650
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.834185779094696,
      "learning_rate": 1.6616352844862873e-06,
      "loss": 0.9091,
      "step": 8660
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.7990911602973938,
      "learning_rate": 1.6635540319279573e-06,
      "loss": 0.8953,
      "step": 8670
    },
    {
      "epoch": 3.55,
      "grad_norm": 2.1438841819763184,
      "learning_rate": 1.6654727793696274e-06,
      "loss": 0.8987,
      "step": 8680
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.5368304252624512,
      "learning_rate": 1.6673915268112976e-06,
      "loss": 0.8981,
      "step": 8690
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1824474334716797,
      "learning_rate": 1.6693102742529675e-06,
      "loss": 0.903,
      "step": 8700
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.5258857011795044,
      "learning_rate": 1.6712290216946377e-06,
      "loss": 0.9008,
      "step": 8710
    },
    {
      "epoch": 3.57,
      "grad_norm": 2.2011425495147705,
      "learning_rate": 1.6731477691363078e-06,
      "loss": 0.8925,
      "step": 8720
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.46099767088890076,
      "learning_rate": 1.6750665165779777e-06,
      "loss": 0.9031,
      "step": 8730
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.3836901187896729,
      "learning_rate": 1.6769852640196479e-06,
      "loss": 0.8935,
      "step": 8740
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.8592503070831299,
      "learning_rate": 1.6789040114613178e-06,
      "loss": 0.9113,
      "step": 8750
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7721015214920044,
      "learning_rate": 1.680822758902988e-06,
      "loss": 0.8887,
      "step": 8760
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7770956754684448,
      "learning_rate": 1.6827415063446581e-06,
      "loss": 0.9019,
      "step": 8770
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.3234071731567383,
      "learning_rate": 1.684660253786328e-06,
      "loss": 0.899,
      "step": 8780
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.7908750772476196,
      "learning_rate": 1.6865790012279982e-06,
      "loss": 0.9055,
      "step": 8790
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.3943448066711426,
      "learning_rate": 1.6884977486696684e-06,
      "loss": 0.8847,
      "step": 8800
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.8751834630966187,
      "learning_rate": 1.6904164961113383e-06,
      "loss": 0.9079,
      "step": 8810
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.165566325187683,
      "learning_rate": 1.6923352435530085e-06,
      "loss": 0.9068,
      "step": 8820
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.021663784980774,
      "learning_rate": 1.6942539909946786e-06,
      "loss": 0.8798,
      "step": 8830
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.411052942276001,
      "learning_rate": 1.6961727384363486e-06,
      "loss": 0.887,
      "step": 8840
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.464910864830017,
      "learning_rate": 1.6980914858780187e-06,
      "loss": 0.8966,
      "step": 8850
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.3666223287582397,
      "learning_rate": 1.7000102333196886e-06,
      "loss": 0.8994,
      "step": 8860
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.1797021627426147,
      "learning_rate": 1.7019289807613588e-06,
      "loss": 0.907,
      "step": 8870
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.3983454704284668,
      "learning_rate": 1.703847728203029e-06,
      "loss": 0.8928,
      "step": 8880
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.3407589197158813,
      "learning_rate": 1.7057664756446989e-06,
      "loss": 0.9161,
      "step": 8890
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.8589609861373901,
      "learning_rate": 1.707685223086369e-06,
      "loss": 0.9039,
      "step": 8900
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.8015274405479431,
      "learning_rate": 1.7096039705280392e-06,
      "loss": 0.8956,
      "step": 8910
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.5755670070648193,
      "learning_rate": 1.7115227179697091e-06,
      "loss": 0.89,
      "step": 8920
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.253976583480835,
      "learning_rate": 1.7134414654113793e-06,
      "loss": 0.8929,
      "step": 8930
    },
    {
      "epoch": 3.66,
      "grad_norm": 2.0126864910125732,
      "learning_rate": 1.7153602128530494e-06,
      "loss": 0.9089,
      "step": 8940
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.170444130897522,
      "learning_rate": 1.7172789602947194e-06,
      "loss": 0.9071,
      "step": 8950
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.1227970123291016,
      "learning_rate": 1.7191977077363895e-06,
      "loss": 0.8983,
      "step": 8960
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.4143755435943604,
      "learning_rate": 1.7211164551780595e-06,
      "loss": 0.9004,
      "step": 8970
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.4665100574493408,
      "learning_rate": 1.7230352026197296e-06,
      "loss": 0.9105,
      "step": 8980
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.5829731225967407,
      "learning_rate": 1.7249539500613998e-06,
      "loss": 0.8948,
      "step": 8990
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.3789006471633911,
      "learning_rate": 1.7268726975030697e-06,
      "loss": 0.8923,
      "step": 9000
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.0195058584213257,
      "learning_rate": 1.7287914449447398e-06,
      "loss": 0.8806,
      "step": 9010
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.9694207906723022,
      "learning_rate": 1.73071019238641e-06,
      "loss": 0.907,
      "step": 9020
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.0620990991592407,
      "learning_rate": 1.73262893982808e-06,
      "loss": 0.9057,
      "step": 9030
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.525031328201294,
      "learning_rate": 1.73454768726975e-06,
      "loss": 0.9043,
      "step": 9040
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.0531690120697021,
      "learning_rate": 1.7364664347114202e-06,
      "loss": 0.9093,
      "step": 9050
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0925312042236328,
      "learning_rate": 1.7383851821530902e-06,
      "loss": 0.8954,
      "step": 9060
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.8937349319458008,
      "learning_rate": 1.7403039295947603e-06,
      "loss": 0.8989,
      "step": 9070
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.4082143306732178,
      "learning_rate": 1.7422226770364303e-06,
      "loss": 0.8961,
      "step": 9080
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.9719319343566895,
      "learning_rate": 1.7441414244781004e-06,
      "loss": 0.8902,
      "step": 9090
    },
    {
      "epoch": 3.72,
      "grad_norm": 2.0578103065490723,
      "learning_rate": 1.7460601719197706e-06,
      "loss": 0.9033,
      "step": 9100
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.5442516803741455,
      "learning_rate": 1.7479789193614405e-06,
      "loss": 0.8943,
      "step": 9110
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.9265874028205872,
      "learning_rate": 1.7498976668031107e-06,
      "loss": 0.8986,
      "step": 9120
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.0423319339752197,
      "learning_rate": 1.7518164142447808e-06,
      "loss": 0.8853,
      "step": 9130
    },
    {
      "epoch": 3.74,
      "grad_norm": 2.1167335510253906,
      "learning_rate": 1.7537351616864507e-06,
      "loss": 0.9126,
      "step": 9140
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.3518155813217163,
      "learning_rate": 1.755653909128121e-06,
      "loss": 0.8967,
      "step": 9150
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.171280026435852,
      "learning_rate": 1.757572656569791e-06,
      "loss": 0.8948,
      "step": 9160
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.082364559173584,
      "learning_rate": 1.7594914040114612e-06,
      "loss": 0.8959,
      "step": 9170
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.9348938465118408,
      "learning_rate": 1.7614101514531314e-06,
      "loss": 0.9012,
      "step": 9180
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.7983856201171875,
      "learning_rate": 1.7633288988948015e-06,
      "loss": 0.8875,
      "step": 9190
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.7732130289077759,
      "learning_rate": 1.7652476463364714e-06,
      "loss": 0.9057,
      "step": 9200
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.704710602760315,
      "learning_rate": 1.7671663937781416e-06,
      "loss": 0.8932,
      "step": 9210
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.4435302019119263,
      "learning_rate": 1.7690851412198117e-06,
      "loss": 0.8852,
      "step": 9220
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.4680230617523193,
      "learning_rate": 1.7710038886614817e-06,
      "loss": 0.9021,
      "step": 9230
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.7827247977256775,
      "learning_rate": 1.7729226361031518e-06,
      "loss": 0.8978,
      "step": 9240
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.9886126518249512,
      "learning_rate": 1.774841383544822e-06,
      "loss": 0.895,
      "step": 9250
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.910647451877594,
      "learning_rate": 1.776760130986492e-06,
      "loss": 0.8859,
      "step": 9260
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.2409913539886475,
      "learning_rate": 1.778678878428162e-06,
      "loss": 0.8863,
      "step": 9270
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.2593765258789062,
      "learning_rate": 1.780597625869832e-06,
      "loss": 0.8897,
      "step": 9280
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.5780937671661377,
      "learning_rate": 1.7825163733115022e-06,
      "loss": 0.8936,
      "step": 9290
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.1310447454452515,
      "learning_rate": 1.7844351207531723e-06,
      "loss": 0.9025,
      "step": 9300
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.2277438640594482,
      "learning_rate": 1.7863538681948423e-06,
      "loss": 0.9021,
      "step": 9310
    },
    {
      "epoch": 3.81,
      "grad_norm": 2.2154178619384766,
      "learning_rate": 1.7882726156365124e-06,
      "loss": 0.8869,
      "step": 9320
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.9902834296226501,
      "learning_rate": 1.7901913630781826e-06,
      "loss": 0.8934,
      "step": 9330
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.4567925930023193,
      "learning_rate": 1.7921101105198525e-06,
      "loss": 0.8863,
      "step": 9340
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.4804329872131348,
      "learning_rate": 1.7940288579615226e-06,
      "loss": 0.8923,
      "step": 9350
    },
    {
      "epoch": 3.83,
      "grad_norm": 2.1118886470794678,
      "learning_rate": 1.7959476054031928e-06,
      "loss": 0.8982,
      "step": 9360
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.7590174674987793,
      "learning_rate": 1.7978663528448627e-06,
      "loss": 0.8965,
      "step": 9370
    },
    {
      "epoch": 3.84,
      "grad_norm": 2.159075975418091,
      "learning_rate": 1.7997851002865329e-06,
      "loss": 0.8894,
      "step": 9380
    },
    {
      "epoch": 3.84,
      "grad_norm": 2.7859482765197754,
      "learning_rate": 1.8017038477282028e-06,
      "loss": 0.8845,
      "step": 9390
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.2064670324325562,
      "learning_rate": 1.803622595169873e-06,
      "loss": 0.8984,
      "step": 9400
    },
    {
      "epoch": 3.85,
      "grad_norm": 2.399698257446289,
      "learning_rate": 1.8055413426115431e-06,
      "loss": 0.8957,
      "step": 9410
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.7884770631790161,
      "learning_rate": 1.807460090053213e-06,
      "loss": 0.8866,
      "step": 9420
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.3903491497039795,
      "learning_rate": 1.8093788374948832e-06,
      "loss": 0.9028,
      "step": 9430
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.3823890686035156,
      "learning_rate": 1.8112975849365534e-06,
      "loss": 0.8921,
      "step": 9440
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.8637466430664062,
      "learning_rate": 1.8132163323782233e-06,
      "loss": 0.9013,
      "step": 9450
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.1230412721633911,
      "learning_rate": 1.8151350798198935e-06,
      "loss": 0.8845,
      "step": 9460
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.2249157428741455,
      "learning_rate": 1.8170538272615636e-06,
      "loss": 0.905,
      "step": 9470
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.9346128702163696,
      "learning_rate": 1.8189725747032335e-06,
      "loss": 0.8725,
      "step": 9480
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.8348875045776367,
      "learning_rate": 1.8208913221449037e-06,
      "loss": 0.8896,
      "step": 9490
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.232509970664978,
      "learning_rate": 1.8228100695865736e-06,
      "loss": 0.9032,
      "step": 9500
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.8332470655441284,
      "learning_rate": 1.8247288170282438e-06,
      "loss": 0.89,
      "step": 9510
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.6756689548492432,
      "learning_rate": 1.826647564469914e-06,
      "loss": 0.8631,
      "step": 9520
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.5219858884811401,
      "learning_rate": 1.8285663119115839e-06,
      "loss": 0.8991,
      "step": 9530
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.3194503784179688,
      "learning_rate": 1.830485059353254e-06,
      "loss": 0.8839,
      "step": 9540
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.8980822563171387,
      "learning_rate": 1.8324038067949242e-06,
      "loss": 0.9242,
      "step": 9550
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.5661436319351196,
      "learning_rate": 1.8343225542365941e-06,
      "loss": 0.8881,
      "step": 9560
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.1143420934677124,
      "learning_rate": 1.8362413016782643e-06,
      "loss": 0.8936,
      "step": 9570
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.5260075330734253,
      "learning_rate": 1.8381600491199342e-06,
      "loss": 0.8881,
      "step": 9580
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.2940798997879028,
      "learning_rate": 1.8400787965616044e-06,
      "loss": 0.8971,
      "step": 9590
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.160205364227295,
      "learning_rate": 1.8419975440032745e-06,
      "loss": 0.8726,
      "step": 9600
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.7810388803482056,
      "learning_rate": 1.8439162914449445e-06,
      "loss": 0.8799,
      "step": 9610
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.9253838062286377,
      "learning_rate": 1.8458350388866146e-06,
      "loss": 0.8794,
      "step": 9620
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.8170150518417358,
      "learning_rate": 1.8477537863282848e-06,
      "loss": 0.8804,
      "step": 9630
    },
    {
      "epoch": 3.95,
      "grad_norm": 2.1621105670928955,
      "learning_rate": 1.8496725337699547e-06,
      "loss": 0.8788,
      "step": 9640
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.0520083904266357,
      "learning_rate": 1.8515912812116248e-06,
      "loss": 0.8804,
      "step": 9650
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.2904253005981445,
      "learning_rate": 1.853510028653295e-06,
      "loss": 0.8954,
      "step": 9660
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.8181043863296509,
      "learning_rate": 1.855428776094965e-06,
      "loss": 0.8824,
      "step": 9670
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.2460153102874756,
      "learning_rate": 1.857347523536635e-06,
      "loss": 0.8927,
      "step": 9680
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.235089659690857,
      "learning_rate": 1.859266270978305e-06,
      "loss": 0.8837,
      "step": 9690
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.6758712530136108,
      "learning_rate": 1.8611850184199752e-06,
      "loss": 0.8892,
      "step": 9700
    },
    {
      "epoch": 3.97,
      "grad_norm": 2.238931655883789,
      "learning_rate": 1.8631037658616453e-06,
      "loss": 0.9066,
      "step": 9710
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.7030560970306396,
      "learning_rate": 1.8650225133033153e-06,
      "loss": 0.8593,
      "step": 9720
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.4164009094238281,
      "learning_rate": 1.8669412607449854e-06,
      "loss": 0.8817,
      "step": 9730
    },
    {
      "epoch": 3.99,
      "grad_norm": 2.1455655097961426,
      "learning_rate": 1.8688600081866556e-06,
      "loss": 0.8766,
      "step": 9740
    },
    {
      "epoch": 3.99,
      "grad_norm": 2.7223548889160156,
      "learning_rate": 1.8707787556283255e-06,
      "loss": 0.9026,
      "step": 9750
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.2641315460205078,
      "learning_rate": 1.8726975030699957e-06,
      "loss": 0.892,
      "step": 9760
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.652611255645752,
      "learning_rate": 1.8746162505116658e-06,
      "loss": 0.8979,
      "step": 9770
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.8854859471321106,
      "eval_runtime": 52.3041,
      "eval_samples_per_second": 65.941,
      "eval_steps_per_second": 8.259,
      "step": 9772
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.0300142765045166,
      "learning_rate": 1.8765349979533357e-06,
      "loss": 0.8953,
      "step": 9780
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.8501081466674805,
      "learning_rate": 1.878453745395006e-06,
      "loss": 0.8762,
      "step": 9790
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.150575876235962,
      "learning_rate": 1.8803724928366758e-06,
      "loss": 0.8714,
      "step": 9800
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.455268144607544,
      "learning_rate": 1.882291240278346e-06,
      "loss": 0.8733,
      "step": 9810
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.3499019145965576,
      "learning_rate": 1.8842099877200161e-06,
      "loss": 0.8865,
      "step": 9820
    },
    {
      "epoch": 4.02,
      "grad_norm": 2.1394078731536865,
      "learning_rate": 1.886128735161686e-06,
      "loss": 0.8943,
      "step": 9830
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.3077867031097412,
      "learning_rate": 1.8880474826033564e-06,
      "loss": 0.8947,
      "step": 9840
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.426698088645935,
      "learning_rate": 1.8899662300450266e-06,
      "loss": 0.9022,
      "step": 9850
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.9866126775741577,
      "learning_rate": 1.8918849774866967e-06,
      "loss": 0.8901,
      "step": 9860
    },
    {
      "epoch": 4.04,
      "grad_norm": 2.0486161708831787,
      "learning_rate": 1.8938037249283667e-06,
      "loss": 0.8909,
      "step": 9870
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.9288359880447388,
      "learning_rate": 1.8957224723700368e-06,
      "loss": 0.8774,
      "step": 9880
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.5774805545806885,
      "learning_rate": 1.8976412198117068e-06,
      "loss": 0.877,
      "step": 9890
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.119106411933899,
      "learning_rate": 1.899559967253377e-06,
      "loss": 0.8986,
      "step": 9900
    },
    {
      "epoch": 4.06,
      "grad_norm": 2.438201427459717,
      "learning_rate": 1.901478714695047e-06,
      "loss": 0.8636,
      "step": 9910
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.2229620218276978,
      "learning_rate": 1.903397462136717e-06,
      "loss": 0.8877,
      "step": 9920
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.7282814979553223,
      "learning_rate": 1.9053162095783872e-06,
      "loss": 0.8454,
      "step": 9930
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.4656469821929932,
      "learning_rate": 1.9072349570200573e-06,
      "loss": 0.8708,
      "step": 9940
    },
    {
      "epoch": 4.07,
      "grad_norm": 2.4168996810913086,
      "learning_rate": 1.9091537044617275e-06,
      "loss": 0.8648,
      "step": 9950
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.7174862623214722,
      "learning_rate": 1.9110724519033974e-06,
      "loss": 0.8748,
      "step": 9960
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.5010780096054077,
      "learning_rate": 1.9129911993450673e-06,
      "loss": 0.8881,
      "step": 9970
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.4121907949447632,
      "learning_rate": 1.9149099467867377e-06,
      "loss": 0.8974,
      "step": 9980
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.6281142234802246,
      "learning_rate": 1.9168286942284076e-06,
      "loss": 0.8898,
      "step": 9990
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.0111535787582397,
      "learning_rate": 1.9187474416700776e-06,
      "loss": 0.8917,
      "step": 10000
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.9711953401565552,
      "learning_rate": 1.920666189111748e-06,
      "loss": 0.8659,
      "step": 10010
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.7657796144485474,
      "learning_rate": 1.922584936553418e-06,
      "loss": 0.8851,
      "step": 10020
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.4116597175598145,
      "learning_rate": 1.924503683995088e-06,
      "loss": 0.8732,
      "step": 10030
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.9057385921478271,
      "learning_rate": 1.9264224314367578e-06,
      "loss": 0.8847,
      "step": 10040
    },
    {
      "epoch": 4.11,
      "grad_norm": 3.7514235973358154,
      "learning_rate": 1.928341178878428e-06,
      "loss": 0.8399,
      "step": 10050
    },
    {
      "epoch": 4.12,
      "grad_norm": 2.1740989685058594,
      "learning_rate": 1.930259926320098e-06,
      "loss": 0.8743,
      "step": 10060
    },
    {
      "epoch": 4.12,
      "grad_norm": 3.4817705154418945,
      "learning_rate": 1.932178673761768e-06,
      "loss": 0.866,
      "step": 10070
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.5491514205932617,
      "learning_rate": 1.9340974212034384e-06,
      "loss": 0.8709,
      "step": 10080
    },
    {
      "epoch": 4.13,
      "grad_norm": 2.1954100131988525,
      "learning_rate": 1.9360161686451083e-06,
      "loss": 0.867,
      "step": 10090
    },
    {
      "epoch": 4.13,
      "grad_norm": 2.855323314666748,
      "learning_rate": 1.9379349160867782e-06,
      "loss": 0.8822,
      "step": 10100
    },
    {
      "epoch": 4.14,
      "grad_norm": 1.7502684593200684,
      "learning_rate": 1.9398536635284486e-06,
      "loss": 0.8651,
      "step": 10110
    },
    {
      "epoch": 4.14,
      "grad_norm": 1.2171279191970825,
      "learning_rate": 1.9417724109701185e-06,
      "loss": 0.8697,
      "step": 10120
    },
    {
      "epoch": 4.15,
      "grad_norm": 2.0511090755462646,
      "learning_rate": 1.9436911584117885e-06,
      "loss": 0.8508,
      "step": 10130
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.9104670286178589,
      "learning_rate": 1.945609905853459e-06,
      "loss": 0.8441,
      "step": 10140
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.392409086227417,
      "learning_rate": 1.9475286532951288e-06,
      "loss": 0.8484,
      "step": 10150
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.417162299156189,
      "learning_rate": 1.9494474007367987e-06,
      "loss": 0.8496,
      "step": 10160
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.999629259109497,
      "learning_rate": 1.951366148178469e-06,
      "loss": 0.853,
      "step": 10170
    },
    {
      "epoch": 4.17,
      "grad_norm": 3.4665515422821045,
      "learning_rate": 1.953284895620139e-06,
      "loss": 0.8601,
      "step": 10180
    },
    {
      "epoch": 4.17,
      "grad_norm": 2.441781520843506,
      "learning_rate": 1.955203643061809e-06,
      "loss": 0.8579,
      "step": 10190
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.6167982816696167,
      "learning_rate": 1.9571223905034793e-06,
      "loss": 0.8509,
      "step": 10200
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.7828446626663208,
      "learning_rate": 1.9590411379451493e-06,
      "loss": 0.861,
      "step": 10210
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.6556615829467773,
      "learning_rate": 1.960959885386819e-06,
      "loss": 0.8505,
      "step": 10220
    },
    {
      "epoch": 4.19,
      "grad_norm": 3.093745231628418,
      "learning_rate": 1.9628786328284896e-06,
      "loss": 0.8417,
      "step": 10230
    },
    {
      "epoch": 4.19,
      "grad_norm": 2.690294027328491,
      "learning_rate": 1.9647973802701595e-06,
      "loss": 0.8342,
      "step": 10240
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.7136330604553223,
      "learning_rate": 1.9667161277118294e-06,
      "loss": 0.829,
      "step": 10250
    },
    {
      "epoch": 4.2,
      "grad_norm": 2.1278891563415527,
      "learning_rate": 1.9686348751534994e-06,
      "loss": 0.8532,
      "step": 10260
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.7383137941360474,
      "learning_rate": 1.9705536225951697e-06,
      "loss": 0.8281,
      "step": 10270
    },
    {
      "epoch": 4.21,
      "grad_norm": 2.6986732482910156,
      "learning_rate": 1.9724723700368397e-06,
      "loss": 0.846,
      "step": 10280
    },
    {
      "epoch": 4.21,
      "grad_norm": 1.9539088010787964,
      "learning_rate": 1.9743911174785096e-06,
      "loss": 0.8672,
      "step": 10290
    },
    {
      "epoch": 4.22,
      "grad_norm": 1.7775877714157104,
      "learning_rate": 1.97630986492018e-06,
      "loss": 0.8508,
      "step": 10300
    },
    {
      "epoch": 4.22,
      "grad_norm": 3.0674169063568115,
      "learning_rate": 1.97822861236185e-06,
      "loss": 0.8664,
      "step": 10310
    },
    {
      "epoch": 4.22,
      "grad_norm": 1.4642376899719238,
      "learning_rate": 1.98014735980352e-06,
      "loss": 0.8568,
      "step": 10320
    },
    {
      "epoch": 4.23,
      "grad_norm": 2.531977415084839,
      "learning_rate": 1.9820661072451902e-06,
      "loss": 0.8675,
      "step": 10330
    },
    {
      "epoch": 4.23,
      "grad_norm": 4.868284225463867,
      "learning_rate": 1.98398485468686e-06,
      "loss": 0.8512,
      "step": 10340
    },
    {
      "epoch": 4.24,
      "grad_norm": 1.631285309791565,
      "learning_rate": 1.98590360212853e-06,
      "loss": 0.8631,
      "step": 10350
    },
    {
      "epoch": 4.24,
      "grad_norm": 2.7865867614746094,
      "learning_rate": 1.9878223495702005e-06,
      "loss": 0.8511,
      "step": 10360
    },
    {
      "epoch": 4.24,
      "grad_norm": 3.8086905479431152,
      "learning_rate": 1.9897410970118704e-06,
      "loss": 0.8278,
      "step": 10370
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.9171888828277588,
      "learning_rate": 1.9916598444535403e-06,
      "loss": 0.8461,
      "step": 10380
    },
    {
      "epoch": 4.25,
      "grad_norm": 2.313276529312134,
      "learning_rate": 1.9935785918952107e-06,
      "loss": 0.8361,
      "step": 10390
    },
    {
      "epoch": 4.26,
      "grad_norm": 3.1053855419158936,
      "learning_rate": 1.9954973393368807e-06,
      "loss": 0.8399,
      "step": 10400
    },
    {
      "epoch": 4.26,
      "grad_norm": 1.6656945943832397,
      "learning_rate": 1.9974160867785506e-06,
      "loss": 0.8472,
      "step": 10410
    },
    {
      "epoch": 4.27,
      "grad_norm": 2.940582513809204,
      "learning_rate": 1.999334834220221e-06,
      "loss": 0.8521,
      "step": 10420
    },
    {
      "epoch": 4.27,
      "grad_norm": 1.3946442604064941,
      "learning_rate": 2.001253581661891e-06,
      "loss": 0.8372,
      "step": 10430
    },
    {
      "epoch": 4.27,
      "grad_norm": 3.2168376445770264,
      "learning_rate": 2.003172329103561e-06,
      "loss": 0.8459,
      "step": 10440
    },
    {
      "epoch": 4.28,
      "grad_norm": 2.1453802585601807,
      "learning_rate": 2.005091076545231e-06,
      "loss": 0.8408,
      "step": 10450
    },
    {
      "epoch": 4.28,
      "grad_norm": 4.674014568328857,
      "learning_rate": 2.007009823986901e-06,
      "loss": 0.8352,
      "step": 10460
    },
    {
      "epoch": 4.29,
      "grad_norm": 1.4227583408355713,
      "learning_rate": 2.008928571428571e-06,
      "loss": 0.8439,
      "step": 10470
    },
    {
      "epoch": 4.29,
      "grad_norm": 2.2779273986816406,
      "learning_rate": 2.010847318870241e-06,
      "loss": 0.8484,
      "step": 10480
    },
    {
      "epoch": 4.29,
      "grad_norm": 2.2230465412139893,
      "learning_rate": 2.0127660663119114e-06,
      "loss": 0.8518,
      "step": 10490
    },
    {
      "epoch": 4.3,
      "grad_norm": 3.28336238861084,
      "learning_rate": 2.0146848137535813e-06,
      "loss": 0.8317,
      "step": 10500
    },
    {
      "epoch": 4.3,
      "grad_norm": 2.1029462814331055,
      "learning_rate": 2.0166035611952513e-06,
      "loss": 0.8282,
      "step": 10510
    },
    {
      "epoch": 4.31,
      "grad_norm": 2.8194525241851807,
      "learning_rate": 2.0185223086369216e-06,
      "loss": 0.8395,
      "step": 10520
    },
    {
      "epoch": 4.31,
      "grad_norm": 2.8839609622955322,
      "learning_rate": 2.020441056078592e-06,
      "loss": 0.8391,
      "step": 10530
    },
    {
      "epoch": 4.31,
      "grad_norm": 1.8259915113449097,
      "learning_rate": 2.022359803520262e-06,
      "loss": 0.8519,
      "step": 10540
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.0376909971237183,
      "learning_rate": 2.024278550961932e-06,
      "loss": 0.8542,
      "step": 10550
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.7315489053726196,
      "learning_rate": 2.0261972984036022e-06,
      "loss": 0.8289,
      "step": 10560
    },
    {
      "epoch": 4.33,
      "grad_norm": 1.8692108392715454,
      "learning_rate": 2.028116045845272e-06,
      "loss": 0.8441,
      "step": 10570
    },
    {
      "epoch": 4.33,
      "grad_norm": 3.440800666809082,
      "learning_rate": 2.030034793286942e-06,
      "loss": 0.8396,
      "step": 10580
    },
    {
      "epoch": 4.33,
      "grad_norm": 2.5794034004211426,
      "learning_rate": 2.0319535407286125e-06,
      "loss": 0.8367,
      "step": 10590
    },
    {
      "epoch": 4.34,
      "grad_norm": 2.8944766521453857,
      "learning_rate": 2.0338722881702824e-06,
      "loss": 0.8378,
      "step": 10600
    },
    {
      "epoch": 4.34,
      "grad_norm": 2.239302635192871,
      "learning_rate": 2.0357910356119523e-06,
      "loss": 0.8202,
      "step": 10610
    },
    {
      "epoch": 4.35,
      "grad_norm": 1.5158023834228516,
      "learning_rate": 2.0377097830536227e-06,
      "loss": 0.8258,
      "step": 10620
    },
    {
      "epoch": 4.35,
      "grad_norm": 2.1082098484039307,
      "learning_rate": 2.0396285304952926e-06,
      "loss": 0.8304,
      "step": 10630
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.7366771697998047,
      "learning_rate": 2.0415472779369626e-06,
      "loss": 0.8302,
      "step": 10640
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.9140199422836304,
      "learning_rate": 2.043466025378633e-06,
      "loss": 0.8377,
      "step": 10650
    },
    {
      "epoch": 4.36,
      "grad_norm": 4.046729564666748,
      "learning_rate": 2.045384772820303e-06,
      "loss": 0.8214,
      "step": 10660
    },
    {
      "epoch": 4.37,
      "grad_norm": 3.5839829444885254,
      "learning_rate": 2.047303520261973e-06,
      "loss": 0.8557,
      "step": 10670
    },
    {
      "epoch": 4.37,
      "grad_norm": 4.174716949462891,
      "learning_rate": 2.0492222677036428e-06,
      "loss": 0.8468,
      "step": 10680
    },
    {
      "epoch": 4.38,
      "grad_norm": 1.4674769639968872,
      "learning_rate": 2.051141015145313e-06,
      "loss": 0.8302,
      "step": 10690
    },
    {
      "epoch": 4.38,
      "grad_norm": 2.5353543758392334,
      "learning_rate": 2.053059762586983e-06,
      "loss": 0.8405,
      "step": 10700
    },
    {
      "epoch": 4.38,
      "grad_norm": 3.1172847747802734,
      "learning_rate": 2.054978510028653e-06,
      "loss": 0.8439,
      "step": 10710
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.9490360021591187,
      "learning_rate": 2.0568972574703234e-06,
      "loss": 0.8428,
      "step": 10720
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.5649007558822632,
      "learning_rate": 2.0588160049119933e-06,
      "loss": 0.8262,
      "step": 10730
    },
    {
      "epoch": 4.4,
      "grad_norm": 2.984516143798828,
      "learning_rate": 2.0607347523536632e-06,
      "loss": 0.8343,
      "step": 10740
    },
    {
      "epoch": 4.4,
      "grad_norm": 3.6094202995300293,
      "learning_rate": 2.0626534997953336e-06,
      "loss": 0.8378,
      "step": 10750
    },
    {
      "epoch": 4.4,
      "grad_norm": 2.1126303672790527,
      "learning_rate": 2.0645722472370035e-06,
      "loss": 0.8336,
      "step": 10760
    },
    {
      "epoch": 4.41,
      "grad_norm": 2.615879535675049,
      "learning_rate": 2.0664909946786735e-06,
      "loss": 0.8474,
      "step": 10770
    },
    {
      "epoch": 4.41,
      "grad_norm": 4.030284404754639,
      "learning_rate": 2.068409742120344e-06,
      "loss": 0.8244,
      "step": 10780
    },
    {
      "epoch": 4.42,
      "grad_norm": 2.7683985233306885,
      "learning_rate": 2.0703284895620138e-06,
      "loss": 0.825,
      "step": 10790
    },
    {
      "epoch": 4.42,
      "grad_norm": 2.6300718784332275,
      "learning_rate": 2.0722472370036837e-06,
      "loss": 0.8552,
      "step": 10800
    },
    {
      "epoch": 4.42,
      "grad_norm": 1.2624963521957397,
      "learning_rate": 2.074165984445354e-06,
      "loss": 0.8468,
      "step": 10810
    },
    {
      "epoch": 4.43,
      "grad_norm": 2.361304998397827,
      "learning_rate": 2.076084731887024e-06,
      "loss": 0.8251,
      "step": 10820
    },
    {
      "epoch": 4.43,
      "grad_norm": 2.5215606689453125,
      "learning_rate": 2.078003479328694e-06,
      "loss": 0.8274,
      "step": 10830
    },
    {
      "epoch": 4.44,
      "grad_norm": 2.048161506652832,
      "learning_rate": 2.0799222267703643e-06,
      "loss": 0.8399,
      "step": 10840
    },
    {
      "epoch": 4.44,
      "grad_norm": 2.029184579849243,
      "learning_rate": 2.0818409742120343e-06,
      "loss": 0.8223,
      "step": 10850
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.6783809661865234,
      "learning_rate": 2.083759721653704e-06,
      "loss": 0.8455,
      "step": 10860
    },
    {
      "epoch": 4.45,
      "grad_norm": 2.226811647415161,
      "learning_rate": 2.085678469095374e-06,
      "loss": 0.8428,
      "step": 10870
    },
    {
      "epoch": 4.45,
      "grad_norm": 2.152172803878784,
      "learning_rate": 2.0875972165370445e-06,
      "loss": 0.8422,
      "step": 10880
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.5861471891403198,
      "learning_rate": 2.0895159639787144e-06,
      "loss": 0.8497,
      "step": 10890
    },
    {
      "epoch": 4.46,
      "grad_norm": 2.5695559978485107,
      "learning_rate": 2.0914347114203844e-06,
      "loss": 0.8341,
      "step": 10900
    },
    {
      "epoch": 4.47,
      "grad_norm": 2.5512049198150635,
      "learning_rate": 2.0933534588620547e-06,
      "loss": 0.8347,
      "step": 10910
    },
    {
      "epoch": 4.47,
      "grad_norm": 2.70626163482666,
      "learning_rate": 2.0952722063037247e-06,
      "loss": 0.8297,
      "step": 10920
    },
    {
      "epoch": 4.47,
      "grad_norm": 3.1733784675598145,
      "learning_rate": 2.0971909537453946e-06,
      "loss": 0.8053,
      "step": 10930
    },
    {
      "epoch": 4.48,
      "grad_norm": 2.360039472579956,
      "learning_rate": 2.099109701187065e-06,
      "loss": 0.8416,
      "step": 10940
    },
    {
      "epoch": 4.48,
      "grad_norm": 7.980179786682129,
      "learning_rate": 2.101028448628735e-06,
      "loss": 0.8192,
      "step": 10950
    },
    {
      "epoch": 4.49,
      "grad_norm": 3.570753335952759,
      "learning_rate": 2.102947196070405e-06,
      "loss": 0.8195,
      "step": 10960
    },
    {
      "epoch": 4.49,
      "grad_norm": 1.9135212898254395,
      "learning_rate": 2.1048659435120752e-06,
      "loss": 0.8231,
      "step": 10970
    },
    {
      "epoch": 4.49,
      "grad_norm": 3.3701584339141846,
      "learning_rate": 2.106784690953745e-06,
      "loss": 0.8165,
      "step": 10980
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.9031100273132324,
      "learning_rate": 2.108703438395415e-06,
      "loss": 0.8372,
      "step": 10990
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.691236138343811,
      "learning_rate": 2.1106221858370855e-06,
      "loss": 0.8666,
      "step": 11000
    },
    {
      "epoch": 4.51,
      "grad_norm": 3.1010076999664307,
      "learning_rate": 2.1125409332787554e-06,
      "loss": 0.8348,
      "step": 11010
    },
    {
      "epoch": 4.51,
      "grad_norm": 2.916713237762451,
      "learning_rate": 2.1144596807204253e-06,
      "loss": 0.8121,
      "step": 11020
    },
    {
      "epoch": 4.51,
      "grad_norm": 2.4510338306427,
      "learning_rate": 2.1163784281620957e-06,
      "loss": 0.8425,
      "step": 11030
    },
    {
      "epoch": 4.52,
      "grad_norm": 2.4495058059692383,
      "learning_rate": 2.1182971756037656e-06,
      "loss": 0.8223,
      "step": 11040
    },
    {
      "epoch": 4.52,
      "grad_norm": 3.1752285957336426,
      "learning_rate": 2.1202159230454356e-06,
      "loss": 0.8162,
      "step": 11050
    },
    {
      "epoch": 4.53,
      "grad_norm": 2.258697509765625,
      "learning_rate": 2.122134670487106e-06,
      "loss": 0.8271,
      "step": 11060
    },
    {
      "epoch": 4.53,
      "grad_norm": 1.2882459163665771,
      "learning_rate": 2.124053417928776e-06,
      "loss": 0.8187,
      "step": 11070
    },
    {
      "epoch": 4.54,
      "grad_norm": 2.2767605781555176,
      "learning_rate": 2.125972165370446e-06,
      "loss": 0.8117,
      "step": 11080
    },
    {
      "epoch": 4.54,
      "grad_norm": 2.038740396499634,
      "learning_rate": 2.1278909128121158e-06,
      "loss": 0.8191,
      "step": 11090
    },
    {
      "epoch": 4.54,
      "grad_norm": 1.094367265701294,
      "learning_rate": 2.129809660253786e-06,
      "loss": 0.8307,
      "step": 11100
    },
    {
      "epoch": 4.55,
      "grad_norm": 2.0124831199645996,
      "learning_rate": 2.131728407695456e-06,
      "loss": 0.8279,
      "step": 11110
    },
    {
      "epoch": 4.55,
      "grad_norm": 2.476423978805542,
      "learning_rate": 2.133647155137126e-06,
      "loss": 0.8194,
      "step": 11120
    },
    {
      "epoch": 4.56,
      "grad_norm": 2.415168523788452,
      "learning_rate": 2.1355659025787964e-06,
      "loss": 0.8342,
      "step": 11130
    },
    {
      "epoch": 4.56,
      "grad_norm": 3.2764835357666016,
      "learning_rate": 2.1374846500204663e-06,
      "loss": 0.8279,
      "step": 11140
    },
    {
      "epoch": 4.56,
      "grad_norm": 1.8996762037277222,
      "learning_rate": 2.1394033974621362e-06,
      "loss": 0.8268,
      "step": 11150
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.549127221107483,
      "learning_rate": 2.1413221449038066e-06,
      "loss": 0.8104,
      "step": 11160
    },
    {
      "epoch": 4.57,
      "grad_norm": 2.846280813217163,
      "learning_rate": 2.1432408923454765e-06,
      "loss": 0.8302,
      "step": 11170
    },
    {
      "epoch": 4.58,
      "grad_norm": 3.0073843002319336,
      "learning_rate": 2.1451596397871465e-06,
      "loss": 0.8355,
      "step": 11180
    },
    {
      "epoch": 4.58,
      "grad_norm": 2.107754707336426,
      "learning_rate": 2.147078387228817e-06,
      "loss": 0.8125,
      "step": 11190
    },
    {
      "epoch": 4.58,
      "grad_norm": 1.936684012413025,
      "learning_rate": 2.1489971346704872e-06,
      "loss": 0.8325,
      "step": 11200
    },
    {
      "epoch": 4.59,
      "grad_norm": 2.1821322441101074,
      "learning_rate": 2.150915882112157e-06,
      "loss": 0.8187,
      "step": 11210
    },
    {
      "epoch": 4.59,
      "grad_norm": 2.136704683303833,
      "learning_rate": 2.152834629553827e-06,
      "loss": 0.8098,
      "step": 11220
    },
    {
      "epoch": 4.6,
      "grad_norm": 3.026359796524048,
      "learning_rate": 2.1547533769954975e-06,
      "loss": 0.8136,
      "step": 11230
    },
    {
      "epoch": 4.6,
      "grad_norm": 2.588404655456543,
      "learning_rate": 2.1566721244371674e-06,
      "loss": 0.8255,
      "step": 11240
    },
    {
      "epoch": 4.6,
      "grad_norm": 2.722329616546631,
      "learning_rate": 2.1585908718788373e-06,
      "loss": 0.8177,
      "step": 11250
    },
    {
      "epoch": 4.61,
      "grad_norm": 2.9263763427734375,
      "learning_rate": 2.1605096193205077e-06,
      "loss": 0.8278,
      "step": 11260
    },
    {
      "epoch": 4.61,
      "grad_norm": 1.9063698053359985,
      "learning_rate": 2.1624283667621776e-06,
      "loss": 0.8165,
      "step": 11270
    },
    {
      "epoch": 4.62,
      "grad_norm": 1.4650136232376099,
      "learning_rate": 2.1643471142038476e-06,
      "loss": 0.8325,
      "step": 11280
    },
    {
      "epoch": 4.62,
      "grad_norm": 1.5212070941925049,
      "learning_rate": 2.1662658616455175e-06,
      "loss": 0.8298,
      "step": 11290
    },
    {
      "epoch": 4.63,
      "grad_norm": 2.288539409637451,
      "learning_rate": 2.168184609087188e-06,
      "loss": 0.8351,
      "step": 11300
    },
    {
      "epoch": 4.63,
      "grad_norm": 2.533893585205078,
      "learning_rate": 2.170103356528858e-06,
      "loss": 0.8118,
      "step": 11310
    },
    {
      "epoch": 4.63,
      "grad_norm": 2.846958637237549,
      "learning_rate": 2.1720221039705278e-06,
      "loss": 0.8157,
      "step": 11320
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.9751179218292236,
      "learning_rate": 2.173940851412198e-06,
      "loss": 0.8156,
      "step": 11330
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.5782331228256226,
      "learning_rate": 2.175859598853868e-06,
      "loss": 0.8193,
      "step": 11340
    },
    {
      "epoch": 4.65,
      "grad_norm": 2.449537515640259,
      "learning_rate": 2.177778346295538e-06,
      "loss": 0.8266,
      "step": 11350
    },
    {
      "epoch": 4.65,
      "grad_norm": 3.4539830684661865,
      "learning_rate": 2.1796970937372084e-06,
      "loss": 0.8026,
      "step": 11360
    },
    {
      "epoch": 4.65,
      "grad_norm": 3.084021806716919,
      "learning_rate": 2.1816158411788783e-06,
      "loss": 0.8185,
      "step": 11370
    },
    {
      "epoch": 4.66,
      "grad_norm": 1.6091853380203247,
      "learning_rate": 2.1835345886205482e-06,
      "loss": 0.8478,
      "step": 11380
    },
    {
      "epoch": 4.66,
      "grad_norm": 1.7991818189620972,
      "learning_rate": 2.1854533360622186e-06,
      "loss": 0.8285,
      "step": 11390
    },
    {
      "epoch": 4.67,
      "grad_norm": 1.761474370956421,
      "learning_rate": 2.1873720835038885e-06,
      "loss": 0.8077,
      "step": 11400
    },
    {
      "epoch": 4.67,
      "grad_norm": 2.9270784854888916,
      "learning_rate": 2.1892908309455585e-06,
      "loss": 0.841,
      "step": 11410
    },
    {
      "epoch": 4.67,
      "grad_norm": 2.5215795040130615,
      "learning_rate": 2.191209578387229e-06,
      "loss": 0.8319,
      "step": 11420
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.6843537092208862,
      "learning_rate": 2.1931283258288988e-06,
      "loss": 0.8424,
      "step": 11430
    },
    {
      "epoch": 4.68,
      "grad_norm": 2.4913716316223145,
      "learning_rate": 2.1950470732705687e-06,
      "loss": 0.8125,
      "step": 11440
    },
    {
      "epoch": 4.69,
      "grad_norm": 2.8558645248413086,
      "learning_rate": 2.196965820712239e-06,
      "loss": 0.8246,
      "step": 11450
    },
    {
      "epoch": 4.69,
      "grad_norm": 2.5223891735076904,
      "learning_rate": 2.198884568153909e-06,
      "loss": 0.8175,
      "step": 11460
    },
    {
      "epoch": 4.7,
      "grad_norm": 2.632451057434082,
      "learning_rate": 2.200803315595579e-06,
      "loss": 0.8225,
      "step": 11470
    },
    {
      "epoch": 4.7,
      "grad_norm": 2.649582624435425,
      "learning_rate": 2.2027220630372493e-06,
      "loss": 0.8134,
      "step": 11480
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.9486398696899414,
      "learning_rate": 2.2046408104789193e-06,
      "loss": 0.8212,
      "step": 11490
    },
    {
      "epoch": 4.71,
      "grad_norm": 2.905886650085449,
      "learning_rate": 2.206559557920589e-06,
      "loss": 0.8147,
      "step": 11500
    },
    {
      "epoch": 4.71,
      "grad_norm": 2.3274035453796387,
      "learning_rate": 2.208478305362259e-06,
      "loss": 0.7961,
      "step": 11510
    },
    {
      "epoch": 4.72,
      "grad_norm": 3.9558913707733154,
      "learning_rate": 2.2103970528039295e-06,
      "loss": 0.8137,
      "step": 11520
    },
    {
      "epoch": 4.72,
      "grad_norm": 3.140258312225342,
      "learning_rate": 2.2123158002455994e-06,
      "loss": 0.815,
      "step": 11530
    },
    {
      "epoch": 4.72,
      "grad_norm": 2.501121759414673,
      "learning_rate": 2.2142345476872694e-06,
      "loss": 0.8472,
      "step": 11540
    },
    {
      "epoch": 4.73,
      "grad_norm": 1.918448567390442,
      "learning_rate": 2.2161532951289397e-06,
      "loss": 0.8052,
      "step": 11550
    },
    {
      "epoch": 4.73,
      "grad_norm": 4.0419487953186035,
      "learning_rate": 2.2180720425706097e-06,
      "loss": 0.8187,
      "step": 11560
    },
    {
      "epoch": 4.74,
      "grad_norm": 2.201221466064453,
      "learning_rate": 2.2199907900122796e-06,
      "loss": 0.8148,
      "step": 11570
    },
    {
      "epoch": 4.74,
      "grad_norm": 2.0684802532196045,
      "learning_rate": 2.22190953745395e-06,
      "loss": 0.805,
      "step": 11580
    },
    {
      "epoch": 4.74,
      "grad_norm": 3.38741397857666,
      "learning_rate": 2.22382828489562e-06,
      "loss": 0.8115,
      "step": 11590
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.9394025802612305,
      "learning_rate": 2.22574703233729e-06,
      "loss": 0.8412,
      "step": 11600
    },
    {
      "epoch": 4.75,
      "grad_norm": 2.5204226970672607,
      "learning_rate": 2.2276657797789602e-06,
      "loss": 0.8191,
      "step": 11610
    },
    {
      "epoch": 4.76,
      "grad_norm": 2.34157657623291,
      "learning_rate": 2.22958452722063e-06,
      "loss": 0.8032,
      "step": 11620
    },
    {
      "epoch": 4.76,
      "grad_norm": 3.5512959957122803,
      "learning_rate": 2.2315032746623e-06,
      "loss": 0.8243,
      "step": 11630
    },
    {
      "epoch": 4.76,
      "grad_norm": 2.2808852195739746,
      "learning_rate": 2.2334220221039705e-06,
      "loss": 0.818,
      "step": 11640
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.8244966268539429,
      "learning_rate": 2.2353407695456404e-06,
      "loss": 0.8298,
      "step": 11650
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.9737284183502197,
      "learning_rate": 2.2372595169873103e-06,
      "loss": 0.8231,
      "step": 11660
    },
    {
      "epoch": 4.78,
      "grad_norm": 3.448364019393921,
      "learning_rate": 2.2391782644289807e-06,
      "loss": 0.8219,
      "step": 11670
    },
    {
      "epoch": 4.78,
      "grad_norm": 2.9290993213653564,
      "learning_rate": 2.2410970118706506e-06,
      "loss": 0.8017,
      "step": 11680
    },
    {
      "epoch": 4.79,
      "grad_norm": 2.018939256668091,
      "learning_rate": 2.2430157593123206e-06,
      "loss": 0.8184,
      "step": 11690
    },
    {
      "epoch": 4.79,
      "grad_norm": 2.5841424465179443,
      "learning_rate": 2.244934506753991e-06,
      "loss": 0.8199,
      "step": 11700
    },
    {
      "epoch": 4.79,
      "grad_norm": 1.8721470832824707,
      "learning_rate": 2.246853254195661e-06,
      "loss": 0.7969,
      "step": 11710
    },
    {
      "epoch": 4.8,
      "grad_norm": 5.8440046310424805,
      "learning_rate": 2.248772001637331e-06,
      "loss": 0.8151,
      "step": 11720
    },
    {
      "epoch": 4.8,
      "grad_norm": 2.99206805229187,
      "learning_rate": 2.2506907490790008e-06,
      "loss": 0.8102,
      "step": 11730
    },
    {
      "epoch": 4.81,
      "grad_norm": 2.135830879211426,
      "learning_rate": 2.252609496520671e-06,
      "loss": 0.8291,
      "step": 11740
    },
    {
      "epoch": 4.81,
      "grad_norm": 2.373750686645508,
      "learning_rate": 2.254528243962341e-06,
      "loss": 0.8171,
      "step": 11750
    },
    {
      "epoch": 4.81,
      "grad_norm": 2.0906171798706055,
      "learning_rate": 2.256446991404011e-06,
      "loss": 0.8214,
      "step": 11760
    },
    {
      "epoch": 4.82,
      "grad_norm": 2.317537546157837,
      "learning_rate": 2.2583657388456814e-06,
      "loss": 0.8003,
      "step": 11770
    },
    {
      "epoch": 4.82,
      "grad_norm": 2.0348172187805176,
      "learning_rate": 2.2602844862873513e-06,
      "loss": 0.824,
      "step": 11780
    },
    {
      "epoch": 4.83,
      "grad_norm": 2.6010680198669434,
      "learning_rate": 2.2622032337290212e-06,
      "loss": 0.819,
      "step": 11790
    },
    {
      "epoch": 4.83,
      "grad_norm": 4.83579683303833,
      "learning_rate": 2.2641219811706916e-06,
      "loss": 0.8093,
      "step": 11800
    },
    {
      "epoch": 4.83,
      "grad_norm": 2.190183639526367,
      "learning_rate": 2.2660407286123615e-06,
      "loss": 0.806,
      "step": 11810
    },
    {
      "epoch": 4.84,
      "grad_norm": 2.404283046722412,
      "learning_rate": 2.2679594760540315e-06,
      "loss": 0.8219,
      "step": 11820
    },
    {
      "epoch": 4.84,
      "grad_norm": 2.900611639022827,
      "learning_rate": 2.269878223495702e-06,
      "loss": 0.8121,
      "step": 11830
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.6961768865585327,
      "learning_rate": 2.2717969709373718e-06,
      "loss": 0.7989,
      "step": 11840
    },
    {
      "epoch": 4.85,
      "grad_norm": 3.5821242332458496,
      "learning_rate": 2.2737157183790417e-06,
      "loss": 0.8085,
      "step": 11850
    },
    {
      "epoch": 4.85,
      "grad_norm": 2.345088243484497,
      "learning_rate": 2.275634465820712e-06,
      "loss": 0.7891,
      "step": 11860
    },
    {
      "epoch": 4.86,
      "grad_norm": 2.643059253692627,
      "learning_rate": 2.277553213262382e-06,
      "loss": 0.8066,
      "step": 11870
    },
    {
      "epoch": 4.86,
      "grad_norm": 2.200460195541382,
      "learning_rate": 2.2794719607040524e-06,
      "loss": 0.8071,
      "step": 11880
    },
    {
      "epoch": 4.87,
      "grad_norm": 1.8169441223144531,
      "learning_rate": 2.2813907081457223e-06,
      "loss": 0.8226,
      "step": 11890
    },
    {
      "epoch": 4.87,
      "grad_norm": 2.5214550495147705,
      "learning_rate": 2.2833094555873923e-06,
      "loss": 0.8104,
      "step": 11900
    },
    {
      "epoch": 4.88,
      "grad_norm": 2.2613413333892822,
      "learning_rate": 2.2852282030290626e-06,
      "loss": 0.8003,
      "step": 11910
    },
    {
      "epoch": 4.88,
      "grad_norm": 2.4181933403015137,
      "learning_rate": 2.2871469504707326e-06,
      "loss": 0.7999,
      "step": 11920
    },
    {
      "epoch": 4.88,
      "grad_norm": 3.104966640472412,
      "learning_rate": 2.2890656979124025e-06,
      "loss": 0.8026,
      "step": 11930
    },
    {
      "epoch": 4.89,
      "grad_norm": 2.4289438724517822,
      "learning_rate": 2.290984445354073e-06,
      "loss": 0.8049,
      "step": 11940
    },
    {
      "epoch": 4.89,
      "grad_norm": 1.8626949787139893,
      "learning_rate": 2.292903192795743e-06,
      "loss": 0.8102,
      "step": 11950
    },
    {
      "epoch": 4.9,
      "grad_norm": 3.290153741836548,
      "learning_rate": 2.2948219402374127e-06,
      "loss": 0.8165,
      "step": 11960
    },
    {
      "epoch": 4.9,
      "grad_norm": 3.0965116024017334,
      "learning_rate": 2.296740687679083e-06,
      "loss": 0.795,
      "step": 11970
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.816860556602478,
      "learning_rate": 2.298659435120753e-06,
      "loss": 0.8286,
      "step": 11980
    },
    {
      "epoch": 4.91,
      "grad_norm": 2.873619794845581,
      "learning_rate": 2.300578182562423e-06,
      "loss": 0.8191,
      "step": 11990
    },
    {
      "epoch": 4.91,
      "grad_norm": 1.720089316368103,
      "learning_rate": 2.3024969300040934e-06,
      "loss": 0.7851,
      "step": 12000
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.7483805418014526,
      "learning_rate": 2.3044156774457633e-06,
      "loss": 0.8306,
      "step": 12010
    },
    {
      "epoch": 4.92,
      "grad_norm": 2.1875736713409424,
      "learning_rate": 2.3063344248874332e-06,
      "loss": 0.7953,
      "step": 12020
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.7056258916854858,
      "learning_rate": 2.3082531723291036e-06,
      "loss": 0.8124,
      "step": 12030
    },
    {
      "epoch": 4.93,
      "grad_norm": 1.8559974431991577,
      "learning_rate": 2.3101719197707735e-06,
      "loss": 0.8084,
      "step": 12040
    },
    {
      "epoch": 4.93,
      "grad_norm": 2.6848037242889404,
      "learning_rate": 2.3120906672124435e-06,
      "loss": 0.8061,
      "step": 12050
    },
    {
      "epoch": 4.94,
      "grad_norm": 2.953944206237793,
      "learning_rate": 2.314009414654114e-06,
      "loss": 0.8011,
      "step": 12060
    },
    {
      "epoch": 4.94,
      "grad_norm": 1.7438480854034424,
      "learning_rate": 2.3159281620957838e-06,
      "loss": 0.8204,
      "step": 12070
    },
    {
      "epoch": 4.94,
      "grad_norm": 2.1849615573883057,
      "learning_rate": 2.3178469095374537e-06,
      "loss": 0.802,
      "step": 12080
    },
    {
      "epoch": 4.95,
      "grad_norm": 1.2791792154312134,
      "learning_rate": 2.319765656979124e-06,
      "loss": 0.7959,
      "step": 12090
    },
    {
      "epoch": 4.95,
      "grad_norm": 1.8999049663543701,
      "learning_rate": 2.321684404420794e-06,
      "loss": 0.8149,
      "step": 12100
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.9907535314559937,
      "learning_rate": 2.323603151862464e-06,
      "loss": 0.7913,
      "step": 12110
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.9084516763687134,
      "learning_rate": 2.325521899304134e-06,
      "loss": 0.8112,
      "step": 12120
    },
    {
      "epoch": 4.97,
      "grad_norm": 1.701418399810791,
      "learning_rate": 2.3274406467458043e-06,
      "loss": 0.8309,
      "step": 12130
    },
    {
      "epoch": 4.97,
      "grad_norm": 2.2903552055358887,
      "learning_rate": 2.329359394187474e-06,
      "loss": 0.8021,
      "step": 12140
    },
    {
      "epoch": 4.97,
      "grad_norm": 2.610959053039551,
      "learning_rate": 2.331278141629144e-06,
      "loss": 0.8077,
      "step": 12150
    },
    {
      "epoch": 4.98,
      "grad_norm": 2.766728162765503,
      "learning_rate": 2.3331968890708145e-06,
      "loss": 0.8037,
      "step": 12160
    },
    {
      "epoch": 4.98,
      "grad_norm": 1.8133352994918823,
      "learning_rate": 2.3351156365124844e-06,
      "loss": 0.7946,
      "step": 12170
    },
    {
      "epoch": 4.99,
      "grad_norm": 3.714430809020996,
      "learning_rate": 2.3370343839541544e-06,
      "loss": 0.8282,
      "step": 12180
    },
    {
      "epoch": 4.99,
      "grad_norm": 2.19608736038208,
      "learning_rate": 2.3389531313958247e-06,
      "loss": 0.8043,
      "step": 12190
    },
    {
      "epoch": 4.99,
      "grad_norm": 2.854498863220215,
      "learning_rate": 2.3408718788374947e-06,
      "loss": 0.8054,
      "step": 12200
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.5339925289154053,
      "learning_rate": 2.3427906262791646e-06,
      "loss": 0.8307,
      "step": 12210
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.8076708316802979,
      "eval_runtime": 52.2044,
      "eval_samples_per_second": 66.067,
      "eval_steps_per_second": 8.275,
      "step": 12215
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.805203676223755,
      "learning_rate": 2.344709373720835e-06,
      "loss": 0.8222,
      "step": 12220
    },
    {
      "epoch": 5.01,
      "grad_norm": 2.3252837657928467,
      "learning_rate": 2.346628121162505e-06,
      "loss": 0.7997,
      "step": 12230
    },
    {
      "epoch": 5.01,
      "grad_norm": 2.330155611038208,
      "learning_rate": 2.348546868604175e-06,
      "loss": 0.8027,
      "step": 12240
    },
    {
      "epoch": 5.01,
      "grad_norm": 1.8571124076843262,
      "learning_rate": 2.3504656160458452e-06,
      "loss": 0.7781,
      "step": 12250
    },
    {
      "epoch": 5.02,
      "grad_norm": 1.7819281816482544,
      "learning_rate": 2.352384363487515e-06,
      "loss": 0.8061,
      "step": 12260
    },
    {
      "epoch": 5.02,
      "grad_norm": 2.018772840499878,
      "learning_rate": 2.354303110929185e-06,
      "loss": 0.8193,
      "step": 12270
    },
    {
      "epoch": 5.03,
      "grad_norm": 3.123751640319824,
      "learning_rate": 2.3562218583708555e-06,
      "loss": 0.7949,
      "step": 12280
    },
    {
      "epoch": 5.03,
      "grad_norm": 1.9870251417160034,
      "learning_rate": 2.3581406058125254e-06,
      "loss": 0.8023,
      "step": 12290
    },
    {
      "epoch": 5.03,
      "grad_norm": 1.7678509950637817,
      "learning_rate": 2.3600593532541953e-06,
      "loss": 0.8098,
      "step": 12300
    },
    {
      "epoch": 5.04,
      "grad_norm": 3.1165521144866943,
      "learning_rate": 2.3619781006958657e-06,
      "loss": 0.7919,
      "step": 12310
    },
    {
      "epoch": 5.04,
      "grad_norm": 2.8455400466918945,
      "learning_rate": 2.3638968481375356e-06,
      "loss": 0.8183,
      "step": 12320
    },
    {
      "epoch": 5.05,
      "grad_norm": 2.5421769618988037,
      "learning_rate": 2.3658155955792056e-06,
      "loss": 0.808,
      "step": 12330
    },
    {
      "epoch": 5.05,
      "grad_norm": 2.513643503189087,
      "learning_rate": 2.3677343430208755e-06,
      "loss": 0.8202,
      "step": 12340
    },
    {
      "epoch": 5.06,
      "grad_norm": 1.870002031326294,
      "learning_rate": 2.369653090462546e-06,
      "loss": 0.8067,
      "step": 12350
    },
    {
      "epoch": 5.06,
      "grad_norm": 2.3300178050994873,
      "learning_rate": 2.371571837904216e-06,
      "loss": 0.7964,
      "step": 12360
    },
    {
      "epoch": 5.06,
      "grad_norm": 2.5590097904205322,
      "learning_rate": 2.3734905853458858e-06,
      "loss": 0.8107,
      "step": 12370
    },
    {
      "epoch": 5.07,
      "grad_norm": 3.336146831512451,
      "learning_rate": 2.375409332787556e-06,
      "loss": 0.8125,
      "step": 12380
    },
    {
      "epoch": 5.07,
      "grad_norm": 2.6458935737609863,
      "learning_rate": 2.377328080229226e-06,
      "loss": 0.8023,
      "step": 12390
    },
    {
      "epoch": 5.08,
      "grad_norm": 1.9496732950210571,
      "learning_rate": 2.379246827670896e-06,
      "loss": 0.7989,
      "step": 12400
    },
    {
      "epoch": 5.08,
      "grad_norm": 4.910450458526611,
      "learning_rate": 2.3811655751125664e-06,
      "loss": 0.8025,
      "step": 12410
    },
    {
      "epoch": 5.08,
      "grad_norm": 4.030574321746826,
      "learning_rate": 2.3830843225542363e-06,
      "loss": 0.8078,
      "step": 12420
    },
    {
      "epoch": 5.09,
      "grad_norm": 2.7242910861968994,
      "learning_rate": 2.3850030699959062e-06,
      "loss": 0.7952,
      "step": 12430
    },
    {
      "epoch": 5.09,
      "grad_norm": 2.5644755363464355,
      "learning_rate": 2.3869218174375766e-06,
      "loss": 0.8131,
      "step": 12440
    },
    {
      "epoch": 5.1,
      "grad_norm": 2.0156025886535645,
      "learning_rate": 2.3888405648792465e-06,
      "loss": 0.794,
      "step": 12450
    },
    {
      "epoch": 5.1,
      "grad_norm": 1.8938802480697632,
      "learning_rate": 2.3907593123209165e-06,
      "loss": 0.817,
      "step": 12460
    },
    {
      "epoch": 5.1,
      "grad_norm": 2.3447282314300537,
      "learning_rate": 2.392678059762587e-06,
      "loss": 0.8079,
      "step": 12470
    },
    {
      "epoch": 5.11,
      "grad_norm": 2.958418369293213,
      "learning_rate": 2.3945968072042568e-06,
      "loss": 0.7927,
      "step": 12480
    },
    {
      "epoch": 5.11,
      "grad_norm": 3.6668879985809326,
      "learning_rate": 2.3965155546459267e-06,
      "loss": 0.7951,
      "step": 12490
    },
    {
      "epoch": 5.12,
      "grad_norm": 1.74611234664917,
      "learning_rate": 2.398434302087597e-06,
      "loss": 0.8004,
      "step": 12500
    },
    {
      "epoch": 5.12,
      "grad_norm": 2.6111724376678467,
      "learning_rate": 2.400353049529267e-06,
      "loss": 0.8215,
      "step": 12510
    },
    {
      "epoch": 5.12,
      "grad_norm": 1.8024252653121948,
      "learning_rate": 2.402271796970937e-06,
      "loss": 0.7898,
      "step": 12520
    },
    {
      "epoch": 5.13,
      "grad_norm": 1.905828833580017,
      "learning_rate": 2.4041905444126073e-06,
      "loss": 0.7953,
      "step": 12530
    },
    {
      "epoch": 5.13,
      "grad_norm": 3.5009686946868896,
      "learning_rate": 2.4061092918542773e-06,
      "loss": 0.8033,
      "step": 12540
    },
    {
      "epoch": 5.14,
      "grad_norm": 1.522625207901001,
      "learning_rate": 2.408028039295947e-06,
      "loss": 0.8105,
      "step": 12550
    },
    {
      "epoch": 5.14,
      "grad_norm": 2.2906570434570312,
      "learning_rate": 2.409946786737617e-06,
      "loss": 0.808,
      "step": 12560
    },
    {
      "epoch": 5.15,
      "grad_norm": 2.0269408226013184,
      "learning_rate": 2.4118655341792875e-06,
      "loss": 0.7845,
      "step": 12570
    },
    {
      "epoch": 5.15,
      "grad_norm": 2.653125762939453,
      "learning_rate": 2.4137842816209574e-06,
      "loss": 0.8078,
      "step": 12580
    },
    {
      "epoch": 5.15,
      "grad_norm": 2.7181434631347656,
      "learning_rate": 2.4157030290626274e-06,
      "loss": 0.797,
      "step": 12590
    },
    {
      "epoch": 5.16,
      "grad_norm": 2.1443395614624023,
      "learning_rate": 2.4176217765042977e-06,
      "loss": 0.7788,
      "step": 12600
    },
    {
      "epoch": 5.16,
      "grad_norm": 3.0488932132720947,
      "learning_rate": 2.4195405239459677e-06,
      "loss": 0.8153,
      "step": 12610
    },
    {
      "epoch": 5.17,
      "grad_norm": 3.565387487411499,
      "learning_rate": 2.4214592713876376e-06,
      "loss": 0.7838,
      "step": 12620
    },
    {
      "epoch": 5.17,
      "grad_norm": 2.129504442214966,
      "learning_rate": 2.423378018829308e-06,
      "loss": 0.7878,
      "step": 12630
    },
    {
      "epoch": 5.17,
      "grad_norm": 2.225531578063965,
      "learning_rate": 2.425296766270978e-06,
      "loss": 0.7948,
      "step": 12640
    },
    {
      "epoch": 5.18,
      "grad_norm": 1.8932526111602783,
      "learning_rate": 2.427215513712648e-06,
      "loss": 0.7979,
      "step": 12650
    },
    {
      "epoch": 5.18,
      "grad_norm": 3.2925751209259033,
      "learning_rate": 2.4291342611543182e-06,
      "loss": 0.799,
      "step": 12660
    },
    {
      "epoch": 5.19,
      "grad_norm": 2.557081937789917,
      "learning_rate": 2.431053008595988e-06,
      "loss": 0.7995,
      "step": 12670
    },
    {
      "epoch": 5.19,
      "grad_norm": 3.621591567993164,
      "learning_rate": 2.432971756037658e-06,
      "loss": 0.788,
      "step": 12680
    },
    {
      "epoch": 5.19,
      "grad_norm": 2.5594003200531006,
      "learning_rate": 2.4348905034793285e-06,
      "loss": 0.797,
      "step": 12690
    },
    {
      "epoch": 5.2,
      "grad_norm": 2.2280361652374268,
      "learning_rate": 2.4368092509209984e-06,
      "loss": 0.7846,
      "step": 12700
    },
    {
      "epoch": 5.2,
      "grad_norm": 2.6865930557250977,
      "learning_rate": 2.4387279983626683e-06,
      "loss": 0.7913,
      "step": 12710
    },
    {
      "epoch": 5.21,
      "grad_norm": 2.6459529399871826,
      "learning_rate": 2.4406467458043387e-06,
      "loss": 0.8019,
      "step": 12720
    },
    {
      "epoch": 5.21,
      "grad_norm": 3.2846932411193848,
      "learning_rate": 2.4425654932460086e-06,
      "loss": 0.7829,
      "step": 12730
    },
    {
      "epoch": 5.21,
      "grad_norm": 2.4437286853790283,
      "learning_rate": 2.4444842406876786e-06,
      "loss": 0.7945,
      "step": 12740
    },
    {
      "epoch": 5.22,
      "grad_norm": 1.5402343273162842,
      "learning_rate": 2.446402988129349e-06,
      "loss": 0.7842,
      "step": 12750
    },
    {
      "epoch": 5.22,
      "grad_norm": 2.954831838607788,
      "learning_rate": 2.448321735571019e-06,
      "loss": 0.8167,
      "step": 12760
    },
    {
      "epoch": 5.23,
      "grad_norm": 2.271108865737915,
      "learning_rate": 2.450240483012689e-06,
      "loss": 0.7901,
      "step": 12770
    },
    {
      "epoch": 5.23,
      "grad_norm": 2.7889199256896973,
      "learning_rate": 2.4521592304543588e-06,
      "loss": 0.7867,
      "step": 12780
    },
    {
      "epoch": 5.24,
      "grad_norm": 3.1606285572052,
      "learning_rate": 2.454077977896029e-06,
      "loss": 0.8073,
      "step": 12790
    },
    {
      "epoch": 5.24,
      "grad_norm": 3.32816743850708,
      "learning_rate": 2.455996725337699e-06,
      "loss": 0.7956,
      "step": 12800
    },
    {
      "epoch": 5.24,
      "grad_norm": 1.3588625192642212,
      "learning_rate": 2.457915472779369e-06,
      "loss": 0.8143,
      "step": 12810
    },
    {
      "epoch": 5.25,
      "grad_norm": 2.5756661891937256,
      "learning_rate": 2.4598342202210394e-06,
      "loss": 0.7797,
      "step": 12820
    },
    {
      "epoch": 5.25,
      "grad_norm": 3.1179187297821045,
      "learning_rate": 2.4617529676627093e-06,
      "loss": 0.8028,
      "step": 12830
    },
    {
      "epoch": 5.26,
      "grad_norm": 2.8271539211273193,
      "learning_rate": 2.4636717151043792e-06,
      "loss": 0.82,
      "step": 12840
    },
    {
      "epoch": 5.26,
      "grad_norm": 2.774052381515503,
      "learning_rate": 2.4655904625460496e-06,
      "loss": 0.791,
      "step": 12850
    },
    {
      "epoch": 5.26,
      "grad_norm": 2.8205552101135254,
      "learning_rate": 2.4675092099877195e-06,
      "loss": 0.8053,
      "step": 12860
    },
    {
      "epoch": 5.27,
      "grad_norm": 2.7559378147125244,
      "learning_rate": 2.4694279574293895e-06,
      "loss": 0.8037,
      "step": 12870
    },
    {
      "epoch": 5.27,
      "grad_norm": 3.0554440021514893,
      "learning_rate": 2.47134670487106e-06,
      "loss": 0.8072,
      "step": 12880
    },
    {
      "epoch": 5.28,
      "grad_norm": 1.893148422241211,
      "learning_rate": 2.47326545231273e-06,
      "loss": 0.8012,
      "step": 12890
    },
    {
      "epoch": 5.28,
      "grad_norm": 1.6404892206192017,
      "learning_rate": 2.4751841997544006e-06,
      "loss": 0.8007,
      "step": 12900
    },
    {
      "epoch": 5.28,
      "grad_norm": 2.0144965648651123,
      "learning_rate": 2.4771029471960705e-06,
      "loss": 0.7944,
      "step": 12910
    },
    {
      "epoch": 5.29,
      "grad_norm": 2.4606211185455322,
      "learning_rate": 2.4790216946377405e-06,
      "loss": 0.7809,
      "step": 12920
    },
    {
      "epoch": 5.29,
      "grad_norm": 3.729370594024658,
      "learning_rate": 2.4809404420794104e-06,
      "loss": 0.7735,
      "step": 12930
    },
    {
      "epoch": 5.3,
      "grad_norm": 2.9163951873779297,
      "learning_rate": 2.4828591895210808e-06,
      "loss": 0.7813,
      "step": 12940
    },
    {
      "epoch": 5.3,
      "grad_norm": 2.6349785327911377,
      "learning_rate": 2.4847779369627507e-06,
      "loss": 0.7909,
      "step": 12950
    },
    {
      "epoch": 5.3,
      "grad_norm": 2.251701593399048,
      "learning_rate": 2.4866966844044206e-06,
      "loss": 0.805,
      "step": 12960
    },
    {
      "epoch": 5.31,
      "grad_norm": 1.562479853630066,
      "learning_rate": 2.488615431846091e-06,
      "loss": 0.798,
      "step": 12970
    },
    {
      "epoch": 5.31,
      "grad_norm": 2.8158960342407227,
      "learning_rate": 2.490534179287761e-06,
      "loss": 0.8122,
      "step": 12980
    },
    {
      "epoch": 5.32,
      "grad_norm": 2.7544050216674805,
      "learning_rate": 2.492452926729431e-06,
      "loss": 0.7758,
      "step": 12990
    },
    {
      "epoch": 5.32,
      "grad_norm": 3.119272470474243,
      "learning_rate": 2.4943716741711012e-06,
      "loss": 0.8108,
      "step": 13000
    },
    {
      "epoch": 5.33,
      "grad_norm": 1.9506605863571167,
      "learning_rate": 2.496290421612771e-06,
      "loss": 0.7788,
      "step": 13010
    },
    {
      "epoch": 5.33,
      "grad_norm": 1.906472086906433,
      "learning_rate": 2.498209169054441e-06,
      "loss": 0.7887,
      "step": 13020
    },
    {
      "epoch": 5.33,
      "grad_norm": 3.296656847000122,
      "learning_rate": 2.5001279164961115e-06,
      "loss": 0.7923,
      "step": 13030
    },
    {
      "epoch": 5.34,
      "grad_norm": 2.7215940952301025,
      "learning_rate": 2.5020466639377814e-06,
      "loss": 0.8004,
      "step": 13040
    },
    {
      "epoch": 5.34,
      "grad_norm": 2.1867363452911377,
      "learning_rate": 2.5039654113794514e-06,
      "loss": 0.8052,
      "step": 13050
    },
    {
      "epoch": 5.35,
      "grad_norm": 2.9296164512634277,
      "learning_rate": 2.5058841588211217e-06,
      "loss": 0.7934,
      "step": 13060
    },
    {
      "epoch": 5.35,
      "grad_norm": 2.726388454437256,
      "learning_rate": 2.5078029062627917e-06,
      "loss": 0.7781,
      "step": 13070
    },
    {
      "epoch": 5.35,
      "grad_norm": 2.8639771938323975,
      "learning_rate": 2.5097216537044616e-06,
      "loss": 0.7855,
      "step": 13080
    },
    {
      "epoch": 5.36,
      "grad_norm": 2.169019937515259,
      "learning_rate": 2.511640401146132e-06,
      "loss": 0.8145,
      "step": 13090
    },
    {
      "epoch": 5.36,
      "grad_norm": 2.488469362258911,
      "learning_rate": 2.513559148587802e-06,
      "loss": 0.7914,
      "step": 13100
    },
    {
      "epoch": 5.37,
      "grad_norm": 2.369385004043579,
      "learning_rate": 2.515477896029472e-06,
      "loss": 0.8021,
      "step": 13110
    },
    {
      "epoch": 5.37,
      "grad_norm": 2.9302170276641846,
      "learning_rate": 2.517396643471142e-06,
      "loss": 0.7915,
      "step": 13120
    },
    {
      "epoch": 5.37,
      "grad_norm": 3.508714199066162,
      "learning_rate": 2.519315390912812e-06,
      "loss": 0.7976,
      "step": 13130
    },
    {
      "epoch": 5.38,
      "grad_norm": 1.6736503839492798,
      "learning_rate": 2.521234138354482e-06,
      "loss": 0.8054,
      "step": 13140
    },
    {
      "epoch": 5.38,
      "grad_norm": 2.262608528137207,
      "learning_rate": 2.523152885796152e-06,
      "loss": 0.7989,
      "step": 13150
    },
    {
      "epoch": 5.39,
      "grad_norm": 3.040647506713867,
      "learning_rate": 2.5250716332378224e-06,
      "loss": 0.8172,
      "step": 13160
    },
    {
      "epoch": 5.39,
      "grad_norm": 2.8032798767089844,
      "learning_rate": 2.5269903806794923e-06,
      "loss": 0.8138,
      "step": 13170
    },
    {
      "epoch": 5.4,
      "grad_norm": 3.7263875007629395,
      "learning_rate": 2.5289091281211623e-06,
      "loss": 0.7878,
      "step": 13180
    },
    {
      "epoch": 5.4,
      "grad_norm": 1.341167688369751,
      "learning_rate": 2.5308278755628326e-06,
      "loss": 0.8133,
      "step": 13190
    },
    {
      "epoch": 5.4,
      "grad_norm": 3.3761534690856934,
      "learning_rate": 2.5327466230045026e-06,
      "loss": 0.8023,
      "step": 13200
    },
    {
      "epoch": 5.41,
      "grad_norm": 2.5009732246398926,
      "learning_rate": 2.5346653704461725e-06,
      "loss": 0.786,
      "step": 13210
    },
    {
      "epoch": 5.41,
      "grad_norm": 2.7589144706726074,
      "learning_rate": 2.536584117887843e-06,
      "loss": 0.7856,
      "step": 13220
    },
    {
      "epoch": 5.42,
      "grad_norm": 4.429553508758545,
      "learning_rate": 2.538502865329513e-06,
      "loss": 0.7951,
      "step": 13230
    },
    {
      "epoch": 5.42,
      "grad_norm": 1.370830774307251,
      "learning_rate": 2.5404216127711827e-06,
      "loss": 0.8053,
      "step": 13240
    },
    {
      "epoch": 5.42,
      "grad_norm": 1.7929476499557495,
      "learning_rate": 2.542340360212853e-06,
      "loss": 0.7762,
      "step": 13250
    },
    {
      "epoch": 5.43,
      "grad_norm": 2.19031023979187,
      "learning_rate": 2.544259107654523e-06,
      "loss": 0.7937,
      "step": 13260
    },
    {
      "epoch": 5.43,
      "grad_norm": 2.974792718887329,
      "learning_rate": 2.546177855096193e-06,
      "loss": 0.7829,
      "step": 13270
    },
    {
      "epoch": 5.44,
      "grad_norm": 2.500720262527466,
      "learning_rate": 2.5480966025378633e-06,
      "loss": 0.7835,
      "step": 13280
    },
    {
      "epoch": 5.44,
      "grad_norm": 3.38653564453125,
      "learning_rate": 2.5500153499795333e-06,
      "loss": 0.8032,
      "step": 13290
    },
    {
      "epoch": 5.44,
      "grad_norm": 2.539163589477539,
      "learning_rate": 2.5519340974212032e-06,
      "loss": 0.8036,
      "step": 13300
    },
    {
      "epoch": 5.45,
      "grad_norm": 1.9775768518447876,
      "learning_rate": 2.5538528448628736e-06,
      "loss": 0.7968,
      "step": 13310
    },
    {
      "epoch": 5.45,
      "grad_norm": 3.385003089904785,
      "learning_rate": 2.5557715923045435e-06,
      "loss": 0.7929,
      "step": 13320
    },
    {
      "epoch": 5.46,
      "grad_norm": 2.3745057582855225,
      "learning_rate": 2.5576903397462135e-06,
      "loss": 0.7956,
      "step": 13330
    },
    {
      "epoch": 5.46,
      "grad_norm": 2.7914650440216064,
      "learning_rate": 2.559609087187884e-06,
      "loss": 0.8157,
      "step": 13340
    },
    {
      "epoch": 5.46,
      "grad_norm": 2.63881778717041,
      "learning_rate": 2.5615278346295538e-06,
      "loss": 0.7979,
      "step": 13350
    },
    {
      "epoch": 5.47,
      "grad_norm": 2.492649555206299,
      "learning_rate": 2.5634465820712237e-06,
      "loss": 0.8016,
      "step": 13360
    },
    {
      "epoch": 5.47,
      "grad_norm": 2.3091604709625244,
      "learning_rate": 2.5653653295128936e-06,
      "loss": 0.7963,
      "step": 13370
    },
    {
      "epoch": 5.48,
      "grad_norm": 4.018183708190918,
      "learning_rate": 2.567284076954564e-06,
      "loss": 0.809,
      "step": 13380
    },
    {
      "epoch": 5.48,
      "grad_norm": 3.455778121948242,
      "learning_rate": 2.569202824396234e-06,
      "loss": 0.8136,
      "step": 13390
    },
    {
      "epoch": 5.49,
      "grad_norm": 2.243960380554199,
      "learning_rate": 2.571121571837904e-06,
      "loss": 0.7969,
      "step": 13400
    },
    {
      "epoch": 5.49,
      "grad_norm": 1.9592015743255615,
      "learning_rate": 2.5730403192795742e-06,
      "loss": 0.7949,
      "step": 13410
    },
    {
      "epoch": 5.49,
      "grad_norm": 2.488445997238159,
      "learning_rate": 2.574959066721244e-06,
      "loss": 0.7891,
      "step": 13420
    },
    {
      "epoch": 5.5,
      "grad_norm": 2.8160006999969482,
      "learning_rate": 2.576877814162914e-06,
      "loss": 0.7789,
      "step": 13430
    },
    {
      "epoch": 5.5,
      "grad_norm": 2.856961488723755,
      "learning_rate": 2.5787965616045845e-06,
      "loss": 0.7741,
      "step": 13440
    },
    {
      "epoch": 5.51,
      "grad_norm": 2.6501734256744385,
      "learning_rate": 2.5807153090462544e-06,
      "loss": 0.7839,
      "step": 13450
    },
    {
      "epoch": 5.51,
      "grad_norm": 2.255722761154175,
      "learning_rate": 2.5826340564879244e-06,
      "loss": 0.7959,
      "step": 13460
    },
    {
      "epoch": 5.51,
      "grad_norm": 3.313262701034546,
      "learning_rate": 2.5845528039295947e-06,
      "loss": 0.7889,
      "step": 13470
    },
    {
      "epoch": 5.52,
      "grad_norm": 2.0520341396331787,
      "learning_rate": 2.5864715513712647e-06,
      "loss": 0.7709,
      "step": 13480
    },
    {
      "epoch": 5.52,
      "grad_norm": 2.3238039016723633,
      "learning_rate": 2.5883902988129346e-06,
      "loss": 0.7852,
      "step": 13490
    },
    {
      "epoch": 5.53,
      "grad_norm": 2.534193992614746,
      "learning_rate": 2.590309046254605e-06,
      "loss": 0.7856,
      "step": 13500
    },
    {
      "epoch": 5.53,
      "grad_norm": 2.866605281829834,
      "learning_rate": 2.592227793696275e-06,
      "loss": 0.7918,
      "step": 13510
    },
    {
      "epoch": 5.53,
      "grad_norm": 4.253636360168457,
      "learning_rate": 2.594146541137945e-06,
      "loss": 0.794,
      "step": 13520
    },
    {
      "epoch": 5.54,
      "grad_norm": 4.083089828491211,
      "learning_rate": 2.596065288579615e-06,
      "loss": 0.7923,
      "step": 13530
    },
    {
      "epoch": 5.54,
      "grad_norm": 3.9399573802948,
      "learning_rate": 2.597984036021285e-06,
      "loss": 0.7865,
      "step": 13540
    },
    {
      "epoch": 5.55,
      "grad_norm": 2.8253235816955566,
      "learning_rate": 2.599902783462955e-06,
      "loss": 0.8105,
      "step": 13550
    },
    {
      "epoch": 5.55,
      "grad_norm": 1.944982647895813,
      "learning_rate": 2.6018215309046254e-06,
      "loss": 0.7765,
      "step": 13560
    },
    {
      "epoch": 5.55,
      "grad_norm": 4.147810935974121,
      "learning_rate": 2.6037402783462954e-06,
      "loss": 0.7868,
      "step": 13570
    },
    {
      "epoch": 5.56,
      "grad_norm": 2.420058012008667,
      "learning_rate": 2.6056590257879653e-06,
      "loss": 0.7873,
      "step": 13580
    },
    {
      "epoch": 5.56,
      "grad_norm": 2.6404054164886475,
      "learning_rate": 2.6075777732296353e-06,
      "loss": 0.7714,
      "step": 13590
    },
    {
      "epoch": 5.57,
      "grad_norm": 2.2874796390533447,
      "learning_rate": 2.6094965206713056e-06,
      "loss": 0.8,
      "step": 13600
    },
    {
      "epoch": 5.57,
      "grad_norm": 2.775066375732422,
      "learning_rate": 2.6114152681129756e-06,
      "loss": 0.8096,
      "step": 13610
    },
    {
      "epoch": 5.58,
      "grad_norm": 2.299852132797241,
      "learning_rate": 2.6133340155546455e-06,
      "loss": 0.7696,
      "step": 13620
    },
    {
      "epoch": 5.58,
      "grad_norm": 2.9270899295806885,
      "learning_rate": 2.615252762996316e-06,
      "loss": 0.7941,
      "step": 13630
    },
    {
      "epoch": 5.58,
      "grad_norm": 3.471794843673706,
      "learning_rate": 2.617171510437986e-06,
      "loss": 0.7798,
      "step": 13640
    },
    {
      "epoch": 5.59,
      "grad_norm": 2.678206205368042,
      "learning_rate": 2.6190902578796557e-06,
      "loss": 0.7959,
      "step": 13650
    },
    {
      "epoch": 5.59,
      "grad_norm": 2.263636350631714,
      "learning_rate": 2.621009005321326e-06,
      "loss": 0.7799,
      "step": 13660
    },
    {
      "epoch": 5.6,
      "grad_norm": 3.783001184463501,
      "learning_rate": 2.622927752762996e-06,
      "loss": 0.7694,
      "step": 13670
    },
    {
      "epoch": 5.6,
      "grad_norm": 2.669189214706421,
      "learning_rate": 2.624846500204666e-06,
      "loss": 0.7719,
      "step": 13680
    },
    {
      "epoch": 5.6,
      "grad_norm": 2.771479845046997,
      "learning_rate": 2.6267652476463364e-06,
      "loss": 0.7931,
      "step": 13690
    },
    {
      "epoch": 5.61,
      "grad_norm": 2.1468303203582764,
      "learning_rate": 2.6286839950880063e-06,
      "loss": 0.815,
      "step": 13700
    },
    {
      "epoch": 5.61,
      "grad_norm": 2.503983497619629,
      "learning_rate": 2.6306027425296762e-06,
      "loss": 0.7748,
      "step": 13710
    },
    {
      "epoch": 5.62,
      "grad_norm": 1.853161096572876,
      "learning_rate": 2.6325214899713466e-06,
      "loss": 0.7845,
      "step": 13720
    },
    {
      "epoch": 5.62,
      "grad_norm": 2.17806077003479,
      "learning_rate": 2.6344402374130165e-06,
      "loss": 0.7884,
      "step": 13730
    },
    {
      "epoch": 5.62,
      "grad_norm": 2.3602237701416016,
      "learning_rate": 2.6363589848546865e-06,
      "loss": 0.7755,
      "step": 13740
    },
    {
      "epoch": 5.63,
      "grad_norm": 2.5360372066497803,
      "learning_rate": 2.638277732296357e-06,
      "loss": 0.7843,
      "step": 13750
    },
    {
      "epoch": 5.63,
      "grad_norm": 2.672706127166748,
      "learning_rate": 2.6401964797380268e-06,
      "loss": 0.7798,
      "step": 13760
    },
    {
      "epoch": 5.64,
      "grad_norm": 3.329430103302002,
      "learning_rate": 2.6421152271796967e-06,
      "loss": 0.768,
      "step": 13770
    },
    {
      "epoch": 5.64,
      "grad_norm": 3.304800510406494,
      "learning_rate": 2.6440339746213667e-06,
      "loss": 0.7793,
      "step": 13780
    },
    {
      "epoch": 5.64,
      "grad_norm": 2.054910659790039,
      "learning_rate": 2.645952722063037e-06,
      "loss": 0.7996,
      "step": 13790
    },
    {
      "epoch": 5.65,
      "grad_norm": 2.42056941986084,
      "learning_rate": 2.647871469504707e-06,
      "loss": 0.8041,
      "step": 13800
    },
    {
      "epoch": 5.65,
      "grad_norm": 2.1319570541381836,
      "learning_rate": 2.649790216946377e-06,
      "loss": 0.7859,
      "step": 13810
    },
    {
      "epoch": 5.66,
      "grad_norm": 1.5367980003356934,
      "learning_rate": 2.6517089643880473e-06,
      "loss": 0.7735,
      "step": 13820
    },
    {
      "epoch": 5.66,
      "grad_norm": 3.520789861679077,
      "learning_rate": 2.653627711829717e-06,
      "loss": 0.7863,
      "step": 13830
    },
    {
      "epoch": 5.67,
      "grad_norm": 2.6606638431549072,
      "learning_rate": 2.655546459271387e-06,
      "loss": 0.7982,
      "step": 13840
    },
    {
      "epoch": 5.67,
      "grad_norm": 3.025753974914551,
      "learning_rate": 2.6574652067130575e-06,
      "loss": 0.7838,
      "step": 13850
    },
    {
      "epoch": 5.67,
      "grad_norm": 2.2710211277008057,
      "learning_rate": 2.6593839541547274e-06,
      "loss": 0.7936,
      "step": 13860
    },
    {
      "epoch": 5.68,
      "grad_norm": 3.0301766395568848,
      "learning_rate": 2.6613027015963974e-06,
      "loss": 0.7924,
      "step": 13870
    },
    {
      "epoch": 5.68,
      "grad_norm": 2.538386344909668,
      "learning_rate": 2.6632214490380677e-06,
      "loss": 0.7878,
      "step": 13880
    },
    {
      "epoch": 5.69,
      "grad_norm": 1.749726414680481,
      "learning_rate": 2.6651401964797377e-06,
      "loss": 0.7838,
      "step": 13890
    },
    {
      "epoch": 5.69,
      "grad_norm": 1.6454633474349976,
      "learning_rate": 2.6670589439214076e-06,
      "loss": 0.7871,
      "step": 13900
    },
    {
      "epoch": 5.69,
      "grad_norm": 2.533597230911255,
      "learning_rate": 2.668977691363078e-06,
      "loss": 0.8129,
      "step": 13910
    },
    {
      "epoch": 5.7,
      "grad_norm": 3.9042129516601562,
      "learning_rate": 2.670896438804748e-06,
      "loss": 0.7825,
      "step": 13920
    },
    {
      "epoch": 5.7,
      "grad_norm": 2.8434014320373535,
      "learning_rate": 2.672815186246418e-06,
      "loss": 0.7692,
      "step": 13930
    },
    {
      "epoch": 5.71,
      "grad_norm": 1.304732322692871,
      "learning_rate": 2.6747339336880882e-06,
      "loss": 0.8034,
      "step": 13940
    },
    {
      "epoch": 5.71,
      "grad_norm": 2.693507671356201,
      "learning_rate": 2.676652681129758e-06,
      "loss": 0.7892,
      "step": 13950
    },
    {
      "epoch": 5.71,
      "grad_norm": 2.584059715270996,
      "learning_rate": 2.678571428571428e-06,
      "loss": 0.7736,
      "step": 13960
    },
    {
      "epoch": 5.72,
      "grad_norm": 2.6466386318206787,
      "learning_rate": 2.6804901760130985e-06,
      "loss": 0.7807,
      "step": 13970
    },
    {
      "epoch": 5.72,
      "grad_norm": 1.5940560102462769,
      "learning_rate": 2.6824089234547684e-06,
      "loss": 0.7968,
      "step": 13980
    },
    {
      "epoch": 5.73,
      "grad_norm": 2.867429256439209,
      "learning_rate": 2.6843276708964383e-06,
      "loss": 0.7921,
      "step": 13990
    },
    {
      "epoch": 5.73,
      "grad_norm": 3.255525827407837,
      "learning_rate": 2.6862464183381083e-06,
      "loss": 0.7825,
      "step": 14000
    },
    {
      "epoch": 5.73,
      "grad_norm": 3.505155086517334,
      "learning_rate": 2.6881651657797786e-06,
      "loss": 0.762,
      "step": 14010
    },
    {
      "epoch": 5.74,
      "grad_norm": 2.8005926609039307,
      "learning_rate": 2.6900839132214486e-06,
      "loss": 0.7885,
      "step": 14020
    },
    {
      "epoch": 5.74,
      "grad_norm": 2.0159568786621094,
      "learning_rate": 2.6920026606631185e-06,
      "loss": 0.7796,
      "step": 14030
    },
    {
      "epoch": 5.75,
      "grad_norm": 3.392059326171875,
      "learning_rate": 2.693921408104789e-06,
      "loss": 0.7824,
      "step": 14040
    },
    {
      "epoch": 5.75,
      "grad_norm": 3.555330276489258,
      "learning_rate": 2.695840155546459e-06,
      "loss": 0.7592,
      "step": 14050
    },
    {
      "epoch": 5.76,
      "grad_norm": 3.015880584716797,
      "learning_rate": 2.6977589029881288e-06,
      "loss": 0.7789,
      "step": 14060
    },
    {
      "epoch": 5.76,
      "grad_norm": 3.9418251514434814,
      "learning_rate": 2.699677650429799e-06,
      "loss": 0.8011,
      "step": 14070
    },
    {
      "epoch": 5.76,
      "grad_norm": 2.806443452835083,
      "learning_rate": 2.701596397871469e-06,
      "loss": 0.786,
      "step": 14080
    },
    {
      "epoch": 5.77,
      "grad_norm": 3.9728965759277344,
      "learning_rate": 2.703515145313139e-06,
      "loss": 0.7829,
      "step": 14090
    },
    {
      "epoch": 5.77,
      "grad_norm": 1.840956211090088,
      "learning_rate": 2.7054338927548094e-06,
      "loss": 0.7757,
      "step": 14100
    },
    {
      "epoch": 5.78,
      "grad_norm": 3.642364740371704,
      "learning_rate": 2.7073526401964793e-06,
      "loss": 0.7957,
      "step": 14110
    },
    {
      "epoch": 5.78,
      "grad_norm": 2.416808605194092,
      "learning_rate": 2.7092713876381492e-06,
      "loss": 0.7811,
      "step": 14120
    },
    {
      "epoch": 5.78,
      "grad_norm": 3.470505475997925,
      "learning_rate": 2.7111901350798196e-06,
      "loss": 0.7913,
      "step": 14130
    },
    {
      "epoch": 5.79,
      "grad_norm": 2.88626766204834,
      "learning_rate": 2.7131088825214895e-06,
      "loss": 0.7813,
      "step": 14140
    },
    {
      "epoch": 5.79,
      "grad_norm": 2.713491678237915,
      "learning_rate": 2.7150276299631595e-06,
      "loss": 0.7954,
      "step": 14150
    },
    {
      "epoch": 5.8,
      "grad_norm": 1.8468754291534424,
      "learning_rate": 2.71694637740483e-06,
      "loss": 0.791,
      "step": 14160
    },
    {
      "epoch": 5.8,
      "grad_norm": 3.992105007171631,
      "learning_rate": 2.7188651248464998e-06,
      "loss": 0.7726,
      "step": 14170
    },
    {
      "epoch": 5.8,
      "grad_norm": 3.424274444580078,
      "learning_rate": 2.7207838722881697e-06,
      "loss": 0.7711,
      "step": 14180
    },
    {
      "epoch": 5.81,
      "grad_norm": 2.553164482116699,
      "learning_rate": 2.72270261972984e-06,
      "loss": 0.8128,
      "step": 14190
    },
    {
      "epoch": 5.81,
      "grad_norm": 2.429574489593506,
      "learning_rate": 2.72462136717151e-06,
      "loss": 0.7811,
      "step": 14200
    },
    {
      "epoch": 5.82,
      "grad_norm": 1.8846369981765747,
      "learning_rate": 2.72654011461318e-06,
      "loss": 0.7708,
      "step": 14210
    },
    {
      "epoch": 5.82,
      "grad_norm": 3.1772098541259766,
      "learning_rate": 2.72845886205485e-06,
      "loss": 0.7942,
      "step": 14220
    },
    {
      "epoch": 5.82,
      "grad_norm": 3.492969512939453,
      "learning_rate": 2.7303776094965203e-06,
      "loss": 0.7761,
      "step": 14230
    },
    {
      "epoch": 5.83,
      "grad_norm": 2.405064344406128,
      "learning_rate": 2.73229635693819e-06,
      "loss": 0.7857,
      "step": 14240
    },
    {
      "epoch": 5.83,
      "grad_norm": 2.8850061893463135,
      "learning_rate": 2.73421510437986e-06,
      "loss": 0.7694,
      "step": 14250
    },
    {
      "epoch": 5.84,
      "grad_norm": 2.4860079288482666,
      "learning_rate": 2.736133851821531e-06,
      "loss": 0.7832,
      "step": 14260
    },
    {
      "epoch": 5.84,
      "grad_norm": 2.14864182472229,
      "learning_rate": 2.738052599263201e-06,
      "loss": 0.7602,
      "step": 14270
    },
    {
      "epoch": 5.85,
      "grad_norm": 2.771458625793457,
      "learning_rate": 2.7399713467048712e-06,
      "loss": 0.7771,
      "step": 14280
    },
    {
      "epoch": 5.85,
      "grad_norm": 2.781034469604492,
      "learning_rate": 2.741890094146541e-06,
      "loss": 0.7551,
      "step": 14290
    },
    {
      "epoch": 5.85,
      "grad_norm": 2.4335598945617676,
      "learning_rate": 2.743808841588211e-06,
      "loss": 0.7859,
      "step": 14300
    },
    {
      "epoch": 5.86,
      "grad_norm": 4.053839206695557,
      "learning_rate": 2.7457275890298815e-06,
      "loss": 0.7767,
      "step": 14310
    },
    {
      "epoch": 5.86,
      "grad_norm": 1.8645654916763306,
      "learning_rate": 2.7476463364715514e-06,
      "loss": 0.7649,
      "step": 14320
    },
    {
      "epoch": 5.87,
      "grad_norm": 2.4538209438323975,
      "learning_rate": 2.7495650839132213e-06,
      "loss": 0.7645,
      "step": 14330
    },
    {
      "epoch": 5.87,
      "grad_norm": 3.2419824600219727,
      "learning_rate": 2.7514838313548917e-06,
      "loss": 0.7552,
      "step": 14340
    },
    {
      "epoch": 5.87,
      "grad_norm": 1.7934889793395996,
      "learning_rate": 2.7534025787965616e-06,
      "loss": 0.7811,
      "step": 14350
    },
    {
      "epoch": 5.88,
      "grad_norm": 2.494438886642456,
      "learning_rate": 2.7553213262382316e-06,
      "loss": 0.779,
      "step": 14360
    },
    {
      "epoch": 5.88,
      "grad_norm": 2.3107337951660156,
      "learning_rate": 2.7572400736799015e-06,
      "loss": 0.7912,
      "step": 14370
    },
    {
      "epoch": 5.89,
      "grad_norm": 2.4750170707702637,
      "learning_rate": 2.759158821121572e-06,
      "loss": 0.7793,
      "step": 14380
    },
    {
      "epoch": 5.89,
      "grad_norm": 2.5577809810638428,
      "learning_rate": 2.761077568563242e-06,
      "loss": 0.7775,
      "step": 14390
    },
    {
      "epoch": 5.89,
      "grad_norm": 2.441610097885132,
      "learning_rate": 2.7629963160049118e-06,
      "loss": 0.7714,
      "step": 14400
    },
    {
      "epoch": 5.9,
      "grad_norm": 2.8030357360839844,
      "learning_rate": 2.764915063446582e-06,
      "loss": 0.7724,
      "step": 14410
    },
    {
      "epoch": 5.9,
      "grad_norm": 2.7506821155548096,
      "learning_rate": 2.766833810888252e-06,
      "loss": 0.7868,
      "step": 14420
    },
    {
      "epoch": 5.91,
      "grad_norm": 2.429654836654663,
      "learning_rate": 2.768752558329922e-06,
      "loss": 0.7851,
      "step": 14430
    },
    {
      "epoch": 5.91,
      "grad_norm": 2.0533480644226074,
      "learning_rate": 2.7706713057715924e-06,
      "loss": 0.776,
      "step": 14440
    },
    {
      "epoch": 5.91,
      "grad_norm": 2.3666677474975586,
      "learning_rate": 2.7725900532132623e-06,
      "loss": 0.7755,
      "step": 14450
    },
    {
      "epoch": 5.92,
      "grad_norm": 1.9106203317642212,
      "learning_rate": 2.7745088006549322e-06,
      "loss": 0.786,
      "step": 14460
    },
    {
      "epoch": 5.92,
      "grad_norm": 2.18278431892395,
      "learning_rate": 2.7764275480966026e-06,
      "loss": 0.7845,
      "step": 14470
    },
    {
      "epoch": 5.93,
      "grad_norm": 2.269351005554199,
      "learning_rate": 2.7783462955382726e-06,
      "loss": 0.7629,
      "step": 14480
    },
    {
      "epoch": 5.93,
      "grad_norm": 2.573800563812256,
      "learning_rate": 2.7802650429799425e-06,
      "loss": 0.7754,
      "step": 14490
    },
    {
      "epoch": 5.94,
      "grad_norm": 2.217215061187744,
      "learning_rate": 2.782183790421613e-06,
      "loss": 0.7737,
      "step": 14500
    },
    {
      "epoch": 5.94,
      "grad_norm": 3.4900832176208496,
      "learning_rate": 2.784102537863283e-06,
      "loss": 0.7571,
      "step": 14510
    },
    {
      "epoch": 5.94,
      "grad_norm": 5.059676170349121,
      "learning_rate": 2.7860212853049527e-06,
      "loss": 0.7817,
      "step": 14520
    },
    {
      "epoch": 5.95,
      "grad_norm": 2.95550799369812,
      "learning_rate": 2.787940032746623e-06,
      "loss": 0.7973,
      "step": 14530
    },
    {
      "epoch": 5.95,
      "grad_norm": 2.690537452697754,
      "learning_rate": 2.789858780188293e-06,
      "loss": 0.7785,
      "step": 14540
    },
    {
      "epoch": 5.96,
      "grad_norm": 3.09035325050354,
      "learning_rate": 2.791777527629963e-06,
      "loss": 0.7814,
      "step": 14550
    },
    {
      "epoch": 5.96,
      "grad_norm": 4.139139652252197,
      "learning_rate": 2.7936962750716333e-06,
      "loss": 0.781,
      "step": 14560
    },
    {
      "epoch": 5.96,
      "grad_norm": 1.9883226156234741,
      "learning_rate": 2.7956150225133033e-06,
      "loss": 0.768,
      "step": 14570
    },
    {
      "epoch": 5.97,
      "grad_norm": 2.2655980587005615,
      "learning_rate": 2.7975337699549732e-06,
      "loss": 0.7779,
      "step": 14580
    },
    {
      "epoch": 5.97,
      "grad_norm": 3.581342935562134,
      "learning_rate": 2.799452517396643e-06,
      "loss": 0.7537,
      "step": 14590
    },
    {
      "epoch": 5.98,
      "grad_norm": 2.2316997051239014,
      "learning_rate": 2.8013712648383135e-06,
      "loss": 0.7745,
      "step": 14600
    },
    {
      "epoch": 5.98,
      "grad_norm": 2.503054141998291,
      "learning_rate": 2.8032900122799835e-06,
      "loss": 0.7741,
      "step": 14610
    },
    {
      "epoch": 5.98,
      "grad_norm": 1.8823162317276,
      "learning_rate": 2.8052087597216534e-06,
      "loss": 0.7876,
      "step": 14620
    },
    {
      "epoch": 5.99,
      "grad_norm": 3.4678351879119873,
      "learning_rate": 2.8071275071633238e-06,
      "loss": 0.7656,
      "step": 14630
    },
    {
      "epoch": 5.99,
      "grad_norm": 2.3853352069854736,
      "learning_rate": 2.8090462546049937e-06,
      "loss": 0.7737,
      "step": 14640
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.567446231842041,
      "learning_rate": 2.8109650020466636e-06,
      "loss": 0.7861,
      "step": 14650
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.777580738067627,
      "eval_runtime": 52.2719,
      "eval_samples_per_second": 65.982,
      "eval_steps_per_second": 8.264,
      "step": 14658
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.717240571975708,
      "learning_rate": 2.812883749488334e-06,
      "loss": 0.7964,
      "step": 14660
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.060931444168091,
      "learning_rate": 2.814802496930004e-06,
      "loss": 0.7661,
      "step": 14670
    },
    {
      "epoch": 6.01,
      "grad_norm": 2.5603959560394287,
      "learning_rate": 2.816721244371674e-06,
      "loss": 0.7775,
      "step": 14680
    },
    {
      "epoch": 6.01,
      "grad_norm": 2.8944742679595947,
      "learning_rate": 2.8186399918133442e-06,
      "loss": 0.7725,
      "step": 14690
    },
    {
      "epoch": 6.02,
      "grad_norm": 2.1897172927856445,
      "learning_rate": 2.820558739255014e-06,
      "loss": 0.7627,
      "step": 14700
    },
    {
      "epoch": 6.02,
      "grad_norm": 1.9863823652267456,
      "learning_rate": 2.822477486696684e-06,
      "loss": 0.7804,
      "step": 14710
    },
    {
      "epoch": 6.03,
      "grad_norm": 2.654351234436035,
      "learning_rate": 2.8243962341383545e-06,
      "loss": 0.7739,
      "step": 14720
    },
    {
      "epoch": 6.03,
      "grad_norm": 2.767806053161621,
      "learning_rate": 2.8263149815800244e-06,
      "loss": 0.7695,
      "step": 14730
    },
    {
      "epoch": 6.03,
      "grad_norm": 2.344273567199707,
      "learning_rate": 2.8282337290216944e-06,
      "loss": 0.7675,
      "step": 14740
    },
    {
      "epoch": 6.04,
      "grad_norm": 3.6826040744781494,
      "learning_rate": 2.8301524764633647e-06,
      "loss": 0.7809,
      "step": 14750
    },
    {
      "epoch": 6.04,
      "grad_norm": 3.1599371433258057,
      "learning_rate": 2.8320712239050347e-06,
      "loss": 0.78,
      "step": 14760
    },
    {
      "epoch": 6.05,
      "grad_norm": 3.2084827423095703,
      "learning_rate": 2.8339899713467046e-06,
      "loss": 0.7605,
      "step": 14770
    },
    {
      "epoch": 6.05,
      "grad_norm": 2.479198932647705,
      "learning_rate": 2.835908718788375e-06,
      "loss": 0.7777,
      "step": 14780
    },
    {
      "epoch": 6.05,
      "grad_norm": 2.8978075981140137,
      "learning_rate": 2.837827466230045e-06,
      "loss": 0.7682,
      "step": 14790
    },
    {
      "epoch": 6.06,
      "grad_norm": 2.548261880874634,
      "learning_rate": 2.839746213671715e-06,
      "loss": 0.7752,
      "step": 14800
    },
    {
      "epoch": 6.06,
      "grad_norm": 3.627479076385498,
      "learning_rate": 2.8416649611133848e-06,
      "loss": 0.776,
      "step": 14810
    },
    {
      "epoch": 6.07,
      "grad_norm": 2.188840627670288,
      "learning_rate": 2.843583708555055e-06,
      "loss": 0.7885,
      "step": 14820
    },
    {
      "epoch": 6.07,
      "grad_norm": 4.024833679199219,
      "learning_rate": 2.845502455996725e-06,
      "loss": 0.7655,
      "step": 14830
    },
    {
      "epoch": 6.07,
      "grad_norm": 1.923360824584961,
      "learning_rate": 2.847421203438395e-06,
      "loss": 0.781,
      "step": 14840
    },
    {
      "epoch": 6.08,
      "grad_norm": 2.8507771492004395,
      "learning_rate": 2.8493399508800654e-06,
      "loss": 0.7693,
      "step": 14850
    },
    {
      "epoch": 6.08,
      "grad_norm": 2.314345359802246,
      "learning_rate": 2.8512586983217353e-06,
      "loss": 0.7629,
      "step": 14860
    },
    {
      "epoch": 6.09,
      "grad_norm": 3.4305214881896973,
      "learning_rate": 2.8531774457634053e-06,
      "loss": 0.7806,
      "step": 14870
    },
    {
      "epoch": 6.09,
      "grad_norm": 2.141120672225952,
      "learning_rate": 2.8550961932050756e-06,
      "loss": 0.7838,
      "step": 14880
    },
    {
      "epoch": 6.09,
      "grad_norm": 2.4338905811309814,
      "learning_rate": 2.8570149406467456e-06,
      "loss": 0.7911,
      "step": 14890
    },
    {
      "epoch": 6.1,
      "grad_norm": 1.9450461864471436,
      "learning_rate": 2.8589336880884155e-06,
      "loss": 0.7655,
      "step": 14900
    },
    {
      "epoch": 6.1,
      "grad_norm": 1.3048748970031738,
      "learning_rate": 2.860852435530086e-06,
      "loss": 0.7615,
      "step": 14910
    },
    {
      "epoch": 6.11,
      "grad_norm": 2.478168249130249,
      "learning_rate": 2.862771182971756e-06,
      "loss": 0.7724,
      "step": 14920
    },
    {
      "epoch": 6.11,
      "grad_norm": 2.618717670440674,
      "learning_rate": 2.8646899304134257e-06,
      "loss": 0.7736,
      "step": 14930
    },
    {
      "epoch": 6.12,
      "grad_norm": 1.9780261516571045,
      "learning_rate": 2.866608677855096e-06,
      "loss": 0.7806,
      "step": 14940
    },
    {
      "epoch": 6.12,
      "grad_norm": 2.6273412704467773,
      "learning_rate": 2.868527425296766e-06,
      "loss": 0.7744,
      "step": 14950
    },
    {
      "epoch": 6.12,
      "grad_norm": 2.834407329559326,
      "learning_rate": 2.870446172738436e-06,
      "loss": 0.7688,
      "step": 14960
    },
    {
      "epoch": 6.13,
      "grad_norm": 2.8161730766296387,
      "learning_rate": 2.8723649201801063e-06,
      "loss": 0.752,
      "step": 14970
    },
    {
      "epoch": 6.13,
      "grad_norm": 2.5402815341949463,
      "learning_rate": 2.8742836676217763e-06,
      "loss": 0.7628,
      "step": 14980
    },
    {
      "epoch": 6.14,
      "grad_norm": 1.8559153079986572,
      "learning_rate": 2.8762024150634462e-06,
      "loss": 0.7805,
      "step": 14990
    },
    {
      "epoch": 6.14,
      "grad_norm": 3.492316246032715,
      "learning_rate": 2.8781211625051166e-06,
      "loss": 0.7553,
      "step": 15000
    },
    {
      "epoch": 6.14,
      "grad_norm": 2.1520566940307617,
      "learning_rate": 2.8800399099467865e-06,
      "loss": 0.785,
      "step": 15010
    },
    {
      "epoch": 6.15,
      "grad_norm": 2.311715841293335,
      "learning_rate": 2.8819586573884565e-06,
      "loss": 0.7836,
      "step": 15020
    },
    {
      "epoch": 6.15,
      "grad_norm": 3.4831624031066895,
      "learning_rate": 2.8838774048301264e-06,
      "loss": 0.7721,
      "step": 15030
    },
    {
      "epoch": 6.16,
      "grad_norm": 2.312450408935547,
      "learning_rate": 2.8857961522717968e-06,
      "loss": 0.7658,
      "step": 15040
    },
    {
      "epoch": 6.16,
      "grad_norm": 2.1205151081085205,
      "learning_rate": 2.8877148997134667e-06,
      "loss": 0.7691,
      "step": 15050
    },
    {
      "epoch": 6.16,
      "grad_norm": 1.9189293384552002,
      "learning_rate": 2.8896336471551366e-06,
      "loss": 0.7757,
      "step": 15060
    },
    {
      "epoch": 6.17,
      "grad_norm": 3.157071113586426,
      "learning_rate": 2.891552394596807e-06,
      "loss": 0.7668,
      "step": 15070
    },
    {
      "epoch": 6.17,
      "grad_norm": 3.328136920928955,
      "learning_rate": 2.893471142038477e-06,
      "loss": 0.7778,
      "step": 15080
    },
    {
      "epoch": 6.18,
      "grad_norm": 2.3976597785949707,
      "learning_rate": 2.895389889480147e-06,
      "loss": 0.7708,
      "step": 15090
    },
    {
      "epoch": 6.18,
      "grad_norm": 2.6935553550720215,
      "learning_rate": 2.8973086369218172e-06,
      "loss": 0.7565,
      "step": 15100
    },
    {
      "epoch": 6.19,
      "grad_norm": 3.265864610671997,
      "learning_rate": 2.899227384363487e-06,
      "loss": 0.7669,
      "step": 15110
    },
    {
      "epoch": 6.19,
      "grad_norm": 2.2624411582946777,
      "learning_rate": 2.901146131805157e-06,
      "loss": 0.7739,
      "step": 15120
    },
    {
      "epoch": 6.19,
      "grad_norm": 2.6436920166015625,
      "learning_rate": 2.9030648792468275e-06,
      "loss": 0.7452,
      "step": 15130
    },
    {
      "epoch": 6.2,
      "grad_norm": 2.2511792182922363,
      "learning_rate": 2.9049836266884974e-06,
      "loss": 0.7969,
      "step": 15140
    },
    {
      "epoch": 6.2,
      "grad_norm": 1.4517951011657715,
      "learning_rate": 2.9069023741301674e-06,
      "loss": 0.7797,
      "step": 15150
    },
    {
      "epoch": 6.21,
      "grad_norm": 2.0579066276550293,
      "learning_rate": 2.9088211215718377e-06,
      "loss": 0.7932,
      "step": 15160
    },
    {
      "epoch": 6.21,
      "grad_norm": 2.466897964477539,
      "learning_rate": 2.9107398690135077e-06,
      "loss": 0.7699,
      "step": 15170
    },
    {
      "epoch": 6.21,
      "grad_norm": 1.7705568075180054,
      "learning_rate": 2.9126586164551776e-06,
      "loss": 0.775,
      "step": 15180
    },
    {
      "epoch": 6.22,
      "grad_norm": 1.6391358375549316,
      "learning_rate": 2.914577363896848e-06,
      "loss": 0.7763,
      "step": 15190
    },
    {
      "epoch": 6.22,
      "grad_norm": 2.3864552974700928,
      "learning_rate": 2.916496111338518e-06,
      "loss": 0.7614,
      "step": 15200
    },
    {
      "epoch": 6.23,
      "grad_norm": 1.8034425973892212,
      "learning_rate": 2.918414858780188e-06,
      "loss": 0.7561,
      "step": 15210
    },
    {
      "epoch": 6.23,
      "grad_norm": 2.114386796951294,
      "learning_rate": 2.920333606221858e-06,
      "loss": 0.7816,
      "step": 15220
    },
    {
      "epoch": 6.23,
      "grad_norm": 3.869657278060913,
      "learning_rate": 2.922252353663528e-06,
      "loss": 0.7698,
      "step": 15230
    },
    {
      "epoch": 6.24,
      "grad_norm": 3.0928056240081787,
      "learning_rate": 2.924171101105198e-06,
      "loss": 0.7673,
      "step": 15240
    },
    {
      "epoch": 6.24,
      "grad_norm": 4.207550525665283,
      "learning_rate": 2.926089848546868e-06,
      "loss": 0.7676,
      "step": 15250
    },
    {
      "epoch": 6.25,
      "grad_norm": 2.09708833694458,
      "learning_rate": 2.9280085959885384e-06,
      "loss": 0.7572,
      "step": 15260
    },
    {
      "epoch": 6.25,
      "grad_norm": 1.9295166730880737,
      "learning_rate": 2.9299273434302083e-06,
      "loss": 0.7691,
      "step": 15270
    },
    {
      "epoch": 6.25,
      "grad_norm": 2.167201042175293,
      "learning_rate": 2.9318460908718783e-06,
      "loss": 0.78,
      "step": 15280
    },
    {
      "epoch": 6.26,
      "grad_norm": 1.543018102645874,
      "learning_rate": 2.9337648383135486e-06,
      "loss": 0.7681,
      "step": 15290
    },
    {
      "epoch": 6.26,
      "grad_norm": 2.1089582443237305,
      "learning_rate": 2.9356835857552186e-06,
      "loss": 0.7751,
      "step": 15300
    },
    {
      "epoch": 6.27,
      "grad_norm": 2.534921407699585,
      "learning_rate": 2.9376023331968885e-06,
      "loss": 0.7623,
      "step": 15310
    },
    {
      "epoch": 6.27,
      "grad_norm": 2.2386105060577393,
      "learning_rate": 2.939521080638559e-06,
      "loss": 0.7804,
      "step": 15320
    },
    {
      "epoch": 6.28,
      "grad_norm": 3.0916101932525635,
      "learning_rate": 2.941439828080229e-06,
      "loss": 0.7666,
      "step": 15330
    },
    {
      "epoch": 6.28,
      "grad_norm": 2.0775389671325684,
      "learning_rate": 2.9433585755218987e-06,
      "loss": 0.7761,
      "step": 15340
    },
    {
      "epoch": 6.28,
      "grad_norm": 2.397599697113037,
      "learning_rate": 2.945277322963569e-06,
      "loss": 0.7654,
      "step": 15350
    },
    {
      "epoch": 6.29,
      "grad_norm": 3.383653163909912,
      "learning_rate": 2.947196070405239e-06,
      "loss": 0.7614,
      "step": 15360
    },
    {
      "epoch": 6.29,
      "grad_norm": 1.7273633480072021,
      "learning_rate": 2.949114817846909e-06,
      "loss": 0.7696,
      "step": 15370
    },
    {
      "epoch": 6.3,
      "grad_norm": 4.318434238433838,
      "learning_rate": 2.9510335652885794e-06,
      "loss": 0.7524,
      "step": 15380
    },
    {
      "epoch": 6.3,
      "grad_norm": 3.79327392578125,
      "learning_rate": 2.9529523127302493e-06,
      "loss": 0.7721,
      "step": 15390
    },
    {
      "epoch": 6.3,
      "grad_norm": 2.8107383251190186,
      "learning_rate": 2.9548710601719192e-06,
      "loss": 0.7829,
      "step": 15400
    },
    {
      "epoch": 6.31,
      "grad_norm": 1.7366441488265991,
      "learning_rate": 2.9567898076135896e-06,
      "loss": 0.7575,
      "step": 15410
    },
    {
      "epoch": 6.31,
      "grad_norm": 2.988767623901367,
      "learning_rate": 2.9587085550552595e-06,
      "loss": 0.7708,
      "step": 15420
    },
    {
      "epoch": 6.32,
      "grad_norm": 2.635784387588501,
      "learning_rate": 2.9606273024969295e-06,
      "loss": 0.7883,
      "step": 15430
    },
    {
      "epoch": 6.32,
      "grad_norm": 1.9691996574401855,
      "learning_rate": 2.9625460499386e-06,
      "loss": 0.7687,
      "step": 15440
    },
    {
      "epoch": 6.32,
      "grad_norm": 3.7135398387908936,
      "learning_rate": 2.9644647973802698e-06,
      "loss": 0.7614,
      "step": 15450
    },
    {
      "epoch": 6.33,
      "grad_norm": 2.489161252975464,
      "learning_rate": 2.9663835448219397e-06,
      "loss": 0.7658,
      "step": 15460
    },
    {
      "epoch": 6.33,
      "grad_norm": 3.5501813888549805,
      "learning_rate": 2.9683022922636096e-06,
      "loss": 0.7669,
      "step": 15470
    },
    {
      "epoch": 6.34,
      "grad_norm": 3.2848498821258545,
      "learning_rate": 2.97022103970528e-06,
      "loss": 0.7656,
      "step": 15480
    },
    {
      "epoch": 6.34,
      "grad_norm": 3.0273361206054688,
      "learning_rate": 2.97213978714695e-06,
      "loss": 0.7763,
      "step": 15490
    },
    {
      "epoch": 6.34,
      "grad_norm": 2.097384452819824,
      "learning_rate": 2.97405853458862e-06,
      "loss": 0.7663,
      "step": 15500
    },
    {
      "epoch": 6.35,
      "grad_norm": 1.9104182720184326,
      "learning_rate": 2.9759772820302903e-06,
      "loss": 0.7641,
      "step": 15510
    },
    {
      "epoch": 6.35,
      "grad_norm": 3.1675984859466553,
      "learning_rate": 2.97789602947196e-06,
      "loss": 0.756,
      "step": 15520
    },
    {
      "epoch": 6.36,
      "grad_norm": 2.803044080734253,
      "learning_rate": 2.97981477691363e-06,
      "loss": 0.7835,
      "step": 15530
    },
    {
      "epoch": 6.36,
      "grad_norm": 1.7402795553207397,
      "learning_rate": 2.9817335243553005e-06,
      "loss": 0.7634,
      "step": 15540
    },
    {
      "epoch": 6.37,
      "grad_norm": 3.7954461574554443,
      "learning_rate": 2.9836522717969704e-06,
      "loss": 0.7817,
      "step": 15550
    },
    {
      "epoch": 6.37,
      "grad_norm": 3.095856189727783,
      "learning_rate": 2.9855710192386404e-06,
      "loss": 0.7615,
      "step": 15560
    },
    {
      "epoch": 6.37,
      "grad_norm": 2.9421355724334717,
      "learning_rate": 2.9874897666803107e-06,
      "loss": 0.781,
      "step": 15570
    },
    {
      "epoch": 6.38,
      "grad_norm": 2.33064603805542,
      "learning_rate": 2.9894085141219807e-06,
      "loss": 0.7759,
      "step": 15580
    },
    {
      "epoch": 6.38,
      "grad_norm": 2.150329828262329,
      "learning_rate": 2.9913272615636506e-06,
      "loss": 0.7568,
      "step": 15590
    },
    {
      "epoch": 6.39,
      "grad_norm": 2.894834280014038,
      "learning_rate": 2.993246009005321e-06,
      "loss": 0.7587,
      "step": 15600
    },
    {
      "epoch": 6.39,
      "grad_norm": 1.8753210306167603,
      "learning_rate": 2.9951647564469913e-06,
      "loss": 0.7474,
      "step": 15610
    },
    {
      "epoch": 6.39,
      "grad_norm": 2.935446262359619,
      "learning_rate": 2.9970835038886613e-06,
      "loss": 0.7624,
      "step": 15620
    },
    {
      "epoch": 6.4,
      "grad_norm": 2.34149169921875,
      "learning_rate": 2.9990022513303316e-06,
      "loss": 0.7893,
      "step": 15630
    },
    {
      "epoch": 6.4,
      "grad_norm": 1.8908036947250366,
      "learning_rate": 3.0009209987720016e-06,
      "loss": 0.7656,
      "step": 15640
    },
    {
      "epoch": 6.41,
      "grad_norm": 2.286386728286743,
      "learning_rate": 3.0028397462136715e-06,
      "loss": 0.7638,
      "step": 15650
    },
    {
      "epoch": 6.41,
      "grad_norm": 2.718564510345459,
      "learning_rate": 3.004758493655342e-06,
      "loss": 0.7451,
      "step": 15660
    },
    {
      "epoch": 6.41,
      "grad_norm": 3.3132286071777344,
      "learning_rate": 3.006677241097012e-06,
      "loss": 0.7461,
      "step": 15670
    },
    {
      "epoch": 6.42,
      "grad_norm": 3.327641248703003,
      "learning_rate": 3.0085959885386818e-06,
      "loss": 0.7676,
      "step": 15680
    },
    {
      "epoch": 6.42,
      "grad_norm": 1.8121187686920166,
      "learning_rate": 3.010514735980352e-06,
      "loss": 0.7571,
      "step": 15690
    },
    {
      "epoch": 6.43,
      "grad_norm": 3.380042791366577,
      "learning_rate": 3.012433483422022e-06,
      "loss": 0.7937,
      "step": 15700
    },
    {
      "epoch": 6.43,
      "grad_norm": 2.5139613151550293,
      "learning_rate": 3.014352230863692e-06,
      "loss": 0.7533,
      "step": 15710
    },
    {
      "epoch": 6.43,
      "grad_norm": 3.270634174346924,
      "learning_rate": 3.0162709783053624e-06,
      "loss": 0.7621,
      "step": 15720
    },
    {
      "epoch": 6.44,
      "grad_norm": 2.105031728744507,
      "learning_rate": 3.0181897257470323e-06,
      "loss": 0.7739,
      "step": 15730
    },
    {
      "epoch": 6.44,
      "grad_norm": 2.112600088119507,
      "learning_rate": 3.0201084731887022e-06,
      "loss": 0.7739,
      "step": 15740
    },
    {
      "epoch": 6.45,
      "grad_norm": 2.2988336086273193,
      "learning_rate": 3.0220272206303726e-06,
      "loss": 0.7541,
      "step": 15750
    },
    {
      "epoch": 6.45,
      "grad_norm": 3.5293619632720947,
      "learning_rate": 3.0239459680720425e-06,
      "loss": 0.7647,
      "step": 15760
    },
    {
      "epoch": 6.46,
      "grad_norm": 2.096374273300171,
      "learning_rate": 3.0258647155137125e-06,
      "loss": 0.7457,
      "step": 15770
    },
    {
      "epoch": 6.46,
      "grad_norm": 4.635911464691162,
      "learning_rate": 3.027783462955383e-06,
      "loss": 0.7562,
      "step": 15780
    },
    {
      "epoch": 6.46,
      "grad_norm": 2.096519947052002,
      "learning_rate": 3.0297022103970528e-06,
      "loss": 0.7675,
      "step": 15790
    },
    {
      "epoch": 6.47,
      "grad_norm": 1.583013892173767,
      "learning_rate": 3.0316209578387227e-06,
      "loss": 0.7772,
      "step": 15800
    },
    {
      "epoch": 6.47,
      "grad_norm": 3.27933669090271,
      "learning_rate": 3.033539705280393e-06,
      "loss": 0.7411,
      "step": 15810
    },
    {
      "epoch": 6.48,
      "grad_norm": 2.8124866485595703,
      "learning_rate": 3.035458452722063e-06,
      "loss": 0.7626,
      "step": 15820
    },
    {
      "epoch": 6.48,
      "grad_norm": 2.8983452320098877,
      "learning_rate": 3.037377200163733e-06,
      "loss": 0.7512,
      "step": 15830
    },
    {
      "epoch": 6.48,
      "grad_norm": 2.3934171199798584,
      "learning_rate": 3.039295947605403e-06,
      "loss": 0.7604,
      "step": 15840
    },
    {
      "epoch": 6.49,
      "grad_norm": 3.1153371334075928,
      "learning_rate": 3.0412146950470733e-06,
      "loss": 0.7657,
      "step": 15850
    },
    {
      "epoch": 6.49,
      "grad_norm": 1.8334221839904785,
      "learning_rate": 3.043133442488743e-06,
      "loss": 0.7602,
      "step": 15860
    },
    {
      "epoch": 6.5,
      "grad_norm": 4.355269432067871,
      "learning_rate": 3.045052189930413e-06,
      "loss": 0.7549,
      "step": 15870
    },
    {
      "epoch": 6.5,
      "grad_norm": 3.143521547317505,
      "learning_rate": 3.0469709373720835e-06,
      "loss": 0.7614,
      "step": 15880
    },
    {
      "epoch": 6.5,
      "grad_norm": 3.0166854858398438,
      "learning_rate": 3.0488896848137534e-06,
      "loss": 0.7588,
      "step": 15890
    },
    {
      "epoch": 6.51,
      "grad_norm": 2.043825387954712,
      "learning_rate": 3.0508084322554234e-06,
      "loss": 0.7725,
      "step": 15900
    },
    {
      "epoch": 6.51,
      "grad_norm": 3.619720220565796,
      "learning_rate": 3.0527271796970937e-06,
      "loss": 0.7498,
      "step": 15910
    },
    {
      "epoch": 6.52,
      "grad_norm": 3.08268141746521,
      "learning_rate": 3.0546459271387637e-06,
      "loss": 0.756,
      "step": 15920
    },
    {
      "epoch": 6.52,
      "grad_norm": 2.4415392875671387,
      "learning_rate": 3.0565646745804336e-06,
      "loss": 0.7663,
      "step": 15930
    },
    {
      "epoch": 6.52,
      "grad_norm": 2.1758649349212646,
      "learning_rate": 3.058483422022104e-06,
      "loss": 0.7543,
      "step": 15940
    },
    {
      "epoch": 6.53,
      "grad_norm": 3.5335183143615723,
      "learning_rate": 3.060402169463774e-06,
      "loss": 0.7568,
      "step": 15950
    },
    {
      "epoch": 6.53,
      "grad_norm": 2.287285089492798,
      "learning_rate": 3.062320916905444e-06,
      "loss": 0.7592,
      "step": 15960
    },
    {
      "epoch": 6.54,
      "grad_norm": 2.962360143661499,
      "learning_rate": 3.0642396643471142e-06,
      "loss": 0.7495,
      "step": 15970
    },
    {
      "epoch": 6.54,
      "grad_norm": 3.042888641357422,
      "learning_rate": 3.066158411788784e-06,
      "loss": 0.7527,
      "step": 15980
    },
    {
      "epoch": 6.55,
      "grad_norm": 2.9540188312530518,
      "learning_rate": 3.068077159230454e-06,
      "loss": 0.7636,
      "step": 15990
    },
    {
      "epoch": 6.55,
      "grad_norm": 2.4094038009643555,
      "learning_rate": 3.0699959066721245e-06,
      "loss": 0.761,
      "step": 16000
    },
    {
      "epoch": 6.55,
      "grad_norm": 3.8240692615509033,
      "learning_rate": 3.0719146541137944e-06,
      "loss": 0.7458,
      "step": 16010
    },
    {
      "epoch": 6.56,
      "grad_norm": 3.1179778575897217,
      "learning_rate": 3.0738334015554643e-06,
      "loss": 0.7512,
      "step": 16020
    },
    {
      "epoch": 6.56,
      "grad_norm": 3.1762168407440186,
      "learning_rate": 3.0757521489971347e-06,
      "loss": 0.7693,
      "step": 16030
    },
    {
      "epoch": 6.57,
      "grad_norm": 3.2232062816619873,
      "learning_rate": 3.0776708964388046e-06,
      "loss": 0.7584,
      "step": 16040
    },
    {
      "epoch": 6.57,
      "grad_norm": 3.539438486099243,
      "learning_rate": 3.0795896438804746e-06,
      "loss": 0.7567,
      "step": 16050
    },
    {
      "epoch": 6.57,
      "grad_norm": 2.6698906421661377,
      "learning_rate": 3.0815083913221445e-06,
      "loss": 0.7549,
      "step": 16060
    },
    {
      "epoch": 6.58,
      "grad_norm": 3.297420024871826,
      "learning_rate": 3.083427138763815e-06,
      "loss": 0.7745,
      "step": 16070
    },
    {
      "epoch": 6.58,
      "grad_norm": 2.138915538787842,
      "learning_rate": 3.085345886205485e-06,
      "loss": 0.7426,
      "step": 16080
    },
    {
      "epoch": 6.59,
      "grad_norm": 2.1028149127960205,
      "learning_rate": 3.0872646336471548e-06,
      "loss": 0.768,
      "step": 16090
    },
    {
      "epoch": 6.59,
      "grad_norm": 3.183014392852783,
      "learning_rate": 3.089183381088825e-06,
      "loss": 0.7648,
      "step": 16100
    },
    {
      "epoch": 6.59,
      "grad_norm": 2.1172757148742676,
      "learning_rate": 3.091102128530495e-06,
      "loss": 0.7559,
      "step": 16110
    },
    {
      "epoch": 6.6,
      "grad_norm": 2.2746706008911133,
      "learning_rate": 3.093020875972165e-06,
      "loss": 0.7661,
      "step": 16120
    },
    {
      "epoch": 6.6,
      "grad_norm": 2.4887850284576416,
      "learning_rate": 3.0949396234138354e-06,
      "loss": 0.7592,
      "step": 16130
    },
    {
      "epoch": 6.61,
      "grad_norm": 2.568481922149658,
      "learning_rate": 3.0968583708555053e-06,
      "loss": 0.7588,
      "step": 16140
    },
    {
      "epoch": 6.61,
      "grad_norm": 2.731459140777588,
      "learning_rate": 3.0987771182971752e-06,
      "loss": 0.7585,
      "step": 16150
    },
    {
      "epoch": 6.61,
      "grad_norm": 2.2372260093688965,
      "learning_rate": 3.1006958657388456e-06,
      "loss": 0.7615,
      "step": 16160
    },
    {
      "epoch": 6.62,
      "grad_norm": 2.2858192920684814,
      "learning_rate": 3.1026146131805156e-06,
      "loss": 0.7627,
      "step": 16170
    },
    {
      "epoch": 6.62,
      "grad_norm": 2.2722463607788086,
      "learning_rate": 3.1045333606221855e-06,
      "loss": 0.7554,
      "step": 16180
    },
    {
      "epoch": 6.63,
      "grad_norm": 2.2383816242218018,
      "learning_rate": 3.106452108063856e-06,
      "loss": 0.7743,
      "step": 16190
    },
    {
      "epoch": 6.63,
      "grad_norm": 4.011078834533691,
      "learning_rate": 3.108370855505526e-06,
      "loss": 0.7562,
      "step": 16200
    },
    {
      "epoch": 6.64,
      "grad_norm": 2.152545690536499,
      "learning_rate": 3.1102896029471957e-06,
      "loss": 0.7576,
      "step": 16210
    },
    {
      "epoch": 6.64,
      "grad_norm": 2.2871713638305664,
      "learning_rate": 3.112208350388866e-06,
      "loss": 0.7538,
      "step": 16220
    },
    {
      "epoch": 6.64,
      "grad_norm": 4.34039306640625,
      "learning_rate": 3.114127097830536e-06,
      "loss": 0.7561,
      "step": 16230
    },
    {
      "epoch": 6.65,
      "grad_norm": 3.535323143005371,
      "learning_rate": 3.116045845272206e-06,
      "loss": 0.7687,
      "step": 16240
    },
    {
      "epoch": 6.65,
      "grad_norm": 3.2254700660705566,
      "learning_rate": 3.1179645927138763e-06,
      "loss": 0.7648,
      "step": 16250
    },
    {
      "epoch": 6.66,
      "grad_norm": 2.86507511138916,
      "learning_rate": 3.1198833401555463e-06,
      "loss": 0.7433,
      "step": 16260
    },
    {
      "epoch": 6.66,
      "grad_norm": 2.031582832336426,
      "learning_rate": 3.1218020875972162e-06,
      "loss": 0.7543,
      "step": 16270
    },
    {
      "epoch": 6.66,
      "grad_norm": 3.5594401359558105,
      "learning_rate": 3.123720835038886e-06,
      "loss": 0.7763,
      "step": 16280
    },
    {
      "epoch": 6.67,
      "grad_norm": 2.9394099712371826,
      "learning_rate": 3.1256395824805565e-06,
      "loss": 0.7635,
      "step": 16290
    },
    {
      "epoch": 6.67,
      "grad_norm": 2.6670889854431152,
      "learning_rate": 3.1275583299222265e-06,
      "loss": 0.7648,
      "step": 16300
    },
    {
      "epoch": 6.68,
      "grad_norm": 2.4579999446868896,
      "learning_rate": 3.1294770773638964e-06,
      "loss": 0.7601,
      "step": 16310
    },
    {
      "epoch": 6.68,
      "grad_norm": 1.8945528268814087,
      "learning_rate": 3.1313958248055668e-06,
      "loss": 0.7743,
      "step": 16320
    },
    {
      "epoch": 6.68,
      "grad_norm": 2.753787040710449,
      "learning_rate": 3.1333145722472367e-06,
      "loss": 0.7432,
      "step": 16330
    },
    {
      "epoch": 6.69,
      "grad_norm": 2.147348642349243,
      "learning_rate": 3.1352333196889066e-06,
      "loss": 0.7592,
      "step": 16340
    },
    {
      "epoch": 6.69,
      "grad_norm": 2.693119764328003,
      "learning_rate": 3.137152067130577e-06,
      "loss": 0.7481,
      "step": 16350
    },
    {
      "epoch": 6.7,
      "grad_norm": 2.2184321880340576,
      "learning_rate": 3.139070814572247e-06,
      "loss": 0.7407,
      "step": 16360
    },
    {
      "epoch": 6.7,
      "grad_norm": 2.6956818103790283,
      "learning_rate": 3.140989562013917e-06,
      "loss": 0.7561,
      "step": 16370
    },
    {
      "epoch": 6.7,
      "grad_norm": 2.252462863922119,
      "learning_rate": 3.1429083094555872e-06,
      "loss": 0.7307,
      "step": 16380
    },
    {
      "epoch": 6.71,
      "grad_norm": 2.601914644241333,
      "learning_rate": 3.144827056897257e-06,
      "loss": 0.7563,
      "step": 16390
    },
    {
      "epoch": 6.71,
      "grad_norm": 2.240199089050293,
      "learning_rate": 3.146745804338927e-06,
      "loss": 0.7669,
      "step": 16400
    },
    {
      "epoch": 6.72,
      "grad_norm": 2.209289312362671,
      "learning_rate": 3.1486645517805975e-06,
      "loss": 0.7591,
      "step": 16410
    },
    {
      "epoch": 6.72,
      "grad_norm": 1.688342809677124,
      "learning_rate": 3.1505832992222674e-06,
      "loss": 0.7674,
      "step": 16420
    },
    {
      "epoch": 6.73,
      "grad_norm": 2.917839765548706,
      "learning_rate": 3.1525020466639374e-06,
      "loss": 0.756,
      "step": 16430
    },
    {
      "epoch": 6.73,
      "grad_norm": 3.350118398666382,
      "learning_rate": 3.1544207941056077e-06,
      "loss": 0.7584,
      "step": 16440
    },
    {
      "epoch": 6.73,
      "grad_norm": 2.959583044052124,
      "learning_rate": 3.1563395415472777e-06,
      "loss": 0.775,
      "step": 16450
    },
    {
      "epoch": 6.74,
      "grad_norm": 2.625608205795288,
      "learning_rate": 3.1582582889889476e-06,
      "loss": 0.7629,
      "step": 16460
    },
    {
      "epoch": 6.74,
      "grad_norm": 1.9467147588729858,
      "learning_rate": 3.1601770364306175e-06,
      "loss": 0.7556,
      "step": 16470
    },
    {
      "epoch": 6.75,
      "grad_norm": 3.7482120990753174,
      "learning_rate": 3.162095783872288e-06,
      "loss": 0.7514,
      "step": 16480
    },
    {
      "epoch": 6.75,
      "grad_norm": 2.689896583557129,
      "learning_rate": 3.164014531313958e-06,
      "loss": 0.7439,
      "step": 16490
    },
    {
      "epoch": 6.75,
      "grad_norm": 3.250370502471924,
      "learning_rate": 3.1659332787556278e-06,
      "loss": 0.7247,
      "step": 16500
    },
    {
      "epoch": 6.76,
      "grad_norm": 2.246769666671753,
      "learning_rate": 3.167852026197298e-06,
      "loss": 0.7569,
      "step": 16510
    },
    {
      "epoch": 6.76,
      "grad_norm": 2.0491251945495605,
      "learning_rate": 3.169770773638968e-06,
      "loss": 0.7632,
      "step": 16520
    },
    {
      "epoch": 6.77,
      "grad_norm": 2.2286932468414307,
      "learning_rate": 3.171689521080638e-06,
      "loss": 0.7507,
      "step": 16530
    },
    {
      "epoch": 6.77,
      "grad_norm": 2.499175786972046,
      "learning_rate": 3.1736082685223084e-06,
      "loss": 0.7678,
      "step": 16540
    },
    {
      "epoch": 6.77,
      "grad_norm": 2.311312437057495,
      "learning_rate": 3.1755270159639783e-06,
      "loss": 0.7532,
      "step": 16550
    },
    {
      "epoch": 6.78,
      "grad_norm": 2.996553421020508,
      "learning_rate": 3.1774457634056483e-06,
      "loss": 0.75,
      "step": 16560
    },
    {
      "epoch": 6.78,
      "grad_norm": 2.033567190170288,
      "learning_rate": 3.1793645108473186e-06,
      "loss": 0.7525,
      "step": 16570
    },
    {
      "epoch": 6.79,
      "grad_norm": 2.1622135639190674,
      "learning_rate": 3.1812832582889886e-06,
      "loss": 0.7497,
      "step": 16580
    },
    {
      "epoch": 6.79,
      "grad_norm": 3.009873628616333,
      "learning_rate": 3.1832020057306585e-06,
      "loss": 0.7657,
      "step": 16590
    },
    {
      "epoch": 6.79,
      "grad_norm": 2.0477192401885986,
      "learning_rate": 3.185120753172329e-06,
      "loss": 0.7753,
      "step": 16600
    },
    {
      "epoch": 6.8,
      "grad_norm": 2.996192216873169,
      "learning_rate": 3.187039500613999e-06,
      "loss": 0.7743,
      "step": 16610
    },
    {
      "epoch": 6.8,
      "grad_norm": 3.331855297088623,
      "learning_rate": 3.1889582480556687e-06,
      "loss": 0.7524,
      "step": 16620
    },
    {
      "epoch": 6.81,
      "grad_norm": 2.355739116668701,
      "learning_rate": 3.190876995497339e-06,
      "loss": 0.7413,
      "step": 16630
    },
    {
      "epoch": 6.81,
      "grad_norm": 4.056034564971924,
      "learning_rate": 3.192795742939009e-06,
      "loss": 0.7452,
      "step": 16640
    },
    {
      "epoch": 6.82,
      "grad_norm": 3.1106748580932617,
      "learning_rate": 3.194714490380679e-06,
      "loss": 0.7542,
      "step": 16650
    },
    {
      "epoch": 6.82,
      "grad_norm": 2.058023691177368,
      "learning_rate": 3.1966332378223493e-06,
      "loss": 0.7517,
      "step": 16660
    },
    {
      "epoch": 6.82,
      "grad_norm": 1.9937883615493774,
      "learning_rate": 3.1985519852640193e-06,
      "loss": 0.7459,
      "step": 16670
    },
    {
      "epoch": 6.83,
      "grad_norm": 5.156467437744141,
      "learning_rate": 3.2004707327056892e-06,
      "loss": 0.7585,
      "step": 16680
    },
    {
      "epoch": 6.83,
      "grad_norm": 2.1261539459228516,
      "learning_rate": 3.202389480147359e-06,
      "loss": 0.7583,
      "step": 16690
    },
    {
      "epoch": 6.84,
      "grad_norm": 2.5210392475128174,
      "learning_rate": 3.2043082275890295e-06,
      "loss": 0.7535,
      "step": 16700
    },
    {
      "epoch": 6.84,
      "grad_norm": 3.665950059890747,
      "learning_rate": 3.2062269750306995e-06,
      "loss": 0.7441,
      "step": 16710
    },
    {
      "epoch": 6.84,
      "grad_norm": 2.3136425018310547,
      "learning_rate": 3.2081457224723694e-06,
      "loss": 0.7512,
      "step": 16720
    },
    {
      "epoch": 6.85,
      "grad_norm": 1.8373973369598389,
      "learning_rate": 3.2100644699140398e-06,
      "loss": 0.7557,
      "step": 16730
    },
    {
      "epoch": 6.85,
      "grad_norm": 2.6783833503723145,
      "learning_rate": 3.2119832173557097e-06,
      "loss": 0.7435,
      "step": 16740
    },
    {
      "epoch": 6.86,
      "grad_norm": 2.5607290267944336,
      "learning_rate": 3.2139019647973796e-06,
      "loss": 0.757,
      "step": 16750
    },
    {
      "epoch": 6.86,
      "grad_norm": 3.817046642303467,
      "learning_rate": 3.21582071223905e-06,
      "loss": 0.7608,
      "step": 16760
    },
    {
      "epoch": 6.86,
      "grad_norm": 2.553675413131714,
      "learning_rate": 3.21773945968072e-06,
      "loss": 0.7495,
      "step": 16770
    },
    {
      "epoch": 6.87,
      "grad_norm": 4.220304012298584,
      "learning_rate": 3.21965820712239e-06,
      "loss": 0.7447,
      "step": 16780
    },
    {
      "epoch": 6.87,
      "grad_norm": 2.270993709564209,
      "learning_rate": 3.2215769545640602e-06,
      "loss": 0.752,
      "step": 16790
    },
    {
      "epoch": 6.88,
      "grad_norm": 2.0005345344543457,
      "learning_rate": 3.22349570200573e-06,
      "loss": 0.7668,
      "step": 16800
    },
    {
      "epoch": 6.88,
      "grad_norm": 2.2027926445007324,
      "learning_rate": 3.2254144494474e-06,
      "loss": 0.7576,
      "step": 16810
    },
    {
      "epoch": 6.88,
      "grad_norm": 3.0597970485687256,
      "learning_rate": 3.2273331968890705e-06,
      "loss": 0.7347,
      "step": 16820
    },
    {
      "epoch": 6.89,
      "grad_norm": 1.8922749757766724,
      "learning_rate": 3.2292519443307404e-06,
      "loss": 0.7456,
      "step": 16830
    },
    {
      "epoch": 6.89,
      "grad_norm": 3.5203187465667725,
      "learning_rate": 3.2311706917724104e-06,
      "loss": 0.7435,
      "step": 16840
    },
    {
      "epoch": 6.9,
      "grad_norm": 4.061552047729492,
      "learning_rate": 3.2330894392140807e-06,
      "loss": 0.7604,
      "step": 16850
    },
    {
      "epoch": 6.9,
      "grad_norm": 2.155548572540283,
      "learning_rate": 3.2350081866557507e-06,
      "loss": 0.7656,
      "step": 16860
    },
    {
      "epoch": 6.91,
      "grad_norm": 3.6593382358551025,
      "learning_rate": 3.2369269340974206e-06,
      "loss": 0.7509,
      "step": 16870
    },
    {
      "epoch": 6.91,
      "grad_norm": 2.7485640048980713,
      "learning_rate": 3.238845681539091e-06,
      "loss": 0.7534,
      "step": 16880
    },
    {
      "epoch": 6.91,
      "grad_norm": 2.654322624206543,
      "learning_rate": 3.240764428980761e-06,
      "loss": 0.7627,
      "step": 16890
    },
    {
      "epoch": 6.92,
      "grad_norm": 2.5452237129211426,
      "learning_rate": 3.242683176422431e-06,
      "loss": 0.7715,
      "step": 16900
    },
    {
      "epoch": 6.92,
      "grad_norm": 2.1424388885498047,
      "learning_rate": 3.2446019238641008e-06,
      "loss": 0.7614,
      "step": 16910
    },
    {
      "epoch": 6.93,
      "grad_norm": 2.392946481704712,
      "learning_rate": 3.246520671305771e-06,
      "loss": 0.7703,
      "step": 16920
    },
    {
      "epoch": 6.93,
      "grad_norm": 1.8647314310073853,
      "learning_rate": 3.248439418747441e-06,
      "loss": 0.749,
      "step": 16930
    },
    {
      "epoch": 6.93,
      "grad_norm": 2.218024730682373,
      "learning_rate": 3.250358166189111e-06,
      "loss": 0.7422,
      "step": 16940
    },
    {
      "epoch": 6.94,
      "grad_norm": 3.145475149154663,
      "learning_rate": 3.2522769136307814e-06,
      "loss": 0.75,
      "step": 16950
    },
    {
      "epoch": 6.94,
      "grad_norm": 3.508521795272827,
      "learning_rate": 3.2541956610724513e-06,
      "loss": 0.7689,
      "step": 16960
    },
    {
      "epoch": 6.95,
      "grad_norm": 2.5453622341156006,
      "learning_rate": 3.256114408514122e-06,
      "loss": 0.7516,
      "step": 16970
    },
    {
      "epoch": 6.95,
      "grad_norm": 3.751232624053955,
      "learning_rate": 3.258033155955792e-06,
      "loss": 0.7498,
      "step": 16980
    },
    {
      "epoch": 6.95,
      "grad_norm": 1.9610122442245483,
      "learning_rate": 3.259951903397462e-06,
      "loss": 0.7546,
      "step": 16990
    },
    {
      "epoch": 6.96,
      "grad_norm": 2.81490421295166,
      "learning_rate": 3.2618706508391324e-06,
      "loss": 0.745,
      "step": 17000
    },
    {
      "epoch": 6.96,
      "grad_norm": 1.9492006301879883,
      "learning_rate": 3.2637893982808023e-06,
      "loss": 0.7561,
      "step": 17010
    },
    {
      "epoch": 6.97,
      "grad_norm": 2.0786566734313965,
      "learning_rate": 3.2657081457224722e-06,
      "loss": 0.7405,
      "step": 17020
    },
    {
      "epoch": 6.97,
      "grad_norm": 2.0161635875701904,
      "learning_rate": 3.2676268931641426e-06,
      "loss": 0.7576,
      "step": 17030
    },
    {
      "epoch": 6.98,
      "grad_norm": 2.5451653003692627,
      "learning_rate": 3.2695456406058125e-06,
      "loss": 0.7327,
      "step": 17040
    },
    {
      "epoch": 6.98,
      "grad_norm": 2.1772642135620117,
      "learning_rate": 3.2714643880474825e-06,
      "loss": 0.7456,
      "step": 17050
    },
    {
      "epoch": 6.98,
      "grad_norm": 3.1234772205352783,
      "learning_rate": 3.2733831354891524e-06,
      "loss": 0.7565,
      "step": 17060
    },
    {
      "epoch": 6.99,
      "grad_norm": 3.409940719604492,
      "learning_rate": 3.2753018829308228e-06,
      "loss": 0.7515,
      "step": 17070
    },
    {
      "epoch": 6.99,
      "grad_norm": 3.215548276901245,
      "learning_rate": 3.2772206303724927e-06,
      "loss": 0.7558,
      "step": 17080
    },
    {
      "epoch": 7.0,
      "grad_norm": 3.8236539363861084,
      "learning_rate": 3.2791393778141627e-06,
      "loss": 0.7473,
      "step": 17090
    },
    {
      "epoch": 7.0,
      "grad_norm": 3.3544819355010986,
      "learning_rate": 3.281058125255833e-06,
      "loss": 0.7665,
      "step": 17100
    },
    {
      "epoch": 7.0,
      "eval_loss": 0.7557178139686584,
      "eval_runtime": 52.8665,
      "eval_samples_per_second": 65.24,
      "eval_steps_per_second": 8.172,
      "step": 17101
    },
    {
      "epoch": 7.0,
      "grad_norm": 2.1801183223724365,
      "learning_rate": 3.282976872697503e-06,
      "loss": 0.7362,
      "step": 17110
    },
    {
      "epoch": 7.01,
      "grad_norm": 2.3891899585723877,
      "learning_rate": 3.284895620139173e-06,
      "loss": 0.7641,
      "step": 17120
    },
    {
      "epoch": 7.01,
      "grad_norm": 2.3542580604553223,
      "learning_rate": 3.2868143675808433e-06,
      "loss": 0.7617,
      "step": 17130
    },
    {
      "epoch": 7.02,
      "grad_norm": 1.5127707719802856,
      "learning_rate": 3.288733115022513e-06,
      "loss": 0.774,
      "step": 17140
    },
    {
      "epoch": 7.02,
      "grad_norm": 2.3853108882904053,
      "learning_rate": 3.290651862464183e-06,
      "loss": 0.763,
      "step": 17150
    },
    {
      "epoch": 7.02,
      "grad_norm": 3.0682082176208496,
      "learning_rate": 3.2925706099058535e-06,
      "loss": 0.7364,
      "step": 17160
    },
    {
      "epoch": 7.03,
      "grad_norm": 1.801768183708191,
      "learning_rate": 3.2944893573475234e-06,
      "loss": 0.7441,
      "step": 17170
    },
    {
      "epoch": 7.03,
      "grad_norm": 2.923715353012085,
      "learning_rate": 3.2964081047891934e-06,
      "loss": 0.7622,
      "step": 17180
    },
    {
      "epoch": 7.04,
      "grad_norm": 1.4154754877090454,
      "learning_rate": 3.2983268522308637e-06,
      "loss": 0.7554,
      "step": 17190
    },
    {
      "epoch": 7.04,
      "grad_norm": 1.6102502346038818,
      "learning_rate": 3.3002455996725337e-06,
      "loss": 0.7532,
      "step": 17200
    },
    {
      "epoch": 7.04,
      "grad_norm": 2.584019660949707,
      "learning_rate": 3.3021643471142036e-06,
      "loss": 0.7455,
      "step": 17210
    },
    {
      "epoch": 7.05,
      "grad_norm": 2.7486348152160645,
      "learning_rate": 3.304083094555874e-06,
      "loss": 0.7619,
      "step": 17220
    },
    {
      "epoch": 7.05,
      "grad_norm": 3.118945360183716,
      "learning_rate": 3.306001841997544e-06,
      "loss": 0.7283,
      "step": 17230
    },
    {
      "epoch": 7.06,
      "grad_norm": 2.4034273624420166,
      "learning_rate": 3.307920589439214e-06,
      "loss": 0.748,
      "step": 17240
    },
    {
      "epoch": 7.06,
      "grad_norm": 2.044762134552002,
      "learning_rate": 3.3098393368808842e-06,
      "loss": 0.7435,
      "step": 17250
    },
    {
      "epoch": 7.07,
      "grad_norm": 2.276139259338379,
      "learning_rate": 3.311758084322554e-06,
      "loss": 0.7265,
      "step": 17260
    },
    {
      "epoch": 7.07,
      "grad_norm": 3.4864351749420166,
      "learning_rate": 3.313676831764224e-06,
      "loss": 0.7452,
      "step": 17270
    },
    {
      "epoch": 7.07,
      "grad_norm": 1.8422060012817383,
      "learning_rate": 3.315595579205894e-06,
      "loss": 0.7536,
      "step": 17280
    },
    {
      "epoch": 7.08,
      "grad_norm": 2.111621141433716,
      "learning_rate": 3.3175143266475644e-06,
      "loss": 0.7544,
      "step": 17290
    },
    {
      "epoch": 7.08,
      "grad_norm": 3.523524761199951,
      "learning_rate": 3.3194330740892343e-06,
      "loss": 0.7455,
      "step": 17300
    },
    {
      "epoch": 7.09,
      "grad_norm": 3.214712619781494,
      "learning_rate": 3.3213518215309043e-06,
      "loss": 0.756,
      "step": 17310
    },
    {
      "epoch": 7.09,
      "grad_norm": 2.6165661811828613,
      "learning_rate": 3.3232705689725746e-06,
      "loss": 0.7522,
      "step": 17320
    },
    {
      "epoch": 7.09,
      "grad_norm": 2.3321261405944824,
      "learning_rate": 3.3251893164142446e-06,
      "loss": 0.7459,
      "step": 17330
    },
    {
      "epoch": 7.1,
      "grad_norm": 2.5449564456939697,
      "learning_rate": 3.3271080638559145e-06,
      "loss": 0.7487,
      "step": 17340
    },
    {
      "epoch": 7.1,
      "grad_norm": 2.4104113578796387,
      "learning_rate": 3.329026811297585e-06,
      "loss": 0.7544,
      "step": 17350
    },
    {
      "epoch": 7.11,
      "grad_norm": 3.357416868209839,
      "learning_rate": 3.330945558739255e-06,
      "loss": 0.7447,
      "step": 17360
    },
    {
      "epoch": 7.11,
      "grad_norm": 2.1500532627105713,
      "learning_rate": 3.3328643061809248e-06,
      "loss": 0.7706,
      "step": 17370
    },
    {
      "epoch": 7.11,
      "grad_norm": 2.7865288257598877,
      "learning_rate": 3.334783053622595e-06,
      "loss": 0.7536,
      "step": 17380
    },
    {
      "epoch": 7.12,
      "grad_norm": 1.5741996765136719,
      "learning_rate": 3.336701801064265e-06,
      "loss": 0.7442,
      "step": 17390
    },
    {
      "epoch": 7.12,
      "grad_norm": 2.6656205654144287,
      "learning_rate": 3.338620548505935e-06,
      "loss": 0.7492,
      "step": 17400
    },
    {
      "epoch": 7.13,
      "grad_norm": 2.390498161315918,
      "learning_rate": 3.3405392959476054e-06,
      "loss": 0.7523,
      "step": 17410
    },
    {
      "epoch": 7.13,
      "grad_norm": 2.911205291748047,
      "learning_rate": 3.3424580433892753e-06,
      "loss": 0.7629,
      "step": 17420
    },
    {
      "epoch": 7.13,
      "grad_norm": 2.911555290222168,
      "learning_rate": 3.3443767908309452e-06,
      "loss": 0.7566,
      "step": 17430
    },
    {
      "epoch": 7.14,
      "grad_norm": 2.317328929901123,
      "learning_rate": 3.3462955382726156e-06,
      "loss": 0.7407,
      "step": 17440
    },
    {
      "epoch": 7.14,
      "grad_norm": 2.837859869003296,
      "learning_rate": 3.3482142857142855e-06,
      "loss": 0.7476,
      "step": 17450
    },
    {
      "epoch": 7.15,
      "grad_norm": 2.2656006813049316,
      "learning_rate": 3.3501330331559555e-06,
      "loss": 0.756,
      "step": 17460
    },
    {
      "epoch": 7.15,
      "grad_norm": 1.8521032333374023,
      "learning_rate": 3.352051780597626e-06,
      "loss": 0.7539,
      "step": 17470
    },
    {
      "epoch": 7.16,
      "grad_norm": 3.109588384628296,
      "learning_rate": 3.3539705280392958e-06,
      "loss": 0.7596,
      "step": 17480
    },
    {
      "epoch": 7.16,
      "grad_norm": 2.4375548362731934,
      "learning_rate": 3.3558892754809657e-06,
      "loss": 0.7494,
      "step": 17490
    },
    {
      "epoch": 7.16,
      "grad_norm": 2.3906686305999756,
      "learning_rate": 3.3578080229226357e-06,
      "loss": 0.7446,
      "step": 17500
    },
    {
      "epoch": 7.17,
      "grad_norm": 2.521421194076538,
      "learning_rate": 3.359726770364306e-06,
      "loss": 0.7506,
      "step": 17510
    },
    {
      "epoch": 7.17,
      "grad_norm": 3.17262864112854,
      "learning_rate": 3.361645517805976e-06,
      "loss": 0.7574,
      "step": 17520
    },
    {
      "epoch": 7.18,
      "grad_norm": 3.3760437965393066,
      "learning_rate": 3.363564265247646e-06,
      "loss": 0.7533,
      "step": 17530
    },
    {
      "epoch": 7.18,
      "grad_norm": 2.3390889167785645,
      "learning_rate": 3.3654830126893163e-06,
      "loss": 0.7718,
      "step": 17540
    },
    {
      "epoch": 7.18,
      "grad_norm": 2.22184157371521,
      "learning_rate": 3.367401760130986e-06,
      "loss": 0.7499,
      "step": 17550
    },
    {
      "epoch": 7.19,
      "grad_norm": 2.378211498260498,
      "learning_rate": 3.369320507572656e-06,
      "loss": 0.7512,
      "step": 17560
    },
    {
      "epoch": 7.19,
      "grad_norm": 3.146496295928955,
      "learning_rate": 3.3712392550143265e-06,
      "loss": 0.7466,
      "step": 17570
    },
    {
      "epoch": 7.2,
      "grad_norm": 2.3085110187530518,
      "learning_rate": 3.3731580024559964e-06,
      "loss": 0.7393,
      "step": 17580
    },
    {
      "epoch": 7.2,
      "grad_norm": 2.611835479736328,
      "learning_rate": 3.3750767498976664e-06,
      "loss": 0.7465,
      "step": 17590
    },
    {
      "epoch": 7.2,
      "grad_norm": 2.2689032554626465,
      "learning_rate": 3.3769954973393367e-06,
      "loss": 0.7568,
      "step": 17600
    },
    {
      "epoch": 7.21,
      "grad_norm": 3.54691219329834,
      "learning_rate": 3.3789142447810067e-06,
      "loss": 0.7481,
      "step": 17610
    },
    {
      "epoch": 7.21,
      "grad_norm": 3.9220199584960938,
      "learning_rate": 3.3808329922226766e-06,
      "loss": 0.7575,
      "step": 17620
    },
    {
      "epoch": 7.22,
      "grad_norm": 2.2888216972351074,
      "learning_rate": 3.382751739664347e-06,
      "loss": 0.7542,
      "step": 17630
    },
    {
      "epoch": 7.22,
      "grad_norm": 1.9730502367019653,
      "learning_rate": 3.384670487106017e-06,
      "loss": 0.7542,
      "step": 17640
    },
    {
      "epoch": 7.22,
      "grad_norm": 2.4520962238311768,
      "learning_rate": 3.386589234547687e-06,
      "loss": 0.7294,
      "step": 17650
    },
    {
      "epoch": 7.23,
      "grad_norm": 2.707348108291626,
      "learning_rate": 3.3885079819893572e-06,
      "loss": 0.7482,
      "step": 17660
    },
    {
      "epoch": 7.23,
      "grad_norm": 3.2043731212615967,
      "learning_rate": 3.390426729431027e-06,
      "loss": 0.7621,
      "step": 17670
    },
    {
      "epoch": 7.24,
      "grad_norm": 1.9953136444091797,
      "learning_rate": 3.392345476872697e-06,
      "loss": 0.7521,
      "step": 17680
    },
    {
      "epoch": 7.24,
      "grad_norm": 2.4026331901550293,
      "learning_rate": 3.3942642243143675e-06,
      "loss": 0.7521,
      "step": 17690
    },
    {
      "epoch": 7.25,
      "grad_norm": 2.2547717094421387,
      "learning_rate": 3.3961829717560374e-06,
      "loss": 0.7451,
      "step": 17700
    },
    {
      "epoch": 7.25,
      "grad_norm": 1.7371487617492676,
      "learning_rate": 3.3981017191977073e-06,
      "loss": 0.7511,
      "step": 17710
    },
    {
      "epoch": 7.25,
      "grad_norm": 2.268935203552246,
      "learning_rate": 3.4000204666393773e-06,
      "loss": 0.7577,
      "step": 17720
    },
    {
      "epoch": 7.26,
      "grad_norm": 2.3801703453063965,
      "learning_rate": 3.4019392140810476e-06,
      "loss": 0.7536,
      "step": 17730
    },
    {
      "epoch": 7.26,
      "grad_norm": 1.966671109199524,
      "learning_rate": 3.4038579615227176e-06,
      "loss": 0.7368,
      "step": 17740
    },
    {
      "epoch": 7.27,
      "grad_norm": 2.4131100177764893,
      "learning_rate": 3.4057767089643875e-06,
      "loss": 0.7551,
      "step": 17750
    },
    {
      "epoch": 7.27,
      "grad_norm": 2.9243404865264893,
      "learning_rate": 3.407695456406058e-06,
      "loss": 0.7531,
      "step": 17760
    },
    {
      "epoch": 7.27,
      "grad_norm": 2.727149486541748,
      "learning_rate": 3.409614203847728e-06,
      "loss": 0.7403,
      "step": 17770
    },
    {
      "epoch": 7.28,
      "grad_norm": 2.710214138031006,
      "learning_rate": 3.4115329512893978e-06,
      "loss": 0.73,
      "step": 17780
    },
    {
      "epoch": 7.28,
      "grad_norm": 2.9263267517089844,
      "learning_rate": 3.413451698731068e-06,
      "loss": 0.7428,
      "step": 17790
    },
    {
      "epoch": 7.29,
      "grad_norm": 2.217193126678467,
      "learning_rate": 3.415370446172738e-06,
      "loss": 0.747,
      "step": 17800
    },
    {
      "epoch": 7.29,
      "grad_norm": 2.9583792686462402,
      "learning_rate": 3.417289193614408e-06,
      "loss": 0.7299,
      "step": 17810
    },
    {
      "epoch": 7.29,
      "grad_norm": 2.4935789108276367,
      "learning_rate": 3.4192079410560784e-06,
      "loss": 0.7675,
      "step": 17820
    },
    {
      "epoch": 7.3,
      "grad_norm": 3.3064022064208984,
      "learning_rate": 3.4211266884977483e-06,
      "loss": 0.7356,
      "step": 17830
    },
    {
      "epoch": 7.3,
      "grad_norm": 3.7965643405914307,
      "learning_rate": 3.4230454359394182e-06,
      "loss": 0.7327,
      "step": 17840
    },
    {
      "epoch": 7.31,
      "grad_norm": 3.566161632537842,
      "learning_rate": 3.4249641833810886e-06,
      "loss": 0.7584,
      "step": 17850
    },
    {
      "epoch": 7.31,
      "grad_norm": 2.8680901527404785,
      "learning_rate": 3.4268829308227586e-06,
      "loss": 0.738,
      "step": 17860
    },
    {
      "epoch": 7.31,
      "grad_norm": 2.477924346923828,
      "learning_rate": 3.4288016782644285e-06,
      "loss": 0.7391,
      "step": 17870
    },
    {
      "epoch": 7.32,
      "grad_norm": 2.022663116455078,
      "learning_rate": 3.430720425706099e-06,
      "loss": 0.7522,
      "step": 17880
    },
    {
      "epoch": 7.32,
      "grad_norm": 2.8619349002838135,
      "learning_rate": 3.432639173147769e-06,
      "loss": 0.7398,
      "step": 17890
    },
    {
      "epoch": 7.33,
      "grad_norm": 2.338188409805298,
      "learning_rate": 3.4345579205894387e-06,
      "loss": 0.7515,
      "step": 17900
    },
    {
      "epoch": 7.33,
      "grad_norm": 1.8188354969024658,
      "learning_rate": 3.436476668031109e-06,
      "loss": 0.7598,
      "step": 17910
    },
    {
      "epoch": 7.34,
      "grad_norm": 2.3921992778778076,
      "learning_rate": 3.438395415472779e-06,
      "loss": 0.7656,
      "step": 17920
    },
    {
      "epoch": 7.34,
      "grad_norm": 2.428778886795044,
      "learning_rate": 3.440314162914449e-06,
      "loss": 0.7369,
      "step": 17930
    },
    {
      "epoch": 7.34,
      "grad_norm": 1.9612889289855957,
      "learning_rate": 3.442232910356119e-06,
      "loss": 0.7533,
      "step": 17940
    },
    {
      "epoch": 7.35,
      "grad_norm": 3.000141143798828,
      "learning_rate": 3.4441516577977893e-06,
      "loss": 0.7586,
      "step": 17950
    },
    {
      "epoch": 7.35,
      "grad_norm": 3.2501280307769775,
      "learning_rate": 3.4460704052394592e-06,
      "loss": 0.7451,
      "step": 17960
    },
    {
      "epoch": 7.36,
      "grad_norm": 1.8310980796813965,
      "learning_rate": 3.447989152681129e-06,
      "loss": 0.7462,
      "step": 17970
    },
    {
      "epoch": 7.36,
      "grad_norm": 1.6861623525619507,
      "learning_rate": 3.4499079001227995e-06,
      "loss": 0.7496,
      "step": 17980
    },
    {
      "epoch": 7.36,
      "grad_norm": 2.4857711791992188,
      "learning_rate": 3.4518266475644695e-06,
      "loss": 0.7631,
      "step": 17990
    },
    {
      "epoch": 7.37,
      "grad_norm": 2.470285654067993,
      "learning_rate": 3.4537453950061394e-06,
      "loss": 0.7484,
      "step": 18000
    },
    {
      "epoch": 7.37,
      "grad_norm": 2.1441779136657715,
      "learning_rate": 3.4556641424478098e-06,
      "loss": 0.7498,
      "step": 18010
    },
    {
      "epoch": 7.38,
      "grad_norm": 2.1529030799865723,
      "learning_rate": 3.4575828898894797e-06,
      "loss": 0.7822,
      "step": 18020
    },
    {
      "epoch": 7.38,
      "grad_norm": 2.17756986618042,
      "learning_rate": 3.4595016373311496e-06,
      "loss": 0.7379,
      "step": 18030
    },
    {
      "epoch": 7.38,
      "grad_norm": 3.705406665802002,
      "learning_rate": 3.46142038477282e-06,
      "loss": 0.749,
      "step": 18040
    },
    {
      "epoch": 7.39,
      "grad_norm": 3.919879674911499,
      "learning_rate": 3.46333913221449e-06,
      "loss": 0.7345,
      "step": 18050
    },
    {
      "epoch": 7.39,
      "grad_norm": 2.5731003284454346,
      "learning_rate": 3.46525787965616e-06,
      "loss": 0.7502,
      "step": 18060
    },
    {
      "epoch": 7.4,
      "grad_norm": 2.0312068462371826,
      "learning_rate": 3.4671766270978302e-06,
      "loss": 0.7283,
      "step": 18070
    },
    {
      "epoch": 7.4,
      "grad_norm": 1.9008715152740479,
      "learning_rate": 3.4690953745395e-06,
      "loss": 0.7366,
      "step": 18080
    },
    {
      "epoch": 7.4,
      "grad_norm": 2.20147967338562,
      "learning_rate": 3.47101412198117e-06,
      "loss": 0.7407,
      "step": 18090
    },
    {
      "epoch": 7.41,
      "grad_norm": 3.049466371536255,
      "learning_rate": 3.4729328694228405e-06,
      "loss": 0.7298,
      "step": 18100
    },
    {
      "epoch": 7.41,
      "grad_norm": 1.4198180437088013,
      "learning_rate": 3.4748516168645104e-06,
      "loss": 0.7495,
      "step": 18110
    },
    {
      "epoch": 7.42,
      "grad_norm": 2.2756857872009277,
      "learning_rate": 3.4767703643061804e-06,
      "loss": 0.7339,
      "step": 18120
    },
    {
      "epoch": 7.42,
      "grad_norm": 3.139470338821411,
      "learning_rate": 3.4786891117478507e-06,
      "loss": 0.7523,
      "step": 18130
    },
    {
      "epoch": 7.43,
      "grad_norm": 2.5963659286499023,
      "learning_rate": 3.4806078591895207e-06,
      "loss": 0.7474,
      "step": 18140
    },
    {
      "epoch": 7.43,
      "grad_norm": 2.1993749141693115,
      "learning_rate": 3.4825266066311906e-06,
      "loss": 0.7455,
      "step": 18150
    },
    {
      "epoch": 7.43,
      "grad_norm": 2.489441394805908,
      "learning_rate": 3.4844453540728605e-06,
      "loss": 0.7431,
      "step": 18160
    },
    {
      "epoch": 7.44,
      "grad_norm": 3.833237648010254,
      "learning_rate": 3.486364101514531e-06,
      "loss": 0.7567,
      "step": 18170
    },
    {
      "epoch": 7.44,
      "grad_norm": 2.5943360328674316,
      "learning_rate": 3.488282848956201e-06,
      "loss": 0.7537,
      "step": 18180
    },
    {
      "epoch": 7.45,
      "grad_norm": 2.978222370147705,
      "learning_rate": 3.4902015963978708e-06,
      "loss": 0.7398,
      "step": 18190
    },
    {
      "epoch": 7.45,
      "grad_norm": 2.3400490283966064,
      "learning_rate": 3.492120343839541e-06,
      "loss": 0.7298,
      "step": 18200
    },
    {
      "epoch": 7.45,
      "grad_norm": 2.8937556743621826,
      "learning_rate": 3.494039091281211e-06,
      "loss": 0.7339,
      "step": 18210
    },
    {
      "epoch": 7.46,
      "grad_norm": 1.928074598312378,
      "learning_rate": 3.495957838722881e-06,
      "loss": 0.755,
      "step": 18220
    },
    {
      "epoch": 7.46,
      "grad_norm": 2.7971529960632324,
      "learning_rate": 3.4978765861645514e-06,
      "loss": 0.7397,
      "step": 18230
    },
    {
      "epoch": 7.47,
      "grad_norm": 2.118241310119629,
      "learning_rate": 3.4997953336062213e-06,
      "loss": 0.7498,
      "step": 18240
    },
    {
      "epoch": 7.47,
      "grad_norm": 2.6184754371643066,
      "learning_rate": 3.5017140810478913e-06,
      "loss": 0.7465,
      "step": 18250
    },
    {
      "epoch": 7.47,
      "grad_norm": 2.1411590576171875,
      "learning_rate": 3.5036328284895616e-06,
      "loss": 0.7674,
      "step": 18260
    },
    {
      "epoch": 7.48,
      "grad_norm": 3.162458896636963,
      "learning_rate": 3.5055515759312316e-06,
      "loss": 0.7411,
      "step": 18270
    },
    {
      "epoch": 7.48,
      "grad_norm": 2.4051504135131836,
      "learning_rate": 3.5074703233729015e-06,
      "loss": 0.7563,
      "step": 18280
    },
    {
      "epoch": 7.49,
      "grad_norm": 3.624148368835449,
      "learning_rate": 3.509389070814572e-06,
      "loss": 0.7326,
      "step": 18290
    },
    {
      "epoch": 7.49,
      "grad_norm": 3.069826126098633,
      "learning_rate": 3.511307818256242e-06,
      "loss": 0.7204,
      "step": 18300
    },
    {
      "epoch": 7.49,
      "grad_norm": 2.2082855701446533,
      "learning_rate": 3.5132265656979117e-06,
      "loss": 0.73,
      "step": 18310
    },
    {
      "epoch": 7.5,
      "grad_norm": 2.445910930633545,
      "learning_rate": 3.515145313139582e-06,
      "loss": 0.7402,
      "step": 18320
    },
    {
      "epoch": 7.5,
      "grad_norm": 1.8297665119171143,
      "learning_rate": 3.5170640605812525e-06,
      "loss": 0.7618,
      "step": 18330
    },
    {
      "epoch": 7.51,
      "grad_norm": 2.4674296379089355,
      "learning_rate": 3.5189828080229224e-06,
      "loss": 0.7565,
      "step": 18340
    },
    {
      "epoch": 7.51,
      "grad_norm": 2.1850197315216064,
      "learning_rate": 3.5209015554645928e-06,
      "loss": 0.7387,
      "step": 18350
    },
    {
      "epoch": 7.52,
      "grad_norm": 2.919203281402588,
      "learning_rate": 3.5228203029062627e-06,
      "loss": 0.7473,
      "step": 18360
    },
    {
      "epoch": 7.52,
      "grad_norm": 2.4457521438598633,
      "learning_rate": 3.5247390503479326e-06,
      "loss": 0.7438,
      "step": 18370
    },
    {
      "epoch": 7.52,
      "grad_norm": 2.0924339294433594,
      "learning_rate": 3.526657797789603e-06,
      "loss": 0.747,
      "step": 18380
    },
    {
      "epoch": 7.53,
      "grad_norm": 2.3261003494262695,
      "learning_rate": 3.528576545231273e-06,
      "loss": 0.7412,
      "step": 18390
    },
    {
      "epoch": 7.53,
      "grad_norm": 3.364964485168457,
      "learning_rate": 3.530495292672943e-06,
      "loss": 0.7347,
      "step": 18400
    },
    {
      "epoch": 7.54,
      "grad_norm": 3.0130534172058105,
      "learning_rate": 3.5324140401146132e-06,
      "loss": 0.7192,
      "step": 18410
    },
    {
      "epoch": 7.54,
      "grad_norm": 1.9294109344482422,
      "learning_rate": 3.534332787556283e-06,
      "loss": 0.7415,
      "step": 18420
    },
    {
      "epoch": 7.54,
      "grad_norm": 2.765706777572632,
      "learning_rate": 3.536251534997953e-06,
      "loss": 0.7528,
      "step": 18430
    },
    {
      "epoch": 7.55,
      "grad_norm": 1.935694694519043,
      "learning_rate": 3.5381702824396235e-06,
      "loss": 0.7195,
      "step": 18440
    },
    {
      "epoch": 7.55,
      "grad_norm": 2.5445408821105957,
      "learning_rate": 3.5400890298812934e-06,
      "loss": 0.7393,
      "step": 18450
    },
    {
      "epoch": 7.56,
      "grad_norm": 3.177288293838501,
      "learning_rate": 3.5420077773229634e-06,
      "loss": 0.7412,
      "step": 18460
    },
    {
      "epoch": 7.56,
      "grad_norm": 2.0527708530426025,
      "learning_rate": 3.5439265247646337e-06,
      "loss": 0.7494,
      "step": 18470
    },
    {
      "epoch": 7.56,
      "grad_norm": 2.073740005493164,
      "learning_rate": 3.5458452722063037e-06,
      "loss": 0.7392,
      "step": 18480
    },
    {
      "epoch": 7.57,
      "grad_norm": 2.005242347717285,
      "learning_rate": 3.5477640196479736e-06,
      "loss": 0.7548,
      "step": 18490
    },
    {
      "epoch": 7.57,
      "grad_norm": 2.553558111190796,
      "learning_rate": 3.549682767089644e-06,
      "loss": 0.7518,
      "step": 18500
    },
    {
      "epoch": 7.58,
      "grad_norm": 3.659876823425293,
      "learning_rate": 3.551601514531314e-06,
      "loss": 0.7296,
      "step": 18510
    },
    {
      "epoch": 7.58,
      "grad_norm": 2.0127222537994385,
      "learning_rate": 3.553520261972984e-06,
      "loss": 0.7343,
      "step": 18520
    },
    {
      "epoch": 7.58,
      "grad_norm": 1.7387890815734863,
      "learning_rate": 3.5554390094146538e-06,
      "loss": 0.7523,
      "step": 18530
    },
    {
      "epoch": 7.59,
      "grad_norm": 3.131810426712036,
      "learning_rate": 3.557357756856324e-06,
      "loss": 0.7335,
      "step": 18540
    },
    {
      "epoch": 7.59,
      "grad_norm": 3.092700719833374,
      "learning_rate": 3.559276504297994e-06,
      "loss": 0.756,
      "step": 18550
    },
    {
      "epoch": 7.6,
      "grad_norm": 2.852936267852783,
      "learning_rate": 3.561195251739664e-06,
      "loss": 0.7219,
      "step": 18560
    },
    {
      "epoch": 7.6,
      "grad_norm": 2.8484230041503906,
      "learning_rate": 3.5631139991813344e-06,
      "loss": 0.7314,
      "step": 18570
    },
    {
      "epoch": 7.61,
      "grad_norm": 2.803412914276123,
      "learning_rate": 3.5650327466230043e-06,
      "loss": 0.7404,
      "step": 18580
    },
    {
      "epoch": 7.61,
      "grad_norm": 2.1542115211486816,
      "learning_rate": 3.5669514940646743e-06,
      "loss": 0.7477,
      "step": 18590
    },
    {
      "epoch": 7.61,
      "grad_norm": 2.6055939197540283,
      "learning_rate": 3.5688702415063446e-06,
      "loss": 0.7501,
      "step": 18600
    },
    {
      "epoch": 7.62,
      "grad_norm": 2.08142352104187,
      "learning_rate": 3.5707889889480146e-06,
      "loss": 0.7212,
      "step": 18610
    },
    {
      "epoch": 7.62,
      "grad_norm": 3.4083633422851562,
      "learning_rate": 3.5727077363896845e-06,
      "loss": 0.7396,
      "step": 18620
    },
    {
      "epoch": 7.63,
      "grad_norm": 2.71659517288208,
      "learning_rate": 3.574626483831355e-06,
      "loss": 0.7298,
      "step": 18630
    },
    {
      "epoch": 7.63,
      "grad_norm": 1.8406809568405151,
      "learning_rate": 3.576545231273025e-06,
      "loss": 0.7495,
      "step": 18640
    },
    {
      "epoch": 7.63,
      "grad_norm": 2.950094699859619,
      "learning_rate": 3.5784639787146948e-06,
      "loss": 0.7357,
      "step": 18650
    },
    {
      "epoch": 7.64,
      "grad_norm": 2.327704668045044,
      "learning_rate": 3.580382726156365e-06,
      "loss": 0.7404,
      "step": 18660
    },
    {
      "epoch": 7.64,
      "grad_norm": 2.7809853553771973,
      "learning_rate": 3.582301473598035e-06,
      "loss": 0.7267,
      "step": 18670
    },
    {
      "epoch": 7.65,
      "grad_norm": 2.351701259613037,
      "learning_rate": 3.584220221039705e-06,
      "loss": 0.737,
      "step": 18680
    },
    {
      "epoch": 7.65,
      "grad_norm": 3.577531576156616,
      "learning_rate": 3.5861389684813754e-06,
      "loss": 0.7188,
      "step": 18690
    },
    {
      "epoch": 7.65,
      "grad_norm": 2.599696159362793,
      "learning_rate": 3.5880577159230453e-06,
      "loss": 0.7561,
      "step": 18700
    },
    {
      "epoch": 7.66,
      "grad_norm": 2.2597334384918213,
      "learning_rate": 3.5899764633647152e-06,
      "loss": 0.7272,
      "step": 18710
    },
    {
      "epoch": 7.66,
      "grad_norm": 3.771239757537842,
      "learning_rate": 3.5918952108063856e-06,
      "loss": 0.7428,
      "step": 18720
    },
    {
      "epoch": 7.67,
      "grad_norm": 2.3528127670288086,
      "learning_rate": 3.5938139582480555e-06,
      "loss": 0.7402,
      "step": 18730
    },
    {
      "epoch": 7.67,
      "grad_norm": 2.7123818397521973,
      "learning_rate": 3.5957327056897255e-06,
      "loss": 0.7429,
      "step": 18740
    },
    {
      "epoch": 7.67,
      "grad_norm": 3.1658742427825928,
      "learning_rate": 3.5976514531313954e-06,
      "loss": 0.7339,
      "step": 18750
    },
    {
      "epoch": 7.68,
      "grad_norm": 3.19943904876709,
      "learning_rate": 3.5995702005730658e-06,
      "loss": 0.7209,
      "step": 18760
    },
    {
      "epoch": 7.68,
      "grad_norm": 2.2596957683563232,
      "learning_rate": 3.6014889480147357e-06,
      "loss": 0.7328,
      "step": 18770
    },
    {
      "epoch": 7.69,
      "grad_norm": 1.8075937032699585,
      "learning_rate": 3.6034076954564057e-06,
      "loss": 0.7259,
      "step": 18780
    },
    {
      "epoch": 7.69,
      "grad_norm": 3.212245464324951,
      "learning_rate": 3.605326442898076e-06,
      "loss": 0.7464,
      "step": 18790
    },
    {
      "epoch": 7.7,
      "grad_norm": 2.4385218620300293,
      "learning_rate": 3.607245190339746e-06,
      "loss": 0.7462,
      "step": 18800
    },
    {
      "epoch": 7.7,
      "grad_norm": 3.546638011932373,
      "learning_rate": 3.609163937781416e-06,
      "loss": 0.7419,
      "step": 18810
    },
    {
      "epoch": 7.7,
      "grad_norm": 3.5286381244659424,
      "learning_rate": 3.6110826852230863e-06,
      "loss": 0.7411,
      "step": 18820
    },
    {
      "epoch": 7.71,
      "grad_norm": 2.477532148361206,
      "learning_rate": 3.613001432664756e-06,
      "loss": 0.7353,
      "step": 18830
    },
    {
      "epoch": 7.71,
      "grad_norm": 2.3625662326812744,
      "learning_rate": 3.614920180106426e-06,
      "loss": 0.748,
      "step": 18840
    },
    {
      "epoch": 7.72,
      "grad_norm": 3.0101239681243896,
      "learning_rate": 3.6168389275480965e-06,
      "loss": 0.7207,
      "step": 18850
    },
    {
      "epoch": 7.72,
      "grad_norm": 1.8601728677749634,
      "learning_rate": 3.6187576749897664e-06,
      "loss": 0.7445,
      "step": 18860
    },
    {
      "epoch": 7.72,
      "grad_norm": 3.25374698638916,
      "learning_rate": 3.6206764224314364e-06,
      "loss": 0.7355,
      "step": 18870
    },
    {
      "epoch": 7.73,
      "grad_norm": 2.4937376976013184,
      "learning_rate": 3.6225951698731067e-06,
      "loss": 0.7283,
      "step": 18880
    },
    {
      "epoch": 7.73,
      "grad_norm": 2.578594446182251,
      "learning_rate": 3.6245139173147767e-06,
      "loss": 0.7485,
      "step": 18890
    },
    {
      "epoch": 7.74,
      "grad_norm": 2.0486013889312744,
      "learning_rate": 3.6264326647564466e-06,
      "loss": 0.7485,
      "step": 18900
    },
    {
      "epoch": 7.74,
      "grad_norm": 2.7974467277526855,
      "learning_rate": 3.628351412198117e-06,
      "loss": 0.7359,
      "step": 18910
    },
    {
      "epoch": 7.74,
      "grad_norm": 3.6336586475372314,
      "learning_rate": 3.630270159639787e-06,
      "loss": 0.7541,
      "step": 18920
    },
    {
      "epoch": 7.75,
      "grad_norm": 2.7304091453552246,
      "learning_rate": 3.632188907081457e-06,
      "loss": 0.7574,
      "step": 18930
    },
    {
      "epoch": 7.75,
      "grad_norm": 3.638920783996582,
      "learning_rate": 3.6341076545231272e-06,
      "loss": 0.7672,
      "step": 18940
    },
    {
      "epoch": 7.76,
      "grad_norm": 2.6557812690734863,
      "learning_rate": 3.636026401964797e-06,
      "loss": 0.7159,
      "step": 18950
    },
    {
      "epoch": 7.76,
      "grad_norm": 3.0026936531066895,
      "learning_rate": 3.637945149406467e-06,
      "loss": 0.7346,
      "step": 18960
    },
    {
      "epoch": 7.77,
      "grad_norm": 2.104013204574585,
      "learning_rate": 3.639863896848137e-06,
      "loss": 0.7427,
      "step": 18970
    },
    {
      "epoch": 7.77,
      "grad_norm": 2.915466070175171,
      "learning_rate": 3.6417826442898074e-06,
      "loss": 0.7449,
      "step": 18980
    },
    {
      "epoch": 7.77,
      "grad_norm": 2.579841375350952,
      "learning_rate": 3.6437013917314773e-06,
      "loss": 0.7378,
      "step": 18990
    },
    {
      "epoch": 7.78,
      "grad_norm": 2.1532042026519775,
      "learning_rate": 3.6456201391731473e-06,
      "loss": 0.745,
      "step": 19000
    },
    {
      "epoch": 7.78,
      "grad_norm": 2.887186050415039,
      "learning_rate": 3.6475388866148176e-06,
      "loss": 0.7136,
      "step": 19010
    },
    {
      "epoch": 7.79,
      "grad_norm": 2.2763407230377197,
      "learning_rate": 3.6494576340564876e-06,
      "loss": 0.7663,
      "step": 19020
    },
    {
      "epoch": 7.79,
      "grad_norm": 2.4176902770996094,
      "learning_rate": 3.6513763814981575e-06,
      "loss": 0.7615,
      "step": 19030
    },
    {
      "epoch": 7.79,
      "grad_norm": 2.5506932735443115,
      "learning_rate": 3.653295128939828e-06,
      "loss": 0.7427,
      "step": 19040
    },
    {
      "epoch": 7.8,
      "grad_norm": 2.942591428756714,
      "learning_rate": 3.655213876381498e-06,
      "loss": 0.7398,
      "step": 19050
    },
    {
      "epoch": 7.8,
      "grad_norm": 1.882542610168457,
      "learning_rate": 3.6571326238231678e-06,
      "loss": 0.7462,
      "step": 19060
    },
    {
      "epoch": 7.81,
      "grad_norm": 2.2738473415374756,
      "learning_rate": 3.659051371264838e-06,
      "loss": 0.754,
      "step": 19070
    },
    {
      "epoch": 7.81,
      "grad_norm": 2.9294309616088867,
      "learning_rate": 3.660970118706508e-06,
      "loss": 0.736,
      "step": 19080
    },
    {
      "epoch": 7.81,
      "grad_norm": 2.1973984241485596,
      "learning_rate": 3.662888866148178e-06,
      "loss": 0.7553,
      "step": 19090
    },
    {
      "epoch": 7.82,
      "grad_norm": 1.81239914894104,
      "learning_rate": 3.6648076135898484e-06,
      "loss": 0.7422,
      "step": 19100
    },
    {
      "epoch": 7.82,
      "grad_norm": 3.04838490486145,
      "learning_rate": 3.6667263610315183e-06,
      "loss": 0.7139,
      "step": 19110
    },
    {
      "epoch": 7.83,
      "grad_norm": 3.7952518463134766,
      "learning_rate": 3.6686451084731882e-06,
      "loss": 0.7492,
      "step": 19120
    },
    {
      "epoch": 7.83,
      "grad_norm": 1.7786036729812622,
      "learning_rate": 3.6705638559148586e-06,
      "loss": 0.7555,
      "step": 19130
    },
    {
      "epoch": 7.83,
      "grad_norm": 2.722230911254883,
      "learning_rate": 3.6724826033565285e-06,
      "loss": 0.7242,
      "step": 19140
    },
    {
      "epoch": 7.84,
      "grad_norm": 1.8700278997421265,
      "learning_rate": 3.6744013507981985e-06,
      "loss": 0.7416,
      "step": 19150
    },
    {
      "epoch": 7.84,
      "grad_norm": 2.2618072032928467,
      "learning_rate": 3.6763200982398684e-06,
      "loss": 0.7307,
      "step": 19160
    },
    {
      "epoch": 7.85,
      "grad_norm": 2.506493091583252,
      "learning_rate": 3.6782388456815388e-06,
      "loss": 0.7284,
      "step": 19170
    },
    {
      "epoch": 7.85,
      "grad_norm": 2.8855278491973877,
      "learning_rate": 3.6801575931232087e-06,
      "loss": 0.7118,
      "step": 19180
    },
    {
      "epoch": 7.86,
      "grad_norm": 2.4437901973724365,
      "learning_rate": 3.6820763405648787e-06,
      "loss": 0.734,
      "step": 19190
    },
    {
      "epoch": 7.86,
      "grad_norm": 2.6283721923828125,
      "learning_rate": 3.683995088006549e-06,
      "loss": 0.7276,
      "step": 19200
    },
    {
      "epoch": 7.86,
      "grad_norm": 2.0561740398406982,
      "learning_rate": 3.685913835448219e-06,
      "loss": 0.7376,
      "step": 19210
    },
    {
      "epoch": 7.87,
      "grad_norm": 2.856940984725952,
      "learning_rate": 3.687832582889889e-06,
      "loss": 0.7216,
      "step": 19220
    },
    {
      "epoch": 7.87,
      "grad_norm": 2.768277168273926,
      "learning_rate": 3.6897513303315593e-06,
      "loss": 0.712,
      "step": 19230
    },
    {
      "epoch": 7.88,
      "grad_norm": 2.3005192279815674,
      "learning_rate": 3.691670077773229e-06,
      "loss": 0.7279,
      "step": 19240
    },
    {
      "epoch": 7.88,
      "grad_norm": 2.4589807987213135,
      "learning_rate": 3.693588825214899e-06,
      "loss": 0.7341,
      "step": 19250
    },
    {
      "epoch": 7.88,
      "grad_norm": 2.1743433475494385,
      "learning_rate": 3.6955075726565695e-06,
      "loss": 0.7418,
      "step": 19260
    },
    {
      "epoch": 7.89,
      "grad_norm": 1.7972573041915894,
      "learning_rate": 3.6974263200982394e-06,
      "loss": 0.7262,
      "step": 19270
    },
    {
      "epoch": 7.89,
      "grad_norm": 2.033193826675415,
      "learning_rate": 3.6993450675399094e-06,
      "loss": 0.7498,
      "step": 19280
    },
    {
      "epoch": 7.9,
      "grad_norm": 3.122403383255005,
      "learning_rate": 3.7012638149815797e-06,
      "loss": 0.7329,
      "step": 19290
    },
    {
      "epoch": 7.9,
      "grad_norm": 1.9098644256591797,
      "learning_rate": 3.7031825624232497e-06,
      "loss": 0.7604,
      "step": 19300
    },
    {
      "epoch": 7.9,
      "grad_norm": 2.0980002880096436,
      "learning_rate": 3.7051013098649196e-06,
      "loss": 0.726,
      "step": 19310
    },
    {
      "epoch": 7.91,
      "grad_norm": 3.0517590045928955,
      "learning_rate": 3.70702005730659e-06,
      "loss": 0.7242,
      "step": 19320
    },
    {
      "epoch": 7.91,
      "grad_norm": 2.333381414413452,
      "learning_rate": 3.70893880474826e-06,
      "loss": 0.7263,
      "step": 19330
    },
    {
      "epoch": 7.92,
      "grad_norm": 3.725365161895752,
      "learning_rate": 3.71085755218993e-06,
      "loss": 0.7549,
      "step": 19340
    },
    {
      "epoch": 7.92,
      "grad_norm": 2.35508131980896,
      "learning_rate": 3.7127762996316002e-06,
      "loss": 0.735,
      "step": 19350
    },
    {
      "epoch": 7.92,
      "grad_norm": 3.2422571182250977,
      "learning_rate": 3.71469504707327e-06,
      "loss": 0.7744,
      "step": 19360
    },
    {
      "epoch": 7.93,
      "grad_norm": 2.435206890106201,
      "learning_rate": 3.71661379451494e-06,
      "loss": 0.7319,
      "step": 19370
    },
    {
      "epoch": 7.93,
      "grad_norm": 2.3231966495513916,
      "learning_rate": 3.71853254195661e-06,
      "loss": 0.7355,
      "step": 19380
    },
    {
      "epoch": 7.94,
      "grad_norm": 3.0547983646392822,
      "learning_rate": 3.7204512893982804e-06,
      "loss": 0.7315,
      "step": 19390
    },
    {
      "epoch": 7.94,
      "grad_norm": 2.3223750591278076,
      "learning_rate": 3.7223700368399503e-06,
      "loss": 0.7362,
      "step": 19400
    },
    {
      "epoch": 7.95,
      "grad_norm": 2.6958742141723633,
      "learning_rate": 3.7242887842816203e-06,
      "loss": 0.7303,
      "step": 19410
    },
    {
      "epoch": 7.95,
      "grad_norm": 2.1155409812927246,
      "learning_rate": 3.7262075317232906e-06,
      "loss": 0.7271,
      "step": 19420
    },
    {
      "epoch": 7.95,
      "grad_norm": 2.3192453384399414,
      "learning_rate": 3.7281262791649606e-06,
      "loss": 0.7226,
      "step": 19430
    },
    {
      "epoch": 7.96,
      "grad_norm": 2.0813770294189453,
      "learning_rate": 3.7300450266066305e-06,
      "loss": 0.7487,
      "step": 19440
    },
    {
      "epoch": 7.96,
      "grad_norm": 2.736722946166992,
      "learning_rate": 3.731963774048301e-06,
      "loss": 0.7215,
      "step": 19450
    },
    {
      "epoch": 7.97,
      "grad_norm": 1.607688546180725,
      "learning_rate": 3.733882521489971e-06,
      "loss": 0.7366,
      "step": 19460
    },
    {
      "epoch": 7.97,
      "grad_norm": 3.448106050491333,
      "learning_rate": 3.7358012689316408e-06,
      "loss": 0.7501,
      "step": 19470
    },
    {
      "epoch": 7.97,
      "grad_norm": 3.102379083633423,
      "learning_rate": 3.737720016373311e-06,
      "loss": 0.7344,
      "step": 19480
    },
    {
      "epoch": 7.98,
      "grad_norm": 1.9554345607757568,
      "learning_rate": 3.739638763814981e-06,
      "loss": 0.7513,
      "step": 19490
    },
    {
      "epoch": 7.98,
      "grad_norm": 2.47356915473938,
      "learning_rate": 3.741557511256651e-06,
      "loss": 0.7423,
      "step": 19500
    },
    {
      "epoch": 7.99,
      "grad_norm": 2.693352222442627,
      "learning_rate": 3.7434762586983214e-06,
      "loss": 0.7387,
      "step": 19510
    },
    {
      "epoch": 7.99,
      "grad_norm": 2.886892080307007,
      "learning_rate": 3.7453950061399913e-06,
      "loss": 0.7262,
      "step": 19520
    },
    {
      "epoch": 7.99,
      "grad_norm": 2.2040047645568848,
      "learning_rate": 3.7473137535816612e-06,
      "loss": 0.7555,
      "step": 19530
    },
    {
      "epoch": 8.0,
      "grad_norm": 2.6085331439971924,
      "learning_rate": 3.7492325010233316e-06,
      "loss": 0.7421,
      "step": 19540
    },
    {
      "epoch": 8.0,
      "eval_loss": 0.7336708307266235,
      "eval_runtime": 52.216,
      "eval_samples_per_second": 66.053,
      "eval_steps_per_second": 8.273,
      "step": 19544
    },
    {
      "epoch": 8.0,
      "grad_norm": 2.56042218208313,
      "learning_rate": 3.7511512484650016e-06,
      "loss": 0.7266,
      "step": 19550
    },
    {
      "epoch": 8.01,
      "grad_norm": 2.152508020401001,
      "learning_rate": 3.7530699959066715e-06,
      "loss": 0.724,
      "step": 19560
    },
    {
      "epoch": 8.01,
      "grad_norm": 2.0230259895324707,
      "learning_rate": 3.754988743348342e-06,
      "loss": 0.7227,
      "step": 19570
    },
    {
      "epoch": 8.01,
      "grad_norm": 2.0087363719940186,
      "learning_rate": 3.756907490790012e-06,
      "loss": 0.7478,
      "step": 19580
    },
    {
      "epoch": 8.02,
      "grad_norm": 2.0666065216064453,
      "learning_rate": 3.7588262382316817e-06,
      "loss": 0.7273,
      "step": 19590
    },
    {
      "epoch": 8.02,
      "grad_norm": 2.376610517501831,
      "learning_rate": 3.7607449856733517e-06,
      "loss": 0.7419,
      "step": 19600
    },
    {
      "epoch": 8.03,
      "grad_norm": 1.902840256690979,
      "learning_rate": 3.762663733115022e-06,
      "loss": 0.7184,
      "step": 19610
    },
    {
      "epoch": 8.03,
      "grad_norm": 2.0478739738464355,
      "learning_rate": 3.764582480556692e-06,
      "loss": 0.7382,
      "step": 19620
    },
    {
      "epoch": 8.04,
      "grad_norm": 2.0761988162994385,
      "learning_rate": 3.766501227998362e-06,
      "loss": 0.7533,
      "step": 19630
    },
    {
      "epoch": 8.04,
      "grad_norm": 2.2151663303375244,
      "learning_rate": 3.7684199754400323e-06,
      "loss": 0.7286,
      "step": 19640
    },
    {
      "epoch": 8.04,
      "grad_norm": 2.9513864517211914,
      "learning_rate": 3.770338722881702e-06,
      "loss": 0.7251,
      "step": 19650
    },
    {
      "epoch": 8.05,
      "grad_norm": 2.138482093811035,
      "learning_rate": 3.772257470323372e-06,
      "loss": 0.727,
      "step": 19660
    },
    {
      "epoch": 8.05,
      "grad_norm": 4.219424724578857,
      "learning_rate": 3.7741762177650425e-06,
      "loss": 0.7248,
      "step": 19670
    },
    {
      "epoch": 8.06,
      "grad_norm": 2.037632703781128,
      "learning_rate": 3.776094965206713e-06,
      "loss": 0.7363,
      "step": 19680
    },
    {
      "epoch": 8.06,
      "grad_norm": 2.8531479835510254,
      "learning_rate": 3.7780137126483832e-06,
      "loss": 0.7438,
      "step": 19690
    },
    {
      "epoch": 8.06,
      "grad_norm": 2.3327910900115967,
      "learning_rate": 3.779932460090053e-06,
      "loss": 0.7158,
      "step": 19700
    },
    {
      "epoch": 8.07,
      "grad_norm": 3.0103955268859863,
      "learning_rate": 3.781851207531723e-06,
      "loss": 0.7453,
      "step": 19710
    },
    {
      "epoch": 8.07,
      "grad_norm": 3.7236547470092773,
      "learning_rate": 3.7837699549733935e-06,
      "loss": 0.7267,
      "step": 19720
    },
    {
      "epoch": 8.08,
      "grad_norm": 1.9048521518707275,
      "learning_rate": 3.7856887024150634e-06,
      "loss": 0.7407,
      "step": 19730
    },
    {
      "epoch": 8.08,
      "grad_norm": 1.5924760103225708,
      "learning_rate": 3.7876074498567334e-06,
      "loss": 0.7305,
      "step": 19740
    },
    {
      "epoch": 8.08,
      "grad_norm": 3.1282122135162354,
      "learning_rate": 3.7895261972984033e-06,
      "loss": 0.7314,
      "step": 19750
    },
    {
      "epoch": 8.09,
      "grad_norm": 2.4540796279907227,
      "learning_rate": 3.7914449447400737e-06,
      "loss": 0.7069,
      "step": 19760
    },
    {
      "epoch": 8.09,
      "grad_norm": 2.9068872928619385,
      "learning_rate": 3.7933636921817436e-06,
      "loss": 0.7394,
      "step": 19770
    },
    {
      "epoch": 8.1,
      "grad_norm": 4.446603775024414,
      "learning_rate": 3.7952824396234135e-06,
      "loss": 0.7184,
      "step": 19780
    },
    {
      "epoch": 8.1,
      "grad_norm": 3.307251453399658,
      "learning_rate": 3.797201187065084e-06,
      "loss": 0.747,
      "step": 19790
    },
    {
      "epoch": 8.1,
      "grad_norm": 2.128770589828491,
      "learning_rate": 3.799119934506754e-06,
      "loss": 0.7293,
      "step": 19800
    },
    {
      "epoch": 8.11,
      "grad_norm": 1.7730234861373901,
      "learning_rate": 3.8010386819484238e-06,
      "loss": 0.7383,
      "step": 19810
    },
    {
      "epoch": 8.11,
      "grad_norm": 1.9136766195297241,
      "learning_rate": 3.802957429390094e-06,
      "loss": 0.7436,
      "step": 19820
    },
    {
      "epoch": 8.12,
      "grad_norm": 2.040583610534668,
      "learning_rate": 3.804876176831764e-06,
      "loss": 0.7443,
      "step": 19830
    },
    {
      "epoch": 8.12,
      "grad_norm": 3.0554463863372803,
      "learning_rate": 3.806794924273434e-06,
      "loss": 0.7252,
      "step": 19840
    },
    {
      "epoch": 8.13,
      "grad_norm": 4.265105724334717,
      "learning_rate": 3.8087136717151044e-06,
      "loss": 0.7381,
      "step": 19850
    },
    {
      "epoch": 8.13,
      "grad_norm": 3.785093069076538,
      "learning_rate": 3.8106324191567743e-06,
      "loss": 0.7439,
      "step": 19860
    },
    {
      "epoch": 8.13,
      "grad_norm": 7.095179557800293,
      "learning_rate": 3.8125511665984443e-06,
      "loss": 0.7244,
      "step": 19870
    },
    {
      "epoch": 8.14,
      "grad_norm": 1.9620656967163086,
      "learning_rate": 3.8144699140401146e-06,
      "loss": 0.7293,
      "step": 19880
    },
    {
      "epoch": 8.14,
      "grad_norm": 2.6351447105407715,
      "learning_rate": 3.8163886614817846e-06,
      "loss": 0.7143,
      "step": 19890
    },
    {
      "epoch": 8.15,
      "grad_norm": 2.4895803928375244,
      "learning_rate": 3.818307408923455e-06,
      "loss": 0.7402,
      "step": 19900
    },
    {
      "epoch": 8.15,
      "grad_norm": 2.789872646331787,
      "learning_rate": 3.8202261563651244e-06,
      "loss": 0.7259,
      "step": 19910
    },
    {
      "epoch": 8.15,
      "grad_norm": 2.9943158626556396,
      "learning_rate": 3.822144903806795e-06,
      "loss": 0.7291,
      "step": 19920
    },
    {
      "epoch": 8.16,
      "grad_norm": 2.1581099033355713,
      "learning_rate": 3.824063651248465e-06,
      "loss": 0.7378,
      "step": 19930
    },
    {
      "epoch": 8.16,
      "grad_norm": 3.6043710708618164,
      "learning_rate": 3.825982398690135e-06,
      "loss": 0.7381,
      "step": 19940
    },
    {
      "epoch": 8.17,
      "grad_norm": 2.5474650859832764,
      "learning_rate": 3.827901146131805e-06,
      "loss": 0.7297,
      "step": 19950
    },
    {
      "epoch": 8.17,
      "grad_norm": 1.650131344795227,
      "learning_rate": 3.829819893573475e-06,
      "loss": 0.7431,
      "step": 19960
    },
    {
      "epoch": 8.17,
      "grad_norm": 2.332686185836792,
      "learning_rate": 3.831738641015145e-06,
      "loss": 0.7279,
      "step": 19970
    },
    {
      "epoch": 8.18,
      "grad_norm": 2.5812113285064697,
      "learning_rate": 3.833657388456815e-06,
      "loss": 0.7069,
      "step": 19980
    },
    {
      "epoch": 8.18,
      "grad_norm": 2.3241212368011475,
      "learning_rate": 3.835576135898486e-06,
      "loss": 0.7313,
      "step": 19990
    },
    {
      "epoch": 8.19,
      "grad_norm": 2.564049482345581,
      "learning_rate": 3.837494883340155e-06,
      "loss": 0.7157,
      "step": 20000
    },
    {
      "epoch": 8.19,
      "grad_norm": 2.991683006286621,
      "learning_rate": 3.8394136307818255e-06,
      "loss": 0.7289,
      "step": 20010
    },
    {
      "epoch": 8.19,
      "grad_norm": 2.2241599559783936,
      "learning_rate": 3.841332378223496e-06,
      "loss": 0.7412,
      "step": 20020
    },
    {
      "epoch": 8.2,
      "grad_norm": 2.564943552017212,
      "learning_rate": 3.843251125665165e-06,
      "loss": 0.7481,
      "step": 20030
    },
    {
      "epoch": 8.2,
      "grad_norm": 2.2530438899993896,
      "learning_rate": 3.845169873106836e-06,
      "loss": 0.7231,
      "step": 20040
    },
    {
      "epoch": 8.21,
      "grad_norm": 3.2632648944854736,
      "learning_rate": 3.847088620548506e-06,
      "loss": 0.7335,
      "step": 20050
    },
    {
      "epoch": 8.21,
      "grad_norm": 2.34401535987854,
      "learning_rate": 3.849007367990176e-06,
      "loss": 0.7416,
      "step": 20060
    },
    {
      "epoch": 8.22,
      "grad_norm": 2.6299943923950195,
      "learning_rate": 3.850926115431846e-06,
      "loss": 0.7278,
      "step": 20070
    },
    {
      "epoch": 8.22,
      "grad_norm": 3.8678479194641113,
      "learning_rate": 3.8528448628735155e-06,
      "loss": 0.7249,
      "step": 20080
    },
    {
      "epoch": 8.22,
      "grad_norm": 1.9892170429229736,
      "learning_rate": 3.854763610315186e-06,
      "loss": 0.7412,
      "step": 20090
    },
    {
      "epoch": 8.23,
      "grad_norm": 2.1796672344207764,
      "learning_rate": 3.856682357756856e-06,
      "loss": 0.7295,
      "step": 20100
    },
    {
      "epoch": 8.23,
      "grad_norm": 2.625791072845459,
      "learning_rate": 3.858601105198526e-06,
      "loss": 0.7384,
      "step": 20110
    },
    {
      "epoch": 8.24,
      "grad_norm": 3.241093635559082,
      "learning_rate": 3.860519852640196e-06,
      "loss": 0.7393,
      "step": 20120
    },
    {
      "epoch": 8.24,
      "grad_norm": 2.435210943222046,
      "learning_rate": 3.8624386000818665e-06,
      "loss": 0.7389,
      "step": 20130
    },
    {
      "epoch": 8.24,
      "grad_norm": 2.2280890941619873,
      "learning_rate": 3.864357347523536e-06,
      "loss": 0.7182,
      "step": 20140
    },
    {
      "epoch": 8.25,
      "grad_norm": 3.72634220123291,
      "learning_rate": 3.866276094965206e-06,
      "loss": 0.7158,
      "step": 20150
    },
    {
      "epoch": 8.25,
      "grad_norm": 2.2097134590148926,
      "learning_rate": 3.868194842406877e-06,
      "loss": 0.7335,
      "step": 20160
    },
    {
      "epoch": 8.26,
      "grad_norm": 2.6305625438690186,
      "learning_rate": 3.870113589848546e-06,
      "loss": 0.7169,
      "step": 20170
    },
    {
      "epoch": 8.26,
      "grad_norm": 2.502753734588623,
      "learning_rate": 3.872032337290217e-06,
      "loss": 0.7209,
      "step": 20180
    },
    {
      "epoch": 8.26,
      "grad_norm": 3.42859148979187,
      "learning_rate": 3.873951084731887e-06,
      "loss": 0.7517,
      "step": 20190
    },
    {
      "epoch": 8.27,
      "grad_norm": 3.1996562480926514,
      "learning_rate": 3.8758698321735565e-06,
      "loss": 0.7133,
      "step": 20200
    },
    {
      "epoch": 8.27,
      "grad_norm": 2.6163480281829834,
      "learning_rate": 3.877788579615227e-06,
      "loss": 0.7358,
      "step": 20210
    },
    {
      "epoch": 8.28,
      "grad_norm": 2.4282922744750977,
      "learning_rate": 3.879707327056897e-06,
      "loss": 0.7213,
      "step": 20220
    },
    {
      "epoch": 8.28,
      "grad_norm": 3.0119805335998535,
      "learning_rate": 3.881626074498567e-06,
      "loss": 0.7213,
      "step": 20230
    },
    {
      "epoch": 8.28,
      "grad_norm": 1.9046536684036255,
      "learning_rate": 3.883544821940237e-06,
      "loss": 0.758,
      "step": 20240
    },
    {
      "epoch": 8.29,
      "grad_norm": 2.650923728942871,
      "learning_rate": 3.8854635693819075e-06,
      "loss": 0.7252,
      "step": 20250
    },
    {
      "epoch": 8.29,
      "grad_norm": 2.4393467903137207,
      "learning_rate": 3.887382316823577e-06,
      "loss": 0.7468,
      "step": 20260
    },
    {
      "epoch": 8.3,
      "grad_norm": 1.942947268486023,
      "learning_rate": 3.889301064265247e-06,
      "loss": 0.7107,
      "step": 20270
    },
    {
      "epoch": 8.3,
      "grad_norm": 4.347216606140137,
      "learning_rate": 3.891219811706918e-06,
      "loss": 0.7384,
      "step": 20280
    },
    {
      "epoch": 8.31,
      "grad_norm": 2.666201114654541,
      "learning_rate": 3.893138559148587e-06,
      "loss": 0.7225,
      "step": 20290
    },
    {
      "epoch": 8.31,
      "grad_norm": 2.7504570484161377,
      "learning_rate": 3.8950573065902576e-06,
      "loss": 0.7347,
      "step": 20300
    },
    {
      "epoch": 8.31,
      "grad_norm": 2.3514912128448486,
      "learning_rate": 3.896976054031928e-06,
      "loss": 0.7438,
      "step": 20310
    },
    {
      "epoch": 8.32,
      "grad_norm": 2.2555506229400635,
      "learning_rate": 3.8988948014735974e-06,
      "loss": 0.7127,
      "step": 20320
    },
    {
      "epoch": 8.32,
      "grad_norm": 3.2026984691619873,
      "learning_rate": 3.900813548915268e-06,
      "loss": 0.7156,
      "step": 20330
    },
    {
      "epoch": 8.33,
      "grad_norm": 3.060145854949951,
      "learning_rate": 3.902732296356938e-06,
      "loss": 0.7051,
      "step": 20340
    },
    {
      "epoch": 8.33,
      "grad_norm": 2.4551594257354736,
      "learning_rate": 3.904651043798608e-06,
      "loss": 0.7312,
      "step": 20350
    },
    {
      "epoch": 8.33,
      "grad_norm": 2.229356527328491,
      "learning_rate": 3.906569791240278e-06,
      "loss": 0.736,
      "step": 20360
    },
    {
      "epoch": 8.34,
      "grad_norm": 2.089663028717041,
      "learning_rate": 3.908488538681948e-06,
      "loss": 0.7125,
      "step": 20370
    },
    {
      "epoch": 8.34,
      "grad_norm": 3.623534679412842,
      "learning_rate": 3.910407286123618e-06,
      "loss": 0.7224,
      "step": 20380
    },
    {
      "epoch": 8.35,
      "grad_norm": 2.442401647567749,
      "learning_rate": 3.912326033565288e-06,
      "loss": 0.7351,
      "step": 20390
    },
    {
      "epoch": 8.35,
      "grad_norm": 3.378669261932373,
      "learning_rate": 3.914244781006959e-06,
      "loss": 0.7213,
      "step": 20400
    },
    {
      "epoch": 8.35,
      "grad_norm": 3.234954595565796,
      "learning_rate": 3.916163528448628e-06,
      "loss": 0.7266,
      "step": 20410
    },
    {
      "epoch": 8.36,
      "grad_norm": 1.927108883857727,
      "learning_rate": 3.9180822758902985e-06,
      "loss": 0.7322,
      "step": 20420
    },
    {
      "epoch": 8.36,
      "grad_norm": 2.763307809829712,
      "learning_rate": 3.920001023331969e-06,
      "loss": 0.7312,
      "step": 20430
    },
    {
      "epoch": 8.37,
      "grad_norm": 2.326509714126587,
      "learning_rate": 3.921919770773638e-06,
      "loss": 0.7406,
      "step": 20440
    },
    {
      "epoch": 8.37,
      "grad_norm": 3.3372039794921875,
      "learning_rate": 3.923838518215309e-06,
      "loss": 0.7372,
      "step": 20450
    },
    {
      "epoch": 8.37,
      "grad_norm": 1.9242358207702637,
      "learning_rate": 3.925757265656979e-06,
      "loss": 0.7373,
      "step": 20460
    },
    {
      "epoch": 8.38,
      "grad_norm": 2.0390613079071045,
      "learning_rate": 3.927676013098649e-06,
      "loss": 0.7221,
      "step": 20470
    },
    {
      "epoch": 8.38,
      "grad_norm": 2.4324305057525635,
      "learning_rate": 3.929594760540319e-06,
      "loss": 0.7197,
      "step": 20480
    },
    {
      "epoch": 8.39,
      "grad_norm": 3.3997457027435303,
      "learning_rate": 3.931513507981989e-06,
      "loss": 0.7338,
      "step": 20490
    },
    {
      "epoch": 8.39,
      "grad_norm": 2.174779176712036,
      "learning_rate": 3.933432255423659e-06,
      "loss": 0.7498,
      "step": 20500
    },
    {
      "epoch": 8.4,
      "grad_norm": 2.2485849857330322,
      "learning_rate": 3.935351002865329e-06,
      "loss": 0.7068,
      "step": 20510
    },
    {
      "epoch": 8.4,
      "grad_norm": 2.777301788330078,
      "learning_rate": 3.937269750306999e-06,
      "loss": 0.7334,
      "step": 20520
    },
    {
      "epoch": 8.4,
      "grad_norm": 2.0368428230285645,
      "learning_rate": 3.939188497748669e-06,
      "loss": 0.7294,
      "step": 20530
    },
    {
      "epoch": 8.41,
      "grad_norm": 3.2429988384246826,
      "learning_rate": 3.9411072451903395e-06,
      "loss": 0.7288,
      "step": 20540
    },
    {
      "epoch": 8.41,
      "grad_norm": 3.6201446056365967,
      "learning_rate": 3.943025992632009e-06,
      "loss": 0.722,
      "step": 20550
    },
    {
      "epoch": 8.42,
      "grad_norm": 1.875707983970642,
      "learning_rate": 3.944944740073679e-06,
      "loss": 0.7071,
      "step": 20560
    },
    {
      "epoch": 8.42,
      "grad_norm": 1.9971251487731934,
      "learning_rate": 3.94686348751535e-06,
      "loss": 0.7202,
      "step": 20570
    },
    {
      "epoch": 8.42,
      "grad_norm": 2.977357864379883,
      "learning_rate": 3.948782234957019e-06,
      "loss": 0.7393,
      "step": 20580
    },
    {
      "epoch": 8.43,
      "grad_norm": 2.661376476287842,
      "learning_rate": 3.95070098239869e-06,
      "loss": 0.7192,
      "step": 20590
    },
    {
      "epoch": 8.43,
      "grad_norm": 1.8311821222305298,
      "learning_rate": 3.95261972984036e-06,
      "loss": 0.7239,
      "step": 20600
    },
    {
      "epoch": 8.44,
      "grad_norm": 3.034759759902954,
      "learning_rate": 3.9545384772820295e-06,
      "loss": 0.7313,
      "step": 20610
    },
    {
      "epoch": 8.44,
      "grad_norm": 2.759415626525879,
      "learning_rate": 3.9564572247237e-06,
      "loss": 0.7404,
      "step": 20620
    },
    {
      "epoch": 8.44,
      "grad_norm": 3.043029308319092,
      "learning_rate": 3.95837597216537e-06,
      "loss": 0.7296,
      "step": 20630
    },
    {
      "epoch": 8.45,
      "grad_norm": 2.7643020153045654,
      "learning_rate": 3.96029471960704e-06,
      "loss": 0.727,
      "step": 20640
    },
    {
      "epoch": 8.45,
      "grad_norm": 2.7193267345428467,
      "learning_rate": 3.96221346704871e-06,
      "loss": 0.7022,
      "step": 20650
    },
    {
      "epoch": 8.46,
      "grad_norm": 2.4776768684387207,
      "learning_rate": 3.9641322144903805e-06,
      "loss": 0.7138,
      "step": 20660
    },
    {
      "epoch": 8.46,
      "grad_norm": 3.718618392944336,
      "learning_rate": 3.96605096193205e-06,
      "loss": 0.7236,
      "step": 20670
    },
    {
      "epoch": 8.47,
      "grad_norm": 2.3279030323028564,
      "learning_rate": 3.96796970937372e-06,
      "loss": 0.7276,
      "step": 20680
    },
    {
      "epoch": 8.47,
      "grad_norm": 2.1661293506622314,
      "learning_rate": 3.969888456815391e-06,
      "loss": 0.7295,
      "step": 20690
    },
    {
      "epoch": 8.47,
      "grad_norm": 2.5474932193756104,
      "learning_rate": 3.97180720425706e-06,
      "loss": 0.7333,
      "step": 20700
    },
    {
      "epoch": 8.48,
      "grad_norm": 2.467068910598755,
      "learning_rate": 3.973725951698731e-06,
      "loss": 0.7154,
      "step": 20710
    },
    {
      "epoch": 8.48,
      "grad_norm": 2.110631227493286,
      "learning_rate": 3.975644699140401e-06,
      "loss": 0.7262,
      "step": 20720
    },
    {
      "epoch": 8.49,
      "grad_norm": 2.7104814052581787,
      "learning_rate": 3.9775634465820705e-06,
      "loss": 0.7195,
      "step": 20730
    },
    {
      "epoch": 8.49,
      "grad_norm": 2.4017930030822754,
      "learning_rate": 3.979482194023741e-06,
      "loss": 0.7404,
      "step": 20740
    },
    {
      "epoch": 8.49,
      "grad_norm": 1.8047349452972412,
      "learning_rate": 3.981400941465411e-06,
      "loss": 0.715,
      "step": 20750
    },
    {
      "epoch": 8.5,
      "grad_norm": 2.094972610473633,
      "learning_rate": 3.983319688907081e-06,
      "loss": 0.721,
      "step": 20760
    },
    {
      "epoch": 8.5,
      "grad_norm": 3.1265017986297607,
      "learning_rate": 3.985238436348751e-06,
      "loss": 0.7292,
      "step": 20770
    },
    {
      "epoch": 8.51,
      "grad_norm": 2.696521282196045,
      "learning_rate": 3.987157183790421e-06,
      "loss": 0.7089,
      "step": 20780
    },
    {
      "epoch": 8.51,
      "grad_norm": 2.5515353679656982,
      "learning_rate": 3.989075931232091e-06,
      "loss": 0.7267,
      "step": 20790
    },
    {
      "epoch": 8.51,
      "grad_norm": 2.281379222869873,
      "learning_rate": 3.990994678673761e-06,
      "loss": 0.7193,
      "step": 20800
    },
    {
      "epoch": 8.52,
      "grad_norm": 2.0666353702545166,
      "learning_rate": 3.992913426115432e-06,
      "loss": 0.7321,
      "step": 20810
    },
    {
      "epoch": 8.52,
      "grad_norm": 2.129612684249878,
      "learning_rate": 3.994832173557101e-06,
      "loss": 0.7091,
      "step": 20820
    },
    {
      "epoch": 8.53,
      "grad_norm": 2.1035525798797607,
      "learning_rate": 3.9967509209987715e-06,
      "loss": 0.7399,
      "step": 20830
    },
    {
      "epoch": 8.53,
      "grad_norm": 4.1486334800720215,
      "learning_rate": 3.998669668440442e-06,
      "loss": 0.7343,
      "step": 20840
    },
    {
      "epoch": 8.53,
      "grad_norm": 2.4414174556732178,
      "learning_rate": 4.000588415882111e-06,
      "loss": 0.7279,
      "step": 20850
    },
    {
      "epoch": 8.54,
      "grad_norm": 2.476081371307373,
      "learning_rate": 4.002507163323782e-06,
      "loss": 0.7256,
      "step": 20860
    },
    {
      "epoch": 8.54,
      "grad_norm": 2.8631772994995117,
      "learning_rate": 4.004425910765452e-06,
      "loss": 0.7312,
      "step": 20870
    },
    {
      "epoch": 8.55,
      "grad_norm": 2.940929651260376,
      "learning_rate": 4.006344658207122e-06,
      "loss": 0.7018,
      "step": 20880
    },
    {
      "epoch": 8.55,
      "grad_norm": 4.535207748413086,
      "learning_rate": 4.008263405648792e-06,
      "loss": 0.7419,
      "step": 20890
    },
    {
      "epoch": 8.56,
      "grad_norm": 2.1698646545410156,
      "learning_rate": 4.010182153090462e-06,
      "loss": 0.7143,
      "step": 20900
    },
    {
      "epoch": 8.56,
      "grad_norm": 2.6687281131744385,
      "learning_rate": 4.012100900532132e-06,
      "loss": 0.7262,
      "step": 20910
    },
    {
      "epoch": 8.56,
      "grad_norm": 2.3931009769439697,
      "learning_rate": 4.014019647973802e-06,
      "loss": 0.7105,
      "step": 20920
    },
    {
      "epoch": 8.57,
      "grad_norm": 2.4640796184539795,
      "learning_rate": 4.015938395415473e-06,
      "loss": 0.7244,
      "step": 20930
    },
    {
      "epoch": 8.57,
      "grad_norm": 2.844505786895752,
      "learning_rate": 4.017857142857142e-06,
      "loss": 0.7378,
      "step": 20940
    },
    {
      "epoch": 8.58,
      "grad_norm": 1.877651572227478,
      "learning_rate": 4.0197758902988125e-06,
      "loss": 0.7317,
      "step": 20950
    },
    {
      "epoch": 8.58,
      "grad_norm": 1.6317135095596313,
      "learning_rate": 4.021694637740482e-06,
      "loss": 0.7151,
      "step": 20960
    },
    {
      "epoch": 8.58,
      "grad_norm": 2.6988914012908936,
      "learning_rate": 4.023613385182152e-06,
      "loss": 0.7186,
      "step": 20970
    },
    {
      "epoch": 8.59,
      "grad_norm": 2.094912528991699,
      "learning_rate": 4.025532132623823e-06,
      "loss": 0.7208,
      "step": 20980
    },
    {
      "epoch": 8.59,
      "grad_norm": 1.6666728258132935,
      "learning_rate": 4.027450880065492e-06,
      "loss": 0.7446,
      "step": 20990
    },
    {
      "epoch": 8.6,
      "grad_norm": 2.9577677249908447,
      "learning_rate": 4.029369627507163e-06,
      "loss": 0.7316,
      "step": 21000
    },
    {
      "epoch": 8.6,
      "grad_norm": 2.6536056995391846,
      "learning_rate": 4.031288374948833e-06,
      "loss": 0.7208,
      "step": 21010
    },
    {
      "epoch": 8.6,
      "grad_norm": 1.79371178150177,
      "learning_rate": 4.0332071223905025e-06,
      "loss": 0.7251,
      "step": 21020
    },
    {
      "epoch": 8.61,
      "grad_norm": 2.71095609664917,
      "learning_rate": 4.035125869832173e-06,
      "loss": 0.7141,
      "step": 21030
    },
    {
      "epoch": 8.61,
      "grad_norm": 2.880289316177368,
      "learning_rate": 4.037044617273843e-06,
      "loss": 0.7204,
      "step": 21040
    },
    {
      "epoch": 8.62,
      "grad_norm": 1.9460537433624268,
      "learning_rate": 4.038963364715514e-06,
      "loss": 0.7217,
      "step": 21050
    },
    {
      "epoch": 8.62,
      "grad_norm": 1.582702398300171,
      "learning_rate": 4.040882112157184e-06,
      "loss": 0.7102,
      "step": 21060
    },
    {
      "epoch": 8.62,
      "grad_norm": 1.793157935142517,
      "learning_rate": 4.0428008595988535e-06,
      "loss": 0.745,
      "step": 21070
    },
    {
      "epoch": 8.63,
      "grad_norm": 2.3298089504241943,
      "learning_rate": 4.044719607040524e-06,
      "loss": 0.7241,
      "step": 21080
    },
    {
      "epoch": 8.63,
      "grad_norm": 1.9684972763061523,
      "learning_rate": 4.046638354482194e-06,
      "loss": 0.7296,
      "step": 21090
    },
    {
      "epoch": 8.64,
      "grad_norm": 2.369952440261841,
      "learning_rate": 4.048557101923864e-06,
      "loss": 0.7218,
      "step": 21100
    },
    {
      "epoch": 8.64,
      "grad_norm": 2.8646955490112305,
      "learning_rate": 4.050475849365534e-06,
      "loss": 0.7121,
      "step": 21110
    },
    {
      "epoch": 8.65,
      "grad_norm": 3.122514009475708,
      "learning_rate": 4.0523945968072044e-06,
      "loss": 0.7379,
      "step": 21120
    },
    {
      "epoch": 8.65,
      "grad_norm": 3.1944386959075928,
      "learning_rate": 4.054313344248874e-06,
      "loss": 0.7148,
      "step": 21130
    },
    {
      "epoch": 8.65,
      "grad_norm": 2.9429609775543213,
      "learning_rate": 4.056232091690544e-06,
      "loss": 0.7054,
      "step": 21140
    },
    {
      "epoch": 8.66,
      "grad_norm": 3.037121057510376,
      "learning_rate": 4.058150839132215e-06,
      "loss": 0.7098,
      "step": 21150
    },
    {
      "epoch": 8.66,
      "grad_norm": 2.333237648010254,
      "learning_rate": 4.060069586573884e-06,
      "loss": 0.7195,
      "step": 21160
    },
    {
      "epoch": 8.67,
      "grad_norm": 2.7827398777008057,
      "learning_rate": 4.0619883340155546e-06,
      "loss": 0.7339,
      "step": 21170
    },
    {
      "epoch": 8.67,
      "grad_norm": 4.795051097869873,
      "learning_rate": 4.063907081457225e-06,
      "loss": 0.7297,
      "step": 21180
    },
    {
      "epoch": 8.67,
      "grad_norm": 2.4155845642089844,
      "learning_rate": 4.0658258288988944e-06,
      "loss": 0.7391,
      "step": 21190
    },
    {
      "epoch": 8.68,
      "grad_norm": 2.614243984222412,
      "learning_rate": 4.067744576340565e-06,
      "loss": 0.7385,
      "step": 21200
    },
    {
      "epoch": 8.68,
      "grad_norm": 1.98758864402771,
      "learning_rate": 4.069663323782235e-06,
      "loss": 0.7418,
      "step": 21210
    },
    {
      "epoch": 8.69,
      "grad_norm": 1.9135496616363525,
      "learning_rate": 4.071582071223905e-06,
      "loss": 0.7298,
      "step": 21220
    },
    {
      "epoch": 8.69,
      "grad_norm": 3.142744302749634,
      "learning_rate": 4.073500818665575e-06,
      "loss": 0.7234,
      "step": 21230
    },
    {
      "epoch": 8.69,
      "grad_norm": 2.776364803314209,
      "learning_rate": 4.075419566107245e-06,
      "loss": 0.7195,
      "step": 21240
    },
    {
      "epoch": 8.7,
      "grad_norm": 2.9243950843811035,
      "learning_rate": 4.077338313548915e-06,
      "loss": 0.7071,
      "step": 21250
    },
    {
      "epoch": 8.7,
      "grad_norm": 1.8371212482452393,
      "learning_rate": 4.079257060990585e-06,
      "loss": 0.7213,
      "step": 21260
    },
    {
      "epoch": 8.71,
      "grad_norm": 2.629533052444458,
      "learning_rate": 4.081175808432256e-06,
      "loss": 0.7275,
      "step": 21270
    },
    {
      "epoch": 8.71,
      "grad_norm": 2.1641323566436768,
      "learning_rate": 4.083094555873925e-06,
      "loss": 0.7265,
      "step": 21280
    },
    {
      "epoch": 8.71,
      "grad_norm": 1.3506728410720825,
      "learning_rate": 4.0850133033155955e-06,
      "loss": 0.7408,
      "step": 21290
    },
    {
      "epoch": 8.72,
      "grad_norm": 1.9706242084503174,
      "learning_rate": 4.086932050757266e-06,
      "loss": 0.723,
      "step": 21300
    },
    {
      "epoch": 8.72,
      "grad_norm": 3.390139102935791,
      "learning_rate": 4.088850798198935e-06,
      "loss": 0.706,
      "step": 21310
    },
    {
      "epoch": 8.73,
      "grad_norm": 1.9741244316101074,
      "learning_rate": 4.090769545640606e-06,
      "loss": 0.7003,
      "step": 21320
    },
    {
      "epoch": 8.73,
      "grad_norm": 1.9972070455551147,
      "learning_rate": 4.092688293082275e-06,
      "loss": 0.7213,
      "step": 21330
    },
    {
      "epoch": 8.74,
      "grad_norm": 3.066802501678467,
      "learning_rate": 4.094607040523946e-06,
      "loss": 0.7363,
      "step": 21340
    },
    {
      "epoch": 8.74,
      "grad_norm": 1.8371870517730713,
      "learning_rate": 4.096525787965616e-06,
      "loss": 0.7119,
      "step": 21350
    },
    {
      "epoch": 8.74,
      "grad_norm": 3.4932351112365723,
      "learning_rate": 4.0984445354072855e-06,
      "loss": 0.7274,
      "step": 21360
    },
    {
      "epoch": 8.75,
      "grad_norm": 1.7755094766616821,
      "learning_rate": 4.100363282848956e-06,
      "loss": 0.7289,
      "step": 21370
    },
    {
      "epoch": 8.75,
      "grad_norm": 2.2712507247924805,
      "learning_rate": 4.102282030290626e-06,
      "loss": 0.7257,
      "step": 21380
    },
    {
      "epoch": 8.76,
      "grad_norm": 3.040832996368408,
      "learning_rate": 4.104200777732296e-06,
      "loss": 0.7402,
      "step": 21390
    },
    {
      "epoch": 8.76,
      "grad_norm": 2.075758695602417,
      "learning_rate": 4.106119525173966e-06,
      "loss": 0.7111,
      "step": 21400
    },
    {
      "epoch": 8.76,
      "grad_norm": 1.9851584434509277,
      "learning_rate": 4.1080382726156365e-06,
      "loss": 0.7104,
      "step": 21410
    },
    {
      "epoch": 8.77,
      "grad_norm": 2.715775489807129,
      "learning_rate": 4.109957020057306e-06,
      "loss": 0.7173,
      "step": 21420
    },
    {
      "epoch": 8.77,
      "grad_norm": 2.043574571609497,
      "learning_rate": 4.111875767498976e-06,
      "loss": 0.7241,
      "step": 21430
    },
    {
      "epoch": 8.78,
      "grad_norm": 2.3404693603515625,
      "learning_rate": 4.113794514940647e-06,
      "loss": 0.7412,
      "step": 21440
    },
    {
      "epoch": 8.78,
      "grad_norm": 2.2759346961975098,
      "learning_rate": 4.115713262382316e-06,
      "loss": 0.7375,
      "step": 21450
    },
    {
      "epoch": 8.78,
      "grad_norm": 3.3243088722229004,
      "learning_rate": 4.117632009823987e-06,
      "loss": 0.746,
      "step": 21460
    },
    {
      "epoch": 8.79,
      "grad_norm": 3.271702289581299,
      "learning_rate": 4.119550757265657e-06,
      "loss": 0.7265,
      "step": 21470
    },
    {
      "epoch": 8.79,
      "grad_norm": 1.958943486213684,
      "learning_rate": 4.1214695047073265e-06,
      "loss": 0.7157,
      "step": 21480
    },
    {
      "epoch": 8.8,
      "grad_norm": 1.7642419338226318,
      "learning_rate": 4.123388252148997e-06,
      "loss": 0.7115,
      "step": 21490
    },
    {
      "epoch": 8.8,
      "grad_norm": 2.7286362648010254,
      "learning_rate": 4.125306999590667e-06,
      "loss": 0.7366,
      "step": 21500
    },
    {
      "epoch": 8.8,
      "grad_norm": 2.184067487716675,
      "learning_rate": 4.127225747032337e-06,
      "loss": 0.7281,
      "step": 21510
    },
    {
      "epoch": 8.81,
      "grad_norm": 3.0401461124420166,
      "learning_rate": 4.129144494474007e-06,
      "loss": 0.7307,
      "step": 21520
    },
    {
      "epoch": 8.81,
      "grad_norm": 1.791648507118225,
      "learning_rate": 4.1310632419156774e-06,
      "loss": 0.7272,
      "step": 21530
    },
    {
      "epoch": 8.82,
      "grad_norm": 2.324040174484253,
      "learning_rate": 4.132981989357347e-06,
      "loss": 0.7237,
      "step": 21540
    },
    {
      "epoch": 8.82,
      "grad_norm": 3.326018810272217,
      "learning_rate": 4.134900736799017e-06,
      "loss": 0.7351,
      "step": 21550
    },
    {
      "epoch": 8.83,
      "grad_norm": 2.788395881652832,
      "learning_rate": 4.136819484240688e-06,
      "loss": 0.7186,
      "step": 21560
    },
    {
      "epoch": 8.83,
      "grad_norm": 2.7202885150909424,
      "learning_rate": 4.138738231682357e-06,
      "loss": 0.7272,
      "step": 21570
    },
    {
      "epoch": 8.83,
      "grad_norm": 2.43361234664917,
      "learning_rate": 4.1406569791240276e-06,
      "loss": 0.7084,
      "step": 21580
    },
    {
      "epoch": 8.84,
      "grad_norm": 2.7894279956817627,
      "learning_rate": 4.142575726565698e-06,
      "loss": 0.7283,
      "step": 21590
    },
    {
      "epoch": 8.84,
      "grad_norm": 2.618901491165161,
      "learning_rate": 4.1444944740073674e-06,
      "loss": 0.7174,
      "step": 21600
    },
    {
      "epoch": 8.85,
      "grad_norm": 2.17155122756958,
      "learning_rate": 4.146413221449038e-06,
      "loss": 0.7085,
      "step": 21610
    },
    {
      "epoch": 8.85,
      "grad_norm": 2.977491855621338,
      "learning_rate": 4.148331968890708e-06,
      "loss": 0.7033,
      "step": 21620
    },
    {
      "epoch": 8.85,
      "grad_norm": 2.1285035610198975,
      "learning_rate": 4.150250716332378e-06,
      "loss": 0.7343,
      "step": 21630
    },
    {
      "epoch": 8.86,
      "grad_norm": 3.194075345993042,
      "learning_rate": 4.152169463774048e-06,
      "loss": 0.7241,
      "step": 21640
    },
    {
      "epoch": 8.86,
      "grad_norm": 2.8879151344299316,
      "learning_rate": 4.154088211215718e-06,
      "loss": 0.732,
      "step": 21650
    },
    {
      "epoch": 8.87,
      "grad_norm": 3.1293957233428955,
      "learning_rate": 4.156006958657388e-06,
      "loss": 0.7284,
      "step": 21660
    },
    {
      "epoch": 8.87,
      "grad_norm": 1.9909446239471436,
      "learning_rate": 4.157925706099058e-06,
      "loss": 0.7248,
      "step": 21670
    },
    {
      "epoch": 8.87,
      "grad_norm": 2.502197265625,
      "learning_rate": 4.159844453540729e-06,
      "loss": 0.7123,
      "step": 21680
    },
    {
      "epoch": 8.88,
      "grad_norm": 1.3876023292541504,
      "learning_rate": 4.161763200982398e-06,
      "loss": 0.7196,
      "step": 21690
    },
    {
      "epoch": 8.88,
      "grad_norm": 2.4209847450256348,
      "learning_rate": 4.1636819484240685e-06,
      "loss": 0.7128,
      "step": 21700
    },
    {
      "epoch": 8.89,
      "grad_norm": 3.996748924255371,
      "learning_rate": 4.165600695865739e-06,
      "loss": 0.7172,
      "step": 21710
    },
    {
      "epoch": 8.89,
      "grad_norm": 2.4194164276123047,
      "learning_rate": 4.167519443307408e-06,
      "loss": 0.7236,
      "step": 21720
    },
    {
      "epoch": 8.89,
      "grad_norm": 3.1892812252044678,
      "learning_rate": 4.169438190749079e-06,
      "loss": 0.7173,
      "step": 21730
    },
    {
      "epoch": 8.9,
      "grad_norm": 2.6256115436553955,
      "learning_rate": 4.171356938190748e-06,
      "loss": 0.7254,
      "step": 21740
    },
    {
      "epoch": 8.9,
      "grad_norm": 4.017436981201172,
      "learning_rate": 4.173275685632419e-06,
      "loss": 0.7065,
      "step": 21750
    },
    {
      "epoch": 8.91,
      "grad_norm": 3.2507646083831787,
      "learning_rate": 4.175194433074089e-06,
      "loss": 0.7142,
      "step": 21760
    },
    {
      "epoch": 8.91,
      "grad_norm": 2.3758420944213867,
      "learning_rate": 4.1771131805157585e-06,
      "loss": 0.7225,
      "step": 21770
    },
    {
      "epoch": 8.92,
      "grad_norm": 3.0346741676330566,
      "learning_rate": 4.179031927957429e-06,
      "loss": 0.7246,
      "step": 21780
    },
    {
      "epoch": 8.92,
      "grad_norm": 3.8513708114624023,
      "learning_rate": 4.180950675399099e-06,
      "loss": 0.706,
      "step": 21790
    },
    {
      "epoch": 8.92,
      "grad_norm": 3.149413824081421,
      "learning_rate": 4.182869422840769e-06,
      "loss": 0.7161,
      "step": 21800
    },
    {
      "epoch": 8.93,
      "grad_norm": 2.459153652191162,
      "learning_rate": 4.184788170282439e-06,
      "loss": 0.7143,
      "step": 21810
    },
    {
      "epoch": 8.93,
      "grad_norm": 2.1341023445129395,
      "learning_rate": 4.1867069177241095e-06,
      "loss": 0.7042,
      "step": 21820
    },
    {
      "epoch": 8.94,
      "grad_norm": 2.0369839668273926,
      "learning_rate": 4.188625665165779e-06,
      "loss": 0.7252,
      "step": 21830
    },
    {
      "epoch": 8.94,
      "grad_norm": 3.0848796367645264,
      "learning_rate": 4.190544412607449e-06,
      "loss": 0.7156,
      "step": 21840
    },
    {
      "epoch": 8.94,
      "grad_norm": 2.6845169067382812,
      "learning_rate": 4.19246316004912e-06,
      "loss": 0.707,
      "step": 21850
    },
    {
      "epoch": 8.95,
      "grad_norm": 3.9522571563720703,
      "learning_rate": 4.194381907490789e-06,
      "loss": 0.7351,
      "step": 21860
    },
    {
      "epoch": 8.95,
      "grad_norm": 1.4499951601028442,
      "learning_rate": 4.19630065493246e-06,
      "loss": 0.7305,
      "step": 21870
    },
    {
      "epoch": 8.96,
      "grad_norm": 2.1964919567108154,
      "learning_rate": 4.19821940237413e-06,
      "loss": 0.7138,
      "step": 21880
    },
    {
      "epoch": 8.96,
      "grad_norm": 2.1569950580596924,
      "learning_rate": 4.2001381498157995e-06,
      "loss": 0.7315,
      "step": 21890
    },
    {
      "epoch": 8.96,
      "grad_norm": 2.6480350494384766,
      "learning_rate": 4.20205689725747e-06,
      "loss": 0.7017,
      "step": 21900
    },
    {
      "epoch": 8.97,
      "grad_norm": 2.143859624862671,
      "learning_rate": 4.20397564469914e-06,
      "loss": 0.7142,
      "step": 21910
    },
    {
      "epoch": 8.97,
      "grad_norm": 2.197686195373535,
      "learning_rate": 4.20589439214081e-06,
      "loss": 0.7113,
      "step": 21920
    },
    {
      "epoch": 8.98,
      "grad_norm": 2.1893532276153564,
      "learning_rate": 4.20781313958248e-06,
      "loss": 0.7347,
      "step": 21930
    },
    {
      "epoch": 8.98,
      "grad_norm": 3.0514943599700928,
      "learning_rate": 4.2097318870241505e-06,
      "loss": 0.732,
      "step": 21940
    },
    {
      "epoch": 8.98,
      "grad_norm": 2.904005527496338,
      "learning_rate": 4.21165063446582e-06,
      "loss": 0.714,
      "step": 21950
    },
    {
      "epoch": 8.99,
      "grad_norm": 2.258946418762207,
      "learning_rate": 4.21356938190749e-06,
      "loss": 0.6936,
      "step": 21960
    },
    {
      "epoch": 8.99,
      "grad_norm": 1.6918946504592896,
      "learning_rate": 4.215488129349161e-06,
      "loss": 0.7233,
      "step": 21970
    },
    {
      "epoch": 9.0,
      "grad_norm": 2.1126232147216797,
      "learning_rate": 4.21740687679083e-06,
      "loss": 0.6841,
      "step": 21980
    },
    {
      "epoch": 9.0,
      "eval_loss": 0.7133491039276123,
      "eval_runtime": 52.3045,
      "eval_samples_per_second": 65.941,
      "eval_steps_per_second": 8.259,
      "step": 21987
    },
    {
      "epoch": 9.0,
      "grad_norm": 2.5512614250183105,
      "learning_rate": 4.2193256242325006e-06,
      "loss": 0.7195,
      "step": 21990
    },
    {
      "epoch": 9.01,
      "grad_norm": 2.4196958541870117,
      "learning_rate": 4.221244371674171e-06,
      "loss": 0.7009,
      "step": 22000
    },
    {
      "epoch": 9.01,
      "grad_norm": 2.3718903064727783,
      "learning_rate": 4.2231631191158404e-06,
      "loss": 0.731,
      "step": 22010
    },
    {
      "epoch": 9.01,
      "grad_norm": 2.3349390029907227,
      "learning_rate": 4.225081866557511e-06,
      "loss": 0.7253,
      "step": 22020
    },
    {
      "epoch": 9.02,
      "grad_norm": 3.1551074981689453,
      "learning_rate": 4.227000613999181e-06,
      "loss": 0.7421,
      "step": 22030
    },
    {
      "epoch": 9.02,
      "grad_norm": 1.931774377822876,
      "learning_rate": 4.228919361440851e-06,
      "loss": 0.7188,
      "step": 22040
    },
    {
      "epoch": 9.03,
      "grad_norm": 2.2866690158843994,
      "learning_rate": 4.230838108882521e-06,
      "loss": 0.715,
      "step": 22050
    },
    {
      "epoch": 9.03,
      "grad_norm": 4.015485763549805,
      "learning_rate": 4.232756856324191e-06,
      "loss": 0.7253,
      "step": 22060
    },
    {
      "epoch": 9.03,
      "grad_norm": 3.6578426361083984,
      "learning_rate": 4.234675603765861e-06,
      "loss": 0.697,
      "step": 22070
    },
    {
      "epoch": 9.04,
      "grad_norm": 1.967729926109314,
      "learning_rate": 4.236594351207531e-06,
      "loss": 0.7368,
      "step": 22080
    },
    {
      "epoch": 9.04,
      "grad_norm": 2.207995891571045,
      "learning_rate": 4.238513098649202e-06,
      "loss": 0.7065,
      "step": 22090
    },
    {
      "epoch": 9.05,
      "grad_norm": 1.9597671031951904,
      "learning_rate": 4.240431846090871e-06,
      "loss": 0.7076,
      "step": 22100
    },
    {
      "epoch": 9.05,
      "grad_norm": 2.3054440021514893,
      "learning_rate": 4.2423505935325415e-06,
      "loss": 0.7126,
      "step": 22110
    },
    {
      "epoch": 9.05,
      "grad_norm": 1.7577553987503052,
      "learning_rate": 4.244269340974212e-06,
      "loss": 0.7331,
      "step": 22120
    },
    {
      "epoch": 9.06,
      "grad_norm": 2.7117369174957275,
      "learning_rate": 4.246188088415881e-06,
      "loss": 0.7163,
      "step": 22130
    },
    {
      "epoch": 9.06,
      "grad_norm": 2.6443862915039062,
      "learning_rate": 4.248106835857552e-06,
      "loss": 0.7164,
      "step": 22140
    },
    {
      "epoch": 9.07,
      "grad_norm": 2.5779972076416016,
      "learning_rate": 4.250025583299222e-06,
      "loss": 0.7096,
      "step": 22150
    },
    {
      "epoch": 9.07,
      "grad_norm": 2.2581629753112793,
      "learning_rate": 4.251944330740892e-06,
      "loss": 0.7181,
      "step": 22160
    },
    {
      "epoch": 9.07,
      "grad_norm": 2.4101061820983887,
      "learning_rate": 4.253863078182562e-06,
      "loss": 0.7205,
      "step": 22170
    },
    {
      "epoch": 9.08,
      "grad_norm": 3.243023157119751,
      "learning_rate": 4.2557818256242315e-06,
      "loss": 0.7106,
      "step": 22180
    },
    {
      "epoch": 9.08,
      "grad_norm": 2.4235546588897705,
      "learning_rate": 4.257700573065902e-06,
      "loss": 0.7034,
      "step": 22190
    },
    {
      "epoch": 9.09,
      "grad_norm": 3.058833599090576,
      "learning_rate": 4.259619320507572e-06,
      "loss": 0.7084,
      "step": 22200
    },
    {
      "epoch": 9.09,
      "grad_norm": 2.746656656265259,
      "learning_rate": 4.261538067949242e-06,
      "loss": 0.7049,
      "step": 22210
    },
    {
      "epoch": 9.1,
      "grad_norm": 2.2618253231048584,
      "learning_rate": 4.263456815390912e-06,
      "loss": 0.7398,
      "step": 22220
    },
    {
      "epoch": 9.1,
      "grad_norm": 2.552424907684326,
      "learning_rate": 4.2653755628325825e-06,
      "loss": 0.7247,
      "step": 22230
    },
    {
      "epoch": 9.1,
      "grad_norm": 1.595353364944458,
      "learning_rate": 4.267294310274252e-06,
      "loss": 0.7335,
      "step": 22240
    },
    {
      "epoch": 9.11,
      "grad_norm": 3.1151340007781982,
      "learning_rate": 4.269213057715922e-06,
      "loss": 0.7215,
      "step": 22250
    },
    {
      "epoch": 9.11,
      "grad_norm": 2.286672592163086,
      "learning_rate": 4.271131805157593e-06,
      "loss": 0.7134,
      "step": 22260
    },
    {
      "epoch": 9.12,
      "grad_norm": 2.4985768795013428,
      "learning_rate": 4.273050552599262e-06,
      "loss": 0.7002,
      "step": 22270
    },
    {
      "epoch": 9.12,
      "grad_norm": 3.0474138259887695,
      "learning_rate": 4.274969300040933e-06,
      "loss": 0.7229,
      "step": 22280
    },
    {
      "epoch": 9.12,
      "grad_norm": 2.7329602241516113,
      "learning_rate": 4.276888047482603e-06,
      "loss": 0.7159,
      "step": 22290
    },
    {
      "epoch": 9.13,
      "grad_norm": 2.3272902965545654,
      "learning_rate": 4.2788067949242725e-06,
      "loss": 0.7146,
      "step": 22300
    },
    {
      "epoch": 9.13,
      "grad_norm": 2.527872085571289,
      "learning_rate": 4.280725542365943e-06,
      "loss": 0.7284,
      "step": 22310
    },
    {
      "epoch": 9.14,
      "grad_norm": 2.4518818855285645,
      "learning_rate": 4.282644289807613e-06,
      "loss": 0.7105,
      "step": 22320
    },
    {
      "epoch": 9.14,
      "grad_norm": 3.238588333129883,
      "learning_rate": 4.284563037249283e-06,
      "loss": 0.7287,
      "step": 22330
    },
    {
      "epoch": 9.14,
      "grad_norm": 2.2045576572418213,
      "learning_rate": 4.286481784690953e-06,
      "loss": 0.7218,
      "step": 22340
    },
    {
      "epoch": 9.15,
      "grad_norm": 2.443382978439331,
      "learning_rate": 4.2884005321326235e-06,
      "loss": 0.727,
      "step": 22350
    },
    {
      "epoch": 9.15,
      "grad_norm": 2.0073421001434326,
      "learning_rate": 4.290319279574293e-06,
      "loss": 0.7388,
      "step": 22360
    },
    {
      "epoch": 9.16,
      "grad_norm": 3.0396225452423096,
      "learning_rate": 4.292238027015963e-06,
      "loss": 0.7211,
      "step": 22370
    },
    {
      "epoch": 9.16,
      "grad_norm": 2.654780864715576,
      "learning_rate": 4.294156774457634e-06,
      "loss": 0.7351,
      "step": 22380
    },
    {
      "epoch": 9.16,
      "grad_norm": 3.633798599243164,
      "learning_rate": 4.296075521899303e-06,
      "loss": 0.7233,
      "step": 22390
    },
    {
      "epoch": 9.17,
      "grad_norm": 3.0976569652557373,
      "learning_rate": 4.2979942693409744e-06,
      "loss": 0.7026,
      "step": 22400
    },
    {
      "epoch": 9.17,
      "grad_norm": 2.121330499649048,
      "learning_rate": 4.299913016782644e-06,
      "loss": 0.7057,
      "step": 22410
    },
    {
      "epoch": 9.18,
      "grad_norm": 2.1180713176727295,
      "learning_rate": 4.301831764224314e-06,
      "loss": 0.7255,
      "step": 22420
    },
    {
      "epoch": 9.18,
      "grad_norm": 2.807095527648926,
      "learning_rate": 4.303750511665985e-06,
      "loss": 0.7173,
      "step": 22430
    },
    {
      "epoch": 9.19,
      "grad_norm": 2.731031894683838,
      "learning_rate": 4.305669259107654e-06,
      "loss": 0.7073,
      "step": 22440
    },
    {
      "epoch": 9.19,
      "grad_norm": 3.3777427673339844,
      "learning_rate": 4.3075880065493245e-06,
      "loss": 0.7294,
      "step": 22450
    },
    {
      "epoch": 9.19,
      "grad_norm": 2.443511486053467,
      "learning_rate": 4.309506753990995e-06,
      "loss": 0.7058,
      "step": 22460
    },
    {
      "epoch": 9.2,
      "grad_norm": 3.0472683906555176,
      "learning_rate": 4.311425501432664e-06,
      "loss": 0.7292,
      "step": 22470
    },
    {
      "epoch": 9.2,
      "grad_norm": 2.3293099403381348,
      "learning_rate": 4.313344248874335e-06,
      "loss": 0.7048,
      "step": 22480
    },
    {
      "epoch": 9.21,
      "grad_norm": 2.0895471572875977,
      "learning_rate": 4.315262996316005e-06,
      "loss": 0.7055,
      "step": 22490
    },
    {
      "epoch": 9.21,
      "grad_norm": 2.9650275707244873,
      "learning_rate": 4.317181743757675e-06,
      "loss": 0.6963,
      "step": 22500
    },
    {
      "epoch": 9.21,
      "grad_norm": 2.7142257690429688,
      "learning_rate": 4.319100491199345e-06,
      "loss": 0.7058,
      "step": 22510
    },
    {
      "epoch": 9.22,
      "grad_norm": 2.5387771129608154,
      "learning_rate": 4.321019238641015e-06,
      "loss": 0.7202,
      "step": 22520
    },
    {
      "epoch": 9.22,
      "grad_norm": 2.390747308731079,
      "learning_rate": 4.322937986082685e-06,
      "loss": 0.7201,
      "step": 22530
    },
    {
      "epoch": 9.23,
      "grad_norm": 2.351577043533325,
      "learning_rate": 4.324856733524355e-06,
      "loss": 0.7139,
      "step": 22540
    },
    {
      "epoch": 9.23,
      "grad_norm": 3.174872398376465,
      "learning_rate": 4.326775480966025e-06,
      "loss": 0.7182,
      "step": 22550
    },
    {
      "epoch": 9.23,
      "grad_norm": 2.595942735671997,
      "learning_rate": 4.328694228407695e-06,
      "loss": 0.7198,
      "step": 22560
    },
    {
      "epoch": 9.24,
      "grad_norm": 2.373372793197632,
      "learning_rate": 4.3306129758493655e-06,
      "loss": 0.7113,
      "step": 22570
    },
    {
      "epoch": 9.24,
      "grad_norm": 1.9330629110336304,
      "learning_rate": 4.332531723291035e-06,
      "loss": 0.7189,
      "step": 22580
    },
    {
      "epoch": 9.25,
      "grad_norm": 2.394287109375,
      "learning_rate": 4.334450470732705e-06,
      "loss": 0.6965,
      "step": 22590
    },
    {
      "epoch": 9.25,
      "grad_norm": 2.4346113204956055,
      "learning_rate": 4.336369218174376e-06,
      "loss": 0.7117,
      "step": 22600
    },
    {
      "epoch": 9.26,
      "grad_norm": 2.4179160594940186,
      "learning_rate": 4.338287965616045e-06,
      "loss": 0.7355,
      "step": 22610
    },
    {
      "epoch": 9.26,
      "grad_norm": 1.9753488302230835,
      "learning_rate": 4.340206713057716e-06,
      "loss": 0.733,
      "step": 22620
    },
    {
      "epoch": 9.26,
      "grad_norm": 2.6475179195404053,
      "learning_rate": 4.342125460499386e-06,
      "loss": 0.7182,
      "step": 22630
    },
    {
      "epoch": 9.27,
      "grad_norm": 2.3281009197235107,
      "learning_rate": 4.3440442079410555e-06,
      "loss": 0.7145,
      "step": 22640
    },
    {
      "epoch": 9.27,
      "grad_norm": 3.107667922973633,
      "learning_rate": 4.345962955382726e-06,
      "loss": 0.7401,
      "step": 22650
    },
    {
      "epoch": 9.28,
      "grad_norm": 1.9574437141418457,
      "learning_rate": 4.347881702824396e-06,
      "loss": 0.6907,
      "step": 22660
    },
    {
      "epoch": 9.28,
      "grad_norm": 2.073007345199585,
      "learning_rate": 4.349800450266066e-06,
      "loss": 0.7302,
      "step": 22670
    },
    {
      "epoch": 9.28,
      "grad_norm": 3.4495351314544678,
      "learning_rate": 4.351719197707736e-06,
      "loss": 0.7045,
      "step": 22680
    },
    {
      "epoch": 9.29,
      "grad_norm": 2.078268051147461,
      "learning_rate": 4.3536379451494065e-06,
      "loss": 0.7188,
      "step": 22690
    },
    {
      "epoch": 9.29,
      "grad_norm": 2.1011979579925537,
      "learning_rate": 4.355556692591076e-06,
      "loss": 0.7031,
      "step": 22700
    },
    {
      "epoch": 9.3,
      "grad_norm": 2.1475424766540527,
      "learning_rate": 4.357475440032746e-06,
      "loss": 0.71,
      "step": 22710
    },
    {
      "epoch": 9.3,
      "grad_norm": 1.8050274848937988,
      "learning_rate": 4.359394187474417e-06,
      "loss": 0.7157,
      "step": 22720
    },
    {
      "epoch": 9.3,
      "grad_norm": 2.7244813442230225,
      "learning_rate": 4.361312934916086e-06,
      "loss": 0.7299,
      "step": 22730
    },
    {
      "epoch": 9.31,
      "grad_norm": 2.0111117362976074,
      "learning_rate": 4.363231682357757e-06,
      "loss": 0.7064,
      "step": 22740
    },
    {
      "epoch": 9.31,
      "grad_norm": 2.215186834335327,
      "learning_rate": 4.365150429799427e-06,
      "loss": 0.7435,
      "step": 22750
    },
    {
      "epoch": 9.32,
      "grad_norm": 2.9119040966033936,
      "learning_rate": 4.3670691772410965e-06,
      "loss": 0.7085,
      "step": 22760
    },
    {
      "epoch": 9.32,
      "grad_norm": 2.231187582015991,
      "learning_rate": 4.368987924682767e-06,
      "loss": 0.7039,
      "step": 22770
    },
    {
      "epoch": 9.32,
      "grad_norm": 2.1570348739624023,
      "learning_rate": 4.370906672124437e-06,
      "loss": 0.7262,
      "step": 22780
    },
    {
      "epoch": 9.33,
      "grad_norm": 2.7184457778930664,
      "learning_rate": 4.372825419566107e-06,
      "loss": 0.7176,
      "step": 22790
    },
    {
      "epoch": 9.33,
      "grad_norm": 2.7784264087677,
      "learning_rate": 4.374744167007777e-06,
      "loss": 0.7095,
      "step": 22800
    },
    {
      "epoch": 9.34,
      "grad_norm": 2.188140630722046,
      "learning_rate": 4.3766629144494474e-06,
      "loss": 0.7056,
      "step": 22810
    },
    {
      "epoch": 9.34,
      "grad_norm": 2.7368955612182617,
      "learning_rate": 4.378581661891117e-06,
      "loss": 0.7123,
      "step": 22820
    },
    {
      "epoch": 9.35,
      "grad_norm": 2.621732234954834,
      "learning_rate": 4.380500409332787e-06,
      "loss": 0.7122,
      "step": 22830
    },
    {
      "epoch": 9.35,
      "grad_norm": 1.8192980289459229,
      "learning_rate": 4.382419156774458e-06,
      "loss": 0.7286,
      "step": 22840
    },
    {
      "epoch": 9.35,
      "grad_norm": 2.697456121444702,
      "learning_rate": 4.384337904216127e-06,
      "loss": 0.7156,
      "step": 22850
    },
    {
      "epoch": 9.36,
      "grad_norm": 1.788084626197815,
      "learning_rate": 4.3862566516577976e-06,
      "loss": 0.7288,
      "step": 22860
    },
    {
      "epoch": 9.36,
      "grad_norm": 2.564004898071289,
      "learning_rate": 4.388175399099468e-06,
      "loss": 0.7149,
      "step": 22870
    },
    {
      "epoch": 9.37,
      "grad_norm": 3.814164400100708,
      "learning_rate": 4.3900941465411374e-06,
      "loss": 0.7135,
      "step": 22880
    },
    {
      "epoch": 9.37,
      "grad_norm": 3.9797768592834473,
      "learning_rate": 4.392012893982808e-06,
      "loss": 0.703,
      "step": 22890
    },
    {
      "epoch": 9.37,
      "grad_norm": 2.462801933288574,
      "learning_rate": 4.393931641424478e-06,
      "loss": 0.7124,
      "step": 22900
    },
    {
      "epoch": 9.38,
      "grad_norm": 3.1810266971588135,
      "learning_rate": 4.395850388866148e-06,
      "loss": 0.7148,
      "step": 22910
    },
    {
      "epoch": 9.38,
      "grad_norm": 2.3487749099731445,
      "learning_rate": 4.397769136307818e-06,
      "loss": 0.7222,
      "step": 22920
    },
    {
      "epoch": 9.39,
      "grad_norm": 1.7973308563232422,
      "learning_rate": 4.399687883749488e-06,
      "loss": 0.7174,
      "step": 22930
    },
    {
      "epoch": 9.39,
      "grad_norm": 2.663794994354248,
      "learning_rate": 4.401606631191158e-06,
      "loss": 0.7148,
      "step": 22940
    },
    {
      "epoch": 9.39,
      "grad_norm": 2.6543002128601074,
      "learning_rate": 4.403525378632828e-06,
      "loss": 0.6878,
      "step": 22950
    },
    {
      "epoch": 9.4,
      "grad_norm": 2.4021263122558594,
      "learning_rate": 4.405444126074499e-06,
      "loss": 0.7096,
      "step": 22960
    },
    {
      "epoch": 9.4,
      "grad_norm": 2.039127826690674,
      "learning_rate": 4.407362873516168e-06,
      "loss": 0.6991,
      "step": 22970
    },
    {
      "epoch": 9.41,
      "grad_norm": 2.689523220062256,
      "learning_rate": 4.4092816209578385e-06,
      "loss": 0.705,
      "step": 22980
    },
    {
      "epoch": 9.41,
      "grad_norm": 2.8206300735473633,
      "learning_rate": 4.411200368399508e-06,
      "loss": 0.697,
      "step": 22990
    },
    {
      "epoch": 9.41,
      "grad_norm": 2.3331527709960938,
      "learning_rate": 4.413119115841178e-06,
      "loss": 0.6908,
      "step": 23000
    },
    {
      "epoch": 9.42,
      "grad_norm": 2.148888349533081,
      "learning_rate": 4.415037863282849e-06,
      "loss": 0.7102,
      "step": 23010
    },
    {
      "epoch": 9.42,
      "grad_norm": 3.0522611141204834,
      "learning_rate": 4.416956610724518e-06,
      "loss": 0.713,
      "step": 23020
    },
    {
      "epoch": 9.43,
      "grad_norm": 1.819650650024414,
      "learning_rate": 4.418875358166189e-06,
      "loss": 0.7228,
      "step": 23030
    },
    {
      "epoch": 9.43,
      "grad_norm": 2.2914063930511475,
      "learning_rate": 4.420794105607859e-06,
      "loss": 0.7107,
      "step": 23040
    },
    {
      "epoch": 9.44,
      "grad_norm": 2.4086549282073975,
      "learning_rate": 4.4227128530495285e-06,
      "loss": 0.6968,
      "step": 23050
    },
    {
      "epoch": 9.44,
      "grad_norm": 3.2707316875457764,
      "learning_rate": 4.424631600491199e-06,
      "loss": 0.6995,
      "step": 23060
    },
    {
      "epoch": 9.44,
      "grad_norm": 2.1338770389556885,
      "learning_rate": 4.426550347932869e-06,
      "loss": 0.7135,
      "step": 23070
    },
    {
      "epoch": 9.45,
      "grad_norm": 2.6822640895843506,
      "learning_rate": 4.428469095374539e-06,
      "loss": 0.7153,
      "step": 23080
    },
    {
      "epoch": 9.45,
      "grad_norm": 2.8068795204162598,
      "learning_rate": 4.430387842816209e-06,
      "loss": 0.728,
      "step": 23090
    },
    {
      "epoch": 9.46,
      "grad_norm": 2.2797255516052246,
      "learning_rate": 4.4323065902578795e-06,
      "loss": 0.6955,
      "step": 23100
    },
    {
      "epoch": 9.46,
      "grad_norm": 2.00830078125,
      "learning_rate": 4.434225337699549e-06,
      "loss": 0.7125,
      "step": 23110
    },
    {
      "epoch": 9.46,
      "grad_norm": 2.1277565956115723,
      "learning_rate": 4.436144085141219e-06,
      "loss": 0.6924,
      "step": 23120
    },
    {
      "epoch": 9.47,
      "grad_norm": 2.7016983032226562,
      "learning_rate": 4.43806283258289e-06,
      "loss": 0.7213,
      "step": 23130
    },
    {
      "epoch": 9.47,
      "grad_norm": 2.756830930709839,
      "learning_rate": 4.439981580024559e-06,
      "loss": 0.7301,
      "step": 23140
    },
    {
      "epoch": 9.48,
      "grad_norm": 2.8072965145111084,
      "learning_rate": 4.44190032746623e-06,
      "loss": 0.7196,
      "step": 23150
    },
    {
      "epoch": 9.48,
      "grad_norm": 2.2121713161468506,
      "learning_rate": 4.4438190749079e-06,
      "loss": 0.7251,
      "step": 23160
    },
    {
      "epoch": 9.48,
      "grad_norm": 2.180386781692505,
      "learning_rate": 4.4457378223495695e-06,
      "loss": 0.6917,
      "step": 23170
    },
    {
      "epoch": 9.49,
      "grad_norm": 1.8096939325332642,
      "learning_rate": 4.44765656979124e-06,
      "loss": 0.7207,
      "step": 23180
    },
    {
      "epoch": 9.49,
      "grad_norm": 2.5491385459899902,
      "learning_rate": 4.44957531723291e-06,
      "loss": 0.7241,
      "step": 23190
    },
    {
      "epoch": 9.5,
      "grad_norm": 3.302410125732422,
      "learning_rate": 4.45149406467458e-06,
      "loss": 0.7066,
      "step": 23200
    },
    {
      "epoch": 9.5,
      "grad_norm": 2.9777417182922363,
      "learning_rate": 4.45341281211625e-06,
      "loss": 0.7304,
      "step": 23210
    },
    {
      "epoch": 9.5,
      "grad_norm": 1.8830287456512451,
      "learning_rate": 4.4553315595579204e-06,
      "loss": 0.7143,
      "step": 23220
    },
    {
      "epoch": 9.51,
      "grad_norm": 1.933571696281433,
      "learning_rate": 4.45725030699959e-06,
      "loss": 0.7098,
      "step": 23230
    },
    {
      "epoch": 9.51,
      "grad_norm": 2.849372148513794,
      "learning_rate": 4.45916905444126e-06,
      "loss": 0.7218,
      "step": 23240
    },
    {
      "epoch": 9.52,
      "grad_norm": 3.5900747776031494,
      "learning_rate": 4.461087801882931e-06,
      "loss": 0.7162,
      "step": 23250
    },
    {
      "epoch": 9.52,
      "grad_norm": 1.935198187828064,
      "learning_rate": 4.4630065493246e-06,
      "loss": 0.7076,
      "step": 23260
    },
    {
      "epoch": 9.53,
      "grad_norm": 2.8796277046203613,
      "learning_rate": 4.4649252967662706e-06,
      "loss": 0.7208,
      "step": 23270
    },
    {
      "epoch": 9.53,
      "grad_norm": 1.954736351966858,
      "learning_rate": 4.466844044207941e-06,
      "loss": 0.7194,
      "step": 23280
    },
    {
      "epoch": 9.53,
      "grad_norm": 2.0738229751586914,
      "learning_rate": 4.4687627916496104e-06,
      "loss": 0.6946,
      "step": 23290
    },
    {
      "epoch": 9.54,
      "grad_norm": 2.298250913619995,
      "learning_rate": 4.470681539091281e-06,
      "loss": 0.7193,
      "step": 23300
    },
    {
      "epoch": 9.54,
      "grad_norm": 3.1078011989593506,
      "learning_rate": 4.472600286532951e-06,
      "loss": 0.7178,
      "step": 23310
    },
    {
      "epoch": 9.55,
      "grad_norm": 2.217550754547119,
      "learning_rate": 4.474519033974621e-06,
      "loss": 0.7094,
      "step": 23320
    },
    {
      "epoch": 9.55,
      "grad_norm": 2.354552745819092,
      "learning_rate": 4.476437781416291e-06,
      "loss": 0.7217,
      "step": 23330
    },
    {
      "epoch": 9.55,
      "grad_norm": 2.821611166000366,
      "learning_rate": 4.478356528857961e-06,
      "loss": 0.7238,
      "step": 23340
    },
    {
      "epoch": 9.56,
      "grad_norm": 2.1091275215148926,
      "learning_rate": 4.480275276299631e-06,
      "loss": 0.6937,
      "step": 23350
    },
    {
      "epoch": 9.56,
      "grad_norm": 2.1754839420318604,
      "learning_rate": 4.482194023741301e-06,
      "loss": 0.7051,
      "step": 23360
    },
    {
      "epoch": 9.57,
      "grad_norm": 1.6845911741256714,
      "learning_rate": 4.484112771182972e-06,
      "loss": 0.7052,
      "step": 23370
    },
    {
      "epoch": 9.57,
      "grad_norm": 3.22454833984375,
      "learning_rate": 4.486031518624641e-06,
      "loss": 0.7105,
      "step": 23380
    },
    {
      "epoch": 9.57,
      "grad_norm": 3.2758629322052,
      "learning_rate": 4.4879502660663115e-06,
      "loss": 0.7022,
      "step": 23390
    },
    {
      "epoch": 9.58,
      "grad_norm": 1.9400813579559326,
      "learning_rate": 4.489869013507982e-06,
      "loss": 0.7171,
      "step": 23400
    },
    {
      "epoch": 9.58,
      "grad_norm": 1.8663121461868286,
      "learning_rate": 4.491787760949651e-06,
      "loss": 0.6981,
      "step": 23410
    },
    {
      "epoch": 9.59,
      "grad_norm": 2.333252429962158,
      "learning_rate": 4.493706508391322e-06,
      "loss": 0.6944,
      "step": 23420
    },
    {
      "epoch": 9.59,
      "grad_norm": 2.6542837619781494,
      "learning_rate": 4.495625255832991e-06,
      "loss": 0.7001,
      "step": 23430
    },
    {
      "epoch": 9.59,
      "grad_norm": 2.650667905807495,
      "learning_rate": 4.497544003274662e-06,
      "loss": 0.7091,
      "step": 23440
    },
    {
      "epoch": 9.6,
      "grad_norm": 2.4116506576538086,
      "learning_rate": 4.499462750716332e-06,
      "loss": 0.6969,
      "step": 23450
    },
    {
      "epoch": 9.6,
      "grad_norm": 2.158320188522339,
      "learning_rate": 4.5013814981580015e-06,
      "loss": 0.7245,
      "step": 23460
    },
    {
      "epoch": 9.61,
      "grad_norm": 3.1943752765655518,
      "learning_rate": 4.503300245599672e-06,
      "loss": 0.7123,
      "step": 23470
    },
    {
      "epoch": 9.61,
      "grad_norm": 2.2453572750091553,
      "learning_rate": 4.505218993041342e-06,
      "loss": 0.7004,
      "step": 23480
    },
    {
      "epoch": 9.62,
      "grad_norm": 1.987513542175293,
      "learning_rate": 4.507137740483012e-06,
      "loss": 0.6975,
      "step": 23490
    },
    {
      "epoch": 9.62,
      "grad_norm": 2.0267202854156494,
      "learning_rate": 4.509056487924682e-06,
      "loss": 0.7117,
      "step": 23500
    },
    {
      "epoch": 9.62,
      "grad_norm": 2.658677816390991,
      "learning_rate": 4.5109752353663525e-06,
      "loss": 0.6913,
      "step": 23510
    },
    {
      "epoch": 9.63,
      "grad_norm": 2.689695358276367,
      "learning_rate": 4.512893982808022e-06,
      "loss": 0.7017,
      "step": 23520
    },
    {
      "epoch": 9.63,
      "grad_norm": 2.0881760120391846,
      "learning_rate": 4.514812730249692e-06,
      "loss": 0.7099,
      "step": 23530
    },
    {
      "epoch": 9.64,
      "grad_norm": 3.263606309890747,
      "learning_rate": 4.516731477691363e-06,
      "loss": 0.7255,
      "step": 23540
    },
    {
      "epoch": 9.64,
      "grad_norm": 2.2726597785949707,
      "learning_rate": 4.518650225133032e-06,
      "loss": 0.7052,
      "step": 23550
    },
    {
      "epoch": 9.64,
      "grad_norm": 3.3644869327545166,
      "learning_rate": 4.520568972574703e-06,
      "loss": 0.7179,
      "step": 23560
    },
    {
      "epoch": 9.65,
      "grad_norm": 2.3356449604034424,
      "learning_rate": 4.522487720016373e-06,
      "loss": 0.716,
      "step": 23570
    },
    {
      "epoch": 9.65,
      "grad_norm": 2.2184672355651855,
      "learning_rate": 4.5244064674580425e-06,
      "loss": 0.6966,
      "step": 23580
    },
    {
      "epoch": 9.66,
      "grad_norm": 3.884279489517212,
      "learning_rate": 4.526325214899713e-06,
      "loss": 0.7131,
      "step": 23590
    },
    {
      "epoch": 9.66,
      "grad_norm": 2.733175039291382,
      "learning_rate": 4.528243962341383e-06,
      "loss": 0.7143,
      "step": 23600
    },
    {
      "epoch": 9.66,
      "grad_norm": 2.4611618518829346,
      "learning_rate": 4.530162709783053e-06,
      "loss": 0.6799,
      "step": 23610
    },
    {
      "epoch": 9.67,
      "grad_norm": 2.8863019943237305,
      "learning_rate": 4.532081457224723e-06,
      "loss": 0.7112,
      "step": 23620
    },
    {
      "epoch": 9.67,
      "grad_norm": 2.29516863822937,
      "learning_rate": 4.5340002046663935e-06,
      "loss": 0.7037,
      "step": 23630
    },
    {
      "epoch": 9.68,
      "grad_norm": 2.1514596939086914,
      "learning_rate": 4.535918952108063e-06,
      "loss": 0.7064,
      "step": 23640
    },
    {
      "epoch": 9.68,
      "grad_norm": 2.272219181060791,
      "learning_rate": 4.537837699549733e-06,
      "loss": 0.7216,
      "step": 23650
    },
    {
      "epoch": 9.68,
      "grad_norm": 2.74214768409729,
      "learning_rate": 4.539756446991404e-06,
      "loss": 0.7002,
      "step": 23660
    },
    {
      "epoch": 9.69,
      "grad_norm": 1.9549198150634766,
      "learning_rate": 4.541675194433073e-06,
      "loss": 0.7126,
      "step": 23670
    },
    {
      "epoch": 9.69,
      "grad_norm": 2.3909482955932617,
      "learning_rate": 4.5435939418747436e-06,
      "loss": 0.7138,
      "step": 23680
    },
    {
      "epoch": 9.7,
      "grad_norm": 2.2318239212036133,
      "learning_rate": 4.545512689316414e-06,
      "loss": 0.7036,
      "step": 23690
    },
    {
      "epoch": 9.7,
      "grad_norm": 3.2179932594299316,
      "learning_rate": 4.5474314367580834e-06,
      "loss": 0.7038,
      "step": 23700
    },
    {
      "epoch": 9.71,
      "grad_norm": 2.445946455001831,
      "learning_rate": 4.549350184199754e-06,
      "loss": 0.7015,
      "step": 23710
    },
    {
      "epoch": 9.71,
      "grad_norm": 2.4000437259674072,
      "learning_rate": 4.551268931641424e-06,
      "loss": 0.682,
      "step": 23720
    },
    {
      "epoch": 9.71,
      "grad_norm": 2.3583409786224365,
      "learning_rate": 4.553187679083094e-06,
      "loss": 0.7047,
      "step": 23730
    },
    {
      "epoch": 9.72,
      "grad_norm": 3.046295166015625,
      "learning_rate": 4.555106426524764e-06,
      "loss": 0.7167,
      "step": 23740
    },
    {
      "epoch": 9.72,
      "grad_norm": 2.0321531295776367,
      "learning_rate": 4.557025173966434e-06,
      "loss": 0.692,
      "step": 23750
    },
    {
      "epoch": 9.73,
      "grad_norm": 2.4237303733825684,
      "learning_rate": 4.558943921408105e-06,
      "loss": 0.7053,
      "step": 23760
    },
    {
      "epoch": 9.73,
      "grad_norm": 2.1437554359436035,
      "learning_rate": 4.560862668849775e-06,
      "loss": 0.7113,
      "step": 23770
    },
    {
      "epoch": 9.73,
      "grad_norm": 3.043562173843384,
      "learning_rate": 4.562781416291445e-06,
      "loss": 0.6758,
      "step": 23780
    },
    {
      "epoch": 9.74,
      "grad_norm": 2.35640549659729,
      "learning_rate": 4.564700163733115e-06,
      "loss": 0.705,
      "step": 23790
    },
    {
      "epoch": 9.74,
      "grad_norm": 2.6797432899475098,
      "learning_rate": 4.5666189111747845e-06,
      "loss": 0.7204,
      "step": 23800
    },
    {
      "epoch": 9.75,
      "grad_norm": 2.3586103916168213,
      "learning_rate": 4.568537658616455e-06,
      "loss": 0.6935,
      "step": 23810
    },
    {
      "epoch": 9.75,
      "grad_norm": 2.8074581623077393,
      "learning_rate": 4.570456406058125e-06,
      "loss": 0.7042,
      "step": 23820
    },
    {
      "epoch": 9.75,
      "grad_norm": 2.609004259109497,
      "learning_rate": 4.572375153499795e-06,
      "loss": 0.7045,
      "step": 23830
    },
    {
      "epoch": 9.76,
      "grad_norm": 2.1478710174560547,
      "learning_rate": 4.574293900941465e-06,
      "loss": 0.7076,
      "step": 23840
    },
    {
      "epoch": 9.76,
      "grad_norm": 2.599799633026123,
      "learning_rate": 4.5762126483831355e-06,
      "loss": 0.6849,
      "step": 23850
    },
    {
      "epoch": 9.77,
      "grad_norm": 2.4808359146118164,
      "learning_rate": 4.578131395824805e-06,
      "loss": 0.7084,
      "step": 23860
    },
    {
      "epoch": 9.77,
      "grad_norm": 3.250492811203003,
      "learning_rate": 4.580050143266475e-06,
      "loss": 0.7138,
      "step": 23870
    },
    {
      "epoch": 9.77,
      "grad_norm": 2.251833200454712,
      "learning_rate": 4.581968890708146e-06,
      "loss": 0.7067,
      "step": 23880
    },
    {
      "epoch": 9.78,
      "grad_norm": 3.004631996154785,
      "learning_rate": 4.583887638149815e-06,
      "loss": 0.6904,
      "step": 23890
    },
    {
      "epoch": 9.78,
      "grad_norm": 2.4478819370269775,
      "learning_rate": 4.585806385591486e-06,
      "loss": 0.709,
      "step": 23900
    },
    {
      "epoch": 9.79,
      "grad_norm": 2.5518958568573,
      "learning_rate": 4.587725133033156e-06,
      "loss": 0.6933,
      "step": 23910
    },
    {
      "epoch": 9.79,
      "grad_norm": 2.3364224433898926,
      "learning_rate": 4.5896438804748255e-06,
      "loss": 0.6936,
      "step": 23920
    },
    {
      "epoch": 9.8,
      "grad_norm": 3.009132146835327,
      "learning_rate": 4.591562627916496e-06,
      "loss": 0.7134,
      "step": 23930
    },
    {
      "epoch": 9.8,
      "grad_norm": 2.641307830810547,
      "learning_rate": 4.593481375358166e-06,
      "loss": 0.7035,
      "step": 23940
    },
    {
      "epoch": 9.8,
      "grad_norm": 2.547075033187866,
      "learning_rate": 4.595400122799836e-06,
      "loss": 0.7008,
      "step": 23950
    },
    {
      "epoch": 9.81,
      "grad_norm": 2.2828924655914307,
      "learning_rate": 4.597318870241506e-06,
      "loss": 0.7219,
      "step": 23960
    },
    {
      "epoch": 9.81,
      "grad_norm": 4.278975009918213,
      "learning_rate": 4.5992376176831765e-06,
      "loss": 0.7189,
      "step": 23970
    },
    {
      "epoch": 9.82,
      "grad_norm": 2.408287763595581,
      "learning_rate": 4.601156365124846e-06,
      "loss": 0.7085,
      "step": 23980
    },
    {
      "epoch": 9.82,
      "grad_norm": 2.7354071140289307,
      "learning_rate": 4.603075112566516e-06,
      "loss": 0.704,
      "step": 23990
    },
    {
      "epoch": 9.82,
      "grad_norm": 2.4840292930603027,
      "learning_rate": 4.604993860008187e-06,
      "loss": 0.7159,
      "step": 24000
    },
    {
      "epoch": 9.83,
      "grad_norm": 2.397722005844116,
      "learning_rate": 4.606912607449856e-06,
      "loss": 0.7007,
      "step": 24010
    },
    {
      "epoch": 9.83,
      "grad_norm": 2.4426732063293457,
      "learning_rate": 4.608831354891527e-06,
      "loss": 0.7,
      "step": 24020
    },
    {
      "epoch": 9.84,
      "grad_norm": 1.7754263877868652,
      "learning_rate": 4.610750102333197e-06,
      "loss": 0.7067,
      "step": 24030
    },
    {
      "epoch": 9.84,
      "grad_norm": 2.488572120666504,
      "learning_rate": 4.6126688497748665e-06,
      "loss": 0.681,
      "step": 24040
    },
    {
      "epoch": 9.84,
      "grad_norm": 2.1728098392486572,
      "learning_rate": 4.614587597216537e-06,
      "loss": 0.7122,
      "step": 24050
    },
    {
      "epoch": 9.85,
      "grad_norm": 2.2914984226226807,
      "learning_rate": 4.616506344658207e-06,
      "loss": 0.7113,
      "step": 24060
    },
    {
      "epoch": 9.85,
      "grad_norm": 2.509848117828369,
      "learning_rate": 4.618425092099877e-06,
      "loss": 0.7126,
      "step": 24070
    },
    {
      "epoch": 9.86,
      "grad_norm": 3.785348415374756,
      "learning_rate": 4.620343839541547e-06,
      "loss": 0.7134,
      "step": 24080
    },
    {
      "epoch": 9.86,
      "grad_norm": 2.3799755573272705,
      "learning_rate": 4.6222625869832174e-06,
      "loss": 0.7178,
      "step": 24090
    },
    {
      "epoch": 9.86,
      "grad_norm": 2.1607892513275146,
      "learning_rate": 4.624181334424887e-06,
      "loss": 0.7008,
      "step": 24100
    },
    {
      "epoch": 9.87,
      "grad_norm": 3.163114547729492,
      "learning_rate": 4.626100081866557e-06,
      "loss": 0.7006,
      "step": 24110
    },
    {
      "epoch": 9.87,
      "grad_norm": 2.456725835800171,
      "learning_rate": 4.628018829308228e-06,
      "loss": 0.7046,
      "step": 24120
    },
    {
      "epoch": 9.88,
      "grad_norm": 2.0399396419525146,
      "learning_rate": 4.629937576749897e-06,
      "loss": 0.7132,
      "step": 24130
    },
    {
      "epoch": 9.88,
      "grad_norm": 2.465848445892334,
      "learning_rate": 4.6318563241915675e-06,
      "loss": 0.6976,
      "step": 24140
    },
    {
      "epoch": 9.89,
      "grad_norm": 2.5167524814605713,
      "learning_rate": 4.633775071633238e-06,
      "loss": 0.7155,
      "step": 24150
    },
    {
      "epoch": 9.89,
      "grad_norm": 2.2722558975219727,
      "learning_rate": 4.635693819074907e-06,
      "loss": 0.7131,
      "step": 24160
    },
    {
      "epoch": 9.89,
      "grad_norm": 2.1503348350524902,
      "learning_rate": 4.637612566516578e-06,
      "loss": 0.7166,
      "step": 24170
    },
    {
      "epoch": 9.9,
      "grad_norm": 1.7920317649841309,
      "learning_rate": 4.639531313958248e-06,
      "loss": 0.7187,
      "step": 24180
    },
    {
      "epoch": 9.9,
      "grad_norm": 2.429795742034912,
      "learning_rate": 4.641450061399918e-06,
      "loss": 0.6896,
      "step": 24190
    },
    {
      "epoch": 9.91,
      "grad_norm": 2.0433402061462402,
      "learning_rate": 4.643368808841588e-06,
      "loss": 0.7138,
      "step": 24200
    },
    {
      "epoch": 9.91,
      "grad_norm": 1.7700269222259521,
      "learning_rate": 4.645287556283258e-06,
      "loss": 0.7102,
      "step": 24210
    },
    {
      "epoch": 9.91,
      "grad_norm": 2.7100043296813965,
      "learning_rate": 4.647206303724928e-06,
      "loss": 0.6998,
      "step": 24220
    },
    {
      "epoch": 9.92,
      "grad_norm": 2.647489309310913,
      "learning_rate": 4.649125051166598e-06,
      "loss": 0.6895,
      "step": 24230
    },
    {
      "epoch": 9.92,
      "grad_norm": 2.5273914337158203,
      "learning_rate": 4.651043798608268e-06,
      "loss": 0.7125,
      "step": 24240
    },
    {
      "epoch": 9.93,
      "grad_norm": 2.8394665718078613,
      "learning_rate": 4.652962546049938e-06,
      "loss": 0.7442,
      "step": 24250
    },
    {
      "epoch": 9.93,
      "grad_norm": 2.2772345542907715,
      "learning_rate": 4.6548812934916085e-06,
      "loss": 0.7012,
      "step": 24260
    },
    {
      "epoch": 9.93,
      "grad_norm": 2.1043944358825684,
      "learning_rate": 4.656800040933278e-06,
      "loss": 0.6895,
      "step": 24270
    },
    {
      "epoch": 9.94,
      "grad_norm": 1.6131404638290405,
      "learning_rate": 4.658718788374948e-06,
      "loss": 0.7223,
      "step": 24280
    },
    {
      "epoch": 9.94,
      "grad_norm": 2.8914451599121094,
      "learning_rate": 4.660637535816619e-06,
      "loss": 0.6987,
      "step": 24290
    },
    {
      "epoch": 9.95,
      "grad_norm": 2.5789406299591064,
      "learning_rate": 4.662556283258288e-06,
      "loss": 0.7039,
      "step": 24300
    },
    {
      "epoch": 9.95,
      "grad_norm": 2.161411762237549,
      "learning_rate": 4.664475030699959e-06,
      "loss": 0.6982,
      "step": 24310
    },
    {
      "epoch": 9.95,
      "grad_norm": 2.430344820022583,
      "learning_rate": 4.666393778141629e-06,
      "loss": 0.7147,
      "step": 24320
    },
    {
      "epoch": 9.96,
      "grad_norm": 2.707850217819214,
      "learning_rate": 4.6683125255832985e-06,
      "loss": 0.7097,
      "step": 24330
    },
    {
      "epoch": 9.96,
      "grad_norm": 4.319214820861816,
      "learning_rate": 4.670231273024969e-06,
      "loss": 0.7228,
      "step": 24340
    },
    {
      "epoch": 9.97,
      "grad_norm": 4.298519611358643,
      "learning_rate": 4.672150020466639e-06,
      "loss": 0.6962,
      "step": 24350
    },
    {
      "epoch": 9.97,
      "grad_norm": 2.4596810340881348,
      "learning_rate": 4.674068767908309e-06,
      "loss": 0.687,
      "step": 24360
    },
    {
      "epoch": 9.98,
      "grad_norm": 3.026045799255371,
      "learning_rate": 4.675987515349979e-06,
      "loss": 0.7037,
      "step": 24370
    },
    {
      "epoch": 9.98,
      "grad_norm": 2.8847506046295166,
      "learning_rate": 4.6779062627916495e-06,
      "loss": 0.7037,
      "step": 24380
    },
    {
      "epoch": 9.98,
      "grad_norm": 2.6147871017456055,
      "learning_rate": 4.679825010233319e-06,
      "loss": 0.7073,
      "step": 24390
    },
    {
      "epoch": 9.99,
      "grad_norm": 2.5384294986724854,
      "learning_rate": 4.681743757674989e-06,
      "loss": 0.702,
      "step": 24400
    },
    {
      "epoch": 9.99,
      "grad_norm": 2.532017230987549,
      "learning_rate": 4.68366250511666e-06,
      "loss": 0.6989,
      "step": 24410
    },
    {
      "epoch": 10.0,
      "grad_norm": 2.0054571628570557,
      "learning_rate": 4.685581252558329e-06,
      "loss": 0.7326,
      "step": 24420
    },
    {
      "epoch": 10.0,
      "grad_norm": 2.593109130859375,
      "learning_rate": 4.6875e-06,
      "loss": 0.6875,
      "step": 24430
    },
    {
      "epoch": 10.0,
      "eval_loss": 0.7001115679740906,
      "eval_runtime": 52.1281,
      "eval_samples_per_second": 66.164,
      "eval_steps_per_second": 8.287,
      "step": 24430
    },
    {
      "epoch": 10.0,
      "grad_norm": 1.9081238508224487,
      "learning_rate": 4.6874999946318216e-06,
      "loss": 0.7194,
      "step": 24440
    },
    {
      "epoch": 10.01,
      "grad_norm": 1.9246702194213867,
      "learning_rate": 4.687499978527288e-06,
      "loss": 0.7054,
      "step": 24450
    },
    {
      "epoch": 10.01,
      "grad_norm": 2.2403645515441895,
      "learning_rate": 4.687499951686398e-06,
      "loss": 0.7214,
      "step": 24460
    },
    {
      "epoch": 10.02,
      "grad_norm": 2.611132860183716,
      "learning_rate": 4.6874999141091545e-06,
      "loss": 0.6924,
      "step": 24470
    },
    {
      "epoch": 10.02,
      "grad_norm": 2.3304314613342285,
      "learning_rate": 4.687499865795554e-06,
      "loss": 0.6797,
      "step": 24480
    },
    {
      "epoch": 10.02,
      "grad_norm": 2.0320703983306885,
      "learning_rate": 4.687499806745598e-06,
      "loss": 0.6977,
      "step": 24490
    },
    {
      "epoch": 10.03,
      "grad_norm": 2.8083138465881348,
      "learning_rate": 4.687499736959288e-06,
      "loss": 0.697,
      "step": 24500
    },
    {
      "epoch": 10.03,
      "grad_norm": 2.580742359161377,
      "learning_rate": 4.687499656436623e-06,
      "loss": 0.7251,
      "step": 24510
    },
    {
      "epoch": 10.04,
      "grad_norm": 2.0508370399475098,
      "learning_rate": 4.687499565177604e-06,
      "loss": 0.7061,
      "step": 24520
    },
    {
      "epoch": 10.04,
      "grad_norm": 3.1649487018585205,
      "learning_rate": 4.687499463182231e-06,
      "loss": 0.6826,
      "step": 24530
    },
    {
      "epoch": 10.05,
      "grad_norm": 2.2241134643554688,
      "learning_rate": 4.6874993504505055e-06,
      "loss": 0.7241,
      "step": 24540
    },
    {
      "epoch": 10.05,
      "grad_norm": 2.777484893798828,
      "learning_rate": 4.687499226982426e-06,
      "loss": 0.6815,
      "step": 24550
    },
    {
      "epoch": 10.05,
      "grad_norm": 2.170311689376831,
      "learning_rate": 4.687499092777995e-06,
      "loss": 0.7092,
      "step": 24560
    },
    {
      "epoch": 10.06,
      "grad_norm": 1.8103601932525635,
      "learning_rate": 4.687498947837212e-06,
      "loss": 0.7172,
      "step": 24570
    },
    {
      "epoch": 10.06,
      "grad_norm": 2.289992570877075,
      "learning_rate": 4.6874987921600785e-06,
      "loss": 0.6988,
      "step": 24580
    },
    {
      "epoch": 10.07,
      "grad_norm": 4.2217936515808105,
      "learning_rate": 4.687498625746594e-06,
      "loss": 0.7014,
      "step": 24590
    },
    {
      "epoch": 10.07,
      "grad_norm": 2.7406299114227295,
      "learning_rate": 4.68749844859676e-06,
      "loss": 0.6997,
      "step": 24600
    },
    {
      "epoch": 10.07,
      "grad_norm": 2.355210304260254,
      "learning_rate": 4.687498260710579e-06,
      "loss": 0.689,
      "step": 24610
    },
    {
      "epoch": 10.08,
      "grad_norm": 1.854041576385498,
      "learning_rate": 4.687498062088049e-06,
      "loss": 0.7024,
      "step": 24620
    },
    {
      "epoch": 10.08,
      "grad_norm": 2.505941867828369,
      "learning_rate": 4.6874978527291715e-06,
      "loss": 0.6833,
      "step": 24630
    },
    {
      "epoch": 10.09,
      "grad_norm": 2.877213478088379,
      "learning_rate": 4.687497632633949e-06,
      "loss": 0.7122,
      "step": 24640
    },
    {
      "epoch": 10.09,
      "grad_norm": 2.705911874771118,
      "learning_rate": 4.687497401802381e-06,
      "loss": 0.7082,
      "step": 24650
    },
    {
      "epoch": 10.09,
      "grad_norm": 2.109062433242798,
      "learning_rate": 4.687497160234469e-06,
      "loss": 0.7099,
      "step": 24660
    },
    {
      "epoch": 10.1,
      "grad_norm": 2.1501412391662598,
      "learning_rate": 4.687496907930215e-06,
      "loss": 0.7079,
      "step": 24670
    },
    {
      "epoch": 10.1,
      "grad_norm": 2.3345303535461426,
      "learning_rate": 4.68749664488962e-06,
      "loss": 0.6954,
      "step": 24680
    },
    {
      "epoch": 10.11,
      "grad_norm": 4.034430503845215,
      "learning_rate": 4.6874963711126826e-06,
      "loss": 0.7122,
      "step": 24690
    },
    {
      "epoch": 10.11,
      "grad_norm": 2.1008243560791016,
      "learning_rate": 4.687496086599407e-06,
      "loss": 0.7126,
      "step": 24700
    },
    {
      "epoch": 10.11,
      "grad_norm": 2.8016743659973145,
      "learning_rate": 4.687495791349794e-06,
      "loss": 0.7141,
      "step": 24710
    },
    {
      "epoch": 10.12,
      "grad_norm": 3.009551763534546,
      "learning_rate": 4.687495485363844e-06,
      "loss": 0.6863,
      "step": 24720
    },
    {
      "epoch": 10.12,
      "grad_norm": 2.7265067100524902,
      "learning_rate": 4.6874951686415595e-06,
      "loss": 0.7091,
      "step": 24730
    },
    {
      "epoch": 10.13,
      "grad_norm": 1.8920705318450928,
      "learning_rate": 4.68749484118294e-06,
      "loss": 0.6918,
      "step": 24740
    },
    {
      "epoch": 10.13,
      "grad_norm": 2.3659470081329346,
      "learning_rate": 4.68749450298799e-06,
      "loss": 0.6999,
      "step": 24750
    },
    {
      "epoch": 10.14,
      "grad_norm": 2.4942498207092285,
      "learning_rate": 4.687494154056709e-06,
      "loss": 0.6914,
      "step": 24760
    },
    {
      "epoch": 10.14,
      "grad_norm": 2.014601707458496,
      "learning_rate": 4.687493794389098e-06,
      "loss": 0.7092,
      "step": 24770
    },
    {
      "epoch": 10.14,
      "grad_norm": 2.759960412979126,
      "learning_rate": 4.68749342398516e-06,
      "loss": 0.7074,
      "step": 24780
    },
    {
      "epoch": 10.15,
      "grad_norm": 2.1190052032470703,
      "learning_rate": 4.687493042844896e-06,
      "loss": 0.6826,
      "step": 24790
    },
    {
      "epoch": 10.15,
      "grad_norm": 1.8002903461456299,
      "learning_rate": 4.687492650968309e-06,
      "loss": 0.6909,
      "step": 24800
    },
    {
      "epoch": 10.16,
      "grad_norm": 2.651128053665161,
      "learning_rate": 4.6874922483554e-06,
      "loss": 0.7131,
      "step": 24810
    },
    {
      "epoch": 10.16,
      "grad_norm": 2.9366824626922607,
      "learning_rate": 4.68749183500617e-06,
      "loss": 0.6988,
      "step": 24820
    },
    {
      "epoch": 10.16,
      "grad_norm": 3.2236030101776123,
      "learning_rate": 4.687491410920623e-06,
      "loss": 0.7188,
      "step": 24830
    },
    {
      "epoch": 10.17,
      "grad_norm": 3.826709747314453,
      "learning_rate": 4.687490976098758e-06,
      "loss": 0.6798,
      "step": 24840
    },
    {
      "epoch": 10.17,
      "grad_norm": 1.746989130973816,
      "learning_rate": 4.6874905305405796e-06,
      "loss": 0.6969,
      "step": 24850
    },
    {
      "epoch": 10.18,
      "grad_norm": 2.6046674251556396,
      "learning_rate": 4.687490074246088e-06,
      "loss": 0.7109,
      "step": 24860
    },
    {
      "epoch": 10.18,
      "grad_norm": 2.537201166152954,
      "learning_rate": 4.687489607215287e-06,
      "loss": 0.686,
      "step": 24870
    },
    {
      "epoch": 10.18,
      "grad_norm": 3.3388583660125732,
      "learning_rate": 4.6874891294481764e-06,
      "loss": 0.6952,
      "step": 24880
    },
    {
      "epoch": 10.19,
      "grad_norm": 2.4384498596191406,
      "learning_rate": 4.687488640944761e-06,
      "loss": 0.6948,
      "step": 24890
    },
    {
      "epoch": 10.19,
      "grad_norm": 2.289560079574585,
      "learning_rate": 4.687488141705042e-06,
      "loss": 0.6894,
      "step": 24900
    },
    {
      "epoch": 10.2,
      "grad_norm": 1.9784636497497559,
      "learning_rate": 4.68748763172902e-06,
      "loss": 0.7155,
      "step": 24910
    },
    {
      "epoch": 10.2,
      "grad_norm": 1.7699781656265259,
      "learning_rate": 4.687487111016701e-06,
      "loss": 0.7043,
      "step": 24920
    },
    {
      "epoch": 10.2,
      "grad_norm": 2.044834613800049,
      "learning_rate": 4.687486579568084e-06,
      "loss": 0.7169,
      "step": 24930
    },
    {
      "epoch": 10.21,
      "grad_norm": 2.2041678428649902,
      "learning_rate": 4.687486037383173e-06,
      "loss": 0.7077,
      "step": 24940
    },
    {
      "epoch": 10.21,
      "grad_norm": 1.9904277324676514,
      "learning_rate": 4.68748548446197e-06,
      "loss": 0.709,
      "step": 24950
    },
    {
      "epoch": 10.22,
      "grad_norm": 2.0793983936309814,
      "learning_rate": 4.687484920804478e-06,
      "loss": 0.7087,
      "step": 24960
    },
    {
      "epoch": 10.22,
      "grad_norm": 3.6744906902313232,
      "learning_rate": 4.6874843464106985e-06,
      "loss": 0.6742,
      "step": 24970
    },
    {
      "epoch": 10.23,
      "grad_norm": 2.2659261226654053,
      "learning_rate": 4.687483761280636e-06,
      "loss": 0.6829,
      "step": 24980
    },
    {
      "epoch": 10.23,
      "grad_norm": 1.9555814266204834,
      "learning_rate": 4.687483165414292e-06,
      "loss": 0.6906,
      "step": 24990
    },
    {
      "epoch": 10.23,
      "grad_norm": 2.8483846187591553,
      "learning_rate": 4.6874825588116685e-06,
      "loss": 0.6916,
      "step": 25000
    },
    {
      "epoch": 10.24,
      "grad_norm": 2.169543981552124,
      "learning_rate": 4.68748194147277e-06,
      "loss": 0.7002,
      "step": 25010
    },
    {
      "epoch": 10.24,
      "grad_norm": 2.1171963214874268,
      "learning_rate": 4.687481313397598e-06,
      "loss": 0.7096,
      "step": 25020
    },
    {
      "epoch": 10.25,
      "grad_norm": 2.130540132522583,
      "learning_rate": 4.687480674586156e-06,
      "loss": 0.7148,
      "step": 25030
    },
    {
      "epoch": 10.25,
      "grad_norm": 1.749649167060852,
      "learning_rate": 4.687480025038446e-06,
      "loss": 0.6923,
      "step": 25040
    },
    {
      "epoch": 10.25,
      "grad_norm": 3.3276140689849854,
      "learning_rate": 4.687479364754473e-06,
      "loss": 0.7046,
      "step": 25050
    },
    {
      "epoch": 10.26,
      "grad_norm": 2.849626302719116,
      "learning_rate": 4.687478693734239e-06,
      "loss": 0.7015,
      "step": 25060
    },
    {
      "epoch": 10.26,
      "grad_norm": 2.9141781330108643,
      "learning_rate": 4.687478011977745e-06,
      "loss": 0.6982,
      "step": 25070
    },
    {
      "epoch": 10.27,
      "grad_norm": 2.741922616958618,
      "learning_rate": 4.687477319484997e-06,
      "loss": 0.7131,
      "step": 25080
    },
    {
      "epoch": 10.27,
      "grad_norm": 3.241647481918335,
      "learning_rate": 4.687476616255997e-06,
      "loss": 0.6983,
      "step": 25090
    },
    {
      "epoch": 10.27,
      "grad_norm": 2.182096242904663,
      "learning_rate": 4.687475902290749e-06,
      "loss": 0.6687,
      "step": 25100
    },
    {
      "epoch": 10.28,
      "grad_norm": 2.8611714839935303,
      "learning_rate": 4.687475177589254e-06,
      "loss": 0.712,
      "step": 25110
    },
    {
      "epoch": 10.28,
      "grad_norm": 1.9566127061843872,
      "learning_rate": 4.687474442151518e-06,
      "loss": 0.718,
      "step": 25120
    },
    {
      "epoch": 10.29,
      "grad_norm": 1.9644968509674072,
      "learning_rate": 4.687473695977543e-06,
      "loss": 0.7012,
      "step": 25130
    },
    {
      "epoch": 10.29,
      "grad_norm": 4.157597064971924,
      "learning_rate": 4.687472939067334e-06,
      "loss": 0.7047,
      "step": 25140
    },
    {
      "epoch": 10.29,
      "grad_norm": 2.114457368850708,
      "learning_rate": 4.687472171420892e-06,
      "loss": 0.6887,
      "step": 25150
    },
    {
      "epoch": 10.3,
      "grad_norm": 2.040167808532715,
      "learning_rate": 4.687471393038222e-06,
      "loss": 0.7027,
      "step": 25160
    },
    {
      "epoch": 10.3,
      "grad_norm": 2.1352407932281494,
      "learning_rate": 4.687470603919327e-06,
      "loss": 0.6802,
      "step": 25170
    },
    {
      "epoch": 10.31,
      "grad_norm": 2.606066942214966,
      "learning_rate": 4.687469804064211e-06,
      "loss": 0.6811,
      "step": 25180
    },
    {
      "epoch": 10.31,
      "grad_norm": 2.324995994567871,
      "learning_rate": 4.687468993472877e-06,
      "loss": 0.7049,
      "step": 25190
    },
    {
      "epoch": 10.32,
      "grad_norm": 3.0513381958007812,
      "learning_rate": 4.68746817214533e-06,
      "loss": 0.6984,
      "step": 25200
    },
    {
      "epoch": 10.32,
      "grad_norm": 1.997780680656433,
      "learning_rate": 4.687467340081573e-06,
      "loss": 0.7061,
      "step": 25210
    },
    {
      "epoch": 10.32,
      "grad_norm": 2.3499419689178467,
      "learning_rate": 4.687466497281609e-06,
      "loss": 0.694,
      "step": 25220
    },
    {
      "epoch": 10.33,
      "grad_norm": 2.974126100540161,
      "learning_rate": 4.6874656437454445e-06,
      "loss": 0.6856,
      "step": 25230
    },
    {
      "epoch": 10.33,
      "grad_norm": 3.309898614883423,
      "learning_rate": 4.6874647794730804e-06,
      "loss": 0.693,
      "step": 25240
    },
    {
      "epoch": 10.34,
      "grad_norm": 1.9142029285430908,
      "learning_rate": 4.687463904464521e-06,
      "loss": 0.7241,
      "step": 25250
    },
    {
      "epoch": 10.34,
      "grad_norm": 3.0542056560516357,
      "learning_rate": 4.687463018719773e-06,
      "loss": 0.6931,
      "step": 25260
    },
    {
      "epoch": 10.34,
      "grad_norm": 3.363588809967041,
      "learning_rate": 4.687462122238838e-06,
      "loss": 0.7006,
      "step": 25270
    },
    {
      "epoch": 10.35,
      "grad_norm": 1.9926533699035645,
      "learning_rate": 4.68746121502172e-06,
      "loss": 0.7124,
      "step": 25280
    },
    {
      "epoch": 10.35,
      "grad_norm": 3.2510571479797363,
      "learning_rate": 4.687460297068424e-06,
      "loss": 0.7013,
      "step": 25290
    },
    {
      "epoch": 10.36,
      "grad_norm": 2.293304681777954,
      "learning_rate": 4.6874593683789545e-06,
      "loss": 0.6886,
      "step": 25300
    },
    {
      "epoch": 10.36,
      "grad_norm": 2.622661828994751,
      "learning_rate": 4.687458428953315e-06,
      "loss": 0.7045,
      "step": 25310
    },
    {
      "epoch": 10.36,
      "grad_norm": 2.0288281440734863,
      "learning_rate": 4.687457478791511e-06,
      "loss": 0.7035,
      "step": 25320
    },
    {
      "epoch": 10.37,
      "grad_norm": 2.0383684635162354,
      "learning_rate": 4.687456517893545e-06,
      "loss": 0.6937,
      "step": 25330
    },
    {
      "epoch": 10.37,
      "grad_norm": 3.229706048965454,
      "learning_rate": 4.687455546259423e-06,
      "loss": 0.7179,
      "step": 25340
    },
    {
      "epoch": 10.38,
      "grad_norm": 2.3551180362701416,
      "learning_rate": 4.687454563889149e-06,
      "loss": 0.684,
      "step": 25350
    },
    {
      "epoch": 10.38,
      "grad_norm": 3.378962755203247,
      "learning_rate": 4.687453570782727e-06,
      "loss": 0.6757,
      "step": 25360
    },
    {
      "epoch": 10.38,
      "grad_norm": 2.1525557041168213,
      "learning_rate": 4.687452566940162e-06,
      "loss": 0.7067,
      "step": 25370
    },
    {
      "epoch": 10.39,
      "grad_norm": 2.1182291507720947,
      "learning_rate": 4.6874515523614595e-06,
      "loss": 0.7102,
      "step": 25380
    },
    {
      "epoch": 10.39,
      "grad_norm": 2.4387550354003906,
      "learning_rate": 4.687450527046622e-06,
      "loss": 0.6986,
      "step": 25390
    },
    {
      "epoch": 10.4,
      "grad_norm": 2.6862294673919678,
      "learning_rate": 4.687449490995656e-06,
      "loss": 0.6807,
      "step": 25400
    },
    {
      "epoch": 10.4,
      "grad_norm": 3.1112961769104004,
      "learning_rate": 4.687448444208564e-06,
      "loss": 0.7127,
      "step": 25410
    },
    {
      "epoch": 10.41,
      "grad_norm": 2.4685683250427246,
      "learning_rate": 4.687447386685354e-06,
      "loss": 0.7001,
      "step": 25420
    },
    {
      "epoch": 10.41,
      "grad_norm": 2.3555707931518555,
      "learning_rate": 4.6874463184260295e-06,
      "loss": 0.6809,
      "step": 25430
    },
    {
      "epoch": 10.41,
      "grad_norm": 3.9830291271209717,
      "learning_rate": 4.687445239430593e-06,
      "loss": 0.6997,
      "step": 25440
    },
    {
      "epoch": 10.42,
      "grad_norm": 2.8116583824157715,
      "learning_rate": 4.687444149699054e-06,
      "loss": 0.6985,
      "step": 25450
    },
    {
      "epoch": 10.42,
      "grad_norm": 1.9374250173568726,
      "learning_rate": 4.6874430492314145e-06,
      "loss": 0.7127,
      "step": 25460
    },
    {
      "epoch": 10.43,
      "grad_norm": 1.6921192407608032,
      "learning_rate": 4.687441938027679e-06,
      "loss": 0.7028,
      "step": 25470
    },
    {
      "epoch": 10.43,
      "grad_norm": 2.468806028366089,
      "learning_rate": 4.687440816087854e-06,
      "loss": 0.6567,
      "step": 25480
    },
    {
      "epoch": 10.43,
      "grad_norm": 2.828578233718872,
      "learning_rate": 4.6874396834119454e-06,
      "loss": 0.6801,
      "step": 25490
    },
    {
      "epoch": 10.44,
      "grad_norm": 2.9606072902679443,
      "learning_rate": 4.687438539999956e-06,
      "loss": 0.702,
      "step": 25500
    },
    {
      "epoch": 10.44,
      "grad_norm": 3.40293025970459,
      "learning_rate": 4.687437385851894e-06,
      "loss": 0.6978,
      "step": 25510
    },
    {
      "epoch": 10.45,
      "grad_norm": 2.406680107116699,
      "learning_rate": 4.687436220967762e-06,
      "loss": 0.6851,
      "step": 25520
    },
    {
      "epoch": 10.45,
      "grad_norm": 2.9803128242492676,
      "learning_rate": 4.687435045347566e-06,
      "loss": 0.693,
      "step": 25530
    },
    {
      "epoch": 10.45,
      "grad_norm": 3.1892952919006348,
      "learning_rate": 4.687433858991312e-06,
      "loss": 0.6855,
      "step": 25540
    },
    {
      "epoch": 10.46,
      "grad_norm": 1.8335274457931519,
      "learning_rate": 4.687432661899005e-06,
      "loss": 0.6829,
      "step": 25550
    },
    {
      "epoch": 10.46,
      "grad_norm": 2.269484281539917,
      "learning_rate": 4.687431454070651e-06,
      "loss": 0.7019,
      "step": 25560
    },
    {
      "epoch": 10.47,
      "grad_norm": 2.6734249591827393,
      "learning_rate": 4.687430235506256e-06,
      "loss": 0.6889,
      "step": 25570
    },
    {
      "epoch": 10.47,
      "grad_norm": 4.091238975524902,
      "learning_rate": 4.687429006205824e-06,
      "loss": 0.6724,
      "step": 25580
    },
    {
      "epoch": 10.47,
      "grad_norm": 1.9147281646728516,
      "learning_rate": 4.687427766169362e-06,
      "loss": 0.7078,
      "step": 25590
    },
    {
      "epoch": 10.48,
      "grad_norm": 1.990689754486084,
      "learning_rate": 4.687426515396874e-06,
      "loss": 0.6836,
      "step": 25600
    },
    {
      "epoch": 10.48,
      "grad_norm": 2.3589296340942383,
      "learning_rate": 4.687425253888368e-06,
      "loss": 0.7137,
      "step": 25610
    },
    {
      "epoch": 10.49,
      "grad_norm": 1.9114412069320679,
      "learning_rate": 4.687423981643848e-06,
      "loss": 0.7031,
      "step": 25620
    },
    {
      "epoch": 10.49,
      "grad_norm": 3.007416248321533,
      "learning_rate": 4.687422698663321e-06,
      "loss": 0.7084,
      "step": 25630
    },
    {
      "epoch": 10.5,
      "grad_norm": 2.0107460021972656,
      "learning_rate": 4.687421404946791e-06,
      "loss": 0.7062,
      "step": 25640
    },
    {
      "epoch": 10.5,
      "grad_norm": 2.6612470149993896,
      "learning_rate": 4.687420100494267e-06,
      "loss": 0.6968,
      "step": 25650
    },
    {
      "epoch": 10.5,
      "grad_norm": 2.7106728553771973,
      "learning_rate": 4.687418785305752e-06,
      "loss": 0.6994,
      "step": 25660
    },
    {
      "epoch": 10.51,
      "grad_norm": 3.147707223892212,
      "learning_rate": 4.687417459381254e-06,
      "loss": 0.6722,
      "step": 25670
    },
    {
      "epoch": 10.51,
      "grad_norm": 2.362344741821289,
      "learning_rate": 4.687416122720778e-06,
      "loss": 0.7073,
      "step": 25680
    },
    {
      "epoch": 10.52,
      "grad_norm": 2.4987714290618896,
      "learning_rate": 4.687414775324331e-06,
      "loss": 0.6816,
      "step": 25690
    },
    {
      "epoch": 10.52,
      "grad_norm": 2.430905342102051,
      "learning_rate": 4.687413417191917e-06,
      "loss": 0.6969,
      "step": 25700
    },
    {
      "epoch": 10.52,
      "grad_norm": 2.1117959022521973,
      "learning_rate": 4.687412048323545e-06,
      "loss": 0.7051,
      "step": 25710
    },
    {
      "epoch": 10.53,
      "grad_norm": 2.781822919845581,
      "learning_rate": 4.687410668719221e-06,
      "loss": 0.6988,
      "step": 25720
    },
    {
      "epoch": 10.53,
      "grad_norm": 1.80507230758667,
      "learning_rate": 4.687409278378949e-06,
      "loss": 0.7082,
      "step": 25730
    },
    {
      "epoch": 10.54,
      "grad_norm": 3.3810460567474365,
      "learning_rate": 4.687407877302737e-06,
      "loss": 0.6703,
      "step": 25740
    },
    {
      "epoch": 10.54,
      "grad_norm": 2.356417655944824,
      "learning_rate": 4.6874064654905924e-06,
      "loss": 0.6946,
      "step": 25750
    },
    {
      "epoch": 10.54,
      "grad_norm": 2.5889596939086914,
      "learning_rate": 4.687405042942519e-06,
      "loss": 0.6805,
      "step": 25760
    },
    {
      "epoch": 10.55,
      "grad_norm": 2.492133140563965,
      "learning_rate": 4.687403609658525e-06,
      "loss": 0.7115,
      "step": 25770
    },
    {
      "epoch": 10.55,
      "grad_norm": 3.0978050231933594,
      "learning_rate": 4.6874021656386174e-06,
      "loss": 0.6852,
      "step": 25780
    },
    {
      "epoch": 10.56,
      "grad_norm": 2.7205240726470947,
      "learning_rate": 4.687400710882803e-06,
      "loss": 0.6959,
      "step": 25790
    },
    {
      "epoch": 10.56,
      "grad_norm": 2.214216470718384,
      "learning_rate": 4.687399245391086e-06,
      "loss": 0.6864,
      "step": 25800
    },
    {
      "epoch": 10.56,
      "grad_norm": 2.6653225421905518,
      "learning_rate": 4.687397769163475e-06,
      "loss": 0.6904,
      "step": 25810
    },
    {
      "epoch": 10.57,
      "grad_norm": 2.3047094345092773,
      "learning_rate": 4.687396282199978e-06,
      "loss": 0.6832,
      "step": 25820
    },
    {
      "epoch": 10.57,
      "grad_norm": 2.4146740436553955,
      "learning_rate": 4.687394784500599e-06,
      "loss": 0.701,
      "step": 25830
    },
    {
      "epoch": 10.58,
      "grad_norm": 2.0295214653015137,
      "learning_rate": 4.687393276065346e-06,
      "loss": 0.6702,
      "step": 25840
    },
    {
      "epoch": 10.58,
      "grad_norm": 3.377100706100464,
      "learning_rate": 4.6873917568942275e-06,
      "loss": 0.6961,
      "step": 25850
    },
    {
      "epoch": 10.59,
      "grad_norm": 3.0117759704589844,
      "learning_rate": 4.687390226987248e-06,
      "loss": 0.6884,
      "step": 25860
    },
    {
      "epoch": 10.59,
      "grad_norm": 2.7361230850219727,
      "learning_rate": 4.6873886863444155e-06,
      "loss": 0.7114,
      "step": 25870
    },
    {
      "epoch": 10.59,
      "grad_norm": 2.821955919265747,
      "learning_rate": 4.687387134965737e-06,
      "loss": 0.6953,
      "step": 25880
    },
    {
      "epoch": 10.6,
      "grad_norm": 2.2120916843414307,
      "learning_rate": 4.687385572851221e-06,
      "loss": 0.6953,
      "step": 25890
    },
    {
      "epoch": 10.6,
      "grad_norm": 2.662487030029297,
      "learning_rate": 4.687384000000872e-06,
      "loss": 0.7089,
      "step": 25900
    },
    {
      "epoch": 10.61,
      "grad_norm": 2.066650867462158,
      "learning_rate": 4.687382416414699e-06,
      "loss": 0.6982,
      "step": 25910
    },
    {
      "epoch": 10.61,
      "grad_norm": 2.3808786869049072,
      "learning_rate": 4.687380822092709e-06,
      "loss": 0.6839,
      "step": 25920
    },
    {
      "epoch": 10.61,
      "grad_norm": 3.525953769683838,
      "learning_rate": 4.687379217034909e-06,
      "loss": 0.6522,
      "step": 25930
    },
    {
      "epoch": 10.62,
      "grad_norm": 2.2790279388427734,
      "learning_rate": 4.687377601241307e-06,
      "loss": 0.707,
      "step": 25940
    },
    {
      "epoch": 10.62,
      "grad_norm": 2.803959608078003,
      "learning_rate": 4.68737597471191e-06,
      "loss": 0.6981,
      "step": 25950
    },
    {
      "epoch": 10.63,
      "grad_norm": 2.7468698024749756,
      "learning_rate": 4.687374337446725e-06,
      "loss": 0.683,
      "step": 25960
    },
    {
      "epoch": 10.63,
      "grad_norm": 3.0344924926757812,
      "learning_rate": 4.687372689445759e-06,
      "loss": 0.6873,
      "step": 25970
    },
    {
      "epoch": 10.63,
      "grad_norm": 3.0883588790893555,
      "learning_rate": 4.687371030709022e-06,
      "loss": 0.6888,
      "step": 25980
    },
    {
      "epoch": 10.64,
      "grad_norm": 2.3354992866516113,
      "learning_rate": 4.687369361236519e-06,
      "loss": 0.6968,
      "step": 25990
    },
    {
      "epoch": 10.64,
      "grad_norm": 2.2876029014587402,
      "learning_rate": 4.687367681028259e-06,
      "loss": 0.6947,
      "step": 26000
    },
    {
      "epoch": 10.65,
      "grad_norm": 1.6666908264160156,
      "learning_rate": 4.687365990084249e-06,
      "loss": 0.6928,
      "step": 26010
    },
    {
      "epoch": 10.65,
      "grad_norm": 2.593799352645874,
      "learning_rate": 4.6873642884044975e-06,
      "loss": 0.6914,
      "step": 26020
    },
    {
      "epoch": 10.65,
      "grad_norm": 3.4643771648406982,
      "learning_rate": 4.687362575989012e-06,
      "loss": 0.6799,
      "step": 26030
    },
    {
      "epoch": 10.66,
      "grad_norm": 2.3690757751464844,
      "learning_rate": 4.6873608528377995e-06,
      "loss": 0.6812,
      "step": 26040
    },
    {
      "epoch": 10.66,
      "grad_norm": 2.9256157875061035,
      "learning_rate": 4.687359118950869e-06,
      "loss": 0.6879,
      "step": 26050
    },
    {
      "epoch": 10.67,
      "grad_norm": 2.122204065322876,
      "learning_rate": 4.687357374328229e-06,
      "loss": 0.7013,
      "step": 26060
    },
    {
      "epoch": 10.67,
      "grad_norm": 2.8819291591644287,
      "learning_rate": 4.687355618969886e-06,
      "loss": 0.7188,
      "step": 26070
    },
    {
      "epoch": 10.68,
      "grad_norm": 2.8127224445343018,
      "learning_rate": 4.687353852875848e-06,
      "loss": 0.6907,
      "step": 26080
    },
    {
      "epoch": 10.68,
      "grad_norm": 2.286961317062378,
      "learning_rate": 4.687352076046124e-06,
      "loss": 0.707,
      "step": 26090
    },
    {
      "epoch": 10.68,
      "grad_norm": 2.345113515853882,
      "learning_rate": 4.687350288480722e-06,
      "loss": 0.7042,
      "step": 26100
    },
    {
      "epoch": 10.69,
      "grad_norm": 1.8122316598892212,
      "learning_rate": 4.68734849017965e-06,
      "loss": 0.6802,
      "step": 26110
    },
    {
      "epoch": 10.69,
      "grad_norm": 3.062197208404541,
      "learning_rate": 4.687346681142917e-06,
      "loss": 0.6909,
      "step": 26120
    },
    {
      "epoch": 10.7,
      "grad_norm": 2.3265304565429688,
      "learning_rate": 4.6873448613705295e-06,
      "loss": 0.6925,
      "step": 26130
    },
    {
      "epoch": 10.7,
      "grad_norm": 1.913813591003418,
      "learning_rate": 4.687343030862497e-06,
      "loss": 0.6944,
      "step": 26140
    },
    {
      "epoch": 10.7,
      "grad_norm": 2.394777774810791,
      "learning_rate": 4.687341189618828e-06,
      "loss": 0.687,
      "step": 26150
    },
    {
      "epoch": 10.71,
      "grad_norm": 2.4483203887939453,
      "learning_rate": 4.68733933763953e-06,
      "loss": 0.6882,
      "step": 26160
    },
    {
      "epoch": 10.71,
      "grad_norm": 2.6383721828460693,
      "learning_rate": 4.687337474924613e-06,
      "loss": 0.6847,
      "step": 26170
    },
    {
      "epoch": 10.72,
      "grad_norm": 2.125107765197754,
      "learning_rate": 4.687335601474085e-06,
      "loss": 0.6604,
      "step": 26180
    },
    {
      "epoch": 10.72,
      "grad_norm": 2.9519612789154053,
      "learning_rate": 4.687333717287954e-06,
      "loss": 0.6721,
      "step": 26190
    },
    {
      "epoch": 10.72,
      "grad_norm": 2.510152578353882,
      "learning_rate": 4.687331822366229e-06,
      "loss": 0.695,
      "step": 26200
    },
    {
      "epoch": 10.73,
      "grad_norm": 2.3917295932769775,
      "learning_rate": 4.687329916708918e-06,
      "loss": 0.684,
      "step": 26210
    },
    {
      "epoch": 10.73,
      "grad_norm": 3.3235976696014404,
      "learning_rate": 4.687328000316031e-06,
      "loss": 0.6862,
      "step": 26220
    },
    {
      "epoch": 10.74,
      "grad_norm": 1.98955500125885,
      "learning_rate": 4.6873260731875755e-06,
      "loss": 0.697,
      "step": 26230
    },
    {
      "epoch": 10.74,
      "grad_norm": 2.309033155441284,
      "learning_rate": 4.6873241353235615e-06,
      "loss": 0.6779,
      "step": 26240
    },
    {
      "epoch": 10.74,
      "grad_norm": 2.8523757457733154,
      "learning_rate": 4.687322186723997e-06,
      "loss": 0.6724,
      "step": 26250
    },
    {
      "epoch": 10.75,
      "grad_norm": 2.0541000366210938,
      "learning_rate": 4.687320227388892e-06,
      "loss": 0.6926,
      "step": 26260
    },
    {
      "epoch": 10.75,
      "grad_norm": 2.5302186012268066,
      "learning_rate": 4.6873182573182545e-06,
      "loss": 0.6886,
      "step": 26270
    },
    {
      "epoch": 10.76,
      "grad_norm": 3.2330284118652344,
      "learning_rate": 4.687316276512093e-06,
      "loss": 0.6775,
      "step": 26280
    },
    {
      "epoch": 10.76,
      "grad_norm": 2.587613821029663,
      "learning_rate": 4.687314284970418e-06,
      "loss": 0.7038,
      "step": 26290
    },
    {
      "epoch": 10.77,
      "grad_norm": 4.054261207580566,
      "learning_rate": 4.687312282693238e-06,
      "loss": 0.7015,
      "step": 26300
    },
    {
      "epoch": 10.77,
      "grad_norm": 3.2147157192230225,
      "learning_rate": 4.687310269680562e-06,
      "loss": 0.6619,
      "step": 26310
    },
    {
      "epoch": 10.77,
      "grad_norm": 2.8611838817596436,
      "learning_rate": 4.687308245932399e-06,
      "loss": 0.694,
      "step": 26320
    },
    {
      "epoch": 10.78,
      "grad_norm": 2.479928970336914,
      "learning_rate": 4.6873062114487595e-06,
      "loss": 0.6919,
      "step": 26330
    },
    {
      "epoch": 10.78,
      "grad_norm": 2.95701003074646,
      "learning_rate": 4.687304166229651e-06,
      "loss": 0.694,
      "step": 26340
    },
    {
      "epoch": 10.79,
      "grad_norm": 2.561511516571045,
      "learning_rate": 4.687302110275084e-06,
      "loss": 0.6911,
      "step": 26350
    },
    {
      "epoch": 10.79,
      "grad_norm": 3.5412354469299316,
      "learning_rate": 4.687300043585068e-06,
      "loss": 0.6914,
      "step": 26360
    },
    {
      "epoch": 10.79,
      "grad_norm": 2.560453176498413,
      "learning_rate": 4.687297966159612e-06,
      "loss": 0.6736,
      "step": 26370
    },
    {
      "epoch": 10.8,
      "grad_norm": 2.9427082538604736,
      "learning_rate": 4.687295877998726e-06,
      "loss": 0.6878,
      "step": 26380
    },
    {
      "epoch": 10.8,
      "grad_norm": 1.5904902219772339,
      "learning_rate": 4.687293779102419e-06,
      "loss": 0.7226,
      "step": 26390
    },
    {
      "epoch": 10.81,
      "grad_norm": 2.834355592727661,
      "learning_rate": 4.6872916694707015e-06,
      "loss": 0.7099,
      "step": 26400
    },
    {
      "epoch": 10.81,
      "grad_norm": 2.741042375564575,
      "learning_rate": 4.687289549103582e-06,
      "loss": 0.692,
      "step": 26410
    },
    {
      "epoch": 10.81,
      "grad_norm": 3.8194193840026855,
      "learning_rate": 4.6872874180010705e-06,
      "loss": 0.7053,
      "step": 26420
    },
    {
      "epoch": 10.82,
      "grad_norm": 3.8415262699127197,
      "learning_rate": 4.687285276163177e-06,
      "loss": 0.6894,
      "step": 26430
    },
    {
      "epoch": 10.82,
      "grad_norm": 3.256091833114624,
      "learning_rate": 4.687283123589912e-06,
      "loss": 0.6882,
      "step": 26440
    },
    {
      "epoch": 10.83,
      "grad_norm": 2.541295289993286,
      "learning_rate": 4.687280960281285e-06,
      "loss": 0.7123,
      "step": 26450
    },
    {
      "epoch": 10.83,
      "grad_norm": 3.148853302001953,
      "learning_rate": 4.687278786237305e-06,
      "loss": 0.6843,
      "step": 26460
    },
    {
      "epoch": 10.84,
      "grad_norm": 2.044623851776123,
      "learning_rate": 4.6872766014579815e-06,
      "loss": 0.6924,
      "step": 26470
    },
    {
      "epoch": 10.84,
      "grad_norm": 2.3053784370422363,
      "learning_rate": 4.687274405943326e-06,
      "loss": 0.6756,
      "step": 26480
    },
    {
      "epoch": 10.84,
      "grad_norm": 2.7232205867767334,
      "learning_rate": 4.6872721996933495e-06,
      "loss": 0.6696,
      "step": 26490
    },
    {
      "epoch": 10.85,
      "grad_norm": 2.9650683403015137,
      "learning_rate": 4.68726998270806e-06,
      "loss": 0.7126,
      "step": 26500
    },
    {
      "epoch": 10.85,
      "grad_norm": 3.887782096862793,
      "learning_rate": 4.687267754987469e-06,
      "loss": 0.6794,
      "step": 26510
    },
    {
      "epoch": 10.86,
      "grad_norm": 2.35567569732666,
      "learning_rate": 4.687265516531585e-06,
      "loss": 0.6864,
      "step": 26520
    },
    {
      "epoch": 10.86,
      "grad_norm": 2.241211414337158,
      "learning_rate": 4.68726326734042e-06,
      "loss": 0.6695,
      "step": 26530
    },
    {
      "epoch": 10.86,
      "grad_norm": 2.642488956451416,
      "learning_rate": 4.687261007413984e-06,
      "loss": 0.7026,
      "step": 26540
    },
    {
      "epoch": 10.87,
      "grad_norm": 3.0003321170806885,
      "learning_rate": 4.687258736752287e-06,
      "loss": 0.689,
      "step": 26550
    },
    {
      "epoch": 10.87,
      "grad_norm": 2.8683817386627197,
      "learning_rate": 4.687256455355338e-06,
      "loss": 0.6979,
      "step": 26560
    },
    {
      "epoch": 10.88,
      "grad_norm": 2.787285089492798,
      "learning_rate": 4.68725416322315e-06,
      "loss": 0.6711,
      "step": 26570
    },
    {
      "epoch": 10.88,
      "grad_norm": 2.740954637527466,
      "learning_rate": 4.687251860355732e-06,
      "loss": 0.684,
      "step": 26580
    },
    {
      "epoch": 10.88,
      "grad_norm": 2.5303289890289307,
      "learning_rate": 4.687249546753095e-06,
      "loss": 0.7011,
      "step": 26590
    },
    {
      "epoch": 10.89,
      "grad_norm": 1.838620901107788,
      "learning_rate": 4.68724722241525e-06,
      "loss": 0.6777,
      "step": 26600
    },
    {
      "epoch": 10.89,
      "grad_norm": 3.0334596633911133,
      "learning_rate": 4.687244887342207e-06,
      "loss": 0.6713,
      "step": 26610
    },
    {
      "epoch": 10.9,
      "grad_norm": 2.409771680831909,
      "learning_rate": 4.687242541533976e-06,
      "loss": 0.7002,
      "step": 26620
    },
    {
      "epoch": 10.9,
      "grad_norm": 2.034080743789673,
      "learning_rate": 4.687240184990569e-06,
      "loss": 0.6744,
      "step": 26630
    },
    {
      "epoch": 10.9,
      "grad_norm": 2.2195982933044434,
      "learning_rate": 4.6872378177119965e-06,
      "loss": 0.6959,
      "step": 26640
    },
    {
      "epoch": 10.91,
      "grad_norm": 1.8292652368545532,
      "learning_rate": 4.68723543969827e-06,
      "loss": 0.6863,
      "step": 26650
    },
    {
      "epoch": 10.91,
      "grad_norm": 3.1448676586151123,
      "learning_rate": 4.687233050949399e-06,
      "loss": 0.6994,
      "step": 26660
    },
    {
      "epoch": 10.92,
      "grad_norm": 2.7486188411712646,
      "learning_rate": 4.687230651465395e-06,
      "loss": 0.6933,
      "step": 26670
    },
    {
      "epoch": 10.92,
      "grad_norm": 2.3364832401275635,
      "learning_rate": 4.687228241246269e-06,
      "loss": 0.6926,
      "step": 26680
    },
    {
      "epoch": 10.93,
      "grad_norm": 2.8298938274383545,
      "learning_rate": 4.687225820292032e-06,
      "loss": 0.6755,
      "step": 26690
    },
    {
      "epoch": 10.93,
      "grad_norm": 2.234604597091675,
      "learning_rate": 4.687223388602696e-06,
      "loss": 0.7017,
      "step": 26700
    },
    {
      "epoch": 10.93,
      "grad_norm": 1.7761273384094238,
      "learning_rate": 4.68722094617827e-06,
      "loss": 0.7024,
      "step": 26710
    },
    {
      "epoch": 10.94,
      "grad_norm": 1.9667928218841553,
      "learning_rate": 4.6872184930187675e-06,
      "loss": 0.6638,
      "step": 26720
    },
    {
      "epoch": 10.94,
      "grad_norm": 2.359996795654297,
      "learning_rate": 4.687216029124199e-06,
      "loss": 0.6883,
      "step": 26730
    },
    {
      "epoch": 10.95,
      "grad_norm": 3.571237564086914,
      "learning_rate": 4.687213554494575e-06,
      "loss": 0.6708,
      "step": 26740
    },
    {
      "epoch": 10.95,
      "grad_norm": 3.065512180328369,
      "learning_rate": 4.687211069129908e-06,
      "loss": 0.6637,
      "step": 26750
    },
    {
      "epoch": 10.95,
      "grad_norm": 2.6516222953796387,
      "learning_rate": 4.687208573030208e-06,
      "loss": 0.7016,
      "step": 26760
    },
    {
      "epoch": 10.96,
      "grad_norm": 2.6633400917053223,
      "learning_rate": 4.687206066195487e-06,
      "loss": 0.6888,
      "step": 26770
    },
    {
      "epoch": 10.96,
      "grad_norm": 2.7876064777374268,
      "learning_rate": 4.687203548625758e-06,
      "loss": 0.6759,
      "step": 26780
    },
    {
      "epoch": 10.97,
      "grad_norm": 1.7773833274841309,
      "learning_rate": 4.68720102032103e-06,
      "loss": 0.6881,
      "step": 26790
    },
    {
      "epoch": 10.97,
      "grad_norm": 2.6893227100372314,
      "learning_rate": 4.687198481281316e-06,
      "loss": 0.6822,
      "step": 26800
    },
    {
      "epoch": 10.97,
      "grad_norm": 2.5288686752319336,
      "learning_rate": 4.687195931506628e-06,
      "loss": 0.7168,
      "step": 26810
    },
    {
      "epoch": 10.98,
      "grad_norm": 3.5325584411621094,
      "learning_rate": 4.6871933709969765e-06,
      "loss": 0.6978,
      "step": 26820
    },
    {
      "epoch": 10.98,
      "grad_norm": 2.1284050941467285,
      "learning_rate": 4.6871907997523744e-06,
      "loss": 0.6935,
      "step": 26830
    },
    {
      "epoch": 10.99,
      "grad_norm": 2.6065151691436768,
      "learning_rate": 4.6871882177728326e-06,
      "loss": 0.6779,
      "step": 26840
    },
    {
      "epoch": 10.99,
      "grad_norm": 2.5276975631713867,
      "learning_rate": 4.687185625058364e-06,
      "loss": 0.6997,
      "step": 26850
    },
    {
      "epoch": 10.99,
      "grad_norm": 3.192046642303467,
      "learning_rate": 4.6871830216089785e-06,
      "loss": 0.688,
      "step": 26860
    },
    {
      "epoch": 11.0,
      "grad_norm": 4.969594478607178,
      "learning_rate": 4.68718040742469e-06,
      "loss": 0.6991,
      "step": 26870
    },
    {
      "epoch": 11.0,
      "eval_loss": 0.6886573433876038,
      "eval_runtime": 52.1466,
      "eval_samples_per_second": 66.14,
      "eval_steps_per_second": 8.284,
      "step": 26873
    },
    {
      "epoch": 11.0,
      "grad_norm": 2.1909523010253906,
      "learning_rate": 4.687177782505509e-06,
      "loss": 0.6599,
      "step": 26880
    },
    {
      "epoch": 11.01,
      "grad_norm": 3.52243709564209,
      "learning_rate": 4.687175146851449e-06,
      "loss": 0.6886,
      "step": 26890
    },
    {
      "epoch": 11.01,
      "grad_norm": 3.5877573490142822,
      "learning_rate": 4.687172500462521e-06,
      "loss": 0.6784,
      "step": 26900
    },
    {
      "epoch": 11.02,
      "grad_norm": 2.596696376800537,
      "learning_rate": 4.6871698433387376e-06,
      "loss": 0.688,
      "step": 26910
    },
    {
      "epoch": 11.02,
      "grad_norm": 2.3700926303863525,
      "learning_rate": 4.687167175480111e-06,
      "loss": 0.6783,
      "step": 26920
    },
    {
      "epoch": 11.02,
      "grad_norm": 2.760956287384033,
      "learning_rate": 4.6871644968866535e-06,
      "loss": 0.6937,
      "step": 26930
    },
    {
      "epoch": 11.03,
      "grad_norm": 2.4728496074676514,
      "learning_rate": 4.687161807558376e-06,
      "loss": 0.6933,
      "step": 26940
    },
    {
      "epoch": 11.03,
      "grad_norm": 3.493293046951294,
      "learning_rate": 4.687159107495293e-06,
      "loss": 0.7083,
      "step": 26950
    },
    {
      "epoch": 11.04,
      "grad_norm": 3.16406512260437,
      "learning_rate": 4.687156396697415e-06,
      "loss": 0.6872,
      "step": 26960
    },
    {
      "epoch": 11.04,
      "grad_norm": 2.839670419692993,
      "learning_rate": 4.687153675164755e-06,
      "loss": 0.6932,
      "step": 26970
    },
    {
      "epoch": 11.04,
      "grad_norm": 2.73439621925354,
      "learning_rate": 4.687150942897326e-06,
      "loss": 0.6914,
      "step": 26980
    },
    {
      "epoch": 11.05,
      "grad_norm": 2.606069803237915,
      "learning_rate": 4.6871481998951415e-06,
      "loss": 0.7088,
      "step": 26990
    },
    {
      "epoch": 11.05,
      "grad_norm": 2.301542282104492,
      "learning_rate": 4.687145446158212e-06,
      "loss": 0.7274,
      "step": 27000
    },
    {
      "epoch": 11.06,
      "grad_norm": 3.360227346420288,
      "learning_rate": 4.68714268168655e-06,
      "loss": 0.6993,
      "step": 27010
    },
    {
      "epoch": 11.06,
      "grad_norm": 2.511772394180298,
      "learning_rate": 4.6871399064801695e-06,
      "loss": 0.7004,
      "step": 27020
    },
    {
      "epoch": 11.06,
      "grad_norm": 3.617428779602051,
      "learning_rate": 4.6871371205390835e-06,
      "loss": 0.6803,
      "step": 27030
    },
    {
      "epoch": 11.07,
      "grad_norm": 2.5989906787872314,
      "learning_rate": 4.687134323863303e-06,
      "loss": 0.7137,
      "step": 27040
    },
    {
      "epoch": 11.07,
      "grad_norm": 2.679950475692749,
      "learning_rate": 4.687131516452842e-06,
      "loss": 0.6834,
      "step": 27050
    },
    {
      "epoch": 11.08,
      "grad_norm": 2.178095579147339,
      "learning_rate": 4.687128698307714e-06,
      "loss": 0.6951,
      "step": 27060
    },
    {
      "epoch": 11.08,
      "grad_norm": 2.7041449546813965,
      "learning_rate": 4.6871258694279305e-06,
      "loss": 0.6796,
      "step": 27070
    },
    {
      "epoch": 11.08,
      "grad_norm": 3.1579508781433105,
      "learning_rate": 4.687123029813505e-06,
      "loss": 0.6748,
      "step": 27080
    },
    {
      "epoch": 11.09,
      "grad_norm": 2.5578224658966064,
      "learning_rate": 4.687120179464451e-06,
      "loss": 0.6829,
      "step": 27090
    },
    {
      "epoch": 11.09,
      "grad_norm": 3.4495041370391846,
      "learning_rate": 4.68711731838078e-06,
      "loss": 0.6745,
      "step": 27100
    },
    {
      "epoch": 11.1,
      "grad_norm": 2.713408946990967,
      "learning_rate": 4.6871144465625075e-06,
      "loss": 0.6848,
      "step": 27110
    },
    {
      "epoch": 11.1,
      "grad_norm": 3.3378944396972656,
      "learning_rate": 4.687111564009645e-06,
      "loss": 0.6697,
      "step": 27120
    },
    {
      "epoch": 11.11,
      "grad_norm": 2.2726471424102783,
      "learning_rate": 4.687108670722206e-06,
      "loss": 0.6883,
      "step": 27130
    },
    {
      "epoch": 11.11,
      "grad_norm": 2.504693031311035,
      "learning_rate": 4.6871057667002035e-06,
      "loss": 0.7059,
      "step": 27140
    },
    {
      "epoch": 11.11,
      "grad_norm": 2.3830270767211914,
      "learning_rate": 4.687102851943652e-06,
      "loss": 0.6933,
      "step": 27150
    },
    {
      "epoch": 11.12,
      "grad_norm": 2.148947238922119,
      "learning_rate": 4.687099926452563e-06,
      "loss": 0.6764,
      "step": 27160
    },
    {
      "epoch": 11.12,
      "grad_norm": 2.427180051803589,
      "learning_rate": 4.687096990226952e-06,
      "loss": 0.6898,
      "step": 27170
    },
    {
      "epoch": 11.13,
      "grad_norm": 2.6315064430236816,
      "learning_rate": 4.687094043266831e-06,
      "loss": 0.6827,
      "step": 27180
    },
    {
      "epoch": 11.13,
      "grad_norm": 1.7869350910186768,
      "learning_rate": 4.687091085572213e-06,
      "loss": 0.7025,
      "step": 27190
    },
    {
      "epoch": 11.13,
      "grad_norm": 2.8560404777526855,
      "learning_rate": 4.687088117143113e-06,
      "loss": 0.6735,
      "step": 27200
    },
    {
      "epoch": 11.14,
      "grad_norm": 1.6930092573165894,
      "learning_rate": 4.687085137979545e-06,
      "loss": 0.6605,
      "step": 27210
    },
    {
      "epoch": 11.14,
      "grad_norm": 2.26949143409729,
      "learning_rate": 4.68708214808152e-06,
      "loss": 0.6639,
      "step": 27220
    },
    {
      "epoch": 11.15,
      "grad_norm": 3.624324083328247,
      "learning_rate": 4.687079147449054e-06,
      "loss": 0.658,
      "step": 27230
    },
    {
      "epoch": 11.15,
      "grad_norm": 2.137108564376831,
      "learning_rate": 4.68707613608216e-06,
      "loss": 0.6933,
      "step": 27240
    },
    {
      "epoch": 11.15,
      "grad_norm": 2.1613664627075195,
      "learning_rate": 4.687073113980852e-06,
      "loss": 0.6805,
      "step": 27250
    },
    {
      "epoch": 11.16,
      "grad_norm": 1.9811687469482422,
      "learning_rate": 4.687070081145144e-06,
      "loss": 0.6882,
      "step": 27260
    },
    {
      "epoch": 11.16,
      "grad_norm": 1.7563241720199585,
      "learning_rate": 4.687067037575049e-06,
      "loss": 0.6951,
      "step": 27270
    },
    {
      "epoch": 11.17,
      "grad_norm": 3.646071195602417,
      "learning_rate": 4.687063983270582e-06,
      "loss": 0.6647,
      "step": 27280
    },
    {
      "epoch": 11.17,
      "grad_norm": 2.089448928833008,
      "learning_rate": 4.687060918231756e-06,
      "loss": 0.6853,
      "step": 27290
    },
    {
      "epoch": 11.17,
      "grad_norm": 2.1503045558929443,
      "learning_rate": 4.687057842458587e-06,
      "loss": 0.6852,
      "step": 27300
    },
    {
      "epoch": 11.18,
      "grad_norm": 2.280374526977539,
      "learning_rate": 4.6870547559510866e-06,
      "loss": 0.6796,
      "step": 27310
    },
    {
      "epoch": 11.18,
      "grad_norm": 2.4563345909118652,
      "learning_rate": 4.68705165870927e-06,
      "loss": 0.696,
      "step": 27320
    },
    {
      "epoch": 11.19,
      "grad_norm": 1.7647799253463745,
      "learning_rate": 4.6870485507331505e-06,
      "loss": 0.6755,
      "step": 27330
    },
    {
      "epoch": 11.19,
      "grad_norm": 2.9854204654693604,
      "learning_rate": 4.687045432022745e-06,
      "loss": 0.6827,
      "step": 27340
    },
    {
      "epoch": 11.2,
      "grad_norm": 2.4094719886779785,
      "learning_rate": 4.687042302578065e-06,
      "loss": 0.6823,
      "step": 27350
    },
    {
      "epoch": 11.2,
      "grad_norm": 1.9572683572769165,
      "learning_rate": 4.687039162399126e-06,
      "loss": 0.6834,
      "step": 27360
    },
    {
      "epoch": 11.2,
      "grad_norm": 2.280850648880005,
      "learning_rate": 4.6870360114859424e-06,
      "loss": 0.6846,
      "step": 27370
    },
    {
      "epoch": 11.21,
      "grad_norm": 2.6644997596740723,
      "learning_rate": 4.687032849838529e-06,
      "loss": 0.6811,
      "step": 27380
    },
    {
      "epoch": 11.21,
      "grad_norm": 2.5357985496520996,
      "learning_rate": 4.687029677456899e-06,
      "loss": 0.6655,
      "step": 27390
    },
    {
      "epoch": 11.22,
      "grad_norm": 2.342198610305786,
      "learning_rate": 4.687026494341068e-06,
      "loss": 0.6678,
      "step": 27400
    },
    {
      "epoch": 11.22,
      "grad_norm": 2.731424331665039,
      "learning_rate": 4.68702330049105e-06,
      "loss": 0.7003,
      "step": 27410
    },
    {
      "epoch": 11.22,
      "grad_norm": 3.5028717517852783,
      "learning_rate": 4.6870200959068605e-06,
      "loss": 0.6793,
      "step": 27420
    },
    {
      "epoch": 11.23,
      "grad_norm": 2.430804967880249,
      "learning_rate": 4.687016880588513e-06,
      "loss": 0.7034,
      "step": 27430
    },
    {
      "epoch": 11.23,
      "grad_norm": 3.008087635040283,
      "learning_rate": 4.687013654536024e-06,
      "loss": 0.7121,
      "step": 27440
    },
    {
      "epoch": 11.24,
      "grad_norm": 1.7742416858673096,
      "learning_rate": 4.687010417749405e-06,
      "loss": 0.6839,
      "step": 27450
    },
    {
      "epoch": 11.24,
      "grad_norm": 2.3631935119628906,
      "learning_rate": 4.687007170228674e-06,
      "loss": 0.6917,
      "step": 27460
    },
    {
      "epoch": 11.24,
      "grad_norm": 2.0613272190093994,
      "learning_rate": 4.687003911973845e-06,
      "loss": 0.6807,
      "step": 27470
    },
    {
      "epoch": 11.25,
      "grad_norm": 2.5094480514526367,
      "learning_rate": 4.687000642984933e-06,
      "loss": 0.6729,
      "step": 27480
    },
    {
      "epoch": 11.25,
      "grad_norm": 1.9880434274673462,
      "learning_rate": 4.686997363261953e-06,
      "loss": 0.687,
      "step": 27490
    },
    {
      "epoch": 11.26,
      "grad_norm": 2.2001800537109375,
      "learning_rate": 4.686994072804918e-06,
      "loss": 0.6953,
      "step": 27500
    },
    {
      "epoch": 11.26,
      "grad_norm": 2.044351816177368,
      "learning_rate": 4.686990771613846e-06,
      "loss": 0.693,
      "step": 27510
    },
    {
      "epoch": 11.26,
      "grad_norm": 1.8860735893249512,
      "learning_rate": 4.686987459688751e-06,
      "loss": 0.685,
      "step": 27520
    },
    {
      "epoch": 11.27,
      "grad_norm": 2.402376651763916,
      "learning_rate": 4.6869841370296476e-06,
      "loss": 0.6884,
      "step": 27530
    },
    {
      "epoch": 11.27,
      "grad_norm": 2.513446569442749,
      "learning_rate": 4.686980803636551e-06,
      "loss": 0.6848,
      "step": 27540
    },
    {
      "epoch": 11.28,
      "grad_norm": 2.554297685623169,
      "learning_rate": 4.686977459509478e-06,
      "loss": 0.6669,
      "step": 27550
    },
    {
      "epoch": 11.28,
      "grad_norm": 2.368933916091919,
      "learning_rate": 4.686974104648444e-06,
      "loss": 0.7016,
      "step": 27560
    },
    {
      "epoch": 11.29,
      "grad_norm": 2.4268438816070557,
      "learning_rate": 4.686970739053461e-06,
      "loss": 0.6775,
      "step": 27570
    },
    {
      "epoch": 11.29,
      "grad_norm": 3.153019428253174,
      "learning_rate": 4.686967362724548e-06,
      "loss": 0.6698,
      "step": 27580
    },
    {
      "epoch": 11.29,
      "grad_norm": 2.4414565563201904,
      "learning_rate": 4.686963975661718e-06,
      "loss": 0.6656,
      "step": 27590
    },
    {
      "epoch": 11.3,
      "grad_norm": 2.724841356277466,
      "learning_rate": 4.686960577864989e-06,
      "loss": 0.6872,
      "step": 27600
    },
    {
      "epoch": 11.3,
      "grad_norm": 3.7658498287200928,
      "learning_rate": 4.686957169334375e-06,
      "loss": 0.6895,
      "step": 27610
    },
    {
      "epoch": 11.31,
      "grad_norm": 2.3328750133514404,
      "learning_rate": 4.6869537500698916e-06,
      "loss": 0.6906,
      "step": 27620
    },
    {
      "epoch": 11.31,
      "grad_norm": 1.8621257543563843,
      "learning_rate": 4.686950320071554e-06,
      "loss": 0.6839,
      "step": 27630
    },
    {
      "epoch": 11.31,
      "grad_norm": 2.2076165676116943,
      "learning_rate": 4.686946879339379e-06,
      "loss": 0.675,
      "step": 27640
    },
    {
      "epoch": 11.32,
      "grad_norm": 1.959406852722168,
      "learning_rate": 4.6869434278733825e-06,
      "loss": 0.6805,
      "step": 27650
    },
    {
      "epoch": 11.32,
      "grad_norm": 3.065192699432373,
      "learning_rate": 4.68693996567358e-06,
      "loss": 0.6784,
      "step": 27660
    },
    {
      "epoch": 11.33,
      "grad_norm": 2.1844379901885986,
      "learning_rate": 4.686936492739986e-06,
      "loss": 0.6871,
      "step": 27670
    },
    {
      "epoch": 11.33,
      "grad_norm": 3.059173345565796,
      "learning_rate": 4.686933009072619e-06,
      "loss": 0.6928,
      "step": 27680
    },
    {
      "epoch": 11.33,
      "grad_norm": 2.992701530456543,
      "learning_rate": 4.686929514671492e-06,
      "loss": 0.6828,
      "step": 27690
    },
    {
      "epoch": 11.34,
      "grad_norm": 4.036897659301758,
      "learning_rate": 4.686926009536624e-06,
      "loss": 0.693,
      "step": 27700
    },
    {
      "epoch": 11.34,
      "grad_norm": 2.83743953704834,
      "learning_rate": 4.686922493668029e-06,
      "loss": 0.6936,
      "step": 27710
    },
    {
      "epoch": 11.35,
      "grad_norm": 2.121931552886963,
      "learning_rate": 4.686918967065724e-06,
      "loss": 0.6887,
      "step": 27720
    },
    {
      "epoch": 11.35,
      "grad_norm": 2.925499200820923,
      "learning_rate": 4.686915429729724e-06,
      "loss": 0.6863,
      "step": 27730
    },
    {
      "epoch": 11.35,
      "grad_norm": 3.0832624435424805,
      "learning_rate": 4.686911881660046e-06,
      "loss": 0.6749,
      "step": 27740
    },
    {
      "epoch": 11.36,
      "grad_norm": 2.3408312797546387,
      "learning_rate": 4.686908322856707e-06,
      "loss": 0.6948,
      "step": 27750
    },
    {
      "epoch": 11.36,
      "grad_norm": 2.539793014526367,
      "learning_rate": 4.6869047533197225e-06,
      "loss": 0.6829,
      "step": 27760
    },
    {
      "epoch": 11.37,
      "grad_norm": 3.1257271766662598,
      "learning_rate": 4.686901173049109e-06,
      "loss": 0.6896,
      "step": 27770
    },
    {
      "epoch": 11.37,
      "grad_norm": 2.0589401721954346,
      "learning_rate": 4.6868975820448825e-06,
      "loss": 0.6925,
      "step": 27780
    },
    {
      "epoch": 11.38,
      "grad_norm": 3.5457634925842285,
      "learning_rate": 4.68689398030706e-06,
      "loss": 0.6775,
      "step": 27790
    },
    {
      "epoch": 11.38,
      "grad_norm": 2.568138837814331,
      "learning_rate": 4.686890367835658e-06,
      "loss": 0.6832,
      "step": 27800
    },
    {
      "epoch": 11.38,
      "grad_norm": 2.7961463928222656,
      "learning_rate": 4.6868867446306924e-06,
      "loss": 0.6694,
      "step": 27810
    },
    {
      "epoch": 11.39,
      "grad_norm": 2.212909698486328,
      "learning_rate": 4.686883110692181e-06,
      "loss": 0.677,
      "step": 27820
    },
    {
      "epoch": 11.39,
      "grad_norm": 3.4182991981506348,
      "learning_rate": 4.686879466020139e-06,
      "loss": 0.6913,
      "step": 27830
    },
    {
      "epoch": 11.4,
      "grad_norm": 2.392984628677368,
      "learning_rate": 4.686875810614584e-06,
      "loss": 0.6808,
      "step": 27840
    },
    {
      "epoch": 11.4,
      "grad_norm": 1.8032842874526978,
      "learning_rate": 4.6868721444755325e-06,
      "loss": 0.6722,
      "step": 27850
    },
    {
      "epoch": 11.4,
      "grad_norm": 2.0573062896728516,
      "learning_rate": 4.686868467603001e-06,
      "loss": 0.6789,
      "step": 27860
    },
    {
      "epoch": 11.41,
      "grad_norm": 2.362434148788452,
      "learning_rate": 4.686864779997008e-06,
      "loss": 0.67,
      "step": 27870
    },
    {
      "epoch": 11.41,
      "grad_norm": 2.8705732822418213,
      "learning_rate": 4.686861081657568e-06,
      "loss": 0.7094,
      "step": 27880
    },
    {
      "epoch": 11.42,
      "grad_norm": 3.1803247928619385,
      "learning_rate": 4.686857372584699e-06,
      "loss": 0.6887,
      "step": 27890
    },
    {
      "epoch": 11.42,
      "grad_norm": 2.1321051120758057,
      "learning_rate": 4.686853652778419e-06,
      "loss": 0.6715,
      "step": 27900
    },
    {
      "epoch": 11.42,
      "grad_norm": 3.3412773609161377,
      "learning_rate": 4.686849922238742e-06,
      "loss": 0.6823,
      "step": 27910
    },
    {
      "epoch": 11.43,
      "grad_norm": 2.4408955574035645,
      "learning_rate": 4.686846180965689e-06,
      "loss": 0.684,
      "step": 27920
    },
    {
      "epoch": 11.43,
      "grad_norm": 3.3059091567993164,
      "learning_rate": 4.686842428959275e-06,
      "loss": 0.6787,
      "step": 27930
    },
    {
      "epoch": 11.44,
      "grad_norm": 2.5585458278656006,
      "learning_rate": 4.686838666219516e-06,
      "loss": 0.6769,
      "step": 27940
    },
    {
      "epoch": 11.44,
      "grad_norm": 2.2485756874084473,
      "learning_rate": 4.686834892746432e-06,
      "loss": 0.6723,
      "step": 27950
    },
    {
      "epoch": 11.44,
      "grad_norm": 2.7146801948547363,
      "learning_rate": 4.68683110854004e-06,
      "loss": 0.6591,
      "step": 27960
    },
    {
      "epoch": 11.45,
      "grad_norm": 2.030329465866089,
      "learning_rate": 4.686827313600354e-06,
      "loss": 0.6647,
      "step": 27970
    },
    {
      "epoch": 11.45,
      "grad_norm": 2.774529457092285,
      "learning_rate": 4.686823507927395e-06,
      "loss": 0.6916,
      "step": 27980
    },
    {
      "epoch": 11.46,
      "grad_norm": 2.988539218902588,
      "learning_rate": 4.686819691521179e-06,
      "loss": 0.6777,
      "step": 27990
    },
    {
      "epoch": 11.46,
      "grad_norm": 3.1057229042053223,
      "learning_rate": 4.6868158643817236e-06,
      "loss": 0.6929,
      "step": 28000
    },
    {
      "epoch": 11.47,
      "grad_norm": 3.8644051551818848,
      "learning_rate": 4.6868120265090464e-06,
      "loss": 0.673,
      "step": 28010
    },
    {
      "epoch": 11.47,
      "grad_norm": 2.855414867401123,
      "learning_rate": 4.686808177903165e-06,
      "loss": 0.6671,
      "step": 28020
    },
    {
      "epoch": 11.47,
      "grad_norm": 2.6637566089630127,
      "learning_rate": 4.686804318564096e-06,
      "loss": 0.6817,
      "step": 28030
    },
    {
      "epoch": 11.48,
      "grad_norm": 2.1276779174804688,
      "learning_rate": 4.686800448491859e-06,
      "loss": 0.6945,
      "step": 28040
    },
    {
      "epoch": 11.48,
      "grad_norm": 3.080754041671753,
      "learning_rate": 4.686796567686471e-06,
      "loss": 0.6763,
      "step": 28050
    },
    {
      "epoch": 11.49,
      "grad_norm": 2.4650816917419434,
      "learning_rate": 4.6867926761479486e-06,
      "loss": 0.6686,
      "step": 28060
    },
    {
      "epoch": 11.49,
      "grad_norm": 1.9025065898895264,
      "learning_rate": 4.68678877387631e-06,
      "loss": 0.665,
      "step": 28070
    },
    {
      "epoch": 11.49,
      "grad_norm": 2.6453938484191895,
      "learning_rate": 4.686784860871576e-06,
      "loss": 0.6906,
      "step": 28080
    },
    {
      "epoch": 11.5,
      "grad_norm": 2.286628484725952,
      "learning_rate": 4.6867809371337604e-06,
      "loss": 0.6813,
      "step": 28090
    },
    {
      "epoch": 11.5,
      "grad_norm": 3.1457364559173584,
      "learning_rate": 4.686777002662884e-06,
      "loss": 0.6781,
      "step": 28100
    },
    {
      "epoch": 11.51,
      "grad_norm": 2.296091079711914,
      "learning_rate": 4.686773057458962e-06,
      "loss": 0.6909,
      "step": 28110
    },
    {
      "epoch": 11.51,
      "grad_norm": 2.2914862632751465,
      "learning_rate": 4.6867691015220165e-06,
      "loss": 0.6887,
      "step": 28120
    },
    {
      "epoch": 11.51,
      "grad_norm": 3.010692834854126,
      "learning_rate": 4.6867651348520615e-06,
      "loss": 0.6902,
      "step": 28130
    },
    {
      "epoch": 11.52,
      "grad_norm": 2.207268476486206,
      "learning_rate": 4.686761157449117e-06,
      "loss": 0.675,
      "step": 28140
    },
    {
      "epoch": 11.52,
      "grad_norm": 2.1728105545043945,
      "learning_rate": 4.686757169313203e-06,
      "loss": 0.6681,
      "step": 28150
    },
    {
      "epoch": 11.53,
      "grad_norm": 1.9767253398895264,
      "learning_rate": 4.686753170444335e-06,
      "loss": 0.6873,
      "step": 28160
    },
    {
      "epoch": 11.53,
      "grad_norm": 2.312825918197632,
      "learning_rate": 4.686749160842533e-06,
      "loss": 0.6572,
      "step": 28170
    },
    {
      "epoch": 11.53,
      "grad_norm": 1.9445289373397827,
      "learning_rate": 4.686745140507813e-06,
      "loss": 0.6924,
      "step": 28180
    },
    {
      "epoch": 11.54,
      "grad_norm": 3.1931371688842773,
      "learning_rate": 4.6867411094401966e-06,
      "loss": 0.6954,
      "step": 28190
    },
    {
      "epoch": 11.54,
      "grad_norm": 2.5471410751342773,
      "learning_rate": 4.686737067639701e-06,
      "loss": 0.6678,
      "step": 28200
    },
    {
      "epoch": 11.55,
      "grad_norm": 1.8074887990951538,
      "learning_rate": 4.686733015106343e-06,
      "loss": 0.6677,
      "step": 28210
    },
    {
      "epoch": 11.55,
      "grad_norm": 3.3204891681671143,
      "learning_rate": 4.686728951840144e-06,
      "loss": 0.6857,
      "step": 28220
    },
    {
      "epoch": 11.56,
      "grad_norm": 2.6832776069641113,
      "learning_rate": 4.6867248778411205e-06,
      "loss": 0.6821,
      "step": 28230
    },
    {
      "epoch": 11.56,
      "grad_norm": 2.6067144870758057,
      "learning_rate": 4.6867207931092925e-06,
      "loss": 0.6935,
      "step": 28240
    },
    {
      "epoch": 11.56,
      "grad_norm": 2.140408992767334,
      "learning_rate": 4.686716697644678e-06,
      "loss": 0.6672,
      "step": 28250
    },
    {
      "epoch": 11.57,
      "grad_norm": 2.6425442695617676,
      "learning_rate": 4.686712591447295e-06,
      "loss": 0.6688,
      "step": 28260
    },
    {
      "epoch": 11.57,
      "grad_norm": 3.699582576751709,
      "learning_rate": 4.686708474517165e-06,
      "loss": 0.6622,
      "step": 28270
    },
    {
      "epoch": 11.58,
      "grad_norm": 2.707691192626953,
      "learning_rate": 4.686704346854304e-06,
      "loss": 0.6774,
      "step": 28280
    },
    {
      "epoch": 11.58,
      "grad_norm": 2.89558744430542,
      "learning_rate": 4.686700208458731e-06,
      "loss": 0.6831,
      "step": 28290
    },
    {
      "epoch": 11.58,
      "grad_norm": 2.6950831413269043,
      "learning_rate": 4.686696059330468e-06,
      "loss": 0.665,
      "step": 28300
    },
    {
      "epoch": 11.59,
      "grad_norm": 3.2357113361358643,
      "learning_rate": 4.686691899469531e-06,
      "loss": 0.686,
      "step": 28310
    },
    {
      "epoch": 11.59,
      "grad_norm": 1.8723901510238647,
      "learning_rate": 4.68668772887594e-06,
      "loss": 0.6717,
      "step": 28320
    },
    {
      "epoch": 11.6,
      "grad_norm": 2.536506414413452,
      "learning_rate": 4.686683547549714e-06,
      "loss": 0.6787,
      "step": 28330
    },
    {
      "epoch": 11.6,
      "grad_norm": 2.5745291709899902,
      "learning_rate": 4.686679355490873e-06,
      "loss": 0.6577,
      "step": 28340
    },
    {
      "epoch": 11.6,
      "grad_norm": 2.588627576828003,
      "learning_rate": 4.686675152699434e-06,
      "loss": 0.6835,
      "step": 28350
    },
    {
      "epoch": 11.61,
      "grad_norm": 2.320115804672241,
      "learning_rate": 4.686670939175419e-06,
      "loss": 0.7059,
      "step": 28360
    },
    {
      "epoch": 11.61,
      "grad_norm": 2.594197988510132,
      "learning_rate": 4.686666714918845e-06,
      "loss": 0.6696,
      "step": 28370
    },
    {
      "epoch": 11.62,
      "grad_norm": 2.5139822959899902,
      "learning_rate": 4.686662479929734e-06,
      "loss": 0.6941,
      "step": 28380
    },
    {
      "epoch": 11.62,
      "grad_norm": 2.821859359741211,
      "learning_rate": 4.686658234208103e-06,
      "loss": 0.6709,
      "step": 28390
    },
    {
      "epoch": 11.63,
      "grad_norm": 3.0010721683502197,
      "learning_rate": 4.686653977753972e-06,
      "loss": 0.6737,
      "step": 28400
    },
    {
      "epoch": 11.63,
      "grad_norm": 2.6950082778930664,
      "learning_rate": 4.68664971056736e-06,
      "loss": 0.6876,
      "step": 28410
    },
    {
      "epoch": 11.63,
      "grad_norm": 3.8451743125915527,
      "learning_rate": 4.686645432648288e-06,
      "loss": 0.6656,
      "step": 28420
    },
    {
      "epoch": 11.64,
      "grad_norm": 2.403587579727173,
      "learning_rate": 4.686641143996775e-06,
      "loss": 0.6776,
      "step": 28430
    },
    {
      "epoch": 11.64,
      "grad_norm": 2.8374884128570557,
      "learning_rate": 4.686636844612841e-06,
      "loss": 0.6929,
      "step": 28440
    },
    {
      "epoch": 11.65,
      "grad_norm": 2.8849236965179443,
      "learning_rate": 4.686632534496504e-06,
      "loss": 0.6735,
      "step": 28450
    },
    {
      "epoch": 11.65,
      "grad_norm": 2.7496225833892822,
      "learning_rate": 4.686628213647785e-06,
      "loss": 0.6622,
      "step": 28460
    },
    {
      "epoch": 11.65,
      "grad_norm": 1.990335464477539,
      "learning_rate": 4.686623882066705e-06,
      "loss": 0.6867,
      "step": 28470
    },
    {
      "epoch": 11.66,
      "grad_norm": 2.8490123748779297,
      "learning_rate": 4.686619539753282e-06,
      "loss": 0.6738,
      "step": 28480
    },
    {
      "epoch": 11.66,
      "grad_norm": 2.5191152095794678,
      "learning_rate": 4.686615186707536e-06,
      "loss": 0.657,
      "step": 28490
    },
    {
      "epoch": 11.67,
      "grad_norm": 3.053858757019043,
      "learning_rate": 4.686610822929488e-06,
      "loss": 0.6735,
      "step": 28500
    },
    {
      "epoch": 11.67,
      "grad_norm": 2.1488187313079834,
      "learning_rate": 4.686606448419157e-06,
      "loss": 0.6765,
      "step": 28510
    },
    {
      "epoch": 11.67,
      "grad_norm": 3.32307767868042,
      "learning_rate": 4.686602063176564e-06,
      "loss": 0.665,
      "step": 28520
    },
    {
      "epoch": 11.68,
      "grad_norm": 2.598156690597534,
      "learning_rate": 4.686597667201727e-06,
      "loss": 0.6812,
      "step": 28530
    },
    {
      "epoch": 11.68,
      "grad_norm": 1.8842664957046509,
      "learning_rate": 4.68659326049467e-06,
      "loss": 0.7118,
      "step": 28540
    },
    {
      "epoch": 11.69,
      "grad_norm": 2.349071741104126,
      "learning_rate": 4.686588843055409e-06,
      "loss": 0.6736,
      "step": 28550
    },
    {
      "epoch": 11.69,
      "grad_norm": 3.0039420127868652,
      "learning_rate": 4.686584414883967e-06,
      "loss": 0.6949,
      "step": 28560
    },
    {
      "epoch": 11.69,
      "grad_norm": 2.55741548538208,
      "learning_rate": 4.686579975980363e-06,
      "loss": 0.6855,
      "step": 28570
    },
    {
      "epoch": 11.7,
      "grad_norm": 2.1997241973876953,
      "learning_rate": 4.6865755263446184e-06,
      "loss": 0.6892,
      "step": 28580
    },
    {
      "epoch": 11.7,
      "grad_norm": 3.565005302429199,
      "learning_rate": 4.6865710659767515e-06,
      "loss": 0.6692,
      "step": 28590
    },
    {
      "epoch": 11.71,
      "grad_norm": 2.8091580867767334,
      "learning_rate": 4.686566594876785e-06,
      "loss": 0.6986,
      "step": 28600
    },
    {
      "epoch": 11.71,
      "grad_norm": 2.8188135623931885,
      "learning_rate": 4.686562113044739e-06,
      "loss": 0.675,
      "step": 28610
    },
    {
      "epoch": 11.72,
      "grad_norm": 2.2534983158111572,
      "learning_rate": 4.686557620480633e-06,
      "loss": 0.6716,
      "step": 28620
    },
    {
      "epoch": 11.72,
      "grad_norm": 2.1515860557556152,
      "learning_rate": 4.686553117184487e-06,
      "loss": 0.6822,
      "step": 28630
    },
    {
      "epoch": 11.72,
      "grad_norm": 3.0715677738189697,
      "learning_rate": 4.686548603156325e-06,
      "loss": 0.6639,
      "step": 28640
    },
    {
      "epoch": 11.73,
      "grad_norm": 2.096536874771118,
      "learning_rate": 4.686544078396165e-06,
      "loss": 0.6937,
      "step": 28650
    },
    {
      "epoch": 11.73,
      "grad_norm": 2.7585742473602295,
      "learning_rate": 4.686539542904027e-06,
      "loss": 0.667,
      "step": 28660
    },
    {
      "epoch": 11.74,
      "grad_norm": 2.4776599407196045,
      "learning_rate": 4.686534996679933e-06,
      "loss": 0.6864,
      "step": 28670
    },
    {
      "epoch": 11.74,
      "grad_norm": 2.05702805519104,
      "learning_rate": 4.686530439723904e-06,
      "loss": 0.6636,
      "step": 28680
    },
    {
      "epoch": 11.74,
      "grad_norm": 2.30255126953125,
      "learning_rate": 4.6865258720359606e-06,
      "loss": 0.6659,
      "step": 28690
    },
    {
      "epoch": 11.75,
      "grad_norm": 2.068040370941162,
      "learning_rate": 4.686521293616125e-06,
      "loss": 0.6687,
      "step": 28700
    },
    {
      "epoch": 11.75,
      "grad_norm": 2.3919811248779297,
      "learning_rate": 4.686516704464415e-06,
      "loss": 0.6711,
      "step": 28710
    },
    {
      "epoch": 11.76,
      "grad_norm": 2.308415412902832,
      "learning_rate": 4.686512104580854e-06,
      "loss": 0.7056,
      "step": 28720
    },
    {
      "epoch": 11.76,
      "grad_norm": 2.647996187210083,
      "learning_rate": 4.686507493965464e-06,
      "loss": 0.6732,
      "step": 28730
    },
    {
      "epoch": 11.76,
      "grad_norm": 2.3793694972991943,
      "learning_rate": 4.686502872618263e-06,
      "loss": 0.6643,
      "step": 28740
    },
    {
      "epoch": 11.77,
      "grad_norm": 2.873857259750366,
      "learning_rate": 4.686498240539276e-06,
      "loss": 0.6858,
      "step": 28750
    },
    {
      "epoch": 11.77,
      "grad_norm": 2.4835989475250244,
      "learning_rate": 4.68649359772852e-06,
      "loss": 0.6694,
      "step": 28760
    },
    {
      "epoch": 11.78,
      "grad_norm": 2.2369980812072754,
      "learning_rate": 4.686488944186019e-06,
      "loss": 0.6757,
      "step": 28770
    },
    {
      "epoch": 11.78,
      "grad_norm": 2.128610610961914,
      "learning_rate": 4.686484279911794e-06,
      "loss": 0.6771,
      "step": 28780
    },
    {
      "epoch": 11.78,
      "grad_norm": 2.2555489540100098,
      "learning_rate": 4.686479604905866e-06,
      "loss": 0.654,
      "step": 28790
    },
    {
      "epoch": 11.79,
      "grad_norm": 2.4723503589630127,
      "learning_rate": 4.686474919168257e-06,
      "loss": 0.6659,
      "step": 28800
    },
    {
      "epoch": 11.79,
      "grad_norm": 3.2864153385162354,
      "learning_rate": 4.686470222698987e-06,
      "loss": 0.671,
      "step": 28810
    },
    {
      "epoch": 11.8,
      "grad_norm": 2.112687349319458,
      "learning_rate": 4.686465515498079e-06,
      "loss": 0.6911,
      "step": 28820
    },
    {
      "epoch": 11.8,
      "grad_norm": 2.209982395172119,
      "learning_rate": 4.6864607975655545e-06,
      "loss": 0.6598,
      "step": 28830
    },
    {
      "epoch": 11.81,
      "grad_norm": 2.171377420425415,
      "learning_rate": 4.686456068901434e-06,
      "loss": 0.6729,
      "step": 28840
    },
    {
      "epoch": 11.81,
      "grad_norm": 2.992546558380127,
      "learning_rate": 4.68645132950574e-06,
      "loss": 0.6813,
      "step": 28850
    },
    {
      "epoch": 11.81,
      "grad_norm": 2.8591158390045166,
      "learning_rate": 4.6864465793784935e-06,
      "loss": 0.6853,
      "step": 28860
    },
    {
      "epoch": 11.82,
      "grad_norm": 2.437727451324463,
      "learning_rate": 4.686441818519718e-06,
      "loss": 0.6696,
      "step": 28870
    },
    {
      "epoch": 11.82,
      "grad_norm": 2.8211281299591064,
      "learning_rate": 4.686437046929434e-06,
      "loss": 0.6717,
      "step": 28880
    },
    {
      "epoch": 11.83,
      "grad_norm": 3.0020124912261963,
      "learning_rate": 4.686432264607662e-06,
      "loss": 0.6792,
      "step": 28890
    },
    {
      "epoch": 11.83,
      "grad_norm": 1.8744521141052246,
      "learning_rate": 4.686427471554426e-06,
      "loss": 0.6819,
      "step": 28900
    },
    {
      "epoch": 11.83,
      "grad_norm": 2.084411144256592,
      "learning_rate": 4.686422667769748e-06,
      "loss": 0.685,
      "step": 28910
    },
    {
      "epoch": 11.84,
      "grad_norm": 4.023205280303955,
      "learning_rate": 4.686417853253648e-06,
      "loss": 0.6884,
      "step": 28920
    },
    {
      "epoch": 11.84,
      "grad_norm": 3.4320478439331055,
      "learning_rate": 4.686413028006151e-06,
      "loss": 0.6553,
      "step": 28930
    },
    {
      "epoch": 11.85,
      "grad_norm": 3.0021841526031494,
      "learning_rate": 4.686408192027276e-06,
      "loss": 0.6733,
      "step": 28940
    },
    {
      "epoch": 11.85,
      "grad_norm": 2.851950168609619,
      "learning_rate": 4.686403345317047e-06,
      "loss": 0.6901,
      "step": 28950
    },
    {
      "epoch": 11.85,
      "grad_norm": 2.7090988159179688,
      "learning_rate": 4.686398487875487e-06,
      "loss": 0.6704,
      "step": 28960
    },
    {
      "epoch": 11.86,
      "grad_norm": 2.295802116394043,
      "learning_rate": 4.686393619702616e-06,
      "loss": 0.6518,
      "step": 28970
    },
    {
      "epoch": 11.86,
      "grad_norm": 2.5318174362182617,
      "learning_rate": 4.686388740798457e-06,
      "loss": 0.6799,
      "step": 28980
    },
    {
      "epoch": 11.87,
      "grad_norm": 2.5765106678009033,
      "learning_rate": 4.686383851163034e-06,
      "loss": 0.6658,
      "step": 28990
    },
    {
      "epoch": 11.87,
      "grad_norm": 2.777561902999878,
      "learning_rate": 4.6863789507963675e-06,
      "loss": 0.676,
      "step": 29000
    },
    {
      "epoch": 11.87,
      "grad_norm": 2.0378518104553223,
      "learning_rate": 4.6863740396984796e-06,
      "loss": 0.6849,
      "step": 29010
    },
    {
      "epoch": 11.88,
      "grad_norm": 3.4186882972717285,
      "learning_rate": 4.686369117869395e-06,
      "loss": 0.6751,
      "step": 29020
    },
    {
      "epoch": 11.88,
      "grad_norm": 2.611907720565796,
      "learning_rate": 4.6863641853091346e-06,
      "loss": 0.669,
      "step": 29030
    },
    {
      "epoch": 11.89,
      "grad_norm": 3.9669578075408936,
      "learning_rate": 4.686359242017722e-06,
      "loss": 0.6661,
      "step": 29040
    },
    {
      "epoch": 11.89,
      "grad_norm": 2.270503282546997,
      "learning_rate": 4.686354287995179e-06,
      "loss": 0.6837,
      "step": 29050
    },
    {
      "epoch": 11.9,
      "grad_norm": 2.3392181396484375,
      "learning_rate": 4.686349323241528e-06,
      "loss": 0.6707,
      "step": 29060
    },
    {
      "epoch": 11.9,
      "grad_norm": 2.60707950592041,
      "learning_rate": 4.686344347756792e-06,
      "loss": 0.6747,
      "step": 29070
    },
    {
      "epoch": 11.9,
      "grad_norm": 3.896052837371826,
      "learning_rate": 4.686339361540995e-06,
      "loss": 0.658,
      "step": 29080
    },
    {
      "epoch": 11.91,
      "grad_norm": 3.361243486404419,
      "learning_rate": 4.686334364594159e-06,
      "loss": 0.6539,
      "step": 29090
    },
    {
      "epoch": 11.91,
      "grad_norm": 3.0365099906921387,
      "learning_rate": 4.686329356916306e-06,
      "loss": 0.6561,
      "step": 29100
    },
    {
      "epoch": 11.92,
      "grad_norm": 2.190608501434326,
      "learning_rate": 4.686324338507461e-06,
      "loss": 0.6764,
      "step": 29110
    },
    {
      "epoch": 11.92,
      "grad_norm": 2.2061471939086914,
      "learning_rate": 4.6863193093676445e-06,
      "loss": 0.6795,
      "step": 29120
    },
    {
      "epoch": 11.92,
      "grad_norm": 2.784367084503174,
      "learning_rate": 4.686314269496882e-06,
      "loss": 0.7027,
      "step": 29130
    },
    {
      "epoch": 11.93,
      "grad_norm": 2.3404417037963867,
      "learning_rate": 4.686309218895194e-06,
      "loss": 0.6721,
      "step": 29140
    },
    {
      "epoch": 11.93,
      "grad_norm": 3.0136823654174805,
      "learning_rate": 4.686304157562606e-06,
      "loss": 0.683,
      "step": 29150
    },
    {
      "epoch": 11.94,
      "grad_norm": 2.4996211528778076,
      "learning_rate": 4.68629908549914e-06,
      "loss": 0.6715,
      "step": 29160
    },
    {
      "epoch": 11.94,
      "grad_norm": 2.2622878551483154,
      "learning_rate": 4.686294002704819e-06,
      "loss": 0.6883,
      "step": 29170
    },
    {
      "epoch": 11.94,
      "grad_norm": 2.4224658012390137,
      "learning_rate": 4.686288909179667e-06,
      "loss": 0.6737,
      "step": 29180
    },
    {
      "epoch": 11.95,
      "grad_norm": 2.4692132472991943,
      "learning_rate": 4.686283804923707e-06,
      "loss": 0.6566,
      "step": 29190
    },
    {
      "epoch": 11.95,
      "grad_norm": 2.018043041229248,
      "learning_rate": 4.686278689936962e-06,
      "loss": 0.6745,
      "step": 29200
    },
    {
      "epoch": 11.96,
      "grad_norm": 4.171597957611084,
      "learning_rate": 4.686273564219457e-06,
      "loss": 0.6815,
      "step": 29210
    },
    {
      "epoch": 11.96,
      "grad_norm": 2.440502643585205,
      "learning_rate": 4.686268427771214e-06,
      "loss": 0.6863,
      "step": 29220
    },
    {
      "epoch": 11.96,
      "grad_norm": 2.644491672515869,
      "learning_rate": 4.686263280592256e-06,
      "loss": 0.6705,
      "step": 29230
    },
    {
      "epoch": 11.97,
      "grad_norm": 2.395806074142456,
      "learning_rate": 4.686258122682608e-06,
      "loss": 0.6784,
      "step": 29240
    },
    {
      "epoch": 11.97,
      "grad_norm": 2.204671859741211,
      "learning_rate": 4.686252954042293e-06,
      "loss": 0.6747,
      "step": 29250
    },
    {
      "epoch": 11.98,
      "grad_norm": 2.8391714096069336,
      "learning_rate": 4.686247774671335e-06,
      "loss": 0.6685,
      "step": 29260
    },
    {
      "epoch": 11.98,
      "grad_norm": 3.313053607940674,
      "learning_rate": 4.686242584569757e-06,
      "loss": 0.6725,
      "step": 29270
    },
    {
      "epoch": 11.99,
      "grad_norm": 2.0325779914855957,
      "learning_rate": 4.6862373837375835e-06,
      "loss": 0.6614,
      "step": 29280
    },
    {
      "epoch": 11.99,
      "grad_norm": 2.627485990524292,
      "learning_rate": 4.686232172174838e-06,
      "loss": 0.6282,
      "step": 29290
    },
    {
      "epoch": 11.99,
      "grad_norm": 2.3500289916992188,
      "learning_rate": 4.686226949881545e-06,
      "loss": 0.6851,
      "step": 29300
    },
    {
      "epoch": 12.0,
      "grad_norm": 2.264674186706543,
      "learning_rate": 4.6862217168577264e-06,
      "loss": 0.6991,
      "step": 29310
    },
    {
      "epoch": 12.0,
      "eval_loss": 0.6711130142211914,
      "eval_runtime": 52.7448,
      "eval_samples_per_second": 65.39,
      "eval_steps_per_second": 8.19,
      "step": 29316
    },
    {
      "epoch": 12.0,
      "grad_norm": 2.645700693130493,
      "learning_rate": 4.686216473103409e-06,
      "loss": 0.6832,
      "step": 29320
    },
    {
      "epoch": 12.01,
      "grad_norm": 3.2575230598449707,
      "learning_rate": 4.686211218618615e-06,
      "loss": 0.6504,
      "step": 29330
    },
    {
      "epoch": 12.01,
      "grad_norm": 2.9525673389434814,
      "learning_rate": 4.686205953403369e-06,
      "loss": 0.706,
      "step": 29340
    },
    {
      "epoch": 12.01,
      "grad_norm": 2.694793939590454,
      "learning_rate": 4.686200677457696e-06,
      "loss": 0.6811,
      "step": 29350
    },
    {
      "epoch": 12.02,
      "grad_norm": 2.271235704421997,
      "learning_rate": 4.686195390781618e-06,
      "loss": 0.6729,
      "step": 29360
    },
    {
      "epoch": 12.02,
      "grad_norm": 1.8445475101470947,
      "learning_rate": 4.686190093375161e-06,
      "loss": 0.6895,
      "step": 29370
    },
    {
      "epoch": 12.03,
      "grad_norm": 3.6876072883605957,
      "learning_rate": 4.686184785238349e-06,
      "loss": 0.6605,
      "step": 29380
    },
    {
      "epoch": 12.03,
      "grad_norm": 2.9007620811462402,
      "learning_rate": 4.6861794663712045e-06,
      "loss": 0.6861,
      "step": 29390
    },
    {
      "epoch": 12.03,
      "grad_norm": 2.5568716526031494,
      "learning_rate": 4.686174136773755e-06,
      "loss": 0.6915,
      "step": 29400
    },
    {
      "epoch": 12.04,
      "grad_norm": 2.6606388092041016,
      "learning_rate": 4.686168796446022e-06,
      "loss": 0.6701,
      "step": 29410
    },
    {
      "epoch": 12.04,
      "grad_norm": 2.3103907108306885,
      "learning_rate": 4.686163445388032e-06,
      "loss": 0.671,
      "step": 29420
    },
    {
      "epoch": 12.05,
      "grad_norm": 2.7277610301971436,
      "learning_rate": 4.686158083599808e-06,
      "loss": 0.681,
      "step": 29430
    },
    {
      "epoch": 12.05,
      "grad_norm": 2.2430458068847656,
      "learning_rate": 4.686152711081377e-06,
      "loss": 0.6889,
      "step": 29440
    },
    {
      "epoch": 12.05,
      "grad_norm": 2.918774366378784,
      "learning_rate": 4.686147327832761e-06,
      "loss": 0.6739,
      "step": 29450
    },
    {
      "epoch": 12.06,
      "grad_norm": 2.5713071823120117,
      "learning_rate": 4.686141933853986e-06,
      "loss": 0.6932,
      "step": 29460
    },
    {
      "epoch": 12.06,
      "grad_norm": 2.7357006072998047,
      "learning_rate": 4.686136529145075e-06,
      "loss": 0.6769,
      "step": 29470
    },
    {
      "epoch": 12.07,
      "grad_norm": 2.215442657470703,
      "learning_rate": 4.686131113706056e-06,
      "loss": 0.685,
      "step": 29480
    },
    {
      "epoch": 12.07,
      "grad_norm": 3.027937412261963,
      "learning_rate": 4.68612568753695e-06,
      "loss": 0.6763,
      "step": 29490
    },
    {
      "epoch": 12.08,
      "grad_norm": 3.1777503490448,
      "learning_rate": 4.686120250637785e-06,
      "loss": 0.6634,
      "step": 29500
    },
    {
      "epoch": 12.08,
      "grad_norm": 1.93949294090271,
      "learning_rate": 4.686114803008584e-06,
      "loss": 0.6766,
      "step": 29510
    },
    {
      "epoch": 12.08,
      "grad_norm": 2.587702512741089,
      "learning_rate": 4.6861093446493735e-06,
      "loss": 0.6613,
      "step": 29520
    },
    {
      "epoch": 12.09,
      "grad_norm": 2.1227972507476807,
      "learning_rate": 4.6861038755601764e-06,
      "loss": 0.6723,
      "step": 29530
    },
    {
      "epoch": 12.09,
      "grad_norm": 3.390068769454956,
      "learning_rate": 4.68609839574102e-06,
      "loss": 0.6679,
      "step": 29540
    },
    {
      "epoch": 12.1,
      "grad_norm": 3.3965823650360107,
      "learning_rate": 4.686092905191927e-06,
      "loss": 0.6684,
      "step": 29550
    },
    {
      "epoch": 12.1,
      "grad_norm": 2.3959786891937256,
      "learning_rate": 4.686087403912925e-06,
      "loss": 0.6578,
      "step": 29560
    },
    {
      "epoch": 12.1,
      "grad_norm": 3.4446280002593994,
      "learning_rate": 4.686081891904039e-06,
      "loss": 0.6725,
      "step": 29570
    },
    {
      "epoch": 12.11,
      "grad_norm": 2.8014285564422607,
      "learning_rate": 4.686076369165291e-06,
      "loss": 0.6443,
      "step": 29580
    },
    {
      "epoch": 12.11,
      "grad_norm": 3.3196187019348145,
      "learning_rate": 4.686070835696711e-06,
      "loss": 0.6787,
      "step": 29590
    },
    {
      "epoch": 12.12,
      "grad_norm": 2.926875352859497,
      "learning_rate": 4.686065291498321e-06,
      "loss": 0.6671,
      "step": 29600
    },
    {
      "epoch": 12.12,
      "grad_norm": 3.1989941596984863,
      "learning_rate": 4.686059736570147e-06,
      "loss": 0.6757,
      "step": 29610
    },
    {
      "epoch": 12.12,
      "grad_norm": 2.4917449951171875,
      "learning_rate": 4.686054170912214e-06,
      "loss": 0.6686,
      "step": 29620
    },
    {
      "epoch": 12.13,
      "grad_norm": 2.7208433151245117,
      "learning_rate": 4.68604859452455e-06,
      "loss": 0.6618,
      "step": 29630
    },
    {
      "epoch": 12.13,
      "grad_norm": 2.686689853668213,
      "learning_rate": 4.6860430074071784e-06,
      "loss": 0.6745,
      "step": 29640
    },
    {
      "epoch": 12.14,
      "grad_norm": 3.301839828491211,
      "learning_rate": 4.686037409560125e-06,
      "loss": 0.6697,
      "step": 29650
    },
    {
      "epoch": 12.14,
      "grad_norm": 2.343724489212036,
      "learning_rate": 4.686031800983417e-06,
      "loss": 0.6617,
      "step": 29660
    },
    {
      "epoch": 12.14,
      "grad_norm": 2.7055211067199707,
      "learning_rate": 4.686026181677077e-06,
      "loss": 0.6871,
      "step": 29670
    },
    {
      "epoch": 12.15,
      "grad_norm": 3.3805010318756104,
      "learning_rate": 4.686020551641134e-06,
      "loss": 0.7039,
      "step": 29680
    },
    {
      "epoch": 12.15,
      "grad_norm": 1.9929989576339722,
      "learning_rate": 4.686014910875612e-06,
      "loss": 0.6659,
      "step": 29690
    },
    {
      "epoch": 12.16,
      "grad_norm": 2.472377061843872,
      "learning_rate": 4.686009259380537e-06,
      "loss": 0.6523,
      "step": 29700
    },
    {
      "epoch": 12.16,
      "grad_norm": 1.822196364402771,
      "learning_rate": 4.6860035971559345e-06,
      "loss": 0.6757,
      "step": 29710
    },
    {
      "epoch": 12.17,
      "grad_norm": 1.8304330110549927,
      "learning_rate": 4.685997924201831e-06,
      "loss": 0.6628,
      "step": 29720
    },
    {
      "epoch": 12.17,
      "grad_norm": 1.8061050176620483,
      "learning_rate": 4.685992240518253e-06,
      "loss": 0.6787,
      "step": 29730
    },
    {
      "epoch": 12.17,
      "grad_norm": 2.6812679767608643,
      "learning_rate": 4.685986546105227e-06,
      "loss": 0.6633,
      "step": 29740
    },
    {
      "epoch": 12.18,
      "grad_norm": 2.6815764904022217,
      "learning_rate": 4.685980840962777e-06,
      "loss": 0.6846,
      "step": 29750
    },
    {
      "epoch": 12.18,
      "grad_norm": 3.197829484939575,
      "learning_rate": 4.68597512509093e-06,
      "loss": 0.6883,
      "step": 29760
    },
    {
      "epoch": 12.19,
      "grad_norm": 2.3871593475341797,
      "learning_rate": 4.685969398489713e-06,
      "loss": 0.6585,
      "step": 29770
    },
    {
      "epoch": 12.19,
      "grad_norm": 1.5592076778411865,
      "learning_rate": 4.6859636611591515e-06,
      "loss": 0.6645,
      "step": 29780
    },
    {
      "epoch": 12.19,
      "grad_norm": 2.1190459728240967,
      "learning_rate": 4.685957913099272e-06,
      "loss": 0.6813,
      "step": 29790
    },
    {
      "epoch": 12.2,
      "grad_norm": 2.296759843826294,
      "learning_rate": 4.685952154310101e-06,
      "loss": 0.6562,
      "step": 29800
    },
    {
      "epoch": 12.2,
      "grad_norm": 2.7196226119995117,
      "learning_rate": 4.685946384791664e-06,
      "loss": 0.6899,
      "step": 29810
    },
    {
      "epoch": 12.21,
      "grad_norm": 2.542187452316284,
      "learning_rate": 4.685940604543988e-06,
      "loss": 0.654,
      "step": 29820
    },
    {
      "epoch": 12.21,
      "grad_norm": 2.125054121017456,
      "learning_rate": 4.685934813567101e-06,
      "loss": 0.6616,
      "step": 29830
    },
    {
      "epoch": 12.21,
      "grad_norm": 2.255758047103882,
      "learning_rate": 4.6859290118610266e-06,
      "loss": 0.6688,
      "step": 29840
    },
    {
      "epoch": 12.22,
      "grad_norm": 3.462245464324951,
      "learning_rate": 4.685923199425794e-06,
      "loss": 0.6712,
      "step": 29850
    },
    {
      "epoch": 12.22,
      "grad_norm": 2.2858755588531494,
      "learning_rate": 4.685917376261428e-06,
      "loss": 0.6628,
      "step": 29860
    },
    {
      "epoch": 12.23,
      "grad_norm": 2.561598300933838,
      "learning_rate": 4.685911542367956e-06,
      "loss": 0.6453,
      "step": 29870
    },
    {
      "epoch": 12.23,
      "grad_norm": 2.2986137866973877,
      "learning_rate": 4.685905697745405e-06,
      "loss": 0.6717,
      "step": 29880
    },
    {
      "epoch": 12.23,
      "grad_norm": 3.1693313121795654,
      "learning_rate": 4.685899842393802e-06,
      "loss": 0.6784,
      "step": 29890
    },
    {
      "epoch": 12.24,
      "grad_norm": 2.466346263885498,
      "learning_rate": 4.685893976313172e-06,
      "loss": 0.6835,
      "step": 29900
    },
    {
      "epoch": 12.24,
      "grad_norm": 2.106217384338379,
      "learning_rate": 4.685888099503544e-06,
      "loss": 0.6665,
      "step": 29910
    },
    {
      "epoch": 12.25,
      "grad_norm": 2.5877163410186768,
      "learning_rate": 4.6858822119649444e-06,
      "loss": 0.6733,
      "step": 29920
    },
    {
      "epoch": 12.25,
      "grad_norm": 2.4662435054779053,
      "learning_rate": 4.685876313697399e-06,
      "loss": 0.6567,
      "step": 29930
    },
    {
      "epoch": 12.26,
      "grad_norm": 3.3671975135803223,
      "learning_rate": 4.685870404700936e-06,
      "loss": 0.6648,
      "step": 29940
    },
    {
      "epoch": 12.26,
      "grad_norm": 1.738213062286377,
      "learning_rate": 4.685864484975583e-06,
      "loss": 0.6838,
      "step": 29950
    },
    {
      "epoch": 12.26,
      "grad_norm": 2.429872751235962,
      "learning_rate": 4.6858585545213646e-06,
      "loss": 0.6648,
      "step": 29960
    },
    {
      "epoch": 12.27,
      "grad_norm": 1.795306921005249,
      "learning_rate": 4.685852613338311e-06,
      "loss": 0.6781,
      "step": 29970
    },
    {
      "epoch": 12.27,
      "grad_norm": 2.5622870922088623,
      "learning_rate": 4.685846661426447e-06,
      "loss": 0.6601,
      "step": 29980
    },
    {
      "epoch": 12.28,
      "grad_norm": 4.642294883728027,
      "learning_rate": 4.685840698785802e-06,
      "loss": 0.6796,
      "step": 29990
    },
    {
      "epoch": 12.28,
      "grad_norm": 2.8618056774139404,
      "learning_rate": 4.685834725416401e-06,
      "loss": 0.6828,
      "step": 30000
    },
    {
      "epoch": 12.28,
      "grad_norm": 1.742167353630066,
      "learning_rate": 4.6858287413182735e-06,
      "loss": 0.6675,
      "step": 30010
    },
    {
      "epoch": 12.29,
      "grad_norm": 2.366070032119751,
      "learning_rate": 4.685822746491446e-06,
      "loss": 0.6504,
      "step": 30020
    },
    {
      "epoch": 12.29,
      "grad_norm": 2.9276463985443115,
      "learning_rate": 4.685816740935946e-06,
      "loss": 0.6912,
      "step": 30030
    },
    {
      "epoch": 12.3,
      "grad_norm": 2.652231216430664,
      "learning_rate": 4.6858107246518e-06,
      "loss": 0.6808,
      "step": 30040
    },
    {
      "epoch": 12.3,
      "grad_norm": 2.616842746734619,
      "learning_rate": 4.685804697639038e-06,
      "loss": 0.6572,
      "step": 30050
    },
    {
      "epoch": 12.3,
      "grad_norm": 2.168264865875244,
      "learning_rate": 4.685798659897685e-06,
      "loss": 0.6774,
      "step": 30060
    },
    {
      "epoch": 12.31,
      "grad_norm": 2.050017833709717,
      "learning_rate": 4.68579261142777e-06,
      "loss": 0.6709,
      "step": 30070
    },
    {
      "epoch": 12.31,
      "grad_norm": 2.0432608127593994,
      "learning_rate": 4.685786552229321e-06,
      "loss": 0.6715,
      "step": 30080
    },
    {
      "epoch": 12.32,
      "grad_norm": 2.5490713119506836,
      "learning_rate": 4.685780482302365e-06,
      "loss": 0.6572,
      "step": 30090
    },
    {
      "epoch": 12.32,
      "grad_norm": 2.1273443698883057,
      "learning_rate": 4.685774401646931e-06,
      "loss": 0.6583,
      "step": 30100
    },
    {
      "epoch": 12.33,
      "grad_norm": 2.552093744277954,
      "learning_rate": 4.685768310263044e-06,
      "loss": 0.6727,
      "step": 30110
    },
    {
      "epoch": 12.33,
      "grad_norm": 2.4735212326049805,
      "learning_rate": 4.685762208150735e-06,
      "loss": 0.6577,
      "step": 30120
    },
    {
      "epoch": 12.33,
      "grad_norm": 2.567812204360962,
      "learning_rate": 4.685756095310031e-06,
      "loss": 0.6778,
      "step": 30130
    },
    {
      "epoch": 12.34,
      "grad_norm": 3.117168426513672,
      "learning_rate": 4.685749971740959e-06,
      "loss": 0.6425,
      "step": 30140
    },
    {
      "epoch": 12.34,
      "grad_norm": 1.600926160812378,
      "learning_rate": 4.685743837443548e-06,
      "loss": 0.6588,
      "step": 30150
    },
    {
      "epoch": 12.35,
      "grad_norm": 2.811508893966675,
      "learning_rate": 4.685737692417827e-06,
      "loss": 0.6737,
      "step": 30160
    },
    {
      "epoch": 12.35,
      "grad_norm": 3.0757999420166016,
      "learning_rate": 4.685731536663822e-06,
      "loss": 0.6799,
      "step": 30170
    },
    {
      "epoch": 12.35,
      "grad_norm": 2.141832113265991,
      "learning_rate": 4.6857253701815626e-06,
      "loss": 0.6728,
      "step": 30180
    },
    {
      "epoch": 12.36,
      "grad_norm": 3.855226516723633,
      "learning_rate": 4.685719192971077e-06,
      "loss": 0.6784,
      "step": 30190
    },
    {
      "epoch": 12.36,
      "grad_norm": 2.7736051082611084,
      "learning_rate": 4.685713005032393e-06,
      "loss": 0.6639,
      "step": 30200
    },
    {
      "epoch": 12.37,
      "grad_norm": 3.8796744346618652,
      "learning_rate": 4.685706806365538e-06,
      "loss": 0.6679,
      "step": 30210
    },
    {
      "epoch": 12.37,
      "grad_norm": 3.192965030670166,
      "learning_rate": 4.685700596970543e-06,
      "loss": 0.6713,
      "step": 30220
    },
    {
      "epoch": 12.37,
      "grad_norm": 2.1753153800964355,
      "learning_rate": 4.685694376847434e-06,
      "loss": 0.6625,
      "step": 30230
    },
    {
      "epoch": 12.38,
      "grad_norm": 4.035234451293945,
      "learning_rate": 4.685688145996242e-06,
      "loss": 0.6819,
      "step": 30240
    },
    {
      "epoch": 12.38,
      "grad_norm": 2.0782392024993896,
      "learning_rate": 4.685681904416993e-06,
      "loss": 0.656,
      "step": 30250
    },
    {
      "epoch": 12.39,
      "grad_norm": 2.436244249343872,
      "learning_rate": 4.685675652109717e-06,
      "loss": 0.6784,
      "step": 30260
    },
    {
      "epoch": 12.39,
      "grad_norm": 1.8709379434585571,
      "learning_rate": 4.685669389074442e-06,
      "loss": 0.6812,
      "step": 30270
    },
    {
      "epoch": 12.39,
      "grad_norm": 3.8939177989959717,
      "learning_rate": 4.6856631153111965e-06,
      "loss": 0.674,
      "step": 30280
    },
    {
      "epoch": 12.4,
      "grad_norm": 2.3485260009765625,
      "learning_rate": 4.685656830820011e-06,
      "loss": 0.6534,
      "step": 30290
    },
    {
      "epoch": 12.4,
      "grad_norm": 3.0558435916900635,
      "learning_rate": 4.685650535600912e-06,
      "loss": 0.665,
      "step": 30300
    },
    {
      "epoch": 12.41,
      "grad_norm": 2.1952695846557617,
      "learning_rate": 4.68564422965393e-06,
      "loss": 0.6751,
      "step": 30310
    },
    {
      "epoch": 12.41,
      "grad_norm": 2.5254087448120117,
      "learning_rate": 4.685637912979092e-06,
      "loss": 0.6895,
      "step": 30320
    },
    {
      "epoch": 12.42,
      "grad_norm": 2.498849630355835,
      "learning_rate": 4.685631585576429e-06,
      "loss": 0.6774,
      "step": 30330
    },
    {
      "epoch": 12.42,
      "grad_norm": 2.032374858856201,
      "learning_rate": 4.685625247445969e-06,
      "loss": 0.667,
      "step": 30340
    },
    {
      "epoch": 12.42,
      "grad_norm": 2.0213916301727295,
      "learning_rate": 4.685618898587741e-06,
      "loss": 0.673,
      "step": 30350
    },
    {
      "epoch": 12.43,
      "grad_norm": 3.01774525642395,
      "learning_rate": 4.685612539001775e-06,
      "loss": 0.6541,
      "step": 30360
    },
    {
      "epoch": 12.43,
      "grad_norm": 3.2606353759765625,
      "learning_rate": 4.685606168688099e-06,
      "loss": 0.6591,
      "step": 30370
    },
    {
      "epoch": 12.44,
      "grad_norm": 2.492962598800659,
      "learning_rate": 4.6855997876467425e-06,
      "loss": 0.678,
      "step": 30380
    },
    {
      "epoch": 12.44,
      "grad_norm": 1.9821053743362427,
      "learning_rate": 4.685593395877735e-06,
      "loss": 0.6848,
      "step": 30390
    },
    {
      "epoch": 12.44,
      "grad_norm": 2.6097476482391357,
      "learning_rate": 4.685586993381106e-06,
      "loss": 0.6816,
      "step": 30400
    },
    {
      "epoch": 12.45,
      "grad_norm": 2.7064058780670166,
      "learning_rate": 4.685580580156883e-06,
      "loss": 0.6811,
      "step": 30410
    },
    {
      "epoch": 12.45,
      "grad_norm": 2.5484910011291504,
      "learning_rate": 4.6855741562050975e-06,
      "loss": 0.6491,
      "step": 30420
    },
    {
      "epoch": 12.46,
      "grad_norm": 2.279108762741089,
      "learning_rate": 4.6855677215257785e-06,
      "loss": 0.676,
      "step": 30430
    },
    {
      "epoch": 12.46,
      "grad_norm": 2.039202928543091,
      "learning_rate": 4.685561276118955e-06,
      "loss": 0.6575,
      "step": 30440
    },
    {
      "epoch": 12.46,
      "grad_norm": 2.3328628540039062,
      "learning_rate": 4.685554819984657e-06,
      "loss": 0.675,
      "step": 30450
    },
    {
      "epoch": 12.47,
      "grad_norm": 2.0112931728363037,
      "learning_rate": 4.685548353122914e-06,
      "loss": 0.6831,
      "step": 30460
    },
    {
      "epoch": 12.47,
      "grad_norm": 3.2488648891448975,
      "learning_rate": 4.685541875533754e-06,
      "loss": 0.669,
      "step": 30470
    },
    {
      "epoch": 12.48,
      "grad_norm": 3.475508689880371,
      "learning_rate": 4.68553538721721e-06,
      "loss": 0.6866,
      "step": 30480
    },
    {
      "epoch": 12.48,
      "grad_norm": 2.6662261486053467,
      "learning_rate": 4.685528888173309e-06,
      "loss": 0.6679,
      "step": 30490
    },
    {
      "epoch": 12.48,
      "grad_norm": 2.7923340797424316,
      "learning_rate": 4.6855223784020805e-06,
      "loss": 0.6637,
      "step": 30500
    },
    {
      "epoch": 12.49,
      "grad_norm": 2.0264642238616943,
      "learning_rate": 4.685515857903557e-06,
      "loss": 0.6467,
      "step": 30510
    },
    {
      "epoch": 12.49,
      "grad_norm": 2.6365456581115723,
      "learning_rate": 4.685509326677766e-06,
      "loss": 0.6754,
      "step": 30520
    },
    {
      "epoch": 12.5,
      "grad_norm": 3.2759852409362793,
      "learning_rate": 4.6855027847247386e-06,
      "loss": 0.6719,
      "step": 30530
    },
    {
      "epoch": 12.5,
      "grad_norm": 2.1001954078674316,
      "learning_rate": 4.685496232044503e-06,
      "loss": 0.6466,
      "step": 30540
    },
    {
      "epoch": 12.51,
      "grad_norm": 2.773414373397827,
      "learning_rate": 4.685489668637092e-06,
      "loss": 0.6481,
      "step": 30550
    },
    {
      "epoch": 12.51,
      "grad_norm": 3.472073793411255,
      "learning_rate": 4.685483094502535e-06,
      "loss": 0.6568,
      "step": 30560
    },
    {
      "epoch": 12.51,
      "grad_norm": 2.5427019596099854,
      "learning_rate": 4.68547650964086e-06,
      "loss": 0.6865,
      "step": 30570
    },
    {
      "epoch": 12.52,
      "grad_norm": 2.993234157562256,
      "learning_rate": 4.685469914052099e-06,
      "loss": 0.682,
      "step": 30580
    },
    {
      "epoch": 12.52,
      "grad_norm": 3.159841537475586,
      "learning_rate": 4.685463307736281e-06,
      "loss": 0.6473,
      "step": 30590
    },
    {
      "epoch": 12.53,
      "grad_norm": 1.8131588697433472,
      "learning_rate": 4.685456690693437e-06,
      "loss": 0.6679,
      "step": 30600
    },
    {
      "epoch": 12.53,
      "grad_norm": 2.869661569595337,
      "learning_rate": 4.685450062923598e-06,
      "loss": 0.6584,
      "step": 30610
    },
    {
      "epoch": 12.53,
      "grad_norm": 2.478978157043457,
      "learning_rate": 4.685443424426794e-06,
      "loss": 0.6695,
      "step": 30620
    },
    {
      "epoch": 12.54,
      "grad_norm": 2.338756799697876,
      "learning_rate": 4.685436775203054e-06,
      "loss": 0.654,
      "step": 30630
    },
    {
      "epoch": 12.54,
      "grad_norm": 3.131338119506836,
      "learning_rate": 4.68543011525241e-06,
      "loss": 0.6509,
      "step": 30640
    },
    {
      "epoch": 12.55,
      "grad_norm": 4.07857608795166,
      "learning_rate": 4.685423444574892e-06,
      "loss": 0.6759,
      "step": 30650
    },
    {
      "epoch": 12.55,
      "grad_norm": 2.6658935546875,
      "learning_rate": 4.6854167631705314e-06,
      "loss": 0.6591,
      "step": 30660
    },
    {
      "epoch": 12.55,
      "grad_norm": 2.756594181060791,
      "learning_rate": 4.685410071039357e-06,
      "loss": 0.6734,
      "step": 30670
    },
    {
      "epoch": 12.56,
      "grad_norm": 3.7178871631622314,
      "learning_rate": 4.685403368181401e-06,
      "loss": 0.6526,
      "step": 30680
    },
    {
      "epoch": 12.56,
      "grad_norm": 2.225250005722046,
      "learning_rate": 4.685396654596693e-06,
      "loss": 0.6537,
      "step": 30690
    },
    {
      "epoch": 12.57,
      "grad_norm": 2.051584005355835,
      "learning_rate": 4.685389930285265e-06,
      "loss": 0.6698,
      "step": 30700
    },
    {
      "epoch": 12.57,
      "grad_norm": 2.8568005561828613,
      "learning_rate": 4.685383195247147e-06,
      "loss": 0.6555,
      "step": 30710
    },
    {
      "epoch": 12.57,
      "grad_norm": 2.720226764678955,
      "learning_rate": 4.68537644948237e-06,
      "loss": 0.6516,
      "step": 30720
    },
    {
      "epoch": 12.58,
      "grad_norm": 3.011594533920288,
      "learning_rate": 4.685369692990965e-06,
      "loss": 0.6709,
      "step": 30730
    },
    {
      "epoch": 12.58,
      "grad_norm": 2.333681106567383,
      "learning_rate": 4.685362925772963e-06,
      "loss": 0.6481,
      "step": 30740
    },
    {
      "epoch": 12.59,
      "grad_norm": 2.0420689582824707,
      "learning_rate": 4.6853561478283935e-06,
      "loss": 0.6651,
      "step": 30750
    },
    {
      "epoch": 12.59,
      "grad_norm": 2.579300880432129,
      "learning_rate": 4.68534935915729e-06,
      "loss": 0.6581,
      "step": 30760
    },
    {
      "epoch": 12.6,
      "grad_norm": 2.3170018196105957,
      "learning_rate": 4.6853425597596824e-06,
      "loss": 0.672,
      "step": 30770
    },
    {
      "epoch": 12.6,
      "grad_norm": 2.7877206802368164,
      "learning_rate": 4.6853357496356015e-06,
      "loss": 0.6755,
      "step": 30780
    },
    {
      "epoch": 12.6,
      "grad_norm": 2.0693514347076416,
      "learning_rate": 4.685328928785079e-06,
      "loss": 0.6623,
      "step": 30790
    },
    {
      "epoch": 12.61,
      "grad_norm": 2.9328532218933105,
      "learning_rate": 4.685322097208146e-06,
      "loss": 0.677,
      "step": 30800
    },
    {
      "epoch": 12.61,
      "grad_norm": 2.524653196334839,
      "learning_rate": 4.685315254904834e-06,
      "loss": 0.6819,
      "step": 30810
    },
    {
      "epoch": 12.62,
      "grad_norm": 2.0958242416381836,
      "learning_rate": 4.685308401875175e-06,
      "loss": 0.6607,
      "step": 30820
    },
    {
      "epoch": 12.62,
      "grad_norm": 1.9483205080032349,
      "learning_rate": 4.685301538119199e-06,
      "loss": 0.6797,
      "step": 30830
    },
    {
      "epoch": 12.62,
      "grad_norm": 1.935376524925232,
      "learning_rate": 4.685294663636937e-06,
      "loss": 0.6542,
      "step": 30840
    },
    {
      "epoch": 12.63,
      "grad_norm": 2.3340108394622803,
      "learning_rate": 4.685287778428423e-06,
      "loss": 0.6539,
      "step": 30850
    },
    {
      "epoch": 12.63,
      "grad_norm": 2.6455063819885254,
      "learning_rate": 4.685280882493687e-06,
      "loss": 0.6886,
      "step": 30860
    },
    {
      "epoch": 12.64,
      "grad_norm": 2.3321034908294678,
      "learning_rate": 4.68527397583276e-06,
      "loss": 0.6556,
      "step": 30870
    },
    {
      "epoch": 12.64,
      "grad_norm": 3.0671815872192383,
      "learning_rate": 4.6852670584456744e-06,
      "loss": 0.6733,
      "step": 30880
    },
    {
      "epoch": 12.64,
      "grad_norm": 2.9470293521881104,
      "learning_rate": 4.685260130332462e-06,
      "loss": 0.6669,
      "step": 30890
    },
    {
      "epoch": 12.65,
      "grad_norm": 2.7951252460479736,
      "learning_rate": 4.6852531914931544e-06,
      "loss": 0.6697,
      "step": 30900
    },
    {
      "epoch": 12.65,
      "grad_norm": 3.0360143184661865,
      "learning_rate": 4.685246241927782e-06,
      "loss": 0.6857,
      "step": 30910
    },
    {
      "epoch": 12.66,
      "grad_norm": 2.7035746574401855,
      "learning_rate": 4.68523928163638e-06,
      "loss": 0.6517,
      "step": 30920
    },
    {
      "epoch": 12.66,
      "grad_norm": 2.1031596660614014,
      "learning_rate": 4.685232310618977e-06,
      "loss": 0.6392,
      "step": 30930
    },
    {
      "epoch": 12.66,
      "grad_norm": 2.2874274253845215,
      "learning_rate": 4.685225328875607e-06,
      "loss": 0.6649,
      "step": 30940
    },
    {
      "epoch": 12.67,
      "grad_norm": 2.5761284828186035,
      "learning_rate": 4.685218336406301e-06,
      "loss": 0.6584,
      "step": 30950
    },
    {
      "epoch": 12.67,
      "grad_norm": 2.4932453632354736,
      "learning_rate": 4.68521133321109e-06,
      "loss": 0.6601,
      "step": 30960
    },
    {
      "epoch": 12.68,
      "grad_norm": 2.7040343284606934,
      "learning_rate": 4.685204319290008e-06,
      "loss": 0.6526,
      "step": 30970
    },
    {
      "epoch": 12.68,
      "grad_norm": 2.7153708934783936,
      "learning_rate": 4.685197294643086e-06,
      "loss": 0.6459,
      "step": 30980
    },
    {
      "epoch": 12.69,
      "grad_norm": 2.4150333404541016,
      "learning_rate": 4.685190259270357e-06,
      "loss": 0.6723,
      "step": 30990
    },
    {
      "epoch": 12.69,
      "grad_norm": 2.4496138095855713,
      "learning_rate": 4.685183213171853e-06,
      "loss": 0.6587,
      "step": 31000
    },
    {
      "epoch": 12.69,
      "grad_norm": 2.6260106563568115,
      "learning_rate": 4.685176156347605e-06,
      "loss": 0.6557,
      "step": 31010
    },
    {
      "epoch": 12.7,
      "grad_norm": 2.9827115535736084,
      "learning_rate": 4.685169088797648e-06,
      "loss": 0.6628,
      "step": 31020
    },
    {
      "epoch": 12.7,
      "grad_norm": 2.2384583950042725,
      "learning_rate": 4.6851620105220125e-06,
      "loss": 0.6408,
      "step": 31030
    },
    {
      "epoch": 12.71,
      "grad_norm": 3.0177807807922363,
      "learning_rate": 4.68515492152073e-06,
      "loss": 0.6528,
      "step": 31040
    },
    {
      "epoch": 12.71,
      "grad_norm": 2.219064712524414,
      "learning_rate": 4.685147821793835e-06,
      "loss": 0.6586,
      "step": 31050
    },
    {
      "epoch": 12.71,
      "grad_norm": 2.065432548522949,
      "learning_rate": 4.685140711341359e-06,
      "loss": 0.667,
      "step": 31060
    },
    {
      "epoch": 12.72,
      "grad_norm": 2.4061458110809326,
      "learning_rate": 4.685133590163335e-06,
      "loss": 0.6606,
      "step": 31070
    },
    {
      "epoch": 12.72,
      "grad_norm": 2.3313381671905518,
      "learning_rate": 4.685126458259795e-06,
      "loss": 0.6698,
      "step": 31080
    },
    {
      "epoch": 12.73,
      "grad_norm": 2.433762311935425,
      "learning_rate": 4.685119315630772e-06,
      "loss": 0.6635,
      "step": 31090
    },
    {
      "epoch": 12.73,
      "grad_norm": 2.440617561340332,
      "learning_rate": 4.685112162276299e-06,
      "loss": 0.6743,
      "step": 31100
    },
    {
      "epoch": 12.73,
      "grad_norm": 3.106081962585449,
      "learning_rate": 4.685104998196409e-06,
      "loss": 0.6533,
      "step": 31110
    },
    {
      "epoch": 12.74,
      "grad_norm": 2.6843535900115967,
      "learning_rate": 4.685097823391134e-06,
      "loss": 0.6785,
      "step": 31120
    },
    {
      "epoch": 12.74,
      "grad_norm": 2.2845876216888428,
      "learning_rate": 4.685090637860507e-06,
      "loss": 0.663,
      "step": 31130
    },
    {
      "epoch": 12.75,
      "grad_norm": 2.0385916233062744,
      "learning_rate": 4.685083441604561e-06,
      "loss": 0.6763,
      "step": 31140
    },
    {
      "epoch": 12.75,
      "grad_norm": 2.1687653064727783,
      "learning_rate": 4.685076234623329e-06,
      "loss": 0.6792,
      "step": 31150
    },
    {
      "epoch": 12.75,
      "grad_norm": 1.8994872570037842,
      "learning_rate": 4.685069016916845e-06,
      "loss": 0.6751,
      "step": 31160
    },
    {
      "epoch": 12.76,
      "grad_norm": 2.2031655311584473,
      "learning_rate": 4.685061788485141e-06,
      "loss": 0.6604,
      "step": 31170
    },
    {
      "epoch": 12.76,
      "grad_norm": 2.6642348766326904,
      "learning_rate": 4.6850545493282494e-06,
      "loss": 0.6454,
      "step": 31180
    },
    {
      "epoch": 12.77,
      "grad_norm": 2.50234055519104,
      "learning_rate": 4.685047299446205e-06,
      "loss": 0.6518,
      "step": 31190
    },
    {
      "epoch": 12.77,
      "grad_norm": 2.382669687271118,
      "learning_rate": 4.68504003883904e-06,
      "loss": 0.6397,
      "step": 31200
    },
    {
      "epoch": 12.78,
      "grad_norm": 2.4062693119049072,
      "learning_rate": 4.685032767506788e-06,
      "loss": 0.6653,
      "step": 31210
    },
    {
      "epoch": 12.78,
      "grad_norm": 2.7892329692840576,
      "learning_rate": 4.685025485449482e-06,
      "loss": 0.6494,
      "step": 31220
    },
    {
      "epoch": 12.78,
      "grad_norm": 3.0177175998687744,
      "learning_rate": 4.685018192667156e-06,
      "loss": 0.6804,
      "step": 31230
    },
    {
      "epoch": 12.79,
      "grad_norm": 2.106290817260742,
      "learning_rate": 4.685010889159843e-06,
      "loss": 0.6335,
      "step": 31240
    },
    {
      "epoch": 12.79,
      "grad_norm": 2.673464298248291,
      "learning_rate": 4.685003574927576e-06,
      "loss": 0.6593,
      "step": 31250
    },
    {
      "epoch": 12.8,
      "grad_norm": 3.1878628730773926,
      "learning_rate": 4.684996249970389e-06,
      "loss": 0.66,
      "step": 31260
    },
    {
      "epoch": 12.8,
      "grad_norm": 2.5250344276428223,
      "learning_rate": 4.684988914288317e-06,
      "loss": 0.6808,
      "step": 31270
    },
    {
      "epoch": 12.8,
      "grad_norm": 2.8931291103363037,
      "learning_rate": 4.68498156788139e-06,
      "loss": 0.653,
      "step": 31280
    },
    {
      "epoch": 12.81,
      "grad_norm": 2.7000625133514404,
      "learning_rate": 4.684974210749645e-06,
      "loss": 0.6674,
      "step": 31290
    },
    {
      "epoch": 12.81,
      "grad_norm": 2.0653698444366455,
      "learning_rate": 4.684966842893114e-06,
      "loss": 0.6585,
      "step": 31300
    },
    {
      "epoch": 12.82,
      "grad_norm": 2.302689790725708,
      "learning_rate": 4.684959464311831e-06,
      "loss": 0.6747,
      "step": 31310
    },
    {
      "epoch": 12.82,
      "grad_norm": 2.4082961082458496,
      "learning_rate": 4.684952075005831e-06,
      "loss": 0.6624,
      "step": 31320
    },
    {
      "epoch": 12.82,
      "grad_norm": 3.1344332695007324,
      "learning_rate": 4.684944674975146e-06,
      "loss": 0.6704,
      "step": 31330
    },
    {
      "epoch": 12.83,
      "grad_norm": 2.1553428173065186,
      "learning_rate": 4.684937264219811e-06,
      "loss": 0.6456,
      "step": 31340
    },
    {
      "epoch": 12.83,
      "grad_norm": 3.0212912559509277,
      "learning_rate": 4.68492984273986e-06,
      "loss": 0.6713,
      "step": 31350
    },
    {
      "epoch": 12.84,
      "grad_norm": 3.3104069232940674,
      "learning_rate": 4.684922410535326e-06,
      "loss": 0.6724,
      "step": 31360
    },
    {
      "epoch": 12.84,
      "grad_norm": 2.6273727416992188,
      "learning_rate": 4.684914967606245e-06,
      "loss": 0.668,
      "step": 31370
    },
    {
      "epoch": 12.84,
      "grad_norm": 3.09881854057312,
      "learning_rate": 4.684907513952649e-06,
      "loss": 0.6743,
      "step": 31380
    },
    {
      "epoch": 12.85,
      "grad_norm": 2.3655130863189697,
      "learning_rate": 4.684900049574573e-06,
      "loss": 0.6386,
      "step": 31390
    },
    {
      "epoch": 12.85,
      "grad_norm": 2.6265883445739746,
      "learning_rate": 4.684892574472051e-06,
      "loss": 0.6832,
      "step": 31400
    },
    {
      "epoch": 12.86,
      "grad_norm": 2.7381110191345215,
      "learning_rate": 4.684885088645118e-06,
      "loss": 0.6608,
      "step": 31410
    },
    {
      "epoch": 12.86,
      "grad_norm": 2.4614346027374268,
      "learning_rate": 4.684877592093807e-06,
      "loss": 0.6792,
      "step": 31420
    },
    {
      "epoch": 12.87,
      "grad_norm": 2.403258800506592,
      "learning_rate": 4.684870084818154e-06,
      "loss": 0.667,
      "step": 31430
    },
    {
      "epoch": 12.87,
      "grad_norm": 3.5404088497161865,
      "learning_rate": 4.684862566818192e-06,
      "loss": 0.6779,
      "step": 31440
    },
    {
      "epoch": 12.87,
      "grad_norm": 2.2119648456573486,
      "learning_rate": 4.684855038093956e-06,
      "loss": 0.6775,
      "step": 31450
    },
    {
      "epoch": 12.88,
      "grad_norm": 2.6994051933288574,
      "learning_rate": 4.68484749864548e-06,
      "loss": 0.6679,
      "step": 31460
    },
    {
      "epoch": 12.88,
      "grad_norm": 2.5414209365844727,
      "learning_rate": 4.6848399484727995e-06,
      "loss": 0.6808,
      "step": 31470
    },
    {
      "epoch": 12.89,
      "grad_norm": 2.7007625102996826,
      "learning_rate": 4.684832387575948e-06,
      "loss": 0.6567,
      "step": 31480
    },
    {
      "epoch": 12.89,
      "grad_norm": 2.460989236831665,
      "learning_rate": 4.6848248159549605e-06,
      "loss": 0.6561,
      "step": 31490
    },
    {
      "epoch": 12.89,
      "grad_norm": 3.4988927841186523,
      "learning_rate": 4.684817233609873e-06,
      "loss": 0.6812,
      "step": 31500
    },
    {
      "epoch": 12.9,
      "grad_norm": 2.1286613941192627,
      "learning_rate": 4.684809640540718e-06,
      "loss": 0.6629,
      "step": 31510
    },
    {
      "epoch": 12.9,
      "grad_norm": 3.653158664703369,
      "learning_rate": 4.684802036747531e-06,
      "loss": 0.6554,
      "step": 31520
    },
    {
      "epoch": 12.91,
      "grad_norm": 3.139120578765869,
      "learning_rate": 4.684794422230348e-06,
      "loss": 0.6752,
      "step": 31530
    },
    {
      "epoch": 12.91,
      "grad_norm": 2.4316112995147705,
      "learning_rate": 4.684786796989203e-06,
      "loss": 0.6594,
      "step": 31540
    },
    {
      "epoch": 12.91,
      "grad_norm": 2.980776309967041,
      "learning_rate": 4.684779161024131e-06,
      "loss": 0.6812,
      "step": 31550
    },
    {
      "epoch": 12.92,
      "grad_norm": 2.2683005332946777,
      "learning_rate": 4.684771514335167e-06,
      "loss": 0.6678,
      "step": 31560
    },
    {
      "epoch": 12.92,
      "grad_norm": 2.2397735118865967,
      "learning_rate": 4.684763856922345e-06,
      "loss": 0.651,
      "step": 31570
    },
    {
      "epoch": 12.93,
      "grad_norm": 2.507154703140259,
      "learning_rate": 4.684756188785702e-06,
      "loss": 0.6555,
      "step": 31580
    },
    {
      "epoch": 12.93,
      "grad_norm": 3.0041563510894775,
      "learning_rate": 4.6847485099252724e-06,
      "loss": 0.6615,
      "step": 31590
    },
    {
      "epoch": 12.93,
      "grad_norm": 2.2024455070495605,
      "learning_rate": 4.6847408203410905e-06,
      "loss": 0.6721,
      "step": 31600
    },
    {
      "epoch": 12.94,
      "grad_norm": 1.941605806350708,
      "learning_rate": 4.684733120033193e-06,
      "loss": 0.6565,
      "step": 31610
    },
    {
      "epoch": 12.94,
      "grad_norm": 3.4661617279052734,
      "learning_rate": 4.684725409001614e-06,
      "loss": 0.6741,
      "step": 31620
    },
    {
      "epoch": 12.95,
      "grad_norm": 2.363065004348755,
      "learning_rate": 4.684717687246389e-06,
      "loss": 0.6533,
      "step": 31630
    },
    {
      "epoch": 12.95,
      "grad_norm": 2.3792777061462402,
      "learning_rate": 4.684709954767554e-06,
      "loss": 0.6574,
      "step": 31640
    },
    {
      "epoch": 12.96,
      "grad_norm": 2.600226402282715,
      "learning_rate": 4.684702211565143e-06,
      "loss": 0.6735,
      "step": 31650
    },
    {
      "epoch": 12.96,
      "grad_norm": 2.9732701778411865,
      "learning_rate": 4.684694457639193e-06,
      "loss": 0.6816,
      "step": 31660
    },
    {
      "epoch": 12.96,
      "grad_norm": 2.255815029144287,
      "learning_rate": 4.684686692989739e-06,
      "loss": 0.6566,
      "step": 31670
    },
    {
      "epoch": 12.97,
      "grad_norm": 2.1581695079803467,
      "learning_rate": 4.684678917616817e-06,
      "loss": 0.6636,
      "step": 31680
    },
    {
      "epoch": 12.97,
      "grad_norm": 2.1507198810577393,
      "learning_rate": 4.684671131520462e-06,
      "loss": 0.6579,
      "step": 31690
    },
    {
      "epoch": 12.98,
      "grad_norm": 2.452609062194824,
      "learning_rate": 4.684663334700709e-06,
      "loss": 0.6613,
      "step": 31700
    },
    {
      "epoch": 12.98,
      "grad_norm": 2.6883673667907715,
      "learning_rate": 4.684655527157596e-06,
      "loss": 0.6733,
      "step": 31710
    },
    {
      "epoch": 12.98,
      "grad_norm": 2.560027599334717,
      "learning_rate": 4.6846477088911564e-06,
      "loss": 0.6619,
      "step": 31720
    },
    {
      "epoch": 12.99,
      "grad_norm": 2.94258451461792,
      "learning_rate": 4.684639879901427e-06,
      "loss": 0.6426,
      "step": 31730
    },
    {
      "epoch": 12.99,
      "grad_norm": 1.796175479888916,
      "learning_rate": 4.684632040188443e-06,
      "loss": 0.6674,
      "step": 31740
    },
    {
      "epoch": 13.0,
      "grad_norm": 2.4567084312438965,
      "learning_rate": 4.684624189752242e-06,
      "loss": 0.6584,
      "step": 31750
    },
    {
      "epoch": 13.0,
      "eval_loss": 0.6674320697784424,
      "eval_runtime": 51.8139,
      "eval_samples_per_second": 66.565,
      "eval_steps_per_second": 8.338,
      "step": 31759
    },
    {
      "epoch": 13.0,
      "grad_norm": 3.1880452632904053,
      "learning_rate": 4.684616328592858e-06,
      "loss": 0.6999,
      "step": 31760
    },
    {
      "epoch": 13.0,
      "grad_norm": 2.789010763168335,
      "learning_rate": 4.684608456710329e-06,
      "loss": 0.6598,
      "step": 31770
    },
    {
      "epoch": 13.01,
      "grad_norm": 2.9529471397399902,
      "learning_rate": 4.684600574104689e-06,
      "loss": 0.6699,
      "step": 31780
    },
    {
      "epoch": 13.01,
      "grad_norm": 2.460350513458252,
      "learning_rate": 4.684592680775975e-06,
      "loss": 0.6611,
      "step": 31790
    },
    {
      "epoch": 13.02,
      "grad_norm": 2.2524607181549072,
      "learning_rate": 4.684584776724224e-06,
      "loss": 0.653,
      "step": 31800
    },
    {
      "epoch": 13.02,
      "grad_norm": 2.65313720703125,
      "learning_rate": 4.68457686194947e-06,
      "loss": 0.6523,
      "step": 31810
    },
    {
      "epoch": 13.02,
      "grad_norm": 2.3350419998168945,
      "learning_rate": 4.684568936451752e-06,
      "loss": 0.672,
      "step": 31820
    },
    {
      "epoch": 13.03,
      "grad_norm": 2.5113625526428223,
      "learning_rate": 4.6845610002311045e-06,
      "loss": 0.6541,
      "step": 31830
    },
    {
      "epoch": 13.03,
      "grad_norm": 2.4080424308776855,
      "learning_rate": 4.684553053287565e-06,
      "loss": 0.6627,
      "step": 31840
    },
    {
      "epoch": 13.04,
      "grad_norm": 2.989555835723877,
      "learning_rate": 4.684545095621169e-06,
      "loss": 0.6752,
      "step": 31850
    },
    {
      "epoch": 13.04,
      "grad_norm": 1.9665181636810303,
      "learning_rate": 4.684537127231952e-06,
      "loss": 0.6776,
      "step": 31860
    },
    {
      "epoch": 13.05,
      "grad_norm": 2.6204428672790527,
      "learning_rate": 4.684529148119953e-06,
      "loss": 0.684,
      "step": 31870
    },
    {
      "epoch": 13.05,
      "grad_norm": 2.9900643825531006,
      "learning_rate": 4.684521158285206e-06,
      "loss": 0.6623,
      "step": 31880
    },
    {
      "epoch": 13.05,
      "grad_norm": 2.4726779460906982,
      "learning_rate": 4.6845131577277495e-06,
      "loss": 0.6824,
      "step": 31890
    },
    {
      "epoch": 13.06,
      "grad_norm": 3.128265857696533,
      "learning_rate": 4.68450514644762e-06,
      "loss": 0.6443,
      "step": 31900
    },
    {
      "epoch": 13.06,
      "grad_norm": 2.098865270614624,
      "learning_rate": 4.684497124444852e-06,
      "loss": 0.6521,
      "step": 31910
    },
    {
      "epoch": 13.07,
      "grad_norm": 2.6776316165924072,
      "learning_rate": 4.684489091719486e-06,
      "loss": 0.6484,
      "step": 31920
    },
    {
      "epoch": 13.07,
      "grad_norm": 2.7430996894836426,
      "learning_rate": 4.684481048271556e-06,
      "loss": 0.6583,
      "step": 31930
    },
    {
      "epoch": 13.07,
      "grad_norm": 2.711198329925537,
      "learning_rate": 4.684472994101099e-06,
      "loss": 0.6367,
      "step": 31940
    },
    {
      "epoch": 13.08,
      "grad_norm": 3.0871570110321045,
      "learning_rate": 4.684464929208153e-06,
      "loss": 0.6811,
      "step": 31950
    },
    {
      "epoch": 13.08,
      "grad_norm": 2.6214141845703125,
      "learning_rate": 4.684456853592754e-06,
      "loss": 0.6587,
      "step": 31960
    },
    {
      "epoch": 13.09,
      "grad_norm": 3.1938328742980957,
      "learning_rate": 4.68444876725494e-06,
      "loss": 0.6685,
      "step": 31970
    },
    {
      "epoch": 13.09,
      "grad_norm": 2.672274589538574,
      "learning_rate": 4.684440670194748e-06,
      "loss": 0.6625,
      "step": 31980
    },
    {
      "epoch": 13.09,
      "grad_norm": 2.1890387535095215,
      "learning_rate": 4.684432562412213e-06,
      "loss": 0.67,
      "step": 31990
    },
    {
      "epoch": 13.1,
      "grad_norm": 2.7750186920166016,
      "learning_rate": 4.684424443907375e-06,
      "loss": 0.6519,
      "step": 32000
    },
    {
      "epoch": 13.1,
      "grad_norm": 2.1876583099365234,
      "learning_rate": 4.684416314680269e-06,
      "loss": 0.6747,
      "step": 32010
    },
    {
      "epoch": 13.11,
      "grad_norm": 2.9716169834136963,
      "learning_rate": 4.684408174730933e-06,
      "loss": 0.6653,
      "step": 32020
    },
    {
      "epoch": 13.11,
      "grad_norm": 2.8033103942871094,
      "learning_rate": 4.684400024059405e-06,
      "loss": 0.6609,
      "step": 32030
    },
    {
      "epoch": 13.12,
      "grad_norm": 3.2833306789398193,
      "learning_rate": 4.684391862665722e-06,
      "loss": 0.6453,
      "step": 32040
    },
    {
      "epoch": 13.12,
      "grad_norm": 2.9911270141601562,
      "learning_rate": 4.68438369054992e-06,
      "loss": 0.6668,
      "step": 32050
    },
    {
      "epoch": 13.12,
      "grad_norm": 2.5160303115844727,
      "learning_rate": 4.6843755077120385e-06,
      "loss": 0.6951,
      "step": 32060
    },
    {
      "epoch": 13.13,
      "grad_norm": 1.9436931610107422,
      "learning_rate": 4.684367314152114e-06,
      "loss": 0.676,
      "step": 32070
    },
    {
      "epoch": 13.13,
      "grad_norm": 3.99041748046875,
      "learning_rate": 4.684359109870183e-06,
      "loss": 0.6593,
      "step": 32080
    },
    {
      "epoch": 13.14,
      "grad_norm": 2.6674015522003174,
      "learning_rate": 4.6843508948662855e-06,
      "loss": 0.6644,
      "step": 32090
    },
    {
      "epoch": 13.14,
      "grad_norm": 2.5172131061553955,
      "learning_rate": 4.684342669140457e-06,
      "loss": 0.6813,
      "step": 32100
    },
    {
      "epoch": 13.14,
      "grad_norm": 2.70955753326416,
      "learning_rate": 4.684334432692736e-06,
      "loss": 0.6439,
      "step": 32110
    },
    {
      "epoch": 13.15,
      "grad_norm": 2.4852919578552246,
      "learning_rate": 4.684326185523161e-06,
      "loss": 0.6543,
      "step": 32120
    },
    {
      "epoch": 13.15,
      "grad_norm": 2.950007438659668,
      "learning_rate": 4.6843179276317685e-06,
      "loss": 0.6656,
      "step": 32130
    },
    {
      "epoch": 13.16,
      "grad_norm": 2.683166265487671,
      "learning_rate": 4.684309659018596e-06,
      "loss": 0.6611,
      "step": 32140
    },
    {
      "epoch": 13.16,
      "grad_norm": 2.7903332710266113,
      "learning_rate": 4.684301379683684e-06,
      "loss": 0.6749,
      "step": 32150
    },
    {
      "epoch": 13.16,
      "grad_norm": 3.2268526554107666,
      "learning_rate": 4.684293089627067e-06,
      "loss": 0.6725,
      "step": 32160
    },
    {
      "epoch": 13.17,
      "grad_norm": 2.7576920986175537,
      "learning_rate": 4.684284788848785e-06,
      "loss": 0.6447,
      "step": 32170
    },
    {
      "epoch": 13.17,
      "grad_norm": 2.451178789138794,
      "learning_rate": 4.684276477348876e-06,
      "loss": 0.6481,
      "step": 32180
    },
    {
      "epoch": 13.18,
      "grad_norm": 4.0581231117248535,
      "learning_rate": 4.684268155127378e-06,
      "loss": 0.6608,
      "step": 32190
    },
    {
      "epoch": 13.18,
      "grad_norm": 2.542081117630005,
      "learning_rate": 4.6842598221843274e-06,
      "loss": 0.664,
      "step": 32200
    },
    {
      "epoch": 13.18,
      "grad_norm": 2.1464602947235107,
      "learning_rate": 4.6842514785197646e-06,
      "loss": 0.6626,
      "step": 32210
    },
    {
      "epoch": 13.19,
      "grad_norm": 2.255553722381592,
      "learning_rate": 4.684243124133727e-06,
      "loss": 0.6752,
      "step": 32220
    },
    {
      "epoch": 13.19,
      "grad_norm": 2.1285581588745117,
      "learning_rate": 4.6842347590262525e-06,
      "loss": 0.6774,
      "step": 32230
    },
    {
      "epoch": 13.2,
      "grad_norm": 2.6956264972686768,
      "learning_rate": 4.68422638319738e-06,
      "loss": 0.6613,
      "step": 32240
    },
    {
      "epoch": 13.2,
      "grad_norm": 1.7254616022109985,
      "learning_rate": 4.684217996647148e-06,
      "loss": 0.647,
      "step": 32250
    },
    {
      "epoch": 13.21,
      "grad_norm": 2.3817336559295654,
      "learning_rate": 4.6842095993755945e-06,
      "loss": 0.6379,
      "step": 32260
    },
    {
      "epoch": 13.21,
      "grad_norm": 3.537195920944214,
      "learning_rate": 4.684201191382758e-06,
      "loss": 0.6736,
      "step": 32270
    },
    {
      "epoch": 13.21,
      "grad_norm": 2.7386856079101562,
      "learning_rate": 4.684192772668677e-06,
      "loss": 0.6691,
      "step": 32280
    },
    {
      "epoch": 13.22,
      "grad_norm": 2.8707754611968994,
      "learning_rate": 4.68418434323339e-06,
      "loss": 0.6565,
      "step": 32290
    },
    {
      "epoch": 13.22,
      "grad_norm": 2.283473253250122,
      "learning_rate": 4.684175903076936e-06,
      "loss": 0.6592,
      "step": 32300
    },
    {
      "epoch": 13.23,
      "grad_norm": 2.5607595443725586,
      "learning_rate": 4.6841674521993535e-06,
      "loss": 0.659,
      "step": 32310
    },
    {
      "epoch": 13.23,
      "grad_norm": 3.296360969543457,
      "learning_rate": 4.684158990600681e-06,
      "loss": 0.6466,
      "step": 32320
    },
    {
      "epoch": 13.23,
      "grad_norm": 2.174675703048706,
      "learning_rate": 4.684150518280956e-06,
      "loss": 0.6966,
      "step": 32330
    },
    {
      "epoch": 13.24,
      "grad_norm": 2.2852935791015625,
      "learning_rate": 4.684142035240221e-06,
      "loss": 0.6487,
      "step": 32340
    },
    {
      "epoch": 13.24,
      "grad_norm": 2.072695016860962,
      "learning_rate": 4.684133541478511e-06,
      "loss": 0.6778,
      "step": 32350
    },
    {
      "epoch": 13.25,
      "grad_norm": 3.402087450027466,
      "learning_rate": 4.684125036995867e-06,
      "loss": 0.6529,
      "step": 32360
    },
    {
      "epoch": 13.25,
      "grad_norm": 2.15165638923645,
      "learning_rate": 4.684116521792327e-06,
      "loss": 0.6436,
      "step": 32370
    },
    {
      "epoch": 13.25,
      "grad_norm": 2.7767865657806396,
      "learning_rate": 4.684107995867931e-06,
      "loss": 0.6629,
      "step": 32380
    },
    {
      "epoch": 13.26,
      "grad_norm": 2.4546120166778564,
      "learning_rate": 4.684099459222717e-06,
      "loss": 0.6395,
      "step": 32390
    },
    {
      "epoch": 13.26,
      "grad_norm": 2.623356580734253,
      "learning_rate": 4.684090911856724e-06,
      "loss": 0.6525,
      "step": 32400
    },
    {
      "epoch": 13.27,
      "grad_norm": 3.011035442352295,
      "learning_rate": 4.684082353769994e-06,
      "loss": 0.6479,
      "step": 32410
    },
    {
      "epoch": 13.27,
      "grad_norm": 2.5077929496765137,
      "learning_rate": 4.684073784962562e-06,
      "loss": 0.6595,
      "step": 32420
    },
    {
      "epoch": 13.27,
      "grad_norm": 2.6247284412384033,
      "learning_rate": 4.684065205434469e-06,
      "loss": 0.6586,
      "step": 32430
    },
    {
      "epoch": 13.28,
      "grad_norm": 2.7465288639068604,
      "learning_rate": 4.684056615185756e-06,
      "loss": 0.6851,
      "step": 32440
    },
    {
      "epoch": 13.28,
      "grad_norm": 2.329559803009033,
      "learning_rate": 4.6840480142164595e-06,
      "loss": 0.6465,
      "step": 32450
    },
    {
      "epoch": 13.29,
      "grad_norm": 2.9254040718078613,
      "learning_rate": 4.68403940252662e-06,
      "loss": 0.66,
      "step": 32460
    },
    {
      "epoch": 13.29,
      "grad_norm": 3.148763656616211,
      "learning_rate": 4.684030780116278e-06,
      "loss": 0.6449,
      "step": 32470
    },
    {
      "epoch": 13.3,
      "grad_norm": 2.6530776023864746,
      "learning_rate": 4.684022146985472e-06,
      "loss": 0.6478,
      "step": 32480
    },
    {
      "epoch": 13.3,
      "grad_norm": 2.8269670009613037,
      "learning_rate": 4.684013503134242e-06,
      "loss": 0.671,
      "step": 32490
    },
    {
      "epoch": 13.3,
      "grad_norm": 2.789647102355957,
      "learning_rate": 4.6840048485626266e-06,
      "loss": 0.666,
      "step": 32500
    },
    {
      "epoch": 13.31,
      "grad_norm": 3.057908296585083,
      "learning_rate": 4.683996183270667e-06,
      "loss": 0.6809,
      "step": 32510
    },
    {
      "epoch": 13.31,
      "grad_norm": 2.740487813949585,
      "learning_rate": 4.683987507258402e-06,
      "loss": 0.6456,
      "step": 32520
    },
    {
      "epoch": 13.32,
      "grad_norm": 2.552154541015625,
      "learning_rate": 4.683978820525871e-06,
      "loss": 0.6695,
      "step": 32530
    },
    {
      "epoch": 13.32,
      "grad_norm": 2.70100998878479,
      "learning_rate": 4.683970123073115e-06,
      "loss": 0.6565,
      "step": 32540
    },
    {
      "epoch": 13.32,
      "grad_norm": 2.4881575107574463,
      "learning_rate": 4.683961414900172e-06,
      "loss": 0.6551,
      "step": 32550
    },
    {
      "epoch": 13.33,
      "grad_norm": 3.17816162109375,
      "learning_rate": 4.683952696007083e-06,
      "loss": 0.6575,
      "step": 32560
    },
    {
      "epoch": 13.33,
      "grad_norm": 2.431304931640625,
      "learning_rate": 4.683943966393888e-06,
      "loss": 0.6528,
      "step": 32570
    },
    {
      "epoch": 13.34,
      "grad_norm": 3.522827625274658,
      "learning_rate": 4.683935226060627e-06,
      "loss": 0.6551,
      "step": 32580
    },
    {
      "epoch": 13.34,
      "grad_norm": 2.589707374572754,
      "learning_rate": 4.68392647500734e-06,
      "loss": 0.6624,
      "step": 32590
    },
    {
      "epoch": 13.34,
      "grad_norm": 1.9343104362487793,
      "learning_rate": 4.683917713234067e-06,
      "loss": 0.6806,
      "step": 32600
    },
    {
      "epoch": 13.35,
      "grad_norm": 3.68924617767334,
      "learning_rate": 4.683908940740848e-06,
      "loss": 0.6509,
      "step": 32610
    },
    {
      "epoch": 13.35,
      "grad_norm": 2.762190341949463,
      "learning_rate": 4.683900157527724e-06,
      "loss": 0.6601,
      "step": 32620
    },
    {
      "epoch": 13.36,
      "grad_norm": 2.2442073822021484,
      "learning_rate": 4.6838913635947335e-06,
      "loss": 0.6764,
      "step": 32630
    },
    {
      "epoch": 13.36,
      "grad_norm": 2.3010294437408447,
      "learning_rate": 4.683882558941918e-06,
      "loss": 0.6723,
      "step": 32640
    },
    {
      "epoch": 13.36,
      "grad_norm": 3.110269546508789,
      "learning_rate": 4.683873743569318e-06,
      "loss": 0.6737,
      "step": 32650
    },
    {
      "epoch": 13.37,
      "grad_norm": 2.0538887977600098,
      "learning_rate": 4.683864917476974e-06,
      "loss": 0.6787,
      "step": 32660
    },
    {
      "epoch": 13.37,
      "grad_norm": 2.050041913986206,
      "learning_rate": 4.683856080664925e-06,
      "loss": 0.6715,
      "step": 32670
    },
    {
      "epoch": 13.38,
      "grad_norm": 2.3093013763427734,
      "learning_rate": 4.683847233133213e-06,
      "loss": 0.6679,
      "step": 32680
    },
    {
      "epoch": 13.38,
      "grad_norm": 2.466024398803711,
      "learning_rate": 4.683838374881878e-06,
      "loss": 0.6722,
      "step": 32690
    },
    {
      "epoch": 13.39,
      "grad_norm": 2.2931981086730957,
      "learning_rate": 4.683829505910961e-06,
      "loss": 0.6589,
      "step": 32700
    },
    {
      "epoch": 13.39,
      "grad_norm": 2.956547737121582,
      "learning_rate": 4.683820626220501e-06,
      "loss": 0.6842,
      "step": 32710
    },
    {
      "epoch": 13.39,
      "grad_norm": 2.2797415256500244,
      "learning_rate": 4.683811735810541e-06,
      "loss": 0.6755,
      "step": 32720
    },
    {
      "epoch": 13.4,
      "grad_norm": 4.443408012390137,
      "learning_rate": 4.683802834681119e-06,
      "loss": 0.6623,
      "step": 32730
    },
    {
      "epoch": 13.4,
      "grad_norm": 2.0650534629821777,
      "learning_rate": 4.68379392283228e-06,
      "loss": 0.644,
      "step": 32740
    },
    {
      "epoch": 13.41,
      "grad_norm": 2.6695287227630615,
      "learning_rate": 4.68378500026406e-06,
      "loss": 0.6534,
      "step": 32750
    },
    {
      "epoch": 13.41,
      "grad_norm": 2.815199613571167,
      "learning_rate": 4.683776066976503e-06,
      "loss": 0.6401,
      "step": 32760
    },
    {
      "epoch": 13.41,
      "grad_norm": 3.640427589416504,
      "learning_rate": 4.6837671229696485e-06,
      "loss": 0.6479,
      "step": 32770
    },
    {
      "epoch": 13.42,
      "grad_norm": 3.28542423248291,
      "learning_rate": 4.683758168243538e-06,
      "loss": 0.6501,
      "step": 32780
    },
    {
      "epoch": 13.42,
      "grad_norm": 3.1022636890411377,
      "learning_rate": 4.6837492027982135e-06,
      "loss": 0.647,
      "step": 32790
    },
    {
      "epoch": 13.43,
      "grad_norm": 2.2274892330169678,
      "learning_rate": 4.683740226633713e-06,
      "loss": 0.6784,
      "step": 32800
    },
    {
      "epoch": 13.43,
      "grad_norm": 2.26322603225708,
      "learning_rate": 4.683731239750082e-06,
      "loss": 0.631,
      "step": 32810
    },
    {
      "epoch": 13.43,
      "grad_norm": 2.7603771686553955,
      "learning_rate": 4.683722242147357e-06,
      "loss": 0.6478,
      "step": 32820
    },
    {
      "epoch": 13.44,
      "grad_norm": 2.60090970993042,
      "learning_rate": 4.683713233825583e-06,
      "loss": 0.6734,
      "step": 32830
    },
    {
      "epoch": 13.44,
      "grad_norm": 2.551750421524048,
      "learning_rate": 4.6837042147848e-06,
      "loss": 0.6487,
      "step": 32840
    },
    {
      "epoch": 13.45,
      "grad_norm": 3.832637071609497,
      "learning_rate": 4.683695185025048e-06,
      "loss": 0.6574,
      "step": 32850
    },
    {
      "epoch": 13.45,
      "grad_norm": 2.301872968673706,
      "learning_rate": 4.68368614454637e-06,
      "loss": 0.6553,
      "step": 32860
    },
    {
      "epoch": 13.45,
      "grad_norm": 2.1875035762786865,
      "learning_rate": 4.683677093348808e-06,
      "loss": 0.6516,
      "step": 32870
    },
    {
      "epoch": 13.46,
      "grad_norm": 2.718621253967285,
      "learning_rate": 4.683668031432401e-06,
      "loss": 0.6675,
      "step": 32880
    },
    {
      "epoch": 13.46,
      "grad_norm": 3.215789556503296,
      "learning_rate": 4.683658958797192e-06,
      "loss": 0.6502,
      "step": 32890
    },
    {
      "epoch": 13.47,
      "grad_norm": 3.091325044631958,
      "learning_rate": 4.683649875443223e-06,
      "loss": 0.6646,
      "step": 32900
    },
    {
      "epoch": 13.47,
      "grad_norm": 2.940284013748169,
      "learning_rate": 4.683640781370535e-06,
      "loss": 0.6532,
      "step": 32910
    },
    {
      "epoch": 13.48,
      "grad_norm": 2.8248679637908936,
      "learning_rate": 4.68363167657917e-06,
      "loss": 0.6577,
      "step": 32920
    },
    {
      "epoch": 13.48,
      "grad_norm": 2.5920310020446777,
      "learning_rate": 4.683622561069169e-06,
      "loss": 0.637,
      "step": 32930
    },
    {
      "epoch": 13.48,
      "grad_norm": 2.9520204067230225,
      "learning_rate": 4.683613434840574e-06,
      "loss": 0.6551,
      "step": 32940
    },
    {
      "epoch": 13.49,
      "grad_norm": 2.6336913108825684,
      "learning_rate": 4.683604297893427e-06,
      "loss": 0.6734,
      "step": 32950
    },
    {
      "epoch": 13.49,
      "grad_norm": 2.6234889030456543,
      "learning_rate": 4.68359515022777e-06,
      "loss": 0.6617,
      "step": 32960
    },
    {
      "epoch": 13.5,
      "grad_norm": 2.6595733165740967,
      "learning_rate": 4.683585991843644e-06,
      "loss": 0.6492,
      "step": 32970
    },
    {
      "epoch": 13.5,
      "grad_norm": 2.9924373626708984,
      "learning_rate": 4.683576822741093e-06,
      "loss": 0.6714,
      "step": 32980
    },
    {
      "epoch": 13.5,
      "grad_norm": 1.9448816776275635,
      "learning_rate": 4.683567642920157e-06,
      "loss": 0.6376,
      "step": 32990
    },
    {
      "epoch": 13.51,
      "grad_norm": 2.1171014308929443,
      "learning_rate": 4.6835584523808785e-06,
      "loss": 0.6486,
      "step": 33000
    },
    {
      "epoch": 13.51,
      "grad_norm": 2.0180633068084717,
      "learning_rate": 4.6835492511233e-06,
      "loss": 0.6675,
      "step": 33010
    },
    {
      "epoch": 13.52,
      "grad_norm": 2.236424446105957,
      "learning_rate": 4.6835400391474636e-06,
      "loss": 0.6644,
      "step": 33020
    },
    {
      "epoch": 13.52,
      "grad_norm": 4.893224716186523,
      "learning_rate": 4.683530816453412e-06,
      "loss": 0.6584,
      "step": 33030
    },
    {
      "epoch": 13.52,
      "grad_norm": 2.213085889816284,
      "learning_rate": 4.683521583041186e-06,
      "loss": 0.6641,
      "step": 33040
    },
    {
      "epoch": 13.53,
      "grad_norm": 2.882657051086426,
      "learning_rate": 4.683512338910829e-06,
      "loss": 0.6676,
      "step": 33050
    },
    {
      "epoch": 13.53,
      "grad_norm": 1.97665536403656,
      "learning_rate": 4.683503084062383e-06,
      "loss": 0.6641,
      "step": 33060
    },
    {
      "epoch": 13.54,
      "grad_norm": 2.606344223022461,
      "learning_rate": 4.683493818495891e-06,
      "loss": 0.6375,
      "step": 33070
    },
    {
      "epoch": 13.54,
      "grad_norm": 2.894026279449463,
      "learning_rate": 4.683484542211394e-06,
      "loss": 0.671,
      "step": 33080
    },
    {
      "epoch": 13.54,
      "grad_norm": 2.5605151653289795,
      "learning_rate": 4.683475255208937e-06,
      "loss": 0.652,
      "step": 33090
    },
    {
      "epoch": 13.55,
      "grad_norm": 2.966402053833008,
      "learning_rate": 4.683465957488559e-06,
      "loss": 0.644,
      "step": 33100
    },
    {
      "epoch": 13.55,
      "grad_norm": 2.1215322017669678,
      "learning_rate": 4.683456649050305e-06,
      "loss": 0.6428,
      "step": 33110
    },
    {
      "epoch": 13.56,
      "grad_norm": 2.0573103427886963,
      "learning_rate": 4.683447329894218e-06,
      "loss": 0.6293,
      "step": 33120
    },
    {
      "epoch": 13.56,
      "grad_norm": 2.7311837673187256,
      "learning_rate": 4.6834380000203395e-06,
      "loss": 0.6543,
      "step": 33130
    },
    {
      "epoch": 13.57,
      "grad_norm": 2.576164484024048,
      "learning_rate": 4.683428659428713e-06,
      "loss": 0.653,
      "step": 33140
    },
    {
      "epoch": 13.57,
      "grad_norm": 2.2703516483306885,
      "learning_rate": 4.68341930811938e-06,
      "loss": 0.6371,
      "step": 33150
    },
    {
      "epoch": 13.57,
      "grad_norm": 2.7844345569610596,
      "learning_rate": 4.683409946092385e-06,
      "loss": 0.6429,
      "step": 33160
    },
    {
      "epoch": 13.58,
      "grad_norm": 2.298175573348999,
      "learning_rate": 4.68340057334777e-06,
      "loss": 0.6437,
      "step": 33170
    },
    {
      "epoch": 13.58,
      "grad_norm": 2.1241962909698486,
      "learning_rate": 4.683391189885577e-06,
      "loss": 0.6642,
      "step": 33180
    },
    {
      "epoch": 13.59,
      "grad_norm": 2.5471248626708984,
      "learning_rate": 4.683381795705852e-06,
      "loss": 0.6237,
      "step": 33190
    },
    {
      "epoch": 13.59,
      "grad_norm": 3.056684970855713,
      "learning_rate": 4.683372390808634e-06,
      "loss": 0.6506,
      "step": 33200
    },
    {
      "epoch": 13.59,
      "grad_norm": 2.517435312271118,
      "learning_rate": 4.68336297519397e-06,
      "loss": 0.6352,
      "step": 33210
    },
    {
      "epoch": 13.6,
      "grad_norm": 2.122680425643921,
      "learning_rate": 4.6833535488619005e-06,
      "loss": 0.654,
      "step": 33220
    },
    {
      "epoch": 13.6,
      "grad_norm": 2.703778028488159,
      "learning_rate": 4.68334411181247e-06,
      "loss": 0.671,
      "step": 33230
    },
    {
      "epoch": 13.61,
      "grad_norm": 2.5015218257904053,
      "learning_rate": 4.6833346640457195e-06,
      "loss": 0.6456,
      "step": 33240
    },
    {
      "epoch": 13.61,
      "grad_norm": 2.08272123336792,
      "learning_rate": 4.683325205561696e-06,
      "loss": 0.6775,
      "step": 33250
    },
    {
      "epoch": 13.61,
      "grad_norm": 2.0302157402038574,
      "learning_rate": 4.68331573636044e-06,
      "loss": 0.6554,
      "step": 33260
    },
    {
      "epoch": 13.62,
      "grad_norm": 2.2979228496551514,
      "learning_rate": 4.683306256441995e-06,
      "loss": 0.6503,
      "step": 33270
    },
    {
      "epoch": 13.62,
      "grad_norm": 2.233821153640747,
      "learning_rate": 4.683296765806406e-06,
      "loss": 0.6512,
      "step": 33280
    },
    {
      "epoch": 13.63,
      "grad_norm": 1.9383947849273682,
      "learning_rate": 4.683287264453716e-06,
      "loss": 0.669,
      "step": 33290
    },
    {
      "epoch": 13.63,
      "grad_norm": 2.9829444885253906,
      "learning_rate": 4.683277752383967e-06,
      "loss": 0.6424,
      "step": 33300
    },
    {
      "epoch": 13.63,
      "grad_norm": 2.177283763885498,
      "learning_rate": 4.683268229597204e-06,
      "loss": 0.6393,
      "step": 33310
    },
    {
      "epoch": 13.64,
      "grad_norm": 3.055663585662842,
      "learning_rate": 4.6832586960934716e-06,
      "loss": 0.6532,
      "step": 33320
    },
    {
      "epoch": 13.64,
      "grad_norm": 2.9312477111816406,
      "learning_rate": 4.68324915187281e-06,
      "loss": 0.6682,
      "step": 33330
    },
    {
      "epoch": 13.65,
      "grad_norm": 2.153606414794922,
      "learning_rate": 4.683239596935267e-06,
      "loss": 0.6388,
      "step": 33340
    },
    {
      "epoch": 13.65,
      "grad_norm": 2.806257963180542,
      "learning_rate": 4.6832300312808834e-06,
      "loss": 0.6585,
      "step": 33350
    },
    {
      "epoch": 13.66,
      "grad_norm": 2.647712469100952,
      "learning_rate": 4.683220454909704e-06,
      "loss": 0.6509,
      "step": 33360
    },
    {
      "epoch": 13.66,
      "grad_norm": 2.757913589477539,
      "learning_rate": 4.683210867821774e-06,
      "loss": 0.647,
      "step": 33370
    },
    {
      "epoch": 13.66,
      "grad_norm": 2.559781074523926,
      "learning_rate": 4.683201270017135e-06,
      "loss": 0.6354,
      "step": 33380
    },
    {
      "epoch": 13.67,
      "grad_norm": 3.1971709728240967,
      "learning_rate": 4.6831916614958315e-06,
      "loss": 0.6465,
      "step": 33390
    },
    {
      "epoch": 13.67,
      "grad_norm": 2.279564142227173,
      "learning_rate": 4.683182042257909e-06,
      "loss": 0.6463,
      "step": 33400
    },
    {
      "epoch": 13.68,
      "grad_norm": 2.9230093955993652,
      "learning_rate": 4.68317241230341e-06,
      "loss": 0.64,
      "step": 33410
    },
    {
      "epoch": 13.68,
      "grad_norm": 2.660216808319092,
      "learning_rate": 4.6831627716323805e-06,
      "loss": 0.6352,
      "step": 33420
    },
    {
      "epoch": 13.68,
      "grad_norm": 2.8584258556365967,
      "learning_rate": 4.6831531202448625e-06,
      "loss": 0.6364,
      "step": 33430
    },
    {
      "epoch": 13.69,
      "grad_norm": 2.549792766571045,
      "learning_rate": 4.683143458140901e-06,
      "loss": 0.6513,
      "step": 33440
    },
    {
      "epoch": 13.69,
      "grad_norm": 3.2838051319122314,
      "learning_rate": 4.68313378532054e-06,
      "loss": 0.6762,
      "step": 33450
    },
    {
      "epoch": 13.7,
      "grad_norm": 3.0990216732025146,
      "learning_rate": 4.683124101783825e-06,
      "loss": 0.6812,
      "step": 33460
    },
    {
      "epoch": 13.7,
      "grad_norm": 3.3470299243927,
      "learning_rate": 4.6831144075307995e-06,
      "loss": 0.6539,
      "step": 33470
    },
    {
      "epoch": 13.7,
      "grad_norm": 2.9801204204559326,
      "learning_rate": 4.683104702561508e-06,
      "loss": 0.6606,
      "step": 33480
    },
    {
      "epoch": 13.71,
      "grad_norm": 2.0826523303985596,
      "learning_rate": 4.683094986875994e-06,
      "loss": 0.6261,
      "step": 33490
    },
    {
      "epoch": 13.71,
      "grad_norm": 2.40470552444458,
      "learning_rate": 4.683085260474304e-06,
      "loss": 0.6412,
      "step": 33500
    },
    {
      "epoch": 13.72,
      "grad_norm": 2.4586222171783447,
      "learning_rate": 4.68307552335648e-06,
      "loss": 0.667,
      "step": 33510
    },
    {
      "epoch": 13.72,
      "grad_norm": 2.89026141166687,
      "learning_rate": 4.6830657755225695e-06,
      "loss": 0.6496,
      "step": 33520
    },
    {
      "epoch": 13.72,
      "grad_norm": 2.9770567417144775,
      "learning_rate": 4.6830560169726155e-06,
      "loss": 0.6528,
      "step": 33530
    },
    {
      "epoch": 13.73,
      "grad_norm": 2.559288263320923,
      "learning_rate": 4.683046247706662e-06,
      "loss": 0.6612,
      "step": 33540
    },
    {
      "epoch": 13.73,
      "grad_norm": 2.3180224895477295,
      "learning_rate": 4.683036467724756e-06,
      "loss": 0.6618,
      "step": 33550
    },
    {
      "epoch": 13.74,
      "grad_norm": 2.307389736175537,
      "learning_rate": 4.683026677026941e-06,
      "loss": 0.6648,
      "step": 33560
    },
    {
      "epoch": 13.74,
      "grad_norm": 3.901636838912964,
      "learning_rate": 4.6830168756132605e-06,
      "loss": 0.6541,
      "step": 33570
    },
    {
      "epoch": 13.75,
      "grad_norm": 2.125566244125366,
      "learning_rate": 4.6830070634837615e-06,
      "loss": 0.6645,
      "step": 33580
    },
    {
      "epoch": 13.75,
      "grad_norm": 2.5972673892974854,
      "learning_rate": 4.682997240638488e-06,
      "loss": 0.6429,
      "step": 33590
    },
    {
      "epoch": 13.75,
      "grad_norm": 2.8895742893218994,
      "learning_rate": 4.682987407077485e-06,
      "loss": 0.6494,
      "step": 33600
    },
    {
      "epoch": 13.76,
      "grad_norm": 1.8422245979309082,
      "learning_rate": 4.682977562800798e-06,
      "loss": 0.6526,
      "step": 33610
    },
    {
      "epoch": 13.76,
      "grad_norm": 2.73919677734375,
      "learning_rate": 4.6829677078084715e-06,
      "loss": 0.658,
      "step": 33620
    },
    {
      "epoch": 13.77,
      "grad_norm": 2.6168220043182373,
      "learning_rate": 4.682957842100551e-06,
      "loss": 0.6594,
      "step": 33630
    },
    {
      "epoch": 13.77,
      "grad_norm": 2.4466164112091064,
      "learning_rate": 4.6829479656770825e-06,
      "loss": 0.6601,
      "step": 33640
    },
    {
      "epoch": 13.77,
      "grad_norm": 2.338501214981079,
      "learning_rate": 4.68293807853811e-06,
      "loss": 0.6711,
      "step": 33650
    },
    {
      "epoch": 13.78,
      "grad_norm": 2.3570830821990967,
      "learning_rate": 4.682928180683678e-06,
      "loss": 0.6633,
      "step": 33660
    },
    {
      "epoch": 13.78,
      "grad_norm": 2.3255271911621094,
      "learning_rate": 4.682918272113834e-06,
      "loss": 0.6537,
      "step": 33670
    },
    {
      "epoch": 13.79,
      "grad_norm": 2.6744542121887207,
      "learning_rate": 4.682908352828622e-06,
      "loss": 0.6548,
      "step": 33680
    },
    {
      "epoch": 13.79,
      "grad_norm": 2.5721144676208496,
      "learning_rate": 4.682898422828089e-06,
      "loss": 0.6664,
      "step": 33690
    },
    {
      "epoch": 13.79,
      "grad_norm": 2.555354356765747,
      "learning_rate": 4.6828884821122785e-06,
      "loss": 0.6604,
      "step": 33700
    },
    {
      "epoch": 13.8,
      "grad_norm": 2.5295112133026123,
      "learning_rate": 4.682878530681237e-06,
      "loss": 0.6348,
      "step": 33710
    },
    {
      "epoch": 13.8,
      "grad_norm": 3.304725408554077,
      "learning_rate": 4.68286856853501e-06,
      "loss": 0.6435,
      "step": 33720
    },
    {
      "epoch": 13.81,
      "grad_norm": 2.1617672443389893,
      "learning_rate": 4.682858595673643e-06,
      "loss": 0.6548,
      "step": 33730
    },
    {
      "epoch": 13.81,
      "grad_norm": 2.1933634281158447,
      "learning_rate": 4.682848612097182e-06,
      "loss": 0.6442,
      "step": 33740
    },
    {
      "epoch": 13.81,
      "grad_norm": 2.364704132080078,
      "learning_rate": 4.6828386178056716e-06,
      "loss": 0.6589,
      "step": 33750
    },
    {
      "epoch": 13.82,
      "grad_norm": 2.623358726501465,
      "learning_rate": 4.682828612799159e-06,
      "loss": 0.6448,
      "step": 33760
    },
    {
      "epoch": 13.82,
      "grad_norm": 2.4616646766662598,
      "learning_rate": 4.68281859707769e-06,
      "loss": 0.6634,
      "step": 33770
    },
    {
      "epoch": 13.83,
      "grad_norm": 3.5198051929473877,
      "learning_rate": 4.68280857064131e-06,
      "loss": 0.6546,
      "step": 33780
    },
    {
      "epoch": 13.83,
      "grad_norm": 1.921637773513794,
      "learning_rate": 4.682798533490064e-06,
      "loss": 0.6812,
      "step": 33790
    },
    {
      "epoch": 13.84,
      "grad_norm": 2.492175340652466,
      "learning_rate": 4.682788485624e-06,
      "loss": 0.65,
      "step": 33800
    },
    {
      "epoch": 13.84,
      "grad_norm": 4.524777412414551,
      "learning_rate": 4.6827784270431615e-06,
      "loss": 0.6263,
      "step": 33810
    },
    {
      "epoch": 13.84,
      "grad_norm": 2.720367670059204,
      "learning_rate": 4.682768357747598e-06,
      "loss": 0.6736,
      "step": 33820
    },
    {
      "epoch": 13.85,
      "grad_norm": 2.5834171772003174,
      "learning_rate": 4.682758277737353e-06,
      "loss": 0.6591,
      "step": 33830
    },
    {
      "epoch": 13.85,
      "grad_norm": 2.277924060821533,
      "learning_rate": 4.682748187012472e-06,
      "loss": 0.6562,
      "step": 33840
    },
    {
      "epoch": 13.86,
      "grad_norm": 3.224459409713745,
      "learning_rate": 4.6827380855730034e-06,
      "loss": 0.6233,
      "step": 33850
    },
    {
      "epoch": 13.86,
      "grad_norm": 2.580604314804077,
      "learning_rate": 4.682727973418993e-06,
      "loss": 0.6554,
      "step": 33860
    },
    {
      "epoch": 13.86,
      "grad_norm": 3.0203654766082764,
      "learning_rate": 4.682717850550487e-06,
      "loss": 0.6364,
      "step": 33870
    },
    {
      "epoch": 13.87,
      "grad_norm": 2.0844428539276123,
      "learning_rate": 4.682707716967531e-06,
      "loss": 0.6699,
      "step": 33880
    },
    {
      "epoch": 13.87,
      "grad_norm": 2.1735262870788574,
      "learning_rate": 4.682697572670171e-06,
      "loss": 0.6487,
      "step": 33890
    },
    {
      "epoch": 13.88,
      "grad_norm": 2.0789601802825928,
      "learning_rate": 4.682687417658456e-06,
      "loss": 0.6436,
      "step": 33900
    },
    {
      "epoch": 13.88,
      "grad_norm": 2.308412790298462,
      "learning_rate": 4.68267725193243e-06,
      "loss": 0.6603,
      "step": 33910
    },
    {
      "epoch": 13.88,
      "grad_norm": 2.210557222366333,
      "learning_rate": 4.682667075492141e-06,
      "loss": 0.655,
      "step": 33920
    },
    {
      "epoch": 13.89,
      "grad_norm": 2.466417074203491,
      "learning_rate": 4.6826568883376345e-06,
      "loss": 0.6524,
      "step": 33930
    },
    {
      "epoch": 13.89,
      "grad_norm": 2.468797206878662,
      "learning_rate": 4.6826466904689586e-06,
      "loss": 0.686,
      "step": 33940
    },
    {
      "epoch": 13.9,
      "grad_norm": 2.2287962436676025,
      "learning_rate": 4.682636481886159e-06,
      "loss": 0.6621,
      "step": 33950
    },
    {
      "epoch": 13.9,
      "grad_norm": 2.732647657394409,
      "learning_rate": 4.682626262589281e-06,
      "loss": 0.6622,
      "step": 33960
    },
    {
      "epoch": 13.91,
      "grad_norm": 1.985825777053833,
      "learning_rate": 4.682616032578375e-06,
      "loss": 0.6475,
      "step": 33970
    },
    {
      "epoch": 13.91,
      "grad_norm": 2.393925189971924,
      "learning_rate": 4.682605791853485e-06,
      "loss": 0.6499,
      "step": 33980
    },
    {
      "epoch": 13.91,
      "grad_norm": 1.8884669542312622,
      "learning_rate": 4.682595540414659e-06,
      "loss": 0.6473,
      "step": 33990
    },
    {
      "epoch": 13.92,
      "grad_norm": 3.102710485458374,
      "learning_rate": 4.682585278261944e-06,
      "loss": 0.658,
      "step": 34000
    },
    {
      "epoch": 13.92,
      "grad_norm": 3.3464303016662598,
      "learning_rate": 4.682575005395387e-06,
      "loss": 0.6321,
      "step": 34010
    },
    {
      "epoch": 13.93,
      "grad_norm": 2.658479690551758,
      "learning_rate": 4.682564721815035e-06,
      "loss": 0.6674,
      "step": 34020
    },
    {
      "epoch": 13.93,
      "grad_norm": 2.704326629638672,
      "learning_rate": 4.682554427520934e-06,
      "loss": 0.6668,
      "step": 34030
    },
    {
      "epoch": 13.93,
      "grad_norm": 3.434011459350586,
      "learning_rate": 4.682544122513133e-06,
      "loss": 0.6702,
      "step": 34040
    },
    {
      "epoch": 13.94,
      "grad_norm": 2.423036575317383,
      "learning_rate": 4.682533806791677e-06,
      "loss": 0.6443,
      "step": 34050
    },
    {
      "epoch": 13.94,
      "grad_norm": 2.162327527999878,
      "learning_rate": 4.682523480356616e-06,
      "loss": 0.6448,
      "step": 34060
    },
    {
      "epoch": 13.95,
      "grad_norm": 2.544757604598999,
      "learning_rate": 4.682513143207996e-06,
      "loss": 0.6284,
      "step": 34070
    },
    {
      "epoch": 13.95,
      "grad_norm": 2.9272189140319824,
      "learning_rate": 4.682502795345864e-06,
      "loss": 0.6503,
      "step": 34080
    },
    {
      "epoch": 13.95,
      "grad_norm": 2.963308095932007,
      "learning_rate": 4.682492436770267e-06,
      "loss": 0.6351,
      "step": 34090
    },
    {
      "epoch": 13.96,
      "grad_norm": 2.2274858951568604,
      "learning_rate": 4.6824820674812535e-06,
      "loss": 0.6537,
      "step": 34100
    },
    {
      "epoch": 13.96,
      "grad_norm": 3.0173842906951904,
      "learning_rate": 4.682471687478871e-06,
      "loss": 0.6719,
      "step": 34110
    },
    {
      "epoch": 13.97,
      "grad_norm": 2.6692869663238525,
      "learning_rate": 4.682461296763166e-06,
      "loss": 0.6603,
      "step": 34120
    },
    {
      "epoch": 13.97,
      "grad_norm": 2.4639194011688232,
      "learning_rate": 4.6824508953341874e-06,
      "loss": 0.6476,
      "step": 34130
    },
    {
      "epoch": 13.97,
      "grad_norm": 3.186432123184204,
      "learning_rate": 4.6824404831919816e-06,
      "loss": 0.6373,
      "step": 34140
    },
    {
      "epoch": 13.98,
      "grad_norm": 2.7917702198028564,
      "learning_rate": 4.682430060336597e-06,
      "loss": 0.6383,
      "step": 34150
    },
    {
      "epoch": 13.98,
      "grad_norm": 2.2113122940063477,
      "learning_rate": 4.682419626768081e-06,
      "loss": 0.656,
      "step": 34160
    },
    {
      "epoch": 13.99,
      "grad_norm": 3.0229218006134033,
      "learning_rate": 4.682409182486482e-06,
      "loss": 0.6618,
      "step": 34170
    },
    {
      "epoch": 13.99,
      "grad_norm": 2.7553746700286865,
      "learning_rate": 4.682398727491848e-06,
      "loss": 0.6531,
      "step": 34180
    },
    {
      "epoch": 14.0,
      "grad_norm": 2.823422908782959,
      "learning_rate": 4.682388261784225e-06,
      "loss": 0.6305,
      "step": 34190
    },
    {
      "epoch": 14.0,
      "grad_norm": 2.357386589050293,
      "learning_rate": 4.682377785363664e-06,
      "loss": 0.6619,
      "step": 34200
    },
    {
      "epoch": 14.0,
      "eval_loss": 0.6506578922271729,
      "eval_runtime": 51.9376,
      "eval_samples_per_second": 66.407,
      "eval_steps_per_second": 8.318,
      "step": 34202
    },
    {
      "epoch": 14.0,
      "grad_norm": 2.7678706645965576,
      "learning_rate": 4.68236729823021e-06,
      "loss": 0.6557,
      "step": 34210
    },
    {
      "epoch": 14.01,
      "grad_norm": 2.8846094608306885,
      "learning_rate": 4.682356800383912e-06,
      "loss": 0.6373,
      "step": 34220
    },
    {
      "epoch": 14.01,
      "grad_norm": 2.7043070793151855,
      "learning_rate": 4.68234629182482e-06,
      "loss": 0.6387,
      "step": 34230
    },
    {
      "epoch": 14.02,
      "grad_norm": 2.67177677154541,
      "learning_rate": 4.682335772552979e-06,
      "loss": 0.6593,
      "step": 34240
    },
    {
      "epoch": 14.02,
      "grad_norm": 2.6412062644958496,
      "learning_rate": 4.68232524256844e-06,
      "loss": 0.659,
      "step": 34250
    },
    {
      "epoch": 14.02,
      "grad_norm": 1.9349182844161987,
      "learning_rate": 4.6823147018712495e-06,
      "loss": 0.6595,
      "step": 34260
    },
    {
      "epoch": 14.03,
      "grad_norm": 3.148360013961792,
      "learning_rate": 4.682304150461456e-06,
      "loss": 0.6421,
      "step": 34270
    },
    {
      "epoch": 14.03,
      "grad_norm": 2.2587623596191406,
      "learning_rate": 4.682293588339109e-06,
      "loss": 0.6516,
      "step": 34280
    },
    {
      "epoch": 14.04,
      "grad_norm": 2.5797595977783203,
      "learning_rate": 4.682283015504255e-06,
      "loss": 0.6392,
      "step": 34290
    },
    {
      "epoch": 14.04,
      "grad_norm": 2.5658493041992188,
      "learning_rate": 4.682272431956944e-06,
      "loss": 0.6314,
      "step": 34300
    },
    {
      "epoch": 14.04,
      "grad_norm": 2.8536229133605957,
      "learning_rate": 4.682261837697223e-06,
      "loss": 0.6601,
      "step": 34310
    },
    {
      "epoch": 14.05,
      "grad_norm": 2.7613229751586914,
      "learning_rate": 4.6822512327251425e-06,
      "loss": 0.6533,
      "step": 34320
    },
    {
      "epoch": 14.05,
      "grad_norm": 2.110917091369629,
      "learning_rate": 4.682240617040749e-06,
      "loss": 0.667,
      "step": 34330
    },
    {
      "epoch": 14.06,
      "grad_norm": 2.4745664596557617,
      "learning_rate": 4.682229990644093e-06,
      "loss": 0.6522,
      "step": 34340
    },
    {
      "epoch": 14.06,
      "grad_norm": 3.2785232067108154,
      "learning_rate": 4.682219353535222e-06,
      "loss": 0.6325,
      "step": 34350
    },
    {
      "epoch": 14.06,
      "grad_norm": 3.0514659881591797,
      "learning_rate": 4.682208705714185e-06,
      "loss": 0.6612,
      "step": 34360
    },
    {
      "epoch": 14.07,
      "grad_norm": 2.2828497886657715,
      "learning_rate": 4.6821980471810305e-06,
      "loss": 0.6371,
      "step": 34370
    },
    {
      "epoch": 14.07,
      "grad_norm": 3.003390073776245,
      "learning_rate": 4.682187377935809e-06,
      "loss": 0.6599,
      "step": 34380
    },
    {
      "epoch": 14.08,
      "grad_norm": 2.8996264934539795,
      "learning_rate": 4.6821766979785665e-06,
      "loss": 0.6535,
      "step": 34390
    },
    {
      "epoch": 14.08,
      "grad_norm": 2.844715118408203,
      "learning_rate": 4.682166007309354e-06,
      "loss": 0.6615,
      "step": 34400
    },
    {
      "epoch": 14.09,
      "grad_norm": 2.905771493911743,
      "learning_rate": 4.68215530592822e-06,
      "loss": 0.6316,
      "step": 34410
    },
    {
      "epoch": 14.09,
      "grad_norm": 2.4071171283721924,
      "learning_rate": 4.682144593835213e-06,
      "loss": 0.6442,
      "step": 34420
    },
    {
      "epoch": 14.09,
      "grad_norm": 2.4862635135650635,
      "learning_rate": 4.682133871030383e-06,
      "loss": 0.6456,
      "step": 34430
    },
    {
      "epoch": 14.1,
      "grad_norm": 2.1384739875793457,
      "learning_rate": 4.68212313751378e-06,
      "loss": 0.6558,
      "step": 34440
    },
    {
      "epoch": 14.1,
      "grad_norm": 2.783815860748291,
      "learning_rate": 4.682112393285449e-06,
      "loss": 0.6474,
      "step": 34450
    },
    {
      "epoch": 14.11,
      "grad_norm": 2.465681552886963,
      "learning_rate": 4.6821016383454444e-06,
      "loss": 0.6478,
      "step": 34460
    },
    {
      "epoch": 14.11,
      "grad_norm": 2.9384381771087646,
      "learning_rate": 4.682090872693812e-06,
      "loss": 0.6655,
      "step": 34470
    },
    {
      "epoch": 14.11,
      "grad_norm": 2.848644971847534,
      "learning_rate": 4.6820800963306014e-06,
      "loss": 0.6461,
      "step": 34480
    },
    {
      "epoch": 14.12,
      "grad_norm": 2.681107997894287,
      "learning_rate": 4.682069309255864e-06,
      "loss": 0.6248,
      "step": 34490
    },
    {
      "epoch": 14.12,
      "grad_norm": 3.1919915676116943,
      "learning_rate": 4.682058511469647e-06,
      "loss": 0.6513,
      "step": 34500
    },
    {
      "epoch": 14.13,
      "grad_norm": 2.5211615562438965,
      "learning_rate": 4.6820477029720015e-06,
      "loss": 0.6483,
      "step": 34510
    },
    {
      "epoch": 14.13,
      "grad_norm": 2.404459238052368,
      "learning_rate": 4.682036883762977e-06,
      "loss": 0.6688,
      "step": 34520
    },
    {
      "epoch": 14.13,
      "grad_norm": 2.2706730365753174,
      "learning_rate": 4.682026053842622e-06,
      "loss": 0.6741,
      "step": 34530
    },
    {
      "epoch": 14.14,
      "grad_norm": 3.804226875305176,
      "learning_rate": 4.682015213210986e-06,
      "loss": 0.6501,
      "step": 34540
    },
    {
      "epoch": 14.14,
      "grad_norm": 2.956646680831909,
      "learning_rate": 4.682004361868118e-06,
      "loss": 0.64,
      "step": 34550
    },
    {
      "epoch": 14.15,
      "grad_norm": 2.786208391189575,
      "learning_rate": 4.681993499814071e-06,
      "loss": 0.6382,
      "step": 34560
    },
    {
      "epoch": 14.15,
      "grad_norm": 2.3201019763946533,
      "learning_rate": 4.6819826270488924e-06,
      "loss": 0.6721,
      "step": 34570
    },
    {
      "epoch": 14.15,
      "grad_norm": 1.95960533618927,
      "learning_rate": 4.6819717435726314e-06,
      "loss": 0.6532,
      "step": 34580
    },
    {
      "epoch": 14.16,
      "grad_norm": 2.611140251159668,
      "learning_rate": 4.681960849385339e-06,
      "loss": 0.6484,
      "step": 34590
    },
    {
      "epoch": 14.16,
      "grad_norm": 2.730512857437134,
      "learning_rate": 4.681949944487065e-06,
      "loss": 0.6377,
      "step": 34600
    },
    {
      "epoch": 14.17,
      "grad_norm": 2.764904499053955,
      "learning_rate": 4.6819390288778596e-06,
      "loss": 0.6489,
      "step": 34610
    },
    {
      "epoch": 14.17,
      "grad_norm": 2.208890914916992,
      "learning_rate": 4.681928102557772e-06,
      "loss": 0.6649,
      "step": 34620
    },
    {
      "epoch": 14.18,
      "grad_norm": 2.4107108116149902,
      "learning_rate": 4.681917165526852e-06,
      "loss": 0.6403,
      "step": 34630
    },
    {
      "epoch": 14.18,
      "grad_norm": 2.463090419769287,
      "learning_rate": 4.681906217785151e-06,
      "loss": 0.678,
      "step": 34640
    },
    {
      "epoch": 14.18,
      "grad_norm": 2.844182252883911,
      "learning_rate": 4.681895259332719e-06,
      "loss": 0.6435,
      "step": 34650
    },
    {
      "epoch": 14.19,
      "grad_norm": 2.436920404434204,
      "learning_rate": 4.681884290169604e-06,
      "loss": 0.6441,
      "step": 34660
    },
    {
      "epoch": 14.19,
      "grad_norm": 1.9941515922546387,
      "learning_rate": 4.681873310295859e-06,
      "loss": 0.6543,
      "step": 34670
    },
    {
      "epoch": 14.2,
      "grad_norm": 2.165769100189209,
      "learning_rate": 4.681862319711533e-06,
      "loss": 0.6701,
      "step": 34680
    },
    {
      "epoch": 14.2,
      "grad_norm": 2.509922742843628,
      "learning_rate": 4.681851318416677e-06,
      "loss": 0.6448,
      "step": 34690
    },
    {
      "epoch": 14.2,
      "grad_norm": 3.517256498336792,
      "learning_rate": 4.681840306411341e-06,
      "loss": 0.6391,
      "step": 34700
    },
    {
      "epoch": 14.21,
      "grad_norm": 1.9166582822799683,
      "learning_rate": 4.681829283695574e-06,
      "loss": 0.6495,
      "step": 34710
    },
    {
      "epoch": 14.21,
      "grad_norm": 2.4867920875549316,
      "learning_rate": 4.68181825026943e-06,
      "loss": 0.6367,
      "step": 34720
    },
    {
      "epoch": 14.22,
      "grad_norm": 2.3544652462005615,
      "learning_rate": 4.681807206132956e-06,
      "loss": 0.6455,
      "step": 34730
    },
    {
      "epoch": 14.22,
      "grad_norm": 2.3575873374938965,
      "learning_rate": 4.681796151286205e-06,
      "loss": 0.6748,
      "step": 34740
    },
    {
      "epoch": 14.22,
      "grad_norm": 2.6190993785858154,
      "learning_rate": 4.681785085729226e-06,
      "loss": 0.6489,
      "step": 34750
    },
    {
      "epoch": 14.23,
      "grad_norm": 3.9520230293273926,
      "learning_rate": 4.681774009462071e-06,
      "loss": 0.6538,
      "step": 34760
    },
    {
      "epoch": 14.23,
      "grad_norm": 2.8494374752044678,
      "learning_rate": 4.681762922484789e-06,
      "loss": 0.643,
      "step": 34770
    },
    {
      "epoch": 14.24,
      "grad_norm": 2.2677457332611084,
      "learning_rate": 4.681751824797432e-06,
      "loss": 0.6529,
      "step": 34780
    },
    {
      "epoch": 14.24,
      "grad_norm": 2.7508926391601562,
      "learning_rate": 4.6817407164000514e-06,
      "loss": 0.6528,
      "step": 34790
    },
    {
      "epoch": 14.24,
      "grad_norm": 2.158733367919922,
      "learning_rate": 4.6817295972926974e-06,
      "loss": 0.6522,
      "step": 34800
    },
    {
      "epoch": 14.25,
      "grad_norm": 2.174143075942993,
      "learning_rate": 4.681718467475421e-06,
      "loss": 0.6451,
      "step": 34810
    },
    {
      "epoch": 14.25,
      "grad_norm": 2.083007335662842,
      "learning_rate": 4.681707326948273e-06,
      "loss": 0.6491,
      "step": 34820
    },
    {
      "epoch": 14.26,
      "grad_norm": 2.7498929500579834,
      "learning_rate": 4.681696175711304e-06,
      "loss": 0.6375,
      "step": 34830
    },
    {
      "epoch": 14.26,
      "grad_norm": 3.172348737716675,
      "learning_rate": 4.681685013764566e-06,
      "loss": 0.6582,
      "step": 34840
    },
    {
      "epoch": 14.27,
      "grad_norm": 2.567404270172119,
      "learning_rate": 4.68167384110811e-06,
      "loss": 0.6718,
      "step": 34850
    },
    {
      "epoch": 14.27,
      "grad_norm": 2.2572946548461914,
      "learning_rate": 4.681662657741986e-06,
      "loss": 0.6534,
      "step": 34860
    },
    {
      "epoch": 14.27,
      "grad_norm": 4.170894622802734,
      "learning_rate": 4.6816514636662475e-06,
      "loss": 0.6542,
      "step": 34870
    },
    {
      "epoch": 14.28,
      "grad_norm": 2.9049792289733887,
      "learning_rate": 4.681640258880944e-06,
      "loss": 0.648,
      "step": 34880
    },
    {
      "epoch": 14.28,
      "grad_norm": 1.7794837951660156,
      "learning_rate": 4.681629043386127e-06,
      "loss": 0.6706,
      "step": 34890
    },
    {
      "epoch": 14.29,
      "grad_norm": 2.3333306312561035,
      "learning_rate": 4.681617817181849e-06,
      "loss": 0.6358,
      "step": 34900
    },
    {
      "epoch": 14.29,
      "grad_norm": 2.6288836002349854,
      "learning_rate": 4.6816065802681596e-06,
      "loss": 0.6419,
      "step": 34910
    },
    {
      "epoch": 14.29,
      "grad_norm": 1.9248683452606201,
      "learning_rate": 4.681595332645112e-06,
      "loss": 0.66,
      "step": 34920
    },
    {
      "epoch": 14.3,
      "grad_norm": 2.201653242111206,
      "learning_rate": 4.6815840743127566e-06,
      "loss": 0.6401,
      "step": 34930
    },
    {
      "epoch": 14.3,
      "grad_norm": 2.6822285652160645,
      "learning_rate": 4.681572805271146e-06,
      "loss": 0.668,
      "step": 34940
    },
    {
      "epoch": 14.31,
      "grad_norm": 2.843212127685547,
      "learning_rate": 4.68156152552033e-06,
      "loss": 0.6546,
      "step": 34950
    },
    {
      "epoch": 14.31,
      "grad_norm": 2.6442081928253174,
      "learning_rate": 4.681550235060362e-06,
      "loss": 0.6271,
      "step": 34960
    },
    {
      "epoch": 14.31,
      "grad_norm": 2.29300856590271,
      "learning_rate": 4.681538933891294e-06,
      "loss": 0.637,
      "step": 34970
    },
    {
      "epoch": 14.32,
      "grad_norm": 2.3590197563171387,
      "learning_rate": 4.681527622013176e-06,
      "loss": 0.648,
      "step": 34980
    },
    {
      "epoch": 14.32,
      "grad_norm": 2.0335381031036377,
      "learning_rate": 4.681516299426061e-06,
      "loss": 0.6226,
      "step": 34990
    },
    {
      "epoch": 14.33,
      "grad_norm": 2.344862222671509,
      "learning_rate": 4.6815049661300005e-06,
      "loss": 0.6533,
      "step": 35000
    },
    {
      "epoch": 14.33,
      "grad_norm": 2.609060049057007,
      "learning_rate": 4.6814936221250466e-06,
      "loss": 0.6535,
      "step": 35010
    },
    {
      "epoch": 14.33,
      "grad_norm": 3.260382652282715,
      "learning_rate": 4.6814822674112516e-06,
      "loss": 0.6434,
      "step": 35020
    },
    {
      "epoch": 14.34,
      "grad_norm": 2.803299903869629,
      "learning_rate": 4.681470901988666e-06,
      "loss": 0.6513,
      "step": 35030
    },
    {
      "epoch": 14.34,
      "grad_norm": 3.155914068222046,
      "learning_rate": 4.681459525857344e-06,
      "loss": 0.6318,
      "step": 35040
    },
    {
      "epoch": 14.35,
      "grad_norm": 4.0265021324157715,
      "learning_rate": 4.681448139017337e-06,
      "loss": 0.6433,
      "step": 35050
    },
    {
      "epoch": 14.35,
      "grad_norm": 2.935347557067871,
      "learning_rate": 4.681436741468696e-06,
      "loss": 0.6497,
      "step": 35060
    },
    {
      "epoch": 14.36,
      "grad_norm": 4.599347114562988,
      "learning_rate": 4.6814253332114746e-06,
      "loss": 0.627,
      "step": 35070
    },
    {
      "epoch": 14.36,
      "grad_norm": 3.4731128215789795,
      "learning_rate": 4.681413914245724e-06,
      "loss": 0.6573,
      "step": 35080
    },
    {
      "epoch": 14.36,
      "grad_norm": 2.6818654537200928,
      "learning_rate": 4.681402484571498e-06,
      "loss": 0.6409,
      "step": 35090
    },
    {
      "epoch": 14.37,
      "grad_norm": 2.9041452407836914,
      "learning_rate": 4.681391044188848e-06,
      "loss": 0.638,
      "step": 35100
    },
    {
      "epoch": 14.37,
      "grad_norm": 2.4370267391204834,
      "learning_rate": 4.681379593097825e-06,
      "loss": 0.6292,
      "step": 35110
    },
    {
      "epoch": 14.38,
      "grad_norm": 2.8333911895751953,
      "learning_rate": 4.681368131298484e-06,
      "loss": 0.6513,
      "step": 35120
    },
    {
      "epoch": 14.38,
      "grad_norm": 2.47269868850708,
      "learning_rate": 4.681356658790875e-06,
      "loss": 0.661,
      "step": 35130
    },
    {
      "epoch": 14.38,
      "grad_norm": 2.6451475620269775,
      "learning_rate": 4.681345175575054e-06,
      "loss": 0.6482,
      "step": 35140
    },
    {
      "epoch": 14.39,
      "grad_norm": 3.4428458213806152,
      "learning_rate": 4.68133368165107e-06,
      "loss": 0.6428,
      "step": 35150
    },
    {
      "epoch": 14.39,
      "grad_norm": 3.252986431121826,
      "learning_rate": 4.681322177018978e-06,
      "loss": 0.6592,
      "step": 35160
    },
    {
      "epoch": 14.4,
      "grad_norm": 3.245223045349121,
      "learning_rate": 4.681310661678829e-06,
      "loss": 0.6223,
      "step": 35170
    },
    {
      "epoch": 14.4,
      "grad_norm": 2.3761348724365234,
      "learning_rate": 4.681299135630677e-06,
      "loss": 0.6441,
      "step": 35180
    },
    {
      "epoch": 14.4,
      "grad_norm": 2.2636401653289795,
      "learning_rate": 4.681287598874575e-06,
      "loss": 0.6512,
      "step": 35190
    },
    {
      "epoch": 14.41,
      "grad_norm": 4.962052345275879,
      "learning_rate": 4.681276051410575e-06,
      "loss": 0.6381,
      "step": 35200
    },
    {
      "epoch": 14.41,
      "grad_norm": 2.569711685180664,
      "learning_rate": 4.6812644932387305e-06,
      "loss": 0.6586,
      "step": 35210
    },
    {
      "epoch": 14.42,
      "grad_norm": 2.19870924949646,
      "learning_rate": 4.681252924359092e-06,
      "loss": 0.6242,
      "step": 35220
    },
    {
      "epoch": 14.42,
      "grad_norm": 2.780505418777466,
      "learning_rate": 4.6812413447717175e-06,
      "loss": 0.6491,
      "step": 35230
    },
    {
      "epoch": 14.42,
      "grad_norm": 2.2342777252197266,
      "learning_rate": 4.681229754476656e-06,
      "loss": 0.6636,
      "step": 35240
    },
    {
      "epoch": 14.43,
      "grad_norm": 2.157261848449707,
      "learning_rate": 4.681218153473961e-06,
      "loss": 0.6511,
      "step": 35250
    },
    {
      "epoch": 14.43,
      "grad_norm": 2.5384209156036377,
      "learning_rate": 4.6812065417636875e-06,
      "loss": 0.6428,
      "step": 35260
    },
    {
      "epoch": 14.44,
      "grad_norm": 2.069634437561035,
      "learning_rate": 4.681194919345887e-06,
      "loss": 0.6356,
      "step": 35270
    },
    {
      "epoch": 14.44,
      "grad_norm": 3.2153077125549316,
      "learning_rate": 4.681183286220614e-06,
      "loss": 0.6553,
      "step": 35280
    },
    {
      "epoch": 14.45,
      "grad_norm": 2.418431282043457,
      "learning_rate": 4.68117164238792e-06,
      "loss": 0.6589,
      "step": 35290
    },
    {
      "epoch": 14.45,
      "grad_norm": 2.8493711948394775,
      "learning_rate": 4.6811599878478605e-06,
      "loss": 0.6484,
      "step": 35300
    },
    {
      "epoch": 14.45,
      "grad_norm": 1.7294400930404663,
      "learning_rate": 4.681148322600487e-06,
      "loss": 0.646,
      "step": 35310
    },
    {
      "epoch": 14.46,
      "grad_norm": 3.088406562805176,
      "learning_rate": 4.681136646645854e-06,
      "loss": 0.6485,
      "step": 35320
    },
    {
      "epoch": 14.46,
      "grad_norm": 2.0081660747528076,
      "learning_rate": 4.6811249599840155e-06,
      "loss": 0.6555,
      "step": 35330
    },
    {
      "epoch": 14.47,
      "grad_norm": 2.315141439437866,
      "learning_rate": 4.681113262615023e-06,
      "loss": 0.6567,
      "step": 35340
    },
    {
      "epoch": 14.47,
      "grad_norm": 3.163759708404541,
      "learning_rate": 4.681101554538932e-06,
      "loss": 0.6556,
      "step": 35350
    },
    {
      "epoch": 14.47,
      "grad_norm": 3.262887716293335,
      "learning_rate": 4.681089835755795e-06,
      "loss": 0.6508,
      "step": 35360
    },
    {
      "epoch": 14.48,
      "grad_norm": 2.6541616916656494,
      "learning_rate": 4.681078106265668e-06,
      "loss": 0.6537,
      "step": 35370
    },
    {
      "epoch": 14.48,
      "grad_norm": 2.73645281791687,
      "learning_rate": 4.681066366068601e-06,
      "loss": 0.654,
      "step": 35380
    },
    {
      "epoch": 14.49,
      "grad_norm": 2.6658966541290283,
      "learning_rate": 4.681054615164651e-06,
      "loss": 0.645,
      "step": 35390
    },
    {
      "epoch": 14.49,
      "grad_norm": 2.777017831802368,
      "learning_rate": 4.68104285355387e-06,
      "loss": 0.6629,
      "step": 35400
    },
    {
      "epoch": 14.49,
      "grad_norm": 2.19930100440979,
      "learning_rate": 4.681031081236312e-06,
      "loss": 0.6555,
      "step": 35410
    },
    {
      "epoch": 14.5,
      "grad_norm": 2.4320080280303955,
      "learning_rate": 4.681019298212032e-06,
      "loss": 0.6304,
      "step": 35420
    },
    {
      "epoch": 14.5,
      "grad_norm": 1.5679337978363037,
      "learning_rate": 4.681007504481082e-06,
      "loss": 0.6499,
      "step": 35430
    },
    {
      "epoch": 14.51,
      "grad_norm": 2.735349178314209,
      "learning_rate": 4.6809957000435195e-06,
      "loss": 0.6622,
      "step": 35440
    },
    {
      "epoch": 14.51,
      "grad_norm": 2.963207960128784,
      "learning_rate": 4.680983884899395e-06,
      "loss": 0.635,
      "step": 35450
    },
    {
      "epoch": 14.51,
      "grad_norm": 3.5011470317840576,
      "learning_rate": 4.680972059048764e-06,
      "loss": 0.6374,
      "step": 35460
    },
    {
      "epoch": 14.52,
      "grad_norm": 3.349263906478882,
      "learning_rate": 4.680960222491681e-06,
      "loss": 0.6603,
      "step": 35470
    },
    {
      "epoch": 14.52,
      "grad_norm": 2.0512049198150635,
      "learning_rate": 4.6809483752282e-06,
      "loss": 0.6513,
      "step": 35480
    },
    {
      "epoch": 14.53,
      "grad_norm": 2.2612357139587402,
      "learning_rate": 4.680936517258375e-06,
      "loss": 0.6656,
      "step": 35490
    },
    {
      "epoch": 14.53,
      "grad_norm": 4.06624698638916,
      "learning_rate": 4.680924648582261e-06,
      "loss": 0.6194,
      "step": 35500
    },
    {
      "epoch": 14.54,
      "grad_norm": 2.4576187133789062,
      "learning_rate": 4.680912769199911e-06,
      "loss": 0.6601,
      "step": 35510
    },
    {
      "epoch": 14.54,
      "grad_norm": 2.713346004486084,
      "learning_rate": 4.680900879111381e-06,
      "loss": 0.6626,
      "step": 35520
    },
    {
      "epoch": 14.54,
      "grad_norm": 2.559565544128418,
      "learning_rate": 4.680888978316725e-06,
      "loss": 0.6669,
      "step": 35530
    },
    {
      "epoch": 14.55,
      "grad_norm": 4.188915729522705,
      "learning_rate": 4.680877066815997e-06,
      "loss": 0.6544,
      "step": 35540
    },
    {
      "epoch": 14.55,
      "grad_norm": 2.067352056503296,
      "learning_rate": 4.6808651446092514e-06,
      "loss": 0.6654,
      "step": 35550
    },
    {
      "epoch": 14.56,
      "grad_norm": 3.17819881439209,
      "learning_rate": 4.680853211696544e-06,
      "loss": 0.6403,
      "step": 35560
    },
    {
      "epoch": 14.56,
      "grad_norm": 3.228236198425293,
      "learning_rate": 4.680841268077929e-06,
      "loss": 0.6483,
      "step": 35570
    },
    {
      "epoch": 14.56,
      "grad_norm": 2.454573631286621,
      "learning_rate": 4.68082931375346e-06,
      "loss": 0.6391,
      "step": 35580
    },
    {
      "epoch": 14.57,
      "grad_norm": 2.1698086261749268,
      "learning_rate": 4.6808173487231925e-06,
      "loss": 0.6498,
      "step": 35590
    },
    {
      "epoch": 14.57,
      "grad_norm": 2.044499158859253,
      "learning_rate": 4.680805372987182e-06,
      "loss": 0.6346,
      "step": 35600
    },
    {
      "epoch": 14.58,
      "grad_norm": 3.6990504264831543,
      "learning_rate": 4.680793386545482e-06,
      "loss": 0.6496,
      "step": 35610
    },
    {
      "epoch": 14.58,
      "grad_norm": 2.265289783477783,
      "learning_rate": 4.680781389398149e-06,
      "loss": 0.6414,
      "step": 35620
    },
    {
      "epoch": 14.58,
      "grad_norm": 3.1002933979034424,
      "learning_rate": 4.680769381545237e-06,
      "loss": 0.6589,
      "step": 35630
    },
    {
      "epoch": 14.59,
      "grad_norm": 2.2903261184692383,
      "learning_rate": 4.680757362986801e-06,
      "loss": 0.6674,
      "step": 35640
    },
    {
      "epoch": 14.59,
      "grad_norm": 2.346153974533081,
      "learning_rate": 4.680745333722896e-06,
      "loss": 0.6434,
      "step": 35650
    },
    {
      "epoch": 14.6,
      "grad_norm": 2.4271907806396484,
      "learning_rate": 4.6807332937535776e-06,
      "loss": 0.6566,
      "step": 35660
    },
    {
      "epoch": 14.6,
      "grad_norm": 2.6505916118621826,
      "learning_rate": 4.6807212430789e-06,
      "loss": 0.6523,
      "step": 35670
    },
    {
      "epoch": 14.6,
      "grad_norm": 2.85648512840271,
      "learning_rate": 4.68070918169892e-06,
      "loss": 0.6197,
      "step": 35680
    },
    {
      "epoch": 14.61,
      "grad_norm": 4.168972492218018,
      "learning_rate": 4.680697109613692e-06,
      "loss": 0.6505,
      "step": 35690
    },
    {
      "epoch": 14.61,
      "grad_norm": 3.1736748218536377,
      "learning_rate": 4.680685026823271e-06,
      "loss": 0.6419,
      "step": 35700
    },
    {
      "epoch": 14.62,
      "grad_norm": 2.7329251766204834,
      "learning_rate": 4.680672933327712e-06,
      "loss": 0.6435,
      "step": 35710
    },
    {
      "epoch": 14.62,
      "grad_norm": 2.6328179836273193,
      "learning_rate": 4.680660829127072e-06,
      "loss": 0.6352,
      "step": 35720
    },
    {
      "epoch": 14.63,
      "grad_norm": 2.6504528522491455,
      "learning_rate": 4.680648714221404e-06,
      "loss": 0.6391,
      "step": 35730
    },
    {
      "epoch": 14.63,
      "grad_norm": 2.4992265701293945,
      "learning_rate": 4.6806365886107665e-06,
      "loss": 0.6624,
      "step": 35740
    },
    {
      "epoch": 14.63,
      "grad_norm": 2.191312551498413,
      "learning_rate": 4.680624452295211e-06,
      "loss": 0.6401,
      "step": 35750
    },
    {
      "epoch": 14.64,
      "grad_norm": 2.3341331481933594,
      "learning_rate": 4.6806123052747984e-06,
      "loss": 0.633,
      "step": 35760
    },
    {
      "epoch": 14.64,
      "grad_norm": 5.860114574432373,
      "learning_rate": 4.68060014754958e-06,
      "loss": 0.6424,
      "step": 35770
    },
    {
      "epoch": 14.65,
      "grad_norm": 3.38424015045166,
      "learning_rate": 4.680587979119613e-06,
      "loss": 0.6408,
      "step": 35780
    },
    {
      "epoch": 14.65,
      "grad_norm": 2.3662109375,
      "learning_rate": 4.6805757999849525e-06,
      "loss": 0.6534,
      "step": 35790
    },
    {
      "epoch": 14.65,
      "grad_norm": 2.3070101737976074,
      "learning_rate": 4.680563610145656e-06,
      "loss": 0.6635,
      "step": 35800
    },
    {
      "epoch": 14.66,
      "grad_norm": 2.979036331176758,
      "learning_rate": 4.680551409601778e-06,
      "loss": 0.6507,
      "step": 35810
    },
    {
      "epoch": 14.66,
      "grad_norm": 3.0694689750671387,
      "learning_rate": 4.680539198353375e-06,
      "loss": 0.6582,
      "step": 35820
    },
    {
      "epoch": 14.67,
      "grad_norm": 3.33314847946167,
      "learning_rate": 4.6805269764005025e-06,
      "loss": 0.645,
      "step": 35830
    },
    {
      "epoch": 14.67,
      "grad_norm": 2.7218434810638428,
      "learning_rate": 4.680514743743216e-06,
      "loss": 0.6487,
      "step": 35840
    },
    {
      "epoch": 14.67,
      "grad_norm": 3.807948350906372,
      "learning_rate": 4.680502500381572e-06,
      "loss": 0.6397,
      "step": 35850
    },
    {
      "epoch": 14.68,
      "grad_norm": 2.5786056518554688,
      "learning_rate": 4.680490246315627e-06,
      "loss": 0.645,
      "step": 35860
    },
    {
      "epoch": 14.68,
      "grad_norm": 2.2762980461120605,
      "learning_rate": 4.6804779815454375e-06,
      "loss": 0.6739,
      "step": 35870
    },
    {
      "epoch": 14.69,
      "grad_norm": 3.320650339126587,
      "learning_rate": 4.680465706071058e-06,
      "loss": 0.6241,
      "step": 35880
    },
    {
      "epoch": 14.69,
      "grad_norm": 3.3023667335510254,
      "learning_rate": 4.680453419892546e-06,
      "loss": 0.6395,
      "step": 35890
    },
    {
      "epoch": 14.7,
      "grad_norm": 3.645552158355713,
      "learning_rate": 4.680441123009958e-06,
      "loss": 0.659,
      "step": 35900
    },
    {
      "epoch": 14.7,
      "grad_norm": 2.1049866676330566,
      "learning_rate": 4.680428815423349e-06,
      "loss": 0.645,
      "step": 35910
    },
    {
      "epoch": 14.7,
      "grad_norm": 3.540074348449707,
      "learning_rate": 4.680416497132777e-06,
      "loss": 0.6465,
      "step": 35920
    },
    {
      "epoch": 14.71,
      "grad_norm": 2.4521875381469727,
      "learning_rate": 4.680404168138297e-06,
      "loss": 0.6453,
      "step": 35930
    },
    {
      "epoch": 14.71,
      "grad_norm": 3.7289230823516846,
      "learning_rate": 4.680391828439966e-06,
      "loss": 0.658,
      "step": 35940
    },
    {
      "epoch": 14.72,
      "grad_norm": 3.0376927852630615,
      "learning_rate": 4.680379478037842e-06,
      "loss": 0.633,
      "step": 35950
    },
    {
      "epoch": 14.72,
      "grad_norm": 3.2328803539276123,
      "learning_rate": 4.680367116931979e-06,
      "loss": 0.6419,
      "step": 35960
    },
    {
      "epoch": 14.72,
      "grad_norm": 2.321375608444214,
      "learning_rate": 4.680354745122436e-06,
      "loss": 0.6442,
      "step": 35970
    },
    {
      "epoch": 14.73,
      "grad_norm": 2.6497485637664795,
      "learning_rate": 4.6803423626092674e-06,
      "loss": 0.6295,
      "step": 35980
    },
    {
      "epoch": 14.73,
      "grad_norm": 2.7506556510925293,
      "learning_rate": 4.680329969392532e-06,
      "loss": 0.6543,
      "step": 35990
    },
    {
      "epoch": 14.74,
      "grad_norm": 2.012561321258545,
      "learning_rate": 4.680317565472284e-06,
      "loss": 0.6654,
      "step": 36000
    },
    {
      "epoch": 14.74,
      "grad_norm": 2.0483994483947754,
      "learning_rate": 4.680305150848583e-06,
      "loss": 0.639,
      "step": 36010
    },
    {
      "epoch": 14.74,
      "grad_norm": 2.593186378479004,
      "learning_rate": 4.680292725521485e-06,
      "loss": 0.6329,
      "step": 36020
    },
    {
      "epoch": 14.75,
      "grad_norm": 2.228065013885498,
      "learning_rate": 4.680280289491046e-06,
      "loss": 0.6381,
      "step": 36030
    },
    {
      "epoch": 14.75,
      "grad_norm": 2.639357089996338,
      "learning_rate": 4.680267842757324e-06,
      "loss": 0.6407,
      "step": 36040
    },
    {
      "epoch": 14.76,
      "grad_norm": 2.3455917835235596,
      "learning_rate": 4.680255385320375e-06,
      "loss": 0.6361,
      "step": 36050
    },
    {
      "epoch": 14.76,
      "grad_norm": 2.4800124168395996,
      "learning_rate": 4.680242917180258e-06,
      "loss": 0.6487,
      "step": 36060
    },
    {
      "epoch": 14.76,
      "grad_norm": 2.9227263927459717,
      "learning_rate": 4.680230438337027e-06,
      "loss": 0.6395,
      "step": 36070
    },
    {
      "epoch": 14.77,
      "grad_norm": 2.5717883110046387,
      "learning_rate": 4.6802179487907425e-06,
      "loss": 0.6399,
      "step": 36080
    },
    {
      "epoch": 14.77,
      "grad_norm": 2.2143020629882812,
      "learning_rate": 4.68020544854146e-06,
      "loss": 0.6186,
      "step": 36090
    },
    {
      "epoch": 14.78,
      "grad_norm": 2.528214693069458,
      "learning_rate": 4.680192937589236e-06,
      "loss": 0.6273,
      "step": 36100
    },
    {
      "epoch": 14.78,
      "grad_norm": 2.622649908065796,
      "learning_rate": 4.680180415934129e-06,
      "loss": 0.6373,
      "step": 36110
    },
    {
      "epoch": 14.79,
      "grad_norm": 2.6190953254699707,
      "learning_rate": 4.680167883576197e-06,
      "loss": 0.6683,
      "step": 36120
    },
    {
      "epoch": 14.79,
      "grad_norm": 3.141246795654297,
      "learning_rate": 4.680155340515496e-06,
      "loss": 0.6532,
      "step": 36130
    },
    {
      "epoch": 14.79,
      "grad_norm": 2.3182358741760254,
      "learning_rate": 4.680142786752083e-06,
      "loss": 0.6336,
      "step": 36140
    },
    {
      "epoch": 14.8,
      "grad_norm": 2.035496950149536,
      "learning_rate": 4.680130222286018e-06,
      "loss": 0.6443,
      "step": 36150
    },
    {
      "epoch": 14.8,
      "grad_norm": 2.2022974491119385,
      "learning_rate": 4.680117647117356e-06,
      "loss": 0.6296,
      "step": 36160
    },
    {
      "epoch": 14.81,
      "grad_norm": 2.4725139141082764,
      "learning_rate": 4.680105061246156e-06,
      "loss": 0.6494,
      "step": 36170
    },
    {
      "epoch": 14.81,
      "grad_norm": 3.2038795948028564,
      "learning_rate": 4.680092464672476e-06,
      "loss": 0.6592,
      "step": 36180
    },
    {
      "epoch": 14.81,
      "grad_norm": 3.213621139526367,
      "learning_rate": 4.680079857396371e-06,
      "loss": 0.6492,
      "step": 36190
    },
    {
      "epoch": 14.82,
      "grad_norm": 2.021979570388794,
      "learning_rate": 4.680067239417903e-06,
      "loss": 0.6367,
      "step": 36200
    },
    {
      "epoch": 14.82,
      "grad_norm": 2.928845167160034,
      "learning_rate": 4.680054610737127e-06,
      "loss": 0.6471,
      "step": 36210
    },
    {
      "epoch": 14.83,
      "grad_norm": 2.599571466445923,
      "learning_rate": 4.6800419713540996e-06,
      "loss": 0.6406,
      "step": 36220
    },
    {
      "epoch": 14.83,
      "grad_norm": 2.4462664127349854,
      "learning_rate": 4.680029321268882e-06,
      "loss": 0.617,
      "step": 36230
    },
    {
      "epoch": 14.83,
      "grad_norm": 2.924823522567749,
      "learning_rate": 4.680016660481531e-06,
      "loss": 0.6615,
      "step": 36240
    },
    {
      "epoch": 14.84,
      "grad_norm": 2.8672115802764893,
      "learning_rate": 4.680003988992104e-06,
      "loss": 0.6722,
      "step": 36250
    },
    {
      "epoch": 14.84,
      "grad_norm": 3.0188329219818115,
      "learning_rate": 4.679991306800658e-06,
      "loss": 0.6451,
      "step": 36260
    },
    {
      "epoch": 14.85,
      "grad_norm": 3.1395673751831055,
      "learning_rate": 4.679978613907255e-06,
      "loss": 0.679,
      "step": 36270
    },
    {
      "epoch": 14.85,
      "grad_norm": 2.1710164546966553,
      "learning_rate": 4.679965910311948e-06,
      "loss": 0.6414,
      "step": 36280
    },
    {
      "epoch": 14.85,
      "grad_norm": 3.824141502380371,
      "learning_rate": 4.6799531960147995e-06,
      "loss": 0.6435,
      "step": 36290
    },
    {
      "epoch": 14.86,
      "grad_norm": 3.566154718399048,
      "learning_rate": 4.679940471015865e-06,
      "loss": 0.6398,
      "step": 36300
    },
    {
      "epoch": 14.86,
      "grad_norm": 2.6624159812927246,
      "learning_rate": 4.679927735315204e-06,
      "loss": 0.6568,
      "step": 36310
    },
    {
      "epoch": 14.87,
      "grad_norm": 2.708275556564331,
      "learning_rate": 4.679914988912874e-06,
      "loss": 0.6421,
      "step": 36320
    },
    {
      "epoch": 14.87,
      "grad_norm": 2.5683906078338623,
      "learning_rate": 4.679902231808934e-06,
      "loss": 0.6507,
      "step": 36330
    },
    {
      "epoch": 14.88,
      "grad_norm": 1.8510404825210571,
      "learning_rate": 4.679889464003444e-06,
      "loss": 0.6591,
      "step": 36340
    },
    {
      "epoch": 14.88,
      "grad_norm": 2.7744596004486084,
      "learning_rate": 4.6798766854964595e-06,
      "loss": 0.6295,
      "step": 36350
    },
    {
      "epoch": 14.88,
      "grad_norm": 2.490017890930176,
      "learning_rate": 4.6798638962880405e-06,
      "loss": 0.6361,
      "step": 36360
    },
    {
      "epoch": 14.89,
      "grad_norm": 2.551062822341919,
      "learning_rate": 4.679851096378246e-06,
      "loss": 0.6801,
      "step": 36370
    },
    {
      "epoch": 14.89,
      "grad_norm": 3.4952151775360107,
      "learning_rate": 4.679838285767134e-06,
      "loss": 0.6579,
      "step": 36380
    },
    {
      "epoch": 14.9,
      "grad_norm": 2.5512828826904297,
      "learning_rate": 4.679825464454763e-06,
      "loss": 0.642,
      "step": 36390
    },
    {
      "epoch": 14.9,
      "grad_norm": 2.858006000518799,
      "learning_rate": 4.679812632441193e-06,
      "loss": 0.6546,
      "step": 36400
    },
    {
      "epoch": 14.9,
      "grad_norm": 2.8193564414978027,
      "learning_rate": 4.679799789726481e-06,
      "loss": 0.6444,
      "step": 36410
    },
    {
      "epoch": 14.91,
      "grad_norm": 2.736570358276367,
      "learning_rate": 4.679786936310687e-06,
      "loss": 0.6347,
      "step": 36420
    },
    {
      "epoch": 14.91,
      "grad_norm": 2.782397747039795,
      "learning_rate": 4.67977407219387e-06,
      "loss": 0.6561,
      "step": 36430
    },
    {
      "epoch": 14.92,
      "grad_norm": 2.2507262229919434,
      "learning_rate": 4.679761197376088e-06,
      "loss": 0.6528,
      "step": 36440
    },
    {
      "epoch": 14.92,
      "grad_norm": 2.6802890300750732,
      "learning_rate": 4.6797483118574005e-06,
      "loss": 0.6401,
      "step": 36450
    },
    {
      "epoch": 14.92,
      "grad_norm": 2.349867582321167,
      "learning_rate": 4.679735415637867e-06,
      "loss": 0.6076,
      "step": 36460
    },
    {
      "epoch": 14.93,
      "grad_norm": 2.9843921661376953,
      "learning_rate": 4.6797225087175455e-06,
      "loss": 0.6395,
      "step": 36470
    },
    {
      "epoch": 14.93,
      "grad_norm": 2.0014138221740723,
      "learning_rate": 4.679709591096496e-06,
      "loss": 0.6449,
      "step": 36480
    },
    {
      "epoch": 14.94,
      "grad_norm": 2.6121132373809814,
      "learning_rate": 4.679696662774777e-06,
      "loss": 0.6442,
      "step": 36490
    },
    {
      "epoch": 14.94,
      "grad_norm": 2.2840330600738525,
      "learning_rate": 4.679683723752448e-06,
      "loss": 0.6242,
      "step": 36500
    },
    {
      "epoch": 14.94,
      "grad_norm": 3.4731364250183105,
      "learning_rate": 4.6796707740295686e-06,
      "loss": 0.643,
      "step": 36510
    },
    {
      "epoch": 14.95,
      "grad_norm": 2.2348451614379883,
      "learning_rate": 4.679657813606199e-06,
      "loss": 0.6453,
      "step": 36520
    },
    {
      "epoch": 14.95,
      "grad_norm": 2.5833959579467773,
      "learning_rate": 4.679644842482396e-06,
      "loss": 0.644,
      "step": 36530
    },
    {
      "epoch": 14.96,
      "grad_norm": 3.203850746154785,
      "learning_rate": 4.679631860658221e-06,
      "loss": 0.6511,
      "step": 36540
    },
    {
      "epoch": 14.96,
      "grad_norm": 2.1786184310913086,
      "learning_rate": 4.679618868133733e-06,
      "loss": 0.6526,
      "step": 36550
    },
    {
      "epoch": 14.97,
      "grad_norm": 4.245147705078125,
      "learning_rate": 4.679605864908992e-06,
      "loss": 0.6538,
      "step": 36560
    },
    {
      "epoch": 14.97,
      "grad_norm": 2.189160108566284,
      "learning_rate": 4.679592850984056e-06,
      "loss": 0.6488,
      "step": 36570
    },
    {
      "epoch": 14.97,
      "grad_norm": 2.2348644733428955,
      "learning_rate": 4.679579826358986e-06,
      "loss": 0.6573,
      "step": 36580
    },
    {
      "epoch": 14.98,
      "grad_norm": 3.1479709148406982,
      "learning_rate": 4.679566791033841e-06,
      "loss": 0.6256,
      "step": 36590
    },
    {
      "epoch": 14.98,
      "grad_norm": 2.9433555603027344,
      "learning_rate": 4.6795537450086815e-06,
      "loss": 0.6469,
      "step": 36600
    },
    {
      "epoch": 14.99,
      "grad_norm": 2.8420803546905518,
      "learning_rate": 4.679540688283567e-06,
      "loss": 0.6166,
      "step": 36610
    },
    {
      "epoch": 14.99,
      "grad_norm": 2.422976016998291,
      "learning_rate": 4.6795276208585565e-06,
      "loss": 0.6489,
      "step": 36620
    },
    {
      "epoch": 14.99,
      "grad_norm": 2.552764892578125,
      "learning_rate": 4.679514542733711e-06,
      "loss": 0.6314,
      "step": 36630
    },
    {
      "epoch": 15.0,
      "grad_norm": 2.214667558670044,
      "learning_rate": 4.67950145390909e-06,
      "loss": 0.6389,
      "step": 36640
    },
    {
      "epoch": 15.0,
      "eval_loss": 0.6462235450744629,
      "eval_runtime": 52.2813,
      "eval_samples_per_second": 65.97,
      "eval_steps_per_second": 8.263,
      "step": 36645
    },
    {
      "epoch": 15.0,
      "grad_norm": 2.9350345134735107,
      "learning_rate": 4.6794883543847534e-06,
      "loss": 0.6505,
      "step": 36650
    },
    {
      "epoch": 15.01,
      "grad_norm": 2.7219526767730713,
      "learning_rate": 4.67947524416076e-06,
      "loss": 0.6523,
      "step": 36660
    },
    {
      "epoch": 15.01,
      "grad_norm": 2.7301502227783203,
      "learning_rate": 4.679462123237172e-06,
      "loss": 0.6654,
      "step": 36670
    },
    {
      "epoch": 15.01,
      "grad_norm": 2.303694009780884,
      "learning_rate": 4.679448991614048e-06,
      "loss": 0.6247,
      "step": 36680
    },
    {
      "epoch": 15.02,
      "grad_norm": 2.346456527709961,
      "learning_rate": 4.679435849291449e-06,
      "loss": 0.6459,
      "step": 36690
    },
    {
      "epoch": 15.02,
      "grad_norm": 2.264143943786621,
      "learning_rate": 4.679422696269434e-06,
      "loss": 0.637,
      "step": 36700
    },
    {
      "epoch": 15.03,
      "grad_norm": 2.5736851692199707,
      "learning_rate": 4.679409532548065e-06,
      "loss": 0.6251,
      "step": 36710
    },
    {
      "epoch": 15.03,
      "grad_norm": 3.3104782104492188,
      "learning_rate": 4.6793963581274015e-06,
      "loss": 0.6275,
      "step": 36720
    },
    {
      "epoch": 15.03,
      "grad_norm": 3.290489912033081,
      "learning_rate": 4.679383173007503e-06,
      "loss": 0.6288,
      "step": 36730
    },
    {
      "epoch": 15.04,
      "grad_norm": 2.3405392169952393,
      "learning_rate": 4.679369977188432e-06,
      "loss": 0.6273,
      "step": 36740
    },
    {
      "epoch": 15.04,
      "grad_norm": 2.5705978870391846,
      "learning_rate": 4.679356770670246e-06,
      "loss": 0.6379,
      "step": 36750
    },
    {
      "epoch": 15.05,
      "grad_norm": 3.2942399978637695,
      "learning_rate": 4.679343553453008e-06,
      "loss": 0.6571,
      "step": 36760
    },
    {
      "epoch": 15.05,
      "grad_norm": 2.665719747543335,
      "learning_rate": 4.679330325536778e-06,
      "loss": 0.6486,
      "step": 36770
    },
    {
      "epoch": 15.06,
      "grad_norm": 3.74979829788208,
      "learning_rate": 4.679317086921616e-06,
      "loss": 0.6644,
      "step": 36780
    },
    {
      "epoch": 15.06,
      "grad_norm": 2.3158562183380127,
      "learning_rate": 4.679303837607582e-06,
      "loss": 0.6578,
      "step": 36790
    },
    {
      "epoch": 15.06,
      "grad_norm": 1.6398507356643677,
      "learning_rate": 4.6792905775947385e-06,
      "loss": 0.637,
      "step": 36800
    },
    {
      "epoch": 15.07,
      "grad_norm": 2.7343099117279053,
      "learning_rate": 4.679277306883144e-06,
      "loss": 0.6072,
      "step": 36810
    },
    {
      "epoch": 15.07,
      "grad_norm": 2.4265103340148926,
      "learning_rate": 4.679264025472862e-06,
      "loss": 0.6251,
      "step": 36820
    },
    {
      "epoch": 15.08,
      "grad_norm": 3.335085153579712,
      "learning_rate": 4.6792507333639516e-06,
      "loss": 0.6335,
      "step": 36830
    },
    {
      "epoch": 15.08,
      "grad_norm": 3.4933652877807617,
      "learning_rate": 4.679237430556474e-06,
      "loss": 0.6244,
      "step": 36840
    },
    {
      "epoch": 15.08,
      "grad_norm": 1.9773304462432861,
      "learning_rate": 4.67922411705049e-06,
      "loss": 0.6303,
      "step": 36850
    },
    {
      "epoch": 15.09,
      "grad_norm": 3.9593305587768555,
      "learning_rate": 4.679210792846061e-06,
      "loss": 0.6219,
      "step": 36860
    },
    {
      "epoch": 15.09,
      "grad_norm": 2.8605427742004395,
      "learning_rate": 4.679197457943247e-06,
      "loss": 0.6374,
      "step": 36870
    },
    {
      "epoch": 15.1,
      "grad_norm": 2.442253828048706,
      "learning_rate": 4.6791841123421115e-06,
      "loss": 0.6746,
      "step": 36880
    },
    {
      "epoch": 15.1,
      "grad_norm": 2.1548354625701904,
      "learning_rate": 4.679170756042713e-06,
      "loss": 0.658,
      "step": 36890
    },
    {
      "epoch": 15.1,
      "grad_norm": 2.8559603691101074,
      "learning_rate": 4.679157389045113e-06,
      "loss": 0.6329,
      "step": 36900
    },
    {
      "epoch": 15.11,
      "grad_norm": 2.7055695056915283,
      "learning_rate": 4.679144011349375e-06,
      "loss": 0.6425,
      "step": 36910
    },
    {
      "epoch": 15.11,
      "grad_norm": 2.1821401119232178,
      "learning_rate": 4.6791306229555575e-06,
      "loss": 0.6467,
      "step": 36920
    },
    {
      "epoch": 15.12,
      "grad_norm": 2.436379909515381,
      "learning_rate": 4.679117223863724e-06,
      "loss": 0.6464,
      "step": 36930
    },
    {
      "epoch": 15.12,
      "grad_norm": 1.8587702512741089,
      "learning_rate": 4.679103814073934e-06,
      "loss": 0.6504,
      "step": 36940
    },
    {
      "epoch": 15.12,
      "grad_norm": 2.425185441970825,
      "learning_rate": 4.679090393586251e-06,
      "loss": 0.65,
      "step": 36950
    },
    {
      "epoch": 15.13,
      "grad_norm": 1.8428348302841187,
      "learning_rate": 4.6790769624007346e-06,
      "loss": 0.6324,
      "step": 36960
    },
    {
      "epoch": 15.13,
      "grad_norm": 2.9639689922332764,
      "learning_rate": 4.679063520517447e-06,
      "loss": 0.6386,
      "step": 36970
    },
    {
      "epoch": 15.14,
      "grad_norm": 2.2485358715057373,
      "learning_rate": 4.679050067936449e-06,
      "loss": 0.6375,
      "step": 36980
    },
    {
      "epoch": 15.14,
      "grad_norm": 2.4355671405792236,
      "learning_rate": 4.679036604657805e-06,
      "loss": 0.6236,
      "step": 36990
    },
    {
      "epoch": 15.15,
      "grad_norm": 2.3844046592712402,
      "learning_rate": 4.679023130681573e-06,
      "loss": 0.6474,
      "step": 37000
    },
    {
      "epoch": 15.15,
      "grad_norm": 2.554058074951172,
      "learning_rate": 4.679009646007817e-06,
      "loss": 0.6419,
      "step": 37010
    },
    {
      "epoch": 15.15,
      "grad_norm": 2.363457202911377,
      "learning_rate": 4.678996150636599e-06,
      "loss": 0.6419,
      "step": 37020
    },
    {
      "epoch": 15.16,
      "grad_norm": 2.3852646350860596,
      "learning_rate": 4.678982644567979e-06,
      "loss": 0.6198,
      "step": 37030
    },
    {
      "epoch": 15.16,
      "grad_norm": 2.042476177215576,
      "learning_rate": 4.678969127802021e-06,
      "loss": 0.6615,
      "step": 37040
    },
    {
      "epoch": 15.17,
      "grad_norm": 2.3960652351379395,
      "learning_rate": 4.678955600338785e-06,
      "loss": 0.6609,
      "step": 37050
    },
    {
      "epoch": 15.17,
      "grad_norm": 2.4129226207733154,
      "learning_rate": 4.678942062178334e-06,
      "loss": 0.6395,
      "step": 37060
    },
    {
      "epoch": 15.17,
      "grad_norm": 3.346850872039795,
      "learning_rate": 4.678928513320729e-06,
      "loss": 0.627,
      "step": 37070
    },
    {
      "epoch": 15.18,
      "grad_norm": 3.3310763835906982,
      "learning_rate": 4.678914953766034e-06,
      "loss": 0.6344,
      "step": 37080
    },
    {
      "epoch": 15.18,
      "grad_norm": 2.258216142654419,
      "learning_rate": 4.67890138351431e-06,
      "loss": 0.6472,
      "step": 37090
    },
    {
      "epoch": 15.19,
      "grad_norm": 2.1631669998168945,
      "learning_rate": 4.678887802565619e-06,
      "loss": 0.6445,
      "step": 37100
    },
    {
      "epoch": 15.19,
      "grad_norm": 2.348505973815918,
      "learning_rate": 4.678874210920022e-06,
      "loss": 0.6525,
      "step": 37110
    },
    {
      "epoch": 15.19,
      "grad_norm": 2.8831119537353516,
      "learning_rate": 4.678860608577584e-06,
      "loss": 0.6524,
      "step": 37120
    },
    {
      "epoch": 15.2,
      "grad_norm": 2.2034366130828857,
      "learning_rate": 4.678846995538366e-06,
      "loss": 0.6342,
      "step": 37130
    },
    {
      "epoch": 15.2,
      "grad_norm": 2.2306067943573,
      "learning_rate": 4.67883337180243e-06,
      "loss": 0.6519,
      "step": 37140
    },
    {
      "epoch": 15.21,
      "grad_norm": 2.7733266353607178,
      "learning_rate": 4.678819737369839e-06,
      "loss": 0.6343,
      "step": 37150
    },
    {
      "epoch": 15.21,
      "grad_norm": 3.451815128326416,
      "learning_rate": 4.678806092240654e-06,
      "loss": 0.6526,
      "step": 37160
    },
    {
      "epoch": 15.21,
      "grad_norm": 2.2963740825653076,
      "learning_rate": 4.6787924364149395e-06,
      "loss": 0.6387,
      "step": 37170
    },
    {
      "epoch": 15.22,
      "grad_norm": 2.9263644218444824,
      "learning_rate": 4.678778769892758e-06,
      "loss": 0.6742,
      "step": 37180
    },
    {
      "epoch": 15.22,
      "grad_norm": 3.511812210083008,
      "learning_rate": 4.6787650926741705e-06,
      "loss": 0.6226,
      "step": 37190
    },
    {
      "epoch": 15.23,
      "grad_norm": 2.502896547317505,
      "learning_rate": 4.678751404759241e-06,
      "loss": 0.6378,
      "step": 37200
    },
    {
      "epoch": 15.23,
      "grad_norm": 3.957930564880371,
      "learning_rate": 4.678737706148031e-06,
      "loss": 0.631,
      "step": 37210
    },
    {
      "epoch": 15.24,
      "grad_norm": 3.008279800415039,
      "learning_rate": 4.678723996840604e-06,
      "loss": 0.645,
      "step": 37220
    },
    {
      "epoch": 15.24,
      "grad_norm": 2.4505245685577393,
      "learning_rate": 4.678710276837023e-06,
      "loss": 0.6264,
      "step": 37230
    },
    {
      "epoch": 15.24,
      "grad_norm": 2.224538803100586,
      "learning_rate": 4.678696546137351e-06,
      "loss": 0.6473,
      "step": 37240
    },
    {
      "epoch": 15.25,
      "grad_norm": 3.1759631633758545,
      "learning_rate": 4.678682804741649e-06,
      "loss": 0.6425,
      "step": 37250
    },
    {
      "epoch": 15.25,
      "grad_norm": 2.4125118255615234,
      "learning_rate": 4.678669052649983e-06,
      "loss": 0.64,
      "step": 37260
    },
    {
      "epoch": 15.26,
      "grad_norm": 2.414349317550659,
      "learning_rate": 4.678655289862414e-06,
      "loss": 0.6313,
      "step": 37270
    },
    {
      "epoch": 15.26,
      "grad_norm": 2.4708008766174316,
      "learning_rate": 4.6786415163790055e-06,
      "loss": 0.645,
      "step": 37280
    },
    {
      "epoch": 15.26,
      "grad_norm": 2.734443187713623,
      "learning_rate": 4.67862773219982e-06,
      "loss": 0.6396,
      "step": 37290
    },
    {
      "epoch": 15.27,
      "grad_norm": 2.7304725646972656,
      "learning_rate": 4.6786139373249214e-06,
      "loss": 0.6363,
      "step": 37300
    },
    {
      "epoch": 15.27,
      "grad_norm": 2.0416626930236816,
      "learning_rate": 4.678600131754374e-06,
      "loss": 0.6379,
      "step": 37310
    },
    {
      "epoch": 15.28,
      "grad_norm": 2.1328928470611572,
      "learning_rate": 4.678586315488238e-06,
      "loss": 0.6468,
      "step": 37320
    },
    {
      "epoch": 15.28,
      "grad_norm": 2.2641305923461914,
      "learning_rate": 4.678572488526579e-06,
      "loss": 0.6607,
      "step": 37330
    },
    {
      "epoch": 15.28,
      "grad_norm": 3.137418508529663,
      "learning_rate": 4.67855865086946e-06,
      "loss": 0.6342,
      "step": 37340
    },
    {
      "epoch": 15.29,
      "grad_norm": 2.7155375480651855,
      "learning_rate": 4.678544802516944e-06,
      "loss": 0.6566,
      "step": 37350
    },
    {
      "epoch": 15.29,
      "grad_norm": 2.5758516788482666,
      "learning_rate": 4.678530943469094e-06,
      "loss": 0.6635,
      "step": 37360
    },
    {
      "epoch": 15.3,
      "grad_norm": 3.622558355331421,
      "learning_rate": 4.678517073725976e-06,
      "loss": 0.6408,
      "step": 37370
    },
    {
      "epoch": 15.3,
      "grad_norm": 3.874131202697754,
      "learning_rate": 4.67850319328765e-06,
      "loss": 0.6415,
      "step": 37380
    },
    {
      "epoch": 15.3,
      "grad_norm": 2.3160178661346436,
      "learning_rate": 4.678489302154182e-06,
      "loss": 0.6477,
      "step": 37390
    },
    {
      "epoch": 15.31,
      "grad_norm": 2.030543088912964,
      "learning_rate": 4.678475400325634e-06,
      "loss": 0.6637,
      "step": 37400
    },
    {
      "epoch": 15.31,
      "grad_norm": 2.9680418968200684,
      "learning_rate": 4.678461487802071e-06,
      "loss": 0.6538,
      "step": 37410
    },
    {
      "epoch": 15.32,
      "grad_norm": 2.692657232284546,
      "learning_rate": 4.678447564583556e-06,
      "loss": 0.6473,
      "step": 37420
    },
    {
      "epoch": 15.32,
      "grad_norm": 2.451820135116577,
      "learning_rate": 4.678433630670153e-06,
      "loss": 0.6283,
      "step": 37430
    },
    {
      "epoch": 15.33,
      "grad_norm": 3.023911476135254,
      "learning_rate": 4.678419686061926e-06,
      "loss": 0.6375,
      "step": 37440
    },
    {
      "epoch": 15.33,
      "grad_norm": 2.5617918968200684,
      "learning_rate": 4.678405730758939e-06,
      "loss": 0.6481,
      "step": 37450
    },
    {
      "epoch": 15.33,
      "grad_norm": 2.535799264907837,
      "learning_rate": 4.678391764761256e-06,
      "loss": 0.6234,
      "step": 37460
    },
    {
      "epoch": 15.34,
      "grad_norm": 2.8340463638305664,
      "learning_rate": 4.6783777880689394e-06,
      "loss": 0.6386,
      "step": 37470
    },
    {
      "epoch": 15.34,
      "grad_norm": 2.653744697570801,
      "learning_rate": 4.6783638006820554e-06,
      "loss": 0.6464,
      "step": 37480
    },
    {
      "epoch": 15.35,
      "grad_norm": 2.636568546295166,
      "learning_rate": 4.678349802600666e-06,
      "loss": 0.6636,
      "step": 37490
    },
    {
      "epoch": 15.35,
      "grad_norm": 1.910236120223999,
      "learning_rate": 4.678335793824838e-06,
      "loss": 0.6512,
      "step": 37500
    },
    {
      "epoch": 15.35,
      "grad_norm": 2.6408345699310303,
      "learning_rate": 4.678321774354633e-06,
      "loss": 0.623,
      "step": 37510
    },
    {
      "epoch": 15.36,
      "grad_norm": 2.607816696166992,
      "learning_rate": 4.678307744190117e-06,
      "loss": 0.6329,
      "step": 37520
    },
    {
      "epoch": 15.36,
      "grad_norm": 2.0712738037109375,
      "learning_rate": 4.678293703331353e-06,
      "loss": 0.6491,
      "step": 37530
    },
    {
      "epoch": 15.37,
      "grad_norm": 2.3651459217071533,
      "learning_rate": 4.678279651778406e-06,
      "loss": 0.6299,
      "step": 37540
    },
    {
      "epoch": 15.37,
      "grad_norm": 2.9104421138763428,
      "learning_rate": 4.6782655895313406e-06,
      "loss": 0.6254,
      "step": 37550
    },
    {
      "epoch": 15.37,
      "grad_norm": 2.9341516494750977,
      "learning_rate": 4.678251516590221e-06,
      "loss": 0.6524,
      "step": 37560
    },
    {
      "epoch": 15.38,
      "grad_norm": 3.5955147743225098,
      "learning_rate": 4.67823743295511e-06,
      "loss": 0.626,
      "step": 37570
    },
    {
      "epoch": 15.38,
      "grad_norm": 3.43900990486145,
      "learning_rate": 4.678223338626074e-06,
      "loss": 0.6228,
      "step": 37580
    },
    {
      "epoch": 15.39,
      "grad_norm": 4.3602166175842285,
      "learning_rate": 4.678209233603178e-06,
      "loss": 0.616,
      "step": 37590
    },
    {
      "epoch": 15.39,
      "grad_norm": 3.0831704139709473,
      "learning_rate": 4.678195117886486e-06,
      "loss": 0.6288,
      "step": 37600
    },
    {
      "epoch": 15.4,
      "grad_norm": 1.749302864074707,
      "learning_rate": 4.678180991476062e-06,
      "loss": 0.6288,
      "step": 37610
    },
    {
      "epoch": 15.4,
      "grad_norm": 2.772141218185425,
      "learning_rate": 4.678166854371971e-06,
      "loss": 0.6289,
      "step": 37620
    },
    {
      "epoch": 15.4,
      "grad_norm": 2.872340679168701,
      "learning_rate": 4.678152706574278e-06,
      "loss": 0.6495,
      "step": 37630
    },
    {
      "epoch": 15.41,
      "grad_norm": 2.829632043838501,
      "learning_rate": 4.678138548083048e-06,
      "loss": 0.6458,
      "step": 37640
    },
    {
      "epoch": 15.41,
      "grad_norm": 2.7115883827209473,
      "learning_rate": 4.678124378898345e-06,
      "loss": 0.6188,
      "step": 37650
    },
    {
      "epoch": 15.42,
      "grad_norm": 3.086014986038208,
      "learning_rate": 4.678110199020235e-06,
      "loss": 0.6416,
      "step": 37660
    },
    {
      "epoch": 15.42,
      "grad_norm": 2.366971969604492,
      "learning_rate": 4.678096008448783e-06,
      "loss": 0.6513,
      "step": 37670
    },
    {
      "epoch": 15.42,
      "grad_norm": 4.38345193862915,
      "learning_rate": 4.678081807184052e-06,
      "loss": 0.6275,
      "step": 37680
    },
    {
      "epoch": 15.43,
      "grad_norm": 2.2899975776672363,
      "learning_rate": 4.678067595226109e-06,
      "loss": 0.6327,
      "step": 37690
    },
    {
      "epoch": 15.43,
      "grad_norm": 2.2081737518310547,
      "learning_rate": 4.678053372575019e-06,
      "loss": 0.6368,
      "step": 37700
    },
    {
      "epoch": 15.44,
      "grad_norm": 2.208292245864868,
      "learning_rate": 4.678039139230847e-06,
      "loss": 0.6473,
      "step": 37710
    },
    {
      "epoch": 15.44,
      "grad_norm": 3.395296096801758,
      "learning_rate": 4.678024895193657e-06,
      "loss": 0.6515,
      "step": 37720
    },
    {
      "epoch": 15.44,
      "grad_norm": 3.248231887817383,
      "learning_rate": 4.678010640463516e-06,
      "loss": 0.6465,
      "step": 37730
    },
    {
      "epoch": 15.45,
      "grad_norm": 1.7916611433029175,
      "learning_rate": 4.677996375040488e-06,
      "loss": 0.6295,
      "step": 37740
    },
    {
      "epoch": 15.45,
      "grad_norm": 2.7292120456695557,
      "learning_rate": 4.67798209892464e-06,
      "loss": 0.6638,
      "step": 37750
    },
    {
      "epoch": 15.46,
      "grad_norm": 4.074592113494873,
      "learning_rate": 4.6779678121160355e-06,
      "loss": 0.6295,
      "step": 37760
    },
    {
      "epoch": 15.46,
      "grad_norm": 2.637927770614624,
      "learning_rate": 4.677953514614741e-06,
      "loss": 0.6452,
      "step": 37770
    },
    {
      "epoch": 15.46,
      "grad_norm": 2.573075532913208,
      "learning_rate": 4.677939206420821e-06,
      "loss": 0.6552,
      "step": 37780
    },
    {
      "epoch": 15.47,
      "grad_norm": 3.1384024620056152,
      "learning_rate": 4.677924887534342e-06,
      "loss": 0.6343,
      "step": 37790
    },
    {
      "epoch": 15.47,
      "grad_norm": 2.297433376312256,
      "learning_rate": 4.677910557955369e-06,
      "loss": 0.6377,
      "step": 37800
    },
    {
      "epoch": 15.48,
      "grad_norm": 2.7836954593658447,
      "learning_rate": 4.677896217683969e-06,
      "loss": 0.6391,
      "step": 37810
    },
    {
      "epoch": 15.48,
      "grad_norm": 2.7749757766723633,
      "learning_rate": 4.677881866720206e-06,
      "loss": 0.6465,
      "step": 37820
    },
    {
      "epoch": 15.49,
      "grad_norm": 2.9328773021698,
      "learning_rate": 4.6778675050641465e-06,
      "loss": 0.6629,
      "step": 37830
    },
    {
      "epoch": 15.49,
      "grad_norm": 2.5167312622070312,
      "learning_rate": 4.677853132715856e-06,
      "loss": 0.6306,
      "step": 37840
    },
    {
      "epoch": 15.49,
      "grad_norm": 3.0655248165130615,
      "learning_rate": 4.6778387496754005e-06,
      "loss": 0.5997,
      "step": 37850
    },
    {
      "epoch": 15.5,
      "grad_norm": 2.8894455432891846,
      "learning_rate": 4.677824355942846e-06,
      "loss": 0.6357,
      "step": 37860
    },
    {
      "epoch": 15.5,
      "grad_norm": 2.958745002746582,
      "learning_rate": 4.6778099515182585e-06,
      "loss": 0.625,
      "step": 37870
    },
    {
      "epoch": 15.51,
      "grad_norm": 2.22891902923584,
      "learning_rate": 4.677795536401704e-06,
      "loss": 0.6303,
      "step": 37880
    },
    {
      "epoch": 15.51,
      "grad_norm": 3.342756509780884,
      "learning_rate": 4.677781110593248e-06,
      "loss": 0.6336,
      "step": 37890
    },
    {
      "epoch": 15.51,
      "grad_norm": 2.0002388954162598,
      "learning_rate": 4.677766674092957e-06,
      "loss": 0.6058,
      "step": 37900
    },
    {
      "epoch": 15.52,
      "grad_norm": 2.6668031215667725,
      "learning_rate": 4.677752226900897e-06,
      "loss": 0.6336,
      "step": 37910
    },
    {
      "epoch": 15.52,
      "grad_norm": 2.601634979248047,
      "learning_rate": 4.677737769017135e-06,
      "loss": 0.6357,
      "step": 37920
    },
    {
      "epoch": 15.53,
      "grad_norm": 1.969595193862915,
      "learning_rate": 4.677723300441736e-06,
      "loss": 0.639,
      "step": 37930
    },
    {
      "epoch": 15.53,
      "grad_norm": 2.522322416305542,
      "learning_rate": 4.6777088211747664e-06,
      "loss": 0.6543,
      "step": 37940
    },
    {
      "epoch": 15.53,
      "grad_norm": 2.4146347045898438,
      "learning_rate": 4.677694331216293e-06,
      "loss": 0.6354,
      "step": 37950
    },
    {
      "epoch": 15.54,
      "grad_norm": 3.8376314640045166,
      "learning_rate": 4.677679830566382e-06,
      "loss": 0.6162,
      "step": 37960
    },
    {
      "epoch": 15.54,
      "grad_norm": 3.677581548690796,
      "learning_rate": 4.677665319225101e-06,
      "loss": 0.6326,
      "step": 37970
    },
    {
      "epoch": 15.55,
      "grad_norm": 3.9611306190490723,
      "learning_rate": 4.677650797192515e-06,
      "loss": 0.6316,
      "step": 37980
    },
    {
      "epoch": 15.55,
      "grad_norm": 3.3678243160247803,
      "learning_rate": 4.67763626446869e-06,
      "loss": 0.6398,
      "step": 37990
    },
    {
      "epoch": 15.55,
      "grad_norm": 2.2688636779785156,
      "learning_rate": 4.6776217210536945e-06,
      "loss": 0.6288,
      "step": 38000
    },
    {
      "epoch": 15.56,
      "grad_norm": 1.8157750368118286,
      "learning_rate": 4.6776071669475926e-06,
      "loss": 0.638,
      "step": 38010
    },
    {
      "epoch": 15.56,
      "grad_norm": 2.2736546993255615,
      "learning_rate": 4.677592602150453e-06,
      "loss": 0.6418,
      "step": 38020
    },
    {
      "epoch": 15.57,
      "grad_norm": 2.6549596786499023,
      "learning_rate": 4.677578026662344e-06,
      "loss": 0.6514,
      "step": 38030
    },
    {
      "epoch": 15.57,
      "grad_norm": 3.0405237674713135,
      "learning_rate": 4.6775634404833274e-06,
      "loss": 0.6512,
      "step": 38040
    },
    {
      "epoch": 15.58,
      "grad_norm": 2.9353432655334473,
      "learning_rate": 4.6775488436134746e-06,
      "loss": 0.6481,
      "step": 38050
    },
    {
      "epoch": 15.58,
      "grad_norm": 3.9060561656951904,
      "learning_rate": 4.6775342360528496e-06,
      "loss": 0.6401,
      "step": 38060
    },
    {
      "epoch": 15.58,
      "grad_norm": 2.163753032684326,
      "learning_rate": 4.677519617801522e-06,
      "loss": 0.637,
      "step": 38070
    },
    {
      "epoch": 15.59,
      "grad_norm": 2.452270984649658,
      "learning_rate": 4.677504988859555e-06,
      "loss": 0.6251,
      "step": 38080
    },
    {
      "epoch": 15.59,
      "grad_norm": 2.5162065029144287,
      "learning_rate": 4.677490349227019e-06,
      "loss": 0.6497,
      "step": 38090
    },
    {
      "epoch": 15.6,
      "grad_norm": 2.3488962650299072,
      "learning_rate": 4.67747569890398e-06,
      "loss": 0.6414,
      "step": 38100
    },
    {
      "epoch": 15.6,
      "grad_norm": 3.23279070854187,
      "learning_rate": 4.677461037890505e-06,
      "loss": 0.6213,
      "step": 38110
    },
    {
      "epoch": 15.6,
      "grad_norm": 2.706007957458496,
      "learning_rate": 4.677446366186661e-06,
      "loss": 0.6354,
      "step": 38120
    },
    {
      "epoch": 15.61,
      "grad_norm": 2.129687786102295,
      "learning_rate": 4.6774316837925155e-06,
      "loss": 0.6443,
      "step": 38130
    },
    {
      "epoch": 15.61,
      "grad_norm": 2.6376688480377197,
      "learning_rate": 4.677416990708135e-06,
      "loss": 0.6201,
      "step": 38140
    },
    {
      "epoch": 15.62,
      "grad_norm": 2.4130866527557373,
      "learning_rate": 4.677402286933588e-06,
      "loss": 0.6295,
      "step": 38150
    },
    {
      "epoch": 15.62,
      "grad_norm": 2.751508951187134,
      "learning_rate": 4.677387572468942e-06,
      "loss": 0.6248,
      "step": 38160
    },
    {
      "epoch": 15.62,
      "grad_norm": 2.4812912940979004,
      "learning_rate": 4.6773728473142625e-06,
      "loss": 0.6577,
      "step": 38170
    },
    {
      "epoch": 15.63,
      "grad_norm": 2.492661237716675,
      "learning_rate": 4.677358111469619e-06,
      "loss": 0.6464,
      "step": 38180
    },
    {
      "epoch": 15.63,
      "grad_norm": 2.6696054935455322,
      "learning_rate": 4.677343364935077e-06,
      "loss": 0.6354,
      "step": 38190
    },
    {
      "epoch": 15.64,
      "grad_norm": 3.1513259410858154,
      "learning_rate": 4.677328607710707e-06,
      "loss": 0.6227,
      "step": 38200
    },
    {
      "epoch": 15.64,
      "grad_norm": 2.8931515216827393,
      "learning_rate": 4.6773138397965735e-06,
      "loss": 0.6373,
      "step": 38210
    },
    {
      "epoch": 15.64,
      "grad_norm": 2.636235237121582,
      "learning_rate": 4.677299061192746e-06,
      "loss": 0.6693,
      "step": 38220
    },
    {
      "epoch": 15.65,
      "grad_norm": 3.244530439376831,
      "learning_rate": 4.67728427189929e-06,
      "loss": 0.6459,
      "step": 38230
    },
    {
      "epoch": 15.65,
      "grad_norm": 2.8265018463134766,
      "learning_rate": 4.677269471916277e-06,
      "loss": 0.6314,
      "step": 38240
    },
    {
      "epoch": 15.66,
      "grad_norm": 2.599531412124634,
      "learning_rate": 4.677254661243772e-06,
      "loss": 0.6547,
      "step": 38250
    },
    {
      "epoch": 15.66,
      "grad_norm": 3.089181661605835,
      "learning_rate": 4.677239839881843e-06,
      "loss": 0.6644,
      "step": 38260
    },
    {
      "epoch": 15.67,
      "grad_norm": 2.3977115154266357,
      "learning_rate": 4.677225007830559e-06,
      "loss": 0.6441,
      "step": 38270
    },
    {
      "epoch": 15.67,
      "grad_norm": 2.775545835494995,
      "learning_rate": 4.677210165089987e-06,
      "loss": 0.6168,
      "step": 38280
    },
    {
      "epoch": 15.67,
      "grad_norm": 2.6185996532440186,
      "learning_rate": 4.677195311660196e-06,
      "loss": 0.6261,
      "step": 38290
    },
    {
      "epoch": 15.68,
      "grad_norm": 3.6245453357696533,
      "learning_rate": 4.677180447541253e-06,
      "loss": 0.6099,
      "step": 38300
    },
    {
      "epoch": 15.68,
      "grad_norm": 2.6158335208892822,
      "learning_rate": 4.677165572733226e-06,
      "loss": 0.6318,
      "step": 38310
    },
    {
      "epoch": 15.69,
      "grad_norm": 2.201587200164795,
      "learning_rate": 4.677150687236184e-06,
      "loss": 0.6291,
      "step": 38320
    },
    {
      "epoch": 15.69,
      "grad_norm": 2.508404016494751,
      "learning_rate": 4.677135791050196e-06,
      "loss": 0.6187,
      "step": 38330
    },
    {
      "epoch": 15.69,
      "grad_norm": 2.7960495948791504,
      "learning_rate": 4.677120884175328e-06,
      "loss": 0.6338,
      "step": 38340
    },
    {
      "epoch": 15.7,
      "grad_norm": 3.8192224502563477,
      "learning_rate": 4.677105966611649e-06,
      "loss": 0.6451,
      "step": 38350
    },
    {
      "epoch": 15.7,
      "grad_norm": 2.7206761837005615,
      "learning_rate": 4.677091038359228e-06,
      "loss": 0.6091,
      "step": 38360
    },
    {
      "epoch": 15.71,
      "grad_norm": 3.287867784500122,
      "learning_rate": 4.677076099418133e-06,
      "loss": 0.6362,
      "step": 38370
    },
    {
      "epoch": 15.71,
      "grad_norm": 2.7635843753814697,
      "learning_rate": 4.677061149788433e-06,
      "loss": 0.6635,
      "step": 38380
    },
    {
      "epoch": 15.71,
      "grad_norm": 2.5834579467773438,
      "learning_rate": 4.677046189470196e-06,
      "loss": 0.6172,
      "step": 38390
    },
    {
      "epoch": 15.72,
      "grad_norm": 2.801835536956787,
      "learning_rate": 4.6770312184634895e-06,
      "loss": 0.6406,
      "step": 38400
    },
    {
      "epoch": 15.72,
      "grad_norm": 2.9632420539855957,
      "learning_rate": 4.677016236768384e-06,
      "loss": 0.6501,
      "step": 38410
    },
    {
      "epoch": 15.73,
      "grad_norm": 2.8932414054870605,
      "learning_rate": 4.677001244384947e-06,
      "loss": 0.6233,
      "step": 38420
    },
    {
      "epoch": 15.73,
      "grad_norm": 2.880448818206787,
      "learning_rate": 4.676986241313247e-06,
      "loss": 0.636,
      "step": 38430
    },
    {
      "epoch": 15.73,
      "grad_norm": 2.84251070022583,
      "learning_rate": 4.676971227553353e-06,
      "loss": 0.6368,
      "step": 38440
    },
    {
      "epoch": 15.74,
      "grad_norm": 2.435663938522339,
      "learning_rate": 4.676956203105335e-06,
      "loss": 0.6319,
      "step": 38450
    },
    {
      "epoch": 15.74,
      "grad_norm": 2.577286720275879,
      "learning_rate": 4.67694116796926e-06,
      "loss": 0.6318,
      "step": 38460
    },
    {
      "epoch": 15.75,
      "grad_norm": 2.293910503387451,
      "learning_rate": 4.676926122145197e-06,
      "loss": 0.637,
      "step": 38470
    },
    {
      "epoch": 15.75,
      "grad_norm": 3.7945303916931152,
      "learning_rate": 4.676911065633217e-06,
      "loss": 0.6269,
      "step": 38480
    },
    {
      "epoch": 15.76,
      "grad_norm": 2.313807249069214,
      "learning_rate": 4.676895998433386e-06,
      "loss": 0.6313,
      "step": 38490
    },
    {
      "epoch": 15.76,
      "grad_norm": 3.368680715560913,
      "learning_rate": 4.676880920545776e-06,
      "loss": 0.6271,
      "step": 38500
    },
    {
      "epoch": 15.76,
      "grad_norm": 3.033524990081787,
      "learning_rate": 4.676865831970454e-06,
      "loss": 0.6383,
      "step": 38510
    },
    {
      "epoch": 15.77,
      "grad_norm": 1.9793490171432495,
      "learning_rate": 4.676850732707489e-06,
      "loss": 0.6385,
      "step": 38520
    },
    {
      "epoch": 15.77,
      "grad_norm": 2.4702305793762207,
      "learning_rate": 4.6768356227569515e-06,
      "loss": 0.6454,
      "step": 38530
    },
    {
      "epoch": 15.78,
      "grad_norm": 4.10009241104126,
      "learning_rate": 4.67682050211891e-06,
      "loss": 0.6412,
      "step": 38540
    },
    {
      "epoch": 15.78,
      "grad_norm": 2.743800640106201,
      "learning_rate": 4.676805370793433e-06,
      "loss": 0.6225,
      "step": 38550
    },
    {
      "epoch": 15.78,
      "grad_norm": 2.471928834915161,
      "learning_rate": 4.676790228780592e-06,
      "loss": 0.6316,
      "step": 38560
    },
    {
      "epoch": 15.79,
      "grad_norm": 2.789595603942871,
      "learning_rate": 4.676775076080455e-06,
      "loss": 0.6379,
      "step": 38570
    },
    {
      "epoch": 15.79,
      "grad_norm": 1.9383084774017334,
      "learning_rate": 4.67675991269309e-06,
      "loss": 0.6323,
      "step": 38580
    },
    {
      "epoch": 15.8,
      "grad_norm": 3.1053483486175537,
      "learning_rate": 4.6767447386185694e-06,
      "loss": 0.6402,
      "step": 38590
    },
    {
      "epoch": 15.8,
      "grad_norm": 3.7602105140686035,
      "learning_rate": 4.676729553856961e-06,
      "loss": 0.6511,
      "step": 38600
    },
    {
      "epoch": 15.8,
      "grad_norm": 2.1643974781036377,
      "learning_rate": 4.676714358408334e-06,
      "loss": 0.6327,
      "step": 38610
    },
    {
      "epoch": 15.81,
      "grad_norm": 3.2693920135498047,
      "learning_rate": 4.676699152272759e-06,
      "loss": 0.6434,
      "step": 38620
    },
    {
      "epoch": 15.81,
      "grad_norm": 2.82348370552063,
      "learning_rate": 4.676683935450304e-06,
      "loss": 0.661,
      "step": 38630
    },
    {
      "epoch": 15.82,
      "grad_norm": 2.6154518127441406,
      "learning_rate": 4.6766687079410415e-06,
      "loss": 0.61,
      "step": 38640
    },
    {
      "epoch": 15.82,
      "grad_norm": 2.636479377746582,
      "learning_rate": 4.67665346974504e-06,
      "loss": 0.6147,
      "step": 38650
    },
    {
      "epoch": 15.82,
      "grad_norm": 2.570660352706909,
      "learning_rate": 4.676638220862367e-06,
      "loss": 0.6428,
      "step": 38660
    },
    {
      "epoch": 15.83,
      "grad_norm": 3.0540761947631836,
      "learning_rate": 4.676622961293096e-06,
      "loss": 0.6438,
      "step": 38670
    },
    {
      "epoch": 15.83,
      "grad_norm": 2.1755733489990234,
      "learning_rate": 4.676607691037295e-06,
      "loss": 0.6306,
      "step": 38680
    },
    {
      "epoch": 15.84,
      "grad_norm": 2.426553249359131,
      "learning_rate": 4.676592410095034e-06,
      "loss": 0.644,
      "step": 38690
    },
    {
      "epoch": 15.84,
      "grad_norm": 2.5056521892547607,
      "learning_rate": 4.676577118466383e-06,
      "loss": 0.6214,
      "step": 38700
    },
    {
      "epoch": 15.85,
      "grad_norm": 2.56709623336792,
      "learning_rate": 4.676561816151413e-06,
      "loss": 0.638,
      "step": 38710
    },
    {
      "epoch": 15.85,
      "grad_norm": 2.777885913848877,
      "learning_rate": 4.676546503150192e-06,
      "loss": 0.6249,
      "step": 38720
    },
    {
      "epoch": 15.85,
      "grad_norm": 2.33632755279541,
      "learning_rate": 4.676531179462793e-06,
      "loss": 0.6663,
      "step": 38730
    },
    {
      "epoch": 15.86,
      "grad_norm": 2.221935510635376,
      "learning_rate": 4.676515845089284e-06,
      "loss": 0.6152,
      "step": 38740
    },
    {
      "epoch": 15.86,
      "grad_norm": 2.729332208633423,
      "learning_rate": 4.676500500029736e-06,
      "loss": 0.6325,
      "step": 38750
    },
    {
      "epoch": 15.87,
      "grad_norm": 2.510345935821533,
      "learning_rate": 4.676485144284219e-06,
      "loss": 0.6367,
      "step": 38760
    },
    {
      "epoch": 15.87,
      "grad_norm": 2.585667848587036,
      "learning_rate": 4.676469777852804e-06,
      "loss": 0.6236,
      "step": 38770
    },
    {
      "epoch": 15.87,
      "grad_norm": 2.9674534797668457,
      "learning_rate": 4.676454400735561e-06,
      "loss": 0.6509,
      "step": 38780
    },
    {
      "epoch": 15.88,
      "grad_norm": 1.891034722328186,
      "learning_rate": 4.676439012932559e-06,
      "loss": 0.658,
      "step": 38790
    },
    {
      "epoch": 15.88,
      "grad_norm": 2.2516093254089355,
      "learning_rate": 4.676423614443871e-06,
      "loss": 0.634,
      "step": 38800
    },
    {
      "epoch": 15.89,
      "grad_norm": 2.8608574867248535,
      "learning_rate": 4.676408205269567e-06,
      "loss": 0.638,
      "step": 38810
    },
    {
      "epoch": 15.89,
      "grad_norm": 2.662395477294922,
      "learning_rate": 4.676392785409717e-06,
      "loss": 0.6318,
      "step": 38820
    },
    {
      "epoch": 15.89,
      "grad_norm": 2.1882822513580322,
      "learning_rate": 4.676377354864391e-06,
      "loss": 0.6507,
      "step": 38830
    },
    {
      "epoch": 15.9,
      "grad_norm": 2.1387557983398438,
      "learning_rate": 4.67636191363366e-06,
      "loss": 0.6447,
      "step": 38840
    },
    {
      "epoch": 15.9,
      "grad_norm": 2.666137456893921,
      "learning_rate": 4.676346461717596e-06,
      "loss": 0.6347,
      "step": 38850
    },
    {
      "epoch": 15.91,
      "grad_norm": 2.297271728515625,
      "learning_rate": 4.676330999116268e-06,
      "loss": 0.626,
      "step": 38860
    },
    {
      "epoch": 15.91,
      "grad_norm": 2.3416476249694824,
      "learning_rate": 4.676315525829748e-06,
      "loss": 0.6004,
      "step": 38870
    },
    {
      "epoch": 15.91,
      "grad_norm": 2.8095180988311768,
      "learning_rate": 4.676300041858106e-06,
      "loss": 0.6263,
      "step": 38880
    },
    {
      "epoch": 15.92,
      "grad_norm": 2.5307090282440186,
      "learning_rate": 4.676284547201415e-06,
      "loss": 0.6241,
      "step": 38890
    },
    {
      "epoch": 15.92,
      "grad_norm": 3.1775896549224854,
      "learning_rate": 4.676269041859744e-06,
      "loss": 0.6278,
      "step": 38900
    },
    {
      "epoch": 15.93,
      "grad_norm": 2.5590693950653076,
      "learning_rate": 4.676253525833164e-06,
      "loss": 0.6483,
      "step": 38910
    },
    {
      "epoch": 15.93,
      "grad_norm": 2.907219886779785,
      "learning_rate": 4.676237999121746e-06,
      "loss": 0.6449,
      "step": 38920
    },
    {
      "epoch": 15.94,
      "grad_norm": 3.070040225982666,
      "learning_rate": 4.676222461725563e-06,
      "loss": 0.6155,
      "step": 38930
    },
    {
      "epoch": 15.94,
      "grad_norm": 2.2149603366851807,
      "learning_rate": 4.676206913644684e-06,
      "loss": 0.6697,
      "step": 38940
    },
    {
      "epoch": 15.94,
      "grad_norm": 3.421022653579712,
      "learning_rate": 4.676191354879181e-06,
      "loss": 0.6222,
      "step": 38950
    },
    {
      "epoch": 15.95,
      "grad_norm": 2.6392087936401367,
      "learning_rate": 4.676175785429126e-06,
      "loss": 0.6276,
      "step": 38960
    },
    {
      "epoch": 15.95,
      "grad_norm": 2.170921802520752,
      "learning_rate": 4.676160205294589e-06,
      "loss": 0.6362,
      "step": 38970
    },
    {
      "epoch": 15.96,
      "grad_norm": 3.241060733795166,
      "learning_rate": 4.676144614475643e-06,
      "loss": 0.6415,
      "step": 38980
    },
    {
      "epoch": 15.96,
      "grad_norm": 2.5890228748321533,
      "learning_rate": 4.676129012972358e-06,
      "loss": 0.6473,
      "step": 38990
    },
    {
      "epoch": 15.96,
      "grad_norm": 2.326591968536377,
      "learning_rate": 4.676113400784807e-06,
      "loss": 0.6285,
      "step": 39000
    },
    {
      "epoch": 15.97,
      "grad_norm": 2.5049796104431152,
      "learning_rate": 4.676097777913059e-06,
      "loss": 0.6208,
      "step": 39010
    },
    {
      "epoch": 15.97,
      "grad_norm": 2.48635196685791,
      "learning_rate": 4.6760821443571875e-06,
      "loss": 0.6234,
      "step": 39020
    },
    {
      "epoch": 15.98,
      "grad_norm": 2.3220996856689453,
      "learning_rate": 4.676066500117263e-06,
      "loss": 0.6628,
      "step": 39030
    },
    {
      "epoch": 15.98,
      "grad_norm": 2.4100208282470703,
      "learning_rate": 4.676050845193359e-06,
      "loss": 0.6427,
      "step": 39040
    },
    {
      "epoch": 15.98,
      "grad_norm": 2.3879222869873047,
      "learning_rate": 4.676035179585545e-06,
      "loss": 0.6521,
      "step": 39050
    },
    {
      "epoch": 15.99,
      "grad_norm": 3.2815864086151123,
      "learning_rate": 4.676019503293895e-06,
      "loss": 0.6227,
      "step": 39060
    },
    {
      "epoch": 15.99,
      "grad_norm": 1.9563926458358765,
      "learning_rate": 4.6760038163184785e-06,
      "loss": 0.6442,
      "step": 39070
    },
    {
      "epoch": 16.0,
      "grad_norm": 2.6378002166748047,
      "learning_rate": 4.6759881186593695e-06,
      "loss": 0.6381,
      "step": 39080
    },
    {
      "epoch": 16.0,
      "eval_loss": 0.6369776129722595,
      "eval_runtime": 51.9876,
      "eval_samples_per_second": 66.343,
      "eval_steps_per_second": 8.31,
      "step": 39088
    },
    {
      "epoch": 16.0,
      "grad_norm": 2.1257152557373047,
      "learning_rate": 4.675972410316637e-06,
      "loss": 0.6547,
      "step": 39090
    },
    {
      "epoch": 16.0,
      "grad_norm": 2.6416778564453125,
      "learning_rate": 4.6759566912903565e-06,
      "loss": 0.6496,
      "step": 39100
    },
    {
      "epoch": 16.01,
      "grad_norm": 2.597236156463623,
      "learning_rate": 4.675940961580598e-06,
      "loss": 0.6224,
      "step": 39110
    },
    {
      "epoch": 16.01,
      "grad_norm": 2.697856903076172,
      "learning_rate": 4.675925221187434e-06,
      "loss": 0.6244,
      "step": 39120
    },
    {
      "epoch": 16.02,
      "grad_norm": 3.5281527042388916,
      "learning_rate": 4.675909470110936e-06,
      "loss": 0.6232,
      "step": 39130
    },
    {
      "epoch": 16.02,
      "grad_norm": 2.253023862838745,
      "learning_rate": 4.675893708351177e-06,
      "loss": 0.613,
      "step": 39140
    },
    {
      "epoch": 16.03,
      "grad_norm": 2.36112117767334,
      "learning_rate": 4.675877935908228e-06,
      "loss": 0.6464,
      "step": 39150
    },
    {
      "epoch": 16.03,
      "grad_norm": 2.6240501403808594,
      "learning_rate": 4.675862152782164e-06,
      "loss": 0.6317,
      "step": 39160
    },
    {
      "epoch": 16.03,
      "grad_norm": 3.0427863597869873,
      "learning_rate": 4.675846358973054e-06,
      "loss": 0.6209,
      "step": 39170
    },
    {
      "epoch": 16.04,
      "grad_norm": 1.8524128198623657,
      "learning_rate": 4.675830554480971e-06,
      "loss": 0.6575,
      "step": 39180
    },
    {
      "epoch": 16.04,
      "grad_norm": 1.994423508644104,
      "learning_rate": 4.67581473930599e-06,
      "loss": 0.648,
      "step": 39190
    },
    {
      "epoch": 16.05,
      "grad_norm": 2.3256564140319824,
      "learning_rate": 4.675798913448181e-06,
      "loss": 0.6245,
      "step": 39200
    },
    {
      "epoch": 16.05,
      "grad_norm": 2.238955497741699,
      "learning_rate": 4.675783076907616e-06,
      "loss": 0.6513,
      "step": 39210
    },
    {
      "epoch": 16.05,
      "grad_norm": 2.1902456283569336,
      "learning_rate": 4.67576722968437e-06,
      "loss": 0.6363,
      "step": 39220
    },
    {
      "epoch": 16.06,
      "grad_norm": 2.007943630218506,
      "learning_rate": 4.675751371778514e-06,
      "loss": 0.642,
      "step": 39230
    },
    {
      "epoch": 16.06,
      "grad_norm": 2.3084075450897217,
      "learning_rate": 4.6757355031901205e-06,
      "loss": 0.6264,
      "step": 39240
    },
    {
      "epoch": 16.07,
      "grad_norm": 2.419182538986206,
      "learning_rate": 4.6757196239192625e-06,
      "loss": 0.6393,
      "step": 39250
    },
    {
      "epoch": 16.07,
      "grad_norm": 3.16894793510437,
      "learning_rate": 4.675703733966013e-06,
      "loss": 0.6378,
      "step": 39260
    },
    {
      "epoch": 16.07,
      "grad_norm": 2.176063060760498,
      "learning_rate": 4.675687833330445e-06,
      "loss": 0.6344,
      "step": 39270
    },
    {
      "epoch": 16.08,
      "grad_norm": 3.0196797847747803,
      "learning_rate": 4.6756719220126305e-06,
      "loss": 0.6287,
      "step": 39280
    },
    {
      "epoch": 16.08,
      "grad_norm": 2.8592491149902344,
      "learning_rate": 4.675656000012642e-06,
      "loss": 0.6143,
      "step": 39290
    },
    {
      "epoch": 16.09,
      "grad_norm": 2.5911366939544678,
      "learning_rate": 4.675640067330555e-06,
      "loss": 0.6226,
      "step": 39300
    },
    {
      "epoch": 16.09,
      "grad_norm": 2.8130874633789062,
      "learning_rate": 4.675624123966439e-06,
      "loss": 0.6469,
      "step": 39310
    },
    {
      "epoch": 16.09,
      "grad_norm": 2.420806407928467,
      "learning_rate": 4.67560816992037e-06,
      "loss": 0.6365,
      "step": 39320
    },
    {
      "epoch": 16.1,
      "grad_norm": 3.992384433746338,
      "learning_rate": 4.67559220519242e-06,
      "loss": 0.6505,
      "step": 39330
    },
    {
      "epoch": 16.1,
      "grad_norm": 2.2845940589904785,
      "learning_rate": 4.675576229782662e-06,
      "loss": 0.6458,
      "step": 39340
    },
    {
      "epoch": 16.11,
      "grad_norm": 2.9288320541381836,
      "learning_rate": 4.6755602436911675e-06,
      "loss": 0.6256,
      "step": 39350
    },
    {
      "epoch": 16.11,
      "grad_norm": 2.6003663539886475,
      "learning_rate": 4.675544246918013e-06,
      "loss": 0.6156,
      "step": 39360
    },
    {
      "epoch": 16.12,
      "grad_norm": 2.6880879402160645,
      "learning_rate": 4.675528239463269e-06,
      "loss": 0.6209,
      "step": 39370
    },
    {
      "epoch": 16.12,
      "grad_norm": 2.8459904193878174,
      "learning_rate": 4.675512221327011e-06,
      "loss": 0.6235,
      "step": 39380
    },
    {
      "epoch": 16.12,
      "grad_norm": 2.0926408767700195,
      "learning_rate": 4.675496192509311e-06,
      "loss": 0.6469,
      "step": 39390
    },
    {
      "epoch": 16.13,
      "grad_norm": 2.3571057319641113,
      "learning_rate": 4.675480153010243e-06,
      "loss": 0.6288,
      "step": 39400
    },
    {
      "epoch": 16.13,
      "grad_norm": 2.853297472000122,
      "learning_rate": 4.6754641028298805e-06,
      "loss": 0.618,
      "step": 39410
    },
    {
      "epoch": 16.14,
      "grad_norm": 3.556408405303955,
      "learning_rate": 4.675448041968296e-06,
      "loss": 0.6412,
      "step": 39420
    },
    {
      "epoch": 16.14,
      "grad_norm": 3.331052780151367,
      "learning_rate": 4.6754319704255645e-06,
      "loss": 0.6642,
      "step": 39430
    },
    {
      "epoch": 16.14,
      "grad_norm": 2.336488962173462,
      "learning_rate": 4.675415888201758e-06,
      "loss": 0.6248,
      "step": 39440
    },
    {
      "epoch": 16.15,
      "grad_norm": 2.139361619949341,
      "learning_rate": 4.675399795296952e-06,
      "loss": 0.6199,
      "step": 39450
    },
    {
      "epoch": 16.15,
      "grad_norm": 2.1644880771636963,
      "learning_rate": 4.67538369171122e-06,
      "loss": 0.6283,
      "step": 39460
    },
    {
      "epoch": 16.16,
      "grad_norm": 2.526815176010132,
      "learning_rate": 4.6753675774446345e-06,
      "loss": 0.6244,
      "step": 39470
    },
    {
      "epoch": 16.16,
      "grad_norm": 2.819751024246216,
      "learning_rate": 4.67535145249727e-06,
      "loss": 0.6573,
      "step": 39480
    },
    {
      "epoch": 16.16,
      "grad_norm": 2.7410683631896973,
      "learning_rate": 4.6753353168692e-06,
      "loss": 0.6386,
      "step": 39490
    },
    {
      "epoch": 16.17,
      "grad_norm": 3.122727632522583,
      "learning_rate": 4.675319170560499e-06,
      "loss": 0.6483,
      "step": 39500
    },
    {
      "epoch": 16.17,
      "grad_norm": 2.8993635177612305,
      "learning_rate": 4.675303013571241e-06,
      "loss": 0.6372,
      "step": 39510
    },
    {
      "epoch": 16.18,
      "grad_norm": 2.412125825881958,
      "learning_rate": 4.6752868459015e-06,
      "loss": 0.634,
      "step": 39520
    },
    {
      "epoch": 16.18,
      "grad_norm": 3.278061866760254,
      "learning_rate": 4.675270667551349e-06,
      "loss": 0.6305,
      "step": 39530
    },
    {
      "epoch": 16.19,
      "grad_norm": 2.2966017723083496,
      "learning_rate": 4.6752544785208645e-06,
      "loss": 0.6359,
      "step": 39540
    },
    {
      "epoch": 16.19,
      "grad_norm": 3.456667184829712,
      "learning_rate": 4.6752382788101174e-06,
      "loss": 0.6565,
      "step": 39550
    },
    {
      "epoch": 16.19,
      "grad_norm": 2.3349833488464355,
      "learning_rate": 4.675222068419184e-06,
      "loss": 0.6314,
      "step": 39560
    },
    {
      "epoch": 16.2,
      "grad_norm": 2.482239246368408,
      "learning_rate": 4.675205847348138e-06,
      "loss": 0.6635,
      "step": 39570
    },
    {
      "epoch": 16.2,
      "grad_norm": 2.760077714920044,
      "learning_rate": 4.675189615597054e-06,
      "loss": 0.6472,
      "step": 39580
    },
    {
      "epoch": 16.21,
      "grad_norm": 2.6523308753967285,
      "learning_rate": 4.675173373166006e-06,
      "loss": 0.6475,
      "step": 39590
    },
    {
      "epoch": 16.21,
      "grad_norm": 2.8609752655029297,
      "learning_rate": 4.675157120055068e-06,
      "loss": 0.6346,
      "step": 39600
    },
    {
      "epoch": 16.21,
      "grad_norm": 2.671121835708618,
      "learning_rate": 4.675140856264317e-06,
      "loss": 0.6238,
      "step": 39610
    },
    {
      "epoch": 16.22,
      "grad_norm": 2.9296152591705322,
      "learning_rate": 4.675124581793824e-06,
      "loss": 0.6527,
      "step": 39620
    },
    {
      "epoch": 16.22,
      "grad_norm": 2.550976514816284,
      "learning_rate": 4.675108296643665e-06,
      "loss": 0.6425,
      "step": 39630
    },
    {
      "epoch": 16.23,
      "grad_norm": 2.9832041263580322,
      "learning_rate": 4.675092000813916e-06,
      "loss": 0.6441,
      "step": 39640
    },
    {
      "epoch": 16.23,
      "grad_norm": 2.9334380626678467,
      "learning_rate": 4.6750756943046494e-06,
      "loss": 0.6048,
      "step": 39650
    },
    {
      "epoch": 16.23,
      "grad_norm": 2.8638486862182617,
      "learning_rate": 4.675059377115941e-06,
      "loss": 0.6297,
      "step": 39660
    },
    {
      "epoch": 16.24,
      "grad_norm": 2.680793523788452,
      "learning_rate": 4.675043049247865e-06,
      "loss": 0.6275,
      "step": 39670
    },
    {
      "epoch": 16.24,
      "grad_norm": 2.591808557510376,
      "learning_rate": 4.675026710700497e-06,
      "loss": 0.6378,
      "step": 39680
    },
    {
      "epoch": 16.25,
      "grad_norm": 3.5518667697906494,
      "learning_rate": 4.6750103614739115e-06,
      "loss": 0.636,
      "step": 39690
    },
    {
      "epoch": 16.25,
      "grad_norm": 3.0787596702575684,
      "learning_rate": 4.674994001568183e-06,
      "loss": 0.6274,
      "step": 39700
    },
    {
      "epoch": 16.25,
      "grad_norm": 2.8650662899017334,
      "learning_rate": 4.674977630983387e-06,
      "loss": 0.6276,
      "step": 39710
    },
    {
      "epoch": 16.26,
      "grad_norm": 3.079967975616455,
      "learning_rate": 4.674961249719598e-06,
      "loss": 0.6259,
      "step": 39720
    },
    {
      "epoch": 16.26,
      "grad_norm": 2.228783369064331,
      "learning_rate": 4.674944857776891e-06,
      "loss": 0.6257,
      "step": 39730
    },
    {
      "epoch": 16.27,
      "grad_norm": 2.1830849647521973,
      "learning_rate": 4.674928455155343e-06,
      "loss": 0.6091,
      "step": 39740
    },
    {
      "epoch": 16.27,
      "grad_norm": 2.866966724395752,
      "learning_rate": 4.674912041855025e-06,
      "loss": 0.6284,
      "step": 39750
    },
    {
      "epoch": 16.28,
      "grad_norm": 2.292307138442993,
      "learning_rate": 4.674895617876017e-06,
      "loss": 0.6369,
      "step": 39760
    },
    {
      "epoch": 16.28,
      "grad_norm": 2.4524922370910645,
      "learning_rate": 4.674879183218391e-06,
      "loss": 0.6139,
      "step": 39770
    },
    {
      "epoch": 16.28,
      "grad_norm": 2.583810567855835,
      "learning_rate": 4.6748627378822235e-06,
      "loss": 0.6241,
      "step": 39780
    },
    {
      "epoch": 16.29,
      "grad_norm": 3.384042263031006,
      "learning_rate": 4.67484628186759e-06,
      "loss": 0.6379,
      "step": 39790
    },
    {
      "epoch": 16.29,
      "grad_norm": 3.443862199783325,
      "learning_rate": 4.674829815174563e-06,
      "loss": 0.6471,
      "step": 39800
    },
    {
      "epoch": 16.3,
      "grad_norm": 2.3390204906463623,
      "learning_rate": 4.674813337803224e-06,
      "loss": 0.6294,
      "step": 39810
    },
    {
      "epoch": 16.3,
      "grad_norm": 2.495562791824341,
      "learning_rate": 4.674796849753643e-06,
      "loss": 0.6192,
      "step": 39820
    },
    {
      "epoch": 16.3,
      "grad_norm": 3.1347858905792236,
      "learning_rate": 4.6747803510258964e-06,
      "loss": 0.6428,
      "step": 39830
    },
    {
      "epoch": 16.31,
      "grad_norm": 3.2555506229400635,
      "learning_rate": 4.674763841620062e-06,
      "loss": 0.6543,
      "step": 39840
    },
    {
      "epoch": 16.31,
      "grad_norm": 2.812868356704712,
      "learning_rate": 4.6747473215362146e-06,
      "loss": 0.6346,
      "step": 39850
    },
    {
      "epoch": 16.32,
      "grad_norm": 2.453119993209839,
      "learning_rate": 4.674730790774429e-06,
      "loss": 0.6264,
      "step": 39860
    },
    {
      "epoch": 16.32,
      "grad_norm": 2.8440566062927246,
      "learning_rate": 4.67471424933478e-06,
      "loss": 0.6402,
      "step": 39870
    },
    {
      "epoch": 16.32,
      "grad_norm": 2.733213186264038,
      "learning_rate": 4.674697697217347e-06,
      "loss": 0.6396,
      "step": 39880
    },
    {
      "epoch": 16.33,
      "grad_norm": 2.2371487617492676,
      "learning_rate": 4.674681134422203e-06,
      "loss": 0.6488,
      "step": 39890
    },
    {
      "epoch": 16.33,
      "grad_norm": 2.490842342376709,
      "learning_rate": 4.674664560949423e-06,
      "loss": 0.6217,
      "step": 39900
    },
    {
      "epoch": 16.34,
      "grad_norm": 2.789400339126587,
      "learning_rate": 4.6746479767990854e-06,
      "loss": 0.6291,
      "step": 39910
    },
    {
      "epoch": 16.34,
      "grad_norm": 2.6276421546936035,
      "learning_rate": 4.674631381971266e-06,
      "loss": 0.6444,
      "step": 39920
    },
    {
      "epoch": 16.34,
      "grad_norm": 3.2386882305145264,
      "learning_rate": 4.674614776466039e-06,
      "loss": 0.6251,
      "step": 39930
    },
    {
      "epoch": 16.35,
      "grad_norm": 3.6080894470214844,
      "learning_rate": 4.674598160283482e-06,
      "loss": 0.6488,
      "step": 39940
    },
    {
      "epoch": 16.35,
      "grad_norm": 2.998136043548584,
      "learning_rate": 4.67458153342367e-06,
      "loss": 0.6356,
      "step": 39950
    },
    {
      "epoch": 16.36,
      "grad_norm": 2.8533682823181152,
      "learning_rate": 4.6745648958866795e-06,
      "loss": 0.6289,
      "step": 39960
    },
    {
      "epoch": 16.36,
      "grad_norm": 2.932708740234375,
      "learning_rate": 4.6745482476725875e-06,
      "loss": 0.6112,
      "step": 39970
    },
    {
      "epoch": 16.37,
      "grad_norm": 2.218787670135498,
      "learning_rate": 4.67453158878147e-06,
      "loss": 0.6318,
      "step": 39980
    },
    {
      "epoch": 16.37,
      "grad_norm": 2.81288480758667,
      "learning_rate": 4.674514919213402e-06,
      "loss": 0.6243,
      "step": 39990
    },
    {
      "epoch": 16.37,
      "grad_norm": 3.4398534297943115,
      "learning_rate": 4.6744982389684615e-06,
      "loss": 0.6269,
      "step": 40000
    },
    {
      "epoch": 16.38,
      "grad_norm": 2.5767054557800293,
      "learning_rate": 4.674481548046724e-06,
      "loss": 0.6297,
      "step": 40010
    },
    {
      "epoch": 16.38,
      "grad_norm": 3.4289467334747314,
      "learning_rate": 4.674464846448266e-06,
      "loss": 0.6299,
      "step": 40020
    },
    {
      "epoch": 16.39,
      "grad_norm": 2.658027410507202,
      "learning_rate": 4.674448134173166e-06,
      "loss": 0.6447,
      "step": 40030
    },
    {
      "epoch": 16.39,
      "grad_norm": 3.723451614379883,
      "learning_rate": 4.674431411221497e-06,
      "loss": 0.6232,
      "step": 40040
    },
    {
      "epoch": 16.39,
      "grad_norm": 2.561666250228882,
      "learning_rate": 4.674414677593338e-06,
      "loss": 0.6155,
      "step": 40050
    },
    {
      "epoch": 16.4,
      "grad_norm": 2.382859230041504,
      "learning_rate": 4.674397933288765e-06,
      "loss": 0.6293,
      "step": 40060
    },
    {
      "epoch": 16.4,
      "grad_norm": 2.7440319061279297,
      "learning_rate": 4.674381178307855e-06,
      "loss": 0.6591,
      "step": 40070
    },
    {
      "epoch": 16.41,
      "grad_norm": 2.684281826019287,
      "learning_rate": 4.674364412650684e-06,
      "loss": 0.647,
      "step": 40080
    },
    {
      "epoch": 16.41,
      "grad_norm": 3.091235399246216,
      "learning_rate": 4.674347636317329e-06,
      "loss": 0.6224,
      "step": 40090
    },
    {
      "epoch": 16.41,
      "grad_norm": 2.959432601928711,
      "learning_rate": 4.674330849307868e-06,
      "loss": 0.6362,
      "step": 40100
    },
    {
      "epoch": 16.42,
      "grad_norm": 3.493481159210205,
      "learning_rate": 4.674314051622377e-06,
      "loss": 0.6171,
      "step": 40110
    },
    {
      "epoch": 16.42,
      "grad_norm": 2.6441779136657715,
      "learning_rate": 4.6742972432609324e-06,
      "loss": 0.6286,
      "step": 40120
    },
    {
      "epoch": 16.43,
      "grad_norm": 2.8310019969940186,
      "learning_rate": 4.674280424223613e-06,
      "loss": 0.6167,
      "step": 40130
    },
    {
      "epoch": 16.43,
      "grad_norm": 2.5857961177825928,
      "learning_rate": 4.674263594510493e-06,
      "loss": 0.6274,
      "step": 40140
    },
    {
      "epoch": 16.43,
      "grad_norm": 2.3608434200286865,
      "learning_rate": 4.674246754121653e-06,
      "loss": 0.6621,
      "step": 40150
    },
    {
      "epoch": 16.44,
      "grad_norm": 2.860731363296509,
      "learning_rate": 4.674229903057167e-06,
      "loss": 0.6428,
      "step": 40160
    },
    {
      "epoch": 16.44,
      "grad_norm": 2.7436795234680176,
      "learning_rate": 4.674213041317113e-06,
      "loss": 0.6151,
      "step": 40170
    },
    {
      "epoch": 16.45,
      "grad_norm": 3.6307241916656494,
      "learning_rate": 4.674196168901571e-06,
      "loss": 0.6094,
      "step": 40180
    },
    {
      "epoch": 16.45,
      "grad_norm": 3.038180351257324,
      "learning_rate": 4.674179285810614e-06,
      "loss": 0.6197,
      "step": 40190
    },
    {
      "epoch": 16.46,
      "grad_norm": 2.0962319374084473,
      "learning_rate": 4.674162392044322e-06,
      "loss": 0.6377,
      "step": 40200
    },
    {
      "epoch": 16.46,
      "grad_norm": 2.056903600692749,
      "learning_rate": 4.674145487602771e-06,
      "loss": 0.6285,
      "step": 40210
    },
    {
      "epoch": 16.46,
      "grad_norm": 2.5865702629089355,
      "learning_rate": 4.67412857248604e-06,
      "loss": 0.6365,
      "step": 40220
    },
    {
      "epoch": 16.47,
      "grad_norm": 2.966190814971924,
      "learning_rate": 4.674111646694205e-06,
      "loss": 0.6335,
      "step": 40230
    },
    {
      "epoch": 16.47,
      "grad_norm": 2.4834630489349365,
      "learning_rate": 4.674094710227345e-06,
      "loss": 0.6373,
      "step": 40240
    },
    {
      "epoch": 16.48,
      "grad_norm": 2.8888027667999268,
      "learning_rate": 4.6740777630855365e-06,
      "loss": 0.6065,
      "step": 40250
    },
    {
      "epoch": 16.48,
      "grad_norm": 3.4693241119384766,
      "learning_rate": 4.6740608052688574e-06,
      "loss": 0.6271,
      "step": 40260
    },
    {
      "epoch": 16.48,
      "grad_norm": 3.0739691257476807,
      "learning_rate": 4.6740438367773855e-06,
      "loss": 0.6258,
      "step": 40270
    },
    {
      "epoch": 16.49,
      "grad_norm": 2.8108625411987305,
      "learning_rate": 4.674026857611199e-06,
      "loss": 0.6341,
      "step": 40280
    },
    {
      "epoch": 16.49,
      "grad_norm": 2.6938838958740234,
      "learning_rate": 4.674009867770373e-06,
      "loss": 0.6341,
      "step": 40290
    },
    {
      "epoch": 16.5,
      "grad_norm": 2.3706955909729004,
      "learning_rate": 4.673992867254989e-06,
      "loss": 0.6355,
      "step": 40300
    },
    {
      "epoch": 16.5,
      "grad_norm": 3.5354833602905273,
      "learning_rate": 4.6739758560651234e-06,
      "loss": 0.6332,
      "step": 40310
    },
    {
      "epoch": 16.5,
      "grad_norm": 2.8727540969848633,
      "learning_rate": 4.673958834200854e-06,
      "loss": 0.6348,
      "step": 40320
    },
    {
      "epoch": 16.51,
      "grad_norm": 2.9338319301605225,
      "learning_rate": 4.673941801662259e-06,
      "loss": 0.635,
      "step": 40330
    },
    {
      "epoch": 16.51,
      "grad_norm": 3.037937879562378,
      "learning_rate": 4.673924758449415e-06,
      "loss": 0.6157,
      "step": 40340
    },
    {
      "epoch": 16.52,
      "grad_norm": 3.2044382095336914,
      "learning_rate": 4.673907704562403e-06,
      "loss": 0.6227,
      "step": 40350
    },
    {
      "epoch": 16.52,
      "grad_norm": 2.5161080360412598,
      "learning_rate": 4.673890640001298e-06,
      "loss": 0.6256,
      "step": 40360
    },
    {
      "epoch": 16.52,
      "grad_norm": 3.39228892326355,
      "learning_rate": 4.673873564766181e-06,
      "loss": 0.6425,
      "step": 40370
    },
    {
      "epoch": 16.53,
      "grad_norm": 2.9907917976379395,
      "learning_rate": 4.673856478857127e-06,
      "loss": 0.6508,
      "step": 40380
    },
    {
      "epoch": 16.53,
      "grad_norm": 3.1126906871795654,
      "learning_rate": 4.6738393822742176e-06,
      "loss": 0.6212,
      "step": 40390
    },
    {
      "epoch": 16.54,
      "grad_norm": 2.6153972148895264,
      "learning_rate": 4.673822275017529e-06,
      "loss": 0.6469,
      "step": 40400
    },
    {
      "epoch": 16.54,
      "grad_norm": 2.3463079929351807,
      "learning_rate": 4.67380515708714e-06,
      "loss": 0.6392,
      "step": 40410
    },
    {
      "epoch": 16.55,
      "grad_norm": 2.346590757369995,
      "learning_rate": 4.673788028483129e-06,
      "loss": 0.636,
      "step": 40420
    },
    {
      "epoch": 16.55,
      "grad_norm": 2.077249526977539,
      "learning_rate": 4.6737708892055755e-06,
      "loss": 0.6363,
      "step": 40430
    },
    {
      "epoch": 16.55,
      "grad_norm": 3.046227216720581,
      "learning_rate": 4.673753739254556e-06,
      "loss": 0.6473,
      "step": 40440
    },
    {
      "epoch": 16.56,
      "grad_norm": 2.426783561706543,
      "learning_rate": 4.673736578630151e-06,
      "loss": 0.6373,
      "step": 40450
    },
    {
      "epoch": 16.56,
      "grad_norm": 1.9311848878860474,
      "learning_rate": 4.673719407332438e-06,
      "loss": 0.6495,
      "step": 40460
    },
    {
      "epoch": 16.57,
      "grad_norm": 2.5166404247283936,
      "learning_rate": 4.6737022253614965e-06,
      "loss": 0.621,
      "step": 40470
    },
    {
      "epoch": 16.57,
      "grad_norm": 3.290925979614258,
      "learning_rate": 4.6736850327174045e-06,
      "loss": 0.6342,
      "step": 40480
    },
    {
      "epoch": 16.57,
      "grad_norm": 3.477708578109741,
      "learning_rate": 4.673667829400242e-06,
      "loss": 0.6197,
      "step": 40490
    },
    {
      "epoch": 16.58,
      "grad_norm": 2.8881311416625977,
      "learning_rate": 4.673650615410084e-06,
      "loss": 0.635,
      "step": 40500
    },
    {
      "epoch": 16.58,
      "grad_norm": 3.412010908126831,
      "learning_rate": 4.673633390747014e-06,
      "loss": 0.6294,
      "step": 40510
    },
    {
      "epoch": 16.59,
      "grad_norm": 2.7617568969726562,
      "learning_rate": 4.673616155411109e-06,
      "loss": 0.6149,
      "step": 40520
    },
    {
      "epoch": 16.59,
      "grad_norm": 1.9163484573364258,
      "learning_rate": 4.673598909402448e-06,
      "loss": 0.6213,
      "step": 40530
    },
    {
      "epoch": 16.59,
      "grad_norm": 2.5813655853271484,
      "learning_rate": 4.673581652721109e-06,
      "loss": 0.619,
      "step": 40540
    },
    {
      "epoch": 16.6,
      "grad_norm": 1.954668641090393,
      "learning_rate": 4.673564385367173e-06,
      "loss": 0.6458,
      "step": 40550
    },
    {
      "epoch": 16.6,
      "grad_norm": 1.9352928400039673,
      "learning_rate": 4.673547107340718e-06,
      "loss": 0.6402,
      "step": 40560
    },
    {
      "epoch": 16.61,
      "grad_norm": 1.8504124879837036,
      "learning_rate": 4.673529818641823e-06,
      "loss": 0.6421,
      "step": 40570
    },
    {
      "epoch": 16.61,
      "grad_norm": 2.493896722793579,
      "learning_rate": 4.673512519270567e-06,
      "loss": 0.6225,
      "step": 40580
    },
    {
      "epoch": 16.61,
      "grad_norm": 2.6711435317993164,
      "learning_rate": 4.67349520922703e-06,
      "loss": 0.6525,
      "step": 40590
    },
    {
      "epoch": 16.62,
      "grad_norm": 2.4250078201293945,
      "learning_rate": 4.673477888511292e-06,
      "loss": 0.6344,
      "step": 40600
    },
    {
      "epoch": 16.62,
      "grad_norm": 3.211810827255249,
      "learning_rate": 4.673460557123431e-06,
      "loss": 0.6387,
      "step": 40610
    },
    {
      "epoch": 16.63,
      "grad_norm": 2.5269486904144287,
      "learning_rate": 4.6734432150635255e-06,
      "loss": 0.6356,
      "step": 40620
    },
    {
      "epoch": 16.63,
      "grad_norm": 3.310922861099243,
      "learning_rate": 4.6734258623316575e-06,
      "loss": 0.5961,
      "step": 40630
    },
    {
      "epoch": 16.64,
      "grad_norm": 2.248422622680664,
      "learning_rate": 4.673408498927904e-06,
      "loss": 0.633,
      "step": 40640
    },
    {
      "epoch": 16.64,
      "grad_norm": 3.743500232696533,
      "learning_rate": 4.673391124852347e-06,
      "loss": 0.6454,
      "step": 40650
    },
    {
      "epoch": 16.64,
      "grad_norm": 3.6572775840759277,
      "learning_rate": 4.673373740105064e-06,
      "loss": 0.6268,
      "step": 40660
    },
    {
      "epoch": 16.65,
      "grad_norm": 2.703291654586792,
      "learning_rate": 4.6733563446861365e-06,
      "loss": 0.6362,
      "step": 40670
    },
    {
      "epoch": 16.65,
      "grad_norm": 2.7020528316497803,
      "learning_rate": 4.673338938595641e-06,
      "loss": 0.6245,
      "step": 40680
    },
    {
      "epoch": 16.66,
      "grad_norm": 2.3137667179107666,
      "learning_rate": 4.673321521833661e-06,
      "loss": 0.6205,
      "step": 40690
    },
    {
      "epoch": 16.66,
      "grad_norm": 3.062298059463501,
      "learning_rate": 4.673304094400274e-06,
      "loss": 0.6272,
      "step": 40700
    },
    {
      "epoch": 16.66,
      "grad_norm": 1.7908852100372314,
      "learning_rate": 4.673286656295562e-06,
      "loss": 0.6386,
      "step": 40710
    },
    {
      "epoch": 16.67,
      "grad_norm": 3.2941861152648926,
      "learning_rate": 4.673269207519601e-06,
      "loss": 0.6212,
      "step": 40720
    },
    {
      "epoch": 16.67,
      "grad_norm": 2.971942663192749,
      "learning_rate": 4.673251748072474e-06,
      "loss": 0.6296,
      "step": 40730
    },
    {
      "epoch": 16.68,
      "grad_norm": 3.000331163406372,
      "learning_rate": 4.6732342779542615e-06,
      "loss": 0.6332,
      "step": 40740
    },
    {
      "epoch": 16.68,
      "grad_norm": 2.7669239044189453,
      "learning_rate": 4.673216797165041e-06,
      "loss": 0.6168,
      "step": 40750
    },
    {
      "epoch": 16.68,
      "grad_norm": 2.8148610591888428,
      "learning_rate": 4.673199305704895e-06,
      "loss": 0.6311,
      "step": 40760
    },
    {
      "epoch": 16.69,
      "grad_norm": 3.056281089782715,
      "learning_rate": 4.673181803573901e-06,
      "loss": 0.659,
      "step": 40770
    },
    {
      "epoch": 16.69,
      "grad_norm": 3.009730339050293,
      "learning_rate": 4.673164290772141e-06,
      "loss": 0.632,
      "step": 40780
    },
    {
      "epoch": 16.7,
      "grad_norm": 3.5450830459594727,
      "learning_rate": 4.673146767299696e-06,
      "loss": 0.6105,
      "step": 40790
    },
    {
      "epoch": 16.7,
      "grad_norm": 2.9563255310058594,
      "learning_rate": 4.673129233156644e-06,
      "loss": 0.6364,
      "step": 40800
    },
    {
      "epoch": 16.7,
      "grad_norm": 2.501181125640869,
      "learning_rate": 4.673111688343067e-06,
      "loss": 0.5985,
      "step": 40810
    },
    {
      "epoch": 16.71,
      "grad_norm": 3.0818252563476562,
      "learning_rate": 4.673094132859045e-06,
      "loss": 0.6448,
      "step": 40820
    },
    {
      "epoch": 16.71,
      "grad_norm": 2.1727187633514404,
      "learning_rate": 4.673076566704658e-06,
      "loss": 0.6385,
      "step": 40830
    },
    {
      "epoch": 16.72,
      "grad_norm": 2.4796266555786133,
      "learning_rate": 4.673058989879987e-06,
      "loss": 0.6354,
      "step": 40840
    },
    {
      "epoch": 16.72,
      "grad_norm": 3.8527441024780273,
      "learning_rate": 4.673041402385112e-06,
      "loss": 0.6277,
      "step": 40850
    },
    {
      "epoch": 16.73,
      "grad_norm": 2.5642917156219482,
      "learning_rate": 4.6730238042201146e-06,
      "loss": 0.6268,
      "step": 40860
    },
    {
      "epoch": 16.73,
      "grad_norm": 2.740166425704956,
      "learning_rate": 4.673006195385073e-06,
      "loss": 0.6527,
      "step": 40870
    },
    {
      "epoch": 16.73,
      "grad_norm": 2.6636719703674316,
      "learning_rate": 4.67298857588007e-06,
      "loss": 0.6379,
      "step": 40880
    },
    {
      "epoch": 16.74,
      "grad_norm": 2.7293097972869873,
      "learning_rate": 4.672970945705186e-06,
      "loss": 0.6463,
      "step": 40890
    },
    {
      "epoch": 16.74,
      "grad_norm": 2.6560068130493164,
      "learning_rate": 4.6729533048605025e-06,
      "loss": 0.6109,
      "step": 40900
    },
    {
      "epoch": 16.75,
      "grad_norm": 2.5107154846191406,
      "learning_rate": 4.672935653346099e-06,
      "loss": 0.6118,
      "step": 40910
    },
    {
      "epoch": 16.75,
      "grad_norm": 2.464170455932617,
      "learning_rate": 4.6729179911620555e-06,
      "loss": 0.6173,
      "step": 40920
    },
    {
      "epoch": 16.75,
      "grad_norm": 3.093353509902954,
      "learning_rate": 4.672900318308456e-06,
      "loss": 0.6324,
      "step": 40930
    },
    {
      "epoch": 16.76,
      "grad_norm": 2.6310336589813232,
      "learning_rate": 4.672882634785378e-06,
      "loss": 0.6364,
      "step": 40940
    },
    {
      "epoch": 16.76,
      "grad_norm": 2.585989475250244,
      "learning_rate": 4.672864940592905e-06,
      "loss": 0.6301,
      "step": 40950
    },
    {
      "epoch": 16.77,
      "grad_norm": 2.1008803844451904,
      "learning_rate": 4.672847235731118e-06,
      "loss": 0.6149,
      "step": 40960
    },
    {
      "epoch": 16.77,
      "grad_norm": 2.329951286315918,
      "learning_rate": 4.672829520200095e-06,
      "loss": 0.6392,
      "step": 40970
    },
    {
      "epoch": 16.77,
      "grad_norm": 2.4371306896209717,
      "learning_rate": 4.672811793999921e-06,
      "loss": 0.6559,
      "step": 40980
    },
    {
      "epoch": 16.78,
      "grad_norm": 2.9164509773254395,
      "learning_rate": 4.6727940571306755e-06,
      "loss": 0.6183,
      "step": 40990
    },
    {
      "epoch": 16.78,
      "grad_norm": 2.3859641551971436,
      "learning_rate": 4.67277630959244e-06,
      "loss": 0.6462,
      "step": 41000
    },
    {
      "epoch": 16.79,
      "grad_norm": 3.2888410091400146,
      "learning_rate": 4.672758551385295e-06,
      "loss": 0.6085,
      "step": 41010
    },
    {
      "epoch": 16.79,
      "grad_norm": 2.4154438972473145,
      "learning_rate": 4.672740782509323e-06,
      "loss": 0.6202,
      "step": 41020
    },
    {
      "epoch": 16.79,
      "grad_norm": 2.360740900039673,
      "learning_rate": 4.672723002964605e-06,
      "loss": 0.6159,
      "step": 41030
    },
    {
      "epoch": 16.8,
      "grad_norm": 3.4106767177581787,
      "learning_rate": 4.672705212751222e-06,
      "loss": 0.6473,
      "step": 41040
    },
    {
      "epoch": 16.8,
      "grad_norm": 3.0249481201171875,
      "learning_rate": 4.672687411869256e-06,
      "loss": 0.6329,
      "step": 41050
    },
    {
      "epoch": 16.81,
      "grad_norm": 3.7006819248199463,
      "learning_rate": 4.6726696003187885e-06,
      "loss": 0.6293,
      "step": 41060
    },
    {
      "epoch": 16.81,
      "grad_norm": 2.4615135192871094,
      "learning_rate": 4.6726517780999e-06,
      "loss": 0.6196,
      "step": 41070
    },
    {
      "epoch": 16.82,
      "grad_norm": 2.1946604251861572,
      "learning_rate": 4.6726339452126745e-06,
      "loss": 0.6252,
      "step": 41080
    },
    {
      "epoch": 16.82,
      "grad_norm": 2.9533843994140625,
      "learning_rate": 4.6726161016571915e-06,
      "loss": 0.643,
      "step": 41090
    },
    {
      "epoch": 16.82,
      "grad_norm": 2.699117422103882,
      "learning_rate": 4.6725982474335335e-06,
      "loss": 0.6089,
      "step": 41100
    },
    {
      "epoch": 16.83,
      "grad_norm": 2.5780181884765625,
      "learning_rate": 4.672580382541784e-06,
      "loss": 0.6175,
      "step": 41110
    },
    {
      "epoch": 16.83,
      "grad_norm": 1.9123973846435547,
      "learning_rate": 4.672562506982021e-06,
      "loss": 0.6371,
      "step": 41120
    },
    {
      "epoch": 16.84,
      "grad_norm": 2.006287097930908,
      "learning_rate": 4.67254462075433e-06,
      "loss": 0.63,
      "step": 41130
    },
    {
      "epoch": 16.84,
      "grad_norm": 1.8381320238113403,
      "learning_rate": 4.672526723858791e-06,
      "loss": 0.6356,
      "step": 41140
    },
    {
      "epoch": 16.84,
      "grad_norm": 3.3069005012512207,
      "learning_rate": 4.672508816295486e-06,
      "loss": 0.6341,
      "step": 41150
    },
    {
      "epoch": 16.85,
      "grad_norm": 2.620209217071533,
      "learning_rate": 4.672490898064498e-06,
      "loss": 0.6244,
      "step": 41160
    },
    {
      "epoch": 16.85,
      "grad_norm": 2.6476645469665527,
      "learning_rate": 4.672472969165909e-06,
      "loss": 0.6175,
      "step": 41170
    },
    {
      "epoch": 16.86,
      "grad_norm": 1.979089379310608,
      "learning_rate": 4.672455029599799e-06,
      "loss": 0.6504,
      "step": 41180
    },
    {
      "epoch": 16.86,
      "grad_norm": 1.8788820505142212,
      "learning_rate": 4.672437079366254e-06,
      "loss": 0.6299,
      "step": 41190
    },
    {
      "epoch": 16.86,
      "grad_norm": 1.9548617601394653,
      "learning_rate": 4.672419118465354e-06,
      "loss": 0.6216,
      "step": 41200
    },
    {
      "epoch": 16.87,
      "grad_norm": 2.8851242065429688,
      "learning_rate": 4.67240114689718e-06,
      "loss": 0.6405,
      "step": 41210
    },
    {
      "epoch": 16.87,
      "grad_norm": 2.472264528274536,
      "learning_rate": 4.672383164661817e-06,
      "loss": 0.6304,
      "step": 41220
    },
    {
      "epoch": 16.88,
      "grad_norm": 2.174154281616211,
      "learning_rate": 4.672365171759345e-06,
      "loss": 0.6456,
      "step": 41230
    },
    {
      "epoch": 16.88,
      "grad_norm": 2.53020977973938,
      "learning_rate": 4.672347168189848e-06,
      "loss": 0.6386,
      "step": 41240
    },
    {
      "epoch": 16.88,
      "grad_norm": 2.2868285179138184,
      "learning_rate": 4.672329153953408e-06,
      "loss": 0.6274,
      "step": 41250
    },
    {
      "epoch": 16.89,
      "grad_norm": 3.418560743331909,
      "learning_rate": 4.672311129050108e-06,
      "loss": 0.6357,
      "step": 41260
    },
    {
      "epoch": 16.89,
      "grad_norm": 2.410273551940918,
      "learning_rate": 4.67229309348003e-06,
      "loss": 0.6333,
      "step": 41270
    },
    {
      "epoch": 16.9,
      "grad_norm": 2.809844970703125,
      "learning_rate": 4.672275047243256e-06,
      "loss": 0.6368,
      "step": 41280
    },
    {
      "epoch": 16.9,
      "grad_norm": 2.824192523956299,
      "learning_rate": 4.6722569903398705e-06,
      "loss": 0.6293,
      "step": 41290
    },
    {
      "epoch": 16.91,
      "grad_norm": 2.099339246749878,
      "learning_rate": 4.672238922769954e-06,
      "loss": 0.6116,
      "step": 41300
    },
    {
      "epoch": 16.91,
      "grad_norm": 2.4553298950195312,
      "learning_rate": 4.672220844533591e-06,
      "loss": 0.6083,
      "step": 41310
    },
    {
      "epoch": 16.91,
      "grad_norm": 2.4075536727905273,
      "learning_rate": 4.672202755630863e-06,
      "loss": 0.6313,
      "step": 41320
    },
    {
      "epoch": 16.92,
      "grad_norm": 2.5116493701934814,
      "learning_rate": 4.672184656061854e-06,
      "loss": 0.6091,
      "step": 41330
    },
    {
      "epoch": 16.92,
      "grad_norm": 3.6368319988250732,
      "learning_rate": 4.6721665458266466e-06,
      "loss": 0.6127,
      "step": 41340
    },
    {
      "epoch": 16.93,
      "grad_norm": 2.905099391937256,
      "learning_rate": 4.672148424925323e-06,
      "loss": 0.6405,
      "step": 41350
    },
    {
      "epoch": 16.93,
      "grad_norm": 2.792470932006836,
      "learning_rate": 4.672130293357967e-06,
      "loss": 0.6297,
      "step": 41360
    },
    {
      "epoch": 16.93,
      "grad_norm": 2.314438581466675,
      "learning_rate": 4.672112151124662e-06,
      "loss": 0.6105,
      "step": 41370
    },
    {
      "epoch": 16.94,
      "grad_norm": 3.5689537525177,
      "learning_rate": 4.6720939982254905e-06,
      "loss": 0.6105,
      "step": 41380
    },
    {
      "epoch": 16.94,
      "grad_norm": 2.5622265338897705,
      "learning_rate": 4.672075834660536e-06,
      "loss": 0.6326,
      "step": 41390
    },
    {
      "epoch": 16.95,
      "grad_norm": 2.4021966457366943,
      "learning_rate": 4.67205766042988e-06,
      "loss": 0.6412,
      "step": 41400
    },
    {
      "epoch": 16.95,
      "grad_norm": 2.208129644393921,
      "learning_rate": 4.672039475533608e-06,
      "loss": 0.6312,
      "step": 41410
    },
    {
      "epoch": 16.95,
      "grad_norm": 2.675083637237549,
      "learning_rate": 4.672021279971803e-06,
      "loss": 0.614,
      "step": 41420
    },
    {
      "epoch": 16.96,
      "grad_norm": 2.0750160217285156,
      "learning_rate": 4.672003073744548e-06,
      "loss": 0.6302,
      "step": 41430
    },
    {
      "epoch": 16.96,
      "grad_norm": 2.975332260131836,
      "learning_rate": 4.671984856851925e-06,
      "loss": 0.6545,
      "step": 41440
    },
    {
      "epoch": 16.97,
      "grad_norm": 2.5062873363494873,
      "learning_rate": 4.67196662929402e-06,
      "loss": 0.6254,
      "step": 41450
    },
    {
      "epoch": 16.97,
      "grad_norm": 2.636556386947632,
      "learning_rate": 4.6719483910709144e-06,
      "loss": 0.6456,
      "step": 41460
    },
    {
      "epoch": 16.98,
      "grad_norm": 3.3247101306915283,
      "learning_rate": 4.671930142182693e-06,
      "loss": 0.6273,
      "step": 41470
    },
    {
      "epoch": 16.98,
      "grad_norm": 2.655284881591797,
      "learning_rate": 4.6719118826294385e-06,
      "loss": 0.6363,
      "step": 41480
    },
    {
      "epoch": 16.98,
      "grad_norm": 3.3021459579467773,
      "learning_rate": 4.671893612411235e-06,
      "loss": 0.6379,
      "step": 41490
    },
    {
      "epoch": 16.99,
      "grad_norm": 2.078291177749634,
      "learning_rate": 4.671875331528166e-06,
      "loss": 0.6389,
      "step": 41500
    },
    {
      "epoch": 16.99,
      "grad_norm": 2.729867696762085,
      "learning_rate": 4.671857039980316e-06,
      "loss": 0.6256,
      "step": 41510
    },
    {
      "epoch": 17.0,
      "grad_norm": 3.2257533073425293,
      "learning_rate": 4.671838737767768e-06,
      "loss": 0.6387,
      "step": 41520
    },
    {
      "epoch": 17.0,
      "grad_norm": 2.788879871368408,
      "learning_rate": 4.671820424890606e-06,
      "loss": 0.616,
      "step": 41530
    },
    {
      "epoch": 17.0,
      "eval_loss": 0.6247778534889221,
      "eval_runtime": 51.9252,
      "eval_samples_per_second": 66.423,
      "eval_steps_per_second": 8.32,
      "step": 41531
    },
    {
      "epoch": 17.0,
      "grad_norm": 2.625559091567993,
      "learning_rate": 4.671802101348914e-06,
      "loss": 0.635,
      "step": 41540
    },
    {
      "epoch": 17.01,
      "grad_norm": 2.8129303455352783,
      "learning_rate": 4.671783767142775e-06,
      "loss": 0.6296,
      "step": 41550
    },
    {
      "epoch": 17.01,
      "grad_norm": 2.82104754447937,
      "learning_rate": 4.671765422272275e-06,
      "loss": 0.608,
      "step": 41560
    },
    {
      "epoch": 17.02,
      "grad_norm": 2.9449517726898193,
      "learning_rate": 4.671747066737497e-06,
      "loss": 0.6274,
      "step": 41570
    },
    {
      "epoch": 17.02,
      "grad_norm": 2.1526966094970703,
      "learning_rate": 4.671728700538524e-06,
      "loss": 0.6255,
      "step": 41580
    },
    {
      "epoch": 17.02,
      "grad_norm": 4.146010875701904,
      "learning_rate": 4.671710323675441e-06,
      "loss": 0.6135,
      "step": 41590
    },
    {
      "epoch": 17.03,
      "grad_norm": 2.5624043941497803,
      "learning_rate": 4.671691936148333e-06,
      "loss": 0.6518,
      "step": 41600
    },
    {
      "epoch": 17.03,
      "grad_norm": 2.8378078937530518,
      "learning_rate": 4.671673537957284e-06,
      "loss": 0.6321,
      "step": 41610
    },
    {
      "epoch": 17.04,
      "grad_norm": 2.7883551120758057,
      "learning_rate": 4.671655129102377e-06,
      "loss": 0.6288,
      "step": 41620
    },
    {
      "epoch": 17.04,
      "grad_norm": 4.437140941619873,
      "learning_rate": 4.6716367095836976e-06,
      "loss": 0.605,
      "step": 41630
    },
    {
      "epoch": 17.04,
      "grad_norm": 3.263578176498413,
      "learning_rate": 4.6716182794013295e-06,
      "loss": 0.6272,
      "step": 41640
    },
    {
      "epoch": 17.05,
      "grad_norm": 3.043933629989624,
      "learning_rate": 4.671599838555357e-06,
      "loss": 0.6358,
      "step": 41650
    },
    {
      "epoch": 17.05,
      "grad_norm": 2.09592866897583,
      "learning_rate": 4.671581387045865e-06,
      "loss": 0.6352,
      "step": 41660
    },
    {
      "epoch": 17.06,
      "grad_norm": 1.9605940580368042,
      "learning_rate": 4.671562924872938e-06,
      "loss": 0.6298,
      "step": 41670
    },
    {
      "epoch": 17.06,
      "grad_norm": 2.0397441387176514,
      "learning_rate": 4.671544452036661e-06,
      "loss": 0.6424,
      "step": 41680
    },
    {
      "epoch": 17.07,
      "grad_norm": 2.2524173259735107,
      "learning_rate": 4.671525968537117e-06,
      "loss": 0.6252,
      "step": 41690
    },
    {
      "epoch": 17.07,
      "grad_norm": 2.4913909435272217,
      "learning_rate": 4.6715074743743935e-06,
      "loss": 0.6287,
      "step": 41700
    },
    {
      "epoch": 17.07,
      "grad_norm": 3.0758097171783447,
      "learning_rate": 4.671488969548572e-06,
      "loss": 0.6079,
      "step": 41710
    },
    {
      "epoch": 17.08,
      "grad_norm": 2.9529647827148438,
      "learning_rate": 4.671470454059739e-06,
      "loss": 0.626,
      "step": 41720
    },
    {
      "epoch": 17.08,
      "grad_norm": 3.1184799671173096,
      "learning_rate": 4.671451927907979e-06,
      "loss": 0.6272,
      "step": 41730
    },
    {
      "epoch": 17.09,
      "grad_norm": 3.975536823272705,
      "learning_rate": 4.671433391093378e-06,
      "loss": 0.6084,
      "step": 41740
    },
    {
      "epoch": 17.09,
      "grad_norm": 2.111569404602051,
      "learning_rate": 4.671414843616019e-06,
      "loss": 0.6251,
      "step": 41750
    },
    {
      "epoch": 17.09,
      "grad_norm": 3.151411771774292,
      "learning_rate": 4.671396285475988e-06,
      "loss": 0.6193,
      "step": 41760
    },
    {
      "epoch": 17.1,
      "grad_norm": 3.4067704677581787,
      "learning_rate": 4.67137771667337e-06,
      "loss": 0.5981,
      "step": 41770
    },
    {
      "epoch": 17.1,
      "grad_norm": 2.901466131210327,
      "learning_rate": 4.67135913720825e-06,
      "loss": 0.6151,
      "step": 41780
    },
    {
      "epoch": 17.11,
      "grad_norm": 3.7907955646514893,
      "learning_rate": 4.671340547080713e-06,
      "loss": 0.6241,
      "step": 41790
    },
    {
      "epoch": 17.11,
      "grad_norm": 2.93078875541687,
      "learning_rate": 4.6713219462908434e-06,
      "loss": 0.6273,
      "step": 41800
    },
    {
      "epoch": 17.11,
      "grad_norm": 2.4878592491149902,
      "learning_rate": 4.671303334838728e-06,
      "loss": 0.6284,
      "step": 41810
    },
    {
      "epoch": 17.12,
      "grad_norm": 2.6600046157836914,
      "learning_rate": 4.671284712724451e-06,
      "loss": 0.6269,
      "step": 41820
    },
    {
      "epoch": 17.12,
      "grad_norm": 2.880016565322876,
      "learning_rate": 4.6712660799480975e-06,
      "loss": 0.6395,
      "step": 41830
    },
    {
      "epoch": 17.13,
      "grad_norm": 3.1200199127197266,
      "learning_rate": 4.6712474365097546e-06,
      "loss": 0.6418,
      "step": 41840
    },
    {
      "epoch": 17.13,
      "grad_norm": 2.49627685546875,
      "learning_rate": 4.6712287824095056e-06,
      "loss": 0.6397,
      "step": 41850
    },
    {
      "epoch": 17.13,
      "grad_norm": 3.141289472579956,
      "learning_rate": 4.671210117647436e-06,
      "loss": 0.605,
      "step": 41860
    },
    {
      "epoch": 17.14,
      "grad_norm": 2.8338205814361572,
      "learning_rate": 4.671191442223633e-06,
      "loss": 0.6243,
      "step": 41870
    },
    {
      "epoch": 17.14,
      "grad_norm": 2.9247958660125732,
      "learning_rate": 4.671172756138181e-06,
      "loss": 0.6187,
      "step": 41880
    },
    {
      "epoch": 17.15,
      "grad_norm": 3.2263875007629395,
      "learning_rate": 4.6711540593911655e-06,
      "loss": 0.6204,
      "step": 41890
    },
    {
      "epoch": 17.15,
      "grad_norm": 2.5973522663116455,
      "learning_rate": 4.671135351982673e-06,
      "loss": 0.6227,
      "step": 41900
    },
    {
      "epoch": 17.16,
      "grad_norm": 4.824460983276367,
      "learning_rate": 4.671116633912788e-06,
      "loss": 0.6286,
      "step": 41910
    },
    {
      "epoch": 17.16,
      "grad_norm": 3.433946132659912,
      "learning_rate": 4.6710979051815975e-06,
      "loss": 0.5993,
      "step": 41920
    },
    {
      "epoch": 17.16,
      "grad_norm": 2.7653136253356934,
      "learning_rate": 4.671079165789186e-06,
      "loss": 0.6184,
      "step": 41930
    },
    {
      "epoch": 17.17,
      "grad_norm": 2.5552315711975098,
      "learning_rate": 4.6710604157356415e-06,
      "loss": 0.6252,
      "step": 41940
    },
    {
      "epoch": 17.17,
      "grad_norm": 2.5415596961975098,
      "learning_rate": 4.671041655021047e-06,
      "loss": 0.6054,
      "step": 41950
    },
    {
      "epoch": 17.18,
      "grad_norm": 2.649989366531372,
      "learning_rate": 4.67102288364549e-06,
      "loss": 0.6596,
      "step": 41960
    },
    {
      "epoch": 17.18,
      "grad_norm": 3.5396556854248047,
      "learning_rate": 4.671004101609057e-06,
      "loss": 0.6144,
      "step": 41970
    },
    {
      "epoch": 17.18,
      "grad_norm": 3.1518871784210205,
      "learning_rate": 4.670985308911832e-06,
      "loss": 0.6184,
      "step": 41980
    },
    {
      "epoch": 17.19,
      "grad_norm": 2.8564388751983643,
      "learning_rate": 4.670966505553904e-06,
      "loss": 0.6274,
      "step": 41990
    },
    {
      "epoch": 17.19,
      "grad_norm": 2.847649097442627,
      "learning_rate": 4.670947691535357e-06,
      "loss": 0.6305,
      "step": 42000
    },
    {
      "epoch": 17.2,
      "grad_norm": 2.6876564025878906,
      "learning_rate": 4.670928866856277e-06,
      "loss": 0.6202,
      "step": 42010
    },
    {
      "epoch": 17.2,
      "grad_norm": 2.3393521308898926,
      "learning_rate": 4.670910031516753e-06,
      "loss": 0.6222,
      "step": 42020
    },
    {
      "epoch": 17.2,
      "grad_norm": 2.182497262954712,
      "learning_rate": 4.670891185516867e-06,
      "loss": 0.6356,
      "step": 42030
    },
    {
      "epoch": 17.21,
      "grad_norm": 2.470573902130127,
      "learning_rate": 4.670872328856709e-06,
      "loss": 0.6256,
      "step": 42040
    },
    {
      "epoch": 17.21,
      "grad_norm": 2.7409424781799316,
      "learning_rate": 4.670853461536364e-06,
      "loss": 0.6193,
      "step": 42050
    },
    {
      "epoch": 17.22,
      "grad_norm": 2.5860419273376465,
      "learning_rate": 4.670834583555919e-06,
      "loss": 0.6288,
      "step": 42060
    },
    {
      "epoch": 17.22,
      "grad_norm": 3.186702013015747,
      "learning_rate": 4.670815694915459e-06,
      "loss": 0.6148,
      "step": 42070
    },
    {
      "epoch": 17.22,
      "grad_norm": 3.1170589923858643,
      "learning_rate": 4.670796795615071e-06,
      "loss": 0.6434,
      "step": 42080
    },
    {
      "epoch": 17.23,
      "grad_norm": 3.295675277709961,
      "learning_rate": 4.670777885654842e-06,
      "loss": 0.6333,
      "step": 42090
    },
    {
      "epoch": 17.23,
      "grad_norm": 2.2935972213745117,
      "learning_rate": 4.670758965034859e-06,
      "loss": 0.6389,
      "step": 42100
    },
    {
      "epoch": 17.24,
      "grad_norm": 2.249312400817871,
      "learning_rate": 4.670740033755209e-06,
      "loss": 0.6351,
      "step": 42110
    },
    {
      "epoch": 17.24,
      "grad_norm": 2.730619192123413,
      "learning_rate": 4.670721091815978e-06,
      "loss": 0.6242,
      "step": 42120
    },
    {
      "epoch": 17.25,
      "grad_norm": 3.3609354496002197,
      "learning_rate": 4.670702139217252e-06,
      "loss": 0.6369,
      "step": 42130
    },
    {
      "epoch": 17.25,
      "grad_norm": 2.0149714946746826,
      "learning_rate": 4.67068317595912e-06,
      "loss": 0.6252,
      "step": 42140
    },
    {
      "epoch": 17.25,
      "grad_norm": 2.4016058444976807,
      "learning_rate": 4.670664202041665e-06,
      "loss": 0.6295,
      "step": 42150
    },
    {
      "epoch": 17.26,
      "grad_norm": 2.516399383544922,
      "learning_rate": 4.6706452174649786e-06,
      "loss": 0.6031,
      "step": 42160
    },
    {
      "epoch": 17.26,
      "grad_norm": 2.850085496902466,
      "learning_rate": 4.670626222229145e-06,
      "loss": 0.635,
      "step": 42170
    },
    {
      "epoch": 17.27,
      "grad_norm": 2.2770707607269287,
      "learning_rate": 4.670607216334252e-06,
      "loss": 0.6272,
      "step": 42180
    },
    {
      "epoch": 17.27,
      "grad_norm": 3.7687036991119385,
      "learning_rate": 4.670588199780386e-06,
      "loss": 0.6229,
      "step": 42190
    },
    {
      "epoch": 17.27,
      "grad_norm": 1.9274401664733887,
      "learning_rate": 4.670569172567634e-06,
      "loss": 0.6495,
      "step": 42200
    },
    {
      "epoch": 17.28,
      "grad_norm": 2.8491783142089844,
      "learning_rate": 4.6705501346960844e-06,
      "loss": 0.6164,
      "step": 42210
    },
    {
      "epoch": 17.28,
      "grad_norm": 2.6286022663116455,
      "learning_rate": 4.670531086165824e-06,
      "loss": 0.6065,
      "step": 42220
    },
    {
      "epoch": 17.29,
      "grad_norm": 2.9341838359832764,
      "learning_rate": 4.67051202697694e-06,
      "loss": 0.6222,
      "step": 42230
    },
    {
      "epoch": 17.29,
      "grad_norm": 2.9625937938690186,
      "learning_rate": 4.670492957129518e-06,
      "loss": 0.6174,
      "step": 42240
    },
    {
      "epoch": 17.29,
      "grad_norm": 2.840226411819458,
      "learning_rate": 4.6704738766236484e-06,
      "loss": 0.5971,
      "step": 42250
    },
    {
      "epoch": 17.3,
      "grad_norm": 2.121798276901245,
      "learning_rate": 4.6704547854594165e-06,
      "loss": 0.6362,
      "step": 42260
    },
    {
      "epoch": 17.3,
      "grad_norm": 3.3375189304351807,
      "learning_rate": 4.6704356836369105e-06,
      "loss": 0.6282,
      "step": 42270
    },
    {
      "epoch": 17.31,
      "grad_norm": 2.9978880882263184,
      "learning_rate": 4.670416571156217e-06,
      "loss": 0.6393,
      "step": 42280
    },
    {
      "epoch": 17.31,
      "grad_norm": 2.8358500003814697,
      "learning_rate": 4.6703974480174254e-06,
      "loss": 0.6184,
      "step": 42290
    },
    {
      "epoch": 17.31,
      "grad_norm": 3.8797905445098877,
      "learning_rate": 4.670378314220622e-06,
      "loss": 0.6144,
      "step": 42300
    },
    {
      "epoch": 17.32,
      "grad_norm": 2.5733933448791504,
      "learning_rate": 4.670359169765894e-06,
      "loss": 0.6172,
      "step": 42310
    },
    {
      "epoch": 17.32,
      "grad_norm": 3.1839606761932373,
      "learning_rate": 4.67034001465333e-06,
      "loss": 0.6156,
      "step": 42320
    },
    {
      "epoch": 17.33,
      "grad_norm": 2.7866809368133545,
      "learning_rate": 4.670320848883017e-06,
      "loss": 0.6205,
      "step": 42330
    },
    {
      "epoch": 17.33,
      "grad_norm": 2.2720675468444824,
      "learning_rate": 4.670301672455044e-06,
      "loss": 0.6217,
      "step": 42340
    },
    {
      "epoch": 17.34,
      "grad_norm": 2.595435619354248,
      "learning_rate": 4.670282485369498e-06,
      "loss": 0.6079,
      "step": 42350
    },
    {
      "epoch": 17.34,
      "grad_norm": 2.0351054668426514,
      "learning_rate": 4.670263287626467e-06,
      "loss": 0.6221,
      "step": 42360
    },
    {
      "epoch": 17.34,
      "grad_norm": 1.976860523223877,
      "learning_rate": 4.6702440792260385e-06,
      "loss": 0.609,
      "step": 42370
    },
    {
      "epoch": 17.35,
      "grad_norm": 2.255293130874634,
      "learning_rate": 4.6702248601683e-06,
      "loss": 0.6288,
      "step": 42380
    },
    {
      "epoch": 17.35,
      "grad_norm": 1.8683220148086548,
      "learning_rate": 4.670205630453342e-06,
      "loss": 0.6336,
      "step": 42390
    },
    {
      "epoch": 17.36,
      "grad_norm": 2.62512469291687,
      "learning_rate": 4.670186390081251e-06,
      "loss": 0.6086,
      "step": 42400
    },
    {
      "epoch": 17.36,
      "grad_norm": 2.0973212718963623,
      "learning_rate": 4.670167139052115e-06,
      "loss": 0.6445,
      "step": 42410
    },
    {
      "epoch": 17.36,
      "grad_norm": 3.021014451980591,
      "learning_rate": 4.670147877366022e-06,
      "loss": 0.6385,
      "step": 42420
    },
    {
      "epoch": 17.37,
      "grad_norm": 2.238415002822876,
      "learning_rate": 4.67012860502306e-06,
      "loss": 0.6219,
      "step": 42430
    },
    {
      "epoch": 17.37,
      "grad_norm": 3.6200993061065674,
      "learning_rate": 4.670109322023319e-06,
      "loss": 0.6248,
      "step": 42440
    },
    {
      "epoch": 17.38,
      "grad_norm": 3.415721893310547,
      "learning_rate": 4.670090028366886e-06,
      "loss": 0.6044,
      "step": 42450
    },
    {
      "epoch": 17.38,
      "grad_norm": 3.1785502433776855,
      "learning_rate": 4.67007072405385e-06,
      "loss": 0.6202,
      "step": 42460
    },
    {
      "epoch": 17.38,
      "grad_norm": 2.4161064624786377,
      "learning_rate": 4.670051409084298e-06,
      "loss": 0.636,
      "step": 42470
    },
    {
      "epoch": 17.39,
      "grad_norm": 2.3695693016052246,
      "learning_rate": 4.670032083458321e-06,
      "loss": 0.6319,
      "step": 42480
    },
    {
      "epoch": 17.39,
      "grad_norm": 2.922131299972534,
      "learning_rate": 4.670012747176005e-06,
      "loss": 0.6163,
      "step": 42490
    },
    {
      "epoch": 17.4,
      "grad_norm": 2.4480254650115967,
      "learning_rate": 4.669993400237439e-06,
      "loss": 0.6314,
      "step": 42500
    },
    {
      "epoch": 17.4,
      "grad_norm": 2.7731828689575195,
      "learning_rate": 4.669974042642713e-06,
      "loss": 0.6,
      "step": 42510
    },
    {
      "epoch": 17.4,
      "grad_norm": 2.659881353378296,
      "learning_rate": 4.669954674391915e-06,
      "loss": 0.6367,
      "step": 42520
    },
    {
      "epoch": 17.41,
      "grad_norm": 2.616454839706421,
      "learning_rate": 4.669935295485134e-06,
      "loss": 0.6277,
      "step": 42530
    },
    {
      "epoch": 17.41,
      "grad_norm": 2.915560722351074,
      "learning_rate": 4.669915905922458e-06,
      "loss": 0.6224,
      "step": 42540
    },
    {
      "epoch": 17.42,
      "grad_norm": 2.840632915496826,
      "learning_rate": 4.669896505703975e-06,
      "loss": 0.6377,
      "step": 42550
    },
    {
      "epoch": 17.42,
      "grad_norm": 2.5774691104888916,
      "learning_rate": 4.669877094829777e-06,
      "loss": 0.6232,
      "step": 42560
    },
    {
      "epoch": 17.43,
      "grad_norm": 2.8789961338043213,
      "learning_rate": 4.669857673299951e-06,
      "loss": 0.6211,
      "step": 42570
    },
    {
      "epoch": 17.43,
      "grad_norm": 2.2455832958221436,
      "learning_rate": 4.669838241114585e-06,
      "loss": 0.6212,
      "step": 42580
    },
    {
      "epoch": 17.43,
      "grad_norm": 2.3605761528015137,
      "learning_rate": 4.66981879827377e-06,
      "loss": 0.6325,
      "step": 42590
    },
    {
      "epoch": 17.44,
      "grad_norm": 2.293848752975464,
      "learning_rate": 4.669799344777593e-06,
      "loss": 0.6184,
      "step": 42600
    },
    {
      "epoch": 17.44,
      "grad_norm": 2.5812389850616455,
      "learning_rate": 4.669779880626144e-06,
      "loss": 0.6024,
      "step": 42610
    },
    {
      "epoch": 17.45,
      "grad_norm": 2.6057300567626953,
      "learning_rate": 4.669760405819513e-06,
      "loss": 0.6352,
      "step": 42620
    },
    {
      "epoch": 17.45,
      "grad_norm": 2.723668336868286,
      "learning_rate": 4.669740920357789e-06,
      "loss": 0.6201,
      "step": 42630
    },
    {
      "epoch": 17.45,
      "grad_norm": 2.6056554317474365,
      "learning_rate": 4.669721424241061e-06,
      "loss": 0.6267,
      "step": 42640
    },
    {
      "epoch": 17.46,
      "grad_norm": 3.2938296794891357,
      "learning_rate": 4.669701917469417e-06,
      "loss": 0.6231,
      "step": 42650
    },
    {
      "epoch": 17.46,
      "grad_norm": 3.474188804626465,
      "learning_rate": 4.669682400042948e-06,
      "loss": 0.6128,
      "step": 42660
    },
    {
      "epoch": 17.47,
      "grad_norm": 2.1571972370147705,
      "learning_rate": 4.6696628719617424e-06,
      "loss": 0.6425,
      "step": 42670
    },
    {
      "epoch": 17.47,
      "grad_norm": 2.6750855445861816,
      "learning_rate": 4.669643333225891e-06,
      "loss": 0.6183,
      "step": 42680
    },
    {
      "epoch": 17.47,
      "grad_norm": 2.860943555831909,
      "learning_rate": 4.669623783835482e-06,
      "loss": 0.6253,
      "step": 42690
    },
    {
      "epoch": 17.48,
      "grad_norm": 3.6276602745056152,
      "learning_rate": 4.669604223790605e-06,
      "loss": 0.6162,
      "step": 42700
    },
    {
      "epoch": 17.48,
      "grad_norm": 2.252223253250122,
      "learning_rate": 4.66958465309135e-06,
      "loss": 0.6189,
      "step": 42710
    },
    {
      "epoch": 17.49,
      "grad_norm": 2.437516212463379,
      "learning_rate": 4.669565071737808e-06,
      "loss": 0.6443,
      "step": 42720
    },
    {
      "epoch": 17.49,
      "grad_norm": 3.4831628799438477,
      "learning_rate": 4.6695454797300664e-06,
      "loss": 0.6303,
      "step": 42730
    },
    {
      "epoch": 17.49,
      "grad_norm": 2.540160655975342,
      "learning_rate": 4.6695258770682155e-06,
      "loss": 0.6012,
      "step": 42740
    },
    {
      "epoch": 17.5,
      "grad_norm": 2.199366331100464,
      "learning_rate": 4.669506263752346e-06,
      "loss": 0.6222,
      "step": 42750
    },
    {
      "epoch": 17.5,
      "grad_norm": 2.078773021697998,
      "learning_rate": 4.669486639782547e-06,
      "loss": 0.6179,
      "step": 42760
    },
    {
      "epoch": 17.51,
      "grad_norm": 2.609201669692993,
      "learning_rate": 4.669467005158909e-06,
      "loss": 0.6219,
      "step": 42770
    },
    {
      "epoch": 17.51,
      "grad_norm": 3.1697330474853516,
      "learning_rate": 4.669447359881522e-06,
      "loss": 0.6353,
      "step": 42780
    },
    {
      "epoch": 17.52,
      "grad_norm": 2.3266260623931885,
      "learning_rate": 4.669427703950475e-06,
      "loss": 0.6419,
      "step": 42790
    },
    {
      "epoch": 17.52,
      "grad_norm": 2.71734356880188,
      "learning_rate": 4.669408037365858e-06,
      "loss": 0.6049,
      "step": 42800
    },
    {
      "epoch": 17.52,
      "grad_norm": 2.506720781326294,
      "learning_rate": 4.669388360127763e-06,
      "loss": 0.6588,
      "step": 42810
    },
    {
      "epoch": 17.53,
      "grad_norm": 2.8435072898864746,
      "learning_rate": 4.6693686722362786e-06,
      "loss": 0.6241,
      "step": 42820
    },
    {
      "epoch": 17.53,
      "grad_norm": 3.4991421699523926,
      "learning_rate": 4.669348973691494e-06,
      "loss": 0.6211,
      "step": 42830
    },
    {
      "epoch": 17.54,
      "grad_norm": 1.8544714450836182,
      "learning_rate": 4.6693292644935015e-06,
      "loss": 0.6246,
      "step": 42840
    },
    {
      "epoch": 17.54,
      "grad_norm": 2.5827245712280273,
      "learning_rate": 4.669309544642391e-06,
      "loss": 0.612,
      "step": 42850
    },
    {
      "epoch": 17.54,
      "grad_norm": 2.385155200958252,
      "learning_rate": 4.669289814138252e-06,
      "loss": 0.6573,
      "step": 42860
    },
    {
      "epoch": 17.55,
      "grad_norm": 2.45524263381958,
      "learning_rate": 4.669270072981175e-06,
      "loss": 0.6051,
      "step": 42870
    },
    {
      "epoch": 17.55,
      "grad_norm": 2.529545307159424,
      "learning_rate": 4.669250321171251e-06,
      "loss": 0.6228,
      "step": 42880
    },
    {
      "epoch": 17.56,
      "grad_norm": 3.633680582046509,
      "learning_rate": 4.6692305587085695e-06,
      "loss": 0.6169,
      "step": 42890
    },
    {
      "epoch": 17.56,
      "grad_norm": 3.209381580352783,
      "learning_rate": 4.669210785593222e-06,
      "loss": 0.6278,
      "step": 42900
    },
    {
      "epoch": 17.56,
      "grad_norm": 2.8759372234344482,
      "learning_rate": 4.6691910018253e-06,
      "loss": 0.6227,
      "step": 42910
    },
    {
      "epoch": 17.57,
      "grad_norm": 2.902499198913574,
      "learning_rate": 4.669171207404891e-06,
      "loss": 0.6321,
      "step": 42920
    },
    {
      "epoch": 17.57,
      "grad_norm": 2.7282307147979736,
      "learning_rate": 4.669151402332088e-06,
      "loss": 0.626,
      "step": 42930
    },
    {
      "epoch": 17.58,
      "grad_norm": 2.5285608768463135,
      "learning_rate": 4.669131586606983e-06,
      "loss": 0.6182,
      "step": 42940
    },
    {
      "epoch": 17.58,
      "grad_norm": 3.334681749343872,
      "learning_rate": 4.669111760229664e-06,
      "loss": 0.6334,
      "step": 42950
    },
    {
      "epoch": 17.58,
      "grad_norm": 2.136928081512451,
      "learning_rate": 4.669091923200222e-06,
      "loss": 0.6384,
      "step": 42960
    },
    {
      "epoch": 17.59,
      "grad_norm": 2.3304193019866943,
      "learning_rate": 4.66907207551875e-06,
      "loss": 0.6369,
      "step": 42970
    },
    {
      "epoch": 17.59,
      "grad_norm": 2.4634201526641846,
      "learning_rate": 4.669052217185337e-06,
      "loss": 0.6109,
      "step": 42980
    },
    {
      "epoch": 17.6,
      "grad_norm": 2.6947059631347656,
      "learning_rate": 4.669032348200075e-06,
      "loss": 0.6388,
      "step": 42990
    },
    {
      "epoch": 17.6,
      "grad_norm": 3.2885470390319824,
      "learning_rate": 4.669012468563055e-06,
      "loss": 0.6129,
      "step": 43000
    },
    {
      "epoch": 17.61,
      "grad_norm": 3.2899277210235596,
      "learning_rate": 4.668992578274367e-06,
      "loss": 0.6062,
      "step": 43010
    },
    {
      "epoch": 17.61,
      "grad_norm": 4.132672309875488,
      "learning_rate": 4.668972677334103e-06,
      "loss": 0.6204,
      "step": 43020
    },
    {
      "epoch": 17.61,
      "grad_norm": 3.276355743408203,
      "learning_rate": 4.668952765742354e-06,
      "loss": 0.6164,
      "step": 43030
    },
    {
      "epoch": 17.62,
      "grad_norm": 3.3925371170043945,
      "learning_rate": 4.668932843499212e-06,
      "loss": 0.6294,
      "step": 43040
    },
    {
      "epoch": 17.62,
      "grad_norm": 2.169414520263672,
      "learning_rate": 4.668912910604767e-06,
      "loss": 0.6272,
      "step": 43050
    },
    {
      "epoch": 17.63,
      "grad_norm": 2.6223130226135254,
      "learning_rate": 4.668892967059111e-06,
      "loss": 0.6141,
      "step": 43060
    },
    {
      "epoch": 17.63,
      "grad_norm": 2.435392379760742,
      "learning_rate": 4.668873012862335e-06,
      "loss": 0.6211,
      "step": 43070
    },
    {
      "epoch": 17.63,
      "grad_norm": 2.331810712814331,
      "learning_rate": 4.66885304801453e-06,
      "loss": 0.6203,
      "step": 43080
    },
    {
      "epoch": 17.64,
      "grad_norm": 2.718520402908325,
      "learning_rate": 4.668833072515789e-06,
      "loss": 0.633,
      "step": 43090
    },
    {
      "epoch": 17.64,
      "grad_norm": 2.1611452102661133,
      "learning_rate": 4.668813086366202e-06,
      "loss": 0.6199,
      "step": 43100
    },
    {
      "epoch": 17.65,
      "grad_norm": 4.855103969573975,
      "learning_rate": 4.668793089565861e-06,
      "loss": 0.6297,
      "step": 43110
    },
    {
      "epoch": 17.65,
      "grad_norm": 2.444364070892334,
      "learning_rate": 4.668773082114857e-06,
      "loss": 0.6186,
      "step": 43120
    },
    {
      "epoch": 17.65,
      "grad_norm": 4.482316017150879,
      "learning_rate": 4.668753064013284e-06,
      "loss": 0.6208,
      "step": 43130
    },
    {
      "epoch": 17.66,
      "grad_norm": 2.871796131134033,
      "learning_rate": 4.668733035261231e-06,
      "loss": 0.6204,
      "step": 43140
    },
    {
      "epoch": 17.66,
      "grad_norm": 2.7057993412017822,
      "learning_rate": 4.668712995858791e-06,
      "loss": 0.6297,
      "step": 43150
    },
    {
      "epoch": 17.67,
      "grad_norm": 2.3618710041046143,
      "learning_rate": 4.668692945806056e-06,
      "loss": 0.6147,
      "step": 43160
    },
    {
      "epoch": 17.67,
      "grad_norm": 2.7233240604400635,
      "learning_rate": 4.668672885103117e-06,
      "loss": 0.6032,
      "step": 43170
    },
    {
      "epoch": 17.67,
      "grad_norm": 2.2987401485443115,
      "learning_rate": 4.668652813750067e-06,
      "loss": 0.6539,
      "step": 43180
    },
    {
      "epoch": 17.68,
      "grad_norm": 2.2514572143554688,
      "learning_rate": 4.668632731746996e-06,
      "loss": 0.63,
      "step": 43190
    },
    {
      "epoch": 17.68,
      "grad_norm": 2.774815082550049,
      "learning_rate": 4.668612639093997e-06,
      "loss": 0.6115,
      "step": 43200
    },
    {
      "epoch": 17.69,
      "grad_norm": 2.2299294471740723,
      "learning_rate": 4.668592535791164e-06,
      "loss": 0.6171,
      "step": 43210
    },
    {
      "epoch": 17.69,
      "grad_norm": 2.684882164001465,
      "learning_rate": 4.668572421838587e-06,
      "loss": 0.6466,
      "step": 43220
    },
    {
      "epoch": 17.7,
      "grad_norm": 2.9943466186523438,
      "learning_rate": 4.6685522972363575e-06,
      "loss": 0.6237,
      "step": 43230
    },
    {
      "epoch": 17.7,
      "grad_norm": 3.140561103820801,
      "learning_rate": 4.6685321619845694e-06,
      "loss": 0.6397,
      "step": 43240
    },
    {
      "epoch": 17.7,
      "grad_norm": 2.0244393348693848,
      "learning_rate": 4.668512016083314e-06,
      "loss": 0.6177,
      "step": 43250
    },
    {
      "epoch": 17.71,
      "grad_norm": 2.713763475418091,
      "learning_rate": 4.6684918595326834e-06,
      "loss": 0.6261,
      "step": 43260
    },
    {
      "epoch": 17.71,
      "grad_norm": 4.52509880065918,
      "learning_rate": 4.668471692332771e-06,
      "loss": 0.6005,
      "step": 43270
    },
    {
      "epoch": 17.72,
      "grad_norm": 2.6861133575439453,
      "learning_rate": 4.668451514483669e-06,
      "loss": 0.6173,
      "step": 43280
    },
    {
      "epoch": 17.72,
      "grad_norm": 2.144493818283081,
      "learning_rate": 4.668431325985469e-06,
      "loss": 0.6125,
      "step": 43290
    },
    {
      "epoch": 17.72,
      "grad_norm": 2.8725626468658447,
      "learning_rate": 4.668411126838264e-06,
      "loss": 0.6219,
      "step": 43300
    },
    {
      "epoch": 17.73,
      "grad_norm": 2.361713171005249,
      "learning_rate": 4.668390917042146e-06,
      "loss": 0.624,
      "step": 43310
    },
    {
      "epoch": 17.73,
      "grad_norm": 2.3928825855255127,
      "learning_rate": 4.668370696597208e-06,
      "loss": 0.6384,
      "step": 43320
    },
    {
      "epoch": 17.74,
      "grad_norm": 2.3550233840942383,
      "learning_rate": 4.668350465503543e-06,
      "loss": 0.6017,
      "step": 43330
    },
    {
      "epoch": 17.74,
      "grad_norm": 2.6482019424438477,
      "learning_rate": 4.668330223761243e-06,
      "loss": 0.6296,
      "step": 43340
    },
    {
      "epoch": 17.74,
      "grad_norm": 2.815096855163574,
      "learning_rate": 4.6683099713704004e-06,
      "loss": 0.6299,
      "step": 43350
    },
    {
      "epoch": 17.75,
      "grad_norm": 3.210529088973999,
      "learning_rate": 4.668289708331109e-06,
      "loss": 0.6404,
      "step": 43360
    },
    {
      "epoch": 17.75,
      "grad_norm": 2.8135523796081543,
      "learning_rate": 4.668269434643462e-06,
      "loss": 0.6242,
      "step": 43370
    },
    {
      "epoch": 17.76,
      "grad_norm": 2.0899393558502197,
      "learning_rate": 4.668249150307551e-06,
      "loss": 0.6084,
      "step": 43380
    },
    {
      "epoch": 17.76,
      "grad_norm": 2.2569148540496826,
      "learning_rate": 4.668228855323469e-06,
      "loss": 0.6049,
      "step": 43390
    },
    {
      "epoch": 17.77,
      "grad_norm": 2.398582696914673,
      "learning_rate": 4.668208549691309e-06,
      "loss": 0.6164,
      "step": 43400
    },
    {
      "epoch": 17.77,
      "grad_norm": 2.6059019565582275,
      "learning_rate": 4.6681882334111655e-06,
      "loss": 0.6112,
      "step": 43410
    },
    {
      "epoch": 17.77,
      "grad_norm": 3.026045799255371,
      "learning_rate": 4.66816790648313e-06,
      "loss": 0.6126,
      "step": 43420
    },
    {
      "epoch": 17.78,
      "grad_norm": 2.3957817554473877,
      "learning_rate": 4.668147568907296e-06,
      "loss": 0.612,
      "step": 43430
    },
    {
      "epoch": 17.78,
      "grad_norm": 2.39263653755188,
      "learning_rate": 4.668127220683757e-06,
      "loss": 0.6146,
      "step": 43440
    },
    {
      "epoch": 17.79,
      "grad_norm": 2.533935546875,
      "learning_rate": 4.668106861812605e-06,
      "loss": 0.6328,
      "step": 43450
    },
    {
      "epoch": 17.79,
      "grad_norm": 2.5830328464508057,
      "learning_rate": 4.668086492293935e-06,
      "loss": 0.6122,
      "step": 43460
    },
    {
      "epoch": 17.79,
      "grad_norm": 2.9051520824432373,
      "learning_rate": 4.668066112127839e-06,
      "loss": 0.6159,
      "step": 43470
    },
    {
      "epoch": 17.8,
      "grad_norm": 2.9036242961883545,
      "learning_rate": 4.668045721314411e-06,
      "loss": 0.6242,
      "step": 43480
    },
    {
      "epoch": 17.8,
      "grad_norm": 2.896477699279785,
      "learning_rate": 4.668025319853745e-06,
      "loss": 0.5923,
      "step": 43490
    },
    {
      "epoch": 17.81,
      "grad_norm": 2.850227117538452,
      "learning_rate": 4.668004907745932e-06,
      "loss": 0.6181,
      "step": 43500
    },
    {
      "epoch": 17.81,
      "grad_norm": 2.887592077255249,
      "learning_rate": 4.667984484991069e-06,
      "loss": 0.6007,
      "step": 43510
    },
    {
      "epoch": 17.81,
      "grad_norm": 3.4412803649902344,
      "learning_rate": 4.667964051589247e-06,
      "loss": 0.6238,
      "step": 43520
    },
    {
      "epoch": 17.82,
      "grad_norm": 3.1876134872436523,
      "learning_rate": 4.6679436075405605e-06,
      "loss": 0.6349,
      "step": 43530
    },
    {
      "epoch": 17.82,
      "grad_norm": 2.1346402168273926,
      "learning_rate": 4.667923152845103e-06,
      "loss": 0.6204,
      "step": 43540
    },
    {
      "epoch": 17.83,
      "grad_norm": 2.429006338119507,
      "learning_rate": 4.667902687502969e-06,
      "loss": 0.617,
      "step": 43550
    },
    {
      "epoch": 17.83,
      "grad_norm": 2.8265271186828613,
      "learning_rate": 4.66788221151425e-06,
      "loss": 0.6396,
      "step": 43560
    },
    {
      "epoch": 17.83,
      "grad_norm": 2.5159149169921875,
      "learning_rate": 4.667861724879043e-06,
      "loss": 0.6354,
      "step": 43570
    },
    {
      "epoch": 17.84,
      "grad_norm": 2.9007039070129395,
      "learning_rate": 4.667841227597438e-06,
      "loss": 0.6292,
      "step": 43580
    },
    {
      "epoch": 17.84,
      "grad_norm": 2.3694393634796143,
      "learning_rate": 4.667820719669532e-06,
      "loss": 0.6181,
      "step": 43590
    },
    {
      "epoch": 17.85,
      "grad_norm": 2.857259511947632,
      "learning_rate": 4.667800201095418e-06,
      "loss": 0.6283,
      "step": 43600
    },
    {
      "epoch": 17.85,
      "grad_norm": 4.043162822723389,
      "learning_rate": 4.667779671875191e-06,
      "loss": 0.6222,
      "step": 43610
    },
    {
      "epoch": 17.86,
      "grad_norm": 2.5897843837738037,
      "learning_rate": 4.667759132008943e-06,
      "loss": 0.6222,
      "step": 43620
    },
    {
      "epoch": 17.86,
      "grad_norm": 1.845577359199524,
      "learning_rate": 4.667738581496769e-06,
      "loss": 0.6186,
      "step": 43630
    },
    {
      "epoch": 17.86,
      "grad_norm": 1.9758347272872925,
      "learning_rate": 4.667718020338763e-06,
      "loss": 0.6142,
      "step": 43640
    },
    {
      "epoch": 17.87,
      "grad_norm": 2.162503242492676,
      "learning_rate": 4.66769744853502e-06,
      "loss": 0.6317,
      "step": 43650
    },
    {
      "epoch": 17.87,
      "grad_norm": 2.9383902549743652,
      "learning_rate": 4.667676866085633e-06,
      "loss": 0.6273,
      "step": 43660
    },
    {
      "epoch": 17.88,
      "grad_norm": 2.314662456512451,
      "learning_rate": 4.667656272990697e-06,
      "loss": 0.6242,
      "step": 43670
    },
    {
      "epoch": 17.88,
      "grad_norm": 2.8788392543792725,
      "learning_rate": 4.667635669250308e-06,
      "loss": 0.6409,
      "step": 43680
    },
    {
      "epoch": 17.88,
      "grad_norm": 3.081483840942383,
      "learning_rate": 4.667615054864557e-06,
      "loss": 0.6155,
      "step": 43690
    },
    {
      "epoch": 17.89,
      "grad_norm": 3.4604756832122803,
      "learning_rate": 4.66759442983354e-06,
      "loss": 0.6372,
      "step": 43700
    },
    {
      "epoch": 17.89,
      "grad_norm": 4.381320476531982,
      "learning_rate": 4.667573794157352e-06,
      "loss": 0.622,
      "step": 43710
    },
    {
      "epoch": 17.9,
      "grad_norm": 2.5329580307006836,
      "learning_rate": 4.667553147836087e-06,
      "loss": 0.6298,
      "step": 43720
    },
    {
      "epoch": 17.9,
      "grad_norm": 2.2867469787597656,
      "learning_rate": 4.66753249086984e-06,
      "loss": 0.621,
      "step": 43730
    },
    {
      "epoch": 17.9,
      "grad_norm": 2.203671932220459,
      "learning_rate": 4.667511823258704e-06,
      "loss": 0.617,
      "step": 43740
    },
    {
      "epoch": 17.91,
      "grad_norm": 2.6673431396484375,
      "learning_rate": 4.667491145002777e-06,
      "loss": 0.6098,
      "step": 43750
    },
    {
      "epoch": 17.91,
      "grad_norm": 2.4539198875427246,
      "learning_rate": 4.66747045610215e-06,
      "loss": 0.6395,
      "step": 43760
    },
    {
      "epoch": 17.92,
      "grad_norm": 3.5667030811309814,
      "learning_rate": 4.66744975655692e-06,
      "loss": 0.65,
      "step": 43770
    },
    {
      "epoch": 17.92,
      "grad_norm": 2.7185330390930176,
      "learning_rate": 4.667429046367181e-06,
      "loss": 0.6147,
      "step": 43780
    },
    {
      "epoch": 17.92,
      "grad_norm": 2.11452054977417,
      "learning_rate": 4.6674083255330285e-06,
      "loss": 0.6488,
      "step": 43790
    },
    {
      "epoch": 17.93,
      "grad_norm": 1.950927734375,
      "learning_rate": 4.6673875940545575e-06,
      "loss": 0.631,
      "step": 43800
    },
    {
      "epoch": 17.93,
      "grad_norm": 2.401061534881592,
      "learning_rate": 4.667366851931862e-06,
      "loss": 0.6175,
      "step": 43810
    },
    {
      "epoch": 17.94,
      "grad_norm": 3.724292278289795,
      "learning_rate": 4.667346099165038e-06,
      "loss": 0.6281,
      "step": 43820
    },
    {
      "epoch": 17.94,
      "grad_norm": 2.4363350868225098,
      "learning_rate": 4.66732533575418e-06,
      "loss": 0.6109,
      "step": 43830
    },
    {
      "epoch": 17.95,
      "grad_norm": 2.976309299468994,
      "learning_rate": 4.6673045616993825e-06,
      "loss": 0.615,
      "step": 43840
    },
    {
      "epoch": 17.95,
      "grad_norm": 2.356243133544922,
      "learning_rate": 4.667283777000742e-06,
      "loss": 0.6162,
      "step": 43850
    },
    {
      "epoch": 17.95,
      "grad_norm": 3.1036105155944824,
      "learning_rate": 4.6672629816583534e-06,
      "loss": 0.6244,
      "step": 43860
    },
    {
      "epoch": 17.96,
      "grad_norm": 2.232448101043701,
      "learning_rate": 4.667242175672311e-06,
      "loss": 0.6225,
      "step": 43870
    },
    {
      "epoch": 17.96,
      "grad_norm": 2.3291966915130615,
      "learning_rate": 4.667221359042712e-06,
      "loss": 0.6077,
      "step": 43880
    },
    {
      "epoch": 17.97,
      "grad_norm": 2.5143208503723145,
      "learning_rate": 4.66720053176965e-06,
      "loss": 0.5994,
      "step": 43890
    },
    {
      "epoch": 17.97,
      "grad_norm": 3.480006217956543,
      "learning_rate": 4.6671796938532196e-06,
      "loss": 0.6087,
      "step": 43900
    },
    {
      "epoch": 17.97,
      "grad_norm": 2.635796070098877,
      "learning_rate": 4.6671588452935194e-06,
      "loss": 0.6138,
      "step": 43910
    },
    {
      "epoch": 17.98,
      "grad_norm": 2.2396748065948486,
      "learning_rate": 4.667137986090642e-06,
      "loss": 0.643,
      "step": 43920
    },
    {
      "epoch": 17.98,
      "grad_norm": 3.4154253005981445,
      "learning_rate": 4.667117116244684e-06,
      "loss": 0.6243,
      "step": 43930
    },
    {
      "epoch": 17.99,
      "grad_norm": 3.078568935394287,
      "learning_rate": 4.667096235755741e-06,
      "loss": 0.6247,
      "step": 43940
    },
    {
      "epoch": 17.99,
      "grad_norm": 2.8962488174438477,
      "learning_rate": 4.66707534462391e-06,
      "loss": 0.6597,
      "step": 43950
    },
    {
      "epoch": 17.99,
      "grad_norm": 2.725235939025879,
      "learning_rate": 4.667054442849285e-06,
      "loss": 0.6234,
      "step": 43960
    },
    {
      "epoch": 18.0,
      "grad_norm": 2.036424160003662,
      "learning_rate": 4.66703353043196e-06,
      "loss": 0.627,
      "step": 43970
    },
    {
      "epoch": 18.0,
      "eval_loss": 0.6213150024414062,
      "eval_runtime": 51.9608,
      "eval_samples_per_second": 66.377,
      "eval_steps_per_second": 8.314,
      "step": 43974
    },
    {
      "epoch": 18.0,
      "grad_norm": 1.9710125923156738,
      "learning_rate": 4.6670126073720356e-06,
      "loss": 0.6242,
      "step": 43980
    },
    {
      "epoch": 18.01,
      "grad_norm": 3.971695899963379,
      "learning_rate": 4.666991673669604e-06,
      "loss": 0.596,
      "step": 43990
    },
    {
      "epoch": 18.01,
      "grad_norm": 3.2845818996429443,
      "learning_rate": 4.666970729324762e-06,
      "loss": 0.6173,
      "step": 44000
    },
    {
      "epoch": 18.01,
      "grad_norm": 3.6261889934539795,
      "learning_rate": 4.666949774337606e-06,
      "loss": 0.6279,
      "step": 44010
    },
    {
      "epoch": 18.02,
      "grad_norm": 3.0978920459747314,
      "learning_rate": 4.666928808708232e-06,
      "loss": 0.6308,
      "step": 44020
    },
    {
      "epoch": 18.02,
      "grad_norm": 2.5469236373901367,
      "learning_rate": 4.666907832436735e-06,
      "loss": 0.6579,
      "step": 44030
    },
    {
      "epoch": 18.03,
      "grad_norm": 2.3233695030212402,
      "learning_rate": 4.666886845523211e-06,
      "loss": 0.6355,
      "step": 44040
    },
    {
      "epoch": 18.03,
      "grad_norm": 2.9361095428466797,
      "learning_rate": 4.6668658479677585e-06,
      "loss": 0.6184,
      "step": 44050
    },
    {
      "epoch": 18.04,
      "grad_norm": 2.1188793182373047,
      "learning_rate": 4.666844839770471e-06,
      "loss": 0.6105,
      "step": 44060
    },
    {
      "epoch": 18.04,
      "grad_norm": 2.5089128017425537,
      "learning_rate": 4.666823820931446e-06,
      "loss": 0.622,
      "step": 44070
    },
    {
      "epoch": 18.04,
      "grad_norm": 3.3219099044799805,
      "learning_rate": 4.6668027914507795e-06,
      "loss": 0.606,
      "step": 44080
    },
    {
      "epoch": 18.05,
      "grad_norm": 2.737504720687866,
      "learning_rate": 4.666781751328569e-06,
      "loss": 0.6234,
      "step": 44090
    },
    {
      "epoch": 18.05,
      "grad_norm": 2.7171709537506104,
      "learning_rate": 4.666760700564908e-06,
      "loss": 0.6188,
      "step": 44100
    },
    {
      "epoch": 18.06,
      "grad_norm": 2.599039316177368,
      "learning_rate": 4.666739639159896e-06,
      "loss": 0.6229,
      "step": 44110
    },
    {
      "epoch": 18.06,
      "grad_norm": 2.700549364089966,
      "learning_rate": 4.6667185671136285e-06,
      "loss": 0.6313,
      "step": 44120
    },
    {
      "epoch": 18.06,
      "grad_norm": 2.717362403869629,
      "learning_rate": 4.666697484426201e-06,
      "loss": 0.6219,
      "step": 44130
    },
    {
      "epoch": 18.07,
      "grad_norm": 2.798164129257202,
      "learning_rate": 4.6666763910977105e-06,
      "loss": 0.6278,
      "step": 44140
    },
    {
      "epoch": 18.07,
      "grad_norm": 3.7747087478637695,
      "learning_rate": 4.666655287128254e-06,
      "loss": 0.6202,
      "step": 44150
    },
    {
      "epoch": 18.08,
      "grad_norm": 2.77167010307312,
      "learning_rate": 4.6666341725179295e-06,
      "loss": 0.6228,
      "step": 44160
    },
    {
      "epoch": 18.08,
      "grad_norm": 2.78114914894104,
      "learning_rate": 4.666613047266831e-06,
      "loss": 0.6152,
      "step": 44170
    },
    {
      "epoch": 18.08,
      "grad_norm": 3.2439374923706055,
      "learning_rate": 4.666591911375057e-06,
      "loss": 0.6215,
      "step": 44180
    },
    {
      "epoch": 18.09,
      "grad_norm": 2.4767534732818604,
      "learning_rate": 4.666570764842704e-06,
      "loss": 0.6129,
      "step": 44190
    },
    {
      "epoch": 18.09,
      "grad_norm": 2.8653724193573,
      "learning_rate": 4.666549607669869e-06,
      "loss": 0.6237,
      "step": 44200
    },
    {
      "epoch": 18.1,
      "grad_norm": 3.3697896003723145,
      "learning_rate": 4.666528439856649e-06,
      "loss": 0.637,
      "step": 44210
    },
    {
      "epoch": 18.1,
      "grad_norm": 2.3782665729522705,
      "learning_rate": 4.6665072614031395e-06,
      "loss": 0.6082,
      "step": 44220
    },
    {
      "epoch": 18.1,
      "grad_norm": 2.7801766395568848,
      "learning_rate": 4.66648607230944e-06,
      "loss": 0.6239,
      "step": 44230
    },
    {
      "epoch": 18.11,
      "grad_norm": 2.647127628326416,
      "learning_rate": 4.666464872575646e-06,
      "loss": 0.6212,
      "step": 44240
    },
    {
      "epoch": 18.11,
      "grad_norm": 2.9150326251983643,
      "learning_rate": 4.666443662201854e-06,
      "loss": 0.5986,
      "step": 44250
    },
    {
      "epoch": 18.12,
      "grad_norm": 2.9387340545654297,
      "learning_rate": 4.666422441188163e-06,
      "loss": 0.628,
      "step": 44260
    },
    {
      "epoch": 18.12,
      "grad_norm": 2.736926317214966,
      "learning_rate": 4.666401209534669e-06,
      "loss": 0.6386,
      "step": 44270
    },
    {
      "epoch": 18.13,
      "grad_norm": 2.413418769836426,
      "learning_rate": 4.666379967241469e-06,
      "loss": 0.616,
      "step": 44280
    },
    {
      "epoch": 18.13,
      "grad_norm": 2.838409185409546,
      "learning_rate": 4.666358714308661e-06,
      "loss": 0.6267,
      "step": 44290
    },
    {
      "epoch": 18.13,
      "grad_norm": 1.967199444770813,
      "learning_rate": 4.6663374507363425e-06,
      "loss": 0.614,
      "step": 44300
    },
    {
      "epoch": 18.14,
      "grad_norm": 2.1109976768493652,
      "learning_rate": 4.6663161765246115e-06,
      "loss": 0.6159,
      "step": 44310
    },
    {
      "epoch": 18.14,
      "grad_norm": 2.409731864929199,
      "learning_rate": 4.666294891673563e-06,
      "loss": 0.6266,
      "step": 44320
    },
    {
      "epoch": 18.15,
      "grad_norm": 2.932211399078369,
      "learning_rate": 4.666273596183297e-06,
      "loss": 0.6159,
      "step": 44330
    },
    {
      "epoch": 18.15,
      "grad_norm": 2.6398935317993164,
      "learning_rate": 4.6662522900539095e-06,
      "loss": 0.6215,
      "step": 44340
    },
    {
      "epoch": 18.15,
      "grad_norm": 2.542475461959839,
      "learning_rate": 4.666230973285499e-06,
      "loss": 0.6311,
      "step": 44350
    },
    {
      "epoch": 18.16,
      "grad_norm": 2.867619514465332,
      "learning_rate": 4.666209645878163e-06,
      "loss": 0.6021,
      "step": 44360
    },
    {
      "epoch": 18.16,
      "grad_norm": 2.134991407394409,
      "learning_rate": 4.6661883078319985e-06,
      "loss": 0.6257,
      "step": 44370
    },
    {
      "epoch": 18.17,
      "grad_norm": 2.824293613433838,
      "learning_rate": 4.666166959147104e-06,
      "loss": 0.618,
      "step": 44380
    },
    {
      "epoch": 18.17,
      "grad_norm": 2.2940478324890137,
      "learning_rate": 4.666145599823577e-06,
      "loss": 0.6198,
      "step": 44390
    },
    {
      "epoch": 18.17,
      "grad_norm": 2.568084239959717,
      "learning_rate": 4.666124229861516e-06,
      "loss": 0.6158,
      "step": 44400
    },
    {
      "epoch": 18.18,
      "grad_norm": 2.724968433380127,
      "learning_rate": 4.666102849261017e-06,
      "loss": 0.6381,
      "step": 44410
    },
    {
      "epoch": 18.18,
      "grad_norm": 2.8086204528808594,
      "learning_rate": 4.666081458022181e-06,
      "loss": 0.6095,
      "step": 44420
    },
    {
      "epoch": 18.19,
      "grad_norm": 2.6196932792663574,
      "learning_rate": 4.666060056145103e-06,
      "loss": 0.6169,
      "step": 44430
    },
    {
      "epoch": 18.19,
      "grad_norm": 3.5124118328094482,
      "learning_rate": 4.666038643629883e-06,
      "loss": 0.6167,
      "step": 44440
    },
    {
      "epoch": 18.19,
      "grad_norm": 3.177664041519165,
      "learning_rate": 4.666017220476618e-06,
      "loss": 0.6245,
      "step": 44450
    },
    {
      "epoch": 18.2,
      "grad_norm": 2.7578983306884766,
      "learning_rate": 4.665995786685406e-06,
      "loss": 0.6281,
      "step": 44460
    },
    {
      "epoch": 18.2,
      "grad_norm": 3.028489351272583,
      "learning_rate": 4.665974342256346e-06,
      "loss": 0.6412,
      "step": 44470
    },
    {
      "epoch": 18.21,
      "grad_norm": 2.8327622413635254,
      "learning_rate": 4.665952887189537e-06,
      "loss": 0.6281,
      "step": 44480
    },
    {
      "epoch": 18.21,
      "grad_norm": 2.9543118476867676,
      "learning_rate": 4.6659314214850755e-06,
      "loss": 0.6276,
      "step": 44490
    },
    {
      "epoch": 18.22,
      "grad_norm": 3.159104108810425,
      "learning_rate": 4.66590994514306e-06,
      "loss": 0.6331,
      "step": 44500
    },
    {
      "epoch": 18.22,
      "grad_norm": 2.9579451084136963,
      "learning_rate": 4.66588845816359e-06,
      "loss": 0.6311,
      "step": 44510
    },
    {
      "epoch": 18.22,
      "grad_norm": 3.5007221698760986,
      "learning_rate": 4.665866960546763e-06,
      "loss": 0.6073,
      "step": 44520
    },
    {
      "epoch": 18.23,
      "grad_norm": 3.011828899383545,
      "learning_rate": 4.665845452292678e-06,
      "loss": 0.6207,
      "step": 44530
    },
    {
      "epoch": 18.23,
      "grad_norm": 2.680316686630249,
      "learning_rate": 4.665823933401434e-06,
      "loss": 0.6134,
      "step": 44540
    },
    {
      "epoch": 18.24,
      "grad_norm": 1.9188518524169922,
      "learning_rate": 4.665802403873128e-06,
      "loss": 0.6083,
      "step": 44550
    },
    {
      "epoch": 18.24,
      "grad_norm": 2.672187328338623,
      "learning_rate": 4.66578086370786e-06,
      "loss": 0.6157,
      "step": 44560
    },
    {
      "epoch": 18.24,
      "grad_norm": 2.9504618644714355,
      "learning_rate": 4.665759312905728e-06,
      "loss": 0.6253,
      "step": 44570
    },
    {
      "epoch": 18.25,
      "grad_norm": 3.0627994537353516,
      "learning_rate": 4.6657377514668315e-06,
      "loss": 0.6267,
      "step": 44580
    },
    {
      "epoch": 18.25,
      "grad_norm": 2.7017769813537598,
      "learning_rate": 4.665716179391269e-06,
      "loss": 0.6194,
      "step": 44590
    },
    {
      "epoch": 18.26,
      "grad_norm": 2.924828052520752,
      "learning_rate": 4.665694596679137e-06,
      "loss": 0.6319,
      "step": 44600
    },
    {
      "epoch": 18.26,
      "grad_norm": 2.1966567039489746,
      "learning_rate": 4.665673003330538e-06,
      "loss": 0.6055,
      "step": 44610
    },
    {
      "epoch": 18.26,
      "grad_norm": 3.340968132019043,
      "learning_rate": 4.665651399345569e-06,
      "loss": 0.5992,
      "step": 44620
    },
    {
      "epoch": 18.27,
      "grad_norm": 2.9040446281433105,
      "learning_rate": 4.6656297847243295e-06,
      "loss": 0.6202,
      "step": 44630
    },
    {
      "epoch": 18.27,
      "grad_norm": 2.427736520767212,
      "learning_rate": 4.665608159466918e-06,
      "loss": 0.6147,
      "step": 44640
    },
    {
      "epoch": 18.28,
      "grad_norm": 2.980614423751831,
      "learning_rate": 4.665586523573434e-06,
      "loss": 0.5935,
      "step": 44650
    },
    {
      "epoch": 18.28,
      "grad_norm": 3.7624073028564453,
      "learning_rate": 4.6655648770439765e-06,
      "loss": 0.6165,
      "step": 44660
    },
    {
      "epoch": 18.28,
      "grad_norm": 2.81839656829834,
      "learning_rate": 4.665543219878645e-06,
      "loss": 0.6167,
      "step": 44670
    },
    {
      "epoch": 18.29,
      "grad_norm": 2.304288864135742,
      "learning_rate": 4.665521552077537e-06,
      "loss": 0.61,
      "step": 44680
    },
    {
      "epoch": 18.29,
      "grad_norm": 2.6458308696746826,
      "learning_rate": 4.6654998736407545e-06,
      "loss": 0.5959,
      "step": 44690
    },
    {
      "epoch": 18.3,
      "grad_norm": 2.2134499549865723,
      "learning_rate": 4.665478184568395e-06,
      "loss": 0.5997,
      "step": 44700
    },
    {
      "epoch": 18.3,
      "grad_norm": 2.6837525367736816,
      "learning_rate": 4.665456484860558e-06,
      "loss": 0.5963,
      "step": 44710
    },
    {
      "epoch": 18.31,
      "grad_norm": 2.6629815101623535,
      "learning_rate": 4.6654347745173436e-06,
      "loss": 0.6148,
      "step": 44720
    },
    {
      "epoch": 18.31,
      "grad_norm": 3.3514890670776367,
      "learning_rate": 4.66541305353885e-06,
      "loss": 0.6235,
      "step": 44730
    },
    {
      "epoch": 18.31,
      "grad_norm": 2.7049756050109863,
      "learning_rate": 4.665391321925178e-06,
      "loss": 0.6193,
      "step": 44740
    },
    {
      "epoch": 18.32,
      "grad_norm": 2.347421169281006,
      "learning_rate": 4.665369579676427e-06,
      "loss": 0.6082,
      "step": 44750
    },
    {
      "epoch": 18.32,
      "grad_norm": 2.4403584003448486,
      "learning_rate": 4.6653478267926955e-06,
      "loss": 0.5938,
      "step": 44760
    },
    {
      "epoch": 18.33,
      "grad_norm": 2.7638020515441895,
      "learning_rate": 4.665326063274085e-06,
      "loss": 0.6241,
      "step": 44770
    },
    {
      "epoch": 18.33,
      "grad_norm": 3.73042368888855,
      "learning_rate": 4.665304289120693e-06,
      "loss": 0.5999,
      "step": 44780
    },
    {
      "epoch": 18.33,
      "grad_norm": 2.6957809925079346,
      "learning_rate": 4.665282504332621e-06,
      "loss": 0.635,
      "step": 44790
    },
    {
      "epoch": 18.34,
      "grad_norm": 3.095593214035034,
      "learning_rate": 4.665260708909968e-06,
      "loss": 0.6215,
      "step": 44800
    },
    {
      "epoch": 18.34,
      "grad_norm": 2.1964151859283447,
      "learning_rate": 4.665238902852833e-06,
      "loss": 0.6075,
      "step": 44810
    },
    {
      "epoch": 18.35,
      "grad_norm": 2.7182247638702393,
      "learning_rate": 4.665217086161318e-06,
      "loss": 0.6152,
      "step": 44820
    },
    {
      "epoch": 18.35,
      "grad_norm": 3.562770128250122,
      "learning_rate": 4.665195258835522e-06,
      "loss": 0.6004,
      "step": 44830
    },
    {
      "epoch": 18.35,
      "grad_norm": 2.219142436981201,
      "learning_rate": 4.665173420875544e-06,
      "loss": 0.6091,
      "step": 44840
    },
    {
      "epoch": 18.36,
      "grad_norm": 2.474789619445801,
      "learning_rate": 4.665151572281485e-06,
      "loss": 0.6444,
      "step": 44850
    },
    {
      "epoch": 18.36,
      "grad_norm": 2.271311044692993,
      "learning_rate": 4.665129713053445e-06,
      "loss": 0.6246,
      "step": 44860
    },
    {
      "epoch": 18.37,
      "grad_norm": 2.1991143226623535,
      "learning_rate": 4.665107843191524e-06,
      "loss": 0.6238,
      "step": 44870
    },
    {
      "epoch": 18.37,
      "grad_norm": 2.5032339096069336,
      "learning_rate": 4.6650859626958225e-06,
      "loss": 0.6198,
      "step": 44880
    },
    {
      "epoch": 18.37,
      "grad_norm": 2.4886016845703125,
      "learning_rate": 4.66506407156644e-06,
      "loss": 0.6061,
      "step": 44890
    },
    {
      "epoch": 18.38,
      "grad_norm": 2.356757164001465,
      "learning_rate": 4.665042169803478e-06,
      "loss": 0.6144,
      "step": 44900
    },
    {
      "epoch": 18.38,
      "grad_norm": 2.435798168182373,
      "learning_rate": 4.665020257407035e-06,
      "loss": 0.6287,
      "step": 44910
    },
    {
      "epoch": 18.39,
      "grad_norm": 3.2744925022125244,
      "learning_rate": 4.664998334377214e-06,
      "loss": 0.5956,
      "step": 44920
    },
    {
      "epoch": 18.39,
      "grad_norm": 2.25762677192688,
      "learning_rate": 4.664976400714112e-06,
      "loss": 0.646,
      "step": 44930
    },
    {
      "epoch": 18.4,
      "grad_norm": 2.585526943206787,
      "learning_rate": 4.664954456417832e-06,
      "loss": 0.6221,
      "step": 44940
    },
    {
      "epoch": 18.4,
      "grad_norm": 2.4297986030578613,
      "learning_rate": 4.664932501488475e-06,
      "loss": 0.6076,
      "step": 44950
    },
    {
      "epoch": 18.4,
      "grad_norm": 2.531993865966797,
      "learning_rate": 4.66491053592614e-06,
      "loss": 0.6054,
      "step": 44960
    },
    {
      "epoch": 18.41,
      "grad_norm": 2.8238749504089355,
      "learning_rate": 4.6648885597309264e-06,
      "loss": 0.6173,
      "step": 44970
    },
    {
      "epoch": 18.41,
      "grad_norm": 1.9831948280334473,
      "learning_rate": 4.664866572902938e-06,
      "loss": 0.624,
      "step": 44980
    },
    {
      "epoch": 18.42,
      "grad_norm": 2.551365613937378,
      "learning_rate": 4.664844575442274e-06,
      "loss": 0.6345,
      "step": 44990
    },
    {
      "epoch": 18.42,
      "grad_norm": 2.825012683868408,
      "learning_rate": 4.664822567349035e-06,
      "loss": 0.6218,
      "step": 45000
    },
    {
      "epoch": 18.42,
      "grad_norm": 2.160788059234619,
      "learning_rate": 4.664800548623322e-06,
      "loss": 0.6411,
      "step": 45010
    },
    {
      "epoch": 18.43,
      "grad_norm": 2.4672138690948486,
      "learning_rate": 4.664778519265236e-06,
      "loss": 0.6405,
      "step": 45020
    },
    {
      "epoch": 18.43,
      "grad_norm": 2.7365939617156982,
      "learning_rate": 4.6647564792748774e-06,
      "loss": 0.6057,
      "step": 45030
    },
    {
      "epoch": 18.44,
      "grad_norm": 2.6297717094421387,
      "learning_rate": 4.664734428652349e-06,
      "loss": 0.6236,
      "step": 45040
    },
    {
      "epoch": 18.44,
      "grad_norm": 2.682029962539673,
      "learning_rate": 4.664712367397748e-06,
      "loss": 0.6023,
      "step": 45050
    },
    {
      "epoch": 18.44,
      "grad_norm": 2.5098049640655518,
      "learning_rate": 4.66469029551118e-06,
      "loss": 0.6455,
      "step": 45060
    },
    {
      "epoch": 18.45,
      "grad_norm": 2.303281784057617,
      "learning_rate": 4.664668212992743e-06,
      "loss": 0.611,
      "step": 45070
    },
    {
      "epoch": 18.45,
      "grad_norm": 2.406381607055664,
      "learning_rate": 4.664646119842539e-06,
      "loss": 0.6159,
      "step": 45080
    },
    {
      "epoch": 18.46,
      "grad_norm": 2.9279303550720215,
      "learning_rate": 4.66462401606067e-06,
      "loss": 0.6024,
      "step": 45090
    },
    {
      "epoch": 18.46,
      "grad_norm": 2.675325632095337,
      "learning_rate": 4.664601901647236e-06,
      "loss": 0.6239,
      "step": 45100
    },
    {
      "epoch": 18.47,
      "grad_norm": 3.081406831741333,
      "learning_rate": 4.664579776602338e-06,
      "loss": 0.6032,
      "step": 45110
    },
    {
      "epoch": 18.47,
      "grad_norm": 3.8092682361602783,
      "learning_rate": 4.66455764092608e-06,
      "loss": 0.6376,
      "step": 45120
    },
    {
      "epoch": 18.47,
      "grad_norm": 2.1805031299591064,
      "learning_rate": 4.664535494618561e-06,
      "loss": 0.6193,
      "step": 45130
    },
    {
      "epoch": 18.48,
      "grad_norm": 2.971754312515259,
      "learning_rate": 4.664513337679882e-06,
      "loss": 0.628,
      "step": 45140
    },
    {
      "epoch": 18.48,
      "grad_norm": 3.374119281768799,
      "learning_rate": 4.664491170110146e-06,
      "loss": 0.6051,
      "step": 45150
    },
    {
      "epoch": 18.49,
      "grad_norm": 1.9082196950912476,
      "learning_rate": 4.664468991909456e-06,
      "loss": 0.6371,
      "step": 45160
    },
    {
      "epoch": 18.49,
      "grad_norm": 3.154221296310425,
      "learning_rate": 4.6644468030779095e-06,
      "loss": 0.6117,
      "step": 45170
    },
    {
      "epoch": 18.49,
      "grad_norm": 2.8416600227355957,
      "learning_rate": 4.664424603615611e-06,
      "loss": 0.628,
      "step": 45180
    },
    {
      "epoch": 18.5,
      "grad_norm": 1.928298830986023,
      "learning_rate": 4.664402393522662e-06,
      "loss": 0.6431,
      "step": 45190
    },
    {
      "epoch": 18.5,
      "grad_norm": 4.013290882110596,
      "learning_rate": 4.664380172799163e-06,
      "loss": 0.6235,
      "step": 45200
    },
    {
      "epoch": 18.51,
      "grad_norm": 2.408033609390259,
      "learning_rate": 4.664357941445217e-06,
      "loss": 0.6314,
      "step": 45210
    },
    {
      "epoch": 18.51,
      "grad_norm": 3.8093724250793457,
      "learning_rate": 4.664335699460926e-06,
      "loss": 0.6141,
      "step": 45220
    },
    {
      "epoch": 18.51,
      "grad_norm": 2.8292226791381836,
      "learning_rate": 4.66431344684639e-06,
      "loss": 0.6098,
      "step": 45230
    },
    {
      "epoch": 18.52,
      "grad_norm": 2.715176582336426,
      "learning_rate": 4.664291183601714e-06,
      "loss": 0.6217,
      "step": 45240
    },
    {
      "epoch": 18.52,
      "grad_norm": 2.641578197479248,
      "learning_rate": 4.664268909726997e-06,
      "loss": 0.6122,
      "step": 45250
    },
    {
      "epoch": 18.53,
      "grad_norm": 2.6785364151000977,
      "learning_rate": 4.664246625222342e-06,
      "loss": 0.6116,
      "step": 45260
    },
    {
      "epoch": 18.53,
      "grad_norm": 2.613618850708008,
      "learning_rate": 4.664224330087851e-06,
      "loss": 0.6336,
      "step": 45270
    },
    {
      "epoch": 18.53,
      "grad_norm": 2.4450860023498535,
      "learning_rate": 4.664202024323628e-06,
      "loss": 0.617,
      "step": 45280
    },
    {
      "epoch": 18.54,
      "grad_norm": 3.175358295440674,
      "learning_rate": 4.664179707929772e-06,
      "loss": 0.6054,
      "step": 45290
    },
    {
      "epoch": 18.54,
      "grad_norm": 2.6295440196990967,
      "learning_rate": 4.664157380906388e-06,
      "loss": 0.6394,
      "step": 45300
    },
    {
      "epoch": 18.55,
      "grad_norm": 2.581779718399048,
      "learning_rate": 4.664135043253577e-06,
      "loss": 0.6095,
      "step": 45310
    },
    {
      "epoch": 18.55,
      "grad_norm": 3.188375949859619,
      "learning_rate": 4.664112694971441e-06,
      "loss": 0.637,
      "step": 45320
    },
    {
      "epoch": 18.56,
      "grad_norm": 2.7806246280670166,
      "learning_rate": 4.664090336060084e-06,
      "loss": 0.6178,
      "step": 45330
    },
    {
      "epoch": 18.56,
      "grad_norm": 3.7592732906341553,
      "learning_rate": 4.664067966519606e-06,
      "loss": 0.6193,
      "step": 45340
    },
    {
      "epoch": 18.56,
      "grad_norm": 3.155977249145508,
      "learning_rate": 4.6640455863501115e-06,
      "loss": 0.6113,
      "step": 45350
    },
    {
      "epoch": 18.57,
      "grad_norm": 3.1651101112365723,
      "learning_rate": 4.664023195551702e-06,
      "loss": 0.6063,
      "step": 45360
    },
    {
      "epoch": 18.57,
      "grad_norm": 2.2720577716827393,
      "learning_rate": 4.664000794124481e-06,
      "loss": 0.634,
      "step": 45370
    },
    {
      "epoch": 18.58,
      "grad_norm": 4.156402111053467,
      "learning_rate": 4.663978382068548e-06,
      "loss": 0.6022,
      "step": 45380
    },
    {
      "epoch": 18.58,
      "grad_norm": 1.9614917039871216,
      "learning_rate": 4.663955959384011e-06,
      "loss": 0.6225,
      "step": 45390
    },
    {
      "epoch": 18.58,
      "grad_norm": 2.4997828006744385,
      "learning_rate": 4.663933526070969e-06,
      "loss": 0.603,
      "step": 45400
    },
    {
      "epoch": 18.59,
      "grad_norm": 2.540187358856201,
      "learning_rate": 4.663911082129525e-06,
      "loss": 0.6126,
      "step": 45410
    },
    {
      "epoch": 18.59,
      "grad_norm": 3.305238723754883,
      "learning_rate": 4.663888627559783e-06,
      "loss": 0.6078,
      "step": 45420
    },
    {
      "epoch": 18.6,
      "grad_norm": 2.196474552154541,
      "learning_rate": 4.663866162361844e-06,
      "loss": 0.5982,
      "step": 45430
    },
    {
      "epoch": 18.6,
      "grad_norm": 2.479691505432129,
      "learning_rate": 4.663843686535814e-06,
      "loss": 0.633,
      "step": 45440
    },
    {
      "epoch": 18.6,
      "grad_norm": 2.902337074279785,
      "learning_rate": 4.663821200081793e-06,
      "loss": 0.6057,
      "step": 45450
    },
    {
      "epoch": 18.61,
      "grad_norm": 2.8839898109436035,
      "learning_rate": 4.663798702999885e-06,
      "loss": 0.6139,
      "step": 45460
    },
    {
      "epoch": 18.61,
      "grad_norm": 2.975221633911133,
      "learning_rate": 4.663776195290194e-06,
      "loss": 0.6211,
      "step": 45470
    },
    {
      "epoch": 18.62,
      "grad_norm": 2.4566774368286133,
      "learning_rate": 4.663753676952822e-06,
      "loss": 0.6145,
      "step": 45480
    },
    {
      "epoch": 18.62,
      "grad_norm": 2.990253448486328,
      "learning_rate": 4.663731147987872e-06,
      "loss": 0.6156,
      "step": 45490
    },
    {
      "epoch": 18.62,
      "grad_norm": 2.4909920692443848,
      "learning_rate": 4.663708608395448e-06,
      "loss": 0.625,
      "step": 45500
    },
    {
      "epoch": 18.63,
      "grad_norm": 2.539499282836914,
      "learning_rate": 4.6636860581756526e-06,
      "loss": 0.6045,
      "step": 45510
    },
    {
      "epoch": 18.63,
      "grad_norm": 2.35300350189209,
      "learning_rate": 4.66366349732859e-06,
      "loss": 0.6053,
      "step": 45520
    },
    {
      "epoch": 18.64,
      "grad_norm": 2.329617500305176,
      "learning_rate": 4.663640925854363e-06,
      "loss": 0.6081,
      "step": 45530
    },
    {
      "epoch": 18.64,
      "grad_norm": 2.5914883613586426,
      "learning_rate": 4.663618343753074e-06,
      "loss": 0.6036,
      "step": 45540
    },
    {
      "epoch": 18.65,
      "grad_norm": 2.9918694496154785,
      "learning_rate": 4.663595751024828e-06,
      "loss": 0.6214,
      "step": 45550
    },
    {
      "epoch": 18.65,
      "grad_norm": 3.7219862937927246,
      "learning_rate": 4.6635731476697284e-06,
      "loss": 0.6037,
      "step": 45560
    },
    {
      "epoch": 18.65,
      "grad_norm": 2.6789932250976562,
      "learning_rate": 4.663550533687878e-06,
      "loss": 0.6308,
      "step": 45570
    },
    {
      "epoch": 18.66,
      "grad_norm": 2.612170934677124,
      "learning_rate": 4.6635279090793795e-06,
      "loss": 0.6185,
      "step": 45580
    },
    {
      "epoch": 18.66,
      "grad_norm": 2.6224253177642822,
      "learning_rate": 4.663505273844339e-06,
      "loss": 0.6144,
      "step": 45590
    },
    {
      "epoch": 18.67,
      "grad_norm": 2.3635201454162598,
      "learning_rate": 4.663482627982858e-06,
      "loss": 0.6534,
      "step": 45600
    },
    {
      "epoch": 18.67,
      "grad_norm": 2.1160900592803955,
      "learning_rate": 4.663459971495042e-06,
      "loss": 0.6341,
      "step": 45610
    },
    {
      "epoch": 18.67,
      "grad_norm": 2.8772220611572266,
      "learning_rate": 4.663437304380993e-06,
      "loss": 0.6179,
      "step": 45620
    },
    {
      "epoch": 18.68,
      "grad_norm": 2.7729289531707764,
      "learning_rate": 4.6634146266408164e-06,
      "loss": 0.5988,
      "step": 45630
    },
    {
      "epoch": 18.68,
      "grad_norm": 2.8975698947906494,
      "learning_rate": 4.663391938274615e-06,
      "loss": 0.6137,
      "step": 45640
    },
    {
      "epoch": 18.69,
      "grad_norm": 3.1233599185943604,
      "learning_rate": 4.663369239282492e-06,
      "loss": 0.6003,
      "step": 45650
    },
    {
      "epoch": 18.69,
      "grad_norm": 2.3128607273101807,
      "learning_rate": 4.6633465296645545e-06,
      "loss": 0.6128,
      "step": 45660
    },
    {
      "epoch": 18.69,
      "grad_norm": 1.7108218669891357,
      "learning_rate": 4.6633238094209036e-06,
      "loss": 0.616,
      "step": 45670
    },
    {
      "epoch": 18.7,
      "grad_norm": 2.8481900691986084,
      "learning_rate": 4.663301078551644e-06,
      "loss": 0.6252,
      "step": 45680
    },
    {
      "epoch": 18.7,
      "grad_norm": 2.684048891067505,
      "learning_rate": 4.663278337056881e-06,
      "loss": 0.605,
      "step": 45690
    },
    {
      "epoch": 18.71,
      "grad_norm": 2.013585090637207,
      "learning_rate": 4.663255584936717e-06,
      "loss": 0.625,
      "step": 45700
    },
    {
      "epoch": 18.71,
      "grad_norm": 2.3205020427703857,
      "learning_rate": 4.663232822191258e-06,
      "loss": 0.6143,
      "step": 45710
    },
    {
      "epoch": 18.71,
      "grad_norm": 2.9482500553131104,
      "learning_rate": 4.663210048820607e-06,
      "loss": 0.6136,
      "step": 45720
    },
    {
      "epoch": 18.72,
      "grad_norm": 3.6359052658081055,
      "learning_rate": 4.663187264824869e-06,
      "loss": 0.6186,
      "step": 45730
    },
    {
      "epoch": 18.72,
      "grad_norm": 2.2956671714782715,
      "learning_rate": 4.663164470204147e-06,
      "loss": 0.6472,
      "step": 45740
    },
    {
      "epoch": 18.73,
      "grad_norm": 2.8815512657165527,
      "learning_rate": 4.663141664958548e-06,
      "loss": 0.6404,
      "step": 45750
    },
    {
      "epoch": 18.73,
      "grad_norm": 3.012413501739502,
      "learning_rate": 4.663118849088174e-06,
      "loss": 0.5977,
      "step": 45760
    },
    {
      "epoch": 18.74,
      "grad_norm": 2.8404502868652344,
      "learning_rate": 4.663096022593131e-06,
      "loss": 0.6044,
      "step": 45770
    },
    {
      "epoch": 18.74,
      "grad_norm": 3.5906360149383545,
      "learning_rate": 4.663073185473523e-06,
      "loss": 0.6222,
      "step": 45780
    },
    {
      "epoch": 18.74,
      "grad_norm": 2.37439227104187,
      "learning_rate": 4.663050337729455e-06,
      "loss": 0.6371,
      "step": 45790
    },
    {
      "epoch": 18.75,
      "grad_norm": 2.77580189704895,
      "learning_rate": 4.66302747936103e-06,
      "loss": 0.6078,
      "step": 45800
    },
    {
      "epoch": 18.75,
      "grad_norm": 2.2662713527679443,
      "learning_rate": 4.663004610368355e-06,
      "loss": 0.61,
      "step": 45810
    },
    {
      "epoch": 18.76,
      "grad_norm": 2.892908811569214,
      "learning_rate": 4.662981730751534e-06,
      "loss": 0.6206,
      "step": 45820
    },
    {
      "epoch": 18.76,
      "grad_norm": 2.5980570316314697,
      "learning_rate": 4.662958840510671e-06,
      "loss": 0.6234,
      "step": 45830
    },
    {
      "epoch": 18.76,
      "grad_norm": 2.3556227684020996,
      "learning_rate": 4.662935939645873e-06,
      "loss": 0.5946,
      "step": 45840
    },
    {
      "epoch": 18.77,
      "grad_norm": 2.3758718967437744,
      "learning_rate": 4.662913028157242e-06,
      "loss": 0.6106,
      "step": 45850
    },
    {
      "epoch": 18.77,
      "grad_norm": 2.7516674995422363,
      "learning_rate": 4.6628901060448845e-06,
      "loss": 0.6214,
      "step": 45860
    },
    {
      "epoch": 18.78,
      "grad_norm": 2.4033522605895996,
      "learning_rate": 4.6628671733089046e-06,
      "loss": 0.6304,
      "step": 45870
    },
    {
      "epoch": 18.78,
      "grad_norm": 2.9340600967407227,
      "learning_rate": 4.6628442299494095e-06,
      "loss": 0.6121,
      "step": 45880
    },
    {
      "epoch": 18.78,
      "grad_norm": 3.0846939086914062,
      "learning_rate": 4.6628212759665025e-06,
      "loss": 0.6134,
      "step": 45890
    },
    {
      "epoch": 18.79,
      "grad_norm": 2.1527719497680664,
      "learning_rate": 4.662798311360288e-06,
      "loss": 0.6057,
      "step": 45900
    },
    {
      "epoch": 18.79,
      "grad_norm": 2.6508891582489014,
      "learning_rate": 4.662775336130874e-06,
      "loss": 0.6286,
      "step": 45910
    },
    {
      "epoch": 18.8,
      "grad_norm": 4.061558723449707,
      "learning_rate": 4.662752350278364e-06,
      "loss": 0.6083,
      "step": 45920
    },
    {
      "epoch": 18.8,
      "grad_norm": 2.958080530166626,
      "learning_rate": 4.662729353802862e-06,
      "loss": 0.6114,
      "step": 45930
    },
    {
      "epoch": 18.8,
      "grad_norm": 2.17171573638916,
      "learning_rate": 4.662706346704476e-06,
      "loss": 0.6386,
      "step": 45940
    },
    {
      "epoch": 18.81,
      "grad_norm": 2.523005962371826,
      "learning_rate": 4.662683328983309e-06,
      "loss": 0.6212,
      "step": 45950
    },
    {
      "epoch": 18.81,
      "grad_norm": 2.379143714904785,
      "learning_rate": 4.662660300639468e-06,
      "loss": 0.6227,
      "step": 45960
    },
    {
      "epoch": 18.82,
      "grad_norm": 1.978300929069519,
      "learning_rate": 4.662637261673058e-06,
      "loss": 0.6006,
      "step": 45970
    },
    {
      "epoch": 18.82,
      "grad_norm": 1.9077990055084229,
      "learning_rate": 4.662614212084185e-06,
      "loss": 0.613,
      "step": 45980
    },
    {
      "epoch": 18.83,
      "grad_norm": 2.5195326805114746,
      "learning_rate": 4.662591151872954e-06,
      "loss": 0.6236,
      "step": 45990
    },
    {
      "epoch": 18.83,
      "grad_norm": 3.2027366161346436,
      "learning_rate": 4.662568081039471e-06,
      "loss": 0.6093,
      "step": 46000
    },
    {
      "epoch": 18.83,
      "grad_norm": 2.2900161743164062,
      "learning_rate": 4.6625449995838404e-06,
      "loss": 0.6148,
      "step": 46010
    },
    {
      "epoch": 18.84,
      "grad_norm": 3.0602664947509766,
      "learning_rate": 4.66252190750617e-06,
      "loss": 0.6093,
      "step": 46020
    },
    {
      "epoch": 18.84,
      "grad_norm": 2.177457571029663,
      "learning_rate": 4.662498804806564e-06,
      "loss": 0.6362,
      "step": 46030
    },
    {
      "epoch": 18.85,
      "grad_norm": 3.5954525470733643,
      "learning_rate": 4.6624756914851295e-06,
      "loss": 0.6221,
      "step": 46040
    },
    {
      "epoch": 18.85,
      "grad_norm": 2.565624952316284,
      "learning_rate": 4.662452567541971e-06,
      "loss": 0.612,
      "step": 46050
    },
    {
      "epoch": 18.85,
      "grad_norm": 2.4701523780822754,
      "learning_rate": 4.662429432977196e-06,
      "loss": 0.6502,
      "step": 46060
    },
    {
      "epoch": 18.86,
      "grad_norm": 4.291499614715576,
      "learning_rate": 4.6624062877909095e-06,
      "loss": 0.6115,
      "step": 46070
    },
    {
      "epoch": 18.86,
      "grad_norm": 3.249579668045044,
      "learning_rate": 4.6623831319832176e-06,
      "loss": 0.6162,
      "step": 46080
    },
    {
      "epoch": 18.87,
      "grad_norm": 2.276186943054199,
      "learning_rate": 4.6623599655542255e-06,
      "loss": 0.6245,
      "step": 46090
    },
    {
      "epoch": 18.87,
      "grad_norm": 2.230895519256592,
      "learning_rate": 4.662336788504042e-06,
      "loss": 0.5977,
      "step": 46100
    },
    {
      "epoch": 18.87,
      "grad_norm": 2.321554660797119,
      "learning_rate": 4.66231360083277e-06,
      "loss": 0.6226,
      "step": 46110
    },
    {
      "epoch": 18.88,
      "grad_norm": 2.9911231994628906,
      "learning_rate": 4.662290402540517e-06,
      "loss": 0.6059,
      "step": 46120
    },
    {
      "epoch": 18.88,
      "grad_norm": 2.958256483078003,
      "learning_rate": 4.662267193627391e-06,
      "loss": 0.6189,
      "step": 46130
    },
    {
      "epoch": 18.89,
      "grad_norm": 2.5748207569122314,
      "learning_rate": 4.662243974093494e-06,
      "loss": 0.6302,
      "step": 46140
    },
    {
      "epoch": 18.89,
      "grad_norm": 2.680699348449707,
      "learning_rate": 4.662220743938938e-06,
      "loss": 0.6219,
      "step": 46150
    },
    {
      "epoch": 18.89,
      "grad_norm": 3.0752174854278564,
      "learning_rate": 4.6621975031638254e-06,
      "loss": 0.621,
      "step": 46160
    },
    {
      "epoch": 18.9,
      "grad_norm": 2.6030538082122803,
      "learning_rate": 4.662174251768264e-06,
      "loss": 0.6148,
      "step": 46170
    },
    {
      "epoch": 18.9,
      "grad_norm": 2.4063210487365723,
      "learning_rate": 4.662150989752361e-06,
      "loss": 0.6249,
      "step": 46180
    },
    {
      "epoch": 18.91,
      "grad_norm": 2.477529525756836,
      "learning_rate": 4.662127717116221e-06,
      "loss": 0.631,
      "step": 46190
    },
    {
      "epoch": 18.91,
      "grad_norm": 2.662778615951538,
      "learning_rate": 4.662104433859952e-06,
      "loss": 0.599,
      "step": 46200
    },
    {
      "epoch": 18.92,
      "grad_norm": 2.728100538253784,
      "learning_rate": 4.66208113998366e-06,
      "loss": 0.6228,
      "step": 46210
    },
    {
      "epoch": 18.92,
      "grad_norm": 3.4923245906829834,
      "learning_rate": 4.662057835487453e-06,
      "loss": 0.6193,
      "step": 46220
    },
    {
      "epoch": 18.92,
      "grad_norm": 1.9268333911895752,
      "learning_rate": 4.662034520371436e-06,
      "loss": 0.63,
      "step": 46230
    },
    {
      "epoch": 18.93,
      "grad_norm": 1.8112379312515259,
      "learning_rate": 4.662011194635718e-06,
      "loss": 0.615,
      "step": 46240
    },
    {
      "epoch": 18.93,
      "grad_norm": 3.7669661045074463,
      "learning_rate": 4.661987858280403e-06,
      "loss": 0.6192,
      "step": 46250
    },
    {
      "epoch": 18.94,
      "grad_norm": 3.9847941398620605,
      "learning_rate": 4.6619645113055994e-06,
      "loss": 0.5954,
      "step": 46260
    },
    {
      "epoch": 18.94,
      "grad_norm": 2.387336492538452,
      "learning_rate": 4.661941153711415e-06,
      "loss": 0.6155,
      "step": 46270
    },
    {
      "epoch": 18.94,
      "grad_norm": 2.8800981044769287,
      "learning_rate": 4.661917785497955e-06,
      "loss": 0.5864,
      "step": 46280
    },
    {
      "epoch": 18.95,
      "grad_norm": 2.9817912578582764,
      "learning_rate": 4.661894406665328e-06,
      "loss": 0.6146,
      "step": 46290
    },
    {
      "epoch": 18.95,
      "grad_norm": 2.765449047088623,
      "learning_rate": 4.661871017213641e-06,
      "loss": 0.6284,
      "step": 46300
    },
    {
      "epoch": 18.96,
      "grad_norm": 3.875731945037842,
      "learning_rate": 4.661847617143e-06,
      "loss": 0.6146,
      "step": 46310
    },
    {
      "epoch": 18.96,
      "grad_norm": 2.6812632083892822,
      "learning_rate": 4.661824206453512e-06,
      "loss": 0.6195,
      "step": 46320
    },
    {
      "epoch": 18.96,
      "grad_norm": 2.9835150241851807,
      "learning_rate": 4.661800785145286e-06,
      "loss": 0.6141,
      "step": 46330
    },
    {
      "epoch": 18.97,
      "grad_norm": 2.2737317085266113,
      "learning_rate": 4.66177735321843e-06,
      "loss": 0.5994,
      "step": 46340
    },
    {
      "epoch": 18.97,
      "grad_norm": 2.9864466190338135,
      "learning_rate": 4.661753910673047e-06,
      "loss": 0.6315,
      "step": 46350
    },
    {
      "epoch": 18.98,
      "grad_norm": 3.8378074169158936,
      "learning_rate": 4.661730457509248e-06,
      "loss": 0.6309,
      "step": 46360
    },
    {
      "epoch": 18.98,
      "grad_norm": 2.7213730812072754,
      "learning_rate": 4.66170699372714e-06,
      "loss": 0.6214,
      "step": 46370
    },
    {
      "epoch": 18.98,
      "grad_norm": 2.8434619903564453,
      "learning_rate": 4.661683519326829e-06,
      "loss": 0.6097,
      "step": 46380
    },
    {
      "epoch": 18.99,
      "grad_norm": 2.434220314025879,
      "learning_rate": 4.6616600343084244e-06,
      "loss": 0.6184,
      "step": 46390
    },
    {
      "epoch": 18.99,
      "grad_norm": 2.624678134918213,
      "learning_rate": 4.661636538672032e-06,
      "loss": 0.5931,
      "step": 46400
    },
    {
      "epoch": 19.0,
      "grad_norm": 2.0941596031188965,
      "learning_rate": 4.661613032417761e-06,
      "loss": 0.6179,
      "step": 46410
    },
    {
      "epoch": 19.0,
      "eval_loss": 0.6149799823760986,
      "eval_runtime": 51.9153,
      "eval_samples_per_second": 66.435,
      "eval_steps_per_second": 8.321,
      "step": 46417
    },
    {
      "epoch": 19.0,
      "grad_norm": 2.436239719390869,
      "learning_rate": 4.6615895155457185e-06,
      "loss": 0.625,
      "step": 46420
    },
    {
      "epoch": 19.01,
      "grad_norm": 2.774595022201538,
      "learning_rate": 4.661565988056012e-06,
      "loss": 0.6164,
      "step": 46430
    },
    {
      "epoch": 19.01,
      "grad_norm": 2.985701560974121,
      "learning_rate": 4.66154244994875e-06,
      "loss": 0.6151,
      "step": 46440
    },
    {
      "epoch": 19.01,
      "grad_norm": 2.309112548828125,
      "learning_rate": 4.661518901224039e-06,
      "loss": 0.6233,
      "step": 46450
    },
    {
      "epoch": 19.02,
      "grad_norm": 2.688619375228882,
      "learning_rate": 4.6614953418819876e-06,
      "loss": 0.6114,
      "step": 46460
    },
    {
      "epoch": 19.02,
      "grad_norm": 3.0587987899780273,
      "learning_rate": 4.661471771922704e-06,
      "loss": 0.6128,
      "step": 46470
    },
    {
      "epoch": 19.03,
      "grad_norm": 2.519796133041382,
      "learning_rate": 4.661448191346296e-06,
      "loss": 0.6284,
      "step": 46480
    },
    {
      "epoch": 19.03,
      "grad_norm": 2.5830345153808594,
      "learning_rate": 4.661424600152872e-06,
      "loss": 0.6144,
      "step": 46490
    },
    {
      "epoch": 19.03,
      "grad_norm": 3.7863590717315674,
      "learning_rate": 4.661400998342538e-06,
      "loss": 0.6165,
      "step": 46500
    },
    {
      "epoch": 19.04,
      "grad_norm": 2.5399718284606934,
      "learning_rate": 4.661377385915405e-06,
      "loss": 0.6091,
      "step": 46510
    },
    {
      "epoch": 19.04,
      "grad_norm": 2.729732036590576,
      "learning_rate": 4.661353762871579e-06,
      "loss": 0.6086,
      "step": 46520
    },
    {
      "epoch": 19.05,
      "grad_norm": 2.831861972808838,
      "learning_rate": 4.66133012921117e-06,
      "loss": 0.6087,
      "step": 46530
    },
    {
      "epoch": 19.05,
      "grad_norm": 2.9288904666900635,
      "learning_rate": 4.661306484934285e-06,
      "loss": 0.6173,
      "step": 46540
    },
    {
      "epoch": 19.05,
      "grad_norm": 2.092949628829956,
      "learning_rate": 4.661282830041032e-06,
      "loss": 0.6243,
      "step": 46550
    },
    {
      "epoch": 19.06,
      "grad_norm": 2.3500094413757324,
      "learning_rate": 4.661259164531521e-06,
      "loss": 0.6068,
      "step": 46560
    },
    {
      "epoch": 19.06,
      "grad_norm": 3.6196470260620117,
      "learning_rate": 4.6612354884058596e-06,
      "loss": 0.6135,
      "step": 46570
    },
    {
      "epoch": 19.07,
      "grad_norm": 2.190734386444092,
      "learning_rate": 4.661211801664155e-06,
      "loss": 0.6333,
      "step": 46580
    },
    {
      "epoch": 19.07,
      "grad_norm": 2.752277135848999,
      "learning_rate": 4.661188104306517e-06,
      "loss": 0.6448,
      "step": 46590
    },
    {
      "epoch": 19.07,
      "grad_norm": 2.587902069091797,
      "learning_rate": 4.661164396333055e-06,
      "loss": 0.6231,
      "step": 46600
    },
    {
      "epoch": 19.08,
      "grad_norm": 2.1029579639434814,
      "learning_rate": 4.6611406777438756e-06,
      "loss": 0.6423,
      "step": 46610
    },
    {
      "epoch": 19.08,
      "grad_norm": 2.371999740600586,
      "learning_rate": 4.661116948539088e-06,
      "loss": 0.6073,
      "step": 46620
    },
    {
      "epoch": 19.09,
      "grad_norm": 2.0032830238342285,
      "learning_rate": 4.661093208718802e-06,
      "loss": 0.5937,
      "step": 46630
    },
    {
      "epoch": 19.09,
      "grad_norm": 4.043903350830078,
      "learning_rate": 4.661069458283126e-06,
      "loss": 0.6316,
      "step": 46640
    },
    {
      "epoch": 19.1,
      "grad_norm": 3.5066301822662354,
      "learning_rate": 4.6610456972321675e-06,
      "loss": 0.6126,
      "step": 46650
    },
    {
      "epoch": 19.1,
      "grad_norm": 2.874279022216797,
      "learning_rate": 4.661021925566038e-06,
      "loss": 0.604,
      "step": 46660
    },
    {
      "epoch": 19.1,
      "grad_norm": 3.0854990482330322,
      "learning_rate": 4.660998143284843e-06,
      "loss": 0.6243,
      "step": 46670
    },
    {
      "epoch": 19.11,
      "grad_norm": 2.780717372894287,
      "learning_rate": 4.6609743503886935e-06,
      "loss": 0.6151,
      "step": 46680
    },
    {
      "epoch": 19.11,
      "grad_norm": 2.5171570777893066,
      "learning_rate": 4.660950546877699e-06,
      "loss": 0.6194,
      "step": 46690
    },
    {
      "epoch": 19.12,
      "grad_norm": 2.6510043144226074,
      "learning_rate": 4.660926732751967e-06,
      "loss": 0.6177,
      "step": 46700
    },
    {
      "epoch": 19.12,
      "grad_norm": 2.50899338722229,
      "learning_rate": 4.6609029080116066e-06,
      "loss": 0.632,
      "step": 46710
    },
    {
      "epoch": 19.12,
      "grad_norm": 2.0906283855438232,
      "learning_rate": 4.660879072656728e-06,
      "loss": 0.6369,
      "step": 46720
    },
    {
      "epoch": 19.13,
      "grad_norm": 3.153125762939453,
      "learning_rate": 4.66085522668744e-06,
      "loss": 0.6028,
      "step": 46730
    },
    {
      "epoch": 19.13,
      "grad_norm": 2.390618085861206,
      "learning_rate": 4.660831370103852e-06,
      "loss": 0.6005,
      "step": 46740
    },
    {
      "epoch": 19.14,
      "grad_norm": 3.8288254737854004,
      "learning_rate": 4.660807502906073e-06,
      "loss": 0.6199,
      "step": 46750
    },
    {
      "epoch": 19.14,
      "grad_norm": 2.004138231277466,
      "learning_rate": 4.660783625094213e-06,
      "loss": 0.6206,
      "step": 46760
    },
    {
      "epoch": 19.14,
      "grad_norm": 2.695652723312378,
      "learning_rate": 4.660759736668379e-06,
      "loss": 0.6097,
      "step": 46770
    },
    {
      "epoch": 19.15,
      "grad_norm": 2.355153799057007,
      "learning_rate": 4.660735837628684e-06,
      "loss": 0.6168,
      "step": 46780
    },
    {
      "epoch": 19.15,
      "grad_norm": 1.9502540826797485,
      "learning_rate": 4.6607119279752355e-06,
      "loss": 0.6205,
      "step": 46790
    },
    {
      "epoch": 19.16,
      "grad_norm": 2.4103033542633057,
      "learning_rate": 4.660688007708143e-06,
      "loss": 0.6297,
      "step": 46800
    },
    {
      "epoch": 19.16,
      "grad_norm": 3.1980040073394775,
      "learning_rate": 4.660664076827516e-06,
      "loss": 0.6171,
      "step": 46810
    },
    {
      "epoch": 19.16,
      "grad_norm": 2.7072665691375732,
      "learning_rate": 4.6606401353334645e-06,
      "loss": 0.627,
      "step": 46820
    },
    {
      "epoch": 19.17,
      "grad_norm": 2.1929714679718018,
      "learning_rate": 4.6606161832260975e-06,
      "loss": 0.6048,
      "step": 46830
    },
    {
      "epoch": 19.17,
      "grad_norm": 2.058436155319214,
      "learning_rate": 4.660592220505526e-06,
      "loss": 0.5887,
      "step": 46840
    },
    {
      "epoch": 19.18,
      "grad_norm": 2.965294122695923,
      "learning_rate": 4.66056824717186e-06,
      "loss": 0.6188,
      "step": 46850
    },
    {
      "epoch": 19.18,
      "grad_norm": 2.8172545433044434,
      "learning_rate": 4.660544263225208e-06,
      "loss": 0.6181,
      "step": 46860
    },
    {
      "epoch": 19.19,
      "grad_norm": 2.5968363285064697,
      "learning_rate": 4.66052026866568e-06,
      "loss": 0.641,
      "step": 46870
    },
    {
      "epoch": 19.19,
      "grad_norm": 2.6014480590820312,
      "learning_rate": 4.660496263493385e-06,
      "loss": 0.6094,
      "step": 46880
    },
    {
      "epoch": 19.19,
      "grad_norm": 1.9482462406158447,
      "learning_rate": 4.660472247708436e-06,
      "loss": 0.6084,
      "step": 46890
    },
    {
      "epoch": 19.2,
      "grad_norm": 2.3161165714263916,
      "learning_rate": 4.6604482213109405e-06,
      "loss": 0.6101,
      "step": 46900
    },
    {
      "epoch": 19.2,
      "grad_norm": 2.4292399883270264,
      "learning_rate": 4.660424184301009e-06,
      "loss": 0.6234,
      "step": 46910
    },
    {
      "epoch": 19.21,
      "grad_norm": 2.1772303581237793,
      "learning_rate": 4.660400136678752e-06,
      "loss": 0.6095,
      "step": 46920
    },
    {
      "epoch": 19.21,
      "grad_norm": 2.123760223388672,
      "learning_rate": 4.660376078444279e-06,
      "loss": 0.6268,
      "step": 46930
    },
    {
      "epoch": 19.21,
      "grad_norm": 2.1830122470855713,
      "learning_rate": 4.660352009597701e-06,
      "loss": 0.6048,
      "step": 46940
    },
    {
      "epoch": 19.22,
      "grad_norm": 2.4655725955963135,
      "learning_rate": 4.660327930139128e-06,
      "loss": 0.5946,
      "step": 46950
    },
    {
      "epoch": 19.22,
      "grad_norm": 2.4974820613861084,
      "learning_rate": 4.66030384006867e-06,
      "loss": 0.6118,
      "step": 46960
    },
    {
      "epoch": 19.23,
      "grad_norm": 2.3704235553741455,
      "learning_rate": 4.660279739386439e-06,
      "loss": 0.6049,
      "step": 46970
    },
    {
      "epoch": 19.23,
      "grad_norm": 2.732973337173462,
      "learning_rate": 4.660255628092543e-06,
      "loss": 0.6259,
      "step": 46980
    },
    {
      "epoch": 19.23,
      "grad_norm": 2.8197944164276123,
      "learning_rate": 4.660231506187093e-06,
      "loss": 0.6274,
      "step": 46990
    },
    {
      "epoch": 19.24,
      "grad_norm": 2.3764069080352783,
      "learning_rate": 4.660207373670201e-06,
      "loss": 0.6071,
      "step": 47000
    },
    {
      "epoch": 19.24,
      "grad_norm": 2.1117358207702637,
      "learning_rate": 4.660183230541976e-06,
      "loss": 0.6088,
      "step": 47010
    },
    {
      "epoch": 19.25,
      "grad_norm": 2.385240316390991,
      "learning_rate": 4.660159076802529e-06,
      "loss": 0.6233,
      "step": 47020
    },
    {
      "epoch": 19.25,
      "grad_norm": 2.254929304122925,
      "learning_rate": 4.660134912451971e-06,
      "loss": 0.6171,
      "step": 47030
    },
    {
      "epoch": 19.26,
      "grad_norm": 2.4035723209381104,
      "learning_rate": 4.6601107374904125e-06,
      "loss": 0.5913,
      "step": 47040
    },
    {
      "epoch": 19.26,
      "grad_norm": 2.972522258758545,
      "learning_rate": 4.660086551917964e-06,
      "loss": 0.6068,
      "step": 47050
    },
    {
      "epoch": 19.26,
      "grad_norm": 2.7816734313964844,
      "learning_rate": 4.660062355734736e-06,
      "loss": 0.6153,
      "step": 47060
    },
    {
      "epoch": 19.27,
      "grad_norm": 2.883770704269409,
      "learning_rate": 4.66003814894084e-06,
      "loss": 0.6144,
      "step": 47070
    },
    {
      "epoch": 19.27,
      "grad_norm": 2.5065712928771973,
      "learning_rate": 4.660013931536388e-06,
      "loss": 0.5937,
      "step": 47080
    },
    {
      "epoch": 19.28,
      "grad_norm": 2.879607677459717,
      "learning_rate": 4.659989703521488e-06,
      "loss": 0.601,
      "step": 47090
    },
    {
      "epoch": 19.28,
      "grad_norm": 2.407590627670288,
      "learning_rate": 4.659965464896254e-06,
      "loss": 0.6029,
      "step": 47100
    },
    {
      "epoch": 19.28,
      "grad_norm": 2.374600648880005,
      "learning_rate": 4.659941215660795e-06,
      "loss": 0.6207,
      "step": 47110
    },
    {
      "epoch": 19.29,
      "grad_norm": 2.8059580326080322,
      "learning_rate": 4.6599169558152225e-06,
      "loss": 0.6302,
      "step": 47120
    },
    {
      "epoch": 19.29,
      "grad_norm": 2.1387240886688232,
      "learning_rate": 4.659892685359648e-06,
      "loss": 0.6057,
      "step": 47130
    },
    {
      "epoch": 19.3,
      "grad_norm": 2.072993516921997,
      "learning_rate": 4.659868404294182e-06,
      "loss": 0.608,
      "step": 47140
    },
    {
      "epoch": 19.3,
      "grad_norm": 2.627539873123169,
      "learning_rate": 4.659844112618938e-06,
      "loss": 0.6222,
      "step": 47150
    },
    {
      "epoch": 19.3,
      "grad_norm": 2.5653557777404785,
      "learning_rate": 4.659819810334025e-06,
      "loss": 0.6281,
      "step": 47160
    },
    {
      "epoch": 19.31,
      "grad_norm": 2.0557658672332764,
      "learning_rate": 4.659795497439554e-06,
      "loss": 0.6068,
      "step": 47170
    },
    {
      "epoch": 19.31,
      "grad_norm": 4.112739562988281,
      "learning_rate": 4.659771173935638e-06,
      "loss": 0.6084,
      "step": 47180
    },
    {
      "epoch": 19.32,
      "grad_norm": 2.7132487297058105,
      "learning_rate": 4.6597468398223875e-06,
      "loss": 0.6237,
      "step": 47190
    },
    {
      "epoch": 19.32,
      "grad_norm": 2.4731669425964355,
      "learning_rate": 4.659722495099913e-06,
      "loss": 0.6278,
      "step": 47200
    },
    {
      "epoch": 19.32,
      "grad_norm": 3.2341768741607666,
      "learning_rate": 4.659698139768329e-06,
      "loss": 0.6337,
      "step": 47210
    },
    {
      "epoch": 19.33,
      "grad_norm": 4.048704624176025,
      "learning_rate": 4.659673773827744e-06,
      "loss": 0.6082,
      "step": 47220
    },
    {
      "epoch": 19.33,
      "grad_norm": 3.0094833374023438,
      "learning_rate": 4.659649397278272e-06,
      "loss": 0.619,
      "step": 47230
    },
    {
      "epoch": 19.34,
      "grad_norm": 3.2735595703125,
      "learning_rate": 4.659625010120023e-06,
      "loss": 0.6175,
      "step": 47240
    },
    {
      "epoch": 19.34,
      "grad_norm": 2.546597480773926,
      "learning_rate": 4.659600612353109e-06,
      "loss": 0.6111,
      "step": 47250
    },
    {
      "epoch": 19.35,
      "grad_norm": 2.5724539756774902,
      "learning_rate": 4.659576203977642e-06,
      "loss": 0.6034,
      "step": 47260
    },
    {
      "epoch": 19.35,
      "grad_norm": 3.1568257808685303,
      "learning_rate": 4.659551784993733e-06,
      "loss": 0.6103,
      "step": 47270
    },
    {
      "epoch": 19.35,
      "grad_norm": 2.415987491607666,
      "learning_rate": 4.6595273554014956e-06,
      "loss": 0.6146,
      "step": 47280
    },
    {
      "epoch": 19.36,
      "grad_norm": 2.9831197261810303,
      "learning_rate": 4.659502915201041e-06,
      "loss": 0.5792,
      "step": 47290
    },
    {
      "epoch": 19.36,
      "grad_norm": 2.569502830505371,
      "learning_rate": 4.659478464392481e-06,
      "loss": 0.595,
      "step": 47300
    },
    {
      "epoch": 19.37,
      "grad_norm": 3.1287758350372314,
      "learning_rate": 4.659454002975927e-06,
      "loss": 0.6352,
      "step": 47310
    },
    {
      "epoch": 19.37,
      "grad_norm": 2.3371636867523193,
      "learning_rate": 4.659429530951492e-06,
      "loss": 0.6017,
      "step": 47320
    },
    {
      "epoch": 19.37,
      "grad_norm": 2.0702733993530273,
      "learning_rate": 4.659405048319288e-06,
      "loss": 0.642,
      "step": 47330
    },
    {
      "epoch": 19.38,
      "grad_norm": 2.5507025718688965,
      "learning_rate": 4.659380555079426e-06,
      "loss": 0.6105,
      "step": 47340
    },
    {
      "epoch": 19.38,
      "grad_norm": 2.70295786857605,
      "learning_rate": 4.659356051232019e-06,
      "loss": 0.6385,
      "step": 47350
    },
    {
      "epoch": 19.39,
      "grad_norm": 2.6481857299804688,
      "learning_rate": 4.6593315367771795e-06,
      "loss": 0.6395,
      "step": 47360
    },
    {
      "epoch": 19.39,
      "grad_norm": 2.35302734375,
      "learning_rate": 4.65930701171502e-06,
      "loss": 0.6347,
      "step": 47370
    },
    {
      "epoch": 19.39,
      "grad_norm": 2.0094985961914062,
      "learning_rate": 4.6592824760456525e-06,
      "loss": 0.5997,
      "step": 47380
    },
    {
      "epoch": 19.4,
      "grad_norm": 2.332158327102661,
      "learning_rate": 4.659257929769189e-06,
      "loss": 0.5934,
      "step": 47390
    },
    {
      "epoch": 19.4,
      "grad_norm": 3.1346535682678223,
      "learning_rate": 4.659233372885742e-06,
      "loss": 0.6026,
      "step": 47400
    },
    {
      "epoch": 19.41,
      "grad_norm": 2.481947660446167,
      "learning_rate": 4.6592088053954245e-06,
      "loss": 0.6193,
      "step": 47410
    },
    {
      "epoch": 19.41,
      "grad_norm": 2.22554349899292,
      "learning_rate": 4.6591842272983494e-06,
      "loss": 0.6274,
      "step": 47420
    },
    {
      "epoch": 19.41,
      "grad_norm": 2.442178964614868,
      "learning_rate": 4.659159638594628e-06,
      "loss": 0.5939,
      "step": 47430
    },
    {
      "epoch": 19.42,
      "grad_norm": 2.947211265563965,
      "learning_rate": 4.659135039284373e-06,
      "loss": 0.6522,
      "step": 47440
    },
    {
      "epoch": 19.42,
      "grad_norm": 2.445514440536499,
      "learning_rate": 4.659110429367699e-06,
      "loss": 0.6208,
      "step": 47450
    },
    {
      "epoch": 19.43,
      "grad_norm": 2.3698723316192627,
      "learning_rate": 4.659085808844717e-06,
      "loss": 0.6223,
      "step": 47460
    },
    {
      "epoch": 19.43,
      "grad_norm": 2.608586311340332,
      "learning_rate": 4.6590611777155405e-06,
      "loss": 0.6109,
      "step": 47470
    },
    {
      "epoch": 19.44,
      "grad_norm": 4.113999843597412,
      "learning_rate": 4.659036535980282e-06,
      "loss": 0.6077,
      "step": 47480
    },
    {
      "epoch": 19.44,
      "grad_norm": 2.3064675331115723,
      "learning_rate": 4.659011883639054e-06,
      "loss": 0.6045,
      "step": 47490
    },
    {
      "epoch": 19.44,
      "grad_norm": 2.8753879070281982,
      "learning_rate": 4.6589872206919704e-06,
      "loss": 0.6068,
      "step": 47500
    },
    {
      "epoch": 19.45,
      "grad_norm": 2.141946315765381,
      "learning_rate": 4.6589625471391425e-06,
      "loss": 0.5803,
      "step": 47510
    },
    {
      "epoch": 19.45,
      "grad_norm": 2.8189470767974854,
      "learning_rate": 4.658937862980686e-06,
      "loss": 0.6189,
      "step": 47520
    },
    {
      "epoch": 19.46,
      "grad_norm": 1.9986275434494019,
      "learning_rate": 4.658913168216712e-06,
      "loss": 0.6044,
      "step": 47530
    },
    {
      "epoch": 19.46,
      "grad_norm": 3.376312494277954,
      "learning_rate": 4.658888462847333e-06,
      "loss": 0.6304,
      "step": 47540
    },
    {
      "epoch": 19.46,
      "grad_norm": 2.6260883808135986,
      "learning_rate": 4.658863746872665e-06,
      "loss": 0.6194,
      "step": 47550
    },
    {
      "epoch": 19.47,
      "grad_norm": 2.4174087047576904,
      "learning_rate": 4.658839020292818e-06,
      "loss": 0.6235,
      "step": 47560
    },
    {
      "epoch": 19.47,
      "grad_norm": 2.3858304023742676,
      "learning_rate": 4.658814283107907e-06,
      "loss": 0.6095,
      "step": 47570
    },
    {
      "epoch": 19.48,
      "grad_norm": 2.2939364910125732,
      "learning_rate": 4.658789535318046e-06,
      "loss": 0.6171,
      "step": 47580
    },
    {
      "epoch": 19.48,
      "grad_norm": 3.1558289527893066,
      "learning_rate": 4.658764776923347e-06,
      "loss": 0.6124,
      "step": 47590
    },
    {
      "epoch": 19.48,
      "grad_norm": 4.456482887268066,
      "learning_rate": 4.658740007923924e-06,
      "loss": 0.6256,
      "step": 47600
    },
    {
      "epoch": 19.49,
      "grad_norm": 2.2347030639648438,
      "learning_rate": 4.658715228319889e-06,
      "loss": 0.6112,
      "step": 47610
    },
    {
      "epoch": 19.49,
      "grad_norm": 2.857664108276367,
      "learning_rate": 4.658690438111358e-06,
      "loss": 0.6097,
      "step": 47620
    },
    {
      "epoch": 19.5,
      "grad_norm": 2.249950408935547,
      "learning_rate": 4.6586656372984425e-06,
      "loss": 0.614,
      "step": 47630
    },
    {
      "epoch": 19.5,
      "grad_norm": 3.0335307121276855,
      "learning_rate": 4.658640825881258e-06,
      "loss": 0.6274,
      "step": 47640
    },
    {
      "epoch": 19.5,
      "grad_norm": 2.5967094898223877,
      "learning_rate": 4.658616003859917e-06,
      "loss": 0.6154,
      "step": 47650
    },
    {
      "epoch": 19.51,
      "grad_norm": 2.6291122436523438,
      "learning_rate": 4.658591171234533e-06,
      "loss": 0.6052,
      "step": 47660
    },
    {
      "epoch": 19.51,
      "grad_norm": 2.3488800525665283,
      "learning_rate": 4.65856632800522e-06,
      "loss": 0.6174,
      "step": 47670
    },
    {
      "epoch": 19.52,
      "grad_norm": 2.4888105392456055,
      "learning_rate": 4.658541474172092e-06,
      "loss": 0.6092,
      "step": 47680
    },
    {
      "epoch": 19.52,
      "grad_norm": 2.6394336223602295,
      "learning_rate": 4.658516609735263e-06,
      "loss": 0.6148,
      "step": 47690
    },
    {
      "epoch": 19.53,
      "grad_norm": 2.6617836952209473,
      "learning_rate": 4.658491734694846e-06,
      "loss": 0.6039,
      "step": 47700
    },
    {
      "epoch": 19.53,
      "grad_norm": 3.53413724899292,
      "learning_rate": 4.658466849050956e-06,
      "loss": 0.6332,
      "step": 47710
    },
    {
      "epoch": 19.53,
      "grad_norm": 2.8065741062164307,
      "learning_rate": 4.658441952803707e-06,
      "loss": 0.6304,
      "step": 47720
    },
    {
      "epoch": 19.54,
      "grad_norm": 3.1485133171081543,
      "learning_rate": 4.658417045953212e-06,
      "loss": 0.6172,
      "step": 47730
    },
    {
      "epoch": 19.54,
      "grad_norm": 3.919890880584717,
      "learning_rate": 4.658392128499586e-06,
      "loss": 0.6018,
      "step": 47740
    },
    {
      "epoch": 19.55,
      "grad_norm": 3.1859002113342285,
      "learning_rate": 4.658367200442943e-06,
      "loss": 0.6409,
      "step": 47750
    },
    {
      "epoch": 19.55,
      "grad_norm": 2.8785085678100586,
      "learning_rate": 4.658342261783397e-06,
      "loss": 0.6187,
      "step": 47760
    },
    {
      "epoch": 19.55,
      "grad_norm": 2.738903522491455,
      "learning_rate": 4.658317312521062e-06,
      "loss": 0.6035,
      "step": 47770
    },
    {
      "epoch": 19.56,
      "grad_norm": 3.5504086017608643,
      "learning_rate": 4.6582923526560526e-06,
      "loss": 0.6294,
      "step": 47780
    },
    {
      "epoch": 19.56,
      "grad_norm": 2.7487881183624268,
      "learning_rate": 4.658267382188484e-06,
      "loss": 0.6227,
      "step": 47790
    },
    {
      "epoch": 19.57,
      "grad_norm": 2.6297547817230225,
      "learning_rate": 4.658242401118469e-06,
      "loss": 0.6458,
      "step": 47800
    },
    {
      "epoch": 19.57,
      "grad_norm": 2.4487884044647217,
      "learning_rate": 4.658217409446123e-06,
      "loss": 0.6065,
      "step": 47810
    },
    {
      "epoch": 19.57,
      "grad_norm": 2.242750644683838,
      "learning_rate": 4.6581924071715596e-06,
      "loss": 0.6107,
      "step": 47820
    },
    {
      "epoch": 19.58,
      "grad_norm": 2.812012195587158,
      "learning_rate": 4.658167394294894e-06,
      "loss": 0.6305,
      "step": 47830
    },
    {
      "epoch": 19.58,
      "grad_norm": 2.274477481842041,
      "learning_rate": 4.658142370816241e-06,
      "loss": 0.5962,
      "step": 47840
    },
    {
      "epoch": 19.59,
      "grad_norm": 4.71469259262085,
      "learning_rate": 4.6581173367357164e-06,
      "loss": 0.601,
      "step": 47850
    },
    {
      "epoch": 19.59,
      "grad_norm": 2.36432147026062,
      "learning_rate": 4.658092292053432e-06,
      "loss": 0.5888,
      "step": 47860
    },
    {
      "epoch": 19.59,
      "grad_norm": 2.589533567428589,
      "learning_rate": 4.658067236769504e-06,
      "loss": 0.6193,
      "step": 47870
    },
    {
      "epoch": 19.6,
      "grad_norm": 2.5181000232696533,
      "learning_rate": 4.658042170884047e-06,
      "loss": 0.6157,
      "step": 47880
    },
    {
      "epoch": 19.6,
      "grad_norm": 2.13157057762146,
      "learning_rate": 4.658017094397177e-06,
      "loss": 0.6083,
      "step": 47890
    },
    {
      "epoch": 19.61,
      "grad_norm": 2.665475606918335,
      "learning_rate": 4.6579920073090065e-06,
      "loss": 0.6176,
      "step": 47900
    },
    {
      "epoch": 19.61,
      "grad_norm": 2.3097431659698486,
      "learning_rate": 4.657966909619652e-06,
      "loss": 0.6063,
      "step": 47910
    },
    {
      "epoch": 19.62,
      "grad_norm": 4.5918097496032715,
      "learning_rate": 4.65794180132923e-06,
      "loss": 0.6176,
      "step": 47920
    },
    {
      "epoch": 19.62,
      "grad_norm": 2.158450126647949,
      "learning_rate": 4.657916682437852e-06,
      "loss": 0.6178,
      "step": 47930
    },
    {
      "epoch": 19.62,
      "grad_norm": 2.340618848800659,
      "learning_rate": 4.657891552945635e-06,
      "loss": 0.6099,
      "step": 47940
    },
    {
      "epoch": 19.63,
      "grad_norm": 2.774646759033203,
      "learning_rate": 4.657866412852694e-06,
      "loss": 0.6131,
      "step": 47950
    },
    {
      "epoch": 19.63,
      "grad_norm": 3.920132637023926,
      "learning_rate": 4.6578412621591444e-06,
      "loss": 0.6099,
      "step": 47960
    },
    {
      "epoch": 19.64,
      "grad_norm": 2.552603244781494,
      "learning_rate": 4.657816100865102e-06,
      "loss": 0.6013,
      "step": 47970
    },
    {
      "epoch": 19.64,
      "grad_norm": 2.5895798206329346,
      "learning_rate": 4.65779092897068e-06,
      "loss": 0.6266,
      "step": 47980
    },
    {
      "epoch": 19.64,
      "grad_norm": 2.608275890350342,
      "learning_rate": 4.657765746475994e-06,
      "loss": 0.6255,
      "step": 47990
    },
    {
      "epoch": 19.65,
      "grad_norm": 2.8876616954803467,
      "learning_rate": 4.657740553381162e-06,
      "loss": 0.6084,
      "step": 48000
    },
    {
      "epoch": 19.65,
      "grad_norm": 3.465656280517578,
      "learning_rate": 4.6577153496862966e-06,
      "loss": 0.6085,
      "step": 48010
    },
    {
      "epoch": 19.66,
      "grad_norm": 3.224877119064331,
      "learning_rate": 4.657690135391515e-06,
      "loss": 0.5922,
      "step": 48020
    },
    {
      "epoch": 19.66,
      "grad_norm": 2.449028491973877,
      "learning_rate": 4.657664910496932e-06,
      "loss": 0.6179,
      "step": 48030
    },
    {
      "epoch": 19.66,
      "grad_norm": 2.185112476348877,
      "learning_rate": 4.657639675002662e-06,
      "loss": 0.5839,
      "step": 48040
    },
    {
      "epoch": 19.67,
      "grad_norm": 2.348985433578491,
      "learning_rate": 4.6576144289088235e-06,
      "loss": 0.6107,
      "step": 48050
    },
    {
      "epoch": 19.67,
      "grad_norm": 2.905299425125122,
      "learning_rate": 4.6575891722155294e-06,
      "loss": 0.6052,
      "step": 48060
    },
    {
      "epoch": 19.68,
      "grad_norm": 2.779005289077759,
      "learning_rate": 4.657563904922896e-06,
      "loss": 0.6065,
      "step": 48070
    },
    {
      "epoch": 19.68,
      "grad_norm": 2.5193564891815186,
      "learning_rate": 4.65753862703104e-06,
      "loss": 0.5919,
      "step": 48080
    },
    {
      "epoch": 19.68,
      "grad_norm": 2.5216543674468994,
      "learning_rate": 4.6575133385400764e-06,
      "loss": 0.6177,
      "step": 48090
    },
    {
      "epoch": 19.69,
      "grad_norm": 2.087618827819824,
      "learning_rate": 4.657488039450121e-06,
      "loss": 0.6114,
      "step": 48100
    },
    {
      "epoch": 19.69,
      "grad_norm": 2.6171295642852783,
      "learning_rate": 4.657462729761291e-06,
      "loss": 0.6208,
      "step": 48110
    },
    {
      "epoch": 19.7,
      "grad_norm": 2.6228511333465576,
      "learning_rate": 4.657437409473701e-06,
      "loss": 0.6251,
      "step": 48120
    },
    {
      "epoch": 19.7,
      "grad_norm": 2.1870346069335938,
      "learning_rate": 4.657412078587466e-06,
      "loss": 0.6168,
      "step": 48130
    },
    {
      "epoch": 19.71,
      "grad_norm": 3.0167040824890137,
      "learning_rate": 4.657386737102705e-06,
      "loss": 0.6156,
      "step": 48140
    },
    {
      "epoch": 19.71,
      "grad_norm": 1.987465262413025,
      "learning_rate": 4.657361385019531e-06,
      "loss": 0.613,
      "step": 48150
    },
    {
      "epoch": 19.71,
      "grad_norm": 2.433825969696045,
      "learning_rate": 4.657336022338063e-06,
      "loss": 0.6027,
      "step": 48160
    },
    {
      "epoch": 19.72,
      "grad_norm": 2.1514732837677,
      "learning_rate": 4.657310649058415e-06,
      "loss": 0.6116,
      "step": 48170
    },
    {
      "epoch": 19.72,
      "grad_norm": 2.832423210144043,
      "learning_rate": 4.657285265180704e-06,
      "loss": 0.6049,
      "step": 48180
    },
    {
      "epoch": 19.73,
      "grad_norm": 2.2930214405059814,
      "learning_rate": 4.657259870705046e-06,
      "loss": 0.6159,
      "step": 48190
    },
    {
      "epoch": 19.73,
      "grad_norm": 2.6692779064178467,
      "learning_rate": 4.657234465631558e-06,
      "loss": 0.613,
      "step": 48200
    },
    {
      "epoch": 19.73,
      "grad_norm": 1.7814031839370728,
      "learning_rate": 4.657209049960356e-06,
      "loss": 0.6193,
      "step": 48210
    },
    {
      "epoch": 19.74,
      "grad_norm": 2.4973554611206055,
      "learning_rate": 4.6571836236915554e-06,
      "loss": 0.6145,
      "step": 48220
    },
    {
      "epoch": 19.74,
      "grad_norm": 2.074580192565918,
      "learning_rate": 4.657158186825275e-06,
      "loss": 0.5993,
      "step": 48230
    },
    {
      "epoch": 19.75,
      "grad_norm": 2.446573257446289,
      "learning_rate": 4.657132739361629e-06,
      "loss": 0.6123,
      "step": 48240
    },
    {
      "epoch": 19.75,
      "grad_norm": 2.716614007949829,
      "learning_rate": 4.657107281300735e-06,
      "loss": 0.6005,
      "step": 48250
    },
    {
      "epoch": 19.75,
      "grad_norm": 3.5814554691314697,
      "learning_rate": 4.65708181264271e-06,
      "loss": 0.6095,
      "step": 48260
    },
    {
      "epoch": 19.76,
      "grad_norm": 2.6704866886138916,
      "learning_rate": 4.65705633338767e-06,
      "loss": 0.6027,
      "step": 48270
    },
    {
      "epoch": 19.76,
      "grad_norm": 2.8460943698883057,
      "learning_rate": 4.657030843535732e-06,
      "loss": 0.6106,
      "step": 48280
    },
    {
      "epoch": 19.77,
      "grad_norm": 3.2489166259765625,
      "learning_rate": 4.657005343087013e-06,
      "loss": 0.5959,
      "step": 48290
    },
    {
      "epoch": 19.77,
      "grad_norm": 3.1840174198150635,
      "learning_rate": 4.656979832041629e-06,
      "loss": 0.6132,
      "step": 48300
    },
    {
      "epoch": 19.77,
      "grad_norm": 2.532818555831909,
      "learning_rate": 4.656954310399697e-06,
      "loss": 0.6151,
      "step": 48310
    },
    {
      "epoch": 19.78,
      "grad_norm": 2.2236969470977783,
      "learning_rate": 4.6569287781613355e-06,
      "loss": 0.6131,
      "step": 48320
    },
    {
      "epoch": 19.78,
      "grad_norm": 2.1508941650390625,
      "learning_rate": 4.65690323532666e-06,
      "loss": 0.6185,
      "step": 48330
    },
    {
      "epoch": 19.79,
      "grad_norm": 3.554762601852417,
      "learning_rate": 4.656877681895787e-06,
      "loss": 0.6222,
      "step": 48340
    },
    {
      "epoch": 19.79,
      "grad_norm": 2.1705482006073,
      "learning_rate": 4.656852117868835e-06,
      "loss": 0.6327,
      "step": 48350
    },
    {
      "epoch": 19.8,
      "grad_norm": 2.553241014480591,
      "learning_rate": 4.656826543245921e-06,
      "loss": 0.6112,
      "step": 48360
    },
    {
      "epoch": 19.8,
      "grad_norm": 2.648557662963867,
      "learning_rate": 4.6568009580271604e-06,
      "loss": 0.6233,
      "step": 48370
    },
    {
      "epoch": 19.8,
      "grad_norm": 3.511692523956299,
      "learning_rate": 4.6567753622126716e-06,
      "loss": 0.6105,
      "step": 48380
    },
    {
      "epoch": 19.81,
      "grad_norm": 2.302412271499634,
      "learning_rate": 4.656749755802572e-06,
      "loss": 0.6192,
      "step": 48390
    },
    {
      "epoch": 19.81,
      "grad_norm": 2.9514691829681396,
      "learning_rate": 4.656724138796979e-06,
      "loss": 0.5944,
      "step": 48400
    },
    {
      "epoch": 19.82,
      "grad_norm": 3.5429508686065674,
      "learning_rate": 4.65669851119601e-06,
      "loss": 0.6339,
      "step": 48410
    },
    {
      "epoch": 19.82,
      "grad_norm": 2.912858724594116,
      "learning_rate": 4.6566728729997815e-06,
      "loss": 0.6017,
      "step": 48420
    },
    {
      "epoch": 19.82,
      "grad_norm": 2.4793918132781982,
      "learning_rate": 4.656647224208412e-06,
      "loss": 0.6252,
      "step": 48430
    },
    {
      "epoch": 19.83,
      "grad_norm": 3.5138542652130127,
      "learning_rate": 4.656621564822019e-06,
      "loss": 0.615,
      "step": 48440
    },
    {
      "epoch": 19.83,
      "grad_norm": 3.351743459701538,
      "learning_rate": 4.6565958948407185e-06,
      "loss": 0.6003,
      "step": 48450
    },
    {
      "epoch": 19.84,
      "grad_norm": 1.998471736907959,
      "learning_rate": 4.656570214264629e-06,
      "loss": 0.6135,
      "step": 48460
    },
    {
      "epoch": 19.84,
      "grad_norm": 2.929429531097412,
      "learning_rate": 4.656544523093868e-06,
      "loss": 0.6099,
      "step": 48470
    },
    {
      "epoch": 19.84,
      "grad_norm": 2.1453051567077637,
      "learning_rate": 4.6565188213285545e-06,
      "loss": 0.6186,
      "step": 48480
    },
    {
      "epoch": 19.85,
      "grad_norm": 3.2126717567443848,
      "learning_rate": 4.656493108968805e-06,
      "loss": 0.6172,
      "step": 48490
    },
    {
      "epoch": 19.85,
      "grad_norm": 2.3319060802459717,
      "learning_rate": 4.656467386014738e-06,
      "loss": 0.5958,
      "step": 48500
    },
    {
      "epoch": 19.86,
      "grad_norm": 3.6425533294677734,
      "learning_rate": 4.656441652466469e-06,
      "loss": 0.6257,
      "step": 48510
    },
    {
      "epoch": 19.86,
      "grad_norm": 2.2921762466430664,
      "learning_rate": 4.656415908324119e-06,
      "loss": 0.6364,
      "step": 48520
    },
    {
      "epoch": 19.86,
      "grad_norm": 2.6619205474853516,
      "learning_rate": 4.6563901535878055e-06,
      "loss": 0.6071,
      "step": 48530
    },
    {
      "epoch": 19.87,
      "grad_norm": 3.1227149963378906,
      "learning_rate": 4.6563643882576445e-06,
      "loss": 0.6059,
      "step": 48540
    },
    {
      "epoch": 19.87,
      "grad_norm": 2.676790952682495,
      "learning_rate": 4.656338612333755e-06,
      "loss": 0.62,
      "step": 48550
    },
    {
      "epoch": 19.88,
      "grad_norm": 2.8856897354125977,
      "learning_rate": 4.656312825816256e-06,
      "loss": 0.5834,
      "step": 48560
    },
    {
      "epoch": 19.88,
      "grad_norm": 3.2321174144744873,
      "learning_rate": 4.656287028705265e-06,
      "loss": 0.5927,
      "step": 48570
    },
    {
      "epoch": 19.89,
      "grad_norm": 1.983233094215393,
      "learning_rate": 4.656261221000898e-06,
      "loss": 0.5936,
      "step": 48580
    },
    {
      "epoch": 19.89,
      "grad_norm": 1.9271085262298584,
      "learning_rate": 4.656235402703278e-06,
      "loss": 0.5915,
      "step": 48590
    },
    {
      "epoch": 19.89,
      "grad_norm": 2.5423383712768555,
      "learning_rate": 4.656209573812518e-06,
      "loss": 0.6106,
      "step": 48600
    },
    {
      "epoch": 19.9,
      "grad_norm": 2.036158323287964,
      "learning_rate": 4.65618373432874e-06,
      "loss": 0.5962,
      "step": 48610
    },
    {
      "epoch": 19.9,
      "grad_norm": 2.338893175125122,
      "learning_rate": 4.6561578842520614e-06,
      "loss": 0.6133,
      "step": 48620
    },
    {
      "epoch": 19.91,
      "grad_norm": 2.897400140762329,
      "learning_rate": 4.6561320235826e-06,
      "loss": 0.5949,
      "step": 48630
    },
    {
      "epoch": 19.91,
      "grad_norm": 2.3653903007507324,
      "learning_rate": 4.6561061523204746e-06,
      "loss": 0.6369,
      "step": 48640
    },
    {
      "epoch": 19.91,
      "grad_norm": 2.420940637588501,
      "learning_rate": 4.656080270465805e-06,
      "loss": 0.6391,
      "step": 48650
    },
    {
      "epoch": 19.92,
      "grad_norm": 3.095151662826538,
      "learning_rate": 4.656054378018707e-06,
      "loss": 0.5975,
      "step": 48660
    },
    {
      "epoch": 19.92,
      "grad_norm": 2.9543745517730713,
      "learning_rate": 4.656028474979301e-06,
      "loss": 0.6336,
      "step": 48670
    },
    {
      "epoch": 19.93,
      "grad_norm": 2.7645180225372314,
      "learning_rate": 4.6560025613477064e-06,
      "loss": 0.6036,
      "step": 48680
    },
    {
      "epoch": 19.93,
      "grad_norm": 3.0565195083618164,
      "learning_rate": 4.655976637124039e-06,
      "loss": 0.6082,
      "step": 48690
    },
    {
      "epoch": 19.93,
      "grad_norm": 2.79772686958313,
      "learning_rate": 4.655950702308422e-06,
      "loss": 0.6198,
      "step": 48700
    },
    {
      "epoch": 19.94,
      "grad_norm": 2.794126272201538,
      "learning_rate": 4.65592475690097e-06,
      "loss": 0.6339,
      "step": 48710
    },
    {
      "epoch": 19.94,
      "grad_norm": 2.6271626949310303,
      "learning_rate": 4.6558988009018034e-06,
      "loss": 0.6192,
      "step": 48720
    },
    {
      "epoch": 19.95,
      "grad_norm": 1.9844321012496948,
      "learning_rate": 4.6558728343110425e-06,
      "loss": 0.6087,
      "step": 48730
    },
    {
      "epoch": 19.95,
      "grad_norm": 2.4461753368377686,
      "learning_rate": 4.655846857128804e-06,
      "loss": 0.6023,
      "step": 48740
    },
    {
      "epoch": 19.95,
      "grad_norm": 3.095381259918213,
      "learning_rate": 4.655820869355207e-06,
      "loss": 0.643,
      "step": 48750
    },
    {
      "epoch": 19.96,
      "grad_norm": 2.5203022956848145,
      "learning_rate": 4.655794870990373e-06,
      "loss": 0.6225,
      "step": 48760
    },
    {
      "epoch": 19.96,
      "grad_norm": 2.8584210872650146,
      "learning_rate": 4.655768862034419e-06,
      "loss": 0.6154,
      "step": 48770
    },
    {
      "epoch": 19.97,
      "grad_norm": 2.214662551879883,
      "learning_rate": 4.6557428424874644e-06,
      "loss": 0.6045,
      "step": 48780
    },
    {
      "epoch": 19.97,
      "grad_norm": 2.45184063911438,
      "learning_rate": 4.6557168123496295e-06,
      "loss": 0.6209,
      "step": 48790
    },
    {
      "epoch": 19.98,
      "grad_norm": 3.454333543777466,
      "learning_rate": 4.655690771621032e-06,
      "loss": 0.6154,
      "step": 48800
    },
    {
      "epoch": 19.98,
      "grad_norm": 2.0609471797943115,
      "learning_rate": 4.655664720301791e-06,
      "loss": 0.6155,
      "step": 48810
    },
    {
      "epoch": 19.98,
      "grad_norm": 2.4876439571380615,
      "learning_rate": 4.655638658392027e-06,
      "loss": 0.6303,
      "step": 48820
    },
    {
      "epoch": 19.99,
      "grad_norm": 2.3354501724243164,
      "learning_rate": 4.655612585891861e-06,
      "loss": 0.613,
      "step": 48830
    },
    {
      "epoch": 19.99,
      "grad_norm": 3.1106631755828857,
      "learning_rate": 4.655586502801409e-06,
      "loss": 0.5998,
      "step": 48840
    },
    {
      "epoch": 20.0,
      "grad_norm": 4.918071746826172,
      "learning_rate": 4.655560409120793e-06,
      "loss": 0.6037,
      "step": 48850
    },
    {
      "epoch": 20.0,
      "grad_norm": 2.9327523708343506,
      "learning_rate": 4.65553430485013e-06,
      "loss": 0.6226,
      "step": 48860
    },
    {
      "epoch": 20.0,
      "eval_loss": 0.6112197637557983,
      "eval_runtime": 52.5543,
      "eval_samples_per_second": 65.627,
      "eval_steps_per_second": 8.22,
      "step": 48860
    },
    {
      "epoch": 20.0,
      "grad_norm": 2.2187039852142334,
      "learning_rate": 4.655508189989542e-06,
      "loss": 0.5916,
      "step": 48870
    },
    {
      "epoch": 20.01,
      "grad_norm": 3.1554036140441895,
      "learning_rate": 4.655482064539147e-06,
      "loss": 0.627,
      "step": 48880
    },
    {
      "epoch": 20.01,
      "grad_norm": 4.267574310302734,
      "learning_rate": 4.6554559284990675e-06,
      "loss": 0.6167,
      "step": 48890
    },
    {
      "epoch": 20.02,
      "grad_norm": 2.189537286758423,
      "learning_rate": 4.655429781869419e-06,
      "loss": 0.6282,
      "step": 48900
    },
    {
      "epoch": 20.02,
      "grad_norm": 2.763394594192505,
      "learning_rate": 4.655403624650325e-06,
      "loss": 0.6107,
      "step": 48910
    },
    {
      "epoch": 20.02,
      "grad_norm": 2.2998809814453125,
      "learning_rate": 4.655377456841902e-06,
      "loss": 0.6127,
      "step": 48920
    },
    {
      "epoch": 20.03,
      "grad_norm": 2.2307631969451904,
      "learning_rate": 4.655351278444273e-06,
      "loss": 0.6063,
      "step": 48930
    },
    {
      "epoch": 20.03,
      "grad_norm": 2.4937453269958496,
      "learning_rate": 4.655325089457556e-06,
      "loss": 0.6094,
      "step": 48940
    },
    {
      "epoch": 20.04,
      "grad_norm": 2.401639699935913,
      "learning_rate": 4.655298889881872e-06,
      "loss": 0.5886,
      "step": 48950
    },
    {
      "epoch": 20.04,
      "grad_norm": 2.748295307159424,
      "learning_rate": 4.655272679717339e-06,
      "loss": 0.6161,
      "step": 48960
    },
    {
      "epoch": 20.05,
      "grad_norm": 4.269063472747803,
      "learning_rate": 4.655246458964081e-06,
      "loss": 0.5901,
      "step": 48970
    },
    {
      "epoch": 20.05,
      "grad_norm": 2.0426583290100098,
      "learning_rate": 4.655220227622213e-06,
      "loss": 0.6026,
      "step": 48980
    },
    {
      "epoch": 20.05,
      "grad_norm": 2.6257965564727783,
      "learning_rate": 4.65519398569186e-06,
      "loss": 0.5953,
      "step": 48990
    },
    {
      "epoch": 20.06,
      "grad_norm": 2.1095573902130127,
      "learning_rate": 4.655167733173139e-06,
      "loss": 0.6052,
      "step": 49000
    },
    {
      "epoch": 20.06,
      "grad_norm": 2.858630418777466,
      "learning_rate": 4.655141470066172e-06,
      "loss": 0.6386,
      "step": 49010
    },
    {
      "epoch": 20.07,
      "grad_norm": 2.893306016921997,
      "learning_rate": 4.6551151963710785e-06,
      "loss": 0.6207,
      "step": 49020
    },
    {
      "epoch": 20.07,
      "grad_norm": 3.1099541187286377,
      "learning_rate": 4.655088912087979e-06,
      "loss": 0.6219,
      "step": 49030
    },
    {
      "epoch": 20.07,
      "grad_norm": 3.098616600036621,
      "learning_rate": 4.655062617216994e-06,
      "loss": 0.6379,
      "step": 49040
    },
    {
      "epoch": 20.08,
      "grad_norm": 2.733222484588623,
      "learning_rate": 4.655036311758243e-06,
      "loss": 0.6146,
      "step": 49050
    },
    {
      "epoch": 20.08,
      "grad_norm": 3.833287239074707,
      "learning_rate": 4.6550099957118484e-06,
      "loss": 0.6088,
      "step": 49060
    },
    {
      "epoch": 20.09,
      "grad_norm": 2.014287233352661,
      "learning_rate": 4.654983669077929e-06,
      "loss": 0.5969,
      "step": 49070
    },
    {
      "epoch": 20.09,
      "grad_norm": 2.817556381225586,
      "learning_rate": 4.654957331856607e-06,
      "loss": 0.607,
      "step": 49080
    },
    {
      "epoch": 20.09,
      "grad_norm": 2.7954623699188232,
      "learning_rate": 4.654930984048001e-06,
      "loss": 0.6018,
      "step": 49090
    },
    {
      "epoch": 20.1,
      "grad_norm": 2.7607264518737793,
      "learning_rate": 4.654904625652233e-06,
      "loss": 0.5893,
      "step": 49100
    },
    {
      "epoch": 20.1,
      "grad_norm": 2.800565242767334,
      "learning_rate": 4.654878256669424e-06,
      "loss": 0.6056,
      "step": 49110
    },
    {
      "epoch": 20.11,
      "grad_norm": 3.3991963863372803,
      "learning_rate": 4.654851877099694e-06,
      "loss": 0.6104,
      "step": 49120
    },
    {
      "epoch": 20.11,
      "grad_norm": 3.0588316917419434,
      "learning_rate": 4.654825486943165e-06,
      "loss": 0.6389,
      "step": 49130
    },
    {
      "epoch": 20.11,
      "grad_norm": 1.916343331336975,
      "learning_rate": 4.654799086199957e-06,
      "loss": 0.603,
      "step": 49140
    },
    {
      "epoch": 20.12,
      "grad_norm": 2.65602707862854,
      "learning_rate": 4.6547726748701905e-06,
      "loss": 0.6086,
      "step": 49150
    },
    {
      "epoch": 20.12,
      "grad_norm": 2.087334156036377,
      "learning_rate": 4.654746252953987e-06,
      "loss": 0.6013,
      "step": 49160
    },
    {
      "epoch": 20.13,
      "grad_norm": 2.3243629932403564,
      "learning_rate": 4.654719820451468e-06,
      "loss": 0.6227,
      "step": 49170
    },
    {
      "epoch": 20.13,
      "grad_norm": 2.472363233566284,
      "learning_rate": 4.654693377362754e-06,
      "loss": 0.6201,
      "step": 49180
    },
    {
      "epoch": 20.14,
      "grad_norm": 2.7747368812561035,
      "learning_rate": 4.654666923687966e-06,
      "loss": 0.6073,
      "step": 49190
    },
    {
      "epoch": 20.14,
      "grad_norm": 2.4132156372070312,
      "learning_rate": 4.654640459427225e-06,
      "loss": 0.6003,
      "step": 49200
    },
    {
      "epoch": 20.14,
      "grad_norm": 2.8337621688842773,
      "learning_rate": 4.654613984580654e-06,
      "loss": 0.6142,
      "step": 49210
    },
    {
      "epoch": 20.15,
      "grad_norm": 3.8775339126586914,
      "learning_rate": 4.654587499148372e-06,
      "loss": 0.5938,
      "step": 49220
    },
    {
      "epoch": 20.15,
      "grad_norm": 3.662557363510132,
      "learning_rate": 4.654561003130502e-06,
      "loss": 0.6163,
      "step": 49230
    },
    {
      "epoch": 20.16,
      "grad_norm": 2.409585475921631,
      "learning_rate": 4.654534496527164e-06,
      "loss": 0.5982,
      "step": 49240
    },
    {
      "epoch": 20.16,
      "grad_norm": 2.3955342769622803,
      "learning_rate": 4.654507979338481e-06,
      "loss": 0.6158,
      "step": 49250
    },
    {
      "epoch": 20.16,
      "grad_norm": 2.043921709060669,
      "learning_rate": 4.654481451564573e-06,
      "loss": 0.5896,
      "step": 49260
    },
    {
      "epoch": 20.17,
      "grad_norm": 3.5631022453308105,
      "learning_rate": 4.654454913205562e-06,
      "loss": 0.6299,
      "step": 49270
    },
    {
      "epoch": 20.17,
      "grad_norm": 1.924441933631897,
      "learning_rate": 4.6544283642615695e-06,
      "loss": 0.6178,
      "step": 49280
    },
    {
      "epoch": 20.18,
      "grad_norm": 3.351980209350586,
      "learning_rate": 4.6544018047327174e-06,
      "loss": 0.6345,
      "step": 49290
    },
    {
      "epoch": 20.18,
      "grad_norm": 3.055736541748047,
      "learning_rate": 4.654375234619127e-06,
      "loss": 0.6084,
      "step": 49300
    },
    {
      "epoch": 20.18,
      "grad_norm": 2.401033639907837,
      "learning_rate": 4.654348653920921e-06,
      "loss": 0.6063,
      "step": 49310
    },
    {
      "epoch": 20.19,
      "grad_norm": 3.0297796726226807,
      "learning_rate": 4.654322062638219e-06,
      "loss": 0.5982,
      "step": 49320
    },
    {
      "epoch": 20.19,
      "grad_norm": 2.5959670543670654,
      "learning_rate": 4.654295460771146e-06,
      "loss": 0.6152,
      "step": 49330
    },
    {
      "epoch": 20.2,
      "grad_norm": 2.6638307571411133,
      "learning_rate": 4.654268848319821e-06,
      "loss": 0.6153,
      "step": 49340
    },
    {
      "epoch": 20.2,
      "grad_norm": 3.192486524581909,
      "learning_rate": 4.654242225284367e-06,
      "loss": 0.6137,
      "step": 49350
    },
    {
      "epoch": 20.2,
      "grad_norm": 2.4192252159118652,
      "learning_rate": 4.654215591664907e-06,
      "loss": 0.607,
      "step": 49360
    },
    {
      "epoch": 20.21,
      "grad_norm": 3.710367441177368,
      "learning_rate": 4.654188947461561e-06,
      "loss": 0.5845,
      "step": 49370
    },
    {
      "epoch": 20.21,
      "grad_norm": 2.756887674331665,
      "learning_rate": 4.654162292674452e-06,
      "loss": 0.6098,
      "step": 49380
    },
    {
      "epoch": 20.22,
      "grad_norm": 3.018980026245117,
      "learning_rate": 4.654135627303703e-06,
      "loss": 0.6079,
      "step": 49390
    },
    {
      "epoch": 20.22,
      "grad_norm": 2.4747557640075684,
      "learning_rate": 4.654108951349434e-06,
      "loss": 0.6133,
      "step": 49400
    },
    {
      "epoch": 20.23,
      "grad_norm": 2.2586982250213623,
      "learning_rate": 4.6540822648117685e-06,
      "loss": 0.6205,
      "step": 49410
    },
    {
      "epoch": 20.23,
      "grad_norm": 2.838430881500244,
      "learning_rate": 4.654055567690829e-06,
      "loss": 0.6034,
      "step": 49420
    },
    {
      "epoch": 20.23,
      "grad_norm": 3.209285020828247,
      "learning_rate": 4.654028859986738e-06,
      "loss": 0.6079,
      "step": 49430
    },
    {
      "epoch": 20.24,
      "grad_norm": 2.829092025756836,
      "learning_rate": 4.654002141699616e-06,
      "loss": 0.6248,
      "step": 49440
    },
    {
      "epoch": 20.24,
      "grad_norm": 3.3626184463500977,
      "learning_rate": 4.653975412829588e-06,
      "loss": 0.602,
      "step": 49450
    },
    {
      "epoch": 20.25,
      "grad_norm": 3.1131229400634766,
      "learning_rate": 4.653948673376775e-06,
      "loss": 0.6081,
      "step": 49460
    },
    {
      "epoch": 20.25,
      "grad_norm": 2.344224691390991,
      "learning_rate": 4.6539219233413e-06,
      "loss": 0.5926,
      "step": 49470
    },
    {
      "epoch": 20.25,
      "grad_norm": 2.693185567855835,
      "learning_rate": 4.6538951627232845e-06,
      "loss": 0.6006,
      "step": 49480
    },
    {
      "epoch": 20.26,
      "grad_norm": 2.855224132537842,
      "learning_rate": 4.653868391522851e-06,
      "loss": 0.614,
      "step": 49490
    },
    {
      "epoch": 20.26,
      "grad_norm": 3.3492865562438965,
      "learning_rate": 4.6538416097401246e-06,
      "loss": 0.6193,
      "step": 49500
    },
    {
      "epoch": 20.27,
      "grad_norm": 2.440385580062866,
      "learning_rate": 4.6538148173752255e-06,
      "loss": 0.5854,
      "step": 49510
    },
    {
      "epoch": 20.27,
      "grad_norm": 2.5194225311279297,
      "learning_rate": 4.653788014428277e-06,
      "loss": 0.6129,
      "step": 49520
    },
    {
      "epoch": 20.27,
      "grad_norm": 2.7162535190582275,
      "learning_rate": 4.6537612008994025e-06,
      "loss": 0.5933,
      "step": 49530
    },
    {
      "epoch": 20.28,
      "grad_norm": 1.9636746644973755,
      "learning_rate": 4.653734376788724e-06,
      "loss": 0.6179,
      "step": 49540
    },
    {
      "epoch": 20.28,
      "grad_norm": 2.484835386276245,
      "learning_rate": 4.653707542096366e-06,
      "loss": 0.6068,
      "step": 49550
    },
    {
      "epoch": 20.29,
      "grad_norm": 2.9623498916625977,
      "learning_rate": 4.653680696822449e-06,
      "loss": 0.5991,
      "step": 49560
    },
    {
      "epoch": 20.29,
      "grad_norm": 3.526850938796997,
      "learning_rate": 4.653653840967098e-06,
      "loss": 0.6345,
      "step": 49570
    },
    {
      "epoch": 20.29,
      "grad_norm": 3.0820441246032715,
      "learning_rate": 4.6536269745304345e-06,
      "loss": 0.6101,
      "step": 49580
    },
    {
      "epoch": 20.3,
      "grad_norm": 3.2927608489990234,
      "learning_rate": 4.6536000975125825e-06,
      "loss": 0.5951,
      "step": 49590
    },
    {
      "epoch": 20.3,
      "grad_norm": 2.471799373626709,
      "learning_rate": 4.653573209913665e-06,
      "loss": 0.5968,
      "step": 49600
    },
    {
      "epoch": 20.31,
      "grad_norm": 2.1156413555145264,
      "learning_rate": 4.653546311733806e-06,
      "loss": 0.5994,
      "step": 49610
    },
    {
      "epoch": 20.31,
      "grad_norm": 3.5327401161193848,
      "learning_rate": 4.653519402973127e-06,
      "loss": 0.6149,
      "step": 49620
    },
    {
      "epoch": 20.32,
      "grad_norm": 2.0712056159973145,
      "learning_rate": 4.653492483631751e-06,
      "loss": 0.5985,
      "step": 49630
    },
    {
      "epoch": 20.32,
      "grad_norm": 3.813190221786499,
      "learning_rate": 4.653465553709805e-06,
      "loss": 0.6162,
      "step": 49640
    },
    {
      "epoch": 20.32,
      "grad_norm": 2.5784621238708496,
      "learning_rate": 4.6534386132074074e-06,
      "loss": 0.6093,
      "step": 49650
    },
    {
      "epoch": 20.33,
      "grad_norm": 2.3123390674591064,
      "learning_rate": 4.6534116621246845e-06,
      "loss": 0.6163,
      "step": 49660
    },
    {
      "epoch": 20.33,
      "grad_norm": 2.496755361557007,
      "learning_rate": 4.653384700461759e-06,
      "loss": 0.5933,
      "step": 49670
    },
    {
      "epoch": 20.34,
      "grad_norm": 3.5569069385528564,
      "learning_rate": 4.653357728218756e-06,
      "loss": 0.6137,
      "step": 49680
    },
    {
      "epoch": 20.34,
      "grad_norm": 2.3315980434417725,
      "learning_rate": 4.6533307453957965e-06,
      "loss": 0.6053,
      "step": 49690
    },
    {
      "epoch": 20.34,
      "grad_norm": 2.2236199378967285,
      "learning_rate": 4.653303751993005e-06,
      "loss": 0.6015,
      "step": 49700
    },
    {
      "epoch": 20.35,
      "grad_norm": 2.9508066177368164,
      "learning_rate": 4.6532767480105065e-06,
      "loss": 0.6048,
      "step": 49710
    },
    {
      "epoch": 20.35,
      "grad_norm": 2.8893520832061768,
      "learning_rate": 4.653249733448423e-06,
      "loss": 0.6159,
      "step": 49720
    },
    {
      "epoch": 20.36,
      "grad_norm": 2.835983991622925,
      "learning_rate": 4.65322270830688e-06,
      "loss": 0.6125,
      "step": 49730
    },
    {
      "epoch": 20.36,
      "grad_norm": 2.1193244457244873,
      "learning_rate": 4.6531956725859996e-06,
      "loss": 0.6148,
      "step": 49740
    },
    {
      "epoch": 20.36,
      "grad_norm": 2.4810428619384766,
      "learning_rate": 4.653168626285905e-06,
      "loss": 0.6043,
      "step": 49750
    },
    {
      "epoch": 20.37,
      "grad_norm": 1.9854989051818848,
      "learning_rate": 4.653141569406722e-06,
      "loss": 0.6169,
      "step": 49760
    },
    {
      "epoch": 20.37,
      "grad_norm": 2.3366281986236572,
      "learning_rate": 4.653114501948575e-06,
      "loss": 0.613,
      "step": 49770
    },
    {
      "epoch": 20.38,
      "grad_norm": 3.457343816757202,
      "learning_rate": 4.6530874239115855e-06,
      "loss": 0.6163,
      "step": 49780
    },
    {
      "epoch": 20.38,
      "grad_norm": 2.7412021160125732,
      "learning_rate": 4.65306033529588e-06,
      "loss": 0.5896,
      "step": 49790
    },
    {
      "epoch": 20.38,
      "grad_norm": 2.526332378387451,
      "learning_rate": 4.6530332361015805e-06,
      "loss": 0.623,
      "step": 49800
    },
    {
      "epoch": 20.39,
      "grad_norm": 3.300891637802124,
      "learning_rate": 4.653006126328813e-06,
      "loss": 0.6112,
      "step": 49810
    },
    {
      "epoch": 20.39,
      "grad_norm": 1.8749380111694336,
      "learning_rate": 4.6529790059777e-06,
      "loss": 0.6224,
      "step": 49820
    },
    {
      "epoch": 20.4,
      "grad_norm": 2.8338794708251953,
      "learning_rate": 4.652951875048367e-06,
      "loss": 0.5906,
      "step": 49830
    },
    {
      "epoch": 20.4,
      "grad_norm": 3.671663284301758,
      "learning_rate": 4.652924733540937e-06,
      "loss": 0.5962,
      "step": 49840
    },
    {
      "epoch": 20.41,
      "grad_norm": 2.52007794380188,
      "learning_rate": 4.652897581455536e-06,
      "loss": 0.6111,
      "step": 49850
    },
    {
      "epoch": 20.41,
      "grad_norm": 2.438049793243408,
      "learning_rate": 4.6528704187922875e-06,
      "loss": 0.6014,
      "step": 49860
    },
    {
      "epoch": 20.41,
      "grad_norm": 2.837031841278076,
      "learning_rate": 4.652843245551316e-06,
      "loss": 0.611,
      "step": 49870
    },
    {
      "epoch": 20.42,
      "grad_norm": 2.2125329971313477,
      "learning_rate": 4.652816061732746e-06,
      "loss": 0.6309,
      "step": 49880
    },
    {
      "epoch": 20.42,
      "grad_norm": 4.161118030548096,
      "learning_rate": 4.6527888673367025e-06,
      "loss": 0.5897,
      "step": 49890
    },
    {
      "epoch": 20.43,
      "grad_norm": 2.2780356407165527,
      "learning_rate": 4.652761662363308e-06,
      "loss": 0.6259,
      "step": 49900
    },
    {
      "epoch": 20.43,
      "grad_norm": 3.6665475368499756,
      "learning_rate": 4.65273444681269e-06,
      "loss": 0.5925,
      "step": 49910
    },
    {
      "epoch": 20.43,
      "grad_norm": 2.422081232070923,
      "learning_rate": 4.652707220684971e-06,
      "loss": 0.6035,
      "step": 49920
    },
    {
      "epoch": 20.44,
      "grad_norm": 2.5200295448303223,
      "learning_rate": 4.6526799839802774e-06,
      "loss": 0.614,
      "step": 49930
    },
    {
      "epoch": 20.44,
      "grad_norm": 2.676168918609619,
      "learning_rate": 4.652652736698733e-06,
      "loss": 0.6148,
      "step": 49940
    },
    {
      "epoch": 20.45,
      "grad_norm": 2.19952130317688,
      "learning_rate": 4.652625478840462e-06,
      "loss": 0.6111,
      "step": 49950
    },
    {
      "epoch": 20.45,
      "grad_norm": 2.2399137020111084,
      "learning_rate": 4.652598210405591e-06,
      "loss": 0.6172,
      "step": 49960
    },
    {
      "epoch": 20.45,
      "grad_norm": 2.867305278778076,
      "learning_rate": 4.652570931394243e-06,
      "loss": 0.6097,
      "step": 49970
    },
    {
      "epoch": 20.46,
      "grad_norm": 2.1140811443328857,
      "learning_rate": 4.652543641806544e-06,
      "loss": 0.5947,
      "step": 49980
    },
    {
      "epoch": 20.46,
      "grad_norm": 2.6176018714904785,
      "learning_rate": 4.652516341642619e-06,
      "loss": 0.6111,
      "step": 49990
    },
    {
      "epoch": 20.47,
      "grad_norm": 2.6412699222564697,
      "learning_rate": 4.6524890309025925e-06,
      "loss": 0.6139,
      "step": 50000
    },
    {
      "epoch": 20.47,
      "grad_norm": 2.371073007583618,
      "learning_rate": 4.652461709586591e-06,
      "loss": 0.6018,
      "step": 50010
    },
    {
      "epoch": 20.47,
      "grad_norm": 2.9732015132904053,
      "learning_rate": 4.652434377694738e-06,
      "loss": 0.5966,
      "step": 50020
    },
    {
      "epoch": 20.48,
      "grad_norm": 3.198545455932617,
      "learning_rate": 4.65240703522716e-06,
      "loss": 0.6106,
      "step": 50030
    },
    {
      "epoch": 20.48,
      "grad_norm": 2.7516262531280518,
      "learning_rate": 4.652379682183982e-06,
      "loss": 0.6206,
      "step": 50040
    },
    {
      "epoch": 20.49,
      "grad_norm": 2.870997190475464,
      "learning_rate": 4.652352318565328e-06,
      "loss": 0.5766,
      "step": 50050
    },
    {
      "epoch": 20.49,
      "grad_norm": 2.970608949661255,
      "learning_rate": 4.652324944371325e-06,
      "loss": 0.6021,
      "step": 50060
    },
    {
      "epoch": 20.5,
      "grad_norm": 2.5505785942077637,
      "learning_rate": 4.652297559602097e-06,
      "loss": 0.6095,
      "step": 50070
    },
    {
      "epoch": 20.5,
      "grad_norm": 3.176114320755005,
      "learning_rate": 4.652270164257771e-06,
      "loss": 0.6112,
      "step": 50080
    },
    {
      "epoch": 20.5,
      "grad_norm": 2.4325294494628906,
      "learning_rate": 4.6522427583384706e-06,
      "loss": 0.6074,
      "step": 50090
    },
    {
      "epoch": 20.51,
      "grad_norm": 3.041637897491455,
      "learning_rate": 4.652215341844323e-06,
      "loss": 0.5791,
      "step": 50100
    },
    {
      "epoch": 20.51,
      "grad_norm": 2.0065085887908936,
      "learning_rate": 4.652187914775454e-06,
      "loss": 0.6013,
      "step": 50110
    },
    {
      "epoch": 20.52,
      "grad_norm": 2.1827306747436523,
      "learning_rate": 4.652160477131987e-06,
      "loss": 0.6215,
      "step": 50120
    },
    {
      "epoch": 20.52,
      "grad_norm": 3.8617091178894043,
      "learning_rate": 4.652133028914049e-06,
      "loss": 0.6121,
      "step": 50130
    },
    {
      "epoch": 20.52,
      "grad_norm": 2.9667606353759766,
      "learning_rate": 4.652105570121767e-06,
      "loss": 0.6032,
      "step": 50140
    },
    {
      "epoch": 20.53,
      "grad_norm": 3.105661630630493,
      "learning_rate": 4.652078100755265e-06,
      "loss": 0.6062,
      "step": 50150
    },
    {
      "epoch": 20.53,
      "grad_norm": 3.0765950679779053,
      "learning_rate": 4.65205062081467e-06,
      "loss": 0.5914,
      "step": 50160
    },
    {
      "epoch": 20.54,
      "grad_norm": 2.143843412399292,
      "learning_rate": 4.652023130300107e-06,
      "loss": 0.5943,
      "step": 50170
    },
    {
      "epoch": 20.54,
      "grad_norm": 3.502624034881592,
      "learning_rate": 4.651995629211702e-06,
      "loss": 0.6168,
      "step": 50180
    },
    {
      "epoch": 20.54,
      "grad_norm": 4.308856010437012,
      "learning_rate": 4.651968117549581e-06,
      "loss": 0.5955,
      "step": 50190
    },
    {
      "epoch": 20.55,
      "grad_norm": 2.1025643348693848,
      "learning_rate": 4.651940595313871e-06,
      "loss": 0.6231,
      "step": 50200
    },
    {
      "epoch": 20.55,
      "grad_norm": 2.4392778873443604,
      "learning_rate": 4.651913062504697e-06,
      "loss": 0.6029,
      "step": 50210
    },
    {
      "epoch": 20.56,
      "grad_norm": 2.816253185272217,
      "learning_rate": 4.651885519122186e-06,
      "loss": 0.5918,
      "step": 50220
    },
    {
      "epoch": 20.56,
      "grad_norm": 2.821845054626465,
      "learning_rate": 4.6518579651664625e-06,
      "loss": 0.6024,
      "step": 50230
    },
    {
      "epoch": 20.56,
      "grad_norm": 1.9164875745773315,
      "learning_rate": 4.651830400637655e-06,
      "loss": 0.6062,
      "step": 50240
    },
    {
      "epoch": 20.57,
      "grad_norm": 3.204824924468994,
      "learning_rate": 4.651802825535889e-06,
      "loss": 0.6183,
      "step": 50250
    },
    {
      "epoch": 20.57,
      "grad_norm": 2.1237974166870117,
      "learning_rate": 4.651775239861289e-06,
      "loss": 0.6326,
      "step": 50260
    },
    {
      "epoch": 20.58,
      "grad_norm": 3.0176079273223877,
      "learning_rate": 4.651747643613984e-06,
      "loss": 0.6019,
      "step": 50270
    },
    {
      "epoch": 20.58,
      "grad_norm": 2.2862069606781006,
      "learning_rate": 4.651720036794099e-06,
      "loss": 0.6103,
      "step": 50280
    },
    {
      "epoch": 20.59,
      "grad_norm": 3.2761120796203613,
      "learning_rate": 4.65169241940176e-06,
      "loss": 0.6033,
      "step": 50290
    },
    {
      "epoch": 20.59,
      "grad_norm": 3.406970262527466,
      "learning_rate": 4.651664791437094e-06,
      "loss": 0.6114,
      "step": 50300
    },
    {
      "epoch": 20.59,
      "grad_norm": 2.7840147018432617,
      "learning_rate": 4.651637152900229e-06,
      "loss": 0.6066,
      "step": 50310
    },
    {
      "epoch": 20.6,
      "grad_norm": 3.418011426925659,
      "learning_rate": 4.65160950379129e-06,
      "loss": 0.6371,
      "step": 50320
    },
    {
      "epoch": 20.6,
      "grad_norm": 2.704920768737793,
      "learning_rate": 4.651581844110403e-06,
      "loss": 0.5837,
      "step": 50330
    },
    {
      "epoch": 20.61,
      "grad_norm": 2.3396658897399902,
      "learning_rate": 4.651554173857696e-06,
      "loss": 0.6097,
      "step": 50340
    },
    {
      "epoch": 20.61,
      "grad_norm": 3.2151992321014404,
      "learning_rate": 4.6515264930332966e-06,
      "loss": 0.6138,
      "step": 50350
    },
    {
      "epoch": 20.61,
      "grad_norm": 2.447242021560669,
      "learning_rate": 4.65149880163733e-06,
      "loss": 0.6185,
      "step": 50360
    },
    {
      "epoch": 20.62,
      "grad_norm": 2.508807420730591,
      "learning_rate": 4.651471099669923e-06,
      "loss": 0.606,
      "step": 50370
    },
    {
      "epoch": 20.62,
      "grad_norm": 2.680814504623413,
      "learning_rate": 4.651443387131204e-06,
      "loss": 0.6207,
      "step": 50380
    },
    {
      "epoch": 20.63,
      "grad_norm": 2.3732001781463623,
      "learning_rate": 4.651415664021299e-06,
      "loss": 0.6074,
      "step": 50390
    },
    {
      "epoch": 20.63,
      "grad_norm": 3.231229305267334,
      "learning_rate": 4.651387930340334e-06,
      "loss": 0.6147,
      "step": 50400
    },
    {
      "epoch": 20.63,
      "grad_norm": 2.2538115978240967,
      "learning_rate": 4.651360186088437e-06,
      "loss": 0.6044,
      "step": 50410
    },
    {
      "epoch": 20.64,
      "grad_norm": 2.5945777893066406,
      "learning_rate": 4.6513324312657365e-06,
      "loss": 0.5977,
      "step": 50420
    },
    {
      "epoch": 20.64,
      "grad_norm": 2.365579843521118,
      "learning_rate": 4.651304665872357e-06,
      "loss": 0.6,
      "step": 50430
    },
    {
      "epoch": 20.65,
      "grad_norm": 2.2312607765197754,
      "learning_rate": 4.651276889908428e-06,
      "loss": 0.5977,
      "step": 50440
    },
    {
      "epoch": 20.65,
      "grad_norm": 2.4791676998138428,
      "learning_rate": 4.651249103374074e-06,
      "loss": 0.602,
      "step": 50450
    },
    {
      "epoch": 20.65,
      "grad_norm": 3.0593791007995605,
      "learning_rate": 4.651221306269425e-06,
      "loss": 0.6077,
      "step": 50460
    },
    {
      "epoch": 20.66,
      "grad_norm": 2.3127520084381104,
      "learning_rate": 4.651193498594607e-06,
      "loss": 0.6144,
      "step": 50470
    },
    {
      "epoch": 20.66,
      "grad_norm": 2.564824342727661,
      "learning_rate": 4.651165680349748e-06,
      "loss": 0.63,
      "step": 50480
    },
    {
      "epoch": 20.67,
      "grad_norm": 2.034926652908325,
      "learning_rate": 4.6511378515349754e-06,
      "loss": 0.6072,
      "step": 50490
    },
    {
      "epoch": 20.67,
      "grad_norm": 3.4574646949768066,
      "learning_rate": 4.651110012150417e-06,
      "loss": 0.6067,
      "step": 50500
    },
    {
      "epoch": 20.68,
      "grad_norm": 3.2282423973083496,
      "learning_rate": 4.651082162196198e-06,
      "loss": 0.6069,
      "step": 50510
    },
    {
      "epoch": 20.68,
      "grad_norm": 2.6060945987701416,
      "learning_rate": 4.651054301672448e-06,
      "loss": 0.6024,
      "step": 50520
    },
    {
      "epoch": 20.68,
      "grad_norm": 2.3707072734832764,
      "learning_rate": 4.651026430579295e-06,
      "loss": 0.6102,
      "step": 50530
    },
    {
      "epoch": 20.69,
      "grad_norm": 2.128807306289673,
      "learning_rate": 4.650998548916866e-06,
      "loss": 0.6015,
      "step": 50540
    },
    {
      "epoch": 20.69,
      "grad_norm": 2.55255126953125,
      "learning_rate": 4.650970656685288e-06,
      "loss": 0.6168,
      "step": 50550
    },
    {
      "epoch": 20.7,
      "grad_norm": 2.0944809913635254,
      "learning_rate": 4.6509427538846905e-06,
      "loss": 0.6232,
      "step": 50560
    },
    {
      "epoch": 20.7,
      "grad_norm": 2.879769802093506,
      "learning_rate": 4.6509148405152e-06,
      "loss": 0.6,
      "step": 50570
    },
    {
      "epoch": 20.7,
      "grad_norm": 3.0755529403686523,
      "learning_rate": 4.650886916576944e-06,
      "loss": 0.5901,
      "step": 50580
    },
    {
      "epoch": 20.71,
      "grad_norm": 2.119919538497925,
      "learning_rate": 4.650858982070052e-06,
      "loss": 0.6104,
      "step": 50590
    },
    {
      "epoch": 20.71,
      "grad_norm": 2.4513111114501953,
      "learning_rate": 4.65083103699465e-06,
      "loss": 0.6128,
      "step": 50600
    },
    {
      "epoch": 20.72,
      "grad_norm": 2.390578269958496,
      "learning_rate": 4.650803081350868e-06,
      "loss": 0.5971,
      "step": 50610
    },
    {
      "epoch": 20.72,
      "grad_norm": 2.7751545906066895,
      "learning_rate": 4.650775115138832e-06,
      "loss": 0.6135,
      "step": 50620
    },
    {
      "epoch": 20.72,
      "grad_norm": 2.6616551876068115,
      "learning_rate": 4.6507471383586735e-06,
      "loss": 0.6172,
      "step": 50630
    },
    {
      "epoch": 20.73,
      "grad_norm": 2.561638355255127,
      "learning_rate": 4.650719151010517e-06,
      "loss": 0.6288,
      "step": 50640
    },
    {
      "epoch": 20.73,
      "grad_norm": 3.7658755779266357,
      "learning_rate": 4.650691153094491e-06,
      "loss": 0.6309,
      "step": 50650
    },
    {
      "epoch": 20.74,
      "grad_norm": 3.0662360191345215,
      "learning_rate": 4.650663144610726e-06,
      "loss": 0.6035,
      "step": 50660
    },
    {
      "epoch": 20.74,
      "grad_norm": 3.135788679122925,
      "learning_rate": 4.6506351255593495e-06,
      "loss": 0.5928,
      "step": 50670
    },
    {
      "epoch": 20.74,
      "grad_norm": 3.3446784019470215,
      "learning_rate": 4.650607095940489e-06,
      "loss": 0.6041,
      "step": 50680
    },
    {
      "epoch": 20.75,
      "grad_norm": 2.615553379058838,
      "learning_rate": 4.6505790557542735e-06,
      "loss": 0.604,
      "step": 50690
    },
    {
      "epoch": 20.75,
      "grad_norm": 2.2925331592559814,
      "learning_rate": 4.650551005000831e-06,
      "loss": 0.5964,
      "step": 50700
    },
    {
      "epoch": 20.76,
      "grad_norm": 2.5892140865325928,
      "learning_rate": 4.650522943680291e-06,
      "loss": 0.6097,
      "step": 50710
    },
    {
      "epoch": 20.76,
      "grad_norm": 3.406376361846924,
      "learning_rate": 4.650494871792781e-06,
      "loss": 0.6026,
      "step": 50720
    },
    {
      "epoch": 20.77,
      "grad_norm": 2.4953951835632324,
      "learning_rate": 4.65046678933843e-06,
      "loss": 0.6276,
      "step": 50730
    },
    {
      "epoch": 20.77,
      "grad_norm": 2.809443235397339,
      "learning_rate": 4.6504386963173665e-06,
      "loss": 0.6213,
      "step": 50740
    },
    {
      "epoch": 20.77,
      "grad_norm": 2.1923019886016846,
      "learning_rate": 4.65041059272972e-06,
      "loss": 0.6256,
      "step": 50750
    },
    {
      "epoch": 20.78,
      "grad_norm": 3.3020148277282715,
      "learning_rate": 4.650382478575618e-06,
      "loss": 0.6135,
      "step": 50760
    },
    {
      "epoch": 20.78,
      "grad_norm": 2.0376548767089844,
      "learning_rate": 4.650354353855189e-06,
      "loss": 0.6138,
      "step": 50770
    },
    {
      "epoch": 20.79,
      "grad_norm": 2.723139762878418,
      "learning_rate": 4.650326218568564e-06,
      "loss": 0.6264,
      "step": 50780
    },
    {
      "epoch": 20.79,
      "grad_norm": 2.8121814727783203,
      "learning_rate": 4.650298072715871e-06,
      "loss": 0.6064,
      "step": 50790
    },
    {
      "epoch": 20.79,
      "grad_norm": 2.516010046005249,
      "learning_rate": 4.6502699162972365e-06,
      "loss": 0.6226,
      "step": 50800
    },
    {
      "epoch": 20.8,
      "grad_norm": 2.357743263244629,
      "learning_rate": 4.650241749312793e-06,
      "loss": 0.6412,
      "step": 50810
    },
    {
      "epoch": 20.8,
      "grad_norm": 1.9953217506408691,
      "learning_rate": 4.650213571762667e-06,
      "loss": 0.5854,
      "step": 50820
    },
    {
      "epoch": 20.81,
      "grad_norm": 3.348583936691284,
      "learning_rate": 4.65018538364699e-06,
      "loss": 0.617,
      "step": 50830
    },
    {
      "epoch": 20.81,
      "grad_norm": 2.255136489868164,
      "learning_rate": 4.650157184965888e-06,
      "loss": 0.5983,
      "step": 50840
    },
    {
      "epoch": 20.81,
      "grad_norm": 2.8340187072753906,
      "learning_rate": 4.650128975719494e-06,
      "loss": 0.6165,
      "step": 50850
    },
    {
      "epoch": 20.82,
      "grad_norm": 2.6380717754364014,
      "learning_rate": 4.650100755907933e-06,
      "loss": 0.5946,
      "step": 50860
    },
    {
      "epoch": 20.82,
      "grad_norm": 3.0799825191497803,
      "learning_rate": 4.650072525531338e-06,
      "loss": 0.5946,
      "step": 50870
    },
    {
      "epoch": 20.83,
      "grad_norm": 2.612711191177368,
      "learning_rate": 4.650044284589835e-06,
      "loss": 0.5921,
      "step": 50880
    },
    {
      "epoch": 20.83,
      "grad_norm": 3.020378351211548,
      "learning_rate": 4.650016033083556e-06,
      "loss": 0.6176,
      "step": 50890
    },
    {
      "epoch": 20.84,
      "grad_norm": 2.9356729984283447,
      "learning_rate": 4.64998777101263e-06,
      "loss": 0.6015,
      "step": 50900
    },
    {
      "epoch": 20.84,
      "grad_norm": 2.142566442489624,
      "learning_rate": 4.649959498377186e-06,
      "loss": 0.6113,
      "step": 50910
    },
    {
      "epoch": 20.84,
      "grad_norm": 2.2585649490356445,
      "learning_rate": 4.649931215177352e-06,
      "loss": 0.5895,
      "step": 50920
    },
    {
      "epoch": 20.85,
      "grad_norm": 3.092466354370117,
      "learning_rate": 4.64990292141326e-06,
      "loss": 0.6052,
      "step": 50930
    },
    {
      "epoch": 20.85,
      "grad_norm": 2.6477346420288086,
      "learning_rate": 4.649874617085039e-06,
      "loss": 0.5983,
      "step": 50940
    },
    {
      "epoch": 20.86,
      "grad_norm": 2.4271068572998047,
      "learning_rate": 4.649846302192817e-06,
      "loss": 0.5845,
      "step": 50950
    },
    {
      "epoch": 20.86,
      "grad_norm": 3.47702693939209,
      "learning_rate": 4.649817976736726e-06,
      "loss": 0.6122,
      "step": 50960
    },
    {
      "epoch": 20.86,
      "grad_norm": 2.744121789932251,
      "learning_rate": 4.6497896407168944e-06,
      "loss": 0.6107,
      "step": 50970
    },
    {
      "epoch": 20.87,
      "grad_norm": 2.684072256088257,
      "learning_rate": 4.649761294133452e-06,
      "loss": 0.5946,
      "step": 50980
    },
    {
      "epoch": 20.87,
      "grad_norm": 2.4417436122894287,
      "learning_rate": 4.6497329369865306e-06,
      "loss": 0.6224,
      "step": 50990
    },
    {
      "epoch": 20.88,
      "grad_norm": 3.225612163543701,
      "learning_rate": 4.649704569276257e-06,
      "loss": 0.5824,
      "step": 51000
    },
    {
      "epoch": 20.88,
      "grad_norm": 2.9335639476776123,
      "learning_rate": 4.6496761910027624e-06,
      "loss": 0.5961,
      "step": 51010
    },
    {
      "epoch": 20.88,
      "grad_norm": 2.867795944213867,
      "learning_rate": 4.649647802166178e-06,
      "loss": 0.6128,
      "step": 51020
    },
    {
      "epoch": 20.89,
      "grad_norm": 2.627286672592163,
      "learning_rate": 4.649619402766632e-06,
      "loss": 0.6098,
      "step": 51030
    },
    {
      "epoch": 20.89,
      "grad_norm": 2.536224365234375,
      "learning_rate": 4.649590992804256e-06,
      "loss": 0.629,
      "step": 51040
    },
    {
      "epoch": 20.9,
      "grad_norm": 2.651949405670166,
      "learning_rate": 4.649562572279179e-06,
      "loss": 0.6203,
      "step": 51050
    },
    {
      "epoch": 20.9,
      "grad_norm": 3.21887469291687,
      "learning_rate": 4.649534141191533e-06,
      "loss": 0.608,
      "step": 51060
    },
    {
      "epoch": 20.9,
      "grad_norm": 3.312021255493164,
      "learning_rate": 4.649505699541445e-06,
      "loss": 0.614,
      "step": 51070
    },
    {
      "epoch": 20.91,
      "grad_norm": 2.4911532402038574,
      "learning_rate": 4.649477247329049e-06,
      "loss": 0.6121,
      "step": 51080
    },
    {
      "epoch": 20.91,
      "grad_norm": 3.3769404888153076,
      "learning_rate": 4.649448784554472e-06,
      "loss": 0.592,
      "step": 51090
    },
    {
      "epoch": 20.92,
      "grad_norm": 2.4660942554473877,
      "learning_rate": 4.649420311217847e-06,
      "loss": 0.6288,
      "step": 51100
    },
    {
      "epoch": 20.92,
      "grad_norm": 3.1962695121765137,
      "learning_rate": 4.6493918273193035e-06,
      "loss": 0.5951,
      "step": 51110
    },
    {
      "epoch": 20.93,
      "grad_norm": 1.9835772514343262,
      "learning_rate": 4.649363332858971e-06,
      "loss": 0.594,
      "step": 51120
    },
    {
      "epoch": 20.93,
      "grad_norm": 2.588317394256592,
      "learning_rate": 4.649334827836982e-06,
      "loss": 0.6279,
      "step": 51130
    },
    {
      "epoch": 20.93,
      "grad_norm": 3.1401188373565674,
      "learning_rate": 4.649306312253465e-06,
      "loss": 0.6136,
      "step": 51140
    },
    {
      "epoch": 20.94,
      "grad_norm": 2.8245460987091064,
      "learning_rate": 4.649277786108552e-06,
      "loss": 0.6127,
      "step": 51150
    },
    {
      "epoch": 20.94,
      "grad_norm": 2.6996631622314453,
      "learning_rate": 4.649249249402373e-06,
      "loss": 0.6055,
      "step": 51160
    },
    {
      "epoch": 20.95,
      "grad_norm": 2.4507274627685547,
      "learning_rate": 4.64922070213506e-06,
      "loss": 0.5768,
      "step": 51170
    },
    {
      "epoch": 20.95,
      "grad_norm": 2.6028525829315186,
      "learning_rate": 4.649192144306741e-06,
      "loss": 0.6111,
      "step": 51180
    },
    {
      "epoch": 20.95,
      "grad_norm": 2.397667646408081,
      "learning_rate": 4.64916357591755e-06,
      "loss": 0.6243,
      "step": 51190
    },
    {
      "epoch": 20.96,
      "grad_norm": 2.809457540512085,
      "learning_rate": 4.649134996967616e-06,
      "loss": 0.6194,
      "step": 51200
    },
    {
      "epoch": 20.96,
      "grad_norm": 2.9293384552001953,
      "learning_rate": 4.64910640745707e-06,
      "loss": 0.6238,
      "step": 51210
    },
    {
      "epoch": 20.97,
      "grad_norm": 2.0112643241882324,
      "learning_rate": 4.6490778073860435e-06,
      "loss": 0.6106,
      "step": 51220
    },
    {
      "epoch": 20.97,
      "grad_norm": 2.1976125240325928,
      "learning_rate": 4.649049196754667e-06,
      "loss": 0.6256,
      "step": 51230
    },
    {
      "epoch": 20.97,
      "grad_norm": 2.542097330093384,
      "learning_rate": 4.649020575563073e-06,
      "loss": 0.6017,
      "step": 51240
    },
    {
      "epoch": 20.98,
      "grad_norm": 2.8570780754089355,
      "learning_rate": 4.6489919438113915e-06,
      "loss": 0.6003,
      "step": 51250
    },
    {
      "epoch": 20.98,
      "grad_norm": 3.103435754776001,
      "learning_rate": 4.6489633014997525e-06,
      "loss": 0.6098,
      "step": 51260
    },
    {
      "epoch": 20.99,
      "grad_norm": 2.749793529510498,
      "learning_rate": 4.648934648628289e-06,
      "loss": 0.6032,
      "step": 51270
    },
    {
      "epoch": 20.99,
      "grad_norm": 2.908785104751587,
      "learning_rate": 4.648905985197131e-06,
      "loss": 0.6183,
      "step": 51280
    },
    {
      "epoch": 20.99,
      "grad_norm": 2.1259613037109375,
      "learning_rate": 4.648877311206412e-06,
      "loss": 0.5932,
      "step": 51290
    },
    {
      "epoch": 21.0,
      "grad_norm": 3.053684711456299,
      "learning_rate": 4.64884862665626e-06,
      "loss": 0.5876,
      "step": 51300
    },
    {
      "epoch": 21.0,
      "eval_loss": 0.6061630845069885,
      "eval_runtime": 51.9386,
      "eval_samples_per_second": 66.405,
      "eval_steps_per_second": 8.318,
      "step": 51303
    },
    {
      "epoch": 21.0,
      "grad_norm": 2.6199941635131836,
      "learning_rate": 4.64881993154681e-06,
      "loss": 0.589,
      "step": 51310
    },
    {
      "epoch": 21.01,
      "grad_norm": 2.459686040878296,
      "learning_rate": 4.64879122587819e-06,
      "loss": 0.6373,
      "step": 51320
    },
    {
      "epoch": 21.01,
      "grad_norm": 2.628009080886841,
      "learning_rate": 4.648762509650534e-06,
      "loss": 0.5989,
      "step": 51330
    },
    {
      "epoch": 21.02,
      "grad_norm": 2.2654807567596436,
      "learning_rate": 4.648733782863973e-06,
      "loss": 0.6069,
      "step": 51340
    },
    {
      "epoch": 21.02,
      "grad_norm": 2.216130018234253,
      "learning_rate": 4.648705045518637e-06,
      "loss": 0.6021,
      "step": 51350
    },
    {
      "epoch": 21.02,
      "grad_norm": 3.0564725399017334,
      "learning_rate": 4.6486762976146596e-06,
      "loss": 0.6172,
      "step": 51360
    },
    {
      "epoch": 21.03,
      "grad_norm": 1.755202054977417,
      "learning_rate": 4.648647539152172e-06,
      "loss": 0.5925,
      "step": 51370
    },
    {
      "epoch": 21.03,
      "grad_norm": 1.9876898527145386,
      "learning_rate": 4.648618770131305e-06,
      "loss": 0.601,
      "step": 51380
    },
    {
      "epoch": 21.04,
      "grad_norm": 3.0491507053375244,
      "learning_rate": 4.648589990552191e-06,
      "loss": 0.5981,
      "step": 51390
    },
    {
      "epoch": 21.04,
      "grad_norm": 2.5850765705108643,
      "learning_rate": 4.648561200414963e-06,
      "loss": 0.6126,
      "step": 51400
    },
    {
      "epoch": 21.04,
      "grad_norm": 2.247148275375366,
      "learning_rate": 4.648532399719752e-06,
      "loss": 0.5993,
      "step": 51410
    },
    {
      "epoch": 21.05,
      "grad_norm": 2.1950790882110596,
      "learning_rate": 4.648503588466689e-06,
      "loss": 0.6067,
      "step": 51420
    },
    {
      "epoch": 21.05,
      "grad_norm": 3.855445623397827,
      "learning_rate": 4.6484747666559065e-06,
      "loss": 0.6119,
      "step": 51430
    },
    {
      "epoch": 21.06,
      "grad_norm": 4.07977294921875,
      "learning_rate": 4.648445934287538e-06,
      "loss": 0.6193,
      "step": 51440
    },
    {
      "epoch": 21.06,
      "grad_norm": 2.8911325931549072,
      "learning_rate": 4.648417091361713e-06,
      "loss": 0.6281,
      "step": 51450
    },
    {
      "epoch": 21.06,
      "grad_norm": 1.906150221824646,
      "learning_rate": 4.648388237878565e-06,
      "loss": 0.6086,
      "step": 51460
    },
    {
      "epoch": 21.07,
      "grad_norm": 1.8143348693847656,
      "learning_rate": 4.648359373838227e-06,
      "loss": 0.6116,
      "step": 51470
    },
    {
      "epoch": 21.07,
      "grad_norm": 2.9626526832580566,
      "learning_rate": 4.64833049924083e-06,
      "loss": 0.6126,
      "step": 51480
    },
    {
      "epoch": 21.08,
      "grad_norm": 3.091977596282959,
      "learning_rate": 4.648301614086507e-06,
      "loss": 0.6138,
      "step": 51490
    },
    {
      "epoch": 21.08,
      "grad_norm": 2.593275308609009,
      "learning_rate": 4.64827271837539e-06,
      "loss": 0.6203,
      "step": 51500
    },
    {
      "epoch": 21.08,
      "grad_norm": 2.6811845302581787,
      "learning_rate": 4.64824381210761e-06,
      "loss": 0.6103,
      "step": 51510
    },
    {
      "epoch": 21.09,
      "grad_norm": 2.72843074798584,
      "learning_rate": 4.648214895283302e-06,
      "loss": 0.6216,
      "step": 51520
    },
    {
      "epoch": 21.09,
      "grad_norm": 2.3726582527160645,
      "learning_rate": 4.648185967902597e-06,
      "loss": 0.6264,
      "step": 51530
    },
    {
      "epoch": 21.1,
      "grad_norm": 2.4593043327331543,
      "learning_rate": 4.648157029965628e-06,
      "loss": 0.6132,
      "step": 51540
    },
    {
      "epoch": 21.1,
      "grad_norm": 3.252146005630493,
      "learning_rate": 4.648128081472527e-06,
      "loss": 0.5896,
      "step": 51550
    },
    {
      "epoch": 21.11,
      "grad_norm": 2.453667163848877,
      "learning_rate": 4.648099122423426e-06,
      "loss": 0.5891,
      "step": 51560
    },
    {
      "epoch": 21.11,
      "grad_norm": 2.4126169681549072,
      "learning_rate": 4.64807015281846e-06,
      "loss": 0.6152,
      "step": 51570
    },
    {
      "epoch": 21.11,
      "grad_norm": 2.326063394546509,
      "learning_rate": 4.64804117265776e-06,
      "loss": 0.5974,
      "step": 51580
    },
    {
      "epoch": 21.12,
      "grad_norm": 2.7170424461364746,
      "learning_rate": 4.648012181941458e-06,
      "loss": 0.5992,
      "step": 51590
    },
    {
      "epoch": 21.12,
      "grad_norm": 2.767014741897583,
      "learning_rate": 4.647983180669689e-06,
      "loss": 0.6334,
      "step": 51600
    },
    {
      "epoch": 21.13,
      "grad_norm": 2.9449727535247803,
      "learning_rate": 4.647954168842584e-06,
      "loss": 0.5712,
      "step": 51610
    },
    {
      "epoch": 21.13,
      "grad_norm": 1.993545413017273,
      "learning_rate": 4.647925146460276e-06,
      "loss": 0.6187,
      "step": 51620
    },
    {
      "epoch": 21.13,
      "grad_norm": 2.661362886428833,
      "learning_rate": 4.6478961135228995e-06,
      "loss": 0.6292,
      "step": 51630
    },
    {
      "epoch": 21.14,
      "grad_norm": 3.0280442237854004,
      "learning_rate": 4.647867070030586e-06,
      "loss": 0.6404,
      "step": 51640
    },
    {
      "epoch": 21.14,
      "grad_norm": 3.526599884033203,
      "learning_rate": 4.647838015983469e-06,
      "loss": 0.6115,
      "step": 51650
    },
    {
      "epoch": 21.15,
      "grad_norm": 2.285827398300171,
      "learning_rate": 4.647808951381682e-06,
      "loss": 0.6046,
      "step": 51660
    },
    {
      "epoch": 21.15,
      "grad_norm": 3.2824723720550537,
      "learning_rate": 4.647779876225357e-06,
      "loss": 0.614,
      "step": 51670
    },
    {
      "epoch": 21.15,
      "grad_norm": 2.7981038093566895,
      "learning_rate": 4.6477507905146295e-06,
      "loss": 0.6141,
      "step": 51680
    },
    {
      "epoch": 21.16,
      "grad_norm": 2.1892848014831543,
      "learning_rate": 4.64772169424963e-06,
      "loss": 0.602,
      "step": 51690
    },
    {
      "epoch": 21.16,
      "grad_norm": 2.2738542556762695,
      "learning_rate": 4.647692587430493e-06,
      "loss": 0.6021,
      "step": 51700
    },
    {
      "epoch": 21.17,
      "grad_norm": 2.8434946537017822,
      "learning_rate": 4.647663470057352e-06,
      "loss": 0.6148,
      "step": 51710
    },
    {
      "epoch": 21.17,
      "grad_norm": 2.3843674659729004,
      "learning_rate": 4.64763434213034e-06,
      "loss": 0.6099,
      "step": 51720
    },
    {
      "epoch": 21.17,
      "grad_norm": 2.3430840969085693,
      "learning_rate": 4.647605203649591e-06,
      "loss": 0.5998,
      "step": 51730
    },
    {
      "epoch": 21.18,
      "grad_norm": 2.125304698944092,
      "learning_rate": 4.647576054615238e-06,
      "loss": 0.6086,
      "step": 51740
    },
    {
      "epoch": 21.18,
      "grad_norm": 3.2748653888702393,
      "learning_rate": 4.647546895027415e-06,
      "loss": 0.6024,
      "step": 51750
    },
    {
      "epoch": 21.19,
      "grad_norm": 2.314098358154297,
      "learning_rate": 4.647517724886254e-06,
      "loss": 0.6097,
      "step": 51760
    },
    {
      "epoch": 21.19,
      "grad_norm": 1.9910093545913696,
      "learning_rate": 4.647488544191891e-06,
      "loss": 0.6175,
      "step": 51770
    },
    {
      "epoch": 21.2,
      "grad_norm": 2.623871088027954,
      "learning_rate": 4.647459352944458e-06,
      "loss": 0.6014,
      "step": 51780
    },
    {
      "epoch": 21.2,
      "grad_norm": 2.5678200721740723,
      "learning_rate": 4.647430151144089e-06,
      "loss": 0.607,
      "step": 51790
    },
    {
      "epoch": 21.2,
      "grad_norm": 2.524115562438965,
      "learning_rate": 4.647400938790919e-06,
      "loss": 0.615,
      "step": 51800
    },
    {
      "epoch": 21.21,
      "grad_norm": 3.530247211456299,
      "learning_rate": 4.64737171588508e-06,
      "loss": 0.6021,
      "step": 51810
    },
    {
      "epoch": 21.21,
      "grad_norm": 2.4868228435516357,
      "learning_rate": 4.6473424824267065e-06,
      "loss": 0.6053,
      "step": 51820
    },
    {
      "epoch": 21.22,
      "grad_norm": 3.1810526847839355,
      "learning_rate": 4.647313238415933e-06,
      "loss": 0.6106,
      "step": 51830
    },
    {
      "epoch": 21.22,
      "grad_norm": 2.7231414318084717,
      "learning_rate": 4.647283983852893e-06,
      "loss": 0.5998,
      "step": 51840
    },
    {
      "epoch": 21.22,
      "grad_norm": 3.095803737640381,
      "learning_rate": 4.647254718737721e-06,
      "loss": 0.6116,
      "step": 51850
    },
    {
      "epoch": 21.23,
      "grad_norm": 2.2628841400146484,
      "learning_rate": 4.64722544307055e-06,
      "loss": 0.607,
      "step": 51860
    },
    {
      "epoch": 21.23,
      "grad_norm": 2.927234649658203,
      "learning_rate": 4.647196156851515e-06,
      "loss": 0.5974,
      "step": 51870
    },
    {
      "epoch": 21.24,
      "grad_norm": 2.3840646743774414,
      "learning_rate": 4.64716686008075e-06,
      "loss": 0.5881,
      "step": 51880
    },
    {
      "epoch": 21.24,
      "grad_norm": 2.4934568405151367,
      "learning_rate": 4.647137552758389e-06,
      "loss": 0.5982,
      "step": 51890
    },
    {
      "epoch": 21.24,
      "grad_norm": 2.26725697517395,
      "learning_rate": 4.647108234884566e-06,
      "loss": 0.6132,
      "step": 51900
    },
    {
      "epoch": 21.25,
      "grad_norm": 2.432624101638794,
      "learning_rate": 4.647078906459416e-06,
      "loss": 0.5971,
      "step": 51910
    },
    {
      "epoch": 21.25,
      "grad_norm": 1.8695636987686157,
      "learning_rate": 4.647049567483073e-06,
      "loss": 0.6083,
      "step": 51920
    },
    {
      "epoch": 21.26,
      "grad_norm": 2.319112539291382,
      "learning_rate": 4.647020217955671e-06,
      "loss": 0.6276,
      "step": 51930
    },
    {
      "epoch": 21.26,
      "grad_norm": 2.1816139221191406,
      "learning_rate": 4.646990857877346e-06,
      "loss": 0.5992,
      "step": 51940
    },
    {
      "epoch": 21.26,
      "grad_norm": 3.868347406387329,
      "learning_rate": 4.64696148724823e-06,
      "loss": 0.6071,
      "step": 51950
    },
    {
      "epoch": 21.27,
      "grad_norm": 2.6113362312316895,
      "learning_rate": 4.6469321060684595e-06,
      "loss": 0.6178,
      "step": 51960
    },
    {
      "epoch": 21.27,
      "grad_norm": 3.944986581802368,
      "learning_rate": 4.646902714338168e-06,
      "loss": 0.5922,
      "step": 51970
    },
    {
      "epoch": 21.28,
      "grad_norm": 2.1723475456237793,
      "learning_rate": 4.646873312057492e-06,
      "loss": 0.6104,
      "step": 51980
    },
    {
      "epoch": 21.28,
      "grad_norm": 2.6881277561187744,
      "learning_rate": 4.646843899226563e-06,
      "loss": 0.6091,
      "step": 51990
    },
    {
      "epoch": 21.29,
      "grad_norm": 2.397261619567871,
      "learning_rate": 4.646814475845518e-06,
      "loss": 0.5767,
      "step": 52000
    },
    {
      "epoch": 21.29,
      "grad_norm": 2.6060283184051514,
      "learning_rate": 4.646785041914491e-06,
      "loss": 0.5968,
      "step": 52010
    },
    {
      "epoch": 21.29,
      "grad_norm": 2.3767716884613037,
      "learning_rate": 4.646755597433618e-06,
      "loss": 0.6206,
      "step": 52020
    },
    {
      "epoch": 21.3,
      "grad_norm": 2.506190061569214,
      "learning_rate": 4.646726142403033e-06,
      "loss": 0.6037,
      "step": 52030
    },
    {
      "epoch": 21.3,
      "grad_norm": 3.2920639514923096,
      "learning_rate": 4.646696676822871e-06,
      "loss": 0.5899,
      "step": 52040
    },
    {
      "epoch": 21.31,
      "grad_norm": 3.0170440673828125,
      "learning_rate": 4.646667200693266e-06,
      "loss": 0.613,
      "step": 52050
    },
    {
      "epoch": 21.31,
      "grad_norm": 1.8058903217315674,
      "learning_rate": 4.646637714014355e-06,
      "loss": 0.5854,
      "step": 52060
    },
    {
      "epoch": 21.31,
      "grad_norm": 2.184004545211792,
      "learning_rate": 4.646608216786271e-06,
      "loss": 0.5988,
      "step": 52070
    },
    {
      "epoch": 21.32,
      "grad_norm": 2.0923166275024414,
      "learning_rate": 4.646578709009151e-06,
      "loss": 0.59,
      "step": 52080
    },
    {
      "epoch": 21.32,
      "grad_norm": 2.475970506668091,
      "learning_rate": 4.646549190683129e-06,
      "loss": 0.6153,
      "step": 52090
    },
    {
      "epoch": 21.33,
      "grad_norm": 2.2442400455474854,
      "learning_rate": 4.64651966180834e-06,
      "loss": 0.6013,
      "step": 52100
    },
    {
      "epoch": 21.33,
      "grad_norm": 2.6262099742889404,
      "learning_rate": 4.646490122384921e-06,
      "loss": 0.5991,
      "step": 52110
    },
    {
      "epoch": 21.33,
      "grad_norm": 2.164034366607666,
      "learning_rate": 4.646460572413006e-06,
      "loss": 0.61,
      "step": 52120
    },
    {
      "epoch": 21.34,
      "grad_norm": 2.6509714126586914,
      "learning_rate": 4.6464310118927296e-06,
      "loss": 0.6036,
      "step": 52130
    },
    {
      "epoch": 21.34,
      "grad_norm": 3.278028726577759,
      "learning_rate": 4.6464014408242285e-06,
      "loss": 0.572,
      "step": 52140
    },
    {
      "epoch": 21.35,
      "grad_norm": 2.559216022491455,
      "learning_rate": 4.646371859207638e-06,
      "loss": 0.5941,
      "step": 52150
    },
    {
      "epoch": 21.35,
      "grad_norm": 2.7960681915283203,
      "learning_rate": 4.646342267043093e-06,
      "loss": 0.6099,
      "step": 52160
    },
    {
      "epoch": 21.35,
      "grad_norm": 2.363461494445801,
      "learning_rate": 4.64631266433073e-06,
      "loss": 0.606,
      "step": 52170
    },
    {
      "epoch": 21.36,
      "grad_norm": 2.962318181991577,
      "learning_rate": 4.646283051070683e-06,
      "loss": 0.631,
      "step": 52180
    },
    {
      "epoch": 21.36,
      "grad_norm": 4.617443561553955,
      "learning_rate": 4.646253427263089e-06,
      "loss": 0.594,
      "step": 52190
    },
    {
      "epoch": 21.37,
      "grad_norm": 2.332749843597412,
      "learning_rate": 4.646223792908083e-06,
      "loss": 0.624,
      "step": 52200
    },
    {
      "epoch": 21.37,
      "grad_norm": 2.4522008895874023,
      "learning_rate": 4.646194148005803e-06,
      "loss": 0.6191,
      "step": 52210
    },
    {
      "epoch": 21.38,
      "grad_norm": 2.1252601146698,
      "learning_rate": 4.646164492556381e-06,
      "loss": 0.6259,
      "step": 52220
    },
    {
      "epoch": 21.38,
      "grad_norm": 2.1184377670288086,
      "learning_rate": 4.646134826559956e-06,
      "loss": 0.6076,
      "step": 52230
    },
    {
      "epoch": 21.38,
      "grad_norm": 3.0295681953430176,
      "learning_rate": 4.646105150016662e-06,
      "loss": 0.6098,
      "step": 52240
    },
    {
      "epoch": 21.39,
      "grad_norm": 2.2656917572021484,
      "learning_rate": 4.646075462926636e-06,
      "loss": 0.5857,
      "step": 52250
    },
    {
      "epoch": 21.39,
      "grad_norm": 2.1829304695129395,
      "learning_rate": 4.646045765290013e-06,
      "loss": 0.6017,
      "step": 52260
    },
    {
      "epoch": 21.4,
      "grad_norm": 2.7203752994537354,
      "learning_rate": 4.6460160571069305e-06,
      "loss": 0.5889,
      "step": 52270
    },
    {
      "epoch": 21.4,
      "grad_norm": 2.0423500537872314,
      "learning_rate": 4.645986338377524e-06,
      "loss": 0.6083,
      "step": 52280
    },
    {
      "epoch": 21.4,
      "grad_norm": 2.6479949951171875,
      "learning_rate": 4.645956609101928e-06,
      "loss": 0.6174,
      "step": 52290
    },
    {
      "epoch": 21.41,
      "grad_norm": 2.7271976470947266,
      "learning_rate": 4.645926869280281e-06,
      "loss": 0.5977,
      "step": 52300
    },
    {
      "epoch": 21.41,
      "grad_norm": 2.398966073989868,
      "learning_rate": 4.645897118912718e-06,
      "loss": 0.604,
      "step": 52310
    },
    {
      "epoch": 21.42,
      "grad_norm": 2.177122116088867,
      "learning_rate": 4.645867357999376e-06,
      "loss": 0.6002,
      "step": 52320
    },
    {
      "epoch": 21.42,
      "grad_norm": 2.0115349292755127,
      "learning_rate": 4.64583758654039e-06,
      "loss": 0.5896,
      "step": 52330
    },
    {
      "epoch": 21.42,
      "grad_norm": 2.8315036296844482,
      "learning_rate": 4.645807804535899e-06,
      "loss": 0.595,
      "step": 52340
    },
    {
      "epoch": 21.43,
      "grad_norm": 2.1203999519348145,
      "learning_rate": 4.645778011986036e-06,
      "loss": 0.5892,
      "step": 52350
    },
    {
      "epoch": 21.43,
      "grad_norm": 2.5433568954467773,
      "learning_rate": 4.6457482088909395e-06,
      "loss": 0.635,
      "step": 52360
    },
    {
      "epoch": 21.44,
      "grad_norm": 2.267151117324829,
      "learning_rate": 4.645718395250746e-06,
      "loss": 0.6219,
      "step": 52370
    },
    {
      "epoch": 21.44,
      "grad_norm": 2.437307834625244,
      "learning_rate": 4.645688571065592e-06,
      "loss": 0.5925,
      "step": 52380
    },
    {
      "epoch": 21.44,
      "grad_norm": 2.7098605632781982,
      "learning_rate": 4.645658736335614e-06,
      "loss": 0.6464,
      "step": 52390
    },
    {
      "epoch": 21.45,
      "grad_norm": 1.9666420221328735,
      "learning_rate": 4.645628891060947e-06,
      "loss": 0.6102,
      "step": 52400
    },
    {
      "epoch": 21.45,
      "grad_norm": 2.694688081741333,
      "learning_rate": 4.645599035241731e-06,
      "loss": 0.6183,
      "step": 52410
    },
    {
      "epoch": 21.46,
      "grad_norm": 3.275298833847046,
      "learning_rate": 4.6455691688781004e-06,
      "loss": 0.5931,
      "step": 52420
    },
    {
      "epoch": 21.46,
      "grad_norm": 3.0116512775421143,
      "learning_rate": 4.645539291970193e-06,
      "loss": 0.5975,
      "step": 52430
    },
    {
      "epoch": 21.47,
      "grad_norm": 2.5408935546875,
      "learning_rate": 4.645509404518146e-06,
      "loss": 0.6039,
      "step": 52440
    },
    {
      "epoch": 21.47,
      "grad_norm": 2.458376407623291,
      "learning_rate": 4.645479506522094e-06,
      "loss": 0.6103,
      "step": 52450
    },
    {
      "epoch": 21.47,
      "grad_norm": 2.5800628662109375,
      "learning_rate": 4.645449597982176e-06,
      "loss": 0.5689,
      "step": 52460
    },
    {
      "epoch": 21.48,
      "grad_norm": 2.173206090927124,
      "learning_rate": 4.64541967889853e-06,
      "loss": 0.5947,
      "step": 52470
    },
    {
      "epoch": 21.48,
      "grad_norm": 2.763214349746704,
      "learning_rate": 4.64538974927129e-06,
      "loss": 0.5928,
      "step": 52480
    },
    {
      "epoch": 21.49,
      "grad_norm": 2.4331603050231934,
      "learning_rate": 4.645359809100596e-06,
      "loss": 0.604,
      "step": 52490
    },
    {
      "epoch": 21.49,
      "grad_norm": 2.5670318603515625,
      "learning_rate": 4.645329858386583e-06,
      "loss": 0.5833,
      "step": 52500
    },
    {
      "epoch": 21.49,
      "grad_norm": 2.529154062271118,
      "learning_rate": 4.6452998971293895e-06,
      "loss": 0.6026,
      "step": 52510
    },
    {
      "epoch": 21.5,
      "grad_norm": 2.6390037536621094,
      "learning_rate": 4.645269925329153e-06,
      "loss": 0.6187,
      "step": 52520
    },
    {
      "epoch": 21.5,
      "grad_norm": 2.5069901943206787,
      "learning_rate": 4.645239942986009e-06,
      "loss": 0.6251,
      "step": 52530
    },
    {
      "epoch": 21.51,
      "grad_norm": 2.653940200805664,
      "learning_rate": 4.645209950100097e-06,
      "loss": 0.5942,
      "step": 52540
    },
    {
      "epoch": 21.51,
      "grad_norm": 2.4128024578094482,
      "learning_rate": 4.645179946671552e-06,
      "loss": 0.5853,
      "step": 52550
    },
    {
      "epoch": 21.51,
      "grad_norm": 2.961561679840088,
      "learning_rate": 4.645149932700514e-06,
      "loss": 0.5873,
      "step": 52560
    },
    {
      "epoch": 21.52,
      "grad_norm": 2.672081470489502,
      "learning_rate": 4.645119908187119e-06,
      "loss": 0.5933,
      "step": 52570
    },
    {
      "epoch": 21.52,
      "grad_norm": 2.154693603515625,
      "learning_rate": 4.645089873131505e-06,
      "loss": 0.6083,
      "step": 52580
    },
    {
      "epoch": 21.53,
      "grad_norm": 2.8606085777282715,
      "learning_rate": 4.645059827533809e-06,
      "loss": 0.5944,
      "step": 52590
    },
    {
      "epoch": 21.53,
      "grad_norm": 2.1535720825195312,
      "learning_rate": 4.645029771394169e-06,
      "loss": 0.618,
      "step": 52600
    },
    {
      "epoch": 21.53,
      "grad_norm": 3.1265547275543213,
      "learning_rate": 4.644999704712724e-06,
      "loss": 0.6239,
      "step": 52610
    },
    {
      "epoch": 21.54,
      "grad_norm": 1.9372283220291138,
      "learning_rate": 4.64496962748961e-06,
      "loss": 0.615,
      "step": 52620
    },
    {
      "epoch": 21.54,
      "grad_norm": 2.4981019496917725,
      "learning_rate": 4.644939539724965e-06,
      "loss": 0.6216,
      "step": 52630
    },
    {
      "epoch": 21.55,
      "grad_norm": 3.3753249645233154,
      "learning_rate": 4.644909441418927e-06,
      "loss": 0.6128,
      "step": 52640
    },
    {
      "epoch": 21.55,
      "grad_norm": 1.8681048154830933,
      "learning_rate": 4.644879332571634e-06,
      "loss": 0.59,
      "step": 52650
    },
    {
      "epoch": 21.56,
      "grad_norm": 2.489607095718384,
      "learning_rate": 4.644849213183224e-06,
      "loss": 0.5916,
      "step": 52660
    },
    {
      "epoch": 21.56,
      "grad_norm": 3.6486668586730957,
      "learning_rate": 4.6448190832538345e-06,
      "loss": 0.5968,
      "step": 52670
    },
    {
      "epoch": 21.56,
      "grad_norm": 2.7421839237213135,
      "learning_rate": 4.6447889427836045e-06,
      "loss": 0.6011,
      "step": 52680
    },
    {
      "epoch": 21.57,
      "grad_norm": 3.264948606491089,
      "learning_rate": 4.644758791772671e-06,
      "loss": 0.6043,
      "step": 52690
    },
    {
      "epoch": 21.57,
      "grad_norm": 2.700977087020874,
      "learning_rate": 4.6447286302211735e-06,
      "loss": 0.5862,
      "step": 52700
    },
    {
      "epoch": 21.58,
      "grad_norm": 2.337780714035034,
      "learning_rate": 4.644698458129248e-06,
      "loss": 0.5905,
      "step": 52710
    },
    {
      "epoch": 21.58,
      "grad_norm": 3.3071675300598145,
      "learning_rate": 4.644668275497035e-06,
      "loss": 0.601,
      "step": 52720
    },
    {
      "epoch": 21.58,
      "grad_norm": 2.0577805042266846,
      "learning_rate": 4.644638082324671e-06,
      "loss": 0.6044,
      "step": 52730
    },
    {
      "epoch": 21.59,
      "grad_norm": 2.133074998855591,
      "learning_rate": 4.644607878612295e-06,
      "loss": 0.6008,
      "step": 52740
    },
    {
      "epoch": 21.59,
      "grad_norm": 2.1658968925476074,
      "learning_rate": 4.6445776643600464e-06,
      "loss": 0.6227,
      "step": 52750
    },
    {
      "epoch": 21.6,
      "grad_norm": 2.181391954421997,
      "learning_rate": 4.644547439568061e-06,
      "loss": 0.6166,
      "step": 52760
    },
    {
      "epoch": 21.6,
      "grad_norm": 2.731431245803833,
      "learning_rate": 4.64451720423648e-06,
      "loss": 0.5921,
      "step": 52770
    },
    {
      "epoch": 21.6,
      "grad_norm": 2.7536919116973877,
      "learning_rate": 4.64448695836544e-06,
      "loss": 0.6068,
      "step": 52780
    },
    {
      "epoch": 21.61,
      "grad_norm": 2.134721279144287,
      "learning_rate": 4.644456701955081e-06,
      "loss": 0.6111,
      "step": 52790
    },
    {
      "epoch": 21.61,
      "grad_norm": 3.193549156188965,
      "learning_rate": 4.644426435005541e-06,
      "loss": 0.6205,
      "step": 52800
    },
    {
      "epoch": 21.62,
      "grad_norm": 2.3146119117736816,
      "learning_rate": 4.644396157516958e-06,
      "loss": 0.5783,
      "step": 52810
    },
    {
      "epoch": 21.62,
      "grad_norm": 2.8455922603607178,
      "learning_rate": 4.644365869489471e-06,
      "loss": 0.5955,
      "step": 52820
    },
    {
      "epoch": 21.63,
      "grad_norm": 2.537309408187866,
      "learning_rate": 4.64433557092322e-06,
      "loss": 0.5918,
      "step": 52830
    },
    {
      "epoch": 21.63,
      "grad_norm": 2.3589000701904297,
      "learning_rate": 4.644305261818342e-06,
      "loss": 0.5998,
      "step": 52840
    },
    {
      "epoch": 21.63,
      "grad_norm": 3.8111326694488525,
      "learning_rate": 4.644274942174977e-06,
      "loss": 0.6135,
      "step": 52850
    },
    {
      "epoch": 21.64,
      "grad_norm": 2.4841535091400146,
      "learning_rate": 4.6442446119932635e-06,
      "loss": 0.5926,
      "step": 52860
    },
    {
      "epoch": 21.64,
      "grad_norm": 2.23590350151062,
      "learning_rate": 4.64421427127334e-06,
      "loss": 0.6277,
      "step": 52870
    },
    {
      "epoch": 21.65,
      "grad_norm": 2.662019729614258,
      "learning_rate": 4.644183920015346e-06,
      "loss": 0.6002,
      "step": 52880
    },
    {
      "epoch": 21.65,
      "grad_norm": 2.8142552375793457,
      "learning_rate": 4.644153558219422e-06,
      "loss": 0.6003,
      "step": 52890
    },
    {
      "epoch": 21.65,
      "grad_norm": 2.825611114501953,
      "learning_rate": 4.644123185885704e-06,
      "loss": 0.5919,
      "step": 52900
    },
    {
      "epoch": 21.66,
      "grad_norm": 2.478999137878418,
      "learning_rate": 4.644092803014333e-06,
      "loss": 0.5804,
      "step": 52910
    },
    {
      "epoch": 21.66,
      "grad_norm": 2.4494829177856445,
      "learning_rate": 4.644062409605448e-06,
      "loss": 0.5797,
      "step": 52920
    },
    {
      "epoch": 21.67,
      "grad_norm": 2.4906980991363525,
      "learning_rate": 4.644032005659188e-06,
      "loss": 0.593,
      "step": 52930
    },
    {
      "epoch": 21.67,
      "grad_norm": 3.7918734550476074,
      "learning_rate": 4.644001591175692e-06,
      "loss": 0.5919,
      "step": 52940
    },
    {
      "epoch": 21.67,
      "grad_norm": 2.3186135292053223,
      "learning_rate": 4.643971166155101e-06,
      "loss": 0.6139,
      "step": 52950
    },
    {
      "epoch": 21.68,
      "grad_norm": 2.4908063411712646,
      "learning_rate": 4.643940730597551e-06,
      "loss": 0.6321,
      "step": 52960
    },
    {
      "epoch": 21.68,
      "grad_norm": 2.815767288208008,
      "learning_rate": 4.643910284503185e-06,
      "loss": 0.6118,
      "step": 52970
    },
    {
      "epoch": 21.69,
      "grad_norm": 3.089226007461548,
      "learning_rate": 4.643879827872141e-06,
      "loss": 0.5892,
      "step": 52980
    },
    {
      "epoch": 21.69,
      "grad_norm": 2.28629207611084,
      "learning_rate": 4.6438493607045584e-06,
      "loss": 0.5811,
      "step": 52990
    },
    {
      "epoch": 21.69,
      "grad_norm": 2.1987695693969727,
      "learning_rate": 4.643818883000576e-06,
      "loss": 0.6,
      "step": 53000
    },
    {
      "epoch": 21.7,
      "grad_norm": 2.5925445556640625,
      "learning_rate": 4.643788394760335e-06,
      "loss": 0.6152,
      "step": 53010
    },
    {
      "epoch": 21.7,
      "grad_norm": 2.2842912673950195,
      "learning_rate": 4.643757895983974e-06,
      "loss": 0.6194,
      "step": 53020
    },
    {
      "epoch": 21.71,
      "grad_norm": 2.615117311477661,
      "learning_rate": 4.643727386671633e-06,
      "loss": 0.5891,
      "step": 53030
    },
    {
      "epoch": 21.71,
      "grad_norm": 2.8005106449127197,
      "learning_rate": 4.643696866823452e-06,
      "loss": 0.6342,
      "step": 53040
    },
    {
      "epoch": 21.72,
      "grad_norm": 2.270331621170044,
      "learning_rate": 4.643666336439571e-06,
      "loss": 0.6314,
      "step": 53050
    },
    {
      "epoch": 21.72,
      "grad_norm": 2.2333364486694336,
      "learning_rate": 4.6436357955201285e-06,
      "loss": 0.5942,
      "step": 53060
    },
    {
      "epoch": 21.72,
      "grad_norm": 3.2338149547576904,
      "learning_rate": 4.643605244065266e-06,
      "loss": 0.6038,
      "step": 53070
    },
    {
      "epoch": 21.73,
      "grad_norm": 2.3401455879211426,
      "learning_rate": 4.643574682075123e-06,
      "loss": 0.6125,
      "step": 53080
    },
    {
      "epoch": 21.73,
      "grad_norm": 2.8468117713928223,
      "learning_rate": 4.643544109549838e-06,
      "loss": 0.5836,
      "step": 53090
    },
    {
      "epoch": 21.74,
      "grad_norm": 2.9235594272613525,
      "learning_rate": 4.643513526489554e-06,
      "loss": 0.5893,
      "step": 53100
    },
    {
      "epoch": 21.74,
      "grad_norm": 2.0026333332061768,
      "learning_rate": 4.643482932894408e-06,
      "loss": 0.6153,
      "step": 53110
    },
    {
      "epoch": 21.74,
      "grad_norm": 2.1595122814178467,
      "learning_rate": 4.6434523287645425e-06,
      "loss": 0.6214,
      "step": 53120
    },
    {
      "epoch": 21.75,
      "grad_norm": 2.6710431575775146,
      "learning_rate": 4.643421714100097e-06,
      "loss": 0.6019,
      "step": 53130
    },
    {
      "epoch": 21.75,
      "grad_norm": 2.7428767681121826,
      "learning_rate": 4.643391088901211e-06,
      "loss": 0.6098,
      "step": 53140
    },
    {
      "epoch": 21.76,
      "grad_norm": 4.059986114501953,
      "learning_rate": 4.643360453168026e-06,
      "loss": 0.6047,
      "step": 53150
    },
    {
      "epoch": 21.76,
      "grad_norm": 2.096367359161377,
      "learning_rate": 4.643329806900681e-06,
      "loss": 0.6084,
      "step": 53160
    },
    {
      "epoch": 21.76,
      "grad_norm": 2.699958562850952,
      "learning_rate": 4.643299150099317e-06,
      "loss": 0.588,
      "step": 53170
    },
    {
      "epoch": 21.77,
      "grad_norm": 2.500455141067505,
      "learning_rate": 4.6432684827640744e-06,
      "loss": 0.6106,
      "step": 53180
    },
    {
      "epoch": 21.77,
      "grad_norm": 2.561638593673706,
      "learning_rate": 4.643237804895094e-06,
      "loss": 0.6026,
      "step": 53190
    },
    {
      "epoch": 21.78,
      "grad_norm": 2.7111260890960693,
      "learning_rate": 4.6432071164925165e-06,
      "loss": 0.5902,
      "step": 53200
    },
    {
      "epoch": 21.78,
      "grad_norm": 2.162770986557007,
      "learning_rate": 4.643176417556481e-06,
      "loss": 0.6202,
      "step": 53210
    },
    {
      "epoch": 21.78,
      "grad_norm": 2.7819483280181885,
      "learning_rate": 4.64314570808713e-06,
      "loss": 0.6152,
      "step": 53220
    },
    {
      "epoch": 21.79,
      "grad_norm": 2.9064829349517822,
      "learning_rate": 4.6431149880846036e-06,
      "loss": 0.6056,
      "step": 53230
    },
    {
      "epoch": 21.79,
      "grad_norm": 2.5279006958007812,
      "learning_rate": 4.643084257549041e-06,
      "loss": 0.611,
      "step": 53240
    },
    {
      "epoch": 21.8,
      "grad_norm": 2.5661556720733643,
      "learning_rate": 4.643053516480586e-06,
      "loss": 0.6017,
      "step": 53250
    },
    {
      "epoch": 21.8,
      "grad_norm": 2.7947232723236084,
      "learning_rate": 4.643022764879377e-06,
      "loss": 0.6068,
      "step": 53260
    },
    {
      "epoch": 21.81,
      "grad_norm": 2.043759346008301,
      "learning_rate": 4.642992002745556e-06,
      "loss": 0.6141,
      "step": 53270
    },
    {
      "epoch": 21.81,
      "grad_norm": 3.173354148864746,
      "learning_rate": 4.642961230079262e-06,
      "loss": 0.6009,
      "step": 53280
    },
    {
      "epoch": 21.81,
      "grad_norm": 2.0823192596435547,
      "learning_rate": 4.642930446880638e-06,
      "loss": 0.611,
      "step": 53290
    },
    {
      "epoch": 21.82,
      "grad_norm": 2.12477445602417,
      "learning_rate": 4.642899653149825e-06,
      "loss": 0.5948,
      "step": 53300
    },
    {
      "epoch": 21.82,
      "grad_norm": 2.7761142253875732,
      "learning_rate": 4.6428688488869635e-06,
      "loss": 0.5925,
      "step": 53310
    },
    {
      "epoch": 21.83,
      "grad_norm": 2.3353116512298584,
      "learning_rate": 4.642838034092194e-06,
      "loss": 0.6007,
      "step": 53320
    },
    {
      "epoch": 21.83,
      "grad_norm": 2.099336862564087,
      "learning_rate": 4.642807208765659e-06,
      "loss": 0.6076,
      "step": 53330
    },
    {
      "epoch": 21.83,
      "grad_norm": 2.5204720497131348,
      "learning_rate": 4.6427763729074985e-06,
      "loss": 0.611,
      "step": 53340
    },
    {
      "epoch": 21.84,
      "grad_norm": 2.42917537689209,
      "learning_rate": 4.6427455265178546e-06,
      "loss": 0.5905,
      "step": 53350
    },
    {
      "epoch": 21.84,
      "grad_norm": 2.464709758758545,
      "learning_rate": 4.642714669596868e-06,
      "loss": 0.6064,
      "step": 53360
    },
    {
      "epoch": 21.85,
      "grad_norm": 2.9343252182006836,
      "learning_rate": 4.64268380214468e-06,
      "loss": 0.5922,
      "step": 53370
    },
    {
      "epoch": 21.85,
      "grad_norm": 2.6559267044067383,
      "learning_rate": 4.642652924161432e-06,
      "loss": 0.6298,
      "step": 53380
    },
    {
      "epoch": 21.85,
      "grad_norm": 3.1844232082366943,
      "learning_rate": 4.642622035647267e-06,
      "loss": 0.5932,
      "step": 53390
    },
    {
      "epoch": 21.86,
      "grad_norm": 3.494455337524414,
      "learning_rate": 4.6425911366023246e-06,
      "loss": 0.6281,
      "step": 53400
    },
    {
      "epoch": 21.86,
      "grad_norm": 3.391342878341675,
      "learning_rate": 4.642560227026746e-06,
      "loss": 0.5953,
      "step": 53410
    },
    {
      "epoch": 21.87,
      "grad_norm": 2.7340664863586426,
      "learning_rate": 4.642529306920675e-06,
      "loss": 0.6087,
      "step": 53420
    },
    {
      "epoch": 21.87,
      "grad_norm": 3.4210290908813477,
      "learning_rate": 4.6424983762842525e-06,
      "loss": 0.5844,
      "step": 53430
    },
    {
      "epoch": 21.87,
      "grad_norm": 2.7867043018341064,
      "learning_rate": 4.642467435117618e-06,
      "loss": 0.62,
      "step": 53440
    },
    {
      "epoch": 21.88,
      "grad_norm": 2.372238874435425,
      "learning_rate": 4.642436483420916e-06,
      "loss": 0.6193,
      "step": 53450
    },
    {
      "epoch": 21.88,
      "grad_norm": 2.349907636642456,
      "learning_rate": 4.642405521194287e-06,
      "loss": 0.5991,
      "step": 53460
    },
    {
      "epoch": 21.89,
      "grad_norm": 2.051988363265991,
      "learning_rate": 4.642374548437873e-06,
      "loss": 0.5984,
      "step": 53470
    },
    {
      "epoch": 21.89,
      "grad_norm": 2.087625503540039,
      "learning_rate": 4.642343565151815e-06,
      "loss": 0.5927,
      "step": 53480
    },
    {
      "epoch": 21.9,
      "grad_norm": 3.7741410732269287,
      "learning_rate": 4.642312571336258e-06,
      "loss": 0.602,
      "step": 53490
    },
    {
      "epoch": 21.9,
      "grad_norm": 2.8295419216156006,
      "learning_rate": 4.64228156699134e-06,
      "loss": 0.595,
      "step": 53500
    },
    {
      "epoch": 21.9,
      "grad_norm": 2.427277088165283,
      "learning_rate": 4.642250552117206e-06,
      "loss": 0.6017,
      "step": 53510
    },
    {
      "epoch": 21.91,
      "grad_norm": 2.534773349761963,
      "learning_rate": 4.642219526713997e-06,
      "loss": 0.6162,
      "step": 53520
    },
    {
      "epoch": 21.91,
      "grad_norm": 2.380366086959839,
      "learning_rate": 4.6421884907818535e-06,
      "loss": 0.6185,
      "step": 53530
    },
    {
      "epoch": 21.92,
      "grad_norm": 2.7496471405029297,
      "learning_rate": 4.642157444320921e-06,
      "loss": 0.6099,
      "step": 53540
    },
    {
      "epoch": 21.92,
      "grad_norm": 2.1391496658325195,
      "learning_rate": 4.642126387331339e-06,
      "loss": 0.5893,
      "step": 53550
    },
    {
      "epoch": 21.92,
      "grad_norm": 3.308389663696289,
      "learning_rate": 4.642095319813251e-06,
      "loss": 0.6055,
      "step": 53560
    },
    {
      "epoch": 21.93,
      "grad_norm": 2.659263849258423,
      "learning_rate": 4.642064241766799e-06,
      "loss": 0.6011,
      "step": 53570
    },
    {
      "epoch": 21.93,
      "grad_norm": 2.7604782581329346,
      "learning_rate": 4.642033153192125e-06,
      "loss": 0.6147,
      "step": 53580
    },
    {
      "epoch": 21.94,
      "grad_norm": 2.4266114234924316,
      "learning_rate": 4.642002054089373e-06,
      "loss": 0.5896,
      "step": 53590
    },
    {
      "epoch": 21.94,
      "grad_norm": 2.971859931945801,
      "learning_rate": 4.641970944458684e-06,
      "loss": 0.6032,
      "step": 53600
    },
    {
      "epoch": 21.94,
      "grad_norm": 2.429929733276367,
      "learning_rate": 4.6419398243001995e-06,
      "loss": 0.5976,
      "step": 53610
    },
    {
      "epoch": 21.95,
      "grad_norm": 3.3133466243743896,
      "learning_rate": 4.641908693614064e-06,
      "loss": 0.6072,
      "step": 53620
    },
    {
      "epoch": 21.95,
      "grad_norm": 3.268378734588623,
      "learning_rate": 4.64187755240042e-06,
      "loss": 0.615,
      "step": 53630
    },
    {
      "epoch": 21.96,
      "grad_norm": 2.5340938568115234,
      "learning_rate": 4.641846400659409e-06,
      "loss": 0.6,
      "step": 53640
    },
    {
      "epoch": 21.96,
      "grad_norm": 3.9057223796844482,
      "learning_rate": 4.641815238391174e-06,
      "loss": 0.6005,
      "step": 53650
    },
    {
      "epoch": 21.96,
      "grad_norm": 2.5399587154388428,
      "learning_rate": 4.641784065595858e-06,
      "loss": 0.5935,
      "step": 53660
    },
    {
      "epoch": 21.97,
      "grad_norm": 2.6769587993621826,
      "learning_rate": 4.6417528822736045e-06,
      "loss": 0.641,
      "step": 53670
    },
    {
      "epoch": 21.97,
      "grad_norm": 2.5432510375976562,
      "learning_rate": 4.641721688424556e-06,
      "loss": 0.6144,
      "step": 53680
    },
    {
      "epoch": 21.98,
      "grad_norm": 3.043605327606201,
      "learning_rate": 4.641690484048854e-06,
      "loss": 0.6041,
      "step": 53690
    },
    {
      "epoch": 21.98,
      "grad_norm": 2.418303966522217,
      "learning_rate": 4.641659269146643e-06,
      "loss": 0.5986,
      "step": 53700
    },
    {
      "epoch": 21.99,
      "grad_norm": 1.7483402490615845,
      "learning_rate": 4.641628043718066e-06,
      "loss": 0.6076,
      "step": 53710
    },
    {
      "epoch": 21.99,
      "grad_norm": 1.816882610321045,
      "learning_rate": 4.641596807763265e-06,
      "loss": 0.6045,
      "step": 53720
    },
    {
      "epoch": 21.99,
      "grad_norm": 2.6148853302001953,
      "learning_rate": 4.6415655612823835e-06,
      "loss": 0.5945,
      "step": 53730
    },
    {
      "epoch": 22.0,
      "grad_norm": 3.2719802856445312,
      "learning_rate": 4.641534304275565e-06,
      "loss": 0.613,
      "step": 53740
    },
    {
      "epoch": 22.0,
      "eval_loss": 0.5989567041397095,
      "eval_runtime": 51.9086,
      "eval_samples_per_second": 66.444,
      "eval_steps_per_second": 8.322,
      "step": 53746
    },
    {
      "epoch": 22.0,
      "grad_norm": 1.9633796215057373,
      "learning_rate": 4.641503036742953e-06,
      "loss": 0.5823,
      "step": 53750
    },
    {
      "epoch": 22.01,
      "grad_norm": 2.2457547187805176,
      "learning_rate": 4.6414717586846895e-06,
      "loss": 0.58,
      "step": 53760
    },
    {
      "epoch": 22.01,
      "grad_norm": 2.422656536102295,
      "learning_rate": 4.641440470100919e-06,
      "loss": 0.5768,
      "step": 53770
    },
    {
      "epoch": 22.01,
      "grad_norm": 2.6036124229431152,
      "learning_rate": 4.641409170991784e-06,
      "loss": 0.5994,
      "step": 53780
    },
    {
      "epoch": 22.02,
      "grad_norm": 3.2937815189361572,
      "learning_rate": 4.641377861357428e-06,
      "loss": 0.5864,
      "step": 53790
    },
    {
      "epoch": 22.02,
      "grad_norm": 2.0996813774108887,
      "learning_rate": 4.641346541197995e-06,
      "loss": 0.5846,
      "step": 53800
    },
    {
      "epoch": 22.03,
      "grad_norm": 2.5025291442871094,
      "learning_rate": 4.641315210513628e-06,
      "loss": 0.5696,
      "step": 53810
    },
    {
      "epoch": 22.03,
      "grad_norm": 2.330432653427124,
      "learning_rate": 4.64128386930447e-06,
      "loss": 0.6079,
      "step": 53820
    },
    {
      "epoch": 22.03,
      "grad_norm": 1.7285447120666504,
      "learning_rate": 4.641252517570665e-06,
      "loss": 0.6105,
      "step": 53830
    },
    {
      "epoch": 22.04,
      "grad_norm": 2.412426471710205,
      "learning_rate": 4.641221155312358e-06,
      "loss": 0.5983,
      "step": 53840
    },
    {
      "epoch": 22.04,
      "grad_norm": 2.3169009685516357,
      "learning_rate": 4.6411897825296905e-06,
      "loss": 0.5774,
      "step": 53850
    },
    {
      "epoch": 22.05,
      "grad_norm": 2.1074652671813965,
      "learning_rate": 4.641158399222808e-06,
      "loss": 0.5877,
      "step": 53860
    },
    {
      "epoch": 22.05,
      "grad_norm": 1.9815763235092163,
      "learning_rate": 4.641127005391853e-06,
      "loss": 0.6207,
      "step": 53870
    },
    {
      "epoch": 22.05,
      "grad_norm": 2.0954749584198,
      "learning_rate": 4.641095601036969e-06,
      "loss": 0.6099,
      "step": 53880
    },
    {
      "epoch": 22.06,
      "grad_norm": 3.412156343460083,
      "learning_rate": 4.641064186158301e-06,
      "loss": 0.6057,
      "step": 53890
    },
    {
      "epoch": 22.06,
      "grad_norm": 3.255891799926758,
      "learning_rate": 4.641032760755993e-06,
      "loss": 0.596,
      "step": 53900
    },
    {
      "epoch": 22.07,
      "grad_norm": 2.471822500228882,
      "learning_rate": 4.641001324830188e-06,
      "loss": 0.6003,
      "step": 53910
    },
    {
      "epoch": 22.07,
      "grad_norm": 2.551203727722168,
      "learning_rate": 4.640969878381031e-06,
      "loss": 0.6055,
      "step": 53920
    },
    {
      "epoch": 22.08,
      "grad_norm": 2.411820650100708,
      "learning_rate": 4.640938421408665e-06,
      "loss": 0.6043,
      "step": 53930
    },
    {
      "epoch": 22.08,
      "grad_norm": 2.6811647415161133,
      "learning_rate": 4.6409069539132345e-06,
      "loss": 0.5836,
      "step": 53940
    },
    {
      "epoch": 22.08,
      "grad_norm": 2.820387601852417,
      "learning_rate": 4.640875475894884e-06,
      "loss": 0.5891,
      "step": 53950
    },
    {
      "epoch": 22.09,
      "grad_norm": 2.8476526737213135,
      "learning_rate": 4.6408439873537575e-06,
      "loss": 0.6019,
      "step": 53960
    },
    {
      "epoch": 22.09,
      "grad_norm": 3.0511865615844727,
      "learning_rate": 4.640812488289999e-06,
      "loss": 0.6095,
      "step": 53970
    },
    {
      "epoch": 22.1,
      "grad_norm": 2.2733571529388428,
      "learning_rate": 4.640780978703753e-06,
      "loss": 0.6225,
      "step": 53980
    },
    {
      "epoch": 22.1,
      "grad_norm": 3.173189878463745,
      "learning_rate": 4.640749458595163e-06,
      "loss": 0.5881,
      "step": 53990
    },
    {
      "epoch": 22.1,
      "grad_norm": 2.1834230422973633,
      "learning_rate": 4.6407179279643765e-06,
      "loss": 0.602,
      "step": 54000
    },
    {
      "epoch": 22.11,
      "grad_norm": 3.5207271575927734,
      "learning_rate": 4.6406863868115345e-06,
      "loss": 0.6289,
      "step": 54010
    },
    {
      "epoch": 22.11,
      "grad_norm": 3.046060800552368,
      "learning_rate": 4.640654835136782e-06,
      "loss": 0.6187,
      "step": 54020
    },
    {
      "epoch": 22.12,
      "grad_norm": 3.144700527191162,
      "learning_rate": 4.640623272940265e-06,
      "loss": 0.5963,
      "step": 54030
    },
    {
      "epoch": 22.12,
      "grad_norm": 2.9763362407684326,
      "learning_rate": 4.640591700222126e-06,
      "loss": 0.6003,
      "step": 54040
    },
    {
      "epoch": 22.12,
      "grad_norm": 2.4183638095855713,
      "learning_rate": 4.640560116982512e-06,
      "loss": 0.5979,
      "step": 54050
    },
    {
      "epoch": 22.13,
      "grad_norm": 3.7444143295288086,
      "learning_rate": 4.640528523221567e-06,
      "loss": 0.6025,
      "step": 54060
    },
    {
      "epoch": 22.13,
      "grad_norm": 2.4758682250976562,
      "learning_rate": 4.640496918939434e-06,
      "loss": 0.5973,
      "step": 54070
    },
    {
      "epoch": 22.14,
      "grad_norm": 2.89774489402771,
      "learning_rate": 4.64046530413626e-06,
      "loss": 0.5899,
      "step": 54080
    },
    {
      "epoch": 22.14,
      "grad_norm": 2.707514762878418,
      "learning_rate": 4.640433678812189e-06,
      "loss": 0.6099,
      "step": 54090
    },
    {
      "epoch": 22.14,
      "grad_norm": 2.2383387088775635,
      "learning_rate": 4.640402042967364e-06,
      "loss": 0.6079,
      "step": 54100
    },
    {
      "epoch": 22.15,
      "grad_norm": 2.208153486251831,
      "learning_rate": 4.640370396601933e-06,
      "loss": 0.616,
      "step": 54110
    },
    {
      "epoch": 22.15,
      "grad_norm": 2.561408758163452,
      "learning_rate": 4.6403387397160405e-06,
      "loss": 0.6056,
      "step": 54120
    },
    {
      "epoch": 22.16,
      "grad_norm": 2.961745262145996,
      "learning_rate": 4.640307072309829e-06,
      "loss": 0.6077,
      "step": 54130
    },
    {
      "epoch": 22.16,
      "grad_norm": 2.469843864440918,
      "learning_rate": 4.6402753943834465e-06,
      "loss": 0.6222,
      "step": 54140
    },
    {
      "epoch": 22.17,
      "grad_norm": 3.2640745639801025,
      "learning_rate": 4.640243705937036e-06,
      "loss": 0.5764,
      "step": 54150
    },
    {
      "epoch": 22.17,
      "grad_norm": 2.636615753173828,
      "learning_rate": 4.6402120069707444e-06,
      "loss": 0.5812,
      "step": 54160
    },
    {
      "epoch": 22.17,
      "grad_norm": 2.03910231590271,
      "learning_rate": 4.6401802974847155e-06,
      "loss": 0.5878,
      "step": 54170
    },
    {
      "epoch": 22.18,
      "grad_norm": 2.3816466331481934,
      "learning_rate": 4.640148577479095e-06,
      "loss": 0.5719,
      "step": 54180
    },
    {
      "epoch": 22.18,
      "grad_norm": 2.1869001388549805,
      "learning_rate": 4.640116846954028e-06,
      "loss": 0.5926,
      "step": 54190
    },
    {
      "epoch": 22.19,
      "grad_norm": 2.4445979595184326,
      "learning_rate": 4.6400851059096595e-06,
      "loss": 0.6086,
      "step": 54200
    },
    {
      "epoch": 22.19,
      "grad_norm": 2.4462027549743652,
      "learning_rate": 4.6400533543461365e-06,
      "loss": 0.6097,
      "step": 54210
    },
    {
      "epoch": 22.19,
      "grad_norm": 3.580864906311035,
      "learning_rate": 4.6400215922636034e-06,
      "loss": 0.5879,
      "step": 54220
    },
    {
      "epoch": 22.2,
      "grad_norm": 3.9360013008117676,
      "learning_rate": 4.639989819662206e-06,
      "loss": 0.6097,
      "step": 54230
    },
    {
      "epoch": 22.2,
      "grad_norm": 2.6469197273254395,
      "learning_rate": 4.639958036542089e-06,
      "loss": 0.6025,
      "step": 54240
    },
    {
      "epoch": 22.21,
      "grad_norm": 2.666199207305908,
      "learning_rate": 4.639926242903398e-06,
      "loss": 0.5952,
      "step": 54250
    },
    {
      "epoch": 22.21,
      "grad_norm": 2.478048324584961,
      "learning_rate": 4.6398944387462815e-06,
      "loss": 0.6207,
      "step": 54260
    },
    {
      "epoch": 22.21,
      "grad_norm": 2.6996915340423584,
      "learning_rate": 4.639862624070881e-06,
      "loss": 0.6041,
      "step": 54270
    },
    {
      "epoch": 22.22,
      "grad_norm": 2.419617176055908,
      "learning_rate": 4.639830798877344e-06,
      "loss": 0.5828,
      "step": 54280
    },
    {
      "epoch": 22.22,
      "grad_norm": 2.885646104812622,
      "learning_rate": 4.6397989631658175e-06,
      "loss": 0.5988,
      "step": 54290
    },
    {
      "epoch": 22.23,
      "grad_norm": 2.7448503971099854,
      "learning_rate": 4.639767116936446e-06,
      "loss": 0.5997,
      "step": 54300
    },
    {
      "epoch": 22.23,
      "grad_norm": 1.8662134408950806,
      "learning_rate": 4.639735260189376e-06,
      "loss": 0.5832,
      "step": 54310
    },
    {
      "epoch": 22.23,
      "grad_norm": 2.6233880519866943,
      "learning_rate": 4.639703392924753e-06,
      "loss": 0.6004,
      "step": 54320
    },
    {
      "epoch": 22.24,
      "grad_norm": 2.551647424697876,
      "learning_rate": 4.6396715151427224e-06,
      "loss": 0.5915,
      "step": 54330
    },
    {
      "epoch": 22.24,
      "grad_norm": 3.397712230682373,
      "learning_rate": 4.639639626843431e-06,
      "loss": 0.5825,
      "step": 54340
    },
    {
      "epoch": 22.25,
      "grad_norm": 2.760202407836914,
      "learning_rate": 4.639607728027025e-06,
      "loss": 0.6107,
      "step": 54350
    },
    {
      "epoch": 22.25,
      "grad_norm": 2.4419596195220947,
      "learning_rate": 4.639575818693651e-06,
      "loss": 0.5992,
      "step": 54360
    },
    {
      "epoch": 22.26,
      "grad_norm": 3.0768680572509766,
      "learning_rate": 4.639543898843453e-06,
      "loss": 0.6025,
      "step": 54370
    },
    {
      "epoch": 22.26,
      "grad_norm": 2.7217302322387695,
      "learning_rate": 4.63951196847658e-06,
      "loss": 0.6225,
      "step": 54380
    },
    {
      "epoch": 22.26,
      "grad_norm": 2.407111883163452,
      "learning_rate": 4.639480027593176e-06,
      "loss": 0.5921,
      "step": 54390
    },
    {
      "epoch": 22.27,
      "grad_norm": 2.839176654815674,
      "learning_rate": 4.639448076193389e-06,
      "loss": 0.6158,
      "step": 54400
    },
    {
      "epoch": 22.27,
      "grad_norm": 3.3147943019866943,
      "learning_rate": 4.639416114277365e-06,
      "loss": 0.6055,
      "step": 54410
    },
    {
      "epoch": 22.28,
      "grad_norm": 2.616945266723633,
      "learning_rate": 4.639384141845249e-06,
      "loss": 0.6027,
      "step": 54420
    },
    {
      "epoch": 22.28,
      "grad_norm": 2.2600531578063965,
      "learning_rate": 4.6393521588971895e-06,
      "loss": 0.59,
      "step": 54430
    },
    {
      "epoch": 22.28,
      "grad_norm": 2.991131067276001,
      "learning_rate": 4.639320165433331e-06,
      "loss": 0.6048,
      "step": 54440
    },
    {
      "epoch": 22.29,
      "grad_norm": 3.068213701248169,
      "learning_rate": 4.639288161453822e-06,
      "loss": 0.6199,
      "step": 54450
    },
    {
      "epoch": 22.29,
      "grad_norm": 2.7918126583099365,
      "learning_rate": 4.639256146958807e-06,
      "loss": 0.6187,
      "step": 54460
    },
    {
      "epoch": 22.3,
      "grad_norm": 3.5437965393066406,
      "learning_rate": 4.639224121948436e-06,
      "loss": 0.6015,
      "step": 54470
    },
    {
      "epoch": 22.3,
      "grad_norm": 2.6159071922302246,
      "learning_rate": 4.639192086422851e-06,
      "loss": 0.6017,
      "step": 54480
    },
    {
      "epoch": 22.3,
      "grad_norm": 3.0845730304718018,
      "learning_rate": 4.639160040382204e-06,
      "loss": 0.5908,
      "step": 54490
    },
    {
      "epoch": 22.31,
      "grad_norm": 2.1779515743255615,
      "learning_rate": 4.639127983826637e-06,
      "loss": 0.5939,
      "step": 54500
    },
    {
      "epoch": 22.31,
      "grad_norm": 2.536607265472412,
      "learning_rate": 4.639095916756299e-06,
      "loss": 0.599,
      "step": 54510
    },
    {
      "epoch": 22.32,
      "grad_norm": 2.395620822906494,
      "learning_rate": 4.639063839171338e-06,
      "loss": 0.6001,
      "step": 54520
    },
    {
      "epoch": 22.32,
      "grad_norm": 3.266791582107544,
      "learning_rate": 4.639031751071899e-06,
      "loss": 0.5984,
      "step": 54530
    },
    {
      "epoch": 22.33,
      "grad_norm": 2.720930576324463,
      "learning_rate": 4.6389996524581305e-06,
      "loss": 0.6156,
      "step": 54540
    },
    {
      "epoch": 22.33,
      "grad_norm": 2.1629881858825684,
      "learning_rate": 4.638967543330178e-06,
      "loss": 0.5977,
      "step": 54550
    },
    {
      "epoch": 22.33,
      "grad_norm": 2.459108352661133,
      "learning_rate": 4.638935423688189e-06,
      "loss": 0.5839,
      "step": 54560
    },
    {
      "epoch": 22.34,
      "grad_norm": 2.931753158569336,
      "learning_rate": 4.638903293532312e-06,
      "loss": 0.5989,
      "step": 54570
    },
    {
      "epoch": 22.34,
      "grad_norm": 3.8581204414367676,
      "learning_rate": 4.638871152862693e-06,
      "loss": 0.5996,
      "step": 54580
    },
    {
      "epoch": 22.35,
      "grad_norm": 2.7681710720062256,
      "learning_rate": 4.6388390016794795e-06,
      "loss": 0.5854,
      "step": 54590
    },
    {
      "epoch": 22.35,
      "grad_norm": 3.149510145187378,
      "learning_rate": 4.6388068399828186e-06,
      "loss": 0.6029,
      "step": 54600
    },
    {
      "epoch": 22.35,
      "grad_norm": 2.360271453857422,
      "learning_rate": 4.638774667772857e-06,
      "loss": 0.6038,
      "step": 54610
    },
    {
      "epoch": 22.36,
      "grad_norm": 2.6077146530151367,
      "learning_rate": 4.638742485049744e-06,
      "loss": 0.5994,
      "step": 54620
    },
    {
      "epoch": 22.36,
      "grad_norm": 2.273167848587036,
      "learning_rate": 4.638710291813625e-06,
      "loss": 0.605,
      "step": 54630
    },
    {
      "epoch": 22.37,
      "grad_norm": 3.1189048290252686,
      "learning_rate": 4.638678088064648e-06,
      "loss": 0.6106,
      "step": 54640
    },
    {
      "epoch": 22.37,
      "grad_norm": 2.488642930984497,
      "learning_rate": 4.6386458738029625e-06,
      "loss": 0.5995,
      "step": 54650
    },
    {
      "epoch": 22.37,
      "grad_norm": 3.747843027114868,
      "learning_rate": 4.638613649028713e-06,
      "loss": 0.6085,
      "step": 54660
    },
    {
      "epoch": 22.38,
      "grad_norm": 2.8299765586853027,
      "learning_rate": 4.638581413742049e-06,
      "loss": 0.5986,
      "step": 54670
    },
    {
      "epoch": 22.38,
      "grad_norm": 2.2834081649780273,
      "learning_rate": 4.638549167943118e-06,
      "loss": 0.5898,
      "step": 54680
    },
    {
      "epoch": 22.39,
      "grad_norm": 3.2854528427124023,
      "learning_rate": 4.638516911632067e-06,
      "loss": 0.6148,
      "step": 54690
    },
    {
      "epoch": 22.39,
      "grad_norm": 2.5228495597839355,
      "learning_rate": 4.638484644809044e-06,
      "loss": 0.5819,
      "step": 54700
    },
    {
      "epoch": 22.39,
      "grad_norm": 4.06541633605957,
      "learning_rate": 4.638452367474197e-06,
      "loss": 0.5949,
      "step": 54710
    },
    {
      "epoch": 22.4,
      "grad_norm": 2.8973913192749023,
      "learning_rate": 4.638420079627674e-06,
      "loss": 0.6198,
      "step": 54720
    },
    {
      "epoch": 22.4,
      "grad_norm": 2.1087751388549805,
      "learning_rate": 4.638387781269622e-06,
      "loss": 0.6165,
      "step": 54730
    },
    {
      "epoch": 22.41,
      "grad_norm": 2.0287272930145264,
      "learning_rate": 4.6383554724001915e-06,
      "loss": 0.6143,
      "step": 54740
    },
    {
      "epoch": 22.41,
      "grad_norm": 2.4573376178741455,
      "learning_rate": 4.638323153019527e-06,
      "loss": 0.6126,
      "step": 54750
    },
    {
      "epoch": 22.42,
      "grad_norm": 2.501101016998291,
      "learning_rate": 4.6382908231277796e-06,
      "loss": 0.6037,
      "step": 54760
    },
    {
      "epoch": 22.42,
      "grad_norm": 2.377526044845581,
      "learning_rate": 4.638258482725095e-06,
      "loss": 0.5828,
      "step": 54770
    },
    {
      "epoch": 22.42,
      "grad_norm": 3.0635035037994385,
      "learning_rate": 4.638226131811623e-06,
      "loss": 0.5899,
      "step": 54780
    },
    {
      "epoch": 22.43,
      "grad_norm": 4.07381010055542,
      "learning_rate": 4.638193770387511e-06,
      "loss": 0.6168,
      "step": 54790
    },
    {
      "epoch": 22.43,
      "grad_norm": 2.6352949142456055,
      "learning_rate": 4.6381613984529075e-06,
      "loss": 0.6111,
      "step": 54800
    },
    {
      "epoch": 22.44,
      "grad_norm": 3.8315725326538086,
      "learning_rate": 4.63812901600796e-06,
      "loss": 0.6117,
      "step": 54810
    },
    {
      "epoch": 22.44,
      "grad_norm": 3.290053367614746,
      "learning_rate": 4.638096623052818e-06,
      "loss": 0.5943,
      "step": 54820
    },
    {
      "epoch": 22.44,
      "grad_norm": 3.701139211654663,
      "learning_rate": 4.63806421958763e-06,
      "loss": 0.6103,
      "step": 54830
    },
    {
      "epoch": 22.45,
      "grad_norm": 2.2870023250579834,
      "learning_rate": 4.638031805612544e-06,
      "loss": 0.6454,
      "step": 54840
    },
    {
      "epoch": 22.45,
      "grad_norm": 2.44997501373291,
      "learning_rate": 4.637999381127707e-06,
      "loss": 0.5716,
      "step": 54850
    },
    {
      "epoch": 22.46,
      "grad_norm": 2.1737446784973145,
      "learning_rate": 4.63796694613327e-06,
      "loss": 0.6161,
      "step": 54860
    },
    {
      "epoch": 22.46,
      "grad_norm": 3.1762735843658447,
      "learning_rate": 4.63793450062938e-06,
      "loss": 0.5884,
      "step": 54870
    },
    {
      "epoch": 22.46,
      "grad_norm": 2.2184622287750244,
      "learning_rate": 4.637902044616186e-06,
      "loss": 0.5971,
      "step": 54880
    },
    {
      "epoch": 22.47,
      "grad_norm": 2.176689624786377,
      "learning_rate": 4.637869578093837e-06,
      "loss": 0.6016,
      "step": 54890
    },
    {
      "epoch": 22.47,
      "grad_norm": 1.863013744354248,
      "learning_rate": 4.637837101062482e-06,
      "loss": 0.5837,
      "step": 54900
    },
    {
      "epoch": 22.48,
      "grad_norm": 2.4114768505096436,
      "learning_rate": 4.637804613522268e-06,
      "loss": 0.5979,
      "step": 54910
    },
    {
      "epoch": 22.48,
      "grad_norm": 2.842369318008423,
      "learning_rate": 4.637772115473347e-06,
      "loss": 0.627,
      "step": 54920
    },
    {
      "epoch": 22.48,
      "grad_norm": 2.7876710891723633,
      "learning_rate": 4.637739606915864e-06,
      "loss": 0.5862,
      "step": 54930
    },
    {
      "epoch": 22.49,
      "grad_norm": 2.359374761581421,
      "learning_rate": 4.637707087849971e-06,
      "loss": 0.5983,
      "step": 54940
    },
    {
      "epoch": 22.49,
      "grad_norm": 2.1884677410125732,
      "learning_rate": 4.637674558275816e-06,
      "loss": 0.6133,
      "step": 54950
    },
    {
      "epoch": 22.5,
      "grad_norm": 2.4476301670074463,
      "learning_rate": 4.637642018193548e-06,
      "loss": 0.5878,
      "step": 54960
    },
    {
      "epoch": 22.5,
      "grad_norm": 1.6831071376800537,
      "learning_rate": 4.637609467603315e-06,
      "loss": 0.5919,
      "step": 54970
    },
    {
      "epoch": 22.51,
      "grad_norm": 2.9227449893951416,
      "learning_rate": 4.637576906505267e-06,
      "loss": 0.596,
      "step": 54980
    },
    {
      "epoch": 22.51,
      "grad_norm": 2.767852306365967,
      "learning_rate": 4.637544334899554e-06,
      "loss": 0.5891,
      "step": 54990
    },
    {
      "epoch": 22.51,
      "grad_norm": 2.6544950008392334,
      "learning_rate": 4.637511752786324e-06,
      "loss": 0.5973,
      "step": 55000
    },
    {
      "epoch": 22.52,
      "grad_norm": 2.2871034145355225,
      "learning_rate": 4.637479160165727e-06,
      "loss": 0.606,
      "step": 55010
    },
    {
      "epoch": 22.52,
      "grad_norm": 3.1073644161224365,
      "learning_rate": 4.6374465570379116e-06,
      "loss": 0.5986,
      "step": 55020
    },
    {
      "epoch": 22.53,
      "grad_norm": 3.18865704536438,
      "learning_rate": 4.637413943403028e-06,
      "loss": 0.6225,
      "step": 55030
    },
    {
      "epoch": 22.53,
      "grad_norm": 2.890333414077759,
      "learning_rate": 4.637381319261224e-06,
      "loss": 0.614,
      "step": 55040
    },
    {
      "epoch": 22.53,
      "grad_norm": 2.791966438293457,
      "learning_rate": 4.637348684612651e-06,
      "loss": 0.5948,
      "step": 55050
    },
    {
      "epoch": 22.54,
      "grad_norm": 2.5858609676361084,
      "learning_rate": 4.637316039457457e-06,
      "loss": 0.6041,
      "step": 55060
    },
    {
      "epoch": 22.54,
      "grad_norm": 3.061190128326416,
      "learning_rate": 4.6372833837957926e-06,
      "loss": 0.5877,
      "step": 55070
    },
    {
      "epoch": 22.55,
      "grad_norm": 2.9639296531677246,
      "learning_rate": 4.637250717627807e-06,
      "loss": 0.6236,
      "step": 55080
    },
    {
      "epoch": 22.55,
      "grad_norm": 1.8546390533447266,
      "learning_rate": 4.6372180409536495e-06,
      "loss": 0.5829,
      "step": 55090
    },
    {
      "epoch": 22.55,
      "grad_norm": 2.5872795581817627,
      "learning_rate": 4.63718535377347e-06,
      "loss": 0.5789,
      "step": 55100
    },
    {
      "epoch": 22.56,
      "grad_norm": 2.8991947174072266,
      "learning_rate": 4.6371526560874185e-06,
      "loss": 0.6149,
      "step": 55110
    },
    {
      "epoch": 22.56,
      "grad_norm": 2.583904504776001,
      "learning_rate": 4.637119947895645e-06,
      "loss": 0.6037,
      "step": 55120
    },
    {
      "epoch": 22.57,
      "grad_norm": 3.180152654647827,
      "learning_rate": 4.637087229198298e-06,
      "loss": 0.5833,
      "step": 55130
    },
    {
      "epoch": 22.57,
      "grad_norm": 2.6908488273620605,
      "learning_rate": 4.63705449999553e-06,
      "loss": 0.6205,
      "step": 55140
    },
    {
      "epoch": 22.57,
      "grad_norm": 2.60044264793396,
      "learning_rate": 4.637021760287488e-06,
      "loss": 0.5946,
      "step": 55150
    },
    {
      "epoch": 22.58,
      "grad_norm": 2.176126480102539,
      "learning_rate": 4.636989010074323e-06,
      "loss": 0.5977,
      "step": 55160
    },
    {
      "epoch": 22.58,
      "grad_norm": 3.074875831604004,
      "learning_rate": 4.6369562493561865e-06,
      "loss": 0.5885,
      "step": 55170
    },
    {
      "epoch": 22.59,
      "grad_norm": 2.0304205417633057,
      "learning_rate": 4.636923478133227e-06,
      "loss": 0.6007,
      "step": 55180
    },
    {
      "epoch": 22.59,
      "grad_norm": 3.804194927215576,
      "learning_rate": 4.636890696405594e-06,
      "loss": 0.5859,
      "step": 55190
    },
    {
      "epoch": 22.6,
      "grad_norm": 3.355100631713867,
      "learning_rate": 4.636857904173439e-06,
      "loss": 0.5631,
      "step": 55200
    },
    {
      "epoch": 22.6,
      "grad_norm": 2.912776470184326,
      "learning_rate": 4.636825101436912e-06,
      "loss": 0.5789,
      "step": 55210
    },
    {
      "epoch": 22.6,
      "grad_norm": 2.806786060333252,
      "learning_rate": 4.636792288196163e-06,
      "loss": 0.6218,
      "step": 55220
    },
    {
      "epoch": 22.61,
      "grad_norm": 2.497161626815796,
      "learning_rate": 4.636759464451342e-06,
      "loss": 0.5935,
      "step": 55230
    },
    {
      "epoch": 22.61,
      "grad_norm": 2.7189252376556396,
      "learning_rate": 4.6367266302026e-06,
      "loss": 0.6163,
      "step": 55240
    },
    {
      "epoch": 22.62,
      "grad_norm": 3.7460787296295166,
      "learning_rate": 4.636693785450087e-06,
      "loss": 0.5975,
      "step": 55250
    },
    {
      "epoch": 22.62,
      "grad_norm": 3.4556660652160645,
      "learning_rate": 4.636660930193955e-06,
      "loss": 0.5884,
      "step": 55260
    },
    {
      "epoch": 22.62,
      "grad_norm": 3.8183698654174805,
      "learning_rate": 4.636628064434352e-06,
      "loss": 0.5913,
      "step": 55270
    },
    {
      "epoch": 22.63,
      "grad_norm": 2.8091211318969727,
      "learning_rate": 4.636595188171429e-06,
      "loss": 0.5864,
      "step": 55280
    },
    {
      "epoch": 22.63,
      "grad_norm": 2.3158159255981445,
      "learning_rate": 4.6365623014053384e-06,
      "loss": 0.5828,
      "step": 55290
    },
    {
      "epoch": 22.64,
      "grad_norm": 2.5710084438323975,
      "learning_rate": 4.63652940413623e-06,
      "loss": 0.5918,
      "step": 55300
    },
    {
      "epoch": 22.64,
      "grad_norm": 2.520341157913208,
      "learning_rate": 4.636496496364253e-06,
      "loss": 0.6109,
      "step": 55310
    },
    {
      "epoch": 22.64,
      "grad_norm": 2.3735620975494385,
      "learning_rate": 4.6364635780895596e-06,
      "loss": 0.6012,
      "step": 55320
    },
    {
      "epoch": 22.65,
      "grad_norm": 2.4130659103393555,
      "learning_rate": 4.636430649312301e-06,
      "loss": 0.6015,
      "step": 55330
    },
    {
      "epoch": 22.65,
      "grad_norm": 2.8036038875579834,
      "learning_rate": 4.6363977100326274e-06,
      "loss": 0.6127,
      "step": 55340
    },
    {
      "epoch": 22.66,
      "grad_norm": 3.639603853225708,
      "learning_rate": 4.636364760250689e-06,
      "loss": 0.5932,
      "step": 55350
    },
    {
      "epoch": 22.66,
      "grad_norm": 3.162799596786499,
      "learning_rate": 4.636331799966638e-06,
      "loss": 0.6125,
      "step": 55360
    },
    {
      "epoch": 22.66,
      "grad_norm": 1.9487721920013428,
      "learning_rate": 4.6362988291806246e-06,
      "loss": 0.5956,
      "step": 55370
    },
    {
      "epoch": 22.67,
      "grad_norm": 2.4367873668670654,
      "learning_rate": 4.636265847892801e-06,
      "loss": 0.6177,
      "step": 55380
    },
    {
      "epoch": 22.67,
      "grad_norm": 3.0966928005218506,
      "learning_rate": 4.636232856103316e-06,
      "loss": 0.5965,
      "step": 55390
    },
    {
      "epoch": 22.68,
      "grad_norm": 2.257220506668091,
      "learning_rate": 4.636199853812322e-06,
      "loss": 0.6284,
      "step": 55400
    },
    {
      "epoch": 22.68,
      "grad_norm": 2.4739232063293457,
      "learning_rate": 4.636166841019972e-06,
      "loss": 0.6007,
      "step": 55410
    },
    {
      "epoch": 22.69,
      "grad_norm": 2.646998643875122,
      "learning_rate": 4.6361338177264135e-06,
      "loss": 0.5968,
      "step": 55420
    },
    {
      "epoch": 22.69,
      "grad_norm": 2.6777987480163574,
      "learning_rate": 4.636100783931801e-06,
      "loss": 0.6031,
      "step": 55430
    },
    {
      "epoch": 22.69,
      "grad_norm": 2.998152256011963,
      "learning_rate": 4.636067739636284e-06,
      "loss": 0.5886,
      "step": 55440
    },
    {
      "epoch": 22.7,
      "grad_norm": 2.7068371772766113,
      "learning_rate": 4.6360346848400155e-06,
      "loss": 0.6109,
      "step": 55450
    },
    {
      "epoch": 22.7,
      "grad_norm": 2.7749252319335938,
      "learning_rate": 4.636001619543144e-06,
      "loss": 0.589,
      "step": 55460
    },
    {
      "epoch": 22.71,
      "grad_norm": 2.2509944438934326,
      "learning_rate": 4.6359685437458245e-06,
      "loss": 0.5821,
      "step": 55470
    },
    {
      "epoch": 22.71,
      "grad_norm": 2.3507235050201416,
      "learning_rate": 4.635935457448206e-06,
      "loss": 0.5909,
      "step": 55480
    },
    {
      "epoch": 22.71,
      "grad_norm": 3.2917213439941406,
      "learning_rate": 4.635902360650441e-06,
      "loss": 0.5906,
      "step": 55490
    },
    {
      "epoch": 22.72,
      "grad_norm": 3.2101223468780518,
      "learning_rate": 4.635869253352681e-06,
      "loss": 0.5995,
      "step": 55500
    },
    {
      "epoch": 22.72,
      "grad_norm": 2.7437868118286133,
      "learning_rate": 4.635836135555078e-06,
      "loss": 0.6068,
      "step": 55510
    },
    {
      "epoch": 22.73,
      "grad_norm": 2.9561004638671875,
      "learning_rate": 4.635803007257783e-06,
      "loss": 0.6037,
      "step": 55520
    },
    {
      "epoch": 22.73,
      "grad_norm": 2.330427646636963,
      "learning_rate": 4.635769868460948e-06,
      "loss": 0.5773,
      "step": 55530
    },
    {
      "epoch": 22.73,
      "grad_norm": 1.8991550207138062,
      "learning_rate": 4.635736719164725e-06,
      "loss": 0.6005,
      "step": 55540
    },
    {
      "epoch": 22.74,
      "grad_norm": 2.2861053943634033,
      "learning_rate": 4.635703559369266e-06,
      "loss": 0.5884,
      "step": 55550
    },
    {
      "epoch": 22.74,
      "grad_norm": 2.992968797683716,
      "learning_rate": 4.635670389074723e-06,
      "loss": 0.6093,
      "step": 55560
    },
    {
      "epoch": 22.75,
      "grad_norm": 2.43890380859375,
      "learning_rate": 4.6356372082812476e-06,
      "loss": 0.5853,
      "step": 55570
    },
    {
      "epoch": 22.75,
      "grad_norm": 2.3498916625976562,
      "learning_rate": 4.635604016988992e-06,
      "loss": 0.6105,
      "step": 55580
    },
    {
      "epoch": 22.75,
      "grad_norm": 1.6222425699234009,
      "learning_rate": 4.635570815198107e-06,
      "loss": 0.6317,
      "step": 55590
    },
    {
      "epoch": 22.76,
      "grad_norm": 2.2805075645446777,
      "learning_rate": 4.635537602908747e-06,
      "loss": 0.5847,
      "step": 55600
    },
    {
      "epoch": 22.76,
      "grad_norm": 4.726080894470215,
      "learning_rate": 4.635504380121062e-06,
      "loss": 0.5948,
      "step": 55610
    },
    {
      "epoch": 22.77,
      "grad_norm": 3.182752847671509,
      "learning_rate": 4.635471146835205e-06,
      "loss": 0.6051,
      "step": 55620
    },
    {
      "epoch": 22.77,
      "grad_norm": 2.6251182556152344,
      "learning_rate": 4.63543790305133e-06,
      "loss": 0.5902,
      "step": 55630
    },
    {
      "epoch": 22.78,
      "grad_norm": 2.3756606578826904,
      "learning_rate": 4.635404648769586e-06,
      "loss": 0.5785,
      "step": 55640
    },
    {
      "epoch": 22.78,
      "grad_norm": 2.484748125076294,
      "learning_rate": 4.635371383990128e-06,
      "loss": 0.5948,
      "step": 55650
    },
    {
      "epoch": 22.78,
      "grad_norm": 3.0538597106933594,
      "learning_rate": 4.635338108713106e-06,
      "loss": 0.6001,
      "step": 55660
    },
    {
      "epoch": 22.79,
      "grad_norm": 2.0708601474761963,
      "learning_rate": 4.6353048229386745e-06,
      "loss": 0.5918,
      "step": 55670
    },
    {
      "epoch": 22.79,
      "grad_norm": 2.219141721725464,
      "learning_rate": 4.635271526666986e-06,
      "loss": 0.6027,
      "step": 55680
    },
    {
      "epoch": 22.8,
      "grad_norm": 2.3673579692840576,
      "learning_rate": 4.635238219898191e-06,
      "loss": 0.6208,
      "step": 55690
    },
    {
      "epoch": 22.8,
      "grad_norm": 2.8401591777801514,
      "learning_rate": 4.635204902632443e-06,
      "loss": 0.5974,
      "step": 55700
    },
    {
      "epoch": 22.8,
      "grad_norm": 3.449596881866455,
      "learning_rate": 4.6351715748698974e-06,
      "loss": 0.6036,
      "step": 55710
    },
    {
      "epoch": 22.81,
      "grad_norm": 2.793727159500122,
      "learning_rate": 4.635138236610702e-06,
      "loss": 0.6184,
      "step": 55720
    },
    {
      "epoch": 22.81,
      "grad_norm": 1.9926471710205078,
      "learning_rate": 4.635104887855013e-06,
      "loss": 0.599,
      "step": 55730
    },
    {
      "epoch": 22.82,
      "grad_norm": 2.305854320526123,
      "learning_rate": 4.635071528602982e-06,
      "loss": 0.6031,
      "step": 55740
    },
    {
      "epoch": 22.82,
      "grad_norm": 2.1037847995758057,
      "learning_rate": 4.635038158854762e-06,
      "loss": 0.6059,
      "step": 55750
    },
    {
      "epoch": 22.82,
      "grad_norm": 2.625999927520752,
      "learning_rate": 4.6350047786105065e-06,
      "loss": 0.5915,
      "step": 55760
    },
    {
      "epoch": 22.83,
      "grad_norm": 3.393871307373047,
      "learning_rate": 4.634971387870367e-06,
      "loss": 0.6025,
      "step": 55770
    },
    {
      "epoch": 22.83,
      "grad_norm": 2.252953290939331,
      "learning_rate": 4.634937986634497e-06,
      "loss": 0.5894,
      "step": 55780
    },
    {
      "epoch": 22.84,
      "grad_norm": 3.8334169387817383,
      "learning_rate": 4.63490457490305e-06,
      "loss": 0.5847,
      "step": 55790
    },
    {
      "epoch": 22.84,
      "grad_norm": 2.6096198558807373,
      "learning_rate": 4.634871152676178e-06,
      "loss": 0.6081,
      "step": 55800
    },
    {
      "epoch": 22.84,
      "grad_norm": 2.4959869384765625,
      "learning_rate": 4.6348377199540354e-06,
      "loss": 0.6282,
      "step": 55810
    },
    {
      "epoch": 22.85,
      "grad_norm": 3.0766637325286865,
      "learning_rate": 4.634804276736775e-06,
      "loss": 0.6264,
      "step": 55820
    },
    {
      "epoch": 22.85,
      "grad_norm": 1.7367624044418335,
      "learning_rate": 4.634770823024549e-06,
      "loss": 0.6033,
      "step": 55830
    },
    {
      "epoch": 22.86,
      "grad_norm": 3.711714744567871,
      "learning_rate": 4.634737358817512e-06,
      "loss": 0.5966,
      "step": 55840
    },
    {
      "epoch": 22.86,
      "grad_norm": 2.7140936851501465,
      "learning_rate": 4.6347038841158165e-06,
      "loss": 0.5907,
      "step": 55850
    },
    {
      "epoch": 22.87,
      "grad_norm": 1.995566725730896,
      "learning_rate": 4.634670398919616e-06,
      "loss": 0.6005,
      "step": 55860
    },
    {
      "epoch": 22.87,
      "grad_norm": 2.5793235301971436,
      "learning_rate": 4.634636903229064e-06,
      "loss": 0.5964,
      "step": 55870
    },
    {
      "epoch": 22.87,
      "grad_norm": 2.205594301223755,
      "learning_rate": 4.634603397044314e-06,
      "loss": 0.6002,
      "step": 55880
    },
    {
      "epoch": 22.88,
      "grad_norm": 3.922738552093506,
      "learning_rate": 4.63456988036552e-06,
      "loss": 0.607,
      "step": 55890
    },
    {
      "epoch": 22.88,
      "grad_norm": 3.773881435394287,
      "learning_rate": 4.634536353192834e-06,
      "loss": 0.612,
      "step": 55900
    },
    {
      "epoch": 22.89,
      "grad_norm": 2.6831610202789307,
      "learning_rate": 4.634502815526411e-06,
      "loss": 0.6106,
      "step": 55910
    },
    {
      "epoch": 22.89,
      "grad_norm": 2.310535430908203,
      "learning_rate": 4.6344692673664034e-06,
      "loss": 0.6049,
      "step": 55920
    },
    {
      "epoch": 22.89,
      "grad_norm": 2.40559720993042,
      "learning_rate": 4.634435708712967e-06,
      "loss": 0.6069,
      "step": 55930
    },
    {
      "epoch": 22.9,
      "grad_norm": 2.086118459701538,
      "learning_rate": 4.634402139566252e-06,
      "loss": 0.5971,
      "step": 55940
    },
    {
      "epoch": 22.9,
      "grad_norm": 2.064824342727661,
      "learning_rate": 4.634368559926416e-06,
      "loss": 0.5819,
      "step": 55950
    },
    {
      "epoch": 22.91,
      "grad_norm": 2.0218417644500732,
      "learning_rate": 4.6343349697936105e-06,
      "loss": 0.616,
      "step": 55960
    },
    {
      "epoch": 22.91,
      "grad_norm": 3.5472474098205566,
      "learning_rate": 4.63430136916799e-06,
      "loss": 0.584,
      "step": 55970
    },
    {
      "epoch": 22.91,
      "grad_norm": 2.7996537685394287,
      "learning_rate": 4.634267758049708e-06,
      "loss": 0.5759,
      "step": 55980
    },
    {
      "epoch": 22.92,
      "grad_norm": 2.3439574241638184,
      "learning_rate": 4.63423413643892e-06,
      "loss": 0.6013,
      "step": 55990
    },
    {
      "epoch": 22.92,
      "grad_norm": 2.1991679668426514,
      "learning_rate": 4.634200504335778e-06,
      "loss": 0.6094,
      "step": 56000
    },
    {
      "epoch": 22.93,
      "grad_norm": 4.16972017288208,
      "learning_rate": 4.634166861740438e-06,
      "loss": 0.6014,
      "step": 56010
    },
    {
      "epoch": 22.93,
      "grad_norm": 2.0465052127838135,
      "learning_rate": 4.634133208653052e-06,
      "loss": 0.609,
      "step": 56020
    },
    {
      "epoch": 22.93,
      "grad_norm": 2.2409627437591553,
      "learning_rate": 4.6340995450737755e-06,
      "loss": 0.6051,
      "step": 56030
    },
    {
      "epoch": 22.94,
      "grad_norm": 2.65704607963562,
      "learning_rate": 4.634065871002763e-06,
      "loss": 0.6173,
      "step": 56040
    },
    {
      "epoch": 22.94,
      "grad_norm": 2.5577917098999023,
      "learning_rate": 4.634032186440167e-06,
      "loss": 0.6014,
      "step": 56050
    },
    {
      "epoch": 22.95,
      "grad_norm": 4.028266906738281,
      "learning_rate": 4.633998491386144e-06,
      "loss": 0.6027,
      "step": 56060
    },
    {
      "epoch": 22.95,
      "grad_norm": 3.5523383617401123,
      "learning_rate": 4.633964785840847e-06,
      "loss": 0.6022,
      "step": 56070
    },
    {
      "epoch": 22.96,
      "grad_norm": 2.4668712615966797,
      "learning_rate": 4.63393106980443e-06,
      "loss": 0.5903,
      "step": 56080
    },
    {
      "epoch": 22.96,
      "grad_norm": 2.5486066341400146,
      "learning_rate": 4.633897343277049e-06,
      "loss": 0.5841,
      "step": 56090
    },
    {
      "epoch": 22.96,
      "grad_norm": 2.781895875930786,
      "learning_rate": 4.6338636062588585e-06,
      "loss": 0.6188,
      "step": 56100
    },
    {
      "epoch": 22.97,
      "grad_norm": 2.5697648525238037,
      "learning_rate": 4.633829858750011e-06,
      "loss": 0.5838,
      "step": 56110
    },
    {
      "epoch": 22.97,
      "grad_norm": 2.742291212081909,
      "learning_rate": 4.633796100750663e-06,
      "loss": 0.5941,
      "step": 56120
    },
    {
      "epoch": 22.98,
      "grad_norm": 2.8742353916168213,
      "learning_rate": 4.633762332260968e-06,
      "loss": 0.5961,
      "step": 56130
    },
    {
      "epoch": 22.98,
      "grad_norm": 3.7567877769470215,
      "learning_rate": 4.633728553281082e-06,
      "loss": 0.5867,
      "step": 56140
    },
    {
      "epoch": 22.98,
      "grad_norm": 3.432849168777466,
      "learning_rate": 4.6336947638111585e-06,
      "loss": 0.6118,
      "step": 56150
    },
    {
      "epoch": 22.99,
      "grad_norm": 2.5422041416168213,
      "learning_rate": 4.633660963851352e-06,
      "loss": 0.601,
      "step": 56160
    },
    {
      "epoch": 22.99,
      "grad_norm": 2.144768714904785,
      "learning_rate": 4.633627153401819e-06,
      "loss": 0.6083,
      "step": 56170
    },
    {
      "epoch": 23.0,
      "grad_norm": 3.054852247238159,
      "learning_rate": 4.633593332462712e-06,
      "loss": 0.5864,
      "step": 56180
    },
    {
      "epoch": 23.0,
      "eval_loss": 0.5947991609573364,
      "eval_runtime": 51.8267,
      "eval_samples_per_second": 66.549,
      "eval_steps_per_second": 8.335,
      "step": 56189
    },
    {
      "epoch": 23.0,
      "grad_norm": 2.3040552139282227,
      "learning_rate": 4.633559501034189e-06,
      "loss": 0.6062,
      "step": 56190
    },
    {
      "epoch": 23.0,
      "grad_norm": 2.4164469242095947,
      "learning_rate": 4.6335256591164024e-06,
      "loss": 0.6001,
      "step": 56200
    },
    {
      "epoch": 23.01,
      "grad_norm": 2.5962212085723877,
      "learning_rate": 4.633491806709508e-06,
      "loss": 0.5764,
      "step": 56210
    },
    {
      "epoch": 23.01,
      "grad_norm": 1.9167416095733643,
      "learning_rate": 4.633457943813661e-06,
      "loss": 0.5979,
      "step": 56220
    },
    {
      "epoch": 23.02,
      "grad_norm": 2.5372233390808105,
      "learning_rate": 4.633424070429017e-06,
      "loss": 0.612,
      "step": 56230
    },
    {
      "epoch": 23.02,
      "grad_norm": 2.601762294769287,
      "learning_rate": 4.633390186555731e-06,
      "loss": 0.5834,
      "step": 56240
    },
    {
      "epoch": 23.02,
      "grad_norm": 2.080484390258789,
      "learning_rate": 4.633356292193957e-06,
      "loss": 0.5793,
      "step": 56250
    },
    {
      "epoch": 23.03,
      "grad_norm": 2.939650058746338,
      "learning_rate": 4.633322387343851e-06,
      "loss": 0.5987,
      "step": 56260
    },
    {
      "epoch": 23.03,
      "grad_norm": 2.9841625690460205,
      "learning_rate": 4.633288472005569e-06,
      "loss": 0.6112,
      "step": 56270
    },
    {
      "epoch": 23.04,
      "grad_norm": 3.254695177078247,
      "learning_rate": 4.6332545461792655e-06,
      "loss": 0.587,
      "step": 56280
    },
    {
      "epoch": 23.04,
      "grad_norm": 2.3993396759033203,
      "learning_rate": 4.633220609865096e-06,
      "loss": 0.6162,
      "step": 56290
    },
    {
      "epoch": 23.05,
      "grad_norm": 2.9624619483947754,
      "learning_rate": 4.633186663063217e-06,
      "loss": 0.5859,
      "step": 56300
    },
    {
      "epoch": 23.05,
      "grad_norm": 2.057281017303467,
      "learning_rate": 4.633152705773784e-06,
      "loss": 0.6097,
      "step": 56310
    },
    {
      "epoch": 23.05,
      "grad_norm": 2.772706985473633,
      "learning_rate": 4.63311873799695e-06,
      "loss": 0.5761,
      "step": 56320
    },
    {
      "epoch": 23.06,
      "grad_norm": 2.314505100250244,
      "learning_rate": 4.633084759732873e-06,
      "loss": 0.5859,
      "step": 56330
    },
    {
      "epoch": 23.06,
      "grad_norm": 2.4579639434814453,
      "learning_rate": 4.633050770981708e-06,
      "loss": 0.5843,
      "step": 56340
    },
    {
      "epoch": 23.07,
      "grad_norm": 2.840142011642456,
      "learning_rate": 4.633016771743611e-06,
      "loss": 0.6135,
      "step": 56350
    },
    {
      "epoch": 23.07,
      "grad_norm": 2.4613704681396484,
      "learning_rate": 4.6329827620187375e-06,
      "loss": 0.5844,
      "step": 56360
    },
    {
      "epoch": 23.07,
      "grad_norm": 2.5171494483947754,
      "learning_rate": 4.632948741807243e-06,
      "loss": 0.5839,
      "step": 56370
    },
    {
      "epoch": 23.08,
      "grad_norm": 2.4233205318450928,
      "learning_rate": 4.632914711109284e-06,
      "loss": 0.616,
      "step": 56380
    },
    {
      "epoch": 23.08,
      "grad_norm": 2.4953970909118652,
      "learning_rate": 4.632880669925015e-06,
      "loss": 0.5937,
      "step": 56390
    },
    {
      "epoch": 23.09,
      "grad_norm": 3.8224732875823975,
      "learning_rate": 4.632846618254594e-06,
      "loss": 0.5881,
      "step": 56400
    },
    {
      "epoch": 23.09,
      "grad_norm": 2.7069106101989746,
      "learning_rate": 4.6328125560981755e-06,
      "loss": 0.597,
      "step": 56410
    },
    {
      "epoch": 23.09,
      "grad_norm": 2.8784849643707275,
      "learning_rate": 4.632778483455915e-06,
      "loss": 0.5771,
      "step": 56420
    },
    {
      "epoch": 23.1,
      "grad_norm": 2.586336135864258,
      "learning_rate": 4.632744400327971e-06,
      "loss": 0.5737,
      "step": 56430
    },
    {
      "epoch": 23.1,
      "grad_norm": 2.4918034076690674,
      "learning_rate": 4.632710306714498e-06,
      "loss": 0.5856,
      "step": 56440
    },
    {
      "epoch": 23.11,
      "grad_norm": 2.7542762756347656,
      "learning_rate": 4.632676202615651e-06,
      "loss": 0.6032,
      "step": 56450
    },
    {
      "epoch": 23.11,
      "grad_norm": 3.6780474185943604,
      "learning_rate": 4.632642088031589e-06,
      "loss": 0.6037,
      "step": 56460
    },
    {
      "epoch": 23.12,
      "grad_norm": 2.549105167388916,
      "learning_rate": 4.632607962962466e-06,
      "loss": 0.6225,
      "step": 56470
    },
    {
      "epoch": 23.12,
      "grad_norm": 2.415621280670166,
      "learning_rate": 4.632573827408439e-06,
      "loss": 0.5938,
      "step": 56480
    },
    {
      "epoch": 23.12,
      "grad_norm": 2.6592397689819336,
      "learning_rate": 4.632539681369665e-06,
      "loss": 0.5872,
      "step": 56490
    },
    {
      "epoch": 23.13,
      "grad_norm": 2.4957640171051025,
      "learning_rate": 4.6325055248463e-06,
      "loss": 0.5863,
      "step": 56500
    },
    {
      "epoch": 23.13,
      "grad_norm": 2.4292428493499756,
      "learning_rate": 4.6324713578385e-06,
      "loss": 0.5942,
      "step": 56510
    },
    {
      "epoch": 23.14,
      "grad_norm": 2.1307737827301025,
      "learning_rate": 4.632437180346423e-06,
      "loss": 0.5881,
      "step": 56520
    },
    {
      "epoch": 23.14,
      "grad_norm": 2.676502227783203,
      "learning_rate": 4.632402992370223e-06,
      "loss": 0.6026,
      "step": 56530
    },
    {
      "epoch": 23.14,
      "grad_norm": 2.241265058517456,
      "learning_rate": 4.632368793910059e-06,
      "loss": 0.5883,
      "step": 56540
    },
    {
      "epoch": 23.15,
      "grad_norm": 3.795992136001587,
      "learning_rate": 4.632334584966086e-06,
      "loss": 0.592,
      "step": 56550
    },
    {
      "epoch": 23.15,
      "grad_norm": 2.558239459991455,
      "learning_rate": 4.6323003655384614e-06,
      "loss": 0.5863,
      "step": 56560
    },
    {
      "epoch": 23.16,
      "grad_norm": 2.897552490234375,
      "learning_rate": 4.6322661356273424e-06,
      "loss": 0.5974,
      "step": 56570
    },
    {
      "epoch": 23.16,
      "grad_norm": 2.073875665664673,
      "learning_rate": 4.632231895232886e-06,
      "loss": 0.598,
      "step": 56580
    },
    {
      "epoch": 23.16,
      "grad_norm": 2.7652554512023926,
      "learning_rate": 4.632197644355248e-06,
      "loss": 0.5908,
      "step": 56590
    },
    {
      "epoch": 23.17,
      "grad_norm": 2.7059707641601562,
      "learning_rate": 4.632163382994585e-06,
      "loss": 0.5803,
      "step": 56600
    },
    {
      "epoch": 23.17,
      "grad_norm": 2.451819896697998,
      "learning_rate": 4.632129111151056e-06,
      "loss": 0.6111,
      "step": 56610
    },
    {
      "epoch": 23.18,
      "grad_norm": 2.687636375427246,
      "learning_rate": 4.632094828824816e-06,
      "loss": 0.6138,
      "step": 56620
    },
    {
      "epoch": 23.18,
      "grad_norm": 3.4484195709228516,
      "learning_rate": 4.6320605360160226e-06,
      "loss": 0.5983,
      "step": 56630
    },
    {
      "epoch": 23.18,
      "grad_norm": 2.595897674560547,
      "learning_rate": 4.632026232724833e-06,
      "loss": 0.5962,
      "step": 56640
    },
    {
      "epoch": 23.19,
      "grad_norm": 2.4093704223632812,
      "learning_rate": 4.631991918951405e-06,
      "loss": 0.5955,
      "step": 56650
    },
    {
      "epoch": 23.19,
      "grad_norm": 2.523531198501587,
      "learning_rate": 4.631957594695894e-06,
      "loss": 0.5862,
      "step": 56660
    },
    {
      "epoch": 23.2,
      "grad_norm": 2.5281243324279785,
      "learning_rate": 4.63192325995846e-06,
      "loss": 0.5949,
      "step": 56670
    },
    {
      "epoch": 23.2,
      "grad_norm": 2.7353696823120117,
      "learning_rate": 4.631888914739258e-06,
      "loss": 0.5884,
      "step": 56680
    },
    {
      "epoch": 23.21,
      "grad_norm": 3.1248984336853027,
      "learning_rate": 4.631854559038446e-06,
      "loss": 0.5914,
      "step": 56690
    },
    {
      "epoch": 23.21,
      "grad_norm": 1.7498503923416138,
      "learning_rate": 4.63182019285618e-06,
      "loss": 0.6062,
      "step": 56700
    },
    {
      "epoch": 23.21,
      "grad_norm": 2.4833500385284424,
      "learning_rate": 4.631785816192621e-06,
      "loss": 0.6115,
      "step": 56710
    },
    {
      "epoch": 23.22,
      "grad_norm": 2.1735899448394775,
      "learning_rate": 4.631751429047923e-06,
      "loss": 0.5894,
      "step": 56720
    },
    {
      "epoch": 23.22,
      "grad_norm": 2.875035047531128,
      "learning_rate": 4.631717031422245e-06,
      "loss": 0.5844,
      "step": 56730
    },
    {
      "epoch": 23.23,
      "grad_norm": 2.7650859355926514,
      "learning_rate": 4.631682623315744e-06,
      "loss": 0.6097,
      "step": 56740
    },
    {
      "epoch": 23.23,
      "grad_norm": 2.0377137660980225,
      "learning_rate": 4.6316482047285786e-06,
      "loss": 0.5858,
      "step": 56750
    },
    {
      "epoch": 23.23,
      "grad_norm": 2.4379007816314697,
      "learning_rate": 4.6316137756609054e-06,
      "loss": 0.6089,
      "step": 56760
    },
    {
      "epoch": 23.24,
      "grad_norm": 2.0468313694000244,
      "learning_rate": 4.631579336112883e-06,
      "loss": 0.5969,
      "step": 56770
    },
    {
      "epoch": 23.24,
      "grad_norm": 3.2723588943481445,
      "learning_rate": 4.631544886084668e-06,
      "loss": 0.5808,
      "step": 56780
    },
    {
      "epoch": 23.25,
      "grad_norm": 3.0498692989349365,
      "learning_rate": 4.63151042557642e-06,
      "loss": 0.5956,
      "step": 56790
    },
    {
      "epoch": 23.25,
      "grad_norm": 2.113138437271118,
      "learning_rate": 4.631475954588295e-06,
      "loss": 0.5968,
      "step": 56800
    },
    {
      "epoch": 23.25,
      "grad_norm": 2.5385186672210693,
      "learning_rate": 4.6314414731204526e-06,
      "loss": 0.5961,
      "step": 56810
    },
    {
      "epoch": 23.26,
      "grad_norm": 2.421177864074707,
      "learning_rate": 4.631406981173049e-06,
      "loss": 0.588,
      "step": 56820
    },
    {
      "epoch": 23.26,
      "grad_norm": 2.128499984741211,
      "learning_rate": 4.6313724787462425e-06,
      "loss": 0.6027,
      "step": 56830
    },
    {
      "epoch": 23.27,
      "grad_norm": 3.9251363277435303,
      "learning_rate": 4.631337965840193e-06,
      "loss": 0.621,
      "step": 56840
    },
    {
      "epoch": 23.27,
      "grad_norm": 2.3145132064819336,
      "learning_rate": 4.631303442455057e-06,
      "loss": 0.5812,
      "step": 56850
    },
    {
      "epoch": 23.27,
      "grad_norm": 4.490973949432373,
      "learning_rate": 4.631268908590992e-06,
      "loss": 0.5945,
      "step": 56860
    },
    {
      "epoch": 23.28,
      "grad_norm": 1.962918758392334,
      "learning_rate": 4.631234364248158e-06,
      "loss": 0.5806,
      "step": 56870
    },
    {
      "epoch": 23.28,
      "grad_norm": 2.6907577514648438,
      "learning_rate": 4.631199809426713e-06,
      "loss": 0.5885,
      "step": 56880
    },
    {
      "epoch": 23.29,
      "grad_norm": 3.2383477687835693,
      "learning_rate": 4.631165244126813e-06,
      "loss": 0.5894,
      "step": 56890
    },
    {
      "epoch": 23.29,
      "grad_norm": 2.513050079345703,
      "learning_rate": 4.63113066834862e-06,
      "loss": 0.6151,
      "step": 56900
    },
    {
      "epoch": 23.3,
      "grad_norm": 2.674121618270874,
      "learning_rate": 4.631096082092289e-06,
      "loss": 0.6064,
      "step": 56910
    },
    {
      "epoch": 23.3,
      "grad_norm": 2.3148794174194336,
      "learning_rate": 4.63106148535798e-06,
      "loss": 0.5737,
      "step": 56920
    },
    {
      "epoch": 23.3,
      "grad_norm": 2.3161633014678955,
      "learning_rate": 4.631026878145852e-06,
      "loss": 0.5993,
      "step": 56930
    },
    {
      "epoch": 23.31,
      "grad_norm": 2.8029348850250244,
      "learning_rate": 4.630992260456062e-06,
      "loss": 0.6279,
      "step": 56940
    },
    {
      "epoch": 23.31,
      "grad_norm": 2.3133842945098877,
      "learning_rate": 4.6309576322887695e-06,
      "loss": 0.5955,
      "step": 56950
    },
    {
      "epoch": 23.32,
      "grad_norm": 2.4069371223449707,
      "learning_rate": 4.630922993644133e-06,
      "loss": 0.6175,
      "step": 56960
    },
    {
      "epoch": 23.32,
      "grad_norm": 1.8493074178695679,
      "learning_rate": 4.630888344522312e-06,
      "loss": 0.5978,
      "step": 56970
    },
    {
      "epoch": 23.32,
      "grad_norm": 2.4178059101104736,
      "learning_rate": 4.630853684923464e-06,
      "loss": 0.602,
      "step": 56980
    },
    {
      "epoch": 23.33,
      "grad_norm": 2.797922372817993,
      "learning_rate": 4.630819014847748e-06,
      "loss": 0.6107,
      "step": 56990
    },
    {
      "epoch": 23.33,
      "grad_norm": 3.1405603885650635,
      "learning_rate": 4.630784334295323e-06,
      "loss": 0.5838,
      "step": 57000
    },
    {
      "epoch": 23.34,
      "grad_norm": 2.9961230754852295,
      "learning_rate": 4.630749643266349e-06,
      "loss": 0.6036,
      "step": 57010
    },
    {
      "epoch": 23.34,
      "grad_norm": 2.6877994537353516,
      "learning_rate": 4.630714941760982e-06,
      "loss": 0.5989,
      "step": 57020
    },
    {
      "epoch": 23.34,
      "grad_norm": 2.997002124786377,
      "learning_rate": 4.630680229779384e-06,
      "loss": 0.6053,
      "step": 57030
    },
    {
      "epoch": 23.35,
      "grad_norm": 1.905447006225586,
      "learning_rate": 4.6306455073217135e-06,
      "loss": 0.5828,
      "step": 57040
    },
    {
      "epoch": 23.35,
      "grad_norm": 2.3076603412628174,
      "learning_rate": 4.630610774388128e-06,
      "loss": 0.5958,
      "step": 57050
    },
    {
      "epoch": 23.36,
      "grad_norm": 2.453364133834839,
      "learning_rate": 4.630576030978786e-06,
      "loss": 0.5799,
      "step": 57060
    },
    {
      "epoch": 23.36,
      "grad_norm": 2.4953689575195312,
      "learning_rate": 4.630541277093851e-06,
      "loss": 0.603,
      "step": 57070
    },
    {
      "epoch": 23.36,
      "grad_norm": 2.5966579914093018,
      "learning_rate": 4.630506512733477e-06,
      "loss": 0.5962,
      "step": 57080
    },
    {
      "epoch": 23.37,
      "grad_norm": 2.6540894508361816,
      "learning_rate": 4.630471737897826e-06,
      "loss": 0.5868,
      "step": 57090
    },
    {
      "epoch": 23.37,
      "grad_norm": 2.058258533477783,
      "learning_rate": 4.630436952587058e-06,
      "loss": 0.5758,
      "step": 57100
    },
    {
      "epoch": 23.38,
      "grad_norm": 2.6676619052886963,
      "learning_rate": 4.630402156801331e-06,
      "loss": 0.6019,
      "step": 57110
    },
    {
      "epoch": 23.38,
      "grad_norm": 2.7986090183258057,
      "learning_rate": 4.6303673505408035e-06,
      "loss": 0.5942,
      "step": 57120
    },
    {
      "epoch": 23.39,
      "grad_norm": 2.1405746936798096,
      "learning_rate": 4.6303325338056364e-06,
      "loss": 0.6083,
      "step": 57130
    },
    {
      "epoch": 23.39,
      "grad_norm": 2.143599510192871,
      "learning_rate": 4.63029770659599e-06,
      "loss": 0.5782,
      "step": 57140
    },
    {
      "epoch": 23.39,
      "grad_norm": 2.318467140197754,
      "learning_rate": 4.630262868912021e-06,
      "loss": 0.5945,
      "step": 57150
    },
    {
      "epoch": 23.4,
      "grad_norm": 3.5698659420013428,
      "learning_rate": 4.630228020753891e-06,
      "loss": 0.6052,
      "step": 57160
    },
    {
      "epoch": 23.4,
      "grad_norm": 2.8768670558929443,
      "learning_rate": 4.63019316212176e-06,
      "loss": 0.6139,
      "step": 57170
    },
    {
      "epoch": 23.41,
      "grad_norm": 2.3341307640075684,
      "learning_rate": 4.6301582930157865e-06,
      "loss": 0.5877,
      "step": 57180
    },
    {
      "epoch": 23.41,
      "grad_norm": 3.0424697399139404,
      "learning_rate": 4.6301234134361305e-06,
      "loss": 0.6205,
      "step": 57190
    },
    {
      "epoch": 23.41,
      "grad_norm": 2.007647752761841,
      "learning_rate": 4.630088523382953e-06,
      "loss": 0.59,
      "step": 57200
    },
    {
      "epoch": 23.42,
      "grad_norm": 2.5965940952301025,
      "learning_rate": 4.630053622856412e-06,
      "loss": 0.615,
      "step": 57210
    },
    {
      "epoch": 23.42,
      "grad_norm": 3.3322033882141113,
      "learning_rate": 4.630018711856668e-06,
      "loss": 0.6025,
      "step": 57220
    },
    {
      "epoch": 23.43,
      "grad_norm": 2.9120590686798096,
      "learning_rate": 4.629983790383882e-06,
      "loss": 0.6186,
      "step": 57230
    },
    {
      "epoch": 23.43,
      "grad_norm": 3.46122670173645,
      "learning_rate": 4.629948858438212e-06,
      "loss": 0.5897,
      "step": 57240
    },
    {
      "epoch": 23.43,
      "grad_norm": 2.611302137374878,
      "learning_rate": 4.62991391601982e-06,
      "loss": 0.576,
      "step": 57250
    },
    {
      "epoch": 23.44,
      "grad_norm": 1.8929747343063354,
      "learning_rate": 4.6298789631288646e-06,
      "loss": 0.6121,
      "step": 57260
    },
    {
      "epoch": 23.44,
      "grad_norm": 2.5793139934539795,
      "learning_rate": 4.629843999765506e-06,
      "loss": 0.62,
      "step": 57270
    },
    {
      "epoch": 23.45,
      "grad_norm": 2.243441581726074,
      "learning_rate": 4.629809025929906e-06,
      "loss": 0.5943,
      "step": 57280
    },
    {
      "epoch": 23.45,
      "grad_norm": 2.4938149452209473,
      "learning_rate": 4.629774041622223e-06,
      "loss": 0.5787,
      "step": 57290
    },
    {
      "epoch": 23.45,
      "grad_norm": 2.5973474979400635,
      "learning_rate": 4.629739046842618e-06,
      "loss": 0.6188,
      "step": 57300
    },
    {
      "epoch": 23.46,
      "grad_norm": 2.5102896690368652,
      "learning_rate": 4.6297040415912515e-06,
      "loss": 0.6162,
      "step": 57310
    },
    {
      "epoch": 23.46,
      "grad_norm": 2.6967108249664307,
      "learning_rate": 4.629669025868283e-06,
      "loss": 0.6001,
      "step": 57320
    },
    {
      "epoch": 23.47,
      "grad_norm": 2.302431106567383,
      "learning_rate": 4.629633999673874e-06,
      "loss": 0.5867,
      "step": 57330
    },
    {
      "epoch": 23.47,
      "grad_norm": 1.9688215255737305,
      "learning_rate": 4.6295989630081835e-06,
      "loss": 0.6033,
      "step": 57340
    },
    {
      "epoch": 23.48,
      "grad_norm": 2.3675198554992676,
      "learning_rate": 4.629563915871374e-06,
      "loss": 0.5966,
      "step": 57350
    },
    {
      "epoch": 23.48,
      "grad_norm": 2.52168607711792,
      "learning_rate": 4.629528858263604e-06,
      "loss": 0.5875,
      "step": 57360
    },
    {
      "epoch": 23.48,
      "grad_norm": 2.16013765335083,
      "learning_rate": 4.629493790185036e-06,
      "loss": 0.5718,
      "step": 57370
    },
    {
      "epoch": 23.49,
      "grad_norm": 2.0672826766967773,
      "learning_rate": 4.629458711635829e-06,
      "loss": 0.5773,
      "step": 57380
    },
    {
      "epoch": 23.49,
      "grad_norm": 1.8393608331680298,
      "learning_rate": 4.629423622616145e-06,
      "loss": 0.59,
      "step": 57390
    },
    {
      "epoch": 23.5,
      "grad_norm": 2.7658851146698,
      "learning_rate": 4.629388523126144e-06,
      "loss": 0.5879,
      "step": 57400
    },
    {
      "epoch": 23.5,
      "grad_norm": 2.573817729949951,
      "learning_rate": 4.629353413165986e-06,
      "loss": 0.6328,
      "step": 57410
    },
    {
      "epoch": 23.5,
      "grad_norm": 1.8298423290252686,
      "learning_rate": 4.629318292735833e-06,
      "loss": 0.6075,
      "step": 57420
    },
    {
      "epoch": 23.51,
      "grad_norm": 2.5875205993652344,
      "learning_rate": 4.629283161835846e-06,
      "loss": 0.5862,
      "step": 57430
    },
    {
      "epoch": 23.51,
      "grad_norm": 2.279186248779297,
      "learning_rate": 4.629248020466186e-06,
      "loss": 0.5695,
      "step": 57440
    },
    {
      "epoch": 23.52,
      "grad_norm": 2.5764713287353516,
      "learning_rate": 4.629212868627013e-06,
      "loss": 0.5835,
      "step": 57450
    },
    {
      "epoch": 23.52,
      "grad_norm": 2.332254409790039,
      "learning_rate": 4.629177706318489e-06,
      "loss": 0.6083,
      "step": 57460
    },
    {
      "epoch": 23.52,
      "grad_norm": 2.1391568183898926,
      "learning_rate": 4.629142533540773e-06,
      "loss": 0.6024,
      "step": 57470
    },
    {
      "epoch": 23.53,
      "grad_norm": 2.3055052757263184,
      "learning_rate": 4.62910735029403e-06,
      "loss": 0.6002,
      "step": 57480
    },
    {
      "epoch": 23.53,
      "grad_norm": 2.845945358276367,
      "learning_rate": 4.629072156578418e-06,
      "loss": 0.6013,
      "step": 57490
    },
    {
      "epoch": 23.54,
      "grad_norm": 2.848635673522949,
      "learning_rate": 4.6290369523940984e-06,
      "loss": 0.5803,
      "step": 57500
    },
    {
      "epoch": 23.54,
      "grad_norm": 2.7911181449890137,
      "learning_rate": 4.629001737741234e-06,
      "loss": 0.5992,
      "step": 57510
    },
    {
      "epoch": 23.54,
      "grad_norm": 2.157257318496704,
      "learning_rate": 4.6289665126199855e-06,
      "loss": 0.5978,
      "step": 57520
    },
    {
      "epoch": 23.55,
      "grad_norm": 1.8728693723678589,
      "learning_rate": 4.628931277030513e-06,
      "loss": 0.5852,
      "step": 57530
    },
    {
      "epoch": 23.55,
      "grad_norm": 2.3495609760284424,
      "learning_rate": 4.62889603097298e-06,
      "loss": 0.5989,
      "step": 57540
    },
    {
      "epoch": 23.56,
      "grad_norm": 2.85127592086792,
      "learning_rate": 4.628860774447547e-06,
      "loss": 0.5547,
      "step": 57550
    },
    {
      "epoch": 23.56,
      "grad_norm": 2.5057458877563477,
      "learning_rate": 4.628825507454375e-06,
      "loss": 0.6058,
      "step": 57560
    },
    {
      "epoch": 23.57,
      "grad_norm": 1.9822665452957153,
      "learning_rate": 4.628790229993626e-06,
      "loss": 0.6143,
      "step": 57570
    },
    {
      "epoch": 23.57,
      "grad_norm": 1.7732127904891968,
      "learning_rate": 4.628754942065461e-06,
      "loss": 0.5867,
      "step": 57580
    },
    {
      "epoch": 23.57,
      "grad_norm": 2.9051477909088135,
      "learning_rate": 4.6287196436700425e-06,
      "loss": 0.6067,
      "step": 57590
    },
    {
      "epoch": 23.58,
      "grad_norm": 2.7317733764648438,
      "learning_rate": 4.628684334807533e-06,
      "loss": 0.587,
      "step": 57600
    },
    {
      "epoch": 23.58,
      "grad_norm": 3.5913596153259277,
      "learning_rate": 4.628649015478092e-06,
      "loss": 0.6002,
      "step": 57610
    },
    {
      "epoch": 23.59,
      "grad_norm": 1.9539633989334106,
      "learning_rate": 4.628613685681883e-06,
      "loss": 0.5895,
      "step": 57620
    },
    {
      "epoch": 23.59,
      "grad_norm": 2.6794636249542236,
      "learning_rate": 4.628578345419067e-06,
      "loss": 0.6053,
      "step": 57630
    },
    {
      "epoch": 23.59,
      "grad_norm": 2.679809808731079,
      "learning_rate": 4.628542994689806e-06,
      "loss": 0.6126,
      "step": 57640
    },
    {
      "epoch": 23.6,
      "grad_norm": 2.2837371826171875,
      "learning_rate": 4.628507633494263e-06,
      "loss": 0.5841,
      "step": 57650
    },
    {
      "epoch": 23.6,
      "grad_norm": 2.6874845027923584,
      "learning_rate": 4.6284722618325985e-06,
      "loss": 0.5845,
      "step": 57660
    },
    {
      "epoch": 23.61,
      "grad_norm": 2.721409797668457,
      "learning_rate": 4.6284368797049754e-06,
      "loss": 0.6047,
      "step": 57670
    },
    {
      "epoch": 23.61,
      "grad_norm": 2.569456100463867,
      "learning_rate": 4.6284014871115555e-06,
      "loss": 0.6047,
      "step": 57680
    },
    {
      "epoch": 23.61,
      "grad_norm": 2.5649735927581787,
      "learning_rate": 4.6283660840525015e-06,
      "loss": 0.6138,
      "step": 57690
    },
    {
      "epoch": 23.62,
      "grad_norm": 1.9548743963241577,
      "learning_rate": 4.628330670527974e-06,
      "loss": 0.5758,
      "step": 57700
    },
    {
      "epoch": 23.62,
      "grad_norm": 2.602661609649658,
      "learning_rate": 4.628295246538137e-06,
      "loss": 0.6002,
      "step": 57710
    },
    {
      "epoch": 23.63,
      "grad_norm": 2.2357213497161865,
      "learning_rate": 4.628259812083152e-06,
      "loss": 0.59,
      "step": 57720
    },
    {
      "epoch": 23.63,
      "grad_norm": 2.600947856903076,
      "learning_rate": 4.628224367163181e-06,
      "loss": 0.6044,
      "step": 57730
    },
    {
      "epoch": 23.63,
      "grad_norm": 2.3723888397216797,
      "learning_rate": 4.628188911778386e-06,
      "loss": 0.6043,
      "step": 57740
    },
    {
      "epoch": 23.64,
      "grad_norm": 2.9345388412475586,
      "learning_rate": 4.628153445928931e-06,
      "loss": 0.6067,
      "step": 57750
    },
    {
      "epoch": 23.64,
      "grad_norm": 2.7744057178497314,
      "learning_rate": 4.628117969614978e-06,
      "loss": 0.5986,
      "step": 57760
    },
    {
      "epoch": 23.65,
      "grad_norm": 2.7955031394958496,
      "learning_rate": 4.628082482836689e-06,
      "loss": 0.5855,
      "step": 57770
    },
    {
      "epoch": 23.65,
      "grad_norm": 2.192763090133667,
      "learning_rate": 4.628046985594226e-06,
      "loss": 0.5964,
      "step": 57780
    },
    {
      "epoch": 23.66,
      "grad_norm": 2.6248881816864014,
      "learning_rate": 4.628011477887753e-06,
      "loss": 0.5867,
      "step": 57790
    },
    {
      "epoch": 23.66,
      "grad_norm": 2.871560573577881,
      "learning_rate": 4.627975959717431e-06,
      "loss": 0.5944,
      "step": 57800
    },
    {
      "epoch": 23.66,
      "grad_norm": 2.3156275749206543,
      "learning_rate": 4.627940431083424e-06,
      "loss": 0.5862,
      "step": 57810
    },
    {
      "epoch": 23.67,
      "grad_norm": 2.2837634086608887,
      "learning_rate": 4.627904891985895e-06,
      "loss": 0.5814,
      "step": 57820
    },
    {
      "epoch": 23.67,
      "grad_norm": 2.8571951389312744,
      "learning_rate": 4.627869342425006e-06,
      "loss": 0.6231,
      "step": 57830
    },
    {
      "epoch": 23.68,
      "grad_norm": 2.10536789894104,
      "learning_rate": 4.627833782400919e-06,
      "loss": 0.5938,
      "step": 57840
    },
    {
      "epoch": 23.68,
      "grad_norm": 2.651041269302368,
      "learning_rate": 4.627798211913799e-06,
      "loss": 0.5854,
      "step": 57850
    },
    {
      "epoch": 23.68,
      "grad_norm": 3.3652374744415283,
      "learning_rate": 4.627762630963808e-06,
      "loss": 0.6048,
      "step": 57860
    },
    {
      "epoch": 23.69,
      "grad_norm": 2.7196567058563232,
      "learning_rate": 4.6277270395511085e-06,
      "loss": 0.5644,
      "step": 57870
    },
    {
      "epoch": 23.69,
      "grad_norm": 3.0259664058685303,
      "learning_rate": 4.627691437675864e-06,
      "loss": 0.5985,
      "step": 57880
    },
    {
      "epoch": 23.7,
      "grad_norm": 2.5956153869628906,
      "learning_rate": 4.627655825338237e-06,
      "loss": 0.6205,
      "step": 57890
    },
    {
      "epoch": 23.7,
      "grad_norm": 3.160275936126709,
      "learning_rate": 4.627620202538392e-06,
      "loss": 0.6002,
      "step": 57900
    },
    {
      "epoch": 23.7,
      "grad_norm": 2.606424331665039,
      "learning_rate": 4.62758456927649e-06,
      "loss": 0.5993,
      "step": 57910
    },
    {
      "epoch": 23.71,
      "grad_norm": 2.135303258895874,
      "learning_rate": 4.627548925552696e-06,
      "loss": 0.5939,
      "step": 57920
    },
    {
      "epoch": 23.71,
      "grad_norm": 2.293117046356201,
      "learning_rate": 4.627513271367173e-06,
      "loss": 0.5894,
      "step": 57930
    },
    {
      "epoch": 23.72,
      "grad_norm": 2.315826654434204,
      "learning_rate": 4.627477606720084e-06,
      "loss": 0.6084,
      "step": 57940
    },
    {
      "epoch": 23.72,
      "grad_norm": 3.0131845474243164,
      "learning_rate": 4.627441931611593e-06,
      "loss": 0.6,
      "step": 57950
    },
    {
      "epoch": 23.72,
      "grad_norm": 2.619136095046997,
      "learning_rate": 4.627406246041863e-06,
      "loss": 0.5919,
      "step": 57960
    },
    {
      "epoch": 23.73,
      "grad_norm": 2.240922212600708,
      "learning_rate": 4.627370550011057e-06,
      "loss": 0.5755,
      "step": 57970
    },
    {
      "epoch": 23.73,
      "grad_norm": 2.1082613468170166,
      "learning_rate": 4.627334843519339e-06,
      "loss": 0.6013,
      "step": 57980
    },
    {
      "epoch": 23.74,
      "grad_norm": 1.497945785522461,
      "learning_rate": 4.627299126566873e-06,
      "loss": 0.6095,
      "step": 57990
    },
    {
      "epoch": 23.74,
      "grad_norm": 2.7908670902252197,
      "learning_rate": 4.627263399153822e-06,
      "loss": 0.6061,
      "step": 58000
    },
    {
      "epoch": 23.75,
      "grad_norm": 1.7149051427841187,
      "learning_rate": 4.627227661280348e-06,
      "loss": 0.6318,
      "step": 58010
    },
    {
      "epoch": 23.75,
      "grad_norm": 2.9450294971466064,
      "learning_rate": 4.6271919129466184e-06,
      "loss": 0.5858,
      "step": 58020
    },
    {
      "epoch": 23.75,
      "grad_norm": 2.213747024536133,
      "learning_rate": 4.627156154152795e-06,
      "loss": 0.5971,
      "step": 58030
    },
    {
      "epoch": 23.76,
      "grad_norm": 2.708775758743286,
      "learning_rate": 4.627120384899041e-06,
      "loss": 0.6113,
      "step": 58040
    },
    {
      "epoch": 23.76,
      "grad_norm": 2.16001558303833,
      "learning_rate": 4.627084605185521e-06,
      "loss": 0.5818,
      "step": 58050
    },
    {
      "epoch": 23.77,
      "grad_norm": 3.0348851680755615,
      "learning_rate": 4.6270488150123985e-06,
      "loss": 0.606,
      "step": 58060
    },
    {
      "epoch": 23.77,
      "grad_norm": 3.2623279094696045,
      "learning_rate": 4.627013014379838e-06,
      "loss": 0.6071,
      "step": 58070
    },
    {
      "epoch": 23.77,
      "grad_norm": 2.3213655948638916,
      "learning_rate": 4.626977203288003e-06,
      "loss": 0.5996,
      "step": 58080
    },
    {
      "epoch": 23.78,
      "grad_norm": 2.1912167072296143,
      "learning_rate": 4.626941381737058e-06,
      "loss": 0.6027,
      "step": 58090
    },
    {
      "epoch": 23.78,
      "grad_norm": 2.4468836784362793,
      "learning_rate": 4.626905549727167e-06,
      "loss": 0.6049,
      "step": 58100
    },
    {
      "epoch": 23.79,
      "grad_norm": 2.679779052734375,
      "learning_rate": 4.626869707258494e-06,
      "loss": 0.5995,
      "step": 58110
    },
    {
      "epoch": 23.79,
      "grad_norm": 2.5840818881988525,
      "learning_rate": 4.626833854331203e-06,
      "loss": 0.5796,
      "step": 58120
    },
    {
      "epoch": 23.79,
      "grad_norm": 3.5899901390075684,
      "learning_rate": 4.626797990945459e-06,
      "loss": 0.606,
      "step": 58130
    },
    {
      "epoch": 23.8,
      "grad_norm": 2.9732720851898193,
      "learning_rate": 4.6267621171014254e-06,
      "loss": 0.5902,
      "step": 58140
    },
    {
      "epoch": 23.8,
      "grad_norm": 2.29474139213562,
      "learning_rate": 4.626726232799267e-06,
      "loss": 0.5887,
      "step": 58150
    },
    {
      "epoch": 23.81,
      "grad_norm": 2.430844783782959,
      "learning_rate": 4.626690338039147e-06,
      "loss": 0.5932,
      "step": 58160
    },
    {
      "epoch": 23.81,
      "grad_norm": 3.0038492679595947,
      "learning_rate": 4.626654432821232e-06,
      "loss": 0.567,
      "step": 58170
    },
    {
      "epoch": 23.81,
      "grad_norm": 2.87182879447937,
      "learning_rate": 4.6266185171456855e-06,
      "loss": 0.5918,
      "step": 58180
    },
    {
      "epoch": 23.82,
      "grad_norm": 1.9615675210952759,
      "learning_rate": 4.6265825910126716e-06,
      "loss": 0.6216,
      "step": 58190
    },
    {
      "epoch": 23.82,
      "grad_norm": 2.150282382965088,
      "learning_rate": 4.626546654422355e-06,
      "loss": 0.5912,
      "step": 58200
    },
    {
      "epoch": 23.83,
      "grad_norm": 2.493434190750122,
      "learning_rate": 4.6265107073749e-06,
      "loss": 0.6118,
      "step": 58210
    },
    {
      "epoch": 23.83,
      "grad_norm": 2.9842188358306885,
      "learning_rate": 4.626474749870472e-06,
      "loss": 0.5914,
      "step": 58220
    },
    {
      "epoch": 23.84,
      "grad_norm": 2.452833414077759,
      "learning_rate": 4.626438781909236e-06,
      "loss": 0.5897,
      "step": 58230
    },
    {
      "epoch": 23.84,
      "grad_norm": 2.06874942779541,
      "learning_rate": 4.626402803491356e-06,
      "loss": 0.6,
      "step": 58240
    },
    {
      "epoch": 23.84,
      "grad_norm": 25.66543197631836,
      "learning_rate": 4.626366814616996e-06,
      "loss": 0.5823,
      "step": 58250
    },
    {
      "epoch": 23.85,
      "grad_norm": 4.0688700675964355,
      "learning_rate": 4.626330815286323e-06,
      "loss": 0.6139,
      "step": 58260
    },
    {
      "epoch": 23.85,
      "grad_norm": 2.1076104640960693,
      "learning_rate": 4.626294805499501e-06,
      "loss": 0.5809,
      "step": 58270
    },
    {
      "epoch": 23.86,
      "grad_norm": 3.2278685569763184,
      "learning_rate": 4.626258785256694e-06,
      "loss": 0.6031,
      "step": 58280
    },
    {
      "epoch": 23.86,
      "grad_norm": 3.630009651184082,
      "learning_rate": 4.626222754558068e-06,
      "loss": 0.5934,
      "step": 58290
    },
    {
      "epoch": 23.86,
      "grad_norm": 2.207069158554077,
      "learning_rate": 4.626186713403788e-06,
      "loss": 0.6087,
      "step": 58300
    },
    {
      "epoch": 23.87,
      "grad_norm": 2.897097110748291,
      "learning_rate": 4.626150661794018e-06,
      "loss": 0.5837,
      "step": 58310
    },
    {
      "epoch": 23.87,
      "grad_norm": 2.7415804862976074,
      "learning_rate": 4.626114599728925e-06,
      "loss": 0.598,
      "step": 58320
    },
    {
      "epoch": 23.88,
      "grad_norm": 2.5656075477600098,
      "learning_rate": 4.6260785272086735e-06,
      "loss": 0.6078,
      "step": 58330
    },
    {
      "epoch": 23.88,
      "grad_norm": 2.806044340133667,
      "learning_rate": 4.6260424442334275e-06,
      "loss": 0.5818,
      "step": 58340
    },
    {
      "epoch": 23.88,
      "grad_norm": 3.1231157779693604,
      "learning_rate": 4.626006350803354e-06,
      "loss": 0.6067,
      "step": 58350
    },
    {
      "epoch": 23.89,
      "grad_norm": 2.8426382541656494,
      "learning_rate": 4.625970246918617e-06,
      "loss": 0.6085,
      "step": 58360
    },
    {
      "epoch": 23.89,
      "grad_norm": 2.112133502960205,
      "learning_rate": 4.625934132579383e-06,
      "loss": 0.5912,
      "step": 58370
    },
    {
      "epoch": 23.9,
      "grad_norm": 2.206954002380371,
      "learning_rate": 4.625898007785817e-06,
      "loss": 0.591,
      "step": 58380
    },
    {
      "epoch": 23.9,
      "grad_norm": 2.7307138442993164,
      "learning_rate": 4.625861872538083e-06,
      "loss": 0.6169,
      "step": 58390
    },
    {
      "epoch": 23.91,
      "grad_norm": 2.1582493782043457,
      "learning_rate": 4.6258257268363495e-06,
      "loss": 0.6029,
      "step": 58400
    },
    {
      "epoch": 23.91,
      "grad_norm": 3.0610594749450684,
      "learning_rate": 4.625789570680781e-06,
      "loss": 0.5924,
      "step": 58410
    },
    {
      "epoch": 23.91,
      "grad_norm": 2.4300012588500977,
      "learning_rate": 4.625753404071541e-06,
      "loss": 0.6089,
      "step": 58420
    },
    {
      "epoch": 23.92,
      "grad_norm": 1.9596058130264282,
      "learning_rate": 4.625717227008798e-06,
      "loss": 0.5962,
      "step": 58430
    },
    {
      "epoch": 23.92,
      "grad_norm": 2.7609713077545166,
      "learning_rate": 4.6256810394927165e-06,
      "loss": 0.5785,
      "step": 58440
    },
    {
      "epoch": 23.93,
      "grad_norm": 2.683321237564087,
      "learning_rate": 4.625644841523461e-06,
      "loss": 0.6046,
      "step": 58450
    },
    {
      "epoch": 23.93,
      "grad_norm": 2.6441121101379395,
      "learning_rate": 4.6256086331012e-06,
      "loss": 0.5892,
      "step": 58460
    },
    {
      "epoch": 23.93,
      "grad_norm": 2.410578727722168,
      "learning_rate": 4.625572414226097e-06,
      "loss": 0.6082,
      "step": 58470
    },
    {
      "epoch": 23.94,
      "grad_norm": 3.8324222564697266,
      "learning_rate": 4.62553618489832e-06,
      "loss": 0.5936,
      "step": 58480
    },
    {
      "epoch": 23.94,
      "grad_norm": 2.4216980934143066,
      "learning_rate": 4.625499945118033e-06,
      "loss": 0.6012,
      "step": 58490
    },
    {
      "epoch": 23.95,
      "grad_norm": 2.6260106563568115,
      "learning_rate": 4.625463694885403e-06,
      "loss": 0.5989,
      "step": 58500
    },
    {
      "epoch": 23.95,
      "grad_norm": 2.3717501163482666,
      "learning_rate": 4.625427434200597e-06,
      "loss": 0.599,
      "step": 58510
    },
    {
      "epoch": 23.95,
      "grad_norm": 2.5572986602783203,
      "learning_rate": 4.625391163063778e-06,
      "loss": 0.6008,
      "step": 58520
    },
    {
      "epoch": 23.96,
      "grad_norm": 2.5155904293060303,
      "learning_rate": 4.625354881475116e-06,
      "loss": 0.588,
      "step": 58530
    },
    {
      "epoch": 23.96,
      "grad_norm": 2.380793809890747,
      "learning_rate": 4.625318589434774e-06,
      "loss": 0.5998,
      "step": 58540
    },
    {
      "epoch": 23.97,
      "grad_norm": 3.1761295795440674,
      "learning_rate": 4.625282286942921e-06,
      "loss": 0.5955,
      "step": 58550
    },
    {
      "epoch": 23.97,
      "grad_norm": 2.4075088500976562,
      "learning_rate": 4.625245973999722e-06,
      "loss": 0.604,
      "step": 58560
    },
    {
      "epoch": 23.97,
      "grad_norm": 2.518930196762085,
      "learning_rate": 4.625209650605342e-06,
      "loss": 0.5905,
      "step": 58570
    },
    {
      "epoch": 23.98,
      "grad_norm": 4.362074851989746,
      "learning_rate": 4.625173316759949e-06,
      "loss": 0.6118,
      "step": 58580
    },
    {
      "epoch": 23.98,
      "grad_norm": 2.6410598754882812,
      "learning_rate": 4.62513697246371e-06,
      "loss": 0.5828,
      "step": 58590
    },
    {
      "epoch": 23.99,
      "grad_norm": 2.6387345790863037,
      "learning_rate": 4.62510061771679e-06,
      "loss": 0.5897,
      "step": 58600
    },
    {
      "epoch": 23.99,
      "grad_norm": 2.273408889770508,
      "learning_rate": 4.625064252519355e-06,
      "loss": 0.5843,
      "step": 58610
    },
    {
      "epoch": 24.0,
      "grad_norm": 3.406968116760254,
      "learning_rate": 4.6250278768715745e-06,
      "loss": 0.5717,
      "step": 58620
    },
    {
      "epoch": 24.0,
      "grad_norm": 2.883575677871704,
      "learning_rate": 4.624991490773613e-06,
      "loss": 0.5741,
      "step": 58630
    },
    {
      "epoch": 24.0,
      "eval_loss": 0.593991756439209,
      "eval_runtime": 51.9304,
      "eval_samples_per_second": 66.416,
      "eval_steps_per_second": 8.319,
      "step": 58632
    },
    {
      "epoch": 24.0,
      "grad_norm": 2.150768518447876,
      "learning_rate": 4.624955094225637e-06,
      "loss": 0.5728,
      "step": 58640
    },
    {
      "epoch": 24.01,
      "grad_norm": 2.66652250289917,
      "learning_rate": 4.624918687227814e-06,
      "loss": 0.6051,
      "step": 58650
    },
    {
      "epoch": 24.01,
      "grad_norm": 1.9739282131195068,
      "learning_rate": 4.624882269780311e-06,
      "loss": 0.581,
      "step": 58660
    },
    {
      "epoch": 24.02,
      "grad_norm": 3.1051816940307617,
      "learning_rate": 4.624845841883293e-06,
      "loss": 0.6029,
      "step": 58670
    },
    {
      "epoch": 24.02,
      "grad_norm": 2.2540969848632812,
      "learning_rate": 4.624809403536929e-06,
      "loss": 0.6189,
      "step": 58680
    },
    {
      "epoch": 24.02,
      "grad_norm": 2.758230447769165,
      "learning_rate": 4.624772954741385e-06,
      "loss": 0.5929,
      "step": 58690
    },
    {
      "epoch": 24.03,
      "grad_norm": 2.814464569091797,
      "learning_rate": 4.624736495496829e-06,
      "loss": 0.5786,
      "step": 58700
    },
    {
      "epoch": 24.03,
      "grad_norm": 2.558760643005371,
      "learning_rate": 4.624700025803426e-06,
      "loss": 0.5893,
      "step": 58710
    },
    {
      "epoch": 24.04,
      "grad_norm": 2.3374345302581787,
      "learning_rate": 4.624663545661344e-06,
      "loss": 0.579,
      "step": 58720
    },
    {
      "epoch": 24.04,
      "grad_norm": 2.1816012859344482,
      "learning_rate": 4.624627055070751e-06,
      "loss": 0.61,
      "step": 58730
    },
    {
      "epoch": 24.04,
      "grad_norm": 2.7466964721679688,
      "learning_rate": 4.624590554031814e-06,
      "loss": 0.6008,
      "step": 58740
    },
    {
      "epoch": 24.05,
      "grad_norm": 2.3933005332946777,
      "learning_rate": 4.624554042544699e-06,
      "loss": 0.5952,
      "step": 58750
    },
    {
      "epoch": 24.05,
      "grad_norm": 2.6252975463867188,
      "learning_rate": 4.624517520609574e-06,
      "loss": 0.6004,
      "step": 58760
    },
    {
      "epoch": 24.06,
      "grad_norm": 2.844984292984009,
      "learning_rate": 4.6244809882266055e-06,
      "loss": 0.6269,
      "step": 58770
    },
    {
      "epoch": 24.06,
      "grad_norm": 2.3348376750946045,
      "learning_rate": 4.624444445395962e-06,
      "loss": 0.5988,
      "step": 58780
    },
    {
      "epoch": 24.06,
      "grad_norm": 2.503671884536743,
      "learning_rate": 4.624407892117811e-06,
      "loss": 0.5913,
      "step": 58790
    },
    {
      "epoch": 24.07,
      "grad_norm": 2.0964527130126953,
      "learning_rate": 4.624371328392319e-06,
      "loss": 0.6096,
      "step": 58800
    },
    {
      "epoch": 24.07,
      "grad_norm": 2.8249995708465576,
      "learning_rate": 4.624334754219653e-06,
      "loss": 0.5942,
      "step": 58810
    },
    {
      "epoch": 24.08,
      "grad_norm": 2.6817688941955566,
      "learning_rate": 4.6242981695999835e-06,
      "loss": 0.5912,
      "step": 58820
    },
    {
      "epoch": 24.08,
      "grad_norm": 2.3306078910827637,
      "learning_rate": 4.6242615745334746e-06,
      "loss": 0.5702,
      "step": 58830
    },
    {
      "epoch": 24.09,
      "grad_norm": 2.721959352493286,
      "learning_rate": 4.624224969020296e-06,
      "loss": 0.5566,
      "step": 58840
    },
    {
      "epoch": 24.09,
      "grad_norm": 2.987189531326294,
      "learning_rate": 4.624188353060614e-06,
      "loss": 0.5882,
      "step": 58850
    },
    {
      "epoch": 24.09,
      "grad_norm": 2.2458865642547607,
      "learning_rate": 4.624151726654597e-06,
      "loss": 0.5932,
      "step": 58860
    },
    {
      "epoch": 24.1,
      "grad_norm": 3.8850302696228027,
      "learning_rate": 4.624115089802413e-06,
      "loss": 0.5694,
      "step": 58870
    },
    {
      "epoch": 24.1,
      "grad_norm": 2.4799342155456543,
      "learning_rate": 4.624078442504231e-06,
      "loss": 0.5831,
      "step": 58880
    },
    {
      "epoch": 24.11,
      "grad_norm": 2.808885097503662,
      "learning_rate": 4.6240417847602155e-06,
      "loss": 0.5948,
      "step": 58890
    },
    {
      "epoch": 24.11,
      "grad_norm": 3.625253438949585,
      "learning_rate": 4.624005116570538e-06,
      "loss": 0.6022,
      "step": 58900
    },
    {
      "epoch": 24.11,
      "grad_norm": 2.187035322189331,
      "learning_rate": 4.623968437935364e-06,
      "loss": 0.5949,
      "step": 58910
    },
    {
      "epoch": 24.12,
      "grad_norm": 2.5306084156036377,
      "learning_rate": 4.623931748854863e-06,
      "loss": 0.6031,
      "step": 58920
    },
    {
      "epoch": 24.12,
      "grad_norm": 2.22078275680542,
      "learning_rate": 4.623895049329203e-06,
      "loss": 0.5995,
      "step": 58930
    },
    {
      "epoch": 24.13,
      "grad_norm": 2.753115177154541,
      "learning_rate": 4.623858339358551e-06,
      "loss": 0.5823,
      "step": 58940
    },
    {
      "epoch": 24.13,
      "grad_norm": 2.706465482711792,
      "learning_rate": 4.623821618943076e-06,
      "loss": 0.6096,
      "step": 58950
    },
    {
      "epoch": 24.13,
      "grad_norm": 2.6907098293304443,
      "learning_rate": 4.623784888082946e-06,
      "loss": 0.6139,
      "step": 58960
    },
    {
      "epoch": 24.14,
      "grad_norm": 2.4573488235473633,
      "learning_rate": 4.623748146778329e-06,
      "loss": 0.5913,
      "step": 58970
    },
    {
      "epoch": 24.14,
      "grad_norm": 3.507486581802368,
      "learning_rate": 4.623711395029394e-06,
      "loss": 0.5869,
      "step": 58980
    },
    {
      "epoch": 24.15,
      "grad_norm": 2.0839433670043945,
      "learning_rate": 4.623674632836308e-06,
      "loss": 0.5864,
      "step": 58990
    },
    {
      "epoch": 24.15,
      "grad_norm": 2.244410991668701,
      "learning_rate": 4.623637860199241e-06,
      "loss": 0.5975,
      "step": 59000
    },
    {
      "epoch": 24.15,
      "grad_norm": 3.111372232437134,
      "learning_rate": 4.6236010771183615e-06,
      "loss": 0.6181,
      "step": 59010
    },
    {
      "epoch": 24.16,
      "grad_norm": 2.932957887649536,
      "learning_rate": 4.623564283593837e-06,
      "loss": 0.6095,
      "step": 59020
    },
    {
      "epoch": 24.16,
      "grad_norm": 2.987668514251709,
      "learning_rate": 4.623527479625836e-06,
      "loss": 0.5893,
      "step": 59030
    },
    {
      "epoch": 24.17,
      "grad_norm": 2.147298812866211,
      "learning_rate": 4.623490665214527e-06,
      "loss": 0.5863,
      "step": 59040
    },
    {
      "epoch": 24.17,
      "grad_norm": 2.4411234855651855,
      "learning_rate": 4.62345384036008e-06,
      "loss": 0.6041,
      "step": 59050
    },
    {
      "epoch": 24.18,
      "grad_norm": 3.2309377193450928,
      "learning_rate": 4.623417005062663e-06,
      "loss": 0.6098,
      "step": 59060
    },
    {
      "epoch": 24.18,
      "grad_norm": 3.384693145751953,
      "learning_rate": 4.623380159322444e-06,
      "loss": 0.5907,
      "step": 59070
    },
    {
      "epoch": 24.18,
      "grad_norm": 2.5620038509368896,
      "learning_rate": 4.623343303139591e-06,
      "loss": 0.5899,
      "step": 59080
    },
    {
      "epoch": 24.19,
      "grad_norm": 2.732771158218384,
      "learning_rate": 4.6233064365142755e-06,
      "loss": 0.5756,
      "step": 59090
    },
    {
      "epoch": 24.19,
      "grad_norm": 2.460899829864502,
      "learning_rate": 4.623269559446665e-06,
      "loss": 0.5874,
      "step": 59100
    },
    {
      "epoch": 24.2,
      "grad_norm": 2.613072156906128,
      "learning_rate": 4.623232671936928e-06,
      "loss": 0.585,
      "step": 59110
    },
    {
      "epoch": 24.2,
      "grad_norm": 1.4375101327896118,
      "learning_rate": 4.623195773985234e-06,
      "loss": 0.5965,
      "step": 59120
    },
    {
      "epoch": 24.2,
      "grad_norm": 2.280318260192871,
      "learning_rate": 4.623158865591752e-06,
      "loss": 0.5833,
      "step": 59130
    },
    {
      "epoch": 24.21,
      "grad_norm": 2.196949005126953,
      "learning_rate": 4.623121946756651e-06,
      "loss": 0.5938,
      "step": 59140
    },
    {
      "epoch": 24.21,
      "grad_norm": 2.8032472133636475,
      "learning_rate": 4.6230850174801e-06,
      "loss": 0.5822,
      "step": 59150
    },
    {
      "epoch": 24.22,
      "grad_norm": 1.661973476409912,
      "learning_rate": 4.623048077762269e-06,
      "loss": 0.5657,
      "step": 59160
    },
    {
      "epoch": 24.22,
      "grad_norm": 2.433018207550049,
      "learning_rate": 4.623011127603326e-06,
      "loss": 0.6128,
      "step": 59170
    },
    {
      "epoch": 24.22,
      "grad_norm": 2.7512855529785156,
      "learning_rate": 4.62297416700344e-06,
      "loss": 0.6091,
      "step": 59180
    },
    {
      "epoch": 24.23,
      "grad_norm": 2.6918928623199463,
      "learning_rate": 4.622937195962782e-06,
      "loss": 0.6051,
      "step": 59190
    },
    {
      "epoch": 24.23,
      "grad_norm": 2.60770320892334,
      "learning_rate": 4.62290021448152e-06,
      "loss": 0.5823,
      "step": 59200
    },
    {
      "epoch": 24.24,
      "grad_norm": 2.9721930027008057,
      "learning_rate": 4.6228632225598244e-06,
      "loss": 0.6004,
      "step": 59210
    },
    {
      "epoch": 24.24,
      "grad_norm": 2.8976986408233643,
      "learning_rate": 4.622826220197863e-06,
      "loss": 0.592,
      "step": 59220
    },
    {
      "epoch": 24.24,
      "grad_norm": 2.4563393592834473,
      "learning_rate": 4.622789207395808e-06,
      "loss": 0.5985,
      "step": 59230
    },
    {
      "epoch": 24.25,
      "grad_norm": 2.6996326446533203,
      "learning_rate": 4.6227521841538265e-06,
      "loss": 0.5877,
      "step": 59240
    },
    {
      "epoch": 24.25,
      "grad_norm": 1.8839510679244995,
      "learning_rate": 4.622715150472089e-06,
      "loss": 0.5957,
      "step": 59250
    },
    {
      "epoch": 24.26,
      "grad_norm": 2.238508462905884,
      "learning_rate": 4.622678106350764e-06,
      "loss": 0.6112,
      "step": 59260
    },
    {
      "epoch": 24.26,
      "grad_norm": 2.9506754875183105,
      "learning_rate": 4.622641051790024e-06,
      "loss": 0.5981,
      "step": 59270
    },
    {
      "epoch": 24.27,
      "grad_norm": 2.8713884353637695,
      "learning_rate": 4.6226039867900355e-06,
      "loss": 0.5739,
      "step": 59280
    },
    {
      "epoch": 24.27,
      "grad_norm": 3.151628255844116,
      "learning_rate": 4.622566911350971e-06,
      "loss": 0.6008,
      "step": 59290
    },
    {
      "epoch": 24.27,
      "grad_norm": 2.6473677158355713,
      "learning_rate": 4.622529825472999e-06,
      "loss": 0.5806,
      "step": 59300
    },
    {
      "epoch": 24.28,
      "grad_norm": 2.2567055225372314,
      "learning_rate": 4.622492729156289e-06,
      "loss": 0.5935,
      "step": 59310
    },
    {
      "epoch": 24.28,
      "grad_norm": 2.9607839584350586,
      "learning_rate": 4.622455622401012e-06,
      "loss": 0.5884,
      "step": 59320
    },
    {
      "epoch": 24.29,
      "grad_norm": 3.2368712425231934,
      "learning_rate": 4.622418505207337e-06,
      "loss": 0.5648,
      "step": 59330
    },
    {
      "epoch": 24.29,
      "grad_norm": 2.7787365913391113,
      "learning_rate": 4.622381377575434e-06,
      "loss": 0.5895,
      "step": 59340
    },
    {
      "epoch": 24.29,
      "grad_norm": 2.9993062019348145,
      "learning_rate": 4.622344239505474e-06,
      "loss": 0.589,
      "step": 59350
    },
    {
      "epoch": 24.3,
      "grad_norm": 2.498741865158081,
      "learning_rate": 4.622307090997628e-06,
      "loss": 0.6127,
      "step": 59360
    },
    {
      "epoch": 24.3,
      "grad_norm": 2.4091663360595703,
      "learning_rate": 4.622269932052063e-06,
      "loss": 0.6065,
      "step": 59370
    },
    {
      "epoch": 24.31,
      "grad_norm": 2.453223943710327,
      "learning_rate": 4.622232762668952e-06,
      "loss": 0.5882,
      "step": 59380
    },
    {
      "epoch": 24.31,
      "grad_norm": 3.225386142730713,
      "learning_rate": 4.622195582848464e-06,
      "loss": 0.585,
      "step": 59390
    },
    {
      "epoch": 24.31,
      "grad_norm": 2.6273341178894043,
      "learning_rate": 4.62215839259077e-06,
      "loss": 0.5945,
      "step": 59400
    },
    {
      "epoch": 24.32,
      "grad_norm": 2.7760045528411865,
      "learning_rate": 4.62212119189604e-06,
      "loss": 0.5766,
      "step": 59410
    },
    {
      "epoch": 24.32,
      "grad_norm": 2.356285572052002,
      "learning_rate": 4.622083980764444e-06,
      "loss": 0.6076,
      "step": 59420
    },
    {
      "epoch": 24.33,
      "grad_norm": 2.633500337600708,
      "learning_rate": 4.622046759196153e-06,
      "loss": 0.6176,
      "step": 59430
    },
    {
      "epoch": 24.33,
      "grad_norm": 2.497912883758545,
      "learning_rate": 4.6220095271913374e-06,
      "loss": 0.6102,
      "step": 59440
    },
    {
      "epoch": 24.33,
      "grad_norm": 2.715207815170288,
      "learning_rate": 4.621972284750168e-06,
      "loss": 0.5811,
      "step": 59450
    },
    {
      "epoch": 24.34,
      "grad_norm": 2.3064496517181396,
      "learning_rate": 4.621935031872815e-06,
      "loss": 0.6055,
      "step": 59460
    },
    {
      "epoch": 24.34,
      "grad_norm": 2.7833781242370605,
      "learning_rate": 4.6218977685594484e-06,
      "loss": 0.5875,
      "step": 59470
    },
    {
      "epoch": 24.35,
      "grad_norm": 2.3166675567626953,
      "learning_rate": 4.621860494810241e-06,
      "loss": 0.5934,
      "step": 59480
    },
    {
      "epoch": 24.35,
      "grad_norm": 3.1481094360351562,
      "learning_rate": 4.6218232106253615e-06,
      "loss": 0.5796,
      "step": 59490
    },
    {
      "epoch": 24.36,
      "grad_norm": 2.0138473510742188,
      "learning_rate": 4.621785916004981e-06,
      "loss": 0.581,
      "step": 59500
    },
    {
      "epoch": 24.36,
      "grad_norm": 2.556485652923584,
      "learning_rate": 4.6217486109492725e-06,
      "loss": 0.5795,
      "step": 59510
    },
    {
      "epoch": 24.36,
      "grad_norm": 3.13653564453125,
      "learning_rate": 4.621711295458403e-06,
      "loss": 0.5987,
      "step": 59520
    },
    {
      "epoch": 24.37,
      "grad_norm": 2.505686044692993,
      "learning_rate": 4.621673969532547e-06,
      "loss": 0.6213,
      "step": 59530
    },
    {
      "epoch": 24.37,
      "grad_norm": 2.061741590499878,
      "learning_rate": 4.621636633171873e-06,
      "loss": 0.5844,
      "step": 59540
    },
    {
      "epoch": 24.38,
      "grad_norm": 2.974437952041626,
      "learning_rate": 4.6215992863765545e-06,
      "loss": 0.5687,
      "step": 59550
    },
    {
      "epoch": 24.38,
      "grad_norm": 2.347961664199829,
      "learning_rate": 4.6215619291467605e-06,
      "loss": 0.5716,
      "step": 59560
    },
    {
      "epoch": 24.38,
      "grad_norm": 1.9422404766082764,
      "learning_rate": 4.6215245614826624e-06,
      "loss": 0.5927,
      "step": 59570
    },
    {
      "epoch": 24.39,
      "grad_norm": 2.438627004623413,
      "learning_rate": 4.621487183384432e-06,
      "loss": 0.598,
      "step": 59580
    },
    {
      "epoch": 24.39,
      "grad_norm": 2.2336819171905518,
      "learning_rate": 4.621449794852241e-06,
      "loss": 0.5658,
      "step": 59590
    },
    {
      "epoch": 24.4,
      "grad_norm": 2.1614139080047607,
      "learning_rate": 4.6214123958862596e-06,
      "loss": 0.5869,
      "step": 59600
    },
    {
      "epoch": 24.4,
      "grad_norm": 2.787630796432495,
      "learning_rate": 4.6213749864866586e-06,
      "loss": 0.5767,
      "step": 59610
    },
    {
      "epoch": 24.4,
      "grad_norm": 2.978520631790161,
      "learning_rate": 4.6213375666536115e-06,
      "loss": 0.6051,
      "step": 59620
    },
    {
      "epoch": 24.41,
      "grad_norm": 2.6746840476989746,
      "learning_rate": 4.621300136387289e-06,
      "loss": 0.5882,
      "step": 59630
    },
    {
      "epoch": 24.41,
      "grad_norm": 2.451915979385376,
      "learning_rate": 4.62126269568786e-06,
      "loss": 0.593,
      "step": 59640
    },
    {
      "epoch": 24.42,
      "grad_norm": 2.862623691558838,
      "learning_rate": 4.621225244555499e-06,
      "loss": 0.6102,
      "step": 59650
    },
    {
      "epoch": 24.42,
      "grad_norm": 3.3883986473083496,
      "learning_rate": 4.621187782990377e-06,
      "loss": 0.5884,
      "step": 59660
    },
    {
      "epoch": 24.42,
      "grad_norm": 2.507455587387085,
      "learning_rate": 4.621150310992665e-06,
      "loss": 0.5753,
      "step": 59670
    },
    {
      "epoch": 24.43,
      "grad_norm": 3.1415345668792725,
      "learning_rate": 4.621112828562535e-06,
      "loss": 0.5454,
      "step": 59680
    },
    {
      "epoch": 24.43,
      "grad_norm": 2.137523889541626,
      "learning_rate": 4.621075335700159e-06,
      "loss": 0.5818,
      "step": 59690
    },
    {
      "epoch": 24.44,
      "grad_norm": 2.5328781604766846,
      "learning_rate": 4.621037832405707e-06,
      "loss": 0.5731,
      "step": 59700
    },
    {
      "epoch": 24.44,
      "grad_norm": 2.767507314682007,
      "learning_rate": 4.621000318679353e-06,
      "loss": 0.5784,
      "step": 59710
    },
    {
      "epoch": 24.45,
      "grad_norm": 2.5330801010131836,
      "learning_rate": 4.620962794521268e-06,
      "loss": 0.6178,
      "step": 59720
    },
    {
      "epoch": 24.45,
      "grad_norm": 2.0280425548553467,
      "learning_rate": 4.620925259931623e-06,
      "loss": 0.6016,
      "step": 59730
    },
    {
      "epoch": 24.45,
      "grad_norm": 3.0479001998901367,
      "learning_rate": 4.620887714910592e-06,
      "loss": 0.6131,
      "step": 59740
    },
    {
      "epoch": 24.46,
      "grad_norm": 3.364246368408203,
      "learning_rate": 4.620850159458345e-06,
      "loss": 0.6075,
      "step": 59750
    },
    {
      "epoch": 24.46,
      "grad_norm": 1.9128164052963257,
      "learning_rate": 4.620812593575055e-06,
      "loss": 0.5789,
      "step": 59760
    },
    {
      "epoch": 24.47,
      "grad_norm": 2.7415778636932373,
      "learning_rate": 4.620775017260895e-06,
      "loss": 0.5907,
      "step": 59770
    },
    {
      "epoch": 24.47,
      "grad_norm": 3.434145212173462,
      "learning_rate": 4.620737430516035e-06,
      "loss": 0.5774,
      "step": 59780
    },
    {
      "epoch": 24.47,
      "grad_norm": 2.570983648300171,
      "learning_rate": 4.620699833340648e-06,
      "loss": 0.5886,
      "step": 59790
    },
    {
      "epoch": 24.48,
      "grad_norm": 2.3521618843078613,
      "learning_rate": 4.620662225734906e-06,
      "loss": 0.5846,
      "step": 59800
    },
    {
      "epoch": 24.48,
      "grad_norm": 2.295978307723999,
      "learning_rate": 4.620624607698983e-06,
      "loss": 0.5963,
      "step": 59810
    },
    {
      "epoch": 24.49,
      "grad_norm": 2.879103422164917,
      "learning_rate": 4.620586979233049e-06,
      "loss": 0.5678,
      "step": 59820
    },
    {
      "epoch": 24.49,
      "grad_norm": 3.003852367401123,
      "learning_rate": 4.620549340337278e-06,
      "loss": 0.5988,
      "step": 59830
    },
    {
      "epoch": 24.49,
      "grad_norm": 2.6616551876068115,
      "learning_rate": 4.620511691011842e-06,
      "loss": 0.596,
      "step": 59840
    },
    {
      "epoch": 24.5,
      "grad_norm": 2.4249424934387207,
      "learning_rate": 4.620474031256913e-06,
      "loss": 0.6202,
      "step": 59850
    },
    {
      "epoch": 24.5,
      "grad_norm": 3.057983160018921,
      "learning_rate": 4.6204363610726635e-06,
      "loss": 0.59,
      "step": 59860
    },
    {
      "epoch": 24.51,
      "grad_norm": 2.223569393157959,
      "learning_rate": 4.6203986804592665e-06,
      "loss": 0.5883,
      "step": 59870
    },
    {
      "epoch": 24.51,
      "grad_norm": 2.531914234161377,
      "learning_rate": 4.620360989416895e-06,
      "loss": 0.6003,
      "step": 59880
    },
    {
      "epoch": 24.51,
      "grad_norm": 2.5798146724700928,
      "learning_rate": 4.620323287945721e-06,
      "loss": 0.5963,
      "step": 59890
    },
    {
      "epoch": 24.52,
      "grad_norm": 2.0895633697509766,
      "learning_rate": 4.620285576045917e-06,
      "loss": 0.599,
      "step": 59900
    },
    {
      "epoch": 24.52,
      "grad_norm": 3.0898501873016357,
      "learning_rate": 4.6202478537176555e-06,
      "loss": 0.5867,
      "step": 59910
    },
    {
      "epoch": 24.53,
      "grad_norm": 2.1018593311309814,
      "learning_rate": 4.620210120961112e-06,
      "loss": 0.5879,
      "step": 59920
    },
    {
      "epoch": 24.53,
      "grad_norm": 2.9652960300445557,
      "learning_rate": 4.620172377776455e-06,
      "loss": 0.5879,
      "step": 59930
    },
    {
      "epoch": 24.54,
      "grad_norm": 3.1914217472076416,
      "learning_rate": 4.6201346241638605e-06,
      "loss": 0.6151,
      "step": 59940
    },
    {
      "epoch": 24.54,
      "grad_norm": 2.049639940261841,
      "learning_rate": 4.620096860123501e-06,
      "loss": 0.5834,
      "step": 59950
    },
    {
      "epoch": 24.54,
      "grad_norm": 2.701951026916504,
      "learning_rate": 4.620059085655548e-06,
      "loss": 0.5911,
      "step": 59960
    },
    {
      "epoch": 24.55,
      "grad_norm": 1.7881492376327515,
      "learning_rate": 4.620021300760176e-06,
      "loss": 0.5926,
      "step": 59970
    },
    {
      "epoch": 24.55,
      "grad_norm": 2.022913694381714,
      "learning_rate": 4.619983505437559e-06,
      "loss": 0.595,
      "step": 59980
    },
    {
      "epoch": 24.56,
      "grad_norm": 1.9755744934082031,
      "learning_rate": 4.619945699687867e-06,
      "loss": 0.5838,
      "step": 59990
    },
    {
      "epoch": 24.56,
      "grad_norm": 2.8454339504241943,
      "learning_rate": 4.619907883511276e-06,
      "loss": 0.5734,
      "step": 60000
    },
    {
      "epoch": 24.56,
      "grad_norm": 3.333676338195801,
      "learning_rate": 4.619870056907958e-06,
      "loss": 0.58,
      "step": 60010
    },
    {
      "epoch": 24.57,
      "grad_norm": 2.392169237136841,
      "learning_rate": 4.619832219878086e-06,
      "loss": 0.5992,
      "step": 60020
    },
    {
      "epoch": 24.57,
      "grad_norm": 2.1105518341064453,
      "learning_rate": 4.619794372421835e-06,
      "loss": 0.6018,
      "step": 60030
    },
    {
      "epoch": 24.58,
      "grad_norm": 2.7320759296417236,
      "learning_rate": 4.619756514539377e-06,
      "loss": 0.5869,
      "step": 60040
    },
    {
      "epoch": 24.58,
      "grad_norm": 2.1205437183380127,
      "learning_rate": 4.619718646230885e-06,
      "loss": 0.5842,
      "step": 60050
    },
    {
      "epoch": 24.58,
      "grad_norm": 2.941098928451538,
      "learning_rate": 4.6196807674965335e-06,
      "loss": 0.5874,
      "step": 60060
    },
    {
      "epoch": 24.59,
      "grad_norm": 2.8787946701049805,
      "learning_rate": 4.619642878336495e-06,
      "loss": 0.5765,
      "step": 60070
    },
    {
      "epoch": 24.59,
      "grad_norm": 2.097790479660034,
      "learning_rate": 4.6196049787509445e-06,
      "loss": 0.5822,
      "step": 60080
    },
    {
      "epoch": 24.6,
      "grad_norm": 2.0615875720977783,
      "learning_rate": 4.6195670687400546e-06,
      "loss": 0.6014,
      "step": 60090
    },
    {
      "epoch": 24.6,
      "grad_norm": 2.1377265453338623,
      "learning_rate": 4.619529148303999e-06,
      "loss": 0.5807,
      "step": 60100
    },
    {
      "epoch": 24.6,
      "grad_norm": 2.4315319061279297,
      "learning_rate": 4.619491217442951e-06,
      "loss": 0.5892,
      "step": 60110
    },
    {
      "epoch": 24.61,
      "grad_norm": 2.742938756942749,
      "learning_rate": 4.619453276157086e-06,
      "loss": 0.599,
      "step": 60120
    },
    {
      "epoch": 24.61,
      "grad_norm": 3.027212619781494,
      "learning_rate": 4.619415324446576e-06,
      "loss": 0.6126,
      "step": 60130
    },
    {
      "epoch": 24.62,
      "grad_norm": 2.6414339542388916,
      "learning_rate": 4.619377362311596e-06,
      "loss": 0.5997,
      "step": 60140
    },
    {
      "epoch": 24.62,
      "grad_norm": 3.23581600189209,
      "learning_rate": 4.61933938975232e-06,
      "loss": 0.5841,
      "step": 60150
    },
    {
      "epoch": 24.63,
      "grad_norm": 2.273327589035034,
      "learning_rate": 4.619301406768921e-06,
      "loss": 0.5859,
      "step": 60160
    },
    {
      "epoch": 24.63,
      "grad_norm": 2.993046283721924,
      "learning_rate": 4.619263413361572e-06,
      "loss": 0.5982,
      "step": 60170
    },
    {
      "epoch": 24.63,
      "grad_norm": 2.254730224609375,
      "learning_rate": 4.61922540953045e-06,
      "loss": 0.5884,
      "step": 60180
    },
    {
      "epoch": 24.64,
      "grad_norm": 3.0333783626556396,
      "learning_rate": 4.619187395275728e-06,
      "loss": 0.5982,
      "step": 60190
    },
    {
      "epoch": 24.64,
      "grad_norm": 3.9217875003814697,
      "learning_rate": 4.619149370597578e-06,
      "loss": 0.5851,
      "step": 60200
    },
    {
      "epoch": 24.65,
      "grad_norm": 2.3492283821105957,
      "learning_rate": 4.619111335496177e-06,
      "loss": 0.5819,
      "step": 60210
    },
    {
      "epoch": 24.65,
      "grad_norm": 2.5748801231384277,
      "learning_rate": 4.619073289971698e-06,
      "loss": 0.59,
      "step": 60220
    },
    {
      "epoch": 24.65,
      "grad_norm": 2.5514535903930664,
      "learning_rate": 4.619035234024316e-06,
      "loss": 0.5963,
      "step": 60230
    },
    {
      "epoch": 24.66,
      "grad_norm": 3.3274853229522705,
      "learning_rate": 4.618997167654203e-06,
      "loss": 0.5795,
      "step": 60240
    },
    {
      "epoch": 24.66,
      "grad_norm": 2.4126648902893066,
      "learning_rate": 4.6189590908615365e-06,
      "loss": 0.5903,
      "step": 60250
    },
    {
      "epoch": 24.67,
      "grad_norm": 2.0635669231414795,
      "learning_rate": 4.618921003646489e-06,
      "loss": 0.574,
      "step": 60260
    },
    {
      "epoch": 24.67,
      "grad_norm": 3.6609725952148438,
      "learning_rate": 4.618882906009236e-06,
      "loss": 0.5975,
      "step": 60270
    },
    {
      "epoch": 24.67,
      "grad_norm": 2.068117141723633,
      "learning_rate": 4.61884479794995e-06,
      "loss": 0.5973,
      "step": 60280
    },
    {
      "epoch": 24.68,
      "grad_norm": 2.6246724128723145,
      "learning_rate": 4.618806679468808e-06,
      "loss": 0.6004,
      "step": 60290
    },
    {
      "epoch": 24.68,
      "grad_norm": 3.4470627307891846,
      "learning_rate": 4.618768550565984e-06,
      "loss": 0.5962,
      "step": 60300
    },
    {
      "epoch": 24.69,
      "grad_norm": 2.43068528175354,
      "learning_rate": 4.618730411241652e-06,
      "loss": 0.6076,
      "step": 60310
    },
    {
      "epoch": 24.69,
      "grad_norm": 2.6587727069854736,
      "learning_rate": 4.618692261495988e-06,
      "loss": 0.5772,
      "step": 60320
    },
    {
      "epoch": 24.7,
      "grad_norm": 2.696446418762207,
      "learning_rate": 4.618654101329164e-06,
      "loss": 0.5869,
      "step": 60330
    },
    {
      "epoch": 24.7,
      "grad_norm": 2.8625240325927734,
      "learning_rate": 4.618615930741358e-06,
      "loss": 0.5816,
      "step": 60340
    },
    {
      "epoch": 24.7,
      "grad_norm": 1.8537873029708862,
      "learning_rate": 4.618577749732744e-06,
      "loss": 0.6051,
      "step": 60350
    },
    {
      "epoch": 24.71,
      "grad_norm": 2.9984660148620605,
      "learning_rate": 4.618539558303495e-06,
      "loss": 0.6033,
      "step": 60360
    },
    {
      "epoch": 24.71,
      "grad_norm": 2.7449464797973633,
      "learning_rate": 4.618501356453788e-06,
      "loss": 0.5776,
      "step": 60370
    },
    {
      "epoch": 24.72,
      "grad_norm": 3.1936233043670654,
      "learning_rate": 4.618463144183797e-06,
      "loss": 0.5821,
      "step": 60380
    },
    {
      "epoch": 24.72,
      "grad_norm": 2.0530223846435547,
      "learning_rate": 4.618424921493698e-06,
      "loss": 0.6105,
      "step": 60390
    },
    {
      "epoch": 24.72,
      "grad_norm": 2.2763495445251465,
      "learning_rate": 4.618386688383665e-06,
      "loss": 0.6021,
      "step": 60400
    },
    {
      "epoch": 24.73,
      "grad_norm": 3.343837261199951,
      "learning_rate": 4.618348444853874e-06,
      "loss": 0.5922,
      "step": 60410
    },
    {
      "epoch": 24.73,
      "grad_norm": 3.050760507583618,
      "learning_rate": 4.618310190904499e-06,
      "loss": 0.5842,
      "step": 60420
    },
    {
      "epoch": 24.74,
      "grad_norm": 2.1529974937438965,
      "learning_rate": 4.618271926535716e-06,
      "loss": 0.5912,
      "step": 60430
    },
    {
      "epoch": 24.74,
      "grad_norm": 2.343078851699829,
      "learning_rate": 4.618233651747701e-06,
      "loss": 0.5817,
      "step": 60440
    },
    {
      "epoch": 24.74,
      "grad_norm": 2.9840550422668457,
      "learning_rate": 4.6181953665406285e-06,
      "loss": 0.598,
      "step": 60450
    },
    {
      "epoch": 24.75,
      "grad_norm": 2.4751760959625244,
      "learning_rate": 4.618157070914674e-06,
      "loss": 0.5921,
      "step": 60460
    },
    {
      "epoch": 24.75,
      "grad_norm": 2.231354236602783,
      "learning_rate": 4.618118764870013e-06,
      "loss": 0.5965,
      "step": 60470
    },
    {
      "epoch": 24.76,
      "grad_norm": 1.9974365234375,
      "learning_rate": 4.618080448406821e-06,
      "loss": 0.6037,
      "step": 60480
    },
    {
      "epoch": 24.76,
      "grad_norm": 2.928788661956787,
      "learning_rate": 4.618042121525273e-06,
      "loss": 0.5901,
      "step": 60490
    },
    {
      "epoch": 24.76,
      "grad_norm": 2.006270408630371,
      "learning_rate": 4.618003784225544e-06,
      "loss": 0.5903,
      "step": 60500
    },
    {
      "epoch": 24.77,
      "grad_norm": 2.499990940093994,
      "learning_rate": 4.617965436507813e-06,
      "loss": 0.5945,
      "step": 60510
    },
    {
      "epoch": 24.77,
      "grad_norm": 2.2611989974975586,
      "learning_rate": 4.617927078372252e-06,
      "loss": 0.5902,
      "step": 60520
    },
    {
      "epoch": 24.78,
      "grad_norm": 4.371289253234863,
      "learning_rate": 4.6178887098190375e-06,
      "loss": 0.6169,
      "step": 60530
    },
    {
      "epoch": 24.78,
      "grad_norm": 3.832425117492676,
      "learning_rate": 4.617850330848347e-06,
      "loss": 0.5953,
      "step": 60540
    },
    {
      "epoch": 24.79,
      "grad_norm": 2.3898866176605225,
      "learning_rate": 4.617811941460354e-06,
      "loss": 0.5735,
      "step": 60550
    },
    {
      "epoch": 24.79,
      "grad_norm": 2.024714708328247,
      "learning_rate": 4.6177735416552364e-06,
      "loss": 0.5897,
      "step": 60560
    },
    {
      "epoch": 24.79,
      "grad_norm": 3.0678839683532715,
      "learning_rate": 4.617735131433169e-06,
      "loss": 0.5958,
      "step": 60570
    },
    {
      "epoch": 24.8,
      "grad_norm": 2.3101513385772705,
      "learning_rate": 4.617696710794327e-06,
      "loss": 0.5873,
      "step": 60580
    },
    {
      "epoch": 24.8,
      "grad_norm": 2.6491382122039795,
      "learning_rate": 4.617658279738888e-06,
      "loss": 0.6216,
      "step": 60590
    },
    {
      "epoch": 24.81,
      "grad_norm": 2.614271402359009,
      "learning_rate": 4.617619838267028e-06,
      "loss": 0.6127,
      "step": 60600
    },
    {
      "epoch": 24.81,
      "grad_norm": 2.4808437824249268,
      "learning_rate": 4.617581386378921e-06,
      "loss": 0.6008,
      "step": 60610
    },
    {
      "epoch": 24.81,
      "grad_norm": 2.1062002182006836,
      "learning_rate": 4.617542924074744e-06,
      "loss": 0.591,
      "step": 60620
    },
    {
      "epoch": 24.82,
      "grad_norm": 2.9176387786865234,
      "learning_rate": 4.617504451354675e-06,
      "loss": 0.589,
      "step": 60630
    },
    {
      "epoch": 24.82,
      "grad_norm": 2.5076217651367188,
      "learning_rate": 4.61746596821889e-06,
      "loss": 0.5927,
      "step": 60640
    },
    {
      "epoch": 24.83,
      "grad_norm": 1.9077818393707275,
      "learning_rate": 4.617427474667562e-06,
      "loss": 0.582,
      "step": 60650
    },
    {
      "epoch": 24.83,
      "grad_norm": 2.5623257160186768,
      "learning_rate": 4.617388970700871e-06,
      "loss": 0.5919,
      "step": 60660
    },
    {
      "epoch": 24.83,
      "grad_norm": 2.9730770587921143,
      "learning_rate": 4.6173504563189925e-06,
      "loss": 0.5987,
      "step": 60670
    },
    {
      "epoch": 24.84,
      "grad_norm": 2.580195665359497,
      "learning_rate": 4.617311931522102e-06,
      "loss": 0.5966,
      "step": 60680
    },
    {
      "epoch": 24.84,
      "grad_norm": 2.3164889812469482,
      "learning_rate": 4.617273396310375e-06,
      "loss": 0.595,
      "step": 60690
    },
    {
      "epoch": 24.85,
      "grad_norm": 3.0161209106445312,
      "learning_rate": 4.617234850683991e-06,
      "loss": 0.5975,
      "step": 60700
    },
    {
      "epoch": 24.85,
      "grad_norm": 3.0693280696868896,
      "learning_rate": 4.617196294643124e-06,
      "loss": 0.5853,
      "step": 60710
    },
    {
      "epoch": 24.85,
      "grad_norm": 1.9950121641159058,
      "learning_rate": 4.617157728187953e-06,
      "loss": 0.5924,
      "step": 60720
    },
    {
      "epoch": 24.86,
      "grad_norm": 2.630095958709717,
      "learning_rate": 4.617119151318652e-06,
      "loss": 0.6038,
      "step": 60730
    },
    {
      "epoch": 24.86,
      "grad_norm": 2.6172800064086914,
      "learning_rate": 4.6170805640354e-06,
      "loss": 0.5912,
      "step": 60740
    },
    {
      "epoch": 24.87,
      "grad_norm": 2.3822600841522217,
      "learning_rate": 4.617041966338372e-06,
      "loss": 0.5877,
      "step": 60750
    },
    {
      "epoch": 24.87,
      "grad_norm": 3.1806507110595703,
      "learning_rate": 4.617003358227745e-06,
      "loss": 0.5888,
      "step": 60760
    },
    {
      "epoch": 24.88,
      "grad_norm": 3.090963125228882,
      "learning_rate": 4.616964739703697e-06,
      "loss": 0.6009,
      "step": 60770
    },
    {
      "epoch": 24.88,
      "grad_norm": 3.384922504425049,
      "learning_rate": 4.616926110766405e-06,
      "loss": 0.6157,
      "step": 60780
    },
    {
      "epoch": 24.88,
      "grad_norm": 4.050862789154053,
      "learning_rate": 4.6168874714160446e-06,
      "loss": 0.5952,
      "step": 60790
    },
    {
      "epoch": 24.89,
      "grad_norm": 3.2825357913970947,
      "learning_rate": 4.6168488216527934e-06,
      "loss": 0.5876,
      "step": 60800
    },
    {
      "epoch": 24.89,
      "grad_norm": 2.134111166000366,
      "learning_rate": 4.616810161476829e-06,
      "loss": 0.5841,
      "step": 60810
    },
    {
      "epoch": 24.9,
      "grad_norm": 2.106740951538086,
      "learning_rate": 4.616771490888328e-06,
      "loss": 0.5885,
      "step": 60820
    },
    {
      "epoch": 24.9,
      "grad_norm": 2.653394937515259,
      "learning_rate": 4.616732809887467e-06,
      "loss": 0.5887,
      "step": 60830
    },
    {
      "epoch": 24.9,
      "grad_norm": 3.1689863204956055,
      "learning_rate": 4.616694118474424e-06,
      "loss": 0.5815,
      "step": 60840
    },
    {
      "epoch": 24.91,
      "grad_norm": 2.3374595642089844,
      "learning_rate": 4.616655416649376e-06,
      "loss": 0.6029,
      "step": 60850
    },
    {
      "epoch": 24.91,
      "grad_norm": 2.3459384441375732,
      "learning_rate": 4.6166167044125e-06,
      "loss": 0.5819,
      "step": 60860
    },
    {
      "epoch": 24.92,
      "grad_norm": 2.4091312885284424,
      "learning_rate": 4.616577981763975e-06,
      "loss": 0.5791,
      "step": 60870
    },
    {
      "epoch": 24.92,
      "grad_norm": 2.4326114654541016,
      "learning_rate": 4.6165392487039765e-06,
      "loss": 0.5969,
      "step": 60880
    },
    {
      "epoch": 24.92,
      "grad_norm": 2.506054162979126,
      "learning_rate": 4.616500505232682e-06,
      "loss": 0.6025,
      "step": 60890
    },
    {
      "epoch": 24.93,
      "grad_norm": 2.689523220062256,
      "learning_rate": 4.616461751350269e-06,
      "loss": 0.6043,
      "step": 60900
    },
    {
      "epoch": 24.93,
      "grad_norm": 3.178133726119995,
      "learning_rate": 4.6164229870569156e-06,
      "loss": 0.5934,
      "step": 60910
    },
    {
      "epoch": 24.94,
      "grad_norm": 2.708256721496582,
      "learning_rate": 4.6163842123528e-06,
      "loss": 0.6143,
      "step": 60920
    },
    {
      "epoch": 24.94,
      "grad_norm": 2.85685133934021,
      "learning_rate": 4.616345427238098e-06,
      "loss": 0.5901,
      "step": 60930
    },
    {
      "epoch": 24.94,
      "grad_norm": 2.7533111572265625,
      "learning_rate": 4.616306631712989e-06,
      "loss": 0.5885,
      "step": 60940
    },
    {
      "epoch": 24.95,
      "grad_norm": 1.9881963729858398,
      "learning_rate": 4.61626782577765e-06,
      "loss": 0.5954,
      "step": 60950
    },
    {
      "epoch": 24.95,
      "grad_norm": 2.4435391426086426,
      "learning_rate": 4.6162290094322575e-06,
      "loss": 0.5976,
      "step": 60960
    },
    {
      "epoch": 24.96,
      "grad_norm": 1.7041678428649902,
      "learning_rate": 4.616190182676992e-06,
      "loss": 0.5935,
      "step": 60970
    },
    {
      "epoch": 24.96,
      "grad_norm": 2.694366931915283,
      "learning_rate": 4.61615134551203e-06,
      "loss": 0.5916,
      "step": 60980
    },
    {
      "epoch": 24.97,
      "grad_norm": 3.438218593597412,
      "learning_rate": 4.616112497937549e-06,
      "loss": 0.5977,
      "step": 60990
    },
    {
      "epoch": 24.97,
      "grad_norm": 2.5274531841278076,
      "learning_rate": 4.616073639953727e-06,
      "loss": 0.5975,
      "step": 61000
    },
    {
      "epoch": 24.97,
      "grad_norm": 2.626952648162842,
      "learning_rate": 4.616034771560743e-06,
      "loss": 0.5975,
      "step": 61010
    },
    {
      "epoch": 24.98,
      "grad_norm": 3.425762891769409,
      "learning_rate": 4.615995892758774e-06,
      "loss": 0.5932,
      "step": 61020
    },
    {
      "epoch": 24.98,
      "grad_norm": 1.902329444885254,
      "learning_rate": 4.6159570035479985e-06,
      "loss": 0.597,
      "step": 61030
    },
    {
      "epoch": 24.99,
      "grad_norm": 2.539626121520996,
      "learning_rate": 4.615918103928594e-06,
      "loss": 0.5728,
      "step": 61040
    },
    {
      "epoch": 24.99,
      "grad_norm": 2.0318517684936523,
      "learning_rate": 4.61587919390074e-06,
      "loss": 0.588,
      "step": 61050
    },
    {
      "epoch": 24.99,
      "grad_norm": 2.4079043865203857,
      "learning_rate": 4.615840273464614e-06,
      "loss": 0.6058,
      "step": 61060
    },
    {
      "epoch": 25.0,
      "grad_norm": 2.122607946395874,
      "learning_rate": 4.615801342620395e-06,
      "loss": 0.5886,
      "step": 61070
    },
    {
      "epoch": 25.0,
      "eval_loss": 0.5882992148399353,
      "eval_runtime": 52.5198,
      "eval_samples_per_second": 65.67,
      "eval_steps_per_second": 8.225,
      "step": 61075
    },
    {
      "epoch": 25.0,
      "grad_norm": 2.5972743034362793,
      "learning_rate": 4.61576240136826e-06,
      "loss": 0.5992,
      "step": 61080
    },
    {
      "epoch": 25.01,
      "grad_norm": 2.340330123901367,
      "learning_rate": 4.615723449708388e-06,
      "loss": 0.5958,
      "step": 61090
    },
    {
      "epoch": 25.01,
      "grad_norm": 2.1801486015319824,
      "learning_rate": 4.615684487640957e-06,
      "loss": 0.5987,
      "step": 61100
    },
    {
      "epoch": 25.01,
      "grad_norm": 3.0071356296539307,
      "learning_rate": 4.615645515166147e-06,
      "loss": 0.6098,
      "step": 61110
    },
    {
      "epoch": 25.02,
      "grad_norm": 2.7998850345611572,
      "learning_rate": 4.6156065322841345e-06,
      "loss": 0.577,
      "step": 61120
    },
    {
      "epoch": 25.02,
      "grad_norm": 2.0403735637664795,
      "learning_rate": 4.6155675389950995e-06,
      "loss": 0.5684,
      "step": 61130
    },
    {
      "epoch": 25.03,
      "grad_norm": 2.943721294403076,
      "learning_rate": 4.6155285352992194e-06,
      "loss": 0.5775,
      "step": 61140
    },
    {
      "epoch": 25.03,
      "grad_norm": 2.168879747390747,
      "learning_rate": 4.615489521196676e-06,
      "loss": 0.5807,
      "step": 61150
    },
    {
      "epoch": 25.03,
      "grad_norm": 2.820164442062378,
      "learning_rate": 4.6154504966876435e-06,
      "loss": 0.5882,
      "step": 61160
    },
    {
      "epoch": 25.04,
      "grad_norm": 2.774272918701172,
      "learning_rate": 4.615411461772303e-06,
      "loss": 0.5748,
      "step": 61170
    },
    {
      "epoch": 25.04,
      "grad_norm": 2.4273011684417725,
      "learning_rate": 4.615372416450834e-06,
      "loss": 0.5876,
      "step": 61180
    },
    {
      "epoch": 25.05,
      "grad_norm": 2.6968417167663574,
      "learning_rate": 4.615333360723414e-06,
      "loss": 0.5887,
      "step": 61190
    },
    {
      "epoch": 25.05,
      "grad_norm": 3.19333553314209,
      "learning_rate": 4.615294294590222e-06,
      "loss": 0.5974,
      "step": 61200
    },
    {
      "epoch": 25.06,
      "grad_norm": 3.2561397552490234,
      "learning_rate": 4.615255218051439e-06,
      "loss": 0.5902,
      "step": 61210
    },
    {
      "epoch": 25.06,
      "grad_norm": 2.363438367843628,
      "learning_rate": 4.615216131107241e-06,
      "loss": 0.5991,
      "step": 61220
    },
    {
      "epoch": 25.06,
      "grad_norm": 2.693844795227051,
      "learning_rate": 4.6151770337578085e-06,
      "loss": 0.5944,
      "step": 61230
    },
    {
      "epoch": 25.07,
      "grad_norm": 2.1269326210021973,
      "learning_rate": 4.615137926003321e-06,
      "loss": 0.6067,
      "step": 61240
    },
    {
      "epoch": 25.07,
      "grad_norm": 1.9544354677200317,
      "learning_rate": 4.615098807843957e-06,
      "loss": 0.5904,
      "step": 61250
    },
    {
      "epoch": 25.08,
      "grad_norm": 2.3966195583343506,
      "learning_rate": 4.615059679279896e-06,
      "loss": 0.6048,
      "step": 61260
    },
    {
      "epoch": 25.08,
      "grad_norm": 2.746993064880371,
      "learning_rate": 4.615020540311317e-06,
      "loss": 0.5762,
      "step": 61270
    },
    {
      "epoch": 25.08,
      "grad_norm": 1.991735577583313,
      "learning_rate": 4.614981390938399e-06,
      "loss": 0.6072,
      "step": 61280
    },
    {
      "epoch": 25.09,
      "grad_norm": 2.6618900299072266,
      "learning_rate": 4.614942231161322e-06,
      "loss": 0.5581,
      "step": 61290
    },
    {
      "epoch": 25.09,
      "grad_norm": 2.0728650093078613,
      "learning_rate": 4.6149030609802655e-06,
      "loss": 0.5802,
      "step": 61300
    },
    {
      "epoch": 25.1,
      "grad_norm": 2.356198787689209,
      "learning_rate": 4.6148638803954084e-06,
      "loss": 0.6026,
      "step": 61310
    },
    {
      "epoch": 25.1,
      "grad_norm": 2.386166572570801,
      "learning_rate": 4.614824689406929e-06,
      "loss": 0.5712,
      "step": 61320
    },
    {
      "epoch": 25.1,
      "grad_norm": 2.400510311126709,
      "learning_rate": 4.61478548801501e-06,
      "loss": 0.5935,
      "step": 61330
    },
    {
      "epoch": 25.11,
      "grad_norm": 2.5655953884124756,
      "learning_rate": 4.614746276219829e-06,
      "loss": 0.5833,
      "step": 61340
    },
    {
      "epoch": 25.11,
      "grad_norm": 2.729006052017212,
      "learning_rate": 4.614707054021565e-06,
      "loss": 0.6106,
      "step": 61350
    },
    {
      "epoch": 25.12,
      "grad_norm": 2.2389678955078125,
      "learning_rate": 4.6146678214203995e-06,
      "loss": 0.6005,
      "step": 61360
    },
    {
      "epoch": 25.12,
      "grad_norm": 2.7689504623413086,
      "learning_rate": 4.61462857841651e-06,
      "loss": 0.5916,
      "step": 61370
    },
    {
      "epoch": 25.12,
      "grad_norm": 2.70379900932312,
      "learning_rate": 4.614589325010078e-06,
      "loss": 0.5973,
      "step": 61380
    },
    {
      "epoch": 25.13,
      "grad_norm": 2.263287305831909,
      "learning_rate": 4.614550061201283e-06,
      "loss": 0.5825,
      "step": 61390
    },
    {
      "epoch": 25.13,
      "grad_norm": 3.1361498832702637,
      "learning_rate": 4.614510786990304e-06,
      "loss": 0.5752,
      "step": 61400
    },
    {
      "epoch": 25.14,
      "grad_norm": 2.3847813606262207,
      "learning_rate": 4.614471502377322e-06,
      "loss": 0.5929,
      "step": 61410
    },
    {
      "epoch": 25.14,
      "grad_norm": 2.6983819007873535,
      "learning_rate": 4.6144322073625156e-06,
      "loss": 0.592,
      "step": 61420
    },
    {
      "epoch": 25.15,
      "grad_norm": 2.6080617904663086,
      "learning_rate": 4.614392901946066e-06,
      "loss": 0.6014,
      "step": 61430
    },
    {
      "epoch": 25.15,
      "grad_norm": 2.8546481132507324,
      "learning_rate": 4.614353586128153e-06,
      "loss": 0.593,
      "step": 61440
    },
    {
      "epoch": 25.15,
      "grad_norm": 2.041088819503784,
      "learning_rate": 4.614314259908957e-06,
      "loss": 0.5964,
      "step": 61450
    },
    {
      "epoch": 25.16,
      "grad_norm": 2.530186176300049,
      "learning_rate": 4.614274923288658e-06,
      "loss": 0.5974,
      "step": 61460
    },
    {
      "epoch": 25.16,
      "grad_norm": 2.531038284301758,
      "learning_rate": 4.614235576267435e-06,
      "loss": 0.5794,
      "step": 61470
    },
    {
      "epoch": 25.17,
      "grad_norm": 2.4917547702789307,
      "learning_rate": 4.61419621884547e-06,
      "loss": 0.5799,
      "step": 61480
    },
    {
      "epoch": 25.17,
      "grad_norm": 2.371901750564575,
      "learning_rate": 4.614156851022942e-06,
      "loss": 0.5911,
      "step": 61490
    },
    {
      "epoch": 25.17,
      "grad_norm": 2.5410168170928955,
      "learning_rate": 4.614117472800031e-06,
      "loss": 0.5971,
      "step": 61500
    },
    {
      "epoch": 25.18,
      "grad_norm": 2.0495212078094482,
      "learning_rate": 4.61407808417692e-06,
      "loss": 0.5788,
      "step": 61510
    },
    {
      "epoch": 25.18,
      "grad_norm": 2.1840686798095703,
      "learning_rate": 4.614038685153786e-06,
      "loss": 0.5991,
      "step": 61520
    },
    {
      "epoch": 25.19,
      "grad_norm": 2.256408929824829,
      "learning_rate": 4.613999275730813e-06,
      "loss": 0.5815,
      "step": 61530
    },
    {
      "epoch": 25.19,
      "grad_norm": 3.14123797416687,
      "learning_rate": 4.6139598559081785e-06,
      "loss": 0.576,
      "step": 61540
    },
    {
      "epoch": 25.19,
      "grad_norm": 2.984753131866455,
      "learning_rate": 4.613920425686064e-06,
      "loss": 0.5971,
      "step": 61550
    },
    {
      "epoch": 25.2,
      "grad_norm": 2.526759386062622,
      "learning_rate": 4.613880985064651e-06,
      "loss": 0.5948,
      "step": 61560
    },
    {
      "epoch": 25.2,
      "grad_norm": 2.9606246948242188,
      "learning_rate": 4.61384153404412e-06,
      "loss": 0.5926,
      "step": 61570
    },
    {
      "epoch": 25.21,
      "grad_norm": 1.9397469758987427,
      "learning_rate": 4.61380207262465e-06,
      "loss": 0.6038,
      "step": 61580
    },
    {
      "epoch": 25.21,
      "grad_norm": 2.2343688011169434,
      "learning_rate": 4.613762600806424e-06,
      "loss": 0.5846,
      "step": 61590
    },
    {
      "epoch": 25.21,
      "grad_norm": 2.6819558143615723,
      "learning_rate": 4.613723118589621e-06,
      "loss": 0.6073,
      "step": 61600
    },
    {
      "epoch": 25.22,
      "grad_norm": 2.7830731868743896,
      "learning_rate": 4.613683625974423e-06,
      "loss": 0.6169,
      "step": 61610
    },
    {
      "epoch": 25.22,
      "grad_norm": 2.5561716556549072,
      "learning_rate": 4.6136441229610105e-06,
      "loss": 0.5822,
      "step": 61620
    },
    {
      "epoch": 25.23,
      "grad_norm": 2.3185927867889404,
      "learning_rate": 4.613604609549565e-06,
      "loss": 0.5909,
      "step": 61630
    },
    {
      "epoch": 25.23,
      "grad_norm": 3.181248426437378,
      "learning_rate": 4.613565085740267e-06,
      "loss": 0.5864,
      "step": 61640
    },
    {
      "epoch": 25.24,
      "grad_norm": 2.0754029750823975,
      "learning_rate": 4.613525551533297e-06,
      "loss": 0.6194,
      "step": 61650
    },
    {
      "epoch": 25.24,
      "grad_norm": 2.5085501670837402,
      "learning_rate": 4.613486006928838e-06,
      "loss": 0.6073,
      "step": 61660
    },
    {
      "epoch": 25.24,
      "grad_norm": 2.840548515319824,
      "learning_rate": 4.613446451927068e-06,
      "loss": 0.6122,
      "step": 61670
    },
    {
      "epoch": 25.25,
      "grad_norm": 2.0484426021575928,
      "learning_rate": 4.613406886528172e-06,
      "loss": 0.5835,
      "step": 61680
    },
    {
      "epoch": 25.25,
      "grad_norm": 2.06144118309021,
      "learning_rate": 4.613367310732328e-06,
      "loss": 0.5918,
      "step": 61690
    },
    {
      "epoch": 25.26,
      "grad_norm": 2.3780007362365723,
      "learning_rate": 4.613327724539719e-06,
      "loss": 0.5943,
      "step": 61700
    },
    {
      "epoch": 25.26,
      "grad_norm": 2.360013246536255,
      "learning_rate": 4.613288127950526e-06,
      "loss": 0.5977,
      "step": 61710
    },
    {
      "epoch": 25.26,
      "grad_norm": 2.5979321002960205,
      "learning_rate": 4.61324852096493e-06,
      "loss": 0.6033,
      "step": 61720
    },
    {
      "epoch": 25.27,
      "grad_norm": 1.7557597160339355,
      "learning_rate": 4.613208903583113e-06,
      "loss": 0.5848,
      "step": 61730
    },
    {
      "epoch": 25.27,
      "grad_norm": 2.461954355239868,
      "learning_rate": 4.613169275805256e-06,
      "loss": 0.5855,
      "step": 61740
    },
    {
      "epoch": 25.28,
      "grad_norm": 2.6721513271331787,
      "learning_rate": 4.613129637631541e-06,
      "loss": 0.5897,
      "step": 61750
    },
    {
      "epoch": 25.28,
      "grad_norm": 2.1154329776763916,
      "learning_rate": 4.613089989062149e-06,
      "loss": 0.5936,
      "step": 61760
    },
    {
      "epoch": 25.28,
      "grad_norm": 3.272127628326416,
      "learning_rate": 4.613050330097262e-06,
      "loss": 0.5847,
      "step": 61770
    },
    {
      "epoch": 25.29,
      "grad_norm": 2.039609909057617,
      "learning_rate": 4.6130106607370625e-06,
      "loss": 0.5873,
      "step": 61780
    },
    {
      "epoch": 25.29,
      "grad_norm": 2.414293050765991,
      "learning_rate": 4.612970980981731e-06,
      "loss": 0.5854,
      "step": 61790
    },
    {
      "epoch": 25.3,
      "grad_norm": 2.155630350112915,
      "learning_rate": 4.612931290831449e-06,
      "loss": 0.5814,
      "step": 61800
    },
    {
      "epoch": 25.3,
      "grad_norm": 1.7207211256027222,
      "learning_rate": 4.6128915902864e-06,
      "loss": 0.5845,
      "step": 61810
    },
    {
      "epoch": 25.3,
      "grad_norm": 2.160749673843384,
      "learning_rate": 4.612851879346763e-06,
      "loss": 0.5837,
      "step": 61820
    },
    {
      "epoch": 25.31,
      "grad_norm": 2.560330390930176,
      "learning_rate": 4.612812158012724e-06,
      "loss": 0.595,
      "step": 61830
    },
    {
      "epoch": 25.31,
      "grad_norm": 2.044217824935913,
      "learning_rate": 4.612772426284461e-06,
      "loss": 0.5838,
      "step": 61840
    },
    {
      "epoch": 25.32,
      "grad_norm": 2.4569270610809326,
      "learning_rate": 4.612732684162158e-06,
      "loss": 0.5846,
      "step": 61850
    },
    {
      "epoch": 25.32,
      "grad_norm": 2.0554864406585693,
      "learning_rate": 4.612692931645997e-06,
      "loss": 0.5911,
      "step": 61860
    },
    {
      "epoch": 25.33,
      "grad_norm": 2.162343978881836,
      "learning_rate": 4.61265316873616e-06,
      "loss": 0.6017,
      "step": 61870
    },
    {
      "epoch": 25.33,
      "grad_norm": 2.2674810886383057,
      "learning_rate": 4.612613395432829e-06,
      "loss": 0.602,
      "step": 61880
    },
    {
      "epoch": 25.33,
      "grad_norm": 2.4041759967803955,
      "learning_rate": 4.612573611736185e-06,
      "loss": 0.5743,
      "step": 61890
    },
    {
      "epoch": 25.34,
      "grad_norm": 2.2678604125976562,
      "learning_rate": 4.612533817646412e-06,
      "loss": 0.6025,
      "step": 61900
    },
    {
      "epoch": 25.34,
      "grad_norm": 3.6557846069335938,
      "learning_rate": 4.612494013163692e-06,
      "loss": 0.577,
      "step": 61910
    },
    {
      "epoch": 25.35,
      "grad_norm": 2.1348698139190674,
      "learning_rate": 4.612454198288207e-06,
      "loss": 0.6136,
      "step": 61920
    },
    {
      "epoch": 25.35,
      "grad_norm": 2.5947322845458984,
      "learning_rate": 4.6124143730201395e-06,
      "loss": 0.6085,
      "step": 61930
    },
    {
      "epoch": 25.35,
      "grad_norm": 2.3352880477905273,
      "learning_rate": 4.612374537359672e-06,
      "loss": 0.5837,
      "step": 61940
    },
    {
      "epoch": 25.36,
      "grad_norm": 2.1992974281311035,
      "learning_rate": 4.612334691306986e-06,
      "loss": 0.5937,
      "step": 61950
    },
    {
      "epoch": 25.36,
      "grad_norm": 2.5591225624084473,
      "learning_rate": 4.6122948348622655e-06,
      "loss": 0.5532,
      "step": 61960
    },
    {
      "epoch": 25.37,
      "grad_norm": 2.5843753814697266,
      "learning_rate": 4.612254968025692e-06,
      "loss": 0.6197,
      "step": 61970
    },
    {
      "epoch": 25.37,
      "grad_norm": 2.306248426437378,
      "learning_rate": 4.612215090797449e-06,
      "loss": 0.5985,
      "step": 61980
    },
    {
      "epoch": 25.37,
      "grad_norm": 2.371432304382324,
      "learning_rate": 4.612175203177719e-06,
      "loss": 0.5887,
      "step": 61990
    },
    {
      "epoch": 25.38,
      "grad_norm": 2.426820755004883,
      "learning_rate": 4.612135305166684e-06,
      "loss": 0.5877,
      "step": 62000
    },
    {
      "epoch": 25.38,
      "grad_norm": 2.275012493133545,
      "learning_rate": 4.612095396764528e-06,
      "loss": 0.5779,
      "step": 62010
    },
    {
      "epoch": 25.39,
      "grad_norm": 2.7757718563079834,
      "learning_rate": 4.612055477971432e-06,
      "loss": 0.597,
      "step": 62020
    },
    {
      "epoch": 25.39,
      "grad_norm": 3.043597459793091,
      "learning_rate": 4.61201554878758e-06,
      "loss": 0.5943,
      "step": 62030
    },
    {
      "epoch": 25.4,
      "grad_norm": 2.282487392425537,
      "learning_rate": 4.611975609213155e-06,
      "loss": 0.5835,
      "step": 62040
    },
    {
      "epoch": 25.4,
      "grad_norm": 2.0716476440429688,
      "learning_rate": 4.6119356592483404e-06,
      "loss": 0.5987,
      "step": 62050
    },
    {
      "epoch": 25.4,
      "grad_norm": 2.6177127361297607,
      "learning_rate": 4.6118956988933175e-06,
      "loss": 0.5876,
      "step": 62060
    },
    {
      "epoch": 25.41,
      "grad_norm": 3.1176440715789795,
      "learning_rate": 4.611855728148271e-06,
      "loss": 0.583,
      "step": 62070
    },
    {
      "epoch": 25.41,
      "grad_norm": 3.043649911880493,
      "learning_rate": 4.611815747013383e-06,
      "loss": 0.5919,
      "step": 62080
    },
    {
      "epoch": 25.42,
      "grad_norm": 3.4783661365509033,
      "learning_rate": 4.611775755488838e-06,
      "loss": 0.5993,
      "step": 62090
    },
    {
      "epoch": 25.42,
      "grad_norm": 2.8220064640045166,
      "learning_rate": 4.6117357535748175e-06,
      "loss": 0.6121,
      "step": 62100
    },
    {
      "epoch": 25.42,
      "grad_norm": 2.738837957382202,
      "learning_rate": 4.611695741271505e-06,
      "loss": 0.6087,
      "step": 62110
    },
    {
      "epoch": 25.43,
      "grad_norm": 3.0741875171661377,
      "learning_rate": 4.611655718579086e-06,
      "loss": 0.6126,
      "step": 62120
    },
    {
      "epoch": 25.43,
      "grad_norm": 3.5591561794281006,
      "learning_rate": 4.611615685497741e-06,
      "loss": 0.5811,
      "step": 62130
    },
    {
      "epoch": 25.44,
      "grad_norm": 1.8928683996200562,
      "learning_rate": 4.611575642027654e-06,
      "loss": 0.5814,
      "step": 62140
    },
    {
      "epoch": 25.44,
      "grad_norm": 2.164926052093506,
      "learning_rate": 4.61153558816901e-06,
      "loss": 0.6041,
      "step": 62150
    },
    {
      "epoch": 25.44,
      "grad_norm": 2.726032018661499,
      "learning_rate": 4.611495523921991e-06,
      "loss": 0.6081,
      "step": 62160
    },
    {
      "epoch": 25.45,
      "grad_norm": 3.3877182006835938,
      "learning_rate": 4.611455449286781e-06,
      "loss": 0.5744,
      "step": 62170
    },
    {
      "epoch": 25.45,
      "grad_norm": 2.3457181453704834,
      "learning_rate": 4.6114153642635635e-06,
      "loss": 0.5956,
      "step": 62180
    },
    {
      "epoch": 25.46,
      "grad_norm": 1.9302353858947754,
      "learning_rate": 4.611375268852522e-06,
      "loss": 0.598,
      "step": 62190
    },
    {
      "epoch": 25.46,
      "grad_norm": 2.091045618057251,
      "learning_rate": 4.611335163053841e-06,
      "loss": 0.5793,
      "step": 62200
    },
    {
      "epoch": 25.46,
      "grad_norm": 2.8769686222076416,
      "learning_rate": 4.6112950468677025e-06,
      "loss": 0.5925,
      "step": 62210
    },
    {
      "epoch": 25.47,
      "grad_norm": 2.350794553756714,
      "learning_rate": 4.611254920294292e-06,
      "loss": 0.5785,
      "step": 62220
    },
    {
      "epoch": 25.47,
      "grad_norm": 2.6117517948150635,
      "learning_rate": 4.611214783333793e-06,
      "loss": 0.5828,
      "step": 62230
    },
    {
      "epoch": 25.48,
      "grad_norm": 2.2483112812042236,
      "learning_rate": 4.611174635986387e-06,
      "loss": 0.5925,
      "step": 62240
    },
    {
      "epoch": 25.48,
      "grad_norm": 2.2800521850585938,
      "learning_rate": 4.611134478252262e-06,
      "loss": 0.5947,
      "step": 62250
    },
    {
      "epoch": 25.49,
      "grad_norm": 1.9778516292572021,
      "learning_rate": 4.611094310131599e-06,
      "loss": 0.5955,
      "step": 62260
    },
    {
      "epoch": 25.49,
      "grad_norm": 2.4535374641418457,
      "learning_rate": 4.611054131624584e-06,
      "loss": 0.5858,
      "step": 62270
    },
    {
      "epoch": 25.49,
      "grad_norm": 3.1881537437438965,
      "learning_rate": 4.611013942731399e-06,
      "loss": 0.6043,
      "step": 62280
    },
    {
      "epoch": 25.5,
      "grad_norm": 2.4916975498199463,
      "learning_rate": 4.610973743452229e-06,
      "loss": 0.5841,
      "step": 62290
    },
    {
      "epoch": 25.5,
      "grad_norm": 2.5947799682617188,
      "learning_rate": 4.610933533787258e-06,
      "loss": 0.5843,
      "step": 62300
    },
    {
      "epoch": 25.51,
      "grad_norm": 2.557464361190796,
      "learning_rate": 4.6108933137366704e-06,
      "loss": 0.5878,
      "step": 62310
    },
    {
      "epoch": 25.51,
      "grad_norm": 2.3597965240478516,
      "learning_rate": 4.6108530833006505e-06,
      "loss": 0.5937,
      "step": 62320
    },
    {
      "epoch": 25.51,
      "grad_norm": 2.8698980808258057,
      "learning_rate": 4.610812842479382e-06,
      "loss": 0.6146,
      "step": 62330
    },
    {
      "epoch": 25.52,
      "grad_norm": 2.698808431625366,
      "learning_rate": 4.6107725912730506e-06,
      "loss": 0.5836,
      "step": 62340
    },
    {
      "epoch": 25.52,
      "grad_norm": 2.413623809814453,
      "learning_rate": 4.610732329681839e-06,
      "loss": 0.6097,
      "step": 62350
    },
    {
      "epoch": 25.53,
      "grad_norm": 2.436450958251953,
      "learning_rate": 4.6106920577059326e-06,
      "loss": 0.6044,
      "step": 62360
    },
    {
      "epoch": 25.53,
      "grad_norm": 2.888434886932373,
      "learning_rate": 4.6106517753455164e-06,
      "loss": 0.5871,
      "step": 62370
    },
    {
      "epoch": 25.53,
      "grad_norm": 2.1017391681671143,
      "learning_rate": 4.610611482600774e-06,
      "loss": 0.5523,
      "step": 62380
    },
    {
      "epoch": 25.54,
      "grad_norm": 2.0830631256103516,
      "learning_rate": 4.61057117947189e-06,
      "loss": 0.5902,
      "step": 62390
    },
    {
      "epoch": 25.54,
      "grad_norm": 2.7333202362060547,
      "learning_rate": 4.6105308659590494e-06,
      "loss": 0.6055,
      "step": 62400
    },
    {
      "epoch": 25.55,
      "grad_norm": 3.3696439266204834,
      "learning_rate": 4.6104905420624364e-06,
      "loss": 0.5941,
      "step": 62410
    },
    {
      "epoch": 25.55,
      "grad_norm": 2.7859227657318115,
      "learning_rate": 4.610450207782235e-06,
      "loss": 0.5953,
      "step": 62420
    },
    {
      "epoch": 25.55,
      "grad_norm": 1.9506361484527588,
      "learning_rate": 4.610409863118633e-06,
      "loss": 0.6173,
      "step": 62430
    },
    {
      "epoch": 25.56,
      "grad_norm": 2.6838443279266357,
      "learning_rate": 4.610369508071812e-06,
      "loss": 0.6026,
      "step": 62440
    },
    {
      "epoch": 25.56,
      "grad_norm": 2.981332778930664,
      "learning_rate": 4.610329142641958e-06,
      "loss": 0.586,
      "step": 62450
    },
    {
      "epoch": 25.57,
      "grad_norm": 2.1863036155700684,
      "learning_rate": 4.610288766829257e-06,
      "loss": 0.5732,
      "step": 62460
    },
    {
      "epoch": 25.57,
      "grad_norm": 2.5644116401672363,
      "learning_rate": 4.6102483806338924e-06,
      "loss": 0.5736,
      "step": 62470
    },
    {
      "epoch": 25.58,
      "grad_norm": 1.8476924896240234,
      "learning_rate": 4.6102079840560495e-06,
      "loss": 0.58,
      "step": 62480
    },
    {
      "epoch": 25.58,
      "grad_norm": 2.7948405742645264,
      "learning_rate": 4.610167577095914e-06,
      "loss": 0.5947,
      "step": 62490
    },
    {
      "epoch": 25.58,
      "grad_norm": 2.1081786155700684,
      "learning_rate": 4.6101271597536705e-06,
      "loss": 0.5821,
      "step": 62500
    },
    {
      "epoch": 25.59,
      "grad_norm": 2.3648340702056885,
      "learning_rate": 4.610086732029504e-06,
      "loss": 0.596,
      "step": 62510
    },
    {
      "epoch": 25.59,
      "grad_norm": 2.3509609699249268,
      "learning_rate": 4.6100462939236e-06,
      "loss": 0.565,
      "step": 62520
    },
    {
      "epoch": 25.6,
      "grad_norm": 2.182117223739624,
      "learning_rate": 4.610005845436143e-06,
      "loss": 0.5787,
      "step": 62530
    },
    {
      "epoch": 25.6,
      "grad_norm": 3.5247442722320557,
      "learning_rate": 4.60996538656732e-06,
      "loss": 0.5869,
      "step": 62540
    },
    {
      "epoch": 25.6,
      "grad_norm": 2.8042869567871094,
      "learning_rate": 4.609924917317315e-06,
      "loss": 0.5684,
      "step": 62550
    },
    {
      "epoch": 25.61,
      "grad_norm": 2.1299796104431152,
      "learning_rate": 4.609884437686313e-06,
      "loss": 0.5844,
      "step": 62560
    },
    {
      "epoch": 25.61,
      "grad_norm": 2.1889307498931885,
      "learning_rate": 4.6098439476745006e-06,
      "loss": 0.6041,
      "step": 62570
    },
    {
      "epoch": 25.62,
      "grad_norm": 2.430565118789673,
      "learning_rate": 4.609803447282063e-06,
      "loss": 0.5704,
      "step": 62580
    },
    {
      "epoch": 25.62,
      "grad_norm": 2.8273732662200928,
      "learning_rate": 4.609762936509185e-06,
      "loss": 0.583,
      "step": 62590
    },
    {
      "epoch": 25.62,
      "grad_norm": 2.455731153488159,
      "learning_rate": 4.6097224153560535e-06,
      "loss": 0.5869,
      "step": 62600
    },
    {
      "epoch": 25.63,
      "grad_norm": 2.815732479095459,
      "learning_rate": 4.609681883822852e-06,
      "loss": 0.597,
      "step": 62610
    },
    {
      "epoch": 25.63,
      "grad_norm": 2.760115385055542,
      "learning_rate": 4.609641341909768e-06,
      "loss": 0.5827,
      "step": 62620
    },
    {
      "epoch": 25.64,
      "grad_norm": 2.6002509593963623,
      "learning_rate": 4.609600789616987e-06,
      "loss": 0.5747,
      "step": 62630
    },
    {
      "epoch": 25.64,
      "grad_norm": 2.3618836402893066,
      "learning_rate": 4.609560226944694e-06,
      "loss": 0.5925,
      "step": 62640
    },
    {
      "epoch": 25.64,
      "grad_norm": 2.8703370094299316,
      "learning_rate": 4.6095196538930765e-06,
      "loss": 0.5874,
      "step": 62650
    },
    {
      "epoch": 25.65,
      "grad_norm": 2.2947850227355957,
      "learning_rate": 4.609479070462318e-06,
      "loss": 0.6113,
      "step": 62660
    },
    {
      "epoch": 25.65,
      "grad_norm": 2.0403425693511963,
      "learning_rate": 4.6094384766526055e-06,
      "loss": 0.5756,
      "step": 62670
    },
    {
      "epoch": 25.66,
      "grad_norm": 3.1516900062561035,
      "learning_rate": 4.609397872464126e-06,
      "loss": 0.5769,
      "step": 62680
    },
    {
      "epoch": 25.66,
      "grad_norm": 3.208814859390259,
      "learning_rate": 4.609357257897063e-06,
      "loss": 0.5888,
      "step": 62690
    },
    {
      "epoch": 25.67,
      "grad_norm": 2.412358045578003,
      "learning_rate": 4.6093166329516045e-06,
      "loss": 0.5891,
      "step": 62700
    },
    {
      "epoch": 25.67,
      "grad_norm": 2.8993184566497803,
      "learning_rate": 4.609275997627936e-06,
      "loss": 0.5878,
      "step": 62710
    },
    {
      "epoch": 25.67,
      "grad_norm": 2.391162633895874,
      "learning_rate": 4.609235351926245e-06,
      "loss": 0.6098,
      "step": 62720
    },
    {
      "epoch": 25.68,
      "grad_norm": 1.8684790134429932,
      "learning_rate": 4.609194695846715e-06,
      "loss": 0.5643,
      "step": 62730
    },
    {
      "epoch": 25.68,
      "grad_norm": 2.9994421005249023,
      "learning_rate": 4.609154029389534e-06,
      "loss": 0.5771,
      "step": 62740
    },
    {
      "epoch": 25.69,
      "grad_norm": 2.336073398590088,
      "learning_rate": 4.6091133525548885e-06,
      "loss": 0.6083,
      "step": 62750
    },
    {
      "epoch": 25.69,
      "grad_norm": 2.035984992980957,
      "learning_rate": 4.609072665342964e-06,
      "loss": 0.5679,
      "step": 62760
    },
    {
      "epoch": 25.69,
      "grad_norm": 1.8894591331481934,
      "learning_rate": 4.609031967753947e-06,
      "loss": 0.5911,
      "step": 62770
    },
    {
      "epoch": 25.7,
      "grad_norm": 2.4973363876342773,
      "learning_rate": 4.608991259788024e-06,
      "loss": 0.5874,
      "step": 62780
    },
    {
      "epoch": 25.7,
      "grad_norm": 2.7023253440856934,
      "learning_rate": 4.608950541445383e-06,
      "loss": 0.5738,
      "step": 62790
    },
    {
      "epoch": 25.71,
      "grad_norm": 2.031989336013794,
      "learning_rate": 4.6089098127262076e-06,
      "loss": 0.5638,
      "step": 62800
    },
    {
      "epoch": 25.71,
      "grad_norm": 2.1473634243011475,
      "learning_rate": 4.608869073630686e-06,
      "loss": 0.5697,
      "step": 62810
    },
    {
      "epoch": 25.71,
      "grad_norm": 2.5727639198303223,
      "learning_rate": 4.6088283241590055e-06,
      "loss": 0.5809,
      "step": 62820
    },
    {
      "epoch": 25.72,
      "grad_norm": 2.384099006652832,
      "learning_rate": 4.608787564311352e-06,
      "loss": 0.6171,
      "step": 62830
    },
    {
      "epoch": 25.72,
      "grad_norm": 2.006639242172241,
      "learning_rate": 4.6087467940879124e-06,
      "loss": 0.5864,
      "step": 62840
    },
    {
      "epoch": 25.73,
      "grad_norm": 2.6616616249084473,
      "learning_rate": 4.608706013488872e-06,
      "loss": 0.5857,
      "step": 62850
    },
    {
      "epoch": 25.73,
      "grad_norm": 2.0416860580444336,
      "learning_rate": 4.608665222514419e-06,
      "loss": 0.588,
      "step": 62860
    },
    {
      "epoch": 25.73,
      "grad_norm": 2.049234390258789,
      "learning_rate": 4.608624421164741e-06,
      "loss": 0.6102,
      "step": 62870
    },
    {
      "epoch": 25.74,
      "grad_norm": 2.5861809253692627,
      "learning_rate": 4.608583609440024e-06,
      "loss": 0.5664,
      "step": 62880
    },
    {
      "epoch": 25.74,
      "grad_norm": 2.7331933975219727,
      "learning_rate": 4.608542787340454e-06,
      "loss": 0.5906,
      "step": 62890
    },
    {
      "epoch": 25.75,
      "grad_norm": 2.1915175914764404,
      "learning_rate": 4.60850195486622e-06,
      "loss": 0.5871,
      "step": 62900
    },
    {
      "epoch": 25.75,
      "grad_norm": 2.3223612308502197,
      "learning_rate": 4.608461112017508e-06,
      "loss": 0.5593,
      "step": 62910
    },
    {
      "epoch": 25.76,
      "grad_norm": 2.2991557121276855,
      "learning_rate": 4.608420258794505e-06,
      "loss": 0.5972,
      "step": 62920
    },
    {
      "epoch": 25.76,
      "grad_norm": 2.376861810684204,
      "learning_rate": 4.608379395197398e-06,
      "loss": 0.5638,
      "step": 62930
    },
    {
      "epoch": 25.76,
      "grad_norm": 3.1873531341552734,
      "learning_rate": 4.6083385212263745e-06,
      "loss": 0.5906,
      "step": 62940
    },
    {
      "epoch": 25.77,
      "grad_norm": 3.188281774520874,
      "learning_rate": 4.608297636881622e-06,
      "loss": 0.5916,
      "step": 62950
    },
    {
      "epoch": 25.77,
      "grad_norm": 2.3781981468200684,
      "learning_rate": 4.608256742163328e-06,
      "loss": 0.5861,
      "step": 62960
    },
    {
      "epoch": 25.78,
      "grad_norm": 2.8018243312835693,
      "learning_rate": 4.608215837071678e-06,
      "loss": 0.5881,
      "step": 62970
    },
    {
      "epoch": 25.78,
      "grad_norm": 2.9095895290374756,
      "learning_rate": 4.608174921606861e-06,
      "loss": 0.6003,
      "step": 62980
    },
    {
      "epoch": 25.78,
      "grad_norm": 2.308727502822876,
      "learning_rate": 4.608133995769064e-06,
      "loss": 0.5714,
      "step": 62990
    },
    {
      "epoch": 25.79,
      "grad_norm": 2.664405584335327,
      "learning_rate": 4.608093059558475e-06,
      "loss": 0.5798,
      "step": 63000
    },
    {
      "epoch": 25.79,
      "grad_norm": 2.1189868450164795,
      "learning_rate": 4.608052112975281e-06,
      "loss": 0.5788,
      "step": 63010
    },
    {
      "epoch": 25.8,
      "grad_norm": 3.4241561889648438,
      "learning_rate": 4.608011156019671e-06,
      "loss": 0.6123,
      "step": 63020
    },
    {
      "epoch": 25.8,
      "grad_norm": 1.8407684564590454,
      "learning_rate": 4.60797018869183e-06,
      "loss": 0.5947,
      "step": 63030
    },
    {
      "epoch": 25.8,
      "grad_norm": 2.6460299491882324,
      "learning_rate": 4.607929210991947e-06,
      "loss": 0.5669,
      "step": 63040
    },
    {
      "epoch": 25.81,
      "grad_norm": 2.7086989879608154,
      "learning_rate": 4.60788822292021e-06,
      "loss": 0.5838,
      "step": 63050
    },
    {
      "epoch": 25.81,
      "grad_norm": 2.278975248336792,
      "learning_rate": 4.607847224476806e-06,
      "loss": 0.5584,
      "step": 63060
    },
    {
      "epoch": 25.82,
      "grad_norm": 2.940380096435547,
      "learning_rate": 4.607806215661923e-06,
      "loss": 0.5852,
      "step": 63070
    },
    {
      "epoch": 25.82,
      "grad_norm": 2.1899635791778564,
      "learning_rate": 4.60776519647575e-06,
      "loss": 0.5924,
      "step": 63080
    },
    {
      "epoch": 25.82,
      "grad_norm": 2.369157075881958,
      "learning_rate": 4.607724166918474e-06,
      "loss": 0.5778,
      "step": 63090
    },
    {
      "epoch": 25.83,
      "grad_norm": 2.2109620571136475,
      "learning_rate": 4.607683126990282e-06,
      "loss": 0.6176,
      "step": 63100
    },
    {
      "epoch": 25.83,
      "grad_norm": 2.274301290512085,
      "learning_rate": 4.607642076691364e-06,
      "loss": 0.5793,
      "step": 63110
    },
    {
      "epoch": 25.84,
      "grad_norm": 2.190276622772217,
      "learning_rate": 4.607601016021907e-06,
      "loss": 0.5895,
      "step": 63120
    },
    {
      "epoch": 25.84,
      "grad_norm": 2.3038291931152344,
      "learning_rate": 4.607559944982098e-06,
      "loss": 0.5865,
      "step": 63130
    },
    {
      "epoch": 25.85,
      "grad_norm": 2.5982532501220703,
      "learning_rate": 4.6075188635721266e-06,
      "loss": 0.5961,
      "step": 63140
    },
    {
      "epoch": 25.85,
      "grad_norm": 2.4895806312561035,
      "learning_rate": 4.607477771792181e-06,
      "loss": 0.6098,
      "step": 63150
    },
    {
      "epoch": 25.85,
      "grad_norm": 2.3597183227539062,
      "learning_rate": 4.607436669642449e-06,
      "loss": 0.6125,
      "step": 63160
    },
    {
      "epoch": 25.86,
      "grad_norm": 2.035334587097168,
      "learning_rate": 4.607395557123119e-06,
      "loss": 0.5847,
      "step": 63170
    },
    {
      "epoch": 25.86,
      "grad_norm": 3.345358371734619,
      "learning_rate": 4.60735443423438e-06,
      "loss": 0.5738,
      "step": 63180
    },
    {
      "epoch": 25.87,
      "grad_norm": 2.4131839275360107,
      "learning_rate": 4.607313300976418e-06,
      "loss": 0.575,
      "step": 63190
    },
    {
      "epoch": 25.87,
      "grad_norm": 2.682488203048706,
      "learning_rate": 4.6072721573494236e-06,
      "loss": 0.5993,
      "step": 63200
    },
    {
      "epoch": 25.87,
      "grad_norm": 2.7142746448516846,
      "learning_rate": 4.607231003353585e-06,
      "loss": 0.6039,
      "step": 63210
    },
    {
      "epoch": 25.88,
      "grad_norm": 2.653055191040039,
      "learning_rate": 4.607189838989091e-06,
      "loss": 0.5696,
      "step": 63220
    },
    {
      "epoch": 25.88,
      "grad_norm": 2.542630910873413,
      "learning_rate": 4.607148664256129e-06,
      "loss": 0.5815,
      "step": 63230
    },
    {
      "epoch": 25.89,
      "grad_norm": 2.772284507751465,
      "learning_rate": 4.6071074791548885e-06,
      "loss": 0.5803,
      "step": 63240
    },
    {
      "epoch": 25.89,
      "grad_norm": 3.278907537460327,
      "learning_rate": 4.607066283685558e-06,
      "loss": 0.596,
      "step": 63250
    },
    {
      "epoch": 25.89,
      "grad_norm": 2.2494349479675293,
      "learning_rate": 4.607025077848326e-06,
      "loss": 0.595,
      "step": 63260
    },
    {
      "epoch": 25.9,
      "grad_norm": 3.447662830352783,
      "learning_rate": 4.6069838616433816e-06,
      "loss": 0.5878,
      "step": 63270
    },
    {
      "epoch": 25.9,
      "grad_norm": 2.9573276042938232,
      "learning_rate": 4.606942635070912e-06,
      "loss": 0.5821,
      "step": 63280
    },
    {
      "epoch": 25.91,
      "grad_norm": 2.4100728034973145,
      "learning_rate": 4.606901398131109e-06,
      "loss": 0.5888,
      "step": 63290
    },
    {
      "epoch": 25.91,
      "grad_norm": 2.1388823986053467,
      "learning_rate": 4.60686015082416e-06,
      "loss": 0.5889,
      "step": 63300
    },
    {
      "epoch": 25.91,
      "grad_norm": 3.013948678970337,
      "learning_rate": 4.606818893150253e-06,
      "loss": 0.567,
      "step": 63310
    },
    {
      "epoch": 25.92,
      "grad_norm": 3.063791275024414,
      "learning_rate": 4.606777625109578e-06,
      "loss": 0.5923,
      "step": 63320
    },
    {
      "epoch": 25.92,
      "grad_norm": 3.0744502544403076,
      "learning_rate": 4.606736346702325e-06,
      "loss": 0.5833,
      "step": 63330
    },
    {
      "epoch": 25.93,
      "grad_norm": 2.1884543895721436,
      "learning_rate": 4.606695057928681e-06,
      "loss": 0.5801,
      "step": 63340
    },
    {
      "epoch": 25.93,
      "grad_norm": 2.314875841140747,
      "learning_rate": 4.606653758788836e-06,
      "loss": 0.583,
      "step": 63350
    },
    {
      "epoch": 25.94,
      "grad_norm": 4.364059925079346,
      "learning_rate": 4.60661244928298e-06,
      "loss": 0.5846,
      "step": 63360
    },
    {
      "epoch": 25.94,
      "grad_norm": 2.0088906288146973,
      "learning_rate": 4.606571129411301e-06,
      "loss": 0.6034,
      "step": 63370
    },
    {
      "epoch": 25.94,
      "grad_norm": 2.6557724475860596,
      "learning_rate": 4.606529799173989e-06,
      "loss": 0.601,
      "step": 63380
    },
    {
      "epoch": 25.95,
      "grad_norm": 2.2303051948547363,
      "learning_rate": 4.606488458571233e-06,
      "loss": 0.591,
      "step": 63390
    },
    {
      "epoch": 25.95,
      "grad_norm": 2.535505533218384,
      "learning_rate": 4.606447107603223e-06,
      "loss": 0.5904,
      "step": 63400
    },
    {
      "epoch": 25.96,
      "grad_norm": 2.5815935134887695,
      "learning_rate": 4.606405746270148e-06,
      "loss": 0.57,
      "step": 63410
    },
    {
      "epoch": 25.96,
      "grad_norm": 2.1493079662323,
      "learning_rate": 4.606364374572197e-06,
      "loss": 0.6018,
      "step": 63420
    },
    {
      "epoch": 25.96,
      "grad_norm": 2.400487184524536,
      "learning_rate": 4.6063229925095604e-06,
      "loss": 0.583,
      "step": 63430
    },
    {
      "epoch": 25.97,
      "grad_norm": 4.070574760437012,
      "learning_rate": 4.606281600082427e-06,
      "loss": 0.5834,
      "step": 63440
    },
    {
      "epoch": 25.97,
      "grad_norm": 2.306959390640259,
      "learning_rate": 4.606240197290987e-06,
      "loss": 0.5905,
      "step": 63450
    },
    {
      "epoch": 25.98,
      "grad_norm": 2.5929036140441895,
      "learning_rate": 4.60619878413543e-06,
      "loss": 0.5804,
      "step": 63460
    },
    {
      "epoch": 25.98,
      "grad_norm": 2.465999126434326,
      "learning_rate": 4.606157360615945e-06,
      "loss": 0.5734,
      "step": 63470
    },
    {
      "epoch": 25.98,
      "grad_norm": 2.3877053260803223,
      "learning_rate": 4.606115926732723e-06,
      "loss": 0.599,
      "step": 63480
    },
    {
      "epoch": 25.99,
      "grad_norm": 2.083092451095581,
      "learning_rate": 4.606074482485952e-06,
      "loss": 0.5765,
      "step": 63490
    },
    {
      "epoch": 25.99,
      "grad_norm": 3.7605881690979004,
      "learning_rate": 4.606033027875824e-06,
      "loss": 0.5815,
      "step": 63500
    },
    {
      "epoch": 26.0,
      "grad_norm": 3.042029619216919,
      "learning_rate": 4.6059915629025264e-06,
      "loss": 0.6028,
      "step": 63510
    },
    {
      "epoch": 26.0,
      "eval_loss": 0.5889604091644287,
      "eval_runtime": 51.8833,
      "eval_samples_per_second": 66.476,
      "eval_steps_per_second": 8.326,
      "step": 63518
    },
    {
      "epoch": 26.0,
      "grad_norm": 2.6577908992767334,
      "learning_rate": 4.605950087566251e-06,
      "loss": 0.5927,
      "step": 63520
    },
    {
      "epoch": 26.0,
      "grad_norm": 2.4132049083709717,
      "learning_rate": 4.605908601867188e-06,
      "loss": 0.5927,
      "step": 63530
    },
    {
      "epoch": 26.01,
      "grad_norm": 1.9549251794815063,
      "learning_rate": 4.6058671058055264e-06,
      "loss": 0.5838,
      "step": 63540
    },
    {
      "epoch": 26.01,
      "grad_norm": 2.0437827110290527,
      "learning_rate": 4.605825599381457e-06,
      "loss": 0.5961,
      "step": 63550
    },
    {
      "epoch": 26.02,
      "grad_norm": 2.9682188034057617,
      "learning_rate": 4.605784082595169e-06,
      "loss": 0.5878,
      "step": 63560
    },
    {
      "epoch": 26.02,
      "grad_norm": 2.3959288597106934,
      "learning_rate": 4.605742555446854e-06,
      "loss": 0.5893,
      "step": 63570
    },
    {
      "epoch": 26.03,
      "grad_norm": 3.150092840194702,
      "learning_rate": 4.605701017936701e-06,
      "loss": 0.5909,
      "step": 63580
    },
    {
      "epoch": 26.03,
      "grad_norm": 3.204085111618042,
      "learning_rate": 4.605659470064901e-06,
      "loss": 0.5901,
      "step": 63590
    },
    {
      "epoch": 26.03,
      "grad_norm": 3.2513551712036133,
      "learning_rate": 4.605617911831643e-06,
      "loss": 0.6007,
      "step": 63600
    },
    {
      "epoch": 26.04,
      "grad_norm": 2.1802706718444824,
      "learning_rate": 4.605576343237119e-06,
      "loss": 0.5789,
      "step": 63610
    },
    {
      "epoch": 26.04,
      "grad_norm": 2.542496681213379,
      "learning_rate": 4.605534764281519e-06,
      "loss": 0.5934,
      "step": 63620
    },
    {
      "epoch": 26.05,
      "grad_norm": 2.1428849697113037,
      "learning_rate": 4.605493174965033e-06,
      "loss": 0.5896,
      "step": 63630
    },
    {
      "epoch": 26.05,
      "grad_norm": 1.9358797073364258,
      "learning_rate": 4.605451575287852e-06,
      "loss": 0.5843,
      "step": 63640
    },
    {
      "epoch": 26.05,
      "grad_norm": 2.810445785522461,
      "learning_rate": 4.6054099652501666e-06,
      "loss": 0.5897,
      "step": 63650
    },
    {
      "epoch": 26.06,
      "grad_norm": 2.0990512371063232,
      "learning_rate": 4.605368344852167e-06,
      "loss": 0.6046,
      "step": 63660
    },
    {
      "epoch": 26.06,
      "grad_norm": 2.846411943435669,
      "learning_rate": 4.6053267140940434e-06,
      "loss": 0.5789,
      "step": 63670
    },
    {
      "epoch": 26.07,
      "grad_norm": 2.7333431243896484,
      "learning_rate": 4.605285072975988e-06,
      "loss": 0.5806,
      "step": 63680
    },
    {
      "epoch": 26.07,
      "grad_norm": 2.1549861431121826,
      "learning_rate": 4.60524342149819e-06,
      "loss": 0.6112,
      "step": 63690
    },
    {
      "epoch": 26.07,
      "grad_norm": 2.22004771232605,
      "learning_rate": 4.605201759660841e-06,
      "loss": 0.5988,
      "step": 63700
    },
    {
      "epoch": 26.08,
      "grad_norm": 3.037325620651245,
      "learning_rate": 4.605160087464132e-06,
      "loss": 0.5741,
      "step": 63710
    },
    {
      "epoch": 26.08,
      "grad_norm": 2.466583251953125,
      "learning_rate": 4.6051184049082535e-06,
      "loss": 0.5951,
      "step": 63720
    },
    {
      "epoch": 26.09,
      "grad_norm": 2.575087070465088,
      "learning_rate": 4.605076711993396e-06,
      "loss": 0.5736,
      "step": 63730
    },
    {
      "epoch": 26.09,
      "grad_norm": 1.8846720457077026,
      "learning_rate": 4.605035008719752e-06,
      "loss": 0.5953,
      "step": 63740
    },
    {
      "epoch": 26.09,
      "grad_norm": 2.018261194229126,
      "learning_rate": 4.604993295087512e-06,
      "loss": 0.574,
      "step": 63750
    },
    {
      "epoch": 26.1,
      "grad_norm": 2.887545108795166,
      "learning_rate": 4.6049515710968656e-06,
      "loss": 0.5815,
      "step": 63760
    },
    {
      "epoch": 26.1,
      "grad_norm": 2.0266971588134766,
      "learning_rate": 4.604909836748005e-06,
      "loss": 0.5614,
      "step": 63770
    },
    {
      "epoch": 26.11,
      "grad_norm": 2.6045494079589844,
      "learning_rate": 4.604868092041121e-06,
      "loss": 0.5962,
      "step": 63780
    },
    {
      "epoch": 26.11,
      "grad_norm": 2.2292821407318115,
      "learning_rate": 4.604826336976406e-06,
      "loss": 0.5895,
      "step": 63790
    },
    {
      "epoch": 26.12,
      "grad_norm": 2.5783045291900635,
      "learning_rate": 4.604784571554051e-06,
      "loss": 0.5894,
      "step": 63800
    },
    {
      "epoch": 26.12,
      "grad_norm": 2.2197320461273193,
      "learning_rate": 4.604742795774245e-06,
      "loss": 0.6074,
      "step": 63810
    },
    {
      "epoch": 26.12,
      "grad_norm": 2.5682787895202637,
      "learning_rate": 4.604701009637183e-06,
      "loss": 0.6037,
      "step": 63820
    },
    {
      "epoch": 26.13,
      "grad_norm": 2.3893613815307617,
      "learning_rate": 4.604659213143053e-06,
      "loss": 0.5837,
      "step": 63830
    },
    {
      "epoch": 26.13,
      "grad_norm": 2.32830810546875,
      "learning_rate": 4.604617406292049e-06,
      "loss": 0.5926,
      "step": 63840
    },
    {
      "epoch": 26.14,
      "grad_norm": 2.7925305366516113,
      "learning_rate": 4.604575589084362e-06,
      "loss": 0.5884,
      "step": 63850
    },
    {
      "epoch": 26.14,
      "grad_norm": 2.129794120788574,
      "learning_rate": 4.604533761520182e-06,
      "loss": 0.6066,
      "step": 63860
    },
    {
      "epoch": 26.14,
      "grad_norm": 2.743002414703369,
      "learning_rate": 4.604491923599702e-06,
      "loss": 0.5959,
      "step": 63870
    },
    {
      "epoch": 26.15,
      "grad_norm": 2.4530768394470215,
      "learning_rate": 4.604450075323114e-06,
      "loss": 0.5748,
      "step": 63880
    },
    {
      "epoch": 26.15,
      "grad_norm": 2.655290365219116,
      "learning_rate": 4.6044082166906085e-06,
      "loss": 0.5785,
      "step": 63890
    },
    {
      "epoch": 26.16,
      "grad_norm": 3.2886478900909424,
      "learning_rate": 4.604366347702379e-06,
      "loss": 0.589,
      "step": 63900
    },
    {
      "epoch": 26.16,
      "grad_norm": 3.2711548805236816,
      "learning_rate": 4.604324468358614e-06,
      "loss": 0.5953,
      "step": 63910
    },
    {
      "epoch": 26.16,
      "grad_norm": 3.12032413482666,
      "learning_rate": 4.6042825786595086e-06,
      "loss": 0.5896,
      "step": 63920
    },
    {
      "epoch": 26.17,
      "grad_norm": 3.0210673809051514,
      "learning_rate": 4.604240678605254e-06,
      "loss": 0.5922,
      "step": 63930
    },
    {
      "epoch": 26.17,
      "grad_norm": 2.691580057144165,
      "learning_rate": 4.60419876819604e-06,
      "loss": 0.5898,
      "step": 63940
    },
    {
      "epoch": 26.18,
      "grad_norm": 2.446791410446167,
      "learning_rate": 4.604156847432062e-06,
      "loss": 0.6033,
      "step": 63950
    },
    {
      "epoch": 26.18,
      "grad_norm": 2.185208559036255,
      "learning_rate": 4.60411491631351e-06,
      "loss": 0.5818,
      "step": 63960
    },
    {
      "epoch": 26.19,
      "grad_norm": 2.578155040740967,
      "learning_rate": 4.604072974840575e-06,
      "loss": 0.6021,
      "step": 63970
    },
    {
      "epoch": 26.19,
      "grad_norm": 2.4198226928710938,
      "learning_rate": 4.6040310230134524e-06,
      "loss": 0.5967,
      "step": 63980
    },
    {
      "epoch": 26.19,
      "grad_norm": 2.2691030502319336,
      "learning_rate": 4.603989060832332e-06,
      "loss": 0.5922,
      "step": 63990
    },
    {
      "epoch": 26.2,
      "grad_norm": 3.170750379562378,
      "learning_rate": 4.603947088297406e-06,
      "loss": 0.5592,
      "step": 64000
    },
    {
      "epoch": 26.2,
      "grad_norm": 3.0179190635681152,
      "learning_rate": 4.603905105408867e-06,
      "loss": 0.5982,
      "step": 64010
    },
    {
      "epoch": 26.21,
      "grad_norm": 2.571206569671631,
      "learning_rate": 4.603863112166909e-06,
      "loss": 0.578,
      "step": 64020
    },
    {
      "epoch": 26.21,
      "grad_norm": 2.298262596130371,
      "learning_rate": 4.603821108571721e-06,
      "loss": 0.5842,
      "step": 64030
    },
    {
      "epoch": 26.21,
      "grad_norm": 2.5995357036590576,
      "learning_rate": 4.603779094623497e-06,
      "loss": 0.5976,
      "step": 64040
    },
    {
      "epoch": 26.22,
      "grad_norm": 2.347205638885498,
      "learning_rate": 4.603737070322431e-06,
      "loss": 0.5892,
      "step": 64050
    },
    {
      "epoch": 26.22,
      "grad_norm": 2.550273895263672,
      "learning_rate": 4.603695035668714e-06,
      "loss": 0.5937,
      "step": 64060
    },
    {
      "epoch": 26.23,
      "grad_norm": 1.6008129119873047,
      "learning_rate": 4.6036529906625394e-06,
      "loss": 0.6139,
      "step": 64070
    },
    {
      "epoch": 26.23,
      "grad_norm": 2.5836756229400635,
      "learning_rate": 4.603610935304098e-06,
      "loss": 0.5747,
      "step": 64080
    },
    {
      "epoch": 26.23,
      "grad_norm": 2.5035438537597656,
      "learning_rate": 4.6035688695935845e-06,
      "loss": 0.5994,
      "step": 64090
    },
    {
      "epoch": 26.24,
      "grad_norm": 2.245128870010376,
      "learning_rate": 4.603526793531191e-06,
      "loss": 0.602,
      "step": 64100
    },
    {
      "epoch": 26.24,
      "grad_norm": 2.837834596633911,
      "learning_rate": 4.603484707117109e-06,
      "loss": 0.5877,
      "step": 64110
    },
    {
      "epoch": 26.25,
      "grad_norm": 2.531755208969116,
      "learning_rate": 4.603442610351532e-06,
      "loss": 0.5913,
      "step": 64120
    },
    {
      "epoch": 26.25,
      "grad_norm": 2.2798221111297607,
      "learning_rate": 4.603400503234654e-06,
      "loss": 0.5916,
      "step": 64130
    },
    {
      "epoch": 26.25,
      "grad_norm": 2.0423049926757812,
      "learning_rate": 4.6033583857666664e-06,
      "loss": 0.5811,
      "step": 64140
    },
    {
      "epoch": 26.26,
      "grad_norm": 2.4454853534698486,
      "learning_rate": 4.603316257947763e-06,
      "loss": 0.5996,
      "step": 64150
    },
    {
      "epoch": 26.26,
      "grad_norm": 2.8854384422302246,
      "learning_rate": 4.603274119778137e-06,
      "loss": 0.5878,
      "step": 64160
    },
    {
      "epoch": 26.27,
      "grad_norm": 2.0531396865844727,
      "learning_rate": 4.60323197125798e-06,
      "loss": 0.5588,
      "step": 64170
    },
    {
      "epoch": 26.27,
      "grad_norm": 1.8551973104476929,
      "learning_rate": 4.603189812387487e-06,
      "loss": 0.5798,
      "step": 64180
    },
    {
      "epoch": 26.28,
      "grad_norm": 1.8157198429107666,
      "learning_rate": 4.6031476431668496e-06,
      "loss": 0.5746,
      "step": 64190
    },
    {
      "epoch": 26.28,
      "grad_norm": 2.164794921875,
      "learning_rate": 4.603105463596262e-06,
      "loss": 0.576,
      "step": 64200
    },
    {
      "epoch": 26.28,
      "grad_norm": 2.3756062984466553,
      "learning_rate": 4.603063273675916e-06,
      "loss": 0.5998,
      "step": 64210
    },
    {
      "epoch": 26.29,
      "grad_norm": 2.5721917152404785,
      "learning_rate": 4.603021073406006e-06,
      "loss": 0.5857,
      "step": 64220
    },
    {
      "epoch": 26.29,
      "grad_norm": 2.4375689029693604,
      "learning_rate": 4.602978862786726e-06,
      "loss": 0.5979,
      "step": 64230
    },
    {
      "epoch": 26.3,
      "grad_norm": 3.4882826805114746,
      "learning_rate": 4.6029366418182675e-06,
      "loss": 0.5787,
      "step": 64240
    },
    {
      "epoch": 26.3,
      "grad_norm": 1.7898123264312744,
      "learning_rate": 4.6028944105008254e-06,
      "loss": 0.5862,
      "step": 64250
    },
    {
      "epoch": 26.3,
      "grad_norm": 2.860419988632202,
      "learning_rate": 4.602852168834593e-06,
      "loss": 0.5716,
      "step": 64260
    },
    {
      "epoch": 26.31,
      "grad_norm": 2.3381052017211914,
      "learning_rate": 4.602809916819762e-06,
      "loss": 0.5832,
      "step": 64270
    },
    {
      "epoch": 26.31,
      "grad_norm": 2.6228373050689697,
      "learning_rate": 4.6027676544565296e-06,
      "loss": 0.5983,
      "step": 64280
    },
    {
      "epoch": 26.32,
      "grad_norm": 1.915860891342163,
      "learning_rate": 4.602725381745085e-06,
      "loss": 0.5903,
      "step": 64290
    },
    {
      "epoch": 26.32,
      "grad_norm": 2.9678702354431152,
      "learning_rate": 4.602683098685625e-06,
      "loss": 0.5555,
      "step": 64300
    },
    {
      "epoch": 26.32,
      "grad_norm": 2.285081148147583,
      "learning_rate": 4.602640805278342e-06,
      "loss": 0.5992,
      "step": 64310
    },
    {
      "epoch": 26.33,
      "grad_norm": 2.369788408279419,
      "learning_rate": 4.60259850152343e-06,
      "loss": 0.5806,
      "step": 64320
    },
    {
      "epoch": 26.33,
      "grad_norm": 2.5574731826782227,
      "learning_rate": 4.602556187421084e-06,
      "loss": 0.5845,
      "step": 64330
    },
    {
      "epoch": 26.34,
      "grad_norm": 3.437265157699585,
      "learning_rate": 4.602513862971495e-06,
      "loss": 0.5906,
      "step": 64340
    },
    {
      "epoch": 26.34,
      "grad_norm": 2.215071201324463,
      "learning_rate": 4.602471528174859e-06,
      "loss": 0.5698,
      "step": 64350
    },
    {
      "epoch": 26.34,
      "grad_norm": 3.1552162170410156,
      "learning_rate": 4.6024291830313706e-06,
      "loss": 0.6081,
      "step": 64360
    },
    {
      "epoch": 26.35,
      "grad_norm": 2.775192975997925,
      "learning_rate": 4.602386827541222e-06,
      "loss": 0.5807,
      "step": 64370
    },
    {
      "epoch": 26.35,
      "grad_norm": 2.2663025856018066,
      "learning_rate": 4.602344461704607e-06,
      "loss": 0.5844,
      "step": 64380
    },
    {
      "epoch": 26.36,
      "grad_norm": 2.623349905014038,
      "learning_rate": 4.602302085521722e-06,
      "loss": 0.5974,
      "step": 64390
    },
    {
      "epoch": 26.36,
      "grad_norm": 2.053270101547241,
      "learning_rate": 4.602259698992759e-06,
      "loss": 0.5766,
      "step": 64400
    },
    {
      "epoch": 26.37,
      "grad_norm": 2.09000563621521,
      "learning_rate": 4.602217302117913e-06,
      "loss": 0.5934,
      "step": 64410
    },
    {
      "epoch": 26.37,
      "grad_norm": 2.346587896347046,
      "learning_rate": 4.6021748948973776e-06,
      "loss": 0.5946,
      "step": 64420
    },
    {
      "epoch": 26.37,
      "grad_norm": 2.3426425457000732,
      "learning_rate": 4.602132477331348e-06,
      "loss": 0.5837,
      "step": 64430
    },
    {
      "epoch": 26.38,
      "grad_norm": 2.206392288208008,
      "learning_rate": 4.6020900494200175e-06,
      "loss": 0.5754,
      "step": 64440
    },
    {
      "epoch": 26.38,
      "grad_norm": 2.0954995155334473,
      "learning_rate": 4.602047611163582e-06,
      "loss": 0.5692,
      "step": 64450
    },
    {
      "epoch": 26.39,
      "grad_norm": 2.4944546222686768,
      "learning_rate": 4.602005162562233e-06,
      "loss": 0.5957,
      "step": 64460
    },
    {
      "epoch": 26.39,
      "grad_norm": 3.8140692710876465,
      "learning_rate": 4.601962703616168e-06,
      "loss": 0.5782,
      "step": 64470
    },
    {
      "epoch": 26.39,
      "grad_norm": 2.315495491027832,
      "learning_rate": 4.601920234325581e-06,
      "loss": 0.5704,
      "step": 64480
    },
    {
      "epoch": 26.4,
      "grad_norm": 3.1686933040618896,
      "learning_rate": 4.601877754690665e-06,
      "loss": 0.5762,
      "step": 64490
    },
    {
      "epoch": 26.4,
      "grad_norm": 2.483968734741211,
      "learning_rate": 4.601835264711616e-06,
      "loss": 0.5788,
      "step": 64500
    },
    {
      "epoch": 26.41,
      "grad_norm": 3.1084463596343994,
      "learning_rate": 4.601792764388628e-06,
      "loss": 0.6049,
      "step": 64510
    },
    {
      "epoch": 26.41,
      "grad_norm": 2.524602174758911,
      "learning_rate": 4.601750253721895e-06,
      "loss": 0.5622,
      "step": 64520
    },
    {
      "epoch": 26.41,
      "grad_norm": 1.8096909523010254,
      "learning_rate": 4.6017077327116136e-06,
      "loss": 0.5906,
      "step": 64530
    },
    {
      "epoch": 26.42,
      "grad_norm": 3.2209572792053223,
      "learning_rate": 4.601665201357977e-06,
      "loss": 0.5988,
      "step": 64540
    },
    {
      "epoch": 26.42,
      "grad_norm": 2.683875799179077,
      "learning_rate": 4.601622659661181e-06,
      "loss": 0.5847,
      "step": 64550
    },
    {
      "epoch": 26.43,
      "grad_norm": 2.1780271530151367,
      "learning_rate": 4.60158010762142e-06,
      "loss": 0.5932,
      "step": 64560
    },
    {
      "epoch": 26.43,
      "grad_norm": 1.8996206521987915,
      "learning_rate": 4.601537545238888e-06,
      "loss": 0.5879,
      "step": 64570
    },
    {
      "epoch": 26.43,
      "grad_norm": 2.577967643737793,
      "learning_rate": 4.601494972513781e-06,
      "loss": 0.5733,
      "step": 64580
    },
    {
      "epoch": 26.44,
      "grad_norm": 2.764862537384033,
      "learning_rate": 4.601452389446295e-06,
      "loss": 0.5972,
      "step": 64590
    },
    {
      "epoch": 26.44,
      "grad_norm": 2.646465539932251,
      "learning_rate": 4.601409796036623e-06,
      "loss": 0.5741,
      "step": 64600
    },
    {
      "epoch": 26.45,
      "grad_norm": 2.591874122619629,
      "learning_rate": 4.601367192284962e-06,
      "loss": 0.5758,
      "step": 64610
    },
    {
      "epoch": 26.45,
      "grad_norm": 2.00358510017395,
      "learning_rate": 4.601324578191506e-06,
      "loss": 0.5708,
      "step": 64620
    },
    {
      "epoch": 26.46,
      "grad_norm": 2.09951114654541,
      "learning_rate": 4.60128195375645e-06,
      "loss": 0.5737,
      "step": 64630
    },
    {
      "epoch": 26.46,
      "grad_norm": 2.882769823074341,
      "learning_rate": 4.60123931897999e-06,
      "loss": 0.6082,
      "step": 64640
    },
    {
      "epoch": 26.46,
      "grad_norm": 2.9914755821228027,
      "learning_rate": 4.601196673862321e-06,
      "loss": 0.5904,
      "step": 64650
    },
    {
      "epoch": 26.47,
      "grad_norm": 2.677677631378174,
      "learning_rate": 4.601154018403639e-06,
      "loss": 0.574,
      "step": 64660
    },
    {
      "epoch": 26.47,
      "grad_norm": 2.4335708618164062,
      "learning_rate": 4.601111352604138e-06,
      "loss": 0.6021,
      "step": 64670
    },
    {
      "epoch": 26.48,
      "grad_norm": 2.748739719390869,
      "learning_rate": 4.601068676464015e-06,
      "loss": 0.5865,
      "step": 64680
    },
    {
      "epoch": 26.48,
      "grad_norm": 3.229912757873535,
      "learning_rate": 4.601025989983464e-06,
      "loss": 0.5894,
      "step": 64690
    },
    {
      "epoch": 26.48,
      "grad_norm": 2.8713161945343018,
      "learning_rate": 4.600983293162681e-06,
      "loss": 0.5619,
      "step": 64700
    },
    {
      "epoch": 26.49,
      "grad_norm": 2.3011257648468018,
      "learning_rate": 4.600940586001863e-06,
      "loss": 0.5961,
      "step": 64710
    },
    {
      "epoch": 26.49,
      "grad_norm": 1.8428481817245483,
      "learning_rate": 4.600897868501204e-06,
      "loss": 0.5679,
      "step": 64720
    },
    {
      "epoch": 26.5,
      "grad_norm": 2.7054009437561035,
      "learning_rate": 4.6008551406609e-06,
      "loss": 0.586,
      "step": 64730
    },
    {
      "epoch": 26.5,
      "grad_norm": 2.5933172702789307,
      "learning_rate": 4.600812402481147e-06,
      "loss": 0.588,
      "step": 64740
    },
    {
      "epoch": 26.5,
      "grad_norm": 2.164201021194458,
      "learning_rate": 4.6007696539621405e-06,
      "loss": 0.6084,
      "step": 64750
    },
    {
      "epoch": 26.51,
      "grad_norm": 3.026634693145752,
      "learning_rate": 4.600726895104077e-06,
      "loss": 0.5942,
      "step": 64760
    },
    {
      "epoch": 26.51,
      "grad_norm": 1.9430389404296875,
      "learning_rate": 4.600684125907152e-06,
      "loss": 0.571,
      "step": 64770
    },
    {
      "epoch": 26.52,
      "grad_norm": 4.052549839019775,
      "learning_rate": 4.600641346371561e-06,
      "loss": 0.5929,
      "step": 64780
    },
    {
      "epoch": 26.52,
      "grad_norm": 3.7321078777313232,
      "learning_rate": 4.6005985564975e-06,
      "loss": 0.5726,
      "step": 64790
    },
    {
      "epoch": 26.52,
      "grad_norm": 4.018692970275879,
      "learning_rate": 4.600555756285165e-06,
      "loss": 0.5892,
      "step": 64800
    },
    {
      "epoch": 26.53,
      "grad_norm": 2.8260035514831543,
      "learning_rate": 4.600512945734753e-06,
      "loss": 0.5816,
      "step": 64810
    },
    {
      "epoch": 26.53,
      "grad_norm": 1.979804277420044,
      "learning_rate": 4.6004701248464585e-06,
      "loss": 0.5885,
      "step": 64820
    },
    {
      "epoch": 26.54,
      "grad_norm": 2.274881601333618,
      "learning_rate": 4.60042729362048e-06,
      "loss": 0.5786,
      "step": 64830
    },
    {
      "epoch": 26.54,
      "grad_norm": 3.7638204097747803,
      "learning_rate": 4.600384452057011e-06,
      "loss": 0.5878,
      "step": 64840
    },
    {
      "epoch": 26.55,
      "grad_norm": 2.902127265930176,
      "learning_rate": 4.60034160015625e-06,
      "loss": 0.5783,
      "step": 64850
    },
    {
      "epoch": 26.55,
      "grad_norm": 2.2838237285614014,
      "learning_rate": 4.600298737918392e-06,
      "loss": 0.5945,
      "step": 64860
    },
    {
      "epoch": 26.55,
      "grad_norm": 2.6787703037261963,
      "learning_rate": 4.600255865343634e-06,
      "loss": 0.5734,
      "step": 64870
    },
    {
      "epoch": 26.56,
      "grad_norm": 2.924039840698242,
      "learning_rate": 4.600212982432172e-06,
      "loss": 0.5884,
      "step": 64880
    },
    {
      "epoch": 26.56,
      "grad_norm": 1.937320351600647,
      "learning_rate": 4.600170089184202e-06,
      "loss": 0.5659,
      "step": 64890
    },
    {
      "epoch": 26.57,
      "grad_norm": 2.96585750579834,
      "learning_rate": 4.600127185599921e-06,
      "loss": 0.5974,
      "step": 64900
    },
    {
      "epoch": 26.57,
      "grad_norm": 2.5876145362854004,
      "learning_rate": 4.600084271679526e-06,
      "loss": 0.5811,
      "step": 64910
    },
    {
      "epoch": 26.57,
      "grad_norm": 2.267530679702759,
      "learning_rate": 4.600041347423213e-06,
      "loss": 0.5769,
      "step": 64920
    },
    {
      "epoch": 26.58,
      "grad_norm": 2.9892871379852295,
      "learning_rate": 4.59999841283118e-06,
      "loss": 0.5682,
      "step": 64930
    },
    {
      "epoch": 26.58,
      "grad_norm": 2.7670960426330566,
      "learning_rate": 4.599955467903621e-06,
      "loss": 0.5747,
      "step": 64940
    },
    {
      "epoch": 26.59,
      "grad_norm": 2.539649724960327,
      "learning_rate": 4.599912512640734e-06,
      "loss": 0.5765,
      "step": 64950
    },
    {
      "epoch": 26.59,
      "grad_norm": 2.7136805057525635,
      "learning_rate": 4.599869547042716e-06,
      "loss": 0.6008,
      "step": 64960
    },
    {
      "epoch": 26.59,
      "grad_norm": 3.1279807090759277,
      "learning_rate": 4.599826571109765e-06,
      "loss": 0.5823,
      "step": 64970
    },
    {
      "epoch": 26.6,
      "grad_norm": 2.699831247329712,
      "learning_rate": 4.599783584842075e-06,
      "loss": 0.5998,
      "step": 64980
    },
    {
      "epoch": 26.6,
      "grad_norm": 2.1904501914978027,
      "learning_rate": 4.5997405882398455e-06,
      "loss": 0.5979,
      "step": 64990
    },
    {
      "epoch": 26.61,
      "grad_norm": 2.472644567489624,
      "learning_rate": 4.599697581303272e-06,
      "loss": 0.6046,
      "step": 65000
    },
    {
      "epoch": 26.61,
      "grad_norm": 2.529417037963867,
      "learning_rate": 4.5996545640325524e-06,
      "loss": 0.5775,
      "step": 65010
    },
    {
      "epoch": 26.61,
      "grad_norm": 2.8588244915008545,
      "learning_rate": 4.599611536427884e-06,
      "loss": 0.5974,
      "step": 65020
    },
    {
      "epoch": 26.62,
      "grad_norm": 2.163729667663574,
      "learning_rate": 4.599568498489463e-06,
      "loss": 0.5781,
      "step": 65030
    },
    {
      "epoch": 26.62,
      "grad_norm": 2.89454984664917,
      "learning_rate": 4.599525450217485e-06,
      "loss": 0.6156,
      "step": 65040
    },
    {
      "epoch": 26.63,
      "grad_norm": 2.724163293838501,
      "learning_rate": 4.599482391612151e-06,
      "loss": 0.5955,
      "step": 65050
    },
    {
      "epoch": 26.63,
      "grad_norm": 2.96451735496521,
      "learning_rate": 4.599439322673656e-06,
      "loss": 0.5866,
      "step": 65060
    },
    {
      "epoch": 26.64,
      "grad_norm": 2.420903205871582,
      "learning_rate": 4.599396243402197e-06,
      "loss": 0.5816,
      "step": 65070
    },
    {
      "epoch": 26.64,
      "grad_norm": 2.4332332611083984,
      "learning_rate": 4.599353153797972e-06,
      "loss": 0.6062,
      "step": 65080
    },
    {
      "epoch": 26.64,
      "grad_norm": 2.9721853733062744,
      "learning_rate": 4.5993100538611785e-06,
      "loss": 0.59,
      "step": 65090
    },
    {
      "epoch": 26.65,
      "grad_norm": 2.6637113094329834,
      "learning_rate": 4.5992669435920136e-06,
      "loss": 0.586,
      "step": 65100
    },
    {
      "epoch": 26.65,
      "grad_norm": 2.435166358947754,
      "learning_rate": 4.599223822990675e-06,
      "loss": 0.5752,
      "step": 65110
    },
    {
      "epoch": 26.66,
      "grad_norm": 2.2657437324523926,
      "learning_rate": 4.59918069205736e-06,
      "loss": 0.5846,
      "step": 65120
    },
    {
      "epoch": 26.66,
      "grad_norm": 2.963033437728882,
      "learning_rate": 4.599137550792266e-06,
      "loss": 0.593,
      "step": 65130
    },
    {
      "epoch": 26.66,
      "grad_norm": 2.1135308742523193,
      "learning_rate": 4.599094399195591e-06,
      "loss": 0.5933,
      "step": 65140
    },
    {
      "epoch": 26.67,
      "grad_norm": 2.544434070587158,
      "learning_rate": 4.599051237267533e-06,
      "loss": 0.5978,
      "step": 65150
    },
    {
      "epoch": 26.67,
      "grad_norm": 2.480036973953247,
      "learning_rate": 4.599008065008289e-06,
      "loss": 0.6029,
      "step": 65160
    },
    {
      "epoch": 26.68,
      "grad_norm": 3.34873628616333,
      "learning_rate": 4.598964882418057e-06,
      "loss": 0.578,
      "step": 65170
    },
    {
      "epoch": 26.68,
      "grad_norm": 2.8610401153564453,
      "learning_rate": 4.598921689497036e-06,
      "loss": 0.5789,
      "step": 65180
    },
    {
      "epoch": 26.68,
      "grad_norm": 2.174022674560547,
      "learning_rate": 4.598878486245421e-06,
      "loss": 0.5755,
      "step": 65190
    },
    {
      "epoch": 26.69,
      "grad_norm": 2.3187015056610107,
      "learning_rate": 4.598835272663413e-06,
      "loss": 0.5715,
      "step": 65200
    },
    {
      "epoch": 26.69,
      "grad_norm": 1.9500644207000732,
      "learning_rate": 4.598792048751208e-06,
      "loss": 0.5722,
      "step": 65210
    },
    {
      "epoch": 26.7,
      "grad_norm": 2.7872681617736816,
      "learning_rate": 4.598748814509005e-06,
      "loss": 0.5781,
      "step": 65220
    },
    {
      "epoch": 26.7,
      "grad_norm": 3.179877996444702,
      "learning_rate": 4.598705569937001e-06,
      "loss": 0.5941,
      "step": 65230
    },
    {
      "epoch": 26.7,
      "grad_norm": 2.219376564025879,
      "learning_rate": 4.598662315035395e-06,
      "loss": 0.5744,
      "step": 65240
    },
    {
      "epoch": 26.71,
      "grad_norm": 1.9772663116455078,
      "learning_rate": 4.598619049804385e-06,
      "loss": 0.5554,
      "step": 65250
    },
    {
      "epoch": 26.71,
      "grad_norm": 2.498405694961548,
      "learning_rate": 4.598575774244169e-06,
      "loss": 0.5774,
      "step": 65260
    },
    {
      "epoch": 26.72,
      "grad_norm": 2.2685234546661377,
      "learning_rate": 4.598532488354946e-06,
      "loss": 0.6092,
      "step": 65270
    },
    {
      "epoch": 26.72,
      "grad_norm": 2.007657527923584,
      "learning_rate": 4.5984891921369126e-06,
      "loss": 0.5954,
      "step": 65280
    },
    {
      "epoch": 26.73,
      "grad_norm": 2.4716498851776123,
      "learning_rate": 4.598445885590269e-06,
      "loss": 0.5611,
      "step": 65290
    },
    {
      "epoch": 26.73,
      "grad_norm": 3.3413093090057373,
      "learning_rate": 4.5984025687152125e-06,
      "loss": 0.583,
      "step": 65300
    },
    {
      "epoch": 26.73,
      "grad_norm": 2.0799455642700195,
      "learning_rate": 4.598359241511941e-06,
      "loss": 0.57,
      "step": 65310
    },
    {
      "epoch": 26.74,
      "grad_norm": 3.152517318725586,
      "learning_rate": 4.598315903980655e-06,
      "loss": 0.5812,
      "step": 65320
    },
    {
      "epoch": 26.74,
      "grad_norm": 3.240442991256714,
      "learning_rate": 4.5982725561215506e-06,
      "loss": 0.6026,
      "step": 65330
    },
    {
      "epoch": 26.75,
      "grad_norm": 2.355131149291992,
      "learning_rate": 4.598229197934828e-06,
      "loss": 0.5728,
      "step": 65340
    },
    {
      "epoch": 26.75,
      "grad_norm": 2.300872325897217,
      "learning_rate": 4.598185829420685e-06,
      "loss": 0.5901,
      "step": 65350
    },
    {
      "epoch": 26.75,
      "grad_norm": 2.304330348968506,
      "learning_rate": 4.598142450579321e-06,
      "loss": 0.5845,
      "step": 65360
    },
    {
      "epoch": 26.76,
      "grad_norm": 2.4844422340393066,
      "learning_rate": 4.598099061410935e-06,
      "loss": 0.5775,
      "step": 65370
    },
    {
      "epoch": 26.76,
      "grad_norm": 3.3015198707580566,
      "learning_rate": 4.598055661915724e-06,
      "loss": 0.5684,
      "step": 65380
    },
    {
      "epoch": 26.77,
      "grad_norm": 2.065453290939331,
      "learning_rate": 4.598012252093888e-06,
      "loss": 0.5638,
      "step": 65390
    },
    {
      "epoch": 26.77,
      "grad_norm": 2.2708206176757812,
      "learning_rate": 4.597968831945627e-06,
      "loss": 0.5875,
      "step": 65400
    },
    {
      "epoch": 26.77,
      "grad_norm": 2.957180976867676,
      "learning_rate": 4.597925401471137e-06,
      "loss": 0.5942,
      "step": 65410
    },
    {
      "epoch": 26.78,
      "grad_norm": 2.557723045349121,
      "learning_rate": 4.59788196067062e-06,
      "loss": 0.58,
      "step": 65420
    },
    {
      "epoch": 26.78,
      "grad_norm": 3.7977466583251953,
      "learning_rate": 4.597838509544272e-06,
      "loss": 0.5895,
      "step": 65430
    },
    {
      "epoch": 26.79,
      "grad_norm": 3.4368412494659424,
      "learning_rate": 4.5977950480922945e-06,
      "loss": 0.5706,
      "step": 65440
    },
    {
      "epoch": 26.79,
      "grad_norm": 2.5999808311462402,
      "learning_rate": 4.597751576314885e-06,
      "loss": 0.5949,
      "step": 65450
    },
    {
      "epoch": 26.79,
      "grad_norm": 2.5293774604797363,
      "learning_rate": 4.597708094212245e-06,
      "loss": 0.5968,
      "step": 65460
    },
    {
      "epoch": 26.8,
      "grad_norm": 2.6413726806640625,
      "learning_rate": 4.597664601784571e-06,
      "loss": 0.5787,
      "step": 65470
    },
    {
      "epoch": 26.8,
      "grad_norm": 2.5204367637634277,
      "learning_rate": 4.597621099032063e-06,
      "loss": 0.5863,
      "step": 65480
    },
    {
      "epoch": 26.81,
      "grad_norm": 2.286494493484497,
      "learning_rate": 4.597577585954921e-06,
      "loss": 0.5884,
      "step": 65490
    },
    {
      "epoch": 26.81,
      "grad_norm": 2.41163969039917,
      "learning_rate": 4.5975340625533435e-06,
      "loss": 0.5953,
      "step": 65500
    },
    {
      "epoch": 26.82,
      "grad_norm": 2.2693963050842285,
      "learning_rate": 4.597490528827531e-06,
      "loss": 0.5872,
      "step": 65510
    },
    {
      "epoch": 26.82,
      "grad_norm": 1.9510881900787354,
      "learning_rate": 4.597446984777682e-06,
      "loss": 0.5696,
      "step": 65520
    },
    {
      "epoch": 26.82,
      "grad_norm": 2.9583780765533447,
      "learning_rate": 4.597403430403995e-06,
      "loss": 0.5822,
      "step": 65530
    },
    {
      "epoch": 26.83,
      "grad_norm": 3.5570616722106934,
      "learning_rate": 4.597359865706672e-06,
      "loss": 0.6055,
      "step": 65540
    },
    {
      "epoch": 26.83,
      "grad_norm": 2.6275746822357178,
      "learning_rate": 4.597316290685911e-06,
      "loss": 0.577,
      "step": 65550
    },
    {
      "epoch": 26.84,
      "grad_norm": 2.0372955799102783,
      "learning_rate": 4.597272705341911e-06,
      "loss": 0.5833,
      "step": 65560
    },
    {
      "epoch": 26.84,
      "grad_norm": 2.101938009262085,
      "learning_rate": 4.597229109674874e-06,
      "loss": 0.6102,
      "step": 65570
    },
    {
      "epoch": 26.84,
      "grad_norm": 2.363659620285034,
      "learning_rate": 4.597185503684997e-06,
      "loss": 0.5889,
      "step": 65580
    },
    {
      "epoch": 26.85,
      "grad_norm": 1.9501316547393799,
      "learning_rate": 4.597141887372481e-06,
      "loss": 0.5702,
      "step": 65590
    },
    {
      "epoch": 26.85,
      "grad_norm": 2.402965545654297,
      "learning_rate": 4.597098260737526e-06,
      "loss": 0.5881,
      "step": 65600
    },
    {
      "epoch": 26.86,
      "grad_norm": 2.4379072189331055,
      "learning_rate": 4.5970546237803315e-06,
      "loss": 0.5746,
      "step": 65610
    },
    {
      "epoch": 26.86,
      "grad_norm": 2.123140573501587,
      "learning_rate": 4.597010976501097e-06,
      "loss": 0.5756,
      "step": 65620
    },
    {
      "epoch": 26.86,
      "grad_norm": 2.264314651489258,
      "learning_rate": 4.596967318900024e-06,
      "loss": 0.5865,
      "step": 65630
    },
    {
      "epoch": 26.87,
      "grad_norm": 2.5032784938812256,
      "learning_rate": 4.596923650977311e-06,
      "loss": 0.6108,
      "step": 65640
    },
    {
      "epoch": 26.87,
      "grad_norm": 2.3153634071350098,
      "learning_rate": 4.596879972733158e-06,
      "loss": 0.5905,
      "step": 65650
    },
    {
      "epoch": 26.88,
      "grad_norm": 2.4617133140563965,
      "learning_rate": 4.5968362841677656e-06,
      "loss": 0.571,
      "step": 65660
    },
    {
      "epoch": 26.88,
      "grad_norm": 1.949966549873352,
      "learning_rate": 4.596792585281334e-06,
      "loss": 0.6051,
      "step": 65670
    },
    {
      "epoch": 26.88,
      "grad_norm": 1.8564488887786865,
      "learning_rate": 4.596748876074063e-06,
      "loss": 0.5806,
      "step": 65680
    },
    {
      "epoch": 26.89,
      "grad_norm": 1.964913010597229,
      "learning_rate": 4.596705156546153e-06,
      "loss": 0.6007,
      "step": 65690
    },
    {
      "epoch": 26.89,
      "grad_norm": 2.231417417526245,
      "learning_rate": 4.596661426697805e-06,
      "loss": 0.6005,
      "step": 65700
    },
    {
      "epoch": 26.9,
      "grad_norm": 2.2621569633483887,
      "learning_rate": 4.596617686529219e-06,
      "loss": 0.5849,
      "step": 65710
    },
    {
      "epoch": 26.9,
      "grad_norm": 2.7737975120544434,
      "learning_rate": 4.596573936040594e-06,
      "loss": 0.5721,
      "step": 65720
    },
    {
      "epoch": 26.91,
      "grad_norm": 2.6225762367248535,
      "learning_rate": 4.596530175232132e-06,
      "loss": 0.5756,
      "step": 65730
    },
    {
      "epoch": 26.91,
      "grad_norm": 2.3247697353363037,
      "learning_rate": 4.596486404104033e-06,
      "loss": 0.5914,
      "step": 65740
    },
    {
      "epoch": 26.91,
      "grad_norm": 2.795745611190796,
      "learning_rate": 4.5964426226564975e-06,
      "loss": 0.5919,
      "step": 65750
    },
    {
      "epoch": 26.92,
      "grad_norm": 2.750155210494995,
      "learning_rate": 4.596398830889725e-06,
      "loss": 0.5842,
      "step": 65760
    },
    {
      "epoch": 26.92,
      "grad_norm": 2.598484992980957,
      "learning_rate": 4.596355028803918e-06,
      "loss": 0.5902,
      "step": 65770
    },
    {
      "epoch": 26.93,
      "grad_norm": 2.303997278213501,
      "learning_rate": 4.596311216399276e-06,
      "loss": 0.5948,
      "step": 65780
    },
    {
      "epoch": 26.93,
      "grad_norm": 1.9822213649749756,
      "learning_rate": 4.596267393676e-06,
      "loss": 0.5939,
      "step": 65790
    },
    {
      "epoch": 26.93,
      "grad_norm": 1.6979349851608276,
      "learning_rate": 4.59622356063429e-06,
      "loss": 0.5924,
      "step": 65800
    },
    {
      "epoch": 26.94,
      "grad_norm": 2.0917201042175293,
      "learning_rate": 4.596179717274348e-06,
      "loss": 0.5765,
      "step": 65810
    },
    {
      "epoch": 26.94,
      "grad_norm": 2.0653419494628906,
      "learning_rate": 4.596135863596375e-06,
      "loss": 0.5849,
      "step": 65820
    },
    {
      "epoch": 26.95,
      "grad_norm": 2.651024580001831,
      "learning_rate": 4.59609199960057e-06,
      "loss": 0.5761,
      "step": 65830
    },
    {
      "epoch": 26.95,
      "grad_norm": 2.449526786804199,
      "learning_rate": 4.596048125287135e-06,
      "loss": 0.6026,
      "step": 65840
    },
    {
      "epoch": 26.95,
      "grad_norm": 2.6210854053497314,
      "learning_rate": 4.5960042406562725e-06,
      "loss": 0.6138,
      "step": 65850
    },
    {
      "epoch": 26.96,
      "grad_norm": 3.059197425842285,
      "learning_rate": 4.5959603457081815e-06,
      "loss": 0.5528,
      "step": 65860
    },
    {
      "epoch": 26.96,
      "grad_norm": 2.9524974822998047,
      "learning_rate": 4.595916440443063e-06,
      "loss": 0.589,
      "step": 65870
    },
    {
      "epoch": 26.97,
      "grad_norm": 2.114811420440674,
      "learning_rate": 4.595872524861119e-06,
      "loss": 0.5832,
      "step": 65880
    },
    {
      "epoch": 26.97,
      "grad_norm": 2.6549062728881836,
      "learning_rate": 4.595828598962552e-06,
      "loss": 0.5872,
      "step": 65890
    },
    {
      "epoch": 26.98,
      "grad_norm": 2.2119228839874268,
      "learning_rate": 4.59578466274756e-06,
      "loss": 0.5746,
      "step": 65900
    },
    {
      "epoch": 26.98,
      "grad_norm": 2.5638198852539062,
      "learning_rate": 4.595740716216347e-06,
      "loss": 0.6046,
      "step": 65910
    },
    {
      "epoch": 26.98,
      "grad_norm": 2.5121865272521973,
      "learning_rate": 4.595696759369113e-06,
      "loss": 0.5924,
      "step": 65920
    },
    {
      "epoch": 26.99,
      "grad_norm": 2.4614615440368652,
      "learning_rate": 4.595652792206059e-06,
      "loss": 0.5808,
      "step": 65930
    },
    {
      "epoch": 26.99,
      "grad_norm": 2.5851287841796875,
      "learning_rate": 4.5956088147273875e-06,
      "loss": 0.5669,
      "step": 65940
    },
    {
      "epoch": 27.0,
      "grad_norm": 2.995889663696289,
      "learning_rate": 4.5955648269333e-06,
      "loss": 0.5813,
      "step": 65950
    },
    {
      "epoch": 27.0,
      "grad_norm": 2.2368102073669434,
      "learning_rate": 4.595520828823997e-06,
      "loss": 0.578,
      "step": 65960
    },
    {
      "epoch": 27.0,
      "eval_loss": 0.58408522605896,
      "eval_runtime": 51.9527,
      "eval_samples_per_second": 66.387,
      "eval_steps_per_second": 8.315,
      "step": 65961
    },
    {
      "epoch": 27.0,
      "grad_norm": 2.6810076236724854,
      "learning_rate": 4.59547682039968e-06,
      "loss": 0.5672,
      "step": 65970
    },
    {
      "epoch": 27.01,
      "grad_norm": 2.6999025344848633,
      "learning_rate": 4.595432801660553e-06,
      "loss": 0.5799,
      "step": 65980
    },
    {
      "epoch": 27.01,
      "grad_norm": 2.323049783706665,
      "learning_rate": 4.595388772606814e-06,
      "loss": 0.5846,
      "step": 65990
    },
    {
      "epoch": 27.02,
      "grad_norm": 2.474163055419922,
      "learning_rate": 4.595344733238667e-06,
      "loss": 0.6017,
      "step": 66000
    },
    {
      "epoch": 27.02,
      "grad_norm": 2.4596662521362305,
      "learning_rate": 4.595300683556313e-06,
      "loss": 0.5897,
      "step": 66010
    },
    {
      "epoch": 27.02,
      "grad_norm": 2.3970885276794434,
      "learning_rate": 4.595256623559955e-06,
      "loss": 0.569,
      "step": 66020
    },
    {
      "epoch": 27.03,
      "grad_norm": 3.2757420539855957,
      "learning_rate": 4.595212553249793e-06,
      "loss": 0.5644,
      "step": 66030
    },
    {
      "epoch": 27.03,
      "grad_norm": 3.2896578311920166,
      "learning_rate": 4.59516847262603e-06,
      "loss": 0.5924,
      "step": 66040
    },
    {
      "epoch": 27.04,
      "grad_norm": 1.982426404953003,
      "learning_rate": 4.595124381688868e-06,
      "loss": 0.5838,
      "step": 66050
    },
    {
      "epoch": 27.04,
      "grad_norm": 2.7899768352508545,
      "learning_rate": 4.595080280438508e-06,
      "loss": 0.6017,
      "step": 66060
    },
    {
      "epoch": 27.04,
      "grad_norm": 3.5511486530303955,
      "learning_rate": 4.595036168875153e-06,
      "loss": 0.5795,
      "step": 66070
    },
    {
      "epoch": 27.05,
      "grad_norm": 3.252838611602783,
      "learning_rate": 4.594992046999004e-06,
      "loss": 0.5845,
      "step": 66080
    },
    {
      "epoch": 27.05,
      "grad_norm": 2.2237751483917236,
      "learning_rate": 4.594947914810265e-06,
      "loss": 0.5993,
      "step": 66090
    },
    {
      "epoch": 27.06,
      "grad_norm": 2.6542768478393555,
      "learning_rate": 4.594903772309136e-06,
      "loss": 0.593,
      "step": 66100
    },
    {
      "epoch": 27.06,
      "grad_norm": 2.220909595489502,
      "learning_rate": 4.594859619495821e-06,
      "loss": 0.574,
      "step": 66110
    },
    {
      "epoch": 27.07,
      "grad_norm": 3.4343104362487793,
      "learning_rate": 4.594815456370521e-06,
      "loss": 0.5951,
      "step": 66120
    },
    {
      "epoch": 27.07,
      "grad_norm": 2.847127676010132,
      "learning_rate": 4.594771282933438e-06,
      "loss": 0.5771,
      "step": 66130
    },
    {
      "epoch": 27.07,
      "grad_norm": 3.54176926612854,
      "learning_rate": 4.594727099184776e-06,
      "loss": 0.5806,
      "step": 66140
    },
    {
      "epoch": 27.08,
      "grad_norm": 3.183084487915039,
      "learning_rate": 4.594682905124737e-06,
      "loss": 0.5964,
      "step": 66150
    },
    {
      "epoch": 27.08,
      "grad_norm": 2.8089802265167236,
      "learning_rate": 4.5946387007535224e-06,
      "loss": 0.5788,
      "step": 66160
    },
    {
      "epoch": 27.09,
      "grad_norm": 2.0744669437408447,
      "learning_rate": 4.594594486071335e-06,
      "loss": 0.5863,
      "step": 66170
    },
    {
      "epoch": 27.09,
      "grad_norm": 2.0071403980255127,
      "learning_rate": 4.594550261078378e-06,
      "loss": 0.5884,
      "step": 66180
    },
    {
      "epoch": 27.09,
      "grad_norm": 2.9962854385375977,
      "learning_rate": 4.594506025774853e-06,
      "loss": 0.5598,
      "step": 66190
    },
    {
      "epoch": 27.1,
      "grad_norm": 2.264137029647827,
      "learning_rate": 4.594461780160965e-06,
      "loss": 0.5813,
      "step": 66200
    },
    {
      "epoch": 27.1,
      "grad_norm": 2.774778127670288,
      "learning_rate": 4.594417524236913e-06,
      "loss": 0.5913,
      "step": 66210
    },
    {
      "epoch": 27.11,
      "grad_norm": 2.2463316917419434,
      "learning_rate": 4.594373258002902e-06,
      "loss": 0.5635,
      "step": 66220
    },
    {
      "epoch": 27.11,
      "grad_norm": 2.6385741233825684,
      "learning_rate": 4.594328981459135e-06,
      "loss": 0.5773,
      "step": 66230
    },
    {
      "epoch": 27.11,
      "grad_norm": 2.599226951599121,
      "learning_rate": 4.594284694605814e-06,
      "loss": 0.5852,
      "step": 66240
    },
    {
      "epoch": 27.12,
      "grad_norm": 2.332261323928833,
      "learning_rate": 4.594240397443141e-06,
      "loss": 0.5599,
      "step": 66250
    },
    {
      "epoch": 27.12,
      "grad_norm": 2.226022481918335,
      "learning_rate": 4.594196089971321e-06,
      "loss": 0.5754,
      "step": 66260
    },
    {
      "epoch": 27.13,
      "grad_norm": 2.401409387588501,
      "learning_rate": 4.594151772190557e-06,
      "loss": 0.5924,
      "step": 66270
    },
    {
      "epoch": 27.13,
      "grad_norm": 1.8972067832946777,
      "learning_rate": 4.5941074441010494e-06,
      "loss": 0.5649,
      "step": 66280
    },
    {
      "epoch": 27.13,
      "grad_norm": 2.4941787719726562,
      "learning_rate": 4.5940631057030035e-06,
      "loss": 0.5699,
      "step": 66290
    },
    {
      "epoch": 27.14,
      "grad_norm": 2.093916893005371,
      "learning_rate": 4.594018756996622e-06,
      "loss": 0.5637,
      "step": 66300
    },
    {
      "epoch": 27.14,
      "grad_norm": 3.1496784687042236,
      "learning_rate": 4.593974397982107e-06,
      "loss": 0.5764,
      "step": 66310
    },
    {
      "epoch": 27.15,
      "grad_norm": 1.9216974973678589,
      "learning_rate": 4.593930028659664e-06,
      "loss": 0.6012,
      "step": 66320
    },
    {
      "epoch": 27.15,
      "grad_norm": 2.7372944355010986,
      "learning_rate": 4.593885649029493e-06,
      "loss": 0.5666,
      "step": 66330
    },
    {
      "epoch": 27.16,
      "grad_norm": 2.1159775257110596,
      "learning_rate": 4.5938412590918e-06,
      "loss": 0.599,
      "step": 66340
    },
    {
      "epoch": 27.16,
      "grad_norm": 2.2961678504943848,
      "learning_rate": 4.5937968588467875e-06,
      "loss": 0.566,
      "step": 66350
    },
    {
      "epoch": 27.16,
      "grad_norm": 2.5653557777404785,
      "learning_rate": 4.5937524482946585e-06,
      "loss": 0.5975,
      "step": 66360
    },
    {
      "epoch": 27.17,
      "grad_norm": 2.4457263946533203,
      "learning_rate": 4.593708027435617e-06,
      "loss": 0.5771,
      "step": 66370
    },
    {
      "epoch": 27.17,
      "grad_norm": 2.373891592025757,
      "learning_rate": 4.593663596269867e-06,
      "loss": 0.5861,
      "step": 66380
    },
    {
      "epoch": 27.18,
      "grad_norm": 2.238943099975586,
      "learning_rate": 4.59361915479761e-06,
      "loss": 0.5951,
      "step": 66390
    },
    {
      "epoch": 27.18,
      "grad_norm": 2.2288432121276855,
      "learning_rate": 4.593574703019051e-06,
      "loss": 0.5534,
      "step": 66400
    },
    {
      "epoch": 27.18,
      "grad_norm": 3.3182647228240967,
      "learning_rate": 4.593530240934393e-06,
      "loss": 0.5949,
      "step": 66410
    },
    {
      "epoch": 27.19,
      "grad_norm": 2.0192487239837646,
      "learning_rate": 4.59348576854384e-06,
      "loss": 0.5668,
      "step": 66420
    },
    {
      "epoch": 27.19,
      "grad_norm": 2.703206777572632,
      "learning_rate": 4.593441285847598e-06,
      "loss": 0.5879,
      "step": 66430
    },
    {
      "epoch": 27.2,
      "grad_norm": 2.406898021697998,
      "learning_rate": 4.593396792845867e-06,
      "loss": 0.5761,
      "step": 66440
    },
    {
      "epoch": 27.2,
      "grad_norm": 2.357755661010742,
      "learning_rate": 4.593352289538853e-06,
      "loss": 0.5777,
      "step": 66450
    },
    {
      "epoch": 27.2,
      "grad_norm": 2.8086278438568115,
      "learning_rate": 4.593307775926759e-06,
      "loss": 0.5669,
      "step": 66460
    },
    {
      "epoch": 27.21,
      "grad_norm": 2.2488162517547607,
      "learning_rate": 4.593263252009789e-06,
      "loss": 0.5901,
      "step": 66470
    },
    {
      "epoch": 27.21,
      "grad_norm": 3.0034029483795166,
      "learning_rate": 4.593218717788148e-06,
      "loss": 0.5907,
      "step": 66480
    },
    {
      "epoch": 27.22,
      "grad_norm": 2.6778054237365723,
      "learning_rate": 4.593174173262038e-06,
      "loss": 0.5957,
      "step": 66490
    },
    {
      "epoch": 27.22,
      "grad_norm": 2.0931520462036133,
      "learning_rate": 4.593129618431665e-06,
      "loss": 0.6086,
      "step": 66500
    },
    {
      "epoch": 27.22,
      "grad_norm": 2.676164150238037,
      "learning_rate": 4.5930850532972315e-06,
      "loss": 0.5936,
      "step": 66510
    },
    {
      "epoch": 27.23,
      "grad_norm": 3.5133461952209473,
      "learning_rate": 4.593040477858944e-06,
      "loss": 0.5986,
      "step": 66520
    },
    {
      "epoch": 27.23,
      "grad_norm": 2.0953283309936523,
      "learning_rate": 4.592995892117004e-06,
      "loss": 0.5917,
      "step": 66530
    },
    {
      "epoch": 27.24,
      "grad_norm": 2.9142866134643555,
      "learning_rate": 4.592951296071617e-06,
      "loss": 0.5964,
      "step": 66540
    },
    {
      "epoch": 27.24,
      "grad_norm": 2.1057260036468506,
      "learning_rate": 4.592906689722987e-06,
      "loss": 0.5686,
      "step": 66550
    },
    {
      "epoch": 27.25,
      "grad_norm": 2.138216495513916,
      "learning_rate": 4.59286207307132e-06,
      "loss": 0.5766,
      "step": 66560
    },
    {
      "epoch": 27.25,
      "grad_norm": 2.3695709705352783,
      "learning_rate": 4.592817446116817e-06,
      "loss": 0.588,
      "step": 66570
    },
    {
      "epoch": 27.25,
      "grad_norm": 2.683547258377075,
      "learning_rate": 4.592772808859685e-06,
      "loss": 0.5838,
      "step": 66580
    },
    {
      "epoch": 27.26,
      "grad_norm": 2.458611249923706,
      "learning_rate": 4.5927281613001276e-06,
      "loss": 0.5964,
      "step": 66590
    },
    {
      "epoch": 27.26,
      "grad_norm": 2.4661340713500977,
      "learning_rate": 4.592683503438351e-06,
      "loss": 0.58,
      "step": 66600
    },
    {
      "epoch": 27.27,
      "grad_norm": 1.8298261165618896,
      "learning_rate": 4.592638835274557e-06,
      "loss": 0.5689,
      "step": 66610
    },
    {
      "epoch": 27.27,
      "grad_norm": 2.752507448196411,
      "learning_rate": 4.5925941568089515e-06,
      "loss": 0.6038,
      "step": 66620
    },
    {
      "epoch": 27.27,
      "grad_norm": 2.5824804306030273,
      "learning_rate": 4.59254946804174e-06,
      "loss": 0.5489,
      "step": 66630
    },
    {
      "epoch": 27.28,
      "grad_norm": 2.1861627101898193,
      "learning_rate": 4.592504768973126e-06,
      "loss": 0.5981,
      "step": 66640
    },
    {
      "epoch": 27.28,
      "grad_norm": 2.145334005355835,
      "learning_rate": 4.592460059603314e-06,
      "loss": 0.5833,
      "step": 66650
    },
    {
      "epoch": 27.29,
      "grad_norm": 2.916844129562378,
      "learning_rate": 4.59241533993251e-06,
      "loss": 0.5775,
      "step": 66660
    },
    {
      "epoch": 27.29,
      "grad_norm": 2.142490863800049,
      "learning_rate": 4.592370609960918e-06,
      "loss": 0.5859,
      "step": 66670
    },
    {
      "epoch": 27.29,
      "grad_norm": 2.351132869720459,
      "learning_rate": 4.5923258696887434e-06,
      "loss": 0.5563,
      "step": 66680
    },
    {
      "epoch": 27.3,
      "grad_norm": 2.1197328567504883,
      "learning_rate": 4.592281119116192e-06,
      "loss": 0.5667,
      "step": 66690
    },
    {
      "epoch": 27.3,
      "grad_norm": 2.26568341255188,
      "learning_rate": 4.592236358243467e-06,
      "loss": 0.5859,
      "step": 66700
    },
    {
      "epoch": 27.31,
      "grad_norm": 2.457569122314453,
      "learning_rate": 4.592191587070773e-06,
      "loss": 0.5624,
      "step": 66710
    },
    {
      "epoch": 27.31,
      "grad_norm": 2.033147096633911,
      "learning_rate": 4.592146805598318e-06,
      "loss": 0.5727,
      "step": 66720
    },
    {
      "epoch": 27.31,
      "grad_norm": 2.2898826599121094,
      "learning_rate": 4.592102013826304e-06,
      "loss": 0.5962,
      "step": 66730
    },
    {
      "epoch": 27.32,
      "grad_norm": 1.9584959745407104,
      "learning_rate": 4.59205721175494e-06,
      "loss": 0.5742,
      "step": 66740
    },
    {
      "epoch": 27.32,
      "grad_norm": 2.425617218017578,
      "learning_rate": 4.592012399384427e-06,
      "loss": 0.603,
      "step": 66750
    },
    {
      "epoch": 27.33,
      "grad_norm": 2.0847063064575195,
      "learning_rate": 4.5919675767149716e-06,
      "loss": 0.5947,
      "step": 66760
    },
    {
      "epoch": 27.33,
      "grad_norm": 2.704186201095581,
      "learning_rate": 4.591922743746781e-06,
      "loss": 0.565,
      "step": 66770
    },
    {
      "epoch": 27.34,
      "grad_norm": 2.415328025817871,
      "learning_rate": 4.591877900480059e-06,
      "loss": 0.5716,
      "step": 66780
    },
    {
      "epoch": 27.34,
      "grad_norm": 1.9384593963623047,
      "learning_rate": 4.591833046915011e-06,
      "loss": 0.5676,
      "step": 66790
    },
    {
      "epoch": 27.34,
      "grad_norm": 2.5692994594573975,
      "learning_rate": 4.5917881830518425e-06,
      "loss": 0.5992,
      "step": 66800
    },
    {
      "epoch": 27.35,
      "grad_norm": 3.357964038848877,
      "learning_rate": 4.591743308890759e-06,
      "loss": 0.5877,
      "step": 66810
    },
    {
      "epoch": 27.35,
      "grad_norm": 3.3223683834075928,
      "learning_rate": 4.591698424431967e-06,
      "loss": 0.5992,
      "step": 66820
    },
    {
      "epoch": 27.36,
      "grad_norm": 2.480137586593628,
      "learning_rate": 4.591653529675672e-06,
      "loss": 0.5673,
      "step": 66830
    },
    {
      "epoch": 27.36,
      "grad_norm": 3.3118607997894287,
      "learning_rate": 4.591608624622078e-06,
      "loss": 0.5818,
      "step": 66840
    },
    {
      "epoch": 27.36,
      "grad_norm": 2.5160980224609375,
      "learning_rate": 4.5915637092713925e-06,
      "loss": 0.5589,
      "step": 66850
    },
    {
      "epoch": 27.37,
      "grad_norm": 2.4273886680603027,
      "learning_rate": 4.5915187836238195e-06,
      "loss": 0.5716,
      "step": 66860
    },
    {
      "epoch": 27.37,
      "grad_norm": 2.5923895835876465,
      "learning_rate": 4.591473847679566e-06,
      "loss": 0.5967,
      "step": 66870
    },
    {
      "epoch": 27.38,
      "grad_norm": 2.362306594848633,
      "learning_rate": 4.591428901438839e-06,
      "loss": 0.5952,
      "step": 66880
    },
    {
      "epoch": 27.38,
      "grad_norm": 2.241628408432007,
      "learning_rate": 4.591383944901842e-06,
      "loss": 0.5874,
      "step": 66890
    },
    {
      "epoch": 27.38,
      "grad_norm": 2.7960996627807617,
      "learning_rate": 4.591338978068782e-06,
      "loss": 0.5707,
      "step": 66900
    },
    {
      "epoch": 27.39,
      "grad_norm": 2.2409772872924805,
      "learning_rate": 4.591294000939864e-06,
      "loss": 0.5916,
      "step": 66910
    },
    {
      "epoch": 27.39,
      "grad_norm": 2.720475196838379,
      "learning_rate": 4.591249013515297e-06,
      "loss": 0.5835,
      "step": 66920
    },
    {
      "epoch": 27.4,
      "grad_norm": 2.3943779468536377,
      "learning_rate": 4.5912040157952835e-06,
      "loss": 0.5868,
      "step": 66930
    },
    {
      "epoch": 27.4,
      "grad_norm": 2.132868528366089,
      "learning_rate": 4.591159007780032e-06,
      "loss": 0.6009,
      "step": 66940
    },
    {
      "epoch": 27.4,
      "grad_norm": 2.6898105144500732,
      "learning_rate": 4.591113989469748e-06,
      "loss": 0.5806,
      "step": 66950
    },
    {
      "epoch": 27.41,
      "grad_norm": 2.1587483882904053,
      "learning_rate": 4.591068960864637e-06,
      "loss": 0.5755,
      "step": 66960
    },
    {
      "epoch": 27.41,
      "grad_norm": 2.051943063735962,
      "learning_rate": 4.5910239219649065e-06,
      "loss": 0.5903,
      "step": 66970
    },
    {
      "epoch": 27.42,
      "grad_norm": 2.1157443523406982,
      "learning_rate": 4.590978872770762e-06,
      "loss": 0.5766,
      "step": 66980
    },
    {
      "epoch": 27.42,
      "grad_norm": 2.2556047439575195,
      "learning_rate": 4.590933813282409e-06,
      "loss": 0.5678,
      "step": 66990
    },
    {
      "epoch": 27.43,
      "grad_norm": 2.5993103981018066,
      "learning_rate": 4.590888743500057e-06,
      "loss": 0.5907,
      "step": 67000
    },
    {
      "epoch": 27.43,
      "grad_norm": 2.303706169128418,
      "learning_rate": 4.590843663423909e-06,
      "loss": 0.5762,
      "step": 67010
    },
    {
      "epoch": 27.43,
      "grad_norm": 2.4515163898468018,
      "learning_rate": 4.590798573054174e-06,
      "loss": 0.5825,
      "step": 67020
    },
    {
      "epoch": 27.44,
      "grad_norm": 2.0109004974365234,
      "learning_rate": 4.590753472391056e-06,
      "loss": 0.5691,
      "step": 67030
    },
    {
      "epoch": 27.44,
      "grad_norm": 1.921759843826294,
      "learning_rate": 4.590708361434764e-06,
      "loss": 0.5654,
      "step": 67040
    },
    {
      "epoch": 27.45,
      "grad_norm": 3.2124011516571045,
      "learning_rate": 4.590663240185503e-06,
      "loss": 0.6047,
      "step": 67050
    },
    {
      "epoch": 27.45,
      "grad_norm": 2.9316742420196533,
      "learning_rate": 4.590618108643482e-06,
      "loss": 0.5889,
      "step": 67060
    },
    {
      "epoch": 27.45,
      "grad_norm": 2.710090398788452,
      "learning_rate": 4.590572966808905e-06,
      "loss": 0.6008,
      "step": 67070
    },
    {
      "epoch": 27.46,
      "grad_norm": 2.587642192840576,
      "learning_rate": 4.59052781468198e-06,
      "loss": 0.6023,
      "step": 67080
    },
    {
      "epoch": 27.46,
      "grad_norm": 2.766483783721924,
      "learning_rate": 4.5904826522629145e-06,
      "loss": 0.5825,
      "step": 67090
    },
    {
      "epoch": 27.47,
      "grad_norm": 1.8637744188308716,
      "learning_rate": 4.590437479551913e-06,
      "loss": 0.5869,
      "step": 67100
    },
    {
      "epoch": 27.47,
      "grad_norm": 2.6341302394866943,
      "learning_rate": 4.590392296549186e-06,
      "loss": 0.5814,
      "step": 67110
    },
    {
      "epoch": 27.47,
      "grad_norm": 2.431735038757324,
      "learning_rate": 4.590347103254938e-06,
      "loss": 0.5703,
      "step": 67120
    },
    {
      "epoch": 27.48,
      "grad_norm": 2.848735809326172,
      "learning_rate": 4.590301899669376e-06,
      "loss": 0.602,
      "step": 67130
    },
    {
      "epoch": 27.48,
      "grad_norm": 2.1094768047332764,
      "learning_rate": 4.590256685792709e-06,
      "loss": 0.5779,
      "step": 67140
    },
    {
      "epoch": 27.49,
      "grad_norm": 2.6847357749938965,
      "learning_rate": 4.590211461625141e-06,
      "loss": 0.6233,
      "step": 67150
    },
    {
      "epoch": 27.49,
      "grad_norm": 3.005518913269043,
      "learning_rate": 4.5901662271668825e-06,
      "loss": 0.5748,
      "step": 67160
    },
    {
      "epoch": 27.49,
      "grad_norm": 2.2139108180999756,
      "learning_rate": 4.590120982418137e-06,
      "loss": 0.5505,
      "step": 67170
    },
    {
      "epoch": 27.5,
      "grad_norm": 3.263174295425415,
      "learning_rate": 4.590075727379116e-06,
      "loss": 0.5993,
      "step": 67180
    },
    {
      "epoch": 27.5,
      "grad_norm": 2.0546412467956543,
      "learning_rate": 4.590030462050025e-06,
      "loss": 0.5867,
      "step": 67190
    },
    {
      "epoch": 27.51,
      "grad_norm": 3.4921202659606934,
      "learning_rate": 4.58998518643107e-06,
      "loss": 0.5805,
      "step": 67200
    },
    {
      "epoch": 27.51,
      "grad_norm": 2.0846519470214844,
      "learning_rate": 4.5899399005224594e-06,
      "loss": 0.6002,
      "step": 67210
    },
    {
      "epoch": 27.52,
      "grad_norm": 2.7269785404205322,
      "learning_rate": 4.589894604324401e-06,
      "loss": 0.5837,
      "step": 67220
    },
    {
      "epoch": 27.52,
      "grad_norm": 3.751565933227539,
      "learning_rate": 4.589849297837102e-06,
      "loss": 0.5654,
      "step": 67230
    },
    {
      "epoch": 27.52,
      "grad_norm": 2.700023651123047,
      "learning_rate": 4.58980398106077e-06,
      "loss": 0.5928,
      "step": 67240
    },
    {
      "epoch": 27.53,
      "grad_norm": 3.2694191932678223,
      "learning_rate": 4.589758653995613e-06,
      "loss": 0.5932,
      "step": 67250
    },
    {
      "epoch": 27.53,
      "grad_norm": 3.0028600692749023,
      "learning_rate": 4.589713316641837e-06,
      "loss": 0.578,
      "step": 67260
    },
    {
      "epoch": 27.54,
      "grad_norm": 2.9158754348754883,
      "learning_rate": 4.589667968999652e-06,
      "loss": 0.5692,
      "step": 67270
    },
    {
      "epoch": 27.54,
      "grad_norm": 2.436375617980957,
      "learning_rate": 4.589622611069264e-06,
      "loss": 0.5909,
      "step": 67280
    },
    {
      "epoch": 27.54,
      "grad_norm": 2.3100390434265137,
      "learning_rate": 4.589577242850882e-06,
      "loss": 0.5788,
      "step": 67290
    },
    {
      "epoch": 27.55,
      "grad_norm": 2.8782601356506348,
      "learning_rate": 4.589531864344713e-06,
      "loss": 0.5923,
      "step": 67300
    },
    {
      "epoch": 27.55,
      "grad_norm": 2.8071908950805664,
      "learning_rate": 4.589486475550965e-06,
      "loss": 0.5809,
      "step": 67310
    },
    {
      "epoch": 27.56,
      "grad_norm": 2.7333338260650635,
      "learning_rate": 4.589441076469846e-06,
      "loss": 0.5856,
      "step": 67320
    },
    {
      "epoch": 27.56,
      "grad_norm": 3.1800804138183594,
      "learning_rate": 4.589395667101564e-06,
      "loss": 0.59,
      "step": 67330
    },
    {
      "epoch": 27.56,
      "grad_norm": 3.2731645107269287,
      "learning_rate": 4.589350247446326e-06,
      "loss": 0.5919,
      "step": 67340
    },
    {
      "epoch": 27.57,
      "grad_norm": 2.3218417167663574,
      "learning_rate": 4.589304817504342e-06,
      "loss": 0.589,
      "step": 67350
    },
    {
      "epoch": 27.57,
      "grad_norm": 3.222141981124878,
      "learning_rate": 4.58925937727582e-06,
      "loss": 0.5736,
      "step": 67360
    },
    {
      "epoch": 27.58,
      "grad_norm": 2.523014545440674,
      "learning_rate": 4.589213926760966e-06,
      "loss": 0.5886,
      "step": 67370
    },
    {
      "epoch": 27.58,
      "grad_norm": 3.1015572547912598,
      "learning_rate": 4.58916846595999e-06,
      "loss": 0.5806,
      "step": 67380
    },
    {
      "epoch": 27.58,
      "grad_norm": 3.851696252822876,
      "learning_rate": 4.5891229948730995e-06,
      "loss": 0.5852,
      "step": 67390
    },
    {
      "epoch": 27.59,
      "grad_norm": 2.6097700595855713,
      "learning_rate": 4.589077513500503e-06,
      "loss": 0.5789,
      "step": 67400
    },
    {
      "epoch": 27.59,
      "grad_norm": 2.557791233062744,
      "learning_rate": 4.5890320218424094e-06,
      "loss": 0.5832,
      "step": 67410
    },
    {
      "epoch": 27.6,
      "grad_norm": 2.6645925045013428,
      "learning_rate": 4.588986519899026e-06,
      "loss": 0.5845,
      "step": 67420
    },
    {
      "epoch": 27.6,
      "grad_norm": 2.503688335418701,
      "learning_rate": 4.5889410076705625e-06,
      "loss": 0.5986,
      "step": 67430
    },
    {
      "epoch": 27.61,
      "grad_norm": 2.2636020183563232,
      "learning_rate": 4.588895485157226e-06,
      "loss": 0.5964,
      "step": 67440
    },
    {
      "epoch": 27.61,
      "grad_norm": 2.9520375728607178,
      "learning_rate": 4.588849952359226e-06,
      "loss": 0.5744,
      "step": 67450
    },
    {
      "epoch": 27.61,
      "grad_norm": 2.588653802871704,
      "learning_rate": 4.588804409276771e-06,
      "loss": 0.5742,
      "step": 67460
    },
    {
      "epoch": 27.62,
      "grad_norm": 2.44184947013855,
      "learning_rate": 4.5887588559100695e-06,
      "loss": 0.5965,
      "step": 67470
    },
    {
      "epoch": 27.62,
      "grad_norm": 2.0838496685028076,
      "learning_rate": 4.58871329225933e-06,
      "loss": 0.5703,
      "step": 67480
    },
    {
      "epoch": 27.63,
      "grad_norm": 2.6264076232910156,
      "learning_rate": 4.5886677183247615e-06,
      "loss": 0.5658,
      "step": 67490
    },
    {
      "epoch": 27.63,
      "grad_norm": 2.1799919605255127,
      "learning_rate": 4.588622134106572e-06,
      "loss": 0.5794,
      "step": 67500
    },
    {
      "epoch": 27.63,
      "grad_norm": 2.9824047088623047,
      "learning_rate": 4.588576539604971e-06,
      "loss": 0.6044,
      "step": 67510
    },
    {
      "epoch": 27.64,
      "grad_norm": 2.864619016647339,
      "learning_rate": 4.5885309348201676e-06,
      "loss": 0.5826,
      "step": 67520
    },
    {
      "epoch": 27.64,
      "grad_norm": 2.674821615219116,
      "learning_rate": 4.58848531975237e-06,
      "loss": 0.5782,
      "step": 67530
    },
    {
      "epoch": 27.65,
      "grad_norm": 2.8309152126312256,
      "learning_rate": 4.5884396944017886e-06,
      "loss": 0.5717,
      "step": 67540
    },
    {
      "epoch": 27.65,
      "grad_norm": 2.1712000370025635,
      "learning_rate": 4.58839405876863e-06,
      "loss": 0.5866,
      "step": 67550
    },
    {
      "epoch": 27.65,
      "grad_norm": 2.387674570083618,
      "learning_rate": 4.588348412853105e-06,
      "loss": 0.5786,
      "step": 67560
    },
    {
      "epoch": 27.66,
      "grad_norm": 2.800722599029541,
      "learning_rate": 4.588302756655423e-06,
      "loss": 0.6026,
      "step": 67570
    },
    {
      "epoch": 27.66,
      "grad_norm": 2.8325142860412598,
      "learning_rate": 4.588257090175792e-06,
      "loss": 0.6083,
      "step": 67580
    },
    {
      "epoch": 27.67,
      "grad_norm": 2.670851945877075,
      "learning_rate": 4.588211413414421e-06,
      "loss": 0.5682,
      "step": 67590
    },
    {
      "epoch": 27.67,
      "grad_norm": 2.433642625808716,
      "learning_rate": 4.588165726371521e-06,
      "loss": 0.5895,
      "step": 67600
    },
    {
      "epoch": 27.67,
      "grad_norm": 2.133350133895874,
      "learning_rate": 4.5881200290473e-06,
      "loss": 0.5803,
      "step": 67610
    },
    {
      "epoch": 27.68,
      "grad_norm": 2.356417179107666,
      "learning_rate": 4.588074321441967e-06,
      "loss": 0.5981,
      "step": 67620
    },
    {
      "epoch": 27.68,
      "grad_norm": 2.20060133934021,
      "learning_rate": 4.588028603555732e-06,
      "loss": 0.5787,
      "step": 67630
    },
    {
      "epoch": 27.69,
      "grad_norm": 2.634934902191162,
      "learning_rate": 4.587982875388804e-06,
      "loss": 0.5899,
      "step": 67640
    },
    {
      "epoch": 27.69,
      "grad_norm": 2.239924430847168,
      "learning_rate": 4.587937136941393e-06,
      "loss": 0.575,
      "step": 67650
    },
    {
      "epoch": 27.7,
      "grad_norm": 1.973973274230957,
      "learning_rate": 4.5878913882137075e-06,
      "loss": 0.561,
      "step": 67660
    },
    {
      "epoch": 27.7,
      "grad_norm": 2.4688355922698975,
      "learning_rate": 4.587845629205959e-06,
      "loss": 0.5886,
      "step": 67670
    },
    {
      "epoch": 27.7,
      "grad_norm": 2.9080984592437744,
      "learning_rate": 4.587799859918356e-06,
      "loss": 0.5769,
      "step": 67680
    },
    {
      "epoch": 27.71,
      "grad_norm": 1.9092503786087036,
      "learning_rate": 4.587754080351107e-06,
      "loss": 0.5972,
      "step": 67690
    },
    {
      "epoch": 27.71,
      "grad_norm": 1.8105696439743042,
      "learning_rate": 4.587708290504424e-06,
      "loss": 0.5881,
      "step": 67700
    },
    {
      "epoch": 27.72,
      "grad_norm": 3.1496481895446777,
      "learning_rate": 4.587662490378515e-06,
      "loss": 0.5866,
      "step": 67710
    },
    {
      "epoch": 27.72,
      "grad_norm": 3.38995361328125,
      "learning_rate": 4.587616679973591e-06,
      "loss": 0.5854,
      "step": 67720
    },
    {
      "epoch": 27.72,
      "grad_norm": 2.057680130004883,
      "learning_rate": 4.58757085928986e-06,
      "loss": 0.5872,
      "step": 67730
    },
    {
      "epoch": 27.73,
      "grad_norm": 1.9775458574295044,
      "learning_rate": 4.5875250283275335e-06,
      "loss": 0.5901,
      "step": 67740
    },
    {
      "epoch": 27.73,
      "grad_norm": 2.675952911376953,
      "learning_rate": 4.587479187086822e-06,
      "loss": 0.5665,
      "step": 67750
    },
    {
      "epoch": 27.74,
      "grad_norm": 2.2983102798461914,
      "learning_rate": 4.587433335567934e-06,
      "loss": 0.577,
      "step": 67760
    },
    {
      "epoch": 27.74,
      "grad_norm": 2.1365251541137695,
      "learning_rate": 4.58738747377108e-06,
      "loss": 0.6067,
      "step": 67770
    },
    {
      "epoch": 27.74,
      "grad_norm": 1.7172960042953491,
      "learning_rate": 4.58734160169647e-06,
      "loss": 0.5898,
      "step": 67780
    },
    {
      "epoch": 27.75,
      "grad_norm": 3.1542935371398926,
      "learning_rate": 4.587295719344315e-06,
      "loss": 0.5993,
      "step": 67790
    },
    {
      "epoch": 27.75,
      "grad_norm": 2.3610424995422363,
      "learning_rate": 4.587249826714825e-06,
      "loss": 0.5818,
      "step": 67800
    },
    {
      "epoch": 27.76,
      "grad_norm": 2.222867250442505,
      "learning_rate": 4.587203923808208e-06,
      "loss": 0.5886,
      "step": 67810
    },
    {
      "epoch": 27.76,
      "grad_norm": 1.9055564403533936,
      "learning_rate": 4.587158010624677e-06,
      "loss": 0.5548,
      "step": 67820
    },
    {
      "epoch": 27.77,
      "grad_norm": 1.8213815689086914,
      "learning_rate": 4.587112087164442e-06,
      "loss": 0.5832,
      "step": 67830
    },
    {
      "epoch": 27.77,
      "grad_norm": 2.387878894805908,
      "learning_rate": 4.587066153427712e-06,
      "loss": 0.6061,
      "step": 67840
    },
    {
      "epoch": 27.77,
      "grad_norm": 2.5530383586883545,
      "learning_rate": 4.587020209414698e-06,
      "loss": 0.5995,
      "step": 67850
    },
    {
      "epoch": 27.78,
      "grad_norm": 2.6031405925750732,
      "learning_rate": 4.5869742551256105e-06,
      "loss": 0.5985,
      "step": 67860
    },
    {
      "epoch": 27.78,
      "grad_norm": 2.5875816345214844,
      "learning_rate": 4.58692829056066e-06,
      "loss": 0.5865,
      "step": 67870
    },
    {
      "epoch": 27.79,
      "grad_norm": 3.1013500690460205,
      "learning_rate": 4.586882315720058e-06,
      "loss": 0.5924,
      "step": 67880
    },
    {
      "epoch": 27.79,
      "grad_norm": 2.341179370880127,
      "learning_rate": 4.586836330604014e-06,
      "loss": 0.5897,
      "step": 67890
    },
    {
      "epoch": 27.79,
      "grad_norm": 2.7049591541290283,
      "learning_rate": 4.586790335212738e-06,
      "loss": 0.5843,
      "step": 67900
    },
    {
      "epoch": 27.8,
      "grad_norm": 2.615035057067871,
      "learning_rate": 4.586744329546443e-06,
      "loss": 0.5688,
      "step": 67910
    },
    {
      "epoch": 27.8,
      "grad_norm": 3.3685455322265625,
      "learning_rate": 4.586698313605337e-06,
      "loss": 0.5831,
      "step": 67920
    },
    {
      "epoch": 27.81,
      "grad_norm": 2.6254289150238037,
      "learning_rate": 4.586652287389633e-06,
      "loss": 0.5871,
      "step": 67930
    },
    {
      "epoch": 27.81,
      "grad_norm": 2.3152973651885986,
      "learning_rate": 4.586606250899541e-06,
      "loss": 0.5856,
      "step": 67940
    },
    {
      "epoch": 27.81,
      "grad_norm": 2.292336940765381,
      "learning_rate": 4.5865602041352715e-06,
      "loss": 0.5691,
      "step": 67950
    },
    {
      "epoch": 27.82,
      "grad_norm": 2.783828020095825,
      "learning_rate": 4.586514147097035e-06,
      "loss": 0.578,
      "step": 67960
    },
    {
      "epoch": 27.82,
      "grad_norm": 2.2893290519714355,
      "learning_rate": 4.5864680797850445e-06,
      "loss": 0.5564,
      "step": 67970
    },
    {
      "epoch": 27.83,
      "grad_norm": 2.0337512493133545,
      "learning_rate": 4.58642200219951e-06,
      "loss": 0.5809,
      "step": 67980
    },
    {
      "epoch": 27.83,
      "grad_norm": 2.504361152648926,
      "learning_rate": 4.586375914340642e-06,
      "loss": 0.5961,
      "step": 67990
    },
    {
      "epoch": 27.83,
      "grad_norm": 3.0243399143218994,
      "learning_rate": 4.586329816208651e-06,
      "loss": 0.5835,
      "step": 68000
    },
    {
      "epoch": 27.84,
      "grad_norm": 2.9374454021453857,
      "learning_rate": 4.58628370780375e-06,
      "loss": 0.5937,
      "step": 68010
    },
    {
      "epoch": 27.84,
      "grad_norm": 2.0671401023864746,
      "learning_rate": 4.58623758912615e-06,
      "loss": 0.5812,
      "step": 68020
    },
    {
      "epoch": 27.85,
      "grad_norm": 2.479219913482666,
      "learning_rate": 4.586191460176061e-06,
      "loss": 0.5771,
      "step": 68030
    },
    {
      "epoch": 27.85,
      "grad_norm": 2.865544319152832,
      "learning_rate": 4.586145320953695e-06,
      "loss": 0.5717,
      "step": 68040
    },
    {
      "epoch": 27.86,
      "grad_norm": 2.0403738021850586,
      "learning_rate": 4.5860991714592635e-06,
      "loss": 0.5856,
      "step": 68050
    },
    {
      "epoch": 27.86,
      "grad_norm": 2.3207826614379883,
      "learning_rate": 4.586053011692977e-06,
      "loss": 0.5832,
      "step": 68060
    },
    {
      "epoch": 27.86,
      "grad_norm": 3.0370986461639404,
      "learning_rate": 4.586006841655048e-06,
      "loss": 0.5862,
      "step": 68070
    },
    {
      "epoch": 27.87,
      "grad_norm": 2.3879458904266357,
      "learning_rate": 4.585960661345689e-06,
      "loss": 0.5763,
      "step": 68080
    },
    {
      "epoch": 27.87,
      "grad_norm": 2.329840898513794,
      "learning_rate": 4.5859144707651085e-06,
      "loss": 0.5759,
      "step": 68090
    },
    {
      "epoch": 27.88,
      "grad_norm": 2.516235828399658,
      "learning_rate": 4.58586826991352e-06,
      "loss": 0.5832,
      "step": 68100
    },
    {
      "epoch": 27.88,
      "grad_norm": 3.4155871868133545,
      "learning_rate": 4.585822058791135e-06,
      "loss": 0.5734,
      "step": 68110
    },
    {
      "epoch": 27.88,
      "grad_norm": 2.307238817214966,
      "learning_rate": 4.585775837398166e-06,
      "loss": 0.5646,
      "step": 68120
    },
    {
      "epoch": 27.89,
      "grad_norm": 2.7684736251831055,
      "learning_rate": 4.585729605734824e-06,
      "loss": 0.5748,
      "step": 68130
    },
    {
      "epoch": 27.89,
      "grad_norm": 2.0953853130340576,
      "learning_rate": 4.585683363801319e-06,
      "loss": 0.5858,
      "step": 68140
    },
    {
      "epoch": 27.9,
      "grad_norm": 2.912316083908081,
      "learning_rate": 4.585637111597866e-06,
      "loss": 0.5762,
      "step": 68150
    },
    {
      "epoch": 27.9,
      "grad_norm": 2.381852149963379,
      "learning_rate": 4.585590849124675e-06,
      "loss": 0.575,
      "step": 68160
    },
    {
      "epoch": 27.9,
      "grad_norm": 3.4596807956695557,
      "learning_rate": 4.585544576381958e-06,
      "loss": 0.57,
      "step": 68170
    },
    {
      "epoch": 27.91,
      "grad_norm": 4.10275936126709,
      "learning_rate": 4.585498293369927e-06,
      "loss": 0.5817,
      "step": 68180
    },
    {
      "epoch": 27.91,
      "grad_norm": 2.5246946811676025,
      "learning_rate": 4.585452000088794e-06,
      "loss": 0.5786,
      "step": 68190
    },
    {
      "epoch": 27.92,
      "grad_norm": 2.590128183364868,
      "learning_rate": 4.585405696538772e-06,
      "loss": 0.5887,
      "step": 68200
    },
    {
      "epoch": 27.92,
      "grad_norm": 2.689340829849243,
      "learning_rate": 4.5853593827200725e-06,
      "loss": 0.586,
      "step": 68210
    },
    {
      "epoch": 27.92,
      "grad_norm": 2.4298832416534424,
      "learning_rate": 4.585313058632907e-06,
      "loss": 0.5689,
      "step": 68220
    },
    {
      "epoch": 27.93,
      "grad_norm": 2.1441571712493896,
      "learning_rate": 4.585266724277489e-06,
      "loss": 0.5833,
      "step": 68230
    },
    {
      "epoch": 27.93,
      "grad_norm": 2.0663609504699707,
      "learning_rate": 4.585220379654029e-06,
      "loss": 0.5754,
      "step": 68240
    },
    {
      "epoch": 27.94,
      "grad_norm": 2.325388193130493,
      "learning_rate": 4.585174024762742e-06,
      "loss": 0.5759,
      "step": 68250
    },
    {
      "epoch": 27.94,
      "grad_norm": 2.5987746715545654,
      "learning_rate": 4.585127659603837e-06,
      "loss": 0.5665,
      "step": 68260
    },
    {
      "epoch": 27.95,
      "grad_norm": 2.251317024230957,
      "learning_rate": 4.585081284177529e-06,
      "loss": 0.5792,
      "step": 68270
    },
    {
      "epoch": 27.95,
      "grad_norm": 2.208597183227539,
      "learning_rate": 4.585034898484029e-06,
      "loss": 0.5827,
      "step": 68280
    },
    {
      "epoch": 27.95,
      "grad_norm": 3.04976487159729,
      "learning_rate": 4.5849885025235495e-06,
      "loss": 0.5909,
      "step": 68290
    },
    {
      "epoch": 27.96,
      "grad_norm": 2.5523502826690674,
      "learning_rate": 4.584942096296304e-06,
      "loss": 0.5774,
      "step": 68300
    },
    {
      "epoch": 27.96,
      "grad_norm": 2.446016311645508,
      "learning_rate": 4.584895679802505e-06,
      "loss": 0.5733,
      "step": 68310
    },
    {
      "epoch": 27.97,
      "grad_norm": 2.384085178375244,
      "learning_rate": 4.584849253042365e-06,
      "loss": 0.5604,
      "step": 68320
    },
    {
      "epoch": 27.97,
      "grad_norm": 2.91213059425354,
      "learning_rate": 4.584802816016096e-06,
      "loss": 0.6023,
      "step": 68330
    },
    {
      "epoch": 27.97,
      "grad_norm": 2.6379823684692383,
      "learning_rate": 4.584756368723911e-06,
      "loss": 0.5734,
      "step": 68340
    },
    {
      "epoch": 27.98,
      "grad_norm": 3.6771862506866455,
      "learning_rate": 4.584709911166022e-06,
      "loss": 0.5889,
      "step": 68350
    },
    {
      "epoch": 27.98,
      "grad_norm": 2.9824180603027344,
      "learning_rate": 4.584663443342644e-06,
      "loss": 0.5632,
      "step": 68360
    },
    {
      "epoch": 27.99,
      "grad_norm": 1.7875107526779175,
      "learning_rate": 4.5846169652539885e-06,
      "loss": 0.594,
      "step": 68370
    },
    {
      "epoch": 27.99,
      "grad_norm": 2.2685208320617676,
      "learning_rate": 4.584570476900268e-06,
      "loss": 0.6121,
      "step": 68380
    },
    {
      "epoch": 27.99,
      "grad_norm": 2.7732417583465576,
      "learning_rate": 4.584523978281696e-06,
      "loss": 0.573,
      "step": 68390
    },
    {
      "epoch": 28.0,
      "grad_norm": 2.256373167037964,
      "learning_rate": 4.584477469398486e-06,
      "loss": 0.5846,
      "step": 68400
    },
    {
      "epoch": 28.0,
      "eval_loss": 0.5779260993003845,
      "eval_runtime": 52.1601,
      "eval_samples_per_second": 66.123,
      "eval_steps_per_second": 8.282,
      "step": 68404
    },
    {
      "epoch": 28.0,
      "grad_norm": 3.9606616497039795,
      "learning_rate": 4.584430950250849e-06,
      "loss": 0.588,
      "step": 68410
    },
    {
      "epoch": 28.01,
      "grad_norm": 2.458935499191284,
      "learning_rate": 4.584384420839001e-06,
      "loss": 0.6136,
      "step": 68420
    },
    {
      "epoch": 28.01,
      "grad_norm": 2.0916872024536133,
      "learning_rate": 4.584337881163154e-06,
      "loss": 0.5422,
      "step": 68430
    },
    {
      "epoch": 28.01,
      "grad_norm": 2.587193012237549,
      "learning_rate": 4.58429133122352e-06,
      "loss": 0.5717,
      "step": 68440
    },
    {
      "epoch": 28.02,
      "grad_norm": 3.2079200744628906,
      "learning_rate": 4.584244771020313e-06,
      "loss": 0.577,
      "step": 68450
    },
    {
      "epoch": 28.02,
      "grad_norm": 2.2806036472320557,
      "learning_rate": 4.5841982005537464e-06,
      "loss": 0.5685,
      "step": 68460
    },
    {
      "epoch": 28.03,
      "grad_norm": 2.0429046154022217,
      "learning_rate": 4.584151619824034e-06,
      "loss": 0.5768,
      "step": 68470
    },
    {
      "epoch": 28.03,
      "grad_norm": 2.2978267669677734,
      "learning_rate": 4.58410502883139e-06,
      "loss": 0.5664,
      "step": 68480
    },
    {
      "epoch": 28.04,
      "grad_norm": 2.3944411277770996,
      "learning_rate": 4.584058427576025e-06,
      "loss": 0.5734,
      "step": 68490
    },
    {
      "epoch": 28.04,
      "grad_norm": 1.9164291620254517,
      "learning_rate": 4.584011816058154e-06,
      "loss": 0.5749,
      "step": 68500
    },
    {
      "epoch": 28.04,
      "grad_norm": 2.1840522289276123,
      "learning_rate": 4.5839651942779905e-06,
      "loss": 0.5558,
      "step": 68510
    },
    {
      "epoch": 28.05,
      "grad_norm": 2.3330318927764893,
      "learning_rate": 4.583918562235748e-06,
      "loss": 0.573,
      "step": 68520
    },
    {
      "epoch": 28.05,
      "grad_norm": 2.2231051921844482,
      "learning_rate": 4.583871919931642e-06,
      "loss": 0.5888,
      "step": 68530
    },
    {
      "epoch": 28.06,
      "grad_norm": 2.9199211597442627,
      "learning_rate": 4.583825267365883e-06,
      "loss": 0.5726,
      "step": 68540
    },
    {
      "epoch": 28.06,
      "grad_norm": 2.9070305824279785,
      "learning_rate": 4.583778604538687e-06,
      "loss": 0.5922,
      "step": 68550
    },
    {
      "epoch": 28.06,
      "grad_norm": 3.110372543334961,
      "learning_rate": 4.583731931450265e-06,
      "loss": 0.5735,
      "step": 68560
    },
    {
      "epoch": 28.07,
      "grad_norm": 2.5523812770843506,
      "learning_rate": 4.583685248100834e-06,
      "loss": 0.5821,
      "step": 68570
    },
    {
      "epoch": 28.07,
      "grad_norm": 2.261971950531006,
      "learning_rate": 4.583638554490607e-06,
      "loss": 0.5836,
      "step": 68580
    },
    {
      "epoch": 28.08,
      "grad_norm": 2.307955026626587,
      "learning_rate": 4.583591850619796e-06,
      "loss": 0.58,
      "step": 68590
    },
    {
      "epoch": 28.08,
      "grad_norm": 3.0860984325408936,
      "learning_rate": 4.583545136488617e-06,
      "loss": 0.5753,
      "step": 68600
    },
    {
      "epoch": 28.08,
      "grad_norm": 1.9695097208023071,
      "learning_rate": 4.583498412097284e-06,
      "loss": 0.5919,
      "step": 68610
    },
    {
      "epoch": 28.09,
      "grad_norm": 3.635868549346924,
      "learning_rate": 4.5834516774460105e-06,
      "loss": 0.576,
      "step": 68620
    },
    {
      "epoch": 28.09,
      "grad_norm": 2.3753278255462646,
      "learning_rate": 4.5834049325350094e-06,
      "loss": 0.5772,
      "step": 68630
    },
    {
      "epoch": 28.1,
      "grad_norm": 2.877265453338623,
      "learning_rate": 4.583358177364497e-06,
      "loss": 0.5866,
      "step": 68640
    },
    {
      "epoch": 28.1,
      "grad_norm": 2.6262803077697754,
      "learning_rate": 4.583311411934685e-06,
      "loss": 0.5797,
      "step": 68650
    },
    {
      "epoch": 28.1,
      "grad_norm": 2.2838995456695557,
      "learning_rate": 4.58326463624579e-06,
      "loss": 0.6005,
      "step": 68660
    },
    {
      "epoch": 28.11,
      "grad_norm": 2.114212989807129,
      "learning_rate": 4.583217850298026e-06,
      "loss": 0.5988,
      "step": 68670
    },
    {
      "epoch": 28.11,
      "grad_norm": 2.292900800704956,
      "learning_rate": 4.583171054091606e-06,
      "loss": 0.5966,
      "step": 68680
    },
    {
      "epoch": 28.12,
      "grad_norm": 1.8580093383789062,
      "learning_rate": 4.583124247626744e-06,
      "loss": 0.5801,
      "step": 68690
    },
    {
      "epoch": 28.12,
      "grad_norm": 3.205181360244751,
      "learning_rate": 4.583077430903656e-06,
      "loss": 0.5727,
      "step": 68700
    },
    {
      "epoch": 28.13,
      "grad_norm": 2.1314306259155273,
      "learning_rate": 4.583030603922556e-06,
      "loss": 0.5917,
      "step": 68710
    },
    {
      "epoch": 28.13,
      "grad_norm": 3.368614435195923,
      "learning_rate": 4.582983766683658e-06,
      "loss": 0.5762,
      "step": 68720
    },
    {
      "epoch": 28.13,
      "grad_norm": 3.5484845638275146,
      "learning_rate": 4.582936919187178e-06,
      "loss": 0.5962,
      "step": 68730
    },
    {
      "epoch": 28.14,
      "grad_norm": 2.7648303508758545,
      "learning_rate": 4.582890061433328e-06,
      "loss": 0.6092,
      "step": 68740
    },
    {
      "epoch": 28.14,
      "grad_norm": 2.922126531600952,
      "learning_rate": 4.5828431934223246e-06,
      "loss": 0.582,
      "step": 68750
    },
    {
      "epoch": 28.15,
      "grad_norm": 2.366877794265747,
      "learning_rate": 4.582796315154383e-06,
      "loss": 0.5695,
      "step": 68760
    },
    {
      "epoch": 28.15,
      "grad_norm": 2.123638391494751,
      "learning_rate": 4.582749426629715e-06,
      "loss": 0.5689,
      "step": 68770
    },
    {
      "epoch": 28.15,
      "grad_norm": 2.342921257019043,
      "learning_rate": 4.5827025278485395e-06,
      "loss": 0.5757,
      "step": 68780
    },
    {
      "epoch": 28.16,
      "grad_norm": 2.1884241104125977,
      "learning_rate": 4.582655618811069e-06,
      "loss": 0.571,
      "step": 68790
    },
    {
      "epoch": 28.16,
      "grad_norm": 2.6117727756500244,
      "learning_rate": 4.582608699517517e-06,
      "loss": 0.5751,
      "step": 68800
    },
    {
      "epoch": 28.17,
      "grad_norm": 2.589921236038208,
      "learning_rate": 4.582561769968101e-06,
      "loss": 0.5649,
      "step": 68810
    },
    {
      "epoch": 28.17,
      "grad_norm": 3.629105806350708,
      "learning_rate": 4.582514830163035e-06,
      "loss": 0.5748,
      "step": 68820
    },
    {
      "epoch": 28.17,
      "grad_norm": 2.4400784969329834,
      "learning_rate": 4.582467880102534e-06,
      "loss": 0.5782,
      "step": 68830
    },
    {
      "epoch": 28.18,
      "grad_norm": 2.0883729457855225,
      "learning_rate": 4.582420919786814e-06,
      "loss": 0.582,
      "step": 68840
    },
    {
      "epoch": 28.18,
      "grad_norm": 2.8454530239105225,
      "learning_rate": 4.582373949216088e-06,
      "loss": 0.5696,
      "step": 68850
    },
    {
      "epoch": 28.19,
      "grad_norm": 3.140186071395874,
      "learning_rate": 4.5823269683905725e-06,
      "loss": 0.5791,
      "step": 68860
    },
    {
      "epoch": 28.19,
      "grad_norm": 2.4818856716156006,
      "learning_rate": 4.582279977310483e-06,
      "loss": 0.6067,
      "step": 68870
    },
    {
      "epoch": 28.19,
      "grad_norm": 2.0405797958374023,
      "learning_rate": 4.582232975976033e-06,
      "loss": 0.5802,
      "step": 68880
    },
    {
      "epoch": 28.2,
      "grad_norm": 2.130727529525757,
      "learning_rate": 4.582185964387442e-06,
      "loss": 0.6003,
      "step": 68890
    },
    {
      "epoch": 28.2,
      "grad_norm": 2.529867172241211,
      "learning_rate": 4.58213894254492e-06,
      "loss": 0.5845,
      "step": 68900
    },
    {
      "epoch": 28.21,
      "grad_norm": 1.9750200510025024,
      "learning_rate": 4.5820919104486854e-06,
      "loss": 0.596,
      "step": 68910
    },
    {
      "epoch": 28.21,
      "grad_norm": 2.1557984352111816,
      "learning_rate": 4.582044868098954e-06,
      "loss": 0.5812,
      "step": 68920
    },
    {
      "epoch": 28.22,
      "grad_norm": 2.331679344177246,
      "learning_rate": 4.581997815495939e-06,
      "loss": 0.5705,
      "step": 68930
    },
    {
      "epoch": 28.22,
      "grad_norm": 2.536944627761841,
      "learning_rate": 4.581950752639859e-06,
      "loss": 0.6088,
      "step": 68940
    },
    {
      "epoch": 28.22,
      "grad_norm": 3.4429142475128174,
      "learning_rate": 4.5819036795309275e-06,
      "loss": 0.5765,
      "step": 68950
    },
    {
      "epoch": 28.23,
      "grad_norm": 2.1892294883728027,
      "learning_rate": 4.581856596169361e-06,
      "loss": 0.5874,
      "step": 68960
    },
    {
      "epoch": 28.23,
      "grad_norm": 2.7762887477874756,
      "learning_rate": 4.581809502555374e-06,
      "loss": 0.5897,
      "step": 68970
    },
    {
      "epoch": 28.24,
      "grad_norm": 2.9630839824676514,
      "learning_rate": 4.581762398689183e-06,
      "loss": 0.5657,
      "step": 68980
    },
    {
      "epoch": 28.24,
      "grad_norm": 2.465541362762451,
      "learning_rate": 4.581715284571004e-06,
      "loss": 0.5649,
      "step": 68990
    },
    {
      "epoch": 28.24,
      "grad_norm": 2.2850584983825684,
      "learning_rate": 4.581668160201053e-06,
      "loss": 0.5989,
      "step": 69000
    },
    {
      "epoch": 28.25,
      "grad_norm": 3.2569031715393066,
      "learning_rate": 4.581621025579545e-06,
      "loss": 0.5884,
      "step": 69010
    },
    {
      "epoch": 28.25,
      "grad_norm": 2.073845863342285,
      "learning_rate": 4.581573880706697e-06,
      "loss": 0.5905,
      "step": 69020
    },
    {
      "epoch": 28.26,
      "grad_norm": 2.564276933670044,
      "learning_rate": 4.5815267255827235e-06,
      "loss": 0.5693,
      "step": 69030
    },
    {
      "epoch": 28.26,
      "grad_norm": 2.681135654449463,
      "learning_rate": 4.581479560207842e-06,
      "loss": 0.5499,
      "step": 69040
    },
    {
      "epoch": 28.26,
      "grad_norm": 2.3337056636810303,
      "learning_rate": 4.581432384582268e-06,
      "loss": 0.5789,
      "step": 69050
    },
    {
      "epoch": 28.27,
      "grad_norm": 3.045982599258423,
      "learning_rate": 4.581385198706217e-06,
      "loss": 0.5703,
      "step": 69060
    },
    {
      "epoch": 28.27,
      "grad_norm": 3.218207359313965,
      "learning_rate": 4.581338002579907e-06,
      "loss": 0.5813,
      "step": 69070
    },
    {
      "epoch": 28.28,
      "grad_norm": 2.6889023780822754,
      "learning_rate": 4.581290796203551e-06,
      "loss": 0.5805,
      "step": 69080
    },
    {
      "epoch": 28.28,
      "grad_norm": 2.449221611022949,
      "learning_rate": 4.581243579577368e-06,
      "loss": 0.5891,
      "step": 69090
    },
    {
      "epoch": 28.28,
      "grad_norm": 3.0097615718841553,
      "learning_rate": 4.5811963527015734e-06,
      "loss": 0.5829,
      "step": 69100
    },
    {
      "epoch": 28.29,
      "grad_norm": 2.605600118637085,
      "learning_rate": 4.5811491155763835e-06,
      "loss": 0.5932,
      "step": 69110
    },
    {
      "epoch": 28.29,
      "grad_norm": 2.826936721801758,
      "learning_rate": 4.581101868202015e-06,
      "loss": 0.5838,
      "step": 69120
    },
    {
      "epoch": 28.3,
      "grad_norm": 2.306689977645874,
      "learning_rate": 4.581054610578685e-06,
      "loss": 0.5811,
      "step": 69130
    },
    {
      "epoch": 28.3,
      "grad_norm": 2.9031946659088135,
      "learning_rate": 4.581007342706608e-06,
      "loss": 0.5893,
      "step": 69140
    },
    {
      "epoch": 28.31,
      "grad_norm": 2.695763349533081,
      "learning_rate": 4.580960064586002e-06,
      "loss": 0.5936,
      "step": 69150
    },
    {
      "epoch": 28.31,
      "grad_norm": 2.3786370754241943,
      "learning_rate": 4.580912776217082e-06,
      "loss": 0.5616,
      "step": 69160
    },
    {
      "epoch": 28.31,
      "grad_norm": 2.3940281867980957,
      "learning_rate": 4.580865477600067e-06,
      "loss": 0.5664,
      "step": 69170
    },
    {
      "epoch": 28.32,
      "grad_norm": 2.1905765533447266,
      "learning_rate": 4.580818168735172e-06,
      "loss": 0.5615,
      "step": 69180
    },
    {
      "epoch": 28.32,
      "grad_norm": 2.1325368881225586,
      "learning_rate": 4.580770849622615e-06,
      "loss": 0.5937,
      "step": 69190
    },
    {
      "epoch": 28.33,
      "grad_norm": 2.513066053390503,
      "learning_rate": 4.5807235202626105e-06,
      "loss": 0.5546,
      "step": 69200
    },
    {
      "epoch": 28.33,
      "grad_norm": 2.56050968170166,
      "learning_rate": 4.580676180655377e-06,
      "loss": 0.5804,
      "step": 69210
    },
    {
      "epoch": 28.33,
      "grad_norm": 2.2757856845855713,
      "learning_rate": 4.580628830801132e-06,
      "loss": 0.5829,
      "step": 69220
    },
    {
      "epoch": 28.34,
      "grad_norm": 3.100836753845215,
      "learning_rate": 4.580581470700091e-06,
      "loss": 0.5719,
      "step": 69230
    },
    {
      "epoch": 28.34,
      "grad_norm": 3.1076858043670654,
      "learning_rate": 4.580534100352472e-06,
      "loss": 0.5737,
      "step": 69240
    },
    {
      "epoch": 28.35,
      "grad_norm": 2.0417659282684326,
      "learning_rate": 4.58048671975849e-06,
      "loss": 0.568,
      "step": 69250
    },
    {
      "epoch": 28.35,
      "grad_norm": 2.584803819656372,
      "learning_rate": 4.580439328918365e-06,
      "loss": 0.5646,
      "step": 69260
    },
    {
      "epoch": 28.35,
      "grad_norm": 1.9929956197738647,
      "learning_rate": 4.580391927832313e-06,
      "loss": 0.57,
      "step": 69270
    },
    {
      "epoch": 28.36,
      "grad_norm": 2.920858383178711,
      "learning_rate": 4.580344516500548e-06,
      "loss": 0.5736,
      "step": 69280
    },
    {
      "epoch": 28.36,
      "grad_norm": 2.829538345336914,
      "learning_rate": 4.580297094923293e-06,
      "loss": 0.5823,
      "step": 69290
    },
    {
      "epoch": 28.37,
      "grad_norm": 2.433185577392578,
      "learning_rate": 4.5802496631007605e-06,
      "loss": 0.5855,
      "step": 69300
    },
    {
      "epoch": 28.37,
      "grad_norm": 2.042401075363159,
      "learning_rate": 4.58020222103317e-06,
      "loss": 0.6098,
      "step": 69310
    },
    {
      "epoch": 28.37,
      "grad_norm": 2.406740665435791,
      "learning_rate": 4.580154768720737e-06,
      "loss": 0.592,
      "step": 69320
    },
    {
      "epoch": 28.38,
      "grad_norm": 1.9493086338043213,
      "learning_rate": 4.580107306163681e-06,
      "loss": 0.5939,
      "step": 69330
    },
    {
      "epoch": 28.38,
      "grad_norm": 2.479764223098755,
      "learning_rate": 4.580059833362219e-06,
      "loss": 0.59,
      "step": 69340
    },
    {
      "epoch": 28.39,
      "grad_norm": 2.144939422607422,
      "learning_rate": 4.580012350316567e-06,
      "loss": 0.5925,
      "step": 69350
    },
    {
      "epoch": 28.39,
      "grad_norm": 2.5803985595703125,
      "learning_rate": 4.579964857026944e-06,
      "loss": 0.5618,
      "step": 69360
    },
    {
      "epoch": 28.4,
      "grad_norm": 2.476367950439453,
      "learning_rate": 4.5799173534935675e-06,
      "loss": 0.5814,
      "step": 69370
    },
    {
      "epoch": 28.4,
      "grad_norm": 2.593306303024292,
      "learning_rate": 4.579869839716655e-06,
      "loss": 0.5988,
      "step": 69380
    },
    {
      "epoch": 28.4,
      "grad_norm": 2.319071054458618,
      "learning_rate": 4.579822315696422e-06,
      "loss": 0.5899,
      "step": 69390
    },
    {
      "epoch": 28.41,
      "grad_norm": 2.216701030731201,
      "learning_rate": 4.579774781433089e-06,
      "loss": 0.5793,
      "step": 69400
    },
    {
      "epoch": 28.41,
      "grad_norm": 2.4467592239379883,
      "learning_rate": 4.579727236926873e-06,
      "loss": 0.5706,
      "step": 69410
    },
    {
      "epoch": 28.42,
      "grad_norm": 2.646374225616455,
      "learning_rate": 4.5796796821779914e-06,
      "loss": 0.5813,
      "step": 69420
    },
    {
      "epoch": 28.42,
      "grad_norm": 2.148625135421753,
      "learning_rate": 4.579632117186663e-06,
      "loss": 0.5769,
      "step": 69430
    },
    {
      "epoch": 28.42,
      "grad_norm": 1.9938709735870361,
      "learning_rate": 4.579584541953103e-06,
      "loss": 0.5848,
      "step": 69440
    },
    {
      "epoch": 28.43,
      "grad_norm": 2.484159469604492,
      "learning_rate": 4.579536956477533e-06,
      "loss": 0.5779,
      "step": 69450
    },
    {
      "epoch": 28.43,
      "grad_norm": 2.423060178756714,
      "learning_rate": 4.579489360760169e-06,
      "loss": 0.5951,
      "step": 69460
    },
    {
      "epoch": 28.44,
      "grad_norm": 3.371776580810547,
      "learning_rate": 4.579441754801228e-06,
      "loss": 0.5908,
      "step": 69470
    },
    {
      "epoch": 28.44,
      "grad_norm": 2.013073682785034,
      "learning_rate": 4.57939413860093e-06,
      "loss": 0.5637,
      "step": 69480
    },
    {
      "epoch": 28.44,
      "grad_norm": 2.781296730041504,
      "learning_rate": 4.5793465121594934e-06,
      "loss": 0.6027,
      "step": 69490
    },
    {
      "epoch": 28.45,
      "grad_norm": 2.792523145675659,
      "learning_rate": 4.579298875477134e-06,
      "loss": 0.5788,
      "step": 69500
    },
    {
      "epoch": 28.45,
      "grad_norm": 2.1564223766326904,
      "learning_rate": 4.579251228554072e-06,
      "loss": 0.58,
      "step": 69510
    },
    {
      "epoch": 28.46,
      "grad_norm": 3.0102970600128174,
      "learning_rate": 4.579203571390526e-06,
      "loss": 0.5722,
      "step": 69520
    },
    {
      "epoch": 28.46,
      "grad_norm": 2.2874672412872314,
      "learning_rate": 4.579155903986712e-06,
      "loss": 0.5965,
      "step": 69530
    },
    {
      "epoch": 28.47,
      "grad_norm": 2.843888759613037,
      "learning_rate": 4.57910822634285e-06,
      "loss": 0.5905,
      "step": 69540
    },
    {
      "epoch": 28.47,
      "grad_norm": 2.887383222579956,
      "learning_rate": 4.5790605384591585e-06,
      "loss": 0.5983,
      "step": 69550
    },
    {
      "epoch": 28.47,
      "grad_norm": 2.024369478225708,
      "learning_rate": 4.579012840335855e-06,
      "loss": 0.597,
      "step": 69560
    },
    {
      "epoch": 28.48,
      "grad_norm": 2.4213433265686035,
      "learning_rate": 4.578965131973159e-06,
      "loss": 0.6014,
      "step": 69570
    },
    {
      "epoch": 28.48,
      "grad_norm": 2.718463897705078,
      "learning_rate": 4.578917413371289e-06,
      "loss": 0.6031,
      "step": 69580
    },
    {
      "epoch": 28.49,
      "grad_norm": 2.8098905086517334,
      "learning_rate": 4.578869684530463e-06,
      "loss": 0.5871,
      "step": 69590
    },
    {
      "epoch": 28.49,
      "grad_norm": 2.3924410343170166,
      "learning_rate": 4.5788219454509e-06,
      "loss": 0.6009,
      "step": 69600
    },
    {
      "epoch": 28.49,
      "grad_norm": 2.031252145767212,
      "learning_rate": 4.578774196132818e-06,
      "loss": 0.5855,
      "step": 69610
    },
    {
      "epoch": 28.5,
      "grad_norm": 2.431867837905884,
      "learning_rate": 4.578726436576437e-06,
      "loss": 0.5927,
      "step": 69620
    },
    {
      "epoch": 28.5,
      "grad_norm": 2.38498854637146,
      "learning_rate": 4.578678666781975e-06,
      "loss": 0.5954,
      "step": 69630
    },
    {
      "epoch": 28.51,
      "grad_norm": 1.8978025913238525,
      "learning_rate": 4.5786308867496505e-06,
      "loss": 0.5761,
      "step": 69640
    },
    {
      "epoch": 28.51,
      "grad_norm": 2.8365113735198975,
      "learning_rate": 4.5785830964796826e-06,
      "loss": 0.5846,
      "step": 69650
    },
    {
      "epoch": 28.51,
      "grad_norm": 1.6591887474060059,
      "learning_rate": 4.578535295972291e-06,
      "loss": 0.6183,
      "step": 69660
    },
    {
      "epoch": 28.52,
      "grad_norm": 3.0088412761688232,
      "learning_rate": 4.578487485227693e-06,
      "loss": 0.5854,
      "step": 69670
    },
    {
      "epoch": 28.52,
      "grad_norm": 2.9868862628936768,
      "learning_rate": 4.5784396642461105e-06,
      "loss": 0.6054,
      "step": 69680
    },
    {
      "epoch": 28.53,
      "grad_norm": 2.8302884101867676,
      "learning_rate": 4.5783918330277585e-06,
      "loss": 0.5987,
      "step": 69690
    },
    {
      "epoch": 28.53,
      "grad_norm": 3.4276742935180664,
      "learning_rate": 4.57834399157286e-06,
      "loss": 0.5842,
      "step": 69700
    },
    {
      "epoch": 28.53,
      "grad_norm": 4.225866794586182,
      "learning_rate": 4.578296139881632e-06,
      "loss": 0.6062,
      "step": 69710
    },
    {
      "epoch": 28.54,
      "grad_norm": 2.78940486907959,
      "learning_rate": 4.578248277954295e-06,
      "loss": 0.5638,
      "step": 69720
    },
    {
      "epoch": 28.54,
      "grad_norm": 2.7788100242614746,
      "learning_rate": 4.578200405791066e-06,
      "loss": 0.5815,
      "step": 69730
    },
    {
      "epoch": 28.55,
      "grad_norm": 2.3397843837738037,
      "learning_rate": 4.578152523392166e-06,
      "loss": 0.5878,
      "step": 69740
    },
    {
      "epoch": 28.55,
      "grad_norm": 1.8254475593566895,
      "learning_rate": 4.578104630757815e-06,
      "loss": 0.5799,
      "step": 69750
    },
    {
      "epoch": 28.56,
      "grad_norm": 2.1628355979919434,
      "learning_rate": 4.578056727888231e-06,
      "loss": 0.5701,
      "step": 69760
    },
    {
      "epoch": 28.56,
      "grad_norm": 2.1238458156585693,
      "learning_rate": 4.578008814783633e-06,
      "loss": 0.5871,
      "step": 69770
    },
    {
      "epoch": 28.56,
      "grad_norm": 3.6298553943634033,
      "learning_rate": 4.577960891444243e-06,
      "loss": 0.5834,
      "step": 69780
    },
    {
      "epoch": 28.57,
      "grad_norm": 2.9205849170684814,
      "learning_rate": 4.577912957870278e-06,
      "loss": 0.539,
      "step": 69790
    },
    {
      "epoch": 28.57,
      "grad_norm": 2.4615302085876465,
      "learning_rate": 4.577865014061959e-06,
      "loss": 0.5747,
      "step": 69800
    },
    {
      "epoch": 28.58,
      "grad_norm": 2.8314731121063232,
      "learning_rate": 4.577817060019505e-06,
      "loss": 0.5724,
      "step": 69810
    },
    {
      "epoch": 28.58,
      "grad_norm": 2.132296323776245,
      "learning_rate": 4.577769095743136e-06,
      "loss": 0.6005,
      "step": 69820
    },
    {
      "epoch": 28.58,
      "grad_norm": 2.399879217147827,
      "learning_rate": 4.57772112123307e-06,
      "loss": 0.5909,
      "step": 69830
    },
    {
      "epoch": 28.59,
      "grad_norm": 2.677614212036133,
      "learning_rate": 4.57767313648953e-06,
      "loss": 0.5843,
      "step": 69840
    },
    {
      "epoch": 28.59,
      "grad_norm": 2.0954861640930176,
      "learning_rate": 4.577625141512733e-06,
      "loss": 0.5711,
      "step": 69850
    },
    {
      "epoch": 28.6,
      "grad_norm": 2.5707221031188965,
      "learning_rate": 4.577577136302901e-06,
      "loss": 0.5523,
      "step": 69860
    },
    {
      "epoch": 28.6,
      "grad_norm": 2.558105707168579,
      "learning_rate": 4.577529120860251e-06,
      "loss": 0.5978,
      "step": 69870
    },
    {
      "epoch": 28.6,
      "grad_norm": 2.7181782722473145,
      "learning_rate": 4.577481095185007e-06,
      "loss": 0.5784,
      "step": 69880
    },
    {
      "epoch": 28.61,
      "grad_norm": 2.6383979320526123,
      "learning_rate": 4.577433059277385e-06,
      "loss": 0.5711,
      "step": 69890
    },
    {
      "epoch": 28.61,
      "grad_norm": 2.285989999771118,
      "learning_rate": 4.577385013137608e-06,
      "loss": 0.5971,
      "step": 69900
    },
    {
      "epoch": 28.62,
      "grad_norm": 2.7754268646240234,
      "learning_rate": 4.577336956765894e-06,
      "loss": 0.5957,
      "step": 69910
    },
    {
      "epoch": 28.62,
      "grad_norm": 1.877310037612915,
      "learning_rate": 4.577288890162465e-06,
      "loss": 0.5637,
      "step": 69920
    },
    {
      "epoch": 28.62,
      "grad_norm": 2.4849398136138916,
      "learning_rate": 4.57724081332754e-06,
      "loss": 0.599,
      "step": 69930
    },
    {
      "epoch": 28.63,
      "grad_norm": 3.1078429222106934,
      "learning_rate": 4.577192726261339e-06,
      "loss": 0.5631,
      "step": 69940
    },
    {
      "epoch": 28.63,
      "grad_norm": 2.4688756465911865,
      "learning_rate": 4.577144628964083e-06,
      "loss": 0.5798,
      "step": 69950
    },
    {
      "epoch": 28.64,
      "grad_norm": 2.2050793170928955,
      "learning_rate": 4.577096521435993e-06,
      "loss": 0.5684,
      "step": 69960
    },
    {
      "epoch": 28.64,
      "grad_norm": 2.3577077388763428,
      "learning_rate": 4.5770484036772874e-06,
      "loss": 0.5749,
      "step": 69970
    },
    {
      "epoch": 28.65,
      "grad_norm": 2.5937576293945312,
      "learning_rate": 4.577000275688188e-06,
      "loss": 0.5883,
      "step": 69980
    },
    {
      "epoch": 28.65,
      "grad_norm": 1.9555785655975342,
      "learning_rate": 4.576952137468915e-06,
      "loss": 0.564,
      "step": 69990
    },
    {
      "epoch": 28.65,
      "grad_norm": 2.362508773803711,
      "learning_rate": 4.5769039890196886e-06,
      "loss": 0.5985,
      "step": 70000
    },
    {
      "epoch": 28.66,
      "grad_norm": 2.2349660396575928,
      "learning_rate": 4.576855830340729e-06,
      "loss": 0.5909,
      "step": 70010
    },
    {
      "epoch": 28.66,
      "grad_norm": 1.614231824874878,
      "learning_rate": 4.576807661432258e-06,
      "loss": 0.5917,
      "step": 70020
    },
    {
      "epoch": 28.67,
      "grad_norm": 2.499480724334717,
      "learning_rate": 4.5767594822944966e-06,
      "loss": 0.5878,
      "step": 70030
    },
    {
      "epoch": 28.67,
      "grad_norm": 2.2878572940826416,
      "learning_rate": 4.576711292927663e-06,
      "loss": 0.584,
      "step": 70040
    },
    {
      "epoch": 28.67,
      "grad_norm": 2.246985912322998,
      "learning_rate": 4.5766630933319814e-06,
      "loss": 0.6023,
      "step": 70050
    },
    {
      "epoch": 28.68,
      "grad_norm": 1.8873820304870605,
      "learning_rate": 4.5766148835076694e-06,
      "loss": 0.5794,
      "step": 70060
    },
    {
      "epoch": 28.68,
      "grad_norm": 2.042201280593872,
      "learning_rate": 4.57656666345495e-06,
      "loss": 0.5804,
      "step": 70070
    },
    {
      "epoch": 28.69,
      "grad_norm": 2.209235668182373,
      "learning_rate": 4.576518433174042e-06,
      "loss": 0.5881,
      "step": 70080
    },
    {
      "epoch": 28.69,
      "grad_norm": 2.637965679168701,
      "learning_rate": 4.576470192665169e-06,
      "loss": 0.587,
      "step": 70090
    },
    {
      "epoch": 28.69,
      "grad_norm": 2.100491523742676,
      "learning_rate": 4.5764219419285495e-06,
      "loss": 0.5853,
      "step": 70100
    },
    {
      "epoch": 28.7,
      "grad_norm": 2.266683340072632,
      "learning_rate": 4.576373680964406e-06,
      "loss": 0.5615,
      "step": 70110
    },
    {
      "epoch": 28.7,
      "grad_norm": 2.2070038318634033,
      "learning_rate": 4.576325409772959e-06,
      "loss": 0.5885,
      "step": 70120
    },
    {
      "epoch": 28.71,
      "grad_norm": 2.21581768989563,
      "learning_rate": 4.5762771283544305e-06,
      "loss": 0.5758,
      "step": 70130
    },
    {
      "epoch": 28.71,
      "grad_norm": 2.869645357131958,
      "learning_rate": 4.57622883670904e-06,
      "loss": 0.5807,
      "step": 70140
    },
    {
      "epoch": 28.71,
      "grad_norm": 2.3460805416107178,
      "learning_rate": 4.576180534837011e-06,
      "loss": 0.5786,
      "step": 70150
    },
    {
      "epoch": 28.72,
      "grad_norm": 2.2759900093078613,
      "learning_rate": 4.576132222738562e-06,
      "loss": 0.5769,
      "step": 70160
    },
    {
      "epoch": 28.72,
      "grad_norm": 2.497122049331665,
      "learning_rate": 4.576083900413917e-06,
      "loss": 0.5714,
      "step": 70170
    },
    {
      "epoch": 28.73,
      "grad_norm": 2.399740219116211,
      "learning_rate": 4.576035567863295e-06,
      "loss": 0.6003,
      "step": 70180
    },
    {
      "epoch": 28.73,
      "grad_norm": 2.0008857250213623,
      "learning_rate": 4.57598722508692e-06,
      "loss": 0.5922,
      "step": 70190
    },
    {
      "epoch": 28.74,
      "grad_norm": 3.1732676029205322,
      "learning_rate": 4.575938872085011e-06,
      "loss": 0.5776,
      "step": 70200
    },
    {
      "epoch": 28.74,
      "grad_norm": 2.0884499549865723,
      "learning_rate": 4.57589050885779e-06,
      "loss": 0.5616,
      "step": 70210
    },
    {
      "epoch": 28.74,
      "grad_norm": 1.6607708930969238,
      "learning_rate": 4.575842135405479e-06,
      "loss": 0.5657,
      "step": 70220
    },
    {
      "epoch": 28.75,
      "grad_norm": 2.836808919906616,
      "learning_rate": 4.5757937517283e-06,
      "loss": 0.577,
      "step": 70230
    },
    {
      "epoch": 28.75,
      "grad_norm": 2.8290607929229736,
      "learning_rate": 4.575745357826475e-06,
      "loss": 0.5838,
      "step": 70240
    },
    {
      "epoch": 28.76,
      "grad_norm": 3.0203123092651367,
      "learning_rate": 4.575696953700225e-06,
      "loss": 0.5847,
      "step": 70250
    },
    {
      "epoch": 28.76,
      "grad_norm": 1.9807488918304443,
      "learning_rate": 4.57564853934977e-06,
      "loss": 0.566,
      "step": 70260
    },
    {
      "epoch": 28.76,
      "grad_norm": 2.3054471015930176,
      "learning_rate": 4.575600114775335e-06,
      "loss": 0.5812,
      "step": 70270
    },
    {
      "epoch": 28.77,
      "grad_norm": 1.9795836210250854,
      "learning_rate": 4.575551679977139e-06,
      "loss": 0.5737,
      "step": 70280
    },
    {
      "epoch": 28.77,
      "grad_norm": 2.3258795738220215,
      "learning_rate": 4.575503234955407e-06,
      "loss": 0.5728,
      "step": 70290
    },
    {
      "epoch": 28.78,
      "grad_norm": 2.211745262145996,
      "learning_rate": 4.5754547797103574e-06,
      "loss": 0.5837,
      "step": 70300
    },
    {
      "epoch": 28.78,
      "grad_norm": 2.500685691833496,
      "learning_rate": 4.575406314242213e-06,
      "loss": 0.5675,
      "step": 70310
    },
    {
      "epoch": 28.78,
      "grad_norm": 2.769144296646118,
      "learning_rate": 4.575357838551199e-06,
      "loss": 0.5639,
      "step": 70320
    },
    {
      "epoch": 28.79,
      "grad_norm": 2.404688835144043,
      "learning_rate": 4.575309352637534e-06,
      "loss": 0.5778,
      "step": 70330
    },
    {
      "epoch": 28.79,
      "grad_norm": 2.692525625228882,
      "learning_rate": 4.575260856501441e-06,
      "loss": 0.5636,
      "step": 70340
    },
    {
      "epoch": 28.8,
      "grad_norm": 2.158555269241333,
      "learning_rate": 4.575212350143143e-06,
      "loss": 0.5472,
      "step": 70350
    },
    {
      "epoch": 28.8,
      "grad_norm": 2.3770785331726074,
      "learning_rate": 4.575163833562861e-06,
      "loss": 0.5963,
      "step": 70360
    },
    {
      "epoch": 28.8,
      "grad_norm": 2.2872843742370605,
      "learning_rate": 4.575115306760818e-06,
      "loss": 0.5829,
      "step": 70370
    },
    {
      "epoch": 28.81,
      "grad_norm": 2.6304335594177246,
      "learning_rate": 4.575066769737237e-06,
      "loss": 0.557,
      "step": 70380
    },
    {
      "epoch": 28.81,
      "grad_norm": 2.99874210357666,
      "learning_rate": 4.575018222492338e-06,
      "loss": 0.5776,
      "step": 70390
    },
    {
      "epoch": 28.82,
      "grad_norm": 2.1872496604919434,
      "learning_rate": 4.574969665026346e-06,
      "loss": 0.5741,
      "step": 70400
    },
    {
      "epoch": 28.82,
      "grad_norm": 2.4430344104766846,
      "learning_rate": 4.5749210973394804e-06,
      "loss": 0.5762,
      "step": 70410
    },
    {
      "epoch": 28.83,
      "grad_norm": 2.3740131855010986,
      "learning_rate": 4.574872519431967e-06,
      "loss": 0.5796,
      "step": 70420
    },
    {
      "epoch": 28.83,
      "grad_norm": 3.3430962562561035,
      "learning_rate": 4.574823931304027e-06,
      "loss": 0.5876,
      "step": 70430
    },
    {
      "epoch": 28.83,
      "grad_norm": 2.6070165634155273,
      "learning_rate": 4.574775332955882e-06,
      "loss": 0.5894,
      "step": 70440
    },
    {
      "epoch": 28.84,
      "grad_norm": 3.2122349739074707,
      "learning_rate": 4.574726724387756e-06,
      "loss": 0.6057,
      "step": 70450
    },
    {
      "epoch": 28.84,
      "grad_norm": 2.325284957885742,
      "learning_rate": 4.574678105599872e-06,
      "loss": 0.591,
      "step": 70460
    },
    {
      "epoch": 28.85,
      "grad_norm": 2.992959976196289,
      "learning_rate": 4.57462947659245e-06,
      "loss": 0.5659,
      "step": 70470
    },
    {
      "epoch": 28.85,
      "grad_norm": 2.529855489730835,
      "learning_rate": 4.574580837365716e-06,
      "loss": 0.5661,
      "step": 70480
    },
    {
      "epoch": 28.85,
      "grad_norm": 2.1059532165527344,
      "learning_rate": 4.574532187919891e-06,
      "loss": 0.6087,
      "step": 70490
    },
    {
      "epoch": 28.86,
      "grad_norm": 2.5507147312164307,
      "learning_rate": 4.574483528255199e-06,
      "loss": 0.5645,
      "step": 70500
    },
    {
      "epoch": 28.86,
      "grad_norm": 2.7286055088043213,
      "learning_rate": 4.574434858371861e-06,
      "loss": 0.596,
      "step": 70510
    },
    {
      "epoch": 28.87,
      "grad_norm": 2.0733022689819336,
      "learning_rate": 4.574386178270101e-06,
      "loss": 0.5825,
      "step": 70520
    },
    {
      "epoch": 28.87,
      "grad_norm": 2.568404197692871,
      "learning_rate": 4.574337487950143e-06,
      "loss": 0.5561,
      "step": 70530
    },
    {
      "epoch": 28.87,
      "grad_norm": 2.2620091438293457,
      "learning_rate": 4.574288787412209e-06,
      "loss": 0.6015,
      "step": 70540
    },
    {
      "epoch": 28.88,
      "grad_norm": 1.8343054056167603,
      "learning_rate": 4.574240076656522e-06,
      "loss": 0.5703,
      "step": 70550
    },
    {
      "epoch": 28.88,
      "grad_norm": 2.384406805038452,
      "learning_rate": 4.574191355683305e-06,
      "loss": 0.5548,
      "step": 70560
    },
    {
      "epoch": 28.89,
      "grad_norm": 2.3690969944000244,
      "learning_rate": 4.574142624492783e-06,
      "loss": 0.5511,
      "step": 70570
    },
    {
      "epoch": 28.89,
      "grad_norm": 2.6162149906158447,
      "learning_rate": 4.574093883085177e-06,
      "loss": 0.5976,
      "step": 70580
    },
    {
      "epoch": 28.89,
      "grad_norm": 2.6729538440704346,
      "learning_rate": 4.574045131460711e-06,
      "loss": 0.5777,
      "step": 70590
    },
    {
      "epoch": 28.9,
      "grad_norm": 2.541349172592163,
      "learning_rate": 4.5739963696196084e-06,
      "loss": 0.5603,
      "step": 70600
    },
    {
      "epoch": 28.9,
      "grad_norm": 2.051339864730835,
      "learning_rate": 4.573947597562092e-06,
      "loss": 0.5918,
      "step": 70610
    },
    {
      "epoch": 28.91,
      "grad_norm": 2.7883551120758057,
      "learning_rate": 4.573898815288387e-06,
      "loss": 0.5636,
      "step": 70620
    },
    {
      "epoch": 28.91,
      "grad_norm": 2.5914227962493896,
      "learning_rate": 4.5738500227987145e-06,
      "loss": 0.5812,
      "step": 70630
    },
    {
      "epoch": 28.92,
      "grad_norm": 2.4950883388519287,
      "learning_rate": 4.5738012200933e-06,
      "loss": 0.6088,
      "step": 70640
    },
    {
      "epoch": 28.92,
      "grad_norm": 2.348675489425659,
      "learning_rate": 4.573752407172365e-06,
      "loss": 0.5717,
      "step": 70650
    },
    {
      "epoch": 28.92,
      "grad_norm": 2.080198287963867,
      "learning_rate": 4.5737035840361355e-06,
      "loss": 0.5678,
      "step": 70660
    },
    {
      "epoch": 28.93,
      "grad_norm": 2.600152015686035,
      "learning_rate": 4.573654750684834e-06,
      "loss": 0.6104,
      "step": 70670
    },
    {
      "epoch": 28.93,
      "grad_norm": 2.2147715091705322,
      "learning_rate": 4.573605907118685e-06,
      "loss": 0.5822,
      "step": 70680
    },
    {
      "epoch": 28.94,
      "grad_norm": 2.433356285095215,
      "learning_rate": 4.573557053337909e-06,
      "loss": 0.5817,
      "step": 70690
    },
    {
      "epoch": 28.94,
      "grad_norm": 2.5955417156219482,
      "learning_rate": 4.573508189342734e-06,
      "loss": 0.5739,
      "step": 70700
    },
    {
      "epoch": 28.94,
      "grad_norm": 2.448540687561035,
      "learning_rate": 4.573459315133382e-06,
      "loss": 0.5806,
      "step": 70710
    },
    {
      "epoch": 28.95,
      "grad_norm": 2.606130599975586,
      "learning_rate": 4.573410430710076e-06,
      "loss": 0.5802,
      "step": 70720
    },
    {
      "epoch": 28.95,
      "grad_norm": 2.1920950412750244,
      "learning_rate": 4.573361536073042e-06,
      "loss": 0.5545,
      "step": 70730
    },
    {
      "epoch": 28.96,
      "grad_norm": 2.817218065261841,
      "learning_rate": 4.573312631222502e-06,
      "loss": 0.5835,
      "step": 70740
    },
    {
      "epoch": 28.96,
      "grad_norm": 2.336266040802002,
      "learning_rate": 4.5732637161586815e-06,
      "loss": 0.6029,
      "step": 70750
    },
    {
      "epoch": 28.96,
      "grad_norm": 2.1345441341400146,
      "learning_rate": 4.573214790881804e-06,
      "loss": 0.5875,
      "step": 70760
    },
    {
      "epoch": 28.97,
      "grad_norm": 2.0455543994903564,
      "learning_rate": 4.573165855392092e-06,
      "loss": 0.5606,
      "step": 70770
    },
    {
      "epoch": 28.97,
      "grad_norm": 3.7079179286956787,
      "learning_rate": 4.5731169096897725e-06,
      "loss": 0.5788,
      "step": 70780
    },
    {
      "epoch": 28.98,
      "grad_norm": 2.9399454593658447,
      "learning_rate": 4.573067953775069e-06,
      "loss": 0.5798,
      "step": 70790
    },
    {
      "epoch": 28.98,
      "grad_norm": 1.555329442024231,
      "learning_rate": 4.573018987648204e-06,
      "loss": 0.5752,
      "step": 70800
    },
    {
      "epoch": 28.98,
      "grad_norm": 1.603709101676941,
      "learning_rate": 4.572970011309403e-06,
      "loss": 0.5781,
      "step": 70810
    },
    {
      "epoch": 28.99,
      "grad_norm": 2.0542097091674805,
      "learning_rate": 4.5729210247588915e-06,
      "loss": 0.572,
      "step": 70820
    },
    {
      "epoch": 28.99,
      "grad_norm": 2.1158945560455322,
      "learning_rate": 4.572872027996893e-06,
      "loss": 0.5827,
      "step": 70830
    },
    {
      "epoch": 29.0,
      "grad_norm": 2.014787435531616,
      "learning_rate": 4.5728230210236306e-06,
      "loss": 0.5725,
      "step": 70840
    },
    {
      "epoch": 29.0,
      "eval_loss": 0.5766116976737976,
      "eval_runtime": 51.9723,
      "eval_samples_per_second": 66.362,
      "eval_steps_per_second": 8.312,
      "step": 70847
    },
    {
      "epoch": 29.0,
      "grad_norm": 2.271862506866455,
      "learning_rate": 4.57277400383933e-06,
      "loss": 0.5843,
      "step": 70850
    },
    {
      "epoch": 29.01,
      "grad_norm": 2.246933698654175,
      "learning_rate": 4.572724976444216e-06,
      "loss": 0.5847,
      "step": 70860
    },
    {
      "epoch": 29.01,
      "grad_norm": 2.941682815551758,
      "learning_rate": 4.572675938838513e-06,
      "loss": 0.5524,
      "step": 70870
    },
    {
      "epoch": 29.01,
      "grad_norm": 2.548579692840576,
      "learning_rate": 4.572626891022444e-06,
      "loss": 0.5985,
      "step": 70880
    },
    {
      "epoch": 29.02,
      "grad_norm": 3.2462711334228516,
      "learning_rate": 4.572577832996237e-06,
      "loss": 0.6004,
      "step": 70890
    },
    {
      "epoch": 29.02,
      "grad_norm": 2.745675563812256,
      "learning_rate": 4.572528764760114e-06,
      "loss": 0.5861,
      "step": 70900
    },
    {
      "epoch": 29.03,
      "grad_norm": 1.9024534225463867,
      "learning_rate": 4.572479686314301e-06,
      "loss": 0.5845,
      "step": 70910
    },
    {
      "epoch": 29.03,
      "grad_norm": 2.650691509246826,
      "learning_rate": 4.5724305976590225e-06,
      "loss": 0.5751,
      "step": 70920
    },
    {
      "epoch": 29.03,
      "grad_norm": 2.408609390258789,
      "learning_rate": 4.572381498794503e-06,
      "loss": 0.5862,
      "step": 70930
    },
    {
      "epoch": 29.04,
      "grad_norm": 2.543994903564453,
      "learning_rate": 4.572332389720968e-06,
      "loss": 0.5887,
      "step": 70940
    },
    {
      "epoch": 29.04,
      "grad_norm": 3.108060359954834,
      "learning_rate": 4.572283270438643e-06,
      "loss": 0.5912,
      "step": 70950
    },
    {
      "epoch": 29.05,
      "grad_norm": 2.847994804382324,
      "learning_rate": 4.572234140947751e-06,
      "loss": 0.5643,
      "step": 70960
    },
    {
      "epoch": 29.05,
      "grad_norm": 1.8662376403808594,
      "learning_rate": 4.572185001248518e-06,
      "loss": 0.5773,
      "step": 70970
    },
    {
      "epoch": 29.05,
      "grad_norm": 2.9129905700683594,
      "learning_rate": 4.57213585134117e-06,
      "loss": 0.5825,
      "step": 70980
    },
    {
      "epoch": 29.06,
      "grad_norm": 2.921266555786133,
      "learning_rate": 4.5720866912259316e-06,
      "loss": 0.5799,
      "step": 70990
    },
    {
      "epoch": 29.06,
      "grad_norm": 3.335434913635254,
      "learning_rate": 4.572037520903028e-06,
      "loss": 0.5817,
      "step": 71000
    },
    {
      "epoch": 29.07,
      "grad_norm": 1.752600073814392,
      "learning_rate": 4.5719883403726836e-06,
      "loss": 0.5661,
      "step": 71010
    },
    {
      "epoch": 29.07,
      "grad_norm": 2.705221176147461,
      "learning_rate": 4.5719391496351255e-06,
      "loss": 0.6043,
      "step": 71020
    },
    {
      "epoch": 29.07,
      "grad_norm": 2.505911350250244,
      "learning_rate": 4.571889948690578e-06,
      "loss": 0.576,
      "step": 71030
    },
    {
      "epoch": 29.08,
      "grad_norm": 2.6451845169067383,
      "learning_rate": 4.571840737539266e-06,
      "loss": 0.5827,
      "step": 71040
    },
    {
      "epoch": 29.08,
      "grad_norm": 2.9062798023223877,
      "learning_rate": 4.571791516181415e-06,
      "loss": 0.5928,
      "step": 71050
    },
    {
      "epoch": 29.09,
      "grad_norm": 2.383395195007324,
      "learning_rate": 4.571742284617251e-06,
      "loss": 0.5621,
      "step": 71060
    },
    {
      "epoch": 29.09,
      "grad_norm": 1.8702890872955322,
      "learning_rate": 4.571693042847e-06,
      "loss": 0.5839,
      "step": 71070
    },
    {
      "epoch": 29.1,
      "grad_norm": 2.258488416671753,
      "learning_rate": 4.5716437908708865e-06,
      "loss": 0.56,
      "step": 71080
    },
    {
      "epoch": 29.1,
      "grad_norm": 2.2204504013061523,
      "learning_rate": 4.571594528689137e-06,
      "loss": 0.5931,
      "step": 71090
    },
    {
      "epoch": 29.1,
      "grad_norm": 2.0292882919311523,
      "learning_rate": 4.571545256301977e-06,
      "loss": 0.5608,
      "step": 71100
    },
    {
      "epoch": 29.11,
      "grad_norm": 2.9361066818237305,
      "learning_rate": 4.5714959737096305e-06,
      "loss": 0.5686,
      "step": 71110
    },
    {
      "epoch": 29.11,
      "grad_norm": 2.6338844299316406,
      "learning_rate": 4.5714466809123255e-06,
      "loss": 0.5543,
      "step": 71120
    },
    {
      "epoch": 29.12,
      "grad_norm": 2.0703284740448,
      "learning_rate": 4.571397377910287e-06,
      "loss": 0.5878,
      "step": 71130
    },
    {
      "epoch": 29.12,
      "grad_norm": 2.6841647624969482,
      "learning_rate": 4.571348064703741e-06,
      "loss": 0.5629,
      "step": 71140
    },
    {
      "epoch": 29.12,
      "grad_norm": 1.9581048488616943,
      "learning_rate": 4.5712987412929135e-06,
      "loss": 0.5713,
      "step": 71150
    },
    {
      "epoch": 29.13,
      "grad_norm": 2.3812761306762695,
      "learning_rate": 4.57124940767803e-06,
      "loss": 0.5704,
      "step": 71160
    },
    {
      "epoch": 29.13,
      "grad_norm": 2.541740894317627,
      "learning_rate": 4.571200063859316e-06,
      "loss": 0.5884,
      "step": 71170
    },
    {
      "epoch": 29.14,
      "grad_norm": 2.997887372970581,
      "learning_rate": 4.571150709836999e-06,
      "loss": 0.5696,
      "step": 71180
    },
    {
      "epoch": 29.14,
      "grad_norm": 2.2105913162231445,
      "learning_rate": 4.571101345611305e-06,
      "loss": 0.5843,
      "step": 71190
    },
    {
      "epoch": 29.14,
      "grad_norm": 2.6944637298583984,
      "learning_rate": 4.571051971182458e-06,
      "loss": 0.5635,
      "step": 71200
    },
    {
      "epoch": 29.15,
      "grad_norm": 1.742593765258789,
      "learning_rate": 4.5710025865506856e-06,
      "loss": 0.6069,
      "step": 71210
    },
    {
      "epoch": 29.15,
      "grad_norm": 2.8597402572631836,
      "learning_rate": 4.570953191716215e-06,
      "loss": 0.5821,
      "step": 71220
    },
    {
      "epoch": 29.16,
      "grad_norm": 2.1156294345855713,
      "learning_rate": 4.570903786679271e-06,
      "loss": 0.5762,
      "step": 71230
    },
    {
      "epoch": 29.16,
      "grad_norm": 2.0896260738372803,
      "learning_rate": 4.5708543714400805e-06,
      "loss": 0.5681,
      "step": 71240
    },
    {
      "epoch": 29.16,
      "grad_norm": 2.389904022216797,
      "learning_rate": 4.5708049459988695e-06,
      "loss": 0.5753,
      "step": 71250
    },
    {
      "epoch": 29.17,
      "grad_norm": 2.7088992595672607,
      "learning_rate": 4.570755510355865e-06,
      "loss": 0.5696,
      "step": 71260
    },
    {
      "epoch": 29.17,
      "grad_norm": 2.0331826210021973,
      "learning_rate": 4.570706064511293e-06,
      "loss": 0.5875,
      "step": 71270
    },
    {
      "epoch": 29.18,
      "grad_norm": 2.5618393421173096,
      "learning_rate": 4.5706566084653805e-06,
      "loss": 0.5782,
      "step": 71280
    },
    {
      "epoch": 29.18,
      "grad_norm": 2.143242359161377,
      "learning_rate": 4.570607142218354e-06,
      "loss": 0.559,
      "step": 71290
    },
    {
      "epoch": 29.19,
      "grad_norm": 3.9928054809570312,
      "learning_rate": 4.570557665770439e-06,
      "loss": 0.5951,
      "step": 71300
    },
    {
      "epoch": 29.19,
      "grad_norm": 2.8436598777770996,
      "learning_rate": 4.570508179121863e-06,
      "loss": 0.6092,
      "step": 71310
    },
    {
      "epoch": 29.19,
      "grad_norm": 1.9162946939468384,
      "learning_rate": 4.570458682272853e-06,
      "loss": 0.5893,
      "step": 71320
    },
    {
      "epoch": 29.2,
      "grad_norm": 2.1251866817474365,
      "learning_rate": 4.570409175223636e-06,
      "loss": 0.5919,
      "step": 71330
    },
    {
      "epoch": 29.2,
      "grad_norm": 2.861319065093994,
      "learning_rate": 4.570359657974437e-06,
      "loss": 0.6054,
      "step": 71340
    },
    {
      "epoch": 29.21,
      "grad_norm": 2.1545605659484863,
      "learning_rate": 4.570310130525484e-06,
      "loss": 0.5798,
      "step": 71350
    },
    {
      "epoch": 29.21,
      "grad_norm": 3.0752501487731934,
      "learning_rate": 4.570260592877005e-06,
      "loss": 0.5561,
      "step": 71360
    },
    {
      "epoch": 29.21,
      "grad_norm": 2.0803334712982178,
      "learning_rate": 4.570211045029224e-06,
      "loss": 0.5933,
      "step": 71370
    },
    {
      "epoch": 29.22,
      "grad_norm": 2.3253414630889893,
      "learning_rate": 4.570161486982372e-06,
      "loss": 0.5719,
      "step": 71380
    },
    {
      "epoch": 29.22,
      "grad_norm": 2.174823760986328,
      "learning_rate": 4.570111918736672e-06,
      "loss": 0.5743,
      "step": 71390
    },
    {
      "epoch": 29.23,
      "grad_norm": 2.0717687606811523,
      "learning_rate": 4.570062340292353e-06,
      "loss": 0.5908,
      "step": 71400
    },
    {
      "epoch": 29.23,
      "grad_norm": 2.8798999786376953,
      "learning_rate": 4.570012751649643e-06,
      "loss": 0.5835,
      "step": 71410
    },
    {
      "epoch": 29.23,
      "grad_norm": 2.3238136768341064,
      "learning_rate": 4.569963152808768e-06,
      "loss": 0.5848,
      "step": 71420
    },
    {
      "epoch": 29.24,
      "grad_norm": 3.2775776386260986,
      "learning_rate": 4.569913543769954e-06,
      "loss": 0.5711,
      "step": 71430
    },
    {
      "epoch": 29.24,
      "grad_norm": 2.5908799171447754,
      "learning_rate": 4.569863924533431e-06,
      "loss": 0.6106,
      "step": 71440
    },
    {
      "epoch": 29.25,
      "grad_norm": 3.1535961627960205,
      "learning_rate": 4.5698142950994244e-06,
      "loss": 0.5803,
      "step": 71450
    },
    {
      "epoch": 29.25,
      "grad_norm": 2.48071026802063,
      "learning_rate": 4.569764655468162e-06,
      "loss": 0.5889,
      "step": 71460
    },
    {
      "epoch": 29.26,
      "grad_norm": 2.0865986347198486,
      "learning_rate": 4.569715005639871e-06,
      "loss": 0.6183,
      "step": 71470
    },
    {
      "epoch": 29.26,
      "grad_norm": 2.208576202392578,
      "learning_rate": 4.569665345614779e-06,
      "loss": 0.5834,
      "step": 71480
    },
    {
      "epoch": 29.26,
      "grad_norm": 2.3382890224456787,
      "learning_rate": 4.569615675393114e-06,
      "loss": 0.5851,
      "step": 71490
    },
    {
      "epoch": 29.27,
      "grad_norm": 2.1446449756622314,
      "learning_rate": 4.569565994975102e-06,
      "loss": 0.5807,
      "step": 71500
    },
    {
      "epoch": 29.27,
      "grad_norm": 2.1922717094421387,
      "learning_rate": 4.569516304360972e-06,
      "loss": 0.5724,
      "step": 71510
    },
    {
      "epoch": 29.28,
      "grad_norm": 2.2889797687530518,
      "learning_rate": 4.569466603550953e-06,
      "loss": 0.5566,
      "step": 71520
    },
    {
      "epoch": 29.28,
      "grad_norm": 1.9467328786849976,
      "learning_rate": 4.5694168925452695e-06,
      "loss": 0.5913,
      "step": 71530
    },
    {
      "epoch": 29.28,
      "grad_norm": 2.0420167446136475,
      "learning_rate": 4.569367171344151e-06,
      "loss": 0.563,
      "step": 71540
    },
    {
      "epoch": 29.29,
      "grad_norm": 2.0345633029937744,
      "learning_rate": 4.569317439947825e-06,
      "loss": 0.5832,
      "step": 71550
    },
    {
      "epoch": 29.29,
      "grad_norm": 1.9382789134979248,
      "learning_rate": 4.569267698356518e-06,
      "loss": 0.5779,
      "step": 71560
    },
    {
      "epoch": 29.3,
      "grad_norm": 2.9059951305389404,
      "learning_rate": 4.56921794657046e-06,
      "loss": 0.5731,
      "step": 71570
    },
    {
      "epoch": 29.3,
      "grad_norm": 3.0459158420562744,
      "learning_rate": 4.569168184589878e-06,
      "loss": 0.5808,
      "step": 71580
    },
    {
      "epoch": 29.3,
      "grad_norm": 2.845625638961792,
      "learning_rate": 4.569118412415e-06,
      "loss": 0.5666,
      "step": 71590
    },
    {
      "epoch": 29.31,
      "grad_norm": 2.4159228801727295,
      "learning_rate": 4.569068630046054e-06,
      "loss": 0.58,
      "step": 71600
    },
    {
      "epoch": 29.31,
      "grad_norm": 2.7850840091705322,
      "learning_rate": 4.569018837483268e-06,
      "loss": 0.5781,
      "step": 71610
    },
    {
      "epoch": 29.32,
      "grad_norm": 2.158647060394287,
      "learning_rate": 4.568969034726869e-06,
      "loss": 0.5773,
      "step": 71620
    },
    {
      "epoch": 29.32,
      "grad_norm": 2.418156862258911,
      "learning_rate": 4.568919221777088e-06,
      "loss": 0.5723,
      "step": 71630
    },
    {
      "epoch": 29.32,
      "grad_norm": 2.1640079021453857,
      "learning_rate": 4.5688693986341504e-06,
      "loss": 0.5605,
      "step": 71640
    },
    {
      "epoch": 29.33,
      "grad_norm": 2.87373948097229,
      "learning_rate": 4.568819565298286e-06,
      "loss": 0.5779,
      "step": 71650
    },
    {
      "epoch": 29.33,
      "grad_norm": 3.074082136154175,
      "learning_rate": 4.568769721769722e-06,
      "loss": 0.5812,
      "step": 71660
    },
    {
      "epoch": 29.34,
      "grad_norm": 3.371215343475342,
      "learning_rate": 4.568719868048687e-06,
      "loss": 0.5631,
      "step": 71670
    },
    {
      "epoch": 29.34,
      "grad_norm": 2.4845237731933594,
      "learning_rate": 4.56867000413541e-06,
      "loss": 0.5795,
      "step": 71680
    },
    {
      "epoch": 29.35,
      "grad_norm": 2.401554584503174,
      "learning_rate": 4.5686201300301185e-06,
      "loss": 0.5887,
      "step": 71690
    },
    {
      "epoch": 29.35,
      "grad_norm": 2.589092493057251,
      "learning_rate": 4.568570245733043e-06,
      "loss": 0.604,
      "step": 71700
    },
    {
      "epoch": 29.35,
      "grad_norm": 1.9529869556427002,
      "learning_rate": 4.568520351244409e-06,
      "loss": 0.5726,
      "step": 71710
    },
    {
      "epoch": 29.36,
      "grad_norm": 2.2949554920196533,
      "learning_rate": 4.568470446564447e-06,
      "loss": 0.5751,
      "step": 71720
    },
    {
      "epoch": 29.36,
      "grad_norm": 3.0051605701446533,
      "learning_rate": 4.568420531693386e-06,
      "loss": 0.5823,
      "step": 71730
    },
    {
      "epoch": 29.37,
      "grad_norm": 2.0935745239257812,
      "learning_rate": 4.5683706066314525e-06,
      "loss": 0.5717,
      "step": 71740
    },
    {
      "epoch": 29.37,
      "grad_norm": 2.8900444507598877,
      "learning_rate": 4.568320671378877e-06,
      "loss": 0.585,
      "step": 71750
    },
    {
      "epoch": 29.37,
      "grad_norm": 2.221048593521118,
      "learning_rate": 4.568270725935888e-06,
      "loss": 0.5726,
      "step": 71760
    },
    {
      "epoch": 29.38,
      "grad_norm": 1.825355887413025,
      "learning_rate": 4.568220770302714e-06,
      "loss": 0.5885,
      "step": 71770
    },
    {
      "epoch": 29.38,
      "grad_norm": 2.224696636199951,
      "learning_rate": 4.568170804479583e-06,
      "loss": 0.5999,
      "step": 71780
    },
    {
      "epoch": 29.39,
      "grad_norm": 2.1220316886901855,
      "learning_rate": 4.568120828466726e-06,
      "loss": 0.5656,
      "step": 71790
    },
    {
      "epoch": 29.39,
      "grad_norm": 2.1940722465515137,
      "learning_rate": 4.568070842264371e-06,
      "loss": 0.5793,
      "step": 71800
    },
    {
      "epoch": 29.39,
      "grad_norm": 1.894890546798706,
      "learning_rate": 4.568020845872746e-06,
      "loss": 0.5626,
      "step": 71810
    },
    {
      "epoch": 29.4,
      "grad_norm": 2.687429189682007,
      "learning_rate": 4.567970839292081e-06,
      "loss": 0.581,
      "step": 71820
    },
    {
      "epoch": 29.4,
      "grad_norm": 1.7818210124969482,
      "learning_rate": 4.567920822522604e-06,
      "loss": 0.5925,
      "step": 71830
    },
    {
      "epoch": 29.41,
      "grad_norm": 2.374974250793457,
      "learning_rate": 4.567870795564546e-06,
      "loss": 0.5758,
      "step": 71840
    },
    {
      "epoch": 29.41,
      "grad_norm": 2.900052309036255,
      "learning_rate": 4.567820758418134e-06,
      "loss": 0.5731,
      "step": 71850
    },
    {
      "epoch": 29.41,
      "grad_norm": 2.847477436065674,
      "learning_rate": 4.5677707110835995e-06,
      "loss": 0.5756,
      "step": 71860
    },
    {
      "epoch": 29.42,
      "grad_norm": 2.3613874912261963,
      "learning_rate": 4.56772065356117e-06,
      "loss": 0.5996,
      "step": 71870
    },
    {
      "epoch": 29.42,
      "grad_norm": 2.410592555999756,
      "learning_rate": 4.567670585851076e-06,
      "loss": 0.5705,
      "step": 71880
    },
    {
      "epoch": 29.43,
      "grad_norm": 2.6751415729522705,
      "learning_rate": 4.567620507953545e-06,
      "loss": 0.5963,
      "step": 71890
    },
    {
      "epoch": 29.43,
      "grad_norm": 2.65508770942688,
      "learning_rate": 4.567570419868807e-06,
      "loss": 0.6056,
      "step": 71900
    },
    {
      "epoch": 29.44,
      "grad_norm": 2.59393048286438,
      "learning_rate": 4.567520321597094e-06,
      "loss": 0.5825,
      "step": 71910
    },
    {
      "epoch": 29.44,
      "grad_norm": 2.1942460536956787,
      "learning_rate": 4.567470213138633e-06,
      "loss": 0.5956,
      "step": 71920
    },
    {
      "epoch": 29.44,
      "grad_norm": 2.3789918422698975,
      "learning_rate": 4.567420094493653e-06,
      "loss": 0.5721,
      "step": 71930
    },
    {
      "epoch": 29.45,
      "grad_norm": 3.287738561630249,
      "learning_rate": 4.567369965662386e-06,
      "loss": 0.5761,
      "step": 71940
    },
    {
      "epoch": 29.45,
      "grad_norm": 2.877326726913452,
      "learning_rate": 4.56731982664506e-06,
      "loss": 0.5755,
      "step": 71950
    },
    {
      "epoch": 29.46,
      "grad_norm": 2.593134880065918,
      "learning_rate": 4.567269677441905e-06,
      "loss": 0.5732,
      "step": 71960
    },
    {
      "epoch": 29.46,
      "grad_norm": 2.8460140228271484,
      "learning_rate": 4.5672195180531496e-06,
      "loss": 0.5744,
      "step": 71970
    },
    {
      "epoch": 29.46,
      "grad_norm": 2.504505157470703,
      "learning_rate": 4.567169348479026e-06,
      "loss": 0.5649,
      "step": 71980
    },
    {
      "epoch": 29.47,
      "grad_norm": 3.1448943614959717,
      "learning_rate": 4.567119168719762e-06,
      "loss": 0.6074,
      "step": 71990
    },
    {
      "epoch": 29.47,
      "grad_norm": 2.623483896255493,
      "learning_rate": 4.567068978775589e-06,
      "loss": 0.5702,
      "step": 72000
    },
    {
      "epoch": 29.48,
      "grad_norm": 2.220673084259033,
      "learning_rate": 4.567018778646736e-06,
      "loss": 0.6139,
      "step": 72010
    },
    {
      "epoch": 29.48,
      "grad_norm": 2.9833390712738037,
      "learning_rate": 4.566968568333432e-06,
      "loss": 0.5812,
      "step": 72020
    },
    {
      "epoch": 29.48,
      "grad_norm": 2.82568621635437,
      "learning_rate": 4.566918347835909e-06,
      "loss": 0.5933,
      "step": 72030
    },
    {
      "epoch": 29.49,
      "grad_norm": 2.828386068344116,
      "learning_rate": 4.566868117154395e-06,
      "loss": 0.5888,
      "step": 72040
    },
    {
      "epoch": 29.49,
      "grad_norm": 2.942668914794922,
      "learning_rate": 4.566817876289122e-06,
      "loss": 0.5686,
      "step": 72050
    },
    {
      "epoch": 29.5,
      "grad_norm": 1.881329894065857,
      "learning_rate": 4.56676762524032e-06,
      "loss": 0.5616,
      "step": 72060
    },
    {
      "epoch": 29.5,
      "grad_norm": 1.7961565256118774,
      "learning_rate": 4.566717364008218e-06,
      "loss": 0.5598,
      "step": 72070
    },
    {
      "epoch": 29.5,
      "grad_norm": 3.526948928833008,
      "learning_rate": 4.566667092593046e-06,
      "loss": 0.5995,
      "step": 72080
    },
    {
      "epoch": 29.51,
      "grad_norm": 2.3920068740844727,
      "learning_rate": 4.566616810995036e-06,
      "loss": 0.5708,
      "step": 72090
    },
    {
      "epoch": 29.51,
      "grad_norm": 2.9037201404571533,
      "learning_rate": 4.566566519214417e-06,
      "loss": 0.5686,
      "step": 72100
    },
    {
      "epoch": 29.52,
      "grad_norm": 2.3800032138824463,
      "learning_rate": 4.56651621725142e-06,
      "loss": 0.5812,
      "step": 72110
    },
    {
      "epoch": 29.52,
      "grad_norm": 2.026334047317505,
      "learning_rate": 4.566465905106275e-06,
      "loss": 0.5752,
      "step": 72120
    },
    {
      "epoch": 29.53,
      "grad_norm": 2.6250386238098145,
      "learning_rate": 4.566415582779213e-06,
      "loss": 0.5927,
      "step": 72130
    },
    {
      "epoch": 29.53,
      "grad_norm": 2.9407436847686768,
      "learning_rate": 4.5663652502704635e-06,
      "loss": 0.5552,
      "step": 72140
    },
    {
      "epoch": 29.53,
      "grad_norm": 2.052851915359497,
      "learning_rate": 4.566314907580258e-06,
      "loss": 0.5769,
      "step": 72150
    },
    {
      "epoch": 29.54,
      "grad_norm": 2.499818801879883,
      "learning_rate": 4.566264554708827e-06,
      "loss": 0.5667,
      "step": 72160
    },
    {
      "epoch": 29.54,
      "grad_norm": 3.097428798675537,
      "learning_rate": 4.5662141916564005e-06,
      "loss": 0.5701,
      "step": 72170
    },
    {
      "epoch": 29.55,
      "grad_norm": 2.4174482822418213,
      "learning_rate": 4.566163818423211e-06,
      "loss": 0.5656,
      "step": 72180
    },
    {
      "epoch": 29.55,
      "grad_norm": 2.1266372203826904,
      "learning_rate": 4.5661134350094875e-06,
      "loss": 0.5922,
      "step": 72190
    },
    {
      "epoch": 29.55,
      "grad_norm": 3.0795388221740723,
      "learning_rate": 4.566063041415461e-06,
      "loss": 0.5615,
      "step": 72200
    },
    {
      "epoch": 29.56,
      "grad_norm": 2.7044854164123535,
      "learning_rate": 4.566012637641362e-06,
      "loss": 0.5989,
      "step": 72210
    },
    {
      "epoch": 29.56,
      "grad_norm": 2.1626334190368652,
      "learning_rate": 4.565962223687423e-06,
      "loss": 0.5639,
      "step": 72220
    },
    {
      "epoch": 29.57,
      "grad_norm": 2.3809869289398193,
      "learning_rate": 4.565911799553874e-06,
      "loss": 0.5607,
      "step": 72230
    },
    {
      "epoch": 29.57,
      "grad_norm": 2.9039885997772217,
      "learning_rate": 4.5658613652409446e-06,
      "loss": 0.5778,
      "step": 72240
    },
    {
      "epoch": 29.57,
      "grad_norm": 2.318729877471924,
      "learning_rate": 4.565810920748869e-06,
      "loss": 0.5619,
      "step": 72250
    },
    {
      "epoch": 29.58,
      "grad_norm": 2.0275683403015137,
      "learning_rate": 4.565760466077875e-06,
      "loss": 0.5928,
      "step": 72260
    },
    {
      "epoch": 29.58,
      "grad_norm": 2.4986820220947266,
      "learning_rate": 4.565710001228196e-06,
      "loss": 0.5554,
      "step": 72270
    },
    {
      "epoch": 29.59,
      "grad_norm": 2.448634624481201,
      "learning_rate": 4.565659526200062e-06,
      "loss": 0.5711,
      "step": 72280
    },
    {
      "epoch": 29.59,
      "grad_norm": 1.8928955793380737,
      "learning_rate": 4.565609040993705e-06,
      "loss": 0.5743,
      "step": 72290
    },
    {
      "epoch": 29.59,
      "grad_norm": 2.70684552192688,
      "learning_rate": 4.565558545609355e-06,
      "loss": 0.5918,
      "step": 72300
    },
    {
      "epoch": 29.6,
      "grad_norm": 3.1057863235473633,
      "learning_rate": 4.565508040047245e-06,
      "loss": 0.5785,
      "step": 72310
    },
    {
      "epoch": 29.6,
      "grad_norm": 1.8760830163955688,
      "learning_rate": 4.565457524307605e-06,
      "loss": 0.5809,
      "step": 72320
    },
    {
      "epoch": 29.61,
      "grad_norm": 2.3474280834198,
      "learning_rate": 4.565406998390667e-06,
      "loss": 0.5607,
      "step": 72330
    },
    {
      "epoch": 29.61,
      "grad_norm": 2.641402244567871,
      "learning_rate": 4.565356462296663e-06,
      "loss": 0.5824,
      "step": 72340
    },
    {
      "epoch": 29.62,
      "grad_norm": 2.4302637577056885,
      "learning_rate": 4.5653059160258235e-06,
      "loss": 0.5915,
      "step": 72350
    },
    {
      "epoch": 29.62,
      "grad_norm": 2.8914783000946045,
      "learning_rate": 4.56525535957838e-06,
      "loss": 0.5694,
      "step": 72360
    },
    {
      "epoch": 29.62,
      "grad_norm": 2.1028552055358887,
      "learning_rate": 4.565204792954564e-06,
      "loss": 0.5579,
      "step": 72370
    },
    {
      "epoch": 29.63,
      "grad_norm": 2.241917371749878,
      "learning_rate": 4.565154216154609e-06,
      "loss": 0.576,
      "step": 72380
    },
    {
      "epoch": 29.63,
      "grad_norm": 1.9999436140060425,
      "learning_rate": 4.565103629178744e-06,
      "loss": 0.5602,
      "step": 72390
    },
    {
      "epoch": 29.64,
      "grad_norm": 2.745612144470215,
      "learning_rate": 4.565053032027203e-06,
      "loss": 0.5908,
      "step": 72400
    },
    {
      "epoch": 29.64,
      "grad_norm": 2.6384685039520264,
      "learning_rate": 4.5650024247002165e-06,
      "loss": 0.5954,
      "step": 72410
    },
    {
      "epoch": 29.64,
      "grad_norm": 2.564356565475464,
      "learning_rate": 4.5649518071980165e-06,
      "loss": 0.5944,
      "step": 72420
    },
    {
      "epoch": 29.65,
      "grad_norm": 2.3103950023651123,
      "learning_rate": 4.564901179520836e-06,
      "loss": 0.577,
      "step": 72430
    },
    {
      "epoch": 29.65,
      "grad_norm": 2.6941123008728027,
      "learning_rate": 4.564850541668904e-06,
      "loss": 0.5858,
      "step": 72440
    },
    {
      "epoch": 29.66,
      "grad_norm": 2.1754872798919678,
      "learning_rate": 4.564799893642455e-06,
      "loss": 0.5754,
      "step": 72450
    },
    {
      "epoch": 29.66,
      "grad_norm": 2.336486339569092,
      "learning_rate": 4.564749235441722e-06,
      "loss": 0.5617,
      "step": 72460
    },
    {
      "epoch": 29.66,
      "grad_norm": 2.570953607559204,
      "learning_rate": 4.5646985670669335e-06,
      "loss": 0.5769,
      "step": 72470
    },
    {
      "epoch": 29.67,
      "grad_norm": 2.7975282669067383,
      "learning_rate": 4.564647888518324e-06,
      "loss": 0.5788,
      "step": 72480
    },
    {
      "epoch": 29.67,
      "grad_norm": 2.2056398391723633,
      "learning_rate": 4.564597199796127e-06,
      "loss": 0.5933,
      "step": 72490
    },
    {
      "epoch": 29.68,
      "grad_norm": 2.446845054626465,
      "learning_rate": 4.564546500900571e-06,
      "loss": 0.5707,
      "step": 72500
    },
    {
      "epoch": 29.68,
      "grad_norm": 3.3222999572753906,
      "learning_rate": 4.56449579183189e-06,
      "loss": 0.5832,
      "step": 72510
    },
    {
      "epoch": 29.68,
      "grad_norm": 2.219311475753784,
      "learning_rate": 4.564445072590318e-06,
      "loss": 0.5748,
      "step": 72520
    },
    {
      "epoch": 29.69,
      "grad_norm": 2.590453863143921,
      "learning_rate": 4.564394343176085e-06,
      "loss": 0.5856,
      "step": 72530
    },
    {
      "epoch": 29.69,
      "grad_norm": 2.4813549518585205,
      "learning_rate": 4.564343603589424e-06,
      "loss": 0.5673,
      "step": 72540
    },
    {
      "epoch": 29.7,
      "grad_norm": 8.248137474060059,
      "learning_rate": 4.564292853830568e-06,
      "loss": 0.5889,
      "step": 72550
    },
    {
      "epoch": 29.7,
      "grad_norm": 2.865710735321045,
      "learning_rate": 4.5642420938997484e-06,
      "loss": 0.5796,
      "step": 72560
    },
    {
      "epoch": 29.71,
      "grad_norm": 2.3818843364715576,
      "learning_rate": 4.564191323797199e-06,
      "loss": 0.5747,
      "step": 72570
    },
    {
      "epoch": 29.71,
      "grad_norm": 3.285947322845459,
      "learning_rate": 4.564140543523152e-06,
      "loss": 0.5754,
      "step": 72580
    },
    {
      "epoch": 29.71,
      "grad_norm": 1.9647902250289917,
      "learning_rate": 4.5640897530778386e-06,
      "loss": 0.597,
      "step": 72590
    },
    {
      "epoch": 29.72,
      "grad_norm": 2.1870975494384766,
      "learning_rate": 4.564038952461494e-06,
      "loss": 0.5476,
      "step": 72600
    },
    {
      "epoch": 29.72,
      "grad_norm": 3.5122227668762207,
      "learning_rate": 4.563988141674348e-06,
      "loss": 0.5802,
      "step": 72610
    },
    {
      "epoch": 29.73,
      "grad_norm": 2.4046573638916016,
      "learning_rate": 4.563937320716637e-06,
      "loss": 0.5619,
      "step": 72620
    },
    {
      "epoch": 29.73,
      "grad_norm": 2.77846097946167,
      "learning_rate": 4.5638864895885905e-06,
      "loss": 0.5822,
      "step": 72630
    },
    {
      "epoch": 29.73,
      "grad_norm": 2.835479259490967,
      "learning_rate": 4.563835648290443e-06,
      "loss": 0.584,
      "step": 72640
    },
    {
      "epoch": 29.74,
      "grad_norm": 2.187511444091797,
      "learning_rate": 4.563784796822426e-06,
      "loss": 0.568,
      "step": 72650
    },
    {
      "epoch": 29.74,
      "grad_norm": 2.681859016418457,
      "learning_rate": 4.563733935184775e-06,
      "loss": 0.5895,
      "step": 72660
    },
    {
      "epoch": 29.75,
      "grad_norm": 3.1909191608428955,
      "learning_rate": 4.56368306337772e-06,
      "loss": 0.5904,
      "step": 72670
    },
    {
      "epoch": 29.75,
      "grad_norm": 2.194833993911743,
      "learning_rate": 4.5636321814014966e-06,
      "loss": 0.6025,
      "step": 72680
    },
    {
      "epoch": 29.75,
      "grad_norm": 2.7407703399658203,
      "learning_rate": 4.563581289256336e-06,
      "loss": 0.5985,
      "step": 72690
    },
    {
      "epoch": 29.76,
      "grad_norm": 1.8689193725585938,
      "learning_rate": 4.563530386942472e-06,
      "loss": 0.5509,
      "step": 72700
    },
    {
      "epoch": 29.76,
      "grad_norm": 3.2153730392456055,
      "learning_rate": 4.563479474460137e-06,
      "loss": 0.5797,
      "step": 72710
    },
    {
      "epoch": 29.77,
      "grad_norm": 1.8540688753128052,
      "learning_rate": 4.563428551809567e-06,
      "loss": 0.5723,
      "step": 72720
    },
    {
      "epoch": 29.77,
      "grad_norm": 2.766801595687866,
      "learning_rate": 4.563377618990992e-06,
      "loss": 0.5799,
      "step": 72730
    },
    {
      "epoch": 29.77,
      "grad_norm": 2.1953694820404053,
      "learning_rate": 4.5633266760046475e-06,
      "loss": 0.5732,
      "step": 72740
    },
    {
      "epoch": 29.78,
      "grad_norm": 2.00839900970459,
      "learning_rate": 4.563275722850765e-06,
      "loss": 0.5895,
      "step": 72750
    },
    {
      "epoch": 29.78,
      "grad_norm": 2.083139657974243,
      "learning_rate": 4.56322475952958e-06,
      "loss": 0.5699,
      "step": 72760
    },
    {
      "epoch": 29.79,
      "grad_norm": 2.0881574153900146,
      "learning_rate": 4.5631737860413235e-06,
      "loss": 0.5788,
      "step": 72770
    },
    {
      "epoch": 29.79,
      "grad_norm": 1.7758471965789795,
      "learning_rate": 4.563122802386231e-06,
      "loss": 0.5667,
      "step": 72780
    },
    {
      "epoch": 29.8,
      "grad_norm": 2.9534451961517334,
      "learning_rate": 4.563071808564535e-06,
      "loss": 0.592,
      "step": 72790
    },
    {
      "epoch": 29.8,
      "grad_norm": 3.8028476238250732,
      "learning_rate": 4.56302080457647e-06,
      "loss": 0.5779,
      "step": 72800
    },
    {
      "epoch": 29.8,
      "grad_norm": 2.075554132461548,
      "learning_rate": 4.56296979042227e-06,
      "loss": 0.5679,
      "step": 72810
    },
    {
      "epoch": 29.81,
      "grad_norm": 3.0720136165618896,
      "learning_rate": 4.562918766102166e-06,
      "loss": 0.5691,
      "step": 72820
    },
    {
      "epoch": 29.81,
      "grad_norm": 2.2169508934020996,
      "learning_rate": 4.562867731616395e-06,
      "loss": 0.5603,
      "step": 72830
    },
    {
      "epoch": 29.82,
      "grad_norm": 2.008148431777954,
      "learning_rate": 4.562816686965189e-06,
      "loss": 0.5835,
      "step": 72840
    },
    {
      "epoch": 29.82,
      "grad_norm": 2.226095676422119,
      "learning_rate": 4.562765632148782e-06,
      "loss": 0.573,
      "step": 72850
    },
    {
      "epoch": 29.82,
      "grad_norm": 2.504221200942993,
      "learning_rate": 4.562714567167407e-06,
      "loss": 0.5627,
      "step": 72860
    },
    {
      "epoch": 29.83,
      "grad_norm": 2.8917784690856934,
      "learning_rate": 4.562663492021301e-06,
      "loss": 0.5818,
      "step": 72870
    },
    {
      "epoch": 29.83,
      "grad_norm": 2.511821746826172,
      "learning_rate": 4.562612406710694e-06,
      "loss": 0.5509,
      "step": 72880
    },
    {
      "epoch": 29.84,
      "grad_norm": 2.947701930999756,
      "learning_rate": 4.562561311235823e-06,
      "loss": 0.5832,
      "step": 72890
    },
    {
      "epoch": 29.84,
      "grad_norm": 2.644174098968506,
      "learning_rate": 4.5625102055969215e-06,
      "loss": 0.5719,
      "step": 72900
    },
    {
      "epoch": 29.84,
      "grad_norm": 2.153553009033203,
      "learning_rate": 4.562459089794223e-06,
      "loss": 0.5745,
      "step": 72910
    },
    {
      "epoch": 29.85,
      "grad_norm": 2.045011281967163,
      "learning_rate": 4.562407963827961e-06,
      "loss": 0.5717,
      "step": 72920
    },
    {
      "epoch": 29.85,
      "grad_norm": 2.6821722984313965,
      "learning_rate": 4.56235682769837e-06,
      "loss": 0.5973,
      "step": 72930
    },
    {
      "epoch": 29.86,
      "grad_norm": 2.250880479812622,
      "learning_rate": 4.562305681405686e-06,
      "loss": 0.5698,
      "step": 72940
    },
    {
      "epoch": 29.86,
      "grad_norm": 3.645805597305298,
      "learning_rate": 4.5622545249501416e-06,
      "loss": 0.5712,
      "step": 72950
    },
    {
      "epoch": 29.86,
      "grad_norm": 2.664524793624878,
      "learning_rate": 4.5622033583319715e-06,
      "loss": 0.5868,
      "step": 72960
    },
    {
      "epoch": 29.87,
      "grad_norm": 2.156508445739746,
      "learning_rate": 4.56215218155141e-06,
      "loss": 0.5667,
      "step": 72970
    },
    {
      "epoch": 29.87,
      "grad_norm": 3.1901910305023193,
      "learning_rate": 4.562100994608691e-06,
      "loss": 0.5877,
      "step": 72980
    },
    {
      "epoch": 29.88,
      "grad_norm": 1.9873735904693604,
      "learning_rate": 4.562049797504051e-06,
      "loss": 0.5741,
      "step": 72990
    },
    {
      "epoch": 29.88,
      "grad_norm": 3.143439292907715,
      "learning_rate": 4.561998590237723e-06,
      "loss": 0.5916,
      "step": 73000
    },
    {
      "epoch": 29.89,
      "grad_norm": 2.3698179721832275,
      "learning_rate": 4.56194737280994e-06,
      "loss": 0.5869,
      "step": 73010
    },
    {
      "epoch": 29.89,
      "grad_norm": 2.589972972869873,
      "learning_rate": 4.561896145220941e-06,
      "loss": 0.5932,
      "step": 73020
    },
    {
      "epoch": 29.89,
      "grad_norm": 3.4469707012176514,
      "learning_rate": 4.561844907470956e-06,
      "loss": 0.5809,
      "step": 73030
    },
    {
      "epoch": 29.9,
      "grad_norm": 2.1657252311706543,
      "learning_rate": 4.561793659560223e-06,
      "loss": 0.5777,
      "step": 73040
    },
    {
      "epoch": 29.9,
      "grad_norm": 1.9601832628250122,
      "learning_rate": 4.561742401488974e-06,
      "loss": 0.576,
      "step": 73050
    },
    {
      "epoch": 29.91,
      "grad_norm": 2.174518585205078,
      "learning_rate": 4.561691133257447e-06,
      "loss": 0.5629,
      "step": 73060
    },
    {
      "epoch": 29.91,
      "grad_norm": 2.0852580070495605,
      "learning_rate": 4.561639854865874e-06,
      "loss": 0.5698,
      "step": 73070
    },
    {
      "epoch": 29.91,
      "grad_norm": 2.6609115600585938,
      "learning_rate": 4.561588566314492e-06,
      "loss": 0.542,
      "step": 73080
    },
    {
      "epoch": 29.92,
      "grad_norm": 2.656994342803955,
      "learning_rate": 4.561537267603535e-06,
      "loss": 0.5705,
      "step": 73090
    },
    {
      "epoch": 29.92,
      "grad_norm": 2.2250518798828125,
      "learning_rate": 4.561485958733237e-06,
      "loss": 0.5826,
      "step": 73100
    },
    {
      "epoch": 29.93,
      "grad_norm": 2.5529050827026367,
      "learning_rate": 4.5614346397038344e-06,
      "loss": 0.5701,
      "step": 73110
    },
    {
      "epoch": 29.93,
      "grad_norm": 2.318167209625244,
      "learning_rate": 4.561383310515562e-06,
      "loss": 0.592,
      "step": 73120
    },
    {
      "epoch": 29.93,
      "grad_norm": 1.557431936264038,
      "learning_rate": 4.561331971168655e-06,
      "loss": 0.564,
      "step": 73130
    },
    {
      "epoch": 29.94,
      "grad_norm": 2.3632912635803223,
      "learning_rate": 4.5612806216633485e-06,
      "loss": 0.5677,
      "step": 73140
    },
    {
      "epoch": 29.94,
      "grad_norm": 2.408649444580078,
      "learning_rate": 4.561229261999877e-06,
      "loss": 0.5709,
      "step": 73150
    },
    {
      "epoch": 29.95,
      "grad_norm": 1.948576807975769,
      "learning_rate": 4.561177892178477e-06,
      "loss": 0.5628,
      "step": 73160
    },
    {
      "epoch": 29.95,
      "grad_norm": 1.687437891960144,
      "learning_rate": 4.561126512199384e-06,
      "loss": 0.5732,
      "step": 73170
    },
    {
      "epoch": 29.95,
      "grad_norm": 2.814894676208496,
      "learning_rate": 4.561075122062831e-06,
      "loss": 0.5877,
      "step": 73180
    },
    {
      "epoch": 29.96,
      "grad_norm": 2.661585569381714,
      "learning_rate": 4.561023721769056e-06,
      "loss": 0.5692,
      "step": 73190
    },
    {
      "epoch": 29.96,
      "grad_norm": 2.4577362537384033,
      "learning_rate": 4.560972311318292e-06,
      "loss": 0.5835,
      "step": 73200
    },
    {
      "epoch": 29.97,
      "grad_norm": 2.142359495162964,
      "learning_rate": 4.5609208907107775e-06,
      "loss": 0.603,
      "step": 73210
    },
    {
      "epoch": 29.97,
      "grad_norm": 2.5918545722961426,
      "learning_rate": 4.560869459946747e-06,
      "loss": 0.5861,
      "step": 73220
    },
    {
      "epoch": 29.98,
      "grad_norm": 2.6544229984283447,
      "learning_rate": 4.560818019026435e-06,
      "loss": 0.5809,
      "step": 73230
    },
    {
      "epoch": 29.98,
      "grad_norm": 2.575986385345459,
      "learning_rate": 4.560766567950077e-06,
      "loss": 0.5919,
      "step": 73240
    },
    {
      "epoch": 29.98,
      "grad_norm": 2.0974297523498535,
      "learning_rate": 4.56071510671791e-06,
      "loss": 0.5796,
      "step": 73250
    },
    {
      "epoch": 29.99,
      "grad_norm": 2.4291253089904785,
      "learning_rate": 4.56066363533017e-06,
      "loss": 0.5658,
      "step": 73260
    },
    {
      "epoch": 29.99,
      "grad_norm": 2.1309540271759033,
      "learning_rate": 4.5606121537870905e-06,
      "loss": 0.5925,
      "step": 73270
    },
    {
      "epoch": 30.0,
      "grad_norm": 1.9518986940383911,
      "learning_rate": 4.5605606620889095e-06,
      "loss": 0.5544,
      "step": 73280
    },
    {
      "epoch": 30.0,
      "grad_norm": 2.19706130027771,
      "learning_rate": 4.560509160235861e-06,
      "loss": 0.5684,
      "step": 73290
    },
    {
      "epoch": 30.0,
      "eval_loss": 0.5790737867355347,
      "eval_runtime": 52.2782,
      "eval_samples_per_second": 65.974,
      "eval_steps_per_second": 8.263,
      "step": 73290
    },
    {
      "epoch": 30.0,
      "grad_norm": 3.2700023651123047,
      "learning_rate": 4.560457648228185e-06,
      "loss": 0.5909,
      "step": 73300
    },
    {
      "epoch": 30.01,
      "grad_norm": 2.3640944957733154,
      "learning_rate": 4.560406126066113e-06,
      "loss": 0.5776,
      "step": 73310
    },
    {
      "epoch": 30.01,
      "grad_norm": 3.206660032272339,
      "learning_rate": 4.560354593749882e-06,
      "loss": 0.5702,
      "step": 73320
    },
    {
      "epoch": 30.02,
      "grad_norm": 2.690838575363159,
      "learning_rate": 4.560303051279728e-06,
      "loss": 0.5729,
      "step": 73330
    },
    {
      "epoch": 30.02,
      "grad_norm": 2.6422202587127686,
      "learning_rate": 4.56025149865589e-06,
      "loss": 0.5532,
      "step": 73340
    },
    {
      "epoch": 30.02,
      "grad_norm": 2.6151857376098633,
      "learning_rate": 4.5601999358786004e-06,
      "loss": 0.5887,
      "step": 73350
    },
    {
      "epoch": 30.03,
      "grad_norm": 3.0804219245910645,
      "learning_rate": 4.560148362948097e-06,
      "loss": 0.5772,
      "step": 73360
    },
    {
      "epoch": 30.03,
      "grad_norm": 2.403714179992676,
      "learning_rate": 4.5600967798646165e-06,
      "loss": 0.5605,
      "step": 73370
    },
    {
      "epoch": 30.04,
      "grad_norm": 2.4061760902404785,
      "learning_rate": 4.560045186628394e-06,
      "loss": 0.5798,
      "step": 73380
    },
    {
      "epoch": 30.04,
      "grad_norm": 2.2632126808166504,
      "learning_rate": 4.559993583239667e-06,
      "loss": 0.5801,
      "step": 73390
    },
    {
      "epoch": 30.05,
      "grad_norm": 2.2022600173950195,
      "learning_rate": 4.559941969698671e-06,
      "loss": 0.5559,
      "step": 73400
    },
    {
      "epoch": 30.05,
      "grad_norm": 2.372058153152466,
      "learning_rate": 4.559890346005643e-06,
      "loss": 0.5959,
      "step": 73410
    },
    {
      "epoch": 30.05,
      "grad_norm": 2.1215367317199707,
      "learning_rate": 4.559838712160819e-06,
      "loss": 0.5786,
      "step": 73420
    },
    {
      "epoch": 30.06,
      "grad_norm": 2.126641273498535,
      "learning_rate": 4.559787068164436e-06,
      "loss": 0.5886,
      "step": 73430
    },
    {
      "epoch": 30.06,
      "grad_norm": 2.5751242637634277,
      "learning_rate": 4.55973541401673e-06,
      "loss": 0.5586,
      "step": 73440
    },
    {
      "epoch": 30.07,
      "grad_norm": 2.074856758117676,
      "learning_rate": 4.559683749717938e-06,
      "loss": 0.5969,
      "step": 73450
    },
    {
      "epoch": 30.07,
      "grad_norm": 3.0837364196777344,
      "learning_rate": 4.5596320752682985e-06,
      "loss": 0.5729,
      "step": 73460
    },
    {
      "epoch": 30.07,
      "grad_norm": 2.1131973266601562,
      "learning_rate": 4.5595803906680445e-06,
      "loss": 0.581,
      "step": 73470
    },
    {
      "epoch": 30.08,
      "grad_norm": 2.848081588745117,
      "learning_rate": 4.559528695917415e-06,
      "loss": 0.5762,
      "step": 73480
    },
    {
      "epoch": 30.08,
      "grad_norm": 1.8090896606445312,
      "learning_rate": 4.559476991016647e-06,
      "loss": 0.567,
      "step": 73490
    },
    {
      "epoch": 30.09,
      "grad_norm": 3.000033378601074,
      "learning_rate": 4.559425275965976e-06,
      "loss": 0.5678,
      "step": 73500
    },
    {
      "epoch": 30.09,
      "grad_norm": 2.333256483078003,
      "learning_rate": 4.55937355076564e-06,
      "loss": 0.5637,
      "step": 73510
    },
    {
      "epoch": 30.09,
      "grad_norm": 2.2035868167877197,
      "learning_rate": 4.559321815415876e-06,
      "loss": 0.59,
      "step": 73520
    },
    {
      "epoch": 30.1,
      "grad_norm": 2.3246994018554688,
      "learning_rate": 4.55927006991692e-06,
      "loss": 0.5698,
      "step": 73530
    },
    {
      "epoch": 30.1,
      "grad_norm": 2.316525459289551,
      "learning_rate": 4.5592183142690105e-06,
      "loss": 0.5724,
      "step": 73540
    },
    {
      "epoch": 30.11,
      "grad_norm": 2.769268751144409,
      "learning_rate": 4.559166548472383e-06,
      "loss": 0.5626,
      "step": 73550
    },
    {
      "epoch": 30.11,
      "grad_norm": 1.8924822807312012,
      "learning_rate": 4.559114772527275e-06,
      "loss": 0.5679,
      "step": 73560
    },
    {
      "epoch": 30.11,
      "grad_norm": 2.7834696769714355,
      "learning_rate": 4.559062986433926e-06,
      "loss": 0.6131,
      "step": 73570
    },
    {
      "epoch": 30.12,
      "grad_norm": 2.7152109146118164,
      "learning_rate": 4.55901119019257e-06,
      "loss": 0.5669,
      "step": 73580
    },
    {
      "epoch": 30.12,
      "grad_norm": 2.6272034645080566,
      "learning_rate": 4.558959383803445e-06,
      "loss": 0.5769,
      "step": 73590
    },
    {
      "epoch": 30.13,
      "grad_norm": 2.340977668762207,
      "learning_rate": 4.55890756726679e-06,
      "loss": 0.5655,
      "step": 73600
    },
    {
      "epoch": 30.13,
      "grad_norm": 2.340299129486084,
      "learning_rate": 4.558855740582841e-06,
      "loss": 0.5855,
      "step": 73610
    },
    {
      "epoch": 30.14,
      "grad_norm": 2.1903343200683594,
      "learning_rate": 4.5588039037518355e-06,
      "loss": 0.5682,
      "step": 73620
    },
    {
      "epoch": 30.14,
      "grad_norm": 2.222724676132202,
      "learning_rate": 4.558752056774011e-06,
      "loss": 0.5772,
      "step": 73630
    },
    {
      "epoch": 30.14,
      "grad_norm": 3.691084861755371,
      "learning_rate": 4.5587001996496044e-06,
      "loss": 0.556,
      "step": 73640
    },
    {
      "epoch": 30.15,
      "grad_norm": 2.0930399894714355,
      "learning_rate": 4.558648332378855e-06,
      "loss": 0.559,
      "step": 73650
    },
    {
      "epoch": 30.15,
      "grad_norm": 2.7741613388061523,
      "learning_rate": 4.558596454962e-06,
      "loss": 0.58,
      "step": 73660
    },
    {
      "epoch": 30.16,
      "grad_norm": 3.0881500244140625,
      "learning_rate": 4.558544567399275e-06,
      "loss": 0.5596,
      "step": 73670
    },
    {
      "epoch": 30.16,
      "grad_norm": 2.159381866455078,
      "learning_rate": 4.558492669690919e-06,
      "loss": 0.5842,
      "step": 73680
    },
    {
      "epoch": 30.16,
      "grad_norm": 1.8411682844161987,
      "learning_rate": 4.55844076183717e-06,
      "loss": 0.5852,
      "step": 73690
    },
    {
      "epoch": 30.17,
      "grad_norm": 2.2383952140808105,
      "learning_rate": 4.558388843838267e-06,
      "loss": 0.5784,
      "step": 73700
    },
    {
      "epoch": 30.17,
      "grad_norm": 3.0344245433807373,
      "learning_rate": 4.558336915694445e-06,
      "loss": 0.5835,
      "step": 73710
    },
    {
      "epoch": 30.18,
      "grad_norm": 2.4709203243255615,
      "learning_rate": 4.558284977405943e-06,
      "loss": 0.5694,
      "step": 73720
    },
    {
      "epoch": 30.18,
      "grad_norm": 2.511636257171631,
      "learning_rate": 4.558233028973001e-06,
      "loss": 0.5859,
      "step": 73730
    },
    {
      "epoch": 30.18,
      "grad_norm": 2.3553483486175537,
      "learning_rate": 4.558181070395854e-06,
      "loss": 0.5699,
      "step": 73740
    },
    {
      "epoch": 30.19,
      "grad_norm": 3.3476462364196777,
      "learning_rate": 4.558129101674741e-06,
      "loss": 0.5691,
      "step": 73750
    },
    {
      "epoch": 30.19,
      "grad_norm": 1.4787366390228271,
      "learning_rate": 4.558077122809901e-06,
      "loss": 0.5727,
      "step": 73760
    },
    {
      "epoch": 30.2,
      "grad_norm": 3.3781380653381348,
      "learning_rate": 4.558025133801571e-06,
      "loss": 0.56,
      "step": 73770
    },
    {
      "epoch": 30.2,
      "grad_norm": 2.498798131942749,
      "learning_rate": 4.55797313464999e-06,
      "loss": 0.573,
      "step": 73780
    },
    {
      "epoch": 30.2,
      "grad_norm": 2.4885988235473633,
      "learning_rate": 4.557921125355395e-06,
      "loss": 0.5634,
      "step": 73790
    },
    {
      "epoch": 30.21,
      "grad_norm": 1.9989150762557983,
      "learning_rate": 4.557869105918025e-06,
      "loss": 0.5889,
      "step": 73800
    },
    {
      "epoch": 30.21,
      "grad_norm": 2.4775662422180176,
      "learning_rate": 4.557817076338119e-06,
      "loss": 0.5813,
      "step": 73810
    },
    {
      "epoch": 30.22,
      "grad_norm": 2.697298049926758,
      "learning_rate": 4.557765036615914e-06,
      "loss": 0.5732,
      "step": 73820
    },
    {
      "epoch": 30.22,
      "grad_norm": 2.0503687858581543,
      "learning_rate": 4.557712986751649e-06,
      "loss": 0.5739,
      "step": 73830
    },
    {
      "epoch": 30.23,
      "grad_norm": 2.0388567447662354,
      "learning_rate": 4.557660926745563e-06,
      "loss": 0.5873,
      "step": 73840
    },
    {
      "epoch": 30.23,
      "grad_norm": 2.110725164413452,
      "learning_rate": 4.557608856597894e-06,
      "loss": 0.5611,
      "step": 73850
    },
    {
      "epoch": 30.23,
      "grad_norm": 2.320300579071045,
      "learning_rate": 4.55755677630888e-06,
      "loss": 0.5797,
      "step": 73860
    },
    {
      "epoch": 30.24,
      "grad_norm": 2.012662410736084,
      "learning_rate": 4.557504685878759e-06,
      "loss": 0.5819,
      "step": 73870
    },
    {
      "epoch": 30.24,
      "grad_norm": 2.41023325920105,
      "learning_rate": 4.557452585307772e-06,
      "loss": 0.5578,
      "step": 73880
    },
    {
      "epoch": 30.25,
      "grad_norm": 2.574998378753662,
      "learning_rate": 4.557400474596156e-06,
      "loss": 0.5647,
      "step": 73890
    },
    {
      "epoch": 30.25,
      "grad_norm": 1.9031906127929688,
      "learning_rate": 4.55734835374415e-06,
      "loss": 0.5759,
      "step": 73900
    },
    {
      "epoch": 30.25,
      "grad_norm": 2.373345136642456,
      "learning_rate": 4.557296222751992e-06,
      "loss": 0.5736,
      "step": 73910
    },
    {
      "epoch": 30.26,
      "grad_norm": 2.87752366065979,
      "learning_rate": 4.557244081619923e-06,
      "loss": 0.569,
      "step": 73920
    },
    {
      "epoch": 30.26,
      "grad_norm": 2.1687734127044678,
      "learning_rate": 4.557191930348179e-06,
      "loss": 0.5857,
      "step": 73930
    },
    {
      "epoch": 30.27,
      "grad_norm": 2.0666568279266357,
      "learning_rate": 4.557139768937002e-06,
      "loss": 0.5771,
      "step": 73940
    },
    {
      "epoch": 30.27,
      "grad_norm": 2.570608377456665,
      "learning_rate": 4.557087597386628e-06,
      "loss": 0.5646,
      "step": 73950
    },
    {
      "epoch": 30.27,
      "grad_norm": 2.191754102706909,
      "learning_rate": 4.557035415697297e-06,
      "loss": 0.5764,
      "step": 73960
    },
    {
      "epoch": 30.28,
      "grad_norm": 2.2734830379486084,
      "learning_rate": 4.556983223869249e-06,
      "loss": 0.6186,
      "step": 73970
    },
    {
      "epoch": 30.28,
      "grad_norm": 2.1498634815216064,
      "learning_rate": 4.556931021902722e-06,
      "loss": 0.598,
      "step": 73980
    },
    {
      "epoch": 30.29,
      "grad_norm": 2.445107936859131,
      "learning_rate": 4.556878809797956e-06,
      "loss": 0.5735,
      "step": 73990
    },
    {
      "epoch": 30.29,
      "grad_norm": 2.5302047729492188,
      "learning_rate": 4.556826587555189e-06,
      "loss": 0.5808,
      "step": 74000
    },
    {
      "epoch": 30.29,
      "grad_norm": 2.231691360473633,
      "learning_rate": 4.556774355174661e-06,
      "loss": 0.5712,
      "step": 74010
    },
    {
      "epoch": 30.3,
      "grad_norm": 2.2796597480773926,
      "learning_rate": 4.556722112656611e-06,
      "loss": 0.5881,
      "step": 74020
    },
    {
      "epoch": 30.3,
      "grad_norm": 3.283581018447876,
      "learning_rate": 4.556669860001279e-06,
      "loss": 0.5557,
      "step": 74030
    },
    {
      "epoch": 30.31,
      "grad_norm": 2.777658700942993,
      "learning_rate": 4.556617597208904e-06,
      "loss": 0.5874,
      "step": 74040
    },
    {
      "epoch": 30.31,
      "grad_norm": 2.0651144981384277,
      "learning_rate": 4.556565324279724e-06,
      "loss": 0.5738,
      "step": 74050
    },
    {
      "epoch": 30.32,
      "grad_norm": 2.158032178878784,
      "learning_rate": 4.556513041213981e-06,
      "loss": 0.5779,
      "step": 74060
    },
    {
      "epoch": 30.32,
      "grad_norm": 1.8682752847671509,
      "learning_rate": 4.556460748011913e-06,
      "loss": 0.5643,
      "step": 74070
    },
    {
      "epoch": 30.32,
      "grad_norm": 3.466402053833008,
      "learning_rate": 4.5564084446737595e-06,
      "loss": 0.5853,
      "step": 74080
    },
    {
      "epoch": 30.33,
      "grad_norm": 2.983581304550171,
      "learning_rate": 4.55635613119976e-06,
      "loss": 0.5715,
      "step": 74090
    },
    {
      "epoch": 30.33,
      "grad_norm": 2.0929057598114014,
      "learning_rate": 4.556303807590155e-06,
      "loss": 0.5652,
      "step": 74100
    },
    {
      "epoch": 30.34,
      "grad_norm": 3.127767562866211,
      "learning_rate": 4.556251473845183e-06,
      "loss": 0.5733,
      "step": 74110
    },
    {
      "epoch": 30.34,
      "grad_norm": 2.3442671298980713,
      "learning_rate": 4.556199129965085e-06,
      "loss": 0.5884,
      "step": 74120
    },
    {
      "epoch": 30.34,
      "grad_norm": 2.6741421222686768,
      "learning_rate": 4.5561467759501e-06,
      "loss": 0.6065,
      "step": 74130
    },
    {
      "epoch": 30.35,
      "grad_norm": 2.7216811180114746,
      "learning_rate": 4.556094411800469e-06,
      "loss": 0.5938,
      "step": 74140
    },
    {
      "epoch": 30.35,
      "grad_norm": 2.048032283782959,
      "learning_rate": 4.556042037516429e-06,
      "loss": 0.5631,
      "step": 74150
    },
    {
      "epoch": 30.36,
      "grad_norm": 3.5092525482177734,
      "learning_rate": 4.5559896530982225e-06,
      "loss": 0.5949,
      "step": 74160
    },
    {
      "epoch": 30.36,
      "grad_norm": 2.241354465484619,
      "learning_rate": 4.55593725854609e-06,
      "loss": 0.6003,
      "step": 74170
    },
    {
      "epoch": 30.36,
      "grad_norm": 2.6240363121032715,
      "learning_rate": 4.5558848538602686e-06,
      "loss": 0.5791,
      "step": 74180
    },
    {
      "epoch": 30.37,
      "grad_norm": 2.9397215843200684,
      "learning_rate": 4.555832439041002e-06,
      "loss": 0.5667,
      "step": 74190
    },
    {
      "epoch": 30.37,
      "grad_norm": 3.4376559257507324,
      "learning_rate": 4.555780014088526e-06,
      "loss": 0.5936,
      "step": 74200
    },
    {
      "epoch": 30.38,
      "grad_norm": 2.702097177505493,
      "learning_rate": 4.555727579003084e-06,
      "loss": 0.5693,
      "step": 74210
    },
    {
      "epoch": 30.38,
      "grad_norm": 2.516784429550171,
      "learning_rate": 4.5556751337849155e-06,
      "loss": 0.6073,
      "step": 74220
    },
    {
      "epoch": 30.38,
      "grad_norm": 2.6835145950317383,
      "learning_rate": 4.555622678434261e-06,
      "loss": 0.5663,
      "step": 74230
    },
    {
      "epoch": 30.39,
      "grad_norm": 2.9480905532836914,
      "learning_rate": 4.5555702129513595e-06,
      "loss": 0.5872,
      "step": 74240
    },
    {
      "epoch": 30.39,
      "grad_norm": 2.980289936065674,
      "learning_rate": 4.555517737336452e-06,
      "loss": 0.5922,
      "step": 74250
    },
    {
      "epoch": 30.4,
      "grad_norm": 2.4471209049224854,
      "learning_rate": 4.555465251589779e-06,
      "loss": 0.5791,
      "step": 74260
    },
    {
      "epoch": 30.4,
      "grad_norm": 2.876150131225586,
      "learning_rate": 4.555412755711582e-06,
      "loss": 0.6049,
      "step": 74270
    },
    {
      "epoch": 30.41,
      "grad_norm": 2.2854583263397217,
      "learning_rate": 4.555360249702099e-06,
      "loss": 0.5603,
      "step": 74280
    },
    {
      "epoch": 30.41,
      "grad_norm": 2.2792065143585205,
      "learning_rate": 4.555307733561574e-06,
      "loss": 0.5694,
      "step": 74290
    },
    {
      "epoch": 30.41,
      "grad_norm": 2.779060125350952,
      "learning_rate": 4.5552552072902435e-06,
      "loss": 0.5577,
      "step": 74300
    },
    {
      "epoch": 30.42,
      "grad_norm": 2.6928765773773193,
      "learning_rate": 4.555202670888351e-06,
      "loss": 0.5977,
      "step": 74310
    },
    {
      "epoch": 30.42,
      "grad_norm": 3.7244319915771484,
      "learning_rate": 4.555150124356136e-06,
      "loss": 0.5598,
      "step": 74320
    },
    {
      "epoch": 30.43,
      "grad_norm": 2.6279804706573486,
      "learning_rate": 4.555097567693839e-06,
      "loss": 0.5603,
      "step": 74330
    },
    {
      "epoch": 30.43,
      "grad_norm": 2.1546220779418945,
      "learning_rate": 4.555045000901702e-06,
      "loss": 0.5753,
      "step": 74340
    },
    {
      "epoch": 30.43,
      "grad_norm": 2.18668270111084,
      "learning_rate": 4.554992423979966e-06,
      "loss": 0.5811,
      "step": 74350
    },
    {
      "epoch": 30.44,
      "grad_norm": 2.7277791500091553,
      "learning_rate": 4.554939836928869e-06,
      "loss": 0.5747,
      "step": 74360
    },
    {
      "epoch": 30.44,
      "grad_norm": 2.695892095565796,
      "learning_rate": 4.554887239748655e-06,
      "loss": 0.5803,
      "step": 74370
    },
    {
      "epoch": 30.45,
      "grad_norm": 2.057778835296631,
      "learning_rate": 4.554834632439564e-06,
      "loss": 0.569,
      "step": 74380
    },
    {
      "epoch": 30.45,
      "grad_norm": 2.553964614868164,
      "learning_rate": 4.554782015001836e-06,
      "loss": 0.5536,
      "step": 74390
    },
    {
      "epoch": 30.45,
      "grad_norm": 2.432440996170044,
      "learning_rate": 4.5547293874357135e-06,
      "loss": 0.5568,
      "step": 74400
    },
    {
      "epoch": 30.46,
      "grad_norm": 2.3676955699920654,
      "learning_rate": 4.554676749741436e-06,
      "loss": 0.6118,
      "step": 74410
    },
    {
      "epoch": 30.46,
      "grad_norm": 4.0132341384887695,
      "learning_rate": 4.5546241019192455e-06,
      "loss": 0.5736,
      "step": 74420
    },
    {
      "epoch": 30.47,
      "grad_norm": 2.649637222290039,
      "learning_rate": 4.554571443969384e-06,
      "loss": 0.5586,
      "step": 74430
    },
    {
      "epoch": 30.47,
      "grad_norm": 2.1438987255096436,
      "learning_rate": 4.554518775892091e-06,
      "loss": 0.5556,
      "step": 74440
    },
    {
      "epoch": 30.47,
      "grad_norm": 2.677448272705078,
      "learning_rate": 4.554466097687609e-06,
      "loss": 0.5686,
      "step": 74450
    },
    {
      "epoch": 30.48,
      "grad_norm": 2.2688777446746826,
      "learning_rate": 4.554413409356179e-06,
      "loss": 0.5818,
      "step": 74460
    },
    {
      "epoch": 30.48,
      "grad_norm": 2.205437421798706,
      "learning_rate": 4.554360710898042e-06,
      "loss": 0.5668,
      "step": 74470
    },
    {
      "epoch": 30.49,
      "grad_norm": 2.0453600883483887,
      "learning_rate": 4.554308002313439e-06,
      "loss": 0.5658,
      "step": 74480
    },
    {
      "epoch": 30.49,
      "grad_norm": 2.380373954772949,
      "learning_rate": 4.554255283602613e-06,
      "loss": 0.5761,
      "step": 74490
    },
    {
      "epoch": 30.5,
      "grad_norm": 2.8123109340667725,
      "learning_rate": 4.554202554765805e-06,
      "loss": 0.5682,
      "step": 74500
    },
    {
      "epoch": 30.5,
      "grad_norm": 2.864351272583008,
      "learning_rate": 4.554149815803256e-06,
      "loss": 0.5891,
      "step": 74510
    },
    {
      "epoch": 30.5,
      "grad_norm": 1.867017388343811,
      "learning_rate": 4.554097066715208e-06,
      "loss": 0.5887,
      "step": 74520
    },
    {
      "epoch": 30.51,
      "grad_norm": 2.4433937072753906,
      "learning_rate": 4.554044307501901e-06,
      "loss": 0.5895,
      "step": 74530
    },
    {
      "epoch": 30.51,
      "grad_norm": 2.4415080547332764,
      "learning_rate": 4.553991538163579e-06,
      "loss": 0.5713,
      "step": 74540
    },
    {
      "epoch": 30.52,
      "grad_norm": 2.3055803775787354,
      "learning_rate": 4.5539387587004835e-06,
      "loss": 0.596,
      "step": 74550
    },
    {
      "epoch": 30.52,
      "grad_norm": 2.3569793701171875,
      "learning_rate": 4.553885969112854e-06,
      "loss": 0.5607,
      "step": 74560
    },
    {
      "epoch": 30.52,
      "grad_norm": 2.3773605823516846,
      "learning_rate": 4.553833169400935e-06,
      "loss": 0.553,
      "step": 74570
    },
    {
      "epoch": 30.53,
      "grad_norm": 2.3246963024139404,
      "learning_rate": 4.5537803595649675e-06,
      "loss": 0.5758,
      "step": 74580
    },
    {
      "epoch": 30.53,
      "grad_norm": 1.905239224433899,
      "learning_rate": 4.5537275396051925e-06,
      "loss": 0.5667,
      "step": 74590
    },
    {
      "epoch": 30.54,
      "grad_norm": 2.3587794303894043,
      "learning_rate": 4.5536747095218525e-06,
      "loss": 0.5647,
      "step": 74600
    },
    {
      "epoch": 30.54,
      "grad_norm": 2.68255877494812,
      "learning_rate": 4.553621869315191e-06,
      "loss": 0.5958,
      "step": 74610
    },
    {
      "epoch": 30.54,
      "grad_norm": 2.6750564575195312,
      "learning_rate": 4.553569018985448e-06,
      "loss": 0.5601,
      "step": 74620
    },
    {
      "epoch": 30.55,
      "grad_norm": 2.252241849899292,
      "learning_rate": 4.553516158532866e-06,
      "loss": 0.5628,
      "step": 74630
    },
    {
      "epoch": 30.55,
      "grad_norm": 4.490936279296875,
      "learning_rate": 4.553463287957688e-06,
      "loss": 0.5635,
      "step": 74640
    },
    {
      "epoch": 30.56,
      "grad_norm": 1.8225915431976318,
      "learning_rate": 4.553410407260155e-06,
      "loss": 0.5689,
      "step": 74650
    },
    {
      "epoch": 30.56,
      "grad_norm": 3.6606783866882324,
      "learning_rate": 4.55335751644051e-06,
      "loss": 0.5807,
      "step": 74660
    },
    {
      "epoch": 30.56,
      "grad_norm": 2.767699718475342,
      "learning_rate": 4.553304615498995e-06,
      "loss": 0.5529,
      "step": 74670
    },
    {
      "epoch": 30.57,
      "grad_norm": 3.337402105331421,
      "learning_rate": 4.553251704435853e-06,
      "loss": 0.5591,
      "step": 74680
    },
    {
      "epoch": 30.57,
      "grad_norm": 1.982401728630066,
      "learning_rate": 4.553198783251325e-06,
      "loss": 0.5838,
      "step": 74690
    },
    {
      "epoch": 30.58,
      "grad_norm": 2.3252599239349365,
      "learning_rate": 4.553145851945656e-06,
      "loss": 0.5745,
      "step": 74700
    },
    {
      "epoch": 30.58,
      "grad_norm": 4.440683364868164,
      "learning_rate": 4.553092910519084e-06,
      "loss": 0.5961,
      "step": 74710
    },
    {
      "epoch": 30.59,
      "grad_norm": 2.1068239212036133,
      "learning_rate": 4.5530399589718566e-06,
      "loss": 0.6042,
      "step": 74720
    },
    {
      "epoch": 30.59,
      "grad_norm": 2.0317816734313965,
      "learning_rate": 4.552986997304213e-06,
      "loss": 0.5712,
      "step": 74730
    },
    {
      "epoch": 30.59,
      "grad_norm": 2.496258497238159,
      "learning_rate": 4.552934025516398e-06,
      "loss": 0.5932,
      "step": 74740
    },
    {
      "epoch": 30.6,
      "grad_norm": 2.0969982147216797,
      "learning_rate": 4.5528810436086525e-06,
      "loss": 0.5452,
      "step": 74750
    },
    {
      "epoch": 30.6,
      "grad_norm": 2.6826844215393066,
      "learning_rate": 4.55282805158122e-06,
      "loss": 0.5574,
      "step": 74760
    },
    {
      "epoch": 30.61,
      "grad_norm": 2.30128812789917,
      "learning_rate": 4.552775049434343e-06,
      "loss": 0.5853,
      "step": 74770
    },
    {
      "epoch": 30.61,
      "grad_norm": 2.986767292022705,
      "learning_rate": 4.552722037168264e-06,
      "loss": 0.5791,
      "step": 74780
    },
    {
      "epoch": 30.61,
      "grad_norm": 2.285210609436035,
      "learning_rate": 4.552669014783225e-06,
      "loss": 0.5639,
      "step": 74790
    },
    {
      "epoch": 30.62,
      "grad_norm": 1.9808436632156372,
      "learning_rate": 4.552615982279472e-06,
      "loss": 0.576,
      "step": 74800
    },
    {
      "epoch": 30.62,
      "grad_norm": 2.4967541694641113,
      "learning_rate": 4.552562939657246e-06,
      "loss": 0.5761,
      "step": 74810
    },
    {
      "epoch": 30.63,
      "grad_norm": 2.1020140647888184,
      "learning_rate": 4.552509886916789e-06,
      "loss": 0.574,
      "step": 74820
    },
    {
      "epoch": 30.63,
      "grad_norm": 2.2514936923980713,
      "learning_rate": 4.5524568240583454e-06,
      "loss": 0.5783,
      "step": 74830
    },
    {
      "epoch": 30.63,
      "grad_norm": 2.163752317428589,
      "learning_rate": 4.5524037510821575e-06,
      "loss": 0.5854,
      "step": 74840
    },
    {
      "epoch": 30.64,
      "grad_norm": 2.192033290863037,
      "learning_rate": 4.55235066798847e-06,
      "loss": 0.573,
      "step": 74850
    },
    {
      "epoch": 30.64,
      "grad_norm": 2.6855530738830566,
      "learning_rate": 4.552297574777524e-06,
      "loss": 0.5848,
      "step": 74860
    },
    {
      "epoch": 30.65,
      "grad_norm": 2.7709007263183594,
      "learning_rate": 4.552244471449564e-06,
      "loss": 0.5815,
      "step": 74870
    },
    {
      "epoch": 30.65,
      "grad_norm": 3.017925500869751,
      "learning_rate": 4.552191358004832e-06,
      "loss": 0.5877,
      "step": 74880
    },
    {
      "epoch": 30.65,
      "grad_norm": 2.4505646228790283,
      "learning_rate": 4.552138234443574e-06,
      "loss": 0.5665,
      "step": 74890
    },
    {
      "epoch": 30.66,
      "grad_norm": 2.506415605545044,
      "learning_rate": 4.552085100766029e-06,
      "loss": 0.5843,
      "step": 74900
    },
    {
      "epoch": 30.66,
      "grad_norm": 1.9410254955291748,
      "learning_rate": 4.552031956972445e-06,
      "loss": 0.5472,
      "step": 74910
    },
    {
      "epoch": 30.67,
      "grad_norm": 2.2084405422210693,
      "learning_rate": 4.5519788030630625e-06,
      "loss": 0.5726,
      "step": 74920
    },
    {
      "epoch": 30.67,
      "grad_norm": 1.9183833599090576,
      "learning_rate": 4.5519256390381266e-06,
      "loss": 0.5596,
      "step": 74930
    },
    {
      "epoch": 30.68,
      "grad_norm": 1.8430728912353516,
      "learning_rate": 4.551872464897878e-06,
      "loss": 0.5843,
      "step": 74940
    },
    {
      "epoch": 30.68,
      "grad_norm": 2.342449903488159,
      "learning_rate": 4.551819280642565e-06,
      "loss": 0.5905,
      "step": 74950
    },
    {
      "epoch": 30.68,
      "grad_norm": 3.326606512069702,
      "learning_rate": 4.551766086272427e-06,
      "loss": 0.5737,
      "step": 74960
    },
    {
      "epoch": 30.69,
      "grad_norm": 2.227644920349121,
      "learning_rate": 4.55171288178771e-06,
      "loss": 0.575,
      "step": 74970
    },
    {
      "epoch": 30.69,
      "grad_norm": 2.4077794551849365,
      "learning_rate": 4.551659667188656e-06,
      "loss": 0.5777,
      "step": 74980
    },
    {
      "epoch": 30.7,
      "grad_norm": 2.7337543964385986,
      "learning_rate": 4.551606442475511e-06,
      "loss": 0.5676,
      "step": 74990
    },
    {
      "epoch": 30.7,
      "grad_norm": 2.6525464057922363,
      "learning_rate": 4.5515532076485155e-06,
      "loss": 0.5669,
      "step": 75000
    },
    {
      "epoch": 30.7,
      "grad_norm": 2.2046985626220703,
      "learning_rate": 4.551499962707918e-06,
      "loss": 0.5759,
      "step": 75010
    },
    {
      "epoch": 30.71,
      "grad_norm": 2.8223674297332764,
      "learning_rate": 4.551446707653958e-06,
      "loss": 0.5747,
      "step": 75020
    },
    {
      "epoch": 30.71,
      "grad_norm": 3.349761962890625,
      "learning_rate": 4.551393442486882e-06,
      "loss": 0.5902,
      "step": 75030
    },
    {
      "epoch": 30.72,
      "grad_norm": 2.389418601989746,
      "learning_rate": 4.551340167206933e-06,
      "loss": 0.5808,
      "step": 75040
    },
    {
      "epoch": 30.72,
      "grad_norm": 2.8466567993164062,
      "learning_rate": 4.551286881814355e-06,
      "loss": 0.5776,
      "step": 75050
    },
    {
      "epoch": 30.72,
      "grad_norm": 2.257169485092163,
      "learning_rate": 4.551233586309394e-06,
      "loss": 0.5908,
      "step": 75060
    },
    {
      "epoch": 30.73,
      "grad_norm": 1.8082133531570435,
      "learning_rate": 4.551180280692292e-06,
      "loss": 0.557,
      "step": 75070
    },
    {
      "epoch": 30.73,
      "grad_norm": 2.5755302906036377,
      "learning_rate": 4.551126964963292e-06,
      "loss": 0.5709,
      "step": 75080
    },
    {
      "epoch": 30.74,
      "grad_norm": 1.92020845413208,
      "learning_rate": 4.551073639122642e-06,
      "loss": 0.5957,
      "step": 75090
    },
    {
      "epoch": 30.74,
      "grad_norm": 2.5148606300354004,
      "learning_rate": 4.551020303170584e-06,
      "loss": 0.5888,
      "step": 75100
    },
    {
      "epoch": 30.74,
      "grad_norm": 2.3836076259613037,
      "learning_rate": 4.550966957107362e-06,
      "loss": 0.5787,
      "step": 75110
    },
    {
      "epoch": 30.75,
      "grad_norm": 2.3643367290496826,
      "learning_rate": 4.55091360093322e-06,
      "loss": 0.5681,
      "step": 75120
    },
    {
      "epoch": 30.75,
      "grad_norm": 2.4374237060546875,
      "learning_rate": 4.550860234648405e-06,
      "loss": 0.5756,
      "step": 75130
    },
    {
      "epoch": 30.76,
      "grad_norm": 2.161043643951416,
      "learning_rate": 4.550806858253159e-06,
      "loss": 0.5595,
      "step": 75140
    },
    {
      "epoch": 30.76,
      "grad_norm": 3.6138901710510254,
      "learning_rate": 4.550753471747728e-06,
      "loss": 0.5797,
      "step": 75150
    },
    {
      "epoch": 30.77,
      "grad_norm": 2.2370765209198,
      "learning_rate": 4.5507000751323555e-06,
      "loss": 0.5983,
      "step": 75160
    },
    {
      "epoch": 30.77,
      "grad_norm": 2.805847406387329,
      "learning_rate": 4.550646668407286e-06,
      "loss": 0.5699,
      "step": 75170
    },
    {
      "epoch": 30.77,
      "grad_norm": 2.1407277584075928,
      "learning_rate": 4.550593251572766e-06,
      "loss": 0.566,
      "step": 75180
    },
    {
      "epoch": 30.78,
      "grad_norm": 1.8173235654830933,
      "learning_rate": 4.5505398246290375e-06,
      "loss": 0.5763,
      "step": 75190
    },
    {
      "epoch": 30.78,
      "grad_norm": 2.1162264347076416,
      "learning_rate": 4.550486387576347e-06,
      "loss": 0.5893,
      "step": 75200
    },
    {
      "epoch": 30.79,
      "grad_norm": 2.9505553245544434,
      "learning_rate": 4.550432940414939e-06,
      "loss": 0.5797,
      "step": 75210
    },
    {
      "epoch": 30.79,
      "grad_norm": 2.560417890548706,
      "learning_rate": 4.550379483145058e-06,
      "loss": 0.5662,
      "step": 75220
    },
    {
      "epoch": 30.79,
      "grad_norm": 3.17655611038208,
      "learning_rate": 4.55032601576695e-06,
      "loss": 0.5666,
      "step": 75230
    },
    {
      "epoch": 30.8,
      "grad_norm": 2.2040352821350098,
      "learning_rate": 4.550272538280858e-06,
      "loss": 0.5789,
      "step": 75240
    },
    {
      "epoch": 30.8,
      "grad_norm": 2.47017765045166,
      "learning_rate": 4.550219050687028e-06,
      "loss": 0.589,
      "step": 75250
    },
    {
      "epoch": 30.81,
      "grad_norm": 2.400482654571533,
      "learning_rate": 4.5501655529857046e-06,
      "loss": 0.5643,
      "step": 75260
    },
    {
      "epoch": 30.81,
      "grad_norm": 2.0534629821777344,
      "learning_rate": 4.550112045177134e-06,
      "loss": 0.6044,
      "step": 75270
    },
    {
      "epoch": 30.81,
      "grad_norm": 2.528550148010254,
      "learning_rate": 4.55005852726156e-06,
      "loss": 0.5709,
      "step": 75280
    },
    {
      "epoch": 30.82,
      "grad_norm": 2.1085009574890137,
      "learning_rate": 4.550004999239229e-06,
      "loss": 0.5798,
      "step": 75290
    },
    {
      "epoch": 30.82,
      "grad_norm": 2.27551531791687,
      "learning_rate": 4.549951461110385e-06,
      "loss": 0.5923,
      "step": 75300
    },
    {
      "epoch": 30.83,
      "grad_norm": 1.9497147798538208,
      "learning_rate": 4.549897912875274e-06,
      "loss": 0.5818,
      "step": 75310
    },
    {
      "epoch": 30.83,
      "grad_norm": 2.035292625427246,
      "learning_rate": 4.549844354534141e-06,
      "loss": 0.5774,
      "step": 75320
    },
    {
      "epoch": 30.84,
      "grad_norm": 2.4390041828155518,
      "learning_rate": 4.549790786087232e-06,
      "loss": 0.5925,
      "step": 75330
    },
    {
      "epoch": 30.84,
      "grad_norm": 2.8930397033691406,
      "learning_rate": 4.549737207534792e-06,
      "loss": 0.5922,
      "step": 75340
    },
    {
      "epoch": 30.84,
      "grad_norm": 2.3456063270568848,
      "learning_rate": 4.549683618877065e-06,
      "loss": 0.5785,
      "step": 75350
    },
    {
      "epoch": 30.85,
      "grad_norm": 2.26542067527771,
      "learning_rate": 4.549630020114298e-06,
      "loss": 0.5899,
      "step": 75360
    },
    {
      "epoch": 30.85,
      "grad_norm": 2.273632764816284,
      "learning_rate": 4.549576411246737e-06,
      "loss": 0.585,
      "step": 75370
    },
    {
      "epoch": 30.86,
      "grad_norm": 2.6031084060668945,
      "learning_rate": 4.549522792274626e-06,
      "loss": 0.5675,
      "step": 75380
    },
    {
      "epoch": 30.86,
      "grad_norm": 2.7040061950683594,
      "learning_rate": 4.549469163198212e-06,
      "loss": 0.5719,
      "step": 75390
    },
    {
      "epoch": 30.86,
      "grad_norm": 3.9018125534057617,
      "learning_rate": 4.549415524017739e-06,
      "loss": 0.5402,
      "step": 75400
    },
    {
      "epoch": 30.87,
      "grad_norm": 1.8170453310012817,
      "learning_rate": 4.5493618747334556e-06,
      "loss": 0.5694,
      "step": 75410
    },
    {
      "epoch": 30.87,
      "grad_norm": 2.3879737854003906,
      "learning_rate": 4.549308215345605e-06,
      "loss": 0.6007,
      "step": 75420
    },
    {
      "epoch": 30.88,
      "grad_norm": 2.5892696380615234,
      "learning_rate": 4.549254545854434e-06,
      "loss": 0.5739,
      "step": 75430
    },
    {
      "epoch": 30.88,
      "grad_norm": 2.4192121028900146,
      "learning_rate": 4.549200866260187e-06,
      "loss": 0.5732,
      "step": 75440
    },
    {
      "epoch": 30.88,
      "grad_norm": 2.437983274459839,
      "learning_rate": 4.549147176563113e-06,
      "loss": 0.5687,
      "step": 75450
    },
    {
      "epoch": 30.89,
      "grad_norm": 2.3913111686706543,
      "learning_rate": 4.549093476763454e-06,
      "loss": 0.5701,
      "step": 75460
    },
    {
      "epoch": 30.89,
      "grad_norm": 1.7930376529693604,
      "learning_rate": 4.54903976686146e-06,
      "loss": 0.5518,
      "step": 75470
    },
    {
      "epoch": 30.9,
      "grad_norm": 2.5952117443084717,
      "learning_rate": 4.548986046857374e-06,
      "loss": 0.595,
      "step": 75480
    },
    {
      "epoch": 30.9,
      "grad_norm": 2.153053045272827,
      "learning_rate": 4.548932316751444e-06,
      "loss": 0.5815,
      "step": 75490
    },
    {
      "epoch": 30.9,
      "grad_norm": 2.9057652950286865,
      "learning_rate": 4.5488785765439146e-06,
      "loss": 0.5865,
      "step": 75500
    },
    {
      "epoch": 30.91,
      "grad_norm": 2.0903263092041016,
      "learning_rate": 4.5488248262350335e-06,
      "loss": 0.5937,
      "step": 75510
    },
    {
      "epoch": 30.91,
      "grad_norm": 2.4210212230682373,
      "learning_rate": 4.548771065825045e-06,
      "loss": 0.5866,
      "step": 75520
    },
    {
      "epoch": 30.92,
      "grad_norm": 1.92710280418396,
      "learning_rate": 4.548717295314197e-06,
      "loss": 0.5502,
      "step": 75530
    },
    {
      "epoch": 30.92,
      "grad_norm": 2.2351176738739014,
      "learning_rate": 4.5486635147027355e-06,
      "loss": 0.5867,
      "step": 75540
    },
    {
      "epoch": 30.93,
      "grad_norm": 1.9764585494995117,
      "learning_rate": 4.548609723990907e-06,
      "loss": 0.559,
      "step": 75550
    },
    {
      "epoch": 30.93,
      "grad_norm": 1.9813809394836426,
      "learning_rate": 4.548555923178957e-06,
      "loss": 0.5672,
      "step": 75560
    },
    {
      "epoch": 30.93,
      "grad_norm": 2.8276419639587402,
      "learning_rate": 4.548502112267133e-06,
      "loss": 0.5544,
      "step": 75570
    },
    {
      "epoch": 30.94,
      "grad_norm": 2.1965391635894775,
      "learning_rate": 4.54844829125568e-06,
      "loss": 0.5797,
      "step": 75580
    },
    {
      "epoch": 30.94,
      "grad_norm": 2.28007435798645,
      "learning_rate": 4.548394460144846e-06,
      "loss": 0.5808,
      "step": 75590
    },
    {
      "epoch": 30.95,
      "grad_norm": 2.223693370819092,
      "learning_rate": 4.548340618934878e-06,
      "loss": 0.5746,
      "step": 75600
    },
    {
      "epoch": 30.95,
      "grad_norm": 2.1608691215515137,
      "learning_rate": 4.548286767626021e-06,
      "loss": 0.5645,
      "step": 75610
    },
    {
      "epoch": 30.95,
      "grad_norm": 2.3781490325927734,
      "learning_rate": 4.548232906218523e-06,
      "loss": 0.5824,
      "step": 75620
    },
    {
      "epoch": 30.96,
      "grad_norm": 2.9070911407470703,
      "learning_rate": 4.54817903471263e-06,
      "loss": 0.5906,
      "step": 75630
    },
    {
      "epoch": 30.96,
      "grad_norm": 2.616732120513916,
      "learning_rate": 4.5481251531085886e-06,
      "loss": 0.5778,
      "step": 75640
    },
    {
      "epoch": 30.97,
      "grad_norm": 2.8526835441589355,
      "learning_rate": 4.548071261406646e-06,
      "loss": 0.5383,
      "step": 75650
    },
    {
      "epoch": 30.97,
      "grad_norm": 2.348051071166992,
      "learning_rate": 4.548017359607049e-06,
      "loss": 0.5719,
      "step": 75660
    },
    {
      "epoch": 30.97,
      "grad_norm": 2.925445079803467,
      "learning_rate": 4.547963447710045e-06,
      "loss": 0.587,
      "step": 75670
    },
    {
      "epoch": 30.98,
      "grad_norm": 2.1671810150146484,
      "learning_rate": 4.547909525715881e-06,
      "loss": 0.5751,
      "step": 75680
    },
    {
      "epoch": 30.98,
      "grad_norm": 2.161707639694214,
      "learning_rate": 4.547855593624803e-06,
      "loss": 0.5667,
      "step": 75690
    },
    {
      "epoch": 30.99,
      "grad_norm": 1.9977715015411377,
      "learning_rate": 4.547801651437059e-06,
      "loss": 0.5744,
      "step": 75700
    },
    {
      "epoch": 30.99,
      "grad_norm": 3.13431453704834,
      "learning_rate": 4.547747699152895e-06,
      "loss": 0.5658,
      "step": 75710
    },
    {
      "epoch": 30.99,
      "grad_norm": 2.908728837966919,
      "learning_rate": 4.547693736772559e-06,
      "loss": 0.5638,
      "step": 75720
    },
    {
      "epoch": 31.0,
      "grad_norm": 1.8795976638793945,
      "learning_rate": 4.547639764296299e-06,
      "loss": 0.5689,
      "step": 75730
    },
    {
      "epoch": 31.0,
      "eval_loss": 0.5725856423377991,
      "eval_runtime": 52.0036,
      "eval_samples_per_second": 66.322,
      "eval_steps_per_second": 8.307,
      "step": 75733
    },
    {
      "epoch": 31.0,
      "grad_norm": 2.164332866668701,
      "learning_rate": 4.547585781724361e-06,
      "loss": 0.5531,
      "step": 75740
    },
    {
      "epoch": 31.01,
      "grad_norm": 2.709390163421631,
      "learning_rate": 4.547531789056993e-06,
      "loss": 0.5845,
      "step": 75750
    },
    {
      "epoch": 31.01,
      "grad_norm": 3.316249132156372,
      "learning_rate": 4.54747778629444e-06,
      "loss": 0.5706,
      "step": 75760
    },
    {
      "epoch": 31.02,
      "grad_norm": 2.4369077682495117,
      "learning_rate": 4.5474237734369526e-06,
      "loss": 0.5769,
      "step": 75770
    },
    {
      "epoch": 31.02,
      "grad_norm": 2.8332245349884033,
      "learning_rate": 4.547369750484777e-06,
      "loss": 0.5856,
      "step": 75780
    },
    {
      "epoch": 31.02,
      "grad_norm": 2.260035514831543,
      "learning_rate": 4.547315717438161e-06,
      "loss": 0.5635,
      "step": 75790
    },
    {
      "epoch": 31.03,
      "grad_norm": 2.0845789909362793,
      "learning_rate": 4.54726167429735e-06,
      "loss": 0.5621,
      "step": 75800
    },
    {
      "epoch": 31.03,
      "grad_norm": 2.108210802078247,
      "learning_rate": 4.547207621062596e-06,
      "loss": 0.589,
      "step": 75810
    },
    {
      "epoch": 31.04,
      "grad_norm": 2.436055898666382,
      "learning_rate": 4.5471535577341415e-06,
      "loss": 0.5924,
      "step": 75820
    },
    {
      "epoch": 31.04,
      "grad_norm": 2.659986734390259,
      "learning_rate": 4.547099484312238e-06,
      "loss": 0.5767,
      "step": 75830
    },
    {
      "epoch": 31.04,
      "grad_norm": 2.283848524093628,
      "learning_rate": 4.54704540079713e-06,
      "loss": 0.572,
      "step": 75840
    },
    {
      "epoch": 31.05,
      "grad_norm": 2.271489143371582,
      "learning_rate": 4.5469913071890685e-06,
      "loss": 0.5587,
      "step": 75850
    },
    {
      "epoch": 31.05,
      "grad_norm": 3.0745134353637695,
      "learning_rate": 4.546937203488299e-06,
      "loss": 0.588,
      "step": 75860
    },
    {
      "epoch": 31.06,
      "grad_norm": 2.224890947341919,
      "learning_rate": 4.5468830896950706e-06,
      "loss": 0.5738,
      "step": 75870
    },
    {
      "epoch": 31.06,
      "grad_norm": 2.6272037029266357,
      "learning_rate": 4.546828965809631e-06,
      "loss": 0.5915,
      "step": 75880
    },
    {
      "epoch": 31.06,
      "grad_norm": 2.4687235355377197,
      "learning_rate": 4.546774831832228e-06,
      "loss": 0.577,
      "step": 75890
    },
    {
      "epoch": 31.07,
      "grad_norm": 2.634247064590454,
      "learning_rate": 4.546720687763108e-06,
      "loss": 0.5704,
      "step": 75900
    },
    {
      "epoch": 31.07,
      "grad_norm": 2.4815638065338135,
      "learning_rate": 4.546666533602522e-06,
      "loss": 0.5954,
      "step": 75910
    },
    {
      "epoch": 31.08,
      "grad_norm": 1.8934990167617798,
      "learning_rate": 4.546612369350716e-06,
      "loss": 0.5732,
      "step": 75920
    },
    {
      "epoch": 31.08,
      "grad_norm": 2.5747933387756348,
      "learning_rate": 4.5465581950079374e-06,
      "loss": 0.5693,
      "step": 75930
    },
    {
      "epoch": 31.08,
      "grad_norm": 2.49442982673645,
      "learning_rate": 4.546504010574437e-06,
      "loss": 0.5623,
      "step": 75940
    },
    {
      "epoch": 31.09,
      "grad_norm": 2.5276758670806885,
      "learning_rate": 4.546449816050462e-06,
      "loss": 0.5692,
      "step": 75950
    },
    {
      "epoch": 31.09,
      "grad_norm": 3.4642865657806396,
      "learning_rate": 4.546395611436259e-06,
      "loss": 0.5731,
      "step": 75960
    },
    {
      "epoch": 31.1,
      "grad_norm": 2.771402597427368,
      "learning_rate": 4.5463413967320785e-06,
      "loss": 0.5672,
      "step": 75970
    },
    {
      "epoch": 31.1,
      "grad_norm": 2.3732752799987793,
      "learning_rate": 4.546287171938168e-06,
      "loss": 0.5792,
      "step": 75980
    },
    {
      "epoch": 31.11,
      "grad_norm": 2.402432680130005,
      "learning_rate": 4.546232937054775e-06,
      "loss": 0.5577,
      "step": 75990
    },
    {
      "epoch": 31.11,
      "grad_norm": 2.6702330112457275,
      "learning_rate": 4.5461786920821495e-06,
      "loss": 0.5658,
      "step": 76000
    },
    {
      "epoch": 31.11,
      "grad_norm": 1.814618468284607,
      "learning_rate": 4.546124437020539e-06,
      "loss": 0.5635,
      "step": 76010
    },
    {
      "epoch": 31.12,
      "grad_norm": 3.2482070922851562,
      "learning_rate": 4.5460701718701916e-06,
      "loss": 0.5671,
      "step": 76020
    },
    {
      "epoch": 31.12,
      "grad_norm": 2.435990571975708,
      "learning_rate": 4.5460158966313574e-06,
      "loss": 0.5649,
      "step": 76030
    },
    {
      "epoch": 31.13,
      "grad_norm": 2.0929975509643555,
      "learning_rate": 4.545961611304284e-06,
      "loss": 0.5782,
      "step": 76040
    },
    {
      "epoch": 31.13,
      "grad_norm": 1.9548481702804565,
      "learning_rate": 4.545907315889221e-06,
      "loss": 0.587,
      "step": 76050
    },
    {
      "epoch": 31.13,
      "grad_norm": 1.9045339822769165,
      "learning_rate": 4.545853010386415e-06,
      "loss": 0.5981,
      "step": 76060
    },
    {
      "epoch": 31.14,
      "grad_norm": 2.7935357093811035,
      "learning_rate": 4.545798694796116e-06,
      "loss": 0.5763,
      "step": 76070
    },
    {
      "epoch": 31.14,
      "grad_norm": 2.3899617195129395,
      "learning_rate": 4.545744369118574e-06,
      "loss": 0.5827,
      "step": 76080
    },
    {
      "epoch": 31.15,
      "grad_norm": 2.6186559200286865,
      "learning_rate": 4.545690033354037e-06,
      "loss": 0.5517,
      "step": 76090
    },
    {
      "epoch": 31.15,
      "grad_norm": 2.2882862091064453,
      "learning_rate": 4.5456356875027536e-06,
      "loss": 0.5622,
      "step": 76100
    },
    {
      "epoch": 31.15,
      "grad_norm": 2.0619802474975586,
      "learning_rate": 4.545581331564972e-06,
      "loss": 0.583,
      "step": 76110
    },
    {
      "epoch": 31.16,
      "grad_norm": 1.8320918083190918,
      "learning_rate": 4.545526965540943e-06,
      "loss": 0.5492,
      "step": 76120
    },
    {
      "epoch": 31.16,
      "grad_norm": 2.6791443824768066,
      "learning_rate": 4.545472589430915e-06,
      "loss": 0.5939,
      "step": 76130
    },
    {
      "epoch": 31.17,
      "grad_norm": 2.4422121047973633,
      "learning_rate": 4.545418203235135e-06,
      "loss": 0.5625,
      "step": 76140
    },
    {
      "epoch": 31.17,
      "grad_norm": 3.057253837585449,
      "learning_rate": 4.545363806953856e-06,
      "loss": 0.5745,
      "step": 76150
    },
    {
      "epoch": 31.17,
      "grad_norm": 2.272834062576294,
      "learning_rate": 4.545309400587324e-06,
      "loss": 0.5851,
      "step": 76160
    },
    {
      "epoch": 31.18,
      "grad_norm": 2.4692983627319336,
      "learning_rate": 4.545254984135789e-06,
      "loss": 0.5697,
      "step": 76170
    },
    {
      "epoch": 31.18,
      "grad_norm": 3.209567070007324,
      "learning_rate": 4.5452005575995015e-06,
      "loss": 0.5819,
      "step": 76180
    },
    {
      "epoch": 31.19,
      "grad_norm": 2.1351287364959717,
      "learning_rate": 4.545146120978709e-06,
      "loss": 0.5709,
      "step": 76190
    },
    {
      "epoch": 31.19,
      "grad_norm": 2.935373067855835,
      "learning_rate": 4.545091674273663e-06,
      "loss": 0.5797,
      "step": 76200
    },
    {
      "epoch": 31.2,
      "grad_norm": 1.892096996307373,
      "learning_rate": 4.545037217484611e-06,
      "loss": 0.5604,
      "step": 76210
    },
    {
      "epoch": 31.2,
      "grad_norm": 1.6477441787719727,
      "learning_rate": 4.544982750611803e-06,
      "loss": 0.5597,
      "step": 76220
    },
    {
      "epoch": 31.2,
      "grad_norm": 2.8919622898101807,
      "learning_rate": 4.544928273655489e-06,
      "loss": 0.5657,
      "step": 76230
    },
    {
      "epoch": 31.21,
      "grad_norm": 3.011021375656128,
      "learning_rate": 4.544873786615918e-06,
      "loss": 0.5721,
      "step": 76240
    },
    {
      "epoch": 31.21,
      "grad_norm": 3.221341848373413,
      "learning_rate": 4.54481928949334e-06,
      "loss": 0.5794,
      "step": 76250
    },
    {
      "epoch": 31.22,
      "grad_norm": 2.129549741744995,
      "learning_rate": 4.544764782288004e-06,
      "loss": 0.5552,
      "step": 76260
    },
    {
      "epoch": 31.22,
      "grad_norm": 1.8665895462036133,
      "learning_rate": 4.544710265000161e-06,
      "loss": 0.5881,
      "step": 76270
    },
    {
      "epoch": 31.22,
      "grad_norm": 1.737165927886963,
      "learning_rate": 4.544655737630059e-06,
      "loss": 0.5643,
      "step": 76280
    },
    {
      "epoch": 31.23,
      "grad_norm": 1.9905686378479004,
      "learning_rate": 4.5446012001779495e-06,
      "loss": 0.5699,
      "step": 76290
    },
    {
      "epoch": 31.23,
      "grad_norm": 2.115753173828125,
      "learning_rate": 4.544546652644081e-06,
      "loss": 0.5701,
      "step": 76300
    },
    {
      "epoch": 31.24,
      "grad_norm": 2.339611530303955,
      "learning_rate": 4.544492095028704e-06,
      "loss": 0.5804,
      "step": 76310
    },
    {
      "epoch": 31.24,
      "grad_norm": 2.3794381618499756,
      "learning_rate": 4.544437527332069e-06,
      "loss": 0.5729,
      "step": 76320
    },
    {
      "epoch": 31.24,
      "grad_norm": 3.2007150650024414,
      "learning_rate": 4.544382949554424e-06,
      "loss": 0.5772,
      "step": 76330
    },
    {
      "epoch": 31.25,
      "grad_norm": 2.2157773971557617,
      "learning_rate": 4.54432836169602e-06,
      "loss": 0.5702,
      "step": 76340
    },
    {
      "epoch": 31.25,
      "grad_norm": 2.389065742492676,
      "learning_rate": 4.544273763757109e-06,
      "loss": 0.5943,
      "step": 76350
    },
    {
      "epoch": 31.26,
      "grad_norm": 2.6629583835601807,
      "learning_rate": 4.544219155737939e-06,
      "loss": 0.5693,
      "step": 76360
    },
    {
      "epoch": 31.26,
      "grad_norm": 2.2606441974639893,
      "learning_rate": 4.54416453763876e-06,
      "loss": 0.5746,
      "step": 76370
    },
    {
      "epoch": 31.26,
      "grad_norm": 3.0230088233947754,
      "learning_rate": 4.5441099094598225e-06,
      "loss": 0.5938,
      "step": 76380
    },
    {
      "epoch": 31.27,
      "grad_norm": 2.3410427570343018,
      "learning_rate": 4.544055271201377e-06,
      "loss": 0.5604,
      "step": 76390
    },
    {
      "epoch": 31.27,
      "grad_norm": 2.1754820346832275,
      "learning_rate": 4.5440006228636746e-06,
      "loss": 0.5693,
      "step": 76400
    },
    {
      "epoch": 31.28,
      "grad_norm": 2.2407283782958984,
      "learning_rate": 4.543945964446965e-06,
      "loss": 0.5806,
      "step": 76410
    },
    {
      "epoch": 31.28,
      "grad_norm": 2.1248013973236084,
      "learning_rate": 4.543891295951498e-06,
      "loss": 0.5861,
      "step": 76420
    },
    {
      "epoch": 31.29,
      "grad_norm": 2.1569368839263916,
      "learning_rate": 4.543836617377524e-06,
      "loss": 0.5529,
      "step": 76430
    },
    {
      "epoch": 31.29,
      "grad_norm": 2.0581533908843994,
      "learning_rate": 4.543781928725295e-06,
      "loss": 0.5897,
      "step": 76440
    },
    {
      "epoch": 31.29,
      "grad_norm": 2.505950450897217,
      "learning_rate": 4.543727229995059e-06,
      "loss": 0.5692,
      "step": 76450
    },
    {
      "epoch": 31.3,
      "grad_norm": 2.6573567390441895,
      "learning_rate": 4.543672521187069e-06,
      "loss": 0.5823,
      "step": 76460
    },
    {
      "epoch": 31.3,
      "grad_norm": 2.340207576751709,
      "learning_rate": 4.543617802301574e-06,
      "loss": 0.5683,
      "step": 76470
    },
    {
      "epoch": 31.31,
      "grad_norm": 2.487767219543457,
      "learning_rate": 4.543563073338826e-06,
      "loss": 0.548,
      "step": 76480
    },
    {
      "epoch": 31.31,
      "grad_norm": 3.497995615005493,
      "learning_rate": 4.543508334299075e-06,
      "loss": 0.5761,
      "step": 76490
    },
    {
      "epoch": 31.31,
      "grad_norm": 2.3676018714904785,
      "learning_rate": 4.5434535851825715e-06,
      "loss": 0.5805,
      "step": 76500
    },
    {
      "epoch": 31.32,
      "grad_norm": 2.199384927749634,
      "learning_rate": 4.543398825989566e-06,
      "loss": 0.5581,
      "step": 76510
    },
    {
      "epoch": 31.32,
      "grad_norm": 3.1854658126831055,
      "learning_rate": 4.543344056720311e-06,
      "loss": 0.5725,
      "step": 76520
    },
    {
      "epoch": 31.33,
      "grad_norm": 2.3231208324432373,
      "learning_rate": 4.543289277375055e-06,
      "loss": 0.571,
      "step": 76530
    },
    {
      "epoch": 31.33,
      "grad_norm": 2.425029754638672,
      "learning_rate": 4.5432344879540516e-06,
      "loss": 0.5825,
      "step": 76540
    },
    {
      "epoch": 31.33,
      "grad_norm": 2.300889253616333,
      "learning_rate": 4.543179688457549e-06,
      "loss": 0.56,
      "step": 76550
    },
    {
      "epoch": 31.34,
      "grad_norm": 1.9588024616241455,
      "learning_rate": 4.543124878885801e-06,
      "loss": 0.5745,
      "step": 76560
    },
    {
      "epoch": 31.34,
      "grad_norm": 2.317955493927002,
      "learning_rate": 4.543070059239056e-06,
      "loss": 0.596,
      "step": 76570
    },
    {
      "epoch": 31.35,
      "grad_norm": 1.5829962491989136,
      "learning_rate": 4.543015229517567e-06,
      "loss": 0.5852,
      "step": 76580
    },
    {
      "epoch": 31.35,
      "grad_norm": 2.658400297164917,
      "learning_rate": 4.542960389721585e-06,
      "loss": 0.5766,
      "step": 76590
    },
    {
      "epoch": 31.35,
      "grad_norm": 3.1956632137298584,
      "learning_rate": 4.5429055398513605e-06,
      "loss": 0.5645,
      "step": 76600
    },
    {
      "epoch": 31.36,
      "grad_norm": 2.634322166442871,
      "learning_rate": 4.542850679907145e-06,
      "loss": 0.5412,
      "step": 76610
    },
    {
      "epoch": 31.36,
      "grad_norm": 2.2597625255584717,
      "learning_rate": 4.54279580988919e-06,
      "loss": 0.5835,
      "step": 76620
    },
    {
      "epoch": 31.37,
      "grad_norm": 2.4257848262786865,
      "learning_rate": 4.542740929797746e-06,
      "loss": 0.5901,
      "step": 76630
    },
    {
      "epoch": 31.37,
      "grad_norm": 2.0716235637664795,
      "learning_rate": 4.542686039633065e-06,
      "loss": 0.562,
      "step": 76640
    },
    {
      "epoch": 31.38,
      "grad_norm": 2.504542827606201,
      "learning_rate": 4.5426311393954e-06,
      "loss": 0.5801,
      "step": 76650
    },
    {
      "epoch": 31.38,
      "grad_norm": 2.2724902629852295,
      "learning_rate": 4.542576229085e-06,
      "loss": 0.5903,
      "step": 76660
    },
    {
      "epoch": 31.38,
      "grad_norm": 2.646052837371826,
      "learning_rate": 4.542521308702118e-06,
      "loss": 0.5771,
      "step": 76670
    },
    {
      "epoch": 31.39,
      "grad_norm": 1.7721720933914185,
      "learning_rate": 4.542466378247005e-06,
      "loss": 0.57,
      "step": 76680
    },
    {
      "epoch": 31.39,
      "grad_norm": 2.6154870986938477,
      "learning_rate": 4.5424114377199126e-06,
      "loss": 0.5782,
      "step": 76690
    },
    {
      "epoch": 31.4,
      "grad_norm": 2.4898324012756348,
      "learning_rate": 4.542356487121093e-06,
      "loss": 0.5634,
      "step": 76700
    },
    {
      "epoch": 31.4,
      "grad_norm": 2.3260135650634766,
      "learning_rate": 4.542301526450797e-06,
      "loss": 0.5796,
      "step": 76710
    },
    {
      "epoch": 31.4,
      "grad_norm": 2.495974063873291,
      "learning_rate": 4.542246555709277e-06,
      "loss": 0.588,
      "step": 76720
    },
    {
      "epoch": 31.41,
      "grad_norm": 2.0790321826934814,
      "learning_rate": 4.542191574896786e-06,
      "loss": 0.5867,
      "step": 76730
    },
    {
      "epoch": 31.41,
      "grad_norm": 2.4879274368286133,
      "learning_rate": 4.542136584013573e-06,
      "loss": 0.5594,
      "step": 76740
    },
    {
      "epoch": 31.42,
      "grad_norm": 2.439357042312622,
      "learning_rate": 4.542081583059893e-06,
      "loss": 0.5681,
      "step": 76750
    },
    {
      "epoch": 31.42,
      "grad_norm": 1.8033827543258667,
      "learning_rate": 4.542026572035996e-06,
      "loss": 0.5896,
      "step": 76760
    },
    {
      "epoch": 31.42,
      "grad_norm": 2.062725782394409,
      "learning_rate": 4.541971550942133e-06,
      "loss": 0.5776,
      "step": 76770
    },
    {
      "epoch": 31.43,
      "grad_norm": 2.0388619899749756,
      "learning_rate": 4.541916519778559e-06,
      "loss": 0.5804,
      "step": 76780
    },
    {
      "epoch": 31.43,
      "grad_norm": 1.9818555116653442,
      "learning_rate": 4.541861478545524e-06,
      "loss": 0.563,
      "step": 76790
    },
    {
      "epoch": 31.44,
      "grad_norm": 1.7316927909851074,
      "learning_rate": 4.541806427243281e-06,
      "loss": 0.5855,
      "step": 76800
    },
    {
      "epoch": 31.44,
      "grad_norm": 2.4759738445281982,
      "learning_rate": 4.541751365872082e-06,
      "loss": 0.5596,
      "step": 76810
    },
    {
      "epoch": 31.44,
      "grad_norm": 2.730722665786743,
      "learning_rate": 4.5416962944321795e-06,
      "loss": 0.5851,
      "step": 76820
    },
    {
      "epoch": 31.45,
      "grad_norm": 1.8637498617172241,
      "learning_rate": 4.541641212923825e-06,
      "loss": 0.5718,
      "step": 76830
    },
    {
      "epoch": 31.45,
      "grad_norm": 2.212344169616699,
      "learning_rate": 4.541586121347271e-06,
      "loss": 0.5742,
      "step": 76840
    },
    {
      "epoch": 31.46,
      "grad_norm": 2.1511731147766113,
      "learning_rate": 4.541531019702771e-06,
      "loss": 0.5968,
      "step": 76850
    },
    {
      "epoch": 31.46,
      "grad_norm": 2.1522369384765625,
      "learning_rate": 4.541475907990575e-06,
      "loss": 0.5802,
      "step": 76860
    },
    {
      "epoch": 31.47,
      "grad_norm": 2.7235748767852783,
      "learning_rate": 4.541420786210938e-06,
      "loss": 0.5697,
      "step": 76870
    },
    {
      "epoch": 31.47,
      "grad_norm": 1.8658909797668457,
      "learning_rate": 4.541365654364112e-06,
      "loss": 0.5936,
      "step": 76880
    },
    {
      "epoch": 31.47,
      "grad_norm": 2.3234546184539795,
      "learning_rate": 4.541310512450347e-06,
      "loss": 0.593,
      "step": 76890
    },
    {
      "epoch": 31.48,
      "grad_norm": 2.2663674354553223,
      "learning_rate": 4.5412553604699e-06,
      "loss": 0.5718,
      "step": 76900
    },
    {
      "epoch": 31.48,
      "grad_norm": 2.6641218662261963,
      "learning_rate": 4.54120019842302e-06,
      "loss": 0.5522,
      "step": 76910
    },
    {
      "epoch": 31.49,
      "grad_norm": 2.948761463165283,
      "learning_rate": 4.541145026309961e-06,
      "loss": 0.5706,
      "step": 76920
    },
    {
      "epoch": 31.49,
      "grad_norm": 2.408862590789795,
      "learning_rate": 4.541089844130976e-06,
      "loss": 0.5885,
      "step": 76930
    },
    {
      "epoch": 31.49,
      "grad_norm": 1.7309269905090332,
      "learning_rate": 4.541034651886317e-06,
      "loss": 0.5784,
      "step": 76940
    },
    {
      "epoch": 31.5,
      "grad_norm": 2.815073013305664,
      "learning_rate": 4.540979449576238e-06,
      "loss": 0.5873,
      "step": 76950
    },
    {
      "epoch": 31.5,
      "grad_norm": 1.561689853668213,
      "learning_rate": 4.540924237200991e-06,
      "loss": 0.5584,
      "step": 76960
    },
    {
      "epoch": 31.51,
      "grad_norm": 2.097637176513672,
      "learning_rate": 4.540869014760829e-06,
      "loss": 0.6068,
      "step": 76970
    },
    {
      "epoch": 31.51,
      "grad_norm": 2.1061923503875732,
      "learning_rate": 4.540813782256005e-06,
      "loss": 0.573,
      "step": 76980
    },
    {
      "epoch": 31.51,
      "grad_norm": 1.9503662586212158,
      "learning_rate": 4.540758539686772e-06,
      "loss": 0.5564,
      "step": 76990
    },
    {
      "epoch": 31.52,
      "grad_norm": 2.115018844604492,
      "learning_rate": 4.540703287053383e-06,
      "loss": 0.5672,
      "step": 77000
    },
    {
      "epoch": 31.52,
      "grad_norm": 2.246546745300293,
      "learning_rate": 4.540648024356092e-06,
      "loss": 0.5745,
      "step": 77010
    },
    {
      "epoch": 31.53,
      "grad_norm": 2.1644041538238525,
      "learning_rate": 4.540592751595151e-06,
      "loss": 0.5711,
      "step": 77020
    },
    {
      "epoch": 31.53,
      "grad_norm": 2.4040234088897705,
      "learning_rate": 4.540537468770813e-06,
      "loss": 0.5844,
      "step": 77030
    },
    {
      "epoch": 31.53,
      "grad_norm": 2.24261736869812,
      "learning_rate": 4.540482175883333e-06,
      "loss": 0.5739,
      "step": 77040
    },
    {
      "epoch": 31.54,
      "grad_norm": 2.625284194946289,
      "learning_rate": 4.540426872932961e-06,
      "loss": 0.5634,
      "step": 77050
    },
    {
      "epoch": 31.54,
      "grad_norm": 1.889898419380188,
      "learning_rate": 4.5403715599199545e-06,
      "loss": 0.5813,
      "step": 77060
    },
    {
      "epoch": 31.55,
      "grad_norm": 2.4729955196380615,
      "learning_rate": 4.540316236844563e-06,
      "loss": 0.5538,
      "step": 77070
    },
    {
      "epoch": 31.55,
      "grad_norm": 2.0000569820404053,
      "learning_rate": 4.5402609037070435e-06,
      "loss": 0.5665,
      "step": 77080
    },
    {
      "epoch": 31.56,
      "grad_norm": 1.986828088760376,
      "learning_rate": 4.540205560507647e-06,
      "loss": 0.5787,
      "step": 77090
    },
    {
      "epoch": 31.56,
      "grad_norm": 2.338529109954834,
      "learning_rate": 4.540150207246628e-06,
      "loss": 0.5558,
      "step": 77100
    },
    {
      "epoch": 31.56,
      "grad_norm": 2.189509153366089,
      "learning_rate": 4.540094843924239e-06,
      "loss": 0.5724,
      "step": 77110
    },
    {
      "epoch": 31.57,
      "grad_norm": 3.6627399921417236,
      "learning_rate": 4.540039470540736e-06,
      "loss": 0.5809,
      "step": 77120
    },
    {
      "epoch": 31.57,
      "grad_norm": 2.572185516357422,
      "learning_rate": 4.53998408709637e-06,
      "loss": 0.5883,
      "step": 77130
    },
    {
      "epoch": 31.58,
      "grad_norm": 2.2565393447875977,
      "learning_rate": 4.539928693591395e-06,
      "loss": 0.5868,
      "step": 77140
    },
    {
      "epoch": 31.58,
      "grad_norm": 2.5019466876983643,
      "learning_rate": 4.539873290026067e-06,
      "loss": 0.5664,
      "step": 77150
    },
    {
      "epoch": 31.58,
      "grad_norm": 2.5421979427337646,
      "learning_rate": 4.5398178764006375e-06,
      "loss": 0.5917,
      "step": 77160
    },
    {
      "epoch": 31.59,
      "grad_norm": 2.163879156112671,
      "learning_rate": 4.5397624527153616e-06,
      "loss": 0.562,
      "step": 77170
    },
    {
      "epoch": 31.59,
      "grad_norm": 2.17181134223938,
      "learning_rate": 4.539707018970492e-06,
      "loss": 0.5833,
      "step": 77180
    },
    {
      "epoch": 31.6,
      "grad_norm": 2.661303758621216,
      "learning_rate": 4.5396515751662846e-06,
      "loss": 0.5601,
      "step": 77190
    },
    {
      "epoch": 31.6,
      "grad_norm": 2.8073840141296387,
      "learning_rate": 4.539596121302991e-06,
      "loss": 0.5624,
      "step": 77200
    },
    {
      "epoch": 31.6,
      "grad_norm": 3.045600414276123,
      "learning_rate": 4.539540657380867e-06,
      "loss": 0.5951,
      "step": 77210
    },
    {
      "epoch": 31.61,
      "grad_norm": 2.8773744106292725,
      "learning_rate": 4.539485183400166e-06,
      "loss": 0.5892,
      "step": 77220
    },
    {
      "epoch": 31.61,
      "grad_norm": 2.2719907760620117,
      "learning_rate": 4.539429699361142e-06,
      "loss": 0.5801,
      "step": 77230
    },
    {
      "epoch": 31.62,
      "grad_norm": 2.3094987869262695,
      "learning_rate": 4.539374205264049e-06,
      "loss": 0.5929,
      "step": 77240
    },
    {
      "epoch": 31.62,
      "grad_norm": 2.019087791442871,
      "learning_rate": 4.539318701109143e-06,
      "loss": 0.5969,
      "step": 77250
    },
    {
      "epoch": 31.63,
      "grad_norm": 2.4786269664764404,
      "learning_rate": 4.539263186896676e-06,
      "loss": 0.5632,
      "step": 77260
    },
    {
      "epoch": 31.63,
      "grad_norm": 2.90187668800354,
      "learning_rate": 4.539207662626902e-06,
      "loss": 0.5703,
      "step": 77270
    },
    {
      "epoch": 31.63,
      "grad_norm": 2.290241241455078,
      "learning_rate": 4.539152128300078e-06,
      "loss": 0.5561,
      "step": 77280
    },
    {
      "epoch": 31.64,
      "grad_norm": 2.2418668270111084,
      "learning_rate": 4.539096583916456e-06,
      "loss": 0.5572,
      "step": 77290
    },
    {
      "epoch": 31.64,
      "grad_norm": 2.103022575378418,
      "learning_rate": 4.5390410294762914e-06,
      "loss": 0.5879,
      "step": 77300
    },
    {
      "epoch": 31.65,
      "grad_norm": 1.6368123292922974,
      "learning_rate": 4.538985464979838e-06,
      "loss": 0.5861,
      "step": 77310
    },
    {
      "epoch": 31.65,
      "grad_norm": 2.4374563694000244,
      "learning_rate": 4.538929890427352e-06,
      "loss": 0.5602,
      "step": 77320
    },
    {
      "epoch": 31.65,
      "grad_norm": 2.4645092487335205,
      "learning_rate": 4.538874305819087e-06,
      "loss": 0.5941,
      "step": 77330
    },
    {
      "epoch": 31.66,
      "grad_norm": 2.6349050998687744,
      "learning_rate": 4.538818711155297e-06,
      "loss": 0.5491,
      "step": 77340
    },
    {
      "epoch": 31.66,
      "grad_norm": 1.8972736597061157,
      "learning_rate": 4.5387631064362375e-06,
      "loss": 0.5512,
      "step": 77350
    },
    {
      "epoch": 31.67,
      "grad_norm": 2.6865344047546387,
      "learning_rate": 4.5387074916621625e-06,
      "loss": 0.5895,
      "step": 77360
    },
    {
      "epoch": 31.67,
      "grad_norm": 2.317434549331665,
      "learning_rate": 4.538651866833327e-06,
      "loss": 0.58,
      "step": 77370
    },
    {
      "epoch": 31.67,
      "grad_norm": 2.133683919906616,
      "learning_rate": 4.538596231949987e-06,
      "loss": 0.5831,
      "step": 77380
    },
    {
      "epoch": 31.68,
      "grad_norm": 1.741848111152649,
      "learning_rate": 4.538540587012394e-06,
      "loss": 0.576,
      "step": 77390
    },
    {
      "epoch": 31.68,
      "grad_norm": 2.5498111248016357,
      "learning_rate": 4.538484932020807e-06,
      "loss": 0.5916,
      "step": 77400
    },
    {
      "epoch": 31.69,
      "grad_norm": 3.289304256439209,
      "learning_rate": 4.53842926697548e-06,
      "loss": 0.5654,
      "step": 77410
    },
    {
      "epoch": 31.69,
      "grad_norm": 2.5364677906036377,
      "learning_rate": 4.538373591876666e-06,
      "loss": 0.5791,
      "step": 77420
    },
    {
      "epoch": 31.69,
      "grad_norm": 2.430962085723877,
      "learning_rate": 4.53831790672462e-06,
      "loss": 0.5709,
      "step": 77430
    },
    {
      "epoch": 31.7,
      "grad_norm": 2.0554065704345703,
      "learning_rate": 4.538262211519601e-06,
      "loss": 0.5947,
      "step": 77440
    },
    {
      "epoch": 31.7,
      "grad_norm": 1.8155182600021362,
      "learning_rate": 4.53820650626186e-06,
      "loss": 0.5829,
      "step": 77450
    },
    {
      "epoch": 31.71,
      "grad_norm": 2.2525899410247803,
      "learning_rate": 4.538150790951653e-06,
      "loss": 0.5479,
      "step": 77460
    },
    {
      "epoch": 31.71,
      "grad_norm": 2.9169981479644775,
      "learning_rate": 4.538095065589237e-06,
      "loss": 0.5614,
      "step": 77470
    },
    {
      "epoch": 31.72,
      "grad_norm": 2.409651041030884,
      "learning_rate": 4.5380393301748655e-06,
      "loss": 0.5694,
      "step": 77480
    },
    {
      "epoch": 31.72,
      "grad_norm": 2.37583327293396,
      "learning_rate": 4.537983584708794e-06,
      "loss": 0.5572,
      "step": 77490
    },
    {
      "epoch": 31.72,
      "grad_norm": 2.3066699504852295,
      "learning_rate": 4.537927829191279e-06,
      "loss": 0.5669,
      "step": 77500
    },
    {
      "epoch": 31.73,
      "grad_norm": 2.4225502014160156,
      "learning_rate": 4.5378720636225745e-06,
      "loss": 0.5771,
      "step": 77510
    },
    {
      "epoch": 31.73,
      "grad_norm": 2.0879576206207275,
      "learning_rate": 4.537816288002937e-06,
      "loss": 0.5829,
      "step": 77520
    },
    {
      "epoch": 31.74,
      "grad_norm": 2.505873918533325,
      "learning_rate": 4.537760502332622e-06,
      "loss": 0.5744,
      "step": 77530
    },
    {
      "epoch": 31.74,
      "grad_norm": 2.55961537361145,
      "learning_rate": 4.5377047066118844e-06,
      "loss": 0.5712,
      "step": 77540
    },
    {
      "epoch": 31.74,
      "grad_norm": 2.8579204082489014,
      "learning_rate": 4.5376489008409795e-06,
      "loss": 0.5792,
      "step": 77550
    },
    {
      "epoch": 31.75,
      "grad_norm": 1.9515380859375,
      "learning_rate": 4.537593085020164e-06,
      "loss": 0.5787,
      "step": 77560
    },
    {
      "epoch": 31.75,
      "grad_norm": 2.035230875015259,
      "learning_rate": 4.5375372591496935e-06,
      "loss": 0.5645,
      "step": 77570
    },
    {
      "epoch": 31.76,
      "grad_norm": 2.356057643890381,
      "learning_rate": 4.537481423229823e-06,
      "loss": 0.5752,
      "step": 77580
    },
    {
      "epoch": 31.76,
      "grad_norm": 2.47991681098938,
      "learning_rate": 4.537425577260808e-06,
      "loss": 0.5657,
      "step": 77590
    },
    {
      "epoch": 31.76,
      "grad_norm": 1.507152795791626,
      "learning_rate": 4.537369721242906e-06,
      "loss": 0.5662,
      "step": 77600
    },
    {
      "epoch": 31.77,
      "grad_norm": 2.485774517059326,
      "learning_rate": 4.5373138551763714e-06,
      "loss": 0.5611,
      "step": 77610
    },
    {
      "epoch": 31.77,
      "grad_norm": 2.0253264904022217,
      "learning_rate": 4.53725797906146e-06,
      "loss": 0.5482,
      "step": 77620
    },
    {
      "epoch": 31.78,
      "grad_norm": 2.5656135082244873,
      "learning_rate": 4.537202092898429e-06,
      "loss": 0.5709,
      "step": 77630
    },
    {
      "epoch": 31.78,
      "grad_norm": 2.908400058746338,
      "learning_rate": 4.537146196687532e-06,
      "loss": 0.5736,
      "step": 77640
    },
    {
      "epoch": 31.78,
      "grad_norm": 2.3138222694396973,
      "learning_rate": 4.537090290429028e-06,
      "loss": 0.5769,
      "step": 77650
    },
    {
      "epoch": 31.79,
      "grad_norm": 2.281005620956421,
      "learning_rate": 4.5370343741231725e-06,
      "loss": 0.5644,
      "step": 77660
    },
    {
      "epoch": 31.79,
      "grad_norm": 1.9247472286224365,
      "learning_rate": 4.53697844777022e-06,
      "loss": 0.5765,
      "step": 77670
    },
    {
      "epoch": 31.8,
      "grad_norm": 1.9253733158111572,
      "learning_rate": 4.536922511370428e-06,
      "loss": 0.597,
      "step": 77680
    },
    {
      "epoch": 31.8,
      "grad_norm": 2.437220811843872,
      "learning_rate": 4.536866564924053e-06,
      "loss": 0.5865,
      "step": 77690
    },
    {
      "epoch": 31.81,
      "grad_norm": 2.3225252628326416,
      "learning_rate": 4.536810608431349e-06,
      "loss": 0.5595,
      "step": 77700
    },
    {
      "epoch": 31.81,
      "grad_norm": 2.551130533218384,
      "learning_rate": 4.536754641892575e-06,
      "loss": 0.5713,
      "step": 77710
    },
    {
      "epoch": 31.81,
      "grad_norm": 2.553041458129883,
      "learning_rate": 4.536698665307986e-06,
      "loss": 0.5602,
      "step": 77720
    },
    {
      "epoch": 31.82,
      "grad_norm": 2.9090471267700195,
      "learning_rate": 4.53664267867784e-06,
      "loss": 0.5666,
      "step": 77730
    },
    {
      "epoch": 31.82,
      "grad_norm": 3.5417416095733643,
      "learning_rate": 4.536586682002391e-06,
      "loss": 0.5484,
      "step": 77740
    },
    {
      "epoch": 31.83,
      "grad_norm": 2.4691998958587646,
      "learning_rate": 4.536530675281896e-06,
      "loss": 0.5665,
      "step": 77750
    },
    {
      "epoch": 31.83,
      "grad_norm": 2.2983193397521973,
      "learning_rate": 4.536474658516614e-06,
      "loss": 0.5737,
      "step": 77760
    },
    {
      "epoch": 31.83,
      "grad_norm": 2.4776713848114014,
      "learning_rate": 4.536418631706799e-06,
      "loss": 0.5799,
      "step": 77770
    },
    {
      "epoch": 31.84,
      "grad_norm": 1.8546667098999023,
      "learning_rate": 4.536362594852709e-06,
      "loss": 0.5666,
      "step": 77780
    },
    {
      "epoch": 31.84,
      "grad_norm": 2.3963258266448975,
      "learning_rate": 4.5363065479546e-06,
      "loss": 0.579,
      "step": 77790
    },
    {
      "epoch": 31.85,
      "grad_norm": 3.2379374504089355,
      "learning_rate": 4.536250491012729e-06,
      "loss": 0.5719,
      "step": 77800
    },
    {
      "epoch": 31.85,
      "grad_norm": 2.9189014434814453,
      "learning_rate": 4.536194424027352e-06,
      "loss": 0.5795,
      "step": 77810
    },
    {
      "epoch": 31.85,
      "grad_norm": 2.4249961376190186,
      "learning_rate": 4.5361383469987285e-06,
      "loss": 0.5688,
      "step": 77820
    },
    {
      "epoch": 31.86,
      "grad_norm": 2.766535520553589,
      "learning_rate": 4.536082259927113e-06,
      "loss": 0.5654,
      "step": 77830
    },
    {
      "epoch": 31.86,
      "grad_norm": 2.33119535446167,
      "learning_rate": 4.536026162812762e-06,
      "loss": 0.5695,
      "step": 77840
    },
    {
      "epoch": 31.87,
      "grad_norm": 2.805884599685669,
      "learning_rate": 4.5359700556559344e-06,
      "loss": 0.5719,
      "step": 77850
    },
    {
      "epoch": 31.87,
      "grad_norm": 2.2704708576202393,
      "learning_rate": 4.535913938456885e-06,
      "loss": 0.5788,
      "step": 77860
    },
    {
      "epoch": 31.87,
      "grad_norm": 2.231809139251709,
      "learning_rate": 4.5358578112158736e-06,
      "loss": 0.5843,
      "step": 77870
    },
    {
      "epoch": 31.88,
      "grad_norm": 2.572713851928711,
      "learning_rate": 4.5358016739331545e-06,
      "loss": 0.5754,
      "step": 77880
    },
    {
      "epoch": 31.88,
      "grad_norm": 1.767374873161316,
      "learning_rate": 4.535745526608987e-06,
      "loss": 0.5715,
      "step": 77890
    },
    {
      "epoch": 31.89,
      "grad_norm": 2.0047314167022705,
      "learning_rate": 4.535689369243627e-06,
      "loss": 0.5794,
      "step": 77900
    },
    {
      "epoch": 31.89,
      "grad_norm": 2.2405343055725098,
      "learning_rate": 4.535633201837333e-06,
      "loss": 0.5528,
      "step": 77910
    },
    {
      "epoch": 31.9,
      "grad_norm": 1.814526915550232,
      "learning_rate": 4.53557702439036e-06,
      "loss": 0.5943,
      "step": 77920
    },
    {
      "epoch": 31.9,
      "grad_norm": 2.8693385124206543,
      "learning_rate": 4.535520836902968e-06,
      "loss": 0.5558,
      "step": 77930
    },
    {
      "epoch": 31.9,
      "grad_norm": 2.0985875129699707,
      "learning_rate": 4.535464639375414e-06,
      "loss": 0.5783,
      "step": 77940
    },
    {
      "epoch": 31.91,
      "grad_norm": 2.5139031410217285,
      "learning_rate": 4.535408431807953e-06,
      "loss": 0.5743,
      "step": 77950
    },
    {
      "epoch": 31.91,
      "grad_norm": 2.350576400756836,
      "learning_rate": 4.535352214200845e-06,
      "loss": 0.5728,
      "step": 77960
    },
    {
      "epoch": 31.92,
      "grad_norm": 1.8828614950180054,
      "learning_rate": 4.535295986554347e-06,
      "loss": 0.571,
      "step": 77970
    },
    {
      "epoch": 31.92,
      "grad_norm": 2.2771241664886475,
      "learning_rate": 4.535239748868715e-06,
      "loss": 0.5603,
      "step": 77980
    },
    {
      "epoch": 31.92,
      "grad_norm": 2.2770912647247314,
      "learning_rate": 4.535183501144209e-06,
      "loss": 0.5941,
      "step": 77990
    },
    {
      "epoch": 31.93,
      "grad_norm": 3.173064947128296,
      "learning_rate": 4.535127243381085e-06,
      "loss": 0.5587,
      "step": 78000
    },
    {
      "epoch": 31.93,
      "grad_norm": 2.0130958557128906,
      "learning_rate": 4.535070975579601e-06,
      "loss": 0.5843,
      "step": 78010
    },
    {
      "epoch": 31.94,
      "grad_norm": 2.3311166763305664,
      "learning_rate": 4.5350146977400155e-06,
      "loss": 0.5839,
      "step": 78020
    },
    {
      "epoch": 31.94,
      "grad_norm": 2.490753650665283,
      "learning_rate": 4.534958409862585e-06,
      "loss": 0.5887,
      "step": 78030
    },
    {
      "epoch": 31.94,
      "grad_norm": 2.552746057510376,
      "learning_rate": 4.534902111947569e-06,
      "loss": 0.5676,
      "step": 78040
    },
    {
      "epoch": 31.95,
      "grad_norm": 2.2610538005828857,
      "learning_rate": 4.534845803995224e-06,
      "loss": 0.5765,
      "step": 78050
    },
    {
      "epoch": 31.95,
      "grad_norm": 2.370037078857422,
      "learning_rate": 4.534789486005809e-06,
      "loss": 0.5639,
      "step": 78060
    },
    {
      "epoch": 31.96,
      "grad_norm": 2.174553632736206,
      "learning_rate": 4.534733157979582e-06,
      "loss": 0.5482,
      "step": 78070
    },
    {
      "epoch": 31.96,
      "grad_norm": 2.8781890869140625,
      "learning_rate": 4.5346768199167995e-06,
      "loss": 0.5763,
      "step": 78080
    },
    {
      "epoch": 31.96,
      "grad_norm": 2.667067050933838,
      "learning_rate": 4.5346204718177205e-06,
      "loss": 0.5621,
      "step": 78090
    },
    {
      "epoch": 31.97,
      "grad_norm": 1.8985546827316284,
      "learning_rate": 4.534564113682603e-06,
      "loss": 0.5981,
      "step": 78100
    },
    {
      "epoch": 31.97,
      "grad_norm": 2.8948700428009033,
      "learning_rate": 4.534507745511705e-06,
      "loss": 0.5873,
      "step": 78110
    },
    {
      "epoch": 31.98,
      "grad_norm": 2.1463510990142822,
      "learning_rate": 4.534451367305287e-06,
      "loss": 0.5637,
      "step": 78120
    },
    {
      "epoch": 31.98,
      "grad_norm": 2.5407488346099854,
      "learning_rate": 4.5343949790636035e-06,
      "loss": 0.577,
      "step": 78130
    },
    {
      "epoch": 31.99,
      "grad_norm": 2.200059652328491,
      "learning_rate": 4.534338580786915e-06,
      "loss": 0.5738,
      "step": 78140
    },
    {
      "epoch": 31.99,
      "grad_norm": 2.2656843662261963,
      "learning_rate": 4.5342821724754795e-06,
      "loss": 0.5603,
      "step": 78150
    },
    {
      "epoch": 31.99,
      "grad_norm": 3.078342914581299,
      "learning_rate": 4.534225754129555e-06,
      "loss": 0.5873,
      "step": 78160
    },
    {
      "epoch": 32.0,
      "grad_norm": 1.8605928421020508,
      "learning_rate": 4.5341693257494015e-06,
      "loss": 0.5478,
      "step": 78170
    },
    {
      "epoch": 32.0,
      "eval_loss": 0.5708456635475159,
      "eval_runtime": 51.8924,
      "eval_samples_per_second": 66.465,
      "eval_steps_per_second": 8.325,
      "step": 78176
    },
    {
      "epoch": 32.0,
      "grad_norm": 2.8978796005249023,
      "learning_rate": 4.534112887335276e-06,
      "loss": 0.5588,
      "step": 78180
    },
    {
      "epoch": 32.01,
      "grad_norm": 2.341273307800293,
      "learning_rate": 4.5340564388874365e-06,
      "loss": 0.5646,
      "step": 78190
    },
    {
      "epoch": 32.01,
      "grad_norm": 2.4099271297454834,
      "learning_rate": 4.533999980406143e-06,
      "loss": 0.5725,
      "step": 78200
    },
    {
      "epoch": 32.01,
      "grad_norm": 2.4224345684051514,
      "learning_rate": 4.533943511891654e-06,
      "loss": 0.5837,
      "step": 78210
    },
    {
      "epoch": 32.02,
      "grad_norm": 3.1175742149353027,
      "learning_rate": 4.533887033344226e-06,
      "loss": 0.5719,
      "step": 78220
    },
    {
      "epoch": 32.02,
      "grad_norm": 2.6168365478515625,
      "learning_rate": 4.533830544764121e-06,
      "loss": 0.5742,
      "step": 78230
    },
    {
      "epoch": 32.03,
      "grad_norm": 2.726379156112671,
      "learning_rate": 4.533774046151595e-06,
      "loss": 0.5694,
      "step": 78240
    },
    {
      "epoch": 32.03,
      "grad_norm": 2.440891981124878,
      "learning_rate": 4.533717537506909e-06,
      "loss": 0.5757,
      "step": 78250
    },
    {
      "epoch": 32.03,
      "grad_norm": 2.1942832469940186,
      "learning_rate": 4.53366101883032e-06,
      "loss": 0.5856,
      "step": 78260
    },
    {
      "epoch": 32.04,
      "grad_norm": 2.3110694885253906,
      "learning_rate": 4.533604490122089e-06,
      "loss": 0.5697,
      "step": 78270
    },
    {
      "epoch": 32.04,
      "grad_norm": 2.960789442062378,
      "learning_rate": 4.533547951382473e-06,
      "loss": 0.5523,
      "step": 78280
    },
    {
      "epoch": 32.05,
      "grad_norm": 2.167685031890869,
      "learning_rate": 4.533491402611732e-06,
      "loss": 0.5553,
      "step": 78290
    },
    {
      "epoch": 32.05,
      "grad_norm": 2.319051742553711,
      "learning_rate": 4.533434843810124e-06,
      "loss": 0.5755,
      "step": 78300
    },
    {
      "epoch": 32.05,
      "grad_norm": 1.9528942108154297,
      "learning_rate": 4.53337827497791e-06,
      "loss": 0.6026,
      "step": 78310
    },
    {
      "epoch": 32.06,
      "grad_norm": 2.4348461627960205,
      "learning_rate": 4.5333216961153475e-06,
      "loss": 0.5683,
      "step": 78320
    },
    {
      "epoch": 32.06,
      "grad_norm": 3.278635025024414,
      "learning_rate": 4.533265107222696e-06,
      "loss": 0.5767,
      "step": 78330
    },
    {
      "epoch": 32.07,
      "grad_norm": 2.9780285358428955,
      "learning_rate": 4.533208508300215e-06,
      "loss": 0.5633,
      "step": 78340
    },
    {
      "epoch": 32.07,
      "grad_norm": 2.075239419937134,
      "learning_rate": 4.533151899348164e-06,
      "loss": 0.5826,
      "step": 78350
    },
    {
      "epoch": 32.08,
      "grad_norm": 2.926633596420288,
      "learning_rate": 4.5330952803668005e-06,
      "loss": 0.5763,
      "step": 78360
    },
    {
      "epoch": 32.08,
      "grad_norm": 2.4455573558807373,
      "learning_rate": 4.533038651356387e-06,
      "loss": 0.6005,
      "step": 78370
    },
    {
      "epoch": 32.08,
      "grad_norm": 1.8441160917282104,
      "learning_rate": 4.532982012317181e-06,
      "loss": 0.5524,
      "step": 78380
    },
    {
      "epoch": 32.09,
      "grad_norm": 2.5217792987823486,
      "learning_rate": 4.532925363249441e-06,
      "loss": 0.5659,
      "step": 78390
    },
    {
      "epoch": 32.09,
      "grad_norm": 2.2588839530944824,
      "learning_rate": 4.532868704153429e-06,
      "loss": 0.5907,
      "step": 78400
    },
    {
      "epoch": 32.1,
      "grad_norm": 2.086413860321045,
      "learning_rate": 4.532812035029403e-06,
      "loss": 0.5717,
      "step": 78410
    },
    {
      "epoch": 32.1,
      "grad_norm": 2.826814651489258,
      "learning_rate": 4.532755355877622e-06,
      "loss": 0.5642,
      "step": 78420
    },
    {
      "epoch": 32.1,
      "grad_norm": 3.8566818237304688,
      "learning_rate": 4.532698666698347e-06,
      "loss": 0.5666,
      "step": 78430
    },
    {
      "epoch": 32.11,
      "grad_norm": 2.089017868041992,
      "learning_rate": 4.532641967491837e-06,
      "loss": 0.5963,
      "step": 78440
    },
    {
      "epoch": 32.11,
      "grad_norm": 3.5393404960632324,
      "learning_rate": 4.532585258258353e-06,
      "loss": 0.5704,
      "step": 78450
    },
    {
      "epoch": 32.12,
      "grad_norm": 2.3434314727783203,
      "learning_rate": 4.532528538998152e-06,
      "loss": 0.5873,
      "step": 78460
    },
    {
      "epoch": 32.12,
      "grad_norm": 2.487184524536133,
      "learning_rate": 4.532471809711497e-06,
      "loss": 0.5573,
      "step": 78470
    },
    {
      "epoch": 32.12,
      "grad_norm": 3.2127344608306885,
      "learning_rate": 4.532415070398645e-06,
      "loss": 0.585,
      "step": 78480
    },
    {
      "epoch": 32.13,
      "grad_norm": 3.452444314956665,
      "learning_rate": 4.532358321059858e-06,
      "loss": 0.5603,
      "step": 78490
    },
    {
      "epoch": 32.13,
      "grad_norm": 2.219619035720825,
      "learning_rate": 4.532301561695395e-06,
      "loss": 0.5728,
      "step": 78500
    },
    {
      "epoch": 32.14,
      "grad_norm": 2.77653169631958,
      "learning_rate": 4.532244792305517e-06,
      "loss": 0.5643,
      "step": 78510
    },
    {
      "epoch": 32.14,
      "grad_norm": 3.2155203819274902,
      "learning_rate": 4.532188012890482e-06,
      "loss": 0.5586,
      "step": 78520
    },
    {
      "epoch": 32.14,
      "grad_norm": 2.671781539916992,
      "learning_rate": 4.5321312234505526e-06,
      "loss": 0.5889,
      "step": 78530
    },
    {
      "epoch": 32.15,
      "grad_norm": 2.2025811672210693,
      "learning_rate": 4.532074423985987e-06,
      "loss": 0.5859,
      "step": 78540
    },
    {
      "epoch": 32.15,
      "grad_norm": 2.6175737380981445,
      "learning_rate": 4.532017614497046e-06,
      "loss": 0.5715,
      "step": 78550
    },
    {
      "epoch": 32.16,
      "grad_norm": 2.2038381099700928,
      "learning_rate": 4.53196079498399e-06,
      "loss": 0.5783,
      "step": 78560
    },
    {
      "epoch": 32.16,
      "grad_norm": 2.267577648162842,
      "learning_rate": 4.5319039654470806e-06,
      "loss": 0.5606,
      "step": 78570
    },
    {
      "epoch": 32.17,
      "grad_norm": 2.288553237915039,
      "learning_rate": 4.531847125886575e-06,
      "loss": 0.5842,
      "step": 78580
    },
    {
      "epoch": 32.17,
      "grad_norm": 2.408884286880493,
      "learning_rate": 4.531790276302736e-06,
      "loss": 0.5597,
      "step": 78590
    },
    {
      "epoch": 32.17,
      "grad_norm": 2.3284554481506348,
      "learning_rate": 4.531733416695823e-06,
      "loss": 0.5629,
      "step": 78600
    },
    {
      "epoch": 32.18,
      "grad_norm": 2.7667722702026367,
      "learning_rate": 4.531676547066097e-06,
      "loss": 0.5731,
      "step": 78610
    },
    {
      "epoch": 32.18,
      "grad_norm": 2.6436328887939453,
      "learning_rate": 4.531619667413819e-06,
      "loss": 0.5795,
      "step": 78620
    },
    {
      "epoch": 32.19,
      "grad_norm": 2.53243350982666,
      "learning_rate": 4.531562777739249e-06,
      "loss": 0.5796,
      "step": 78630
    },
    {
      "epoch": 32.19,
      "grad_norm": 2.404256820678711,
      "learning_rate": 4.5315058780426465e-06,
      "loss": 0.5688,
      "step": 78640
    },
    {
      "epoch": 32.19,
      "grad_norm": 2.1382741928100586,
      "learning_rate": 4.531448968324273e-06,
      "loss": 0.5527,
      "step": 78650
    },
    {
      "epoch": 32.2,
      "grad_norm": 1.9422252178192139,
      "learning_rate": 4.53139204858439e-06,
      "loss": 0.5701,
      "step": 78660
    },
    {
      "epoch": 32.2,
      "grad_norm": 2.254448175430298,
      "learning_rate": 4.531335118823257e-06,
      "loss": 0.5618,
      "step": 78670
    },
    {
      "epoch": 32.21,
      "grad_norm": 2.1632778644561768,
      "learning_rate": 4.531278179041136e-06,
      "loss": 0.5513,
      "step": 78680
    },
    {
      "epoch": 32.21,
      "grad_norm": 2.583832263946533,
      "learning_rate": 4.531221229238288e-06,
      "loss": 0.5639,
      "step": 78690
    },
    {
      "epoch": 32.21,
      "grad_norm": 2.3683342933654785,
      "learning_rate": 4.531164269414971e-06,
      "loss": 0.5841,
      "step": 78700
    },
    {
      "epoch": 32.22,
      "grad_norm": 2.031176805496216,
      "learning_rate": 4.531107299571449e-06,
      "loss": 0.5922,
      "step": 78710
    },
    {
      "epoch": 32.22,
      "grad_norm": 2.1544415950775146,
      "learning_rate": 4.531050319707982e-06,
      "loss": 0.5545,
      "step": 78720
    },
    {
      "epoch": 32.23,
      "grad_norm": 2.9084930419921875,
      "learning_rate": 4.5309933298248315e-06,
      "loss": 0.5553,
      "step": 78730
    },
    {
      "epoch": 32.23,
      "grad_norm": 3.0726747512817383,
      "learning_rate": 4.530936329922257e-06,
      "loss": 0.56,
      "step": 78740
    },
    {
      "epoch": 32.23,
      "grad_norm": 1.7806328535079956,
      "learning_rate": 4.530879320000521e-06,
      "loss": 0.5523,
      "step": 78750
    },
    {
      "epoch": 32.24,
      "grad_norm": 2.038102865219116,
      "learning_rate": 4.530822300059885e-06,
      "loss": 0.5672,
      "step": 78760
    },
    {
      "epoch": 32.24,
      "grad_norm": 2.6864638328552246,
      "learning_rate": 4.530765270100609e-06,
      "loss": 0.5617,
      "step": 78770
    },
    {
      "epoch": 32.25,
      "grad_norm": 2.3900375366210938,
      "learning_rate": 4.5307082301229545e-06,
      "loss": 0.5852,
      "step": 78780
    },
    {
      "epoch": 32.25,
      "grad_norm": 2.7428057193756104,
      "learning_rate": 4.530651180127184e-06,
      "loss": 0.5539,
      "step": 78790
    },
    {
      "epoch": 32.26,
      "grad_norm": 1.9741833209991455,
      "learning_rate": 4.530594120113557e-06,
      "loss": 0.5859,
      "step": 78800
    },
    {
      "epoch": 32.26,
      "grad_norm": 2.544848918914795,
      "learning_rate": 4.530537050082336e-06,
      "loss": 0.572,
      "step": 78810
    },
    {
      "epoch": 32.26,
      "grad_norm": 2.735474109649658,
      "learning_rate": 4.530479970033781e-06,
      "loss": 0.5736,
      "step": 78820
    },
    {
      "epoch": 32.27,
      "grad_norm": 2.013164520263672,
      "learning_rate": 4.5304228799681565e-06,
      "loss": 0.5878,
      "step": 78830
    },
    {
      "epoch": 32.27,
      "grad_norm": 1.9862143993377686,
      "learning_rate": 4.5303657798857215e-06,
      "loss": 0.5814,
      "step": 78840
    },
    {
      "epoch": 32.28,
      "grad_norm": 3.01301646232605,
      "learning_rate": 4.530308669786738e-06,
      "loss": 0.5683,
      "step": 78850
    },
    {
      "epoch": 32.28,
      "grad_norm": 2.7917206287384033,
      "learning_rate": 4.530251549671468e-06,
      "loss": 0.5749,
      "step": 78860
    },
    {
      "epoch": 32.28,
      "grad_norm": 2.31095027923584,
      "learning_rate": 4.530194419540174e-06,
      "loss": 0.5784,
      "step": 78870
    },
    {
      "epoch": 32.29,
      "grad_norm": 2.085369825363159,
      "learning_rate": 4.530137279393115e-06,
      "loss": 0.5522,
      "step": 78880
    },
    {
      "epoch": 32.29,
      "grad_norm": 2.227648973464966,
      "learning_rate": 4.530080129230555e-06,
      "loss": 0.5637,
      "step": 78890
    },
    {
      "epoch": 32.3,
      "grad_norm": 2.1896138191223145,
      "learning_rate": 4.530022969052756e-06,
      "loss": 0.5848,
      "step": 78900
    },
    {
      "epoch": 32.3,
      "grad_norm": 2.6325268745422363,
      "learning_rate": 4.5299657988599795e-06,
      "loss": 0.5652,
      "step": 78910
    },
    {
      "epoch": 32.3,
      "grad_norm": 2.411480665206909,
      "learning_rate": 4.529908618652485e-06,
      "loss": 0.5576,
      "step": 78920
    },
    {
      "epoch": 32.31,
      "grad_norm": 3.4247801303863525,
      "learning_rate": 4.529851428430538e-06,
      "loss": 0.5654,
      "step": 78930
    },
    {
      "epoch": 32.31,
      "grad_norm": 1.9097952842712402,
      "learning_rate": 4.529794228194398e-06,
      "loss": 0.5585,
      "step": 78940
    },
    {
      "epoch": 32.32,
      "grad_norm": 2.7631354331970215,
      "learning_rate": 4.529737017944329e-06,
      "loss": 0.549,
      "step": 78950
    },
    {
      "epoch": 32.32,
      "grad_norm": 2.2565722465515137,
      "learning_rate": 4.529679797680592e-06,
      "loss": 0.5837,
      "step": 78960
    },
    {
      "epoch": 32.33,
      "grad_norm": 2.888117551803589,
      "learning_rate": 4.529622567403449e-06,
      "loss": 0.5667,
      "step": 78970
    },
    {
      "epoch": 32.33,
      "grad_norm": 2.7144651412963867,
      "learning_rate": 4.529565327113161e-06,
      "loss": 0.5494,
      "step": 78980
    },
    {
      "epoch": 32.33,
      "grad_norm": 2.5252368450164795,
      "learning_rate": 4.529508076809993e-06,
      "loss": 0.5823,
      "step": 78990
    },
    {
      "epoch": 32.34,
      "grad_norm": 2.4887940883636475,
      "learning_rate": 4.529450816494205e-06,
      "loss": 0.5671,
      "step": 79000
    },
    {
      "epoch": 32.34,
      "grad_norm": 2.1581616401672363,
      "learning_rate": 4.52939354616606e-06,
      "loss": 0.5646,
      "step": 79010
    },
    {
      "epoch": 32.35,
      "grad_norm": 2.8574304580688477,
      "learning_rate": 4.52933626582582e-06,
      "loss": 0.592,
      "step": 79020
    },
    {
      "epoch": 32.35,
      "grad_norm": 2.4782943725585938,
      "learning_rate": 4.529278975473749e-06,
      "loss": 0.5708,
      "step": 79030
    },
    {
      "epoch": 32.35,
      "grad_norm": 2.2000346183776855,
      "learning_rate": 4.529221675110107e-06,
      "loss": 0.5856,
      "step": 79040
    },
    {
      "epoch": 32.36,
      "grad_norm": 3.081641912460327,
      "learning_rate": 4.529164364735159e-06,
      "loss": 0.5654,
      "step": 79050
    },
    {
      "epoch": 32.36,
      "grad_norm": 1.7153711318969727,
      "learning_rate": 4.529107044349166e-06,
      "loss": 0.5547,
      "step": 79060
    },
    {
      "epoch": 32.37,
      "grad_norm": 2.5136420726776123,
      "learning_rate": 4.52904971395239e-06,
      "loss": 0.5843,
      "step": 79070
    },
    {
      "epoch": 32.37,
      "grad_norm": 2.3042187690734863,
      "learning_rate": 4.528992373545096e-06,
      "loss": 0.5477,
      "step": 79080
    },
    {
      "epoch": 32.37,
      "grad_norm": 2.2451343536376953,
      "learning_rate": 4.528935023127543e-06,
      "loss": 0.5865,
      "step": 79090
    },
    {
      "epoch": 32.38,
      "grad_norm": 2.1043152809143066,
      "learning_rate": 4.528877662699997e-06,
      "loss": 0.5487,
      "step": 79100
    },
    {
      "epoch": 32.38,
      "grad_norm": 2.2813093662261963,
      "learning_rate": 4.52882029226272e-06,
      "loss": 0.5691,
      "step": 79110
    },
    {
      "epoch": 32.39,
      "grad_norm": 2.1587564945220947,
      "learning_rate": 4.5287629118159735e-06,
      "loss": 0.5489,
      "step": 79120
    },
    {
      "epoch": 32.39,
      "grad_norm": 2.3655855655670166,
      "learning_rate": 4.528705521360022e-06,
      "loss": 0.5756,
      "step": 79130
    },
    {
      "epoch": 32.39,
      "grad_norm": 2.734523057937622,
      "learning_rate": 4.528648120895127e-06,
      "loss": 0.5615,
      "step": 79140
    },
    {
      "epoch": 32.4,
      "grad_norm": 2.349912643432617,
      "learning_rate": 4.528590710421552e-06,
      "loss": 0.5843,
      "step": 79150
    },
    {
      "epoch": 32.4,
      "grad_norm": 1.87503182888031,
      "learning_rate": 4.52853328993956e-06,
      "loss": 0.5675,
      "step": 79160
    },
    {
      "epoch": 32.41,
      "grad_norm": 2.389723539352417,
      "learning_rate": 4.528475859449415e-06,
      "loss": 0.545,
      "step": 79170
    },
    {
      "epoch": 32.41,
      "grad_norm": 2.3604161739349365,
      "learning_rate": 4.528418418951379e-06,
      "loss": 0.5729,
      "step": 79180
    },
    {
      "epoch": 32.42,
      "grad_norm": 2.4876620769500732,
      "learning_rate": 4.528360968445715e-06,
      "loss": 0.5916,
      "step": 79190
    },
    {
      "epoch": 32.42,
      "grad_norm": 2.6266369819641113,
      "learning_rate": 4.528303507932686e-06,
      "loss": 0.5791,
      "step": 79200
    },
    {
      "epoch": 32.42,
      "grad_norm": 2.325303554534912,
      "learning_rate": 4.528246037412557e-06,
      "loss": 0.5688,
      "step": 79210
    },
    {
      "epoch": 32.43,
      "grad_norm": 2.3686985969543457,
      "learning_rate": 4.528188556885588e-06,
      "loss": 0.5792,
      "step": 79220
    },
    {
      "epoch": 32.43,
      "grad_norm": 1.9922786951065063,
      "learning_rate": 4.528131066352046e-06,
      "loss": 0.578,
      "step": 79230
    },
    {
      "epoch": 32.44,
      "grad_norm": 3.792658567428589,
      "learning_rate": 4.5280735658121916e-06,
      "loss": 0.6079,
      "step": 79240
    },
    {
      "epoch": 32.44,
      "grad_norm": 2.518186330795288,
      "learning_rate": 4.52801605526629e-06,
      "loss": 0.5746,
      "step": 79250
    },
    {
      "epoch": 32.44,
      "grad_norm": 2.435699462890625,
      "learning_rate": 4.527958534714604e-06,
      "loss": 0.5738,
      "step": 79260
    },
    {
      "epoch": 32.45,
      "grad_norm": 2.6744449138641357,
      "learning_rate": 4.527901004157396e-06,
      "loss": 0.5656,
      "step": 79270
    },
    {
      "epoch": 32.45,
      "grad_norm": 2.148310661315918,
      "learning_rate": 4.527843463594931e-06,
      "loss": 0.5787,
      "step": 79280
    },
    {
      "epoch": 32.46,
      "grad_norm": 2.0822737216949463,
      "learning_rate": 4.527785913027472e-06,
      "loss": 0.5598,
      "step": 79290
    },
    {
      "epoch": 32.46,
      "grad_norm": 2.5404787063598633,
      "learning_rate": 4.527728352455283e-06,
      "loss": 0.5682,
      "step": 79300
    },
    {
      "epoch": 32.46,
      "grad_norm": 2.232741355895996,
      "learning_rate": 4.527670781878627e-06,
      "loss": 0.5531,
      "step": 79310
    },
    {
      "epoch": 32.47,
      "grad_norm": 1.8488729000091553,
      "learning_rate": 4.527613201297768e-06,
      "loss": 0.597,
      "step": 79320
    },
    {
      "epoch": 32.47,
      "grad_norm": 2.2174623012542725,
      "learning_rate": 4.527555610712971e-06,
      "loss": 0.5694,
      "step": 79330
    },
    {
      "epoch": 32.48,
      "grad_norm": 2.3771815299987793,
      "learning_rate": 4.527498010124497e-06,
      "loss": 0.5701,
      "step": 79340
    },
    {
      "epoch": 32.48,
      "grad_norm": 8.219464302062988,
      "learning_rate": 4.527440399532613e-06,
      "loss": 0.5542,
      "step": 79350
    },
    {
      "epoch": 32.48,
      "grad_norm": 2.366297960281372,
      "learning_rate": 4.527382778937582e-06,
      "loss": 0.5588,
      "step": 79360
    },
    {
      "epoch": 32.49,
      "grad_norm": 2.169647693634033,
      "learning_rate": 4.527325148339665e-06,
      "loss": 0.5512,
      "step": 79370
    },
    {
      "epoch": 32.49,
      "grad_norm": 2.06571626663208,
      "learning_rate": 4.52726750773913e-06,
      "loss": 0.5794,
      "step": 79380
    },
    {
      "epoch": 32.5,
      "grad_norm": 3.6722209453582764,
      "learning_rate": 4.5272098571362394e-06,
      "loss": 0.5687,
      "step": 79390
    },
    {
      "epoch": 32.5,
      "grad_norm": 1.8959429264068604,
      "learning_rate": 4.527152196531258e-06,
      "loss": 0.5634,
      "step": 79400
    },
    {
      "epoch": 32.51,
      "grad_norm": 2.7817440032958984,
      "learning_rate": 4.527094525924448e-06,
      "loss": 0.5601,
      "step": 79410
    },
    {
      "epoch": 32.51,
      "grad_norm": 3.025219202041626,
      "learning_rate": 4.527036845316076e-06,
      "loss": 0.6014,
      "step": 79420
    },
    {
      "epoch": 32.51,
      "grad_norm": 2.738072395324707,
      "learning_rate": 4.526979154706404e-06,
      "loss": 0.5629,
      "step": 79430
    },
    {
      "epoch": 32.52,
      "grad_norm": 1.828729510307312,
      "learning_rate": 4.526921454095698e-06,
      "loss": 0.5474,
      "step": 79440
    },
    {
      "epoch": 32.52,
      "grad_norm": 2.3830478191375732,
      "learning_rate": 4.526863743484221e-06,
      "loss": 0.5743,
      "step": 79450
    },
    {
      "epoch": 32.53,
      "grad_norm": 2.5712714195251465,
      "learning_rate": 4.5268060228722386e-06,
      "loss": 0.5433,
      "step": 79460
    },
    {
      "epoch": 32.53,
      "grad_norm": 3.214953660964966,
      "learning_rate": 4.526748292260014e-06,
      "loss": 0.5681,
      "step": 79470
    },
    {
      "epoch": 32.53,
      "grad_norm": 1.8825982809066772,
      "learning_rate": 4.526690551647813e-06,
      "loss": 0.5799,
      "step": 79480
    },
    {
      "epoch": 32.54,
      "grad_norm": 2.5292439460754395,
      "learning_rate": 4.5266328010359e-06,
      "loss": 0.5713,
      "step": 79490
    },
    {
      "epoch": 32.54,
      "grad_norm": 2.3819618225097656,
      "learning_rate": 4.526575040424538e-06,
      "loss": 0.5634,
      "step": 79500
    },
    {
      "epoch": 32.55,
      "grad_norm": 2.5317163467407227,
      "learning_rate": 4.526517269813992e-06,
      "loss": 0.5859,
      "step": 79510
    },
    {
      "epoch": 32.55,
      "grad_norm": 2.7877540588378906,
      "learning_rate": 4.5264594892045285e-06,
      "loss": 0.5843,
      "step": 79520
    },
    {
      "epoch": 32.55,
      "grad_norm": 2.950493097305298,
      "learning_rate": 4.52640169859641e-06,
      "loss": 0.5743,
      "step": 79530
    },
    {
      "epoch": 32.56,
      "grad_norm": 2.4212565422058105,
      "learning_rate": 4.5263438979899025e-06,
      "loss": 0.5617,
      "step": 79540
    },
    {
      "epoch": 32.56,
      "grad_norm": 2.4199798107147217,
      "learning_rate": 4.526286087385271e-06,
      "loss": 0.5773,
      "step": 79550
    },
    {
      "epoch": 32.57,
      "grad_norm": 2.2469708919525146,
      "learning_rate": 4.5262282667827785e-06,
      "loss": 0.567,
      "step": 79560
    },
    {
      "epoch": 32.57,
      "grad_norm": 2.107842445373535,
      "learning_rate": 4.526170436182691e-06,
      "loss": 0.5826,
      "step": 79570
    },
    {
      "epoch": 32.57,
      "grad_norm": 2.1406137943267822,
      "learning_rate": 4.526112595585274e-06,
      "loss": 0.5695,
      "step": 79580
    },
    {
      "epoch": 32.58,
      "grad_norm": 2.525019884109497,
      "learning_rate": 4.526054744990792e-06,
      "loss": 0.5757,
      "step": 79590
    },
    {
      "epoch": 32.58,
      "grad_norm": 3.070769786834717,
      "learning_rate": 4.52599688439951e-06,
      "loss": 0.5748,
      "step": 79600
    },
    {
      "epoch": 32.59,
      "grad_norm": 2.275784492492676,
      "learning_rate": 4.525939013811693e-06,
      "loss": 0.5754,
      "step": 79610
    },
    {
      "epoch": 32.59,
      "grad_norm": 3.7507500648498535,
      "learning_rate": 4.5258811332276064e-06,
      "loss": 0.587,
      "step": 79620
    },
    {
      "epoch": 32.6,
      "grad_norm": 1.8768279552459717,
      "learning_rate": 4.5258232426475145e-06,
      "loss": 0.5762,
      "step": 79630
    },
    {
      "epoch": 32.6,
      "grad_norm": 2.375994920730591,
      "learning_rate": 4.5257653420716825e-06,
      "loss": 0.5605,
      "step": 79640
    },
    {
      "epoch": 32.6,
      "grad_norm": 2.2939963340759277,
      "learning_rate": 4.525707431500377e-06,
      "loss": 0.5793,
      "step": 79650
    },
    {
      "epoch": 32.61,
      "grad_norm": 3.7926535606384277,
      "learning_rate": 4.525649510933862e-06,
      "loss": 0.5855,
      "step": 79660
    },
    {
      "epoch": 32.61,
      "grad_norm": 2.21986722946167,
      "learning_rate": 4.525591580372403e-06,
      "loss": 0.5507,
      "step": 79670
    },
    {
      "epoch": 32.62,
      "grad_norm": 1.9681432247161865,
      "learning_rate": 4.525533639816266e-06,
      "loss": 0.5889,
      "step": 79680
    },
    {
      "epoch": 32.62,
      "grad_norm": 2.5946402549743652,
      "learning_rate": 4.525475689265716e-06,
      "loss": 0.5637,
      "step": 79690
    },
    {
      "epoch": 32.62,
      "grad_norm": 2.9246609210968018,
      "learning_rate": 4.525417728721019e-06,
      "loss": 0.5564,
      "step": 79700
    },
    {
      "epoch": 32.63,
      "grad_norm": 2.1642487049102783,
      "learning_rate": 4.52535975818244e-06,
      "loss": 0.5705,
      "step": 79710
    },
    {
      "epoch": 32.63,
      "grad_norm": 2.8758442401885986,
      "learning_rate": 4.525301777650244e-06,
      "loss": 0.555,
      "step": 79720
    },
    {
      "epoch": 32.64,
      "grad_norm": 2.289092540740967,
      "learning_rate": 4.525243787124697e-06,
      "loss": 0.5833,
      "step": 79730
    },
    {
      "epoch": 32.64,
      "grad_norm": 2.7831413745880127,
      "learning_rate": 4.525185786606065e-06,
      "loss": 0.5487,
      "step": 79740
    },
    {
      "epoch": 32.64,
      "grad_norm": 2.538559913635254,
      "learning_rate": 4.525127776094614e-06,
      "loss": 0.5864,
      "step": 79750
    },
    {
      "epoch": 32.65,
      "grad_norm": 1.7036164999008179,
      "learning_rate": 4.525069755590609e-06,
      "loss": 0.5749,
      "step": 79760
    },
    {
      "epoch": 32.65,
      "grad_norm": 2.7783408164978027,
      "learning_rate": 4.525011725094315e-06,
      "loss": 0.5572,
      "step": 79770
    },
    {
      "epoch": 32.66,
      "grad_norm": 1.809090256690979,
      "learning_rate": 4.524953684606e-06,
      "loss": 0.5648,
      "step": 79780
    },
    {
      "epoch": 32.66,
      "grad_norm": 2.1846537590026855,
      "learning_rate": 4.524895634125929e-06,
      "loss": 0.5592,
      "step": 79790
    },
    {
      "epoch": 32.66,
      "grad_norm": 2.228660821914673,
      "learning_rate": 4.524837573654366e-06,
      "loss": 0.5687,
      "step": 79800
    },
    {
      "epoch": 32.67,
      "grad_norm": 2.0903384685516357,
      "learning_rate": 4.52477950319158e-06,
      "loss": 0.5833,
      "step": 79810
    },
    {
      "epoch": 32.67,
      "grad_norm": 2.177515983581543,
      "learning_rate": 4.524721422737835e-06,
      "loss": 0.5804,
      "step": 79820
    },
    {
      "epoch": 32.68,
      "grad_norm": 2.6227259635925293,
      "learning_rate": 4.524663332293398e-06,
      "loss": 0.5856,
      "step": 79830
    },
    {
      "epoch": 32.68,
      "grad_norm": 2.6298837661743164,
      "learning_rate": 4.524605231858535e-06,
      "loss": 0.5808,
      "step": 79840
    },
    {
      "epoch": 32.69,
      "grad_norm": 2.1750905513763428,
      "learning_rate": 4.524547121433511e-06,
      "loss": 0.5553,
      "step": 79850
    },
    {
      "epoch": 32.69,
      "grad_norm": 2.1797008514404297,
      "learning_rate": 4.524489001018595e-06,
      "loss": 0.5694,
      "step": 79860
    },
    {
      "epoch": 32.69,
      "grad_norm": 2.332679271697998,
      "learning_rate": 4.5244308706140495e-06,
      "loss": 0.552,
      "step": 79870
    },
    {
      "epoch": 32.7,
      "grad_norm": 2.4238321781158447,
      "learning_rate": 4.524372730220143e-06,
      "loss": 0.5826,
      "step": 79880
    },
    {
      "epoch": 32.7,
      "grad_norm": 2.654261589050293,
      "learning_rate": 4.524314579837143e-06,
      "loss": 0.55,
      "step": 79890
    },
    {
      "epoch": 32.71,
      "grad_norm": 2.038548231124878,
      "learning_rate": 4.524256419465313e-06,
      "loss": 0.5869,
      "step": 79900
    },
    {
      "epoch": 32.71,
      "grad_norm": 2.6137020587921143,
      "learning_rate": 4.524198249104921e-06,
      "loss": 0.5686,
      "step": 79910
    },
    {
      "epoch": 32.71,
      "grad_norm": 2.3958358764648438,
      "learning_rate": 4.524140068756232e-06,
      "loss": 0.5628,
      "step": 79920
    },
    {
      "epoch": 32.72,
      "grad_norm": 2.413422107696533,
      "learning_rate": 4.524081878419516e-06,
      "loss": 0.5862,
      "step": 79930
    },
    {
      "epoch": 32.72,
      "grad_norm": 2.635110378265381,
      "learning_rate": 4.524023678095036e-06,
      "loss": 0.5438,
      "step": 79940
    },
    {
      "epoch": 32.73,
      "grad_norm": 2.4285643100738525,
      "learning_rate": 4.52396546778306e-06,
      "loss": 0.5676,
      "step": 79950
    },
    {
      "epoch": 32.73,
      "grad_norm": 2.5896201133728027,
      "learning_rate": 4.523907247483855e-06,
      "loss": 0.5773,
      "step": 79960
    },
    {
      "epoch": 32.73,
      "grad_norm": 2.411398410797119,
      "learning_rate": 4.5238490171976865e-06,
      "loss": 0.5907,
      "step": 79970
    },
    {
      "epoch": 32.74,
      "grad_norm": 2.079883337020874,
      "learning_rate": 4.523790776924823e-06,
      "loss": 0.5535,
      "step": 79980
    },
    {
      "epoch": 32.74,
      "grad_norm": 2.9092438220977783,
      "learning_rate": 4.52373252666553e-06,
      "loss": 0.5537,
      "step": 79990
    },
    {
      "epoch": 32.75,
      "grad_norm": 1.9728728532791138,
      "learning_rate": 4.523674266420075e-06,
      "loss": 0.5559,
      "step": 80000
    },
    {
      "epoch": 32.75,
      "grad_norm": 2.2280313968658447,
      "learning_rate": 4.523615996188724e-06,
      "loss": 0.5518,
      "step": 80010
    },
    {
      "epoch": 32.75,
      "grad_norm": 2.010087013244629,
      "learning_rate": 4.523557715971745e-06,
      "loss": 0.5882,
      "step": 80020
    },
    {
      "epoch": 32.76,
      "grad_norm": 2.203242301940918,
      "learning_rate": 4.523499425769404e-06,
      "loss": 0.5516,
      "step": 80030
    },
    {
      "epoch": 32.76,
      "grad_norm": 2.070305824279785,
      "learning_rate": 4.523441125581968e-06,
      "loss": 0.6148,
      "step": 80040
    },
    {
      "epoch": 32.77,
      "grad_norm": 2.6460530757904053,
      "learning_rate": 4.523382815409705e-06,
      "loss": 0.5553,
      "step": 80050
    },
    {
      "epoch": 32.77,
      "grad_norm": 1.9216537475585938,
      "learning_rate": 4.523324495252882e-06,
      "loss": 0.5555,
      "step": 80060
    },
    {
      "epoch": 32.78,
      "grad_norm": 2.219714403152466,
      "learning_rate": 4.5232661651117655e-06,
      "loss": 0.5756,
      "step": 80070
    },
    {
      "epoch": 32.78,
      "grad_norm": 2.4635510444641113,
      "learning_rate": 4.523207824986624e-06,
      "loss": 0.572,
      "step": 80080
    },
    {
      "epoch": 32.78,
      "grad_norm": 2.090485095977783,
      "learning_rate": 4.523149474877722e-06,
      "loss": 0.5688,
      "step": 80090
    },
    {
      "epoch": 32.79,
      "grad_norm": 2.088209867477417,
      "learning_rate": 4.52309111478533e-06,
      "loss": 0.5698,
      "step": 80100
    },
    {
      "epoch": 32.79,
      "grad_norm": 2.6270318031311035,
      "learning_rate": 4.523032744709712e-06,
      "loss": 0.564,
      "step": 80110
    },
    {
      "epoch": 32.8,
      "grad_norm": 2.4413647651672363,
      "learning_rate": 4.522974364651139e-06,
      "loss": 0.5575,
      "step": 80120
    },
    {
      "epoch": 32.8,
      "grad_norm": 1.7867045402526855,
      "learning_rate": 4.522915974609876e-06,
      "loss": 0.5904,
      "step": 80130
    },
    {
      "epoch": 32.8,
      "grad_norm": 2.3951165676116943,
      "learning_rate": 4.522857574586191e-06,
      "loss": 0.5661,
      "step": 80140
    },
    {
      "epoch": 32.81,
      "grad_norm": 2.4086673259735107,
      "learning_rate": 4.5227991645803515e-06,
      "loss": 0.5753,
      "step": 80150
    },
    {
      "epoch": 32.81,
      "grad_norm": 2.878032684326172,
      "learning_rate": 4.522740744592626e-06,
      "loss": 0.5787,
      "step": 80160
    },
    {
      "epoch": 32.82,
      "grad_norm": 1.9349865913391113,
      "learning_rate": 4.522682314623281e-06,
      "loss": 0.5456,
      "step": 80170
    },
    {
      "epoch": 32.82,
      "grad_norm": 2.503779649734497,
      "learning_rate": 4.5226238746725845e-06,
      "loss": 0.5752,
      "step": 80180
    },
    {
      "epoch": 32.82,
      "grad_norm": 2.0198755264282227,
      "learning_rate": 4.5225654247408045e-06,
      "loss": 0.5774,
      "step": 80190
    },
    {
      "epoch": 32.83,
      "grad_norm": 3.502448558807373,
      "learning_rate": 4.522506964828207e-06,
      "loss": 0.583,
      "step": 80200
    },
    {
      "epoch": 32.83,
      "grad_norm": 1.8344517946243286,
      "learning_rate": 4.5224484949350634e-06,
      "loss": 0.5698,
      "step": 80210
    },
    {
      "epoch": 32.84,
      "grad_norm": 2.368838310241699,
      "learning_rate": 4.522390015061638e-06,
      "loss": 0.5997,
      "step": 80220
    },
    {
      "epoch": 32.84,
      "grad_norm": 2.574176073074341,
      "learning_rate": 4.5223315252082005e-06,
      "loss": 0.5696,
      "step": 80230
    },
    {
      "epoch": 32.84,
      "grad_norm": 2.1829264163970947,
      "learning_rate": 4.5222730253750185e-06,
      "loss": 0.558,
      "step": 80240
    },
    {
      "epoch": 32.85,
      "grad_norm": 2.2293152809143066,
      "learning_rate": 4.52221451556236e-06,
      "loss": 0.564,
      "step": 80250
    },
    {
      "epoch": 32.85,
      "grad_norm": 2.300067901611328,
      "learning_rate": 4.522155995770493e-06,
      "loss": 0.5612,
      "step": 80260
    },
    {
      "epoch": 32.86,
      "grad_norm": 2.6249022483825684,
      "learning_rate": 4.522097465999685e-06,
      "loss": 0.5705,
      "step": 80270
    },
    {
      "epoch": 32.86,
      "grad_norm": 2.381531238555908,
      "learning_rate": 4.522038926250205e-06,
      "loss": 0.5765,
      "step": 80280
    },
    {
      "epoch": 32.87,
      "grad_norm": 3.0491557121276855,
      "learning_rate": 4.52198037652232e-06,
      "loss": 0.5736,
      "step": 80290
    },
    {
      "epoch": 32.87,
      "grad_norm": 1.9924544095993042,
      "learning_rate": 4.5219218168163e-06,
      "loss": 0.562,
      "step": 80300
    },
    {
      "epoch": 32.87,
      "grad_norm": 2.999743938446045,
      "learning_rate": 4.5218632471324125e-06,
      "loss": 0.5805,
      "step": 80310
    },
    {
      "epoch": 32.88,
      "grad_norm": 2.921898603439331,
      "learning_rate": 4.521804667470925e-06,
      "loss": 0.5836,
      "step": 80320
    },
    {
      "epoch": 32.88,
      "grad_norm": 2.3374531269073486,
      "learning_rate": 4.521746077832106e-06,
      "loss": 0.5741,
      "step": 80330
    },
    {
      "epoch": 32.89,
      "grad_norm": 2.919755220413208,
      "learning_rate": 4.5216874782162254e-06,
      "loss": 0.5927,
      "step": 80340
    },
    {
      "epoch": 32.89,
      "grad_norm": 2.870084524154663,
      "learning_rate": 4.521628868623549e-06,
      "loss": 0.5689,
      "step": 80350
    },
    {
      "epoch": 32.89,
      "grad_norm": 2.0789873600006104,
      "learning_rate": 4.521570249054348e-06,
      "loss": 0.5532,
      "step": 80360
    },
    {
      "epoch": 32.9,
      "grad_norm": 1.9921703338623047,
      "learning_rate": 4.5215116195088895e-06,
      "loss": 0.5514,
      "step": 80370
    },
    {
      "epoch": 32.9,
      "grad_norm": 2.184020519256592,
      "learning_rate": 4.521452979987441e-06,
      "loss": 0.591,
      "step": 80380
    },
    {
      "epoch": 32.91,
      "grad_norm": 2.208270311355591,
      "learning_rate": 4.521394330490274e-06,
      "loss": 0.5588,
      "step": 80390
    },
    {
      "epoch": 32.91,
      "grad_norm": 2.2428510189056396,
      "learning_rate": 4.521335671017655e-06,
      "loss": 0.5819,
      "step": 80400
    },
    {
      "epoch": 32.91,
      "grad_norm": 1.9114233255386353,
      "learning_rate": 4.521277001569853e-06,
      "loss": 0.5638,
      "step": 80410
    },
    {
      "epoch": 32.92,
      "grad_norm": 3.166583299636841,
      "learning_rate": 4.5212183221471374e-06,
      "loss": 0.5702,
      "step": 80420
    },
    {
      "epoch": 32.92,
      "grad_norm": 2.0446321964263916,
      "learning_rate": 4.521159632749777e-06,
      "loss": 0.5797,
      "step": 80430
    },
    {
      "epoch": 32.93,
      "grad_norm": 3.7447471618652344,
      "learning_rate": 4.52110093337804e-06,
      "loss": 0.6117,
      "step": 80440
    },
    {
      "epoch": 32.93,
      "grad_norm": 2.4027624130249023,
      "learning_rate": 4.521042224032195e-06,
      "loss": 0.553,
      "step": 80450
    },
    {
      "epoch": 32.93,
      "grad_norm": 1.933945655822754,
      "learning_rate": 4.520983504712512e-06,
      "loss": 0.5649,
      "step": 80460
    },
    {
      "epoch": 32.94,
      "grad_norm": 1.9642751216888428,
      "learning_rate": 4.520924775419259e-06,
      "loss": 0.581,
      "step": 80470
    },
    {
      "epoch": 32.94,
      "grad_norm": 1.9667515754699707,
      "learning_rate": 4.520866036152707e-06,
      "loss": 0.5846,
      "step": 80480
    },
    {
      "epoch": 32.95,
      "grad_norm": 2.413195848464966,
      "learning_rate": 4.520807286913122e-06,
      "loss": 0.5874,
      "step": 80490
    },
    {
      "epoch": 32.95,
      "grad_norm": 2.166651964187622,
      "learning_rate": 4.520748527700776e-06,
      "loss": 0.5936,
      "step": 80500
    },
    {
      "epoch": 32.96,
      "grad_norm": 2.049217462539673,
      "learning_rate": 4.520689758515936e-06,
      "loss": 0.5654,
      "step": 80510
    },
    {
      "epoch": 32.96,
      "grad_norm": 2.0839152336120605,
      "learning_rate": 4.520630979358872e-06,
      "loss": 0.5836,
      "step": 80520
    },
    {
      "epoch": 32.96,
      "grad_norm": 1.7499841451644897,
      "learning_rate": 4.520572190229854e-06,
      "loss": 0.5628,
      "step": 80530
    },
    {
      "epoch": 32.97,
      "grad_norm": 2.4926185607910156,
      "learning_rate": 4.520513391129151e-06,
      "loss": 0.5729,
      "step": 80540
    },
    {
      "epoch": 32.97,
      "grad_norm": 2.742215156555176,
      "learning_rate": 4.520454582057031e-06,
      "loss": 0.5776,
      "step": 80550
    },
    {
      "epoch": 32.98,
      "grad_norm": 1.9524561166763306,
      "learning_rate": 4.520395763013764e-06,
      "loss": 0.5426,
      "step": 80560
    },
    {
      "epoch": 32.98,
      "grad_norm": 2.3469574451446533,
      "learning_rate": 4.520336933999621e-06,
      "loss": 0.5734,
      "step": 80570
    },
    {
      "epoch": 32.98,
      "grad_norm": 3.5676755905151367,
      "learning_rate": 4.52027809501487e-06,
      "loss": 0.5625,
      "step": 80580
    },
    {
      "epoch": 32.99,
      "grad_norm": 2.69901442527771,
      "learning_rate": 4.520219246059781e-06,
      "loss": 0.5819,
      "step": 80590
    },
    {
      "epoch": 32.99,
      "grad_norm": 2.7370223999023438,
      "learning_rate": 4.520160387134624e-06,
      "loss": 0.5483,
      "step": 80600
    },
    {
      "epoch": 33.0,
      "grad_norm": 2.665503740310669,
      "learning_rate": 4.520101518239667e-06,
      "loss": 0.5739,
      "step": 80610
    },
    {
      "epoch": 33.0,
      "eval_loss": 0.5696834921836853,
      "eval_runtime": 52.3568,
      "eval_samples_per_second": 65.875,
      "eval_steps_per_second": 8.251,
      "step": 80619
    },
    {
      "epoch": 33.0,
      "grad_norm": 2.0022501945495605,
      "learning_rate": 4.520042639375181e-06,
      "loss": 0.5594,
      "step": 80620
    },
    {
      "epoch": 33.0,
      "grad_norm": 2.46246075630188,
      "learning_rate": 4.519983750541435e-06,
      "loss": 0.5708,
      "step": 80630
    },
    {
      "epoch": 33.01,
      "grad_norm": 1.764562964439392,
      "learning_rate": 4.5199248517387e-06,
      "loss": 0.5567,
      "step": 80640
    },
    {
      "epoch": 33.01,
      "grad_norm": 2.5414159297943115,
      "learning_rate": 4.519865942967246e-06,
      "loss": 0.5748,
      "step": 80650
    },
    {
      "epoch": 33.02,
      "grad_norm": 2.109189987182617,
      "learning_rate": 4.51980702422734e-06,
      "loss": 0.5756,
      "step": 80660
    },
    {
      "epoch": 33.02,
      "grad_norm": 2.0724501609802246,
      "learning_rate": 4.519748095519255e-06,
      "loss": 0.5848,
      "step": 80670
    },
    {
      "epoch": 33.02,
      "grad_norm": 2.345142364501953,
      "learning_rate": 4.519689156843259e-06,
      "loss": 0.5733,
      "step": 80680
    },
    {
      "epoch": 33.03,
      "grad_norm": 2.0577547550201416,
      "learning_rate": 4.519630208199623e-06,
      "loss": 0.5599,
      "step": 80690
    },
    {
      "epoch": 33.03,
      "grad_norm": 2.559893846511841,
      "learning_rate": 4.519571249588617e-06,
      "loss": 0.5956,
      "step": 80700
    },
    {
      "epoch": 33.04,
      "grad_norm": 2.2680039405822754,
      "learning_rate": 4.51951228101051e-06,
      "loss": 0.5745,
      "step": 80710
    },
    {
      "epoch": 33.04,
      "grad_norm": 2.6919803619384766,
      "learning_rate": 4.519453302465573e-06,
      "loss": 0.5541,
      "step": 80720
    },
    {
      "epoch": 33.05,
      "grad_norm": 2.6409225463867188,
      "learning_rate": 4.519394313954077e-06,
      "loss": 0.5662,
      "step": 80730
    },
    {
      "epoch": 33.05,
      "grad_norm": 2.7872352600097656,
      "learning_rate": 4.51933531547629e-06,
      "loss": 0.5606,
      "step": 80740
    },
    {
      "epoch": 33.05,
      "grad_norm": 2.3439583778381348,
      "learning_rate": 4.519276307032485e-06,
      "loss": 0.5692,
      "step": 80750
    },
    {
      "epoch": 33.06,
      "grad_norm": 2.2231431007385254,
      "learning_rate": 4.519217288622931e-06,
      "loss": 0.5761,
      "step": 80760
    },
    {
      "epoch": 33.06,
      "grad_norm": 2.6437411308288574,
      "learning_rate": 4.519158260247897e-06,
      "loss": 0.5504,
      "step": 80770
    },
    {
      "epoch": 33.07,
      "grad_norm": 2.5579655170440674,
      "learning_rate": 4.519099221907656e-06,
      "loss": 0.5829,
      "step": 80780
    },
    {
      "epoch": 33.07,
      "grad_norm": 2.3179819583892822,
      "learning_rate": 4.519040173602476e-06,
      "loss": 0.5569,
      "step": 80790
    },
    {
      "epoch": 33.07,
      "grad_norm": 2.010173797607422,
      "learning_rate": 4.5189811153326295e-06,
      "loss": 0.5618,
      "step": 80800
    },
    {
      "epoch": 33.08,
      "grad_norm": 2.6635115146636963,
      "learning_rate": 4.518922047098386e-06,
      "loss": 0.5605,
      "step": 80810
    },
    {
      "epoch": 33.08,
      "grad_norm": 2.5018715858459473,
      "learning_rate": 4.518862968900016e-06,
      "loss": 0.5977,
      "step": 80820
    },
    {
      "epoch": 33.09,
      "grad_norm": 2.645136833190918,
      "learning_rate": 4.51880388073779e-06,
      "loss": 0.5664,
      "step": 80830
    },
    {
      "epoch": 33.09,
      "grad_norm": 2.4854257106781006,
      "learning_rate": 4.518744782611979e-06,
      "loss": 0.5437,
      "step": 80840
    },
    {
      "epoch": 33.09,
      "grad_norm": 2.4340009689331055,
      "learning_rate": 4.518685674522855e-06,
      "loss": 0.584,
      "step": 80850
    },
    {
      "epoch": 33.1,
      "grad_norm": 2.3600847721099854,
      "learning_rate": 4.518626556470686e-06,
      "loss": 0.5601,
      "step": 80860
    },
    {
      "epoch": 33.1,
      "grad_norm": 2.5373964309692383,
      "learning_rate": 4.518567428455745e-06,
      "loss": 0.5531,
      "step": 80870
    },
    {
      "epoch": 33.11,
      "grad_norm": 2.1029789447784424,
      "learning_rate": 4.518508290478302e-06,
      "loss": 0.5788,
      "step": 80880
    },
    {
      "epoch": 33.11,
      "grad_norm": 1.872618556022644,
      "learning_rate": 4.518449142538628e-06,
      "loss": 0.5822,
      "step": 80890
    },
    {
      "epoch": 33.12,
      "grad_norm": 2.0734684467315674,
      "learning_rate": 4.518389984636994e-06,
      "loss": 0.5716,
      "step": 80900
    },
    {
      "epoch": 33.12,
      "grad_norm": 2.2746524810791016,
      "learning_rate": 4.518330816773672e-06,
      "loss": 0.5585,
      "step": 80910
    },
    {
      "epoch": 33.12,
      "grad_norm": 2.6297950744628906,
      "learning_rate": 4.51827163894893e-06,
      "loss": 0.583,
      "step": 80920
    },
    {
      "epoch": 33.13,
      "grad_norm": 1.9041601419448853,
      "learning_rate": 4.5182124511630426e-06,
      "loss": 0.5662,
      "step": 80930
    },
    {
      "epoch": 33.13,
      "grad_norm": 2.453462839126587,
      "learning_rate": 4.518153253416279e-06,
      "loss": 0.5689,
      "step": 80940
    },
    {
      "epoch": 33.14,
      "grad_norm": 2.7026257514953613,
      "learning_rate": 4.518094045708912e-06,
      "loss": 0.5545,
      "step": 80950
    },
    {
      "epoch": 33.14,
      "grad_norm": 2.3178791999816895,
      "learning_rate": 4.51803482804121e-06,
      "loss": 0.5713,
      "step": 80960
    },
    {
      "epoch": 33.14,
      "grad_norm": 2.2219622135162354,
      "learning_rate": 4.517975600413447e-06,
      "loss": 0.5849,
      "step": 80970
    },
    {
      "epoch": 33.15,
      "grad_norm": 1.6573436260223389,
      "learning_rate": 4.517916362825892e-06,
      "loss": 0.5633,
      "step": 80980
    },
    {
      "epoch": 33.15,
      "grad_norm": 1.539802074432373,
      "learning_rate": 4.5178571152788195e-06,
      "loss": 0.5888,
      "step": 80990
    },
    {
      "epoch": 33.16,
      "grad_norm": 2.5450334548950195,
      "learning_rate": 4.517797857772499e-06,
      "loss": 0.6018,
      "step": 81000
    },
    {
      "epoch": 33.16,
      "grad_norm": 1.8975216150283813,
      "learning_rate": 4.5177385903072e-06,
      "loss": 0.5998,
      "step": 81010
    },
    {
      "epoch": 33.16,
      "grad_norm": 1.8749902248382568,
      "learning_rate": 4.5176793128831965e-06,
      "loss": 0.5542,
      "step": 81020
    },
    {
      "epoch": 33.17,
      "grad_norm": 1.7131067514419556,
      "learning_rate": 4.517620025500761e-06,
      "loss": 0.552,
      "step": 81030
    },
    {
      "epoch": 33.17,
      "grad_norm": 3.078540802001953,
      "learning_rate": 4.517560728160162e-06,
      "loss": 0.5671,
      "step": 81040
    },
    {
      "epoch": 33.18,
      "grad_norm": 2.0964372158050537,
      "learning_rate": 4.517501420861673e-06,
      "loss": 0.5945,
      "step": 81050
    },
    {
      "epoch": 33.18,
      "grad_norm": 2.1396710872650146,
      "learning_rate": 4.517442103605566e-06,
      "loss": 0.5523,
      "step": 81060
    },
    {
      "epoch": 33.18,
      "grad_norm": 2.1230363845825195,
      "learning_rate": 4.517382776392111e-06,
      "loss": 0.588,
      "step": 81070
    },
    {
      "epoch": 33.19,
      "grad_norm": 1.9607888460159302,
      "learning_rate": 4.517323439221582e-06,
      "loss": 0.5735,
      "step": 81080
    },
    {
      "epoch": 33.19,
      "grad_norm": 2.479729413986206,
      "learning_rate": 4.517264092094249e-06,
      "loss": 0.5602,
      "step": 81090
    },
    {
      "epoch": 33.2,
      "grad_norm": 2.4987478256225586,
      "learning_rate": 4.517204735010385e-06,
      "loss": 0.5565,
      "step": 81100
    },
    {
      "epoch": 33.2,
      "grad_norm": 2.6253974437713623,
      "learning_rate": 4.517145367970261e-06,
      "loss": 0.5684,
      "step": 81110
    },
    {
      "epoch": 33.21,
      "grad_norm": 2.1521334648132324,
      "learning_rate": 4.51708599097415e-06,
      "loss": 0.5773,
      "step": 81120
    },
    {
      "epoch": 33.21,
      "grad_norm": 2.9170148372650146,
      "learning_rate": 4.517026604022323e-06,
      "loss": 0.5796,
      "step": 81130
    },
    {
      "epoch": 33.21,
      "grad_norm": 1.845820426940918,
      "learning_rate": 4.516967207115053e-06,
      "loss": 0.5608,
      "step": 81140
    },
    {
      "epoch": 33.22,
      "grad_norm": 2.1525168418884277,
      "learning_rate": 4.516907800252611e-06,
      "loss": 0.5987,
      "step": 81150
    },
    {
      "epoch": 33.22,
      "grad_norm": 1.894493818283081,
      "learning_rate": 4.5168483834352694e-06,
      "loss": 0.5742,
      "step": 81160
    },
    {
      "epoch": 33.23,
      "grad_norm": 2.1823794841766357,
      "learning_rate": 4.516788956663301e-06,
      "loss": 0.5803,
      "step": 81170
    },
    {
      "epoch": 33.23,
      "grad_norm": 2.0449066162109375,
      "learning_rate": 4.516729519936978e-06,
      "loss": 0.5645,
      "step": 81180
    },
    {
      "epoch": 33.23,
      "grad_norm": 2.3440184593200684,
      "learning_rate": 4.516670073256572e-06,
      "loss": 0.5616,
      "step": 81190
    },
    {
      "epoch": 33.24,
      "grad_norm": 2.3939270973205566,
      "learning_rate": 4.516610616622356e-06,
      "loss": 0.5504,
      "step": 81200
    },
    {
      "epoch": 33.24,
      "grad_norm": 2.462376832962036,
      "learning_rate": 4.5165511500346014e-06,
      "loss": 0.5696,
      "step": 81210
    },
    {
      "epoch": 33.25,
      "grad_norm": 2.3585712909698486,
      "learning_rate": 4.516491673493582e-06,
      "loss": 0.5608,
      "step": 81220
    },
    {
      "epoch": 33.25,
      "grad_norm": 2.3970634937286377,
      "learning_rate": 4.5164321869995685e-06,
      "loss": 0.5651,
      "step": 81230
    },
    {
      "epoch": 33.25,
      "grad_norm": 2.0794317722320557,
      "learning_rate": 4.516372690552835e-06,
      "loss": 0.5654,
      "step": 81240
    },
    {
      "epoch": 33.26,
      "grad_norm": 2.4707424640655518,
      "learning_rate": 4.516313184153653e-06,
      "loss": 0.5875,
      "step": 81250
    },
    {
      "epoch": 33.26,
      "grad_norm": 1.938523530960083,
      "learning_rate": 4.516253667802297e-06,
      "loss": 0.5582,
      "step": 81260
    },
    {
      "epoch": 33.27,
      "grad_norm": 1.968554139137268,
      "learning_rate": 4.516194141499036e-06,
      "loss": 0.572,
      "step": 81270
    },
    {
      "epoch": 33.27,
      "grad_norm": 2.609492063522339,
      "learning_rate": 4.516134605244146e-06,
      "loss": 0.5563,
      "step": 81280
    },
    {
      "epoch": 33.27,
      "grad_norm": 2.2817227840423584,
      "learning_rate": 4.516075059037899e-06,
      "loss": 0.572,
      "step": 81290
    },
    {
      "epoch": 33.28,
      "grad_norm": 2.847752094268799,
      "learning_rate": 4.516015502880567e-06,
      "loss": 0.5567,
      "step": 81300
    },
    {
      "epoch": 33.28,
      "grad_norm": 2.041821002960205,
      "learning_rate": 4.515955936772423e-06,
      "loss": 0.573,
      "step": 81310
    },
    {
      "epoch": 33.29,
      "grad_norm": 2.278549909591675,
      "learning_rate": 4.51589636071374e-06,
      "loss": 0.5503,
      "step": 81320
    },
    {
      "epoch": 33.29,
      "grad_norm": 2.1164395809173584,
      "learning_rate": 4.5158367747047905e-06,
      "loss": 0.5764,
      "step": 81330
    },
    {
      "epoch": 33.3,
      "grad_norm": 2.4244723320007324,
      "learning_rate": 4.515777178745848e-06,
      "loss": 0.5551,
      "step": 81340
    },
    {
      "epoch": 33.3,
      "grad_norm": 1.971292495727539,
      "learning_rate": 4.515717572837186e-06,
      "loss": 0.576,
      "step": 81350
    },
    {
      "epoch": 33.3,
      "grad_norm": 1.9673115015029907,
      "learning_rate": 4.515657956979076e-06,
      "loss": 0.5724,
      "step": 81360
    },
    {
      "epoch": 33.31,
      "grad_norm": 2.645352840423584,
      "learning_rate": 4.515598331171792e-06,
      "loss": 0.5967,
      "step": 81370
    },
    {
      "epoch": 33.31,
      "grad_norm": 2.449984550476074,
      "learning_rate": 4.515538695415608e-06,
      "loss": 0.5681,
      "step": 81380
    },
    {
      "epoch": 33.32,
      "grad_norm": 1.7042206525802612,
      "learning_rate": 4.515479049710797e-06,
      "loss": 0.5732,
      "step": 81390
    },
    {
      "epoch": 33.32,
      "grad_norm": 2.111052989959717,
      "learning_rate": 4.51541939405763e-06,
      "loss": 0.5706,
      "step": 81400
    },
    {
      "epoch": 33.32,
      "grad_norm": 2.6258761882781982,
      "learning_rate": 4.515359728456383e-06,
      "loss": 0.5709,
      "step": 81410
    },
    {
      "epoch": 33.33,
      "grad_norm": 1.7309129238128662,
      "learning_rate": 4.515300052907327e-06,
      "loss": 0.5855,
      "step": 81420
    },
    {
      "epoch": 33.33,
      "grad_norm": 2.111159563064575,
      "learning_rate": 4.515240367410737e-06,
      "loss": 0.5645,
      "step": 81430
    },
    {
      "epoch": 33.34,
      "grad_norm": 2.3877432346343994,
      "learning_rate": 4.515180671966886e-06,
      "loss": 0.565,
      "step": 81440
    },
    {
      "epoch": 33.34,
      "grad_norm": 2.136976957321167,
      "learning_rate": 4.515120966576048e-06,
      "loss": 0.5712,
      "step": 81450
    },
    {
      "epoch": 33.34,
      "grad_norm": 2.6504266262054443,
      "learning_rate": 4.515061251238494e-06,
      "loss": 0.5604,
      "step": 81460
    },
    {
      "epoch": 33.35,
      "grad_norm": 2.338526964187622,
      "learning_rate": 4.515001525954501e-06,
      "loss": 0.5869,
      "step": 81470
    },
    {
      "epoch": 33.35,
      "grad_norm": 2.354191303253174,
      "learning_rate": 4.514941790724341e-06,
      "loss": 0.585,
      "step": 81480
    },
    {
      "epoch": 33.36,
      "grad_norm": 2.0077426433563232,
      "learning_rate": 4.5148820455482876e-06,
      "loss": 0.5933,
      "step": 81490
    },
    {
      "epoch": 33.36,
      "grad_norm": 2.5718986988067627,
      "learning_rate": 4.514822290426615e-06,
      "loss": 0.5765,
      "step": 81500
    },
    {
      "epoch": 33.36,
      "grad_norm": 2.426896810531616,
      "learning_rate": 4.514762525359595e-06,
      "loss": 0.5755,
      "step": 81510
    },
    {
      "epoch": 33.37,
      "grad_norm": 2.2509565353393555,
      "learning_rate": 4.514702750347503e-06,
      "loss": 0.5796,
      "step": 81520
    },
    {
      "epoch": 33.37,
      "grad_norm": 2.4742467403411865,
      "learning_rate": 4.514642965390614e-06,
      "loss": 0.5773,
      "step": 81530
    },
    {
      "epoch": 33.38,
      "grad_norm": 2.353567361831665,
      "learning_rate": 4.514583170489199e-06,
      "loss": 0.559,
      "step": 81540
    },
    {
      "epoch": 33.38,
      "grad_norm": 2.439802885055542,
      "learning_rate": 4.5145233656435355e-06,
      "loss": 0.5651,
      "step": 81550
    },
    {
      "epoch": 33.39,
      "grad_norm": 2.6457223892211914,
      "learning_rate": 4.514463550853894e-06,
      "loss": 0.5785,
      "step": 81560
    },
    {
      "epoch": 33.39,
      "grad_norm": 2.2971527576446533,
      "learning_rate": 4.51440372612055e-06,
      "loss": 0.5705,
      "step": 81570
    },
    {
      "epoch": 33.39,
      "grad_norm": 2.3307735919952393,
      "learning_rate": 4.514343891443777e-06,
      "loss": 0.5857,
      "step": 81580
    },
    {
      "epoch": 33.4,
      "grad_norm": 2.4896035194396973,
      "learning_rate": 4.5142840468238506e-06,
      "loss": 0.5693,
      "step": 81590
    },
    {
      "epoch": 33.4,
      "grad_norm": 2.0443379878997803,
      "learning_rate": 4.5142241922610424e-06,
      "loss": 0.5892,
      "step": 81600
    },
    {
      "epoch": 33.41,
      "grad_norm": 2.394651412963867,
      "learning_rate": 4.514164327755629e-06,
      "loss": 0.561,
      "step": 81610
    },
    {
      "epoch": 33.41,
      "grad_norm": 2.3419198989868164,
      "learning_rate": 4.514104453307884e-06,
      "loss": 0.5342,
      "step": 81620
    },
    {
      "epoch": 33.41,
      "grad_norm": 2.192152976989746,
      "learning_rate": 4.5140445689180814e-06,
      "loss": 0.5687,
      "step": 81630
    },
    {
      "epoch": 33.42,
      "grad_norm": 2.3378171920776367,
      "learning_rate": 4.513984674586495e-06,
      "loss": 0.5665,
      "step": 81640
    },
    {
      "epoch": 33.42,
      "grad_norm": 2.5451247692108154,
      "learning_rate": 4.5139247703134005e-06,
      "loss": 0.5757,
      "step": 81650
    },
    {
      "epoch": 33.43,
      "grad_norm": 2.365722894668579,
      "learning_rate": 4.51386485609907e-06,
      "loss": 0.5546,
      "step": 81660
    },
    {
      "epoch": 33.43,
      "grad_norm": 1.7964543104171753,
      "learning_rate": 4.51380493194378e-06,
      "loss": 0.5873,
      "step": 81670
    },
    {
      "epoch": 33.43,
      "grad_norm": 1.5127311944961548,
      "learning_rate": 4.513744997847806e-06,
      "loss": 0.5658,
      "step": 81680
    },
    {
      "epoch": 33.44,
      "grad_norm": 1.7185274362564087,
      "learning_rate": 4.513685053811419e-06,
      "loss": 0.5686,
      "step": 81690
    },
    {
      "epoch": 33.44,
      "grad_norm": 2.581165075302124,
      "learning_rate": 4.513625099834896e-06,
      "loss": 0.5864,
      "step": 81700
    },
    {
      "epoch": 33.45,
      "grad_norm": 2.8554444313049316,
      "learning_rate": 4.513565135918511e-06,
      "loss": 0.5528,
      "step": 81710
    },
    {
      "epoch": 33.45,
      "grad_norm": 2.188807725906372,
      "learning_rate": 4.51350516206254e-06,
      "loss": 0.5672,
      "step": 81720
    },
    {
      "epoch": 33.45,
      "grad_norm": 2.252811908721924,
      "learning_rate": 4.513445178267255e-06,
      "loss": 0.5639,
      "step": 81730
    },
    {
      "epoch": 33.46,
      "grad_norm": 2.4398365020751953,
      "learning_rate": 4.513385184532934e-06,
      "loss": 0.5577,
      "step": 81740
    },
    {
      "epoch": 33.46,
      "grad_norm": 2.5340981483459473,
      "learning_rate": 4.51332518085985e-06,
      "loss": 0.5775,
      "step": 81750
    },
    {
      "epoch": 33.47,
      "grad_norm": 2.322295904159546,
      "learning_rate": 4.513265167248277e-06,
      "loss": 0.5854,
      "step": 81760
    },
    {
      "epoch": 33.47,
      "grad_norm": 2.104288101196289,
      "learning_rate": 4.513205143698492e-06,
      "loss": 0.5752,
      "step": 81770
    },
    {
      "epoch": 33.48,
      "grad_norm": 2.2933120727539062,
      "learning_rate": 4.5131451102107685e-06,
      "loss": 0.5726,
      "step": 81780
    },
    {
      "epoch": 33.48,
      "grad_norm": 3.096698760986328,
      "learning_rate": 4.513085066785382e-06,
      "loss": 0.5644,
      "step": 81790
    },
    {
      "epoch": 33.48,
      "grad_norm": 2.43894100189209,
      "learning_rate": 4.513025013422608e-06,
      "loss": 0.557,
      "step": 81800
    },
    {
      "epoch": 33.49,
      "grad_norm": 2.065760374069214,
      "learning_rate": 4.512964950122721e-06,
      "loss": 0.5761,
      "step": 81810
    },
    {
      "epoch": 33.49,
      "grad_norm": 3.004152774810791,
      "learning_rate": 4.5129048768859955e-06,
      "loss": 0.5575,
      "step": 81820
    },
    {
      "epoch": 33.5,
      "grad_norm": 1.9976308345794678,
      "learning_rate": 4.512844793712708e-06,
      "loss": 0.5713,
      "step": 81830
    },
    {
      "epoch": 33.5,
      "grad_norm": 2.41913104057312,
      "learning_rate": 4.512784700603133e-06,
      "loss": 0.5587,
      "step": 81840
    },
    {
      "epoch": 33.5,
      "grad_norm": 2.4399452209472656,
      "learning_rate": 4.512724597557547e-06,
      "loss": 0.5722,
      "step": 81850
    },
    {
      "epoch": 33.51,
      "grad_norm": 2.5883193016052246,
      "learning_rate": 4.512664484576223e-06,
      "loss": 0.5505,
      "step": 81860
    },
    {
      "epoch": 33.51,
      "grad_norm": 1.6997798681259155,
      "learning_rate": 4.512604361659439e-06,
      "loss": 0.5529,
      "step": 81870
    },
    {
      "epoch": 33.52,
      "grad_norm": 1.6762553453445435,
      "learning_rate": 4.5125442288074675e-06,
      "loss": 0.5597,
      "step": 81880
    },
    {
      "epoch": 33.52,
      "grad_norm": 2.398606061935425,
      "learning_rate": 4.512484086020586e-06,
      "loss": 0.5902,
      "step": 81890
    },
    {
      "epoch": 33.52,
      "grad_norm": 1.9740405082702637,
      "learning_rate": 4.512423933299069e-06,
      "loss": 0.5537,
      "step": 81900
    },
    {
      "epoch": 33.53,
      "grad_norm": 2.5529041290283203,
      "learning_rate": 4.5123637706431935e-06,
      "loss": 0.5618,
      "step": 81910
    },
    {
      "epoch": 33.53,
      "grad_norm": 2.7212488651275635,
      "learning_rate": 4.512303598053234e-06,
      "loss": 0.5742,
      "step": 81920
    },
    {
      "epoch": 33.54,
      "grad_norm": 2.1318371295928955,
      "learning_rate": 4.512243415529466e-06,
      "loss": 0.601,
      "step": 81930
    },
    {
      "epoch": 33.54,
      "grad_norm": 2.3871748447418213,
      "learning_rate": 4.512183223072166e-06,
      "loss": 0.5591,
      "step": 81940
    },
    {
      "epoch": 33.54,
      "grad_norm": 2.6808021068573,
      "learning_rate": 4.512123020681609e-06,
      "loss": 0.5441,
      "step": 81950
    },
    {
      "epoch": 33.55,
      "grad_norm": 2.579111337661743,
      "learning_rate": 4.512062808358071e-06,
      "loss": 0.5838,
      "step": 81960
    },
    {
      "epoch": 33.55,
      "grad_norm": 2.435392379760742,
      "learning_rate": 4.512002586101828e-06,
      "loss": 0.5725,
      "step": 81970
    },
    {
      "epoch": 33.56,
      "grad_norm": 2.349623680114746,
      "learning_rate": 4.511942353913155e-06,
      "loss": 0.5924,
      "step": 81980
    },
    {
      "epoch": 33.56,
      "grad_norm": 3.0131008625030518,
      "learning_rate": 4.511882111792329e-06,
      "loss": 0.576,
      "step": 81990
    },
    {
      "epoch": 33.57,
      "grad_norm": 2.902272939682007,
      "learning_rate": 4.511821859739626e-06,
      "loss": 0.5676,
      "step": 82000
    },
    {
      "epoch": 33.57,
      "grad_norm": 2.594003677368164,
      "learning_rate": 4.511761597755321e-06,
      "loss": 0.5689,
      "step": 82010
    },
    {
      "epoch": 33.57,
      "grad_norm": 2.2812066078186035,
      "learning_rate": 4.511701325839691e-06,
      "loss": 0.5618,
      "step": 82020
    },
    {
      "epoch": 33.58,
      "grad_norm": 2.006598472595215,
      "learning_rate": 4.511641043993011e-06,
      "loss": 0.573,
      "step": 82030
    },
    {
      "epoch": 33.58,
      "grad_norm": 2.8231449127197266,
      "learning_rate": 4.511580752215559e-06,
      "loss": 0.558,
      "step": 82040
    },
    {
      "epoch": 33.59,
      "grad_norm": 1.7393100261688232,
      "learning_rate": 4.511520450507609e-06,
      "loss": 0.5529,
      "step": 82050
    },
    {
      "epoch": 33.59,
      "grad_norm": 2.061246395111084,
      "learning_rate": 4.511460138869439e-06,
      "loss": 0.5742,
      "step": 82060
    },
    {
      "epoch": 33.59,
      "grad_norm": 1.5852147340774536,
      "learning_rate": 4.5113998173013245e-06,
      "loss": 0.589,
      "step": 82070
    },
    {
      "epoch": 33.6,
      "grad_norm": 2.682499408721924,
      "learning_rate": 4.5113394858035415e-06,
      "loss": 0.5637,
      "step": 82080
    },
    {
      "epoch": 33.6,
      "grad_norm": 2.932990550994873,
      "learning_rate": 4.5112791443763664e-06,
      "loss": 0.5632,
      "step": 82090
    },
    {
      "epoch": 33.61,
      "grad_norm": 3.185204029083252,
      "learning_rate": 4.511218793020077e-06,
      "loss": 0.568,
      "step": 82100
    },
    {
      "epoch": 33.61,
      "grad_norm": 2.5690176486968994,
      "learning_rate": 4.511158431734948e-06,
      "loss": 0.5726,
      "step": 82110
    },
    {
      "epoch": 33.61,
      "grad_norm": 2.541593313217163,
      "learning_rate": 4.511098060521257e-06,
      "loss": 0.569,
      "step": 82120
    },
    {
      "epoch": 33.62,
      "grad_norm": 2.020610809326172,
      "learning_rate": 4.511037679379281e-06,
      "loss": 0.5524,
      "step": 82130
    },
    {
      "epoch": 33.62,
      "grad_norm": 2.1347787380218506,
      "learning_rate": 4.510977288309294e-06,
      "loss": 0.551,
      "step": 82140
    },
    {
      "epoch": 33.63,
      "grad_norm": 2.097787857055664,
      "learning_rate": 4.510916887311575e-06,
      "loss": 0.5626,
      "step": 82150
    },
    {
      "epoch": 33.63,
      "grad_norm": 1.7551569938659668,
      "learning_rate": 4.5108564763864016e-06,
      "loss": 0.5545,
      "step": 82160
    },
    {
      "epoch": 33.63,
      "grad_norm": 2.4950900077819824,
      "learning_rate": 4.510796055534047e-06,
      "loss": 0.566,
      "step": 82170
    },
    {
      "epoch": 33.64,
      "grad_norm": 2.0767152309417725,
      "learning_rate": 4.510735624754791e-06,
      "loss": 0.566,
      "step": 82180
    },
    {
      "epoch": 33.64,
      "grad_norm": 2.2506401538848877,
      "learning_rate": 4.51067518404891e-06,
      "loss": 0.5639,
      "step": 82190
    },
    {
      "epoch": 33.65,
      "grad_norm": 1.688443899154663,
      "learning_rate": 4.51061473341668e-06,
      "loss": 0.5806,
      "step": 82200
    },
    {
      "epoch": 33.65,
      "grad_norm": 2.277127265930176,
      "learning_rate": 4.510554272858378e-06,
      "loss": 0.578,
      "step": 82210
    },
    {
      "epoch": 33.66,
      "grad_norm": 2.9572761058807373,
      "learning_rate": 4.510493802374281e-06,
      "loss": 0.5694,
      "step": 82220
    },
    {
      "epoch": 33.66,
      "grad_norm": 2.0723259449005127,
      "learning_rate": 4.510433321964666e-06,
      "loss": 0.5914,
      "step": 82230
    },
    {
      "epoch": 33.66,
      "grad_norm": 2.2896673679351807,
      "learning_rate": 4.510372831629811e-06,
      "loss": 0.5397,
      "step": 82240
    },
    {
      "epoch": 33.67,
      "grad_norm": 2.3916149139404297,
      "learning_rate": 4.510312331369991e-06,
      "loss": 0.5495,
      "step": 82250
    },
    {
      "epoch": 33.67,
      "grad_norm": 2.51042103767395,
      "learning_rate": 4.510251821185486e-06,
      "loss": 0.5773,
      "step": 82260
    },
    {
      "epoch": 33.68,
      "grad_norm": 1.9283647537231445,
      "learning_rate": 4.510191301076571e-06,
      "loss": 0.5488,
      "step": 82270
    },
    {
      "epoch": 33.68,
      "grad_norm": 2.693887710571289,
      "learning_rate": 4.510130771043524e-06,
      "loss": 0.5597,
      "step": 82280
    },
    {
      "epoch": 33.68,
      "grad_norm": 2.668839931488037,
      "learning_rate": 4.510070231086623e-06,
      "loss": 0.5417,
      "step": 82290
    },
    {
      "epoch": 33.69,
      "grad_norm": 2.106733798980713,
      "learning_rate": 4.510009681206143e-06,
      "loss": 0.5918,
      "step": 82300
    },
    {
      "epoch": 33.69,
      "grad_norm": 1.8615902662277222,
      "learning_rate": 4.509949121402365e-06,
      "loss": 0.5725,
      "step": 82310
    },
    {
      "epoch": 33.7,
      "grad_norm": 1.4988443851470947,
      "learning_rate": 4.5098885516755624e-06,
      "loss": 0.5628,
      "step": 82320
    },
    {
      "epoch": 33.7,
      "grad_norm": 1.862416386604309,
      "learning_rate": 4.509827972026015e-06,
      "loss": 0.5919,
      "step": 82330
    },
    {
      "epoch": 33.7,
      "grad_norm": 2.0885980129241943,
      "learning_rate": 4.5097673824539995e-06,
      "loss": 0.5574,
      "step": 82340
    },
    {
      "epoch": 33.71,
      "grad_norm": 1.8583736419677734,
      "learning_rate": 4.509706782959795e-06,
      "loss": 0.5601,
      "step": 82350
    },
    {
      "epoch": 33.71,
      "grad_norm": 2.3543968200683594,
      "learning_rate": 4.509646173543677e-06,
      "loss": 0.5552,
      "step": 82360
    },
    {
      "epoch": 33.72,
      "grad_norm": 2.437808036804199,
      "learning_rate": 4.509585554205924e-06,
      "loss": 0.583,
      "step": 82370
    },
    {
      "epoch": 33.72,
      "grad_norm": 2.2807250022888184,
      "learning_rate": 4.509524924946814e-06,
      "loss": 0.5693,
      "step": 82380
    },
    {
      "epoch": 33.72,
      "grad_norm": 2.4007956981658936,
      "learning_rate": 4.509464285766625e-06,
      "loss": 0.5806,
      "step": 82390
    },
    {
      "epoch": 33.73,
      "grad_norm": 2.767054557800293,
      "learning_rate": 4.509403636665634e-06,
      "loss": 0.5683,
      "step": 82400
    },
    {
      "epoch": 33.73,
      "grad_norm": 2.365220546722412,
      "learning_rate": 4.509342977644118e-06,
      "loss": 0.5477,
      "step": 82410
    },
    {
      "epoch": 33.74,
      "grad_norm": 2.932762384414673,
      "learning_rate": 4.509282308702357e-06,
      "loss": 0.5801,
      "step": 82420
    },
    {
      "epoch": 33.74,
      "grad_norm": 2.6474812030792236,
      "learning_rate": 4.509221629840628e-06,
      "loss": 0.5652,
      "step": 82430
    },
    {
      "epoch": 33.75,
      "grad_norm": 2.9426376819610596,
      "learning_rate": 4.509160941059208e-06,
      "loss": 0.5443,
      "step": 82440
    },
    {
      "epoch": 33.75,
      "grad_norm": 2.4494853019714355,
      "learning_rate": 4.509100242358377e-06,
      "loss": 0.5668,
      "step": 82450
    },
    {
      "epoch": 33.75,
      "grad_norm": 1.8569635152816772,
      "learning_rate": 4.509039533738411e-06,
      "loss": 0.5926,
      "step": 82460
    },
    {
      "epoch": 33.76,
      "grad_norm": 1.7504140138626099,
      "learning_rate": 4.508978815199589e-06,
      "loss": 0.5807,
      "step": 82470
    },
    {
      "epoch": 33.76,
      "grad_norm": 2.490365982055664,
      "learning_rate": 4.50891808674219e-06,
      "loss": 0.5589,
      "step": 82480
    },
    {
      "epoch": 33.77,
      "grad_norm": 1.989537000656128,
      "learning_rate": 4.50885734836649e-06,
      "loss": 0.5634,
      "step": 82490
    },
    {
      "epoch": 33.77,
      "grad_norm": 3.030958652496338,
      "learning_rate": 4.50879660007277e-06,
      "loss": 0.5651,
      "step": 82500
    },
    {
      "epoch": 33.77,
      "grad_norm": 2.6214334964752197,
      "learning_rate": 4.508735841861306e-06,
      "loss": 0.6044,
      "step": 82510
    },
    {
      "epoch": 33.78,
      "grad_norm": 2.030789852142334,
      "learning_rate": 4.508675073732377e-06,
      "loss": 0.5518,
      "step": 82520
    },
    {
      "epoch": 33.78,
      "grad_norm": 2.533996343612671,
      "learning_rate": 4.508614295686261e-06,
      "loss": 0.5482,
      "step": 82530
    },
    {
      "epoch": 33.79,
      "grad_norm": 1.9699816703796387,
      "learning_rate": 4.508553507723238e-06,
      "loss": 0.5707,
      "step": 82540
    },
    {
      "epoch": 33.79,
      "grad_norm": 2.702392578125,
      "learning_rate": 4.508492709843585e-06,
      "loss": 0.5456,
      "step": 82550
    },
    {
      "epoch": 33.79,
      "grad_norm": 2.3848702907562256,
      "learning_rate": 4.508431902047581e-06,
      "loss": 0.5589,
      "step": 82560
    },
    {
      "epoch": 33.8,
      "grad_norm": 2.4658002853393555,
      "learning_rate": 4.508371084335504e-06,
      "loss": 0.5523,
      "step": 82570
    },
    {
      "epoch": 33.8,
      "grad_norm": 1.8516641855239868,
      "learning_rate": 4.508310256707634e-06,
      "loss": 0.5724,
      "step": 82580
    },
    {
      "epoch": 33.81,
      "grad_norm": 2.732351064682007,
      "learning_rate": 4.508249419164248e-06,
      "loss": 0.5842,
      "step": 82590
    },
    {
      "epoch": 33.81,
      "grad_norm": 1.774175763130188,
      "learning_rate": 4.5081885717056255e-06,
      "loss": 0.5615,
      "step": 82600
    },
    {
      "epoch": 33.81,
      "grad_norm": 2.311005115509033,
      "learning_rate": 4.508127714332045e-06,
      "loss": 0.5652,
      "step": 82610
    },
    {
      "epoch": 33.82,
      "grad_norm": 2.8478755950927734,
      "learning_rate": 4.5080668470437855e-06,
      "loss": 0.5662,
      "step": 82620
    },
    {
      "epoch": 33.82,
      "grad_norm": 2.6949074268341064,
      "learning_rate": 4.5080059698411256e-06,
      "loss": 0.5791,
      "step": 82630
    },
    {
      "epoch": 33.83,
      "grad_norm": 2.278404951095581,
      "learning_rate": 4.507945082724345e-06,
      "loss": 0.569,
      "step": 82640
    },
    {
      "epoch": 33.83,
      "grad_norm": 3.194368600845337,
      "learning_rate": 4.50788418569372e-06,
      "loss": 0.5612,
      "step": 82650
    },
    {
      "epoch": 33.84,
      "grad_norm": 5.663681983947754,
      "learning_rate": 4.507823278749533e-06,
      "loss": 0.563,
      "step": 82660
    },
    {
      "epoch": 33.84,
      "grad_norm": 2.325523614883423,
      "learning_rate": 4.507762361892062e-06,
      "loss": 0.5792,
      "step": 82670
    },
    {
      "epoch": 33.84,
      "grad_norm": 2.3285276889801025,
      "learning_rate": 4.507701435121584e-06,
      "loss": 0.587,
      "step": 82680
    },
    {
      "epoch": 33.85,
      "grad_norm": 2.419893264770508,
      "learning_rate": 4.50764049843838e-06,
      "loss": 0.582,
      "step": 82690
    },
    {
      "epoch": 33.85,
      "grad_norm": 2.2894582748413086,
      "learning_rate": 4.50757955184273e-06,
      "loss": 0.5665,
      "step": 82700
    },
    {
      "epoch": 33.86,
      "grad_norm": 1.9209036827087402,
      "learning_rate": 4.50751859533491e-06,
      "loss": 0.5966,
      "step": 82710
    },
    {
      "epoch": 33.86,
      "grad_norm": 2.1715691089630127,
      "learning_rate": 4.507457628915203e-06,
      "loss": 0.5667,
      "step": 82720
    },
    {
      "epoch": 33.86,
      "grad_norm": 2.495070219039917,
      "learning_rate": 4.507396652583885e-06,
      "loss": 0.575,
      "step": 82730
    },
    {
      "epoch": 33.87,
      "grad_norm": 2.0476386547088623,
      "learning_rate": 4.507335666341238e-06,
      "loss": 0.5641,
      "step": 82740
    },
    {
      "epoch": 33.87,
      "grad_norm": 2.4762654304504395,
      "learning_rate": 4.507274670187539e-06,
      "loss": 0.561,
      "step": 82750
    },
    {
      "epoch": 33.88,
      "grad_norm": 2.4965293407440186,
      "learning_rate": 4.507213664123069e-06,
      "loss": 0.5892,
      "step": 82760
    },
    {
      "epoch": 33.88,
      "grad_norm": 2.25319766998291,
      "learning_rate": 4.507152648148107e-06,
      "loss": 0.5549,
      "step": 82770
    },
    {
      "epoch": 33.88,
      "grad_norm": 2.274531602859497,
      "learning_rate": 4.5070916222629325e-06,
      "loss": 0.5774,
      "step": 82780
    },
    {
      "epoch": 33.89,
      "grad_norm": 1.760613203048706,
      "learning_rate": 4.507030586467826e-06,
      "loss": 0.5843,
      "step": 82790
    },
    {
      "epoch": 33.89,
      "grad_norm": 1.8340805768966675,
      "learning_rate": 4.506969540763065e-06,
      "loss": 0.5595,
      "step": 82800
    },
    {
      "epoch": 33.9,
      "grad_norm": 2.7518649101257324,
      "learning_rate": 4.506908485148931e-06,
      "loss": 0.5841,
      "step": 82810
    },
    {
      "epoch": 33.9,
      "grad_norm": 3.105833053588867,
      "learning_rate": 4.5068474196257015e-06,
      "loss": 0.553,
      "step": 82820
    },
    {
      "epoch": 33.91,
      "grad_norm": 3.0679125785827637,
      "learning_rate": 4.506786344193659e-06,
      "loss": 0.5604,
      "step": 82830
    },
    {
      "epoch": 33.91,
      "grad_norm": 2.2097842693328857,
      "learning_rate": 4.506725258853082e-06,
      "loss": 0.5763,
      "step": 82840
    },
    {
      "epoch": 33.91,
      "grad_norm": 2.289839267730713,
      "learning_rate": 4.5066641636042505e-06,
      "loss": 0.5746,
      "step": 82850
    },
    {
      "epoch": 33.92,
      "grad_norm": 2.3013627529144287,
      "learning_rate": 4.506603058447443e-06,
      "loss": 0.5701,
      "step": 82860
    },
    {
      "epoch": 33.92,
      "grad_norm": 2.8291964530944824,
      "learning_rate": 4.506541943382942e-06,
      "loss": 0.5609,
      "step": 82870
    },
    {
      "epoch": 33.93,
      "grad_norm": 2.4038209915161133,
      "learning_rate": 4.506480818411025e-06,
      "loss": 0.5683,
      "step": 82880
    },
    {
      "epoch": 33.93,
      "grad_norm": 2.0311453342437744,
      "learning_rate": 4.506419683531973e-06,
      "loss": 0.5639,
      "step": 82890
    },
    {
      "epoch": 33.93,
      "grad_norm": 2.027108907699585,
      "learning_rate": 4.506358538746067e-06,
      "loss": 0.5671,
      "step": 82900
    },
    {
      "epoch": 33.94,
      "grad_norm": 2.6042404174804688,
      "learning_rate": 4.506297384053586e-06,
      "loss": 0.5561,
      "step": 82910
    },
    {
      "epoch": 33.94,
      "grad_norm": 1.6538764238357544,
      "learning_rate": 4.5062362194548105e-06,
      "loss": 0.5517,
      "step": 82920
    },
    {
      "epoch": 33.95,
      "grad_norm": 2.4133493900299072,
      "learning_rate": 4.50617504495002e-06,
      "loss": 0.5841,
      "step": 82930
    },
    {
      "epoch": 33.95,
      "grad_norm": 2.2795188426971436,
      "learning_rate": 4.5061138605394954e-06,
      "loss": 0.5532,
      "step": 82940
    },
    {
      "epoch": 33.95,
      "grad_norm": 2.797187089920044,
      "learning_rate": 4.506052666223517e-06,
      "loss": 0.5885,
      "step": 82950
    },
    {
      "epoch": 33.96,
      "grad_norm": 1.985876202583313,
      "learning_rate": 4.505991462002365e-06,
      "loss": 0.5671,
      "step": 82960
    },
    {
      "epoch": 33.96,
      "grad_norm": 2.1521694660186768,
      "learning_rate": 4.505930247876319e-06,
      "loss": 0.546,
      "step": 82970
    },
    {
      "epoch": 33.97,
      "grad_norm": 1.8064290285110474,
      "learning_rate": 4.505869023845661e-06,
      "loss": 0.587,
      "step": 82980
    },
    {
      "epoch": 33.97,
      "grad_norm": 3.7523739337921143,
      "learning_rate": 4.505807789910671e-06,
      "loss": 0.5634,
      "step": 82990
    },
    {
      "epoch": 33.97,
      "grad_norm": 1.5777552127838135,
      "learning_rate": 4.5057465460716285e-06,
      "loss": 0.5576,
      "step": 83000
    },
    {
      "epoch": 33.98,
      "grad_norm": 3.3032569885253906,
      "learning_rate": 4.5056852923288134e-06,
      "loss": 0.5918,
      "step": 83010
    },
    {
      "epoch": 33.98,
      "grad_norm": 2.343996524810791,
      "learning_rate": 4.5056240286825095e-06,
      "loss": 0.5596,
      "step": 83020
    },
    {
      "epoch": 33.99,
      "grad_norm": 2.163341760635376,
      "learning_rate": 4.5055627551329945e-06,
      "loss": 0.5677,
      "step": 83030
    },
    {
      "epoch": 33.99,
      "grad_norm": 1.9412665367126465,
      "learning_rate": 4.5055014716805506e-06,
      "loss": 0.5826,
      "step": 83040
    },
    {
      "epoch": 34.0,
      "grad_norm": 2.6160452365875244,
      "learning_rate": 4.505440178325457e-06,
      "loss": 0.5818,
      "step": 83050
    },
    {
      "epoch": 34.0,
      "grad_norm": 2.2312357425689697,
      "learning_rate": 4.505378875067997e-06,
      "loss": 0.5578,
      "step": 83060
    },
    {
      "epoch": 34.0,
      "eval_loss": 0.5629244446754456,
      "eval_runtime": 51.9993,
      "eval_samples_per_second": 66.328,
      "eval_steps_per_second": 8.308,
      "step": 83062
    },
    {
      "epoch": 34.0,
      "grad_norm": 2.084984064102173,
      "learning_rate": 4.505317561908448e-06,
      "loss": 0.5844,
      "step": 83070
    },
    {
      "epoch": 34.01,
      "grad_norm": 2.4533538818359375,
      "learning_rate": 4.5052562388470945e-06,
      "loss": 0.567,
      "step": 83080
    },
    {
      "epoch": 34.01,
      "grad_norm": 2.196040630340576,
      "learning_rate": 4.5051949058842144e-06,
      "loss": 0.5554,
      "step": 83090
    },
    {
      "epoch": 34.02,
      "grad_norm": 3.0541791915893555,
      "learning_rate": 4.505133563020091e-06,
      "loss": 0.56,
      "step": 83100
    },
    {
      "epoch": 34.02,
      "grad_norm": 2.200789451599121,
      "learning_rate": 4.5050722102550036e-06,
      "loss": 0.5872,
      "step": 83110
    },
    {
      "epoch": 34.02,
      "grad_norm": 2.449901580810547,
      "learning_rate": 4.505010847589234e-06,
      "loss": 0.5636,
      "step": 83120
    },
    {
      "epoch": 34.03,
      "grad_norm": 2.306931257247925,
      "learning_rate": 4.504949475023064e-06,
      "loss": 0.5662,
      "step": 83130
    },
    {
      "epoch": 34.03,
      "grad_norm": 2.7570338249206543,
      "learning_rate": 4.504888092556772e-06,
      "loss": 0.5695,
      "step": 83140
    },
    {
      "epoch": 34.04,
      "grad_norm": 2.0866804122924805,
      "learning_rate": 4.504826700190643e-06,
      "loss": 0.569,
      "step": 83150
    },
    {
      "epoch": 34.04,
      "grad_norm": 2.0620415210723877,
      "learning_rate": 4.504765297924955e-06,
      "loss": 0.5571,
      "step": 83160
    },
    {
      "epoch": 34.04,
      "grad_norm": 1.9383130073547363,
      "learning_rate": 4.504703885759992e-06,
      "loss": 0.5595,
      "step": 83170
    },
    {
      "epoch": 34.05,
      "grad_norm": 2.0723607540130615,
      "learning_rate": 4.504642463696033e-06,
      "loss": 0.5709,
      "step": 83180
    },
    {
      "epoch": 34.05,
      "grad_norm": 2.689443588256836,
      "learning_rate": 4.50458103173336e-06,
      "loss": 0.569,
      "step": 83190
    },
    {
      "epoch": 34.06,
      "grad_norm": 2.4644534587860107,
      "learning_rate": 4.504519589872255e-06,
      "loss": 0.5789,
      "step": 83200
    },
    {
      "epoch": 34.06,
      "grad_norm": 2.283989191055298,
      "learning_rate": 4.504458138112999e-06,
      "loss": 0.5827,
      "step": 83210
    },
    {
      "epoch": 34.06,
      "grad_norm": 1.9574998617172241,
      "learning_rate": 4.504396676455874e-06,
      "loss": 0.5842,
      "step": 83220
    },
    {
      "epoch": 34.07,
      "grad_norm": 3.020861864089966,
      "learning_rate": 4.5043352049011605e-06,
      "loss": 0.5643,
      "step": 83230
    },
    {
      "epoch": 34.07,
      "grad_norm": 2.5475962162017822,
      "learning_rate": 4.504273723449142e-06,
      "loss": 0.5765,
      "step": 83240
    },
    {
      "epoch": 34.08,
      "grad_norm": 3.2815277576446533,
      "learning_rate": 4.504212232100098e-06,
      "loss": 0.5784,
      "step": 83250
    },
    {
      "epoch": 34.08,
      "grad_norm": 3.042719841003418,
      "learning_rate": 4.5041507308543104e-06,
      "loss": 0.5956,
      "step": 83260
    },
    {
      "epoch": 34.09,
      "grad_norm": 2.6575684547424316,
      "learning_rate": 4.504089219712063e-06,
      "loss": 0.5632,
      "step": 83270
    },
    {
      "epoch": 34.09,
      "grad_norm": 2.150094985961914,
      "learning_rate": 4.504027698673636e-06,
      "loss": 0.5541,
      "step": 83280
    },
    {
      "epoch": 34.09,
      "grad_norm": 2.148131847381592,
      "learning_rate": 4.503966167739311e-06,
      "loss": 0.5555,
      "step": 83290
    },
    {
      "epoch": 34.1,
      "grad_norm": 2.5778379440307617,
      "learning_rate": 4.50390462690937e-06,
      "loss": 0.5676,
      "step": 83300
    },
    {
      "epoch": 34.1,
      "grad_norm": 2.5036025047302246,
      "learning_rate": 4.503843076184096e-06,
      "loss": 0.5765,
      "step": 83310
    },
    {
      "epoch": 34.11,
      "grad_norm": 2.6518051624298096,
      "learning_rate": 4.503781515563769e-06,
      "loss": 0.5673,
      "step": 83320
    },
    {
      "epoch": 34.11,
      "grad_norm": 2.2789981365203857,
      "learning_rate": 4.5037199450486735e-06,
      "loss": 0.561,
      "step": 83330
    },
    {
      "epoch": 34.11,
      "grad_norm": 2.7880542278289795,
      "learning_rate": 4.50365836463909e-06,
      "loss": 0.5732,
      "step": 83340
    },
    {
      "epoch": 34.12,
      "grad_norm": 2.5296220779418945,
      "learning_rate": 4.5035967743353e-06,
      "loss": 0.5855,
      "step": 83350
    },
    {
      "epoch": 34.12,
      "grad_norm": 2.1965837478637695,
      "learning_rate": 4.503535174137586e-06,
      "loss": 0.5517,
      "step": 83360
    },
    {
      "epoch": 34.13,
      "grad_norm": 2.563058376312256,
      "learning_rate": 4.503473564046232e-06,
      "loss": 0.566,
      "step": 83370
    },
    {
      "epoch": 34.13,
      "grad_norm": 2.837421417236328,
      "learning_rate": 4.503411944061518e-06,
      "loss": 0.5895,
      "step": 83380
    },
    {
      "epoch": 34.13,
      "grad_norm": 1.8446687459945679,
      "learning_rate": 4.503350314183728e-06,
      "loss": 0.5891,
      "step": 83390
    },
    {
      "epoch": 34.14,
      "grad_norm": 3.0778112411499023,
      "learning_rate": 4.503288674413143e-06,
      "loss": 0.573,
      "step": 83400
    },
    {
      "epoch": 34.14,
      "grad_norm": 2.4312331676483154,
      "learning_rate": 4.503227024750045e-06,
      "loss": 0.5584,
      "step": 83410
    },
    {
      "epoch": 34.15,
      "grad_norm": 3.4574971199035645,
      "learning_rate": 4.5031653651947184e-06,
      "loss": 0.5862,
      "step": 83420
    },
    {
      "epoch": 34.15,
      "grad_norm": 2.5949971675872803,
      "learning_rate": 4.5031036957474436e-06,
      "loss": 0.5737,
      "step": 83430
    },
    {
      "epoch": 34.15,
      "grad_norm": 2.6685798168182373,
      "learning_rate": 4.5030420164085035e-06,
      "loss": 0.5804,
      "step": 83440
    },
    {
      "epoch": 34.16,
      "grad_norm": 2.447844982147217,
      "learning_rate": 4.502980327178182e-06,
      "loss": 0.5512,
      "step": 83450
    },
    {
      "epoch": 34.16,
      "grad_norm": 2.1422717571258545,
      "learning_rate": 4.5029186280567605e-06,
      "loss": 0.5512,
      "step": 83460
    },
    {
      "epoch": 34.17,
      "grad_norm": 2.0065321922302246,
      "learning_rate": 4.5028569190445215e-06,
      "loss": 0.5571,
      "step": 83470
    },
    {
      "epoch": 34.17,
      "grad_norm": 2.7079527378082275,
      "learning_rate": 4.502795200141749e-06,
      "loss": 0.5667,
      "step": 83480
    },
    {
      "epoch": 34.18,
      "grad_norm": 2.4882991313934326,
      "learning_rate": 4.502733471348724e-06,
      "loss": 0.5589,
      "step": 83490
    },
    {
      "epoch": 34.18,
      "grad_norm": 2.856297731399536,
      "learning_rate": 4.50267173266573e-06,
      "loss": 0.5772,
      "step": 83500
    },
    {
      "epoch": 34.18,
      "grad_norm": 2.1484534740448,
      "learning_rate": 4.5026099840930505e-06,
      "loss": 0.569,
      "step": 83510
    },
    {
      "epoch": 34.19,
      "grad_norm": 2.2379939556121826,
      "learning_rate": 4.502548225630968e-06,
      "loss": 0.5782,
      "step": 83520
    },
    {
      "epoch": 34.19,
      "grad_norm": 1.7736546993255615,
      "learning_rate": 4.502486457279764e-06,
      "loss": 0.5562,
      "step": 83530
    },
    {
      "epoch": 34.2,
      "grad_norm": 2.550743818283081,
      "learning_rate": 4.502424679039724e-06,
      "loss": 0.5665,
      "step": 83540
    },
    {
      "epoch": 34.2,
      "grad_norm": 2.8118085861206055,
      "learning_rate": 4.502362890911128e-06,
      "loss": 0.5781,
      "step": 83550
    },
    {
      "epoch": 34.2,
      "grad_norm": 2.1982157230377197,
      "learning_rate": 4.502301092894262e-06,
      "loss": 0.59,
      "step": 83560
    },
    {
      "epoch": 34.21,
      "grad_norm": 2.011183500289917,
      "learning_rate": 4.502239284989407e-06,
      "loss": 0.5951,
      "step": 83570
    },
    {
      "epoch": 34.21,
      "grad_norm": 2.520460605621338,
      "learning_rate": 4.5021774671968475e-06,
      "loss": 0.5792,
      "step": 83580
    },
    {
      "epoch": 34.22,
      "grad_norm": 1.9923748970031738,
      "learning_rate": 4.502115639516866e-06,
      "loss": 0.5779,
      "step": 83590
    },
    {
      "epoch": 34.22,
      "grad_norm": 2.402695417404175,
      "learning_rate": 4.502053801949745e-06,
      "loss": 0.5874,
      "step": 83600
    },
    {
      "epoch": 34.22,
      "grad_norm": 2.7527477741241455,
      "learning_rate": 4.501991954495769e-06,
      "loss": 0.5798,
      "step": 83610
    },
    {
      "epoch": 34.23,
      "grad_norm": 2.0956685543060303,
      "learning_rate": 4.501930097155221e-06,
      "loss": 0.5629,
      "step": 83620
    },
    {
      "epoch": 34.23,
      "grad_norm": 2.4639880657196045,
      "learning_rate": 4.5018682299283836e-06,
      "loss": 0.575,
      "step": 83630
    },
    {
      "epoch": 34.24,
      "grad_norm": 1.8427313566207886,
      "learning_rate": 4.501806352815541e-06,
      "loss": 0.577,
      "step": 83640
    },
    {
      "epoch": 34.24,
      "grad_norm": 2.181307315826416,
      "learning_rate": 4.501744465816977e-06,
      "loss": 0.5689,
      "step": 83650
    },
    {
      "epoch": 34.24,
      "grad_norm": 1.6982369422912598,
      "learning_rate": 4.5016825689329745e-06,
      "loss": 0.5737,
      "step": 83660
    },
    {
      "epoch": 34.25,
      "grad_norm": 2.2065653800964355,
      "learning_rate": 4.501620662163817e-06,
      "loss": 0.5778,
      "step": 83670
    },
    {
      "epoch": 34.25,
      "grad_norm": 2.4075376987457275,
      "learning_rate": 4.501558745509788e-06,
      "loss": 0.5947,
      "step": 83680
    },
    {
      "epoch": 34.26,
      "grad_norm": 2.4106810092926025,
      "learning_rate": 4.501496818971172e-06,
      "loss": 0.5656,
      "step": 83690
    },
    {
      "epoch": 34.26,
      "grad_norm": 2.0300233364105225,
      "learning_rate": 4.501434882548251e-06,
      "loss": 0.5556,
      "step": 83700
    },
    {
      "epoch": 34.27,
      "grad_norm": 1.862162709236145,
      "learning_rate": 4.50137293624131e-06,
      "loss": 0.5706,
      "step": 83710
    },
    {
      "epoch": 34.27,
      "grad_norm": 1.8989619016647339,
      "learning_rate": 4.501310980050632e-06,
      "loss": 0.5556,
      "step": 83720
    },
    {
      "epoch": 34.27,
      "grad_norm": 3.0474064350128174,
      "learning_rate": 4.501249013976502e-06,
      "loss": 0.5794,
      "step": 83730
    },
    {
      "epoch": 34.28,
      "grad_norm": 2.1992270946502686,
      "learning_rate": 4.501187038019204e-06,
      "loss": 0.565,
      "step": 83740
    },
    {
      "epoch": 34.28,
      "grad_norm": 1.9642821550369263,
      "learning_rate": 4.501125052179019e-06,
      "loss": 0.5608,
      "step": 83750
    },
    {
      "epoch": 34.29,
      "grad_norm": 2.3618061542510986,
      "learning_rate": 4.501063056456235e-06,
      "loss": 0.5604,
      "step": 83760
    },
    {
      "epoch": 34.29,
      "grad_norm": 3.1420042514801025,
      "learning_rate": 4.5010010508511326e-06,
      "loss": 0.5686,
      "step": 83770
    },
    {
      "epoch": 34.29,
      "grad_norm": 1.925045371055603,
      "learning_rate": 4.500939035363998e-06,
      "loss": 0.5703,
      "step": 83780
    },
    {
      "epoch": 34.3,
      "grad_norm": 3.3474555015563965,
      "learning_rate": 4.500877009995114e-06,
      "loss": 0.5553,
      "step": 83790
    },
    {
      "epoch": 34.3,
      "grad_norm": 2.104809045791626,
      "learning_rate": 4.500814974744765e-06,
      "loss": 0.5719,
      "step": 83800
    },
    {
      "epoch": 34.31,
      "grad_norm": 2.0741891860961914,
      "learning_rate": 4.500752929613237e-06,
      "loss": 0.5391,
      "step": 83810
    },
    {
      "epoch": 34.31,
      "grad_norm": 2.665926456451416,
      "learning_rate": 4.50069087460081e-06,
      "loss": 0.5517,
      "step": 83820
    },
    {
      "epoch": 34.31,
      "grad_norm": 2.2159714698791504,
      "learning_rate": 4.5006288097077735e-06,
      "loss": 0.5459,
      "step": 83830
    },
    {
      "epoch": 34.32,
      "grad_norm": 2.5707850456237793,
      "learning_rate": 4.500566734934407e-06,
      "loss": 0.5647,
      "step": 83840
    },
    {
      "epoch": 34.32,
      "grad_norm": 2.3666884899139404,
      "learning_rate": 4.500504650280998e-06,
      "loss": 0.5565,
      "step": 83850
    },
    {
      "epoch": 34.33,
      "grad_norm": 2.7887094020843506,
      "learning_rate": 4.5004425557478314e-06,
      "loss": 0.5601,
      "step": 83860
    },
    {
      "epoch": 34.33,
      "grad_norm": 1.9626038074493408,
      "learning_rate": 4.500380451335188e-06,
      "loss": 0.5606,
      "step": 83870
    },
    {
      "epoch": 34.33,
      "grad_norm": 2.397308349609375,
      "learning_rate": 4.500318337043356e-06,
      "loss": 0.5617,
      "step": 83880
    },
    {
      "epoch": 34.34,
      "grad_norm": 2.511850595474243,
      "learning_rate": 4.5002562128726175e-06,
      "loss": 0.554,
      "step": 83890
    },
    {
      "epoch": 34.34,
      "grad_norm": 2.3202342987060547,
      "learning_rate": 4.500194078823259e-06,
      "loss": 0.5562,
      "step": 83900
    },
    {
      "epoch": 34.35,
      "grad_norm": 1.884649395942688,
      "learning_rate": 4.500131934895563e-06,
      "loss": 0.5714,
      "step": 83910
    },
    {
      "epoch": 34.35,
      "grad_norm": 2.5460870265960693,
      "learning_rate": 4.500069781089816e-06,
      "loss": 0.5776,
      "step": 83920
    },
    {
      "epoch": 34.36,
      "grad_norm": 2.0485808849334717,
      "learning_rate": 4.500007617406302e-06,
      "loss": 0.568,
      "step": 83930
    },
    {
      "epoch": 34.36,
      "grad_norm": 1.9019063711166382,
      "learning_rate": 4.4999454438453055e-06,
      "loss": 0.5703,
      "step": 83940
    },
    {
      "epoch": 34.36,
      "grad_norm": 3.147703170776367,
      "learning_rate": 4.499883260407112e-06,
      "loss": 0.5476,
      "step": 83950
    },
    {
      "epoch": 34.37,
      "grad_norm": 2.3530423641204834,
      "learning_rate": 4.4998210670920055e-06,
      "loss": 0.5531,
      "step": 83960
    },
    {
      "epoch": 34.37,
      "grad_norm": 2.398447036743164,
      "learning_rate": 4.499758863900271e-06,
      "loss": 0.5513,
      "step": 83970
    },
    {
      "epoch": 34.38,
      "grad_norm": 1.5923625230789185,
      "learning_rate": 4.499696650832194e-06,
      "loss": 0.5549,
      "step": 83980
    },
    {
      "epoch": 34.38,
      "grad_norm": 1.9556424617767334,
      "learning_rate": 4.49963442788806e-06,
      "loss": 0.554,
      "step": 83990
    },
    {
      "epoch": 34.38,
      "grad_norm": 2.2342991828918457,
      "learning_rate": 4.499572195068153e-06,
      "loss": 0.5766,
      "step": 84000
    },
    {
      "epoch": 34.39,
      "grad_norm": 1.9873147010803223,
      "learning_rate": 4.499509952372758e-06,
      "loss": 0.5758,
      "step": 84010
    },
    {
      "epoch": 34.39,
      "grad_norm": 2.8186209201812744,
      "learning_rate": 4.499447699802161e-06,
      "loss": 0.5562,
      "step": 84020
    },
    {
      "epoch": 34.4,
      "grad_norm": 2.474529981613159,
      "learning_rate": 4.4993854373566475e-06,
      "loss": 0.5631,
      "step": 84030
    },
    {
      "epoch": 34.4,
      "grad_norm": 2.23685622215271,
      "learning_rate": 4.499323165036501e-06,
      "loss": 0.5687,
      "step": 84040
    },
    {
      "epoch": 34.4,
      "grad_norm": 1.8063546419143677,
      "learning_rate": 4.499260882842007e-06,
      "loss": 0.5742,
      "step": 84050
    },
    {
      "epoch": 34.41,
      "grad_norm": 1.9214212894439697,
      "learning_rate": 4.499198590773452e-06,
      "loss": 0.5542,
      "step": 84060
    },
    {
      "epoch": 34.41,
      "grad_norm": 2.5034210681915283,
      "learning_rate": 4.499136288831121e-06,
      "loss": 0.5852,
      "step": 84070
    },
    {
      "epoch": 34.42,
      "grad_norm": 2.7393124103546143,
      "learning_rate": 4.499073977015299e-06,
      "loss": 0.5463,
      "step": 84080
    },
    {
      "epoch": 34.42,
      "grad_norm": 1.7990577220916748,
      "learning_rate": 4.499011655326272e-06,
      "loss": 0.5474,
      "step": 84090
    },
    {
      "epoch": 34.42,
      "grad_norm": 3.1844570636749268,
      "learning_rate": 4.498949323764325e-06,
      "loss": 0.5748,
      "step": 84100
    },
    {
      "epoch": 34.43,
      "grad_norm": 2.4454238414764404,
      "learning_rate": 4.498886982329744e-06,
      "loss": 0.5664,
      "step": 84110
    },
    {
      "epoch": 34.43,
      "grad_norm": 1.9724844694137573,
      "learning_rate": 4.498824631022813e-06,
      "loss": 0.5644,
      "step": 84120
    },
    {
      "epoch": 34.44,
      "grad_norm": 2.19851016998291,
      "learning_rate": 4.4987622698438196e-06,
      "loss": 0.5556,
      "step": 84130
    },
    {
      "epoch": 34.44,
      "grad_norm": 2.6408498287200928,
      "learning_rate": 4.498699898793049e-06,
      "loss": 0.5675,
      "step": 84140
    },
    {
      "epoch": 34.45,
      "grad_norm": 2.3320648670196533,
      "learning_rate": 4.498637517870786e-06,
      "loss": 0.5814,
      "step": 84150
    },
    {
      "epoch": 34.45,
      "grad_norm": 1.7711008787155151,
      "learning_rate": 4.4985751270773185e-06,
      "loss": 0.5658,
      "step": 84160
    },
    {
      "epoch": 34.45,
      "grad_norm": 2.1856658458709717,
      "learning_rate": 4.498512726412929e-06,
      "loss": 0.5689,
      "step": 84170
    },
    {
      "epoch": 34.46,
      "grad_norm": 2.2338438034057617,
      "learning_rate": 4.498450315877906e-06,
      "loss": 0.5906,
      "step": 84180
    },
    {
      "epoch": 34.46,
      "grad_norm": 2.4722647666931152,
      "learning_rate": 4.498387895472535e-06,
      "loss": 0.5744,
      "step": 84190
    },
    {
      "epoch": 34.47,
      "grad_norm": 2.253589153289795,
      "learning_rate": 4.498325465197101e-06,
      "loss": 0.5763,
      "step": 84200
    },
    {
      "epoch": 34.47,
      "grad_norm": 2.14093279838562,
      "learning_rate": 4.498263025051891e-06,
      "loss": 0.5636,
      "step": 84210
    },
    {
      "epoch": 34.47,
      "grad_norm": 1.7579559087753296,
      "learning_rate": 4.4982005750371895e-06,
      "loss": 0.5528,
      "step": 84220
    },
    {
      "epoch": 34.48,
      "grad_norm": 2.2433481216430664,
      "learning_rate": 4.498138115153284e-06,
      "loss": 0.5668,
      "step": 84230
    },
    {
      "epoch": 34.48,
      "grad_norm": 2.163489580154419,
      "learning_rate": 4.49807564540046e-06,
      "loss": 0.5668,
      "step": 84240
    },
    {
      "epoch": 34.49,
      "grad_norm": 2.672640323638916,
      "learning_rate": 4.4980131657790045e-06,
      "loss": 0.5663,
      "step": 84250
    },
    {
      "epoch": 34.49,
      "grad_norm": 2.4213194847106934,
      "learning_rate": 4.497950676289203e-06,
      "loss": 0.5524,
      "step": 84260
    },
    {
      "epoch": 34.49,
      "grad_norm": 1.9754881858825684,
      "learning_rate": 4.497888176931341e-06,
      "loss": 0.5578,
      "step": 84270
    },
    {
      "epoch": 34.5,
      "grad_norm": 1.9242138862609863,
      "learning_rate": 4.497825667705706e-06,
      "loss": 0.586,
      "step": 84280
    },
    {
      "epoch": 34.5,
      "grad_norm": 3.225085973739624,
      "learning_rate": 4.4977631486125845e-06,
      "loss": 0.5416,
      "step": 84290
    },
    {
      "epoch": 34.51,
      "grad_norm": 2.3491785526275635,
      "learning_rate": 4.497700619652262e-06,
      "loss": 0.5881,
      "step": 84300
    },
    {
      "epoch": 34.51,
      "grad_norm": 2.2156457901000977,
      "learning_rate": 4.4976380808250245e-06,
      "loss": 0.55,
      "step": 84310
    },
    {
      "epoch": 34.51,
      "grad_norm": 2.1743526458740234,
      "learning_rate": 4.497575532131161e-06,
      "loss": 0.5642,
      "step": 84320
    },
    {
      "epoch": 34.52,
      "grad_norm": 2.408738374710083,
      "learning_rate": 4.4975129735709545e-06,
      "loss": 0.5715,
      "step": 84330
    },
    {
      "epoch": 34.52,
      "grad_norm": 2.174107551574707,
      "learning_rate": 4.4974504051446945e-06,
      "loss": 0.5822,
      "step": 84340
    },
    {
      "epoch": 34.53,
      "grad_norm": 2.914175033569336,
      "learning_rate": 4.497387826852667e-06,
      "loss": 0.5535,
      "step": 84350
    },
    {
      "epoch": 34.53,
      "grad_norm": 4.1847243309021,
      "learning_rate": 4.497325238695157e-06,
      "loss": 0.5713,
      "step": 84360
    },
    {
      "epoch": 34.54,
      "grad_norm": 1.8495010137557983,
      "learning_rate": 4.497262640672452e-06,
      "loss": 0.5508,
      "step": 84370
    },
    {
      "epoch": 34.54,
      "grad_norm": 1.9461454153060913,
      "learning_rate": 4.4972000327848396e-06,
      "loss": 0.5608,
      "step": 84380
    },
    {
      "epoch": 34.54,
      "grad_norm": 2.379945993423462,
      "learning_rate": 4.497137415032606e-06,
      "loss": 0.5473,
      "step": 84390
    },
    {
      "epoch": 34.55,
      "grad_norm": 2.246286392211914,
      "learning_rate": 4.497074787416039e-06,
      "loss": 0.5822,
      "step": 84400
    },
    {
      "epoch": 34.55,
      "grad_norm": 1.9813677072525024,
      "learning_rate": 4.497012149935424e-06,
      "loss": 0.5484,
      "step": 84410
    },
    {
      "epoch": 34.56,
      "grad_norm": 3.2760396003723145,
      "learning_rate": 4.496949502591049e-06,
      "loss": 0.559,
      "step": 84420
    },
    {
      "epoch": 34.56,
      "grad_norm": 2.4454028606414795,
      "learning_rate": 4.4968868453832e-06,
      "loss": 0.5696,
      "step": 84430
    },
    {
      "epoch": 34.56,
      "grad_norm": 2.5118653774261475,
      "learning_rate": 4.496824178312165e-06,
      "loss": 0.5829,
      "step": 84440
    },
    {
      "epoch": 34.57,
      "grad_norm": 1.9584550857543945,
      "learning_rate": 4.49676150137823e-06,
      "loss": 0.565,
      "step": 84450
    },
    {
      "epoch": 34.57,
      "grad_norm": 1.955704689025879,
      "learning_rate": 4.496698814581683e-06,
      "loss": 0.5532,
      "step": 84460
    },
    {
      "epoch": 34.58,
      "grad_norm": 1.6284536123275757,
      "learning_rate": 4.496636117922811e-06,
      "loss": 0.5548,
      "step": 84470
    },
    {
      "epoch": 34.58,
      "grad_norm": 2.924957513809204,
      "learning_rate": 4.496573411401901e-06,
      "loss": 0.5589,
      "step": 84480
    },
    {
      "epoch": 34.58,
      "grad_norm": 2.027667760848999,
      "learning_rate": 4.496510695019241e-06,
      "loss": 0.572,
      "step": 84490
    },
    {
      "epoch": 34.59,
      "grad_norm": 3.034834623336792,
      "learning_rate": 4.496447968775117e-06,
      "loss": 0.5355,
      "step": 84500
    },
    {
      "epoch": 34.59,
      "grad_norm": 2.4442684650421143,
      "learning_rate": 4.496385232669817e-06,
      "loss": 0.5877,
      "step": 84510
    },
    {
      "epoch": 34.6,
      "grad_norm": 2.6603524684906006,
      "learning_rate": 4.496322486703629e-06,
      "loss": 0.5769,
      "step": 84520
    },
    {
      "epoch": 34.6,
      "grad_norm": 1.8714003562927246,
      "learning_rate": 4.496259730876838e-06,
      "loss": 0.5506,
      "step": 84530
    },
    {
      "epoch": 34.6,
      "grad_norm": 2.377241849899292,
      "learning_rate": 4.4961969651897355e-06,
      "loss": 0.5599,
      "step": 84540
    },
    {
      "epoch": 34.61,
      "grad_norm": 2.6475305557250977,
      "learning_rate": 4.496134189642606e-06,
      "loss": 0.5545,
      "step": 84550
    },
    {
      "epoch": 34.61,
      "grad_norm": 2.2546515464782715,
      "learning_rate": 4.496071404235738e-06,
      "loss": 0.5569,
      "step": 84560
    },
    {
      "epoch": 34.62,
      "grad_norm": 1.984316110610962,
      "learning_rate": 4.496008608969419e-06,
      "loss": 0.5615,
      "step": 84570
    },
    {
      "epoch": 34.62,
      "grad_norm": 2.4404985904693604,
      "learning_rate": 4.495945803843936e-06,
      "loss": 0.5686,
      "step": 84580
    },
    {
      "epoch": 34.63,
      "grad_norm": 2.28354549407959,
      "learning_rate": 4.495882988859578e-06,
      "loss": 0.5859,
      "step": 84590
    },
    {
      "epoch": 34.63,
      "grad_norm": 2.5634963512420654,
      "learning_rate": 4.495820164016632e-06,
      "loss": 0.5794,
      "step": 84600
    },
    {
      "epoch": 34.63,
      "grad_norm": 1.7660819292068481,
      "learning_rate": 4.495757329315384e-06,
      "loss": 0.5671,
      "step": 84610
    },
    {
      "epoch": 34.64,
      "grad_norm": 2.674086093902588,
      "learning_rate": 4.495694484756126e-06,
      "loss": 0.5856,
      "step": 84620
    },
    {
      "epoch": 34.64,
      "grad_norm": 1.901663899421692,
      "learning_rate": 4.495631630339143e-06,
      "loss": 0.5682,
      "step": 84630
    },
    {
      "epoch": 34.65,
      "grad_norm": 3.3835084438323975,
      "learning_rate": 4.495568766064724e-06,
      "loss": 0.5707,
      "step": 84640
    },
    {
      "epoch": 34.65,
      "grad_norm": 2.6889309883117676,
      "learning_rate": 4.495505891933156e-06,
      "loss": 0.5599,
      "step": 84650
    },
    {
      "epoch": 34.65,
      "grad_norm": 2.6528499126434326,
      "learning_rate": 4.495443007944727e-06,
      "loss": 0.5605,
      "step": 84660
    },
    {
      "epoch": 34.66,
      "grad_norm": 2.228358030319214,
      "learning_rate": 4.495380114099726e-06,
      "loss": 0.5549,
      "step": 84670
    },
    {
      "epoch": 34.66,
      "grad_norm": 2.919539451599121,
      "learning_rate": 4.49531721039844e-06,
      "loss": 0.5568,
      "step": 84680
    },
    {
      "epoch": 34.67,
      "grad_norm": 1.7108219861984253,
      "learning_rate": 4.49525429684116e-06,
      "loss": 0.5932,
      "step": 84690
    },
    {
      "epoch": 34.67,
      "grad_norm": 2.109978199005127,
      "learning_rate": 4.49519137342817e-06,
      "loss": 0.5784,
      "step": 84700
    },
    {
      "epoch": 34.67,
      "grad_norm": 2.2045559883117676,
      "learning_rate": 4.495128440159761e-06,
      "loss": 0.5674,
      "step": 84710
    },
    {
      "epoch": 34.68,
      "grad_norm": 2.372645139694214,
      "learning_rate": 4.495065497036219e-06,
      "loss": 0.5622,
      "step": 84720
    },
    {
      "epoch": 34.68,
      "grad_norm": 2.4504308700561523,
      "learning_rate": 4.495002544057836e-06,
      "loss": 0.5688,
      "step": 84730
    },
    {
      "epoch": 34.69,
      "grad_norm": 3.382962703704834,
      "learning_rate": 4.494939581224897e-06,
      "loss": 0.5941,
      "step": 84740
    },
    {
      "epoch": 34.69,
      "grad_norm": 2.1846652030944824,
      "learning_rate": 4.494876608537693e-06,
      "loss": 0.5633,
      "step": 84750
    },
    {
      "epoch": 34.7,
      "grad_norm": 2.2751426696777344,
      "learning_rate": 4.49481362599651e-06,
      "loss": 0.5548,
      "step": 84760
    },
    {
      "epoch": 34.7,
      "grad_norm": 2.273183822631836,
      "learning_rate": 4.494750633601638e-06,
      "loss": 0.58,
      "step": 84770
    },
    {
      "epoch": 34.7,
      "grad_norm": 2.0835211277008057,
      "learning_rate": 4.494687631353364e-06,
      "loss": 0.5715,
      "step": 84780
    },
    {
      "epoch": 34.71,
      "grad_norm": 2.0900094509124756,
      "learning_rate": 4.49462461925198e-06,
      "loss": 0.562,
      "step": 84790
    },
    {
      "epoch": 34.71,
      "grad_norm": 2.1151490211486816,
      "learning_rate": 4.4945615972977715e-06,
      "loss": 0.5576,
      "step": 84800
    },
    {
      "epoch": 34.72,
      "grad_norm": 2.8513572216033936,
      "learning_rate": 4.494498565491028e-06,
      "loss": 0.5903,
      "step": 84810
    },
    {
      "epoch": 34.72,
      "grad_norm": 2.8384411334991455,
      "learning_rate": 4.494435523832039e-06,
      "loss": 0.5628,
      "step": 84820
    },
    {
      "epoch": 34.72,
      "grad_norm": 2.1118428707122803,
      "learning_rate": 4.494372472321092e-06,
      "loss": 0.5686,
      "step": 84830
    },
    {
      "epoch": 34.73,
      "grad_norm": 2.890911340713501,
      "learning_rate": 4.494309410958477e-06,
      "loss": 0.5738,
      "step": 84840
    },
    {
      "epoch": 34.73,
      "grad_norm": 2.6222434043884277,
      "learning_rate": 4.4942463397444816e-06,
      "loss": 0.5839,
      "step": 84850
    },
    {
      "epoch": 34.74,
      "grad_norm": 1.9203988313674927,
      "learning_rate": 4.494183258679396e-06,
      "loss": 0.5819,
      "step": 84860
    },
    {
      "epoch": 34.74,
      "grad_norm": 2.7117950916290283,
      "learning_rate": 4.4941201677635094e-06,
      "loss": 0.584,
      "step": 84870
    },
    {
      "epoch": 34.74,
      "grad_norm": 3.2783150672912598,
      "learning_rate": 4.494057066997109e-06,
      "loss": 0.5821,
      "step": 84880
    },
    {
      "epoch": 34.75,
      "grad_norm": 1.950311541557312,
      "learning_rate": 4.493993956380486e-06,
      "loss": 0.5595,
      "step": 84890
    },
    {
      "epoch": 34.75,
      "grad_norm": 2.038668632507324,
      "learning_rate": 4.493930835913927e-06,
      "loss": 0.5574,
      "step": 84900
    },
    {
      "epoch": 34.76,
      "grad_norm": 1.96699857711792,
      "learning_rate": 4.4938677055977235e-06,
      "loss": 0.5462,
      "step": 84910
    },
    {
      "epoch": 34.76,
      "grad_norm": 1.849463701248169,
      "learning_rate": 4.493804565432164e-06,
      "loss": 0.5767,
      "step": 84920
    },
    {
      "epoch": 34.76,
      "grad_norm": 1.7447079420089722,
      "learning_rate": 4.493741415417537e-06,
      "loss": 0.5673,
      "step": 84930
    },
    {
      "epoch": 34.77,
      "grad_norm": 2.3556385040283203,
      "learning_rate": 4.493678255554133e-06,
      "loss": 0.5543,
      "step": 84940
    },
    {
      "epoch": 34.77,
      "grad_norm": 2.2607262134552,
      "learning_rate": 4.49361508584224e-06,
      "loss": 0.5691,
      "step": 84950
    },
    {
      "epoch": 34.78,
      "grad_norm": 2.2491605281829834,
      "learning_rate": 4.493551906282148e-06,
      "loss": 0.5867,
      "step": 84960
    },
    {
      "epoch": 34.78,
      "grad_norm": 2.627399206161499,
      "learning_rate": 4.493488716874146e-06,
      "loss": 0.5497,
      "step": 84970
    },
    {
      "epoch": 34.79,
      "grad_norm": 1.9784607887268066,
      "learning_rate": 4.493425517618525e-06,
      "loss": 0.5599,
      "step": 84980
    },
    {
      "epoch": 34.79,
      "grad_norm": 2.9604172706604004,
      "learning_rate": 4.493362308515573e-06,
      "loss": 0.5979,
      "step": 84990
    },
    {
      "epoch": 34.79,
      "grad_norm": 2.696425437927246,
      "learning_rate": 4.493299089565579e-06,
      "loss": 0.5614,
      "step": 85000
    },
    {
      "epoch": 34.8,
      "grad_norm": 3.0626654624938965,
      "learning_rate": 4.493235860768834e-06,
      "loss": 0.5923,
      "step": 85010
    },
    {
      "epoch": 34.8,
      "grad_norm": 1.9462599754333496,
      "learning_rate": 4.493172622125627e-06,
      "loss": 0.5619,
      "step": 85020
    },
    {
      "epoch": 34.81,
      "grad_norm": 1.9550384283065796,
      "learning_rate": 4.493109373636249e-06,
      "loss": 0.5594,
      "step": 85030
    },
    {
      "epoch": 34.81,
      "grad_norm": 2.5195016860961914,
      "learning_rate": 4.493046115300988e-06,
      "loss": 0.5618,
      "step": 85040
    },
    {
      "epoch": 34.81,
      "grad_norm": 2.8233518600463867,
      "learning_rate": 4.492982847120134e-06,
      "loss": 0.567,
      "step": 85050
    },
    {
      "epoch": 34.82,
      "grad_norm": 2.0928165912628174,
      "learning_rate": 4.492919569093977e-06,
      "loss": 0.572,
      "step": 85060
    },
    {
      "epoch": 34.82,
      "grad_norm": 3.686784267425537,
      "learning_rate": 4.492856281222808e-06,
      "loss": 0.5816,
      "step": 85070
    },
    {
      "epoch": 34.83,
      "grad_norm": 1.9349483251571655,
      "learning_rate": 4.492792983506915e-06,
      "loss": 0.5637,
      "step": 85080
    },
    {
      "epoch": 34.83,
      "grad_norm": 2.4487500190734863,
      "learning_rate": 4.4927296759465885e-06,
      "loss": 0.554,
      "step": 85090
    },
    {
      "epoch": 34.83,
      "grad_norm": 2.8304128646850586,
      "learning_rate": 4.49266635854212e-06,
      "loss": 0.5603,
      "step": 85100
    },
    {
      "epoch": 34.84,
      "grad_norm": 2.4188928604125977,
      "learning_rate": 4.492603031293798e-06,
      "loss": 0.5699,
      "step": 85110
    },
    {
      "epoch": 34.84,
      "grad_norm": 2.7484798431396484,
      "learning_rate": 4.492539694201913e-06,
      "loss": 0.5592,
      "step": 85120
    },
    {
      "epoch": 34.85,
      "grad_norm": 2.483092784881592,
      "learning_rate": 4.492476347266756e-06,
      "loss": 0.5605,
      "step": 85130
    },
    {
      "epoch": 34.85,
      "grad_norm": 2.23539137840271,
      "learning_rate": 4.4924129904886155e-06,
      "loss": 0.5633,
      "step": 85140
    },
    {
      "epoch": 34.85,
      "grad_norm": 2.86722731590271,
      "learning_rate": 4.492349623867783e-06,
      "loss": 0.5612,
      "step": 85150
    },
    {
      "epoch": 34.86,
      "grad_norm": 2.109886884689331,
      "learning_rate": 4.492286247404549e-06,
      "loss": 0.5564,
      "step": 85160
    },
    {
      "epoch": 34.86,
      "grad_norm": 1.8919273614883423,
      "learning_rate": 4.492222861099201e-06,
      "loss": 0.561,
      "step": 85170
    },
    {
      "epoch": 34.87,
      "grad_norm": 2.4622278213500977,
      "learning_rate": 4.492159464952034e-06,
      "loss": 0.5655,
      "step": 85180
    },
    {
      "epoch": 34.87,
      "grad_norm": 2.126413583755493,
      "learning_rate": 4.4920960589633344e-06,
      "loss": 0.5633,
      "step": 85190
    },
    {
      "epoch": 34.88,
      "grad_norm": 1.8167738914489746,
      "learning_rate": 4.4920326431333944e-06,
      "loss": 0.5798,
      "step": 85200
    },
    {
      "epoch": 34.88,
      "grad_norm": 2.86224627494812,
      "learning_rate": 4.491969217462505e-06,
      "loss": 0.5734,
      "step": 85210
    },
    {
      "epoch": 34.88,
      "grad_norm": 2.701204776763916,
      "learning_rate": 4.491905781950956e-06,
      "loss": 0.546,
      "step": 85220
    },
    {
      "epoch": 34.89,
      "grad_norm": 2.240469455718994,
      "learning_rate": 4.491842336599037e-06,
      "loss": 0.5567,
      "step": 85230
    },
    {
      "epoch": 34.89,
      "grad_norm": 2.81125807762146,
      "learning_rate": 4.491778881407041e-06,
      "loss": 0.5682,
      "step": 85240
    },
    {
      "epoch": 34.9,
      "grad_norm": 2.183816432952881,
      "learning_rate": 4.4917154163752565e-06,
      "loss": 0.5599,
      "step": 85250
    },
    {
      "epoch": 34.9,
      "grad_norm": 2.3115265369415283,
      "learning_rate": 4.491651941503975e-06,
      "loss": 0.5553,
      "step": 85260
    },
    {
      "epoch": 34.9,
      "grad_norm": 1.970459222793579,
      "learning_rate": 4.4915884567934886e-06,
      "loss": 0.5576,
      "step": 85270
    },
    {
      "epoch": 34.91,
      "grad_norm": 2.5005202293395996,
      "learning_rate": 4.491524962244086e-06,
      "loss": 0.5633,
      "step": 85280
    },
    {
      "epoch": 34.91,
      "grad_norm": 2.1164631843566895,
      "learning_rate": 4.491461457856059e-06,
      "loss": 0.5773,
      "step": 85290
    },
    {
      "epoch": 34.92,
      "grad_norm": 2.2958056926727295,
      "learning_rate": 4.4913979436296984e-06,
      "loss": 0.5596,
      "step": 85300
    },
    {
      "epoch": 34.92,
      "grad_norm": 1.6544073820114136,
      "learning_rate": 4.491334419565295e-06,
      "loss": 0.5639,
      "step": 85310
    },
    {
      "epoch": 34.92,
      "grad_norm": 2.217711925506592,
      "learning_rate": 4.49127088566314e-06,
      "loss": 0.5701,
      "step": 85320
    },
    {
      "epoch": 34.93,
      "grad_norm": 2.9970591068267822,
      "learning_rate": 4.491207341923524e-06,
      "loss": 0.5564,
      "step": 85330
    },
    {
      "epoch": 34.93,
      "grad_norm": 2.789825439453125,
      "learning_rate": 4.49114378834674e-06,
      "loss": 0.5539,
      "step": 85340
    },
    {
      "epoch": 34.94,
      "grad_norm": 1.997558355331421,
      "learning_rate": 4.491080224933076e-06,
      "loss": 0.5624,
      "step": 85350
    },
    {
      "epoch": 34.94,
      "grad_norm": 2.5521063804626465,
      "learning_rate": 4.491016651682825e-06,
      "loss": 0.5507,
      "step": 85360
    },
    {
      "epoch": 34.94,
      "grad_norm": 2.436234474182129,
      "learning_rate": 4.490953068596279e-06,
      "loss": 0.5688,
      "step": 85370
    },
    {
      "epoch": 34.95,
      "grad_norm": 2.0248711109161377,
      "learning_rate": 4.490889475673727e-06,
      "loss": 0.5503,
      "step": 85380
    },
    {
      "epoch": 34.95,
      "grad_norm": 2.1825945377349854,
      "learning_rate": 4.490825872915463e-06,
      "loss": 0.5716,
      "step": 85390
    },
    {
      "epoch": 34.96,
      "grad_norm": 2.313704013824463,
      "learning_rate": 4.490762260321776e-06,
      "loss": 0.5526,
      "step": 85400
    },
    {
      "epoch": 34.96,
      "grad_norm": 2.466364622116089,
      "learning_rate": 4.490698637892959e-06,
      "loss": 0.5536,
      "step": 85410
    },
    {
      "epoch": 34.97,
      "grad_norm": 2.5362789630889893,
      "learning_rate": 4.490635005629302e-06,
      "loss": 0.565,
      "step": 85420
    },
    {
      "epoch": 34.97,
      "grad_norm": 2.3384666442871094,
      "learning_rate": 4.4905713635310975e-06,
      "loss": 0.5567,
      "step": 85430
    },
    {
      "epoch": 34.97,
      "grad_norm": 2.595944881439209,
      "learning_rate": 4.490507711598637e-06,
      "loss": 0.5788,
      "step": 85440
    },
    {
      "epoch": 34.98,
      "grad_norm": 2.487816572189331,
      "learning_rate": 4.490444049832213e-06,
      "loss": 0.5828,
      "step": 85450
    },
    {
      "epoch": 34.98,
      "grad_norm": 1.9130275249481201,
      "learning_rate": 4.490380378232113e-06,
      "loss": 0.5768,
      "step": 85460
    },
    {
      "epoch": 34.99,
      "grad_norm": 2.07182240486145,
      "learning_rate": 4.490316696798634e-06,
      "loss": 0.5637,
      "step": 85470
    },
    {
      "epoch": 34.99,
      "grad_norm": 2.266730308532715,
      "learning_rate": 4.490253005532065e-06,
      "loss": 0.5695,
      "step": 85480
    },
    {
      "epoch": 34.99,
      "grad_norm": 1.9226611852645874,
      "learning_rate": 4.490189304432699e-06,
      "loss": 0.5562,
      "step": 85490
    },
    {
      "epoch": 35.0,
      "grad_norm": 2.9584028720855713,
      "learning_rate": 4.490125593500826e-06,
      "loss": 0.568,
      "step": 85500
    },
    {
      "epoch": 35.0,
      "eval_loss": 0.5695931315422058,
      "eval_runtime": 51.8769,
      "eval_samples_per_second": 66.484,
      "eval_steps_per_second": 8.327,
      "step": 85505
    },
    {
      "epoch": 35.0,
      "grad_norm": 1.832360863685608,
      "learning_rate": 4.4900618727367384e-06,
      "loss": 0.5614,
      "step": 85510
    },
    {
      "epoch": 35.01,
      "grad_norm": 2.1999869346618652,
      "learning_rate": 4.489998142140729e-06,
      "loss": 0.5852,
      "step": 85520
    },
    {
      "epoch": 35.01,
      "grad_norm": 2.6158065795898438,
      "learning_rate": 4.48993440171309e-06,
      "loss": 0.5641,
      "step": 85530
    },
    {
      "epoch": 35.01,
      "grad_norm": 2.376134157180786,
      "learning_rate": 4.489870651454112e-06,
      "loss": 0.5668,
      "step": 85540
    },
    {
      "epoch": 35.02,
      "grad_norm": 2.346254348754883,
      "learning_rate": 4.489806891364086e-06,
      "loss": 0.5917,
      "step": 85550
    },
    {
      "epoch": 35.02,
      "grad_norm": 2.0877840518951416,
      "learning_rate": 4.489743121443308e-06,
      "loss": 0.5509,
      "step": 85560
    },
    {
      "epoch": 35.03,
      "grad_norm": 3.5030322074890137,
      "learning_rate": 4.489679341692067e-06,
      "loss": 0.5807,
      "step": 85570
    },
    {
      "epoch": 35.03,
      "grad_norm": 1.75956130027771,
      "learning_rate": 4.489615552110656e-06,
      "loss": 0.5645,
      "step": 85580
    },
    {
      "epoch": 35.03,
      "grad_norm": 2.9784343242645264,
      "learning_rate": 4.489551752699368e-06,
      "loss": 0.5692,
      "step": 85590
    },
    {
      "epoch": 35.04,
      "grad_norm": 2.0631985664367676,
      "learning_rate": 4.489487943458494e-06,
      "loss": 0.5686,
      "step": 85600
    },
    {
      "epoch": 35.04,
      "grad_norm": 2.0291426181793213,
      "learning_rate": 4.489424124388327e-06,
      "loss": 0.5745,
      "step": 85610
    },
    {
      "epoch": 35.05,
      "grad_norm": 2.3433918952941895,
      "learning_rate": 4.489360295489158e-06,
      "loss": 0.5435,
      "step": 85620
    },
    {
      "epoch": 35.05,
      "grad_norm": 2.9505550861358643,
      "learning_rate": 4.489296456761281e-06,
      "loss": 0.5814,
      "step": 85630
    },
    {
      "epoch": 35.06,
      "grad_norm": 2.998577356338501,
      "learning_rate": 4.489232608204989e-06,
      "loss": 0.5649,
      "step": 85640
    },
    {
      "epoch": 35.06,
      "grad_norm": 2.184720754623413,
      "learning_rate": 4.489168749820573e-06,
      "loss": 0.56,
      "step": 85650
    },
    {
      "epoch": 35.06,
      "grad_norm": 3.0225226879119873,
      "learning_rate": 4.489104881608325e-06,
      "loss": 0.5819,
      "step": 85660
    },
    {
      "epoch": 35.07,
      "grad_norm": 2.3041024208068848,
      "learning_rate": 4.48904100356854e-06,
      "loss": 0.5716,
      "step": 85670
    },
    {
      "epoch": 35.07,
      "grad_norm": 2.3456907272338867,
      "learning_rate": 4.488977115701507e-06,
      "loss": 0.5724,
      "step": 85680
    },
    {
      "epoch": 35.08,
      "grad_norm": 2.702995777130127,
      "learning_rate": 4.488913218007522e-06,
      "loss": 0.5709,
      "step": 85690
    },
    {
      "epoch": 35.08,
      "grad_norm": 2.237413167953491,
      "learning_rate": 4.4888493104868776e-06,
      "loss": 0.5691,
      "step": 85700
    },
    {
      "epoch": 35.08,
      "grad_norm": 2.757506847381592,
      "learning_rate": 4.488785393139865e-06,
      "loss": 0.5572,
      "step": 85710
    },
    {
      "epoch": 35.09,
      "grad_norm": 2.345628023147583,
      "learning_rate": 4.488721465966777e-06,
      "loss": 0.5518,
      "step": 85720
    },
    {
      "epoch": 35.09,
      "grad_norm": 2.345705270767212,
      "learning_rate": 4.488657528967906e-06,
      "loss": 0.57,
      "step": 85730
    },
    {
      "epoch": 35.1,
      "grad_norm": 1.8203827142715454,
      "learning_rate": 4.488593582143548e-06,
      "loss": 0.565,
      "step": 85740
    },
    {
      "epoch": 35.1,
      "grad_norm": 3.692695140838623,
      "learning_rate": 4.488529625493992e-06,
      "loss": 0.5535,
      "step": 85750
    },
    {
      "epoch": 35.1,
      "grad_norm": 2.136023759841919,
      "learning_rate": 4.488465659019534e-06,
      "loss": 0.5626,
      "step": 85760
    },
    {
      "epoch": 35.11,
      "grad_norm": 2.380319118499756,
      "learning_rate": 4.488401682720464e-06,
      "loss": 0.5504,
      "step": 85770
    },
    {
      "epoch": 35.11,
      "grad_norm": 1.8866980075836182,
      "learning_rate": 4.488337696597077e-06,
      "loss": 0.5622,
      "step": 85780
    },
    {
      "epoch": 35.12,
      "grad_norm": 1.9375782012939453,
      "learning_rate": 4.488273700649667e-06,
      "loss": 0.5799,
      "step": 85790
    },
    {
      "epoch": 35.12,
      "grad_norm": 2.4091367721557617,
      "learning_rate": 4.488209694878526e-06,
      "loss": 0.5589,
      "step": 85800
    },
    {
      "epoch": 35.12,
      "grad_norm": 2.840338945388794,
      "learning_rate": 4.488145679283947e-06,
      "loss": 0.5734,
      "step": 85810
    },
    {
      "epoch": 35.13,
      "grad_norm": 2.1664934158325195,
      "learning_rate": 4.488081653866223e-06,
      "loss": 0.576,
      "step": 85820
    },
    {
      "epoch": 35.13,
      "grad_norm": 2.8238461017608643,
      "learning_rate": 4.4880176186256476e-06,
      "loss": 0.5691,
      "step": 85830
    },
    {
      "epoch": 35.14,
      "grad_norm": 1.9636839628219604,
      "learning_rate": 4.4879535735625155e-06,
      "loss": 0.5509,
      "step": 85840
    },
    {
      "epoch": 35.14,
      "grad_norm": 2.8839290142059326,
      "learning_rate": 4.487889518677118e-06,
      "loss": 0.5543,
      "step": 85850
    },
    {
      "epoch": 35.15,
      "grad_norm": 2.287278175354004,
      "learning_rate": 4.48782545396975e-06,
      "loss": 0.5699,
      "step": 85860
    },
    {
      "epoch": 35.15,
      "grad_norm": 3.077028751373291,
      "learning_rate": 4.487761379440704e-06,
      "loss": 0.5475,
      "step": 85870
    },
    {
      "epoch": 35.15,
      "grad_norm": 3.103543281555176,
      "learning_rate": 4.4876972950902745e-06,
      "loss": 0.5772,
      "step": 85880
    },
    {
      "epoch": 35.16,
      "grad_norm": 2.6175782680511475,
      "learning_rate": 4.487633200918754e-06,
      "loss": 0.5783,
      "step": 85890
    },
    {
      "epoch": 35.16,
      "grad_norm": 2.005488872528076,
      "learning_rate": 4.487569096926437e-06,
      "loss": 0.5735,
      "step": 85900
    },
    {
      "epoch": 35.17,
      "grad_norm": 2.2169923782348633,
      "learning_rate": 4.487504983113615e-06,
      "loss": 0.5577,
      "step": 85910
    },
    {
      "epoch": 35.17,
      "grad_norm": 2.1295483112335205,
      "learning_rate": 4.487440859480585e-06,
      "loss": 0.5522,
      "step": 85920
    },
    {
      "epoch": 35.17,
      "grad_norm": 2.4763426780700684,
      "learning_rate": 4.487376726027639e-06,
      "loss": 0.5811,
      "step": 85930
    },
    {
      "epoch": 35.18,
      "grad_norm": 2.338071823120117,
      "learning_rate": 4.487312582755071e-06,
      "loss": 0.5558,
      "step": 85940
    },
    {
      "epoch": 35.18,
      "grad_norm": 1.8255925178527832,
      "learning_rate": 4.4872484296631745e-06,
      "loss": 0.559,
      "step": 85950
    },
    {
      "epoch": 35.19,
      "grad_norm": 1.5998831987380981,
      "learning_rate": 4.4871842667522435e-06,
      "loss": 0.5582,
      "step": 85960
    },
    {
      "epoch": 35.19,
      "grad_norm": 2.5722827911376953,
      "learning_rate": 4.487120094022573e-06,
      "loss": 0.5744,
      "step": 85970
    },
    {
      "epoch": 35.19,
      "grad_norm": 2.4407081604003906,
      "learning_rate": 4.487055911474455e-06,
      "loss": 0.5614,
      "step": 85980
    },
    {
      "epoch": 35.2,
      "grad_norm": 2.368853807449341,
      "learning_rate": 4.4869917191081844e-06,
      "loss": 0.5739,
      "step": 85990
    },
    {
      "epoch": 35.2,
      "grad_norm": 3.2641196250915527,
      "learning_rate": 4.486927516924056e-06,
      "loss": 0.558,
      "step": 86000
    },
    {
      "epoch": 35.21,
      "grad_norm": 2.514544725418091,
      "learning_rate": 4.486863304922364e-06,
      "loss": 0.5832,
      "step": 86010
    },
    {
      "epoch": 35.21,
      "grad_norm": 2.601128339767456,
      "learning_rate": 4.486799083103401e-06,
      "loss": 0.5512,
      "step": 86020
    },
    {
      "epoch": 35.21,
      "grad_norm": 2.2393136024475098,
      "learning_rate": 4.486734851467462e-06,
      "loss": 0.554,
      "step": 86030
    },
    {
      "epoch": 35.22,
      "grad_norm": 1.6036810874938965,
      "learning_rate": 4.486670610014841e-06,
      "loss": 0.5445,
      "step": 86040
    },
    {
      "epoch": 35.22,
      "grad_norm": 2.4681179523468018,
      "learning_rate": 4.486606358745834e-06,
      "loss": 0.5643,
      "step": 86050
    },
    {
      "epoch": 35.23,
      "grad_norm": 2.3685991764068604,
      "learning_rate": 4.486542097660732e-06,
      "loss": 0.5777,
      "step": 86060
    },
    {
      "epoch": 35.23,
      "grad_norm": 2.00649094581604,
      "learning_rate": 4.486477826759833e-06,
      "loss": 0.5705,
      "step": 86070
    },
    {
      "epoch": 35.24,
      "grad_norm": 2.6468114852905273,
      "learning_rate": 4.486413546043428e-06,
      "loss": 0.5849,
      "step": 86080
    },
    {
      "epoch": 35.24,
      "grad_norm": 2.431558847427368,
      "learning_rate": 4.486349255511814e-06,
      "loss": 0.5591,
      "step": 86090
    },
    {
      "epoch": 35.24,
      "grad_norm": 1.8529571294784546,
      "learning_rate": 4.486284955165284e-06,
      "loss": 0.5889,
      "step": 86100
    },
    {
      "epoch": 35.25,
      "grad_norm": 2.1349189281463623,
      "learning_rate": 4.486220645004135e-06,
      "loss": 0.5705,
      "step": 86110
    },
    {
      "epoch": 35.25,
      "grad_norm": 2.618565559387207,
      "learning_rate": 4.486156325028658e-06,
      "loss": 0.5635,
      "step": 86120
    },
    {
      "epoch": 35.26,
      "grad_norm": 2.276939868927002,
      "learning_rate": 4.48609199523915e-06,
      "loss": 0.5536,
      "step": 86130
    },
    {
      "epoch": 35.26,
      "grad_norm": 2.1186280250549316,
      "learning_rate": 4.486027655635905e-06,
      "loss": 0.5477,
      "step": 86140
    },
    {
      "epoch": 35.26,
      "grad_norm": 2.2937052249908447,
      "learning_rate": 4.485963306219218e-06,
      "loss": 0.588,
      "step": 86150
    },
    {
      "epoch": 35.27,
      "grad_norm": 2.7797553539276123,
      "learning_rate": 4.4858989469893835e-06,
      "loss": 0.5473,
      "step": 86160
    },
    {
      "epoch": 35.27,
      "grad_norm": 2.1971044540405273,
      "learning_rate": 4.485834577946696e-06,
      "loss": 0.5625,
      "step": 86170
    },
    {
      "epoch": 35.28,
      "grad_norm": 2.761702060699463,
      "learning_rate": 4.485770199091451e-06,
      "loss": 0.5766,
      "step": 86180
    },
    {
      "epoch": 35.28,
      "grad_norm": 2.438554286956787,
      "learning_rate": 4.485705810423944e-06,
      "loss": 0.5356,
      "step": 86190
    },
    {
      "epoch": 35.28,
      "grad_norm": 2.295199394226074,
      "learning_rate": 4.485641411944468e-06,
      "loss": 0.588,
      "step": 86200
    },
    {
      "epoch": 35.29,
      "grad_norm": 2.4342663288116455,
      "learning_rate": 4.48557700365332e-06,
      "loss": 0.5611,
      "step": 86210
    },
    {
      "epoch": 35.29,
      "grad_norm": 2.36822772026062,
      "learning_rate": 4.485512585550794e-06,
      "loss": 0.5626,
      "step": 86220
    },
    {
      "epoch": 35.3,
      "grad_norm": 2.7322072982788086,
      "learning_rate": 4.485448157637185e-06,
      "loss": 0.5445,
      "step": 86230
    },
    {
      "epoch": 35.3,
      "grad_norm": 2.3436567783355713,
      "learning_rate": 4.485383719912788e-06,
      "loss": 0.5799,
      "step": 86240
    },
    {
      "epoch": 35.3,
      "grad_norm": 2.188225030899048,
      "learning_rate": 4.4853192723779e-06,
      "loss": 0.5536,
      "step": 86250
    },
    {
      "epoch": 35.31,
      "grad_norm": 2.027200937271118,
      "learning_rate": 4.485254815032813e-06,
      "loss": 0.5562,
      "step": 86260
    },
    {
      "epoch": 35.31,
      "grad_norm": 1.8918648958206177,
      "learning_rate": 4.4851903478778265e-06,
      "loss": 0.5721,
      "step": 86270
    },
    {
      "epoch": 35.32,
      "grad_norm": 2.3493611812591553,
      "learning_rate": 4.485125870913232e-06,
      "loss": 0.5731,
      "step": 86280
    },
    {
      "epoch": 35.32,
      "grad_norm": 2.7626638412475586,
      "learning_rate": 4.485061384139326e-06,
      "loss": 0.5717,
      "step": 86290
    },
    {
      "epoch": 35.33,
      "grad_norm": 2.719089984893799,
      "learning_rate": 4.484996887556405e-06,
      "loss": 0.565,
      "step": 86300
    },
    {
      "epoch": 35.33,
      "grad_norm": 2.1661148071289062,
      "learning_rate": 4.484932381164763e-06,
      "loss": 0.5721,
      "step": 86310
    },
    {
      "epoch": 35.33,
      "grad_norm": 2.440931558609009,
      "learning_rate": 4.484867864964696e-06,
      "loss": 0.5732,
      "step": 86320
    },
    {
      "epoch": 35.34,
      "grad_norm": 1.9979538917541504,
      "learning_rate": 4.4848033389565e-06,
      "loss": 0.5431,
      "step": 86330
    },
    {
      "epoch": 35.34,
      "grad_norm": 2.368234157562256,
      "learning_rate": 4.48473880314047e-06,
      "loss": 0.5585,
      "step": 86340
    },
    {
      "epoch": 35.35,
      "grad_norm": 2.583651542663574,
      "learning_rate": 4.4846742575169016e-06,
      "loss": 0.585,
      "step": 86350
    },
    {
      "epoch": 35.35,
      "grad_norm": 2.2605514526367188,
      "learning_rate": 4.484609702086092e-06,
      "loss": 0.5558,
      "step": 86360
    },
    {
      "epoch": 35.35,
      "grad_norm": 2.5415518283843994,
      "learning_rate": 4.484545136848334e-06,
      "loss": 0.5501,
      "step": 86370
    },
    {
      "epoch": 35.36,
      "grad_norm": 2.560800075531006,
      "learning_rate": 4.484480561803927e-06,
      "loss": 0.5479,
      "step": 86380
    },
    {
      "epoch": 35.36,
      "grad_norm": 2.2624475955963135,
      "learning_rate": 4.484415976953163e-06,
      "loss": 0.5741,
      "step": 86390
    },
    {
      "epoch": 35.37,
      "grad_norm": 2.6283161640167236,
      "learning_rate": 4.48435138229634e-06,
      "loss": 0.5459,
      "step": 86400
    },
    {
      "epoch": 35.37,
      "grad_norm": 2.2869858741760254,
      "learning_rate": 4.484286777833753e-06,
      "loss": 0.5578,
      "step": 86410
    },
    {
      "epoch": 35.37,
      "grad_norm": 2.3615834712982178,
      "learning_rate": 4.4842221635657005e-06,
      "loss": 0.5762,
      "step": 86420
    },
    {
      "epoch": 35.38,
      "grad_norm": 2.5102899074554443,
      "learning_rate": 4.4841575394924756e-06,
      "loss": 0.5546,
      "step": 86430
    },
    {
      "epoch": 35.38,
      "grad_norm": 2.361090898513794,
      "learning_rate": 4.484092905614375e-06,
      "loss": 0.5869,
      "step": 86440
    },
    {
      "epoch": 35.39,
      "grad_norm": 1.9407821893692017,
      "learning_rate": 4.484028261931695e-06,
      "loss": 0.5548,
      "step": 86450
    },
    {
      "epoch": 35.39,
      "grad_norm": 1.9426100254058838,
      "learning_rate": 4.483963608444731e-06,
      "loss": 0.5695,
      "step": 86460
    },
    {
      "epoch": 35.4,
      "grad_norm": 2.6036665439605713,
      "learning_rate": 4.483898945153782e-06,
      "loss": 0.5405,
      "step": 86470
    },
    {
      "epoch": 35.4,
      "grad_norm": 2.8699076175689697,
      "learning_rate": 4.483834272059141e-06,
      "loss": 0.5568,
      "step": 86480
    },
    {
      "epoch": 35.4,
      "grad_norm": 2.270376443862915,
      "learning_rate": 4.483769589161105e-06,
      "loss": 0.5657,
      "step": 86490
    },
    {
      "epoch": 35.41,
      "grad_norm": 2.004094362258911,
      "learning_rate": 4.483704896459971e-06,
      "loss": 0.5507,
      "step": 86500
    },
    {
      "epoch": 35.41,
      "grad_norm": 1.9624602794647217,
      "learning_rate": 4.483640193956035e-06,
      "loss": 0.5666,
      "step": 86510
    },
    {
      "epoch": 35.42,
      "grad_norm": 2.553138017654419,
      "learning_rate": 4.483575481649593e-06,
      "loss": 0.5514,
      "step": 86520
    },
    {
      "epoch": 35.42,
      "grad_norm": 2.420971393585205,
      "learning_rate": 4.4835107595409424e-06,
      "loss": 0.5667,
      "step": 86530
    },
    {
      "epoch": 35.42,
      "grad_norm": 2.2938780784606934,
      "learning_rate": 4.483446027630379e-06,
      "loss": 0.5554,
      "step": 86540
    },
    {
      "epoch": 35.43,
      "grad_norm": 2.620809316635132,
      "learning_rate": 4.4833812859182e-06,
      "loss": 0.5664,
      "step": 86550
    },
    {
      "epoch": 35.43,
      "grad_norm": 3.3682641983032227,
      "learning_rate": 4.4833165344047005e-06,
      "loss": 0.538,
      "step": 86560
    },
    {
      "epoch": 35.44,
      "grad_norm": 1.748834252357483,
      "learning_rate": 4.483251773090179e-06,
      "loss": 0.5404,
      "step": 86570
    },
    {
      "epoch": 35.44,
      "grad_norm": 2.0325212478637695,
      "learning_rate": 4.48318700197493e-06,
      "loss": 0.5677,
      "step": 86580
    },
    {
      "epoch": 35.44,
      "grad_norm": 2.2087655067443848,
      "learning_rate": 4.483122221059253e-06,
      "loss": 0.5691,
      "step": 86590
    },
    {
      "epoch": 35.45,
      "grad_norm": 2.2869017124176025,
      "learning_rate": 4.483057430343442e-06,
      "loss": 0.5559,
      "step": 86600
    },
    {
      "epoch": 35.45,
      "grad_norm": 2.3221323490142822,
      "learning_rate": 4.4829926298277955e-06,
      "loss": 0.5654,
      "step": 86610
    },
    {
      "epoch": 35.46,
      "grad_norm": 1.888546109199524,
      "learning_rate": 4.482927819512609e-06,
      "loss": 0.5488,
      "step": 86620
    },
    {
      "epoch": 35.46,
      "grad_norm": 2.8151724338531494,
      "learning_rate": 4.482862999398181e-06,
      "loss": 0.5703,
      "step": 86630
    },
    {
      "epoch": 35.46,
      "grad_norm": 2.50296688079834,
      "learning_rate": 4.482798169484808e-06,
      "loss": 0.5667,
      "step": 86640
    },
    {
      "epoch": 35.47,
      "grad_norm": 2.2808542251586914,
      "learning_rate": 4.482733329772786e-06,
      "loss": 0.576,
      "step": 86650
    },
    {
      "epoch": 35.47,
      "grad_norm": 2.1046769618988037,
      "learning_rate": 4.482668480262413e-06,
      "loss": 0.5849,
      "step": 86660
    },
    {
      "epoch": 35.48,
      "grad_norm": 2.484034776687622,
      "learning_rate": 4.4826036209539845e-06,
      "loss": 0.5927,
      "step": 86670
    },
    {
      "epoch": 35.48,
      "grad_norm": 2.6181235313415527,
      "learning_rate": 4.4825387518478e-06,
      "loss": 0.5675,
      "step": 86680
    },
    {
      "epoch": 35.49,
      "grad_norm": 2.049117088317871,
      "learning_rate": 4.482473872944155e-06,
      "loss": 0.5712,
      "step": 86690
    },
    {
      "epoch": 35.49,
      "grad_norm": 2.6154611110687256,
      "learning_rate": 4.4824089842433475e-06,
      "loss": 0.56,
      "step": 86700
    },
    {
      "epoch": 35.49,
      "grad_norm": 2.180521249771118,
      "learning_rate": 4.4823440857456744e-06,
      "loss": 0.5838,
      "step": 86710
    },
    {
      "epoch": 35.5,
      "grad_norm": 2.186352491378784,
      "learning_rate": 4.4822791774514326e-06,
      "loss": 0.575,
      "step": 86720
    },
    {
      "epoch": 35.5,
      "grad_norm": 2.8667221069335938,
      "learning_rate": 4.48221425936092e-06,
      "loss": 0.5682,
      "step": 86730
    },
    {
      "epoch": 35.51,
      "grad_norm": 2.296041965484619,
      "learning_rate": 4.482149331474434e-06,
      "loss": 0.5672,
      "step": 86740
    },
    {
      "epoch": 35.51,
      "grad_norm": 2.0413668155670166,
      "learning_rate": 4.482084393792272e-06,
      "loss": 0.5722,
      "step": 86750
    },
    {
      "epoch": 35.51,
      "grad_norm": 2.853044033050537,
      "learning_rate": 4.48201944631473e-06,
      "loss": 0.5718,
      "step": 86760
    },
    {
      "epoch": 35.52,
      "grad_norm": 2.53739595413208,
      "learning_rate": 4.4819544890421086e-06,
      "loss": 0.5711,
      "step": 86770
    },
    {
      "epoch": 35.52,
      "grad_norm": 2.1747868061065674,
      "learning_rate": 4.481889521974703e-06,
      "loss": 0.5699,
      "step": 86780
    },
    {
      "epoch": 35.53,
      "grad_norm": 1.8861708641052246,
      "learning_rate": 4.481824545112812e-06,
      "loss": 0.5667,
      "step": 86790
    },
    {
      "epoch": 35.53,
      "grad_norm": 2.5131618976593018,
      "learning_rate": 4.481759558456731e-06,
      "loss": 0.5734,
      "step": 86800
    },
    {
      "epoch": 35.53,
      "grad_norm": 2.1002299785614014,
      "learning_rate": 4.4816945620067595e-06,
      "loss": 0.5905,
      "step": 86810
    },
    {
      "epoch": 35.54,
      "grad_norm": 1.988388180732727,
      "learning_rate": 4.481629555763196e-06,
      "loss": 0.5788,
      "step": 86820
    },
    {
      "epoch": 35.54,
      "grad_norm": 2.141496419906616,
      "learning_rate": 4.481564539726337e-06,
      "loss": 0.5785,
      "step": 86830
    },
    {
      "epoch": 35.55,
      "grad_norm": 2.97581148147583,
      "learning_rate": 4.4814995138964815e-06,
      "loss": 0.5431,
      "step": 86840
    },
    {
      "epoch": 35.55,
      "grad_norm": 2.541496992111206,
      "learning_rate": 4.481434478273925e-06,
      "loss": 0.5476,
      "step": 86850
    },
    {
      "epoch": 35.55,
      "grad_norm": 2.357070207595825,
      "learning_rate": 4.481369432858968e-06,
      "loss": 0.5697,
      "step": 86860
    },
    {
      "epoch": 35.56,
      "grad_norm": 2.325693368911743,
      "learning_rate": 4.481304377651907e-06,
      "loss": 0.5379,
      "step": 86870
    },
    {
      "epoch": 35.56,
      "grad_norm": 2.436344861984253,
      "learning_rate": 4.481239312653041e-06,
      "loss": 0.5384,
      "step": 86880
    },
    {
      "epoch": 35.57,
      "grad_norm": 2.6130597591400146,
      "learning_rate": 4.481174237862667e-06,
      "loss": 0.5726,
      "step": 86890
    },
    {
      "epoch": 35.57,
      "grad_norm": 2.316218614578247,
      "learning_rate": 4.4811091532810835e-06,
      "loss": 0.5501,
      "step": 86900
    },
    {
      "epoch": 35.58,
      "grad_norm": 2.014246702194214,
      "learning_rate": 4.481044058908589e-06,
      "loss": 0.5705,
      "step": 86910
    },
    {
      "epoch": 35.58,
      "grad_norm": 2.230898380279541,
      "learning_rate": 4.480978954745482e-06,
      "loss": 0.5714,
      "step": 86920
    },
    {
      "epoch": 35.58,
      "grad_norm": 2.3832898139953613,
      "learning_rate": 4.480913840792059e-06,
      "loss": 0.5655,
      "step": 86930
    },
    {
      "epoch": 35.59,
      "grad_norm": 2.3953359127044678,
      "learning_rate": 4.4808487170486205e-06,
      "loss": 0.5719,
      "step": 86940
    },
    {
      "epoch": 35.59,
      "grad_norm": 1.8235281705856323,
      "learning_rate": 4.4807835835154625e-06,
      "loss": 0.5289,
      "step": 86950
    },
    {
      "epoch": 35.6,
      "grad_norm": 1.7906370162963867,
      "learning_rate": 4.480718440192886e-06,
      "loss": 0.5826,
      "step": 86960
    },
    {
      "epoch": 35.6,
      "grad_norm": 2.2216243743896484,
      "learning_rate": 4.480653287081186e-06,
      "loss": 0.5675,
      "step": 86970
    },
    {
      "epoch": 35.6,
      "grad_norm": 2.1871707439422607,
      "learning_rate": 4.480588124180665e-06,
      "loss": 0.5758,
      "step": 86980
    },
    {
      "epoch": 35.61,
      "grad_norm": 2.653785467147827,
      "learning_rate": 4.480522951491618e-06,
      "loss": 0.5502,
      "step": 86990
    },
    {
      "epoch": 35.61,
      "grad_norm": 3.013496160507202,
      "learning_rate": 4.480457769014346e-06,
      "loss": 0.5655,
      "step": 87000
    },
    {
      "epoch": 35.62,
      "grad_norm": 1.8090764284133911,
      "learning_rate": 4.480392576749147e-06,
      "loss": 0.5519,
      "step": 87010
    },
    {
      "epoch": 35.62,
      "grad_norm": 2.3103718757629395,
      "learning_rate": 4.480327374696319e-06,
      "loss": 0.5656,
      "step": 87020
    },
    {
      "epoch": 35.62,
      "grad_norm": 2.26910400390625,
      "learning_rate": 4.48026216285616e-06,
      "loss": 0.5714,
      "step": 87030
    },
    {
      "epoch": 35.63,
      "grad_norm": 2.4086596965789795,
      "learning_rate": 4.4801969412289705e-06,
      "loss": 0.5756,
      "step": 87040
    },
    {
      "epoch": 35.63,
      "grad_norm": 2.1010608673095703,
      "learning_rate": 4.480131709815048e-06,
      "loss": 0.5698,
      "step": 87050
    },
    {
      "epoch": 35.64,
      "grad_norm": 2.452174186706543,
      "learning_rate": 4.480066468614692e-06,
      "loss": 0.5585,
      "step": 87060
    },
    {
      "epoch": 35.64,
      "grad_norm": 2.4448165893554688,
      "learning_rate": 4.480001217628201e-06,
      "loss": 0.5453,
      "step": 87070
    },
    {
      "epoch": 35.64,
      "grad_norm": 2.09761118888855,
      "learning_rate": 4.479935956855874e-06,
      "loss": 0.5632,
      "step": 87080
    },
    {
      "epoch": 35.65,
      "grad_norm": 2.1126067638397217,
      "learning_rate": 4.479870686298011e-06,
      "loss": 0.5636,
      "step": 87090
    },
    {
      "epoch": 35.65,
      "grad_norm": 2.2350120544433594,
      "learning_rate": 4.479805405954908e-06,
      "loss": 0.5666,
      "step": 87100
    },
    {
      "epoch": 35.66,
      "grad_norm": 2.9200339317321777,
      "learning_rate": 4.479740115826867e-06,
      "loss": 0.588,
      "step": 87110
    },
    {
      "epoch": 35.66,
      "grad_norm": 2.6417012214660645,
      "learning_rate": 4.479674815914187e-06,
      "loss": 0.5671,
      "step": 87120
    },
    {
      "epoch": 35.67,
      "grad_norm": 2.4416627883911133,
      "learning_rate": 4.479609506217165e-06,
      "loss": 0.5603,
      "step": 87130
    },
    {
      "epoch": 35.67,
      "grad_norm": 3.5627472400665283,
      "learning_rate": 4.479544186736102e-06,
      "loss": 0.5743,
      "step": 87140
    },
    {
      "epoch": 35.67,
      "grad_norm": 3.0720696449279785,
      "learning_rate": 4.479478857471296e-06,
      "loss": 0.5722,
      "step": 87150
    },
    {
      "epoch": 35.68,
      "grad_norm": 2.662715435028076,
      "learning_rate": 4.479413518423046e-06,
      "loss": 0.5616,
      "step": 87160
    },
    {
      "epoch": 35.68,
      "grad_norm": 1.7154282331466675,
      "learning_rate": 4.479348169591654e-06,
      "loss": 0.5816,
      "step": 87170
    },
    {
      "epoch": 35.69,
      "grad_norm": 2.1513354778289795,
      "learning_rate": 4.479282810977417e-06,
      "loss": 0.5696,
      "step": 87180
    },
    {
      "epoch": 35.69,
      "grad_norm": 2.194183349609375,
      "learning_rate": 4.479217442580635e-06,
      "loss": 0.5619,
      "step": 87190
    },
    {
      "epoch": 35.69,
      "grad_norm": 2.363264799118042,
      "learning_rate": 4.479152064401607e-06,
      "loss": 0.5558,
      "step": 87200
    },
    {
      "epoch": 35.7,
      "grad_norm": 2.3988232612609863,
      "learning_rate": 4.479086676440633e-06,
      "loss": 0.5357,
      "step": 87210
    },
    {
      "epoch": 35.7,
      "grad_norm": 1.8573381900787354,
      "learning_rate": 4.479021278698012e-06,
      "loss": 0.5587,
      "step": 87220
    },
    {
      "epoch": 35.71,
      "grad_norm": 3.228128671646118,
      "learning_rate": 4.4789558711740444e-06,
      "loss": 0.5644,
      "step": 87230
    },
    {
      "epoch": 35.71,
      "grad_norm": 1.7322030067443848,
      "learning_rate": 4.47889045386903e-06,
      "loss": 0.5596,
      "step": 87240
    },
    {
      "epoch": 35.71,
      "grad_norm": 2.263268232345581,
      "learning_rate": 4.478825026783267e-06,
      "loss": 0.5542,
      "step": 87250
    },
    {
      "epoch": 35.72,
      "grad_norm": 2.2331762313842773,
      "learning_rate": 4.478759589917056e-06,
      "loss": 0.5653,
      "step": 87260
    },
    {
      "epoch": 35.72,
      "grad_norm": 2.4775142669677734,
      "learning_rate": 4.478694143270696e-06,
      "loss": 0.5513,
      "step": 87270
    },
    {
      "epoch": 35.73,
      "grad_norm": 2.1893699169158936,
      "learning_rate": 4.478628686844489e-06,
      "loss": 0.5571,
      "step": 87280
    },
    {
      "epoch": 35.73,
      "grad_norm": 2.039356231689453,
      "learning_rate": 4.478563220638733e-06,
      "loss": 0.5593,
      "step": 87290
    },
    {
      "epoch": 35.73,
      "grad_norm": 2.357783794403076,
      "learning_rate": 4.478497744653728e-06,
      "loss": 0.5893,
      "step": 87300
    },
    {
      "epoch": 35.74,
      "grad_norm": 2.516711473464966,
      "learning_rate": 4.4784322588897735e-06,
      "loss": 0.58,
      "step": 87310
    },
    {
      "epoch": 35.74,
      "grad_norm": 1.8585165739059448,
      "learning_rate": 4.478366763347171e-06,
      "loss": 0.5615,
      "step": 87320
    },
    {
      "epoch": 35.75,
      "grad_norm": 1.8747690916061401,
      "learning_rate": 4.478301258026219e-06,
      "loss": 0.5707,
      "step": 87330
    },
    {
      "epoch": 35.75,
      "grad_norm": 2.0323450565338135,
      "learning_rate": 4.478235742927219e-06,
      "loss": 0.5619,
      "step": 87340
    },
    {
      "epoch": 35.76,
      "grad_norm": 1.880407691001892,
      "learning_rate": 4.47817021805047e-06,
      "loss": 0.5592,
      "step": 87350
    },
    {
      "epoch": 35.76,
      "grad_norm": 2.175931453704834,
      "learning_rate": 4.478104683396273e-06,
      "loss": 0.578,
      "step": 87360
    },
    {
      "epoch": 35.76,
      "grad_norm": 2.3377809524536133,
      "learning_rate": 4.478039138964928e-06,
      "loss": 0.547,
      "step": 87370
    },
    {
      "epoch": 35.77,
      "grad_norm": 2.140963554382324,
      "learning_rate": 4.477973584756734e-06,
      "loss": 0.5688,
      "step": 87380
    },
    {
      "epoch": 35.77,
      "grad_norm": 2.0462427139282227,
      "learning_rate": 4.477908020771993e-06,
      "loss": 0.5278,
      "step": 87390
    },
    {
      "epoch": 35.78,
      "grad_norm": 2.554647207260132,
      "learning_rate": 4.477842447011004e-06,
      "loss": 0.5686,
      "step": 87400
    },
    {
      "epoch": 35.78,
      "grad_norm": 2.434882640838623,
      "learning_rate": 4.477776863474068e-06,
      "loss": 0.573,
      "step": 87410
    },
    {
      "epoch": 35.78,
      "grad_norm": 2.2753655910491943,
      "learning_rate": 4.477711270161486e-06,
      "loss": 0.5495,
      "step": 87420
    },
    {
      "epoch": 35.79,
      "grad_norm": 2.3674087524414062,
      "learning_rate": 4.477645667073558e-06,
      "loss": 0.5695,
      "step": 87430
    },
    {
      "epoch": 35.79,
      "grad_norm": 1.9571279287338257,
      "learning_rate": 4.477580054210584e-06,
      "loss": 0.5753,
      "step": 87440
    },
    {
      "epoch": 35.8,
      "grad_norm": 2.3525161743164062,
      "learning_rate": 4.477514431572865e-06,
      "loss": 0.5707,
      "step": 87450
    },
    {
      "epoch": 35.8,
      "grad_norm": 2.79392147064209,
      "learning_rate": 4.477448799160701e-06,
      "loss": 0.5531,
      "step": 87460
    },
    {
      "epoch": 35.8,
      "grad_norm": 1.853076457977295,
      "learning_rate": 4.4773831569743945e-06,
      "loss": 0.558,
      "step": 87470
    },
    {
      "epoch": 35.81,
      "grad_norm": 2.255819082260132,
      "learning_rate": 4.477317505014244e-06,
      "loss": 0.5465,
      "step": 87480
    },
    {
      "epoch": 35.81,
      "grad_norm": 2.3820431232452393,
      "learning_rate": 4.477251843280552e-06,
      "loss": 0.5864,
      "step": 87490
    },
    {
      "epoch": 35.82,
      "grad_norm": 1.5193156003952026,
      "learning_rate": 4.477186171773618e-06,
      "loss": 0.5826,
      "step": 87500
    },
    {
      "epoch": 35.82,
      "grad_norm": 2.4957776069641113,
      "learning_rate": 4.4771204904937434e-06,
      "loss": 0.5491,
      "step": 87510
    },
    {
      "epoch": 35.82,
      "grad_norm": 2.424799680709839,
      "learning_rate": 4.477054799441229e-06,
      "loss": 0.5744,
      "step": 87520
    },
    {
      "epoch": 35.83,
      "grad_norm": 2.7837612628936768,
      "learning_rate": 4.476989098616375e-06,
      "loss": 0.5621,
      "step": 87530
    },
    {
      "epoch": 35.83,
      "grad_norm": 1.9221190214157104,
      "learning_rate": 4.476923388019484e-06,
      "loss": 0.5648,
      "step": 87540
    },
    {
      "epoch": 35.84,
      "grad_norm": 2.4253745079040527,
      "learning_rate": 4.476857667650855e-06,
      "loss": 0.59,
      "step": 87550
    },
    {
      "epoch": 35.84,
      "grad_norm": 2.3480136394500732,
      "learning_rate": 4.476791937510791e-06,
      "loss": 0.5438,
      "step": 87560
    },
    {
      "epoch": 35.85,
      "grad_norm": 2.3474512100219727,
      "learning_rate": 4.4767261975995925e-06,
      "loss": 0.5793,
      "step": 87570
    },
    {
      "epoch": 35.85,
      "grad_norm": 2.514681577682495,
      "learning_rate": 4.47666044791756e-06,
      "loss": 0.5732,
      "step": 87580
    },
    {
      "epoch": 35.85,
      "grad_norm": 3.181330919265747,
      "learning_rate": 4.4765946884649955e-06,
      "loss": 0.5617,
      "step": 87590
    },
    {
      "epoch": 35.86,
      "grad_norm": 2.2748138904571533,
      "learning_rate": 4.476528919242199e-06,
      "loss": 0.5706,
      "step": 87600
    },
    {
      "epoch": 35.86,
      "grad_norm": 2.335641860961914,
      "learning_rate": 4.476463140249472e-06,
      "loss": 0.5548,
      "step": 87610
    },
    {
      "epoch": 35.87,
      "grad_norm": 2.549197196960449,
      "learning_rate": 4.476397351487118e-06,
      "loss": 0.5599,
      "step": 87620
    },
    {
      "epoch": 35.87,
      "grad_norm": 2.627976655960083,
      "learning_rate": 4.476331552955436e-06,
      "loss": 0.5698,
      "step": 87630
    },
    {
      "epoch": 35.87,
      "grad_norm": 2.17928409576416,
      "learning_rate": 4.476265744654729e-06,
      "loss": 0.5956,
      "step": 87640
    },
    {
      "epoch": 35.88,
      "grad_norm": 1.8768956661224365,
      "learning_rate": 4.476199926585296e-06,
      "loss": 0.5619,
      "step": 87650
    },
    {
      "epoch": 35.88,
      "grad_norm": 2.004706859588623,
      "learning_rate": 4.476134098747441e-06,
      "loss": 0.5728,
      "step": 87660
    },
    {
      "epoch": 35.89,
      "grad_norm": 2.6745500564575195,
      "learning_rate": 4.476068261141465e-06,
      "loss": 0.5828,
      "step": 87670
    },
    {
      "epoch": 35.89,
      "grad_norm": 2.720985174179077,
      "learning_rate": 4.476002413767669e-06,
      "loss": 0.5725,
      "step": 87680
    },
    {
      "epoch": 35.89,
      "grad_norm": 2.6746954917907715,
      "learning_rate": 4.4759365566263544e-06,
      "loss": 0.5479,
      "step": 87690
    },
    {
      "epoch": 35.9,
      "grad_norm": 1.870742678642273,
      "learning_rate": 4.4758706897178245e-06,
      "loss": 0.5792,
      "step": 87700
    },
    {
      "epoch": 35.9,
      "grad_norm": 1.6830322742462158,
      "learning_rate": 4.475804813042379e-06,
      "loss": 0.5493,
      "step": 87710
    },
    {
      "epoch": 35.91,
      "grad_norm": 1.9418057203292847,
      "learning_rate": 4.475738926600321e-06,
      "loss": 0.5503,
      "step": 87720
    },
    {
      "epoch": 35.91,
      "grad_norm": 2.668222665786743,
      "learning_rate": 4.475673030391952e-06,
      "loss": 0.5721,
      "step": 87730
    },
    {
      "epoch": 35.91,
      "grad_norm": 2.9826855659484863,
      "learning_rate": 4.475607124417574e-06,
      "loss": 0.5571,
      "step": 87740
    },
    {
      "epoch": 35.92,
      "grad_norm": 2.7961015701293945,
      "learning_rate": 4.475541208677488e-06,
      "loss": 0.5684,
      "step": 87750
    },
    {
      "epoch": 35.92,
      "grad_norm": 2.5610172748565674,
      "learning_rate": 4.475475283171996e-06,
      "loss": 0.575,
      "step": 87760
    },
    {
      "epoch": 35.93,
      "grad_norm": 2.2418580055236816,
      "learning_rate": 4.475409347901402e-06,
      "loss": 0.555,
      "step": 87770
    },
    {
      "epoch": 35.93,
      "grad_norm": 2.835965633392334,
      "learning_rate": 4.4753434028660055e-06,
      "loss": 0.576,
      "step": 87780
    },
    {
      "epoch": 35.94,
      "grad_norm": 2.710134506225586,
      "learning_rate": 4.475277448066111e-06,
      "loss": 0.5731,
      "step": 87790
    },
    {
      "epoch": 35.94,
      "grad_norm": 1.8012471199035645,
      "learning_rate": 4.475211483502019e-06,
      "loss": 0.5599,
      "step": 87800
    },
    {
      "epoch": 35.94,
      "grad_norm": 2.7810206413269043,
      "learning_rate": 4.475145509174032e-06,
      "loss": 0.5658,
      "step": 87810
    },
    {
      "epoch": 35.95,
      "grad_norm": 2.1382811069488525,
      "learning_rate": 4.475079525082452e-06,
      "loss": 0.5596,
      "step": 87820
    },
    {
      "epoch": 35.95,
      "grad_norm": 2.1623077392578125,
      "learning_rate": 4.475013531227581e-06,
      "loss": 0.5765,
      "step": 87830
    },
    {
      "epoch": 35.96,
      "grad_norm": 2.2895545959472656,
      "learning_rate": 4.4749475276097225e-06,
      "loss": 0.569,
      "step": 87840
    },
    {
      "epoch": 35.96,
      "grad_norm": 3.4951963424682617,
      "learning_rate": 4.474881514229178e-06,
      "loss": 0.5626,
      "step": 87850
    },
    {
      "epoch": 35.96,
      "grad_norm": 2.550459384918213,
      "learning_rate": 4.47481549108625e-06,
      "loss": 0.5849,
      "step": 87860
    },
    {
      "epoch": 35.97,
      "grad_norm": 2.644376754760742,
      "learning_rate": 4.4747494581812415e-06,
      "loss": 0.5754,
      "step": 87870
    },
    {
      "epoch": 35.97,
      "grad_norm": 2.1128299236297607,
      "learning_rate": 4.474683415514454e-06,
      "loss": 0.5647,
      "step": 87880
    },
    {
      "epoch": 35.98,
      "grad_norm": 3.157797336578369,
      "learning_rate": 4.4746173630861906e-06,
      "loss": 0.5677,
      "step": 87890
    },
    {
      "epoch": 35.98,
      "grad_norm": 1.970737338066101,
      "learning_rate": 4.474551300896754e-06,
      "loss": 0.6015,
      "step": 87900
    },
    {
      "epoch": 35.98,
      "grad_norm": 2.7733473777770996,
      "learning_rate": 4.474485228946447e-06,
      "loss": 0.5666,
      "step": 87910
    },
    {
      "epoch": 35.99,
      "grad_norm": 2.2031211853027344,
      "learning_rate": 4.474419147235571e-06,
      "loss": 0.5846,
      "step": 87920
    },
    {
      "epoch": 35.99,
      "grad_norm": 2.2403371334075928,
      "learning_rate": 4.47435305576443e-06,
      "loss": 0.5608,
      "step": 87930
    },
    {
      "epoch": 36.0,
      "grad_norm": 2.6960647106170654,
      "learning_rate": 4.474286954533326e-06,
      "loss": 0.5819,
      "step": 87940
    },
    {
      "epoch": 36.0,
      "eval_loss": 0.5648784637451172,
      "eval_runtime": 51.8976,
      "eval_samples_per_second": 66.458,
      "eval_steps_per_second": 8.324,
      "step": 87948
    },
    {
      "epoch": 36.0,
      "grad_norm": 2.123206377029419,
      "learning_rate": 4.474220843542563e-06,
      "loss": 0.5752,
      "step": 87950
    },
    {
      "epoch": 36.0,
      "grad_norm": 1.9945365190505981,
      "learning_rate": 4.474154722792443e-06,
      "loss": 0.5686,
      "step": 87960
    },
    {
      "epoch": 36.01,
      "grad_norm": 2.096179723739624,
      "learning_rate": 4.474088592283268e-06,
      "loss": 0.563,
      "step": 87970
    },
    {
      "epoch": 36.01,
      "grad_norm": 3.2936463356018066,
      "learning_rate": 4.474022452015343e-06,
      "loss": 0.5782,
      "step": 87980
    },
    {
      "epoch": 36.02,
      "grad_norm": 2.3693108558654785,
      "learning_rate": 4.473956301988969e-06,
      "loss": 0.58,
      "step": 87990
    },
    {
      "epoch": 36.02,
      "grad_norm": 3.4931013584136963,
      "learning_rate": 4.47389014220445e-06,
      "loss": 0.5391,
      "step": 88000
    },
    {
      "epoch": 36.03,
      "grad_norm": 2.3363590240478516,
      "learning_rate": 4.473823972662089e-06,
      "loss": 0.5759,
      "step": 88010
    },
    {
      "epoch": 36.03,
      "grad_norm": 2.077911853790283,
      "learning_rate": 4.4737577933621895e-06,
      "loss": 0.5845,
      "step": 88020
    },
    {
      "epoch": 36.03,
      "grad_norm": 1.885273814201355,
      "learning_rate": 4.473691604305053e-06,
      "loss": 0.5553,
      "step": 88030
    },
    {
      "epoch": 36.04,
      "grad_norm": 2.6469244956970215,
      "learning_rate": 4.473625405490985e-06,
      "loss": 0.5591,
      "step": 88040
    },
    {
      "epoch": 36.04,
      "grad_norm": 1.7904694080352783,
      "learning_rate": 4.4735591969202865e-06,
      "loss": 0.5691,
      "step": 88050
    },
    {
      "epoch": 36.05,
      "grad_norm": 2.0376639366149902,
      "learning_rate": 4.473492978593262e-06,
      "loss": 0.5528,
      "step": 88060
    },
    {
      "epoch": 36.05,
      "grad_norm": 2.905411720275879,
      "learning_rate": 4.4734267505102154e-06,
      "loss": 0.5591,
      "step": 88070
    },
    {
      "epoch": 36.05,
      "grad_norm": 2.293128728866577,
      "learning_rate": 4.47336051267145e-06,
      "loss": 0.5594,
      "step": 88080
    },
    {
      "epoch": 36.06,
      "grad_norm": 3.065314531326294,
      "learning_rate": 4.473294265077268e-06,
      "loss": 0.5655,
      "step": 88090
    },
    {
      "epoch": 36.06,
      "grad_norm": 2.3504562377929688,
      "learning_rate": 4.4732280077279734e-06,
      "loss": 0.5729,
      "step": 88100
    },
    {
      "epoch": 36.07,
      "grad_norm": 3.2212603092193604,
      "learning_rate": 4.47316174062387e-06,
      "loss": 0.5418,
      "step": 88110
    },
    {
      "epoch": 36.07,
      "grad_norm": 1.9927722215652466,
      "learning_rate": 4.473095463765261e-06,
      "loss": 0.5583,
      "step": 88120
    },
    {
      "epoch": 36.07,
      "grad_norm": 2.082709550857544,
      "learning_rate": 4.4730291771524495e-06,
      "loss": 0.5439,
      "step": 88130
    },
    {
      "epoch": 36.08,
      "grad_norm": 4.452851295471191,
      "learning_rate": 4.472962880785741e-06,
      "loss": 0.559,
      "step": 88140
    },
    {
      "epoch": 36.08,
      "grad_norm": 1.66939377784729,
      "learning_rate": 4.472896574665438e-06,
      "loss": 0.5616,
      "step": 88150
    },
    {
      "epoch": 36.09,
      "grad_norm": 2.2649636268615723,
      "learning_rate": 4.472830258791844e-06,
      "loss": 0.559,
      "step": 88160
    },
    {
      "epoch": 36.09,
      "grad_norm": 2.7034149169921875,
      "learning_rate": 4.472763933165263e-06,
      "loss": 0.5568,
      "step": 88170
    },
    {
      "epoch": 36.09,
      "grad_norm": 2.1735923290252686,
      "learning_rate": 4.472697597785998e-06,
      "loss": 0.5704,
      "step": 88180
    },
    {
      "epoch": 36.1,
      "grad_norm": 2.094324827194214,
      "learning_rate": 4.472631252654356e-06,
      "loss": 0.5526,
      "step": 88190
    },
    {
      "epoch": 36.1,
      "grad_norm": 2.178032875061035,
      "learning_rate": 4.472564897770637e-06,
      "loss": 0.5552,
      "step": 88200
    },
    {
      "epoch": 36.11,
      "grad_norm": 2.2321364879608154,
      "learning_rate": 4.472498533135147e-06,
      "loss": 0.5614,
      "step": 88210
    },
    {
      "epoch": 36.11,
      "grad_norm": 2.4520742893218994,
      "learning_rate": 4.472432158748189e-06,
      "loss": 0.5646,
      "step": 88220
    },
    {
      "epoch": 36.12,
      "grad_norm": 2.660107135772705,
      "learning_rate": 4.472365774610069e-06,
      "loss": 0.5691,
      "step": 88230
    },
    {
      "epoch": 36.12,
      "grad_norm": 2.7895822525024414,
      "learning_rate": 4.472299380721089e-06,
      "loss": 0.5668,
      "step": 88240
    },
    {
      "epoch": 36.12,
      "grad_norm": 2.207852363586426,
      "learning_rate": 4.472232977081554e-06,
      "loss": 0.5606,
      "step": 88250
    },
    {
      "epoch": 36.13,
      "grad_norm": 2.1232779026031494,
      "learning_rate": 4.4721665636917685e-06,
      "loss": 0.5769,
      "step": 88260
    },
    {
      "epoch": 36.13,
      "grad_norm": 2.409306049346924,
      "learning_rate": 4.472100140552036e-06,
      "loss": 0.5612,
      "step": 88270
    },
    {
      "epoch": 36.14,
      "grad_norm": 2.2127938270568848,
      "learning_rate": 4.472033707662661e-06,
      "loss": 0.5725,
      "step": 88280
    },
    {
      "epoch": 36.14,
      "grad_norm": 2.450268507003784,
      "learning_rate": 4.471967265023949e-06,
      "loss": 0.5806,
      "step": 88290
    },
    {
      "epoch": 36.14,
      "grad_norm": 1.97626531124115,
      "learning_rate": 4.471900812636202e-06,
      "loss": 0.5689,
      "step": 88300
    },
    {
      "epoch": 36.15,
      "grad_norm": 2.276599168777466,
      "learning_rate": 4.471834350499726e-06,
      "loss": 0.5666,
      "step": 88310
    },
    {
      "epoch": 36.15,
      "grad_norm": 2.2498204708099365,
      "learning_rate": 4.4717678786148255e-06,
      "loss": 0.5777,
      "step": 88320
    },
    {
      "epoch": 36.16,
      "grad_norm": 1.9898531436920166,
      "learning_rate": 4.471701396981804e-06,
      "loss": 0.5577,
      "step": 88330
    },
    {
      "epoch": 36.16,
      "grad_norm": 2.117755651473999,
      "learning_rate": 4.471634905600968e-06,
      "loss": 0.5654,
      "step": 88340
    },
    {
      "epoch": 36.16,
      "grad_norm": 2.083737373352051,
      "learning_rate": 4.471568404472619e-06,
      "loss": 0.5601,
      "step": 88350
    },
    {
      "epoch": 36.17,
      "grad_norm": 1.7022159099578857,
      "learning_rate": 4.4715018935970645e-06,
      "loss": 0.582,
      "step": 88360
    },
    {
      "epoch": 36.17,
      "grad_norm": 2.1089718341827393,
      "learning_rate": 4.471435372974608e-06,
      "loss": 0.575,
      "step": 88370
    },
    {
      "epoch": 36.18,
      "grad_norm": 2.1210010051727295,
      "learning_rate": 4.471368842605554e-06,
      "loss": 0.5599,
      "step": 88380
    },
    {
      "epoch": 36.18,
      "grad_norm": 2.0956385135650635,
      "learning_rate": 4.471302302490208e-06,
      "loss": 0.5566,
      "step": 88390
    },
    {
      "epoch": 36.19,
      "grad_norm": 2.1308786869049072,
      "learning_rate": 4.471235752628875e-06,
      "loss": 0.5571,
      "step": 88400
    },
    {
      "epoch": 36.19,
      "grad_norm": 2.1035449504852295,
      "learning_rate": 4.4711691930218586e-06,
      "loss": 0.5571,
      "step": 88410
    },
    {
      "epoch": 36.19,
      "grad_norm": 2.6736631393432617,
      "learning_rate": 4.471102623669464e-06,
      "loss": 0.5574,
      "step": 88420
    },
    {
      "epoch": 36.2,
      "grad_norm": 1.9099608659744263,
      "learning_rate": 4.471036044571997e-06,
      "loss": 0.5591,
      "step": 88430
    },
    {
      "epoch": 36.2,
      "grad_norm": 2.6123297214508057,
      "learning_rate": 4.470969455729762e-06,
      "loss": 0.5549,
      "step": 88440
    },
    {
      "epoch": 36.21,
      "grad_norm": 1.9166232347488403,
      "learning_rate": 4.470902857143063e-06,
      "loss": 0.5711,
      "step": 88450
    },
    {
      "epoch": 36.21,
      "grad_norm": 1.9927693605422974,
      "learning_rate": 4.470836248812208e-06,
      "loss": 0.5641,
      "step": 88460
    },
    {
      "epoch": 36.21,
      "grad_norm": 2.554863691329956,
      "learning_rate": 4.4707696307374995e-06,
      "loss": 0.5432,
      "step": 88470
    },
    {
      "epoch": 36.22,
      "grad_norm": 2.56022572517395,
      "learning_rate": 4.470703002919243e-06,
      "loss": 0.555,
      "step": 88480
    },
    {
      "epoch": 36.22,
      "grad_norm": 3.130126953125,
      "learning_rate": 4.470636365357745e-06,
      "loss": 0.5537,
      "step": 88490
    },
    {
      "epoch": 36.23,
      "grad_norm": 1.710602045059204,
      "learning_rate": 4.470569718053309e-06,
      "loss": 0.5818,
      "step": 88500
    },
    {
      "epoch": 36.23,
      "grad_norm": 2.1725003719329834,
      "learning_rate": 4.470503061006242e-06,
      "loss": 0.5746,
      "step": 88510
    },
    {
      "epoch": 36.23,
      "grad_norm": 1.9459336996078491,
      "learning_rate": 4.470436394216849e-06,
      "loss": 0.5581,
      "step": 88520
    },
    {
      "epoch": 36.24,
      "grad_norm": 2.2195851802825928,
      "learning_rate": 4.470369717685434e-06,
      "loss": 0.568,
      "step": 88530
    },
    {
      "epoch": 36.24,
      "grad_norm": 2.65337872505188,
      "learning_rate": 4.470303031412304e-06,
      "loss": 0.5765,
      "step": 88540
    },
    {
      "epoch": 36.25,
      "grad_norm": 2.132981538772583,
      "learning_rate": 4.470236335397764e-06,
      "loss": 0.5664,
      "step": 88550
    },
    {
      "epoch": 36.25,
      "grad_norm": 2.220593214035034,
      "learning_rate": 4.4701696296421186e-06,
      "loss": 0.5776,
      "step": 88560
    },
    {
      "epoch": 36.25,
      "grad_norm": 2.0746331214904785,
      "learning_rate": 4.470102914145675e-06,
      "loss": 0.56,
      "step": 88570
    },
    {
      "epoch": 36.26,
      "grad_norm": 1.9280575513839722,
      "learning_rate": 4.470036188908738e-06,
      "loss": 0.5561,
      "step": 88580
    },
    {
      "epoch": 36.26,
      "grad_norm": 1.995223045349121,
      "learning_rate": 4.469969453931613e-06,
      "loss": 0.5761,
      "step": 88590
    },
    {
      "epoch": 36.27,
      "grad_norm": 2.082857847213745,
      "learning_rate": 4.469902709214606e-06,
      "loss": 0.5254,
      "step": 88600
    },
    {
      "epoch": 36.27,
      "grad_norm": 1.671557903289795,
      "learning_rate": 4.469835954758023e-06,
      "loss": 0.5491,
      "step": 88610
    },
    {
      "epoch": 36.28,
      "grad_norm": 2.463373899459839,
      "learning_rate": 4.469769190562169e-06,
      "loss": 0.5594,
      "step": 88620
    },
    {
      "epoch": 36.28,
      "grad_norm": 2.8811521530151367,
      "learning_rate": 4.469702416627351e-06,
      "loss": 0.5701,
      "step": 88630
    },
    {
      "epoch": 36.28,
      "grad_norm": 1.9648739099502563,
      "learning_rate": 4.4696356329538735e-06,
      "loss": 0.5723,
      "step": 88640
    },
    {
      "epoch": 36.29,
      "grad_norm": 2.496537685394287,
      "learning_rate": 4.469568839542043e-06,
      "loss": 0.5734,
      "step": 88650
    },
    {
      "epoch": 36.29,
      "grad_norm": 2.6528067588806152,
      "learning_rate": 4.469502036392167e-06,
      "loss": 0.5603,
      "step": 88660
    },
    {
      "epoch": 36.3,
      "grad_norm": 2.982306718826294,
      "learning_rate": 4.469435223504548e-06,
      "loss": 0.5729,
      "step": 88670
    },
    {
      "epoch": 36.3,
      "grad_norm": 2.9117541313171387,
      "learning_rate": 4.469368400879496e-06,
      "loss": 0.5682,
      "step": 88680
    },
    {
      "epoch": 36.3,
      "grad_norm": 2.1507961750030518,
      "learning_rate": 4.469301568517315e-06,
      "loss": 0.5686,
      "step": 88690
    },
    {
      "epoch": 36.31,
      "grad_norm": 1.9422916173934937,
      "learning_rate": 4.469234726418311e-06,
      "loss": 0.5682,
      "step": 88700
    },
    {
      "epoch": 36.31,
      "grad_norm": 2.0210418701171875,
      "learning_rate": 4.4691678745827905e-06,
      "loss": 0.5535,
      "step": 88710
    },
    {
      "epoch": 36.32,
      "grad_norm": 2.4127535820007324,
      "learning_rate": 4.46910101301106e-06,
      "loss": 0.563,
      "step": 88720
    },
    {
      "epoch": 36.32,
      "grad_norm": 2.2621731758117676,
      "learning_rate": 4.469034141703426e-06,
      "loss": 0.564,
      "step": 88730
    },
    {
      "epoch": 36.32,
      "grad_norm": 3.08292293548584,
      "learning_rate": 4.4689672606601946e-06,
      "loss": 0.5838,
      "step": 88740
    },
    {
      "epoch": 36.33,
      "grad_norm": 2.627034902572632,
      "learning_rate": 4.468900369881672e-06,
      "loss": 0.5468,
      "step": 88750
    },
    {
      "epoch": 36.33,
      "grad_norm": 2.204749822616577,
      "learning_rate": 4.468833469368164e-06,
      "loss": 0.5547,
      "step": 88760
    },
    {
      "epoch": 36.34,
      "grad_norm": 2.347198009490967,
      "learning_rate": 4.468766559119979e-06,
      "loss": 0.5556,
      "step": 88770
    },
    {
      "epoch": 36.34,
      "grad_norm": 2.2998929023742676,
      "learning_rate": 4.46869963913742e-06,
      "loss": 0.5721,
      "step": 88780
    },
    {
      "epoch": 36.34,
      "grad_norm": 2.5992467403411865,
      "learning_rate": 4.468632709420799e-06,
      "loss": 0.5747,
      "step": 88790
    },
    {
      "epoch": 36.35,
      "grad_norm": 2.3613829612731934,
      "learning_rate": 4.4685657699704165e-06,
      "loss": 0.5707,
      "step": 88800
    },
    {
      "epoch": 36.35,
      "grad_norm": 2.5992772579193115,
      "learning_rate": 4.468498820786584e-06,
      "loss": 0.5577,
      "step": 88810
    },
    {
      "epoch": 36.36,
      "grad_norm": 2.253361940383911,
      "learning_rate": 4.468431861869605e-06,
      "loss": 0.5562,
      "step": 88820
    },
    {
      "epoch": 36.36,
      "grad_norm": 2.1338467597961426,
      "learning_rate": 4.468364893219788e-06,
      "loss": 0.5978,
      "step": 88830
    },
    {
      "epoch": 36.37,
      "grad_norm": 2.0399856567382812,
      "learning_rate": 4.468297914837439e-06,
      "loss": 0.5695,
      "step": 88840
    },
    {
      "epoch": 36.37,
      "grad_norm": 2.4345028400421143,
      "learning_rate": 4.468230926722865e-06,
      "loss": 0.5671,
      "step": 88850
    },
    {
      "epoch": 36.37,
      "grad_norm": 2.1729915142059326,
      "learning_rate": 4.468163928876373e-06,
      "loss": 0.5582,
      "step": 88860
    },
    {
      "epoch": 36.38,
      "grad_norm": 2.043806791305542,
      "learning_rate": 4.46809692129827e-06,
      "loss": 0.5761,
      "step": 88870
    },
    {
      "epoch": 36.38,
      "grad_norm": 12.566461563110352,
      "learning_rate": 4.468029903988863e-06,
      "loss": 0.5603,
      "step": 88880
    },
    {
      "epoch": 36.39,
      "grad_norm": 2.6667439937591553,
      "learning_rate": 4.467962876948457e-06,
      "loss": 0.5782,
      "step": 88890
    },
    {
      "epoch": 36.39,
      "grad_norm": 2.5512795448303223,
      "learning_rate": 4.467895840177362e-06,
      "loss": 0.5645,
      "step": 88900
    },
    {
      "epoch": 36.39,
      "grad_norm": 2.263735294342041,
      "learning_rate": 4.467828793675884e-06,
      "loss": 0.5571,
      "step": 88910
    },
    {
      "epoch": 36.4,
      "grad_norm": 2.1437771320343018,
      "learning_rate": 4.46776173744433e-06,
      "loss": 0.567,
      "step": 88920
    },
    {
      "epoch": 36.4,
      "grad_norm": 2.4637441635131836,
      "learning_rate": 4.467694671483007e-06,
      "loss": 0.5807,
      "step": 88930
    },
    {
      "epoch": 36.41,
      "grad_norm": 1.9301565885543823,
      "learning_rate": 4.467627595792223e-06,
      "loss": 0.5602,
      "step": 88940
    },
    {
      "epoch": 36.41,
      "grad_norm": 1.9769819974899292,
      "learning_rate": 4.467560510372283e-06,
      "loss": 0.5568,
      "step": 88950
    },
    {
      "epoch": 36.41,
      "grad_norm": 2.3835437297821045,
      "learning_rate": 4.467493415223498e-06,
      "loss": 0.5609,
      "step": 88960
    },
    {
      "epoch": 36.42,
      "grad_norm": 2.0615427494049072,
      "learning_rate": 4.467426310346172e-06,
      "loss": 0.5776,
      "step": 88970
    },
    {
      "epoch": 36.42,
      "grad_norm": 2.193817138671875,
      "learning_rate": 4.467359195740614e-06,
      "loss": 0.5718,
      "step": 88980
    },
    {
      "epoch": 36.43,
      "grad_norm": 2.248039484024048,
      "learning_rate": 4.467292071407132e-06,
      "loss": 0.5674,
      "step": 88990
    },
    {
      "epoch": 36.43,
      "grad_norm": 2.071897029876709,
      "learning_rate": 4.467224937346031e-06,
      "loss": 0.5455,
      "step": 89000
    },
    {
      "epoch": 36.43,
      "grad_norm": 1.7884570360183716,
      "learning_rate": 4.467157793557621e-06,
      "loss": 0.5383,
      "step": 89010
    },
    {
      "epoch": 36.44,
      "grad_norm": 2.026808500289917,
      "learning_rate": 4.467090640042209e-06,
      "loss": 0.5713,
      "step": 89020
    },
    {
      "epoch": 36.44,
      "grad_norm": 2.092759370803833,
      "learning_rate": 4.467023476800103e-06,
      "loss": 0.5617,
      "step": 89030
    },
    {
      "epoch": 36.45,
      "grad_norm": 2.2352826595306396,
      "learning_rate": 4.466956303831609e-06,
      "loss": 0.5588,
      "step": 89040
    },
    {
      "epoch": 36.45,
      "grad_norm": 2.6644673347473145,
      "learning_rate": 4.466889121137036e-06,
      "loss": 0.5573,
      "step": 89050
    },
    {
      "epoch": 36.46,
      "grad_norm": 3.0079166889190674,
      "learning_rate": 4.466821928716692e-06,
      "loss": 0.5656,
      "step": 89060
    },
    {
      "epoch": 36.46,
      "grad_norm": 1.9322224855422974,
      "learning_rate": 4.466754726570884e-06,
      "loss": 0.552,
      "step": 89070
    },
    {
      "epoch": 36.46,
      "grad_norm": 1.937069296836853,
      "learning_rate": 4.46668751469992e-06,
      "loss": 0.5524,
      "step": 89080
    },
    {
      "epoch": 36.47,
      "grad_norm": 1.8491543531417847,
      "learning_rate": 4.466620293104107e-06,
      "loss": 0.5401,
      "step": 89090
    },
    {
      "epoch": 36.47,
      "grad_norm": 1.9143052101135254,
      "learning_rate": 4.466553061783755e-06,
      "loss": 0.5818,
      "step": 89100
    },
    {
      "epoch": 36.48,
      "grad_norm": 1.894637107849121,
      "learning_rate": 4.46648582073917e-06,
      "loss": 0.5539,
      "step": 89110
    },
    {
      "epoch": 36.48,
      "grad_norm": 2.71488618850708,
      "learning_rate": 4.4664185699706625e-06,
      "loss": 0.5878,
      "step": 89120
    },
    {
      "epoch": 36.48,
      "grad_norm": 2.49200177192688,
      "learning_rate": 4.4663513094785385e-06,
      "loss": 0.5454,
      "step": 89130
    },
    {
      "epoch": 36.49,
      "grad_norm": 2.262532949447632,
      "learning_rate": 4.466284039263106e-06,
      "loss": 0.5464,
      "step": 89140
    },
    {
      "epoch": 36.49,
      "grad_norm": 1.9218255281448364,
      "learning_rate": 4.4662167593246745e-06,
      "loss": 0.5901,
      "step": 89150
    },
    {
      "epoch": 36.5,
      "grad_norm": 2.591017246246338,
      "learning_rate": 4.4661494696635505e-06,
      "loss": 0.5709,
      "step": 89160
    },
    {
      "epoch": 36.5,
      "grad_norm": 3.4444994926452637,
      "learning_rate": 4.466082170280045e-06,
      "loss": 0.5503,
      "step": 89170
    },
    {
      "epoch": 36.5,
      "grad_norm": 2.0403802394866943,
      "learning_rate": 4.466014861174463e-06,
      "loss": 0.559,
      "step": 89180
    },
    {
      "epoch": 36.51,
      "grad_norm": 1.8206381797790527,
      "learning_rate": 4.465947542347115e-06,
      "loss": 0.5497,
      "step": 89190
    },
    {
      "epoch": 36.51,
      "grad_norm": 1.9797663688659668,
      "learning_rate": 4.465880213798308e-06,
      "loss": 0.5657,
      "step": 89200
    },
    {
      "epoch": 36.52,
      "grad_norm": 2.6870224475860596,
      "learning_rate": 4.465812875528352e-06,
      "loss": 0.5486,
      "step": 89210
    },
    {
      "epoch": 36.52,
      "grad_norm": 2.0354561805725098,
      "learning_rate": 4.465745527537554e-06,
      "loss": 0.5521,
      "step": 89220
    },
    {
      "epoch": 36.52,
      "grad_norm": 2.55759859085083,
      "learning_rate": 4.465678169826223e-06,
      "loss": 0.5814,
      "step": 89230
    },
    {
      "epoch": 36.53,
      "grad_norm": 2.2930736541748047,
      "learning_rate": 4.465610802394668e-06,
      "loss": 0.5941,
      "step": 89240
    },
    {
      "epoch": 36.53,
      "grad_norm": 3.0021443367004395,
      "learning_rate": 4.465543425243198e-06,
      "loss": 0.5415,
      "step": 89250
    },
    {
      "epoch": 36.54,
      "grad_norm": 1.688470482826233,
      "learning_rate": 4.465476038372121e-06,
      "loss": 0.5614,
      "step": 89260
    },
    {
      "epoch": 36.54,
      "grad_norm": 4.17825174331665,
      "learning_rate": 4.4654086417817444e-06,
      "loss": 0.5651,
      "step": 89270
    },
    {
      "epoch": 36.55,
      "grad_norm": 1.5601180791854858,
      "learning_rate": 4.4653412354723784e-06,
      "loss": 0.5551,
      "step": 89280
    },
    {
      "epoch": 36.55,
      "grad_norm": 2.409230947494507,
      "learning_rate": 4.465273819444331e-06,
      "loss": 0.5814,
      "step": 89290
    },
    {
      "epoch": 36.55,
      "grad_norm": 2.5048117637634277,
      "learning_rate": 4.465206393697913e-06,
      "loss": 0.5806,
      "step": 89300
    },
    {
      "epoch": 36.56,
      "grad_norm": 2.7447469234466553,
      "learning_rate": 4.46513895823343e-06,
      "loss": 0.553,
      "step": 89310
    },
    {
      "epoch": 36.56,
      "grad_norm": 2.4295310974121094,
      "learning_rate": 4.465071513051195e-06,
      "loss": 0.5569,
      "step": 89320
    },
    {
      "epoch": 36.57,
      "grad_norm": 1.943763017654419,
      "learning_rate": 4.465004058151513e-06,
      "loss": 0.5599,
      "step": 89330
    },
    {
      "epoch": 36.57,
      "grad_norm": 2.675427198410034,
      "learning_rate": 4.464936593534695e-06,
      "loss": 0.542,
      "step": 89340
    },
    {
      "epoch": 36.57,
      "grad_norm": 2.34822416305542,
      "learning_rate": 4.46486911920105e-06,
      "loss": 0.5484,
      "step": 89350
    },
    {
      "epoch": 36.58,
      "grad_norm": 2.759913682937622,
      "learning_rate": 4.464801635150887e-06,
      "loss": 0.5849,
      "step": 89360
    },
    {
      "epoch": 36.58,
      "grad_norm": 2.1496341228485107,
      "learning_rate": 4.464734141384514e-06,
      "loss": 0.5491,
      "step": 89370
    },
    {
      "epoch": 36.59,
      "grad_norm": 2.41324520111084,
      "learning_rate": 4.464666637902242e-06,
      "loss": 0.5672,
      "step": 89380
    },
    {
      "epoch": 36.59,
      "grad_norm": 2.3938188552856445,
      "learning_rate": 4.464599124704379e-06,
      "loss": 0.5771,
      "step": 89390
    },
    {
      "epoch": 36.59,
      "grad_norm": 2.106269359588623,
      "learning_rate": 4.464531601791235e-06,
      "loss": 0.562,
      "step": 89400
    },
    {
      "epoch": 36.6,
      "grad_norm": 2.6523075103759766,
      "learning_rate": 4.464464069163118e-06,
      "loss": 0.5849,
      "step": 89410
    },
    {
      "epoch": 36.6,
      "grad_norm": 1.543311357498169,
      "learning_rate": 4.464396526820339e-06,
      "loss": 0.5579,
      "step": 89420
    },
    {
      "epoch": 36.61,
      "grad_norm": 2.4445741176605225,
      "learning_rate": 4.464328974763206e-06,
      "loss": 0.5643,
      "step": 89430
    },
    {
      "epoch": 36.61,
      "grad_norm": 2.6060562133789062,
      "learning_rate": 4.4642614129920306e-06,
      "loss": 0.5612,
      "step": 89440
    },
    {
      "epoch": 36.61,
      "grad_norm": 2.146773099899292,
      "learning_rate": 4.46419384150712e-06,
      "loss": 0.5617,
      "step": 89450
    },
    {
      "epoch": 36.62,
      "grad_norm": 2.9087307453155518,
      "learning_rate": 4.464126260308784e-06,
      "loss": 0.5189,
      "step": 89460
    },
    {
      "epoch": 36.62,
      "grad_norm": 3.6888720989227295,
      "learning_rate": 4.464058669397334e-06,
      "loss": 0.567,
      "step": 89470
    },
    {
      "epoch": 36.63,
      "grad_norm": 2.0198049545288086,
      "learning_rate": 4.463991068773077e-06,
      "loss": 0.5481,
      "step": 89480
    },
    {
      "epoch": 36.63,
      "grad_norm": 1.6286029815673828,
      "learning_rate": 4.463923458436325e-06,
      "loss": 0.559,
      "step": 89490
    },
    {
      "epoch": 36.64,
      "grad_norm": 2.1304609775543213,
      "learning_rate": 4.463855838387386e-06,
      "loss": 0.5565,
      "step": 89500
    },
    {
      "epoch": 36.64,
      "grad_norm": 2.426461935043335,
      "learning_rate": 4.463788208626571e-06,
      "loss": 0.5606,
      "step": 89510
    },
    {
      "epoch": 36.64,
      "grad_norm": 2.5624217987060547,
      "learning_rate": 4.463720569154189e-06,
      "loss": 0.5709,
      "step": 89520
    },
    {
      "epoch": 36.65,
      "grad_norm": 2.494029998779297,
      "learning_rate": 4.463652919970551e-06,
      "loss": 0.542,
      "step": 89530
    },
    {
      "epoch": 36.65,
      "grad_norm": 1.9811656475067139,
      "learning_rate": 4.463585261075965e-06,
      "loss": 0.5421,
      "step": 89540
    },
    {
      "epoch": 36.66,
      "grad_norm": 1.9603689908981323,
      "learning_rate": 4.463517592470743e-06,
      "loss": 0.5575,
      "step": 89550
    },
    {
      "epoch": 36.66,
      "grad_norm": 2.342514991760254,
      "learning_rate": 4.4634499141551924e-06,
      "loss": 0.5729,
      "step": 89560
    },
    {
      "epoch": 36.66,
      "grad_norm": 1.935390830039978,
      "learning_rate": 4.463382226129626e-06,
      "loss": 0.5555,
      "step": 89570
    },
    {
      "epoch": 36.67,
      "grad_norm": 2.3697540760040283,
      "learning_rate": 4.463314528394352e-06,
      "loss": 0.5572,
      "step": 89580
    },
    {
      "epoch": 36.67,
      "grad_norm": 3.163139581680298,
      "learning_rate": 4.463246820949681e-06,
      "loss": 0.5588,
      "step": 89590
    },
    {
      "epoch": 36.68,
      "grad_norm": 2.512399673461914,
      "learning_rate": 4.463179103795924e-06,
      "loss": 0.5704,
      "step": 89600
    },
    {
      "epoch": 36.68,
      "grad_norm": 1.944942593574524,
      "learning_rate": 4.46311137693339e-06,
      "loss": 0.5727,
      "step": 89610
    },
    {
      "epoch": 36.68,
      "grad_norm": 2.37804913520813,
      "learning_rate": 4.46304364036239e-06,
      "loss": 0.5353,
      "step": 89620
    },
    {
      "epoch": 36.69,
      "grad_norm": 2.67744779586792,
      "learning_rate": 4.462975894083234e-06,
      "loss": 0.5462,
      "step": 89630
    },
    {
      "epoch": 36.69,
      "grad_norm": 2.5485692024230957,
      "learning_rate": 4.462908138096233e-06,
      "loss": 0.5731,
      "step": 89640
    },
    {
      "epoch": 36.7,
      "grad_norm": 2.264639377593994,
      "learning_rate": 4.462840372401695e-06,
      "loss": 0.57,
      "step": 89650
    },
    {
      "epoch": 36.7,
      "grad_norm": 2.2313191890716553,
      "learning_rate": 4.462772596999933e-06,
      "loss": 0.5581,
      "step": 89660
    },
    {
      "epoch": 36.7,
      "grad_norm": 2.1408677101135254,
      "learning_rate": 4.462704811891258e-06,
      "loss": 0.56,
      "step": 89670
    },
    {
      "epoch": 36.71,
      "grad_norm": 2.5645406246185303,
      "learning_rate": 4.462637017075978e-06,
      "loss": 0.5399,
      "step": 89680
    },
    {
      "epoch": 36.71,
      "grad_norm": 2.220127582550049,
      "learning_rate": 4.462569212554405e-06,
      "loss": 0.5478,
      "step": 89690
    },
    {
      "epoch": 36.72,
      "grad_norm": 2.1818580627441406,
      "learning_rate": 4.462501398326849e-06,
      "loss": 0.5614,
      "step": 89700
    },
    {
      "epoch": 36.72,
      "grad_norm": 2.5990936756134033,
      "learning_rate": 4.462433574393621e-06,
      "loss": 0.5534,
      "step": 89710
    },
    {
      "epoch": 36.73,
      "grad_norm": 2.7728161811828613,
      "learning_rate": 4.462365740755032e-06,
      "loss": 0.5478,
      "step": 89720
    },
    {
      "epoch": 36.73,
      "grad_norm": 1.636177659034729,
      "learning_rate": 4.462297897411392e-06,
      "loss": 0.565,
      "step": 89730
    },
    {
      "epoch": 36.73,
      "grad_norm": 3.046090602874756,
      "learning_rate": 4.462230044363013e-06,
      "loss": 0.5625,
      "step": 89740
    },
    {
      "epoch": 36.74,
      "grad_norm": 2.1703269481658936,
      "learning_rate": 4.462162181610204e-06,
      "loss": 0.5833,
      "step": 89750
    },
    {
      "epoch": 36.74,
      "grad_norm": 2.0181260108947754,
      "learning_rate": 4.462094309153278e-06,
      "loss": 0.5547,
      "step": 89760
    },
    {
      "epoch": 36.75,
      "grad_norm": 2.4185118675231934,
      "learning_rate": 4.462026426992544e-06,
      "loss": 0.5565,
      "step": 89770
    },
    {
      "epoch": 36.75,
      "grad_norm": 2.027101516723633,
      "learning_rate": 4.4619585351283135e-06,
      "loss": 0.5537,
      "step": 89780
    },
    {
      "epoch": 36.75,
      "grad_norm": 2.4287919998168945,
      "learning_rate": 4.461890633560898e-06,
      "loss": 0.5833,
      "step": 89790
    },
    {
      "epoch": 36.76,
      "grad_norm": 2.0122785568237305,
      "learning_rate": 4.461822722290609e-06,
      "loss": 0.5674,
      "step": 89800
    },
    {
      "epoch": 36.76,
      "grad_norm": 2.349625825881958,
      "learning_rate": 4.461754801317756e-06,
      "loss": 0.5773,
      "step": 89810
    },
    {
      "epoch": 36.77,
      "grad_norm": 1.9449796676635742,
      "learning_rate": 4.461686870642651e-06,
      "loss": 0.5525,
      "step": 89820
    },
    {
      "epoch": 36.77,
      "grad_norm": 2.2889935970306396,
      "learning_rate": 4.4616189302656054e-06,
      "loss": 0.5723,
      "step": 89830
    },
    {
      "epoch": 36.77,
      "grad_norm": 2.110405683517456,
      "learning_rate": 4.46155098018693e-06,
      "loss": 0.5628,
      "step": 89840
    },
    {
      "epoch": 36.78,
      "grad_norm": 2.34987735748291,
      "learning_rate": 4.461483020406938e-06,
      "loss": 0.5533,
      "step": 89850
    },
    {
      "epoch": 36.78,
      "grad_norm": 2.1326656341552734,
      "learning_rate": 4.461415050925936e-06,
      "loss": 0.5742,
      "step": 89860
    },
    {
      "epoch": 36.79,
      "grad_norm": 2.157888889312744,
      "learning_rate": 4.4613470717442406e-06,
      "loss": 0.5758,
      "step": 89870
    },
    {
      "epoch": 36.79,
      "grad_norm": 2.227729320526123,
      "learning_rate": 4.461279082862161e-06,
      "loss": 0.5637,
      "step": 89880
    },
    {
      "epoch": 36.79,
      "grad_norm": 2.033935785293579,
      "learning_rate": 4.461211084280008e-06,
      "loss": 0.5575,
      "step": 89890
    },
    {
      "epoch": 36.8,
      "grad_norm": 2.450827121734619,
      "learning_rate": 4.461143075998093e-06,
      "loss": 0.5573,
      "step": 89900
    },
    {
      "epoch": 36.8,
      "grad_norm": 2.140942335128784,
      "learning_rate": 4.461075058016729e-06,
      "loss": 0.5711,
      "step": 89910
    },
    {
      "epoch": 36.81,
      "grad_norm": 2.009971857070923,
      "learning_rate": 4.461007030336227e-06,
      "loss": 0.575,
      "step": 89920
    },
    {
      "epoch": 36.81,
      "grad_norm": 2.1378772258758545,
      "learning_rate": 4.460938992956898e-06,
      "loss": 0.5562,
      "step": 89930
    },
    {
      "epoch": 36.82,
      "grad_norm": 2.102527141571045,
      "learning_rate": 4.460870945879054e-06,
      "loss": 0.5688,
      "step": 89940
    },
    {
      "epoch": 36.82,
      "grad_norm": 2.261326313018799,
      "learning_rate": 4.460802889103007e-06,
      "loss": 0.5729,
      "step": 89950
    },
    {
      "epoch": 36.82,
      "grad_norm": 2.0952165126800537,
      "learning_rate": 4.460734822629068e-06,
      "loss": 0.5415,
      "step": 89960
    },
    {
      "epoch": 36.83,
      "grad_norm": 2.130779504776001,
      "learning_rate": 4.460666746457551e-06,
      "loss": 0.5718,
      "step": 89970
    },
    {
      "epoch": 36.83,
      "grad_norm": 2.213078022003174,
      "learning_rate": 4.460598660588765e-06,
      "loss": 0.588,
      "step": 89980
    },
    {
      "epoch": 36.84,
      "grad_norm": 1.826370358467102,
      "learning_rate": 4.460530565023024e-06,
      "loss": 0.5622,
      "step": 89990
    },
    {
      "epoch": 36.84,
      "grad_norm": 2.6210551261901855,
      "learning_rate": 4.460462459760638e-06,
      "loss": 0.5601,
      "step": 90000
    },
    {
      "epoch": 36.84,
      "grad_norm": 2.9786484241485596,
      "learning_rate": 4.46039434480192e-06,
      "loss": 0.5664,
      "step": 90010
    },
    {
      "epoch": 36.85,
      "grad_norm": 2.0362114906311035,
      "learning_rate": 4.460326220147183e-06,
      "loss": 0.5619,
      "step": 90020
    },
    {
      "epoch": 36.85,
      "grad_norm": 1.9911549091339111,
      "learning_rate": 4.460258085796738e-06,
      "loss": 0.5741,
      "step": 90030
    },
    {
      "epoch": 36.86,
      "grad_norm": 3.3821277618408203,
      "learning_rate": 4.460189941750896e-06,
      "loss": 0.5719,
      "step": 90040
    },
    {
      "epoch": 36.86,
      "grad_norm": 1.913404941558838,
      "learning_rate": 4.460121788009972e-06,
      "loss": 0.5492,
      "step": 90050
    },
    {
      "epoch": 36.86,
      "grad_norm": 2.776268243789673,
      "learning_rate": 4.460053624574276e-06,
      "loss": 0.5709,
      "step": 90060
    },
    {
      "epoch": 36.87,
      "grad_norm": 2.502254009246826,
      "learning_rate": 4.459985451444121e-06,
      "loss": 0.5568,
      "step": 90070
    },
    {
      "epoch": 36.87,
      "grad_norm": 2.0593979358673096,
      "learning_rate": 4.459917268619819e-06,
      "loss": 0.5636,
      "step": 90080
    },
    {
      "epoch": 36.88,
      "grad_norm": 2.0704450607299805,
      "learning_rate": 4.459849076101682e-06,
      "loss": 0.5666,
      "step": 90090
    },
    {
      "epoch": 36.88,
      "grad_norm": 3.0708272457122803,
      "learning_rate": 4.459780873890023e-06,
      "loss": 0.596,
      "step": 90100
    },
    {
      "epoch": 36.88,
      "grad_norm": 2.074918746948242,
      "learning_rate": 4.459712661985154e-06,
      "loss": 0.5419,
      "step": 90110
    },
    {
      "epoch": 36.89,
      "grad_norm": 1.8009194135665894,
      "learning_rate": 4.459644440387389e-06,
      "loss": 0.5759,
      "step": 90120
    },
    {
      "epoch": 36.89,
      "grad_norm": 2.2169649600982666,
      "learning_rate": 4.459576209097039e-06,
      "loss": 0.5776,
      "step": 90130
    },
    {
      "epoch": 36.9,
      "grad_norm": 2.184906244277954,
      "learning_rate": 4.459507968114416e-06,
      "loss": 0.542,
      "step": 90140
    },
    {
      "epoch": 36.9,
      "grad_norm": 2.353790760040283,
      "learning_rate": 4.459439717439834e-06,
      "loss": 0.5424,
      "step": 90150
    },
    {
      "epoch": 36.91,
      "grad_norm": 1.9311912059783936,
      "learning_rate": 4.459371457073604e-06,
      "loss": 0.5506,
      "step": 90160
    },
    {
      "epoch": 36.91,
      "grad_norm": 2.026370048522949,
      "learning_rate": 4.459303187016041e-06,
      "loss": 0.5702,
      "step": 90170
    },
    {
      "epoch": 36.91,
      "grad_norm": 2.9925332069396973,
      "learning_rate": 4.459234907267456e-06,
      "loss": 0.6094,
      "step": 90180
    },
    {
      "epoch": 36.92,
      "grad_norm": 3.0662693977355957,
      "learning_rate": 4.459166617828162e-06,
      "loss": 0.5652,
      "step": 90190
    },
    {
      "epoch": 36.92,
      "grad_norm": 2.071260452270508,
      "learning_rate": 4.459098318698473e-06,
      "loss": 0.5705,
      "step": 90200
    },
    {
      "epoch": 36.93,
      "grad_norm": 3.053273916244507,
      "learning_rate": 4.4590300098787e-06,
      "loss": 0.5413,
      "step": 90210
    },
    {
      "epoch": 36.93,
      "grad_norm": 2.232642412185669,
      "learning_rate": 4.4589616913691565e-06,
      "loss": 0.5399,
      "step": 90220
    },
    {
      "epoch": 36.93,
      "grad_norm": 3.3023757934570312,
      "learning_rate": 4.458893363170157e-06,
      "loss": 0.5791,
      "step": 90230
    },
    {
      "epoch": 36.94,
      "grad_norm": 2.6369364261627197,
      "learning_rate": 4.458825025282013e-06,
      "loss": 0.5686,
      "step": 90240
    },
    {
      "epoch": 36.94,
      "grad_norm": 2.386253595352173,
      "learning_rate": 4.458756677705038e-06,
      "loss": 0.5698,
      "step": 90250
    },
    {
      "epoch": 36.95,
      "grad_norm": 2.2434515953063965,
      "learning_rate": 4.458688320439544e-06,
      "loss": 0.5503,
      "step": 90260
    },
    {
      "epoch": 36.95,
      "grad_norm": 1.9596196413040161,
      "learning_rate": 4.458619953485846e-06,
      "loss": 0.5594,
      "step": 90270
    },
    {
      "epoch": 36.95,
      "grad_norm": 2.2673847675323486,
      "learning_rate": 4.458551576844256e-06,
      "loss": 0.5769,
      "step": 90280
    },
    {
      "epoch": 36.96,
      "grad_norm": 1.8476232290267944,
      "learning_rate": 4.458483190515088e-06,
      "loss": 0.5558,
      "step": 90290
    },
    {
      "epoch": 36.96,
      "grad_norm": 1.6711151599884033,
      "learning_rate": 4.458414794498654e-06,
      "loss": 0.5779,
      "step": 90300
    },
    {
      "epoch": 36.97,
      "grad_norm": 2.298553705215454,
      "learning_rate": 4.4583463887952685e-06,
      "loss": 0.5521,
      "step": 90310
    },
    {
      "epoch": 36.97,
      "grad_norm": 2.913994312286377,
      "learning_rate": 4.458277973405244e-06,
      "loss": 0.5702,
      "step": 90320
    },
    {
      "epoch": 36.98,
      "grad_norm": 2.76578426361084,
      "learning_rate": 4.4582095483288955e-06,
      "loss": 0.5587,
      "step": 90330
    },
    {
      "epoch": 36.98,
      "grad_norm": 2.1027820110321045,
      "learning_rate": 4.458141113566534e-06,
      "loss": 0.5554,
      "step": 90340
    },
    {
      "epoch": 36.98,
      "grad_norm": 2.102033853530884,
      "learning_rate": 4.458072669118474e-06,
      "loss": 0.5558,
      "step": 90350
    },
    {
      "epoch": 36.99,
      "grad_norm": 2.151698589324951,
      "learning_rate": 4.4580042149850305e-06,
      "loss": 0.5484,
      "step": 90360
    },
    {
      "epoch": 36.99,
      "grad_norm": 3.1027121543884277,
      "learning_rate": 4.4579357511665156e-06,
      "loss": 0.5465,
      "step": 90370
    },
    {
      "epoch": 37.0,
      "grad_norm": 2.1476473808288574,
      "learning_rate": 4.457867277663243e-06,
      "loss": 0.5907,
      "step": 90380
    },
    {
      "epoch": 37.0,
      "grad_norm": 2.9468626976013184,
      "learning_rate": 4.457798794475526e-06,
      "loss": 0.5442,
      "step": 90390
    },
    {
      "epoch": 37.0,
      "eval_loss": 0.5648902058601379,
      "eval_runtime": 51.9539,
      "eval_samples_per_second": 66.386,
      "eval_steps_per_second": 8.315,
      "step": 90391
    },
    {
      "epoch": 37.0,
      "grad_norm": 1.7409180402755737,
      "learning_rate": 4.457730301603679e-06,
      "loss": 0.5331,
      "step": 90400
    },
    {
      "epoch": 37.01,
      "grad_norm": 2.3098230361938477,
      "learning_rate": 4.457661799048017e-06,
      "loss": 0.5686,
      "step": 90410
    },
    {
      "epoch": 37.01,
      "grad_norm": 3.179309844970703,
      "learning_rate": 4.457593286808851e-06,
      "loss": 0.5464,
      "step": 90420
    },
    {
      "epoch": 37.02,
      "grad_norm": 3.1861093044281006,
      "learning_rate": 4.457524764886497e-06,
      "loss": 0.5929,
      "step": 90430
    },
    {
      "epoch": 37.02,
      "grad_norm": 2.0161218643188477,
      "learning_rate": 4.4574562332812675e-06,
      "loss": 0.5591,
      "step": 90440
    },
    {
      "epoch": 37.02,
      "grad_norm": 2.2761738300323486,
      "learning_rate": 4.4573876919934775e-06,
      "loss": 0.5803,
      "step": 90450
    },
    {
      "epoch": 37.03,
      "grad_norm": 2.205005407333374,
      "learning_rate": 4.45731914102344e-06,
      "loss": 0.5545,
      "step": 90460
    },
    {
      "epoch": 37.03,
      "grad_norm": 2.667076826095581,
      "learning_rate": 4.457250580371471e-06,
      "loss": 0.5816,
      "step": 90470
    },
    {
      "epoch": 37.04,
      "grad_norm": 2.5660619735717773,
      "learning_rate": 4.457182010037882e-06,
      "loss": 0.5725,
      "step": 90480
    },
    {
      "epoch": 37.04,
      "grad_norm": 1.640669822692871,
      "learning_rate": 4.457113430022988e-06,
      "loss": 0.5665,
      "step": 90490
    },
    {
      "epoch": 37.04,
      "grad_norm": 2.6099424362182617,
      "learning_rate": 4.4570448403271045e-06,
      "loss": 0.5494,
      "step": 90500
    },
    {
      "epoch": 37.05,
      "grad_norm": 1.9594553709030151,
      "learning_rate": 4.456976240950544e-06,
      "loss": 0.5643,
      "step": 90510
    },
    {
      "epoch": 37.05,
      "grad_norm": 2.2600491046905518,
      "learning_rate": 4.456907631893621e-06,
      "loss": 0.5566,
      "step": 90520
    },
    {
      "epoch": 37.06,
      "grad_norm": 3.159895896911621,
      "learning_rate": 4.456839013156651e-06,
      "loss": 0.5558,
      "step": 90530
    },
    {
      "epoch": 37.06,
      "grad_norm": 2.4475820064544678,
      "learning_rate": 4.456770384739947e-06,
      "loss": 0.5719,
      "step": 90540
    },
    {
      "epoch": 37.07,
      "grad_norm": 2.1023411750793457,
      "learning_rate": 4.456701746643824e-06,
      "loss": 0.5709,
      "step": 90550
    },
    {
      "epoch": 37.07,
      "grad_norm": 2.2326807975769043,
      "learning_rate": 4.456633098868597e-06,
      "loss": 0.5617,
      "step": 90560
    },
    {
      "epoch": 37.07,
      "grad_norm": 1.846453070640564,
      "learning_rate": 4.4565644414145785e-06,
      "loss": 0.563,
      "step": 90570
    },
    {
      "epoch": 37.08,
      "grad_norm": 1.9042118787765503,
      "learning_rate": 4.4564957742820846e-06,
      "loss": 0.5443,
      "step": 90580
    },
    {
      "epoch": 37.08,
      "grad_norm": 2.023357391357422,
      "learning_rate": 4.4564270974714295e-06,
      "loss": 0.5482,
      "step": 90590
    },
    {
      "epoch": 37.09,
      "grad_norm": 1.989528775215149,
      "learning_rate": 4.456358410982928e-06,
      "loss": 0.5769,
      "step": 90600
    },
    {
      "epoch": 37.09,
      "grad_norm": 2.3197553157806396,
      "learning_rate": 4.456289714816894e-06,
      "loss": 0.5819,
      "step": 90610
    },
    {
      "epoch": 37.09,
      "grad_norm": 2.7347915172576904,
      "learning_rate": 4.456221008973644e-06,
      "loss": 0.5826,
      "step": 90620
    },
    {
      "epoch": 37.1,
      "grad_norm": 2.2557027339935303,
      "learning_rate": 4.456152293453491e-06,
      "loss": 0.5624,
      "step": 90630
    },
    {
      "epoch": 37.1,
      "grad_norm": 2.4489548206329346,
      "learning_rate": 4.456083568256749e-06,
      "loss": 0.5727,
      "step": 90640
    },
    {
      "epoch": 37.11,
      "grad_norm": 2.530925989151001,
      "learning_rate": 4.456014833383736e-06,
      "loss": 0.561,
      "step": 90650
    },
    {
      "epoch": 37.11,
      "grad_norm": 1.907025694847107,
      "learning_rate": 4.455946088834763e-06,
      "loss": 0.5418,
      "step": 90660
    },
    {
      "epoch": 37.11,
      "grad_norm": 1.9415768384933472,
      "learning_rate": 4.4558773346101485e-06,
      "loss": 0.5618,
      "step": 90670
    },
    {
      "epoch": 37.12,
      "grad_norm": 1.9417822360992432,
      "learning_rate": 4.455808570710205e-06,
      "loss": 0.5503,
      "step": 90680
    },
    {
      "epoch": 37.12,
      "grad_norm": 1.7699646949768066,
      "learning_rate": 4.455739797135248e-06,
      "loss": 0.564,
      "step": 90690
    },
    {
      "epoch": 37.13,
      "grad_norm": 2.092294692993164,
      "learning_rate": 4.4556710138855934e-06,
      "loss": 0.5478,
      "step": 90700
    },
    {
      "epoch": 37.13,
      "grad_norm": 2.2084319591522217,
      "learning_rate": 4.455602220961555e-06,
      "loss": 0.5658,
      "step": 90710
    },
    {
      "epoch": 37.13,
      "grad_norm": 2.00958514213562,
      "learning_rate": 4.45553341836345e-06,
      "loss": 0.575,
      "step": 90720
    },
    {
      "epoch": 37.14,
      "grad_norm": 1.6599243879318237,
      "learning_rate": 4.455464606091591e-06,
      "loss": 0.5529,
      "step": 90730
    },
    {
      "epoch": 37.14,
      "grad_norm": 2.5735976696014404,
      "learning_rate": 4.455395784146294e-06,
      "loss": 0.5736,
      "step": 90740
    },
    {
      "epoch": 37.15,
      "grad_norm": 2.1188673973083496,
      "learning_rate": 4.455326952527876e-06,
      "loss": 0.5749,
      "step": 90750
    },
    {
      "epoch": 37.15,
      "grad_norm": 2.098083734512329,
      "learning_rate": 4.4552581112366505e-06,
      "loss": 0.571,
      "step": 90760
    },
    {
      "epoch": 37.16,
      "grad_norm": 2.657477855682373,
      "learning_rate": 4.455189260272933e-06,
      "loss": 0.5757,
      "step": 90770
    },
    {
      "epoch": 37.16,
      "grad_norm": 2.119770050048828,
      "learning_rate": 4.45512039963704e-06,
      "loss": 0.5531,
      "step": 90780
    },
    {
      "epoch": 37.16,
      "grad_norm": 3.212977409362793,
      "learning_rate": 4.455051529329285e-06,
      "loss": 0.5571,
      "step": 90790
    },
    {
      "epoch": 37.17,
      "grad_norm": 1.989200234413147,
      "learning_rate": 4.454982649349985e-06,
      "loss": 0.5532,
      "step": 90800
    },
    {
      "epoch": 37.17,
      "grad_norm": 1.8235126733779907,
      "learning_rate": 4.454913759699456e-06,
      "loss": 0.5775,
      "step": 90810
    },
    {
      "epoch": 37.18,
      "grad_norm": 3.282895565032959,
      "learning_rate": 4.454844860378013e-06,
      "loss": 0.5634,
      "step": 90820
    },
    {
      "epoch": 37.18,
      "grad_norm": 2.1310527324676514,
      "learning_rate": 4.45477595138597e-06,
      "loss": 0.5753,
      "step": 90830
    },
    {
      "epoch": 37.18,
      "grad_norm": 2.12117600440979,
      "learning_rate": 4.454707032723645e-06,
      "loss": 0.5471,
      "step": 90840
    },
    {
      "epoch": 37.19,
      "grad_norm": 3.0475335121154785,
      "learning_rate": 4.454638104391353e-06,
      "loss": 0.5591,
      "step": 90850
    },
    {
      "epoch": 37.19,
      "grad_norm": 2.1875853538513184,
      "learning_rate": 4.454569166389409e-06,
      "loss": 0.5545,
      "step": 90860
    },
    {
      "epoch": 37.2,
      "grad_norm": 2.6883673667907715,
      "learning_rate": 4.454500218718129e-06,
      "loss": 0.5458,
      "step": 90870
    },
    {
      "epoch": 37.2,
      "grad_norm": 2.0118930339813232,
      "learning_rate": 4.45443126137783e-06,
      "loss": 0.5522,
      "step": 90880
    },
    {
      "epoch": 37.2,
      "grad_norm": 2.159234046936035,
      "learning_rate": 4.454362294368826e-06,
      "loss": 0.5794,
      "step": 90890
    },
    {
      "epoch": 37.21,
      "grad_norm": 1.758739948272705,
      "learning_rate": 4.454293317691435e-06,
      "loss": 0.5736,
      "step": 90900
    },
    {
      "epoch": 37.21,
      "grad_norm": 2.6872589588165283,
      "learning_rate": 4.454224331345971e-06,
      "loss": 0.5695,
      "step": 90910
    },
    {
      "epoch": 37.22,
      "grad_norm": 2.092066526412964,
      "learning_rate": 4.454155335332751e-06,
      "loss": 0.5638,
      "step": 90920
    },
    {
      "epoch": 37.22,
      "grad_norm": 2.066354513168335,
      "learning_rate": 4.454086329652092e-06,
      "loss": 0.5608,
      "step": 90930
    },
    {
      "epoch": 37.22,
      "grad_norm": 1.7824848890304565,
      "learning_rate": 4.4540173143043076e-06,
      "loss": 0.5583,
      "step": 90940
    },
    {
      "epoch": 37.23,
      "grad_norm": 2.4378366470336914,
      "learning_rate": 4.453948289289717e-06,
      "loss": 0.56,
      "step": 90950
    },
    {
      "epoch": 37.23,
      "grad_norm": 1.9086850881576538,
      "learning_rate": 4.453879254608634e-06,
      "loss": 0.5532,
      "step": 90960
    },
    {
      "epoch": 37.24,
      "grad_norm": 3.093496084213257,
      "learning_rate": 4.453810210261376e-06,
      "loss": 0.5768,
      "step": 90970
    },
    {
      "epoch": 37.24,
      "grad_norm": 2.276506185531616,
      "learning_rate": 4.453741156248258e-06,
      "loss": 0.5378,
      "step": 90980
    },
    {
      "epoch": 37.25,
      "grad_norm": 2.1495561599731445,
      "learning_rate": 4.4536720925695985e-06,
      "loss": 0.5482,
      "step": 90990
    },
    {
      "epoch": 37.25,
      "grad_norm": 2.2645461559295654,
      "learning_rate": 4.4536030192257115e-06,
      "loss": 0.5805,
      "step": 91000
    },
    {
      "epoch": 37.25,
      "grad_norm": 1.9161796569824219,
      "learning_rate": 4.453533936216916e-06,
      "loss": 0.5622,
      "step": 91010
    },
    {
      "epoch": 37.26,
      "grad_norm": 3.960627555847168,
      "learning_rate": 4.453464843543526e-06,
      "loss": 0.5647,
      "step": 91020
    },
    {
      "epoch": 37.26,
      "grad_norm": 2.383471965789795,
      "learning_rate": 4.453395741205859e-06,
      "loss": 0.5825,
      "step": 91030
    },
    {
      "epoch": 37.27,
      "grad_norm": 2.122941493988037,
      "learning_rate": 4.4533266292042325e-06,
      "loss": 0.5714,
      "step": 91040
    },
    {
      "epoch": 37.27,
      "grad_norm": 2.6942338943481445,
      "learning_rate": 4.4532575075389615e-06,
      "loss": 0.5839,
      "step": 91050
    },
    {
      "epoch": 37.27,
      "grad_norm": 2.7669527530670166,
      "learning_rate": 4.453188376210363e-06,
      "loss": 0.5723,
      "step": 91060
    },
    {
      "epoch": 37.28,
      "grad_norm": 1.9943013191223145,
      "learning_rate": 4.453119235218754e-06,
      "loss": 0.5582,
      "step": 91070
    },
    {
      "epoch": 37.28,
      "grad_norm": 2.0827033519744873,
      "learning_rate": 4.453050084564452e-06,
      "loss": 0.5622,
      "step": 91080
    },
    {
      "epoch": 37.29,
      "grad_norm": 2.5039303302764893,
      "learning_rate": 4.452980924247773e-06,
      "loss": 0.5557,
      "step": 91090
    },
    {
      "epoch": 37.29,
      "grad_norm": 2.205697536468506,
      "learning_rate": 4.452911754269033e-06,
      "loss": 0.5543,
      "step": 91100
    },
    {
      "epoch": 37.29,
      "grad_norm": 1.838098168373108,
      "learning_rate": 4.45284257462855e-06,
      "loss": 0.5572,
      "step": 91110
    },
    {
      "epoch": 37.3,
      "grad_norm": 2.4301490783691406,
      "learning_rate": 4.45277338532664e-06,
      "loss": 0.5908,
      "step": 91120
    },
    {
      "epoch": 37.3,
      "grad_norm": 3.319154739379883,
      "learning_rate": 4.452704186363621e-06,
      "loss": 0.5634,
      "step": 91130
    },
    {
      "epoch": 37.31,
      "grad_norm": 2.1994993686676025,
      "learning_rate": 4.452634977739809e-06,
      "loss": 0.5409,
      "step": 91140
    },
    {
      "epoch": 37.31,
      "grad_norm": 2.0323774814605713,
      "learning_rate": 4.452565759455523e-06,
      "loss": 0.5591,
      "step": 91150
    },
    {
      "epoch": 37.31,
      "grad_norm": 2.071072816848755,
      "learning_rate": 4.452496531511077e-06,
      "loss": 0.5716,
      "step": 91160
    },
    {
      "epoch": 37.32,
      "grad_norm": 1.968706727027893,
      "learning_rate": 4.452427293906791e-06,
      "loss": 0.5703,
      "step": 91170
    },
    {
      "epoch": 37.32,
      "grad_norm": 2.5198051929473877,
      "learning_rate": 4.452358046642981e-06,
      "loss": 0.549,
      "step": 91180
    },
    {
      "epoch": 37.33,
      "grad_norm": 2.1453750133514404,
      "learning_rate": 4.452288789719962e-06,
      "loss": 0.56,
      "step": 91190
    },
    {
      "epoch": 37.33,
      "grad_norm": 2.2724082469940186,
      "learning_rate": 4.4522195231380555e-06,
      "loss": 0.562,
      "step": 91200
    },
    {
      "epoch": 37.34,
      "grad_norm": 2.523775339126587,
      "learning_rate": 4.452150246897576e-06,
      "loss": 0.5669,
      "step": 91210
    },
    {
      "epoch": 37.34,
      "grad_norm": 2.614192008972168,
      "learning_rate": 4.45208096099884e-06,
      "loss": 0.571,
      "step": 91220
    },
    {
      "epoch": 37.34,
      "grad_norm": 2.1167967319488525,
      "learning_rate": 4.452011665442169e-06,
      "loss": 0.5546,
      "step": 91230
    },
    {
      "epoch": 37.35,
      "grad_norm": 2.6193482875823975,
      "learning_rate": 4.4519423602278765e-06,
      "loss": 0.579,
      "step": 91240
    },
    {
      "epoch": 37.35,
      "grad_norm": 1.8887745141983032,
      "learning_rate": 4.451873045356281e-06,
      "loss": 0.5551,
      "step": 91250
    },
    {
      "epoch": 37.36,
      "grad_norm": 2.81600284576416,
      "learning_rate": 4.4518037208277005e-06,
      "loss": 0.5616,
      "step": 91260
    },
    {
      "epoch": 37.36,
      "grad_norm": 2.3358657360076904,
      "learning_rate": 4.4517343866424526e-06,
      "loss": 0.572,
      "step": 91270
    },
    {
      "epoch": 37.36,
      "grad_norm": 1.742424488067627,
      "learning_rate": 4.4516650428008545e-06,
      "loss": 0.5615,
      "step": 91280
    },
    {
      "epoch": 37.37,
      "grad_norm": 1.8711259365081787,
      "learning_rate": 4.451595689303224e-06,
      "loss": 0.579,
      "step": 91290
    },
    {
      "epoch": 37.37,
      "grad_norm": 1.961875319480896,
      "learning_rate": 4.451526326149879e-06,
      "loss": 0.5486,
      "step": 91300
    },
    {
      "epoch": 37.38,
      "grad_norm": 3.1472342014312744,
      "learning_rate": 4.451456953341138e-06,
      "loss": 0.5777,
      "step": 91310
    },
    {
      "epoch": 37.38,
      "grad_norm": 1.9936649799346924,
      "learning_rate": 4.451387570877317e-06,
      "loss": 0.5568,
      "step": 91320
    },
    {
      "epoch": 37.38,
      "grad_norm": 2.4545657634735107,
      "learning_rate": 4.451318178758734e-06,
      "loss": 0.5502,
      "step": 91330
    },
    {
      "epoch": 37.39,
      "grad_norm": 1.935708999633789,
      "learning_rate": 4.451248776985708e-06,
      "loss": 0.557,
      "step": 91340
    },
    {
      "epoch": 37.39,
      "grad_norm": 2.3863797187805176,
      "learning_rate": 4.451179365558557e-06,
      "loss": 0.5469,
      "step": 91350
    },
    {
      "epoch": 37.4,
      "grad_norm": 2.067732095718384,
      "learning_rate": 4.451109944477598e-06,
      "loss": 0.5579,
      "step": 91360
    },
    {
      "epoch": 37.4,
      "grad_norm": 1.807321548461914,
      "learning_rate": 4.45104051374315e-06,
      "loss": 0.5459,
      "step": 91370
    },
    {
      "epoch": 37.4,
      "grad_norm": 2.599388360977173,
      "learning_rate": 4.45097107335553e-06,
      "loss": 0.5643,
      "step": 91380
    },
    {
      "epoch": 37.41,
      "grad_norm": 3.202481985092163,
      "learning_rate": 4.450901623315056e-06,
      "loss": 0.5914,
      "step": 91390
    },
    {
      "epoch": 37.41,
      "grad_norm": 2.6473817825317383,
      "learning_rate": 4.450832163622048e-06,
      "loss": 0.5603,
      "step": 91400
    },
    {
      "epoch": 37.42,
      "grad_norm": 1.6004638671875,
      "learning_rate": 4.4507626942768224e-06,
      "loss": 0.5566,
      "step": 91410
    },
    {
      "epoch": 37.42,
      "grad_norm": 2.19944429397583,
      "learning_rate": 4.450693215279697e-06,
      "loss": 0.5724,
      "step": 91420
    },
    {
      "epoch": 37.43,
      "grad_norm": 2.0978200435638428,
      "learning_rate": 4.450623726630992e-06,
      "loss": 0.5482,
      "step": 91430
    },
    {
      "epoch": 37.43,
      "grad_norm": 2.0603363513946533,
      "learning_rate": 4.450554228331025e-06,
      "loss": 0.566,
      "step": 91440
    },
    {
      "epoch": 37.43,
      "grad_norm": 1.970702052116394,
      "learning_rate": 4.450484720380115e-06,
      "loss": 0.5603,
      "step": 91450
    },
    {
      "epoch": 37.44,
      "grad_norm": 1.7207789421081543,
      "learning_rate": 4.450415202778578e-06,
      "loss": 0.5525,
      "step": 91460
    },
    {
      "epoch": 37.44,
      "grad_norm": 2.3332226276397705,
      "learning_rate": 4.4503456755267335e-06,
      "loss": 0.5826,
      "step": 91470
    },
    {
      "epoch": 37.45,
      "grad_norm": 2.8150436878204346,
      "learning_rate": 4.450276138624902e-06,
      "loss": 0.5528,
      "step": 91480
    },
    {
      "epoch": 37.45,
      "grad_norm": 2.045396327972412,
      "learning_rate": 4.4502065920734e-06,
      "loss": 0.5627,
      "step": 91490
    },
    {
      "epoch": 37.45,
      "grad_norm": 2.5339841842651367,
      "learning_rate": 4.450137035872546e-06,
      "loss": 0.5812,
      "step": 91500
    },
    {
      "epoch": 37.46,
      "grad_norm": 1.8927080631256104,
      "learning_rate": 4.45006747002266e-06,
      "loss": 0.5465,
      "step": 91510
    },
    {
      "epoch": 37.46,
      "grad_norm": 2.3889214992523193,
      "learning_rate": 4.449997894524059e-06,
      "loss": 0.5876,
      "step": 91520
    },
    {
      "epoch": 37.47,
      "grad_norm": 2.477280616760254,
      "learning_rate": 4.449928309377063e-06,
      "loss": 0.5527,
      "step": 91530
    },
    {
      "epoch": 37.47,
      "grad_norm": 3.0132672786712646,
      "learning_rate": 4.449858714581991e-06,
      "loss": 0.5616,
      "step": 91540
    },
    {
      "epoch": 37.47,
      "grad_norm": 2.734504461288452,
      "learning_rate": 4.44978911013916e-06,
      "loss": 0.5615,
      "step": 91550
    },
    {
      "epoch": 37.48,
      "grad_norm": 2.2211835384368896,
      "learning_rate": 4.44971949604889e-06,
      "loss": 0.565,
      "step": 91560
    },
    {
      "epoch": 37.48,
      "grad_norm": 2.2906105518341064,
      "learning_rate": 4.449649872311501e-06,
      "loss": 0.5479,
      "step": 91570
    },
    {
      "epoch": 37.49,
      "grad_norm": 2.7467215061187744,
      "learning_rate": 4.449580238927311e-06,
      "loss": 0.5594,
      "step": 91580
    },
    {
      "epoch": 37.49,
      "grad_norm": 1.796919345855713,
      "learning_rate": 4.449510595896638e-06,
      "loss": 0.5494,
      "step": 91590
    },
    {
      "epoch": 37.49,
      "grad_norm": 1.9409672021865845,
      "learning_rate": 4.449440943219801e-06,
      "loss": 0.5507,
      "step": 91600
    },
    {
      "epoch": 37.5,
      "grad_norm": 2.0737996101379395,
      "learning_rate": 4.449371280897121e-06,
      "loss": 0.5481,
      "step": 91610
    },
    {
      "epoch": 37.5,
      "grad_norm": 3.004495859146118,
      "learning_rate": 4.449301608928916e-06,
      "loss": 0.5633,
      "step": 91620
    },
    {
      "epoch": 37.51,
      "grad_norm": 1.9126514196395874,
      "learning_rate": 4.4492319273155045e-06,
      "loss": 0.5495,
      "step": 91630
    },
    {
      "epoch": 37.51,
      "grad_norm": 2.013500213623047,
      "learning_rate": 4.449162236057206e-06,
      "loss": 0.5623,
      "step": 91640
    },
    {
      "epoch": 37.52,
      "grad_norm": 1.9315907955169678,
      "learning_rate": 4.449092535154341e-06,
      "loss": 0.5565,
      "step": 91650
    },
    {
      "epoch": 37.52,
      "grad_norm": 2.755403757095337,
      "learning_rate": 4.4490228246072285e-06,
      "loss": 0.5522,
      "step": 91660
    },
    {
      "epoch": 37.52,
      "grad_norm": 2.4049551486968994,
      "learning_rate": 4.4489531044161865e-06,
      "loss": 0.5979,
      "step": 91670
    },
    {
      "epoch": 37.53,
      "grad_norm": 2.0642964839935303,
      "learning_rate": 4.4488833745815355e-06,
      "loss": 0.5593,
      "step": 91680
    },
    {
      "epoch": 37.53,
      "grad_norm": 1.829854965209961,
      "learning_rate": 4.448813635103594e-06,
      "loss": 0.5637,
      "step": 91690
    },
    {
      "epoch": 37.54,
      "grad_norm": 1.778753638267517,
      "learning_rate": 4.448743885982683e-06,
      "loss": 0.5592,
      "step": 91700
    },
    {
      "epoch": 37.54,
      "grad_norm": 1.908333659172058,
      "learning_rate": 4.448674127219119e-06,
      "loss": 0.5572,
      "step": 91710
    },
    {
      "epoch": 37.54,
      "grad_norm": 2.435680627822876,
      "learning_rate": 4.448604358813225e-06,
      "loss": 0.5427,
      "step": 91720
    },
    {
      "epoch": 37.55,
      "grad_norm": 2.104217529296875,
      "learning_rate": 4.448534580765319e-06,
      "loss": 0.5626,
      "step": 91730
    },
    {
      "epoch": 37.55,
      "grad_norm": 2.35968279838562,
      "learning_rate": 4.448464793075721e-06,
      "loss": 0.5774,
      "step": 91740
    },
    {
      "epoch": 37.56,
      "grad_norm": 2.850116729736328,
      "learning_rate": 4.44839499574475e-06,
      "loss": 0.5413,
      "step": 91750
    },
    {
      "epoch": 37.56,
      "grad_norm": 2.074540615081787,
      "learning_rate": 4.448325188772727e-06,
      "loss": 0.554,
      "step": 91760
    },
    {
      "epoch": 37.56,
      "grad_norm": 1.7477155923843384,
      "learning_rate": 4.4482553721599695e-06,
      "loss": 0.5558,
      "step": 91770
    },
    {
      "epoch": 37.57,
      "grad_norm": 2.4066317081451416,
      "learning_rate": 4.4481855459068e-06,
      "loss": 0.5662,
      "step": 91780
    },
    {
      "epoch": 37.57,
      "grad_norm": 2.268728494644165,
      "learning_rate": 4.448115710013537e-06,
      "loss": 0.562,
      "step": 91790
    },
    {
      "epoch": 37.58,
      "grad_norm": 2.350914478302002,
      "learning_rate": 4.4480458644805006e-06,
      "loss": 0.5542,
      "step": 91800
    },
    {
      "epoch": 37.58,
      "grad_norm": 1.727189302444458,
      "learning_rate": 4.447976009308011e-06,
      "loss": 0.5526,
      "step": 91810
    },
    {
      "epoch": 37.58,
      "grad_norm": 1.8314639329910278,
      "learning_rate": 4.447906144496387e-06,
      "loss": 0.5663,
      "step": 91820
    },
    {
      "epoch": 37.59,
      "grad_norm": 2.4349124431610107,
      "learning_rate": 4.44783627004595e-06,
      "loss": 0.5686,
      "step": 91830
    },
    {
      "epoch": 37.59,
      "grad_norm": 2.436824321746826,
      "learning_rate": 4.44776638595702e-06,
      "loss": 0.5689,
      "step": 91840
    },
    {
      "epoch": 37.6,
      "grad_norm": 2.3059298992156982,
      "learning_rate": 4.447696492229916e-06,
      "loss": 0.5523,
      "step": 91850
    },
    {
      "epoch": 37.6,
      "grad_norm": 3.2270407676696777,
      "learning_rate": 4.44762658886496e-06,
      "loss": 0.5481,
      "step": 91860
    },
    {
      "epoch": 37.61,
      "grad_norm": 2.5987768173217773,
      "learning_rate": 4.447556675862471e-06,
      "loss": 0.5671,
      "step": 91870
    },
    {
      "epoch": 37.61,
      "grad_norm": 2.6899240016937256,
      "learning_rate": 4.4474867532227696e-06,
      "loss": 0.5462,
      "step": 91880
    },
    {
      "epoch": 37.61,
      "grad_norm": 2.0213661193847656,
      "learning_rate": 4.447416820946175e-06,
      "loss": 0.5387,
      "step": 91890
    },
    {
      "epoch": 37.62,
      "grad_norm": 1.9534794092178345,
      "learning_rate": 4.447346879033009e-06,
      "loss": 0.5651,
      "step": 91900
    },
    {
      "epoch": 37.62,
      "grad_norm": 2.2411012649536133,
      "learning_rate": 4.447276927483591e-06,
      "loss": 0.5559,
      "step": 91910
    },
    {
      "epoch": 37.63,
      "grad_norm": 3.2102389335632324,
      "learning_rate": 4.447206966298243e-06,
      "loss": 0.5828,
      "step": 91920
    },
    {
      "epoch": 37.63,
      "grad_norm": 2.419139862060547,
      "learning_rate": 4.447136995477283e-06,
      "loss": 0.5526,
      "step": 91930
    },
    {
      "epoch": 37.63,
      "grad_norm": 1.8455878496170044,
      "learning_rate": 4.447067015021034e-06,
      "loss": 0.5765,
      "step": 91940
    },
    {
      "epoch": 37.64,
      "grad_norm": 2.1147148609161377,
      "learning_rate": 4.446997024929815e-06,
      "loss": 0.5512,
      "step": 91950
    },
    {
      "epoch": 37.64,
      "grad_norm": 2.522895574569702,
      "learning_rate": 4.446927025203948e-06,
      "loss": 0.5881,
      "step": 91960
    },
    {
      "epoch": 37.65,
      "grad_norm": 1.8507634401321411,
      "learning_rate": 4.446857015843752e-06,
      "loss": 0.5735,
      "step": 91970
    },
    {
      "epoch": 37.65,
      "grad_norm": 2.2427806854248047,
      "learning_rate": 4.446786996849548e-06,
      "loss": 0.5635,
      "step": 91980
    },
    {
      "epoch": 37.65,
      "grad_norm": 2.8436765670776367,
      "learning_rate": 4.446716968221658e-06,
      "loss": 0.552,
      "step": 91990
    },
    {
      "epoch": 37.66,
      "grad_norm": 2.1376254558563232,
      "learning_rate": 4.446646929960402e-06,
      "loss": 0.5684,
      "step": 92000
    },
    {
      "epoch": 37.66,
      "grad_norm": 1.6952224969863892,
      "learning_rate": 4.4465768820661005e-06,
      "loss": 0.559,
      "step": 92010
    },
    {
      "epoch": 37.67,
      "grad_norm": 2.1908559799194336,
      "learning_rate": 4.4465068245390745e-06,
      "loss": 0.5546,
      "step": 92020
    },
    {
      "epoch": 37.67,
      "grad_norm": 1.9551849365234375,
      "learning_rate": 4.446436757379646e-06,
      "loss": 0.5564,
      "step": 92030
    },
    {
      "epoch": 37.67,
      "grad_norm": 1.8459904193878174,
      "learning_rate": 4.4463666805881345e-06,
      "loss": 0.5596,
      "step": 92040
    },
    {
      "epoch": 37.68,
      "grad_norm": 1.985071063041687,
      "learning_rate": 4.446296594164861e-06,
      "loss": 0.5645,
      "step": 92050
    },
    {
      "epoch": 37.68,
      "grad_norm": 1.7119485139846802,
      "learning_rate": 4.446226498110148e-06,
      "loss": 0.5739,
      "step": 92060
    },
    {
      "epoch": 37.69,
      "grad_norm": 2.3969388008117676,
      "learning_rate": 4.446156392424315e-06,
      "loss": 0.5581,
      "step": 92070
    },
    {
      "epoch": 37.69,
      "grad_norm": 2.069458246231079,
      "learning_rate": 4.446086277107685e-06,
      "loss": 0.5568,
      "step": 92080
    },
    {
      "epoch": 37.7,
      "grad_norm": 2.485933542251587,
      "learning_rate": 4.4460161521605776e-06,
      "loss": 0.5565,
      "step": 92090
    },
    {
      "epoch": 37.7,
      "grad_norm": 3.00553035736084,
      "learning_rate": 4.445946017583315e-06,
      "loss": 0.5526,
      "step": 92100
    },
    {
      "epoch": 37.7,
      "grad_norm": 2.912783145904541,
      "learning_rate": 4.445875873376217e-06,
      "loss": 0.5577,
      "step": 92110
    },
    {
      "epoch": 37.71,
      "grad_norm": 2.320150852203369,
      "learning_rate": 4.4458057195396065e-06,
      "loss": 0.5679,
      "step": 92120
    },
    {
      "epoch": 37.71,
      "grad_norm": 2.23464035987854,
      "learning_rate": 4.445735556073804e-06,
      "loss": 0.5758,
      "step": 92130
    },
    {
      "epoch": 37.72,
      "grad_norm": 2.5032718181610107,
      "learning_rate": 4.445665382979132e-06,
      "loss": 0.56,
      "step": 92140
    },
    {
      "epoch": 37.72,
      "grad_norm": 1.6204193830490112,
      "learning_rate": 4.4455952002559105e-06,
      "loss": 0.5621,
      "step": 92150
    },
    {
      "epoch": 37.72,
      "grad_norm": 1.9222426414489746,
      "learning_rate": 4.4455250079044624e-06,
      "loss": 0.5817,
      "step": 92160
    },
    {
      "epoch": 37.73,
      "grad_norm": 2.49764084815979,
      "learning_rate": 4.445454805925108e-06,
      "loss": 0.5579,
      "step": 92170
    },
    {
      "epoch": 37.73,
      "grad_norm": 2.216549873352051,
      "learning_rate": 4.445384594318169e-06,
      "loss": 0.5464,
      "step": 92180
    },
    {
      "epoch": 37.74,
      "grad_norm": 2.517822742462158,
      "learning_rate": 4.445314373083967e-06,
      "loss": 0.5754,
      "step": 92190
    },
    {
      "epoch": 37.74,
      "grad_norm": 2.2711021900177,
      "learning_rate": 4.445244142222825e-06,
      "loss": 0.5481,
      "step": 92200
    },
    {
      "epoch": 37.74,
      "grad_norm": 2.0841245651245117,
      "learning_rate": 4.445173901735063e-06,
      "loss": 0.5765,
      "step": 92210
    },
    {
      "epoch": 37.75,
      "grad_norm": 2.6387107372283936,
      "learning_rate": 4.4451036516210046e-06,
      "loss": 0.5616,
      "step": 92220
    },
    {
      "epoch": 37.75,
      "grad_norm": 2.28464674949646,
      "learning_rate": 4.44503339188097e-06,
      "loss": 0.5399,
      "step": 92230
    },
    {
      "epoch": 37.76,
      "grad_norm": 2.3746817111968994,
      "learning_rate": 4.444963122515282e-06,
      "loss": 0.5804,
      "step": 92240
    },
    {
      "epoch": 37.76,
      "grad_norm": 2.405324697494507,
      "learning_rate": 4.444892843524262e-06,
      "loss": 0.5414,
      "step": 92250
    },
    {
      "epoch": 37.77,
      "grad_norm": 2.278794288635254,
      "learning_rate": 4.444822554908231e-06,
      "loss": 0.547,
      "step": 92260
    },
    {
      "epoch": 37.77,
      "grad_norm": 3.134263038635254,
      "learning_rate": 4.444752256667513e-06,
      "loss": 0.5548,
      "step": 92270
    },
    {
      "epoch": 37.77,
      "grad_norm": 1.9558311700820923,
      "learning_rate": 4.444681948802429e-06,
      "loss": 0.5575,
      "step": 92280
    },
    {
      "epoch": 37.78,
      "grad_norm": 2.00883150100708,
      "learning_rate": 4.444611631313301e-06,
      "loss": 0.5306,
      "step": 92290
    },
    {
      "epoch": 37.78,
      "grad_norm": 2.591524839401245,
      "learning_rate": 4.4445413042004515e-06,
      "loss": 0.5561,
      "step": 92300
    },
    {
      "epoch": 37.79,
      "grad_norm": 2.4773378372192383,
      "learning_rate": 4.444470967464202e-06,
      "loss": 0.5447,
      "step": 92310
    },
    {
      "epoch": 37.79,
      "grad_norm": 1.9223852157592773,
      "learning_rate": 4.444400621104874e-06,
      "loss": 0.5667,
      "step": 92320
    },
    {
      "epoch": 37.79,
      "grad_norm": 2.195277214050293,
      "learning_rate": 4.444330265122793e-06,
      "loss": 0.5657,
      "step": 92330
    },
    {
      "epoch": 37.8,
      "grad_norm": 2.7521157264709473,
      "learning_rate": 4.444259899518278e-06,
      "loss": 0.5657,
      "step": 92340
    },
    {
      "epoch": 37.8,
      "grad_norm": 2.5975921154022217,
      "learning_rate": 4.444189524291652e-06,
      "loss": 0.5446,
      "step": 92350
    },
    {
      "epoch": 37.81,
      "grad_norm": 2.2947680950164795,
      "learning_rate": 4.444119139443239e-06,
      "loss": 0.5543,
      "step": 92360
    },
    {
      "epoch": 37.81,
      "grad_norm": 2.756300926208496,
      "learning_rate": 4.444048744973359e-06,
      "loss": 0.5592,
      "step": 92370
    },
    {
      "epoch": 37.81,
      "grad_norm": 2.1544458866119385,
      "learning_rate": 4.443978340882337e-06,
      "loss": 0.5512,
      "step": 92380
    },
    {
      "epoch": 37.82,
      "grad_norm": 2.0706074237823486,
      "learning_rate": 4.443907927170494e-06,
      "loss": 0.5544,
      "step": 92390
    },
    {
      "epoch": 37.82,
      "grad_norm": 2.169818639755249,
      "learning_rate": 4.443837503838152e-06,
      "loss": 0.5657,
      "step": 92400
    },
    {
      "epoch": 37.83,
      "grad_norm": 1.9475692510604858,
      "learning_rate": 4.443767070885635e-06,
      "loss": 0.5729,
      "step": 92410
    },
    {
      "epoch": 37.83,
      "grad_norm": 2.289388656616211,
      "learning_rate": 4.4436966283132645e-06,
      "loss": 0.5549,
      "step": 92420
    },
    {
      "epoch": 37.83,
      "grad_norm": 2.1868326663970947,
      "learning_rate": 4.443626176121364e-06,
      "loss": 0.5586,
      "step": 92430
    },
    {
      "epoch": 37.84,
      "grad_norm": 2.4380457401275635,
      "learning_rate": 4.443555714310256e-06,
      "loss": 0.5324,
      "step": 92440
    },
    {
      "epoch": 37.84,
      "grad_norm": 1.9275457859039307,
      "learning_rate": 4.443485242880263e-06,
      "loss": 0.5612,
      "step": 92450
    },
    {
      "epoch": 37.85,
      "grad_norm": 2.4803173542022705,
      "learning_rate": 4.443414761831709e-06,
      "loss": 0.5603,
      "step": 92460
    },
    {
      "epoch": 37.85,
      "grad_norm": 2.2650580406188965,
      "learning_rate": 4.443344271164914e-06,
      "loss": 0.5577,
      "step": 92470
    },
    {
      "epoch": 37.86,
      "grad_norm": 3.4525949954986572,
      "learning_rate": 4.443273770880204e-06,
      "loss": 0.5757,
      "step": 92480
    },
    {
      "epoch": 37.86,
      "grad_norm": 2.0312211513519287,
      "learning_rate": 4.443203260977901e-06,
      "loss": 0.5595,
      "step": 92490
    },
    {
      "epoch": 37.86,
      "grad_norm": 1.684416651725769,
      "learning_rate": 4.443132741458327e-06,
      "loss": 0.559,
      "step": 92500
    },
    {
      "epoch": 37.87,
      "grad_norm": 3.1819286346435547,
      "learning_rate": 4.443062212321806e-06,
      "loss": 0.5403,
      "step": 92510
    },
    {
      "epoch": 37.87,
      "grad_norm": 2.14400053024292,
      "learning_rate": 4.442991673568662e-06,
      "loss": 0.5432,
      "step": 92520
    },
    {
      "epoch": 37.88,
      "grad_norm": 2.5095865726470947,
      "learning_rate": 4.442921125199216e-06,
      "loss": 0.5641,
      "step": 92530
    },
    {
      "epoch": 37.88,
      "grad_norm": 2.216278314590454,
      "learning_rate": 4.442850567213792e-06,
      "loss": 0.5694,
      "step": 92540
    },
    {
      "epoch": 37.88,
      "grad_norm": 1.9216251373291016,
      "learning_rate": 4.442779999612714e-06,
      "loss": 0.557,
      "step": 92550
    },
    {
      "epoch": 37.89,
      "grad_norm": 2.0880773067474365,
      "learning_rate": 4.442709422396304e-06,
      "loss": 0.5548,
      "step": 92560
    },
    {
      "epoch": 37.89,
      "grad_norm": 2.017435073852539,
      "learning_rate": 4.442638835564886e-06,
      "loss": 0.552,
      "step": 92570
    },
    {
      "epoch": 37.9,
      "grad_norm": 2.3426082134246826,
      "learning_rate": 4.442568239118784e-06,
      "loss": 0.5686,
      "step": 92580
    },
    {
      "epoch": 37.9,
      "grad_norm": 2.3320248126983643,
      "learning_rate": 4.4424976330583195e-06,
      "loss": 0.5391,
      "step": 92590
    },
    {
      "epoch": 37.9,
      "grad_norm": 2.0752756595611572,
      "learning_rate": 4.442427017383819e-06,
      "loss": 0.5605,
      "step": 92600
    },
    {
      "epoch": 37.91,
      "grad_norm": 2.3379082679748535,
      "learning_rate": 4.442356392095602e-06,
      "loss": 0.5616,
      "step": 92610
    },
    {
      "epoch": 37.91,
      "grad_norm": 2.039828300476074,
      "learning_rate": 4.442285757193996e-06,
      "loss": 0.5571,
      "step": 92620
    },
    {
      "epoch": 37.92,
      "grad_norm": 2.3882341384887695,
      "learning_rate": 4.4422151126793215e-06,
      "loss": 0.5678,
      "step": 92630
    },
    {
      "epoch": 37.92,
      "grad_norm": 1.637789011001587,
      "learning_rate": 4.442144458551903e-06,
      "loss": 0.5816,
      "step": 92640
    },
    {
      "epoch": 37.92,
      "grad_norm": 2.4617297649383545,
      "learning_rate": 4.4420737948120655e-06,
      "loss": 0.5594,
      "step": 92650
    },
    {
      "epoch": 37.93,
      "grad_norm": 2.1096835136413574,
      "learning_rate": 4.442003121460132e-06,
      "loss": 0.548,
      "step": 92660
    },
    {
      "epoch": 37.93,
      "grad_norm": 2.204312801361084,
      "learning_rate": 4.441932438496425e-06,
      "loss": 0.5366,
      "step": 92670
    },
    {
      "epoch": 37.94,
      "grad_norm": 1.6052560806274414,
      "learning_rate": 4.441861745921268e-06,
      "loss": 0.5623,
      "step": 92680
    },
    {
      "epoch": 37.94,
      "grad_norm": 1.9642878770828247,
      "learning_rate": 4.441791043734988e-06,
      "loss": 0.568,
      "step": 92690
    },
    {
      "epoch": 37.95,
      "grad_norm": 2.1960809230804443,
      "learning_rate": 4.441720331937906e-06,
      "loss": 0.5739,
      "step": 92700
    },
    {
      "epoch": 37.95,
      "grad_norm": 2.4221208095550537,
      "learning_rate": 4.441649610530347e-06,
      "loss": 0.5487,
      "step": 92710
    },
    {
      "epoch": 37.95,
      "grad_norm": 2.2561686038970947,
      "learning_rate": 4.4415788795126355e-06,
      "loss": 0.5533,
      "step": 92720
    },
    {
      "epoch": 37.96,
      "grad_norm": 2.250863552093506,
      "learning_rate": 4.441508138885094e-06,
      "loss": 0.5707,
      "step": 92730
    },
    {
      "epoch": 37.96,
      "grad_norm": 1.6797795295715332,
      "learning_rate": 4.441437388648047e-06,
      "loss": 0.5569,
      "step": 92740
    },
    {
      "epoch": 37.97,
      "grad_norm": 2.223003625869751,
      "learning_rate": 4.441366628801819e-06,
      "loss": 0.5712,
      "step": 92750
    },
    {
      "epoch": 37.97,
      "grad_norm": 1.8716089725494385,
      "learning_rate": 4.441295859346735e-06,
      "loss": 0.5615,
      "step": 92760
    },
    {
      "epoch": 37.97,
      "grad_norm": 1.8375948667526245,
      "learning_rate": 4.441225080283118e-06,
      "loss": 0.5514,
      "step": 92770
    },
    {
      "epoch": 37.98,
      "grad_norm": 3.2548322677612305,
      "learning_rate": 4.441154291611292e-06,
      "loss": 0.554,
      "step": 92780
    },
    {
      "epoch": 37.98,
      "grad_norm": 2.485487222671509,
      "learning_rate": 4.441083493331582e-06,
      "loss": 0.5761,
      "step": 92790
    },
    {
      "epoch": 37.99,
      "grad_norm": 2.0159707069396973,
      "learning_rate": 4.4410126854443125e-06,
      "loss": 0.5671,
      "step": 92800
    },
    {
      "epoch": 37.99,
      "grad_norm": 2.6894493103027344,
      "learning_rate": 4.4409418679498065e-06,
      "loss": 0.5547,
      "step": 92810
    },
    {
      "epoch": 37.99,
      "grad_norm": 2.102858781814575,
      "learning_rate": 4.440871040848391e-06,
      "loss": 0.5524,
      "step": 92820
    },
    {
      "epoch": 38.0,
      "grad_norm": 2.300069570541382,
      "learning_rate": 4.4408002041403875e-06,
      "loss": 0.5616,
      "step": 92830
    },
    {
      "epoch": 38.0,
      "eval_loss": 0.5626325607299805,
      "eval_runtime": 52.0429,
      "eval_samples_per_second": 66.272,
      "eval_steps_per_second": 8.301,
      "step": 92834
    },
    {
      "epoch": 38.0,
      "grad_norm": 2.2236149311065674,
      "learning_rate": 4.440729357826122e-06,
      "loss": 0.5743,
      "step": 92840
    },
    {
      "epoch": 38.01,
      "grad_norm": 2.2606921195983887,
      "learning_rate": 4.440658501905918e-06,
      "loss": 0.5442,
      "step": 92850
    },
    {
      "epoch": 38.01,
      "grad_norm": 2.4950966835021973,
      "learning_rate": 4.440587636380102e-06,
      "loss": 0.5744,
      "step": 92860
    },
    {
      "epoch": 38.01,
      "grad_norm": 2.357168197631836,
      "learning_rate": 4.440516761248997e-06,
      "loss": 0.5479,
      "step": 92870
    },
    {
      "epoch": 38.02,
      "grad_norm": 1.700644850730896,
      "learning_rate": 4.440445876512928e-06,
      "loss": 0.555,
      "step": 92880
    },
    {
      "epoch": 38.02,
      "grad_norm": 2.2739450931549072,
      "learning_rate": 4.440374982172221e-06,
      "loss": 0.5481,
      "step": 92890
    },
    {
      "epoch": 38.03,
      "grad_norm": 2.4296481609344482,
      "learning_rate": 4.440304078227199e-06,
      "loss": 0.5485,
      "step": 92900
    },
    {
      "epoch": 38.03,
      "grad_norm": 3.045168876647949,
      "learning_rate": 4.440233164678187e-06,
      "loss": 0.5453,
      "step": 92910
    },
    {
      "epoch": 38.04,
      "grad_norm": 2.572361469268799,
      "learning_rate": 4.440162241525511e-06,
      "loss": 0.5646,
      "step": 92920
    },
    {
      "epoch": 38.04,
      "grad_norm": 2.1689915657043457,
      "learning_rate": 4.440091308769495e-06,
      "loss": 0.5395,
      "step": 92930
    },
    {
      "epoch": 38.04,
      "grad_norm": 2.376094341278076,
      "learning_rate": 4.440020366410464e-06,
      "loss": 0.5433,
      "step": 92940
    },
    {
      "epoch": 38.05,
      "grad_norm": 2.1975677013397217,
      "learning_rate": 4.439949414448743e-06,
      "loss": 0.5732,
      "step": 92950
    },
    {
      "epoch": 38.05,
      "grad_norm": 1.8448774814605713,
      "learning_rate": 4.439878452884657e-06,
      "loss": 0.5675,
      "step": 92960
    },
    {
      "epoch": 38.06,
      "grad_norm": 1.986367106437683,
      "learning_rate": 4.439807481718532e-06,
      "loss": 0.5716,
      "step": 92970
    },
    {
      "epoch": 38.06,
      "grad_norm": 2.6348822116851807,
      "learning_rate": 4.439736500950692e-06,
      "loss": 0.5738,
      "step": 92980
    },
    {
      "epoch": 38.06,
      "grad_norm": 1.7115345001220703,
      "learning_rate": 4.4396655105814615e-06,
      "loss": 0.565,
      "step": 92990
    },
    {
      "epoch": 38.07,
      "grad_norm": 2.1076812744140625,
      "learning_rate": 4.439594510611168e-06,
      "loss": 0.5489,
      "step": 93000
    },
    {
      "epoch": 38.07,
      "grad_norm": 2.9390227794647217,
      "learning_rate": 4.439523501040134e-06,
      "loss": 0.5515,
      "step": 93010
    },
    {
      "epoch": 38.08,
      "grad_norm": 1.952335238456726,
      "learning_rate": 4.439452481868687e-06,
      "loss": 0.556,
      "step": 93020
    },
    {
      "epoch": 38.08,
      "grad_norm": 2.326477527618408,
      "learning_rate": 4.439381453097151e-06,
      "loss": 0.5606,
      "step": 93030
    },
    {
      "epoch": 38.08,
      "grad_norm": 1.9096190929412842,
      "learning_rate": 4.439310414725853e-06,
      "loss": 0.5397,
      "step": 93040
    },
    {
      "epoch": 38.09,
      "grad_norm": 2.3849520683288574,
      "learning_rate": 4.439239366755117e-06,
      "loss": 0.5691,
      "step": 93050
    },
    {
      "epoch": 38.09,
      "grad_norm": 2.349959135055542,
      "learning_rate": 4.439168309185267e-06,
      "loss": 0.5421,
      "step": 93060
    },
    {
      "epoch": 38.1,
      "grad_norm": 2.2835254669189453,
      "learning_rate": 4.439097242016632e-06,
      "loss": 0.5683,
      "step": 93070
    },
    {
      "epoch": 38.1,
      "grad_norm": 2.7076971530914307,
      "learning_rate": 4.439026165249535e-06,
      "loss": 0.546,
      "step": 93080
    },
    {
      "epoch": 38.1,
      "grad_norm": 2.2326247692108154,
      "learning_rate": 4.438955078884303e-06,
      "loss": 0.5605,
      "step": 93090
    },
    {
      "epoch": 38.11,
      "grad_norm": 1.7831822633743286,
      "learning_rate": 4.438883982921261e-06,
      "loss": 0.5779,
      "step": 93100
    },
    {
      "epoch": 38.11,
      "grad_norm": 2.4191033840179443,
      "learning_rate": 4.438812877360733e-06,
      "loss": 0.5592,
      "step": 93110
    },
    {
      "epoch": 38.12,
      "grad_norm": 1.9041297435760498,
      "learning_rate": 4.438741762203049e-06,
      "loss": 0.5546,
      "step": 93120
    },
    {
      "epoch": 38.12,
      "grad_norm": 3.6552531719207764,
      "learning_rate": 4.438670637448531e-06,
      "loss": 0.5492,
      "step": 93130
    },
    {
      "epoch": 38.13,
      "grad_norm": 3.0210909843444824,
      "learning_rate": 4.438599503097506e-06,
      "loss": 0.5812,
      "step": 93140
    },
    {
      "epoch": 38.13,
      "grad_norm": 2.7568819522857666,
      "learning_rate": 4.438528359150299e-06,
      "loss": 0.5807,
      "step": 93150
    },
    {
      "epoch": 38.13,
      "grad_norm": 1.8749114274978638,
      "learning_rate": 4.438457205607238e-06,
      "loss": 0.5606,
      "step": 93160
    },
    {
      "epoch": 38.14,
      "grad_norm": 2.012373208999634,
      "learning_rate": 4.438386042468648e-06,
      "loss": 0.5608,
      "step": 93170
    },
    {
      "epoch": 38.14,
      "grad_norm": 2.6022493839263916,
      "learning_rate": 4.4383148697348536e-06,
      "loss": 0.5683,
      "step": 93180
    },
    {
      "epoch": 38.15,
      "grad_norm": 2.6277616024017334,
      "learning_rate": 4.438243687406183e-06,
      "loss": 0.556,
      "step": 93190
    },
    {
      "epoch": 38.15,
      "grad_norm": 2.267522096633911,
      "learning_rate": 4.4381724954829595e-06,
      "loss": 0.5686,
      "step": 93200
    },
    {
      "epoch": 38.15,
      "grad_norm": 2.6011855602264404,
      "learning_rate": 4.438101293965512e-06,
      "loss": 0.565,
      "step": 93210
    },
    {
      "epoch": 38.16,
      "grad_norm": 2.0993082523345947,
      "learning_rate": 4.438030082854165e-06,
      "loss": 0.5555,
      "step": 93220
    },
    {
      "epoch": 38.16,
      "grad_norm": 2.2328994274139404,
      "learning_rate": 4.437958862149246e-06,
      "loss": 0.5739,
      "step": 93230
    },
    {
      "epoch": 38.17,
      "grad_norm": 2.9928407669067383,
      "learning_rate": 4.43788763185108e-06,
      "loss": 0.544,
      "step": 93240
    },
    {
      "epoch": 38.17,
      "grad_norm": 2.6577484607696533,
      "learning_rate": 4.437816391959994e-06,
      "loss": 0.5438,
      "step": 93250
    },
    {
      "epoch": 38.17,
      "grad_norm": 2.036163806915283,
      "learning_rate": 4.437745142476315e-06,
      "loss": 0.5567,
      "step": 93260
    },
    {
      "epoch": 38.18,
      "grad_norm": 2.9123756885528564,
      "learning_rate": 4.437673883400367e-06,
      "loss": 0.5554,
      "step": 93270
    },
    {
      "epoch": 38.18,
      "grad_norm": 2.2342121601104736,
      "learning_rate": 4.437602614732479e-06,
      "loss": 0.5483,
      "step": 93280
    },
    {
      "epoch": 38.19,
      "grad_norm": 1.8382275104522705,
      "learning_rate": 4.437531336472977e-06,
      "loss": 0.5715,
      "step": 93290
    },
    {
      "epoch": 38.19,
      "grad_norm": 2.282374620437622,
      "learning_rate": 4.437460048622186e-06,
      "loss": 0.5552,
      "step": 93300
    },
    {
      "epoch": 38.19,
      "grad_norm": 2.8784260749816895,
      "learning_rate": 4.437388751180433e-06,
      "loss": 0.5443,
      "step": 93310
    },
    {
      "epoch": 38.2,
      "grad_norm": 2.0382418632507324,
      "learning_rate": 4.437317444148046e-06,
      "loss": 0.5543,
      "step": 93320
    },
    {
      "epoch": 38.2,
      "grad_norm": 1.5884612798690796,
      "learning_rate": 4.437246127525352e-06,
      "loss": 0.5565,
      "step": 93330
    },
    {
      "epoch": 38.21,
      "grad_norm": 1.928520917892456,
      "learning_rate": 4.437174801312674e-06,
      "loss": 0.5544,
      "step": 93340
    },
    {
      "epoch": 38.21,
      "grad_norm": 3.078294038772583,
      "learning_rate": 4.437103465510342e-06,
      "loss": 0.5506,
      "step": 93350
    },
    {
      "epoch": 38.22,
      "grad_norm": 2.018171787261963,
      "learning_rate": 4.4370321201186825e-06,
      "loss": 0.5624,
      "step": 93360
    },
    {
      "epoch": 38.22,
      "grad_norm": 2.513150453567505,
      "learning_rate": 4.436960765138022e-06,
      "loss": 0.5446,
      "step": 93370
    },
    {
      "epoch": 38.22,
      "grad_norm": 1.9589953422546387,
      "learning_rate": 4.436889400568686e-06,
      "loss": 0.5604,
      "step": 93380
    },
    {
      "epoch": 38.23,
      "grad_norm": 3.045412063598633,
      "learning_rate": 4.436818026411004e-06,
      "loss": 0.5737,
      "step": 93390
    },
    {
      "epoch": 38.23,
      "grad_norm": 1.918710470199585,
      "learning_rate": 4.4367466426653e-06,
      "loss": 0.5492,
      "step": 93400
    },
    {
      "epoch": 38.24,
      "grad_norm": 1.638527512550354,
      "learning_rate": 4.436675249331904e-06,
      "loss": 0.5633,
      "step": 93410
    },
    {
      "epoch": 38.24,
      "grad_norm": 2.556030035018921,
      "learning_rate": 4.43660384641114e-06,
      "loss": 0.5588,
      "step": 93420
    },
    {
      "epoch": 38.24,
      "grad_norm": 2.027575731277466,
      "learning_rate": 4.436532433903338e-06,
      "loss": 0.55,
      "step": 93430
    },
    {
      "epoch": 38.25,
      "grad_norm": 1.62592613697052,
      "learning_rate": 4.436461011808823e-06,
      "loss": 0.5566,
      "step": 93440
    },
    {
      "epoch": 38.25,
      "grad_norm": 2.007873296737671,
      "learning_rate": 4.4363895801279235e-06,
      "loss": 0.5621,
      "step": 93450
    },
    {
      "epoch": 38.26,
      "grad_norm": 2.2384347915649414,
      "learning_rate": 4.4363181388609655e-06,
      "loss": 0.5608,
      "step": 93460
    },
    {
      "epoch": 38.26,
      "grad_norm": 1.802314043045044,
      "learning_rate": 4.436246688008278e-06,
      "loss": 0.5622,
      "step": 93470
    },
    {
      "epoch": 38.26,
      "grad_norm": 1.875178575515747,
      "learning_rate": 4.436175227570186e-06,
      "loss": 0.5532,
      "step": 93480
    },
    {
      "epoch": 38.27,
      "grad_norm": 1.6760729551315308,
      "learning_rate": 4.436103757547019e-06,
      "loss": 0.5683,
      "step": 93490
    },
    {
      "epoch": 38.27,
      "grad_norm": 2.3104333877563477,
      "learning_rate": 4.4360322779391025e-06,
      "loss": 0.5846,
      "step": 93500
    },
    {
      "epoch": 38.28,
      "grad_norm": 2.499420166015625,
      "learning_rate": 4.4359607887467654e-06,
      "loss": 0.5554,
      "step": 93510
    },
    {
      "epoch": 38.28,
      "grad_norm": 2.617450714111328,
      "learning_rate": 4.435889289970335e-06,
      "loss": 0.536,
      "step": 93520
    },
    {
      "epoch": 38.28,
      "grad_norm": 2.3279905319213867,
      "learning_rate": 4.435817781610139e-06,
      "loss": 0.5702,
      "step": 93530
    },
    {
      "epoch": 38.29,
      "grad_norm": 2.275756597518921,
      "learning_rate": 4.4357462636665035e-06,
      "loss": 0.564,
      "step": 93540
    },
    {
      "epoch": 38.29,
      "grad_norm": 2.0033514499664307,
      "learning_rate": 4.4356747361397575e-06,
      "loss": 0.5505,
      "step": 93550
    },
    {
      "epoch": 38.3,
      "grad_norm": 2.1780662536621094,
      "learning_rate": 4.435603199030228e-06,
      "loss": 0.5798,
      "step": 93560
    },
    {
      "epoch": 38.3,
      "grad_norm": 1.789036512374878,
      "learning_rate": 4.435531652338243e-06,
      "loss": 0.5573,
      "step": 93570
    },
    {
      "epoch": 38.31,
      "grad_norm": 2.4158143997192383,
      "learning_rate": 4.435460096064131e-06,
      "loss": 0.5771,
      "step": 93580
    },
    {
      "epoch": 38.31,
      "grad_norm": 2.108459949493408,
      "learning_rate": 4.435388530208218e-06,
      "loss": 0.5623,
      "step": 93590
    },
    {
      "epoch": 38.31,
      "grad_norm": 2.091301441192627,
      "learning_rate": 4.435316954770833e-06,
      "loss": 0.5678,
      "step": 93600
    },
    {
      "epoch": 38.32,
      "grad_norm": 1.9129345417022705,
      "learning_rate": 4.435245369752304e-06,
      "loss": 0.5605,
      "step": 93610
    },
    {
      "epoch": 38.32,
      "grad_norm": 1.9681051969528198,
      "learning_rate": 4.435173775152959e-06,
      "loss": 0.542,
      "step": 93620
    },
    {
      "epoch": 38.33,
      "grad_norm": 2.232431173324585,
      "learning_rate": 4.435102170973125e-06,
      "loss": 0.549,
      "step": 93630
    },
    {
      "epoch": 38.33,
      "grad_norm": 2.5836715698242188,
      "learning_rate": 4.43503055721313e-06,
      "loss": 0.5686,
      "step": 93640
    },
    {
      "epoch": 38.33,
      "grad_norm": 2.268216133117676,
      "learning_rate": 4.434958933873304e-06,
      "loss": 0.5802,
      "step": 93650
    },
    {
      "epoch": 38.34,
      "grad_norm": 2.172283411026001,
      "learning_rate": 4.434887300953973e-06,
      "loss": 0.5658,
      "step": 93660
    },
    {
      "epoch": 38.34,
      "grad_norm": 1.886804223060608,
      "learning_rate": 4.434815658455466e-06,
      "loss": 0.5636,
      "step": 93670
    },
    {
      "epoch": 38.35,
      "grad_norm": 2.3214404582977295,
      "learning_rate": 4.434744006378111e-06,
      "loss": 0.5511,
      "step": 93680
    },
    {
      "epoch": 38.35,
      "grad_norm": 7.830613613128662,
      "learning_rate": 4.434672344722237e-06,
      "loss": 0.5475,
      "step": 93690
    },
    {
      "epoch": 38.35,
      "grad_norm": 2.0191166400909424,
      "learning_rate": 4.43460067348817e-06,
      "loss": 0.5654,
      "step": 93700
    },
    {
      "epoch": 38.36,
      "grad_norm": 3.1199934482574463,
      "learning_rate": 4.434528992676241e-06,
      "loss": 0.5815,
      "step": 93710
    },
    {
      "epoch": 38.36,
      "grad_norm": 2.3681297302246094,
      "learning_rate": 4.4344573022867765e-06,
      "loss": 0.5374,
      "step": 93720
    },
    {
      "epoch": 38.37,
      "grad_norm": 1.7574328184127808,
      "learning_rate": 4.434385602320107e-06,
      "loss": 0.5533,
      "step": 93730
    },
    {
      "epoch": 38.37,
      "grad_norm": 2.2012321949005127,
      "learning_rate": 4.434313892776558e-06,
      "loss": 0.5706,
      "step": 93740
    },
    {
      "epoch": 38.37,
      "grad_norm": 2.4022774696350098,
      "learning_rate": 4.434242173656461e-06,
      "loss": 0.567,
      "step": 93750
    },
    {
      "epoch": 38.38,
      "grad_norm": 1.590698480606079,
      "learning_rate": 4.434170444960142e-06,
      "loss": 0.5474,
      "step": 93760
    },
    {
      "epoch": 38.38,
      "grad_norm": 2.150714159011841,
      "learning_rate": 4.43409870668793e-06,
      "loss": 0.567,
      "step": 93770
    },
    {
      "epoch": 38.39,
      "grad_norm": 1.9331305027008057,
      "learning_rate": 4.434026958840156e-06,
      "loss": 0.5495,
      "step": 93780
    },
    {
      "epoch": 38.39,
      "grad_norm": 1.9741119146347046,
      "learning_rate": 4.433955201417147e-06,
      "loss": 0.5506,
      "step": 93790
    },
    {
      "epoch": 38.4,
      "grad_norm": 2.592766284942627,
      "learning_rate": 4.43388343441923e-06,
      "loss": 0.5568,
      "step": 93800
    },
    {
      "epoch": 38.4,
      "grad_norm": 2.1210503578186035,
      "learning_rate": 4.433811657846737e-06,
      "loss": 0.5493,
      "step": 93810
    },
    {
      "epoch": 38.4,
      "grad_norm": 2.0791430473327637,
      "learning_rate": 4.433739871699995e-06,
      "loss": 0.5688,
      "step": 93820
    },
    {
      "epoch": 38.41,
      "grad_norm": 2.3410050868988037,
      "learning_rate": 4.433668075979332e-06,
      "loss": 0.5335,
      "step": 93830
    },
    {
      "epoch": 38.41,
      "grad_norm": 2.99078631401062,
      "learning_rate": 4.433596270685079e-06,
      "loss": 0.5543,
      "step": 93840
    },
    {
      "epoch": 38.42,
      "grad_norm": 2.27970552444458,
      "learning_rate": 4.433524455817563e-06,
      "loss": 0.5624,
      "step": 93850
    },
    {
      "epoch": 38.42,
      "grad_norm": 1.7769052982330322,
      "learning_rate": 4.433452631377115e-06,
      "loss": 0.5591,
      "step": 93860
    },
    {
      "epoch": 38.42,
      "grad_norm": 2.1825826168060303,
      "learning_rate": 4.433380797364063e-06,
      "loss": 0.5618,
      "step": 93870
    },
    {
      "epoch": 38.43,
      "grad_norm": 1.8763487339019775,
      "learning_rate": 4.433308953778734e-06,
      "loss": 0.568,
      "step": 93880
    },
    {
      "epoch": 38.43,
      "grad_norm": 1.9845465421676636,
      "learning_rate": 4.433237100621461e-06,
      "loss": 0.5746,
      "step": 93890
    },
    {
      "epoch": 38.44,
      "grad_norm": 1.793887734413147,
      "learning_rate": 4.4331652378925705e-06,
      "loss": 0.5636,
      "step": 93900
    },
    {
      "epoch": 38.44,
      "grad_norm": 1.7043395042419434,
      "learning_rate": 4.433093365592392e-06,
      "loss": 0.5562,
      "step": 93910
    },
    {
      "epoch": 38.44,
      "grad_norm": 1.8386138677597046,
      "learning_rate": 4.433021483721256e-06,
      "loss": 0.5669,
      "step": 93920
    },
    {
      "epoch": 38.45,
      "grad_norm": 1.7245311737060547,
      "learning_rate": 4.43294959227949e-06,
      "loss": 0.5649,
      "step": 93930
    },
    {
      "epoch": 38.45,
      "grad_norm": 1.7597767114639282,
      "learning_rate": 4.432877691267424e-06,
      "loss": 0.5478,
      "step": 93940
    },
    {
      "epoch": 38.46,
      "grad_norm": 2.3910930156707764,
      "learning_rate": 4.432805780685389e-06,
      "loss": 0.5565,
      "step": 93950
    },
    {
      "epoch": 38.46,
      "grad_norm": 2.692310094833374,
      "learning_rate": 4.4327338605337124e-06,
      "loss": 0.5724,
      "step": 93960
    },
    {
      "epoch": 38.47,
      "grad_norm": 2.3378002643585205,
      "learning_rate": 4.4326619308127236e-06,
      "loss": 0.5573,
      "step": 93970
    },
    {
      "epoch": 38.47,
      "grad_norm": 2.5010552406311035,
      "learning_rate": 4.432589991522753e-06,
      "loss": 0.5473,
      "step": 93980
    },
    {
      "epoch": 38.47,
      "grad_norm": 2.657499313354492,
      "learning_rate": 4.432518042664131e-06,
      "loss": 0.5684,
      "step": 93990
    },
    {
      "epoch": 38.48,
      "grad_norm": 2.4227466583251953,
      "learning_rate": 4.4324460842371845e-06,
      "loss": 0.5656,
      "step": 94000
    },
    {
      "epoch": 38.48,
      "grad_norm": 2.416016101837158,
      "learning_rate": 4.432374116242245e-06,
      "loss": 0.5587,
      "step": 94010
    },
    {
      "epoch": 38.49,
      "grad_norm": 2.7652175426483154,
      "learning_rate": 4.432302138679643e-06,
      "loss": 0.5396,
      "step": 94020
    },
    {
      "epoch": 38.49,
      "grad_norm": 2.231172561645508,
      "learning_rate": 4.432230151549706e-06,
      "loss": 0.5425,
      "step": 94030
    },
    {
      "epoch": 38.49,
      "grad_norm": 2.466487407684326,
      "learning_rate": 4.432158154852766e-06,
      "loss": 0.5482,
      "step": 94040
    },
    {
      "epoch": 38.5,
      "grad_norm": 1.8204270601272583,
      "learning_rate": 4.43208614858915e-06,
      "loss": 0.5485,
      "step": 94050
    },
    {
      "epoch": 38.5,
      "grad_norm": 2.7044107913970947,
      "learning_rate": 4.432014132759191e-06,
      "loss": 0.5534,
      "step": 94060
    },
    {
      "epoch": 38.51,
      "grad_norm": 1.6854511499404907,
      "learning_rate": 4.431942107363217e-06,
      "loss": 0.5573,
      "step": 94070
    },
    {
      "epoch": 38.51,
      "grad_norm": 1.970005750656128,
      "learning_rate": 4.431870072401558e-06,
      "loss": 0.5844,
      "step": 94080
    },
    {
      "epoch": 38.51,
      "grad_norm": 2.1260788440704346,
      "learning_rate": 4.431798027874545e-06,
      "loss": 0.566,
      "step": 94090
    },
    {
      "epoch": 38.52,
      "grad_norm": 2.3330750465393066,
      "learning_rate": 4.431725973782506e-06,
      "loss": 0.5801,
      "step": 94100
    },
    {
      "epoch": 38.52,
      "grad_norm": 2.091885805130005,
      "learning_rate": 4.4316539101257735e-06,
      "loss": 0.5869,
      "step": 94110
    },
    {
      "epoch": 38.53,
      "grad_norm": 2.5873830318450928,
      "learning_rate": 4.431581836904678e-06,
      "loss": 0.5596,
      "step": 94120
    },
    {
      "epoch": 38.53,
      "grad_norm": 2.8718435764312744,
      "learning_rate": 4.4315097541195454e-06,
      "loss": 0.5529,
      "step": 94130
    },
    {
      "epoch": 38.53,
      "grad_norm": 2.7324817180633545,
      "learning_rate": 4.43143766177071e-06,
      "loss": 0.5824,
      "step": 94140
    },
    {
      "epoch": 38.54,
      "grad_norm": 3.878920316696167,
      "learning_rate": 4.431365559858501e-06,
      "loss": 0.544,
      "step": 94150
    },
    {
      "epoch": 38.54,
      "grad_norm": 2.040640354156494,
      "learning_rate": 4.431293448383247e-06,
      "loss": 0.5802,
      "step": 94160
    },
    {
      "epoch": 38.55,
      "grad_norm": 2.006608486175537,
      "learning_rate": 4.431221327345281e-06,
      "loss": 0.5554,
      "step": 94170
    },
    {
      "epoch": 38.55,
      "grad_norm": 2.60903000831604,
      "learning_rate": 4.431149196744932e-06,
      "loss": 0.5591,
      "step": 94180
    },
    {
      "epoch": 38.56,
      "grad_norm": 2.489908218383789,
      "learning_rate": 4.43107705658253e-06,
      "loss": 0.562,
      "step": 94190
    },
    {
      "epoch": 38.56,
      "grad_norm": 3.3839480876922607,
      "learning_rate": 4.431004906858407e-06,
      "loss": 0.5682,
      "step": 94200
    },
    {
      "epoch": 38.56,
      "grad_norm": 1.7880808115005493,
      "learning_rate": 4.430932747572893e-06,
      "loss": 0.5575,
      "step": 94210
    },
    {
      "epoch": 38.57,
      "grad_norm": 1.9153209924697876,
      "learning_rate": 4.4308605787263165e-06,
      "loss": 0.5768,
      "step": 94220
    },
    {
      "epoch": 38.57,
      "grad_norm": 2.357100248336792,
      "learning_rate": 4.430788400319011e-06,
      "loss": 0.5421,
      "step": 94230
    },
    {
      "epoch": 38.58,
      "grad_norm": 3.033226490020752,
      "learning_rate": 4.430716212351304e-06,
      "loss": 0.5463,
      "step": 94240
    },
    {
      "epoch": 38.58,
      "grad_norm": 2.197927951812744,
      "learning_rate": 4.430644014823529e-06,
      "loss": 0.5732,
      "step": 94250
    },
    {
      "epoch": 38.58,
      "grad_norm": 2.068089485168457,
      "learning_rate": 4.430571807736016e-06,
      "loss": 0.5743,
      "step": 94260
    },
    {
      "epoch": 38.59,
      "grad_norm": 2.364203929901123,
      "learning_rate": 4.430499591089095e-06,
      "loss": 0.5794,
      "step": 94270
    },
    {
      "epoch": 38.59,
      "grad_norm": 1.9388329982757568,
      "learning_rate": 4.4304273648830985e-06,
      "loss": 0.5517,
      "step": 94280
    },
    {
      "epoch": 38.6,
      "grad_norm": 2.1645498275756836,
      "learning_rate": 4.430355129118355e-06,
      "loss": 0.5743,
      "step": 94290
    },
    {
      "epoch": 38.6,
      "grad_norm": 2.1268692016601562,
      "learning_rate": 4.430282883795197e-06,
      "loss": 0.5565,
      "step": 94300
    },
    {
      "epoch": 38.6,
      "grad_norm": 2.5291616916656494,
      "learning_rate": 4.430210628913954e-06,
      "loss": 0.5518,
      "step": 94310
    },
    {
      "epoch": 38.61,
      "grad_norm": 2.053834915161133,
      "learning_rate": 4.430138364474959e-06,
      "loss": 0.5747,
      "step": 94320
    },
    {
      "epoch": 38.61,
      "grad_norm": 2.1155507564544678,
      "learning_rate": 4.430066090478542e-06,
      "loss": 0.5572,
      "step": 94330
    },
    {
      "epoch": 38.62,
      "grad_norm": 2.1081316471099854,
      "learning_rate": 4.429993806925033e-06,
      "loss": 0.5533,
      "step": 94340
    },
    {
      "epoch": 38.62,
      "grad_norm": 2.1223878860473633,
      "learning_rate": 4.429921513814765e-06,
      "loss": 0.5315,
      "step": 94350
    },
    {
      "epoch": 38.62,
      "grad_norm": 2.7545039653778076,
      "learning_rate": 4.429849211148069e-06,
      "loss": 0.5653,
      "step": 94360
    },
    {
      "epoch": 38.63,
      "grad_norm": 2.0076465606689453,
      "learning_rate": 4.429776898925275e-06,
      "loss": 0.5429,
      "step": 94370
    },
    {
      "epoch": 38.63,
      "grad_norm": 2.675330877304077,
      "learning_rate": 4.4297045771467155e-06,
      "loss": 0.5541,
      "step": 94380
    },
    {
      "epoch": 38.64,
      "grad_norm": 2.3889002799987793,
      "learning_rate": 4.42963224581272e-06,
      "loss": 0.55,
      "step": 94390
    },
    {
      "epoch": 38.64,
      "grad_norm": 3.0011796951293945,
      "learning_rate": 4.429559904923622e-06,
      "loss": 0.566,
      "step": 94400
    },
    {
      "epoch": 38.65,
      "grad_norm": 2.0455029010772705,
      "learning_rate": 4.429487554479751e-06,
      "loss": 0.559,
      "step": 94410
    },
    {
      "epoch": 38.65,
      "grad_norm": 1.7263433933258057,
      "learning_rate": 4.42941519448144e-06,
      "loss": 0.5623,
      "step": 94420
    },
    {
      "epoch": 38.65,
      "grad_norm": 2.0100111961364746,
      "learning_rate": 4.429342824929019e-06,
      "loss": 0.5568,
      "step": 94430
    },
    {
      "epoch": 38.66,
      "grad_norm": 2.3170242309570312,
      "learning_rate": 4.429270445822821e-06,
      "loss": 0.5374,
      "step": 94440
    },
    {
      "epoch": 38.66,
      "grad_norm": 2.447873592376709,
      "learning_rate": 4.429198057163177e-06,
      "loss": 0.5679,
      "step": 94450
    },
    {
      "epoch": 38.67,
      "grad_norm": 2.1204824447631836,
      "learning_rate": 4.429125658950418e-06,
      "loss": 0.5596,
      "step": 94460
    },
    {
      "epoch": 38.67,
      "grad_norm": 1.6322040557861328,
      "learning_rate": 4.4290532511848755e-06,
      "loss": 0.5497,
      "step": 94470
    },
    {
      "epoch": 38.67,
      "grad_norm": 2.7074480056762695,
      "learning_rate": 4.428980833866882e-06,
      "loss": 0.5631,
      "step": 94480
    },
    {
      "epoch": 38.68,
      "grad_norm": 2.2039501667022705,
      "learning_rate": 4.428908406996769e-06,
      "loss": 0.542,
      "step": 94490
    },
    {
      "epoch": 38.68,
      "grad_norm": 2.1103525161743164,
      "learning_rate": 4.428835970574869e-06,
      "loss": 0.5605,
      "step": 94500
    },
    {
      "epoch": 38.69,
      "grad_norm": 2.6743478775024414,
      "learning_rate": 4.428763524601513e-06,
      "loss": 0.5739,
      "step": 94510
    },
    {
      "epoch": 38.69,
      "grad_norm": 2.5297067165374756,
      "learning_rate": 4.428691069077032e-06,
      "loss": 0.5468,
      "step": 94520
    },
    {
      "epoch": 38.69,
      "grad_norm": 2.1176440715789795,
      "learning_rate": 4.42861860400176e-06,
      "loss": 0.5442,
      "step": 94530
    },
    {
      "epoch": 38.7,
      "grad_norm": 1.8541535139083862,
      "learning_rate": 4.428546129376027e-06,
      "loss": 0.5678,
      "step": 94540
    },
    {
      "epoch": 38.7,
      "grad_norm": 2.2714176177978516,
      "learning_rate": 4.428473645200165e-06,
      "loss": 0.5448,
      "step": 94550
    },
    {
      "epoch": 38.71,
      "grad_norm": 2.792362689971924,
      "learning_rate": 4.428401151474508e-06,
      "loss": 0.546,
      "step": 94560
    },
    {
      "epoch": 38.71,
      "grad_norm": 2.1693496704101562,
      "learning_rate": 4.4283286481993875e-06,
      "loss": 0.5337,
      "step": 94570
    },
    {
      "epoch": 38.71,
      "grad_norm": 2.3530802726745605,
      "learning_rate": 4.428256135375134e-06,
      "loss": 0.5462,
      "step": 94580
    },
    {
      "epoch": 38.72,
      "grad_norm": 2.7005012035369873,
      "learning_rate": 4.428183613002081e-06,
      "loss": 0.5754,
      "step": 94590
    },
    {
      "epoch": 38.72,
      "grad_norm": 2.4980571269989014,
      "learning_rate": 4.42811108108056e-06,
      "loss": 0.5552,
      "step": 94600
    },
    {
      "epoch": 38.73,
      "grad_norm": 2.159224510192871,
      "learning_rate": 4.428038539610905e-06,
      "loss": 0.5667,
      "step": 94610
    },
    {
      "epoch": 38.73,
      "grad_norm": 1.74956476688385,
      "learning_rate": 4.427965988593445e-06,
      "loss": 0.5544,
      "step": 94620
    },
    {
      "epoch": 38.74,
      "grad_norm": 1.9866652488708496,
      "learning_rate": 4.427893428028516e-06,
      "loss": 0.5705,
      "step": 94630
    },
    {
      "epoch": 38.74,
      "grad_norm": 1.9912914037704468,
      "learning_rate": 4.427820857916448e-06,
      "loss": 0.5544,
      "step": 94640
    },
    {
      "epoch": 38.74,
      "grad_norm": 2.1197900772094727,
      "learning_rate": 4.427748278257574e-06,
      "loss": 0.5844,
      "step": 94650
    },
    {
      "epoch": 38.75,
      "grad_norm": 2.015155792236328,
      "learning_rate": 4.427675689052227e-06,
      "loss": 0.5728,
      "step": 94660
    },
    {
      "epoch": 38.75,
      "grad_norm": 2.184950590133667,
      "learning_rate": 4.42760309030074e-06,
      "loss": 0.5618,
      "step": 94670
    },
    {
      "epoch": 38.76,
      "grad_norm": 2.281992197036743,
      "learning_rate": 4.427530482003443e-06,
      "loss": 0.542,
      "step": 94680
    },
    {
      "epoch": 38.76,
      "grad_norm": 2.634927988052368,
      "learning_rate": 4.427457864160672e-06,
      "loss": 0.5621,
      "step": 94690
    },
    {
      "epoch": 38.76,
      "grad_norm": 2.5438222885131836,
      "learning_rate": 4.427385236772757e-06,
      "loss": 0.5567,
      "step": 94700
    },
    {
      "epoch": 38.77,
      "grad_norm": 2.034604072570801,
      "learning_rate": 4.427312599840032e-06,
      "loss": 0.5662,
      "step": 94710
    },
    {
      "epoch": 38.77,
      "grad_norm": 2.7045063972473145,
      "learning_rate": 4.427239953362829e-06,
      "loss": 0.5473,
      "step": 94720
    },
    {
      "epoch": 38.78,
      "grad_norm": 1.8837864398956299,
      "learning_rate": 4.427167297341481e-06,
      "loss": 0.548,
      "step": 94730
    },
    {
      "epoch": 38.78,
      "grad_norm": 2.6130378246307373,
      "learning_rate": 4.427094631776322e-06,
      "loss": 0.5638,
      "step": 94740
    },
    {
      "epoch": 38.78,
      "grad_norm": 2.2611348628997803,
      "learning_rate": 4.427021956667684e-06,
      "loss": 0.5674,
      "step": 94750
    },
    {
      "epoch": 38.79,
      "grad_norm": 2.1545848846435547,
      "learning_rate": 4.426949272015898e-06,
      "loss": 0.5681,
      "step": 94760
    },
    {
      "epoch": 38.79,
      "grad_norm": 2.0427913665771484,
      "learning_rate": 4.4268765778213e-06,
      "loss": 0.5665,
      "step": 94770
    },
    {
      "epoch": 38.8,
      "grad_norm": 2.4768683910369873,
      "learning_rate": 4.426803874084222e-06,
      "loss": 0.545,
      "step": 94780
    },
    {
      "epoch": 38.8,
      "grad_norm": 1.8982410430908203,
      "learning_rate": 4.426731160804996e-06,
      "loss": 0.5474,
      "step": 94790
    },
    {
      "epoch": 38.8,
      "grad_norm": 1.768310546875,
      "learning_rate": 4.426658437983957e-06,
      "loss": 0.5675,
      "step": 94800
    },
    {
      "epoch": 38.81,
      "grad_norm": 2.148061513900757,
      "learning_rate": 4.426585705621436e-06,
      "loss": 0.5547,
      "step": 94810
    },
    {
      "epoch": 38.81,
      "grad_norm": 2.4398093223571777,
      "learning_rate": 4.426512963717768e-06,
      "loss": 0.5621,
      "step": 94820
    },
    {
      "epoch": 38.82,
      "grad_norm": 2.302076578140259,
      "learning_rate": 4.4264402122732844e-06,
      "loss": 0.5444,
      "step": 94830
    },
    {
      "epoch": 38.82,
      "grad_norm": 2.7480523586273193,
      "learning_rate": 4.42636745128832e-06,
      "loss": 0.547,
      "step": 94840
    },
    {
      "epoch": 38.83,
      "grad_norm": 2.2518503665924072,
      "learning_rate": 4.4262946807632085e-06,
      "loss": 0.5515,
      "step": 94850
    },
    {
      "epoch": 38.83,
      "grad_norm": 2.027383804321289,
      "learning_rate": 4.4262219006982816e-06,
      "loss": 0.5533,
      "step": 94860
    },
    {
      "epoch": 38.83,
      "grad_norm": 1.6705577373504639,
      "learning_rate": 4.426149111093873e-06,
      "loss": 0.5766,
      "step": 94870
    },
    {
      "epoch": 38.84,
      "grad_norm": 1.8778526782989502,
      "learning_rate": 4.426076311950317e-06,
      "loss": 0.5624,
      "step": 94880
    },
    {
      "epoch": 38.84,
      "grad_norm": 3.375864267349243,
      "learning_rate": 4.426003503267947e-06,
      "loss": 0.567,
      "step": 94890
    },
    {
      "epoch": 38.85,
      "grad_norm": 1.5329320430755615,
      "learning_rate": 4.4259306850470955e-06,
      "loss": 0.5694,
      "step": 94900
    },
    {
      "epoch": 38.85,
      "grad_norm": 2.3671517372131348,
      "learning_rate": 4.425857857288096e-06,
      "loss": 0.5498,
      "step": 94910
    },
    {
      "epoch": 38.85,
      "grad_norm": 1.695013403892517,
      "learning_rate": 4.425785019991285e-06,
      "loss": 0.5591,
      "step": 94920
    },
    {
      "epoch": 38.86,
      "grad_norm": 2.5226986408233643,
      "learning_rate": 4.425712173156993e-06,
      "loss": 0.5536,
      "step": 94930
    },
    {
      "epoch": 38.86,
      "grad_norm": 2.151693820953369,
      "learning_rate": 4.425639316785554e-06,
      "loss": 0.5489,
      "step": 94940
    },
    {
      "epoch": 38.87,
      "grad_norm": 2.938695192337036,
      "learning_rate": 4.425566450877303e-06,
      "loss": 0.5513,
      "step": 94950
    },
    {
      "epoch": 38.87,
      "grad_norm": 2.575256824493408,
      "learning_rate": 4.425493575432573e-06,
      "loss": 0.5488,
      "step": 94960
    },
    {
      "epoch": 38.87,
      "grad_norm": 1.8461635112762451,
      "learning_rate": 4.425420690451699e-06,
      "loss": 0.5539,
      "step": 94970
    },
    {
      "epoch": 38.88,
      "grad_norm": 2.481416702270508,
      "learning_rate": 4.425347795935014e-06,
      "loss": 0.5419,
      "step": 94980
    },
    {
      "epoch": 38.88,
      "grad_norm": 1.9846407175064087,
      "learning_rate": 4.425274891882851e-06,
      "loss": 0.5458,
      "step": 94990
    },
    {
      "epoch": 38.89,
      "grad_norm": 2.531930923461914,
      "learning_rate": 4.425201978295544e-06,
      "loss": 0.5557,
      "step": 95000
    },
    {
      "epoch": 38.89,
      "grad_norm": 2.2905797958374023,
      "learning_rate": 4.425129055173429e-06,
      "loss": 0.5488,
      "step": 95010
    },
    {
      "epoch": 38.89,
      "grad_norm": 1.6976946592330933,
      "learning_rate": 4.425056122516839e-06,
      "loss": 0.5809,
      "step": 95020
    },
    {
      "epoch": 38.9,
      "grad_norm": 1.8869117498397827,
      "learning_rate": 4.424983180326108e-06,
      "loss": 0.5429,
      "step": 95030
    },
    {
      "epoch": 38.9,
      "grad_norm": 2.3412957191467285,
      "learning_rate": 4.42491022860157e-06,
      "loss": 0.5458,
      "step": 95040
    },
    {
      "epoch": 38.91,
      "grad_norm": 1.523501992225647,
      "learning_rate": 4.424837267343559e-06,
      "loss": 0.5503,
      "step": 95050
    },
    {
      "epoch": 38.91,
      "grad_norm": 2.0311825275421143,
      "learning_rate": 4.42476429655241e-06,
      "loss": 0.5503,
      "step": 95060
    },
    {
      "epoch": 38.92,
      "grad_norm": 1.9379551410675049,
      "learning_rate": 4.424691316228456e-06,
      "loss": 0.5634,
      "step": 95070
    },
    {
      "epoch": 38.92,
      "grad_norm": 2.9020907878875732,
      "learning_rate": 4.4246183263720335e-06,
      "loss": 0.5413,
      "step": 95080
    },
    {
      "epoch": 38.92,
      "grad_norm": 2.9578425884246826,
      "learning_rate": 4.424545326983474e-06,
      "loss": 0.5602,
      "step": 95090
    },
    {
      "epoch": 38.93,
      "grad_norm": 3.51973819732666,
      "learning_rate": 4.424472318063114e-06,
      "loss": 0.5514,
      "step": 95100
    },
    {
      "epoch": 38.93,
      "grad_norm": 1.890714406967163,
      "learning_rate": 4.424399299611287e-06,
      "loss": 0.5319,
      "step": 95110
    },
    {
      "epoch": 38.94,
      "grad_norm": 2.9808502197265625,
      "learning_rate": 4.4243262716283285e-06,
      "loss": 0.5514,
      "step": 95120
    },
    {
      "epoch": 38.94,
      "grad_norm": 1.9054161310195923,
      "learning_rate": 4.424253234114572e-06,
      "loss": 0.5887,
      "step": 95130
    },
    {
      "epoch": 38.94,
      "grad_norm": 1.9635941982269287,
      "learning_rate": 4.424180187070352e-06,
      "loss": 0.5556,
      "step": 95140
    },
    {
      "epoch": 38.95,
      "grad_norm": 1.5896626710891724,
      "learning_rate": 4.424107130496004e-06,
      "loss": 0.5607,
      "step": 95150
    },
    {
      "epoch": 38.95,
      "grad_norm": 1.9612981081008911,
      "learning_rate": 4.424034064391861e-06,
      "loss": 0.5684,
      "step": 95160
    },
    {
      "epoch": 38.96,
      "grad_norm": 2.852541208267212,
      "learning_rate": 4.42396098875826e-06,
      "loss": 0.5701,
      "step": 95170
    },
    {
      "epoch": 38.96,
      "grad_norm": 2.3321049213409424,
      "learning_rate": 4.423887903595535e-06,
      "loss": 0.5545,
      "step": 95180
    },
    {
      "epoch": 38.96,
      "grad_norm": 1.8141369819641113,
      "learning_rate": 4.42381480890402e-06,
      "loss": 0.549,
      "step": 95190
    },
    {
      "epoch": 38.97,
      "grad_norm": 2.8987221717834473,
      "learning_rate": 4.42374170468405e-06,
      "loss": 0.5619,
      "step": 95200
    },
    {
      "epoch": 38.97,
      "grad_norm": 1.7724915742874146,
      "learning_rate": 4.42366859093596e-06,
      "loss": 0.5716,
      "step": 95210
    },
    {
      "epoch": 38.98,
      "grad_norm": 2.1956613063812256,
      "learning_rate": 4.423595467660085e-06,
      "loss": 0.5744,
      "step": 95220
    },
    {
      "epoch": 38.98,
      "grad_norm": 2.663029193878174,
      "learning_rate": 4.42352233485676e-06,
      "loss": 0.5773,
      "step": 95230
    },
    {
      "epoch": 38.98,
      "grad_norm": 2.3169403076171875,
      "learning_rate": 4.4234491925263204e-06,
      "loss": 0.5665,
      "step": 95240
    },
    {
      "epoch": 38.99,
      "grad_norm": 2.4619126319885254,
      "learning_rate": 4.423376040669101e-06,
      "loss": 0.574,
      "step": 95250
    },
    {
      "epoch": 38.99,
      "grad_norm": 2.700037956237793,
      "learning_rate": 4.423302879285437e-06,
      "loss": 0.5676,
      "step": 95260
    },
    {
      "epoch": 39.0,
      "grad_norm": 2.07149600982666,
      "learning_rate": 4.4232297083756625e-06,
      "loss": 0.5386,
      "step": 95270
    },
    {
      "epoch": 39.0,
      "eval_loss": 0.5617032647132874,
      "eval_runtime": 51.8423,
      "eval_samples_per_second": 66.529,
      "eval_steps_per_second": 8.333,
      "step": 95277
    },
    {
      "epoch": 39.0,
      "grad_norm": 2.264923095703125,
      "learning_rate": 4.423156527940114e-06,
      "loss": 0.5752,
      "step": 95280
    },
    {
      "epoch": 39.01,
      "grad_norm": 2.2622909545898438,
      "learning_rate": 4.423083337979126e-06,
      "loss": 0.5633,
      "step": 95290
    },
    {
      "epoch": 39.01,
      "grad_norm": 1.7359551191329956,
      "learning_rate": 4.423010138493034e-06,
      "loss": 0.5696,
      "step": 95300
    },
    {
      "epoch": 39.01,
      "grad_norm": 2.8561298847198486,
      "learning_rate": 4.4229369294821736e-06,
      "loss": 0.5456,
      "step": 95310
    },
    {
      "epoch": 39.02,
      "grad_norm": 2.5543975830078125,
      "learning_rate": 4.42286371094688e-06,
      "loss": 0.5761,
      "step": 95320
    },
    {
      "epoch": 39.02,
      "grad_norm": 2.003695011138916,
      "learning_rate": 4.4227904828874875e-06,
      "loss": 0.5426,
      "step": 95330
    },
    {
      "epoch": 39.03,
      "grad_norm": 1.694057583808899,
      "learning_rate": 4.422717245304334e-06,
      "loss": 0.5699,
      "step": 95340
    },
    {
      "epoch": 39.03,
      "grad_norm": 2.596752882003784,
      "learning_rate": 4.422643998197752e-06,
      "loss": 0.5577,
      "step": 95350
    },
    {
      "epoch": 39.03,
      "grad_norm": 1.85471510887146,
      "learning_rate": 4.42257074156808e-06,
      "loss": 0.5498,
      "step": 95360
    },
    {
      "epoch": 39.04,
      "grad_norm": 1.444670557975769,
      "learning_rate": 4.422497475415651e-06,
      "loss": 0.5573,
      "step": 95370
    },
    {
      "epoch": 39.04,
      "grad_norm": 1.5373903512954712,
      "learning_rate": 4.422424199740802e-06,
      "loss": 0.556,
      "step": 95380
    },
    {
      "epoch": 39.05,
      "grad_norm": 2.506535291671753,
      "learning_rate": 4.422350914543869e-06,
      "loss": 0.5434,
      "step": 95390
    },
    {
      "epoch": 39.05,
      "grad_norm": 2.6698379516601562,
      "learning_rate": 4.4222776198251865e-06,
      "loss": 0.5385,
      "step": 95400
    },
    {
      "epoch": 39.05,
      "grad_norm": 2.939042329788208,
      "learning_rate": 4.422204315585091e-06,
      "loss": 0.5408,
      "step": 95410
    },
    {
      "epoch": 39.06,
      "grad_norm": 2.5508317947387695,
      "learning_rate": 4.422131001823919e-06,
      "loss": 0.5717,
      "step": 95420
    },
    {
      "epoch": 39.06,
      "grad_norm": 2.0563666820526123,
      "learning_rate": 4.422057678542004e-06,
      "loss": 0.5492,
      "step": 95430
    },
    {
      "epoch": 39.07,
      "grad_norm": 2.0999746322631836,
      "learning_rate": 4.421984345739685e-06,
      "loss": 0.5432,
      "step": 95440
    },
    {
      "epoch": 39.07,
      "grad_norm": 2.4124388694763184,
      "learning_rate": 4.421911003417294e-06,
      "loss": 0.5483,
      "step": 95450
    },
    {
      "epoch": 39.07,
      "grad_norm": 2.120377779006958,
      "learning_rate": 4.421837651575171e-06,
      "loss": 0.5644,
      "step": 95460
    },
    {
      "epoch": 39.08,
      "grad_norm": 2.373776912689209,
      "learning_rate": 4.42176429021365e-06,
      "loss": 0.5643,
      "step": 95470
    },
    {
      "epoch": 39.08,
      "grad_norm": 2.325448751449585,
      "learning_rate": 4.421690919333068e-06,
      "loss": 0.5632,
      "step": 95480
    },
    {
      "epoch": 39.09,
      "grad_norm": 2.0436573028564453,
      "learning_rate": 4.42161753893376e-06,
      "loss": 0.5436,
      "step": 95490
    },
    {
      "epoch": 39.09,
      "grad_norm": 2.279355525970459,
      "learning_rate": 4.421544149016062e-06,
      "loss": 0.5505,
      "step": 95500
    },
    {
      "epoch": 39.1,
      "grad_norm": 2.524021863937378,
      "learning_rate": 4.421470749580311e-06,
      "loss": 0.5562,
      "step": 95510
    },
    {
      "epoch": 39.1,
      "grad_norm": 2.0332834720611572,
      "learning_rate": 4.421397340626844e-06,
      "loss": 0.572,
      "step": 95520
    },
    {
      "epoch": 39.1,
      "grad_norm": 3.2993414402008057,
      "learning_rate": 4.421323922155995e-06,
      "loss": 0.5488,
      "step": 95530
    },
    {
      "epoch": 39.11,
      "grad_norm": 2.2387595176696777,
      "learning_rate": 4.421250494168103e-06,
      "loss": 0.5876,
      "step": 95540
    },
    {
      "epoch": 39.11,
      "grad_norm": 2.3587985038757324,
      "learning_rate": 4.421177056663502e-06,
      "loss": 0.5876,
      "step": 95550
    },
    {
      "epoch": 39.12,
      "grad_norm": 2.5469727516174316,
      "learning_rate": 4.421103609642529e-06,
      "loss": 0.5476,
      "step": 95560
    },
    {
      "epoch": 39.12,
      "grad_norm": 2.6880171298980713,
      "learning_rate": 4.421030153105522e-06,
      "loss": 0.5596,
      "step": 95570
    },
    {
      "epoch": 39.12,
      "grad_norm": 2.4604854583740234,
      "learning_rate": 4.420956687052815e-06,
      "loss": 0.5513,
      "step": 95580
    },
    {
      "epoch": 39.13,
      "grad_norm": 1.9691805839538574,
      "learning_rate": 4.4208832114847464e-06,
      "loss": 0.5524,
      "step": 95590
    },
    {
      "epoch": 39.13,
      "grad_norm": 2.2929582595825195,
      "learning_rate": 4.420809726401653e-06,
      "loss": 0.5547,
      "step": 95600
    },
    {
      "epoch": 39.14,
      "grad_norm": 1.8850692510604858,
      "learning_rate": 4.42073623180387e-06,
      "loss": 0.5405,
      "step": 95610
    },
    {
      "epoch": 39.14,
      "grad_norm": 2.2652220726013184,
      "learning_rate": 4.420662727691735e-06,
      "loss": 0.5686,
      "step": 95620
    },
    {
      "epoch": 39.14,
      "grad_norm": 1.7362018823623657,
      "learning_rate": 4.420589214065584e-06,
      "loss": 0.5618,
      "step": 95630
    },
    {
      "epoch": 39.15,
      "grad_norm": 2.2066872119903564,
      "learning_rate": 4.420515690925755e-06,
      "loss": 0.5863,
      "step": 95640
    },
    {
      "epoch": 39.15,
      "grad_norm": 1.9725528955459595,
      "learning_rate": 4.420442158272583e-06,
      "loss": 0.5405,
      "step": 95650
    },
    {
      "epoch": 39.16,
      "grad_norm": 2.2282822132110596,
      "learning_rate": 4.4203686161064065e-06,
      "loss": 0.5415,
      "step": 95660
    },
    {
      "epoch": 39.16,
      "grad_norm": 2.5499773025512695,
      "learning_rate": 4.420295064427562e-06,
      "loss": 0.5624,
      "step": 95670
    },
    {
      "epoch": 39.16,
      "grad_norm": 2.1957454681396484,
      "learning_rate": 4.420221503236385e-06,
      "loss": 0.5702,
      "step": 95680
    },
    {
      "epoch": 39.17,
      "grad_norm": 2.787978410720825,
      "learning_rate": 4.420147932533214e-06,
      "loss": 0.5627,
      "step": 95690
    },
    {
      "epoch": 39.17,
      "grad_norm": 2.390576124191284,
      "learning_rate": 4.4200743523183865e-06,
      "loss": 0.5577,
      "step": 95700
    },
    {
      "epoch": 39.18,
      "grad_norm": 1.8302830457687378,
      "learning_rate": 4.420000762592238e-06,
      "loss": 0.5603,
      "step": 95710
    },
    {
      "epoch": 39.18,
      "grad_norm": 2.1923630237579346,
      "learning_rate": 4.419927163355106e-06,
      "loss": 0.5643,
      "step": 95720
    },
    {
      "epoch": 39.19,
      "grad_norm": 2.8879201412200928,
      "learning_rate": 4.419853554607329e-06,
      "loss": 0.566,
      "step": 95730
    },
    {
      "epoch": 39.19,
      "grad_norm": 2.5688846111297607,
      "learning_rate": 4.419779936349242e-06,
      "loss": 0.5488,
      "step": 95740
    },
    {
      "epoch": 39.19,
      "grad_norm": 2.4150121212005615,
      "learning_rate": 4.419706308581184e-06,
      "loss": 0.5641,
      "step": 95750
    },
    {
      "epoch": 39.2,
      "grad_norm": 2.49421763420105,
      "learning_rate": 4.419632671303492e-06,
      "loss": 0.5583,
      "step": 95760
    },
    {
      "epoch": 39.2,
      "grad_norm": 1.8771793842315674,
      "learning_rate": 4.4195590245165025e-06,
      "loss": 0.5486,
      "step": 95770
    },
    {
      "epoch": 39.21,
      "grad_norm": 1.9491519927978516,
      "learning_rate": 4.419485368220553e-06,
      "loss": 0.5658,
      "step": 95780
    },
    {
      "epoch": 39.21,
      "grad_norm": 2.2905893325805664,
      "learning_rate": 4.419411702415982e-06,
      "loss": 0.5721,
      "step": 95790
    },
    {
      "epoch": 39.21,
      "grad_norm": 1.783009648323059,
      "learning_rate": 4.419338027103126e-06,
      "loss": 0.5583,
      "step": 95800
    },
    {
      "epoch": 39.22,
      "grad_norm": 3.1436195373535156,
      "learning_rate": 4.419264342282323e-06,
      "loss": 0.5471,
      "step": 95810
    },
    {
      "epoch": 39.22,
      "grad_norm": 2.251999855041504,
      "learning_rate": 4.419190647953909e-06,
      "loss": 0.5521,
      "step": 95820
    },
    {
      "epoch": 39.23,
      "grad_norm": 3.107415199279785,
      "learning_rate": 4.419116944118224e-06,
      "loss": 0.5618,
      "step": 95830
    },
    {
      "epoch": 39.23,
      "grad_norm": 1.7524596452713013,
      "learning_rate": 4.419043230775603e-06,
      "loss": 0.5424,
      "step": 95840
    },
    {
      "epoch": 39.23,
      "grad_norm": 2.0459144115448,
      "learning_rate": 4.418969507926386e-06,
      "loss": 0.5622,
      "step": 95850
    },
    {
      "epoch": 39.24,
      "grad_norm": 2.8554773330688477,
      "learning_rate": 4.41889577557091e-06,
      "loss": 0.5591,
      "step": 95860
    },
    {
      "epoch": 39.24,
      "grad_norm": 2.241609573364258,
      "learning_rate": 4.418822033709513e-06,
      "loss": 0.5449,
      "step": 95870
    },
    {
      "epoch": 39.25,
      "grad_norm": 2.453716516494751,
      "learning_rate": 4.418748282342532e-06,
      "loss": 0.5596,
      "step": 95880
    },
    {
      "epoch": 39.25,
      "grad_norm": 2.5948283672332764,
      "learning_rate": 4.418674521470305e-06,
      "loss": 0.5435,
      "step": 95890
    },
    {
      "epoch": 39.26,
      "grad_norm": 2.412696123123169,
      "learning_rate": 4.4186007510931705e-06,
      "loss": 0.5569,
      "step": 95900
    },
    {
      "epoch": 39.26,
      "grad_norm": 2.7102138996124268,
      "learning_rate": 4.418526971211466e-06,
      "loss": 0.5702,
      "step": 95910
    },
    {
      "epoch": 39.26,
      "grad_norm": 2.232455015182495,
      "learning_rate": 4.418453181825529e-06,
      "loss": 0.5648,
      "step": 95920
    },
    {
      "epoch": 39.27,
      "grad_norm": 2.825080394744873,
      "learning_rate": 4.4183793829356985e-06,
      "loss": 0.5568,
      "step": 95930
    },
    {
      "epoch": 39.27,
      "grad_norm": 2.177931308746338,
      "learning_rate": 4.418305574542312e-06,
      "loss": 0.5405,
      "step": 95940
    },
    {
      "epoch": 39.28,
      "grad_norm": 1.840105652809143,
      "learning_rate": 4.418231756645707e-06,
      "loss": 0.5339,
      "step": 95950
    },
    {
      "epoch": 39.28,
      "grad_norm": 1.9692432880401611,
      "learning_rate": 4.418157929246224e-06,
      "loss": 0.5446,
      "step": 95960
    },
    {
      "epoch": 39.28,
      "grad_norm": 2.111201763153076,
      "learning_rate": 4.418084092344198e-06,
      "loss": 0.5458,
      "step": 95970
    },
    {
      "epoch": 39.29,
      "grad_norm": 2.1094961166381836,
      "learning_rate": 4.418010245939969e-06,
      "loss": 0.5517,
      "step": 95980
    },
    {
      "epoch": 39.29,
      "grad_norm": 2.2949702739715576,
      "learning_rate": 4.417936390033876e-06,
      "loss": 0.5537,
      "step": 95990
    },
    {
      "epoch": 39.3,
      "grad_norm": 1.9410566091537476,
      "learning_rate": 4.417862524626255e-06,
      "loss": 0.5633,
      "step": 96000
    },
    {
      "epoch": 39.3,
      "grad_norm": 1.850609540939331,
      "learning_rate": 4.417788649717446e-06,
      "loss": 0.5532,
      "step": 96010
    },
    {
      "epoch": 39.3,
      "grad_norm": 2.9480185508728027,
      "learning_rate": 4.417714765307788e-06,
      "loss": 0.5789,
      "step": 96020
    },
    {
      "epoch": 39.31,
      "grad_norm": 1.9403311014175415,
      "learning_rate": 4.417640871397618e-06,
      "loss": 0.5493,
      "step": 96030
    },
    {
      "epoch": 39.31,
      "grad_norm": 2.533531665802002,
      "learning_rate": 4.417566967987275e-06,
      "loss": 0.5362,
      "step": 96040
    },
    {
      "epoch": 39.32,
      "grad_norm": 2.278158664703369,
      "learning_rate": 4.417493055077099e-06,
      "loss": 0.5729,
      "step": 96050
    },
    {
      "epoch": 39.32,
      "grad_norm": 2.34187388420105,
      "learning_rate": 4.417419132667425e-06,
      "loss": 0.5324,
      "step": 96060
    },
    {
      "epoch": 39.32,
      "grad_norm": 2.2597756385803223,
      "learning_rate": 4.417345200758596e-06,
      "loss": 0.5458,
      "step": 96070
    },
    {
      "epoch": 39.33,
      "grad_norm": 2.382213592529297,
      "learning_rate": 4.417271259350947e-06,
      "loss": 0.5427,
      "step": 96080
    },
    {
      "epoch": 39.33,
      "grad_norm": 1.8719555139541626,
      "learning_rate": 4.417197308444819e-06,
      "loss": 0.5487,
      "step": 96090
    },
    {
      "epoch": 39.34,
      "grad_norm": 2.333094358444214,
      "learning_rate": 4.417123348040549e-06,
      "loss": 0.5717,
      "step": 96100
    },
    {
      "epoch": 39.34,
      "grad_norm": 2.584233283996582,
      "learning_rate": 4.4170493781384784e-06,
      "loss": 0.5512,
      "step": 96110
    },
    {
      "epoch": 39.35,
      "grad_norm": 2.0027549266815186,
      "learning_rate": 4.416975398738943e-06,
      "loss": 0.5415,
      "step": 96120
    },
    {
      "epoch": 39.35,
      "grad_norm": 1.8509154319763184,
      "learning_rate": 4.416901409842284e-06,
      "loss": 0.5535,
      "step": 96130
    },
    {
      "epoch": 39.35,
      "grad_norm": 2.7363216876983643,
      "learning_rate": 4.416827411448839e-06,
      "loss": 0.5681,
      "step": 96140
    },
    {
      "epoch": 39.36,
      "grad_norm": 2.0865323543548584,
      "learning_rate": 4.416753403558948e-06,
      "loss": 0.563,
      "step": 96150
    },
    {
      "epoch": 39.36,
      "grad_norm": 2.7031307220458984,
      "learning_rate": 4.416679386172949e-06,
      "loss": 0.5656,
      "step": 96160
    },
    {
      "epoch": 39.37,
      "grad_norm": 2.90325927734375,
      "learning_rate": 4.416605359291182e-06,
      "loss": 0.5698,
      "step": 96170
    },
    {
      "epoch": 39.37,
      "grad_norm": 1.8240277767181396,
      "learning_rate": 4.416531322913985e-06,
      "loss": 0.5519,
      "step": 96180
    },
    {
      "epoch": 39.37,
      "grad_norm": 1.8413751125335693,
      "learning_rate": 4.416457277041698e-06,
      "loss": 0.5752,
      "step": 96190
    },
    {
      "epoch": 39.38,
      "grad_norm": 1.3549633026123047,
      "learning_rate": 4.41638322167466e-06,
      "loss": 0.5617,
      "step": 96200
    },
    {
      "epoch": 39.38,
      "grad_norm": 2.295933246612549,
      "learning_rate": 4.41630915681321e-06,
      "loss": 0.5462,
      "step": 96210
    },
    {
      "epoch": 39.39,
      "grad_norm": 2.9375340938568115,
      "learning_rate": 4.416235082457688e-06,
      "loss": 0.5518,
      "step": 96220
    },
    {
      "epoch": 39.39,
      "grad_norm": 1.809618353843689,
      "learning_rate": 4.416160998608433e-06,
      "loss": 0.5404,
      "step": 96230
    },
    {
      "epoch": 39.39,
      "grad_norm": 2.6288952827453613,
      "learning_rate": 4.416086905265782e-06,
      "loss": 0.5557,
      "step": 96240
    },
    {
      "epoch": 39.4,
      "grad_norm": 2.07058048248291,
      "learning_rate": 4.416012802430079e-06,
      "loss": 0.5684,
      "step": 96250
    },
    {
      "epoch": 39.4,
      "grad_norm": 2.4732863903045654,
      "learning_rate": 4.41593869010166e-06,
      "loss": 0.5485,
      "step": 96260
    },
    {
      "epoch": 39.41,
      "grad_norm": 2.0896193981170654,
      "learning_rate": 4.415864568280866e-06,
      "loss": 0.5575,
      "step": 96270
    },
    {
      "epoch": 39.41,
      "grad_norm": 2.39009952545166,
      "learning_rate": 4.415790436968036e-06,
      "loss": 0.5552,
      "step": 96280
    },
    {
      "epoch": 39.41,
      "grad_norm": 2.124246597290039,
      "learning_rate": 4.4157162961635084e-06,
      "loss": 0.5528,
      "step": 96290
    },
    {
      "epoch": 39.42,
      "grad_norm": 1.7168645858764648,
      "learning_rate": 4.415642145867626e-06,
      "loss": 0.5427,
      "step": 96300
    },
    {
      "epoch": 39.42,
      "grad_norm": 2.7497849464416504,
      "learning_rate": 4.415567986080725e-06,
      "loss": 0.5616,
      "step": 96310
    },
    {
      "epoch": 39.43,
      "grad_norm": 1.8863165378570557,
      "learning_rate": 4.415493816803148e-06,
      "loss": 0.5377,
      "step": 96320
    },
    {
      "epoch": 39.43,
      "grad_norm": 3.019791841506958,
      "learning_rate": 4.415419638035233e-06,
      "loss": 0.5557,
      "step": 96330
    },
    {
      "epoch": 39.44,
      "grad_norm": 1.8300076723098755,
      "learning_rate": 4.4153454497773195e-06,
      "loss": 0.5688,
      "step": 96340
    },
    {
      "epoch": 39.44,
      "grad_norm": 1.9392242431640625,
      "learning_rate": 4.415271252029749e-06,
      "loss": 0.5656,
      "step": 96350
    },
    {
      "epoch": 39.44,
      "grad_norm": 3.004774332046509,
      "learning_rate": 4.41519704479286e-06,
      "loss": 0.5682,
      "step": 96360
    },
    {
      "epoch": 39.45,
      "grad_norm": 1.610075831413269,
      "learning_rate": 4.415122828066993e-06,
      "loss": 0.5706,
      "step": 96370
    },
    {
      "epoch": 39.45,
      "grad_norm": 2.74654221534729,
      "learning_rate": 4.4150486018524875e-06,
      "loss": 0.5707,
      "step": 96380
    },
    {
      "epoch": 39.46,
      "grad_norm": 2.8173186779022217,
      "learning_rate": 4.414974366149684e-06,
      "loss": 0.552,
      "step": 96390
    },
    {
      "epoch": 39.46,
      "grad_norm": 1.786303162574768,
      "learning_rate": 4.414900120958923e-06,
      "loss": 0.5457,
      "step": 96400
    },
    {
      "epoch": 39.46,
      "grad_norm": 1.7975064516067505,
      "learning_rate": 4.414825866280543e-06,
      "loss": 0.5542,
      "step": 96410
    },
    {
      "epoch": 39.47,
      "grad_norm": 1.9597491025924683,
      "learning_rate": 4.4147516021148865e-06,
      "loss": 0.5484,
      "step": 96420
    },
    {
      "epoch": 39.47,
      "grad_norm": 2.449479103088379,
      "learning_rate": 4.414677328462291e-06,
      "loss": 0.5513,
      "step": 96430
    },
    {
      "epoch": 39.48,
      "grad_norm": 2.379486322402954,
      "learning_rate": 4.4146030453231e-06,
      "loss": 0.5658,
      "step": 96440
    },
    {
      "epoch": 39.48,
      "grad_norm": 3.369645595550537,
      "learning_rate": 4.41452875269765e-06,
      "loss": 0.5568,
      "step": 96450
    },
    {
      "epoch": 39.48,
      "grad_norm": 2.528153657913208,
      "learning_rate": 4.4144544505862844e-06,
      "loss": 0.5594,
      "step": 96460
    },
    {
      "epoch": 39.49,
      "grad_norm": 2.174588203430176,
      "learning_rate": 4.414380138989342e-06,
      "loss": 0.5676,
      "step": 96470
    },
    {
      "epoch": 39.49,
      "grad_norm": 2.3399670124053955,
      "learning_rate": 4.4143058179071635e-06,
      "loss": 0.5542,
      "step": 96480
    },
    {
      "epoch": 39.5,
      "grad_norm": 2.3001246452331543,
      "learning_rate": 4.414231487340089e-06,
      "loss": 0.5567,
      "step": 96490
    },
    {
      "epoch": 39.5,
      "grad_norm": 1.9988497495651245,
      "learning_rate": 4.414157147288461e-06,
      "loss": 0.5566,
      "step": 96500
    },
    {
      "epoch": 39.5,
      "grad_norm": 2.473780632019043,
      "learning_rate": 4.4140827977526175e-06,
      "loss": 0.5577,
      "step": 96510
    },
    {
      "epoch": 39.51,
      "grad_norm": 2.036475658416748,
      "learning_rate": 4.4140084387329e-06,
      "loss": 0.5384,
      "step": 96520
    },
    {
      "epoch": 39.51,
      "grad_norm": 1.5005289316177368,
      "learning_rate": 4.41393407022965e-06,
      "loss": 0.5606,
      "step": 96530
    },
    {
      "epoch": 39.52,
      "grad_norm": 2.1966958045959473,
      "learning_rate": 4.413859692243206e-06,
      "loss": 0.5734,
      "step": 96540
    },
    {
      "epoch": 39.52,
      "grad_norm": 3.138040781021118,
      "learning_rate": 4.413785304773911e-06,
      "loss": 0.5805,
      "step": 96550
    },
    {
      "epoch": 39.53,
      "grad_norm": 1.8419328927993774,
      "learning_rate": 4.413710907822104e-06,
      "loss": 0.5524,
      "step": 96560
    },
    {
      "epoch": 39.53,
      "grad_norm": 1.7596683502197266,
      "learning_rate": 4.413636501388128e-06,
      "loss": 0.5531,
      "step": 96570
    },
    {
      "epoch": 39.53,
      "grad_norm": 1.8206329345703125,
      "learning_rate": 4.4135620854723225e-06,
      "loss": 0.5558,
      "step": 96580
    },
    {
      "epoch": 39.54,
      "grad_norm": 1.6915926933288574,
      "learning_rate": 4.4134876600750275e-06,
      "loss": 0.5596,
      "step": 96590
    },
    {
      "epoch": 39.54,
      "grad_norm": 1.953357219696045,
      "learning_rate": 4.413413225196585e-06,
      "loss": 0.5587,
      "step": 96600
    },
    {
      "epoch": 39.55,
      "grad_norm": 2.177748680114746,
      "learning_rate": 4.4133387808373355e-06,
      "loss": 0.5564,
      "step": 96610
    },
    {
      "epoch": 39.55,
      "grad_norm": 2.000729560852051,
      "learning_rate": 4.413264326997621e-06,
      "loss": 0.5622,
      "step": 96620
    },
    {
      "epoch": 39.55,
      "grad_norm": 1.9477686882019043,
      "learning_rate": 4.413189863677782e-06,
      "loss": 0.5653,
      "step": 96630
    },
    {
      "epoch": 39.56,
      "grad_norm": 2.5535454750061035,
      "learning_rate": 4.413115390878158e-06,
      "loss": 0.5627,
      "step": 96640
    },
    {
      "epoch": 39.56,
      "grad_norm": 1.9393095970153809,
      "learning_rate": 4.413040908599093e-06,
      "loss": 0.5572,
      "step": 96650
    },
    {
      "epoch": 39.57,
      "grad_norm": 2.612898588180542,
      "learning_rate": 4.412966416840927e-06,
      "loss": 0.5652,
      "step": 96660
    },
    {
      "epoch": 39.57,
      "grad_norm": 1.7338006496429443,
      "learning_rate": 4.412891915604001e-06,
      "loss": 0.5598,
      "step": 96670
    },
    {
      "epoch": 39.57,
      "grad_norm": 1.7664915323257446,
      "learning_rate": 4.412817404888656e-06,
      "loss": 0.5605,
      "step": 96680
    },
    {
      "epoch": 39.58,
      "grad_norm": 2.12214732170105,
      "learning_rate": 4.412742884695233e-06,
      "loss": 0.5682,
      "step": 96690
    },
    {
      "epoch": 39.58,
      "grad_norm": 1.9501521587371826,
      "learning_rate": 4.412668355024074e-06,
      "loss": 0.5699,
      "step": 96700
    },
    {
      "epoch": 39.59,
      "grad_norm": 1.8982857465744019,
      "learning_rate": 4.412593815875521e-06,
      "loss": 0.569,
      "step": 96710
    },
    {
      "epoch": 39.59,
      "grad_norm": 2.057159662246704,
      "learning_rate": 4.412519267249915e-06,
      "loss": 0.544,
      "step": 96720
    },
    {
      "epoch": 39.59,
      "grad_norm": 2.4175336360931396,
      "learning_rate": 4.412444709147598e-06,
      "loss": 0.5699,
      "step": 96730
    },
    {
      "epoch": 39.6,
      "grad_norm": 1.775512933731079,
      "learning_rate": 4.41237014156891e-06,
      "loss": 0.5672,
      "step": 96740
    },
    {
      "epoch": 39.6,
      "grad_norm": 2.2018556594848633,
      "learning_rate": 4.412295564514193e-06,
      "loss": 0.5645,
      "step": 96750
    },
    {
      "epoch": 39.61,
      "grad_norm": 2.4209210872650146,
      "learning_rate": 4.41222097798379e-06,
      "loss": 0.5594,
      "step": 96760
    },
    {
      "epoch": 39.61,
      "grad_norm": 2.876221179962158,
      "learning_rate": 4.412146381978042e-06,
      "loss": 0.5649,
      "step": 96770
    },
    {
      "epoch": 39.62,
      "grad_norm": 4.025433540344238,
      "learning_rate": 4.41207177649729e-06,
      "loss": 0.5632,
      "step": 96780
    },
    {
      "epoch": 39.62,
      "grad_norm": 2.78852915763855,
      "learning_rate": 4.411997161541875e-06,
      "loss": 0.5689,
      "step": 96790
    },
    {
      "epoch": 39.62,
      "grad_norm": 2.7243244647979736,
      "learning_rate": 4.411922537112142e-06,
      "loss": 0.5898,
      "step": 96800
    },
    {
      "epoch": 39.63,
      "grad_norm": 2.10642409324646,
      "learning_rate": 4.411847903208431e-06,
      "loss": 0.5552,
      "step": 96810
    },
    {
      "epoch": 39.63,
      "grad_norm": 2.8779137134552,
      "learning_rate": 4.411773259831083e-06,
      "loss": 0.5461,
      "step": 96820
    },
    {
      "epoch": 39.64,
      "grad_norm": 1.8390789031982422,
      "learning_rate": 4.4116986069804405e-06,
      "loss": 0.5397,
      "step": 96830
    },
    {
      "epoch": 39.64,
      "grad_norm": 2.416450023651123,
      "learning_rate": 4.411623944656846e-06,
      "loss": 0.5518,
      "step": 96840
    },
    {
      "epoch": 39.64,
      "grad_norm": 1.6745247840881348,
      "learning_rate": 4.411549272860641e-06,
      "loss": 0.5512,
      "step": 96850
    },
    {
      "epoch": 39.65,
      "grad_norm": 2.2082934379577637,
      "learning_rate": 4.411474591592168e-06,
      "loss": 0.5789,
      "step": 96860
    },
    {
      "epoch": 39.65,
      "grad_norm": 2.199294328689575,
      "learning_rate": 4.4113999008517694e-06,
      "loss": 0.5376,
      "step": 96870
    },
    {
      "epoch": 39.66,
      "grad_norm": 2.0423574447631836,
      "learning_rate": 4.411325200639786e-06,
      "loss": 0.5495,
      "step": 96880
    },
    {
      "epoch": 39.66,
      "grad_norm": 1.8899281024932861,
      "learning_rate": 4.4112504909565604e-06,
      "loss": 0.5478,
      "step": 96890
    },
    {
      "epoch": 39.66,
      "grad_norm": 2.3421294689178467,
      "learning_rate": 4.411175771802436e-06,
      "loss": 0.5543,
      "step": 96900
    },
    {
      "epoch": 39.67,
      "grad_norm": 1.7643481492996216,
      "learning_rate": 4.411101043177755e-06,
      "loss": 0.5531,
      "step": 96910
    },
    {
      "epoch": 39.67,
      "grad_norm": 2.1831278800964355,
      "learning_rate": 4.411026305082858e-06,
      "loss": 0.5601,
      "step": 96920
    },
    {
      "epoch": 39.68,
      "grad_norm": 1.8483009338378906,
      "learning_rate": 4.410951557518089e-06,
      "loss": 0.5598,
      "step": 96930
    },
    {
      "epoch": 39.68,
      "grad_norm": 2.2913613319396973,
      "learning_rate": 4.41087680048379e-06,
      "loss": 0.557,
      "step": 96940
    },
    {
      "epoch": 39.68,
      "grad_norm": 2.450174331665039,
      "learning_rate": 4.410802033980302e-06,
      "loss": 0.5614,
      "step": 96950
    },
    {
      "epoch": 39.69,
      "grad_norm": 2.0248894691467285,
      "learning_rate": 4.410727258007969e-06,
      "loss": 0.5555,
      "step": 96960
    },
    {
      "epoch": 39.69,
      "grad_norm": 2.144728899002075,
      "learning_rate": 4.410652472567135e-06,
      "loss": 0.5491,
      "step": 96970
    },
    {
      "epoch": 39.7,
      "grad_norm": 2.2161617279052734,
      "learning_rate": 4.4105776776581395e-06,
      "loss": 0.55,
      "step": 96980
    },
    {
      "epoch": 39.7,
      "grad_norm": 2.9318110942840576,
      "learning_rate": 4.410502873281327e-06,
      "loss": 0.5468,
      "step": 96990
    },
    {
      "epoch": 39.71,
      "grad_norm": 2.418473720550537,
      "learning_rate": 4.410428059437039e-06,
      "loss": 0.5394,
      "step": 97000
    },
    {
      "epoch": 39.71,
      "grad_norm": 2.219895124435425,
      "learning_rate": 4.410353236125619e-06,
      "loss": 0.5297,
      "step": 97010
    },
    {
      "epoch": 39.71,
      "grad_norm": 2.2092432975769043,
      "learning_rate": 4.41027840334741e-06,
      "loss": 0.5716,
      "step": 97020
    },
    {
      "epoch": 39.72,
      "grad_norm": 1.6239659786224365,
      "learning_rate": 4.410203561102754e-06,
      "loss": 0.5722,
      "step": 97030
    },
    {
      "epoch": 39.72,
      "grad_norm": 2.403867244720459,
      "learning_rate": 4.410128709391994e-06,
      "loss": 0.5715,
      "step": 97040
    },
    {
      "epoch": 39.73,
      "grad_norm": 1.8286411762237549,
      "learning_rate": 4.410053848215474e-06,
      "loss": 0.5602,
      "step": 97050
    },
    {
      "epoch": 39.73,
      "grad_norm": 2.0162744522094727,
      "learning_rate": 4.409978977573536e-06,
      "loss": 0.5608,
      "step": 97060
    },
    {
      "epoch": 39.73,
      "grad_norm": 1.8024206161499023,
      "learning_rate": 4.409904097466523e-06,
      "loss": 0.5643,
      "step": 97070
    },
    {
      "epoch": 39.74,
      "grad_norm": 1.7627602815628052,
      "learning_rate": 4.409829207894778e-06,
      "loss": 0.5459,
      "step": 97080
    },
    {
      "epoch": 39.74,
      "grad_norm": 1.6754790544509888,
      "learning_rate": 4.409754308858644e-06,
      "loss": 0.5692,
      "step": 97090
    },
    {
      "epoch": 39.75,
      "grad_norm": 2.2245569229125977,
      "learning_rate": 4.409679400358464e-06,
      "loss": 0.5522,
      "step": 97100
    },
    {
      "epoch": 39.75,
      "grad_norm": 2.296043872833252,
      "learning_rate": 4.409604482394582e-06,
      "loss": 0.5545,
      "step": 97110
    },
    {
      "epoch": 39.75,
      "grad_norm": 2.878809690475464,
      "learning_rate": 4.40952955496734e-06,
      "loss": 0.5705,
      "step": 97120
    },
    {
      "epoch": 39.76,
      "grad_norm": 1.9598497152328491,
      "learning_rate": 4.409454618077081e-06,
      "loss": 0.5509,
      "step": 97130
    },
    {
      "epoch": 39.76,
      "grad_norm": 2.203878879547119,
      "learning_rate": 4.409379671724151e-06,
      "loss": 0.5461,
      "step": 97140
    },
    {
      "epoch": 39.77,
      "grad_norm": 1.8853689432144165,
      "learning_rate": 4.40930471590889e-06,
      "loss": 0.5895,
      "step": 97150
    },
    {
      "epoch": 39.77,
      "grad_norm": 2.094694137573242,
      "learning_rate": 4.409229750631643e-06,
      "loss": 0.5584,
      "step": 97160
    },
    {
      "epoch": 39.77,
      "grad_norm": 1.9747021198272705,
      "learning_rate": 4.409154775892753e-06,
      "loss": 0.5447,
      "step": 97170
    },
    {
      "epoch": 39.78,
      "grad_norm": 3.7877588272094727,
      "learning_rate": 4.4090797916925645e-06,
      "loss": 0.5626,
      "step": 97180
    },
    {
      "epoch": 39.78,
      "grad_norm": 2.9122791290283203,
      "learning_rate": 4.409004798031419e-06,
      "loss": 0.5719,
      "step": 97190
    },
    {
      "epoch": 39.79,
      "grad_norm": 1.470812439918518,
      "learning_rate": 4.408929794909661e-06,
      "loss": 0.5785,
      "step": 97200
    },
    {
      "epoch": 39.79,
      "grad_norm": 2.4841296672821045,
      "learning_rate": 4.408854782327636e-06,
      "loss": 0.5527,
      "step": 97210
    },
    {
      "epoch": 39.8,
      "grad_norm": 2.801814317703247,
      "learning_rate": 4.408779760285684e-06,
      "loss": 0.5541,
      "step": 97220
    },
    {
      "epoch": 39.8,
      "grad_norm": 2.5860655307769775,
      "learning_rate": 4.408704728784151e-06,
      "loss": 0.5497,
      "step": 97230
    },
    {
      "epoch": 39.8,
      "grad_norm": 2.9904000759124756,
      "learning_rate": 4.40862968782338e-06,
      "loss": 0.5538,
      "step": 97240
    },
    {
      "epoch": 39.81,
      "grad_norm": 2.4521474838256836,
      "learning_rate": 4.4085546374037155e-06,
      "loss": 0.5311,
      "step": 97250
    },
    {
      "epoch": 39.81,
      "grad_norm": 2.5856549739837646,
      "learning_rate": 4.4084795775255e-06,
      "loss": 0.5641,
      "step": 97260
    },
    {
      "epoch": 39.82,
      "grad_norm": 2.0104427337646484,
      "learning_rate": 4.408404508189079e-06,
      "loss": 0.5788,
      "step": 97270
    },
    {
      "epoch": 39.82,
      "grad_norm": 1.913775086402893,
      "learning_rate": 4.408329429394795e-06,
      "loss": 0.5716,
      "step": 97280
    },
    {
      "epoch": 39.82,
      "grad_norm": 1.8877485990524292,
      "learning_rate": 4.408254341142992e-06,
      "loss": 0.5692,
      "step": 97290
    },
    {
      "epoch": 39.83,
      "grad_norm": 1.811247706413269,
      "learning_rate": 4.408179243434014e-06,
      "loss": 0.5311,
      "step": 97300
    },
    {
      "epoch": 39.83,
      "grad_norm": 2.44661808013916,
      "learning_rate": 4.408104136268206e-06,
      "loss": 0.5595,
      "step": 97310
    },
    {
      "epoch": 39.84,
      "grad_norm": 2.2528128623962402,
      "learning_rate": 4.408029019645912e-06,
      "loss": 0.5616,
      "step": 97320
    },
    {
      "epoch": 39.84,
      "grad_norm": 2.0888590812683105,
      "learning_rate": 4.407953893567474e-06,
      "loss": 0.5442,
      "step": 97330
    },
    {
      "epoch": 39.84,
      "grad_norm": 1.5414245128631592,
      "learning_rate": 4.407878758033239e-06,
      "loss": 0.5668,
      "step": 97340
    },
    {
      "epoch": 39.85,
      "grad_norm": 1.7053515911102295,
      "learning_rate": 4.407803613043549e-06,
      "loss": 0.5493,
      "step": 97350
    },
    {
      "epoch": 39.85,
      "grad_norm": 2.6386353969573975,
      "learning_rate": 4.40772845859875e-06,
      "loss": 0.5574,
      "step": 97360
    },
    {
      "epoch": 39.86,
      "grad_norm": 2.294262409210205,
      "learning_rate": 4.4076532946991846e-06,
      "loss": 0.5502,
      "step": 97370
    },
    {
      "epoch": 39.86,
      "grad_norm": 2.1528806686401367,
      "learning_rate": 4.407578121345199e-06,
      "loss": 0.5541,
      "step": 97380
    },
    {
      "epoch": 39.86,
      "grad_norm": 2.3674004077911377,
      "learning_rate": 4.407502938537135e-06,
      "loss": 0.5623,
      "step": 97390
    },
    {
      "epoch": 39.87,
      "grad_norm": 2.8398075103759766,
      "learning_rate": 4.407427746275338e-06,
      "loss": 0.5715,
      "step": 97400
    },
    {
      "epoch": 39.87,
      "grad_norm": 2.5898475646972656,
      "learning_rate": 4.407352544560154e-06,
      "loss": 0.5468,
      "step": 97410
    },
    {
      "epoch": 39.88,
      "grad_norm": 3.018876552581787,
      "learning_rate": 4.4072773333919255e-06,
      "loss": 0.572,
      "step": 97420
    },
    {
      "epoch": 39.88,
      "grad_norm": 1.9890093803405762,
      "learning_rate": 4.407202112770999e-06,
      "loss": 0.5452,
      "step": 97430
    },
    {
      "epoch": 39.89,
      "grad_norm": 2.0600826740264893,
      "learning_rate": 4.407126882697717e-06,
      "loss": 0.5707,
      "step": 97440
    },
    {
      "epoch": 39.89,
      "grad_norm": 2.6833646297454834,
      "learning_rate": 4.407051643172425e-06,
      "loss": 0.5405,
      "step": 97450
    },
    {
      "epoch": 39.89,
      "grad_norm": 2.5310449600219727,
      "learning_rate": 4.406976394195468e-06,
      "loss": 0.586,
      "step": 97460
    },
    {
      "epoch": 39.9,
      "grad_norm": 2.0237627029418945,
      "learning_rate": 4.4069011357671904e-06,
      "loss": 0.5714,
      "step": 97470
    },
    {
      "epoch": 39.9,
      "grad_norm": 2.5267157554626465,
      "learning_rate": 4.406825867887937e-06,
      "loss": 0.5639,
      "step": 97480
    },
    {
      "epoch": 39.91,
      "grad_norm": 2.97294020652771,
      "learning_rate": 4.406750590558053e-06,
      "loss": 0.5518,
      "step": 97490
    },
    {
      "epoch": 39.91,
      "grad_norm": 1.9656175374984741,
      "learning_rate": 4.406675303777882e-06,
      "loss": 0.5853,
      "step": 97500
    },
    {
      "epoch": 39.91,
      "grad_norm": 2.531808376312256,
      "learning_rate": 4.40660000754777e-06,
      "loss": 0.5757,
      "step": 97510
    },
    {
      "epoch": 39.92,
      "grad_norm": 2.0010151863098145,
      "learning_rate": 4.406524701868062e-06,
      "loss": 0.5273,
      "step": 97520
    },
    {
      "epoch": 39.92,
      "grad_norm": 1.9800455570220947,
      "learning_rate": 4.4064493867391015e-06,
      "loss": 0.565,
      "step": 97530
    },
    {
      "epoch": 39.93,
      "grad_norm": 2.5079166889190674,
      "learning_rate": 4.406374062161236e-06,
      "loss": 0.5562,
      "step": 97540
    },
    {
      "epoch": 39.93,
      "grad_norm": 2.009908437728882,
      "learning_rate": 4.406298728134808e-06,
      "loss": 0.5417,
      "step": 97550
    },
    {
      "epoch": 39.93,
      "grad_norm": 2.328829050064087,
      "learning_rate": 4.406223384660164e-06,
      "loss": 0.555,
      "step": 97560
    },
    {
      "epoch": 39.94,
      "grad_norm": 2.706676721572876,
      "learning_rate": 4.406148031737649e-06,
      "loss": 0.5629,
      "step": 97570
    },
    {
      "epoch": 39.94,
      "grad_norm": 2.1518425941467285,
      "learning_rate": 4.406072669367608e-06,
      "loss": 0.5539,
      "step": 97580
    },
    {
      "epoch": 39.95,
      "grad_norm": 2.291797399520874,
      "learning_rate": 4.4059972975503856e-06,
      "loss": 0.5688,
      "step": 97590
    },
    {
      "epoch": 39.95,
      "grad_norm": 2.0953657627105713,
      "learning_rate": 4.405921916286329e-06,
      "loss": 0.543,
      "step": 97600
    },
    {
      "epoch": 39.95,
      "grad_norm": 3.609057664871216,
      "learning_rate": 4.405846525575781e-06,
      "loss": 0.5831,
      "step": 97610
    },
    {
      "epoch": 39.96,
      "grad_norm": 2.8816604614257812,
      "learning_rate": 4.405771125419089e-06,
      "loss": 0.5508,
      "step": 97620
    },
    {
      "epoch": 39.96,
      "grad_norm": 1.9747153520584106,
      "learning_rate": 4.405695715816597e-06,
      "loss": 0.544,
      "step": 97630
    },
    {
      "epoch": 39.97,
      "grad_norm": 2.888554573059082,
      "learning_rate": 4.405620296768652e-06,
      "loss": 0.5508,
      "step": 97640
    },
    {
      "epoch": 39.97,
      "grad_norm": 2.0136733055114746,
      "learning_rate": 4.405544868275597e-06,
      "loss": 0.5711,
      "step": 97650
    },
    {
      "epoch": 39.98,
      "grad_norm": 2.728132963180542,
      "learning_rate": 4.40546943033778e-06,
      "loss": 0.5666,
      "step": 97660
    },
    {
      "epoch": 39.98,
      "grad_norm": 1.7594304084777832,
      "learning_rate": 4.405393982955545e-06,
      "loss": 0.5658,
      "step": 97670
    },
    {
      "epoch": 39.98,
      "grad_norm": 2.6599433422088623,
      "learning_rate": 4.405318526129239e-06,
      "loss": 0.5564,
      "step": 97680
    },
    {
      "epoch": 39.99,
      "grad_norm": 1.5329645872116089,
      "learning_rate": 4.405243059859206e-06,
      "loss": 0.562,
      "step": 97690
    },
    {
      "epoch": 39.99,
      "grad_norm": 1.7848870754241943,
      "learning_rate": 4.405167584145793e-06,
      "loss": 0.53,
      "step": 97700
    },
    {
      "epoch": 40.0,
      "grad_norm": 3.0303499698638916,
      "learning_rate": 4.405092098989345e-06,
      "loss": 0.5579,
      "step": 97710
    },
    {
      "epoch": 40.0,
      "grad_norm": 1.669691801071167,
      "learning_rate": 4.4050166043902084e-06,
      "loss": 0.5725,
      "step": 97720
    },
    {
      "epoch": 40.0,
      "eval_loss": 0.5551761388778687,
      "eval_runtime": 51.8881,
      "eval_samples_per_second": 66.47,
      "eval_steps_per_second": 8.326,
      "step": 97720
    },
    {
      "epoch": 40.0,
      "grad_norm": 3.00700044631958,
      "learning_rate": 4.404941100348729e-06,
      "loss": 0.5518,
      "step": 97730
    },
    {
      "epoch": 40.01,
      "grad_norm": 1.6129648685455322,
      "learning_rate": 4.404865586865251e-06,
      "loss": 0.5507,
      "step": 97740
    },
    {
      "epoch": 40.01,
      "grad_norm": 2.068580389022827,
      "learning_rate": 4.404790063940123e-06,
      "loss": 0.5546,
      "step": 97750
    },
    {
      "epoch": 40.02,
      "grad_norm": 2.043867588043213,
      "learning_rate": 4.404714531573689e-06,
      "loss": 0.5385,
      "step": 97760
    },
    {
      "epoch": 40.02,
      "grad_norm": 2.1598856449127197,
      "learning_rate": 4.404638989766296e-06,
      "loss": 0.5652,
      "step": 97770
    },
    {
      "epoch": 40.02,
      "grad_norm": 2.5880978107452393,
      "learning_rate": 4.404563438518291e-06,
      "loss": 0.5534,
      "step": 97780
    },
    {
      "epoch": 40.03,
      "grad_norm": 1.9968892335891724,
      "learning_rate": 4.404487877830017e-06,
      "loss": 0.5708,
      "step": 97790
    },
    {
      "epoch": 40.03,
      "grad_norm": 1.96083402633667,
      "learning_rate": 4.404412307701822e-06,
      "loss": 0.5594,
      "step": 97800
    },
    {
      "epoch": 40.04,
      "grad_norm": 1.9830238819122314,
      "learning_rate": 4.404336728134054e-06,
      "loss": 0.5706,
      "step": 97810
    },
    {
      "epoch": 40.04,
      "grad_norm": 1.8457247018814087,
      "learning_rate": 4.404261139127055e-06,
      "loss": 0.5553,
      "step": 97820
    },
    {
      "epoch": 40.05,
      "grad_norm": 1.8598226308822632,
      "learning_rate": 4.404185540681174e-06,
      "loss": 0.5475,
      "step": 97830
    },
    {
      "epoch": 40.05,
      "grad_norm": 2.4975547790527344,
      "learning_rate": 4.404109932796757e-06,
      "loss": 0.5453,
      "step": 97840
    },
    {
      "epoch": 40.05,
      "grad_norm": 2.4830141067504883,
      "learning_rate": 4.404034315474151e-06,
      "loss": 0.5503,
      "step": 97850
    },
    {
      "epoch": 40.06,
      "grad_norm": 2.239849090576172,
      "learning_rate": 4.403958688713701e-06,
      "loss": 0.5771,
      "step": 97860
    },
    {
      "epoch": 40.06,
      "grad_norm": 1.8206992149353027,
      "learning_rate": 4.403883052515754e-06,
      "loss": 0.5634,
      "step": 97870
    },
    {
      "epoch": 40.07,
      "grad_norm": 1.9841712713241577,
      "learning_rate": 4.403807406880657e-06,
      "loss": 0.5566,
      "step": 97880
    },
    {
      "epoch": 40.07,
      "grad_norm": 1.7843647003173828,
      "learning_rate": 4.403731751808756e-06,
      "loss": 0.5601,
      "step": 97890
    },
    {
      "epoch": 40.07,
      "grad_norm": 1.9785667657852173,
      "learning_rate": 4.403656087300398e-06,
      "loss": 0.5477,
      "step": 97900
    },
    {
      "epoch": 40.08,
      "grad_norm": 1.824467658996582,
      "learning_rate": 4.403580413355928e-06,
      "loss": 0.5464,
      "step": 97910
    },
    {
      "epoch": 40.08,
      "grad_norm": 1.8094942569732666,
      "learning_rate": 4.403504729975695e-06,
      "loss": 0.5506,
      "step": 97920
    },
    {
      "epoch": 40.09,
      "grad_norm": 2.0295684337615967,
      "learning_rate": 4.403429037160045e-06,
      "loss": 0.5588,
      "step": 97930
    },
    {
      "epoch": 40.09,
      "grad_norm": 1.5659472942352295,
      "learning_rate": 4.4033533349093225e-06,
      "loss": 0.5451,
      "step": 97940
    },
    {
      "epoch": 40.09,
      "grad_norm": 2.196483850479126,
      "learning_rate": 4.403277623223877e-06,
      "loss": 0.5346,
      "step": 97950
    },
    {
      "epoch": 40.1,
      "grad_norm": 1.9153074026107788,
      "learning_rate": 4.403201902104055e-06,
      "loss": 0.5653,
      "step": 97960
    },
    {
      "epoch": 40.1,
      "grad_norm": 2.092785358428955,
      "learning_rate": 4.4031261715502025e-06,
      "loss": 0.5685,
      "step": 97970
    },
    {
      "epoch": 40.11,
      "grad_norm": 2.5512001514434814,
      "learning_rate": 4.403050431562667e-06,
      "loss": 0.5374,
      "step": 97980
    },
    {
      "epoch": 40.11,
      "grad_norm": 2.431213855743408,
      "learning_rate": 4.4029746821417945e-06,
      "loss": 0.5641,
      "step": 97990
    },
    {
      "epoch": 40.11,
      "grad_norm": 2.460116147994995,
      "learning_rate": 4.402898923287932e-06,
      "loss": 0.5512,
      "step": 98000
    },
    {
      "epoch": 40.12,
      "grad_norm": 2.407264471054077,
      "learning_rate": 4.402823155001428e-06,
      "loss": 0.5678,
      "step": 98010
    },
    {
      "epoch": 40.12,
      "grad_norm": 2.025463581085205,
      "learning_rate": 4.402747377282629e-06,
      "loss": 0.5494,
      "step": 98020
    },
    {
      "epoch": 40.13,
      "grad_norm": 2.0791127681732178,
      "learning_rate": 4.402671590131881e-06,
      "loss": 0.5417,
      "step": 98030
    },
    {
      "epoch": 40.13,
      "grad_norm": 2.891033411026001,
      "learning_rate": 4.402595793549533e-06,
      "loss": 0.5647,
      "step": 98040
    },
    {
      "epoch": 40.14,
      "grad_norm": 1.775838851928711,
      "learning_rate": 4.402519987535931e-06,
      "loss": 0.5468,
      "step": 98050
    },
    {
      "epoch": 40.14,
      "grad_norm": 2.706804037094116,
      "learning_rate": 4.402444172091422e-06,
      "loss": 0.5463,
      "step": 98060
    },
    {
      "epoch": 40.14,
      "grad_norm": 2.0229127407073975,
      "learning_rate": 4.402368347216354e-06,
      "loss": 0.5688,
      "step": 98070
    },
    {
      "epoch": 40.15,
      "grad_norm": 2.4850854873657227,
      "learning_rate": 4.402292512911075e-06,
      "loss": 0.5682,
      "step": 98080
    },
    {
      "epoch": 40.15,
      "grad_norm": 2.1891930103302,
      "learning_rate": 4.402216669175931e-06,
      "loss": 0.5578,
      "step": 98090
    },
    {
      "epoch": 40.16,
      "grad_norm": 2.1293129920959473,
      "learning_rate": 4.40214081601127e-06,
      "loss": 0.5553,
      "step": 98100
    },
    {
      "epoch": 40.16,
      "grad_norm": 2.213042736053467,
      "learning_rate": 4.40206495341744e-06,
      "loss": 0.5515,
      "step": 98110
    },
    {
      "epoch": 40.16,
      "grad_norm": 1.787521243095398,
      "learning_rate": 4.401989081394787e-06,
      "loss": 0.5619,
      "step": 98120
    },
    {
      "epoch": 40.17,
      "grad_norm": 2.2803092002868652,
      "learning_rate": 4.40191319994366e-06,
      "loss": 0.5518,
      "step": 98130
    },
    {
      "epoch": 40.17,
      "grad_norm": 2.4154813289642334,
      "learning_rate": 4.4018373090644055e-06,
      "loss": 0.5558,
      "step": 98140
    },
    {
      "epoch": 40.18,
      "grad_norm": 2.4718029499053955,
      "learning_rate": 4.4017614087573725e-06,
      "loss": 0.5491,
      "step": 98150
    },
    {
      "epoch": 40.18,
      "grad_norm": 2.1084916591644287,
      "learning_rate": 4.401685499022909e-06,
      "loss": 0.5498,
      "step": 98160
    },
    {
      "epoch": 40.18,
      "grad_norm": 2.2590107917785645,
      "learning_rate": 4.401609579861359e-06,
      "loss": 0.5533,
      "step": 98170
    },
    {
      "epoch": 40.19,
      "grad_norm": 2.127387285232544,
      "learning_rate": 4.4015336512730744e-06,
      "loss": 0.5645,
      "step": 98180
    },
    {
      "epoch": 40.19,
      "grad_norm": 1.8826043605804443,
      "learning_rate": 4.401457713258403e-06,
      "loss": 0.5391,
      "step": 98190
    },
    {
      "epoch": 40.2,
      "grad_norm": 2.7659244537353516,
      "learning_rate": 4.401381765817689e-06,
      "loss": 0.5476,
      "step": 98200
    },
    {
      "epoch": 40.2,
      "grad_norm": 2.2436904907226562,
      "learning_rate": 4.401305808951283e-06,
      "loss": 0.5764,
      "step": 98210
    },
    {
      "epoch": 40.2,
      "grad_norm": 2.2267916202545166,
      "learning_rate": 4.401229842659534e-06,
      "loss": 0.5445,
      "step": 98220
    },
    {
      "epoch": 40.21,
      "grad_norm": 2.0723876953125,
      "learning_rate": 4.401153866942787e-06,
      "loss": 0.5471,
      "step": 98230
    },
    {
      "epoch": 40.21,
      "grad_norm": 1.8513784408569336,
      "learning_rate": 4.401077881801392e-06,
      "loss": 0.5668,
      "step": 98240
    },
    {
      "epoch": 40.22,
      "grad_norm": 2.2891640663146973,
      "learning_rate": 4.401001887235696e-06,
      "loss": 0.5581,
      "step": 98250
    },
    {
      "epoch": 40.22,
      "grad_norm": 2.116922616958618,
      "learning_rate": 4.400925883246048e-06,
      "loss": 0.5697,
      "step": 98260
    },
    {
      "epoch": 40.23,
      "grad_norm": 1.839112401008606,
      "learning_rate": 4.400849869832796e-06,
      "loss": 0.5814,
      "step": 98270
    },
    {
      "epoch": 40.23,
      "grad_norm": 2.172283887863159,
      "learning_rate": 4.400773846996289e-06,
      "loss": 0.5477,
      "step": 98280
    },
    {
      "epoch": 40.23,
      "grad_norm": 1.8354928493499756,
      "learning_rate": 4.400697814736874e-06,
      "loss": 0.5519,
      "step": 98290
    },
    {
      "epoch": 40.24,
      "grad_norm": 2.437617063522339,
      "learning_rate": 4.400621773054898e-06,
      "loss": 0.5496,
      "step": 98300
    },
    {
      "epoch": 40.24,
      "grad_norm": 2.213942527770996,
      "learning_rate": 4.400545721950712e-06,
      "loss": 0.5455,
      "step": 98310
    },
    {
      "epoch": 40.25,
      "grad_norm": 2.3473119735717773,
      "learning_rate": 4.400469661424664e-06,
      "loss": 0.5529,
      "step": 98320
    },
    {
      "epoch": 40.25,
      "grad_norm": 2.006028890609741,
      "learning_rate": 4.400393591477101e-06,
      "loss": 0.551,
      "step": 98330
    },
    {
      "epoch": 40.25,
      "grad_norm": 2.3097610473632812,
      "learning_rate": 4.400317512108372e-06,
      "loss": 0.5443,
      "step": 98340
    },
    {
      "epoch": 40.26,
      "grad_norm": 1.820436716079712,
      "learning_rate": 4.4002414233188264e-06,
      "loss": 0.5591,
      "step": 98350
    },
    {
      "epoch": 40.26,
      "grad_norm": 2.252936601638794,
      "learning_rate": 4.400165325108812e-06,
      "loss": 0.5651,
      "step": 98360
    },
    {
      "epoch": 40.27,
      "grad_norm": 1.745079517364502,
      "learning_rate": 4.400089217478677e-06,
      "loss": 0.5767,
      "step": 98370
    },
    {
      "epoch": 40.27,
      "grad_norm": 2.2052884101867676,
      "learning_rate": 4.400013100428771e-06,
      "loss": 0.5691,
      "step": 98380
    },
    {
      "epoch": 40.27,
      "grad_norm": 1.7690073251724243,
      "learning_rate": 4.399936973959442e-06,
      "loss": 0.5547,
      "step": 98390
    },
    {
      "epoch": 40.28,
      "grad_norm": 1.8549712896347046,
      "learning_rate": 4.3998608380710386e-06,
      "loss": 0.5703,
      "step": 98400
    },
    {
      "epoch": 40.28,
      "grad_norm": 2.3443615436553955,
      "learning_rate": 4.39978469276391e-06,
      "loss": 0.5566,
      "step": 98410
    },
    {
      "epoch": 40.29,
      "grad_norm": 2.4698450565338135,
      "learning_rate": 4.399708538038405e-06,
      "loss": 0.5604,
      "step": 98420
    },
    {
      "epoch": 40.29,
      "grad_norm": 1.9678558111190796,
      "learning_rate": 4.399632373894872e-06,
      "loss": 0.5569,
      "step": 98430
    },
    {
      "epoch": 40.29,
      "grad_norm": 2.0479445457458496,
      "learning_rate": 4.39955620033366e-06,
      "loss": 0.5493,
      "step": 98440
    },
    {
      "epoch": 40.3,
      "grad_norm": 1.9546409845352173,
      "learning_rate": 4.399480017355119e-06,
      "loss": 0.5638,
      "step": 98450
    },
    {
      "epoch": 40.3,
      "grad_norm": 2.0343358516693115,
      "learning_rate": 4.399403824959598e-06,
      "loss": 0.5467,
      "step": 98460
    },
    {
      "epoch": 40.31,
      "grad_norm": 2.3138973712921143,
      "learning_rate": 4.399327623147444e-06,
      "loss": 0.5676,
      "step": 98470
    },
    {
      "epoch": 40.31,
      "grad_norm": 2.3741872310638428,
      "learning_rate": 4.3992514119190074e-06,
      "loss": 0.5481,
      "step": 98480
    },
    {
      "epoch": 40.32,
      "grad_norm": 2.339681386947632,
      "learning_rate": 4.399175191274637e-06,
      "loss": 0.5789,
      "step": 98490
    },
    {
      "epoch": 40.32,
      "grad_norm": 2.1910080909729004,
      "learning_rate": 4.399098961214682e-06,
      "loss": 0.5486,
      "step": 98500
    },
    {
      "epoch": 40.32,
      "grad_norm": 2.342935085296631,
      "learning_rate": 4.399022721739492e-06,
      "loss": 0.5423,
      "step": 98510
    },
    {
      "epoch": 40.33,
      "grad_norm": 1.9410814046859741,
      "learning_rate": 4.398946472849416e-06,
      "loss": 0.5472,
      "step": 98520
    },
    {
      "epoch": 40.33,
      "grad_norm": 1.616811752319336,
      "learning_rate": 4.398870214544803e-06,
      "loss": 0.5501,
      "step": 98530
    },
    {
      "epoch": 40.34,
      "grad_norm": 2.229728937149048,
      "learning_rate": 4.398793946826003e-06,
      "loss": 0.5492,
      "step": 98540
    },
    {
      "epoch": 40.34,
      "grad_norm": 1.9771904945373535,
      "learning_rate": 4.398717669693365e-06,
      "loss": 0.5598,
      "step": 98550
    },
    {
      "epoch": 40.34,
      "grad_norm": 2.4479706287384033,
      "learning_rate": 4.3986413831472385e-06,
      "loss": 0.5557,
      "step": 98560
    },
    {
      "epoch": 40.35,
      "grad_norm": 2.3816027641296387,
      "learning_rate": 4.398565087187972e-06,
      "loss": 0.5554,
      "step": 98570
    },
    {
      "epoch": 40.35,
      "grad_norm": 2.3289990425109863,
      "learning_rate": 4.398488781815916e-06,
      "loss": 0.5465,
      "step": 98580
    },
    {
      "epoch": 40.36,
      "grad_norm": 2.535414218902588,
      "learning_rate": 4.3984124670314195e-06,
      "loss": 0.5682,
      "step": 98590
    },
    {
      "epoch": 40.36,
      "grad_norm": 2.190732479095459,
      "learning_rate": 4.398336142834833e-06,
      "loss": 0.5528,
      "step": 98600
    },
    {
      "epoch": 40.36,
      "grad_norm": 2.3745081424713135,
      "learning_rate": 4.398259809226506e-06,
      "loss": 0.5569,
      "step": 98610
    },
    {
      "epoch": 40.37,
      "grad_norm": 1.8234866857528687,
      "learning_rate": 4.3981834662067875e-06,
      "loss": 0.5538,
      "step": 98620
    },
    {
      "epoch": 40.37,
      "grad_norm": 2.647066354751587,
      "learning_rate": 4.398107113776027e-06,
      "loss": 0.5454,
      "step": 98630
    },
    {
      "epoch": 40.38,
      "grad_norm": 1.7782331705093384,
      "learning_rate": 4.398030751934575e-06,
      "loss": 0.5429,
      "step": 98640
    },
    {
      "epoch": 40.38,
      "grad_norm": 1.7228929996490479,
      "learning_rate": 4.397954380682781e-06,
      "loss": 0.5709,
      "step": 98650
    },
    {
      "epoch": 40.38,
      "grad_norm": 1.9904308319091797,
      "learning_rate": 4.397878000020995e-06,
      "loss": 0.5654,
      "step": 98660
    },
    {
      "epoch": 40.39,
      "grad_norm": 2.63542103767395,
      "learning_rate": 4.397801609949567e-06,
      "loss": 0.5524,
      "step": 98670
    },
    {
      "epoch": 40.39,
      "grad_norm": 2.5921051502227783,
      "learning_rate": 4.397725210468846e-06,
      "loss": 0.5389,
      "step": 98680
    },
    {
      "epoch": 40.4,
      "grad_norm": 1.8864338397979736,
      "learning_rate": 4.397648801579183e-06,
      "loss": 0.5648,
      "step": 98690
    },
    {
      "epoch": 40.4,
      "grad_norm": 1.8185008764266968,
      "learning_rate": 4.397572383280928e-06,
      "loss": 0.5376,
      "step": 98700
    },
    {
      "epoch": 40.41,
      "grad_norm": 2.6472837924957275,
      "learning_rate": 4.3974959555744294e-06,
      "loss": 0.5394,
      "step": 98710
    },
    {
      "epoch": 40.41,
      "grad_norm": 1.9837911128997803,
      "learning_rate": 4.39741951846004e-06,
      "loss": 0.5688,
      "step": 98720
    },
    {
      "epoch": 40.41,
      "grad_norm": 1.6872798204421997,
      "learning_rate": 4.397343071938108e-06,
      "loss": 0.5539,
      "step": 98730
    },
    {
      "epoch": 40.42,
      "grad_norm": 2.3362009525299072,
      "learning_rate": 4.397266616008985e-06,
      "loss": 0.5707,
      "step": 98740
    },
    {
      "epoch": 40.42,
      "grad_norm": 2.315833330154419,
      "learning_rate": 4.397190150673019e-06,
      "loss": 0.5531,
      "step": 98750
    },
    {
      "epoch": 40.43,
      "grad_norm": 1.9763776063919067,
      "learning_rate": 4.3971136759305635e-06,
      "loss": 0.5589,
      "step": 98760
    },
    {
      "epoch": 40.43,
      "grad_norm": 2.9646291732788086,
      "learning_rate": 4.397037191781966e-06,
      "loss": 0.5703,
      "step": 98770
    },
    {
      "epoch": 40.43,
      "grad_norm": 2.188274383544922,
      "learning_rate": 4.396960698227577e-06,
      "loss": 0.5741,
      "step": 98780
    },
    {
      "epoch": 40.44,
      "grad_norm": 1.7745658159255981,
      "learning_rate": 4.3968841952677485e-06,
      "loss": 0.5678,
      "step": 98790
    },
    {
      "epoch": 40.44,
      "grad_norm": 3.360374927520752,
      "learning_rate": 4.396807682902831e-06,
      "loss": 0.5686,
      "step": 98800
    },
    {
      "epoch": 40.45,
      "grad_norm": 1.8008559942245483,
      "learning_rate": 4.3967311611331735e-06,
      "loss": 0.5629,
      "step": 98810
    },
    {
      "epoch": 40.45,
      "grad_norm": 2.2752623558044434,
      "learning_rate": 4.396654629959127e-06,
      "loss": 0.554,
      "step": 98820
    },
    {
      "epoch": 40.45,
      "grad_norm": 2.1711034774780273,
      "learning_rate": 4.396578089381043e-06,
      "loss": 0.5723,
      "step": 98830
    },
    {
      "epoch": 40.46,
      "grad_norm": 2.32993483543396,
      "learning_rate": 4.396501539399271e-06,
      "loss": 0.5439,
      "step": 98840
    },
    {
      "epoch": 40.46,
      "grad_norm": 3.0572922229766846,
      "learning_rate": 4.396424980014162e-06,
      "loss": 0.5786,
      "step": 98850
    },
    {
      "epoch": 40.47,
      "grad_norm": 2.9984805583953857,
      "learning_rate": 4.396348411226067e-06,
      "loss": 0.5393,
      "step": 98860
    },
    {
      "epoch": 40.47,
      "grad_norm": 1.5981817245483398,
      "learning_rate": 4.396271833035336e-06,
      "loss": 0.5433,
      "step": 98870
    },
    {
      "epoch": 40.47,
      "grad_norm": 2.0208799839019775,
      "learning_rate": 4.396195245442321e-06,
      "loss": 0.5518,
      "step": 98880
    },
    {
      "epoch": 40.48,
      "grad_norm": 2.5638091564178467,
      "learning_rate": 4.396118648447372e-06,
      "loss": 0.5614,
      "step": 98890
    },
    {
      "epoch": 40.48,
      "grad_norm": 2.3163504600524902,
      "learning_rate": 4.39604204205084e-06,
      "loss": 0.5222,
      "step": 98900
    },
    {
      "epoch": 40.49,
      "grad_norm": 1.583064317703247,
      "learning_rate": 4.395965426253077e-06,
      "loss": 0.54,
      "step": 98910
    },
    {
      "epoch": 40.49,
      "grad_norm": 3.1077919006347656,
      "learning_rate": 4.395888801054431e-06,
      "loss": 0.547,
      "step": 98920
    },
    {
      "epoch": 40.5,
      "grad_norm": 2.246446371078491,
      "learning_rate": 4.395812166455257e-06,
      "loss": 0.5592,
      "step": 98930
    },
    {
      "epoch": 40.5,
      "grad_norm": 1.9649646282196045,
      "learning_rate": 4.395735522455904e-06,
      "loss": 0.5657,
      "step": 98940
    },
    {
      "epoch": 40.5,
      "grad_norm": 1.7454736232757568,
      "learning_rate": 4.3956588690567214e-06,
      "loss": 0.574,
      "step": 98950
    },
    {
      "epoch": 40.51,
      "grad_norm": 2.339891195297241,
      "learning_rate": 4.395582206258063e-06,
      "loss": 0.5756,
      "step": 98960
    },
    {
      "epoch": 40.51,
      "grad_norm": 2.214505672454834,
      "learning_rate": 4.395505534060279e-06,
      "loss": 0.5558,
      "step": 98970
    },
    {
      "epoch": 40.52,
      "grad_norm": 1.7170698642730713,
      "learning_rate": 4.39542885246372e-06,
      "loss": 0.5482,
      "step": 98980
    },
    {
      "epoch": 40.52,
      "grad_norm": 1.932904839515686,
      "learning_rate": 4.395352161468739e-06,
      "loss": 0.5392,
      "step": 98990
    },
    {
      "epoch": 40.52,
      "grad_norm": 2.1776766777038574,
      "learning_rate": 4.395275461075686e-06,
      "loss": 0.5611,
      "step": 99000
    },
    {
      "epoch": 40.53,
      "grad_norm": 1.798287034034729,
      "learning_rate": 4.3951987512849115e-06,
      "loss": 0.5581,
      "step": 99010
    },
    {
      "epoch": 40.53,
      "grad_norm": 3.2417657375335693,
      "learning_rate": 4.395122032096769e-06,
      "loss": 0.5673,
      "step": 99020
    },
    {
      "epoch": 40.54,
      "grad_norm": 2.1278679370880127,
      "learning_rate": 4.395045303511609e-06,
      "loss": 0.5591,
      "step": 99030
    },
    {
      "epoch": 40.54,
      "grad_norm": 2.1454176902770996,
      "learning_rate": 4.394968565529782e-06,
      "loss": 0.5698,
      "step": 99040
    },
    {
      "epoch": 40.54,
      "grad_norm": 2.04142165184021,
      "learning_rate": 4.394891818151641e-06,
      "loss": 0.5575,
      "step": 99050
    },
    {
      "epoch": 40.55,
      "grad_norm": 2.3723554611206055,
      "learning_rate": 4.394815061377538e-06,
      "loss": 0.5612,
      "step": 99060
    },
    {
      "epoch": 40.55,
      "grad_norm": 1.9315167665481567,
      "learning_rate": 4.394738295207822e-06,
      "loss": 0.5505,
      "step": 99070
    },
    {
      "epoch": 40.56,
      "grad_norm": 2.868222951889038,
      "learning_rate": 4.394661519642847e-06,
      "loss": 0.553,
      "step": 99080
    },
    {
      "epoch": 40.56,
      "grad_norm": 2.057589292526245,
      "learning_rate": 4.394584734682964e-06,
      "loss": 0.5678,
      "step": 99090
    },
    {
      "epoch": 40.56,
      "grad_norm": 2.1526474952697754,
      "learning_rate": 4.394507940328524e-06,
      "loss": 0.5435,
      "step": 99100
    },
    {
      "epoch": 40.57,
      "grad_norm": 2.5444676876068115,
      "learning_rate": 4.39443113657988e-06,
      "loss": 0.5579,
      "step": 99110
    },
    {
      "epoch": 40.57,
      "grad_norm": 1.958168864250183,
      "learning_rate": 4.3943543234373835e-06,
      "loss": 0.5392,
      "step": 99120
    },
    {
      "epoch": 40.58,
      "grad_norm": 2.2289891242980957,
      "learning_rate": 4.394277500901386e-06,
      "loss": 0.5543,
      "step": 99130
    },
    {
      "epoch": 40.58,
      "grad_norm": 2.4755699634552,
      "learning_rate": 4.394200668972239e-06,
      "loss": 0.5536,
      "step": 99140
    },
    {
      "epoch": 40.59,
      "grad_norm": 1.9492928981781006,
      "learning_rate": 4.394123827650295e-06,
      "loss": 0.5535,
      "step": 99150
    },
    {
      "epoch": 40.59,
      "grad_norm": 2.514298915863037,
      "learning_rate": 4.394046976935907e-06,
      "loss": 0.5575,
      "step": 99160
    },
    {
      "epoch": 40.59,
      "grad_norm": 3.4449429512023926,
      "learning_rate": 4.393970116829426e-06,
      "loss": 0.5485,
      "step": 99170
    },
    {
      "epoch": 40.6,
      "grad_norm": 2.6215314865112305,
      "learning_rate": 4.393893247331204e-06,
      "loss": 0.5386,
      "step": 99180
    },
    {
      "epoch": 40.6,
      "grad_norm": 2.267146348953247,
      "learning_rate": 4.393816368441592e-06,
      "loss": 0.5607,
      "step": 99190
    },
    {
      "epoch": 40.61,
      "grad_norm": 1.9495929479599,
      "learning_rate": 4.393739480160946e-06,
      "loss": 0.5555,
      "step": 99200
    },
    {
      "epoch": 40.61,
      "grad_norm": 2.151137590408325,
      "learning_rate": 4.393662582489614e-06,
      "loss": 0.5629,
      "step": 99210
    },
    {
      "epoch": 40.61,
      "grad_norm": 1.9232815504074097,
      "learning_rate": 4.393585675427949e-06,
      "loss": 0.5815,
      "step": 99220
    },
    {
      "epoch": 40.62,
      "grad_norm": 2.771672487258911,
      "learning_rate": 4.393508758976306e-06,
      "loss": 0.5521,
      "step": 99230
    },
    {
      "epoch": 40.62,
      "grad_norm": 2.4446651935577393,
      "learning_rate": 4.393431833135035e-06,
      "loss": 0.5679,
      "step": 99240
    },
    {
      "epoch": 40.63,
      "grad_norm": 3.034580945968628,
      "learning_rate": 4.393354897904489e-06,
      "loss": 0.5424,
      "step": 99250
    },
    {
      "epoch": 40.63,
      "grad_norm": 2.0796241760253906,
      "learning_rate": 4.3932779532850206e-06,
      "loss": 0.5516,
      "step": 99260
    },
    {
      "epoch": 40.63,
      "grad_norm": 1.893781304359436,
      "learning_rate": 4.393200999276982e-06,
      "loss": 0.5474,
      "step": 99270
    },
    {
      "epoch": 40.64,
      "grad_norm": 2.05239200592041,
      "learning_rate": 4.393124035880725e-06,
      "loss": 0.5732,
      "step": 99280
    },
    {
      "epoch": 40.64,
      "grad_norm": 2.2195887565612793,
      "learning_rate": 4.3930470630966035e-06,
      "loss": 0.5621,
      "step": 99290
    },
    {
      "epoch": 40.65,
      "grad_norm": 1.7684518098831177,
      "learning_rate": 4.392970080924969e-06,
      "loss": 0.557,
      "step": 99300
    },
    {
      "epoch": 40.65,
      "grad_norm": 2.211395502090454,
      "learning_rate": 4.3928930893661754e-06,
      "loss": 0.5706,
      "step": 99310
    },
    {
      "epoch": 40.65,
      "grad_norm": 2.185025691986084,
      "learning_rate": 4.392816088420574e-06,
      "loss": 0.5461,
      "step": 99320
    },
    {
      "epoch": 40.66,
      "grad_norm": 2.088557004928589,
      "learning_rate": 4.392739078088519e-06,
      "loss": 0.5513,
      "step": 99330
    },
    {
      "epoch": 40.66,
      "grad_norm": 3.4370837211608887,
      "learning_rate": 4.392662058370362e-06,
      "loss": 0.5488,
      "step": 99340
    },
    {
      "epoch": 40.67,
      "grad_norm": 1.8460345268249512,
      "learning_rate": 4.392585029266456e-06,
      "loss": 0.549,
      "step": 99350
    },
    {
      "epoch": 40.67,
      "grad_norm": 1.8957107067108154,
      "learning_rate": 4.392507990777155e-06,
      "loss": 0.5456,
      "step": 99360
    },
    {
      "epoch": 40.68,
      "grad_norm": 2.0258843898773193,
      "learning_rate": 4.3924309429028095e-06,
      "loss": 0.5534,
      "step": 99370
    },
    {
      "epoch": 40.68,
      "grad_norm": 1.9901013374328613,
      "learning_rate": 4.392353885643774e-06,
      "loss": 0.5445,
      "step": 99380
    },
    {
      "epoch": 40.68,
      "grad_norm": 3.124028205871582,
      "learning_rate": 4.392276819000402e-06,
      "loss": 0.5514,
      "step": 99390
    },
    {
      "epoch": 40.69,
      "grad_norm": 2.8150434494018555,
      "learning_rate": 4.392199742973046e-06,
      "loss": 0.5466,
      "step": 99400
    },
    {
      "epoch": 40.69,
      "grad_norm": 2.963984251022339,
      "learning_rate": 4.392122657562059e-06,
      "loss": 0.5619,
      "step": 99410
    },
    {
      "epoch": 40.7,
      "grad_norm": 2.0852036476135254,
      "learning_rate": 4.392045562767793e-06,
      "loss": 0.5729,
      "step": 99420
    },
    {
      "epoch": 40.7,
      "grad_norm": 2.1858439445495605,
      "learning_rate": 4.391968458590604e-06,
      "loss": 0.5607,
      "step": 99430
    },
    {
      "epoch": 40.7,
      "grad_norm": 2.0062286853790283,
      "learning_rate": 4.391891345030842e-06,
      "loss": 0.5545,
      "step": 99440
    },
    {
      "epoch": 40.71,
      "grad_norm": 2.0228495597839355,
      "learning_rate": 4.391814222088862e-06,
      "loss": 0.5735,
      "step": 99450
    },
    {
      "epoch": 40.71,
      "grad_norm": 1.8038291931152344,
      "learning_rate": 4.391737089765017e-06,
      "loss": 0.5407,
      "step": 99460
    },
    {
      "epoch": 40.72,
      "grad_norm": 2.2095441818237305,
      "learning_rate": 4.391659948059661e-06,
      "loss": 0.5447,
      "step": 99470
    },
    {
      "epoch": 40.72,
      "grad_norm": 2.070201873779297,
      "learning_rate": 4.391582796973146e-06,
      "loss": 0.5566,
      "step": 99480
    },
    {
      "epoch": 40.72,
      "grad_norm": 2.360927104949951,
      "learning_rate": 4.391505636505826e-06,
      "loss": 0.543,
      "step": 99490
    },
    {
      "epoch": 40.73,
      "grad_norm": 2.8164560794830322,
      "learning_rate": 4.391428466658055e-06,
      "loss": 0.5468,
      "step": 99500
    },
    {
      "epoch": 40.73,
      "grad_norm": 2.030853509902954,
      "learning_rate": 4.391351287430186e-06,
      "loss": 0.5521,
      "step": 99510
    },
    {
      "epoch": 40.74,
      "grad_norm": 1.7732843160629272,
      "learning_rate": 4.391274098822573e-06,
      "loss": 0.5614,
      "step": 99520
    },
    {
      "epoch": 40.74,
      "grad_norm": 1.5668690204620361,
      "learning_rate": 4.391196900835568e-06,
      "loss": 0.5833,
      "step": 99530
    },
    {
      "epoch": 40.74,
      "grad_norm": 2.642148733139038,
      "learning_rate": 4.391119693469527e-06,
      "loss": 0.5644,
      "step": 99540
    },
    {
      "epoch": 40.75,
      "grad_norm": 2.4142181873321533,
      "learning_rate": 4.391042476724802e-06,
      "loss": 0.5537,
      "step": 99550
    },
    {
      "epoch": 40.75,
      "grad_norm": 2.691389799118042,
      "learning_rate": 4.390965250601747e-06,
      "loss": 0.5633,
      "step": 99560
    },
    {
      "epoch": 40.76,
      "grad_norm": 1.9231219291687012,
      "learning_rate": 4.3908880151007165e-06,
      "loss": 0.5262,
      "step": 99570
    },
    {
      "epoch": 40.76,
      "grad_norm": 2.1611053943634033,
      "learning_rate": 4.3908107702220635e-06,
      "loss": 0.5324,
      "step": 99580
    },
    {
      "epoch": 40.77,
      "grad_norm": 2.2413246631622314,
      "learning_rate": 4.390733515966142e-06,
      "loss": 0.5596,
      "step": 99590
    },
    {
      "epoch": 40.77,
      "grad_norm": 2.104736566543579,
      "learning_rate": 4.390656252333307e-06,
      "loss": 0.5613,
      "step": 99600
    },
    {
      "epoch": 40.77,
      "grad_norm": 1.707595705986023,
      "learning_rate": 4.3905789793239105e-06,
      "loss": 0.5421,
      "step": 99610
    },
    {
      "epoch": 40.78,
      "grad_norm": 1.6450353860855103,
      "learning_rate": 4.3905016969383075e-06,
      "loss": 0.5401,
      "step": 99620
    },
    {
      "epoch": 40.78,
      "grad_norm": 2.3209235668182373,
      "learning_rate": 4.390424405176852e-06,
      "loss": 0.5395,
      "step": 99630
    },
    {
      "epoch": 40.79,
      "grad_norm": 2.597141742706299,
      "learning_rate": 4.3903471040398985e-06,
      "loss": 0.5606,
      "step": 99640
    },
    {
      "epoch": 40.79,
      "grad_norm": 1.746780514717102,
      "learning_rate": 4.3902697935278e-06,
      "loss": 0.531,
      "step": 99650
    },
    {
      "epoch": 40.79,
      "grad_norm": 2.7038986682891846,
      "learning_rate": 4.390192473640912e-06,
      "loss": 0.5445,
      "step": 99660
    },
    {
      "epoch": 40.8,
      "grad_norm": 2.335635185241699,
      "learning_rate": 4.390115144379587e-06,
      "loss": 0.5608,
      "step": 99670
    },
    {
      "epoch": 40.8,
      "grad_norm": 2.389085054397583,
      "learning_rate": 4.39003780574418e-06,
      "loss": 0.5594,
      "step": 99680
    },
    {
      "epoch": 40.81,
      "grad_norm": 2.8634085655212402,
      "learning_rate": 4.389960457735046e-06,
      "loss": 0.5704,
      "step": 99690
    },
    {
      "epoch": 40.81,
      "grad_norm": 1.8718756437301636,
      "learning_rate": 4.3898831003525396e-06,
      "loss": 0.5536,
      "step": 99700
    },
    {
      "epoch": 40.81,
      "grad_norm": 2.153198480606079,
      "learning_rate": 4.389805733597014e-06,
      "loss": 0.5691,
      "step": 99710
    },
    {
      "epoch": 40.82,
      "grad_norm": 2.3821630477905273,
      "learning_rate": 4.389728357468823e-06,
      "loss": 0.5611,
      "step": 99720
    },
    {
      "epoch": 40.82,
      "grad_norm": 1.6920511722564697,
      "learning_rate": 4.389650971968323e-06,
      "loss": 0.5442,
      "step": 99730
    },
    {
      "epoch": 40.83,
      "grad_norm": 2.458378791809082,
      "learning_rate": 4.389573577095867e-06,
      "loss": 0.5725,
      "step": 99740
    },
    {
      "epoch": 40.83,
      "grad_norm": 2.6172878742218018,
      "learning_rate": 4.3894961728518095e-06,
      "loss": 0.5734,
      "step": 99750
    },
    {
      "epoch": 40.84,
      "grad_norm": 2.500044345855713,
      "learning_rate": 4.389418759236507e-06,
      "loss": 0.5634,
      "step": 99760
    },
    {
      "epoch": 40.84,
      "grad_norm": 2.3394765853881836,
      "learning_rate": 4.3893413362503114e-06,
      "loss": 0.5534,
      "step": 99770
    },
    {
      "epoch": 40.84,
      "grad_norm": 1.9660725593566895,
      "learning_rate": 4.38926390389358e-06,
      "loss": 0.5376,
      "step": 99780
    },
    {
      "epoch": 40.85,
      "grad_norm": 2.04080867767334,
      "learning_rate": 4.389186462166665e-06,
      "loss": 0.5625,
      "step": 99790
    },
    {
      "epoch": 40.85,
      "grad_norm": 2.94579815864563,
      "learning_rate": 4.389109011069922e-06,
      "loss": 0.5288,
      "step": 99800
    },
    {
      "epoch": 40.86,
      "grad_norm": 1.6802009344100952,
      "learning_rate": 4.389031550603708e-06,
      "loss": 0.5615,
      "step": 99810
    },
    {
      "epoch": 40.86,
      "grad_norm": 3.2498319149017334,
      "learning_rate": 4.388954080768374e-06,
      "loss": 0.5748,
      "step": 99820
    },
    {
      "epoch": 40.86,
      "grad_norm": 1.8857868909835815,
      "learning_rate": 4.388876601564278e-06,
      "loss": 0.538,
      "step": 99830
    },
    {
      "epoch": 40.87,
      "grad_norm": 2.9269375801086426,
      "learning_rate": 4.3887991129917735e-06,
      "loss": 0.5603,
      "step": 99840
    },
    {
      "epoch": 40.87,
      "grad_norm": 2.425809144973755,
      "learning_rate": 4.388721615051216e-06,
      "loss": 0.5395,
      "step": 99850
    },
    {
      "epoch": 40.88,
      "grad_norm": 2.1650705337524414,
      "learning_rate": 4.38864410774296e-06,
      "loss": 0.5582,
      "step": 99860
    },
    {
      "epoch": 40.88,
      "grad_norm": 2.0051629543304443,
      "learning_rate": 4.388566591067361e-06,
      "loss": 0.5694,
      "step": 99870
    },
    {
      "epoch": 40.88,
      "grad_norm": 2.217139482498169,
      "learning_rate": 4.388489065024773e-06,
      "loss": 0.551,
      "step": 99880
    },
    {
      "epoch": 40.89,
      "grad_norm": 2.138972759246826,
      "learning_rate": 4.388411529615553e-06,
      "loss": 0.5472,
      "step": 99890
    },
    {
      "epoch": 40.89,
      "grad_norm": 1.7537380456924438,
      "learning_rate": 4.388333984840055e-06,
      "loss": 0.5484,
      "step": 99900
    },
    {
      "epoch": 40.9,
      "grad_norm": 2.677037239074707,
      "learning_rate": 4.388256430698634e-06,
      "loss": 0.5447,
      "step": 99910
    },
    {
      "epoch": 40.9,
      "grad_norm": 2.322882652282715,
      "learning_rate": 4.388178867191646e-06,
      "loss": 0.5765,
      "step": 99920
    },
    {
      "epoch": 40.9,
      "grad_norm": 2.4868087768554688,
      "learning_rate": 4.388101294319446e-06,
      "loss": 0.5639,
      "step": 99930
    },
    {
      "epoch": 40.91,
      "grad_norm": 2.2156147956848145,
      "learning_rate": 4.388023712082389e-06,
      "loss": 0.5514,
      "step": 99940
    },
    {
      "epoch": 40.91,
      "grad_norm": 2.4964070320129395,
      "learning_rate": 4.387946120480831e-06,
      "loss": 0.5727,
      "step": 99950
    },
    {
      "epoch": 40.92,
      "grad_norm": 2.4856526851654053,
      "learning_rate": 4.387868519515127e-06,
      "loss": 0.5384,
      "step": 99960
    },
    {
      "epoch": 40.92,
      "grad_norm": 2.4797303676605225,
      "learning_rate": 4.387790909185633e-06,
      "loss": 0.5452,
      "step": 99970
    },
    {
      "epoch": 40.93,
      "grad_norm": 1.8781930208206177,
      "learning_rate": 4.3877132894927036e-06,
      "loss": 0.5496,
      "step": 99980
    },
    {
      "epoch": 40.93,
      "grad_norm": 2.0142569541931152,
      "learning_rate": 4.387635660436696e-06,
      "loss": 0.5346,
      "step": 99990
    },
    {
      "epoch": 40.93,
      "grad_norm": 2.3964672088623047,
      "learning_rate": 4.3875580220179636e-06,
      "loss": 0.5452,
      "step": 100000
    },
    {
      "epoch": 40.94,
      "grad_norm": 1.8390291929244995,
      "learning_rate": 4.3874803742368635e-06,
      "loss": 0.5622,
      "step": 100010
    },
    {
      "epoch": 40.94,
      "grad_norm": 2.1013498306274414,
      "learning_rate": 4.387402717093751e-06,
      "loss": 0.551,
      "step": 100020
    },
    {
      "epoch": 40.95,
      "grad_norm": 1.9248555898666382,
      "learning_rate": 4.387325050588982e-06,
      "loss": 0.5482,
      "step": 100030
    },
    {
      "epoch": 40.95,
      "grad_norm": 2.1670591831207275,
      "learning_rate": 4.387247374722912e-06,
      "loss": 0.5526,
      "step": 100040
    },
    {
      "epoch": 40.95,
      "grad_norm": 2.4056572914123535,
      "learning_rate": 4.387169689495898e-06,
      "loss": 0.5631,
      "step": 100050
    },
    {
      "epoch": 40.96,
      "grad_norm": 2.5102992057800293,
      "learning_rate": 4.387091994908293e-06,
      "loss": 0.5631,
      "step": 100060
    },
    {
      "epoch": 40.96,
      "grad_norm": 2.433762550354004,
      "learning_rate": 4.387014290960456e-06,
      "loss": 0.5381,
      "step": 100070
    },
    {
      "epoch": 40.97,
      "grad_norm": 1.7922359704971313,
      "learning_rate": 4.386936577652742e-06,
      "loss": 0.5573,
      "step": 100080
    },
    {
      "epoch": 40.97,
      "grad_norm": 2.329770088195801,
      "learning_rate": 4.386858854985506e-06,
      "loss": 0.5374,
      "step": 100090
    },
    {
      "epoch": 40.97,
      "grad_norm": 2.3707799911499023,
      "learning_rate": 4.386781122959105e-06,
      "loss": 0.5942,
      "step": 100100
    },
    {
      "epoch": 40.98,
      "grad_norm": 1.7968897819519043,
      "learning_rate": 4.386703381573895e-06,
      "loss": 0.5539,
      "step": 100110
    },
    {
      "epoch": 40.98,
      "grad_norm": 2.182675361633301,
      "learning_rate": 4.386625630830232e-06,
      "loss": 0.5444,
      "step": 100120
    },
    {
      "epoch": 40.99,
      "grad_norm": 2.0334129333496094,
      "learning_rate": 4.386547870728472e-06,
      "loss": 0.5518,
      "step": 100130
    },
    {
      "epoch": 40.99,
      "grad_norm": 2.1937782764434814,
      "learning_rate": 4.38647010126897e-06,
      "loss": 0.5472,
      "step": 100140
    },
    {
      "epoch": 40.99,
      "grad_norm": 2.278658390045166,
      "learning_rate": 4.386392322452085e-06,
      "loss": 0.5541,
      "step": 100150
    },
    {
      "epoch": 41.0,
      "grad_norm": 1.9896142482757568,
      "learning_rate": 4.386314534278172e-06,
      "loss": 0.549,
      "step": 100160
    },
    {
      "epoch": 41.0,
      "eval_loss": 0.5620504021644592,
      "eval_runtime": 52.087,
      "eval_samples_per_second": 66.216,
      "eval_steps_per_second": 8.294,
      "step": 100163
    },
    {
      "epoch": 41.0,
      "grad_norm": 2.3630826473236084,
      "learning_rate": 4.386236736747587e-06,
      "loss": 0.5584,
      "step": 100170
    },
    {
      "epoch": 41.01,
      "grad_norm": 3.1399784088134766,
      "learning_rate": 4.386158929860686e-06,
      "loss": 0.5771,
      "step": 100180
    },
    {
      "epoch": 41.01,
      "grad_norm": 1.958107352256775,
      "learning_rate": 4.3860811136178265e-06,
      "loss": 0.5621,
      "step": 100190
    },
    {
      "epoch": 41.02,
      "grad_norm": 1.7124652862548828,
      "learning_rate": 4.386003288019364e-06,
      "loss": 0.5492,
      "step": 100200
    },
    {
      "epoch": 41.02,
      "grad_norm": 2.4628243446350098,
      "learning_rate": 4.385925453065657e-06,
      "loss": 0.5782,
      "step": 100210
    },
    {
      "epoch": 41.02,
      "grad_norm": 1.961846947669983,
      "learning_rate": 4.385847608757059e-06,
      "loss": 0.5912,
      "step": 100220
    },
    {
      "epoch": 41.03,
      "grad_norm": 1.978297233581543,
      "learning_rate": 4.3857697550939276e-06,
      "loss": 0.5619,
      "step": 100230
    },
    {
      "epoch": 41.03,
      "grad_norm": 2.229363441467285,
      "learning_rate": 4.385691892076621e-06,
      "loss": 0.5572,
      "step": 100240
    },
    {
      "epoch": 41.04,
      "grad_norm": 2.916755437850952,
      "learning_rate": 4.3856140197054956e-06,
      "loss": 0.5555,
      "step": 100250
    },
    {
      "epoch": 41.04,
      "grad_norm": 2.193627119064331,
      "learning_rate": 4.3855361379809055e-06,
      "loss": 0.5609,
      "step": 100260
    },
    {
      "epoch": 41.04,
      "grad_norm": 1.9724171161651611,
      "learning_rate": 4.38545824690321e-06,
      "loss": 0.5578,
      "step": 100270
    },
    {
      "epoch": 41.05,
      "grad_norm": 2.6250669956207275,
      "learning_rate": 4.385380346472766e-06,
      "loss": 0.5696,
      "step": 100280
    },
    {
      "epoch": 41.05,
      "grad_norm": 2.6251401901245117,
      "learning_rate": 4.385302436689929e-06,
      "loss": 0.5607,
      "step": 100290
    },
    {
      "epoch": 41.06,
      "grad_norm": 2.9813876152038574,
      "learning_rate": 4.385224517555056e-06,
      "loss": 0.5382,
      "step": 100300
    },
    {
      "epoch": 41.06,
      "grad_norm": 1.5334131717681885,
      "learning_rate": 4.3851465890685056e-06,
      "loss": 0.5326,
      "step": 100310
    },
    {
      "epoch": 41.06,
      "grad_norm": 1.8893824815750122,
      "learning_rate": 4.385068651230632e-06,
      "loss": 0.5545,
      "step": 100320
    },
    {
      "epoch": 41.07,
      "grad_norm": 1.911344051361084,
      "learning_rate": 4.384990704041796e-06,
      "loss": 0.5327,
      "step": 100330
    },
    {
      "epoch": 41.07,
      "grad_norm": 2.2890520095825195,
      "learning_rate": 4.384912747502351e-06,
      "loss": 0.5257,
      "step": 100340
    },
    {
      "epoch": 41.08,
      "grad_norm": 2.4457943439483643,
      "learning_rate": 4.384834781612655e-06,
      "loss": 0.561,
      "step": 100350
    },
    {
      "epoch": 41.08,
      "grad_norm": 2.0036354064941406,
      "learning_rate": 4.3847568063730665e-06,
      "loss": 0.5216,
      "step": 100360
    },
    {
      "epoch": 41.08,
      "grad_norm": 2.057870626449585,
      "learning_rate": 4.3846788217839416e-06,
      "loss": 0.5557,
      "step": 100370
    },
    {
      "epoch": 41.09,
      "grad_norm": 2.280508518218994,
      "learning_rate": 4.384600827845638e-06,
      "loss": 0.5488,
      "step": 100380
    },
    {
      "epoch": 41.09,
      "grad_norm": 2.2011518478393555,
      "learning_rate": 4.384522824558513e-06,
      "loss": 0.5452,
      "step": 100390
    },
    {
      "epoch": 41.1,
      "grad_norm": 2.953575849533081,
      "learning_rate": 4.384444811922924e-06,
      "loss": 0.5388,
      "step": 100400
    },
    {
      "epoch": 41.1,
      "grad_norm": 2.087479829788208,
      "learning_rate": 4.384366789939227e-06,
      "loss": 0.5782,
      "step": 100410
    },
    {
      "epoch": 41.11,
      "grad_norm": 1.9811322689056396,
      "learning_rate": 4.3842887586077815e-06,
      "loss": 0.5487,
      "step": 100420
    },
    {
      "epoch": 41.11,
      "grad_norm": 2.0992565155029297,
      "learning_rate": 4.384210717928944e-06,
      "loss": 0.5502,
      "step": 100430
    },
    {
      "epoch": 41.11,
      "grad_norm": 1.8914297819137573,
      "learning_rate": 4.384132667903071e-06,
      "loss": 0.5639,
      "step": 100440
    },
    {
      "epoch": 41.12,
      "grad_norm": 2.323819160461426,
      "learning_rate": 4.384054608530522e-06,
      "loss": 0.5699,
      "step": 100450
    },
    {
      "epoch": 41.12,
      "grad_norm": 3.1216301918029785,
      "learning_rate": 4.383976539811653e-06,
      "loss": 0.5569,
      "step": 100460
    },
    {
      "epoch": 41.13,
      "grad_norm": 2.5440189838409424,
      "learning_rate": 4.383898461746822e-06,
      "loss": 0.5463,
      "step": 100470
    },
    {
      "epoch": 41.13,
      "grad_norm": 2.1458637714385986,
      "learning_rate": 4.383820374336388e-06,
      "loss": 0.5447,
      "step": 100480
    },
    {
      "epoch": 41.13,
      "grad_norm": 1.9898253679275513,
      "learning_rate": 4.383742277580706e-06,
      "loss": 0.5504,
      "step": 100490
    },
    {
      "epoch": 41.14,
      "grad_norm": 1.9510834217071533,
      "learning_rate": 4.383664171480137e-06,
      "loss": 0.5471,
      "step": 100500
    },
    {
      "epoch": 41.14,
      "grad_norm": 2.0588347911834717,
      "learning_rate": 4.383586056035036e-06,
      "loss": 0.5536,
      "step": 100510
    },
    {
      "epoch": 41.15,
      "grad_norm": 2.0611190795898438,
      "learning_rate": 4.383507931245762e-06,
      "loss": 0.542,
      "step": 100520
    },
    {
      "epoch": 41.15,
      "grad_norm": 2.1885881423950195,
      "learning_rate": 4.383429797112673e-06,
      "loss": 0.5826,
      "step": 100530
    },
    {
      "epoch": 41.15,
      "grad_norm": 2.226933240890503,
      "learning_rate": 4.383351653636127e-06,
      "loss": 0.553,
      "step": 100540
    },
    {
      "epoch": 41.16,
      "grad_norm": 2.9829864501953125,
      "learning_rate": 4.383273500816481e-06,
      "loss": 0.5334,
      "step": 100550
    },
    {
      "epoch": 41.16,
      "grad_norm": 2.4161598682403564,
      "learning_rate": 4.3831953386540946e-06,
      "loss": 0.5471,
      "step": 100560
    },
    {
      "epoch": 41.17,
      "grad_norm": 1.5816665887832642,
      "learning_rate": 4.383117167149325e-06,
      "loss": 0.535,
      "step": 100570
    },
    {
      "epoch": 41.17,
      "grad_norm": 1.706369161605835,
      "learning_rate": 4.383038986302529e-06,
      "loss": 0.5401,
      "step": 100580
    },
    {
      "epoch": 41.17,
      "grad_norm": 2.6322011947631836,
      "learning_rate": 4.382960796114066e-06,
      "loss": 0.5461,
      "step": 100590
    },
    {
      "epoch": 41.18,
      "grad_norm": 2.290370225906372,
      "learning_rate": 4.382882596584295e-06,
      "loss": 0.5589,
      "step": 100600
    },
    {
      "epoch": 41.18,
      "grad_norm": 1.9363744258880615,
      "learning_rate": 4.382804387713574e-06,
      "loss": 0.5547,
      "step": 100610
    },
    {
      "epoch": 41.19,
      "grad_norm": 2.171556234359741,
      "learning_rate": 4.38272616950226e-06,
      "loss": 0.5671,
      "step": 100620
    },
    {
      "epoch": 41.19,
      "grad_norm": 1.941332221031189,
      "learning_rate": 4.382647941950711e-06,
      "loss": 0.5299,
      "step": 100630
    },
    {
      "epoch": 41.2,
      "grad_norm": 1.9296085834503174,
      "learning_rate": 4.382569705059287e-06,
      "loss": 0.557,
      "step": 100640
    },
    {
      "epoch": 41.2,
      "grad_norm": 2.430645227432251,
      "learning_rate": 4.382491458828346e-06,
      "loss": 0.5398,
      "step": 100650
    },
    {
      "epoch": 41.2,
      "grad_norm": 2.529038190841675,
      "learning_rate": 4.382413203258245e-06,
      "loss": 0.5545,
      "step": 100660
    },
    {
      "epoch": 41.21,
      "grad_norm": 2.6595566272735596,
      "learning_rate": 4.382334938349345e-06,
      "loss": 0.5607,
      "step": 100670
    },
    {
      "epoch": 41.21,
      "grad_norm": 2.560983896255493,
      "learning_rate": 4.382256664102003e-06,
      "loss": 0.5358,
      "step": 100680
    },
    {
      "epoch": 41.22,
      "grad_norm": 1.9091339111328125,
      "learning_rate": 4.3821783805165775e-06,
      "loss": 0.5541,
      "step": 100690
    },
    {
      "epoch": 41.22,
      "grad_norm": 2.691772937774658,
      "learning_rate": 4.382100087593426e-06,
      "loss": 0.5391,
      "step": 100700
    },
    {
      "epoch": 41.22,
      "grad_norm": 1.8869872093200684,
      "learning_rate": 4.3820217853329105e-06,
      "loss": 0.5679,
      "step": 100710
    },
    {
      "epoch": 41.23,
      "grad_norm": 2.365426778793335,
      "learning_rate": 4.3819434737353855e-06,
      "loss": 0.5535,
      "step": 100720
    },
    {
      "epoch": 41.23,
      "grad_norm": 2.4927353858947754,
      "learning_rate": 4.381865152801214e-06,
      "loss": 0.5502,
      "step": 100730
    },
    {
      "epoch": 41.24,
      "grad_norm": 2.0087954998016357,
      "learning_rate": 4.381786822530752e-06,
      "loss": 0.5563,
      "step": 100740
    },
    {
      "epoch": 41.24,
      "grad_norm": 2.4414002895355225,
      "learning_rate": 4.381708482924358e-06,
      "loss": 0.5348,
      "step": 100750
    },
    {
      "epoch": 41.24,
      "grad_norm": 1.8908779621124268,
      "learning_rate": 4.381630133982392e-06,
      "loss": 0.5709,
      "step": 100760
    },
    {
      "epoch": 41.25,
      "grad_norm": 2.3896758556365967,
      "learning_rate": 4.381551775705213e-06,
      "loss": 0.5305,
      "step": 100770
    },
    {
      "epoch": 41.25,
      "grad_norm": 1.852227807044983,
      "learning_rate": 4.381473408093179e-06,
      "loss": 0.5537,
      "step": 100780
    },
    {
      "epoch": 41.26,
      "grad_norm": 1.8936855792999268,
      "learning_rate": 4.3813950311466514e-06,
      "loss": 0.5592,
      "step": 100790
    },
    {
      "epoch": 41.26,
      "grad_norm": 1.6456173658370972,
      "learning_rate": 4.381316644865987e-06,
      "loss": 0.5422,
      "step": 100800
    },
    {
      "epoch": 41.26,
      "grad_norm": 2.319658041000366,
      "learning_rate": 4.381238249251543e-06,
      "loss": 0.5406,
      "step": 100810
    },
    {
      "epoch": 41.27,
      "grad_norm": 1.8079382181167603,
      "learning_rate": 4.3811598443036834e-06,
      "loss": 0.5461,
      "step": 100820
    },
    {
      "epoch": 41.27,
      "grad_norm": 1.9065812826156616,
      "learning_rate": 4.381081430022764e-06,
      "loss": 0.5587,
      "step": 100830
    },
    {
      "epoch": 41.28,
      "grad_norm": 3.2508788108825684,
      "learning_rate": 4.381003006409145e-06,
      "loss": 0.5495,
      "step": 100840
    },
    {
      "epoch": 41.28,
      "grad_norm": 1.7624024152755737,
      "learning_rate": 4.3809245734631855e-06,
      "loss": 0.5476,
      "step": 100850
    },
    {
      "epoch": 41.29,
      "grad_norm": 1.8801305294036865,
      "learning_rate": 4.380846131185245e-06,
      "loss": 0.5599,
      "step": 100860
    },
    {
      "epoch": 41.29,
      "grad_norm": 1.8110815286636353,
      "learning_rate": 4.380767679575682e-06,
      "loss": 0.5501,
      "step": 100870
    },
    {
      "epoch": 41.29,
      "grad_norm": 2.129392623901367,
      "learning_rate": 4.380689218634857e-06,
      "loss": 0.539,
      "step": 100880
    },
    {
      "epoch": 41.3,
      "grad_norm": 2.270228862762451,
      "learning_rate": 4.380610748363129e-06,
      "loss": 0.5522,
      "step": 100890
    },
    {
      "epoch": 41.3,
      "grad_norm": 2.526132822036743,
      "learning_rate": 4.380532268760856e-06,
      "loss": 0.5495,
      "step": 100900
    },
    {
      "epoch": 41.31,
      "grad_norm": 2.362438201904297,
      "learning_rate": 4.3804537798284e-06,
      "loss": 0.5722,
      "step": 100910
    },
    {
      "epoch": 41.31,
      "grad_norm": 1.983559489250183,
      "learning_rate": 4.38037528156612e-06,
      "loss": 0.5481,
      "step": 100920
    },
    {
      "epoch": 41.31,
      "grad_norm": 2.6842410564422607,
      "learning_rate": 4.380296773974374e-06,
      "loss": 0.5609,
      "step": 100930
    },
    {
      "epoch": 41.32,
      "grad_norm": 2.135620355606079,
      "learning_rate": 4.380218257053523e-06,
      "loss": 0.5417,
      "step": 100940
    },
    {
      "epoch": 41.32,
      "grad_norm": 2.71604585647583,
      "learning_rate": 4.3801397308039254e-06,
      "loss": 0.5591,
      "step": 100950
    },
    {
      "epoch": 41.33,
      "grad_norm": 2.7069735527038574,
      "learning_rate": 4.380061195225943e-06,
      "loss": 0.5436,
      "step": 100960
    },
    {
      "epoch": 41.33,
      "grad_norm": 2.2887566089630127,
      "learning_rate": 4.3799826503199345e-06,
      "loss": 0.5582,
      "step": 100970
    },
    {
      "epoch": 41.33,
      "grad_norm": 1.649815320968628,
      "learning_rate": 4.379904096086258e-06,
      "loss": 0.5678,
      "step": 100980
    },
    {
      "epoch": 41.34,
      "grad_norm": 2.2295377254486084,
      "learning_rate": 4.379825532525276e-06,
      "loss": 0.5559,
      "step": 100990
    },
    {
      "epoch": 41.34,
      "grad_norm": 2.4757399559020996,
      "learning_rate": 4.379746959637349e-06,
      "loss": 0.5575,
      "step": 101000
    },
    {
      "epoch": 41.35,
      "grad_norm": 3.2602458000183105,
      "learning_rate": 4.379668377422833e-06,
      "loss": 0.5705,
      "step": 101010
    },
    {
      "epoch": 41.35,
      "grad_norm": 1.745428204536438,
      "learning_rate": 4.379589785882091e-06,
      "loss": 0.5649,
      "step": 101020
    },
    {
      "epoch": 41.35,
      "grad_norm": 2.937786102294922,
      "learning_rate": 4.379511185015482e-06,
      "loss": 0.5438,
      "step": 101030
    },
    {
      "epoch": 41.36,
      "grad_norm": 1.9139961004257202,
      "learning_rate": 4.379432574823366e-06,
      "loss": 0.5818,
      "step": 101040
    },
    {
      "epoch": 41.36,
      "grad_norm": 2.6724555492401123,
      "learning_rate": 4.379353955306103e-06,
      "loss": 0.5671,
      "step": 101050
    },
    {
      "epoch": 41.37,
      "grad_norm": 2.2103281021118164,
      "learning_rate": 4.379275326464055e-06,
      "loss": 0.5582,
      "step": 101060
    },
    {
      "epoch": 41.37,
      "grad_norm": 2.404836416244507,
      "learning_rate": 4.37919668829758e-06,
      "loss": 0.5809,
      "step": 101070
    },
    {
      "epoch": 41.38,
      "grad_norm": 2.4774441719055176,
      "learning_rate": 4.379118040807038e-06,
      "loss": 0.54,
      "step": 101080
    },
    {
      "epoch": 41.38,
      "grad_norm": 2.380223274230957,
      "learning_rate": 4.379039383992792e-06,
      "loss": 0.556,
      "step": 101090
    },
    {
      "epoch": 41.38,
      "grad_norm": 1.9710975885391235,
      "learning_rate": 4.3789607178552e-06,
      "loss": 0.5556,
      "step": 101100
    },
    {
      "epoch": 41.39,
      "grad_norm": 2.404501438140869,
      "learning_rate": 4.378882042394622e-06,
      "loss": 0.5835,
      "step": 101110
    },
    {
      "epoch": 41.39,
      "grad_norm": 1.9822571277618408,
      "learning_rate": 4.378803357611419e-06,
      "loss": 0.5353,
      "step": 101120
    },
    {
      "epoch": 41.4,
      "grad_norm": 2.2189276218414307,
      "learning_rate": 4.378724663505953e-06,
      "loss": 0.5743,
      "step": 101130
    },
    {
      "epoch": 41.4,
      "grad_norm": 2.8009417057037354,
      "learning_rate": 4.378645960078582e-06,
      "loss": 0.567,
      "step": 101140
    },
    {
      "epoch": 41.4,
      "grad_norm": 2.1330726146698,
      "learning_rate": 4.37856724732967e-06,
      "loss": 0.5601,
      "step": 101150
    },
    {
      "epoch": 41.41,
      "grad_norm": 2.4397146701812744,
      "learning_rate": 4.3784885252595735e-06,
      "loss": 0.5511,
      "step": 101160
    },
    {
      "epoch": 41.41,
      "grad_norm": 2.360398292541504,
      "learning_rate": 4.378409793868655e-06,
      "loss": 0.5474,
      "step": 101170
    },
    {
      "epoch": 41.42,
      "grad_norm": 2.117103099822998,
      "learning_rate": 4.378331053157275e-06,
      "loss": 0.5442,
      "step": 101180
    },
    {
      "epoch": 41.42,
      "grad_norm": 1.7895985841751099,
      "learning_rate": 4.378252303125796e-06,
      "loss": 0.571,
      "step": 101190
    },
    {
      "epoch": 41.42,
      "grad_norm": 1.7574806213378906,
      "learning_rate": 4.378173543774575e-06,
      "loss": 0.5733,
      "step": 101200
    },
    {
      "epoch": 41.43,
      "grad_norm": 2.246443033218384,
      "learning_rate": 4.378094775103976e-06,
      "loss": 0.5517,
      "step": 101210
    },
    {
      "epoch": 41.43,
      "grad_norm": 2.2803456783294678,
      "learning_rate": 4.378015997114358e-06,
      "loss": 0.574,
      "step": 101220
    },
    {
      "epoch": 41.44,
      "grad_norm": 2.527630567550659,
      "learning_rate": 4.377937209806082e-06,
      "loss": 0.5601,
      "step": 101230
    },
    {
      "epoch": 41.44,
      "grad_norm": 2.0271708965301514,
      "learning_rate": 4.377858413179511e-06,
      "loss": 0.555,
      "step": 101240
    },
    {
      "epoch": 41.44,
      "grad_norm": 2.1832525730133057,
      "learning_rate": 4.3777796072350034e-06,
      "loss": 0.538,
      "step": 101250
    },
    {
      "epoch": 41.45,
      "grad_norm": 2.110222578048706,
      "learning_rate": 4.377700791972921e-06,
      "loss": 0.5406,
      "step": 101260
    },
    {
      "epoch": 41.45,
      "grad_norm": 1.9932355880737305,
      "learning_rate": 4.377621967393625e-06,
      "loss": 0.5372,
      "step": 101270
    },
    {
      "epoch": 41.46,
      "grad_norm": 1.940951943397522,
      "learning_rate": 4.3775431334974765e-06,
      "loss": 0.5488,
      "step": 101280
    },
    {
      "epoch": 41.46,
      "grad_norm": 2.3375396728515625,
      "learning_rate": 4.3774642902848376e-06,
      "loss": 0.5643,
      "step": 101290
    },
    {
      "epoch": 41.47,
      "grad_norm": 2.0513722896575928,
      "learning_rate": 4.377385437756068e-06,
      "loss": 0.5446,
      "step": 101300
    },
    {
      "epoch": 41.47,
      "grad_norm": 2.0778939723968506,
      "learning_rate": 4.377306575911529e-06,
      "loss": 0.5626,
      "step": 101310
    },
    {
      "epoch": 41.47,
      "grad_norm": 1.7477881908416748,
      "learning_rate": 4.377227704751582e-06,
      "loss": 0.5441,
      "step": 101320
    },
    {
      "epoch": 41.48,
      "grad_norm": 2.0918190479278564,
      "learning_rate": 4.377148824276589e-06,
      "loss": 0.5479,
      "step": 101330
    },
    {
      "epoch": 41.48,
      "grad_norm": 2.5873472690582275,
      "learning_rate": 4.377069934486911e-06,
      "loss": 0.5444,
      "step": 101340
    },
    {
      "epoch": 41.49,
      "grad_norm": 2.4955711364746094,
      "learning_rate": 4.376991035382909e-06,
      "loss": 0.5737,
      "step": 101350
    },
    {
      "epoch": 41.49,
      "grad_norm": 2.857661724090576,
      "learning_rate": 4.3769121269649456e-06,
      "loss": 0.537,
      "step": 101360
    },
    {
      "epoch": 41.49,
      "grad_norm": 1.8342927694320679,
      "learning_rate": 4.37683320923338e-06,
      "loss": 0.5588,
      "step": 101370
    },
    {
      "epoch": 41.5,
      "grad_norm": 2.381561040878296,
      "learning_rate": 4.376754282188576e-06,
      "loss": 0.5328,
      "step": 101380
    },
    {
      "epoch": 41.5,
      "grad_norm": 2.4286997318267822,
      "learning_rate": 4.376675345830894e-06,
      "loss": 0.5707,
      "step": 101390
    },
    {
      "epoch": 41.51,
      "grad_norm": 2.1786789894104004,
      "learning_rate": 4.376596400160696e-06,
      "loss": 0.5571,
      "step": 101400
    },
    {
      "epoch": 41.51,
      "grad_norm": 3.28125,
      "learning_rate": 4.376517445178344e-06,
      "loss": 0.562,
      "step": 101410
    },
    {
      "epoch": 41.51,
      "grad_norm": 2.8682968616485596,
      "learning_rate": 4.376438480884197e-06,
      "loss": 0.5315,
      "step": 101420
    },
    {
      "epoch": 41.52,
      "grad_norm": 1.6961584091186523,
      "learning_rate": 4.376359507278621e-06,
      "loss": 0.5645,
      "step": 101430
    },
    {
      "epoch": 41.52,
      "grad_norm": 1.6710278987884521,
      "learning_rate": 4.376280524361975e-06,
      "loss": 0.5456,
      "step": 101440
    },
    {
      "epoch": 41.53,
      "grad_norm": 2.3218116760253906,
      "learning_rate": 4.376201532134621e-06,
      "loss": 0.5537,
      "step": 101450
    },
    {
      "epoch": 41.53,
      "grad_norm": 1.7218835353851318,
      "learning_rate": 4.376122530596922e-06,
      "loss": 0.5661,
      "step": 101460
    },
    {
      "epoch": 41.53,
      "grad_norm": 3.069349765777588,
      "learning_rate": 4.376043519749239e-06,
      "loss": 0.5503,
      "step": 101470
    },
    {
      "epoch": 41.54,
      "grad_norm": 1.9557075500488281,
      "learning_rate": 4.375964499591935e-06,
      "loss": 0.539,
      "step": 101480
    },
    {
      "epoch": 41.54,
      "grad_norm": 2.0589487552642822,
      "learning_rate": 4.375885470125369e-06,
      "loss": 0.5674,
      "step": 101490
    },
    {
      "epoch": 41.55,
      "grad_norm": 2.4969258308410645,
      "learning_rate": 4.375806431349907e-06,
      "loss": 0.5495,
      "step": 101500
    },
    {
      "epoch": 41.55,
      "grad_norm": 2.1709823608398438,
      "learning_rate": 4.375727383265908e-06,
      "loss": 0.5567,
      "step": 101510
    },
    {
      "epoch": 41.56,
      "grad_norm": 1.7451800107955933,
      "learning_rate": 4.375648325873736e-06,
      "loss": 0.5458,
      "step": 101520
    },
    {
      "epoch": 41.56,
      "grad_norm": 2.0711984634399414,
      "learning_rate": 4.375569259173752e-06,
      "loss": 0.5388,
      "step": 101530
    },
    {
      "epoch": 41.56,
      "grad_norm": 2.2610552310943604,
      "learning_rate": 4.375490183166319e-06,
      "loss": 0.5525,
      "step": 101540
    },
    {
      "epoch": 41.57,
      "grad_norm": 2.8643507957458496,
      "learning_rate": 4.375411097851799e-06,
      "loss": 0.5406,
      "step": 101550
    },
    {
      "epoch": 41.57,
      "grad_norm": 2.2630531787872314,
      "learning_rate": 4.3753320032305535e-06,
      "loss": 0.5478,
      "step": 101560
    },
    {
      "epoch": 41.58,
      "grad_norm": 1.7612196207046509,
      "learning_rate": 4.375252899302946e-06,
      "loss": 0.5346,
      "step": 101570
    },
    {
      "epoch": 41.58,
      "grad_norm": 1.4972169399261475,
      "learning_rate": 4.375173786069338e-06,
      "loss": 0.5571,
      "step": 101580
    },
    {
      "epoch": 41.58,
      "grad_norm": 2.7003753185272217,
      "learning_rate": 4.375094663530092e-06,
      "loss": 0.5486,
      "step": 101590
    },
    {
      "epoch": 41.59,
      "grad_norm": 2.080397844314575,
      "learning_rate": 4.375015531685572e-06,
      "loss": 0.5463,
      "step": 101600
    },
    {
      "epoch": 41.59,
      "grad_norm": 2.0202577114105225,
      "learning_rate": 4.3749363905361375e-06,
      "loss": 0.5476,
      "step": 101610
    },
    {
      "epoch": 41.6,
      "grad_norm": 2.145895004272461,
      "learning_rate": 4.374857240082154e-06,
      "loss": 0.5526,
      "step": 101620
    },
    {
      "epoch": 41.6,
      "grad_norm": 2.1174418926239014,
      "learning_rate": 4.374778080323982e-06,
      "loss": 0.5698,
      "step": 101630
    },
    {
      "epoch": 41.6,
      "grad_norm": 2.268665313720703,
      "learning_rate": 4.374698911261985e-06,
      "loss": 0.521,
      "step": 101640
    },
    {
      "epoch": 41.61,
      "grad_norm": 2.4860341548919678,
      "learning_rate": 4.374619732896526e-06,
      "loss": 0.5911,
      "step": 101650
    },
    {
      "epoch": 41.61,
      "grad_norm": 2.0936670303344727,
      "learning_rate": 4.374540545227967e-06,
      "loss": 0.5516,
      "step": 101660
    },
    {
      "epoch": 41.62,
      "grad_norm": 1.9092131853103638,
      "learning_rate": 4.374461348256671e-06,
      "loss": 0.5483,
      "step": 101670
    },
    {
      "epoch": 41.62,
      "grad_norm": 1.9871978759765625,
      "learning_rate": 4.374382141983001e-06,
      "loss": 0.5539,
      "step": 101680
    },
    {
      "epoch": 41.63,
      "grad_norm": 2.316063165664673,
      "learning_rate": 4.374302926407319e-06,
      "loss": 0.5391,
      "step": 101690
    },
    {
      "epoch": 41.63,
      "grad_norm": 1.998263955116272,
      "learning_rate": 4.374223701529989e-06,
      "loss": 0.5565,
      "step": 101700
    },
    {
      "epoch": 41.63,
      "grad_norm": 2.3504269123077393,
      "learning_rate": 4.374144467351374e-06,
      "loss": 0.5581,
      "step": 101710
    },
    {
      "epoch": 41.64,
      "grad_norm": 2.228036642074585,
      "learning_rate": 4.374065223871836e-06,
      "loss": 0.5539,
      "step": 101720
    },
    {
      "epoch": 41.64,
      "grad_norm": 1.7087600231170654,
      "learning_rate": 4.373985971091738e-06,
      "loss": 0.5685,
      "step": 101730
    },
    {
      "epoch": 41.65,
      "grad_norm": 2.2453393936157227,
      "learning_rate": 4.373906709011443e-06,
      "loss": 0.5751,
      "step": 101740
    },
    {
      "epoch": 41.65,
      "grad_norm": 1.8498214483261108,
      "learning_rate": 4.3738274376313155e-06,
      "loss": 0.5444,
      "step": 101750
    },
    {
      "epoch": 41.65,
      "grad_norm": 2.3793022632598877,
      "learning_rate": 4.373748156951718e-06,
      "loss": 0.5345,
      "step": 101760
    },
    {
      "epoch": 41.66,
      "grad_norm": 2.4428305625915527,
      "learning_rate": 4.373668866973013e-06,
      "loss": 0.5528,
      "step": 101770
    },
    {
      "epoch": 41.66,
      "grad_norm": 1.7596220970153809,
      "learning_rate": 4.373589567695563e-06,
      "loss": 0.5722,
      "step": 101780
    },
    {
      "epoch": 41.67,
      "grad_norm": 2.5800564289093018,
      "learning_rate": 4.373510259119733e-06,
      "loss": 0.5863,
      "step": 101790
    },
    {
      "epoch": 41.67,
      "grad_norm": 2.3289918899536133,
      "learning_rate": 4.373430941245886e-06,
      "loss": 0.5533,
      "step": 101800
    },
    {
      "epoch": 41.67,
      "grad_norm": 2.3519792556762695,
      "learning_rate": 4.373351614074384e-06,
      "loss": 0.5623,
      "step": 101810
    },
    {
      "epoch": 41.68,
      "grad_norm": 1.9051254987716675,
      "learning_rate": 4.373272277605592e-06,
      "loss": 0.5659,
      "step": 101820
    },
    {
      "epoch": 41.68,
      "grad_norm": 1.970668077468872,
      "learning_rate": 4.373192931839873e-06,
      "loss": 0.5352,
      "step": 101830
    },
    {
      "epoch": 41.69,
      "grad_norm": 2.5000154972076416,
      "learning_rate": 4.37311357677759e-06,
      "loss": 0.5731,
      "step": 101840
    },
    {
      "epoch": 41.69,
      "grad_norm": 2.1832947731018066,
      "learning_rate": 4.373034212419106e-06,
      "loss": 0.5585,
      "step": 101850
    },
    {
      "epoch": 41.69,
      "grad_norm": 1.9436522722244263,
      "learning_rate": 4.372954838764786e-06,
      "loss": 0.5456,
      "step": 101860
    },
    {
      "epoch": 41.7,
      "grad_norm": 2.3665411472320557,
      "learning_rate": 4.372875455814994e-06,
      "loss": 0.5557,
      "step": 101870
    },
    {
      "epoch": 41.7,
      "grad_norm": 2.350656747817993,
      "learning_rate": 4.37279606357009e-06,
      "loss": 0.5549,
      "step": 101880
    },
    {
      "epoch": 41.71,
      "grad_norm": 2.071450710296631,
      "learning_rate": 4.372716662030442e-06,
      "loss": 0.5379,
      "step": 101890
    },
    {
      "epoch": 41.71,
      "grad_norm": 1.8112667798995972,
      "learning_rate": 4.372637251196411e-06,
      "loss": 0.5474,
      "step": 101900
    },
    {
      "epoch": 41.72,
      "grad_norm": 2.0238873958587646,
      "learning_rate": 4.3725578310683635e-06,
      "loss": 0.5746,
      "step": 101910
    },
    {
      "epoch": 41.72,
      "grad_norm": 4.426995754241943,
      "learning_rate": 4.37247840164666e-06,
      "loss": 0.5619,
      "step": 101920
    },
    {
      "epoch": 41.72,
      "grad_norm": 1.91187584400177,
      "learning_rate": 4.372398962931666e-06,
      "loss": 0.5479,
      "step": 101930
    },
    {
      "epoch": 41.73,
      "grad_norm": 2.8714804649353027,
      "learning_rate": 4.372319514923744e-06,
      "loss": 0.5674,
      "step": 101940
    },
    {
      "epoch": 41.73,
      "grad_norm": 2.5320844650268555,
      "learning_rate": 4.3722400576232614e-06,
      "loss": 0.5605,
      "step": 101950
    },
    {
      "epoch": 41.74,
      "grad_norm": 2.8190910816192627,
      "learning_rate": 4.37216059103058e-06,
      "loss": 0.5781,
      "step": 101960
    },
    {
      "epoch": 41.74,
      "grad_norm": 2.559926748275757,
      "learning_rate": 4.372081115146063e-06,
      "loss": 0.5602,
      "step": 101970
    },
    {
      "epoch": 41.74,
      "grad_norm": 2.2527337074279785,
      "learning_rate": 4.372001629970075e-06,
      "loss": 0.5739,
      "step": 101980
    },
    {
      "epoch": 41.75,
      "grad_norm": 1.8878283500671387,
      "learning_rate": 4.371922135502981e-06,
      "loss": 0.5582,
      "step": 101990
    },
    {
      "epoch": 41.75,
      "grad_norm": 1.7533632516860962,
      "learning_rate": 4.371842631745144e-06,
      "loss": 0.5546,
      "step": 102000
    },
    {
      "epoch": 41.76,
      "grad_norm": 2.0320749282836914,
      "learning_rate": 4.371763118696929e-06,
      "loss": 0.5561,
      "step": 102010
    },
    {
      "epoch": 41.76,
      "grad_norm": 1.9630481004714966,
      "learning_rate": 4.3716835963587e-06,
      "loss": 0.5651,
      "step": 102020
    },
    {
      "epoch": 41.76,
      "grad_norm": 2.229560375213623,
      "learning_rate": 4.371604064730821e-06,
      "loss": 0.5587,
      "step": 102030
    },
    {
      "epoch": 41.77,
      "grad_norm": 2.0466935634613037,
      "learning_rate": 4.371524523813657e-06,
      "loss": 0.5471,
      "step": 102040
    },
    {
      "epoch": 41.77,
      "grad_norm": 1.8886182308197021,
      "learning_rate": 4.371444973607571e-06,
      "loss": 0.521,
      "step": 102050
    },
    {
      "epoch": 41.78,
      "grad_norm": 2.199096918106079,
      "learning_rate": 4.37136541411293e-06,
      "loss": 0.5474,
      "step": 102060
    },
    {
      "epoch": 41.78,
      "grad_norm": 2.354555130004883,
      "learning_rate": 4.371285845330096e-06,
      "loss": 0.5375,
      "step": 102070
    },
    {
      "epoch": 41.78,
      "grad_norm": 2.263551712036133,
      "learning_rate": 4.371206267259434e-06,
      "loss": 0.5492,
      "step": 102080
    },
    {
      "epoch": 41.79,
      "grad_norm": 1.62924325466156,
      "learning_rate": 4.371126679901309e-06,
      "loss": 0.5653,
      "step": 102090
    },
    {
      "epoch": 41.79,
      "grad_norm": 2.072892665863037,
      "learning_rate": 4.371047083256086e-06,
      "loss": 0.5409,
      "step": 102100
    },
    {
      "epoch": 41.8,
      "grad_norm": 1.578104853630066,
      "learning_rate": 4.370967477324128e-06,
      "loss": 0.5547,
      "step": 102110
    },
    {
      "epoch": 41.8,
      "grad_norm": 2.915766954421997,
      "learning_rate": 4.370887862105801e-06,
      "loss": 0.5353,
      "step": 102120
    },
    {
      "epoch": 41.81,
      "grad_norm": 1.730364203453064,
      "learning_rate": 4.37080823760147e-06,
      "loss": 0.5579,
      "step": 102130
    },
    {
      "epoch": 41.81,
      "grad_norm": 2.1599862575531006,
      "learning_rate": 4.370728603811499e-06,
      "loss": 0.5552,
      "step": 102140
    },
    {
      "epoch": 41.81,
      "grad_norm": 1.933364748954773,
      "learning_rate": 4.370648960736253e-06,
      "loss": 0.5493,
      "step": 102150
    },
    {
      "epoch": 41.82,
      "grad_norm": 2.0858237743377686,
      "learning_rate": 4.370569308376096e-06,
      "loss": 0.5343,
      "step": 102160
    },
    {
      "epoch": 41.82,
      "grad_norm": 2.2332568168640137,
      "learning_rate": 4.370489646731394e-06,
      "loss": 0.5617,
      "step": 102170
    },
    {
      "epoch": 41.83,
      "grad_norm": 3.3140745162963867,
      "learning_rate": 4.370409975802512e-06,
      "loss": 0.5551,
      "step": 102180
    },
    {
      "epoch": 41.83,
      "grad_norm": 2.505732297897339,
      "learning_rate": 4.370330295589814e-06,
      "loss": 0.5507,
      "step": 102190
    },
    {
      "epoch": 41.83,
      "grad_norm": 1.674935221672058,
      "learning_rate": 4.370250606093666e-06,
      "loss": 0.5542,
      "step": 102200
    },
    {
      "epoch": 41.84,
      "grad_norm": 2.1469905376434326,
      "learning_rate": 4.370170907314432e-06,
      "loss": 0.566,
      "step": 102210
    },
    {
      "epoch": 41.84,
      "grad_norm": 1.5046838521957397,
      "learning_rate": 4.370091199252478e-06,
      "loss": 0.5577,
      "step": 102220
    },
    {
      "epoch": 41.85,
      "grad_norm": 2.945429801940918,
      "learning_rate": 4.3700114819081685e-06,
      "loss": 0.5791,
      "step": 102230
    },
    {
      "epoch": 41.85,
      "grad_norm": 1.9648133516311646,
      "learning_rate": 4.369931755281869e-06,
      "loss": 0.5493,
      "step": 102240
    },
    {
      "epoch": 41.85,
      "grad_norm": 2.7348594665527344,
      "learning_rate": 4.3698520193739456e-06,
      "loss": 0.55,
      "step": 102250
    },
    {
      "epoch": 41.86,
      "grad_norm": 2.170743227005005,
      "learning_rate": 4.369772274184762e-06,
      "loss": 0.5499,
      "step": 102260
    },
    {
      "epoch": 41.86,
      "grad_norm": 1.797668695449829,
      "learning_rate": 4.3696925197146836e-06,
      "loss": 0.5424,
      "step": 102270
    },
    {
      "epoch": 41.87,
      "grad_norm": 1.7548863887786865,
      "learning_rate": 4.369612755964076e-06,
      "loss": 0.5353,
      "step": 102280
    },
    {
      "epoch": 41.87,
      "grad_norm": 2.2047410011291504,
      "learning_rate": 4.369532982933306e-06,
      "loss": 0.5619,
      "step": 102290
    },
    {
      "epoch": 41.87,
      "grad_norm": 2.0886764526367188,
      "learning_rate": 4.369453200622737e-06,
      "loss": 0.554,
      "step": 102300
    },
    {
      "epoch": 41.88,
      "grad_norm": 2.1823389530181885,
      "learning_rate": 4.369373409032736e-06,
      "loss": 0.5617,
      "step": 102310
    },
    {
      "epoch": 41.88,
      "grad_norm": 1.8272887468338013,
      "learning_rate": 4.3692936081636674e-06,
      "loss": 0.5759,
      "step": 102320
    },
    {
      "epoch": 41.89,
      "grad_norm": 2.215632200241089,
      "learning_rate": 4.369213798015898e-06,
      "loss": 0.5453,
      "step": 102330
    },
    {
      "epoch": 41.89,
      "grad_norm": 1.9292209148406982,
      "learning_rate": 4.369133978589792e-06,
      "loss": 0.535,
      "step": 102340
    },
    {
      "epoch": 41.9,
      "grad_norm": 2.115920066833496,
      "learning_rate": 4.369054149885716e-06,
      "loss": 0.5495,
      "step": 102350
    },
    {
      "epoch": 41.9,
      "grad_norm": 2.523259401321411,
      "learning_rate": 4.3689743119040346e-06,
      "loss": 0.5898,
      "step": 102360
    },
    {
      "epoch": 41.9,
      "grad_norm": 2.2648870944976807,
      "learning_rate": 4.368894464645115e-06,
      "loss": 0.551,
      "step": 102370
    },
    {
      "epoch": 41.91,
      "grad_norm": 2.105130672454834,
      "learning_rate": 4.368814608109322e-06,
      "loss": 0.5749,
      "step": 102380
    },
    {
      "epoch": 41.91,
      "grad_norm": 1.862319827079773,
      "learning_rate": 4.368734742297022e-06,
      "loss": 0.5561,
      "step": 102390
    },
    {
      "epoch": 41.92,
      "grad_norm": 1.6844046115875244,
      "learning_rate": 4.36865486720858e-06,
      "loss": 0.5538,
      "step": 102400
    },
    {
      "epoch": 41.92,
      "grad_norm": 2.336211681365967,
      "learning_rate": 4.368574982844362e-06,
      "loss": 0.5517,
      "step": 102410
    },
    {
      "epoch": 41.92,
      "grad_norm": 1.9194163084030151,
      "learning_rate": 4.3684950892047355e-06,
      "loss": 0.5391,
      "step": 102420
    },
    {
      "epoch": 41.93,
      "grad_norm": 2.449411153793335,
      "learning_rate": 4.368415186290064e-06,
      "loss": 0.555,
      "step": 102430
    },
    {
      "epoch": 41.93,
      "grad_norm": 1.748058557510376,
      "learning_rate": 4.3683352741007165e-06,
      "loss": 0.5325,
      "step": 102440
    },
    {
      "epoch": 41.94,
      "grad_norm": 1.850451946258545,
      "learning_rate": 4.368255352637056e-06,
      "loss": 0.5395,
      "step": 102450
    },
    {
      "epoch": 41.94,
      "grad_norm": 3.367419481277466,
      "learning_rate": 4.36817542189945e-06,
      "loss": 0.5579,
      "step": 102460
    },
    {
      "epoch": 41.94,
      "grad_norm": 1.8595572710037231,
      "learning_rate": 4.368095481888265e-06,
      "loss": 0.5381,
      "step": 102470
    },
    {
      "epoch": 41.95,
      "grad_norm": 2.5290281772613525,
      "learning_rate": 4.368015532603867e-06,
      "loss": 0.5552,
      "step": 102480
    },
    {
      "epoch": 41.95,
      "grad_norm": 1.993522047996521,
      "learning_rate": 4.367935574046623e-06,
      "loss": 0.5574,
      "step": 102490
    },
    {
      "epoch": 41.96,
      "grad_norm": 2.7138702869415283,
      "learning_rate": 4.3678556062168974e-06,
      "loss": 0.5394,
      "step": 102500
    },
    {
      "epoch": 41.96,
      "grad_norm": 1.7879161834716797,
      "learning_rate": 4.367775629115057e-06,
      "loss": 0.5511,
      "step": 102510
    },
    {
      "epoch": 41.96,
      "grad_norm": 1.5996874570846558,
      "learning_rate": 4.3676956427414695e-06,
      "loss": 0.5422,
      "step": 102520
    },
    {
      "epoch": 41.97,
      "grad_norm": 2.8610165119171143,
      "learning_rate": 4.3676156470965e-06,
      "loss": 0.5579,
      "step": 102530
    },
    {
      "epoch": 41.97,
      "grad_norm": 2.982748508453369,
      "learning_rate": 4.367535642180516e-06,
      "loss": 0.5507,
      "step": 102540
    },
    {
      "epoch": 41.98,
      "grad_norm": 1.5199421644210815,
      "learning_rate": 4.367455627993882e-06,
      "loss": 0.541,
      "step": 102550
    },
    {
      "epoch": 41.98,
      "grad_norm": 2.284716844558716,
      "learning_rate": 4.367375604536968e-06,
      "loss": 0.5584,
      "step": 102560
    },
    {
      "epoch": 41.99,
      "grad_norm": 2.4082443714141846,
      "learning_rate": 4.367295571810137e-06,
      "loss": 0.578,
      "step": 102570
    },
    {
      "epoch": 41.99,
      "grad_norm": 1.9217082262039185,
      "learning_rate": 4.3672155298137575e-06,
      "loss": 0.548,
      "step": 102580
    },
    {
      "epoch": 41.99,
      "grad_norm": 2.8553590774536133,
      "learning_rate": 4.367135478548196e-06,
      "loss": 0.5474,
      "step": 102590
    },
    {
      "epoch": 42.0,
      "grad_norm": 1.9631218910217285,
      "learning_rate": 4.367055418013819e-06,
      "loss": 0.5539,
      "step": 102600
    },
    {
      "epoch": 42.0,
      "eval_loss": 0.5535340309143066,
      "eval_runtime": 51.8423,
      "eval_samples_per_second": 66.529,
      "eval_steps_per_second": 8.333,
      "step": 102606
    },
    {
      "epoch": 42.0,
      "grad_norm": 1.7170987129211426,
      "learning_rate": 4.366975348210992e-06,
      "loss": 0.5512,
      "step": 102610
    },
    {
      "epoch": 42.01,
      "grad_norm": 4.15293025970459,
      "learning_rate": 4.366895269140085e-06,
      "loss": 0.5549,
      "step": 102620
    },
    {
      "epoch": 42.01,
      "grad_norm": 1.7751621007919312,
      "learning_rate": 4.366815180801462e-06,
      "loss": 0.5461,
      "step": 102630
    },
    {
      "epoch": 42.01,
      "grad_norm": 2.2984766960144043,
      "learning_rate": 4.36673508319549e-06,
      "loss": 0.5403,
      "step": 102640
    },
    {
      "epoch": 42.02,
      "grad_norm": 2.913729429244995,
      "learning_rate": 4.366654976322537e-06,
      "loss": 0.5447,
      "step": 102650
    },
    {
      "epoch": 42.02,
      "grad_norm": 1.640069603919983,
      "learning_rate": 4.366574860182971e-06,
      "loss": 0.5628,
      "step": 102660
    },
    {
      "epoch": 42.03,
      "grad_norm": 2.217100143432617,
      "learning_rate": 4.3664947347771555e-06,
      "loss": 0.5552,
      "step": 102670
    },
    {
      "epoch": 42.03,
      "grad_norm": 2.243673324584961,
      "learning_rate": 4.3664146001054605e-06,
      "loss": 0.5472,
      "step": 102680
    },
    {
      "epoch": 42.03,
      "grad_norm": 2.5005948543548584,
      "learning_rate": 4.366334456168253e-06,
      "loss": 0.5534,
      "step": 102690
    },
    {
      "epoch": 42.04,
      "grad_norm": 2.029398202896118,
      "learning_rate": 4.3662543029658976e-06,
      "loss": 0.5484,
      "step": 102700
    },
    {
      "epoch": 42.04,
      "grad_norm": 2.472731351852417,
      "learning_rate": 4.366174140498765e-06,
      "loss": 0.5577,
      "step": 102710
    },
    {
      "epoch": 42.05,
      "grad_norm": 2.340665340423584,
      "learning_rate": 4.366093968767219e-06,
      "loss": 0.5381,
      "step": 102720
    },
    {
      "epoch": 42.05,
      "grad_norm": 2.3125712871551514,
      "learning_rate": 4.36601378777163e-06,
      "loss": 0.5514,
      "step": 102730
    },
    {
      "epoch": 42.05,
      "grad_norm": 1.6697050333023071,
      "learning_rate": 4.365933597512362e-06,
      "loss": 0.5512,
      "step": 102740
    },
    {
      "epoch": 42.06,
      "grad_norm": 2.5373291969299316,
      "learning_rate": 4.365853397989786e-06,
      "loss": 0.5501,
      "step": 102750
    },
    {
      "epoch": 42.06,
      "grad_norm": 2.544093370437622,
      "learning_rate": 4.365773189204267e-06,
      "loss": 0.5685,
      "step": 102760
    },
    {
      "epoch": 42.07,
      "grad_norm": 1.9723312854766846,
      "learning_rate": 4.365692971156172e-06,
      "loss": 0.5639,
      "step": 102770
    },
    {
      "epoch": 42.07,
      "grad_norm": 1.9109318256378174,
      "learning_rate": 4.36561274384587e-06,
      "loss": 0.5427,
      "step": 102780
    },
    {
      "epoch": 42.08,
      "grad_norm": 3.0186846256256104,
      "learning_rate": 4.365532507273728e-06,
      "loss": 0.5668,
      "step": 102790
    },
    {
      "epoch": 42.08,
      "grad_norm": 2.2936878204345703,
      "learning_rate": 4.3654522614401136e-06,
      "loss": 0.5234,
      "step": 102800
    },
    {
      "epoch": 42.08,
      "grad_norm": 2.2797799110412598,
      "learning_rate": 4.365372006345394e-06,
      "loss": 0.5635,
      "step": 102810
    },
    {
      "epoch": 42.09,
      "grad_norm": 2.5432677268981934,
      "learning_rate": 4.365291741989937e-06,
      "loss": 0.5477,
      "step": 102820
    },
    {
      "epoch": 42.09,
      "grad_norm": 2.3378186225891113,
      "learning_rate": 4.3652114683741106e-06,
      "loss": 0.5295,
      "step": 102830
    },
    {
      "epoch": 42.1,
      "grad_norm": 2.0388023853302,
      "learning_rate": 4.365131185498282e-06,
      "loss": 0.54,
      "step": 102840
    },
    {
      "epoch": 42.1,
      "grad_norm": 2.2812297344207764,
      "learning_rate": 4.365050893362819e-06,
      "loss": 0.5663,
      "step": 102850
    },
    {
      "epoch": 42.1,
      "grad_norm": 2.0517632961273193,
      "learning_rate": 4.36497059196809e-06,
      "loss": 0.5401,
      "step": 102860
    },
    {
      "epoch": 42.11,
      "grad_norm": 1.6615303754806519,
      "learning_rate": 4.364890281314463e-06,
      "loss": 0.556,
      "step": 102870
    },
    {
      "epoch": 42.11,
      "grad_norm": 1.533905029296875,
      "learning_rate": 4.364809961402304e-06,
      "loss": 0.5589,
      "step": 102880
    },
    {
      "epoch": 42.12,
      "grad_norm": 2.6940135955810547,
      "learning_rate": 4.3647296322319835e-06,
      "loss": 0.5527,
      "step": 102890
    },
    {
      "epoch": 42.12,
      "grad_norm": 2.4836628437042236,
      "learning_rate": 4.364649293803868e-06,
      "loss": 0.5696,
      "step": 102900
    },
    {
      "epoch": 42.12,
      "grad_norm": 2.3706297874450684,
      "learning_rate": 4.364568946118326e-06,
      "loss": 0.5555,
      "step": 102910
    },
    {
      "epoch": 42.13,
      "grad_norm": 2.65421199798584,
      "learning_rate": 4.364488589175725e-06,
      "loss": 0.5584,
      "step": 102920
    },
    {
      "epoch": 42.13,
      "grad_norm": 2.6253087520599365,
      "learning_rate": 4.364408222976433e-06,
      "loss": 0.5804,
      "step": 102930
    },
    {
      "epoch": 42.14,
      "grad_norm": 2.1401562690734863,
      "learning_rate": 4.36432784752082e-06,
      "loss": 0.5593,
      "step": 102940
    },
    {
      "epoch": 42.14,
      "grad_norm": 2.432751417160034,
      "learning_rate": 4.364247462809252e-06,
      "loss": 0.5292,
      "step": 102950
    },
    {
      "epoch": 42.14,
      "grad_norm": 1.7800401449203491,
      "learning_rate": 4.364167068842097e-06,
      "loss": 0.5614,
      "step": 102960
    },
    {
      "epoch": 42.15,
      "grad_norm": 2.1796369552612305,
      "learning_rate": 4.364086665619726e-06,
      "loss": 0.54,
      "step": 102970
    },
    {
      "epoch": 42.15,
      "grad_norm": 1.6858599185943604,
      "learning_rate": 4.3640062531425054e-06,
      "loss": 0.5455,
      "step": 102980
    },
    {
      "epoch": 42.16,
      "grad_norm": 1.9116766452789307,
      "learning_rate": 4.363925831410803e-06,
      "loss": 0.5576,
      "step": 102990
    },
    {
      "epoch": 42.16,
      "grad_norm": 2.0081682205200195,
      "learning_rate": 4.363845400424989e-06,
      "loss": 0.546,
      "step": 103000
    },
    {
      "epoch": 42.17,
      "grad_norm": 2.8764519691467285,
      "learning_rate": 4.36376496018543e-06,
      "loss": 0.5557,
      "step": 103010
    },
    {
      "epoch": 42.17,
      "grad_norm": 2.547124147415161,
      "learning_rate": 4.363684510692495e-06,
      "loss": 0.5539,
      "step": 103020
    },
    {
      "epoch": 42.17,
      "grad_norm": 1.9980354309082031,
      "learning_rate": 4.363604051946554e-06,
      "loss": 0.5602,
      "step": 103030
    },
    {
      "epoch": 42.18,
      "grad_norm": 2.354275941848755,
      "learning_rate": 4.363523583947973e-06,
      "loss": 0.5417,
      "step": 103040
    },
    {
      "epoch": 42.18,
      "grad_norm": 2.6107518672943115,
      "learning_rate": 4.363443106697123e-06,
      "loss": 0.5428,
      "step": 103050
    },
    {
      "epoch": 42.19,
      "grad_norm": 1.9120956659317017,
      "learning_rate": 4.363362620194371e-06,
      "loss": 0.544,
      "step": 103060
    },
    {
      "epoch": 42.19,
      "grad_norm": 2.0541305541992188,
      "learning_rate": 4.363282124440087e-06,
      "loss": 0.5499,
      "step": 103070
    },
    {
      "epoch": 42.19,
      "grad_norm": 2.2971696853637695,
      "learning_rate": 4.36320161943464e-06,
      "loss": 0.5568,
      "step": 103080
    },
    {
      "epoch": 42.2,
      "grad_norm": 1.8770103454589844,
      "learning_rate": 4.3631211051783966e-06,
      "loss": 0.5446,
      "step": 103090
    },
    {
      "epoch": 42.2,
      "grad_norm": 2.434401512145996,
      "learning_rate": 4.363040581671726e-06,
      "loss": 0.56,
      "step": 103100
    },
    {
      "epoch": 42.21,
      "grad_norm": 2.574631690979004,
      "learning_rate": 4.362960048914999e-06,
      "loss": 0.5721,
      "step": 103110
    },
    {
      "epoch": 42.21,
      "grad_norm": 2.9152374267578125,
      "learning_rate": 4.3628795069085845e-06,
      "loss": 0.5281,
      "step": 103120
    },
    {
      "epoch": 42.21,
      "grad_norm": 1.7310634851455688,
      "learning_rate": 4.362798955652849e-06,
      "loss": 0.57,
      "step": 103130
    },
    {
      "epoch": 42.22,
      "grad_norm": 2.004251003265381,
      "learning_rate": 4.362718395148163e-06,
      "loss": 0.5668,
      "step": 103140
    },
    {
      "epoch": 42.22,
      "grad_norm": 2.2012174129486084,
      "learning_rate": 4.362637825394896e-06,
      "loss": 0.5475,
      "step": 103150
    },
    {
      "epoch": 42.23,
      "grad_norm": 1.8365929126739502,
      "learning_rate": 4.362557246393416e-06,
      "loss": 0.5604,
      "step": 103160
    },
    {
      "epoch": 42.23,
      "grad_norm": 1.7324609756469727,
      "learning_rate": 4.362476658144094e-06,
      "loss": 0.5504,
      "step": 103170
    },
    {
      "epoch": 42.23,
      "grad_norm": 2.5295393466949463,
      "learning_rate": 4.362396060647296e-06,
      "loss": 0.5583,
      "step": 103180
    },
    {
      "epoch": 42.24,
      "grad_norm": 1.9885717630386353,
      "learning_rate": 4.362315453903394e-06,
      "loss": 0.5653,
      "step": 103190
    },
    {
      "epoch": 42.24,
      "grad_norm": 2.3013505935668945,
      "learning_rate": 4.362234837912756e-06,
      "loss": 0.5634,
      "step": 103200
    },
    {
      "epoch": 42.25,
      "grad_norm": 2.4468228816986084,
      "learning_rate": 4.362154212675752e-06,
      "loss": 0.5748,
      "step": 103210
    },
    {
      "epoch": 42.25,
      "grad_norm": 2.6359705924987793,
      "learning_rate": 4.362073578192751e-06,
      "loss": 0.5566,
      "step": 103220
    },
    {
      "epoch": 42.26,
      "grad_norm": 1.6807883977890015,
      "learning_rate": 4.3619929344641214e-06,
      "loss": 0.5553,
      "step": 103230
    },
    {
      "epoch": 42.26,
      "grad_norm": 2.23121976852417,
      "learning_rate": 4.361912281490234e-06,
      "loss": 0.5506,
      "step": 103240
    },
    {
      "epoch": 42.26,
      "grad_norm": 2.9811313152313232,
      "learning_rate": 4.361831619271457e-06,
      "loss": 0.5441,
      "step": 103250
    },
    {
      "epoch": 42.27,
      "grad_norm": 2.4680018424987793,
      "learning_rate": 4.361750947808162e-06,
      "loss": 0.5556,
      "step": 103260
    },
    {
      "epoch": 42.27,
      "grad_norm": 3.168724536895752,
      "learning_rate": 4.361670267100716e-06,
      "loss": 0.5601,
      "step": 103270
    },
    {
      "epoch": 42.28,
      "grad_norm": 2.0134830474853516,
      "learning_rate": 4.361589577149491e-06,
      "loss": 0.5461,
      "step": 103280
    },
    {
      "epoch": 42.28,
      "grad_norm": 2.1958870887756348,
      "learning_rate": 4.361508877954855e-06,
      "loss": 0.5569,
      "step": 103290
    },
    {
      "epoch": 42.28,
      "grad_norm": 2.671189785003662,
      "learning_rate": 4.3614281695171774e-06,
      "loss": 0.5587,
      "step": 103300
    },
    {
      "epoch": 42.29,
      "grad_norm": 2.417792320251465,
      "learning_rate": 4.361347451836828e-06,
      "loss": 0.5417,
      "step": 103310
    },
    {
      "epoch": 42.29,
      "grad_norm": 1.4525600671768188,
      "learning_rate": 4.361266724914179e-06,
      "loss": 0.5632,
      "step": 103320
    },
    {
      "epoch": 42.3,
      "grad_norm": 2.1352813243865967,
      "learning_rate": 4.361185988749597e-06,
      "loss": 0.5548,
      "step": 103330
    },
    {
      "epoch": 42.3,
      "grad_norm": 2.118764638900757,
      "learning_rate": 4.361105243343453e-06,
      "loss": 0.5595,
      "step": 103340
    },
    {
      "epoch": 42.3,
      "grad_norm": 2.2392735481262207,
      "learning_rate": 4.361024488696118e-06,
      "loss": 0.5492,
      "step": 103350
    },
    {
      "epoch": 42.31,
      "grad_norm": 2.3952367305755615,
      "learning_rate": 4.360943724807961e-06,
      "loss": 0.5565,
      "step": 103360
    },
    {
      "epoch": 42.31,
      "grad_norm": 2.2700138092041016,
      "learning_rate": 4.360862951679351e-06,
      "loss": 0.562,
      "step": 103370
    },
    {
      "epoch": 42.32,
      "grad_norm": 2.058467388153076,
      "learning_rate": 4.36078216931066e-06,
      "loss": 0.5446,
      "step": 103380
    },
    {
      "epoch": 42.32,
      "grad_norm": 3.0022356510162354,
      "learning_rate": 4.360701377702257e-06,
      "loss": 0.5572,
      "step": 103390
    },
    {
      "epoch": 42.33,
      "grad_norm": 2.659609794616699,
      "learning_rate": 4.360620576854511e-06,
      "loss": 0.5601,
      "step": 103400
    },
    {
      "epoch": 42.33,
      "grad_norm": 2.8831562995910645,
      "learning_rate": 4.3605397667677935e-06,
      "loss": 0.5617,
      "step": 103410
    },
    {
      "epoch": 42.33,
      "grad_norm": 2.1912131309509277,
      "learning_rate": 4.360458947442475e-06,
      "loss": 0.5562,
      "step": 103420
    },
    {
      "epoch": 42.34,
      "grad_norm": 2.0877528190612793,
      "learning_rate": 4.360378118878925e-06,
      "loss": 0.5562,
      "step": 103430
    },
    {
      "epoch": 42.34,
      "grad_norm": 2.122953414916992,
      "learning_rate": 4.360297281077514e-06,
      "loss": 0.5336,
      "step": 103440
    },
    {
      "epoch": 42.35,
      "grad_norm": 2.8277387619018555,
      "learning_rate": 4.360216434038612e-06,
      "loss": 0.5435,
      "step": 103450
    },
    {
      "epoch": 42.35,
      "grad_norm": 2.1831209659576416,
      "learning_rate": 4.3601355777625894e-06,
      "loss": 0.5506,
      "step": 103460
    },
    {
      "epoch": 42.35,
      "grad_norm": 1.9560247659683228,
      "learning_rate": 4.360054712249817e-06,
      "loss": 0.5451,
      "step": 103470
    },
    {
      "epoch": 42.36,
      "grad_norm": 2.7266159057617188,
      "learning_rate": 4.359973837500665e-06,
      "loss": 0.5791,
      "step": 103480
    },
    {
      "epoch": 42.36,
      "grad_norm": 2.067697763442993,
      "learning_rate": 4.359892953515504e-06,
      "loss": 0.5615,
      "step": 103490
    },
    {
      "epoch": 42.37,
      "grad_norm": 1.9467594623565674,
      "learning_rate": 4.359812060294704e-06,
      "loss": 0.5546,
      "step": 103500
    },
    {
      "epoch": 42.37,
      "grad_norm": 1.771416187286377,
      "learning_rate": 4.3597311578386365e-06,
      "loss": 0.5475,
      "step": 103510
    },
    {
      "epoch": 42.37,
      "grad_norm": 2.1106390953063965,
      "learning_rate": 4.35965024614767e-06,
      "loss": 0.5709,
      "step": 103520
    },
    {
      "epoch": 42.38,
      "grad_norm": 1.9012147188186646,
      "learning_rate": 4.359569325222179e-06,
      "loss": 0.5497,
      "step": 103530
    },
    {
      "epoch": 42.38,
      "grad_norm": 2.790764331817627,
      "learning_rate": 4.3594883950625295e-06,
      "loss": 0.5453,
      "step": 103540
    },
    {
      "epoch": 42.39,
      "grad_norm": 2.653045892715454,
      "learning_rate": 4.3594074556690965e-06,
      "loss": 0.5539,
      "step": 103550
    },
    {
      "epoch": 42.39,
      "grad_norm": 1.9095207452774048,
      "learning_rate": 4.359326507042248e-06,
      "loss": 0.537,
      "step": 103560
    },
    {
      "epoch": 42.39,
      "grad_norm": 2.119849920272827,
      "learning_rate": 4.359245549182356e-06,
      "loss": 0.5597,
      "step": 103570
    },
    {
      "epoch": 42.4,
      "grad_norm": 1.840572476387024,
      "learning_rate": 4.35916458208979e-06,
      "loss": 0.52,
      "step": 103580
    },
    {
      "epoch": 42.4,
      "grad_norm": 2.0734291076660156,
      "learning_rate": 4.3590836057649225e-06,
      "loss": 0.5388,
      "step": 103590
    },
    {
      "epoch": 42.41,
      "grad_norm": 2.2056446075439453,
      "learning_rate": 4.359002620208124e-06,
      "loss": 0.5339,
      "step": 103600
    },
    {
      "epoch": 42.41,
      "grad_norm": 2.676764965057373,
      "learning_rate": 4.358921625419766e-06,
      "loss": 0.566,
      "step": 103610
    },
    {
      "epoch": 42.42,
      "grad_norm": 2.313584327697754,
      "learning_rate": 4.358840621400218e-06,
      "loss": 0.5839,
      "step": 103620
    },
    {
      "epoch": 42.42,
      "grad_norm": 1.9355666637420654,
      "learning_rate": 4.3587596081498525e-06,
      "loss": 0.5459,
      "step": 103630
    },
    {
      "epoch": 42.42,
      "grad_norm": 2.583287239074707,
      "learning_rate": 4.35867858566904e-06,
      "loss": 0.5544,
      "step": 103640
    },
    {
      "epoch": 42.43,
      "grad_norm": 2.401801824569702,
      "learning_rate": 4.358597553958151e-06,
      "loss": 0.5528,
      "step": 103650
    },
    {
      "epoch": 42.43,
      "grad_norm": 2.0010721683502197,
      "learning_rate": 4.358516513017558e-06,
      "loss": 0.5357,
      "step": 103660
    },
    {
      "epoch": 42.44,
      "grad_norm": 2.0748331546783447,
      "learning_rate": 4.3584354628476315e-06,
      "loss": 0.5438,
      "step": 103670
    },
    {
      "epoch": 42.44,
      "grad_norm": 2.1385767459869385,
      "learning_rate": 4.358354403448743e-06,
      "loss": 0.5423,
      "step": 103680
    },
    {
      "epoch": 42.44,
      "grad_norm": 3.0309135913848877,
      "learning_rate": 4.358273334821264e-06,
      "loss": 0.5495,
      "step": 103690
    },
    {
      "epoch": 42.45,
      "grad_norm": 1.8302137851715088,
      "learning_rate": 4.358192256965565e-06,
      "loss": 0.5833,
      "step": 103700
    },
    {
      "epoch": 42.45,
      "grad_norm": 1.8646281957626343,
      "learning_rate": 4.358111169882018e-06,
      "loss": 0.5724,
      "step": 103710
    },
    {
      "epoch": 42.46,
      "grad_norm": 2.072065830230713,
      "learning_rate": 4.358030073570995e-06,
      "loss": 0.5486,
      "step": 103720
    },
    {
      "epoch": 42.46,
      "grad_norm": 2.326615333557129,
      "learning_rate": 4.3579489680328666e-06,
      "loss": 0.5392,
      "step": 103730
    },
    {
      "epoch": 42.46,
      "grad_norm": 2.5910143852233887,
      "learning_rate": 4.357867853268004e-06,
      "loss": 0.5781,
      "step": 103740
    },
    {
      "epoch": 42.47,
      "grad_norm": 2.251269817352295,
      "learning_rate": 4.35778672927678e-06,
      "loss": 0.5545,
      "step": 103750
    },
    {
      "epoch": 42.47,
      "grad_norm": 1.8179733753204346,
      "learning_rate": 4.357705596059566e-06,
      "loss": 0.538,
      "step": 103760
    },
    {
      "epoch": 42.48,
      "grad_norm": 1.6705600023269653,
      "learning_rate": 4.357624453616733e-06,
      "loss": 0.5379,
      "step": 103770
    },
    {
      "epoch": 42.48,
      "grad_norm": 2.1743781566619873,
      "learning_rate": 4.357543301948654e-06,
      "loss": 0.5375,
      "step": 103780
    },
    {
      "epoch": 42.48,
      "grad_norm": 2.0746371746063232,
      "learning_rate": 4.357462141055698e-06,
      "loss": 0.5636,
      "step": 103790
    },
    {
      "epoch": 42.49,
      "grad_norm": 2.1625728607177734,
      "learning_rate": 4.357380970938239e-06,
      "loss": 0.525,
      "step": 103800
    },
    {
      "epoch": 42.49,
      "grad_norm": 2.3547370433807373,
      "learning_rate": 4.357299791596649e-06,
      "loss": 0.5551,
      "step": 103810
    },
    {
      "epoch": 42.5,
      "grad_norm": 1.6504733562469482,
      "learning_rate": 4.357218603031299e-06,
      "loss": 0.555,
      "step": 103820
    },
    {
      "epoch": 42.5,
      "grad_norm": 1.6946985721588135,
      "learning_rate": 4.357137405242562e-06,
      "loss": 0.5614,
      "step": 103830
    },
    {
      "epoch": 42.51,
      "grad_norm": 1.823730230331421,
      "learning_rate": 4.357056198230808e-06,
      "loss": 0.5369,
      "step": 103840
    },
    {
      "epoch": 42.51,
      "grad_norm": 2.0856916904449463,
      "learning_rate": 4.3569749819964095e-06,
      "loss": 0.5528,
      "step": 103850
    },
    {
      "epoch": 42.51,
      "grad_norm": 2.385606527328491,
      "learning_rate": 4.356893756539741e-06,
      "loss": 0.5662,
      "step": 103860
    },
    {
      "epoch": 42.52,
      "grad_norm": 2.151432752609253,
      "learning_rate": 4.356812521861171e-06,
      "loss": 0.5613,
      "step": 103870
    },
    {
      "epoch": 42.52,
      "grad_norm": 2.2595527172088623,
      "learning_rate": 4.356731277961074e-06,
      "loss": 0.5432,
      "step": 103880
    },
    {
      "epoch": 42.53,
      "grad_norm": 2.4208922386169434,
      "learning_rate": 4.3566500248398214e-06,
      "loss": 0.5465,
      "step": 103890
    },
    {
      "epoch": 42.53,
      "grad_norm": 2.4392502307891846,
      "learning_rate": 4.356568762497786e-06,
      "loss": 0.5411,
      "step": 103900
    },
    {
      "epoch": 42.53,
      "grad_norm": 1.5483876466751099,
      "learning_rate": 4.356487490935339e-06,
      "loss": 0.5615,
      "step": 103910
    },
    {
      "epoch": 42.54,
      "grad_norm": 2.1975228786468506,
      "learning_rate": 4.3564062101528545e-06,
      "loss": 0.5336,
      "step": 103920
    },
    {
      "epoch": 42.54,
      "grad_norm": 2.0184555053710938,
      "learning_rate": 4.356324920150702e-06,
      "loss": 0.5579,
      "step": 103930
    },
    {
      "epoch": 42.55,
      "grad_norm": 2.065873861312866,
      "learning_rate": 4.356243620929256e-06,
      "loss": 0.5539,
      "step": 103940
    },
    {
      "epoch": 42.55,
      "grad_norm": 2.1850881576538086,
      "learning_rate": 4.356162312488889e-06,
      "loss": 0.5613,
      "step": 103950
    },
    {
      "epoch": 42.55,
      "grad_norm": 2.077829122543335,
      "learning_rate": 4.356080994829972e-06,
      "loss": 0.5407,
      "step": 103960
    },
    {
      "epoch": 42.56,
      "grad_norm": 2.1799917221069336,
      "learning_rate": 4.355999667952879e-06,
      "loss": 0.5489,
      "step": 103970
    },
    {
      "epoch": 42.56,
      "grad_norm": 2.1136858463287354,
      "learning_rate": 4.3559183318579815e-06,
      "loss": 0.5454,
      "step": 103980
    },
    {
      "epoch": 42.57,
      "grad_norm": 2.4809067249298096,
      "learning_rate": 4.355836986545653e-06,
      "loss": 0.5302,
      "step": 103990
    },
    {
      "epoch": 42.57,
      "grad_norm": 2.4001305103302,
      "learning_rate": 4.3557556320162654e-06,
      "loss": 0.5322,
      "step": 104000
    },
    {
      "epoch": 42.57,
      "grad_norm": 1.7670228481292725,
      "learning_rate": 4.3556742682701916e-06,
      "loss": 0.5473,
      "step": 104010
    },
    {
      "epoch": 42.58,
      "grad_norm": 1.631339430809021,
      "learning_rate": 4.355592895307804e-06,
      "loss": 0.5406,
      "step": 104020
    },
    {
      "epoch": 42.58,
      "grad_norm": 2.5900120735168457,
      "learning_rate": 4.3555115131294755e-06,
      "loss": 0.548,
      "step": 104030
    },
    {
      "epoch": 42.59,
      "grad_norm": 2.8214571475982666,
      "learning_rate": 4.35543012173558e-06,
      "loss": 0.5508,
      "step": 104040
    },
    {
      "epoch": 42.59,
      "grad_norm": 2.2791850566864014,
      "learning_rate": 4.355348721126489e-06,
      "loss": 0.5571,
      "step": 104050
    },
    {
      "epoch": 42.6,
      "grad_norm": 2.7731423377990723,
      "learning_rate": 4.3552673113025756e-06,
      "loss": 0.5506,
      "step": 104060
    },
    {
      "epoch": 42.6,
      "grad_norm": 3.014401435852051,
      "learning_rate": 4.355185892264213e-06,
      "loss": 0.5692,
      "step": 104070
    },
    {
      "epoch": 42.6,
      "grad_norm": 2.3238539695739746,
      "learning_rate": 4.355104464011774e-06,
      "loss": 0.5459,
      "step": 104080
    },
    {
      "epoch": 42.61,
      "grad_norm": 4.370250225067139,
      "learning_rate": 4.355023026545632e-06,
      "loss": 0.5551,
      "step": 104090
    },
    {
      "epoch": 42.61,
      "grad_norm": 2.0881667137145996,
      "learning_rate": 4.354941579866159e-06,
      "loss": 0.5496,
      "step": 104100
    },
    {
      "epoch": 42.62,
      "grad_norm": 2.1645166873931885,
      "learning_rate": 4.354860123973729e-06,
      "loss": 0.5432,
      "step": 104110
    },
    {
      "epoch": 42.62,
      "grad_norm": 1.8333910703659058,
      "learning_rate": 4.354778658868716e-06,
      "loss": 0.537,
      "step": 104120
    },
    {
      "epoch": 42.62,
      "grad_norm": 2.0376546382904053,
      "learning_rate": 4.35469718455149e-06,
      "loss": 0.5505,
      "step": 104130
    },
    {
      "epoch": 42.63,
      "grad_norm": 2.7757413387298584,
      "learning_rate": 4.354615701022429e-06,
      "loss": 0.5535,
      "step": 104140
    },
    {
      "epoch": 42.63,
      "grad_norm": 1.9915199279785156,
      "learning_rate": 4.354534208281902e-06,
      "loss": 0.5558,
      "step": 104150
    },
    {
      "epoch": 42.64,
      "grad_norm": 1.9198294878005981,
      "learning_rate": 4.354452706330285e-06,
      "loss": 0.5441,
      "step": 104160
    },
    {
      "epoch": 42.64,
      "grad_norm": 2.208899736404419,
      "learning_rate": 4.35437119516795e-06,
      "loss": 0.5455,
      "step": 104170
    },
    {
      "epoch": 42.64,
      "grad_norm": 1.8682230710983276,
      "learning_rate": 4.35428967479527e-06,
      "loss": 0.543,
      "step": 104180
    },
    {
      "epoch": 42.65,
      "grad_norm": 2.107853651046753,
      "learning_rate": 4.35420814521262e-06,
      "loss": 0.5463,
      "step": 104190
    },
    {
      "epoch": 42.65,
      "grad_norm": 2.5206046104431152,
      "learning_rate": 4.354126606420372e-06,
      "loss": 0.5439,
      "step": 104200
    },
    {
      "epoch": 42.66,
      "grad_norm": 1.8559825420379639,
      "learning_rate": 4.3540450584189e-06,
      "loss": 0.5652,
      "step": 104210
    },
    {
      "epoch": 42.66,
      "grad_norm": 2.230466365814209,
      "learning_rate": 4.353963501208578e-06,
      "loss": 0.5688,
      "step": 104220
    },
    {
      "epoch": 42.66,
      "grad_norm": 2.3243019580841064,
      "learning_rate": 4.353881934789779e-06,
      "loss": 0.5548,
      "step": 104230
    },
    {
      "epoch": 42.67,
      "grad_norm": 1.931508183479309,
      "learning_rate": 4.3538003591628775e-06,
      "loss": 0.576,
      "step": 104240
    },
    {
      "epoch": 42.67,
      "grad_norm": 2.149968147277832,
      "learning_rate": 4.353718774328246e-06,
      "loss": 0.5467,
      "step": 104250
    },
    {
      "epoch": 42.68,
      "grad_norm": 2.076280355453491,
      "learning_rate": 4.353637180286259e-06,
      "loss": 0.5529,
      "step": 104260
    },
    {
      "epoch": 42.68,
      "grad_norm": 2.6068296432495117,
      "learning_rate": 4.35355557703729e-06,
      "loss": 0.5655,
      "step": 104270
    },
    {
      "epoch": 42.69,
      "grad_norm": 3.4642579555511475,
      "learning_rate": 4.353473964581713e-06,
      "loss": 0.5611,
      "step": 104280
    },
    {
      "epoch": 42.69,
      "grad_norm": 2.629439115524292,
      "learning_rate": 4.3533923429199016e-06,
      "loss": 0.5462,
      "step": 104290
    },
    {
      "epoch": 42.69,
      "grad_norm": 2.1139674186706543,
      "learning_rate": 4.353310712052231e-06,
      "loss": 0.5599,
      "step": 104300
    },
    {
      "epoch": 42.7,
      "grad_norm": 2.2809371948242188,
      "learning_rate": 4.353229071979073e-06,
      "loss": 0.5454,
      "step": 104310
    },
    {
      "epoch": 42.7,
      "grad_norm": 1.9252279996871948,
      "learning_rate": 4.353147422700802e-06,
      "loss": 0.5495,
      "step": 104320
    },
    {
      "epoch": 42.71,
      "grad_norm": 2.012791395187378,
      "learning_rate": 4.353065764217794e-06,
      "loss": 0.5593,
      "step": 104330
    },
    {
      "epoch": 42.71,
      "grad_norm": 1.8258640766143799,
      "learning_rate": 4.35298409653042e-06,
      "loss": 0.5559,
      "step": 104340
    },
    {
      "epoch": 42.71,
      "grad_norm": 2.5004723072052,
      "learning_rate": 4.352902419639057e-06,
      "loss": 0.5591,
      "step": 104350
    },
    {
      "epoch": 42.72,
      "grad_norm": 2.2321949005126953,
      "learning_rate": 4.352820733544078e-06,
      "loss": 0.5708,
      "step": 104360
    },
    {
      "epoch": 42.72,
      "grad_norm": 2.1549220085144043,
      "learning_rate": 4.352739038245857e-06,
      "loss": 0.5655,
      "step": 104370
    },
    {
      "epoch": 42.73,
      "grad_norm": 1.7294414043426514,
      "learning_rate": 4.352657333744768e-06,
      "loss": 0.5349,
      "step": 104380
    },
    {
      "epoch": 42.73,
      "grad_norm": 2.356924057006836,
      "learning_rate": 4.352575620041185e-06,
      "loss": 0.5548,
      "step": 104390
    },
    {
      "epoch": 42.73,
      "grad_norm": 2.027904510498047,
      "learning_rate": 4.352493897135484e-06,
      "loss": 0.5567,
      "step": 104400
    },
    {
      "epoch": 42.74,
      "grad_norm": 2.5202126502990723,
      "learning_rate": 4.3524121650280385e-06,
      "loss": 0.5439,
      "step": 104410
    },
    {
      "epoch": 42.74,
      "grad_norm": 2.2608158588409424,
      "learning_rate": 4.352330423719221e-06,
      "loss": 0.561,
      "step": 104420
    },
    {
      "epoch": 42.75,
      "grad_norm": 2.285729169845581,
      "learning_rate": 4.35224867320941e-06,
      "loss": 0.5419,
      "step": 104430
    },
    {
      "epoch": 42.75,
      "grad_norm": 2.1828155517578125,
      "learning_rate": 4.352166913498976e-06,
      "loss": 0.5587,
      "step": 104440
    },
    {
      "epoch": 42.75,
      "grad_norm": 2.1892106533050537,
      "learning_rate": 4.352085144588295e-06,
      "loss": 0.5451,
      "step": 104450
    },
    {
      "epoch": 42.76,
      "grad_norm": 2.4787437915802,
      "learning_rate": 4.352003366477743e-06,
      "loss": 0.5363,
      "step": 104460
    },
    {
      "epoch": 42.76,
      "grad_norm": 1.9432224035263062,
      "learning_rate": 4.3519215791676924e-06,
      "loss": 0.5598,
      "step": 104470
    },
    {
      "epoch": 42.77,
      "grad_norm": 1.9435245990753174,
      "learning_rate": 4.351839782658519e-06,
      "loss": 0.5597,
      "step": 104480
    },
    {
      "epoch": 42.77,
      "grad_norm": 2.2105538845062256,
      "learning_rate": 4.3517579769505976e-06,
      "loss": 0.547,
      "step": 104490
    },
    {
      "epoch": 42.78,
      "grad_norm": 2.4005208015441895,
      "learning_rate": 4.351676162044302e-06,
      "loss": 0.5226,
      "step": 104500
    },
    {
      "epoch": 42.78,
      "grad_norm": 2.2932546138763428,
      "learning_rate": 4.3515943379400084e-06,
      "loss": 0.5633,
      "step": 104510
    },
    {
      "epoch": 42.78,
      "grad_norm": 2.3473033905029297,
      "learning_rate": 4.3515125046380905e-06,
      "loss": 0.5329,
      "step": 104520
    },
    {
      "epoch": 42.79,
      "grad_norm": 1.9367072582244873,
      "learning_rate": 4.351430662138923e-06,
      "loss": 0.5535,
      "step": 104530
    },
    {
      "epoch": 42.79,
      "grad_norm": 2.112368106842041,
      "learning_rate": 4.351348810442882e-06,
      "loss": 0.5519,
      "step": 104540
    },
    {
      "epoch": 42.8,
      "grad_norm": 2.0629303455352783,
      "learning_rate": 4.3512669495503415e-06,
      "loss": 0.5556,
      "step": 104550
    },
    {
      "epoch": 42.8,
      "grad_norm": 1.9967373609542847,
      "learning_rate": 4.351185079461677e-06,
      "loss": 0.5529,
      "step": 104560
    },
    {
      "epoch": 42.8,
      "grad_norm": 2.7791731357574463,
      "learning_rate": 4.351103200177263e-06,
      "loss": 0.5371,
      "step": 104570
    },
    {
      "epoch": 42.81,
      "grad_norm": 1.855237603187561,
      "learning_rate": 4.351021311697475e-06,
      "loss": 0.5539,
      "step": 104580
    },
    {
      "epoch": 42.81,
      "grad_norm": 2.1558573246002197,
      "learning_rate": 4.350939414022689e-06,
      "loss": 0.5194,
      "step": 104590
    },
    {
      "epoch": 42.82,
      "grad_norm": 1.8953325748443604,
      "learning_rate": 4.350857507153277e-06,
      "loss": 0.5716,
      "step": 104600
    },
    {
      "epoch": 42.82,
      "grad_norm": 2.3424248695373535,
      "learning_rate": 4.350775591089618e-06,
      "loss": 0.5639,
      "step": 104610
    },
    {
      "epoch": 42.82,
      "grad_norm": 2.238318920135498,
      "learning_rate": 4.350693665832084e-06,
      "loss": 0.5416,
      "step": 104620
    },
    {
      "epoch": 42.83,
      "grad_norm": 2.0873911380767822,
      "learning_rate": 4.350611731381054e-06,
      "loss": 0.5448,
      "step": 104630
    },
    {
      "epoch": 42.83,
      "grad_norm": 2.7222695350646973,
      "learning_rate": 4.3505297877369e-06,
      "loss": 0.565,
      "step": 104640
    },
    {
      "epoch": 42.84,
      "grad_norm": 2.1453750133514404,
      "learning_rate": 4.350447834899998e-06,
      "loss": 0.5419,
      "step": 104650
    },
    {
      "epoch": 42.84,
      "grad_norm": 1.834221601486206,
      "learning_rate": 4.350365872870725e-06,
      "loss": 0.554,
      "step": 104660
    },
    {
      "epoch": 42.84,
      "grad_norm": 2.03725004196167,
      "learning_rate": 4.350283901649455e-06,
      "loss": 0.557,
      "step": 104670
    },
    {
      "epoch": 42.85,
      "grad_norm": 2.215298652648926,
      "learning_rate": 4.350201921236565e-06,
      "loss": 0.5476,
      "step": 104680
    },
    {
      "epoch": 42.85,
      "grad_norm": 2.6881110668182373,
      "learning_rate": 4.350119931632428e-06,
      "loss": 0.5436,
      "step": 104690
    },
    {
      "epoch": 42.86,
      "grad_norm": 3.249380588531494,
      "learning_rate": 4.3500379328374215e-06,
      "loss": 0.553,
      "step": 104700
    },
    {
      "epoch": 42.86,
      "grad_norm": 2.33090877532959,
      "learning_rate": 4.349955924851921e-06,
      "loss": 0.5503,
      "step": 104710
    },
    {
      "epoch": 42.87,
      "grad_norm": 2.4503376483917236,
      "learning_rate": 4.3498739076763016e-06,
      "loss": 0.5397,
      "step": 104720
    },
    {
      "epoch": 42.87,
      "grad_norm": 2.1727094650268555,
      "learning_rate": 4.34979188131094e-06,
      "loss": 0.5459,
      "step": 104730
    },
    {
      "epoch": 42.87,
      "grad_norm": 2.217197895050049,
      "learning_rate": 4.349709845756211e-06,
      "loss": 0.5636,
      "step": 104740
    },
    {
      "epoch": 42.88,
      "grad_norm": 2.0321877002716064,
      "learning_rate": 4.34962780101249e-06,
      "loss": 0.5539,
      "step": 104750
    },
    {
      "epoch": 42.88,
      "grad_norm": 2.98907732963562,
      "learning_rate": 4.349545747080153e-06,
      "loss": 0.556,
      "step": 104760
    },
    {
      "epoch": 42.89,
      "grad_norm": 2.7179131507873535,
      "learning_rate": 4.3494636839595775e-06,
      "loss": 0.5543,
      "step": 104770
    },
    {
      "epoch": 42.89,
      "grad_norm": 2.5802524089813232,
      "learning_rate": 4.349381611651138e-06,
      "loss": 0.5472,
      "step": 104780
    },
    {
      "epoch": 42.89,
      "grad_norm": 2.0552783012390137,
      "learning_rate": 4.349299530155209e-06,
      "loss": 0.5538,
      "step": 104790
    },
    {
      "epoch": 42.9,
      "grad_norm": 2.239800453186035,
      "learning_rate": 4.34921743947217e-06,
      "loss": 0.5666,
      "step": 104800
    },
    {
      "epoch": 42.9,
      "grad_norm": 2.1847972869873047,
      "learning_rate": 4.3491353396023946e-06,
      "loss": 0.5549,
      "step": 104810
    },
    {
      "epoch": 42.91,
      "grad_norm": 1.6731722354888916,
      "learning_rate": 4.34905323054626e-06,
      "loss": 0.5468,
      "step": 104820
    },
    {
      "epoch": 42.91,
      "grad_norm": 2.7058372497558594,
      "learning_rate": 4.348971112304142e-06,
      "loss": 0.5621,
      "step": 104830
    },
    {
      "epoch": 42.91,
      "grad_norm": 2.6535842418670654,
      "learning_rate": 4.348888984876416e-06,
      "loss": 0.5382,
      "step": 104840
    },
    {
      "epoch": 42.92,
      "grad_norm": 2.503084659576416,
      "learning_rate": 4.3488068482634595e-06,
      "loss": 0.5627,
      "step": 104850
    },
    {
      "epoch": 42.92,
      "grad_norm": 2.2503280639648438,
      "learning_rate": 4.3487247024656475e-06,
      "loss": 0.5637,
      "step": 104860
    },
    {
      "epoch": 42.93,
      "grad_norm": 2.2432403564453125,
      "learning_rate": 4.348642547483357e-06,
      "loss": 0.5298,
      "step": 104870
    },
    {
      "epoch": 42.93,
      "grad_norm": 1.6469355821609497,
      "learning_rate": 4.348560383316965e-06,
      "loss": 0.5432,
      "step": 104880
    },
    {
      "epoch": 42.93,
      "grad_norm": 2.10941743850708,
      "learning_rate": 4.348478209966846e-06,
      "loss": 0.5302,
      "step": 104890
    },
    {
      "epoch": 42.94,
      "grad_norm": 2.1504292488098145,
      "learning_rate": 4.348396027433378e-06,
      "loss": 0.5608,
      "step": 104900
    },
    {
      "epoch": 42.94,
      "grad_norm": 2.9114010334014893,
      "learning_rate": 4.348313835716937e-06,
      "loss": 0.5454,
      "step": 104910
    },
    {
      "epoch": 42.95,
      "grad_norm": 2.203181743621826,
      "learning_rate": 4.3482316348179e-06,
      "loss": 0.5546,
      "step": 104920
    },
    {
      "epoch": 42.95,
      "grad_norm": 2.4885716438293457,
      "learning_rate": 4.348149424736642e-06,
      "loss": 0.5528,
      "step": 104930
    },
    {
      "epoch": 42.96,
      "grad_norm": 2.046265125274658,
      "learning_rate": 4.348067205473541e-06,
      "loss": 0.5469,
      "step": 104940
    },
    {
      "epoch": 42.96,
      "grad_norm": 2.4884681701660156,
      "learning_rate": 4.347984977028974e-06,
      "loss": 0.5628,
      "step": 104950
    },
    {
      "epoch": 42.96,
      "grad_norm": 2.6999988555908203,
      "learning_rate": 4.347902739403317e-06,
      "loss": 0.5583,
      "step": 104960
    },
    {
      "epoch": 42.97,
      "grad_norm": 2.5527262687683105,
      "learning_rate": 4.347820492596946e-06,
      "loss": 0.5567,
      "step": 104970
    },
    {
      "epoch": 42.97,
      "grad_norm": 2.1673905849456787,
      "learning_rate": 4.347738236610238e-06,
      "loss": 0.5318,
      "step": 104980
    },
    {
      "epoch": 42.98,
      "grad_norm": 1.7513123750686646,
      "learning_rate": 4.347655971443572e-06,
      "loss": 0.5488,
      "step": 104990
    },
    {
      "epoch": 42.98,
      "grad_norm": 2.048619270324707,
      "learning_rate": 4.347573697097321e-06,
      "loss": 0.551,
      "step": 105000
    },
    {
      "epoch": 42.98,
      "grad_norm": 1.5921717882156372,
      "learning_rate": 4.347491413571865e-06,
      "loss": 0.5612,
      "step": 105010
    },
    {
      "epoch": 42.99,
      "grad_norm": 1.9227540493011475,
      "learning_rate": 4.34740912086758e-06,
      "loss": 0.589,
      "step": 105020
    },
    {
      "epoch": 42.99,
      "grad_norm": 1.8870948553085327,
      "learning_rate": 4.347326818984843e-06,
      "loss": 0.5366,
      "step": 105030
    },
    {
      "epoch": 43.0,
      "grad_norm": 1.7581628561019897,
      "learning_rate": 4.34724450792403e-06,
      "loss": 0.5513,
      "step": 105040
    },
    {
      "epoch": 43.0,
      "eval_loss": 0.5513549447059631,
      "eval_runtime": 52.1152,
      "eval_samples_per_second": 66.18,
      "eval_steps_per_second": 8.289,
      "step": 105049
    },
    {
      "epoch": 43.0,
      "grad_norm": 1.8945488929748535,
      "learning_rate": 4.34716218768552e-06,
      "loss": 0.5393,
      "step": 105050
    },
    {
      "epoch": 43.0,
      "grad_norm": 1.5647331476211548,
      "learning_rate": 4.347079858269688e-06,
      "loss": 0.5579,
      "step": 105060
    },
    {
      "epoch": 43.01,
      "grad_norm": 2.359421730041504,
      "learning_rate": 4.346997519676913e-06,
      "loss": 0.5475,
      "step": 105070
    },
    {
      "epoch": 43.01,
      "grad_norm": 1.8757340908050537,
      "learning_rate": 4.346915171907571e-06,
      "loss": 0.559,
      "step": 105080
    },
    {
      "epoch": 43.02,
      "grad_norm": 1.9021663665771484,
      "learning_rate": 4.34683281496204e-06,
      "loss": 0.5455,
      "step": 105090
    },
    {
      "epoch": 43.02,
      "grad_norm": 1.8423805236816406,
      "learning_rate": 4.346750448840697e-06,
      "loss": 0.5587,
      "step": 105100
    },
    {
      "epoch": 43.02,
      "grad_norm": 1.9120105504989624,
      "learning_rate": 4.346668073543918e-06,
      "loss": 0.5542,
      "step": 105110
    },
    {
      "epoch": 43.03,
      "grad_norm": 1.6214053630828857,
      "learning_rate": 4.346585689072083e-06,
      "loss": 0.5759,
      "step": 105120
    },
    {
      "epoch": 43.03,
      "grad_norm": 2.664010524749756,
      "learning_rate": 4.346503295425567e-06,
      "loss": 0.5462,
      "step": 105130
    },
    {
      "epoch": 43.04,
      "grad_norm": 2.092190980911255,
      "learning_rate": 4.346420892604748e-06,
      "loss": 0.568,
      "step": 105140
    },
    {
      "epoch": 43.04,
      "grad_norm": 2.3198843002319336,
      "learning_rate": 4.346338480610004e-06,
      "loss": 0.5383,
      "step": 105150
    },
    {
      "epoch": 43.05,
      "grad_norm": 2.5361602306365967,
      "learning_rate": 4.346256059441713e-06,
      "loss": 0.5839,
      "step": 105160
    },
    {
      "epoch": 43.05,
      "grad_norm": 1.7886834144592285,
      "learning_rate": 4.3461736291002505e-06,
      "loss": 0.5373,
      "step": 105170
    },
    {
      "epoch": 43.05,
      "grad_norm": 2.576643705368042,
      "learning_rate": 4.346091189585997e-06,
      "loss": 0.5496,
      "step": 105180
    },
    {
      "epoch": 43.06,
      "grad_norm": 2.2741453647613525,
      "learning_rate": 4.346008740899327e-06,
      "loss": 0.5555,
      "step": 105190
    },
    {
      "epoch": 43.06,
      "grad_norm": 1.9805856943130493,
      "learning_rate": 4.345926283040622e-06,
      "loss": 0.5214,
      "step": 105200
    },
    {
      "epoch": 43.07,
      "grad_norm": 2.806605339050293,
      "learning_rate": 4.3458438160102555e-06,
      "loss": 0.5607,
      "step": 105210
    },
    {
      "epoch": 43.07,
      "grad_norm": 1.936253547668457,
      "learning_rate": 4.345761339808609e-06,
      "loss": 0.5461,
      "step": 105220
    },
    {
      "epoch": 43.07,
      "grad_norm": 1.6545953750610352,
      "learning_rate": 4.345678854436057e-06,
      "loss": 0.5332,
      "step": 105230
    },
    {
      "epoch": 43.08,
      "grad_norm": 2.4622809886932373,
      "learning_rate": 4.34559635989298e-06,
      "loss": 0.5677,
      "step": 105240
    },
    {
      "epoch": 43.08,
      "grad_norm": 2.311042547225952,
      "learning_rate": 4.3455138561797535e-06,
      "loss": 0.561,
      "step": 105250
    },
    {
      "epoch": 43.09,
      "grad_norm": 2.3140501976013184,
      "learning_rate": 4.345431343296758e-06,
      "loss": 0.546,
      "step": 105260
    },
    {
      "epoch": 43.09,
      "grad_norm": 2.0602433681488037,
      "learning_rate": 4.34534882124437e-06,
      "loss": 0.5522,
      "step": 105270
    },
    {
      "epoch": 43.09,
      "grad_norm": 2.942735195159912,
      "learning_rate": 4.345266290022968e-06,
      "loss": 0.5558,
      "step": 105280
    },
    {
      "epoch": 43.1,
      "grad_norm": 1.8046560287475586,
      "learning_rate": 4.34518374963293e-06,
      "loss": 0.5794,
      "step": 105290
    },
    {
      "epoch": 43.1,
      "grad_norm": 2.438645839691162,
      "learning_rate": 4.345101200074633e-06,
      "loss": 0.5653,
      "step": 105300
    },
    {
      "epoch": 43.11,
      "grad_norm": 2.878575086593628,
      "learning_rate": 4.345018641348457e-06,
      "loss": 0.541,
      "step": 105310
    },
    {
      "epoch": 43.11,
      "grad_norm": 2.3311116695404053,
      "learning_rate": 4.344936073454779e-06,
      "loss": 0.5507,
      "step": 105320
    },
    {
      "epoch": 43.12,
      "grad_norm": 2.425227165222168,
      "learning_rate": 4.3448534963939774e-06,
      "loss": 0.535,
      "step": 105330
    },
    {
      "epoch": 43.12,
      "grad_norm": 1.7720941305160522,
      "learning_rate": 4.344770910166431e-06,
      "loss": 0.5572,
      "step": 105340
    },
    {
      "epoch": 43.12,
      "grad_norm": 2.490846872329712,
      "learning_rate": 4.344688314772517e-06,
      "loss": 0.5514,
      "step": 105350
    },
    {
      "epoch": 43.13,
      "grad_norm": 2.4348714351654053,
      "learning_rate": 4.344605710212615e-06,
      "loss": 0.5531,
      "step": 105360
    },
    {
      "epoch": 43.13,
      "grad_norm": 1.98379647731781,
      "learning_rate": 4.344523096487103e-06,
      "loss": 0.555,
      "step": 105370
    },
    {
      "epoch": 43.14,
      "grad_norm": 2.0298097133636475,
      "learning_rate": 4.344440473596359e-06,
      "loss": 0.5611,
      "step": 105380
    },
    {
      "epoch": 43.14,
      "grad_norm": 2.217482805252075,
      "learning_rate": 4.344357841540761e-06,
      "loss": 0.5526,
      "step": 105390
    },
    {
      "epoch": 43.14,
      "grad_norm": 2.8979785442352295,
      "learning_rate": 4.34427520032069e-06,
      "loss": 0.5252,
      "step": 105400
    },
    {
      "epoch": 43.15,
      "grad_norm": 3.3753011226654053,
      "learning_rate": 4.344192549936522e-06,
      "loss": 0.5645,
      "step": 105410
    },
    {
      "epoch": 43.15,
      "grad_norm": 2.4175915718078613,
      "learning_rate": 4.344109890388636e-06,
      "loss": 0.5588,
      "step": 105420
    },
    {
      "epoch": 43.16,
      "grad_norm": 1.7432974576950073,
      "learning_rate": 4.344027221677412e-06,
      "loss": 0.5423,
      "step": 105430
    },
    {
      "epoch": 43.16,
      "grad_norm": 2.034615993499756,
      "learning_rate": 4.343944543803227e-06,
      "loss": 0.5233,
      "step": 105440
    },
    {
      "epoch": 43.16,
      "grad_norm": 2.7238166332244873,
      "learning_rate": 4.343861856766461e-06,
      "loss": 0.5576,
      "step": 105450
    },
    {
      "epoch": 43.17,
      "grad_norm": 1.5142265558242798,
      "learning_rate": 4.343779160567493e-06,
      "loss": 0.5618,
      "step": 105460
    },
    {
      "epoch": 43.17,
      "grad_norm": 2.111295223236084,
      "learning_rate": 4.3436964552067006e-06,
      "loss": 0.5494,
      "step": 105470
    },
    {
      "epoch": 43.18,
      "grad_norm": 2.774644374847412,
      "learning_rate": 4.343613740684463e-06,
      "loss": 0.5467,
      "step": 105480
    },
    {
      "epoch": 43.18,
      "grad_norm": 1.9513194561004639,
      "learning_rate": 4.34353101700116e-06,
      "loss": 0.5559,
      "step": 105490
    },
    {
      "epoch": 43.18,
      "grad_norm": 1.613394021987915,
      "learning_rate": 4.343448284157169e-06,
      "loss": 0.5411,
      "step": 105500
    },
    {
      "epoch": 43.19,
      "grad_norm": 1.9423842430114746,
      "learning_rate": 4.3433655421528695e-06,
      "loss": 0.5506,
      "step": 105510
    },
    {
      "epoch": 43.19,
      "grad_norm": 2.804666042327881,
      "learning_rate": 4.343282790988642e-06,
      "loss": 0.5505,
      "step": 105520
    },
    {
      "epoch": 43.2,
      "grad_norm": 2.3919193744659424,
      "learning_rate": 4.343200030664864e-06,
      "loss": 0.5518,
      "step": 105530
    },
    {
      "epoch": 43.2,
      "grad_norm": 1.8602100610733032,
      "learning_rate": 4.3431172611819145e-06,
      "loss": 0.5506,
      "step": 105540
    },
    {
      "epoch": 43.21,
      "grad_norm": 2.0942540168762207,
      "learning_rate": 4.343034482540174e-06,
      "loss": 0.5286,
      "step": 105550
    },
    {
      "epoch": 43.21,
      "grad_norm": 1.8856678009033203,
      "learning_rate": 4.3429516947400206e-06,
      "loss": 0.5405,
      "step": 105560
    },
    {
      "epoch": 43.21,
      "grad_norm": 1.6207712888717651,
      "learning_rate": 4.342868897781834e-06,
      "loss": 0.5436,
      "step": 105570
    },
    {
      "epoch": 43.22,
      "grad_norm": 1.8880157470703125,
      "learning_rate": 4.342786091665994e-06,
      "loss": 0.5526,
      "step": 105580
    },
    {
      "epoch": 43.22,
      "grad_norm": 2.7539138793945312,
      "learning_rate": 4.342703276392878e-06,
      "loss": 0.5579,
      "step": 105590
    },
    {
      "epoch": 43.23,
      "grad_norm": 2.084388494491577,
      "learning_rate": 4.3426204519628676e-06,
      "loss": 0.5478,
      "step": 105600
    },
    {
      "epoch": 43.23,
      "grad_norm": 2.9444589614868164,
      "learning_rate": 4.34253761837634e-06,
      "loss": 0.5694,
      "step": 105610
    },
    {
      "epoch": 43.23,
      "grad_norm": 3.4926066398620605,
      "learning_rate": 4.3424547756336775e-06,
      "loss": 0.5623,
      "step": 105620
    },
    {
      "epoch": 43.24,
      "grad_norm": 2.4587128162384033,
      "learning_rate": 4.342371923735257e-06,
      "loss": 0.5394,
      "step": 105630
    },
    {
      "epoch": 43.24,
      "grad_norm": 2.0304038524627686,
      "learning_rate": 4.342289062681459e-06,
      "loss": 0.5623,
      "step": 105640
    },
    {
      "epoch": 43.25,
      "grad_norm": 2.7818007469177246,
      "learning_rate": 4.342206192472664e-06,
      "loss": 0.5595,
      "step": 105650
    },
    {
      "epoch": 43.25,
      "grad_norm": 1.9084720611572266,
      "learning_rate": 4.342123313109249e-06,
      "loss": 0.574,
      "step": 105660
    },
    {
      "epoch": 43.25,
      "grad_norm": 2.4914231300354004,
      "learning_rate": 4.342040424591596e-06,
      "loss": 0.5372,
      "step": 105670
    },
    {
      "epoch": 43.26,
      "grad_norm": 2.4242265224456787,
      "learning_rate": 4.341957526920084e-06,
      "loss": 0.5627,
      "step": 105680
    },
    {
      "epoch": 43.26,
      "grad_norm": 2.031296491622925,
      "learning_rate": 4.3418746200950936e-06,
      "loss": 0.5335,
      "step": 105690
    },
    {
      "epoch": 43.27,
      "grad_norm": 2.3370325565338135,
      "learning_rate": 4.341791704117003e-06,
      "loss": 0.5543,
      "step": 105700
    },
    {
      "epoch": 43.27,
      "grad_norm": 2.1985585689544678,
      "learning_rate": 4.3417087789861926e-06,
      "loss": 0.5654,
      "step": 105710
    },
    {
      "epoch": 43.27,
      "grad_norm": 2.2311134338378906,
      "learning_rate": 4.3416258447030435e-06,
      "loss": 0.5632,
      "step": 105720
    },
    {
      "epoch": 43.28,
      "grad_norm": 2.5465147495269775,
      "learning_rate": 4.341542901267934e-06,
      "loss": 0.5515,
      "step": 105730
    },
    {
      "epoch": 43.28,
      "grad_norm": 1.9520108699798584,
      "learning_rate": 4.341459948681245e-06,
      "loss": 0.5434,
      "step": 105740
    },
    {
      "epoch": 43.29,
      "grad_norm": 2.0506601333618164,
      "learning_rate": 4.341376986943355e-06,
      "loss": 0.5594,
      "step": 105750
    },
    {
      "epoch": 43.29,
      "grad_norm": 2.4093122482299805,
      "learning_rate": 4.341294016054646e-06,
      "loss": 0.5635,
      "step": 105760
    },
    {
      "epoch": 43.3,
      "grad_norm": 1.9182056188583374,
      "learning_rate": 4.341211036015497e-06,
      "loss": 0.5376,
      "step": 105770
    },
    {
      "epoch": 43.3,
      "grad_norm": 2.122018575668335,
      "learning_rate": 4.341128046826288e-06,
      "loss": 0.5641,
      "step": 105780
    },
    {
      "epoch": 43.3,
      "grad_norm": 2.722930908203125,
      "learning_rate": 4.3410450484874e-06,
      "loss": 0.5581,
      "step": 105790
    },
    {
      "epoch": 43.31,
      "grad_norm": 2.4172680377960205,
      "learning_rate": 4.3409620409992126e-06,
      "loss": 0.5286,
      "step": 105800
    },
    {
      "epoch": 43.31,
      "grad_norm": 1.7722753286361694,
      "learning_rate": 4.340879024362106e-06,
      "loss": 0.5636,
      "step": 105810
    },
    {
      "epoch": 43.32,
      "grad_norm": 2.500824451446533,
      "learning_rate": 4.34079599857646e-06,
      "loss": 0.5429,
      "step": 105820
    },
    {
      "epoch": 43.32,
      "grad_norm": 3.2940473556518555,
      "learning_rate": 4.340712963642657e-06,
      "loss": 0.5524,
      "step": 105830
    },
    {
      "epoch": 43.32,
      "grad_norm": 2.6627490520477295,
      "learning_rate": 4.340629919561075e-06,
      "loss": 0.5441,
      "step": 105840
    },
    {
      "epoch": 43.33,
      "grad_norm": 1.8284521102905273,
      "learning_rate": 4.340546866332096e-06,
      "loss": 0.5533,
      "step": 105850
    },
    {
      "epoch": 43.33,
      "grad_norm": 2.080206871032715,
      "learning_rate": 4.340463803956099e-06,
      "loss": 0.5497,
      "step": 105860
    },
    {
      "epoch": 43.34,
      "grad_norm": 1.9274343252182007,
      "learning_rate": 4.340380732433466e-06,
      "loss": 0.5275,
      "step": 105870
    },
    {
      "epoch": 43.34,
      "grad_norm": 2.3494949340820312,
      "learning_rate": 4.340297651764576e-06,
      "loss": 0.5373,
      "step": 105880
    },
    {
      "epoch": 43.34,
      "grad_norm": 1.8510836362838745,
      "learning_rate": 4.340214561949811e-06,
      "loss": 0.5563,
      "step": 105890
    },
    {
      "epoch": 43.35,
      "grad_norm": 1.9602168798446655,
      "learning_rate": 4.340131462989551e-06,
      "loss": 0.5139,
      "step": 105900
    },
    {
      "epoch": 43.35,
      "grad_norm": 1.9786430597305298,
      "learning_rate": 4.340048354884176e-06,
      "loss": 0.5492,
      "step": 105910
    },
    {
      "epoch": 43.36,
      "grad_norm": 2.538933038711548,
      "learning_rate": 4.3399652376340685e-06,
      "loss": 0.5646,
      "step": 105920
    },
    {
      "epoch": 43.36,
      "grad_norm": 1.4968512058258057,
      "learning_rate": 4.339882111239608e-06,
      "loss": 0.5484,
      "step": 105930
    },
    {
      "epoch": 43.36,
      "grad_norm": 1.7644431591033936,
      "learning_rate": 4.3397989757011744e-06,
      "loss": 0.556,
      "step": 105940
    },
    {
      "epoch": 43.37,
      "grad_norm": 2.782944440841675,
      "learning_rate": 4.339715831019151e-06,
      "loss": 0.553,
      "step": 105950
    },
    {
      "epoch": 43.37,
      "grad_norm": 2.0121543407440186,
      "learning_rate": 4.339632677193916e-06,
      "loss": 0.5412,
      "step": 105960
    },
    {
      "epoch": 43.38,
      "grad_norm": 2.251657247543335,
      "learning_rate": 4.339549514225852e-06,
      "loss": 0.5534,
      "step": 105970
    },
    {
      "epoch": 43.38,
      "grad_norm": 2.7540197372436523,
      "learning_rate": 4.339466342115339e-06,
      "loss": 0.5616,
      "step": 105980
    },
    {
      "epoch": 43.39,
      "grad_norm": 1.8542797565460205,
      "learning_rate": 4.33938316086276e-06,
      "loss": 0.5577,
      "step": 105990
    },
    {
      "epoch": 43.39,
      "grad_norm": 2.614039182662964,
      "learning_rate": 4.339299970468494e-06,
      "loss": 0.5585,
      "step": 106000
    },
    {
      "epoch": 43.39,
      "grad_norm": 2.026064395904541,
      "learning_rate": 4.3392167709329225e-06,
      "loss": 0.5448,
      "step": 106010
    },
    {
      "epoch": 43.4,
      "grad_norm": 2.9453108310699463,
      "learning_rate": 4.339133562256426e-06,
      "loss": 0.5835,
      "step": 106020
    },
    {
      "epoch": 43.4,
      "grad_norm": 1.8620723485946655,
      "learning_rate": 4.339050344439387e-06,
      "loss": 0.5653,
      "step": 106030
    },
    {
      "epoch": 43.41,
      "grad_norm": 2.1242318153381348,
      "learning_rate": 4.3389671174821866e-06,
      "loss": 0.5584,
      "step": 106040
    },
    {
      "epoch": 43.41,
      "grad_norm": 1.9102814197540283,
      "learning_rate": 4.338883881385205e-06,
      "loss": 0.5606,
      "step": 106050
    },
    {
      "epoch": 43.41,
      "grad_norm": 2.29653263092041,
      "learning_rate": 4.338800636148824e-06,
      "loss": 0.5413,
      "step": 106060
    },
    {
      "epoch": 43.42,
      "grad_norm": 2.247443437576294,
      "learning_rate": 4.338717381773425e-06,
      "loss": 0.5479,
      "step": 106070
    },
    {
      "epoch": 43.42,
      "grad_norm": 2.2194507122039795,
      "learning_rate": 4.338634118259389e-06,
      "loss": 0.5458,
      "step": 106080
    },
    {
      "epoch": 43.43,
      "grad_norm": 2.251497745513916,
      "learning_rate": 4.338550845607099e-06,
      "loss": 0.548,
      "step": 106090
    },
    {
      "epoch": 43.43,
      "grad_norm": 1.7387555837631226,
      "learning_rate": 4.338467563816935e-06,
      "loss": 0.5664,
      "step": 106100
    },
    {
      "epoch": 43.43,
      "grad_norm": 1.8787070512771606,
      "learning_rate": 4.338384272889278e-06,
      "loss": 0.5362,
      "step": 106110
    },
    {
      "epoch": 43.44,
      "grad_norm": 1.968713641166687,
      "learning_rate": 4.3383009728245105e-06,
      "loss": 0.5674,
      "step": 106120
    },
    {
      "epoch": 43.44,
      "grad_norm": 2.2412025928497314,
      "learning_rate": 4.3382176636230135e-06,
      "loss": 0.5526,
      "step": 106130
    },
    {
      "epoch": 43.45,
      "grad_norm": 1.7029818296432495,
      "learning_rate": 4.33813434528517e-06,
      "loss": 0.543,
      "step": 106140
    },
    {
      "epoch": 43.45,
      "grad_norm": 1.6731222867965698,
      "learning_rate": 4.338051017811361e-06,
      "loss": 0.5539,
      "step": 106150
    },
    {
      "epoch": 43.45,
      "grad_norm": 2.2178821563720703,
      "learning_rate": 4.337967681201966e-06,
      "loss": 0.5312,
      "step": 106160
    },
    {
      "epoch": 43.46,
      "grad_norm": 2.701479911804199,
      "learning_rate": 4.337884335457371e-06,
      "loss": 0.5671,
      "step": 106170
    },
    {
      "epoch": 43.46,
      "grad_norm": 2.070652484893799,
      "learning_rate": 4.337800980577954e-06,
      "loss": 0.5596,
      "step": 106180
    },
    {
      "epoch": 43.47,
      "grad_norm": 3.384176731109619,
      "learning_rate": 4.337717616564099e-06,
      "loss": 0.5467,
      "step": 106190
    },
    {
      "epoch": 43.47,
      "grad_norm": 1.9795973300933838,
      "learning_rate": 4.337634243416186e-06,
      "loss": 0.5619,
      "step": 106200
    },
    {
      "epoch": 43.48,
      "grad_norm": 2.05983304977417,
      "learning_rate": 4.337550861134599e-06,
      "loss": 0.543,
      "step": 106210
    },
    {
      "epoch": 43.48,
      "grad_norm": 2.6576929092407227,
      "learning_rate": 4.337467469719719e-06,
      "loss": 0.5442,
      "step": 106220
    },
    {
      "epoch": 43.48,
      "grad_norm": 2.2208051681518555,
      "learning_rate": 4.337384069171927e-06,
      "loss": 0.5597,
      "step": 106230
    },
    {
      "epoch": 43.49,
      "grad_norm": 2.60244083404541,
      "learning_rate": 4.337300659491607e-06,
      "loss": 0.5572,
      "step": 106240
    },
    {
      "epoch": 43.49,
      "grad_norm": 2.45005464553833,
      "learning_rate": 4.3372172406791405e-06,
      "loss": 0.5273,
      "step": 106250
    },
    {
      "epoch": 43.5,
      "grad_norm": 2.322950601577759,
      "learning_rate": 4.3371338127349084e-06,
      "loss": 0.5301,
      "step": 106260
    },
    {
      "epoch": 43.5,
      "grad_norm": 2.6298000812530518,
      "learning_rate": 4.3370503756592945e-06,
      "loss": 0.5587,
      "step": 106270
    },
    {
      "epoch": 43.5,
      "grad_norm": 2.3884260654449463,
      "learning_rate": 4.33696692945268e-06,
      "loss": 0.5587,
      "step": 106280
    },
    {
      "epoch": 43.51,
      "grad_norm": 2.0567774772644043,
      "learning_rate": 4.336883474115448e-06,
      "loss": 0.535,
      "step": 106290
    },
    {
      "epoch": 43.51,
      "grad_norm": 2.276702404022217,
      "learning_rate": 4.336800009647979e-06,
      "loss": 0.5501,
      "step": 106300
    },
    {
      "epoch": 43.52,
      "grad_norm": 2.1474056243896484,
      "learning_rate": 4.336716536050658e-06,
      "loss": 0.5637,
      "step": 106310
    },
    {
      "epoch": 43.52,
      "grad_norm": 2.203441619873047,
      "learning_rate": 4.336633053323864e-06,
      "loss": 0.5682,
      "step": 106320
    },
    {
      "epoch": 43.52,
      "grad_norm": 2.058638572692871,
      "learning_rate": 4.336549561467983e-06,
      "loss": 0.5553,
      "step": 106330
    },
    {
      "epoch": 43.53,
      "grad_norm": 2.078442335128784,
      "learning_rate": 4.336466060483396e-06,
      "loss": 0.5411,
      "step": 106340
    },
    {
      "epoch": 43.53,
      "grad_norm": 2.1921889781951904,
      "learning_rate": 4.336382550370485e-06,
      "loss": 0.558,
      "step": 106350
    },
    {
      "epoch": 43.54,
      "grad_norm": 2.5808684825897217,
      "learning_rate": 4.336299031129632e-06,
      "loss": 0.539,
      "step": 106360
    },
    {
      "epoch": 43.54,
      "grad_norm": 2.5281474590301514,
      "learning_rate": 4.3362155027612215e-06,
      "loss": 0.5653,
      "step": 106370
    },
    {
      "epoch": 43.54,
      "grad_norm": 1.9313349723815918,
      "learning_rate": 4.336131965265634e-06,
      "loss": 0.5549,
      "step": 106380
    },
    {
      "epoch": 43.55,
      "grad_norm": 1.7395073175430298,
      "learning_rate": 4.336048418643254e-06,
      "loss": 0.541,
      "step": 106390
    },
    {
      "epoch": 43.55,
      "grad_norm": 2.600745916366577,
      "learning_rate": 4.335964862894464e-06,
      "loss": 0.5469,
      "step": 106400
    },
    {
      "epoch": 43.56,
      "grad_norm": 2.77972149848938,
      "learning_rate": 4.335881298019646e-06,
      "loss": 0.5665,
      "step": 106410
    },
    {
      "epoch": 43.56,
      "grad_norm": 1.8843023777008057,
      "learning_rate": 4.335797724019183e-06,
      "loss": 0.5499,
      "step": 106420
    },
    {
      "epoch": 43.57,
      "grad_norm": 2.2189176082611084,
      "learning_rate": 4.335714140893457e-06,
      "loss": 0.5363,
      "step": 106430
    },
    {
      "epoch": 43.57,
      "grad_norm": 1.6918905973434448,
      "learning_rate": 4.335630548642852e-06,
      "loss": 0.5479,
      "step": 106440
    },
    {
      "epoch": 43.57,
      "grad_norm": 1.6777100563049316,
      "learning_rate": 4.335546947267752e-06,
      "loss": 0.5568,
      "step": 106450
    },
    {
      "epoch": 43.58,
      "grad_norm": 2.734445095062256,
      "learning_rate": 4.335463336768537e-06,
      "loss": 0.5725,
      "step": 106460
    },
    {
      "epoch": 43.58,
      "grad_norm": 2.0479376316070557,
      "learning_rate": 4.335379717145592e-06,
      "loss": 0.5418,
      "step": 106470
    },
    {
      "epoch": 43.59,
      "grad_norm": 1.677834391593933,
      "learning_rate": 4.335296088399301e-06,
      "loss": 0.5381,
      "step": 106480
    },
    {
      "epoch": 43.59,
      "grad_norm": 3.0798747539520264,
      "learning_rate": 4.335212450530045e-06,
      "loss": 0.5423,
      "step": 106490
    },
    {
      "epoch": 43.59,
      "grad_norm": 2.3005595207214355,
      "learning_rate": 4.335128803538208e-06,
      "loss": 0.5509,
      "step": 106500
    },
    {
      "epoch": 43.6,
      "grad_norm": 2.2900891304016113,
      "learning_rate": 4.3350451474241725e-06,
      "loss": 0.5389,
      "step": 106510
    },
    {
      "epoch": 43.6,
      "grad_norm": 2.917402744293213,
      "learning_rate": 4.334961482188323e-06,
      "loss": 0.5594,
      "step": 106520
    },
    {
      "epoch": 43.61,
      "grad_norm": 2.4961962699890137,
      "learning_rate": 4.334877807831042e-06,
      "loss": 0.5578,
      "step": 106530
    },
    {
      "epoch": 43.61,
      "grad_norm": 2.3583831787109375,
      "learning_rate": 4.334794124352713e-06,
      "loss": 0.5529,
      "step": 106540
    },
    {
      "epoch": 43.61,
      "grad_norm": 1.8875503540039062,
      "learning_rate": 4.334710431753719e-06,
      "loss": 0.5519,
      "step": 106550
    },
    {
      "epoch": 43.62,
      "grad_norm": 1.7875564098358154,
      "learning_rate": 4.334626730034443e-06,
      "loss": 0.5526,
      "step": 106560
    },
    {
      "epoch": 43.62,
      "grad_norm": 1.7887693643569946,
      "learning_rate": 4.334543019195269e-06,
      "loss": 0.5463,
      "step": 106570
    },
    {
      "epoch": 43.63,
      "grad_norm": 2.245974540710449,
      "learning_rate": 4.334459299236582e-06,
      "loss": 0.5528,
      "step": 106580
    },
    {
      "epoch": 43.63,
      "grad_norm": 2.2521767616271973,
      "learning_rate": 4.334375570158763e-06,
      "loss": 0.5458,
      "step": 106590
    },
    {
      "epoch": 43.63,
      "grad_norm": 2.2756707668304443,
      "learning_rate": 4.334291831962196e-06,
      "loss": 0.5349,
      "step": 106600
    },
    {
      "epoch": 43.64,
      "grad_norm": 1.9267140626907349,
      "learning_rate": 4.334208084647265e-06,
      "loss": 0.5608,
      "step": 106610
    },
    {
      "epoch": 43.64,
      "grad_norm": 1.9978264570236206,
      "learning_rate": 4.334124328214356e-06,
      "loss": 0.5409,
      "step": 106620
    },
    {
      "epoch": 43.65,
      "grad_norm": 1.7450612783432007,
      "learning_rate": 4.334040562663848e-06,
      "loss": 0.5439,
      "step": 106630
    },
    {
      "epoch": 43.65,
      "grad_norm": 2.281721353530884,
      "learning_rate": 4.333956787996129e-06,
      "loss": 0.5553,
      "step": 106640
    },
    {
      "epoch": 43.66,
      "grad_norm": 2.437582015991211,
      "learning_rate": 4.3338730042115795e-06,
      "loss": 0.5284,
      "step": 106650
    },
    {
      "epoch": 43.66,
      "grad_norm": 1.832450270652771,
      "learning_rate": 4.333789211310586e-06,
      "loss": 0.5544,
      "step": 106660
    },
    {
      "epoch": 43.66,
      "grad_norm": 2.5195424556732178,
      "learning_rate": 4.33370540929353e-06,
      "loss": 0.5489,
      "step": 106670
    },
    {
      "epoch": 43.67,
      "grad_norm": 2.4966886043548584,
      "learning_rate": 4.3336215981607974e-06,
      "loss": 0.5451,
      "step": 106680
    },
    {
      "epoch": 43.67,
      "grad_norm": 2.492004156112671,
      "learning_rate": 4.3335377779127706e-06,
      "loss": 0.5294,
      "step": 106690
    },
    {
      "epoch": 43.68,
      "grad_norm": 2.704926013946533,
      "learning_rate": 4.333453948549834e-06,
      "loss": 0.5503,
      "step": 106700
    },
    {
      "epoch": 43.68,
      "grad_norm": 1.9289376735687256,
      "learning_rate": 4.3333701100723714e-06,
      "loss": 0.5453,
      "step": 106710
    },
    {
      "epoch": 43.68,
      "grad_norm": 2.1084976196289062,
      "learning_rate": 4.333286262480768e-06,
      "loss": 0.5689,
      "step": 106720
    },
    {
      "epoch": 43.69,
      "grad_norm": 1.8475368022918701,
      "learning_rate": 4.3332024057754074e-06,
      "loss": 0.5559,
      "step": 106730
    },
    {
      "epoch": 43.69,
      "grad_norm": 2.480288028717041,
      "learning_rate": 4.3331185399566726e-06,
      "loss": 0.5457,
      "step": 106740
    },
    {
      "epoch": 43.7,
      "grad_norm": 2.4299399852752686,
      "learning_rate": 4.333034665024949e-06,
      "loss": 0.5444,
      "step": 106750
    },
    {
      "epoch": 43.7,
      "grad_norm": 2.2881174087524414,
      "learning_rate": 4.33295078098062e-06,
      "loss": 0.5594,
      "step": 106760
    },
    {
      "epoch": 43.7,
      "grad_norm": 2.0268912315368652,
      "learning_rate": 4.33286688782407e-06,
      "loss": 0.5407,
      "step": 106770
    },
    {
      "epoch": 43.71,
      "grad_norm": 1.9623113870620728,
      "learning_rate": 4.332782985555685e-06,
      "loss": 0.5458,
      "step": 106780
    },
    {
      "epoch": 43.71,
      "grad_norm": 2.080784320831299,
      "learning_rate": 4.3326990741758464e-06,
      "loss": 0.5595,
      "step": 106790
    },
    {
      "epoch": 43.72,
      "grad_norm": 1.90704345703125,
      "learning_rate": 4.3326151536849405e-06,
      "loss": 0.5437,
      "step": 106800
    },
    {
      "epoch": 43.72,
      "grad_norm": 2.679779529571533,
      "learning_rate": 4.3325312240833525e-06,
      "loss": 0.5525,
      "step": 106810
    },
    {
      "epoch": 43.72,
      "grad_norm": 2.224443197250366,
      "learning_rate": 4.332447285371465e-06,
      "loss": 0.5623,
      "step": 106820
    },
    {
      "epoch": 43.73,
      "grad_norm": 2.178487539291382,
      "learning_rate": 4.332363337549663e-06,
      "loss": 0.5497,
      "step": 106830
    },
    {
      "epoch": 43.73,
      "grad_norm": 2.050525188446045,
      "learning_rate": 4.332279380618332e-06,
      "loss": 0.5436,
      "step": 106840
    },
    {
      "epoch": 43.74,
      "grad_norm": 2.4488534927368164,
      "learning_rate": 4.332195414577855e-06,
      "loss": 0.5569,
      "step": 106850
    },
    {
      "epoch": 43.74,
      "grad_norm": 2.012861490249634,
      "learning_rate": 4.332111439428617e-06,
      "loss": 0.5477,
      "step": 106860
    },
    {
      "epoch": 43.75,
      "grad_norm": 2.6267454624176025,
      "learning_rate": 4.3320274551710045e-06,
      "loss": 0.5428,
      "step": 106870
    },
    {
      "epoch": 43.75,
      "grad_norm": 2.349313974380493,
      "learning_rate": 4.3319434618054e-06,
      "loss": 0.5324,
      "step": 106880
    },
    {
      "epoch": 43.75,
      "grad_norm": 2.1845386028289795,
      "learning_rate": 4.331859459332189e-06,
      "loss": 0.5368,
      "step": 106890
    },
    {
      "epoch": 43.76,
      "grad_norm": 1.9868245124816895,
      "learning_rate": 4.331775447751757e-06,
      "loss": 0.5793,
      "step": 106900
    },
    {
      "epoch": 43.76,
      "grad_norm": 2.4180469512939453,
      "learning_rate": 4.33169142706449e-06,
      "loss": 0.5599,
      "step": 106910
    },
    {
      "epoch": 43.77,
      "grad_norm": 1.9544670581817627,
      "learning_rate": 4.331607397270768e-06,
      "loss": 0.534,
      "step": 106920
    },
    {
      "epoch": 43.77,
      "grad_norm": 2.150434732437134,
      "learning_rate": 4.331523358370981e-06,
      "loss": 0.5772,
      "step": 106930
    },
    {
      "epoch": 43.77,
      "grad_norm": 2.1745364665985107,
      "learning_rate": 4.331439310365511e-06,
      "loss": 0.5442,
      "step": 106940
    },
    {
      "epoch": 43.78,
      "grad_norm": 2.53592848777771,
      "learning_rate": 4.331355253254746e-06,
      "loss": 0.5691,
      "step": 106950
    },
    {
      "epoch": 43.78,
      "grad_norm": 2.066606283187866,
      "learning_rate": 4.331271187039067e-06,
      "loss": 0.5567,
      "step": 106960
    },
    {
      "epoch": 43.79,
      "grad_norm": 2.0765817165374756,
      "learning_rate": 4.331187111718862e-06,
      "loss": 0.5681,
      "step": 106970
    },
    {
      "epoch": 43.79,
      "grad_norm": 2.027024030685425,
      "learning_rate": 4.331103027294516e-06,
      "loss": 0.5616,
      "step": 106980
    },
    {
      "epoch": 43.79,
      "grad_norm": 1.9031239748001099,
      "learning_rate": 4.331018933766414e-06,
      "loss": 0.5337,
      "step": 106990
    },
    {
      "epoch": 43.8,
      "grad_norm": 1.6774932146072388,
      "learning_rate": 4.330934831134939e-06,
      "loss": 0.542,
      "step": 107000
    },
    {
      "epoch": 43.8,
      "grad_norm": 2.347019910812378,
      "learning_rate": 4.33085071940048e-06,
      "loss": 0.5402,
      "step": 107010
    },
    {
      "epoch": 43.81,
      "grad_norm": 2.297476053237915,
      "learning_rate": 4.330766598563419e-06,
      "loss": 0.5588,
      "step": 107020
    },
    {
      "epoch": 43.81,
      "grad_norm": 2.3378713130950928,
      "learning_rate": 4.330682468624142e-06,
      "loss": 0.5445,
      "step": 107030
    },
    {
      "epoch": 43.81,
      "grad_norm": 1.9958804845809937,
      "learning_rate": 4.330598329583036e-06,
      "loss": 0.5423,
      "step": 107040
    },
    {
      "epoch": 43.82,
      "grad_norm": 2.296072006225586,
      "learning_rate": 4.330514181440486e-06,
      "loss": 0.5455,
      "step": 107050
    },
    {
      "epoch": 43.82,
      "grad_norm": 3.3310799598693848,
      "learning_rate": 4.3304300241968764e-06,
      "loss": 0.5656,
      "step": 107060
    },
    {
      "epoch": 43.83,
      "grad_norm": 2.3347506523132324,
      "learning_rate": 4.330345857852593e-06,
      "loss": 0.5618,
      "step": 107070
    },
    {
      "epoch": 43.83,
      "grad_norm": 2.3900649547576904,
      "learning_rate": 4.330261682408023e-06,
      "loss": 0.5804,
      "step": 107080
    },
    {
      "epoch": 43.84,
      "grad_norm": 1.345093846321106,
      "learning_rate": 4.330177497863549e-06,
      "loss": 0.5495,
      "step": 107090
    },
    {
      "epoch": 43.84,
      "grad_norm": 2.2519237995147705,
      "learning_rate": 4.330093304219559e-06,
      "loss": 0.537,
      "step": 107100
    },
    {
      "epoch": 43.84,
      "grad_norm": 2.374861240386963,
      "learning_rate": 4.330009101476439e-06,
      "loss": 0.5281,
      "step": 107110
    },
    {
      "epoch": 43.85,
      "grad_norm": 3.356278657913208,
      "learning_rate": 4.329924889634572e-06,
      "loss": 0.5477,
      "step": 107120
    },
    {
      "epoch": 43.85,
      "grad_norm": 1.8463152647018433,
      "learning_rate": 4.329840668694346e-06,
      "loss": 0.5492,
      "step": 107130
    },
    {
      "epoch": 43.86,
      "grad_norm": 2.0325193405151367,
      "learning_rate": 4.329756438656147e-06,
      "loss": 0.5442,
      "step": 107140
    },
    {
      "epoch": 43.86,
      "grad_norm": 2.588879108428955,
      "learning_rate": 4.329672199520359e-06,
      "loss": 0.5528,
      "step": 107150
    },
    {
      "epoch": 43.86,
      "grad_norm": 1.8221524953842163,
      "learning_rate": 4.329587951287369e-06,
      "loss": 0.5254,
      "step": 107160
    },
    {
      "epoch": 43.87,
      "grad_norm": 2.2063746452331543,
      "learning_rate": 4.329503693957563e-06,
      "loss": 0.5459,
      "step": 107170
    },
    {
      "epoch": 43.87,
      "grad_norm": 2.0325093269348145,
      "learning_rate": 4.3294194275313275e-06,
      "loss": 0.5617,
      "step": 107180
    },
    {
      "epoch": 43.88,
      "grad_norm": 2.5475077629089355,
      "learning_rate": 4.329335152009047e-06,
      "loss": 0.5416,
      "step": 107190
    },
    {
      "epoch": 43.88,
      "grad_norm": 2.7437705993652344,
      "learning_rate": 4.329250867391109e-06,
      "loss": 0.5387,
      "step": 107200
    },
    {
      "epoch": 43.88,
      "grad_norm": 2.136204719543457,
      "learning_rate": 4.329166573677899e-06,
      "loss": 0.5339,
      "step": 107210
    },
    {
      "epoch": 43.89,
      "grad_norm": 1.7896517515182495,
      "learning_rate": 4.329082270869803e-06,
      "loss": 0.5489,
      "step": 107220
    },
    {
      "epoch": 43.89,
      "grad_norm": 1.7730133533477783,
      "learning_rate": 4.328997958967207e-06,
      "loss": 0.5336,
      "step": 107230
    },
    {
      "epoch": 43.9,
      "grad_norm": 2.0932374000549316,
      "learning_rate": 4.3289136379704986e-06,
      "loss": 0.578,
      "step": 107240
    },
    {
      "epoch": 43.9,
      "grad_norm": 3.177762031555176,
      "learning_rate": 4.328829307880062e-06,
      "loss": 0.5415,
      "step": 107250
    },
    {
      "epoch": 43.91,
      "grad_norm": 2.4535462856292725,
      "learning_rate": 4.328744968696285e-06,
      "loss": 0.5533,
      "step": 107260
    },
    {
      "epoch": 43.91,
      "grad_norm": 2.627932071685791,
      "learning_rate": 4.328660620419554e-06,
      "loss": 0.547,
      "step": 107270
    },
    {
      "epoch": 43.91,
      "grad_norm": 1.812103271484375,
      "learning_rate": 4.328576263050253e-06,
      "loss": 0.5311,
      "step": 107280
    },
    {
      "epoch": 43.92,
      "grad_norm": 1.8374642133712769,
      "learning_rate": 4.328491896588771e-06,
      "loss": 0.5398,
      "step": 107290
    },
    {
      "epoch": 43.92,
      "grad_norm": 2.786407947540283,
      "learning_rate": 4.328407521035495e-06,
      "loss": 0.5464,
      "step": 107300
    },
    {
      "epoch": 43.93,
      "grad_norm": 2.163004159927368,
      "learning_rate": 4.32832313639081e-06,
      "loss": 0.5491,
      "step": 107310
    },
    {
      "epoch": 43.93,
      "grad_norm": 2.1231234073638916,
      "learning_rate": 4.328238742655102e-06,
      "loss": 0.5565,
      "step": 107320
    },
    {
      "epoch": 43.93,
      "grad_norm": 1.788482666015625,
      "learning_rate": 4.328154339828758e-06,
      "loss": 0.5658,
      "step": 107330
    },
    {
      "epoch": 43.94,
      "grad_norm": 2.0199108123779297,
      "learning_rate": 4.3280699279121655e-06,
      "loss": 0.5666,
      "step": 107340
    },
    {
      "epoch": 43.94,
      "grad_norm": 2.3141415119171143,
      "learning_rate": 4.327985506905712e-06,
      "loss": 0.5494,
      "step": 107350
    },
    {
      "epoch": 43.95,
      "grad_norm": 2.292252779006958,
      "learning_rate": 4.327901076809781e-06,
      "loss": 0.567,
      "step": 107360
    },
    {
      "epoch": 43.95,
      "grad_norm": 2.2935378551483154,
      "learning_rate": 4.327816637624763e-06,
      "loss": 0.5691,
      "step": 107370
    },
    {
      "epoch": 43.95,
      "grad_norm": 1.8250490427017212,
      "learning_rate": 4.327732189351042e-06,
      "loss": 0.5334,
      "step": 107380
    },
    {
      "epoch": 43.96,
      "grad_norm": 1.8313517570495605,
      "learning_rate": 4.327647731989005e-06,
      "loss": 0.5288,
      "step": 107390
    },
    {
      "epoch": 43.96,
      "grad_norm": 1.964896559715271,
      "learning_rate": 4.327563265539041e-06,
      "loss": 0.5329,
      "step": 107400
    },
    {
      "epoch": 43.97,
      "grad_norm": 1.7724661827087402,
      "learning_rate": 4.327478790001535e-06,
      "loss": 0.5562,
      "step": 107410
    },
    {
      "epoch": 43.97,
      "grad_norm": 2.3622026443481445,
      "learning_rate": 4.327394305376875e-06,
      "loss": 0.5373,
      "step": 107420
    },
    {
      "epoch": 43.97,
      "grad_norm": 2.672593832015991,
      "learning_rate": 4.327309811665448e-06,
      "loss": 0.5567,
      "step": 107430
    },
    {
      "epoch": 43.98,
      "grad_norm": 2.9288806915283203,
      "learning_rate": 4.32722530886764e-06,
      "loss": 0.5468,
      "step": 107440
    },
    {
      "epoch": 43.98,
      "grad_norm": 1.8221806287765503,
      "learning_rate": 4.327140796983839e-06,
      "loss": 0.5445,
      "step": 107450
    },
    {
      "epoch": 43.99,
      "grad_norm": 2.1983883380889893,
      "learning_rate": 4.327056276014432e-06,
      "loss": 0.5303,
      "step": 107460
    },
    {
      "epoch": 43.99,
      "grad_norm": 1.8697009086608887,
      "learning_rate": 4.326971745959807e-06,
      "loss": 0.5637,
      "step": 107470
    },
    {
      "epoch": 44.0,
      "grad_norm": 2.223104238510132,
      "learning_rate": 4.326887206820349e-06,
      "loss": 0.5542,
      "step": 107480
    },
    {
      "epoch": 44.0,
      "grad_norm": 1.6753230094909668,
      "learning_rate": 4.326802658596447e-06,
      "loss": 0.5538,
      "step": 107490
    },
    {
      "epoch": 44.0,
      "eval_loss": 0.5480167269706726,
      "eval_runtime": 51.8609,
      "eval_samples_per_second": 66.505,
      "eval_steps_per_second": 8.33,
      "step": 107492
    },
    {
      "epoch": 44.0,
      "grad_norm": 2.0307724475860596,
      "learning_rate": 4.3267181012884875e-06,
      "loss": 0.548,
      "step": 107500
    },
    {
      "epoch": 44.01,
      "grad_norm": 1.7434415817260742,
      "learning_rate": 4.326633534896859e-06,
      "loss": 0.5532,
      "step": 107510
    },
    {
      "epoch": 44.01,
      "grad_norm": 2.4911093711853027,
      "learning_rate": 4.326548959421947e-06,
      "loss": 0.548,
      "step": 107520
    },
    {
      "epoch": 44.02,
      "grad_norm": 2.305330514907837,
      "learning_rate": 4.326464374864141e-06,
      "loss": 0.5366,
      "step": 107530
    },
    {
      "epoch": 44.02,
      "grad_norm": 1.8381531238555908,
      "learning_rate": 4.326379781223828e-06,
      "loss": 0.549,
      "step": 107540
    },
    {
      "epoch": 44.02,
      "grad_norm": 2.0150868892669678,
      "learning_rate": 4.326295178501394e-06,
      "loss": 0.5593,
      "step": 107550
    },
    {
      "epoch": 44.03,
      "grad_norm": 2.1827642917633057,
      "learning_rate": 4.326210566697228e-06,
      "loss": 0.5285,
      "step": 107560
    },
    {
      "epoch": 44.03,
      "grad_norm": 1.7318447828292847,
      "learning_rate": 4.326125945811717e-06,
      "loss": 0.5282,
      "step": 107570
    },
    {
      "epoch": 44.04,
      "grad_norm": 2.445266008377075,
      "learning_rate": 4.326041315845248e-06,
      "loss": 0.5711,
      "step": 107580
    },
    {
      "epoch": 44.04,
      "grad_norm": 3.8216536045074463,
      "learning_rate": 4.3259566767982105e-06,
      "loss": 0.5438,
      "step": 107590
    },
    {
      "epoch": 44.04,
      "grad_norm": 1.9671152830123901,
      "learning_rate": 4.325872028670992e-06,
      "loss": 0.5458,
      "step": 107600
    },
    {
      "epoch": 44.05,
      "grad_norm": 1.8745083808898926,
      "learning_rate": 4.325787371463979e-06,
      "loss": 0.554,
      "step": 107610
    },
    {
      "epoch": 44.05,
      "grad_norm": 2.2252748012542725,
      "learning_rate": 4.325702705177558e-06,
      "loss": 0.5518,
      "step": 107620
    },
    {
      "epoch": 44.06,
      "grad_norm": 2.4943835735321045,
      "learning_rate": 4.32561802981212e-06,
      "loss": 0.5549,
      "step": 107630
    },
    {
      "epoch": 44.06,
      "grad_norm": 2.0992088317871094,
      "learning_rate": 4.3255333453680516e-06,
      "loss": 0.5289,
      "step": 107640
    },
    {
      "epoch": 44.06,
      "grad_norm": 2.692920207977295,
      "learning_rate": 4.32544865184574e-06,
      "loss": 0.5442,
      "step": 107650
    },
    {
      "epoch": 44.07,
      "grad_norm": 1.850233554840088,
      "learning_rate": 4.325363949245574e-06,
      "loss": 0.5679,
      "step": 107660
    },
    {
      "epoch": 44.07,
      "grad_norm": 2.031073570251465,
      "learning_rate": 4.325279237567942e-06,
      "loss": 0.5735,
      "step": 107670
    },
    {
      "epoch": 44.08,
      "grad_norm": 1.8133825063705444,
      "learning_rate": 4.325194516813231e-06,
      "loss": 0.5602,
      "step": 107680
    },
    {
      "epoch": 44.08,
      "grad_norm": 2.9698989391326904,
      "learning_rate": 4.32510978698183e-06,
      "loss": 0.5424,
      "step": 107690
    },
    {
      "epoch": 44.09,
      "grad_norm": 2.247582197189331,
      "learning_rate": 4.325025048074125e-06,
      "loss": 0.5429,
      "step": 107700
    },
    {
      "epoch": 44.09,
      "grad_norm": 2.111379623413086,
      "learning_rate": 4.324940300090508e-06,
      "loss": 0.5605,
      "step": 107710
    },
    {
      "epoch": 44.09,
      "grad_norm": 1.884600281715393,
      "learning_rate": 4.324855543031364e-06,
      "loss": 0.5199,
      "step": 107720
    },
    {
      "epoch": 44.1,
      "grad_norm": 2.4833383560180664,
      "learning_rate": 4.324770776897082e-06,
      "loss": 0.5416,
      "step": 107730
    },
    {
      "epoch": 44.1,
      "grad_norm": 2.578876495361328,
      "learning_rate": 4.3246860016880514e-06,
      "loss": 0.5556,
      "step": 107740
    },
    {
      "epoch": 44.11,
      "grad_norm": 2.651845932006836,
      "learning_rate": 4.3246012174046595e-06,
      "loss": 0.5656,
      "step": 107750
    },
    {
      "epoch": 44.11,
      "grad_norm": 1.679566502571106,
      "learning_rate": 4.324516424047294e-06,
      "loss": 0.5475,
      "step": 107760
    },
    {
      "epoch": 44.11,
      "grad_norm": 2.0263705253601074,
      "learning_rate": 4.324431621616346e-06,
      "loss": 0.5506,
      "step": 107770
    },
    {
      "epoch": 44.12,
      "grad_norm": 1.7423003911972046,
      "learning_rate": 4.324346810112201e-06,
      "loss": 0.5603,
      "step": 107780
    },
    {
      "epoch": 44.12,
      "grad_norm": 2.0507025718688965,
      "learning_rate": 4.324261989535249e-06,
      "loss": 0.5428,
      "step": 107790
    },
    {
      "epoch": 44.13,
      "grad_norm": 2.4116368293762207,
      "learning_rate": 4.324177159885878e-06,
      "loss": 0.5632,
      "step": 107800
    },
    {
      "epoch": 44.13,
      "grad_norm": 2.074572801589966,
      "learning_rate": 4.324092321164477e-06,
      "loss": 0.5518,
      "step": 107810
    },
    {
      "epoch": 44.13,
      "grad_norm": 2.4002058506011963,
      "learning_rate": 4.324007473371435e-06,
      "loss": 0.5521,
      "step": 107820
    },
    {
      "epoch": 44.14,
      "grad_norm": 3.1293447017669678,
      "learning_rate": 4.32392261650714e-06,
      "loss": 0.5424,
      "step": 107830
    },
    {
      "epoch": 44.14,
      "grad_norm": 2.1663851737976074,
      "learning_rate": 4.323837750571981e-06,
      "loss": 0.5515,
      "step": 107840
    },
    {
      "epoch": 44.15,
      "grad_norm": 1.9451863765716553,
      "learning_rate": 4.323752875566347e-06,
      "loss": 0.5626,
      "step": 107850
    },
    {
      "epoch": 44.15,
      "grad_norm": 2.426640033721924,
      "learning_rate": 4.323667991490626e-06,
      "loss": 0.5429,
      "step": 107860
    },
    {
      "epoch": 44.15,
      "grad_norm": 2.017890691757202,
      "learning_rate": 4.323583098345207e-06,
      "loss": 0.5571,
      "step": 107870
    },
    {
      "epoch": 44.16,
      "grad_norm": 2.8126473426818848,
      "learning_rate": 4.3234981961304795e-06,
      "loss": 0.5441,
      "step": 107880
    },
    {
      "epoch": 44.16,
      "grad_norm": 2.6095612049102783,
      "learning_rate": 4.3234132848468325e-06,
      "loss": 0.5474,
      "step": 107890
    },
    {
      "epoch": 44.17,
      "grad_norm": 2.083914279937744,
      "learning_rate": 4.323328364494653e-06,
      "loss": 0.5548,
      "step": 107900
    },
    {
      "epoch": 44.17,
      "grad_norm": 2.314030647277832,
      "learning_rate": 4.3232434350743336e-06,
      "loss": 0.56,
      "step": 107910
    },
    {
      "epoch": 44.18,
      "grad_norm": 2.022469997406006,
      "learning_rate": 4.323158496586261e-06,
      "loss": 0.5472,
      "step": 107920
    },
    {
      "epoch": 44.18,
      "grad_norm": 2.4388675689697266,
      "learning_rate": 4.323073549030824e-06,
      "loss": 0.5348,
      "step": 107930
    },
    {
      "epoch": 44.18,
      "grad_norm": 2.5498359203338623,
      "learning_rate": 4.322988592408412e-06,
      "loss": 0.565,
      "step": 107940
    },
    {
      "epoch": 44.19,
      "grad_norm": 2.070490837097168,
      "learning_rate": 4.3229036267194155e-06,
      "loss": 0.5595,
      "step": 107950
    },
    {
      "epoch": 44.19,
      "grad_norm": 2.4853668212890625,
      "learning_rate": 4.322818651964223e-06,
      "loss": 0.5324,
      "step": 107960
    },
    {
      "epoch": 44.2,
      "grad_norm": 2.3438477516174316,
      "learning_rate": 4.322733668143222e-06,
      "loss": 0.5551,
      "step": 107970
    },
    {
      "epoch": 44.2,
      "grad_norm": 2.954362154006958,
      "learning_rate": 4.322648675256804e-06,
      "loss": 0.5442,
      "step": 107980
    },
    {
      "epoch": 44.2,
      "grad_norm": 2.229663610458374,
      "learning_rate": 4.322563673305358e-06,
      "loss": 0.5546,
      "step": 107990
    },
    {
      "epoch": 44.21,
      "grad_norm": 2.2599143981933594,
      "learning_rate": 4.322478662289273e-06,
      "loss": 0.5514,
      "step": 108000
    },
    {
      "epoch": 44.21,
      "grad_norm": 1.3498010635375977,
      "learning_rate": 4.322393642208938e-06,
      "loss": 0.5255,
      "step": 108010
    },
    {
      "epoch": 44.22,
      "grad_norm": 2.2581372261047363,
      "learning_rate": 4.322308613064743e-06,
      "loss": 0.5323,
      "step": 108020
    },
    {
      "epoch": 44.22,
      "grad_norm": 2.026221752166748,
      "learning_rate": 4.322223574857078e-06,
      "loss": 0.5541,
      "step": 108030
    },
    {
      "epoch": 44.22,
      "grad_norm": 4.902153491973877,
      "learning_rate": 4.322138527586331e-06,
      "loss": 0.5405,
      "step": 108040
    },
    {
      "epoch": 44.23,
      "grad_norm": 2.137350559234619,
      "learning_rate": 4.3220534712528935e-06,
      "loss": 0.5651,
      "step": 108050
    },
    {
      "epoch": 44.23,
      "grad_norm": 1.6824506521224976,
      "learning_rate": 4.321968405857153e-06,
      "loss": 0.5648,
      "step": 108060
    },
    {
      "epoch": 44.24,
      "grad_norm": 1.7907352447509766,
      "learning_rate": 4.321883331399502e-06,
      "loss": 0.5358,
      "step": 108070
    },
    {
      "epoch": 44.24,
      "grad_norm": 1.7108482122421265,
      "learning_rate": 4.321798247880327e-06,
      "loss": 0.5317,
      "step": 108080
    },
    {
      "epoch": 44.24,
      "grad_norm": 2.402952194213867,
      "learning_rate": 4.32171315530002e-06,
      "loss": 0.5681,
      "step": 108090
    },
    {
      "epoch": 44.25,
      "grad_norm": 1.7523983716964722,
      "learning_rate": 4.32162805365897e-06,
      "loss": 0.5745,
      "step": 108100
    },
    {
      "epoch": 44.25,
      "grad_norm": 2.00469970703125,
      "learning_rate": 4.321542942957567e-06,
      "loss": 0.5653,
      "step": 108110
    },
    {
      "epoch": 44.26,
      "grad_norm": 2.2119288444519043,
      "learning_rate": 4.321457823196202e-06,
      "loss": 0.5566,
      "step": 108120
    },
    {
      "epoch": 44.26,
      "grad_norm": 2.5384292602539062,
      "learning_rate": 4.321372694375262e-06,
      "loss": 0.5577,
      "step": 108130
    },
    {
      "epoch": 44.27,
      "grad_norm": 2.276777744293213,
      "learning_rate": 4.321287556495139e-06,
      "loss": 0.574,
      "step": 108140
    },
    {
      "epoch": 44.27,
      "grad_norm": 2.8834526538848877,
      "learning_rate": 4.321202409556223e-06,
      "loss": 0.5446,
      "step": 108150
    },
    {
      "epoch": 44.27,
      "grad_norm": 2.3363888263702393,
      "learning_rate": 4.3211172535589035e-06,
      "loss": 0.5466,
      "step": 108160
    },
    {
      "epoch": 44.28,
      "grad_norm": 2.0936367511749268,
      "learning_rate": 4.321032088503571e-06,
      "loss": 0.559,
      "step": 108170
    },
    {
      "epoch": 44.28,
      "grad_norm": 2.1594743728637695,
      "learning_rate": 4.320946914390616e-06,
      "loss": 0.5483,
      "step": 108180
    },
    {
      "epoch": 44.29,
      "grad_norm": 2.4213216304779053,
      "learning_rate": 4.320861731220428e-06,
      "loss": 0.5554,
      "step": 108190
    },
    {
      "epoch": 44.29,
      "grad_norm": 1.8033366203308105,
      "learning_rate": 4.320776538993397e-06,
      "loss": 0.5464,
      "step": 108200
    },
    {
      "epoch": 44.29,
      "grad_norm": 2.228199005126953,
      "learning_rate": 4.320691337709913e-06,
      "loss": 0.5383,
      "step": 108210
    },
    {
      "epoch": 44.3,
      "grad_norm": 1.7751046419143677,
      "learning_rate": 4.320606127370368e-06,
      "loss": 0.5575,
      "step": 108220
    },
    {
      "epoch": 44.3,
      "grad_norm": 2.234679937362671,
      "learning_rate": 4.320520907975151e-06,
      "loss": 0.544,
      "step": 108230
    },
    {
      "epoch": 44.31,
      "grad_norm": 1.9174715280532837,
      "learning_rate": 4.320435679524653e-06,
      "loss": 0.5688,
      "step": 108240
    },
    {
      "epoch": 44.31,
      "grad_norm": 1.8153648376464844,
      "learning_rate": 4.320350442019263e-06,
      "loss": 0.5429,
      "step": 108250
    },
    {
      "epoch": 44.31,
      "grad_norm": 2.903900623321533,
      "learning_rate": 4.320265195459373e-06,
      "loss": 0.5397,
      "step": 108260
    },
    {
      "epoch": 44.32,
      "grad_norm": 1.7249267101287842,
      "learning_rate": 4.320179939845373e-06,
      "loss": 0.5532,
      "step": 108270
    },
    {
      "epoch": 44.32,
      "grad_norm": 2.4439282417297363,
      "learning_rate": 4.320094675177653e-06,
      "loss": 0.5538,
      "step": 108280
    },
    {
      "epoch": 44.33,
      "grad_norm": 1.9263027906417847,
      "learning_rate": 4.320009401456606e-06,
      "loss": 0.5603,
      "step": 108290
    },
    {
      "epoch": 44.33,
      "grad_norm": 2.764537811279297,
      "learning_rate": 4.319924118682618e-06,
      "loss": 0.5663,
      "step": 108300
    },
    {
      "epoch": 44.33,
      "grad_norm": 2.361440420150757,
      "learning_rate": 4.319838826856084e-06,
      "loss": 0.556,
      "step": 108310
    },
    {
      "epoch": 44.34,
      "grad_norm": 2.346947193145752,
      "learning_rate": 4.3197535259773934e-06,
      "loss": 0.528,
      "step": 108320
    },
    {
      "epoch": 44.34,
      "grad_norm": 2.2314820289611816,
      "learning_rate": 4.319668216046936e-06,
      "loss": 0.5396,
      "step": 108330
    },
    {
      "epoch": 44.35,
      "grad_norm": 1.7160637378692627,
      "learning_rate": 4.319582897065103e-06,
      "loss": 0.5444,
      "step": 108340
    },
    {
      "epoch": 44.35,
      "grad_norm": 2.3881285190582275,
      "learning_rate": 4.319497569032285e-06,
      "loss": 0.5689,
      "step": 108350
    },
    {
      "epoch": 44.36,
      "grad_norm": 2.19364595413208,
      "learning_rate": 4.319412231948876e-06,
      "loss": 0.5434,
      "step": 108360
    },
    {
      "epoch": 44.36,
      "grad_norm": 1.7684396505355835,
      "learning_rate": 4.319326885815262e-06,
      "loss": 0.533,
      "step": 108370
    },
    {
      "epoch": 44.36,
      "grad_norm": 1.8815770149230957,
      "learning_rate": 4.319241530631836e-06,
      "loss": 0.534,
      "step": 108380
    },
    {
      "epoch": 44.37,
      "grad_norm": 1.7926762104034424,
      "learning_rate": 4.319156166398991e-06,
      "loss": 0.5426,
      "step": 108390
    },
    {
      "epoch": 44.37,
      "grad_norm": 2.3225646018981934,
      "learning_rate": 4.319070793117115e-06,
      "loss": 0.5353,
      "step": 108400
    },
    {
      "epoch": 44.38,
      "grad_norm": 2.1653707027435303,
      "learning_rate": 4.3189854107866e-06,
      "loss": 0.5516,
      "step": 108410
    },
    {
      "epoch": 44.38,
      "grad_norm": 3.341549873352051,
      "learning_rate": 4.318900019407838e-06,
      "loss": 0.5481,
      "step": 108420
    },
    {
      "epoch": 44.38,
      "grad_norm": 1.6874504089355469,
      "learning_rate": 4.318814618981221e-06,
      "loss": 0.5389,
      "step": 108430
    },
    {
      "epoch": 44.39,
      "grad_norm": 2.2239832878112793,
      "learning_rate": 4.3187292095071375e-06,
      "loss": 0.5215,
      "step": 108440
    },
    {
      "epoch": 44.39,
      "grad_norm": 2.423274040222168,
      "learning_rate": 4.3186437909859805e-06,
      "loss": 0.562,
      "step": 108450
    },
    {
      "epoch": 44.4,
      "grad_norm": 2.4873502254486084,
      "learning_rate": 4.31855836341814e-06,
      "loss": 0.5525,
      "step": 108460
    },
    {
      "epoch": 44.4,
      "grad_norm": 2.117725133895874,
      "learning_rate": 4.318472926804009e-06,
      "loss": 0.545,
      "step": 108470
    },
    {
      "epoch": 44.4,
      "grad_norm": 1.8638283014297485,
      "learning_rate": 4.318387481143978e-06,
      "loss": 0.5549,
      "step": 108480
    },
    {
      "epoch": 44.41,
      "grad_norm": 2.0709116458892822,
      "learning_rate": 4.318302026438439e-06,
      "loss": 0.5536,
      "step": 108490
    },
    {
      "epoch": 44.41,
      "grad_norm": 2.2566349506378174,
      "learning_rate": 4.3182165626877824e-06,
      "loss": 0.5419,
      "step": 108500
    },
    {
      "epoch": 44.42,
      "grad_norm": 2.5464532375335693,
      "learning_rate": 4.318131089892401e-06,
      "loss": 0.5258,
      "step": 108510
    },
    {
      "epoch": 44.42,
      "grad_norm": 2.6636416912078857,
      "learning_rate": 4.318045608052684e-06,
      "loss": 0.5621,
      "step": 108520
    },
    {
      "epoch": 44.42,
      "grad_norm": 1.9785188436508179,
      "learning_rate": 4.3179601171690255e-06,
      "loss": 0.5404,
      "step": 108530
    },
    {
      "epoch": 44.43,
      "grad_norm": 2.3441781997680664,
      "learning_rate": 4.317874617241816e-06,
      "loss": 0.5558,
      "step": 108540
    },
    {
      "epoch": 44.43,
      "grad_norm": 2.199610948562622,
      "learning_rate": 4.317789108271448e-06,
      "loss": 0.5577,
      "step": 108550
    },
    {
      "epoch": 44.44,
      "grad_norm": 2.75887393951416,
      "learning_rate": 4.317703590258312e-06,
      "loss": 0.5301,
      "step": 108560
    },
    {
      "epoch": 44.44,
      "grad_norm": 1.6907038688659668,
      "learning_rate": 4.317618063202801e-06,
      "loss": 0.5614,
      "step": 108570
    },
    {
      "epoch": 44.45,
      "grad_norm": 2.008610725402832,
      "learning_rate": 4.317532527105305e-06,
      "loss": 0.5445,
      "step": 108580
    },
    {
      "epoch": 44.45,
      "grad_norm": 1.9814835786819458,
      "learning_rate": 4.317446981966218e-06,
      "loss": 0.5624,
      "step": 108590
    },
    {
      "epoch": 44.45,
      "grad_norm": 1.6396968364715576,
      "learning_rate": 4.31736142778593e-06,
      "loss": 0.5428,
      "step": 108600
    },
    {
      "epoch": 44.46,
      "grad_norm": 1.873026967048645,
      "learning_rate": 4.317275864564834e-06,
      "loss": 0.5448,
      "step": 108610
    },
    {
      "epoch": 44.46,
      "grad_norm": 2.20161509513855,
      "learning_rate": 4.3171902923033215e-06,
      "loss": 0.5611,
      "step": 108620
    },
    {
      "epoch": 44.47,
      "grad_norm": 2.6161105632781982,
      "learning_rate": 4.317104711001784e-06,
      "loss": 0.56,
      "step": 108630
    },
    {
      "epoch": 44.47,
      "grad_norm": 2.3014237880706787,
      "learning_rate": 4.317019120660616e-06,
      "loss": 0.5555,
      "step": 108640
    },
    {
      "epoch": 44.47,
      "grad_norm": 2.280458450317383,
      "learning_rate": 4.316933521280206e-06,
      "loss": 0.5425,
      "step": 108650
    },
    {
      "epoch": 44.48,
      "grad_norm": 1.9650543928146362,
      "learning_rate": 4.31684791286095e-06,
      "loss": 0.5601,
      "step": 108660
    },
    {
      "epoch": 44.48,
      "grad_norm": 2.3989856243133545,
      "learning_rate": 4.316762295403235e-06,
      "loss": 0.5513,
      "step": 108670
    },
    {
      "epoch": 44.49,
      "grad_norm": 2.3093063831329346,
      "learning_rate": 4.316676668907459e-06,
      "loss": 0.5594,
      "step": 108680
    },
    {
      "epoch": 44.49,
      "grad_norm": 2.2203257083892822,
      "learning_rate": 4.316591033374009e-06,
      "loss": 0.5599,
      "step": 108690
    },
    {
      "epoch": 44.49,
      "grad_norm": 2.32256817817688,
      "learning_rate": 4.316505388803283e-06,
      "loss": 0.5491,
      "step": 108700
    },
    {
      "epoch": 44.5,
      "grad_norm": 2.4586262702941895,
      "learning_rate": 4.316419735195667e-06,
      "loss": 0.5542,
      "step": 108710
    },
    {
      "epoch": 44.5,
      "grad_norm": 1.897262692451477,
      "learning_rate": 4.316334072551558e-06,
      "loss": 0.5481,
      "step": 108720
    },
    {
      "epoch": 44.51,
      "grad_norm": 1.9078428745269775,
      "learning_rate": 4.316248400871347e-06,
      "loss": 0.5456,
      "step": 108730
    },
    {
      "epoch": 44.51,
      "grad_norm": 2.744955539703369,
      "learning_rate": 4.316162720155425e-06,
      "loss": 0.5592,
      "step": 108740
    },
    {
      "epoch": 44.51,
      "grad_norm": 1.6394919157028198,
      "learning_rate": 4.316077030404186e-06,
      "loss": 0.5572,
      "step": 108750
    },
    {
      "epoch": 44.52,
      "grad_norm": 2.8692049980163574,
      "learning_rate": 4.315991331618024e-06,
      "loss": 0.5695,
      "step": 108760
    },
    {
      "epoch": 44.52,
      "grad_norm": 2.772160053253174,
      "learning_rate": 4.315905623797328e-06,
      "loss": 0.5421,
      "step": 108770
    },
    {
      "epoch": 44.53,
      "grad_norm": 2.846226930618286,
      "learning_rate": 4.315819906942493e-06,
      "loss": 0.5455,
      "step": 108780
    },
    {
      "epoch": 44.53,
      "grad_norm": 2.5323944091796875,
      "learning_rate": 4.315734181053912e-06,
      "loss": 0.5516,
      "step": 108790
    },
    {
      "epoch": 44.54,
      "grad_norm": 1.9476298093795776,
      "learning_rate": 4.315648446131977e-06,
      "loss": 0.5592,
      "step": 108800
    },
    {
      "epoch": 44.54,
      "grad_norm": 2.346492052078247,
      "learning_rate": 4.315562702177078e-06,
      "loss": 0.551,
      "step": 108810
    },
    {
      "epoch": 44.54,
      "grad_norm": 1.7780042886734009,
      "learning_rate": 4.315476949189614e-06,
      "loss": 0.5596,
      "step": 108820
    },
    {
      "epoch": 44.55,
      "grad_norm": 2.023686408996582,
      "learning_rate": 4.315391187169972e-06,
      "loss": 0.5333,
      "step": 108830
    },
    {
      "epoch": 44.55,
      "grad_norm": 1.6411041021347046,
      "learning_rate": 4.3153054161185475e-06,
      "loss": 0.5515,
      "step": 108840
    },
    {
      "epoch": 44.56,
      "grad_norm": 2.2317938804626465,
      "learning_rate": 4.315219636035733e-06,
      "loss": 0.5766,
      "step": 108850
    },
    {
      "epoch": 44.56,
      "grad_norm": 2.647650718688965,
      "learning_rate": 4.315133846921922e-06,
      "loss": 0.536,
      "step": 108860
    },
    {
      "epoch": 44.56,
      "grad_norm": 2.1369786262512207,
      "learning_rate": 4.315048048777506e-06,
      "loss": 0.5443,
      "step": 108870
    },
    {
      "epoch": 44.57,
      "grad_norm": 2.042853593826294,
      "learning_rate": 4.31496224160288e-06,
      "loss": 0.5513,
      "step": 108880
    },
    {
      "epoch": 44.57,
      "grad_norm": 2.7897017002105713,
      "learning_rate": 4.314876425398435e-06,
      "loss": 0.5427,
      "step": 108890
    },
    {
      "epoch": 44.58,
      "grad_norm": 2.3141133785247803,
      "learning_rate": 4.314790600164565e-06,
      "loss": 0.5463,
      "step": 108900
    },
    {
      "epoch": 44.58,
      "grad_norm": 1.9481688737869263,
      "learning_rate": 4.314704765901664e-06,
      "loss": 0.5414,
      "step": 108910
    },
    {
      "epoch": 44.58,
      "grad_norm": 2.193514347076416,
      "learning_rate": 4.314618922610124e-06,
      "loss": 0.532,
      "step": 108920
    },
    {
      "epoch": 44.59,
      "grad_norm": 2.3900182247161865,
      "learning_rate": 4.314533070290339e-06,
      "loss": 0.5442,
      "step": 108930
    },
    {
      "epoch": 44.59,
      "grad_norm": 2.1389310359954834,
      "learning_rate": 4.314447208942703e-06,
      "loss": 0.5546,
      "step": 108940
    },
    {
      "epoch": 44.6,
      "grad_norm": 2.1527416706085205,
      "learning_rate": 4.314361338567607e-06,
      "loss": 0.5417,
      "step": 108950
    },
    {
      "epoch": 44.6,
      "grad_norm": 1.5796815156936646,
      "learning_rate": 4.3142754591654445e-06,
      "loss": 0.5464,
      "step": 108960
    },
    {
      "epoch": 44.6,
      "grad_norm": 1.7689909934997559,
      "learning_rate": 4.314189570736611e-06,
      "loss": 0.5481,
      "step": 108970
    },
    {
      "epoch": 44.61,
      "grad_norm": 2.7586870193481445,
      "learning_rate": 4.3141036732815e-06,
      "loss": 0.5462,
      "step": 108980
    },
    {
      "epoch": 44.61,
      "grad_norm": 1.9725319147109985,
      "learning_rate": 4.314017766800504e-06,
      "loss": 0.5549,
      "step": 108990
    },
    {
      "epoch": 44.62,
      "grad_norm": 2.273860454559326,
      "learning_rate": 4.313931851294015e-06,
      "loss": 0.5272,
      "step": 109000
    },
    {
      "epoch": 44.62,
      "grad_norm": 2.3628430366516113,
      "learning_rate": 4.313845926762429e-06,
      "loss": 0.5687,
      "step": 109010
    },
    {
      "epoch": 44.63,
      "grad_norm": 2.2887802124023438,
      "learning_rate": 4.313759993206138e-06,
      "loss": 0.5278,
      "step": 109020
    },
    {
      "epoch": 44.63,
      "grad_norm": 2.193223714828491,
      "learning_rate": 4.313674050625538e-06,
      "loss": 0.548,
      "step": 109030
    },
    {
      "epoch": 44.63,
      "grad_norm": 1.6181426048278809,
      "learning_rate": 4.313588099021019e-06,
      "loss": 0.5461,
      "step": 109040
    },
    {
      "epoch": 44.64,
      "grad_norm": 2.016287326812744,
      "learning_rate": 4.313502138392978e-06,
      "loss": 0.5568,
      "step": 109050
    },
    {
      "epoch": 44.64,
      "grad_norm": 1.8589202165603638,
      "learning_rate": 4.313416168741806e-06,
      "loss": 0.5603,
      "step": 109060
    },
    {
      "epoch": 44.65,
      "grad_norm": 2.8818891048431396,
      "learning_rate": 4.3133301900679e-06,
      "loss": 0.5355,
      "step": 109070
    },
    {
      "epoch": 44.65,
      "grad_norm": 1.8072667121887207,
      "learning_rate": 4.3132442023716515e-06,
      "loss": 0.536,
      "step": 109080
    },
    {
      "epoch": 44.65,
      "grad_norm": 2.1414239406585693,
      "learning_rate": 4.313158205653455e-06,
      "loss": 0.5444,
      "step": 109090
    },
    {
      "epoch": 44.66,
      "grad_norm": 2.0221784114837646,
      "learning_rate": 4.3130721999137045e-06,
      "loss": 0.5494,
      "step": 109100
    },
    {
      "epoch": 44.66,
      "grad_norm": 2.959907054901123,
      "learning_rate": 4.312986185152795e-06,
      "loss": 0.539,
      "step": 109110
    },
    {
      "epoch": 44.67,
      "grad_norm": 1.988181471824646,
      "learning_rate": 4.312900161371118e-06,
      "loss": 0.5387,
      "step": 109120
    },
    {
      "epoch": 44.67,
      "grad_norm": 2.208516836166382,
      "learning_rate": 4.31281412856907e-06,
      "loss": 0.5444,
      "step": 109130
    },
    {
      "epoch": 44.67,
      "grad_norm": 1.7456616163253784,
      "learning_rate": 4.312728086747044e-06,
      "loss": 0.5275,
      "step": 109140
    },
    {
      "epoch": 44.68,
      "grad_norm": 2.141947031021118,
      "learning_rate": 4.3126420359054335e-06,
      "loss": 0.5775,
      "step": 109150
    },
    {
      "epoch": 44.68,
      "grad_norm": 2.077711343765259,
      "learning_rate": 4.312555976044635e-06,
      "loss": 0.5365,
      "step": 109160
    },
    {
      "epoch": 44.69,
      "grad_norm": 1.7984728813171387,
      "learning_rate": 4.31246990716504e-06,
      "loss": 0.5464,
      "step": 109170
    },
    {
      "epoch": 44.69,
      "grad_norm": 2.259777307510376,
      "learning_rate": 4.312383829267045e-06,
      "loss": 0.5183,
      "step": 109180
    },
    {
      "epoch": 44.7,
      "grad_norm": 2.4862120151519775,
      "learning_rate": 4.3122977423510425e-06,
      "loss": 0.556,
      "step": 109190
    },
    {
      "epoch": 44.7,
      "grad_norm": 2.896911382675171,
      "learning_rate": 4.312211646417427e-06,
      "loss": 0.5477,
      "step": 109200
    },
    {
      "epoch": 44.7,
      "grad_norm": 2.3569509983062744,
      "learning_rate": 4.312125541466596e-06,
      "loss": 0.5644,
      "step": 109210
    },
    {
      "epoch": 44.71,
      "grad_norm": 2.0731019973754883,
      "learning_rate": 4.312039427498939e-06,
      "loss": 0.54,
      "step": 109220
    },
    {
      "epoch": 44.71,
      "grad_norm": 1.8476399183273315,
      "learning_rate": 4.311953304514854e-06,
      "loss": 0.5485,
      "step": 109230
    },
    {
      "epoch": 44.72,
      "grad_norm": 2.3031609058380127,
      "learning_rate": 4.311867172514734e-06,
      "loss": 0.5429,
      "step": 109240
    },
    {
      "epoch": 44.72,
      "grad_norm": 1.987261176109314,
      "learning_rate": 4.3117810314989755e-06,
      "loss": 0.5564,
      "step": 109250
    },
    {
      "epoch": 44.72,
      "grad_norm": 2.5526161193847656,
      "learning_rate": 4.31169488146797e-06,
      "loss": 0.557,
      "step": 109260
    },
    {
      "epoch": 44.73,
      "grad_norm": 2.2775919437408447,
      "learning_rate": 4.311608722422114e-06,
      "loss": 0.546,
      "step": 109270
    },
    {
      "epoch": 44.73,
      "grad_norm": 5.339600563049316,
      "learning_rate": 4.311522554361803e-06,
      "loss": 0.5549,
      "step": 109280
    },
    {
      "epoch": 44.74,
      "grad_norm": 1.912207841873169,
      "learning_rate": 4.31143637728743e-06,
      "loss": 0.554,
      "step": 109290
    },
    {
      "epoch": 44.74,
      "grad_norm": 1.875025987625122,
      "learning_rate": 4.31135019119939e-06,
      "loss": 0.5499,
      "step": 109300
    },
    {
      "epoch": 44.74,
      "grad_norm": 2.246645450592041,
      "learning_rate": 4.311263996098079e-06,
      "loss": 0.5386,
      "step": 109310
    },
    {
      "epoch": 44.75,
      "grad_norm": 2.3132810592651367,
      "learning_rate": 4.3111777919838905e-06,
      "loss": 0.5644,
      "step": 109320
    },
    {
      "epoch": 44.75,
      "grad_norm": 2.5216352939605713,
      "learning_rate": 4.31109157885722e-06,
      "loss": 0.5642,
      "step": 109330
    },
    {
      "epoch": 44.76,
      "grad_norm": 1.8188903331756592,
      "learning_rate": 4.311005356718462e-06,
      "loss": 0.5456,
      "step": 109340
    },
    {
      "epoch": 44.76,
      "grad_norm": 1.7064597606658936,
      "learning_rate": 4.310919125568013e-06,
      "loss": 0.5428,
      "step": 109350
    },
    {
      "epoch": 44.76,
      "grad_norm": 2.3840456008911133,
      "learning_rate": 4.310832885406267e-06,
      "loss": 0.5269,
      "step": 109360
    },
    {
      "epoch": 44.77,
      "grad_norm": 2.3252604007720947,
      "learning_rate": 4.310746636233618e-06,
      "loss": 0.5675,
      "step": 109370
    },
    {
      "epoch": 44.77,
      "grad_norm": 2.2425949573516846,
      "learning_rate": 4.310660378050463e-06,
      "loss": 0.5266,
      "step": 109380
    },
    {
      "epoch": 44.78,
      "grad_norm": 2.057595729827881,
      "learning_rate": 4.310574110857195e-06,
      "loss": 0.5558,
      "step": 109390
    },
    {
      "epoch": 44.78,
      "grad_norm": 2.031956672668457,
      "learning_rate": 4.31048783465421e-06,
      "loss": 0.5877,
      "step": 109400
    },
    {
      "epoch": 44.79,
      "grad_norm": 2.378204107284546,
      "learning_rate": 4.310401549441905e-06,
      "loss": 0.5241,
      "step": 109410
    },
    {
      "epoch": 44.79,
      "grad_norm": 2.411025047302246,
      "learning_rate": 4.310315255220673e-06,
      "loss": 0.5663,
      "step": 109420
    },
    {
      "epoch": 44.79,
      "grad_norm": 1.708167314529419,
      "learning_rate": 4.310228951990911e-06,
      "loss": 0.551,
      "step": 109430
    },
    {
      "epoch": 44.8,
      "grad_norm": 1.94674813747406,
      "learning_rate": 4.310142639753013e-06,
      "loss": 0.5566,
      "step": 109440
    },
    {
      "epoch": 44.8,
      "grad_norm": 1.9242627620697021,
      "learning_rate": 4.310056318507375e-06,
      "loss": 0.5546,
      "step": 109450
    },
    {
      "epoch": 44.81,
      "grad_norm": 2.69991135597229,
      "learning_rate": 4.309969988254391e-06,
      "loss": 0.5528,
      "step": 109460
    },
    {
      "epoch": 44.81,
      "grad_norm": 2.737360715866089,
      "learning_rate": 4.309883648994459e-06,
      "loss": 0.5477,
      "step": 109470
    },
    {
      "epoch": 44.81,
      "grad_norm": 2.4183123111724854,
      "learning_rate": 4.309797300727974e-06,
      "loss": 0.5396,
      "step": 109480
    },
    {
      "epoch": 44.82,
      "grad_norm": 2.027576446533203,
      "learning_rate": 4.309710943455329e-06,
      "loss": 0.5523,
      "step": 109490
    },
    {
      "epoch": 44.82,
      "grad_norm": 1.7043060064315796,
      "learning_rate": 4.309624577176923e-06,
      "loss": 0.5494,
      "step": 109500
    },
    {
      "epoch": 44.83,
      "grad_norm": 1.8622519969940186,
      "learning_rate": 4.309538201893148e-06,
      "loss": 0.5497,
      "step": 109510
    },
    {
      "epoch": 44.83,
      "grad_norm": 1.7043960094451904,
      "learning_rate": 4.309451817604403e-06,
      "loss": 0.5472,
      "step": 109520
    },
    {
      "epoch": 44.83,
      "grad_norm": 2.683516502380371,
      "learning_rate": 4.309365424311083e-06,
      "loss": 0.5488,
      "step": 109530
    },
    {
      "epoch": 44.84,
      "grad_norm": 3.3732283115386963,
      "learning_rate": 4.309279022013582e-06,
      "loss": 0.5482,
      "step": 109540
    },
    {
      "epoch": 44.84,
      "grad_norm": 2.241781234741211,
      "learning_rate": 4.309192610712297e-06,
      "loss": 0.5556,
      "step": 109550
    },
    {
      "epoch": 44.85,
      "grad_norm": 2.6414780616760254,
      "learning_rate": 4.309106190407625e-06,
      "loss": 0.5669,
      "step": 109560
    },
    {
      "epoch": 44.85,
      "grad_norm": 2.243196487426758,
      "learning_rate": 4.30901976109996e-06,
      "loss": 0.5604,
      "step": 109570
    },
    {
      "epoch": 44.85,
      "grad_norm": 1.973664402961731,
      "learning_rate": 4.308933322789698e-06,
      "loss": 0.5653,
      "step": 109580
    },
    {
      "epoch": 44.86,
      "grad_norm": 2.7360284328460693,
      "learning_rate": 4.308846875477236e-06,
      "loss": 0.5389,
      "step": 109590
    },
    {
      "epoch": 44.86,
      "grad_norm": 2.4533777236938477,
      "learning_rate": 4.30876041916297e-06,
      "loss": 0.5625,
      "step": 109600
    },
    {
      "epoch": 44.87,
      "grad_norm": 2.320197820663452,
      "learning_rate": 4.3086739538472945e-06,
      "loss": 0.5387,
      "step": 109610
    },
    {
      "epoch": 44.87,
      "grad_norm": 2.6641225814819336,
      "learning_rate": 4.308587479530607e-06,
      "loss": 0.5574,
      "step": 109620
    },
    {
      "epoch": 44.88,
      "grad_norm": 1.9332549571990967,
      "learning_rate": 4.308500996213304e-06,
      "loss": 0.5605,
      "step": 109630
    },
    {
      "epoch": 44.88,
      "grad_norm": 1.8800616264343262,
      "learning_rate": 4.308414503895781e-06,
      "loss": 0.5342,
      "step": 109640
    },
    {
      "epoch": 44.88,
      "grad_norm": 2.329623222351074,
      "learning_rate": 4.308328002578433e-06,
      "loss": 0.5585,
      "step": 109650
    },
    {
      "epoch": 44.89,
      "grad_norm": 2.544847249984741,
      "learning_rate": 4.3082414922616585e-06,
      "loss": 0.5325,
      "step": 109660
    },
    {
      "epoch": 44.89,
      "grad_norm": 2.4323902130126953,
      "learning_rate": 4.308154972945852e-06,
      "loss": 0.5287,
      "step": 109670
    },
    {
      "epoch": 44.9,
      "grad_norm": 1.8719828128814697,
      "learning_rate": 4.308068444631411e-06,
      "loss": 0.5498,
      "step": 109680
    },
    {
      "epoch": 44.9,
      "grad_norm": 2.268378257751465,
      "learning_rate": 4.3079819073187315e-06,
      "loss": 0.5688,
      "step": 109690
    },
    {
      "epoch": 44.9,
      "grad_norm": 2.158275604248047,
      "learning_rate": 4.30789536100821e-06,
      "loss": 0.5312,
      "step": 109700
    },
    {
      "epoch": 44.91,
      "grad_norm": 2.474424362182617,
      "learning_rate": 4.307808805700242e-06,
      "loss": 0.5518,
      "step": 109710
    },
    {
      "epoch": 44.91,
      "grad_norm": 1.8506461381912231,
      "learning_rate": 4.3077222413952255e-06,
      "loss": 0.5386,
      "step": 109720
    },
    {
      "epoch": 44.92,
      "grad_norm": 2.2656350135803223,
      "learning_rate": 4.307635668093556e-06,
      "loss": 0.5637,
      "step": 109730
    },
    {
      "epoch": 44.92,
      "grad_norm": 2.5018398761749268,
      "learning_rate": 4.307549085795631e-06,
      "loss": 0.5387,
      "step": 109740
    },
    {
      "epoch": 44.92,
      "grad_norm": 2.3139941692352295,
      "learning_rate": 4.307462494501845e-06,
      "loss": 0.5523,
      "step": 109750
    },
    {
      "epoch": 44.93,
      "grad_norm": 2.145648717880249,
      "learning_rate": 4.307375894212598e-06,
      "loss": 0.5704,
      "step": 109760
    },
    {
      "epoch": 44.93,
      "grad_norm": 2.095191240310669,
      "learning_rate": 4.307289284928284e-06,
      "loss": 0.5404,
      "step": 109770
    },
    {
      "epoch": 44.94,
      "grad_norm": 2.26285982131958,
      "learning_rate": 4.3072026666493e-06,
      "loss": 0.5386,
      "step": 109780
    },
    {
      "epoch": 44.94,
      "grad_norm": 2.1315371990203857,
      "learning_rate": 4.307116039376044e-06,
      "loss": 0.5556,
      "step": 109790
    },
    {
      "epoch": 44.94,
      "grad_norm": 2.766920804977417,
      "learning_rate": 4.307029403108913e-06,
      "loss": 0.5615,
      "step": 109800
    },
    {
      "epoch": 44.95,
      "grad_norm": 1.8531522750854492,
      "learning_rate": 4.306942757848302e-06,
      "loss": 0.5512,
      "step": 109810
    },
    {
      "epoch": 44.95,
      "grad_norm": 1.5457851886749268,
      "learning_rate": 4.306856103594609e-06,
      "loss": 0.5535,
      "step": 109820
    },
    {
      "epoch": 44.96,
      "grad_norm": 1.9932914972305298,
      "learning_rate": 4.306769440348232e-06,
      "loss": 0.557,
      "step": 109830
    },
    {
      "epoch": 44.96,
      "grad_norm": 2.136106252670288,
      "learning_rate": 4.306682768109566e-06,
      "loss": 0.5617,
      "step": 109840
    },
    {
      "epoch": 44.97,
      "grad_norm": 1.5044283866882324,
      "learning_rate": 4.30659608687901e-06,
      "loss": 0.5608,
      "step": 109850
    },
    {
      "epoch": 44.97,
      "grad_norm": 2.1433963775634766,
      "learning_rate": 4.3065093966569596e-06,
      "loss": 0.5388,
      "step": 109860
    },
    {
      "epoch": 44.97,
      "grad_norm": 2.183047294616699,
      "learning_rate": 4.306422697443812e-06,
      "loss": 0.5432,
      "step": 109870
    },
    {
      "epoch": 44.98,
      "grad_norm": 2.533135175704956,
      "learning_rate": 4.306335989239965e-06,
      "loss": 0.5357,
      "step": 109880
    },
    {
      "epoch": 44.98,
      "grad_norm": 1.9720373153686523,
      "learning_rate": 4.3062492720458156e-06,
      "loss": 0.5361,
      "step": 109890
    },
    {
      "epoch": 44.99,
      "grad_norm": 2.3566806316375732,
      "learning_rate": 4.306162545861761e-06,
      "loss": 0.5581,
      "step": 109900
    },
    {
      "epoch": 44.99,
      "grad_norm": 1.7330695390701294,
      "learning_rate": 4.306075810688198e-06,
      "loss": 0.5529,
      "step": 109910
    },
    {
      "epoch": 44.99,
      "grad_norm": 1.9526864290237427,
      "learning_rate": 4.3059890665255245e-06,
      "loss": 0.5406,
      "step": 109920
    },
    {
      "epoch": 45.0,
      "grad_norm": 2.0619122982025146,
      "learning_rate": 4.305902313374138e-06,
      "loss": 0.5423,
      "step": 109930
    },
    {
      "epoch": 45.0,
      "eval_loss": 0.5487946271896362,
      "eval_runtime": 51.872,
      "eval_samples_per_second": 66.491,
      "eval_steps_per_second": 8.328,
      "step": 109935
    },
    {
      "epoch": 45.0,
      "grad_norm": 2.026308536529541,
      "learning_rate": 4.3058155512344355e-06,
      "loss": 0.5417,
      "step": 109940
    },
    {
      "epoch": 45.01,
      "grad_norm": 1.7971718311309814,
      "learning_rate": 4.305728780106814e-06,
      "loss": 0.565,
      "step": 109950
    },
    {
      "epoch": 45.01,
      "grad_norm": 2.2617807388305664,
      "learning_rate": 4.305641999991673e-06,
      "loss": 0.5339,
      "step": 109960
    },
    {
      "epoch": 45.01,
      "grad_norm": 1.9187620878219604,
      "learning_rate": 4.305555210889407e-06,
      "loss": 0.545,
      "step": 109970
    },
    {
      "epoch": 45.02,
      "grad_norm": 2.2780144214630127,
      "learning_rate": 4.305468412800416e-06,
      "loss": 0.5508,
      "step": 109980
    },
    {
      "epoch": 45.02,
      "grad_norm": 1.7322510480880737,
      "learning_rate": 4.305381605725096e-06,
      "loss": 0.5572,
      "step": 109990
    },
    {
      "epoch": 45.03,
      "grad_norm": 2.8127973079681396,
      "learning_rate": 4.3052947896638465e-06,
      "loss": 0.5485,
      "step": 110000
    },
    {
      "epoch": 45.03,
      "grad_norm": 1.9623520374298096,
      "learning_rate": 4.305207964617064e-06,
      "loss": 0.569,
      "step": 110010
    },
    {
      "epoch": 45.03,
      "grad_norm": 3.0131680965423584,
      "learning_rate": 4.3051211305851455e-06,
      "loss": 0.5286,
      "step": 110020
    },
    {
      "epoch": 45.04,
      "grad_norm": 2.235253095626831,
      "learning_rate": 4.30503428756849e-06,
      "loss": 0.543,
      "step": 110030
    },
    {
      "epoch": 45.04,
      "grad_norm": 1.80426025390625,
      "learning_rate": 4.304947435567495e-06,
      "loss": 0.5405,
      "step": 110040
    },
    {
      "epoch": 45.05,
      "grad_norm": 2.1215932369232178,
      "learning_rate": 4.304860574582558e-06,
      "loss": 0.5615,
      "step": 110050
    },
    {
      "epoch": 45.05,
      "grad_norm": 1.8021246194839478,
      "learning_rate": 4.304773704614078e-06,
      "loss": 0.5602,
      "step": 110060
    },
    {
      "epoch": 45.06,
      "grad_norm": 1.7372227907180786,
      "learning_rate": 4.304686825662451e-06,
      "loss": 0.5413,
      "step": 110070
    },
    {
      "epoch": 45.06,
      "grad_norm": 2.0573885440826416,
      "learning_rate": 4.304599937728078e-06,
      "loss": 0.5539,
      "step": 110080
    },
    {
      "epoch": 45.06,
      "grad_norm": 1.8103958368301392,
      "learning_rate": 4.3045130408113524e-06,
      "loss": 0.5415,
      "step": 110090
    },
    {
      "epoch": 45.07,
      "grad_norm": 2.1660075187683105,
      "learning_rate": 4.304426134912677e-06,
      "loss": 0.5467,
      "step": 110100
    },
    {
      "epoch": 45.07,
      "grad_norm": 2.5124292373657227,
      "learning_rate": 4.3043392200324464e-06,
      "loss": 0.5268,
      "step": 110110
    },
    {
      "epoch": 45.08,
      "grad_norm": 2.4886863231658936,
      "learning_rate": 4.304252296171062e-06,
      "loss": 0.5572,
      "step": 110120
    },
    {
      "epoch": 45.08,
      "grad_norm": 1.8194587230682373,
      "learning_rate": 4.304165363328919e-06,
      "loss": 0.5482,
      "step": 110130
    },
    {
      "epoch": 45.08,
      "grad_norm": 2.095961570739746,
      "learning_rate": 4.304078421506417e-06,
      "loss": 0.5528,
      "step": 110140
    },
    {
      "epoch": 45.09,
      "grad_norm": 2.235553026199341,
      "learning_rate": 4.303991470703954e-06,
      "loss": 0.57,
      "step": 110150
    },
    {
      "epoch": 45.09,
      "grad_norm": 1.9344512224197388,
      "learning_rate": 4.3039045109219285e-06,
      "loss": 0.5277,
      "step": 110160
    },
    {
      "epoch": 45.1,
      "grad_norm": 2.3542873859405518,
      "learning_rate": 4.303817542160739e-06,
      "loss": 0.5207,
      "step": 110170
    },
    {
      "epoch": 45.1,
      "grad_norm": 1.8390291929244995,
      "learning_rate": 4.303730564420783e-06,
      "loss": 0.5774,
      "step": 110180
    },
    {
      "epoch": 45.1,
      "grad_norm": 2.0833122730255127,
      "learning_rate": 4.30364357770246e-06,
      "loss": 0.5512,
      "step": 110190
    },
    {
      "epoch": 45.11,
      "grad_norm": 1.8211326599121094,
      "learning_rate": 4.303556582006169e-06,
      "loss": 0.5364,
      "step": 110200
    },
    {
      "epoch": 45.11,
      "grad_norm": 2.647876024246216,
      "learning_rate": 4.303469577332306e-06,
      "loss": 0.5291,
      "step": 110210
    },
    {
      "epoch": 45.12,
      "grad_norm": 2.6140713691711426,
      "learning_rate": 4.303382563681272e-06,
      "loss": 0.5483,
      "step": 110220
    },
    {
      "epoch": 45.12,
      "grad_norm": 2.139202356338501,
      "learning_rate": 4.303295541053464e-06,
      "loss": 0.5716,
      "step": 110230
    },
    {
      "epoch": 45.12,
      "grad_norm": 2.439274311065674,
      "learning_rate": 4.303208509449282e-06,
      "loss": 0.5551,
      "step": 110240
    },
    {
      "epoch": 45.13,
      "grad_norm": 2.436208724975586,
      "learning_rate": 4.303121468869124e-06,
      "loss": 0.5699,
      "step": 110250
    },
    {
      "epoch": 45.13,
      "grad_norm": 2.469343423843384,
      "learning_rate": 4.303034419313388e-06,
      "loss": 0.558,
      "step": 110260
    },
    {
      "epoch": 45.14,
      "grad_norm": 1.6053508520126343,
      "learning_rate": 4.302947360782474e-06,
      "loss": 0.5459,
      "step": 110270
    },
    {
      "epoch": 45.14,
      "grad_norm": 2.415367603302002,
      "learning_rate": 4.302860293276781e-06,
      "loss": 0.5553,
      "step": 110280
    },
    {
      "epoch": 45.15,
      "grad_norm": 2.1505823135375977,
      "learning_rate": 4.302773216796707e-06,
      "loss": 0.5365,
      "step": 110290
    },
    {
      "epoch": 45.15,
      "grad_norm": 2.9460256099700928,
      "learning_rate": 4.30268613134265e-06,
      "loss": 0.5524,
      "step": 110300
    },
    {
      "epoch": 45.15,
      "grad_norm": 2.1574134826660156,
      "learning_rate": 4.302599036915011e-06,
      "loss": 0.5184,
      "step": 110310
    },
    {
      "epoch": 45.16,
      "grad_norm": 3.1948933601379395,
      "learning_rate": 4.302511933514188e-06,
      "loss": 0.5353,
      "step": 110320
    },
    {
      "epoch": 45.16,
      "grad_norm": 2.123872995376587,
      "learning_rate": 4.3024248211405785e-06,
      "loss": 0.5465,
      "step": 110330
    },
    {
      "epoch": 45.17,
      "grad_norm": 2.9019012451171875,
      "learning_rate": 4.302337699794584e-06,
      "loss": 0.5329,
      "step": 110340
    },
    {
      "epoch": 45.17,
      "grad_norm": 3.028163433074951,
      "learning_rate": 4.302250569476603e-06,
      "loss": 0.5437,
      "step": 110350
    },
    {
      "epoch": 45.17,
      "grad_norm": 2.582411289215088,
      "learning_rate": 4.302163430187034e-06,
      "loss": 0.5136,
      "step": 110360
    },
    {
      "epoch": 45.18,
      "grad_norm": 2.0543854236602783,
      "learning_rate": 4.302076281926275e-06,
      "loss": 0.5371,
      "step": 110370
    },
    {
      "epoch": 45.18,
      "grad_norm": 2.405301570892334,
      "learning_rate": 4.301989124694728e-06,
      "loss": 0.5286,
      "step": 110380
    },
    {
      "epoch": 45.19,
      "grad_norm": 2.584857702255249,
      "learning_rate": 4.301901958492791e-06,
      "loss": 0.5315,
      "step": 110390
    },
    {
      "epoch": 45.19,
      "grad_norm": 2.2718687057495117,
      "learning_rate": 4.301814783320862e-06,
      "loss": 0.5509,
      "step": 110400
    },
    {
      "epoch": 45.19,
      "grad_norm": 2.504284381866455,
      "learning_rate": 4.301727599179342e-06,
      "loss": 0.5333,
      "step": 110410
    },
    {
      "epoch": 45.2,
      "grad_norm": 2.23695707321167,
      "learning_rate": 4.30164040606863e-06,
      "loss": 0.5441,
      "step": 110420
    },
    {
      "epoch": 45.2,
      "grad_norm": 1.708439588546753,
      "learning_rate": 4.3015532039891255e-06,
      "loss": 0.5363,
      "step": 110430
    },
    {
      "epoch": 45.21,
      "grad_norm": 2.1274967193603516,
      "learning_rate": 4.301465992941228e-06,
      "loss": 0.5337,
      "step": 110440
    },
    {
      "epoch": 45.21,
      "grad_norm": 3.0911338329315186,
      "learning_rate": 4.301378772925336e-06,
      "loss": 0.5608,
      "step": 110450
    },
    {
      "epoch": 45.21,
      "grad_norm": 2.016371965408325,
      "learning_rate": 4.30129154394185e-06,
      "loss": 0.5709,
      "step": 110460
    },
    {
      "epoch": 45.22,
      "grad_norm": 2.608445167541504,
      "learning_rate": 4.3012043059911686e-06,
      "loss": 0.5565,
      "step": 110470
    },
    {
      "epoch": 45.22,
      "grad_norm": 1.491384744644165,
      "learning_rate": 4.301117059073693e-06,
      "loss": 0.545,
      "step": 110480
    },
    {
      "epoch": 45.23,
      "grad_norm": 1.9723668098449707,
      "learning_rate": 4.301029803189822e-06,
      "loss": 0.5454,
      "step": 110490
    },
    {
      "epoch": 45.23,
      "grad_norm": 2.072814702987671,
      "learning_rate": 4.3009425383399555e-06,
      "loss": 0.5432,
      "step": 110500
    },
    {
      "epoch": 45.24,
      "grad_norm": 1.9666274785995483,
      "learning_rate": 4.300855264524492e-06,
      "loss": 0.5585,
      "step": 110510
    },
    {
      "epoch": 45.24,
      "grad_norm": 1.888292670249939,
      "learning_rate": 4.300767981743834e-06,
      "loss": 0.5402,
      "step": 110520
    },
    {
      "epoch": 45.24,
      "grad_norm": 1.4480324983596802,
      "learning_rate": 4.300680689998379e-06,
      "loss": 0.5336,
      "step": 110530
    },
    {
      "epoch": 45.25,
      "grad_norm": 2.260263442993164,
      "learning_rate": 4.300593389288527e-06,
      "loss": 0.5431,
      "step": 110540
    },
    {
      "epoch": 45.25,
      "grad_norm": 2.437248945236206,
      "learning_rate": 4.300506079614679e-06,
      "loss": 0.538,
      "step": 110550
    },
    {
      "epoch": 45.26,
      "grad_norm": 1.7950072288513184,
      "learning_rate": 4.3004187609772344e-06,
      "loss": 0.53,
      "step": 110560
    },
    {
      "epoch": 45.26,
      "grad_norm": 2.2952427864074707,
      "learning_rate": 4.3003314333765935e-06,
      "loss": 0.5566,
      "step": 110570
    },
    {
      "epoch": 45.26,
      "grad_norm": 1.767354965209961,
      "learning_rate": 4.300244096813156e-06,
      "loss": 0.542,
      "step": 110580
    },
    {
      "epoch": 45.27,
      "grad_norm": 2.2663774490356445,
      "learning_rate": 4.3001567512873215e-06,
      "loss": 0.5631,
      "step": 110590
    },
    {
      "epoch": 45.27,
      "grad_norm": 2.036248207092285,
      "learning_rate": 4.300069396799491e-06,
      "loss": 0.5517,
      "step": 110600
    },
    {
      "epoch": 45.28,
      "grad_norm": 2.010836362838745,
      "learning_rate": 4.299982033350065e-06,
      "loss": 0.5458,
      "step": 110610
    },
    {
      "epoch": 45.28,
      "grad_norm": 2.0340728759765625,
      "learning_rate": 4.299894660939442e-06,
      "loss": 0.5441,
      "step": 110620
    },
    {
      "epoch": 45.28,
      "grad_norm": 2.0923235416412354,
      "learning_rate": 4.299807279568023e-06,
      "loss": 0.5468,
      "step": 110630
    },
    {
      "epoch": 45.29,
      "grad_norm": 2.2576651573181152,
      "learning_rate": 4.29971988923621e-06,
      "loss": 0.5554,
      "step": 110640
    },
    {
      "epoch": 45.29,
      "grad_norm": 2.076112747192383,
      "learning_rate": 4.2996324899444e-06,
      "loss": 0.5542,
      "step": 110650
    },
    {
      "epoch": 45.3,
      "grad_norm": 2.1004226207733154,
      "learning_rate": 4.299545081692997e-06,
      "loss": 0.5382,
      "step": 110660
    },
    {
      "epoch": 45.3,
      "grad_norm": 2.4579734802246094,
      "learning_rate": 4.299457664482399e-06,
      "loss": 0.562,
      "step": 110670
    },
    {
      "epoch": 45.3,
      "grad_norm": 1.512007236480713,
      "learning_rate": 4.299370238313007e-06,
      "loss": 0.5425,
      "step": 110680
    },
    {
      "epoch": 45.31,
      "grad_norm": 2.8107011318206787,
      "learning_rate": 4.299282803185222e-06,
      "loss": 0.5418,
      "step": 110690
    },
    {
      "epoch": 45.31,
      "grad_norm": 1.8978502750396729,
      "learning_rate": 4.299195359099443e-06,
      "loss": 0.567,
      "step": 110700
    },
    {
      "epoch": 45.32,
      "grad_norm": 2.1702864170074463,
      "learning_rate": 4.2991079060560725e-06,
      "loss": 0.5541,
      "step": 110710
    },
    {
      "epoch": 45.32,
      "grad_norm": 2.212057590484619,
      "learning_rate": 4.299020444055509e-06,
      "loss": 0.5538,
      "step": 110720
    },
    {
      "epoch": 45.33,
      "grad_norm": 2.1710050106048584,
      "learning_rate": 4.2989329730981556e-06,
      "loss": 0.548,
      "step": 110730
    },
    {
      "epoch": 45.33,
      "grad_norm": 2.3033862113952637,
      "learning_rate": 4.298845493184412e-06,
      "loss": 0.5622,
      "step": 110740
    },
    {
      "epoch": 45.33,
      "grad_norm": 1.6807717084884644,
      "learning_rate": 4.298758004314679e-06,
      "loss": 0.5121,
      "step": 110750
    },
    {
      "epoch": 45.34,
      "grad_norm": 2.3271262645721436,
      "learning_rate": 4.298670506489356e-06,
      "loss": 0.5417,
      "step": 110760
    },
    {
      "epoch": 45.34,
      "grad_norm": 1.9805694818496704,
      "learning_rate": 4.298582999708844e-06,
      "loss": 0.5659,
      "step": 110770
    },
    {
      "epoch": 45.35,
      "grad_norm": 2.127378225326538,
      "learning_rate": 4.298495483973548e-06,
      "loss": 0.5562,
      "step": 110780
    },
    {
      "epoch": 45.35,
      "grad_norm": 1.985703706741333,
      "learning_rate": 4.298407959283863e-06,
      "loss": 0.5438,
      "step": 110790
    },
    {
      "epoch": 45.35,
      "grad_norm": 1.9260488748550415,
      "learning_rate": 4.298320425640193e-06,
      "loss": 0.5523,
      "step": 110800
    },
    {
      "epoch": 45.36,
      "grad_norm": 1.8620500564575195,
      "learning_rate": 4.298232883042939e-06,
      "loss": 0.5459,
      "step": 110810
    },
    {
      "epoch": 45.36,
      "grad_norm": 1.7684807777404785,
      "learning_rate": 4.298145331492502e-06,
      "loss": 0.5282,
      "step": 110820
    },
    {
      "epoch": 45.37,
      "grad_norm": 3.5282020568847656,
      "learning_rate": 4.298057770989282e-06,
      "loss": 0.5425,
      "step": 110830
    },
    {
      "epoch": 45.37,
      "grad_norm": 2.2185299396514893,
      "learning_rate": 4.297970201533681e-06,
      "loss": 0.5479,
      "step": 110840
    },
    {
      "epoch": 45.37,
      "grad_norm": 1.8418527841567993,
      "learning_rate": 4.2978826231261e-06,
      "loss": 0.5549,
      "step": 110850
    },
    {
      "epoch": 45.38,
      "grad_norm": 2.5484023094177246,
      "learning_rate": 4.2977950357669405e-06,
      "loss": 0.5693,
      "step": 110860
    },
    {
      "epoch": 45.38,
      "grad_norm": 1.5192712545394897,
      "learning_rate": 4.297707439456602e-06,
      "loss": 0.5462,
      "step": 110870
    },
    {
      "epoch": 45.39,
      "grad_norm": 2.28971529006958,
      "learning_rate": 4.297619834195488e-06,
      "loss": 0.5515,
      "step": 110880
    },
    {
      "epoch": 45.39,
      "grad_norm": 2.154573917388916,
      "learning_rate": 4.297532219983999e-06,
      "loss": 0.5441,
      "step": 110890
    },
    {
      "epoch": 45.4,
      "grad_norm": 2.558171510696411,
      "learning_rate": 4.297444596822536e-06,
      "loss": 0.5486,
      "step": 110900
    },
    {
      "epoch": 45.4,
      "grad_norm": 1.6544229984283447,
      "learning_rate": 4.297356964711501e-06,
      "loss": 0.5402,
      "step": 110910
    },
    {
      "epoch": 45.4,
      "grad_norm": 2.3077855110168457,
      "learning_rate": 4.297269323651294e-06,
      "loss": 0.5537,
      "step": 110920
    },
    {
      "epoch": 45.41,
      "grad_norm": 2.54377818107605,
      "learning_rate": 4.297181673642318e-06,
      "loss": 0.5399,
      "step": 110930
    },
    {
      "epoch": 45.41,
      "grad_norm": 2.1787822246551514,
      "learning_rate": 4.297094014684974e-06,
      "loss": 0.5558,
      "step": 110940
    },
    {
      "epoch": 45.42,
      "grad_norm": 2.3102221488952637,
      "learning_rate": 4.297006346779664e-06,
      "loss": 0.5592,
      "step": 110950
    },
    {
      "epoch": 45.42,
      "grad_norm": 2.2223405838012695,
      "learning_rate": 4.296918669926789e-06,
      "loss": 0.5522,
      "step": 110960
    },
    {
      "epoch": 45.42,
      "grad_norm": 1.8795318603515625,
      "learning_rate": 4.2968309841267515e-06,
      "loss": 0.5297,
      "step": 110970
    },
    {
      "epoch": 45.43,
      "grad_norm": 3.3164172172546387,
      "learning_rate": 4.296743289379951e-06,
      "loss": 0.5432,
      "step": 110980
    },
    {
      "epoch": 45.43,
      "grad_norm": 2.2048513889312744,
      "learning_rate": 4.296655585686792e-06,
      "loss": 0.555,
      "step": 110990
    },
    {
      "epoch": 45.44,
      "grad_norm": 2.1225523948669434,
      "learning_rate": 4.296567873047675e-06,
      "loss": 0.5599,
      "step": 111000
    },
    {
      "epoch": 45.44,
      "grad_norm": 1.8426432609558105,
      "learning_rate": 4.296480151463e-06,
      "loss": 0.5354,
      "step": 111010
    },
    {
      "epoch": 45.44,
      "grad_norm": 2.474797248840332,
      "learning_rate": 4.296392420933172e-06,
      "loss": 0.5278,
      "step": 111020
    },
    {
      "epoch": 45.45,
      "grad_norm": 2.2991600036621094,
      "learning_rate": 4.2963046814585915e-06,
      "loss": 0.5514,
      "step": 111030
    },
    {
      "epoch": 45.45,
      "grad_norm": 1.785092830657959,
      "learning_rate": 4.296216933039661e-06,
      "loss": 0.5393,
      "step": 111040
    },
    {
      "epoch": 45.46,
      "grad_norm": 2.1313116550445557,
      "learning_rate": 4.29612917567678e-06,
      "loss": 0.5534,
      "step": 111050
    },
    {
      "epoch": 45.46,
      "grad_norm": 2.1610770225524902,
      "learning_rate": 4.296041409370353e-06,
      "loss": 0.551,
      "step": 111060
    },
    {
      "epoch": 45.46,
      "grad_norm": 1.757239818572998,
      "learning_rate": 4.295953634120782e-06,
      "loss": 0.567,
      "step": 111070
    },
    {
      "epoch": 45.47,
      "grad_norm": 2.3961117267608643,
      "learning_rate": 4.295865849928468e-06,
      "loss": 0.5576,
      "step": 111080
    },
    {
      "epoch": 45.47,
      "grad_norm": 2.385838747024536,
      "learning_rate": 4.295778056793814e-06,
      "loss": 0.5542,
      "step": 111090
    },
    {
      "epoch": 45.48,
      "grad_norm": 1.647537112236023,
      "learning_rate": 4.2956902547172216e-06,
      "loss": 0.5536,
      "step": 111100
    },
    {
      "epoch": 45.48,
      "grad_norm": 1.9659533500671387,
      "learning_rate": 4.295602443699093e-06,
      "loss": 0.5531,
      "step": 111110
    },
    {
      "epoch": 45.49,
      "grad_norm": 2.858609676361084,
      "learning_rate": 4.295514623739832e-06,
      "loss": 0.5559,
      "step": 111120
    },
    {
      "epoch": 45.49,
      "grad_norm": 2.2174675464630127,
      "learning_rate": 4.295426794839837e-06,
      "loss": 0.5515,
      "step": 111130
    },
    {
      "epoch": 45.49,
      "grad_norm": 2.2201359272003174,
      "learning_rate": 4.2953389569995154e-06,
      "loss": 0.563,
      "step": 111140
    },
    {
      "epoch": 45.5,
      "grad_norm": 1.9860249757766724,
      "learning_rate": 4.295251110219266e-06,
      "loss": 0.5375,
      "step": 111150
    },
    {
      "epoch": 45.5,
      "grad_norm": 2.6984634399414062,
      "learning_rate": 4.295163254499492e-06,
      "loss": 0.5501,
      "step": 111160
    },
    {
      "epoch": 45.51,
      "grad_norm": 2.091843366622925,
      "learning_rate": 4.295075389840596e-06,
      "loss": 0.5553,
      "step": 111170
    },
    {
      "epoch": 45.51,
      "grad_norm": 2.380431652069092,
      "learning_rate": 4.29498751624298e-06,
      "loss": 0.5567,
      "step": 111180
    },
    {
      "epoch": 45.51,
      "grad_norm": 1.7526482343673706,
      "learning_rate": 4.294899633707049e-06,
      "loss": 0.5627,
      "step": 111190
    },
    {
      "epoch": 45.52,
      "grad_norm": 2.2816567420959473,
      "learning_rate": 4.294811742233202e-06,
      "loss": 0.5414,
      "step": 111200
    },
    {
      "epoch": 45.52,
      "grad_norm": 1.7397282123565674,
      "learning_rate": 4.294723841821844e-06,
      "loss": 0.5323,
      "step": 111210
    },
    {
      "epoch": 45.53,
      "grad_norm": 2.957159996032715,
      "learning_rate": 4.294635932473378e-06,
      "loss": 0.5483,
      "step": 111220
    },
    {
      "epoch": 45.53,
      "grad_norm": 2.0356366634368896,
      "learning_rate": 4.294548014188203e-06,
      "loss": 0.5694,
      "step": 111230
    },
    {
      "epoch": 45.53,
      "grad_norm": 1.9789077043533325,
      "learning_rate": 4.294460086966727e-06,
      "loss": 0.5437,
      "step": 111240
    },
    {
      "epoch": 45.54,
      "grad_norm": 2.2746644020080566,
      "learning_rate": 4.294372150809349e-06,
      "loss": 0.5352,
      "step": 111250
    },
    {
      "epoch": 45.54,
      "grad_norm": 2.7351677417755127,
      "learning_rate": 4.294284205716473e-06,
      "loss": 0.5389,
      "step": 111260
    },
    {
      "epoch": 45.55,
      "grad_norm": 2.072643995285034,
      "learning_rate": 4.294196251688502e-06,
      "loss": 0.5442,
      "step": 111270
    },
    {
      "epoch": 45.55,
      "grad_norm": 2.69563889503479,
      "learning_rate": 4.294108288725839e-06,
      "loss": 0.5532,
      "step": 111280
    },
    {
      "epoch": 45.55,
      "grad_norm": 1.9792757034301758,
      "learning_rate": 4.2940203168288864e-06,
      "loss": 0.5513,
      "step": 111290
    },
    {
      "epoch": 45.56,
      "grad_norm": 1.7613186836242676,
      "learning_rate": 4.2939323359980485e-06,
      "loss": 0.5428,
      "step": 111300
    },
    {
      "epoch": 45.56,
      "grad_norm": 2.429133653640747,
      "learning_rate": 4.293844346233726e-06,
      "loss": 0.5593,
      "step": 111310
    },
    {
      "epoch": 45.57,
      "grad_norm": 2.773261070251465,
      "learning_rate": 4.293756347536324e-06,
      "loss": 0.5061,
      "step": 111320
    },
    {
      "epoch": 45.57,
      "grad_norm": 1.8457599878311157,
      "learning_rate": 4.293668339906245e-06,
      "loss": 0.5765,
      "step": 111330
    },
    {
      "epoch": 45.58,
      "grad_norm": 2.829710006713867,
      "learning_rate": 4.293580323343892e-06,
      "loss": 0.5298,
      "step": 111340
    },
    {
      "epoch": 45.58,
      "grad_norm": 2.2396435737609863,
      "learning_rate": 4.293492297849669e-06,
      "loss": 0.564,
      "step": 111350
    },
    {
      "epoch": 45.58,
      "grad_norm": 2.0945465564727783,
      "learning_rate": 4.2934042634239766e-06,
      "loss": 0.5416,
      "step": 111360
    },
    {
      "epoch": 45.59,
      "grad_norm": 1.8886475563049316,
      "learning_rate": 4.293316220067221e-06,
      "loss": 0.533,
      "step": 111370
    },
    {
      "epoch": 45.59,
      "grad_norm": 2.0575854778289795,
      "learning_rate": 4.293228167779805e-06,
      "loss": 0.5685,
      "step": 111380
    },
    {
      "epoch": 45.6,
      "grad_norm": 2.5391762256622314,
      "learning_rate": 4.293140106562131e-06,
      "loss": 0.5525,
      "step": 111390
    },
    {
      "epoch": 45.6,
      "grad_norm": 2.008953094482422,
      "learning_rate": 4.293052036414602e-06,
      "loss": 0.5569,
      "step": 111400
    },
    {
      "epoch": 45.6,
      "grad_norm": 2.2296335697174072,
      "learning_rate": 4.292963957337623e-06,
      "loss": 0.5453,
      "step": 111410
    },
    {
      "epoch": 45.61,
      "grad_norm": 2.0532314777374268,
      "learning_rate": 4.292875869331597e-06,
      "loss": 0.5797,
      "step": 111420
    },
    {
      "epoch": 45.61,
      "grad_norm": 2.0049760341644287,
      "learning_rate": 4.292787772396927e-06,
      "loss": 0.5371,
      "step": 111430
    },
    {
      "epoch": 45.62,
      "grad_norm": 2.212844133377075,
      "learning_rate": 4.292699666534018e-06,
      "loss": 0.54,
      "step": 111440
    },
    {
      "epoch": 45.62,
      "grad_norm": 2.3054840564727783,
      "learning_rate": 4.29261155174327e-06,
      "loss": 0.5603,
      "step": 111450
    },
    {
      "epoch": 45.62,
      "grad_norm": 1.9277817010879517,
      "learning_rate": 4.29252342802509e-06,
      "loss": 0.5424,
      "step": 111460
    },
    {
      "epoch": 45.63,
      "grad_norm": 1.3668383359909058,
      "learning_rate": 4.292435295379881e-06,
      "loss": 0.5493,
      "step": 111470
    },
    {
      "epoch": 45.63,
      "grad_norm": 3.0633113384246826,
      "learning_rate": 4.292347153808047e-06,
      "loss": 0.5227,
      "step": 111480
    },
    {
      "epoch": 45.64,
      "grad_norm": 2.2346019744873047,
      "learning_rate": 4.29225900330999e-06,
      "loss": 0.5634,
      "step": 111490
    },
    {
      "epoch": 45.64,
      "grad_norm": 1.9754503965377808,
      "learning_rate": 4.292170843886116e-06,
      "loss": 0.5487,
      "step": 111500
    },
    {
      "epoch": 45.64,
      "grad_norm": 2.6116344928741455,
      "learning_rate": 4.292082675536827e-06,
      "loss": 0.5614,
      "step": 111510
    },
    {
      "epoch": 45.65,
      "grad_norm": 1.6074714660644531,
      "learning_rate": 4.291994498262529e-06,
      "loss": 0.5667,
      "step": 111520
    },
    {
      "epoch": 45.65,
      "grad_norm": 1.8517141342163086,
      "learning_rate": 4.2919063120636236e-06,
      "loss": 0.5521,
      "step": 111530
    },
    {
      "epoch": 45.66,
      "grad_norm": 2.633669137954712,
      "learning_rate": 4.291818116940517e-06,
      "loss": 0.5394,
      "step": 111540
    },
    {
      "epoch": 45.66,
      "grad_norm": 2.5273728370666504,
      "learning_rate": 4.291729912893611e-06,
      "loss": 0.5625,
      "step": 111550
    },
    {
      "epoch": 45.67,
      "grad_norm": 2.214390993118286,
      "learning_rate": 4.291641699923311e-06,
      "loss": 0.5379,
      "step": 111560
    },
    {
      "epoch": 45.67,
      "grad_norm": 2.084855079650879,
      "learning_rate": 4.29155347803002e-06,
      "loss": 0.5401,
      "step": 111570
    },
    {
      "epoch": 45.67,
      "grad_norm": 2.0175771713256836,
      "learning_rate": 4.291465247214144e-06,
      "loss": 0.5569,
      "step": 111580
    },
    {
      "epoch": 45.68,
      "grad_norm": 2.0368213653564453,
      "learning_rate": 4.291377007476086e-06,
      "loss": 0.5713,
      "step": 111590
    },
    {
      "epoch": 45.68,
      "grad_norm": 1.7109731435775757,
      "learning_rate": 4.29128875881625e-06,
      "loss": 0.5389,
      "step": 111600
    },
    {
      "epoch": 45.69,
      "grad_norm": 2.1033003330230713,
      "learning_rate": 4.291200501235041e-06,
      "loss": 0.5577,
      "step": 111610
    },
    {
      "epoch": 45.69,
      "grad_norm": 2.1998355388641357,
      "learning_rate": 4.291112234732862e-06,
      "loss": 0.5508,
      "step": 111620
    },
    {
      "epoch": 45.69,
      "grad_norm": 1.6130610704421997,
      "learning_rate": 4.291023959310119e-06,
      "loss": 0.5468,
      "step": 111630
    },
    {
      "epoch": 45.7,
      "grad_norm": 2.4529967308044434,
      "learning_rate": 4.290935674967215e-06,
      "loss": 0.5282,
      "step": 111640
    },
    {
      "epoch": 45.7,
      "grad_norm": 1.7721039056777954,
      "learning_rate": 4.2908473817045555e-06,
      "loss": 0.5498,
      "step": 111650
    },
    {
      "epoch": 45.71,
      "grad_norm": 1.7633609771728516,
      "learning_rate": 4.290759079522544e-06,
      "loss": 0.5609,
      "step": 111660
    },
    {
      "epoch": 45.71,
      "grad_norm": 2.2628471851348877,
      "learning_rate": 4.290670768421585e-06,
      "loss": 0.5559,
      "step": 111670
    },
    {
      "epoch": 45.71,
      "grad_norm": 1.8639777898788452,
      "learning_rate": 4.290582448402085e-06,
      "loss": 0.5587,
      "step": 111680
    },
    {
      "epoch": 45.72,
      "grad_norm": 2.2311453819274902,
      "learning_rate": 4.290494119464445e-06,
      "loss": 0.5533,
      "step": 111690
    },
    {
      "epoch": 45.72,
      "grad_norm": 2.3229820728302,
      "learning_rate": 4.290405781609074e-06,
      "loss": 0.5367,
      "step": 111700
    },
    {
      "epoch": 45.73,
      "grad_norm": 2.219374656677246,
      "learning_rate": 4.290317434836373e-06,
      "loss": 0.5323,
      "step": 111710
    },
    {
      "epoch": 45.73,
      "grad_norm": 1.7430914640426636,
      "learning_rate": 4.290229079146749e-06,
      "loss": 0.5338,
      "step": 111720
    },
    {
      "epoch": 45.73,
      "grad_norm": 2.2688183784484863,
      "learning_rate": 4.290140714540604e-06,
      "loss": 0.5385,
      "step": 111730
    },
    {
      "epoch": 45.74,
      "grad_norm": 2.2403886318206787,
      "learning_rate": 4.2900523410183455e-06,
      "loss": 0.5512,
      "step": 111740
    },
    {
      "epoch": 45.74,
      "grad_norm": 2.0415642261505127,
      "learning_rate": 4.289963958580377e-06,
      "loss": 0.5525,
      "step": 111750
    },
    {
      "epoch": 45.75,
      "grad_norm": 1.9206568002700806,
      "learning_rate": 4.289875567227105e-06,
      "loss": 0.5501,
      "step": 111760
    },
    {
      "epoch": 45.75,
      "grad_norm": 2.1618576049804688,
      "learning_rate": 4.289787166958932e-06,
      "loss": 0.5485,
      "step": 111770
    },
    {
      "epoch": 45.76,
      "grad_norm": 1.7907557487487793,
      "learning_rate": 4.289698757776265e-06,
      "loss": 0.5513,
      "step": 111780
    },
    {
      "epoch": 45.76,
      "grad_norm": 1.8333942890167236,
      "learning_rate": 4.2896103396795075e-06,
      "loss": 0.5387,
      "step": 111790
    },
    {
      "epoch": 45.76,
      "grad_norm": 1.9552288055419922,
      "learning_rate": 4.2895219126690654e-06,
      "loss": 0.5487,
      "step": 111800
    },
    {
      "epoch": 45.77,
      "grad_norm": 1.684877872467041,
      "learning_rate": 4.289433476745344e-06,
      "loss": 0.5523,
      "step": 111810
    },
    {
      "epoch": 45.77,
      "grad_norm": 2.6798160076141357,
      "learning_rate": 4.289345031908746e-06,
      "loss": 0.5381,
      "step": 111820
    },
    {
      "epoch": 45.78,
      "grad_norm": 2.614635705947876,
      "learning_rate": 4.2892565781596805e-06,
      "loss": 0.5454,
      "step": 111830
    },
    {
      "epoch": 45.78,
      "grad_norm": 2.571258783340454,
      "learning_rate": 4.28916811549855e-06,
      "loss": 0.5579,
      "step": 111840
    },
    {
      "epoch": 45.78,
      "grad_norm": 2.1559090614318848,
      "learning_rate": 4.28907964392576e-06,
      "loss": 0.5643,
      "step": 111850
    },
    {
      "epoch": 45.79,
      "grad_norm": 2.2023069858551025,
      "learning_rate": 4.288991163441716e-06,
      "loss": 0.5651,
      "step": 111860
    },
    {
      "epoch": 45.79,
      "grad_norm": 2.1512155532836914,
      "learning_rate": 4.288902674046825e-06,
      "loss": 0.517,
      "step": 111870
    },
    {
      "epoch": 45.8,
      "grad_norm": 2.411074161529541,
      "learning_rate": 4.28881417574149e-06,
      "loss": 0.5565,
      "step": 111880
    },
    {
      "epoch": 45.8,
      "grad_norm": 1.70357346534729,
      "learning_rate": 4.288725668526116e-06,
      "loss": 0.5461,
      "step": 111890
    },
    {
      "epoch": 45.8,
      "grad_norm": 2.031454086303711,
      "learning_rate": 4.288637152401112e-06,
      "loss": 0.5586,
      "step": 111900
    },
    {
      "epoch": 45.81,
      "grad_norm": 2.3824374675750732,
      "learning_rate": 4.28854862736688e-06,
      "loss": 0.5274,
      "step": 111910
    },
    {
      "epoch": 45.81,
      "grad_norm": 2.3950088024139404,
      "learning_rate": 4.288460093423826e-06,
      "loss": 0.552,
      "step": 111920
    },
    {
      "epoch": 45.82,
      "grad_norm": 1.7030506134033203,
      "learning_rate": 4.288371550572357e-06,
      "loss": 0.5651,
      "step": 111930
    },
    {
      "epoch": 45.82,
      "grad_norm": 3.1455228328704834,
      "learning_rate": 4.288282998812878e-06,
      "loss": 0.5553,
      "step": 111940
    },
    {
      "epoch": 45.82,
      "grad_norm": 1.740113377571106,
      "learning_rate": 4.288194438145793e-06,
      "loss": 0.5476,
      "step": 111950
    },
    {
      "epoch": 45.83,
      "grad_norm": 2.446333646774292,
      "learning_rate": 4.288105868571511e-06,
      "loss": 0.5339,
      "step": 111960
    },
    {
      "epoch": 45.83,
      "grad_norm": 2.4718031883239746,
      "learning_rate": 4.288017290090435e-06,
      "loss": 0.561,
      "step": 111970
    },
    {
      "epoch": 45.84,
      "grad_norm": 1.898261308670044,
      "learning_rate": 4.287928702702972e-06,
      "loss": 0.5508,
      "step": 111980
    },
    {
      "epoch": 45.84,
      "grad_norm": 3.114190101623535,
      "learning_rate": 4.287840106409527e-06,
      "loss": 0.5577,
      "step": 111990
    },
    {
      "epoch": 45.85,
      "grad_norm": 2.0411739349365234,
      "learning_rate": 4.287751501210507e-06,
      "loss": 0.5425,
      "step": 112000
    },
    {
      "epoch": 45.85,
      "grad_norm": 2.460784673690796,
      "learning_rate": 4.287662887106316e-06,
      "loss": 0.5353,
      "step": 112010
    },
    {
      "epoch": 45.85,
      "grad_norm": 2.0470399856567383,
      "learning_rate": 4.287574264097363e-06,
      "loss": 0.5429,
      "step": 112020
    },
    {
      "epoch": 45.86,
      "grad_norm": 1.9688833951950073,
      "learning_rate": 4.2874856321840505e-06,
      "loss": 0.5643,
      "step": 112030
    },
    {
      "epoch": 45.86,
      "grad_norm": 1.6744104623794556,
      "learning_rate": 4.287396991366787e-06,
      "loss": 0.5266,
      "step": 112040
    },
    {
      "epoch": 45.87,
      "grad_norm": 2.2281973361968994,
      "learning_rate": 4.287308341645977e-06,
      "loss": 0.5303,
      "step": 112050
    },
    {
      "epoch": 45.87,
      "grad_norm": 2.442473888397217,
      "learning_rate": 4.287219683022028e-06,
      "loss": 0.5524,
      "step": 112060
    },
    {
      "epoch": 45.87,
      "grad_norm": 2.481161117553711,
      "learning_rate": 4.287131015495344e-06,
      "loss": 0.5342,
      "step": 112070
    },
    {
      "epoch": 45.88,
      "grad_norm": 2.654134750366211,
      "learning_rate": 4.287042339066334e-06,
      "loss": 0.5491,
      "step": 112080
    },
    {
      "epoch": 45.88,
      "grad_norm": 2.3059804439544678,
      "learning_rate": 4.286953653735402e-06,
      "loss": 0.5464,
      "step": 112090
    },
    {
      "epoch": 45.89,
      "grad_norm": 2.482410430908203,
      "learning_rate": 4.2868649595029555e-06,
      "loss": 0.5661,
      "step": 112100
    },
    {
      "epoch": 45.89,
      "grad_norm": 2.390789747238159,
      "learning_rate": 4.2867762563694e-06,
      "loss": 0.5232,
      "step": 112110
    },
    {
      "epoch": 45.89,
      "grad_norm": 2.8376336097717285,
      "learning_rate": 4.286687544335142e-06,
      "loss": 0.5392,
      "step": 112120
    },
    {
      "epoch": 45.9,
      "grad_norm": 1.690274715423584,
      "learning_rate": 4.2865988234005875e-06,
      "loss": 0.5364,
      "step": 112130
    },
    {
      "epoch": 45.9,
      "grad_norm": 1.8579392433166504,
      "learning_rate": 4.286510093566145e-06,
      "loss": 0.531,
      "step": 112140
    },
    {
      "epoch": 45.91,
      "grad_norm": 1.8283556699752808,
      "learning_rate": 4.286421354832218e-06,
      "loss": 0.554,
      "step": 112150
    },
    {
      "epoch": 45.91,
      "grad_norm": 2.542834997177124,
      "learning_rate": 4.286332607199215e-06,
      "loss": 0.5465,
      "step": 112160
    },
    {
      "epoch": 45.91,
      "grad_norm": 1.72517991065979,
      "learning_rate": 4.286243850667541e-06,
      "loss": 0.5203,
      "step": 112170
    },
    {
      "epoch": 45.92,
      "grad_norm": 2.3249549865722656,
      "learning_rate": 4.286155085237605e-06,
      "loss": 0.5456,
      "step": 112180
    },
    {
      "epoch": 45.92,
      "grad_norm": 1.6573406457901,
      "learning_rate": 4.286066310909811e-06,
      "loss": 0.5428,
      "step": 112190
    },
    {
      "epoch": 45.93,
      "grad_norm": 2.556246757507324,
      "learning_rate": 4.285977527684567e-06,
      "loss": 0.5393,
      "step": 112200
    },
    {
      "epoch": 45.93,
      "grad_norm": 1.5692976713180542,
      "learning_rate": 4.28588873556228e-06,
      "loss": 0.5459,
      "step": 112210
    },
    {
      "epoch": 45.94,
      "grad_norm": 2.5512359142303467,
      "learning_rate": 4.285799934543355e-06,
      "loss": 0.5685,
      "step": 112220
    },
    {
      "epoch": 45.94,
      "grad_norm": 2.4492056369781494,
      "learning_rate": 4.2857111246282016e-06,
      "loss": 0.5507,
      "step": 112230
    },
    {
      "epoch": 45.94,
      "grad_norm": 2.057232141494751,
      "learning_rate": 4.285622305817224e-06,
      "loss": 0.5329,
      "step": 112240
    },
    {
      "epoch": 45.95,
      "grad_norm": 2.054408550262451,
      "learning_rate": 4.2855334781108304e-06,
      "loss": 0.5676,
      "step": 112250
    },
    {
      "epoch": 45.95,
      "grad_norm": 1.795770287513733,
      "learning_rate": 4.285444641509428e-06,
      "loss": 0.5279,
      "step": 112260
    },
    {
      "epoch": 45.96,
      "grad_norm": 2.4650559425354004,
      "learning_rate": 4.285355796013423e-06,
      "loss": 0.5458,
      "step": 112270
    },
    {
      "epoch": 45.96,
      "grad_norm": 2.382462739944458,
      "learning_rate": 4.285266941623221e-06,
      "loss": 0.5483,
      "step": 112280
    },
    {
      "epoch": 45.96,
      "grad_norm": 2.04280686378479,
      "learning_rate": 4.285178078339232e-06,
      "loss": 0.5509,
      "step": 112290
    },
    {
      "epoch": 45.97,
      "grad_norm": 1.7959836721420288,
      "learning_rate": 4.285089206161861e-06,
      "loss": 0.5651,
      "step": 112300
    },
    {
      "epoch": 45.97,
      "grad_norm": 1.8057501316070557,
      "learning_rate": 4.285000325091517e-06,
      "loss": 0.5094,
      "step": 112310
    },
    {
      "epoch": 45.98,
      "grad_norm": 2.1960372924804688,
      "learning_rate": 4.2849114351286045e-06,
      "loss": 0.5596,
      "step": 112320
    },
    {
      "epoch": 45.98,
      "grad_norm": 1.7624294757843018,
      "learning_rate": 4.284822536273533e-06,
      "loss": 0.5436,
      "step": 112330
    },
    {
      "epoch": 45.98,
      "grad_norm": 2.0445127487182617,
      "learning_rate": 4.284733628526708e-06,
      "loss": 0.5244,
      "step": 112340
    },
    {
      "epoch": 45.99,
      "grad_norm": 2.025289535522461,
      "learning_rate": 4.284644711888538e-06,
      "loss": 0.5675,
      "step": 112350
    },
    {
      "epoch": 45.99,
      "grad_norm": 2.599557399749756,
      "learning_rate": 4.28455578635943e-06,
      "loss": 0.5414,
      "step": 112360
    },
    {
      "epoch": 46.0,
      "grad_norm": 1.9528369903564453,
      "learning_rate": 4.284466851939791e-06,
      "loss": 0.5431,
      "step": 112370
    },
    {
      "epoch": 46.0,
      "eval_loss": 0.5465502142906189,
      "eval_runtime": 52.5356,
      "eval_samples_per_second": 65.651,
      "eval_steps_per_second": 8.223,
      "step": 112378
    },
    {
      "epoch": 46.0,
      "grad_norm": 2.2961645126342773,
      "learning_rate": 4.284377908630029e-06,
      "loss": 0.554,
      "step": 112380
    },
    {
      "epoch": 46.0,
      "grad_norm": 2.1333277225494385,
      "learning_rate": 4.284288956430551e-06,
      "loss": 0.551,
      "step": 112390
    },
    {
      "epoch": 46.01,
      "grad_norm": 1.9136372804641724,
      "learning_rate": 4.284199995341765e-06,
      "loss": 0.5485,
      "step": 112400
    },
    {
      "epoch": 46.01,
      "grad_norm": 2.9078574180603027,
      "learning_rate": 4.284111025364077e-06,
      "loss": 0.5462,
      "step": 112410
    },
    {
      "epoch": 46.02,
      "grad_norm": 2.9716603755950928,
      "learning_rate": 4.284022046497896e-06,
      "loss": 0.5637,
      "step": 112420
    },
    {
      "epoch": 46.02,
      "grad_norm": 3.545471668243408,
      "learning_rate": 4.28393305874363e-06,
      "loss": 0.5375,
      "step": 112430
    },
    {
      "epoch": 46.03,
      "grad_norm": 1.887412428855896,
      "learning_rate": 4.283844062101685e-06,
      "loss": 0.5525,
      "step": 112440
    },
    {
      "epoch": 46.03,
      "grad_norm": 3.2251269817352295,
      "learning_rate": 4.28375505657247e-06,
      "loss": 0.5296,
      "step": 112450
    },
    {
      "epoch": 46.03,
      "grad_norm": 2.387582302093506,
      "learning_rate": 4.2836660421563915e-06,
      "loss": 0.5495,
      "step": 112460
    },
    {
      "epoch": 46.04,
      "grad_norm": 2.4835309982299805,
      "learning_rate": 4.283577018853859e-06,
      "loss": 0.5374,
      "step": 112470
    },
    {
      "epoch": 46.04,
      "grad_norm": 1.811398983001709,
      "learning_rate": 4.283487986665279e-06,
      "loss": 0.5513,
      "step": 112480
    },
    {
      "epoch": 46.05,
      "grad_norm": 1.6761281490325928,
      "learning_rate": 4.28339894559106e-06,
      "loss": 0.5583,
      "step": 112490
    },
    {
      "epoch": 46.05,
      "grad_norm": 2.6258699893951416,
      "learning_rate": 4.28330989563161e-06,
      "loss": 0.557,
      "step": 112500
    },
    {
      "epoch": 46.05,
      "grad_norm": 2.223022699356079,
      "learning_rate": 4.2832208367873355e-06,
      "loss": 0.5336,
      "step": 112510
    },
    {
      "epoch": 46.06,
      "grad_norm": 3.017155408859253,
      "learning_rate": 4.283131769058646e-06,
      "loss": 0.5421,
      "step": 112520
    },
    {
      "epoch": 46.06,
      "grad_norm": 2.0312044620513916,
      "learning_rate": 4.283042692445949e-06,
      "loss": 0.5378,
      "step": 112530
    },
    {
      "epoch": 46.07,
      "grad_norm": 2.3128042221069336,
      "learning_rate": 4.282953606949652e-06,
      "loss": 0.5309,
      "step": 112540
    },
    {
      "epoch": 46.07,
      "grad_norm": 1.8273918628692627,
      "learning_rate": 4.282864512570163e-06,
      "loss": 0.5619,
      "step": 112550
    },
    {
      "epoch": 46.07,
      "grad_norm": 1.9847776889801025,
      "learning_rate": 4.282775409307892e-06,
      "loss": 0.5606,
      "step": 112560
    },
    {
      "epoch": 46.08,
      "grad_norm": 2.501532793045044,
      "learning_rate": 4.282686297163246e-06,
      "loss": 0.5551,
      "step": 112570
    },
    {
      "epoch": 46.08,
      "grad_norm": 1.6010257005691528,
      "learning_rate": 4.2825971761366326e-06,
      "loss": 0.5265,
      "step": 112580
    },
    {
      "epoch": 46.09,
      "grad_norm": 2.0331456661224365,
      "learning_rate": 4.28250804622846e-06,
      "loss": 0.5625,
      "step": 112590
    },
    {
      "epoch": 46.09,
      "grad_norm": 2.1980807781219482,
      "learning_rate": 4.282418907439138e-06,
      "loss": 0.5627,
      "step": 112600
    },
    {
      "epoch": 46.09,
      "grad_norm": 2.1405646800994873,
      "learning_rate": 4.2823297597690734e-06,
      "loss": 0.5463,
      "step": 112610
    },
    {
      "epoch": 46.1,
      "grad_norm": 1.8661330938339233,
      "learning_rate": 4.282240603218676e-06,
      "loss": 0.5634,
      "step": 112620
    },
    {
      "epoch": 46.1,
      "grad_norm": 2.1331675052642822,
      "learning_rate": 4.282151437788352e-06,
      "loss": 0.5356,
      "step": 112630
    },
    {
      "epoch": 46.11,
      "grad_norm": 1.7283759117126465,
      "learning_rate": 4.282062263478513e-06,
      "loss": 0.5369,
      "step": 112640
    },
    {
      "epoch": 46.11,
      "grad_norm": 2.0258073806762695,
      "learning_rate": 4.281973080289564e-06,
      "loss": 0.5372,
      "step": 112650
    },
    {
      "epoch": 46.12,
      "grad_norm": 2.359787702560425,
      "learning_rate": 4.2818838882219155e-06,
      "loss": 0.5428,
      "step": 112660
    },
    {
      "epoch": 46.12,
      "grad_norm": 2.226189374923706,
      "learning_rate": 4.281794687275977e-06,
      "loss": 0.5359,
      "step": 112670
    },
    {
      "epoch": 46.12,
      "grad_norm": 1.901937484741211,
      "learning_rate": 4.281705477452154e-06,
      "loss": 0.5309,
      "step": 112680
    },
    {
      "epoch": 46.13,
      "grad_norm": 2.315681219100952,
      "learning_rate": 4.281616258750858e-06,
      "loss": 0.5574,
      "step": 112690
    },
    {
      "epoch": 46.13,
      "grad_norm": 2.256680965423584,
      "learning_rate": 4.281527031172497e-06,
      "loss": 0.5213,
      "step": 112700
    },
    {
      "epoch": 46.14,
      "grad_norm": 2.745357036590576,
      "learning_rate": 4.281437794717481e-06,
      "loss": 0.5324,
      "step": 112710
    },
    {
      "epoch": 46.14,
      "grad_norm": 2.2972054481506348,
      "learning_rate": 4.2813485493862155e-06,
      "loss": 0.5416,
      "step": 112720
    },
    {
      "epoch": 46.14,
      "grad_norm": 2.4134464263916016,
      "learning_rate": 4.281259295179111e-06,
      "loss": 0.553,
      "step": 112730
    },
    {
      "epoch": 46.15,
      "grad_norm": 1.9864673614501953,
      "learning_rate": 4.281170032096577e-06,
      "loss": 0.5561,
      "step": 112740
    },
    {
      "epoch": 46.15,
      "grad_norm": 3.182713747024536,
      "learning_rate": 4.281080760139022e-06,
      "loss": 0.5722,
      "step": 112750
    },
    {
      "epoch": 46.16,
      "grad_norm": 2.2927913665771484,
      "learning_rate": 4.280991479306854e-06,
      "loss": 0.5621,
      "step": 112760
    },
    {
      "epoch": 46.16,
      "grad_norm": 1.6874405145645142,
      "learning_rate": 4.280902189600484e-06,
      "loss": 0.565,
      "step": 112770
    },
    {
      "epoch": 46.16,
      "grad_norm": 2.0551257133483887,
      "learning_rate": 4.2808128910203195e-06,
      "loss": 0.5294,
      "step": 112780
    },
    {
      "epoch": 46.17,
      "grad_norm": 2.4894888401031494,
      "learning_rate": 4.280723583566769e-06,
      "loss": 0.543,
      "step": 112790
    },
    {
      "epoch": 46.17,
      "grad_norm": 2.7306466102600098,
      "learning_rate": 4.280634267240244e-06,
      "loss": 0.5546,
      "step": 112800
    },
    {
      "epoch": 46.18,
      "grad_norm": 2.18550705909729,
      "learning_rate": 4.280544942041151e-06,
      "loss": 0.5496,
      "step": 112810
    },
    {
      "epoch": 46.18,
      "grad_norm": 2.282949924468994,
      "learning_rate": 4.2804556079699005e-06,
      "loss": 0.5427,
      "step": 112820
    },
    {
      "epoch": 46.19,
      "grad_norm": 2.105482339859009,
      "learning_rate": 4.280366265026901e-06,
      "loss": 0.5369,
      "step": 112830
    },
    {
      "epoch": 46.19,
      "grad_norm": 2.127336263656616,
      "learning_rate": 4.280276913212563e-06,
      "loss": 0.5465,
      "step": 112840
    },
    {
      "epoch": 46.19,
      "grad_norm": 2.360576629638672,
      "learning_rate": 4.280187552527296e-06,
      "loss": 0.5491,
      "step": 112850
    },
    {
      "epoch": 46.2,
      "grad_norm": 1.8459577560424805,
      "learning_rate": 4.280098182971507e-06,
      "loss": 0.536,
      "step": 112860
    },
    {
      "epoch": 46.2,
      "grad_norm": 1.7571049928665161,
      "learning_rate": 4.280008804545607e-06,
      "loss": 0.5771,
      "step": 112870
    },
    {
      "epoch": 46.21,
      "grad_norm": 2.149618148803711,
      "learning_rate": 4.279919417250006e-06,
      "loss": 0.5292,
      "step": 112880
    },
    {
      "epoch": 46.21,
      "grad_norm": 1.9342479705810547,
      "learning_rate": 4.279830021085113e-06,
      "loss": 0.545,
      "step": 112890
    },
    {
      "epoch": 46.21,
      "grad_norm": 2.860739231109619,
      "learning_rate": 4.279740616051336e-06,
      "loss": 0.5412,
      "step": 112900
    },
    {
      "epoch": 46.22,
      "grad_norm": 2.0200040340423584,
      "learning_rate": 4.279651202149086e-06,
      "loss": 0.5457,
      "step": 112910
    },
    {
      "epoch": 46.22,
      "grad_norm": 2.161872148513794,
      "learning_rate": 4.279561779378774e-06,
      "loss": 0.5555,
      "step": 112920
    },
    {
      "epoch": 46.23,
      "grad_norm": 1.6996279954910278,
      "learning_rate": 4.2794723477408066e-06,
      "loss": 0.5445,
      "step": 112930
    },
    {
      "epoch": 46.23,
      "grad_norm": 1.936248540878296,
      "learning_rate": 4.2793829072355954e-06,
      "loss": 0.5468,
      "step": 112940
    },
    {
      "epoch": 46.23,
      "grad_norm": 1.9945621490478516,
      "learning_rate": 4.2792934578635485e-06,
      "loss": 0.5313,
      "step": 112950
    },
    {
      "epoch": 46.24,
      "grad_norm": 1.7651760578155518,
      "learning_rate": 4.279203999625078e-06,
      "loss": 0.5446,
      "step": 112960
    },
    {
      "epoch": 46.24,
      "grad_norm": 2.202897071838379,
      "learning_rate": 4.279114532520592e-06,
      "loss": 0.5361,
      "step": 112970
    },
    {
      "epoch": 46.25,
      "grad_norm": 1.960799217224121,
      "learning_rate": 4.279025056550501e-06,
      "loss": 0.5463,
      "step": 112980
    },
    {
      "epoch": 46.25,
      "grad_norm": 2.2067224979400635,
      "learning_rate": 4.278935571715215e-06,
      "loss": 0.5593,
      "step": 112990
    },
    {
      "epoch": 46.25,
      "grad_norm": 1.775704264640808,
      "learning_rate": 4.278846078015143e-06,
      "loss": 0.5257,
      "step": 113000
    },
    {
      "epoch": 46.26,
      "grad_norm": 2.0914158821105957,
      "learning_rate": 4.278756575450696e-06,
      "loss": 0.5559,
      "step": 113010
    },
    {
      "epoch": 46.26,
      "grad_norm": 2.2346394062042236,
      "learning_rate": 4.278667064022284e-06,
      "loss": 0.5612,
      "step": 113020
    },
    {
      "epoch": 46.27,
      "grad_norm": 2.019641637802124,
      "learning_rate": 4.2785775437303156e-06,
      "loss": 0.5466,
      "step": 113030
    },
    {
      "epoch": 46.27,
      "grad_norm": 2.6669890880584717,
      "learning_rate": 4.278488014575203e-06,
      "loss": 0.5769,
      "step": 113040
    },
    {
      "epoch": 46.28,
      "grad_norm": 2.8808634281158447,
      "learning_rate": 4.278398476557354e-06,
      "loss": 0.5506,
      "step": 113050
    },
    {
      "epoch": 46.28,
      "grad_norm": 2.856337070465088,
      "learning_rate": 4.278308929677181e-06,
      "loss": 0.5578,
      "step": 113060
    },
    {
      "epoch": 46.28,
      "grad_norm": 2.5934345722198486,
      "learning_rate": 4.278219373935092e-06,
      "loss": 0.5285,
      "step": 113070
    },
    {
      "epoch": 46.29,
      "grad_norm": 1.8240798711776733,
      "learning_rate": 4.278129809331499e-06,
      "loss": 0.5566,
      "step": 113080
    },
    {
      "epoch": 46.29,
      "grad_norm": 2.55441951751709,
      "learning_rate": 4.278040235866811e-06,
      "loss": 0.5581,
      "step": 113090
    },
    {
      "epoch": 46.3,
      "grad_norm": 2.0076582431793213,
      "learning_rate": 4.27795065354144e-06,
      "loss": 0.5572,
      "step": 113100
    },
    {
      "epoch": 46.3,
      "grad_norm": 1.9292609691619873,
      "learning_rate": 4.277861062355795e-06,
      "loss": 0.5518,
      "step": 113110
    },
    {
      "epoch": 46.3,
      "grad_norm": 2.3604981899261475,
      "learning_rate": 4.277771462310287e-06,
      "loss": 0.5385,
      "step": 113120
    },
    {
      "epoch": 46.31,
      "grad_norm": 1.8493958711624146,
      "learning_rate": 4.2776818534053254e-06,
      "loss": 0.5644,
      "step": 113130
    },
    {
      "epoch": 46.31,
      "grad_norm": 2.1062474250793457,
      "learning_rate": 4.277592235641322e-06,
      "loss": 0.5499,
      "step": 113140
    },
    {
      "epoch": 46.32,
      "grad_norm": 1.6302754878997803,
      "learning_rate": 4.277502609018686e-06,
      "loss": 0.5551,
      "step": 113150
    },
    {
      "epoch": 46.32,
      "grad_norm": 1.676235318183899,
      "learning_rate": 4.277412973537831e-06,
      "loss": 0.5411,
      "step": 113160
    },
    {
      "epoch": 46.32,
      "grad_norm": 2.1222102642059326,
      "learning_rate": 4.277323329199164e-06,
      "loss": 0.5516,
      "step": 113170
    },
    {
      "epoch": 46.33,
      "grad_norm": 2.528083562850952,
      "learning_rate": 4.277233676003096e-06,
      "loss": 0.5397,
      "step": 113180
    },
    {
      "epoch": 46.33,
      "grad_norm": 1.722929835319519,
      "learning_rate": 4.27714401395004e-06,
      "loss": 0.5378,
      "step": 113190
    },
    {
      "epoch": 46.34,
      "grad_norm": 2.124159812927246,
      "learning_rate": 4.277054343040405e-06,
      "loss": 0.5302,
      "step": 113200
    },
    {
      "epoch": 46.34,
      "grad_norm": 2.323035955429077,
      "learning_rate": 4.276964663274603e-06,
      "loss": 0.5533,
      "step": 113210
    },
    {
      "epoch": 46.34,
      "grad_norm": 1.982711911201477,
      "learning_rate": 4.2768749746530435e-06,
      "loss": 0.5523,
      "step": 113220
    },
    {
      "epoch": 46.35,
      "grad_norm": 1.8605588674545288,
      "learning_rate": 4.276785277176137e-06,
      "loss": 0.532,
      "step": 113230
    },
    {
      "epoch": 46.35,
      "grad_norm": 1.8163785934448242,
      "learning_rate": 4.276695570844296e-06,
      "loss": 0.5307,
      "step": 113240
    },
    {
      "epoch": 46.36,
      "grad_norm": 2.4714906215667725,
      "learning_rate": 4.276605855657931e-06,
      "loss": 0.5184,
      "step": 113250
    },
    {
      "epoch": 46.36,
      "grad_norm": 2.130504846572876,
      "learning_rate": 4.276516131617452e-06,
      "loss": 0.5504,
      "step": 113260
    },
    {
      "epoch": 46.37,
      "grad_norm": 1.864415168762207,
      "learning_rate": 4.276426398723272e-06,
      "loss": 0.5391,
      "step": 113270
    },
    {
      "epoch": 46.37,
      "grad_norm": 1.7617889642715454,
      "learning_rate": 4.2763366569758e-06,
      "loss": 0.5451,
      "step": 113280
    },
    {
      "epoch": 46.37,
      "grad_norm": 2.5835115909576416,
      "learning_rate": 4.276246906375447e-06,
      "loss": 0.5265,
      "step": 113290
    },
    {
      "epoch": 46.38,
      "grad_norm": 2.0964269638061523,
      "learning_rate": 4.2761571469226256e-06,
      "loss": 0.5366,
      "step": 113300
    },
    {
      "epoch": 46.38,
      "grad_norm": 1.9098113775253296,
      "learning_rate": 4.276067378617746e-06,
      "loss": 0.5672,
      "step": 113310
    },
    {
      "epoch": 46.39,
      "grad_norm": 2.296308994293213,
      "learning_rate": 4.275977601461221e-06,
      "loss": 0.5562,
      "step": 113320
    },
    {
      "epoch": 46.39,
      "grad_norm": 2.5860655307769775,
      "learning_rate": 4.2758878154534595e-06,
      "loss": 0.5515,
      "step": 113330
    },
    {
      "epoch": 46.39,
      "grad_norm": 2.4995675086975098,
      "learning_rate": 4.275798020594874e-06,
      "loss": 0.5327,
      "step": 113340
    },
    {
      "epoch": 46.4,
      "grad_norm": 2.0752758979797363,
      "learning_rate": 4.275708216885875e-06,
      "loss": 0.5438,
      "step": 113350
    },
    {
      "epoch": 46.4,
      "grad_norm": 1.9697574377059937,
      "learning_rate": 4.2756184043268756e-06,
      "loss": 0.525,
      "step": 113360
    },
    {
      "epoch": 46.41,
      "grad_norm": 2.0400209426879883,
      "learning_rate": 4.275528582918287e-06,
      "loss": 0.5268,
      "step": 113370
    },
    {
      "epoch": 46.41,
      "grad_norm": 1.8039902448654175,
      "learning_rate": 4.275438752660518e-06,
      "loss": 0.5421,
      "step": 113380
    },
    {
      "epoch": 46.41,
      "grad_norm": 2.3744537830352783,
      "learning_rate": 4.275348913553983e-06,
      "loss": 0.5493,
      "step": 113390
    },
    {
      "epoch": 46.42,
      "grad_norm": 2.0047786235809326,
      "learning_rate": 4.2752590655990924e-06,
      "loss": 0.565,
      "step": 113400
    },
    {
      "epoch": 46.42,
      "grad_norm": 2.15767765045166,
      "learning_rate": 4.275169208796258e-06,
      "loss": 0.5323,
      "step": 113410
    },
    {
      "epoch": 46.43,
      "grad_norm": 1.7862350940704346,
      "learning_rate": 4.275079343145891e-06,
      "loss": 0.5472,
      "step": 113420
    },
    {
      "epoch": 46.43,
      "grad_norm": 1.7650080919265747,
      "learning_rate": 4.274989468648405e-06,
      "loss": 0.5622,
      "step": 113430
    },
    {
      "epoch": 46.43,
      "grad_norm": 2.3198325634002686,
      "learning_rate": 4.274899585304208e-06,
      "loss": 0.5373,
      "step": 113440
    },
    {
      "epoch": 46.44,
      "grad_norm": 1.9309988021850586,
      "learning_rate": 4.274809693113715e-06,
      "loss": 0.5287,
      "step": 113450
    },
    {
      "epoch": 46.44,
      "grad_norm": 2.2044856548309326,
      "learning_rate": 4.2747197920773355e-06,
      "loss": 0.5445,
      "step": 113460
    },
    {
      "epoch": 46.45,
      "grad_norm": 2.0657360553741455,
      "learning_rate": 4.274629882195484e-06,
      "loss": 0.5555,
      "step": 113470
    },
    {
      "epoch": 46.45,
      "grad_norm": 2.4303205013275146,
      "learning_rate": 4.274539963468569e-06,
      "loss": 0.534,
      "step": 113480
    },
    {
      "epoch": 46.46,
      "grad_norm": 2.0638413429260254,
      "learning_rate": 4.274450035897005e-06,
      "loss": 0.5486,
      "step": 113490
    },
    {
      "epoch": 46.46,
      "grad_norm": 2.1800007820129395,
      "learning_rate": 4.274360099481203e-06,
      "loss": 0.5448,
      "step": 113500
    },
    {
      "epoch": 46.46,
      "grad_norm": 1.9551115036010742,
      "learning_rate": 4.274270154221575e-06,
      "loss": 0.5552,
      "step": 113510
    },
    {
      "epoch": 46.47,
      "grad_norm": 2.276968002319336,
      "learning_rate": 4.2741802001185345e-06,
      "loss": 0.5716,
      "step": 113520
    },
    {
      "epoch": 46.47,
      "grad_norm": 1.9903260469436646,
      "learning_rate": 4.27409023717249e-06,
      "loss": 0.5543,
      "step": 113530
    },
    {
      "epoch": 46.48,
      "grad_norm": 2.036487340927124,
      "learning_rate": 4.274000265383857e-06,
      "loss": 0.543,
      "step": 113540
    },
    {
      "epoch": 46.48,
      "grad_norm": 2.231795072555542,
      "learning_rate": 4.273910284753047e-06,
      "loss": 0.5289,
      "step": 113550
    },
    {
      "epoch": 46.48,
      "grad_norm": 1.6230343580245972,
      "learning_rate": 4.273820295280471e-06,
      "loss": 0.5495,
      "step": 113560
    },
    {
      "epoch": 46.49,
      "grad_norm": 1.6128817796707153,
      "learning_rate": 4.2737302969665426e-06,
      "loss": 0.535,
      "step": 113570
    },
    {
      "epoch": 46.49,
      "grad_norm": 1.923211693763733,
      "learning_rate": 4.273640289811673e-06,
      "loss": 0.5644,
      "step": 113580
    },
    {
      "epoch": 46.5,
      "grad_norm": 2.1362481117248535,
      "learning_rate": 4.273550273816274e-06,
      "loss": 0.5527,
      "step": 113590
    },
    {
      "epoch": 46.5,
      "grad_norm": 2.5212273597717285,
      "learning_rate": 4.27346024898076e-06,
      "loss": 0.5618,
      "step": 113600
    },
    {
      "epoch": 46.5,
      "grad_norm": 2.4302330017089844,
      "learning_rate": 4.273370215305541e-06,
      "loss": 0.5461,
      "step": 113610
    },
    {
      "epoch": 46.51,
      "grad_norm": 1.744096040725708,
      "learning_rate": 4.273280172791031e-06,
      "loss": 0.5496,
      "step": 113620
    },
    {
      "epoch": 46.51,
      "grad_norm": 2.2610228061676025,
      "learning_rate": 4.2731901214376425e-06,
      "loss": 0.5621,
      "step": 113630
    },
    {
      "epoch": 46.52,
      "grad_norm": 2.144094228744507,
      "learning_rate": 4.273100061245788e-06,
      "loss": 0.5619,
      "step": 113640
    },
    {
      "epoch": 46.52,
      "grad_norm": 1.8981105089187622,
      "learning_rate": 4.273009992215879e-06,
      "loss": 0.5369,
      "step": 113650
    },
    {
      "epoch": 46.52,
      "grad_norm": 1.8585398197174072,
      "learning_rate": 4.272919914348329e-06,
      "loss": 0.5432,
      "step": 113660
    },
    {
      "epoch": 46.53,
      "grad_norm": 1.8403117656707764,
      "learning_rate": 4.272829827643551e-06,
      "loss": 0.5283,
      "step": 113670
    },
    {
      "epoch": 46.53,
      "grad_norm": 2.0200095176696777,
      "learning_rate": 4.272739732101956e-06,
      "loss": 0.5367,
      "step": 113680
    },
    {
      "epoch": 46.54,
      "grad_norm": 2.265479326248169,
      "learning_rate": 4.272649627723958e-06,
      "loss": 0.5577,
      "step": 113690
    },
    {
      "epoch": 46.54,
      "grad_norm": 2.2461605072021484,
      "learning_rate": 4.27255951450997e-06,
      "loss": 0.5444,
      "step": 113700
    },
    {
      "epoch": 46.55,
      "grad_norm": 2.2213969230651855,
      "learning_rate": 4.272469392460404e-06,
      "loss": 0.5553,
      "step": 113710
    },
    {
      "epoch": 46.55,
      "grad_norm": 1.9805645942687988,
      "learning_rate": 4.272379261575673e-06,
      "loss": 0.5444,
      "step": 113720
    },
    {
      "epoch": 46.55,
      "grad_norm": 2.063690662384033,
      "learning_rate": 4.272289121856191e-06,
      "loss": 0.5424,
      "step": 113730
    },
    {
      "epoch": 46.56,
      "grad_norm": 2.5409343242645264,
      "learning_rate": 4.272198973302369e-06,
      "loss": 0.5316,
      "step": 113740
    },
    {
      "epoch": 46.56,
      "grad_norm": 1.9734854698181152,
      "learning_rate": 4.272108815914621e-06,
      "loss": 0.5687,
      "step": 113750
    },
    {
      "epoch": 46.57,
      "grad_norm": 2.7643961906433105,
      "learning_rate": 4.27201864969336e-06,
      "loss": 0.5272,
      "step": 113760
    },
    {
      "epoch": 46.57,
      "grad_norm": 1.6985002756118774,
      "learning_rate": 4.2719284746389986e-06,
      "loss": 0.5465,
      "step": 113770
    },
    {
      "epoch": 46.57,
      "grad_norm": 2.2643160820007324,
      "learning_rate": 4.271838290751951e-06,
      "loss": 0.5541,
      "step": 113780
    },
    {
      "epoch": 46.58,
      "grad_norm": 2.78353214263916,
      "learning_rate": 4.2717480980326284e-06,
      "loss": 0.5488,
      "step": 113790
    },
    {
      "epoch": 46.58,
      "grad_norm": 2.1765270233154297,
      "learning_rate": 4.271657896481447e-06,
      "loss": 0.5676,
      "step": 113800
    },
    {
      "epoch": 46.59,
      "grad_norm": 2.3608739376068115,
      "learning_rate": 4.271567686098816e-06,
      "loss": 0.5528,
      "step": 113810
    },
    {
      "epoch": 46.59,
      "grad_norm": 2.597472667694092,
      "learning_rate": 4.271477466885153e-06,
      "loss": 0.5526,
      "step": 113820
    },
    {
      "epoch": 46.59,
      "grad_norm": 2.2388617992401123,
      "learning_rate": 4.2713872388408665e-06,
      "loss": 0.5559,
      "step": 113830
    },
    {
      "epoch": 46.6,
      "grad_norm": 2.407504081726074,
      "learning_rate": 4.271297001966374e-06,
      "loss": 0.554,
      "step": 113840
    },
    {
      "epoch": 46.6,
      "grad_norm": 2.964568614959717,
      "learning_rate": 4.271206756262087e-06,
      "loss": 0.5355,
      "step": 113850
    },
    {
      "epoch": 46.61,
      "grad_norm": 1.8688580989837646,
      "learning_rate": 4.271116501728419e-06,
      "loss": 0.5424,
      "step": 113860
    },
    {
      "epoch": 46.61,
      "grad_norm": 1.7519097328186035,
      "learning_rate": 4.271026238365783e-06,
      "loss": 0.5232,
      "step": 113870
    },
    {
      "epoch": 46.61,
      "grad_norm": 2.2249915599823,
      "learning_rate": 4.270935966174593e-06,
      "loss": 0.5335,
      "step": 113880
    },
    {
      "epoch": 46.62,
      "grad_norm": 1.9516509771347046,
      "learning_rate": 4.270845685155263e-06,
      "loss": 0.5597,
      "step": 113890
    },
    {
      "epoch": 46.62,
      "grad_norm": 1.823263168334961,
      "learning_rate": 4.2707553953082056e-06,
      "loss": 0.5567,
      "step": 113900
    },
    {
      "epoch": 46.63,
      "grad_norm": 2.3184454441070557,
      "learning_rate": 4.270665096633836e-06,
      "loss": 0.5375,
      "step": 113910
    },
    {
      "epoch": 46.63,
      "grad_norm": 2.804029941558838,
      "learning_rate": 4.270574789132566e-06,
      "loss": 0.5456,
      "step": 113920
    },
    {
      "epoch": 46.64,
      "grad_norm": 2.314847707748413,
      "learning_rate": 4.2704844728048105e-06,
      "loss": 0.5356,
      "step": 113930
    },
    {
      "epoch": 46.64,
      "grad_norm": 1.7618072032928467,
      "learning_rate": 4.270394147650983e-06,
      "loss": 0.5516,
      "step": 113940
    },
    {
      "epoch": 46.64,
      "grad_norm": 2.17891263961792,
      "learning_rate": 4.270303813671496e-06,
      "loss": 0.5616,
      "step": 113950
    },
    {
      "epoch": 46.65,
      "grad_norm": 2.078387975692749,
      "learning_rate": 4.270213470866765e-06,
      "loss": 0.5687,
      "step": 113960
    },
    {
      "epoch": 46.65,
      "grad_norm": 1.9499436616897583,
      "learning_rate": 4.270123119237203e-06,
      "loss": 0.5634,
      "step": 113970
    },
    {
      "epoch": 46.66,
      "grad_norm": 1.9835245609283447,
      "learning_rate": 4.2700327587832245e-06,
      "loss": 0.5535,
      "step": 113980
    },
    {
      "epoch": 46.66,
      "grad_norm": 2.5096051692962646,
      "learning_rate": 4.269942389505242e-06,
      "loss": 0.5305,
      "step": 113990
    },
    {
      "epoch": 46.66,
      "grad_norm": 1.5891451835632324,
      "learning_rate": 4.269852011403671e-06,
      "loss": 0.553,
      "step": 114000
    },
    {
      "epoch": 46.67,
      "grad_norm": 2.5860743522644043,
      "learning_rate": 4.269761624478926e-06,
      "loss": 0.5513,
      "step": 114010
    },
    {
      "epoch": 46.67,
      "grad_norm": 2.2426583766937256,
      "learning_rate": 4.269671228731418e-06,
      "loss": 0.5553,
      "step": 114020
    },
    {
      "epoch": 46.68,
      "grad_norm": 2.0892527103424072,
      "learning_rate": 4.269580824161565e-06,
      "loss": 0.5743,
      "step": 114030
    },
    {
      "epoch": 46.68,
      "grad_norm": 2.784334182739258,
      "learning_rate": 4.2694904107697786e-06,
      "loss": 0.5386,
      "step": 114040
    },
    {
      "epoch": 46.68,
      "grad_norm": 2.343273639678955,
      "learning_rate": 4.269399988556473e-06,
      "loss": 0.5368,
      "step": 114050
    },
    {
      "epoch": 46.69,
      "grad_norm": 1.8400555849075317,
      "learning_rate": 4.2693095575220634e-06,
      "loss": 0.5541,
      "step": 114060
    },
    {
      "epoch": 46.69,
      "grad_norm": 2.0245189666748047,
      "learning_rate": 4.269219117666964e-06,
      "loss": 0.5462,
      "step": 114070
    },
    {
      "epoch": 46.7,
      "grad_norm": 1.8452893495559692,
      "learning_rate": 4.269128668991589e-06,
      "loss": 0.5327,
      "step": 114080
    },
    {
      "epoch": 46.7,
      "grad_norm": 2.577063798904419,
      "learning_rate": 4.269038211496352e-06,
      "loss": 0.5429,
      "step": 114090
    },
    {
      "epoch": 46.7,
      "grad_norm": 1.9651083946228027,
      "learning_rate": 4.268947745181668e-06,
      "loss": 0.5554,
      "step": 114100
    },
    {
      "epoch": 46.71,
      "grad_norm": 2.511725425720215,
      "learning_rate": 4.268857270047951e-06,
      "loss": 0.5389,
      "step": 114110
    },
    {
      "epoch": 46.71,
      "grad_norm": 1.6579915285110474,
      "learning_rate": 4.268766786095616e-06,
      "loss": 0.5471,
      "step": 114120
    },
    {
      "epoch": 46.72,
      "grad_norm": 1.9168168306350708,
      "learning_rate": 4.2686762933250775e-06,
      "loss": 0.5385,
      "step": 114130
    },
    {
      "epoch": 46.72,
      "grad_norm": 2.0244297981262207,
      "learning_rate": 4.268585791736749e-06,
      "loss": 0.5433,
      "step": 114140
    },
    {
      "epoch": 46.73,
      "grad_norm": 2.608765125274658,
      "learning_rate": 4.268495281331047e-06,
      "loss": 0.5302,
      "step": 114150
    },
    {
      "epoch": 46.73,
      "grad_norm": 1.6766093969345093,
      "learning_rate": 4.2684047621083836e-06,
      "loss": 0.5535,
      "step": 114160
    },
    {
      "epoch": 46.73,
      "grad_norm": 2.1082699298858643,
      "learning_rate": 4.268314234069176e-06,
      "loss": 0.5373,
      "step": 114170
    },
    {
      "epoch": 46.74,
      "grad_norm": 2.6665937900543213,
      "learning_rate": 4.268223697213837e-06,
      "loss": 0.535,
      "step": 114180
    },
    {
      "epoch": 46.74,
      "grad_norm": 2.2902252674102783,
      "learning_rate": 4.268133151542782e-06,
      "loss": 0.5534,
      "step": 114190
    },
    {
      "epoch": 46.75,
      "grad_norm": 2.528871536254883,
      "learning_rate": 4.268042597056426e-06,
      "loss": 0.5415,
      "step": 114200
    },
    {
      "epoch": 46.75,
      "grad_norm": 2.042868137359619,
      "learning_rate": 4.267952033755184e-06,
      "loss": 0.5587,
      "step": 114210
    },
    {
      "epoch": 46.75,
      "grad_norm": 1.9119932651519775,
      "learning_rate": 4.26786146163947e-06,
      "loss": 0.5596,
      "step": 114220
    },
    {
      "epoch": 46.76,
      "grad_norm": 2.062178373336792,
      "learning_rate": 4.2677708807096995e-06,
      "loss": 0.5377,
      "step": 114230
    },
    {
      "epoch": 46.76,
      "grad_norm": 2.5837063789367676,
      "learning_rate": 4.267680290966287e-06,
      "loss": 0.5429,
      "step": 114240
    },
    {
      "epoch": 46.77,
      "grad_norm": 2.3096349239349365,
      "learning_rate": 4.267589692409649e-06,
      "loss": 0.5548,
      "step": 114250
    },
    {
      "epoch": 46.77,
      "grad_norm": 2.233051061630249,
      "learning_rate": 4.267499085040197e-06,
      "loss": 0.5443,
      "step": 114260
    },
    {
      "epoch": 46.77,
      "grad_norm": 2.5305674076080322,
      "learning_rate": 4.267408468858351e-06,
      "loss": 0.5355,
      "step": 114270
    },
    {
      "epoch": 46.78,
      "grad_norm": 1.4871004819869995,
      "learning_rate": 4.267317843864522e-06,
      "loss": 0.5331,
      "step": 114280
    },
    {
      "epoch": 46.78,
      "grad_norm": 3.4688479900360107,
      "learning_rate": 4.267227210059126e-06,
      "loss": 0.5545,
      "step": 114290
    },
    {
      "epoch": 46.79,
      "grad_norm": 2.0579867362976074,
      "learning_rate": 4.26713656744258e-06,
      "loss": 0.5542,
      "step": 114300
    },
    {
      "epoch": 46.79,
      "grad_norm": 2.7136237621307373,
      "learning_rate": 4.267045916015298e-06,
      "loss": 0.5337,
      "step": 114310
    },
    {
      "epoch": 46.79,
      "grad_norm": 1.831705927848816,
      "learning_rate": 4.266955255777695e-06,
      "loss": 0.5868,
      "step": 114320
    },
    {
      "epoch": 46.8,
      "grad_norm": 2.253411293029785,
      "learning_rate": 4.266864586730186e-06,
      "loss": 0.538,
      "step": 114330
    },
    {
      "epoch": 46.8,
      "grad_norm": 1.9528348445892334,
      "learning_rate": 4.266773908873188e-06,
      "loss": 0.5504,
      "step": 114340
    },
    {
      "epoch": 46.81,
      "grad_norm": 2.1592414379119873,
      "learning_rate": 4.2666832222071145e-06,
      "loss": 0.5495,
      "step": 114350
    },
    {
      "epoch": 46.81,
      "grad_norm": 2.0405704975128174,
      "learning_rate": 4.266592526732382e-06,
      "loss": 0.5343,
      "step": 114360
    },
    {
      "epoch": 46.82,
      "grad_norm": 3.91690731048584,
      "learning_rate": 4.266501822449406e-06,
      "loss": 0.5357,
      "step": 114370
    },
    {
      "epoch": 46.82,
      "grad_norm": 2.2542545795440674,
      "learning_rate": 4.266411109358601e-06,
      "loss": 0.5625,
      "step": 114380
    },
    {
      "epoch": 46.82,
      "grad_norm": 2.5332698822021484,
      "learning_rate": 4.266320387460383e-06,
      "loss": 0.558,
      "step": 114390
    },
    {
      "epoch": 46.83,
      "grad_norm": 2.6943366527557373,
      "learning_rate": 4.266229656755168e-06,
      "loss": 0.5402,
      "step": 114400
    },
    {
      "epoch": 46.83,
      "grad_norm": 2.0221264362335205,
      "learning_rate": 4.266138917243372e-06,
      "loss": 0.5724,
      "step": 114410
    },
    {
      "epoch": 46.84,
      "grad_norm": 2.003972053527832,
      "learning_rate": 4.26604816892541e-06,
      "loss": 0.5462,
      "step": 114420
    },
    {
      "epoch": 46.84,
      "grad_norm": 1.886199712753296,
      "learning_rate": 4.265957411801697e-06,
      "loss": 0.5324,
      "step": 114430
    },
    {
      "epoch": 46.84,
      "grad_norm": 2.2338576316833496,
      "learning_rate": 4.265866645872651e-06,
      "loss": 0.531,
      "step": 114440
    },
    {
      "epoch": 46.85,
      "grad_norm": 2.877448320388794,
      "learning_rate": 4.265775871138686e-06,
      "loss": 0.5562,
      "step": 114450
    },
    {
      "epoch": 46.85,
      "grad_norm": 2.565589189529419,
      "learning_rate": 4.265685087600216e-06,
      "loss": 0.5472,
      "step": 114460
    },
    {
      "epoch": 46.86,
      "grad_norm": 1.7845412492752075,
      "learning_rate": 4.265594295257661e-06,
      "loss": 0.5505,
      "step": 114470
    },
    {
      "epoch": 46.86,
      "grad_norm": 1.5558210611343384,
      "learning_rate": 4.265503494111435e-06,
      "loss": 0.5422,
      "step": 114480
    },
    {
      "epoch": 46.86,
      "grad_norm": 2.022810697555542,
      "learning_rate": 4.265412684161953e-06,
      "loss": 0.5504,
      "step": 114490
    },
    {
      "epoch": 46.87,
      "grad_norm": 1.652874231338501,
      "learning_rate": 4.265321865409632e-06,
      "loss": 0.5289,
      "step": 114500
    },
    {
      "epoch": 46.87,
      "grad_norm": 1.5757359266281128,
      "learning_rate": 4.265231037854888e-06,
      "loss": 0.5337,
      "step": 114510
    },
    {
      "epoch": 46.88,
      "grad_norm": 1.5482901334762573,
      "learning_rate": 4.265140201498137e-06,
      "loss": 0.5168,
      "step": 114520
    },
    {
      "epoch": 46.88,
      "grad_norm": 1.9137163162231445,
      "learning_rate": 4.2650493563397955e-06,
      "loss": 0.5428,
      "step": 114530
    },
    {
      "epoch": 46.88,
      "grad_norm": 2.096248149871826,
      "learning_rate": 4.26495850238028e-06,
      "loss": 0.5273,
      "step": 114540
    },
    {
      "epoch": 46.89,
      "grad_norm": 1.4305224418640137,
      "learning_rate": 4.264867639620004e-06,
      "loss": 0.5392,
      "step": 114550
    },
    {
      "epoch": 46.89,
      "grad_norm": 2.388575553894043,
      "learning_rate": 4.264776768059386e-06,
      "loss": 0.5636,
      "step": 114560
    },
    {
      "epoch": 46.9,
      "grad_norm": 1.9622207880020142,
      "learning_rate": 4.264685887698842e-06,
      "loss": 0.5385,
      "step": 114570
    },
    {
      "epoch": 46.9,
      "grad_norm": 2.5375072956085205,
      "learning_rate": 4.264594998538789e-06,
      "loss": 0.5591,
      "step": 114580
    },
    {
      "epoch": 46.91,
      "grad_norm": 1.95803964138031,
      "learning_rate": 4.264504100579642e-06,
      "loss": 0.5648,
      "step": 114590
    },
    {
      "epoch": 46.91,
      "grad_norm": 1.8321582078933716,
      "learning_rate": 4.2644131938218186e-06,
      "loss": 0.5221,
      "step": 114600
    },
    {
      "epoch": 46.91,
      "grad_norm": 1.6552740335464478,
      "learning_rate": 4.264322278265733e-06,
      "loss": 0.5511,
      "step": 114610
    },
    {
      "epoch": 46.92,
      "grad_norm": 2.6750380992889404,
      "learning_rate": 4.264231353911804e-06,
      "loss": 0.546,
      "step": 114620
    },
    {
      "epoch": 46.92,
      "grad_norm": 2.471245288848877,
      "learning_rate": 4.264140420760448e-06,
      "loss": 0.5324,
      "step": 114630
    },
    {
      "epoch": 46.93,
      "grad_norm": 2.1757476329803467,
      "learning_rate": 4.26404947881208e-06,
      "loss": 0.5473,
      "step": 114640
    },
    {
      "epoch": 46.93,
      "grad_norm": 2.2591288089752197,
      "learning_rate": 4.263958528067118e-06,
      "loss": 0.5348,
      "step": 114650
    },
    {
      "epoch": 46.93,
      "grad_norm": 2.178130626678467,
      "learning_rate": 4.263867568525978e-06,
      "loss": 0.5476,
      "step": 114660
    },
    {
      "epoch": 46.94,
      "grad_norm": 2.0740649700164795,
      "learning_rate": 4.263776600189077e-06,
      "loss": 0.5483,
      "step": 114670
    },
    {
      "epoch": 46.94,
      "grad_norm": 1.865268588066101,
      "learning_rate": 4.26368562305683e-06,
      "loss": 0.5421,
      "step": 114680
    },
    {
      "epoch": 46.95,
      "grad_norm": 2.0256505012512207,
      "learning_rate": 4.263594637129657e-06,
      "loss": 0.5363,
      "step": 114690
    },
    {
      "epoch": 46.95,
      "grad_norm": 2.0448644161224365,
      "learning_rate": 4.263503642407973e-06,
      "loss": 0.5419,
      "step": 114700
    },
    {
      "epoch": 46.95,
      "grad_norm": 1.898390769958496,
      "learning_rate": 4.263412638892193e-06,
      "loss": 0.5416,
      "step": 114710
    },
    {
      "epoch": 46.96,
      "grad_norm": 2.805645704269409,
      "learning_rate": 4.263321626582737e-06,
      "loss": 0.5679,
      "step": 114720
    },
    {
      "epoch": 46.96,
      "grad_norm": 2.4084060192108154,
      "learning_rate": 4.263230605480022e-06,
      "loss": 0.548,
      "step": 114730
    },
    {
      "epoch": 46.97,
      "grad_norm": 2.008298873901367,
      "learning_rate": 4.263139575584462e-06,
      "loss": 0.5406,
      "step": 114740
    },
    {
      "epoch": 46.97,
      "grad_norm": 2.3252203464508057,
      "learning_rate": 4.2630485368964755e-06,
      "loss": 0.5509,
      "step": 114750
    },
    {
      "epoch": 46.98,
      "grad_norm": 3.2623002529144287,
      "learning_rate": 4.262957489416481e-06,
      "loss": 0.5613,
      "step": 114760
    },
    {
      "epoch": 46.98,
      "grad_norm": 3.2417850494384766,
      "learning_rate": 4.2628664331448934e-06,
      "loss": 0.5366,
      "step": 114770
    },
    {
      "epoch": 46.98,
      "grad_norm": 2.3653054237365723,
      "learning_rate": 4.26277536808213e-06,
      "loss": 0.5383,
      "step": 114780
    },
    {
      "epoch": 46.99,
      "grad_norm": 2.378633499145508,
      "learning_rate": 4.26268429422861e-06,
      "loss": 0.5367,
      "step": 114790
    },
    {
      "epoch": 46.99,
      "grad_norm": 2.746000051498413,
      "learning_rate": 4.262593211584748e-06,
      "loss": 0.5339,
      "step": 114800
    },
    {
      "epoch": 47.0,
      "grad_norm": 2.4852490425109863,
      "learning_rate": 4.262502120150963e-06,
      "loss": 0.5582,
      "step": 114810
    },
    {
      "epoch": 47.0,
      "grad_norm": 1.7529988288879395,
      "learning_rate": 4.262411019927671e-06,
      "loss": 0.5495,
      "step": 114820
    },
    {
      "epoch": 47.0,
      "eval_loss": 0.5442128777503967,
      "eval_runtime": 51.926,
      "eval_samples_per_second": 66.421,
      "eval_steps_per_second": 8.32,
      "step": 114821
    },
    {
      "epoch": 47.0,
      "grad_norm": 2.452923059463501,
      "learning_rate": 4.2623199109152905e-06,
      "loss": 0.5401,
      "step": 114830
    },
    {
      "epoch": 47.01,
      "grad_norm": 1.8032925128936768,
      "learning_rate": 4.262228793114238e-06,
      "loss": 0.5403,
      "step": 114840
    },
    {
      "epoch": 47.01,
      "grad_norm": 1.8847767114639282,
      "learning_rate": 4.262137666524932e-06,
      "loss": 0.5597,
      "step": 114850
    },
    {
      "epoch": 47.02,
      "grad_norm": 2.344705820083618,
      "learning_rate": 4.262046531147788e-06,
      "loss": 0.5488,
      "step": 114860
    },
    {
      "epoch": 47.02,
      "grad_norm": 2.02152943611145,
      "learning_rate": 4.261955386983226e-06,
      "loss": 0.5565,
      "step": 114870
    },
    {
      "epoch": 47.02,
      "grad_norm": 1.7590017318725586,
      "learning_rate": 4.261864234031662e-06,
      "loss": 0.5381,
      "step": 114880
    },
    {
      "epoch": 47.03,
      "grad_norm": 2.393627643585205,
      "learning_rate": 4.261773072293513e-06,
      "loss": 0.5451,
      "step": 114890
    },
    {
      "epoch": 47.03,
      "grad_norm": 2.431776523590088,
      "learning_rate": 4.261681901769198e-06,
      "loss": 0.5301,
      "step": 114900
    },
    {
      "epoch": 47.04,
      "grad_norm": 2.5786185264587402,
      "learning_rate": 4.261590722459134e-06,
      "loss": 0.5587,
      "step": 114910
    },
    {
      "epoch": 47.04,
      "grad_norm": 2.1342945098876953,
      "learning_rate": 4.261499534363739e-06,
      "loss": 0.529,
      "step": 114920
    },
    {
      "epoch": 47.04,
      "grad_norm": 2.4744858741760254,
      "learning_rate": 4.261408337483429e-06,
      "loss": 0.5488,
      "step": 114930
    },
    {
      "epoch": 47.05,
      "grad_norm": 2.0267486572265625,
      "learning_rate": 4.261317131818623e-06,
      "loss": 0.5474,
      "step": 114940
    },
    {
      "epoch": 47.05,
      "grad_norm": 1.9824787378311157,
      "learning_rate": 4.261225917369741e-06,
      "loss": 0.5432,
      "step": 114950
    },
    {
      "epoch": 47.06,
      "grad_norm": 2.7829275131225586,
      "learning_rate": 4.261134694137198e-06,
      "loss": 0.5437,
      "step": 114960
    },
    {
      "epoch": 47.06,
      "grad_norm": 2.5298080444335938,
      "learning_rate": 4.261043462121412e-06,
      "loss": 0.5505,
      "step": 114970
    },
    {
      "epoch": 47.07,
      "grad_norm": 1.6836248636245728,
      "learning_rate": 4.260952221322802e-06,
      "loss": 0.532,
      "step": 114980
    },
    {
      "epoch": 47.07,
      "grad_norm": 3.288264513015747,
      "learning_rate": 4.260860971741786e-06,
      "loss": 0.543,
      "step": 114990
    },
    {
      "epoch": 47.07,
      "grad_norm": 1.8249717950820923,
      "learning_rate": 4.26076971337878e-06,
      "loss": 0.5438,
      "step": 115000
    },
    {
      "epoch": 47.08,
      "grad_norm": 1.5757472515106201,
      "learning_rate": 4.260678446234205e-06,
      "loss": 0.5586,
      "step": 115010
    },
    {
      "epoch": 47.08,
      "grad_norm": 2.01015305519104,
      "learning_rate": 4.260587170308477e-06,
      "loss": 0.5623,
      "step": 115020
    },
    {
      "epoch": 47.09,
      "grad_norm": 1.9164799451828003,
      "learning_rate": 4.2604958856020154e-06,
      "loss": 0.5319,
      "step": 115030
    },
    {
      "epoch": 47.09,
      "grad_norm": 2.3439292907714844,
      "learning_rate": 4.260404592115237e-06,
      "loss": 0.5382,
      "step": 115040
    },
    {
      "epoch": 47.09,
      "grad_norm": 1.7095552682876587,
      "learning_rate": 4.260313289848562e-06,
      "loss": 0.5362,
      "step": 115050
    },
    {
      "epoch": 47.1,
      "grad_norm": 2.156940221786499,
      "learning_rate": 4.2602219788024055e-06,
      "loss": 0.5297,
      "step": 115060
    },
    {
      "epoch": 47.1,
      "grad_norm": 2.7742068767547607,
      "learning_rate": 4.260130658977189e-06,
      "loss": 0.5562,
      "step": 115070
    },
    {
      "epoch": 47.11,
      "grad_norm": 2.5542585849761963,
      "learning_rate": 4.260039330373329e-06,
      "loss": 0.5419,
      "step": 115080
    },
    {
      "epoch": 47.11,
      "grad_norm": 2.3012568950653076,
      "learning_rate": 4.259947992991245e-06,
      "loss": 0.5346,
      "step": 115090
    },
    {
      "epoch": 47.11,
      "grad_norm": 2.0449583530426025,
      "learning_rate": 4.259856646831354e-06,
      "loss": 0.5365,
      "step": 115100
    },
    {
      "epoch": 47.12,
      "grad_norm": 2.355806350708008,
      "learning_rate": 4.259765291894075e-06,
      "loss": 0.5353,
      "step": 115110
    },
    {
      "epoch": 47.12,
      "grad_norm": 2.023500442504883,
      "learning_rate": 4.259673928179827e-06,
      "loss": 0.5403,
      "step": 115120
    },
    {
      "epoch": 47.13,
      "grad_norm": 2.1597375869750977,
      "learning_rate": 4.259582555689029e-06,
      "loss": 0.5594,
      "step": 115130
    },
    {
      "epoch": 47.13,
      "grad_norm": 2.0151782035827637,
      "learning_rate": 4.259491174422098e-06,
      "loss": 0.5553,
      "step": 115140
    },
    {
      "epoch": 47.13,
      "grad_norm": 2.059398889541626,
      "learning_rate": 4.259399784379453e-06,
      "loss": 0.5248,
      "step": 115150
    },
    {
      "epoch": 47.14,
      "grad_norm": 2.039231300354004,
      "learning_rate": 4.259308385561514e-06,
      "loss": 0.5062,
      "step": 115160
    },
    {
      "epoch": 47.14,
      "grad_norm": 2.117868185043335,
      "learning_rate": 4.259216977968698e-06,
      "loss": 0.5329,
      "step": 115170
    },
    {
      "epoch": 47.15,
      "grad_norm": 2.2165534496307373,
      "learning_rate": 4.259125561601424e-06,
      "loss": 0.5333,
      "step": 115180
    },
    {
      "epoch": 47.15,
      "grad_norm": 2.546895742416382,
      "learning_rate": 4.259034136460112e-06,
      "loss": 0.5676,
      "step": 115190
    },
    {
      "epoch": 47.16,
      "grad_norm": 2.63444185256958,
      "learning_rate": 4.25894270254518e-06,
      "loss": 0.5491,
      "step": 115200
    },
    {
      "epoch": 47.16,
      "grad_norm": 1.8254672288894653,
      "learning_rate": 4.258851259857047e-06,
      "loss": 0.5435,
      "step": 115210
    },
    {
      "epoch": 47.16,
      "grad_norm": 2.2720539569854736,
      "learning_rate": 4.2587598083961316e-06,
      "loss": 0.5287,
      "step": 115220
    },
    {
      "epoch": 47.17,
      "grad_norm": 1.9966634511947632,
      "learning_rate": 4.258668348162852e-06,
      "loss": 0.5433,
      "step": 115230
    },
    {
      "epoch": 47.17,
      "grad_norm": 2.4307491779327393,
      "learning_rate": 4.258576879157629e-06,
      "loss": 0.5336,
      "step": 115240
    },
    {
      "epoch": 47.18,
      "grad_norm": 1.9000071287155151,
      "learning_rate": 4.25848540138088e-06,
      "loss": 0.5325,
      "step": 115250
    },
    {
      "epoch": 47.18,
      "grad_norm": 1.8601661920547485,
      "learning_rate": 4.258393914833025e-06,
      "loss": 0.5327,
      "step": 115260
    },
    {
      "epoch": 47.18,
      "grad_norm": 2.518284320831299,
      "learning_rate": 4.258302419514483e-06,
      "loss": 0.5242,
      "step": 115270
    },
    {
      "epoch": 47.19,
      "grad_norm": 2.395437240600586,
      "learning_rate": 4.258210915425672e-06,
      "loss": 0.5313,
      "step": 115280
    },
    {
      "epoch": 47.19,
      "grad_norm": 2.230881690979004,
      "learning_rate": 4.258119402567013e-06,
      "loss": 0.5596,
      "step": 115290
    },
    {
      "epoch": 47.2,
      "grad_norm": 2.303032159805298,
      "learning_rate": 4.258027880938923e-06,
      "loss": 0.5534,
      "step": 115300
    },
    {
      "epoch": 47.2,
      "grad_norm": 1.7685153484344482,
      "learning_rate": 4.2579363505418235e-06,
      "loss": 0.5432,
      "step": 115310
    },
    {
      "epoch": 47.2,
      "grad_norm": 2.292664051055908,
      "learning_rate": 4.257844811376133e-06,
      "loss": 0.5524,
      "step": 115320
    },
    {
      "epoch": 47.21,
      "grad_norm": 2.376493453979492,
      "learning_rate": 4.2577532634422695e-06,
      "loss": 0.5467,
      "step": 115330
    },
    {
      "epoch": 47.21,
      "grad_norm": 2.577629566192627,
      "learning_rate": 4.257661706740655e-06,
      "loss": 0.5433,
      "step": 115340
    },
    {
      "epoch": 47.22,
      "grad_norm": 2.038696527481079,
      "learning_rate": 4.257570141271706e-06,
      "loss": 0.5431,
      "step": 115350
    },
    {
      "epoch": 47.22,
      "grad_norm": 1.9673857688903809,
      "learning_rate": 4.257478567035843e-06,
      "loss": 0.5548,
      "step": 115360
    },
    {
      "epoch": 47.22,
      "grad_norm": 2.2026031017303467,
      "learning_rate": 4.257386984033487e-06,
      "loss": 0.5289,
      "step": 115370
    },
    {
      "epoch": 47.23,
      "grad_norm": 2.077345132827759,
      "learning_rate": 4.257295392265056e-06,
      "loss": 0.5499,
      "step": 115380
    },
    {
      "epoch": 47.23,
      "grad_norm": 1.936621904373169,
      "learning_rate": 4.25720379173097e-06,
      "loss": 0.5458,
      "step": 115390
    },
    {
      "epoch": 47.24,
      "grad_norm": 1.823115587234497,
      "learning_rate": 4.257112182431649e-06,
      "loss": 0.549,
      "step": 115400
    },
    {
      "epoch": 47.24,
      "grad_norm": 1.9199553728103638,
      "learning_rate": 4.257020564367511e-06,
      "loss": 0.5418,
      "step": 115410
    },
    {
      "epoch": 47.25,
      "grad_norm": 1.9546477794647217,
      "learning_rate": 4.256928937538977e-06,
      "loss": 0.5342,
      "step": 115420
    },
    {
      "epoch": 47.25,
      "grad_norm": 2.4521114826202393,
      "learning_rate": 4.256837301946468e-06,
      "loss": 0.5579,
      "step": 115430
    },
    {
      "epoch": 47.25,
      "grad_norm": 2.381892204284668,
      "learning_rate": 4.256745657590401e-06,
      "loss": 0.5481,
      "step": 115440
    },
    {
      "epoch": 47.26,
      "grad_norm": 1.842685580253601,
      "learning_rate": 4.2566540044711975e-06,
      "loss": 0.5393,
      "step": 115450
    },
    {
      "epoch": 47.26,
      "grad_norm": 2.4630815982818604,
      "learning_rate": 4.2565623425892775e-06,
      "loss": 0.5437,
      "step": 115460
    },
    {
      "epoch": 47.27,
      "grad_norm": 1.7695093154907227,
      "learning_rate": 4.256470671945059e-06,
      "loss": 0.5175,
      "step": 115470
    },
    {
      "epoch": 47.27,
      "grad_norm": 2.7984848022460938,
      "learning_rate": 4.256378992538965e-06,
      "loss": 0.549,
      "step": 115480
    },
    {
      "epoch": 47.27,
      "grad_norm": 2.329340696334839,
      "learning_rate": 4.256287304371413e-06,
      "loss": 0.5462,
      "step": 115490
    },
    {
      "epoch": 47.28,
      "grad_norm": 2.303410291671753,
      "learning_rate": 4.256195607442824e-06,
      "loss": 0.5507,
      "step": 115500
    },
    {
      "epoch": 47.28,
      "grad_norm": 2.399829387664795,
      "learning_rate": 4.256103901753617e-06,
      "loss": 0.5461,
      "step": 115510
    },
    {
      "epoch": 47.29,
      "grad_norm": 1.9190504550933838,
      "learning_rate": 4.256012187304213e-06,
      "loss": 0.5441,
      "step": 115520
    },
    {
      "epoch": 47.29,
      "grad_norm": 2.5187225341796875,
      "learning_rate": 4.255920464095034e-06,
      "loss": 0.5579,
      "step": 115530
    },
    {
      "epoch": 47.29,
      "grad_norm": 1.9369007349014282,
      "learning_rate": 4.255828732126496e-06,
      "loss": 0.5373,
      "step": 115540
    },
    {
      "epoch": 47.3,
      "grad_norm": 2.19240140914917,
      "learning_rate": 4.255736991399023e-06,
      "loss": 0.5271,
      "step": 115550
    },
    {
      "epoch": 47.3,
      "grad_norm": 2.2296667098999023,
      "learning_rate": 4.2556452419130326e-06,
      "loss": 0.5403,
      "step": 115560
    },
    {
      "epoch": 47.31,
      "grad_norm": 1.8035290241241455,
      "learning_rate": 4.255553483668947e-06,
      "loss": 0.5605,
      "step": 115570
    },
    {
      "epoch": 47.31,
      "grad_norm": 2.518961191177368,
      "learning_rate": 4.255461716667186e-06,
      "loss": 0.5293,
      "step": 115580
    },
    {
      "epoch": 47.31,
      "grad_norm": 2.2550418376922607,
      "learning_rate": 4.255369940908168e-06,
      "loss": 0.5638,
      "step": 115590
    },
    {
      "epoch": 47.32,
      "grad_norm": 1.9699732065200806,
      "learning_rate": 4.2552781563923166e-06,
      "loss": 0.5557,
      "step": 115600
    },
    {
      "epoch": 47.32,
      "grad_norm": 1.7914597988128662,
      "learning_rate": 4.25518636312005e-06,
      "loss": 0.5589,
      "step": 115610
    },
    {
      "epoch": 47.33,
      "grad_norm": 2.38564133644104,
      "learning_rate": 4.25509456109179e-06,
      "loss": 0.5345,
      "step": 115620
    },
    {
      "epoch": 47.33,
      "grad_norm": 1.3394339084625244,
      "learning_rate": 4.2550027503079566e-06,
      "loss": 0.5393,
      "step": 115630
    },
    {
      "epoch": 47.34,
      "grad_norm": 2.5581774711608887,
      "learning_rate": 4.2549109307689706e-06,
      "loss": 0.5516,
      "step": 115640
    },
    {
      "epoch": 47.34,
      "grad_norm": 2.1297879219055176,
      "learning_rate": 4.254819102475251e-06,
      "loss": 0.5481,
      "step": 115650
    },
    {
      "epoch": 47.34,
      "grad_norm": 1.898861289024353,
      "learning_rate": 4.254727265427221e-06,
      "loss": 0.5407,
      "step": 115660
    },
    {
      "epoch": 47.35,
      "grad_norm": 1.7338430881500244,
      "learning_rate": 4.254635419625299e-06,
      "loss": 0.5424,
      "step": 115670
    },
    {
      "epoch": 47.35,
      "grad_norm": 2.0939407348632812,
      "learning_rate": 4.254543565069908e-06,
      "loss": 0.5432,
      "step": 115680
    },
    {
      "epoch": 47.36,
      "grad_norm": 2.1346099376678467,
      "learning_rate": 4.254451701761468e-06,
      "loss": 0.5242,
      "step": 115690
    },
    {
      "epoch": 47.36,
      "grad_norm": 2.153958559036255,
      "learning_rate": 4.254359829700397e-06,
      "loss": 0.5458,
      "step": 115700
    },
    {
      "epoch": 47.36,
      "grad_norm": 2.3388242721557617,
      "learning_rate": 4.254267948887121e-06,
      "loss": 0.5509,
      "step": 115710
    },
    {
      "epoch": 47.37,
      "grad_norm": 2.395111560821533,
      "learning_rate": 4.254176059322057e-06,
      "loss": 0.5474,
      "step": 115720
    },
    {
      "epoch": 47.37,
      "grad_norm": 2.4084362983703613,
      "learning_rate": 4.254084161005627e-06,
      "loss": 0.5523,
      "step": 115730
    },
    {
      "epoch": 47.38,
      "grad_norm": 2.184136390686035,
      "learning_rate": 4.253992253938252e-06,
      "loss": 0.5792,
      "step": 115740
    },
    {
      "epoch": 47.38,
      "grad_norm": 1.6286282539367676,
      "learning_rate": 4.2539003381203534e-06,
      "loss": 0.5502,
      "step": 115750
    },
    {
      "epoch": 47.38,
      "grad_norm": 1.804930329322815,
      "learning_rate": 4.253808413552352e-06,
      "loss": 0.5332,
      "step": 115760
    },
    {
      "epoch": 47.39,
      "grad_norm": 2.424678325653076,
      "learning_rate": 4.253716480234668e-06,
      "loss": 0.5205,
      "step": 115770
    },
    {
      "epoch": 47.39,
      "grad_norm": 2.3393757343292236,
      "learning_rate": 4.2536245381677245e-06,
      "loss": 0.5612,
      "step": 115780
    },
    {
      "epoch": 47.4,
      "grad_norm": 2.9568216800689697,
      "learning_rate": 4.253532587351941e-06,
      "loss": 0.5501,
      "step": 115790
    },
    {
      "epoch": 47.4,
      "grad_norm": 2.08183217048645,
      "learning_rate": 4.253440627787739e-06,
      "loss": 0.5445,
      "step": 115800
    },
    {
      "epoch": 47.4,
      "grad_norm": 2.2330970764160156,
      "learning_rate": 4.25334865947554e-06,
      "loss": 0.5475,
      "step": 115810
    },
    {
      "epoch": 47.41,
      "grad_norm": 1.9580155611038208,
      "learning_rate": 4.253256682415766e-06,
      "loss": 0.528,
      "step": 115820
    },
    {
      "epoch": 47.41,
      "grad_norm": 2.84720516204834,
      "learning_rate": 4.253164696608837e-06,
      "loss": 0.5455,
      "step": 115830
    },
    {
      "epoch": 47.42,
      "grad_norm": 1.9692858457565308,
      "learning_rate": 4.253072702055175e-06,
      "loss": 0.5471,
      "step": 115840
    },
    {
      "epoch": 47.42,
      "grad_norm": 2.1050679683685303,
      "learning_rate": 4.252980698755201e-06,
      "loss": 0.546,
      "step": 115850
    },
    {
      "epoch": 47.43,
      "grad_norm": 2.239096164703369,
      "learning_rate": 4.252888686709338e-06,
      "loss": 0.5269,
      "step": 115860
    },
    {
      "epoch": 47.43,
      "grad_norm": 1.7395600080490112,
      "learning_rate": 4.252796665918005e-06,
      "loss": 0.547,
      "step": 115870
    },
    {
      "epoch": 47.43,
      "grad_norm": 2.6672909259796143,
      "learning_rate": 4.252704636381626e-06,
      "loss": 0.5633,
      "step": 115880
    },
    {
      "epoch": 47.44,
      "grad_norm": 2.0487594604492188,
      "learning_rate": 4.252612598100621e-06,
      "loss": 0.526,
      "step": 115890
    },
    {
      "epoch": 47.44,
      "grad_norm": 2.3010995388031006,
      "learning_rate": 4.252520551075412e-06,
      "loss": 0.5377,
      "step": 115900
    },
    {
      "epoch": 47.45,
      "grad_norm": 2.8780500888824463,
      "learning_rate": 4.252428495306421e-06,
      "loss": 0.5433,
      "step": 115910
    },
    {
      "epoch": 47.45,
      "grad_norm": 2.0898752212524414,
      "learning_rate": 4.252336430794069e-06,
      "loss": 0.532,
      "step": 115920
    },
    {
      "epoch": 47.45,
      "grad_norm": 1.917283296585083,
      "learning_rate": 4.252244357538778e-06,
      "loss": 0.5433,
      "step": 115930
    },
    {
      "epoch": 47.46,
      "grad_norm": 1.8805937767028809,
      "learning_rate": 4.25215227554097e-06,
      "loss": 0.552,
      "step": 115940
    },
    {
      "epoch": 47.46,
      "grad_norm": 2.407020092010498,
      "learning_rate": 4.252060184801068e-06,
      "loss": 0.5409,
      "step": 115950
    },
    {
      "epoch": 47.47,
      "grad_norm": 2.125303030014038,
      "learning_rate": 4.251968085319491e-06,
      "loss": 0.5314,
      "step": 115960
    },
    {
      "epoch": 47.47,
      "grad_norm": 1.7785282135009766,
      "learning_rate": 4.251875977096663e-06,
      "loss": 0.548,
      "step": 115970
    },
    {
      "epoch": 47.47,
      "grad_norm": 2.156717300415039,
      "learning_rate": 4.251783860133005e-06,
      "loss": 0.5492,
      "step": 115980
    },
    {
      "epoch": 47.48,
      "grad_norm": 2.203136920928955,
      "learning_rate": 4.25169173442894e-06,
      "loss": 0.5679,
      "step": 115990
    },
    {
      "epoch": 47.48,
      "grad_norm": 1.9437212944030762,
      "learning_rate": 4.251599599984889e-06,
      "loss": 0.5534,
      "step": 116000
    },
    {
      "epoch": 47.49,
      "grad_norm": 2.2514894008636475,
      "learning_rate": 4.251507456801274e-06,
      "loss": 0.5685,
      "step": 116010
    },
    {
      "epoch": 47.49,
      "grad_norm": 2.173121452331543,
      "learning_rate": 4.251415304878519e-06,
      "loss": 0.5437,
      "step": 116020
    },
    {
      "epoch": 47.49,
      "grad_norm": 1.869721531867981,
      "learning_rate": 4.251323144217044e-06,
      "loss": 0.5555,
      "step": 116030
    },
    {
      "epoch": 47.5,
      "grad_norm": 2.2363903522491455,
      "learning_rate": 4.2512309748172716e-06,
      "loss": 0.5393,
      "step": 116040
    },
    {
      "epoch": 47.5,
      "grad_norm": 2.5948214530944824,
      "learning_rate": 4.251138796679625e-06,
      "loss": 0.5341,
      "step": 116050
    },
    {
      "epoch": 47.51,
      "grad_norm": 2.382000207901001,
      "learning_rate": 4.251046609804524e-06,
      "loss": 0.548,
      "step": 116060
    },
    {
      "epoch": 47.51,
      "grad_norm": 3.3324148654937744,
      "learning_rate": 4.250954414192395e-06,
      "loss": 0.5405,
      "step": 116070
    },
    {
      "epoch": 47.52,
      "grad_norm": 2.2811989784240723,
      "learning_rate": 4.250862209843656e-06,
      "loss": 0.5496,
      "step": 116080
    },
    {
      "epoch": 47.52,
      "grad_norm": 1.7424139976501465,
      "learning_rate": 4.2507699967587315e-06,
      "loss": 0.5624,
      "step": 116090
    },
    {
      "epoch": 47.52,
      "grad_norm": 2.164106845855713,
      "learning_rate": 4.250677774938045e-06,
      "loss": 0.5471,
      "step": 116100
    },
    {
      "epoch": 47.53,
      "grad_norm": 2.099362850189209,
      "learning_rate": 4.250585544382017e-06,
      "loss": 0.5389,
      "step": 116110
    },
    {
      "epoch": 47.53,
      "grad_norm": 2.175459384918213,
      "learning_rate": 4.2504933050910696e-06,
      "loss": 0.5389,
      "step": 116120
    },
    {
      "epoch": 47.54,
      "grad_norm": 3.0095086097717285,
      "learning_rate": 4.250401057065628e-06,
      "loss": 0.5355,
      "step": 116130
    },
    {
      "epoch": 47.54,
      "grad_norm": 2.4965577125549316,
      "learning_rate": 4.250308800306112e-06,
      "loss": 0.5428,
      "step": 116140
    },
    {
      "epoch": 47.54,
      "grad_norm": 1.9907991886138916,
      "learning_rate": 4.250216534812947e-06,
      "loss": 0.56,
      "step": 116150
    },
    {
      "epoch": 47.55,
      "grad_norm": 3.6532559394836426,
      "learning_rate": 4.250124260586552e-06,
      "loss": 0.5332,
      "step": 116160
    },
    {
      "epoch": 47.55,
      "grad_norm": 3.43347430229187,
      "learning_rate": 4.2500319776273535e-06,
      "loss": 0.5473,
      "step": 116170
    },
    {
      "epoch": 47.56,
      "grad_norm": 2.095547676086426,
      "learning_rate": 4.249939685935771e-06,
      "loss": 0.5419,
      "step": 116180
    },
    {
      "epoch": 47.56,
      "grad_norm": 2.194706916809082,
      "learning_rate": 4.24984738551223e-06,
      "loss": 0.5406,
      "step": 116190
    },
    {
      "epoch": 47.56,
      "grad_norm": 1.9534069299697876,
      "learning_rate": 4.249755076357152e-06,
      "loss": 0.5592,
      "step": 116200
    },
    {
      "epoch": 47.57,
      "grad_norm": 2.0124313831329346,
      "learning_rate": 4.24966275847096e-06,
      "loss": 0.5167,
      "step": 116210
    },
    {
      "epoch": 47.57,
      "grad_norm": 1.7982250452041626,
      "learning_rate": 4.249570431854076e-06,
      "loss": 0.5526,
      "step": 116220
    },
    {
      "epoch": 47.58,
      "grad_norm": 2.1650593280792236,
      "learning_rate": 4.2494780965069245e-06,
      "loss": 0.5246,
      "step": 116230
    },
    {
      "epoch": 47.58,
      "grad_norm": 2.696247100830078,
      "learning_rate": 4.249385752429926e-06,
      "loss": 0.5421,
      "step": 116240
    },
    {
      "epoch": 47.58,
      "grad_norm": 2.0768330097198486,
      "learning_rate": 4.249293399623507e-06,
      "loss": 0.566,
      "step": 116250
    },
    {
      "epoch": 47.59,
      "grad_norm": 1.8064959049224854,
      "learning_rate": 4.2492010380880885e-06,
      "loss": 0.5538,
      "step": 116260
    },
    {
      "epoch": 47.59,
      "grad_norm": 2.016571521759033,
      "learning_rate": 4.249108667824094e-06,
      "loss": 0.5546,
      "step": 116270
    },
    {
      "epoch": 47.6,
      "grad_norm": 2.4557368755340576,
      "learning_rate": 4.2490162888319466e-06,
      "loss": 0.5389,
      "step": 116280
    },
    {
      "epoch": 47.6,
      "grad_norm": 2.1636910438537598,
      "learning_rate": 4.248923901112069e-06,
      "loss": 0.5425,
      "step": 116290
    },
    {
      "epoch": 47.61,
      "grad_norm": 2.7756361961364746,
      "learning_rate": 4.248831504664885e-06,
      "loss": 0.5191,
      "step": 116300
    },
    {
      "epoch": 47.61,
      "grad_norm": 1.7733720541000366,
      "learning_rate": 4.248739099490818e-06,
      "loss": 0.526,
      "step": 116310
    },
    {
      "epoch": 47.61,
      "grad_norm": 2.293433904647827,
      "learning_rate": 4.24864668559029e-06,
      "loss": 0.5627,
      "step": 116320
    },
    {
      "epoch": 47.62,
      "grad_norm": 1.992822289466858,
      "learning_rate": 4.248554262963726e-06,
      "loss": 0.5689,
      "step": 116330
    },
    {
      "epoch": 47.62,
      "grad_norm": 2.022570848464966,
      "learning_rate": 4.248461831611549e-06,
      "loss": 0.5802,
      "step": 116340
    },
    {
      "epoch": 47.63,
      "grad_norm": 1.803492546081543,
      "learning_rate": 4.248369391534181e-06,
      "loss": 0.5246,
      "step": 116350
    },
    {
      "epoch": 47.63,
      "grad_norm": 1.6883150339126587,
      "learning_rate": 4.2482769427320474e-06,
      "loss": 0.522,
      "step": 116360
    },
    {
      "epoch": 47.63,
      "grad_norm": 2.7828314304351807,
      "learning_rate": 4.2481844852055704e-06,
      "loss": 0.5226,
      "step": 116370
    },
    {
      "epoch": 47.64,
      "grad_norm": 1.809862494468689,
      "learning_rate": 4.248092018955173e-06,
      "loss": 0.5639,
      "step": 116380
    },
    {
      "epoch": 47.64,
      "grad_norm": 2.3451638221740723,
      "learning_rate": 4.247999543981282e-06,
      "loss": 0.5361,
      "step": 116390
    },
    {
      "epoch": 47.65,
      "grad_norm": 1.9086172580718994,
      "learning_rate": 4.247907060284318e-06,
      "loss": 0.5472,
      "step": 116400
    },
    {
      "epoch": 47.65,
      "grad_norm": 2.4577202796936035,
      "learning_rate": 4.247814567864705e-06,
      "loss": 0.5414,
      "step": 116410
    },
    {
      "epoch": 47.65,
      "grad_norm": 2.0716350078582764,
      "learning_rate": 4.247722066722867e-06,
      "loss": 0.5586,
      "step": 116420
    },
    {
      "epoch": 47.66,
      "grad_norm": 2.2465479373931885,
      "learning_rate": 4.2476295568592275e-06,
      "loss": 0.5608,
      "step": 116430
    },
    {
      "epoch": 47.66,
      "grad_norm": 1.8528639078140259,
      "learning_rate": 4.247537038274211e-06,
      "loss": 0.5322,
      "step": 116440
    },
    {
      "epoch": 47.67,
      "grad_norm": 2.250638484954834,
      "learning_rate": 4.247444510968241e-06,
      "loss": 0.5323,
      "step": 116450
    },
    {
      "epoch": 47.67,
      "grad_norm": 1.8709025382995605,
      "learning_rate": 4.247351974941741e-06,
      "loss": 0.5233,
      "step": 116460
    },
    {
      "epoch": 47.67,
      "grad_norm": 1.9591701030731201,
      "learning_rate": 4.247259430195136e-06,
      "loss": 0.5432,
      "step": 116470
    },
    {
      "epoch": 47.68,
      "grad_norm": 1.750239372253418,
      "learning_rate": 4.247166876728848e-06,
      "loss": 0.5522,
      "step": 116480
    },
    {
      "epoch": 47.68,
      "grad_norm": 2.737677812576294,
      "learning_rate": 4.247074314543302e-06,
      "loss": 0.5618,
      "step": 116490
    },
    {
      "epoch": 47.69,
      "grad_norm": 2.048828125,
      "learning_rate": 4.246981743638923e-06,
      "loss": 0.5504,
      "step": 116500
    },
    {
      "epoch": 47.69,
      "grad_norm": 2.079401731491089,
      "learning_rate": 4.246889164016134e-06,
      "loss": 0.527,
      "step": 116510
    },
    {
      "epoch": 47.7,
      "grad_norm": 2.523049831390381,
      "learning_rate": 4.246796575675359e-06,
      "loss": 0.5608,
      "step": 116520
    },
    {
      "epoch": 47.7,
      "grad_norm": 2.1345608234405518,
      "learning_rate": 4.246703978617022e-06,
      "loss": 0.565,
      "step": 116530
    },
    {
      "epoch": 47.7,
      "grad_norm": 2.2611966133117676,
      "learning_rate": 4.246611372841548e-06,
      "loss": 0.5577,
      "step": 116540
    },
    {
      "epoch": 47.71,
      "grad_norm": 1.9170880317687988,
      "learning_rate": 4.246518758349361e-06,
      "loss": 0.534,
      "step": 116550
    },
    {
      "epoch": 47.71,
      "grad_norm": 1.7738181352615356,
      "learning_rate": 4.246426135140885e-06,
      "loss": 0.5443,
      "step": 116560
    },
    {
      "epoch": 47.72,
      "grad_norm": 2.0734381675720215,
      "learning_rate": 4.2463335032165435e-06,
      "loss": 0.5393,
      "step": 116570
    },
    {
      "epoch": 47.72,
      "grad_norm": 2.1233229637145996,
      "learning_rate": 4.246240862576762e-06,
      "loss": 0.5385,
      "step": 116580
    },
    {
      "epoch": 47.72,
      "grad_norm": 1.6475456953048706,
      "learning_rate": 4.246148213221965e-06,
      "loss": 0.5442,
      "step": 116590
    },
    {
      "epoch": 47.73,
      "grad_norm": 3.0267961025238037,
      "learning_rate": 4.246055555152576e-06,
      "loss": 0.555,
      "step": 116600
    },
    {
      "epoch": 47.73,
      "grad_norm": 1.959881067276001,
      "learning_rate": 4.24596288836902e-06,
      "loss": 0.5443,
      "step": 116610
    },
    {
      "epoch": 47.74,
      "grad_norm": 1.9695454835891724,
      "learning_rate": 4.245870212871721e-06,
      "loss": 0.549,
      "step": 116620
    },
    {
      "epoch": 47.74,
      "grad_norm": 1.995721459388733,
      "learning_rate": 4.245777528661104e-06,
      "loss": 0.5276,
      "step": 116630
    },
    {
      "epoch": 47.74,
      "grad_norm": 1.9269076585769653,
      "learning_rate": 4.245684835737593e-06,
      "loss": 0.5437,
      "step": 116640
    },
    {
      "epoch": 47.75,
      "grad_norm": 1.8635307550430298,
      "learning_rate": 4.2455921341016136e-06,
      "loss": 0.5437,
      "step": 116650
    },
    {
      "epoch": 47.75,
      "grad_norm": 1.9411789178848267,
      "learning_rate": 4.24549942375359e-06,
      "loss": 0.5602,
      "step": 116660
    },
    {
      "epoch": 47.76,
      "grad_norm": 1.981960654258728,
      "learning_rate": 4.245406704693947e-06,
      "loss": 0.5577,
      "step": 116670
    },
    {
      "epoch": 47.76,
      "grad_norm": 2.021254777908325,
      "learning_rate": 4.245313976923109e-06,
      "loss": 0.5483,
      "step": 116680
    },
    {
      "epoch": 47.77,
      "grad_norm": 1.9866622686386108,
      "learning_rate": 4.245221240441501e-06,
      "loss": 0.5421,
      "step": 116690
    },
    {
      "epoch": 47.77,
      "grad_norm": 2.5005593299865723,
      "learning_rate": 4.245128495249547e-06,
      "loss": 0.535,
      "step": 116700
    },
    {
      "epoch": 47.77,
      "grad_norm": 2.0408759117126465,
      "learning_rate": 4.245035741347672e-06,
      "loss": 0.55,
      "step": 116710
    },
    {
      "epoch": 47.78,
      "grad_norm": 2.1591756343841553,
      "learning_rate": 4.244942978736303e-06,
      "loss": 0.5611,
      "step": 116720
    },
    {
      "epoch": 47.78,
      "grad_norm": 1.5377830266952515,
      "learning_rate": 4.244850207415864e-06,
      "loss": 0.5318,
      "step": 116730
    },
    {
      "epoch": 47.79,
      "grad_norm": 2.1076502799987793,
      "learning_rate": 4.2447574273867775e-06,
      "loss": 0.5433,
      "step": 116740
    },
    {
      "epoch": 47.79,
      "grad_norm": 1.9228854179382324,
      "learning_rate": 4.244664638649471e-06,
      "loss": 0.5419,
      "step": 116750
    },
    {
      "epoch": 47.79,
      "grad_norm": 2.122546672821045,
      "learning_rate": 4.244571841204369e-06,
      "loss": 0.5392,
      "step": 116760
    },
    {
      "epoch": 47.8,
      "grad_norm": 2.003530979156494,
      "learning_rate": 4.244479035051898e-06,
      "loss": 0.5302,
      "step": 116770
    },
    {
      "epoch": 47.8,
      "grad_norm": 2.1302528381347656,
      "learning_rate": 4.24438622019248e-06,
      "loss": 0.5712,
      "step": 116780
    },
    {
      "epoch": 47.81,
      "grad_norm": 2.1283302307128906,
      "learning_rate": 4.244293396626542e-06,
      "loss": 0.5548,
      "step": 116790
    },
    {
      "epoch": 47.81,
      "grad_norm": 2.026996612548828,
      "learning_rate": 4.244200564354509e-06,
      "loss": 0.5345,
      "step": 116800
    },
    {
      "epoch": 47.81,
      "grad_norm": 2.4408273696899414,
      "learning_rate": 4.244107723376807e-06,
      "loss": 0.5131,
      "step": 116810
    },
    {
      "epoch": 47.82,
      "grad_norm": 2.4800057411193848,
      "learning_rate": 4.244014873693859e-06,
      "loss": 0.5508,
      "step": 116820
    },
    {
      "epoch": 47.82,
      "grad_norm": 2.4512317180633545,
      "learning_rate": 4.243922015306093e-06,
      "loss": 0.5369,
      "step": 116830
    },
    {
      "epoch": 47.83,
      "grad_norm": 2.8617641925811768,
      "learning_rate": 4.243829148213934e-06,
      "loss": 0.5336,
      "step": 116840
    },
    {
      "epoch": 47.83,
      "grad_norm": 1.8569233417510986,
      "learning_rate": 4.243736272417806e-06,
      "loss": 0.5578,
      "step": 116850
    },
    {
      "epoch": 47.83,
      "grad_norm": 2.164182186126709,
      "learning_rate": 4.243643387918135e-06,
      "loss": 0.5454,
      "step": 116860
    },
    {
      "epoch": 47.84,
      "grad_norm": 2.168799638748169,
      "learning_rate": 4.243550494715347e-06,
      "loss": 0.5523,
      "step": 116870
    },
    {
      "epoch": 47.84,
      "grad_norm": 2.1230661869049072,
      "learning_rate": 4.2434575928098664e-06,
      "loss": 0.5645,
      "step": 116880
    },
    {
      "epoch": 47.85,
      "grad_norm": 2.323185682296753,
      "learning_rate": 4.243364682202119e-06,
      "loss": 0.5684,
      "step": 116890
    },
    {
      "epoch": 47.85,
      "grad_norm": 2.3433997631073,
      "learning_rate": 4.2432717628925325e-06,
      "loss": 0.5363,
      "step": 116900
    },
    {
      "epoch": 47.86,
      "grad_norm": 1.9925144910812378,
      "learning_rate": 4.2431788348815304e-06,
      "loss": 0.5481,
      "step": 116910
    },
    {
      "epoch": 47.86,
      "grad_norm": 1.9509038925170898,
      "learning_rate": 4.243085898169538e-06,
      "loss": 0.5196,
      "step": 116920
    },
    {
      "epoch": 47.86,
      "grad_norm": 3.856377363204956,
      "learning_rate": 4.242992952756984e-06,
      "loss": 0.5438,
      "step": 116930
    },
    {
      "epoch": 47.87,
      "grad_norm": 1.6350446939468384,
      "learning_rate": 4.242899998644291e-06,
      "loss": 0.5459,
      "step": 116940
    },
    {
      "epoch": 47.87,
      "grad_norm": 2.4064126014709473,
      "learning_rate": 4.242807035831886e-06,
      "loss": 0.5471,
      "step": 116950
    },
    {
      "epoch": 47.88,
      "grad_norm": 2.608287811279297,
      "learning_rate": 4.242714064320196e-06,
      "loss": 0.5518,
      "step": 116960
    },
    {
      "epoch": 47.88,
      "grad_norm": 2.1201250553131104,
      "learning_rate": 4.242621084109644e-06,
      "loss": 0.5589,
      "step": 116970
    },
    {
      "epoch": 47.88,
      "grad_norm": 2.2451040744781494,
      "learning_rate": 4.242528095200658e-06,
      "loss": 0.547,
      "step": 116980
    },
    {
      "epoch": 47.89,
      "grad_norm": 2.7612764835357666,
      "learning_rate": 4.242435097593664e-06,
      "loss": 0.5469,
      "step": 116990
    },
    {
      "epoch": 47.89,
      "grad_norm": 2.795191526412964,
      "learning_rate": 4.242342091289088e-06,
      "loss": 0.5481,
      "step": 117000
    },
    {
      "epoch": 47.9,
      "grad_norm": 2.077340841293335,
      "learning_rate": 4.242249076287355e-06,
      "loss": 0.5192,
      "step": 117010
    },
    {
      "epoch": 47.9,
      "grad_norm": 1.8486026525497437,
      "learning_rate": 4.242156052588891e-06,
      "loss": 0.5368,
      "step": 117020
    },
    {
      "epoch": 47.9,
      "grad_norm": 1.7978527545928955,
      "learning_rate": 4.242063020194124e-06,
      "loss": 0.531,
      "step": 117030
    },
    {
      "epoch": 47.91,
      "grad_norm": 2.1313436031341553,
      "learning_rate": 4.2419699791034785e-06,
      "loss": 0.5419,
      "step": 117040
    },
    {
      "epoch": 47.91,
      "grad_norm": 2.575660467147827,
      "learning_rate": 4.241876929317382e-06,
      "loss": 0.5536,
      "step": 117050
    },
    {
      "epoch": 47.92,
      "grad_norm": 1.7747100591659546,
      "learning_rate": 4.2417838708362595e-06,
      "loss": 0.5579,
      "step": 117060
    },
    {
      "epoch": 47.92,
      "grad_norm": 2.4728286266326904,
      "learning_rate": 4.241690803660538e-06,
      "loss": 0.527,
      "step": 117070
    },
    {
      "epoch": 47.92,
      "grad_norm": 2.958923816680908,
      "learning_rate": 4.2415977277906435e-06,
      "loss": 0.5574,
      "step": 117080
    },
    {
      "epoch": 47.93,
      "grad_norm": 1.8605141639709473,
      "learning_rate": 4.241504643227002e-06,
      "loss": 0.5284,
      "step": 117090
    },
    {
      "epoch": 47.93,
      "grad_norm": 2.0470824241638184,
      "learning_rate": 4.241411549970042e-06,
      "loss": 0.5308,
      "step": 117100
    },
    {
      "epoch": 47.94,
      "grad_norm": 2.194458246231079,
      "learning_rate": 4.241318448020187e-06,
      "loss": 0.5647,
      "step": 117110
    },
    {
      "epoch": 47.94,
      "grad_norm": 1.5726501941680908,
      "learning_rate": 4.241225337377865e-06,
      "loss": 0.5416,
      "step": 117120
    },
    {
      "epoch": 47.95,
      "grad_norm": 2.3452627658843994,
      "learning_rate": 4.241132218043502e-06,
      "loss": 0.5476,
      "step": 117130
    },
    {
      "epoch": 47.95,
      "grad_norm": 2.3877522945404053,
      "learning_rate": 4.241039090017525e-06,
      "loss": 0.5611,
      "step": 117140
    },
    {
      "epoch": 47.95,
      "grad_norm": 2.9885470867156982,
      "learning_rate": 4.240945953300361e-06,
      "loss": 0.5188,
      "step": 117150
    },
    {
      "epoch": 47.96,
      "grad_norm": 1.7208174467086792,
      "learning_rate": 4.2408528078924355e-06,
      "loss": 0.5334,
      "step": 117160
    },
    {
      "epoch": 47.96,
      "grad_norm": 2.2414181232452393,
      "learning_rate": 4.240759653794177e-06,
      "loss": 0.5307,
      "step": 117170
    },
    {
      "epoch": 47.97,
      "grad_norm": 2.0017566680908203,
      "learning_rate": 4.24066649100601e-06,
      "loss": 0.5672,
      "step": 117180
    },
    {
      "epoch": 47.97,
      "grad_norm": 1.8401755094528198,
      "learning_rate": 4.240573319528362e-06,
      "loss": 0.5563,
      "step": 117190
    },
    {
      "epoch": 47.97,
      "grad_norm": 2.1365060806274414,
      "learning_rate": 4.240480139361661e-06,
      "loss": 0.532,
      "step": 117200
    },
    {
      "epoch": 47.98,
      "grad_norm": 2.3976290225982666,
      "learning_rate": 4.240386950506333e-06,
      "loss": 0.5566,
      "step": 117210
    },
    {
      "epoch": 47.98,
      "grad_norm": 1.6923643350601196,
      "learning_rate": 4.240293752962805e-06,
      "loss": 0.5309,
      "step": 117220
    },
    {
      "epoch": 47.99,
      "grad_norm": 2.300494909286499,
      "learning_rate": 4.2402005467315025e-06,
      "loss": 0.5348,
      "step": 117230
    },
    {
      "epoch": 47.99,
      "grad_norm": 1.810646414756775,
      "learning_rate": 4.240107331812855e-06,
      "loss": 0.5455,
      "step": 117240
    },
    {
      "epoch": 47.99,
      "grad_norm": 1.6276077032089233,
      "learning_rate": 4.240014108207288e-06,
      "loss": 0.5346,
      "step": 117250
    },
    {
      "epoch": 48.0,
      "grad_norm": 2.084228992462158,
      "learning_rate": 4.239920875915229e-06,
      "loss": 0.5593,
      "step": 117260
    },
    {
      "epoch": 48.0,
      "eval_loss": 0.5446627140045166,
      "eval_runtime": 52.0344,
      "eval_samples_per_second": 66.283,
      "eval_steps_per_second": 8.302,
      "step": 117264
    },
    {
      "epoch": 48.0,
      "grad_norm": 2.694324016571045,
      "learning_rate": 4.2398276349371044e-06,
      "loss": 0.5296,
      "step": 117270
    },
    {
      "epoch": 48.01,
      "grad_norm": 3.113316535949707,
      "learning_rate": 4.239734385273342e-06,
      "loss": 0.5374,
      "step": 117280
    },
    {
      "epoch": 48.01,
      "grad_norm": 2.0471622943878174,
      "learning_rate": 4.239641126924368e-06,
      "loss": 0.544,
      "step": 117290
    },
    {
      "epoch": 48.01,
      "grad_norm": 1.7456016540527344,
      "learning_rate": 4.239547859890611e-06,
      "loss": 0.5503,
      "step": 117300
    },
    {
      "epoch": 48.02,
      "grad_norm": 2.1301302909851074,
      "learning_rate": 4.239454584172498e-06,
      "loss": 0.559,
      "step": 117310
    },
    {
      "epoch": 48.02,
      "grad_norm": 1.7443585395812988,
      "learning_rate": 4.239361299770455e-06,
      "loss": 0.5564,
      "step": 117320
    },
    {
      "epoch": 48.03,
      "grad_norm": 2.6655161380767822,
      "learning_rate": 4.239268006684911e-06,
      "loss": 0.5494,
      "step": 117330
    },
    {
      "epoch": 48.03,
      "grad_norm": 2.159179210662842,
      "learning_rate": 4.239174704916293e-06,
      "loss": 0.5481,
      "step": 117340
    },
    {
      "epoch": 48.04,
      "grad_norm": 2.5776827335357666,
      "learning_rate": 4.239081394465027e-06,
      "loss": 0.5303,
      "step": 117350
    },
    {
      "epoch": 48.04,
      "grad_norm": 2.1206278800964355,
      "learning_rate": 4.238988075331542e-06,
      "loss": 0.536,
      "step": 117360
    },
    {
      "epoch": 48.04,
      "grad_norm": 1.8438830375671387,
      "learning_rate": 4.238894747516265e-06,
      "loss": 0.5294,
      "step": 117370
    },
    {
      "epoch": 48.05,
      "grad_norm": 2.8446462154388428,
      "learning_rate": 4.2388014110196225e-06,
      "loss": 0.5387,
      "step": 117380
    },
    {
      "epoch": 48.05,
      "grad_norm": 1.8797540664672852,
      "learning_rate": 4.238708065842044e-06,
      "loss": 0.557,
      "step": 117390
    },
    {
      "epoch": 48.06,
      "grad_norm": 1.976084589958191,
      "learning_rate": 4.238614711983955e-06,
      "loss": 0.5511,
      "step": 117400
    },
    {
      "epoch": 48.06,
      "grad_norm": 1.4239602088928223,
      "learning_rate": 4.2385213494457845e-06,
      "loss": 0.5336,
      "step": 117410
    },
    {
      "epoch": 48.06,
      "grad_norm": 1.6045575141906738,
      "learning_rate": 4.238427978227961e-06,
      "loss": 0.5351,
      "step": 117420
    },
    {
      "epoch": 48.07,
      "grad_norm": 2.864057779312134,
      "learning_rate": 4.2383345983309104e-06,
      "loss": 0.5795,
      "step": 117430
    },
    {
      "epoch": 48.07,
      "grad_norm": 2.218346357345581,
      "learning_rate": 4.238241209755061e-06,
      "loss": 0.5441,
      "step": 117440
    },
    {
      "epoch": 48.08,
      "grad_norm": 2.7341432571411133,
      "learning_rate": 4.238147812500841e-06,
      "loss": 0.5381,
      "step": 117450
    },
    {
      "epoch": 48.08,
      "grad_norm": 1.6977571249008179,
      "learning_rate": 4.238054406568677e-06,
      "loss": 0.5368,
      "step": 117460
    },
    {
      "epoch": 48.08,
      "grad_norm": 2.0068106651306152,
      "learning_rate": 4.237960991958999e-06,
      "loss": 0.5447,
      "step": 117470
    },
    {
      "epoch": 48.09,
      "grad_norm": 1.830082893371582,
      "learning_rate": 4.237867568672233e-06,
      "loss": 0.5556,
      "step": 117480
    },
    {
      "epoch": 48.09,
      "grad_norm": 1.8868396282196045,
      "learning_rate": 4.237774136708809e-06,
      "loss": 0.5415,
      "step": 117490
    },
    {
      "epoch": 48.1,
      "grad_norm": 2.5626165866851807,
      "learning_rate": 4.237680696069152e-06,
      "loss": 0.543,
      "step": 117500
    },
    {
      "epoch": 48.1,
      "grad_norm": 2.0024232864379883,
      "learning_rate": 4.237587246753694e-06,
      "loss": 0.5326,
      "step": 117510
    },
    {
      "epoch": 48.1,
      "grad_norm": 2.0914881229400635,
      "learning_rate": 4.237493788762859e-06,
      "loss": 0.5466,
      "step": 117520
    },
    {
      "epoch": 48.11,
      "grad_norm": 3.141117572784424,
      "learning_rate": 4.237400322097078e-06,
      "loss": 0.5388,
      "step": 117530
    },
    {
      "epoch": 48.11,
      "grad_norm": 1.9254624843597412,
      "learning_rate": 4.237306846756778e-06,
      "loss": 0.5403,
      "step": 117540
    },
    {
      "epoch": 48.12,
      "grad_norm": 1.8690128326416016,
      "learning_rate": 4.237213362742387e-06,
      "loss": 0.5278,
      "step": 117550
    },
    {
      "epoch": 48.12,
      "grad_norm": 2.0073819160461426,
      "learning_rate": 4.237119870054334e-06,
      "loss": 0.5328,
      "step": 117560
    },
    {
      "epoch": 48.13,
      "grad_norm": 2.276616096496582,
      "learning_rate": 4.237026368693047e-06,
      "loss": 0.5306,
      "step": 117570
    },
    {
      "epoch": 48.13,
      "grad_norm": 2.00510835647583,
      "learning_rate": 4.236932858658953e-06,
      "loss": 0.5408,
      "step": 117580
    },
    {
      "epoch": 48.13,
      "grad_norm": 2.031028985977173,
      "learning_rate": 4.236839339952482e-06,
      "loss": 0.5201,
      "step": 117590
    },
    {
      "epoch": 48.14,
      "grad_norm": 2.123015880584717,
      "learning_rate": 4.236745812574062e-06,
      "loss": 0.5422,
      "step": 117600
    },
    {
      "epoch": 48.14,
      "grad_norm": 2.185748338699341,
      "learning_rate": 4.236652276524122e-06,
      "loss": 0.5408,
      "step": 117610
    },
    {
      "epoch": 48.15,
      "grad_norm": 2.5237317085266113,
      "learning_rate": 4.2365587318030894e-06,
      "loss": 0.54,
      "step": 117620
    },
    {
      "epoch": 48.15,
      "grad_norm": 1.665619134902954,
      "learning_rate": 4.2364651784113934e-06,
      "loss": 0.5376,
      "step": 117630
    },
    {
      "epoch": 48.15,
      "grad_norm": 2.0567009449005127,
      "learning_rate": 4.236371616349461e-06,
      "loss": 0.541,
      "step": 117640
    },
    {
      "epoch": 48.16,
      "grad_norm": 1.7678266763687134,
      "learning_rate": 4.236278045617724e-06,
      "loss": 0.544,
      "step": 117650
    },
    {
      "epoch": 48.16,
      "grad_norm": 2.090224504470825,
      "learning_rate": 4.236184466216608e-06,
      "loss": 0.5574,
      "step": 117660
    },
    {
      "epoch": 48.17,
      "grad_norm": 1.7617206573486328,
      "learning_rate": 4.236090878146544e-06,
      "loss": 0.5699,
      "step": 117670
    },
    {
      "epoch": 48.17,
      "grad_norm": 2.1656014919281006,
      "learning_rate": 4.235997281407958e-06,
      "loss": 0.5483,
      "step": 117680
    },
    {
      "epoch": 48.17,
      "grad_norm": 2.0065202713012695,
      "learning_rate": 4.23590367600128e-06,
      "loss": 0.5282,
      "step": 117690
    },
    {
      "epoch": 48.18,
      "grad_norm": 2.017519235610962,
      "learning_rate": 4.23581006192694e-06,
      "loss": 0.5283,
      "step": 117700
    },
    {
      "epoch": 48.18,
      "grad_norm": 2.4503514766693115,
      "learning_rate": 4.2357164391853655e-06,
      "loss": 0.5581,
      "step": 117710
    },
    {
      "epoch": 48.19,
      "grad_norm": 2.550290584564209,
      "learning_rate": 4.235622807776986e-06,
      "loss": 0.5492,
      "step": 117720
    },
    {
      "epoch": 48.19,
      "grad_norm": 1.4594924449920654,
      "learning_rate": 4.23552916770223e-06,
      "loss": 0.5228,
      "step": 117730
    },
    {
      "epoch": 48.19,
      "grad_norm": 2.6137046813964844,
      "learning_rate": 4.235435518961527e-06,
      "loss": 0.5472,
      "step": 117740
    },
    {
      "epoch": 48.2,
      "grad_norm": 2.198885917663574,
      "learning_rate": 4.2353418615553055e-06,
      "loss": 0.5671,
      "step": 117750
    },
    {
      "epoch": 48.2,
      "grad_norm": 2.315214157104492,
      "learning_rate": 4.235248195483994e-06,
      "loss": 0.5441,
      "step": 117760
    },
    {
      "epoch": 48.21,
      "grad_norm": 2.177389144897461,
      "learning_rate": 4.235154520748022e-06,
      "loss": 0.5492,
      "step": 117770
    },
    {
      "epoch": 48.21,
      "grad_norm": 2.81748366355896,
      "learning_rate": 4.23506083734782e-06,
      "loss": 0.5457,
      "step": 117780
    },
    {
      "epoch": 48.22,
      "grad_norm": 1.7969931364059448,
      "learning_rate": 4.234967145283815e-06,
      "loss": 0.5379,
      "step": 117790
    },
    {
      "epoch": 48.22,
      "grad_norm": 2.0316824913024902,
      "learning_rate": 4.2348734445564385e-06,
      "loss": 0.5572,
      "step": 117800
    },
    {
      "epoch": 48.22,
      "grad_norm": 1.7660095691680908,
      "learning_rate": 4.234779735166117e-06,
      "loss": 0.5335,
      "step": 117810
    },
    {
      "epoch": 48.23,
      "grad_norm": 2.2650418281555176,
      "learning_rate": 4.234686017113281e-06,
      "loss": 0.5417,
      "step": 117820
    },
    {
      "epoch": 48.23,
      "grad_norm": 2.17319917678833,
      "learning_rate": 4.234592290398361e-06,
      "loss": 0.5442,
      "step": 117830
    },
    {
      "epoch": 48.24,
      "grad_norm": 2.3708407878875732,
      "learning_rate": 4.2344985550217855e-06,
      "loss": 0.5279,
      "step": 117840
    },
    {
      "epoch": 48.24,
      "grad_norm": 2.2089834213256836,
      "learning_rate": 4.234404810983983e-06,
      "loss": 0.5437,
      "step": 117850
    },
    {
      "epoch": 48.24,
      "grad_norm": 2.1026816368103027,
      "learning_rate": 4.2343110582853835e-06,
      "loss": 0.5373,
      "step": 117860
    },
    {
      "epoch": 48.25,
      "grad_norm": 2.1291286945343018,
      "learning_rate": 4.234217296926416e-06,
      "loss": 0.5555,
      "step": 117870
    },
    {
      "epoch": 48.25,
      "grad_norm": 2.3080668449401855,
      "learning_rate": 4.234123526907512e-06,
      "loss": 0.5682,
      "step": 117880
    },
    {
      "epoch": 48.26,
      "grad_norm": 2.1450443267822266,
      "learning_rate": 4.234029748229098e-06,
      "loss": 0.5265,
      "step": 117890
    },
    {
      "epoch": 48.26,
      "grad_norm": 2.0684216022491455,
      "learning_rate": 4.233935960891607e-06,
      "loss": 0.5454,
      "step": 117900
    },
    {
      "epoch": 48.26,
      "grad_norm": 1.895013689994812,
      "learning_rate": 4.2338421648954655e-06,
      "loss": 0.5327,
      "step": 117910
    },
    {
      "epoch": 48.27,
      "grad_norm": 1.8292115926742554,
      "learning_rate": 4.233748360241105e-06,
      "loss": 0.5163,
      "step": 117920
    },
    {
      "epoch": 48.27,
      "grad_norm": 2.1507818698883057,
      "learning_rate": 4.233654546928955e-06,
      "loss": 0.5571,
      "step": 117930
    },
    {
      "epoch": 48.28,
      "grad_norm": 2.359595775604248,
      "learning_rate": 4.233560724959444e-06,
      "loss": 0.555,
      "step": 117940
    },
    {
      "epoch": 48.28,
      "grad_norm": 1.9298169612884521,
      "learning_rate": 4.233466894333004e-06,
      "loss": 0.549,
      "step": 117950
    },
    {
      "epoch": 48.28,
      "grad_norm": 2.2728590965270996,
      "learning_rate": 4.233373055050062e-06,
      "loss": 0.5201,
      "step": 117960
    },
    {
      "epoch": 48.29,
      "grad_norm": 1.6075186729431152,
      "learning_rate": 4.233279207111051e-06,
      "loss": 0.5435,
      "step": 117970
    },
    {
      "epoch": 48.29,
      "grad_norm": 2.2795963287353516,
      "learning_rate": 4.233185350516399e-06,
      "loss": 0.5348,
      "step": 117980
    },
    {
      "epoch": 48.3,
      "grad_norm": 2.119157075881958,
      "learning_rate": 4.233091485266536e-06,
      "loss": 0.5565,
      "step": 117990
    },
    {
      "epoch": 48.3,
      "grad_norm": 1.7376662492752075,
      "learning_rate": 4.232997611361893e-06,
      "loss": 0.5606,
      "step": 118000
    },
    {
      "epoch": 48.31,
      "grad_norm": 1.9138277769088745,
      "learning_rate": 4.232903728802898e-06,
      "loss": 0.5369,
      "step": 118010
    },
    {
      "epoch": 48.31,
      "grad_norm": 2.591014862060547,
      "learning_rate": 4.2328098375899825e-06,
      "loss": 0.5373,
      "step": 118020
    },
    {
      "epoch": 48.31,
      "grad_norm": 1.6429004669189453,
      "learning_rate": 4.232715937723578e-06,
      "loss": 0.5648,
      "step": 118030
    },
    {
      "epoch": 48.32,
      "grad_norm": 2.117587089538574,
      "learning_rate": 4.232622029204111e-06,
      "loss": 0.5491,
      "step": 118040
    },
    {
      "epoch": 48.32,
      "grad_norm": 1.940627932548523,
      "learning_rate": 4.232528112032015e-06,
      "loss": 0.5616,
      "step": 118050
    },
    {
      "epoch": 48.33,
      "grad_norm": 2.1440958976745605,
      "learning_rate": 4.232434186207718e-06,
      "loss": 0.5358,
      "step": 118060
    },
    {
      "epoch": 48.33,
      "grad_norm": 1.8895877599716187,
      "learning_rate": 4.232340251731653e-06,
      "loss": 0.5374,
      "step": 118070
    },
    {
      "epoch": 48.33,
      "grad_norm": 3.2773544788360596,
      "learning_rate": 4.232246308604247e-06,
      "loss": 0.5605,
      "step": 118080
    },
    {
      "epoch": 48.34,
      "grad_norm": 2.166050672531128,
      "learning_rate": 4.232152356825932e-06,
      "loss": 0.5469,
      "step": 118090
    },
    {
      "epoch": 48.34,
      "grad_norm": 1.7312781810760498,
      "learning_rate": 4.232058396397139e-06,
      "loss": 0.5546,
      "step": 118100
    },
    {
      "epoch": 48.35,
      "grad_norm": 1.8151023387908936,
      "learning_rate": 4.231964427318297e-06,
      "loss": 0.5449,
      "step": 118110
    },
    {
      "epoch": 48.35,
      "grad_norm": 2.486586332321167,
      "learning_rate": 4.231870449589837e-06,
      "loss": 0.5482,
      "step": 118120
    },
    {
      "epoch": 48.35,
      "grad_norm": 1.9229165315628052,
      "learning_rate": 4.231776463212191e-06,
      "loss": 0.5497,
      "step": 118130
    },
    {
      "epoch": 48.36,
      "grad_norm": 1.7571473121643066,
      "learning_rate": 4.231682468185787e-06,
      "loss": 0.5279,
      "step": 118140
    },
    {
      "epoch": 48.36,
      "grad_norm": 1.8728989362716675,
      "learning_rate": 4.231588464511056e-06,
      "loss": 0.568,
      "step": 118150
    },
    {
      "epoch": 48.37,
      "grad_norm": 1.381737232208252,
      "learning_rate": 4.231494452188431e-06,
      "loss": 0.5355,
      "step": 118160
    },
    {
      "epoch": 48.37,
      "grad_norm": 1.9984323978424072,
      "learning_rate": 4.23140043121834e-06,
      "loss": 0.5488,
      "step": 118170
    },
    {
      "epoch": 48.37,
      "grad_norm": 1.8377792835235596,
      "learning_rate": 4.231306401601216e-06,
      "loss": 0.5543,
      "step": 118180
    },
    {
      "epoch": 48.38,
      "grad_norm": 2.179736852645874,
      "learning_rate": 4.231212363337487e-06,
      "loss": 0.5438,
      "step": 118190
    },
    {
      "epoch": 48.38,
      "grad_norm": 2.033278703689575,
      "learning_rate": 4.231118316427586e-06,
      "loss": 0.5283,
      "step": 118200
    },
    {
      "epoch": 48.39,
      "grad_norm": 1.8904087543487549,
      "learning_rate": 4.231024260871943e-06,
      "loss": 0.5479,
      "step": 118210
    },
    {
      "epoch": 48.39,
      "grad_norm": 2.1131041049957275,
      "learning_rate": 4.2309301966709884e-06,
      "loss": 0.5314,
      "step": 118220
    },
    {
      "epoch": 48.4,
      "grad_norm": 2.250413179397583,
      "learning_rate": 4.230836123825153e-06,
      "loss": 0.5251,
      "step": 118230
    },
    {
      "epoch": 48.4,
      "grad_norm": 1.9582488536834717,
      "learning_rate": 4.23074204233487e-06,
      "loss": 0.5364,
      "step": 118240
    },
    {
      "epoch": 48.4,
      "grad_norm": 1.8265184164047241,
      "learning_rate": 4.230647952200568e-06,
      "loss": 0.5567,
      "step": 118250
    },
    {
      "epoch": 48.41,
      "grad_norm": 1.7451974153518677,
      "learning_rate": 4.230553853422679e-06,
      "loss": 0.5242,
      "step": 118260
    },
    {
      "epoch": 48.41,
      "grad_norm": 5.201967716217041,
      "learning_rate": 4.230459746001633e-06,
      "loss": 0.5707,
      "step": 118270
    },
    {
      "epoch": 48.42,
      "grad_norm": 1.8996961116790771,
      "learning_rate": 4.230365629937862e-06,
      "loss": 0.5643,
      "step": 118280
    },
    {
      "epoch": 48.42,
      "grad_norm": 2.2708189487457275,
      "learning_rate": 4.230271505231797e-06,
      "loss": 0.5179,
      "step": 118290
    },
    {
      "epoch": 48.42,
      "grad_norm": 1.8259233236312866,
      "learning_rate": 4.230177371883869e-06,
      "loss": 0.5416,
      "step": 118300
    },
    {
      "epoch": 48.43,
      "grad_norm": 1.92729914188385,
      "learning_rate": 4.2300832298945096e-06,
      "loss": 0.5276,
      "step": 118310
    },
    {
      "epoch": 48.43,
      "grad_norm": 2.2665772438049316,
      "learning_rate": 4.22998907926415e-06,
      "loss": 0.5428,
      "step": 118320
    },
    {
      "epoch": 48.44,
      "grad_norm": 2.1477203369140625,
      "learning_rate": 4.229894919993222e-06,
      "loss": 0.525,
      "step": 118330
    },
    {
      "epoch": 48.44,
      "grad_norm": 2.9471850395202637,
      "learning_rate": 4.229800752082155e-06,
      "loss": 0.528,
      "step": 118340
    },
    {
      "epoch": 48.44,
      "grad_norm": 1.9451982975006104,
      "learning_rate": 4.229706575531382e-06,
      "loss": 0.5297,
      "step": 118350
    },
    {
      "epoch": 48.45,
      "grad_norm": 2.1828160285949707,
      "learning_rate": 4.229612390341334e-06,
      "loss": 0.5323,
      "step": 118360
    },
    {
      "epoch": 48.45,
      "grad_norm": 1.657785415649414,
      "learning_rate": 4.2295181965124425e-06,
      "loss": 0.5297,
      "step": 118370
    },
    {
      "epoch": 48.46,
      "grad_norm": 2.335181713104248,
      "learning_rate": 4.2294239940451385e-06,
      "loss": 0.5315,
      "step": 118380
    },
    {
      "epoch": 48.46,
      "grad_norm": 2.4441583156585693,
      "learning_rate": 4.229329782939854e-06,
      "loss": 0.5393,
      "step": 118390
    },
    {
      "epoch": 48.47,
      "grad_norm": 2.1929805278778076,
      "learning_rate": 4.229235563197021e-06,
      "loss": 0.5542,
      "step": 118400
    },
    {
      "epoch": 48.47,
      "grad_norm": 2.171355724334717,
      "learning_rate": 4.229141334817071e-06,
      "loss": 0.5279,
      "step": 118410
    },
    {
      "epoch": 48.47,
      "grad_norm": 2.533348321914673,
      "learning_rate": 4.229047097800434e-06,
      "loss": 0.5499,
      "step": 118420
    },
    {
      "epoch": 48.48,
      "grad_norm": 2.238128662109375,
      "learning_rate": 4.228952852147544e-06,
      "loss": 0.5404,
      "step": 118430
    },
    {
      "epoch": 48.48,
      "grad_norm": 2.1108813285827637,
      "learning_rate": 4.228858597858832e-06,
      "loss": 0.538,
      "step": 118440
    },
    {
      "epoch": 48.49,
      "grad_norm": 2.731388568878174,
      "learning_rate": 4.228764334934729e-06,
      "loss": 0.5278,
      "step": 118450
    },
    {
      "epoch": 48.49,
      "grad_norm": 1.918203353881836,
      "learning_rate": 4.228670063375667e-06,
      "loss": 0.5565,
      "step": 118460
    },
    {
      "epoch": 48.49,
      "grad_norm": 2.2032032012939453,
      "learning_rate": 4.228575783182079e-06,
      "loss": 0.5482,
      "step": 118470
    },
    {
      "epoch": 48.5,
      "grad_norm": 1.9231964349746704,
      "learning_rate": 4.228481494354395e-06,
      "loss": 0.5481,
      "step": 118480
    },
    {
      "epoch": 48.5,
      "grad_norm": 3.1038639545440674,
      "learning_rate": 4.228387196893048e-06,
      "loss": 0.5403,
      "step": 118490
    },
    {
      "epoch": 48.51,
      "grad_norm": 1.8425297737121582,
      "learning_rate": 4.22829289079847e-06,
      "loss": 0.5409,
      "step": 118500
    },
    {
      "epoch": 48.51,
      "grad_norm": 2.701422929763794,
      "learning_rate": 4.2281985760710935e-06,
      "loss": 0.5468,
      "step": 118510
    },
    {
      "epoch": 48.51,
      "grad_norm": 1.9676895141601562,
      "learning_rate": 4.228104252711348e-06,
      "loss": 0.5326,
      "step": 118520
    },
    {
      "epoch": 48.52,
      "grad_norm": 2.1958112716674805,
      "learning_rate": 4.22800992071967e-06,
      "loss": 0.5544,
      "step": 118530
    },
    {
      "epoch": 48.52,
      "grad_norm": 2.1947832107543945,
      "learning_rate": 4.227915580096488e-06,
      "loss": 0.5371,
      "step": 118540
    },
    {
      "epoch": 48.53,
      "grad_norm": 2.0274593830108643,
      "learning_rate": 4.227821230842235e-06,
      "loss": 0.5271,
      "step": 118550
    },
    {
      "epoch": 48.53,
      "grad_norm": 1.80138099193573,
      "learning_rate": 4.227726872957344e-06,
      "loss": 0.5457,
      "step": 118560
    },
    {
      "epoch": 48.53,
      "grad_norm": 2.1681814193725586,
      "learning_rate": 4.227632506442246e-06,
      "loss": 0.5549,
      "step": 118570
    },
    {
      "epoch": 48.54,
      "grad_norm": 2.550734043121338,
      "learning_rate": 4.227538131297375e-06,
      "loss": 0.5588,
      "step": 118580
    },
    {
      "epoch": 48.54,
      "grad_norm": 2.08970308303833,
      "learning_rate": 4.227443747523161e-06,
      "loss": 0.5386,
      "step": 118590
    },
    {
      "epoch": 48.55,
      "grad_norm": 1.871956467628479,
      "learning_rate": 4.227349355120039e-06,
      "loss": 0.5257,
      "step": 118600
    },
    {
      "epoch": 48.55,
      "grad_norm": 2.6864655017852783,
      "learning_rate": 4.22725495408844e-06,
      "loss": 0.5224,
      "step": 118610
    },
    {
      "epoch": 48.56,
      "grad_norm": 1.858124017715454,
      "learning_rate": 4.227160544428795e-06,
      "loss": 0.5596,
      "step": 118620
    },
    {
      "epoch": 48.56,
      "grad_norm": 2.5897862911224365,
      "learning_rate": 4.227066126141539e-06,
      "loss": 0.5463,
      "step": 118630
    },
    {
      "epoch": 48.56,
      "grad_norm": 2.1362388134002686,
      "learning_rate": 4.226971699227103e-06,
      "loss": 0.5513,
      "step": 118640
    },
    {
      "epoch": 48.57,
      "grad_norm": 2.0577731132507324,
      "learning_rate": 4.226877263685922e-06,
      "loss": 0.522,
      "step": 118650
    },
    {
      "epoch": 48.57,
      "grad_norm": 1.8178088665008545,
      "learning_rate": 4.226782819518424e-06,
      "loss": 0.5403,
      "step": 118660
    },
    {
      "epoch": 48.58,
      "grad_norm": 1.6466424465179443,
      "learning_rate": 4.226688366725046e-06,
      "loss": 0.5353,
      "step": 118670
    },
    {
      "epoch": 48.58,
      "grad_norm": 2.734143018722534,
      "learning_rate": 4.226593905306218e-06,
      "loss": 0.5569,
      "step": 118680
    },
    {
      "epoch": 48.58,
      "grad_norm": 2.0410146713256836,
      "learning_rate": 4.226499435262374e-06,
      "loss": 0.5563,
      "step": 118690
    },
    {
      "epoch": 48.59,
      "grad_norm": 1.9541161060333252,
      "learning_rate": 4.226404956593947e-06,
      "loss": 0.5484,
      "step": 118700
    },
    {
      "epoch": 48.59,
      "grad_norm": 2.237515926361084,
      "learning_rate": 4.226310469301369e-06,
      "loss": 0.5569,
      "step": 118710
    },
    {
      "epoch": 48.6,
      "grad_norm": 1.788450837135315,
      "learning_rate": 4.226215973385072e-06,
      "loss": 0.5436,
      "step": 118720
    },
    {
      "epoch": 48.6,
      "grad_norm": 2.0362212657928467,
      "learning_rate": 4.226121468845491e-06,
      "loss": 0.5439,
      "step": 118730
    },
    {
      "epoch": 48.6,
      "grad_norm": 2.8569107055664062,
      "learning_rate": 4.226026955683057e-06,
      "loss": 0.5743,
      "step": 118740
    },
    {
      "epoch": 48.61,
      "grad_norm": 2.40720272064209,
      "learning_rate": 4.225932433898205e-06,
      "loss": 0.5298,
      "step": 118750
    },
    {
      "epoch": 48.61,
      "grad_norm": 1.9352893829345703,
      "learning_rate": 4.225837903491366e-06,
      "loss": 0.5445,
      "step": 118760
    },
    {
      "epoch": 48.62,
      "grad_norm": 2.2062559127807617,
      "learning_rate": 4.225743364462973e-06,
      "loss": 0.5349,
      "step": 118770
    },
    {
      "epoch": 48.62,
      "grad_norm": 1.8122303485870361,
      "learning_rate": 4.225648816813461e-06,
      "loss": 0.5442,
      "step": 118780
    },
    {
      "epoch": 48.62,
      "grad_norm": 2.5718212127685547,
      "learning_rate": 4.225554260543261e-06,
      "loss": 0.5466,
      "step": 118790
    },
    {
      "epoch": 48.63,
      "grad_norm": 2.267925500869751,
      "learning_rate": 4.225459695652809e-06,
      "loss": 0.5381,
      "step": 118800
    },
    {
      "epoch": 48.63,
      "grad_norm": 1.6585965156555176,
      "learning_rate": 4.225365122142535e-06,
      "loss": 0.5199,
      "step": 118810
    },
    {
      "epoch": 48.64,
      "grad_norm": 1.6667039394378662,
      "learning_rate": 4.225270540012873e-06,
      "loss": 0.5514,
      "step": 118820
    },
    {
      "epoch": 48.64,
      "grad_norm": 1.7513923645019531,
      "learning_rate": 4.225175949264257e-06,
      "loss": 0.5567,
      "step": 118830
    },
    {
      "epoch": 48.65,
      "grad_norm": 1.9484024047851562,
      "learning_rate": 4.225081349897121e-06,
      "loss": 0.5408,
      "step": 118840
    },
    {
      "epoch": 48.65,
      "grad_norm": 1.565038800239563,
      "learning_rate": 4.2249867419118965e-06,
      "loss": 0.5325,
      "step": 118850
    },
    {
      "epoch": 48.65,
      "grad_norm": 2.63590145111084,
      "learning_rate": 4.224892125309019e-06,
      "loss": 0.5567,
      "step": 118860
    },
    {
      "epoch": 48.66,
      "grad_norm": 2.71669602394104,
      "learning_rate": 4.224797500088919e-06,
      "loss": 0.5293,
      "step": 118870
    },
    {
      "epoch": 48.66,
      "grad_norm": 2.4157168865203857,
      "learning_rate": 4.2247028662520325e-06,
      "loss": 0.5429,
      "step": 118880
    },
    {
      "epoch": 48.67,
      "grad_norm": 2.958829402923584,
      "learning_rate": 4.224608223798793e-06,
      "loss": 0.5179,
      "step": 118890
    },
    {
      "epoch": 48.67,
      "grad_norm": 1.8231362104415894,
      "learning_rate": 4.224513572729633e-06,
      "loss": 0.5432,
      "step": 118900
    },
    {
      "epoch": 48.67,
      "grad_norm": 2.126033306121826,
      "learning_rate": 4.224418913044986e-06,
      "loss": 0.5278,
      "step": 118910
    },
    {
      "epoch": 48.68,
      "grad_norm": 2.0662519931793213,
      "learning_rate": 4.224324244745286e-06,
      "loss": 0.5181,
      "step": 118920
    },
    {
      "epoch": 48.68,
      "grad_norm": 2.6972036361694336,
      "learning_rate": 4.224229567830967e-06,
      "loss": 0.5555,
      "step": 118930
    },
    {
      "epoch": 48.69,
      "grad_norm": 2.509106159210205,
      "learning_rate": 4.224134882302462e-06,
      "loss": 0.5415,
      "step": 118940
    },
    {
      "epoch": 48.69,
      "grad_norm": 1.7083179950714111,
      "learning_rate": 4.224040188160204e-06,
      "loss": 0.5472,
      "step": 118950
    },
    {
      "epoch": 48.69,
      "grad_norm": 2.4902803897857666,
      "learning_rate": 4.223945485404629e-06,
      "loss": 0.5463,
      "step": 118960
    },
    {
      "epoch": 48.7,
      "grad_norm": 2.4095711708068848,
      "learning_rate": 4.22385077403617e-06,
      "loss": 0.5243,
      "step": 118970
    },
    {
      "epoch": 48.7,
      "grad_norm": 1.9425562620162964,
      "learning_rate": 4.22375605405526e-06,
      "loss": 0.5473,
      "step": 118980
    },
    {
      "epoch": 48.71,
      "grad_norm": 1.686340093612671,
      "learning_rate": 4.2236613254623336e-06,
      "loss": 0.5337,
      "step": 118990
    },
    {
      "epoch": 48.71,
      "grad_norm": 2.5449938774108887,
      "learning_rate": 4.223566588257824e-06,
      "loss": 0.5422,
      "step": 119000
    },
    {
      "epoch": 48.71,
      "grad_norm": 1.5996301174163818,
      "learning_rate": 4.223471842442167e-06,
      "loss": 0.5486,
      "step": 119010
    },
    {
      "epoch": 48.72,
      "grad_norm": 2.1628267765045166,
      "learning_rate": 4.223377088015794e-06,
      "loss": 0.5529,
      "step": 119020
    },
    {
      "epoch": 48.72,
      "grad_norm": 3.1969902515411377,
      "learning_rate": 4.223282324979141e-06,
      "loss": 0.5712,
      "step": 119030
    },
    {
      "epoch": 48.73,
      "grad_norm": 2.147526502609253,
      "learning_rate": 4.223187553332641e-06,
      "loss": 0.5295,
      "step": 119040
    },
    {
      "epoch": 48.73,
      "grad_norm": 1.9333020448684692,
      "learning_rate": 4.2230927730767295e-06,
      "loss": 0.5434,
      "step": 119050
    },
    {
      "epoch": 48.74,
      "grad_norm": 2.4209628105163574,
      "learning_rate": 4.22299798421184e-06,
      "loss": 0.5489,
      "step": 119060
    },
    {
      "epoch": 48.74,
      "grad_norm": 2.2841737270355225,
      "learning_rate": 4.222903186738405e-06,
      "loss": 0.5407,
      "step": 119070
    },
    {
      "epoch": 48.74,
      "grad_norm": 1.9462461471557617,
      "learning_rate": 4.222808380656862e-06,
      "loss": 0.5539,
      "step": 119080
    },
    {
      "epoch": 48.75,
      "grad_norm": 2.21403169631958,
      "learning_rate": 4.222713565967643e-06,
      "loss": 0.5351,
      "step": 119090
    },
    {
      "epoch": 48.75,
      "grad_norm": 2.1723856925964355,
      "learning_rate": 4.222618742671182e-06,
      "loss": 0.5347,
      "step": 119100
    },
    {
      "epoch": 48.76,
      "grad_norm": 1.731661081314087,
      "learning_rate": 4.222523910767915e-06,
      "loss": 0.5545,
      "step": 119110
    },
    {
      "epoch": 48.76,
      "grad_norm": 2.2695229053497314,
      "learning_rate": 4.222429070258276e-06,
      "loss": 0.5403,
      "step": 119120
    },
    {
      "epoch": 48.76,
      "grad_norm": 1.8527096509933472,
      "learning_rate": 4.2223342211426985e-06,
      "loss": 0.5382,
      "step": 119130
    },
    {
      "epoch": 48.77,
      "grad_norm": 1.8767116069793701,
      "learning_rate": 4.222239363421618e-06,
      "loss": 0.5289,
      "step": 119140
    },
    {
      "epoch": 48.77,
      "grad_norm": 1.845015048980713,
      "learning_rate": 4.222144497095469e-06,
      "loss": 0.5577,
      "step": 119150
    },
    {
      "epoch": 48.78,
      "grad_norm": 2.1626834869384766,
      "learning_rate": 4.222049622164685e-06,
      "loss": 0.5429,
      "step": 119160
    },
    {
      "epoch": 48.78,
      "grad_norm": 1.738495945930481,
      "learning_rate": 4.221954738629701e-06,
      "loss": 0.5395,
      "step": 119170
    },
    {
      "epoch": 48.78,
      "grad_norm": 1.8345643281936646,
      "learning_rate": 4.221859846490953e-06,
      "loss": 0.5522,
      "step": 119180
    },
    {
      "epoch": 48.79,
      "grad_norm": 1.9159694910049438,
      "learning_rate": 4.221764945748874e-06,
      "loss": 0.5539,
      "step": 119190
    },
    {
      "epoch": 48.79,
      "grad_norm": 1.8414260149002075,
      "learning_rate": 4.221670036403899e-06,
      "loss": 0.5494,
      "step": 119200
    },
    {
      "epoch": 48.8,
      "grad_norm": 2.00323748588562,
      "learning_rate": 4.221575118456464e-06,
      "loss": 0.5318,
      "step": 119210
    },
    {
      "epoch": 48.8,
      "grad_norm": 1.6734081506729126,
      "learning_rate": 4.221480191907003e-06,
      "loss": 0.5199,
      "step": 119220
    },
    {
      "epoch": 48.8,
      "grad_norm": 2.2009928226470947,
      "learning_rate": 4.2213852567559505e-06,
      "loss": 0.5483,
      "step": 119230
    },
    {
      "epoch": 48.81,
      "grad_norm": 2.4411370754241943,
      "learning_rate": 4.221290313003742e-06,
      "loss": 0.5364,
      "step": 119240
    },
    {
      "epoch": 48.81,
      "grad_norm": 2.1656835079193115,
      "learning_rate": 4.2211953606508114e-06,
      "loss": 0.5424,
      "step": 119250
    },
    {
      "epoch": 48.82,
      "grad_norm": 2.2030344009399414,
      "learning_rate": 4.221100399697595e-06,
      "loss": 0.5463,
      "step": 119260
    },
    {
      "epoch": 48.82,
      "grad_norm": 2.129693031311035,
      "learning_rate": 4.221005430144526e-06,
      "loss": 0.5303,
      "step": 119270
    },
    {
      "epoch": 48.83,
      "grad_norm": 2.2806589603424072,
      "learning_rate": 4.220910451992042e-06,
      "loss": 0.5382,
      "step": 119280
    },
    {
      "epoch": 48.83,
      "grad_norm": 2.075451374053955,
      "learning_rate": 4.220815465240576e-06,
      "loss": 0.5385,
      "step": 119290
    },
    {
      "epoch": 48.83,
      "grad_norm": 1.8231512308120728,
      "learning_rate": 4.220720469890564e-06,
      "loss": 0.533,
      "step": 119300
    },
    {
      "epoch": 48.84,
      "grad_norm": 1.5596528053283691,
      "learning_rate": 4.220625465942441e-06,
      "loss": 0.5412,
      "step": 119310
    },
    {
      "epoch": 48.84,
      "grad_norm": 1.8661913871765137,
      "learning_rate": 4.220530453396642e-06,
      "loss": 0.5508,
      "step": 119320
    },
    {
      "epoch": 48.85,
      "grad_norm": 2.0915000438690186,
      "learning_rate": 4.2204354322536026e-06,
      "loss": 0.5355,
      "step": 119330
    },
    {
      "epoch": 48.85,
      "grad_norm": 1.5893698930740356,
      "learning_rate": 4.2203404025137575e-06,
      "loss": 0.5475,
      "step": 119340
    },
    {
      "epoch": 48.85,
      "grad_norm": 1.876281976699829,
      "learning_rate": 4.220245364177543e-06,
      "loss": 0.5482,
      "step": 119350
    },
    {
      "epoch": 48.86,
      "grad_norm": 2.0897417068481445,
      "learning_rate": 4.220150317245393e-06,
      "loss": 0.5493,
      "step": 119360
    },
    {
      "epoch": 48.86,
      "grad_norm": 2.3914296627044678,
      "learning_rate": 4.220055261717744e-06,
      "loss": 0.5551,
      "step": 119370
    },
    {
      "epoch": 48.87,
      "grad_norm": 2.177720546722412,
      "learning_rate": 4.219960197595031e-06,
      "loss": 0.5466,
      "step": 119380
    },
    {
      "epoch": 48.87,
      "grad_norm": 2.436581611633301,
      "learning_rate": 4.219865124877691e-06,
      "loss": 0.5275,
      "step": 119390
    },
    {
      "epoch": 48.87,
      "grad_norm": 1.6164952516555786,
      "learning_rate": 4.2197700435661565e-06,
      "loss": 0.5534,
      "step": 119400
    },
    {
      "epoch": 48.88,
      "grad_norm": 2.3737871646881104,
      "learning_rate": 4.219674953660866e-06,
      "loss": 0.5555,
      "step": 119410
    },
    {
      "epoch": 48.88,
      "grad_norm": 1.939232349395752,
      "learning_rate": 4.219579855162253e-06,
      "loss": 0.5339,
      "step": 119420
    },
    {
      "epoch": 48.89,
      "grad_norm": 2.164555072784424,
      "learning_rate": 4.2194847480707545e-06,
      "loss": 0.5333,
      "step": 119430
    },
    {
      "epoch": 48.89,
      "grad_norm": 2.664670467376709,
      "learning_rate": 4.219389632386805e-06,
      "loss": 0.5663,
      "step": 119440
    },
    {
      "epoch": 48.89,
      "grad_norm": 2.9159021377563477,
      "learning_rate": 4.2192945081108414e-06,
      "loss": 0.5463,
      "step": 119450
    },
    {
      "epoch": 48.9,
      "grad_norm": 2.205836772918701,
      "learning_rate": 4.2191993752432985e-06,
      "loss": 0.5509,
      "step": 119460
    },
    {
      "epoch": 48.9,
      "grad_norm": 1.8866349458694458,
      "learning_rate": 4.219104233784612e-06,
      "loss": 0.5192,
      "step": 119470
    },
    {
      "epoch": 48.91,
      "grad_norm": 1.5156762599945068,
      "learning_rate": 4.219009083735219e-06,
      "loss": 0.5434,
      "step": 119480
    },
    {
      "epoch": 48.91,
      "grad_norm": 2.067742109298706,
      "learning_rate": 4.218913925095555e-06,
      "loss": 0.5493,
      "step": 119490
    },
    {
      "epoch": 48.92,
      "grad_norm": 1.8910223245620728,
      "learning_rate": 4.218818757866054e-06,
      "loss": 0.5287,
      "step": 119500
    },
    {
      "epoch": 48.92,
      "grad_norm": 2.2150909900665283,
      "learning_rate": 4.2187235820471545e-06,
      "loss": 0.5288,
      "step": 119510
    },
    {
      "epoch": 48.92,
      "grad_norm": 2.063201665878296,
      "learning_rate": 4.2186283976392915e-06,
      "loss": 0.5288,
      "step": 119520
    },
    {
      "epoch": 48.93,
      "grad_norm": 3.1074225902557373,
      "learning_rate": 4.2185332046429e-06,
      "loss": 0.5486,
      "step": 119530
    },
    {
      "epoch": 48.93,
      "grad_norm": 1.8859741687774658,
      "learning_rate": 4.218438003058417e-06,
      "loss": 0.5483,
      "step": 119540
    },
    {
      "epoch": 48.94,
      "grad_norm": 1.8391774892807007,
      "learning_rate": 4.2183427928862795e-06,
      "loss": 0.5547,
      "step": 119550
    },
    {
      "epoch": 48.94,
      "grad_norm": 1.8411756753921509,
      "learning_rate": 4.2182475741269225e-06,
      "loss": 0.5169,
      "step": 119560
    },
    {
      "epoch": 48.94,
      "grad_norm": 1.8113619089126587,
      "learning_rate": 4.218152346780782e-06,
      "loss": 0.5361,
      "step": 119570
    },
    {
      "epoch": 48.95,
      "grad_norm": 1.9306647777557373,
      "learning_rate": 4.218057110848295e-06,
      "loss": 0.5424,
      "step": 119580
    },
    {
      "epoch": 48.95,
      "grad_norm": 1.9867514371871948,
      "learning_rate": 4.217961866329898e-06,
      "loss": 0.5304,
      "step": 119590
    },
    {
      "epoch": 48.96,
      "grad_norm": 1.6197446584701538,
      "learning_rate": 4.2178666132260255e-06,
      "loss": 0.5204,
      "step": 119600
    },
    {
      "epoch": 48.96,
      "grad_norm": 2.0506625175476074,
      "learning_rate": 4.217771351537115e-06,
      "loss": 0.5136,
      "step": 119610
    },
    {
      "epoch": 48.96,
      "grad_norm": 2.540475606918335,
      "learning_rate": 4.217676081263604e-06,
      "loss": 0.5242,
      "step": 119620
    },
    {
      "epoch": 48.97,
      "grad_norm": 1.9807504415512085,
      "learning_rate": 4.217580802405927e-06,
      "loss": 0.5501,
      "step": 119630
    },
    {
      "epoch": 48.97,
      "grad_norm": 1.934715986251831,
      "learning_rate": 4.217485514964521e-06,
      "loss": 0.5491,
      "step": 119640
    },
    {
      "epoch": 48.98,
      "grad_norm": 1.8721446990966797,
      "learning_rate": 4.217390218939823e-06,
      "loss": 0.5514,
      "step": 119650
    },
    {
      "epoch": 48.98,
      "grad_norm": 1.9760644435882568,
      "learning_rate": 4.217294914332271e-06,
      "loss": 0.5428,
      "step": 119660
    },
    {
      "epoch": 48.98,
      "grad_norm": 2.1911964416503906,
      "learning_rate": 4.217199601142298e-06,
      "loss": 0.5429,
      "step": 119670
    },
    {
      "epoch": 48.99,
      "grad_norm": 2.8842127323150635,
      "learning_rate": 4.217104279370343e-06,
      "loss": 0.5339,
      "step": 119680
    },
    {
      "epoch": 48.99,
      "grad_norm": 2.5562868118286133,
      "learning_rate": 4.217008949016842e-06,
      "loss": 0.5384,
      "step": 119690
    },
    {
      "epoch": 49.0,
      "grad_norm": 2.071612596511841,
      "learning_rate": 4.216913610082233e-06,
      "loss": 0.5488,
      "step": 119700
    },
    {
      "epoch": 49.0,
      "eval_loss": 0.5431312322616577,
      "eval_runtime": 52.0089,
      "eval_samples_per_second": 66.316,
      "eval_steps_per_second": 8.306,
      "step": 119707
    },
    {
      "epoch": 49.0,
      "grad_norm": 1.8715763092041016,
      "learning_rate": 4.21681826256695e-06,
      "loss": 0.5432,
      "step": 119710
    },
    {
      "epoch": 49.01,
      "grad_norm": 2.279472827911377,
      "learning_rate": 4.216722906471432e-06,
      "loss": 0.5431,
      "step": 119720
    },
    {
      "epoch": 49.01,
      "grad_norm": 2.8034982681274414,
      "learning_rate": 4.2166275417961155e-06,
      "loss": 0.5543,
      "step": 119730
    },
    {
      "epoch": 49.01,
      "grad_norm": 2.3638412952423096,
      "learning_rate": 4.216532168541437e-06,
      "loss": 0.5468,
      "step": 119740
    },
    {
      "epoch": 49.02,
      "grad_norm": 2.266578197479248,
      "learning_rate": 4.2164367867078335e-06,
      "loss": 0.5355,
      "step": 119750
    },
    {
      "epoch": 49.02,
      "grad_norm": 2.4237589836120605,
      "learning_rate": 4.216341396295741e-06,
      "loss": 0.5295,
      "step": 119760
    },
    {
      "epoch": 49.03,
      "grad_norm": 1.997811198234558,
      "learning_rate": 4.216245997305599e-06,
      "loss": 0.533,
      "step": 119770
    },
    {
      "epoch": 49.03,
      "grad_norm": 2.662438154220581,
      "learning_rate": 4.216150589737841e-06,
      "loss": 0.5458,
      "step": 119780
    },
    {
      "epoch": 49.03,
      "grad_norm": 2.517063617706299,
      "learning_rate": 4.216055173592907e-06,
      "loss": 0.5583,
      "step": 119790
    },
    {
      "epoch": 49.04,
      "grad_norm": 2.411651611328125,
      "learning_rate": 4.215959748871232e-06,
      "loss": 0.5279,
      "step": 119800
    },
    {
      "epoch": 49.04,
      "grad_norm": 2.210735321044922,
      "learning_rate": 4.215864315573255e-06,
      "loss": 0.5454,
      "step": 119810
    },
    {
      "epoch": 49.05,
      "grad_norm": 2.038583517074585,
      "learning_rate": 4.2157688736994125e-06,
      "loss": 0.5204,
      "step": 119820
    },
    {
      "epoch": 49.05,
      "grad_norm": 1.814418077468872,
      "learning_rate": 4.21567342325014e-06,
      "loss": 0.5389,
      "step": 119830
    },
    {
      "epoch": 49.05,
      "grad_norm": 2.063260555267334,
      "learning_rate": 4.2155779642258775e-06,
      "loss": 0.5252,
      "step": 119840
    },
    {
      "epoch": 49.06,
      "grad_norm": 2.367733955383301,
      "learning_rate": 4.21548249662706e-06,
      "loss": 0.5394,
      "step": 119850
    },
    {
      "epoch": 49.06,
      "grad_norm": 2.090754985809326,
      "learning_rate": 4.215387020454127e-06,
      "loss": 0.528,
      "step": 119860
    },
    {
      "epoch": 49.07,
      "grad_norm": 2.0157089233398438,
      "learning_rate": 4.2152915357075135e-06,
      "loss": 0.5559,
      "step": 119870
    },
    {
      "epoch": 49.07,
      "grad_norm": 2.7243807315826416,
      "learning_rate": 4.215196042387659e-06,
      "loss": 0.5499,
      "step": 119880
    },
    {
      "epoch": 49.07,
      "grad_norm": 2.725738763809204,
      "learning_rate": 4.215100540495e-06,
      "loss": 0.5364,
      "step": 119890
    },
    {
      "epoch": 49.08,
      "grad_norm": 1.5104445219039917,
      "learning_rate": 4.215005030029973e-06,
      "loss": 0.5523,
      "step": 119900
    },
    {
      "epoch": 49.08,
      "grad_norm": 1.7753623723983765,
      "learning_rate": 4.214909510993017e-06,
      "loss": 0.546,
      "step": 119910
    },
    {
      "epoch": 49.09,
      "grad_norm": 1.8804975748062134,
      "learning_rate": 4.214813983384569e-06,
      "loss": 0.5379,
      "step": 119920
    },
    {
      "epoch": 49.09,
      "grad_norm": 2.0868263244628906,
      "learning_rate": 4.214718447205067e-06,
      "loss": 0.5372,
      "step": 119930
    },
    {
      "epoch": 49.1,
      "grad_norm": 2.1144680976867676,
      "learning_rate": 4.214622902454948e-06,
      "loss": 0.5217,
      "step": 119940
    },
    {
      "epoch": 49.1,
      "grad_norm": 2.4176440238952637,
      "learning_rate": 4.21452734913465e-06,
      "loss": 0.5438,
      "step": 119950
    },
    {
      "epoch": 49.1,
      "grad_norm": 1.9307265281677246,
      "learning_rate": 4.214431787244611e-06,
      "loss": 0.5237,
      "step": 119960
    },
    {
      "epoch": 49.11,
      "grad_norm": 2.1202218532562256,
      "learning_rate": 4.214336216785269e-06,
      "loss": 0.5233,
      "step": 119970
    },
    {
      "epoch": 49.11,
      "grad_norm": 1.8421213626861572,
      "learning_rate": 4.2142406377570596e-06,
      "loss": 0.5205,
      "step": 119980
    },
    {
      "epoch": 49.12,
      "grad_norm": 2.0148892402648926,
      "learning_rate": 4.2141450501604236e-06,
      "loss": 0.5386,
      "step": 119990
    },
    {
      "epoch": 49.12,
      "grad_norm": 1.8393936157226562,
      "learning_rate": 4.214049453995797e-06,
      "loss": 0.5324,
      "step": 120000
    },
    {
      "epoch": 49.12,
      "grad_norm": 2.3423497676849365,
      "learning_rate": 4.213953849263618e-06,
      "loss": 0.5475,
      "step": 120010
    },
    {
      "epoch": 49.13,
      "grad_norm": 2.8389203548431396,
      "learning_rate": 4.213858235964326e-06,
      "loss": 0.5366,
      "step": 120020
    },
    {
      "epoch": 49.13,
      "grad_norm": 1.653465747833252,
      "learning_rate": 4.213762614098356e-06,
      "loss": 0.5428,
      "step": 120030
    },
    {
      "epoch": 49.14,
      "grad_norm": 1.9589678049087524,
      "learning_rate": 4.213666983666148e-06,
      "loss": 0.528,
      "step": 120040
    },
    {
      "epoch": 49.14,
      "grad_norm": 2.134765386581421,
      "learning_rate": 4.213571344668142e-06,
      "loss": 0.552,
      "step": 120050
    },
    {
      "epoch": 49.14,
      "grad_norm": 1.9380030632019043,
      "learning_rate": 4.213475697104773e-06,
      "loss": 0.5303,
      "step": 120060
    },
    {
      "epoch": 49.15,
      "grad_norm": 1.9440969228744507,
      "learning_rate": 4.21338004097648e-06,
      "loss": 0.5294,
      "step": 120070
    },
    {
      "epoch": 49.15,
      "grad_norm": 1.5917471647262573,
      "learning_rate": 4.2132843762837e-06,
      "loss": 0.5204,
      "step": 120080
    },
    {
      "epoch": 49.16,
      "grad_norm": 2.1703436374664307,
      "learning_rate": 4.213188703026875e-06,
      "loss": 0.5471,
      "step": 120090
    },
    {
      "epoch": 49.16,
      "grad_norm": 2.141885995864868,
      "learning_rate": 4.213093021206438e-06,
      "loss": 0.551,
      "step": 120100
    },
    {
      "epoch": 49.16,
      "grad_norm": 2.051440477371216,
      "learning_rate": 4.212997330822832e-06,
      "loss": 0.5341,
      "step": 120110
    },
    {
      "epoch": 49.17,
      "grad_norm": 1.8094990253448486,
      "learning_rate": 4.212901631876493e-06,
      "loss": 0.5132,
      "step": 120120
    },
    {
      "epoch": 49.17,
      "grad_norm": 2.233011245727539,
      "learning_rate": 4.21280592436786e-06,
      "loss": 0.5341,
      "step": 120130
    },
    {
      "epoch": 49.18,
      "grad_norm": 1.8098098039627075,
      "learning_rate": 4.212710208297371e-06,
      "loss": 0.5444,
      "step": 120140
    },
    {
      "epoch": 49.18,
      "grad_norm": 2.2914485931396484,
      "learning_rate": 4.212614483665465e-06,
      "loss": 0.554,
      "step": 120150
    },
    {
      "epoch": 49.19,
      "grad_norm": 1.7483726739883423,
      "learning_rate": 4.21251875047258e-06,
      "loss": 0.5272,
      "step": 120160
    },
    {
      "epoch": 49.19,
      "grad_norm": 1.54042387008667,
      "learning_rate": 4.212423008719156e-06,
      "loss": 0.5436,
      "step": 120170
    },
    {
      "epoch": 49.19,
      "grad_norm": 2.3582239151000977,
      "learning_rate": 4.212327258405629e-06,
      "loss": 0.5373,
      "step": 120180
    },
    {
      "epoch": 49.2,
      "grad_norm": 2.4148402214050293,
      "learning_rate": 4.212231499532439e-06,
      "loss": 0.5399,
      "step": 120190
    },
    {
      "epoch": 49.2,
      "grad_norm": 2.708719253540039,
      "learning_rate": 4.212135732100025e-06,
      "loss": 0.5488,
      "step": 120200
    },
    {
      "epoch": 49.21,
      "grad_norm": 1.97689950466156,
      "learning_rate": 4.212039956108825e-06,
      "loss": 0.5327,
      "step": 120210
    },
    {
      "epoch": 49.21,
      "grad_norm": 2.2373359203338623,
      "learning_rate": 4.211944171559279e-06,
      "loss": 0.5279,
      "step": 120220
    },
    {
      "epoch": 49.21,
      "grad_norm": 2.2990188598632812,
      "learning_rate": 4.2118483784518235e-06,
      "loss": 0.5422,
      "step": 120230
    },
    {
      "epoch": 49.22,
      "grad_norm": 2.4421517848968506,
      "learning_rate": 4.2117525767868995e-06,
      "loss": 0.5605,
      "step": 120240
    },
    {
      "epoch": 49.22,
      "grad_norm": 2.2329492568969727,
      "learning_rate": 4.211656766564945e-06,
      "loss": 0.5449,
      "step": 120250
    },
    {
      "epoch": 49.23,
      "grad_norm": 2.4934511184692383,
      "learning_rate": 4.2115609477863985e-06,
      "loss": 0.546,
      "step": 120260
    },
    {
      "epoch": 49.23,
      "grad_norm": 2.35933256149292,
      "learning_rate": 4.211465120451699e-06,
      "loss": 0.5328,
      "step": 120270
    },
    {
      "epoch": 49.23,
      "grad_norm": 1.6873127222061157,
      "learning_rate": 4.211369284561286e-06,
      "loss": 0.5306,
      "step": 120280
    },
    {
      "epoch": 49.24,
      "grad_norm": 1.769020915031433,
      "learning_rate": 4.211273440115599e-06,
      "loss": 0.5541,
      "step": 120290
    },
    {
      "epoch": 49.24,
      "grad_norm": 1.7726378440856934,
      "learning_rate": 4.211177587115075e-06,
      "loss": 0.5567,
      "step": 120300
    },
    {
      "epoch": 49.25,
      "grad_norm": 1.9604991674423218,
      "learning_rate": 4.2110817255601555e-06,
      "loss": 0.5426,
      "step": 120310
    },
    {
      "epoch": 49.25,
      "grad_norm": 2.2954623699188232,
      "learning_rate": 4.210985855451279e-06,
      "loss": 0.542,
      "step": 120320
    },
    {
      "epoch": 49.26,
      "grad_norm": 2.1080129146575928,
      "learning_rate": 4.210889976788883e-06,
      "loss": 0.5207,
      "step": 120330
    },
    {
      "epoch": 49.26,
      "grad_norm": 1.8354333639144897,
      "learning_rate": 4.210794089573409e-06,
      "loss": 0.5464,
      "step": 120340
    },
    {
      "epoch": 49.26,
      "grad_norm": 2.051910400390625,
      "learning_rate": 4.2106981938052944e-06,
      "loss": 0.5368,
      "step": 120350
    },
    {
      "epoch": 49.27,
      "grad_norm": 1.6838144063949585,
      "learning_rate": 4.21060228948498e-06,
      "loss": 0.5645,
      "step": 120360
    },
    {
      "epoch": 49.27,
      "grad_norm": 1.9892724752426147,
      "learning_rate": 4.210506376612904e-06,
      "loss": 0.5359,
      "step": 120370
    },
    {
      "epoch": 49.28,
      "grad_norm": 1.7412705421447754,
      "learning_rate": 4.210410455189506e-06,
      "loss": 0.5697,
      "step": 120380
    },
    {
      "epoch": 49.28,
      "grad_norm": 1.6087863445281982,
      "learning_rate": 4.2103145252152245e-06,
      "loss": 0.529,
      "step": 120390
    },
    {
      "epoch": 49.28,
      "grad_norm": 2.164177894592285,
      "learning_rate": 4.2102185866905015e-06,
      "loss": 0.5403,
      "step": 120400
    },
    {
      "epoch": 49.29,
      "grad_norm": 2.464412212371826,
      "learning_rate": 4.2101226396157745e-06,
      "loss": 0.5488,
      "step": 120410
    },
    {
      "epoch": 49.29,
      "grad_norm": 1.7006912231445312,
      "learning_rate": 4.210026683991483e-06,
      "loss": 0.5233,
      "step": 120420
    },
    {
      "epoch": 49.3,
      "grad_norm": 2.3572943210601807,
      "learning_rate": 4.209930719818068e-06,
      "loss": 0.5226,
      "step": 120430
    },
    {
      "epoch": 49.3,
      "grad_norm": 1.5229796171188354,
      "learning_rate": 4.209834747095968e-06,
      "loss": 0.537,
      "step": 120440
    },
    {
      "epoch": 49.3,
      "grad_norm": 2.442357063293457,
      "learning_rate": 4.209738765825622e-06,
      "loss": 0.5295,
      "step": 120450
    },
    {
      "epoch": 49.31,
      "grad_norm": 2.02604079246521,
      "learning_rate": 4.209642776007471e-06,
      "loss": 0.5253,
      "step": 120460
    },
    {
      "epoch": 49.31,
      "grad_norm": 1.6720067262649536,
      "learning_rate": 4.209546777641954e-06,
      "loss": 0.5464,
      "step": 120470
    },
    {
      "epoch": 49.32,
      "grad_norm": 1.8300412893295288,
      "learning_rate": 4.209450770729511e-06,
      "loss": 0.5608,
      "step": 120480
    },
    {
      "epoch": 49.32,
      "grad_norm": 3.0343894958496094,
      "learning_rate": 4.2093547552705815e-06,
      "loss": 0.5446,
      "step": 120490
    },
    {
      "epoch": 49.32,
      "grad_norm": 2.353104591369629,
      "learning_rate": 4.209258731265607e-06,
      "loss": 0.5718,
      "step": 120500
    },
    {
      "epoch": 49.33,
      "grad_norm": 2.6356568336486816,
      "learning_rate": 4.209162698715024e-06,
      "loss": 0.5343,
      "step": 120510
    },
    {
      "epoch": 49.33,
      "grad_norm": 2.089203357696533,
      "learning_rate": 4.209066657619275e-06,
      "loss": 0.555,
      "step": 120520
    },
    {
      "epoch": 49.34,
      "grad_norm": 3.001373767852783,
      "learning_rate": 4.208970607978799e-06,
      "loss": 0.5517,
      "step": 120530
    },
    {
      "epoch": 49.34,
      "grad_norm": 2.3725688457489014,
      "learning_rate": 4.208874549794037e-06,
      "loss": 0.5531,
      "step": 120540
    },
    {
      "epoch": 49.35,
      "grad_norm": 2.552708387374878,
      "learning_rate": 4.208778483065428e-06,
      "loss": 0.5672,
      "step": 120550
    },
    {
      "epoch": 49.35,
      "grad_norm": 2.057373523712158,
      "learning_rate": 4.208682407793411e-06,
      "loss": 0.5382,
      "step": 120560
    },
    {
      "epoch": 49.35,
      "grad_norm": 2.01265287399292,
      "learning_rate": 4.208586323978429e-06,
      "loss": 0.5545,
      "step": 120570
    },
    {
      "epoch": 49.36,
      "grad_norm": 1.776511311531067,
      "learning_rate": 4.20849023162092e-06,
      "loss": 0.5544,
      "step": 120580
    },
    {
      "epoch": 49.36,
      "grad_norm": 1.9104136228561401,
      "learning_rate": 4.2083941307213255e-06,
      "loss": 0.5521,
      "step": 120590
    },
    {
      "epoch": 49.37,
      "grad_norm": 2.07564115524292,
      "learning_rate": 4.2082980212800845e-06,
      "loss": 0.5374,
      "step": 120600
    },
    {
      "epoch": 49.37,
      "grad_norm": 1.899701476097107,
      "learning_rate": 4.208201903297638e-06,
      "loss": 0.524,
      "step": 120610
    },
    {
      "epoch": 49.37,
      "grad_norm": 2.0127809047698975,
      "learning_rate": 4.2081057767744255e-06,
      "loss": 0.5464,
      "step": 120620
    },
    {
      "epoch": 49.38,
      "grad_norm": 1.9689644575119019,
      "learning_rate": 4.208009641710888e-06,
      "loss": 0.5265,
      "step": 120630
    },
    {
      "epoch": 49.38,
      "grad_norm": 2.262986421585083,
      "learning_rate": 4.207913498107466e-06,
      "loss": 0.5365,
      "step": 120640
    },
    {
      "epoch": 49.39,
      "grad_norm": 2.6727516651153564,
      "learning_rate": 4.2078173459646e-06,
      "loss": 0.5316,
      "step": 120650
    },
    {
      "epoch": 49.39,
      "grad_norm": 1.8754621744155884,
      "learning_rate": 4.20772118528273e-06,
      "loss": 0.5532,
      "step": 120660
    },
    {
      "epoch": 49.39,
      "grad_norm": 1.819629430770874,
      "learning_rate": 4.207625016062296e-06,
      "loss": 0.5677,
      "step": 120670
    },
    {
      "epoch": 49.4,
      "grad_norm": 2.2379331588745117,
      "learning_rate": 4.207528838303741e-06,
      "loss": 0.5383,
      "step": 120680
    },
    {
      "epoch": 49.4,
      "grad_norm": 2.7796471118927,
      "learning_rate": 4.207432652007502e-06,
      "loss": 0.5449,
      "step": 120690
    },
    {
      "epoch": 49.41,
      "grad_norm": 2.789504289627075,
      "learning_rate": 4.207336457174022e-06,
      "loss": 0.5395,
      "step": 120700
    },
    {
      "epoch": 49.41,
      "grad_norm": 2.4654932022094727,
      "learning_rate": 4.2072402538037415e-06,
      "loss": 0.5328,
      "step": 120710
    },
    {
      "epoch": 49.41,
      "grad_norm": 2.9568605422973633,
      "learning_rate": 4.2071440418971e-06,
      "loss": 0.5378,
      "step": 120720
    },
    {
      "epoch": 49.42,
      "grad_norm": 2.2636702060699463,
      "learning_rate": 4.207047821454539e-06,
      "loss": 0.5253,
      "step": 120730
    },
    {
      "epoch": 49.42,
      "grad_norm": 2.1534745693206787,
      "learning_rate": 4.2069515924765e-06,
      "loss": 0.545,
      "step": 120740
    },
    {
      "epoch": 49.43,
      "grad_norm": 2.417025089263916,
      "learning_rate": 4.206855354963423e-06,
      "loss": 0.5554,
      "step": 120750
    },
    {
      "epoch": 49.43,
      "grad_norm": 2.110813617706299,
      "learning_rate": 4.206759108915748e-06,
      "loss": 0.5538,
      "step": 120760
    },
    {
      "epoch": 49.44,
      "grad_norm": 2.9910902976989746,
      "learning_rate": 4.206662854333917e-06,
      "loss": 0.5571,
      "step": 120770
    },
    {
      "epoch": 49.44,
      "grad_norm": 2.3178539276123047,
      "learning_rate": 4.206566591218372e-06,
      "loss": 0.5249,
      "step": 120780
    },
    {
      "epoch": 49.44,
      "grad_norm": 2.2103874683380127,
      "learning_rate": 4.206470319569552e-06,
      "loss": 0.5251,
      "step": 120790
    },
    {
      "epoch": 49.45,
      "grad_norm": 2.137355327606201,
      "learning_rate": 4.206374039387898e-06,
      "loss": 0.528,
      "step": 120800
    },
    {
      "epoch": 49.45,
      "grad_norm": 2.6654930114746094,
      "learning_rate": 4.206277750673852e-06,
      "loss": 0.5439,
      "step": 120810
    },
    {
      "epoch": 49.46,
      "grad_norm": 2.6605169773101807,
      "learning_rate": 4.206181453427855e-06,
      "loss": 0.552,
      "step": 120820
    },
    {
      "epoch": 49.46,
      "grad_norm": 1.855904459953308,
      "learning_rate": 4.206085147650348e-06,
      "loss": 0.5225,
      "step": 120830
    },
    {
      "epoch": 49.46,
      "grad_norm": 1.7005754709243774,
      "learning_rate": 4.2059888333417715e-06,
      "loss": 0.5354,
      "step": 120840
    },
    {
      "epoch": 49.47,
      "grad_norm": 1.727523684501648,
      "learning_rate": 4.205892510502568e-06,
      "loss": 0.5317,
      "step": 120850
    },
    {
      "epoch": 49.47,
      "grad_norm": 2.3402254581451416,
      "learning_rate": 4.205796179133178e-06,
      "loss": 0.5363,
      "step": 120860
    },
    {
      "epoch": 49.48,
      "grad_norm": 1.6956862211227417,
      "learning_rate": 4.205699839234043e-06,
      "loss": 0.5429,
      "step": 120870
    },
    {
      "epoch": 49.48,
      "grad_norm": 1.8096027374267578,
      "learning_rate": 4.205603490805604e-06,
      "loss": 0.5261,
      "step": 120880
    },
    {
      "epoch": 49.48,
      "grad_norm": 2.1890869140625,
      "learning_rate": 4.205507133848303e-06,
      "loss": 0.5349,
      "step": 120890
    },
    {
      "epoch": 49.49,
      "grad_norm": 1.7263213396072388,
      "learning_rate": 4.20541076836258e-06,
      "loss": 0.5518,
      "step": 120900
    },
    {
      "epoch": 49.49,
      "grad_norm": 1.7452398538589478,
      "learning_rate": 4.205314394348877e-06,
      "loss": 0.556,
      "step": 120910
    },
    {
      "epoch": 49.5,
      "grad_norm": 2.90043044090271,
      "learning_rate": 4.205218011807637e-06,
      "loss": 0.533,
      "step": 120920
    },
    {
      "epoch": 49.5,
      "grad_norm": 2.1823971271514893,
      "learning_rate": 4.205121620739299e-06,
      "loss": 0.5449,
      "step": 120930
    },
    {
      "epoch": 49.5,
      "grad_norm": 1.4637452363967896,
      "learning_rate": 4.205025221144307e-06,
      "loss": 0.5237,
      "step": 120940
    },
    {
      "epoch": 49.51,
      "grad_norm": 1.7435269355773926,
      "learning_rate": 4.204928813023101e-06,
      "loss": 0.5251,
      "step": 120950
    },
    {
      "epoch": 49.51,
      "grad_norm": 2.040107011795044,
      "learning_rate": 4.2048323963761234e-06,
      "loss": 0.5245,
      "step": 120960
    },
    {
      "epoch": 49.52,
      "grad_norm": 1.5902345180511475,
      "learning_rate": 4.204735971203815e-06,
      "loss": 0.5406,
      "step": 120970
    },
    {
      "epoch": 49.52,
      "grad_norm": 2.2060797214508057,
      "learning_rate": 4.204639537506619e-06,
      "loss": 0.5115,
      "step": 120980
    },
    {
      "epoch": 49.53,
      "grad_norm": 2.1476354598999023,
      "learning_rate": 4.204543095284975e-06,
      "loss": 0.5363,
      "step": 120990
    },
    {
      "epoch": 49.53,
      "grad_norm": 1.665250539779663,
      "learning_rate": 4.204446644539327e-06,
      "loss": 0.5302,
      "step": 121000
    },
    {
      "epoch": 49.53,
      "grad_norm": 1.8617686033248901,
      "learning_rate": 4.204350185270115e-06,
      "loss": 0.557,
      "step": 121010
    },
    {
      "epoch": 49.54,
      "grad_norm": 2.010934829711914,
      "learning_rate": 4.204253717477783e-06,
      "loss": 0.5426,
      "step": 121020
    },
    {
      "epoch": 49.54,
      "grad_norm": 1.9672902822494507,
      "learning_rate": 4.204157241162771e-06,
      "loss": 0.5315,
      "step": 121030
    },
    {
      "epoch": 49.55,
      "grad_norm": 2.008634567260742,
      "learning_rate": 4.204060756325521e-06,
      "loss": 0.5542,
      "step": 121040
    },
    {
      "epoch": 49.55,
      "grad_norm": 1.827244758605957,
      "learning_rate": 4.203964262966476e-06,
      "loss": 0.5301,
      "step": 121050
    },
    {
      "epoch": 49.55,
      "grad_norm": 1.6849406957626343,
      "learning_rate": 4.203867761086077e-06,
      "loss": 0.5394,
      "step": 121060
    },
    {
      "epoch": 49.56,
      "grad_norm": 1.8020480871200562,
      "learning_rate": 4.203771250684767e-06,
      "loss": 0.5437,
      "step": 121070
    },
    {
      "epoch": 49.56,
      "grad_norm": 1.8339407444000244,
      "learning_rate": 4.203674731762988e-06,
      "loss": 0.5597,
      "step": 121080
    },
    {
      "epoch": 49.57,
      "grad_norm": 2.424283027648926,
      "learning_rate": 4.203578204321181e-06,
      "loss": 0.5459,
      "step": 121090
    },
    {
      "epoch": 49.57,
      "grad_norm": 2.280285358428955,
      "learning_rate": 4.2034816683597904e-06,
      "loss": 0.5627,
      "step": 121100
    },
    {
      "epoch": 49.57,
      "grad_norm": 1.889950156211853,
      "learning_rate": 4.203385123879256e-06,
      "loss": 0.5361,
      "step": 121110
    },
    {
      "epoch": 49.58,
      "grad_norm": 1.5442674160003662,
      "learning_rate": 4.203288570880022e-06,
      "loss": 0.547,
      "step": 121120
    },
    {
      "epoch": 49.58,
      "grad_norm": 1.9942654371261597,
      "learning_rate": 4.203192009362528e-06,
      "loss": 0.5232,
      "step": 121130
    },
    {
      "epoch": 49.59,
      "grad_norm": 2.0842697620391846,
      "learning_rate": 4.20309543932722e-06,
      "loss": 0.5571,
      "step": 121140
    },
    {
      "epoch": 49.59,
      "grad_norm": 1.6817423105239868,
      "learning_rate": 4.202998860774538e-06,
      "loss": 0.5448,
      "step": 121150
    },
    {
      "epoch": 49.59,
      "grad_norm": 2.154550075531006,
      "learning_rate": 4.202902273704925e-06,
      "loss": 0.5409,
      "step": 121160
    },
    {
      "epoch": 49.6,
      "grad_norm": 1.8529977798461914,
      "learning_rate": 4.202805678118822e-06,
      "loss": 0.5347,
      "step": 121170
    },
    {
      "epoch": 49.6,
      "grad_norm": 2.041316032409668,
      "learning_rate": 4.202709074016674e-06,
      "loss": 0.543,
      "step": 121180
    },
    {
      "epoch": 49.61,
      "grad_norm": 2.951836109161377,
      "learning_rate": 4.2026124613989224e-06,
      "loss": 0.5316,
      "step": 121190
    },
    {
      "epoch": 49.61,
      "grad_norm": 1.7964472770690918,
      "learning_rate": 4.2025158402660106e-06,
      "loss": 0.5576,
      "step": 121200
    },
    {
      "epoch": 49.62,
      "grad_norm": 2.387003183364868,
      "learning_rate": 4.202419210618379e-06,
      "loss": 0.5447,
      "step": 121210
    },
    {
      "epoch": 49.62,
      "grad_norm": 2.2054409980773926,
      "learning_rate": 4.202322572456472e-06,
      "loss": 0.54,
      "step": 121220
    },
    {
      "epoch": 49.62,
      "grad_norm": 2.0349161624908447,
      "learning_rate": 4.2022259257807315e-06,
      "loss": 0.5384,
      "step": 121230
    },
    {
      "epoch": 49.63,
      "grad_norm": 2.195180654525757,
      "learning_rate": 4.2021292705916005e-06,
      "loss": 0.5468,
      "step": 121240
    },
    {
      "epoch": 49.63,
      "grad_norm": 1.910887360572815,
      "learning_rate": 4.202032606889522e-06,
      "loss": 0.5272,
      "step": 121250
    },
    {
      "epoch": 49.64,
      "grad_norm": 2.447509765625,
      "learning_rate": 4.20193593467494e-06,
      "loss": 0.5614,
      "step": 121260
    },
    {
      "epoch": 49.64,
      "grad_norm": 1.8934621810913086,
      "learning_rate": 4.201839253948295e-06,
      "loss": 0.5473,
      "step": 121270
    },
    {
      "epoch": 49.64,
      "grad_norm": 1.5493533611297607,
      "learning_rate": 4.201742564710031e-06,
      "loss": 0.5647,
      "step": 121280
    },
    {
      "epoch": 49.65,
      "grad_norm": 1.8433918952941895,
      "learning_rate": 4.20164586696059e-06,
      "loss": 0.5614,
      "step": 121290
    },
    {
      "epoch": 49.65,
      "grad_norm": 1.6985023021697998,
      "learning_rate": 4.201549160700417e-06,
      "loss": 0.5497,
      "step": 121300
    },
    {
      "epoch": 49.66,
      "grad_norm": 1.9048694372177124,
      "learning_rate": 4.201452445929953e-06,
      "loss": 0.5423,
      "step": 121310
    },
    {
      "epoch": 49.66,
      "grad_norm": 2.1657607555389404,
      "learning_rate": 4.201355722649642e-06,
      "loss": 0.5272,
      "step": 121320
    },
    {
      "epoch": 49.66,
      "grad_norm": 2.2443838119506836,
      "learning_rate": 4.201258990859928e-06,
      "loss": 0.5399,
      "step": 121330
    },
    {
      "epoch": 49.67,
      "grad_norm": 2.8185577392578125,
      "learning_rate": 4.201162250561251e-06,
      "loss": 0.5235,
      "step": 121340
    },
    {
      "epoch": 49.67,
      "grad_norm": 2.6267988681793213,
      "learning_rate": 4.201065501754057e-06,
      "loss": 0.5195,
      "step": 121350
    },
    {
      "epoch": 49.68,
      "grad_norm": 2.0333540439605713,
      "learning_rate": 4.200968744438789e-06,
      "loss": 0.5622,
      "step": 121360
    },
    {
      "epoch": 49.68,
      "grad_norm": 1.5055603981018066,
      "learning_rate": 4.200871978615888e-06,
      "loss": 0.5475,
      "step": 121370
    },
    {
      "epoch": 49.68,
      "grad_norm": 1.8047524690628052,
      "learning_rate": 4.2007752042858e-06,
      "loss": 0.5495,
      "step": 121380
    },
    {
      "epoch": 49.69,
      "grad_norm": 2.4003007411956787,
      "learning_rate": 4.200678421448967e-06,
      "loss": 0.5159,
      "step": 121390
    },
    {
      "epoch": 49.69,
      "grad_norm": 2.6152164936065674,
      "learning_rate": 4.200581630105832e-06,
      "loss": 0.5454,
      "step": 121400
    },
    {
      "epoch": 49.7,
      "grad_norm": 3.272564172744751,
      "learning_rate": 4.20048483025684e-06,
      "loss": 0.5605,
      "step": 121410
    },
    {
      "epoch": 49.7,
      "grad_norm": 1.6240988969802856,
      "learning_rate": 4.2003880219024316e-06,
      "loss": 0.5375,
      "step": 121420
    },
    {
      "epoch": 49.71,
      "grad_norm": 2.297600507736206,
      "learning_rate": 4.2002912050430534e-06,
      "loss": 0.5406,
      "step": 121430
    },
    {
      "epoch": 49.71,
      "grad_norm": 2.1224265098571777,
      "learning_rate": 4.200194379679148e-06,
      "loss": 0.543,
      "step": 121440
    },
    {
      "epoch": 49.71,
      "grad_norm": 1.8894740343093872,
      "learning_rate": 4.2000975458111566e-06,
      "loss": 0.5292,
      "step": 121450
    },
    {
      "epoch": 49.72,
      "grad_norm": 2.4081311225891113,
      "learning_rate": 4.200000703439525e-06,
      "loss": 0.5419,
      "step": 121460
    },
    {
      "epoch": 49.72,
      "grad_norm": 1.8796931505203247,
      "learning_rate": 4.199903852564698e-06,
      "loss": 0.5198,
      "step": 121470
    },
    {
      "epoch": 49.73,
      "grad_norm": 1.9129172563552856,
      "learning_rate": 4.199806993187115e-06,
      "loss": 0.5464,
      "step": 121480
    },
    {
      "epoch": 49.73,
      "grad_norm": 2.82035756111145,
      "learning_rate": 4.199710125307224e-06,
      "loss": 0.5301,
      "step": 121490
    },
    {
      "epoch": 49.73,
      "grad_norm": 2.353574275970459,
      "learning_rate": 4.199613248925467e-06,
      "loss": 0.5417,
      "step": 121500
    },
    {
      "epoch": 49.74,
      "grad_norm": 2.3505191802978516,
      "learning_rate": 4.199516364042288e-06,
      "loss": 0.5427,
      "step": 121510
    },
    {
      "epoch": 49.74,
      "grad_norm": 1.671582579612732,
      "learning_rate": 4.19941947065813e-06,
      "loss": 0.5412,
      "step": 121520
    },
    {
      "epoch": 49.75,
      "grad_norm": 2.8988146781921387,
      "learning_rate": 4.199322568773438e-06,
      "loss": 0.5318,
      "step": 121530
    },
    {
      "epoch": 49.75,
      "grad_norm": 2.069045066833496,
      "learning_rate": 4.199225658388655e-06,
      "loss": 0.5345,
      "step": 121540
    },
    {
      "epoch": 49.75,
      "grad_norm": 2.318685531616211,
      "learning_rate": 4.199128739504226e-06,
      "loss": 0.5757,
      "step": 121550
    },
    {
      "epoch": 49.76,
      "grad_norm": 2.688971757888794,
      "learning_rate": 4.199031812120594e-06,
      "loss": 0.5313,
      "step": 121560
    },
    {
      "epoch": 49.76,
      "grad_norm": 2.2982518672943115,
      "learning_rate": 4.198934876238204e-06,
      "loss": 0.5392,
      "step": 121570
    },
    {
      "epoch": 49.77,
      "grad_norm": 2.247117519378662,
      "learning_rate": 4.198837931857499e-06,
      "loss": 0.5447,
      "step": 121580
    },
    {
      "epoch": 49.77,
      "grad_norm": 1.9128307104110718,
      "learning_rate": 4.198740978978923e-06,
      "loss": 0.5571,
      "step": 121590
    },
    {
      "epoch": 49.77,
      "grad_norm": 2.4582479000091553,
      "learning_rate": 4.198644017602921e-06,
      "loss": 0.532,
      "step": 121600
    },
    {
      "epoch": 49.78,
      "grad_norm": 1.7849130630493164,
      "learning_rate": 4.198547047729936e-06,
      "loss": 0.5597,
      "step": 121610
    },
    {
      "epoch": 49.78,
      "grad_norm": 2.9665839672088623,
      "learning_rate": 4.198450069360414e-06,
      "loss": 0.5575,
      "step": 121620
    },
    {
      "epoch": 49.79,
      "grad_norm": 1.880323052406311,
      "learning_rate": 4.198353082494799e-06,
      "loss": 0.5627,
      "step": 121630
    },
    {
      "epoch": 49.79,
      "grad_norm": 2.0908150672912598,
      "learning_rate": 4.198256087133532e-06,
      "loss": 0.5329,
      "step": 121640
    },
    {
      "epoch": 49.8,
      "grad_norm": 2.1445915699005127,
      "learning_rate": 4.198159083277062e-06,
      "loss": 0.553,
      "step": 121650
    },
    {
      "epoch": 49.8,
      "grad_norm": 1.3455356359481812,
      "learning_rate": 4.19806207092583e-06,
      "loss": 0.5301,
      "step": 121660
    },
    {
      "epoch": 49.8,
      "grad_norm": 1.8978345394134521,
      "learning_rate": 4.197965050080282e-06,
      "loss": 0.5313,
      "step": 121670
    },
    {
      "epoch": 49.81,
      "grad_norm": 2.356736421585083,
      "learning_rate": 4.197868020740862e-06,
      "loss": 0.5541,
      "step": 121680
    },
    {
      "epoch": 49.81,
      "grad_norm": 2.2126481533050537,
      "learning_rate": 4.197770982908014e-06,
      "loss": 0.5611,
      "step": 121690
    },
    {
      "epoch": 49.82,
      "grad_norm": 2.1065831184387207,
      "learning_rate": 4.197673936582184e-06,
      "loss": 0.5466,
      "step": 121700
    },
    {
      "epoch": 49.82,
      "grad_norm": 2.2534241676330566,
      "learning_rate": 4.197576881763815e-06,
      "loss": 0.5477,
      "step": 121710
    },
    {
      "epoch": 49.82,
      "grad_norm": 2.564521074295044,
      "learning_rate": 4.197479818453352e-06,
      "loss": 0.5256,
      "step": 121720
    },
    {
      "epoch": 49.83,
      "grad_norm": 1.8758074045181274,
      "learning_rate": 4.19738274665124e-06,
      "loss": 0.5708,
      "step": 121730
    },
    {
      "epoch": 49.83,
      "grad_norm": 1.9118707180023193,
      "learning_rate": 4.197285666357924e-06,
      "loss": 0.5474,
      "step": 121740
    },
    {
      "epoch": 49.84,
      "grad_norm": 2.359640598297119,
      "learning_rate": 4.197188577573848e-06,
      "loss": 0.5536,
      "step": 121750
    },
    {
      "epoch": 49.84,
      "grad_norm": 1.9018983840942383,
      "learning_rate": 4.197091480299455e-06,
      "loss": 0.5229,
      "step": 121760
    },
    {
      "epoch": 49.84,
      "grad_norm": 3.1257615089416504,
      "learning_rate": 4.196994374535194e-06,
      "loss": 0.5319,
      "step": 121770
    },
    {
      "epoch": 49.85,
      "grad_norm": 2.272169828414917,
      "learning_rate": 4.196897260281507e-06,
      "loss": 0.5424,
      "step": 121780
    },
    {
      "epoch": 49.85,
      "grad_norm": 1.9817912578582764,
      "learning_rate": 4.196800137538839e-06,
      "loss": 0.5534,
      "step": 121790
    },
    {
      "epoch": 49.86,
      "grad_norm": 2.9890220165252686,
      "learning_rate": 4.196703006307636e-06,
      "loss": 0.5337,
      "step": 121800
    },
    {
      "epoch": 49.86,
      "grad_norm": 2.17195987701416,
      "learning_rate": 4.196605866588342e-06,
      "loss": 0.5614,
      "step": 121810
    },
    {
      "epoch": 49.86,
      "grad_norm": 2.1738266944885254,
      "learning_rate": 4.196508718381401e-06,
      "loss": 0.5083,
      "step": 121820
    },
    {
      "epoch": 49.87,
      "grad_norm": 1.9757596254348755,
      "learning_rate": 4.196411561687261e-06,
      "loss": 0.5417,
      "step": 121830
    },
    {
      "epoch": 49.87,
      "grad_norm": 1.868606686592102,
      "learning_rate": 4.196314396506364e-06,
      "loss": 0.5548,
      "step": 121840
    },
    {
      "epoch": 49.88,
      "grad_norm": 2.1633951663970947,
      "learning_rate": 4.196217222839157e-06,
      "loss": 0.5333,
      "step": 121850
    },
    {
      "epoch": 49.88,
      "grad_norm": 2.0796377658843994,
      "learning_rate": 4.196120040686084e-06,
      "loss": 0.5152,
      "step": 121860
    },
    {
      "epoch": 49.89,
      "grad_norm": 2.1271045207977295,
      "learning_rate": 4.196022850047591e-06,
      "loss": 0.5322,
      "step": 121870
    },
    {
      "epoch": 49.89,
      "grad_norm": 1.8198862075805664,
      "learning_rate": 4.1959256509241235e-06,
      "loss": 0.5743,
      "step": 121880
    },
    {
      "epoch": 49.89,
      "grad_norm": 2.0090270042419434,
      "learning_rate": 4.195828443316126e-06,
      "loss": 0.5318,
      "step": 121890
    },
    {
      "epoch": 49.9,
      "grad_norm": 1.9833732843399048,
      "learning_rate": 4.195731227224044e-06,
      "loss": 0.5405,
      "step": 121900
    },
    {
      "epoch": 49.9,
      "grad_norm": 2.4722976684570312,
      "learning_rate": 4.195634002648322e-06,
      "loss": 0.5697,
      "step": 121910
    },
    {
      "epoch": 49.91,
      "grad_norm": 1.6796042919158936,
      "learning_rate": 4.195536769589407e-06,
      "loss": 0.5325,
      "step": 121920
    },
    {
      "epoch": 49.91,
      "grad_norm": 1.503589153289795,
      "learning_rate": 4.195439528047743e-06,
      "loss": 0.5268,
      "step": 121930
    },
    {
      "epoch": 49.91,
      "grad_norm": 2.294424057006836,
      "learning_rate": 4.195342278023777e-06,
      "loss": 0.541,
      "step": 121940
    },
    {
      "epoch": 49.92,
      "grad_norm": 2.5219833850860596,
      "learning_rate": 4.195245019517953e-06,
      "loss": 0.5498,
      "step": 121950
    },
    {
      "epoch": 49.92,
      "grad_norm": 1.6542359590530396,
      "learning_rate": 4.195147752530716e-06,
      "loss": 0.5557,
      "step": 121960
    },
    {
      "epoch": 49.93,
      "grad_norm": 1.6847577095031738,
      "learning_rate": 4.195050477062514e-06,
      "loss": 0.5613,
      "step": 121970
    },
    {
      "epoch": 49.93,
      "grad_norm": 1.8458982706069946,
      "learning_rate": 4.194953193113791e-06,
      "loss": 0.5453,
      "step": 121980
    },
    {
      "epoch": 49.93,
      "grad_norm": 2.0752370357513428,
      "learning_rate": 4.194855900684993e-06,
      "loss": 0.5305,
      "step": 121990
    },
    {
      "epoch": 49.94,
      "grad_norm": 2.001678943634033,
      "learning_rate": 4.194758599776565e-06,
      "loss": 0.5039,
      "step": 122000
    },
    {
      "epoch": 49.94,
      "grad_norm": 1.543367862701416,
      "learning_rate": 4.194661290388954e-06,
      "loss": 0.5402,
      "step": 122010
    },
    {
      "epoch": 49.95,
      "grad_norm": 1.910407543182373,
      "learning_rate": 4.194563972522605e-06,
      "loss": 0.5358,
      "step": 122020
    },
    {
      "epoch": 49.95,
      "grad_norm": 2.1790664196014404,
      "learning_rate": 4.194466646177963e-06,
      "loss": 0.5377,
      "step": 122030
    },
    {
      "epoch": 49.95,
      "grad_norm": 2.1340582370758057,
      "learning_rate": 4.1943693113554745e-06,
      "loss": 0.5617,
      "step": 122040
    },
    {
      "epoch": 49.96,
      "grad_norm": 1.4845908880233765,
      "learning_rate": 4.194271968055586e-06,
      "loss": 0.536,
      "step": 122050
    },
    {
      "epoch": 49.96,
      "grad_norm": 1.9697530269622803,
      "learning_rate": 4.194174616278743e-06,
      "loss": 0.549,
      "step": 122060
    },
    {
      "epoch": 49.97,
      "grad_norm": 2.1819241046905518,
      "learning_rate": 4.194077256025392e-06,
      "loss": 0.5146,
      "step": 122070
    },
    {
      "epoch": 49.97,
      "grad_norm": 2.866708755493164,
      "learning_rate": 4.1939798872959774e-06,
      "loss": 0.5326,
      "step": 122080
    },
    {
      "epoch": 49.98,
      "grad_norm": 1.961320161819458,
      "learning_rate": 4.193882510090947e-06,
      "loss": 0.5429,
      "step": 122090
    },
    {
      "epoch": 49.98,
      "grad_norm": 1.6701503992080688,
      "learning_rate": 4.193785124410745e-06,
      "loss": 0.5423,
      "step": 122100
    },
    {
      "epoch": 49.98,
      "grad_norm": 3.047852039337158,
      "learning_rate": 4.19368773025582e-06,
      "loss": 0.5486,
      "step": 122110
    },
    {
      "epoch": 49.99,
      "grad_norm": 2.25203013420105,
      "learning_rate": 4.1935903276266165e-06,
      "loss": 0.5456,
      "step": 122120
    },
    {
      "epoch": 49.99,
      "grad_norm": 2.596092462539673,
      "learning_rate": 4.193492916523581e-06,
      "loss": 0.5384,
      "step": 122130
    },
    {
      "epoch": 50.0,
      "grad_norm": 2.390962839126587,
      "learning_rate": 4.193395496947159e-06,
      "loss": 0.5564,
      "step": 122140
    },
    {
      "epoch": 50.0,
      "grad_norm": 2.10553240776062,
      "learning_rate": 4.193298068897798e-06,
      "loss": 0.5203,
      "step": 122150
    },
    {
      "epoch": 50.0,
      "eval_loss": 0.5391061902046204,
      "eval_runtime": 51.893,
      "eval_samples_per_second": 66.464,
      "eval_steps_per_second": 8.325,
      "step": 122150
    },
    {
      "epoch": 50.0,
      "grad_norm": 1.7846674919128418,
      "learning_rate": 4.1932006323759426e-06,
      "loss": 0.5567,
      "step": 122160
    },
    {
      "epoch": 50.01,
      "grad_norm": 1.8221369981765747,
      "learning_rate": 4.193103187382041e-06,
      "loss": 0.5297,
      "step": 122170
    },
    {
      "epoch": 50.01,
      "grad_norm": 1.7331582307815552,
      "learning_rate": 4.193005733916539e-06,
      "loss": 0.5508,
      "step": 122180
    },
    {
      "epoch": 50.02,
      "grad_norm": 1.8656268119812012,
      "learning_rate": 4.1929082719798835e-06,
      "loss": 0.5256,
      "step": 122190
    },
    {
      "epoch": 50.02,
      "grad_norm": 2.2688305377960205,
      "learning_rate": 4.192810801572519e-06,
      "loss": 0.5405,
      "step": 122200
    },
    {
      "epoch": 50.02,
      "grad_norm": 1.7808105945587158,
      "learning_rate": 4.192713322694894e-06,
      "loss": 0.5449,
      "step": 122210
    },
    {
      "epoch": 50.03,
      "grad_norm": 1.823649525642395,
      "learning_rate": 4.192615835347455e-06,
      "loss": 0.5256,
      "step": 122220
    },
    {
      "epoch": 50.03,
      "grad_norm": 2.3442931175231934,
      "learning_rate": 4.1925183395306465e-06,
      "loss": 0.5446,
      "step": 122230
    },
    {
      "epoch": 50.04,
      "grad_norm": 2.578320026397705,
      "learning_rate": 4.192420835244917e-06,
      "loss": 0.5675,
      "step": 122240
    },
    {
      "epoch": 50.04,
      "grad_norm": 1.7100093364715576,
      "learning_rate": 4.1923233224907135e-06,
      "loss": 0.532,
      "step": 122250
    },
    {
      "epoch": 50.05,
      "grad_norm": 2.188908815383911,
      "learning_rate": 4.192225801268482e-06,
      "loss": 0.5436,
      "step": 122260
    },
    {
      "epoch": 50.05,
      "grad_norm": 2.4216980934143066,
      "learning_rate": 4.192128271578668e-06,
      "loss": 0.5491,
      "step": 122270
    },
    {
      "epoch": 50.05,
      "grad_norm": 2.4679861068725586,
      "learning_rate": 4.19203073342172e-06,
      "loss": 0.5439,
      "step": 122280
    },
    {
      "epoch": 50.06,
      "grad_norm": 2.637556791305542,
      "learning_rate": 4.191933186798084e-06,
      "loss": 0.5383,
      "step": 122290
    },
    {
      "epoch": 50.06,
      "grad_norm": 1.6028584241867065,
      "learning_rate": 4.191835631708207e-06,
      "loss": 0.5286,
      "step": 122300
    },
    {
      "epoch": 50.07,
      "grad_norm": 2.0741384029388428,
      "learning_rate": 4.191738068152537e-06,
      "loss": 0.5451,
      "step": 122310
    },
    {
      "epoch": 50.07,
      "grad_norm": 2.043733835220337,
      "learning_rate": 4.191640496131519e-06,
      "loss": 0.5344,
      "step": 122320
    },
    {
      "epoch": 50.07,
      "grad_norm": 3.047941207885742,
      "learning_rate": 4.1915429156456e-06,
      "loss": 0.5387,
      "step": 122330
    },
    {
      "epoch": 50.08,
      "grad_norm": 2.0868582725524902,
      "learning_rate": 4.191445326695229e-06,
      "loss": 0.5305,
      "step": 122340
    },
    {
      "epoch": 50.08,
      "grad_norm": 2.7992262840270996,
      "learning_rate": 4.191347729280851e-06,
      "loss": 0.5405,
      "step": 122350
    },
    {
      "epoch": 50.09,
      "grad_norm": 2.1216979026794434,
      "learning_rate": 4.191250123402915e-06,
      "loss": 0.5419,
      "step": 122360
    },
    {
      "epoch": 50.09,
      "grad_norm": 2.192042350769043,
      "learning_rate": 4.191152509061867e-06,
      "loss": 0.5256,
      "step": 122370
    },
    {
      "epoch": 50.09,
      "grad_norm": 2.001722812652588,
      "learning_rate": 4.191054886258153e-06,
      "loss": 0.5349,
      "step": 122380
    },
    {
      "epoch": 50.1,
      "grad_norm": 2.2016427516937256,
      "learning_rate": 4.190957254992222e-06,
      "loss": 0.5332,
      "step": 122390
    },
    {
      "epoch": 50.1,
      "grad_norm": 1.803343415260315,
      "learning_rate": 4.190859615264521e-06,
      "loss": 0.5294,
      "step": 122400
    },
    {
      "epoch": 50.11,
      "grad_norm": 1.9015214443206787,
      "learning_rate": 4.190761967075497e-06,
      "loss": 0.5594,
      "step": 122410
    },
    {
      "epoch": 50.11,
      "grad_norm": 2.299959897994995,
      "learning_rate": 4.1906643104255975e-06,
      "loss": 0.5369,
      "step": 122420
    },
    {
      "epoch": 50.11,
      "grad_norm": 2.7084124088287354,
      "learning_rate": 4.190566645315268e-06,
      "loss": 0.5477,
      "step": 122430
    },
    {
      "epoch": 50.12,
      "grad_norm": 2.8222484588623047,
      "learning_rate": 4.190468971744959e-06,
      "loss": 0.5196,
      "step": 122440
    },
    {
      "epoch": 50.12,
      "grad_norm": 1.9232608079910278,
      "learning_rate": 4.190371289715115e-06,
      "loss": 0.5385,
      "step": 122450
    },
    {
      "epoch": 50.13,
      "grad_norm": 1.9963696002960205,
      "learning_rate": 4.1902735992261866e-06,
      "loss": 0.5355,
      "step": 122460
    },
    {
      "epoch": 50.13,
      "grad_norm": 2.0984182357788086,
      "learning_rate": 4.190175900278618e-06,
      "loss": 0.5468,
      "step": 122470
    },
    {
      "epoch": 50.14,
      "grad_norm": 2.2711875438690186,
      "learning_rate": 4.190078192872859e-06,
      "loss": 0.5455,
      "step": 122480
    },
    {
      "epoch": 50.14,
      "grad_norm": 1.9526379108428955,
      "learning_rate": 4.189980477009357e-06,
      "loss": 0.5664,
      "step": 122490
    },
    {
      "epoch": 50.14,
      "grad_norm": 1.8573939800262451,
      "learning_rate": 4.189882752688559e-06,
      "loss": 0.5197,
      "step": 122500
    },
    {
      "epoch": 50.15,
      "grad_norm": 1.9013457298278809,
      "learning_rate": 4.189785019910911e-06,
      "loss": 0.5453,
      "step": 122510
    },
    {
      "epoch": 50.15,
      "grad_norm": 1.9417601823806763,
      "learning_rate": 4.1896872786768646e-06,
      "loss": 0.5478,
      "step": 122520
    },
    {
      "epoch": 50.16,
      "grad_norm": 1.8212018013000488,
      "learning_rate": 4.189589528986865e-06,
      "loss": 0.5448,
      "step": 122530
    },
    {
      "epoch": 50.16,
      "grad_norm": 2.200613021850586,
      "learning_rate": 4.189491770841359e-06,
      "loss": 0.5402,
      "step": 122540
    },
    {
      "epoch": 50.16,
      "grad_norm": 2.2937331199645996,
      "learning_rate": 4.189394004240797e-06,
      "loss": 0.5246,
      "step": 122550
    },
    {
      "epoch": 50.17,
      "grad_norm": 2.278667688369751,
      "learning_rate": 4.189296229185626e-06,
      "loss": 0.5422,
      "step": 122560
    },
    {
      "epoch": 50.17,
      "grad_norm": 1.8457409143447876,
      "learning_rate": 4.189198445676293e-06,
      "loss": 0.5289,
      "step": 122570
    },
    {
      "epoch": 50.18,
      "grad_norm": 1.5175503492355347,
      "learning_rate": 4.189100653713247e-06,
      "loss": 0.5364,
      "step": 122580
    },
    {
      "epoch": 50.18,
      "grad_norm": 1.8273704051971436,
      "learning_rate": 4.189002853296934e-06,
      "loss": 0.5345,
      "step": 122590
    },
    {
      "epoch": 50.18,
      "grad_norm": 1.806451678276062,
      "learning_rate": 4.188905044427804e-06,
      "loss": 0.5379,
      "step": 122600
    },
    {
      "epoch": 50.19,
      "grad_norm": 2.059976577758789,
      "learning_rate": 4.188807227106305e-06,
      "loss": 0.5311,
      "step": 122610
    },
    {
      "epoch": 50.19,
      "grad_norm": 2.8539540767669678,
      "learning_rate": 4.188709401332885e-06,
      "loss": 0.5467,
      "step": 122620
    },
    {
      "epoch": 50.2,
      "grad_norm": 2.9009695053100586,
      "learning_rate": 4.188611567107991e-06,
      "loss": 0.548,
      "step": 122630
    },
    {
      "epoch": 50.2,
      "grad_norm": 2.5830061435699463,
      "learning_rate": 4.1885137244320715e-06,
      "loss": 0.5323,
      "step": 122640
    },
    {
      "epoch": 50.2,
      "grad_norm": 2.283228635787964,
      "learning_rate": 4.1884158733055755e-06,
      "loss": 0.5387,
      "step": 122650
    },
    {
      "epoch": 50.21,
      "grad_norm": 1.758600115776062,
      "learning_rate": 4.188318013728951e-06,
      "loss": 0.5623,
      "step": 122660
    },
    {
      "epoch": 50.21,
      "grad_norm": 1.5719480514526367,
      "learning_rate": 4.1882201457026465e-06,
      "loss": 0.5306,
      "step": 122670
    },
    {
      "epoch": 50.22,
      "grad_norm": 2.1646718978881836,
      "learning_rate": 4.188122269227109e-06,
      "loss": 0.5481,
      "step": 122680
    },
    {
      "epoch": 50.22,
      "grad_norm": 2.15242862701416,
      "learning_rate": 4.188024384302788e-06,
      "loss": 0.5421,
      "step": 122690
    },
    {
      "epoch": 50.23,
      "grad_norm": 2.342621326446533,
      "learning_rate": 4.187926490930132e-06,
      "loss": 0.5164,
      "step": 122700
    },
    {
      "epoch": 50.23,
      "grad_norm": 2.1026391983032227,
      "learning_rate": 4.187828589109589e-06,
      "loss": 0.5497,
      "step": 122710
    },
    {
      "epoch": 50.23,
      "grad_norm": 2.1064083576202393,
      "learning_rate": 4.187730678841608e-06,
      "loss": 0.5319,
      "step": 122720
    },
    {
      "epoch": 50.24,
      "grad_norm": 2.057589530944824,
      "learning_rate": 4.187632760126636e-06,
      "loss": 0.551,
      "step": 122730
    },
    {
      "epoch": 50.24,
      "grad_norm": 1.826912522315979,
      "learning_rate": 4.187534832965124e-06,
      "loss": 0.5162,
      "step": 122740
    },
    {
      "epoch": 50.25,
      "grad_norm": 2.221766471862793,
      "learning_rate": 4.1874368973575185e-06,
      "loss": 0.5377,
      "step": 122750
    },
    {
      "epoch": 50.25,
      "grad_norm": 1.8194912672042847,
      "learning_rate": 4.1873389533042685e-06,
      "loss": 0.534,
      "step": 122760
    },
    {
      "epoch": 50.25,
      "grad_norm": 2.3252906799316406,
      "learning_rate": 4.187241000805824e-06,
      "loss": 0.5382,
      "step": 122770
    },
    {
      "epoch": 50.26,
      "grad_norm": 1.9628671407699585,
      "learning_rate": 4.187143039862631e-06,
      "loss": 0.5484,
      "step": 122780
    },
    {
      "epoch": 50.26,
      "grad_norm": 2.259829044342041,
      "learning_rate": 4.187045070475142e-06,
      "loss": 0.5487,
      "step": 122790
    },
    {
      "epoch": 50.27,
      "grad_norm": 2.4564719200134277,
      "learning_rate": 4.1869470926438025e-06,
      "loss": 0.5434,
      "step": 122800
    },
    {
      "epoch": 50.27,
      "grad_norm": 2.1296803951263428,
      "learning_rate": 4.186849106369063e-06,
      "loss": 0.5483,
      "step": 122810
    },
    {
      "epoch": 50.27,
      "grad_norm": 2.4108240604400635,
      "learning_rate": 4.186751111651372e-06,
      "loss": 0.5452,
      "step": 122820
    },
    {
      "epoch": 50.28,
      "grad_norm": 1.8753997087478638,
      "learning_rate": 4.186653108491177e-06,
      "loss": 0.5407,
      "step": 122830
    },
    {
      "epoch": 50.28,
      "grad_norm": 1.791305661201477,
      "learning_rate": 4.18655509688893e-06,
      "loss": 0.5555,
      "step": 122840
    },
    {
      "epoch": 50.29,
      "grad_norm": 3.092879056930542,
      "learning_rate": 4.186457076845078e-06,
      "loss": 0.5595,
      "step": 122850
    },
    {
      "epoch": 50.29,
      "grad_norm": 2.3611690998077393,
      "learning_rate": 4.186359048360069e-06,
      "loss": 0.5338,
      "step": 122860
    },
    {
      "epoch": 50.29,
      "grad_norm": 2.055462598800659,
      "learning_rate": 4.1862610114343544e-06,
      "loss": 0.5408,
      "step": 122870
    },
    {
      "epoch": 50.3,
      "grad_norm": 2.2234466075897217,
      "learning_rate": 4.186162966068383e-06,
      "loss": 0.5427,
      "step": 122880
    },
    {
      "epoch": 50.3,
      "grad_norm": 1.7447748184204102,
      "learning_rate": 4.186064912262602e-06,
      "loss": 0.5339,
      "step": 122890
    },
    {
      "epoch": 50.31,
      "grad_norm": 1.801596760749817,
      "learning_rate": 4.185966850017462e-06,
      "loss": 0.5469,
      "step": 122900
    },
    {
      "epoch": 50.31,
      "grad_norm": 1.9294347763061523,
      "learning_rate": 4.185868779333411e-06,
      "loss": 0.5542,
      "step": 122910
    },
    {
      "epoch": 50.32,
      "grad_norm": 1.9119529724121094,
      "learning_rate": 4.185770700210901e-06,
      "loss": 0.5236,
      "step": 122920
    },
    {
      "epoch": 50.32,
      "grad_norm": 2.313382625579834,
      "learning_rate": 4.185672612650378e-06,
      "loss": 0.5393,
      "step": 122930
    },
    {
      "epoch": 50.32,
      "grad_norm": 1.8285869359970093,
      "learning_rate": 4.185574516652293e-06,
      "loss": 0.5414,
      "step": 122940
    },
    {
      "epoch": 50.33,
      "grad_norm": 1.9304543733596802,
      "learning_rate": 4.185476412217096e-06,
      "loss": 0.5501,
      "step": 122950
    },
    {
      "epoch": 50.33,
      "grad_norm": 2.5590567588806152,
      "learning_rate": 4.1853782993452345e-06,
      "loss": 0.5344,
      "step": 122960
    },
    {
      "epoch": 50.34,
      "grad_norm": 1.9126009941101074,
      "learning_rate": 4.18528017803716e-06,
      "loss": 0.5557,
      "step": 122970
    },
    {
      "epoch": 50.34,
      "grad_norm": 1.8868595361709595,
      "learning_rate": 4.185182048293319e-06,
      "loss": 0.5559,
      "step": 122980
    },
    {
      "epoch": 50.34,
      "grad_norm": 1.5736433267593384,
      "learning_rate": 4.185083910114164e-06,
      "loss": 0.5399,
      "step": 122990
    },
    {
      "epoch": 50.35,
      "grad_norm": 2.4757914543151855,
      "learning_rate": 4.184985763500145e-06,
      "loss": 0.5173,
      "step": 123000
    },
    {
      "epoch": 50.35,
      "grad_norm": 1.6534202098846436,
      "learning_rate": 4.184887608451708e-06,
      "loss": 0.5411,
      "step": 123010
    },
    {
      "epoch": 50.36,
      "grad_norm": 1.9235060214996338,
      "learning_rate": 4.184789444969306e-06,
      "loss": 0.5441,
      "step": 123020
    },
    {
      "epoch": 50.36,
      "grad_norm": 2.0017597675323486,
      "learning_rate": 4.1846912730533876e-06,
      "loss": 0.5414,
      "step": 123030
    },
    {
      "epoch": 50.36,
      "grad_norm": 2.203268527984619,
      "learning_rate": 4.184593092704401e-06,
      "loss": 0.5518,
      "step": 123040
    },
    {
      "epoch": 50.37,
      "grad_norm": 1.4107744693756104,
      "learning_rate": 4.184494903922798e-06,
      "loss": 0.5244,
      "step": 123050
    },
    {
      "epoch": 50.37,
      "grad_norm": 1.8527402877807617,
      "learning_rate": 4.184396706709028e-06,
      "loss": 0.5414,
      "step": 123060
    },
    {
      "epoch": 50.38,
      "grad_norm": 2.627988815307617,
      "learning_rate": 4.1842985010635405e-06,
      "loss": 0.5434,
      "step": 123070
    },
    {
      "epoch": 50.38,
      "grad_norm": 2.2399885654449463,
      "learning_rate": 4.184200286986785e-06,
      "loss": 0.5502,
      "step": 123080
    },
    {
      "epoch": 50.38,
      "grad_norm": 2.445005178451538,
      "learning_rate": 4.184102064479212e-06,
      "loss": 0.5698,
      "step": 123090
    },
    {
      "epoch": 50.39,
      "grad_norm": 1.9943567514419556,
      "learning_rate": 4.18400383354127e-06,
      "loss": 0.5402,
      "step": 123100
    },
    {
      "epoch": 50.39,
      "grad_norm": 2.0148510932922363,
      "learning_rate": 4.183905594173412e-06,
      "loss": 0.5266,
      "step": 123110
    },
    {
      "epoch": 50.4,
      "grad_norm": 1.7264167070388794,
      "learning_rate": 4.183807346376085e-06,
      "loss": 0.5414,
      "step": 123120
    },
    {
      "epoch": 50.4,
      "grad_norm": 1.904407262802124,
      "learning_rate": 4.183709090149741e-06,
      "loss": 0.5412,
      "step": 123130
    },
    {
      "epoch": 50.41,
      "grad_norm": 2.271641969680786,
      "learning_rate": 4.183610825494829e-06,
      "loss": 0.5073,
      "step": 123140
    },
    {
      "epoch": 50.41,
      "grad_norm": 1.856792688369751,
      "learning_rate": 4.1835125524118e-06,
      "loss": 0.5458,
      "step": 123150
    },
    {
      "epoch": 50.41,
      "grad_norm": 2.019817352294922,
      "learning_rate": 4.183414270901103e-06,
      "loss": 0.525,
      "step": 123160
    },
    {
      "epoch": 50.42,
      "grad_norm": 1.707413911819458,
      "learning_rate": 4.1833159809631885e-06,
      "loss": 0.5628,
      "step": 123170
    },
    {
      "epoch": 50.42,
      "grad_norm": 1.9001840353012085,
      "learning_rate": 4.183217682598509e-06,
      "loss": 0.5367,
      "step": 123180
    },
    {
      "epoch": 50.43,
      "grad_norm": 1.972571849822998,
      "learning_rate": 4.1831193758075116e-06,
      "loss": 0.5197,
      "step": 123190
    },
    {
      "epoch": 50.43,
      "grad_norm": 1.6044917106628418,
      "learning_rate": 4.183021060590648e-06,
      "loss": 0.5302,
      "step": 123200
    },
    {
      "epoch": 50.43,
      "grad_norm": 2.495727777481079,
      "learning_rate": 4.182922736948368e-06,
      "loss": 0.5323,
      "step": 123210
    },
    {
      "epoch": 50.44,
      "grad_norm": 1.9420605897903442,
      "learning_rate": 4.1828244048811235e-06,
      "loss": 0.5549,
      "step": 123220
    },
    {
      "epoch": 50.44,
      "grad_norm": 2.1039650440216064,
      "learning_rate": 4.1827260643893636e-06,
      "loss": 0.5508,
      "step": 123230
    },
    {
      "epoch": 50.45,
      "grad_norm": 1.9076030254364014,
      "learning_rate": 4.182627715473538e-06,
      "loss": 0.5549,
      "step": 123240
    },
    {
      "epoch": 50.45,
      "grad_norm": 2.451333522796631,
      "learning_rate": 4.1825293581341e-06,
      "loss": 0.5071,
      "step": 123250
    },
    {
      "epoch": 50.45,
      "grad_norm": 2.454184055328369,
      "learning_rate": 4.182430992371498e-06,
      "loss": 0.5387,
      "step": 123260
    },
    {
      "epoch": 50.46,
      "grad_norm": 1.8435543775558472,
      "learning_rate": 4.182332618186184e-06,
      "loss": 0.5135,
      "step": 123270
    },
    {
      "epoch": 50.46,
      "grad_norm": 1.9810905456542969,
      "learning_rate": 4.182234235578606e-06,
      "loss": 0.5109,
      "step": 123280
    },
    {
      "epoch": 50.47,
      "grad_norm": 1.71976637840271,
      "learning_rate": 4.182135844549218e-06,
      "loss": 0.5139,
      "step": 123290
    },
    {
      "epoch": 50.47,
      "grad_norm": 2.2335610389709473,
      "learning_rate": 4.182037445098468e-06,
      "loss": 0.5613,
      "step": 123300
    },
    {
      "epoch": 50.47,
      "grad_norm": 2.4050869941711426,
      "learning_rate": 4.181939037226808e-06,
      "loss": 0.5429,
      "step": 123310
    },
    {
      "epoch": 50.48,
      "grad_norm": 1.6827794313430786,
      "learning_rate": 4.18184062093469e-06,
      "loss": 0.5323,
      "step": 123320
    },
    {
      "epoch": 50.48,
      "grad_norm": 1.944287896156311,
      "learning_rate": 4.1817421962225624e-06,
      "loss": 0.5308,
      "step": 123330
    },
    {
      "epoch": 50.49,
      "grad_norm": 1.9542365074157715,
      "learning_rate": 4.181643763090878e-06,
      "loss": 0.544,
      "step": 123340
    },
    {
      "epoch": 50.49,
      "grad_norm": 2.1482510566711426,
      "learning_rate": 4.181545321540085e-06,
      "loss": 0.5215,
      "step": 123350
    },
    {
      "epoch": 50.5,
      "grad_norm": 2.2676267623901367,
      "learning_rate": 4.181446871570638e-06,
      "loss": 0.5374,
      "step": 123360
    },
    {
      "epoch": 50.5,
      "grad_norm": 1.8763630390167236,
      "learning_rate": 4.1813484131829865e-06,
      "loss": 0.5381,
      "step": 123370
    },
    {
      "epoch": 50.5,
      "grad_norm": 1.6749510765075684,
      "learning_rate": 4.18124994637758e-06,
      "loss": 0.5344,
      "step": 123380
    },
    {
      "epoch": 50.51,
      "grad_norm": 1.9313483238220215,
      "learning_rate": 4.181151471154872e-06,
      "loss": 0.5331,
      "step": 123390
    },
    {
      "epoch": 50.51,
      "grad_norm": 1.6821274757385254,
      "learning_rate": 4.1810529875153115e-06,
      "loss": 0.5571,
      "step": 123400
    },
    {
      "epoch": 50.52,
      "grad_norm": 1.9962773323059082,
      "learning_rate": 4.18095449545935e-06,
      "loss": 0.546,
      "step": 123410
    },
    {
      "epoch": 50.52,
      "grad_norm": 2.9477157592773438,
      "learning_rate": 4.180855994987441e-06,
      "loss": 0.5511,
      "step": 123420
    },
    {
      "epoch": 50.52,
      "grad_norm": 2.4082181453704834,
      "learning_rate": 4.180757486100033e-06,
      "loss": 0.5503,
      "step": 123430
    },
    {
      "epoch": 50.53,
      "grad_norm": 2.2577764987945557,
      "learning_rate": 4.180658968797578e-06,
      "loss": 0.5569,
      "step": 123440
    },
    {
      "epoch": 50.53,
      "grad_norm": 2.602318286895752,
      "learning_rate": 4.1805604430805274e-06,
      "loss": 0.5401,
      "step": 123450
    },
    {
      "epoch": 50.54,
      "grad_norm": 2.6850335597991943,
      "learning_rate": 4.180461908949334e-06,
      "loss": 0.5371,
      "step": 123460
    },
    {
      "epoch": 50.54,
      "grad_norm": 2.9315693378448486,
      "learning_rate": 4.1803633664044465e-06,
      "loss": 0.5299,
      "step": 123470
    },
    {
      "epoch": 50.54,
      "grad_norm": 1.6630303859710693,
      "learning_rate": 4.180264815446317e-06,
      "loss": 0.5438,
      "step": 123480
    },
    {
      "epoch": 50.55,
      "grad_norm": 2.3666136264801025,
      "learning_rate": 4.1801662560753984e-06,
      "loss": 0.5269,
      "step": 123490
    },
    {
      "epoch": 50.55,
      "grad_norm": 2.108283758163452,
      "learning_rate": 4.180067688292142e-06,
      "loss": 0.5143,
      "step": 123500
    },
    {
      "epoch": 50.56,
      "grad_norm": 1.8700850009918213,
      "learning_rate": 4.179969112096998e-06,
      "loss": 0.5241,
      "step": 123510
    },
    {
      "epoch": 50.56,
      "grad_norm": 1.7913950681686401,
      "learning_rate": 4.179870527490419e-06,
      "loss": 0.5303,
      "step": 123520
    },
    {
      "epoch": 50.56,
      "grad_norm": 2.0600686073303223,
      "learning_rate": 4.1797719344728555e-06,
      "loss": 0.548,
      "step": 123530
    },
    {
      "epoch": 50.57,
      "grad_norm": 2.495814323425293,
      "learning_rate": 4.17967333304476e-06,
      "loss": 0.5392,
      "step": 123540
    },
    {
      "epoch": 50.57,
      "grad_norm": 2.754781484603882,
      "learning_rate": 4.179574723206585e-06,
      "loss": 0.5231,
      "step": 123550
    },
    {
      "epoch": 50.58,
      "grad_norm": 2.5318171977996826,
      "learning_rate": 4.17947610495878e-06,
      "loss": 0.5394,
      "step": 123560
    },
    {
      "epoch": 50.58,
      "grad_norm": 1.999868392944336,
      "learning_rate": 4.179377478301799e-06,
      "loss": 0.5461,
      "step": 123570
    },
    {
      "epoch": 50.59,
      "grad_norm": 2.7804222106933594,
      "learning_rate": 4.179278843236092e-06,
      "loss": 0.5499,
      "step": 123580
    },
    {
      "epoch": 50.59,
      "grad_norm": 2.290982484817505,
      "learning_rate": 4.179180199762112e-06,
      "loss": 0.5395,
      "step": 123590
    },
    {
      "epoch": 50.59,
      "grad_norm": 2.413179636001587,
      "learning_rate": 4.179081547880311e-06,
      "loss": 0.5417,
      "step": 123600
    },
    {
      "epoch": 50.6,
      "grad_norm": 2.1763243675231934,
      "learning_rate": 4.17898288759114e-06,
      "loss": 0.5332,
      "step": 123610
    },
    {
      "epoch": 50.6,
      "grad_norm": 2.0366287231445312,
      "learning_rate": 4.1788842188950505e-06,
      "loss": 0.5441,
      "step": 123620
    },
    {
      "epoch": 50.61,
      "grad_norm": 3.009139060974121,
      "learning_rate": 4.178785541792496e-06,
      "loss": 0.5565,
      "step": 123630
    },
    {
      "epoch": 50.61,
      "grad_norm": 2.0555856227874756,
      "learning_rate": 4.178686856283928e-06,
      "loss": 0.5261,
      "step": 123640
    },
    {
      "epoch": 50.61,
      "grad_norm": 2.2096071243286133,
      "learning_rate": 4.178588162369799e-06,
      "loss": 0.5388,
      "step": 123650
    },
    {
      "epoch": 50.62,
      "grad_norm": 1.9186511039733887,
      "learning_rate": 4.17848946005056e-06,
      "loss": 0.5488,
      "step": 123660
    },
    {
      "epoch": 50.62,
      "grad_norm": 2.1566030979156494,
      "learning_rate": 4.178390749326663e-06,
      "loss": 0.5433,
      "step": 123670
    },
    {
      "epoch": 50.63,
      "grad_norm": 2.283658742904663,
      "learning_rate": 4.178292030198562e-06,
      "loss": 0.5104,
      "step": 123680
    },
    {
      "epoch": 50.63,
      "grad_norm": 1.7992802858352661,
      "learning_rate": 4.178193302666708e-06,
      "loss": 0.5324,
      "step": 123690
    },
    {
      "epoch": 50.63,
      "grad_norm": 2.0801408290863037,
      "learning_rate": 4.1780945667315526e-06,
      "loss": 0.5395,
      "step": 123700
    },
    {
      "epoch": 50.64,
      "grad_norm": 2.163592576980591,
      "learning_rate": 4.177995822393549e-06,
      "loss": 0.5452,
      "step": 123710
    },
    {
      "epoch": 50.64,
      "grad_norm": 1.5996485948562622,
      "learning_rate": 4.177897069653149e-06,
      "loss": 0.5638,
      "step": 123720
    },
    {
      "epoch": 50.65,
      "grad_norm": 1.6055450439453125,
      "learning_rate": 4.177798308510806e-06,
      "loss": 0.5252,
      "step": 123730
    },
    {
      "epoch": 50.65,
      "grad_norm": 3.222384214401245,
      "learning_rate": 4.1776995389669724e-06,
      "loss": 0.5552,
      "step": 123740
    },
    {
      "epoch": 50.65,
      "grad_norm": 1.7135484218597412,
      "learning_rate": 4.177600761022098e-06,
      "loss": 0.5201,
      "step": 123750
    },
    {
      "epoch": 50.66,
      "grad_norm": 1.8039577007293701,
      "learning_rate": 4.177501974676639e-06,
      "loss": 0.5384,
      "step": 123760
    },
    {
      "epoch": 50.66,
      "grad_norm": 2.2558531761169434,
      "learning_rate": 4.177403179931046e-06,
      "loss": 0.5413,
      "step": 123770
    },
    {
      "epoch": 50.67,
      "grad_norm": 1.835016131401062,
      "learning_rate": 4.177304376785771e-06,
      "loss": 0.5451,
      "step": 123780
    },
    {
      "epoch": 50.67,
      "grad_norm": 2.3176796436309814,
      "learning_rate": 4.177205565241267e-06,
      "loss": 0.5437,
      "step": 123790
    },
    {
      "epoch": 50.68,
      "grad_norm": 1.9961597919464111,
      "learning_rate": 4.177106745297989e-06,
      "loss": 0.5391,
      "step": 123800
    },
    {
      "epoch": 50.68,
      "grad_norm": 2.667315721511841,
      "learning_rate": 4.177007916956386e-06,
      "loss": 0.5391,
      "step": 123810
    },
    {
      "epoch": 50.68,
      "grad_norm": 1.8792697191238403,
      "learning_rate": 4.176909080216913e-06,
      "loss": 0.5466,
      "step": 123820
    },
    {
      "epoch": 50.69,
      "grad_norm": 1.8366857767105103,
      "learning_rate": 4.176810235080021e-06,
      "loss": 0.5403,
      "step": 123830
    },
    {
      "epoch": 50.69,
      "grad_norm": 1.751265525817871,
      "learning_rate": 4.176711381546166e-06,
      "loss": 0.5339,
      "step": 123840
    },
    {
      "epoch": 50.7,
      "grad_norm": 2.160780191421509,
      "learning_rate": 4.176612519615797e-06,
      "loss": 0.5294,
      "step": 123850
    },
    {
      "epoch": 50.7,
      "grad_norm": 2.6499714851379395,
      "learning_rate": 4.17651364928937e-06,
      "loss": 0.5429,
      "step": 123860
    },
    {
      "epoch": 50.7,
      "grad_norm": 2.392695665359497,
      "learning_rate": 4.176414770567335e-06,
      "loss": 0.5365,
      "step": 123870
    },
    {
      "epoch": 50.71,
      "grad_norm": 2.150050640106201,
      "learning_rate": 4.1763158834501475e-06,
      "loss": 0.5186,
      "step": 123880
    },
    {
      "epoch": 50.71,
      "grad_norm": 1.7356945276260376,
      "learning_rate": 4.17621698793826e-06,
      "loss": 0.5391,
      "step": 123890
    },
    {
      "epoch": 50.72,
      "grad_norm": 1.583042025566101,
      "learning_rate": 4.176118084032124e-06,
      "loss": 0.5418,
      "step": 123900
    },
    {
      "epoch": 50.72,
      "grad_norm": 2.167564630508423,
      "learning_rate": 4.1760191717321944e-06,
      "loss": 0.5465,
      "step": 123910
    },
    {
      "epoch": 50.72,
      "grad_norm": 1.862276315689087,
      "learning_rate": 4.175920251038923e-06,
      "loss": 0.5322,
      "step": 123920
    },
    {
      "epoch": 50.73,
      "grad_norm": 2.397933006286621,
      "learning_rate": 4.1758213219527635e-06,
      "loss": 0.5337,
      "step": 123930
    },
    {
      "epoch": 50.73,
      "grad_norm": 2.274104595184326,
      "learning_rate": 4.175722384474169e-06,
      "loss": 0.5514,
      "step": 123940
    },
    {
      "epoch": 50.74,
      "grad_norm": 2.122342348098755,
      "learning_rate": 4.175623438603593e-06,
      "loss": 0.5419,
      "step": 123950
    },
    {
      "epoch": 50.74,
      "grad_norm": 2.443134069442749,
      "learning_rate": 4.175524484341488e-06,
      "loss": 0.5391,
      "step": 123960
    },
    {
      "epoch": 50.74,
      "grad_norm": 1.759237289428711,
      "learning_rate": 4.175425521688308e-06,
      "loss": 0.5333,
      "step": 123970
    },
    {
      "epoch": 50.75,
      "grad_norm": 1.901945948600769,
      "learning_rate": 4.175326550644506e-06,
      "loss": 0.5318,
      "step": 123980
    },
    {
      "epoch": 50.75,
      "grad_norm": 1.8923603296279907,
      "learning_rate": 4.1752275712105356e-06,
      "loss": 0.5399,
      "step": 123990
    },
    {
      "epoch": 50.76,
      "grad_norm": 1.9389545917510986,
      "learning_rate": 4.17512858338685e-06,
      "loss": 0.5332,
      "step": 124000
    },
    {
      "epoch": 50.76,
      "grad_norm": 1.9865151643753052,
      "learning_rate": 4.175029587173903e-06,
      "loss": 0.5483,
      "step": 124010
    },
    {
      "epoch": 50.77,
      "grad_norm": 2.199235439300537,
      "learning_rate": 4.174930582572148e-06,
      "loss": 0.536,
      "step": 124020
    },
    {
      "epoch": 50.77,
      "grad_norm": 2.0512304306030273,
      "learning_rate": 4.174831569582038e-06,
      "loss": 0.5439,
      "step": 124030
    },
    {
      "epoch": 50.77,
      "grad_norm": 1.884913682937622,
      "learning_rate": 4.1747325482040265e-06,
      "loss": 0.5529,
      "step": 124040
    },
    {
      "epoch": 50.78,
      "grad_norm": 1.9285293817520142,
      "learning_rate": 4.174633518438568e-06,
      "loss": 0.5286,
      "step": 124050
    },
    {
      "epoch": 50.78,
      "grad_norm": 2.4409878253936768,
      "learning_rate": 4.174534480286116e-06,
      "loss": 0.5509,
      "step": 124060
    },
    {
      "epoch": 50.79,
      "grad_norm": 1.5030794143676758,
      "learning_rate": 4.174435433747122e-06,
      "loss": 0.5386,
      "step": 124070
    },
    {
      "epoch": 50.79,
      "grad_norm": 1.9382922649383545,
      "learning_rate": 4.174336378822043e-06,
      "loss": 0.5491,
      "step": 124080
    },
    {
      "epoch": 50.79,
      "grad_norm": 2.376893997192383,
      "learning_rate": 4.1742373155113315e-06,
      "loss": 0.5708,
      "step": 124090
    },
    {
      "epoch": 50.8,
      "grad_norm": 1.886469841003418,
      "learning_rate": 4.1741382438154415e-06,
      "loss": 0.5381,
      "step": 124100
    },
    {
      "epoch": 50.8,
      "grad_norm": 1.640675663948059,
      "learning_rate": 4.174039163734826e-06,
      "loss": 0.5382,
      "step": 124110
    },
    {
      "epoch": 50.81,
      "grad_norm": 1.725498914718628,
      "learning_rate": 4.173940075269939e-06,
      "loss": 0.5337,
      "step": 124120
    },
    {
      "epoch": 50.81,
      "grad_norm": 2.2842233180999756,
      "learning_rate": 4.1738409784212345e-06,
      "loss": 0.5427,
      "step": 124130
    },
    {
      "epoch": 50.81,
      "grad_norm": 2.1763436794281006,
      "learning_rate": 4.173741873189166e-06,
      "loss": 0.54,
      "step": 124140
    },
    {
      "epoch": 50.82,
      "grad_norm": 2.5556485652923584,
      "learning_rate": 4.17364275957419e-06,
      "loss": 0.5352,
      "step": 124150
    },
    {
      "epoch": 50.82,
      "grad_norm": 2.560326337814331,
      "learning_rate": 4.173543637576758e-06,
      "loss": 0.5326,
      "step": 124160
    },
    {
      "epoch": 50.83,
      "grad_norm": 1.8086234331130981,
      "learning_rate": 4.173444507197324e-06,
      "loss": 0.5477,
      "step": 124170
    },
    {
      "epoch": 50.83,
      "grad_norm": 2.1989119052886963,
      "learning_rate": 4.173345368436343e-06,
      "loss": 0.532,
      "step": 124180
    },
    {
      "epoch": 50.84,
      "grad_norm": 2.4855809211730957,
      "learning_rate": 4.173246221294269e-06,
      "loss": 0.554,
      "step": 124190
    },
    {
      "epoch": 50.84,
      "grad_norm": 3.412393093109131,
      "learning_rate": 4.173147065771557e-06,
      "loss": 0.5464,
      "step": 124200
    },
    {
      "epoch": 50.84,
      "grad_norm": 1.358606219291687,
      "learning_rate": 4.173047901868659e-06,
      "loss": 0.5721,
      "step": 124210
    },
    {
      "epoch": 50.85,
      "grad_norm": 2.2150235176086426,
      "learning_rate": 4.172948729586032e-06,
      "loss": 0.5395,
      "step": 124220
    },
    {
      "epoch": 50.85,
      "grad_norm": 2.0421130657196045,
      "learning_rate": 4.172849548924128e-06,
      "loss": 0.5515,
      "step": 124230
    },
    {
      "epoch": 50.86,
      "grad_norm": 2.1243276596069336,
      "learning_rate": 4.172750359883402e-06,
      "loss": 0.5652,
      "step": 124240
    },
    {
      "epoch": 50.86,
      "grad_norm": 2.5890040397644043,
      "learning_rate": 4.172651162464309e-06,
      "loss": 0.5391,
      "step": 124250
    },
    {
      "epoch": 50.86,
      "grad_norm": 2.774319887161255,
      "learning_rate": 4.172551956667304e-06,
      "loss": 0.5166,
      "step": 124260
    },
    {
      "epoch": 50.87,
      "grad_norm": 2.569335460662842,
      "learning_rate": 4.1724527424928395e-06,
      "loss": 0.5418,
      "step": 124270
    },
    {
      "epoch": 50.87,
      "grad_norm": 1.7879581451416016,
      "learning_rate": 4.172353519941371e-06,
      "loss": 0.5353,
      "step": 124280
    },
    {
      "epoch": 50.88,
      "grad_norm": 2.2071402072906494,
      "learning_rate": 4.172254289013352e-06,
      "loss": 0.5415,
      "step": 124290
    },
    {
      "epoch": 50.88,
      "grad_norm": 1.7273240089416504,
      "learning_rate": 4.17215504970924e-06,
      "loss": 0.5456,
      "step": 124300
    },
    {
      "epoch": 50.88,
      "grad_norm": 2.063244342803955,
      "learning_rate": 4.172055802029487e-06,
      "loss": 0.553,
      "step": 124310
    },
    {
      "epoch": 50.89,
      "grad_norm": 2.1409928798675537,
      "learning_rate": 4.171956545974547e-06,
      "loss": 0.5504,
      "step": 124320
    },
    {
      "epoch": 50.89,
      "grad_norm": 2.530259609222412,
      "learning_rate": 4.171857281544877e-06,
      "loss": 0.5408,
      "step": 124330
    },
    {
      "epoch": 50.9,
      "grad_norm": 1.7912499904632568,
      "learning_rate": 4.171758008740931e-06,
      "loss": 0.5263,
      "step": 124340
    },
    {
      "epoch": 50.9,
      "grad_norm": 2.2306346893310547,
      "learning_rate": 4.171658727563163e-06,
      "loss": 0.5142,
      "step": 124350
    },
    {
      "epoch": 50.9,
      "grad_norm": 1.8884259462356567,
      "learning_rate": 4.1715594380120275e-06,
      "loss": 0.5354,
      "step": 124360
    },
    {
      "epoch": 50.91,
      "grad_norm": 1.804941177368164,
      "learning_rate": 4.171460140087981e-06,
      "loss": 0.5157,
      "step": 124370
    },
    {
      "epoch": 50.91,
      "grad_norm": 2.113887071609497,
      "learning_rate": 4.171360833791477e-06,
      "loss": 0.5441,
      "step": 124380
    },
    {
      "epoch": 50.92,
      "grad_norm": 1.955442190170288,
      "learning_rate": 4.1712615191229714e-06,
      "loss": 0.5403,
      "step": 124390
    },
    {
      "epoch": 50.92,
      "grad_norm": 1.720138430595398,
      "learning_rate": 4.171162196082917e-06,
      "loss": 0.5734,
      "step": 124400
    },
    {
      "epoch": 50.93,
      "grad_norm": 1.9255092144012451,
      "learning_rate": 4.171062864671772e-06,
      "loss": 0.5289,
      "step": 124410
    },
    {
      "epoch": 50.93,
      "grad_norm": 1.93245267868042,
      "learning_rate": 4.170963524889989e-06,
      "loss": 0.5289,
      "step": 124420
    },
    {
      "epoch": 50.93,
      "grad_norm": 2.152470827102661,
      "learning_rate": 4.170864176738024e-06,
      "loss": 0.5375,
      "step": 124430
    },
    {
      "epoch": 50.94,
      "grad_norm": 2.5490357875823975,
      "learning_rate": 4.170764820216332e-06,
      "loss": 0.547,
      "step": 124440
    },
    {
      "epoch": 50.94,
      "grad_norm": 1.5512580871582031,
      "learning_rate": 4.1706654553253685e-06,
      "loss": 0.5276,
      "step": 124450
    },
    {
      "epoch": 50.95,
      "grad_norm": 2.1819419860839844,
      "learning_rate": 4.170566082065587e-06,
      "loss": 0.5343,
      "step": 124460
    },
    {
      "epoch": 50.95,
      "grad_norm": 1.683740258216858,
      "learning_rate": 4.170466700437445e-06,
      "loss": 0.5571,
      "step": 124470
    },
    {
      "epoch": 50.95,
      "grad_norm": 2.150235176086426,
      "learning_rate": 4.170367310441397e-06,
      "loss": 0.528,
      "step": 124480
    },
    {
      "epoch": 50.96,
      "grad_norm": 1.9714065790176392,
      "learning_rate": 4.170267912077897e-06,
      "loss": 0.5529,
      "step": 124490
    },
    {
      "epoch": 50.96,
      "grad_norm": 1.8793692588806152,
      "learning_rate": 4.170168505347401e-06,
      "loss": 0.5402,
      "step": 124500
    },
    {
      "epoch": 50.97,
      "grad_norm": 1.954455852508545,
      "learning_rate": 4.1700690902503656e-06,
      "loss": 0.5249,
      "step": 124510
    },
    {
      "epoch": 50.97,
      "grad_norm": 2.36507248878479,
      "learning_rate": 4.169969666787245e-06,
      "loss": 0.5232,
      "step": 124520
    },
    {
      "epoch": 50.97,
      "grad_norm": 1.6833844184875488,
      "learning_rate": 4.169870234958495e-06,
      "loss": 0.5447,
      "step": 124530
    },
    {
      "epoch": 50.98,
      "grad_norm": 1.9239927530288696,
      "learning_rate": 4.169770794764571e-06,
      "loss": 0.5255,
      "step": 124540
    },
    {
      "epoch": 50.98,
      "grad_norm": 2.165349245071411,
      "learning_rate": 4.169671346205928e-06,
      "loss": 0.5315,
      "step": 124550
    },
    {
      "epoch": 50.99,
      "grad_norm": 1.7437775135040283,
      "learning_rate": 4.169571889283022e-06,
      "loss": 0.5456,
      "step": 124560
    },
    {
      "epoch": 50.99,
      "grad_norm": 1.4674148559570312,
      "learning_rate": 4.169472423996311e-06,
      "loss": 0.5259,
      "step": 124570
    },
    {
      "epoch": 50.99,
      "grad_norm": 1.8638880252838135,
      "learning_rate": 4.169372950346246e-06,
      "loss": 0.5287,
      "step": 124580
    },
    {
      "epoch": 51.0,
      "grad_norm": 2.5492403507232666,
      "learning_rate": 4.169273468333285e-06,
      "loss": 0.5386,
      "step": 124590
    },
    {
      "epoch": 51.0,
      "eval_loss": 0.538404107093811,
      "eval_runtime": 51.9501,
      "eval_samples_per_second": 66.391,
      "eval_steps_per_second": 8.316,
      "step": 124593
    },
    {
      "epoch": 51.0,
      "grad_norm": 1.6457054615020752,
      "learning_rate": 4.169173977957885e-06,
      "loss": 0.5315,
      "step": 124600
    },
    {
      "epoch": 51.01,
      "grad_norm": 2.092977285385132,
      "learning_rate": 4.1690744792204995e-06,
      "loss": 0.5364,
      "step": 124610
    },
    {
      "epoch": 51.01,
      "grad_norm": 1.8963102102279663,
      "learning_rate": 4.168974972121585e-06,
      "loss": 0.5239,
      "step": 124620
    },
    {
      "epoch": 51.02,
      "grad_norm": 2.1369993686676025,
      "learning_rate": 4.168875456661599e-06,
      "loss": 0.5445,
      "step": 124630
    },
    {
      "epoch": 51.02,
      "grad_norm": 2.3140993118286133,
      "learning_rate": 4.168775932840995e-06,
      "loss": 0.5463,
      "step": 124640
    },
    {
      "epoch": 51.02,
      "grad_norm": 2.097980260848999,
      "learning_rate": 4.1686764006602295e-06,
      "loss": 0.5435,
      "step": 124650
    },
    {
      "epoch": 51.03,
      "grad_norm": 2.917383909225464,
      "learning_rate": 4.168576860119759e-06,
      "loss": 0.554,
      "step": 124660
    },
    {
      "epoch": 51.03,
      "grad_norm": 1.6626911163330078,
      "learning_rate": 4.16847731122004e-06,
      "loss": 0.5395,
      "step": 124670
    },
    {
      "epoch": 51.04,
      "grad_norm": 1.8870432376861572,
      "learning_rate": 4.168377753961527e-06,
      "loss": 0.5423,
      "step": 124680
    },
    {
      "epoch": 51.04,
      "grad_norm": 3.014817476272583,
      "learning_rate": 4.168278188344677e-06,
      "loss": 0.5365,
      "step": 124690
    },
    {
      "epoch": 51.04,
      "grad_norm": 1.7405058145523071,
      "learning_rate": 4.168178614369946e-06,
      "loss": 0.5404,
      "step": 124700
    },
    {
      "epoch": 51.05,
      "grad_norm": 1.9874552488327026,
      "learning_rate": 4.16807903203779e-06,
      "loss": 0.5601,
      "step": 124710
    },
    {
      "epoch": 51.05,
      "grad_norm": 3.421097755432129,
      "learning_rate": 4.167979441348665e-06,
      "loss": 0.5561,
      "step": 124720
    },
    {
      "epoch": 51.06,
      "grad_norm": 1.9852291345596313,
      "learning_rate": 4.167879842303028e-06,
      "loss": 0.5397,
      "step": 124730
    },
    {
      "epoch": 51.06,
      "grad_norm": 1.6928669214248657,
      "learning_rate": 4.167780234901335e-06,
      "loss": 0.5073,
      "step": 124740
    },
    {
      "epoch": 51.06,
      "grad_norm": 2.064563751220703,
      "learning_rate": 4.1676806191440405e-06,
      "loss": 0.5307,
      "step": 124750
    },
    {
      "epoch": 51.07,
      "grad_norm": 1.9978883266448975,
      "learning_rate": 4.167580995031603e-06,
      "loss": 0.5196,
      "step": 124760
    },
    {
      "epoch": 51.07,
      "grad_norm": 2.9514589309692383,
      "learning_rate": 4.167481362564479e-06,
      "loss": 0.552,
      "step": 124770
    },
    {
      "epoch": 51.08,
      "grad_norm": 1.9825453758239746,
      "learning_rate": 4.167381721743123e-06,
      "loss": 0.5284,
      "step": 124780
    },
    {
      "epoch": 51.08,
      "grad_norm": 1.9267765283584595,
      "learning_rate": 4.167282072567993e-06,
      "loss": 0.5631,
      "step": 124790
    },
    {
      "epoch": 51.08,
      "grad_norm": 1.7151671648025513,
      "learning_rate": 4.1671824150395445e-06,
      "loss": 0.5467,
      "step": 124800
    },
    {
      "epoch": 51.09,
      "grad_norm": 1.7384345531463623,
      "learning_rate": 4.167082749158235e-06,
      "loss": 0.5508,
      "step": 124810
    },
    {
      "epoch": 51.09,
      "grad_norm": 1.5280425548553467,
      "learning_rate": 4.166983074924521e-06,
      "loss": 0.5155,
      "step": 124820
    },
    {
      "epoch": 51.1,
      "grad_norm": 2.15317964553833,
      "learning_rate": 4.166883392338858e-06,
      "loss": 0.5487,
      "step": 124830
    },
    {
      "epoch": 51.1,
      "grad_norm": 2.2953076362609863,
      "learning_rate": 4.166783701401703e-06,
      "loss": 0.5254,
      "step": 124840
    },
    {
      "epoch": 51.11,
      "grad_norm": 2.0778534412384033,
      "learning_rate": 4.166684002113514e-06,
      "loss": 0.5409,
      "step": 124850
    },
    {
      "epoch": 51.11,
      "grad_norm": 1.9445602893829346,
      "learning_rate": 4.166584294474746e-06,
      "loss": 0.5124,
      "step": 124860
    },
    {
      "epoch": 51.11,
      "grad_norm": 2.645260810852051,
      "learning_rate": 4.166484578485857e-06,
      "loss": 0.552,
      "step": 124870
    },
    {
      "epoch": 51.12,
      "grad_norm": 2.0836024284362793,
      "learning_rate": 4.1663848541473015e-06,
      "loss": 0.5456,
      "step": 124880
    },
    {
      "epoch": 51.12,
      "grad_norm": 2.139491081237793,
      "learning_rate": 4.166285121459539e-06,
      "loss": 0.5468,
      "step": 124890
    },
    {
      "epoch": 51.13,
      "grad_norm": 1.998351812362671,
      "learning_rate": 4.166185380423026e-06,
      "loss": 0.5417,
      "step": 124900
    },
    {
      "epoch": 51.13,
      "grad_norm": 1.6028343439102173,
      "learning_rate": 4.166085631038218e-06,
      "loss": 0.5392,
      "step": 124910
    },
    {
      "epoch": 51.13,
      "grad_norm": 1.8166627883911133,
      "learning_rate": 4.165985873305573e-06,
      "loss": 0.5421,
      "step": 124920
    },
    {
      "epoch": 51.14,
      "grad_norm": 2.933948516845703,
      "learning_rate": 4.165886107225548e-06,
      "loss": 0.5386,
      "step": 124930
    },
    {
      "epoch": 51.14,
      "grad_norm": 1.7322036027908325,
      "learning_rate": 4.165786332798599e-06,
      "loss": 0.5213,
      "step": 124940
    },
    {
      "epoch": 51.15,
      "grad_norm": 2.6145334243774414,
      "learning_rate": 4.1656865500251835e-06,
      "loss": 0.5378,
      "step": 124950
    },
    {
      "epoch": 51.15,
      "grad_norm": 2.7765121459960938,
      "learning_rate": 4.1655867589057594e-06,
      "loss": 0.5334,
      "step": 124960
    },
    {
      "epoch": 51.15,
      "grad_norm": 2.0812134742736816,
      "learning_rate": 4.165486959440783e-06,
      "loss": 0.5341,
      "step": 124970
    },
    {
      "epoch": 51.16,
      "grad_norm": 2.845273733139038,
      "learning_rate": 4.165387151630712e-06,
      "loss": 0.5306,
      "step": 124980
    },
    {
      "epoch": 51.16,
      "grad_norm": 1.9524744749069214,
      "learning_rate": 4.165287335476004e-06,
      "loss": 0.553,
      "step": 124990
    },
    {
      "epoch": 51.17,
      "grad_norm": 2.5043418407440186,
      "learning_rate": 4.1651875109771145e-06,
      "loss": 0.5309,
      "step": 125000
    },
    {
      "epoch": 51.17,
      "grad_norm": 2.1291935443878174,
      "learning_rate": 4.165087678134502e-06,
      "loss": 0.5558,
      "step": 125010
    },
    {
      "epoch": 51.17,
      "grad_norm": 2.670112371444702,
      "learning_rate": 4.164987836948624e-06,
      "loss": 0.5373,
      "step": 125020
    },
    {
      "epoch": 51.18,
      "grad_norm": 2.2229952812194824,
      "learning_rate": 4.164887987419937e-06,
      "loss": 0.5234,
      "step": 125030
    },
    {
      "epoch": 51.18,
      "grad_norm": 1.5131123065948486,
      "learning_rate": 4.1647881295489e-06,
      "loss": 0.5223,
      "step": 125040
    },
    {
      "epoch": 51.19,
      "grad_norm": 3.5618338584899902,
      "learning_rate": 4.164688263335968e-06,
      "loss": 0.5388,
      "step": 125050
    },
    {
      "epoch": 51.19,
      "grad_norm": 2.6732046604156494,
      "learning_rate": 4.164588388781601e-06,
      "loss": 0.5393,
      "step": 125060
    },
    {
      "epoch": 51.2,
      "grad_norm": 2.238999128341675,
      "learning_rate": 4.164488505886255e-06,
      "loss": 0.5388,
      "step": 125070
    },
    {
      "epoch": 51.2,
      "grad_norm": 2.6867475509643555,
      "learning_rate": 4.164388614650387e-06,
      "loss": 0.54,
      "step": 125080
    },
    {
      "epoch": 51.2,
      "grad_norm": 2.5677552223205566,
      "learning_rate": 4.164288715074456e-06,
      "loss": 0.5381,
      "step": 125090
    },
    {
      "epoch": 51.21,
      "grad_norm": 1.6154766082763672,
      "learning_rate": 4.164188807158919e-06,
      "loss": 0.5467,
      "step": 125100
    },
    {
      "epoch": 51.21,
      "grad_norm": 1.761570692062378,
      "learning_rate": 4.1640888909042346e-06,
      "loss": 0.5535,
      "step": 125110
    },
    {
      "epoch": 51.22,
      "grad_norm": 2.0690793991088867,
      "learning_rate": 4.163988966310859e-06,
      "loss": 0.5397,
      "step": 125120
    },
    {
      "epoch": 51.22,
      "grad_norm": 1.6333844661712646,
      "learning_rate": 4.16388903337925e-06,
      "loss": 0.5576,
      "step": 125130
    },
    {
      "epoch": 51.22,
      "grad_norm": 2.2902474403381348,
      "learning_rate": 4.163789092109867e-06,
      "loss": 0.5465,
      "step": 125140
    },
    {
      "epoch": 51.23,
      "grad_norm": 1.412739634513855,
      "learning_rate": 4.163689142503167e-06,
      "loss": 0.5348,
      "step": 125150
    },
    {
      "epoch": 51.23,
      "grad_norm": 1.8311610221862793,
      "learning_rate": 4.163589184559607e-06,
      "loss": 0.5435,
      "step": 125160
    },
    {
      "epoch": 51.24,
      "grad_norm": 2.0663506984710693,
      "learning_rate": 4.163489218279645e-06,
      "loss": 0.5252,
      "step": 125170
    },
    {
      "epoch": 51.24,
      "grad_norm": 2.228217124938965,
      "learning_rate": 4.16338924366374e-06,
      "loss": 0.5569,
      "step": 125180
    },
    {
      "epoch": 51.24,
      "grad_norm": 2.4960410594940186,
      "learning_rate": 4.16328926071235e-06,
      "loss": 0.5401,
      "step": 125190
    },
    {
      "epoch": 51.25,
      "grad_norm": 1.934897541999817,
      "learning_rate": 4.1631892694259315e-06,
      "loss": 0.5465,
      "step": 125200
    },
    {
      "epoch": 51.25,
      "grad_norm": 2.0225839614868164,
      "learning_rate": 4.163089269804943e-06,
      "loss": 0.532,
      "step": 125210
    },
    {
      "epoch": 51.26,
      "grad_norm": 2.187143087387085,
      "learning_rate": 4.162989261849844e-06,
      "loss": 0.5286,
      "step": 125220
    },
    {
      "epoch": 51.26,
      "grad_norm": 2.913637161254883,
      "learning_rate": 4.162889245561092e-06,
      "loss": 0.5407,
      "step": 125230
    },
    {
      "epoch": 51.26,
      "grad_norm": 2.3560783863067627,
      "learning_rate": 4.162789220939145e-06,
      "loss": 0.5192,
      "step": 125240
    },
    {
      "epoch": 51.27,
      "grad_norm": 1.8565216064453125,
      "learning_rate": 4.16268918798446e-06,
      "loss": 0.5542,
      "step": 125250
    },
    {
      "epoch": 51.27,
      "grad_norm": 2.21091890335083,
      "learning_rate": 4.162589146697497e-06,
      "loss": 0.5091,
      "step": 125260
    },
    {
      "epoch": 51.28,
      "grad_norm": 1.7024950981140137,
      "learning_rate": 4.162489097078713e-06,
      "loss": 0.5257,
      "step": 125270
    },
    {
      "epoch": 51.28,
      "grad_norm": 1.604887843132019,
      "learning_rate": 4.162389039128568e-06,
      "loss": 0.545,
      "step": 125280
    },
    {
      "epoch": 51.29,
      "grad_norm": 2.11365008354187,
      "learning_rate": 4.162288972847518e-06,
      "loss": 0.5411,
      "step": 125290
    },
    {
      "epoch": 51.29,
      "grad_norm": 2.3198108673095703,
      "learning_rate": 4.162188898236023e-06,
      "loss": 0.551,
      "step": 125300
    },
    {
      "epoch": 51.29,
      "grad_norm": 2.236104726791382,
      "learning_rate": 4.162088815294541e-06,
      "loss": 0.529,
      "step": 125310
    },
    {
      "epoch": 51.3,
      "grad_norm": 1.8594518899917603,
      "learning_rate": 4.16198872402353e-06,
      "loss": 0.5364,
      "step": 125320
    },
    {
      "epoch": 51.3,
      "grad_norm": 1.9494001865386963,
      "learning_rate": 4.16188862442345e-06,
      "loss": 0.5226,
      "step": 125330
    },
    {
      "epoch": 51.31,
      "grad_norm": 2.050060272216797,
      "learning_rate": 4.161788516494758e-06,
      "loss": 0.5736,
      "step": 125340
    },
    {
      "epoch": 51.31,
      "grad_norm": 1.8835455179214478,
      "learning_rate": 4.1616884002379135e-06,
      "loss": 0.5423,
      "step": 125350
    },
    {
      "epoch": 51.31,
      "grad_norm": 2.4571073055267334,
      "learning_rate": 4.161588275653374e-06,
      "loss": 0.5263,
      "step": 125360
    },
    {
      "epoch": 51.32,
      "grad_norm": 2.3162341117858887,
      "learning_rate": 4.1614881427416e-06,
      "loss": 0.5254,
      "step": 125370
    },
    {
      "epoch": 51.32,
      "grad_norm": 1.758333444595337,
      "learning_rate": 4.1613880015030474e-06,
      "loss": 0.5252,
      "step": 125380
    },
    {
      "epoch": 51.33,
      "grad_norm": 1.7249934673309326,
      "learning_rate": 4.161287851938178e-06,
      "loss": 0.548,
      "step": 125390
    },
    {
      "epoch": 51.33,
      "grad_norm": 1.8141965866088867,
      "learning_rate": 4.161187694047449e-06,
      "loss": 0.5227,
      "step": 125400
    },
    {
      "epoch": 51.33,
      "grad_norm": 2.441368818283081,
      "learning_rate": 4.161087527831319e-06,
      "loss": 0.5484,
      "step": 125410
    },
    {
      "epoch": 51.34,
      "grad_norm": 2.2067689895629883,
      "learning_rate": 4.1609873532902475e-06,
      "loss": 0.5366,
      "step": 125420
    },
    {
      "epoch": 51.34,
      "grad_norm": 2.323026657104492,
      "learning_rate": 4.160887170424693e-06,
      "loss": 0.5436,
      "step": 125430
    },
    {
      "epoch": 51.35,
      "grad_norm": 2.2761659622192383,
      "learning_rate": 4.160786979235115e-06,
      "loss": 0.5378,
      "step": 125440
    },
    {
      "epoch": 51.35,
      "grad_norm": 2.1894500255584717,
      "learning_rate": 4.1606867797219705e-06,
      "loss": 0.545,
      "step": 125450
    },
    {
      "epoch": 51.35,
      "grad_norm": 1.9269508123397827,
      "learning_rate": 4.160586571885721e-06,
      "loss": 0.5478,
      "step": 125460
    },
    {
      "epoch": 51.36,
      "grad_norm": 1.8030608892440796,
      "learning_rate": 4.1604863557268245e-06,
      "loss": 0.5484,
      "step": 125470
    },
    {
      "epoch": 51.36,
      "grad_norm": 2.516540288925171,
      "learning_rate": 4.160386131245741e-06,
      "loss": 0.5155,
      "step": 125480
    },
    {
      "epoch": 51.37,
      "grad_norm": 1.8879491090774536,
      "learning_rate": 4.160285898442928e-06,
      "loss": 0.5469,
      "step": 125490
    },
    {
      "epoch": 51.37,
      "grad_norm": 1.739532470703125,
      "learning_rate": 4.160185657318845e-06,
      "loss": 0.5374,
      "step": 125500
    },
    {
      "epoch": 51.38,
      "grad_norm": 2.0784027576446533,
      "learning_rate": 4.160085407873952e-06,
      "loss": 0.5543,
      "step": 125510
    },
    {
      "epoch": 51.38,
      "grad_norm": 1.8291878700256348,
      "learning_rate": 4.159985150108708e-06,
      "loss": 0.529,
      "step": 125520
    },
    {
      "epoch": 51.38,
      "grad_norm": 2.430835247039795,
      "learning_rate": 4.159884884023571e-06,
      "loss": 0.5403,
      "step": 125530
    },
    {
      "epoch": 51.39,
      "grad_norm": 1.8131996393203735,
      "learning_rate": 4.159784609619002e-06,
      "loss": 0.541,
      "step": 125540
    },
    {
      "epoch": 51.39,
      "grad_norm": 2.223829746246338,
      "learning_rate": 4.15968432689546e-06,
      "loss": 0.5616,
      "step": 125550
    },
    {
      "epoch": 51.4,
      "grad_norm": 1.6638460159301758,
      "learning_rate": 4.159584035853404e-06,
      "loss": 0.5251,
      "step": 125560
    },
    {
      "epoch": 51.4,
      "grad_norm": 2.219618082046509,
      "learning_rate": 4.159483736493293e-06,
      "loss": 0.5312,
      "step": 125570
    },
    {
      "epoch": 51.4,
      "grad_norm": 2.1234428882598877,
      "learning_rate": 4.159383428815588e-06,
      "loss": 0.5482,
      "step": 125580
    },
    {
      "epoch": 51.41,
      "grad_norm": 1.9770973920822144,
      "learning_rate": 4.159283112820747e-06,
      "loss": 0.5413,
      "step": 125590
    },
    {
      "epoch": 51.41,
      "grad_norm": 2.7266952991485596,
      "learning_rate": 4.159182788509229e-06,
      "loss": 0.5415,
      "step": 125600
    },
    {
      "epoch": 51.42,
      "grad_norm": 1.7298322916030884,
      "learning_rate": 4.1590824558814954e-06,
      "loss": 0.5572,
      "step": 125610
    },
    {
      "epoch": 51.42,
      "grad_norm": 1.9682892560958862,
      "learning_rate": 4.158982114938005e-06,
      "loss": 0.5377,
      "step": 125620
    },
    {
      "epoch": 51.42,
      "grad_norm": 2.2011916637420654,
      "learning_rate": 4.158881765679217e-06,
      "loss": 0.5466,
      "step": 125630
    },
    {
      "epoch": 51.43,
      "grad_norm": 2.2077295780181885,
      "learning_rate": 4.158781408105592e-06,
      "loss": 0.5175,
      "step": 125640
    },
    {
      "epoch": 51.43,
      "grad_norm": 1.5463745594024658,
      "learning_rate": 4.1586810422175885e-06,
      "loss": 0.5314,
      "step": 125650
    },
    {
      "epoch": 51.44,
      "grad_norm": 1.8394262790679932,
      "learning_rate": 4.158580668015668e-06,
      "loss": 0.5308,
      "step": 125660
    },
    {
      "epoch": 51.44,
      "grad_norm": 2.221050500869751,
      "learning_rate": 4.158480285500288e-06,
      "loss": 0.5394,
      "step": 125670
    },
    {
      "epoch": 51.44,
      "grad_norm": 3.132953405380249,
      "learning_rate": 4.1583798946719106e-06,
      "loss": 0.5183,
      "step": 125680
    },
    {
      "epoch": 51.45,
      "grad_norm": 1.887831449508667,
      "learning_rate": 4.158279495530994e-06,
      "loss": 0.5458,
      "step": 125690
    },
    {
      "epoch": 51.45,
      "grad_norm": 2.3211352825164795,
      "learning_rate": 4.158179088078e-06,
      "loss": 0.5411,
      "step": 125700
    },
    {
      "epoch": 51.46,
      "grad_norm": 2.059176206588745,
      "learning_rate": 4.158078672313387e-06,
      "loss": 0.5352,
      "step": 125710
    },
    {
      "epoch": 51.46,
      "grad_norm": 2.0691604614257812,
      "learning_rate": 4.1579782482376145e-06,
      "loss": 0.5317,
      "step": 125720
    },
    {
      "epoch": 51.47,
      "grad_norm": 2.5943808555603027,
      "learning_rate": 4.157877815851144e-06,
      "loss": 0.5591,
      "step": 125730
    },
    {
      "epoch": 51.47,
      "grad_norm": 1.8077545166015625,
      "learning_rate": 4.1577773751544344e-06,
      "loss": 0.5348,
      "step": 125740
    },
    {
      "epoch": 51.47,
      "grad_norm": 2.2749364376068115,
      "learning_rate": 4.1576769261479466e-06,
      "loss": 0.5469,
      "step": 125750
    },
    {
      "epoch": 51.48,
      "grad_norm": 1.676824927330017,
      "learning_rate": 4.15757646883214e-06,
      "loss": 0.5443,
      "step": 125760
    },
    {
      "epoch": 51.48,
      "grad_norm": 2.052952766418457,
      "learning_rate": 4.1574760032074755e-06,
      "loss": 0.5451,
      "step": 125770
    },
    {
      "epoch": 51.49,
      "grad_norm": 2.49365496635437,
      "learning_rate": 4.157375529274414e-06,
      "loss": 0.5557,
      "step": 125780
    },
    {
      "epoch": 51.49,
      "grad_norm": 3.10341215133667,
      "learning_rate": 4.157275047033414e-06,
      "loss": 0.5393,
      "step": 125790
    },
    {
      "epoch": 51.49,
      "grad_norm": 1.7555954456329346,
      "learning_rate": 4.157174556484937e-06,
      "loss": 0.5387,
      "step": 125800
    },
    {
      "epoch": 51.5,
      "grad_norm": 1.7981404066085815,
      "learning_rate": 4.157074057629442e-06,
      "loss": 0.5359,
      "step": 125810
    },
    {
      "epoch": 51.5,
      "grad_norm": 2.2730181217193604,
      "learning_rate": 4.156973550467392e-06,
      "loss": 0.5502,
      "step": 125820
    },
    {
      "epoch": 51.51,
      "grad_norm": 1.9455796480178833,
      "learning_rate": 4.156873034999244e-06,
      "loss": 0.5523,
      "step": 125830
    },
    {
      "epoch": 51.51,
      "grad_norm": 1.7812604904174805,
      "learning_rate": 4.156772511225461e-06,
      "loss": 0.5388,
      "step": 125840
    },
    {
      "epoch": 51.51,
      "grad_norm": 2.799679756164551,
      "learning_rate": 4.1566719791465024e-06,
      "loss": 0.5595,
      "step": 125850
    },
    {
      "epoch": 51.52,
      "grad_norm": 1.5637097358703613,
      "learning_rate": 4.1565714387628295e-06,
      "loss": 0.5222,
      "step": 125860
    },
    {
      "epoch": 51.52,
      "grad_norm": 3.245661735534668,
      "learning_rate": 4.156470890074902e-06,
      "loss": 0.5447,
      "step": 125870
    },
    {
      "epoch": 51.53,
      "grad_norm": 2.4584593772888184,
      "learning_rate": 4.156370333083181e-06,
      "loss": 0.5496,
      "step": 125880
    },
    {
      "epoch": 51.53,
      "grad_norm": 1.6846539974212646,
      "learning_rate": 4.156269767788126e-06,
      "loss": 0.5174,
      "step": 125890
    },
    {
      "epoch": 51.53,
      "grad_norm": 1.7655694484710693,
      "learning_rate": 4.1561691941902e-06,
      "loss": 0.5333,
      "step": 125900
    },
    {
      "epoch": 51.54,
      "grad_norm": 1.575495719909668,
      "learning_rate": 4.156068612289861e-06,
      "loss": 0.5503,
      "step": 125910
    },
    {
      "epoch": 51.54,
      "grad_norm": 1.9369069337844849,
      "learning_rate": 4.155968022087573e-06,
      "loss": 0.5049,
      "step": 125920
    },
    {
      "epoch": 51.55,
      "grad_norm": 1.6705371141433716,
      "learning_rate": 4.155867423583793e-06,
      "loss": 0.5437,
      "step": 125930
    },
    {
      "epoch": 51.55,
      "grad_norm": 1.9501193761825562,
      "learning_rate": 4.155766816778985e-06,
      "loss": 0.5318,
      "step": 125940
    },
    {
      "epoch": 51.56,
      "grad_norm": 1.8952550888061523,
      "learning_rate": 4.1556662016736084e-06,
      "loss": 0.535,
      "step": 125950
    },
    {
      "epoch": 51.56,
      "grad_norm": 2.893663167953491,
      "learning_rate": 4.155565578268124e-06,
      "loss": 0.551,
      "step": 125960
    },
    {
      "epoch": 51.56,
      "grad_norm": 1.4592254161834717,
      "learning_rate": 4.155464946562993e-06,
      "loss": 0.5541,
      "step": 125970
    },
    {
      "epoch": 51.57,
      "grad_norm": 2.2126874923706055,
      "learning_rate": 4.155364306558677e-06,
      "loss": 0.5543,
      "step": 125980
    },
    {
      "epoch": 51.57,
      "grad_norm": 2.9685492515563965,
      "learning_rate": 4.155263658255635e-06,
      "loss": 0.5585,
      "step": 125990
    },
    {
      "epoch": 51.58,
      "grad_norm": 2.6634457111358643,
      "learning_rate": 4.15516300165433e-06,
      "loss": 0.5381,
      "step": 126000
    },
    {
      "epoch": 51.58,
      "grad_norm": 1.9173624515533447,
      "learning_rate": 4.155062336755223e-06,
      "loss": 0.5535,
      "step": 126010
    },
    {
      "epoch": 51.58,
      "grad_norm": 1.4075490236282349,
      "learning_rate": 4.154961663558775e-06,
      "loss": 0.5451,
      "step": 126020
    },
    {
      "epoch": 51.59,
      "grad_norm": 2.24863862991333,
      "learning_rate": 4.154860982065447e-06,
      "loss": 0.5434,
      "step": 126030
    },
    {
      "epoch": 51.59,
      "grad_norm": 1.9196189641952515,
      "learning_rate": 4.154760292275699e-06,
      "loss": 0.5237,
      "step": 126040
    },
    {
      "epoch": 51.6,
      "grad_norm": 2.0981860160827637,
      "learning_rate": 4.154659594189994e-06,
      "loss": 0.5359,
      "step": 126050
    },
    {
      "epoch": 51.6,
      "grad_norm": 2.0641095638275146,
      "learning_rate": 4.154558887808792e-06,
      "loss": 0.5389,
      "step": 126060
    },
    {
      "epoch": 51.6,
      "grad_norm": 1.5757942199707031,
      "learning_rate": 4.154458173132556e-06,
      "loss": 0.5308,
      "step": 126070
    },
    {
      "epoch": 51.61,
      "grad_norm": 2.086862325668335,
      "learning_rate": 4.154357450161745e-06,
      "loss": 0.5231,
      "step": 126080
    },
    {
      "epoch": 51.61,
      "grad_norm": 1.6299716234207153,
      "learning_rate": 4.154256718896823e-06,
      "loss": 0.54,
      "step": 126090
    },
    {
      "epoch": 51.62,
      "grad_norm": 1.8405444622039795,
      "learning_rate": 4.1541559793382495e-06,
      "loss": 0.5524,
      "step": 126100
    },
    {
      "epoch": 51.62,
      "grad_norm": 2.1887238025665283,
      "learning_rate": 4.154055231486486e-06,
      "loss": 0.5362,
      "step": 126110
    },
    {
      "epoch": 51.63,
      "grad_norm": 1.6990675926208496,
      "learning_rate": 4.153954475341996e-06,
      "loss": 0.524,
      "step": 126120
    },
    {
      "epoch": 51.63,
      "grad_norm": 1.844632625579834,
      "learning_rate": 4.153853710905239e-06,
      "loss": 0.5622,
      "step": 126130
    },
    {
      "epoch": 51.63,
      "grad_norm": 2.283845901489258,
      "learning_rate": 4.153752938176677e-06,
      "loss": 0.554,
      "step": 126140
    },
    {
      "epoch": 51.64,
      "grad_norm": 1.8289538621902466,
      "learning_rate": 4.153652157156772e-06,
      "loss": 0.5508,
      "step": 126150
    },
    {
      "epoch": 51.64,
      "grad_norm": 2.000547170639038,
      "learning_rate": 4.153551367845986e-06,
      "loss": 0.5453,
      "step": 126160
    },
    {
      "epoch": 51.65,
      "grad_norm": 2.6733899116516113,
      "learning_rate": 4.153450570244779e-06,
      "loss": 0.5447,
      "step": 126170
    },
    {
      "epoch": 51.65,
      "grad_norm": 2.1070306301116943,
      "learning_rate": 4.1533497643536144e-06,
      "loss": 0.5304,
      "step": 126180
    },
    {
      "epoch": 51.65,
      "grad_norm": 2.5500664710998535,
      "learning_rate": 4.153248950172954e-06,
      "loss": 0.5722,
      "step": 126190
    },
    {
      "epoch": 51.66,
      "grad_norm": 2.4888665676116943,
      "learning_rate": 4.153148127703259e-06,
      "loss": 0.5311,
      "step": 126200
    },
    {
      "epoch": 51.66,
      "grad_norm": 1.7885711193084717,
      "learning_rate": 4.153047296944991e-06,
      "loss": 0.5506,
      "step": 126210
    },
    {
      "epoch": 51.67,
      "grad_norm": 2.212153911590576,
      "learning_rate": 4.152946457898612e-06,
      "loss": 0.5266,
      "step": 126220
    },
    {
      "epoch": 51.67,
      "grad_norm": 2.304469108581543,
      "learning_rate": 4.152845610564585e-06,
      "loss": 0.5063,
      "step": 126230
    },
    {
      "epoch": 51.67,
      "grad_norm": 1.5826894044876099,
      "learning_rate": 4.1527447549433715e-06,
      "loss": 0.5483,
      "step": 126240
    },
    {
      "epoch": 51.68,
      "grad_norm": 1.7856807708740234,
      "learning_rate": 4.152643891035433e-06,
      "loss": 0.5336,
      "step": 126250
    },
    {
      "epoch": 51.68,
      "grad_norm": 1.6774712800979614,
      "learning_rate": 4.152543018841231e-06,
      "loss": 0.5393,
      "step": 126260
    },
    {
      "epoch": 51.69,
      "grad_norm": 1.9989269971847534,
      "learning_rate": 4.152442138361229e-06,
      "loss": 0.5536,
      "step": 126270
    },
    {
      "epoch": 51.69,
      "grad_norm": 1.8887757062911987,
      "learning_rate": 4.152341249595888e-06,
      "loss": 0.5387,
      "step": 126280
    },
    {
      "epoch": 51.69,
      "grad_norm": 1.385276436805725,
      "learning_rate": 4.15224035254567e-06,
      "loss": 0.5287,
      "step": 126290
    },
    {
      "epoch": 51.7,
      "grad_norm": 1.8087835311889648,
      "learning_rate": 4.152139447211039e-06,
      "loss": 0.5243,
      "step": 126300
    },
    {
      "epoch": 51.7,
      "grad_norm": 2.3537068367004395,
      "learning_rate": 4.152038533592456e-06,
      "loss": 0.5462,
      "step": 126310
    },
    {
      "epoch": 51.71,
      "grad_norm": 2.0465190410614014,
      "learning_rate": 4.151937611690382e-06,
      "loss": 0.5389,
      "step": 126320
    },
    {
      "epoch": 51.71,
      "grad_norm": 1.626522421836853,
      "learning_rate": 4.151836681505282e-06,
      "loss": 0.5494,
      "step": 126330
    },
    {
      "epoch": 51.72,
      "grad_norm": 1.7519575357437134,
      "learning_rate": 4.151735743037616e-06,
      "loss": 0.5339,
      "step": 126340
    },
    {
      "epoch": 51.72,
      "grad_norm": 2.2741737365722656,
      "learning_rate": 4.151634796287847e-06,
      "loss": 0.5196,
      "step": 126350
    },
    {
      "epoch": 51.72,
      "grad_norm": 1.5675148963928223,
      "learning_rate": 4.151533841256439e-06,
      "loss": 0.5153,
      "step": 126360
    },
    {
      "epoch": 51.73,
      "grad_norm": 2.0781936645507812,
      "learning_rate": 4.1514328779438525e-06,
      "loss": 0.531,
      "step": 126370
    },
    {
      "epoch": 51.73,
      "grad_norm": 2.388582706451416,
      "learning_rate": 4.15133190635055e-06,
      "loss": 0.5357,
      "step": 126380
    },
    {
      "epoch": 51.74,
      "grad_norm": 1.9597883224487305,
      "learning_rate": 4.1512309264769955e-06,
      "loss": 0.5183,
      "step": 126390
    },
    {
      "epoch": 51.74,
      "grad_norm": 2.628281831741333,
      "learning_rate": 4.151129938323651e-06,
      "loss": 0.5499,
      "step": 126400
    },
    {
      "epoch": 51.74,
      "grad_norm": 1.9855296611785889,
      "learning_rate": 4.151028941890978e-06,
      "loss": 0.5307,
      "step": 126410
    },
    {
      "epoch": 51.75,
      "grad_norm": 1.8947638273239136,
      "learning_rate": 4.150927937179441e-06,
      "loss": 0.5431,
      "step": 126420
    },
    {
      "epoch": 51.75,
      "grad_norm": 2.1245155334472656,
      "learning_rate": 4.1508269241895016e-06,
      "loss": 0.5174,
      "step": 126430
    },
    {
      "epoch": 51.76,
      "grad_norm": 2.608677864074707,
      "learning_rate": 4.150725902921623e-06,
      "loss": 0.5412,
      "step": 126440
    },
    {
      "epoch": 51.76,
      "grad_norm": 1.6333361864089966,
      "learning_rate": 4.150624873376267e-06,
      "loss": 0.5533,
      "step": 126450
    },
    {
      "epoch": 51.76,
      "grad_norm": 1.9615975618362427,
      "learning_rate": 4.150523835553897e-06,
      "loss": 0.5526,
      "step": 126460
    },
    {
      "epoch": 51.77,
      "grad_norm": 2.1834988594055176,
      "learning_rate": 4.150422789454976e-06,
      "loss": 0.5341,
      "step": 126470
    },
    {
      "epoch": 51.77,
      "grad_norm": 2.1717522144317627,
      "learning_rate": 4.150321735079967e-06,
      "loss": 0.5124,
      "step": 126480
    },
    {
      "epoch": 51.78,
      "grad_norm": 2.159883975982666,
      "learning_rate": 4.150220672429332e-06,
      "loss": 0.5308,
      "step": 126490
    },
    {
      "epoch": 51.78,
      "grad_norm": 1.9615261554718018,
      "learning_rate": 4.150119601503535e-06,
      "loss": 0.5269,
      "step": 126500
    },
    {
      "epoch": 51.78,
      "grad_norm": 2.1872611045837402,
      "learning_rate": 4.150018522303039e-06,
      "loss": 0.5278,
      "step": 126510
    },
    {
      "epoch": 51.79,
      "grad_norm": 1.7783710956573486,
      "learning_rate": 4.149917434828307e-06,
      "loss": 0.5393,
      "step": 126520
    },
    {
      "epoch": 51.79,
      "grad_norm": 2.7362148761749268,
      "learning_rate": 4.1498163390798e-06,
      "loss": 0.5234,
      "step": 126530
    },
    {
      "epoch": 51.8,
      "grad_norm": 2.2107582092285156,
      "learning_rate": 4.149715235057984e-06,
      "loss": 0.541,
      "step": 126540
    },
    {
      "epoch": 51.8,
      "grad_norm": 1.8854700326919556,
      "learning_rate": 4.149614122763321e-06,
      "loss": 0.5519,
      "step": 126550
    },
    {
      "epoch": 51.81,
      "grad_norm": 2.538468360900879,
      "learning_rate": 4.149513002196274e-06,
      "loss": 0.5578,
      "step": 126560
    },
    {
      "epoch": 51.81,
      "grad_norm": 1.7837519645690918,
      "learning_rate": 4.149411873357306e-06,
      "loss": 0.5496,
      "step": 126570
    },
    {
      "epoch": 51.81,
      "grad_norm": 1.9821178913116455,
      "learning_rate": 4.1493107362468815e-06,
      "loss": 0.5536,
      "step": 126580
    },
    {
      "epoch": 51.82,
      "grad_norm": 1.879075288772583,
      "learning_rate": 4.149209590865462e-06,
      "loss": 0.5573,
      "step": 126590
    },
    {
      "epoch": 51.82,
      "grad_norm": 2.7000961303710938,
      "learning_rate": 4.149108437213512e-06,
      "loss": 0.5265,
      "step": 126600
    },
    {
      "epoch": 51.83,
      "grad_norm": 2.177027940750122,
      "learning_rate": 4.149007275291494e-06,
      "loss": 0.535,
      "step": 126610
    },
    {
      "epoch": 51.83,
      "grad_norm": 2.1944987773895264,
      "learning_rate": 4.1489061050998735e-06,
      "loss": 0.5481,
      "step": 126620
    },
    {
      "epoch": 51.83,
      "grad_norm": 2.4799492359161377,
      "learning_rate": 4.148804926639111e-06,
      "loss": 0.5383,
      "step": 126630
    },
    {
      "epoch": 51.84,
      "grad_norm": 1.9346095323562622,
      "learning_rate": 4.148703739909672e-06,
      "loss": 0.5373,
      "step": 126640
    },
    {
      "epoch": 51.84,
      "grad_norm": 1.8488428592681885,
      "learning_rate": 4.148602544912019e-06,
      "loss": 0.5447,
      "step": 126650
    },
    {
      "epoch": 51.85,
      "grad_norm": 1.7793911695480347,
      "learning_rate": 4.148501341646617e-06,
      "loss": 0.5285,
      "step": 126660
    },
    {
      "epoch": 51.85,
      "grad_norm": 2.166804075241089,
      "learning_rate": 4.148400130113928e-06,
      "loss": 0.5427,
      "step": 126670
    },
    {
      "epoch": 51.85,
      "grad_norm": 2.2949228286743164,
      "learning_rate": 4.148298910314416e-06,
      "loss": 0.5364,
      "step": 126680
    },
    {
      "epoch": 51.86,
      "grad_norm": 2.3529553413391113,
      "learning_rate": 4.148197682248545e-06,
      "loss": 0.5347,
      "step": 126690
    },
    {
      "epoch": 51.86,
      "grad_norm": 2.2041471004486084,
      "learning_rate": 4.148096445916779e-06,
      "loss": 0.5486,
      "step": 126700
    },
    {
      "epoch": 51.87,
      "grad_norm": 2.4096224308013916,
      "learning_rate": 4.147995201319581e-06,
      "loss": 0.5432,
      "step": 126710
    },
    {
      "epoch": 51.87,
      "grad_norm": 1.8648160696029663,
      "learning_rate": 4.147893948457416e-06,
      "loss": 0.5487,
      "step": 126720
    },
    {
      "epoch": 51.87,
      "grad_norm": 2.0057995319366455,
      "learning_rate": 4.147792687330746e-06,
      "loss": 0.531,
      "step": 126730
    },
    {
      "epoch": 51.88,
      "grad_norm": 1.9171607494354248,
      "learning_rate": 4.147691417940036e-06,
      "loss": 0.5421,
      "step": 126740
    },
    {
      "epoch": 51.88,
      "grad_norm": 2.049478769302368,
      "learning_rate": 4.14759014028575e-06,
      "loss": 0.5119,
      "step": 126750
    },
    {
      "epoch": 51.89,
      "grad_norm": 2.855839729309082,
      "learning_rate": 4.147488854368351e-06,
      "loss": 0.5422,
      "step": 126760
    },
    {
      "epoch": 51.89,
      "grad_norm": 1.997131109237671,
      "learning_rate": 4.147387560188305e-06,
      "loss": 0.5254,
      "step": 126770
    },
    {
      "epoch": 51.9,
      "grad_norm": 2.280219078063965,
      "learning_rate": 4.147286257746073e-06,
      "loss": 0.5172,
      "step": 126780
    },
    {
      "epoch": 51.9,
      "grad_norm": 2.5016324520111084,
      "learning_rate": 4.147184947042122e-06,
      "loss": 0.5483,
      "step": 126790
    },
    {
      "epoch": 51.9,
      "grad_norm": 2.6538426876068115,
      "learning_rate": 4.147083628076914e-06,
      "loss": 0.5391,
      "step": 126800
    },
    {
      "epoch": 51.91,
      "grad_norm": 2.2483978271484375,
      "learning_rate": 4.146982300850915e-06,
      "loss": 0.5649,
      "step": 126810
    },
    {
      "epoch": 51.91,
      "grad_norm": 2.240370273590088,
      "learning_rate": 4.146880965364587e-06,
      "loss": 0.5432,
      "step": 126820
    },
    {
      "epoch": 51.92,
      "grad_norm": 2.028135061264038,
      "learning_rate": 4.146779621618395e-06,
      "loss": 0.5316,
      "step": 126830
    },
    {
      "epoch": 51.92,
      "grad_norm": 2.58593487739563,
      "learning_rate": 4.146678269612805e-06,
      "loss": 0.5306,
      "step": 126840
    },
    {
      "epoch": 51.92,
      "grad_norm": 2.3396801948547363,
      "learning_rate": 4.146576909348279e-06,
      "loss": 0.5336,
      "step": 126850
    },
    {
      "epoch": 51.93,
      "grad_norm": 1.8308570384979248,
      "learning_rate": 4.146475540825283e-06,
      "loss": 0.5307,
      "step": 126860
    },
    {
      "epoch": 51.93,
      "grad_norm": 2.666254758834839,
      "learning_rate": 4.146374164044279e-06,
      "loss": 0.5328,
      "step": 126870
    },
    {
      "epoch": 51.94,
      "grad_norm": 2.3525807857513428,
      "learning_rate": 4.146272779005734e-06,
      "loss": 0.5295,
      "step": 126880
    },
    {
      "epoch": 51.94,
      "grad_norm": 2.033179998397827,
      "learning_rate": 4.146171385710111e-06,
      "loss": 0.5552,
      "step": 126890
    },
    {
      "epoch": 51.94,
      "grad_norm": 1.8990610837936401,
      "learning_rate": 4.146069984157874e-06,
      "loss": 0.5339,
      "step": 126900
    },
    {
      "epoch": 51.95,
      "grad_norm": 1.9442200660705566,
      "learning_rate": 4.1459685743494895e-06,
      "loss": 0.5236,
      "step": 126910
    },
    {
      "epoch": 51.95,
      "grad_norm": 1.7188166379928589,
      "learning_rate": 4.14586715628542e-06,
      "loss": 0.549,
      "step": 126920
    },
    {
      "epoch": 51.96,
      "grad_norm": 1.5796815156936646,
      "learning_rate": 4.145765729966131e-06,
      "loss": 0.5425,
      "step": 126930
    },
    {
      "epoch": 51.96,
      "grad_norm": 2.178621530532837,
      "learning_rate": 4.145664295392087e-06,
      "loss": 0.552,
      "step": 126940
    },
    {
      "epoch": 51.96,
      "grad_norm": 2.2983291149139404,
      "learning_rate": 4.145562852563753e-06,
      "loss": 0.5402,
      "step": 126950
    },
    {
      "epoch": 51.97,
      "grad_norm": 2.4858803749084473,
      "learning_rate": 4.145461401481593e-06,
      "loss": 0.5395,
      "step": 126960
    },
    {
      "epoch": 51.97,
      "grad_norm": 2.677644729614258,
      "learning_rate": 4.1453599421460724e-06,
      "loss": 0.5396,
      "step": 126970
    },
    {
      "epoch": 51.98,
      "grad_norm": 1.8611841201782227,
      "learning_rate": 4.145258474557655e-06,
      "loss": 0.5393,
      "step": 126980
    },
    {
      "epoch": 51.98,
      "grad_norm": 2.203974723815918,
      "learning_rate": 4.145156998716807e-06,
      "loss": 0.5424,
      "step": 126990
    },
    {
      "epoch": 51.99,
      "grad_norm": 2.2047996520996094,
      "learning_rate": 4.145055514623993e-06,
      "loss": 0.5455,
      "step": 127000
    },
    {
      "epoch": 51.99,
      "grad_norm": 1.9173345565795898,
      "learning_rate": 4.144954022279676e-06,
      "loss": 0.5357,
      "step": 127010
    },
    {
      "epoch": 51.99,
      "grad_norm": 2.3235323429107666,
      "learning_rate": 4.144852521684324e-06,
      "loss": 0.5317,
      "step": 127020
    },
    {
      "epoch": 52.0,
      "grad_norm": 2.0888619422912598,
      "learning_rate": 4.144751012838399e-06,
      "loss": 0.5498,
      "step": 127030
    },
    {
      "epoch": 52.0,
      "eval_loss": 0.5393477082252502,
      "eval_runtime": 51.961,
      "eval_samples_per_second": 66.377,
      "eval_steps_per_second": 8.314,
      "step": 127036
    },
    {
      "epoch": 52.0,
      "grad_norm": 1.8081159591674805,
      "learning_rate": 4.144649495742367e-06,
      "loss": 0.5272,
      "step": 127040
    },
    {
      "epoch": 52.01,
      "grad_norm": 1.958786964416504,
      "learning_rate": 4.144547970396694e-06,
      "loss": 0.5216,
      "step": 127050
    },
    {
      "epoch": 52.01,
      "grad_norm": 1.672959327697754,
      "learning_rate": 4.1444464368018446e-06,
      "loss": 0.5241,
      "step": 127060
    },
    {
      "epoch": 52.01,
      "grad_norm": 2.2643179893493652,
      "learning_rate": 4.144344894958283e-06,
      "loss": 0.5304,
      "step": 127070
    },
    {
      "epoch": 52.02,
      "grad_norm": 2.4498445987701416,
      "learning_rate": 4.1442433448664755e-06,
      "loss": 0.539,
      "step": 127080
    },
    {
      "epoch": 52.02,
      "grad_norm": 1.8549153804779053,
      "learning_rate": 4.144141786526887e-06,
      "loss": 0.536,
      "step": 127090
    },
    {
      "epoch": 52.03,
      "grad_norm": 1.9536012411117554,
      "learning_rate": 4.144040219939983e-06,
      "loss": 0.531,
      "step": 127100
    },
    {
      "epoch": 52.03,
      "grad_norm": 1.8813276290893555,
      "learning_rate": 4.143938645106228e-06,
      "loss": 0.5615,
      "step": 127110
    },
    {
      "epoch": 52.03,
      "grad_norm": 1.2652217149734497,
      "learning_rate": 4.143837062026087e-06,
      "loss": 0.5341,
      "step": 127120
    },
    {
      "epoch": 52.04,
      "grad_norm": 1.5732296705245972,
      "learning_rate": 4.143735470700026e-06,
      "loss": 0.5144,
      "step": 127130
    },
    {
      "epoch": 52.04,
      "grad_norm": 2.146064519882202,
      "learning_rate": 4.14363387112851e-06,
      "loss": 0.5843,
      "step": 127140
    },
    {
      "epoch": 52.05,
      "grad_norm": 2.1457436084747314,
      "learning_rate": 4.143532263312006e-06,
      "loss": 0.5518,
      "step": 127150
    },
    {
      "epoch": 52.05,
      "grad_norm": 2.2456002235412598,
      "learning_rate": 4.1434306472509775e-06,
      "loss": 0.536,
      "step": 127160
    },
    {
      "epoch": 52.05,
      "grad_norm": 3.257840633392334,
      "learning_rate": 4.14332902294589e-06,
      "loss": 0.5341,
      "step": 127170
    },
    {
      "epoch": 52.06,
      "grad_norm": 2.4279069900512695,
      "learning_rate": 4.143227390397211e-06,
      "loss": 0.5484,
      "step": 127180
    },
    {
      "epoch": 52.06,
      "grad_norm": 2.268521308898926,
      "learning_rate": 4.143125749605404e-06,
      "loss": 0.5324,
      "step": 127190
    },
    {
      "epoch": 52.07,
      "grad_norm": 1.7931311130523682,
      "learning_rate": 4.1430241005709355e-06,
      "loss": 0.5337,
      "step": 127200
    },
    {
      "epoch": 52.07,
      "grad_norm": 2.1247503757476807,
      "learning_rate": 4.142922443294271e-06,
      "loss": 0.5267,
      "step": 127210
    },
    {
      "epoch": 52.08,
      "grad_norm": 2.278062582015991,
      "learning_rate": 4.142820777775876e-06,
      "loss": 0.5192,
      "step": 127220
    },
    {
      "epoch": 52.08,
      "grad_norm": 2.156494140625,
      "learning_rate": 4.142719104016216e-06,
      "loss": 0.5251,
      "step": 127230
    },
    {
      "epoch": 52.08,
      "grad_norm": 1.8657225370407104,
      "learning_rate": 4.142617422015758e-06,
      "loss": 0.5278,
      "step": 127240
    },
    {
      "epoch": 52.09,
      "grad_norm": 2.240706443786621,
      "learning_rate": 4.142515731774967e-06,
      "loss": 0.5304,
      "step": 127250
    },
    {
      "epoch": 52.09,
      "grad_norm": 1.9199343919754028,
      "learning_rate": 4.1424140332943074e-06,
      "loss": 0.5378,
      "step": 127260
    },
    {
      "epoch": 52.1,
      "grad_norm": 2.1747536659240723,
      "learning_rate": 4.142312326574248e-06,
      "loss": 0.5283,
      "step": 127270
    },
    {
      "epoch": 52.1,
      "grad_norm": 2.042478561401367,
      "learning_rate": 4.142210611615253e-06,
      "loss": 0.5492,
      "step": 127280
    },
    {
      "epoch": 52.1,
      "grad_norm": 1.759676218032837,
      "learning_rate": 4.142108888417788e-06,
      "loss": 0.5257,
      "step": 127290
    },
    {
      "epoch": 52.11,
      "grad_norm": 2.2093570232391357,
      "learning_rate": 4.142007156982318e-06,
      "loss": 0.5251,
      "step": 127300
    },
    {
      "epoch": 52.11,
      "grad_norm": 2.0494444370269775,
      "learning_rate": 4.1419054173093125e-06,
      "loss": 0.5293,
      "step": 127310
    },
    {
      "epoch": 52.12,
      "grad_norm": 2.338200807571411,
      "learning_rate": 4.141803669399235e-06,
      "loss": 0.5348,
      "step": 127320
    },
    {
      "epoch": 52.12,
      "grad_norm": 1.973915696144104,
      "learning_rate": 4.141701913252552e-06,
      "loss": 0.5335,
      "step": 127330
    },
    {
      "epoch": 52.12,
      "grad_norm": 1.8066028356552124,
      "learning_rate": 4.14160014886973e-06,
      "loss": 0.5543,
      "step": 127340
    },
    {
      "epoch": 52.13,
      "grad_norm": 2.1483218669891357,
      "learning_rate": 4.141498376251234e-06,
      "loss": 0.5255,
      "step": 127350
    },
    {
      "epoch": 52.13,
      "grad_norm": 2.1648917198181152,
      "learning_rate": 4.1413965953975315e-06,
      "loss": 0.5326,
      "step": 127360
    },
    {
      "epoch": 52.14,
      "grad_norm": 2.167050361633301,
      "learning_rate": 4.14129480630909e-06,
      "loss": 0.5311,
      "step": 127370
    },
    {
      "epoch": 52.14,
      "grad_norm": 1.8924049139022827,
      "learning_rate": 4.141193008986373e-06,
      "loss": 0.5423,
      "step": 127380
    },
    {
      "epoch": 52.14,
      "grad_norm": 2.3317997455596924,
      "learning_rate": 4.141091203429847e-06,
      "loss": 0.5481,
      "step": 127390
    },
    {
      "epoch": 52.15,
      "grad_norm": 2.0239009857177734,
      "learning_rate": 4.140989389639979e-06,
      "loss": 0.5217,
      "step": 127400
    },
    {
      "epoch": 52.15,
      "grad_norm": 2.1841630935668945,
      "learning_rate": 4.140887567617237e-06,
      "loss": 0.5311,
      "step": 127410
    },
    {
      "epoch": 52.16,
      "grad_norm": 1.9779448509216309,
      "learning_rate": 4.140785737362086e-06,
      "loss": 0.5106,
      "step": 127420
    },
    {
      "epoch": 52.16,
      "grad_norm": 2.1584417819976807,
      "learning_rate": 4.1406838988749925e-06,
      "loss": 0.5541,
      "step": 127430
    },
    {
      "epoch": 52.17,
      "grad_norm": 2.395753860473633,
      "learning_rate": 4.140582052156422e-06,
      "loss": 0.5584,
      "step": 127440
    },
    {
      "epoch": 52.17,
      "grad_norm": 1.6927626132965088,
      "learning_rate": 4.140480197206843e-06,
      "loss": 0.5225,
      "step": 127450
    },
    {
      "epoch": 52.17,
      "grad_norm": 2.129190683364868,
      "learning_rate": 4.140378334026722e-06,
      "loss": 0.5332,
      "step": 127460
    },
    {
      "epoch": 52.18,
      "grad_norm": 2.531015634536743,
      "learning_rate": 4.140276462616524e-06,
      "loss": 0.5474,
      "step": 127470
    },
    {
      "epoch": 52.18,
      "grad_norm": 2.7840240001678467,
      "learning_rate": 4.140174582976717e-06,
      "loss": 0.5267,
      "step": 127480
    },
    {
      "epoch": 52.19,
      "grad_norm": 2.4859020709991455,
      "learning_rate": 4.1400726951077664e-06,
      "loss": 0.517,
      "step": 127490
    },
    {
      "epoch": 52.19,
      "grad_norm": 2.0952794551849365,
      "learning_rate": 4.1399707990101405e-06,
      "loss": 0.5307,
      "step": 127500
    },
    {
      "epoch": 52.19,
      "grad_norm": 1.9125665426254272,
      "learning_rate": 4.139868894684305e-06,
      "loss": 0.5297,
      "step": 127510
    },
    {
      "epoch": 52.2,
      "grad_norm": 1.6265395879745483,
      "learning_rate": 4.1397669821307264e-06,
      "loss": 0.5379,
      "step": 127520
    },
    {
      "epoch": 52.2,
      "grad_norm": 2.1291329860687256,
      "learning_rate": 4.1396650613498735e-06,
      "loss": 0.5351,
      "step": 127530
    },
    {
      "epoch": 52.21,
      "grad_norm": 2.1269919872283936,
      "learning_rate": 4.1395631323422104e-06,
      "loss": 0.5291,
      "step": 127540
    },
    {
      "epoch": 52.21,
      "grad_norm": 1.748836636543274,
      "learning_rate": 4.139461195108207e-06,
      "loss": 0.5475,
      "step": 127550
    },
    {
      "epoch": 52.21,
      "grad_norm": 2.258951425552368,
      "learning_rate": 4.139359249648328e-06,
      "loss": 0.5456,
      "step": 127560
    },
    {
      "epoch": 52.22,
      "grad_norm": 1.7675951719284058,
      "learning_rate": 4.139257295963041e-06,
      "loss": 0.5532,
      "step": 127570
    },
    {
      "epoch": 52.22,
      "grad_norm": 1.6509102582931519,
      "learning_rate": 4.139155334052813e-06,
      "loss": 0.5211,
      "step": 127580
    },
    {
      "epoch": 52.23,
      "grad_norm": 2.1728596687316895,
      "learning_rate": 4.139053363918111e-06,
      "loss": 0.5487,
      "step": 127590
    },
    {
      "epoch": 52.23,
      "grad_norm": 2.122559070587158,
      "learning_rate": 4.138951385559402e-06,
      "loss": 0.5442,
      "step": 127600
    },
    {
      "epoch": 52.23,
      "grad_norm": 2.4908952713012695,
      "learning_rate": 4.138849398977155e-06,
      "loss": 0.5474,
      "step": 127610
    },
    {
      "epoch": 52.24,
      "grad_norm": 1.8818970918655396,
      "learning_rate": 4.1387474041718345e-06,
      "loss": 0.5197,
      "step": 127620
    },
    {
      "epoch": 52.24,
      "grad_norm": 2.3291287422180176,
      "learning_rate": 4.1386454011439095e-06,
      "loss": 0.5312,
      "step": 127630
    },
    {
      "epoch": 52.25,
      "grad_norm": 1.8281400203704834,
      "learning_rate": 4.138543389893846e-06,
      "loss": 0.537,
      "step": 127640
    },
    {
      "epoch": 52.25,
      "grad_norm": 3.096074104309082,
      "learning_rate": 4.138441370422112e-06,
      "loss": 0.5587,
      "step": 127650
    },
    {
      "epoch": 52.26,
      "grad_norm": 2.2286453247070312,
      "learning_rate": 4.1383393427291755e-06,
      "loss": 0.5371,
      "step": 127660
    },
    {
      "epoch": 52.26,
      "grad_norm": 2.2772469520568848,
      "learning_rate": 4.138237306815502e-06,
      "loss": 0.5515,
      "step": 127670
    },
    {
      "epoch": 52.26,
      "grad_norm": 2.2626233100891113,
      "learning_rate": 4.138135262681561e-06,
      "loss": 0.5476,
      "step": 127680
    },
    {
      "epoch": 52.27,
      "grad_norm": 2.0017895698547363,
      "learning_rate": 4.138033210327819e-06,
      "loss": 0.5596,
      "step": 127690
    },
    {
      "epoch": 52.27,
      "grad_norm": 2.1985294818878174,
      "learning_rate": 4.137931149754743e-06,
      "loss": 0.5491,
      "step": 127700
    },
    {
      "epoch": 52.28,
      "grad_norm": 1.744323492050171,
      "learning_rate": 4.137829080962801e-06,
      "loss": 0.5374,
      "step": 127710
    },
    {
      "epoch": 52.28,
      "grad_norm": 2.5219061374664307,
      "learning_rate": 4.1377270039524615e-06,
      "loss": 0.5479,
      "step": 127720
    },
    {
      "epoch": 52.28,
      "grad_norm": 1.9632006883621216,
      "learning_rate": 4.13762491872419e-06,
      "loss": 0.5272,
      "step": 127730
    },
    {
      "epoch": 52.29,
      "grad_norm": 1.696826457977295,
      "learning_rate": 4.137522825278456e-06,
      "loss": 0.5416,
      "step": 127740
    },
    {
      "epoch": 52.29,
      "grad_norm": 2.7721080780029297,
      "learning_rate": 4.137420723615726e-06,
      "loss": 0.5372,
      "step": 127750
    },
    {
      "epoch": 52.3,
      "grad_norm": 2.5054636001586914,
      "learning_rate": 4.137318613736469e-06,
      "loss": 0.5336,
      "step": 127760
    },
    {
      "epoch": 52.3,
      "grad_norm": 2.93949818611145,
      "learning_rate": 4.137216495641152e-06,
      "loss": 0.5279,
      "step": 127770
    },
    {
      "epoch": 52.3,
      "grad_norm": 2.274883270263672,
      "learning_rate": 4.137114369330242e-06,
      "loss": 0.5389,
      "step": 127780
    },
    {
      "epoch": 52.31,
      "grad_norm": 1.7299373149871826,
      "learning_rate": 4.1370122348042075e-06,
      "loss": 0.5343,
      "step": 127790
    },
    {
      "epoch": 52.31,
      "grad_norm": 1.9843236207962036,
      "learning_rate": 4.136910092063516e-06,
      "loss": 0.5238,
      "step": 127800
    },
    {
      "epoch": 52.32,
      "grad_norm": 1.8946349620819092,
      "learning_rate": 4.136807941108637e-06,
      "loss": 0.5527,
      "step": 127810
    },
    {
      "epoch": 52.32,
      "grad_norm": 2.251221179962158,
      "learning_rate": 4.1367057819400366e-06,
      "loss": 0.5281,
      "step": 127820
    },
    {
      "epoch": 52.33,
      "grad_norm": 1.6098047494888306,
      "learning_rate": 4.136603614558184e-06,
      "loss": 0.5351,
      "step": 127830
    },
    {
      "epoch": 52.33,
      "grad_norm": 1.856268048286438,
      "learning_rate": 4.136501438963546e-06,
      "loss": 0.5168,
      "step": 127840
    },
    {
      "epoch": 52.33,
      "grad_norm": 2.5549476146698,
      "learning_rate": 4.136399255156592e-06,
      "loss": 0.5594,
      "step": 127850
    },
    {
      "epoch": 52.34,
      "grad_norm": 2.405104637145996,
      "learning_rate": 4.136297063137789e-06,
      "loss": 0.5123,
      "step": 127860
    },
    {
      "epoch": 52.34,
      "grad_norm": 2.7919764518737793,
      "learning_rate": 4.136194862907605e-06,
      "loss": 0.5582,
      "step": 127870
    },
    {
      "epoch": 52.35,
      "grad_norm": 2.1780545711517334,
      "learning_rate": 4.136092654466509e-06,
      "loss": 0.5388,
      "step": 127880
    },
    {
      "epoch": 52.35,
      "grad_norm": 2.3738481998443604,
      "learning_rate": 4.135990437814969e-06,
      "loss": 0.5446,
      "step": 127890
    },
    {
      "epoch": 52.35,
      "grad_norm": 2.3778278827667236,
      "learning_rate": 4.135888212953454e-06,
      "loss": 0.5436,
      "step": 127900
    },
    {
      "epoch": 52.36,
      "grad_norm": 2.2800710201263428,
      "learning_rate": 4.13578597988243e-06,
      "loss": 0.5311,
      "step": 127910
    },
    {
      "epoch": 52.36,
      "grad_norm": 2.4334986209869385,
      "learning_rate": 4.135683738602368e-06,
      "loss": 0.5432,
      "step": 127920
    },
    {
      "epoch": 52.37,
      "grad_norm": 2.8421616554260254,
      "learning_rate": 4.135581489113734e-06,
      "loss": 0.5411,
      "step": 127930
    },
    {
      "epoch": 52.37,
      "grad_norm": 1.858958125114441,
      "learning_rate": 4.135479231416997e-06,
      "loss": 0.5257,
      "step": 127940
    },
    {
      "epoch": 52.37,
      "grad_norm": 2.074000597000122,
      "learning_rate": 4.135376965512627e-06,
      "loss": 0.5333,
      "step": 127950
    },
    {
      "epoch": 52.38,
      "grad_norm": 2.2023611068725586,
      "learning_rate": 4.135274691401091e-06,
      "loss": 0.5535,
      "step": 127960
    },
    {
      "epoch": 52.38,
      "grad_norm": 2.6080400943756104,
      "learning_rate": 4.135172409082858e-06,
      "loss": 0.5211,
      "step": 127970
    },
    {
      "epoch": 52.39,
      "grad_norm": 2.14388370513916,
      "learning_rate": 4.135070118558396e-06,
      "loss": 0.5254,
      "step": 127980
    },
    {
      "epoch": 52.39,
      "grad_norm": 1.8782509565353394,
      "learning_rate": 4.134967819828174e-06,
      "loss": 0.5421,
      "step": 127990
    },
    {
      "epoch": 52.39,
      "grad_norm": 2.128122568130493,
      "learning_rate": 4.134865512892661e-06,
      "loss": 0.5357,
      "step": 128000
    },
    {
      "epoch": 52.4,
      "grad_norm": 1.4299017190933228,
      "learning_rate": 4.1347631977523245e-06,
      "loss": 0.5513,
      "step": 128010
    },
    {
      "epoch": 52.4,
      "grad_norm": 2.189673662185669,
      "learning_rate": 4.134660874407635e-06,
      "loss": 0.5414,
      "step": 128020
    },
    {
      "epoch": 52.41,
      "grad_norm": 2.128826379776001,
      "learning_rate": 4.134558542859059e-06,
      "loss": 0.5257,
      "step": 128030
    },
    {
      "epoch": 52.41,
      "grad_norm": 1.9432885646820068,
      "learning_rate": 4.134456203107067e-06,
      "loss": 0.5119,
      "step": 128040
    },
    {
      "epoch": 52.42,
      "grad_norm": 2.1321845054626465,
      "learning_rate": 4.134353855152127e-06,
      "loss": 0.5201,
      "step": 128050
    },
    {
      "epoch": 52.42,
      "grad_norm": 2.4484217166900635,
      "learning_rate": 4.134251498994708e-06,
      "loss": 0.5318,
      "step": 128060
    },
    {
      "epoch": 52.42,
      "grad_norm": 1.812617540359497,
      "learning_rate": 4.1341491346352795e-06,
      "loss": 0.5466,
      "step": 128070
    },
    {
      "epoch": 52.43,
      "grad_norm": 1.7415724992752075,
      "learning_rate": 4.134046762074309e-06,
      "loss": 0.5418,
      "step": 128080
    },
    {
      "epoch": 52.43,
      "grad_norm": 1.9170063734054565,
      "learning_rate": 4.1339443813122675e-06,
      "loss": 0.5204,
      "step": 128090
    },
    {
      "epoch": 52.44,
      "grad_norm": 1.398492693901062,
      "learning_rate": 4.133841992349621e-06,
      "loss": 0.5421,
      "step": 128100
    },
    {
      "epoch": 52.44,
      "grad_norm": 2.0794310569763184,
      "learning_rate": 4.1337395951868416e-06,
      "loss": 0.5346,
      "step": 128110
    },
    {
      "epoch": 52.44,
      "grad_norm": 2.141857624053955,
      "learning_rate": 4.1336371898243975e-06,
      "loss": 0.5555,
      "step": 128120
    },
    {
      "epoch": 52.45,
      "grad_norm": 1.9400619268417358,
      "learning_rate": 4.133534776262756e-06,
      "loss": 0.5177,
      "step": 128130
    },
    {
      "epoch": 52.45,
      "grad_norm": 2.229395627975464,
      "learning_rate": 4.133432354502389e-06,
      "loss": 0.5426,
      "step": 128140
    },
    {
      "epoch": 52.46,
      "grad_norm": 2.5176749229431152,
      "learning_rate": 4.133329924543763e-06,
      "loss": 0.5533,
      "step": 128150
    },
    {
      "epoch": 52.46,
      "grad_norm": 2.261626720428467,
      "learning_rate": 4.1332274863873495e-06,
      "loss": 0.5354,
      "step": 128160
    },
    {
      "epoch": 52.46,
      "grad_norm": 2.0106401443481445,
      "learning_rate": 4.133125040033617e-06,
      "loss": 0.5305,
      "step": 128170
    },
    {
      "epoch": 52.47,
      "grad_norm": 1.9463300704956055,
      "learning_rate": 4.133022585483033e-06,
      "loss": 0.527,
      "step": 128180
    },
    {
      "epoch": 52.47,
      "grad_norm": 1.9787930250167847,
      "learning_rate": 4.132920122736069e-06,
      "loss": 0.544,
      "step": 128190
    },
    {
      "epoch": 52.48,
      "grad_norm": 2.0918197631835938,
      "learning_rate": 4.132817651793194e-06,
      "loss": 0.5212,
      "step": 128200
    },
    {
      "epoch": 52.48,
      "grad_norm": 2.2111587524414062,
      "learning_rate": 4.132715172654878e-06,
      "loss": 0.5458,
      "step": 128210
    },
    {
      "epoch": 52.48,
      "grad_norm": 1.721646785736084,
      "learning_rate": 4.132612685321589e-06,
      "loss": 0.5376,
      "step": 128220
    },
    {
      "epoch": 52.49,
      "grad_norm": 2.07804536819458,
      "learning_rate": 4.132510189793796e-06,
      "loss": 0.5515,
      "step": 128230
    },
    {
      "epoch": 52.49,
      "grad_norm": 1.999291181564331,
      "learning_rate": 4.132407686071971e-06,
      "loss": 0.5466,
      "step": 128240
    },
    {
      "epoch": 52.5,
      "grad_norm": 1.6448280811309814,
      "learning_rate": 4.132305174156582e-06,
      "loss": 0.5345,
      "step": 128250
    },
    {
      "epoch": 52.5,
      "grad_norm": 1.915460228919983,
      "learning_rate": 4.1322026540480985e-06,
      "loss": 0.5638,
      "step": 128260
    },
    {
      "epoch": 52.51,
      "grad_norm": 2.5951857566833496,
      "learning_rate": 4.13210012574699e-06,
      "loss": 0.5454,
      "step": 128270
    },
    {
      "epoch": 52.51,
      "grad_norm": 2.433394193649292,
      "learning_rate": 4.131997589253727e-06,
      "loss": 0.5337,
      "step": 128280
    },
    {
      "epoch": 52.51,
      "grad_norm": 2.3359627723693848,
      "learning_rate": 4.131895044568778e-06,
      "loss": 0.537,
      "step": 128290
    },
    {
      "epoch": 52.52,
      "grad_norm": 2.113508462905884,
      "learning_rate": 4.1317924916926146e-06,
      "loss": 0.5231,
      "step": 128300
    },
    {
      "epoch": 52.52,
      "grad_norm": 2.4040236473083496,
      "learning_rate": 4.131689930625705e-06,
      "loss": 0.5453,
      "step": 128310
    },
    {
      "epoch": 52.53,
      "grad_norm": 2.2003307342529297,
      "learning_rate": 4.131587361368519e-06,
      "loss": 0.5373,
      "step": 128320
    },
    {
      "epoch": 52.53,
      "grad_norm": 1.6878474950790405,
      "learning_rate": 4.1314847839215275e-06,
      "loss": 0.5403,
      "step": 128330
    },
    {
      "epoch": 52.53,
      "grad_norm": 2.348940372467041,
      "learning_rate": 4.131382198285199e-06,
      "loss": 0.5311,
      "step": 128340
    },
    {
      "epoch": 52.54,
      "grad_norm": 2.5448758602142334,
      "learning_rate": 4.131279604460006e-06,
      "loss": 0.5346,
      "step": 128350
    },
    {
      "epoch": 52.54,
      "grad_norm": 2.3121471405029297,
      "learning_rate": 4.131177002446416e-06,
      "loss": 0.5313,
      "step": 128360
    },
    {
      "epoch": 52.55,
      "grad_norm": 2.307143211364746,
      "learning_rate": 4.131074392244899e-06,
      "loss": 0.5157,
      "step": 128370
    },
    {
      "epoch": 52.55,
      "grad_norm": 2.4454598426818848,
      "learning_rate": 4.130971773855926e-06,
      "loss": 0.5538,
      "step": 128380
    },
    {
      "epoch": 52.55,
      "grad_norm": 2.4087271690368652,
      "learning_rate": 4.130869147279967e-06,
      "loss": 0.5567,
      "step": 128390
    },
    {
      "epoch": 52.56,
      "grad_norm": 3.0033023357391357,
      "learning_rate": 4.130766512517491e-06,
      "loss": 0.5449,
      "step": 128400
    },
    {
      "epoch": 52.56,
      "grad_norm": 1.8437811136245728,
      "learning_rate": 4.13066386956897e-06,
      "loss": 0.5429,
      "step": 128410
    },
    {
      "epoch": 52.57,
      "grad_norm": 2.0784409046173096,
      "learning_rate": 4.130561218434874e-06,
      "loss": 0.5459,
      "step": 128420
    },
    {
      "epoch": 52.57,
      "grad_norm": 2.186429023742676,
      "learning_rate": 4.1304585591156715e-06,
      "loss": 0.545,
      "step": 128430
    },
    {
      "epoch": 52.57,
      "grad_norm": 2.096132755279541,
      "learning_rate": 4.130355891611834e-06,
      "loss": 0.5426,
      "step": 128440
    },
    {
      "epoch": 52.58,
      "grad_norm": 2.1507723331451416,
      "learning_rate": 4.130253215923832e-06,
      "loss": 0.5523,
      "step": 128450
    },
    {
      "epoch": 52.58,
      "grad_norm": 2.4593522548675537,
      "learning_rate": 4.130150532052135e-06,
      "loss": 0.5438,
      "step": 128460
    },
    {
      "epoch": 52.59,
      "grad_norm": 2.002850294113159,
      "learning_rate": 4.130047839997214e-06,
      "loss": 0.5525,
      "step": 128470
    },
    {
      "epoch": 52.59,
      "grad_norm": 2.2916641235351562,
      "learning_rate": 4.129945139759539e-06,
      "loss": 0.5281,
      "step": 128480
    },
    {
      "epoch": 52.6,
      "grad_norm": 2.694851875305176,
      "learning_rate": 4.1298424313395816e-06,
      "loss": 0.5587,
      "step": 128490
    },
    {
      "epoch": 52.6,
      "grad_norm": 1.8550286293029785,
      "learning_rate": 4.12973971473781e-06,
      "loss": 0.527,
      "step": 128500
    },
    {
      "epoch": 52.6,
      "grad_norm": 2.239934206008911,
      "learning_rate": 4.129636989954697e-06,
      "loss": 0.53,
      "step": 128510
    },
    {
      "epoch": 52.61,
      "grad_norm": 1.9434194564819336,
      "learning_rate": 4.129534256990712e-06,
      "loss": 0.5313,
      "step": 128520
    },
    {
      "epoch": 52.61,
      "grad_norm": 1.8038644790649414,
      "learning_rate": 4.1294315158463256e-06,
      "loss": 0.5501,
      "step": 128530
    },
    {
      "epoch": 52.62,
      "grad_norm": 1.8709787130355835,
      "learning_rate": 4.12932876652201e-06,
      "loss": 0.5593,
      "step": 128540
    },
    {
      "epoch": 52.62,
      "grad_norm": 1.7976232767105103,
      "learning_rate": 4.129226009018233e-06,
      "loss": 0.5428,
      "step": 128550
    },
    {
      "epoch": 52.62,
      "grad_norm": 2.7507002353668213,
      "learning_rate": 4.129123243335468e-06,
      "loss": 0.534,
      "step": 128560
    },
    {
      "epoch": 52.63,
      "grad_norm": 1.848792552947998,
      "learning_rate": 4.1290204694741845e-06,
      "loss": 0.5527,
      "step": 128570
    },
    {
      "epoch": 52.63,
      "grad_norm": 2.6732165813446045,
      "learning_rate": 4.128917687434853e-06,
      "loss": 0.5187,
      "step": 128580
    },
    {
      "epoch": 52.64,
      "grad_norm": 2.724902391433716,
      "learning_rate": 4.128814897217945e-06,
      "loss": 0.5411,
      "step": 128590
    },
    {
      "epoch": 52.64,
      "grad_norm": 1.7459540367126465,
      "learning_rate": 4.128712098823931e-06,
      "loss": 0.5291,
      "step": 128600
    },
    {
      "epoch": 52.64,
      "grad_norm": 1.4494379758834839,
      "learning_rate": 4.128609292253282e-06,
      "loss": 0.526,
      "step": 128610
    },
    {
      "epoch": 52.65,
      "grad_norm": 2.4241700172424316,
      "learning_rate": 4.12850647750647e-06,
      "loss": 0.5425,
      "step": 128620
    },
    {
      "epoch": 52.65,
      "grad_norm": 2.256782054901123,
      "learning_rate": 4.128403654583964e-06,
      "loss": 0.542,
      "step": 128630
    },
    {
      "epoch": 52.66,
      "grad_norm": 2.1435606479644775,
      "learning_rate": 4.128300823486236e-06,
      "loss": 0.5661,
      "step": 128640
    },
    {
      "epoch": 52.66,
      "grad_norm": 2.093785524368286,
      "learning_rate": 4.128197984213758e-06,
      "loss": 0.5493,
      "step": 128650
    },
    {
      "epoch": 52.66,
      "grad_norm": 2.6590287685394287,
      "learning_rate": 4.128095136766999e-06,
      "loss": 0.5388,
      "step": 128660
    },
    {
      "epoch": 52.67,
      "grad_norm": 2.434588670730591,
      "learning_rate": 4.127992281146431e-06,
      "loss": 0.5331,
      "step": 128670
    },
    {
      "epoch": 52.67,
      "grad_norm": 2.2718870639801025,
      "learning_rate": 4.127889417352527e-06,
      "loss": 0.5548,
      "step": 128680
    },
    {
      "epoch": 52.68,
      "grad_norm": 6.858951091766357,
      "learning_rate": 4.127786545385755e-06,
      "loss": 0.536,
      "step": 128690
    },
    {
      "epoch": 52.68,
      "grad_norm": 1.9840024709701538,
      "learning_rate": 4.12768366524659e-06,
      "loss": 0.5173,
      "step": 128700
    },
    {
      "epoch": 52.69,
      "grad_norm": 2.2652204036712646,
      "learning_rate": 4.1275807769354986e-06,
      "loss": 0.5232,
      "step": 128710
    },
    {
      "epoch": 52.69,
      "grad_norm": 2.098259210586548,
      "learning_rate": 4.127477880452956e-06,
      "loss": 0.5375,
      "step": 128720
    },
    {
      "epoch": 52.69,
      "grad_norm": 2.0957064628601074,
      "learning_rate": 4.127374975799432e-06,
      "loss": 0.5476,
      "step": 128730
    },
    {
      "epoch": 52.7,
      "grad_norm": 2.04123854637146,
      "learning_rate": 4.127272062975399e-06,
      "loss": 0.5377,
      "step": 128740
    },
    {
      "epoch": 52.7,
      "grad_norm": 1.778164267539978,
      "learning_rate": 4.127169141981327e-06,
      "loss": 0.5447,
      "step": 128750
    },
    {
      "epoch": 52.71,
      "grad_norm": 2.4177956581115723,
      "learning_rate": 4.127066212817688e-06,
      "loss": 0.5506,
      "step": 128760
    },
    {
      "epoch": 52.71,
      "grad_norm": 1.953973412513733,
      "learning_rate": 4.126963275484953e-06,
      "loss": 0.5377,
      "step": 128770
    },
    {
      "epoch": 52.71,
      "grad_norm": 1.5651710033416748,
      "learning_rate": 4.126860329983595e-06,
      "loss": 0.5264,
      "step": 128780
    },
    {
      "epoch": 52.72,
      "grad_norm": 2.4552271366119385,
      "learning_rate": 4.1267573763140845e-06,
      "loss": 0.5223,
      "step": 128790
    },
    {
      "epoch": 52.72,
      "grad_norm": 2.014594554901123,
      "learning_rate": 4.126654414476894e-06,
      "loss": 0.5329,
      "step": 128800
    },
    {
      "epoch": 52.73,
      "grad_norm": 1.7605834007263184,
      "learning_rate": 4.126551444472494e-06,
      "loss": 0.5649,
      "step": 128810
    },
    {
      "epoch": 52.73,
      "grad_norm": 2.4797232151031494,
      "learning_rate": 4.126448466301356e-06,
      "loss": 0.5292,
      "step": 128820
    },
    {
      "epoch": 52.73,
      "grad_norm": 2.1252074241638184,
      "learning_rate": 4.126345479963953e-06,
      "loss": 0.549,
      "step": 128830
    },
    {
      "epoch": 52.74,
      "grad_norm": 2.192409038543701,
      "learning_rate": 4.126242485460756e-06,
      "loss": 0.5451,
      "step": 128840
    },
    {
      "epoch": 52.74,
      "grad_norm": 1.9626829624176025,
      "learning_rate": 4.126139482792238e-06,
      "loss": 0.5321,
      "step": 128850
    },
    {
      "epoch": 52.75,
      "grad_norm": 1.5427162647247314,
      "learning_rate": 4.126036471958868e-06,
      "loss": 0.5515,
      "step": 128860
    },
    {
      "epoch": 52.75,
      "grad_norm": 3.0285654067993164,
      "learning_rate": 4.125933452961121e-06,
      "loss": 0.5486,
      "step": 128870
    },
    {
      "epoch": 52.75,
      "grad_norm": 1.8968013525009155,
      "learning_rate": 4.125830425799467e-06,
      "loss": 0.5478,
      "step": 128880
    },
    {
      "epoch": 52.76,
      "grad_norm": 2.2404487133026123,
      "learning_rate": 4.12572739047438e-06,
      "loss": 0.5467,
      "step": 128890
    },
    {
      "epoch": 52.76,
      "grad_norm": 1.3559801578521729,
      "learning_rate": 4.125624346986329e-06,
      "loss": 0.5384,
      "step": 128900
    },
    {
      "epoch": 52.77,
      "grad_norm": 1.9235140085220337,
      "learning_rate": 4.125521295335788e-06,
      "loss": 0.5266,
      "step": 128910
    },
    {
      "epoch": 52.77,
      "grad_norm": 1.7390902042388916,
      "learning_rate": 4.125418235523228e-06,
      "loss": 0.5072,
      "step": 128920
    },
    {
      "epoch": 52.78,
      "grad_norm": 2.0171189308166504,
      "learning_rate": 4.125315167549123e-06,
      "loss": 0.546,
      "step": 128930
    },
    {
      "epoch": 52.78,
      "grad_norm": 1.455106258392334,
      "learning_rate": 4.125212091413943e-06,
      "loss": 0.5254,
      "step": 128940
    },
    {
      "epoch": 52.78,
      "grad_norm": 2.4573872089385986,
      "learning_rate": 4.125109007118161e-06,
      "loss": 0.5319,
      "step": 128950
    },
    {
      "epoch": 52.79,
      "grad_norm": 2.454526662826538,
      "learning_rate": 4.125005914662249e-06,
      "loss": 0.5239,
      "step": 128960
    },
    {
      "epoch": 52.79,
      "grad_norm": 2.450904130935669,
      "learning_rate": 4.12490281404668e-06,
      "loss": 0.5274,
      "step": 128970
    },
    {
      "epoch": 52.8,
      "grad_norm": 1.629469394683838,
      "learning_rate": 4.124799705271926e-06,
      "loss": 0.5397,
      "step": 128980
    },
    {
      "epoch": 52.8,
      "grad_norm": 2.041963577270508,
      "learning_rate": 4.124696588338459e-06,
      "loss": 0.5596,
      "step": 128990
    },
    {
      "epoch": 52.8,
      "grad_norm": 2.7260971069335938,
      "learning_rate": 4.124593463246752e-06,
      "loss": 0.5359,
      "step": 129000
    },
    {
      "epoch": 52.81,
      "grad_norm": 2.590731143951416,
      "learning_rate": 4.124490329997276e-06,
      "loss": 0.5209,
      "step": 129010
    },
    {
      "epoch": 52.81,
      "grad_norm": 1.8595746755599976,
      "learning_rate": 4.124387188590505e-06,
      "loss": 0.5379,
      "step": 129020
    },
    {
      "epoch": 52.82,
      "grad_norm": 1.7056502103805542,
      "learning_rate": 4.12428403902691e-06,
      "loss": 0.5394,
      "step": 129030
    },
    {
      "epoch": 52.82,
      "grad_norm": 1.3596895933151245,
      "learning_rate": 4.1241808813069646e-06,
      "loss": 0.5577,
      "step": 129040
    },
    {
      "epoch": 52.82,
      "grad_norm": 1.8088551759719849,
      "learning_rate": 4.124077715431141e-06,
      "loss": 0.5211,
      "step": 129050
    },
    {
      "epoch": 52.83,
      "grad_norm": 1.843775987625122,
      "learning_rate": 4.123974541399913e-06,
      "loss": 0.5513,
      "step": 129060
    },
    {
      "epoch": 52.83,
      "grad_norm": 1.5650714635849,
      "learning_rate": 4.123871359213751e-06,
      "loss": 0.5321,
      "step": 129070
    },
    {
      "epoch": 52.84,
      "grad_norm": 2.9004390239715576,
      "learning_rate": 4.123768168873129e-06,
      "loss": 0.5401,
      "step": 129080
    },
    {
      "epoch": 52.84,
      "grad_norm": 1.854828953742981,
      "learning_rate": 4.123664970378519e-06,
      "loss": 0.52,
      "step": 129090
    },
    {
      "epoch": 52.84,
      "grad_norm": 2.087582588195801,
      "learning_rate": 4.123561763730395e-06,
      "loss": 0.5276,
      "step": 129100
    },
    {
      "epoch": 52.85,
      "grad_norm": 1.831321358680725,
      "learning_rate": 4.123458548929228e-06,
      "loss": 0.5347,
      "step": 129110
    },
    {
      "epoch": 52.85,
      "grad_norm": 2.497443199157715,
      "learning_rate": 4.123355325975492e-06,
      "loss": 0.5528,
      "step": 129120
    },
    {
      "epoch": 52.86,
      "grad_norm": 2.0933420658111572,
      "learning_rate": 4.123252094869661e-06,
      "loss": 0.5536,
      "step": 129130
    },
    {
      "epoch": 52.86,
      "grad_norm": 2.3695473670959473,
      "learning_rate": 4.123148855612205e-06,
      "loss": 0.5507,
      "step": 129140
    },
    {
      "epoch": 52.87,
      "grad_norm": 2.0574824810028076,
      "learning_rate": 4.123045608203598e-06,
      "loss": 0.5422,
      "step": 129150
    },
    {
      "epoch": 52.87,
      "grad_norm": 2.1628785133361816,
      "learning_rate": 4.1229423526443145e-06,
      "loss": 0.5614,
      "step": 129160
    },
    {
      "epoch": 52.87,
      "grad_norm": 2.2817165851593018,
      "learning_rate": 4.122839088934826e-06,
      "loss": 0.5526,
      "step": 129170
    },
    {
      "epoch": 52.88,
      "grad_norm": 1.4805638790130615,
      "learning_rate": 4.122735817075606e-06,
      "loss": 0.5379,
      "step": 129180
    },
    {
      "epoch": 52.88,
      "grad_norm": 3.0586116313934326,
      "learning_rate": 4.122632537067128e-06,
      "loss": 0.5443,
      "step": 129190
    },
    {
      "epoch": 52.89,
      "grad_norm": 1.4699900150299072,
      "learning_rate": 4.1225292489098634e-06,
      "loss": 0.5094,
      "step": 129200
    },
    {
      "epoch": 52.89,
      "grad_norm": 2.112851142883301,
      "learning_rate": 4.1224259526042875e-06,
      "loss": 0.5196,
      "step": 129210
    },
    {
      "epoch": 52.89,
      "grad_norm": 1.5912171602249146,
      "learning_rate": 4.122322648150872e-06,
      "loss": 0.5346,
      "step": 129220
    },
    {
      "epoch": 52.9,
      "grad_norm": 2.094653367996216,
      "learning_rate": 4.122219335550092e-06,
      "loss": 0.5359,
      "step": 129230
    },
    {
      "epoch": 52.9,
      "grad_norm": 1.9887269735336304,
      "learning_rate": 4.122116014802417e-06,
      "loss": 0.5413,
      "step": 129240
    },
    {
      "epoch": 52.91,
      "grad_norm": 2.2644615173339844,
      "learning_rate": 4.122012685908324e-06,
      "loss": 0.5355,
      "step": 129250
    },
    {
      "epoch": 52.91,
      "grad_norm": 2.687237024307251,
      "learning_rate": 4.121909348868286e-06,
      "loss": 0.5414,
      "step": 129260
    },
    {
      "epoch": 52.91,
      "grad_norm": 1.7248551845550537,
      "learning_rate": 4.1218060036827746e-06,
      "loss": 0.525,
      "step": 129270
    },
    {
      "epoch": 52.92,
      "grad_norm": 2.206404209136963,
      "learning_rate": 4.121702650352262e-06,
      "loss": 0.5365,
      "step": 129280
    },
    {
      "epoch": 52.92,
      "grad_norm": 2.6524901390075684,
      "learning_rate": 4.121599288877227e-06,
      "loss": 0.5475,
      "step": 129290
    },
    {
      "epoch": 52.93,
      "grad_norm": 2.0358240604400635,
      "learning_rate": 4.121495919258138e-06,
      "loss": 0.5309,
      "step": 129300
    },
    {
      "epoch": 52.93,
      "grad_norm": 1.557915210723877,
      "learning_rate": 4.12139254149547e-06,
      "loss": 0.5576,
      "step": 129310
    },
    {
      "epoch": 52.93,
      "grad_norm": 1.7550997734069824,
      "learning_rate": 4.121289155589697e-06,
      "loss": 0.5491,
      "step": 129320
    },
    {
      "epoch": 52.94,
      "grad_norm": 2.444406509399414,
      "learning_rate": 4.121185761541293e-06,
      "loss": 0.5502,
      "step": 129330
    },
    {
      "epoch": 52.94,
      "grad_norm": 1.6472176313400269,
      "learning_rate": 4.121082359350731e-06,
      "loss": 0.5389,
      "step": 129340
    },
    {
      "epoch": 52.95,
      "grad_norm": 2.558274507522583,
      "learning_rate": 4.120978949018485e-06,
      "loss": 0.5513,
      "step": 129350
    },
    {
      "epoch": 52.95,
      "grad_norm": 2.1403427124023438,
      "learning_rate": 4.1208755305450265e-06,
      "loss": 0.5358,
      "step": 129360
    },
    {
      "epoch": 52.96,
      "grad_norm": 2.0075716972351074,
      "learning_rate": 4.120772103930833e-06,
      "loss": 0.5571,
      "step": 129370
    },
    {
      "epoch": 52.96,
      "grad_norm": 2.2509098052978516,
      "learning_rate": 4.120668669176376e-06,
      "loss": 0.5467,
      "step": 129380
    },
    {
      "epoch": 52.96,
      "grad_norm": 2.103726625442505,
      "learning_rate": 4.1205652262821296e-06,
      "loss": 0.5253,
      "step": 129390
    },
    {
      "epoch": 52.97,
      "grad_norm": 1.6224430799484253,
      "learning_rate": 4.120461775248568e-06,
      "loss": 0.5191,
      "step": 129400
    },
    {
      "epoch": 52.97,
      "grad_norm": 1.4951756000518799,
      "learning_rate": 4.120358316076166e-06,
      "loss": 0.5329,
      "step": 129410
    },
    {
      "epoch": 52.98,
      "grad_norm": 1.9314732551574707,
      "learning_rate": 4.120254848765395e-06,
      "loss": 0.5414,
      "step": 129420
    },
    {
      "epoch": 52.98,
      "grad_norm": 2.253376007080078,
      "learning_rate": 4.120151373316731e-06,
      "loss": 0.5386,
      "step": 129430
    },
    {
      "epoch": 52.98,
      "grad_norm": 1.7629787921905518,
      "learning_rate": 4.120047889730646e-06,
      "loss": 0.5315,
      "step": 129440
    },
    {
      "epoch": 52.99,
      "grad_norm": 1.9585024118423462,
      "learning_rate": 4.119944398007618e-06,
      "loss": 0.5171,
      "step": 129450
    },
    {
      "epoch": 52.99,
      "grad_norm": 2.4795353412628174,
      "learning_rate": 4.119840898148116e-06,
      "loss": 0.5444,
      "step": 129460
    },
    {
      "epoch": 53.0,
      "grad_norm": 1.7188283205032349,
      "learning_rate": 4.119737390152618e-06,
      "loss": 0.5391,
      "step": 129470
    },
    {
      "epoch": 53.0,
      "eval_loss": 0.5372152924537659,
      "eval_runtime": 52.013,
      "eval_samples_per_second": 66.31,
      "eval_steps_per_second": 8.306,
      "step": 129479
    },
    {
      "epoch": 53.0,
      "grad_norm": 1.418126106262207,
      "learning_rate": 4.119633874021597e-06,
      "loss": 0.5271,
      "step": 129480
    },
    {
      "epoch": 53.0,
      "grad_norm": 2.206214666366577,
      "learning_rate": 4.119530349755527e-06,
      "loss": 0.5299,
      "step": 129490
    },
    {
      "epoch": 53.01,
      "grad_norm": 1.7881218194961548,
      "learning_rate": 4.1194268173548815e-06,
      "loss": 0.5412,
      "step": 129500
    },
    {
      "epoch": 53.01,
      "grad_norm": 2.3237662315368652,
      "learning_rate": 4.119323276820135e-06,
      "loss": 0.5362,
      "step": 129510
    },
    {
      "epoch": 53.02,
      "grad_norm": 2.202423334121704,
      "learning_rate": 4.1192197281517645e-06,
      "loss": 0.5197,
      "step": 129520
    },
    {
      "epoch": 53.02,
      "grad_norm": 1.7879750728607178,
      "learning_rate": 4.119116171350241e-06,
      "loss": 0.5489,
      "step": 129530
    },
    {
      "epoch": 53.02,
      "grad_norm": 1.7757891416549683,
      "learning_rate": 4.11901260641604e-06,
      "loss": 0.5211,
      "step": 129540
    },
    {
      "epoch": 53.03,
      "grad_norm": 2.6310126781463623,
      "learning_rate": 4.1189090333496346e-06,
      "loss": 0.5292,
      "step": 129550
    },
    {
      "epoch": 53.03,
      "grad_norm": 2.139122247695923,
      "learning_rate": 4.118805452151502e-06,
      "loss": 0.5068,
      "step": 129560
    },
    {
      "epoch": 53.04,
      "grad_norm": 1.9057587385177612,
      "learning_rate": 4.118701862822115e-06,
      "loss": 0.5469,
      "step": 129570
    },
    {
      "epoch": 53.04,
      "grad_norm": 2.0544848442077637,
      "learning_rate": 4.118598265361949e-06,
      "loss": 0.5341,
      "step": 129580
    },
    {
      "epoch": 53.05,
      "grad_norm": 1.9561761617660522,
      "learning_rate": 4.1184946597714774e-06,
      "loss": 0.5433,
      "step": 129590
    },
    {
      "epoch": 53.05,
      "grad_norm": 2.200988531112671,
      "learning_rate": 4.118391046051175e-06,
      "loss": 0.5192,
      "step": 129600
    },
    {
      "epoch": 53.05,
      "grad_norm": 1.529646873474121,
      "learning_rate": 4.118287424201518e-06,
      "loss": 0.5323,
      "step": 129610
    },
    {
      "epoch": 53.06,
      "grad_norm": 2.1244425773620605,
      "learning_rate": 4.118183794222979e-06,
      "loss": 0.5466,
      "step": 129620
    },
    {
      "epoch": 53.06,
      "grad_norm": 2.050938367843628,
      "learning_rate": 4.118080156116033e-06,
      "loss": 0.5372,
      "step": 129630
    },
    {
      "epoch": 53.07,
      "grad_norm": 1.8838655948638916,
      "learning_rate": 4.117976509881156e-06,
      "loss": 0.5434,
      "step": 129640
    },
    {
      "epoch": 53.07,
      "grad_norm": 2.148404121398926,
      "learning_rate": 4.117872855518822e-06,
      "loss": 0.5314,
      "step": 129650
    },
    {
      "epoch": 53.07,
      "grad_norm": 2.0624306201934814,
      "learning_rate": 4.1177691930295065e-06,
      "loss": 0.5508,
      "step": 129660
    },
    {
      "epoch": 53.08,
      "grad_norm": 2.121849536895752,
      "learning_rate": 4.117665522413683e-06,
      "loss": 0.5385,
      "step": 129670
    },
    {
      "epoch": 53.08,
      "grad_norm": 2.144188642501831,
      "learning_rate": 4.1175618436718275e-06,
      "loss": 0.5288,
      "step": 129680
    },
    {
      "epoch": 53.09,
      "grad_norm": 2.2858972549438477,
      "learning_rate": 4.1174581568044155e-06,
      "loss": 0.5226,
      "step": 129690
    },
    {
      "epoch": 53.09,
      "grad_norm": 1.617043137550354,
      "learning_rate": 4.11735446181192e-06,
      "loss": 0.5469,
      "step": 129700
    },
    {
      "epoch": 53.09,
      "grad_norm": 2.338932514190674,
      "learning_rate": 4.1172507586948176e-06,
      "loss": 0.5175,
      "step": 129710
    },
    {
      "epoch": 53.1,
      "grad_norm": 2.335045099258423,
      "learning_rate": 4.117147047453584e-06,
      "loss": 0.5356,
      "step": 129720
    },
    {
      "epoch": 53.1,
      "grad_norm": 2.055644989013672,
      "learning_rate": 4.1170433280886905e-06,
      "loss": 0.5218,
      "step": 129730
    },
    {
      "epoch": 53.11,
      "grad_norm": 2.166411876678467,
      "learning_rate": 4.116939600600617e-06,
      "loss": 0.5473,
      "step": 129740
    },
    {
      "epoch": 53.11,
      "grad_norm": 2.084181070327759,
      "learning_rate": 4.116835864989836e-06,
      "loss": 0.5483,
      "step": 129750
    },
    {
      "epoch": 53.12,
      "grad_norm": 2.313981294631958,
      "learning_rate": 4.116732121256823e-06,
      "loss": 0.5477,
      "step": 129760
    },
    {
      "epoch": 53.12,
      "grad_norm": 2.396108865737915,
      "learning_rate": 4.116628369402054e-06,
      "loss": 0.5115,
      "step": 129770
    },
    {
      "epoch": 53.12,
      "grad_norm": 1.7525317668914795,
      "learning_rate": 4.116524609426003e-06,
      "loss": 0.5287,
      "step": 129780
    },
    {
      "epoch": 53.13,
      "grad_norm": 1.9644984006881714,
      "learning_rate": 4.116420841329147e-06,
      "loss": 0.5334,
      "step": 129790
    },
    {
      "epoch": 53.13,
      "grad_norm": 2.7552549839019775,
      "learning_rate": 4.116317065111959e-06,
      "loss": 0.5172,
      "step": 129800
    },
    {
      "epoch": 53.14,
      "grad_norm": 2.5882866382598877,
      "learning_rate": 4.116213280774917e-06,
      "loss": 0.5331,
      "step": 129810
    },
    {
      "epoch": 53.14,
      "grad_norm": 2.575160026550293,
      "learning_rate": 4.116109488318495e-06,
      "loss": 0.5269,
      "step": 129820
    },
    {
      "epoch": 53.14,
      "grad_norm": 1.8610707521438599,
      "learning_rate": 4.116005687743169e-06,
      "loss": 0.5574,
      "step": 129830
    },
    {
      "epoch": 53.15,
      "grad_norm": 2.1322567462921143,
      "learning_rate": 4.115901879049414e-06,
      "loss": 0.5357,
      "step": 129840
    },
    {
      "epoch": 53.15,
      "grad_norm": 2.07621693611145,
      "learning_rate": 4.115798062237705e-06,
      "loss": 0.5383,
      "step": 129850
    },
    {
      "epoch": 53.16,
      "grad_norm": 2.3821442127227783,
      "learning_rate": 4.115694237308519e-06,
      "loss": 0.5258,
      "step": 129860
    },
    {
      "epoch": 53.16,
      "grad_norm": 1.8344240188598633,
      "learning_rate": 4.11559040426233e-06,
      "loss": 0.5355,
      "step": 129870
    },
    {
      "epoch": 53.16,
      "grad_norm": 1.904490351676941,
      "learning_rate": 4.115486563099616e-06,
      "loss": 0.5166,
      "step": 129880
    },
    {
      "epoch": 53.17,
      "grad_norm": 1.970550775527954,
      "learning_rate": 4.115382713820849e-06,
      "loss": 0.5317,
      "step": 129890
    },
    {
      "epoch": 53.17,
      "grad_norm": 2.232191562652588,
      "learning_rate": 4.115278856426509e-06,
      "loss": 0.5292,
      "step": 129900
    },
    {
      "epoch": 53.18,
      "grad_norm": 2.3609461784362793,
      "learning_rate": 4.115174990917068e-06,
      "loss": 0.5458,
      "step": 129910
    },
    {
      "epoch": 53.18,
      "grad_norm": 1.778395414352417,
      "learning_rate": 4.115071117293005e-06,
      "loss": 0.5515,
      "step": 129920
    },
    {
      "epoch": 53.18,
      "grad_norm": 1.7120954990386963,
      "learning_rate": 4.114967235554793e-06,
      "loss": 0.5258,
      "step": 129930
    },
    {
      "epoch": 53.19,
      "grad_norm": 2.644707202911377,
      "learning_rate": 4.11486334570291e-06,
      "loss": 0.5563,
      "step": 129940
    },
    {
      "epoch": 53.19,
      "grad_norm": 1.893597960472107,
      "learning_rate": 4.114759447737832e-06,
      "loss": 0.5462,
      "step": 129950
    },
    {
      "epoch": 53.2,
      "grad_norm": 2.3769495487213135,
      "learning_rate": 4.114655541660032e-06,
      "loss": 0.5193,
      "step": 129960
    },
    {
      "epoch": 53.2,
      "grad_norm": 1.9384154081344604,
      "learning_rate": 4.1145516274699884e-06,
      "loss": 0.5222,
      "step": 129970
    },
    {
      "epoch": 53.21,
      "grad_norm": 1.8825854063034058,
      "learning_rate": 4.114447705168177e-06,
      "loss": 0.5103,
      "step": 129980
    },
    {
      "epoch": 53.21,
      "grad_norm": 1.9129047393798828,
      "learning_rate": 4.114343774755074e-06,
      "loss": 0.5261,
      "step": 129990
    },
    {
      "epoch": 53.21,
      "grad_norm": 1.5638599395751953,
      "learning_rate": 4.114239836231155e-06,
      "loss": 0.5336,
      "step": 130000
    },
    {
      "epoch": 53.22,
      "grad_norm": 2.0257294178009033,
      "learning_rate": 4.114135889596897e-06,
      "loss": 0.5413,
      "step": 130010
    },
    {
      "epoch": 53.22,
      "grad_norm": 2.3118722438812256,
      "learning_rate": 4.114031934852773e-06,
      "loss": 0.5126,
      "step": 130020
    },
    {
      "epoch": 53.23,
      "grad_norm": 2.036576747894287,
      "learning_rate": 4.113927971999264e-06,
      "loss": 0.5166,
      "step": 130030
    },
    {
      "epoch": 53.23,
      "grad_norm": 2.303623676300049,
      "learning_rate": 4.113824001036844e-06,
      "loss": 0.5384,
      "step": 130040
    },
    {
      "epoch": 53.23,
      "grad_norm": 1.594509482383728,
      "learning_rate": 4.1137200219659875e-06,
      "loss": 0.5153,
      "step": 130050
    },
    {
      "epoch": 53.24,
      "grad_norm": 1.8494830131530762,
      "learning_rate": 4.113616034787173e-06,
      "loss": 0.5179,
      "step": 130060
    },
    {
      "epoch": 53.24,
      "grad_norm": 2.2413454055786133,
      "learning_rate": 4.113512039500877e-06,
      "loss": 0.5419,
      "step": 130070
    },
    {
      "epoch": 53.25,
      "grad_norm": 1.9755288362503052,
      "learning_rate": 4.113408036107574e-06,
      "loss": 0.5457,
      "step": 130080
    },
    {
      "epoch": 53.25,
      "grad_norm": 2.021106481552124,
      "learning_rate": 4.113304024607742e-06,
      "loss": 0.5382,
      "step": 130090
    },
    {
      "epoch": 53.25,
      "grad_norm": 1.6339225769042969,
      "learning_rate": 4.113200005001858e-06,
      "loss": 0.5369,
      "step": 130100
    },
    {
      "epoch": 53.26,
      "grad_norm": 2.112360715866089,
      "learning_rate": 4.1130959772903965e-06,
      "loss": 0.5167,
      "step": 130110
    },
    {
      "epoch": 53.26,
      "grad_norm": 1.9229835271835327,
      "learning_rate": 4.112991941473836e-06,
      "loss": 0.5622,
      "step": 130120
    },
    {
      "epoch": 53.27,
      "grad_norm": 2.51758074760437,
      "learning_rate": 4.1128878975526505e-06,
      "loss": 0.525,
      "step": 130130
    },
    {
      "epoch": 53.27,
      "grad_norm": 1.3055565357208252,
      "learning_rate": 4.11278384552732e-06,
      "loss": 0.5327,
      "step": 130140
    },
    {
      "epoch": 53.27,
      "grad_norm": 1.6897886991500854,
      "learning_rate": 4.112679785398318e-06,
      "loss": 0.5491,
      "step": 130150
    },
    {
      "epoch": 53.28,
      "grad_norm": 1.9799623489379883,
      "learning_rate": 4.112575717166124e-06,
      "loss": 0.5484,
      "step": 130160
    },
    {
      "epoch": 53.28,
      "grad_norm": 1.6657739877700806,
      "learning_rate": 4.1124716408312125e-06,
      "loss": 0.551,
      "step": 130170
    },
    {
      "epoch": 53.29,
      "grad_norm": 2.576878786087036,
      "learning_rate": 4.112367556394062e-06,
      "loss": 0.5371,
      "step": 130180
    },
    {
      "epoch": 53.29,
      "grad_norm": 1.974401593208313,
      "learning_rate": 4.112263463855147e-06,
      "loss": 0.5456,
      "step": 130190
    },
    {
      "epoch": 53.3,
      "grad_norm": 2.06732177734375,
      "learning_rate": 4.112159363214947e-06,
      "loss": 0.5323,
      "step": 130200
    },
    {
      "epoch": 53.3,
      "grad_norm": 2.5701706409454346,
      "learning_rate": 4.1120552544739375e-06,
      "loss": 0.5506,
      "step": 130210
    },
    {
      "epoch": 53.3,
      "grad_norm": 1.8649768829345703,
      "learning_rate": 4.111951137632594e-06,
      "loss": 0.5419,
      "step": 130220
    },
    {
      "epoch": 53.31,
      "grad_norm": 1.8155912160873413,
      "learning_rate": 4.111847012691396e-06,
      "loss": 0.5531,
      "step": 130230
    },
    {
      "epoch": 53.31,
      "grad_norm": 2.242689371109009,
      "learning_rate": 4.11174287965082e-06,
      "loss": 0.5378,
      "step": 130240
    },
    {
      "epoch": 53.32,
      "grad_norm": 2.7665934562683105,
      "learning_rate": 4.111638738511341e-06,
      "loss": 0.5253,
      "step": 130250
    },
    {
      "epoch": 53.32,
      "grad_norm": 2.136147975921631,
      "learning_rate": 4.111534589273439e-06,
      "loss": 0.527,
      "step": 130260
    },
    {
      "epoch": 53.32,
      "grad_norm": 2.4961867332458496,
      "learning_rate": 4.1114304319375885e-06,
      "loss": 0.5388,
      "step": 130270
    },
    {
      "epoch": 53.33,
      "grad_norm": 1.816428780555725,
      "learning_rate": 4.111326266504268e-06,
      "loss": 0.5371,
      "step": 130280
    },
    {
      "epoch": 53.33,
      "grad_norm": 2.3076140880584717,
      "learning_rate": 4.111222092973955e-06,
      "loss": 0.5498,
      "step": 130290
    },
    {
      "epoch": 53.34,
      "grad_norm": 2.426457405090332,
      "learning_rate": 4.111117911347125e-06,
      "loss": 0.5579,
      "step": 130300
    },
    {
      "epoch": 53.34,
      "grad_norm": 1.5915424823760986,
      "learning_rate": 4.111013721624257e-06,
      "loss": 0.5424,
      "step": 130310
    },
    {
      "epoch": 53.34,
      "grad_norm": 1.8856333494186401,
      "learning_rate": 4.110909523805827e-06,
      "loss": 0.5497,
      "step": 130320
    },
    {
      "epoch": 53.35,
      "grad_norm": 1.6750260591506958,
      "learning_rate": 4.110805317892313e-06,
      "loss": 0.5515,
      "step": 130330
    },
    {
      "epoch": 53.35,
      "grad_norm": 1.7378233671188354,
      "learning_rate": 4.110701103884192e-06,
      "loss": 0.5485,
      "step": 130340
    },
    {
      "epoch": 53.36,
      "grad_norm": 1.786706566810608,
      "learning_rate": 4.110596881781943e-06,
      "loss": 0.5477,
      "step": 130350
    },
    {
      "epoch": 53.36,
      "grad_norm": 2.387860059738159,
      "learning_rate": 4.1104926515860405e-06,
      "loss": 0.5277,
      "step": 130360
    },
    {
      "epoch": 53.36,
      "grad_norm": 2.4245657920837402,
      "learning_rate": 4.1103884132969645e-06,
      "loss": 0.5506,
      "step": 130370
    },
    {
      "epoch": 53.37,
      "grad_norm": 1.6851359605789185,
      "learning_rate": 4.110284166915191e-06,
      "loss": 0.5414,
      "step": 130380
    },
    {
      "epoch": 53.37,
      "grad_norm": 2.7339906692504883,
      "learning_rate": 4.110179912441199e-06,
      "loss": 0.5275,
      "step": 130390
    },
    {
      "epoch": 53.38,
      "grad_norm": 1.6602234840393066,
      "learning_rate": 4.1100756498754635e-06,
      "loss": 0.527,
      "step": 130400
    },
    {
      "epoch": 53.38,
      "grad_norm": 2.2487189769744873,
      "learning_rate": 4.109971379218465e-06,
      "loss": 0.5237,
      "step": 130410
    },
    {
      "epoch": 53.39,
      "grad_norm": 1.8629957437515259,
      "learning_rate": 4.10986710047068e-06,
      "loss": 0.5416,
      "step": 130420
    },
    {
      "epoch": 53.39,
      "grad_norm": 2.3861424922943115,
      "learning_rate": 4.109762813632585e-06,
      "loss": 0.54,
      "step": 130430
    },
    {
      "epoch": 53.39,
      "grad_norm": 2.4895448684692383,
      "learning_rate": 4.1096585187046606e-06,
      "loss": 0.5349,
      "step": 130440
    },
    {
      "epoch": 53.4,
      "grad_norm": 2.0059664249420166,
      "learning_rate": 4.109554215687381e-06,
      "loss": 0.5504,
      "step": 130450
    },
    {
      "epoch": 53.4,
      "grad_norm": 2.6087005138397217,
      "learning_rate": 4.109449904581226e-06,
      "loss": 0.534,
      "step": 130460
    },
    {
      "epoch": 53.41,
      "grad_norm": 2.8210179805755615,
      "learning_rate": 4.109345585386674e-06,
      "loss": 0.5453,
      "step": 130470
    },
    {
      "epoch": 53.41,
      "grad_norm": 2.7843902111053467,
      "learning_rate": 4.109241258104202e-06,
      "loss": 0.5236,
      "step": 130480
    },
    {
      "epoch": 53.41,
      "grad_norm": 2.2338950634002686,
      "learning_rate": 4.109136922734287e-06,
      "loss": 0.5332,
      "step": 130490
    },
    {
      "epoch": 53.42,
      "grad_norm": 1.671015977859497,
      "learning_rate": 4.109032579277409e-06,
      "loss": 0.5451,
      "step": 130500
    },
    {
      "epoch": 53.42,
      "grad_norm": 1.9919427633285522,
      "learning_rate": 4.108928227734044e-06,
      "loss": 0.5203,
      "step": 130510
    },
    {
      "epoch": 53.43,
      "grad_norm": 1.9885010719299316,
      "learning_rate": 4.108823868104671e-06,
      "loss": 0.5078,
      "step": 130520
    },
    {
      "epoch": 53.43,
      "grad_norm": 1.9887489080429077,
      "learning_rate": 4.108719500389769e-06,
      "loss": 0.5396,
      "step": 130530
    },
    {
      "epoch": 53.43,
      "grad_norm": 2.6432392597198486,
      "learning_rate": 4.108615124589814e-06,
      "loss": 0.5396,
      "step": 130540
    },
    {
      "epoch": 53.44,
      "grad_norm": 1.8026330471038818,
      "learning_rate": 4.1085107407052855e-06,
      "loss": 0.532,
      "step": 130550
    },
    {
      "epoch": 53.44,
      "grad_norm": 2.3788886070251465,
      "learning_rate": 4.108406348736661e-06,
      "loss": 0.5337,
      "step": 130560
    },
    {
      "epoch": 53.45,
      "grad_norm": 2.0478527545928955,
      "learning_rate": 4.10830194868442e-06,
      "loss": 0.5271,
      "step": 130570
    },
    {
      "epoch": 53.45,
      "grad_norm": 2.079078435897827,
      "learning_rate": 4.10819754054904e-06,
      "loss": 0.5288,
      "step": 130580
    },
    {
      "epoch": 53.45,
      "grad_norm": 1.607348918914795,
      "learning_rate": 4.108093124330997e-06,
      "loss": 0.5256,
      "step": 130590
    },
    {
      "epoch": 53.46,
      "grad_norm": 2.0617966651916504,
      "learning_rate": 4.107988700030773e-06,
      "loss": 0.5298,
      "step": 130600
    },
    {
      "epoch": 53.46,
      "grad_norm": 2.342381477355957,
      "learning_rate": 4.107884267648845e-06,
      "loss": 0.512,
      "step": 130610
    },
    {
      "epoch": 53.47,
      "grad_norm": 1.5919331312179565,
      "learning_rate": 4.1077798271856904e-06,
      "loss": 0.5305,
      "step": 130620
    },
    {
      "epoch": 53.47,
      "grad_norm": 1.849762201309204,
      "learning_rate": 4.107675378641788e-06,
      "loss": 0.5206,
      "step": 130630
    },
    {
      "epoch": 53.48,
      "grad_norm": 2.177990198135376,
      "learning_rate": 4.107570922017618e-06,
      "loss": 0.5163,
      "step": 130640
    },
    {
      "epoch": 53.48,
      "grad_norm": 1.9588124752044678,
      "learning_rate": 4.107466457313656e-06,
      "loss": 0.5303,
      "step": 130650
    },
    {
      "epoch": 53.48,
      "grad_norm": 2.1081602573394775,
      "learning_rate": 4.107361984530383e-06,
      "loss": 0.5528,
      "step": 130660
    },
    {
      "epoch": 53.49,
      "grad_norm": 2.3787336349487305,
      "learning_rate": 4.107257503668276e-06,
      "loss": 0.512,
      "step": 130670
    },
    {
      "epoch": 53.49,
      "grad_norm": 2.3444631099700928,
      "learning_rate": 4.107153014727815e-06,
      "loss": 0.5207,
      "step": 130680
    },
    {
      "epoch": 53.5,
      "grad_norm": 2.5285634994506836,
      "learning_rate": 4.107048517709478e-06,
      "loss": 0.5356,
      "step": 130690
    },
    {
      "epoch": 53.5,
      "grad_norm": 2.1409623622894287,
      "learning_rate": 4.106944012613743e-06,
      "loss": 0.536,
      "step": 130700
    },
    {
      "epoch": 53.5,
      "grad_norm": 1.6789137125015259,
      "learning_rate": 4.106839499441089e-06,
      "loss": 0.5576,
      "step": 130710
    },
    {
      "epoch": 53.51,
      "grad_norm": 2.1962785720825195,
      "learning_rate": 4.106734978191996e-06,
      "loss": 0.534,
      "step": 130720
    },
    {
      "epoch": 53.51,
      "grad_norm": 1.7370569705963135,
      "learning_rate": 4.106630448866941e-06,
      "loss": 0.5557,
      "step": 130730
    },
    {
      "epoch": 53.52,
      "grad_norm": 1.8132394552230835,
      "learning_rate": 4.1065259114664035e-06,
      "loss": 0.526,
      "step": 130740
    },
    {
      "epoch": 53.52,
      "grad_norm": 1.767246127128601,
      "learning_rate": 4.106421365990863e-06,
      "loss": 0.5343,
      "step": 130750
    },
    {
      "epoch": 53.52,
      "grad_norm": 1.8113172054290771,
      "learning_rate": 4.106316812440798e-06,
      "loss": 0.5305,
      "step": 130760
    },
    {
      "epoch": 53.53,
      "grad_norm": 1.6920878887176514,
      "learning_rate": 4.106212250816688e-06,
      "loss": 0.5399,
      "step": 130770
    },
    {
      "epoch": 53.53,
      "grad_norm": 2.333723306655884,
      "learning_rate": 4.10610768111901e-06,
      "loss": 0.5432,
      "step": 130780
    },
    {
      "epoch": 53.54,
      "grad_norm": 2.2408461570739746,
      "learning_rate": 4.106003103348245e-06,
      "loss": 0.5293,
      "step": 130790
    },
    {
      "epoch": 53.54,
      "grad_norm": 1.684464693069458,
      "learning_rate": 4.105898517504872e-06,
      "loss": 0.5255,
      "step": 130800
    },
    {
      "epoch": 53.54,
      "grad_norm": 1.9940649271011353,
      "learning_rate": 4.105793923589369e-06,
      "loss": 0.5429,
      "step": 130810
    },
    {
      "epoch": 53.55,
      "grad_norm": 1.5818500518798828,
      "learning_rate": 4.105689321602216e-06,
      "loss": 0.5287,
      "step": 130820
    },
    {
      "epoch": 53.55,
      "grad_norm": 1.8626751899719238,
      "learning_rate": 4.105584711543891e-06,
      "loss": 0.5274,
      "step": 130830
    },
    {
      "epoch": 53.56,
      "grad_norm": 2.0329484939575195,
      "learning_rate": 4.105480093414875e-06,
      "loss": 0.5483,
      "step": 130840
    },
    {
      "epoch": 53.56,
      "grad_norm": 1.8919103145599365,
      "learning_rate": 4.105375467215646e-06,
      "loss": 0.5352,
      "step": 130850
    },
    {
      "epoch": 53.57,
      "grad_norm": 2.1965596675872803,
      "learning_rate": 4.105270832946683e-06,
      "loss": 0.5454,
      "step": 130860
    },
    {
      "epoch": 53.57,
      "grad_norm": 2.0983316898345947,
      "learning_rate": 4.105166190608467e-06,
      "loss": 0.5437,
      "step": 130870
    },
    {
      "epoch": 53.57,
      "grad_norm": 1.6476445198059082,
      "learning_rate": 4.105061540201475e-06,
      "loss": 0.5339,
      "step": 130880
    },
    {
      "epoch": 53.58,
      "grad_norm": 1.880295991897583,
      "learning_rate": 4.104956881726189e-06,
      "loss": 0.5476,
      "step": 130890
    },
    {
      "epoch": 53.58,
      "grad_norm": 1.5067224502563477,
      "learning_rate": 4.1048522151830864e-06,
      "loss": 0.5304,
      "step": 130900
    },
    {
      "epoch": 53.59,
      "grad_norm": 1.6013808250427246,
      "learning_rate": 4.104747540572646e-06,
      "loss": 0.5245,
      "step": 130910
    },
    {
      "epoch": 53.59,
      "grad_norm": 2.0569944381713867,
      "learning_rate": 4.10464285789535e-06,
      "loss": 0.5374,
      "step": 130920
    },
    {
      "epoch": 53.59,
      "grad_norm": 2.3344249725341797,
      "learning_rate": 4.104538167151677e-06,
      "loss": 0.5379,
      "step": 130930
    },
    {
      "epoch": 53.6,
      "grad_norm": 1.7246387004852295,
      "learning_rate": 4.104433468342105e-06,
      "loss": 0.5174,
      "step": 130940
    },
    {
      "epoch": 53.6,
      "grad_norm": 2.2781102657318115,
      "learning_rate": 4.104328761467114e-06,
      "loss": 0.5336,
      "step": 130950
    },
    {
      "epoch": 53.61,
      "grad_norm": 2.195009231567383,
      "learning_rate": 4.104224046527186e-06,
      "loss": 0.5287,
      "step": 130960
    },
    {
      "epoch": 53.61,
      "grad_norm": 3.0830750465393066,
      "learning_rate": 4.104119323522799e-06,
      "loss": 0.518,
      "step": 130970
    },
    {
      "epoch": 53.61,
      "grad_norm": 2.176765203475952,
      "learning_rate": 4.104014592454432e-06,
      "loss": 0.5412,
      "step": 130980
    },
    {
      "epoch": 53.62,
      "grad_norm": 2.612248182296753,
      "learning_rate": 4.103909853322566e-06,
      "loss": 0.536,
      "step": 130990
    },
    {
      "epoch": 53.62,
      "grad_norm": 2.3078625202178955,
      "learning_rate": 4.103805106127679e-06,
      "loss": 0.5366,
      "step": 131000
    },
    {
      "epoch": 53.63,
      "grad_norm": 1.8441004753112793,
      "learning_rate": 4.1037003508702535e-06,
      "loss": 0.522,
      "step": 131010
    },
    {
      "epoch": 53.63,
      "grad_norm": 2.0736887454986572,
      "learning_rate": 4.103595587550768e-06,
      "loss": 0.528,
      "step": 131020
    },
    {
      "epoch": 53.63,
      "grad_norm": 2.467144012451172,
      "learning_rate": 4.103490816169702e-06,
      "loss": 0.5225,
      "step": 131030
    },
    {
      "epoch": 53.64,
      "grad_norm": 1.9570574760437012,
      "learning_rate": 4.103386036727536e-06,
      "loss": 0.5226,
      "step": 131040
    },
    {
      "epoch": 53.64,
      "grad_norm": 2.7577731609344482,
      "learning_rate": 4.103281249224749e-06,
      "loss": 0.5337,
      "step": 131050
    },
    {
      "epoch": 53.65,
      "grad_norm": 1.7731448411941528,
      "learning_rate": 4.103176453661823e-06,
      "loss": 0.5412,
      "step": 131060
    },
    {
      "epoch": 53.65,
      "grad_norm": 2.0100457668304443,
      "learning_rate": 4.103071650039237e-06,
      "loss": 0.5294,
      "step": 131070
    },
    {
      "epoch": 53.66,
      "grad_norm": 2.2238128185272217,
      "learning_rate": 4.1029668383574705e-06,
      "loss": 0.5402,
      "step": 131080
    },
    {
      "epoch": 53.66,
      "grad_norm": 1.442427635192871,
      "learning_rate": 4.102862018617004e-06,
      "loss": 0.5353,
      "step": 131090
    },
    {
      "epoch": 53.66,
      "grad_norm": 1.6602896451950073,
      "learning_rate": 4.102757190818318e-06,
      "loss": 0.5354,
      "step": 131100
    },
    {
      "epoch": 53.67,
      "grad_norm": 1.8474018573760986,
      "learning_rate": 4.102652354961892e-06,
      "loss": 0.5198,
      "step": 131110
    },
    {
      "epoch": 53.67,
      "grad_norm": 1.6155773401260376,
      "learning_rate": 4.102547511048207e-06,
      "loss": 0.536,
      "step": 131120
    },
    {
      "epoch": 53.68,
      "grad_norm": 1.8934085369110107,
      "learning_rate": 4.102442659077743e-06,
      "loss": 0.5337,
      "step": 131130
    },
    {
      "epoch": 53.68,
      "grad_norm": 2.2813873291015625,
      "learning_rate": 4.1023377990509805e-06,
      "loss": 0.5465,
      "step": 131140
    },
    {
      "epoch": 53.68,
      "grad_norm": 1.8337377309799194,
      "learning_rate": 4.102232930968399e-06,
      "loss": 0.5436,
      "step": 131150
    },
    {
      "epoch": 53.69,
      "grad_norm": 2.760676145553589,
      "learning_rate": 4.102128054830481e-06,
      "loss": 0.536,
      "step": 131160
    },
    {
      "epoch": 53.69,
      "grad_norm": 2.044743061065674,
      "learning_rate": 4.102023170637704e-06,
      "loss": 0.54,
      "step": 131170
    },
    {
      "epoch": 53.7,
      "grad_norm": 1.8173608779907227,
      "learning_rate": 4.101918278390551e-06,
      "loss": 0.5151,
      "step": 131180
    },
    {
      "epoch": 53.7,
      "grad_norm": 2.004732847213745,
      "learning_rate": 4.1018133780895e-06,
      "loss": 0.5364,
      "step": 131190
    },
    {
      "epoch": 53.7,
      "grad_norm": 1.7489913702011108,
      "learning_rate": 4.101708469735034e-06,
      "loss": 0.56,
      "step": 131200
    },
    {
      "epoch": 53.71,
      "grad_norm": 2.5879251956939697,
      "learning_rate": 4.101603553327632e-06,
      "loss": 0.5241,
      "step": 131210
    },
    {
      "epoch": 53.71,
      "grad_norm": 1.741219162940979,
      "learning_rate": 4.101498628867776e-06,
      "loss": 0.5213,
      "step": 131220
    },
    {
      "epoch": 53.72,
      "grad_norm": 1.8267412185668945,
      "learning_rate": 4.101393696355945e-06,
      "loss": 0.5382,
      "step": 131230
    },
    {
      "epoch": 53.72,
      "grad_norm": 1.6962080001831055,
      "learning_rate": 4.10128875579262e-06,
      "loss": 0.5505,
      "step": 131240
    },
    {
      "epoch": 53.72,
      "grad_norm": 2.097263813018799,
      "learning_rate": 4.101183807178283e-06,
      "loss": 0.5368,
      "step": 131250
    },
    {
      "epoch": 53.73,
      "grad_norm": 2.0730388164520264,
      "learning_rate": 4.1010788505134135e-06,
      "loss": 0.5282,
      "step": 131260
    },
    {
      "epoch": 53.73,
      "grad_norm": 1.6506909132003784,
      "learning_rate": 4.100973885798492e-06,
      "loss": 0.5399,
      "step": 131270
    },
    {
      "epoch": 53.74,
      "grad_norm": 2.376579523086548,
      "learning_rate": 4.100868913034002e-06,
      "loss": 0.5331,
      "step": 131280
    },
    {
      "epoch": 53.74,
      "grad_norm": 3.142632484436035,
      "learning_rate": 4.100763932220421e-06,
      "loss": 0.5266,
      "step": 131290
    },
    {
      "epoch": 53.75,
      "grad_norm": 2.5411927700042725,
      "learning_rate": 4.100658943358231e-06,
      "loss": 0.5506,
      "step": 131300
    },
    {
      "epoch": 53.75,
      "grad_norm": 2.4266505241394043,
      "learning_rate": 4.100553946447913e-06,
      "loss": 0.5164,
      "step": 131310
    },
    {
      "epoch": 53.75,
      "grad_norm": 2.0954666137695312,
      "learning_rate": 4.100448941489949e-06,
      "loss": 0.5343,
      "step": 131320
    },
    {
      "epoch": 53.76,
      "grad_norm": 2.256610631942749,
      "learning_rate": 4.100343928484819e-06,
      "loss": 0.527,
      "step": 131330
    },
    {
      "epoch": 53.76,
      "grad_norm": 2.228660821914673,
      "learning_rate": 4.100238907433005e-06,
      "loss": 0.5481,
      "step": 131340
    },
    {
      "epoch": 53.77,
      "grad_norm": 1.7080707550048828,
      "learning_rate": 4.1001338783349865e-06,
      "loss": 0.5476,
      "step": 131350
    },
    {
      "epoch": 53.77,
      "grad_norm": 1.7870699167251587,
      "learning_rate": 4.100028841191245e-06,
      "loss": 0.5593,
      "step": 131360
    },
    {
      "epoch": 53.77,
      "grad_norm": 2.963419198989868,
      "learning_rate": 4.099923796002263e-06,
      "loss": 0.5286,
      "step": 131370
    },
    {
      "epoch": 53.78,
      "grad_norm": 1.4928855895996094,
      "learning_rate": 4.099818742768521e-06,
      "loss": 0.528,
      "step": 131380
    },
    {
      "epoch": 53.78,
      "grad_norm": 2.81071138381958,
      "learning_rate": 4.0997136814905e-06,
      "loss": 0.5358,
      "step": 131390
    },
    {
      "epoch": 53.79,
      "grad_norm": 2.118422269821167,
      "learning_rate": 4.099608612168681e-06,
      "loss": 0.5167,
      "step": 131400
    },
    {
      "epoch": 53.79,
      "grad_norm": 1.9199930429458618,
      "learning_rate": 4.099503534803545e-06,
      "loss": 0.5411,
      "step": 131410
    },
    {
      "epoch": 53.79,
      "grad_norm": 2.711331367492676,
      "learning_rate": 4.099398449395574e-06,
      "loss": 0.5203,
      "step": 131420
    },
    {
      "epoch": 53.8,
      "grad_norm": 1.9525789022445679,
      "learning_rate": 4.099293355945251e-06,
      "loss": 0.5352,
      "step": 131430
    },
    {
      "epoch": 53.8,
      "grad_norm": 1.6553603410720825,
      "learning_rate": 4.099188254453054e-06,
      "loss": 0.5397,
      "step": 131440
    },
    {
      "epoch": 53.81,
      "grad_norm": 2.317357301712036,
      "learning_rate": 4.099083144919466e-06,
      "loss": 0.536,
      "step": 131450
    },
    {
      "epoch": 53.81,
      "grad_norm": 1.4652787446975708,
      "learning_rate": 4.09897802734497e-06,
      "loss": 0.548,
      "step": 131460
    },
    {
      "epoch": 53.81,
      "grad_norm": 1.9934643507003784,
      "learning_rate": 4.098872901730046e-06,
      "loss": 0.5546,
      "step": 131470
    },
    {
      "epoch": 53.82,
      "grad_norm": 1.819674015045166,
      "learning_rate": 4.098767768075176e-06,
      "loss": 0.5402,
      "step": 131480
    },
    {
      "epoch": 53.82,
      "grad_norm": 1.7664133310317993,
      "learning_rate": 4.098662626380841e-06,
      "loss": 0.5604,
      "step": 131490
    },
    {
      "epoch": 53.83,
      "grad_norm": 1.6842982769012451,
      "learning_rate": 4.098557476647523e-06,
      "loss": 0.5408,
      "step": 131500
    },
    {
      "epoch": 53.83,
      "grad_norm": 2.0951077938079834,
      "learning_rate": 4.098452318875704e-06,
      "loss": 0.5359,
      "step": 131510
    },
    {
      "epoch": 53.84,
      "grad_norm": 2.1442060470581055,
      "learning_rate": 4.098347153065865e-06,
      "loss": 0.5105,
      "step": 131520
    },
    {
      "epoch": 53.84,
      "grad_norm": 1.755609154701233,
      "learning_rate": 4.098241979218489e-06,
      "loss": 0.5614,
      "step": 131530
    },
    {
      "epoch": 53.84,
      "grad_norm": 2.6256051063537598,
      "learning_rate": 4.0981367973340556e-06,
      "loss": 0.5559,
      "step": 131540
    },
    {
      "epoch": 53.85,
      "grad_norm": 2.263367176055908,
      "learning_rate": 4.098031607413049e-06,
      "loss": 0.531,
      "step": 131550
    },
    {
      "epoch": 53.85,
      "grad_norm": 2.1843032836914062,
      "learning_rate": 4.09792640945595e-06,
      "loss": 0.5407,
      "step": 131560
    },
    {
      "epoch": 53.86,
      "grad_norm": 1.4728186130523682,
      "learning_rate": 4.09782120346324e-06,
      "loss": 0.5291,
      "step": 131570
    },
    {
      "epoch": 53.86,
      "grad_norm": 1.875675916671753,
      "learning_rate": 4.097715989435402e-06,
      "loss": 0.5434,
      "step": 131580
    },
    {
      "epoch": 53.86,
      "grad_norm": 1.992032527923584,
      "learning_rate": 4.0976107673729166e-06,
      "loss": 0.5277,
      "step": 131590
    },
    {
      "epoch": 53.87,
      "grad_norm": 1.47902250289917,
      "learning_rate": 4.097505537276268e-06,
      "loss": 0.5553,
      "step": 131600
    },
    {
      "epoch": 53.87,
      "grad_norm": 2.1429176330566406,
      "learning_rate": 4.097400299145936e-06,
      "loss": 0.5301,
      "step": 131610
    },
    {
      "epoch": 53.88,
      "grad_norm": 1.657186508178711,
      "learning_rate": 4.097295052982404e-06,
      "loss": 0.5285,
      "step": 131620
    },
    {
      "epoch": 53.88,
      "grad_norm": 1.9960581064224243,
      "learning_rate": 4.097189798786153e-06,
      "loss": 0.5487,
      "step": 131630
    },
    {
      "epoch": 53.88,
      "grad_norm": 2.0214545726776123,
      "learning_rate": 4.097084536557666e-06,
      "loss": 0.5463,
      "step": 131640
    },
    {
      "epoch": 53.89,
      "grad_norm": 2.566554546356201,
      "learning_rate": 4.096979266297425e-06,
      "loss": 0.5337,
      "step": 131650
    },
    {
      "epoch": 53.89,
      "grad_norm": 1.8684601783752441,
      "learning_rate": 4.096873988005913e-06,
      "loss": 0.537,
      "step": 131660
    },
    {
      "epoch": 53.9,
      "grad_norm": 1.615590214729309,
      "learning_rate": 4.096768701683611e-06,
      "loss": 0.5198,
      "step": 131670
    },
    {
      "epoch": 53.9,
      "grad_norm": 1.8027219772338867,
      "learning_rate": 4.096663407331002e-06,
      "loss": 0.5279,
      "step": 131680
    },
    {
      "epoch": 53.91,
      "grad_norm": 2.1248035430908203,
      "learning_rate": 4.096558104948568e-06,
      "loss": 0.5249,
      "step": 131690
    },
    {
      "epoch": 53.91,
      "grad_norm": 2.3004727363586426,
      "learning_rate": 4.096452794536792e-06,
      "loss": 0.5537,
      "step": 131700
    },
    {
      "epoch": 53.91,
      "grad_norm": 1.9199029207229614,
      "learning_rate": 4.096347476096155e-06,
      "loss": 0.538,
      "step": 131710
    },
    {
      "epoch": 53.92,
      "grad_norm": 2.2601587772369385,
      "learning_rate": 4.096242149627141e-06,
      "loss": 0.5328,
      "step": 131720
    },
    {
      "epoch": 53.92,
      "grad_norm": 1.674875259399414,
      "learning_rate": 4.096136815130231e-06,
      "loss": 0.5219,
      "step": 131730
    },
    {
      "epoch": 53.93,
      "grad_norm": 1.5322611331939697,
      "learning_rate": 4.09603147260591e-06,
      "loss": 0.5442,
      "step": 131740
    },
    {
      "epoch": 53.93,
      "grad_norm": 1.6339166164398193,
      "learning_rate": 4.095926122054658e-06,
      "loss": 0.5277,
      "step": 131750
    },
    {
      "epoch": 53.93,
      "grad_norm": 2.2396483421325684,
      "learning_rate": 4.095820763476958e-06,
      "loss": 0.534,
      "step": 131760
    },
    {
      "epoch": 53.94,
      "grad_norm": 1.4854429960250854,
      "learning_rate": 4.095715396873294e-06,
      "loss": 0.5432,
      "step": 131770
    },
    {
      "epoch": 53.94,
      "grad_norm": 1.6399927139282227,
      "learning_rate": 4.0956100222441475e-06,
      "loss": 0.5253,
      "step": 131780
    },
    {
      "epoch": 53.95,
      "grad_norm": 1.8175926208496094,
      "learning_rate": 4.095504639590002e-06,
      "loss": 0.5279,
      "step": 131790
    },
    {
      "epoch": 53.95,
      "grad_norm": 2.5753300189971924,
      "learning_rate": 4.095399248911339e-06,
      "loss": 0.5392,
      "step": 131800
    },
    {
      "epoch": 53.95,
      "grad_norm": 1.6052839756011963,
      "learning_rate": 4.095293850208643e-06,
      "loss": 0.5334,
      "step": 131810
    },
    {
      "epoch": 53.96,
      "grad_norm": 1.8720393180847168,
      "learning_rate": 4.0951884434823945e-06,
      "loss": 0.5483,
      "step": 131820
    },
    {
      "epoch": 53.96,
      "grad_norm": 2.540104866027832,
      "learning_rate": 4.09508302873308e-06,
      "loss": 0.5482,
      "step": 131830
    },
    {
      "epoch": 53.97,
      "grad_norm": 2.249361276626587,
      "learning_rate": 4.094977605961178e-06,
      "loss": 0.5288,
      "step": 131840
    },
    {
      "epoch": 53.97,
      "grad_norm": 1.9183067083358765,
      "learning_rate": 4.094872175167175e-06,
      "loss": 0.5387,
      "step": 131850
    },
    {
      "epoch": 53.97,
      "grad_norm": 1.7959696054458618,
      "learning_rate": 4.0947667363515515e-06,
      "loss": 0.5097,
      "step": 131860
    },
    {
      "epoch": 53.98,
      "grad_norm": 2.167626142501831,
      "learning_rate": 4.0946612895147915e-06,
      "loss": 0.5507,
      "step": 131870
    },
    {
      "epoch": 53.98,
      "grad_norm": 1.6272742748260498,
      "learning_rate": 4.094555834657378e-06,
      "loss": 0.535,
      "step": 131880
    },
    {
      "epoch": 53.99,
      "grad_norm": 3.331141471862793,
      "learning_rate": 4.094450371779795e-06,
      "loss": 0.5514,
      "step": 131890
    },
    {
      "epoch": 53.99,
      "grad_norm": 2.042294502258301,
      "learning_rate": 4.094344900882524e-06,
      "loss": 0.5258,
      "step": 131900
    },
    {
      "epoch": 54.0,
      "grad_norm": 2.88619065284729,
      "learning_rate": 4.094239421966049e-06,
      "loss": 0.5201,
      "step": 131910
    },
    {
      "epoch": 54.0,
      "grad_norm": 1.7992806434631348,
      "learning_rate": 4.0941339350308535e-06,
      "loss": 0.5361,
      "step": 131920
    },
    {
      "epoch": 54.0,
      "eval_loss": 0.5362539887428284,
      "eval_runtime": 52.8387,
      "eval_samples_per_second": 65.274,
      "eval_steps_per_second": 8.176,
      "step": 131922
    },
    {
      "epoch": 54.0,
      "grad_norm": 1.5607244968414307,
      "learning_rate": 4.09402844007742e-06,
      "loss": 0.5461,
      "step": 131930
    },
    {
      "epoch": 54.01,
      "grad_norm": 2.5603160858154297,
      "learning_rate": 4.093922937106232e-06,
      "loss": 0.5229,
      "step": 131940
    },
    {
      "epoch": 54.01,
      "grad_norm": 2.4396049976348877,
      "learning_rate": 4.093817426117773e-06,
      "loss": 0.5347,
      "step": 131950
    },
    {
      "epoch": 54.02,
      "grad_norm": 1.9233179092407227,
      "learning_rate": 4.093711907112526e-06,
      "loss": 0.5358,
      "step": 131960
    },
    {
      "epoch": 54.02,
      "grad_norm": 1.9937955141067505,
      "learning_rate": 4.093606380090974e-06,
      "loss": 0.5345,
      "step": 131970
    },
    {
      "epoch": 54.02,
      "grad_norm": 1.7645889520645142,
      "learning_rate": 4.093500845053601e-06,
      "loss": 0.5304,
      "step": 131980
    },
    {
      "epoch": 54.03,
      "grad_norm": 2.3067638874053955,
      "learning_rate": 4.093395302000891e-06,
      "loss": 0.5318,
      "step": 131990
    },
    {
      "epoch": 54.03,
      "grad_norm": 2.5835325717926025,
      "learning_rate": 4.093289750933327e-06,
      "loss": 0.5283,
      "step": 132000
    },
    {
      "epoch": 54.04,
      "grad_norm": 2.305009126663208,
      "learning_rate": 4.093184191851392e-06,
      "loss": 0.5493,
      "step": 132010
    },
    {
      "epoch": 54.04,
      "grad_norm": 1.9420347213745117,
      "learning_rate": 4.09307862475557e-06,
      "loss": 0.5287,
      "step": 132020
    },
    {
      "epoch": 54.04,
      "grad_norm": 1.9856210947036743,
      "learning_rate": 4.092973049646343e-06,
      "loss": 0.5494,
      "step": 132030
    },
    {
      "epoch": 54.05,
      "grad_norm": 2.31531023979187,
      "learning_rate": 4.092867466524199e-06,
      "loss": 0.5477,
      "step": 132040
    },
    {
      "epoch": 54.05,
      "grad_norm": 1.8805971145629883,
      "learning_rate": 4.092761875389616e-06,
      "loss": 0.5161,
      "step": 132050
    },
    {
      "epoch": 54.06,
      "grad_norm": 1.8470211029052734,
      "learning_rate": 4.092656276243082e-06,
      "loss": 0.5416,
      "step": 132060
    },
    {
      "epoch": 54.06,
      "grad_norm": 2.216646909713745,
      "learning_rate": 4.092550669085079e-06,
      "loss": 0.5219,
      "step": 132070
    },
    {
      "epoch": 54.06,
      "grad_norm": 2.0311903953552246,
      "learning_rate": 4.0924450539160905e-06,
      "loss": 0.5288,
      "step": 132080
    },
    {
      "epoch": 54.07,
      "grad_norm": 2.4247989654541016,
      "learning_rate": 4.092339430736602e-06,
      "loss": 0.5205,
      "step": 132090
    },
    {
      "epoch": 54.07,
      "grad_norm": 2.1639010906219482,
      "learning_rate": 4.092233799547095e-06,
      "loss": 0.5482,
      "step": 132100
    },
    {
      "epoch": 54.08,
      "grad_norm": 2.2134876251220703,
      "learning_rate": 4.092128160348054e-06,
      "loss": 0.5371,
      "step": 132110
    },
    {
      "epoch": 54.08,
      "grad_norm": 1.7724883556365967,
      "learning_rate": 4.092022513139964e-06,
      "loss": 0.5279,
      "step": 132120
    },
    {
      "epoch": 54.09,
      "grad_norm": 2.2667973041534424,
      "learning_rate": 4.091916857923309e-06,
      "loss": 0.522,
      "step": 132130
    },
    {
      "epoch": 54.09,
      "grad_norm": 2.7037994861602783,
      "learning_rate": 4.091811194698572e-06,
      "loss": 0.5381,
      "step": 132140
    },
    {
      "epoch": 54.09,
      "grad_norm": 1.616349697113037,
      "learning_rate": 4.091705523466237e-06,
      "loss": 0.5448,
      "step": 132150
    },
    {
      "epoch": 54.1,
      "grad_norm": 1.7340495586395264,
      "learning_rate": 4.091599844226789e-06,
      "loss": 0.5094,
      "step": 132160
    },
    {
      "epoch": 54.1,
      "grad_norm": 2.538762092590332,
      "learning_rate": 4.091494156980711e-06,
      "loss": 0.5346,
      "step": 132170
    },
    {
      "epoch": 54.11,
      "grad_norm": 2.4858627319335938,
      "learning_rate": 4.091388461728488e-06,
      "loss": 0.5389,
      "step": 132180
    },
    {
      "epoch": 54.11,
      "grad_norm": 1.641795039176941,
      "learning_rate": 4.091282758470604e-06,
      "loss": 0.5252,
      "step": 132190
    },
    {
      "epoch": 54.11,
      "grad_norm": 1.9453188180923462,
      "learning_rate": 4.091177047207543e-06,
      "loss": 0.5365,
      "step": 132200
    },
    {
      "epoch": 54.12,
      "grad_norm": 2.794846773147583,
      "learning_rate": 4.091071327939789e-06,
      "loss": 0.5056,
      "step": 132210
    },
    {
      "epoch": 54.12,
      "grad_norm": 2.1459693908691406,
      "learning_rate": 4.090965600667827e-06,
      "loss": 0.5392,
      "step": 132220
    },
    {
      "epoch": 54.13,
      "grad_norm": 2.622171640396118,
      "learning_rate": 4.09085986539214e-06,
      "loss": 0.5413,
      "step": 132230
    },
    {
      "epoch": 54.13,
      "grad_norm": 1.9678378105163574,
      "learning_rate": 4.0907541221132145e-06,
      "loss": 0.5261,
      "step": 132240
    },
    {
      "epoch": 54.13,
      "grad_norm": 2.0295372009277344,
      "learning_rate": 4.090648370831533e-06,
      "loss": 0.5291,
      "step": 132250
    },
    {
      "epoch": 54.14,
      "grad_norm": 1.7830121517181396,
      "learning_rate": 4.090542611547581e-06,
      "loss": 0.5318,
      "step": 132260
    },
    {
      "epoch": 54.14,
      "grad_norm": 1.9694373607635498,
      "learning_rate": 4.090436844261843e-06,
      "loss": 0.518,
      "step": 132270
    },
    {
      "epoch": 54.15,
      "grad_norm": 2.9264159202575684,
      "learning_rate": 4.090331068974802e-06,
      "loss": 0.5544,
      "step": 132280
    },
    {
      "epoch": 54.15,
      "grad_norm": 1.7573243379592896,
      "learning_rate": 4.090225285686944e-06,
      "loss": 0.5378,
      "step": 132290
    },
    {
      "epoch": 54.15,
      "grad_norm": 2.095101833343506,
      "learning_rate": 4.090119494398753e-06,
      "loss": 0.5261,
      "step": 132300
    },
    {
      "epoch": 54.16,
      "grad_norm": 1.9145665168762207,
      "learning_rate": 4.090013695110715e-06,
      "loss": 0.5503,
      "step": 132310
    },
    {
      "epoch": 54.16,
      "grad_norm": 2.037325620651245,
      "learning_rate": 4.0899078878233126e-06,
      "loss": 0.5191,
      "step": 132320
    },
    {
      "epoch": 54.17,
      "grad_norm": 2.2786028385162354,
      "learning_rate": 4.089802072537031e-06,
      "loss": 0.5506,
      "step": 132330
    },
    {
      "epoch": 54.17,
      "grad_norm": 1.811763048171997,
      "learning_rate": 4.089696249252355e-06,
      "loss": 0.5209,
      "step": 132340
    },
    {
      "epoch": 54.18,
      "grad_norm": 1.9078130722045898,
      "learning_rate": 4.089590417969771e-06,
      "loss": 0.5194,
      "step": 132350
    },
    {
      "epoch": 54.18,
      "grad_norm": 2.1736810207366943,
      "learning_rate": 4.089484578689761e-06,
      "loss": 0.5436,
      "step": 132360
    },
    {
      "epoch": 54.18,
      "grad_norm": 2.719566583633423,
      "learning_rate": 4.089378731412812e-06,
      "loss": 0.5441,
      "step": 132370
    },
    {
      "epoch": 54.19,
      "grad_norm": 2.1934287548065186,
      "learning_rate": 4.089272876139408e-06,
      "loss": 0.5434,
      "step": 132380
    },
    {
      "epoch": 54.19,
      "grad_norm": 2.049351930618286,
      "learning_rate": 4.089167012870034e-06,
      "loss": 0.5232,
      "step": 132390
    },
    {
      "epoch": 54.2,
      "grad_norm": 2.1897382736206055,
      "learning_rate": 4.0890611416051745e-06,
      "loss": 0.5236,
      "step": 132400
    },
    {
      "epoch": 54.2,
      "grad_norm": 2.381525754928589,
      "learning_rate": 4.088955262345316e-06,
      "loss": 0.5107,
      "step": 132410
    },
    {
      "epoch": 54.2,
      "grad_norm": 2.29940128326416,
      "learning_rate": 4.0888493750909415e-06,
      "loss": 0.5107,
      "step": 132420
    },
    {
      "epoch": 54.21,
      "grad_norm": 2.2307653427124023,
      "learning_rate": 4.088743479842538e-06,
      "loss": 0.5353,
      "step": 132430
    },
    {
      "epoch": 54.21,
      "grad_norm": 2.0094447135925293,
      "learning_rate": 4.088637576600588e-06,
      "loss": 0.5192,
      "step": 132440
    },
    {
      "epoch": 54.22,
      "grad_norm": 1.9600239992141724,
      "learning_rate": 4.08853166536558e-06,
      "loss": 0.5423,
      "step": 132450
    },
    {
      "epoch": 54.22,
      "grad_norm": 1.8489822149276733,
      "learning_rate": 4.088425746137997e-06,
      "loss": 0.5581,
      "step": 132460
    },
    {
      "epoch": 54.22,
      "grad_norm": 2.2629213333129883,
      "learning_rate": 4.0883198189183235e-06,
      "loss": 0.5251,
      "step": 132470
    },
    {
      "epoch": 54.23,
      "grad_norm": 2.110200881958008,
      "learning_rate": 4.088213883707047e-06,
      "loss": 0.5163,
      "step": 132480
    },
    {
      "epoch": 54.23,
      "grad_norm": 1.543508529663086,
      "learning_rate": 4.0881079405046515e-06,
      "loss": 0.5208,
      "step": 132490
    },
    {
      "epoch": 54.24,
      "grad_norm": 1.7613426446914673,
      "learning_rate": 4.088001989311622e-06,
      "loss": 0.5125,
      "step": 132500
    },
    {
      "epoch": 54.24,
      "grad_norm": 2.1860663890838623,
      "learning_rate": 4.0878960301284446e-06,
      "loss": 0.5716,
      "step": 132510
    },
    {
      "epoch": 54.24,
      "grad_norm": 2.0222525596618652,
      "learning_rate": 4.087790062955604e-06,
      "loss": 0.5486,
      "step": 132520
    },
    {
      "epoch": 54.25,
      "grad_norm": 2.2513506412506104,
      "learning_rate": 4.087684087793586e-06,
      "loss": 0.5358,
      "step": 132530
    },
    {
      "epoch": 54.25,
      "grad_norm": 3.2749197483062744,
      "learning_rate": 4.087578104642877e-06,
      "loss": 0.5483,
      "step": 132540
    },
    {
      "epoch": 54.26,
      "grad_norm": 1.6074497699737549,
      "learning_rate": 4.087472113503959e-06,
      "loss": 0.5349,
      "step": 132550
    },
    {
      "epoch": 54.26,
      "grad_norm": 2.273597478866577,
      "learning_rate": 4.087366114377323e-06,
      "loss": 0.5486,
      "step": 132560
    },
    {
      "epoch": 54.27,
      "grad_norm": 1.8053004741668701,
      "learning_rate": 4.08726010726345e-06,
      "loss": 0.5499,
      "step": 132570
    },
    {
      "epoch": 54.27,
      "grad_norm": 1.8421564102172852,
      "learning_rate": 4.087154092162827e-06,
      "loss": 0.5385,
      "step": 132580
    },
    {
      "epoch": 54.27,
      "grad_norm": 1.637410283088684,
      "learning_rate": 4.087048069075941e-06,
      "loss": 0.5401,
      "step": 132590
    },
    {
      "epoch": 54.28,
      "grad_norm": 1.686780571937561,
      "learning_rate": 4.086942038003276e-06,
      "loss": 0.5354,
      "step": 132600
    },
    {
      "epoch": 54.28,
      "grad_norm": 2.001370668411255,
      "learning_rate": 4.086835998945318e-06,
      "loss": 0.5512,
      "step": 132610
    },
    {
      "epoch": 54.29,
      "grad_norm": 2.334343433380127,
      "learning_rate": 4.0867299519025535e-06,
      "loss": 0.5271,
      "step": 132620
    },
    {
      "epoch": 54.29,
      "grad_norm": 2.2855708599090576,
      "learning_rate": 4.086623896875467e-06,
      "loss": 0.5538,
      "step": 132630
    },
    {
      "epoch": 54.29,
      "grad_norm": 1.8451474905014038,
      "learning_rate": 4.086517833864547e-06,
      "loss": 0.5264,
      "step": 132640
    },
    {
      "epoch": 54.3,
      "grad_norm": 2.0867700576782227,
      "learning_rate": 4.086411762870275e-06,
      "loss": 0.5296,
      "step": 132650
    },
    {
      "epoch": 54.3,
      "grad_norm": 2.1457343101501465,
      "learning_rate": 4.086305683893141e-06,
      "loss": 0.5304,
      "step": 132660
    },
    {
      "epoch": 54.31,
      "grad_norm": 2.399653434753418,
      "learning_rate": 4.086199596933628e-06,
      "loss": 0.5579,
      "step": 132670
    },
    {
      "epoch": 54.31,
      "grad_norm": 2.3101134300231934,
      "learning_rate": 4.086093501992224e-06,
      "loss": 0.5408,
      "step": 132680
    },
    {
      "epoch": 54.31,
      "grad_norm": 2.627755880355835,
      "learning_rate": 4.085987399069414e-06,
      "loss": 0.5399,
      "step": 132690
    },
    {
      "epoch": 54.32,
      "grad_norm": 2.1456449031829834,
      "learning_rate": 4.085881288165685e-06,
      "loss": 0.534,
      "step": 132700
    },
    {
      "epoch": 54.32,
      "grad_norm": 2.343768835067749,
      "learning_rate": 4.0857751692815215e-06,
      "loss": 0.5251,
      "step": 132710
    },
    {
      "epoch": 54.33,
      "grad_norm": 2.2711267471313477,
      "learning_rate": 4.085669042417411e-06,
      "loss": 0.5331,
      "step": 132720
    },
    {
      "epoch": 54.33,
      "grad_norm": 2.1992857456207275,
      "learning_rate": 4.085562907573838e-06,
      "loss": 0.5162,
      "step": 132730
    },
    {
      "epoch": 54.33,
      "grad_norm": 1.9983510971069336,
      "learning_rate": 4.0854567647512915e-06,
      "loss": 0.5265,
      "step": 132740
    },
    {
      "epoch": 54.34,
      "grad_norm": 1.8390541076660156,
      "learning_rate": 4.085350613950256e-06,
      "loss": 0.5509,
      "step": 132750
    },
    {
      "epoch": 54.34,
      "grad_norm": 1.7842164039611816,
      "learning_rate": 4.085244455171216e-06,
      "loss": 0.5417,
      "step": 132760
    },
    {
      "epoch": 54.35,
      "grad_norm": 2.2504842281341553,
      "learning_rate": 4.085138288414661e-06,
      "loss": 0.5473,
      "step": 132770
    },
    {
      "epoch": 54.35,
      "grad_norm": 1.8920990228652954,
      "learning_rate": 4.085032113681076e-06,
      "loss": 0.5469,
      "step": 132780
    },
    {
      "epoch": 54.36,
      "grad_norm": 1.8297216892242432,
      "learning_rate": 4.0849259309709466e-06,
      "loss": 0.5134,
      "step": 132790
    },
    {
      "epoch": 54.36,
      "grad_norm": 1.8035697937011719,
      "learning_rate": 4.084819740284761e-06,
      "loss": 0.5188,
      "step": 132800
    },
    {
      "epoch": 54.36,
      "grad_norm": 1.5295602083206177,
      "learning_rate": 4.084713541623004e-06,
      "loss": 0.5241,
      "step": 132810
    },
    {
      "epoch": 54.37,
      "grad_norm": 2.0473248958587646,
      "learning_rate": 4.084607334986162e-06,
      "loss": 0.5533,
      "step": 132820
    },
    {
      "epoch": 54.37,
      "grad_norm": 2.2024986743927,
      "learning_rate": 4.0845011203747226e-06,
      "loss": 0.5363,
      "step": 132830
    },
    {
      "epoch": 54.38,
      "grad_norm": 1.5544005632400513,
      "learning_rate": 4.084394897789172e-06,
      "loss": 0.5241,
      "step": 132840
    },
    {
      "epoch": 54.38,
      "grad_norm": 1.7530572414398193,
      "learning_rate": 4.084288667229997e-06,
      "loss": 0.5316,
      "step": 132850
    },
    {
      "epoch": 54.38,
      "grad_norm": 2.2396254539489746,
      "learning_rate": 4.084182428697682e-06,
      "loss": 0.5519,
      "step": 132860
    },
    {
      "epoch": 54.39,
      "grad_norm": 1.8237143754959106,
      "learning_rate": 4.0840761821927176e-06,
      "loss": 0.5484,
      "step": 132870
    },
    {
      "epoch": 54.39,
      "grad_norm": 1.8381891250610352,
      "learning_rate": 4.083969927715589e-06,
      "loss": 0.5572,
      "step": 132880
    },
    {
      "epoch": 54.4,
      "grad_norm": 1.6406580209732056,
      "learning_rate": 4.083863665266781e-06,
      "loss": 0.5337,
      "step": 132890
    },
    {
      "epoch": 54.4,
      "grad_norm": 1.8359028100967407,
      "learning_rate": 4.083757394846782e-06,
      "loss": 0.5655,
      "step": 132900
    },
    {
      "epoch": 54.4,
      "grad_norm": 2.3635377883911133,
      "learning_rate": 4.083651116456079e-06,
      "loss": 0.5327,
      "step": 132910
    },
    {
      "epoch": 54.41,
      "grad_norm": 1.8841079473495483,
      "learning_rate": 4.083544830095158e-06,
      "loss": 0.5377,
      "step": 132920
    },
    {
      "epoch": 54.41,
      "grad_norm": 2.105937957763672,
      "learning_rate": 4.083438535764507e-06,
      "loss": 0.5305,
      "step": 132930
    },
    {
      "epoch": 54.42,
      "grad_norm": 2.7389109134674072,
      "learning_rate": 4.083332233464612e-06,
      "loss": 0.5315,
      "step": 132940
    },
    {
      "epoch": 54.42,
      "grad_norm": 1.6071257591247559,
      "learning_rate": 4.08322592319596e-06,
      "loss": 0.5365,
      "step": 132950
    },
    {
      "epoch": 54.42,
      "grad_norm": 2.3018951416015625,
      "learning_rate": 4.083119604959038e-06,
      "loss": 0.5144,
      "step": 132960
    },
    {
      "epoch": 54.43,
      "grad_norm": 1.8000035285949707,
      "learning_rate": 4.0830132787543335e-06,
      "loss": 0.5285,
      "step": 132970
    },
    {
      "epoch": 54.43,
      "grad_norm": 1.898988962173462,
      "learning_rate": 4.0829069445823336e-06,
      "loss": 0.5338,
      "step": 132980
    },
    {
      "epoch": 54.44,
      "grad_norm": 2.300812244415283,
      "learning_rate": 4.082800602443524e-06,
      "loss": 0.538,
      "step": 132990
    },
    {
      "epoch": 54.44,
      "grad_norm": 2.206430196762085,
      "learning_rate": 4.082694252338394e-06,
      "loss": 0.5094,
      "step": 133000
    },
    {
      "epoch": 54.45,
      "grad_norm": 2.8352644443511963,
      "learning_rate": 4.0825878942674295e-06,
      "loss": 0.5287,
      "step": 133010
    },
    {
      "epoch": 54.45,
      "grad_norm": 1.8631689548492432,
      "learning_rate": 4.082481528231117e-06,
      "loss": 0.5327,
      "step": 133020
    },
    {
      "epoch": 54.45,
      "grad_norm": 2.080360174179077,
      "learning_rate": 4.082375154229946e-06,
      "loss": 0.5241,
      "step": 133030
    },
    {
      "epoch": 54.46,
      "grad_norm": 2.156557559967041,
      "learning_rate": 4.082268772264402e-06,
      "loss": 0.5277,
      "step": 133040
    },
    {
      "epoch": 54.46,
      "grad_norm": 2.5027637481689453,
      "learning_rate": 4.0821623823349725e-06,
      "loss": 0.5405,
      "step": 133050
    },
    {
      "epoch": 54.47,
      "grad_norm": 1.6393911838531494,
      "learning_rate": 4.082055984442145e-06,
      "loss": 0.5217,
      "step": 133060
    },
    {
      "epoch": 54.47,
      "grad_norm": 1.9940671920776367,
      "learning_rate": 4.081949578586408e-06,
      "loss": 0.5713,
      "step": 133070
    },
    {
      "epoch": 54.47,
      "grad_norm": 2.112793207168579,
      "learning_rate": 4.081843164768246e-06,
      "loss": 0.5478,
      "step": 133080
    },
    {
      "epoch": 54.48,
      "grad_norm": 2.759962320327759,
      "learning_rate": 4.081736742988149e-06,
      "loss": 0.5449,
      "step": 133090
    },
    {
      "epoch": 54.48,
      "grad_norm": 2.1210713386535645,
      "learning_rate": 4.0816303132466045e-06,
      "loss": 0.5454,
      "step": 133100
    },
    {
      "epoch": 54.49,
      "grad_norm": 2.0251259803771973,
      "learning_rate": 4.081523875544099e-06,
      "loss": 0.5272,
      "step": 133110
    },
    {
      "epoch": 54.49,
      "grad_norm": 2.3555827140808105,
      "learning_rate": 4.081417429881122e-06,
      "loss": 0.5563,
      "step": 133120
    },
    {
      "epoch": 54.49,
      "grad_norm": 1.7611894607543945,
      "learning_rate": 4.081310976258157e-06,
      "loss": 0.543,
      "step": 133130
    },
    {
      "epoch": 54.5,
      "grad_norm": 2.5128424167633057,
      "learning_rate": 4.081204514675697e-06,
      "loss": 0.5425,
      "step": 133140
    },
    {
      "epoch": 54.5,
      "grad_norm": 1.973768949508667,
      "learning_rate": 4.081098045134225e-06,
      "loss": 0.5123,
      "step": 133150
    },
    {
      "epoch": 54.51,
      "grad_norm": 1.7009539604187012,
      "learning_rate": 4.080991567634231e-06,
      "loss": 0.5248,
      "step": 133160
    },
    {
      "epoch": 54.51,
      "grad_norm": 1.9691892862319946,
      "learning_rate": 4.080885082176202e-06,
      "loss": 0.5379,
      "step": 133170
    },
    {
      "epoch": 54.51,
      "grad_norm": 1.942969560623169,
      "learning_rate": 4.080778588760629e-06,
      "loss": 0.5138,
      "step": 133180
    },
    {
      "epoch": 54.52,
      "grad_norm": 1.8666118383407593,
      "learning_rate": 4.080672087387993e-06,
      "loss": 0.5412,
      "step": 133190
    },
    {
      "epoch": 54.52,
      "grad_norm": 1.6690406799316406,
      "learning_rate": 4.080565578058789e-06,
      "loss": 0.5329,
      "step": 133200
    },
    {
      "epoch": 54.53,
      "grad_norm": 2.568164110183716,
      "learning_rate": 4.080459060773501e-06,
      "loss": 0.5289,
      "step": 133210
    },
    {
      "epoch": 54.53,
      "grad_norm": 1.7784686088562012,
      "learning_rate": 4.080352535532617e-06,
      "loss": 0.5201,
      "step": 133220
    },
    {
      "epoch": 54.54,
      "grad_norm": 2.1463520526885986,
      "learning_rate": 4.080246002336626e-06,
      "loss": 0.5265,
      "step": 133230
    },
    {
      "epoch": 54.54,
      "grad_norm": 1.964368462562561,
      "learning_rate": 4.080139461186016e-06,
      "loss": 0.5252,
      "step": 133240
    },
    {
      "epoch": 54.54,
      "grad_norm": 1.7985502481460571,
      "learning_rate": 4.080032912081275e-06,
      "loss": 0.5212,
      "step": 133250
    },
    {
      "epoch": 54.55,
      "grad_norm": 2.7908377647399902,
      "learning_rate": 4.07992635502289e-06,
      "loss": 0.515,
      "step": 133260
    },
    {
      "epoch": 54.55,
      "grad_norm": 2.336662530899048,
      "learning_rate": 4.0798197900113506e-06,
      "loss": 0.5107,
      "step": 133270
    },
    {
      "epoch": 54.56,
      "grad_norm": 1.9692816734313965,
      "learning_rate": 4.079713217047144e-06,
      "loss": 0.5401,
      "step": 133280
    },
    {
      "epoch": 54.56,
      "grad_norm": 2.0603551864624023,
      "learning_rate": 4.07960663613076e-06,
      "loss": 0.5323,
      "step": 133290
    },
    {
      "epoch": 54.56,
      "grad_norm": 1.792649745941162,
      "learning_rate": 4.079500047262684e-06,
      "loss": 0.5089,
      "step": 133300
    },
    {
      "epoch": 54.57,
      "grad_norm": 1.571146011352539,
      "learning_rate": 4.0793934504434065e-06,
      "loss": 0.5393,
      "step": 133310
    },
    {
      "epoch": 54.57,
      "grad_norm": 1.8739310503005981,
      "learning_rate": 4.0792868456734154e-06,
      "loss": 0.5397,
      "step": 133320
    },
    {
      "epoch": 54.58,
      "grad_norm": 2.5302069187164307,
      "learning_rate": 4.0791802329531985e-06,
      "loss": 0.5497,
      "step": 133330
    },
    {
      "epoch": 54.58,
      "grad_norm": 2.1350114345550537,
      "learning_rate": 4.0790736122832445e-06,
      "loss": 0.5582,
      "step": 133340
    },
    {
      "epoch": 54.58,
      "grad_norm": 1.821750283241272,
      "learning_rate": 4.078966983664042e-06,
      "loss": 0.5433,
      "step": 133350
    },
    {
      "epoch": 54.59,
      "grad_norm": 1.4178102016448975,
      "learning_rate": 4.078860347096079e-06,
      "loss": 0.5228,
      "step": 133360
    },
    {
      "epoch": 54.59,
      "grad_norm": 1.8305050134658813,
      "learning_rate": 4.078753702579845e-06,
      "loss": 0.5421,
      "step": 133370
    },
    {
      "epoch": 54.6,
      "grad_norm": 2.3619613647460938,
      "learning_rate": 4.078647050115826e-06,
      "loss": 0.5374,
      "step": 133380
    },
    {
      "epoch": 54.6,
      "grad_norm": 1.9150344133377075,
      "learning_rate": 4.078540389704514e-06,
      "loss": 0.538,
      "step": 133390
    },
    {
      "epoch": 54.6,
      "grad_norm": 2.0995829105377197,
      "learning_rate": 4.0784337213463946e-06,
      "loss": 0.5333,
      "step": 133400
    },
    {
      "epoch": 54.61,
      "grad_norm": 2.0314981937408447,
      "learning_rate": 4.078327045041959e-06,
      "loss": 0.5305,
      "step": 133410
    },
    {
      "epoch": 54.61,
      "grad_norm": 1.8224544525146484,
      "learning_rate": 4.078220360791693e-06,
      "loss": 0.519,
      "step": 133420
    },
    {
      "epoch": 54.62,
      "grad_norm": 2.3127808570861816,
      "learning_rate": 4.078113668596088e-06,
      "loss": 0.5336,
      "step": 133430
    },
    {
      "epoch": 54.62,
      "grad_norm": 1.9038070440292358,
      "learning_rate": 4.078006968455631e-06,
      "loss": 0.519,
      "step": 133440
    },
    {
      "epoch": 54.63,
      "grad_norm": 2.2784974575042725,
      "learning_rate": 4.0779002603708124e-06,
      "loss": 0.5433,
      "step": 133450
    },
    {
      "epoch": 54.63,
      "grad_norm": 1.5350738763809204,
      "learning_rate": 4.07779354434212e-06,
      "loss": 0.5199,
      "step": 133460
    },
    {
      "epoch": 54.63,
      "grad_norm": 2.6689674854278564,
      "learning_rate": 4.0776868203700416e-06,
      "loss": 0.5281,
      "step": 133470
    },
    {
      "epoch": 54.64,
      "grad_norm": 2.0412213802337646,
      "learning_rate": 4.077580088455067e-06,
      "loss": 0.5409,
      "step": 133480
    },
    {
      "epoch": 54.64,
      "grad_norm": 2.2648520469665527,
      "learning_rate": 4.077473348597687e-06,
      "loss": 0.534,
      "step": 133490
    },
    {
      "epoch": 54.65,
      "grad_norm": 1.688694953918457,
      "learning_rate": 4.077366600798388e-06,
      "loss": 0.5277,
      "step": 133500
    },
    {
      "epoch": 54.65,
      "grad_norm": 1.7163445949554443,
      "learning_rate": 4.077259845057659e-06,
      "loss": 0.5305,
      "step": 133510
    },
    {
      "epoch": 54.65,
      "grad_norm": 2.0181636810302734,
      "learning_rate": 4.07715308137599e-06,
      "loss": 0.537,
      "step": 133520
    },
    {
      "epoch": 54.66,
      "grad_norm": 1.6784673929214478,
      "learning_rate": 4.077046309753871e-06,
      "loss": 0.5402,
      "step": 133530
    },
    {
      "epoch": 54.66,
      "grad_norm": 1.393716812133789,
      "learning_rate": 4.076939530191789e-06,
      "loss": 0.5244,
      "step": 133540
    },
    {
      "epoch": 54.67,
      "grad_norm": 2.3712282180786133,
      "learning_rate": 4.076832742690235e-06,
      "loss": 0.5346,
      "step": 133550
    },
    {
      "epoch": 54.67,
      "grad_norm": 1.6586114168167114,
      "learning_rate": 4.076725947249697e-06,
      "loss": 0.526,
      "step": 133560
    },
    {
      "epoch": 54.67,
      "grad_norm": 2.2151641845703125,
      "learning_rate": 4.076619143870665e-06,
      "loss": 0.5384,
      "step": 133570
    },
    {
      "epoch": 54.68,
      "grad_norm": 1.9261212348937988,
      "learning_rate": 4.076512332553627e-06,
      "loss": 0.5227,
      "step": 133580
    },
    {
      "epoch": 54.68,
      "grad_norm": 2.427041530609131,
      "learning_rate": 4.076405513299074e-06,
      "loss": 0.5414,
      "step": 133590
    },
    {
      "epoch": 54.69,
      "grad_norm": 2.0884742736816406,
      "learning_rate": 4.0762986861074935e-06,
      "loss": 0.5642,
      "step": 133600
    },
    {
      "epoch": 54.69,
      "grad_norm": 2.3020849227905273,
      "learning_rate": 4.076191850979377e-06,
      "loss": 0.5511,
      "step": 133610
    },
    {
      "epoch": 54.7,
      "grad_norm": 2.4062914848327637,
      "learning_rate": 4.076085007915211e-06,
      "loss": 0.5345,
      "step": 133620
    },
    {
      "epoch": 54.7,
      "grad_norm": 2.3907647132873535,
      "learning_rate": 4.075978156915488e-06,
      "loss": 0.5349,
      "step": 133630
    },
    {
      "epoch": 54.7,
      "grad_norm": 2.0016045570373535,
      "learning_rate": 4.075871297980695e-06,
      "loss": 0.5569,
      "step": 133640
    },
    {
      "epoch": 54.71,
      "grad_norm": 1.8497984409332275,
      "learning_rate": 4.075764431111324e-06,
      "loss": 0.5515,
      "step": 133650
    },
    {
      "epoch": 54.71,
      "grad_norm": 3.2602202892303467,
      "learning_rate": 4.075657556307862e-06,
      "loss": 0.5124,
      "step": 133660
    },
    {
      "epoch": 54.72,
      "grad_norm": 2.288120985031128,
      "learning_rate": 4.0755506735708e-06,
      "loss": 0.5445,
      "step": 133670
    },
    {
      "epoch": 54.72,
      "grad_norm": 2.2289798259735107,
      "learning_rate": 4.075443782900628e-06,
      "loss": 0.5416,
      "step": 133680
    },
    {
      "epoch": 54.72,
      "grad_norm": 1.7433102130889893,
      "learning_rate": 4.075336884297834e-06,
      "loss": 0.5327,
      "step": 133690
    },
    {
      "epoch": 54.73,
      "grad_norm": 2.5789144039154053,
      "learning_rate": 4.075229977762909e-06,
      "loss": 0.5168,
      "step": 133700
    },
    {
      "epoch": 54.73,
      "grad_norm": 2.5322017669677734,
      "learning_rate": 4.075123063296342e-06,
      "loss": 0.5537,
      "step": 133710
    },
    {
      "epoch": 54.74,
      "grad_norm": 1.4493319988250732,
      "learning_rate": 4.075016140898624e-06,
      "loss": 0.5362,
      "step": 133720
    },
    {
      "epoch": 54.74,
      "grad_norm": 1.5946317911148071,
      "learning_rate": 4.074909210570243e-06,
      "loss": 0.5195,
      "step": 133730
    },
    {
      "epoch": 54.74,
      "grad_norm": 2.479342222213745,
      "learning_rate": 4.0748022723116905e-06,
      "loss": 0.5195,
      "step": 133740
    },
    {
      "epoch": 54.75,
      "grad_norm": 1.7998663187026978,
      "learning_rate": 4.074695326123455e-06,
      "loss": 0.5379,
      "step": 133750
    },
    {
      "epoch": 54.75,
      "grad_norm": 1.7624695301055908,
      "learning_rate": 4.074588372006027e-06,
      "loss": 0.5494,
      "step": 133760
    },
    {
      "epoch": 54.76,
      "grad_norm": 1.7939856052398682,
      "learning_rate": 4.074481409959897e-06,
      "loss": 0.5284,
      "step": 133770
    },
    {
      "epoch": 54.76,
      "grad_norm": 1.638863205909729,
      "learning_rate": 4.074374439985555e-06,
      "loss": 0.5319,
      "step": 133780
    },
    {
      "epoch": 54.76,
      "grad_norm": 1.99439537525177,
      "learning_rate": 4.0742674620834895e-06,
      "loss": 0.5208,
      "step": 133790
    },
    {
      "epoch": 54.77,
      "grad_norm": 2.0504186153411865,
      "learning_rate": 4.074160476254191e-06,
      "loss": 0.5567,
      "step": 133800
    },
    {
      "epoch": 54.77,
      "grad_norm": 1.7470414638519287,
      "learning_rate": 4.074053482498151e-06,
      "loss": 0.5465,
      "step": 133810
    },
    {
      "epoch": 54.78,
      "grad_norm": 2.0260815620422363,
      "learning_rate": 4.073946480815859e-06,
      "loss": 0.5369,
      "step": 133820
    },
    {
      "epoch": 54.78,
      "grad_norm": 2.126343011856079,
      "learning_rate": 4.073839471207804e-06,
      "loss": 0.5419,
      "step": 133830
    },
    {
      "epoch": 54.79,
      "grad_norm": 2.1492481231689453,
      "learning_rate": 4.073732453674478e-06,
      "loss": 0.5306,
      "step": 133840
    },
    {
      "epoch": 54.79,
      "grad_norm": 2.146960973739624,
      "learning_rate": 4.073625428216369e-06,
      "loss": 0.5327,
      "step": 133850
    },
    {
      "epoch": 54.79,
      "grad_norm": 2.272033214569092,
      "learning_rate": 4.07351839483397e-06,
      "loss": 0.5245,
      "step": 133860
    },
    {
      "epoch": 54.8,
      "grad_norm": 2.153716564178467,
      "learning_rate": 4.073411353527769e-06,
      "loss": 0.5524,
      "step": 133870
    },
    {
      "epoch": 54.8,
      "grad_norm": 1.6957409381866455,
      "learning_rate": 4.073304304298257e-06,
      "loss": 0.5505,
      "step": 133880
    },
    {
      "epoch": 54.81,
      "grad_norm": 1.910563588142395,
      "learning_rate": 4.073197247145926e-06,
      "loss": 0.5324,
      "step": 133890
    },
    {
      "epoch": 54.81,
      "grad_norm": 2.07053279876709,
      "learning_rate": 4.073090182071264e-06,
      "loss": 0.5405,
      "step": 133900
    },
    {
      "epoch": 54.81,
      "grad_norm": 1.918337106704712,
      "learning_rate": 4.072983109074762e-06,
      "loss": 0.5358,
      "step": 133910
    },
    {
      "epoch": 54.82,
      "grad_norm": 1.675337553024292,
      "learning_rate": 4.0728760281569114e-06,
      "loss": 0.5374,
      "step": 133920
    },
    {
      "epoch": 54.82,
      "grad_norm": 1.7865301370620728,
      "learning_rate": 4.072768939318203e-06,
      "loss": 0.5497,
      "step": 133930
    },
    {
      "epoch": 54.83,
      "grad_norm": 2.0851552486419678,
      "learning_rate": 4.072661842559125e-06,
      "loss": 0.55,
      "step": 133940
    },
    {
      "epoch": 54.83,
      "grad_norm": 1.4794867038726807,
      "learning_rate": 4.072554737880171e-06,
      "loss": 0.524,
      "step": 133950
    },
    {
      "epoch": 54.83,
      "grad_norm": 1.8373342752456665,
      "learning_rate": 4.072447625281829e-06,
      "loss": 0.5451,
      "step": 133960
    },
    {
      "epoch": 54.84,
      "grad_norm": 1.9441771507263184,
      "learning_rate": 4.0723405047645924e-06,
      "loss": 0.5357,
      "step": 133970
    },
    {
      "epoch": 54.84,
      "grad_norm": 2.101915121078491,
      "learning_rate": 4.072233376328949e-06,
      "loss": 0.5236,
      "step": 133980
    },
    {
      "epoch": 54.85,
      "grad_norm": 1.4175454378128052,
      "learning_rate": 4.072126239975392e-06,
      "loss": 0.5112,
      "step": 133990
    },
    {
      "epoch": 54.85,
      "grad_norm": 2.2177207469940186,
      "learning_rate": 4.07201909570441e-06,
      "loss": 0.5269,
      "step": 134000
    },
    {
      "epoch": 54.85,
      "grad_norm": 2.0571632385253906,
      "learning_rate": 4.071911943516496e-06,
      "loss": 0.557,
      "step": 134010
    },
    {
      "epoch": 54.86,
      "grad_norm": 2.4051356315612793,
      "learning_rate": 4.07180478341214e-06,
      "loss": 0.5189,
      "step": 134020
    },
    {
      "epoch": 54.86,
      "grad_norm": 2.1116621494293213,
      "learning_rate": 4.071697615391831e-06,
      "loss": 0.5353,
      "step": 134030
    },
    {
      "epoch": 54.87,
      "grad_norm": 1.747316837310791,
      "learning_rate": 4.071590439456063e-06,
      "loss": 0.5358,
      "step": 134040
    },
    {
      "epoch": 54.87,
      "grad_norm": 1.7513960599899292,
      "learning_rate": 4.071483255605325e-06,
      "loss": 0.5217,
      "step": 134050
    },
    {
      "epoch": 54.88,
      "grad_norm": 2.186032772064209,
      "learning_rate": 4.071376063840109e-06,
      "loss": 0.533,
      "step": 134060
    },
    {
      "epoch": 54.88,
      "grad_norm": 1.5992006063461304,
      "learning_rate": 4.071268864160905e-06,
      "loss": 0.543,
      "step": 134070
    },
    {
      "epoch": 54.88,
      "grad_norm": 2.142103910446167,
      "learning_rate": 4.071161656568205e-06,
      "loss": 0.5545,
      "step": 134080
    },
    {
      "epoch": 54.89,
      "grad_norm": 2.049201726913452,
      "learning_rate": 4.071054441062499e-06,
      "loss": 0.525,
      "step": 134090
    },
    {
      "epoch": 54.89,
      "grad_norm": 1.9264851808547974,
      "learning_rate": 4.0709472176442794e-06,
      "loss": 0.5438,
      "step": 134100
    },
    {
      "epoch": 54.9,
      "grad_norm": 2.591552257537842,
      "learning_rate": 4.0708399863140375e-06,
      "loss": 0.5286,
      "step": 134110
    },
    {
      "epoch": 54.9,
      "grad_norm": 1.5544551610946655,
      "learning_rate": 4.070732747072263e-06,
      "loss": 0.5375,
      "step": 134120
    },
    {
      "epoch": 54.9,
      "grad_norm": 2.2550594806671143,
      "learning_rate": 4.070625499919448e-06,
      "loss": 0.5092,
      "step": 134130
    },
    {
      "epoch": 54.91,
      "grad_norm": 1.8610217571258545,
      "learning_rate": 4.070518244856084e-06,
      "loss": 0.5151,
      "step": 134140
    },
    {
      "epoch": 54.91,
      "grad_norm": 2.42795991897583,
      "learning_rate": 4.0704109818826625e-06,
      "loss": 0.5339,
      "step": 134150
    },
    {
      "epoch": 54.92,
      "grad_norm": 1.9194391965866089,
      "learning_rate": 4.070303710999674e-06,
      "loss": 0.5282,
      "step": 134160
    },
    {
      "epoch": 54.92,
      "grad_norm": 2.0001184940338135,
      "learning_rate": 4.0701964322076105e-06,
      "loss": 0.5399,
      "step": 134170
    },
    {
      "epoch": 54.92,
      "grad_norm": 1.905078649520874,
      "learning_rate": 4.070089145506963e-06,
      "loss": 0.5455,
      "step": 134180
    },
    {
      "epoch": 54.93,
      "grad_norm": 1.9658211469650269,
      "learning_rate": 4.069981850898223e-06,
      "loss": 0.5136,
      "step": 134190
    },
    {
      "epoch": 54.93,
      "grad_norm": 2.240562677383423,
      "learning_rate": 4.069874548381883e-06,
      "loss": 0.5435,
      "step": 134200
    },
    {
      "epoch": 54.94,
      "grad_norm": 2.9245107173919678,
      "learning_rate": 4.069767237958434e-06,
      "loss": 0.5273,
      "step": 134210
    },
    {
      "epoch": 54.94,
      "grad_norm": 1.6216827630996704,
      "learning_rate": 4.0696599196283665e-06,
      "loss": 0.5521,
      "step": 134220
    },
    {
      "epoch": 54.94,
      "grad_norm": 2.1467161178588867,
      "learning_rate": 4.069552593392173e-06,
      "loss": 0.5385,
      "step": 134230
    },
    {
      "epoch": 54.95,
      "grad_norm": 1.6803362369537354,
      "learning_rate": 4.069445259250345e-06,
      "loss": 0.5501,
      "step": 134240
    },
    {
      "epoch": 54.95,
      "grad_norm": 2.1655826568603516,
      "learning_rate": 4.069337917203375e-06,
      "loss": 0.5529,
      "step": 134250
    },
    {
      "epoch": 54.96,
      "grad_norm": 1.6631871461868286,
      "learning_rate": 4.069230567251753e-06,
      "loss": 0.541,
      "step": 134260
    },
    {
      "epoch": 54.96,
      "grad_norm": 1.8249664306640625,
      "learning_rate": 4.069123209395973e-06,
      "loss": 0.5409,
      "step": 134270
    },
    {
      "epoch": 54.97,
      "grad_norm": 1.9835152626037598,
      "learning_rate": 4.069015843636524e-06,
      "loss": 0.5335,
      "step": 134280
    },
    {
      "epoch": 54.97,
      "grad_norm": 1.551106572151184,
      "learning_rate": 4.068908469973901e-06,
      "loss": 0.5183,
      "step": 134290
    },
    {
      "epoch": 54.97,
      "grad_norm": 2.8822789192199707,
      "learning_rate": 4.068801088408593e-06,
      "loss": 0.5628,
      "step": 134300
    },
    {
      "epoch": 54.98,
      "grad_norm": 1.7391079664230347,
      "learning_rate": 4.068693698941094e-06,
      "loss": 0.5353,
      "step": 134310
    },
    {
      "epoch": 54.98,
      "grad_norm": 2.308398723602295,
      "learning_rate": 4.068586301571894e-06,
      "loss": 0.54,
      "step": 134320
    },
    {
      "epoch": 54.99,
      "grad_norm": 1.815447211265564,
      "learning_rate": 4.068478896301487e-06,
      "loss": 0.5599,
      "step": 134330
    },
    {
      "epoch": 54.99,
      "grad_norm": 1.987694501876831,
      "learning_rate": 4.068371483130363e-06,
      "loss": 0.5117,
      "step": 134340
    },
    {
      "epoch": 54.99,
      "grad_norm": 2.0397775173187256,
      "learning_rate": 4.068264062059016e-06,
      "loss": 0.5132,
      "step": 134350
    },
    {
      "epoch": 55.0,
      "grad_norm": 1.8312160968780518,
      "learning_rate": 4.068156633087936e-06,
      "loss": 0.5295,
      "step": 134360
    },
    {
      "epoch": 55.0,
      "eval_loss": 0.5343015193939209,
      "eval_runtime": 52.0614,
      "eval_samples_per_second": 66.249,
      "eval_steps_per_second": 8.298,
      "step": 134365
    },
    {
      "epoch": 55.0,
      "grad_norm": 2.142251968383789,
      "learning_rate": 4.068049196217618e-06,
      "loss": 0.5364,
      "step": 134370
    },
    {
      "epoch": 55.01,
      "grad_norm": 1.5132172107696533,
      "learning_rate": 4.067941751448551e-06,
      "loss": 0.5388,
      "step": 134380
    },
    {
      "epoch": 55.01,
      "grad_norm": 2.1260735988616943,
      "learning_rate": 4.06783429878123e-06,
      "loss": 0.5479,
      "step": 134390
    },
    {
      "epoch": 55.01,
      "grad_norm": 2.019447088241577,
      "learning_rate": 4.067726838216145e-06,
      "loss": 0.5287,
      "step": 134400
    },
    {
      "epoch": 55.02,
      "grad_norm": 1.8660378456115723,
      "learning_rate": 4.067619369753789e-06,
      "loss": 0.5309,
      "step": 134410
    },
    {
      "epoch": 55.02,
      "grad_norm": 1.956706166267395,
      "learning_rate": 4.067511893394655e-06,
      "loss": 0.5535,
      "step": 134420
    },
    {
      "epoch": 55.03,
      "grad_norm": 1.749645709991455,
      "learning_rate": 4.067404409139234e-06,
      "loss": 0.5285,
      "step": 134430
    },
    {
      "epoch": 55.03,
      "grad_norm": 2.7851388454437256,
      "learning_rate": 4.067296916988019e-06,
      "loss": 0.5369,
      "step": 134440
    },
    {
      "epoch": 55.03,
      "grad_norm": 1.329525351524353,
      "learning_rate": 4.067189416941503e-06,
      "loss": 0.546,
      "step": 134450
    },
    {
      "epoch": 55.04,
      "grad_norm": 2.126760482788086,
      "learning_rate": 4.067081909000178e-06,
      "loss": 0.5371,
      "step": 134460
    },
    {
      "epoch": 55.04,
      "grad_norm": 2.4068610668182373,
      "learning_rate": 4.066974393164535e-06,
      "loss": 0.5228,
      "step": 134470
    },
    {
      "epoch": 55.05,
      "grad_norm": 2.326212167739868,
      "learning_rate": 4.06686686943507e-06,
      "loss": 0.5183,
      "step": 134480
    },
    {
      "epoch": 55.05,
      "grad_norm": 2.5850887298583984,
      "learning_rate": 4.066759337812273e-06,
      "loss": 0.4977,
      "step": 134490
    },
    {
      "epoch": 55.06,
      "grad_norm": 1.4370707273483276,
      "learning_rate": 4.066651798296637e-06,
      "loss": 0.5131,
      "step": 134500
    },
    {
      "epoch": 55.06,
      "grad_norm": 2.067875385284424,
      "learning_rate": 4.066544250888654e-06,
      "loss": 0.5398,
      "step": 134510
    },
    {
      "epoch": 55.06,
      "grad_norm": 2.3489644527435303,
      "learning_rate": 4.066436695588817e-06,
      "loss": 0.5314,
      "step": 134520
    },
    {
      "epoch": 55.07,
      "grad_norm": 1.861151099205017,
      "learning_rate": 4.066329132397621e-06,
      "loss": 0.5384,
      "step": 134530
    },
    {
      "epoch": 55.07,
      "grad_norm": 2.7617406845092773,
      "learning_rate": 4.066221561315554e-06,
      "loss": 0.5184,
      "step": 134540
    },
    {
      "epoch": 55.08,
      "grad_norm": 2.1747968196868896,
      "learning_rate": 4.066113982343114e-06,
      "loss": 0.5274,
      "step": 134550
    },
    {
      "epoch": 55.08,
      "grad_norm": 2.172673463821411,
      "learning_rate": 4.0660063954807904e-06,
      "loss": 0.5212,
      "step": 134560
    },
    {
      "epoch": 55.08,
      "grad_norm": 2.0242867469787598,
      "learning_rate": 4.065898800729076e-06,
      "loss": 0.5455,
      "step": 134570
    },
    {
      "epoch": 55.09,
      "grad_norm": 2.181318759918213,
      "learning_rate": 4.0657911980884655e-06,
      "loss": 0.5341,
      "step": 134580
    },
    {
      "epoch": 55.09,
      "grad_norm": 1.99505615234375,
      "learning_rate": 4.0656835875594515e-06,
      "loss": 0.5277,
      "step": 134590
    },
    {
      "epoch": 55.1,
      "grad_norm": 2.106019973754883,
      "learning_rate": 4.065575969142526e-06,
      "loss": 0.5258,
      "step": 134600
    },
    {
      "epoch": 55.1,
      "grad_norm": 1.8273030519485474,
      "learning_rate": 4.065468342838182e-06,
      "loss": 0.5397,
      "step": 134610
    },
    {
      "epoch": 55.1,
      "grad_norm": 1.9924904108047485,
      "learning_rate": 4.065360708646913e-06,
      "loss": 0.555,
      "step": 134620
    },
    {
      "epoch": 55.11,
      "grad_norm": 1.441427230834961,
      "learning_rate": 4.065253066569213e-06,
      "loss": 0.5389,
      "step": 134630
    },
    {
      "epoch": 55.11,
      "grad_norm": 2.591911792755127,
      "learning_rate": 4.065145416605573e-06,
      "loss": 0.5375,
      "step": 134640
    },
    {
      "epoch": 55.12,
      "grad_norm": 1.8852084875106812,
      "learning_rate": 4.065037758756488e-06,
      "loss": 0.5421,
      "step": 134650
    },
    {
      "epoch": 55.12,
      "grad_norm": 1.8838448524475098,
      "learning_rate": 4.06493009302245e-06,
      "loss": 0.5241,
      "step": 134660
    },
    {
      "epoch": 55.12,
      "grad_norm": 1.5720655918121338,
      "learning_rate": 4.064822419403952e-06,
      "loss": 0.5214,
      "step": 134670
    },
    {
      "epoch": 55.13,
      "grad_norm": 1.5988795757293701,
      "learning_rate": 4.064714737901488e-06,
      "loss": 0.5339,
      "step": 134680
    },
    {
      "epoch": 55.13,
      "grad_norm": 2.2383906841278076,
      "learning_rate": 4.064607048515551e-06,
      "loss": 0.5219,
      "step": 134690
    },
    {
      "epoch": 55.14,
      "grad_norm": 2.1665048599243164,
      "learning_rate": 4.064499351246635e-06,
      "loss": 0.5441,
      "step": 134700
    },
    {
      "epoch": 55.14,
      "grad_norm": 2.3454339504241943,
      "learning_rate": 4.064391646095233e-06,
      "loss": 0.5396,
      "step": 134710
    },
    {
      "epoch": 55.15,
      "grad_norm": 1.6700067520141602,
      "learning_rate": 4.064283933061837e-06,
      "loss": 0.5298,
      "step": 134720
    },
    {
      "epoch": 55.15,
      "grad_norm": 1.8344975709915161,
      "learning_rate": 4.064176212146942e-06,
      "loss": 0.5067,
      "step": 134730
    },
    {
      "epoch": 55.15,
      "grad_norm": 1.8788723945617676,
      "learning_rate": 4.064068483351042e-06,
      "loss": 0.5368,
      "step": 134740
    },
    {
      "epoch": 55.16,
      "grad_norm": 2.2651147842407227,
      "learning_rate": 4.063960746674627e-06,
      "loss": 0.5157,
      "step": 134750
    },
    {
      "epoch": 55.16,
      "grad_norm": 2.158264398574829,
      "learning_rate": 4.063853002118195e-06,
      "loss": 0.539,
      "step": 134760
    },
    {
      "epoch": 55.17,
      "grad_norm": 1.8040670156478882,
      "learning_rate": 4.063745249682237e-06,
      "loss": 0.5179,
      "step": 134770
    },
    {
      "epoch": 55.17,
      "grad_norm": 2.034822463989258,
      "learning_rate": 4.063637489367247e-06,
      "loss": 0.5309,
      "step": 134780
    },
    {
      "epoch": 55.17,
      "grad_norm": 1.670730710029602,
      "learning_rate": 4.063529721173718e-06,
      "loss": 0.5448,
      "step": 134790
    },
    {
      "epoch": 55.18,
      "grad_norm": 1.9785492420196533,
      "learning_rate": 4.063421945102146e-06,
      "loss": 0.5423,
      "step": 134800
    },
    {
      "epoch": 55.18,
      "grad_norm": 2.0952422618865967,
      "learning_rate": 4.063314161153022e-06,
      "loss": 0.5314,
      "step": 134810
    },
    {
      "epoch": 55.19,
      "grad_norm": 2.662372589111328,
      "learning_rate": 4.06320636932684e-06,
      "loss": 0.5325,
      "step": 134820
    },
    {
      "epoch": 55.19,
      "grad_norm": 2.3505756855010986,
      "learning_rate": 4.063098569624098e-06,
      "loss": 0.5357,
      "step": 134830
    },
    {
      "epoch": 55.19,
      "grad_norm": 2.0574605464935303,
      "learning_rate": 4.062990762045283e-06,
      "loss": 0.5494,
      "step": 134840
    },
    {
      "epoch": 55.2,
      "grad_norm": 1.9828345775604248,
      "learning_rate": 4.062882946590894e-06,
      "loss": 0.5236,
      "step": 134850
    },
    {
      "epoch": 55.2,
      "grad_norm": 1.9266396760940552,
      "learning_rate": 4.062775123261422e-06,
      "loss": 0.5505,
      "step": 134860
    },
    {
      "epoch": 55.21,
      "grad_norm": 2.298731803894043,
      "learning_rate": 4.062667292057364e-06,
      "loss": 0.533,
      "step": 134870
    },
    {
      "epoch": 55.21,
      "grad_norm": 1.8787710666656494,
      "learning_rate": 4.062559452979211e-06,
      "loss": 0.5432,
      "step": 134880
    },
    {
      "epoch": 55.21,
      "grad_norm": 1.9545676708221436,
      "learning_rate": 4.062451606027458e-06,
      "loss": 0.5503,
      "step": 134890
    },
    {
      "epoch": 55.22,
      "grad_norm": 2.3733668327331543,
      "learning_rate": 4.062343751202599e-06,
      "loss": 0.5227,
      "step": 134900
    },
    {
      "epoch": 55.22,
      "grad_norm": 1.3950257301330566,
      "learning_rate": 4.062235888505129e-06,
      "loss": 0.5109,
      "step": 134910
    },
    {
      "epoch": 55.23,
      "grad_norm": 1.4353824853897095,
      "learning_rate": 4.062128017935542e-06,
      "loss": 0.5273,
      "step": 134920
    },
    {
      "epoch": 55.23,
      "grad_norm": 1.318987250328064,
      "learning_rate": 4.062020139494329e-06,
      "loss": 0.533,
      "step": 134930
    },
    {
      "epoch": 55.24,
      "grad_norm": 3.535403251647949,
      "learning_rate": 4.061912253181988e-06,
      "loss": 0.5211,
      "step": 134940
    },
    {
      "epoch": 55.24,
      "grad_norm": 1.9571341276168823,
      "learning_rate": 4.0618043589990116e-06,
      "loss": 0.5338,
      "step": 134950
    },
    {
      "epoch": 55.24,
      "grad_norm": 2.0499050617218018,
      "learning_rate": 4.061696456945895e-06,
      "loss": 0.527,
      "step": 134960
    },
    {
      "epoch": 55.25,
      "grad_norm": 2.0552773475646973,
      "learning_rate": 4.061588547023131e-06,
      "loss": 0.5172,
      "step": 134970
    },
    {
      "epoch": 55.25,
      "grad_norm": 2.380800485610962,
      "learning_rate": 4.0614806292312145e-06,
      "loss": 0.522,
      "step": 134980
    },
    {
      "epoch": 55.26,
      "grad_norm": 1.8453654050827026,
      "learning_rate": 4.06137270357064e-06,
      "loss": 0.5395,
      "step": 134990
    },
    {
      "epoch": 55.26,
      "grad_norm": 2.074673652648926,
      "learning_rate": 4.061264770041902e-06,
      "loss": 0.5429,
      "step": 135000
    },
    {
      "epoch": 55.26,
      "grad_norm": 1.7614322900772095,
      "learning_rate": 4.061156828645496e-06,
      "loss": 0.5225,
      "step": 135010
    },
    {
      "epoch": 55.27,
      "grad_norm": 1.805139183998108,
      "learning_rate": 4.061048879381913e-06,
      "loss": 0.541,
      "step": 135020
    },
    {
      "epoch": 55.27,
      "grad_norm": 2.0633764266967773,
      "learning_rate": 4.060940922251651e-06,
      "loss": 0.5379,
      "step": 135030
    },
    {
      "epoch": 55.28,
      "grad_norm": 2.4210293292999268,
      "learning_rate": 4.0608329572552036e-06,
      "loss": 0.5117,
      "step": 135040
    },
    {
      "epoch": 55.28,
      "grad_norm": 2.2109568119049072,
      "learning_rate": 4.060724984393065e-06,
      "loss": 0.5402,
      "step": 135050
    },
    {
      "epoch": 55.28,
      "grad_norm": 2.609501838684082,
      "learning_rate": 4.06061700366573e-06,
      "loss": 0.5312,
      "step": 135060
    },
    {
      "epoch": 55.29,
      "grad_norm": 1.6726939678192139,
      "learning_rate": 4.060509015073693e-06,
      "loss": 0.5312,
      "step": 135070
    },
    {
      "epoch": 55.29,
      "grad_norm": 2.091695547103882,
      "learning_rate": 4.060401018617448e-06,
      "loss": 0.5181,
      "step": 135080
    },
    {
      "epoch": 55.3,
      "grad_norm": 1.869715690612793,
      "learning_rate": 4.060293014297492e-06,
      "loss": 0.5336,
      "step": 135090
    },
    {
      "epoch": 55.3,
      "grad_norm": 1.7155338525772095,
      "learning_rate": 4.060185002114317e-06,
      "loss": 0.5378,
      "step": 135100
    },
    {
      "epoch": 55.3,
      "grad_norm": 2.555711269378662,
      "learning_rate": 4.06007698206842e-06,
      "loss": 0.518,
      "step": 135110
    },
    {
      "epoch": 55.31,
      "grad_norm": 1.9277845621109009,
      "learning_rate": 4.059968954160295e-06,
      "loss": 0.5278,
      "step": 135120
    },
    {
      "epoch": 55.31,
      "grad_norm": 1.9349288940429688,
      "learning_rate": 4.059860918390435e-06,
      "loss": 0.5138,
      "step": 135130
    },
    {
      "epoch": 55.32,
      "grad_norm": 1.9292645454406738,
      "learning_rate": 4.059752874759338e-06,
      "loss": 0.5209,
      "step": 135140
    },
    {
      "epoch": 55.32,
      "grad_norm": 1.9484515190124512,
      "learning_rate": 4.059644823267498e-06,
      "loss": 0.5136,
      "step": 135150
    },
    {
      "epoch": 55.33,
      "grad_norm": 2.1349964141845703,
      "learning_rate": 4.059536763915409e-06,
      "loss": 0.5356,
      "step": 135160
    },
    {
      "epoch": 55.33,
      "grad_norm": 1.9487546682357788,
      "learning_rate": 4.059428696703567e-06,
      "loss": 0.512,
      "step": 135170
    },
    {
      "epoch": 55.33,
      "grad_norm": 2.3934285640716553,
      "learning_rate": 4.059320621632466e-06,
      "loss": 0.5292,
      "step": 135180
    },
    {
      "epoch": 55.34,
      "grad_norm": 1.707032561302185,
      "learning_rate": 4.0592125387026015e-06,
      "loss": 0.5374,
      "step": 135190
    },
    {
      "epoch": 55.34,
      "grad_norm": 2.0743513107299805,
      "learning_rate": 4.05910444791447e-06,
      "loss": 0.5249,
      "step": 135200
    },
    {
      "epoch": 55.35,
      "grad_norm": 1.846676230430603,
      "learning_rate": 4.058996349268565e-06,
      "loss": 0.554,
      "step": 135210
    },
    {
      "epoch": 55.35,
      "grad_norm": 2.0014188289642334,
      "learning_rate": 4.058888242765381e-06,
      "loss": 0.5178,
      "step": 135220
    },
    {
      "epoch": 55.35,
      "grad_norm": 1.8531467914581299,
      "learning_rate": 4.058780128405415e-06,
      "loss": 0.5233,
      "step": 135230
    },
    {
      "epoch": 55.36,
      "grad_norm": 2.056790590286255,
      "learning_rate": 4.058672006189162e-06,
      "loss": 0.5415,
      "step": 135240
    },
    {
      "epoch": 55.36,
      "grad_norm": 1.9431037902832031,
      "learning_rate": 4.058563876117117e-06,
      "loss": 0.5473,
      "step": 135250
    },
    {
      "epoch": 55.37,
      "grad_norm": 1.984246850013733,
      "learning_rate": 4.058455738189775e-06,
      "loss": 0.5225,
      "step": 135260
    },
    {
      "epoch": 55.37,
      "grad_norm": 1.8972570896148682,
      "learning_rate": 4.05834759240763e-06,
      "loss": 0.531,
      "step": 135270
    },
    {
      "epoch": 55.37,
      "grad_norm": 1.9926832914352417,
      "learning_rate": 4.058239438771181e-06,
      "loss": 0.5509,
      "step": 135280
    },
    {
      "epoch": 55.38,
      "grad_norm": 2.1729576587677,
      "learning_rate": 4.058131277280921e-06,
      "loss": 0.5259,
      "step": 135290
    },
    {
      "epoch": 55.38,
      "grad_norm": 1.9833488464355469,
      "learning_rate": 4.058023107937345e-06,
      "loss": 0.509,
      "step": 135300
    },
    {
      "epoch": 55.39,
      "grad_norm": 1.7176710367202759,
      "learning_rate": 4.05791493074095e-06,
      "loss": 0.5398,
      "step": 135310
    },
    {
      "epoch": 55.39,
      "grad_norm": 2.0084245204925537,
      "learning_rate": 4.0578067456922315e-06,
      "loss": 0.5395,
      "step": 135320
    },
    {
      "epoch": 55.4,
      "grad_norm": 2.354872226715088,
      "learning_rate": 4.0576985527916835e-06,
      "loss": 0.5297,
      "step": 135330
    },
    {
      "epoch": 55.4,
      "grad_norm": 2.007570266723633,
      "learning_rate": 4.057590352039803e-06,
      "loss": 0.5323,
      "step": 135340
    },
    {
      "epoch": 55.4,
      "grad_norm": 1.9335254430770874,
      "learning_rate": 4.057482143437086e-06,
      "loss": 0.5319,
      "step": 135350
    },
    {
      "epoch": 55.41,
      "grad_norm": 1.978175401687622,
      "learning_rate": 4.057373926984026e-06,
      "loss": 0.531,
      "step": 135360
    },
    {
      "epoch": 55.41,
      "grad_norm": 1.9995925426483154,
      "learning_rate": 4.057265702681121e-06,
      "loss": 0.5223,
      "step": 135370
    },
    {
      "epoch": 55.42,
      "grad_norm": 1.6401677131652832,
      "learning_rate": 4.057157470528865e-06,
      "loss": 0.527,
      "step": 135380
    },
    {
      "epoch": 55.42,
      "grad_norm": 1.8831430673599243,
      "learning_rate": 4.057049230527757e-06,
      "loss": 0.5269,
      "step": 135390
    },
    {
      "epoch": 55.42,
      "grad_norm": 2.050532579421997,
      "learning_rate": 4.056940982678289e-06,
      "loss": 0.5185,
      "step": 135400
    },
    {
      "epoch": 55.43,
      "grad_norm": 2.17182993888855,
      "learning_rate": 4.0568327269809575e-06,
      "loss": 0.5306,
      "step": 135410
    },
    {
      "epoch": 55.43,
      "grad_norm": 1.960013747215271,
      "learning_rate": 4.056724463436261e-06,
      "loss": 0.5148,
      "step": 135420
    },
    {
      "epoch": 55.44,
      "grad_norm": 1.9098531007766724,
      "learning_rate": 4.056616192044693e-06,
      "loss": 0.5329,
      "step": 135430
    },
    {
      "epoch": 55.44,
      "grad_norm": 1.653612732887268,
      "learning_rate": 4.0565079128067505e-06,
      "loss": 0.5475,
      "step": 135440
    },
    {
      "epoch": 55.44,
      "grad_norm": 1.9950798749923706,
      "learning_rate": 4.0563996257229295e-06,
      "loss": 0.5198,
      "step": 135450
    },
    {
      "epoch": 55.45,
      "grad_norm": 1.9337490797042847,
      "learning_rate": 4.056291330793725e-06,
      "loss": 0.5247,
      "step": 135460
    },
    {
      "epoch": 55.45,
      "grad_norm": 2.0159804821014404,
      "learning_rate": 4.056183028019635e-06,
      "loss": 0.5282,
      "step": 135470
    },
    {
      "epoch": 55.46,
      "grad_norm": 2.3988075256347656,
      "learning_rate": 4.056074717401154e-06,
      "loss": 0.534,
      "step": 135480
    },
    {
      "epoch": 55.46,
      "grad_norm": 1.8080482482910156,
      "learning_rate": 4.055966398938778e-06,
      "loss": 0.5345,
      "step": 135490
    },
    {
      "epoch": 55.46,
      "grad_norm": 2.055481195449829,
      "learning_rate": 4.055858072633005e-06,
      "loss": 0.542,
      "step": 135500
    },
    {
      "epoch": 55.47,
      "grad_norm": 1.6567028760910034,
      "learning_rate": 4.055749738484329e-06,
      "loss": 0.5284,
      "step": 135510
    },
    {
      "epoch": 55.47,
      "grad_norm": 1.6150267124176025,
      "learning_rate": 4.055641396493247e-06,
      "loss": 0.5283,
      "step": 135520
    },
    {
      "epoch": 55.48,
      "grad_norm": 2.4665310382843018,
      "learning_rate": 4.055533046660257e-06,
      "loss": 0.5159,
      "step": 135530
    },
    {
      "epoch": 55.48,
      "grad_norm": 1.6996208429336548,
      "learning_rate": 4.055424688985853e-06,
      "loss": 0.5197,
      "step": 135540
    },
    {
      "epoch": 55.49,
      "grad_norm": 1.683939814567566,
      "learning_rate": 4.055316323470533e-06,
      "loss": 0.5294,
      "step": 135550
    },
    {
      "epoch": 55.49,
      "grad_norm": 2.878119707107544,
      "learning_rate": 4.055207950114792e-06,
      "loss": 0.5155,
      "step": 135560
    },
    {
      "epoch": 55.49,
      "grad_norm": 1.4430367946624756,
      "learning_rate": 4.055099568919128e-06,
      "loss": 0.5512,
      "step": 135570
    },
    {
      "epoch": 55.5,
      "grad_norm": 3.3085927963256836,
      "learning_rate": 4.054991179884037e-06,
      "loss": 0.5404,
      "step": 135580
    },
    {
      "epoch": 55.5,
      "grad_norm": 1.7394930124282837,
      "learning_rate": 4.0548827830100135e-06,
      "loss": 0.5271,
      "step": 135590
    },
    {
      "epoch": 55.51,
      "grad_norm": 2.157357692718506,
      "learning_rate": 4.0547743782975565e-06,
      "loss": 0.5272,
      "step": 135600
    },
    {
      "epoch": 55.51,
      "grad_norm": 1.9230842590332031,
      "learning_rate": 4.054665965747162e-06,
      "loss": 0.5196,
      "step": 135610
    },
    {
      "epoch": 55.51,
      "grad_norm": 1.9936206340789795,
      "learning_rate": 4.054557545359327e-06,
      "loss": 0.5171,
      "step": 135620
    },
    {
      "epoch": 55.52,
      "grad_norm": 2.12422776222229,
      "learning_rate": 4.054449117134547e-06,
      "loss": 0.5358,
      "step": 135630
    },
    {
      "epoch": 55.52,
      "grad_norm": 2.47271466255188,
      "learning_rate": 4.054340681073319e-06,
      "loss": 0.5399,
      "step": 135640
    },
    {
      "epoch": 55.53,
      "grad_norm": 1.9980785846710205,
      "learning_rate": 4.0542322371761405e-06,
      "loss": 0.5613,
      "step": 135650
    },
    {
      "epoch": 55.53,
      "grad_norm": 1.875928282737732,
      "learning_rate": 4.054123785443507e-06,
      "loss": 0.5639,
      "step": 135660
    },
    {
      "epoch": 55.53,
      "grad_norm": 2.0821409225463867,
      "learning_rate": 4.054015325875917e-06,
      "loss": 0.5506,
      "step": 135670
    },
    {
      "epoch": 55.54,
      "grad_norm": 1.9748274087905884,
      "learning_rate": 4.053906858473866e-06,
      "loss": 0.5415,
      "step": 135680
    },
    {
      "epoch": 55.54,
      "grad_norm": 2.044341802597046,
      "learning_rate": 4.053798383237851e-06,
      "loss": 0.5322,
      "step": 135690
    },
    {
      "epoch": 55.55,
      "grad_norm": 1.8547399044036865,
      "learning_rate": 4.05368990016837e-06,
      "loss": 0.5119,
      "step": 135700
    },
    {
      "epoch": 55.55,
      "grad_norm": 2.0335586071014404,
      "learning_rate": 4.053581409265919e-06,
      "loss": 0.5163,
      "step": 135710
    },
    {
      "epoch": 55.55,
      "grad_norm": 1.3724557161331177,
      "learning_rate": 4.053472910530994e-06,
      "loss": 0.5572,
      "step": 135720
    },
    {
      "epoch": 55.56,
      "grad_norm": 2.4363279342651367,
      "learning_rate": 4.053364403964094e-06,
      "loss": 0.5205,
      "step": 135730
    },
    {
      "epoch": 55.56,
      "grad_norm": 1.4020159244537354,
      "learning_rate": 4.053255889565716e-06,
      "loss": 0.496,
      "step": 135740
    },
    {
      "epoch": 55.57,
      "grad_norm": 2.044250011444092,
      "learning_rate": 4.053147367336355e-06,
      "loss": 0.5335,
      "step": 135750
    },
    {
      "epoch": 55.57,
      "grad_norm": 2.0637874603271484,
      "learning_rate": 4.053038837276509e-06,
      "loss": 0.5338,
      "step": 135760
    },
    {
      "epoch": 55.58,
      "grad_norm": 2.0419139862060547,
      "learning_rate": 4.052930299386677e-06,
      "loss": 0.5143,
      "step": 135770
    },
    {
      "epoch": 55.58,
      "grad_norm": 1.5379703044891357,
      "learning_rate": 4.052821753667354e-06,
      "loss": 0.5233,
      "step": 135780
    },
    {
      "epoch": 55.58,
      "grad_norm": 2.0058228969573975,
      "learning_rate": 4.052713200119038e-06,
      "loss": 0.5222,
      "step": 135790
    },
    {
      "epoch": 55.59,
      "grad_norm": 1.8417258262634277,
      "learning_rate": 4.052604638742226e-06,
      "loss": 0.5364,
      "step": 135800
    },
    {
      "epoch": 55.59,
      "grad_norm": 2.458630323410034,
      "learning_rate": 4.052496069537416e-06,
      "loss": 0.5367,
      "step": 135810
    },
    {
      "epoch": 55.6,
      "grad_norm": 1.5455245971679688,
      "learning_rate": 4.052387492505105e-06,
      "loss": 0.5457,
      "step": 135820
    },
    {
      "epoch": 55.6,
      "grad_norm": 2.3548827171325684,
      "learning_rate": 4.05227890764579e-06,
      "loss": 0.5462,
      "step": 135830
    },
    {
      "epoch": 55.6,
      "grad_norm": 2.0880651473999023,
      "learning_rate": 4.052170314959968e-06,
      "loss": 0.5287,
      "step": 135840
    },
    {
      "epoch": 55.61,
      "grad_norm": 2.215331554412842,
      "learning_rate": 4.052061714448138e-06,
      "loss": 0.5171,
      "step": 135850
    },
    {
      "epoch": 55.61,
      "grad_norm": 2.432131052017212,
      "learning_rate": 4.051953106110796e-06,
      "loss": 0.5319,
      "step": 135860
    },
    {
      "epoch": 55.62,
      "grad_norm": 1.9953253269195557,
      "learning_rate": 4.051844489948441e-06,
      "loss": 0.533,
      "step": 135870
    },
    {
      "epoch": 55.62,
      "grad_norm": 2.1713385581970215,
      "learning_rate": 4.051735865961569e-06,
      "loss": 0.5463,
      "step": 135880
    },
    {
      "epoch": 55.62,
      "grad_norm": 2.172294855117798,
      "learning_rate": 4.0516272341506785e-06,
      "loss": 0.5504,
      "step": 135890
    },
    {
      "epoch": 55.63,
      "grad_norm": 2.052274703979492,
      "learning_rate": 4.051518594516267e-06,
      "loss": 0.5373,
      "step": 135900
    },
    {
      "epoch": 55.63,
      "grad_norm": 2.2391762733459473,
      "learning_rate": 4.051409947058832e-06,
      "loss": 0.5283,
      "step": 135910
    },
    {
      "epoch": 55.64,
      "grad_norm": 2.4095282554626465,
      "learning_rate": 4.051301291778871e-06,
      "loss": 0.5279,
      "step": 135920
    },
    {
      "epoch": 55.64,
      "grad_norm": 1.9076900482177734,
      "learning_rate": 4.051192628676882e-06,
      "loss": 0.5161,
      "step": 135930
    },
    {
      "epoch": 55.64,
      "grad_norm": 1.7826648950576782,
      "learning_rate": 4.051083957753363e-06,
      "loss": 0.5363,
      "step": 135940
    },
    {
      "epoch": 55.65,
      "grad_norm": 1.989090085029602,
      "learning_rate": 4.050975279008811e-06,
      "loss": 0.52,
      "step": 135950
    },
    {
      "epoch": 55.65,
      "grad_norm": 2.114499807357788,
      "learning_rate": 4.050866592443725e-06,
      "loss": 0.533,
      "step": 135960
    },
    {
      "epoch": 55.66,
      "grad_norm": 2.122990369796753,
      "learning_rate": 4.050757898058603e-06,
      "loss": 0.5382,
      "step": 135970
    },
    {
      "epoch": 55.66,
      "grad_norm": 2.369263172149658,
      "learning_rate": 4.05064919585394e-06,
      "loss": 0.5345,
      "step": 135980
    },
    {
      "epoch": 55.67,
      "grad_norm": 3.1587185859680176,
      "learning_rate": 4.050540485830238e-06,
      "loss": 0.5158,
      "step": 135990
    },
    {
      "epoch": 55.67,
      "grad_norm": 1.544399380683899,
      "learning_rate": 4.050431767987992e-06,
      "loss": 0.5505,
      "step": 136000
    },
    {
      "epoch": 55.67,
      "grad_norm": 2.7160284519195557,
      "learning_rate": 4.050323042327702e-06,
      "loss": 0.5506,
      "step": 136010
    },
    {
      "epoch": 55.68,
      "grad_norm": 1.8240742683410645,
      "learning_rate": 4.050214308849864e-06,
      "loss": 0.5342,
      "step": 136020
    },
    {
      "epoch": 55.68,
      "grad_norm": 2.5455591678619385,
      "learning_rate": 4.050105567554979e-06,
      "loss": 0.5423,
      "step": 136030
    },
    {
      "epoch": 55.69,
      "grad_norm": 1.6666382551193237,
      "learning_rate": 4.049996818443542e-06,
      "loss": 0.5433,
      "step": 136040
    },
    {
      "epoch": 55.69,
      "grad_norm": 1.7199040651321411,
      "learning_rate": 4.049888061516054e-06,
      "loss": 0.5421,
      "step": 136050
    },
    {
      "epoch": 55.69,
      "grad_norm": 2.0920679569244385,
      "learning_rate": 4.04977929677301e-06,
      "loss": 0.5266,
      "step": 136060
    },
    {
      "epoch": 55.7,
      "grad_norm": 2.0501670837402344,
      "learning_rate": 4.049670524214911e-06,
      "loss": 0.5531,
      "step": 136070
    },
    {
      "epoch": 55.7,
      "grad_norm": 2.1676957607269287,
      "learning_rate": 4.049561743842255e-06,
      "loss": 0.5246,
      "step": 136080
    },
    {
      "epoch": 55.71,
      "grad_norm": 2.2675299644470215,
      "learning_rate": 4.049452955655538e-06,
      "loss": 0.5416,
      "step": 136090
    },
    {
      "epoch": 55.71,
      "grad_norm": 1.8399213552474976,
      "learning_rate": 4.049344159655261e-06,
      "loss": 0.5387,
      "step": 136100
    },
    {
      "epoch": 55.71,
      "grad_norm": 1.881492257118225,
      "learning_rate": 4.049235355841921e-06,
      "loss": 0.5385,
      "step": 136110
    },
    {
      "epoch": 55.72,
      "grad_norm": 1.7074302434921265,
      "learning_rate": 4.049126544216016e-06,
      "loss": 0.537,
      "step": 136120
    },
    {
      "epoch": 55.72,
      "grad_norm": 1.7061890363693237,
      "learning_rate": 4.049017724778046e-06,
      "loss": 0.549,
      "step": 136130
    },
    {
      "epoch": 55.73,
      "grad_norm": 2.0617451667785645,
      "learning_rate": 4.048908897528508e-06,
      "loss": 0.5283,
      "step": 136140
    },
    {
      "epoch": 55.73,
      "grad_norm": 2.265617847442627,
      "learning_rate": 4.0488000624679015e-06,
      "loss": 0.5127,
      "step": 136150
    },
    {
      "epoch": 55.73,
      "grad_norm": 2.1158506870269775,
      "learning_rate": 4.048691219596725e-06,
      "loss": 0.5208,
      "step": 136160
    },
    {
      "epoch": 55.74,
      "grad_norm": 2.236088991165161,
      "learning_rate": 4.048582368915477e-06,
      "loss": 0.5386,
      "step": 136170
    },
    {
      "epoch": 55.74,
      "grad_norm": 2.010789155960083,
      "learning_rate": 4.048473510424656e-06,
      "loss": 0.5401,
      "step": 136180
    },
    {
      "epoch": 55.75,
      "grad_norm": 1.9389013051986694,
      "learning_rate": 4.04836464412476e-06,
      "loss": 0.5284,
      "step": 136190
    },
    {
      "epoch": 55.75,
      "grad_norm": 2.1468307971954346,
      "learning_rate": 4.0482557700162886e-06,
      "loss": 0.5454,
      "step": 136200
    },
    {
      "epoch": 55.76,
      "grad_norm": 1.9472072124481201,
      "learning_rate": 4.048146888099739e-06,
      "loss": 0.5316,
      "step": 136210
    },
    {
      "epoch": 55.76,
      "grad_norm": 1.8125245571136475,
      "learning_rate": 4.048037998375613e-06,
      "loss": 0.5277,
      "step": 136220
    },
    {
      "epoch": 55.76,
      "grad_norm": 1.9155887365341187,
      "learning_rate": 4.047929100844407e-06,
      "loss": 0.5753,
      "step": 136230
    },
    {
      "epoch": 55.77,
      "grad_norm": 2.202608108520508,
      "learning_rate": 4.047820195506621e-06,
      "loss": 0.5236,
      "step": 136240
    },
    {
      "epoch": 55.77,
      "grad_norm": 2.073714017868042,
      "learning_rate": 4.047711282362752e-06,
      "loss": 0.5477,
      "step": 136250
    },
    {
      "epoch": 55.78,
      "grad_norm": 2.9607064723968506,
      "learning_rate": 4.047602361413301e-06,
      "loss": 0.5342,
      "step": 136260
    },
    {
      "epoch": 55.78,
      "grad_norm": 2.084113597869873,
      "learning_rate": 4.047493432658767e-06,
      "loss": 0.5219,
      "step": 136270
    },
    {
      "epoch": 55.78,
      "grad_norm": 2.5144946575164795,
      "learning_rate": 4.047384496099647e-06,
      "loss": 0.512,
      "step": 136280
    },
    {
      "epoch": 55.79,
      "grad_norm": 2.6372809410095215,
      "learning_rate": 4.047275551736442e-06,
      "loss": 0.5116,
      "step": 136290
    },
    {
      "epoch": 55.79,
      "grad_norm": 1.9200433492660522,
      "learning_rate": 4.04716659956965e-06,
      "loss": 0.5282,
      "step": 136300
    },
    {
      "epoch": 55.8,
      "grad_norm": 2.4944252967834473,
      "learning_rate": 4.0470576395997696e-06,
      "loss": 0.5288,
      "step": 136310
    },
    {
      "epoch": 55.8,
      "grad_norm": 2.1683189868927,
      "learning_rate": 4.0469486718273015e-06,
      "loss": 0.5266,
      "step": 136320
    },
    {
      "epoch": 55.8,
      "grad_norm": 1.9525281190872192,
      "learning_rate": 4.046839696252744e-06,
      "loss": 0.5189,
      "step": 136330
    },
    {
      "epoch": 55.81,
      "grad_norm": 2.128206253051758,
      "learning_rate": 4.046730712876596e-06,
      "loss": 0.5133,
      "step": 136340
    },
    {
      "epoch": 55.81,
      "grad_norm": 1.6573268175125122,
      "learning_rate": 4.046621721699358e-06,
      "loss": 0.5435,
      "step": 136350
    },
    {
      "epoch": 55.82,
      "grad_norm": 1.5216768980026245,
      "learning_rate": 4.0465127227215276e-06,
      "loss": 0.5299,
      "step": 136360
    },
    {
      "epoch": 55.82,
      "grad_norm": 2.179058074951172,
      "learning_rate": 4.046403715943606e-06,
      "loss": 0.53,
      "step": 136370
    },
    {
      "epoch": 55.82,
      "grad_norm": 1.6263887882232666,
      "learning_rate": 4.046294701366089e-06,
      "loss": 0.5131,
      "step": 136380
    },
    {
      "epoch": 55.83,
      "grad_norm": 2.893829107284546,
      "learning_rate": 4.0461856789894804e-06,
      "loss": 0.5277,
      "step": 136390
    },
    {
      "epoch": 55.83,
      "grad_norm": 1.9546772241592407,
      "learning_rate": 4.046076648814277e-06,
      "loss": 0.5329,
      "step": 136400
    },
    {
      "epoch": 55.84,
      "grad_norm": 2.674103260040283,
      "learning_rate": 4.045967610840979e-06,
      "loss": 0.5426,
      "step": 136410
    },
    {
      "epoch": 55.84,
      "grad_norm": 1.643548607826233,
      "learning_rate": 4.045858565070086e-06,
      "loss": 0.5355,
      "step": 136420
    },
    {
      "epoch": 55.85,
      "grad_norm": 2.8451831340789795,
      "learning_rate": 4.045749511502097e-06,
      "loss": 0.5295,
      "step": 136430
    },
    {
      "epoch": 55.85,
      "grad_norm": 2.3521742820739746,
      "learning_rate": 4.045640450137511e-06,
      "loss": 0.5375,
      "step": 136440
    },
    {
      "epoch": 55.85,
      "grad_norm": 2.1984164714813232,
      "learning_rate": 4.045531380976829e-06,
      "loss": 0.5294,
      "step": 136450
    },
    {
      "epoch": 55.86,
      "grad_norm": 1.881373405456543,
      "learning_rate": 4.0454223040205506e-06,
      "loss": 0.5524,
      "step": 136460
    },
    {
      "epoch": 55.86,
      "grad_norm": 2.448354721069336,
      "learning_rate": 4.045313219269175e-06,
      "loss": 0.521,
      "step": 136470
    },
    {
      "epoch": 55.87,
      "grad_norm": 2.518087387084961,
      "learning_rate": 4.0452041267232016e-06,
      "loss": 0.5415,
      "step": 136480
    },
    {
      "epoch": 55.87,
      "grad_norm": 2.682429552078247,
      "learning_rate": 4.04509502638313e-06,
      "loss": 0.5237,
      "step": 136490
    },
    {
      "epoch": 55.87,
      "grad_norm": 1.9603558778762817,
      "learning_rate": 4.04498591824946e-06,
      "loss": 0.5287,
      "step": 136500
    },
    {
      "epoch": 55.88,
      "grad_norm": 2.3283567428588867,
      "learning_rate": 4.044876802322692e-06,
      "loss": 0.5254,
      "step": 136510
    },
    {
      "epoch": 55.88,
      "grad_norm": 1.9264150857925415,
      "learning_rate": 4.044767678603326e-06,
      "loss": 0.5331,
      "step": 136520
    },
    {
      "epoch": 55.89,
      "grad_norm": 1.747584342956543,
      "learning_rate": 4.044658547091861e-06,
      "loss": 0.5135,
      "step": 136530
    },
    {
      "epoch": 55.89,
      "grad_norm": 2.394484043121338,
      "learning_rate": 4.044549407788798e-06,
      "loss": 0.5403,
      "step": 136540
    },
    {
      "epoch": 55.89,
      "grad_norm": 2.0312888622283936,
      "learning_rate": 4.044440260694635e-06,
      "loss": 0.5331,
      "step": 136550
    },
    {
      "epoch": 55.9,
      "grad_norm": 2.1000866889953613,
      "learning_rate": 4.0443311058098745e-06,
      "loss": 0.5205,
      "step": 136560
    },
    {
      "epoch": 55.9,
      "grad_norm": 1.6662406921386719,
      "learning_rate": 4.044221943135015e-06,
      "loss": 0.5204,
      "step": 136570
    },
    {
      "epoch": 55.91,
      "grad_norm": 1.8441665172576904,
      "learning_rate": 4.044112772670557e-06,
      "loss": 0.5531,
      "step": 136580
    },
    {
      "epoch": 55.91,
      "grad_norm": 1.7032793760299683,
      "learning_rate": 4.044003594417e-06,
      "loss": 0.528,
      "step": 136590
    },
    {
      "epoch": 55.91,
      "grad_norm": 1.7827634811401367,
      "learning_rate": 4.0438944083748455e-06,
      "loss": 0.5374,
      "step": 136600
    },
    {
      "epoch": 55.92,
      "grad_norm": 1.8277794122695923,
      "learning_rate": 4.043785214544592e-06,
      "loss": 0.5349,
      "step": 136610
    },
    {
      "epoch": 55.92,
      "grad_norm": 2.0948286056518555,
      "learning_rate": 4.043676012926741e-06,
      "loss": 0.524,
      "step": 136620
    },
    {
      "epoch": 55.93,
      "grad_norm": 2.2315893173217773,
      "learning_rate": 4.0435668035217914e-06,
      "loss": 0.5483,
      "step": 136630
    },
    {
      "epoch": 55.93,
      "grad_norm": 1.5945042371749878,
      "learning_rate": 4.043457586330245e-06,
      "loss": 0.526,
      "step": 136640
    },
    {
      "epoch": 55.94,
      "grad_norm": 2.1425931453704834,
      "learning_rate": 4.043348361352601e-06,
      "loss": 0.5448,
      "step": 136650
    },
    {
      "epoch": 55.94,
      "grad_norm": 2.4164223670959473,
      "learning_rate": 4.043239128589361e-06,
      "loss": 0.5387,
      "step": 136660
    },
    {
      "epoch": 55.94,
      "grad_norm": 1.7309859991073608,
      "learning_rate": 4.043129888041023e-06,
      "loss": 0.5219,
      "step": 136670
    },
    {
      "epoch": 55.95,
      "grad_norm": 2.10347580909729,
      "learning_rate": 4.043020639708089e-06,
      "loss": 0.5635,
      "step": 136680
    },
    {
      "epoch": 55.95,
      "grad_norm": 1.6370898485183716,
      "learning_rate": 4.0429113835910615e-06,
      "loss": 0.5443,
      "step": 136690
    },
    {
      "epoch": 55.96,
      "grad_norm": 2.277146339416504,
      "learning_rate": 4.042802119690437e-06,
      "loss": 0.536,
      "step": 136700
    },
    {
      "epoch": 55.96,
      "grad_norm": 2.1949715614318848,
      "learning_rate": 4.0426928480067176e-06,
      "loss": 0.5471,
      "step": 136710
    },
    {
      "epoch": 55.96,
      "grad_norm": 2.0722594261169434,
      "learning_rate": 4.042583568540405e-06,
      "loss": 0.5372,
      "step": 136720
    },
    {
      "epoch": 55.97,
      "grad_norm": 2.7383315563201904,
      "learning_rate": 4.0424742812919985e-06,
      "loss": 0.5229,
      "step": 136730
    },
    {
      "epoch": 55.97,
      "grad_norm": 2.5388336181640625,
      "learning_rate": 4.0423649862619994e-06,
      "loss": 0.5209,
      "step": 136740
    },
    {
      "epoch": 55.98,
      "grad_norm": 3.1685359477996826,
      "learning_rate": 4.042255683450908e-06,
      "loss": 0.5311,
      "step": 136750
    },
    {
      "epoch": 55.98,
      "grad_norm": 1.6487935781478882,
      "learning_rate": 4.042146372859225e-06,
      "loss": 0.5408,
      "step": 136760
    },
    {
      "epoch": 55.98,
      "grad_norm": 1.9244451522827148,
      "learning_rate": 4.042037054487452e-06,
      "loss": 0.5224,
      "step": 136770
    },
    {
      "epoch": 55.99,
      "grad_norm": 1.6784995794296265,
      "learning_rate": 4.041927728336088e-06,
      "loss": 0.521,
      "step": 136780
    },
    {
      "epoch": 55.99,
      "grad_norm": 1.8422915935516357,
      "learning_rate": 4.041818394405635e-06,
      "loss": 0.5358,
      "step": 136790
    },
    {
      "epoch": 56.0,
      "grad_norm": 1.8843945264816284,
      "learning_rate": 4.041709052696595e-06,
      "loss": 0.5227,
      "step": 136800
    },
    {
      "epoch": 56.0,
      "eval_loss": 0.5344858765602112,
      "eval_runtime": 52.6062,
      "eval_samples_per_second": 65.563,
      "eval_steps_per_second": 8.212,
      "step": 136808
    },
    {
      "epoch": 56.0,
      "grad_norm": 2.5536906719207764,
      "learning_rate": 4.041599703209466e-06,
      "loss": 0.5384,
      "step": 136810
    },
    {
      "epoch": 56.0,
      "grad_norm": 1.8830580711364746,
      "learning_rate": 4.04149034594475e-06,
      "loss": 0.5316,
      "step": 136820
    },
    {
      "epoch": 56.01,
      "grad_norm": 1.7267619371414185,
      "learning_rate": 4.04138098090295e-06,
      "loss": 0.5224,
      "step": 136830
    },
    {
      "epoch": 56.01,
      "grad_norm": 1.9529515504837036,
      "learning_rate": 4.0412716080845645e-06,
      "loss": 0.5256,
      "step": 136840
    },
    {
      "epoch": 56.02,
      "grad_norm": 2.002340078353882,
      "learning_rate": 4.0411622274900965e-06,
      "loss": 0.5502,
      "step": 136850
    },
    {
      "epoch": 56.02,
      "grad_norm": 1.9914350509643555,
      "learning_rate": 4.041052839120045e-06,
      "loss": 0.5335,
      "step": 136860
    },
    {
      "epoch": 56.03,
      "grad_norm": 2.118866205215454,
      "learning_rate": 4.040943442974912e-06,
      "loss": 0.5226,
      "step": 136870
    },
    {
      "epoch": 56.03,
      "grad_norm": 2.2203528881073,
      "learning_rate": 4.0408340390551984e-06,
      "loss": 0.5446,
      "step": 136880
    },
    {
      "epoch": 56.03,
      "grad_norm": 1.989494800567627,
      "learning_rate": 4.040724627361406e-06,
      "loss": 0.5526,
      "step": 136890
    },
    {
      "epoch": 56.04,
      "grad_norm": 1.752736210823059,
      "learning_rate": 4.040615207894035e-06,
      "loss": 0.5386,
      "step": 136900
    },
    {
      "epoch": 56.04,
      "grad_norm": 1.602746844291687,
      "learning_rate": 4.040505780653588e-06,
      "loss": 0.5233,
      "step": 136910
    },
    {
      "epoch": 56.05,
      "grad_norm": 2.7420103549957275,
      "learning_rate": 4.040396345640566e-06,
      "loss": 0.5151,
      "step": 136920
    },
    {
      "epoch": 56.05,
      "grad_norm": 1.6452746391296387,
      "learning_rate": 4.040286902855469e-06,
      "loss": 0.5293,
      "step": 136930
    },
    {
      "epoch": 56.05,
      "grad_norm": 2.3762917518615723,
      "learning_rate": 4.0401774522988e-06,
      "loss": 0.5389,
      "step": 136940
    },
    {
      "epoch": 56.06,
      "grad_norm": 2.1215083599090576,
      "learning_rate": 4.040067993971059e-06,
      "loss": 0.5538,
      "step": 136950
    },
    {
      "epoch": 56.06,
      "grad_norm": 2.268995523452759,
      "learning_rate": 4.039958527872748e-06,
      "loss": 0.5184,
      "step": 136960
    },
    {
      "epoch": 56.07,
      "grad_norm": 2.4404947757720947,
      "learning_rate": 4.039849054004368e-06,
      "loss": 0.5233,
      "step": 136970
    },
    {
      "epoch": 56.07,
      "grad_norm": 2.0622968673706055,
      "learning_rate": 4.039739572366422e-06,
      "loss": 0.5304,
      "step": 136980
    },
    {
      "epoch": 56.07,
      "grad_norm": 2.3209357261657715,
      "learning_rate": 4.0396300829594096e-06,
      "loss": 0.5381,
      "step": 136990
    },
    {
      "epoch": 56.08,
      "grad_norm": 2.8317244052886963,
      "learning_rate": 4.039520585783834e-06,
      "loss": 0.5464,
      "step": 137000
    },
    {
      "epoch": 56.08,
      "grad_norm": 1.9648089408874512,
      "learning_rate": 4.039411080840195e-06,
      "loss": 0.5139,
      "step": 137010
    },
    {
      "epoch": 56.09,
      "grad_norm": 1.5653977394104004,
      "learning_rate": 4.039301568128996e-06,
      "loss": 0.5322,
      "step": 137020
    },
    {
      "epoch": 56.09,
      "grad_norm": 1.992710828781128,
      "learning_rate": 4.039192047650737e-06,
      "loss": 0.5205,
      "step": 137030
    },
    {
      "epoch": 56.09,
      "grad_norm": 1.9464211463928223,
      "learning_rate": 4.039082519405922e-06,
      "loss": 0.5245,
      "step": 137040
    },
    {
      "epoch": 56.1,
      "grad_norm": 2.282118558883667,
      "learning_rate": 4.03897298339505e-06,
      "loss": 0.5171,
      "step": 137050
    },
    {
      "epoch": 56.1,
      "grad_norm": 2.2964026927948,
      "learning_rate": 4.038863439618624e-06,
      "loss": 0.522,
      "step": 137060
    },
    {
      "epoch": 56.11,
      "grad_norm": 1.7826911211013794,
      "learning_rate": 4.038753888077146e-06,
      "loss": 0.5462,
      "step": 137070
    },
    {
      "epoch": 56.11,
      "grad_norm": 1.7753455638885498,
      "learning_rate": 4.038644328771117e-06,
      "loss": 0.5455,
      "step": 137080
    },
    {
      "epoch": 56.12,
      "grad_norm": 1.8744306564331055,
      "learning_rate": 4.0385347617010415e-06,
      "loss": 0.5479,
      "step": 137090
    },
    {
      "epoch": 56.12,
      "grad_norm": 2.0466468334198,
      "learning_rate": 4.038425186867418e-06,
      "loss": 0.5283,
      "step": 137100
    },
    {
      "epoch": 56.12,
      "grad_norm": 1.6193373203277588,
      "learning_rate": 4.03831560427075e-06,
      "loss": 0.5351,
      "step": 137110
    },
    {
      "epoch": 56.13,
      "grad_norm": 1.6740479469299316,
      "learning_rate": 4.03820601391154e-06,
      "loss": 0.5385,
      "step": 137120
    },
    {
      "epoch": 56.13,
      "grad_norm": 1.7000023126602173,
      "learning_rate": 4.03809641579029e-06,
      "loss": 0.5394,
      "step": 137130
    },
    {
      "epoch": 56.14,
      "grad_norm": 1.788615345954895,
      "learning_rate": 4.0379868099075e-06,
      "loss": 0.5491,
      "step": 137140
    },
    {
      "epoch": 56.14,
      "grad_norm": 1.9398750066757202,
      "learning_rate": 4.037877196263674e-06,
      "loss": 0.532,
      "step": 137150
    },
    {
      "epoch": 56.14,
      "grad_norm": 1.754002332687378,
      "learning_rate": 4.037767574859315e-06,
      "loss": 0.528,
      "step": 137160
    },
    {
      "epoch": 56.15,
      "grad_norm": 1.9923298358917236,
      "learning_rate": 4.0376579456949225e-06,
      "loss": 0.5362,
      "step": 137170
    },
    {
      "epoch": 56.15,
      "grad_norm": 1.6683595180511475,
      "learning_rate": 4.037548308770999e-06,
      "loss": 0.5245,
      "step": 137180
    },
    {
      "epoch": 56.16,
      "grad_norm": 2.4421162605285645,
      "learning_rate": 4.03743866408805e-06,
      "loss": 0.5124,
      "step": 137190
    },
    {
      "epoch": 56.16,
      "grad_norm": 1.590203881263733,
      "learning_rate": 4.037329011646574e-06,
      "loss": 0.5129,
      "step": 137200
    },
    {
      "epoch": 56.16,
      "grad_norm": 1.930492639541626,
      "learning_rate": 4.037219351447076e-06,
      "loss": 0.5343,
      "step": 137210
    },
    {
      "epoch": 56.17,
      "grad_norm": 1.4190938472747803,
      "learning_rate": 4.037109683490056e-06,
      "loss": 0.5157,
      "step": 137220
    },
    {
      "epoch": 56.17,
      "grad_norm": 1.8097760677337646,
      "learning_rate": 4.037000007776019e-06,
      "loss": 0.5469,
      "step": 137230
    },
    {
      "epoch": 56.18,
      "grad_norm": 1.5224460363388062,
      "learning_rate": 4.036890324305465e-06,
      "loss": 0.5163,
      "step": 137240
    },
    {
      "epoch": 56.18,
      "grad_norm": 1.6990249156951904,
      "learning_rate": 4.036780633078897e-06,
      "loss": 0.5291,
      "step": 137250
    },
    {
      "epoch": 56.19,
      "grad_norm": 2.056739091873169,
      "learning_rate": 4.036670934096818e-06,
      "loss": 0.5346,
      "step": 137260
    },
    {
      "epoch": 56.19,
      "grad_norm": 2.8117685317993164,
      "learning_rate": 4.036561227359731e-06,
      "loss": 0.5525,
      "step": 137270
    },
    {
      "epoch": 56.19,
      "grad_norm": 2.1467485427856445,
      "learning_rate": 4.036451512868137e-06,
      "loss": 0.5444,
      "step": 137280
    },
    {
      "epoch": 56.2,
      "grad_norm": 1.9348708391189575,
      "learning_rate": 4.036341790622541e-06,
      "loss": 0.5091,
      "step": 137290
    },
    {
      "epoch": 56.2,
      "grad_norm": 2.317841053009033,
      "learning_rate": 4.036232060623443e-06,
      "loss": 0.5314,
      "step": 137300
    },
    {
      "epoch": 56.21,
      "grad_norm": 1.8876854181289673,
      "learning_rate": 4.0361223228713475e-06,
      "loss": 0.4956,
      "step": 137310
    },
    {
      "epoch": 56.21,
      "grad_norm": 2.0123703479766846,
      "learning_rate": 4.036012577366755e-06,
      "loss": 0.5335,
      "step": 137320
    },
    {
      "epoch": 56.21,
      "grad_norm": 2.0690975189208984,
      "learning_rate": 4.035902824110172e-06,
      "loss": 0.5312,
      "step": 137330
    },
    {
      "epoch": 56.22,
      "grad_norm": 2.071775436401367,
      "learning_rate": 4.0357930631020974e-06,
      "loss": 0.5393,
      "step": 137340
    },
    {
      "epoch": 56.22,
      "grad_norm": 2.223008394241333,
      "learning_rate": 4.0356832943430356e-06,
      "loss": 0.5401,
      "step": 137350
    },
    {
      "epoch": 56.23,
      "grad_norm": 2.1077451705932617,
      "learning_rate": 4.035573517833489e-06,
      "loss": 0.5396,
      "step": 137360
    },
    {
      "epoch": 56.23,
      "grad_norm": 1.2777507305145264,
      "learning_rate": 4.0354637335739624e-06,
      "loss": 0.5239,
      "step": 137370
    },
    {
      "epoch": 56.23,
      "grad_norm": 2.087411403656006,
      "learning_rate": 4.035353941564957e-06,
      "loss": 0.5213,
      "step": 137380
    },
    {
      "epoch": 56.24,
      "grad_norm": 2.499988079071045,
      "learning_rate": 4.035244141806975e-06,
      "loss": 0.5288,
      "step": 137390
    },
    {
      "epoch": 56.24,
      "grad_norm": 1.4877325296401978,
      "learning_rate": 4.0351343343005205e-06,
      "loss": 0.5535,
      "step": 137400
    },
    {
      "epoch": 56.25,
      "grad_norm": 2.633960247039795,
      "learning_rate": 4.035024519046096e-06,
      "loss": 0.5171,
      "step": 137410
    },
    {
      "epoch": 56.25,
      "grad_norm": 2.005901575088501,
      "learning_rate": 4.034914696044206e-06,
      "loss": 0.5275,
      "step": 137420
    },
    {
      "epoch": 56.25,
      "grad_norm": 1.8879810571670532,
      "learning_rate": 4.034804865295351e-06,
      "loss": 0.5242,
      "step": 137430
    },
    {
      "epoch": 56.26,
      "grad_norm": 2.2120449542999268,
      "learning_rate": 4.034695026800036e-06,
      "loss": 0.5197,
      "step": 137440
    },
    {
      "epoch": 56.26,
      "grad_norm": 2.316632032394409,
      "learning_rate": 4.034585180558764e-06,
      "loss": 0.5392,
      "step": 137450
    },
    {
      "epoch": 56.27,
      "grad_norm": 2.9741384983062744,
      "learning_rate": 4.034475326572038e-06,
      "loss": 0.5212,
      "step": 137460
    },
    {
      "epoch": 56.27,
      "grad_norm": 1.949123740196228,
      "learning_rate": 4.034365464840362e-06,
      "loss": 0.5353,
      "step": 137470
    },
    {
      "epoch": 56.28,
      "grad_norm": 1.9297316074371338,
      "learning_rate": 4.0342555953642365e-06,
      "loss": 0.5284,
      "step": 137480
    },
    {
      "epoch": 56.28,
      "grad_norm": 2.0877370834350586,
      "learning_rate": 4.034145718144168e-06,
      "loss": 0.5328,
      "step": 137490
    },
    {
      "epoch": 56.28,
      "grad_norm": 2.829275608062744,
      "learning_rate": 4.034035833180657e-06,
      "loss": 0.5095,
      "step": 137500
    },
    {
      "epoch": 56.29,
      "grad_norm": 1.8155518770217896,
      "learning_rate": 4.033925940474211e-06,
      "loss": 0.5304,
      "step": 137510
    },
    {
      "epoch": 56.29,
      "grad_norm": 2.2382118701934814,
      "learning_rate": 4.033816040025328e-06,
      "loss": 0.5242,
      "step": 137520
    },
    {
      "epoch": 56.3,
      "grad_norm": 2.706611156463623,
      "learning_rate": 4.033706131834515e-06,
      "loss": 0.5276,
      "step": 137530
    },
    {
      "epoch": 56.3,
      "grad_norm": 1.1759880781173706,
      "learning_rate": 4.033596215902276e-06,
      "loss": 0.5318,
      "step": 137540
    },
    {
      "epoch": 56.3,
      "grad_norm": 2.84123158454895,
      "learning_rate": 4.0334862922291125e-06,
      "loss": 0.5505,
      "step": 137550
    },
    {
      "epoch": 56.31,
      "grad_norm": 2.0464775562286377,
      "learning_rate": 4.033376360815528e-06,
      "loss": 0.5507,
      "step": 137560
    },
    {
      "epoch": 56.31,
      "grad_norm": 2.2488527297973633,
      "learning_rate": 4.0332664216620274e-06,
      "loss": 0.5226,
      "step": 137570
    },
    {
      "epoch": 56.32,
      "grad_norm": 2.3271565437316895,
      "learning_rate": 4.033156474769114e-06,
      "loss": 0.5193,
      "step": 137580
    },
    {
      "epoch": 56.32,
      "grad_norm": 2.0350730419158936,
      "learning_rate": 4.03304652013729e-06,
      "loss": 0.5219,
      "step": 137590
    },
    {
      "epoch": 56.32,
      "grad_norm": 2.201167106628418,
      "learning_rate": 4.0329365577670615e-06,
      "loss": 0.5638,
      "step": 137600
    },
    {
      "epoch": 56.33,
      "grad_norm": 1.9762582778930664,
      "learning_rate": 4.032826587658931e-06,
      "loss": 0.5284,
      "step": 137610
    },
    {
      "epoch": 56.33,
      "grad_norm": 3.0091543197631836,
      "learning_rate": 4.0327166098134015e-06,
      "loss": 0.5516,
      "step": 137620
    },
    {
      "epoch": 56.34,
      "grad_norm": 1.9605762958526611,
      "learning_rate": 4.032606624230978e-06,
      "loss": 0.5328,
      "step": 137630
    },
    {
      "epoch": 56.34,
      "grad_norm": 2.2347240447998047,
      "learning_rate": 4.032496630912163e-06,
      "loss": 0.5334,
      "step": 137640
    },
    {
      "epoch": 56.34,
      "grad_norm": 1.908545970916748,
      "learning_rate": 4.032386629857463e-06,
      "loss": 0.5478,
      "step": 137650
    },
    {
      "epoch": 56.35,
      "grad_norm": 1.7600693702697754,
      "learning_rate": 4.032276621067378e-06,
      "loss": 0.5161,
      "step": 137660
    },
    {
      "epoch": 56.35,
      "grad_norm": 1.956824779510498,
      "learning_rate": 4.032166604542416e-06,
      "loss": 0.5432,
      "step": 137670
    },
    {
      "epoch": 56.36,
      "grad_norm": 2.0291810035705566,
      "learning_rate": 4.032056580283077e-06,
      "loss": 0.5437,
      "step": 137680
    },
    {
      "epoch": 56.36,
      "grad_norm": 1.599443793296814,
      "learning_rate": 4.031946548289869e-06,
      "loss": 0.5449,
      "step": 137690
    },
    {
      "epoch": 56.37,
      "grad_norm": 2.045743227005005,
      "learning_rate": 4.031836508563293e-06,
      "loss": 0.5144,
      "step": 137700
    },
    {
      "epoch": 56.37,
      "grad_norm": 1.9277071952819824,
      "learning_rate": 4.031726461103855e-06,
      "loss": 0.5378,
      "step": 137710
    },
    {
      "epoch": 56.37,
      "grad_norm": 1.9754278659820557,
      "learning_rate": 4.031616405912057e-06,
      "loss": 0.5424,
      "step": 137720
    },
    {
      "epoch": 56.38,
      "grad_norm": 1.3389607667922974,
      "learning_rate": 4.031506342988404e-06,
      "loss": 0.5317,
      "step": 137730
    },
    {
      "epoch": 56.38,
      "grad_norm": 1.876191258430481,
      "learning_rate": 4.0313962723334016e-06,
      "loss": 0.5275,
      "step": 137740
    },
    {
      "epoch": 56.39,
      "grad_norm": 2.2880966663360596,
      "learning_rate": 4.031286193947554e-06,
      "loss": 0.5193,
      "step": 137750
    },
    {
      "epoch": 56.39,
      "grad_norm": 1.3791800737380981,
      "learning_rate": 4.031176107831363e-06,
      "loss": 0.5539,
      "step": 137760
    },
    {
      "epoch": 56.39,
      "grad_norm": 1.6778146028518677,
      "learning_rate": 4.0310660139853345e-06,
      "loss": 0.5325,
      "step": 137770
    },
    {
      "epoch": 56.4,
      "grad_norm": 1.7618095874786377,
      "learning_rate": 4.030955912409973e-06,
      "loss": 0.5358,
      "step": 137780
    },
    {
      "epoch": 56.4,
      "grad_norm": 1.904000163078308,
      "learning_rate": 4.030845803105782e-06,
      "loss": 0.5162,
      "step": 137790
    },
    {
      "epoch": 56.41,
      "grad_norm": 1.8304076194763184,
      "learning_rate": 4.030735686073268e-06,
      "loss": 0.535,
      "step": 137800
    },
    {
      "epoch": 56.41,
      "grad_norm": 2.5491271018981934,
      "learning_rate": 4.030625561312932e-06,
      "loss": 0.5222,
      "step": 137810
    },
    {
      "epoch": 56.41,
      "grad_norm": 7.6976141929626465,
      "learning_rate": 4.0305154288252806e-06,
      "loss": 0.5169,
      "step": 137820
    },
    {
      "epoch": 56.42,
      "grad_norm": 1.945586919784546,
      "learning_rate": 4.030405288610818e-06,
      "loss": 0.5165,
      "step": 137830
    },
    {
      "epoch": 56.42,
      "grad_norm": 2.0950725078582764,
      "learning_rate": 4.0302951406700496e-06,
      "loss": 0.5155,
      "step": 137840
    },
    {
      "epoch": 56.43,
      "grad_norm": 2.2841200828552246,
      "learning_rate": 4.030184985003479e-06,
      "loss": 0.5327,
      "step": 137850
    },
    {
      "epoch": 56.43,
      "grad_norm": 1.8153095245361328,
      "learning_rate": 4.03007482161161e-06,
      "loss": 0.5193,
      "step": 137860
    },
    {
      "epoch": 56.43,
      "grad_norm": 1.6317369937896729,
      "learning_rate": 4.029964650494948e-06,
      "loss": 0.528,
      "step": 137870
    },
    {
      "epoch": 56.44,
      "grad_norm": 1.9350669384002686,
      "learning_rate": 4.029854471653999e-06,
      "loss": 0.5289,
      "step": 137880
    },
    {
      "epoch": 56.44,
      "grad_norm": 2.579334020614624,
      "learning_rate": 4.029744285089266e-06,
      "loss": 0.5337,
      "step": 137890
    },
    {
      "epoch": 56.45,
      "grad_norm": 2.5861194133758545,
      "learning_rate": 4.029634090801254e-06,
      "loss": 0.5218,
      "step": 137900
    },
    {
      "epoch": 56.45,
      "grad_norm": 1.7976597547531128,
      "learning_rate": 4.0295238887904685e-06,
      "loss": 0.5147,
      "step": 137910
    },
    {
      "epoch": 56.46,
      "grad_norm": 1.5143406391143799,
      "learning_rate": 4.029413679057414e-06,
      "loss": 0.5252,
      "step": 137920
    },
    {
      "epoch": 56.46,
      "grad_norm": 1.816734790802002,
      "learning_rate": 4.029303461602595e-06,
      "loss": 0.5309,
      "step": 137930
    },
    {
      "epoch": 56.46,
      "grad_norm": 2.5266611576080322,
      "learning_rate": 4.029193236426517e-06,
      "loss": 0.5374,
      "step": 137940
    },
    {
      "epoch": 56.47,
      "grad_norm": 1.880052924156189,
      "learning_rate": 4.029083003529683e-06,
      "loss": 0.5094,
      "step": 137950
    },
    {
      "epoch": 56.47,
      "grad_norm": 2.257080554962158,
      "learning_rate": 4.028972762912601e-06,
      "loss": 0.5097,
      "step": 137960
    },
    {
      "epoch": 56.48,
      "grad_norm": 1.7113410234451294,
      "learning_rate": 4.028862514575775e-06,
      "loss": 0.5047,
      "step": 137970
    },
    {
      "epoch": 56.48,
      "grad_norm": 1.8939409255981445,
      "learning_rate": 4.028752258519708e-06,
      "loss": 0.5106,
      "step": 137980
    },
    {
      "epoch": 56.48,
      "grad_norm": 1.6739375591278076,
      "learning_rate": 4.028641994744907e-06,
      "loss": 0.5391,
      "step": 137990
    },
    {
      "epoch": 56.49,
      "grad_norm": 1.7505210638046265,
      "learning_rate": 4.0285317232518776e-06,
      "loss": 0.5135,
      "step": 138000
    },
    {
      "epoch": 56.49,
      "grad_norm": 2.1291167736053467,
      "learning_rate": 4.028421444041123e-06,
      "loss": 0.5084,
      "step": 138010
    },
    {
      "epoch": 56.5,
      "grad_norm": 1.5452266931533813,
      "learning_rate": 4.02831115711315e-06,
      "loss": 0.5416,
      "step": 138020
    },
    {
      "epoch": 56.5,
      "grad_norm": 2.0375313758850098,
      "learning_rate": 4.028200862468464e-06,
      "loss": 0.5396,
      "step": 138030
    },
    {
      "epoch": 56.5,
      "grad_norm": 2.058053731918335,
      "learning_rate": 4.028090560107567e-06,
      "loss": 0.5423,
      "step": 138040
    },
    {
      "epoch": 56.51,
      "grad_norm": 1.549747109413147,
      "learning_rate": 4.02798025003097e-06,
      "loss": 0.5155,
      "step": 138050
    },
    {
      "epoch": 56.51,
      "grad_norm": 1.946806788444519,
      "learning_rate": 4.027869932239173e-06,
      "loss": 0.5124,
      "step": 138060
    },
    {
      "epoch": 56.52,
      "grad_norm": 2.0420687198638916,
      "learning_rate": 4.0277596067326835e-06,
      "loss": 0.5265,
      "step": 138070
    },
    {
      "epoch": 56.52,
      "grad_norm": 1.8439974784851074,
      "learning_rate": 4.027649273512007e-06,
      "loss": 0.5293,
      "step": 138080
    },
    {
      "epoch": 56.52,
      "grad_norm": 2.153489828109741,
      "learning_rate": 4.0275389325776495e-06,
      "loss": 0.5111,
      "step": 138090
    },
    {
      "epoch": 56.53,
      "grad_norm": 2.603569269180298,
      "learning_rate": 4.027428583930115e-06,
      "loss": 0.5427,
      "step": 138100
    },
    {
      "epoch": 56.53,
      "grad_norm": 2.205723524093628,
      "learning_rate": 4.02731822756991e-06,
      "loss": 0.5373,
      "step": 138110
    },
    {
      "epoch": 56.54,
      "grad_norm": 1.779257893562317,
      "learning_rate": 4.027207863497539e-06,
      "loss": 0.5163,
      "step": 138120
    },
    {
      "epoch": 56.54,
      "grad_norm": 2.07289719581604,
      "learning_rate": 4.027097491713509e-06,
      "loss": 0.5354,
      "step": 138130
    },
    {
      "epoch": 56.55,
      "grad_norm": 2.6157076358795166,
      "learning_rate": 4.026987112218324e-06,
      "loss": 0.5635,
      "step": 138140
    },
    {
      "epoch": 56.55,
      "grad_norm": 2.1665701866149902,
      "learning_rate": 4.026876725012491e-06,
      "loss": 0.5225,
      "step": 138150
    },
    {
      "epoch": 56.55,
      "grad_norm": 2.287041187286377,
      "learning_rate": 4.026766330096514e-06,
      "loss": 0.5212,
      "step": 138160
    },
    {
      "epoch": 56.56,
      "grad_norm": 1.7019613981246948,
      "learning_rate": 4.026655927470902e-06,
      "loss": 0.5184,
      "step": 138170
    },
    {
      "epoch": 56.56,
      "grad_norm": 1.4696412086486816,
      "learning_rate": 4.026545517136157e-06,
      "loss": 0.5248,
      "step": 138180
    },
    {
      "epoch": 56.57,
      "grad_norm": 1.8063342571258545,
      "learning_rate": 4.026435099092787e-06,
      "loss": 0.5297,
      "step": 138190
    },
    {
      "epoch": 56.57,
      "grad_norm": 1.9185309410095215,
      "learning_rate": 4.026324673341297e-06,
      "loss": 0.5149,
      "step": 138200
    },
    {
      "epoch": 56.57,
      "grad_norm": 2.062462091445923,
      "learning_rate": 4.0262142398821935e-06,
      "loss": 0.528,
      "step": 138210
    },
    {
      "epoch": 56.58,
      "grad_norm": 1.564926028251648,
      "learning_rate": 4.026103798715981e-06,
      "loss": 0.5245,
      "step": 138220
    },
    {
      "epoch": 56.58,
      "grad_norm": 2.3494253158569336,
      "learning_rate": 4.025993349843167e-06,
      "loss": 0.5356,
      "step": 138230
    },
    {
      "epoch": 56.59,
      "grad_norm": 1.5877894163131714,
      "learning_rate": 4.025882893264255e-06,
      "loss": 0.5297,
      "step": 138240
    },
    {
      "epoch": 56.59,
      "grad_norm": 2.4812424182891846,
      "learning_rate": 4.0257724289797545e-06,
      "loss": 0.5241,
      "step": 138250
    },
    {
      "epoch": 56.59,
      "grad_norm": 2.0888519287109375,
      "learning_rate": 4.0256619569901696e-06,
      "loss": 0.5436,
      "step": 138260
    },
    {
      "epoch": 56.6,
      "grad_norm": 1.83463716506958,
      "learning_rate": 4.025551477296006e-06,
      "loss": 0.5508,
      "step": 138270
    },
    {
      "epoch": 56.6,
      "grad_norm": 1.4657785892486572,
      "learning_rate": 4.025440989897771e-06,
      "loss": 0.5275,
      "step": 138280
    },
    {
      "epoch": 56.61,
      "grad_norm": 1.9293850660324097,
      "learning_rate": 4.02533049479597e-06,
      "loss": 0.5234,
      "step": 138290
    },
    {
      "epoch": 56.61,
      "grad_norm": 1.9380254745483398,
      "learning_rate": 4.025219991991108e-06,
      "loss": 0.5085,
      "step": 138300
    },
    {
      "epoch": 56.61,
      "grad_norm": 2.1245338916778564,
      "learning_rate": 4.025109481483693e-06,
      "loss": 0.5507,
      "step": 138310
    },
    {
      "epoch": 56.62,
      "grad_norm": 1.9779489040374756,
      "learning_rate": 4.024998963274231e-06,
      "loss": 0.5476,
      "step": 138320
    },
    {
      "epoch": 56.62,
      "grad_norm": 2.561290740966797,
      "learning_rate": 4.024888437363228e-06,
      "loss": 0.5232,
      "step": 138330
    },
    {
      "epoch": 56.63,
      "grad_norm": 1.841191053390503,
      "learning_rate": 4.024777903751189e-06,
      "loss": 0.5267,
      "step": 138340
    },
    {
      "epoch": 56.63,
      "grad_norm": 1.7968101501464844,
      "learning_rate": 4.024667362438622e-06,
      "loss": 0.5118,
      "step": 138350
    },
    {
      "epoch": 56.64,
      "grad_norm": 1.6510469913482666,
      "learning_rate": 4.024556813426034e-06,
      "loss": 0.5374,
      "step": 138360
    },
    {
      "epoch": 56.64,
      "grad_norm": 2.5000834465026855,
      "learning_rate": 4.024446256713929e-06,
      "loss": 0.52,
      "step": 138370
    },
    {
      "epoch": 56.64,
      "grad_norm": 2.6214563846588135,
      "learning_rate": 4.024335692302815e-06,
      "loss": 0.5158,
      "step": 138380
    },
    {
      "epoch": 56.65,
      "grad_norm": 2.00930118560791,
      "learning_rate": 4.024225120193199e-06,
      "loss": 0.5177,
      "step": 138390
    },
    {
      "epoch": 56.65,
      "grad_norm": 2.2981088161468506,
      "learning_rate": 4.024114540385586e-06,
      "loss": 0.5685,
      "step": 138400
    },
    {
      "epoch": 56.66,
      "grad_norm": 1.6046146154403687,
      "learning_rate": 4.024003952880484e-06,
      "loss": 0.5298,
      "step": 138410
    },
    {
      "epoch": 56.66,
      "grad_norm": 2.1347858905792236,
      "learning_rate": 4.023893357678398e-06,
      "loss": 0.532,
      "step": 138420
    },
    {
      "epoch": 56.66,
      "grad_norm": 1.6597535610198975,
      "learning_rate": 4.0237827547798355e-06,
      "loss": 0.553,
      "step": 138430
    },
    {
      "epoch": 56.67,
      "grad_norm": 2.199218511581421,
      "learning_rate": 4.0236721441853036e-06,
      "loss": 0.5403,
      "step": 138440
    },
    {
      "epoch": 56.67,
      "grad_norm": 1.8773300647735596,
      "learning_rate": 4.023561525895308e-06,
      "loss": 0.5136,
      "step": 138450
    },
    {
      "epoch": 56.68,
      "grad_norm": 2.7535624504089355,
      "learning_rate": 4.023450899910356e-06,
      "loss": 0.544,
      "step": 138460
    },
    {
      "epoch": 56.68,
      "grad_norm": 1.7957642078399658,
      "learning_rate": 4.023340266230956e-06,
      "loss": 0.5522,
      "step": 138470
    },
    {
      "epoch": 56.68,
      "grad_norm": 2.2374119758605957,
      "learning_rate": 4.0232296248576105e-06,
      "loss": 0.5235,
      "step": 138480
    },
    {
      "epoch": 56.69,
      "grad_norm": 2.117408514022827,
      "learning_rate": 4.02311897579083e-06,
      "loss": 0.5485,
      "step": 138490
    },
    {
      "epoch": 56.69,
      "grad_norm": 2.2580041885375977,
      "learning_rate": 4.023008319031121e-06,
      "loss": 0.54,
      "step": 138500
    },
    {
      "epoch": 56.7,
      "grad_norm": 2.254387617111206,
      "learning_rate": 4.02289765457899e-06,
      "loss": 0.5118,
      "step": 138510
    },
    {
      "epoch": 56.7,
      "grad_norm": 1.9768071174621582,
      "learning_rate": 4.022786982434942e-06,
      "loss": 0.5364,
      "step": 138520
    },
    {
      "epoch": 56.7,
      "grad_norm": 2.1470224857330322,
      "learning_rate": 4.022676302599486e-06,
      "loss": 0.5351,
      "step": 138530
    },
    {
      "epoch": 56.71,
      "grad_norm": 2.022779703140259,
      "learning_rate": 4.022565615073129e-06,
      "loss": 0.5375,
      "step": 138540
    },
    {
      "epoch": 56.71,
      "grad_norm": 1.9702043533325195,
      "learning_rate": 4.022454919856378e-06,
      "loss": 0.5417,
      "step": 138550
    },
    {
      "epoch": 56.72,
      "grad_norm": 1.8242136240005493,
      "learning_rate": 4.022344216949739e-06,
      "loss": 0.5341,
      "step": 138560
    },
    {
      "epoch": 56.72,
      "grad_norm": 1.6725006103515625,
      "learning_rate": 4.0222335063537205e-06,
      "loss": 0.5331,
      "step": 138570
    },
    {
      "epoch": 56.73,
      "grad_norm": 2.298245906829834,
      "learning_rate": 4.022122788068829e-06,
      "loss": 0.5356,
      "step": 138580
    },
    {
      "epoch": 56.73,
      "grad_norm": 1.7803122997283936,
      "learning_rate": 4.022012062095571e-06,
      "loss": 0.5155,
      "step": 138590
    },
    {
      "epoch": 56.73,
      "grad_norm": 2.339909076690674,
      "learning_rate": 4.021901328434455e-06,
      "loss": 0.5432,
      "step": 138600
    },
    {
      "epoch": 56.74,
      "grad_norm": 2.855778694152832,
      "learning_rate": 4.021790587085988e-06,
      "loss": 0.5436,
      "step": 138610
    },
    {
      "epoch": 56.74,
      "grad_norm": 2.0673446655273438,
      "learning_rate": 4.021679838050676e-06,
      "loss": 0.5149,
      "step": 138620
    },
    {
      "epoch": 56.75,
      "grad_norm": 1.6827905178070068,
      "learning_rate": 4.021569081329028e-06,
      "loss": 0.5169,
      "step": 138630
    },
    {
      "epoch": 56.75,
      "grad_norm": 2.260279417037964,
      "learning_rate": 4.02145831692155e-06,
      "loss": 0.5234,
      "step": 138640
    },
    {
      "epoch": 56.75,
      "grad_norm": 2.409102439880371,
      "learning_rate": 4.021347544828751e-06,
      "loss": 0.5317,
      "step": 138650
    },
    {
      "epoch": 56.76,
      "grad_norm": 1.9020583629608154,
      "learning_rate": 4.021236765051137e-06,
      "loss": 0.5247,
      "step": 138660
    },
    {
      "epoch": 56.76,
      "grad_norm": 1.691524624824524,
      "learning_rate": 4.021125977589216e-06,
      "loss": 0.5458,
      "step": 138670
    },
    {
      "epoch": 56.77,
      "grad_norm": 2.254277467727661,
      "learning_rate": 4.021015182443495e-06,
      "loss": 0.5218,
      "step": 138680
    },
    {
      "epoch": 56.77,
      "grad_norm": 1.5580775737762451,
      "learning_rate": 4.020904379614482e-06,
      "loss": 0.5406,
      "step": 138690
    },
    {
      "epoch": 56.77,
      "grad_norm": 1.9404765367507935,
      "learning_rate": 4.020793569102685e-06,
      "loss": 0.5106,
      "step": 138700
    },
    {
      "epoch": 56.78,
      "grad_norm": 2.655550003051758,
      "learning_rate": 4.02068275090861e-06,
      "loss": 0.5438,
      "step": 138710
    },
    {
      "epoch": 56.78,
      "grad_norm": 2.0104403495788574,
      "learning_rate": 4.020571925032767e-06,
      "loss": 0.5481,
      "step": 138720
    },
    {
      "epoch": 56.79,
      "grad_norm": 2.1486027240753174,
      "learning_rate": 4.020461091475663e-06,
      "loss": 0.541,
      "step": 138730
    },
    {
      "epoch": 56.79,
      "grad_norm": 2.42091965675354,
      "learning_rate": 4.020350250237804e-06,
      "loss": 0.5306,
      "step": 138740
    },
    {
      "epoch": 56.79,
      "grad_norm": 1.8738620281219482,
      "learning_rate": 4.020239401319698e-06,
      "loss": 0.5166,
      "step": 138750
    },
    {
      "epoch": 56.8,
      "grad_norm": 1.572594404220581,
      "learning_rate": 4.020128544721856e-06,
      "loss": 0.5151,
      "step": 138760
    },
    {
      "epoch": 56.8,
      "grad_norm": 2.5485680103302,
      "learning_rate": 4.020017680444782e-06,
      "loss": 0.5255,
      "step": 138770
    },
    {
      "epoch": 56.81,
      "grad_norm": 2.3040950298309326,
      "learning_rate": 4.019906808488986e-06,
      "loss": 0.5249,
      "step": 138780
    },
    {
      "epoch": 56.81,
      "grad_norm": 1.8457982540130615,
      "learning_rate": 4.019795928854974e-06,
      "loss": 0.5104,
      "step": 138790
    },
    {
      "epoch": 56.82,
      "grad_norm": 2.157106876373291,
      "learning_rate": 4.019685041543257e-06,
      "loss": 0.528,
      "step": 138800
    },
    {
      "epoch": 56.82,
      "grad_norm": 2.214296579360962,
      "learning_rate": 4.0195741465543395e-06,
      "loss": 0.5454,
      "step": 138810
    },
    {
      "epoch": 56.82,
      "grad_norm": 2.5698065757751465,
      "learning_rate": 4.019463243888732e-06,
      "loss": 0.5364,
      "step": 138820
    },
    {
      "epoch": 56.83,
      "grad_norm": 1.8101478815078735,
      "learning_rate": 4.0193523335469425e-06,
      "loss": 0.5453,
      "step": 138830
    },
    {
      "epoch": 56.83,
      "grad_norm": 2.573920965194702,
      "learning_rate": 4.0192414155294765e-06,
      "loss": 0.5224,
      "step": 138840
    },
    {
      "epoch": 56.84,
      "grad_norm": 2.081847906112671,
      "learning_rate": 4.019130489836844e-06,
      "loss": 0.5366,
      "step": 138850
    },
    {
      "epoch": 56.84,
      "grad_norm": 1.6998358964920044,
      "learning_rate": 4.019019556469555e-06,
      "loss": 0.5453,
      "step": 138860
    },
    {
      "epoch": 56.84,
      "grad_norm": 1.6267542839050293,
      "learning_rate": 4.0189086154281135e-06,
      "loss": 0.533,
      "step": 138870
    },
    {
      "epoch": 56.85,
      "grad_norm": 2.1959357261657715,
      "learning_rate": 4.01879766671303e-06,
      "loss": 0.5137,
      "step": 138880
    },
    {
      "epoch": 56.85,
      "grad_norm": 2.01275897026062,
      "learning_rate": 4.018686710324813e-06,
      "loss": 0.537,
      "step": 138890
    },
    {
      "epoch": 56.86,
      "grad_norm": 2.084705114364624,
      "learning_rate": 4.01857574626397e-06,
      "loss": 0.5379,
      "step": 138900
    },
    {
      "epoch": 56.86,
      "grad_norm": 2.0161359310150146,
      "learning_rate": 4.01846477453101e-06,
      "loss": 0.5178,
      "step": 138910
    },
    {
      "epoch": 56.86,
      "grad_norm": 1.8173242807388306,
      "learning_rate": 4.0183537951264406e-06,
      "loss": 0.5273,
      "step": 138920
    },
    {
      "epoch": 56.87,
      "grad_norm": 2.544257402420044,
      "learning_rate": 4.018242808050771e-06,
      "loss": 0.5394,
      "step": 138930
    },
    {
      "epoch": 56.87,
      "grad_norm": 2.0345799922943115,
      "learning_rate": 4.018131813304509e-06,
      "loss": 0.5387,
      "step": 138940
    },
    {
      "epoch": 56.88,
      "grad_norm": 2.520099401473999,
      "learning_rate": 4.018020810888163e-06,
      "loss": 0.527,
      "step": 138950
    },
    {
      "epoch": 56.88,
      "grad_norm": 2.2540385723114014,
      "learning_rate": 4.017909800802242e-06,
      "loss": 0.5406,
      "step": 138960
    },
    {
      "epoch": 56.88,
      "grad_norm": 1.6950103044509888,
      "learning_rate": 4.017798783047254e-06,
      "loss": 0.5456,
      "step": 138970
    },
    {
      "epoch": 56.89,
      "grad_norm": 2.2279322147369385,
      "learning_rate": 4.0176877576237085e-06,
      "loss": 0.5318,
      "step": 138980
    },
    {
      "epoch": 56.89,
      "grad_norm": 1.9265005588531494,
      "learning_rate": 4.017576724532112e-06,
      "loss": 0.5393,
      "step": 138990
    },
    {
      "epoch": 56.9,
      "grad_norm": 1.7795392274856567,
      "learning_rate": 4.017465683772976e-06,
      "loss": 0.5299,
      "step": 139000
    },
    {
      "epoch": 56.9,
      "grad_norm": 2.041080951690674,
      "learning_rate": 4.017354635346807e-06,
      "loss": 0.5306,
      "step": 139010
    },
    {
      "epoch": 56.91,
      "grad_norm": 2.761605978012085,
      "learning_rate": 4.017243579254114e-06,
      "loss": 0.5112,
      "step": 139020
    },
    {
      "epoch": 56.91,
      "grad_norm": 2.036954641342163,
      "learning_rate": 4.017132515495407e-06,
      "loss": 0.5202,
      "step": 139030
    },
    {
      "epoch": 56.91,
      "grad_norm": 1.603628158569336,
      "learning_rate": 4.017021444071193e-06,
      "loss": 0.5385,
      "step": 139040
    },
    {
      "epoch": 56.92,
      "grad_norm": 1.9316861629486084,
      "learning_rate": 4.016910364981982e-06,
      "loss": 0.5186,
      "step": 139050
    },
    {
      "epoch": 56.92,
      "grad_norm": 2.5485894680023193,
      "learning_rate": 4.0167992782282825e-06,
      "loss": 0.5431,
      "step": 139060
    },
    {
      "epoch": 56.93,
      "grad_norm": 1.9920539855957031,
      "learning_rate": 4.016688183810604e-06,
      "loss": 0.5169,
      "step": 139070
    },
    {
      "epoch": 56.93,
      "grad_norm": 2.4592854976654053,
      "learning_rate": 4.016577081729453e-06,
      "loss": 0.5405,
      "step": 139080
    },
    {
      "epoch": 56.93,
      "grad_norm": 1.8826534748077393,
      "learning_rate": 4.0164659719853415e-06,
      "loss": 0.555,
      "step": 139090
    },
    {
      "epoch": 56.94,
      "grad_norm": 2.271054267883301,
      "learning_rate": 4.016354854578777e-06,
      "loss": 0.5468,
      "step": 139100
    },
    {
      "epoch": 56.94,
      "grad_norm": 1.7015947103500366,
      "learning_rate": 4.016243729510268e-06,
      "loss": 0.5256,
      "step": 139110
    },
    {
      "epoch": 56.95,
      "grad_norm": 1.8397655487060547,
      "learning_rate": 4.016132596780325e-06,
      "loss": 0.5327,
      "step": 139120
    },
    {
      "epoch": 56.95,
      "grad_norm": 2.4936671257019043,
      "learning_rate": 4.016021456389457e-06,
      "loss": 0.5227,
      "step": 139130
    },
    {
      "epoch": 56.95,
      "grad_norm": 2.0185842514038086,
      "learning_rate": 4.015910308338172e-06,
      "loss": 0.5484,
      "step": 139140
    },
    {
      "epoch": 56.96,
      "grad_norm": 2.3500096797943115,
      "learning_rate": 4.015799152626979e-06,
      "loss": 0.5598,
      "step": 139150
    },
    {
      "epoch": 56.96,
      "grad_norm": 2.539191246032715,
      "learning_rate": 4.0156879892563874e-06,
      "loss": 0.5341,
      "step": 139160
    },
    {
      "epoch": 56.97,
      "grad_norm": 2.0919501781463623,
      "learning_rate": 4.015576818226908e-06,
      "loss": 0.5131,
      "step": 139170
    },
    {
      "epoch": 56.97,
      "grad_norm": 1.6524875164031982,
      "learning_rate": 4.015465639539048e-06,
      "loss": 0.536,
      "step": 139180
    },
    {
      "epoch": 56.98,
      "grad_norm": 2.003009557723999,
      "learning_rate": 4.015354453193318e-06,
      "loss": 0.5238,
      "step": 139190
    },
    {
      "epoch": 56.98,
      "grad_norm": 1.8544435501098633,
      "learning_rate": 4.015243259190226e-06,
      "loss": 0.5372,
      "step": 139200
    },
    {
      "epoch": 56.98,
      "grad_norm": 2.1714305877685547,
      "learning_rate": 4.015132057530284e-06,
      "loss": 0.5418,
      "step": 139210
    },
    {
      "epoch": 56.99,
      "grad_norm": 1.7244879007339478,
      "learning_rate": 4.0150208482139986e-06,
      "loss": 0.5455,
      "step": 139220
    },
    {
      "epoch": 56.99,
      "grad_norm": 1.8909789323806763,
      "learning_rate": 4.01490963124188e-06,
      "loss": 0.5238,
      "step": 139230
    },
    {
      "epoch": 57.0,
      "grad_norm": 1.8302143812179565,
      "learning_rate": 4.014798406614438e-06,
      "loss": 0.5424,
      "step": 139240
    },
    {
      "epoch": 57.0,
      "grad_norm": 2.164508819580078,
      "learning_rate": 4.014687174332182e-06,
      "loss": 0.5182,
      "step": 139250
    },
    {
      "epoch": 57.0,
      "eval_loss": 0.528732180595398,
      "eval_runtime": 52.1244,
      "eval_samples_per_second": 66.169,
      "eval_steps_per_second": 8.288,
      "step": 139251
    },
    {
      "epoch": 57.0,
      "grad_norm": 2.059551477432251,
      "learning_rate": 4.014575934395622e-06,
      "loss": 0.5161,
      "step": 139260
    },
    {
      "epoch": 57.01,
      "grad_norm": 2.5525577068328857,
      "learning_rate": 4.014464686805267e-06,
      "loss": 0.5323,
      "step": 139270
    },
    {
      "epoch": 57.01,
      "grad_norm": 2.182507038116455,
      "learning_rate": 4.014353431561627e-06,
      "loss": 0.5188,
      "step": 139280
    },
    {
      "epoch": 57.02,
      "grad_norm": 2.7313623428344727,
      "learning_rate": 4.0142421686652115e-06,
      "loss": 0.537,
      "step": 139290
    },
    {
      "epoch": 57.02,
      "grad_norm": 1.7792997360229492,
      "learning_rate": 4.014130898116529e-06,
      "loss": 0.5088,
      "step": 139300
    },
    {
      "epoch": 57.02,
      "grad_norm": 1.3785182237625122,
      "learning_rate": 4.014019619916092e-06,
      "loss": 0.527,
      "step": 139310
    },
    {
      "epoch": 57.03,
      "grad_norm": 2.3238065242767334,
      "learning_rate": 4.013908334064407e-06,
      "loss": 0.5242,
      "step": 139320
    },
    {
      "epoch": 57.03,
      "grad_norm": 2.3989288806915283,
      "learning_rate": 4.013797040561986e-06,
      "loss": 0.5477,
      "step": 139330
    },
    {
      "epoch": 57.04,
      "grad_norm": 1.6099677085876465,
      "learning_rate": 4.013685739409337e-06,
      "loss": 0.536,
      "step": 139340
    },
    {
      "epoch": 57.04,
      "grad_norm": 1.874420166015625,
      "learning_rate": 4.0135744306069726e-06,
      "loss": 0.5406,
      "step": 139350
    },
    {
      "epoch": 57.04,
      "grad_norm": 2.065828800201416,
      "learning_rate": 4.0134631141554005e-06,
      "loss": 0.5365,
      "step": 139360
    },
    {
      "epoch": 57.05,
      "grad_norm": 1.7260509729385376,
      "learning_rate": 4.013351790055131e-06,
      "loss": 0.5211,
      "step": 139370
    },
    {
      "epoch": 57.05,
      "grad_norm": 2.080810070037842,
      "learning_rate": 4.013240458306675e-06,
      "loss": 0.5463,
      "step": 139380
    },
    {
      "epoch": 57.06,
      "grad_norm": 1.6470102071762085,
      "learning_rate": 4.0131291189105415e-06,
      "loss": 0.5237,
      "step": 139390
    },
    {
      "epoch": 57.06,
      "grad_norm": 2.2240195274353027,
      "learning_rate": 4.01301777186724e-06,
      "loss": 0.5169,
      "step": 139400
    },
    {
      "epoch": 57.07,
      "grad_norm": 2.0057342052459717,
      "learning_rate": 4.012906417177283e-06,
      "loss": 0.5262,
      "step": 139410
    },
    {
      "epoch": 57.07,
      "grad_norm": 2.05601167678833,
      "learning_rate": 4.012795054841177e-06,
      "loss": 0.5355,
      "step": 139420
    },
    {
      "epoch": 57.07,
      "grad_norm": 1.9616584777832031,
      "learning_rate": 4.0126836848594354e-06,
      "loss": 0.5335,
      "step": 139430
    },
    {
      "epoch": 57.08,
      "grad_norm": 1.6956528425216675,
      "learning_rate": 4.012572307232567e-06,
      "loss": 0.5442,
      "step": 139440
    },
    {
      "epoch": 57.08,
      "grad_norm": 2.547447443008423,
      "learning_rate": 4.012460921961082e-06,
      "loss": 0.5468,
      "step": 139450
    },
    {
      "epoch": 57.09,
      "grad_norm": 2.1206259727478027,
      "learning_rate": 4.01234952904549e-06,
      "loss": 0.5429,
      "step": 139460
    },
    {
      "epoch": 57.09,
      "grad_norm": 2.084868907928467,
      "learning_rate": 4.0122381284863025e-06,
      "loss": 0.5514,
      "step": 139470
    },
    {
      "epoch": 57.09,
      "grad_norm": 2.5018322467803955,
      "learning_rate": 4.012126720284029e-06,
      "loss": 0.535,
      "step": 139480
    },
    {
      "epoch": 57.1,
      "grad_norm": 1.8952959775924683,
      "learning_rate": 4.01201530443918e-06,
      "loss": 0.5156,
      "step": 139490
    },
    {
      "epoch": 57.1,
      "grad_norm": 1.7204926013946533,
      "learning_rate": 4.011903880952266e-06,
      "loss": 0.5182,
      "step": 139500
    },
    {
      "epoch": 57.11,
      "grad_norm": 1.9744511842727661,
      "learning_rate": 4.011792449823798e-06,
      "loss": 0.5324,
      "step": 139510
    },
    {
      "epoch": 57.11,
      "grad_norm": 1.8842500448226929,
      "learning_rate": 4.011681011054286e-06,
      "loss": 0.5429,
      "step": 139520
    },
    {
      "epoch": 57.11,
      "grad_norm": 1.5645216703414917,
      "learning_rate": 4.01156956464424e-06,
      "loss": 0.5191,
      "step": 139530
    },
    {
      "epoch": 57.12,
      "grad_norm": 2.2070798873901367,
      "learning_rate": 4.01145811059417e-06,
      "loss": 0.5182,
      "step": 139540
    },
    {
      "epoch": 57.12,
      "grad_norm": 1.6141564846038818,
      "learning_rate": 4.011346648904589e-06,
      "loss": 0.546,
      "step": 139550
    },
    {
      "epoch": 57.13,
      "grad_norm": 2.2421157360076904,
      "learning_rate": 4.011235179576005e-06,
      "loss": 0.5328,
      "step": 139560
    },
    {
      "epoch": 57.13,
      "grad_norm": 2.1970698833465576,
      "learning_rate": 4.01112370260893e-06,
      "loss": 0.5257,
      "step": 139570
    },
    {
      "epoch": 57.13,
      "grad_norm": 2.2865793704986572,
      "learning_rate": 4.011012218003874e-06,
      "loss": 0.5356,
      "step": 139580
    },
    {
      "epoch": 57.14,
      "grad_norm": 1.96378493309021,
      "learning_rate": 4.010900725761348e-06,
      "loss": 0.5228,
      "step": 139590
    },
    {
      "epoch": 57.14,
      "grad_norm": 2.1455602645874023,
      "learning_rate": 4.010789225881864e-06,
      "loss": 0.5452,
      "step": 139600
    },
    {
      "epoch": 57.15,
      "grad_norm": 1.7620649337768555,
      "learning_rate": 4.0106777183659295e-06,
      "loss": 0.5179,
      "step": 139610
    },
    {
      "epoch": 57.15,
      "grad_norm": 1.8574024438858032,
      "learning_rate": 4.0105662032140585e-06,
      "loss": 0.5195,
      "step": 139620
    },
    {
      "epoch": 57.16,
      "grad_norm": 2.0847580432891846,
      "learning_rate": 4.01045468042676e-06,
      "loss": 0.5484,
      "step": 139630
    },
    {
      "epoch": 57.16,
      "grad_norm": 2.3749465942382812,
      "learning_rate": 4.010343150004546e-06,
      "loss": 0.5258,
      "step": 139640
    },
    {
      "epoch": 57.16,
      "grad_norm": 2.2687759399414062,
      "learning_rate": 4.010231611947926e-06,
      "loss": 0.5228,
      "step": 139650
    },
    {
      "epoch": 57.17,
      "grad_norm": 2.130129337310791,
      "learning_rate": 4.010120066257413e-06,
      "loss": 0.5325,
      "step": 139660
    },
    {
      "epoch": 57.17,
      "grad_norm": 1.5875575542449951,
      "learning_rate": 4.010008512933516e-06,
      "loss": 0.539,
      "step": 139670
    },
    {
      "epoch": 57.18,
      "grad_norm": 2.0581281185150146,
      "learning_rate": 4.009896951976747e-06,
      "loss": 0.5334,
      "step": 139680
    },
    {
      "epoch": 57.18,
      "grad_norm": 1.5521079301834106,
      "learning_rate": 4.009785383387617e-06,
      "loss": 0.5317,
      "step": 139690
    },
    {
      "epoch": 57.18,
      "grad_norm": 1.6732277870178223,
      "learning_rate": 4.009673807166636e-06,
      "loss": 0.5275,
      "step": 139700
    },
    {
      "epoch": 57.19,
      "grad_norm": 2.3154478073120117,
      "learning_rate": 4.0095622233143165e-06,
      "loss": 0.5431,
      "step": 139710
    },
    {
      "epoch": 57.19,
      "grad_norm": 2.0956571102142334,
      "learning_rate": 4.00945063183117e-06,
      "loss": 0.5426,
      "step": 139720
    },
    {
      "epoch": 57.2,
      "grad_norm": 1.9805794954299927,
      "learning_rate": 4.009339032717707e-06,
      "loss": 0.5359,
      "step": 139730
    },
    {
      "epoch": 57.2,
      "grad_norm": 2.009016275405884,
      "learning_rate": 4.009227425974438e-06,
      "loss": 0.5176,
      "step": 139740
    },
    {
      "epoch": 57.2,
      "grad_norm": 1.9500540494918823,
      "learning_rate": 4.009115811601874e-06,
      "loss": 0.5382,
      "step": 139750
    },
    {
      "epoch": 57.21,
      "grad_norm": 2.2700676918029785,
      "learning_rate": 4.0090041896005285e-06,
      "loss": 0.5085,
      "step": 139760
    },
    {
      "epoch": 57.21,
      "grad_norm": 2.0888493061065674,
      "learning_rate": 4.008892559970911e-06,
      "loss": 0.5257,
      "step": 139770
    },
    {
      "epoch": 57.22,
      "grad_norm": 1.9359713792800903,
      "learning_rate": 4.008780922713533e-06,
      "loss": 0.5336,
      "step": 139780
    },
    {
      "epoch": 57.22,
      "grad_norm": 1.6704044342041016,
      "learning_rate": 4.008669277828906e-06,
      "loss": 0.5353,
      "step": 139790
    },
    {
      "epoch": 57.22,
      "grad_norm": 2.445241928100586,
      "learning_rate": 4.0085576253175424e-06,
      "loss": 0.5311,
      "step": 139800
    },
    {
      "epoch": 57.23,
      "grad_norm": 2.2235536575317383,
      "learning_rate": 4.008445965179953e-06,
      "loss": 0.5134,
      "step": 139810
    },
    {
      "epoch": 57.23,
      "grad_norm": 2.2498226165771484,
      "learning_rate": 4.008334297416648e-06,
      "loss": 0.5326,
      "step": 139820
    },
    {
      "epoch": 57.24,
      "grad_norm": 2.057508707046509,
      "learning_rate": 4.008222622028141e-06,
      "loss": 0.5199,
      "step": 139830
    },
    {
      "epoch": 57.24,
      "grad_norm": 2.85711669921875,
      "learning_rate": 4.008110939014943e-06,
      "loss": 0.5278,
      "step": 139840
    },
    {
      "epoch": 57.25,
      "grad_norm": 2.6315088272094727,
      "learning_rate": 4.007999248377564e-06,
      "loss": 0.536,
      "step": 139850
    },
    {
      "epoch": 57.25,
      "grad_norm": 2.168414354324341,
      "learning_rate": 4.007887550116519e-06,
      "loss": 0.5311,
      "step": 139860
    },
    {
      "epoch": 57.25,
      "grad_norm": 1.4965802431106567,
      "learning_rate": 4.007775844232316e-06,
      "loss": 0.5487,
      "step": 139870
    },
    {
      "epoch": 57.26,
      "grad_norm": 1.6916923522949219,
      "learning_rate": 4.007664130725468e-06,
      "loss": 0.5229,
      "step": 139880
    },
    {
      "epoch": 57.26,
      "grad_norm": 1.6080209016799927,
      "learning_rate": 4.007552409596488e-06,
      "loss": 0.5327,
      "step": 139890
    },
    {
      "epoch": 57.27,
      "grad_norm": 2.2257776260375977,
      "learning_rate": 4.007440680845887e-06,
      "loss": 0.5371,
      "step": 139900
    },
    {
      "epoch": 57.27,
      "grad_norm": 2.0702099800109863,
      "learning_rate": 4.007328944474176e-06,
      "loss": 0.5209,
      "step": 139910
    },
    {
      "epoch": 57.27,
      "grad_norm": 1.8556020259857178,
      "learning_rate": 4.007217200481869e-06,
      "loss": 0.5353,
      "step": 139920
    },
    {
      "epoch": 57.28,
      "grad_norm": 2.0496020317077637,
      "learning_rate": 4.007105448869475e-06,
      "loss": 0.509,
      "step": 139930
    },
    {
      "epoch": 57.28,
      "grad_norm": 2.3341434001922607,
      "learning_rate": 4.006993689637507e-06,
      "loss": 0.5245,
      "step": 139940
    },
    {
      "epoch": 57.29,
      "grad_norm": 1.7711873054504395,
      "learning_rate": 4.006881922786478e-06,
      "loss": 0.5288,
      "step": 139950
    },
    {
      "epoch": 57.29,
      "grad_norm": 2.8653724193573,
      "learning_rate": 4.006770148316899e-06,
      "loss": 0.5069,
      "step": 139960
    },
    {
      "epoch": 57.29,
      "grad_norm": 1.7651770114898682,
      "learning_rate": 4.006658366229282e-06,
      "loss": 0.5208,
      "step": 139970
    },
    {
      "epoch": 57.3,
      "grad_norm": 2.9605581760406494,
      "learning_rate": 4.006546576524139e-06,
      "loss": 0.5183,
      "step": 139980
    },
    {
      "epoch": 57.3,
      "grad_norm": 2.3421220779418945,
      "learning_rate": 4.006434779201983e-06,
      "loss": 0.536,
      "step": 139990
    },
    {
      "epoch": 57.31,
      "grad_norm": 1.6516526937484741,
      "learning_rate": 4.006322974263325e-06,
      "loss": 0.524,
      "step": 140000
    },
    {
      "epoch": 57.31,
      "grad_norm": 2.346087694168091,
      "learning_rate": 4.006211161708679e-06,
      "loss": 0.5201,
      "step": 140010
    },
    {
      "epoch": 57.31,
      "grad_norm": 1.458109974861145,
      "learning_rate": 4.006099341538554e-06,
      "loss": 0.5454,
      "step": 140020
    },
    {
      "epoch": 57.32,
      "grad_norm": 2.6836373805999756,
      "learning_rate": 4.005987513753466e-06,
      "loss": 0.5464,
      "step": 140030
    },
    {
      "epoch": 57.32,
      "grad_norm": 2.3230910301208496,
      "learning_rate": 4.005875678353924e-06,
      "loss": 0.5285,
      "step": 140040
    },
    {
      "epoch": 57.33,
      "grad_norm": 2.40586256980896,
      "learning_rate": 4.005763835340441e-06,
      "loss": 0.5267,
      "step": 140050
    },
    {
      "epoch": 57.33,
      "grad_norm": 2.160883665084839,
      "learning_rate": 4.005651984713532e-06,
      "loss": 0.531,
      "step": 140060
    },
    {
      "epoch": 57.34,
      "grad_norm": 2.1052632331848145,
      "learning_rate": 4.005540126473707e-06,
      "loss": 0.5341,
      "step": 140070
    },
    {
      "epoch": 57.34,
      "grad_norm": 1.8414039611816406,
      "learning_rate": 4.005428260621478e-06,
      "loss": 0.5362,
      "step": 140080
    },
    {
      "epoch": 57.34,
      "grad_norm": 1.5226635932922363,
      "learning_rate": 4.005316387157358e-06,
      "loss": 0.5207,
      "step": 140090
    },
    {
      "epoch": 57.35,
      "grad_norm": 2.186962127685547,
      "learning_rate": 4.005204506081861e-06,
      "loss": 0.5379,
      "step": 140100
    },
    {
      "epoch": 57.35,
      "grad_norm": 2.2764244079589844,
      "learning_rate": 4.005092617395497e-06,
      "loss": 0.5395,
      "step": 140110
    },
    {
      "epoch": 57.36,
      "grad_norm": 2.1869733333587646,
      "learning_rate": 4.004980721098781e-06,
      "loss": 0.5341,
      "step": 140120
    },
    {
      "epoch": 57.36,
      "grad_norm": 2.054652452468872,
      "learning_rate": 4.0048688171922225e-06,
      "loss": 0.5378,
      "step": 140130
    },
    {
      "epoch": 57.36,
      "grad_norm": 1.9802594184875488,
      "learning_rate": 4.004756905676337e-06,
      "loss": 0.5257,
      "step": 140140
    },
    {
      "epoch": 57.37,
      "grad_norm": 1.7477993965148926,
      "learning_rate": 4.004644986551637e-06,
      "loss": 0.5272,
      "step": 140150
    },
    {
      "epoch": 57.37,
      "grad_norm": 2.0979163646698,
      "learning_rate": 4.004533059818634e-06,
      "loss": 0.5298,
      "step": 140160
    },
    {
      "epoch": 57.38,
      "grad_norm": 1.8266257047653198,
      "learning_rate": 4.004421125477839e-06,
      "loss": 0.5273,
      "step": 140170
    },
    {
      "epoch": 57.38,
      "grad_norm": 1.98978590965271,
      "learning_rate": 4.004309183529769e-06,
      "loss": 0.5253,
      "step": 140180
    },
    {
      "epoch": 57.38,
      "grad_norm": 1.840692162513733,
      "learning_rate": 4.004197233974933e-06,
      "loss": 0.5346,
      "step": 140190
    },
    {
      "epoch": 57.39,
      "grad_norm": 1.7052537202835083,
      "learning_rate": 4.004085276813846e-06,
      "loss": 0.5341,
      "step": 140200
    },
    {
      "epoch": 57.39,
      "grad_norm": 2.134944200515747,
      "learning_rate": 4.003973312047021e-06,
      "loss": 0.5498,
      "step": 140210
    },
    {
      "epoch": 57.4,
      "grad_norm": 1.552662968635559,
      "learning_rate": 4.00386133967497e-06,
      "loss": 0.524,
      "step": 140220
    },
    {
      "epoch": 57.4,
      "grad_norm": 1.9136937856674194,
      "learning_rate": 4.003749359698205e-06,
      "loss": 0.5251,
      "step": 140230
    },
    {
      "epoch": 57.4,
      "grad_norm": 1.8588043451309204,
      "learning_rate": 4.00363737211724e-06,
      "loss": 0.5145,
      "step": 140240
    },
    {
      "epoch": 57.41,
      "grad_norm": 1.9319782257080078,
      "learning_rate": 4.003525376932589e-06,
      "loss": 0.5331,
      "step": 140250
    },
    {
      "epoch": 57.41,
      "grad_norm": 2.075047254562378,
      "learning_rate": 4.003413374144764e-06,
      "loss": 0.5423,
      "step": 140260
    },
    {
      "epoch": 57.42,
      "grad_norm": 1.7735567092895508,
      "learning_rate": 4.003301363754277e-06,
      "loss": 0.5078,
      "step": 140270
    },
    {
      "epoch": 57.42,
      "grad_norm": 2.2624528408050537,
      "learning_rate": 4.003189345761643e-06,
      "loss": 0.5347,
      "step": 140280
    },
    {
      "epoch": 57.43,
      "grad_norm": 1.6158965826034546,
      "learning_rate": 4.003077320167375e-06,
      "loss": 0.5491,
      "step": 140290
    },
    {
      "epoch": 57.43,
      "grad_norm": 1.7360293865203857,
      "learning_rate": 4.002965286971985e-06,
      "loss": 0.5051,
      "step": 140300
    },
    {
      "epoch": 57.43,
      "grad_norm": 2.4476802349090576,
      "learning_rate": 4.002853246175986e-06,
      "loss": 0.5403,
      "step": 140310
    },
    {
      "epoch": 57.44,
      "grad_norm": 1.797548532485962,
      "learning_rate": 4.002741197779893e-06,
      "loss": 0.5456,
      "step": 140320
    },
    {
      "epoch": 57.44,
      "grad_norm": 1.7902637720108032,
      "learning_rate": 4.002629141784218e-06,
      "loss": 0.5422,
      "step": 140330
    },
    {
      "epoch": 57.45,
      "grad_norm": 2.7983005046844482,
      "learning_rate": 4.002517078189473e-06,
      "loss": 0.5221,
      "step": 140340
    },
    {
      "epoch": 57.45,
      "grad_norm": 2.173356771469116,
      "learning_rate": 4.002405006996175e-06,
      "loss": 0.5209,
      "step": 140350
    },
    {
      "epoch": 57.45,
      "grad_norm": 2.181858539581299,
      "learning_rate": 4.002292928204834e-06,
      "loss": 0.5359,
      "step": 140360
    },
    {
      "epoch": 57.46,
      "grad_norm": 2.0950241088867188,
      "learning_rate": 4.002180841815966e-06,
      "loss": 0.5349,
      "step": 140370
    },
    {
      "epoch": 57.46,
      "grad_norm": 1.967257022857666,
      "learning_rate": 4.002068747830082e-06,
      "loss": 0.5375,
      "step": 140380
    },
    {
      "epoch": 57.47,
      "grad_norm": 2.1239309310913086,
      "learning_rate": 4.0019566462476975e-06,
      "loss": 0.5519,
      "step": 140390
    },
    {
      "epoch": 57.47,
      "grad_norm": 2.25631046295166,
      "learning_rate": 4.001844537069324e-06,
      "loss": 0.5343,
      "step": 140400
    },
    {
      "epoch": 57.47,
      "grad_norm": 1.7726320028305054,
      "learning_rate": 4.001732420295477e-06,
      "loss": 0.5414,
      "step": 140410
    },
    {
      "epoch": 57.48,
      "grad_norm": 2.053417682647705,
      "learning_rate": 4.00162029592667e-06,
      "loss": 0.5065,
      "step": 140420
    },
    {
      "epoch": 57.48,
      "grad_norm": 1.6599162817001343,
      "learning_rate": 4.001508163963414e-06,
      "loss": 0.5232,
      "step": 140430
    },
    {
      "epoch": 57.49,
      "grad_norm": 2.0525453090667725,
      "learning_rate": 4.001396024406226e-06,
      "loss": 0.5389,
      "step": 140440
    },
    {
      "epoch": 57.49,
      "grad_norm": 1.4186937808990479,
      "learning_rate": 4.0012838772556184e-06,
      "loss": 0.5528,
      "step": 140450
    },
    {
      "epoch": 57.49,
      "grad_norm": 2.2218689918518066,
      "learning_rate": 4.001171722512105e-06,
      "loss": 0.5223,
      "step": 140460
    },
    {
      "epoch": 57.5,
      "grad_norm": 2.6569063663482666,
      "learning_rate": 4.001059560176198e-06,
      "loss": 0.5098,
      "step": 140470
    },
    {
      "epoch": 57.5,
      "grad_norm": 1.690772294998169,
      "learning_rate": 4.000947390248413e-06,
      "loss": 0.5387,
      "step": 140480
    },
    {
      "epoch": 57.51,
      "grad_norm": 2.0765020847320557,
      "learning_rate": 4.000835212729264e-06,
      "loss": 0.5235,
      "step": 140490
    },
    {
      "epoch": 57.51,
      "grad_norm": 2.6174156665802,
      "learning_rate": 4.0007230276192635e-06,
      "loss": 0.5387,
      "step": 140500
    },
    {
      "epoch": 57.52,
      "grad_norm": 1.990055799484253,
      "learning_rate": 4.000610834918928e-06,
      "loss": 0.5291,
      "step": 140510
    },
    {
      "epoch": 57.52,
      "grad_norm": 1.773476243019104,
      "learning_rate": 4.000498634628768e-06,
      "loss": 0.5257,
      "step": 140520
    },
    {
      "epoch": 57.52,
      "grad_norm": 2.0059545040130615,
      "learning_rate": 4.000386426749299e-06,
      "loss": 0.5385,
      "step": 140530
    },
    {
      "epoch": 57.53,
      "grad_norm": 2.1800808906555176,
      "learning_rate": 4.0002742112810356e-06,
      "loss": 0.5342,
      "step": 140540
    },
    {
      "epoch": 57.53,
      "grad_norm": 1.5630053281784058,
      "learning_rate": 4.000161988224492e-06,
      "loss": 0.5293,
      "step": 140550
    },
    {
      "epoch": 57.54,
      "grad_norm": 1.6731058359146118,
      "learning_rate": 4.00004975758018e-06,
      "loss": 0.5196,
      "step": 140560
    },
    {
      "epoch": 57.54,
      "grad_norm": 1.9129605293273926,
      "learning_rate": 3.999937519348617e-06,
      "loss": 0.5258,
      "step": 140570
    },
    {
      "epoch": 57.54,
      "grad_norm": 1.5459049940109253,
      "learning_rate": 3.999825273530314e-06,
      "loss": 0.5356,
      "step": 140580
    },
    {
      "epoch": 57.55,
      "grad_norm": 1.8946959972381592,
      "learning_rate": 3.999713020125787e-06,
      "loss": 0.5288,
      "step": 140590
    },
    {
      "epoch": 57.55,
      "grad_norm": 1.861288070678711,
      "learning_rate": 3.99960075913555e-06,
      "loss": 0.5373,
      "step": 140600
    },
    {
      "epoch": 57.56,
      "grad_norm": 2.108436107635498,
      "learning_rate": 3.999488490560118e-06,
      "loss": 0.5522,
      "step": 140610
    },
    {
      "epoch": 57.56,
      "grad_norm": 2.445754051208496,
      "learning_rate": 3.9993762144000025e-06,
      "loss": 0.5389,
      "step": 140620
    },
    {
      "epoch": 57.56,
      "grad_norm": 1.507636308670044,
      "learning_rate": 3.9992639306557215e-06,
      "loss": 0.5451,
      "step": 140630
    },
    {
      "epoch": 57.57,
      "grad_norm": 2.0543859004974365,
      "learning_rate": 3.9991516393277875e-06,
      "loss": 0.5204,
      "step": 140640
    },
    {
      "epoch": 57.57,
      "grad_norm": 1.7064220905303955,
      "learning_rate": 3.999039340416714e-06,
      "loss": 0.5346,
      "step": 140650
    },
    {
      "epoch": 57.58,
      "grad_norm": 2.154602527618408,
      "learning_rate": 3.998927033923016e-06,
      "loss": 0.5318,
      "step": 140660
    },
    {
      "epoch": 57.58,
      "grad_norm": 2.0518128871917725,
      "learning_rate": 3.99881471984721e-06,
      "loss": 0.5164,
      "step": 140670
    },
    {
      "epoch": 57.58,
      "grad_norm": 2.6124460697174072,
      "learning_rate": 3.998702398189808e-06,
      "loss": 0.5462,
      "step": 140680
    },
    {
      "epoch": 57.59,
      "grad_norm": 1.9067440032958984,
      "learning_rate": 3.998590068951326e-06,
      "loss": 0.5471,
      "step": 140690
    },
    {
      "epoch": 57.59,
      "grad_norm": 2.123028039932251,
      "learning_rate": 3.998477732132276e-06,
      "loss": 0.5455,
      "step": 140700
    },
    {
      "epoch": 57.6,
      "grad_norm": 1.5540186166763306,
      "learning_rate": 3.998365387733176e-06,
      "loss": 0.524,
      "step": 140710
    },
    {
      "epoch": 57.6,
      "grad_norm": 1.954728603363037,
      "learning_rate": 3.998253035754539e-06,
      "loss": 0.5258,
      "step": 140720
    },
    {
      "epoch": 57.61,
      "grad_norm": 2.7220940589904785,
      "learning_rate": 3.99814067619688e-06,
      "loss": 0.5137,
      "step": 140730
    },
    {
      "epoch": 57.61,
      "grad_norm": 1.8340129852294922,
      "learning_rate": 3.998028309060713e-06,
      "loss": 0.5364,
      "step": 140740
    },
    {
      "epoch": 57.61,
      "grad_norm": 1.9323256015777588,
      "learning_rate": 3.997915934346554e-06,
      "loss": 0.5337,
      "step": 140750
    },
    {
      "epoch": 57.62,
      "grad_norm": 2.4691920280456543,
      "learning_rate": 3.997803552054916e-06,
      "loss": 0.535,
      "step": 140760
    },
    {
      "epoch": 57.62,
      "grad_norm": 2.6756343841552734,
      "learning_rate": 3.997691162186316e-06,
      "loss": 0.5312,
      "step": 140770
    },
    {
      "epoch": 57.63,
      "grad_norm": 2.480633020401001,
      "learning_rate": 3.997578764741267e-06,
      "loss": 0.5341,
      "step": 140780
    },
    {
      "epoch": 57.63,
      "grad_norm": 2.110583543777466,
      "learning_rate": 3.9974663597202845e-06,
      "loss": 0.5415,
      "step": 140790
    },
    {
      "epoch": 57.63,
      "grad_norm": 1.7226399183273315,
      "learning_rate": 3.997353947123884e-06,
      "loss": 0.5062,
      "step": 140800
    },
    {
      "epoch": 57.64,
      "grad_norm": 1.97298264503479,
      "learning_rate": 3.997241526952579e-06,
      "loss": 0.5411,
      "step": 140810
    },
    {
      "epoch": 57.64,
      "grad_norm": 2.2905290126800537,
      "learning_rate": 3.997129099206886e-06,
      "loss": 0.5297,
      "step": 140820
    },
    {
      "epoch": 57.65,
      "grad_norm": 1.6359604597091675,
      "learning_rate": 3.9970166638873195e-06,
      "loss": 0.5421,
      "step": 140830
    },
    {
      "epoch": 57.65,
      "grad_norm": 1.5784211158752441,
      "learning_rate": 3.996904220994395e-06,
      "loss": 0.5372,
      "step": 140840
    },
    {
      "epoch": 57.65,
      "grad_norm": 1.9396051168441772,
      "learning_rate": 3.996791770528626e-06,
      "loss": 0.5203,
      "step": 140850
    },
    {
      "epoch": 57.66,
      "grad_norm": 1.6730278730392456,
      "learning_rate": 3.996679312490528e-06,
      "loss": 0.538,
      "step": 140860
    },
    {
      "epoch": 57.66,
      "grad_norm": 1.9604527950286865,
      "learning_rate": 3.996566846880618e-06,
      "loss": 0.5285,
      "step": 140870
    },
    {
      "epoch": 57.67,
      "grad_norm": 2.719438076019287,
      "learning_rate": 3.9964543736994105e-06,
      "loss": 0.5453,
      "step": 140880
    },
    {
      "epoch": 57.67,
      "grad_norm": 2.449883222579956,
      "learning_rate": 3.996341892947419e-06,
      "loss": 0.5503,
      "step": 140890
    },
    {
      "epoch": 57.67,
      "grad_norm": 1.7790554761886597,
      "learning_rate": 3.996229404625161e-06,
      "loss": 0.5286,
      "step": 140900
    },
    {
      "epoch": 57.68,
      "grad_norm": 2.2739791870117188,
      "learning_rate": 3.99611690873315e-06,
      "loss": 0.5598,
      "step": 140910
    },
    {
      "epoch": 57.68,
      "grad_norm": 2.067718982696533,
      "learning_rate": 3.996004405271903e-06,
      "loss": 0.5535,
      "step": 140920
    },
    {
      "epoch": 57.69,
      "grad_norm": 1.6295499801635742,
      "learning_rate": 3.995891894241933e-06,
      "loss": 0.5528,
      "step": 140930
    },
    {
      "epoch": 57.69,
      "grad_norm": 2.1578774452209473,
      "learning_rate": 3.995779375643758e-06,
      "loss": 0.5203,
      "step": 140940
    },
    {
      "epoch": 57.7,
      "grad_norm": 1.6887149810791016,
      "learning_rate": 3.995666849477891e-06,
      "loss": 0.5295,
      "step": 140950
    },
    {
      "epoch": 57.7,
      "grad_norm": 1.8655633926391602,
      "learning_rate": 3.99555431574485e-06,
      "loss": 0.5098,
      "step": 140960
    },
    {
      "epoch": 57.7,
      "grad_norm": 1.551093339920044,
      "learning_rate": 3.99544177444515e-06,
      "loss": 0.5267,
      "step": 140970
    },
    {
      "epoch": 57.71,
      "grad_norm": 1.4325119256973267,
      "learning_rate": 3.995329225579304e-06,
      "loss": 0.5092,
      "step": 140980
    },
    {
      "epoch": 57.71,
      "grad_norm": 2.43379807472229,
      "learning_rate": 3.99521666914783e-06,
      "loss": 0.5145,
      "step": 140990
    },
    {
      "epoch": 57.72,
      "grad_norm": 1.9105377197265625,
      "learning_rate": 3.995104105151243e-06,
      "loss": 0.5162,
      "step": 141000
    },
    {
      "epoch": 57.72,
      "grad_norm": 2.0142500400543213,
      "learning_rate": 3.994991533590059e-06,
      "loss": 0.5171,
      "step": 141010
    },
    {
      "epoch": 57.72,
      "grad_norm": 2.0611305236816406,
      "learning_rate": 3.9948789544647926e-06,
      "loss": 0.5384,
      "step": 141020
    },
    {
      "epoch": 57.73,
      "grad_norm": 2.5198240280151367,
      "learning_rate": 3.99476636777596e-06,
      "loss": 0.525,
      "step": 141030
    },
    {
      "epoch": 57.73,
      "grad_norm": 1.8229949474334717,
      "learning_rate": 3.994653773524078e-06,
      "loss": 0.5625,
      "step": 141040
    },
    {
      "epoch": 57.74,
      "grad_norm": 2.3529322147369385,
      "learning_rate": 3.99454117170966e-06,
      "loss": 0.5236,
      "step": 141050
    },
    {
      "epoch": 57.74,
      "grad_norm": 1.846038579940796,
      "learning_rate": 3.994428562333225e-06,
      "loss": 0.5437,
      "step": 141060
    },
    {
      "epoch": 57.74,
      "grad_norm": 1.9682024717330933,
      "learning_rate": 3.994315945395286e-06,
      "loss": 0.549,
      "step": 141070
    },
    {
      "epoch": 57.75,
      "grad_norm": 2.0604381561279297,
      "learning_rate": 3.994203320896361e-06,
      "loss": 0.5345,
      "step": 141080
    },
    {
      "epoch": 57.75,
      "grad_norm": 1.3625407218933105,
      "learning_rate": 3.994090688836965e-06,
      "loss": 0.5319,
      "step": 141090
    },
    {
      "epoch": 57.76,
      "grad_norm": 2.106774091720581,
      "learning_rate": 3.993978049217613e-06,
      "loss": 0.5169,
      "step": 141100
    },
    {
      "epoch": 57.76,
      "grad_norm": 1.9696844816207886,
      "learning_rate": 3.993865402038822e-06,
      "loss": 0.527,
      "step": 141110
    },
    {
      "epoch": 57.77,
      "grad_norm": 3.1604154109954834,
      "learning_rate": 3.993752747301108e-06,
      "loss": 0.5308,
      "step": 141120
    },
    {
      "epoch": 57.77,
      "grad_norm": 2.031601667404175,
      "learning_rate": 3.993640085004987e-06,
      "loss": 0.5206,
      "step": 141130
    },
    {
      "epoch": 57.77,
      "grad_norm": 2.0636050701141357,
      "learning_rate": 3.993527415150975e-06,
      "loss": 0.5182,
      "step": 141140
    },
    {
      "epoch": 57.78,
      "grad_norm": 1.813675045967102,
      "learning_rate": 3.993414737739588e-06,
      "loss": 0.5256,
      "step": 141150
    },
    {
      "epoch": 57.78,
      "grad_norm": 1.6389819383621216,
      "learning_rate": 3.993302052771343e-06,
      "loss": 0.5119,
      "step": 141160
    },
    {
      "epoch": 57.79,
      "grad_norm": 1.9369585514068604,
      "learning_rate": 3.993189360246755e-06,
      "loss": 0.5374,
      "step": 141170
    },
    {
      "epoch": 57.79,
      "grad_norm": 2.232398271560669,
      "learning_rate": 3.993076660166341e-06,
      "loss": 0.5152,
      "step": 141180
    },
    {
      "epoch": 57.79,
      "grad_norm": 1.704695463180542,
      "learning_rate": 3.992963952530617e-06,
      "loss": 0.5085,
      "step": 141190
    },
    {
      "epoch": 57.8,
      "grad_norm": 1.6819056272506714,
      "learning_rate": 3.9928512373400985e-06,
      "loss": 0.5411,
      "step": 141200
    },
    {
      "epoch": 57.8,
      "grad_norm": 1.5779565572738647,
      "learning_rate": 3.992738514595304e-06,
      "loss": 0.5538,
      "step": 141210
    },
    {
      "epoch": 57.81,
      "grad_norm": 1.5049299001693726,
      "learning_rate": 3.992625784296747e-06,
      "loss": 0.5163,
      "step": 141220
    },
    {
      "epoch": 57.81,
      "grad_norm": 2.4814116954803467,
      "learning_rate": 3.992513046444946e-06,
      "loss": 0.539,
      "step": 141230
    },
    {
      "epoch": 57.81,
      "grad_norm": 2.430180549621582,
      "learning_rate": 3.992400301040417e-06,
      "loss": 0.5325,
      "step": 141240
    },
    {
      "epoch": 57.82,
      "grad_norm": 1.3939590454101562,
      "learning_rate": 3.992287548083676e-06,
      "loss": 0.4874,
      "step": 141250
    },
    {
      "epoch": 57.82,
      "grad_norm": 2.2892274856567383,
      "learning_rate": 3.99217478757524e-06,
      "loss": 0.5371,
      "step": 141260
    },
    {
      "epoch": 57.83,
      "grad_norm": 1.7666866779327393,
      "learning_rate": 3.992062019515626e-06,
      "loss": 0.5246,
      "step": 141270
    },
    {
      "epoch": 57.83,
      "grad_norm": 2.2434964179992676,
      "learning_rate": 3.9919492439053485e-06,
      "loss": 0.5452,
      "step": 141280
    },
    {
      "epoch": 57.83,
      "grad_norm": 2.2682037353515625,
      "learning_rate": 3.991836460744927e-06,
      "loss": 0.5256,
      "step": 141290
    },
    {
      "epoch": 57.84,
      "grad_norm": 2.3360579013824463,
      "learning_rate": 3.991723670034876e-06,
      "loss": 0.5163,
      "step": 141300
    },
    {
      "epoch": 57.84,
      "grad_norm": 1.6736477613449097,
      "learning_rate": 3.991610871775712e-06,
      "loss": 0.5355,
      "step": 141310
    },
    {
      "epoch": 57.85,
      "grad_norm": 2.028616428375244,
      "learning_rate": 3.9914980659679536e-06,
      "loss": 0.5312,
      "step": 141320
    },
    {
      "epoch": 57.85,
      "grad_norm": 1.7062835693359375,
      "learning_rate": 3.991385252612117e-06,
      "loss": 0.5359,
      "step": 141330
    },
    {
      "epoch": 57.86,
      "grad_norm": 2.410841703414917,
      "learning_rate": 3.9912724317087166e-06,
      "loss": 0.5313,
      "step": 141340
    },
    {
      "epoch": 57.86,
      "grad_norm": 1.723294973373413,
      "learning_rate": 3.991159603258273e-06,
      "loss": 0.5403,
      "step": 141350
    },
    {
      "epoch": 57.86,
      "grad_norm": 2.7273635864257812,
      "learning_rate": 3.991046767261301e-06,
      "loss": 0.5263,
      "step": 141360
    },
    {
      "epoch": 57.87,
      "grad_norm": 2.235865592956543,
      "learning_rate": 3.990933923718316e-06,
      "loss": 0.5627,
      "step": 141370
    },
    {
      "epoch": 57.87,
      "grad_norm": 2.0619890689849854,
      "learning_rate": 3.990821072629838e-06,
      "loss": 0.5284,
      "step": 141380
    },
    {
      "epoch": 57.88,
      "grad_norm": 2.1445116996765137,
      "learning_rate": 3.990708213996382e-06,
      "loss": 0.5269,
      "step": 141390
    },
    {
      "epoch": 57.88,
      "grad_norm": 1.939335584640503,
      "learning_rate": 3.990595347818466e-06,
      "loss": 0.5208,
      "step": 141400
    },
    {
      "epoch": 57.88,
      "grad_norm": 1.488251805305481,
      "learning_rate": 3.990482474096606e-06,
      "loss": 0.5351,
      "step": 141410
    },
    {
      "epoch": 57.89,
      "grad_norm": 2.5282466411590576,
      "learning_rate": 3.9903695928313185e-06,
      "loss": 0.5188,
      "step": 141420
    },
    {
      "epoch": 57.89,
      "grad_norm": 2.3487048149108887,
      "learning_rate": 3.990256704023123e-06,
      "loss": 0.5351,
      "step": 141430
    },
    {
      "epoch": 57.9,
      "grad_norm": 1.7332481145858765,
      "learning_rate": 3.990143807672535e-06,
      "loss": 0.5283,
      "step": 141440
    },
    {
      "epoch": 57.9,
      "grad_norm": 2.7559139728546143,
      "learning_rate": 3.990030903780072e-06,
      "loss": 0.5338,
      "step": 141450
    },
    {
      "epoch": 57.9,
      "grad_norm": 1.5923222303390503,
      "learning_rate": 3.9899179923462516e-06,
      "loss": 0.5333,
      "step": 141460
    },
    {
      "epoch": 57.91,
      "grad_norm": 2.1425435543060303,
      "learning_rate": 3.9898050733715895e-06,
      "loss": 0.527,
      "step": 141470
    },
    {
      "epoch": 57.91,
      "grad_norm": 1.921281099319458,
      "learning_rate": 3.9896921468566055e-06,
      "loss": 0.5205,
      "step": 141480
    },
    {
      "epoch": 57.92,
      "grad_norm": 1.845396637916565,
      "learning_rate": 3.989579212801814e-06,
      "loss": 0.5302,
      "step": 141490
    },
    {
      "epoch": 57.92,
      "grad_norm": 1.9414451122283936,
      "learning_rate": 3.989466271207734e-06,
      "loss": 0.5365,
      "step": 141500
    },
    {
      "epoch": 57.92,
      "grad_norm": 2.022838830947876,
      "learning_rate": 3.989353322074883e-06,
      "loss": 0.5195,
      "step": 141510
    },
    {
      "epoch": 57.93,
      "grad_norm": 1.8776718378067017,
      "learning_rate": 3.989240365403778e-06,
      "loss": 0.5375,
      "step": 141520
    },
    {
      "epoch": 57.93,
      "grad_norm": 1.9963849782943726,
      "learning_rate": 3.989127401194936e-06,
      "loss": 0.5033,
      "step": 141530
    },
    {
      "epoch": 57.94,
      "grad_norm": 2.1538422107696533,
      "learning_rate": 3.989014429448876e-06,
      "loss": 0.5268,
      "step": 141540
    },
    {
      "epoch": 57.94,
      "grad_norm": 1.6979529857635498,
      "learning_rate": 3.988901450166114e-06,
      "loss": 0.5608,
      "step": 141550
    },
    {
      "epoch": 57.95,
      "grad_norm": 2.687537670135498,
      "learning_rate": 3.9887884633471675e-06,
      "loss": 0.5359,
      "step": 141560
    },
    {
      "epoch": 57.95,
      "grad_norm": 2.12424898147583,
      "learning_rate": 3.988675468992555e-06,
      "loss": 0.5252,
      "step": 141570
    },
    {
      "epoch": 57.95,
      "grad_norm": 1.9364567995071411,
      "learning_rate": 3.988562467102794e-06,
      "loss": 0.526,
      "step": 141580
    },
    {
      "epoch": 57.96,
      "grad_norm": 1.5652704238891602,
      "learning_rate": 3.988449457678401e-06,
      "loss": 0.5561,
      "step": 141590
    },
    {
      "epoch": 57.96,
      "grad_norm": 2.2228329181671143,
      "learning_rate": 3.988336440719895e-06,
      "loss": 0.5384,
      "step": 141600
    },
    {
      "epoch": 57.97,
      "grad_norm": 2.021738052368164,
      "learning_rate": 3.988223416227793e-06,
      "loss": 0.5391,
      "step": 141610
    },
    {
      "epoch": 57.97,
      "grad_norm": 1.8923991918563843,
      "learning_rate": 3.988110384202612e-06,
      "loss": 0.5237,
      "step": 141620
    },
    {
      "epoch": 57.97,
      "grad_norm": 2.072960615158081,
      "learning_rate": 3.987997344644873e-06,
      "loss": 0.5448,
      "step": 141630
    },
    {
      "epoch": 57.98,
      "grad_norm": 2.3950982093811035,
      "learning_rate": 3.98788429755509e-06,
      "loss": 0.5223,
      "step": 141640
    },
    {
      "epoch": 57.98,
      "grad_norm": 2.458256244659424,
      "learning_rate": 3.987771242933783e-06,
      "loss": 0.5183,
      "step": 141650
    },
    {
      "epoch": 57.99,
      "grad_norm": 1.9785735607147217,
      "learning_rate": 3.987658180781469e-06,
      "loss": 0.5303,
      "step": 141660
    },
    {
      "epoch": 57.99,
      "grad_norm": 2.269448757171631,
      "learning_rate": 3.987545111098666e-06,
      "loss": 0.5238,
      "step": 141670
    },
    {
      "epoch": 57.99,
      "grad_norm": 1.923596978187561,
      "learning_rate": 3.987432033885893e-06,
      "loss": 0.5253,
      "step": 141680
    },
    {
      "epoch": 58.0,
      "grad_norm": 2.353534698486328,
      "learning_rate": 3.987318949143667e-06,
      "loss": 0.5103,
      "step": 141690
    },
    {
      "epoch": 58.0,
      "eval_loss": 0.5302824974060059,
      "eval_runtime": 52.2093,
      "eval_samples_per_second": 66.061,
      "eval_steps_per_second": 8.274,
      "step": 141694
    },
    {
      "epoch": 58.0,
      "grad_norm": 1.6419942378997803,
      "learning_rate": 3.9872058568725055e-06,
      "loss": 0.5193,
      "step": 141700
    },
    {
      "epoch": 58.01,
      "grad_norm": 2.5997366905212402,
      "learning_rate": 3.987092757072927e-06,
      "loss": 0.5133,
      "step": 141710
    },
    {
      "epoch": 58.01,
      "grad_norm": 1.9856581687927246,
      "learning_rate": 3.98697964974545e-06,
      "loss": 0.5183,
      "step": 141720
    },
    {
      "epoch": 58.01,
      "grad_norm": 1.8273108005523682,
      "learning_rate": 3.986866534890593e-06,
      "loss": 0.5278,
      "step": 141730
    },
    {
      "epoch": 58.02,
      "grad_norm": 1.9301966428756714,
      "learning_rate": 3.986753412508873e-06,
      "loss": 0.5308,
      "step": 141740
    },
    {
      "epoch": 58.02,
      "grad_norm": 2.48218035697937,
      "learning_rate": 3.98664028260081e-06,
      "loss": 0.5194,
      "step": 141750
    },
    {
      "epoch": 58.03,
      "grad_norm": 2.127315044403076,
      "learning_rate": 3.98652714516692e-06,
      "loss": 0.5387,
      "step": 141760
    },
    {
      "epoch": 58.03,
      "grad_norm": 2.313267469406128,
      "learning_rate": 3.9864140002077234e-06,
      "loss": 0.5228,
      "step": 141770
    },
    {
      "epoch": 58.04,
      "grad_norm": 2.7507026195526123,
      "learning_rate": 3.986300847723737e-06,
      "loss": 0.539,
      "step": 141780
    },
    {
      "epoch": 58.04,
      "grad_norm": 2.1157374382019043,
      "learning_rate": 3.9861876877154785e-06,
      "loss": 0.5132,
      "step": 141790
    },
    {
      "epoch": 58.04,
      "grad_norm": 2.8906733989715576,
      "learning_rate": 3.986074520183468e-06,
      "loss": 0.5389,
      "step": 141800
    },
    {
      "epoch": 58.05,
      "grad_norm": 2.715057373046875,
      "learning_rate": 3.9859613451282225e-06,
      "loss": 0.5168,
      "step": 141810
    },
    {
      "epoch": 58.05,
      "grad_norm": 2.111326217651367,
      "learning_rate": 3.985848162550262e-06,
      "loss": 0.5141,
      "step": 141820
    },
    {
      "epoch": 58.06,
      "grad_norm": 2.285568952560425,
      "learning_rate": 3.985734972450105e-06,
      "loss": 0.5338,
      "step": 141830
    },
    {
      "epoch": 58.06,
      "grad_norm": 2.0122601985931396,
      "learning_rate": 3.985621774828268e-06,
      "loss": 0.5555,
      "step": 141840
    },
    {
      "epoch": 58.06,
      "grad_norm": 2.1924829483032227,
      "learning_rate": 3.98550856968527e-06,
      "loss": 0.545,
      "step": 141850
    },
    {
      "epoch": 58.07,
      "grad_norm": 1.8186103105545044,
      "learning_rate": 3.9853953570216315e-06,
      "loss": 0.5229,
      "step": 141860
    },
    {
      "epoch": 58.07,
      "grad_norm": 1.9636372327804565,
      "learning_rate": 3.98528213683787e-06,
      "loss": 0.5132,
      "step": 141870
    },
    {
      "epoch": 58.08,
      "grad_norm": 1.9819700717926025,
      "learning_rate": 3.985168909134503e-06,
      "loss": 0.5422,
      "step": 141880
    },
    {
      "epoch": 58.08,
      "grad_norm": 2.0043201446533203,
      "learning_rate": 3.98505567391205e-06,
      "loss": 0.5297,
      "step": 141890
    },
    {
      "epoch": 58.08,
      "grad_norm": 1.2110614776611328,
      "learning_rate": 3.984942431171031e-06,
      "loss": 0.5519,
      "step": 141900
    },
    {
      "epoch": 58.09,
      "grad_norm": 2.2224478721618652,
      "learning_rate": 3.984829180911963e-06,
      "loss": 0.5267,
      "step": 141910
    },
    {
      "epoch": 58.09,
      "grad_norm": 1.6277096271514893,
      "learning_rate": 3.984715923135366e-06,
      "loss": 0.5069,
      "step": 141920
    },
    {
      "epoch": 58.1,
      "grad_norm": 2.0473763942718506,
      "learning_rate": 3.984602657841757e-06,
      "loss": 0.5287,
      "step": 141930
    },
    {
      "epoch": 58.1,
      "grad_norm": 2.007382869720459,
      "learning_rate": 3.984489385031657e-06,
      "loss": 0.51,
      "step": 141940
    },
    {
      "epoch": 58.1,
      "grad_norm": 2.286846160888672,
      "learning_rate": 3.9843761047055844e-06,
      "loss": 0.5208,
      "step": 141950
    },
    {
      "epoch": 58.11,
      "grad_norm": 1.7104655504226685,
      "learning_rate": 3.9842628168640565e-06,
      "loss": 0.5322,
      "step": 141960
    },
    {
      "epoch": 58.11,
      "grad_norm": 2.2725205421447754,
      "learning_rate": 3.984149521507595e-06,
      "loss": 0.5511,
      "step": 141970
    },
    {
      "epoch": 58.12,
      "grad_norm": 1.8557558059692383,
      "learning_rate": 3.9840362186367166e-06,
      "loss": 0.5202,
      "step": 141980
    },
    {
      "epoch": 58.12,
      "grad_norm": 2.3695428371429443,
      "learning_rate": 3.98392290825194e-06,
      "loss": 0.5355,
      "step": 141990
    },
    {
      "epoch": 58.13,
      "grad_norm": 2.233051300048828,
      "learning_rate": 3.983809590353787e-06,
      "loss": 0.5204,
      "step": 142000
    },
    {
      "epoch": 58.13,
      "grad_norm": 2.9949300289154053,
      "learning_rate": 3.9836962649427744e-06,
      "loss": 0.5146,
      "step": 142010
    },
    {
      "epoch": 58.13,
      "grad_norm": 2.0086960792541504,
      "learning_rate": 3.983582932019421e-06,
      "loss": 0.5201,
      "step": 142020
    },
    {
      "epoch": 58.14,
      "grad_norm": 2.2076923847198486,
      "learning_rate": 3.983469591584248e-06,
      "loss": 0.5192,
      "step": 142030
    },
    {
      "epoch": 58.14,
      "grad_norm": 2.0816667079925537,
      "learning_rate": 3.983356243637773e-06,
      "loss": 0.5403,
      "step": 142040
    },
    {
      "epoch": 58.15,
      "grad_norm": 1.9858579635620117,
      "learning_rate": 3.983242888180516e-06,
      "loss": 0.5335,
      "step": 142050
    },
    {
      "epoch": 58.15,
      "grad_norm": 1.6930714845657349,
      "learning_rate": 3.983129525212996e-06,
      "loss": 0.523,
      "step": 142060
    },
    {
      "epoch": 58.15,
      "grad_norm": 2.053487777709961,
      "learning_rate": 3.9830161547357326e-06,
      "loss": 0.5321,
      "step": 142070
    },
    {
      "epoch": 58.16,
      "grad_norm": 1.515338659286499,
      "learning_rate": 3.982902776749244e-06,
      "loss": 0.5433,
      "step": 142080
    },
    {
      "epoch": 58.16,
      "grad_norm": 1.754008412361145,
      "learning_rate": 3.982789391254051e-06,
      "loss": 0.496,
      "step": 142090
    },
    {
      "epoch": 58.17,
      "grad_norm": 2.2168474197387695,
      "learning_rate": 3.982675998250672e-06,
      "loss": 0.5261,
      "step": 142100
    },
    {
      "epoch": 58.17,
      "grad_norm": 2.3951709270477295,
      "learning_rate": 3.982562597739628e-06,
      "loss": 0.548,
      "step": 142110
    },
    {
      "epoch": 58.17,
      "grad_norm": 2.026401996612549,
      "learning_rate": 3.9824491897214355e-06,
      "loss": 0.5173,
      "step": 142120
    },
    {
      "epoch": 58.18,
      "grad_norm": 2.5173611640930176,
      "learning_rate": 3.982335774196617e-06,
      "loss": 0.5129,
      "step": 142130
    },
    {
      "epoch": 58.18,
      "grad_norm": 2.0257601737976074,
      "learning_rate": 3.98222235116569e-06,
      "loss": 0.5242,
      "step": 142140
    },
    {
      "epoch": 58.19,
      "grad_norm": 2.3167331218719482,
      "learning_rate": 3.982108920629176e-06,
      "loss": 0.5245,
      "step": 142150
    },
    {
      "epoch": 58.19,
      "grad_norm": 2.2946834564208984,
      "learning_rate": 3.981995482587592e-06,
      "loss": 0.5208,
      "step": 142160
    },
    {
      "epoch": 58.19,
      "grad_norm": 2.125154972076416,
      "learning_rate": 3.9818820370414605e-06,
      "loss": 0.5472,
      "step": 142170
    },
    {
      "epoch": 58.2,
      "grad_norm": 1.9930895566940308,
      "learning_rate": 3.981768583991299e-06,
      "loss": 0.5226,
      "step": 142180
    },
    {
      "epoch": 58.2,
      "grad_norm": 1.5428298711776733,
      "learning_rate": 3.981655123437628e-06,
      "loss": 0.5295,
      "step": 142190
    },
    {
      "epoch": 58.21,
      "grad_norm": 2.2325081825256348,
      "learning_rate": 3.981541655380968e-06,
      "loss": 0.5242,
      "step": 142200
    },
    {
      "epoch": 58.21,
      "grad_norm": 2.141568899154663,
      "learning_rate": 3.981428179821838e-06,
      "loss": 0.5441,
      "step": 142210
    },
    {
      "epoch": 58.22,
      "grad_norm": 2.3033032417297363,
      "learning_rate": 3.981314696760758e-06,
      "loss": 0.5273,
      "step": 142220
    },
    {
      "epoch": 58.22,
      "grad_norm": 1.6594101190567017,
      "learning_rate": 3.981201206198247e-06,
      "loss": 0.5142,
      "step": 142230
    },
    {
      "epoch": 58.22,
      "grad_norm": 1.928079605102539,
      "learning_rate": 3.981087708134826e-06,
      "loss": 0.5376,
      "step": 142240
    },
    {
      "epoch": 58.23,
      "grad_norm": 2.754426956176758,
      "learning_rate": 3.9809742025710144e-06,
      "loss": 0.5392,
      "step": 142250
    },
    {
      "epoch": 58.23,
      "grad_norm": 1.9252488613128662,
      "learning_rate": 3.980860689507332e-06,
      "loss": 0.5458,
      "step": 142260
    },
    {
      "epoch": 58.24,
      "grad_norm": 1.7777600288391113,
      "learning_rate": 3.9807471689443e-06,
      "loss": 0.5274,
      "step": 142270
    },
    {
      "epoch": 58.24,
      "grad_norm": 1.7301316261291504,
      "learning_rate": 3.980633640882437e-06,
      "loss": 0.5222,
      "step": 142280
    },
    {
      "epoch": 58.24,
      "grad_norm": 1.6889069080352783,
      "learning_rate": 3.980520105322263e-06,
      "loss": 0.5173,
      "step": 142290
    },
    {
      "epoch": 58.25,
      "grad_norm": 2.365895986557007,
      "learning_rate": 3.9804065622643e-06,
      "loss": 0.5185,
      "step": 142300
    },
    {
      "epoch": 58.25,
      "grad_norm": 1.9779706001281738,
      "learning_rate": 3.980293011709066e-06,
      "loss": 0.5319,
      "step": 142310
    },
    {
      "epoch": 58.26,
      "grad_norm": 1.5098317861557007,
      "learning_rate": 3.980179453657081e-06,
      "loss": 0.538,
      "step": 142320
    },
    {
      "epoch": 58.26,
      "grad_norm": 1.7173717021942139,
      "learning_rate": 3.980065888108868e-06,
      "loss": 0.535,
      "step": 142330
    },
    {
      "epoch": 58.26,
      "grad_norm": 1.9380388259887695,
      "learning_rate": 3.979952315064944e-06,
      "loss": 0.5357,
      "step": 142340
    },
    {
      "epoch": 58.27,
      "grad_norm": 1.7614710330963135,
      "learning_rate": 3.97983873452583e-06,
      "loss": 0.5327,
      "step": 142350
    },
    {
      "epoch": 58.27,
      "grad_norm": 2.4783411026000977,
      "learning_rate": 3.979725146492048e-06,
      "loss": 0.5623,
      "step": 142360
    },
    {
      "epoch": 58.28,
      "grad_norm": 2.647291660308838,
      "learning_rate": 3.979611550964117e-06,
      "loss": 0.5373,
      "step": 142370
    },
    {
      "epoch": 58.28,
      "grad_norm": 2.551680564880371,
      "learning_rate": 3.979497947942558e-06,
      "loss": 0.5454,
      "step": 142380
    },
    {
      "epoch": 58.28,
      "grad_norm": 2.3971099853515625,
      "learning_rate": 3.9793843374278895e-06,
      "loss": 0.5377,
      "step": 142390
    },
    {
      "epoch": 58.29,
      "grad_norm": 1.9494589567184448,
      "learning_rate": 3.979270719420634e-06,
      "loss": 0.5286,
      "step": 142400
    },
    {
      "epoch": 58.29,
      "grad_norm": 1.9027605056762695,
      "learning_rate": 3.979157093921312e-06,
      "loss": 0.542,
      "step": 142410
    },
    {
      "epoch": 58.3,
      "grad_norm": 1.8969441652297974,
      "learning_rate": 3.9790434609304446e-06,
      "loss": 0.5086,
      "step": 142420
    },
    {
      "epoch": 58.3,
      "grad_norm": 1.949062466621399,
      "learning_rate": 3.978929820448549e-06,
      "loss": 0.5635,
      "step": 142430
    },
    {
      "epoch": 58.31,
      "grad_norm": 2.1252710819244385,
      "learning_rate": 3.978816172476148e-06,
      "loss": 0.5067,
      "step": 142440
    },
    {
      "epoch": 58.31,
      "grad_norm": 2.0065293312072754,
      "learning_rate": 3.978702517013763e-06,
      "loss": 0.5495,
      "step": 142450
    },
    {
      "epoch": 58.31,
      "grad_norm": 1.883963704109192,
      "learning_rate": 3.978588854061914e-06,
      "loss": 0.5311,
      "step": 142460
    },
    {
      "epoch": 58.32,
      "grad_norm": 2.305581569671631,
      "learning_rate": 3.97847518362112e-06,
      "loss": 0.5093,
      "step": 142470
    },
    {
      "epoch": 58.32,
      "grad_norm": 2.420950174331665,
      "learning_rate": 3.978361505691904e-06,
      "loss": 0.548,
      "step": 142480
    },
    {
      "epoch": 58.33,
      "grad_norm": 1.8072316646575928,
      "learning_rate": 3.978247820274786e-06,
      "loss": 0.5444,
      "step": 142490
    },
    {
      "epoch": 58.33,
      "grad_norm": 1.6469933986663818,
      "learning_rate": 3.978134127370286e-06,
      "loss": 0.5313,
      "step": 142500
    },
    {
      "epoch": 58.33,
      "grad_norm": 1.9895689487457275,
      "learning_rate": 3.978020426978926e-06,
      "loss": 0.535,
      "step": 142510
    },
    {
      "epoch": 58.34,
      "grad_norm": 2.7502660751342773,
      "learning_rate": 3.977906719101226e-06,
      "loss": 0.533,
      "step": 142520
    },
    {
      "epoch": 58.34,
      "grad_norm": 2.2844583988189697,
      "learning_rate": 3.977793003737708e-06,
      "loss": 0.5066,
      "step": 142530
    },
    {
      "epoch": 58.35,
      "grad_norm": 2.0699596405029297,
      "learning_rate": 3.97767928088889e-06,
      "loss": 0.5164,
      "step": 142540
    },
    {
      "epoch": 58.35,
      "grad_norm": 1.9669156074523926,
      "learning_rate": 3.977565550555297e-06,
      "loss": 0.5313,
      "step": 142550
    },
    {
      "epoch": 58.35,
      "grad_norm": 1.9595869779586792,
      "learning_rate": 3.977451812737447e-06,
      "loss": 0.5195,
      "step": 142560
    },
    {
      "epoch": 58.36,
      "grad_norm": 1.8495500087738037,
      "learning_rate": 3.977338067435862e-06,
      "loss": 0.5299,
      "step": 142570
    },
    {
      "epoch": 58.36,
      "grad_norm": 1.6337206363677979,
      "learning_rate": 3.977224314651063e-06,
      "loss": 0.5268,
      "step": 142580
    },
    {
      "epoch": 58.37,
      "grad_norm": 1.8993711471557617,
      "learning_rate": 3.977110554383572e-06,
      "loss": 0.5187,
      "step": 142590
    },
    {
      "epoch": 58.37,
      "grad_norm": 1.6686993837356567,
      "learning_rate": 3.976996786633909e-06,
      "loss": 0.5247,
      "step": 142600
    },
    {
      "epoch": 58.37,
      "grad_norm": 1.5930451154708862,
      "learning_rate": 3.976883011402595e-06,
      "loss": 0.5329,
      "step": 142610
    },
    {
      "epoch": 58.38,
      "grad_norm": 1.9472556114196777,
      "learning_rate": 3.976769228690152e-06,
      "loss": 0.5433,
      "step": 142620
    },
    {
      "epoch": 58.38,
      "grad_norm": 1.583888292312622,
      "learning_rate": 3.9766554384971e-06,
      "loss": 0.5225,
      "step": 142630
    },
    {
      "epoch": 58.39,
      "grad_norm": 2.0077455043792725,
      "learning_rate": 3.976541640823962e-06,
      "loss": 0.5212,
      "step": 142640
    },
    {
      "epoch": 58.39,
      "grad_norm": 1.9144527912139893,
      "learning_rate": 3.9764278356712575e-06,
      "loss": 0.527,
      "step": 142650
    },
    {
      "epoch": 58.4,
      "grad_norm": 2.1850852966308594,
      "learning_rate": 3.976314023039509e-06,
      "loss": 0.5324,
      "step": 142660
    },
    {
      "epoch": 58.4,
      "grad_norm": 1.8229278326034546,
      "learning_rate": 3.976200202929238e-06,
      "loss": 0.5394,
      "step": 142670
    },
    {
      "epoch": 58.4,
      "grad_norm": 2.364422082901001,
      "learning_rate": 3.976086375340966e-06,
      "loss": 0.5161,
      "step": 142680
    },
    {
      "epoch": 58.41,
      "grad_norm": 1.8139195442199707,
      "learning_rate": 3.9759725402752125e-06,
      "loss": 0.5424,
      "step": 142690
    },
    {
      "epoch": 58.41,
      "grad_norm": 2.1107373237609863,
      "learning_rate": 3.9758586977325004e-06,
      "loss": 0.5181,
      "step": 142700
    },
    {
      "epoch": 58.42,
      "grad_norm": 2.212252378463745,
      "learning_rate": 3.975744847713352e-06,
      "loss": 0.5411,
      "step": 142710
    },
    {
      "epoch": 58.42,
      "grad_norm": 1.910419225692749,
      "learning_rate": 3.975630990218287e-06,
      "loss": 0.5343,
      "step": 142720
    },
    {
      "epoch": 58.42,
      "grad_norm": 1.694434404373169,
      "learning_rate": 3.975517125247829e-06,
      "loss": 0.5208,
      "step": 142730
    },
    {
      "epoch": 58.43,
      "grad_norm": 1.6122747659683228,
      "learning_rate": 3.975403252802497e-06,
      "loss": 0.5186,
      "step": 142740
    },
    {
      "epoch": 58.43,
      "grad_norm": 1.8885414600372314,
      "learning_rate": 3.975289372882815e-06,
      "loss": 0.5433,
      "step": 142750
    },
    {
      "epoch": 58.44,
      "grad_norm": 2.0170257091522217,
      "learning_rate": 3.975175485489304e-06,
      "loss": 0.5443,
      "step": 142760
    },
    {
      "epoch": 58.44,
      "grad_norm": 2.0629546642303467,
      "learning_rate": 3.9750615906224854e-06,
      "loss": 0.5008,
      "step": 142770
    },
    {
      "epoch": 58.44,
      "grad_norm": 2.0944266319274902,
      "learning_rate": 3.974947688282882e-06,
      "loss": 0.5143,
      "step": 142780
    },
    {
      "epoch": 58.45,
      "grad_norm": 1.3590171337127686,
      "learning_rate": 3.974833778471013e-06,
      "loss": 0.5235,
      "step": 142790
    },
    {
      "epoch": 58.45,
      "grad_norm": 2.852595567703247,
      "learning_rate": 3.974719861187403e-06,
      "loss": 0.5249,
      "step": 142800
    },
    {
      "epoch": 58.46,
      "grad_norm": 2.21073842048645,
      "learning_rate": 3.974605936432572e-06,
      "loss": 0.5047,
      "step": 142810
    },
    {
      "epoch": 58.46,
      "grad_norm": 1.6570061445236206,
      "learning_rate": 3.974492004207043e-06,
      "loss": 0.5161,
      "step": 142820
    },
    {
      "epoch": 58.47,
      "grad_norm": 2.133294105529785,
      "learning_rate": 3.9743780645113365e-06,
      "loss": 0.5191,
      "step": 142830
    },
    {
      "epoch": 58.47,
      "grad_norm": 1.5680954456329346,
      "learning_rate": 3.974264117345976e-06,
      "loss": 0.5319,
      "step": 142840
    },
    {
      "epoch": 58.47,
      "grad_norm": 2.287567377090454,
      "learning_rate": 3.9741501627114825e-06,
      "loss": 0.5285,
      "step": 142850
    },
    {
      "epoch": 58.48,
      "grad_norm": 1.8092002868652344,
      "learning_rate": 3.9740362006083785e-06,
      "loss": 0.5244,
      "step": 142860
    },
    {
      "epoch": 58.48,
      "grad_norm": 2.222963571548462,
      "learning_rate": 3.9739222310371865e-06,
      "loss": 0.5489,
      "step": 142870
    },
    {
      "epoch": 58.49,
      "grad_norm": 2.018723487854004,
      "learning_rate": 3.9738082539984275e-06,
      "loss": 0.5257,
      "step": 142880
    },
    {
      "epoch": 58.49,
      "grad_norm": 2.2503292560577393,
      "learning_rate": 3.973694269492624e-06,
      "loss": 0.5274,
      "step": 142890
    },
    {
      "epoch": 58.49,
      "grad_norm": 1.9513928890228271,
      "learning_rate": 3.973580277520298e-06,
      "loss": 0.5486,
      "step": 142900
    },
    {
      "epoch": 58.5,
      "grad_norm": 1.7305337190628052,
      "learning_rate": 3.9734662780819725e-06,
      "loss": 0.5361,
      "step": 142910
    },
    {
      "epoch": 58.5,
      "grad_norm": 2.0460264682769775,
      "learning_rate": 3.973352271178169e-06,
      "loss": 0.5198,
      "step": 142920
    },
    {
      "epoch": 58.51,
      "grad_norm": 2.2331655025482178,
      "learning_rate": 3.97323825680941e-06,
      "loss": 0.5461,
      "step": 142930
    },
    {
      "epoch": 58.51,
      "grad_norm": 2.397364854812622,
      "learning_rate": 3.973124234976216e-06,
      "loss": 0.5222,
      "step": 142940
    },
    {
      "epoch": 58.51,
      "grad_norm": 1.7991840839385986,
      "learning_rate": 3.9730102056791125e-06,
      "loss": 0.5349,
      "step": 142950
    },
    {
      "epoch": 58.52,
      "grad_norm": 1.8076329231262207,
      "learning_rate": 3.972896168918621e-06,
      "loss": 0.5252,
      "step": 142960
    },
    {
      "epoch": 58.52,
      "grad_norm": 1.9209214448928833,
      "learning_rate": 3.972782124695262e-06,
      "loss": 0.5466,
      "step": 142970
    },
    {
      "epoch": 58.53,
      "grad_norm": 1.909952998161316,
      "learning_rate": 3.972668073009559e-06,
      "loss": 0.5474,
      "step": 142980
    },
    {
      "epoch": 58.53,
      "grad_norm": 1.7906204462051392,
      "learning_rate": 3.972554013862035e-06,
      "loss": 0.5392,
      "step": 142990
    },
    {
      "epoch": 58.53,
      "grad_norm": 2.9913265705108643,
      "learning_rate": 3.972439947253213e-06,
      "loss": 0.5289,
      "step": 143000
    },
    {
      "epoch": 58.54,
      "grad_norm": 1.8858689069747925,
      "learning_rate": 3.972325873183614e-06,
      "loss": 0.5274,
      "step": 143010
    },
    {
      "epoch": 58.54,
      "grad_norm": 1.4074314832687378,
      "learning_rate": 3.972211791653761e-06,
      "loss": 0.5231,
      "step": 143020
    },
    {
      "epoch": 58.55,
      "grad_norm": 1.6206252574920654,
      "learning_rate": 3.972097702664176e-06,
      "loss": 0.5211,
      "step": 143030
    },
    {
      "epoch": 58.55,
      "grad_norm": 2.5172855854034424,
      "learning_rate": 3.971983606215383e-06,
      "loss": 0.5065,
      "step": 143040
    },
    {
      "epoch": 58.56,
      "grad_norm": 1.8819330930709839,
      "learning_rate": 3.971869502307904e-06,
      "loss": 0.5317,
      "step": 143050
    },
    {
      "epoch": 58.56,
      "grad_norm": 2.6076865196228027,
      "learning_rate": 3.971755390942262e-06,
      "loss": 0.5291,
      "step": 143060
    },
    {
      "epoch": 58.56,
      "grad_norm": 2.2426702976226807,
      "learning_rate": 3.971641272118979e-06,
      "loss": 0.5386,
      "step": 143070
    },
    {
      "epoch": 58.57,
      "grad_norm": 2.8639583587646484,
      "learning_rate": 3.971527145838578e-06,
      "loss": 0.5223,
      "step": 143080
    },
    {
      "epoch": 58.57,
      "grad_norm": 2.236696243286133,
      "learning_rate": 3.971413012101582e-06,
      "loss": 0.516,
      "step": 143090
    },
    {
      "epoch": 58.58,
      "grad_norm": 2.26788067817688,
      "learning_rate": 3.971298870908514e-06,
      "loss": 0.5309,
      "step": 143100
    },
    {
      "epoch": 58.58,
      "grad_norm": 1.7508559226989746,
      "learning_rate": 3.971184722259897e-06,
      "loss": 0.5304,
      "step": 143110
    },
    {
      "epoch": 58.58,
      "grad_norm": 1.4855762720108032,
      "learning_rate": 3.971070566156254e-06,
      "loss": 0.5447,
      "step": 143120
    },
    {
      "epoch": 58.59,
      "grad_norm": 1.9939467906951904,
      "learning_rate": 3.970956402598106e-06,
      "loss": 0.519,
      "step": 143130
    },
    {
      "epoch": 58.59,
      "grad_norm": 2.137012481689453,
      "learning_rate": 3.970842231585979e-06,
      "loss": 0.5088,
      "step": 143140
    },
    {
      "epoch": 58.6,
      "grad_norm": 2.1155591011047363,
      "learning_rate": 3.970728053120394e-06,
      "loss": 0.5331,
      "step": 143150
    },
    {
      "epoch": 58.6,
      "grad_norm": 1.7892650365829468,
      "learning_rate": 3.9706138672018735e-06,
      "loss": 0.5199,
      "step": 143160
    },
    {
      "epoch": 58.6,
      "grad_norm": 1.966902732849121,
      "learning_rate": 3.970499673830943e-06,
      "loss": 0.5114,
      "step": 143170
    },
    {
      "epoch": 58.61,
      "grad_norm": 2.231707811355591,
      "learning_rate": 3.970385473008123e-06,
      "loss": 0.5194,
      "step": 143180
    },
    {
      "epoch": 58.61,
      "grad_norm": 2.221517562866211,
      "learning_rate": 3.970271264733939e-06,
      "loss": 0.5228,
      "step": 143190
    },
    {
      "epoch": 58.62,
      "grad_norm": 1.6774882078170776,
      "learning_rate": 3.9701570490089125e-06,
      "loss": 0.5268,
      "step": 143200
    },
    {
      "epoch": 58.62,
      "grad_norm": 3.135699510574341,
      "learning_rate": 3.970042825833567e-06,
      "loss": 0.5214,
      "step": 143210
    },
    {
      "epoch": 58.62,
      "grad_norm": 2.0218498706817627,
      "learning_rate": 3.9699285952084255e-06,
      "loss": 0.5222,
      "step": 143220
    },
    {
      "epoch": 58.63,
      "grad_norm": 1.6851969957351685,
      "learning_rate": 3.9698143571340124e-06,
      "loss": 0.5251,
      "step": 143230
    },
    {
      "epoch": 58.63,
      "grad_norm": 1.7263283729553223,
      "learning_rate": 3.96970011161085e-06,
      "loss": 0.5196,
      "step": 143240
    },
    {
      "epoch": 58.64,
      "grad_norm": 2.137221097946167,
      "learning_rate": 3.969585858639462e-06,
      "loss": 0.5084,
      "step": 143250
    },
    {
      "epoch": 58.64,
      "grad_norm": 1.933714509010315,
      "learning_rate": 3.969471598220372e-06,
      "loss": 0.5241,
      "step": 143260
    },
    {
      "epoch": 58.65,
      "grad_norm": 2.2292866706848145,
      "learning_rate": 3.9693573303541035e-06,
      "loss": 0.5541,
      "step": 143270
    },
    {
      "epoch": 58.65,
      "grad_norm": 3.2688148021698,
      "learning_rate": 3.9692430550411786e-06,
      "loss": 0.5411,
      "step": 143280
    },
    {
      "epoch": 58.65,
      "grad_norm": 1.9785619974136353,
      "learning_rate": 3.9691287722821225e-06,
      "loss": 0.5386,
      "step": 143290
    },
    {
      "epoch": 58.66,
      "grad_norm": 1.9848538637161255,
      "learning_rate": 3.969014482077457e-06,
      "loss": 0.5239,
      "step": 143300
    },
    {
      "epoch": 58.66,
      "grad_norm": 2.4814884662628174,
      "learning_rate": 3.968900184427708e-06,
      "loss": 0.5359,
      "step": 143310
    },
    {
      "epoch": 58.67,
      "grad_norm": 1.9619741439819336,
      "learning_rate": 3.9687858793333965e-06,
      "loss": 0.5309,
      "step": 143320
    },
    {
      "epoch": 58.67,
      "grad_norm": 2.6907715797424316,
      "learning_rate": 3.968671566795047e-06,
      "loss": 0.5174,
      "step": 143330
    },
    {
      "epoch": 58.67,
      "grad_norm": 2.4653797149658203,
      "learning_rate": 3.968557246813185e-06,
      "loss": 0.5382,
      "step": 143340
    },
    {
      "epoch": 58.68,
      "grad_norm": 1.8420928716659546,
      "learning_rate": 3.968442919388332e-06,
      "loss": 0.5454,
      "step": 143350
    },
    {
      "epoch": 58.68,
      "grad_norm": 2.080352544784546,
      "learning_rate": 3.968328584521012e-06,
      "loss": 0.5109,
      "step": 143360
    },
    {
      "epoch": 58.69,
      "grad_norm": 2.1606361865997314,
      "learning_rate": 3.968214242211749e-06,
      "loss": 0.5381,
      "step": 143370
    },
    {
      "epoch": 58.69,
      "grad_norm": 1.6623913049697876,
      "learning_rate": 3.968099892461067e-06,
      "loss": 0.5268,
      "step": 143380
    },
    {
      "epoch": 58.69,
      "grad_norm": 1.9302785396575928,
      "learning_rate": 3.96798553526949e-06,
      "loss": 0.5432,
      "step": 143390
    },
    {
      "epoch": 58.7,
      "grad_norm": 1.8639527559280396,
      "learning_rate": 3.967871170637541e-06,
      "loss": 0.53,
      "step": 143400
    },
    {
      "epoch": 58.7,
      "grad_norm": 2.0029783248901367,
      "learning_rate": 3.967756798565745e-06,
      "loss": 0.5117,
      "step": 143410
    },
    {
      "epoch": 58.71,
      "grad_norm": 2.182023286819458,
      "learning_rate": 3.967642419054626e-06,
      "loss": 0.5268,
      "step": 143420
    },
    {
      "epoch": 58.71,
      "grad_norm": 2.351064682006836,
      "learning_rate": 3.967528032104706e-06,
      "loss": 0.5128,
      "step": 143430
    },
    {
      "epoch": 58.71,
      "grad_norm": 2.0683071613311768,
      "learning_rate": 3.967413637716511e-06,
      "loss": 0.5185,
      "step": 143440
    },
    {
      "epoch": 58.72,
      "grad_norm": 1.7215001583099365,
      "learning_rate": 3.967299235890564e-06,
      "loss": 0.5235,
      "step": 143450
    },
    {
      "epoch": 58.72,
      "grad_norm": 1.8036288022994995,
      "learning_rate": 3.967184826627389e-06,
      "loss": 0.5445,
      "step": 143460
    },
    {
      "epoch": 58.73,
      "grad_norm": 2.008671283721924,
      "learning_rate": 3.967070409927512e-06,
      "loss": 0.5423,
      "step": 143470
    },
    {
      "epoch": 58.73,
      "grad_norm": 2.19372820854187,
      "learning_rate": 3.966955985791454e-06,
      "loss": 0.5179,
      "step": 143480
    },
    {
      "epoch": 58.74,
      "grad_norm": 2.317241668701172,
      "learning_rate": 3.966841554219742e-06,
      "loss": 0.5217,
      "step": 143490
    },
    {
      "epoch": 58.74,
      "grad_norm": 1.5386618375778198,
      "learning_rate": 3.966727115212898e-06,
      "loss": 0.543,
      "step": 143500
    },
    {
      "epoch": 58.74,
      "grad_norm": 5.03570032119751,
      "learning_rate": 3.966612668771447e-06,
      "loss": 0.5346,
      "step": 143510
    },
    {
      "epoch": 58.75,
      "grad_norm": 1.6037195920944214,
      "learning_rate": 3.966498214895914e-06,
      "loss": 0.5343,
      "step": 143520
    },
    {
      "epoch": 58.75,
      "grad_norm": 2.428983449935913,
      "learning_rate": 3.966383753586823e-06,
      "loss": 0.5305,
      "step": 143530
    },
    {
      "epoch": 58.76,
      "grad_norm": 1.7887957096099854,
      "learning_rate": 3.966269284844698e-06,
      "loss": 0.5174,
      "step": 143540
    },
    {
      "epoch": 58.76,
      "grad_norm": 1.8681401014328003,
      "learning_rate": 3.966154808670064e-06,
      "loss": 0.5301,
      "step": 143550
    },
    {
      "epoch": 58.76,
      "grad_norm": 1.868898630142212,
      "learning_rate": 3.966040325063444e-06,
      "loss": 0.5292,
      "step": 143560
    },
    {
      "epoch": 58.77,
      "grad_norm": 1.3129762411117554,
      "learning_rate": 3.965925834025363e-06,
      "loss": 0.5109,
      "step": 143570
    },
    {
      "epoch": 58.77,
      "grad_norm": 1.8932524919509888,
      "learning_rate": 3.965811335556346e-06,
      "loss": 0.5325,
      "step": 143580
    },
    {
      "epoch": 58.78,
      "grad_norm": 1.9790372848510742,
      "learning_rate": 3.9656968296569175e-06,
      "loss": 0.5342,
      "step": 143590
    },
    {
      "epoch": 58.78,
      "grad_norm": 2.09902286529541,
      "learning_rate": 3.9655823163276014e-06,
      "loss": 0.5346,
      "step": 143600
    },
    {
      "epoch": 58.78,
      "grad_norm": 2.4924073219299316,
      "learning_rate": 3.965467795568923e-06,
      "loss": 0.5331,
      "step": 143610
    },
    {
      "epoch": 58.79,
      "grad_norm": 2.458806276321411,
      "learning_rate": 3.965353267381406e-06,
      "loss": 0.5207,
      "step": 143620
    },
    {
      "epoch": 58.79,
      "grad_norm": 1.7078965902328491,
      "learning_rate": 3.965238731765576e-06,
      "loss": 0.5135,
      "step": 143630
    },
    {
      "epoch": 58.8,
      "grad_norm": 1.9914335012435913,
      "learning_rate": 3.965124188721957e-06,
      "loss": 0.5265,
      "step": 143640
    },
    {
      "epoch": 58.8,
      "grad_norm": 1.7870779037475586,
      "learning_rate": 3.965009638251075e-06,
      "loss": 0.5409,
      "step": 143650
    },
    {
      "epoch": 58.8,
      "grad_norm": 1.9621738195419312,
      "learning_rate": 3.964895080353452e-06,
      "loss": 0.5213,
      "step": 143660
    },
    {
      "epoch": 58.81,
      "grad_norm": 2.120826482772827,
      "learning_rate": 3.964780515029615e-06,
      "loss": 0.5367,
      "step": 143670
    },
    {
      "epoch": 58.81,
      "grad_norm": 2.197340726852417,
      "learning_rate": 3.964665942280088e-06,
      "loss": 0.5736,
      "step": 143680
    },
    {
      "epoch": 58.82,
      "grad_norm": 1.2372779846191406,
      "learning_rate": 3.964551362105397e-06,
      "loss": 0.5305,
      "step": 143690
    },
    {
      "epoch": 58.82,
      "grad_norm": 1.8432384729385376,
      "learning_rate": 3.964436774506066e-06,
      "loss": 0.5183,
      "step": 143700
    },
    {
      "epoch": 58.83,
      "grad_norm": 2.361260414123535,
      "learning_rate": 3.9643221794826194e-06,
      "loss": 0.5308,
      "step": 143710
    },
    {
      "epoch": 58.83,
      "grad_norm": 2.2928993701934814,
      "learning_rate": 3.964207577035582e-06,
      "loss": 0.5245,
      "step": 143720
    },
    {
      "epoch": 58.83,
      "grad_norm": 2.648019552230835,
      "learning_rate": 3.96409296716548e-06,
      "loss": 0.52,
      "step": 143730
    },
    {
      "epoch": 58.84,
      "grad_norm": 2.442563533782959,
      "learning_rate": 3.963978349872838e-06,
      "loss": 0.5187,
      "step": 143740
    },
    {
      "epoch": 58.84,
      "grad_norm": 2.0026421546936035,
      "learning_rate": 3.963863725158181e-06,
      "loss": 0.5243,
      "step": 143750
    },
    {
      "epoch": 58.85,
      "grad_norm": 2.3879098892211914,
      "learning_rate": 3.963749093022034e-06,
      "loss": 0.5138,
      "step": 143760
    },
    {
      "epoch": 58.85,
      "grad_norm": 2.054311990737915,
      "learning_rate": 3.963634453464922e-06,
      "loss": 0.5264,
      "step": 143770
    },
    {
      "epoch": 58.85,
      "grad_norm": 1.9768189191818237,
      "learning_rate": 3.9635198064873695e-06,
      "loss": 0.5082,
      "step": 143780
    },
    {
      "epoch": 58.86,
      "grad_norm": 2.1987040042877197,
      "learning_rate": 3.9634051520899025e-06,
      "loss": 0.5348,
      "step": 143790
    },
    {
      "epoch": 58.86,
      "grad_norm": 1.614118218421936,
      "learning_rate": 3.963290490273046e-06,
      "loss": 0.523,
      "step": 143800
    },
    {
      "epoch": 58.87,
      "grad_norm": 2.4024927616119385,
      "learning_rate": 3.963175821037326e-06,
      "loss": 0.5508,
      "step": 143810
    },
    {
      "epoch": 58.87,
      "grad_norm": 1.6336511373519897,
      "learning_rate": 3.963061144383266e-06,
      "loss": 0.5373,
      "step": 143820
    },
    {
      "epoch": 58.87,
      "grad_norm": 2.2252564430236816,
      "learning_rate": 3.962946460311394e-06,
      "loss": 0.5372,
      "step": 143830
    },
    {
      "epoch": 58.88,
      "grad_norm": 1.8351138830184937,
      "learning_rate": 3.962831768822233e-06,
      "loss": 0.5361,
      "step": 143840
    },
    {
      "epoch": 58.88,
      "grad_norm": 1.9577707052230835,
      "learning_rate": 3.9627170699163086e-06,
      "loss": 0.5312,
      "step": 143850
    },
    {
      "epoch": 58.89,
      "grad_norm": 2.1391005516052246,
      "learning_rate": 3.962602363594148e-06,
      "loss": 0.5332,
      "step": 143860
    },
    {
      "epoch": 58.89,
      "grad_norm": 2.447092056274414,
      "learning_rate": 3.962487649856274e-06,
      "loss": 0.5227,
      "step": 143870
    },
    {
      "epoch": 58.89,
      "grad_norm": 1.96880042552948,
      "learning_rate": 3.962372928703215e-06,
      "loss": 0.5363,
      "step": 143880
    },
    {
      "epoch": 58.9,
      "grad_norm": 1.8358572721481323,
      "learning_rate": 3.962258200135494e-06,
      "loss": 0.534,
      "step": 143890
    },
    {
      "epoch": 58.9,
      "grad_norm": 2.4979021549224854,
      "learning_rate": 3.962143464153638e-06,
      "loss": 0.514,
      "step": 143900
    },
    {
      "epoch": 58.91,
      "grad_norm": 1.9340497255325317,
      "learning_rate": 3.962028720758172e-06,
      "loss": 0.523,
      "step": 143910
    },
    {
      "epoch": 58.91,
      "grad_norm": 2.1336746215820312,
      "learning_rate": 3.961913969949622e-06,
      "loss": 0.5234,
      "step": 143920
    },
    {
      "epoch": 58.92,
      "grad_norm": 1.8386826515197754,
      "learning_rate": 3.961799211728514e-06,
      "loss": 0.5385,
      "step": 143930
    },
    {
      "epoch": 58.92,
      "grad_norm": 1.938880443572998,
      "learning_rate": 3.961684446095372e-06,
      "loss": 0.532,
      "step": 143940
    },
    {
      "epoch": 58.92,
      "grad_norm": 1.489012598991394,
      "learning_rate": 3.961569673050724e-06,
      "loss": 0.5278,
      "step": 143950
    },
    {
      "epoch": 58.93,
      "grad_norm": 2.498570203781128,
      "learning_rate": 3.9614548925950935e-06,
      "loss": 0.5226,
      "step": 143960
    },
    {
      "epoch": 58.93,
      "grad_norm": 2.2273950576782227,
      "learning_rate": 3.961340104729008e-06,
      "loss": 0.5391,
      "step": 143970
    },
    {
      "epoch": 58.94,
      "grad_norm": 2.3358240127563477,
      "learning_rate": 3.961225309452993e-06,
      "loss": 0.5485,
      "step": 143980
    },
    {
      "epoch": 58.94,
      "grad_norm": 1.8295124769210815,
      "learning_rate": 3.961110506767573e-06,
      "loss": 0.5221,
      "step": 143990
    },
    {
      "epoch": 58.94,
      "grad_norm": 1.785682201385498,
      "learning_rate": 3.9609956966732755e-06,
      "loss": 0.5204,
      "step": 144000
    },
    {
      "epoch": 58.95,
      "grad_norm": 2.4189040660858154,
      "learning_rate": 3.9608808791706266e-06,
      "loss": 0.5226,
      "step": 144010
    },
    {
      "epoch": 58.95,
      "grad_norm": 1.7168282270431519,
      "learning_rate": 3.960766054260151e-06,
      "loss": 0.5099,
      "step": 144020
    },
    {
      "epoch": 58.96,
      "grad_norm": 1.545819878578186,
      "learning_rate": 3.960651221942376e-06,
      "loss": 0.5336,
      "step": 144030
    },
    {
      "epoch": 58.96,
      "grad_norm": 2.028578519821167,
      "learning_rate": 3.960536382217826e-06,
      "loss": 0.551,
      "step": 144040
    },
    {
      "epoch": 58.96,
      "grad_norm": 1.7718162536621094,
      "learning_rate": 3.9604215350870285e-06,
      "loss": 0.5468,
      "step": 144050
    },
    {
      "epoch": 58.97,
      "grad_norm": 2.329446792602539,
      "learning_rate": 3.960306680550508e-06,
      "loss": 0.528,
      "step": 144060
    },
    {
      "epoch": 58.97,
      "grad_norm": 2.0353684425354004,
      "learning_rate": 3.9601918186087935e-06,
      "loss": 0.5164,
      "step": 144070
    },
    {
      "epoch": 58.98,
      "grad_norm": 1.7973123788833618,
      "learning_rate": 3.960076949262409e-06,
      "loss": 0.5403,
      "step": 144080
    },
    {
      "epoch": 58.98,
      "grad_norm": 2.4850995540618896,
      "learning_rate": 3.95996207251188e-06,
      "loss": 0.5089,
      "step": 144090
    },
    {
      "epoch": 58.98,
      "grad_norm": 3.183542251586914,
      "learning_rate": 3.9598471883577345e-06,
      "loss": 0.5261,
      "step": 144100
    },
    {
      "epoch": 58.99,
      "grad_norm": 1.7932755947113037,
      "learning_rate": 3.959732296800498e-06,
      "loss": 0.5445,
      "step": 144110
    },
    {
      "epoch": 58.99,
      "grad_norm": 1.8444385528564453,
      "learning_rate": 3.959617397840697e-06,
      "loss": 0.5506,
      "step": 144120
    },
    {
      "epoch": 59.0,
      "grad_norm": 2.0827271938323975,
      "learning_rate": 3.959502491478857e-06,
      "loss": 0.5411,
      "step": 144130
    },
    {
      "epoch": 59.0,
      "eval_loss": 0.5278185606002808,
      "eval_runtime": 52.8856,
      "eval_samples_per_second": 65.216,
      "eval_steps_per_second": 8.169,
      "step": 144137
    },
    {
      "epoch": 59.0,
      "grad_norm": 1.9734784364700317,
      "learning_rate": 3.9593875777155056e-06,
      "loss": 0.5373,
      "step": 144140
    },
    {
      "epoch": 59.01,
      "grad_norm": 1.405210018157959,
      "learning_rate": 3.959272656551169e-06,
      "loss": 0.5241,
      "step": 144150
    },
    {
      "epoch": 59.01,
      "grad_norm": 2.057932138442993,
      "learning_rate": 3.959157727986372e-06,
      "loss": 0.54,
      "step": 144160
    },
    {
      "epoch": 59.01,
      "grad_norm": 2.3481438159942627,
      "learning_rate": 3.959042792021643e-06,
      "loss": 0.5312,
      "step": 144170
    },
    {
      "epoch": 59.02,
      "grad_norm": 2.0339996814727783,
      "learning_rate": 3.958927848657509e-06,
      "loss": 0.537,
      "step": 144180
    },
    {
      "epoch": 59.02,
      "grad_norm": 1.8685791492462158,
      "learning_rate": 3.9588128978944945e-06,
      "loss": 0.5462,
      "step": 144190
    },
    {
      "epoch": 59.03,
      "grad_norm": 2.151822328567505,
      "learning_rate": 3.9586979397331275e-06,
      "loss": 0.5462,
      "step": 144200
    },
    {
      "epoch": 59.03,
      "grad_norm": 2.1662323474884033,
      "learning_rate": 3.958582974173933e-06,
      "loss": 0.5124,
      "step": 144210
    },
    {
      "epoch": 59.03,
      "grad_norm": 1.5922681093215942,
      "learning_rate": 3.95846800121744e-06,
      "loss": 0.5124,
      "step": 144220
    },
    {
      "epoch": 59.04,
      "grad_norm": 2.2547242641448975,
      "learning_rate": 3.958353020864174e-06,
      "loss": 0.5261,
      "step": 144230
    },
    {
      "epoch": 59.04,
      "grad_norm": 1.779331088066101,
      "learning_rate": 3.95823803311466e-06,
      "loss": 0.5204,
      "step": 144240
    },
    {
      "epoch": 59.05,
      "grad_norm": 2.0876545906066895,
      "learning_rate": 3.958123037969427e-06,
      "loss": 0.5201,
      "step": 144250
    },
    {
      "epoch": 59.05,
      "grad_norm": 2.6513450145721436,
      "learning_rate": 3.9580080354290015e-06,
      "loss": 0.5256,
      "step": 144260
    },
    {
      "epoch": 59.05,
      "grad_norm": 1.9410111904144287,
      "learning_rate": 3.95789302549391e-06,
      "loss": 0.5204,
      "step": 144270
    },
    {
      "epoch": 59.06,
      "grad_norm": 2.350358009338379,
      "learning_rate": 3.957778008164679e-06,
      "loss": 0.5511,
      "step": 144280
    },
    {
      "epoch": 59.06,
      "grad_norm": 1.7180311679840088,
      "learning_rate": 3.9576629834418364e-06,
      "loss": 0.4923,
      "step": 144290
    },
    {
      "epoch": 59.07,
      "grad_norm": 2.0334835052490234,
      "learning_rate": 3.9575479513259076e-06,
      "loss": 0.5502,
      "step": 144300
    },
    {
      "epoch": 59.07,
      "grad_norm": 2.4011528491973877,
      "learning_rate": 3.95743291181742e-06,
      "loss": 0.5082,
      "step": 144310
    },
    {
      "epoch": 59.07,
      "grad_norm": 1.754312515258789,
      "learning_rate": 3.957317864916902e-06,
      "loss": 0.5125,
      "step": 144320
    },
    {
      "epoch": 59.08,
      "grad_norm": 1.6116002798080444,
      "learning_rate": 3.957202810624878e-06,
      "loss": 0.5164,
      "step": 144330
    },
    {
      "epoch": 59.08,
      "grad_norm": 2.4139811992645264,
      "learning_rate": 3.957087748941878e-06,
      "loss": 0.5201,
      "step": 144340
    },
    {
      "epoch": 59.09,
      "grad_norm": 1.8928351402282715,
      "learning_rate": 3.956972679868427e-06,
      "loss": 0.5244,
      "step": 144350
    },
    {
      "epoch": 59.09,
      "grad_norm": 2.062747001647949,
      "learning_rate": 3.956857603405053e-06,
      "loss": 0.5108,
      "step": 144360
    },
    {
      "epoch": 59.1,
      "grad_norm": 2.18442964553833,
      "learning_rate": 3.956742519552283e-06,
      "loss": 0.4954,
      "step": 144370
    },
    {
      "epoch": 59.1,
      "grad_norm": 1.641286849975586,
      "learning_rate": 3.956627428310643e-06,
      "loss": 0.5273,
      "step": 144380
    },
    {
      "epoch": 59.1,
      "grad_norm": 2.1983938217163086,
      "learning_rate": 3.956512329680662e-06,
      "loss": 0.528,
      "step": 144390
    },
    {
      "epoch": 59.11,
      "grad_norm": 2.1132962703704834,
      "learning_rate": 3.956397223662867e-06,
      "loss": 0.5508,
      "step": 144400
    },
    {
      "epoch": 59.11,
      "grad_norm": 2.273843765258789,
      "learning_rate": 3.956282110257784e-06,
      "loss": 0.5236,
      "step": 144410
    },
    {
      "epoch": 59.12,
      "grad_norm": 2.4306275844573975,
      "learning_rate": 3.956166989465942e-06,
      "loss": 0.5358,
      "step": 144420
    },
    {
      "epoch": 59.12,
      "grad_norm": 1.6391364336013794,
      "learning_rate": 3.9560518612878666e-06,
      "loss": 0.5433,
      "step": 144430
    },
    {
      "epoch": 59.12,
      "grad_norm": 1.554274082183838,
      "learning_rate": 3.955936725724087e-06,
      "loss": 0.5399,
      "step": 144440
    },
    {
      "epoch": 59.13,
      "grad_norm": 2.8176229000091553,
      "learning_rate": 3.955821582775128e-06,
      "loss": 0.5222,
      "step": 144450
    },
    {
      "epoch": 59.13,
      "grad_norm": 1.7682697772979736,
      "learning_rate": 3.95570643244152e-06,
      "loss": 0.5131,
      "step": 144460
    },
    {
      "epoch": 59.14,
      "grad_norm": 2.2484190464019775,
      "learning_rate": 3.95559127472379e-06,
      "loss": 0.5404,
      "step": 144470
    },
    {
      "epoch": 59.14,
      "grad_norm": 2.0958244800567627,
      "learning_rate": 3.9554761096224635e-06,
      "loss": 0.5184,
      "step": 144480
    },
    {
      "epoch": 59.14,
      "grad_norm": 2.6351747512817383,
      "learning_rate": 3.95536093713807e-06,
      "loss": 0.5351,
      "step": 144490
    },
    {
      "epoch": 59.15,
      "grad_norm": 2.1911520957946777,
      "learning_rate": 3.955245757271136e-06,
      "loss": 0.5697,
      "step": 144500
    },
    {
      "epoch": 59.15,
      "grad_norm": 1.6727211475372314,
      "learning_rate": 3.955130570022189e-06,
      "loss": 0.5378,
      "step": 144510
    },
    {
      "epoch": 59.16,
      "grad_norm": 2.0820648670196533,
      "learning_rate": 3.9550153753917585e-06,
      "loss": 0.5177,
      "step": 144520
    },
    {
      "epoch": 59.16,
      "grad_norm": 2.023057460784912,
      "learning_rate": 3.95490017338037e-06,
      "loss": 0.5515,
      "step": 144530
    },
    {
      "epoch": 59.16,
      "grad_norm": 1.826634407043457,
      "learning_rate": 3.954784963988552e-06,
      "loss": 0.5244,
      "step": 144540
    },
    {
      "epoch": 59.17,
      "grad_norm": 1.7641994953155518,
      "learning_rate": 3.954669747216833e-06,
      "loss": 0.5196,
      "step": 144550
    },
    {
      "epoch": 59.17,
      "grad_norm": 2.9824492931365967,
      "learning_rate": 3.95455452306574e-06,
      "loss": 0.5223,
      "step": 144560
    },
    {
      "epoch": 59.18,
      "grad_norm": 2.123081922531128,
      "learning_rate": 3.954439291535799e-06,
      "loss": 0.5325,
      "step": 144570
    },
    {
      "epoch": 59.18,
      "grad_norm": 2.2038352489471436,
      "learning_rate": 3.954324052627542e-06,
      "loss": 0.5319,
      "step": 144580
    },
    {
      "epoch": 59.19,
      "grad_norm": 1.9250140190124512,
      "learning_rate": 3.954208806341494e-06,
      "loss": 0.5146,
      "step": 144590
    },
    {
      "epoch": 59.19,
      "grad_norm": 2.0873115062713623,
      "learning_rate": 3.954093552678184e-06,
      "loss": 0.5228,
      "step": 144600
    },
    {
      "epoch": 59.19,
      "grad_norm": 1.5839757919311523,
      "learning_rate": 3.953978291638139e-06,
      "loss": 0.5143,
      "step": 144610
    },
    {
      "epoch": 59.2,
      "grad_norm": 1.8016234636306763,
      "learning_rate": 3.9538630232218885e-06,
      "loss": 0.5239,
      "step": 144620
    },
    {
      "epoch": 59.2,
      "grad_norm": 2.023494243621826,
      "learning_rate": 3.953747747429959e-06,
      "loss": 0.5183,
      "step": 144630
    },
    {
      "epoch": 59.21,
      "grad_norm": 1.6765527725219727,
      "learning_rate": 3.953632464262878e-06,
      "loss": 0.53,
      "step": 144640
    },
    {
      "epoch": 59.21,
      "grad_norm": 1.4390060901641846,
      "learning_rate": 3.953517173721176e-06,
      "loss": 0.5385,
      "step": 144650
    },
    {
      "epoch": 59.21,
      "grad_norm": 2.1154932975769043,
      "learning_rate": 3.95340187580538e-06,
      "loss": 0.5431,
      "step": 144660
    },
    {
      "epoch": 59.22,
      "grad_norm": 1.8681049346923828,
      "learning_rate": 3.953286570516018e-06,
      "loss": 0.5059,
      "step": 144670
    },
    {
      "epoch": 59.22,
      "grad_norm": 2.0279595851898193,
      "learning_rate": 3.953171257853618e-06,
      "loss": 0.5274,
      "step": 144680
    },
    {
      "epoch": 59.23,
      "grad_norm": 2.527442455291748,
      "learning_rate": 3.953055937818708e-06,
      "loss": 0.5329,
      "step": 144690
    },
    {
      "epoch": 59.23,
      "grad_norm": 1.9868649244308472,
      "learning_rate": 3.952940610411818e-06,
      "loss": 0.5434,
      "step": 144700
    },
    {
      "epoch": 59.23,
      "grad_norm": 1.8749221563339233,
      "learning_rate": 3.952825275633473e-06,
      "loss": 0.538,
      "step": 144710
    },
    {
      "epoch": 59.24,
      "grad_norm": 1.9413255453109741,
      "learning_rate": 3.952709933484205e-06,
      "loss": 0.5356,
      "step": 144720
    },
    {
      "epoch": 59.24,
      "grad_norm": 2.382045030593872,
      "learning_rate": 3.9525945839645406e-06,
      "loss": 0.5377,
      "step": 144730
    },
    {
      "epoch": 59.25,
      "grad_norm": 2.1557648181915283,
      "learning_rate": 3.952479227075008e-06,
      "loss": 0.5311,
      "step": 144740
    },
    {
      "epoch": 59.25,
      "grad_norm": 2.0098824501037598,
      "learning_rate": 3.952363862816136e-06,
      "loss": 0.5288,
      "step": 144750
    },
    {
      "epoch": 59.26,
      "grad_norm": 1.8809672594070435,
      "learning_rate": 3.952248491188453e-06,
      "loss": 0.5232,
      "step": 144760
    },
    {
      "epoch": 59.26,
      "grad_norm": 1.9878432750701904,
      "learning_rate": 3.952133112192488e-06,
      "loss": 0.5146,
      "step": 144770
    },
    {
      "epoch": 59.26,
      "grad_norm": 2.391862154006958,
      "learning_rate": 3.952017725828768e-06,
      "loss": 0.5272,
      "step": 144780
    },
    {
      "epoch": 59.27,
      "grad_norm": 1.8121700286865234,
      "learning_rate": 3.951902332097823e-06,
      "loss": 0.5384,
      "step": 144790
    },
    {
      "epoch": 59.27,
      "grad_norm": 1.9147939682006836,
      "learning_rate": 3.951786931000181e-06,
      "loss": 0.5271,
      "step": 144800
    },
    {
      "epoch": 59.28,
      "grad_norm": 2.090498208999634,
      "learning_rate": 3.951671522536372e-06,
      "loss": 0.535,
      "step": 144810
    },
    {
      "epoch": 59.28,
      "grad_norm": 1.8357521295547485,
      "learning_rate": 3.951556106706922e-06,
      "loss": 0.5374,
      "step": 144820
    },
    {
      "epoch": 59.28,
      "grad_norm": 2.0333542823791504,
      "learning_rate": 3.951440683512363e-06,
      "loss": 0.5337,
      "step": 144830
    },
    {
      "epoch": 59.29,
      "grad_norm": 1.7653450965881348,
      "learning_rate": 3.951325252953221e-06,
      "loss": 0.5296,
      "step": 144840
    },
    {
      "epoch": 59.29,
      "grad_norm": 1.5315841436386108,
      "learning_rate": 3.951209815030025e-06,
      "loss": 0.5222,
      "step": 144850
    },
    {
      "epoch": 59.3,
      "grad_norm": 1.8044562339782715,
      "learning_rate": 3.951094369743305e-06,
      "loss": 0.5376,
      "step": 144860
    },
    {
      "epoch": 59.3,
      "grad_norm": 1.8851652145385742,
      "learning_rate": 3.95097891709359e-06,
      "loss": 0.5269,
      "step": 144870
    },
    {
      "epoch": 59.3,
      "grad_norm": 2.1117358207702637,
      "learning_rate": 3.950863457081409e-06,
      "loss": 0.5428,
      "step": 144880
    },
    {
      "epoch": 59.31,
      "grad_norm": 1.7998906373977661,
      "learning_rate": 3.950747989707288e-06,
      "loss": 0.5214,
      "step": 144890
    },
    {
      "epoch": 59.31,
      "grad_norm": 2.17352557182312,
      "learning_rate": 3.950632514971759e-06,
      "loss": 0.5254,
      "step": 144900
    },
    {
      "epoch": 59.32,
      "grad_norm": 1.520215392112732,
      "learning_rate": 3.95051703287535e-06,
      "loss": 0.5319,
      "step": 144910
    },
    {
      "epoch": 59.32,
      "grad_norm": 1.985074758529663,
      "learning_rate": 3.950401543418591e-06,
      "loss": 0.5246,
      "step": 144920
    },
    {
      "epoch": 59.32,
      "grad_norm": 3.2087926864624023,
      "learning_rate": 3.950286046602009e-06,
      "loss": 0.5405,
      "step": 144930
    },
    {
      "epoch": 59.33,
      "grad_norm": 1.485747218132019,
      "learning_rate": 3.950170542426134e-06,
      "loss": 0.5064,
      "step": 144940
    },
    {
      "epoch": 59.33,
      "grad_norm": 1.6867204904556274,
      "learning_rate": 3.950055030891496e-06,
      "loss": 0.5163,
      "step": 144950
    },
    {
      "epoch": 59.34,
      "grad_norm": 2.668144941329956,
      "learning_rate": 3.949939511998623e-06,
      "loss": 0.5216,
      "step": 144960
    },
    {
      "epoch": 59.34,
      "grad_norm": 2.1948509216308594,
      "learning_rate": 3.949823985748045e-06,
      "loss": 0.5171,
      "step": 144970
    },
    {
      "epoch": 59.35,
      "grad_norm": 1.9618701934814453,
      "learning_rate": 3.94970845214029e-06,
      "loss": 0.5292,
      "step": 144980
    },
    {
      "epoch": 59.35,
      "grad_norm": 1.4080243110656738,
      "learning_rate": 3.9495929111758894e-06,
      "loss": 0.5004,
      "step": 144990
    },
    {
      "epoch": 59.35,
      "grad_norm": 1.8715784549713135,
      "learning_rate": 3.94947736285537e-06,
      "loss": 0.5335,
      "step": 145000
    },
    {
      "epoch": 59.36,
      "grad_norm": 2.458235502243042,
      "learning_rate": 3.9493618071792625e-06,
      "loss": 0.5326,
      "step": 145010
    },
    {
      "epoch": 59.36,
      "grad_norm": 1.9842692613601685,
      "learning_rate": 3.949246244148097e-06,
      "loss": 0.5138,
      "step": 145020
    },
    {
      "epoch": 59.37,
      "grad_norm": 2.3015799522399902,
      "learning_rate": 3.9491306737624e-06,
      "loss": 0.5436,
      "step": 145030
    },
    {
      "epoch": 59.37,
      "grad_norm": 1.9860080480575562,
      "learning_rate": 3.949015096022703e-06,
      "loss": 0.531,
      "step": 145040
    },
    {
      "epoch": 59.37,
      "grad_norm": 2.1067209243774414,
      "learning_rate": 3.948899510929537e-06,
      "loss": 0.5406,
      "step": 145050
    },
    {
      "epoch": 59.38,
      "grad_norm": 1.6428266763687134,
      "learning_rate": 3.948783918483428e-06,
      "loss": 0.5257,
      "step": 145060
    },
    {
      "epoch": 59.38,
      "grad_norm": 1.6807973384857178,
      "learning_rate": 3.948668318684908e-06,
      "loss": 0.5203,
      "step": 145070
    },
    {
      "epoch": 59.39,
      "grad_norm": 1.6618794202804565,
      "learning_rate": 3.948552711534506e-06,
      "loss": 0.5323,
      "step": 145080
    },
    {
      "epoch": 59.39,
      "grad_norm": 2.3751983642578125,
      "learning_rate": 3.94843709703275e-06,
      "loss": 0.5283,
      "step": 145090
    },
    {
      "epoch": 59.39,
      "grad_norm": 2.1635992527008057,
      "learning_rate": 3.9483214751801716e-06,
      "loss": 0.5347,
      "step": 145100
    },
    {
      "epoch": 59.4,
      "grad_norm": 1.7945879697799683,
      "learning_rate": 3.9482058459773e-06,
      "loss": 0.5213,
      "step": 145110
    },
    {
      "epoch": 59.4,
      "grad_norm": 2.2606403827667236,
      "learning_rate": 3.948090209424665e-06,
      "loss": 0.5264,
      "step": 145120
    },
    {
      "epoch": 59.41,
      "grad_norm": 1.8313671350479126,
      "learning_rate": 3.947974565522795e-06,
      "loss": 0.5398,
      "step": 145130
    },
    {
      "epoch": 59.41,
      "grad_norm": 2.0358729362487793,
      "learning_rate": 3.947858914272222e-06,
      "loss": 0.5557,
      "step": 145140
    },
    {
      "epoch": 59.41,
      "grad_norm": 2.0050065517425537,
      "learning_rate": 3.947743255673473e-06,
      "loss": 0.5504,
      "step": 145150
    },
    {
      "epoch": 59.42,
      "grad_norm": 2.3457095623016357,
      "learning_rate": 3.94762758972708e-06,
      "loss": 0.5414,
      "step": 145160
    },
    {
      "epoch": 59.42,
      "grad_norm": 1.7676211595535278,
      "learning_rate": 3.947511916433572e-06,
      "loss": 0.5316,
      "step": 145170
    },
    {
      "epoch": 59.43,
      "grad_norm": 1.805027723312378,
      "learning_rate": 3.947396235793479e-06,
      "loss": 0.5368,
      "step": 145180
    },
    {
      "epoch": 59.43,
      "grad_norm": 1.888519525527954,
      "learning_rate": 3.947280547807331e-06,
      "loss": 0.5105,
      "step": 145190
    },
    {
      "epoch": 59.44,
      "grad_norm": 2.6626293659210205,
      "learning_rate": 3.947164852475658e-06,
      "loss": 0.5403,
      "step": 145200
    },
    {
      "epoch": 59.44,
      "grad_norm": 2.5487303733825684,
      "learning_rate": 3.94704914979899e-06,
      "loss": 0.531,
      "step": 145210
    },
    {
      "epoch": 59.44,
      "grad_norm": 2.21832013130188,
      "learning_rate": 3.946933439777857e-06,
      "loss": 0.5195,
      "step": 145220
    },
    {
      "epoch": 59.45,
      "grad_norm": 2.2427382469177246,
      "learning_rate": 3.9468177224127886e-06,
      "loss": 0.5124,
      "step": 145230
    },
    {
      "epoch": 59.45,
      "grad_norm": 2.238204002380371,
      "learning_rate": 3.946701997704316e-06,
      "loss": 0.5232,
      "step": 145240
    },
    {
      "epoch": 59.46,
      "grad_norm": 2.2942769527435303,
      "learning_rate": 3.946586265652968e-06,
      "loss": 0.5018,
      "step": 145250
    },
    {
      "epoch": 59.46,
      "grad_norm": 2.2409985065460205,
      "learning_rate": 3.9464705262592745e-06,
      "loss": 0.5167,
      "step": 145260
    },
    {
      "epoch": 59.46,
      "grad_norm": 2.7614855766296387,
      "learning_rate": 3.9463547795237675e-06,
      "loss": 0.5192,
      "step": 145270
    },
    {
      "epoch": 59.47,
      "grad_norm": 2.016165018081665,
      "learning_rate": 3.946239025446976e-06,
      "loss": 0.5259,
      "step": 145280
    },
    {
      "epoch": 59.47,
      "grad_norm": 1.767465353012085,
      "learning_rate": 3.94612326402943e-06,
      "loss": 0.5235,
      "step": 145290
    },
    {
      "epoch": 59.48,
      "grad_norm": 1.8627527952194214,
      "learning_rate": 3.946007495271661e-06,
      "loss": 0.5163,
      "step": 145300
    },
    {
      "epoch": 59.48,
      "grad_norm": 1.7511571645736694,
      "learning_rate": 3.945891719174198e-06,
      "loss": 0.5184,
      "step": 145310
    },
    {
      "epoch": 59.48,
      "grad_norm": 1.655996561050415,
      "learning_rate": 3.945775935737572e-06,
      "loss": 0.5242,
      "step": 145320
    },
    {
      "epoch": 59.49,
      "grad_norm": 2.522472858428955,
      "learning_rate": 3.945660144962314e-06,
      "loss": 0.5084,
      "step": 145330
    },
    {
      "epoch": 59.49,
      "grad_norm": 1.645694613456726,
      "learning_rate": 3.945544346848953e-06,
      "loss": 0.5184,
      "step": 145340
    },
    {
      "epoch": 59.5,
      "grad_norm": 2.147660493850708,
      "learning_rate": 3.9454285413980196e-06,
      "loss": 0.5329,
      "step": 145350
    },
    {
      "epoch": 59.5,
      "grad_norm": 2.1592321395874023,
      "learning_rate": 3.9453127286100456e-06,
      "loss": 0.5272,
      "step": 145360
    },
    {
      "epoch": 59.5,
      "grad_norm": 1.8887406587600708,
      "learning_rate": 3.945196908485561e-06,
      "loss": 0.5243,
      "step": 145370
    },
    {
      "epoch": 59.51,
      "grad_norm": 1.791155219078064,
      "learning_rate": 3.945081081025095e-06,
      "loss": 0.5274,
      "step": 145380
    },
    {
      "epoch": 59.51,
      "grad_norm": 2.385009527206421,
      "learning_rate": 3.94496524622918e-06,
      "loss": 0.5438,
      "step": 145390
    },
    {
      "epoch": 59.52,
      "grad_norm": 1.6563138961791992,
      "learning_rate": 3.9448494040983464e-06,
      "loss": 0.5346,
      "step": 145400
    },
    {
      "epoch": 59.52,
      "grad_norm": 2.091464042663574,
      "learning_rate": 3.944733554633124e-06,
      "loss": 0.5535,
      "step": 145410
    },
    {
      "epoch": 59.53,
      "grad_norm": 2.411006212234497,
      "learning_rate": 3.944617697834043e-06,
      "loss": 0.5076,
      "step": 145420
    },
    {
      "epoch": 59.53,
      "grad_norm": 2.530747413635254,
      "learning_rate": 3.944501833701637e-06,
      "loss": 0.5288,
      "step": 145430
    },
    {
      "epoch": 59.53,
      "grad_norm": 2.0374646186828613,
      "learning_rate": 3.9443859622364336e-06,
      "loss": 0.5468,
      "step": 145440
    },
    {
      "epoch": 59.54,
      "grad_norm": 1.8936132192611694,
      "learning_rate": 3.944270083438964e-06,
      "loss": 0.5383,
      "step": 145450
    },
    {
      "epoch": 59.54,
      "grad_norm": 2.5510451793670654,
      "learning_rate": 3.944154197309762e-06,
      "loss": 0.5305,
      "step": 145460
    },
    {
      "epoch": 59.55,
      "grad_norm": 2.223367214202881,
      "learning_rate": 3.944038303849354e-06,
      "loss": 0.5259,
      "step": 145470
    },
    {
      "epoch": 59.55,
      "grad_norm": 1.7335152626037598,
      "learning_rate": 3.943922403058274e-06,
      "loss": 0.5204,
      "step": 145480
    },
    {
      "epoch": 59.55,
      "grad_norm": 1.7115134000778198,
      "learning_rate": 3.943806494937052e-06,
      "loss": 0.5366,
      "step": 145490
    },
    {
      "epoch": 59.56,
      "grad_norm": 2.2141177654266357,
      "learning_rate": 3.943690579486219e-06,
      "loss": 0.5171,
      "step": 145500
    },
    {
      "epoch": 59.56,
      "grad_norm": 2.100630521774292,
      "learning_rate": 3.943574656706306e-06,
      "loss": 0.5377,
      "step": 145510
    },
    {
      "epoch": 59.57,
      "grad_norm": 1.8765835762023926,
      "learning_rate": 3.9434587265978435e-06,
      "loss": 0.5041,
      "step": 145520
    },
    {
      "epoch": 59.57,
      "grad_norm": 2.0926263332366943,
      "learning_rate": 3.943342789161364e-06,
      "loss": 0.5152,
      "step": 145530
    },
    {
      "epoch": 59.57,
      "grad_norm": 1.1641263961791992,
      "learning_rate": 3.9432268443973975e-06,
      "loss": 0.5092,
      "step": 145540
    },
    {
      "epoch": 59.58,
      "grad_norm": 2.2879180908203125,
      "learning_rate": 3.943110892306475e-06,
      "loss": 0.5306,
      "step": 145550
    },
    {
      "epoch": 59.58,
      "grad_norm": 1.9286651611328125,
      "learning_rate": 3.942994932889127e-06,
      "loss": 0.5194,
      "step": 145560
    },
    {
      "epoch": 59.59,
      "grad_norm": 1.9378994703292847,
      "learning_rate": 3.942878966145886e-06,
      "loss": 0.5319,
      "step": 145570
    },
    {
      "epoch": 59.59,
      "grad_norm": 2.143786668777466,
      "learning_rate": 3.9427629920772835e-06,
      "loss": 0.5029,
      "step": 145580
    },
    {
      "epoch": 59.59,
      "grad_norm": 1.7515292167663574,
      "learning_rate": 3.942647010683851e-06,
      "loss": 0.5187,
      "step": 145590
    },
    {
      "epoch": 59.6,
      "grad_norm": 2.5728206634521484,
      "learning_rate": 3.942531021966118e-06,
      "loss": 0.5082,
      "step": 145600
    },
    {
      "epoch": 59.6,
      "grad_norm": 2.208190679550171,
      "learning_rate": 3.942415025924616e-06,
      "loss": 0.5281,
      "step": 145610
    },
    {
      "epoch": 59.61,
      "grad_norm": 2.514043092727661,
      "learning_rate": 3.942299022559878e-06,
      "loss": 0.5342,
      "step": 145620
    },
    {
      "epoch": 59.61,
      "grad_norm": 2.602191209793091,
      "learning_rate": 3.942183011872434e-06,
      "loss": 0.5145,
      "step": 145630
    },
    {
      "epoch": 59.62,
      "grad_norm": 3.077498435974121,
      "learning_rate": 3.942066993862817e-06,
      "loss": 0.5247,
      "step": 145640
    },
    {
      "epoch": 59.62,
      "grad_norm": 2.6155455112457275,
      "learning_rate": 3.941950968531557e-06,
      "loss": 0.5338,
      "step": 145650
    },
    {
      "epoch": 59.62,
      "grad_norm": 2.1149559020996094,
      "learning_rate": 3.941834935879185e-06,
      "loss": 0.5087,
      "step": 145660
    },
    {
      "epoch": 59.63,
      "grad_norm": 2.0688562393188477,
      "learning_rate": 3.941718895906233e-06,
      "loss": 0.54,
      "step": 145670
    },
    {
      "epoch": 59.63,
      "grad_norm": 2.059370756149292,
      "learning_rate": 3.941602848613235e-06,
      "loss": 0.5476,
      "step": 145680
    },
    {
      "epoch": 59.64,
      "grad_norm": 1.8964766263961792,
      "learning_rate": 3.941486794000719e-06,
      "loss": 0.5027,
      "step": 145690
    },
    {
      "epoch": 59.64,
      "grad_norm": 2.770251512527466,
      "learning_rate": 3.941370732069219e-06,
      "loss": 0.5574,
      "step": 145700
    },
    {
      "epoch": 59.64,
      "grad_norm": 1.6840776205062866,
      "learning_rate": 3.941254662819265e-06,
      "loss": 0.5175,
      "step": 145710
    },
    {
      "epoch": 59.65,
      "grad_norm": 1.2632743120193481,
      "learning_rate": 3.94113858625139e-06,
      "loss": 0.5079,
      "step": 145720
    },
    {
      "epoch": 59.65,
      "grad_norm": 2.0081543922424316,
      "learning_rate": 3.9410225023661255e-06,
      "loss": 0.5354,
      "step": 145730
    },
    {
      "epoch": 59.66,
      "grad_norm": 1.8794777393341064,
      "learning_rate": 3.940906411164003e-06,
      "loss": 0.543,
      "step": 145740
    },
    {
      "epoch": 59.66,
      "grad_norm": 1.8477516174316406,
      "learning_rate": 3.9407903126455535e-06,
      "loss": 0.5353,
      "step": 145750
    },
    {
      "epoch": 59.66,
      "grad_norm": 1.6896291971206665,
      "learning_rate": 3.94067420681131e-06,
      "loss": 0.5219,
      "step": 145760
    },
    {
      "epoch": 59.67,
      "grad_norm": 2.144566059112549,
      "learning_rate": 3.940558093661804e-06,
      "loss": 0.5302,
      "step": 145770
    },
    {
      "epoch": 59.67,
      "grad_norm": 1.9504992961883545,
      "learning_rate": 3.940441973197568e-06,
      "loss": 0.5263,
      "step": 145780
    },
    {
      "epoch": 59.68,
      "grad_norm": 2.2942051887512207,
      "learning_rate": 3.940325845419133e-06,
      "loss": 0.5198,
      "step": 145790
    },
    {
      "epoch": 59.68,
      "grad_norm": 1.948601245880127,
      "learning_rate": 3.940209710327032e-06,
      "loss": 0.5198,
      "step": 145800
    },
    {
      "epoch": 59.68,
      "grad_norm": 2.034637689590454,
      "learning_rate": 3.940093567921796e-06,
      "loss": 0.5369,
      "step": 145810
    },
    {
      "epoch": 59.69,
      "grad_norm": 1.7708200216293335,
      "learning_rate": 3.939977418203957e-06,
      "loss": 0.5029,
      "step": 145820
    },
    {
      "epoch": 59.69,
      "grad_norm": 1.62108314037323,
      "learning_rate": 3.939861261174047e-06,
      "loss": 0.5322,
      "step": 145830
    },
    {
      "epoch": 59.7,
      "grad_norm": 2.031768798828125,
      "learning_rate": 3.939745096832599e-06,
      "loss": 0.5125,
      "step": 145840
    },
    {
      "epoch": 59.7,
      "grad_norm": 2.2900521755218506,
      "learning_rate": 3.939628925180145e-06,
      "loss": 0.5037,
      "step": 145850
    },
    {
      "epoch": 59.71,
      "grad_norm": 1.5696383714675903,
      "learning_rate": 3.939512746217215e-06,
      "loss": 0.5164,
      "step": 145860
    },
    {
      "epoch": 59.71,
      "grad_norm": 2.139594554901123,
      "learning_rate": 3.939396559944345e-06,
      "loss": 0.5388,
      "step": 145870
    },
    {
      "epoch": 59.71,
      "grad_norm": 2.3940069675445557,
      "learning_rate": 3.939280366362065e-06,
      "loss": 0.528,
      "step": 145880
    },
    {
      "epoch": 59.72,
      "grad_norm": 1.6779179573059082,
      "learning_rate": 3.939164165470907e-06,
      "loss": 0.5275,
      "step": 145890
    },
    {
      "epoch": 59.72,
      "grad_norm": 1.8433700799942017,
      "learning_rate": 3.939047957271404e-06,
      "loss": 0.5096,
      "step": 145900
    },
    {
      "epoch": 59.73,
      "grad_norm": 1.5058954954147339,
      "learning_rate": 3.9389317417640885e-06,
      "loss": 0.5343,
      "step": 145910
    },
    {
      "epoch": 59.73,
      "grad_norm": 2.0581493377685547,
      "learning_rate": 3.938815518949491e-06,
      "loss": 0.5207,
      "step": 145920
    },
    {
      "epoch": 59.73,
      "grad_norm": 1.727989912033081,
      "learning_rate": 3.938699288828147e-06,
      "loss": 0.5209,
      "step": 145930
    },
    {
      "epoch": 59.74,
      "grad_norm": 1.7128041982650757,
      "learning_rate": 3.938583051400586e-06,
      "loss": 0.544,
      "step": 145940
    },
    {
      "epoch": 59.74,
      "grad_norm": 1.614738941192627,
      "learning_rate": 3.938466806667342e-06,
      "loss": 0.5516,
      "step": 145950
    },
    {
      "epoch": 59.75,
      "grad_norm": 1.760521411895752,
      "learning_rate": 3.9383505546289485e-06,
      "loss": 0.535,
      "step": 145960
    },
    {
      "epoch": 59.75,
      "grad_norm": 1.9770622253417969,
      "learning_rate": 3.9382342952859366e-06,
      "loss": 0.5385,
      "step": 145970
    },
    {
      "epoch": 59.75,
      "grad_norm": 1.8513051271438599,
      "learning_rate": 3.9381180286388375e-06,
      "loss": 0.5314,
      "step": 145980
    },
    {
      "epoch": 59.76,
      "grad_norm": 1.2801852226257324,
      "learning_rate": 3.938001754688187e-06,
      "loss": 0.5212,
      "step": 145990
    },
    {
      "epoch": 59.76,
      "grad_norm": 2.759018898010254,
      "learning_rate": 3.937885473434515e-06,
      "loss": 0.5337,
      "step": 146000
    },
    {
      "epoch": 59.77,
      "grad_norm": 2.0700788497924805,
      "learning_rate": 3.937769184878356e-06,
      "loss": 0.5427,
      "step": 146010
    },
    {
      "epoch": 59.77,
      "grad_norm": 2.051129102706909,
      "learning_rate": 3.937652889020242e-06,
      "loss": 0.5313,
      "step": 146020
    },
    {
      "epoch": 59.77,
      "grad_norm": 1.7288304567337036,
      "learning_rate": 3.937536585860706e-06,
      "loss": 0.5216,
      "step": 146030
    },
    {
      "epoch": 59.78,
      "grad_norm": 2.011627435684204,
      "learning_rate": 3.937420275400279e-06,
      "loss": 0.5471,
      "step": 146040
    },
    {
      "epoch": 59.78,
      "grad_norm": 1.6672422885894775,
      "learning_rate": 3.9373039576394966e-06,
      "loss": 0.533,
      "step": 146050
    },
    {
      "epoch": 59.79,
      "grad_norm": 1.7438997030258179,
      "learning_rate": 3.93718763257889e-06,
      "loss": 0.532,
      "step": 146060
    },
    {
      "epoch": 59.79,
      "grad_norm": 1.504694938659668,
      "learning_rate": 3.937071300218992e-06,
      "loss": 0.5318,
      "step": 146070
    },
    {
      "epoch": 59.8,
      "grad_norm": 2.012892007827759,
      "learning_rate": 3.936954960560337e-06,
      "loss": 0.528,
      "step": 146080
    },
    {
      "epoch": 59.8,
      "grad_norm": 1.5479754209518433,
      "learning_rate": 3.9368386136034554e-06,
      "loss": 0.5208,
      "step": 146090
    },
    {
      "epoch": 59.8,
      "grad_norm": 2.088191032409668,
      "learning_rate": 3.936722259348882e-06,
      "loss": 0.5188,
      "step": 146100
    },
    {
      "epoch": 59.81,
      "grad_norm": 2.2936410903930664,
      "learning_rate": 3.93660589779715e-06,
      "loss": 0.5618,
      "step": 146110
    },
    {
      "epoch": 59.81,
      "grad_norm": 1.683990478515625,
      "learning_rate": 3.936489528948792e-06,
      "loss": 0.5239,
      "step": 146120
    },
    {
      "epoch": 59.82,
      "grad_norm": 2.1905124187469482,
      "learning_rate": 3.93637315280434e-06,
      "loss": 0.5252,
      "step": 146130
    },
    {
      "epoch": 59.82,
      "grad_norm": 2.2507834434509277,
      "learning_rate": 3.936256769364328e-06,
      "loss": 0.5263,
      "step": 146140
    },
    {
      "epoch": 59.82,
      "grad_norm": 1.6476905345916748,
      "learning_rate": 3.936140378629289e-06,
      "loss": 0.5244,
      "step": 146150
    },
    {
      "epoch": 59.83,
      "grad_norm": 1.8413646221160889,
      "learning_rate": 3.936023980599756e-06,
      "loss": 0.539,
      "step": 146160
    },
    {
      "epoch": 59.83,
      "grad_norm": 1.8637280464172363,
      "learning_rate": 3.935907575276264e-06,
      "loss": 0.5118,
      "step": 146170
    },
    {
      "epoch": 59.84,
      "grad_norm": 2.194049835205078,
      "learning_rate": 3.935791162659344e-06,
      "loss": 0.534,
      "step": 146180
    },
    {
      "epoch": 59.84,
      "grad_norm": 1.7583624124526978,
      "learning_rate": 3.93567474274953e-06,
      "loss": 0.5241,
      "step": 146190
    },
    {
      "epoch": 59.84,
      "grad_norm": 2.092137336730957,
      "learning_rate": 3.935558315547355e-06,
      "loss": 0.5314,
      "step": 146200
    },
    {
      "epoch": 59.85,
      "grad_norm": 1.5780525207519531,
      "learning_rate": 3.935441881053353e-06,
      "loss": 0.5298,
      "step": 146210
    },
    {
      "epoch": 59.85,
      "grad_norm": 2.148286819458008,
      "learning_rate": 3.935325439268057e-06,
      "loss": 0.5344,
      "step": 146220
    },
    {
      "epoch": 59.86,
      "grad_norm": 1.8498613834381104,
      "learning_rate": 3.935208990192e-06,
      "loss": 0.5294,
      "step": 146230
    },
    {
      "epoch": 59.86,
      "grad_norm": 2.124082565307617,
      "learning_rate": 3.935092533825715e-06,
      "loss": 0.5269,
      "step": 146240
    },
    {
      "epoch": 59.86,
      "grad_norm": 2.060939311981201,
      "learning_rate": 3.934976070169738e-06,
      "loss": 0.5512,
      "step": 146250
    },
    {
      "epoch": 59.87,
      "grad_norm": 2.125941514968872,
      "learning_rate": 3.934859599224601e-06,
      "loss": 0.5286,
      "step": 146260
    },
    {
      "epoch": 59.87,
      "grad_norm": 2.330625534057617,
      "learning_rate": 3.934743120990837e-06,
      "loss": 0.5208,
      "step": 146270
    },
    {
      "epoch": 59.88,
      "grad_norm": 1.8975894451141357,
      "learning_rate": 3.934626635468979e-06,
      "loss": 0.5381,
      "step": 146280
    },
    {
      "epoch": 59.88,
      "grad_norm": 2.671685218811035,
      "learning_rate": 3.934510142659562e-06,
      "loss": 0.5535,
      "step": 146290
    },
    {
      "epoch": 59.89,
      "grad_norm": 1.6919411420822144,
      "learning_rate": 3.934393642563119e-06,
      "loss": 0.5031,
      "step": 146300
    },
    {
      "epoch": 59.89,
      "grad_norm": 2.042910099029541,
      "learning_rate": 3.934277135180184e-06,
      "loss": 0.5221,
      "step": 146310
    },
    {
      "epoch": 59.89,
      "grad_norm": 2.295673131942749,
      "learning_rate": 3.93416062051129e-06,
      "loss": 0.5342,
      "step": 146320
    },
    {
      "epoch": 59.9,
      "grad_norm": 1.9701184034347534,
      "learning_rate": 3.934044098556972e-06,
      "loss": 0.5242,
      "step": 146330
    },
    {
      "epoch": 59.9,
      "grad_norm": 2.17779541015625,
      "learning_rate": 3.933927569317762e-06,
      "loss": 0.5353,
      "step": 146340
    },
    {
      "epoch": 59.91,
      "grad_norm": 1.9445070028305054,
      "learning_rate": 3.933811032794196e-06,
      "loss": 0.5386,
      "step": 146350
    },
    {
      "epoch": 59.91,
      "grad_norm": 2.369988441467285,
      "learning_rate": 3.933694488986806e-06,
      "loss": 0.5289,
      "step": 146360
    },
    {
      "epoch": 59.91,
      "grad_norm": 1.816972255706787,
      "learning_rate": 3.933577937896126e-06,
      "loss": 0.5098,
      "step": 146370
    },
    {
      "epoch": 59.92,
      "grad_norm": 1.699519157409668,
      "learning_rate": 3.933461379522692e-06,
      "loss": 0.5288,
      "step": 146380
    },
    {
      "epoch": 59.92,
      "grad_norm": 1.6743016242980957,
      "learning_rate": 3.933344813867035e-06,
      "loss": 0.5142,
      "step": 146390
    },
    {
      "epoch": 59.93,
      "grad_norm": 1.614898920059204,
      "learning_rate": 3.933228240929691e-06,
      "loss": 0.5335,
      "step": 146400
    },
    {
      "epoch": 59.93,
      "grad_norm": 1.7262076139450073,
      "learning_rate": 3.9331116607111935e-06,
      "loss": 0.516,
      "step": 146410
    },
    {
      "epoch": 59.93,
      "grad_norm": 1.9658387899398804,
      "learning_rate": 3.9329950732120756e-06,
      "loss": 0.5269,
      "step": 146420
    },
    {
      "epoch": 59.94,
      "grad_norm": 1.8298795223236084,
      "learning_rate": 3.932878478432873e-06,
      "loss": 0.5194,
      "step": 146430
    },
    {
      "epoch": 59.94,
      "grad_norm": 1.3034955263137817,
      "learning_rate": 3.932761876374118e-06,
      "loss": 0.5016,
      "step": 146440
    },
    {
      "epoch": 59.95,
      "grad_norm": 1.9778622388839722,
      "learning_rate": 3.9326452670363464e-06,
      "loss": 0.5148,
      "step": 146450
    },
    {
      "epoch": 59.95,
      "grad_norm": 2.2311999797821045,
      "learning_rate": 3.9325286504200915e-06,
      "loss": 0.5389,
      "step": 146460
    },
    {
      "epoch": 59.95,
      "grad_norm": 2.004326105117798,
      "learning_rate": 3.932412026525888e-06,
      "loss": 0.5295,
      "step": 146470
    },
    {
      "epoch": 59.96,
      "grad_norm": 2.11438250541687,
      "learning_rate": 3.932295395354269e-06,
      "loss": 0.5275,
      "step": 146480
    },
    {
      "epoch": 59.96,
      "grad_norm": 2.1737828254699707,
      "learning_rate": 3.93217875690577e-06,
      "loss": 0.5145,
      "step": 146490
    },
    {
      "epoch": 59.97,
      "grad_norm": 2.0812594890594482,
      "learning_rate": 3.932062111180925e-06,
      "loss": 0.5186,
      "step": 146500
    },
    {
      "epoch": 59.97,
      "grad_norm": 2.1401760578155518,
      "learning_rate": 3.931945458180268e-06,
      "loss": 0.5446,
      "step": 146510
    },
    {
      "epoch": 59.98,
      "grad_norm": 2.7695696353912354,
      "learning_rate": 3.931828797904333e-06,
      "loss": 0.5215,
      "step": 146520
    },
    {
      "epoch": 59.98,
      "grad_norm": 2.1241631507873535,
      "learning_rate": 3.931712130353656e-06,
      "loss": 0.5249,
      "step": 146530
    },
    {
      "epoch": 59.98,
      "grad_norm": 1.9486967325210571,
      "learning_rate": 3.93159545552877e-06,
      "loss": 0.5095,
      "step": 146540
    },
    {
      "epoch": 59.99,
      "grad_norm": 1.5422002077102661,
      "learning_rate": 3.93147877343021e-06,
      "loss": 0.5254,
      "step": 146550
    },
    {
      "epoch": 59.99,
      "grad_norm": 1.6606467962265015,
      "learning_rate": 3.93136208405851e-06,
      "loss": 0.5101,
      "step": 146560
    },
    {
      "epoch": 60.0,
      "grad_norm": 2.111661195755005,
      "learning_rate": 3.931245387414206e-06,
      "loss": 0.5028,
      "step": 146570
    },
    {
      "epoch": 60.0,
      "grad_norm": 2.104800224304199,
      "learning_rate": 3.93112868349783e-06,
      "loss": 0.5187,
      "step": 146580
    },
    {
      "epoch": 60.0,
      "eval_loss": 0.5259493589401245,
      "eval_runtime": 52.2285,
      "eval_samples_per_second": 66.037,
      "eval_steps_per_second": 8.271,
      "step": 146580
    },
    {
      "epoch": 60.0,
      "grad_norm": 1.9624266624450684,
      "learning_rate": 3.93101197230992e-06,
      "loss": 0.5474,
      "step": 146590
    },
    {
      "epoch": 60.01,
      "grad_norm": 1.924638271331787,
      "learning_rate": 3.930895253851007e-06,
      "loss": 0.5351,
      "step": 146600
    },
    {
      "epoch": 60.01,
      "grad_norm": 2.308030128479004,
      "learning_rate": 3.930778528121627e-06,
      "loss": 0.5246,
      "step": 146610
    },
    {
      "epoch": 60.02,
      "grad_norm": 2.561755895614624,
      "learning_rate": 3.9306617951223166e-06,
      "loss": 0.5243,
      "step": 146620
    },
    {
      "epoch": 60.02,
      "grad_norm": 2.144747734069824,
      "learning_rate": 3.9305450548536085e-06,
      "loss": 0.5265,
      "step": 146630
    },
    {
      "epoch": 60.02,
      "grad_norm": 1.8236632347106934,
      "learning_rate": 3.930428307316037e-06,
      "loss": 0.5334,
      "step": 146640
    },
    {
      "epoch": 60.03,
      "grad_norm": 2.0845346450805664,
      "learning_rate": 3.930311552510139e-06,
      "loss": 0.5371,
      "step": 146650
    },
    {
      "epoch": 60.03,
      "grad_norm": 1.783719539642334,
      "learning_rate": 3.930194790436448e-06,
      "loss": 0.5252,
      "step": 146660
    },
    {
      "epoch": 60.04,
      "grad_norm": 1.831024169921875,
      "learning_rate": 3.9300780210955e-06,
      "loss": 0.5214,
      "step": 146670
    },
    {
      "epoch": 60.04,
      "grad_norm": 2.2327044010162354,
      "learning_rate": 3.9299612444878275e-06,
      "loss": 0.5152,
      "step": 146680
    },
    {
      "epoch": 60.05,
      "grad_norm": 1.7434097528457642,
      "learning_rate": 3.9298444606139676e-06,
      "loss": 0.5328,
      "step": 146690
    },
    {
      "epoch": 60.05,
      "grad_norm": 2.4430251121520996,
      "learning_rate": 3.929727669474455e-06,
      "loss": 0.5289,
      "step": 146700
    },
    {
      "epoch": 60.05,
      "grad_norm": 2.140800714492798,
      "learning_rate": 3.929610871069823e-06,
      "loss": 0.5047,
      "step": 146710
    },
    {
      "epoch": 60.06,
      "grad_norm": 1.7099665403366089,
      "learning_rate": 3.9294940654006095e-06,
      "loss": 0.528,
      "step": 146720
    },
    {
      "epoch": 60.06,
      "grad_norm": 2.381749153137207,
      "learning_rate": 3.929377252467347e-06,
      "loss": 0.5288,
      "step": 146730
    },
    {
      "epoch": 60.07,
      "grad_norm": 2.487415313720703,
      "learning_rate": 3.929260432270572e-06,
      "loss": 0.5478,
      "step": 146740
    },
    {
      "epoch": 60.07,
      "grad_norm": 1.8153539896011353,
      "learning_rate": 3.929143604810819e-06,
      "loss": 0.5361,
      "step": 146750
    },
    {
      "epoch": 60.07,
      "grad_norm": 2.037903308868408,
      "learning_rate": 3.9290267700886245e-06,
      "loss": 0.5157,
      "step": 146760
    },
    {
      "epoch": 60.08,
      "grad_norm": 2.197158098220825,
      "learning_rate": 3.9289099281045215e-06,
      "loss": 0.5169,
      "step": 146770
    },
    {
      "epoch": 60.08,
      "grad_norm": 1.926330327987671,
      "learning_rate": 3.928793078859047e-06,
      "loss": 0.5466,
      "step": 146780
    },
    {
      "epoch": 60.09,
      "grad_norm": 1.6316425800323486,
      "learning_rate": 3.9286762223527355e-06,
      "loss": 0.5129,
      "step": 146790
    },
    {
      "epoch": 60.09,
      "grad_norm": 1.9414902925491333,
      "learning_rate": 3.9285593585861215e-06,
      "loss": 0.5183,
      "step": 146800
    },
    {
      "epoch": 60.09,
      "grad_norm": 2.488680839538574,
      "learning_rate": 3.928442487559742e-06,
      "loss": 0.5294,
      "step": 146810
    },
    {
      "epoch": 60.1,
      "grad_norm": 3.5961198806762695,
      "learning_rate": 3.928325609274132e-06,
      "loss": 0.5216,
      "step": 146820
    },
    {
      "epoch": 60.1,
      "grad_norm": 1.9588319063186646,
      "learning_rate": 3.928208723729826e-06,
      "loss": 0.5276,
      "step": 146830
    },
    {
      "epoch": 60.11,
      "grad_norm": 2.311995267868042,
      "learning_rate": 3.92809183092736e-06,
      "loss": 0.5215,
      "step": 146840
    },
    {
      "epoch": 60.11,
      "grad_norm": 2.830687999725342,
      "learning_rate": 3.92797493086727e-06,
      "loss": 0.516,
      "step": 146850
    },
    {
      "epoch": 60.11,
      "grad_norm": 2.6164400577545166,
      "learning_rate": 3.92785802355009e-06,
      "loss": 0.5264,
      "step": 146860
    },
    {
      "epoch": 60.12,
      "grad_norm": 1.966853380203247,
      "learning_rate": 3.927741108976357e-06,
      "loss": 0.5201,
      "step": 146870
    },
    {
      "epoch": 60.12,
      "grad_norm": 1.6993621587753296,
      "learning_rate": 3.927624187146607e-06,
      "loss": 0.5296,
      "step": 146880
    },
    {
      "epoch": 60.13,
      "grad_norm": 1.5007429122924805,
      "learning_rate": 3.9275072580613726e-06,
      "loss": 0.5397,
      "step": 146890
    },
    {
      "epoch": 60.13,
      "grad_norm": 1.780963659286499,
      "learning_rate": 3.927390321721193e-06,
      "loss": 0.5091,
      "step": 146900
    },
    {
      "epoch": 60.14,
      "grad_norm": 2.1265692710876465,
      "learning_rate": 3.927273378126602e-06,
      "loss": 0.5364,
      "step": 146910
    },
    {
      "epoch": 60.14,
      "grad_norm": 2.2339413166046143,
      "learning_rate": 3.927156427278136e-06,
      "loss": 0.5368,
      "step": 146920
    },
    {
      "epoch": 60.14,
      "grad_norm": 2.3380415439605713,
      "learning_rate": 3.92703946917633e-06,
      "loss": 0.5376,
      "step": 146930
    },
    {
      "epoch": 60.15,
      "grad_norm": 2.73292875289917,
      "learning_rate": 3.92692250382172e-06,
      "loss": 0.5394,
      "step": 146940
    },
    {
      "epoch": 60.15,
      "grad_norm": 1.9091094732284546,
      "learning_rate": 3.926805531214842e-06,
      "loss": 0.502,
      "step": 146950
    },
    {
      "epoch": 60.16,
      "grad_norm": 1.8724603652954102,
      "learning_rate": 3.926688551356232e-06,
      "loss": 0.5339,
      "step": 146960
    },
    {
      "epoch": 60.16,
      "grad_norm": 2.069401741027832,
      "learning_rate": 3.926571564246425e-06,
      "loss": 0.521,
      "step": 146970
    },
    {
      "epoch": 60.16,
      "grad_norm": 2.1580405235290527,
      "learning_rate": 3.926454569885959e-06,
      "loss": 0.5221,
      "step": 146980
    },
    {
      "epoch": 60.17,
      "grad_norm": 2.1326282024383545,
      "learning_rate": 3.926337568275367e-06,
      "loss": 0.5241,
      "step": 146990
    },
    {
      "epoch": 60.17,
      "grad_norm": 2.1742355823516846,
      "learning_rate": 3.926220559415186e-06,
      "loss": 0.5353,
      "step": 147000
    },
    {
      "epoch": 60.18,
      "grad_norm": 1.8202630281448364,
      "learning_rate": 3.926103543305954e-06,
      "loss": 0.5155,
      "step": 147010
    },
    {
      "epoch": 60.18,
      "grad_norm": 1.4533802270889282,
      "learning_rate": 3.925986519948205e-06,
      "loss": 0.5225,
      "step": 147020
    },
    {
      "epoch": 60.18,
      "grad_norm": 1.6583033800125122,
      "learning_rate": 3.925869489342475e-06,
      "loss": 0.5208,
      "step": 147030
    },
    {
      "epoch": 60.19,
      "grad_norm": 1.86183500289917,
      "learning_rate": 3.9257524514893015e-06,
      "loss": 0.5218,
      "step": 147040
    },
    {
      "epoch": 60.19,
      "grad_norm": 1.7671043872833252,
      "learning_rate": 3.9256354063892185e-06,
      "loss": 0.5163,
      "step": 147050
    },
    {
      "epoch": 60.2,
      "grad_norm": 2.3672432899475098,
      "learning_rate": 3.925518354042764e-06,
      "loss": 0.5265,
      "step": 147060
    },
    {
      "epoch": 60.2,
      "grad_norm": 2.8823330402374268,
      "learning_rate": 3.925401294450475e-06,
      "loss": 0.5108,
      "step": 147070
    },
    {
      "epoch": 60.2,
      "grad_norm": 2.07491135597229,
      "learning_rate": 3.925284227612884e-06,
      "loss": 0.5202,
      "step": 147080
    },
    {
      "epoch": 60.21,
      "grad_norm": 3.2676827907562256,
      "learning_rate": 3.925167153530532e-06,
      "loss": 0.5393,
      "step": 147090
    },
    {
      "epoch": 60.21,
      "grad_norm": 2.1601343154907227,
      "learning_rate": 3.925050072203951e-06,
      "loss": 0.5031,
      "step": 147100
    },
    {
      "epoch": 60.22,
      "grad_norm": 2.064645290374756,
      "learning_rate": 3.924932983633681e-06,
      "loss": 0.5162,
      "step": 147110
    },
    {
      "epoch": 60.22,
      "grad_norm": 2.3297345638275146,
      "learning_rate": 3.9248158878202555e-06,
      "loss": 0.5156,
      "step": 147120
    },
    {
      "epoch": 60.23,
      "grad_norm": 2.069058895111084,
      "learning_rate": 3.924698784764212e-06,
      "loss": 0.53,
      "step": 147130
    },
    {
      "epoch": 60.23,
      "grad_norm": 1.8926788568496704,
      "learning_rate": 3.924581674466087e-06,
      "loss": 0.5328,
      "step": 147140
    },
    {
      "epoch": 60.23,
      "grad_norm": 1.4059243202209473,
      "learning_rate": 3.924464556926418e-06,
      "loss": 0.5114,
      "step": 147150
    },
    {
      "epoch": 60.24,
      "grad_norm": 1.96304190158844,
      "learning_rate": 3.924347432145739e-06,
      "loss": 0.5231,
      "step": 147160
    },
    {
      "epoch": 60.24,
      "grad_norm": 1.7230898141860962,
      "learning_rate": 3.924230300124588e-06,
      "loss": 0.5224,
      "step": 147170
    },
    {
      "epoch": 60.25,
      "grad_norm": 2.7845513820648193,
      "learning_rate": 3.924113160863502e-06,
      "loss": 0.5283,
      "step": 147180
    },
    {
      "epoch": 60.25,
      "grad_norm": 2.1235923767089844,
      "learning_rate": 3.923996014363017e-06,
      "loss": 0.5099,
      "step": 147190
    },
    {
      "epoch": 60.25,
      "grad_norm": 2.1845474243164062,
      "learning_rate": 3.92387886062367e-06,
      "loss": 0.545,
      "step": 147200
    },
    {
      "epoch": 60.26,
      "grad_norm": 1.747135043144226,
      "learning_rate": 3.923761699645997e-06,
      "loss": 0.5112,
      "step": 147210
    },
    {
      "epoch": 60.26,
      "grad_norm": 2.252396583557129,
      "learning_rate": 3.923644531430536e-06,
      "loss": 0.5261,
      "step": 147220
    },
    {
      "epoch": 60.27,
      "grad_norm": 1.587704062461853,
      "learning_rate": 3.923527355977822e-06,
      "loss": 0.5137,
      "step": 147230
    },
    {
      "epoch": 60.27,
      "grad_norm": 1.9994832277297974,
      "learning_rate": 3.923410173288393e-06,
      "loss": 0.5099,
      "step": 147240
    },
    {
      "epoch": 60.27,
      "grad_norm": 1.5359052419662476,
      "learning_rate": 3.923292983362785e-06,
      "loss": 0.5238,
      "step": 147250
    },
    {
      "epoch": 60.28,
      "grad_norm": 2.1973283290863037,
      "learning_rate": 3.923175786201535e-06,
      "loss": 0.519,
      "step": 147260
    },
    {
      "epoch": 60.28,
      "grad_norm": 1.4605460166931152,
      "learning_rate": 3.92305858180518e-06,
      "loss": 0.5285,
      "step": 147270
    },
    {
      "epoch": 60.29,
      "grad_norm": 2.124847412109375,
      "learning_rate": 3.922941370174258e-06,
      "loss": 0.5214,
      "step": 147280
    },
    {
      "epoch": 60.29,
      "grad_norm": 2.660280704498291,
      "learning_rate": 3.922824151309304e-06,
      "loss": 0.537,
      "step": 147290
    },
    {
      "epoch": 60.29,
      "grad_norm": 2.210808038711548,
      "learning_rate": 3.922706925210856e-06,
      "loss": 0.5548,
      "step": 147300
    },
    {
      "epoch": 60.3,
      "grad_norm": 2.2197554111480713,
      "learning_rate": 3.922589691879452e-06,
      "loss": 0.5292,
      "step": 147310
    },
    {
      "epoch": 60.3,
      "grad_norm": 2.0112335681915283,
      "learning_rate": 3.922472451315626e-06,
      "loss": 0.5193,
      "step": 147320
    },
    {
      "epoch": 60.31,
      "grad_norm": 2.454773426055908,
      "learning_rate": 3.922355203519918e-06,
      "loss": 0.5069,
      "step": 147330
    },
    {
      "epoch": 60.31,
      "grad_norm": 1.8027523756027222,
      "learning_rate": 3.922237948492864e-06,
      "loss": 0.5333,
      "step": 147340
    },
    {
      "epoch": 60.32,
      "grad_norm": 1.9449490308761597,
      "learning_rate": 3.922120686235002e-06,
      "loss": 0.5505,
      "step": 147350
    },
    {
      "epoch": 60.32,
      "grad_norm": 2.4741907119750977,
      "learning_rate": 3.922003416746867e-06,
      "loss": 0.5309,
      "step": 147360
    },
    {
      "epoch": 60.32,
      "grad_norm": 1.787335991859436,
      "learning_rate": 3.921886140028998e-06,
      "loss": 0.5249,
      "step": 147370
    },
    {
      "epoch": 60.33,
      "grad_norm": 2.446528911590576,
      "learning_rate": 3.921768856081931e-06,
      "loss": 0.5419,
      "step": 147380
    },
    {
      "epoch": 60.33,
      "grad_norm": 2.2279629707336426,
      "learning_rate": 3.921651564906205e-06,
      "loss": 0.5291,
      "step": 147390
    },
    {
      "epoch": 60.34,
      "grad_norm": 2.3665101528167725,
      "learning_rate": 3.921534266502356e-06,
      "loss": 0.5152,
      "step": 147400
    },
    {
      "epoch": 60.34,
      "grad_norm": 1.6508928537368774,
      "learning_rate": 3.921416960870922e-06,
      "loss": 0.5259,
      "step": 147410
    },
    {
      "epoch": 60.34,
      "grad_norm": 1.5905380249023438,
      "learning_rate": 3.921299648012439e-06,
      "loss": 0.5215,
      "step": 147420
    },
    {
      "epoch": 60.35,
      "grad_norm": 1.842990517616272,
      "learning_rate": 3.921182327927447e-06,
      "loss": 0.5284,
      "step": 147430
    },
    {
      "epoch": 60.35,
      "grad_norm": 2.4347405433654785,
      "learning_rate": 3.9210650006164804e-06,
      "loss": 0.5232,
      "step": 147440
    },
    {
      "epoch": 60.36,
      "grad_norm": 1.403730034828186,
      "learning_rate": 3.920947666080078e-06,
      "loss": 0.5319,
      "step": 147450
    },
    {
      "epoch": 60.36,
      "grad_norm": 1.5640461444854736,
      "learning_rate": 3.9208303243187776e-06,
      "loss": 0.5193,
      "step": 147460
    },
    {
      "epoch": 60.36,
      "grad_norm": 1.9187341928482056,
      "learning_rate": 3.920712975333117e-06,
      "loss": 0.5282,
      "step": 147470
    },
    {
      "epoch": 60.37,
      "grad_norm": 1.8832422494888306,
      "learning_rate": 3.920595619123633e-06,
      "loss": 0.5017,
      "step": 147480
    },
    {
      "epoch": 60.37,
      "grad_norm": 2.2777791023254395,
      "learning_rate": 3.920478255690863e-06,
      "loss": 0.4945,
      "step": 147490
    },
    {
      "epoch": 60.38,
      "grad_norm": 2.1726627349853516,
      "learning_rate": 3.920360885035346e-06,
      "loss": 0.5295,
      "step": 147500
    },
    {
      "epoch": 60.38,
      "grad_norm": 2.132610321044922,
      "learning_rate": 3.920243507157618e-06,
      "loss": 0.5264,
      "step": 147510
    },
    {
      "epoch": 60.38,
      "grad_norm": 1.6271083354949951,
      "learning_rate": 3.920126122058217e-06,
      "loss": 0.5226,
      "step": 147520
    },
    {
      "epoch": 60.39,
      "grad_norm": 1.7168937921524048,
      "learning_rate": 3.920008729737681e-06,
      "loss": 0.5209,
      "step": 147530
    },
    {
      "epoch": 60.39,
      "grad_norm": 2.314058542251587,
      "learning_rate": 3.919891330196549e-06,
      "loss": 0.5528,
      "step": 147540
    },
    {
      "epoch": 60.4,
      "grad_norm": 1.9545884132385254,
      "learning_rate": 3.919773923435357e-06,
      "loss": 0.505,
      "step": 147550
    },
    {
      "epoch": 60.4,
      "grad_norm": 1.9648298025131226,
      "learning_rate": 3.919656509454644e-06,
      "loss": 0.5157,
      "step": 147560
    },
    {
      "epoch": 60.41,
      "grad_norm": 1.7151566743850708,
      "learning_rate": 3.919539088254947e-06,
      "loss": 0.5185,
      "step": 147570
    },
    {
      "epoch": 60.41,
      "grad_norm": 2.084446668624878,
      "learning_rate": 3.919421659836803e-06,
      "loss": 0.5102,
      "step": 147580
    },
    {
      "epoch": 60.41,
      "grad_norm": 1.759864330291748,
      "learning_rate": 3.919304224200753e-06,
      "loss": 0.5089,
      "step": 147590
    },
    {
      "epoch": 60.42,
      "grad_norm": 1.6352488994598389,
      "learning_rate": 3.9191867813473325e-06,
      "loss": 0.5588,
      "step": 147600
    },
    {
      "epoch": 60.42,
      "grad_norm": 2.477160930633545,
      "learning_rate": 3.919069331277081e-06,
      "loss": 0.5194,
      "step": 147610
    },
    {
      "epoch": 60.43,
      "grad_norm": 1.7655742168426514,
      "learning_rate": 3.918951873990534e-06,
      "loss": 0.5276,
      "step": 147620
    },
    {
      "epoch": 60.43,
      "grad_norm": 1.9704993963241577,
      "learning_rate": 3.918834409488232e-06,
      "loss": 0.5233,
      "step": 147630
    },
    {
      "epoch": 60.43,
      "grad_norm": 1.8844190835952759,
      "learning_rate": 3.918716937770711e-06,
      "loss": 0.5318,
      "step": 147640
    },
    {
      "epoch": 60.44,
      "grad_norm": 2.294440984725952,
      "learning_rate": 3.918599458838512e-06,
      "loss": 0.5201,
      "step": 147650
    },
    {
      "epoch": 60.44,
      "grad_norm": 1.678584337234497,
      "learning_rate": 3.918481972692171e-06,
      "loss": 0.5067,
      "step": 147660
    },
    {
      "epoch": 60.45,
      "grad_norm": 1.533780813217163,
      "learning_rate": 3.918364479332227e-06,
      "loss": 0.5191,
      "step": 147670
    },
    {
      "epoch": 60.45,
      "grad_norm": 1.8060848712921143,
      "learning_rate": 3.918246978759218e-06,
      "loss": 0.5355,
      "step": 147680
    },
    {
      "epoch": 60.45,
      "grad_norm": 1.9604344367980957,
      "learning_rate": 3.918129470973681e-06,
      "loss": 0.5342,
      "step": 147690
    },
    {
      "epoch": 60.46,
      "grad_norm": 2.539306163787842,
      "learning_rate": 3.918011955976156e-06,
      "loss": 0.5106,
      "step": 147700
    },
    {
      "epoch": 60.46,
      "grad_norm": 1.9796165227890015,
      "learning_rate": 3.917894433767181e-06,
      "loss": 0.5306,
      "step": 147710
    },
    {
      "epoch": 60.47,
      "grad_norm": 2.365973949432373,
      "learning_rate": 3.917776904347295e-06,
      "loss": 0.5259,
      "step": 147720
    },
    {
      "epoch": 60.47,
      "grad_norm": 1.8050364255905151,
      "learning_rate": 3.917659367717035e-06,
      "loss": 0.5143,
      "step": 147730
    },
    {
      "epoch": 60.47,
      "grad_norm": 2.2083828449249268,
      "learning_rate": 3.917541823876939e-06,
      "loss": 0.5479,
      "step": 147740
    },
    {
      "epoch": 60.48,
      "grad_norm": 2.143416404724121,
      "learning_rate": 3.917424272827547e-06,
      "loss": 0.5347,
      "step": 147750
    },
    {
      "epoch": 60.48,
      "grad_norm": 2.5549685955047607,
      "learning_rate": 3.917306714569397e-06,
      "loss": 0.5281,
      "step": 147760
    },
    {
      "epoch": 60.49,
      "grad_norm": 1.9960992336273193,
      "learning_rate": 3.917189149103027e-06,
      "loss": 0.5157,
      "step": 147770
    },
    {
      "epoch": 60.49,
      "grad_norm": 1.8900936841964722,
      "learning_rate": 3.917071576428977e-06,
      "loss": 0.5443,
      "step": 147780
    },
    {
      "epoch": 60.5,
      "grad_norm": 2.5906152725219727,
      "learning_rate": 3.9169539965477825e-06,
      "loss": 0.5192,
      "step": 147790
    },
    {
      "epoch": 60.5,
      "grad_norm": 2.3723976612091064,
      "learning_rate": 3.916836409459985e-06,
      "loss": 0.5463,
      "step": 147800
    },
    {
      "epoch": 60.5,
      "grad_norm": 2.2251956462860107,
      "learning_rate": 3.916718815166123e-06,
      "loss": 0.5318,
      "step": 147810
    },
    {
      "epoch": 60.51,
      "grad_norm": 1.7864595651626587,
      "learning_rate": 3.916601213666734e-06,
      "loss": 0.538,
      "step": 147820
    },
    {
      "epoch": 60.51,
      "grad_norm": 1.699102759361267,
      "learning_rate": 3.916483604962358e-06,
      "loss": 0.5113,
      "step": 147830
    },
    {
      "epoch": 60.52,
      "grad_norm": 3.0931966304779053,
      "learning_rate": 3.916365989053532e-06,
      "loss": 0.4997,
      "step": 147840
    },
    {
      "epoch": 60.52,
      "grad_norm": 1.8793466091156006,
      "learning_rate": 3.916248365940796e-06,
      "loss": 0.523,
      "step": 147850
    },
    {
      "epoch": 60.52,
      "grad_norm": 1.8172178268432617,
      "learning_rate": 3.916130735624688e-06,
      "loss": 0.5264,
      "step": 147860
    },
    {
      "epoch": 60.53,
      "grad_norm": 2.003999710083008,
      "learning_rate": 3.916013098105748e-06,
      "loss": 0.5331,
      "step": 147870
    },
    {
      "epoch": 60.53,
      "grad_norm": 1.8415734767913818,
      "learning_rate": 3.915895453384514e-06,
      "loss": 0.5141,
      "step": 147880
    },
    {
      "epoch": 60.54,
      "grad_norm": 2.0555672645568848,
      "learning_rate": 3.9157778014615255e-06,
      "loss": 0.5407,
      "step": 147890
    },
    {
      "epoch": 60.54,
      "grad_norm": 1.6719735860824585,
      "learning_rate": 3.915660142337321e-06,
      "loss": 0.5277,
      "step": 147900
    },
    {
      "epoch": 60.54,
      "grad_norm": 1.7983061075210571,
      "learning_rate": 3.915542476012439e-06,
      "loss": 0.5199,
      "step": 147910
    },
    {
      "epoch": 60.55,
      "grad_norm": 1.9401086568832397,
      "learning_rate": 3.915424802487419e-06,
      "loss": 0.5486,
      "step": 147920
    },
    {
      "epoch": 60.55,
      "grad_norm": 1.6224911212921143,
      "learning_rate": 3.915307121762801e-06,
      "loss": 0.5238,
      "step": 147930
    },
    {
      "epoch": 60.56,
      "grad_norm": 2.33786940574646,
      "learning_rate": 3.915189433839123e-06,
      "loss": 0.5129,
      "step": 147940
    },
    {
      "epoch": 60.56,
      "grad_norm": 2.5312881469726562,
      "learning_rate": 3.915071738716923e-06,
      "loss": 0.5316,
      "step": 147950
    },
    {
      "epoch": 60.56,
      "grad_norm": 2.645758628845215,
      "learning_rate": 3.914954036396744e-06,
      "loss": 0.5428,
      "step": 147960
    },
    {
      "epoch": 60.57,
      "grad_norm": 2.2195541858673096,
      "learning_rate": 3.91483632687912e-06,
      "loss": 0.5388,
      "step": 147970
    },
    {
      "epoch": 60.57,
      "grad_norm": 1.5009630918502808,
      "learning_rate": 3.914718610164595e-06,
      "loss": 0.5211,
      "step": 147980
    },
    {
      "epoch": 60.58,
      "grad_norm": 1.7501661777496338,
      "learning_rate": 3.914600886253705e-06,
      "loss": 0.55,
      "step": 147990
    },
    {
      "epoch": 60.58,
      "grad_norm": 2.4271247386932373,
      "learning_rate": 3.91448315514699e-06,
      "loss": 0.5149,
      "step": 148000
    },
    {
      "epoch": 60.59,
      "grad_norm": 2.0956568717956543,
      "learning_rate": 3.91436541684499e-06,
      "loss": 0.5139,
      "step": 148010
    },
    {
      "epoch": 60.59,
      "grad_norm": 2.4719443321228027,
      "learning_rate": 3.914247671348245e-06,
      "loss": 0.5376,
      "step": 148020
    },
    {
      "epoch": 60.59,
      "grad_norm": 2.0987155437469482,
      "learning_rate": 3.914129918657292e-06,
      "loss": 0.5452,
      "step": 148030
    },
    {
      "epoch": 60.6,
      "grad_norm": 2.308415412902832,
      "learning_rate": 3.914012158772672e-06,
      "loss": 0.532,
      "step": 148040
    },
    {
      "epoch": 60.6,
      "grad_norm": 2.1914355754852295,
      "learning_rate": 3.913894391694925e-06,
      "loss": 0.5284,
      "step": 148050
    },
    {
      "epoch": 60.61,
      "grad_norm": 1.9321107864379883,
      "learning_rate": 3.9137766174245896e-06,
      "loss": 0.5318,
      "step": 148060
    },
    {
      "epoch": 60.61,
      "grad_norm": 1.78590989112854,
      "learning_rate": 3.913658835962205e-06,
      "loss": 0.537,
      "step": 148070
    },
    {
      "epoch": 60.61,
      "grad_norm": 2.227842330932617,
      "learning_rate": 3.913541047308312e-06,
      "loss": 0.5025,
      "step": 148080
    },
    {
      "epoch": 60.62,
      "grad_norm": 1.918877124786377,
      "learning_rate": 3.9134232514634476e-06,
      "loss": 0.5286,
      "step": 148090
    },
    {
      "epoch": 60.62,
      "grad_norm": 1.983197808265686,
      "learning_rate": 3.9133054484281546e-06,
      "loss": 0.5339,
      "step": 148100
    },
    {
      "epoch": 60.63,
      "grad_norm": 2.1933279037475586,
      "learning_rate": 3.913187638202971e-06,
      "loss": 0.4926,
      "step": 148110
    },
    {
      "epoch": 60.63,
      "grad_norm": 1.6473252773284912,
      "learning_rate": 3.913069820788436e-06,
      "loss": 0.524,
      "step": 148120
    },
    {
      "epoch": 60.63,
      "grad_norm": 1.96395742893219,
      "learning_rate": 3.912951996185091e-06,
      "loss": 0.5222,
      "step": 148130
    },
    {
      "epoch": 60.64,
      "grad_norm": 2.6004533767700195,
      "learning_rate": 3.912834164393474e-06,
      "loss": 0.5426,
      "step": 148140
    },
    {
      "epoch": 60.64,
      "grad_norm": 2.9170379638671875,
      "learning_rate": 3.912716325414125e-06,
      "loss": 0.5253,
      "step": 148150
    },
    {
      "epoch": 60.65,
      "grad_norm": 2.399832248687744,
      "learning_rate": 3.912598479247584e-06,
      "loss": 0.5309,
      "step": 148160
    },
    {
      "epoch": 60.65,
      "grad_norm": 2.089594602584839,
      "learning_rate": 3.912480625894392e-06,
      "loss": 0.5066,
      "step": 148170
    },
    {
      "epoch": 60.65,
      "grad_norm": 2.227804183959961,
      "learning_rate": 3.9123627653550884e-06,
      "loss": 0.5279,
      "step": 148180
    },
    {
      "epoch": 60.66,
      "grad_norm": 1.625158429145813,
      "learning_rate": 3.912244897630212e-06,
      "loss": 0.5269,
      "step": 148190
    },
    {
      "epoch": 60.66,
      "grad_norm": 2.0818850994110107,
      "learning_rate": 3.9121270227203034e-06,
      "loss": 0.5011,
      "step": 148200
    },
    {
      "epoch": 60.67,
      "grad_norm": 2.3196983337402344,
      "learning_rate": 3.912009140625903e-06,
      "loss": 0.541,
      "step": 148210
    },
    {
      "epoch": 60.67,
      "grad_norm": 1.3852612972259521,
      "learning_rate": 3.911891251347549e-06,
      "loss": 0.5202,
      "step": 148220
    },
    {
      "epoch": 60.68,
      "grad_norm": 1.6041873693466187,
      "learning_rate": 3.911773354885784e-06,
      "loss": 0.5326,
      "step": 148230
    },
    {
      "epoch": 60.68,
      "grad_norm": 1.6357883214950562,
      "learning_rate": 3.9116554512411465e-06,
      "loss": 0.5516,
      "step": 148240
    },
    {
      "epoch": 60.68,
      "grad_norm": 2.3832242488861084,
      "learning_rate": 3.911537540414176e-06,
      "loss": 0.5197,
      "step": 148250
    },
    {
      "epoch": 60.69,
      "grad_norm": 2.3546502590179443,
      "learning_rate": 3.911419622405415e-06,
      "loss": 0.5369,
      "step": 148260
    },
    {
      "epoch": 60.69,
      "grad_norm": 1.9531115293502808,
      "learning_rate": 3.911301697215402e-06,
      "loss": 0.5392,
      "step": 148270
    },
    {
      "epoch": 60.7,
      "grad_norm": 2.0469746589660645,
      "learning_rate": 3.911183764844677e-06,
      "loss": 0.5002,
      "step": 148280
    },
    {
      "epoch": 60.7,
      "grad_norm": 1.9275726079940796,
      "learning_rate": 3.911065825293781e-06,
      "loss": 0.556,
      "step": 148290
    },
    {
      "epoch": 60.7,
      "grad_norm": 1.7910842895507812,
      "learning_rate": 3.9109478785632536e-06,
      "loss": 0.4963,
      "step": 148300
    },
    {
      "epoch": 60.71,
      "grad_norm": 2.034682512283325,
      "learning_rate": 3.9108299246536354e-06,
      "loss": 0.5325,
      "step": 148310
    },
    {
      "epoch": 60.71,
      "grad_norm": 2.1975574493408203,
      "learning_rate": 3.910711963565467e-06,
      "loss": 0.5157,
      "step": 148320
    },
    {
      "epoch": 60.72,
      "grad_norm": 1.769086480140686,
      "learning_rate": 3.910593995299289e-06,
      "loss": 0.5458,
      "step": 148330
    },
    {
      "epoch": 60.72,
      "grad_norm": 1.9976410865783691,
      "learning_rate": 3.910476019855641e-06,
      "loss": 0.5366,
      "step": 148340
    },
    {
      "epoch": 60.72,
      "grad_norm": 1.2584025859832764,
      "learning_rate": 3.9103580372350635e-06,
      "loss": 0.5373,
      "step": 148350
    },
    {
      "epoch": 60.73,
      "grad_norm": 1.8714625835418701,
      "learning_rate": 3.910240047438098e-06,
      "loss": 0.5271,
      "step": 148360
    },
    {
      "epoch": 60.73,
      "grad_norm": 1.9051318168640137,
      "learning_rate": 3.910122050465283e-06,
      "loss": 0.5088,
      "step": 148370
    },
    {
      "epoch": 60.74,
      "grad_norm": 2.3942477703094482,
      "learning_rate": 3.910004046317161e-06,
      "loss": 0.5305,
      "step": 148380
    },
    {
      "epoch": 60.74,
      "grad_norm": 2.15460205078125,
      "learning_rate": 3.909886034994271e-06,
      "loss": 0.5101,
      "step": 148390
    },
    {
      "epoch": 60.74,
      "grad_norm": 1.6250513792037964,
      "learning_rate": 3.909768016497155e-06,
      "loss": 0.523,
      "step": 148400
    },
    {
      "epoch": 60.75,
      "grad_norm": 2.209716558456421,
      "learning_rate": 3.909649990826353e-06,
      "loss": 0.5198,
      "step": 148410
    },
    {
      "epoch": 60.75,
      "grad_norm": 1.8178528547286987,
      "learning_rate": 3.909531957982405e-06,
      "loss": 0.5107,
      "step": 148420
    },
    {
      "epoch": 60.76,
      "grad_norm": 1.8713730573654175,
      "learning_rate": 3.909413917965853e-06,
      "loss": 0.5326,
      "step": 148430
    },
    {
      "epoch": 60.76,
      "grad_norm": 2.2073307037353516,
      "learning_rate": 3.909295870777237e-06,
      "loss": 0.5411,
      "step": 148440
    },
    {
      "epoch": 60.77,
      "grad_norm": 1.789268970489502,
      "learning_rate": 3.9091778164170974e-06,
      "loss": 0.5218,
      "step": 148450
    },
    {
      "epoch": 60.77,
      "grad_norm": 1.7285833358764648,
      "learning_rate": 3.909059754885976e-06,
      "loss": 0.5129,
      "step": 148460
    },
    {
      "epoch": 60.77,
      "grad_norm": 1.9744305610656738,
      "learning_rate": 3.908941686184413e-06,
      "loss": 0.5167,
      "step": 148470
    },
    {
      "epoch": 60.78,
      "grad_norm": 1.9662052392959595,
      "learning_rate": 3.9088236103129495e-06,
      "loss": 0.5295,
      "step": 148480
    },
    {
      "epoch": 60.78,
      "grad_norm": 1.9198089838027954,
      "learning_rate": 3.908705527272124e-06,
      "loss": 0.5304,
      "step": 148490
    },
    {
      "epoch": 60.79,
      "grad_norm": 1.8641101121902466,
      "learning_rate": 3.9085874370624824e-06,
      "loss": 0.5067,
      "step": 148500
    },
    {
      "epoch": 60.79,
      "grad_norm": 1.7147514820098877,
      "learning_rate": 3.908469339684562e-06,
      "loss": 0.5291,
      "step": 148510
    },
    {
      "epoch": 60.79,
      "grad_norm": 2.2141013145446777,
      "learning_rate": 3.908351235138904e-06,
      "loss": 0.5161,
      "step": 148520
    },
    {
      "epoch": 60.8,
      "grad_norm": 2.195711851119995,
      "learning_rate": 3.90823312342605e-06,
      "loss": 0.5285,
      "step": 148530
    },
    {
      "epoch": 60.8,
      "grad_norm": 2.2256953716278076,
      "learning_rate": 3.908115004546543e-06,
      "loss": 0.5444,
      "step": 148540
    },
    {
      "epoch": 60.81,
      "grad_norm": 2.020106554031372,
      "learning_rate": 3.90799687850092e-06,
      "loss": 0.5384,
      "step": 148550
    },
    {
      "epoch": 60.81,
      "grad_norm": 1.8431802988052368,
      "learning_rate": 3.907878745289725e-06,
      "loss": 0.5255,
      "step": 148560
    },
    {
      "epoch": 60.81,
      "grad_norm": 3.140125274658203,
      "learning_rate": 3.907760604913498e-06,
      "loss": 0.5292,
      "step": 148570
    },
    {
      "epoch": 60.82,
      "grad_norm": 2.2899081707000732,
      "learning_rate": 3.9076424573727816e-06,
      "loss": 0.547,
      "step": 148580
    },
    {
      "epoch": 60.82,
      "grad_norm": 2.254031181335449,
      "learning_rate": 3.907524302668116e-06,
      "loss": 0.5322,
      "step": 148590
    },
    {
      "epoch": 60.83,
      "grad_norm": 1.7303667068481445,
      "learning_rate": 3.907406140800041e-06,
      "loss": 0.517,
      "step": 148600
    },
    {
      "epoch": 60.83,
      "grad_norm": 1.8214185237884521,
      "learning_rate": 3.907287971769101e-06,
      "loss": 0.5179,
      "step": 148610
    },
    {
      "epoch": 60.84,
      "grad_norm": 1.7043308019638062,
      "learning_rate": 3.907169795575835e-06,
      "loss": 0.5107,
      "step": 148620
    },
    {
      "epoch": 60.84,
      "grad_norm": 1.9565465450286865,
      "learning_rate": 3.9070516122207856e-06,
      "loss": 0.5468,
      "step": 148630
    },
    {
      "epoch": 60.84,
      "grad_norm": 2.2754740715026855,
      "learning_rate": 3.906933421704493e-06,
      "loss": 0.5168,
      "step": 148640
    },
    {
      "epoch": 60.85,
      "grad_norm": 2.0157854557037354,
      "learning_rate": 3.9068152240274995e-06,
      "loss": 0.516,
      "step": 148650
    },
    {
      "epoch": 60.85,
      "grad_norm": 2.3274776935577393,
      "learning_rate": 3.906697019190347e-06,
      "loss": 0.5218,
      "step": 148660
    },
    {
      "epoch": 60.86,
      "grad_norm": 1.7202814817428589,
      "learning_rate": 3.906578807193575e-06,
      "loss": 0.5516,
      "step": 148670
    },
    {
      "epoch": 60.86,
      "grad_norm": 2.151352882385254,
      "learning_rate": 3.906460588037726e-06,
      "loss": 0.5207,
      "step": 148680
    },
    {
      "epoch": 60.86,
      "grad_norm": 2.054600238800049,
      "learning_rate": 3.9063423617233424e-06,
      "loss": 0.5058,
      "step": 148690
    },
    {
      "epoch": 60.87,
      "grad_norm": 2.9440419673919678,
      "learning_rate": 3.906224128250966e-06,
      "loss": 0.5383,
      "step": 148700
    },
    {
      "epoch": 60.87,
      "grad_norm": 1.9456760883331299,
      "learning_rate": 3.9061058876211365e-06,
      "loss": 0.5355,
      "step": 148710
    },
    {
      "epoch": 60.88,
      "grad_norm": 2.267853260040283,
      "learning_rate": 3.905987639834397e-06,
      "loss": 0.5412,
      "step": 148720
    },
    {
      "epoch": 60.88,
      "grad_norm": 1.7875173091888428,
      "learning_rate": 3.90586938489129e-06,
      "loss": 0.5349,
      "step": 148730
    },
    {
      "epoch": 60.88,
      "grad_norm": 2.0590832233428955,
      "learning_rate": 3.905751122792354e-06,
      "loss": 0.5229,
      "step": 148740
    },
    {
      "epoch": 60.89,
      "grad_norm": 1.9650698900222778,
      "learning_rate": 3.9056328535381336e-06,
      "loss": 0.5208,
      "step": 148750
    },
    {
      "epoch": 60.89,
      "grad_norm": 2.2495763301849365,
      "learning_rate": 3.905514577129171e-06,
      "loss": 0.5224,
      "step": 148760
    },
    {
      "epoch": 60.9,
      "grad_norm": 1.9499313831329346,
      "learning_rate": 3.905396293566005e-06,
      "loss": 0.5327,
      "step": 148770
    },
    {
      "epoch": 60.9,
      "grad_norm": 1.8597673177719116,
      "learning_rate": 3.9052780028491795e-06,
      "loss": 0.5211,
      "step": 148780
    },
    {
      "epoch": 60.9,
      "grad_norm": 2.055572271347046,
      "learning_rate": 3.905159704979237e-06,
      "loss": 0.5439,
      "step": 148790
    },
    {
      "epoch": 60.91,
      "grad_norm": 1.9284230470657349,
      "learning_rate": 3.905041399956718e-06,
      "loss": 0.5152,
      "step": 148800
    },
    {
      "epoch": 60.91,
      "grad_norm": 3.2171638011932373,
      "learning_rate": 3.904923087782164e-06,
      "loss": 0.5077,
      "step": 148810
    },
    {
      "epoch": 60.92,
      "grad_norm": 1.8568235635757446,
      "learning_rate": 3.904804768456118e-06,
      "loss": 0.5156,
      "step": 148820
    },
    {
      "epoch": 60.92,
      "grad_norm": 1.9155155420303345,
      "learning_rate": 3.904686441979123e-06,
      "loss": 0.5192,
      "step": 148830
    },
    {
      "epoch": 60.93,
      "grad_norm": 1.8392459154129028,
      "learning_rate": 3.904568108351719e-06,
      "loss": 0.53,
      "step": 148840
    },
    {
      "epoch": 60.93,
      "grad_norm": 2.5904300212860107,
      "learning_rate": 3.904449767574449e-06,
      "loss": 0.5375,
      "step": 148850
    },
    {
      "epoch": 60.93,
      "grad_norm": 1.9533882141113281,
      "learning_rate": 3.9043314196478555e-06,
      "loss": 0.4997,
      "step": 148860
    },
    {
      "epoch": 60.94,
      "grad_norm": 1.6799346208572388,
      "learning_rate": 3.9042130645724805e-06,
      "loss": 0.5253,
      "step": 148870
    },
    {
      "epoch": 60.94,
      "grad_norm": 2.254737377166748,
      "learning_rate": 3.904094702348865e-06,
      "loss": 0.5096,
      "step": 148880
    },
    {
      "epoch": 60.95,
      "grad_norm": 1.9623664617538452,
      "learning_rate": 3.903976332977552e-06,
      "loss": 0.5342,
      "step": 148890
    },
    {
      "epoch": 60.95,
      "grad_norm": 1.9933924674987793,
      "learning_rate": 3.903857956459085e-06,
      "loss": 0.5153,
      "step": 148900
    },
    {
      "epoch": 60.95,
      "grad_norm": 1.6673473119735718,
      "learning_rate": 3.903739572794004e-06,
      "loss": 0.5301,
      "step": 148910
    },
    {
      "epoch": 60.96,
      "grad_norm": 2.2381956577301025,
      "learning_rate": 3.903621181982853e-06,
      "loss": 0.5235,
      "step": 148920
    },
    {
      "epoch": 60.96,
      "grad_norm": 2.3949239253997803,
      "learning_rate": 3.903502784026173e-06,
      "loss": 0.5015,
      "step": 148930
    },
    {
      "epoch": 60.97,
      "grad_norm": 1.6225765943527222,
      "learning_rate": 3.903384378924508e-06,
      "loss": 0.5217,
      "step": 148940
    },
    {
      "epoch": 60.97,
      "grad_norm": 1.6904921531677246,
      "learning_rate": 3.903265966678399e-06,
      "loss": 0.5213,
      "step": 148950
    },
    {
      "epoch": 60.97,
      "grad_norm": 1.7444099187850952,
      "learning_rate": 3.903147547288388e-06,
      "loss": 0.5436,
      "step": 148960
    },
    {
      "epoch": 60.98,
      "grad_norm": 2.2845160961151123,
      "learning_rate": 3.90302912075502e-06,
      "loss": 0.5257,
      "step": 148970
    },
    {
      "epoch": 60.98,
      "grad_norm": 1.4771589040756226,
      "learning_rate": 3.902910687078834e-06,
      "loss": 0.519,
      "step": 148980
    },
    {
      "epoch": 60.99,
      "grad_norm": 1.8090368509292603,
      "learning_rate": 3.9027922462603756e-06,
      "loss": 0.5221,
      "step": 148990
    },
    {
      "epoch": 60.99,
      "grad_norm": 2.1290719509124756,
      "learning_rate": 3.902673798300186e-06,
      "loss": 0.5469,
      "step": 149000
    },
    {
      "epoch": 60.99,
      "grad_norm": 1.8217201232910156,
      "learning_rate": 3.902555343198808e-06,
      "loss": 0.5224,
      "step": 149010
    },
    {
      "epoch": 61.0,
      "grad_norm": 1.7975667715072632,
      "learning_rate": 3.902436880956784e-06,
      "loss": 0.5272,
      "step": 149020
    },
    {
      "epoch": 61.0,
      "eval_loss": 0.5254462361335754,
      "eval_runtime": 52.1415,
      "eval_samples_per_second": 66.147,
      "eval_steps_per_second": 8.285,
      "step": 149023
    },
    {
      "epoch": 61.0,
      "grad_norm": 1.9354302883148193,
      "learning_rate": 3.902318411574657e-06,
      "loss": 0.522,
      "step": 149030
    },
    {
      "epoch": 61.01,
      "grad_norm": 1.8914636373519897,
      "learning_rate": 3.902199935052968e-06,
      "loss": 0.5243,
      "step": 149040
    },
    {
      "epoch": 61.01,
      "grad_norm": 2.001845359802246,
      "learning_rate": 3.902081451392263e-06,
      "loss": 0.5321,
      "step": 149050
    },
    {
      "epoch": 61.02,
      "grad_norm": 1.9023045301437378,
      "learning_rate": 3.901962960593082e-06,
      "loss": 0.5335,
      "step": 149060
    },
    {
      "epoch": 61.02,
      "grad_norm": 1.6744990348815918,
      "learning_rate": 3.90184446265597e-06,
      "loss": 0.5087,
      "step": 149070
    },
    {
      "epoch": 61.02,
      "grad_norm": 1.7969200611114502,
      "learning_rate": 3.901725957581467e-06,
      "loss": 0.4985,
      "step": 149080
    },
    {
      "epoch": 61.03,
      "grad_norm": 1.5852986574172974,
      "learning_rate": 3.901607445370118e-06,
      "loss": 0.5297,
      "step": 149090
    },
    {
      "epoch": 61.03,
      "grad_norm": 1.810746669769287,
      "learning_rate": 3.901488926022465e-06,
      "loss": 0.5263,
      "step": 149100
    },
    {
      "epoch": 61.04,
      "grad_norm": 1.6492966413497925,
      "learning_rate": 3.901370399539051e-06,
      "loss": 0.5356,
      "step": 149110
    },
    {
      "epoch": 61.04,
      "grad_norm": 1.6105139255523682,
      "learning_rate": 3.90125186592042e-06,
      "loss": 0.5246,
      "step": 149120
    },
    {
      "epoch": 61.04,
      "grad_norm": 1.4314250946044922,
      "learning_rate": 3.901133325167114e-06,
      "loss": 0.522,
      "step": 149130
    },
    {
      "epoch": 61.05,
      "grad_norm": 1.6904234886169434,
      "learning_rate": 3.901014777279676e-06,
      "loss": 0.5109,
      "step": 149140
    },
    {
      "epoch": 61.05,
      "grad_norm": 2.4007480144500732,
      "learning_rate": 3.90089622225865e-06,
      "loss": 0.5336,
      "step": 149150
    },
    {
      "epoch": 61.06,
      "grad_norm": 1.9902594089508057,
      "learning_rate": 3.900777660104577e-06,
      "loss": 0.5156,
      "step": 149160
    },
    {
      "epoch": 61.06,
      "grad_norm": 2.0292913913726807,
      "learning_rate": 3.900659090818002e-06,
      "loss": 0.5442,
      "step": 149170
    },
    {
      "epoch": 61.06,
      "grad_norm": 1.6091729402542114,
      "learning_rate": 3.900540514399468e-06,
      "loss": 0.5275,
      "step": 149180
    },
    {
      "epoch": 61.07,
      "grad_norm": 2.226715087890625,
      "learning_rate": 3.900421930849517e-06,
      "loss": 0.5248,
      "step": 149190
    },
    {
      "epoch": 61.07,
      "grad_norm": 2.205320119857788,
      "learning_rate": 3.900303340168694e-06,
      "loss": 0.5266,
      "step": 149200
    },
    {
      "epoch": 61.08,
      "grad_norm": 1.8638442754745483,
      "learning_rate": 3.90018474235754e-06,
      "loss": 0.5356,
      "step": 149210
    },
    {
      "epoch": 61.08,
      "grad_norm": 1.5351519584655762,
      "learning_rate": 3.900066137416601e-06,
      "loss": 0.5166,
      "step": 149220
    },
    {
      "epoch": 61.08,
      "grad_norm": 1.8093605041503906,
      "learning_rate": 3.899947525346417e-06,
      "loss": 0.5325,
      "step": 149230
    },
    {
      "epoch": 61.09,
      "grad_norm": 2.0532543659210205,
      "learning_rate": 3.8998289061475345e-06,
      "loss": 0.5365,
      "step": 149240
    },
    {
      "epoch": 61.09,
      "grad_norm": 2.247298240661621,
      "learning_rate": 3.899710279820494e-06,
      "loss": 0.5143,
      "step": 149250
    },
    {
      "epoch": 61.1,
      "grad_norm": 2.1404213905334473,
      "learning_rate": 3.899591646365842e-06,
      "loss": 0.5152,
      "step": 149260
    },
    {
      "epoch": 61.1,
      "grad_norm": 1.9986776113510132,
      "learning_rate": 3.8994730057841195e-06,
      "loss": 0.5413,
      "step": 149270
    },
    {
      "epoch": 61.11,
      "grad_norm": 1.918906807899475,
      "learning_rate": 3.899354358075871e-06,
      "loss": 0.5249,
      "step": 149280
    },
    {
      "epoch": 61.11,
      "grad_norm": 2.984485149383545,
      "learning_rate": 3.899235703241641e-06,
      "loss": 0.527,
      "step": 149290
    },
    {
      "epoch": 61.11,
      "grad_norm": 2.1739213466644287,
      "learning_rate": 3.8991170412819704e-06,
      "loss": 0.5084,
      "step": 149300
    },
    {
      "epoch": 61.12,
      "grad_norm": 2.193294048309326,
      "learning_rate": 3.898998372197405e-06,
      "loss": 0.5095,
      "step": 149310
    },
    {
      "epoch": 61.12,
      "grad_norm": 1.4679383039474487,
      "learning_rate": 3.898879695988487e-06,
      "loss": 0.5336,
      "step": 149320
    },
    {
      "epoch": 61.13,
      "grad_norm": 2.0574162006378174,
      "learning_rate": 3.898761012655761e-06,
      "loss": 0.5337,
      "step": 149330
    },
    {
      "epoch": 61.13,
      "grad_norm": 2.0945639610290527,
      "learning_rate": 3.8986423221997704e-06,
      "loss": 0.524,
      "step": 149340
    },
    {
      "epoch": 61.13,
      "grad_norm": 2.458066701889038,
      "learning_rate": 3.898523624621059e-06,
      "loss": 0.5252,
      "step": 149350
    },
    {
      "epoch": 61.14,
      "grad_norm": 2.360100746154785,
      "learning_rate": 3.89840491992017e-06,
      "loss": 0.5063,
      "step": 149360
    },
    {
      "epoch": 61.14,
      "grad_norm": 2.581796884536743,
      "learning_rate": 3.898286208097649e-06,
      "loss": 0.5092,
      "step": 149370
    },
    {
      "epoch": 61.15,
      "grad_norm": 1.6511516571044922,
      "learning_rate": 3.898167489154037e-06,
      "loss": 0.5358,
      "step": 149380
    },
    {
      "epoch": 61.15,
      "grad_norm": 2.055567502975464,
      "learning_rate": 3.89804876308988e-06,
      "loss": 0.5413,
      "step": 149390
    },
    {
      "epoch": 61.15,
      "grad_norm": 1.528664469718933,
      "learning_rate": 3.8979300299057204e-06,
      "loss": 0.5216,
      "step": 149400
    },
    {
      "epoch": 61.16,
      "grad_norm": 1.672919511795044,
      "learning_rate": 3.897811289602103e-06,
      "loss": 0.5407,
      "step": 149410
    },
    {
      "epoch": 61.16,
      "grad_norm": 2.771254539489746,
      "learning_rate": 3.897692542179572e-06,
      "loss": 0.5217,
      "step": 149420
    },
    {
      "epoch": 61.17,
      "grad_norm": 1.3645808696746826,
      "learning_rate": 3.89757378763867e-06,
      "loss": 0.5218,
      "step": 149430
    },
    {
      "epoch": 61.17,
      "grad_norm": 1.9973028898239136,
      "learning_rate": 3.897455025979942e-06,
      "loss": 0.5211,
      "step": 149440
    },
    {
      "epoch": 61.17,
      "grad_norm": 1.7294325828552246,
      "learning_rate": 3.897336257203932e-06,
      "loss": 0.5233,
      "step": 149450
    },
    {
      "epoch": 61.18,
      "grad_norm": 1.866542100906372,
      "learning_rate": 3.897217481311185e-06,
      "loss": 0.5252,
      "step": 149460
    },
    {
      "epoch": 61.18,
      "grad_norm": 2.030226230621338,
      "learning_rate": 3.897098698302243e-06,
      "loss": 0.5589,
      "step": 149470
    },
    {
      "epoch": 61.19,
      "grad_norm": 2.0322349071502686,
      "learning_rate": 3.896979908177651e-06,
      "loss": 0.539,
      "step": 149480
    },
    {
      "epoch": 61.19,
      "grad_norm": 1.8524900674819946,
      "learning_rate": 3.896861110937954e-06,
      "loss": 0.515,
      "step": 149490
    },
    {
      "epoch": 61.2,
      "grad_norm": 1.5293279886245728,
      "learning_rate": 3.896742306583696e-06,
      "loss": 0.5107,
      "step": 149500
    },
    {
      "epoch": 61.2,
      "grad_norm": 2.015462636947632,
      "learning_rate": 3.8966234951154196e-06,
      "loss": 0.5337,
      "step": 149510
    },
    {
      "epoch": 61.2,
      "grad_norm": 2.060805320739746,
      "learning_rate": 3.89650467653367e-06,
      "loss": 0.516,
      "step": 149520
    },
    {
      "epoch": 61.21,
      "grad_norm": 2.61027193069458,
      "learning_rate": 3.896385850838992e-06,
      "loss": 0.5383,
      "step": 149530
    },
    {
      "epoch": 61.21,
      "grad_norm": 1.9353723526000977,
      "learning_rate": 3.89626701803193e-06,
      "loss": 0.5235,
      "step": 149540
    },
    {
      "epoch": 61.22,
      "grad_norm": 2.5129146575927734,
      "learning_rate": 3.896148178113029e-06,
      "loss": 0.5318,
      "step": 149550
    },
    {
      "epoch": 61.22,
      "grad_norm": 2.2719008922576904,
      "learning_rate": 3.89602933108283e-06,
      "loss": 0.5248,
      "step": 149560
    },
    {
      "epoch": 61.22,
      "grad_norm": 2.127509117126465,
      "learning_rate": 3.895910476941881e-06,
      "loss": 0.534,
      "step": 149570
    },
    {
      "epoch": 61.23,
      "grad_norm": 2.5056586265563965,
      "learning_rate": 3.895791615690726e-06,
      "loss": 0.5373,
      "step": 149580
    },
    {
      "epoch": 61.23,
      "grad_norm": 1.6723711490631104,
      "learning_rate": 3.895672747329908e-06,
      "loss": 0.5348,
      "step": 149590
    },
    {
      "epoch": 61.24,
      "grad_norm": 1.6758863925933838,
      "learning_rate": 3.895553871859972e-06,
      "loss": 0.5204,
      "step": 149600
    },
    {
      "epoch": 61.24,
      "grad_norm": 1.6513692140579224,
      "learning_rate": 3.895434989281463e-06,
      "loss": 0.5142,
      "step": 149610
    },
    {
      "epoch": 61.24,
      "grad_norm": 2.2724740505218506,
      "learning_rate": 3.895316099594926e-06,
      "loss": 0.5316,
      "step": 149620
    },
    {
      "epoch": 61.25,
      "grad_norm": 2.7179787158966064,
      "learning_rate": 3.895197202800904e-06,
      "loss": 0.5481,
      "step": 149630
    },
    {
      "epoch": 61.25,
      "grad_norm": 1.58810293674469,
      "learning_rate": 3.895078298899943e-06,
      "loss": 0.5313,
      "step": 149640
    },
    {
      "epoch": 61.26,
      "grad_norm": 1.8906463384628296,
      "learning_rate": 3.894959387892587e-06,
      "loss": 0.5194,
      "step": 149650
    },
    {
      "epoch": 61.26,
      "grad_norm": 2.0896129608154297,
      "learning_rate": 3.894840469779382e-06,
      "loss": 0.5175,
      "step": 149660
    },
    {
      "epoch": 61.26,
      "grad_norm": 2.190880060195923,
      "learning_rate": 3.894721544560871e-06,
      "loss": 0.5194,
      "step": 149670
    },
    {
      "epoch": 61.27,
      "grad_norm": 1.4158987998962402,
      "learning_rate": 3.894602612237599e-06,
      "loss": 0.5279,
      "step": 149680
    },
    {
      "epoch": 61.27,
      "grad_norm": 1.6969763040542603,
      "learning_rate": 3.894483672810113e-06,
      "loss": 0.5348,
      "step": 149690
    },
    {
      "epoch": 61.28,
      "grad_norm": 1.6249113082885742,
      "learning_rate": 3.894364726278955e-06,
      "loss": 0.5142,
      "step": 149700
    },
    {
      "epoch": 61.28,
      "grad_norm": 2.491525888442993,
      "learning_rate": 3.894245772644671e-06,
      "loss": 0.5377,
      "step": 149710
    },
    {
      "epoch": 61.29,
      "grad_norm": 2.290942668914795,
      "learning_rate": 3.894126811907806e-06,
      "loss": 0.5146,
      "step": 149720
    },
    {
      "epoch": 61.29,
      "grad_norm": 1.74600088596344,
      "learning_rate": 3.8940078440689065e-06,
      "loss": 0.5118,
      "step": 149730
    },
    {
      "epoch": 61.29,
      "grad_norm": 2.7547504901885986,
      "learning_rate": 3.893888869128514e-06,
      "loss": 0.5307,
      "step": 149740
    },
    {
      "epoch": 61.3,
      "grad_norm": 1.9821728467941284,
      "learning_rate": 3.893769887087176e-06,
      "loss": 0.5236,
      "step": 149750
    },
    {
      "epoch": 61.3,
      "grad_norm": 1.7301459312438965,
      "learning_rate": 3.893650897945437e-06,
      "loss": 0.5324,
      "step": 149760
    },
    {
      "epoch": 61.31,
      "grad_norm": 2.4103641510009766,
      "learning_rate": 3.8935319017038425e-06,
      "loss": 0.5214,
      "step": 149770
    },
    {
      "epoch": 61.31,
      "grad_norm": 2.6497561931610107,
      "learning_rate": 3.8934128983629365e-06,
      "loss": 0.5288,
      "step": 149780
    },
    {
      "epoch": 61.31,
      "grad_norm": 2.163553237915039,
      "learning_rate": 3.893293887923265e-06,
      "loss": 0.53,
      "step": 149790
    },
    {
      "epoch": 61.32,
      "grad_norm": 1.9536266326904297,
      "learning_rate": 3.893174870385373e-06,
      "loss": 0.5168,
      "step": 149800
    },
    {
      "epoch": 61.32,
      "grad_norm": 1.7325743436813354,
      "learning_rate": 3.893055845749805e-06,
      "loss": 0.5417,
      "step": 149810
    },
    {
      "epoch": 61.33,
      "grad_norm": 1.9185357093811035,
      "learning_rate": 3.892936814017107e-06,
      "loss": 0.5227,
      "step": 149820
    },
    {
      "epoch": 61.33,
      "grad_norm": 2.1412620544433594,
      "learning_rate": 3.8928177751878245e-06,
      "loss": 0.5135,
      "step": 149830
    },
    {
      "epoch": 61.33,
      "grad_norm": 2.1234357357025146,
      "learning_rate": 3.892698729262503e-06,
      "loss": 0.5394,
      "step": 149840
    },
    {
      "epoch": 61.34,
      "grad_norm": 2.3013460636138916,
      "learning_rate": 3.892579676241687e-06,
      "loss": 0.5202,
      "step": 149850
    },
    {
      "epoch": 61.34,
      "grad_norm": 2.321150779724121,
      "learning_rate": 3.8924606161259215e-06,
      "loss": 0.512,
      "step": 149860
    },
    {
      "epoch": 61.35,
      "grad_norm": 1.583156704902649,
      "learning_rate": 3.892341548915752e-06,
      "loss": 0.5015,
      "step": 149870
    },
    {
      "epoch": 61.35,
      "grad_norm": 2.8111789226531982,
      "learning_rate": 3.892222474611726e-06,
      "loss": 0.5131,
      "step": 149880
    },
    {
      "epoch": 61.35,
      "grad_norm": 1.9289350509643555,
      "learning_rate": 3.892103393214387e-06,
      "loss": 0.536,
      "step": 149890
    },
    {
      "epoch": 61.36,
      "grad_norm": 1.6897947788238525,
      "learning_rate": 3.89198430472428e-06,
      "loss": 0.5223,
      "step": 149900
    },
    {
      "epoch": 61.36,
      "grad_norm": 1.5647634267807007,
      "learning_rate": 3.891865209141952e-06,
      "loss": 0.5241,
      "step": 149910
    },
    {
      "epoch": 61.37,
      "grad_norm": 1.7749254703521729,
      "learning_rate": 3.891746106467948e-06,
      "loss": 0.5255,
      "step": 149920
    },
    {
      "epoch": 61.37,
      "grad_norm": 2.0004780292510986,
      "learning_rate": 3.891626996702814e-06,
      "loss": 0.5148,
      "step": 149930
    },
    {
      "epoch": 61.38,
      "grad_norm": 1.9405887126922607,
      "learning_rate": 3.891507879847094e-06,
      "loss": 0.5283,
      "step": 149940
    },
    {
      "epoch": 61.38,
      "grad_norm": 1.5243803262710571,
      "learning_rate": 3.891388755901337e-06,
      "loss": 0.5276,
      "step": 149950
    },
    {
      "epoch": 61.38,
      "grad_norm": 2.1392884254455566,
      "learning_rate": 3.891269624866084e-06,
      "loss": 0.538,
      "step": 149960
    },
    {
      "epoch": 61.39,
      "grad_norm": 2.215944766998291,
      "learning_rate": 3.891150486741885e-06,
      "loss": 0.5154,
      "step": 149970
    },
    {
      "epoch": 61.39,
      "grad_norm": 1.7291767597198486,
      "learning_rate": 3.891031341529283e-06,
      "loss": 0.5297,
      "step": 149980
    },
    {
      "epoch": 61.4,
      "grad_norm": 1.645116925239563,
      "learning_rate": 3.8909121892288255e-06,
      "loss": 0.5444,
      "step": 149990
    },
    {
      "epoch": 61.4,
      "grad_norm": 2.0054376125335693,
      "learning_rate": 3.890793029841056e-06,
      "loss": 0.5351,
      "step": 150000
    },
    {
      "epoch": 61.4,
      "grad_norm": 1.8083542585372925,
      "learning_rate": 3.8906738633665235e-06,
      "loss": 0.5255,
      "step": 150010
    },
    {
      "epoch": 61.41,
      "grad_norm": 2.0681025981903076,
      "learning_rate": 3.890554689805773e-06,
      "loss": 0.5096,
      "step": 150020
    },
    {
      "epoch": 61.41,
      "grad_norm": 1.931192398071289,
      "learning_rate": 3.890435509159349e-06,
      "loss": 0.5172,
      "step": 150030
    },
    {
      "epoch": 61.42,
      "grad_norm": 1.717603087425232,
      "learning_rate": 3.890316321427798e-06,
      "loss": 0.5269,
      "step": 150040
    },
    {
      "epoch": 61.42,
      "grad_norm": 1.988484501838684,
      "learning_rate": 3.890197126611667e-06,
      "loss": 0.5117,
      "step": 150050
    },
    {
      "epoch": 61.42,
      "grad_norm": 2.408557891845703,
      "learning_rate": 3.8900779247115e-06,
      "loss": 0.4966,
      "step": 150060
    },
    {
      "epoch": 61.43,
      "grad_norm": 1.8395270109176636,
      "learning_rate": 3.889958715727845e-06,
      "loss": 0.5184,
      "step": 150070
    },
    {
      "epoch": 61.43,
      "grad_norm": 1.9281558990478516,
      "learning_rate": 3.889839499661247e-06,
      "loss": 0.5231,
      "step": 150080
    },
    {
      "epoch": 61.44,
      "grad_norm": 2.1529948711395264,
      "learning_rate": 3.889720276512252e-06,
      "loss": 0.5342,
      "step": 150090
    },
    {
      "epoch": 61.44,
      "grad_norm": 2.219062089920044,
      "learning_rate": 3.889601046281408e-06,
      "loss": 0.5393,
      "step": 150100
    },
    {
      "epoch": 61.44,
      "grad_norm": 2.016726493835449,
      "learning_rate": 3.889481808969258e-06,
      "loss": 0.5421,
      "step": 150110
    },
    {
      "epoch": 61.45,
      "grad_norm": 1.8129618167877197,
      "learning_rate": 3.8893625645763506e-06,
      "loss": 0.5344,
      "step": 150120
    },
    {
      "epoch": 61.45,
      "grad_norm": 1.497687578201294,
      "learning_rate": 3.889243313103233e-06,
      "loss": 0.5523,
      "step": 150130
    },
    {
      "epoch": 61.46,
      "grad_norm": 2.6400845050811768,
      "learning_rate": 3.889124054550448e-06,
      "loss": 0.5202,
      "step": 150140
    },
    {
      "epoch": 61.46,
      "grad_norm": 1.7705916166305542,
      "learning_rate": 3.889004788918544e-06,
      "loss": 0.5269,
      "step": 150150
    },
    {
      "epoch": 61.47,
      "grad_norm": 2.723491907119751,
      "learning_rate": 3.888885516208068e-06,
      "loss": 0.5304,
      "step": 150160
    },
    {
      "epoch": 61.47,
      "grad_norm": 1.876911997795105,
      "learning_rate": 3.888766236419565e-06,
      "loss": 0.5212,
      "step": 150170
    },
    {
      "epoch": 61.47,
      "grad_norm": 2.1301887035369873,
      "learning_rate": 3.888646949553582e-06,
      "loss": 0.5247,
      "step": 150180
    },
    {
      "epoch": 61.48,
      "grad_norm": 2.187023878097534,
      "learning_rate": 3.888527655610664e-06,
      "loss": 0.521,
      "step": 150190
    },
    {
      "epoch": 61.48,
      "grad_norm": 2.2489867210388184,
      "learning_rate": 3.888408354591361e-06,
      "loss": 0.5392,
      "step": 150200
    },
    {
      "epoch": 61.49,
      "grad_norm": 2.226954698562622,
      "learning_rate": 3.8882890464962164e-06,
      "loss": 0.5234,
      "step": 150210
    },
    {
      "epoch": 61.49,
      "grad_norm": 2.185279607772827,
      "learning_rate": 3.888169731325777e-06,
      "loss": 0.513,
      "step": 150220
    },
    {
      "epoch": 61.49,
      "grad_norm": 1.6149805784225464,
      "learning_rate": 3.888050409080591e-06,
      "loss": 0.5307,
      "step": 150230
    },
    {
      "epoch": 61.5,
      "grad_norm": 1.8638451099395752,
      "learning_rate": 3.887931079761203e-06,
      "loss": 0.5206,
      "step": 150240
    },
    {
      "epoch": 61.5,
      "grad_norm": 2.3423595428466797,
      "learning_rate": 3.887811743368162e-06,
      "loss": 0.5165,
      "step": 150250
    },
    {
      "epoch": 61.51,
      "grad_norm": 1.811574935913086,
      "learning_rate": 3.887692399902012e-06,
      "loss": 0.5383,
      "step": 150260
    },
    {
      "epoch": 61.51,
      "grad_norm": 1.771540641784668,
      "learning_rate": 3.8875730493633016e-06,
      "loss": 0.4995,
      "step": 150270
    },
    {
      "epoch": 61.51,
      "grad_norm": 1.882521152496338,
      "learning_rate": 3.887453691752576e-06,
      "loss": 0.5307,
      "step": 150280
    },
    {
      "epoch": 61.52,
      "grad_norm": 1.8816471099853516,
      "learning_rate": 3.887334327070384e-06,
      "loss": 0.5169,
      "step": 150290
    },
    {
      "epoch": 61.52,
      "grad_norm": 2.1041440963745117,
      "learning_rate": 3.887214955317271e-06,
      "loss": 0.5185,
      "step": 150300
    },
    {
      "epoch": 61.53,
      "grad_norm": 1.6706866025924683,
      "learning_rate": 3.887095576493784e-06,
      "loss": 0.5172,
      "step": 150310
    },
    {
      "epoch": 61.53,
      "grad_norm": 1.4796086549758911,
      "learning_rate": 3.88697619060047e-06,
      "loss": 0.5292,
      "step": 150320
    },
    {
      "epoch": 61.53,
      "grad_norm": 1.7811814546585083,
      "learning_rate": 3.886856797637876e-06,
      "loss": 0.5384,
      "step": 150330
    },
    {
      "epoch": 61.54,
      "grad_norm": 1.795473337173462,
      "learning_rate": 3.886737397606548e-06,
      "loss": 0.5346,
      "step": 150340
    },
    {
      "epoch": 61.54,
      "grad_norm": 1.7210962772369385,
      "learning_rate": 3.886617990507034e-06,
      "loss": 0.5183,
      "step": 150350
    },
    {
      "epoch": 61.55,
      "grad_norm": 2.4097585678100586,
      "learning_rate": 3.886498576339881e-06,
      "loss": 0.5108,
      "step": 150360
    },
    {
      "epoch": 61.55,
      "grad_norm": 1.9858952760696411,
      "learning_rate": 3.886379155105635e-06,
      "loss": 0.5316,
      "step": 150370
    },
    {
      "epoch": 61.56,
      "grad_norm": 2.1378822326660156,
      "learning_rate": 3.886259726804844e-06,
      "loss": 0.5148,
      "step": 150380
    },
    {
      "epoch": 61.56,
      "grad_norm": 1.94449782371521,
      "learning_rate": 3.886140291438055e-06,
      "loss": 0.5144,
      "step": 150390
    },
    {
      "epoch": 61.56,
      "grad_norm": 1.668578028678894,
      "learning_rate": 3.886020849005815e-06,
      "loss": 0.5211,
      "step": 150400
    },
    {
      "epoch": 61.57,
      "grad_norm": 2.0120203495025635,
      "learning_rate": 3.885901399508671e-06,
      "loss": 0.5126,
      "step": 150410
    },
    {
      "epoch": 61.57,
      "grad_norm": 1.3771389722824097,
      "learning_rate": 3.885781942947171e-06,
      "loss": 0.5303,
      "step": 150420
    },
    {
      "epoch": 61.58,
      "grad_norm": 2.250089645385742,
      "learning_rate": 3.8856624793218605e-06,
      "loss": 0.5404,
      "step": 150430
    },
    {
      "epoch": 61.58,
      "grad_norm": 1.9726399183273315,
      "learning_rate": 3.885543008633288e-06,
      "loss": 0.5177,
      "step": 150440
    },
    {
      "epoch": 61.58,
      "grad_norm": 1.5682333707809448,
      "learning_rate": 3.885423530882001e-06,
      "loss": 0.5194,
      "step": 150450
    },
    {
      "epoch": 61.59,
      "grad_norm": 2.2868876457214355,
      "learning_rate": 3.885304046068546e-06,
      "loss": 0.5282,
      "step": 150460
    },
    {
      "epoch": 61.59,
      "grad_norm": 2.1570396423339844,
      "learning_rate": 3.88518455419347e-06,
      "loss": 0.5286,
      "step": 150470
    },
    {
      "epoch": 61.6,
      "grad_norm": 2.018754005432129,
      "learning_rate": 3.885065055257321e-06,
      "loss": 0.5316,
      "step": 150480
    },
    {
      "epoch": 61.6,
      "grad_norm": 1.7077301740646362,
      "learning_rate": 3.884945549260648e-06,
      "loss": 0.5364,
      "step": 150490
    },
    {
      "epoch": 61.6,
      "grad_norm": 2.5028367042541504,
      "learning_rate": 3.884826036203996e-06,
      "loss": 0.5306,
      "step": 150500
    },
    {
      "epoch": 61.61,
      "grad_norm": 2.152695417404175,
      "learning_rate": 3.8847065160879125e-06,
      "loss": 0.5253,
      "step": 150510
    },
    {
      "epoch": 61.61,
      "grad_norm": 2.0307259559631348,
      "learning_rate": 3.884586988912946e-06,
      "loss": 0.5408,
      "step": 150520
    },
    {
      "epoch": 61.62,
      "grad_norm": 2.395172119140625,
      "learning_rate": 3.884467454679645e-06,
      "loss": 0.5319,
      "step": 150530
    },
    {
      "epoch": 61.62,
      "grad_norm": 1.8193535804748535,
      "learning_rate": 3.884347913388555e-06,
      "loss": 0.5181,
      "step": 150540
    },
    {
      "epoch": 61.63,
      "grad_norm": 2.3817641735076904,
      "learning_rate": 3.884228365040225e-06,
      "loss": 0.5148,
      "step": 150550
    },
    {
      "epoch": 61.63,
      "grad_norm": 2.1619911193847656,
      "learning_rate": 3.884108809635203e-06,
      "loss": 0.5441,
      "step": 150560
    },
    {
      "epoch": 61.63,
      "grad_norm": 1.7783458232879639,
      "learning_rate": 3.883989247174034e-06,
      "loss": 0.5385,
      "step": 150570
    },
    {
      "epoch": 61.64,
      "grad_norm": 1.7573124170303345,
      "learning_rate": 3.883869677657269e-06,
      "loss": 0.5184,
      "step": 150580
    },
    {
      "epoch": 61.64,
      "grad_norm": 1.6309067010879517,
      "learning_rate": 3.883750101085453e-06,
      "loss": 0.5286,
      "step": 150590
    },
    {
      "epoch": 61.65,
      "grad_norm": 2.071073532104492,
      "learning_rate": 3.883630517459137e-06,
      "loss": 0.5276,
      "step": 150600
    },
    {
      "epoch": 61.65,
      "grad_norm": 1.966497540473938,
      "learning_rate": 3.883510926778865e-06,
      "loss": 0.5403,
      "step": 150610
    },
    {
      "epoch": 61.65,
      "grad_norm": 1.7112663984298706,
      "learning_rate": 3.883391329045187e-06,
      "loss": 0.531,
      "step": 150620
    },
    {
      "epoch": 61.66,
      "grad_norm": 2.008841037750244,
      "learning_rate": 3.883271724258651e-06,
      "loss": 0.543,
      "step": 150630
    },
    {
      "epoch": 61.66,
      "grad_norm": 1.6403803825378418,
      "learning_rate": 3.883152112419805e-06,
      "loss": 0.5159,
      "step": 150640
    },
    {
      "epoch": 61.67,
      "grad_norm": 2.467533588409424,
      "learning_rate": 3.883032493529196e-06,
      "loss": 0.5126,
      "step": 150650
    },
    {
      "epoch": 61.67,
      "grad_norm": 2.095567464828491,
      "learning_rate": 3.882912867587371e-06,
      "loss": 0.5187,
      "step": 150660
    },
    {
      "epoch": 61.67,
      "grad_norm": 1.9959877729415894,
      "learning_rate": 3.88279323459488e-06,
      "loss": 0.5249,
      "step": 150670
    },
    {
      "epoch": 61.68,
      "grad_norm": 2.178438425064087,
      "learning_rate": 3.88267359455227e-06,
      "loss": 0.5116,
      "step": 150680
    },
    {
      "epoch": 61.68,
      "grad_norm": 1.5299681425094604,
      "learning_rate": 3.88255394746009e-06,
      "loss": 0.5284,
      "step": 150690
    },
    {
      "epoch": 61.69,
      "grad_norm": 1.7405372858047485,
      "learning_rate": 3.882434293318887e-06,
      "loss": 0.5193,
      "step": 150700
    },
    {
      "epoch": 61.69,
      "grad_norm": 1.6497186422348022,
      "learning_rate": 3.8823146321292104e-06,
      "loss": 0.505,
      "step": 150710
    },
    {
      "epoch": 61.69,
      "grad_norm": 2.368659257888794,
      "learning_rate": 3.882194963891607e-06,
      "loss": 0.5337,
      "step": 150720
    },
    {
      "epoch": 61.7,
      "grad_norm": 1.996516227722168,
      "learning_rate": 3.8820752886066245e-06,
      "loss": 0.5096,
      "step": 150730
    },
    {
      "epoch": 61.7,
      "grad_norm": 2.1482772827148438,
      "learning_rate": 3.8819556062748134e-06,
      "loss": 0.53,
      "step": 150740
    },
    {
      "epoch": 61.71,
      "grad_norm": 1.9372533559799194,
      "learning_rate": 3.88183591689672e-06,
      "loss": 0.5406,
      "step": 150750
    },
    {
      "epoch": 61.71,
      "grad_norm": 1.668088674545288,
      "learning_rate": 3.881716220472893e-06,
      "loss": 0.5345,
      "step": 150760
    },
    {
      "epoch": 61.72,
      "grad_norm": 2.0272042751312256,
      "learning_rate": 3.881596517003881e-06,
      "loss": 0.5124,
      "step": 150770
    },
    {
      "epoch": 61.72,
      "grad_norm": 2.5758605003356934,
      "learning_rate": 3.881476806490233e-06,
      "loss": 0.5228,
      "step": 150780
    },
    {
      "epoch": 61.72,
      "grad_norm": 1.6942273378372192,
      "learning_rate": 3.881357088932497e-06,
      "loss": 0.5087,
      "step": 150790
    },
    {
      "epoch": 61.73,
      "grad_norm": 1.822277545928955,
      "learning_rate": 3.881237364331219e-06,
      "loss": 0.5142,
      "step": 150800
    },
    {
      "epoch": 61.73,
      "grad_norm": 1.5457683801651,
      "learning_rate": 3.881117632686951e-06,
      "loss": 0.5076,
      "step": 150810
    },
    {
      "epoch": 61.74,
      "grad_norm": 1.9316630363464355,
      "learning_rate": 3.8809978940002405e-06,
      "loss": 0.548,
      "step": 150820
    },
    {
      "epoch": 61.74,
      "grad_norm": 2.4272749423980713,
      "learning_rate": 3.8808781482716345e-06,
      "loss": 0.5368,
      "step": 150830
    },
    {
      "epoch": 61.74,
      "grad_norm": 1.773330569267273,
      "learning_rate": 3.8807583955016825e-06,
      "loss": 0.5177,
      "step": 150840
    },
    {
      "epoch": 61.75,
      "grad_norm": 1.8053003549575806,
      "learning_rate": 3.880638635690933e-06,
      "loss": 0.5311,
      "step": 150850
    },
    {
      "epoch": 61.75,
      "grad_norm": 1.7200530767440796,
      "learning_rate": 3.8805188688399354e-06,
      "loss": 0.5116,
      "step": 150860
    },
    {
      "epoch": 61.76,
      "grad_norm": 2.6540780067443848,
      "learning_rate": 3.880399094949237e-06,
      "loss": 0.5086,
      "step": 150870
    },
    {
      "epoch": 61.76,
      "grad_norm": 2.2340357303619385,
      "learning_rate": 3.880279314019387e-06,
      "loss": 0.5264,
      "step": 150880
    },
    {
      "epoch": 61.76,
      "grad_norm": 1.601706624031067,
      "learning_rate": 3.880159526050935e-06,
      "loss": 0.524,
      "step": 150890
    },
    {
      "epoch": 61.77,
      "grad_norm": 1.9865368604660034,
      "learning_rate": 3.880039731044427e-06,
      "loss": 0.5068,
      "step": 150900
    },
    {
      "epoch": 61.77,
      "grad_norm": 2.0326011180877686,
      "learning_rate": 3.8799199290004156e-06,
      "loss": 0.5431,
      "step": 150910
    },
    {
      "epoch": 61.78,
      "grad_norm": 2.1464295387268066,
      "learning_rate": 3.879800119919447e-06,
      "loss": 0.5517,
      "step": 150920
    },
    {
      "epoch": 61.78,
      "grad_norm": 1.6363083124160767,
      "learning_rate": 3.879680303802071e-06,
      "loss": 0.5223,
      "step": 150930
    },
    {
      "epoch": 61.78,
      "grad_norm": 2.1601500511169434,
      "learning_rate": 3.879560480648836e-06,
      "loss": 0.5399,
      "step": 150940
    },
    {
      "epoch": 61.79,
      "grad_norm": 2.0271859169006348,
      "learning_rate": 3.87944065046029e-06,
      "loss": 0.5252,
      "step": 150950
    },
    {
      "epoch": 61.79,
      "grad_norm": 1.755632758140564,
      "learning_rate": 3.879320813236985e-06,
      "loss": 0.5512,
      "step": 150960
    },
    {
      "epoch": 61.8,
      "grad_norm": 1.5777640342712402,
      "learning_rate": 3.879200968979466e-06,
      "loss": 0.5411,
      "step": 150970
    },
    {
      "epoch": 61.8,
      "grad_norm": 1.5317068099975586,
      "learning_rate": 3.879081117688285e-06,
      "loss": 0.5186,
      "step": 150980
    },
    {
      "epoch": 61.81,
      "grad_norm": 1.8013278245925903,
      "learning_rate": 3.878961259363989e-06,
      "loss": 0.5177,
      "step": 150990
    },
    {
      "epoch": 61.81,
      "grad_norm": 1.8313578367233276,
      "learning_rate": 3.8788413940071295e-06,
      "loss": 0.5417,
      "step": 151000
    },
    {
      "epoch": 61.81,
      "grad_norm": 2.127776861190796,
      "learning_rate": 3.878721521618253e-06,
      "loss": 0.5344,
      "step": 151010
    },
    {
      "epoch": 61.82,
      "grad_norm": 1.9516078233718872,
      "learning_rate": 3.87860164219791e-06,
      "loss": 0.5208,
      "step": 151020
    },
    {
      "epoch": 61.82,
      "grad_norm": 1.839292287826538,
      "learning_rate": 3.878481755746649e-06,
      "loss": 0.5313,
      "step": 151030
    },
    {
      "epoch": 61.83,
      "grad_norm": 2.0395443439483643,
      "learning_rate": 3.87836186226502e-06,
      "loss": 0.5132,
      "step": 151040
    },
    {
      "epoch": 61.83,
      "grad_norm": 1.8940119743347168,
      "learning_rate": 3.878241961753571e-06,
      "loss": 0.532,
      "step": 151050
    },
    {
      "epoch": 61.83,
      "grad_norm": 2.731226921081543,
      "learning_rate": 3.8781220542128525e-06,
      "loss": 0.5209,
      "step": 151060
    },
    {
      "epoch": 61.84,
      "grad_norm": 1.9027016162872314,
      "learning_rate": 3.878002139643413e-06,
      "loss": 0.5332,
      "step": 151070
    },
    {
      "epoch": 61.84,
      "grad_norm": 1.8062318563461304,
      "learning_rate": 3.877882218045803e-06,
      "loss": 0.5287,
      "step": 151080
    },
    {
      "epoch": 61.85,
      "grad_norm": 2.0838961601257324,
      "learning_rate": 3.877762289420569e-06,
      "loss": 0.5232,
      "step": 151090
    },
    {
      "epoch": 61.85,
      "grad_norm": 2.1688530445098877,
      "learning_rate": 3.877642353768263e-06,
      "loss": 0.5294,
      "step": 151100
    },
    {
      "epoch": 61.85,
      "grad_norm": 1.680254340171814,
      "learning_rate": 3.877522411089435e-06,
      "loss": 0.5373,
      "step": 151110
    },
    {
      "epoch": 61.86,
      "grad_norm": 2.6453654766082764,
      "learning_rate": 3.877402461384632e-06,
      "loss": 0.5195,
      "step": 151120
    },
    {
      "epoch": 61.86,
      "grad_norm": 2.0553834438323975,
      "learning_rate": 3.877282504654404e-06,
      "loss": 0.5054,
      "step": 151130
    },
    {
      "epoch": 61.87,
      "grad_norm": 1.9330852031707764,
      "learning_rate": 3.8771625408993025e-06,
      "loss": 0.5244,
      "step": 151140
    },
    {
      "epoch": 61.87,
      "grad_norm": 1.5203200578689575,
      "learning_rate": 3.877042570119875e-06,
      "loss": 0.5213,
      "step": 151150
    },
    {
      "epoch": 61.87,
      "grad_norm": 2.2064783573150635,
      "learning_rate": 3.876922592316672e-06,
      "loss": 0.5379,
      "step": 151160
    },
    {
      "epoch": 61.88,
      "grad_norm": 2.650312900543213,
      "learning_rate": 3.876802607490242e-06,
      "loss": 0.5287,
      "step": 151170
    },
    {
      "epoch": 61.88,
      "grad_norm": 2.1615805625915527,
      "learning_rate": 3.876682615641136e-06,
      "loss": 0.5349,
      "step": 151180
    },
    {
      "epoch": 61.89,
      "grad_norm": 2.405921697616577,
      "learning_rate": 3.876562616769902e-06,
      "loss": 0.5239,
      "step": 151190
    },
    {
      "epoch": 61.89,
      "grad_norm": 2.1587233543395996,
      "learning_rate": 3.8764426108770925e-06,
      "loss": 0.5321,
      "step": 151200
    },
    {
      "epoch": 61.9,
      "grad_norm": 1.9966261386871338,
      "learning_rate": 3.876322597963255e-06,
      "loss": 0.5211,
      "step": 151210
    },
    {
      "epoch": 61.9,
      "grad_norm": 1.7234680652618408,
      "learning_rate": 3.876202578028939e-06,
      "loss": 0.5256,
      "step": 151220
    },
    {
      "epoch": 61.9,
      "grad_norm": 5.025945663452148,
      "learning_rate": 3.8760825510746956e-06,
      "loss": 0.5388,
      "step": 151230
    },
    {
      "epoch": 61.91,
      "grad_norm": 2.167874813079834,
      "learning_rate": 3.875962517101075e-06,
      "loss": 0.5265,
      "step": 151240
    },
    {
      "epoch": 61.91,
      "grad_norm": 2.2046449184417725,
      "learning_rate": 3.875842476108625e-06,
      "loss": 0.5293,
      "step": 151250
    },
    {
      "epoch": 61.92,
      "grad_norm": 1.8970714807510376,
      "learning_rate": 3.875722428097897e-06,
      "loss": 0.5227,
      "step": 151260
    },
    {
      "epoch": 61.92,
      "grad_norm": 1.4749358892440796,
      "learning_rate": 3.8756023730694404e-06,
      "loss": 0.5267,
      "step": 151270
    },
    {
      "epoch": 61.92,
      "grad_norm": 1.8844729661941528,
      "learning_rate": 3.875482311023805e-06,
      "loss": 0.5284,
      "step": 151280
    },
    {
      "epoch": 61.93,
      "grad_norm": 2.818701982498169,
      "learning_rate": 3.8753622419615416e-06,
      "loss": 0.5349,
      "step": 151290
    },
    {
      "epoch": 61.93,
      "grad_norm": 1.6745208501815796,
      "learning_rate": 3.8752421658832005e-06,
      "loss": 0.5324,
      "step": 151300
    },
    {
      "epoch": 61.94,
      "grad_norm": 1.9570318460464478,
      "learning_rate": 3.8751220827893295e-06,
      "loss": 0.5284,
      "step": 151310
    },
    {
      "epoch": 61.94,
      "grad_norm": 1.9855762720108032,
      "learning_rate": 3.875001992680481e-06,
      "loss": 0.536,
      "step": 151320
    },
    {
      "epoch": 61.94,
      "grad_norm": 1.5116163492202759,
      "learning_rate": 3.874881895557203e-06,
      "loss": 0.5372,
      "step": 151330
    },
    {
      "epoch": 61.95,
      "grad_norm": 2.2885966300964355,
      "learning_rate": 3.8747617914200484e-06,
      "loss": 0.5139,
      "step": 151340
    },
    {
      "epoch": 61.95,
      "grad_norm": 2.2124104499816895,
      "learning_rate": 3.874641680269565e-06,
      "loss": 0.534,
      "step": 151350
    },
    {
      "epoch": 61.96,
      "grad_norm": 1.918231725692749,
      "learning_rate": 3.874521562106305e-06,
      "loss": 0.5307,
      "step": 151360
    },
    {
      "epoch": 61.96,
      "grad_norm": 1.853096604347229,
      "learning_rate": 3.874401436930817e-06,
      "loss": 0.5397,
      "step": 151370
    },
    {
      "epoch": 61.96,
      "grad_norm": 1.9706991910934448,
      "learning_rate": 3.874281304743651e-06,
      "loss": 0.5428,
      "step": 151380
    },
    {
      "epoch": 61.97,
      "grad_norm": 1.5117031335830688,
      "learning_rate": 3.874161165545359e-06,
      "loss": 0.5305,
      "step": 151390
    },
    {
      "epoch": 61.97,
      "grad_norm": 2.145629405975342,
      "learning_rate": 3.87404101933649e-06,
      "loss": 0.5291,
      "step": 151400
    },
    {
      "epoch": 61.98,
      "grad_norm": 1.8175499439239502,
      "learning_rate": 3.873920866117596e-06,
      "loss": 0.537,
      "step": 151410
    },
    {
      "epoch": 61.98,
      "grad_norm": 1.690921664237976,
      "learning_rate": 3.873800705889225e-06,
      "loss": 0.5327,
      "step": 151420
    },
    {
      "epoch": 61.99,
      "grad_norm": 1.4447622299194336,
      "learning_rate": 3.8736805386519284e-06,
      "loss": 0.5235,
      "step": 151430
    },
    {
      "epoch": 61.99,
      "grad_norm": 2.111146926879883,
      "learning_rate": 3.8735603644062585e-06,
      "loss": 0.4964,
      "step": 151440
    },
    {
      "epoch": 61.99,
      "grad_norm": 3.4365696907043457,
      "learning_rate": 3.873440183152762e-06,
      "loss": 0.502,
      "step": 151450
    },
    {
      "epoch": 62.0,
      "grad_norm": 1.864060878753662,
      "learning_rate": 3.873319994891994e-06,
      "loss": 0.5352,
      "step": 151460
    },
    {
      "epoch": 62.0,
      "eval_loss": 0.5263904929161072,
      "eval_runtime": 53.079,
      "eval_samples_per_second": 64.979,
      "eval_steps_per_second": 8.139,
      "step": 151466
    },
    {
      "epoch": 62.0,
      "grad_norm": 1.9559988975524902,
      "learning_rate": 3.8731997996245016e-06,
      "loss": 0.5292,
      "step": 151470
    },
    {
      "epoch": 62.01,
      "grad_norm": 1.700243353843689,
      "learning_rate": 3.873079597350837e-06,
      "loss": 0.5237,
      "step": 151480
    },
    {
      "epoch": 62.01,
      "grad_norm": 1.7002853155136108,
      "learning_rate": 3.872959388071549e-06,
      "loss": 0.5366,
      "step": 151490
    },
    {
      "epoch": 62.01,
      "grad_norm": 2.033539056777954,
      "learning_rate": 3.872839171787191e-06,
      "loss": 0.5312,
      "step": 151500
    },
    {
      "epoch": 62.02,
      "grad_norm": 1.5000505447387695,
      "learning_rate": 3.872718948498312e-06,
      "loss": 0.5151,
      "step": 151510
    },
    {
      "epoch": 62.02,
      "grad_norm": 1.5782763957977295,
      "learning_rate": 3.872598718205463e-06,
      "loss": 0.5098,
      "step": 151520
    },
    {
      "epoch": 62.03,
      "grad_norm": 1.6860828399658203,
      "learning_rate": 3.872478480909195e-06,
      "loss": 0.5036,
      "step": 151530
    },
    {
      "epoch": 62.03,
      "grad_norm": 1.6567349433898926,
      "learning_rate": 3.872358236610059e-06,
      "loss": 0.5432,
      "step": 151540
    },
    {
      "epoch": 62.03,
      "grad_norm": 1.9934245347976685,
      "learning_rate": 3.872237985308604e-06,
      "loss": 0.5315,
      "step": 151550
    },
    {
      "epoch": 62.04,
      "grad_norm": 1.8039385080337524,
      "learning_rate": 3.872117727005382e-06,
      "loss": 0.5466,
      "step": 151560
    },
    {
      "epoch": 62.04,
      "grad_norm": 2.057943105697632,
      "learning_rate": 3.871997461700946e-06,
      "loss": 0.5311,
      "step": 151570
    },
    {
      "epoch": 62.05,
      "grad_norm": 2.3186194896698,
      "learning_rate": 3.871877189395843e-06,
      "loss": 0.5111,
      "step": 151580
    },
    {
      "epoch": 62.05,
      "grad_norm": 1.945086121559143,
      "learning_rate": 3.871756910090629e-06,
      "loss": 0.5506,
      "step": 151590
    },
    {
      "epoch": 62.05,
      "grad_norm": 2.244349241256714,
      "learning_rate": 3.87163662378585e-06,
      "loss": 0.5377,
      "step": 151600
    },
    {
      "epoch": 62.06,
      "grad_norm": 1.6765704154968262,
      "learning_rate": 3.871516330482059e-06,
      "loss": 0.5264,
      "step": 151610
    },
    {
      "epoch": 62.06,
      "grad_norm": 2.1508243083953857,
      "learning_rate": 3.8713960301798066e-06,
      "loss": 0.5315,
      "step": 151620
    },
    {
      "epoch": 62.07,
      "grad_norm": 1.9282382726669312,
      "learning_rate": 3.871275722879645e-06,
      "loss": 0.518,
      "step": 151630
    },
    {
      "epoch": 62.07,
      "grad_norm": 2.0601179599761963,
      "learning_rate": 3.8711554085821246e-06,
      "loss": 0.5262,
      "step": 151640
    },
    {
      "epoch": 62.08,
      "grad_norm": 1.7481873035430908,
      "learning_rate": 3.871035087287797e-06,
      "loss": 0.5166,
      "step": 151650
    },
    {
      "epoch": 62.08,
      "grad_norm": 1.7192457914352417,
      "learning_rate": 3.870914758997213e-06,
      "loss": 0.5184,
      "step": 151660
    },
    {
      "epoch": 62.08,
      "grad_norm": 1.9794511795043945,
      "learning_rate": 3.8707944237109224e-06,
      "loss": 0.5139,
      "step": 151670
    },
    {
      "epoch": 62.09,
      "grad_norm": 1.5283211469650269,
      "learning_rate": 3.870674081429479e-06,
      "loss": 0.5329,
      "step": 151680
    },
    {
      "epoch": 62.09,
      "grad_norm": 1.7096457481384277,
      "learning_rate": 3.8705537321534325e-06,
      "loss": 0.5208,
      "step": 151690
    },
    {
      "epoch": 62.1,
      "grad_norm": 1.8405213356018066,
      "learning_rate": 3.870433375883334e-06,
      "loss": 0.5251,
      "step": 151700
    },
    {
      "epoch": 62.1,
      "grad_norm": 1.665083408355713,
      "learning_rate": 3.870313012619736e-06,
      "loss": 0.5349,
      "step": 151710
    },
    {
      "epoch": 62.1,
      "grad_norm": 2.172706127166748,
      "learning_rate": 3.87019264236319e-06,
      "loss": 0.5322,
      "step": 151720
    },
    {
      "epoch": 62.11,
      "grad_norm": 2.3452601432800293,
      "learning_rate": 3.870072265114245e-06,
      "loss": 0.5525,
      "step": 151730
    },
    {
      "epoch": 62.11,
      "grad_norm": 2.0081071853637695,
      "learning_rate": 3.8699518808734554e-06,
      "loss": 0.5264,
      "step": 151740
    },
    {
      "epoch": 62.12,
      "grad_norm": 2.026357412338257,
      "learning_rate": 3.869831489641371e-06,
      "loss": 0.5358,
      "step": 151750
    },
    {
      "epoch": 62.12,
      "grad_norm": 2.01367449760437,
      "learning_rate": 3.869711091418544e-06,
      "loss": 0.5181,
      "step": 151760
    },
    {
      "epoch": 62.12,
      "grad_norm": 2.2139828205108643,
      "learning_rate": 3.8695906862055246e-06,
      "loss": 0.511,
      "step": 151770
    },
    {
      "epoch": 62.13,
      "grad_norm": 1.6562870740890503,
      "learning_rate": 3.869470274002866e-06,
      "loss": 0.5386,
      "step": 151780
    },
    {
      "epoch": 62.13,
      "grad_norm": 1.604129433631897,
      "learning_rate": 3.869349854811119e-06,
      "loss": 0.519,
      "step": 151790
    },
    {
      "epoch": 62.14,
      "grad_norm": 2.0224461555480957,
      "learning_rate": 3.869229428630835e-06,
      "loss": 0.5253,
      "step": 151800
    },
    {
      "epoch": 62.14,
      "grad_norm": 1.7049827575683594,
      "learning_rate": 3.869108995462566e-06,
      "loss": 0.5179,
      "step": 151810
    },
    {
      "epoch": 62.14,
      "grad_norm": 2.5049831867218018,
      "learning_rate": 3.868988555306864e-06,
      "loss": 0.5338,
      "step": 151820
    },
    {
      "epoch": 62.15,
      "grad_norm": 2.355252981185913,
      "learning_rate": 3.868868108164281e-06,
      "loss": 0.5027,
      "step": 151830
    },
    {
      "epoch": 62.15,
      "grad_norm": 1.8900772333145142,
      "learning_rate": 3.868747654035367e-06,
      "loss": 0.5412,
      "step": 151840
    },
    {
      "epoch": 62.16,
      "grad_norm": 2.2201156616210938,
      "learning_rate": 3.868627192920675e-06,
      "loss": 0.5129,
      "step": 151850
    },
    {
      "epoch": 62.16,
      "grad_norm": 1.7482364177703857,
      "learning_rate": 3.868506724820757e-06,
      "loss": 0.5253,
      "step": 151860
    },
    {
      "epoch": 62.17,
      "grad_norm": 1.7953252792358398,
      "learning_rate": 3.868386249736165e-06,
      "loss": 0.5368,
      "step": 151870
    },
    {
      "epoch": 62.17,
      "grad_norm": 1.7478828430175781,
      "learning_rate": 3.868265767667451e-06,
      "loss": 0.5054,
      "step": 151880
    },
    {
      "epoch": 62.17,
      "grad_norm": 1.5260462760925293,
      "learning_rate": 3.868145278615165e-06,
      "loss": 0.5241,
      "step": 151890
    },
    {
      "epoch": 62.18,
      "grad_norm": 1.868094563484192,
      "learning_rate": 3.8680247825798605e-06,
      "loss": 0.5298,
      "step": 151900
    },
    {
      "epoch": 62.18,
      "grad_norm": 2.3297207355499268,
      "learning_rate": 3.86790427956209e-06,
      "loss": 0.5065,
      "step": 151910
    },
    {
      "epoch": 62.19,
      "grad_norm": 2.269955635070801,
      "learning_rate": 3.867783769562404e-06,
      "loss": 0.537,
      "step": 151920
    },
    {
      "epoch": 62.19,
      "grad_norm": 1.5671446323394775,
      "learning_rate": 3.8676632525813555e-06,
      "loss": 0.5462,
      "step": 151930
    },
    {
      "epoch": 62.19,
      "grad_norm": 1.923513412475586,
      "learning_rate": 3.867542728619497e-06,
      "loss": 0.5218,
      "step": 151940
    },
    {
      "epoch": 62.2,
      "grad_norm": 2.083667278289795,
      "learning_rate": 3.867422197677379e-06,
      "loss": 0.5066,
      "step": 151950
    },
    {
      "epoch": 62.2,
      "grad_norm": 1.9991533756256104,
      "learning_rate": 3.867301659755556e-06,
      "loss": 0.5284,
      "step": 151960
    },
    {
      "epoch": 62.21,
      "grad_norm": 1.660483956336975,
      "learning_rate": 3.867181114854578e-06,
      "loss": 0.5082,
      "step": 151970
    },
    {
      "epoch": 62.21,
      "grad_norm": 2.307002067565918,
      "learning_rate": 3.867060562974998e-06,
      "loss": 0.5074,
      "step": 151980
    },
    {
      "epoch": 62.21,
      "grad_norm": 1.8546297550201416,
      "learning_rate": 3.866940004117368e-06,
      "loss": 0.5226,
      "step": 151990
    },
    {
      "epoch": 62.22,
      "grad_norm": 2.1263506412506104,
      "learning_rate": 3.866819438282241e-06,
      "loss": 0.5418,
      "step": 152000
    },
    {
      "epoch": 62.22,
      "grad_norm": 2.1594505310058594,
      "learning_rate": 3.866698865470169e-06,
      "loss": 0.5229,
      "step": 152010
    },
    {
      "epoch": 62.23,
      "grad_norm": 2.151752471923828,
      "learning_rate": 3.866578285681703e-06,
      "loss": 0.5258,
      "step": 152020
    },
    {
      "epoch": 62.23,
      "grad_norm": 1.8566099405288696,
      "learning_rate": 3.8664576989173975e-06,
      "loss": 0.539,
      "step": 152030
    },
    {
      "epoch": 62.23,
      "grad_norm": 2.0505707263946533,
      "learning_rate": 3.866337105177803e-06,
      "loss": 0.5132,
      "step": 152040
    },
    {
      "epoch": 62.24,
      "grad_norm": 1.5982199907302856,
      "learning_rate": 3.866216504463474e-06,
      "loss": 0.519,
      "step": 152050
    },
    {
      "epoch": 62.24,
      "grad_norm": 2.0845279693603516,
      "learning_rate": 3.86609589677496e-06,
      "loss": 0.5318,
      "step": 152060
    },
    {
      "epoch": 62.25,
      "grad_norm": 2.5469868183135986,
      "learning_rate": 3.865975282112817e-06,
      "loss": 0.5266,
      "step": 152070
    },
    {
      "epoch": 62.25,
      "grad_norm": 2.1977927684783936,
      "learning_rate": 3.865854660477594e-06,
      "loss": 0.5136,
      "step": 152080
    },
    {
      "epoch": 62.26,
      "grad_norm": 1.5853837728500366,
      "learning_rate": 3.8657340318698465e-06,
      "loss": 0.5154,
      "step": 152090
    },
    {
      "epoch": 62.26,
      "grad_norm": 1.8560855388641357,
      "learning_rate": 3.865613396290126e-06,
      "loss": 0.5139,
      "step": 152100
    },
    {
      "epoch": 62.26,
      "grad_norm": 1.6821988821029663,
      "learning_rate": 3.8654927537389845e-06,
      "loss": 0.505,
      "step": 152110
    },
    {
      "epoch": 62.27,
      "grad_norm": 2.208148956298828,
      "learning_rate": 3.865372104216975e-06,
      "loss": 0.5266,
      "step": 152120
    },
    {
      "epoch": 62.27,
      "grad_norm": 1.739022135734558,
      "learning_rate": 3.865251447724651e-06,
      "loss": 0.5158,
      "step": 152130
    },
    {
      "epoch": 62.28,
      "grad_norm": 1.7600430250167847,
      "learning_rate": 3.865130784262564e-06,
      "loss": 0.5258,
      "step": 152140
    },
    {
      "epoch": 62.28,
      "grad_norm": 1.6983846426010132,
      "learning_rate": 3.865010113831267e-06,
      "loss": 0.5458,
      "step": 152150
    },
    {
      "epoch": 62.28,
      "grad_norm": 1.9008738994598389,
      "learning_rate": 3.864889436431313e-06,
      "loss": 0.5136,
      "step": 152160
    },
    {
      "epoch": 62.29,
      "grad_norm": 1.8611698150634766,
      "learning_rate": 3.864768752063255e-06,
      "loss": 0.5008,
      "step": 152170
    },
    {
      "epoch": 62.29,
      "grad_norm": 1.5213782787322998,
      "learning_rate": 3.864648060727646e-06,
      "loss": 0.5327,
      "step": 152180
    },
    {
      "epoch": 62.3,
      "grad_norm": 1.8942979574203491,
      "learning_rate": 3.864527362425038e-06,
      "loss": 0.5427,
      "step": 152190
    },
    {
      "epoch": 62.3,
      "grad_norm": 2.21745228767395,
      "learning_rate": 3.864406657155984e-06,
      "loss": 0.5481,
      "step": 152200
    },
    {
      "epoch": 62.3,
      "grad_norm": 2.4637763500213623,
      "learning_rate": 3.8642859449210385e-06,
      "loss": 0.5121,
      "step": 152210
    },
    {
      "epoch": 62.31,
      "grad_norm": 1.8165162801742554,
      "learning_rate": 3.864165225720753e-06,
      "loss": 0.5156,
      "step": 152220
    },
    {
      "epoch": 62.31,
      "grad_norm": 2.0406508445739746,
      "learning_rate": 3.86404449955568e-06,
      "loss": 0.518,
      "step": 152230
    },
    {
      "epoch": 62.32,
      "grad_norm": 1.627376914024353,
      "learning_rate": 3.863923766426375e-06,
      "loss": 0.5257,
      "step": 152240
    },
    {
      "epoch": 62.32,
      "grad_norm": 1.8290590047836304,
      "learning_rate": 3.863803026333388e-06,
      "loss": 0.5154,
      "step": 152250
    },
    {
      "epoch": 62.33,
      "grad_norm": 1.8764934539794922,
      "learning_rate": 3.863682279277273e-06,
      "loss": 0.5234,
      "step": 152260
    },
    {
      "epoch": 62.33,
      "grad_norm": 1.9816648960113525,
      "learning_rate": 3.863561525258585e-06,
      "loss": 0.5078,
      "step": 152270
    },
    {
      "epoch": 62.33,
      "grad_norm": 1.7717125415802002,
      "learning_rate": 3.863440764277875e-06,
      "loss": 0.5291,
      "step": 152280
    },
    {
      "epoch": 62.34,
      "grad_norm": 2.603956699371338,
      "learning_rate": 3.863319996335698e-06,
      "loss": 0.5343,
      "step": 152290
    },
    {
      "epoch": 62.34,
      "grad_norm": 2.3217415809631348,
      "learning_rate": 3.863199221432605e-06,
      "loss": 0.5268,
      "step": 152300
    },
    {
      "epoch": 62.35,
      "grad_norm": 2.558468818664551,
      "learning_rate": 3.863078439569151e-06,
      "loss": 0.5168,
      "step": 152310
    },
    {
      "epoch": 62.35,
      "grad_norm": 2.622220039367676,
      "learning_rate": 3.862957650745888e-06,
      "loss": 0.5208,
      "step": 152320
    },
    {
      "epoch": 62.35,
      "grad_norm": 2.095717191696167,
      "learning_rate": 3.862836854963371e-06,
      "loss": 0.5234,
      "step": 152330
    },
    {
      "epoch": 62.36,
      "grad_norm": 1.7304052114486694,
      "learning_rate": 3.862716052222152e-06,
      "loss": 0.5101,
      "step": 152340
    },
    {
      "epoch": 62.36,
      "grad_norm": 1.6146916151046753,
      "learning_rate": 3.862595242522785e-06,
      "loss": 0.526,
      "step": 152350
    },
    {
      "epoch": 62.37,
      "grad_norm": 2.4420619010925293,
      "learning_rate": 3.862474425865822e-06,
      "loss": 0.5186,
      "step": 152360
    },
    {
      "epoch": 62.37,
      "grad_norm": 1.8684892654418945,
      "learning_rate": 3.862353602251819e-06,
      "loss": 0.5427,
      "step": 152370
    },
    {
      "epoch": 62.37,
      "grad_norm": 2.1146492958068848,
      "learning_rate": 3.8622327716813275e-06,
      "loss": 0.5237,
      "step": 152380
    },
    {
      "epoch": 62.38,
      "grad_norm": 2.067031145095825,
      "learning_rate": 3.862111934154902e-06,
      "loss": 0.5052,
      "step": 152390
    },
    {
      "epoch": 62.38,
      "grad_norm": 2.065354585647583,
      "learning_rate": 3.8619910896730945e-06,
      "loss": 0.5236,
      "step": 152400
    },
    {
      "epoch": 62.39,
      "grad_norm": 2.709503173828125,
      "learning_rate": 3.861870238236461e-06,
      "loss": 0.5166,
      "step": 152410
    },
    {
      "epoch": 62.39,
      "grad_norm": 2.296949625015259,
      "learning_rate": 3.8617493798455535e-06,
      "loss": 0.5607,
      "step": 152420
    },
    {
      "epoch": 62.39,
      "grad_norm": 2.419957160949707,
      "learning_rate": 3.8616285145009254e-06,
      "loss": 0.5264,
      "step": 152430
    },
    {
      "epoch": 62.4,
      "grad_norm": 1.9900017976760864,
      "learning_rate": 3.861507642203132e-06,
      "loss": 0.5429,
      "step": 152440
    },
    {
      "epoch": 62.4,
      "grad_norm": 2.020221471786499,
      "learning_rate": 3.861386762952724e-06,
      "loss": 0.53,
      "step": 152450
    },
    {
      "epoch": 62.41,
      "grad_norm": 2.2632293701171875,
      "learning_rate": 3.861265876750259e-06,
      "loss": 0.5178,
      "step": 152460
    },
    {
      "epoch": 62.41,
      "grad_norm": 2.100618600845337,
      "learning_rate": 3.861144983596287e-06,
      "loss": 0.5282,
      "step": 152470
    },
    {
      "epoch": 62.42,
      "grad_norm": 2.2004692554473877,
      "learning_rate": 3.861024083491366e-06,
      "loss": 0.5303,
      "step": 152480
    },
    {
      "epoch": 62.42,
      "grad_norm": 1.7107163667678833,
      "learning_rate": 3.860903176436045e-06,
      "loss": 0.5203,
      "step": 152490
    },
    {
      "epoch": 62.42,
      "grad_norm": 1.8865206241607666,
      "learning_rate": 3.860782262430881e-06,
      "loss": 0.5308,
      "step": 152500
    },
    {
      "epoch": 62.43,
      "grad_norm": 2.313443899154663,
      "learning_rate": 3.860661341476428e-06,
      "loss": 0.538,
      "step": 152510
    },
    {
      "epoch": 62.43,
      "grad_norm": 1.712270975112915,
      "learning_rate": 3.860540413573238e-06,
      "loss": 0.5215,
      "step": 152520
    },
    {
      "epoch": 62.44,
      "grad_norm": 2.4015276432037354,
      "learning_rate": 3.860419478721867e-06,
      "loss": 0.5043,
      "step": 152530
    },
    {
      "epoch": 62.44,
      "grad_norm": 2.1336989402770996,
      "learning_rate": 3.860298536922867e-06,
      "loss": 0.5298,
      "step": 152540
    },
    {
      "epoch": 62.44,
      "grad_norm": 2.269120693206787,
      "learning_rate": 3.860177588176793e-06,
      "loss": 0.5276,
      "step": 152550
    },
    {
      "epoch": 62.45,
      "grad_norm": 2.1323280334472656,
      "learning_rate": 3.8600566324841995e-06,
      "loss": 0.5258,
      "step": 152560
    },
    {
      "epoch": 62.45,
      "grad_norm": 2.1778242588043213,
      "learning_rate": 3.85993566984564e-06,
      "loss": 0.5096,
      "step": 152570
    },
    {
      "epoch": 62.46,
      "grad_norm": 1.7719613313674927,
      "learning_rate": 3.859814700261669e-06,
      "loss": 0.505,
      "step": 152580
    },
    {
      "epoch": 62.46,
      "grad_norm": 2.269141674041748,
      "learning_rate": 3.85969372373284e-06,
      "loss": 0.5288,
      "step": 152590
    },
    {
      "epoch": 62.46,
      "grad_norm": 2.02099347114563,
      "learning_rate": 3.859572740259708e-06,
      "loss": 0.542,
      "step": 152600
    },
    {
      "epoch": 62.47,
      "grad_norm": 2.034489870071411,
      "learning_rate": 3.859451749842828e-06,
      "loss": 0.5366,
      "step": 152610
    },
    {
      "epoch": 62.47,
      "grad_norm": 1.9196324348449707,
      "learning_rate": 3.8593307524827506e-06,
      "loss": 0.5174,
      "step": 152620
    },
    {
      "epoch": 62.48,
      "grad_norm": 1.9371095895767212,
      "learning_rate": 3.859209748180034e-06,
      "loss": 0.5162,
      "step": 152630
    },
    {
      "epoch": 62.48,
      "grad_norm": 2.805856704711914,
      "learning_rate": 3.85908873693523e-06,
      "loss": 0.5199,
      "step": 152640
    },
    {
      "epoch": 62.48,
      "grad_norm": 2.378471612930298,
      "learning_rate": 3.858967718748895e-06,
      "loss": 0.5108,
      "step": 152650
    },
    {
      "epoch": 62.49,
      "grad_norm": 1.5900890827178955,
      "learning_rate": 3.858846693621582e-06,
      "loss": 0.5291,
      "step": 152660
    },
    {
      "epoch": 62.49,
      "grad_norm": 1.8926775455474854,
      "learning_rate": 3.858725661553845e-06,
      "loss": 0.5048,
      "step": 152670
    },
    {
      "epoch": 62.5,
      "grad_norm": 2.4184491634368896,
      "learning_rate": 3.8586046225462395e-06,
      "loss": 0.5297,
      "step": 152680
    },
    {
      "epoch": 62.5,
      "grad_norm": 1.7183283567428589,
      "learning_rate": 3.85848357659932e-06,
      "loss": 0.5105,
      "step": 152690
    },
    {
      "epoch": 62.51,
      "grad_norm": 2.1175947189331055,
      "learning_rate": 3.8583625237136395e-06,
      "loss": 0.5102,
      "step": 152700
    },
    {
      "epoch": 62.51,
      "grad_norm": 1.9735617637634277,
      "learning_rate": 3.858241463889755e-06,
      "loss": 0.513,
      "step": 152710
    },
    {
      "epoch": 62.51,
      "grad_norm": 2.3087756633758545,
      "learning_rate": 3.8581203971282185e-06,
      "loss": 0.5241,
      "step": 152720
    },
    {
      "epoch": 62.52,
      "grad_norm": 2.297300100326538,
      "learning_rate": 3.857999323429587e-06,
      "loss": 0.5322,
      "step": 152730
    },
    {
      "epoch": 62.52,
      "grad_norm": 1.9721741676330566,
      "learning_rate": 3.857878242794412e-06,
      "loss": 0.5254,
      "step": 152740
    },
    {
      "epoch": 62.53,
      "grad_norm": 3.179591417312622,
      "learning_rate": 3.857757155223252e-06,
      "loss": 0.5462,
      "step": 152750
    },
    {
      "epoch": 62.53,
      "grad_norm": 2.108252763748169,
      "learning_rate": 3.8576360607166586e-06,
      "loss": 0.5256,
      "step": 152760
    },
    {
      "epoch": 62.53,
      "grad_norm": 2.2372195720672607,
      "learning_rate": 3.857514959275187e-06,
      "loss": 0.544,
      "step": 152770
    },
    {
      "epoch": 62.54,
      "grad_norm": 2.263946056365967,
      "learning_rate": 3.857393850899394e-06,
      "loss": 0.5277,
      "step": 152780
    },
    {
      "epoch": 62.54,
      "grad_norm": 1.7459290027618408,
      "learning_rate": 3.8572727355898324e-06,
      "loss": 0.5314,
      "step": 152790
    },
    {
      "epoch": 62.55,
      "grad_norm": 1.6709874868392944,
      "learning_rate": 3.857151613347057e-06,
      "loss": 0.5211,
      "step": 152800
    },
    {
      "epoch": 62.55,
      "grad_norm": 1.8027143478393555,
      "learning_rate": 3.857030484171624e-06,
      "loss": 0.5075,
      "step": 152810
    },
    {
      "epoch": 62.55,
      "grad_norm": 2.055286407470703,
      "learning_rate": 3.856909348064087e-06,
      "loss": 0.5114,
      "step": 152820
    },
    {
      "epoch": 62.56,
      "grad_norm": 2.034414768218994,
      "learning_rate": 3.856788205025001e-06,
      "loss": 0.5211,
      "step": 152830
    },
    {
      "epoch": 62.56,
      "grad_norm": 2.373507022857666,
      "learning_rate": 3.856667055054922e-06,
      "loss": 0.5124,
      "step": 152840
    },
    {
      "epoch": 62.57,
      "grad_norm": 1.6696120500564575,
      "learning_rate": 3.8565458981544045e-06,
      "loss": 0.5101,
      "step": 152850
    },
    {
      "epoch": 62.57,
      "grad_norm": 1.8170456886291504,
      "learning_rate": 3.856424734324002e-06,
      "loss": 0.533,
      "step": 152860
    },
    {
      "epoch": 62.57,
      "grad_norm": 2.005957841873169,
      "learning_rate": 3.856303563564272e-06,
      "loss": 0.5073,
      "step": 152870
    },
    {
      "epoch": 62.58,
      "grad_norm": 1.948272466659546,
      "learning_rate": 3.856182385875768e-06,
      "loss": 0.5084,
      "step": 152880
    },
    {
      "epoch": 62.58,
      "grad_norm": 1.9397344589233398,
      "learning_rate": 3.856061201259046e-06,
      "loss": 0.5164,
      "step": 152890
    },
    {
      "epoch": 62.59,
      "grad_norm": 2.283458948135376,
      "learning_rate": 3.85594000971466e-06,
      "loss": 0.503,
      "step": 152900
    },
    {
      "epoch": 62.59,
      "grad_norm": 1.621848225593567,
      "learning_rate": 3.8558188112431655e-06,
      "loss": 0.5157,
      "step": 152910
    },
    {
      "epoch": 62.6,
      "grad_norm": 1.6048800945281982,
      "learning_rate": 3.8556976058451185e-06,
      "loss": 0.5058,
      "step": 152920
    },
    {
      "epoch": 62.6,
      "grad_norm": 2.1734092235565186,
      "learning_rate": 3.855576393521073e-06,
      "loss": 0.5358,
      "step": 152930
    },
    {
      "epoch": 62.6,
      "grad_norm": 2.043874979019165,
      "learning_rate": 3.855455174271585e-06,
      "loss": 0.5134,
      "step": 152940
    },
    {
      "epoch": 62.61,
      "grad_norm": 2.253553628921509,
      "learning_rate": 3.855333948097211e-06,
      "loss": 0.5484,
      "step": 152950
    },
    {
      "epoch": 62.61,
      "grad_norm": 1.8333956003189087,
      "learning_rate": 3.855212714998504e-06,
      "loss": 0.5105,
      "step": 152960
    },
    {
      "epoch": 62.62,
      "grad_norm": 2.4551310539245605,
      "learning_rate": 3.85509147497602e-06,
      "loss": 0.5249,
      "step": 152970
    },
    {
      "epoch": 62.62,
      "grad_norm": 1.5580192804336548,
      "learning_rate": 3.854970228030316e-06,
      "loss": 0.5045,
      "step": 152980
    },
    {
      "epoch": 62.62,
      "grad_norm": 3.4524295330047607,
      "learning_rate": 3.854848974161945e-06,
      "loss": 0.5348,
      "step": 152990
    },
    {
      "epoch": 62.63,
      "grad_norm": 1.8980329036712646,
      "learning_rate": 3.854727713371464e-06,
      "loss": 0.5269,
      "step": 153000
    },
    {
      "epoch": 62.63,
      "grad_norm": 1.5592045783996582,
      "learning_rate": 3.854606445659428e-06,
      "loss": 0.5239,
      "step": 153010
    },
    {
      "epoch": 62.64,
      "grad_norm": 2.021977186203003,
      "learning_rate": 3.854485171026393e-06,
      "loss": 0.5101,
      "step": 153020
    },
    {
      "epoch": 62.64,
      "grad_norm": 1.8011465072631836,
      "learning_rate": 3.854363889472914e-06,
      "loss": 0.5286,
      "step": 153030
    },
    {
      "epoch": 62.64,
      "grad_norm": 2.8269145488739014,
      "learning_rate": 3.854242600999547e-06,
      "loss": 0.5096,
      "step": 153040
    },
    {
      "epoch": 62.65,
      "grad_norm": 1.487863302230835,
      "learning_rate": 3.854121305606846e-06,
      "loss": 0.5291,
      "step": 153050
    },
    {
      "epoch": 62.65,
      "grad_norm": 1.8670555353164673,
      "learning_rate": 3.85400000329537e-06,
      "loss": 0.5227,
      "step": 153060
    },
    {
      "epoch": 62.66,
      "grad_norm": 2.0971810817718506,
      "learning_rate": 3.853878694065671e-06,
      "loss": 0.5196,
      "step": 153070
    },
    {
      "epoch": 62.66,
      "grad_norm": 1.7204021215438843,
      "learning_rate": 3.853757377918307e-06,
      "loss": 0.5181,
      "step": 153080
    },
    {
      "epoch": 62.66,
      "grad_norm": 1.7797824144363403,
      "learning_rate": 3.853636054853833e-06,
      "loss": 0.5252,
      "step": 153090
    },
    {
      "epoch": 62.67,
      "grad_norm": 2.030808925628662,
      "learning_rate": 3.853514724872805e-06,
      "loss": 0.5177,
      "step": 153100
    },
    {
      "epoch": 62.67,
      "grad_norm": 2.2973408699035645,
      "learning_rate": 3.853393387975778e-06,
      "loss": 0.5446,
      "step": 153110
    },
    {
      "epoch": 62.68,
      "grad_norm": 1.8853328227996826,
      "learning_rate": 3.853272044163309e-06,
      "loss": 0.5054,
      "step": 153120
    },
    {
      "epoch": 62.68,
      "grad_norm": 1.6929268836975098,
      "learning_rate": 3.853150693435953e-06,
      "loss": 0.5233,
      "step": 153130
    },
    {
      "epoch": 62.69,
      "grad_norm": 1.8260849714279175,
      "learning_rate": 3.853029335794266e-06,
      "loss": 0.5356,
      "step": 153140
    },
    {
      "epoch": 62.69,
      "grad_norm": 2.76318097114563,
      "learning_rate": 3.852907971238804e-06,
      "loss": 0.5332,
      "step": 153150
    },
    {
      "epoch": 62.69,
      "grad_norm": 2.819082736968994,
      "learning_rate": 3.852786599770123e-06,
      "loss": 0.5197,
      "step": 153160
    },
    {
      "epoch": 62.7,
      "grad_norm": 1.9261492490768433,
      "learning_rate": 3.85266522138878e-06,
      "loss": 0.5019,
      "step": 153170
    },
    {
      "epoch": 62.7,
      "grad_norm": 2.1732773780822754,
      "learning_rate": 3.8525438360953295e-06,
      "loss": 0.5272,
      "step": 153180
    },
    {
      "epoch": 62.71,
      "grad_norm": 2.361672878265381,
      "learning_rate": 3.852422443890329e-06,
      "loss": 0.5179,
      "step": 153190
    },
    {
      "epoch": 62.71,
      "grad_norm": 2.4236016273498535,
      "learning_rate": 3.8523010447743325e-06,
      "loss": 0.5311,
      "step": 153200
    },
    {
      "epoch": 62.71,
      "grad_norm": 2.4033761024475098,
      "learning_rate": 3.852179638747897e-06,
      "loss": 0.5072,
      "step": 153210
    },
    {
      "epoch": 62.72,
      "grad_norm": 2.013639450073242,
      "learning_rate": 3.85205822581158e-06,
      "loss": 0.5384,
      "step": 153220
    },
    {
      "epoch": 62.72,
      "grad_norm": 1.9955494403839111,
      "learning_rate": 3.851936805965936e-06,
      "loss": 0.5305,
      "step": 153230
    },
    {
      "epoch": 62.73,
      "grad_norm": 1.5225107669830322,
      "learning_rate": 3.851815379211522e-06,
      "loss": 0.5149,
      "step": 153240
    },
    {
      "epoch": 62.73,
      "grad_norm": 2.616730213165283,
      "learning_rate": 3.8516939455488946e-06,
      "loss": 0.5235,
      "step": 153250
    },
    {
      "epoch": 62.73,
      "grad_norm": 1.976671576499939,
      "learning_rate": 3.851572504978608e-06,
      "loss": 0.5204,
      "step": 153260
    },
    {
      "epoch": 62.74,
      "grad_norm": 2.2143805027008057,
      "learning_rate": 3.851451057501222e-06,
      "loss": 0.5257,
      "step": 153270
    },
    {
      "epoch": 62.74,
      "grad_norm": 2.8792340755462646,
      "learning_rate": 3.851329603117289e-06,
      "loss": 0.5394,
      "step": 153280
    },
    {
      "epoch": 62.75,
      "grad_norm": 2.212501049041748,
      "learning_rate": 3.851208141827369e-06,
      "loss": 0.5284,
      "step": 153290
    },
    {
      "epoch": 62.75,
      "grad_norm": 2.2805516719818115,
      "learning_rate": 3.851086673632015e-06,
      "loss": 0.5415,
      "step": 153300
    },
    {
      "epoch": 62.75,
      "grad_norm": 1.7109800577163696,
      "learning_rate": 3.850965198531787e-06,
      "loss": 0.5324,
      "step": 153310
    },
    {
      "epoch": 62.76,
      "grad_norm": 1.915814757347107,
      "learning_rate": 3.8508437165272384e-06,
      "loss": 0.5205,
      "step": 153320
    },
    {
      "epoch": 62.76,
      "grad_norm": 1.8345024585723877,
      "learning_rate": 3.850722227618927e-06,
      "loss": 0.5211,
      "step": 153330
    },
    {
      "epoch": 62.77,
      "grad_norm": 1.9898796081542969,
      "learning_rate": 3.850600731807409e-06,
      "loss": 0.5371,
      "step": 153340
    },
    {
      "epoch": 62.77,
      "grad_norm": 1.8161448240280151,
      "learning_rate": 3.850479229093242e-06,
      "loss": 0.5068,
      "step": 153350
    },
    {
      "epoch": 62.78,
      "grad_norm": 1.7139427661895752,
      "learning_rate": 3.850357719476981e-06,
      "loss": 0.5136,
      "step": 153360
    },
    {
      "epoch": 62.78,
      "grad_norm": 2.219841480255127,
      "learning_rate": 3.850236202959184e-06,
      "loss": 0.5291,
      "step": 153370
    },
    {
      "epoch": 62.78,
      "grad_norm": 1.5543795824050903,
      "learning_rate": 3.850114679540407e-06,
      "loss": 0.5092,
      "step": 153380
    },
    {
      "epoch": 62.79,
      "grad_norm": 1.6054044961929321,
      "learning_rate": 3.849993149221206e-06,
      "loss": 0.5179,
      "step": 153390
    },
    {
      "epoch": 62.79,
      "grad_norm": 2.1606369018554688,
      "learning_rate": 3.849871612002139e-06,
      "loss": 0.5191,
      "step": 153400
    },
    {
      "epoch": 62.8,
      "grad_norm": 1.770628809928894,
      "learning_rate": 3.849750067883761e-06,
      "loss": 0.529,
      "step": 153410
    },
    {
      "epoch": 62.8,
      "grad_norm": 1.8861695528030396,
      "learning_rate": 3.849628516866631e-06,
      "loss": 0.5108,
      "step": 153420
    },
    {
      "epoch": 62.8,
      "grad_norm": 2.1386096477508545,
      "learning_rate": 3.849506958951305e-06,
      "loss": 0.5234,
      "step": 153430
    },
    {
      "epoch": 62.81,
      "grad_norm": 1.881076455116272,
      "learning_rate": 3.849385394138339e-06,
      "loss": 0.5066,
      "step": 153440
    },
    {
      "epoch": 62.81,
      "grad_norm": 2.050318717956543,
      "learning_rate": 3.84926382242829e-06,
      "loss": 0.5206,
      "step": 153450
    },
    {
      "epoch": 62.82,
      "grad_norm": 1.8267072439193726,
      "learning_rate": 3.849142243821717e-06,
      "loss": 0.5207,
      "step": 153460
    },
    {
      "epoch": 62.82,
      "grad_norm": 1.886654019355774,
      "learning_rate": 3.849020658319174e-06,
      "loss": 0.5266,
      "step": 153470
    },
    {
      "epoch": 62.82,
      "grad_norm": 1.6426925659179688,
      "learning_rate": 3.8488990659212195e-06,
      "loss": 0.5035,
      "step": 153480
    },
    {
      "epoch": 62.83,
      "grad_norm": 1.6848313808441162,
      "learning_rate": 3.84877746662841e-06,
      "loss": 0.5179,
      "step": 153490
    },
    {
      "epoch": 62.83,
      "grad_norm": 2.2070162296295166,
      "learning_rate": 3.848655860441303e-06,
      "loss": 0.5321,
      "step": 153500
    },
    {
      "epoch": 62.84,
      "grad_norm": 1.8147435188293457,
      "learning_rate": 3.848534247360455e-06,
      "loss": 0.5481,
      "step": 153510
    },
    {
      "epoch": 62.84,
      "grad_norm": 1.7589521408081055,
      "learning_rate": 3.848412627386424e-06,
      "loss": 0.5098,
      "step": 153520
    },
    {
      "epoch": 62.84,
      "grad_norm": 2.092207908630371,
      "learning_rate": 3.848291000519765e-06,
      "loss": 0.5354,
      "step": 153530
    },
    {
      "epoch": 62.85,
      "grad_norm": 1.571529507637024,
      "learning_rate": 3.848169366761039e-06,
      "loss": 0.5285,
      "step": 153540
    },
    {
      "epoch": 62.85,
      "grad_norm": 2.245805501937866,
      "learning_rate": 3.8480477261107995e-06,
      "loss": 0.5247,
      "step": 153550
    },
    {
      "epoch": 62.86,
      "grad_norm": 1.4933196306228638,
      "learning_rate": 3.847926078569605e-06,
      "loss": 0.5213,
      "step": 153560
    },
    {
      "epoch": 62.86,
      "grad_norm": 2.372983455657959,
      "learning_rate": 3.847804424138013e-06,
      "loss": 0.5149,
      "step": 153570
    },
    {
      "epoch": 62.87,
      "grad_norm": 2.056072473526001,
      "learning_rate": 3.847682762816582e-06,
      "loss": 0.5205,
      "step": 153580
    },
    {
      "epoch": 62.87,
      "grad_norm": 2.224860906600952,
      "learning_rate": 3.847561094605866e-06,
      "loss": 0.5333,
      "step": 153590
    },
    {
      "epoch": 62.87,
      "grad_norm": 1.8365356922149658,
      "learning_rate": 3.847439419506425e-06,
      "loss": 0.5252,
      "step": 153600
    },
    {
      "epoch": 62.88,
      "grad_norm": 1.5912280082702637,
      "learning_rate": 3.847317737518815e-06,
      "loss": 0.5175,
      "step": 153610
    },
    {
      "epoch": 62.88,
      "grad_norm": 1.6756290197372437,
      "learning_rate": 3.8471960486435945e-06,
      "loss": 0.5036,
      "step": 153620
    },
    {
      "epoch": 62.89,
      "grad_norm": 1.7708929777145386,
      "learning_rate": 3.847074352881321e-06,
      "loss": 0.5366,
      "step": 153630
    },
    {
      "epoch": 62.89,
      "grad_norm": 1.8581264019012451,
      "learning_rate": 3.846952650232551e-06,
      "loss": 0.5167,
      "step": 153640
    },
    {
      "epoch": 62.89,
      "grad_norm": 2.014554262161255,
      "learning_rate": 3.846830940697843e-06,
      "loss": 0.5574,
      "step": 153650
    },
    {
      "epoch": 62.9,
      "grad_norm": 1.7698471546173096,
      "learning_rate": 3.8467092242777535e-06,
      "loss": 0.536,
      "step": 153660
    },
    {
      "epoch": 62.9,
      "grad_norm": 2.1276628971099854,
      "learning_rate": 3.84658750097284e-06,
      "loss": 0.5315,
      "step": 153670
    },
    {
      "epoch": 62.91,
      "grad_norm": 1.9627842903137207,
      "learning_rate": 3.846465770783661e-06,
      "loss": 0.513,
      "step": 153680
    },
    {
      "epoch": 62.91,
      "grad_norm": 1.706617832183838,
      "learning_rate": 3.846344033710774e-06,
      "loss": 0.5043,
      "step": 153690
    },
    {
      "epoch": 62.91,
      "grad_norm": 1.544722318649292,
      "learning_rate": 3.846222289754736e-06,
      "loss": 0.5408,
      "step": 153700
    },
    {
      "epoch": 62.92,
      "grad_norm": 1.604065179824829,
      "learning_rate": 3.846100538916106e-06,
      "loss": 0.516,
      "step": 153710
    },
    {
      "epoch": 62.92,
      "grad_norm": 2.3161981105804443,
      "learning_rate": 3.845978781195441e-06,
      "loss": 0.5168,
      "step": 153720
    },
    {
      "epoch": 62.93,
      "grad_norm": 1.5279181003570557,
      "learning_rate": 3.845857016593298e-06,
      "loss": 0.5083,
      "step": 153730
    },
    {
      "epoch": 62.93,
      "grad_norm": 1.6874017715454102,
      "learning_rate": 3.8457352451102355e-06,
      "loss": 0.5135,
      "step": 153740
    },
    {
      "epoch": 62.93,
      "grad_norm": 2.704735517501831,
      "learning_rate": 3.845613466746811e-06,
      "loss": 0.5133,
      "step": 153750
    },
    {
      "epoch": 62.94,
      "grad_norm": 1.9571069478988647,
      "learning_rate": 3.845491681503583e-06,
      "loss": 0.5164,
      "step": 153760
    },
    {
      "epoch": 62.94,
      "grad_norm": 1.765388011932373,
      "learning_rate": 3.845369889381107e-06,
      "loss": 0.5154,
      "step": 153770
    },
    {
      "epoch": 62.95,
      "grad_norm": 1.896937608718872,
      "learning_rate": 3.845248090379946e-06,
      "loss": 0.5155,
      "step": 153780
    },
    {
      "epoch": 62.95,
      "grad_norm": 2.0234415531158447,
      "learning_rate": 3.845126284500653e-06,
      "loss": 0.5346,
      "step": 153790
    },
    {
      "epoch": 62.96,
      "grad_norm": 1.9686203002929688,
      "learning_rate": 3.8450044717437875e-06,
      "loss": 0.5158,
      "step": 153800
    },
    {
      "epoch": 62.96,
      "grad_norm": 1.6036406755447388,
      "learning_rate": 3.844882652109908e-06,
      "loss": 0.5194,
      "step": 153810
    },
    {
      "epoch": 62.96,
      "grad_norm": 2.351006031036377,
      "learning_rate": 3.844760825599574e-06,
      "loss": 0.5408,
      "step": 153820
    },
    {
      "epoch": 62.97,
      "grad_norm": 1.5948584079742432,
      "learning_rate": 3.844638992213339e-06,
      "loss": 0.5223,
      "step": 153830
    },
    {
      "epoch": 62.97,
      "grad_norm": 2.150912284851074,
      "learning_rate": 3.8445171519517655e-06,
      "loss": 0.5154,
      "step": 153840
    },
    {
      "epoch": 62.98,
      "grad_norm": 1.7975856065750122,
      "learning_rate": 3.8443953048154106e-06,
      "loss": 0.5217,
      "step": 153850
    },
    {
      "epoch": 62.98,
      "grad_norm": 2.541957378387451,
      "learning_rate": 3.844273450804832e-06,
      "loss": 0.5313,
      "step": 153860
    },
    {
      "epoch": 62.98,
      "grad_norm": 2.2061898708343506,
      "learning_rate": 3.844151589920587e-06,
      "loss": 0.5146,
      "step": 153870
    },
    {
      "epoch": 62.99,
      "grad_norm": 2.503051996231079,
      "learning_rate": 3.844029722163234e-06,
      "loss": 0.5159,
      "step": 153880
    },
    {
      "epoch": 62.99,
      "grad_norm": 2.0023229122161865,
      "learning_rate": 3.8439078475333325e-06,
      "loss": 0.5249,
      "step": 153890
    },
    {
      "epoch": 63.0,
      "grad_norm": 2.458746910095215,
      "learning_rate": 3.8437859660314415e-06,
      "loss": 0.5243,
      "step": 153900
    },
    {
      "epoch": 63.0,
      "eval_loss": 0.5213854312896729,
      "eval_runtime": 52.0975,
      "eval_samples_per_second": 66.203,
      "eval_steps_per_second": 8.292,
      "step": 153909
    },
    {
      "epoch": 63.0,
      "grad_norm": 2.3028454780578613,
      "learning_rate": 3.843664077658117e-06,
      "loss": 0.5301,
      "step": 153910
    },
    {
      "epoch": 63.0,
      "grad_norm": 2.1404635906219482,
      "learning_rate": 3.843542182413918e-06,
      "loss": 0.5422,
      "step": 153920
    },
    {
      "epoch": 63.01,
      "grad_norm": 1.9652609825134277,
      "learning_rate": 3.843420280299404e-06,
      "loss": 0.5196,
      "step": 153930
    },
    {
      "epoch": 63.01,
      "grad_norm": 2.4163002967834473,
      "learning_rate": 3.843298371315133e-06,
      "loss": 0.5557,
      "step": 153940
    },
    {
      "epoch": 63.02,
      "grad_norm": 1.6972812414169312,
      "learning_rate": 3.843176455461662e-06,
      "loss": 0.5143,
      "step": 153950
    },
    {
      "epoch": 63.02,
      "grad_norm": 2.498721122741699,
      "learning_rate": 3.843054532739552e-06,
      "loss": 0.5302,
      "step": 153960
    },
    {
      "epoch": 63.02,
      "grad_norm": 1.794740915298462,
      "learning_rate": 3.842932603149359e-06,
      "loss": 0.5177,
      "step": 153970
    },
    {
      "epoch": 63.03,
      "grad_norm": 1.579284429550171,
      "learning_rate": 3.8428106666916424e-06,
      "loss": 0.5233,
      "step": 153980
    },
    {
      "epoch": 63.03,
      "grad_norm": 2.029716968536377,
      "learning_rate": 3.842688723366962e-06,
      "loss": 0.5367,
      "step": 153990
    },
    {
      "epoch": 63.04,
      "grad_norm": 2.1959171295166016,
      "learning_rate": 3.842566773175875e-06,
      "loss": 0.5264,
      "step": 154000
    },
    {
      "epoch": 63.04,
      "grad_norm": 1.5788792371749878,
      "learning_rate": 3.842444816118942e-06,
      "loss": 0.5119,
      "step": 154010
    },
    {
      "epoch": 63.05,
      "grad_norm": 2.0667192935943604,
      "learning_rate": 3.842322852196718e-06,
      "loss": 0.5192,
      "step": 154020
    },
    {
      "epoch": 63.05,
      "grad_norm": 2.0939767360687256,
      "learning_rate": 3.842200881409764e-06,
      "loss": 0.5384,
      "step": 154030
    },
    {
      "epoch": 63.05,
      "grad_norm": 2.552046537399292,
      "learning_rate": 3.842078903758639e-06,
      "loss": 0.5131,
      "step": 154040
    },
    {
      "epoch": 63.06,
      "grad_norm": 1.8587696552276611,
      "learning_rate": 3.841956919243902e-06,
      "loss": 0.5169,
      "step": 154050
    },
    {
      "epoch": 63.06,
      "grad_norm": 1.7978720664978027,
      "learning_rate": 3.841834927866111e-06,
      "loss": 0.5179,
      "step": 154060
    },
    {
      "epoch": 63.07,
      "grad_norm": 2.224349021911621,
      "learning_rate": 3.841712929625825e-06,
      "loss": 0.4982,
      "step": 154070
    },
    {
      "epoch": 63.07,
      "grad_norm": 2.0182995796203613,
      "learning_rate": 3.841590924523602e-06,
      "loss": 0.5351,
      "step": 154080
    },
    {
      "epoch": 63.07,
      "grad_norm": 2.0486016273498535,
      "learning_rate": 3.8414689125600026e-06,
      "loss": 0.516,
      "step": 154090
    },
    {
      "epoch": 63.08,
      "grad_norm": 2.330697536468506,
      "learning_rate": 3.841346893735584e-06,
      "loss": 0.5178,
      "step": 154100
    },
    {
      "epoch": 63.08,
      "grad_norm": 2.3463191986083984,
      "learning_rate": 3.8412248680509064e-06,
      "loss": 0.5388,
      "step": 154110
    },
    {
      "epoch": 63.09,
      "grad_norm": 1.8573769330978394,
      "learning_rate": 3.841102835506529e-06,
      "loss": 0.5232,
      "step": 154120
    },
    {
      "epoch": 63.09,
      "grad_norm": 3.1497957706451416,
      "learning_rate": 3.840980796103009e-06,
      "loss": 0.5302,
      "step": 154130
    },
    {
      "epoch": 63.09,
      "grad_norm": 1.592226505279541,
      "learning_rate": 3.8408587498409065e-06,
      "loss": 0.5038,
      "step": 154140
    },
    {
      "epoch": 63.1,
      "grad_norm": 2.0424652099609375,
      "learning_rate": 3.8407366967207815e-06,
      "loss": 0.5351,
      "step": 154150
    },
    {
      "epoch": 63.1,
      "grad_norm": 2.030907154083252,
      "learning_rate": 3.840614636743192e-06,
      "loss": 0.5369,
      "step": 154160
    },
    {
      "epoch": 63.11,
      "grad_norm": 2.105900764465332,
      "learning_rate": 3.8404925699086965e-06,
      "loss": 0.5253,
      "step": 154170
    },
    {
      "epoch": 63.11,
      "grad_norm": 1.6907026767730713,
      "learning_rate": 3.840370496217856e-06,
      "loss": 0.5359,
      "step": 154180
    },
    {
      "epoch": 63.12,
      "grad_norm": 1.578169584274292,
      "learning_rate": 3.84024841567123e-06,
      "loss": 0.5306,
      "step": 154190
    },
    {
      "epoch": 63.12,
      "grad_norm": 1.9190820455551147,
      "learning_rate": 3.840126328269375e-06,
      "loss": 0.5277,
      "step": 154200
    },
    {
      "epoch": 63.12,
      "grad_norm": 1.702454686164856,
      "learning_rate": 3.8400042340128524e-06,
      "loss": 0.5001,
      "step": 154210
    },
    {
      "epoch": 63.13,
      "grad_norm": 1.8119860887527466,
      "learning_rate": 3.83988213290222e-06,
      "loss": 0.5249,
      "step": 154220
    },
    {
      "epoch": 63.13,
      "grad_norm": 2.1994571685791016,
      "learning_rate": 3.839760024938039e-06,
      "loss": 0.523,
      "step": 154230
    },
    {
      "epoch": 63.14,
      "grad_norm": 2.165076971054077,
      "learning_rate": 3.8396379101208675e-06,
      "loss": 0.5081,
      "step": 154240
    },
    {
      "epoch": 63.14,
      "grad_norm": 2.3863885402679443,
      "learning_rate": 3.839515788451264e-06,
      "loss": 0.5406,
      "step": 154250
    },
    {
      "epoch": 63.14,
      "grad_norm": 2.3831849098205566,
      "learning_rate": 3.839393659929791e-06,
      "loss": 0.5273,
      "step": 154260
    },
    {
      "epoch": 63.15,
      "grad_norm": 2.0297489166259766,
      "learning_rate": 3.839271524557004e-06,
      "loss": 0.5299,
      "step": 154270
    },
    {
      "epoch": 63.15,
      "grad_norm": 2.3090097904205322,
      "learning_rate": 3.839149382333466e-06,
      "loss": 0.5095,
      "step": 154280
    },
    {
      "epoch": 63.16,
      "grad_norm": 1.6854894161224365,
      "learning_rate": 3.839027233259734e-06,
      "loss": 0.5283,
      "step": 154290
    },
    {
      "epoch": 63.16,
      "grad_norm": 2.3792779445648193,
      "learning_rate": 3.83890507733637e-06,
      "loss": 0.4735,
      "step": 154300
    },
    {
      "epoch": 63.16,
      "grad_norm": 1.5550364255905151,
      "learning_rate": 3.838782914563931e-06,
      "loss": 0.5388,
      "step": 154310
    },
    {
      "epoch": 63.17,
      "grad_norm": 2.122439384460449,
      "learning_rate": 3.838660744942978e-06,
      "loss": 0.5208,
      "step": 154320
    },
    {
      "epoch": 63.17,
      "grad_norm": 2.3252716064453125,
      "learning_rate": 3.838538568474071e-06,
      "loss": 0.5399,
      "step": 154330
    },
    {
      "epoch": 63.18,
      "grad_norm": 1.7349647283554077,
      "learning_rate": 3.8384163851577675e-06,
      "loss": 0.5567,
      "step": 154340
    },
    {
      "epoch": 63.18,
      "grad_norm": 2.269141674041748,
      "learning_rate": 3.83829419499463e-06,
      "loss": 0.5032,
      "step": 154350
    },
    {
      "epoch": 63.18,
      "grad_norm": 1.6489853858947754,
      "learning_rate": 3.838171997985217e-06,
      "loss": 0.5424,
      "step": 154360
    },
    {
      "epoch": 63.19,
      "grad_norm": 1.9372787475585938,
      "learning_rate": 3.838049794130088e-06,
      "loss": 0.5327,
      "step": 154370
    },
    {
      "epoch": 63.19,
      "grad_norm": 2.2754135131835938,
      "learning_rate": 3.837927583429803e-06,
      "loss": 0.5365,
      "step": 154380
    },
    {
      "epoch": 63.2,
      "grad_norm": 2.123208522796631,
      "learning_rate": 3.837805365884922e-06,
      "loss": 0.5351,
      "step": 154390
    },
    {
      "epoch": 63.2,
      "grad_norm": 1.6010267734527588,
      "learning_rate": 3.837683141496005e-06,
      "loss": 0.5157,
      "step": 154400
    },
    {
      "epoch": 63.21,
      "grad_norm": 2.1367335319519043,
      "learning_rate": 3.837560910263611e-06,
      "loss": 0.5242,
      "step": 154410
    },
    {
      "epoch": 63.21,
      "grad_norm": 1.5892281532287598,
      "learning_rate": 3.837438672188302e-06,
      "loss": 0.5142,
      "step": 154420
    },
    {
      "epoch": 63.21,
      "grad_norm": 1.7886559963226318,
      "learning_rate": 3.837316427270634e-06,
      "loss": 0.5264,
      "step": 154430
    },
    {
      "epoch": 63.22,
      "grad_norm": 1.6137055158615112,
      "learning_rate": 3.837194175511172e-06,
      "loss": 0.5152,
      "step": 154440
    },
    {
      "epoch": 63.22,
      "grad_norm": 2.087533473968506,
      "learning_rate": 3.837071916910472e-06,
      "loss": 0.5209,
      "step": 154450
    },
    {
      "epoch": 63.23,
      "grad_norm": 1.8312593698501587,
      "learning_rate": 3.8369496514690964e-06,
      "loss": 0.528,
      "step": 154460
    },
    {
      "epoch": 63.23,
      "grad_norm": 1.767499566078186,
      "learning_rate": 3.836827379187604e-06,
      "loss": 0.5286,
      "step": 154470
    },
    {
      "epoch": 63.23,
      "grad_norm": 2.1063177585601807,
      "learning_rate": 3.836705100066556e-06,
      "loss": 0.5212,
      "step": 154480
    },
    {
      "epoch": 63.24,
      "grad_norm": 2.242258071899414,
      "learning_rate": 3.836582814106511e-06,
      "loss": 0.5093,
      "step": 154490
    },
    {
      "epoch": 63.24,
      "grad_norm": 2.5361998081207275,
      "learning_rate": 3.836460521308031e-06,
      "loss": 0.5243,
      "step": 154500
    },
    {
      "epoch": 63.25,
      "grad_norm": 1.8862754106521606,
      "learning_rate": 3.836338221671674e-06,
      "loss": 0.529,
      "step": 154510
    },
    {
      "epoch": 63.25,
      "grad_norm": 2.365825891494751,
      "learning_rate": 3.8362159151980025e-06,
      "loss": 0.5208,
      "step": 154520
    },
    {
      "epoch": 63.25,
      "grad_norm": 2.063392400741577,
      "learning_rate": 3.836093601887576e-06,
      "loss": 0.5375,
      "step": 154530
    },
    {
      "epoch": 63.26,
      "grad_norm": 1.729468822479248,
      "learning_rate": 3.835971281740954e-06,
      "loss": 0.5246,
      "step": 154540
    },
    {
      "epoch": 63.26,
      "grad_norm": 2.288893222808838,
      "learning_rate": 3.835848954758697e-06,
      "loss": 0.5079,
      "step": 154550
    },
    {
      "epoch": 63.27,
      "grad_norm": 2.1453423500061035,
      "learning_rate": 3.835726620941366e-06,
      "loss": 0.5035,
      "step": 154560
    },
    {
      "epoch": 63.27,
      "grad_norm": 1.8985414505004883,
      "learning_rate": 3.835604280289521e-06,
      "loss": 0.5335,
      "step": 154570
    },
    {
      "epoch": 63.27,
      "grad_norm": 1.7659515142440796,
      "learning_rate": 3.835481932803723e-06,
      "loss": 0.5226,
      "step": 154580
    },
    {
      "epoch": 63.28,
      "grad_norm": 2.6953957080841064,
      "learning_rate": 3.835359578484533e-06,
      "loss": 0.5192,
      "step": 154590
    },
    {
      "epoch": 63.28,
      "grad_norm": 1.9710198640823364,
      "learning_rate": 3.8352372173325095e-06,
      "loss": 0.5339,
      "step": 154600
    },
    {
      "epoch": 63.29,
      "grad_norm": 2.048135757446289,
      "learning_rate": 3.835114849348214e-06,
      "loss": 0.5444,
      "step": 154610
    },
    {
      "epoch": 63.29,
      "grad_norm": 1.7210663557052612,
      "learning_rate": 3.834992474532207e-06,
      "loss": 0.5189,
      "step": 154620
    },
    {
      "epoch": 63.3,
      "grad_norm": 2.0363218784332275,
      "learning_rate": 3.834870092885049e-06,
      "loss": 0.5284,
      "step": 154630
    },
    {
      "epoch": 63.3,
      "grad_norm": 2.5939149856567383,
      "learning_rate": 3.834747704407302e-06,
      "loss": 0.5094,
      "step": 154640
    },
    {
      "epoch": 63.3,
      "grad_norm": 1.9840617179870605,
      "learning_rate": 3.8346253090995246e-06,
      "loss": 0.5376,
      "step": 154650
    },
    {
      "epoch": 63.31,
      "grad_norm": 2.115389108657837,
      "learning_rate": 3.834502906962278e-06,
      "loss": 0.5317,
      "step": 154660
    },
    {
      "epoch": 63.31,
      "grad_norm": 2.1627442836761475,
      "learning_rate": 3.834380497996124e-06,
      "loss": 0.5144,
      "step": 154670
    },
    {
      "epoch": 63.32,
      "grad_norm": 1.8907686471939087,
      "learning_rate": 3.834258082201622e-06,
      "loss": 0.5094,
      "step": 154680
    },
    {
      "epoch": 63.32,
      "grad_norm": 1.977022409439087,
      "learning_rate": 3.834135659579333e-06,
      "loss": 0.5079,
      "step": 154690
    },
    {
      "epoch": 63.32,
      "grad_norm": 2.1422717571258545,
      "learning_rate": 3.834013230129819e-06,
      "loss": 0.5424,
      "step": 154700
    },
    {
      "epoch": 63.33,
      "grad_norm": 2.1456708908081055,
      "learning_rate": 3.83389079385364e-06,
      "loss": 0.5014,
      "step": 154710
    },
    {
      "epoch": 63.33,
      "grad_norm": 1.3851913213729858,
      "learning_rate": 3.833768350751355e-06,
      "loss": 0.5356,
      "step": 154720
    },
    {
      "epoch": 63.34,
      "grad_norm": 1.7585424184799194,
      "learning_rate": 3.833645900823529e-06,
      "loss": 0.5138,
      "step": 154730
    },
    {
      "epoch": 63.34,
      "grad_norm": 2.7179224491119385,
      "learning_rate": 3.8335234440707185e-06,
      "loss": 0.513,
      "step": 154740
    },
    {
      "epoch": 63.34,
      "grad_norm": 1.7472459077835083,
      "learning_rate": 3.8334009804934875e-06,
      "loss": 0.5135,
      "step": 154750
    },
    {
      "epoch": 63.35,
      "grad_norm": 1.72080397605896,
      "learning_rate": 3.8332785100923964e-06,
      "loss": 0.5081,
      "step": 154760
    },
    {
      "epoch": 63.35,
      "grad_norm": 1.4066303968429565,
      "learning_rate": 3.833156032868005e-06,
      "loss": 0.526,
      "step": 154770
    },
    {
      "epoch": 63.36,
      "grad_norm": 1.5516793727874756,
      "learning_rate": 3.833033548820877e-06,
      "loss": 0.5197,
      "step": 154780
    },
    {
      "epoch": 63.36,
      "grad_norm": 1.82698392868042,
      "learning_rate": 3.83291105795157e-06,
      "loss": 0.5396,
      "step": 154790
    },
    {
      "epoch": 63.36,
      "grad_norm": 2.264711618423462,
      "learning_rate": 3.832788560260648e-06,
      "loss": 0.5063,
      "step": 154800
    },
    {
      "epoch": 63.37,
      "grad_norm": 1.713713526725769,
      "learning_rate": 3.8326660557486695e-06,
      "loss": 0.5132,
      "step": 154810
    },
    {
      "epoch": 63.37,
      "grad_norm": 2.5454366207122803,
      "learning_rate": 3.832543544416198e-06,
      "loss": 0.5132,
      "step": 154820
    },
    {
      "epoch": 63.38,
      "grad_norm": 1.602426528930664,
      "learning_rate": 3.832421026263795e-06,
      "loss": 0.5105,
      "step": 154830
    },
    {
      "epoch": 63.38,
      "grad_norm": 2.8643150329589844,
      "learning_rate": 3.832298501292019e-06,
      "loss": 0.5218,
      "step": 154840
    },
    {
      "epoch": 63.39,
      "grad_norm": 2.1145718097686768,
      "learning_rate": 3.832175969501432e-06,
      "loss": 0.5001,
      "step": 154850
    },
    {
      "epoch": 63.39,
      "grad_norm": 2.252556085586548,
      "learning_rate": 3.832053430892598e-06,
      "loss": 0.5114,
      "step": 154860
    },
    {
      "epoch": 63.39,
      "grad_norm": 2.399937391281128,
      "learning_rate": 3.831930885466076e-06,
      "loss": 0.5066,
      "step": 154870
    },
    {
      "epoch": 63.4,
      "grad_norm": 1.9446866512298584,
      "learning_rate": 3.831808333222428e-06,
      "loss": 0.5268,
      "step": 154880
    },
    {
      "epoch": 63.4,
      "grad_norm": 2.122464418411255,
      "learning_rate": 3.831685774162215e-06,
      "loss": 0.5184,
      "step": 154890
    },
    {
      "epoch": 63.41,
      "grad_norm": 2.083463191986084,
      "learning_rate": 3.831563208285999e-06,
      "loss": 0.5347,
      "step": 154900
    },
    {
      "epoch": 63.41,
      "grad_norm": 2.463449239730835,
      "learning_rate": 3.83144063559434e-06,
      "loss": 0.4931,
      "step": 154910
    },
    {
      "epoch": 63.41,
      "grad_norm": 1.646995186805725,
      "learning_rate": 3.831318056087801e-06,
      "loss": 0.5446,
      "step": 154920
    },
    {
      "epoch": 63.42,
      "grad_norm": 1.8385868072509766,
      "learning_rate": 3.831195469766944e-06,
      "loss": 0.5444,
      "step": 154930
    },
    {
      "epoch": 63.42,
      "grad_norm": 1.808009147644043,
      "learning_rate": 3.831072876632329e-06,
      "loss": 0.5452,
      "step": 154940
    },
    {
      "epoch": 63.43,
      "grad_norm": 2.133786201477051,
      "learning_rate": 3.830950276684518e-06,
      "loss": 0.515,
      "step": 154950
    },
    {
      "epoch": 63.43,
      "grad_norm": 2.2418642044067383,
      "learning_rate": 3.830827669924074e-06,
      "loss": 0.5357,
      "step": 154960
    },
    {
      "epoch": 63.43,
      "grad_norm": 2.014094829559326,
      "learning_rate": 3.830705056351556e-06,
      "loss": 0.5276,
      "step": 154970
    },
    {
      "epoch": 63.44,
      "grad_norm": 2.023563861846924,
      "learning_rate": 3.830582435967528e-06,
      "loss": 0.541,
      "step": 154980
    },
    {
      "epoch": 63.44,
      "grad_norm": 1.9492883682250977,
      "learning_rate": 3.830459808772551e-06,
      "loss": 0.5353,
      "step": 154990
    },
    {
      "epoch": 63.45,
      "grad_norm": 2.1130402088165283,
      "learning_rate": 3.8303371747671865e-06,
      "loss": 0.5178,
      "step": 155000
    },
    {
      "epoch": 63.45,
      "grad_norm": 1.711883783340454,
      "learning_rate": 3.830214533951995e-06,
      "loss": 0.5204,
      "step": 155010
    },
    {
      "epoch": 63.45,
      "grad_norm": 2.0820584297180176,
      "learning_rate": 3.830091886327541e-06,
      "loss": 0.5333,
      "step": 155020
    },
    {
      "epoch": 63.46,
      "grad_norm": 1.837984561920166,
      "learning_rate": 3.829969231894386e-06,
      "loss": 0.5133,
      "step": 155030
    },
    {
      "epoch": 63.46,
      "grad_norm": 2.2091450691223145,
      "learning_rate": 3.829846570653089e-06,
      "loss": 0.5301,
      "step": 155040
    },
    {
      "epoch": 63.47,
      "grad_norm": 2.0613021850585938,
      "learning_rate": 3.8297239026042145e-06,
      "loss": 0.5326,
      "step": 155050
    },
    {
      "epoch": 63.47,
      "grad_norm": 1.8615461587905884,
      "learning_rate": 3.829601227748324e-06,
      "loss": 0.5311,
      "step": 155060
    },
    {
      "epoch": 63.48,
      "grad_norm": 1.7959221601486206,
      "learning_rate": 3.829478546085979e-06,
      "loss": 0.5326,
      "step": 155070
    },
    {
      "epoch": 63.48,
      "grad_norm": 1.615783929824829,
      "learning_rate": 3.8293558576177415e-06,
      "loss": 0.53,
      "step": 155080
    },
    {
      "epoch": 63.48,
      "grad_norm": 1.6013422012329102,
      "learning_rate": 3.8292331623441735e-06,
      "loss": 0.517,
      "step": 155090
    },
    {
      "epoch": 63.49,
      "grad_norm": 2.265652894973755,
      "learning_rate": 3.829110460265838e-06,
      "loss": 0.5255,
      "step": 155100
    },
    {
      "epoch": 63.49,
      "grad_norm": 1.9415132999420166,
      "learning_rate": 3.8289877513832945e-06,
      "loss": 0.5091,
      "step": 155110
    },
    {
      "epoch": 63.5,
      "grad_norm": 1.9366763830184937,
      "learning_rate": 3.828865035697108e-06,
      "loss": 0.5053,
      "step": 155120
    },
    {
      "epoch": 63.5,
      "grad_norm": 1.828979253768921,
      "learning_rate": 3.82874231320784e-06,
      "loss": 0.5306,
      "step": 155130
    },
    {
      "epoch": 63.5,
      "grad_norm": 1.702422022819519,
      "learning_rate": 3.828619583916051e-06,
      "loss": 0.5026,
      "step": 155140
    },
    {
      "epoch": 63.51,
      "grad_norm": 1.7762246131896973,
      "learning_rate": 3.8284968478223055e-06,
      "loss": 0.5169,
      "step": 155150
    },
    {
      "epoch": 63.51,
      "grad_norm": 2.0820181369781494,
      "learning_rate": 3.8283741049271645e-06,
      "loss": 0.5311,
      "step": 155160
    },
    {
      "epoch": 63.52,
      "grad_norm": 1.872010588645935,
      "learning_rate": 3.82825135523119e-06,
      "loss": 0.5272,
      "step": 155170
    },
    {
      "epoch": 63.52,
      "grad_norm": 2.0576834678649902,
      "learning_rate": 3.828128598734945e-06,
      "loss": 0.5221,
      "step": 155180
    },
    {
      "epoch": 63.52,
      "grad_norm": 1.643836498260498,
      "learning_rate": 3.828005835438991e-06,
      "loss": 0.508,
      "step": 155190
    },
    {
      "epoch": 63.53,
      "grad_norm": 2.0170342922210693,
      "learning_rate": 3.82788306534389e-06,
      "loss": 0.5215,
      "step": 155200
    },
    {
      "epoch": 63.53,
      "grad_norm": 1.5773594379425049,
      "learning_rate": 3.827760288450207e-06,
      "loss": 0.5181,
      "step": 155210
    },
    {
      "epoch": 63.54,
      "grad_norm": 2.1865956783294678,
      "learning_rate": 3.827637504758503e-06,
      "loss": 0.5061,
      "step": 155220
    },
    {
      "epoch": 63.54,
      "grad_norm": 1.9585094451904297,
      "learning_rate": 3.827514714269339e-06,
      "loss": 0.5023,
      "step": 155230
    },
    {
      "epoch": 63.54,
      "grad_norm": 1.9280760288238525,
      "learning_rate": 3.827391916983279e-06,
      "loss": 0.5258,
      "step": 155240
    },
    {
      "epoch": 63.55,
      "grad_norm": 1.8366661071777344,
      "learning_rate": 3.827269112900885e-06,
      "loss": 0.5408,
      "step": 155250
    },
    {
      "epoch": 63.55,
      "grad_norm": 2.3319361209869385,
      "learning_rate": 3.82714630202272e-06,
      "loss": 0.4961,
      "step": 155260
    },
    {
      "epoch": 63.56,
      "grad_norm": 1.9491615295410156,
      "learning_rate": 3.827023484349346e-06,
      "loss": 0.5238,
      "step": 155270
    },
    {
      "epoch": 63.56,
      "grad_norm": 1.9515528678894043,
      "learning_rate": 3.826900659881326e-06,
      "loss": 0.5377,
      "step": 155280
    },
    {
      "epoch": 63.57,
      "grad_norm": 1.845450758934021,
      "learning_rate": 3.8267778286192226e-06,
      "loss": 0.5259,
      "step": 155290
    },
    {
      "epoch": 63.57,
      "grad_norm": 1.8102916479110718,
      "learning_rate": 3.826654990563598e-06,
      "loss": 0.5037,
      "step": 155300
    },
    {
      "epoch": 63.57,
      "grad_norm": 2.397449493408203,
      "learning_rate": 3.826532145715016e-06,
      "loss": 0.5133,
      "step": 155310
    },
    {
      "epoch": 63.58,
      "grad_norm": 2.2831649780273438,
      "learning_rate": 3.826409294074038e-06,
      "loss": 0.5223,
      "step": 155320
    },
    {
      "epoch": 63.58,
      "grad_norm": 2.4172580242156982,
      "learning_rate": 3.826286435641227e-06,
      "loss": 0.5053,
      "step": 155330
    },
    {
      "epoch": 63.59,
      "grad_norm": 1.7146492004394531,
      "learning_rate": 3.826163570417147e-06,
      "loss": 0.5146,
      "step": 155340
    },
    {
      "epoch": 63.59,
      "grad_norm": 2.0278682708740234,
      "learning_rate": 3.82604069840236e-06,
      "loss": 0.5103,
      "step": 155350
    },
    {
      "epoch": 63.59,
      "grad_norm": 1.6495859622955322,
      "learning_rate": 3.825917819597429e-06,
      "loss": 0.526,
      "step": 155360
    },
    {
      "epoch": 63.6,
      "grad_norm": 1.8869777917861938,
      "learning_rate": 3.825794934002916e-06,
      "loss": 0.51,
      "step": 155370
    },
    {
      "epoch": 63.6,
      "grad_norm": 1.6490978002548218,
      "learning_rate": 3.825672041619385e-06,
      "loss": 0.5079,
      "step": 155380
    },
    {
      "epoch": 63.61,
      "grad_norm": 2.289264678955078,
      "learning_rate": 3.825549142447399e-06,
      "loss": 0.5456,
      "step": 155390
    },
    {
      "epoch": 63.61,
      "grad_norm": 1.5526530742645264,
      "learning_rate": 3.8254262364875205e-06,
      "loss": 0.5107,
      "step": 155400
    },
    {
      "epoch": 63.61,
      "grad_norm": 2.171860456466675,
      "learning_rate": 3.825303323740313e-06,
      "loss": 0.5429,
      "step": 155410
    },
    {
      "epoch": 63.62,
      "grad_norm": 1.7381356954574585,
      "learning_rate": 3.825180404206338e-06,
      "loss": 0.4888,
      "step": 155420
    },
    {
      "epoch": 63.62,
      "grad_norm": 1.7838133573532104,
      "learning_rate": 3.825057477886161e-06,
      "loss": 0.5249,
      "step": 155430
    },
    {
      "epoch": 63.63,
      "grad_norm": 2.944317102432251,
      "learning_rate": 3.824934544780342e-06,
      "loss": 0.5404,
      "step": 155440
    },
    {
      "epoch": 63.63,
      "grad_norm": 2.1825497150421143,
      "learning_rate": 3.824811604889448e-06,
      "loss": 0.5199,
      "step": 155450
    },
    {
      "epoch": 63.63,
      "grad_norm": 1.8586359024047852,
      "learning_rate": 3.8246886582140394e-06,
      "loss": 0.5205,
      "step": 155460
    },
    {
      "epoch": 63.64,
      "grad_norm": 1.9976279735565186,
      "learning_rate": 3.824565704754679e-06,
      "loss": 0.5133,
      "step": 155470
    },
    {
      "epoch": 63.64,
      "grad_norm": 6.667086124420166,
      "learning_rate": 3.824442744511933e-06,
      "loss": 0.5205,
      "step": 155480
    },
    {
      "epoch": 63.65,
      "grad_norm": 2.777783155441284,
      "learning_rate": 3.824319777486362e-06,
      "loss": 0.5247,
      "step": 155490
    },
    {
      "epoch": 63.65,
      "grad_norm": 2.4009616374969482,
      "learning_rate": 3.82419680367853e-06,
      "loss": 0.5329,
      "step": 155500
    },
    {
      "epoch": 63.66,
      "grad_norm": 1.9027026891708374,
      "learning_rate": 3.824073823089002e-06,
      "loss": 0.5175,
      "step": 155510
    },
    {
      "epoch": 63.66,
      "grad_norm": 1.958800196647644,
      "learning_rate": 3.823950835718338e-06,
      "loss": 0.5402,
      "step": 155520
    },
    {
      "epoch": 63.66,
      "grad_norm": 2.202186107635498,
      "learning_rate": 3.823827841567104e-06,
      "loss": 0.501,
      "step": 155530
    },
    {
      "epoch": 63.67,
      "grad_norm": 1.8114458322525024,
      "learning_rate": 3.823704840635863e-06,
      "loss": 0.5231,
      "step": 155540
    },
    {
      "epoch": 63.67,
      "grad_norm": 2.3051488399505615,
      "learning_rate": 3.823581832925177e-06,
      "loss": 0.5268,
      "step": 155550
    },
    {
      "epoch": 63.68,
      "grad_norm": 2.2117128372192383,
      "learning_rate": 3.8234588184356115e-06,
      "loss": 0.517,
      "step": 155560
    },
    {
      "epoch": 63.68,
      "grad_norm": 2.153409957885742,
      "learning_rate": 3.823335797167728e-06,
      "loss": 0.5189,
      "step": 155570
    },
    {
      "epoch": 63.68,
      "grad_norm": 1.714379906654358,
      "learning_rate": 3.823212769122092e-06,
      "loss": 0.532,
      "step": 155580
    },
    {
      "epoch": 63.69,
      "grad_norm": 2.2561185359954834,
      "learning_rate": 3.823089734299266e-06,
      "loss": 0.5133,
      "step": 155590
    },
    {
      "epoch": 63.69,
      "grad_norm": 2.11336612701416,
      "learning_rate": 3.822966692699814e-06,
      "loss": 0.5283,
      "step": 155600
    },
    {
      "epoch": 63.7,
      "grad_norm": 1.8597952127456665,
      "learning_rate": 3.822843644324299e-06,
      "loss": 0.5247,
      "step": 155610
    },
    {
      "epoch": 63.7,
      "grad_norm": 2.0181236267089844,
      "learning_rate": 3.822720589173285e-06,
      "loss": 0.498,
      "step": 155620
    },
    {
      "epoch": 63.7,
      "grad_norm": 1.7574529647827148,
      "learning_rate": 3.822597527247336e-06,
      "loss": 0.506,
      "step": 155630
    },
    {
      "epoch": 63.71,
      "grad_norm": 1.6089386940002441,
      "learning_rate": 3.822474458547016e-06,
      "loss": 0.5171,
      "step": 155640
    },
    {
      "epoch": 63.71,
      "grad_norm": 1.4325276613235474,
      "learning_rate": 3.8223513830728876e-06,
      "loss": 0.5219,
      "step": 155650
    },
    {
      "epoch": 63.72,
      "grad_norm": 1.5617752075195312,
      "learning_rate": 3.822228300825515e-06,
      "loss": 0.5238,
      "step": 155660
    },
    {
      "epoch": 63.72,
      "grad_norm": 2.202331781387329,
      "learning_rate": 3.822105211805463e-06,
      "loss": 0.518,
      "step": 155670
    },
    {
      "epoch": 63.72,
      "grad_norm": 2.26603102684021,
      "learning_rate": 3.821982116013294e-06,
      "loss": 0.5137,
      "step": 155680
    },
    {
      "epoch": 63.73,
      "grad_norm": 1.8465023040771484,
      "learning_rate": 3.821859013449573e-06,
      "loss": 0.5307,
      "step": 155690
    },
    {
      "epoch": 63.73,
      "grad_norm": 1.9394177198410034,
      "learning_rate": 3.8217359041148644e-06,
      "loss": 0.5309,
      "step": 155700
    },
    {
      "epoch": 63.74,
      "grad_norm": 1.9813733100891113,
      "learning_rate": 3.82161278800973e-06,
      "loss": 0.5421,
      "step": 155710
    },
    {
      "epoch": 63.74,
      "grad_norm": 1.3916929960250854,
      "learning_rate": 3.821489665134736e-06,
      "loss": 0.5136,
      "step": 155720
    },
    {
      "epoch": 63.75,
      "grad_norm": 2.044358491897583,
      "learning_rate": 3.8213665354904446e-06,
      "loss": 0.5365,
      "step": 155730
    },
    {
      "epoch": 63.75,
      "grad_norm": 2.0695252418518066,
      "learning_rate": 3.821243399077421e-06,
      "loss": 0.5313,
      "step": 155740
    },
    {
      "epoch": 63.75,
      "grad_norm": 1.6747117042541504,
      "learning_rate": 3.821120255896229e-06,
      "loss": 0.5329,
      "step": 155750
    },
    {
      "epoch": 63.76,
      "grad_norm": 2.1479265689849854,
      "learning_rate": 3.820997105947432e-06,
      "loss": 0.52,
      "step": 155760
    },
    {
      "epoch": 63.76,
      "grad_norm": 1.942146897315979,
      "learning_rate": 3.820873949231595e-06,
      "loss": 0.5231,
      "step": 155770
    },
    {
      "epoch": 63.77,
      "grad_norm": 1.4068529605865479,
      "learning_rate": 3.820750785749283e-06,
      "loss": 0.5005,
      "step": 155780
    },
    {
      "epoch": 63.77,
      "grad_norm": 1.9046214818954468,
      "learning_rate": 3.820627615501059e-06,
      "loss": 0.5248,
      "step": 155790
    },
    {
      "epoch": 63.77,
      "grad_norm": 2.689561367034912,
      "learning_rate": 3.820504438487486e-06,
      "loss": 0.5197,
      "step": 155800
    },
    {
      "epoch": 63.78,
      "grad_norm": 1.5144431591033936,
      "learning_rate": 3.820381254709132e-06,
      "loss": 0.5462,
      "step": 155810
    },
    {
      "epoch": 63.78,
      "grad_norm": 1.846634864807129,
      "learning_rate": 3.820258064166556e-06,
      "loss": 0.5163,
      "step": 155820
    },
    {
      "epoch": 63.79,
      "grad_norm": 1.9912246465682983,
      "learning_rate": 3.820134866860327e-06,
      "loss": 0.532,
      "step": 155830
    },
    {
      "epoch": 63.79,
      "grad_norm": 2.31644868850708,
      "learning_rate": 3.820011662791008e-06,
      "loss": 0.5249,
      "step": 155840
    },
    {
      "epoch": 63.79,
      "grad_norm": 1.7381772994995117,
      "learning_rate": 3.819888451959162e-06,
      "loss": 0.5345,
      "step": 155850
    },
    {
      "epoch": 63.8,
      "grad_norm": 1.6873555183410645,
      "learning_rate": 3.819765234365354e-06,
      "loss": 0.534,
      "step": 155860
    },
    {
      "epoch": 63.8,
      "grad_norm": 1.6260446310043335,
      "learning_rate": 3.8196420100101495e-06,
      "loss": 0.5193,
      "step": 155870
    },
    {
      "epoch": 63.81,
      "grad_norm": 2.0521903038024902,
      "learning_rate": 3.819518778894112e-06,
      "loss": 0.5321,
      "step": 155880
    },
    {
      "epoch": 63.81,
      "grad_norm": 2.874046564102173,
      "learning_rate": 3.819395541017806e-06,
      "loss": 0.5321,
      "step": 155890
    },
    {
      "epoch": 63.81,
      "grad_norm": 1.9425513744354248,
      "learning_rate": 3.819272296381797e-06,
      "loss": 0.5354,
      "step": 155900
    },
    {
      "epoch": 63.82,
      "grad_norm": 2.013113021850586,
      "learning_rate": 3.8191490449866486e-06,
      "loss": 0.5407,
      "step": 155910
    },
    {
      "epoch": 63.82,
      "grad_norm": 2.5969417095184326,
      "learning_rate": 3.819025786832925e-06,
      "loss": 0.5324,
      "step": 155920
    },
    {
      "epoch": 63.83,
      "grad_norm": 1.6298719644546509,
      "learning_rate": 3.8189025219211925e-06,
      "loss": 0.5283,
      "step": 155930
    },
    {
      "epoch": 63.83,
      "grad_norm": 1.810927152633667,
      "learning_rate": 3.818779250252014e-06,
      "loss": 0.5327,
      "step": 155940
    },
    {
      "epoch": 63.84,
      "grad_norm": 1.9705318212509155,
      "learning_rate": 3.818655971825955e-06,
      "loss": 0.5087,
      "step": 155950
    },
    {
      "epoch": 63.84,
      "grad_norm": 1.8412926197052002,
      "learning_rate": 3.818532686643582e-06,
      "loss": 0.5164,
      "step": 155960
    },
    {
      "epoch": 63.84,
      "grad_norm": 1.8432652950286865,
      "learning_rate": 3.818409394705456e-06,
      "loss": 0.5257,
      "step": 155970
    },
    {
      "epoch": 63.85,
      "grad_norm": 2.2821717262268066,
      "learning_rate": 3.818286096012144e-06,
      "loss": 0.5432,
      "step": 155980
    },
    {
      "epoch": 63.85,
      "grad_norm": 1.456052541732788,
      "learning_rate": 3.81816279056421e-06,
      "loss": 0.5164,
      "step": 155990
    },
    {
      "epoch": 63.86,
      "grad_norm": 3.3953537940979004,
      "learning_rate": 3.81803947836222e-06,
      "loss": 0.5119,
      "step": 156000
    },
    {
      "epoch": 63.86,
      "grad_norm": 2.742408275604248,
      "learning_rate": 3.817916159406738e-06,
      "loss": 0.5198,
      "step": 156010
    },
    {
      "epoch": 63.86,
      "grad_norm": 1.9668079614639282,
      "learning_rate": 3.817792833698329e-06,
      "loss": 0.5407,
      "step": 156020
    },
    {
      "epoch": 63.87,
      "grad_norm": 2.386472463607788,
      "learning_rate": 3.817669501237559e-06,
      "loss": 0.5471,
      "step": 156030
    },
    {
      "epoch": 63.87,
      "grad_norm": 1.595240831375122,
      "learning_rate": 3.8175461620249915e-06,
      "loss": 0.5314,
      "step": 156040
    },
    {
      "epoch": 63.88,
      "grad_norm": 2.273665189743042,
      "learning_rate": 3.817422816061192e-06,
      "loss": 0.5063,
      "step": 156050
    },
    {
      "epoch": 63.88,
      "grad_norm": 2.725404977798462,
      "learning_rate": 3.817299463346725e-06,
      "loss": 0.5046,
      "step": 156060
    },
    {
      "epoch": 63.88,
      "grad_norm": 1.8941868543624878,
      "learning_rate": 3.817176103882157e-06,
      "loss": 0.5358,
      "step": 156070
    },
    {
      "epoch": 63.89,
      "grad_norm": 1.700403094291687,
      "learning_rate": 3.817052737668052e-06,
      "loss": 0.5168,
      "step": 156080
    },
    {
      "epoch": 63.89,
      "grad_norm": 2.5565543174743652,
      "learning_rate": 3.816929364704975e-06,
      "loss": 0.5193,
      "step": 156090
    },
    {
      "epoch": 63.9,
      "grad_norm": 2.1096320152282715,
      "learning_rate": 3.816805984993491e-06,
      "loss": 0.5191,
      "step": 156100
    },
    {
      "epoch": 63.9,
      "grad_norm": 2.270777940750122,
      "learning_rate": 3.816682598534166e-06,
      "loss": 0.5565,
      "step": 156110
    },
    {
      "epoch": 63.91,
      "grad_norm": 1.5686225891113281,
      "learning_rate": 3.816559205327566e-06,
      "loss": 0.5115,
      "step": 156120
    },
    {
      "epoch": 63.91,
      "grad_norm": 1.870316982269287,
      "learning_rate": 3.816435805374254e-06,
      "loss": 0.5077,
      "step": 156130
    },
    {
      "epoch": 63.91,
      "grad_norm": 1.9572515487670898,
      "learning_rate": 3.816312398674797e-06,
      "loss": 0.5348,
      "step": 156140
    },
    {
      "epoch": 63.92,
      "grad_norm": 1.880399227142334,
      "learning_rate": 3.816188985229759e-06,
      "loss": 0.5059,
      "step": 156150
    },
    {
      "epoch": 63.92,
      "grad_norm": 1.3947865962982178,
      "learning_rate": 3.8160655650397065e-06,
      "loss": 0.5066,
      "step": 156160
    },
    {
      "epoch": 63.93,
      "grad_norm": 1.8165700435638428,
      "learning_rate": 3.815942138105204e-06,
      "loss": 0.5243,
      "step": 156170
    },
    {
      "epoch": 63.93,
      "grad_norm": 2.1612496376037598,
      "learning_rate": 3.815818704426817e-06,
      "loss": 0.5077,
      "step": 156180
    },
    {
      "epoch": 63.93,
      "grad_norm": 2.06174898147583,
      "learning_rate": 3.815695264005112e-06,
      "loss": 0.5141,
      "step": 156190
    },
    {
      "epoch": 63.94,
      "grad_norm": 1.7415764331817627,
      "learning_rate": 3.8155718168406546e-06,
      "loss": 0.5326,
      "step": 156200
    },
    {
      "epoch": 63.94,
      "grad_norm": 2.0553998947143555,
      "learning_rate": 3.815448362934009e-06,
      "loss": 0.514,
      "step": 156210
    },
    {
      "epoch": 63.95,
      "grad_norm": 2.144745111465454,
      "learning_rate": 3.815324902285739e-06,
      "loss": 0.5476,
      "step": 156220
    },
    {
      "epoch": 63.95,
      "grad_norm": 1.7569316625595093,
      "learning_rate": 3.815201434896414e-06,
      "loss": 0.544,
      "step": 156230
    },
    {
      "epoch": 63.95,
      "grad_norm": 1.7468961477279663,
      "learning_rate": 3.815077960766598e-06,
      "loss": 0.537,
      "step": 156240
    },
    {
      "epoch": 63.96,
      "grad_norm": 2.177065849304199,
      "learning_rate": 3.814954479896856e-06,
      "loss": 0.5189,
      "step": 156250
    },
    {
      "epoch": 63.96,
      "grad_norm": 2.6777892112731934,
      "learning_rate": 3.814830992287754e-06,
      "loss": 0.5318,
      "step": 156260
    },
    {
      "epoch": 63.97,
      "grad_norm": 1.7500073909759521,
      "learning_rate": 3.814707497939858e-06,
      "loss": 0.4942,
      "step": 156270
    },
    {
      "epoch": 63.97,
      "grad_norm": 1.8245478868484497,
      "learning_rate": 3.8145839968537337e-06,
      "loss": 0.5202,
      "step": 156280
    },
    {
      "epoch": 63.97,
      "grad_norm": 1.923181414604187,
      "learning_rate": 3.8144604890299466e-06,
      "loss": 0.5063,
      "step": 156290
    },
    {
      "epoch": 63.98,
      "grad_norm": 1.7800675630569458,
      "learning_rate": 3.8143369744690624e-06,
      "loss": 0.5132,
      "step": 156300
    },
    {
      "epoch": 63.98,
      "grad_norm": 1.5916087627410889,
      "learning_rate": 3.8142134531716472e-06,
      "loss": 0.5302,
      "step": 156310
    },
    {
      "epoch": 63.99,
      "grad_norm": 1.8680126667022705,
      "learning_rate": 3.814089925138266e-06,
      "loss": 0.5226,
      "step": 156320
    },
    {
      "epoch": 63.99,
      "grad_norm": 1.5279895067214966,
      "learning_rate": 3.8139663903694856e-06,
      "loss": 0.5365,
      "step": 156330
    },
    {
      "epoch": 64.0,
      "grad_norm": 2.4622550010681152,
      "learning_rate": 3.813842848865872e-06,
      "loss": 0.5212,
      "step": 156340
    },
    {
      "epoch": 64.0,
      "grad_norm": 1.6249574422836304,
      "learning_rate": 3.8137193006279904e-06,
      "loss": 0.5134,
      "step": 156350
    },
    {
      "epoch": 64.0,
      "eval_loss": 0.5210059285163879,
      "eval_runtime": 52.2135,
      "eval_samples_per_second": 66.056,
      "eval_steps_per_second": 8.274,
      "step": 156352
    },
    {
      "epoch": 64.0,
      "grad_norm": 1.728102207183838,
      "learning_rate": 3.8135957456564077e-06,
      "loss": 0.5249,
      "step": 156360
    },
    {
      "epoch": 64.01,
      "grad_norm": 1.7006560564041138,
      "learning_rate": 3.813472183951689e-06,
      "loss": 0.5347,
      "step": 156370
    },
    {
      "epoch": 64.01,
      "grad_norm": 2.145892858505249,
      "learning_rate": 3.8133486155144004e-06,
      "loss": 0.5082,
      "step": 156380
    },
    {
      "epoch": 64.02,
      "grad_norm": 2.086371421813965,
      "learning_rate": 3.8132250403451074e-06,
      "loss": 0.5101,
      "step": 156390
    },
    {
      "epoch": 64.02,
      "grad_norm": 2.273766040802002,
      "learning_rate": 3.8131014584443776e-06,
      "loss": 0.5081,
      "step": 156400
    },
    {
      "epoch": 64.02,
      "grad_norm": 1.8440519571304321,
      "learning_rate": 3.8129778698127766e-06,
      "loss": 0.5123,
      "step": 156410
    },
    {
      "epoch": 64.03,
      "grad_norm": 2.502145528793335,
      "learning_rate": 3.8128542744508696e-06,
      "loss": 0.5069,
      "step": 156420
    },
    {
      "epoch": 64.03,
      "grad_norm": 2.208252191543579,
      "learning_rate": 3.812730672359224e-06,
      "loss": 0.5265,
      "step": 156430
    },
    {
      "epoch": 64.04,
      "grad_norm": 1.4714715480804443,
      "learning_rate": 3.812607063538405e-06,
      "loss": 0.5071,
      "step": 156440
    },
    {
      "epoch": 64.04,
      "grad_norm": 2.25866961479187,
      "learning_rate": 3.8124834479889797e-06,
      "loss": 0.527,
      "step": 156450
    },
    {
      "epoch": 64.04,
      "grad_norm": 1.9933676719665527,
      "learning_rate": 3.812359825711513e-06,
      "loss": 0.529,
      "step": 156460
    },
    {
      "epoch": 64.05,
      "grad_norm": 2.3075313568115234,
      "learning_rate": 3.812236196706574e-06,
      "loss": 0.5279,
      "step": 156470
    },
    {
      "epoch": 64.05,
      "grad_norm": 1.8234258890151978,
      "learning_rate": 3.8121125609747254e-06,
      "loss": 0.5341,
      "step": 156480
    },
    {
      "epoch": 64.06,
      "grad_norm": 1.8222872018814087,
      "learning_rate": 3.811988918516536e-06,
      "loss": 0.5064,
      "step": 156490
    },
    {
      "epoch": 64.06,
      "grad_norm": 1.6561223268508911,
      "learning_rate": 3.8118652693325714e-06,
      "loss": 0.5147,
      "step": 156500
    },
    {
      "epoch": 64.06,
      "grad_norm": 1.6922423839569092,
      "learning_rate": 3.8117416134233974e-06,
      "loss": 0.5351,
      "step": 156510
    },
    {
      "epoch": 64.07,
      "grad_norm": 2.6577377319335938,
      "learning_rate": 3.8116179507895823e-06,
      "loss": 0.5317,
      "step": 156520
    },
    {
      "epoch": 64.07,
      "grad_norm": 1.5774935483932495,
      "learning_rate": 3.8114942814316914e-06,
      "loss": 0.518,
      "step": 156530
    },
    {
      "epoch": 64.08,
      "grad_norm": 1.9194680452346802,
      "learning_rate": 3.811370605350291e-06,
      "loss": 0.531,
      "step": 156540
    },
    {
      "epoch": 64.08,
      "grad_norm": 1.723948359489441,
      "learning_rate": 3.8112469225459472e-06,
      "loss": 0.5252,
      "step": 156550
    },
    {
      "epoch": 64.09,
      "grad_norm": 2.2452657222747803,
      "learning_rate": 3.8111232330192286e-06,
      "loss": 0.5165,
      "step": 156560
    },
    {
      "epoch": 64.09,
      "grad_norm": 1.5363959074020386,
      "learning_rate": 3.8109995367706992e-06,
      "loss": 0.5084,
      "step": 156570
    },
    {
      "epoch": 64.09,
      "grad_norm": 1.6494003534317017,
      "learning_rate": 3.8108758338009274e-06,
      "loss": 0.5222,
      "step": 156580
    },
    {
      "epoch": 64.1,
      "grad_norm": 1.789835810661316,
      "learning_rate": 3.8107521241104798e-06,
      "loss": 0.523,
      "step": 156590
    },
    {
      "epoch": 64.1,
      "grad_norm": 2.127107858657837,
      "learning_rate": 3.8106284076999218e-06,
      "loss": 0.5081,
      "step": 156600
    },
    {
      "epoch": 64.11,
      "grad_norm": 2.1830224990844727,
      "learning_rate": 3.8105046845698218e-06,
      "loss": 0.5312,
      "step": 156610
    },
    {
      "epoch": 64.11,
      "grad_norm": 1.6277155876159668,
      "learning_rate": 3.8103809547207455e-06,
      "loss": 0.5094,
      "step": 156620
    },
    {
      "epoch": 64.11,
      "grad_norm": 2.0708835124969482,
      "learning_rate": 3.8102572181532594e-06,
      "loss": 0.5323,
      "step": 156630
    },
    {
      "epoch": 64.12,
      "grad_norm": 2.022948980331421,
      "learning_rate": 3.8101334748679316e-06,
      "loss": 0.525,
      "step": 156640
    },
    {
      "epoch": 64.12,
      "grad_norm": 1.583343505859375,
      "learning_rate": 3.810009724865328e-06,
      "loss": 0.5193,
      "step": 156650
    },
    {
      "epoch": 64.13,
      "grad_norm": 1.667233943939209,
      "learning_rate": 3.809885968146016e-06,
      "loss": 0.5198,
      "step": 156660
    },
    {
      "epoch": 64.13,
      "grad_norm": 1.5218716859817505,
      "learning_rate": 3.8097622047105607e-06,
      "loss": 0.5376,
      "step": 156670
    },
    {
      "epoch": 64.13,
      "grad_norm": 1.7417831420898438,
      "learning_rate": 3.8096384345595324e-06,
      "loss": 0.5241,
      "step": 156680
    },
    {
      "epoch": 64.14,
      "grad_norm": 1.8291833400726318,
      "learning_rate": 3.809514657693495e-06,
      "loss": 0.5077,
      "step": 156690
    },
    {
      "epoch": 64.14,
      "grad_norm": 2.737529993057251,
      "learning_rate": 3.8093908741130167e-06,
      "loss": 0.5184,
      "step": 156700
    },
    {
      "epoch": 64.15,
      "grad_norm": 2.211887836456299,
      "learning_rate": 3.809267083818665e-06,
      "loss": 0.5153,
      "step": 156710
    },
    {
      "epoch": 64.15,
      "grad_norm": 1.7136168479919434,
      "learning_rate": 3.809143286811007e-06,
      "loss": 0.5472,
      "step": 156720
    },
    {
      "epoch": 64.15,
      "grad_norm": 2.096021890640259,
      "learning_rate": 3.8090194830906087e-06,
      "loss": 0.5231,
      "step": 156730
    },
    {
      "epoch": 64.16,
      "grad_norm": 2.1480069160461426,
      "learning_rate": 3.808895672658038e-06,
      "loss": 0.5597,
      "step": 156740
    },
    {
      "epoch": 64.16,
      "grad_norm": 1.6285524368286133,
      "learning_rate": 3.8087718555138614e-06,
      "loss": 0.5113,
      "step": 156750
    },
    {
      "epoch": 64.17,
      "grad_norm": 2.3310205936431885,
      "learning_rate": 3.808648031658647e-06,
      "loss": 0.5251,
      "step": 156760
    },
    {
      "epoch": 64.17,
      "grad_norm": 2.3258044719696045,
      "learning_rate": 3.8085242010929615e-06,
      "loss": 0.5284,
      "step": 156770
    },
    {
      "epoch": 64.18,
      "grad_norm": 2.5621657371520996,
      "learning_rate": 3.8084003638173724e-06,
      "loss": 0.5216,
      "step": 156780
    },
    {
      "epoch": 64.18,
      "grad_norm": 2.1793768405914307,
      "learning_rate": 3.8082765198324463e-06,
      "loss": 0.5205,
      "step": 156790
    },
    {
      "epoch": 64.18,
      "grad_norm": 1.855533242225647,
      "learning_rate": 3.808152669138751e-06,
      "loss": 0.5254,
      "step": 156800
    },
    {
      "epoch": 64.19,
      "grad_norm": 1.626361608505249,
      "learning_rate": 3.8080288117368545e-06,
      "loss": 0.5171,
      "step": 156810
    },
    {
      "epoch": 64.19,
      "grad_norm": 1.9128714799880981,
      "learning_rate": 3.807904947627323e-06,
      "loss": 0.5241,
      "step": 156820
    },
    {
      "epoch": 64.2,
      "grad_norm": 2.3022544384002686,
      "learning_rate": 3.807781076810725e-06,
      "loss": 0.4968,
      "step": 156830
    },
    {
      "epoch": 64.2,
      "grad_norm": 1.5816304683685303,
      "learning_rate": 3.8076571992876272e-06,
      "loss": 0.5053,
      "step": 156840
    },
    {
      "epoch": 64.2,
      "grad_norm": 1.3903001546859741,
      "learning_rate": 3.807533315058597e-06,
      "loss": 0.5054,
      "step": 156850
    },
    {
      "epoch": 64.21,
      "grad_norm": 1.9389647245407104,
      "learning_rate": 3.807409424124202e-06,
      "loss": 0.5505,
      "step": 156860
    },
    {
      "epoch": 64.21,
      "grad_norm": 2.5942790508270264,
      "learning_rate": 3.8072855264850104e-06,
      "loss": 0.4993,
      "step": 156870
    },
    {
      "epoch": 64.22,
      "grad_norm": 2.341620445251465,
      "learning_rate": 3.807161622141589e-06,
      "loss": 0.514,
      "step": 156880
    },
    {
      "epoch": 64.22,
      "grad_norm": 1.6475862264633179,
      "learning_rate": 3.807037711094505e-06,
      "loss": 0.5065,
      "step": 156890
    },
    {
      "epoch": 64.22,
      "grad_norm": 1.8875422477722168,
      "learning_rate": 3.8069137933443272e-06,
      "loss": 0.5466,
      "step": 156900
    },
    {
      "epoch": 64.23,
      "grad_norm": 2.392578125,
      "learning_rate": 3.8067898688916227e-06,
      "loss": 0.5147,
      "step": 156910
    },
    {
      "epoch": 64.23,
      "grad_norm": 1.6341570615768433,
      "learning_rate": 3.806665937736959e-06,
      "loss": 0.5209,
      "step": 156920
    },
    {
      "epoch": 64.24,
      "grad_norm": 1.9247890710830688,
      "learning_rate": 3.806541999880904e-06,
      "loss": 0.5409,
      "step": 156930
    },
    {
      "epoch": 64.24,
      "grad_norm": 2.7057385444641113,
      "learning_rate": 3.8064180553240257e-06,
      "loss": 0.5247,
      "step": 156940
    },
    {
      "epoch": 64.24,
      "grad_norm": 1.7402249574661255,
      "learning_rate": 3.806294104066891e-06,
      "loss": 0.5157,
      "step": 156950
    },
    {
      "epoch": 64.25,
      "grad_norm": 1.399778127670288,
      "learning_rate": 3.806170146110068e-06,
      "loss": 0.5072,
      "step": 156960
    },
    {
      "epoch": 64.25,
      "grad_norm": 1.9548840522766113,
      "learning_rate": 3.806046181454126e-06,
      "loss": 0.4966,
      "step": 156970
    },
    {
      "epoch": 64.26,
      "grad_norm": 2.7493557929992676,
      "learning_rate": 3.8059222100996303e-06,
      "loss": 0.5444,
      "step": 156980
    },
    {
      "epoch": 64.26,
      "grad_norm": 1.7117786407470703,
      "learning_rate": 3.805798232047151e-06,
      "loss": 0.512,
      "step": 156990
    },
    {
      "epoch": 64.27,
      "grad_norm": 1.552498459815979,
      "learning_rate": 3.805674247297255e-06,
      "loss": 0.5286,
      "step": 157000
    },
    {
      "epoch": 64.27,
      "grad_norm": 1.7905727624893188,
      "learning_rate": 3.8055502558505106e-06,
      "loss": 0.5223,
      "step": 157010
    },
    {
      "epoch": 64.27,
      "grad_norm": 1.9913655519485474,
      "learning_rate": 3.805426257707485e-06,
      "loss": 0.5164,
      "step": 157020
    },
    {
      "epoch": 64.28,
      "grad_norm": 2.1985795497894287,
      "learning_rate": 3.805302252868747e-06,
      "loss": 0.5363,
      "step": 157030
    },
    {
      "epoch": 64.28,
      "grad_norm": 1.9802491664886475,
      "learning_rate": 3.8051782413348653e-06,
      "loss": 0.529,
      "step": 157040
    },
    {
      "epoch": 64.29,
      "grad_norm": 2.0527803897857666,
      "learning_rate": 3.8050542231064067e-06,
      "loss": 0.4965,
      "step": 157050
    },
    {
      "epoch": 64.29,
      "grad_norm": 1.6914242506027222,
      "learning_rate": 3.8049301981839394e-06,
      "loss": 0.5074,
      "step": 157060
    },
    {
      "epoch": 64.29,
      "grad_norm": 2.067511558532715,
      "learning_rate": 3.8048061665680323e-06,
      "loss": 0.5216,
      "step": 157070
    },
    {
      "epoch": 64.3,
      "grad_norm": 2.6189284324645996,
      "learning_rate": 3.8046821282592532e-06,
      "loss": 0.5218,
      "step": 157080
    },
    {
      "epoch": 64.3,
      "grad_norm": 1.9451695680618286,
      "learning_rate": 3.8045580832581698e-06,
      "loss": 0.5153,
      "step": 157090
    },
    {
      "epoch": 64.31,
      "grad_norm": 1.680678367614746,
      "learning_rate": 3.8044340315653507e-06,
      "loss": 0.5183,
      "step": 157100
    },
    {
      "epoch": 64.31,
      "grad_norm": 1.4652711153030396,
      "learning_rate": 3.8043099731813652e-06,
      "loss": 0.521,
      "step": 157110
    },
    {
      "epoch": 64.31,
      "grad_norm": 2.326519250869751,
      "learning_rate": 3.8041859081067796e-06,
      "loss": 0.505,
      "step": 157120
    },
    {
      "epoch": 64.32,
      "grad_norm": 1.7913323640823364,
      "learning_rate": 3.8040618363421642e-06,
      "loss": 0.5277,
      "step": 157130
    },
    {
      "epoch": 64.32,
      "grad_norm": 2.0813066959381104,
      "learning_rate": 3.8039377578880863e-06,
      "loss": 0.5194,
      "step": 157140
    },
    {
      "epoch": 64.33,
      "grad_norm": 2.2152132987976074,
      "learning_rate": 3.803813672745114e-06,
      "loss": 0.5065,
      "step": 157150
    },
    {
      "epoch": 64.33,
      "grad_norm": 2.2561676502227783,
      "learning_rate": 3.803689580913816e-06,
      "loss": 0.5032,
      "step": 157160
    },
    {
      "epoch": 64.33,
      "grad_norm": 1.8620222806930542,
      "learning_rate": 3.8035654823947612e-06,
      "loss": 0.5075,
      "step": 157170
    },
    {
      "epoch": 64.34,
      "grad_norm": 1.6731189489364624,
      "learning_rate": 3.8034413771885173e-06,
      "loss": 0.5196,
      "step": 157180
    },
    {
      "epoch": 64.34,
      "grad_norm": 1.7533595561981201,
      "learning_rate": 3.8033172652956542e-06,
      "loss": 0.5108,
      "step": 157190
    },
    {
      "epoch": 64.35,
      "grad_norm": 1.8017247915267944,
      "learning_rate": 3.8031931467167387e-06,
      "loss": 0.5142,
      "step": 157200
    },
    {
      "epoch": 64.35,
      "grad_norm": 2.0812392234802246,
      "learning_rate": 3.8030690214523404e-06,
      "loss": 0.5314,
      "step": 157210
    },
    {
      "epoch": 64.36,
      "grad_norm": 2.163050413131714,
      "learning_rate": 3.8029448895030284e-06,
      "loss": 0.54,
      "step": 157220
    },
    {
      "epoch": 64.36,
      "grad_norm": 1.3686084747314453,
      "learning_rate": 3.802820750869369e-06,
      "loss": 0.4975,
      "step": 157230
    },
    {
      "epoch": 64.36,
      "grad_norm": 1.43633234500885,
      "learning_rate": 3.8026966055519338e-06,
      "loss": 0.5157,
      "step": 157240
    },
    {
      "epoch": 64.37,
      "grad_norm": 2.5548946857452393,
      "learning_rate": 3.8025724535512903e-06,
      "loss": 0.5165,
      "step": 157250
    },
    {
      "epoch": 64.37,
      "grad_norm": 1.8237823247909546,
      "learning_rate": 3.8024482948680058e-06,
      "loss": 0.4966,
      "step": 157260
    },
    {
      "epoch": 64.38,
      "grad_norm": 2.237967014312744,
      "learning_rate": 3.802324129502651e-06,
      "loss": 0.5146,
      "step": 157270
    },
    {
      "epoch": 64.38,
      "grad_norm": 2.2077674865722656,
      "learning_rate": 3.8021999574557943e-06,
      "loss": 0.5207,
      "step": 157280
    },
    {
      "epoch": 64.38,
      "grad_norm": 1.6995055675506592,
      "learning_rate": 3.8020757787280036e-06,
      "loss": 0.5388,
      "step": 157290
    },
    {
      "epoch": 64.39,
      "grad_norm": 1.5688397884368896,
      "learning_rate": 3.8019515933198484e-06,
      "loss": 0.5442,
      "step": 157300
    },
    {
      "epoch": 64.39,
      "grad_norm": 1.7213021516799927,
      "learning_rate": 3.8018274012318985e-06,
      "loss": 0.4941,
      "step": 157310
    },
    {
      "epoch": 64.4,
      "grad_norm": 1.997789740562439,
      "learning_rate": 3.8017032024647207e-06,
      "loss": 0.5224,
      "step": 157320
    },
    {
      "epoch": 64.4,
      "grad_norm": 2.0727362632751465,
      "learning_rate": 3.801578997018886e-06,
      "loss": 0.5186,
      "step": 157330
    },
    {
      "epoch": 64.4,
      "grad_norm": 2.1745095252990723,
      "learning_rate": 3.8014547848949617e-06,
      "loss": 0.5277,
      "step": 157340
    },
    {
      "epoch": 64.41,
      "grad_norm": 1.8505866527557373,
      "learning_rate": 3.801330566093518e-06,
      "loss": 0.5313,
      "step": 157350
    },
    {
      "epoch": 64.41,
      "grad_norm": 2.4346742630004883,
      "learning_rate": 3.8012063406151228e-06,
      "loss": 0.5335,
      "step": 157360
    },
    {
      "epoch": 64.42,
      "grad_norm": 2.215548276901245,
      "learning_rate": 3.8010821084603465e-06,
      "loss": 0.5048,
      "step": 157370
    },
    {
      "epoch": 64.42,
      "grad_norm": 1.9393659830093384,
      "learning_rate": 3.800957869629757e-06,
      "loss": 0.5157,
      "step": 157380
    },
    {
      "epoch": 64.42,
      "grad_norm": 1.5473308563232422,
      "learning_rate": 3.800833624123925e-06,
      "loss": 0.5055,
      "step": 157390
    },
    {
      "epoch": 64.43,
      "grad_norm": 1.9320075511932373,
      "learning_rate": 3.8007093719434173e-06,
      "loss": 0.5352,
      "step": 157400
    },
    {
      "epoch": 64.43,
      "grad_norm": 2.676973819732666,
      "learning_rate": 3.800585113088805e-06,
      "loss": 0.5133,
      "step": 157410
    },
    {
      "epoch": 64.44,
      "grad_norm": 1.5580945014953613,
      "learning_rate": 3.8004608475606564e-06,
      "loss": 0.5262,
      "step": 157420
    },
    {
      "epoch": 64.44,
      "grad_norm": 1.9922089576721191,
      "learning_rate": 3.800336575359541e-06,
      "loss": 0.516,
      "step": 157430
    },
    {
      "epoch": 64.45,
      "grad_norm": 1.9472416639328003,
      "learning_rate": 3.800212296486028e-06,
      "loss": 0.5249,
      "step": 157440
    },
    {
      "epoch": 64.45,
      "grad_norm": 2.3392820358276367,
      "learning_rate": 3.800088010940687e-06,
      "loss": 0.5403,
      "step": 157450
    },
    {
      "epoch": 64.45,
      "grad_norm": 1.463423252105713,
      "learning_rate": 3.7999637187240873e-06,
      "loss": 0.5312,
      "step": 157460
    },
    {
      "epoch": 64.46,
      "grad_norm": 1.8383907079696655,
      "learning_rate": 3.799839419836798e-06,
      "loss": 0.5099,
      "step": 157470
    },
    {
      "epoch": 64.46,
      "grad_norm": 2.1438121795654297,
      "learning_rate": 3.799715114279389e-06,
      "loss": 0.5437,
      "step": 157480
    },
    {
      "epoch": 64.47,
      "grad_norm": 2.4336254596710205,
      "learning_rate": 3.799590802052428e-06,
      "loss": 0.5323,
      "step": 157490
    },
    {
      "epoch": 64.47,
      "grad_norm": 1.45754075050354,
      "learning_rate": 3.7994664831564862e-06,
      "loss": 0.5221,
      "step": 157500
    },
    {
      "epoch": 64.47,
      "grad_norm": 1.9868764877319336,
      "learning_rate": 3.7993421575921333e-06,
      "loss": 0.5225,
      "step": 157510
    },
    {
      "epoch": 64.48,
      "grad_norm": 1.5930348634719849,
      "learning_rate": 3.799217825359938e-06,
      "loss": 0.5133,
      "step": 157520
    },
    {
      "epoch": 64.48,
      "grad_norm": 1.9698797464370728,
      "learning_rate": 3.79909348646047e-06,
      "loss": 0.5118,
      "step": 157530
    },
    {
      "epoch": 64.49,
      "grad_norm": 2.458934783935547,
      "learning_rate": 3.7989691408942985e-06,
      "loss": 0.533,
      "step": 157540
    },
    {
      "epoch": 64.49,
      "grad_norm": 1.6852531433105469,
      "learning_rate": 3.7988447886619933e-06,
      "loss": 0.4979,
      "step": 157550
    },
    {
      "epoch": 64.49,
      "grad_norm": 2.3803374767303467,
      "learning_rate": 3.798720429764124e-06,
      "loss": 0.5121,
      "step": 157560
    },
    {
      "epoch": 64.5,
      "grad_norm": 2.1697463989257812,
      "learning_rate": 3.7985960642012614e-06,
      "loss": 0.5303,
      "step": 157570
    },
    {
      "epoch": 64.5,
      "grad_norm": 2.501605987548828,
      "learning_rate": 3.7984716919739736e-06,
      "loss": 0.5219,
      "step": 157580
    },
    {
      "epoch": 64.51,
      "grad_norm": 1.6358041763305664,
      "learning_rate": 3.798347313082832e-06,
      "loss": 0.5102,
      "step": 157590
    },
    {
      "epoch": 64.51,
      "grad_norm": 1.7073436975479126,
      "learning_rate": 3.7982229275284047e-06,
      "loss": 0.5169,
      "step": 157600
    },
    {
      "epoch": 64.51,
      "grad_norm": 1.6747126579284668,
      "learning_rate": 3.7980985353112626e-06,
      "loss": 0.519,
      "step": 157610
    },
    {
      "epoch": 64.52,
      "grad_norm": 1.8190209865570068,
      "learning_rate": 3.797974136431974e-06,
      "loss": 0.5148,
      "step": 157620
    },
    {
      "epoch": 64.52,
      "grad_norm": 2.038886785507202,
      "learning_rate": 3.7978497308911105e-06,
      "loss": 0.5135,
      "step": 157630
    },
    {
      "epoch": 64.53,
      "grad_norm": 2.180445432662964,
      "learning_rate": 3.7977253186892415e-06,
      "loss": 0.5318,
      "step": 157640
    },
    {
      "epoch": 64.53,
      "grad_norm": 1.582968831062317,
      "learning_rate": 3.7976008998269365e-06,
      "loss": 0.5247,
      "step": 157650
    },
    {
      "epoch": 64.54,
      "grad_norm": 1.8574382066726685,
      "learning_rate": 3.7974764743047663e-06,
      "loss": 0.5268,
      "step": 157660
    },
    {
      "epoch": 64.54,
      "grad_norm": 1.607961654663086,
      "learning_rate": 3.7973520421232994e-06,
      "loss": 0.5172,
      "step": 157670
    },
    {
      "epoch": 64.54,
      "grad_norm": 1.6781924962997437,
      "learning_rate": 3.7972276032831073e-06,
      "loss": 0.5065,
      "step": 157680
    },
    {
      "epoch": 64.55,
      "grad_norm": 2.2277307510375977,
      "learning_rate": 3.797103157784759e-06,
      "loss": 0.4967,
      "step": 157690
    },
    {
      "epoch": 64.55,
      "grad_norm": 1.8272686004638672,
      "learning_rate": 3.796978705628825e-06,
      "loss": 0.5054,
      "step": 157700
    },
    {
      "epoch": 64.56,
      "grad_norm": 2.104318857192993,
      "learning_rate": 3.7968542468158755e-06,
      "loss": 0.528,
      "step": 157710
    },
    {
      "epoch": 64.56,
      "grad_norm": 1.8171093463897705,
      "learning_rate": 3.796729781346481e-06,
      "loss": 0.5323,
      "step": 157720
    },
    {
      "epoch": 64.56,
      "grad_norm": 2.253207206726074,
      "learning_rate": 3.7966053092212106e-06,
      "loss": 0.5095,
      "step": 157730
    },
    {
      "epoch": 64.57,
      "grad_norm": 1.9810445308685303,
      "learning_rate": 3.7964808304406347e-06,
      "loss": 0.5346,
      "step": 157740
    },
    {
      "epoch": 64.57,
      "grad_norm": 1.5879015922546387,
      "learning_rate": 3.7963563450053237e-06,
      "loss": 0.5327,
      "step": 157750
    },
    {
      "epoch": 64.58,
      "grad_norm": 1.9072142839431763,
      "learning_rate": 3.796231852915849e-06,
      "loss": 0.5034,
      "step": 157760
    },
    {
      "epoch": 64.58,
      "grad_norm": 1.5114232301712036,
      "learning_rate": 3.796107354172779e-06,
      "loss": 0.5254,
      "step": 157770
    },
    {
      "epoch": 64.58,
      "grad_norm": 2.0516867637634277,
      "learning_rate": 3.7959828487766854e-06,
      "loss": 0.5205,
      "step": 157780
    },
    {
      "epoch": 64.59,
      "grad_norm": 1.5065984725952148,
      "learning_rate": 3.795858336728138e-06,
      "loss": 0.5064,
      "step": 157790
    },
    {
      "epoch": 64.59,
      "grad_norm": 2.4750311374664307,
      "learning_rate": 3.795733818027707e-06,
      "loss": 0.5382,
      "step": 157800
    },
    {
      "epoch": 64.6,
      "grad_norm": 2.178083896636963,
      "learning_rate": 3.7956092926759627e-06,
      "loss": 0.5305,
      "step": 157810
    },
    {
      "epoch": 64.6,
      "grad_norm": 2.0675065517425537,
      "learning_rate": 3.795484760673477e-06,
      "loss": 0.5237,
      "step": 157820
    },
    {
      "epoch": 64.6,
      "grad_norm": 2.3473153114318848,
      "learning_rate": 3.7953602220208176e-06,
      "loss": 0.5404,
      "step": 157830
    },
    {
      "epoch": 64.61,
      "grad_norm": 1.625840425491333,
      "learning_rate": 3.7952356767185575e-06,
      "loss": 0.5312,
      "step": 157840
    },
    {
      "epoch": 64.61,
      "grad_norm": 2.5259077548980713,
      "learning_rate": 3.795111124767266e-06,
      "loss": 0.513,
      "step": 157850
    },
    {
      "epoch": 64.62,
      "grad_norm": 1.748494029045105,
      "learning_rate": 3.794986566167514e-06,
      "loss": 0.493,
      "step": 157860
    },
    {
      "epoch": 64.62,
      "grad_norm": 2.158263921737671,
      "learning_rate": 3.7948620009198724e-06,
      "loss": 0.5059,
      "step": 157870
    },
    {
      "epoch": 64.63,
      "grad_norm": 6.759503364562988,
      "learning_rate": 3.7947374290249117e-06,
      "loss": 0.5161,
      "step": 157880
    },
    {
      "epoch": 64.63,
      "grad_norm": 2.237151861190796,
      "learning_rate": 3.794612850483202e-06,
      "loss": 0.5107,
      "step": 157890
    },
    {
      "epoch": 64.63,
      "grad_norm": 1.5740149021148682,
      "learning_rate": 3.7944882652953137e-06,
      "loss": 0.5061,
      "step": 157900
    },
    {
      "epoch": 64.64,
      "grad_norm": 1.9208900928497314,
      "learning_rate": 3.7943636734618183e-06,
      "loss": 0.5386,
      "step": 157910
    },
    {
      "epoch": 64.64,
      "grad_norm": 2.5172882080078125,
      "learning_rate": 3.7942390749832863e-06,
      "loss": 0.5023,
      "step": 157920
    },
    {
      "epoch": 64.65,
      "grad_norm": 1.830405831336975,
      "learning_rate": 3.7941144698602883e-06,
      "loss": 0.5233,
      "step": 157930
    },
    {
      "epoch": 64.65,
      "grad_norm": 2.8359787464141846,
      "learning_rate": 3.793989858093396e-06,
      "loss": 0.5384,
      "step": 157940
    },
    {
      "epoch": 64.65,
      "grad_norm": 2.6596243381500244,
      "learning_rate": 3.793865239683179e-06,
      "loss": 0.528,
      "step": 157950
    },
    {
      "epoch": 64.66,
      "grad_norm": 2.283097267150879,
      "learning_rate": 3.7937406146302094e-06,
      "loss": 0.5194,
      "step": 157960
    },
    {
      "epoch": 64.66,
      "grad_norm": 1.8615808486938477,
      "learning_rate": 3.7936159829350568e-06,
      "loss": 0.5324,
      "step": 157970
    },
    {
      "epoch": 64.67,
      "grad_norm": 1.7773933410644531,
      "learning_rate": 3.7934913445982925e-06,
      "loss": 0.5287,
      "step": 157980
    },
    {
      "epoch": 64.67,
      "grad_norm": 2.51450777053833,
      "learning_rate": 3.793366699620488e-06,
      "loss": 0.5185,
      "step": 157990
    },
    {
      "epoch": 64.67,
      "grad_norm": 2.2145230770111084,
      "learning_rate": 3.7932420480022137e-06,
      "loss": 0.5452,
      "step": 158000
    },
    {
      "epoch": 64.68,
      "grad_norm": 2.004405975341797,
      "learning_rate": 3.7931173897440405e-06,
      "loss": 0.5015,
      "step": 158010
    },
    {
      "epoch": 64.68,
      "grad_norm": 1.6584433317184448,
      "learning_rate": 3.7929927248465398e-06,
      "loss": 0.5187,
      "step": 158020
    },
    {
      "epoch": 64.69,
      "grad_norm": 1.9257322549819946,
      "learning_rate": 3.792868053310283e-06,
      "loss": 0.518,
      "step": 158030
    },
    {
      "epoch": 64.69,
      "grad_norm": 1.7118544578552246,
      "learning_rate": 3.792743375135841e-06,
      "loss": 0.5295,
      "step": 158040
    },
    {
      "epoch": 64.7,
      "grad_norm": 2.0533318519592285,
      "learning_rate": 3.7926186903237844e-06,
      "loss": 0.5085,
      "step": 158050
    },
    {
      "epoch": 64.7,
      "grad_norm": 1.627707600593567,
      "learning_rate": 3.7924939988746845e-06,
      "loss": 0.5187,
      "step": 158060
    },
    {
      "epoch": 64.7,
      "grad_norm": 2.4957358837127686,
      "learning_rate": 3.792369300789113e-06,
      "loss": 0.5153,
      "step": 158070
    },
    {
      "epoch": 64.71,
      "grad_norm": 2.293165683746338,
      "learning_rate": 3.792244596067641e-06,
      "loss": 0.53,
      "step": 158080
    },
    {
      "epoch": 64.71,
      "grad_norm": 1.8185440301895142,
      "learning_rate": 3.792119884710839e-06,
      "loss": 0.5158,
      "step": 158090
    },
    {
      "epoch": 64.72,
      "grad_norm": 1.600371241569519,
      "learning_rate": 3.79199516671928e-06,
      "loss": 0.5434,
      "step": 158100
    },
    {
      "epoch": 64.72,
      "grad_norm": 2.544764757156372,
      "learning_rate": 3.7918704420935325e-06,
      "loss": 0.5491,
      "step": 158110
    },
    {
      "epoch": 64.72,
      "grad_norm": 2.094207525253296,
      "learning_rate": 3.791745710834171e-06,
      "loss": 0.5187,
      "step": 158120
    },
    {
      "epoch": 64.73,
      "grad_norm": 1.9715584516525269,
      "learning_rate": 3.7916209729417644e-06,
      "loss": 0.5127,
      "step": 158130
    },
    {
      "epoch": 64.73,
      "grad_norm": 2.442898750305176,
      "learning_rate": 3.7914962284168854e-06,
      "loss": 0.5446,
      "step": 158140
    },
    {
      "epoch": 64.74,
      "grad_norm": 1.9021944999694824,
      "learning_rate": 3.7913714772601056e-06,
      "loss": 0.5341,
      "step": 158150
    },
    {
      "epoch": 64.74,
      "grad_norm": 1.750558614730835,
      "learning_rate": 3.791246719471996e-06,
      "loss": 0.5204,
      "step": 158160
    },
    {
      "epoch": 64.74,
      "grad_norm": 1.7232595682144165,
      "learning_rate": 3.7911219550531273e-06,
      "loss": 0.5456,
      "step": 158170
    },
    {
      "epoch": 64.75,
      "grad_norm": 2.3789923191070557,
      "learning_rate": 3.7909971840040727e-06,
      "loss": 0.5302,
      "step": 158180
    },
    {
      "epoch": 64.75,
      "grad_norm": 1.9170992374420166,
      "learning_rate": 3.790872406325402e-06,
      "loss": 0.5136,
      "step": 158190
    },
    {
      "epoch": 64.76,
      "grad_norm": 2.188831090927124,
      "learning_rate": 3.7907476220176883e-06,
      "loss": 0.5194,
      "step": 158200
    },
    {
      "epoch": 64.76,
      "grad_norm": 2.036184787750244,
      "learning_rate": 3.790622831081503e-06,
      "loss": 0.5121,
      "step": 158210
    },
    {
      "epoch": 64.76,
      "grad_norm": 2.0240397453308105,
      "learning_rate": 3.7904980335174168e-06,
      "loss": 0.5274,
      "step": 158220
    },
    {
      "epoch": 64.77,
      "grad_norm": 2.2895572185516357,
      "learning_rate": 3.7903732293260013e-06,
      "loss": 0.5109,
      "step": 158230
    },
    {
      "epoch": 64.77,
      "grad_norm": 2.1132726669311523,
      "learning_rate": 3.7902484185078295e-06,
      "loss": 0.5045,
      "step": 158240
    },
    {
      "epoch": 64.78,
      "grad_norm": 1.9394965171813965,
      "learning_rate": 3.7901236010634723e-06,
      "loss": 0.5161,
      "step": 158250
    },
    {
      "epoch": 64.78,
      "grad_norm": 2.114136219024658,
      "learning_rate": 3.7899987769935027e-06,
      "loss": 0.5162,
      "step": 158260
    },
    {
      "epoch": 64.79,
      "grad_norm": 2.0538573265075684,
      "learning_rate": 3.78987394629849e-06,
      "loss": 0.5259,
      "step": 158270
    },
    {
      "epoch": 64.79,
      "grad_norm": 1.8969502449035645,
      "learning_rate": 3.7897491089790076e-06,
      "loss": 0.4931,
      "step": 158280
    },
    {
      "epoch": 64.79,
      "grad_norm": 2.189554214477539,
      "learning_rate": 3.7896242650356274e-06,
      "loss": 0.5207,
      "step": 158290
    },
    {
      "epoch": 64.8,
      "grad_norm": 2.2674880027770996,
      "learning_rate": 3.7894994144689218e-06,
      "loss": 0.5306,
      "step": 158300
    },
    {
      "epoch": 64.8,
      "grad_norm": 2.7991583347320557,
      "learning_rate": 3.789374557279461e-06,
      "loss": 0.5145,
      "step": 158310
    },
    {
      "epoch": 64.81,
      "grad_norm": 1.8454166650772095,
      "learning_rate": 3.789249693467819e-06,
      "loss": 0.5175,
      "step": 158320
    },
    {
      "epoch": 64.81,
      "grad_norm": 1.7948552370071411,
      "learning_rate": 3.7891248230345664e-06,
      "loss": 0.525,
      "step": 158330
    },
    {
      "epoch": 64.81,
      "grad_norm": 1.5889182090759277,
      "learning_rate": 3.7889999459802752e-06,
      "loss": 0.5294,
      "step": 158340
    },
    {
      "epoch": 64.82,
      "grad_norm": 1.8785125017166138,
      "learning_rate": 3.7888750623055174e-06,
      "loss": 0.5259,
      "step": 158350
    },
    {
      "epoch": 64.82,
      "grad_norm": 1.9535754919052124,
      "learning_rate": 3.7887501720108667e-06,
      "loss": 0.5291,
      "step": 158360
    },
    {
      "epoch": 64.83,
      "grad_norm": 1.9128645658493042,
      "learning_rate": 3.7886252750968937e-06,
      "loss": 0.4926,
      "step": 158370
    },
    {
      "epoch": 64.83,
      "grad_norm": 1.4649617671966553,
      "learning_rate": 3.78850037156417e-06,
      "loss": 0.5328,
      "step": 158380
    },
    {
      "epoch": 64.83,
      "grad_norm": 1.5930343866348267,
      "learning_rate": 3.7883754614132693e-06,
      "loss": 0.5458,
      "step": 158390
    },
    {
      "epoch": 64.84,
      "grad_norm": 2.087029218673706,
      "learning_rate": 3.788250544644763e-06,
      "loss": 0.5144,
      "step": 158400
    },
    {
      "epoch": 64.84,
      "grad_norm": 1.6897697448730469,
      "learning_rate": 3.7881256212592237e-06,
      "loss": 0.5019,
      "step": 158410
    },
    {
      "epoch": 64.85,
      "grad_norm": 1.8216501474380493,
      "learning_rate": 3.7880006912572228e-06,
      "loss": 0.5235,
      "step": 158420
    },
    {
      "epoch": 64.85,
      "grad_norm": 1.9198799133300781,
      "learning_rate": 3.787875754639334e-06,
      "loss": 0.5069,
      "step": 158430
    },
    {
      "epoch": 64.85,
      "grad_norm": 1.8584983348846436,
      "learning_rate": 3.7877508114061277e-06,
      "loss": 0.5145,
      "step": 158440
    },
    {
      "epoch": 64.86,
      "grad_norm": 2.1442151069641113,
      "learning_rate": 3.7876258615581774e-06,
      "loss": 0.5097,
      "step": 158450
    },
    {
      "epoch": 64.86,
      "grad_norm": 1.791730523109436,
      "learning_rate": 3.7875009050960564e-06,
      "loss": 0.5177,
      "step": 158460
    },
    {
      "epoch": 64.87,
      "grad_norm": 2.1214749813079834,
      "learning_rate": 3.7873759420203357e-06,
      "loss": 0.5146,
      "step": 158470
    },
    {
      "epoch": 64.87,
      "grad_norm": 1.9063340425491333,
      "learning_rate": 3.7872509723315873e-06,
      "loss": 0.5096,
      "step": 158480
    },
    {
      "epoch": 64.88,
      "grad_norm": 1.6122405529022217,
      "learning_rate": 3.787125996030385e-06,
      "loss": 0.5348,
      "step": 158490
    },
    {
      "epoch": 64.88,
      "grad_norm": 1.7030928134918213,
      "learning_rate": 3.7870010131173014e-06,
      "loss": 0.528,
      "step": 158500
    },
    {
      "epoch": 64.88,
      "grad_norm": 2.2036046981811523,
      "learning_rate": 3.786876023592908e-06,
      "loss": 0.5175,
      "step": 158510
    },
    {
      "epoch": 64.89,
      "grad_norm": 1.7082620859146118,
      "learning_rate": 3.7867510274577775e-06,
      "loss": 0.5247,
      "step": 158520
    },
    {
      "epoch": 64.89,
      "grad_norm": 1.4692864418029785,
      "learning_rate": 3.786626024712483e-06,
      "loss": 0.5346,
      "step": 158530
    },
    {
      "epoch": 64.9,
      "grad_norm": 2.361672878265381,
      "learning_rate": 3.7865010153575966e-06,
      "loss": 0.5087,
      "step": 158540
    },
    {
      "epoch": 64.9,
      "grad_norm": 2.069108486175537,
      "learning_rate": 3.7863759993936917e-06,
      "loss": 0.5305,
      "step": 158550
    },
    {
      "epoch": 64.9,
      "grad_norm": 2.0938565731048584,
      "learning_rate": 3.7862509768213404e-06,
      "loss": 0.5264,
      "step": 158560
    },
    {
      "epoch": 64.91,
      "grad_norm": 1.765430212020874,
      "learning_rate": 3.7861259476411157e-06,
      "loss": 0.5133,
      "step": 158570
    },
    {
      "epoch": 64.91,
      "grad_norm": 1.9779269695281982,
      "learning_rate": 3.7860009118535897e-06,
      "loss": 0.5177,
      "step": 158580
    },
    {
      "epoch": 64.92,
      "grad_norm": 1.909786581993103,
      "learning_rate": 3.7858758694593354e-06,
      "loss": 0.5206,
      "step": 158590
    },
    {
      "epoch": 64.92,
      "grad_norm": 2.249729633331299,
      "learning_rate": 3.785750820458927e-06,
      "loss": 0.5278,
      "step": 158600
    },
    {
      "epoch": 64.92,
      "grad_norm": 2.1689774990081787,
      "learning_rate": 3.785625764852935e-06,
      "loss": 0.5241,
      "step": 158610
    },
    {
      "epoch": 64.93,
      "grad_norm": 2.265632152557373,
      "learning_rate": 3.785500702641934e-06,
      "loss": 0.5387,
      "step": 158620
    },
    {
      "epoch": 64.93,
      "grad_norm": 1.9661688804626465,
      "learning_rate": 3.785375633826496e-06,
      "loss": 0.5039,
      "step": 158630
    },
    {
      "epoch": 64.94,
      "grad_norm": 1.9934335947036743,
      "learning_rate": 3.785250558407194e-06,
      "loss": 0.5327,
      "step": 158640
    },
    {
      "epoch": 64.94,
      "grad_norm": 2.057652235031128,
      "learning_rate": 3.7851254763846026e-06,
      "loss": 0.5305,
      "step": 158650
    },
    {
      "epoch": 64.94,
      "grad_norm": 1.9257174730300903,
      "learning_rate": 3.785000387759292e-06,
      "loss": 0.5485,
      "step": 158660
    },
    {
      "epoch": 64.95,
      "grad_norm": 2.400026798248291,
      "learning_rate": 3.7848752925318373e-06,
      "loss": 0.5169,
      "step": 158670
    },
    {
      "epoch": 64.95,
      "grad_norm": 2.230132579803467,
      "learning_rate": 3.78475019070281e-06,
      "loss": 0.5145,
      "step": 158680
    },
    {
      "epoch": 64.96,
      "grad_norm": 2.109868288040161,
      "learning_rate": 3.7846250822727843e-06,
      "loss": 0.5356,
      "step": 158690
    },
    {
      "epoch": 64.96,
      "grad_norm": 1.603721261024475,
      "learning_rate": 3.7844999672423335e-06,
      "loss": 0.5283,
      "step": 158700
    },
    {
      "epoch": 64.97,
      "grad_norm": 1.8916832208633423,
      "learning_rate": 3.7843748456120294e-06,
      "loss": 0.4962,
      "step": 158710
    },
    {
      "epoch": 64.97,
      "grad_norm": 1.7945116758346558,
      "learning_rate": 3.784249717382447e-06,
      "loss": 0.509,
      "step": 158720
    },
    {
      "epoch": 64.97,
      "grad_norm": 1.944307804107666,
      "learning_rate": 3.784124582554158e-06,
      "loss": 0.5012,
      "step": 158730
    },
    {
      "epoch": 64.98,
      "grad_norm": 2.1088390350341797,
      "learning_rate": 3.7839994411277356e-06,
      "loss": 0.5518,
      "step": 158740
    },
    {
      "epoch": 64.98,
      "grad_norm": 1.5642118453979492,
      "learning_rate": 3.7838742931037545e-06,
      "loss": 0.4925,
      "step": 158750
    },
    {
      "epoch": 64.99,
      "grad_norm": 1.901861310005188,
      "learning_rate": 3.7837491384827864e-06,
      "loss": 0.4974,
      "step": 158760
    },
    {
      "epoch": 64.99,
      "grad_norm": 1.6414539813995361,
      "learning_rate": 3.7836239772654056e-06,
      "loss": 0.5288,
      "step": 158770
    },
    {
      "epoch": 64.99,
      "grad_norm": 1.9333956241607666,
      "learning_rate": 3.783498809452185e-06,
      "loss": 0.5281,
      "step": 158780
    },
    {
      "epoch": 65.0,
      "grad_norm": 1.9522759914398193,
      "learning_rate": 3.783373635043698e-06,
      "loss": 0.5305,
      "step": 158790
    },
    {
      "epoch": 65.0,
      "eval_loss": 0.523764431476593,
      "eval_runtime": 52.8968,
      "eval_samples_per_second": 65.202,
      "eval_steps_per_second": 8.167,
      "step": 158795
    },
    {
      "epoch": 65.0,
      "grad_norm": 2.4279377460479736,
      "learning_rate": 3.7832484540405177e-06,
      "loss": 0.5112,
      "step": 158800
    },
    {
      "epoch": 65.01,
      "grad_norm": 1.927599549293518,
      "learning_rate": 3.7831232664432186e-06,
      "loss": 0.5488,
      "step": 158810
    },
    {
      "epoch": 65.01,
      "grad_norm": 1.6416816711425781,
      "learning_rate": 3.7829980722523733e-06,
      "loss": 0.5419,
      "step": 158820
    },
    {
      "epoch": 65.01,
      "grad_norm": 1.7188806533813477,
      "learning_rate": 3.7828728714685555e-06,
      "loss": 0.5169,
      "step": 158830
    },
    {
      "epoch": 65.02,
      "grad_norm": 2.0331783294677734,
      "learning_rate": 3.7827476640923384e-06,
      "loss": 0.5347,
      "step": 158840
    },
    {
      "epoch": 65.02,
      "grad_norm": 1.8934650421142578,
      "learning_rate": 3.782622450124296e-06,
      "loss": 0.5194,
      "step": 158850
    },
    {
      "epoch": 65.03,
      "grad_norm": 2.206455707550049,
      "learning_rate": 3.7824972295650017e-06,
      "loss": 0.5234,
      "step": 158860
    },
    {
      "epoch": 65.03,
      "grad_norm": 1.7830156087875366,
      "learning_rate": 3.7823720024150286e-06,
      "loss": 0.5241,
      "step": 158870
    },
    {
      "epoch": 65.03,
      "grad_norm": 2.4519026279449463,
      "learning_rate": 3.7822467686749516e-06,
      "loss": 0.4904,
      "step": 158880
    },
    {
      "epoch": 65.04,
      "grad_norm": 1.8869450092315674,
      "learning_rate": 3.782121528345344e-06,
      "loss": 0.5201,
      "step": 158890
    },
    {
      "epoch": 65.04,
      "grad_norm": 1.6349377632141113,
      "learning_rate": 3.781996281426779e-06,
      "loss": 0.5215,
      "step": 158900
    },
    {
      "epoch": 65.05,
      "grad_norm": 1.3915135860443115,
      "learning_rate": 3.7818710279198305e-06,
      "loss": 0.5287,
      "step": 158910
    },
    {
      "epoch": 65.05,
      "grad_norm": 2.171539306640625,
      "learning_rate": 3.781745767825072e-06,
      "loss": 0.5164,
      "step": 158920
    },
    {
      "epoch": 65.06,
      "grad_norm": 1.3699145317077637,
      "learning_rate": 3.7816205011430775e-06,
      "loss": 0.5326,
      "step": 158930
    },
    {
      "epoch": 65.06,
      "grad_norm": 2.280742645263672,
      "learning_rate": 3.7814952278744216e-06,
      "loss": 0.5252,
      "step": 158940
    },
    {
      "epoch": 65.06,
      "grad_norm": 3.2266929149627686,
      "learning_rate": 3.7813699480196767e-06,
      "loss": 0.519,
      "step": 158950
    },
    {
      "epoch": 65.07,
      "grad_norm": 2.359330415725708,
      "learning_rate": 3.781244661579418e-06,
      "loss": 0.5165,
      "step": 158960
    },
    {
      "epoch": 65.07,
      "grad_norm": 1.8520656824111938,
      "learning_rate": 3.7811193685542184e-06,
      "loss": 0.5154,
      "step": 158970
    },
    {
      "epoch": 65.08,
      "grad_norm": 2.1587915420532227,
      "learning_rate": 3.7809940689446526e-06,
      "loss": 0.5215,
      "step": 158980
    },
    {
      "epoch": 65.08,
      "grad_norm": 1.8245469331741333,
      "learning_rate": 3.780868762751294e-06,
      "loss": 0.5144,
      "step": 158990
    },
    {
      "epoch": 65.08,
      "grad_norm": 1.99760901927948,
      "learning_rate": 3.7807434499747173e-06,
      "loss": 0.5138,
      "step": 159000
    },
    {
      "epoch": 65.09,
      "grad_norm": 1.7115962505340576,
      "learning_rate": 3.7806181306154967e-06,
      "loss": 0.5016,
      "step": 159010
    },
    {
      "epoch": 65.09,
      "grad_norm": 2.099111557006836,
      "learning_rate": 3.780492804674205e-06,
      "loss": 0.5069,
      "step": 159020
    },
    {
      "epoch": 65.1,
      "grad_norm": 1.9103292226791382,
      "learning_rate": 3.780367472151417e-06,
      "loss": 0.5006,
      "step": 159030
    },
    {
      "epoch": 65.1,
      "grad_norm": 1.762911319732666,
      "learning_rate": 3.7802421330477067e-06,
      "loss": 0.5219,
      "step": 159040
    },
    {
      "epoch": 65.1,
      "grad_norm": 1.2881319522857666,
      "learning_rate": 3.7801167873636486e-06,
      "loss": 0.5239,
      "step": 159050
    },
    {
      "epoch": 65.11,
      "grad_norm": 2.1586098670959473,
      "learning_rate": 3.7799914350998167e-06,
      "loss": 0.5194,
      "step": 159060
    },
    {
      "epoch": 65.11,
      "grad_norm": 2.0037643909454346,
      "learning_rate": 3.7798660762567854e-06,
      "loss": 0.5213,
      "step": 159070
    },
    {
      "epoch": 65.12,
      "grad_norm": 1.9476205110549927,
      "learning_rate": 3.7797407108351284e-06,
      "loss": 0.527,
      "step": 159080
    },
    {
      "epoch": 65.12,
      "grad_norm": 1.8438626527786255,
      "learning_rate": 3.7796153388354207e-06,
      "loss": 0.5086,
      "step": 159090
    },
    {
      "epoch": 65.12,
      "grad_norm": 1.7755497694015503,
      "learning_rate": 3.779489960258236e-06,
      "loss": 0.5325,
      "step": 159100
    },
    {
      "epoch": 65.13,
      "grad_norm": 1.9015839099884033,
      "learning_rate": 3.779364575104149e-06,
      "loss": 0.523,
      "step": 159110
    },
    {
      "epoch": 65.13,
      "grad_norm": 2.5040738582611084,
      "learning_rate": 3.7792391833737343e-06,
      "loss": 0.5297,
      "step": 159120
    },
    {
      "epoch": 65.14,
      "grad_norm": 2.2963154315948486,
      "learning_rate": 3.7791137850675656e-06,
      "loss": 0.5199,
      "step": 159130
    },
    {
      "epoch": 65.14,
      "grad_norm": 2.8717434406280518,
      "learning_rate": 3.7789883801862176e-06,
      "loss": 0.5193,
      "step": 159140
    },
    {
      "epoch": 65.15,
      "grad_norm": 1.6273247003555298,
      "learning_rate": 3.778862968730265e-06,
      "loss": 0.5085,
      "step": 159150
    },
    {
      "epoch": 65.15,
      "grad_norm": 2.196399688720703,
      "learning_rate": 3.7787375507002823e-06,
      "loss": 0.5244,
      "step": 159160
    },
    {
      "epoch": 65.15,
      "grad_norm": 1.5647743940353394,
      "learning_rate": 3.7786121260968437e-06,
      "loss": 0.5355,
      "step": 159170
    },
    {
      "epoch": 65.16,
      "grad_norm": 1.9543116092681885,
      "learning_rate": 3.7784866949205238e-06,
      "loss": 0.5161,
      "step": 159180
    },
    {
      "epoch": 65.16,
      "grad_norm": 1.7725850343704224,
      "learning_rate": 3.7783612571718974e-06,
      "loss": 0.5319,
      "step": 159190
    },
    {
      "epoch": 65.17,
      "grad_norm": 1.9567608833312988,
      "learning_rate": 3.778235812851539e-06,
      "loss": 0.501,
      "step": 159200
    },
    {
      "epoch": 65.17,
      "grad_norm": 1.595030665397644,
      "learning_rate": 3.7781103619600234e-06,
      "loss": 0.5248,
      "step": 159210
    },
    {
      "epoch": 65.17,
      "grad_norm": 1.9133599996566772,
      "learning_rate": 3.777984904497925e-06,
      "loss": 0.532,
      "step": 159220
    },
    {
      "epoch": 65.18,
      "grad_norm": 1.7963534593582153,
      "learning_rate": 3.7778594404658183e-06,
      "loss": 0.5332,
      "step": 159230
    },
    {
      "epoch": 65.18,
      "grad_norm": 1.7187259197235107,
      "learning_rate": 3.7777339698642787e-06,
      "loss": 0.5002,
      "step": 159240
    },
    {
      "epoch": 65.19,
      "grad_norm": 2.7646172046661377,
      "learning_rate": 3.777608492693881e-06,
      "loss": 0.5402,
      "step": 159250
    },
    {
      "epoch": 65.19,
      "grad_norm": 2.17448353767395,
      "learning_rate": 3.7774830089551987e-06,
      "loss": 0.5108,
      "step": 159260
    },
    {
      "epoch": 65.19,
      "grad_norm": 1.6824469566345215,
      "learning_rate": 3.7773575186488085e-06,
      "loss": 0.5308,
      "step": 159270
    },
    {
      "epoch": 65.2,
      "grad_norm": 1.5287319421768188,
      "learning_rate": 3.7772320217752838e-06,
      "loss": 0.5148,
      "step": 159280
    },
    {
      "epoch": 65.2,
      "grad_norm": 1.9704774618148804,
      "learning_rate": 3.7771065183352e-06,
      "loss": 0.5132,
      "step": 159290
    },
    {
      "epoch": 65.21,
      "grad_norm": 1.9011273384094238,
      "learning_rate": 3.7769810083291317e-06,
      "loss": 0.5404,
      "step": 159300
    },
    {
      "epoch": 65.21,
      "grad_norm": 1.892511010169983,
      "learning_rate": 3.7768554917576547e-06,
      "loss": 0.5129,
      "step": 159310
    },
    {
      "epoch": 65.21,
      "grad_norm": 1.9292417764663696,
      "learning_rate": 3.7767299686213428e-06,
      "loss": 0.5238,
      "step": 159320
    },
    {
      "epoch": 65.22,
      "grad_norm": 2.549097776412964,
      "learning_rate": 3.7766044389207716e-06,
      "loss": 0.5315,
      "step": 159330
    },
    {
      "epoch": 65.22,
      "grad_norm": 1.4856767654418945,
      "learning_rate": 3.776478902656516e-06,
      "loss": 0.5324,
      "step": 159340
    },
    {
      "epoch": 65.23,
      "grad_norm": 2.373217821121216,
      "learning_rate": 3.776353359829152e-06,
      "loss": 0.5304,
      "step": 159350
    },
    {
      "epoch": 65.23,
      "grad_norm": 1.6046212911605835,
      "learning_rate": 3.776227810439253e-06,
      "loss": 0.5183,
      "step": 159360
    },
    {
      "epoch": 65.24,
      "grad_norm": 1.9278889894485474,
      "learning_rate": 3.7761022544873955e-06,
      "loss": 0.518,
      "step": 159370
    },
    {
      "epoch": 65.24,
      "grad_norm": 1.970039963722229,
      "learning_rate": 3.7759766919741546e-06,
      "loss": 0.5332,
      "step": 159380
    },
    {
      "epoch": 65.24,
      "grad_norm": 2.109560012817383,
      "learning_rate": 3.7758511229001047e-06,
      "loss": 0.5178,
      "step": 159390
    },
    {
      "epoch": 65.25,
      "grad_norm": 1.6459572315216064,
      "learning_rate": 3.7757255472658207e-06,
      "loss": 0.5129,
      "step": 159400
    },
    {
      "epoch": 65.25,
      "grad_norm": 1.5146909952163696,
      "learning_rate": 3.775599965071879e-06,
      "loss": 0.5112,
      "step": 159410
    },
    {
      "epoch": 65.26,
      "grad_norm": 2.1506237983703613,
      "learning_rate": 3.7754743763188544e-06,
      "loss": 0.5212,
      "step": 159420
    },
    {
      "epoch": 65.26,
      "grad_norm": 2.10945725440979,
      "learning_rate": 3.7753487810073217e-06,
      "loss": 0.5407,
      "step": 159430
    },
    {
      "epoch": 65.26,
      "grad_norm": 1.9828486442565918,
      "learning_rate": 3.775223179137857e-06,
      "loss": 0.4943,
      "step": 159440
    },
    {
      "epoch": 65.27,
      "grad_norm": 1.7316837310791016,
      "learning_rate": 3.7750975707110354e-06,
      "loss": 0.5113,
      "step": 159450
    },
    {
      "epoch": 65.27,
      "grad_norm": 1.8147146701812744,
      "learning_rate": 3.7749719557274326e-06,
      "loss": 0.5269,
      "step": 159460
    },
    {
      "epoch": 65.28,
      "grad_norm": 1.8089708089828491,
      "learning_rate": 3.774846334187623e-06,
      "loss": 0.5273,
      "step": 159470
    },
    {
      "epoch": 65.28,
      "grad_norm": 1.7805765867233276,
      "learning_rate": 3.7747207060921834e-06,
      "loss": 0.5198,
      "step": 159480
    },
    {
      "epoch": 65.28,
      "grad_norm": 1.741111159324646,
      "learning_rate": 3.774595071441688e-06,
      "loss": 0.5365,
      "step": 159490
    },
    {
      "epoch": 65.29,
      "grad_norm": 1.9443687200546265,
      "learning_rate": 3.7744694302367134e-06,
      "loss": 0.5311,
      "step": 159500
    },
    {
      "epoch": 65.29,
      "grad_norm": 1.8513412475585938,
      "learning_rate": 3.774343782477834e-06,
      "loss": 0.52,
      "step": 159510
    },
    {
      "epoch": 65.3,
      "grad_norm": 2.130981206893921,
      "learning_rate": 3.7742181281656266e-06,
      "loss": 0.5303,
      "step": 159520
    },
    {
      "epoch": 65.3,
      "grad_norm": 1.7139203548431396,
      "learning_rate": 3.7740924673006665e-06,
      "loss": 0.524,
      "step": 159530
    },
    {
      "epoch": 65.3,
      "grad_norm": 1.9122682809829712,
      "learning_rate": 3.7739667998835284e-06,
      "loss": 0.5158,
      "step": 159540
    },
    {
      "epoch": 65.31,
      "grad_norm": 1.7599760293960571,
      "learning_rate": 3.773841125914789e-06,
      "loss": 0.5269,
      "step": 159550
    },
    {
      "epoch": 65.31,
      "grad_norm": 2.8605329990386963,
      "learning_rate": 3.773715445395023e-06,
      "loss": 0.5195,
      "step": 159560
    },
    {
      "epoch": 65.32,
      "grad_norm": 2.119436502456665,
      "learning_rate": 3.773589758324808e-06,
      "loss": 0.5285,
      "step": 159570
    },
    {
      "epoch": 65.32,
      "grad_norm": 1.4145503044128418,
      "learning_rate": 3.7734640647047175e-06,
      "loss": 0.5302,
      "step": 159580
    },
    {
      "epoch": 65.33,
      "grad_norm": 2.1519558429718018,
      "learning_rate": 3.773338364535329e-06,
      "loss": 0.4941,
      "step": 159590
    },
    {
      "epoch": 65.33,
      "grad_norm": 1.7540310621261597,
      "learning_rate": 3.7732126578172168e-06,
      "loss": 0.5268,
      "step": 159600
    },
    {
      "epoch": 65.33,
      "grad_norm": 1.9083731174468994,
      "learning_rate": 3.7730869445509585e-06,
      "loss": 0.5252,
      "step": 159610
    },
    {
      "epoch": 65.34,
      "grad_norm": 1.9714711904525757,
      "learning_rate": 3.7729612247371275e-06,
      "loss": 0.494,
      "step": 159620
    },
    {
      "epoch": 65.34,
      "grad_norm": 1.8077915906906128,
      "learning_rate": 3.7728354983763028e-06,
      "loss": 0.5117,
      "step": 159630
    },
    {
      "epoch": 65.35,
      "grad_norm": 1.7333120107650757,
      "learning_rate": 3.7727097654690577e-06,
      "loss": 0.5313,
      "step": 159640
    },
    {
      "epoch": 65.35,
      "grad_norm": 1.8359068632125854,
      "learning_rate": 3.7725840260159695e-06,
      "loss": 0.4973,
      "step": 159650
    },
    {
      "epoch": 65.35,
      "grad_norm": 2.209078788757324,
      "learning_rate": 3.7724582800176146e-06,
      "loss": 0.5161,
      "step": 159660
    },
    {
      "epoch": 65.36,
      "grad_norm": 2.0012707710266113,
      "learning_rate": 3.772332527474568e-06,
      "loss": 0.5204,
      "step": 159670
    },
    {
      "epoch": 65.36,
      "grad_norm": 2.1719818115234375,
      "learning_rate": 3.772206768387406e-06,
      "loss": 0.4979,
      "step": 159680
    },
    {
      "epoch": 65.37,
      "grad_norm": 2.0557799339294434,
      "learning_rate": 3.7720810027567043e-06,
      "loss": 0.5308,
      "step": 159690
    },
    {
      "epoch": 65.37,
      "grad_norm": 2.570707321166992,
      "learning_rate": 3.7719552305830404e-06,
      "loss": 0.5342,
      "step": 159700
    },
    {
      "epoch": 65.37,
      "grad_norm": 1.6896501779556274,
      "learning_rate": 3.7718294518669884e-06,
      "loss": 0.5523,
      "step": 159710
    },
    {
      "epoch": 65.38,
      "grad_norm": 1.9985198974609375,
      "learning_rate": 3.7717036666091257e-06,
      "loss": 0.5223,
      "step": 159720
    },
    {
      "epoch": 65.38,
      "grad_norm": 2.002073049545288,
      "learning_rate": 3.771577874810029e-06,
      "loss": 0.5277,
      "step": 159730
    },
    {
      "epoch": 65.39,
      "grad_norm": 2.0385589599609375,
      "learning_rate": 3.771452076470274e-06,
      "loss": 0.5299,
      "step": 159740
    },
    {
      "epoch": 65.39,
      "grad_norm": 2.0027365684509277,
      "learning_rate": 3.771326271590437e-06,
      "loss": 0.5345,
      "step": 159750
    },
    {
      "epoch": 65.4,
      "grad_norm": 1.9893819093704224,
      "learning_rate": 3.771200460171094e-06,
      "loss": 0.5236,
      "step": 159760
    },
    {
      "epoch": 65.4,
      "grad_norm": 2.0674266815185547,
      "learning_rate": 3.771074642212821e-06,
      "loss": 0.5171,
      "step": 159770
    },
    {
      "epoch": 65.4,
      "grad_norm": 1.8723706007003784,
      "learning_rate": 3.7709488177161953e-06,
      "loss": 0.4986,
      "step": 159780
    },
    {
      "epoch": 65.41,
      "grad_norm": 2.030294418334961,
      "learning_rate": 3.7708229866817926e-06,
      "loss": 0.5004,
      "step": 159790
    },
    {
      "epoch": 65.41,
      "grad_norm": 1.7294349670410156,
      "learning_rate": 3.77069714911019e-06,
      "loss": 0.5299,
      "step": 159800
    },
    {
      "epoch": 65.42,
      "grad_norm": 1.828220248222351,
      "learning_rate": 3.7705713050019634e-06,
      "loss": 0.5163,
      "step": 159810
    },
    {
      "epoch": 65.42,
      "grad_norm": 2.165083885192871,
      "learning_rate": 3.7704454543576887e-06,
      "loss": 0.5104,
      "step": 159820
    },
    {
      "epoch": 65.42,
      "grad_norm": 1.6806823015213013,
      "learning_rate": 3.7703195971779434e-06,
      "loss": 0.5224,
      "step": 159830
    },
    {
      "epoch": 65.43,
      "grad_norm": 2.6214141845703125,
      "learning_rate": 3.7701937334633045e-06,
      "loss": 0.5365,
      "step": 159840
    },
    {
      "epoch": 65.43,
      "grad_norm": 1.8055998086929321,
      "learning_rate": 3.770067863214347e-06,
      "loss": 0.5232,
      "step": 159850
    },
    {
      "epoch": 65.44,
      "grad_norm": 2.105255126953125,
      "learning_rate": 3.7699419864316486e-06,
      "loss": 0.5125,
      "step": 159860
    },
    {
      "epoch": 65.44,
      "grad_norm": 2.6089932918548584,
      "learning_rate": 3.7698161031157856e-06,
      "loss": 0.5424,
      "step": 159870
    },
    {
      "epoch": 65.44,
      "grad_norm": 2.255347728729248,
      "learning_rate": 3.7696902132673347e-06,
      "loss": 0.5163,
      "step": 159880
    },
    {
      "epoch": 65.45,
      "grad_norm": 1.49593985080719,
      "learning_rate": 3.769564316886872e-06,
      "loss": 0.5245,
      "step": 159890
    },
    {
      "epoch": 65.45,
      "grad_norm": 2.3969614505767822,
      "learning_rate": 3.769438413974974e-06,
      "loss": 0.5176,
      "step": 159900
    },
    {
      "epoch": 65.46,
      "grad_norm": 1.838632345199585,
      "learning_rate": 3.7693125045322197e-06,
      "loss": 0.5164,
      "step": 159910
    },
    {
      "epoch": 65.46,
      "grad_norm": 2.1025991439819336,
      "learning_rate": 3.769186588559184e-06,
      "loss": 0.5077,
      "step": 159920
    },
    {
      "epoch": 65.46,
      "grad_norm": 1.870849609375,
      "learning_rate": 3.7690606660564436e-06,
      "loss": 0.5381,
      "step": 159930
    },
    {
      "epoch": 65.47,
      "grad_norm": 2.199915647506714,
      "learning_rate": 3.768934737024576e-06,
      "loss": 0.5126,
      "step": 159940
    },
    {
      "epoch": 65.47,
      "grad_norm": 2.2772560119628906,
      "learning_rate": 3.7688088014641583e-06,
      "loss": 0.5278,
      "step": 159950
    },
    {
      "epoch": 65.48,
      "grad_norm": 1.9343807697296143,
      "learning_rate": 3.768682859375767e-06,
      "loss": 0.5254,
      "step": 159960
    },
    {
      "epoch": 65.48,
      "grad_norm": 2.21915602684021,
      "learning_rate": 3.768556910759978e-06,
      "loss": 0.5308,
      "step": 159970
    },
    {
      "epoch": 65.49,
      "grad_norm": 2.1318161487579346,
      "learning_rate": 3.7684309556173692e-06,
      "loss": 0.5133,
      "step": 159980
    },
    {
      "epoch": 65.49,
      "grad_norm": 2.0348727703094482,
      "learning_rate": 3.7683049939485187e-06,
      "loss": 0.5242,
      "step": 159990
    },
    {
      "epoch": 65.49,
      "grad_norm": 2.3601887226104736,
      "learning_rate": 3.768179025754001e-06,
      "loss": 0.5348,
      "step": 160000
    },
    {
      "epoch": 65.5,
      "grad_norm": 1.815392017364502,
      "learning_rate": 3.7680530510343955e-06,
      "loss": 0.5104,
      "step": 160010
    },
    {
      "epoch": 65.5,
      "grad_norm": 1.9624706506729126,
      "learning_rate": 3.7679270697902777e-06,
      "loss": 0.5163,
      "step": 160020
    },
    {
      "epoch": 65.51,
      "grad_norm": 1.374582052230835,
      "learning_rate": 3.767801082022226e-06,
      "loss": 0.5226,
      "step": 160030
    },
    {
      "epoch": 65.51,
      "grad_norm": 2.1068274974823,
      "learning_rate": 3.767675087730817e-06,
      "loss": 0.5232,
      "step": 160040
    },
    {
      "epoch": 65.51,
      "grad_norm": 1.5067917108535767,
      "learning_rate": 3.7675490869166268e-06,
      "loss": 0.5036,
      "step": 160050
    },
    {
      "epoch": 65.52,
      "grad_norm": 1.7803311347961426,
      "learning_rate": 3.767423079580234e-06,
      "loss": 0.5092,
      "step": 160060
    },
    {
      "epoch": 65.52,
      "grad_norm": 2.4808743000030518,
      "learning_rate": 3.7672970657222148e-06,
      "loss": 0.4924,
      "step": 160070
    },
    {
      "epoch": 65.53,
      "grad_norm": 1.854393482208252,
      "learning_rate": 3.7671710453431475e-06,
      "loss": 0.517,
      "step": 160080
    },
    {
      "epoch": 65.53,
      "grad_norm": 1.5778898000717163,
      "learning_rate": 3.7670450184436087e-06,
      "loss": 0.5128,
      "step": 160090
    },
    {
      "epoch": 65.53,
      "grad_norm": 2.1059370040893555,
      "learning_rate": 3.766918985024176e-06,
      "loss": 0.5325,
      "step": 160100
    },
    {
      "epoch": 65.54,
      "grad_norm": 1.6118097305297852,
      "learning_rate": 3.766792945085426e-06,
      "loss": 0.5195,
      "step": 160110
    },
    {
      "epoch": 65.54,
      "grad_norm": 1.8326703310012817,
      "learning_rate": 3.7666668986279376e-06,
      "loss": 0.5284,
      "step": 160120
    },
    {
      "epoch": 65.55,
      "grad_norm": 2.1632800102233887,
      "learning_rate": 3.766540845652286e-06,
      "loss": 0.5137,
      "step": 160130
    },
    {
      "epoch": 65.55,
      "grad_norm": 2.0374200344085693,
      "learning_rate": 3.7664147861590513e-06,
      "loss": 0.5303,
      "step": 160140
    },
    {
      "epoch": 65.55,
      "grad_norm": 1.6745631694793701,
      "learning_rate": 3.766288720148809e-06,
      "loss": 0.5362,
      "step": 160150
    },
    {
      "epoch": 65.56,
      "grad_norm": 1.7551242113113403,
      "learning_rate": 3.766162647622137e-06,
      "loss": 0.5286,
      "step": 160160
    },
    {
      "epoch": 65.56,
      "grad_norm": 1.9402484893798828,
      "learning_rate": 3.766036568579613e-06,
      "loss": 0.4994,
      "step": 160170
    },
    {
      "epoch": 65.57,
      "grad_norm": 1.6000590324401855,
      "learning_rate": 3.7659104830218143e-06,
      "loss": 0.5076,
      "step": 160180
    },
    {
      "epoch": 65.57,
      "grad_norm": 1.752607822418213,
      "learning_rate": 3.7657843909493185e-06,
      "loss": 0.5275,
      "step": 160190
    },
    {
      "epoch": 65.58,
      "grad_norm": 1.7431803941726685,
      "learning_rate": 3.7656582923627034e-06,
      "loss": 0.4925,
      "step": 160200
    },
    {
      "epoch": 65.58,
      "grad_norm": 2.265411615371704,
      "learning_rate": 3.7655321872625474e-06,
      "loss": 0.5336,
      "step": 160210
    },
    {
      "epoch": 65.58,
      "grad_norm": 1.976563572883606,
      "learning_rate": 3.765406075649426e-06,
      "loss": 0.513,
      "step": 160220
    },
    {
      "epoch": 65.59,
      "grad_norm": 2.2940611839294434,
      "learning_rate": 3.7652799575239197e-06,
      "loss": 0.5274,
      "step": 160230
    },
    {
      "epoch": 65.59,
      "grad_norm": 2.160508155822754,
      "learning_rate": 3.7651538328866044e-06,
      "loss": 0.5328,
      "step": 160240
    },
    {
      "epoch": 65.6,
      "grad_norm": 2.059155225753784,
      "learning_rate": 3.7650277017380583e-06,
      "loss": 0.5362,
      "step": 160250
    },
    {
      "epoch": 65.6,
      "grad_norm": 1.728066086769104,
      "learning_rate": 3.764901564078859e-06,
      "loss": 0.5074,
      "step": 160260
    },
    {
      "epoch": 65.6,
      "grad_norm": 2.1085329055786133,
      "learning_rate": 3.7647754199095844e-06,
      "loss": 0.525,
      "step": 160270
    },
    {
      "epoch": 65.61,
      "grad_norm": 1.8234130144119263,
      "learning_rate": 3.7646492692308125e-06,
      "loss": 0.525,
      "step": 160280
    },
    {
      "epoch": 65.61,
      "grad_norm": 2.2810819149017334,
      "learning_rate": 3.7645231120431213e-06,
      "loss": 0.5169,
      "step": 160290
    },
    {
      "epoch": 65.62,
      "grad_norm": 2.578972578048706,
      "learning_rate": 3.764396948347088e-06,
      "loss": 0.5245,
      "step": 160300
    },
    {
      "epoch": 65.62,
      "grad_norm": 1.6731674671173096,
      "learning_rate": 3.764270778143292e-06,
      "loss": 0.5142,
      "step": 160310
    },
    {
      "epoch": 65.62,
      "grad_norm": 1.4967045783996582,
      "learning_rate": 3.7641446014323095e-06,
      "loss": 0.5542,
      "step": 160320
    },
    {
      "epoch": 65.63,
      "grad_norm": 2.413766384124756,
      "learning_rate": 3.7640184182147194e-06,
      "loss": 0.5473,
      "step": 160330
    },
    {
      "epoch": 65.63,
      "grad_norm": 2.1620566844940186,
      "learning_rate": 3.7638922284911e-06,
      "loss": 0.5213,
      "step": 160340
    },
    {
      "epoch": 65.64,
      "grad_norm": 2.3012185096740723,
      "learning_rate": 3.7637660322620287e-06,
      "loss": 0.5261,
      "step": 160350
    },
    {
      "epoch": 65.64,
      "grad_norm": 1.7017109394073486,
      "learning_rate": 3.763639829528084e-06,
      "loss": 0.5113,
      "step": 160360
    },
    {
      "epoch": 65.64,
      "grad_norm": 1.8174209594726562,
      "learning_rate": 3.7635136202898436e-06,
      "loss": 0.5391,
      "step": 160370
    },
    {
      "epoch": 65.65,
      "grad_norm": 1.5174411535263062,
      "learning_rate": 3.7633874045478858e-06,
      "loss": 0.5243,
      "step": 160380
    },
    {
      "epoch": 65.65,
      "grad_norm": 2.2132372856140137,
      "learning_rate": 3.7632611823027893e-06,
      "loss": 0.5212,
      "step": 160390
    },
    {
      "epoch": 65.66,
      "grad_norm": 2.1328279972076416,
      "learning_rate": 3.7631349535551315e-06,
      "loss": 0.5128,
      "step": 160400
    },
    {
      "epoch": 65.66,
      "grad_norm": 2.1462011337280273,
      "learning_rate": 3.7630087183054916e-06,
      "loss": 0.5254,
      "step": 160410
    },
    {
      "epoch": 65.67,
      "grad_norm": 1.8725584745407104,
      "learning_rate": 3.7628824765544474e-06,
      "loss": 0.4934,
      "step": 160420
    },
    {
      "epoch": 65.67,
      "grad_norm": 1.9953992366790771,
      "learning_rate": 3.762756228302577e-06,
      "loss": 0.5229,
      "step": 160430
    },
    {
      "epoch": 65.67,
      "grad_norm": 1.285796046257019,
      "learning_rate": 3.7626299735504583e-06,
      "loss": 0.5279,
      "step": 160440
    },
    {
      "epoch": 65.68,
      "grad_norm": 1.9859158992767334,
      "learning_rate": 3.7625037122986713e-06,
      "loss": 0.5105,
      "step": 160450
    },
    {
      "epoch": 65.68,
      "grad_norm": 1.9743437767028809,
      "learning_rate": 3.762377444547792e-06,
      "loss": 0.5105,
      "step": 160460
    },
    {
      "epoch": 65.69,
      "grad_norm": 2.948133945465088,
      "learning_rate": 3.7622511702984005e-06,
      "loss": 0.5256,
      "step": 160470
    },
    {
      "epoch": 65.69,
      "grad_norm": 1.5331958532333374,
      "learning_rate": 3.762124889551075e-06,
      "loss": 0.5197,
      "step": 160480
    },
    {
      "epoch": 65.69,
      "grad_norm": 1.7571991682052612,
      "learning_rate": 3.761998602306394e-06,
      "loss": 0.523,
      "step": 160490
    },
    {
      "epoch": 65.7,
      "grad_norm": 2.3156094551086426,
      "learning_rate": 3.7618723085649353e-06,
      "loss": 0.5061,
      "step": 160500
    },
    {
      "epoch": 65.7,
      "grad_norm": 1.7426714897155762,
      "learning_rate": 3.7617460083272782e-06,
      "loss": 0.5318,
      "step": 160510
    },
    {
      "epoch": 65.71,
      "grad_norm": 2.10658597946167,
      "learning_rate": 3.761619701594001e-06,
      "loss": 0.5073,
      "step": 160520
    },
    {
      "epoch": 65.71,
      "grad_norm": 1.6788039207458496,
      "learning_rate": 3.761493388365682e-06,
      "loss": 0.5198,
      "step": 160530
    },
    {
      "epoch": 65.71,
      "grad_norm": 1.698270320892334,
      "learning_rate": 3.7613670686429e-06,
      "loss": 0.525,
      "step": 160540
    },
    {
      "epoch": 65.72,
      "grad_norm": 1.8095710277557373,
      "learning_rate": 3.7612407424262345e-06,
      "loss": 0.5179,
      "step": 160550
    },
    {
      "epoch": 65.72,
      "grad_norm": 2.085430860519409,
      "learning_rate": 3.7611144097162624e-06,
      "loss": 0.5073,
      "step": 160560
    },
    {
      "epoch": 65.73,
      "grad_norm": 1.8931721448898315,
      "learning_rate": 3.760988070513564e-06,
      "loss": 0.5086,
      "step": 160570
    },
    {
      "epoch": 65.73,
      "grad_norm": 2.161367177963257,
      "learning_rate": 3.760861724818718e-06,
      "loss": 0.5095,
      "step": 160580
    },
    {
      "epoch": 65.73,
      "grad_norm": 1.9492826461791992,
      "learning_rate": 3.760735372632302e-06,
      "loss": 0.5175,
      "step": 160590
    },
    {
      "epoch": 65.74,
      "grad_norm": 2.3809072971343994,
      "learning_rate": 3.7606090139548954e-06,
      "loss": 0.5414,
      "step": 160600
    },
    {
      "epoch": 65.74,
      "grad_norm": 2.1043002605438232,
      "learning_rate": 3.7604826487870774e-06,
      "loss": 0.5375,
      "step": 160610
    },
    {
      "epoch": 65.75,
      "grad_norm": 2.7296206951141357,
      "learning_rate": 3.7603562771294263e-06,
      "loss": 0.5364,
      "step": 160620
    },
    {
      "epoch": 65.75,
      "grad_norm": 2.2247865200042725,
      "learning_rate": 3.7602298989825216e-06,
      "loss": 0.5141,
      "step": 160630
    },
    {
      "epoch": 65.76,
      "grad_norm": 1.7630302906036377,
      "learning_rate": 3.7601035143469418e-06,
      "loss": 0.5007,
      "step": 160640
    },
    {
      "epoch": 65.76,
      "grad_norm": 1.4094973802566528,
      "learning_rate": 3.7599771232232653e-06,
      "loss": 0.5323,
      "step": 160650
    },
    {
      "epoch": 65.76,
      "grad_norm": 2.0631825923919678,
      "learning_rate": 3.759850725612072e-06,
      "loss": 0.5058,
      "step": 160660
    },
    {
      "epoch": 65.77,
      "grad_norm": 1.5139726400375366,
      "learning_rate": 3.759724321513941e-06,
      "loss": 0.4892,
      "step": 160670
    },
    {
      "epoch": 65.77,
      "grad_norm": 1.9421958923339844,
      "learning_rate": 3.7595979109294507e-06,
      "loss": 0.5153,
      "step": 160680
    },
    {
      "epoch": 65.78,
      "grad_norm": 2.0361547470092773,
      "learning_rate": 3.75947149385918e-06,
      "loss": 0.5195,
      "step": 160690
    },
    {
      "epoch": 65.78,
      "grad_norm": 1.9076507091522217,
      "learning_rate": 3.7593450703037087e-06,
      "loss": 0.512,
      "step": 160700
    },
    {
      "epoch": 65.78,
      "grad_norm": 1.9644118547439575,
      "learning_rate": 3.7592186402636164e-06,
      "loss": 0.5087,
      "step": 160710
    },
    {
      "epoch": 65.79,
      "grad_norm": 1.7317613363265991,
      "learning_rate": 3.7590922037394804e-06,
      "loss": 0.5119,
      "step": 160720
    },
    {
      "epoch": 65.79,
      "grad_norm": 1.7036908864974976,
      "learning_rate": 3.7589657607318807e-06,
      "loss": 0.5169,
      "step": 160730
    },
    {
      "epoch": 65.8,
      "grad_norm": 1.8773236274719238,
      "learning_rate": 3.7588393112413975e-06,
      "loss": 0.5474,
      "step": 160740
    },
    {
      "epoch": 65.8,
      "grad_norm": 1.6794438362121582,
      "learning_rate": 3.7587128552686092e-06,
      "loss": 0.5213,
      "step": 160750
    },
    {
      "epoch": 65.8,
      "grad_norm": 1.7268445491790771,
      "learning_rate": 3.7585863928140953e-06,
      "loss": 0.5338,
      "step": 160760
    },
    {
      "epoch": 65.81,
      "grad_norm": 1.9863386154174805,
      "learning_rate": 3.7584599238784347e-06,
      "loss": 0.5422,
      "step": 160770
    },
    {
      "epoch": 65.81,
      "grad_norm": 2.4939258098602295,
      "learning_rate": 3.758333448462207e-06,
      "loss": 0.5345,
      "step": 160780
    },
    {
      "epoch": 65.82,
      "grad_norm": 1.6942665576934814,
      "learning_rate": 3.758206966565992e-06,
      "loss": 0.5331,
      "step": 160790
    },
    {
      "epoch": 65.82,
      "grad_norm": 1.6922852993011475,
      "learning_rate": 3.7580804781903683e-06,
      "loss": 0.5179,
      "step": 160800
    },
    {
      "epoch": 65.82,
      "grad_norm": 2.208866834640503,
      "learning_rate": 3.757953983335916e-06,
      "loss": 0.5171,
      "step": 160810
    },
    {
      "epoch": 65.83,
      "grad_norm": 2.3139498233795166,
      "learning_rate": 3.7578274820032143e-06,
      "loss": 0.5201,
      "step": 160820
    },
    {
      "epoch": 65.83,
      "grad_norm": 2.3302419185638428,
      "learning_rate": 3.757700974192843e-06,
      "loss": 0.5118,
      "step": 160830
    },
    {
      "epoch": 65.84,
      "grad_norm": 2.1244521141052246,
      "learning_rate": 3.75757445990538e-06,
      "loss": 0.5003,
      "step": 160840
    },
    {
      "epoch": 65.84,
      "grad_norm": 2.1460413932800293,
      "learning_rate": 3.7574479391414072e-06,
      "loss": 0.5204,
      "step": 160850
    },
    {
      "epoch": 65.85,
      "grad_norm": 2.5159685611724854,
      "learning_rate": 3.7573214119015025e-06,
      "loss": 0.5174,
      "step": 160860
    },
    {
      "epoch": 65.85,
      "grad_norm": 1.4597959518432617,
      "learning_rate": 3.7571948781862465e-06,
      "loss": 0.5101,
      "step": 160870
    },
    {
      "epoch": 65.85,
      "grad_norm": 1.4614347219467163,
      "learning_rate": 3.757068337996218e-06,
      "loss": 0.519,
      "step": 160880
    },
    {
      "epoch": 65.86,
      "grad_norm": 2.040304660797119,
      "learning_rate": 3.7569417913319976e-06,
      "loss": 0.5537,
      "step": 160890
    },
    {
      "epoch": 65.86,
      "grad_norm": 1.7127131223678589,
      "learning_rate": 3.756815238194164e-06,
      "loss": 0.5327,
      "step": 160900
    },
    {
      "epoch": 65.87,
      "grad_norm": 1.7492133378982544,
      "learning_rate": 3.7566886785832976e-06,
      "loss": 0.5004,
      "step": 160910
    },
    {
      "epoch": 65.87,
      "grad_norm": 2.0356016159057617,
      "learning_rate": 3.756562112499978e-06,
      "loss": 0.5214,
      "step": 160920
    },
    {
      "epoch": 65.87,
      "grad_norm": 1.6583337783813477,
      "learning_rate": 3.7564355399447848e-06,
      "loss": 0.5268,
      "step": 160930
    },
    {
      "epoch": 65.88,
      "grad_norm": 1.8120691776275635,
      "learning_rate": 3.756308960918298e-06,
      "loss": 0.5258,
      "step": 160940
    },
    {
      "epoch": 65.88,
      "grad_norm": 1.7281568050384521,
      "learning_rate": 3.7561823754210968e-06,
      "loss": 0.5225,
      "step": 160950
    },
    {
      "epoch": 65.89,
      "grad_norm": 2.0692763328552246,
      "learning_rate": 3.7560557834537625e-06,
      "loss": 0.5101,
      "step": 160960
    },
    {
      "epoch": 65.89,
      "grad_norm": 2.37337589263916,
      "learning_rate": 3.7559291850168744e-06,
      "loss": 0.5248,
      "step": 160970
    },
    {
      "epoch": 65.89,
      "grad_norm": 2.1072897911071777,
      "learning_rate": 3.755802580111011e-06,
      "loss": 0.5266,
      "step": 160980
    },
    {
      "epoch": 65.9,
      "grad_norm": 1.9085707664489746,
      "learning_rate": 3.755675968736754e-06,
      "loss": 0.5113,
      "step": 160990
    },
    {
      "epoch": 65.9,
      "grad_norm": 2.290072202682495,
      "learning_rate": 3.7555493508946827e-06,
      "loss": 0.5155,
      "step": 161000
    },
    {
      "epoch": 65.91,
      "grad_norm": 1.6654584407806396,
      "learning_rate": 3.7554227265853774e-06,
      "loss": 0.5168,
      "step": 161010
    },
    {
      "epoch": 65.91,
      "grad_norm": 2.2759411334991455,
      "learning_rate": 3.7552960958094177e-06,
      "loss": 0.5191,
      "step": 161020
    },
    {
      "epoch": 65.91,
      "grad_norm": 2.5385522842407227,
      "learning_rate": 3.7551694585673838e-06,
      "loss": 0.5148,
      "step": 161030
    },
    {
      "epoch": 65.92,
      "grad_norm": 2.178891897201538,
      "learning_rate": 3.7550428148598563e-06,
      "loss": 0.5278,
      "step": 161040
    },
    {
      "epoch": 65.92,
      "grad_norm": 1.9366703033447266,
      "learning_rate": 3.7549161646874147e-06,
      "loss": 0.5286,
      "step": 161050
    },
    {
      "epoch": 65.93,
      "grad_norm": 1.8144599199295044,
      "learning_rate": 3.75478950805064e-06,
      "loss": 0.5295,
      "step": 161060
    },
    {
      "epoch": 65.93,
      "grad_norm": 1.517522931098938,
      "learning_rate": 3.754662844950111e-06,
      "loss": 0.5331,
      "step": 161070
    },
    {
      "epoch": 65.94,
      "grad_norm": 2.1124441623687744,
      "learning_rate": 3.7545361753864095e-06,
      "loss": 0.5132,
      "step": 161080
    },
    {
      "epoch": 65.94,
      "grad_norm": 2.713104724884033,
      "learning_rate": 3.754409499360115e-06,
      "loss": 0.5016,
      "step": 161090
    },
    {
      "epoch": 65.94,
      "grad_norm": 2.2603869438171387,
      "learning_rate": 3.7542828168718075e-06,
      "loss": 0.5083,
      "step": 161100
    },
    {
      "epoch": 65.95,
      "grad_norm": 2.513066291809082,
      "learning_rate": 3.754156127922068e-06,
      "loss": 0.5057,
      "step": 161110
    },
    {
      "epoch": 65.95,
      "grad_norm": 1.718752145767212,
      "learning_rate": 3.754029432511476e-06,
      "loss": 0.5051,
      "step": 161120
    },
    {
      "epoch": 65.96,
      "grad_norm": 1.572532296180725,
      "learning_rate": 3.753902730640612e-06,
      "loss": 0.5281,
      "step": 161130
    },
    {
      "epoch": 65.96,
      "grad_norm": 1.7328671216964722,
      "learning_rate": 3.7537760223100573e-06,
      "loss": 0.5206,
      "step": 161140
    },
    {
      "epoch": 65.96,
      "grad_norm": 1.9733197689056396,
      "learning_rate": 3.7536493075203922e-06,
      "loss": 0.5312,
      "step": 161150
    },
    {
      "epoch": 65.97,
      "grad_norm": 2.4460673332214355,
      "learning_rate": 3.753522586272196e-06,
      "loss": 0.534,
      "step": 161160
    },
    {
      "epoch": 65.97,
      "grad_norm": 2.1492135524749756,
      "learning_rate": 3.7533958585660496e-06,
      "loss": 0.4978,
      "step": 161170
    },
    {
      "epoch": 65.98,
      "grad_norm": 1.7419095039367676,
      "learning_rate": 3.753269124402535e-06,
      "loss": 0.5183,
      "step": 161180
    },
    {
      "epoch": 65.98,
      "grad_norm": 1.865354299545288,
      "learning_rate": 3.753142383782231e-06,
      "loss": 0.5256,
      "step": 161190
    },
    {
      "epoch": 65.98,
      "grad_norm": 2.517225742340088,
      "learning_rate": 3.7530156367057185e-06,
      "loss": 0.5292,
      "step": 161200
    },
    {
      "epoch": 65.99,
      "grad_norm": 2.0522661209106445,
      "learning_rate": 3.7528888831735783e-06,
      "loss": 0.517,
      "step": 161210
    },
    {
      "epoch": 65.99,
      "grad_norm": 1.7383190393447876,
      "learning_rate": 3.7527621231863915e-06,
      "loss": 0.4993,
      "step": 161220
    },
    {
      "epoch": 66.0,
      "grad_norm": 1.6500543355941772,
      "learning_rate": 3.752635356744738e-06,
      "loss": 0.5507,
      "step": 161230
    },
    {
      "epoch": 66.0,
      "eval_loss": 0.5209594368934631,
      "eval_runtime": 52.2463,
      "eval_samples_per_second": 66.014,
      "eval_steps_per_second": 8.269,
      "step": 161238
    },
    {
      "epoch": 66.0,
      "grad_norm": 1.7050164937973022,
      "learning_rate": 3.7525085838491988e-06,
      "loss": 0.501,
      "step": 161240
    },
    {
      "epoch": 66.0,
      "grad_norm": 2.7078537940979004,
      "learning_rate": 3.752381804500355e-06,
      "loss": 0.5146,
      "step": 161250
    },
    {
      "epoch": 66.01,
      "grad_norm": 1.8335692882537842,
      "learning_rate": 3.7522550186987875e-06,
      "loss": 0.527,
      "step": 161260
    },
    {
      "epoch": 66.01,
      "grad_norm": 1.829372763633728,
      "learning_rate": 3.752128226445076e-06,
      "loss": 0.5217,
      "step": 161270
    },
    {
      "epoch": 66.02,
      "grad_norm": 1.9517320394515991,
      "learning_rate": 3.752001427739802e-06,
      "loss": 0.507,
      "step": 161280
    },
    {
      "epoch": 66.02,
      "grad_norm": 1.8141040802001953,
      "learning_rate": 3.7518746225835465e-06,
      "loss": 0.5267,
      "step": 161290
    },
    {
      "epoch": 66.03,
      "grad_norm": 1.667573094367981,
      "learning_rate": 3.7517478109768903e-06,
      "loss": 0.5219,
      "step": 161300
    },
    {
      "epoch": 66.03,
      "grad_norm": 2.101719379425049,
      "learning_rate": 3.751620992920414e-06,
      "loss": 0.5116,
      "step": 161310
    },
    {
      "epoch": 66.03,
      "grad_norm": 1.9858441352844238,
      "learning_rate": 3.751494168414699e-06,
      "loss": 0.5262,
      "step": 161320
    },
    {
      "epoch": 66.04,
      "grad_norm": 1.877804160118103,
      "learning_rate": 3.7513673374603253e-06,
      "loss": 0.4942,
      "step": 161330
    },
    {
      "epoch": 66.04,
      "grad_norm": 1.8046300411224365,
      "learning_rate": 3.751240500057875e-06,
      "loss": 0.4997,
      "step": 161340
    },
    {
      "epoch": 66.05,
      "grad_norm": 1.6464478969573975,
      "learning_rate": 3.7511136562079288e-06,
      "loss": 0.525,
      "step": 161350
    },
    {
      "epoch": 66.05,
      "grad_norm": 1.6554739475250244,
      "learning_rate": 3.7509868059110674e-06,
      "loss": 0.5334,
      "step": 161360
    },
    {
      "epoch": 66.05,
      "grad_norm": 1.5163322687149048,
      "learning_rate": 3.7508599491678724e-06,
      "loss": 0.5207,
      "step": 161370
    },
    {
      "epoch": 66.06,
      "grad_norm": 1.9661401510238647,
      "learning_rate": 3.750733085978925e-06,
      "loss": 0.5328,
      "step": 161380
    },
    {
      "epoch": 66.06,
      "grad_norm": 1.6433913707733154,
      "learning_rate": 3.750606216344805e-06,
      "loss": 0.523,
      "step": 161390
    },
    {
      "epoch": 66.07,
      "grad_norm": 1.9054954051971436,
      "learning_rate": 3.7504793402660953e-06,
      "loss": 0.5352,
      "step": 161400
    },
    {
      "epoch": 66.07,
      "grad_norm": 1.7367595434188843,
      "learning_rate": 3.7503524577433755e-06,
      "loss": 0.5173,
      "step": 161410
    },
    {
      "epoch": 66.07,
      "grad_norm": 1.9734525680541992,
      "learning_rate": 3.7502255687772286e-06,
      "loss": 0.5309,
      "step": 161420
    },
    {
      "epoch": 66.08,
      "grad_norm": 1.6909300088882446,
      "learning_rate": 3.7500986733682345e-06,
      "loss": 0.5107,
      "step": 161430
    },
    {
      "epoch": 66.08,
      "grad_norm": 1.9628175497055054,
      "learning_rate": 3.7499717715169754e-06,
      "loss": 0.5191,
      "step": 161440
    },
    {
      "epoch": 66.09,
      "grad_norm": 1.3540505170822144,
      "learning_rate": 3.7498448632240316e-06,
      "loss": 0.5297,
      "step": 161450
    },
    {
      "epoch": 66.09,
      "grad_norm": 1.6158506870269775,
      "learning_rate": 3.7497179484899854e-06,
      "loss": 0.5276,
      "step": 161460
    },
    {
      "epoch": 66.09,
      "grad_norm": 2.541893720626831,
      "learning_rate": 3.7495910273154175e-06,
      "loss": 0.5259,
      "step": 161470
    },
    {
      "epoch": 66.1,
      "grad_norm": 1.9474446773529053,
      "learning_rate": 3.7494640997009098e-06,
      "loss": 0.5462,
      "step": 161480
    },
    {
      "epoch": 66.1,
      "grad_norm": 2.6627607345581055,
      "learning_rate": 3.7493371656470432e-06,
      "loss": 0.5271,
      "step": 161490
    },
    {
      "epoch": 66.11,
      "grad_norm": 2.190999984741211,
      "learning_rate": 3.7492102251544e-06,
      "loss": 0.5177,
      "step": 161500
    },
    {
      "epoch": 66.11,
      "grad_norm": 2.000091075897217,
      "learning_rate": 3.7490832782235603e-06,
      "loss": 0.5016,
      "step": 161510
    },
    {
      "epoch": 66.12,
      "grad_norm": 1.6158391237258911,
      "learning_rate": 3.748956324855107e-06,
      "loss": 0.5191,
      "step": 161520
    },
    {
      "epoch": 66.12,
      "grad_norm": 2.0075442790985107,
      "learning_rate": 3.748829365049621e-06,
      "loss": 0.5099,
      "step": 161530
    },
    {
      "epoch": 66.12,
      "grad_norm": 2.0169804096221924,
      "learning_rate": 3.7487023988076843e-06,
      "loss": 0.5077,
      "step": 161540
    },
    {
      "epoch": 66.13,
      "grad_norm": 2.6241414546966553,
      "learning_rate": 3.7485754261298777e-06,
      "loss": 0.513,
      "step": 161550
    },
    {
      "epoch": 66.13,
      "grad_norm": 1.4966152906417847,
      "learning_rate": 3.7484484470167844e-06,
      "loss": 0.5375,
      "step": 161560
    },
    {
      "epoch": 66.14,
      "grad_norm": 1.8487598896026611,
      "learning_rate": 3.7483214614689844e-06,
      "loss": 0.5083,
      "step": 161570
    },
    {
      "epoch": 66.14,
      "grad_norm": 1.8149845600128174,
      "learning_rate": 3.7481944694870597e-06,
      "loss": 0.5334,
      "step": 161580
    },
    {
      "epoch": 66.14,
      "grad_norm": 1.896308183670044,
      "learning_rate": 3.7480674710715938e-06,
      "loss": 0.5132,
      "step": 161590
    },
    {
      "epoch": 66.15,
      "grad_norm": 1.5436452627182007,
      "learning_rate": 3.747940466223165e-06,
      "loss": 0.5173,
      "step": 161600
    },
    {
      "epoch": 66.15,
      "grad_norm": 1.7797625064849854,
      "learning_rate": 3.7478134549423587e-06,
      "loss": 0.5439,
      "step": 161610
    },
    {
      "epoch": 66.16,
      "grad_norm": 1.5950238704681396,
      "learning_rate": 3.7476864372297547e-06,
      "loss": 0.5231,
      "step": 161620
    },
    {
      "epoch": 66.16,
      "grad_norm": 2.0843403339385986,
      "learning_rate": 3.747559413085935e-06,
      "loss": 0.5312,
      "step": 161630
    },
    {
      "epoch": 66.16,
      "grad_norm": 1.8038420677185059,
      "learning_rate": 3.7474323825114824e-06,
      "loss": 0.5148,
      "step": 161640
    },
    {
      "epoch": 66.17,
      "grad_norm": 1.8828322887420654,
      "learning_rate": 3.747305345506978e-06,
      "loss": 0.4994,
      "step": 161650
    },
    {
      "epoch": 66.17,
      "grad_norm": 1.476609468460083,
      "learning_rate": 3.7471783020730034e-06,
      "loss": 0.5107,
      "step": 161660
    },
    {
      "epoch": 66.18,
      "grad_norm": 1.8080319166183472,
      "learning_rate": 3.7470512522101415e-06,
      "loss": 0.51,
      "step": 161670
    },
    {
      "epoch": 66.18,
      "grad_norm": 1.8829355239868164,
      "learning_rate": 3.7469241959189743e-06,
      "loss": 0.5197,
      "step": 161680
    },
    {
      "epoch": 66.19,
      "grad_norm": 2.1230640411376953,
      "learning_rate": 3.746797133200083e-06,
      "loss": 0.5284,
      "step": 161690
    },
    {
      "epoch": 66.19,
      "grad_norm": 2.178161382675171,
      "learning_rate": 3.7466700640540495e-06,
      "loss": 0.5108,
      "step": 161700
    },
    {
      "epoch": 66.19,
      "grad_norm": 2.456211805343628,
      "learning_rate": 3.7465429884814566e-06,
      "loss": 0.5178,
      "step": 161710
    },
    {
      "epoch": 66.2,
      "grad_norm": 1.8229655027389526,
      "learning_rate": 3.7464159064828877e-06,
      "loss": 0.5144,
      "step": 161720
    },
    {
      "epoch": 66.2,
      "grad_norm": 1.9162583351135254,
      "learning_rate": 3.7462888180589216e-06,
      "loss": 0.5291,
      "step": 161730
    },
    {
      "epoch": 66.21,
      "grad_norm": 1.7229281663894653,
      "learning_rate": 3.7461617232101438e-06,
      "loss": 0.5162,
      "step": 161740
    },
    {
      "epoch": 66.21,
      "grad_norm": 2.447347640991211,
      "learning_rate": 3.7460346219371347e-06,
      "loss": 0.5355,
      "step": 161750
    },
    {
      "epoch": 66.21,
      "grad_norm": 3.7714343070983887,
      "learning_rate": 3.7459075142404764e-06,
      "loss": 0.5122,
      "step": 161760
    },
    {
      "epoch": 66.22,
      "grad_norm": 1.6414997577667236,
      "learning_rate": 3.745780400120752e-06,
      "loss": 0.5249,
      "step": 161770
    },
    {
      "epoch": 66.22,
      "grad_norm": 1.905366063117981,
      "learning_rate": 3.745653279578544e-06,
      "loss": 0.5109,
      "step": 161780
    },
    {
      "epoch": 66.23,
      "grad_norm": 2.2827231884002686,
      "learning_rate": 3.745526152614433e-06,
      "loss": 0.5197,
      "step": 161790
    },
    {
      "epoch": 66.23,
      "grad_norm": 1.8383525609970093,
      "learning_rate": 3.7453990192290035e-06,
      "loss": 0.5019,
      "step": 161800
    },
    {
      "epoch": 66.23,
      "grad_norm": 1.6410248279571533,
      "learning_rate": 3.745271879422837e-06,
      "loss": 0.5135,
      "step": 161810
    },
    {
      "epoch": 66.24,
      "grad_norm": 1.9579154253005981,
      "learning_rate": 3.7451447331965158e-06,
      "loss": 0.5124,
      "step": 161820
    },
    {
      "epoch": 66.24,
      "grad_norm": 1.6674606800079346,
      "learning_rate": 3.7450175805506218e-06,
      "loss": 0.5159,
      "step": 161830
    },
    {
      "epoch": 66.25,
      "grad_norm": 1.933714747428894,
      "learning_rate": 3.744890421485738e-06,
      "loss": 0.5308,
      "step": 161840
    },
    {
      "epoch": 66.25,
      "grad_norm": 1.7262871265411377,
      "learning_rate": 3.7447632560024477e-06,
      "loss": 0.5155,
      "step": 161850
    },
    {
      "epoch": 66.25,
      "grad_norm": 2.0524656772613525,
      "learning_rate": 3.7446360841013315e-06,
      "loss": 0.5592,
      "step": 161860
    },
    {
      "epoch": 66.26,
      "grad_norm": 1.697703242301941,
      "learning_rate": 3.744508905782974e-06,
      "loss": 0.5234,
      "step": 161870
    },
    {
      "epoch": 66.26,
      "grad_norm": 1.9134451150894165,
      "learning_rate": 3.7443817210479562e-06,
      "loss": 0.5137,
      "step": 161880
    },
    {
      "epoch": 66.27,
      "grad_norm": 2.382877826690674,
      "learning_rate": 3.7442545298968618e-06,
      "loss": 0.5182,
      "step": 161890
    },
    {
      "epoch": 66.27,
      "grad_norm": 2.269325017929077,
      "learning_rate": 3.7441273323302737e-06,
      "loss": 0.5112,
      "step": 161900
    },
    {
      "epoch": 66.28,
      "grad_norm": 2.184312343597412,
      "learning_rate": 3.7440001283487735e-06,
      "loss": 0.5477,
      "step": 161910
    },
    {
      "epoch": 66.28,
      "grad_norm": 1.9081894159317017,
      "learning_rate": 3.7438729179529436e-06,
      "loss": 0.5407,
      "step": 161920
    },
    {
      "epoch": 66.28,
      "grad_norm": 1.7218971252441406,
      "learning_rate": 3.7437457011433683e-06,
      "loss": 0.5411,
      "step": 161930
    },
    {
      "epoch": 66.29,
      "grad_norm": 1.775390863418579,
      "learning_rate": 3.7436184779206296e-06,
      "loss": 0.5294,
      "step": 161940
    },
    {
      "epoch": 66.29,
      "grad_norm": 2.4251294136047363,
      "learning_rate": 3.7434912482853093e-06,
      "loss": 0.5189,
      "step": 161950
    },
    {
      "epoch": 66.3,
      "grad_norm": 2.0600173473358154,
      "learning_rate": 3.7433640122379916e-06,
      "loss": 0.5213,
      "step": 161960
    },
    {
      "epoch": 66.3,
      "grad_norm": 1.632379174232483,
      "learning_rate": 3.743236769779259e-06,
      "loss": 0.5193,
      "step": 161970
    },
    {
      "epoch": 66.3,
      "grad_norm": 1.6354812383651733,
      "learning_rate": 3.7431095209096943e-06,
      "loss": 0.5219,
      "step": 161980
    },
    {
      "epoch": 66.31,
      "grad_norm": 2.256479501724243,
      "learning_rate": 3.7429822656298795e-06,
      "loss": 0.5229,
      "step": 161990
    },
    {
      "epoch": 66.31,
      "grad_norm": 2.3407034873962402,
      "learning_rate": 3.7428550039403994e-06,
      "loss": 0.5341,
      "step": 162000
    },
    {
      "epoch": 66.32,
      "grad_norm": 1.8847897052764893,
      "learning_rate": 3.7427277358418357e-06,
      "loss": 0.5037,
      "step": 162010
    },
    {
      "epoch": 66.32,
      "grad_norm": 2.4451005458831787,
      "learning_rate": 3.7426004613347714e-06,
      "loss": 0.5059,
      "step": 162020
    },
    {
      "epoch": 66.32,
      "grad_norm": 2.2532432079315186,
      "learning_rate": 3.74247318041979e-06,
      "loss": 0.5252,
      "step": 162030
    },
    {
      "epoch": 66.33,
      "grad_norm": 2.1348769664764404,
      "learning_rate": 3.742345893097474e-06,
      "loss": 0.527,
      "step": 162040
    },
    {
      "epoch": 66.33,
      "grad_norm": 1.6023895740509033,
      "learning_rate": 3.7422185993684074e-06,
      "loss": 0.4872,
      "step": 162050
    },
    {
      "epoch": 66.34,
      "grad_norm": 1.9470175504684448,
      "learning_rate": 3.7420912992331725e-06,
      "loss": 0.5248,
      "step": 162060
    },
    {
      "epoch": 66.34,
      "grad_norm": 1.6290168762207031,
      "learning_rate": 3.7419639926923526e-06,
      "loss": 0.5181,
      "step": 162070
    },
    {
      "epoch": 66.34,
      "grad_norm": 1.551485538482666,
      "learning_rate": 3.7418366797465304e-06,
      "loss": 0.528,
      "step": 162080
    },
    {
      "epoch": 66.35,
      "grad_norm": 1.8660162687301636,
      "learning_rate": 3.7417093603962905e-06,
      "loss": 0.5217,
      "step": 162090
    },
    {
      "epoch": 66.35,
      "grad_norm": 1.9788849353790283,
      "learning_rate": 3.741582034642215e-06,
      "loss": 0.5171,
      "step": 162100
    },
    {
      "epoch": 66.36,
      "grad_norm": 2.3323755264282227,
      "learning_rate": 3.7414547024848877e-06,
      "loss": 0.5254,
      "step": 162110
    },
    {
      "epoch": 66.36,
      "grad_norm": 2.195263385772705,
      "learning_rate": 3.7413273639248913e-06,
      "loss": 0.5017,
      "step": 162120
    },
    {
      "epoch": 66.37,
      "grad_norm": 1.9764469861984253,
      "learning_rate": 3.7412000189628094e-06,
      "loss": 0.5224,
      "step": 162130
    },
    {
      "epoch": 66.37,
      "grad_norm": 1.6750379800796509,
      "learning_rate": 3.741072667599225e-06,
      "loss": 0.5205,
      "step": 162140
    },
    {
      "epoch": 66.37,
      "grad_norm": 1.8337970972061157,
      "learning_rate": 3.7409453098347226e-06,
      "loss": 0.5484,
      "step": 162150
    },
    {
      "epoch": 66.38,
      "grad_norm": 2.703272581100464,
      "learning_rate": 3.7408179456698845e-06,
      "loss": 0.4927,
      "step": 162160
    },
    {
      "epoch": 66.38,
      "grad_norm": 1.844727873802185,
      "learning_rate": 3.740690575105294e-06,
      "loss": 0.523,
      "step": 162170
    },
    {
      "epoch": 66.39,
      "grad_norm": 2.0743441581726074,
      "learning_rate": 3.7405631981415358e-06,
      "loss": 0.5088,
      "step": 162180
    },
    {
      "epoch": 66.39,
      "grad_norm": 1.5064594745635986,
      "learning_rate": 3.740435814779193e-06,
      "loss": 0.5249,
      "step": 162190
    },
    {
      "epoch": 66.39,
      "grad_norm": 1.9022363424301147,
      "learning_rate": 3.7403084250188477e-06,
      "loss": 0.5077,
      "step": 162200
    },
    {
      "epoch": 66.4,
      "grad_norm": 2.1161277294158936,
      "learning_rate": 3.7401810288610852e-06,
      "loss": 0.539,
      "step": 162210
    },
    {
      "epoch": 66.4,
      "grad_norm": 2.0256197452545166,
      "learning_rate": 3.740053626306488e-06,
      "loss": 0.5204,
      "step": 162220
    },
    {
      "epoch": 66.41,
      "grad_norm": 2.0753040313720703,
      "learning_rate": 3.7399262173556413e-06,
      "loss": 0.532,
      "step": 162230
    },
    {
      "epoch": 66.41,
      "grad_norm": 2.167173385620117,
      "learning_rate": 3.739798802009126e-06,
      "loss": 0.5177,
      "step": 162240
    },
    {
      "epoch": 66.41,
      "grad_norm": 2.029301881790161,
      "learning_rate": 3.739671380267528e-06,
      "loss": 0.5222,
      "step": 162250
    },
    {
      "epoch": 66.42,
      "grad_norm": 1.5835984945297241,
      "learning_rate": 3.739543952131431e-06,
      "loss": 0.5213,
      "step": 162260
    },
    {
      "epoch": 66.42,
      "grad_norm": 2.6785264015197754,
      "learning_rate": 3.7394165176014173e-06,
      "loss": 0.5462,
      "step": 162270
    },
    {
      "epoch": 66.43,
      "grad_norm": 1.788845419883728,
      "learning_rate": 3.7392890766780716e-06,
      "loss": 0.5161,
      "step": 162280
    },
    {
      "epoch": 66.43,
      "grad_norm": 2.046670913696289,
      "learning_rate": 3.739161629361977e-06,
      "loss": 0.5346,
      "step": 162290
    },
    {
      "epoch": 66.43,
      "grad_norm": 2.0137739181518555,
      "learning_rate": 3.739034175653719e-06,
      "loss": 0.526,
      "step": 162300
    },
    {
      "epoch": 66.44,
      "grad_norm": 2.1781511306762695,
      "learning_rate": 3.7389067155538797e-06,
      "loss": 0.5541,
      "step": 162310
    },
    {
      "epoch": 66.44,
      "grad_norm": 1.7766857147216797,
      "learning_rate": 3.7387792490630435e-06,
      "loss": 0.5109,
      "step": 162320
    },
    {
      "epoch": 66.45,
      "grad_norm": 1.8421907424926758,
      "learning_rate": 3.738651776181794e-06,
      "loss": 0.5147,
      "step": 162330
    },
    {
      "epoch": 66.45,
      "grad_norm": 2.1240949630737305,
      "learning_rate": 3.738524296910716e-06,
      "loss": 0.5233,
      "step": 162340
    },
    {
      "epoch": 66.46,
      "grad_norm": 1.6873018741607666,
      "learning_rate": 3.738396811250393e-06,
      "loss": 0.5213,
      "step": 162350
    },
    {
      "epoch": 66.46,
      "grad_norm": 1.9249346256256104,
      "learning_rate": 3.7382693192014084e-06,
      "loss": 0.5211,
      "step": 162360
    },
    {
      "epoch": 66.46,
      "grad_norm": 2.1729867458343506,
      "learning_rate": 3.738141820764347e-06,
      "loss": 0.5137,
      "step": 162370
    },
    {
      "epoch": 66.47,
      "grad_norm": 1.7384358644485474,
      "learning_rate": 3.7380143159397928e-06,
      "loss": 0.5316,
      "step": 162380
    },
    {
      "epoch": 66.47,
      "grad_norm": 2.317877769470215,
      "learning_rate": 3.73788680472833e-06,
      "loss": 0.5128,
      "step": 162390
    },
    {
      "epoch": 66.48,
      "grad_norm": 2.3525502681732178,
      "learning_rate": 3.7377592871305416e-06,
      "loss": 0.5255,
      "step": 162400
    },
    {
      "epoch": 66.48,
      "grad_norm": 2.244065523147583,
      "learning_rate": 3.7376317631470133e-06,
      "loss": 0.5054,
      "step": 162410
    },
    {
      "epoch": 66.48,
      "grad_norm": 2.0546276569366455,
      "learning_rate": 3.737504232778328e-06,
      "loss": 0.5159,
      "step": 162420
    },
    {
      "epoch": 66.49,
      "grad_norm": 1.862156629562378,
      "learning_rate": 3.7373766960250706e-06,
      "loss": 0.513,
      "step": 162430
    },
    {
      "epoch": 66.49,
      "grad_norm": 1.870280146598816,
      "learning_rate": 3.7372491528878256e-06,
      "loss": 0.5207,
      "step": 162440
    },
    {
      "epoch": 66.5,
      "grad_norm": 1.7109650373458862,
      "learning_rate": 3.737121603367176e-06,
      "loss": 0.4963,
      "step": 162450
    },
    {
      "epoch": 66.5,
      "grad_norm": 2.0048351287841797,
      "learning_rate": 3.736994047463707e-06,
      "loss": 0.4997,
      "step": 162460
    },
    {
      "epoch": 66.5,
      "grad_norm": 1.546486258506775,
      "learning_rate": 3.7368664851780027e-06,
      "loss": 0.5073,
      "step": 162470
    },
    {
      "epoch": 66.51,
      "grad_norm": 1.5506236553192139,
      "learning_rate": 3.736738916510648e-06,
      "loss": 0.5142,
      "step": 162480
    },
    {
      "epoch": 66.51,
      "grad_norm": 1.8734714984893799,
      "learning_rate": 3.736611341462227e-06,
      "loss": 0.5013,
      "step": 162490
    },
    {
      "epoch": 66.52,
      "grad_norm": 1.9365122318267822,
      "learning_rate": 3.7364837600333232e-06,
      "loss": 0.5178,
      "step": 162500
    },
    {
      "epoch": 66.52,
      "grad_norm": 1.6660337448120117,
      "learning_rate": 3.7363561722245224e-06,
      "loss": 0.5308,
      "step": 162510
    },
    {
      "epoch": 66.52,
      "grad_norm": 2.045686721801758,
      "learning_rate": 3.7362285780364076e-06,
      "loss": 0.5135,
      "step": 162520
    },
    {
      "epoch": 66.53,
      "grad_norm": 1.7522259950637817,
      "learning_rate": 3.7361009774695647e-06,
      "loss": 0.5064,
      "step": 162530
    },
    {
      "epoch": 66.53,
      "grad_norm": 1.7806099653244019,
      "learning_rate": 3.735973370524577e-06,
      "loss": 0.5227,
      "step": 162540
    },
    {
      "epoch": 66.54,
      "grad_norm": 1.7797702550888062,
      "learning_rate": 3.7358457572020302e-06,
      "loss": 0.5219,
      "step": 162550
    },
    {
      "epoch": 66.54,
      "grad_norm": 2.526339530944824,
      "learning_rate": 3.7357181375025088e-06,
      "loss": 0.519,
      "step": 162560
    },
    {
      "epoch": 66.55,
      "grad_norm": 1.8043558597564697,
      "learning_rate": 3.7355905114265966e-06,
      "loss": 0.5118,
      "step": 162570
    },
    {
      "epoch": 66.55,
      "grad_norm": 1.5250487327575684,
      "learning_rate": 3.735462878974878e-06,
      "loss": 0.5016,
      "step": 162580
    },
    {
      "epoch": 66.55,
      "grad_norm": 1.9646965265274048,
      "learning_rate": 3.7353352401479385e-06,
      "loss": 0.5457,
      "step": 162590
    },
    {
      "epoch": 66.56,
      "grad_norm": 2.264699697494507,
      "learning_rate": 3.735207594946363e-06,
      "loss": 0.5113,
      "step": 162600
    },
    {
      "epoch": 66.56,
      "grad_norm": 1.903663992881775,
      "learning_rate": 3.735079943370735e-06,
      "loss": 0.5123,
      "step": 162610
    },
    {
      "epoch": 66.57,
      "grad_norm": 2.8463189601898193,
      "learning_rate": 3.7349522854216406e-06,
      "loss": 0.527,
      "step": 162620
    },
    {
      "epoch": 66.57,
      "grad_norm": 1.6665655374526978,
      "learning_rate": 3.7348246210996637e-06,
      "loss": 0.5371,
      "step": 162630
    },
    {
      "epoch": 66.57,
      "grad_norm": 1.7201529741287231,
      "learning_rate": 3.7346969504053897e-06,
      "loss": 0.5099,
      "step": 162640
    },
    {
      "epoch": 66.58,
      "grad_norm": 2.2946882247924805,
      "learning_rate": 3.7345692733394027e-06,
      "loss": 0.5056,
      "step": 162650
    },
    {
      "epoch": 66.58,
      "grad_norm": 2.375084638595581,
      "learning_rate": 3.734441589902288e-06,
      "loss": 0.5315,
      "step": 162660
    },
    {
      "epoch": 66.59,
      "grad_norm": 1.7700111865997314,
      "learning_rate": 3.734313900094631e-06,
      "loss": 0.5278,
      "step": 162670
    },
    {
      "epoch": 66.59,
      "grad_norm": 1.8135744333267212,
      "learning_rate": 3.734186203917016e-06,
      "loss": 0.5165,
      "step": 162680
    },
    {
      "epoch": 66.59,
      "grad_norm": 2.585543632507324,
      "learning_rate": 3.7340585013700277e-06,
      "loss": 0.5049,
      "step": 162690
    },
    {
      "epoch": 66.6,
      "grad_norm": 1.953345775604248,
      "learning_rate": 3.7339307924542517e-06,
      "loss": 0.4883,
      "step": 162700
    },
    {
      "epoch": 66.6,
      "grad_norm": 1.8662282228469849,
      "learning_rate": 3.7338030771702733e-06,
      "loss": 0.4942,
      "step": 162710
    },
    {
      "epoch": 66.61,
      "grad_norm": 2.54609751701355,
      "learning_rate": 3.733675355518676e-06,
      "loss": 0.5235,
      "step": 162720
    },
    {
      "epoch": 66.61,
      "grad_norm": 2.2189109325408936,
      "learning_rate": 3.733547627500046e-06,
      "loss": 0.5059,
      "step": 162730
    },
    {
      "epoch": 66.61,
      "grad_norm": 1.93467378616333,
      "learning_rate": 3.733419893114969e-06,
      "loss": 0.5279,
      "step": 162740
    },
    {
      "epoch": 66.62,
      "grad_norm": 2.0014214515686035,
      "learning_rate": 3.7332921523640287e-06,
      "loss": 0.52,
      "step": 162750
    },
    {
      "epoch": 66.62,
      "grad_norm": 1.7741968631744385,
      "learning_rate": 3.733164405247812e-06,
      "loss": 0.5133,
      "step": 162760
    },
    {
      "epoch": 66.63,
      "grad_norm": 1.7779476642608643,
      "learning_rate": 3.7330366517669025e-06,
      "loss": 0.507,
      "step": 162770
    },
    {
      "epoch": 66.63,
      "grad_norm": 2.0998904705047607,
      "learning_rate": 3.7329088919218857e-06,
      "loss": 0.5188,
      "step": 162780
    },
    {
      "epoch": 66.64,
      "grad_norm": 1.992871880531311,
      "learning_rate": 3.7327811257133475e-06,
      "loss": 0.5235,
      "step": 162790
    },
    {
      "epoch": 66.64,
      "grad_norm": 2.534804105758667,
      "learning_rate": 3.7326533531418733e-06,
      "loss": 0.5084,
      "step": 162800
    },
    {
      "epoch": 66.64,
      "grad_norm": 1.9126691818237305,
      "learning_rate": 3.7325255742080475e-06,
      "loss": 0.523,
      "step": 162810
    },
    {
      "epoch": 66.65,
      "grad_norm": 2.011096477508545,
      "learning_rate": 3.732397788912456e-06,
      "loss": 0.5161,
      "step": 162820
    },
    {
      "epoch": 66.65,
      "grad_norm": 2.2161080837249756,
      "learning_rate": 3.7322699972556834e-06,
      "loss": 0.524,
      "step": 162830
    },
    {
      "epoch": 66.66,
      "grad_norm": 1.7375171184539795,
      "learning_rate": 3.732142199238316e-06,
      "loss": 0.5004,
      "step": 162840
    },
    {
      "epoch": 66.66,
      "grad_norm": 1.6423108577728271,
      "learning_rate": 3.7320143948609395e-06,
      "loss": 0.5143,
      "step": 162850
    },
    {
      "epoch": 66.66,
      "grad_norm": 1.9826428890228271,
      "learning_rate": 3.7318865841241384e-06,
      "loss": 0.4874,
      "step": 162860
    },
    {
      "epoch": 66.67,
      "grad_norm": 2.1086838245391846,
      "learning_rate": 3.7317587670284985e-06,
      "loss": 0.5223,
      "step": 162870
    },
    {
      "epoch": 66.67,
      "grad_norm": 2.0041253566741943,
      "learning_rate": 3.7316309435746057e-06,
      "loss": 0.5136,
      "step": 162880
    },
    {
      "epoch": 66.68,
      "grad_norm": 2.515733242034912,
      "learning_rate": 3.7315031137630453e-06,
      "loss": 0.5206,
      "step": 162890
    },
    {
      "epoch": 66.68,
      "grad_norm": 1.6375852823257446,
      "learning_rate": 3.7313752775944024e-06,
      "loss": 0.5578,
      "step": 162900
    },
    {
      "epoch": 66.68,
      "grad_norm": 2.0377657413482666,
      "learning_rate": 3.7312474350692633e-06,
      "loss": 0.5304,
      "step": 162910
    },
    {
      "epoch": 66.69,
      "grad_norm": 2.044630527496338,
      "learning_rate": 3.7311195861882127e-06,
      "loss": 0.5176,
      "step": 162920
    },
    {
      "epoch": 66.69,
      "grad_norm": 1.938283920288086,
      "learning_rate": 3.7309917309518374e-06,
      "loss": 0.4963,
      "step": 162930
    },
    {
      "epoch": 66.7,
      "grad_norm": 2.5545451641082764,
      "learning_rate": 3.7308638693607224e-06,
      "loss": 0.5333,
      "step": 162940
    },
    {
      "epoch": 66.7,
      "grad_norm": 2.104527473449707,
      "learning_rate": 3.730736001415454e-06,
      "loss": 0.5146,
      "step": 162950
    },
    {
      "epoch": 66.7,
      "grad_norm": 2.151521921157837,
      "learning_rate": 3.7306081271166177e-06,
      "loss": 0.5262,
      "step": 162960
    },
    {
      "epoch": 66.71,
      "grad_norm": 1.9623637199401855,
      "learning_rate": 3.730480246464799e-06,
      "loss": 0.5273,
      "step": 162970
    },
    {
      "epoch": 66.71,
      "grad_norm": 2.908804416656494,
      "learning_rate": 3.730352359460583e-06,
      "loss": 0.5444,
      "step": 162980
    },
    {
      "epoch": 66.72,
      "grad_norm": 2.1571521759033203,
      "learning_rate": 3.730224466104557e-06,
      "loss": 0.5329,
      "step": 162990
    },
    {
      "epoch": 66.72,
      "grad_norm": 2.1889007091522217,
      "learning_rate": 3.7300965663973064e-06,
      "loss": 0.5114,
      "step": 163000
    },
    {
      "epoch": 66.73,
      "grad_norm": 4.1281514167785645,
      "learning_rate": 3.729968660339417e-06,
      "loss": 0.515,
      "step": 163010
    },
    {
      "epoch": 66.73,
      "grad_norm": 1.7722864151000977,
      "learning_rate": 3.7298407479314728e-06,
      "loss": 0.5236,
      "step": 163020
    },
    {
      "epoch": 66.73,
      "grad_norm": 1.8191660642623901,
      "learning_rate": 3.7297128291740635e-06,
      "loss": 0.5139,
      "step": 163030
    },
    {
      "epoch": 66.74,
      "grad_norm": 1.6777315139770508,
      "learning_rate": 3.729584904067772e-06,
      "loss": 0.5325,
      "step": 163040
    },
    {
      "epoch": 66.74,
      "grad_norm": 2.290637969970703,
      "learning_rate": 3.729456972613186e-06,
      "loss": 0.514,
      "step": 163050
    },
    {
      "epoch": 66.75,
      "grad_norm": 2.029474973678589,
      "learning_rate": 3.729329034810891e-06,
      "loss": 0.5246,
      "step": 163060
    },
    {
      "epoch": 66.75,
      "grad_norm": 1.823245644569397,
      "learning_rate": 3.729201090661473e-06,
      "loss": 0.5219,
      "step": 163070
    },
    {
      "epoch": 66.75,
      "grad_norm": 1.8871407508850098,
      "learning_rate": 3.729073140165518e-06,
      "loss": 0.5307,
      "step": 163080
    },
    {
      "epoch": 66.76,
      "grad_norm": 1.841001272201538,
      "learning_rate": 3.728945183323612e-06,
      "loss": 0.5081,
      "step": 163090
    },
    {
      "epoch": 66.76,
      "grad_norm": 2.4676029682159424,
      "learning_rate": 3.728817220136342e-06,
      "loss": 0.5108,
      "step": 163100
    },
    {
      "epoch": 66.77,
      "grad_norm": 3.027071237564087,
      "learning_rate": 3.728689250604293e-06,
      "loss": 0.5031,
      "step": 163110
    },
    {
      "epoch": 66.77,
      "grad_norm": 2.582350015640259,
      "learning_rate": 3.728561274728052e-06,
      "loss": 0.5173,
      "step": 163120
    },
    {
      "epoch": 66.77,
      "grad_norm": 1.8493521213531494,
      "learning_rate": 3.728433292508205e-06,
      "loss": 0.5116,
      "step": 163130
    },
    {
      "epoch": 66.78,
      "grad_norm": 2.3444409370422363,
      "learning_rate": 3.728305303945339e-06,
      "loss": 0.5053,
      "step": 163140
    },
    {
      "epoch": 66.78,
      "grad_norm": 2.4672281742095947,
      "learning_rate": 3.7281773090400383e-06,
      "loss": 0.5001,
      "step": 163150
    },
    {
      "epoch": 66.79,
      "grad_norm": 1.8334919214248657,
      "learning_rate": 3.728049307792891e-06,
      "loss": 0.5217,
      "step": 163160
    },
    {
      "epoch": 66.79,
      "grad_norm": 2.3989417552948,
      "learning_rate": 3.7279213002044836e-06,
      "loss": 0.5088,
      "step": 163170
    },
    {
      "epoch": 66.79,
      "grad_norm": 1.5519084930419922,
      "learning_rate": 3.7277932862754014e-06,
      "loss": 0.5192,
      "step": 163180
    },
    {
      "epoch": 66.8,
      "grad_norm": 1.739135503768921,
      "learning_rate": 3.7276652660062303e-06,
      "loss": 0.5207,
      "step": 163190
    },
    {
      "epoch": 66.8,
      "grad_norm": 1.844954490661621,
      "learning_rate": 3.7275372393975587e-06,
      "loss": 0.5048,
      "step": 163200
    },
    {
      "epoch": 66.81,
      "grad_norm": 2.0457632541656494,
      "learning_rate": 3.727409206449972e-06,
      "loss": 0.5258,
      "step": 163210
    },
    {
      "epoch": 66.81,
      "grad_norm": 1.8762537240982056,
      "learning_rate": 3.7272811671640566e-06,
      "loss": 0.5176,
      "step": 163220
    },
    {
      "epoch": 66.82,
      "grad_norm": 2.145038604736328,
      "learning_rate": 3.7271531215403996e-06,
      "loss": 0.5225,
      "step": 163230
    },
    {
      "epoch": 66.82,
      "grad_norm": 1.6153318881988525,
      "learning_rate": 3.7270250695795864e-06,
      "loss": 0.5118,
      "step": 163240
    },
    {
      "epoch": 66.82,
      "grad_norm": 1.6340720653533936,
      "learning_rate": 3.7268970112822044e-06,
      "loss": 0.5258,
      "step": 163250
    },
    {
      "epoch": 66.83,
      "grad_norm": 1.9981385469436646,
      "learning_rate": 3.726768946648841e-06,
      "loss": 0.5036,
      "step": 163260
    },
    {
      "epoch": 66.83,
      "grad_norm": 2.0772602558135986,
      "learning_rate": 3.7266408756800812e-06,
      "loss": 0.5272,
      "step": 163270
    },
    {
      "epoch": 66.84,
      "grad_norm": 1.704903244972229,
      "learning_rate": 3.7265127983765128e-06,
      "loss": 0.525,
      "step": 163280
    },
    {
      "epoch": 66.84,
      "grad_norm": 1.542123556137085,
      "learning_rate": 3.726384714738722e-06,
      "loss": 0.5314,
      "step": 163290
    },
    {
      "epoch": 66.84,
      "grad_norm": 1.8027572631835938,
      "learning_rate": 3.7262566247672954e-06,
      "loss": 0.5405,
      "step": 163300
    },
    {
      "epoch": 66.85,
      "grad_norm": 2.5650382041931152,
      "learning_rate": 3.726128528462821e-06,
      "loss": 0.5247,
      "step": 163310
    },
    {
      "epoch": 66.85,
      "grad_norm": 1.8782795667648315,
      "learning_rate": 3.726000425825884e-06,
      "loss": 0.5096,
      "step": 163320
    },
    {
      "epoch": 66.86,
      "grad_norm": 6.355854034423828,
      "learning_rate": 3.7258723168570717e-06,
      "loss": 0.5247,
      "step": 163330
    },
    {
      "epoch": 66.86,
      "grad_norm": 1.9037855863571167,
      "learning_rate": 3.7257442015569715e-06,
      "loss": 0.5329,
      "step": 163340
    },
    {
      "epoch": 66.86,
      "grad_norm": 1.9658572673797607,
      "learning_rate": 3.72561607992617e-06,
      "loss": 0.5218,
      "step": 163350
    },
    {
      "epoch": 66.87,
      "grad_norm": 1.896005392074585,
      "learning_rate": 3.725487951965254e-06,
      "loss": 0.5109,
      "step": 163360
    },
    {
      "epoch": 66.87,
      "grad_norm": 2.016667366027832,
      "learning_rate": 3.72535981767481e-06,
      "loss": 0.5155,
      "step": 163370
    },
    {
      "epoch": 66.88,
      "grad_norm": 2.9624969959259033,
      "learning_rate": 3.7252316770554256e-06,
      "loss": 0.5329,
      "step": 163380
    },
    {
      "epoch": 66.88,
      "grad_norm": 2.0356802940368652,
      "learning_rate": 3.7251035301076876e-06,
      "loss": 0.5352,
      "step": 163390
    },
    {
      "epoch": 66.88,
      "grad_norm": 1.697953224182129,
      "learning_rate": 3.7249753768321827e-06,
      "loss": 0.5162,
      "step": 163400
    },
    {
      "epoch": 66.89,
      "grad_norm": 1.7351183891296387,
      "learning_rate": 3.724847217229499e-06,
      "loss": 0.5078,
      "step": 163410
    },
    {
      "epoch": 66.89,
      "grad_norm": 1.6900397539138794,
      "learning_rate": 3.7247190513002222e-06,
      "loss": 0.5205,
      "step": 163420
    },
    {
      "epoch": 66.9,
      "grad_norm": 2.3682546615600586,
      "learning_rate": 3.7245908790449397e-06,
      "loss": 0.5112,
      "step": 163430
    },
    {
      "epoch": 66.9,
      "grad_norm": 1.6726558208465576,
      "learning_rate": 3.72446270046424e-06,
      "loss": 0.5057,
      "step": 163440
    },
    {
      "epoch": 66.91,
      "grad_norm": 1.9356063604354858,
      "learning_rate": 3.7243345155587085e-06,
      "loss": 0.5415,
      "step": 163450
    },
    {
      "epoch": 66.91,
      "grad_norm": 2.098719835281372,
      "learning_rate": 3.7242063243289337e-06,
      "loss": 0.5173,
      "step": 163460
    },
    {
      "epoch": 66.91,
      "grad_norm": 2.0204622745513916,
      "learning_rate": 3.7240781267755024e-06,
      "loss": 0.5215,
      "step": 163470
    },
    {
      "epoch": 66.92,
      "grad_norm": 2.0538835525512695,
      "learning_rate": 3.723949922899001e-06,
      "loss": 0.5377,
      "step": 163480
    },
    {
      "epoch": 66.92,
      "grad_norm": 1.8190734386444092,
      "learning_rate": 3.7238217127000175e-06,
      "loss": 0.4856,
      "step": 163490
    },
    {
      "epoch": 66.93,
      "grad_norm": 1.4186017513275146,
      "learning_rate": 3.7236934961791396e-06,
      "loss": 0.503,
      "step": 163500
    },
    {
      "epoch": 66.93,
      "grad_norm": 2.044713020324707,
      "learning_rate": 3.7235652733369543e-06,
      "loss": 0.5101,
      "step": 163510
    },
    {
      "epoch": 66.93,
      "grad_norm": 2.0592129230499268,
      "learning_rate": 3.7234370441740487e-06,
      "loss": 0.5369,
      "step": 163520
    },
    {
      "epoch": 66.94,
      "grad_norm": 2.4708313941955566,
      "learning_rate": 3.72330880869101e-06,
      "loss": 0.5239,
      "step": 163530
    },
    {
      "epoch": 66.94,
      "grad_norm": 2.101121425628662,
      "learning_rate": 3.7231805668884272e-06,
      "loss": 0.5221,
      "step": 163540
    },
    {
      "epoch": 66.95,
      "grad_norm": 1.7151298522949219,
      "learning_rate": 3.723052318766886e-06,
      "loss": 0.5136,
      "step": 163550
    },
    {
      "epoch": 66.95,
      "grad_norm": 1.941340446472168,
      "learning_rate": 3.7229240643269742e-06,
      "loss": 0.5235,
      "step": 163560
    },
    {
      "epoch": 66.95,
      "grad_norm": 1.747175931930542,
      "learning_rate": 3.72279580356928e-06,
      "loss": 0.518,
      "step": 163570
    },
    {
      "epoch": 66.96,
      "grad_norm": 2.4390196800231934,
      "learning_rate": 3.7226675364943898e-06,
      "loss": 0.5112,
      "step": 163580
    },
    {
      "epoch": 66.96,
      "grad_norm": 1.860976219177246,
      "learning_rate": 3.7225392631028926e-06,
      "loss": 0.5311,
      "step": 163590
    },
    {
      "epoch": 66.97,
      "grad_norm": 1.8919860124588013,
      "learning_rate": 3.7224109833953758e-06,
      "loss": 0.5217,
      "step": 163600
    },
    {
      "epoch": 66.97,
      "grad_norm": 1.7672470808029175,
      "learning_rate": 3.722282697372426e-06,
      "loss": 0.5165,
      "step": 163610
    },
    {
      "epoch": 66.98,
      "grad_norm": 2.5508220195770264,
      "learning_rate": 3.7221544050346316e-06,
      "loss": 0.5216,
      "step": 163620
    },
    {
      "epoch": 66.98,
      "grad_norm": 2.121492862701416,
      "learning_rate": 3.722026106382579e-06,
      "loss": 0.5192,
      "step": 163630
    },
    {
      "epoch": 66.98,
      "grad_norm": 2.100919008255005,
      "learning_rate": 3.7218978014168586e-06,
      "loss": 0.5214,
      "step": 163640
    },
    {
      "epoch": 66.99,
      "grad_norm": 2.2954697608947754,
      "learning_rate": 3.7217694901380555e-06,
      "loss": 0.5128,
      "step": 163650
    },
    {
      "epoch": 66.99,
      "grad_norm": 1.9307425022125244,
      "learning_rate": 3.7216411725467593e-06,
      "loss": 0.537,
      "step": 163660
    },
    {
      "epoch": 67.0,
      "grad_norm": 1.9747374057769775,
      "learning_rate": 3.721512848643557e-06,
      "loss": 0.5167,
      "step": 163670
    },
    {
      "epoch": 67.0,
      "grad_norm": 1.8690680265426636,
      "learning_rate": 3.7213845184290353e-06,
      "loss": 0.5179,
      "step": 163680
    },
    {
      "epoch": 67.0,
      "eval_loss": 0.5217322111129761,
      "eval_runtime": 52.3453,
      "eval_samples_per_second": 65.889,
      "eval_steps_per_second": 8.253,
      "step": 163681
    },
    {
      "epoch": 67.0,
      "grad_norm": 2.308143138885498,
      "learning_rate": 3.721256181903784e-06,
      "loss": 0.496,
      "step": 163690
    },
    {
      "epoch": 67.01,
      "grad_norm": 1.7654649019241333,
      "learning_rate": 3.7211278390683904e-06,
      "loss": 0.5187,
      "step": 163700
    },
    {
      "epoch": 67.01,
      "grad_norm": 1.7208713293075562,
      "learning_rate": 3.7209994899234423e-06,
      "loss": 0.5163,
      "step": 163710
    },
    {
      "epoch": 67.02,
      "grad_norm": 2.035889148712158,
      "learning_rate": 3.7208711344695272e-06,
      "loss": 0.5105,
      "step": 163720
    },
    {
      "epoch": 67.02,
      "grad_norm": 1.4723163843154907,
      "learning_rate": 3.7207427727072335e-06,
      "loss": 0.5248,
      "step": 163730
    },
    {
      "epoch": 67.02,
      "grad_norm": 1.8788270950317383,
      "learning_rate": 3.7206144046371494e-06,
      "loss": 0.5118,
      "step": 163740
    },
    {
      "epoch": 67.03,
      "grad_norm": 1.4541510343551636,
      "learning_rate": 3.7204860302598627e-06,
      "loss": 0.5415,
      "step": 163750
    },
    {
      "epoch": 67.03,
      "grad_norm": 1.9189454317092896,
      "learning_rate": 3.7203576495759612e-06,
      "loss": 0.5167,
      "step": 163760
    },
    {
      "epoch": 67.04,
      "grad_norm": 2.5767550468444824,
      "learning_rate": 3.7202292625860333e-06,
      "loss": 0.5172,
      "step": 163770
    },
    {
      "epoch": 67.04,
      "grad_norm": 1.647993564605713,
      "learning_rate": 3.7201008692906672e-06,
      "loss": 0.5218,
      "step": 163780
    },
    {
      "epoch": 67.04,
      "grad_norm": 1.7176419496536255,
      "learning_rate": 3.719972469690451e-06,
      "loss": 0.5192,
      "step": 163790
    },
    {
      "epoch": 67.05,
      "grad_norm": 1.7052279710769653,
      "learning_rate": 3.7198440637859723e-06,
      "loss": 0.5199,
      "step": 163800
    },
    {
      "epoch": 67.05,
      "grad_norm": 1.7070562839508057,
      "learning_rate": 3.71971565157782e-06,
      "loss": 0.5231,
      "step": 163810
    },
    {
      "epoch": 67.06,
      "grad_norm": 1.678242564201355,
      "learning_rate": 3.719587233066583e-06,
      "loss": 0.5163,
      "step": 163820
    },
    {
      "epoch": 67.06,
      "grad_norm": 2.1876347064971924,
      "learning_rate": 3.719458808252848e-06,
      "loss": 0.5166,
      "step": 163830
    },
    {
      "epoch": 67.07,
      "grad_norm": 1.9517803192138672,
      "learning_rate": 3.7193303771372038e-06,
      "loss": 0.5368,
      "step": 163840
    },
    {
      "epoch": 67.07,
      "grad_norm": 1.4278165102005005,
      "learning_rate": 3.719201939720239e-06,
      "loss": 0.5269,
      "step": 163850
    },
    {
      "epoch": 67.07,
      "grad_norm": 2.741824150085449,
      "learning_rate": 3.719073496002542e-06,
      "loss": 0.4896,
      "step": 163860
    },
    {
      "epoch": 67.08,
      "grad_norm": 1.570056676864624,
      "learning_rate": 3.718945045984701e-06,
      "loss": 0.5136,
      "step": 163870
    },
    {
      "epoch": 67.08,
      "grad_norm": 2.3027803897857666,
      "learning_rate": 3.7188165896673042e-06,
      "loss": 0.5262,
      "step": 163880
    },
    {
      "epoch": 67.09,
      "grad_norm": 1.9769786596298218,
      "learning_rate": 3.7186881270509413e-06,
      "loss": 0.5292,
      "step": 163890
    },
    {
      "epoch": 67.09,
      "grad_norm": 2.274935245513916,
      "learning_rate": 3.7185596581361993e-06,
      "loss": 0.519,
      "step": 163900
    },
    {
      "epoch": 67.09,
      "grad_norm": 1.9686565399169922,
      "learning_rate": 3.7184311829236666e-06,
      "loss": 0.5321,
      "step": 163910
    },
    {
      "epoch": 67.1,
      "grad_norm": 1.9159752130508423,
      "learning_rate": 3.7183027014139327e-06,
      "loss": 0.5067,
      "step": 163920
    },
    {
      "epoch": 67.1,
      "grad_norm": 1.5372092723846436,
      "learning_rate": 3.7181742136075862e-06,
      "loss": 0.519,
      "step": 163930
    },
    {
      "epoch": 67.11,
      "grad_norm": 1.5988961458206177,
      "learning_rate": 3.7180457195052147e-06,
      "loss": 0.5245,
      "step": 163940
    },
    {
      "epoch": 67.11,
      "grad_norm": 2.2737951278686523,
      "learning_rate": 3.7179172191074075e-06,
      "loss": 0.5049,
      "step": 163950
    },
    {
      "epoch": 67.11,
      "grad_norm": 1.8738001585006714,
      "learning_rate": 3.717788712414753e-06,
      "loss": 0.4905,
      "step": 163960
    },
    {
      "epoch": 67.12,
      "grad_norm": 2.2611889839172363,
      "learning_rate": 3.71766019942784e-06,
      "loss": 0.5141,
      "step": 163970
    },
    {
      "epoch": 67.12,
      "grad_norm": 1.7071605920791626,
      "learning_rate": 3.717531680147257e-06,
      "loss": 0.519,
      "step": 163980
    },
    {
      "epoch": 67.13,
      "grad_norm": 1.6504459381103516,
      "learning_rate": 3.717403154573593e-06,
      "loss": 0.5218,
      "step": 163990
    },
    {
      "epoch": 67.13,
      "grad_norm": 1.82159423828125,
      "learning_rate": 3.7172746227074364e-06,
      "loss": 0.5172,
      "step": 164000
    },
    {
      "epoch": 67.13,
      "grad_norm": 1.6522397994995117,
      "learning_rate": 3.7171460845493774e-06,
      "loss": 0.5113,
      "step": 164010
    },
    {
      "epoch": 67.14,
      "grad_norm": 2.032892942428589,
      "learning_rate": 3.7170175401000022e-06,
      "loss": 0.5376,
      "step": 164020
    },
    {
      "epoch": 67.14,
      "grad_norm": 1.6510088443756104,
      "learning_rate": 3.7168889893599015e-06,
      "loss": 0.5161,
      "step": 164030
    },
    {
      "epoch": 67.15,
      "grad_norm": 2.5010886192321777,
      "learning_rate": 3.716760432329664e-06,
      "loss": 0.5344,
      "step": 164040
    },
    {
      "epoch": 67.15,
      "grad_norm": 1.7430275678634644,
      "learning_rate": 3.716631869009878e-06,
      "loss": 0.5316,
      "step": 164050
    },
    {
      "epoch": 67.16,
      "grad_norm": 1.5809043645858765,
      "learning_rate": 3.7165032994011325e-06,
      "loss": 0.5248,
      "step": 164060
    },
    {
      "epoch": 67.16,
      "grad_norm": 1.613500952720642,
      "learning_rate": 3.716374723504018e-06,
      "loss": 0.5347,
      "step": 164070
    },
    {
      "epoch": 67.16,
      "grad_norm": 1.9429718255996704,
      "learning_rate": 3.7162461413191214e-06,
      "loss": 0.5152,
      "step": 164080
    },
    {
      "epoch": 67.17,
      "grad_norm": 1.7385178804397583,
      "learning_rate": 3.7161175528470325e-06,
      "loss": 0.504,
      "step": 164090
    },
    {
      "epoch": 67.17,
      "grad_norm": 2.1536612510681152,
      "learning_rate": 3.7159889580883397e-06,
      "loss": 0.5307,
      "step": 164100
    },
    {
      "epoch": 67.18,
      "grad_norm": 1.565955400466919,
      "learning_rate": 3.715860357043634e-06,
      "loss": 0.5031,
      "step": 164110
    },
    {
      "epoch": 67.18,
      "grad_norm": 1.9447258710861206,
      "learning_rate": 3.7157317497135026e-06,
      "loss": 0.5255,
      "step": 164120
    },
    {
      "epoch": 67.18,
      "grad_norm": 2.209205389022827,
      "learning_rate": 3.7156031360985358e-06,
      "loss": 0.4983,
      "step": 164130
    },
    {
      "epoch": 67.19,
      "grad_norm": 2.378222703933716,
      "learning_rate": 3.715474516199322e-06,
      "loss": 0.5124,
      "step": 164140
    },
    {
      "epoch": 67.19,
      "grad_norm": 1.623551368713379,
      "learning_rate": 3.7153458900164503e-06,
      "loss": 0.4982,
      "step": 164150
    },
    {
      "epoch": 67.2,
      "grad_norm": 2.6195106506347656,
      "learning_rate": 3.71521725755051e-06,
      "loss": 0.5207,
      "step": 164160
    },
    {
      "epoch": 67.2,
      "grad_norm": 2.1026408672332764,
      "learning_rate": 3.715088618802091e-06,
      "loss": 0.5275,
      "step": 164170
    },
    {
      "epoch": 67.2,
      "grad_norm": 1.5657973289489746,
      "learning_rate": 3.714959973771782e-06,
      "loss": 0.5212,
      "step": 164180
    },
    {
      "epoch": 67.21,
      "grad_norm": 1.830666184425354,
      "learning_rate": 3.714831322460173e-06,
      "loss": 0.5391,
      "step": 164190
    },
    {
      "epoch": 67.21,
      "grad_norm": 2.484311580657959,
      "learning_rate": 3.7147026648678527e-06,
      "loss": 0.5177,
      "step": 164200
    },
    {
      "epoch": 67.22,
      "grad_norm": 1.7866426706314087,
      "learning_rate": 3.71457400099541e-06,
      "loss": 0.4973,
      "step": 164210
    },
    {
      "epoch": 67.22,
      "grad_norm": 2.183983564376831,
      "learning_rate": 3.714445330843435e-06,
      "loss": 0.5095,
      "step": 164220
    },
    {
      "epoch": 67.22,
      "grad_norm": 1.922357201576233,
      "learning_rate": 3.7143166544125176e-06,
      "loss": 0.5298,
      "step": 164230
    },
    {
      "epoch": 67.23,
      "grad_norm": 1.5752779245376587,
      "learning_rate": 3.7141879717032458e-06,
      "loss": 0.5187,
      "step": 164240
    },
    {
      "epoch": 67.23,
      "grad_norm": 1.688913106918335,
      "learning_rate": 3.7140592827162104e-06,
      "loss": 0.5188,
      "step": 164250
    },
    {
      "epoch": 67.24,
      "grad_norm": 2.220311164855957,
      "learning_rate": 3.7139305874520003e-06,
      "loss": 0.5051,
      "step": 164260
    },
    {
      "epoch": 67.24,
      "grad_norm": 2.137350082397461,
      "learning_rate": 3.713801885911205e-06,
      "loss": 0.5088,
      "step": 164270
    },
    {
      "epoch": 67.25,
      "grad_norm": 1.8837082386016846,
      "learning_rate": 3.7136731780944137e-06,
      "loss": 0.5224,
      "step": 164280
    },
    {
      "epoch": 67.25,
      "grad_norm": 1.8714067935943604,
      "learning_rate": 3.713544464002218e-06,
      "loss": 0.5404,
      "step": 164290
    },
    {
      "epoch": 67.25,
      "grad_norm": 2.6914689540863037,
      "learning_rate": 3.7134157436352048e-06,
      "loss": 0.5166,
      "step": 164300
    },
    {
      "epoch": 67.26,
      "grad_norm": 1.7557387351989746,
      "learning_rate": 3.7132870169939656e-06,
      "loss": 0.512,
      "step": 164310
    },
    {
      "epoch": 67.26,
      "grad_norm": 1.5549237728118896,
      "learning_rate": 3.7131582840790885e-06,
      "loss": 0.5216,
      "step": 164320
    },
    {
      "epoch": 67.27,
      "grad_norm": 1.5777097940444946,
      "learning_rate": 3.713029544891165e-06,
      "loss": 0.5313,
      "step": 164330
    },
    {
      "epoch": 67.27,
      "grad_norm": 1.985503077507019,
      "learning_rate": 3.7129007994307833e-06,
      "loss": 0.5127,
      "step": 164340
    },
    {
      "epoch": 67.27,
      "grad_norm": 2.1538138389587402,
      "learning_rate": 3.712772047698534e-06,
      "loss": 0.5132,
      "step": 164350
    },
    {
      "epoch": 67.28,
      "grad_norm": 2.0169100761413574,
      "learning_rate": 3.712643289695007e-06,
      "loss": 0.534,
      "step": 164360
    },
    {
      "epoch": 67.28,
      "grad_norm": 1.4662748575210571,
      "learning_rate": 3.7125145254207917e-06,
      "loss": 0.5338,
      "step": 164370
    },
    {
      "epoch": 67.29,
      "grad_norm": 1.7886643409729004,
      "learning_rate": 3.712385754876478e-06,
      "loss": 0.5282,
      "step": 164380
    },
    {
      "epoch": 67.29,
      "grad_norm": 1.9072705507278442,
      "learning_rate": 3.7122569780626563e-06,
      "loss": 0.5065,
      "step": 164390
    },
    {
      "epoch": 67.29,
      "grad_norm": 1.7571221590042114,
      "learning_rate": 3.7121281949799155e-06,
      "loss": 0.4906,
      "step": 164400
    },
    {
      "epoch": 67.3,
      "grad_norm": 1.8716779947280884,
      "learning_rate": 3.7119994056288462e-06,
      "loss": 0.5229,
      "step": 164410
    },
    {
      "epoch": 67.3,
      "grad_norm": 2.0441734790802,
      "learning_rate": 3.711870610010039e-06,
      "loss": 0.5178,
      "step": 164420
    },
    {
      "epoch": 67.31,
      "grad_norm": 1.9614347219467163,
      "learning_rate": 3.711741808124082e-06,
      "loss": 0.5221,
      "step": 164430
    },
    {
      "epoch": 67.31,
      "grad_norm": 1.8875072002410889,
      "learning_rate": 3.711612999971567e-06,
      "loss": 0.5171,
      "step": 164440
    },
    {
      "epoch": 67.31,
      "grad_norm": 2.0950334072113037,
      "learning_rate": 3.711484185553083e-06,
      "loss": 0.5268,
      "step": 164450
    },
    {
      "epoch": 67.32,
      "grad_norm": 1.7901960611343384,
      "learning_rate": 3.711355364869221e-06,
      "loss": 0.5383,
      "step": 164460
    },
    {
      "epoch": 67.32,
      "grad_norm": 2.1895506381988525,
      "learning_rate": 3.711226537920571e-06,
      "loss": 0.5344,
      "step": 164470
    },
    {
      "epoch": 67.33,
      "grad_norm": 2.6144859790802,
      "learning_rate": 3.711097704707722e-06,
      "loss": 0.5338,
      "step": 164480
    },
    {
      "epoch": 67.33,
      "grad_norm": 1.654393196105957,
      "learning_rate": 3.7109688652312656e-06,
      "loss": 0.5246,
      "step": 164490
    },
    {
      "epoch": 67.34,
      "grad_norm": 2.219557523727417,
      "learning_rate": 3.710840019491791e-06,
      "loss": 0.5042,
      "step": 164500
    },
    {
      "epoch": 67.34,
      "grad_norm": 2.063389301300049,
      "learning_rate": 3.710711167489888e-06,
      "loss": 0.5111,
      "step": 164510
    },
    {
      "epoch": 67.34,
      "grad_norm": 1.629746913909912,
      "learning_rate": 3.7105823092261487e-06,
      "loss": 0.5173,
      "step": 164520
    },
    {
      "epoch": 67.35,
      "grad_norm": 1.5234248638153076,
      "learning_rate": 3.710453444701161e-06,
      "loss": 0.5296,
      "step": 164530
    },
    {
      "epoch": 67.35,
      "grad_norm": 1.9888503551483154,
      "learning_rate": 3.7103245739155176e-06,
      "loss": 0.5119,
      "step": 164540
    },
    {
      "epoch": 67.36,
      "grad_norm": 2.056359052658081,
      "learning_rate": 3.7101956968698077e-06,
      "loss": 0.5024,
      "step": 164550
    },
    {
      "epoch": 67.36,
      "grad_norm": 2.659677505493164,
      "learning_rate": 3.7100668135646216e-06,
      "loss": 0.5029,
      "step": 164560
    },
    {
      "epoch": 67.36,
      "grad_norm": 1.8194419145584106,
      "learning_rate": 3.7099379240005495e-06,
      "loss": 0.5177,
      "step": 164570
    },
    {
      "epoch": 67.37,
      "grad_norm": 1.9391165971755981,
      "learning_rate": 3.7098090281781825e-06,
      "loss": 0.4981,
      "step": 164580
    },
    {
      "epoch": 67.37,
      "grad_norm": 1.6548259258270264,
      "learning_rate": 3.70968012609811e-06,
      "loss": 0.5049,
      "step": 164590
    },
    {
      "epoch": 67.38,
      "grad_norm": 1.4266412258148193,
      "learning_rate": 3.709551217760924e-06,
      "loss": 0.4924,
      "step": 164600
    },
    {
      "epoch": 67.38,
      "grad_norm": 1.8899401426315308,
      "learning_rate": 3.7094223031672137e-06,
      "loss": 0.528,
      "step": 164610
    },
    {
      "epoch": 67.38,
      "grad_norm": 1.6154158115386963,
      "learning_rate": 3.7092933823175693e-06,
      "loss": 0.5386,
      "step": 164620
    },
    {
      "epoch": 67.39,
      "grad_norm": 2.619267225265503,
      "learning_rate": 3.709164455212583e-06,
      "loss": 0.5338,
      "step": 164630
    },
    {
      "epoch": 67.39,
      "grad_norm": 1.7865644693374634,
      "learning_rate": 3.7090355218528444e-06,
      "loss": 0.5076,
      "step": 164640
    },
    {
      "epoch": 67.4,
      "grad_norm": 1.762885570526123,
      "learning_rate": 3.708906582238944e-06,
      "loss": 0.5113,
      "step": 164650
    },
    {
      "epoch": 67.4,
      "grad_norm": 1.6287609338760376,
      "learning_rate": 3.7087776363714732e-06,
      "loss": 0.5438,
      "step": 164660
    },
    {
      "epoch": 67.4,
      "grad_norm": 1.722760558128357,
      "learning_rate": 3.7086486842510224e-06,
      "loss": 0.5255,
      "step": 164670
    },
    {
      "epoch": 67.41,
      "grad_norm": 1.5333980321884155,
      "learning_rate": 3.7085197258781818e-06,
      "loss": 0.5403,
      "step": 164680
    },
    {
      "epoch": 67.41,
      "grad_norm": 1.5854469537734985,
      "learning_rate": 3.7083907612535417e-06,
      "loss": 0.5192,
      "step": 164690
    },
    {
      "epoch": 67.42,
      "grad_norm": 1.7034205198287964,
      "learning_rate": 3.7082617903776948e-06,
      "loss": 0.5051,
      "step": 164700
    },
    {
      "epoch": 67.42,
      "grad_norm": 1.7160115242004395,
      "learning_rate": 3.7081328132512296e-06,
      "loss": 0.5088,
      "step": 164710
    },
    {
      "epoch": 67.43,
      "grad_norm": 1.5112508535385132,
      "learning_rate": 3.7080038298747387e-06,
      "loss": 0.5139,
      "step": 164720
    },
    {
      "epoch": 67.43,
      "grad_norm": 2.5060644149780273,
      "learning_rate": 3.7078748402488124e-06,
      "loss": 0.5186,
      "step": 164730
    },
    {
      "epoch": 67.43,
      "grad_norm": 1.4779397249221802,
      "learning_rate": 3.707745844374041e-06,
      "loss": 0.5326,
      "step": 164740
    },
    {
      "epoch": 67.44,
      "grad_norm": 1.7831745147705078,
      "learning_rate": 3.7076168422510167e-06,
      "loss": 0.5201,
      "step": 164750
    },
    {
      "epoch": 67.44,
      "grad_norm": 1.9808001518249512,
      "learning_rate": 3.7074878338803284e-06,
      "loss": 0.5246,
      "step": 164760
    },
    {
      "epoch": 67.45,
      "grad_norm": 2.280791759490967,
      "learning_rate": 3.7073588192625693e-06,
      "loss": 0.5071,
      "step": 164770
    },
    {
      "epoch": 67.45,
      "grad_norm": 2.2813851833343506,
      "learning_rate": 3.7072297983983293e-06,
      "loss": 0.4887,
      "step": 164780
    },
    {
      "epoch": 67.45,
      "grad_norm": 1.748660683631897,
      "learning_rate": 3.7071007712881987e-06,
      "loss": 0.5345,
      "step": 164790
    },
    {
      "epoch": 67.46,
      "grad_norm": 1.7751667499542236,
      "learning_rate": 3.70697173793277e-06,
      "loss": 0.5205,
      "step": 164800
    },
    {
      "epoch": 67.46,
      "grad_norm": 1.959740400314331,
      "learning_rate": 3.7068426983326336e-06,
      "loss": 0.518,
      "step": 164810
    },
    {
      "epoch": 67.47,
      "grad_norm": 1.6311426162719727,
      "learning_rate": 3.7067136524883806e-06,
      "loss": 0.5162,
      "step": 164820
    },
    {
      "epoch": 67.47,
      "grad_norm": 2.646336793899536,
      "learning_rate": 3.706584600400602e-06,
      "loss": 0.5193,
      "step": 164830
    },
    {
      "epoch": 67.47,
      "grad_norm": 1.7220600843429565,
      "learning_rate": 3.7064555420698894e-06,
      "loss": 0.5422,
      "step": 164840
    },
    {
      "epoch": 67.48,
      "grad_norm": 2.3958685398101807,
      "learning_rate": 3.7063264774968337e-06,
      "loss": 0.5375,
      "step": 164850
    },
    {
      "epoch": 67.48,
      "grad_norm": 1.8423398733139038,
      "learning_rate": 3.7061974066820257e-06,
      "loss": 0.5277,
      "step": 164860
    },
    {
      "epoch": 67.49,
      "grad_norm": 1.7798388004302979,
      "learning_rate": 3.706068329626058e-06,
      "loss": 0.5097,
      "step": 164870
    },
    {
      "epoch": 67.49,
      "grad_norm": 2.3492860794067383,
      "learning_rate": 3.7059392463295206e-06,
      "loss": 0.5179,
      "step": 164880
    },
    {
      "epoch": 67.49,
      "grad_norm": 1.7827212810516357,
      "learning_rate": 3.705810156793005e-06,
      "loss": 0.5401,
      "step": 164890
    },
    {
      "epoch": 67.5,
      "grad_norm": 2.0511069297790527,
      "learning_rate": 3.7056810610171026e-06,
      "loss": 0.5303,
      "step": 164900
    },
    {
      "epoch": 67.5,
      "grad_norm": 1.9956461191177368,
      "learning_rate": 3.7055519590024056e-06,
      "loss": 0.5135,
      "step": 164910
    },
    {
      "epoch": 67.51,
      "grad_norm": 2.465836524963379,
      "learning_rate": 3.7054228507495045e-06,
      "loss": 0.5337,
      "step": 164920
    },
    {
      "epoch": 67.51,
      "grad_norm": 2.5856592655181885,
      "learning_rate": 3.7052937362589905e-06,
      "loss": 0.5435,
      "step": 164930
    },
    {
      "epoch": 67.52,
      "grad_norm": 1.9885362386703491,
      "learning_rate": 3.705164615531456e-06,
      "loss": 0.5212,
      "step": 164940
    },
    {
      "epoch": 67.52,
      "grad_norm": 1.484127402305603,
      "learning_rate": 3.7050354885674923e-06,
      "loss": 0.5419,
      "step": 164950
    },
    {
      "epoch": 67.52,
      "grad_norm": 1.8049677610397339,
      "learning_rate": 3.70490635536769e-06,
      "loss": 0.519,
      "step": 164960
    },
    {
      "epoch": 67.53,
      "grad_norm": 1.9925248622894287,
      "learning_rate": 3.7047772159326414e-06,
      "loss": 0.5189,
      "step": 164970
    },
    {
      "epoch": 67.53,
      "grad_norm": 2.2136073112487793,
      "learning_rate": 3.704648070262938e-06,
      "loss": 0.5209,
      "step": 164980
    },
    {
      "epoch": 67.54,
      "grad_norm": 1.8933662176132202,
      "learning_rate": 3.7045189183591713e-06,
      "loss": 0.5129,
      "step": 164990
    },
    {
      "epoch": 67.54,
      "grad_norm": 2.110398530960083,
      "learning_rate": 3.7043897602219326e-06,
      "loss": 0.5102,
      "step": 165000
    },
    {
      "epoch": 67.54,
      "grad_norm": 2.137322187423706,
      "learning_rate": 3.704260595851814e-06,
      "loss": 0.5098,
      "step": 165010
    },
    {
      "epoch": 67.55,
      "grad_norm": 2.023683786392212,
      "learning_rate": 3.704131425249408e-06,
      "loss": 0.51,
      "step": 165020
    },
    {
      "epoch": 67.55,
      "grad_norm": 1.9929571151733398,
      "learning_rate": 3.704002248415304e-06,
      "loss": 0.5069,
      "step": 165030
    },
    {
      "epoch": 67.56,
      "grad_norm": 1.6636452674865723,
      "learning_rate": 3.703873065350096e-06,
      "loss": 0.523,
      "step": 165040
    },
    {
      "epoch": 67.56,
      "grad_norm": 2.0230791568756104,
      "learning_rate": 3.703743876054374e-06,
      "loss": 0.5131,
      "step": 165050
    },
    {
      "epoch": 67.56,
      "grad_norm": 2.3495819568634033,
      "learning_rate": 3.703614680528732e-06,
      "loss": 0.5043,
      "step": 165060
    },
    {
      "epoch": 67.57,
      "grad_norm": 2.007122039794922,
      "learning_rate": 3.703485478773759e-06,
      "loss": 0.5125,
      "step": 165070
    },
    {
      "epoch": 67.57,
      "grad_norm": 1.9169447422027588,
      "learning_rate": 3.703356270790049e-06,
      "loss": 0.5032,
      "step": 165080
    },
    {
      "epoch": 67.58,
      "grad_norm": 1.817618489265442,
      "learning_rate": 3.7032270565781934e-06,
      "loss": 0.5097,
      "step": 165090
    },
    {
      "epoch": 67.58,
      "grad_norm": 2.116849184036255,
      "learning_rate": 3.703097836138784e-06,
      "loss": 0.5213,
      "step": 165100
    },
    {
      "epoch": 67.58,
      "grad_norm": 1.7196482419967651,
      "learning_rate": 3.702968609472412e-06,
      "loss": 0.5307,
      "step": 165110
    },
    {
      "epoch": 67.59,
      "grad_norm": 2.165236234664917,
      "learning_rate": 3.7028393765796713e-06,
      "loss": 0.51,
      "step": 165120
    },
    {
      "epoch": 67.59,
      "grad_norm": 1.908217191696167,
      "learning_rate": 3.7027101374611514e-06,
      "loss": 0.501,
      "step": 165130
    },
    {
      "epoch": 67.6,
      "grad_norm": 1.758704423904419,
      "learning_rate": 3.7025808921174467e-06,
      "loss": 0.5145,
      "step": 165140
    },
    {
      "epoch": 67.6,
      "grad_norm": 1.751267671585083,
      "learning_rate": 3.7024516405491476e-06,
      "loss": 0.5305,
      "step": 165150
    },
    {
      "epoch": 67.61,
      "grad_norm": 1.793142557144165,
      "learning_rate": 3.7023223827568466e-06,
      "loss": 0.534,
      "step": 165160
    },
    {
      "epoch": 67.61,
      "grad_norm": 1.4735095500946045,
      "learning_rate": 3.7021931187411357e-06,
      "loss": 0.5206,
      "step": 165170
    },
    {
      "epoch": 67.61,
      "grad_norm": 2.1389920711517334,
      "learning_rate": 3.7020638485026076e-06,
      "loss": 0.5347,
      "step": 165180
    },
    {
      "epoch": 67.62,
      "grad_norm": 1.9728261232376099,
      "learning_rate": 3.7019345720418537e-06,
      "loss": 0.4964,
      "step": 165190
    },
    {
      "epoch": 67.62,
      "grad_norm": 1.750260829925537,
      "learning_rate": 3.701805289359467e-06,
      "loss": 0.511,
      "step": 165200
    },
    {
      "epoch": 67.63,
      "grad_norm": 2.1953015327453613,
      "learning_rate": 3.7016760004560394e-06,
      "loss": 0.5376,
      "step": 165210
    },
    {
      "epoch": 67.63,
      "grad_norm": 2.4648613929748535,
      "learning_rate": 3.7015467053321627e-06,
      "loss": 0.5091,
      "step": 165220
    },
    {
      "epoch": 67.63,
      "grad_norm": 2.003817319869995,
      "learning_rate": 3.7014174039884302e-06,
      "loss": 0.5343,
      "step": 165230
    },
    {
      "epoch": 67.64,
      "grad_norm": 2.4435911178588867,
      "learning_rate": 3.7012880964254338e-06,
      "loss": 0.524,
      "step": 165240
    },
    {
      "epoch": 67.64,
      "grad_norm": 1.4572042226791382,
      "learning_rate": 3.701158782643765e-06,
      "loss": 0.5057,
      "step": 165250
    },
    {
      "epoch": 67.65,
      "grad_norm": 1.703974962234497,
      "learning_rate": 3.7010294626440167e-06,
      "loss": 0.5003,
      "step": 165260
    },
    {
      "epoch": 67.65,
      "grad_norm": 1.640300989151001,
      "learning_rate": 3.7009001364267814e-06,
      "loss": 0.511,
      "step": 165270
    },
    {
      "epoch": 67.65,
      "grad_norm": 1.8646494150161743,
      "learning_rate": 3.7007708039926513e-06,
      "loss": 0.5249,
      "step": 165280
    },
    {
      "epoch": 67.66,
      "grad_norm": 2.831761598587036,
      "learning_rate": 3.7006414653422197e-06,
      "loss": 0.5017,
      "step": 165290
    },
    {
      "epoch": 67.66,
      "grad_norm": 2.0933799743652344,
      "learning_rate": 3.7005121204760777e-06,
      "loss": 0.5276,
      "step": 165300
    },
    {
      "epoch": 67.67,
      "grad_norm": 1.8497416973114014,
      "learning_rate": 3.7003827693948197e-06,
      "loss": 0.5072,
      "step": 165310
    },
    {
      "epoch": 67.67,
      "grad_norm": 1.6866116523742676,
      "learning_rate": 3.7002534120990363e-06,
      "loss": 0.521,
      "step": 165320
    },
    {
      "epoch": 67.67,
      "grad_norm": 1.9423078298568726,
      "learning_rate": 3.7001240485893206e-06,
      "loss": 0.5274,
      "step": 165330
    },
    {
      "epoch": 67.68,
      "grad_norm": 1.9378598928451538,
      "learning_rate": 3.6999946788662654e-06,
      "loss": 0.5159,
      "step": 165340
    },
    {
      "epoch": 67.68,
      "grad_norm": 1.7725999355316162,
      "learning_rate": 3.6998653029304645e-06,
      "loss": 0.5046,
      "step": 165350
    },
    {
      "epoch": 67.69,
      "grad_norm": 1.7970491647720337,
      "learning_rate": 3.6997359207825087e-06,
      "loss": 0.5166,
      "step": 165360
    },
    {
      "epoch": 67.69,
      "grad_norm": 1.6808040142059326,
      "learning_rate": 3.6996065324229906e-06,
      "loss": 0.5167,
      "step": 165370
    },
    {
      "epoch": 67.7,
      "grad_norm": 1.6731882095336914,
      "learning_rate": 3.699477137852505e-06,
      "loss": 0.5238,
      "step": 165380
    },
    {
      "epoch": 67.7,
      "grad_norm": 2.039808750152588,
      "learning_rate": 3.6993477370716426e-06,
      "loss": 0.5262,
      "step": 165390
    },
    {
      "epoch": 67.7,
      "grad_norm": 1.9442884922027588,
      "learning_rate": 3.699218330080997e-06,
      "loss": 0.4918,
      "step": 165400
    },
    {
      "epoch": 67.71,
      "grad_norm": 1.6815309524536133,
      "learning_rate": 3.699088916881161e-06,
      "loss": 0.5268,
      "step": 165410
    },
    {
      "epoch": 67.71,
      "grad_norm": 2.2724812030792236,
      "learning_rate": 3.6989594974727274e-06,
      "loss": 0.5231,
      "step": 165420
    },
    {
      "epoch": 67.72,
      "grad_norm": 2.344792127609253,
      "learning_rate": 3.698830071856289e-06,
      "loss": 0.5091,
      "step": 165430
    },
    {
      "epoch": 67.72,
      "grad_norm": 1.959713101387024,
      "learning_rate": 3.6987006400324384e-06,
      "loss": 0.5478,
      "step": 165440
    },
    {
      "epoch": 67.72,
      "grad_norm": 2.052273750305176,
      "learning_rate": 3.6985712020017694e-06,
      "loss": 0.5403,
      "step": 165450
    },
    {
      "epoch": 67.73,
      "grad_norm": 2.2403411865234375,
      "learning_rate": 3.6984417577648735e-06,
      "loss": 0.5245,
      "step": 165460
    },
    {
      "epoch": 67.73,
      "grad_norm": 1.7191983461380005,
      "learning_rate": 3.6983123073223444e-06,
      "loss": 0.4945,
      "step": 165470
    },
    {
      "epoch": 67.74,
      "grad_norm": 1.7452651262283325,
      "learning_rate": 3.698182850674776e-06,
      "loss": 0.4948,
      "step": 165480
    },
    {
      "epoch": 67.74,
      "grad_norm": 2.0395760536193848,
      "learning_rate": 3.69805338782276e-06,
      "loss": 0.5007,
      "step": 165490
    },
    {
      "epoch": 67.74,
      "grad_norm": 1.7056245803833008,
      "learning_rate": 3.6979239187668896e-06,
      "loss": 0.5457,
      "step": 165500
    },
    {
      "epoch": 67.75,
      "grad_norm": 1.715720772743225,
      "learning_rate": 3.6977944435077584e-06,
      "loss": 0.5128,
      "step": 165510
    },
    {
      "epoch": 67.75,
      "grad_norm": 1.7317038774490356,
      "learning_rate": 3.6976649620459595e-06,
      "loss": 0.5267,
      "step": 165520
    },
    {
      "epoch": 67.76,
      "grad_norm": 2.024616003036499,
      "learning_rate": 3.697535474382086e-06,
      "loss": 0.5173,
      "step": 165530
    },
    {
      "epoch": 67.76,
      "grad_norm": 1.9730311632156372,
      "learning_rate": 3.697405980516731e-06,
      "loss": 0.5133,
      "step": 165540
    },
    {
      "epoch": 67.77,
      "grad_norm": 1.7073360681533813,
      "learning_rate": 3.697276480450487e-06,
      "loss": 0.5111,
      "step": 165550
    },
    {
      "epoch": 67.77,
      "grad_norm": 1.7858625650405884,
      "learning_rate": 3.6971469741839477e-06,
      "loss": 0.5233,
      "step": 165560
    },
    {
      "epoch": 67.77,
      "grad_norm": 1.7851561307907104,
      "learning_rate": 3.6970174617177063e-06,
      "loss": 0.5344,
      "step": 165570
    },
    {
      "epoch": 67.78,
      "grad_norm": 1.907270073890686,
      "learning_rate": 3.6968879430523573e-06,
      "loss": 0.5105,
      "step": 165580
    },
    {
      "epoch": 67.78,
      "grad_norm": 1.838478446006775,
      "learning_rate": 3.6967584181884924e-06,
      "loss": 0.5209,
      "step": 165590
    },
    {
      "epoch": 67.79,
      "grad_norm": 1.9058631658554077,
      "learning_rate": 3.6966288871267057e-06,
      "loss": 0.5179,
      "step": 165600
    },
    {
      "epoch": 67.79,
      "grad_norm": 1.860472559928894,
      "learning_rate": 3.6964993498675898e-06,
      "loss": 0.523,
      "step": 165610
    },
    {
      "epoch": 67.79,
      "grad_norm": 1.7134041786193848,
      "learning_rate": 3.6963698064117387e-06,
      "loss": 0.5177,
      "step": 165620
    },
    {
      "epoch": 67.8,
      "grad_norm": 1.881276249885559,
      "learning_rate": 3.6962402567597464e-06,
      "loss": 0.5136,
      "step": 165630
    },
    {
      "epoch": 67.8,
      "grad_norm": 1.7754952907562256,
      "learning_rate": 3.6961107009122057e-06,
      "loss": 0.5471,
      "step": 165640
    },
    {
      "epoch": 67.81,
      "grad_norm": 2.0098392963409424,
      "learning_rate": 3.695981138869709e-06,
      "loss": 0.5195,
      "step": 165650
    },
    {
      "epoch": 67.81,
      "grad_norm": 2.1923277378082275,
      "learning_rate": 3.6958515706328517e-06,
      "loss": 0.5368,
      "step": 165660
    },
    {
      "epoch": 67.81,
      "grad_norm": 2.761054277420044,
      "learning_rate": 3.695721996202226e-06,
      "loss": 0.5268,
      "step": 165670
    },
    {
      "epoch": 67.82,
      "grad_norm": 1.6584081649780273,
      "learning_rate": 3.695592415578427e-06,
      "loss": 0.5043,
      "step": 165680
    },
    {
      "epoch": 67.82,
      "grad_norm": 1.5281227827072144,
      "learning_rate": 3.695462828762046e-06,
      "loss": 0.5162,
      "step": 165690
    },
    {
      "epoch": 67.83,
      "grad_norm": 1.6956661939620972,
      "learning_rate": 3.6953332357536785e-06,
      "loss": 0.5206,
      "step": 165700
    },
    {
      "epoch": 67.83,
      "grad_norm": 1.8187230825424194,
      "learning_rate": 3.695203636553918e-06,
      "loss": 0.5185,
      "step": 165710
    },
    {
      "epoch": 67.83,
      "grad_norm": 2.2715237140655518,
      "learning_rate": 3.6950740311633566e-06,
      "loss": 0.5481,
      "step": 165720
    },
    {
      "epoch": 67.84,
      "grad_norm": 1.838904857635498,
      "learning_rate": 3.6949444195825895e-06,
      "loss": 0.5184,
      "step": 165730
    },
    {
      "epoch": 67.84,
      "grad_norm": 2.0274648666381836,
      "learning_rate": 3.69481480181221e-06,
      "loss": 0.5186,
      "step": 165740
    },
    {
      "epoch": 67.85,
      "grad_norm": 2.2104713916778564,
      "learning_rate": 3.694685177852812e-06,
      "loss": 0.5144,
      "step": 165750
    },
    {
      "epoch": 67.85,
      "grad_norm": 1.8750264644622803,
      "learning_rate": 3.6945555477049893e-06,
      "loss": 0.5199,
      "step": 165760
    },
    {
      "epoch": 67.86,
      "grad_norm": 1.5890424251556396,
      "learning_rate": 3.694425911369336e-06,
      "loss": 0.5313,
      "step": 165770
    },
    {
      "epoch": 67.86,
      "grad_norm": 1.7558655738830566,
      "learning_rate": 3.6942962688464446e-06,
      "loss": 0.5313,
      "step": 165780
    },
    {
      "epoch": 67.86,
      "grad_norm": 1.681584358215332,
      "learning_rate": 3.69416662013691e-06,
      "loss": 0.5113,
      "step": 165790
    },
    {
      "epoch": 67.87,
      "grad_norm": 1.6961504220962524,
      "learning_rate": 3.6940369652413267e-06,
      "loss": 0.5107,
      "step": 165800
    },
    {
      "epoch": 67.87,
      "grad_norm": 2.040473461151123,
      "learning_rate": 3.6939073041602876e-06,
      "loss": 0.5406,
      "step": 165810
    },
    {
      "epoch": 67.88,
      "grad_norm": 2.043955087661743,
      "learning_rate": 3.693777636894387e-06,
      "loss": 0.5119,
      "step": 165820
    },
    {
      "epoch": 67.88,
      "grad_norm": 2.509394407272339,
      "learning_rate": 3.693647963444219e-06,
      "loss": 0.5265,
      "step": 165830
    },
    {
      "epoch": 67.88,
      "grad_norm": 1.648661494255066,
      "learning_rate": 3.693518283810377e-06,
      "loss": 0.5245,
      "step": 165840
    },
    {
      "epoch": 67.89,
      "grad_norm": 1.986120343208313,
      "learning_rate": 3.693388597993456e-06,
      "loss": 0.5108,
      "step": 165850
    },
    {
      "epoch": 67.89,
      "grad_norm": 1.856823444366455,
      "learning_rate": 3.6932589059940492e-06,
      "loss": 0.5194,
      "step": 165860
    },
    {
      "epoch": 67.9,
      "grad_norm": 1.7262011766433716,
      "learning_rate": 3.6931292078127514e-06,
      "loss": 0.5142,
      "step": 165870
    },
    {
      "epoch": 67.9,
      "grad_norm": 2.2397913932800293,
      "learning_rate": 3.692999503450156e-06,
      "loss": 0.5418,
      "step": 165880
    },
    {
      "epoch": 67.9,
      "grad_norm": 1.5680365562438965,
      "learning_rate": 3.6928697929068584e-06,
      "loss": 0.5132,
      "step": 165890
    },
    {
      "epoch": 67.91,
      "grad_norm": 1.5841145515441895,
      "learning_rate": 3.692740076183451e-06,
      "loss": 0.4981,
      "step": 165900
    },
    {
      "epoch": 67.91,
      "grad_norm": 1.521743893623352,
      "learning_rate": 3.6926103532805295e-06,
      "loss": 0.525,
      "step": 165910
    },
    {
      "epoch": 67.92,
      "grad_norm": 2.1734132766723633,
      "learning_rate": 3.6924806241986875e-06,
      "loss": 0.4984,
      "step": 165920
    },
    {
      "epoch": 67.92,
      "grad_norm": 2.8712165355682373,
      "learning_rate": 3.6923508889385194e-06,
      "loss": 0.5157,
      "step": 165930
    },
    {
      "epoch": 67.92,
      "grad_norm": 1.905537724494934,
      "learning_rate": 3.6922211475006195e-06,
      "loss": 0.5233,
      "step": 165940
    },
    {
      "epoch": 67.93,
      "grad_norm": 1.618035912513733,
      "learning_rate": 3.6920913998855823e-06,
      "loss": 0.5127,
      "step": 165950
    },
    {
      "epoch": 67.93,
      "grad_norm": 1.509650707244873,
      "learning_rate": 3.691961646094002e-06,
      "loss": 0.5181,
      "step": 165960
    },
    {
      "epoch": 67.94,
      "grad_norm": 1.6188535690307617,
      "learning_rate": 3.6918318861264724e-06,
      "loss": 0.4953,
      "step": 165970
    },
    {
      "epoch": 67.94,
      "grad_norm": 1.868493676185608,
      "learning_rate": 3.6917021199835894e-06,
      "loss": 0.5204,
      "step": 165980
    },
    {
      "epoch": 67.95,
      "grad_norm": 2.4201605319976807,
      "learning_rate": 3.691572347665946e-06,
      "loss": 0.5128,
      "step": 165990
    },
    {
      "epoch": 67.95,
      "grad_norm": 1.5553830862045288,
      "learning_rate": 3.691442569174137e-06,
      "loss": 0.5195,
      "step": 166000
    },
    {
      "epoch": 67.95,
      "grad_norm": 1.9628665447235107,
      "learning_rate": 3.691312784508757e-06,
      "loss": 0.538,
      "step": 166010
    },
    {
      "epoch": 67.96,
      "grad_norm": 2.2754151821136475,
      "learning_rate": 3.6911829936704007e-06,
      "loss": 0.5333,
      "step": 166020
    },
    {
      "epoch": 67.96,
      "grad_norm": 1.9193801879882812,
      "learning_rate": 3.6910531966596622e-06,
      "loss": 0.5057,
      "step": 166030
    },
    {
      "epoch": 67.97,
      "grad_norm": 1.9156173467636108,
      "learning_rate": 3.6909233934771368e-06,
      "loss": 0.5198,
      "step": 166040
    },
    {
      "epoch": 67.97,
      "grad_norm": 2.184762716293335,
      "learning_rate": 3.6907935841234185e-06,
      "loss": 0.5349,
      "step": 166050
    },
    {
      "epoch": 67.97,
      "grad_norm": 1.9152852296829224,
      "learning_rate": 3.690663768599103e-06,
      "loss": 0.5198,
      "step": 166060
    },
    {
      "epoch": 67.98,
      "grad_norm": 1.8652141094207764,
      "learning_rate": 3.6905339469047834e-06,
      "loss": 0.5142,
      "step": 166070
    },
    {
      "epoch": 67.98,
      "grad_norm": 2.2311511039733887,
      "learning_rate": 3.690404119041055e-06,
      "loss": 0.5166,
      "step": 166080
    },
    {
      "epoch": 67.99,
      "grad_norm": 1.7803921699523926,
      "learning_rate": 3.690274285008513e-06,
      "loss": 0.5154,
      "step": 166090
    },
    {
      "epoch": 67.99,
      "grad_norm": 1.8641011714935303,
      "learning_rate": 3.6901444448077513e-06,
      "loss": 0.5262,
      "step": 166100
    },
    {
      "epoch": 67.99,
      "grad_norm": 2.7277615070343018,
      "learning_rate": 3.6900145984393658e-06,
      "loss": 0.5274,
      "step": 166110
    },
    {
      "epoch": 68.0,
      "grad_norm": 1.8178457021713257,
      "learning_rate": 3.6898847459039507e-06,
      "loss": 0.5162,
      "step": 166120
    },
    {
      "epoch": 68.0,
      "eval_loss": 0.5165771842002869,
      "eval_runtime": 61.5374,
      "eval_samples_per_second": 56.047,
      "eval_steps_per_second": 7.02,
      "step": 166124
    },
    {
      "epoch": 68.0,
      "grad_norm": 1.9875566959381104,
      "learning_rate": 3.6897548872021e-06,
      "loss": 0.5057,
      "step": 166130
    },
    {
      "epoch": 68.01,
      "grad_norm": 2.008148193359375,
      "learning_rate": 3.6896250223344096e-06,
      "loss": 0.5309,
      "step": 166140
    },
    {
      "epoch": 68.01,
      "grad_norm": 2.12567138671875,
      "learning_rate": 3.6894951513014746e-06,
      "loss": 0.5311,
      "step": 166150
    },
    {
      "epoch": 68.01,
      "grad_norm": 1.7640740871429443,
      "learning_rate": 3.689365274103889e-06,
      "loss": 0.5371,
      "step": 166160
    },
    {
      "epoch": 68.02,
      "grad_norm": 2.020801067352295,
      "learning_rate": 3.6892353907422486e-06,
      "loss": 0.5219,
      "step": 166170
    },
    {
      "epoch": 68.02,
      "grad_norm": 1.7917999029159546,
      "learning_rate": 3.689105501217148e-06,
      "loss": 0.5167,
      "step": 166180
    },
    {
      "epoch": 68.03,
      "grad_norm": 2.0969948768615723,
      "learning_rate": 3.6889756055291822e-06,
      "loss": 0.5105,
      "step": 166190
    },
    {
      "epoch": 68.03,
      "grad_norm": 1.8129383325576782,
      "learning_rate": 3.6888457036789456e-06,
      "loss": 0.514,
      "step": 166200
    },
    {
      "epoch": 68.04,
      "grad_norm": 2.1108932495117188,
      "learning_rate": 3.6887157956670344e-06,
      "loss": 0.5136,
      "step": 166210
    },
    {
      "epoch": 68.04,
      "grad_norm": 2.172776460647583,
      "learning_rate": 3.688585881494043e-06,
      "loss": 0.523,
      "step": 166220
    },
    {
      "epoch": 68.04,
      "grad_norm": 2.158778667449951,
      "learning_rate": 3.6884559611605675e-06,
      "loss": 0.508,
      "step": 166230
    },
    {
      "epoch": 68.05,
      "grad_norm": 1.793658971786499,
      "learning_rate": 3.688326034667201e-06,
      "loss": 0.5328,
      "step": 166240
    },
    {
      "epoch": 68.05,
      "grad_norm": 2.033906936645508,
      "learning_rate": 3.688196102014541e-06,
      "loss": 0.5403,
      "step": 166250
    },
    {
      "epoch": 68.06,
      "grad_norm": 1.95950448513031,
      "learning_rate": 3.6880661632031813e-06,
      "loss": 0.5189,
      "step": 166260
    },
    {
      "epoch": 68.06,
      "grad_norm": 2.0430471897125244,
      "learning_rate": 3.6879362182337177e-06,
      "loss": 0.5201,
      "step": 166270
    },
    {
      "epoch": 68.06,
      "grad_norm": 2.4698071479797363,
      "learning_rate": 3.6878062671067444e-06,
      "loss": 0.5124,
      "step": 166280
    },
    {
      "epoch": 68.07,
      "grad_norm": 1.6574985980987549,
      "learning_rate": 3.687676309822858e-06,
      "loss": 0.524,
      "step": 166290
    },
    {
      "epoch": 68.07,
      "grad_norm": 2.0504233837127686,
      "learning_rate": 3.687546346382653e-06,
      "loss": 0.5121,
      "step": 166300
    },
    {
      "epoch": 68.08,
      "grad_norm": 1.8994089365005493,
      "learning_rate": 3.6874163767867256e-06,
      "loss": 0.5108,
      "step": 166310
    },
    {
      "epoch": 68.08,
      "grad_norm": 1.9893492460250854,
      "learning_rate": 3.68728640103567e-06,
      "loss": 0.5261,
      "step": 166320
    },
    {
      "epoch": 68.08,
      "grad_norm": 1.7836723327636719,
      "learning_rate": 3.6871564191300823e-06,
      "loss": 0.5384,
      "step": 166330
    },
    {
      "epoch": 68.09,
      "grad_norm": 2.040889024734497,
      "learning_rate": 3.6870264310705586e-06,
      "loss": 0.4979,
      "step": 166340
    },
    {
      "epoch": 68.09,
      "grad_norm": 2.8802034854888916,
      "learning_rate": 3.686896436857693e-06,
      "loss": 0.5257,
      "step": 166350
    },
    {
      "epoch": 68.1,
      "grad_norm": 1.6861262321472168,
      "learning_rate": 3.686766436492082e-06,
      "loss": 0.518,
      "step": 166360
    },
    {
      "epoch": 68.1,
      "grad_norm": 2.4341602325439453,
      "learning_rate": 3.6866364299743203e-06,
      "loss": 0.502,
      "step": 166370
    },
    {
      "epoch": 68.1,
      "grad_norm": 1.6411771774291992,
      "learning_rate": 3.686506417305003e-06,
      "loss": 0.5243,
      "step": 166380
    },
    {
      "epoch": 68.11,
      "grad_norm": 2.088787078857422,
      "learning_rate": 3.686376398484728e-06,
      "loss": 0.5122,
      "step": 166390
    },
    {
      "epoch": 68.11,
      "grad_norm": 1.7543195486068726,
      "learning_rate": 3.6862463735140884e-06,
      "loss": 0.5196,
      "step": 166400
    },
    {
      "epoch": 68.12,
      "grad_norm": 2.134930372238159,
      "learning_rate": 3.6861163423936814e-06,
      "loss": 0.5249,
      "step": 166410
    },
    {
      "epoch": 68.12,
      "grad_norm": 2.3341522216796875,
      "learning_rate": 3.6859863051241015e-06,
      "loss": 0.508,
      "step": 166420
    },
    {
      "epoch": 68.13,
      "grad_norm": 1.7053998708724976,
      "learning_rate": 3.685856261705946e-06,
      "loss": 0.523,
      "step": 166430
    },
    {
      "epoch": 68.13,
      "grad_norm": 1.9353183507919312,
      "learning_rate": 3.6857262121398086e-06,
      "loss": 0.5082,
      "step": 166440
    },
    {
      "epoch": 68.13,
      "grad_norm": 1.59817636013031,
      "learning_rate": 3.6855961564262862e-06,
      "loss": 0.5299,
      "step": 166450
    },
    {
      "epoch": 68.14,
      "grad_norm": 1.7951509952545166,
      "learning_rate": 3.685466094565975e-06,
      "loss": 0.5051,
      "step": 166460
    },
    {
      "epoch": 68.14,
      "grad_norm": 2.7191519737243652,
      "learning_rate": 3.6853360265594696e-06,
      "loss": 0.5142,
      "step": 166470
    },
    {
      "epoch": 68.15,
      "grad_norm": 1.626170039176941,
      "learning_rate": 3.685205952407367e-06,
      "loss": 0.512,
      "step": 166480
    },
    {
      "epoch": 68.15,
      "grad_norm": 2.314171314239502,
      "learning_rate": 3.685075872110261e-06,
      "loss": 0.4979,
      "step": 166490
    },
    {
      "epoch": 68.15,
      "grad_norm": 1.868282437324524,
      "learning_rate": 3.68494578566875e-06,
      "loss": 0.5111,
      "step": 166500
    },
    {
      "epoch": 68.16,
      "grad_norm": 2.66656231880188,
      "learning_rate": 3.6848156930834284e-06,
      "loss": 0.5206,
      "step": 166510
    },
    {
      "epoch": 68.16,
      "grad_norm": 1.6640002727508545,
      "learning_rate": 3.6846855943548925e-06,
      "loss": 0.5036,
      "step": 166520
    },
    {
      "epoch": 68.17,
      "grad_norm": 2.0781211853027344,
      "learning_rate": 3.684555489483739e-06,
      "loss": 0.5341,
      "step": 166530
    },
    {
      "epoch": 68.17,
      "grad_norm": 1.6765565872192383,
      "learning_rate": 3.6844253784705626e-06,
      "loss": 0.5261,
      "step": 166540
    },
    {
      "epoch": 68.17,
      "grad_norm": 2.1655845642089844,
      "learning_rate": 3.6842952613159603e-06,
      "loss": 0.5202,
      "step": 166550
    },
    {
      "epoch": 68.18,
      "grad_norm": 1.9227001667022705,
      "learning_rate": 3.684165138020527e-06,
      "loss": 0.5233,
      "step": 166560
    },
    {
      "epoch": 68.18,
      "grad_norm": 2.089853286743164,
      "learning_rate": 3.6840350085848607e-06,
      "loss": 0.5172,
      "step": 166570
    },
    {
      "epoch": 68.19,
      "grad_norm": 2.429070472717285,
      "learning_rate": 3.683904873009555e-06,
      "loss": 0.5106,
      "step": 166580
    },
    {
      "epoch": 68.19,
      "grad_norm": 1.9567664861679077,
      "learning_rate": 3.6837747312952085e-06,
      "loss": 0.5308,
      "step": 166590
    },
    {
      "epoch": 68.19,
      "grad_norm": 2.3709359169006348,
      "learning_rate": 3.6836445834424157e-06,
      "loss": 0.5079,
      "step": 166600
    },
    {
      "epoch": 68.2,
      "grad_norm": 2.8623836040496826,
      "learning_rate": 3.683514429451773e-06,
      "loss": 0.5102,
      "step": 166610
    },
    {
      "epoch": 68.2,
      "grad_norm": 2.6235177516937256,
      "learning_rate": 3.6833842693238776e-06,
      "loss": 0.5313,
      "step": 166620
    },
    {
      "epoch": 68.21,
      "grad_norm": 1.9042965173721313,
      "learning_rate": 3.6832541030593254e-06,
      "loss": 0.5065,
      "step": 166630
    },
    {
      "epoch": 68.21,
      "grad_norm": 1.8191732168197632,
      "learning_rate": 3.6831239306587114e-06,
      "loss": 0.5222,
      "step": 166640
    },
    {
      "epoch": 68.22,
      "grad_norm": 1.902307152748108,
      "learning_rate": 3.6829937521226338e-06,
      "loss": 0.5265,
      "step": 166650
    },
    {
      "epoch": 68.22,
      "grad_norm": 1.7004755735397339,
      "learning_rate": 3.6828635674516875e-06,
      "loss": 0.5205,
      "step": 166660
    },
    {
      "epoch": 68.22,
      "grad_norm": 2.0640106201171875,
      "learning_rate": 3.6827333766464694e-06,
      "loss": 0.4932,
      "step": 166670
    },
    {
      "epoch": 68.23,
      "grad_norm": 2.022430896759033,
      "learning_rate": 3.6826031797075756e-06,
      "loss": 0.5096,
      "step": 166680
    },
    {
      "epoch": 68.23,
      "grad_norm": 2.6353518962860107,
      "learning_rate": 3.6824729766356026e-06,
      "loss": 0.515,
      "step": 166690
    },
    {
      "epoch": 68.24,
      "grad_norm": 1.7754340171813965,
      "learning_rate": 3.682342767431147e-06,
      "loss": 0.5011,
      "step": 166700
    },
    {
      "epoch": 68.24,
      "grad_norm": 1.6680160760879517,
      "learning_rate": 3.682212552094806e-06,
      "loss": 0.5227,
      "step": 166710
    },
    {
      "epoch": 68.24,
      "grad_norm": 1.8972878456115723,
      "learning_rate": 3.682082330627175e-06,
      "loss": 0.5051,
      "step": 166720
    },
    {
      "epoch": 68.25,
      "grad_norm": 2.1073532104492188,
      "learning_rate": 3.68195210302885e-06,
      "loss": 0.5118,
      "step": 166730
    },
    {
      "epoch": 68.25,
      "grad_norm": 1.7538539171218872,
      "learning_rate": 3.6818218693004295e-06,
      "loss": 0.5233,
      "step": 166740
    },
    {
      "epoch": 68.26,
      "grad_norm": 1.688747763633728,
      "learning_rate": 3.6816916294425085e-06,
      "loss": 0.5006,
      "step": 166750
    },
    {
      "epoch": 68.26,
      "grad_norm": 2.2674190998077393,
      "learning_rate": 3.6815613834556845e-06,
      "loss": 0.5118,
      "step": 166760
    },
    {
      "epoch": 68.26,
      "grad_norm": 1.999815583229065,
      "learning_rate": 3.681431131340554e-06,
      "loss": 0.5009,
      "step": 166770
    },
    {
      "epoch": 68.27,
      "grad_norm": 1.7168196439743042,
      "learning_rate": 3.6813008730977124e-06,
      "loss": 0.5156,
      "step": 166780
    },
    {
      "epoch": 68.27,
      "grad_norm": 2.2674105167388916,
      "learning_rate": 3.6811706087277573e-06,
      "loss": 0.5314,
      "step": 166790
    },
    {
      "epoch": 68.28,
      "grad_norm": 1.7703748941421509,
      "learning_rate": 3.681040338231286e-06,
      "loss": 0.5172,
      "step": 166800
    },
    {
      "epoch": 68.28,
      "grad_norm": 1.442775011062622,
      "learning_rate": 3.6809100616088945e-06,
      "loss": 0.4966,
      "step": 166810
    },
    {
      "epoch": 68.28,
      "grad_norm": 2.2804250717163086,
      "learning_rate": 3.68077977886118e-06,
      "loss": 0.5102,
      "step": 166820
    },
    {
      "epoch": 68.29,
      "grad_norm": 1.7153456211090088,
      "learning_rate": 3.6806494899887397e-06,
      "loss": 0.501,
      "step": 166830
    },
    {
      "epoch": 68.29,
      "grad_norm": 2.4788169860839844,
      "learning_rate": 3.6805191949921696e-06,
      "loss": 0.5129,
      "step": 166840
    },
    {
      "epoch": 68.3,
      "grad_norm": 1.432257890701294,
      "learning_rate": 3.6803888938720672e-06,
      "loss": 0.5191,
      "step": 166850
    },
    {
      "epoch": 68.3,
      "grad_norm": 2.0064094066619873,
      "learning_rate": 3.680258586629028e-06,
      "loss": 0.5178,
      "step": 166860
    },
    {
      "epoch": 68.31,
      "grad_norm": 2.0976295471191406,
      "learning_rate": 3.680128273263651e-06,
      "loss": 0.5189,
      "step": 166870
    },
    {
      "epoch": 68.31,
      "grad_norm": 1.6260864734649658,
      "learning_rate": 3.679997953776531e-06,
      "loss": 0.5263,
      "step": 166880
    },
    {
      "epoch": 68.31,
      "grad_norm": 1.7345645427703857,
      "learning_rate": 3.6798676281682665e-06,
      "loss": 0.5324,
      "step": 166890
    },
    {
      "epoch": 68.32,
      "grad_norm": 1.467527985572815,
      "learning_rate": 3.6797372964394545e-06,
      "loss": 0.5369,
      "step": 166900
    },
    {
      "epoch": 68.32,
      "grad_norm": 1.8095988035202026,
      "learning_rate": 3.679606958590691e-06,
      "loss": 0.524,
      "step": 166910
    },
    {
      "epoch": 68.33,
      "grad_norm": 1.9390407800674438,
      "learning_rate": 3.679476614622574e-06,
      "loss": 0.5349,
      "step": 166920
    },
    {
      "epoch": 68.33,
      "grad_norm": 1.9792749881744385,
      "learning_rate": 3.6793462645357007e-06,
      "loss": 0.5247,
      "step": 166930
    },
    {
      "epoch": 68.33,
      "grad_norm": 1.9770488739013672,
      "learning_rate": 3.6792159083306672e-06,
      "loss": 0.5264,
      "step": 166940
    },
    {
      "epoch": 68.34,
      "grad_norm": 2.240675210952759,
      "learning_rate": 3.679085546008071e-06,
      "loss": 0.5191,
      "step": 166950
    },
    {
      "epoch": 68.34,
      "grad_norm": 1.7286708354949951,
      "learning_rate": 3.6789551775685097e-06,
      "loss": 0.5116,
      "step": 166960
    },
    {
      "epoch": 68.35,
      "grad_norm": 2.4341423511505127,
      "learning_rate": 3.6788248030125803e-06,
      "loss": 0.4996,
      "step": 166970
    },
    {
      "epoch": 68.35,
      "grad_norm": 2.1926982402801514,
      "learning_rate": 3.67869442234088e-06,
      "loss": 0.5344,
      "step": 166980
    },
    {
      "epoch": 68.35,
      "grad_norm": 2.684056043624878,
      "learning_rate": 3.678564035554006e-06,
      "loss": 0.5177,
      "step": 166990
    },
    {
      "epoch": 68.36,
      "grad_norm": 1.8505326509475708,
      "learning_rate": 3.678433642652556e-06,
      "loss": 0.5421,
      "step": 167000
    },
    {
      "epoch": 68.36,
      "grad_norm": 1.8015952110290527,
      "learning_rate": 3.6783032436371268e-06,
      "loss": 0.5175,
      "step": 167010
    },
    {
      "epoch": 68.37,
      "grad_norm": 1.8550705909729004,
      "learning_rate": 3.6781728385083156e-06,
      "loss": 0.5512,
      "step": 167020
    },
    {
      "epoch": 68.37,
      "grad_norm": 2.078677177429199,
      "learning_rate": 3.6780424272667197e-06,
      "loss": 0.4976,
      "step": 167030
    },
    {
      "epoch": 68.37,
      "grad_norm": 1.8761659860610962,
      "learning_rate": 3.6779120099129374e-06,
      "loss": 0.5096,
      "step": 167040
    },
    {
      "epoch": 68.38,
      "grad_norm": 1.866276502609253,
      "learning_rate": 3.677781586447565e-06,
      "loss": 0.5139,
      "step": 167050
    },
    {
      "epoch": 68.38,
      "grad_norm": 2.084005117416382,
      "learning_rate": 3.6776511568712016e-06,
      "loss": 0.5394,
      "step": 167060
    },
    {
      "epoch": 68.39,
      "grad_norm": 1.4761348962783813,
      "learning_rate": 3.6775207211844423e-06,
      "loss": 0.5158,
      "step": 167070
    },
    {
      "epoch": 68.39,
      "grad_norm": 2.050755739212036,
      "learning_rate": 3.6773902793878866e-06,
      "loss": 0.5274,
      "step": 167080
    },
    {
      "epoch": 68.4,
      "grad_norm": 1.7466065883636475,
      "learning_rate": 3.6772598314821316e-06,
      "loss": 0.5258,
      "step": 167090
    },
    {
      "epoch": 68.4,
      "grad_norm": 1.6788420677185059,
      "learning_rate": 3.677129377467774e-06,
      "loss": 0.5292,
      "step": 167100
    },
    {
      "epoch": 68.4,
      "grad_norm": 1.7625962495803833,
      "learning_rate": 3.6769989173454115e-06,
      "loss": 0.5241,
      "step": 167110
    },
    {
      "epoch": 68.41,
      "grad_norm": 2.170992851257324,
      "learning_rate": 3.6768684511156428e-06,
      "loss": 0.5105,
      "step": 167120
    },
    {
      "epoch": 68.41,
      "grad_norm": 1.7357274293899536,
      "learning_rate": 3.6767379787790655e-06,
      "loss": 0.5102,
      "step": 167130
    },
    {
      "epoch": 68.42,
      "grad_norm": 2.088090419769287,
      "learning_rate": 3.676607500336276e-06,
      "loss": 0.538,
      "step": 167140
    },
    {
      "epoch": 68.42,
      "grad_norm": 1.6543947458267212,
      "learning_rate": 3.676477015787873e-06,
      "loss": 0.5293,
      "step": 167150
    },
    {
      "epoch": 68.42,
      "grad_norm": 1.5782333612442017,
      "learning_rate": 3.6763465251344537e-06,
      "loss": 0.5152,
      "step": 167160
    },
    {
      "epoch": 68.43,
      "grad_norm": 2.433795213699341,
      "learning_rate": 3.676216028376616e-06,
      "loss": 0.4928,
      "step": 167170
    },
    {
      "epoch": 68.43,
      "grad_norm": 1.8484013080596924,
      "learning_rate": 3.6760855255149577e-06,
      "loss": 0.5189,
      "step": 167180
    },
    {
      "epoch": 68.44,
      "grad_norm": 1.98676598072052,
      "learning_rate": 3.675955016550078e-06,
      "loss": 0.5098,
      "step": 167190
    },
    {
      "epoch": 68.44,
      "grad_norm": 2.532933235168457,
      "learning_rate": 3.6758245014825723e-06,
      "loss": 0.501,
      "step": 167200
    },
    {
      "epoch": 68.44,
      "grad_norm": 1.895599126815796,
      "learning_rate": 3.6756939803130397e-06,
      "loss": 0.5223,
      "step": 167210
    },
    {
      "epoch": 68.45,
      "grad_norm": 2.3139703273773193,
      "learning_rate": 3.675563453042078e-06,
      "loss": 0.5304,
      "step": 167220
    },
    {
      "epoch": 68.45,
      "grad_norm": 1.9782471656799316,
      "learning_rate": 3.675432919670285e-06,
      "loss": 0.5304,
      "step": 167230
    },
    {
      "epoch": 68.46,
      "grad_norm": 2.206688404083252,
      "learning_rate": 3.675302380198259e-06,
      "loss": 0.5026,
      "step": 167240
    },
    {
      "epoch": 68.46,
      "grad_norm": 1.3745182752609253,
      "learning_rate": 3.6751718346265974e-06,
      "loss": 0.5197,
      "step": 167250
    },
    {
      "epoch": 68.47,
      "grad_norm": 1.524267315864563,
      "learning_rate": 3.6750412829558987e-06,
      "loss": 0.5219,
      "step": 167260
    },
    {
      "epoch": 68.47,
      "grad_norm": 2.4799437522888184,
      "learning_rate": 3.674910725186761e-06,
      "loss": 0.5286,
      "step": 167270
    },
    {
      "epoch": 68.47,
      "grad_norm": 2.015390634536743,
      "learning_rate": 3.6747801613197828e-06,
      "loss": 0.5043,
      "step": 167280
    },
    {
      "epoch": 68.48,
      "grad_norm": 2.365990400314331,
      "learning_rate": 3.6746495913555607e-06,
      "loss": 0.5233,
      "step": 167290
    },
    {
      "epoch": 68.48,
      "grad_norm": 2.236342430114746,
      "learning_rate": 3.6745190152946935e-06,
      "loss": 0.5354,
      "step": 167300
    },
    {
      "epoch": 68.49,
      "grad_norm": 1.9199893474578857,
      "learning_rate": 3.6743884331377803e-06,
      "loss": 0.5159,
      "step": 167310
    },
    {
      "epoch": 68.49,
      "grad_norm": 1.762232780456543,
      "learning_rate": 3.6742578448854185e-06,
      "loss": 0.5256,
      "step": 167320
    },
    {
      "epoch": 68.49,
      "grad_norm": 1.4671273231506348,
      "learning_rate": 3.6741272505382057e-06,
      "loss": 0.5023,
      "step": 167330
    },
    {
      "epoch": 68.5,
      "grad_norm": 1.5905050039291382,
      "learning_rate": 3.673996650096741e-06,
      "loss": 0.5183,
      "step": 167340
    },
    {
      "epoch": 68.5,
      "grad_norm": 2.143939971923828,
      "learning_rate": 3.6738660435616225e-06,
      "loss": 0.501,
      "step": 167350
    },
    {
      "epoch": 68.51,
      "grad_norm": 2.4217376708984375,
      "learning_rate": 3.6737354309334477e-06,
      "loss": 0.5044,
      "step": 167360
    },
    {
      "epoch": 68.51,
      "grad_norm": 1.5997843742370605,
      "learning_rate": 3.6736048122128166e-06,
      "loss": 0.5161,
      "step": 167370
    },
    {
      "epoch": 68.51,
      "grad_norm": 1.6885987520217896,
      "learning_rate": 3.6734741874003265e-06,
      "loss": 0.5226,
      "step": 167380
    },
    {
      "epoch": 68.52,
      "grad_norm": 1.8118740320205688,
      "learning_rate": 3.6733435564965757e-06,
      "loss": 0.4858,
      "step": 167390
    },
    {
      "epoch": 68.52,
      "grad_norm": 1.6683069467544556,
      "learning_rate": 3.6732129195021624e-06,
      "loss": 0.5236,
      "step": 167400
    },
    {
      "epoch": 68.53,
      "grad_norm": 1.6334158182144165,
      "learning_rate": 3.673082276417685e-06,
      "loss": 0.539,
      "step": 167410
    },
    {
      "epoch": 68.53,
      "grad_norm": 1.8189091682434082,
      "learning_rate": 3.6729516272437433e-06,
      "loss": 0.5081,
      "step": 167420
    },
    {
      "epoch": 68.53,
      "grad_norm": 2.000413417816162,
      "learning_rate": 3.6728209719809346e-06,
      "loss": 0.5297,
      "step": 167430
    },
    {
      "epoch": 68.54,
      "grad_norm": 1.6072101593017578,
      "learning_rate": 3.6726903106298573e-06,
      "loss": 0.4909,
      "step": 167440
    },
    {
      "epoch": 68.54,
      "grad_norm": 1.8176138401031494,
      "learning_rate": 3.67255964319111e-06,
      "loss": 0.5126,
      "step": 167450
    },
    {
      "epoch": 68.55,
      "grad_norm": 1.678409218788147,
      "learning_rate": 3.6724289696652924e-06,
      "loss": 0.5196,
      "step": 167460
    },
    {
      "epoch": 68.55,
      "grad_norm": 1.625403642654419,
      "learning_rate": 3.6722982900530013e-06,
      "loss": 0.5237,
      "step": 167470
    },
    {
      "epoch": 68.56,
      "grad_norm": 2.2956597805023193,
      "learning_rate": 3.6721676043548365e-06,
      "loss": 0.5318,
      "step": 167480
    },
    {
      "epoch": 68.56,
      "grad_norm": 2.1421291828155518,
      "learning_rate": 3.6720369125713964e-06,
      "loss": 0.498,
      "step": 167490
    },
    {
      "epoch": 68.56,
      "grad_norm": 1.990388035774231,
      "learning_rate": 3.6719062147032794e-06,
      "loss": 0.5117,
      "step": 167500
    },
    {
      "epoch": 68.57,
      "grad_norm": 2.0620017051696777,
      "learning_rate": 3.6717755107510853e-06,
      "loss": 0.5177,
      "step": 167510
    },
    {
      "epoch": 68.57,
      "grad_norm": 1.8748136758804321,
      "learning_rate": 3.6716448007154117e-06,
      "loss": 0.525,
      "step": 167520
    },
    {
      "epoch": 68.58,
      "grad_norm": 2.1289753913879395,
      "learning_rate": 3.6715140845968573e-06,
      "loss": 0.517,
      "step": 167530
    },
    {
      "epoch": 68.58,
      "grad_norm": 2.0633668899536133,
      "learning_rate": 3.6713833623960215e-06,
      "loss": 0.5226,
      "step": 167540
    },
    {
      "epoch": 68.58,
      "grad_norm": 1.9208135604858398,
      "learning_rate": 3.6712526341135027e-06,
      "loss": 0.5158,
      "step": 167550
    },
    {
      "epoch": 68.59,
      "grad_norm": 2.0983712673187256,
      "learning_rate": 3.6711218997499e-06,
      "loss": 0.5017,
      "step": 167560
    },
    {
      "epoch": 68.59,
      "grad_norm": 1.8083652257919312,
      "learning_rate": 3.6709911593058127e-06,
      "loss": 0.5169,
      "step": 167570
    },
    {
      "epoch": 68.6,
      "grad_norm": 2.1270804405212402,
      "learning_rate": 3.6708604127818384e-06,
      "loss": 0.5096,
      "step": 167580
    },
    {
      "epoch": 68.6,
      "grad_norm": 1.920868992805481,
      "learning_rate": 3.6707296601785773e-06,
      "loss": 0.5191,
      "step": 167590
    },
    {
      "epoch": 68.6,
      "grad_norm": 2.4209656715393066,
      "learning_rate": 3.6705989014966275e-06,
      "loss": 0.5282,
      "step": 167600
    },
    {
      "epoch": 68.61,
      "grad_norm": 1.5356770753860474,
      "learning_rate": 3.670468136736588e-06,
      "loss": 0.4937,
      "step": 167610
    },
    {
      "epoch": 68.61,
      "grad_norm": 1.7651562690734863,
      "learning_rate": 3.6703373658990596e-06,
      "loss": 0.5291,
      "step": 167620
    },
    {
      "epoch": 68.62,
      "grad_norm": 1.8501415252685547,
      "learning_rate": 3.670206588984639e-06,
      "loss": 0.5154,
      "step": 167630
    },
    {
      "epoch": 68.62,
      "grad_norm": 1.9163931608200073,
      "learning_rate": 3.6700758059939263e-06,
      "loss": 0.4959,
      "step": 167640
    },
    {
      "epoch": 68.62,
      "grad_norm": 2.0223186016082764,
      "learning_rate": 3.66994501692752e-06,
      "loss": 0.5191,
      "step": 167650
    },
    {
      "epoch": 68.63,
      "grad_norm": 1.8542628288269043,
      "learning_rate": 3.6698142217860206e-06,
      "loss": 0.5135,
      "step": 167660
    },
    {
      "epoch": 68.63,
      "grad_norm": 2.1695210933685303,
      "learning_rate": 3.6696834205700263e-06,
      "loss": 0.5281,
      "step": 167670
    },
    {
      "epoch": 68.64,
      "grad_norm": 1.7210034132003784,
      "learning_rate": 3.6695526132801357e-06,
      "loss": 0.5186,
      "step": 167680
    },
    {
      "epoch": 68.64,
      "grad_norm": 1.805031657218933,
      "learning_rate": 3.669421799916949e-06,
      "loss": 0.5112,
      "step": 167690
    },
    {
      "epoch": 68.65,
      "grad_norm": 1.725960373878479,
      "learning_rate": 3.6692909804810656e-06,
      "loss": 0.5087,
      "step": 167700
    },
    {
      "epoch": 68.65,
      "grad_norm": 1.6794283390045166,
      "learning_rate": 3.6691601549730834e-06,
      "loss": 0.503,
      "step": 167710
    },
    {
      "epoch": 68.65,
      "grad_norm": 2.301342010498047,
      "learning_rate": 3.6690293233936028e-06,
      "loss": 0.5112,
      "step": 167720
    },
    {
      "epoch": 68.66,
      "grad_norm": 2.1042919158935547,
      "learning_rate": 3.6688984857432228e-06,
      "loss": 0.5407,
      "step": 167730
    },
    {
      "epoch": 68.66,
      "grad_norm": 1.8017059564590454,
      "learning_rate": 3.6687676420225434e-06,
      "loss": 0.5339,
      "step": 167740
    },
    {
      "epoch": 68.67,
      "grad_norm": 1.8548738956451416,
      "learning_rate": 3.6686367922321624e-06,
      "loss": 0.5165,
      "step": 167750
    },
    {
      "epoch": 68.67,
      "grad_norm": 1.6302103996276855,
      "learning_rate": 3.668505936372681e-06,
      "loss": 0.5071,
      "step": 167760
    },
    {
      "epoch": 68.67,
      "grad_norm": 1.9539752006530762,
      "learning_rate": 3.6683750744446973e-06,
      "loss": 0.5046,
      "step": 167770
    },
    {
      "epoch": 68.68,
      "grad_norm": 1.5880697965621948,
      "learning_rate": 3.6682442064488114e-06,
      "loss": 0.5239,
      "step": 167780
    },
    {
      "epoch": 68.68,
      "grad_norm": 2.276564836502075,
      "learning_rate": 3.6681133323856233e-06,
      "loss": 0.5449,
      "step": 167790
    },
    {
      "epoch": 68.69,
      "grad_norm": 1.286165475845337,
      "learning_rate": 3.6679824522557313e-06,
      "loss": 0.5161,
      "step": 167800
    },
    {
      "epoch": 68.69,
      "grad_norm": 2.0185587406158447,
      "learning_rate": 3.6678515660597356e-06,
      "loss": 0.5194,
      "step": 167810
    },
    {
      "epoch": 68.69,
      "grad_norm": 1.6280481815338135,
      "learning_rate": 3.6677206737982358e-06,
      "loss": 0.5299,
      "step": 167820
    },
    {
      "epoch": 68.7,
      "grad_norm": 1.6432547569274902,
      "learning_rate": 3.667589775471831e-06,
      "loss": 0.5024,
      "step": 167830
    },
    {
      "epoch": 68.7,
      "grad_norm": 2.128570556640625,
      "learning_rate": 3.6674588710811216e-06,
      "loss": 0.4978,
      "step": 167840
    },
    {
      "epoch": 68.71,
      "grad_norm": 2.00246000289917,
      "learning_rate": 3.6673279606267064e-06,
      "loss": 0.5219,
      "step": 167850
    },
    {
      "epoch": 68.71,
      "grad_norm": 1.8785628080368042,
      "learning_rate": 3.667197044109186e-06,
      "loss": 0.4923,
      "step": 167860
    },
    {
      "epoch": 68.71,
      "grad_norm": 2.0208892822265625,
      "learning_rate": 3.66706612152916e-06,
      "loss": 0.5161,
      "step": 167870
    },
    {
      "epoch": 68.72,
      "grad_norm": 2.288175344467163,
      "learning_rate": 3.666935192887227e-06,
      "loss": 0.5214,
      "step": 167880
    },
    {
      "epoch": 68.72,
      "grad_norm": 1.55963134765625,
      "learning_rate": 3.6668042581839884e-06,
      "loss": 0.5308,
      "step": 167890
    },
    {
      "epoch": 68.73,
      "grad_norm": 1.4599690437316895,
      "learning_rate": 3.666673317420042e-06,
      "loss": 0.5048,
      "step": 167900
    },
    {
      "epoch": 68.73,
      "grad_norm": 2.1762452125549316,
      "learning_rate": 3.6665423705959895e-06,
      "loss": 0.5218,
      "step": 167910
    },
    {
      "epoch": 68.74,
      "grad_norm": 2.1189310550689697,
      "learning_rate": 3.6664114177124296e-06,
      "loss": 0.5284,
      "step": 167920
    },
    {
      "epoch": 68.74,
      "grad_norm": 2.3718225955963135,
      "learning_rate": 3.666280458769963e-06,
      "loss": 0.5072,
      "step": 167930
    },
    {
      "epoch": 68.74,
      "grad_norm": 1.8563077449798584,
      "learning_rate": 3.6661494937691887e-06,
      "loss": 0.5077,
      "step": 167940
    },
    {
      "epoch": 68.75,
      "grad_norm": 2.550926685333252,
      "learning_rate": 3.6660185227107083e-06,
      "loss": 0.5344,
      "step": 167950
    },
    {
      "epoch": 68.75,
      "grad_norm": 1.793186902999878,
      "learning_rate": 3.6658875455951194e-06,
      "loss": 0.5,
      "step": 167960
    },
    {
      "epoch": 68.76,
      "grad_norm": 1.8772286176681519,
      "learning_rate": 3.6657565624230234e-06,
      "loss": 0.5125,
      "step": 167970
    },
    {
      "epoch": 68.76,
      "grad_norm": 2.2231407165527344,
      "learning_rate": 3.6656255731950205e-06,
      "loss": 0.5207,
      "step": 167980
    },
    {
      "epoch": 68.76,
      "grad_norm": 2.040614604949951,
      "learning_rate": 3.6654945779117103e-06,
      "loss": 0.5302,
      "step": 167990
    },
    {
      "epoch": 68.77,
      "grad_norm": 2.3555943965911865,
      "learning_rate": 3.665363576573692e-06,
      "loss": 0.5322,
      "step": 168000
    },
    {
      "epoch": 68.77,
      "grad_norm": 1.815832257270813,
      "learning_rate": 3.665232569181567e-06,
      "loss": 0.5195,
      "step": 168010
    },
    {
      "epoch": 68.78,
      "grad_norm": 1.5627553462982178,
      "learning_rate": 3.6651015557359358e-06,
      "loss": 0.5084,
      "step": 168020
    },
    {
      "epoch": 68.78,
      "grad_norm": 3.5158886909484863,
      "learning_rate": 3.6649705362373974e-06,
      "loss": 0.5154,
      "step": 168030
    },
    {
      "epoch": 68.78,
      "grad_norm": 2.0117905139923096,
      "learning_rate": 3.6648395106865516e-06,
      "loss": 0.5072,
      "step": 168040
    },
    {
      "epoch": 68.79,
      "grad_norm": 1.949467658996582,
      "learning_rate": 3.664708479084e-06,
      "loss": 0.5155,
      "step": 168050
    },
    {
      "epoch": 68.79,
      "grad_norm": 1.5207014083862305,
      "learning_rate": 3.664577441430342e-06,
      "loss": 0.5182,
      "step": 168060
    },
    {
      "epoch": 68.8,
      "grad_norm": 1.8126776218414307,
      "learning_rate": 3.664446397726178e-06,
      "loss": 0.5093,
      "step": 168070
    },
    {
      "epoch": 68.8,
      "grad_norm": 2.295260429382324,
      "learning_rate": 3.664315347972109e-06,
      "loss": 0.4914,
      "step": 168080
    },
    {
      "epoch": 68.8,
      "grad_norm": 1.5757449865341187,
      "learning_rate": 3.6641842921687334e-06,
      "loss": 0.5079,
      "step": 168090
    },
    {
      "epoch": 68.81,
      "grad_norm": 1.7308592796325684,
      "learning_rate": 3.6640532303166537e-06,
      "loss": 0.5275,
      "step": 168100
    },
    {
      "epoch": 68.81,
      "grad_norm": 1.5149073600769043,
      "learning_rate": 3.663922162416469e-06,
      "loss": 0.5283,
      "step": 168110
    },
    {
      "epoch": 68.82,
      "grad_norm": 1.7547905445098877,
      "learning_rate": 3.66379108846878e-06,
      "loss": 0.5189,
      "step": 168120
    },
    {
      "epoch": 68.82,
      "grad_norm": 2.03395676612854,
      "learning_rate": 3.6636600084741876e-06,
      "loss": 0.5061,
      "step": 168130
    },
    {
      "epoch": 68.83,
      "grad_norm": 2.081010341644287,
      "learning_rate": 3.6635289224332915e-06,
      "loss": 0.5193,
      "step": 168140
    },
    {
      "epoch": 68.83,
      "grad_norm": 1.8105391263961792,
      "learning_rate": 3.663397830346693e-06,
      "loss": 0.5107,
      "step": 168150
    },
    {
      "epoch": 68.83,
      "grad_norm": 1.8764100074768066,
      "learning_rate": 3.663266732214992e-06,
      "loss": 0.5163,
      "step": 168160
    },
    {
      "epoch": 68.84,
      "grad_norm": 2.0073041915893555,
      "learning_rate": 3.663135628038789e-06,
      "loss": 0.5151,
      "step": 168170
    },
    {
      "epoch": 68.84,
      "grad_norm": 1.7250288724899292,
      "learning_rate": 3.663004517818685e-06,
      "loss": 0.5099,
      "step": 168180
    },
    {
      "epoch": 68.85,
      "grad_norm": 1.8717375993728638,
      "learning_rate": 3.66287340155528e-06,
      "loss": 0.5218,
      "step": 168190
    },
    {
      "epoch": 68.85,
      "grad_norm": 1.9334479570388794,
      "learning_rate": 3.6627422792491758e-06,
      "loss": 0.5287,
      "step": 168200
    },
    {
      "epoch": 68.85,
      "grad_norm": 2.5614635944366455,
      "learning_rate": 3.662611150900971e-06,
      "loss": 0.5135,
      "step": 168210
    },
    {
      "epoch": 68.86,
      "grad_norm": 1.5387217998504639,
      "learning_rate": 3.6624800165112687e-06,
      "loss": 0.517,
      "step": 168220
    },
    {
      "epoch": 68.86,
      "grad_norm": 2.028118371963501,
      "learning_rate": 3.6623488760806675e-06,
      "loss": 0.5318,
      "step": 168230
    },
    {
      "epoch": 68.87,
      "grad_norm": 2.020049810409546,
      "learning_rate": 3.66221772960977e-06,
      "loss": 0.5093,
      "step": 168240
    },
    {
      "epoch": 68.87,
      "grad_norm": 1.4976662397384644,
      "learning_rate": 3.6620865770991752e-06,
      "loss": 0.5229,
      "step": 168250
    },
    {
      "epoch": 68.87,
      "grad_norm": 2.484949827194214,
      "learning_rate": 3.6619554185494856e-06,
      "loss": 0.5061,
      "step": 168260
    },
    {
      "epoch": 68.88,
      "grad_norm": 1.8620312213897705,
      "learning_rate": 3.6618242539613e-06,
      "loss": 0.5231,
      "step": 168270
    },
    {
      "epoch": 68.88,
      "grad_norm": 1.2630243301391602,
      "learning_rate": 3.6616930833352214e-06,
      "loss": 0.5062,
      "step": 168280
    },
    {
      "epoch": 68.89,
      "grad_norm": 2.4657552242279053,
      "learning_rate": 3.661561906671849e-06,
      "loss": 0.5234,
      "step": 168290
    },
    {
      "epoch": 68.89,
      "grad_norm": 2.258730411529541,
      "learning_rate": 3.661430723971784e-06,
      "loss": 0.503,
      "step": 168300
    },
    {
      "epoch": 68.89,
      "grad_norm": 2.190545082092285,
      "learning_rate": 3.661299535235628e-06,
      "loss": 0.52,
      "step": 168310
    },
    {
      "epoch": 68.9,
      "grad_norm": 2.3466272354125977,
      "learning_rate": 3.6611683404639823e-06,
      "loss": 0.5234,
      "step": 168320
    },
    {
      "epoch": 68.9,
      "grad_norm": 2.375977039337158,
      "learning_rate": 3.6610371396574466e-06,
      "loss": 0.512,
      "step": 168330
    },
    {
      "epoch": 68.91,
      "grad_norm": 1.64376962184906,
      "learning_rate": 3.6609059328166227e-06,
      "loss": 0.5367,
      "step": 168340
    },
    {
      "epoch": 68.91,
      "grad_norm": 1.8666095733642578,
      "learning_rate": 3.660774719942111e-06,
      "loss": 0.5173,
      "step": 168350
    },
    {
      "epoch": 68.92,
      "grad_norm": 1.6309049129486084,
      "learning_rate": 3.660643501034514e-06,
      "loss": 0.5244,
      "step": 168360
    },
    {
      "epoch": 68.92,
      "grad_norm": 1.898639440536499,
      "learning_rate": 3.660512276094431e-06,
      "loss": 0.5142,
      "step": 168370
    },
    {
      "epoch": 68.92,
      "grad_norm": 2.6228888034820557,
      "learning_rate": 3.6603810451224643e-06,
      "loss": 0.5177,
      "step": 168380
    },
    {
      "epoch": 68.93,
      "grad_norm": 1.6614938974380493,
      "learning_rate": 3.6602498081192143e-06,
      "loss": 0.5138,
      "step": 168390
    },
    {
      "epoch": 68.93,
      "grad_norm": 1.5886632204055786,
      "learning_rate": 3.6601185650852825e-06,
      "loss": 0.493,
      "step": 168400
    },
    {
      "epoch": 68.94,
      "grad_norm": 2.014446496963501,
      "learning_rate": 3.6599873160212705e-06,
      "loss": 0.5211,
      "step": 168410
    },
    {
      "epoch": 68.94,
      "grad_norm": 1.7651704549789429,
      "learning_rate": 3.65985606092778e-06,
      "loss": 0.5047,
      "step": 168420
    },
    {
      "epoch": 68.94,
      "grad_norm": 1.9140803813934326,
      "learning_rate": 3.6597247998054103e-06,
      "loss": 0.5217,
      "step": 168430
    },
    {
      "epoch": 68.95,
      "grad_norm": 2.266035795211792,
      "learning_rate": 3.6595935326547644e-06,
      "loss": 0.499,
      "step": 168440
    },
    {
      "epoch": 68.95,
      "grad_norm": 1.7552826404571533,
      "learning_rate": 3.6594622594764428e-06,
      "loss": 0.5132,
      "step": 168450
    },
    {
      "epoch": 68.96,
      "grad_norm": 2.422133445739746,
      "learning_rate": 3.6593309802710467e-06,
      "loss": 0.5349,
      "step": 168460
    },
    {
      "epoch": 68.96,
      "grad_norm": 2.6095638275146484,
      "learning_rate": 3.659199695039179e-06,
      "loss": 0.5422,
      "step": 168470
    },
    {
      "epoch": 68.96,
      "grad_norm": 1.8573352098464966,
      "learning_rate": 3.6590684037814392e-06,
      "loss": 0.5228,
      "step": 168480
    },
    {
      "epoch": 68.97,
      "grad_norm": 2.2716779708862305,
      "learning_rate": 3.6589371064984293e-06,
      "loss": 0.5318,
      "step": 168490
    },
    {
      "epoch": 68.97,
      "grad_norm": 2.80733323097229,
      "learning_rate": 3.658805803190751e-06,
      "loss": 0.5146,
      "step": 168500
    },
    {
      "epoch": 68.98,
      "grad_norm": 1.4937752485275269,
      "learning_rate": 3.658674493859006e-06,
      "loss": 0.5089,
      "step": 168510
    },
    {
      "epoch": 68.98,
      "grad_norm": 2.233340263366699,
      "learning_rate": 3.6585431785037956e-06,
      "loss": 0.5183,
      "step": 168520
    },
    {
      "epoch": 68.98,
      "grad_norm": 1.5435309410095215,
      "learning_rate": 3.6584118571257214e-06,
      "loss": 0.5241,
      "step": 168530
    },
    {
      "epoch": 68.99,
      "grad_norm": 1.8286635875701904,
      "learning_rate": 3.658280529725385e-06,
      "loss": 0.5062,
      "step": 168540
    },
    {
      "epoch": 68.99,
      "grad_norm": 1.7409433126449585,
      "learning_rate": 3.6581491963033876e-06,
      "loss": 0.5088,
      "step": 168550
    },
    {
      "epoch": 69.0,
      "grad_norm": 1.916264295578003,
      "learning_rate": 3.658017856860331e-06,
      "loss": 0.5192,
      "step": 168560
    },
    {
      "epoch": 69.0,
      "eval_loss": 0.5201067328453064,
      "eval_runtime": 63.0012,
      "eval_samples_per_second": 54.745,
      "eval_steps_per_second": 6.857,
      "step": 168567
    },
    {
      "epoch": 69.0,
      "grad_norm": 1.803868055343628,
      "learning_rate": 3.657886511396816e-06,
      "loss": 0.5094,
      "step": 168570
    },
    {
      "epoch": 69.01,
      "grad_norm": 2.0029263496398926,
      "learning_rate": 3.657755159913446e-06,
      "loss": 0.5279,
      "step": 168580
    },
    {
      "epoch": 69.01,
      "grad_norm": 1.973289966583252,
      "learning_rate": 3.6576238024108217e-06,
      "loss": 0.52,
      "step": 168590
    },
    {
      "epoch": 69.01,
      "grad_norm": 1.8785669803619385,
      "learning_rate": 3.657492438889545e-06,
      "loss": 0.5091,
      "step": 168600
    },
    {
      "epoch": 69.02,
      "grad_norm": 1.937422513961792,
      "learning_rate": 3.657361069350218e-06,
      "loss": 0.5244,
      "step": 168610
    },
    {
      "epoch": 69.02,
      "grad_norm": 2.421718120574951,
      "learning_rate": 3.6572296937934424e-06,
      "loss": 0.4997,
      "step": 168620
    },
    {
      "epoch": 69.03,
      "grad_norm": 1.9990407228469849,
      "learning_rate": 3.6570983122198185e-06,
      "loss": 0.5141,
      "step": 168630
    },
    {
      "epoch": 69.03,
      "grad_norm": 1.6310193538665771,
      "learning_rate": 3.6569669246299504e-06,
      "loss": 0.5203,
      "step": 168640
    },
    {
      "epoch": 69.03,
      "grad_norm": 2.0916507244110107,
      "learning_rate": 3.656835531024439e-06,
      "loss": 0.5128,
      "step": 168650
    },
    {
      "epoch": 69.04,
      "grad_norm": 2.2201786041259766,
      "learning_rate": 3.6567041314038856e-06,
      "loss": 0.5031,
      "step": 168660
    },
    {
      "epoch": 69.04,
      "grad_norm": 1.9520000219345093,
      "learning_rate": 3.6565727257688924e-06,
      "loss": 0.5018,
      "step": 168670
    },
    {
      "epoch": 69.05,
      "grad_norm": 2.1352195739746094,
      "learning_rate": 3.656441314120062e-06,
      "loss": 0.5343,
      "step": 168680
    },
    {
      "epoch": 69.05,
      "grad_norm": 1.7707923650741577,
      "learning_rate": 3.656309896457996e-06,
      "loss": 0.5053,
      "step": 168690
    },
    {
      "epoch": 69.05,
      "grad_norm": 1.8526016473770142,
      "learning_rate": 3.6561784727832968e-06,
      "loss": 0.4974,
      "step": 168700
    },
    {
      "epoch": 69.06,
      "grad_norm": 1.6041837930679321,
      "learning_rate": 3.6560470430965655e-06,
      "loss": 0.495,
      "step": 168710
    },
    {
      "epoch": 69.06,
      "grad_norm": 1.8082586526870728,
      "learning_rate": 3.655915607398405e-06,
      "loss": 0.5069,
      "step": 168720
    },
    {
      "epoch": 69.07,
      "grad_norm": 1.6035884618759155,
      "learning_rate": 3.655784165689417e-06,
      "loss": 0.5194,
      "step": 168730
    },
    {
      "epoch": 69.07,
      "grad_norm": 1.8118507862091064,
      "learning_rate": 3.6556527179702033e-06,
      "loss": 0.5071,
      "step": 168740
    },
    {
      "epoch": 69.07,
      "grad_norm": 1.7526663541793823,
      "learning_rate": 3.6555212642413667e-06,
      "loss": 0.5185,
      "step": 168750
    },
    {
      "epoch": 69.08,
      "grad_norm": 1.5816203355789185,
      "learning_rate": 3.655389804503509e-06,
      "loss": 0.5138,
      "step": 168760
    },
    {
      "epoch": 69.08,
      "grad_norm": 1.7546021938323975,
      "learning_rate": 3.655258338757232e-06,
      "loss": 0.5078,
      "step": 168770
    },
    {
      "epoch": 69.09,
      "grad_norm": 2.451540946960449,
      "learning_rate": 3.6551268670031392e-06,
      "loss": 0.5015,
      "step": 168780
    },
    {
      "epoch": 69.09,
      "grad_norm": 2.0312981605529785,
      "learning_rate": 3.654995389241832e-06,
      "loss": 0.512,
      "step": 168790
    },
    {
      "epoch": 69.1,
      "grad_norm": 2.0013203620910645,
      "learning_rate": 3.6548639054739126e-06,
      "loss": 0.494,
      "step": 168800
    },
    {
      "epoch": 69.1,
      "grad_norm": 1.641224980354309,
      "learning_rate": 3.6547324156999832e-06,
      "loss": 0.5171,
      "step": 168810
    },
    {
      "epoch": 69.1,
      "grad_norm": 1.5629029273986816,
      "learning_rate": 3.654600919920647e-06,
      "loss": 0.5245,
      "step": 168820
    },
    {
      "epoch": 69.11,
      "grad_norm": 2.1369788646698,
      "learning_rate": 3.6544694181365054e-06,
      "loss": 0.5124,
      "step": 168830
    },
    {
      "epoch": 69.11,
      "grad_norm": 1.9950865507125854,
      "learning_rate": 3.654337910348161e-06,
      "loss": 0.513,
      "step": 168840
    },
    {
      "epoch": 69.12,
      "grad_norm": 1.8568675518035889,
      "learning_rate": 3.6542063965562165e-06,
      "loss": 0.5086,
      "step": 168850
    },
    {
      "epoch": 69.12,
      "grad_norm": 2.0640761852264404,
      "learning_rate": 3.654074876761274e-06,
      "loss": 0.5407,
      "step": 168860
    },
    {
      "epoch": 69.12,
      "grad_norm": 1.9337785243988037,
      "learning_rate": 3.6539433509639364e-06,
      "loss": 0.516,
      "step": 168870
    },
    {
      "epoch": 69.13,
      "grad_norm": 1.4431846141815186,
      "learning_rate": 3.653811819164806e-06,
      "loss": 0.5216,
      "step": 168880
    },
    {
      "epoch": 69.13,
      "grad_norm": 2.2419826984405518,
      "learning_rate": 3.6536802813644853e-06,
      "loss": 0.5076,
      "step": 168890
    },
    {
      "epoch": 69.14,
      "grad_norm": 2.2807323932647705,
      "learning_rate": 3.6535487375635767e-06,
      "loss": 0.5242,
      "step": 168900
    },
    {
      "epoch": 69.14,
      "grad_norm": 1.5232183933258057,
      "learning_rate": 3.653417187762683e-06,
      "loss": 0.538,
      "step": 168910
    },
    {
      "epoch": 69.14,
      "grad_norm": 1.5711005926132202,
      "learning_rate": 3.6532856319624068e-06,
      "loss": 0.5137,
      "step": 168920
    },
    {
      "epoch": 69.15,
      "grad_norm": 2.0027542114257812,
      "learning_rate": 3.653154070163351e-06,
      "loss": 0.5196,
      "step": 168930
    },
    {
      "epoch": 69.15,
      "grad_norm": 1.4712902307510376,
      "learning_rate": 3.653022502366117e-06,
      "loss": 0.5247,
      "step": 168940
    },
    {
      "epoch": 69.16,
      "grad_norm": 1.8894505500793457,
      "learning_rate": 3.6528909285713092e-06,
      "loss": 0.5214,
      "step": 168950
    },
    {
      "epoch": 69.16,
      "grad_norm": 1.7518973350524902,
      "learning_rate": 3.6527593487795286e-06,
      "loss": 0.5325,
      "step": 168960
    },
    {
      "epoch": 69.16,
      "grad_norm": 2.2554855346679688,
      "learning_rate": 3.65262776299138e-06,
      "loss": 0.524,
      "step": 168970
    },
    {
      "epoch": 69.17,
      "grad_norm": 1.8030320405960083,
      "learning_rate": 3.6524961712074646e-06,
      "loss": 0.5274,
      "step": 168980
    },
    {
      "epoch": 69.17,
      "grad_norm": 1.796789288520813,
      "learning_rate": 3.652364573428386e-06,
      "loss": 0.4934,
      "step": 168990
    },
    {
      "epoch": 69.18,
      "grad_norm": 1.641850233078003,
      "learning_rate": 3.6522329696547463e-06,
      "loss": 0.5153,
      "step": 169000
    },
    {
      "epoch": 69.18,
      "grad_norm": 1.50547456741333,
      "learning_rate": 3.6521013598871494e-06,
      "loss": 0.5248,
      "step": 169010
    },
    {
      "epoch": 69.19,
      "grad_norm": 1.7206064462661743,
      "learning_rate": 3.6519697441261968e-06,
      "loss": 0.5264,
      "step": 169020
    },
    {
      "epoch": 69.19,
      "grad_norm": 1.8421592712402344,
      "learning_rate": 3.6518381223724924e-06,
      "loss": 0.5114,
      "step": 169030
    },
    {
      "epoch": 69.19,
      "grad_norm": 2.1957449913024902,
      "learning_rate": 3.651706494626639e-06,
      "loss": 0.5165,
      "step": 169040
    },
    {
      "epoch": 69.2,
      "grad_norm": 1.7037229537963867,
      "learning_rate": 3.651574860889239e-06,
      "loss": 0.507,
      "step": 169050
    },
    {
      "epoch": 69.2,
      "grad_norm": 2.111154556274414,
      "learning_rate": 3.651443221160896e-06,
      "loss": 0.5112,
      "step": 169060
    },
    {
      "epoch": 69.21,
      "grad_norm": 1.9972413778305054,
      "learning_rate": 3.6513115754422133e-06,
      "loss": 0.5413,
      "step": 169070
    },
    {
      "epoch": 69.21,
      "grad_norm": 1.5889630317687988,
      "learning_rate": 3.651179923733794e-06,
      "loss": 0.5347,
      "step": 169080
    },
    {
      "epoch": 69.21,
      "grad_norm": 2.764378786087036,
      "learning_rate": 3.6510482660362393e-06,
      "loss": 0.5206,
      "step": 169090
    },
    {
      "epoch": 69.22,
      "grad_norm": 2.009706735610962,
      "learning_rate": 3.6509166023501545e-06,
      "loss": 0.4932,
      "step": 169100
    },
    {
      "epoch": 69.22,
      "grad_norm": 1.4572490453720093,
      "learning_rate": 3.650784932676142e-06,
      "loss": 0.5145,
      "step": 169110
    },
    {
      "epoch": 69.23,
      "grad_norm": 2.1287500858306885,
      "learning_rate": 3.650653257014805e-06,
      "loss": 0.4963,
      "step": 169120
    },
    {
      "epoch": 69.23,
      "grad_norm": 1.6912565231323242,
      "learning_rate": 3.6505215753667463e-06,
      "loss": 0.5166,
      "step": 169130
    },
    {
      "epoch": 69.23,
      "grad_norm": 2.4100289344787598,
      "learning_rate": 3.6503898877325697e-06,
      "loss": 0.5405,
      "step": 169140
    },
    {
      "epoch": 69.24,
      "grad_norm": 1.9643322229385376,
      "learning_rate": 3.6502581941128775e-06,
      "loss": 0.5042,
      "step": 169150
    },
    {
      "epoch": 69.24,
      "grad_norm": 1.9839701652526855,
      "learning_rate": 3.6501264945082737e-06,
      "loss": 0.5168,
      "step": 169160
    },
    {
      "epoch": 69.25,
      "grad_norm": 1.6282308101654053,
      "learning_rate": 3.6499947889193614e-06,
      "loss": 0.5216,
      "step": 169170
    },
    {
      "epoch": 69.25,
      "grad_norm": 1.980109691619873,
      "learning_rate": 3.649863077346745e-06,
      "loss": 0.5136,
      "step": 169180
    },
    {
      "epoch": 69.26,
      "grad_norm": 2.3207249641418457,
      "learning_rate": 3.6497313597910266e-06,
      "loss": 0.5356,
      "step": 169190
    },
    {
      "epoch": 69.26,
      "grad_norm": 1.886428713798523,
      "learning_rate": 3.649599636252809e-06,
      "loss": 0.5241,
      "step": 169200
    },
    {
      "epoch": 69.26,
      "grad_norm": 1.8183485269546509,
      "learning_rate": 3.6494679067326973e-06,
      "loss": 0.506,
      "step": 169210
    },
    {
      "epoch": 69.27,
      "grad_norm": 1.8198177814483643,
      "learning_rate": 3.6493361712312933e-06,
      "loss": 0.5306,
      "step": 169220
    },
    {
      "epoch": 69.27,
      "grad_norm": 2.0681145191192627,
      "learning_rate": 3.649204429749202e-06,
      "loss": 0.5154,
      "step": 169230
    },
    {
      "epoch": 69.28,
      "grad_norm": 2.4769184589385986,
      "learning_rate": 3.6490726822870257e-06,
      "loss": 0.5034,
      "step": 169240
    },
    {
      "epoch": 69.28,
      "grad_norm": 1.9176905155181885,
      "learning_rate": 3.6489409288453683e-06,
      "loss": 0.5091,
      "step": 169250
    },
    {
      "epoch": 69.28,
      "grad_norm": 1.8490629196166992,
      "learning_rate": 3.648809169424834e-06,
      "loss": 0.5062,
      "step": 169260
    },
    {
      "epoch": 69.29,
      "grad_norm": 2.1972620487213135,
      "learning_rate": 3.6486774040260248e-06,
      "loss": 0.5182,
      "step": 169270
    },
    {
      "epoch": 69.29,
      "grad_norm": 2.042834758758545,
      "learning_rate": 3.648545632649546e-06,
      "loss": 0.5329,
      "step": 169280
    },
    {
      "epoch": 69.3,
      "grad_norm": 2.102982997894287,
      "learning_rate": 3.6484138552960004e-06,
      "loss": 0.5159,
      "step": 169290
    },
    {
      "epoch": 69.3,
      "grad_norm": 2.4345145225524902,
      "learning_rate": 3.6482820719659917e-06,
      "loss": 0.4935,
      "step": 169300
    },
    {
      "epoch": 69.3,
      "grad_norm": 2.215097427368164,
      "learning_rate": 3.6481502826601236e-06,
      "loss": 0.5212,
      "step": 169310
    },
    {
      "epoch": 69.31,
      "grad_norm": 1.8764452934265137,
      "learning_rate": 3.6480184873789993e-06,
      "loss": 0.5228,
      "step": 169320
    },
    {
      "epoch": 69.31,
      "grad_norm": 1.7015701532363892,
      "learning_rate": 3.6478866861232234e-06,
      "loss": 0.5271,
      "step": 169330
    },
    {
      "epoch": 69.32,
      "grad_norm": 1.4087823629379272,
      "learning_rate": 3.647754878893399e-06,
      "loss": 0.5026,
      "step": 169340
    },
    {
      "epoch": 69.32,
      "grad_norm": 2.1659340858459473,
      "learning_rate": 3.647623065690131e-06,
      "loss": 0.5197,
      "step": 169350
    },
    {
      "epoch": 69.32,
      "grad_norm": 1.894351601600647,
      "learning_rate": 3.647491246514022e-06,
      "loss": 0.5033,
      "step": 169360
    },
    {
      "epoch": 69.33,
      "grad_norm": 1.7563953399658203,
      "learning_rate": 3.647359421365676e-06,
      "loss": 0.5104,
      "step": 169370
    },
    {
      "epoch": 69.33,
      "grad_norm": 3.1505770683288574,
      "learning_rate": 3.647227590245698e-06,
      "loss": 0.5157,
      "step": 169380
    },
    {
      "epoch": 69.34,
      "grad_norm": 2.025341033935547,
      "learning_rate": 3.64709575315469e-06,
      "loss": 0.5063,
      "step": 169390
    },
    {
      "epoch": 69.34,
      "grad_norm": 1.973093867301941,
      "learning_rate": 3.646963910093257e-06,
      "loss": 0.5195,
      "step": 169400
    },
    {
      "epoch": 69.35,
      "grad_norm": 2.226799249649048,
      "learning_rate": 3.6468320610620038e-06,
      "loss": 0.5382,
      "step": 169410
    },
    {
      "epoch": 69.35,
      "grad_norm": 1.6354612112045288,
      "learning_rate": 3.6467002060615326e-06,
      "loss": 0.5247,
      "step": 169420
    },
    {
      "epoch": 69.35,
      "grad_norm": 2.0408310890197754,
      "learning_rate": 3.6465683450924484e-06,
      "loss": 0.5267,
      "step": 169430
    },
    {
      "epoch": 69.36,
      "grad_norm": 1.5483555793762207,
      "learning_rate": 3.646436478155356e-06,
      "loss": 0.5022,
      "step": 169440
    },
    {
      "epoch": 69.36,
      "grad_norm": 1.9982478618621826,
      "learning_rate": 3.646304605250858e-06,
      "loss": 0.5307,
      "step": 169450
    },
    {
      "epoch": 69.37,
      "grad_norm": 2.0530803203582764,
      "learning_rate": 3.6461727263795587e-06,
      "loss": 0.5171,
      "step": 169460
    },
    {
      "epoch": 69.37,
      "grad_norm": 1.6273795366287231,
      "learning_rate": 3.646040841542063e-06,
      "loss": 0.5189,
      "step": 169470
    },
    {
      "epoch": 69.37,
      "grad_norm": 2.1086037158966064,
      "learning_rate": 3.645908950738975e-06,
      "loss": 0.5141,
      "step": 169480
    },
    {
      "epoch": 69.38,
      "grad_norm": 1.8693017959594727,
      "learning_rate": 3.6457770539708975e-06,
      "loss": 0.5316,
      "step": 169490
    },
    {
      "epoch": 69.38,
      "grad_norm": 2.064138889312744,
      "learning_rate": 3.6456451512384367e-06,
      "loss": 0.482,
      "step": 169500
    },
    {
      "epoch": 69.39,
      "grad_norm": 2.229382038116455,
      "learning_rate": 3.645513242542195e-06,
      "loss": 0.5182,
      "step": 169510
    },
    {
      "epoch": 69.39,
      "grad_norm": 1.8454827070236206,
      "learning_rate": 3.645381327882778e-06,
      "loss": 0.5149,
      "step": 169520
    },
    {
      "epoch": 69.39,
      "grad_norm": 2.0350029468536377,
      "learning_rate": 3.6452494072607896e-06,
      "loss": 0.4924,
      "step": 169530
    },
    {
      "epoch": 69.4,
      "grad_norm": 1.7253150939941406,
      "learning_rate": 3.6451174806768335e-06,
      "loss": 0.512,
      "step": 169540
    },
    {
      "epoch": 69.4,
      "grad_norm": 1.867302417755127,
      "learning_rate": 3.644985548131515e-06,
      "loss": 0.5377,
      "step": 169550
    },
    {
      "epoch": 69.41,
      "grad_norm": 1.8056247234344482,
      "learning_rate": 3.644853609625438e-06,
      "loss": 0.5161,
      "step": 169560
    },
    {
      "epoch": 69.41,
      "grad_norm": 1.9933289289474487,
      "learning_rate": 3.6447216651592063e-06,
      "loss": 0.5159,
      "step": 169570
    },
    {
      "epoch": 69.41,
      "grad_norm": 1.8984575271606445,
      "learning_rate": 3.6445897147334255e-06,
      "loss": 0.5311,
      "step": 169580
    },
    {
      "epoch": 69.42,
      "grad_norm": 2.0109996795654297,
      "learning_rate": 3.644457758348699e-06,
      "loss": 0.5174,
      "step": 169590
    },
    {
      "epoch": 69.42,
      "grad_norm": 1.584787130355835,
      "learning_rate": 3.6443257960056315e-06,
      "loss": 0.5002,
      "step": 169600
    },
    {
      "epoch": 69.43,
      "grad_norm": 2.29555082321167,
      "learning_rate": 3.6441938277048288e-06,
      "loss": 0.5061,
      "step": 169610
    },
    {
      "epoch": 69.43,
      "grad_norm": 2.208853244781494,
      "learning_rate": 3.6440618534468933e-06,
      "loss": 0.5282,
      "step": 169620
    },
    {
      "epoch": 69.44,
      "grad_norm": 1.4359393119812012,
      "learning_rate": 3.6439298732324307e-06,
      "loss": 0.5147,
      "step": 169630
    },
    {
      "epoch": 69.44,
      "grad_norm": 2.321845054626465,
      "learning_rate": 3.643797887062045e-06,
      "loss": 0.5243,
      "step": 169640
    },
    {
      "epoch": 69.44,
      "grad_norm": 2.1503748893737793,
      "learning_rate": 3.6436658949363425e-06,
      "loss": 0.52,
      "step": 169650
    },
    {
      "epoch": 69.45,
      "grad_norm": 1.824794888496399,
      "learning_rate": 3.643533896855926e-06,
      "loss": 0.5171,
      "step": 169660
    },
    {
      "epoch": 69.45,
      "grad_norm": 1.5946425199508667,
      "learning_rate": 3.6434018928214007e-06,
      "loss": 0.5244,
      "step": 169670
    },
    {
      "epoch": 69.46,
      "grad_norm": 2.039926052093506,
      "learning_rate": 3.6432698828333713e-06,
      "loss": 0.5201,
      "step": 169680
    },
    {
      "epoch": 69.46,
      "grad_norm": 1.8214614391326904,
      "learning_rate": 3.643137866892442e-06,
      "loss": 0.5194,
      "step": 169690
    },
    {
      "epoch": 69.46,
      "grad_norm": 1.847261667251587,
      "learning_rate": 3.6430058449992195e-06,
      "loss": 0.5154,
      "step": 169700
    },
    {
      "epoch": 69.47,
      "grad_norm": 1.564008116722107,
      "learning_rate": 3.642873817154306e-06,
      "loss": 0.5259,
      "step": 169710
    },
    {
      "epoch": 69.47,
      "grad_norm": 1.871103286743164,
      "learning_rate": 3.642741783358308e-06,
      "loss": 0.516,
      "step": 169720
    },
    {
      "epoch": 69.48,
      "grad_norm": 1.9221936464309692,
      "learning_rate": 3.64260974361183e-06,
      "loss": 0.5329,
      "step": 169730
    },
    {
      "epoch": 69.48,
      "grad_norm": 1.9723957777023315,
      "learning_rate": 3.6424776979154763e-06,
      "loss": 0.5424,
      "step": 169740
    },
    {
      "epoch": 69.48,
      "grad_norm": 1.6871243715286255,
      "learning_rate": 3.6423456462698525e-06,
      "loss": 0.5111,
      "step": 169750
    },
    {
      "epoch": 69.49,
      "grad_norm": 2.5368592739105225,
      "learning_rate": 3.6422135886755628e-06,
      "loss": 0.5238,
      "step": 169760
    },
    {
      "epoch": 69.49,
      "grad_norm": 1.856323003768921,
      "learning_rate": 3.642081525133213e-06,
      "loss": 0.5292,
      "step": 169770
    },
    {
      "epoch": 69.5,
      "grad_norm": 1.7560290098190308,
      "learning_rate": 3.6419494556434072e-06,
      "loss": 0.5038,
      "step": 169780
    },
    {
      "epoch": 69.5,
      "grad_norm": 3.3413431644439697,
      "learning_rate": 3.6418173802067505e-06,
      "loss": 0.5072,
      "step": 169790
    },
    {
      "epoch": 69.5,
      "grad_norm": 2.3894898891448975,
      "learning_rate": 3.6416852988238484e-06,
      "loss": 0.5231,
      "step": 169800
    },
    {
      "epoch": 69.51,
      "grad_norm": 1.9756507873535156,
      "learning_rate": 3.6415532114953057e-06,
      "loss": 0.5262,
      "step": 169810
    },
    {
      "epoch": 69.51,
      "grad_norm": 1.9382472038269043,
      "learning_rate": 3.641421118221727e-06,
      "loss": 0.5179,
      "step": 169820
    },
    {
      "epoch": 69.52,
      "grad_norm": 1.718388557434082,
      "learning_rate": 3.6412890190037187e-06,
      "loss": 0.5379,
      "step": 169830
    },
    {
      "epoch": 69.52,
      "grad_norm": 1.5756971836090088,
      "learning_rate": 3.641156913841885e-06,
      "loss": 0.5337,
      "step": 169840
    },
    {
      "epoch": 69.53,
      "grad_norm": 2.4874043464660645,
      "learning_rate": 3.641024802736831e-06,
      "loss": 0.5096,
      "step": 169850
    },
    {
      "epoch": 69.53,
      "grad_norm": 2.299428939819336,
      "learning_rate": 3.6408926856891615e-06,
      "loss": 0.5133,
      "step": 169860
    },
    {
      "epoch": 69.53,
      "grad_norm": 1.6924515962600708,
      "learning_rate": 3.640760562699483e-06,
      "loss": 0.5189,
      "step": 169870
    },
    {
      "epoch": 69.54,
      "grad_norm": 1.9548441171646118,
      "learning_rate": 3.6406284337684e-06,
      "loss": 0.5169,
      "step": 169880
    },
    {
      "epoch": 69.54,
      "grad_norm": 1.8829536437988281,
      "learning_rate": 3.640496298896517e-06,
      "loss": 0.5076,
      "step": 169890
    },
    {
      "epoch": 69.55,
      "grad_norm": 1.7297552824020386,
      "learning_rate": 3.6403641580844404e-06,
      "loss": 0.5062,
      "step": 169900
    },
    {
      "epoch": 69.55,
      "grad_norm": 1.8417885303497314,
      "learning_rate": 3.6402320113327756e-06,
      "loss": 0.5304,
      "step": 169910
    },
    {
      "epoch": 69.55,
      "grad_norm": 1.8816310167312622,
      "learning_rate": 3.6400998586421266e-06,
      "loss": 0.5317,
      "step": 169920
    },
    {
      "epoch": 69.56,
      "grad_norm": 2.1290361881256104,
      "learning_rate": 3.6399677000131002e-06,
      "loss": 0.5396,
      "step": 169930
    },
    {
      "epoch": 69.56,
      "grad_norm": 1.6184325218200684,
      "learning_rate": 3.6398355354463014e-06,
      "loss": 0.509,
      "step": 169940
    },
    {
      "epoch": 69.57,
      "grad_norm": 2.0171730518341064,
      "learning_rate": 3.6397033649423356e-06,
      "loss": 0.5108,
      "step": 169950
    },
    {
      "epoch": 69.57,
      "grad_norm": 1.8724548816680908,
      "learning_rate": 3.6395711885018072e-06,
      "loss": 0.5072,
      "step": 169960
    },
    {
      "epoch": 69.57,
      "grad_norm": 2.1723339557647705,
      "learning_rate": 3.6394390061253233e-06,
      "loss": 0.5252,
      "step": 169970
    },
    {
      "epoch": 69.58,
      "grad_norm": 2.124002456665039,
      "learning_rate": 3.6393068178134893e-06,
      "loss": 0.529,
      "step": 169980
    },
    {
      "epoch": 69.58,
      "grad_norm": 1.7250117063522339,
      "learning_rate": 3.639174623566909e-06,
      "loss": 0.5045,
      "step": 169990
    },
    {
      "epoch": 69.59,
      "grad_norm": 2.2457187175750732,
      "learning_rate": 3.6390424233861895e-06,
      "loss": 0.5137,
      "step": 170000
    },
    {
      "epoch": 69.59,
      "grad_norm": 1.565205693244934,
      "learning_rate": 3.6389102172719367e-06,
      "loss": 0.4901,
      "step": 170010
    },
    {
      "epoch": 69.59,
      "grad_norm": 2.0586917400360107,
      "learning_rate": 3.6387780052247546e-06,
      "loss": 0.51,
      "step": 170020
    },
    {
      "epoch": 69.6,
      "grad_norm": 2.454493284225464,
      "learning_rate": 3.63864578724525e-06,
      "loss": 0.4996,
      "step": 170030
    },
    {
      "epoch": 69.6,
      "grad_norm": 1.7533605098724365,
      "learning_rate": 3.638513563334029e-06,
      "loss": 0.4983,
      "step": 170040
    },
    {
      "epoch": 69.61,
      "grad_norm": 2.308037281036377,
      "learning_rate": 3.638381333491696e-06,
      "loss": 0.5136,
      "step": 170050
    },
    {
      "epoch": 69.61,
      "grad_norm": 1.4973082542419434,
      "learning_rate": 3.638249097718858e-06,
      "loss": 0.5059,
      "step": 170060
    },
    {
      "epoch": 69.62,
      "grad_norm": 1.6320573091506958,
      "learning_rate": 3.638116856016119e-06,
      "loss": 0.5189,
      "step": 170070
    },
    {
      "epoch": 69.62,
      "grad_norm": 1.4886764287948608,
      "learning_rate": 3.637984608384087e-06,
      "loss": 0.521,
      "step": 170080
    },
    {
      "epoch": 69.62,
      "grad_norm": 1.7677181959152222,
      "learning_rate": 3.6378523548233657e-06,
      "loss": 0.5163,
      "step": 170090
    },
    {
      "epoch": 69.63,
      "grad_norm": 2.0003483295440674,
      "learning_rate": 3.637720095334563e-06,
      "loss": 0.5133,
      "step": 170100
    },
    {
      "epoch": 69.63,
      "grad_norm": 1.543048620223999,
      "learning_rate": 3.6375878299182827e-06,
      "loss": 0.4938,
      "step": 170110
    },
    {
      "epoch": 69.64,
      "grad_norm": 1.9996098279953003,
      "learning_rate": 3.6374555585751322e-06,
      "loss": 0.5181,
      "step": 170120
    },
    {
      "epoch": 69.64,
      "grad_norm": 1.8663899898529053,
      "learning_rate": 3.6373232813057166e-06,
      "loss": 0.5037,
      "step": 170130
    },
    {
      "epoch": 69.64,
      "grad_norm": 1.9757425785064697,
      "learning_rate": 3.6371909981106427e-06,
      "loss": 0.5049,
      "step": 170140
    },
    {
      "epoch": 69.65,
      "grad_norm": 1.4962881803512573,
      "learning_rate": 3.6370587089905157e-06,
      "loss": 0.5106,
      "step": 170150
    },
    {
      "epoch": 69.65,
      "grad_norm": 1.565460205078125,
      "learning_rate": 3.6369264139459413e-06,
      "loss": 0.5347,
      "step": 170160
    },
    {
      "epoch": 69.66,
      "grad_norm": 2.004842758178711,
      "learning_rate": 3.6367941129775263e-06,
      "loss": 0.5421,
      "step": 170170
    },
    {
      "epoch": 69.66,
      "grad_norm": 1.525967001914978,
      "learning_rate": 3.6366618060858764e-06,
      "loss": 0.4987,
      "step": 170180
    },
    {
      "epoch": 69.66,
      "grad_norm": 1.7242146730422974,
      "learning_rate": 3.6365294932715973e-06,
      "loss": 0.5256,
      "step": 170190
    },
    {
      "epoch": 69.67,
      "grad_norm": 1.6458202600479126,
      "learning_rate": 3.6363971745352966e-06,
      "loss": 0.5303,
      "step": 170200
    },
    {
      "epoch": 69.67,
      "grad_norm": 2.170722007751465,
      "learning_rate": 3.6362648498775787e-06,
      "loss": 0.5048,
      "step": 170210
    },
    {
      "epoch": 69.68,
      "grad_norm": 1.3979406356811523,
      "learning_rate": 3.6361325192990515e-06,
      "loss": 0.4975,
      "step": 170220
    },
    {
      "epoch": 69.68,
      "grad_norm": 2.411165952682495,
      "learning_rate": 3.636000182800319e-06,
      "loss": 0.5179,
      "step": 170230
    },
    {
      "epoch": 69.68,
      "grad_norm": 1.5746458768844604,
      "learning_rate": 3.6358678403819886e-06,
      "loss": 0.5053,
      "step": 170240
    },
    {
      "epoch": 69.69,
      "grad_norm": 2.042362928390503,
      "learning_rate": 3.6357354920446668e-06,
      "loss": 0.5023,
      "step": 170250
    },
    {
      "epoch": 69.69,
      "grad_norm": 2.4566550254821777,
      "learning_rate": 3.6356031377889593e-06,
      "loss": 0.5199,
      "step": 170260
    },
    {
      "epoch": 69.7,
      "grad_norm": 2.3705077171325684,
      "learning_rate": 3.635470777615473e-06,
      "loss": 0.5144,
      "step": 170270
    },
    {
      "epoch": 69.7,
      "grad_norm": 2.3387067317962646,
      "learning_rate": 3.6353384115248134e-06,
      "loss": 0.5125,
      "step": 170280
    },
    {
      "epoch": 69.71,
      "grad_norm": 2.0918781757354736,
      "learning_rate": 3.6352060395175875e-06,
      "loss": 0.5293,
      "step": 170290
    },
    {
      "epoch": 69.71,
      "grad_norm": 1.841825008392334,
      "learning_rate": 3.6350736615944013e-06,
      "loss": 0.5127,
      "step": 170300
    },
    {
      "epoch": 69.71,
      "grad_norm": 2.1142332553863525,
      "learning_rate": 3.634941277755862e-06,
      "loss": 0.5307,
      "step": 170310
    },
    {
      "epoch": 69.72,
      "grad_norm": 1.9095853567123413,
      "learning_rate": 3.6348088880025754e-06,
      "loss": 0.5259,
      "step": 170320
    },
    {
      "epoch": 69.72,
      "grad_norm": 1.4380699396133423,
      "learning_rate": 3.6346764923351476e-06,
      "loss": 0.4904,
      "step": 170330
    },
    {
      "epoch": 69.73,
      "grad_norm": 1.3795573711395264,
      "learning_rate": 3.634544090754185e-06,
      "loss": 0.5096,
      "step": 170340
    },
    {
      "epoch": 69.73,
      "grad_norm": 2.144181251525879,
      "learning_rate": 3.6344116832602958e-06,
      "loss": 0.528,
      "step": 170350
    },
    {
      "epoch": 69.73,
      "grad_norm": 1.9916478395462036,
      "learning_rate": 3.6342792698540847e-06,
      "loss": 0.5164,
      "step": 170360
    },
    {
      "epoch": 69.74,
      "grad_norm": 2.0823562145233154,
      "learning_rate": 3.6341468505361583e-06,
      "loss": 0.5267,
      "step": 170370
    },
    {
      "epoch": 69.74,
      "grad_norm": 2.051327705383301,
      "learning_rate": 3.634014425307125e-06,
      "loss": 0.5277,
      "step": 170380
    },
    {
      "epoch": 69.75,
      "grad_norm": 1.6562771797180176,
      "learning_rate": 3.633881994167589e-06,
      "loss": 0.5076,
      "step": 170390
    },
    {
      "epoch": 69.75,
      "grad_norm": 2.145602226257324,
      "learning_rate": 3.6337495571181587e-06,
      "loss": 0.5383,
      "step": 170400
    },
    {
      "epoch": 69.75,
      "grad_norm": 1.68825101852417,
      "learning_rate": 3.63361711415944e-06,
      "loss": 0.5221,
      "step": 170410
    },
    {
      "epoch": 69.76,
      "grad_norm": 2.6020984649658203,
      "learning_rate": 3.63348466529204e-06,
      "loss": 0.5043,
      "step": 170420
    },
    {
      "epoch": 69.76,
      "grad_norm": 1.5199940204620361,
      "learning_rate": 3.633352210516565e-06,
      "loss": 0.5311,
      "step": 170430
    },
    {
      "epoch": 69.77,
      "grad_norm": 1.8859294652938843,
      "learning_rate": 3.633219749833622e-06,
      "loss": 0.5232,
      "step": 170440
    },
    {
      "epoch": 69.77,
      "grad_norm": 1.730245590209961,
      "learning_rate": 3.633087283243818e-06,
      "loss": 0.505,
      "step": 170450
    },
    {
      "epoch": 69.77,
      "grad_norm": 1.8021621704101562,
      "learning_rate": 3.6329548107477594e-06,
      "loss": 0.5241,
      "step": 170460
    },
    {
      "epoch": 69.78,
      "grad_norm": 2.974109172821045,
      "learning_rate": 3.632822332346053e-06,
      "loss": 0.528,
      "step": 170470
    },
    {
      "epoch": 69.78,
      "grad_norm": 2.530247211456299,
      "learning_rate": 3.632689848039306e-06,
      "loss": 0.5123,
      "step": 170480
    },
    {
      "epoch": 69.79,
      "grad_norm": 1.6922388076782227,
      "learning_rate": 3.632557357828126e-06,
      "loss": 0.5327,
      "step": 170490
    },
    {
      "epoch": 69.79,
      "grad_norm": 1.8099627494812012,
      "learning_rate": 3.6324248617131183e-06,
      "loss": 0.5166,
      "step": 170500
    },
    {
      "epoch": 69.8,
      "grad_norm": 2.0211470127105713,
      "learning_rate": 3.6322923596948913e-06,
      "loss": 0.5188,
      "step": 170510
    },
    {
      "epoch": 69.8,
      "grad_norm": 2.1593234539031982,
      "learning_rate": 3.63215985177405e-06,
      "loss": 0.5192,
      "step": 170520
    },
    {
      "epoch": 69.8,
      "grad_norm": 1.6837291717529297,
      "learning_rate": 3.632027337951204e-06,
      "loss": 0.5316,
      "step": 170530
    },
    {
      "epoch": 69.81,
      "grad_norm": 2.072617530822754,
      "learning_rate": 3.631894818226959e-06,
      "loss": 0.512,
      "step": 170540
    },
    {
      "epoch": 69.81,
      "grad_norm": 1.621170997619629,
      "learning_rate": 3.6317622926019212e-06,
      "loss": 0.511,
      "step": 170550
    },
    {
      "epoch": 69.82,
      "grad_norm": 1.6936626434326172,
      "learning_rate": 3.6316297610766988e-06,
      "loss": 0.5118,
      "step": 170560
    },
    {
      "epoch": 69.82,
      "grad_norm": 2.0262038707733154,
      "learning_rate": 3.6314972236518987e-06,
      "loss": 0.5169,
      "step": 170570
    },
    {
      "epoch": 69.82,
      "grad_norm": 1.7545597553253174,
      "learning_rate": 3.631364680328128e-06,
      "loss": 0.5187,
      "step": 170580
    },
    {
      "epoch": 69.83,
      "grad_norm": 1.877760410308838,
      "learning_rate": 3.631232131105994e-06,
      "loss": 0.5257,
      "step": 170590
    },
    {
      "epoch": 69.83,
      "grad_norm": 2.3002312183380127,
      "learning_rate": 3.6310995759861044e-06,
      "loss": 0.527,
      "step": 170600
    },
    {
      "epoch": 69.84,
      "grad_norm": 1.8074886798858643,
      "learning_rate": 3.630967014969065e-06,
      "loss": 0.5301,
      "step": 170610
    },
    {
      "epoch": 69.84,
      "grad_norm": 1.568618893623352,
      "learning_rate": 3.630834448055484e-06,
      "loss": 0.5283,
      "step": 170620
    },
    {
      "epoch": 69.84,
      "grad_norm": 2.8973968029022217,
      "learning_rate": 3.6307018752459686e-06,
      "loss": 0.5156,
      "step": 170630
    },
    {
      "epoch": 69.85,
      "grad_norm": 2.1317548751831055,
      "learning_rate": 3.6305692965411258e-06,
      "loss": 0.5334,
      "step": 170640
    },
    {
      "epoch": 69.85,
      "grad_norm": 2.125642776489258,
      "learning_rate": 3.630436711941563e-06,
      "loss": 0.5228,
      "step": 170650
    },
    {
      "epoch": 69.86,
      "grad_norm": 1.8348886966705322,
      "learning_rate": 3.630304121447887e-06,
      "loss": 0.5226,
      "step": 170660
    },
    {
      "epoch": 69.86,
      "grad_norm": 2.276887893676758,
      "learning_rate": 3.630171525060707e-06,
      "loss": 0.5138,
      "step": 170670
    },
    {
      "epoch": 69.86,
      "grad_norm": 1.8029861450195312,
      "learning_rate": 3.6300389227806284e-06,
      "loss": 0.5224,
      "step": 170680
    },
    {
      "epoch": 69.87,
      "grad_norm": 1.5477548837661743,
      "learning_rate": 3.6299063146082603e-06,
      "loss": 0.5212,
      "step": 170690
    },
    {
      "epoch": 69.87,
      "grad_norm": 1.5257505178451538,
      "learning_rate": 3.6297737005442084e-06,
      "loss": 0.5252,
      "step": 170700
    },
    {
      "epoch": 69.88,
      "grad_norm": 1.6145219802856445,
      "learning_rate": 3.629641080589082e-06,
      "loss": 0.5156,
      "step": 170710
    },
    {
      "epoch": 69.88,
      "grad_norm": 2.0199973583221436,
      "learning_rate": 3.629508454743487e-06,
      "loss": 0.5169,
      "step": 170720
    },
    {
      "epoch": 69.89,
      "grad_norm": 1.8195793628692627,
      "learning_rate": 3.6293758230080324e-06,
      "loss": 0.5179,
      "step": 170730
    },
    {
      "epoch": 69.89,
      "grad_norm": 1.7472944259643555,
      "learning_rate": 3.6292431853833237e-06,
      "loss": 0.5192,
      "step": 170740
    },
    {
      "epoch": 69.89,
      "grad_norm": 2.929384469985962,
      "learning_rate": 3.6291105418699703e-06,
      "loss": 0.5031,
      "step": 170750
    },
    {
      "epoch": 69.9,
      "grad_norm": 2.3848628997802734,
      "learning_rate": 3.62897789246858e-06,
      "loss": 0.5058,
      "step": 170760
    },
    {
      "epoch": 69.9,
      "grad_norm": 2.117213726043701,
      "learning_rate": 3.6288452371797594e-06,
      "loss": 0.5053,
      "step": 170770
    },
    {
      "epoch": 69.91,
      "grad_norm": 2.557398557662964,
      "learning_rate": 3.628712576004116e-06,
      "loss": 0.5205,
      "step": 170780
    },
    {
      "epoch": 69.91,
      "grad_norm": 1.5995190143585205,
      "learning_rate": 3.628579908942259e-06,
      "loss": 0.521,
      "step": 170790
    },
    {
      "epoch": 69.91,
      "grad_norm": 1.6625934839248657,
      "learning_rate": 3.6284472359947947e-06,
      "loss": 0.5264,
      "step": 170800
    },
    {
      "epoch": 69.92,
      "grad_norm": 2.454153537750244,
      "learning_rate": 3.628314557162331e-06,
      "loss": 0.5312,
      "step": 170810
    },
    {
      "epoch": 69.92,
      "grad_norm": 1.55994713306427,
      "learning_rate": 3.628181872445477e-06,
      "loss": 0.5152,
      "step": 170820
    },
    {
      "epoch": 69.93,
      "grad_norm": 2.4154484272003174,
      "learning_rate": 3.6280491818448385e-06,
      "loss": 0.5209,
      "step": 170830
    },
    {
      "epoch": 69.93,
      "grad_norm": 1.7845513820648193,
      "learning_rate": 3.627916485361025e-06,
      "loss": 0.5133,
      "step": 170840
    },
    {
      "epoch": 69.93,
      "grad_norm": 1.9195003509521484,
      "learning_rate": 3.627783782994643e-06,
      "loss": 0.5082,
      "step": 170850
    },
    {
      "epoch": 69.94,
      "grad_norm": 1.998376488685608,
      "learning_rate": 3.6276510747463014e-06,
      "loss": 0.5021,
      "step": 170860
    },
    {
      "epoch": 69.94,
      "grad_norm": 2.175527572631836,
      "learning_rate": 3.6275183606166083e-06,
      "loss": 0.5047,
      "step": 170870
    },
    {
      "epoch": 69.95,
      "grad_norm": 1.9851799011230469,
      "learning_rate": 3.62738564060617e-06,
      "loss": 0.5283,
      "step": 170880
    },
    {
      "epoch": 69.95,
      "grad_norm": 1.9007480144500732,
      "learning_rate": 3.6272529147155972e-06,
      "loss": 0.5368,
      "step": 170890
    },
    {
      "epoch": 69.95,
      "grad_norm": 1.9999576807022095,
      "learning_rate": 3.627120182945495e-06,
      "loss": 0.5007,
      "step": 170900
    },
    {
      "epoch": 69.96,
      "grad_norm": 2.0458457469940186,
      "learning_rate": 3.626987445296474e-06,
      "loss": 0.5015,
      "step": 170910
    },
    {
      "epoch": 69.96,
      "grad_norm": 1.9910191297531128,
      "learning_rate": 3.62685470176914e-06,
      "loss": 0.5165,
      "step": 170920
    },
    {
      "epoch": 69.97,
      "grad_norm": 1.6315020322799683,
      "learning_rate": 3.626721952364102e-06,
      "loss": 0.5231,
      "step": 170930
    },
    {
      "epoch": 69.97,
      "grad_norm": 1.6000787019729614,
      "learning_rate": 3.6265891970819685e-06,
      "loss": 0.5329,
      "step": 170940
    },
    {
      "epoch": 69.98,
      "grad_norm": 1.8989510536193848,
      "learning_rate": 3.626456435923347e-06,
      "loss": 0.5058,
      "step": 170950
    },
    {
      "epoch": 69.98,
      "grad_norm": 1.6747177839279175,
      "learning_rate": 3.6263236688888465e-06,
      "loss": 0.5195,
      "step": 170960
    },
    {
      "epoch": 69.98,
      "grad_norm": 2.3908209800720215,
      "learning_rate": 3.6261908959790746e-06,
      "loss": 0.5096,
      "step": 170970
    },
    {
      "epoch": 69.99,
      "grad_norm": 1.7491734027862549,
      "learning_rate": 3.6260581171946396e-06,
      "loss": 0.519,
      "step": 170980
    },
    {
      "epoch": 69.99,
      "grad_norm": 1.8482755422592163,
      "learning_rate": 3.6259253325361497e-06,
      "loss": 0.511,
      "step": 170990
    },
    {
      "epoch": 70.0,
      "grad_norm": 2.2414345741271973,
      "learning_rate": 3.625792542004213e-06,
      "loss": 0.5234,
      "step": 171000
    },
    {
      "epoch": 70.0,
      "grad_norm": 1.9816951751708984,
      "learning_rate": 3.6256597455994383e-06,
      "loss": 0.5231,
      "step": 171010
    },
    {
      "epoch": 70.0,
      "eval_loss": 0.5175246596336365,
      "eval_runtime": 61.5978,
      "eval_samples_per_second": 55.992,
      "eval_steps_per_second": 7.013,
      "step": 171010
    },
    {
      "epoch": 70.0,
      "grad_norm": 1.9447940587997437,
      "learning_rate": 3.6255269433224327e-06,
      "loss": 0.5274,
      "step": 171020
    },
    {
      "epoch": 70.01,
      "grad_norm": 1.6237167119979858,
      "learning_rate": 3.6253941351738057e-06,
      "loss": 0.5114,
      "step": 171030
    },
    {
      "epoch": 70.01,
      "grad_norm": 1.867166519165039,
      "learning_rate": 3.625261321154166e-06,
      "loss": 0.5239,
      "step": 171040
    },
    {
      "epoch": 70.02,
      "grad_norm": 2.132662296295166,
      "learning_rate": 3.625128501264121e-06,
      "loss": 0.5162,
      "step": 171050
    },
    {
      "epoch": 70.02,
      "grad_norm": 1.9406626224517822,
      "learning_rate": 3.624995675504279e-06,
      "loss": 0.5066,
      "step": 171060
    },
    {
      "epoch": 70.02,
      "grad_norm": 2.663431167602539,
      "learning_rate": 3.6248628438752504e-06,
      "loss": 0.5128,
      "step": 171070
    },
    {
      "epoch": 70.03,
      "grad_norm": 2.0025172233581543,
      "learning_rate": 3.624730006377641e-06,
      "loss": 0.5455,
      "step": 171080
    },
    {
      "epoch": 70.03,
      "grad_norm": 2.161970615386963,
      "learning_rate": 3.624597163012061e-06,
      "loss": 0.5171,
      "step": 171090
    },
    {
      "epoch": 70.04,
      "grad_norm": 2.0169014930725098,
      "learning_rate": 3.624464313779119e-06,
      "loss": 0.5283,
      "step": 171100
    },
    {
      "epoch": 70.04,
      "grad_norm": 1.902296781539917,
      "learning_rate": 3.6243314586794226e-06,
      "loss": 0.5274,
      "step": 171110
    },
    {
      "epoch": 70.05,
      "grad_norm": 1.4999374151229858,
      "learning_rate": 3.624198597713581e-06,
      "loss": 0.5192,
      "step": 171120
    },
    {
      "epoch": 70.05,
      "grad_norm": 2.2773404121398926,
      "learning_rate": 3.624065730882202e-06,
      "loss": 0.5138,
      "step": 171130
    },
    {
      "epoch": 70.05,
      "grad_norm": 2.9436655044555664,
      "learning_rate": 3.6239328581858955e-06,
      "loss": 0.5143,
      "step": 171140
    },
    {
      "epoch": 70.06,
      "grad_norm": 2.1474735736846924,
      "learning_rate": 3.62379997962527e-06,
      "loss": 0.5214,
      "step": 171150
    },
    {
      "epoch": 70.06,
      "grad_norm": 1.8213127851486206,
      "learning_rate": 3.6236670952009335e-06,
      "loss": 0.5081,
      "step": 171160
    },
    {
      "epoch": 70.07,
      "grad_norm": 1.4545403718948364,
      "learning_rate": 3.6235342049134946e-06,
      "loss": 0.4995,
      "step": 171170
    },
    {
      "epoch": 70.07,
      "grad_norm": 2.093916416168213,
      "learning_rate": 3.623401308763563e-06,
      "loss": 0.5343,
      "step": 171180
    },
    {
      "epoch": 70.07,
      "grad_norm": 2.155012369155884,
      "learning_rate": 3.623268406751747e-06,
      "loss": 0.5367,
      "step": 171190
    },
    {
      "epoch": 70.08,
      "grad_norm": 1.8410884141921997,
      "learning_rate": 3.6231354988786543e-06,
      "loss": 0.5135,
      "step": 171200
    },
    {
      "epoch": 70.08,
      "grad_norm": 1.777658462524414,
      "learning_rate": 3.6230025851448954e-06,
      "loss": 0.5041,
      "step": 171210
    },
    {
      "epoch": 70.09,
      "grad_norm": 1.6987078189849854,
      "learning_rate": 3.6228696655510784e-06,
      "loss": 0.5184,
      "step": 171220
    },
    {
      "epoch": 70.09,
      "grad_norm": 1.8075426816940308,
      "learning_rate": 3.6227367400978128e-06,
      "loss": 0.5295,
      "step": 171230
    },
    {
      "epoch": 70.09,
      "grad_norm": 2.0225462913513184,
      "learning_rate": 3.6226038087857066e-06,
      "loss": 0.5257,
      "step": 171240
    },
    {
      "epoch": 70.1,
      "grad_norm": 2.031608819961548,
      "learning_rate": 3.6224708716153694e-06,
      "loss": 0.5113,
      "step": 171250
    },
    {
      "epoch": 70.1,
      "grad_norm": 1.8180382251739502,
      "learning_rate": 3.6223379285874097e-06,
      "loss": 0.5319,
      "step": 171260
    },
    {
      "epoch": 70.11,
      "grad_norm": 1.7527521848678589,
      "learning_rate": 3.622204979702437e-06,
      "loss": 0.5399,
      "step": 171270
    },
    {
      "epoch": 70.11,
      "grad_norm": 1.8334707021713257,
      "learning_rate": 3.6220720249610594e-06,
      "loss": 0.53,
      "step": 171280
    },
    {
      "epoch": 70.11,
      "grad_norm": 1.6362197399139404,
      "learning_rate": 3.6219390643638873e-06,
      "loss": 0.5277,
      "step": 171290
    },
    {
      "epoch": 70.12,
      "grad_norm": 2.0151209831237793,
      "learning_rate": 3.621806097911529e-06,
      "loss": 0.509,
      "step": 171300
    },
    {
      "epoch": 70.12,
      "grad_norm": 2.024897575378418,
      "learning_rate": 3.621673125604593e-06,
      "loss": 0.5342,
      "step": 171310
    },
    {
      "epoch": 70.13,
      "grad_norm": 1.531581163406372,
      "learning_rate": 3.6215401474436894e-06,
      "loss": 0.5179,
      "step": 171320
    },
    {
      "epoch": 70.13,
      "grad_norm": 2.1736438274383545,
      "learning_rate": 3.6214071634294275e-06,
      "loss": 0.5107,
      "step": 171330
    },
    {
      "epoch": 70.14,
      "grad_norm": 1.3213834762573242,
      "learning_rate": 3.6212741735624152e-06,
      "loss": 0.4947,
      "step": 171340
    },
    {
      "epoch": 70.14,
      "grad_norm": 1.2689183950424194,
      "learning_rate": 3.6211411778432634e-06,
      "loss": 0.5063,
      "step": 171350
    },
    {
      "epoch": 70.14,
      "grad_norm": 2.112739086151123,
      "learning_rate": 3.6210081762725796e-06,
      "loss": 0.5001,
      "step": 171360
    },
    {
      "epoch": 70.15,
      "grad_norm": 1.8485900163650513,
      "learning_rate": 3.6208751688509747e-06,
      "loss": 0.4931,
      "step": 171370
    },
    {
      "epoch": 70.15,
      "grad_norm": 2.3963658809661865,
      "learning_rate": 3.6207421555790563e-06,
      "loss": 0.5243,
      "step": 171380
    },
    {
      "epoch": 70.16,
      "grad_norm": 2.035855770111084,
      "learning_rate": 3.6206091364574357e-06,
      "loss": 0.5225,
      "step": 171390
    },
    {
      "epoch": 70.16,
      "grad_norm": 1.5059353113174438,
      "learning_rate": 3.6204761114867197e-06,
      "loss": 0.5185,
      "step": 171400
    },
    {
      "epoch": 70.16,
      "grad_norm": 1.7206541299819946,
      "learning_rate": 3.62034308066752e-06,
      "loss": 0.5112,
      "step": 171410
    },
    {
      "epoch": 70.17,
      "grad_norm": 2.092515707015991,
      "learning_rate": 3.620210044000445e-06,
      "loss": 0.5127,
      "step": 171420
    },
    {
      "epoch": 70.17,
      "grad_norm": 2.0173885822296143,
      "learning_rate": 3.6200770014861048e-06,
      "loss": 0.5361,
      "step": 171430
    },
    {
      "epoch": 70.18,
      "grad_norm": 1.7226824760437012,
      "learning_rate": 3.6199439531251076e-06,
      "loss": 0.5053,
      "step": 171440
    },
    {
      "epoch": 70.18,
      "grad_norm": 2.45088791847229,
      "learning_rate": 3.619810898918064e-06,
      "loss": 0.513,
      "step": 171450
    },
    {
      "epoch": 70.18,
      "grad_norm": 2.1154253482818604,
      "learning_rate": 3.6196778388655822e-06,
      "loss": 0.5137,
      "step": 171460
    },
    {
      "epoch": 70.19,
      "grad_norm": 1.4598023891448975,
      "learning_rate": 3.619544772968273e-06,
      "loss": 0.5045,
      "step": 171470
    },
    {
      "epoch": 70.19,
      "grad_norm": 2.5676627159118652,
      "learning_rate": 3.619411701226746e-06,
      "loss": 0.508,
      "step": 171480
    },
    {
      "epoch": 70.2,
      "grad_norm": 1.5709517002105713,
      "learning_rate": 3.6192786236416095e-06,
      "loss": 0.5292,
      "step": 171490
    },
    {
      "epoch": 70.2,
      "grad_norm": 2.5291318893432617,
      "learning_rate": 3.619145540213474e-06,
      "loss": 0.5157,
      "step": 171500
    },
    {
      "epoch": 70.2,
      "grad_norm": 1.5946314334869385,
      "learning_rate": 3.6190124509429493e-06,
      "loss": 0.5152,
      "step": 171510
    },
    {
      "epoch": 70.21,
      "grad_norm": 1.8285197019577026,
      "learning_rate": 3.6188793558306445e-06,
      "loss": 0.4871,
      "step": 171520
    },
    {
      "epoch": 70.21,
      "grad_norm": 2.0481700897216797,
      "learning_rate": 3.61874625487717e-06,
      "loss": 0.5289,
      "step": 171530
    },
    {
      "epoch": 70.22,
      "grad_norm": 1.9123485088348389,
      "learning_rate": 3.618613148083136e-06,
      "loss": 0.5061,
      "step": 171540
    },
    {
      "epoch": 70.22,
      "grad_norm": 1.6102648973464966,
      "learning_rate": 3.61848003544915e-06,
      "loss": 0.5157,
      "step": 171550
    },
    {
      "epoch": 70.23,
      "grad_norm": 1.5111855268478394,
      "learning_rate": 3.6183469169758235e-06,
      "loss": 0.5025,
      "step": 171560
    },
    {
      "epoch": 70.23,
      "grad_norm": 1.941924810409546,
      "learning_rate": 3.6182137926637663e-06,
      "loss": 0.5081,
      "step": 171570
    },
    {
      "epoch": 70.23,
      "grad_norm": 2.1054115295410156,
      "learning_rate": 3.6180806625135875e-06,
      "loss": 0.5216,
      "step": 171580
    },
    {
      "epoch": 70.24,
      "grad_norm": 1.8681799173355103,
      "learning_rate": 3.6179475265258977e-06,
      "loss": 0.5009,
      "step": 171590
    },
    {
      "epoch": 70.24,
      "grad_norm": 1.6935224533081055,
      "learning_rate": 3.617814384701305e-06,
      "loss": 0.527,
      "step": 171600
    },
    {
      "epoch": 70.25,
      "grad_norm": 1.7776695489883423,
      "learning_rate": 3.6176812370404222e-06,
      "loss": 0.5061,
      "step": 171610
    },
    {
      "epoch": 70.25,
      "grad_norm": 2.0589025020599365,
      "learning_rate": 3.6175480835438567e-06,
      "loss": 0.5165,
      "step": 171620
    },
    {
      "epoch": 70.25,
      "grad_norm": 2.306497812271118,
      "learning_rate": 3.6174149242122205e-06,
      "loss": 0.497,
      "step": 171630
    },
    {
      "epoch": 70.26,
      "grad_norm": 1.9263488054275513,
      "learning_rate": 3.6172817590461215e-06,
      "loss": 0.5174,
      "step": 171640
    },
    {
      "epoch": 70.26,
      "grad_norm": 2.0652318000793457,
      "learning_rate": 3.6171485880461715e-06,
      "loss": 0.4989,
      "step": 171650
    },
    {
      "epoch": 70.27,
      "grad_norm": 1.658857822418213,
      "learning_rate": 3.6170154112129796e-06,
      "loss": 0.5309,
      "step": 171660
    },
    {
      "epoch": 70.27,
      "grad_norm": 2.0724573135375977,
      "learning_rate": 3.6168822285471557e-06,
      "loss": 0.4962,
      "step": 171670
    },
    {
      "epoch": 70.27,
      "grad_norm": 1.6327968835830688,
      "learning_rate": 3.6167490400493105e-06,
      "loss": 0.5201,
      "step": 171680
    },
    {
      "epoch": 70.28,
      "grad_norm": 2.106119155883789,
      "learning_rate": 3.616615845720054e-06,
      "loss": 0.5487,
      "step": 171690
    },
    {
      "epoch": 70.28,
      "grad_norm": 1.6649856567382812,
      "learning_rate": 3.616482645559996e-06,
      "loss": 0.5047,
      "step": 171700
    },
    {
      "epoch": 70.29,
      "grad_norm": 1.8431799411773682,
      "learning_rate": 3.616349439569746e-06,
      "loss": 0.4929,
      "step": 171710
    },
    {
      "epoch": 70.29,
      "grad_norm": 1.3740098476409912,
      "learning_rate": 3.6162162277499167e-06,
      "loss": 0.5311,
      "step": 171720
    },
    {
      "epoch": 70.29,
      "grad_norm": 2.211353063583374,
      "learning_rate": 3.6160830101011154e-06,
      "loss": 0.5315,
      "step": 171730
    },
    {
      "epoch": 70.3,
      "grad_norm": 1.69297194480896,
      "learning_rate": 3.6159497866239545e-06,
      "loss": 0.498,
      "step": 171740
    },
    {
      "epoch": 70.3,
      "grad_norm": 1.626726508140564,
      "learning_rate": 3.6158165573190426e-06,
      "loss": 0.523,
      "step": 171750
    },
    {
      "epoch": 70.31,
      "grad_norm": 1.569383144378662,
      "learning_rate": 3.6156833221869917e-06,
      "loss": 0.5298,
      "step": 171760
    },
    {
      "epoch": 70.31,
      "grad_norm": 1.9081952571868896,
      "learning_rate": 3.6155500812284103e-06,
      "loss": 0.5196,
      "step": 171770
    },
    {
      "epoch": 70.32,
      "grad_norm": 1.735011100769043,
      "learning_rate": 3.6154168344439105e-06,
      "loss": 0.5062,
      "step": 171780
    },
    {
      "epoch": 70.32,
      "grad_norm": 1.7179756164550781,
      "learning_rate": 3.615283581834101e-06,
      "loss": 0.5213,
      "step": 171790
    },
    {
      "epoch": 70.32,
      "grad_norm": 1.4601582288742065,
      "learning_rate": 3.6151503233995938e-06,
      "loss": 0.5005,
      "step": 171800
    },
    {
      "epoch": 70.33,
      "grad_norm": 2.1883740425109863,
      "learning_rate": 3.615017059140998e-06,
      "loss": 0.5166,
      "step": 171810
    },
    {
      "epoch": 70.33,
      "grad_norm": 1.3439054489135742,
      "learning_rate": 3.6148837890589254e-06,
      "loss": 0.5172,
      "step": 171820
    },
    {
      "epoch": 70.34,
      "grad_norm": 1.7588688135147095,
      "learning_rate": 3.6147505131539856e-06,
      "loss": 0.5208,
      "step": 171830
    },
    {
      "epoch": 70.34,
      "grad_norm": 1.8683375120162964,
      "learning_rate": 3.614617231426789e-06,
      "loss": 0.5195,
      "step": 171840
    },
    {
      "epoch": 70.34,
      "grad_norm": 1.4477083683013916,
      "learning_rate": 3.6144839438779464e-06,
      "loss": 0.5263,
      "step": 171850
    },
    {
      "epoch": 70.35,
      "grad_norm": 1.697092890739441,
      "learning_rate": 3.614350650508069e-06,
      "loss": 0.5056,
      "step": 171860
    },
    {
      "epoch": 70.35,
      "grad_norm": 1.8589787483215332,
      "learning_rate": 3.614217351317766e-06,
      "loss": 0.5059,
      "step": 171870
    },
    {
      "epoch": 70.36,
      "grad_norm": 1.6585134267807007,
      "learning_rate": 3.614084046307649e-06,
      "loss": 0.5301,
      "step": 171880
    },
    {
      "epoch": 70.36,
      "grad_norm": 2.351032018661499,
      "learning_rate": 3.6139507354783286e-06,
      "loss": 0.5075,
      "step": 171890
    },
    {
      "epoch": 70.36,
      "grad_norm": 2.3367185592651367,
      "learning_rate": 3.6138174188304157e-06,
      "loss": 0.5187,
      "step": 171900
    },
    {
      "epoch": 70.37,
      "grad_norm": 1.6415218114852905,
      "learning_rate": 3.6136840963645203e-06,
      "loss": 0.4839,
      "step": 171910
    },
    {
      "epoch": 70.37,
      "grad_norm": 1.9254282712936401,
      "learning_rate": 3.613550768081253e-06,
      "loss": 0.5342,
      "step": 171920
    },
    {
      "epoch": 70.38,
      "grad_norm": 1.686892032623291,
      "learning_rate": 3.6134174339812263e-06,
      "loss": 0.5089,
      "step": 171930
    },
    {
      "epoch": 70.38,
      "grad_norm": 1.8749555349349976,
      "learning_rate": 3.6132840940650486e-06,
      "loss": 0.5199,
      "step": 171940
    },
    {
      "epoch": 70.38,
      "grad_norm": 2.3358466625213623,
      "learning_rate": 3.6131507483333325e-06,
      "loss": 0.5013,
      "step": 171950
    },
    {
      "epoch": 70.39,
      "grad_norm": 1.9411674737930298,
      "learning_rate": 3.6130173967866877e-06,
      "loss": 0.5116,
      "step": 171960
    },
    {
      "epoch": 70.39,
      "grad_norm": 1.6224154233932495,
      "learning_rate": 3.612884039425726e-06,
      "loss": 0.512,
      "step": 171970
    },
    {
      "epoch": 70.4,
      "grad_norm": 1.9573678970336914,
      "learning_rate": 3.6127506762510574e-06,
      "loss": 0.5308,
      "step": 171980
    },
    {
      "epoch": 70.4,
      "grad_norm": 1.5651187896728516,
      "learning_rate": 3.612617307263293e-06,
      "loss": 0.5033,
      "step": 171990
    },
    {
      "epoch": 70.41,
      "grad_norm": 2.048389434814453,
      "learning_rate": 3.6124839324630444e-06,
      "loss": 0.5089,
      "step": 172000
    },
    {
      "epoch": 70.41,
      "grad_norm": 1.3171746730804443,
      "learning_rate": 3.6123505518509227e-06,
      "loss": 0.5245,
      "step": 172010
    },
    {
      "epoch": 70.41,
      "grad_norm": 1.966212272644043,
      "learning_rate": 3.612217165427537e-06,
      "loss": 0.5064,
      "step": 172020
    },
    {
      "epoch": 70.42,
      "grad_norm": 1.6115915775299072,
      "learning_rate": 3.612083773193501e-06,
      "loss": 0.5168,
      "step": 172030
    },
    {
      "epoch": 70.42,
      "grad_norm": 1.9240115880966187,
      "learning_rate": 3.611950375149424e-06,
      "loss": 0.5273,
      "step": 172040
    },
    {
      "epoch": 70.43,
      "grad_norm": 1.38029146194458,
      "learning_rate": 3.611816971295917e-06,
      "loss": 0.5208,
      "step": 172050
    },
    {
      "epoch": 70.43,
      "grad_norm": 1.9915754795074463,
      "learning_rate": 3.6116835616335917e-06,
      "loss": 0.5014,
      "step": 172060
    },
    {
      "epoch": 70.43,
      "grad_norm": 1.5775219202041626,
      "learning_rate": 3.6115501461630586e-06,
      "loss": 0.5168,
      "step": 172070
    },
    {
      "epoch": 70.44,
      "grad_norm": 1.9711686372756958,
      "learning_rate": 3.6114167248849304e-06,
      "loss": 0.5303,
      "step": 172080
    },
    {
      "epoch": 70.44,
      "grad_norm": 1.7402101755142212,
      "learning_rate": 3.611283297799817e-06,
      "loss": 0.5254,
      "step": 172090
    },
    {
      "epoch": 70.45,
      "grad_norm": 1.941143274307251,
      "learning_rate": 3.61114986490833e-06,
      "loss": 0.5339,
      "step": 172100
    },
    {
      "epoch": 70.45,
      "grad_norm": 2.2476439476013184,
      "learning_rate": 3.61101642621108e-06,
      "loss": 0.5297,
      "step": 172110
    },
    {
      "epoch": 70.45,
      "grad_norm": 1.7119433879852295,
      "learning_rate": 3.6108829817086786e-06,
      "loss": 0.5458,
      "step": 172120
    },
    {
      "epoch": 70.46,
      "grad_norm": 1.5959298610687256,
      "learning_rate": 3.610749531401738e-06,
      "loss": 0.5078,
      "step": 172130
    },
    {
      "epoch": 70.46,
      "grad_norm": 3.3957624435424805,
      "learning_rate": 3.6106160752908683e-06,
      "loss": 0.5103,
      "step": 172140
    },
    {
      "epoch": 70.47,
      "grad_norm": 1.5709266662597656,
      "learning_rate": 3.610482613376682e-06,
      "loss": 0.5106,
      "step": 172150
    },
    {
      "epoch": 70.47,
      "grad_norm": 1.8578583002090454,
      "learning_rate": 3.6103491456597885e-06,
      "loss": 0.5189,
      "step": 172160
    },
    {
      "epoch": 70.47,
      "grad_norm": 1.6508901119232178,
      "learning_rate": 3.6102156721408013e-06,
      "loss": 0.5085,
      "step": 172170
    },
    {
      "epoch": 70.48,
      "grad_norm": 2.3852782249450684,
      "learning_rate": 3.610082192820331e-06,
      "loss": 0.5406,
      "step": 172180
    },
    {
      "epoch": 70.48,
      "grad_norm": 2.0864477157592773,
      "learning_rate": 3.609948707698989e-06,
      "loss": 0.5052,
      "step": 172190
    },
    {
      "epoch": 70.49,
      "grad_norm": 1.801318883895874,
      "learning_rate": 3.609815216777387e-06,
      "loss": 0.5252,
      "step": 172200
    },
    {
      "epoch": 70.49,
      "grad_norm": 1.7489069700241089,
      "learning_rate": 3.6096817200561355e-06,
      "loss": 0.5011,
      "step": 172210
    },
    {
      "epoch": 70.5,
      "grad_norm": 1.8159226179122925,
      "learning_rate": 3.609548217535848e-06,
      "loss": 0.497,
      "step": 172220
    },
    {
      "epoch": 70.5,
      "grad_norm": 3.1670358180999756,
      "learning_rate": 3.609414709217134e-06,
      "loss": 0.4963,
      "step": 172230
    },
    {
      "epoch": 70.5,
      "grad_norm": 1.465694785118103,
      "learning_rate": 3.6092811951006067e-06,
      "loss": 0.5107,
      "step": 172240
    },
    {
      "epoch": 70.51,
      "grad_norm": 1.5719417333602905,
      "learning_rate": 3.6091476751868764e-06,
      "loss": 0.4992,
      "step": 172250
    },
    {
      "epoch": 70.51,
      "grad_norm": 2.0537891387939453,
      "learning_rate": 3.6090141494765556e-06,
      "loss": 0.507,
      "step": 172260
    },
    {
      "epoch": 70.52,
      "grad_norm": 2.392252206802368,
      "learning_rate": 3.6088806179702552e-06,
      "loss": 0.515,
      "step": 172270
    },
    {
      "epoch": 70.52,
      "grad_norm": 2.312291383743286,
      "learning_rate": 3.6087470806685876e-06,
      "loss": 0.4965,
      "step": 172280
    },
    {
      "epoch": 70.52,
      "grad_norm": 2.013476848602295,
      "learning_rate": 3.608613537572165e-06,
      "loss": 0.5063,
      "step": 172290
    },
    {
      "epoch": 70.53,
      "grad_norm": 2.064134359359741,
      "learning_rate": 3.6084799886815978e-06,
      "loss": 0.5339,
      "step": 172300
    },
    {
      "epoch": 70.53,
      "grad_norm": 2.094750165939331,
      "learning_rate": 3.6083464339974986e-06,
      "loss": 0.5276,
      "step": 172310
    },
    {
      "epoch": 70.54,
      "grad_norm": 1.3436423540115356,
      "learning_rate": 3.608212873520479e-06,
      "loss": 0.5254,
      "step": 172320
    },
    {
      "epoch": 70.54,
      "grad_norm": 1.706003189086914,
      "learning_rate": 3.60807930725115e-06,
      "loss": 0.513,
      "step": 172330
    },
    {
      "epoch": 70.54,
      "grad_norm": 1.4768399000167847,
      "learning_rate": 3.6079457351901252e-06,
      "loss": 0.5259,
      "step": 172340
    },
    {
      "epoch": 70.55,
      "grad_norm": 1.962741732597351,
      "learning_rate": 3.6078121573380152e-06,
      "loss": 0.5364,
      "step": 172350
    },
    {
      "epoch": 70.55,
      "grad_norm": 1.9031448364257812,
      "learning_rate": 3.6076785736954317e-06,
      "loss": 0.5273,
      "step": 172360
    },
    {
      "epoch": 70.56,
      "grad_norm": 1.9164389371871948,
      "learning_rate": 3.607544984262988e-06,
      "loss": 0.5172,
      "step": 172370
    },
    {
      "epoch": 70.56,
      "grad_norm": 1.6394082307815552,
      "learning_rate": 3.607411389041295e-06,
      "loss": 0.5125,
      "step": 172380
    },
    {
      "epoch": 70.56,
      "grad_norm": 1.7133508920669556,
      "learning_rate": 3.6072777880309646e-06,
      "loss": 0.5226,
      "step": 172390
    },
    {
      "epoch": 70.57,
      "grad_norm": 2.193923234939575,
      "learning_rate": 3.607144181232609e-06,
      "loss": 0.5103,
      "step": 172400
    },
    {
      "epoch": 70.57,
      "grad_norm": 1.651383638381958,
      "learning_rate": 3.6070105686468407e-06,
      "loss": 0.5277,
      "step": 172410
    },
    {
      "epoch": 70.58,
      "grad_norm": 1.9295579195022583,
      "learning_rate": 3.606876950274271e-06,
      "loss": 0.5069,
      "step": 172420
    },
    {
      "epoch": 70.58,
      "grad_norm": 1.8268723487854004,
      "learning_rate": 3.6067433261155132e-06,
      "loss": 0.5201,
      "step": 172430
    },
    {
      "epoch": 70.59,
      "grad_norm": 1.791853427886963,
      "learning_rate": 3.606609696171178e-06,
      "loss": 0.5206,
      "step": 172440
    },
    {
      "epoch": 70.59,
      "grad_norm": 1.791640281677246,
      "learning_rate": 3.6064760604418777e-06,
      "loss": 0.4989,
      "step": 172450
    },
    {
      "epoch": 70.59,
      "grad_norm": 2.1389501094818115,
      "learning_rate": 3.606342418928225e-06,
      "loss": 0.5479,
      "step": 172460
    },
    {
      "epoch": 70.6,
      "grad_norm": 2.750983238220215,
      "learning_rate": 3.6062087716308325e-06,
      "loss": 0.5145,
      "step": 172470
    },
    {
      "epoch": 70.6,
      "grad_norm": 2.5526163578033447,
      "learning_rate": 3.606075118550312e-06,
      "loss": 0.5215,
      "step": 172480
    },
    {
      "epoch": 70.61,
      "grad_norm": 2.231300115585327,
      "learning_rate": 3.605941459687275e-06,
      "loss": 0.5118,
      "step": 172490
    },
    {
      "epoch": 70.61,
      "grad_norm": 1.8459224700927734,
      "learning_rate": 3.605807795042334e-06,
      "loss": 0.5189,
      "step": 172500
    },
    {
      "epoch": 70.61,
      "grad_norm": 1.6037079095840454,
      "learning_rate": 3.605674124616103e-06,
      "loss": 0.5243,
      "step": 172510
    },
    {
      "epoch": 70.62,
      "grad_norm": 1.6794010400772095,
      "learning_rate": 3.605540448409192e-06,
      "loss": 0.5088,
      "step": 172520
    },
    {
      "epoch": 70.62,
      "grad_norm": 2.0910985469818115,
      "learning_rate": 3.605406766422215e-06,
      "loss": 0.5133,
      "step": 172530
    },
    {
      "epoch": 70.63,
      "grad_norm": 1.5266233682632446,
      "learning_rate": 3.605273078655783e-06,
      "loss": 0.5146,
      "step": 172540
    },
    {
      "epoch": 70.63,
      "grad_norm": 1.7375925779342651,
      "learning_rate": 3.60513938511051e-06,
      "loss": 0.5065,
      "step": 172550
    },
    {
      "epoch": 70.63,
      "grad_norm": 2.145730972290039,
      "learning_rate": 3.605005685787007e-06,
      "loss": 0.5109,
      "step": 172560
    },
    {
      "epoch": 70.64,
      "grad_norm": 1.6842138767242432,
      "learning_rate": 3.604871980685887e-06,
      "loss": 0.5076,
      "step": 172570
    },
    {
      "epoch": 70.64,
      "grad_norm": 1.6902481317520142,
      "learning_rate": 3.6047382698077634e-06,
      "loss": 0.525,
      "step": 172580
    },
    {
      "epoch": 70.65,
      "grad_norm": 2.0145106315612793,
      "learning_rate": 3.6046045531532465e-06,
      "loss": 0.5235,
      "step": 172590
    },
    {
      "epoch": 70.65,
      "grad_norm": 1.94094979763031,
      "learning_rate": 3.6044708307229513e-06,
      "loss": 0.5198,
      "step": 172600
    },
    {
      "epoch": 70.65,
      "grad_norm": 1.7788721323013306,
      "learning_rate": 3.604337102517489e-06,
      "loss": 0.5142,
      "step": 172610
    },
    {
      "epoch": 70.66,
      "grad_norm": 1.6440391540527344,
      "learning_rate": 3.6042033685374717e-06,
      "loss": 0.4926,
      "step": 172620
    },
    {
      "epoch": 70.66,
      "grad_norm": 1.8454055786132812,
      "learning_rate": 3.6040696287835134e-06,
      "loss": 0.4982,
      "step": 172630
    },
    {
      "epoch": 70.67,
      "grad_norm": 1.8132729530334473,
      "learning_rate": 3.603935883256226e-06,
      "loss": 0.496,
      "step": 172640
    },
    {
      "epoch": 70.67,
      "grad_norm": 2.2977852821350098,
      "learning_rate": 3.603802131956222e-06,
      "loss": 0.5113,
      "step": 172650
    },
    {
      "epoch": 70.68,
      "grad_norm": 1.693930745124817,
      "learning_rate": 3.603668374884114e-06,
      "loss": 0.5157,
      "step": 172660
    },
    {
      "epoch": 70.68,
      "grad_norm": 1.7910231351852417,
      "learning_rate": 3.6035346120405158e-06,
      "loss": 0.5295,
      "step": 172670
    },
    {
      "epoch": 70.68,
      "grad_norm": 2.3082234859466553,
      "learning_rate": 3.603400843426039e-06,
      "loss": 0.5312,
      "step": 172680
    },
    {
      "epoch": 70.69,
      "grad_norm": 1.8006081581115723,
      "learning_rate": 3.6032670690412964e-06,
      "loss": 0.5156,
      "step": 172690
    },
    {
      "epoch": 70.69,
      "grad_norm": 1.7461137771606445,
      "learning_rate": 3.603133288886902e-06,
      "loss": 0.5119,
      "step": 172700
    },
    {
      "epoch": 70.7,
      "grad_norm": 1.6827090978622437,
      "learning_rate": 3.6029995029634667e-06,
      "loss": 0.5127,
      "step": 172710
    },
    {
      "epoch": 70.7,
      "grad_norm": 2.1413521766662598,
      "learning_rate": 3.6028657112716054e-06,
      "loss": 0.5238,
      "step": 172720
    },
    {
      "epoch": 70.7,
      "grad_norm": 1.4539343118667603,
      "learning_rate": 3.60273191381193e-06,
      "loss": 0.5161,
      "step": 172730
    },
    {
      "epoch": 70.71,
      "grad_norm": 1.414546251296997,
      "learning_rate": 3.602598110585053e-06,
      "loss": 0.5181,
      "step": 172740
    },
    {
      "epoch": 70.71,
      "grad_norm": 1.7199196815490723,
      "learning_rate": 3.602464301591588e-06,
      "loss": 0.4936,
      "step": 172750
    },
    {
      "epoch": 70.72,
      "grad_norm": 1.6606545448303223,
      "learning_rate": 3.6023304868321477e-06,
      "loss": 0.52,
      "step": 172760
    },
    {
      "epoch": 70.72,
      "grad_norm": 1.8952254056930542,
      "learning_rate": 3.602196666307345e-06,
      "loss": 0.5066,
      "step": 172770
    },
    {
      "epoch": 70.72,
      "grad_norm": 1.7723007202148438,
      "learning_rate": 3.602062840017793e-06,
      "loss": 0.5365,
      "step": 172780
    },
    {
      "epoch": 70.73,
      "grad_norm": 2.129183530807495,
      "learning_rate": 3.601929007964105e-06,
      "loss": 0.5199,
      "step": 172790
    },
    {
      "epoch": 70.73,
      "grad_norm": 2.135413408279419,
      "learning_rate": 3.6017951701468933e-06,
      "loss": 0.5285,
      "step": 172800
    },
    {
      "epoch": 70.74,
      "grad_norm": 2.0098090171813965,
      "learning_rate": 3.6016613265667725e-06,
      "loss": 0.5172,
      "step": 172810
    },
    {
      "epoch": 70.74,
      "grad_norm": 2.252859115600586,
      "learning_rate": 3.6015274772243534e-06,
      "loss": 0.512,
      "step": 172820
    },
    {
      "epoch": 70.74,
      "grad_norm": 1.5882495641708374,
      "learning_rate": 3.6013936221202515e-06,
      "loss": 0.5142,
      "step": 172830
    },
    {
      "epoch": 70.75,
      "grad_norm": 1.6562848091125488,
      "learning_rate": 3.6012597612550782e-06,
      "loss": 0.5166,
      "step": 172840
    },
    {
      "epoch": 70.75,
      "grad_norm": 1.6794651746749878,
      "learning_rate": 3.6011258946294474e-06,
      "loss": 0.5249,
      "step": 172850
    },
    {
      "epoch": 70.76,
      "grad_norm": 1.786034107208252,
      "learning_rate": 3.6009920222439734e-06,
      "loss": 0.5257,
      "step": 172860
    },
    {
      "epoch": 70.76,
      "grad_norm": 2.4178919792175293,
      "learning_rate": 3.600858144099268e-06,
      "loss": 0.5306,
      "step": 172870
    },
    {
      "epoch": 70.77,
      "grad_norm": 2.1942923069000244,
      "learning_rate": 3.6007242601959446e-06,
      "loss": 0.5186,
      "step": 172880
    },
    {
      "epoch": 70.77,
      "grad_norm": 1.678991675376892,
      "learning_rate": 3.600590370534617e-06,
      "loss": 0.5173,
      "step": 172890
    },
    {
      "epoch": 70.77,
      "grad_norm": 1.6944676637649536,
      "learning_rate": 3.6004564751158977e-06,
      "loss": 0.5324,
      "step": 172900
    },
    {
      "epoch": 70.78,
      "grad_norm": 1.5906174182891846,
      "learning_rate": 3.6003225739404015e-06,
      "loss": 0.498,
      "step": 172910
    },
    {
      "epoch": 70.78,
      "grad_norm": 2.332313060760498,
      "learning_rate": 3.60018866700874e-06,
      "loss": 0.5028,
      "step": 172920
    },
    {
      "epoch": 70.79,
      "grad_norm": 1.9555046558380127,
      "learning_rate": 3.6000547543215287e-06,
      "loss": 0.5173,
      "step": 172930
    },
    {
      "epoch": 70.79,
      "grad_norm": 2.3173086643218994,
      "learning_rate": 3.599920835879379e-06,
      "loss": 0.5204,
      "step": 172940
    },
    {
      "epoch": 70.79,
      "grad_norm": 1.7702933549880981,
      "learning_rate": 3.599786911682906e-06,
      "loss": 0.513,
      "step": 172950
    },
    {
      "epoch": 70.8,
      "grad_norm": 1.8437227010726929,
      "learning_rate": 3.5996529817327212e-06,
      "loss": 0.5084,
      "step": 172960
    },
    {
      "epoch": 70.8,
      "grad_norm": 1.6298993825912476,
      "learning_rate": 3.5995190460294406e-06,
      "loss": 0.5214,
      "step": 172970
    },
    {
      "epoch": 70.81,
      "grad_norm": 2.401958703994751,
      "learning_rate": 3.5993851045736755e-06,
      "loss": 0.5151,
      "step": 172980
    },
    {
      "epoch": 70.81,
      "grad_norm": 1.814254641532898,
      "learning_rate": 3.5992511573660416e-06,
      "loss": 0.5207,
      "step": 172990
    },
    {
      "epoch": 70.81,
      "grad_norm": 2.1454570293426514,
      "learning_rate": 3.5991172044071502e-06,
      "loss": 0.5159,
      "step": 173000
    },
    {
      "epoch": 70.82,
      "grad_norm": 2.206146478652954,
      "learning_rate": 3.5989832456976166e-06,
      "loss": 0.523,
      "step": 173010
    },
    {
      "epoch": 70.82,
      "grad_norm": 1.8050923347473145,
      "learning_rate": 3.5988492812380536e-06,
      "loss": 0.5084,
      "step": 173020
    },
    {
      "epoch": 70.83,
      "grad_norm": 1.6057446002960205,
      "learning_rate": 3.5987153110290754e-06,
      "loss": 0.5205,
      "step": 173030
    },
    {
      "epoch": 70.83,
      "grad_norm": 1.6986708641052246,
      "learning_rate": 3.5985813350712955e-06,
      "loss": 0.5159,
      "step": 173040
    },
    {
      "epoch": 70.84,
      "grad_norm": 2.316746950149536,
      "learning_rate": 3.5984473533653277e-06,
      "loss": 0.5304,
      "step": 173050
    },
    {
      "epoch": 70.84,
      "grad_norm": 1.7859352827072144,
      "learning_rate": 3.5983133659117852e-06,
      "loss": 0.5112,
      "step": 173060
    },
    {
      "epoch": 70.84,
      "grad_norm": 2.0372390747070312,
      "learning_rate": 3.598179372711282e-06,
      "loss": 0.502,
      "step": 173070
    },
    {
      "epoch": 70.85,
      "grad_norm": 1.8484961986541748,
      "learning_rate": 3.5980453737644335e-06,
      "loss": 0.5136,
      "step": 173080
    },
    {
      "epoch": 70.85,
      "grad_norm": 1.909478783607483,
      "learning_rate": 3.597911369071851e-06,
      "loss": 0.5024,
      "step": 173090
    },
    {
      "epoch": 70.86,
      "grad_norm": 1.5429025888442993,
      "learning_rate": 3.5977773586341506e-06,
      "loss": 0.5073,
      "step": 173100
    },
    {
      "epoch": 70.86,
      "grad_norm": 1.621517300605774,
      "learning_rate": 3.5976433424519437e-06,
      "loss": 0.5187,
      "step": 173110
    },
    {
      "epoch": 70.86,
      "grad_norm": 2.051815986633301,
      "learning_rate": 3.597509320525846e-06,
      "loss": 0.5263,
      "step": 173120
    },
    {
      "epoch": 70.87,
      "grad_norm": 2.03029727935791,
      "learning_rate": 3.5973752928564704e-06,
      "loss": 0.513,
      "step": 173130
    },
    {
      "epoch": 70.87,
      "grad_norm": 1.9333142042160034,
      "learning_rate": 3.5972412594444326e-06,
      "loss": 0.5112,
      "step": 173140
    },
    {
      "epoch": 70.88,
      "grad_norm": 2.976998805999756,
      "learning_rate": 3.5971072202903453e-06,
      "loss": 0.5069,
      "step": 173150
    },
    {
      "epoch": 70.88,
      "grad_norm": 2.2782602310180664,
      "learning_rate": 3.596973175394823e-06,
      "loss": 0.5095,
      "step": 173160
    },
    {
      "epoch": 70.88,
      "grad_norm": 2.039158582687378,
      "learning_rate": 3.5968391247584787e-06,
      "loss": 0.5221,
      "step": 173170
    },
    {
      "epoch": 70.89,
      "grad_norm": 1.913437843322754,
      "learning_rate": 3.596705068381927e-06,
      "loss": 0.5357,
      "step": 173180
    },
    {
      "epoch": 70.89,
      "grad_norm": 1.9251269102096558,
      "learning_rate": 3.596571006265783e-06,
      "loss": 0.5301,
      "step": 173190
    },
    {
      "epoch": 70.9,
      "grad_norm": 1.837496280670166,
      "learning_rate": 3.5964369384106585e-06,
      "loss": 0.5308,
      "step": 173200
    },
    {
      "epoch": 70.9,
      "grad_norm": 1.8752411603927612,
      "learning_rate": 3.596302864817171e-06,
      "loss": 0.5043,
      "step": 173210
    },
    {
      "epoch": 70.9,
      "grad_norm": 3.2717485427856445,
      "learning_rate": 3.5961687854859324e-06,
      "loss": 0.4985,
      "step": 173220
    },
    {
      "epoch": 70.91,
      "grad_norm": 1.9336785078048706,
      "learning_rate": 3.596034700417557e-06,
      "loss": 0.505,
      "step": 173230
    },
    {
      "epoch": 70.91,
      "grad_norm": 2.233778953552246,
      "learning_rate": 3.5959006096126582e-06,
      "loss": 0.5077,
      "step": 173240
    },
    {
      "epoch": 70.92,
      "grad_norm": 1.9061450958251953,
      "learning_rate": 3.5957665130718524e-06,
      "loss": 0.4926,
      "step": 173250
    },
    {
      "epoch": 70.92,
      "grad_norm": 1.609270453453064,
      "learning_rate": 3.5956324107957535e-06,
      "loss": 0.5301,
      "step": 173260
    },
    {
      "epoch": 70.93,
      "grad_norm": 2.0019705295562744,
      "learning_rate": 3.5954983027849748e-06,
      "loss": 0.5293,
      "step": 173270
    },
    {
      "epoch": 70.93,
      "grad_norm": 2.327092409133911,
      "learning_rate": 3.5953641890401307e-06,
      "loss": 0.5147,
      "step": 173280
    },
    {
      "epoch": 70.93,
      "grad_norm": 1.924701452255249,
      "learning_rate": 3.5952300695618353e-06,
      "loss": 0.5113,
      "step": 173290
    },
    {
      "epoch": 70.94,
      "grad_norm": 2.2166216373443604,
      "learning_rate": 3.595095944350704e-06,
      "loss": 0.5195,
      "step": 173300
    },
    {
      "epoch": 70.94,
      "grad_norm": 2.112532138824463,
      "learning_rate": 3.594961813407351e-06,
      "loss": 0.5083,
      "step": 173310
    },
    {
      "epoch": 70.95,
      "grad_norm": 1.6576337814331055,
      "learning_rate": 3.5948276767323903e-06,
      "loss": 0.5087,
      "step": 173320
    },
    {
      "epoch": 70.95,
      "grad_norm": 1.956019639968872,
      "learning_rate": 3.5946935343264365e-06,
      "loss": 0.501,
      "step": 173330
    },
    {
      "epoch": 70.95,
      "grad_norm": 1.925607681274414,
      "learning_rate": 3.5945593861901044e-06,
      "loss": 0.4954,
      "step": 173340
    },
    {
      "epoch": 70.96,
      "grad_norm": 1.633307695388794,
      "learning_rate": 3.594425232324007e-06,
      "loss": 0.5079,
      "step": 173350
    },
    {
      "epoch": 70.96,
      "grad_norm": 1.5029468536376953,
      "learning_rate": 3.5942910727287616e-06,
      "loss": 0.495,
      "step": 173360
    },
    {
      "epoch": 70.97,
      "grad_norm": 1.7466332912445068,
      "learning_rate": 3.5941569074049803e-06,
      "loss": 0.5113,
      "step": 173370
    },
    {
      "epoch": 70.97,
      "grad_norm": 1.8133916854858398,
      "learning_rate": 3.5940227363532794e-06,
      "loss": 0.5246,
      "step": 173380
    },
    {
      "epoch": 70.97,
      "grad_norm": 2.206373453140259,
      "learning_rate": 3.5938885595742715e-06,
      "loss": 0.5193,
      "step": 173390
    },
    {
      "epoch": 70.98,
      "grad_norm": 2.0303192138671875,
      "learning_rate": 3.593754377068573e-06,
      "loss": 0.5145,
      "step": 173400
    },
    {
      "epoch": 70.98,
      "grad_norm": 2.149815559387207,
      "learning_rate": 3.5936201888367984e-06,
      "loss": 0.4987,
      "step": 173410
    },
    {
      "epoch": 70.99,
      "grad_norm": 2.2854111194610596,
      "learning_rate": 3.5934859948795613e-06,
      "loss": 0.52,
      "step": 173420
    },
    {
      "epoch": 70.99,
      "grad_norm": 1.9091823101043701,
      "learning_rate": 3.593351795197478e-06,
      "loss": 0.5455,
      "step": 173430
    },
    {
      "epoch": 70.99,
      "grad_norm": 2.2806074619293213,
      "learning_rate": 3.5932175897911617e-06,
      "loss": 0.5436,
      "step": 173440
    },
    {
      "epoch": 71.0,
      "grad_norm": 1.7222899198532104,
      "learning_rate": 3.593083378661229e-06,
      "loss": 0.5095,
      "step": 173450
    },
    {
      "epoch": 71.0,
      "eval_loss": 0.5137829184532166,
      "eval_runtime": 52.2999,
      "eval_samples_per_second": 65.947,
      "eval_steps_per_second": 8.26,
      "step": 173453
    },
    {
      "epoch": 71.0,
      "grad_norm": 3.087026596069336,
      "learning_rate": 3.592949161808293e-06,
      "loss": 0.5199,
      "step": 173460
    },
    {
      "epoch": 71.01,
      "grad_norm": 2.1763687133789062,
      "learning_rate": 3.592814939232968e-06,
      "loss": 0.4969,
      "step": 173470
    },
    {
      "epoch": 71.01,
      "grad_norm": 1.875874638557434,
      "learning_rate": 3.592680710935871e-06,
      "loss": 0.5145,
      "step": 173480
    },
    {
      "epoch": 71.02,
      "grad_norm": 2.0256927013397217,
      "learning_rate": 3.5925464769176157e-06,
      "loss": 0.4944,
      "step": 173490
    },
    {
      "epoch": 71.02,
      "grad_norm": 1.6964994668960571,
      "learning_rate": 3.5924122371788175e-06,
      "loss": 0.5045,
      "step": 173500
    },
    {
      "epoch": 71.02,
      "grad_norm": 1.4706275463104248,
      "learning_rate": 3.5922779917200904e-06,
      "loss": 0.5093,
      "step": 173510
    },
    {
      "epoch": 71.03,
      "grad_norm": 1.7284878492355347,
      "learning_rate": 3.592143740542051e-06,
      "loss": 0.5071,
      "step": 173520
    },
    {
      "epoch": 71.03,
      "grad_norm": 2.146057367324829,
      "learning_rate": 3.5920094836453115e-06,
      "loss": 0.5064,
      "step": 173530
    },
    {
      "epoch": 71.04,
      "grad_norm": 2.208031177520752,
      "learning_rate": 3.5918752210304895e-06,
      "loss": 0.5059,
      "step": 173540
    },
    {
      "epoch": 71.04,
      "grad_norm": 1.7870739698410034,
      "learning_rate": 3.5917409526981993e-06,
      "loss": 0.5226,
      "step": 173550
    },
    {
      "epoch": 71.04,
      "grad_norm": 1.7364948987960815,
      "learning_rate": 3.5916066786490554e-06,
      "loss": 0.5149,
      "step": 173560
    },
    {
      "epoch": 71.05,
      "grad_norm": 1.9529255628585815,
      "learning_rate": 3.5914723988836738e-06,
      "loss": 0.5126,
      "step": 173570
    },
    {
      "epoch": 71.05,
      "grad_norm": 2.278374671936035,
      "learning_rate": 3.5913381134026688e-06,
      "loss": 0.5084,
      "step": 173580
    },
    {
      "epoch": 71.06,
      "grad_norm": 1.6643363237380981,
      "learning_rate": 3.5912038222066563e-06,
      "loss": 0.5202,
      "step": 173590
    },
    {
      "epoch": 71.06,
      "grad_norm": 1.7713758945465088,
      "learning_rate": 3.5910695252962503e-06,
      "loss": 0.5058,
      "step": 173600
    },
    {
      "epoch": 71.06,
      "grad_norm": 2.2607409954071045,
      "learning_rate": 3.590935222672067e-06,
      "loss": 0.5109,
      "step": 173610
    },
    {
      "epoch": 71.07,
      "grad_norm": 2.4818949699401855,
      "learning_rate": 3.5908009143347217e-06,
      "loss": 0.4928,
      "step": 173620
    },
    {
      "epoch": 71.07,
      "grad_norm": 1.9316891431808472,
      "learning_rate": 3.590666600284829e-06,
      "loss": 0.5201,
      "step": 173630
    },
    {
      "epoch": 71.08,
      "grad_norm": 2.3409595489501953,
      "learning_rate": 3.5905322805230047e-06,
      "loss": 0.5239,
      "step": 173640
    },
    {
      "epoch": 71.08,
      "grad_norm": 2.0228078365325928,
      "learning_rate": 3.5903979550498634e-06,
      "loss": 0.4921,
      "step": 173650
    },
    {
      "epoch": 71.08,
      "grad_norm": 1.3877507448196411,
      "learning_rate": 3.5902636238660213e-06,
      "loss": 0.4991,
      "step": 173660
    },
    {
      "epoch": 71.09,
      "grad_norm": 1.5929336547851562,
      "learning_rate": 3.590129286972093e-06,
      "loss": 0.5148,
      "step": 173670
    },
    {
      "epoch": 71.09,
      "grad_norm": 2.1648447513580322,
      "learning_rate": 3.5899949443686946e-06,
      "loss": 0.4975,
      "step": 173680
    },
    {
      "epoch": 71.1,
      "grad_norm": 2.0655624866485596,
      "learning_rate": 3.5898605960564407e-06,
      "loss": 0.519,
      "step": 173690
    },
    {
      "epoch": 71.1,
      "grad_norm": 2.062994956970215,
      "learning_rate": 3.5897262420359475e-06,
      "loss": 0.5005,
      "step": 173700
    },
    {
      "epoch": 71.11,
      "grad_norm": 2.487497329711914,
      "learning_rate": 3.5895918823078294e-06,
      "loss": 0.5206,
      "step": 173710
    },
    {
      "epoch": 71.11,
      "grad_norm": 2.6886918544769287,
      "learning_rate": 3.5894575168727033e-06,
      "loss": 0.5181,
      "step": 173720
    },
    {
      "epoch": 71.11,
      "grad_norm": 2.59141206741333,
      "learning_rate": 3.589323145731184e-06,
      "loss": 0.5223,
      "step": 173730
    },
    {
      "epoch": 71.12,
      "grad_norm": 1.7025175094604492,
      "learning_rate": 3.589188768883887e-06,
      "loss": 0.5186,
      "step": 173740
    },
    {
      "epoch": 71.12,
      "grad_norm": 2.2087504863739014,
      "learning_rate": 3.5890543863314287e-06,
      "loss": 0.4913,
      "step": 173750
    },
    {
      "epoch": 71.13,
      "grad_norm": 2.3370070457458496,
      "learning_rate": 3.5889199980744224e-06,
      "loss": 0.5359,
      "step": 173760
    },
    {
      "epoch": 71.13,
      "grad_norm": 1.6185047626495361,
      "learning_rate": 3.5887856041134856e-06,
      "loss": 0.5287,
      "step": 173770
    },
    {
      "epoch": 71.13,
      "grad_norm": 1.8714975118637085,
      "learning_rate": 3.5886512044492337e-06,
      "loss": 0.5078,
      "step": 173780
    },
    {
      "epoch": 71.14,
      "grad_norm": 1.9417617321014404,
      "learning_rate": 3.5885167990822817e-06,
      "loss": 0.5127,
      "step": 173790
    },
    {
      "epoch": 71.14,
      "grad_norm": 2.5377888679504395,
      "learning_rate": 3.588382388013246e-06,
      "loss": 0.5156,
      "step": 173800
    },
    {
      "epoch": 71.15,
      "grad_norm": 1.8306280374526978,
      "learning_rate": 3.588247971242743e-06,
      "loss": 0.5233,
      "step": 173810
    },
    {
      "epoch": 71.15,
      "grad_norm": 2.0477097034454346,
      "learning_rate": 3.5881135487713866e-06,
      "loss": 0.531,
      "step": 173820
    },
    {
      "epoch": 71.15,
      "grad_norm": 1.9636871814727783,
      "learning_rate": 3.5879791205997936e-06,
      "loss": 0.5312,
      "step": 173830
    },
    {
      "epoch": 71.16,
      "grad_norm": 1.7481279373168945,
      "learning_rate": 3.5878446867285795e-06,
      "loss": 0.5246,
      "step": 173840
    },
    {
      "epoch": 71.16,
      "grad_norm": 1.7877999544143677,
      "learning_rate": 3.5877102471583604e-06,
      "loss": 0.5027,
      "step": 173850
    },
    {
      "epoch": 71.17,
      "grad_norm": 1.4915335178375244,
      "learning_rate": 3.587575801889752e-06,
      "loss": 0.505,
      "step": 173860
    },
    {
      "epoch": 71.17,
      "grad_norm": 1.8789058923721313,
      "learning_rate": 3.5874413509233705e-06,
      "loss": 0.4996,
      "step": 173870
    },
    {
      "epoch": 71.17,
      "grad_norm": 2.3769500255584717,
      "learning_rate": 3.587306894259832e-06,
      "loss": 0.5126,
      "step": 173880
    },
    {
      "epoch": 71.18,
      "grad_norm": 2.136632204055786,
      "learning_rate": 3.587172431899751e-06,
      "loss": 0.5175,
      "step": 173890
    },
    {
      "epoch": 71.18,
      "grad_norm": 1.7636173963546753,
      "learning_rate": 3.587037963843745e-06,
      "loss": 0.5349,
      "step": 173900
    },
    {
      "epoch": 71.19,
      "grad_norm": 2.1990489959716797,
      "learning_rate": 3.586903490092429e-06,
      "loss": 0.5305,
      "step": 173910
    },
    {
      "epoch": 71.19,
      "grad_norm": 1.943557858467102,
      "learning_rate": 3.58676901064642e-06,
      "loss": 0.5299,
      "step": 173920
    },
    {
      "epoch": 71.2,
      "grad_norm": 2.113485336303711,
      "learning_rate": 3.5866345255063333e-06,
      "loss": 0.5009,
      "step": 173930
    },
    {
      "epoch": 71.2,
      "grad_norm": 1.7647106647491455,
      "learning_rate": 3.586500034672784e-06,
      "loss": 0.5251,
      "step": 173940
    },
    {
      "epoch": 71.2,
      "grad_norm": 1.6667368412017822,
      "learning_rate": 3.5863655381463904e-06,
      "loss": 0.5023,
      "step": 173950
    },
    {
      "epoch": 71.21,
      "grad_norm": 1.7312004566192627,
      "learning_rate": 3.5862310359277676e-06,
      "loss": 0.5203,
      "step": 173960
    },
    {
      "epoch": 71.21,
      "grad_norm": 2.3927371501922607,
      "learning_rate": 3.586096528017531e-06,
      "loss": 0.5092,
      "step": 173970
    },
    {
      "epoch": 71.22,
      "grad_norm": 2.2408711910247803,
      "learning_rate": 3.5859620144162973e-06,
      "loss": 0.5319,
      "step": 173980
    },
    {
      "epoch": 71.22,
      "grad_norm": 1.5939730405807495,
      "learning_rate": 3.585827495124683e-06,
      "loss": 0.4987,
      "step": 173990
    },
    {
      "epoch": 71.22,
      "grad_norm": 2.124218463897705,
      "learning_rate": 3.5856929701433035e-06,
      "loss": 0.5186,
      "step": 174000
    },
    {
      "epoch": 71.23,
      "grad_norm": 1.6600486040115356,
      "learning_rate": 3.5855584394727762e-06,
      "loss": 0.5257,
      "step": 174010
    },
    {
      "epoch": 71.23,
      "grad_norm": 2.1059858798980713,
      "learning_rate": 3.5854239031137167e-06,
      "loss": 0.5039,
      "step": 174020
    },
    {
      "epoch": 71.24,
      "grad_norm": 1.8866755962371826,
      "learning_rate": 3.585289361066742e-06,
      "loss": 0.524,
      "step": 174030
    },
    {
      "epoch": 71.24,
      "grad_norm": 1.6350626945495605,
      "learning_rate": 3.585154813332467e-06,
      "loss": 0.4942,
      "step": 174040
    },
    {
      "epoch": 71.24,
      "grad_norm": 2.234893560409546,
      "learning_rate": 3.585020259911508e-06,
      "loss": 0.5045,
      "step": 174050
    },
    {
      "epoch": 71.25,
      "grad_norm": 1.7663875818252563,
      "learning_rate": 3.584885700804483e-06,
      "loss": 0.5027,
      "step": 174060
    },
    {
      "epoch": 71.25,
      "grad_norm": 2.1836118698120117,
      "learning_rate": 3.584751136012008e-06,
      "loss": 0.5087,
      "step": 174070
    },
    {
      "epoch": 71.26,
      "grad_norm": 1.8201473951339722,
      "learning_rate": 3.5846165655346987e-06,
      "loss": 0.4976,
      "step": 174080
    },
    {
      "epoch": 71.26,
      "grad_norm": 2.031881809234619,
      "learning_rate": 3.5844819893731717e-06,
      "loss": 0.5185,
      "step": 174090
    },
    {
      "epoch": 71.26,
      "grad_norm": 2.236668586730957,
      "learning_rate": 3.5843474075280436e-06,
      "loss": 0.5207,
      "step": 174100
    },
    {
      "epoch": 71.27,
      "grad_norm": 1.6826053857803345,
      "learning_rate": 3.5842128199999312e-06,
      "loss": 0.4942,
      "step": 174110
    },
    {
      "epoch": 71.27,
      "grad_norm": 2.038055658340454,
      "learning_rate": 3.5840782267894505e-06,
      "loss": 0.4985,
      "step": 174120
    },
    {
      "epoch": 71.28,
      "grad_norm": 1.9942944049835205,
      "learning_rate": 3.5839436278972185e-06,
      "loss": 0.5081,
      "step": 174130
    },
    {
      "epoch": 71.28,
      "grad_norm": 1.5831477642059326,
      "learning_rate": 3.583809023323851e-06,
      "loss": 0.507,
      "step": 174140
    },
    {
      "epoch": 71.29,
      "grad_norm": 2.0989882946014404,
      "learning_rate": 3.5836744130699644e-06,
      "loss": 0.5252,
      "step": 174150
    },
    {
      "epoch": 71.29,
      "grad_norm": 2.027679443359375,
      "learning_rate": 3.5835397971361774e-06,
      "loss": 0.5186,
      "step": 174160
    },
    {
      "epoch": 71.29,
      "grad_norm": 2.6466853618621826,
      "learning_rate": 3.583405175523106e-06,
      "loss": 0.5111,
      "step": 174170
    },
    {
      "epoch": 71.3,
      "grad_norm": 1.4806411266326904,
      "learning_rate": 3.5832705482313653e-06,
      "loss": 0.5063,
      "step": 174180
    },
    {
      "epoch": 71.3,
      "grad_norm": 1.537156105041504,
      "learning_rate": 3.5831359152615727e-06,
      "loss": 0.5166,
      "step": 174190
    },
    {
      "epoch": 71.31,
      "grad_norm": 1.7925784587860107,
      "learning_rate": 3.583001276614345e-06,
      "loss": 0.4979,
      "step": 174200
    },
    {
      "epoch": 71.31,
      "grad_norm": 1.8398619890213013,
      "learning_rate": 3.5828666322903e-06,
      "loss": 0.5181,
      "step": 174210
    },
    {
      "epoch": 71.31,
      "grad_norm": 2.1030325889587402,
      "learning_rate": 3.582731982290053e-06,
      "loss": 0.5227,
      "step": 174220
    },
    {
      "epoch": 71.32,
      "grad_norm": 1.8998501300811768,
      "learning_rate": 3.5825973266142218e-06,
      "loss": 0.5069,
      "step": 174230
    },
    {
      "epoch": 71.32,
      "grad_norm": 1.6837736368179321,
      "learning_rate": 3.582462665263422e-06,
      "loss": 0.5078,
      "step": 174240
    },
    {
      "epoch": 71.33,
      "grad_norm": 1.8735827207565308,
      "learning_rate": 3.582327998238272e-06,
      "loss": 0.526,
      "step": 174250
    },
    {
      "epoch": 71.33,
      "grad_norm": 1.7679470777511597,
      "learning_rate": 3.5821933255393885e-06,
      "loss": 0.509,
      "step": 174260
    },
    {
      "epoch": 71.33,
      "grad_norm": 2.2503156661987305,
      "learning_rate": 3.5820586471673877e-06,
      "loss": 0.5219,
      "step": 174270
    },
    {
      "epoch": 71.34,
      "grad_norm": 1.7880799770355225,
      "learning_rate": 3.581923963122886e-06,
      "loss": 0.5114,
      "step": 174280
    },
    {
      "epoch": 71.34,
      "grad_norm": 1.6037079095840454,
      "learning_rate": 3.581789273406502e-06,
      "loss": 0.5129,
      "step": 174290
    },
    {
      "epoch": 71.35,
      "grad_norm": 1.9347645044326782,
      "learning_rate": 3.581654578018851e-06,
      "loss": 0.5246,
      "step": 174300
    },
    {
      "epoch": 71.35,
      "grad_norm": 1.7122406959533691,
      "learning_rate": 3.5815198769605515e-06,
      "loss": 0.5007,
      "step": 174310
    },
    {
      "epoch": 71.35,
      "grad_norm": 2.2618234157562256,
      "learning_rate": 3.5813851702322195e-06,
      "loss": 0.5108,
      "step": 174320
    },
    {
      "epoch": 71.36,
      "grad_norm": 2.0127627849578857,
      "learning_rate": 3.581250457834472e-06,
      "loss": 0.5263,
      "step": 174330
    },
    {
      "epoch": 71.36,
      "grad_norm": 2.561680555343628,
      "learning_rate": 3.581115739767927e-06,
      "loss": 0.5007,
      "step": 174340
    },
    {
      "epoch": 71.37,
      "grad_norm": 1.7048282623291016,
      "learning_rate": 3.5809810160332014e-06,
      "loss": 0.5271,
      "step": 174350
    },
    {
      "epoch": 71.37,
      "grad_norm": 1.8662551641464233,
      "learning_rate": 3.580846286630913e-06,
      "loss": 0.5295,
      "step": 174360
    },
    {
      "epoch": 71.38,
      "grad_norm": 1.541641354560852,
      "learning_rate": 3.580711551561677e-06,
      "loss": 0.5164,
      "step": 174370
    },
    {
      "epoch": 71.38,
      "grad_norm": 1.7387380599975586,
      "learning_rate": 3.5805768108261113e-06,
      "loss": 0.5019,
      "step": 174380
    },
    {
      "epoch": 71.38,
      "grad_norm": 1.815083622932434,
      "learning_rate": 3.5804420644248346e-06,
      "loss": 0.5136,
      "step": 174390
    },
    {
      "epoch": 71.39,
      "grad_norm": 2.0608630180358887,
      "learning_rate": 3.5803073123584628e-06,
      "loss": 0.5089,
      "step": 174400
    },
    {
      "epoch": 71.39,
      "grad_norm": 2.034149169921875,
      "learning_rate": 3.580172554627613e-06,
      "loss": 0.4918,
      "step": 174410
    },
    {
      "epoch": 71.4,
      "grad_norm": 2.0457923412323,
      "learning_rate": 3.580037791232903e-06,
      "loss": 0.5088,
      "step": 174420
    },
    {
      "epoch": 71.4,
      "grad_norm": 1.660903811454773,
      "learning_rate": 3.57990302217495e-06,
      "loss": 0.5273,
      "step": 174430
    },
    {
      "epoch": 71.4,
      "grad_norm": 1.6182690858840942,
      "learning_rate": 3.579768247454372e-06,
      "loss": 0.5217,
      "step": 174440
    },
    {
      "epoch": 71.41,
      "grad_norm": 2.1480870246887207,
      "learning_rate": 3.5796334670717847e-06,
      "loss": 0.5272,
      "step": 174450
    },
    {
      "epoch": 71.41,
      "grad_norm": 1.9073779582977295,
      "learning_rate": 3.5794986810278085e-06,
      "loss": 0.5009,
      "step": 174460
    },
    {
      "epoch": 71.42,
      "grad_norm": 1.7897999286651611,
      "learning_rate": 3.579363889323058e-06,
      "loss": 0.5216,
      "step": 174470
    },
    {
      "epoch": 71.42,
      "grad_norm": 2.204073429107666,
      "learning_rate": 3.579229091958151e-06,
      "loss": 0.5042,
      "step": 174480
    },
    {
      "epoch": 71.42,
      "grad_norm": 1.786015272140503,
      "learning_rate": 3.5790942889337062e-06,
      "loss": 0.5294,
      "step": 174490
    },
    {
      "epoch": 71.43,
      "grad_norm": 2.1387569904327393,
      "learning_rate": 3.578959480250341e-06,
      "loss": 0.5107,
      "step": 174500
    },
    {
      "epoch": 71.43,
      "grad_norm": 2.4214277267456055,
      "learning_rate": 3.5788246659086715e-06,
      "loss": 0.5145,
      "step": 174510
    },
    {
      "epoch": 71.44,
      "grad_norm": 2.082735538482666,
      "learning_rate": 3.5786898459093166e-06,
      "loss": 0.5425,
      "step": 174520
    },
    {
      "epoch": 71.44,
      "grad_norm": 1.821678638458252,
      "learning_rate": 3.5785550202528938e-06,
      "loss": 0.5155,
      "step": 174530
    },
    {
      "epoch": 71.44,
      "grad_norm": 2.124225378036499,
      "learning_rate": 3.578420188940021e-06,
      "loss": 0.5213,
      "step": 174540
    },
    {
      "epoch": 71.45,
      "grad_norm": 1.924385905265808,
      "learning_rate": 3.578285351971314e-06,
      "loss": 0.5163,
      "step": 174550
    },
    {
      "epoch": 71.45,
      "grad_norm": 1.7501142024993896,
      "learning_rate": 3.578150509347392e-06,
      "loss": 0.522,
      "step": 174560
    },
    {
      "epoch": 71.46,
      "grad_norm": 2.75797176361084,
      "learning_rate": 3.5780156610688734e-06,
      "loss": 0.5129,
      "step": 174570
    },
    {
      "epoch": 71.46,
      "grad_norm": 2.6224372386932373,
      "learning_rate": 3.577880807136374e-06,
      "loss": 0.5047,
      "step": 174580
    },
    {
      "epoch": 71.47,
      "grad_norm": 1.5511270761489868,
      "learning_rate": 3.577745947550512e-06,
      "loss": 0.507,
      "step": 174590
    },
    {
      "epoch": 71.47,
      "grad_norm": 1.8100028038024902,
      "learning_rate": 3.577611082311907e-06,
      "loss": 0.5177,
      "step": 174600
    },
    {
      "epoch": 71.47,
      "grad_norm": 1.5155272483825684,
      "learning_rate": 3.5774762114211755e-06,
      "loss": 0.5093,
      "step": 174610
    },
    {
      "epoch": 71.48,
      "grad_norm": 1.8886725902557373,
      "learning_rate": 3.5773413348789346e-06,
      "loss": 0.5198,
      "step": 174620
    },
    {
      "epoch": 71.48,
      "grad_norm": 2.422222375869751,
      "learning_rate": 3.5772064526858027e-06,
      "loss": 0.5133,
      "step": 174630
    },
    {
      "epoch": 71.49,
      "grad_norm": 1.7495664358139038,
      "learning_rate": 3.5770715648423978e-06,
      "loss": 0.5224,
      "step": 174640
    },
    {
      "epoch": 71.49,
      "grad_norm": 1.6211025714874268,
      "learning_rate": 3.576936671349338e-06,
      "loss": 0.5147,
      "step": 174650
    },
    {
      "epoch": 71.49,
      "grad_norm": 1.7966593503952026,
      "learning_rate": 3.5768017722072417e-06,
      "loss": 0.5172,
      "step": 174660
    },
    {
      "epoch": 71.5,
      "grad_norm": 1.6932518482208252,
      "learning_rate": 3.576666867416725e-06,
      "loss": 0.515,
      "step": 174670
    },
    {
      "epoch": 71.5,
      "grad_norm": 2.1963961124420166,
      "learning_rate": 3.576531956978408e-06,
      "loss": 0.523,
      "step": 174680
    },
    {
      "epoch": 71.51,
      "grad_norm": 1.5773640871047974,
      "learning_rate": 3.576397040892908e-06,
      "loss": 0.4916,
      "step": 174690
    },
    {
      "epoch": 71.51,
      "grad_norm": 2.204252004623413,
      "learning_rate": 3.5762621191608417e-06,
      "loss": 0.4975,
      "step": 174700
    },
    {
      "epoch": 71.51,
      "grad_norm": 2.1074137687683105,
      "learning_rate": 3.5761271917828287e-06,
      "loss": 0.5188,
      "step": 174710
    },
    {
      "epoch": 71.52,
      "grad_norm": 1.781429648399353,
      "learning_rate": 3.575992258759487e-06,
      "loss": 0.5244,
      "step": 174720
    },
    {
      "epoch": 71.52,
      "grad_norm": 1.9419289827346802,
      "learning_rate": 3.575857320091434e-06,
      "loss": 0.5136,
      "step": 174730
    },
    {
      "epoch": 71.53,
      "grad_norm": 1.7157130241394043,
      "learning_rate": 3.5757223757792884e-06,
      "loss": 0.5102,
      "step": 174740
    },
    {
      "epoch": 71.53,
      "grad_norm": 2.007230520248413,
      "learning_rate": 3.5755874258236685e-06,
      "loss": 0.5332,
      "step": 174750
    },
    {
      "epoch": 71.53,
      "grad_norm": 1.7048348188400269,
      "learning_rate": 3.575452470225192e-06,
      "loss": 0.5314,
      "step": 174760
    },
    {
      "epoch": 71.54,
      "grad_norm": 1.7273293733596802,
      "learning_rate": 3.5753175089844766e-06,
      "loss": 0.5365,
      "step": 174770
    },
    {
      "epoch": 71.54,
      "grad_norm": 1.5168628692626953,
      "learning_rate": 3.5751825421021417e-06,
      "loss": 0.5414,
      "step": 174780
    },
    {
      "epoch": 71.55,
      "grad_norm": 1.5518547296524048,
      "learning_rate": 3.5750475695788045e-06,
      "loss": 0.5019,
      "step": 174790
    },
    {
      "epoch": 71.55,
      "grad_norm": 2.622097969055176,
      "learning_rate": 3.5749125914150833e-06,
      "loss": 0.5242,
      "step": 174800
    },
    {
      "epoch": 71.56,
      "grad_norm": 2.106936454772949,
      "learning_rate": 3.5747776076115983e-06,
      "loss": 0.5064,
      "step": 174810
    },
    {
      "epoch": 71.56,
      "grad_norm": 1.9325834512710571,
      "learning_rate": 3.5746426181689664e-06,
      "loss": 0.5025,
      "step": 174820
    },
    {
      "epoch": 71.56,
      "grad_norm": 2.163973093032837,
      "learning_rate": 3.5745076230878056e-06,
      "loss": 0.4991,
      "step": 174830
    },
    {
      "epoch": 71.57,
      "grad_norm": 2.176790952682495,
      "learning_rate": 3.5743726223687347e-06,
      "loss": 0.5168,
      "step": 174840
    },
    {
      "epoch": 71.57,
      "grad_norm": 1.6668533086776733,
      "learning_rate": 3.5742376160123723e-06,
      "loss": 0.5168,
      "step": 174850
    },
    {
      "epoch": 71.58,
      "grad_norm": 1.7160903215408325,
      "learning_rate": 3.574102604019336e-06,
      "loss": 0.5146,
      "step": 174860
    },
    {
      "epoch": 71.58,
      "grad_norm": 1.701614260673523,
      "learning_rate": 3.5739675863902462e-06,
      "loss": 0.5105,
      "step": 174870
    },
    {
      "epoch": 71.58,
      "grad_norm": 1.920664668083191,
      "learning_rate": 3.5738325631257192e-06,
      "loss": 0.5085,
      "step": 174880
    },
    {
      "epoch": 71.59,
      "grad_norm": 2.282538652420044,
      "learning_rate": 3.573697534226375e-06,
      "loss": 0.5143,
      "step": 174890
    },
    {
      "epoch": 71.59,
      "grad_norm": 2.658278226852417,
      "learning_rate": 3.5735624996928314e-06,
      "loss": 0.503,
      "step": 174900
    },
    {
      "epoch": 71.6,
      "grad_norm": 1.7156859636306763,
      "learning_rate": 3.573427459525707e-06,
      "loss": 0.5318,
      "step": 174910
    },
    {
      "epoch": 71.6,
      "grad_norm": 1.9686206579208374,
      "learning_rate": 3.573292413725621e-06,
      "loss": 0.5077,
      "step": 174920
    },
    {
      "epoch": 71.6,
      "grad_norm": 2.3099191188812256,
      "learning_rate": 3.5731573622931916e-06,
      "loss": 0.5184,
      "step": 174930
    },
    {
      "epoch": 71.61,
      "grad_norm": 2.868741273880005,
      "learning_rate": 3.5730223052290375e-06,
      "loss": 0.5184,
      "step": 174940
    },
    {
      "epoch": 71.61,
      "grad_norm": 1.7671544551849365,
      "learning_rate": 3.572887242533777e-06,
      "loss": 0.5038,
      "step": 174950
    },
    {
      "epoch": 71.62,
      "grad_norm": 1.7305997610092163,
      "learning_rate": 3.5727521742080295e-06,
      "loss": 0.5247,
      "step": 174960
    },
    {
      "epoch": 71.62,
      "grad_norm": 1.7562085390090942,
      "learning_rate": 3.572617100252413e-06,
      "loss": 0.5015,
      "step": 174970
    },
    {
      "epoch": 71.63,
      "grad_norm": 2.127716302871704,
      "learning_rate": 3.572482020667547e-06,
      "loss": 0.5312,
      "step": 174980
    },
    {
      "epoch": 71.63,
      "grad_norm": 1.812743902206421,
      "learning_rate": 3.5723469354540492e-06,
      "loss": 0.5235,
      "step": 174990
    },
    {
      "epoch": 71.63,
      "grad_norm": 1.7083741426467896,
      "learning_rate": 3.57221184461254e-06,
      "loss": 0.5066,
      "step": 175000
    },
    {
      "epoch": 71.64,
      "grad_norm": 1.5285757780075073,
      "learning_rate": 3.572076748143637e-06,
      "loss": 0.4971,
      "step": 175010
    },
    {
      "epoch": 71.64,
      "grad_norm": 2.8019180297851562,
      "learning_rate": 3.571941646047959e-06,
      "loss": 0.4953,
      "step": 175020
    },
    {
      "epoch": 71.65,
      "grad_norm": 1.8033336400985718,
      "learning_rate": 3.571806538326126e-06,
      "loss": 0.5153,
      "step": 175030
    },
    {
      "epoch": 71.65,
      "grad_norm": 1.757455587387085,
      "learning_rate": 3.5716714249787565e-06,
      "loss": 0.4959,
      "step": 175040
    },
    {
      "epoch": 71.65,
      "grad_norm": 1.8182259798049927,
      "learning_rate": 3.571536306006468e-06,
      "loss": 0.5048,
      "step": 175050
    },
    {
      "epoch": 71.66,
      "grad_norm": 2.3441364765167236,
      "learning_rate": 3.5714011814098815e-06,
      "loss": 0.5261,
      "step": 175060
    },
    {
      "epoch": 71.66,
      "grad_norm": 1.7252485752105713,
      "learning_rate": 3.571266051189614e-06,
      "loss": 0.5,
      "step": 175070
    },
    {
      "epoch": 71.67,
      "grad_norm": 1.6498630046844482,
      "learning_rate": 3.571130915346286e-06,
      "loss": 0.535,
      "step": 175080
    },
    {
      "epoch": 71.67,
      "grad_norm": 1.8165826797485352,
      "learning_rate": 3.570995773880516e-06,
      "loss": 0.5361,
      "step": 175090
    },
    {
      "epoch": 71.67,
      "grad_norm": 1.9207240343093872,
      "learning_rate": 3.5708606267929237e-06,
      "loss": 0.5204,
      "step": 175100
    },
    {
      "epoch": 71.68,
      "grad_norm": 1.5083160400390625,
      "learning_rate": 3.5707254740841276e-06,
      "loss": 0.5353,
      "step": 175110
    },
    {
      "epoch": 71.68,
      "grad_norm": 1.6959972381591797,
      "learning_rate": 3.570590315754747e-06,
      "loss": 0.5271,
      "step": 175120
    },
    {
      "epoch": 71.69,
      "grad_norm": 1.68105149269104,
      "learning_rate": 3.5704551518054004e-06,
      "loss": 0.4883,
      "step": 175130
    },
    {
      "epoch": 71.69,
      "grad_norm": 2.144681930541992,
      "learning_rate": 3.570319982236708e-06,
      "loss": 0.5228,
      "step": 175140
    },
    {
      "epoch": 71.69,
      "grad_norm": 1.8872309923171997,
      "learning_rate": 3.5701848070492878e-06,
      "loss": 0.5096,
      "step": 175150
    },
    {
      "epoch": 71.7,
      "grad_norm": 2.125620126724243,
      "learning_rate": 3.5700496262437603e-06,
      "loss": 0.4887,
      "step": 175160
    },
    {
      "epoch": 71.7,
      "grad_norm": 1.663162112236023,
      "learning_rate": 3.569914439820744e-06,
      "loss": 0.5002,
      "step": 175170
    },
    {
      "epoch": 71.71,
      "grad_norm": 2.3412013053894043,
      "learning_rate": 3.569779247780858e-06,
      "loss": 0.5155,
      "step": 175180
    },
    {
      "epoch": 71.71,
      "grad_norm": 1.8302414417266846,
      "learning_rate": 3.5696440501247222e-06,
      "loss": 0.5207,
      "step": 175190
    },
    {
      "epoch": 71.72,
      "grad_norm": 1.7511208057403564,
      "learning_rate": 3.5695088468529554e-06,
      "loss": 0.5297,
      "step": 175200
    },
    {
      "epoch": 71.72,
      "grad_norm": 1.60996675491333,
      "learning_rate": 3.5693736379661767e-06,
      "loss": 0.5109,
      "step": 175210
    },
    {
      "epoch": 71.72,
      "grad_norm": 1.5162571668624878,
      "learning_rate": 3.5692384234650065e-06,
      "loss": 0.5241,
      "step": 175220
    },
    {
      "epoch": 71.73,
      "grad_norm": 1.6307917833328247,
      "learning_rate": 3.5691032033500636e-06,
      "loss": 0.4979,
      "step": 175230
    },
    {
      "epoch": 71.73,
      "grad_norm": 2.2213187217712402,
      "learning_rate": 3.5689679776219676e-06,
      "loss": 0.5125,
      "step": 175240
    },
    {
      "epoch": 71.74,
      "grad_norm": 1.9521739482879639,
      "learning_rate": 3.5688327462813374e-06,
      "loss": 0.5336,
      "step": 175250
    },
    {
      "epoch": 71.74,
      "grad_norm": 1.6828515529632568,
      "learning_rate": 3.5686975093287936e-06,
      "loss": 0.5274,
      "step": 175260
    },
    {
      "epoch": 71.74,
      "grad_norm": 1.890272855758667,
      "learning_rate": 3.5685622667649548e-06,
      "loss": 0.5181,
      "step": 175270
    },
    {
      "epoch": 71.75,
      "grad_norm": 1.6035982370376587,
      "learning_rate": 3.5684270185904395e-06,
      "loss": 0.5099,
      "step": 175280
    },
    {
      "epoch": 71.75,
      "grad_norm": 2.0598134994506836,
      "learning_rate": 3.5682917648058692e-06,
      "loss": 0.5279,
      "step": 175290
    },
    {
      "epoch": 71.76,
      "grad_norm": 1.781844139099121,
      "learning_rate": 3.5681565054118626e-06,
      "loss": 0.5111,
      "step": 175300
    },
    {
      "epoch": 71.76,
      "grad_norm": 2.2002148628234863,
      "learning_rate": 3.56802124040904e-06,
      "loss": 0.5196,
      "step": 175310
    },
    {
      "epoch": 71.76,
      "grad_norm": 1.8048726320266724,
      "learning_rate": 3.5678859697980202e-06,
      "loss": 0.5234,
      "step": 175320
    },
    {
      "epoch": 71.77,
      "grad_norm": 2.120936155319214,
      "learning_rate": 3.567750693579423e-06,
      "loss": 0.4918,
      "step": 175330
    },
    {
      "epoch": 71.77,
      "grad_norm": 1.8402751684188843,
      "learning_rate": 3.5676154117538687e-06,
      "loss": 0.5132,
      "step": 175340
    },
    {
      "epoch": 71.78,
      "grad_norm": 2.3214051723480225,
      "learning_rate": 3.567480124321976e-06,
      "loss": 0.5215,
      "step": 175350
    },
    {
      "epoch": 71.78,
      "grad_norm": 1.595168113708496,
      "learning_rate": 3.5673448312843644e-06,
      "loss": 0.5146,
      "step": 175360
    },
    {
      "epoch": 71.78,
      "grad_norm": 1.876544713973999,
      "learning_rate": 3.567209532641655e-06,
      "loss": 0.5238,
      "step": 175370
    },
    {
      "epoch": 71.79,
      "grad_norm": 1.7734501361846924,
      "learning_rate": 3.5670742283944664e-06,
      "loss": 0.4957,
      "step": 175380
    },
    {
      "epoch": 71.79,
      "grad_norm": 1.9137849807739258,
      "learning_rate": 3.56693891854342e-06,
      "loss": 0.4987,
      "step": 175390
    },
    {
      "epoch": 71.8,
      "grad_norm": 2.115408182144165,
      "learning_rate": 3.5668036030891345e-06,
      "loss": 0.5007,
      "step": 175400
    },
    {
      "epoch": 71.8,
      "grad_norm": 1.6007535457611084,
      "learning_rate": 3.5666682820322296e-06,
      "loss": 0.4946,
      "step": 175410
    },
    {
      "epoch": 71.81,
      "grad_norm": 2.0196988582611084,
      "learning_rate": 3.5665329553733254e-06,
      "loss": 0.5079,
      "step": 175420
    },
    {
      "epoch": 71.81,
      "grad_norm": 1.5272825956344604,
      "learning_rate": 3.566397623113042e-06,
      "loss": 0.4919,
      "step": 175430
    },
    {
      "epoch": 71.81,
      "grad_norm": 1.4807097911834717,
      "learning_rate": 3.566262285251999e-06,
      "loss": 0.5124,
      "step": 175440
    },
    {
      "epoch": 71.82,
      "grad_norm": 2.043401002883911,
      "learning_rate": 3.566126941790816e-06,
      "loss": 0.499,
      "step": 175450
    },
    {
      "epoch": 71.82,
      "grad_norm": 2.262449264526367,
      "learning_rate": 3.5659915927301143e-06,
      "loss": 0.5174,
      "step": 175460
    },
    {
      "epoch": 71.83,
      "grad_norm": 1.4609724283218384,
      "learning_rate": 3.565856238070513e-06,
      "loss": 0.5051,
      "step": 175470
    },
    {
      "epoch": 71.83,
      "grad_norm": 2.0263612270355225,
      "learning_rate": 3.5657208778126322e-06,
      "loss": 0.4977,
      "step": 175480
    },
    {
      "epoch": 71.83,
      "grad_norm": 4.24222469329834,
      "learning_rate": 3.565585511957092e-06,
      "loss": 0.5105,
      "step": 175490
    },
    {
      "epoch": 71.84,
      "grad_norm": 2.2006499767303467,
      "learning_rate": 3.565450140504513e-06,
      "loss": 0.4904,
      "step": 175500
    },
    {
      "epoch": 71.84,
      "grad_norm": 1.8950488567352295,
      "learning_rate": 3.5653147634555143e-06,
      "loss": 0.4848,
      "step": 175510
    },
    {
      "epoch": 71.85,
      "grad_norm": 2.618391513824463,
      "learning_rate": 3.5651793808107168e-06,
      "loss": 0.4972,
      "step": 175520
    },
    {
      "epoch": 71.85,
      "grad_norm": 1.792765736579895,
      "learning_rate": 3.5650439925707406e-06,
      "loss": 0.5266,
      "step": 175530
    },
    {
      "epoch": 71.85,
      "grad_norm": 2.2061142921447754,
      "learning_rate": 3.5649085987362055e-06,
      "loss": 0.5279,
      "step": 175540
    },
    {
      "epoch": 71.86,
      "grad_norm": 2.1014516353607178,
      "learning_rate": 3.5647731993077326e-06,
      "loss": 0.5406,
      "step": 175550
    },
    {
      "epoch": 71.86,
      "grad_norm": 2.1440834999084473,
      "learning_rate": 3.5646377942859403e-06,
      "loss": 0.5109,
      "step": 175560
    },
    {
      "epoch": 71.87,
      "grad_norm": 1.8492186069488525,
      "learning_rate": 3.5645023836714513e-06,
      "loss": 0.5109,
      "step": 175570
    },
    {
      "epoch": 71.87,
      "grad_norm": 2.438079357147217,
      "learning_rate": 3.5643669674648837e-06,
      "loss": 0.5328,
      "step": 175580
    },
    {
      "epoch": 71.87,
      "grad_norm": 1.554383397102356,
      "learning_rate": 3.5642315456668594e-06,
      "loss": 0.5142,
      "step": 175590
    },
    {
      "epoch": 71.88,
      "grad_norm": 1.6565170288085938,
      "learning_rate": 3.5640961182779983e-06,
      "loss": 0.5206,
      "step": 175600
    },
    {
      "epoch": 71.88,
      "grad_norm": 2.0021965503692627,
      "learning_rate": 3.56396068529892e-06,
      "loss": 0.5162,
      "step": 175610
    },
    {
      "epoch": 71.89,
      "grad_norm": 2.080357789993286,
      "learning_rate": 3.563825246730246e-06,
      "loss": 0.509,
      "step": 175620
    },
    {
      "epoch": 71.89,
      "grad_norm": 1.4753812551498413,
      "learning_rate": 3.5636898025725967e-06,
      "loss": 0.5102,
      "step": 175630
    },
    {
      "epoch": 71.9,
      "grad_norm": 2.4661946296691895,
      "learning_rate": 3.5635543528265916e-06,
      "loss": 0.5194,
      "step": 175640
    },
    {
      "epoch": 71.9,
      "grad_norm": 1.680365800857544,
      "learning_rate": 3.5634188974928515e-06,
      "loss": 0.5238,
      "step": 175650
    },
    {
      "epoch": 71.9,
      "grad_norm": 1.9698315858840942,
      "learning_rate": 3.563283436571997e-06,
      "loss": 0.5286,
      "step": 175660
    },
    {
      "epoch": 71.91,
      "grad_norm": 2.47580885887146,
      "learning_rate": 3.563147970064649e-06,
      "loss": 0.5205,
      "step": 175670
    },
    {
      "epoch": 71.91,
      "grad_norm": 1.4126431941986084,
      "learning_rate": 3.5630124979714273e-06,
      "loss": 0.5247,
      "step": 175680
    },
    {
      "epoch": 71.92,
      "grad_norm": 1.9077073335647583,
      "learning_rate": 3.562877020292954e-06,
      "loss": 0.4934,
      "step": 175690
    },
    {
      "epoch": 71.92,
      "grad_norm": 1.7599796056747437,
      "learning_rate": 3.5627415370298473e-06,
      "loss": 0.5002,
      "step": 175700
    },
    {
      "epoch": 71.92,
      "grad_norm": 2.2668731212615967,
      "learning_rate": 3.5626060481827303e-06,
      "loss": 0.4966,
      "step": 175710
    },
    {
      "epoch": 71.93,
      "grad_norm": 1.940593957901001,
      "learning_rate": 3.562470553752222e-06,
      "loss": 0.505,
      "step": 175720
    },
    {
      "epoch": 71.93,
      "grad_norm": 1.423608422279358,
      "learning_rate": 3.5623350537389437e-06,
      "loss": 0.5258,
      "step": 175730
    },
    {
      "epoch": 71.94,
      "grad_norm": 2.0388343334198,
      "learning_rate": 3.5621995481435158e-06,
      "loss": 0.5088,
      "step": 175740
    },
    {
      "epoch": 71.94,
      "grad_norm": 1.8277560472488403,
      "learning_rate": 3.5620640369665588e-06,
      "loss": 0.5195,
      "step": 175750
    },
    {
      "epoch": 71.94,
      "grad_norm": 1.7730635404586792,
      "learning_rate": 3.5619285202086948e-06,
      "loss": 0.5002,
      "step": 175760
    },
    {
      "epoch": 71.95,
      "grad_norm": 1.3795918226242065,
      "learning_rate": 3.5617929978705435e-06,
      "loss": 0.528,
      "step": 175770
    },
    {
      "epoch": 71.95,
      "grad_norm": 1.7179700136184692,
      "learning_rate": 3.5616574699527248e-06,
      "loss": 0.5186,
      "step": 175780
    },
    {
      "epoch": 71.96,
      "grad_norm": 2.723188638687134,
      "learning_rate": 3.5615219364558618e-06,
      "loss": 0.5019,
      "step": 175790
    },
    {
      "epoch": 71.96,
      "grad_norm": 2.242609977722168,
      "learning_rate": 3.561386397380573e-06,
      "loss": 0.5112,
      "step": 175800
    },
    {
      "epoch": 71.96,
      "grad_norm": 1.9759925603866577,
      "learning_rate": 3.561250852727481e-06,
      "loss": 0.4991,
      "step": 175810
    },
    {
      "epoch": 71.97,
      "grad_norm": 1.9830753803253174,
      "learning_rate": 3.5611153024972066e-06,
      "loss": 0.5219,
      "step": 175820
    },
    {
      "epoch": 71.97,
      "grad_norm": 1.9852129220962524,
      "learning_rate": 3.56097974669037e-06,
      "loss": 0.5201,
      "step": 175830
    },
    {
      "epoch": 71.98,
      "grad_norm": 1.919233798980713,
      "learning_rate": 3.5608441853075923e-06,
      "loss": 0.5182,
      "step": 175840
    },
    {
      "epoch": 71.98,
      "grad_norm": 1.7894431352615356,
      "learning_rate": 3.560708618349495e-06,
      "loss": 0.5148,
      "step": 175850
    },
    {
      "epoch": 71.99,
      "grad_norm": 2.3143460750579834,
      "learning_rate": 3.560573045816699e-06,
      "loss": 0.5152,
      "step": 175860
    },
    {
      "epoch": 71.99,
      "grad_norm": 1.9077024459838867,
      "learning_rate": 3.560437467709824e-06,
      "loss": 0.53,
      "step": 175870
    },
    {
      "epoch": 71.99,
      "grad_norm": 1.6112412214279175,
      "learning_rate": 3.560301884029493e-06,
      "loss": 0.4925,
      "step": 175880
    },
    {
      "epoch": 72.0,
      "grad_norm": 1.528408169746399,
      "learning_rate": 3.5601662947763258e-06,
      "loss": 0.5205,
      "step": 175890
    },
    {
      "epoch": 72.0,
      "eval_loss": 0.5135322213172913,
      "eval_runtime": 52.1936,
      "eval_samples_per_second": 66.081,
      "eval_steps_per_second": 8.277,
      "step": 175896
    },
    {
      "epoch": 72.0,
      "grad_norm": 2.1572189331054688,
      "learning_rate": 3.560030699950944e-06,
      "loss": 0.519,
      "step": 175900
    },
    {
      "epoch": 72.01,
      "grad_norm": 2.3800766468048096,
      "learning_rate": 3.5598950995539694e-06,
      "loss": 0.5071,
      "step": 175910
    },
    {
      "epoch": 72.01,
      "grad_norm": 1.9657771587371826,
      "learning_rate": 3.5597594935860216e-06,
      "loss": 0.4963,
      "step": 175920
    },
    {
      "epoch": 72.01,
      "grad_norm": 1.5098388195037842,
      "learning_rate": 3.559623882047723e-06,
      "loss": 0.5193,
      "step": 175930
    },
    {
      "epoch": 72.02,
      "grad_norm": 2.000453472137451,
      "learning_rate": 3.5594882649396934e-06,
      "loss": 0.5206,
      "step": 175940
    },
    {
      "epoch": 72.02,
      "grad_norm": 1.7396882772445679,
      "learning_rate": 3.559352642262556e-06,
      "loss": 0.4857,
      "step": 175950
    },
    {
      "epoch": 72.03,
      "grad_norm": 1.9044692516326904,
      "learning_rate": 3.559217014016931e-06,
      "loss": 0.5147,
      "step": 175960
    },
    {
      "epoch": 72.03,
      "grad_norm": 1.890915036201477,
      "learning_rate": 3.5590813802034405e-06,
      "loss": 0.5282,
      "step": 175970
    },
    {
      "epoch": 72.03,
      "grad_norm": 2.0988762378692627,
      "learning_rate": 3.5589457408227047e-06,
      "loss": 0.5313,
      "step": 175980
    },
    {
      "epoch": 72.04,
      "grad_norm": 1.5528085231781006,
      "learning_rate": 3.558810095875345e-06,
      "loss": 0.5297,
      "step": 175990
    },
    {
      "epoch": 72.04,
      "grad_norm": 2.6484415531158447,
      "learning_rate": 3.5586744453619836e-06,
      "loss": 0.505,
      "step": 176000
    },
    {
      "epoch": 72.05,
      "grad_norm": 1.67936110496521,
      "learning_rate": 3.5585387892832414e-06,
      "loss": 0.5177,
      "step": 176010
    },
    {
      "epoch": 72.05,
      "grad_norm": 2.4588441848754883,
      "learning_rate": 3.55840312763974e-06,
      "loss": 0.5062,
      "step": 176020
    },
    {
      "epoch": 72.05,
      "grad_norm": 1.352135419845581,
      "learning_rate": 3.5582674604321003e-06,
      "loss": 0.4942,
      "step": 176030
    },
    {
      "epoch": 72.06,
      "grad_norm": 2.0941853523254395,
      "learning_rate": 3.5581317876609445e-06,
      "loss": 0.512,
      "step": 176040
    },
    {
      "epoch": 72.06,
      "grad_norm": 2.2046754360198975,
      "learning_rate": 3.557996109326894e-06,
      "loss": 0.5141,
      "step": 176050
    },
    {
      "epoch": 72.07,
      "grad_norm": 1.549541711807251,
      "learning_rate": 3.5578604254305703e-06,
      "loss": 0.5332,
      "step": 176060
    },
    {
      "epoch": 72.07,
      "grad_norm": 2.183525800704956,
      "learning_rate": 3.5577247359725945e-06,
      "loss": 0.5413,
      "step": 176070
    },
    {
      "epoch": 72.08,
      "grad_norm": 2.4871768951416016,
      "learning_rate": 3.5575890409535877e-06,
      "loss": 0.5096,
      "step": 176080
    },
    {
      "epoch": 72.08,
      "grad_norm": 1.9787828922271729,
      "learning_rate": 3.557453340374173e-06,
      "loss": 0.5192,
      "step": 176090
    },
    {
      "epoch": 72.08,
      "grad_norm": 1.67916738986969,
      "learning_rate": 3.5573176342349716e-06,
      "loss": 0.5148,
      "step": 176100
    },
    {
      "epoch": 72.09,
      "grad_norm": 1.8478575944900513,
      "learning_rate": 3.557181922536604e-06,
      "loss": 0.5109,
      "step": 176110
    },
    {
      "epoch": 72.09,
      "grad_norm": 1.838218331336975,
      "learning_rate": 3.557046205279693e-06,
      "loss": 0.5139,
      "step": 176120
    },
    {
      "epoch": 72.1,
      "grad_norm": 1.8479543924331665,
      "learning_rate": 3.5569104824648605e-06,
      "loss": 0.5113,
      "step": 176130
    },
    {
      "epoch": 72.1,
      "grad_norm": 1.4916925430297852,
      "learning_rate": 3.556774754092727e-06,
      "loss": 0.5066,
      "step": 176140
    },
    {
      "epoch": 72.1,
      "grad_norm": 1.9725050926208496,
      "learning_rate": 3.556639020163915e-06,
      "loss": 0.5246,
      "step": 176150
    },
    {
      "epoch": 72.11,
      "grad_norm": 1.7406142950057983,
      "learning_rate": 3.556503280679046e-06,
      "loss": 0.5137,
      "step": 176160
    },
    {
      "epoch": 72.11,
      "grad_norm": 1.6621614694595337,
      "learning_rate": 3.556367535638743e-06,
      "loss": 0.5151,
      "step": 176170
    },
    {
      "epoch": 72.12,
      "grad_norm": 1.9365484714508057,
      "learning_rate": 3.556231785043626e-06,
      "loss": 0.5145,
      "step": 176180
    },
    {
      "epoch": 72.12,
      "grad_norm": 1.4162123203277588,
      "learning_rate": 3.5560960288943187e-06,
      "loss": 0.5333,
      "step": 176190
    },
    {
      "epoch": 72.12,
      "grad_norm": 1.8479456901550293,
      "learning_rate": 3.5559602671914408e-06,
      "loss": 0.5096,
      "step": 176200
    },
    {
      "epoch": 72.13,
      "grad_norm": 1.9879409074783325,
      "learning_rate": 3.555824499935617e-06,
      "loss": 0.5179,
      "step": 176210
    },
    {
      "epoch": 72.13,
      "grad_norm": 2.1772801876068115,
      "learning_rate": 3.5556887271274657e-06,
      "loss": 0.5142,
      "step": 176220
    },
    {
      "epoch": 72.14,
      "grad_norm": 1.867307186126709,
      "learning_rate": 3.5555529487676116e-06,
      "loss": 0.5273,
      "step": 176230
    },
    {
      "epoch": 72.14,
      "grad_norm": 2.0672624111175537,
      "learning_rate": 3.555417164856676e-06,
      "loss": 0.547,
      "step": 176240
    },
    {
      "epoch": 72.14,
      "grad_norm": 1.7649664878845215,
      "learning_rate": 3.555281375395281e-06,
      "loss": 0.4929,
      "step": 176250
    },
    {
      "epoch": 72.15,
      "grad_norm": 1.441188097000122,
      "learning_rate": 3.555145580384048e-06,
      "loss": 0.4994,
      "step": 176260
    },
    {
      "epoch": 72.15,
      "grad_norm": 1.9251786470413208,
      "learning_rate": 3.5550097798236e-06,
      "loss": 0.5312,
      "step": 176270
    },
    {
      "epoch": 72.16,
      "grad_norm": 2.022378921508789,
      "learning_rate": 3.554873973714558e-06,
      "loss": 0.5018,
      "step": 176280
    },
    {
      "epoch": 72.16,
      "grad_norm": 1.9271483421325684,
      "learning_rate": 3.5547381620575446e-06,
      "loss": 0.5097,
      "step": 176290
    },
    {
      "epoch": 72.17,
      "grad_norm": 1.6573387384414673,
      "learning_rate": 3.5546023448531823e-06,
      "loss": 0.5197,
      "step": 176300
    },
    {
      "epoch": 72.17,
      "grad_norm": 1.8492329120635986,
      "learning_rate": 3.5544665221020924e-06,
      "loss": 0.5371,
      "step": 176310
    },
    {
      "epoch": 72.17,
      "grad_norm": 2.169104814529419,
      "learning_rate": 3.5543306938048976e-06,
      "loss": 0.5297,
      "step": 176320
    },
    {
      "epoch": 72.18,
      "grad_norm": 2.2249693870544434,
      "learning_rate": 3.5541948599622204e-06,
      "loss": 0.5178,
      "step": 176330
    },
    {
      "epoch": 72.18,
      "grad_norm": 2.2470078468322754,
      "learning_rate": 3.5540590205746835e-06,
      "loss": 0.5023,
      "step": 176340
    },
    {
      "epoch": 72.19,
      "grad_norm": 2.3044042587280273,
      "learning_rate": 3.5539231756429076e-06,
      "loss": 0.4987,
      "step": 176350
    },
    {
      "epoch": 72.19,
      "grad_norm": 1.5730384588241577,
      "learning_rate": 3.553787325167516e-06,
      "loss": 0.5231,
      "step": 176360
    },
    {
      "epoch": 72.19,
      "grad_norm": 1.6652456521987915,
      "learning_rate": 3.5536514691491307e-06,
      "loss": 0.5368,
      "step": 176370
    },
    {
      "epoch": 72.2,
      "grad_norm": 1.8501626253128052,
      "learning_rate": 3.5535156075883744e-06,
      "loss": 0.541,
      "step": 176380
    },
    {
      "epoch": 72.2,
      "grad_norm": 2.2559685707092285,
      "learning_rate": 3.5533797404858686e-06,
      "loss": 0.505,
      "step": 176390
    },
    {
      "epoch": 72.21,
      "grad_norm": 1.920445442199707,
      "learning_rate": 3.553243867842237e-06,
      "loss": 0.5111,
      "step": 176400
    },
    {
      "epoch": 72.21,
      "grad_norm": 2.0654947757720947,
      "learning_rate": 3.553107989658101e-06,
      "loss": 0.5281,
      "step": 176410
    },
    {
      "epoch": 72.21,
      "grad_norm": 2.1887123584747314,
      "learning_rate": 3.5529721059340834e-06,
      "loss": 0.5133,
      "step": 176420
    },
    {
      "epoch": 72.22,
      "grad_norm": 1.586238980293274,
      "learning_rate": 3.5528362166708067e-06,
      "loss": 0.5264,
      "step": 176430
    },
    {
      "epoch": 72.22,
      "grad_norm": 2.228060245513916,
      "learning_rate": 3.5527003218688933e-06,
      "loss": 0.5162,
      "step": 176440
    },
    {
      "epoch": 72.23,
      "grad_norm": 2.10394287109375,
      "learning_rate": 3.5525644215289654e-06,
      "loss": 0.5032,
      "step": 176450
    },
    {
      "epoch": 72.23,
      "grad_norm": 2.034085512161255,
      "learning_rate": 3.552428515651646e-06,
      "loss": 0.5058,
      "step": 176460
    },
    {
      "epoch": 72.23,
      "grad_norm": 1.5599864721298218,
      "learning_rate": 3.5522926042375577e-06,
      "loss": 0.5282,
      "step": 176470
    },
    {
      "epoch": 72.24,
      "grad_norm": 2.067533493041992,
      "learning_rate": 3.5521566872873226e-06,
      "loss": 0.5215,
      "step": 176480
    },
    {
      "epoch": 72.24,
      "grad_norm": 2.954697847366333,
      "learning_rate": 3.5520207648015638e-06,
      "loss": 0.5113,
      "step": 176490
    },
    {
      "epoch": 72.25,
      "grad_norm": 1.9391868114471436,
      "learning_rate": 3.5518848367809033e-06,
      "loss": 0.5069,
      "step": 176500
    },
    {
      "epoch": 72.25,
      "grad_norm": 2.151139259338379,
      "learning_rate": 3.551748903225965e-06,
      "loss": 0.4942,
      "step": 176510
    },
    {
      "epoch": 72.26,
      "grad_norm": 1.973740816116333,
      "learning_rate": 3.55161296413737e-06,
      "loss": 0.5065,
      "step": 176520
    },
    {
      "epoch": 72.26,
      "grad_norm": 1.8713663816452026,
      "learning_rate": 3.5514770195157422e-06,
      "loss": 0.5205,
      "step": 176530
    },
    {
      "epoch": 72.26,
      "grad_norm": 1.885942816734314,
      "learning_rate": 3.5513410693617042e-06,
      "loss": 0.4986,
      "step": 176540
    },
    {
      "epoch": 72.27,
      "grad_norm": 2.012467622756958,
      "learning_rate": 3.5512051136758783e-06,
      "loss": 0.5028,
      "step": 176550
    },
    {
      "epoch": 72.27,
      "grad_norm": 2.6523220539093018,
      "learning_rate": 3.551069152458888e-06,
      "loss": 0.5185,
      "step": 176560
    },
    {
      "epoch": 72.28,
      "grad_norm": 2.221665143966675,
      "learning_rate": 3.550933185711355e-06,
      "loss": 0.5183,
      "step": 176570
    },
    {
      "epoch": 72.28,
      "grad_norm": 1.6368396282196045,
      "learning_rate": 3.5507972134339027e-06,
      "loss": 0.5276,
      "step": 176580
    },
    {
      "epoch": 72.28,
      "grad_norm": 2.7887465953826904,
      "learning_rate": 3.550661235627155e-06,
      "loss": 0.5063,
      "step": 176590
    },
    {
      "epoch": 72.29,
      "grad_norm": 2.9954347610473633,
      "learning_rate": 3.5505252522917324e-06,
      "loss": 0.5057,
      "step": 176600
    },
    {
      "epoch": 72.29,
      "grad_norm": 1.6801401376724243,
      "learning_rate": 3.55038926342826e-06,
      "loss": 0.5039,
      "step": 176610
    },
    {
      "epoch": 72.3,
      "grad_norm": 1.5080832242965698,
      "learning_rate": 3.5502532690373604e-06,
      "loss": 0.517,
      "step": 176620
    },
    {
      "epoch": 72.3,
      "grad_norm": 3.583059549331665,
      "learning_rate": 3.5501172691196557e-06,
      "loss": 0.5142,
      "step": 176630
    },
    {
      "epoch": 72.3,
      "grad_norm": 1.5256661176681519,
      "learning_rate": 3.5499812636757695e-06,
      "loss": 0.511,
      "step": 176640
    },
    {
      "epoch": 72.31,
      "grad_norm": 2.4826297760009766,
      "learning_rate": 3.5498452527063252e-06,
      "loss": 0.5233,
      "step": 176650
    },
    {
      "epoch": 72.31,
      "grad_norm": 1.8233978748321533,
      "learning_rate": 3.5497092362119447e-06,
      "loss": 0.5075,
      "step": 176660
    },
    {
      "epoch": 72.32,
      "grad_norm": 2.7085773944854736,
      "learning_rate": 3.5495732141932517e-06,
      "loss": 0.5253,
      "step": 176670
    },
    {
      "epoch": 72.32,
      "grad_norm": 2.1109864711761475,
      "learning_rate": 3.5494371866508697e-06,
      "loss": 0.5042,
      "step": 176680
    },
    {
      "epoch": 72.33,
      "grad_norm": 2.275263547897339,
      "learning_rate": 3.54930115358542e-06,
      "loss": 0.5172,
      "step": 176690
    },
    {
      "epoch": 72.33,
      "grad_norm": 2.088017702102661,
      "learning_rate": 3.5491651149975284e-06,
      "loss": 0.5081,
      "step": 176700
    },
    {
      "epoch": 72.33,
      "grad_norm": 1.6470623016357422,
      "learning_rate": 3.549029070887817e-06,
      "loss": 0.5122,
      "step": 176710
    },
    {
      "epoch": 72.34,
      "grad_norm": 2.000032424926758,
      "learning_rate": 3.5488930212569085e-06,
      "loss": 0.5301,
      "step": 176720
    },
    {
      "epoch": 72.34,
      "grad_norm": 2.129387855529785,
      "learning_rate": 3.548756966105427e-06,
      "loss": 0.5237,
      "step": 176730
    },
    {
      "epoch": 72.35,
      "grad_norm": 1.8759394884109497,
      "learning_rate": 3.5486209054339944e-06,
      "loss": 0.5112,
      "step": 176740
    },
    {
      "epoch": 72.35,
      "grad_norm": 2.4111533164978027,
      "learning_rate": 3.548484839243235e-06,
      "loss": 0.5053,
      "step": 176750
    },
    {
      "epoch": 72.35,
      "grad_norm": 1.8148119449615479,
      "learning_rate": 3.5483487675337726e-06,
      "loss": 0.5325,
      "step": 176760
    },
    {
      "epoch": 72.36,
      "grad_norm": 1.8289040327072144,
      "learning_rate": 3.548212690306229e-06,
      "loss": 0.5251,
      "step": 176770
    },
    {
      "epoch": 72.36,
      "grad_norm": 1.902539849281311,
      "learning_rate": 3.5480766075612283e-06,
      "loss": 0.5272,
      "step": 176780
    },
    {
      "epoch": 72.37,
      "grad_norm": 2.2519984245300293,
      "learning_rate": 3.547940519299394e-06,
      "loss": 0.5306,
      "step": 176790
    },
    {
      "epoch": 72.37,
      "grad_norm": 1.8801133632659912,
      "learning_rate": 3.54780442552135e-06,
      "loss": 0.5002,
      "step": 176800
    },
    {
      "epoch": 72.37,
      "grad_norm": 1.9176923036575317,
      "learning_rate": 3.547668326227718e-06,
      "loss": 0.5232,
      "step": 176810
    },
    {
      "epoch": 72.38,
      "grad_norm": 1.5257885456085205,
      "learning_rate": 3.547532221419124e-06,
      "loss": 0.5123,
      "step": 176820
    },
    {
      "epoch": 72.38,
      "grad_norm": 1.7671564817428589,
      "learning_rate": 3.5473961110961896e-06,
      "loss": 0.5031,
      "step": 176830
    },
    {
      "epoch": 72.39,
      "grad_norm": 1.6216487884521484,
      "learning_rate": 3.5472599952595387e-06,
      "loss": 0.5154,
      "step": 176840
    },
    {
      "epoch": 72.39,
      "grad_norm": 1.8764665126800537,
      "learning_rate": 3.5471238739097947e-06,
      "loss": 0.5036,
      "step": 176850
    },
    {
      "epoch": 72.39,
      "grad_norm": 1.5202345848083496,
      "learning_rate": 3.5469877470475814e-06,
      "loss": 0.5069,
      "step": 176860
    },
    {
      "epoch": 72.4,
      "grad_norm": 2.1368792057037354,
      "learning_rate": 3.546851614673522e-06,
      "loss": 0.5204,
      "step": 176870
    },
    {
      "epoch": 72.4,
      "grad_norm": 3.3516862392425537,
      "learning_rate": 3.5467154767882405e-06,
      "loss": 0.513,
      "step": 176880
    },
    {
      "epoch": 72.41,
      "grad_norm": 1.5043141841888428,
      "learning_rate": 3.546579333392361e-06,
      "loss": 0.5082,
      "step": 176890
    },
    {
      "epoch": 72.41,
      "grad_norm": 1.920363426208496,
      "learning_rate": 3.546443184486506e-06,
      "loss": 0.5187,
      "step": 176900
    },
    {
      "epoch": 72.42,
      "grad_norm": 2.2718660831451416,
      "learning_rate": 3.5463070300713004e-06,
      "loss": 0.5157,
      "step": 176910
    },
    {
      "epoch": 72.42,
      "grad_norm": 2.931422710418701,
      "learning_rate": 3.5461708701473677e-06,
      "loss": 0.4998,
      "step": 176920
    },
    {
      "epoch": 72.42,
      "grad_norm": 1.7944918870925903,
      "learning_rate": 3.5460347047153305e-06,
      "loss": 0.507,
      "step": 176930
    },
    {
      "epoch": 72.43,
      "grad_norm": 2.1750125885009766,
      "learning_rate": 3.545898533775813e-06,
      "loss": 0.5146,
      "step": 176940
    },
    {
      "epoch": 72.43,
      "grad_norm": 1.7084729671478271,
      "learning_rate": 3.545762357329439e-06,
      "loss": 0.5154,
      "step": 176950
    },
    {
      "epoch": 72.44,
      "grad_norm": 2.104072332382202,
      "learning_rate": 3.545626175376834e-06,
      "loss": 0.5371,
      "step": 176960
    },
    {
      "epoch": 72.44,
      "grad_norm": 1.7211602926254272,
      "learning_rate": 3.5454899879186186e-06,
      "loss": 0.5046,
      "step": 176970
    },
    {
      "epoch": 72.44,
      "grad_norm": 1.8047904968261719,
      "learning_rate": 3.545353794955419e-06,
      "loss": 0.5111,
      "step": 176980
    },
    {
      "epoch": 72.45,
      "grad_norm": 1.8797794580459595,
      "learning_rate": 3.5452175964878595e-06,
      "loss": 0.4952,
      "step": 176990
    },
    {
      "epoch": 72.45,
      "grad_norm": 2.4116764068603516,
      "learning_rate": 3.5450813925165626e-06,
      "loss": 0.5031,
      "step": 177000
    },
    {
      "epoch": 72.46,
      "grad_norm": 1.4912362098693848,
      "learning_rate": 3.5449451830421516e-06,
      "loss": 0.5214,
      "step": 177010
    },
    {
      "epoch": 72.46,
      "grad_norm": 2.0704286098480225,
      "learning_rate": 3.5448089680652527e-06,
      "loss": 0.5124,
      "step": 177020
    },
    {
      "epoch": 72.46,
      "grad_norm": 1.8378890752792358,
      "learning_rate": 3.5446727475864878e-06,
      "loss": 0.5337,
      "step": 177030
    },
    {
      "epoch": 72.47,
      "grad_norm": 1.7747591733932495,
      "learning_rate": 3.5445365216064823e-06,
      "loss": 0.4937,
      "step": 177040
    },
    {
      "epoch": 72.47,
      "grad_norm": 1.9753906726837158,
      "learning_rate": 3.5444002901258597e-06,
      "loss": 0.5234,
      "step": 177050
    },
    {
      "epoch": 72.48,
      "grad_norm": 1.5174062252044678,
      "learning_rate": 3.544264053145244e-06,
      "loss": 0.5149,
      "step": 177060
    },
    {
      "epoch": 72.48,
      "grad_norm": 2.2333824634552,
      "learning_rate": 3.5441278106652588e-06,
      "loss": 0.508,
      "step": 177070
    },
    {
      "epoch": 72.48,
      "grad_norm": 2.0453379154205322,
      "learning_rate": 3.5439915626865295e-06,
      "loss": 0.4986,
      "step": 177080
    },
    {
      "epoch": 72.49,
      "grad_norm": 2.1154074668884277,
      "learning_rate": 3.5438553092096792e-06,
      "loss": 0.501,
      "step": 177090
    },
    {
      "epoch": 72.49,
      "grad_norm": 1.6351268291473389,
      "learning_rate": 3.5437190502353328e-06,
      "loss": 0.5218,
      "step": 177100
    },
    {
      "epoch": 72.5,
      "grad_norm": 2.9550371170043945,
      "learning_rate": 3.5435827857641135e-06,
      "loss": 0.5122,
      "step": 177110
    },
    {
      "epoch": 72.5,
      "grad_norm": 2.0355899333953857,
      "learning_rate": 3.543446515796646e-06,
      "loss": 0.5105,
      "step": 177120
    },
    {
      "epoch": 72.51,
      "grad_norm": 1.5460686683654785,
      "learning_rate": 3.5433102403335547e-06,
      "loss": 0.5205,
      "step": 177130
    },
    {
      "epoch": 72.51,
      "grad_norm": 3.03245210647583,
      "learning_rate": 3.543173959375464e-06,
      "loss": 0.5121,
      "step": 177140
    },
    {
      "epoch": 72.51,
      "grad_norm": 1.7356624603271484,
      "learning_rate": 3.543037672922998e-06,
      "loss": 0.5149,
      "step": 177150
    },
    {
      "epoch": 72.52,
      "grad_norm": 1.9050419330596924,
      "learning_rate": 3.5429013809767797e-06,
      "loss": 0.5286,
      "step": 177160
    },
    {
      "epoch": 72.52,
      "grad_norm": 2.201946258544922,
      "learning_rate": 3.542765083537435e-06,
      "loss": 0.5106,
      "step": 177170
    },
    {
      "epoch": 72.53,
      "grad_norm": 2.841214418411255,
      "learning_rate": 3.5426287806055887e-06,
      "loss": 0.5109,
      "step": 177180
    },
    {
      "epoch": 72.53,
      "grad_norm": 2.1835527420043945,
      "learning_rate": 3.5424924721818645e-06,
      "loss": 0.5118,
      "step": 177190
    },
    {
      "epoch": 72.53,
      "grad_norm": 2.110299825668335,
      "learning_rate": 3.542356158266886e-06,
      "loss": 0.5022,
      "step": 177200
    },
    {
      "epoch": 72.54,
      "grad_norm": 1.8135793209075928,
      "learning_rate": 3.5422198388612787e-06,
      "loss": 0.5127,
      "step": 177210
    },
    {
      "epoch": 72.54,
      "grad_norm": 1.407670021057129,
      "learning_rate": 3.5420835139656664e-06,
      "loss": 0.5194,
      "step": 177220
    },
    {
      "epoch": 72.55,
      "grad_norm": 2.171970844268799,
      "learning_rate": 3.541947183580674e-06,
      "loss": 0.4914,
      "step": 177230
    },
    {
      "epoch": 72.55,
      "grad_norm": 2.1579525470733643,
      "learning_rate": 3.541810847706926e-06,
      "loss": 0.5089,
      "step": 177240
    },
    {
      "epoch": 72.55,
      "grad_norm": 1.8056873083114624,
      "learning_rate": 3.541674506345047e-06,
      "loss": 0.508,
      "step": 177250
    },
    {
      "epoch": 72.56,
      "grad_norm": 1.5794627666473389,
      "learning_rate": 3.54153815949566e-06,
      "loss": 0.5036,
      "step": 177260
    },
    {
      "epoch": 72.56,
      "grad_norm": 1.9303357601165771,
      "learning_rate": 3.541401807159392e-06,
      "loss": 0.5103,
      "step": 177270
    },
    {
      "epoch": 72.57,
      "grad_norm": 1.7918411493301392,
      "learning_rate": 3.5412654493368665e-06,
      "loss": 0.5219,
      "step": 177280
    },
    {
      "epoch": 72.57,
      "grad_norm": 2.2386651039123535,
      "learning_rate": 3.541129086028708e-06,
      "loss": 0.5078,
      "step": 177290
    },
    {
      "epoch": 72.57,
      "grad_norm": 2.215515613555908,
      "learning_rate": 3.540992717235542e-06,
      "loss": 0.499,
      "step": 177300
    },
    {
      "epoch": 72.58,
      "grad_norm": 1.9902763366699219,
      "learning_rate": 3.5408563429579914e-06,
      "loss": 0.4945,
      "step": 177310
    },
    {
      "epoch": 72.58,
      "grad_norm": 1.673478126525879,
      "learning_rate": 3.5407199631966835e-06,
      "loss": 0.5147,
      "step": 177320
    },
    {
      "epoch": 72.59,
      "grad_norm": 2.59617018699646,
      "learning_rate": 3.5405835779522404e-06,
      "loss": 0.5085,
      "step": 177330
    },
    {
      "epoch": 72.59,
      "grad_norm": 1.5846964120864868,
      "learning_rate": 3.5404471872252885e-06,
      "loss": 0.5209,
      "step": 177340
    },
    {
      "epoch": 72.6,
      "grad_norm": 2.814751625061035,
      "learning_rate": 3.5403107910164516e-06,
      "loss": 0.5119,
      "step": 177350
    },
    {
      "epoch": 72.6,
      "grad_norm": 1.7759830951690674,
      "learning_rate": 3.5401743893263556e-06,
      "loss": 0.5161,
      "step": 177360
    },
    {
      "epoch": 72.6,
      "grad_norm": 2.000533103942871,
      "learning_rate": 3.5400379821556244e-06,
      "loss": 0.504,
      "step": 177370
    },
    {
      "epoch": 72.61,
      "grad_norm": 1.650122046470642,
      "learning_rate": 3.539901569504883e-06,
      "loss": 0.5071,
      "step": 177380
    },
    {
      "epoch": 72.61,
      "grad_norm": 2.1966075897216797,
      "learning_rate": 3.5397651513747573e-06,
      "loss": 0.4954,
      "step": 177390
    },
    {
      "epoch": 72.62,
      "grad_norm": 1.7703839540481567,
      "learning_rate": 3.5396287277658708e-06,
      "loss": 0.4932,
      "step": 177400
    },
    {
      "epoch": 72.62,
      "grad_norm": 1.6336500644683838,
      "learning_rate": 3.5394922986788494e-06,
      "loss": 0.511,
      "step": 177410
    },
    {
      "epoch": 72.62,
      "grad_norm": 2.154998540878296,
      "learning_rate": 3.539355864114318e-06,
      "loss": 0.5026,
      "step": 177420
    },
    {
      "epoch": 72.63,
      "grad_norm": 1.9913716316223145,
      "learning_rate": 3.5392194240729004e-06,
      "loss": 0.504,
      "step": 177430
    },
    {
      "epoch": 72.63,
      "grad_norm": 1.6370397806167603,
      "learning_rate": 3.5390829785552238e-06,
      "loss": 0.5162,
      "step": 177440
    },
    {
      "epoch": 72.64,
      "grad_norm": 1.7375096082687378,
      "learning_rate": 3.53894652756191e-06,
      "loss": 0.5149,
      "step": 177450
    },
    {
      "epoch": 72.64,
      "grad_norm": 1.7588140964508057,
      "learning_rate": 3.538810071093587e-06,
      "loss": 0.5292,
      "step": 177460
    },
    {
      "epoch": 72.64,
      "grad_norm": 2.009852170944214,
      "learning_rate": 3.5386736091508794e-06,
      "loss": 0.5109,
      "step": 177470
    },
    {
      "epoch": 72.65,
      "grad_norm": 1.6925522089004517,
      "learning_rate": 3.5385371417344115e-06,
      "loss": 0.5075,
      "step": 177480
    },
    {
      "epoch": 72.65,
      "grad_norm": 1.855358600616455,
      "learning_rate": 3.538400668844809e-06,
      "loss": 0.5102,
      "step": 177490
    },
    {
      "epoch": 72.66,
      "grad_norm": 1.667927861213684,
      "learning_rate": 3.538264190482696e-06,
      "loss": 0.5255,
      "step": 177500
    },
    {
      "epoch": 72.66,
      "grad_norm": 1.985044240951538,
      "learning_rate": 3.5381277066486993e-06,
      "loss": 0.529,
      "step": 177510
    },
    {
      "epoch": 72.66,
      "grad_norm": 1.657232642173767,
      "learning_rate": 3.5379912173434435e-06,
      "loss": 0.5197,
      "step": 177520
    },
    {
      "epoch": 72.67,
      "grad_norm": 1.7843434810638428,
      "learning_rate": 3.537854722567553e-06,
      "loss": 0.4976,
      "step": 177530
    },
    {
      "epoch": 72.67,
      "grad_norm": 1.5267553329467773,
      "learning_rate": 3.537718222321653e-06,
      "loss": 0.5187,
      "step": 177540
    },
    {
      "epoch": 72.68,
      "grad_norm": 2.0094916820526123,
      "learning_rate": 3.5375817166063705e-06,
      "loss": 0.5159,
      "step": 177550
    },
    {
      "epoch": 72.68,
      "grad_norm": 1.913699984550476,
      "learning_rate": 3.5374452054223295e-06,
      "loss": 0.5213,
      "step": 177560
    },
    {
      "epoch": 72.69,
      "grad_norm": 1.9461760520935059,
      "learning_rate": 3.537308688770156e-06,
      "loss": 0.5105,
      "step": 177570
    },
    {
      "epoch": 72.69,
      "grad_norm": 2.0109519958496094,
      "learning_rate": 3.537172166650475e-06,
      "loss": 0.5311,
      "step": 177580
    },
    {
      "epoch": 72.69,
      "grad_norm": 1.6948471069335938,
      "learning_rate": 3.537035639063911e-06,
      "loss": 0.5028,
      "step": 177590
    },
    {
      "epoch": 72.7,
      "grad_norm": 1.739665150642395,
      "learning_rate": 3.536899106011091e-06,
      "loss": 0.5113,
      "step": 177600
    },
    {
      "epoch": 72.7,
      "grad_norm": 1.7810399532318115,
      "learning_rate": 3.53676256749264e-06,
      "loss": 0.5053,
      "step": 177610
    },
    {
      "epoch": 72.71,
      "grad_norm": 2.3361871242523193,
      "learning_rate": 3.5366260235091826e-06,
      "loss": 0.499,
      "step": 177620
    },
    {
      "epoch": 72.71,
      "grad_norm": 1.7341325283050537,
      "learning_rate": 3.5364894740613443e-06,
      "loss": 0.4993,
      "step": 177630
    },
    {
      "epoch": 72.71,
      "grad_norm": 1.7878718376159668,
      "learning_rate": 3.5363529191497526e-06,
      "loss": 0.4907,
      "step": 177640
    },
    {
      "epoch": 72.72,
      "grad_norm": 2.2872314453125,
      "learning_rate": 3.536216358775031e-06,
      "loss": 0.5324,
      "step": 177650
    },
    {
      "epoch": 72.72,
      "grad_norm": 2.631899118423462,
      "learning_rate": 3.5360797929378055e-06,
      "loss": 0.516,
      "step": 177660
    },
    {
      "epoch": 72.73,
      "grad_norm": 1.7560265064239502,
      "learning_rate": 3.535943221638702e-06,
      "loss": 0.4967,
      "step": 177670
    },
    {
      "epoch": 72.73,
      "grad_norm": 2.207972764968872,
      "learning_rate": 3.535806644878346e-06,
      "loss": 0.5118,
      "step": 177680
    },
    {
      "epoch": 72.73,
      "grad_norm": 2.045060396194458,
      "learning_rate": 3.5356700626573635e-06,
      "loss": 0.4979,
      "step": 177690
    },
    {
      "epoch": 72.74,
      "grad_norm": 2.069308042526245,
      "learning_rate": 3.535533474976379e-06,
      "loss": 0.512,
      "step": 177700
    },
    {
      "epoch": 72.74,
      "grad_norm": 1.708174228668213,
      "learning_rate": 3.535396881836019e-06,
      "loss": 0.5056,
      "step": 177710
    },
    {
      "epoch": 72.75,
      "grad_norm": 1.776868462562561,
      "learning_rate": 3.5352602832369103e-06,
      "loss": 0.5146,
      "step": 177720
    },
    {
      "epoch": 72.75,
      "grad_norm": 1.5524497032165527,
      "learning_rate": 3.5351236791796763e-06,
      "loss": 0.5273,
      "step": 177730
    },
    {
      "epoch": 72.75,
      "grad_norm": 1.919212818145752,
      "learning_rate": 3.534987069664945e-06,
      "loss": 0.5106,
      "step": 177740
    },
    {
      "epoch": 72.76,
      "grad_norm": 1.717077374458313,
      "learning_rate": 3.5348504546933404e-06,
      "loss": 0.5037,
      "step": 177750
    },
    {
      "epoch": 72.76,
      "grad_norm": 1.5640628337860107,
      "learning_rate": 3.534713834265489e-06,
      "loss": 0.52,
      "step": 177760
    },
    {
      "epoch": 72.77,
      "grad_norm": 2.1617238521575928,
      "learning_rate": 3.5345772083820173e-06,
      "loss": 0.516,
      "step": 177770
    },
    {
      "epoch": 72.77,
      "grad_norm": 2.7448573112487793,
      "learning_rate": 3.5344405770435503e-06,
      "loss": 0.4982,
      "step": 177780
    },
    {
      "epoch": 72.78,
      "grad_norm": 1.9679293632507324,
      "learning_rate": 3.534303940250714e-06,
      "loss": 0.5122,
      "step": 177790
    },
    {
      "epoch": 72.78,
      "grad_norm": 1.564573049545288,
      "learning_rate": 3.5341672980041354e-06,
      "loss": 0.5054,
      "step": 177800
    },
    {
      "epoch": 72.78,
      "grad_norm": 1.7526273727416992,
      "learning_rate": 3.5340306503044384e-06,
      "loss": 0.4984,
      "step": 177810
    },
    {
      "epoch": 72.79,
      "grad_norm": 2.2301948070526123,
      "learning_rate": 3.5338939971522502e-06,
      "loss": 0.5201,
      "step": 177820
    },
    {
      "epoch": 72.79,
      "grad_norm": 1.723394513130188,
      "learning_rate": 3.533757338548197e-06,
      "loss": 0.5069,
      "step": 177830
    },
    {
      "epoch": 72.8,
      "grad_norm": 2.2574093341827393,
      "learning_rate": 3.5336206744929047e-06,
      "loss": 0.5151,
      "step": 177840
    },
    {
      "epoch": 72.8,
      "grad_norm": 1.5552648305892944,
      "learning_rate": 3.5334840049869985e-06,
      "loss": 0.527,
      "step": 177850
    },
    {
      "epoch": 72.8,
      "grad_norm": 1.9636361598968506,
      "learning_rate": 3.5333473300311057e-06,
      "loss": 0.5323,
      "step": 177860
    },
    {
      "epoch": 72.81,
      "grad_norm": 1.6238114833831787,
      "learning_rate": 3.5332106496258518e-06,
      "loss": 0.5109,
      "step": 177870
    },
    {
      "epoch": 72.81,
      "grad_norm": 3.045382022857666,
      "learning_rate": 3.5330739637718624e-06,
      "loss": 0.5021,
      "step": 177880
    },
    {
      "epoch": 72.82,
      "grad_norm": 2.262413740158081,
      "learning_rate": 3.5329372724697642e-06,
      "loss": 0.5264,
      "step": 177890
    },
    {
      "epoch": 72.82,
      "grad_norm": 1.8065625429153442,
      "learning_rate": 3.5328005757201834e-06,
      "loss": 0.5158,
      "step": 177900
    },
    {
      "epoch": 72.82,
      "grad_norm": 1.55332612991333,
      "learning_rate": 3.532663873523746e-06,
      "loss": 0.5335,
      "step": 177910
    },
    {
      "epoch": 72.83,
      "grad_norm": 1.8235472440719604,
      "learning_rate": 3.532527165881078e-06,
      "loss": 0.533,
      "step": 177920
    },
    {
      "epoch": 72.83,
      "grad_norm": 1.7247445583343506,
      "learning_rate": 3.5323904527928063e-06,
      "loss": 0.5026,
      "step": 177930
    },
    {
      "epoch": 72.84,
      "grad_norm": 2.393190383911133,
      "learning_rate": 3.5322537342595573e-06,
      "loss": 0.5107,
      "step": 177940
    },
    {
      "epoch": 72.84,
      "grad_norm": 1.7154722213745117,
      "learning_rate": 3.5321170102819558e-06,
      "loss": 0.5158,
      "step": 177950
    },
    {
      "epoch": 72.84,
      "grad_norm": 1.9911272525787354,
      "learning_rate": 3.5319802808606294e-06,
      "loss": 0.5176,
      "step": 177960
    },
    {
      "epoch": 72.85,
      "grad_norm": 2.8746187686920166,
      "learning_rate": 3.5318435459962036e-06,
      "loss": 0.5103,
      "step": 177970
    },
    {
      "epoch": 72.85,
      "grad_norm": 1.906441569328308,
      "learning_rate": 3.5317068056893065e-06,
      "loss": 0.5102,
      "step": 177980
    },
    {
      "epoch": 72.86,
      "grad_norm": 2.1912200450897217,
      "learning_rate": 3.531570059940562e-06,
      "loss": 0.5392,
      "step": 177990
    },
    {
      "epoch": 72.86,
      "grad_norm": 2.2654435634613037,
      "learning_rate": 3.5314333087505986e-06,
      "loss": 0.5132,
      "step": 178000
    },
    {
      "epoch": 72.87,
      "grad_norm": 2.0096724033355713,
      "learning_rate": 3.5312965521200413e-06,
      "loss": 0.5191,
      "step": 178010
    },
    {
      "epoch": 72.87,
      "grad_norm": 1.8544847965240479,
      "learning_rate": 3.5311597900495173e-06,
      "loss": 0.522,
      "step": 178020
    },
    {
      "epoch": 72.87,
      "grad_norm": 1.5958881378173828,
      "learning_rate": 3.531023022539653e-06,
      "loss": 0.5094,
      "step": 178030
    },
    {
      "epoch": 72.88,
      "grad_norm": 1.9949380159378052,
      "learning_rate": 3.530886249591075e-06,
      "loss": 0.5143,
      "step": 178040
    },
    {
      "epoch": 72.88,
      "grad_norm": 1.7074942588806152,
      "learning_rate": 3.5307494712044093e-06,
      "loss": 0.5125,
      "step": 178050
    },
    {
      "epoch": 72.89,
      "grad_norm": 1.933872938156128,
      "learning_rate": 3.5306126873802836e-06,
      "loss": 0.5174,
      "step": 178060
    },
    {
      "epoch": 72.89,
      "grad_norm": 2.135573387145996,
      "learning_rate": 3.530475898119323e-06,
      "loss": 0.5217,
      "step": 178070
    },
    {
      "epoch": 72.89,
      "grad_norm": 1.3661600351333618,
      "learning_rate": 3.530339103422155e-06,
      "loss": 0.4975,
      "step": 178080
    },
    {
      "epoch": 72.9,
      "grad_norm": 5.606698036193848,
      "learning_rate": 3.530202303289407e-06,
      "loss": 0.5097,
      "step": 178090
    },
    {
      "epoch": 72.9,
      "grad_norm": 1.9230173826217651,
      "learning_rate": 3.530065497721704e-06,
      "loss": 0.5309,
      "step": 178100
    },
    {
      "epoch": 72.91,
      "grad_norm": 1.6442464590072632,
      "learning_rate": 3.529928686719673e-06,
      "loss": 0.5037,
      "step": 178110
    },
    {
      "epoch": 72.91,
      "grad_norm": 1.5507125854492188,
      "learning_rate": 3.5297918702839415e-06,
      "loss": 0.518,
      "step": 178120
    },
    {
      "epoch": 72.91,
      "grad_norm": 2.262528896331787,
      "learning_rate": 3.529655048415136e-06,
      "loss": 0.5215,
      "step": 178130
    },
    {
      "epoch": 72.92,
      "grad_norm": 2.5658352375030518,
      "learning_rate": 3.529518221113883e-06,
      "loss": 0.5115,
      "step": 178140
    },
    {
      "epoch": 72.92,
      "grad_norm": 2.210589647293091,
      "learning_rate": 3.529381388380809e-06,
      "loss": 0.5345,
      "step": 178150
    },
    {
      "epoch": 72.93,
      "grad_norm": 2.179300308227539,
      "learning_rate": 3.5292445502165428e-06,
      "loss": 0.5155,
      "step": 178160
    },
    {
      "epoch": 72.93,
      "grad_norm": 2.0211708545684814,
      "learning_rate": 3.529107706621708e-06,
      "loss": 0.4966,
      "step": 178170
    },
    {
      "epoch": 72.93,
      "grad_norm": 2.3627126216888428,
      "learning_rate": 3.5289708575969343e-06,
      "loss": 0.4868,
      "step": 178180
    },
    {
      "epoch": 72.94,
      "grad_norm": 2.4924726486206055,
      "learning_rate": 3.5288340031428464e-06,
      "loss": 0.5158,
      "step": 178190
    },
    {
      "epoch": 72.94,
      "grad_norm": 1.895407795906067,
      "learning_rate": 3.5286971432600723e-06,
      "loss": 0.5118,
      "step": 178200
    },
    {
      "epoch": 72.95,
      "grad_norm": 2.091250419616699,
      "learning_rate": 3.528560277949239e-06,
      "loss": 0.5273,
      "step": 178210
    },
    {
      "epoch": 72.95,
      "grad_norm": 2.0520379543304443,
      "learning_rate": 3.528423407210974e-06,
      "loss": 0.5065,
      "step": 178220
    },
    {
      "epoch": 72.96,
      "grad_norm": 1.6198359727859497,
      "learning_rate": 3.528286531045903e-06,
      "loss": 0.5166,
      "step": 178230
    },
    {
      "epoch": 72.96,
      "grad_norm": 2.1155080795288086,
      "learning_rate": 3.528149649454654e-06,
      "loss": 0.5219,
      "step": 178240
    },
    {
      "epoch": 72.96,
      "grad_norm": 1.7739731073379517,
      "learning_rate": 3.5280127624378528e-06,
      "loss": 0.5084,
      "step": 178250
    },
    {
      "epoch": 72.97,
      "grad_norm": 2.179739236831665,
      "learning_rate": 3.527875869996128e-06,
      "loss": 0.5049,
      "step": 178260
    },
    {
      "epoch": 72.97,
      "grad_norm": 1.8327536582946777,
      "learning_rate": 3.5277389721301056e-06,
      "loss": 0.5137,
      "step": 178270
    },
    {
      "epoch": 72.98,
      "grad_norm": 1.9050047397613525,
      "learning_rate": 3.5276020688404136e-06,
      "loss": 0.524,
      "step": 178280
    },
    {
      "epoch": 72.98,
      "grad_norm": 2.083136558532715,
      "learning_rate": 3.527465160127678e-06,
      "loss": 0.5173,
      "step": 178290
    },
    {
      "epoch": 72.98,
      "grad_norm": 1.638922929763794,
      "learning_rate": 3.527328245992527e-06,
      "loss": 0.5045,
      "step": 178300
    },
    {
      "epoch": 72.99,
      "grad_norm": 2.1380908489227295,
      "learning_rate": 3.5271913264355867e-06,
      "loss": 0.5133,
      "step": 178310
    },
    {
      "epoch": 72.99,
      "grad_norm": 1.5151749849319458,
      "learning_rate": 3.5270544014574856e-06,
      "loss": 0.5096,
      "step": 178320
    },
    {
      "epoch": 73.0,
      "grad_norm": 2.473184585571289,
      "learning_rate": 3.52691747105885e-06,
      "loss": 0.5299,
      "step": 178330
    },
    {
      "epoch": 73.0,
      "eval_loss": 0.5147174000740051,
      "eval_runtime": 52.5207,
      "eval_samples_per_second": 65.669,
      "eval_steps_per_second": 8.225,
      "step": 178339
    },
    {
      "epoch": 73.0,
      "grad_norm": 1.528168797492981,
      "learning_rate": 3.526780535240308e-06,
      "loss": 0.5306,
      "step": 178340
    },
    {
      "epoch": 73.0,
      "grad_norm": 1.728865385055542,
      "learning_rate": 3.5266435940024854e-06,
      "loss": 0.5255,
      "step": 178350
    },
    {
      "epoch": 73.01,
      "grad_norm": 1.673398733139038,
      "learning_rate": 3.5265066473460114e-06,
      "loss": 0.5189,
      "step": 178360
    },
    {
      "epoch": 73.01,
      "grad_norm": 1.9696300029754639,
      "learning_rate": 3.5263696952715114e-06,
      "loss": 0.504,
      "step": 178370
    },
    {
      "epoch": 73.02,
      "grad_norm": 2.063037157058716,
      "learning_rate": 3.5262327377796145e-06,
      "loss": 0.5121,
      "step": 178380
    },
    {
      "epoch": 73.02,
      "grad_norm": 2.62868070602417,
      "learning_rate": 3.5260957748709465e-06,
      "loss": 0.5047,
      "step": 178390
    },
    {
      "epoch": 73.02,
      "grad_norm": 1.9693952798843384,
      "learning_rate": 3.5259588065461364e-06,
      "loss": 0.5084,
      "step": 178400
    },
    {
      "epoch": 73.03,
      "grad_norm": 2.0778017044067383,
      "learning_rate": 3.525821832805811e-06,
      "loss": 0.5247,
      "step": 178410
    },
    {
      "epoch": 73.03,
      "grad_norm": 1.7299668788909912,
      "learning_rate": 3.525684853650597e-06,
      "loss": 0.5018,
      "step": 178420
    },
    {
      "epoch": 73.04,
      "grad_norm": 1.5539686679840088,
      "learning_rate": 3.5255478690811227e-06,
      "loss": 0.5168,
      "step": 178430
    },
    {
      "epoch": 73.04,
      "grad_norm": 2.0065951347351074,
      "learning_rate": 3.5254108790980156e-06,
      "loss": 0.5071,
      "step": 178440
    },
    {
      "epoch": 73.05,
      "grad_norm": 1.9407798051834106,
      "learning_rate": 3.525273883701903e-06,
      "loss": 0.5046,
      "step": 178450
    },
    {
      "epoch": 73.05,
      "grad_norm": 1.8930517435073853,
      "learning_rate": 3.5251368828934123e-06,
      "loss": 0.4809,
      "step": 178460
    },
    {
      "epoch": 73.05,
      "grad_norm": 1.5948611497879028,
      "learning_rate": 3.524999876673171e-06,
      "loss": 0.5047,
      "step": 178470
    },
    {
      "epoch": 73.06,
      "grad_norm": 1.9633946418762207,
      "learning_rate": 3.5248628650418068e-06,
      "loss": 0.5007,
      "step": 178480
    },
    {
      "epoch": 73.06,
      "grad_norm": 1.7002511024475098,
      "learning_rate": 3.524725847999948e-06,
      "loss": 0.5061,
      "step": 178490
    },
    {
      "epoch": 73.07,
      "grad_norm": 1.4055380821228027,
      "learning_rate": 3.5245888255482214e-06,
      "loss": 0.5113,
      "step": 178500
    },
    {
      "epoch": 73.07,
      "grad_norm": 1.5488340854644775,
      "learning_rate": 3.5244517976872553e-06,
      "loss": 0.5006,
      "step": 178510
    },
    {
      "epoch": 73.07,
      "grad_norm": 2.0515847206115723,
      "learning_rate": 3.524314764417678e-06,
      "loss": 0.5084,
      "step": 178520
    },
    {
      "epoch": 73.08,
      "grad_norm": 1.8870140314102173,
      "learning_rate": 3.5241777257401146e-06,
      "loss": 0.5328,
      "step": 178530
    },
    {
      "epoch": 73.08,
      "grad_norm": 1.8876363039016724,
      "learning_rate": 3.5240406816551962e-06,
      "loss": 0.508,
      "step": 178540
    },
    {
      "epoch": 73.09,
      "grad_norm": 2.499962568283081,
      "learning_rate": 3.523903632163548e-06,
      "loss": 0.5201,
      "step": 178550
    },
    {
      "epoch": 73.09,
      "grad_norm": 1.5614562034606934,
      "learning_rate": 3.5237665772657983e-06,
      "loss": 0.5129,
      "step": 178560
    },
    {
      "epoch": 73.09,
      "grad_norm": 1.9046630859375,
      "learning_rate": 3.5236295169625762e-06,
      "loss": 0.5163,
      "step": 178570
    },
    {
      "epoch": 73.1,
      "grad_norm": 1.9282280206680298,
      "learning_rate": 3.523492451254509e-06,
      "loss": 0.5205,
      "step": 178580
    },
    {
      "epoch": 73.1,
      "grad_norm": 1.4659335613250732,
      "learning_rate": 3.5233553801422235e-06,
      "loss": 0.505,
      "step": 178590
    },
    {
      "epoch": 73.11,
      "grad_norm": 2.3554093837738037,
      "learning_rate": 3.523218303626349e-06,
      "loss": 0.5091,
      "step": 178600
    },
    {
      "epoch": 73.11,
      "grad_norm": 1.675707459449768,
      "learning_rate": 3.523081221707513e-06,
      "loss": 0.5352,
      "step": 178610
    },
    {
      "epoch": 73.12,
      "grad_norm": 1.6439721584320068,
      "learning_rate": 3.522944134386343e-06,
      "loss": 0.5217,
      "step": 178620
    },
    {
      "epoch": 73.12,
      "grad_norm": 1.4538935422897339,
      "learning_rate": 3.5228070416634673e-06,
      "loss": 0.5173,
      "step": 178630
    },
    {
      "epoch": 73.12,
      "grad_norm": 1.3499195575714111,
      "learning_rate": 3.522669943539514e-06,
      "loss": 0.5155,
      "step": 178640
    },
    {
      "epoch": 73.13,
      "grad_norm": 2.927582025527954,
      "learning_rate": 3.522532840015111e-06,
      "loss": 0.5039,
      "step": 178650
    },
    {
      "epoch": 73.13,
      "grad_norm": 1.6641596555709839,
      "learning_rate": 3.5223957310908864e-06,
      "loss": 0.5101,
      "step": 178660
    },
    {
      "epoch": 73.14,
      "grad_norm": 1.5162057876586914,
      "learning_rate": 3.5222586167674677e-06,
      "loss": 0.4938,
      "step": 178670
    },
    {
      "epoch": 73.14,
      "grad_norm": 1.4814093112945557,
      "learning_rate": 3.5221214970454845e-06,
      "loss": 0.5182,
      "step": 178680
    },
    {
      "epoch": 73.14,
      "grad_norm": 1.689042091369629,
      "learning_rate": 3.521984371925563e-06,
      "loss": 0.5318,
      "step": 178690
    },
    {
      "epoch": 73.15,
      "grad_norm": 1.8026715517044067,
      "learning_rate": 3.521847241408333e-06,
      "loss": 0.5181,
      "step": 178700
    },
    {
      "epoch": 73.15,
      "grad_norm": 1.8368366956710815,
      "learning_rate": 3.521710105494422e-06,
      "loss": 0.4949,
      "step": 178710
    },
    {
      "epoch": 73.16,
      "grad_norm": 2.338204860687256,
      "learning_rate": 3.5215729641844576e-06,
      "loss": 0.5118,
      "step": 178720
    },
    {
      "epoch": 73.16,
      "grad_norm": 1.919116735458374,
      "learning_rate": 3.5214358174790692e-06,
      "loss": 0.5242,
      "step": 178730
    },
    {
      "epoch": 73.16,
      "grad_norm": 1.9887334108352661,
      "learning_rate": 3.521298665378885e-06,
      "loss": 0.524,
      "step": 178740
    },
    {
      "epoch": 73.17,
      "grad_norm": 1.8243403434753418,
      "learning_rate": 3.5211615078845316e-06,
      "loss": 0.5105,
      "step": 178750
    },
    {
      "epoch": 73.17,
      "grad_norm": 1.5834749937057495,
      "learning_rate": 3.521024344996639e-06,
      "loss": 0.4969,
      "step": 178760
    },
    {
      "epoch": 73.18,
      "grad_norm": 1.551837682723999,
      "learning_rate": 3.5208871767158344e-06,
      "loss": 0.51,
      "step": 178770
    },
    {
      "epoch": 73.18,
      "grad_norm": 1.91185462474823,
      "learning_rate": 3.5207500030427466e-06,
      "loss": 0.5004,
      "step": 178780
    },
    {
      "epoch": 73.18,
      "grad_norm": 1.9527881145477295,
      "learning_rate": 3.520612823978005e-06,
      "loss": 0.5018,
      "step": 178790
    },
    {
      "epoch": 73.19,
      "grad_norm": 1.8077017068862915,
      "learning_rate": 3.5204756395222363e-06,
      "loss": 0.5128,
      "step": 178800
    },
    {
      "epoch": 73.19,
      "grad_norm": 1.9531066417694092,
      "learning_rate": 3.520338449676071e-06,
      "loss": 0.4978,
      "step": 178810
    },
    {
      "epoch": 73.2,
      "grad_norm": 1.7785351276397705,
      "learning_rate": 3.5202012544401347e-06,
      "loss": 0.5088,
      "step": 178820
    },
    {
      "epoch": 73.2,
      "grad_norm": 1.4806435108184814,
      "learning_rate": 3.5200640538150574e-06,
      "loss": 0.5184,
      "step": 178830
    },
    {
      "epoch": 73.21,
      "grad_norm": 1.6841667890548706,
      "learning_rate": 3.519926847801468e-06,
      "loss": 0.5114,
      "step": 178840
    },
    {
      "epoch": 73.21,
      "grad_norm": 1.771830439567566,
      "learning_rate": 3.519789636399995e-06,
      "loss": 0.5037,
      "step": 178850
    },
    {
      "epoch": 73.21,
      "grad_norm": 1.8251638412475586,
      "learning_rate": 3.519652419611266e-06,
      "loss": 0.5103,
      "step": 178860
    },
    {
      "epoch": 73.22,
      "grad_norm": 1.5137090682983398,
      "learning_rate": 3.51951519743591e-06,
      "loss": 0.5068,
      "step": 178870
    },
    {
      "epoch": 73.22,
      "grad_norm": 2.8352930545806885,
      "learning_rate": 3.519377969874557e-06,
      "loss": 0.5173,
      "step": 178880
    },
    {
      "epoch": 73.23,
      "grad_norm": 1.657015323638916,
      "learning_rate": 3.5192407369278336e-06,
      "loss": 0.4999,
      "step": 178890
    },
    {
      "epoch": 73.23,
      "grad_norm": 1.537333369255066,
      "learning_rate": 3.519103498596369e-06,
      "loss": 0.5124,
      "step": 178900
    },
    {
      "epoch": 73.23,
      "grad_norm": 1.7860935926437378,
      "learning_rate": 3.5189662548807917e-06,
      "loss": 0.513,
      "step": 178910
    },
    {
      "epoch": 73.24,
      "grad_norm": 1.5741548538208008,
      "learning_rate": 3.518829005781731e-06,
      "loss": 0.5088,
      "step": 178920
    },
    {
      "epoch": 73.24,
      "grad_norm": 1.7167835235595703,
      "learning_rate": 3.518691751299816e-06,
      "loss": 0.5118,
      "step": 178930
    },
    {
      "epoch": 73.25,
      "grad_norm": 1.9327195882797241,
      "learning_rate": 3.5185544914356738e-06,
      "loss": 0.5005,
      "step": 178940
    },
    {
      "epoch": 73.25,
      "grad_norm": 2.100813865661621,
      "learning_rate": 3.518417226189935e-06,
      "loss": 0.5138,
      "step": 178950
    },
    {
      "epoch": 73.25,
      "grad_norm": 2.3143789768218994,
      "learning_rate": 3.518279955563227e-06,
      "loss": 0.5199,
      "step": 178960
    },
    {
      "epoch": 73.26,
      "grad_norm": 1.8485385179519653,
      "learning_rate": 3.5181426795561794e-06,
      "loss": 0.4963,
      "step": 178970
    },
    {
      "epoch": 73.26,
      "grad_norm": 1.6905171871185303,
      "learning_rate": 3.5180053981694205e-06,
      "loss": 0.5031,
      "step": 178980
    },
    {
      "epoch": 73.27,
      "grad_norm": 2.442544460296631,
      "learning_rate": 3.51786811140358e-06,
      "loss": 0.4999,
      "step": 178990
    },
    {
      "epoch": 73.27,
      "grad_norm": 1.5448414087295532,
      "learning_rate": 3.517730819259286e-06,
      "loss": 0.5108,
      "step": 179000
    },
    {
      "epoch": 73.27,
      "grad_norm": 2.188040256500244,
      "learning_rate": 3.5175935217371674e-06,
      "loss": 0.5066,
      "step": 179010
    },
    {
      "epoch": 73.28,
      "grad_norm": 1.8476964235305786,
      "learning_rate": 3.5174562188378536e-06,
      "loss": 0.5168,
      "step": 179020
    },
    {
      "epoch": 73.28,
      "grad_norm": 1.5454577207565308,
      "learning_rate": 3.517318910561974e-06,
      "loss": 0.5559,
      "step": 179030
    },
    {
      "epoch": 73.29,
      "grad_norm": 2.165811538696289,
      "learning_rate": 3.5171815969101565e-06,
      "loss": 0.5362,
      "step": 179040
    },
    {
      "epoch": 73.29,
      "grad_norm": 1.6907565593719482,
      "learning_rate": 3.51704427788303e-06,
      "loss": 0.4814,
      "step": 179050
    },
    {
      "epoch": 73.3,
      "grad_norm": 2.1145246028900146,
      "learning_rate": 3.516906953481225e-06,
      "loss": 0.5016,
      "step": 179060
    },
    {
      "epoch": 73.3,
      "grad_norm": 1.9749836921691895,
      "learning_rate": 3.5167696237053696e-06,
      "loss": 0.5026,
      "step": 179070
    },
    {
      "epoch": 73.3,
      "grad_norm": 2.048952102661133,
      "learning_rate": 3.516632288556093e-06,
      "loss": 0.5273,
      "step": 179080
    },
    {
      "epoch": 73.31,
      "grad_norm": 2.2030889987945557,
      "learning_rate": 3.516494948034023e-06,
      "loss": 0.5248,
      "step": 179090
    },
    {
      "epoch": 73.31,
      "grad_norm": 2.167257070541382,
      "learning_rate": 3.5163576021397923e-06,
      "loss": 0.4938,
      "step": 179100
    },
    {
      "epoch": 73.32,
      "grad_norm": 2.0760207176208496,
      "learning_rate": 3.5162202508740265e-06,
      "loss": 0.5247,
      "step": 179110
    },
    {
      "epoch": 73.32,
      "grad_norm": 1.6689261198043823,
      "learning_rate": 3.5160828942373557e-06,
      "loss": 0.5221,
      "step": 179120
    },
    {
      "epoch": 73.32,
      "grad_norm": 1.759941816329956,
      "learning_rate": 3.51594553223041e-06,
      "loss": 0.5215,
      "step": 179130
    },
    {
      "epoch": 73.33,
      "grad_norm": 1.8903456926345825,
      "learning_rate": 3.515808164853817e-06,
      "loss": 0.5271,
      "step": 179140
    },
    {
      "epoch": 73.33,
      "grad_norm": 2.0571017265319824,
      "learning_rate": 3.5156707921082085e-06,
      "loss": 0.5081,
      "step": 179150
    },
    {
      "epoch": 73.34,
      "grad_norm": 2.8953189849853516,
      "learning_rate": 3.515533413994212e-06,
      "loss": 0.5174,
      "step": 179160
    },
    {
      "epoch": 73.34,
      "grad_norm": 2.2028064727783203,
      "learning_rate": 3.5153960305124575e-06,
      "loss": 0.5152,
      "step": 179170
    },
    {
      "epoch": 73.34,
      "grad_norm": 2.3823721408843994,
      "learning_rate": 3.5152586416635735e-06,
      "loss": 0.5028,
      "step": 179180
    },
    {
      "epoch": 73.35,
      "grad_norm": 1.9800682067871094,
      "learning_rate": 3.51512124744819e-06,
      "loss": 0.5161,
      "step": 179190
    },
    {
      "epoch": 73.35,
      "grad_norm": 2.4306201934814453,
      "learning_rate": 3.514983847866936e-06,
      "loss": 0.5149,
      "step": 179200
    },
    {
      "epoch": 73.36,
      "grad_norm": 1.8805902004241943,
      "learning_rate": 3.514846442920441e-06,
      "loss": 0.5425,
      "step": 179210
    },
    {
      "epoch": 73.36,
      "grad_norm": 1.685359001159668,
      "learning_rate": 3.514709032609335e-06,
      "loss": 0.5132,
      "step": 179220
    },
    {
      "epoch": 73.36,
      "grad_norm": 2.0638911724090576,
      "learning_rate": 3.5145716169342466e-06,
      "loss": 0.4981,
      "step": 179230
    },
    {
      "epoch": 73.37,
      "grad_norm": 1.801527500152588,
      "learning_rate": 3.5144341958958058e-06,
      "loss": 0.5139,
      "step": 179240
    },
    {
      "epoch": 73.37,
      "grad_norm": 1.6796362400054932,
      "learning_rate": 3.514296769494642e-06,
      "loss": 0.5061,
      "step": 179250
    },
    {
      "epoch": 73.38,
      "grad_norm": 2.600708484649658,
      "learning_rate": 3.5141593377313845e-06,
      "loss": 0.4964,
      "step": 179260
    },
    {
      "epoch": 73.38,
      "grad_norm": 1.7275077104568481,
      "learning_rate": 3.5140219006066636e-06,
      "loss": 0.5263,
      "step": 179270
    },
    {
      "epoch": 73.39,
      "grad_norm": 2.020613431930542,
      "learning_rate": 3.5138844581211084e-06,
      "loss": 0.5163,
      "step": 179280
    },
    {
      "epoch": 73.39,
      "grad_norm": 1.8393996953964233,
      "learning_rate": 3.5137470102753484e-06,
      "loss": 0.5196,
      "step": 179290
    },
    {
      "epoch": 73.39,
      "grad_norm": 2.0775134563446045,
      "learning_rate": 3.5136095570700127e-06,
      "loss": 0.5034,
      "step": 179300
    },
    {
      "epoch": 73.4,
      "grad_norm": 2.35683012008667,
      "learning_rate": 3.513472098505732e-06,
      "loss": 0.4917,
      "step": 179310
    },
    {
      "epoch": 73.4,
      "grad_norm": 1.4938393831253052,
      "learning_rate": 3.513334634583136e-06,
      "loss": 0.5092,
      "step": 179320
    },
    {
      "epoch": 73.41,
      "grad_norm": 1.5320994853973389,
      "learning_rate": 3.5131971653028527e-06,
      "loss": 0.5039,
      "step": 179330
    },
    {
      "epoch": 73.41,
      "grad_norm": 1.543013572692871,
      "learning_rate": 3.5130596906655138e-06,
      "loss": 0.4993,
      "step": 179340
    },
    {
      "epoch": 73.41,
      "grad_norm": 2.2579243183135986,
      "learning_rate": 3.5129222106717485e-06,
      "loss": 0.5139,
      "step": 179350
    },
    {
      "epoch": 73.42,
      "grad_norm": 1.347885012626648,
      "learning_rate": 3.512784725322186e-06,
      "loss": 0.514,
      "step": 179360
    },
    {
      "epoch": 73.42,
      "grad_norm": 2.2292304039001465,
      "learning_rate": 3.5126472346174566e-06,
      "loss": 0.506,
      "step": 179370
    },
    {
      "epoch": 73.43,
      "grad_norm": 2.1451632976531982,
      "learning_rate": 3.5125097385581903e-06,
      "loss": 0.5154,
      "step": 179380
    },
    {
      "epoch": 73.43,
      "grad_norm": 1.9909080266952515,
      "learning_rate": 3.5123722371450163e-06,
      "loss": 0.5051,
      "step": 179390
    },
    {
      "epoch": 73.43,
      "grad_norm": 1.917741298675537,
      "learning_rate": 3.512234730378565e-06,
      "loss": 0.5124,
      "step": 179400
    },
    {
      "epoch": 73.44,
      "grad_norm": 1.917618751525879,
      "learning_rate": 3.5120972182594653e-06,
      "loss": 0.4851,
      "step": 179410
    },
    {
      "epoch": 73.44,
      "grad_norm": 1.77726411819458,
      "learning_rate": 3.5119597007883487e-06,
      "loss": 0.5336,
      "step": 179420
    },
    {
      "epoch": 73.45,
      "grad_norm": 2.1695313453674316,
      "learning_rate": 3.511822177965844e-06,
      "loss": 0.5073,
      "step": 179430
    },
    {
      "epoch": 73.45,
      "grad_norm": 1.613104224205017,
      "learning_rate": 3.5116846497925817e-06,
      "loss": 0.5197,
      "step": 179440
    },
    {
      "epoch": 73.45,
      "grad_norm": 2.037132501602173,
      "learning_rate": 3.511547116269192e-06,
      "loss": 0.5261,
      "step": 179450
    },
    {
      "epoch": 73.46,
      "grad_norm": 2.4762866497039795,
      "learning_rate": 3.511409577396305e-06,
      "loss": 0.5265,
      "step": 179460
    },
    {
      "epoch": 73.46,
      "grad_norm": 1.5661070346832275,
      "learning_rate": 3.5112720331745496e-06,
      "loss": 0.485,
      "step": 179470
    },
    {
      "epoch": 73.47,
      "grad_norm": 1.7304723262786865,
      "learning_rate": 3.511134483604556e-06,
      "loss": 0.498,
      "step": 179480
    },
    {
      "epoch": 73.47,
      "grad_norm": 2.1486001014709473,
      "learning_rate": 3.5109969286869562e-06,
      "loss": 0.5233,
      "step": 179490
    },
    {
      "epoch": 73.48,
      "grad_norm": 1.6192699670791626,
      "learning_rate": 3.5108593684223778e-06,
      "loss": 0.5179,
      "step": 179500
    },
    {
      "epoch": 73.48,
      "grad_norm": 1.9392346143722534,
      "learning_rate": 3.510721802811453e-06,
      "loss": 0.5166,
      "step": 179510
    },
    {
      "epoch": 73.48,
      "grad_norm": 1.3655496835708618,
      "learning_rate": 3.5105842318548104e-06,
      "loss": 0.5124,
      "step": 179520
    },
    {
      "epoch": 73.49,
      "grad_norm": 1.5911259651184082,
      "learning_rate": 3.5104466555530814e-06,
      "loss": 0.523,
      "step": 179530
    },
    {
      "epoch": 73.49,
      "grad_norm": 1.7591650485992432,
      "learning_rate": 3.510309073906895e-06,
      "loss": 0.506,
      "step": 179540
    },
    {
      "epoch": 73.5,
      "grad_norm": 1.5179367065429688,
      "learning_rate": 3.510171486916883e-06,
      "loss": 0.5146,
      "step": 179550
    },
    {
      "epoch": 73.5,
      "grad_norm": 2.058058023452759,
      "learning_rate": 3.5100338945836745e-06,
      "loss": 0.5063,
      "step": 179560
    },
    {
      "epoch": 73.5,
      "grad_norm": 1.8423068523406982,
      "learning_rate": 3.5098962969079e-06,
      "loss": 0.5295,
      "step": 179570
    },
    {
      "epoch": 73.51,
      "grad_norm": 2.204238176345825,
      "learning_rate": 3.5097586938901904e-06,
      "loss": 0.5165,
      "step": 179580
    },
    {
      "epoch": 73.51,
      "grad_norm": 1.4333873987197876,
      "learning_rate": 3.5096210855311748e-06,
      "loss": 0.536,
      "step": 179590
    },
    {
      "epoch": 73.52,
      "grad_norm": 1.756763219833374,
      "learning_rate": 3.5094834718314847e-06,
      "loss": 0.5092,
      "step": 179600
    },
    {
      "epoch": 73.52,
      "grad_norm": 2.121772527694702,
      "learning_rate": 3.5093458527917504e-06,
      "loss": 0.541,
      "step": 179610
    },
    {
      "epoch": 73.52,
      "grad_norm": 1.689719796180725,
      "learning_rate": 3.509208228412601e-06,
      "loss": 0.5183,
      "step": 179620
    },
    {
      "epoch": 73.53,
      "grad_norm": 1.9107507467269897,
      "learning_rate": 3.509070598694669e-06,
      "loss": 0.5376,
      "step": 179630
    },
    {
      "epoch": 73.53,
      "grad_norm": 1.7032850980758667,
      "learning_rate": 3.5089329636385834e-06,
      "loss": 0.5105,
      "step": 179640
    },
    {
      "epoch": 73.54,
      "grad_norm": 1.7469881772994995,
      "learning_rate": 3.5087953232449753e-06,
      "loss": 0.5281,
      "step": 179650
    },
    {
      "epoch": 73.54,
      "grad_norm": 1.674124002456665,
      "learning_rate": 3.5086576775144748e-06,
      "loss": 0.507,
      "step": 179660
    },
    {
      "epoch": 73.54,
      "grad_norm": 1.9210090637207031,
      "learning_rate": 3.5085200264477124e-06,
      "loss": 0.5087,
      "step": 179670
    },
    {
      "epoch": 73.55,
      "grad_norm": 2.270164966583252,
      "learning_rate": 3.50838237004532e-06,
      "loss": 0.5164,
      "step": 179680
    },
    {
      "epoch": 73.55,
      "grad_norm": 1.8460747003555298,
      "learning_rate": 3.5082447083079262e-06,
      "loss": 0.5006,
      "step": 179690
    },
    {
      "epoch": 73.56,
      "grad_norm": 1.822083830833435,
      "learning_rate": 3.508107041236162e-06,
      "loss": 0.5129,
      "step": 179700
    },
    {
      "epoch": 73.56,
      "grad_norm": 1.9291431903839111,
      "learning_rate": 3.507969368830659e-06,
      "loss": 0.5127,
      "step": 179710
    },
    {
      "epoch": 73.57,
      "grad_norm": 1.8593988418579102,
      "learning_rate": 3.5078316910920475e-06,
      "loss": 0.5105,
      "step": 179720
    },
    {
      "epoch": 73.57,
      "grad_norm": 1.9662925004959106,
      "learning_rate": 3.507694008020958e-06,
      "loss": 0.5064,
      "step": 179730
    },
    {
      "epoch": 73.57,
      "grad_norm": 1.6287158727645874,
      "learning_rate": 3.507556319618021e-06,
      "loss": 0.5209,
      "step": 179740
    },
    {
      "epoch": 73.58,
      "grad_norm": 1.6923092603683472,
      "learning_rate": 3.507418625883868e-06,
      "loss": 0.4857,
      "step": 179750
    },
    {
      "epoch": 73.58,
      "grad_norm": 1.851263165473938,
      "learning_rate": 3.507280926819129e-06,
      "loss": 0.5287,
      "step": 179760
    },
    {
      "epoch": 73.59,
      "grad_norm": 1.9722298383712769,
      "learning_rate": 3.5071432224244352e-06,
      "loss": 0.4979,
      "step": 179770
    },
    {
      "epoch": 73.59,
      "grad_norm": 1.6080985069274902,
      "learning_rate": 3.5070055127004168e-06,
      "loss": 0.4881,
      "step": 179780
    },
    {
      "epoch": 73.59,
      "grad_norm": 2.042346239089966,
      "learning_rate": 3.5068677976477056e-06,
      "loss": 0.526,
      "step": 179790
    },
    {
      "epoch": 73.6,
      "grad_norm": 2.2901008129119873,
      "learning_rate": 3.506730077266931e-06,
      "loss": 0.5034,
      "step": 179800
    },
    {
      "epoch": 73.6,
      "grad_norm": 2.2458174228668213,
      "learning_rate": 3.5065923515587253e-06,
      "loss": 0.4942,
      "step": 179810
    },
    {
      "epoch": 73.61,
      "grad_norm": 2.343204975128174,
      "learning_rate": 3.5064546205237196e-06,
      "loss": 0.507,
      "step": 179820
    },
    {
      "epoch": 73.61,
      "grad_norm": 2.2081332206726074,
      "learning_rate": 3.506316884162543e-06,
      "loss": 0.5366,
      "step": 179830
    },
    {
      "epoch": 73.61,
      "grad_norm": 1.85776948928833,
      "learning_rate": 3.5061791424758283e-06,
      "loss": 0.539,
      "step": 179840
    },
    {
      "epoch": 73.62,
      "grad_norm": 1.816582441329956,
      "learning_rate": 3.506041395464205e-06,
      "loss": 0.5185,
      "step": 179850
    },
    {
      "epoch": 73.62,
      "grad_norm": 1.3407236337661743,
      "learning_rate": 3.505903643128305e-06,
      "loss": 0.5217,
      "step": 179860
    },
    {
      "epoch": 73.63,
      "grad_norm": 1.7672308683395386,
      "learning_rate": 3.5057658854687597e-06,
      "loss": 0.5425,
      "step": 179870
    },
    {
      "epoch": 73.63,
      "grad_norm": 1.9657049179077148,
      "learning_rate": 3.505628122486199e-06,
      "loss": 0.5138,
      "step": 179880
    },
    {
      "epoch": 73.63,
      "grad_norm": 1.8143097162246704,
      "learning_rate": 3.5054903541812546e-06,
      "loss": 0.5189,
      "step": 179890
    },
    {
      "epoch": 73.64,
      "grad_norm": 1.8259612321853638,
      "learning_rate": 3.505352580554558e-06,
      "loss": 0.5195,
      "step": 179900
    },
    {
      "epoch": 73.64,
      "grad_norm": 1.8508009910583496,
      "learning_rate": 3.5052148016067396e-06,
      "loss": 0.5208,
      "step": 179910
    },
    {
      "epoch": 73.65,
      "grad_norm": 1.5464926958084106,
      "learning_rate": 3.5050770173384304e-06,
      "loss": 0.5157,
      "step": 179920
    },
    {
      "epoch": 73.65,
      "grad_norm": 1.648016095161438,
      "learning_rate": 3.504939227750263e-06,
      "loss": 0.5379,
      "step": 179930
    },
    {
      "epoch": 73.66,
      "grad_norm": 1.371349573135376,
      "learning_rate": 3.504801432842867e-06,
      "loss": 0.5045,
      "step": 179940
    },
    {
      "epoch": 73.66,
      "grad_norm": 1.8854726552963257,
      "learning_rate": 3.504663632616874e-06,
      "loss": 0.5422,
      "step": 179950
    },
    {
      "epoch": 73.66,
      "grad_norm": 1.6069040298461914,
      "learning_rate": 3.5045258270729154e-06,
      "loss": 0.5029,
      "step": 179960
    },
    {
      "epoch": 73.67,
      "grad_norm": 1.4361900091171265,
      "learning_rate": 3.5043880162116236e-06,
      "loss": 0.508,
      "step": 179970
    },
    {
      "epoch": 73.67,
      "grad_norm": 1.759535551071167,
      "learning_rate": 3.504250200033628e-06,
      "loss": 0.5004,
      "step": 179980
    },
    {
      "epoch": 73.68,
      "grad_norm": 2.6580874919891357,
      "learning_rate": 3.50411237853956e-06,
      "loss": 0.508,
      "step": 179990
    },
    {
      "epoch": 73.68,
      "grad_norm": 1.9030036926269531,
      "learning_rate": 3.5039745517300526e-06,
      "loss": 0.5156,
      "step": 180000
    },
    {
      "epoch": 73.68,
      "grad_norm": 1.6015088558197021,
      "learning_rate": 3.503836719605736e-06,
      "loss": 0.511,
      "step": 180010
    },
    {
      "epoch": 73.69,
      "grad_norm": 1.9348883628845215,
      "learning_rate": 3.5036988821672416e-06,
      "loss": 0.5246,
      "step": 180020
    },
    {
      "epoch": 73.69,
      "grad_norm": 1.9199296236038208,
      "learning_rate": 3.5035610394152008e-06,
      "loss": 0.5059,
      "step": 180030
    },
    {
      "epoch": 73.7,
      "grad_norm": 2.24651837348938,
      "learning_rate": 3.5034231913502466e-06,
      "loss": 0.5039,
      "step": 180040
    },
    {
      "epoch": 73.7,
      "grad_norm": 1.575089454650879,
      "learning_rate": 3.5032853379730075e-06,
      "loss": 0.5043,
      "step": 180050
    },
    {
      "epoch": 73.7,
      "grad_norm": 2.287430763244629,
      "learning_rate": 3.5031474792841177e-06,
      "loss": 0.5055,
      "step": 180060
    },
    {
      "epoch": 73.71,
      "grad_norm": 1.909921407699585,
      "learning_rate": 3.5030096152842067e-06,
      "loss": 0.4991,
      "step": 180070
    },
    {
      "epoch": 73.71,
      "grad_norm": 1.7412534952163696,
      "learning_rate": 3.502871745973907e-06,
      "loss": 0.5149,
      "step": 180080
    },
    {
      "epoch": 73.72,
      "grad_norm": 2.245748519897461,
      "learning_rate": 3.5027338713538502e-06,
      "loss": 0.5057,
      "step": 180090
    },
    {
      "epoch": 73.72,
      "grad_norm": 1.7561113834381104,
      "learning_rate": 3.502595991424668e-06,
      "loss": 0.5254,
      "step": 180100
    },
    {
      "epoch": 73.72,
      "grad_norm": 1.560873031616211,
      "learning_rate": 3.5024581061869925e-06,
      "loss": 0.5279,
      "step": 180110
    },
    {
      "epoch": 73.73,
      "grad_norm": 1.646871566772461,
      "learning_rate": 3.502320215641454e-06,
      "loss": 0.5353,
      "step": 180120
    },
    {
      "epoch": 73.73,
      "grad_norm": 2.202826976776123,
      "learning_rate": 3.502182319788684e-06,
      "loss": 0.512,
      "step": 180130
    },
    {
      "epoch": 73.74,
      "grad_norm": 2.258119583129883,
      "learning_rate": 3.502044418629316e-06,
      "loss": 0.4988,
      "step": 180140
    },
    {
      "epoch": 73.74,
      "grad_norm": 2.0150301456451416,
      "learning_rate": 3.50190651216398e-06,
      "loss": 0.495,
      "step": 180150
    },
    {
      "epoch": 73.75,
      "grad_norm": 1.703976035118103,
      "learning_rate": 3.501768600393309e-06,
      "loss": 0.5142,
      "step": 180160
    },
    {
      "epoch": 73.75,
      "grad_norm": 1.6912163496017456,
      "learning_rate": 3.5016306833179335e-06,
      "loss": 0.5076,
      "step": 180170
    },
    {
      "epoch": 73.75,
      "grad_norm": 1.9566316604614258,
      "learning_rate": 3.501492760938486e-06,
      "loss": 0.5163,
      "step": 180180
    },
    {
      "epoch": 73.76,
      "grad_norm": 1.8292723894119263,
      "learning_rate": 3.5013548332555985e-06,
      "loss": 0.5193,
      "step": 180190
    },
    {
      "epoch": 73.76,
      "grad_norm": 1.9885526895523071,
      "learning_rate": 3.5012169002699023e-06,
      "loss": 0.5053,
      "step": 180200
    },
    {
      "epoch": 73.77,
      "grad_norm": 1.6649444103240967,
      "learning_rate": 3.5010789619820297e-06,
      "loss": 0.5019,
      "step": 180210
    },
    {
      "epoch": 73.77,
      "grad_norm": 1.9844133853912354,
      "learning_rate": 3.5009410183926126e-06,
      "loss": 0.5148,
      "step": 180220
    },
    {
      "epoch": 73.77,
      "grad_norm": 2.044407844543457,
      "learning_rate": 3.5008030695022825e-06,
      "loss": 0.5161,
      "step": 180230
    },
    {
      "epoch": 73.78,
      "grad_norm": 1.8204752206802368,
      "learning_rate": 3.500665115311671e-06,
      "loss": 0.5061,
      "step": 180240
    },
    {
      "epoch": 73.78,
      "grad_norm": 2.165658950805664,
      "learning_rate": 3.5005271558214108e-06,
      "loss": 0.5025,
      "step": 180250
    },
    {
      "epoch": 73.79,
      "grad_norm": 2.006044864654541,
      "learning_rate": 3.500389191032134e-06,
      "loss": 0.5094,
      "step": 180260
    },
    {
      "epoch": 73.79,
      "grad_norm": 1.9623075723648071,
      "learning_rate": 3.5002512209444716e-06,
      "loss": 0.527,
      "step": 180270
    },
    {
      "epoch": 73.79,
      "grad_norm": 1.5143942832946777,
      "learning_rate": 3.500113245559056e-06,
      "loss": 0.5422,
      "step": 180280
    },
    {
      "epoch": 73.8,
      "grad_norm": 2.0753204822540283,
      "learning_rate": 3.4999752648765203e-06,
      "loss": 0.5189,
      "step": 180290
    },
    {
      "epoch": 73.8,
      "grad_norm": 2.1243391036987305,
      "learning_rate": 3.4998372788974948e-06,
      "loss": 0.5225,
      "step": 180300
    },
    {
      "epoch": 73.81,
      "grad_norm": 2.0601892471313477,
      "learning_rate": 3.4996992876226135e-06,
      "loss": 0.5077,
      "step": 180310
    },
    {
      "epoch": 73.81,
      "grad_norm": 2.278242349624634,
      "learning_rate": 3.499561291052507e-06,
      "loss": 0.5326,
      "step": 180320
    },
    {
      "epoch": 73.81,
      "grad_norm": 1.9395204782485962,
      "learning_rate": 3.4994232891878084e-06,
      "loss": 0.507,
      "step": 180330
    },
    {
      "epoch": 73.82,
      "grad_norm": 1.8870736360549927,
      "learning_rate": 3.499285282029149e-06,
      "loss": 0.5073,
      "step": 180340
    },
    {
      "epoch": 73.82,
      "grad_norm": 1.9266663789749146,
      "learning_rate": 3.499147269577162e-06,
      "loss": 0.5006,
      "step": 180350
    },
    {
      "epoch": 73.83,
      "grad_norm": 1.8588987588882446,
      "learning_rate": 3.4990092518324785e-06,
      "loss": 0.5027,
      "step": 180360
    },
    {
      "epoch": 73.83,
      "grad_norm": 2.1768362522125244,
      "learning_rate": 3.4988712287957306e-06,
      "loss": 0.5365,
      "step": 180370
    },
    {
      "epoch": 73.84,
      "grad_norm": 1.7999892234802246,
      "learning_rate": 3.498733200467552e-06,
      "loss": 0.5142,
      "step": 180380
    },
    {
      "epoch": 73.84,
      "grad_norm": 1.8331149816513062,
      "learning_rate": 3.4985951668485744e-06,
      "loss": 0.5109,
      "step": 180390
    },
    {
      "epoch": 73.84,
      "grad_norm": 2.03817081451416,
      "learning_rate": 3.49845712793943e-06,
      "loss": 0.5237,
      "step": 180400
    },
    {
      "epoch": 73.85,
      "grad_norm": 1.6330198049545288,
      "learning_rate": 3.4983190837407514e-06,
      "loss": 0.5074,
      "step": 180410
    },
    {
      "epoch": 73.85,
      "grad_norm": 1.6014965772628784,
      "learning_rate": 3.49818103425317e-06,
      "loss": 0.4899,
      "step": 180420
    },
    {
      "epoch": 73.86,
      "grad_norm": 1.9729185104370117,
      "learning_rate": 3.4980429794773187e-06,
      "loss": 0.4888,
      "step": 180430
    },
    {
      "epoch": 73.86,
      "grad_norm": 1.9562269449234009,
      "learning_rate": 3.49790491941383e-06,
      "loss": 0.5187,
      "step": 180440
    },
    {
      "epoch": 73.86,
      "grad_norm": 1.6328171491622925,
      "learning_rate": 3.497766854063337e-06,
      "loss": 0.5089,
      "step": 180450
    },
    {
      "epoch": 73.87,
      "grad_norm": 2.111856460571289,
      "learning_rate": 3.4976287834264706e-06,
      "loss": 0.5292,
      "step": 180460
    },
    {
      "epoch": 73.87,
      "grad_norm": 2.3985886573791504,
      "learning_rate": 3.4974907075038646e-06,
      "loss": 0.4872,
      "step": 180470
    },
    {
      "epoch": 73.88,
      "grad_norm": 1.8921067714691162,
      "learning_rate": 3.497352626296151e-06,
      "loss": 0.5111,
      "step": 180480
    },
    {
      "epoch": 73.88,
      "grad_norm": 1.9766466617584229,
      "learning_rate": 3.497214539803963e-06,
      "loss": 0.5081,
      "step": 180490
    },
    {
      "epoch": 73.88,
      "grad_norm": 1.9821618795394897,
      "learning_rate": 3.497076448027932e-06,
      "loss": 0.5075,
      "step": 180500
    },
    {
      "epoch": 73.89,
      "grad_norm": 2.040321111679077,
      "learning_rate": 3.4969383509686914e-06,
      "loss": 0.5323,
      "step": 180510
    },
    {
      "epoch": 73.89,
      "grad_norm": 1.7814552783966064,
      "learning_rate": 3.496800248626874e-06,
      "loss": 0.5104,
      "step": 180520
    },
    {
      "epoch": 73.9,
      "grad_norm": 2.0420634746551514,
      "learning_rate": 3.4966621410031113e-06,
      "loss": 0.4988,
      "step": 180530
    },
    {
      "epoch": 73.9,
      "grad_norm": 1.8373773097991943,
      "learning_rate": 3.4965240280980367e-06,
      "loss": 0.5085,
      "step": 180540
    },
    {
      "epoch": 73.91,
      "grad_norm": 2.061615228652954,
      "learning_rate": 3.4963859099122834e-06,
      "loss": 0.5095,
      "step": 180550
    },
    {
      "epoch": 73.91,
      "grad_norm": 1.8944613933563232,
      "learning_rate": 3.496247786446482e-06,
      "loss": 0.5049,
      "step": 180560
    },
    {
      "epoch": 73.91,
      "grad_norm": 1.970078468322754,
      "learning_rate": 3.4961096577012684e-06,
      "loss": 0.4951,
      "step": 180570
    },
    {
      "epoch": 73.92,
      "grad_norm": 1.5879778861999512,
      "learning_rate": 3.495971523677273e-06,
      "loss": 0.5202,
      "step": 180580
    },
    {
      "epoch": 73.92,
      "grad_norm": 1.7213258743286133,
      "learning_rate": 3.4958333843751293e-06,
      "loss": 0.5251,
      "step": 180590
    },
    {
      "epoch": 73.93,
      "grad_norm": 1.5684425830841064,
      "learning_rate": 3.49569523979547e-06,
      "loss": 0.4934,
      "step": 180600
    },
    {
      "epoch": 73.93,
      "grad_norm": 1.643649935722351,
      "learning_rate": 3.4955570899389277e-06,
      "loss": 0.5248,
      "step": 180610
    },
    {
      "epoch": 73.93,
      "grad_norm": 1.9815657138824463,
      "learning_rate": 3.495418934806136e-06,
      "loss": 0.5145,
      "step": 180620
    },
    {
      "epoch": 73.94,
      "grad_norm": 1.7697372436523438,
      "learning_rate": 3.4952807743977274e-06,
      "loss": 0.52,
      "step": 180630
    },
    {
      "epoch": 73.94,
      "grad_norm": 2.001324415206909,
      "learning_rate": 3.4951426087143344e-06,
      "loss": 0.4991,
      "step": 180640
    },
    {
      "epoch": 73.95,
      "grad_norm": 1.7373861074447632,
      "learning_rate": 3.4950044377565896e-06,
      "loss": 0.5187,
      "step": 180650
    },
    {
      "epoch": 73.95,
      "grad_norm": 2.139169216156006,
      "learning_rate": 3.494866261525127e-06,
      "loss": 0.5144,
      "step": 180660
    },
    {
      "epoch": 73.95,
      "grad_norm": 2.1545748710632324,
      "learning_rate": 3.494728080020579e-06,
      "loss": 0.5051,
      "step": 180670
    },
    {
      "epoch": 73.96,
      "grad_norm": 1.5723857879638672,
      "learning_rate": 3.4945898932435788e-06,
      "loss": 0.5123,
      "step": 180680
    },
    {
      "epoch": 73.96,
      "grad_norm": 1.5661286115646362,
      "learning_rate": 3.4944517011947598e-06,
      "loss": 0.5059,
      "step": 180690
    },
    {
      "epoch": 73.97,
      "grad_norm": 1.6867563724517822,
      "learning_rate": 3.494313503874754e-06,
      "loss": 0.5161,
      "step": 180700
    },
    {
      "epoch": 73.97,
      "grad_norm": 1.736872673034668,
      "learning_rate": 3.4941753012841944e-06,
      "loss": 0.5183,
      "step": 180710
    },
    {
      "epoch": 73.97,
      "grad_norm": 2.1042115688323975,
      "learning_rate": 3.4940370934237156e-06,
      "loss": 0.5031,
      "step": 180720
    },
    {
      "epoch": 73.98,
      "grad_norm": 1.799613118171692,
      "learning_rate": 3.4938988802939495e-06,
      "loss": 0.5253,
      "step": 180730
    },
    {
      "epoch": 73.98,
      "grad_norm": 2.29744029045105,
      "learning_rate": 3.493760661895529e-06,
      "loss": 0.525,
      "step": 180740
    },
    {
      "epoch": 73.99,
      "grad_norm": 1.6965644359588623,
      "learning_rate": 3.493622438229088e-06,
      "loss": 0.5227,
      "step": 180750
    },
    {
      "epoch": 73.99,
      "grad_norm": 2.2530031204223633,
      "learning_rate": 3.49348420929526e-06,
      "loss": 0.5103,
      "step": 180760
    },
    {
      "epoch": 74.0,
      "grad_norm": 1.9707456827163696,
      "learning_rate": 3.4933459750946768e-06,
      "loss": 0.5082,
      "step": 180770
    },
    {
      "epoch": 74.0,
      "grad_norm": 1.6960926055908203,
      "learning_rate": 3.4932077356279725e-06,
      "loss": 0.4947,
      "step": 180780
    },
    {
      "epoch": 74.0,
      "eval_loss": 0.5112026929855347,
      "eval_runtime": 73.6079,
      "eval_samples_per_second": 46.856,
      "eval_steps_per_second": 5.869,
      "step": 180782
    },
    {
      "epoch": 74.0,
      "grad_norm": 1.546612024307251,
      "learning_rate": 3.493069490895781e-06,
      "loss": 0.5134,
      "step": 180790
    },
    {
      "epoch": 74.01,
      "grad_norm": 2.086601495742798,
      "learning_rate": 3.4929312408987344e-06,
      "loss": 0.5104,
      "step": 180800
    },
    {
      "epoch": 74.01,
      "grad_norm": 1.758173942565918,
      "learning_rate": 3.492792985637467e-06,
      "loss": 0.5,
      "step": 180810
    },
    {
      "epoch": 74.02,
      "grad_norm": 2.2101757526397705,
      "learning_rate": 3.4926547251126114e-06,
      "loss": 0.5326,
      "step": 180820
    },
    {
      "epoch": 74.02,
      "grad_norm": 1.6797789335250854,
      "learning_rate": 3.4925164593248013e-06,
      "loss": 0.5006,
      "step": 180830
    },
    {
      "epoch": 74.02,
      "grad_norm": 1.4378851652145386,
      "learning_rate": 3.49237818827467e-06,
      "loss": 0.5096,
      "step": 180840
    },
    {
      "epoch": 74.03,
      "grad_norm": 1.6502970457077026,
      "learning_rate": 3.49223991196285e-06,
      "loss": 0.4814,
      "step": 180850
    },
    {
      "epoch": 74.03,
      "grad_norm": 2.4484009742736816,
      "learning_rate": 3.4921016303899767e-06,
      "loss": 0.5214,
      "step": 180860
    },
    {
      "epoch": 74.04,
      "grad_norm": 1.684946894645691,
      "learning_rate": 3.491963343556682e-06,
      "loss": 0.5184,
      "step": 180870
    },
    {
      "epoch": 74.04,
      "grad_norm": 1.699235200881958,
      "learning_rate": 3.4918250514635994e-06,
      "loss": 0.5122,
      "step": 180880
    },
    {
      "epoch": 74.04,
      "grad_norm": 1.5414429903030396,
      "learning_rate": 3.4916867541113632e-06,
      "loss": 0.4974,
      "step": 180890
    },
    {
      "epoch": 74.05,
      "grad_norm": 2.0445077419281006,
      "learning_rate": 3.4915484515006063e-06,
      "loss": 0.526,
      "step": 180900
    },
    {
      "epoch": 74.05,
      "grad_norm": 2.088330030441284,
      "learning_rate": 3.491410143631963e-06,
      "loss": 0.528,
      "step": 180910
    },
    {
      "epoch": 74.06,
      "grad_norm": 1.7561269998550415,
      "learning_rate": 3.491271830506066e-06,
      "loss": 0.4932,
      "step": 180920
    },
    {
      "epoch": 74.06,
      "grad_norm": 1.7711400985717773,
      "learning_rate": 3.4911335121235486e-06,
      "loss": 0.4939,
      "step": 180930
    },
    {
      "epoch": 74.06,
      "grad_norm": 1.6499649286270142,
      "learning_rate": 3.4909951884850454e-06,
      "loss": 0.5268,
      "step": 180940
    },
    {
      "epoch": 74.07,
      "grad_norm": 1.9604692459106445,
      "learning_rate": 3.49085685959119e-06,
      "loss": 0.4818,
      "step": 180950
    },
    {
      "epoch": 74.07,
      "grad_norm": 1.667312502861023,
      "learning_rate": 3.490718525442616e-06,
      "loss": 0.521,
      "step": 180960
    },
    {
      "epoch": 74.08,
      "grad_norm": 1.907899260520935,
      "learning_rate": 3.490580186039956e-06,
      "loss": 0.5285,
      "step": 180970
    },
    {
      "epoch": 74.08,
      "grad_norm": 1.3543980121612549,
      "learning_rate": 3.4904418413838454e-06,
      "loss": 0.5178,
      "step": 180980
    },
    {
      "epoch": 74.09,
      "grad_norm": 2.14578914642334,
      "learning_rate": 3.490303491474916e-06,
      "loss": 0.5142,
      "step": 180990
    },
    {
      "epoch": 74.09,
      "grad_norm": 1.7611876726150513,
      "learning_rate": 3.4901651363138035e-06,
      "loss": 0.5017,
      "step": 181000
    },
    {
      "epoch": 74.09,
      "grad_norm": 1.5794827938079834,
      "learning_rate": 3.4900267759011404e-06,
      "loss": 0.5127,
      "step": 181010
    },
    {
      "epoch": 74.1,
      "grad_norm": 1.4873390197753906,
      "learning_rate": 3.4898884102375613e-06,
      "loss": 0.521,
      "step": 181020
    },
    {
      "epoch": 74.1,
      "grad_norm": 2.3338847160339355,
      "learning_rate": 3.489750039323699e-06,
      "loss": 0.5274,
      "step": 181030
    },
    {
      "epoch": 74.11,
      "grad_norm": 1.858809232711792,
      "learning_rate": 3.4896116631601884e-06,
      "loss": 0.5041,
      "step": 181040
    },
    {
      "epoch": 74.11,
      "grad_norm": 1.8012609481811523,
      "learning_rate": 3.489473281747663e-06,
      "loss": 0.5119,
      "step": 181050
    },
    {
      "epoch": 74.11,
      "grad_norm": 1.5944325923919678,
      "learning_rate": 3.4893348950867563e-06,
      "loss": 0.5052,
      "step": 181060
    },
    {
      "epoch": 74.12,
      "grad_norm": 3.8202624320983887,
      "learning_rate": 3.489196503178103e-06,
      "loss": 0.519,
      "step": 181070
    },
    {
      "epoch": 74.12,
      "grad_norm": 2.2271571159362793,
      "learning_rate": 3.4890581060223367e-06,
      "loss": 0.511,
      "step": 181080
    },
    {
      "epoch": 74.13,
      "grad_norm": 2.321662664413452,
      "learning_rate": 3.4889197036200914e-06,
      "loss": 0.4899,
      "step": 181090
    },
    {
      "epoch": 74.13,
      "grad_norm": 1.9230412244796753,
      "learning_rate": 3.4887812959720008e-06,
      "loss": 0.5072,
      "step": 181100
    },
    {
      "epoch": 74.13,
      "grad_norm": 1.7207592725753784,
      "learning_rate": 3.488642883078699e-06,
      "loss": 0.5328,
      "step": 181110
    },
    {
      "epoch": 74.14,
      "grad_norm": 1.4618198871612549,
      "learning_rate": 3.4885044649408205e-06,
      "loss": 0.5135,
      "step": 181120
    },
    {
      "epoch": 74.14,
      "grad_norm": 1.7094838619232178,
      "learning_rate": 3.4883660415589993e-06,
      "loss": 0.5309,
      "step": 181130
    },
    {
      "epoch": 74.15,
      "grad_norm": 3.377530813217163,
      "learning_rate": 3.4882276129338688e-06,
      "loss": 0.5149,
      "step": 181140
    },
    {
      "epoch": 74.15,
      "grad_norm": 1.6602251529693604,
      "learning_rate": 3.488089179066064e-06,
      "loss": 0.5245,
      "step": 181150
    },
    {
      "epoch": 74.15,
      "grad_norm": 1.5257805585861206,
      "learning_rate": 3.4879507399562185e-06,
      "loss": 0.4983,
      "step": 181160
    },
    {
      "epoch": 74.16,
      "grad_norm": 2.2754483222961426,
      "learning_rate": 3.4878122956049663e-06,
      "loss": 0.5276,
      "step": 181170
    },
    {
      "epoch": 74.16,
      "grad_norm": 1.6863104104995728,
      "learning_rate": 3.487673846012942e-06,
      "loss": 0.5082,
      "step": 181180
    },
    {
      "epoch": 74.17,
      "grad_norm": 1.6114226579666138,
      "learning_rate": 3.48753539118078e-06,
      "loss": 0.5233,
      "step": 181190
    },
    {
      "epoch": 74.17,
      "grad_norm": 1.2888234853744507,
      "learning_rate": 3.4873969311091146e-06,
      "loss": 0.5206,
      "step": 181200
    },
    {
      "epoch": 74.18,
      "grad_norm": 2.915100336074829,
      "learning_rate": 3.487258465798579e-06,
      "loss": 0.5026,
      "step": 181210
    },
    {
      "epoch": 74.18,
      "grad_norm": 2.188857316970825,
      "learning_rate": 3.4871199952498085e-06,
      "loss": 0.5038,
      "step": 181220
    },
    {
      "epoch": 74.18,
      "grad_norm": 1.8453795909881592,
      "learning_rate": 3.486981519463437e-06,
      "loss": 0.5103,
      "step": 181230
    },
    {
      "epoch": 74.19,
      "grad_norm": 1.8901112079620361,
      "learning_rate": 3.486843038440099e-06,
      "loss": 0.5058,
      "step": 181240
    },
    {
      "epoch": 74.19,
      "grad_norm": 2.0559515953063965,
      "learning_rate": 3.4867045521804286e-06,
      "loss": 0.505,
      "step": 181250
    },
    {
      "epoch": 74.2,
      "grad_norm": 1.9671460390090942,
      "learning_rate": 3.486566060685061e-06,
      "loss": 0.4926,
      "step": 181260
    },
    {
      "epoch": 74.2,
      "grad_norm": 1.9307334423065186,
      "learning_rate": 3.48642756395463e-06,
      "loss": 0.5162,
      "step": 181270
    },
    {
      "epoch": 74.2,
      "grad_norm": 1.991374135017395,
      "learning_rate": 3.48628906198977e-06,
      "loss": 0.508,
      "step": 181280
    },
    {
      "epoch": 74.21,
      "grad_norm": 1.7440454959869385,
      "learning_rate": 3.486150554791115e-06,
      "loss": 0.5126,
      "step": 181290
    },
    {
      "epoch": 74.21,
      "grad_norm": 1.8472943305969238,
      "learning_rate": 3.4860120423593e-06,
      "loss": 0.5143,
      "step": 181300
    },
    {
      "epoch": 74.22,
      "grad_norm": 1.417321801185608,
      "learning_rate": 3.4858735246949592e-06,
      "loss": 0.5203,
      "step": 181310
    },
    {
      "epoch": 74.22,
      "grad_norm": 2.4749715328216553,
      "learning_rate": 3.485735001798728e-06,
      "loss": 0.506,
      "step": 181320
    },
    {
      "epoch": 74.22,
      "grad_norm": 2.370126247406006,
      "learning_rate": 3.48559647367124e-06,
      "loss": 0.5238,
      "step": 181330
    },
    {
      "epoch": 74.23,
      "grad_norm": 2.1304421424865723,
      "learning_rate": 3.4854579403131314e-06,
      "loss": 0.5378,
      "step": 181340
    },
    {
      "epoch": 74.23,
      "grad_norm": 2.1110026836395264,
      "learning_rate": 3.4853194017250347e-06,
      "loss": 0.5168,
      "step": 181350
    },
    {
      "epoch": 74.24,
      "grad_norm": 1.861999750137329,
      "learning_rate": 3.485180857907585e-06,
      "loss": 0.5326,
      "step": 181360
    },
    {
      "epoch": 74.24,
      "grad_norm": 2.0272586345672607,
      "learning_rate": 3.4850423088614175e-06,
      "loss": 0.5252,
      "step": 181370
    },
    {
      "epoch": 74.24,
      "grad_norm": 1.791474461555481,
      "learning_rate": 3.4849037545871668e-06,
      "loss": 0.5182,
      "step": 181380
    },
    {
      "epoch": 74.25,
      "grad_norm": 2.173034906387329,
      "learning_rate": 3.4847651950854675e-06,
      "loss": 0.5181,
      "step": 181390
    },
    {
      "epoch": 74.25,
      "grad_norm": 1.801550030708313,
      "learning_rate": 3.4846266303569544e-06,
      "loss": 0.5062,
      "step": 181400
    },
    {
      "epoch": 74.26,
      "grad_norm": 1.845140814781189,
      "learning_rate": 3.4844880604022616e-06,
      "loss": 0.5297,
      "step": 181410
    },
    {
      "epoch": 74.26,
      "grad_norm": 2.125101089477539,
      "learning_rate": 3.4843494852220248e-06,
      "loss": 0.5144,
      "step": 181420
    },
    {
      "epoch": 74.27,
      "grad_norm": 1.9716718196868896,
      "learning_rate": 3.4842109048168783e-06,
      "loss": 0.518,
      "step": 181430
    },
    {
      "epoch": 74.27,
      "grad_norm": 1.5263787508010864,
      "learning_rate": 3.4840723191874567e-06,
      "loss": 0.4931,
      "step": 181440
    },
    {
      "epoch": 74.27,
      "grad_norm": 1.938826322555542,
      "learning_rate": 3.483933728334396e-06,
      "loss": 0.5053,
      "step": 181450
    },
    {
      "epoch": 74.28,
      "grad_norm": 1.9077011346817017,
      "learning_rate": 3.4837951322583295e-06,
      "loss": 0.5238,
      "step": 181460
    },
    {
      "epoch": 74.28,
      "grad_norm": 1.6562278270721436,
      "learning_rate": 3.483656530959893e-06,
      "loss": 0.511,
      "step": 181470
    },
    {
      "epoch": 74.29,
      "grad_norm": 1.8341610431671143,
      "learning_rate": 3.4835179244397215e-06,
      "loss": 0.4971,
      "step": 181480
    },
    {
      "epoch": 74.29,
      "grad_norm": 1.6212605237960815,
      "learning_rate": 3.4833793126984497e-06,
      "loss": 0.542,
      "step": 181490
    },
    {
      "epoch": 74.29,
      "grad_norm": 1.9115848541259766,
      "learning_rate": 3.483240695736712e-06,
      "loss": 0.5165,
      "step": 181500
    },
    {
      "epoch": 74.3,
      "grad_norm": 1.7547534704208374,
      "learning_rate": 3.483102073555144e-06,
      "loss": 0.5026,
      "step": 181510
    },
    {
      "epoch": 74.3,
      "grad_norm": 1.8190631866455078,
      "learning_rate": 3.482963446154381e-06,
      "loss": 0.501,
      "step": 181520
    },
    {
      "epoch": 74.31,
      "grad_norm": 2.5011143684387207,
      "learning_rate": 3.482824813535057e-06,
      "loss": 0.5061,
      "step": 181530
    },
    {
      "epoch": 74.31,
      "grad_norm": 1.8816992044448853,
      "learning_rate": 3.4826861756978087e-06,
      "loss": 0.5217,
      "step": 181540
    },
    {
      "epoch": 74.31,
      "grad_norm": 1.542527198791504,
      "learning_rate": 3.4825475326432696e-06,
      "loss": 0.4923,
      "step": 181550
    },
    {
      "epoch": 74.32,
      "grad_norm": 1.9895167350769043,
      "learning_rate": 3.4824088843720753e-06,
      "loss": 0.4967,
      "step": 181560
    },
    {
      "epoch": 74.32,
      "grad_norm": 1.9158002138137817,
      "learning_rate": 3.4822702308848613e-06,
      "loss": 0.5002,
      "step": 181570
    },
    {
      "epoch": 74.33,
      "grad_norm": 1.9942444562911987,
      "learning_rate": 3.482131572182262e-06,
      "loss": 0.5266,
      "step": 181580
    },
    {
      "epoch": 74.33,
      "grad_norm": 1.8064309358596802,
      "learning_rate": 3.481992908264913e-06,
      "loss": 0.517,
      "step": 181590
    },
    {
      "epoch": 74.33,
      "grad_norm": 1.7705682516098022,
      "learning_rate": 3.4818542391334494e-06,
      "loss": 0.5174,
      "step": 181600
    },
    {
      "epoch": 74.34,
      "grad_norm": 1.509734034538269,
      "learning_rate": 3.481715564788507e-06,
      "loss": 0.5023,
      "step": 181610
    },
    {
      "epoch": 74.34,
      "grad_norm": 1.6661548614501953,
      "learning_rate": 3.4815768852307205e-06,
      "loss": 0.5119,
      "step": 181620
    },
    {
      "epoch": 74.35,
      "grad_norm": 1.7602026462554932,
      "learning_rate": 3.4814382004607253e-06,
      "loss": 0.515,
      "step": 181630
    },
    {
      "epoch": 74.35,
      "grad_norm": 2.2652039527893066,
      "learning_rate": 3.4812995104791565e-06,
      "loss": 0.5305,
      "step": 181640
    },
    {
      "epoch": 74.36,
      "grad_norm": 1.8752797842025757,
      "learning_rate": 3.4811608152866496e-06,
      "loss": 0.5058,
      "step": 181650
    },
    {
      "epoch": 74.36,
      "grad_norm": 1.718510389328003,
      "learning_rate": 3.48102211488384e-06,
      "loss": 0.5018,
      "step": 181660
    },
    {
      "epoch": 74.36,
      "grad_norm": 1.7500101327896118,
      "learning_rate": 3.4808834092713625e-06,
      "loss": 0.4898,
      "step": 181670
    },
    {
      "epoch": 74.37,
      "grad_norm": 2.175615072250366,
      "learning_rate": 3.4807446984498534e-06,
      "loss": 0.513,
      "step": 181680
    },
    {
      "epoch": 74.37,
      "grad_norm": 1.7312289476394653,
      "learning_rate": 3.480605982419947e-06,
      "loss": 0.5239,
      "step": 181690
    },
    {
      "epoch": 74.38,
      "grad_norm": 1.9177336692810059,
      "learning_rate": 3.4804672611822807e-06,
      "loss": 0.4982,
      "step": 181700
    },
    {
      "epoch": 74.38,
      "grad_norm": 1.6122397184371948,
      "learning_rate": 3.480328534737488e-06,
      "loss": 0.4992,
      "step": 181710
    },
    {
      "epoch": 74.38,
      "grad_norm": 1.805660605430603,
      "learning_rate": 3.4801898030862048e-06,
      "loss": 0.4899,
      "step": 181720
    },
    {
      "epoch": 74.39,
      "grad_norm": 2.213862895965576,
      "learning_rate": 3.480051066229067e-06,
      "loss": 0.5041,
      "step": 181730
    },
    {
      "epoch": 74.39,
      "grad_norm": 1.8300036191940308,
      "learning_rate": 3.4799123241667102e-06,
      "loss": 0.504,
      "step": 181740
    },
    {
      "epoch": 74.4,
      "grad_norm": 2.0991361141204834,
      "learning_rate": 3.47977357689977e-06,
      "loss": 0.5056,
      "step": 181750
    },
    {
      "epoch": 74.4,
      "grad_norm": 2.601109266281128,
      "learning_rate": 3.479634824428881e-06,
      "loss": 0.5172,
      "step": 181760
    },
    {
      "epoch": 74.4,
      "grad_norm": 1.80966055393219,
      "learning_rate": 3.47949606675468e-06,
      "loss": 0.5136,
      "step": 181770
    },
    {
      "epoch": 74.41,
      "grad_norm": 2.160304307937622,
      "learning_rate": 3.4793573038778017e-06,
      "loss": 0.508,
      "step": 181780
    },
    {
      "epoch": 74.41,
      "grad_norm": 1.7281361818313599,
      "learning_rate": 3.4792185357988826e-06,
      "loss": 0.5037,
      "step": 181790
    },
    {
      "epoch": 74.42,
      "grad_norm": 1.8598543405532837,
      "learning_rate": 3.4790797625185577e-06,
      "loss": 0.4997,
      "step": 181800
    },
    {
      "epoch": 74.42,
      "grad_norm": 1.995863437652588,
      "learning_rate": 3.4789409840374637e-06,
      "loss": 0.5142,
      "step": 181810
    },
    {
      "epoch": 74.42,
      "grad_norm": 2.1679091453552246,
      "learning_rate": 3.478802200356235e-06,
      "loss": 0.5061,
      "step": 181820
    },
    {
      "epoch": 74.43,
      "grad_norm": 2.185302734375,
      "learning_rate": 3.478663411475508e-06,
      "loss": 0.5159,
      "step": 181830
    },
    {
      "epoch": 74.43,
      "grad_norm": 1.6703507900238037,
      "learning_rate": 3.4785246173959185e-06,
      "loss": 0.5019,
      "step": 181840
    },
    {
      "epoch": 74.44,
      "grad_norm": 1.7023253440856934,
      "learning_rate": 3.4783858181181027e-06,
      "loss": 0.5011,
      "step": 181850
    },
    {
      "epoch": 74.44,
      "grad_norm": 1.8930644989013672,
      "learning_rate": 3.4782470136426953e-06,
      "loss": 0.5136,
      "step": 181860
    },
    {
      "epoch": 74.45,
      "grad_norm": 1.8547697067260742,
      "learning_rate": 3.478108203970333e-06,
      "loss": 0.5182,
      "step": 181870
    },
    {
      "epoch": 74.45,
      "grad_norm": 1.729715347290039,
      "learning_rate": 3.477969389101651e-06,
      "loss": 0.5096,
      "step": 181880
    },
    {
      "epoch": 74.45,
      "grad_norm": 1.6344200372695923,
      "learning_rate": 3.4778305690372858e-06,
      "loss": 0.5138,
      "step": 181890
    },
    {
      "epoch": 74.46,
      "grad_norm": 2.3222999572753906,
      "learning_rate": 3.477691743777874e-06,
      "loss": 0.5144,
      "step": 181900
    },
    {
      "epoch": 74.46,
      "grad_norm": 2.320474863052368,
      "learning_rate": 3.4775529133240495e-06,
      "loss": 0.4983,
      "step": 181910
    },
    {
      "epoch": 74.47,
      "grad_norm": 1.698975682258606,
      "learning_rate": 3.4774140776764508e-06,
      "loss": 0.5071,
      "step": 181920
    },
    {
      "epoch": 74.47,
      "grad_norm": 1.8570395708084106,
      "learning_rate": 3.477275236835712e-06,
      "loss": 0.5157,
      "step": 181930
    },
    {
      "epoch": 74.47,
      "grad_norm": 1.8917709589004517,
      "learning_rate": 3.4771363908024695e-06,
      "loss": 0.4884,
      "step": 181940
    },
    {
      "epoch": 74.48,
      "grad_norm": 1.8136547803878784,
      "learning_rate": 3.476997539577359e-06,
      "loss": 0.5346,
      "step": 181950
    },
    {
      "epoch": 74.48,
      "grad_norm": 1.731631875038147,
      "learning_rate": 3.4768586831610174e-06,
      "loss": 0.5055,
      "step": 181960
    },
    {
      "epoch": 74.49,
      "grad_norm": 1.9494621753692627,
      "learning_rate": 3.4767198215540797e-06,
      "loss": 0.5063,
      "step": 181970
    },
    {
      "epoch": 74.49,
      "grad_norm": 1.9862980842590332,
      "learning_rate": 3.4765809547571835e-06,
      "loss": 0.4887,
      "step": 181980
    },
    {
      "epoch": 74.49,
      "grad_norm": 1.465603232383728,
      "learning_rate": 3.476442082770964e-06,
      "loss": 0.4969,
      "step": 181990
    },
    {
      "epoch": 74.5,
      "grad_norm": 3.2349908351898193,
      "learning_rate": 3.476303205596058e-06,
      "loss": 0.5021,
      "step": 182000
    },
    {
      "epoch": 74.5,
      "grad_norm": 2.0971386432647705,
      "learning_rate": 3.4761643232331006e-06,
      "loss": 0.5345,
      "step": 182010
    },
    {
      "epoch": 74.51,
      "grad_norm": 1.861315131187439,
      "learning_rate": 3.4760254356827288e-06,
      "loss": 0.5003,
      "step": 182020
    },
    {
      "epoch": 74.51,
      "grad_norm": 1.3120577335357666,
      "learning_rate": 3.475886542945578e-06,
      "loss": 0.5007,
      "step": 182030
    },
    {
      "epoch": 74.51,
      "grad_norm": 2.3555397987365723,
      "learning_rate": 3.475747645022286e-06,
      "loss": 0.5176,
      "step": 182040
    },
    {
      "epoch": 74.52,
      "grad_norm": 1.5587854385375977,
      "learning_rate": 3.4756087419134874e-06,
      "loss": 0.5137,
      "step": 182050
    },
    {
      "epoch": 74.52,
      "grad_norm": 1.4924819469451904,
      "learning_rate": 3.4754698336198195e-06,
      "loss": 0.5073,
      "step": 182060
    },
    {
      "epoch": 74.53,
      "grad_norm": 2.2358529567718506,
      "learning_rate": 3.475330920141918e-06,
      "loss": 0.4965,
      "step": 182070
    },
    {
      "epoch": 74.53,
      "grad_norm": 1.7951693534851074,
      "learning_rate": 3.47519200148042e-06,
      "loss": 0.5104,
      "step": 182080
    },
    {
      "epoch": 74.54,
      "grad_norm": 2.358665704727173,
      "learning_rate": 3.4750530776359603e-06,
      "loss": 0.5247,
      "step": 182090
    },
    {
      "epoch": 74.54,
      "grad_norm": 2.0304512977600098,
      "learning_rate": 3.4749141486091777e-06,
      "loss": 0.5117,
      "step": 182100
    },
    {
      "epoch": 74.54,
      "grad_norm": 4.036208629608154,
      "learning_rate": 3.474775214400707e-06,
      "loss": 0.5065,
      "step": 182110
    },
    {
      "epoch": 74.55,
      "grad_norm": 2.03338885307312,
      "learning_rate": 3.4746362750111848e-06,
      "loss": 0.5089,
      "step": 182120
    },
    {
      "epoch": 74.55,
      "grad_norm": 1.7917219400405884,
      "learning_rate": 3.4744973304412476e-06,
      "loss": 0.5143,
      "step": 182130
    },
    {
      "epoch": 74.56,
      "grad_norm": 1.8372670412063599,
      "learning_rate": 3.474358380691532e-06,
      "loss": 0.5178,
      "step": 182140
    },
    {
      "epoch": 74.56,
      "grad_norm": 2.0450704097747803,
      "learning_rate": 3.474219425762675e-06,
      "loss": 0.4983,
      "step": 182150
    },
    {
      "epoch": 74.56,
      "grad_norm": 1.6200714111328125,
      "learning_rate": 3.4740804656553118e-06,
      "loss": 0.5002,
      "step": 182160
    },
    {
      "epoch": 74.57,
      "grad_norm": 2.1367292404174805,
      "learning_rate": 3.4739415003700802e-06,
      "loss": 0.4924,
      "step": 182170
    },
    {
      "epoch": 74.57,
      "grad_norm": 1.6413296461105347,
      "learning_rate": 3.4738025299076164e-06,
      "loss": 0.5064,
      "step": 182180
    },
    {
      "epoch": 74.58,
      "grad_norm": 2.2968273162841797,
      "learning_rate": 3.4736635542685573e-06,
      "loss": 0.5152,
      "step": 182190
    },
    {
      "epoch": 74.58,
      "grad_norm": 1.8668729066848755,
      "learning_rate": 3.4735245734535387e-06,
      "loss": 0.5208,
      "step": 182200
    },
    {
      "epoch": 74.58,
      "grad_norm": 1.8301957845687866,
      "learning_rate": 3.4733855874631987e-06,
      "loss": 0.522,
      "step": 182210
    },
    {
      "epoch": 74.59,
      "grad_norm": 1.6829575300216675,
      "learning_rate": 3.4732465962981718e-06,
      "loss": 0.5176,
      "step": 182220
    },
    {
      "epoch": 74.59,
      "grad_norm": 2.1776511669158936,
      "learning_rate": 3.4731075999590962e-06,
      "loss": 0.5014,
      "step": 182230
    },
    {
      "epoch": 74.6,
      "grad_norm": 1.5382559299468994,
      "learning_rate": 3.4729685984466085e-06,
      "loss": 0.5289,
      "step": 182240
    },
    {
      "epoch": 74.6,
      "grad_norm": 2.301532030105591,
      "learning_rate": 3.472829591761345e-06,
      "loss": 0.5235,
      "step": 182250
    },
    {
      "epoch": 74.6,
      "grad_norm": 2.236790657043457,
      "learning_rate": 3.4726905799039434e-06,
      "loss": 0.5275,
      "step": 182260
    },
    {
      "epoch": 74.61,
      "grad_norm": 1.7109403610229492,
      "learning_rate": 3.472551562875039e-06,
      "loss": 0.4977,
      "step": 182270
    },
    {
      "epoch": 74.61,
      "grad_norm": 1.795692801475525,
      "learning_rate": 3.4724125406752705e-06,
      "loss": 0.5175,
      "step": 182280
    },
    {
      "epoch": 74.62,
      "grad_norm": 2.013859272003174,
      "learning_rate": 3.4722735133052734e-06,
      "loss": 0.5093,
      "step": 182290
    },
    {
      "epoch": 74.62,
      "grad_norm": 1.4397846460342407,
      "learning_rate": 3.4721344807656842e-06,
      "loss": 0.5252,
      "step": 182300
    },
    {
      "epoch": 74.63,
      "grad_norm": 1.8029366731643677,
      "learning_rate": 3.4719954430571413e-06,
      "loss": 0.5117,
      "step": 182310
    },
    {
      "epoch": 74.63,
      "grad_norm": 1.3800023794174194,
      "learning_rate": 3.47185640018028e-06,
      "loss": 0.5133,
      "step": 182320
    },
    {
      "epoch": 74.63,
      "grad_norm": 1.6838825941085815,
      "learning_rate": 3.4717173521357387e-06,
      "loss": 0.5106,
      "step": 182330
    },
    {
      "epoch": 74.64,
      "grad_norm": 1.4372845888137817,
      "learning_rate": 3.4715782989241525e-06,
      "loss": 0.5143,
      "step": 182340
    },
    {
      "epoch": 74.64,
      "grad_norm": 1.77462637424469,
      "learning_rate": 3.4714392405461604e-06,
      "loss": 0.5239,
      "step": 182350
    },
    {
      "epoch": 74.65,
      "grad_norm": 2.224322557449341,
      "learning_rate": 3.471300177002398e-06,
      "loss": 0.516,
      "step": 182360
    },
    {
      "epoch": 74.65,
      "grad_norm": 1.5589885711669922,
      "learning_rate": 3.471161108293504e-06,
      "loss": 0.5352,
      "step": 182370
    },
    {
      "epoch": 74.65,
      "grad_norm": 1.6311018466949463,
      "learning_rate": 3.4710220344201135e-06,
      "loss": 0.4858,
      "step": 182380
    },
    {
      "epoch": 74.66,
      "grad_norm": 1.5284788608551025,
      "learning_rate": 3.4708829553828646e-06,
      "loss": 0.5097,
      "step": 182390
    },
    {
      "epoch": 74.66,
      "grad_norm": 2.1497201919555664,
      "learning_rate": 3.470743871182394e-06,
      "loss": 0.5063,
      "step": 182400
    },
    {
      "epoch": 74.67,
      "grad_norm": 1.6944427490234375,
      "learning_rate": 3.4706047818193385e-06,
      "loss": 0.5098,
      "step": 182410
    },
    {
      "epoch": 74.67,
      "grad_norm": 1.7435898780822754,
      "learning_rate": 3.4704656872943365e-06,
      "loss": 0.5221,
      "step": 182420
    },
    {
      "epoch": 74.67,
      "grad_norm": 2.2735161781311035,
      "learning_rate": 3.4703265876080244e-06,
      "loss": 0.5069,
      "step": 182430
    },
    {
      "epoch": 74.68,
      "grad_norm": 2.132511854171753,
      "learning_rate": 3.4701874827610396e-06,
      "loss": 0.5179,
      "step": 182440
    },
    {
      "epoch": 74.68,
      "grad_norm": 1.581534743309021,
      "learning_rate": 3.4700483727540185e-06,
      "loss": 0.5066,
      "step": 182450
    },
    {
      "epoch": 74.69,
      "grad_norm": 1.6704291105270386,
      "learning_rate": 3.469909257587599e-06,
      "loss": 0.5339,
      "step": 182460
    },
    {
      "epoch": 74.69,
      "grad_norm": 1.747265100479126,
      "learning_rate": 3.469770137262419e-06,
      "loss": 0.528,
      "step": 182470
    },
    {
      "epoch": 74.7,
      "grad_norm": 2.0685508251190186,
      "learning_rate": 3.4696310117791146e-06,
      "loss": 0.518,
      "step": 182480
    },
    {
      "epoch": 74.7,
      "grad_norm": 1.770615577697754,
      "learning_rate": 3.4694918811383235e-06,
      "loss": 0.5143,
      "step": 182490
    },
    {
      "epoch": 74.7,
      "grad_norm": 2.058736562728882,
      "learning_rate": 3.469352745340684e-06,
      "loss": 0.5194,
      "step": 182500
    },
    {
      "epoch": 74.71,
      "grad_norm": 1.8473565578460693,
      "learning_rate": 3.4692136043868327e-06,
      "loss": 0.5064,
      "step": 182510
    },
    {
      "epoch": 74.71,
      "grad_norm": 1.9703410863876343,
      "learning_rate": 3.4690744582774065e-06,
      "loss": 0.5015,
      "step": 182520
    },
    {
      "epoch": 74.72,
      "grad_norm": 2.023266315460205,
      "learning_rate": 3.4689353070130433e-06,
      "loss": 0.5037,
      "step": 182530
    },
    {
      "epoch": 74.72,
      "grad_norm": 1.6844991445541382,
      "learning_rate": 3.4687961505943795e-06,
      "loss": 0.5194,
      "step": 182540
    },
    {
      "epoch": 74.72,
      "grad_norm": 1.8326504230499268,
      "learning_rate": 3.468656989022054e-06,
      "loss": 0.5123,
      "step": 182550
    },
    {
      "epoch": 74.73,
      "grad_norm": 1.488412618637085,
      "learning_rate": 3.468517822296704e-06,
      "loss": 0.4993,
      "step": 182560
    },
    {
      "epoch": 74.73,
      "grad_norm": 2.03379225730896,
      "learning_rate": 3.468378650418968e-06,
      "loss": 0.5056,
      "step": 182570
    },
    {
      "epoch": 74.74,
      "grad_norm": 2.000553846359253,
      "learning_rate": 3.468239473389481e-06,
      "loss": 0.5229,
      "step": 182580
    },
    {
      "epoch": 74.74,
      "grad_norm": 1.7542346715927124,
      "learning_rate": 3.4681002912088823e-06,
      "loss": 0.5002,
      "step": 182590
    },
    {
      "epoch": 74.74,
      "grad_norm": 1.9586374759674072,
      "learning_rate": 3.4679611038778087e-06,
      "loss": 0.4987,
      "step": 182600
    },
    {
      "epoch": 74.75,
      "grad_norm": 2.2715647220611572,
      "learning_rate": 3.467821911396898e-06,
      "loss": 0.5087,
      "step": 182610
    },
    {
      "epoch": 74.75,
      "grad_norm": 1.9227900505065918,
      "learning_rate": 3.4676827137667886e-06,
      "loss": 0.5201,
      "step": 182620
    },
    {
      "epoch": 74.76,
      "grad_norm": 2.006740093231201,
      "learning_rate": 3.4675435109881164e-06,
      "loss": 0.5187,
      "step": 182630
    },
    {
      "epoch": 74.76,
      "grad_norm": 1.688560962677002,
      "learning_rate": 3.4674043030615214e-06,
      "loss": 0.5162,
      "step": 182640
    },
    {
      "epoch": 74.76,
      "grad_norm": 1.6493672132492065,
      "learning_rate": 3.4672650899876397e-06,
      "loss": 0.4904,
      "step": 182650
    },
    {
      "epoch": 74.77,
      "grad_norm": 2.4255518913269043,
      "learning_rate": 3.4671258717671088e-06,
      "loss": 0.525,
      "step": 182660
    },
    {
      "epoch": 74.77,
      "grad_norm": 1.595849633216858,
      "learning_rate": 3.4669866484005677e-06,
      "loss": 0.5078,
      "step": 182670
    },
    {
      "epoch": 74.78,
      "grad_norm": 2.25777268409729,
      "learning_rate": 3.4668474198886538e-06,
      "loss": 0.5028,
      "step": 182680
    },
    {
      "epoch": 74.78,
      "grad_norm": 1.7722458839416504,
      "learning_rate": 3.4667081862320036e-06,
      "loss": 0.5186,
      "step": 182690
    },
    {
      "epoch": 74.79,
      "grad_norm": 2.0274581909179688,
      "learning_rate": 3.4665689474312563e-06,
      "loss": 0.51,
      "step": 182700
    },
    {
      "epoch": 74.79,
      "grad_norm": 1.8588038682937622,
      "learning_rate": 3.4664297034870493e-06,
      "loss": 0.5346,
      "step": 182710
    },
    {
      "epoch": 74.79,
      "grad_norm": 1.644139289855957,
      "learning_rate": 3.4662904544000203e-06,
      "loss": 0.4994,
      "step": 182720
    },
    {
      "epoch": 74.8,
      "grad_norm": 1.3596323728561401,
      "learning_rate": 3.4661512001708075e-06,
      "loss": 0.5175,
      "step": 182730
    },
    {
      "epoch": 74.8,
      "grad_norm": 1.7722376585006714,
      "learning_rate": 3.466011940800048e-06,
      "loss": 0.5391,
      "step": 182740
    },
    {
      "epoch": 74.81,
      "grad_norm": 2.049481153488159,
      "learning_rate": 3.465872676288381e-06,
      "loss": 0.5256,
      "step": 182750
    },
    {
      "epoch": 74.81,
      "grad_norm": 1.4979273080825806,
      "learning_rate": 3.4657334066364436e-06,
      "loss": 0.5025,
      "step": 182760
    },
    {
      "epoch": 74.81,
      "grad_norm": 2.2076327800750732,
      "learning_rate": 3.465594131844874e-06,
      "loss": 0.5078,
      "step": 182770
    },
    {
      "epoch": 74.82,
      "grad_norm": 2.1496098041534424,
      "learning_rate": 3.4654548519143106e-06,
      "loss": 0.512,
      "step": 182780
    },
    {
      "epoch": 74.82,
      "grad_norm": 1.3070154190063477,
      "learning_rate": 3.4653155668453906e-06,
      "loss": 0.5164,
      "step": 182790
    },
    {
      "epoch": 74.83,
      "grad_norm": 2.6827163696289062,
      "learning_rate": 3.4651762766387533e-06,
      "loss": 0.5004,
      "step": 182800
    },
    {
      "epoch": 74.83,
      "grad_norm": 1.819034457206726,
      "learning_rate": 3.4650369812950348e-06,
      "loss": 0.5095,
      "step": 182810
    },
    {
      "epoch": 74.83,
      "grad_norm": 1.6029839515686035,
      "learning_rate": 3.4648976808148738e-06,
      "loss": 0.5437,
      "step": 182820
    },
    {
      "epoch": 74.84,
      "grad_norm": 2.5286433696746826,
      "learning_rate": 3.4647583751989097e-06,
      "loss": 0.4933,
      "step": 182830
    },
    {
      "epoch": 74.84,
      "grad_norm": 1.8662406206130981,
      "learning_rate": 3.46461906444778e-06,
      "loss": 0.5083,
      "step": 182840
    },
    {
      "epoch": 74.85,
      "grad_norm": 2.0146307945251465,
      "learning_rate": 3.4644797485621222e-06,
      "loss": 0.5134,
      "step": 182850
    },
    {
      "epoch": 74.85,
      "grad_norm": 1.8795758485794067,
      "learning_rate": 3.4643404275425757e-06,
      "loss": 0.5057,
      "step": 182860
    },
    {
      "epoch": 74.85,
      "grad_norm": 2.4391167163848877,
      "learning_rate": 3.4642011013897778e-06,
      "loss": 0.4937,
      "step": 182870
    },
    {
      "epoch": 74.86,
      "grad_norm": 2.1100986003875732,
      "learning_rate": 3.4640617701043668e-06,
      "loss": 0.5093,
      "step": 182880
    },
    {
      "epoch": 74.86,
      "grad_norm": 1.634259581565857,
      "learning_rate": 3.463922433686981e-06,
      "loss": 0.507,
      "step": 182890
    },
    {
      "epoch": 74.87,
      "grad_norm": 1.8549424409866333,
      "learning_rate": 3.463783092138259e-06,
      "loss": 0.5008,
      "step": 182900
    },
    {
      "epoch": 74.87,
      "grad_norm": 1.7484339475631714,
      "learning_rate": 3.4636437454588377e-06,
      "loss": 0.5101,
      "step": 182910
    },
    {
      "epoch": 74.88,
      "grad_norm": 1.864814281463623,
      "learning_rate": 3.4635043936493576e-06,
      "loss": 0.5131,
      "step": 182920
    },
    {
      "epoch": 74.88,
      "grad_norm": 2.0874276161193848,
      "learning_rate": 3.4633650367104565e-06,
      "loss": 0.5172,
      "step": 182930
    },
    {
      "epoch": 74.88,
      "grad_norm": 2.431506633758545,
      "learning_rate": 3.4632256746427717e-06,
      "loss": 0.5206,
      "step": 182940
    },
    {
      "epoch": 74.89,
      "grad_norm": 2.1675078868865967,
      "learning_rate": 3.4630863074469424e-06,
      "loss": 0.5162,
      "step": 182950
    },
    {
      "epoch": 74.89,
      "grad_norm": 1.5856380462646484,
      "learning_rate": 3.4629469351236063e-06,
      "loss": 0.5172,
      "step": 182960
    },
    {
      "epoch": 74.9,
      "grad_norm": 1.649803876876831,
      "learning_rate": 3.4628075576734034e-06,
      "loss": 0.5228,
      "step": 182970
    },
    {
      "epoch": 74.9,
      "grad_norm": 2.415140151977539,
      "learning_rate": 3.46266817509697e-06,
      "loss": 0.5058,
      "step": 182980
    },
    {
      "epoch": 74.9,
      "grad_norm": 1.8634839057922363,
      "learning_rate": 3.4625287873949463e-06,
      "loss": 0.5161,
      "step": 182990
    },
    {
      "epoch": 74.91,
      "grad_norm": 2.291313409805298,
      "learning_rate": 3.4623893945679703e-06,
      "loss": 0.492,
      "step": 183000
    },
    {
      "epoch": 74.91,
      "grad_norm": 1.8722255229949951,
      "learning_rate": 3.46224999661668e-06,
      "loss": 0.4949,
      "step": 183010
    },
    {
      "epoch": 74.92,
      "grad_norm": 1.7909038066864014,
      "learning_rate": 3.4621105935417153e-06,
      "loss": 0.5103,
      "step": 183020
    },
    {
      "epoch": 74.92,
      "grad_norm": 2.072567939758301,
      "learning_rate": 3.4619711853437133e-06,
      "loss": 0.5011,
      "step": 183030
    },
    {
      "epoch": 74.92,
      "grad_norm": 1.5220504999160767,
      "learning_rate": 3.461831772023313e-06,
      "loss": 0.4967,
      "step": 183040
    },
    {
      "epoch": 74.93,
      "grad_norm": 1.6338863372802734,
      "learning_rate": 3.461692353581154e-06,
      "loss": 0.4944,
      "step": 183050
    },
    {
      "epoch": 74.93,
      "grad_norm": 1.886412262916565,
      "learning_rate": 3.461552930017874e-06,
      "loss": 0.5106,
      "step": 183060
    },
    {
      "epoch": 74.94,
      "grad_norm": 1.9373658895492554,
      "learning_rate": 3.461413501334111e-06,
      "loss": 0.5054,
      "step": 183070
    },
    {
      "epoch": 74.94,
      "grad_norm": 1.6676238775253296,
      "learning_rate": 3.4612740675305055e-06,
      "loss": 0.5054,
      "step": 183080
    },
    {
      "epoch": 74.94,
      "grad_norm": 1.689077377319336,
      "learning_rate": 3.4611346286076958e-06,
      "loss": 0.5161,
      "step": 183090
    },
    {
      "epoch": 74.95,
      "grad_norm": 1.6229476928710938,
      "learning_rate": 3.460995184566319e-06,
      "loss": 0.5158,
      "step": 183100
    },
    {
      "epoch": 74.95,
      "grad_norm": 1.8204216957092285,
      "learning_rate": 3.4608557354070157e-06,
      "loss": 0.5007,
      "step": 183110
    },
    {
      "epoch": 74.96,
      "grad_norm": 1.9619035720825195,
      "learning_rate": 3.460716281130424e-06,
      "loss": 0.5116,
      "step": 183120
    },
    {
      "epoch": 74.96,
      "grad_norm": 1.6055048704147339,
      "learning_rate": 3.460576821737182e-06,
      "loss": 0.5018,
      "step": 183130
    },
    {
      "epoch": 74.97,
      "grad_norm": 1.9814504384994507,
      "learning_rate": 3.46043735722793e-06,
      "loss": 0.5032,
      "step": 183140
    },
    {
      "epoch": 74.97,
      "grad_norm": 1.7688266038894653,
      "learning_rate": 3.4602978876033058e-06,
      "loss": 0.5252,
      "step": 183150
    },
    {
      "epoch": 74.97,
      "grad_norm": 2.1985533237457275,
      "learning_rate": 3.4601584128639493e-06,
      "loss": 0.5113,
      "step": 183160
    },
    {
      "epoch": 74.98,
      "grad_norm": 1.9619799852371216,
      "learning_rate": 3.4600189330104984e-06,
      "loss": 0.5161,
      "step": 183170
    },
    {
      "epoch": 74.98,
      "grad_norm": 1.858881950378418,
      "learning_rate": 3.4598794480435926e-06,
      "loss": 0.497,
      "step": 183180
    },
    {
      "epoch": 74.99,
      "grad_norm": 2.2222084999084473,
      "learning_rate": 3.4597399579638702e-06,
      "loss": 0.4954,
      "step": 183190
    },
    {
      "epoch": 74.99,
      "grad_norm": 1.4101197719573975,
      "learning_rate": 3.4596004627719707e-06,
      "loss": 0.5154,
      "step": 183200
    },
    {
      "epoch": 74.99,
      "grad_norm": 2.1281778812408447,
      "learning_rate": 3.459460962468533e-06,
      "loss": 0.5144,
      "step": 183210
    },
    {
      "epoch": 75.0,
      "grad_norm": 1.8464406728744507,
      "learning_rate": 3.4593214570541964e-06,
      "loss": 0.5133,
      "step": 183220
    },
    {
      "epoch": 75.0,
      "eval_loss": 0.5114964246749878,
      "eval_runtime": 76.6669,
      "eval_samples_per_second": 44.987,
      "eval_steps_per_second": 5.635,
      "step": 183225
    },
    {
      "epoch": 75.0,
      "grad_norm": 1.8936372995376587,
      "learning_rate": 3.4591819465296004e-06,
      "loss": 0.5092,
      "step": 183230
    },
    {
      "epoch": 75.01,
      "grad_norm": 2.333238124847412,
      "learning_rate": 3.4590424308953827e-06,
      "loss": 0.5317,
      "step": 183240
    },
    {
      "epoch": 75.01,
      "grad_norm": 1.4847654104232788,
      "learning_rate": 3.458902910152183e-06,
      "loss": 0.5197,
      "step": 183250
    },
    {
      "epoch": 75.01,
      "grad_norm": 1.4530055522918701,
      "learning_rate": 3.4587633843006403e-06,
      "loss": 0.4949,
      "step": 183260
    },
    {
      "epoch": 75.02,
      "grad_norm": 1.9522125720977783,
      "learning_rate": 3.4586238533413943e-06,
      "loss": 0.4934,
      "step": 183270
    },
    {
      "epoch": 75.02,
      "grad_norm": 2.162660598754883,
      "learning_rate": 3.4584843172750835e-06,
      "loss": 0.5132,
      "step": 183280
    },
    {
      "epoch": 75.03,
      "grad_norm": 2.3492679595947266,
      "learning_rate": 3.458344776102348e-06,
      "loss": 0.5239,
      "step": 183290
    },
    {
      "epoch": 75.03,
      "grad_norm": 1.7068637609481812,
      "learning_rate": 3.458205229823826e-06,
      "loss": 0.5226,
      "step": 183300
    },
    {
      "epoch": 75.03,
      "grad_norm": 2.100043773651123,
      "learning_rate": 3.4580656784401575e-06,
      "loss": 0.4868,
      "step": 183310
    },
    {
      "epoch": 75.04,
      "grad_norm": 2.1104609966278076,
      "learning_rate": 3.4579261219519813e-06,
      "loss": 0.5068,
      "step": 183320
    },
    {
      "epoch": 75.04,
      "grad_norm": 1.908278465270996,
      "learning_rate": 3.4577865603599366e-06,
      "loss": 0.5031,
      "step": 183330
    },
    {
      "epoch": 75.05,
      "grad_norm": 2.4526097774505615,
      "learning_rate": 3.457646993664663e-06,
      "loss": 0.5032,
      "step": 183340
    },
    {
      "epoch": 75.05,
      "grad_norm": 2.409363031387329,
      "learning_rate": 3.4575074218668002e-06,
      "loss": 0.5183,
      "step": 183350
    },
    {
      "epoch": 75.06,
      "grad_norm": 1.8910741806030273,
      "learning_rate": 3.457367844966987e-06,
      "loss": 0.5172,
      "step": 183360
    },
    {
      "epoch": 75.06,
      "grad_norm": 1.7140977382659912,
      "learning_rate": 3.457228262965863e-06,
      "loss": 0.518,
      "step": 183370
    },
    {
      "epoch": 75.06,
      "grad_norm": 2.459437608718872,
      "learning_rate": 3.4570886758640676e-06,
      "loss": 0.5093,
      "step": 183380
    },
    {
      "epoch": 75.07,
      "grad_norm": 2.118678092956543,
      "learning_rate": 3.4569490836622397e-06,
      "loss": 0.4838,
      "step": 183390
    },
    {
      "epoch": 75.07,
      "grad_norm": 1.488796353340149,
      "learning_rate": 3.4568094863610194e-06,
      "loss": 0.4937,
      "step": 183400
    },
    {
      "epoch": 75.08,
      "grad_norm": 2.380005359649658,
      "learning_rate": 3.4566698839610457e-06,
      "loss": 0.5211,
      "step": 183410
    },
    {
      "epoch": 75.08,
      "grad_norm": 1.9345006942749023,
      "learning_rate": 3.4565302764629586e-06,
      "loss": 0.51,
      "step": 183420
    },
    {
      "epoch": 75.08,
      "grad_norm": 1.7444876432418823,
      "learning_rate": 3.456390663867398e-06,
      "loss": 0.522,
      "step": 183430
    },
    {
      "epoch": 75.09,
      "grad_norm": 1.8555760383605957,
      "learning_rate": 3.456251046175002e-06,
      "loss": 0.5099,
      "step": 183440
    },
    {
      "epoch": 75.09,
      "grad_norm": 1.9165478944778442,
      "learning_rate": 3.4561114233864116e-06,
      "loss": 0.5313,
      "step": 183450
    },
    {
      "epoch": 75.1,
      "grad_norm": 2.1383564472198486,
      "learning_rate": 3.455971795502265e-06,
      "loss": 0.5248,
      "step": 183460
    },
    {
      "epoch": 75.1,
      "grad_norm": 1.756066083908081,
      "learning_rate": 3.4558321625232038e-06,
      "loss": 0.5216,
      "step": 183470
    },
    {
      "epoch": 75.1,
      "grad_norm": 2.0642871856689453,
      "learning_rate": 3.455692524449865e-06,
      "loss": 0.5161,
      "step": 183480
    },
    {
      "epoch": 75.11,
      "grad_norm": 1.520211935043335,
      "learning_rate": 3.4555528812828904e-06,
      "loss": 0.4979,
      "step": 183490
    },
    {
      "epoch": 75.11,
      "grad_norm": 2.081627130508423,
      "learning_rate": 3.455413233022919e-06,
      "loss": 0.5097,
      "step": 183500
    },
    {
      "epoch": 75.12,
      "grad_norm": 1.995398998260498,
      "learning_rate": 3.4552735796705902e-06,
      "loss": 0.5164,
      "step": 183510
    },
    {
      "epoch": 75.12,
      "grad_norm": 2.009106159210205,
      "learning_rate": 3.4551339212265448e-06,
      "loss": 0.508,
      "step": 183520
    },
    {
      "epoch": 75.12,
      "grad_norm": 1.4434086084365845,
      "learning_rate": 3.4549942576914213e-06,
      "loss": 0.5063,
      "step": 183530
    },
    {
      "epoch": 75.13,
      "grad_norm": 2.208294153213501,
      "learning_rate": 3.4548545890658594e-06,
      "loss": 0.5186,
      "step": 183540
    },
    {
      "epoch": 75.13,
      "grad_norm": 1.751649260520935,
      "learning_rate": 3.4547149153505003e-06,
      "loss": 0.5207,
      "step": 183550
    },
    {
      "epoch": 75.14,
      "grad_norm": 2.2199621200561523,
      "learning_rate": 3.454575236545983e-06,
      "loss": 0.505,
      "step": 183560
    },
    {
      "epoch": 75.14,
      "grad_norm": 1.502639889717102,
      "learning_rate": 3.454435552652946e-06,
      "loss": 0.494,
      "step": 183570
    },
    {
      "epoch": 75.15,
      "grad_norm": 1.7006467580795288,
      "learning_rate": 3.4542958636720326e-06,
      "loss": 0.5217,
      "step": 183580
    },
    {
      "epoch": 75.15,
      "grad_norm": 2.1218700408935547,
      "learning_rate": 3.4541561696038786e-06,
      "loss": 0.5203,
      "step": 183590
    },
    {
      "epoch": 75.15,
      "grad_norm": 1.7537184953689575,
      "learning_rate": 3.4540164704491275e-06,
      "loss": 0.5247,
      "step": 183600
    },
    {
      "epoch": 75.16,
      "grad_norm": 1.5841248035430908,
      "learning_rate": 3.4538767662084166e-06,
      "loss": 0.5204,
      "step": 183610
    },
    {
      "epoch": 75.16,
      "grad_norm": 2.4200501441955566,
      "learning_rate": 3.453737056882387e-06,
      "loss": 0.5033,
      "step": 183620
    },
    {
      "epoch": 75.17,
      "grad_norm": 2.7062346935272217,
      "learning_rate": 3.453597342471679e-06,
      "loss": 0.5018,
      "step": 183630
    },
    {
      "epoch": 75.17,
      "grad_norm": 1.6821702718734741,
      "learning_rate": 3.453457622976932e-06,
      "loss": 0.5101,
      "step": 183640
    },
    {
      "epoch": 75.17,
      "grad_norm": 1.8907525539398193,
      "learning_rate": 3.4533178983987865e-06,
      "loss": 0.5064,
      "step": 183650
    },
    {
      "epoch": 75.18,
      "grad_norm": 1.7642037868499756,
      "learning_rate": 3.453178168737882e-06,
      "loss": 0.5231,
      "step": 183660
    },
    {
      "epoch": 75.18,
      "grad_norm": 1.4540742635726929,
      "learning_rate": 3.4530384339948586e-06,
      "loss": 0.4979,
      "step": 183670
    },
    {
      "epoch": 75.19,
      "grad_norm": 1.6798452138900757,
      "learning_rate": 3.452898694170357e-06,
      "loss": 0.5221,
      "step": 183680
    },
    {
      "epoch": 75.19,
      "grad_norm": 1.3902274370193481,
      "learning_rate": 3.4527589492650174e-06,
      "loss": 0.5046,
      "step": 183690
    },
    {
      "epoch": 75.19,
      "grad_norm": 2.248809576034546,
      "learning_rate": 3.452619199279479e-06,
      "loss": 0.5063,
      "step": 183700
    },
    {
      "epoch": 75.2,
      "grad_norm": 1.7566829919815063,
      "learning_rate": 3.452479444214383e-06,
      "loss": 0.505,
      "step": 183710
    },
    {
      "epoch": 75.2,
      "grad_norm": 2.0302577018737793,
      "learning_rate": 3.4523396840703685e-06,
      "loss": 0.5142,
      "step": 183720
    },
    {
      "epoch": 75.21,
      "grad_norm": 2.1736936569213867,
      "learning_rate": 3.4521999188480766e-06,
      "loss": 0.4808,
      "step": 183730
    },
    {
      "epoch": 75.21,
      "grad_norm": 1.6326508522033691,
      "learning_rate": 3.4520601485481474e-06,
      "loss": 0.5079,
      "step": 183740
    },
    {
      "epoch": 75.21,
      "grad_norm": 2.0135157108306885,
      "learning_rate": 3.451920373171222e-06,
      "loss": 0.5065,
      "step": 183750
    },
    {
      "epoch": 75.22,
      "grad_norm": 1.8830976486206055,
      "learning_rate": 3.4517805927179375e-06,
      "loss": 0.5005,
      "step": 183760
    },
    {
      "epoch": 75.22,
      "grad_norm": 1.8706759214401245,
      "learning_rate": 3.451640807188938e-06,
      "loss": 0.5149,
      "step": 183770
    },
    {
      "epoch": 75.23,
      "grad_norm": 1.7740566730499268,
      "learning_rate": 3.451501016584862e-06,
      "loss": 0.5248,
      "step": 183780
    },
    {
      "epoch": 75.23,
      "grad_norm": 1.9776808023452759,
      "learning_rate": 3.45136122090635e-06,
      "loss": 0.5321,
      "step": 183790
    },
    {
      "epoch": 75.24,
      "grad_norm": 1.7604562044143677,
      "learning_rate": 3.451221420154043e-06,
      "loss": 0.5016,
      "step": 183800
    },
    {
      "epoch": 75.24,
      "grad_norm": 1.9635684490203857,
      "learning_rate": 3.451081614328581e-06,
      "loss": 0.519,
      "step": 183810
    },
    {
      "epoch": 75.24,
      "grad_norm": 2.272951364517212,
      "learning_rate": 3.450941803430604e-06,
      "loss": 0.4906,
      "step": 183820
    },
    {
      "epoch": 75.25,
      "grad_norm": 1.9056593179702759,
      "learning_rate": 3.4508019874607522e-06,
      "loss": 0.5096,
      "step": 183830
    },
    {
      "epoch": 75.25,
      "grad_norm": 2.496023416519165,
      "learning_rate": 3.4506621664196676e-06,
      "loss": 0.5021,
      "step": 183840
    },
    {
      "epoch": 75.26,
      "grad_norm": 2.0861446857452393,
      "learning_rate": 3.4505223403079896e-06,
      "loss": 0.5008,
      "step": 183850
    },
    {
      "epoch": 75.26,
      "grad_norm": 1.8108376264572144,
      "learning_rate": 3.4503825091263583e-06,
      "loss": 0.5181,
      "step": 183860
    },
    {
      "epoch": 75.26,
      "grad_norm": 1.683005690574646,
      "learning_rate": 3.450242672875415e-06,
      "loss": 0.5087,
      "step": 183870
    },
    {
      "epoch": 75.27,
      "grad_norm": 1.4602974653244019,
      "learning_rate": 3.4501028315558014e-06,
      "loss": 0.503,
      "step": 183880
    },
    {
      "epoch": 75.27,
      "grad_norm": 1.2992804050445557,
      "learning_rate": 3.449962985168156e-06,
      "loss": 0.5145,
      "step": 183890
    },
    {
      "epoch": 75.28,
      "grad_norm": 2.0317089557647705,
      "learning_rate": 3.44982313371312e-06,
      "loss": 0.5131,
      "step": 183900
    },
    {
      "epoch": 75.28,
      "grad_norm": 2.1103382110595703,
      "learning_rate": 3.4496832771913346e-06,
      "loss": 0.519,
      "step": 183910
    },
    {
      "epoch": 75.28,
      "grad_norm": 1.9923456907272339,
      "learning_rate": 3.44954341560344e-06,
      "loss": 0.5075,
      "step": 183920
    },
    {
      "epoch": 75.29,
      "grad_norm": 2.116286039352417,
      "learning_rate": 3.4494035489500767e-06,
      "loss": 0.5166,
      "step": 183930
    },
    {
      "epoch": 75.29,
      "grad_norm": 1.5383920669555664,
      "learning_rate": 3.449263677231886e-06,
      "loss": 0.501,
      "step": 183940
    },
    {
      "epoch": 75.3,
      "grad_norm": 1.6067880392074585,
      "learning_rate": 3.4491238004495086e-06,
      "loss": 0.5154,
      "step": 183950
    },
    {
      "epoch": 75.3,
      "grad_norm": 1.7213698625564575,
      "learning_rate": 3.4489839186035845e-06,
      "loss": 0.5163,
      "step": 183960
    },
    {
      "epoch": 75.3,
      "grad_norm": 1.636719822883606,
      "learning_rate": 3.4488440316947556e-06,
      "loss": 0.5179,
      "step": 183970
    },
    {
      "epoch": 75.31,
      "grad_norm": 1.6573091745376587,
      "learning_rate": 3.4487041397236613e-06,
      "loss": 0.5097,
      "step": 183980
    },
    {
      "epoch": 75.31,
      "grad_norm": 1.9108890295028687,
      "learning_rate": 3.448564242690944e-06,
      "loss": 0.5154,
      "step": 183990
    },
    {
      "epoch": 75.32,
      "grad_norm": 1.5473518371582031,
      "learning_rate": 3.4484243405972426e-06,
      "loss": 0.4979,
      "step": 184000
    },
    {
      "epoch": 75.32,
      "grad_norm": 1.8088786602020264,
      "learning_rate": 3.4482844334432005e-06,
      "loss": 0.502,
      "step": 184010
    },
    {
      "epoch": 75.33,
      "grad_norm": 1.9792829751968384,
      "learning_rate": 3.4481445212294567e-06,
      "loss": 0.5253,
      "step": 184020
    },
    {
      "epoch": 75.33,
      "grad_norm": 1.6606574058532715,
      "learning_rate": 3.4480046039566523e-06,
      "loss": 0.4933,
      "step": 184030
    },
    {
      "epoch": 75.33,
      "grad_norm": 1.582829475402832,
      "learning_rate": 3.4478646816254294e-06,
      "loss": 0.5189,
      "step": 184040
    },
    {
      "epoch": 75.34,
      "grad_norm": 2.007347822189331,
      "learning_rate": 3.447724754236427e-06,
      "loss": 0.4982,
      "step": 184050
    },
    {
      "epoch": 75.34,
      "grad_norm": 1.7428624629974365,
      "learning_rate": 3.447584821790288e-06,
      "loss": 0.5277,
      "step": 184060
    },
    {
      "epoch": 75.35,
      "grad_norm": 1.8118762969970703,
      "learning_rate": 3.4474448842876524e-06,
      "loss": 0.5126,
      "step": 184070
    },
    {
      "epoch": 75.35,
      "grad_norm": 2.111241102218628,
      "learning_rate": 3.447304941729161e-06,
      "loss": 0.5133,
      "step": 184080
    },
    {
      "epoch": 75.35,
      "grad_norm": 2.1878695487976074,
      "learning_rate": 3.4471649941154557e-06,
      "loss": 0.4974,
      "step": 184090
    },
    {
      "epoch": 75.36,
      "grad_norm": 1.6585724353790283,
      "learning_rate": 3.4470250414471782e-06,
      "loss": 0.5069,
      "step": 184100
    },
    {
      "epoch": 75.36,
      "grad_norm": 1.7777029275894165,
      "learning_rate": 3.4468850837249676e-06,
      "loss": 0.4965,
      "step": 184110
    },
    {
      "epoch": 75.37,
      "grad_norm": 1.9729253053665161,
      "learning_rate": 3.4467451209494665e-06,
      "loss": 0.5266,
      "step": 184120
    },
    {
      "epoch": 75.37,
      "grad_norm": 2.87937593460083,
      "learning_rate": 3.446605153121315e-06,
      "loss": 0.5063,
      "step": 184130
    },
    {
      "epoch": 75.37,
      "grad_norm": 2.4313549995422363,
      "learning_rate": 3.446465180241155e-06,
      "loss": 0.5003,
      "step": 184140
    },
    {
      "epoch": 75.38,
      "grad_norm": 1.5472182035446167,
      "learning_rate": 3.446325202309627e-06,
      "loss": 0.5086,
      "step": 184150
    },
    {
      "epoch": 75.38,
      "grad_norm": 2.1920723915100098,
      "learning_rate": 3.4461852193273736e-06,
      "loss": 0.5178,
      "step": 184160
    },
    {
      "epoch": 75.39,
      "grad_norm": 1.7386263608932495,
      "learning_rate": 3.446045231295035e-06,
      "loss": 0.5247,
      "step": 184170
    },
    {
      "epoch": 75.39,
      "grad_norm": 1.6341841220855713,
      "learning_rate": 3.445905238213253e-06,
      "loss": 0.521,
      "step": 184180
    },
    {
      "epoch": 75.4,
      "grad_norm": 1.441680908203125,
      "learning_rate": 3.4457652400826677e-06,
      "loss": 0.5162,
      "step": 184190
    },
    {
      "epoch": 75.4,
      "grad_norm": 1.7836211919784546,
      "learning_rate": 3.4456252369039215e-06,
      "loss": 0.4979,
      "step": 184200
    },
    {
      "epoch": 75.4,
      "grad_norm": 1.5156223773956299,
      "learning_rate": 3.445485228677656e-06,
      "loss": 0.5003,
      "step": 184210
    },
    {
      "epoch": 75.41,
      "grad_norm": 2.190117359161377,
      "learning_rate": 3.4453452154045117e-06,
      "loss": 0.5164,
      "step": 184220
    },
    {
      "epoch": 75.41,
      "grad_norm": 3.454005479812622,
      "learning_rate": 3.4452051970851304e-06,
      "loss": 0.4897,
      "step": 184230
    },
    {
      "epoch": 75.42,
      "grad_norm": 2.71610426902771,
      "learning_rate": 3.4450651737201534e-06,
      "loss": 0.4985,
      "step": 184240
    },
    {
      "epoch": 75.42,
      "grad_norm": 1.901984453201294,
      "learning_rate": 3.4449251453102223e-06,
      "loss": 0.4928,
      "step": 184250
    },
    {
      "epoch": 75.42,
      "grad_norm": 1.850829839706421,
      "learning_rate": 3.444785111855978e-06,
      "loss": 0.5093,
      "step": 184260
    },
    {
      "epoch": 75.43,
      "grad_norm": 2.5574190616607666,
      "learning_rate": 3.4446450733580624e-06,
      "loss": 0.5425,
      "step": 184270
    },
    {
      "epoch": 75.43,
      "grad_norm": 2.0953633785247803,
      "learning_rate": 3.444505029817117e-06,
      "loss": 0.5168,
      "step": 184280
    },
    {
      "epoch": 75.44,
      "grad_norm": 2.0167617797851562,
      "learning_rate": 3.4443649812337836e-06,
      "loss": 0.5149,
      "step": 184290
    },
    {
      "epoch": 75.44,
      "grad_norm": 1.6683558225631714,
      "learning_rate": 3.4442249276087036e-06,
      "loss": 0.4853,
      "step": 184300
    },
    {
      "epoch": 75.44,
      "grad_norm": 2.4419803619384766,
      "learning_rate": 3.4440848689425176e-06,
      "loss": 0.5268,
      "step": 184310
    },
    {
      "epoch": 75.45,
      "grad_norm": 2.5516276359558105,
      "learning_rate": 3.443944805235869e-06,
      "loss": 0.5234,
      "step": 184320
    },
    {
      "epoch": 75.45,
      "grad_norm": 2.2470507621765137,
      "learning_rate": 3.4438047364893976e-06,
      "loss": 0.4965,
      "step": 184330
    },
    {
      "epoch": 75.46,
      "grad_norm": 2.0120370388031006,
      "learning_rate": 3.443664662703746e-06,
      "loss": 0.5162,
      "step": 184340
    },
    {
      "epoch": 75.46,
      "grad_norm": 1.7806328535079956,
      "learning_rate": 3.4435245838795557e-06,
      "loss": 0.5131,
      "step": 184350
    },
    {
      "epoch": 75.46,
      "grad_norm": 1.78562331199646,
      "learning_rate": 3.443384500017468e-06,
      "loss": 0.5268,
      "step": 184360
    },
    {
      "epoch": 75.47,
      "grad_norm": 2.1249358654022217,
      "learning_rate": 3.4432444111181253e-06,
      "loss": 0.5063,
      "step": 184370
    },
    {
      "epoch": 75.47,
      "grad_norm": 1.5671721696853638,
      "learning_rate": 3.4431043171821693e-06,
      "loss": 0.522,
      "step": 184380
    },
    {
      "epoch": 75.48,
      "grad_norm": 1.829930305480957,
      "learning_rate": 3.4429642182102415e-06,
      "loss": 0.5053,
      "step": 184390
    },
    {
      "epoch": 75.48,
      "grad_norm": 1.9557307958602905,
      "learning_rate": 3.4428241142029833e-06,
      "loss": 0.4901,
      "step": 184400
    },
    {
      "epoch": 75.49,
      "grad_norm": 2.050677537918091,
      "learning_rate": 3.4426840051610366e-06,
      "loss": 0.4985,
      "step": 184410
    },
    {
      "epoch": 75.49,
      "grad_norm": 1.7481279373168945,
      "learning_rate": 3.442543891085043e-06,
      "loss": 0.5119,
      "step": 184420
    },
    {
      "epoch": 75.49,
      "grad_norm": 2.3942360877990723,
      "learning_rate": 3.4424037719756452e-06,
      "loss": 0.5159,
      "step": 184430
    },
    {
      "epoch": 75.5,
      "grad_norm": 1.6933594942092896,
      "learning_rate": 3.4422636478334847e-06,
      "loss": 0.4905,
      "step": 184440
    },
    {
      "epoch": 75.5,
      "grad_norm": 1.6187719106674194,
      "learning_rate": 3.442123518659203e-06,
      "loss": 0.5068,
      "step": 184450
    },
    {
      "epoch": 75.51,
      "grad_norm": 1.808019995689392,
      "learning_rate": 3.441983384453443e-06,
      "loss": 0.5198,
      "step": 184460
    },
    {
      "epoch": 75.51,
      "grad_norm": 2.067495107650757,
      "learning_rate": 3.4418432452168454e-06,
      "loss": 0.49,
      "step": 184470
    },
    {
      "epoch": 75.51,
      "grad_norm": 2.016303777694702,
      "learning_rate": 3.441703100950053e-06,
      "loss": 0.5218,
      "step": 184480
    },
    {
      "epoch": 75.52,
      "grad_norm": 1.70279860496521,
      "learning_rate": 3.4415629516537076e-06,
      "loss": 0.5227,
      "step": 184490
    },
    {
      "epoch": 75.52,
      "grad_norm": 2.2240347862243652,
      "learning_rate": 3.4414227973284508e-06,
      "loss": 0.5116,
      "step": 184500
    },
    {
      "epoch": 75.53,
      "grad_norm": 1.658474326133728,
      "learning_rate": 3.4412826379749248e-06,
      "loss": 0.504,
      "step": 184510
    },
    {
      "epoch": 75.53,
      "grad_norm": 1.7433433532714844,
      "learning_rate": 3.4411424735937716e-06,
      "loss": 0.5088,
      "step": 184520
    },
    {
      "epoch": 75.53,
      "grad_norm": 1.551856279373169,
      "learning_rate": 3.441002304185634e-06,
      "loss": 0.516,
      "step": 184530
    },
    {
      "epoch": 75.54,
      "grad_norm": 2.2550246715545654,
      "learning_rate": 3.440862129751153e-06,
      "loss": 0.5101,
      "step": 184540
    },
    {
      "epoch": 75.54,
      "grad_norm": 1.9550455808639526,
      "learning_rate": 3.440721950290972e-06,
      "loss": 0.5028,
      "step": 184550
    },
    {
      "epoch": 75.55,
      "grad_norm": 1.8037341833114624,
      "learning_rate": 3.4405817658057316e-06,
      "loss": 0.5118,
      "step": 184560
    },
    {
      "epoch": 75.55,
      "grad_norm": 2.3025288581848145,
      "learning_rate": 3.440441576296075e-06,
      "loss": 0.5032,
      "step": 184570
    },
    {
      "epoch": 75.55,
      "grad_norm": 1.9040470123291016,
      "learning_rate": 3.4403013817626445e-06,
      "loss": 0.5255,
      "step": 184580
    },
    {
      "epoch": 75.56,
      "grad_norm": 2.0050294399261475,
      "learning_rate": 3.440161182206082e-06,
      "loss": 0.5182,
      "step": 184590
    },
    {
      "epoch": 75.56,
      "grad_norm": 1.7959991693496704,
      "learning_rate": 3.440020977627029e-06,
      "loss": 0.5093,
      "step": 184600
    },
    {
      "epoch": 75.57,
      "grad_norm": 2.2661800384521484,
      "learning_rate": 3.4398807680261284e-06,
      "loss": 0.514,
      "step": 184610
    },
    {
      "epoch": 75.57,
      "grad_norm": 2.294710159301758,
      "learning_rate": 3.4397405534040228e-06,
      "loss": 0.4943,
      "step": 184620
    },
    {
      "epoch": 75.58,
      "grad_norm": 1.900931477546692,
      "learning_rate": 3.4396003337613538e-06,
      "loss": 0.5095,
      "step": 184630
    },
    {
      "epoch": 75.58,
      "grad_norm": 2.0174059867858887,
      "learning_rate": 3.439460109098765e-06,
      "loss": 0.5023,
      "step": 184640
    },
    {
      "epoch": 75.58,
      "grad_norm": 1.9415004253387451,
      "learning_rate": 3.439319879416898e-06,
      "loss": 0.5241,
      "step": 184650
    },
    {
      "epoch": 75.59,
      "grad_norm": 1.5793920755386353,
      "learning_rate": 3.4391796447163937e-06,
      "loss": 0.5132,
      "step": 184660
    },
    {
      "epoch": 75.59,
      "grad_norm": 2.044415235519409,
      "learning_rate": 3.439039404997897e-06,
      "loss": 0.5034,
      "step": 184670
    },
    {
      "epoch": 75.6,
      "grad_norm": 2.29146409034729,
      "learning_rate": 3.438899160262049e-06,
      "loss": 0.5191,
      "step": 184680
    },
    {
      "epoch": 75.6,
      "grad_norm": 1.3489683866500854,
      "learning_rate": 3.4387589105094927e-06,
      "loss": 0.5138,
      "step": 184690
    },
    {
      "epoch": 75.6,
      "grad_norm": 1.975862979888916,
      "learning_rate": 3.438618655740869e-06,
      "loss": 0.5009,
      "step": 184700
    },
    {
      "epoch": 75.61,
      "grad_norm": 1.4505661725997925,
      "learning_rate": 3.438478395956822e-06,
      "loss": 0.5191,
      "step": 184710
    },
    {
      "epoch": 75.61,
      "grad_norm": 2.3901126384735107,
      "learning_rate": 3.4383381311579934e-06,
      "loss": 0.5295,
      "step": 184720
    },
    {
      "epoch": 75.62,
      "grad_norm": 2.012977123260498,
      "learning_rate": 3.438197861345027e-06,
      "loss": 0.5121,
      "step": 184730
    },
    {
      "epoch": 75.62,
      "grad_norm": 1.707664132118225,
      "learning_rate": 3.4380575865185636e-06,
      "loss": 0.5151,
      "step": 184740
    },
    {
      "epoch": 75.62,
      "grad_norm": 1.8401036262512207,
      "learning_rate": 3.4379173066792476e-06,
      "loss": 0.5071,
      "step": 184750
    },
    {
      "epoch": 75.63,
      "grad_norm": 2.256437301635742,
      "learning_rate": 3.43777702182772e-06,
      "loss": 0.5244,
      "step": 184760
    },
    {
      "epoch": 75.63,
      "grad_norm": 1.8034679889678955,
      "learning_rate": 3.4376367319646247e-06,
      "loss": 0.4991,
      "step": 184770
    },
    {
      "epoch": 75.64,
      "grad_norm": 1.437151312828064,
      "learning_rate": 3.4374964370906027e-06,
      "loss": 0.5014,
      "step": 184780
    },
    {
      "epoch": 75.64,
      "grad_norm": 2.0017974376678467,
      "learning_rate": 3.437356137206298e-06,
      "loss": 0.4938,
      "step": 184790
    },
    {
      "epoch": 75.64,
      "grad_norm": 2.2036943435668945,
      "learning_rate": 3.4372158323123524e-06,
      "loss": 0.4983,
      "step": 184800
    },
    {
      "epoch": 75.65,
      "grad_norm": 1.7339673042297363,
      "learning_rate": 3.43707552240941e-06,
      "loss": 0.5126,
      "step": 184810
    },
    {
      "epoch": 75.65,
      "grad_norm": 1.5497915744781494,
      "learning_rate": 3.436935207498113e-06,
      "loss": 0.4929,
      "step": 184820
    },
    {
      "epoch": 75.66,
      "grad_norm": 1.569024682044983,
      "learning_rate": 3.436794887579103e-06,
      "loss": 0.5011,
      "step": 184830
    },
    {
      "epoch": 75.66,
      "grad_norm": 1.7660609483718872,
      "learning_rate": 3.436654562653024e-06,
      "loss": 0.5291,
      "step": 184840
    },
    {
      "epoch": 75.67,
      "grad_norm": 1.9612329006195068,
      "learning_rate": 3.4365142327205184e-06,
      "loss": 0.5024,
      "step": 184850
    },
    {
      "epoch": 75.67,
      "grad_norm": 1.8379865884780884,
      "learning_rate": 3.4363738977822293e-06,
      "loss": 0.4866,
      "step": 184860
    },
    {
      "epoch": 75.67,
      "grad_norm": 1.7327752113342285,
      "learning_rate": 3.4362335578387992e-06,
      "loss": 0.5137,
      "step": 184870
    },
    {
      "epoch": 75.68,
      "grad_norm": 1.5229038000106812,
      "learning_rate": 3.436093212890871e-06,
      "loss": 0.5087,
      "step": 184880
    },
    {
      "epoch": 75.68,
      "grad_norm": 1.8040255308151245,
      "learning_rate": 3.4359528629390878e-06,
      "loss": 0.5216,
      "step": 184890
    },
    {
      "epoch": 75.69,
      "grad_norm": 1.9804974794387817,
      "learning_rate": 3.4358125079840926e-06,
      "loss": 0.5,
      "step": 184900
    },
    {
      "epoch": 75.69,
      "grad_norm": 2.0324294567108154,
      "learning_rate": 3.435672148026528e-06,
      "loss": 0.5169,
      "step": 184910
    },
    {
      "epoch": 75.69,
      "grad_norm": 1.6012415885925293,
      "learning_rate": 3.435531783067037e-06,
      "loss": 0.5183,
      "step": 184920
    },
    {
      "epoch": 75.7,
      "grad_norm": 2.196781873703003,
      "learning_rate": 3.435391413106263e-06,
      "loss": 0.5215,
      "step": 184930
    },
    {
      "epoch": 75.7,
      "grad_norm": 2.15460467338562,
      "learning_rate": 3.4352510381448486e-06,
      "loss": 0.5053,
      "step": 184940
    },
    {
      "epoch": 75.71,
      "grad_norm": 1.5890161991119385,
      "learning_rate": 3.435110658183437e-06,
      "loss": 0.5145,
      "step": 184950
    },
    {
      "epoch": 75.71,
      "grad_norm": 1.9315600395202637,
      "learning_rate": 3.4349702732226716e-06,
      "loss": 0.4958,
      "step": 184960
    },
    {
      "epoch": 75.71,
      "grad_norm": 1.7334187030792236,
      "learning_rate": 3.434829883263195e-06,
      "loss": 0.4878,
      "step": 184970
    },
    {
      "epoch": 75.72,
      "grad_norm": 1.439138650894165,
      "learning_rate": 3.43468948830565e-06,
      "loss": 0.4941,
      "step": 184980
    },
    {
      "epoch": 75.72,
      "grad_norm": 1.684893012046814,
      "learning_rate": 3.43454908835068e-06,
      "loss": 0.5008,
      "step": 184990
    },
    {
      "epoch": 75.73,
      "grad_norm": 2.3283157348632812,
      "learning_rate": 3.4344086833989287e-06,
      "loss": 0.506,
      "step": 185000
    },
    {
      "epoch": 75.73,
      "grad_norm": 1.4292877912521362,
      "learning_rate": 3.434268273451039e-06,
      "loss": 0.5053,
      "step": 185010
    },
    {
      "epoch": 75.73,
      "grad_norm": 1.6337486505508423,
      "learning_rate": 3.434127858507654e-06,
      "loss": 0.4998,
      "step": 185020
    },
    {
      "epoch": 75.74,
      "grad_norm": 2.546355962753296,
      "learning_rate": 3.4339874385694167e-06,
      "loss": 0.4929,
      "step": 185030
    },
    {
      "epoch": 75.74,
      "grad_norm": 1.887971043586731,
      "learning_rate": 3.433847013636971e-06,
      "loss": 0.521,
      "step": 185040
    },
    {
      "epoch": 75.75,
      "grad_norm": 2.5685293674468994,
      "learning_rate": 3.4337065837109593e-06,
      "loss": 0.5251,
      "step": 185050
    },
    {
      "epoch": 75.75,
      "grad_norm": 1.4703242778778076,
      "learning_rate": 3.433566148792025e-06,
      "loss": 0.5204,
      "step": 185060
    },
    {
      "epoch": 75.76,
      "grad_norm": 1.805734395980835,
      "learning_rate": 3.4334257088808123e-06,
      "loss": 0.521,
      "step": 185070
    },
    {
      "epoch": 75.76,
      "grad_norm": 1.9302020072937012,
      "learning_rate": 3.433285263977963e-06,
      "loss": 0.5244,
      "step": 185080
    },
    {
      "epoch": 75.76,
      "grad_norm": 1.6861108541488647,
      "learning_rate": 3.4331448140841217e-06,
      "loss": 0.5094,
      "step": 185090
    },
    {
      "epoch": 75.77,
      "grad_norm": 1.566704273223877,
      "learning_rate": 3.4330043591999318e-06,
      "loss": 0.4849,
      "step": 185100
    },
    {
      "epoch": 75.77,
      "grad_norm": 1.7259972095489502,
      "learning_rate": 3.432863899326037e-06,
      "loss": 0.5192,
      "step": 185110
    },
    {
      "epoch": 75.78,
      "grad_norm": 1.7638764381408691,
      "learning_rate": 3.432723434463079e-06,
      "loss": 0.5114,
      "step": 185120
    },
    {
      "epoch": 75.78,
      "grad_norm": 1.9398349523544312,
      "learning_rate": 3.432582964611703e-06,
      "loss": 0.505,
      "step": 185130
    },
    {
      "epoch": 75.78,
      "grad_norm": 1.7967160940170288,
      "learning_rate": 3.4324424897725515e-06,
      "loss": 0.5088,
      "step": 185140
    },
    {
      "epoch": 75.79,
      "grad_norm": 2.8808999061584473,
      "learning_rate": 3.4323020099462685e-06,
      "loss": 0.5356,
      "step": 185150
    },
    {
      "epoch": 75.79,
      "grad_norm": 2.2090859413146973,
      "learning_rate": 3.4321615251334975e-06,
      "loss": 0.5132,
      "step": 185160
    },
    {
      "epoch": 75.8,
      "grad_norm": 1.6324701309204102,
      "learning_rate": 3.4320210353348818e-06,
      "loss": 0.5296,
      "step": 185170
    },
    {
      "epoch": 75.8,
      "grad_norm": 1.5407265424728394,
      "learning_rate": 3.4318805405510642e-06,
      "loss": 0.5165,
      "step": 185180
    },
    {
      "epoch": 75.8,
      "grad_norm": 1.3278344869613647,
      "learning_rate": 3.43174004078269e-06,
      "loss": 0.5099,
      "step": 185190
    },
    {
      "epoch": 75.81,
      "grad_norm": 2.1623501777648926,
      "learning_rate": 3.4315995360304016e-06,
      "loss": 0.5243,
      "step": 185200
    },
    {
      "epoch": 75.81,
      "grad_norm": 1.8047508001327515,
      "learning_rate": 3.431459026294843e-06,
      "loss": 0.5038,
      "step": 185210
    },
    {
      "epoch": 75.82,
      "grad_norm": 2.0894479751586914,
      "learning_rate": 3.431318511576658e-06,
      "loss": 0.5044,
      "step": 185220
    },
    {
      "epoch": 75.82,
      "grad_norm": 1.823479175567627,
      "learning_rate": 3.43117799187649e-06,
      "loss": 0.5177,
      "step": 185230
    },
    {
      "epoch": 75.82,
      "grad_norm": 2.087805986404419,
      "learning_rate": 3.4310374671949827e-06,
      "loss": 0.52,
      "step": 185240
    },
    {
      "epoch": 75.83,
      "grad_norm": 1.6111645698547363,
      "learning_rate": 3.4308969375327798e-06,
      "loss": 0.522,
      "step": 185250
    },
    {
      "epoch": 75.83,
      "grad_norm": 1.6182700395584106,
      "learning_rate": 3.4307564028905254e-06,
      "loss": 0.4991,
      "step": 185260
    },
    {
      "epoch": 75.84,
      "grad_norm": 1.6213287115097046,
      "learning_rate": 3.4306158632688625e-06,
      "loss": 0.5069,
      "step": 185270
    },
    {
      "epoch": 75.84,
      "grad_norm": 2.110072135925293,
      "learning_rate": 3.430475318668436e-06,
      "loss": 0.5029,
      "step": 185280
    },
    {
      "epoch": 75.85,
      "grad_norm": 2.0557360649108887,
      "learning_rate": 3.4303347690898886e-06,
      "loss": 0.4998,
      "step": 185290
    },
    {
      "epoch": 75.85,
      "grad_norm": 2.3799116611480713,
      "learning_rate": 3.4301942145338647e-06,
      "loss": 0.5143,
      "step": 185300
    },
    {
      "epoch": 75.85,
      "grad_norm": 1.8639415502548218,
      "learning_rate": 3.4300536550010085e-06,
      "loss": 0.5135,
      "step": 185310
    },
    {
      "epoch": 75.86,
      "grad_norm": 2.8924942016601562,
      "learning_rate": 3.4299130904919633e-06,
      "loss": 0.5205,
      "step": 185320
    },
    {
      "epoch": 75.86,
      "grad_norm": 2.2632741928100586,
      "learning_rate": 3.4297725210073738e-06,
      "loss": 0.5027,
      "step": 185330
    },
    {
      "epoch": 75.87,
      "grad_norm": 2.9109253883361816,
      "learning_rate": 3.4296319465478827e-06,
      "loss": 0.5091,
      "step": 185340
    },
    {
      "epoch": 75.87,
      "grad_norm": 1.7053828239440918,
      "learning_rate": 3.4294913671141344e-06,
      "loss": 0.5204,
      "step": 185350
    },
    {
      "epoch": 75.87,
      "grad_norm": 1.626810073852539,
      "learning_rate": 3.4293507827067733e-06,
      "loss": 0.5051,
      "step": 185360
    },
    {
      "epoch": 75.88,
      "grad_norm": 1.741767168045044,
      "learning_rate": 3.429210193326443e-06,
      "loss": 0.5054,
      "step": 185370
    },
    {
      "epoch": 75.88,
      "grad_norm": 1.9935979843139648,
      "learning_rate": 3.429069598973787e-06,
      "loss": 0.5018,
      "step": 185380
    },
    {
      "epoch": 75.89,
      "grad_norm": 2.277594804763794,
      "learning_rate": 3.4289289996494512e-06,
      "loss": 0.5126,
      "step": 185390
    },
    {
      "epoch": 75.89,
      "grad_norm": 1.9404823780059814,
      "learning_rate": 3.4287883953540784e-06,
      "loss": 0.516,
      "step": 185400
    },
    {
      "epoch": 75.89,
      "grad_norm": 2.010507345199585,
      "learning_rate": 3.428647786088312e-06,
      "loss": 0.4851,
      "step": 185410
    },
    {
      "epoch": 75.9,
      "grad_norm": 1.9949945211410522,
      "learning_rate": 3.4285071718527976e-06,
      "loss": 0.5087,
      "step": 185420
    },
    {
      "epoch": 75.9,
      "grad_norm": 2.121263027191162,
      "learning_rate": 3.4283665526481783e-06,
      "loss": 0.5073,
      "step": 185430
    },
    {
      "epoch": 75.91,
      "grad_norm": 1.5507142543792725,
      "learning_rate": 3.4282259284750983e-06,
      "loss": 0.5393,
      "step": 185440
    },
    {
      "epoch": 75.91,
      "grad_norm": 1.6410938501358032,
      "learning_rate": 3.4280852993342023e-06,
      "loss": 0.5113,
      "step": 185450
    },
    {
      "epoch": 75.91,
      "grad_norm": 2.6072702407836914,
      "learning_rate": 3.4279446652261336e-06,
      "loss": 0.4955,
      "step": 185460
    },
    {
      "epoch": 75.92,
      "grad_norm": 2.1335365772247314,
      "learning_rate": 3.427804026151538e-06,
      "loss": 0.4936,
      "step": 185470
    },
    {
      "epoch": 75.92,
      "grad_norm": 1.6921906471252441,
      "learning_rate": 3.427663382111058e-06,
      "loss": 0.5178,
      "step": 185480
    },
    {
      "epoch": 75.93,
      "grad_norm": 1.6630215644836426,
      "learning_rate": 3.4275227331053395e-06,
      "loss": 0.5137,
      "step": 185490
    },
    {
      "epoch": 75.93,
      "grad_norm": 1.7824885845184326,
      "learning_rate": 3.4273820791350254e-06,
      "loss": 0.4989,
      "step": 185500
    },
    {
      "epoch": 75.94,
      "grad_norm": 1.5821634531021118,
      "learning_rate": 3.4272414202007607e-06,
      "loss": 0.5034,
      "step": 185510
    },
    {
      "epoch": 75.94,
      "grad_norm": 1.9504565000534058,
      "learning_rate": 3.427100756303189e-06,
      "loss": 0.4894,
      "step": 185520
    },
    {
      "epoch": 75.94,
      "grad_norm": 2.125878095626831,
      "learning_rate": 3.426960087442956e-06,
      "loss": 0.4946,
      "step": 185530
    },
    {
      "epoch": 75.95,
      "grad_norm": 1.9387215375900269,
      "learning_rate": 3.4268194136207053e-06,
      "loss": 0.4951,
      "step": 185540
    },
    {
      "epoch": 75.95,
      "grad_norm": 2.2444002628326416,
      "learning_rate": 3.4266787348370815e-06,
      "loss": 0.4881,
      "step": 185550
    },
    {
      "epoch": 75.96,
      "grad_norm": 1.7448810338974,
      "learning_rate": 3.426538051092728e-06,
      "loss": 0.4975,
      "step": 185560
    },
    {
      "epoch": 75.96,
      "grad_norm": 2.0842654705047607,
      "learning_rate": 3.4263973623882912e-06,
      "loss": 0.5044,
      "step": 185570
    },
    {
      "epoch": 75.96,
      "grad_norm": 2.0765738487243652,
      "learning_rate": 3.4262566687244136e-06,
      "loss": 0.512,
      "step": 185580
    },
    {
      "epoch": 75.97,
      "grad_norm": 3.0320041179656982,
      "learning_rate": 3.426115970101741e-06,
      "loss": 0.5151,
      "step": 185590
    },
    {
      "epoch": 75.97,
      "grad_norm": 1.7355084419250488,
      "learning_rate": 3.4259752665209174e-06,
      "loss": 0.485,
      "step": 185600
    },
    {
      "epoch": 75.98,
      "grad_norm": 1.576298475265503,
      "learning_rate": 3.4258345579825876e-06,
      "loss": 0.5077,
      "step": 185610
    },
    {
      "epoch": 75.98,
      "grad_norm": 2.3599395751953125,
      "learning_rate": 3.4256938444873965e-06,
      "loss": 0.4954,
      "step": 185620
    },
    {
      "epoch": 75.98,
      "grad_norm": 1.5922318696975708,
      "learning_rate": 3.4255531260359873e-06,
      "loss": 0.4959,
      "step": 185630
    },
    {
      "epoch": 75.99,
      "grad_norm": 1.9739669561386108,
      "learning_rate": 3.425412402629006e-06,
      "loss": 0.5234,
      "step": 185640
    },
    {
      "epoch": 75.99,
      "grad_norm": 1.7315573692321777,
      "learning_rate": 3.4252716742670966e-06,
      "loss": 0.4952,
      "step": 185650
    },
    {
      "epoch": 76.0,
      "grad_norm": 2.789520740509033,
      "learning_rate": 3.4251309409509042e-06,
      "loss": 0.4886,
      "step": 185660
    },
    {
      "epoch": 76.0,
      "eval_loss": 0.5089752078056335,
      "eval_runtime": 75.5869,
      "eval_samples_per_second": 45.63,
      "eval_steps_per_second": 5.715,
      "step": 185668
    },
    {
      "epoch": 76.0,
      "grad_norm": 2.729156255722046,
      "learning_rate": 3.424990202681073e-06,
      "loss": 0.5153,
      "step": 185670
    },
    {
      "epoch": 76.0,
      "grad_norm": 2.148165702819824,
      "learning_rate": 3.4248494594582475e-06,
      "loss": 0.5065,
      "step": 185680
    },
    {
      "epoch": 76.01,
      "grad_norm": 2.0563156604766846,
      "learning_rate": 3.4247087112830737e-06,
      "loss": 0.493,
      "step": 185690
    },
    {
      "epoch": 76.01,
      "grad_norm": 2.4571266174316406,
      "learning_rate": 3.424567958156195e-06,
      "loss": 0.5223,
      "step": 185700
    },
    {
      "epoch": 76.02,
      "grad_norm": 2.0062992572784424,
      "learning_rate": 3.4244272000782563e-06,
      "loss": 0.503,
      "step": 185710
    },
    {
      "epoch": 76.02,
      "grad_norm": 1.8927440643310547,
      "learning_rate": 3.4242864370499027e-06,
      "loss": 0.4985,
      "step": 185720
    },
    {
      "epoch": 76.03,
      "grad_norm": 1.784038782119751,
      "learning_rate": 3.4241456690717796e-06,
      "loss": 0.5066,
      "step": 185730
    },
    {
      "epoch": 76.03,
      "grad_norm": 1.9520834684371948,
      "learning_rate": 3.4240048961445306e-06,
      "loss": 0.527,
      "step": 185740
    },
    {
      "epoch": 76.03,
      "grad_norm": 1.8874828815460205,
      "learning_rate": 3.4238641182688013e-06,
      "loss": 0.523,
      "step": 185750
    },
    {
      "epoch": 76.04,
      "grad_norm": 1.7039682865142822,
      "learning_rate": 3.4237233354452373e-06,
      "loss": 0.5153,
      "step": 185760
    },
    {
      "epoch": 76.04,
      "grad_norm": 1.6699448823928833,
      "learning_rate": 3.423582547674482e-06,
      "loss": 0.4997,
      "step": 185770
    },
    {
      "epoch": 76.05,
      "grad_norm": 1.7332062721252441,
      "learning_rate": 3.423441754957181e-06,
      "loss": 0.507,
      "step": 185780
    },
    {
      "epoch": 76.05,
      "grad_norm": 1.6710129976272583,
      "learning_rate": 3.4233009572939796e-06,
      "loss": 0.494,
      "step": 185790
    },
    {
      "epoch": 76.05,
      "grad_norm": 1.5972532033920288,
      "learning_rate": 3.4231601546855226e-06,
      "loss": 0.5133,
      "step": 185800
    },
    {
      "epoch": 76.06,
      "grad_norm": 1.9126185178756714,
      "learning_rate": 3.423019347132454e-06,
      "loss": 0.5147,
      "step": 185810
    },
    {
      "epoch": 76.06,
      "grad_norm": 1.5796151161193848,
      "learning_rate": 3.42287853463542e-06,
      "loss": 0.5151,
      "step": 185820
    },
    {
      "epoch": 76.07,
      "grad_norm": 1.6681299209594727,
      "learning_rate": 3.422737717195066e-06,
      "loss": 0.5138,
      "step": 185830
    },
    {
      "epoch": 76.07,
      "grad_norm": 1.9803316593170166,
      "learning_rate": 3.422596894812036e-06,
      "loss": 0.4847,
      "step": 185840
    },
    {
      "epoch": 76.07,
      "grad_norm": 2.53311824798584,
      "learning_rate": 3.422456067486975e-06,
      "loss": 0.517,
      "step": 185850
    },
    {
      "epoch": 76.08,
      "grad_norm": 2.0375843048095703,
      "learning_rate": 3.422315235220529e-06,
      "loss": 0.5124,
      "step": 185860
    },
    {
      "epoch": 76.08,
      "grad_norm": 1.8092631101608276,
      "learning_rate": 3.4221743980133426e-06,
      "loss": 0.4939,
      "step": 185870
    },
    {
      "epoch": 76.09,
      "grad_norm": 2.0015602111816406,
      "learning_rate": 3.4220335558660614e-06,
      "loss": 0.4946,
      "step": 185880
    },
    {
      "epoch": 76.09,
      "grad_norm": 1.6739953756332397,
      "learning_rate": 3.4218927087793296e-06,
      "loss": 0.5174,
      "step": 185890
    },
    {
      "epoch": 76.09,
      "grad_norm": 2.116074800491333,
      "learning_rate": 3.421751856753793e-06,
      "loss": 0.5052,
      "step": 185900
    },
    {
      "epoch": 76.1,
      "grad_norm": 1.6694378852844238,
      "learning_rate": 3.4216109997900977e-06,
      "loss": 0.5054,
      "step": 185910
    },
    {
      "epoch": 76.1,
      "grad_norm": 1.8318649530410767,
      "learning_rate": 3.4214701378888876e-06,
      "loss": 0.4979,
      "step": 185920
    },
    {
      "epoch": 76.11,
      "grad_norm": 2.195556640625,
      "learning_rate": 3.421329271050808e-06,
      "loss": 0.5268,
      "step": 185930
    },
    {
      "epoch": 76.11,
      "grad_norm": 1.5963563919067383,
      "learning_rate": 3.4211883992765053e-06,
      "loss": 0.4798,
      "step": 185940
    },
    {
      "epoch": 76.12,
      "grad_norm": 1.716896653175354,
      "learning_rate": 3.4210475225666237e-06,
      "loss": 0.5261,
      "step": 185950
    },
    {
      "epoch": 76.12,
      "grad_norm": 1.8804720640182495,
      "learning_rate": 3.420906640921809e-06,
      "loss": 0.4925,
      "step": 185960
    },
    {
      "epoch": 76.12,
      "grad_norm": 2.140610694885254,
      "learning_rate": 3.420765754342707e-06,
      "loss": 0.5279,
      "step": 185970
    },
    {
      "epoch": 76.13,
      "grad_norm": 1.9212702512741089,
      "learning_rate": 3.420624862829963e-06,
      "loss": 0.4962,
      "step": 185980
    },
    {
      "epoch": 76.13,
      "grad_norm": 1.805693507194519,
      "learning_rate": 3.420483966384221e-06,
      "loss": 0.5068,
      "step": 185990
    },
    {
      "epoch": 76.14,
      "grad_norm": 1.8112190961837769,
      "learning_rate": 3.420343065006128e-06,
      "loss": 0.5179,
      "step": 186000
    },
    {
      "epoch": 76.14,
      "grad_norm": 1.718278169631958,
      "learning_rate": 3.4202021586963287e-06,
      "loss": 0.5047,
      "step": 186010
    },
    {
      "epoch": 76.14,
      "grad_norm": 1.7288070917129517,
      "learning_rate": 3.4200612474554685e-06,
      "loss": 0.4958,
      "step": 186020
    },
    {
      "epoch": 76.15,
      "grad_norm": 2.3882534503936768,
      "learning_rate": 3.419920331284193e-06,
      "loss": 0.5045,
      "step": 186030
    },
    {
      "epoch": 76.15,
      "grad_norm": 2.053746461868286,
      "learning_rate": 3.4197794101831483e-06,
      "loss": 0.518,
      "step": 186040
    },
    {
      "epoch": 76.16,
      "grad_norm": 1.867882490158081,
      "learning_rate": 3.41963848415298e-06,
      "loss": 0.4863,
      "step": 186050
    },
    {
      "epoch": 76.16,
      "grad_norm": 1.8837559223175049,
      "learning_rate": 3.4194975531943335e-06,
      "loss": 0.515,
      "step": 186060
    },
    {
      "epoch": 76.16,
      "grad_norm": 1.462327480316162,
      "learning_rate": 3.4193566173078527e-06,
      "loss": 0.5084,
      "step": 186070
    },
    {
      "epoch": 76.17,
      "grad_norm": 1.5441759824752808,
      "learning_rate": 3.4192156764941856e-06,
      "loss": 0.5035,
      "step": 186080
    },
    {
      "epoch": 76.17,
      "grad_norm": 1.5793848037719727,
      "learning_rate": 3.4190747307539762e-06,
      "loss": 0.5262,
      "step": 186090
    },
    {
      "epoch": 76.18,
      "grad_norm": 1.554738998413086,
      "learning_rate": 3.418933780087871e-06,
      "loss": 0.5235,
      "step": 186100
    },
    {
      "epoch": 76.18,
      "grad_norm": 2.344219923019409,
      "learning_rate": 3.418792824496515e-06,
      "loss": 0.51,
      "step": 186110
    },
    {
      "epoch": 76.19,
      "grad_norm": 1.98622465133667,
      "learning_rate": 3.418651863980555e-06,
      "loss": 0.5083,
      "step": 186120
    },
    {
      "epoch": 76.19,
      "grad_norm": 1.6165235042572021,
      "learning_rate": 3.4185108985406348e-06,
      "loss": 0.5066,
      "step": 186130
    },
    {
      "epoch": 76.19,
      "grad_norm": 2.7865753173828125,
      "learning_rate": 3.418369928177403e-06,
      "loss": 0.5247,
      "step": 186140
    },
    {
      "epoch": 76.2,
      "grad_norm": 1.8020126819610596,
      "learning_rate": 3.4182289528915027e-06,
      "loss": 0.4863,
      "step": 186150
    },
    {
      "epoch": 76.2,
      "grad_norm": 1.7849329710006714,
      "learning_rate": 3.418087972683581e-06,
      "loss": 0.4871,
      "step": 186160
    },
    {
      "epoch": 76.21,
      "grad_norm": 1.3922734260559082,
      "learning_rate": 3.4179469875542833e-06,
      "loss": 0.489,
      "step": 186170
    },
    {
      "epoch": 76.21,
      "grad_norm": 1.6756738424301147,
      "learning_rate": 3.4178059975042554e-06,
      "loss": 0.4884,
      "step": 186180
    },
    {
      "epoch": 76.21,
      "grad_norm": 1.9604741334915161,
      "learning_rate": 3.417665002534144e-06,
      "loss": 0.509,
      "step": 186190
    },
    {
      "epoch": 76.22,
      "grad_norm": 1.6496223211288452,
      "learning_rate": 3.4175240026445937e-06,
      "loss": 0.5066,
      "step": 186200
    },
    {
      "epoch": 76.22,
      "grad_norm": 2.362470865249634,
      "learning_rate": 3.417382997836252e-06,
      "loss": 0.5069,
      "step": 186210
    },
    {
      "epoch": 76.23,
      "grad_norm": 1.7445557117462158,
      "learning_rate": 3.4172419881097623e-06,
      "loss": 0.5292,
      "step": 186220
    },
    {
      "epoch": 76.23,
      "grad_norm": 1.678353190422058,
      "learning_rate": 3.4171009734657723e-06,
      "loss": 0.4807,
      "step": 186230
    },
    {
      "epoch": 76.23,
      "grad_norm": 1.5337040424346924,
      "learning_rate": 3.416959953904929e-06,
      "loss": 0.5349,
      "step": 186240
    },
    {
      "epoch": 76.24,
      "grad_norm": 1.6768934726715088,
      "learning_rate": 3.4168189294278757e-06,
      "loss": 0.5057,
      "step": 186250
    },
    {
      "epoch": 76.24,
      "grad_norm": 1.9376708269119263,
      "learning_rate": 3.41667790003526e-06,
      "loss": 0.501,
      "step": 186260
    },
    {
      "epoch": 76.25,
      "grad_norm": 1.5466808080673218,
      "learning_rate": 3.416536865727729e-06,
      "loss": 0.5021,
      "step": 186270
    },
    {
      "epoch": 76.25,
      "grad_norm": 2.2956435680389404,
      "learning_rate": 3.4163958265059268e-06,
      "loss": 0.5322,
      "step": 186280
    },
    {
      "epoch": 76.25,
      "grad_norm": 1.634133219718933,
      "learning_rate": 3.4162547823704997e-06,
      "loss": 0.5174,
      "step": 186290
    },
    {
      "epoch": 76.26,
      "grad_norm": 2.1295838356018066,
      "learning_rate": 3.416113733322095e-06,
      "loss": 0.5138,
      "step": 186300
    },
    {
      "epoch": 76.26,
      "grad_norm": 1.8098680973052979,
      "learning_rate": 3.415972679361357e-06,
      "loss": 0.4846,
      "step": 186310
    },
    {
      "epoch": 76.27,
      "grad_norm": 1.5860048532485962,
      "learning_rate": 3.415831620488934e-06,
      "loss": 0.5098,
      "step": 186320
    },
    {
      "epoch": 76.27,
      "grad_norm": 1.7330892086029053,
      "learning_rate": 3.4156905567054707e-06,
      "loss": 0.4897,
      "step": 186330
    },
    {
      "epoch": 76.28,
      "grad_norm": 1.3952237367630005,
      "learning_rate": 3.415549488011614e-06,
      "loss": 0.5174,
      "step": 186340
    },
    {
      "epoch": 76.28,
      "grad_norm": 3.1620044708251953,
      "learning_rate": 3.41540841440801e-06,
      "loss": 0.5341,
      "step": 186350
    },
    {
      "epoch": 76.28,
      "grad_norm": 2.3031888008117676,
      "learning_rate": 3.415267335895304e-06,
      "loss": 0.5198,
      "step": 186360
    },
    {
      "epoch": 76.29,
      "grad_norm": 1.772708535194397,
      "learning_rate": 3.4151262524741437e-06,
      "loss": 0.5099,
      "step": 186370
    },
    {
      "epoch": 76.29,
      "grad_norm": 1.8464844226837158,
      "learning_rate": 3.4149851641451746e-06,
      "loss": 0.5131,
      "step": 186380
    },
    {
      "epoch": 76.3,
      "grad_norm": 1.8199799060821533,
      "learning_rate": 3.414844070909043e-06,
      "loss": 0.5298,
      "step": 186390
    },
    {
      "epoch": 76.3,
      "grad_norm": 1.5950448513031006,
      "learning_rate": 3.4147029727663944e-06,
      "loss": 0.5226,
      "step": 186400
    },
    {
      "epoch": 76.3,
      "grad_norm": 1.7259937524795532,
      "learning_rate": 3.4145618697178777e-06,
      "loss": 0.529,
      "step": 186410
    },
    {
      "epoch": 76.31,
      "grad_norm": 2.2508182525634766,
      "learning_rate": 3.414420761764137e-06,
      "loss": 0.4979,
      "step": 186420
    },
    {
      "epoch": 76.31,
      "grad_norm": 1.8221288919448853,
      "learning_rate": 3.4142796489058195e-06,
      "loss": 0.5026,
      "step": 186430
    },
    {
      "epoch": 76.32,
      "grad_norm": 1.6034060716629028,
      "learning_rate": 3.414138531143571e-06,
      "loss": 0.5079,
      "step": 186440
    },
    {
      "epoch": 76.32,
      "grad_norm": 1.9415287971496582,
      "learning_rate": 3.413997408478039e-06,
      "loss": 0.4947,
      "step": 186450
    },
    {
      "epoch": 76.32,
      "grad_norm": 1.728334903717041,
      "learning_rate": 3.413856280909869e-06,
      "loss": 0.5086,
      "step": 186460
    },
    {
      "epoch": 76.33,
      "grad_norm": 1.9811584949493408,
      "learning_rate": 3.4137151484397083e-06,
      "loss": 0.4876,
      "step": 186470
    },
    {
      "epoch": 76.33,
      "grad_norm": 1.5301415920257568,
      "learning_rate": 3.413574011068202e-06,
      "loss": 0.4764,
      "step": 186480
    },
    {
      "epoch": 76.34,
      "grad_norm": 2.078836679458618,
      "learning_rate": 3.4134328687959984e-06,
      "loss": 0.509,
      "step": 186490
    },
    {
      "epoch": 76.34,
      "grad_norm": 1.582627296447754,
      "learning_rate": 3.413291721623743e-06,
      "loss": 0.5176,
      "step": 186500
    },
    {
      "epoch": 76.34,
      "grad_norm": 1.7418407201766968,
      "learning_rate": 3.413150569552083e-06,
      "loss": 0.5273,
      "step": 186510
    },
    {
      "epoch": 76.35,
      "grad_norm": 1.9583230018615723,
      "learning_rate": 3.413009412581664e-06,
      "loss": 0.5303,
      "step": 186520
    },
    {
      "epoch": 76.35,
      "grad_norm": 1.4086339473724365,
      "learning_rate": 3.4128682507131338e-06,
      "loss": 0.4981,
      "step": 186530
    },
    {
      "epoch": 76.36,
      "grad_norm": 2.398127555847168,
      "learning_rate": 3.4127270839471383e-06,
      "loss": 0.5025,
      "step": 186540
    },
    {
      "epoch": 76.36,
      "grad_norm": 2.0597259998321533,
      "learning_rate": 3.412585912284324e-06,
      "loss": 0.5245,
      "step": 186550
    },
    {
      "epoch": 76.37,
      "grad_norm": 1.558879017829895,
      "learning_rate": 3.4124447357253383e-06,
      "loss": 0.5223,
      "step": 186560
    },
    {
      "epoch": 76.37,
      "grad_norm": 2.126598358154297,
      "learning_rate": 3.4123035542708272e-06,
      "loss": 0.5068,
      "step": 186570
    },
    {
      "epoch": 76.37,
      "grad_norm": 2.1833889484405518,
      "learning_rate": 3.412162367921438e-06,
      "loss": 0.5225,
      "step": 186580
    },
    {
      "epoch": 76.38,
      "grad_norm": 1.5427168607711792,
      "learning_rate": 3.4120211766778172e-06,
      "loss": 0.4928,
      "step": 186590
    },
    {
      "epoch": 76.38,
      "grad_norm": 3.98736572265625,
      "learning_rate": 3.411879980540611e-06,
      "loss": 0.5015,
      "step": 186600
    },
    {
      "epoch": 76.39,
      "grad_norm": 2.337307929992676,
      "learning_rate": 3.411738779510467e-06,
      "loss": 0.5192,
      "step": 186610
    },
    {
      "epoch": 76.39,
      "grad_norm": 1.9903286695480347,
      "learning_rate": 3.4115975735880324e-06,
      "loss": 0.4955,
      "step": 186620
    },
    {
      "epoch": 76.39,
      "grad_norm": 2.0030479431152344,
      "learning_rate": 3.4114563627739532e-06,
      "loss": 0.4943,
      "step": 186630
    },
    {
      "epoch": 76.4,
      "grad_norm": 1.509448766708374,
      "learning_rate": 3.4113151470688768e-06,
      "loss": 0.5075,
      "step": 186640
    },
    {
      "epoch": 76.4,
      "grad_norm": 1.6453429460525513,
      "learning_rate": 3.411173926473449e-06,
      "loss": 0.4928,
      "step": 186650
    },
    {
      "epoch": 76.41,
      "grad_norm": 2.2160465717315674,
      "learning_rate": 3.411032700988318e-06,
      "loss": 0.5035,
      "step": 186660
    },
    {
      "epoch": 76.41,
      "grad_norm": 2.2455875873565674,
      "learning_rate": 3.4108914706141302e-06,
      "loss": 0.5256,
      "step": 186670
    },
    {
      "epoch": 76.41,
      "grad_norm": 1.7681750059127808,
      "learning_rate": 3.4107502353515327e-06,
      "loss": 0.5156,
      "step": 186680
    },
    {
      "epoch": 76.42,
      "grad_norm": 1.6624681949615479,
      "learning_rate": 3.4106089952011713e-06,
      "loss": 0.5042,
      "step": 186690
    },
    {
      "epoch": 76.42,
      "grad_norm": 2.4360764026641846,
      "learning_rate": 3.4104677501636954e-06,
      "loss": 0.5155,
      "step": 186700
    },
    {
      "epoch": 76.43,
      "grad_norm": 1.2258139848709106,
      "learning_rate": 3.4103265002397507e-06,
      "loss": 0.5118,
      "step": 186710
    },
    {
      "epoch": 76.43,
      "grad_norm": 1.8856600522994995,
      "learning_rate": 3.4101852454299833e-06,
      "loss": 0.5198,
      "step": 186720
    },
    {
      "epoch": 76.43,
      "grad_norm": 2.161771535873413,
      "learning_rate": 3.4100439857350422e-06,
      "loss": 0.4918,
      "step": 186730
    },
    {
      "epoch": 76.44,
      "grad_norm": 1.6967449188232422,
      "learning_rate": 3.4099027211555726e-06,
      "loss": 0.4939,
      "step": 186740
    },
    {
      "epoch": 76.44,
      "grad_norm": 1.5648967027664185,
      "learning_rate": 3.409761451692223e-06,
      "loss": 0.5032,
      "step": 186750
    },
    {
      "epoch": 76.45,
      "grad_norm": 1.9036953449249268,
      "learning_rate": 3.4096201773456396e-06,
      "loss": 0.4979,
      "step": 186760
    },
    {
      "epoch": 76.45,
      "grad_norm": 1.678420901298523,
      "learning_rate": 3.4094788981164705e-06,
      "loss": 0.4935,
      "step": 186770
    },
    {
      "epoch": 76.46,
      "grad_norm": 1.6389504671096802,
      "learning_rate": 3.4093376140053625e-06,
      "loss": 0.5029,
      "step": 186780
    },
    {
      "epoch": 76.46,
      "grad_norm": 2.034750461578369,
      "learning_rate": 3.4091963250129617e-06,
      "loss": 0.4981,
      "step": 186790
    },
    {
      "epoch": 76.46,
      "grad_norm": 1.9384046792984009,
      "learning_rate": 3.4090550311399177e-06,
      "loss": 0.4946,
      "step": 186800
    },
    {
      "epoch": 76.47,
      "grad_norm": 1.4916393756866455,
      "learning_rate": 3.408913732386875e-06,
      "loss": 0.5171,
      "step": 186810
    },
    {
      "epoch": 76.47,
      "grad_norm": 1.84031081199646,
      "learning_rate": 3.408772428754483e-06,
      "loss": 0.497,
      "step": 186820
    },
    {
      "epoch": 76.48,
      "grad_norm": 2.182386875152588,
      "learning_rate": 3.408631120243388e-06,
      "loss": 0.5059,
      "step": 186830
    },
    {
      "epoch": 76.48,
      "grad_norm": 1.7271485328674316,
      "learning_rate": 3.408489806854237e-06,
      "loss": 0.5091,
      "step": 186840
    },
    {
      "epoch": 76.48,
      "grad_norm": 1.7164435386657715,
      "learning_rate": 3.4083484885876784e-06,
      "loss": 0.4926,
      "step": 186850
    },
    {
      "epoch": 76.49,
      "grad_norm": 1.7102527618408203,
      "learning_rate": 3.4082071654443593e-06,
      "loss": 0.5071,
      "step": 186860
    },
    {
      "epoch": 76.49,
      "grad_norm": 1.6709026098251343,
      "learning_rate": 3.408065837424927e-06,
      "loss": 0.5301,
      "step": 186870
    },
    {
      "epoch": 76.5,
      "grad_norm": 2.016017198562622,
      "learning_rate": 3.407924504530027e-06,
      "loss": 0.5136,
      "step": 186880
    },
    {
      "epoch": 76.5,
      "grad_norm": 1.8539717197418213,
      "learning_rate": 3.40778316676031e-06,
      "loss": 0.5131,
      "step": 186890
    },
    {
      "epoch": 76.5,
      "grad_norm": 2.024707317352295,
      "learning_rate": 3.4076418241164207e-06,
      "loss": 0.5187,
      "step": 186900
    },
    {
      "epoch": 76.51,
      "grad_norm": 1.829707384109497,
      "learning_rate": 3.407500476599008e-06,
      "loss": 0.522,
      "step": 186910
    },
    {
      "epoch": 76.51,
      "grad_norm": 1.6253786087036133,
      "learning_rate": 3.4073591242087205e-06,
      "loss": 0.5203,
      "step": 186920
    },
    {
      "epoch": 76.52,
      "grad_norm": 1.960517406463623,
      "learning_rate": 3.407217766946203e-06,
      "loss": 0.509,
      "step": 186930
    },
    {
      "epoch": 76.52,
      "grad_norm": 1.5736639499664307,
      "learning_rate": 3.4070764048121046e-06,
      "loss": 0.5026,
      "step": 186940
    },
    {
      "epoch": 76.52,
      "grad_norm": 2.1099252700805664,
      "learning_rate": 3.4069350378070728e-06,
      "loss": 0.5205,
      "step": 186950
    },
    {
      "epoch": 76.53,
      "grad_norm": 2.2217979431152344,
      "learning_rate": 3.406793665931755e-06,
      "loss": 0.5071,
      "step": 186960
    },
    {
      "epoch": 76.53,
      "grad_norm": 2.391677141189575,
      "learning_rate": 3.406652289186798e-06,
      "loss": 0.5033,
      "step": 186970
    },
    {
      "epoch": 76.54,
      "grad_norm": 1.674241542816162,
      "learning_rate": 3.406510907572851e-06,
      "loss": 0.522,
      "step": 186980
    },
    {
      "epoch": 76.54,
      "grad_norm": 1.8258332014083862,
      "learning_rate": 3.406369521090561e-06,
      "loss": 0.5176,
      "step": 186990
    },
    {
      "epoch": 76.55,
      "grad_norm": 2.8709921836853027,
      "learning_rate": 3.406228129740575e-06,
      "loss": 0.4901,
      "step": 187000
    },
    {
      "epoch": 76.55,
      "grad_norm": 1.6479841470718384,
      "learning_rate": 3.4060867335235417e-06,
      "loss": 0.5141,
      "step": 187010
    },
    {
      "epoch": 76.55,
      "grad_norm": 1.4907410144805908,
      "learning_rate": 3.405945332440108e-06,
      "loss": 0.5135,
      "step": 187020
    },
    {
      "epoch": 76.56,
      "grad_norm": 1.848317265510559,
      "learning_rate": 3.405803926490922e-06,
      "loss": 0.5045,
      "step": 187030
    },
    {
      "epoch": 76.56,
      "grad_norm": 2.2236318588256836,
      "learning_rate": 3.405662515676632e-06,
      "loss": 0.4966,
      "step": 187040
    },
    {
      "epoch": 76.57,
      "grad_norm": 1.5896514654159546,
      "learning_rate": 3.4055210999978843e-06,
      "loss": 0.4964,
      "step": 187050
    },
    {
      "epoch": 76.57,
      "grad_norm": 1.8628013134002686,
      "learning_rate": 3.405379679455328e-06,
      "loss": 0.4867,
      "step": 187060
    },
    {
      "epoch": 76.57,
      "grad_norm": 1.7702196836471558,
      "learning_rate": 3.4052382540496112e-06,
      "loss": 0.5149,
      "step": 187070
    },
    {
      "epoch": 76.58,
      "grad_norm": 1.6618447303771973,
      "learning_rate": 3.4050968237813805e-06,
      "loss": 0.501,
      "step": 187080
    },
    {
      "epoch": 76.58,
      "grad_norm": 2.5544028282165527,
      "learning_rate": 3.4049553886512844e-06,
      "loss": 0.5081,
      "step": 187090
    },
    {
      "epoch": 76.59,
      "grad_norm": 1.865706443786621,
      "learning_rate": 3.40481394865997e-06,
      "loss": 0.5253,
      "step": 187100
    },
    {
      "epoch": 76.59,
      "grad_norm": 2.507613182067871,
      "learning_rate": 3.4046725038080873e-06,
      "loss": 0.5074,
      "step": 187110
    },
    {
      "epoch": 76.59,
      "grad_norm": 1.7330669164657593,
      "learning_rate": 3.404531054096282e-06,
      "loss": 0.5051,
      "step": 187120
    },
    {
      "epoch": 76.6,
      "grad_norm": 2.022869825363159,
      "learning_rate": 3.4043895995252037e-06,
      "loss": 0.5135,
      "step": 187130
    },
    {
      "epoch": 76.6,
      "grad_norm": 1.6426515579223633,
      "learning_rate": 3.4042481400954997e-06,
      "loss": 0.5247,
      "step": 187140
    },
    {
      "epoch": 76.61,
      "grad_norm": 2.198150157928467,
      "learning_rate": 3.4041066758078173e-06,
      "loss": 0.5224,
      "step": 187150
    },
    {
      "epoch": 76.61,
      "grad_norm": 2.02933931350708,
      "learning_rate": 3.4039652066628055e-06,
      "loss": 0.5093,
      "step": 187160
    },
    {
      "epoch": 76.61,
      "grad_norm": 1.9399168491363525,
      "learning_rate": 3.4038237326611122e-06,
      "loss": 0.5282,
      "step": 187170
    },
    {
      "epoch": 76.62,
      "grad_norm": 2.1092002391815186,
      "learning_rate": 3.403682253803385e-06,
      "loss": 0.4959,
      "step": 187180
    },
    {
      "epoch": 76.62,
      "grad_norm": 2.2774252891540527,
      "learning_rate": 3.4035407700902725e-06,
      "loss": 0.5122,
      "step": 187190
    },
    {
      "epoch": 76.63,
      "grad_norm": 1.881734848022461,
      "learning_rate": 3.403399281522422e-06,
      "loss": 0.5097,
      "step": 187200
    },
    {
      "epoch": 76.63,
      "grad_norm": 2.231881618499756,
      "learning_rate": 3.403257788100484e-06,
      "loss": 0.5058,
      "step": 187210
    },
    {
      "epoch": 76.64,
      "grad_norm": 1.736578106880188,
      "learning_rate": 3.403116289825103e-06,
      "loss": 0.4985,
      "step": 187220
    },
    {
      "epoch": 76.64,
      "grad_norm": 2.130767583847046,
      "learning_rate": 3.40297478669693e-06,
      "loss": 0.5304,
      "step": 187230
    },
    {
      "epoch": 76.64,
      "grad_norm": 2.0060880184173584,
      "learning_rate": 3.402833278716612e-06,
      "loss": 0.5078,
      "step": 187240
    },
    {
      "epoch": 76.65,
      "grad_norm": 1.8854193687438965,
      "learning_rate": 3.402691765884797e-06,
      "loss": 0.5154,
      "step": 187250
    },
    {
      "epoch": 76.65,
      "grad_norm": 1.5768163204193115,
      "learning_rate": 3.4025502482021344e-06,
      "loss": 0.5132,
      "step": 187260
    },
    {
      "epoch": 76.66,
      "grad_norm": 1.6494852304458618,
      "learning_rate": 3.4024087256692717e-06,
      "loss": 0.5068,
      "step": 187270
    },
    {
      "epoch": 76.66,
      "grad_norm": 2.152228355407715,
      "learning_rate": 3.4022671982868577e-06,
      "loss": 0.4961,
      "step": 187280
    },
    {
      "epoch": 76.66,
      "grad_norm": 1.682260274887085,
      "learning_rate": 3.40212566605554e-06,
      "loss": 0.4959,
      "step": 187290
    },
    {
      "epoch": 76.67,
      "grad_norm": 2.0768775939941406,
      "learning_rate": 3.4019841289759677e-06,
      "loss": 0.5114,
      "step": 187300
    },
    {
      "epoch": 76.67,
      "grad_norm": 1.536238193511963,
      "learning_rate": 3.401842587048788e-06,
      "loss": 0.54,
      "step": 187310
    },
    {
      "epoch": 76.68,
      "grad_norm": 2.6671717166900635,
      "learning_rate": 3.4017010402746504e-06,
      "loss": 0.5101,
      "step": 187320
    },
    {
      "epoch": 76.68,
      "grad_norm": 1.9405473470687866,
      "learning_rate": 3.401559488654203e-06,
      "loss": 0.515,
      "step": 187330
    },
    {
      "epoch": 76.68,
      "grad_norm": 2.0558409690856934,
      "learning_rate": 3.4014179321880942e-06,
      "loss": 0.511,
      "step": 187340
    },
    {
      "epoch": 76.69,
      "grad_norm": 1.743296504020691,
      "learning_rate": 3.4012763708769726e-06,
      "loss": 0.5245,
      "step": 187350
    },
    {
      "epoch": 76.69,
      "grad_norm": 1.644816517829895,
      "learning_rate": 3.4011348047214857e-06,
      "loss": 0.4931,
      "step": 187360
    },
    {
      "epoch": 76.7,
      "grad_norm": 2.3793528079986572,
      "learning_rate": 3.4009932337222837e-06,
      "loss": 0.5193,
      "step": 187370
    },
    {
      "epoch": 76.7,
      "grad_norm": 1.8147294521331787,
      "learning_rate": 3.4008516578800136e-06,
      "loss": 0.5122,
      "step": 187380
    },
    {
      "epoch": 76.7,
      "grad_norm": 1.6677591800689697,
      "learning_rate": 3.4007100771953245e-06,
      "loss": 0.52,
      "step": 187390
    },
    {
      "epoch": 76.71,
      "grad_norm": 2.0957093238830566,
      "learning_rate": 3.4005684916688653e-06,
      "loss": 0.4951,
      "step": 187400
    },
    {
      "epoch": 76.71,
      "grad_norm": 2.1270065307617188,
      "learning_rate": 3.4004269013012843e-06,
      "loss": 0.5184,
      "step": 187410
    },
    {
      "epoch": 76.72,
      "grad_norm": 1.7537370920181274,
      "learning_rate": 3.4002853060932297e-06,
      "loss": 0.5259,
      "step": 187420
    },
    {
      "epoch": 76.72,
      "grad_norm": 1.845199704170227,
      "learning_rate": 3.4001437060453504e-06,
      "loss": 0.4951,
      "step": 187430
    },
    {
      "epoch": 76.73,
      "grad_norm": 1.83848237991333,
      "learning_rate": 3.400002101158295e-06,
      "loss": 0.4862,
      "step": 187440
    },
    {
      "epoch": 76.73,
      "grad_norm": 2.1771042346954346,
      "learning_rate": 3.399860491432713e-06,
      "loss": 0.4942,
      "step": 187450
    },
    {
      "epoch": 76.73,
      "grad_norm": 2.518933057785034,
      "learning_rate": 3.3997188768692516e-06,
      "loss": 0.5074,
      "step": 187460
    },
    {
      "epoch": 76.74,
      "grad_norm": 2.032130241394043,
      "learning_rate": 3.399577257468561e-06,
      "loss": 0.4893,
      "step": 187470
    },
    {
      "epoch": 76.74,
      "grad_norm": 2.0669617652893066,
      "learning_rate": 3.399435633231289e-06,
      "loss": 0.4944,
      "step": 187480
    },
    {
      "epoch": 76.75,
      "grad_norm": 1.9964561462402344,
      "learning_rate": 3.399294004158085e-06,
      "loss": 0.5263,
      "step": 187490
    },
    {
      "epoch": 76.75,
      "grad_norm": 1.5732427835464478,
      "learning_rate": 3.3991523702495967e-06,
      "loss": 0.5134,
      "step": 187500
    },
    {
      "epoch": 76.75,
      "grad_norm": 2.076139450073242,
      "learning_rate": 3.399010731506475e-06,
      "loss": 0.4964,
      "step": 187510
    },
    {
      "epoch": 76.76,
      "grad_norm": 1.9604498147964478,
      "learning_rate": 3.3988690879293658e-06,
      "loss": 0.4972,
      "step": 187520
    },
    {
      "epoch": 76.76,
      "grad_norm": 1.4681833982467651,
      "learning_rate": 3.3987274395189198e-06,
      "loss": 0.5144,
      "step": 187530
    },
    {
      "epoch": 76.77,
      "grad_norm": 1.8336663246154785,
      "learning_rate": 3.3985857862757853e-06,
      "loss": 0.5054,
      "step": 187540
    },
    {
      "epoch": 76.77,
      "grad_norm": 1.9154329299926758,
      "learning_rate": 3.398444128200611e-06,
      "loss": 0.5135,
      "step": 187550
    },
    {
      "epoch": 76.77,
      "grad_norm": 1.9133274555206299,
      "learning_rate": 3.3983024652940476e-06,
      "loss": 0.5302,
      "step": 187560
    },
    {
      "epoch": 76.78,
      "grad_norm": 1.996525764465332,
      "learning_rate": 3.3981607975567424e-06,
      "loss": 0.4864,
      "step": 187570
    },
    {
      "epoch": 76.78,
      "grad_norm": 2.0211293697357178,
      "learning_rate": 3.3980191249893445e-06,
      "loss": 0.5256,
      "step": 187580
    },
    {
      "epoch": 76.79,
      "grad_norm": 1.7263622283935547,
      "learning_rate": 3.3978774475925027e-06,
      "loss": 0.5019,
      "step": 187590
    },
    {
      "epoch": 76.79,
      "grad_norm": 1.998857855796814,
      "learning_rate": 3.397735765366866e-06,
      "loss": 0.5013,
      "step": 187600
    },
    {
      "epoch": 76.79,
      "grad_norm": 1.8690227270126343,
      "learning_rate": 3.3975940783130848e-06,
      "loss": 0.4916,
      "step": 187610
    },
    {
      "epoch": 76.8,
      "grad_norm": 1.9552154541015625,
      "learning_rate": 3.3974523864318063e-06,
      "loss": 0.5223,
      "step": 187620
    },
    {
      "epoch": 76.8,
      "grad_norm": 1.8803424835205078,
      "learning_rate": 3.39731068972368e-06,
      "loss": 0.5002,
      "step": 187630
    },
    {
      "epoch": 76.81,
      "grad_norm": 2.1016530990600586,
      "learning_rate": 3.3971689881893563e-06,
      "loss": 0.5119,
      "step": 187640
    },
    {
      "epoch": 76.81,
      "grad_norm": 2.201664686203003,
      "learning_rate": 3.3970272818294833e-06,
      "loss": 0.5172,
      "step": 187650
    },
    {
      "epoch": 76.82,
      "grad_norm": 1.832924485206604,
      "learning_rate": 3.3968855706447094e-06,
      "loss": 0.515,
      "step": 187660
    },
    {
      "epoch": 76.82,
      "grad_norm": 2.216050386428833,
      "learning_rate": 3.3967438546356847e-06,
      "loss": 0.5311,
      "step": 187670
    },
    {
      "epoch": 76.82,
      "grad_norm": 1.8054258823394775,
      "learning_rate": 3.3966021338030586e-06,
      "loss": 0.4974,
      "step": 187680
    },
    {
      "epoch": 76.83,
      "grad_norm": 2.6817984580993652,
      "learning_rate": 3.3964604081474794e-06,
      "loss": 0.5077,
      "step": 187690
    },
    {
      "epoch": 76.83,
      "grad_norm": 1.6597810983657837,
      "learning_rate": 3.3963186776695976e-06,
      "loss": 0.5376,
      "step": 187700
    },
    {
      "epoch": 76.84,
      "grad_norm": 1.6470515727996826,
      "learning_rate": 3.3961769423700613e-06,
      "loss": 0.5064,
      "step": 187710
    },
    {
      "epoch": 76.84,
      "grad_norm": 1.688418984413147,
      "learning_rate": 3.3960352022495204e-06,
      "loss": 0.4983,
      "step": 187720
    },
    {
      "epoch": 76.84,
      "grad_norm": 1.808264136314392,
      "learning_rate": 3.3958934573086235e-06,
      "loss": 0.5113,
      "step": 187730
    },
    {
      "epoch": 76.85,
      "grad_norm": 1.7750649452209473,
      "learning_rate": 3.395751707548021e-06,
      "loss": 0.5008,
      "step": 187740
    },
    {
      "epoch": 76.85,
      "grad_norm": 1.7779148817062378,
      "learning_rate": 3.395609952968361e-06,
      "loss": 0.5204,
      "step": 187750
    },
    {
      "epoch": 76.86,
      "grad_norm": 1.8060944080352783,
      "learning_rate": 3.395468193570294e-06,
      "loss": 0.5167,
      "step": 187760
    },
    {
      "epoch": 76.86,
      "grad_norm": 2.1745169162750244,
      "learning_rate": 3.3953264293544684e-06,
      "loss": 0.5095,
      "step": 187770
    },
    {
      "epoch": 76.86,
      "grad_norm": 1.868662714958191,
      "learning_rate": 3.3951846603215345e-06,
      "loss": 0.5193,
      "step": 187780
    },
    {
      "epoch": 76.87,
      "grad_norm": 1.5929477214813232,
      "learning_rate": 3.395042886472141e-06,
      "loss": 0.4981,
      "step": 187790
    },
    {
      "epoch": 76.87,
      "grad_norm": 2.016767740249634,
      "learning_rate": 3.3949011078069376e-06,
      "loss": 0.4916,
      "step": 187800
    },
    {
      "epoch": 76.88,
      "grad_norm": 1.3518662452697754,
      "learning_rate": 3.394759324326573e-06,
      "loss": 0.4997,
      "step": 187810
    },
    {
      "epoch": 76.88,
      "grad_norm": 1.7642923593521118,
      "learning_rate": 3.3946175360316983e-06,
      "loss": 0.4963,
      "step": 187820
    },
    {
      "epoch": 76.88,
      "grad_norm": 1.4932376146316528,
      "learning_rate": 3.3944757429229616e-06,
      "loss": 0.498,
      "step": 187830
    },
    {
      "epoch": 76.89,
      "grad_norm": 2.12091064453125,
      "learning_rate": 3.3943339450010135e-06,
      "loss": 0.491,
      "step": 187840
    },
    {
      "epoch": 76.89,
      "grad_norm": 1.6813523769378662,
      "learning_rate": 3.394192142266503e-06,
      "loss": 0.5125,
      "step": 187850
    },
    {
      "epoch": 76.9,
      "grad_norm": 2.427401065826416,
      "learning_rate": 3.3940503347200795e-06,
      "loss": 0.5213,
      "step": 187860
    },
    {
      "epoch": 76.9,
      "grad_norm": 2.3858299255371094,
      "learning_rate": 3.3939085223623937e-06,
      "loss": 0.4989,
      "step": 187870
    },
    {
      "epoch": 76.91,
      "grad_norm": 1.704255223274231,
      "learning_rate": 3.3937667051940936e-06,
      "loss": 0.5199,
      "step": 187880
    },
    {
      "epoch": 76.91,
      "grad_norm": 2.4452524185180664,
      "learning_rate": 3.39362488321583e-06,
      "loss": 0.5086,
      "step": 187890
    },
    {
      "epoch": 76.91,
      "grad_norm": 2.4628477096557617,
      "learning_rate": 3.393483056428251e-06,
      "loss": 0.4932,
      "step": 187900
    },
    {
      "epoch": 76.92,
      "grad_norm": 2.2505099773406982,
      "learning_rate": 3.393341224832008e-06,
      "loss": 0.497,
      "step": 187910
    },
    {
      "epoch": 76.92,
      "grad_norm": 1.749938726425171,
      "learning_rate": 3.3931993884277507e-06,
      "loss": 0.5009,
      "step": 187920
    },
    {
      "epoch": 76.93,
      "grad_norm": 1.7710535526275635,
      "learning_rate": 3.3930575472161285e-06,
      "loss": 0.4973,
      "step": 187930
    },
    {
      "epoch": 76.93,
      "grad_norm": 2.248939275741577,
      "learning_rate": 3.3929157011977905e-06,
      "loss": 0.5076,
      "step": 187940
    },
    {
      "epoch": 76.93,
      "grad_norm": 1.4029242992401123,
      "learning_rate": 3.3927738503733867e-06,
      "loss": 0.5021,
      "step": 187950
    },
    {
      "epoch": 76.94,
      "grad_norm": 1.5252137184143066,
      "learning_rate": 3.392631994743567e-06,
      "loss": 0.5053,
      "step": 187960
    },
    {
      "epoch": 76.94,
      "grad_norm": 1.9972271919250488,
      "learning_rate": 3.3924901343089812e-06,
      "loss": 0.5281,
      "step": 187970
    },
    {
      "epoch": 76.95,
      "grad_norm": 1.5707221031188965,
      "learning_rate": 3.39234826907028e-06,
      "loss": 0.5151,
      "step": 187980
    },
    {
      "epoch": 76.95,
      "grad_norm": 2.036635398864746,
      "learning_rate": 3.392206399028112e-06,
      "loss": 0.508,
      "step": 187990
    },
    {
      "epoch": 76.95,
      "grad_norm": 2.1944360733032227,
      "learning_rate": 3.392064524183128e-06,
      "loss": 0.5063,
      "step": 188000
    },
    {
      "epoch": 76.96,
      "grad_norm": 2.062809467315674,
      "learning_rate": 3.391922644535977e-06,
      "loss": 0.5201,
      "step": 188010
    },
    {
      "epoch": 76.96,
      "grad_norm": 1.8259304761886597,
      "learning_rate": 3.3917807600873096e-06,
      "loss": 0.5189,
      "step": 188020
    },
    {
      "epoch": 76.97,
      "grad_norm": 1.8165831565856934,
      "learning_rate": 3.391638870837776e-06,
      "loss": 0.4909,
      "step": 188030
    },
    {
      "epoch": 76.97,
      "grad_norm": 1.864509105682373,
      "learning_rate": 3.3914969767880253e-06,
      "loss": 0.5086,
      "step": 188040
    },
    {
      "epoch": 76.98,
      "grad_norm": 1.6178507804870605,
      "learning_rate": 3.3913550779387084e-06,
      "loss": 0.5207,
      "step": 188050
    },
    {
      "epoch": 76.98,
      "grad_norm": 1.8715617656707764,
      "learning_rate": 3.3912131742904742e-06,
      "loss": 0.5357,
      "step": 188060
    },
    {
      "epoch": 76.98,
      "grad_norm": 1.763301134109497,
      "learning_rate": 3.3910712658439738e-06,
      "loss": 0.5094,
      "step": 188070
    },
    {
      "epoch": 76.99,
      "grad_norm": 1.5800610780715942,
      "learning_rate": 3.3909293525998566e-06,
      "loss": 0.504,
      "step": 188080
    },
    {
      "epoch": 76.99,
      "grad_norm": 2.1346099376678467,
      "learning_rate": 3.3907874345587742e-06,
      "loss": 0.5042,
      "step": 188090
    },
    {
      "epoch": 77.0,
      "grad_norm": 1.9996801614761353,
      "learning_rate": 3.390645511721374e-06,
      "loss": 0.4983,
      "step": 188100
    },
    {
      "epoch": 77.0,
      "grad_norm": 2.0078980922698975,
      "learning_rate": 3.390503584088308e-06,
      "loss": 0.5288,
      "step": 188110
    },
    {
      "epoch": 77.0,
      "eval_loss": 0.510461688041687,
      "eval_runtime": 79.42,
      "eval_samples_per_second": 43.427,
      "eval_steps_per_second": 5.439,
      "step": 188111
    },
    {
      "epoch": 77.0,
      "grad_norm": 1.6557921171188354,
      "learning_rate": 3.3903616516602263e-06,
      "loss": 0.5082,
      "step": 188120
    },
    {
      "epoch": 77.01,
      "grad_norm": 1.7301627397537231,
      "learning_rate": 3.3902197144377783e-06,
      "loss": 0.5117,
      "step": 188130
    },
    {
      "epoch": 77.01,
      "grad_norm": 1.8056702613830566,
      "learning_rate": 3.3900777724216147e-06,
      "loss": 0.492,
      "step": 188140
    },
    {
      "epoch": 77.02,
      "grad_norm": 1.7639795541763306,
      "learning_rate": 3.3899358256123855e-06,
      "loss": 0.4867,
      "step": 188150
    },
    {
      "epoch": 77.02,
      "grad_norm": 1.7611790895462036,
      "learning_rate": 3.3897938740107422e-06,
      "loss": 0.4977,
      "step": 188160
    },
    {
      "epoch": 77.02,
      "grad_norm": 2.885936737060547,
      "learning_rate": 3.3896519176173327e-06,
      "loss": 0.5106,
      "step": 188170
    },
    {
      "epoch": 77.03,
      "grad_norm": 1.8350528478622437,
      "learning_rate": 3.389509956432809e-06,
      "loss": 0.5193,
      "step": 188180
    },
    {
      "epoch": 77.03,
      "grad_norm": 2.0001907348632812,
      "learning_rate": 3.3893679904578207e-06,
      "loss": 0.5075,
      "step": 188190
    },
    {
      "epoch": 77.04,
      "grad_norm": 1.6249191761016846,
      "learning_rate": 3.3892260196930175e-06,
      "loss": 0.5193,
      "step": 188200
    },
    {
      "epoch": 77.04,
      "grad_norm": 2.483853816986084,
      "learning_rate": 3.3890840441390515e-06,
      "loss": 0.5394,
      "step": 188210
    },
    {
      "epoch": 77.04,
      "grad_norm": 1.9826806783676147,
      "learning_rate": 3.3889420637965723e-06,
      "loss": 0.5008,
      "step": 188220
    },
    {
      "epoch": 77.05,
      "grad_norm": 2.0408236980438232,
      "learning_rate": 3.3888000786662297e-06,
      "loss": 0.5092,
      "step": 188230
    },
    {
      "epoch": 77.05,
      "grad_norm": 1.5429325103759766,
      "learning_rate": 3.388658088748675e-06,
      "loss": 0.4902,
      "step": 188240
    },
    {
      "epoch": 77.06,
      "grad_norm": 1.9518086910247803,
      "learning_rate": 3.3885160940445577e-06,
      "loss": 0.5136,
      "step": 188250
    },
    {
      "epoch": 77.06,
      "grad_norm": 1.634042501449585,
      "learning_rate": 3.3883740945545286e-06,
      "loss": 0.4902,
      "step": 188260
    },
    {
      "epoch": 77.07,
      "grad_norm": 2.9550373554229736,
      "learning_rate": 3.3882320902792386e-06,
      "loss": 0.5066,
      "step": 188270
    },
    {
      "epoch": 77.07,
      "grad_norm": 1.5919831991195679,
      "learning_rate": 3.388090081219338e-06,
      "loss": 0.5152,
      "step": 188280
    },
    {
      "epoch": 77.07,
      "grad_norm": 1.627949595451355,
      "learning_rate": 3.3879480673754768e-06,
      "loss": 0.4915,
      "step": 188290
    },
    {
      "epoch": 77.08,
      "grad_norm": 1.491575837135315,
      "learning_rate": 3.3878060487483065e-06,
      "loss": 0.4787,
      "step": 188300
    },
    {
      "epoch": 77.08,
      "grad_norm": 1.594417929649353,
      "learning_rate": 3.3876640253384765e-06,
      "loss": 0.4862,
      "step": 188310
    },
    {
      "epoch": 77.09,
      "grad_norm": 1.611967921257019,
      "learning_rate": 3.387521997146639e-06,
      "loss": 0.4966,
      "step": 188320
    },
    {
      "epoch": 77.09,
      "grad_norm": 2.041907548904419,
      "learning_rate": 3.3873799641734424e-06,
      "loss": 0.494,
      "step": 188330
    },
    {
      "epoch": 77.09,
      "grad_norm": 2.2299928665161133,
      "learning_rate": 3.3872379264195393e-06,
      "loss": 0.5104,
      "step": 188340
    },
    {
      "epoch": 77.1,
      "grad_norm": 2.379166603088379,
      "learning_rate": 3.3870958838855794e-06,
      "loss": 0.5242,
      "step": 188350
    },
    {
      "epoch": 77.1,
      "grad_norm": 1.5469927787780762,
      "learning_rate": 3.386953836572214e-06,
      "loss": 0.5023,
      "step": 188360
    },
    {
      "epoch": 77.11,
      "grad_norm": 1.692086100578308,
      "learning_rate": 3.386811784480093e-06,
      "loss": 0.4967,
      "step": 188370
    },
    {
      "epoch": 77.11,
      "grad_norm": 1.5752257108688354,
      "learning_rate": 3.3866697276098676e-06,
      "loss": 0.5257,
      "step": 188380
    },
    {
      "epoch": 77.11,
      "grad_norm": 2.2764008045196533,
      "learning_rate": 3.3865276659621883e-06,
      "loss": 0.4965,
      "step": 188390
    },
    {
      "epoch": 77.12,
      "grad_norm": 1.6104934215545654,
      "learning_rate": 3.386385599537706e-06,
      "loss": 0.522,
      "step": 188400
    },
    {
      "epoch": 77.12,
      "grad_norm": 1.5767220258712769,
      "learning_rate": 3.386243528337072e-06,
      "loss": 0.5049,
      "step": 188410
    },
    {
      "epoch": 77.13,
      "grad_norm": 1.7125916481018066,
      "learning_rate": 3.386101452360936e-06,
      "loss": 0.5274,
      "step": 188420
    },
    {
      "epoch": 77.13,
      "grad_norm": 2.095287561416626,
      "learning_rate": 3.3859593716099497e-06,
      "loss": 0.4859,
      "step": 188430
    },
    {
      "epoch": 77.13,
      "grad_norm": 1.9454323053359985,
      "learning_rate": 3.3858172860847635e-06,
      "loss": 0.493,
      "step": 188440
    },
    {
      "epoch": 77.14,
      "grad_norm": 1.9982218742370605,
      "learning_rate": 3.3856751957860287e-06,
      "loss": 0.5022,
      "step": 188450
    },
    {
      "epoch": 77.14,
      "grad_norm": 1.7661056518554688,
      "learning_rate": 3.3855331007143963e-06,
      "loss": 0.5003,
      "step": 188460
    },
    {
      "epoch": 77.15,
      "grad_norm": 1.7311944961547852,
      "learning_rate": 3.385391000870516e-06,
      "loss": 0.5082,
      "step": 188470
    },
    {
      "epoch": 77.15,
      "grad_norm": 1.664107322692871,
      "learning_rate": 3.3852488962550396e-06,
      "loss": 0.5445,
      "step": 188480
    },
    {
      "epoch": 77.16,
      "grad_norm": 1.8474040031433105,
      "learning_rate": 3.385106786868618e-06,
      "loss": 0.5079,
      "step": 188490
    },
    {
      "epoch": 77.16,
      "grad_norm": 1.8579457998275757,
      "learning_rate": 3.3849646727119028e-06,
      "loss": 0.5023,
      "step": 188500
    },
    {
      "epoch": 77.16,
      "grad_norm": 2.1007585525512695,
      "learning_rate": 3.384822553785544e-06,
      "loss": 0.5204,
      "step": 188510
    },
    {
      "epoch": 77.17,
      "grad_norm": 2.2074015140533447,
      "learning_rate": 3.384680430090194e-06,
      "loss": 0.5045,
      "step": 188520
    },
    {
      "epoch": 77.17,
      "grad_norm": 2.583173990249634,
      "learning_rate": 3.3845383016265023e-06,
      "loss": 0.5191,
      "step": 188530
    },
    {
      "epoch": 77.18,
      "grad_norm": 1.524420976638794,
      "learning_rate": 3.3843961683951208e-06,
      "loss": 0.4975,
      "step": 188540
    },
    {
      "epoch": 77.18,
      "grad_norm": 2.729262590408325,
      "learning_rate": 3.3842540303966997e-06,
      "loss": 0.5174,
      "step": 188550
    },
    {
      "epoch": 77.18,
      "grad_norm": 1.6561601161956787,
      "learning_rate": 3.384111887631891e-06,
      "loss": 0.5081,
      "step": 188560
    },
    {
      "epoch": 77.19,
      "grad_norm": 1.9315433502197266,
      "learning_rate": 3.383969740101346e-06,
      "loss": 0.5007,
      "step": 188570
    },
    {
      "epoch": 77.19,
      "grad_norm": 1.526045799255371,
      "learning_rate": 3.3838275878057154e-06,
      "loss": 0.5061,
      "step": 188580
    },
    {
      "epoch": 77.2,
      "grad_norm": 1.9803308248519897,
      "learning_rate": 3.38368543074565e-06,
      "loss": 0.5225,
      "step": 188590
    },
    {
      "epoch": 77.2,
      "grad_norm": 1.5692574977874756,
      "learning_rate": 3.383543268921802e-06,
      "loss": 0.5035,
      "step": 188600
    },
    {
      "epoch": 77.2,
      "grad_norm": 1.650346279144287,
      "learning_rate": 3.3834011023348216e-06,
      "loss": 0.4873,
      "step": 188610
    },
    {
      "epoch": 77.21,
      "grad_norm": 1.8396207094192505,
      "learning_rate": 3.383258930985361e-06,
      "loss": 0.5108,
      "step": 188620
    },
    {
      "epoch": 77.21,
      "grad_norm": 1.8092681169509888,
      "learning_rate": 3.3831167548740705e-06,
      "loss": 0.4968,
      "step": 188630
    },
    {
      "epoch": 77.22,
      "grad_norm": 1.6354072093963623,
      "learning_rate": 3.382974574001602e-06,
      "loss": 0.5027,
      "step": 188640
    },
    {
      "epoch": 77.22,
      "grad_norm": 1.7704497575759888,
      "learning_rate": 3.3828323883686065e-06,
      "loss": 0.5052,
      "step": 188650
    },
    {
      "epoch": 77.22,
      "grad_norm": 1.7843724489212036,
      "learning_rate": 3.382690197975736e-06,
      "loss": 0.4981,
      "step": 188660
    },
    {
      "epoch": 77.23,
      "grad_norm": 2.168631076812744,
      "learning_rate": 3.382548002823641e-06,
      "loss": 0.5067,
      "step": 188670
    },
    {
      "epoch": 77.23,
      "grad_norm": 2.308986186981201,
      "learning_rate": 3.3824058029129733e-06,
      "loss": 0.5194,
      "step": 188680
    },
    {
      "epoch": 77.24,
      "grad_norm": 1.9640010595321655,
      "learning_rate": 3.382263598244384e-06,
      "loss": 0.488,
      "step": 188690
    },
    {
      "epoch": 77.24,
      "grad_norm": 1.6445999145507812,
      "learning_rate": 3.3821213888185246e-06,
      "loss": 0.512,
      "step": 188700
    },
    {
      "epoch": 77.25,
      "grad_norm": 1.5775315761566162,
      "learning_rate": 3.3819791746360477e-06,
      "loss": 0.5062,
      "step": 188710
    },
    {
      "epoch": 77.25,
      "grad_norm": 1.6987652778625488,
      "learning_rate": 3.381836955697603e-06,
      "loss": 0.5112,
      "step": 188720
    },
    {
      "epoch": 77.25,
      "grad_norm": 1.6230584383010864,
      "learning_rate": 3.3816947320038427e-06,
      "loss": 0.5128,
      "step": 188730
    },
    {
      "epoch": 77.26,
      "grad_norm": 2.3944249153137207,
      "learning_rate": 3.381552503555419e-06,
      "loss": 0.5173,
      "step": 188740
    },
    {
      "epoch": 77.26,
      "grad_norm": 1.7197859287261963,
      "learning_rate": 3.381410270352982e-06,
      "loss": 0.4981,
      "step": 188750
    },
    {
      "epoch": 77.27,
      "grad_norm": 2.0636610984802246,
      "learning_rate": 3.3812680323971837e-06,
      "loss": 0.5036,
      "step": 188760
    },
    {
      "epoch": 77.27,
      "grad_norm": 1.795047640800476,
      "learning_rate": 3.3811257896886767e-06,
      "loss": 0.5159,
      "step": 188770
    },
    {
      "epoch": 77.27,
      "grad_norm": 2.3659729957580566,
      "learning_rate": 3.380983542228111e-06,
      "loss": 0.5066,
      "step": 188780
    },
    {
      "epoch": 77.28,
      "grad_norm": 5.410919666290283,
      "learning_rate": 3.38084129001614e-06,
      "loss": 0.52,
      "step": 188790
    },
    {
      "epoch": 77.28,
      "grad_norm": 1.8440935611724854,
      "learning_rate": 3.380699033053414e-06,
      "loss": 0.4995,
      "step": 188800
    },
    {
      "epoch": 77.29,
      "grad_norm": 2.1969666481018066,
      "learning_rate": 3.380556771340586e-06,
      "loss": 0.5324,
      "step": 188810
    },
    {
      "epoch": 77.29,
      "grad_norm": 2.066657543182373,
      "learning_rate": 3.380414504878306e-06,
      "loss": 0.4902,
      "step": 188820
    },
    {
      "epoch": 77.29,
      "grad_norm": 1.9838727712631226,
      "learning_rate": 3.380272233667226e-06,
      "loss": 0.5097,
      "step": 188830
    },
    {
      "epoch": 77.3,
      "grad_norm": 1.9227412939071655,
      "learning_rate": 3.3801299577079986e-06,
      "loss": 0.5036,
      "step": 188840
    },
    {
      "epoch": 77.3,
      "grad_norm": 2.4422121047973633,
      "learning_rate": 3.3799876770012755e-06,
      "loss": 0.5159,
      "step": 188850
    },
    {
      "epoch": 77.31,
      "grad_norm": 2.5206100940704346,
      "learning_rate": 3.3798453915477074e-06,
      "loss": 0.5103,
      "step": 188860
    },
    {
      "epoch": 77.31,
      "grad_norm": 2.4056670665740967,
      "learning_rate": 3.3797031013479465e-06,
      "loss": 0.5024,
      "step": 188870
    },
    {
      "epoch": 77.31,
      "grad_norm": 1.7881972789764404,
      "learning_rate": 3.379560806402646e-06,
      "loss": 0.5254,
      "step": 188880
    },
    {
      "epoch": 77.32,
      "grad_norm": 1.660719394683838,
      "learning_rate": 3.3794185067124557e-06,
      "loss": 0.515,
      "step": 188890
    },
    {
      "epoch": 77.32,
      "grad_norm": 1.7537821531295776,
      "learning_rate": 3.379276202278029e-06,
      "loss": 0.5214,
      "step": 188900
    },
    {
      "epoch": 77.33,
      "grad_norm": 2.1677627563476562,
      "learning_rate": 3.379133893100017e-06,
      "loss": 0.5075,
      "step": 188910
    },
    {
      "epoch": 77.33,
      "grad_norm": 1.3256950378417969,
      "learning_rate": 3.378991579179072e-06,
      "loss": 0.5031,
      "step": 188920
    },
    {
      "epoch": 77.34,
      "grad_norm": 1.68873929977417,
      "learning_rate": 3.3788492605158445e-06,
      "loss": 0.5095,
      "step": 188930
    },
    {
      "epoch": 77.34,
      "grad_norm": 1.683236002922058,
      "learning_rate": 3.3787069371109884e-06,
      "loss": 0.5137,
      "step": 188940
    },
    {
      "epoch": 77.34,
      "grad_norm": 2.118739604949951,
      "learning_rate": 3.3785646089651547e-06,
      "loss": 0.4905,
      "step": 188950
    },
    {
      "epoch": 77.35,
      "grad_norm": 1.9656202793121338,
      "learning_rate": 3.3784222760789956e-06,
      "loss": 0.5177,
      "step": 188960
    },
    {
      "epoch": 77.35,
      "grad_norm": 1.823632836341858,
      "learning_rate": 3.378279938453163e-06,
      "loss": 0.4904,
      "step": 188970
    },
    {
      "epoch": 77.36,
      "grad_norm": 1.8374805450439453,
      "learning_rate": 3.378137596088309e-06,
      "loss": 0.5024,
      "step": 188980
    },
    {
      "epoch": 77.36,
      "grad_norm": 2.1511595249176025,
      "learning_rate": 3.3779952489850856e-06,
      "loss": 0.4991,
      "step": 188990
    },
    {
      "epoch": 77.36,
      "grad_norm": 1.7354578971862793,
      "learning_rate": 3.377852897144145e-06,
      "loss": 0.5347,
      "step": 189000
    },
    {
      "epoch": 77.37,
      "grad_norm": 2.085127592086792,
      "learning_rate": 3.3777105405661394e-06,
      "loss": 0.5011,
      "step": 189010
    },
    {
      "epoch": 77.37,
      "grad_norm": 2.0893259048461914,
      "learning_rate": 3.3775681792517203e-06,
      "loss": 0.4965,
      "step": 189020
    },
    {
      "epoch": 77.38,
      "grad_norm": 1.8560116291046143,
      "learning_rate": 3.377425813201541e-06,
      "loss": 0.5061,
      "step": 189030
    },
    {
      "epoch": 77.38,
      "grad_norm": 2.3863279819488525,
      "learning_rate": 3.3772834424162516e-06,
      "loss": 0.5109,
      "step": 189040
    },
    {
      "epoch": 77.38,
      "grad_norm": 2.107030153274536,
      "learning_rate": 3.377141066896506e-06,
      "loss": 0.5255,
      "step": 189050
    },
    {
      "epoch": 77.39,
      "grad_norm": 1.5022897720336914,
      "learning_rate": 3.376998686642956e-06,
      "loss": 0.5107,
      "step": 189060
    },
    {
      "epoch": 77.39,
      "grad_norm": 1.94402015209198,
      "learning_rate": 3.376856301656254e-06,
      "loss": 0.5115,
      "step": 189070
    },
    {
      "epoch": 77.4,
      "grad_norm": 2.2844576835632324,
      "learning_rate": 3.376713911937052e-06,
      "loss": 0.516,
      "step": 189080
    },
    {
      "epoch": 77.4,
      "grad_norm": 1.6073172092437744,
      "learning_rate": 3.376571517486002e-06,
      "loss": 0.5108,
      "step": 189090
    },
    {
      "epoch": 77.4,
      "grad_norm": 1.6420788764953613,
      "learning_rate": 3.3764291183037574e-06,
      "loss": 0.5103,
      "step": 189100
    },
    {
      "epoch": 77.41,
      "grad_norm": 1.5544315576553345,
      "learning_rate": 3.376286714390969e-06,
      "loss": 0.5076,
      "step": 189110
    },
    {
      "epoch": 77.41,
      "grad_norm": 1.5961854457855225,
      "learning_rate": 3.3761443057482903e-06,
      "loss": 0.5151,
      "step": 189120
    },
    {
      "epoch": 77.42,
      "grad_norm": 1.6101351976394653,
      "learning_rate": 3.3760018923763728e-06,
      "loss": 0.5101,
      "step": 189130
    },
    {
      "epoch": 77.42,
      "grad_norm": 1.9311124086380005,
      "learning_rate": 3.3758594742758688e-06,
      "loss": 0.5294,
      "step": 189140
    },
    {
      "epoch": 77.43,
      "grad_norm": 1.6320271492004395,
      "learning_rate": 3.375717051447431e-06,
      "loss": 0.4987,
      "step": 189150
    },
    {
      "epoch": 77.43,
      "grad_norm": 1.8832601308822632,
      "learning_rate": 3.375574623891713e-06,
      "loss": 0.5205,
      "step": 189160
    },
    {
      "epoch": 77.43,
      "grad_norm": 1.7491637468338013,
      "learning_rate": 3.375432191609366e-06,
      "loss": 0.4983,
      "step": 189170
    },
    {
      "epoch": 77.44,
      "grad_norm": 1.6778069734573364,
      "learning_rate": 3.3752897546010427e-06,
      "loss": 0.5114,
      "step": 189180
    },
    {
      "epoch": 77.44,
      "grad_norm": 1.5850061178207397,
      "learning_rate": 3.375147312867395e-06,
      "loss": 0.5086,
      "step": 189190
    },
    {
      "epoch": 77.45,
      "grad_norm": 1.9832072257995605,
      "learning_rate": 3.375004866409076e-06,
      "loss": 0.5141,
      "step": 189200
    },
    {
      "epoch": 77.45,
      "grad_norm": 2.037816286087036,
      "learning_rate": 3.3748624152267387e-06,
      "loss": 0.5163,
      "step": 189210
    },
    {
      "epoch": 77.45,
      "grad_norm": 2.152970314025879,
      "learning_rate": 3.374719959321035e-06,
      "loss": 0.5229,
      "step": 189220
    },
    {
      "epoch": 77.46,
      "grad_norm": 1.4166728258132935,
      "learning_rate": 3.374577498692617e-06,
      "loss": 0.4866,
      "step": 189230
    },
    {
      "epoch": 77.46,
      "grad_norm": 1.6734768152236938,
      "learning_rate": 3.3744350333421384e-06,
      "loss": 0.5275,
      "step": 189240
    },
    {
      "epoch": 77.47,
      "grad_norm": 1.8474743366241455,
      "learning_rate": 3.3742925632702508e-06,
      "loss": 0.5009,
      "step": 189250
    },
    {
      "epoch": 77.47,
      "grad_norm": 2.1156952381134033,
      "learning_rate": 3.374150088477608e-06,
      "loss": 0.511,
      "step": 189260
    },
    {
      "epoch": 77.47,
      "grad_norm": 1.5634934902191162,
      "learning_rate": 3.374007608964861e-06,
      "loss": 0.5166,
      "step": 189270
    },
    {
      "epoch": 77.48,
      "grad_norm": 1.8852864503860474,
      "learning_rate": 3.3738651247326646e-06,
      "loss": 0.5219,
      "step": 189280
    },
    {
      "epoch": 77.48,
      "grad_norm": 2.466768980026245,
      "learning_rate": 3.3737226357816696e-06,
      "loss": 0.5095,
      "step": 189290
    },
    {
      "epoch": 77.49,
      "grad_norm": 1.530553936958313,
      "learning_rate": 3.37358014211253e-06,
      "loss": 0.4876,
      "step": 189300
    },
    {
      "epoch": 77.49,
      "grad_norm": 1.7868657112121582,
      "learning_rate": 3.373437643725897e-06,
      "loss": 0.5022,
      "step": 189310
    },
    {
      "epoch": 77.49,
      "grad_norm": 1.9101674556732178,
      "learning_rate": 3.3732951406224253e-06,
      "loss": 0.5145,
      "step": 189320
    },
    {
      "epoch": 77.5,
      "grad_norm": 1.6492509841918945,
      "learning_rate": 3.3731526328027664e-06,
      "loss": 0.4996,
      "step": 189330
    },
    {
      "epoch": 77.5,
      "grad_norm": 2.223158359527588,
      "learning_rate": 3.3730101202675735e-06,
      "loss": 0.5002,
      "step": 189340
    },
    {
      "epoch": 77.51,
      "grad_norm": 1.5159066915512085,
      "learning_rate": 3.3728676030175e-06,
      "loss": 0.5059,
      "step": 189350
    },
    {
      "epoch": 77.51,
      "grad_norm": 1.804656982421875,
      "learning_rate": 3.3727250810531972e-06,
      "loss": 0.503,
      "step": 189360
    },
    {
      "epoch": 77.52,
      "grad_norm": 2.4357717037200928,
      "learning_rate": 3.372582554375319e-06,
      "loss": 0.5136,
      "step": 189370
    },
    {
      "epoch": 77.52,
      "grad_norm": 1.6746114492416382,
      "learning_rate": 3.3724400229845185e-06,
      "loss": 0.5082,
      "step": 189380
    },
    {
      "epoch": 77.52,
      "grad_norm": 1.72357177734375,
      "learning_rate": 3.3722974868814486e-06,
      "loss": 0.5088,
      "step": 189390
    },
    {
      "epoch": 77.53,
      "grad_norm": 1.9171736240386963,
      "learning_rate": 3.372154946066762e-06,
      "loss": 0.5001,
      "step": 189400
    },
    {
      "epoch": 77.53,
      "grad_norm": 1.510678768157959,
      "learning_rate": 3.3720124005411115e-06,
      "loss": 0.4875,
      "step": 189410
    },
    {
      "epoch": 77.54,
      "grad_norm": 1.9881007671356201,
      "learning_rate": 3.371869850305149e-06,
      "loss": 0.5323,
      "step": 189420
    },
    {
      "epoch": 77.54,
      "grad_norm": 1.9409273862838745,
      "learning_rate": 3.3717272953595298e-06,
      "loss": 0.5161,
      "step": 189430
    },
    {
      "epoch": 77.54,
      "grad_norm": 1.5985857248306274,
      "learning_rate": 3.3715847357049053e-06,
      "loss": 0.4805,
      "step": 189440
    },
    {
      "epoch": 77.55,
      "grad_norm": 2.1715362071990967,
      "learning_rate": 3.37144217134193e-06,
      "loss": 0.5112,
      "step": 189450
    },
    {
      "epoch": 77.55,
      "grad_norm": 1.832220196723938,
      "learning_rate": 3.371299602271256e-06,
      "loss": 0.4934,
      "step": 189460
    },
    {
      "epoch": 77.56,
      "grad_norm": 2.040241003036499,
      "learning_rate": 3.3711570284935356e-06,
      "loss": 0.5052,
      "step": 189470
    },
    {
      "epoch": 77.56,
      "grad_norm": 1.697702407836914,
      "learning_rate": 3.371014450009423e-06,
      "loss": 0.4917,
      "step": 189480
    },
    {
      "epoch": 77.56,
      "grad_norm": 1.5728884935379028,
      "learning_rate": 3.370871866819571e-06,
      "loss": 0.5228,
      "step": 189490
    },
    {
      "epoch": 77.57,
      "grad_norm": 2.21805477142334,
      "learning_rate": 3.370729278924633e-06,
      "loss": 0.5226,
      "step": 189500
    },
    {
      "epoch": 77.57,
      "grad_norm": 2.088027000427246,
      "learning_rate": 3.3705866863252617e-06,
      "loss": 0.5042,
      "step": 189510
    },
    {
      "epoch": 77.58,
      "grad_norm": 1.9944286346435547,
      "learning_rate": 3.3704440890221102e-06,
      "loss": 0.5286,
      "step": 189520
    },
    {
      "epoch": 77.58,
      "grad_norm": 1.591992735862732,
      "learning_rate": 3.370301487015833e-06,
      "loss": 0.4988,
      "step": 189530
    },
    {
      "epoch": 77.58,
      "grad_norm": 1.908322811126709,
      "learning_rate": 3.3701588803070815e-06,
      "loss": 0.5278,
      "step": 189540
    },
    {
      "epoch": 77.59,
      "grad_norm": 1.8932620286941528,
      "learning_rate": 3.3700162688965103e-06,
      "loss": 0.5079,
      "step": 189550
    },
    {
      "epoch": 77.59,
      "grad_norm": 1.9628665447235107,
      "learning_rate": 3.3698736527847725e-06,
      "loss": 0.5229,
      "step": 189560
    },
    {
      "epoch": 77.6,
      "grad_norm": 1.8587496280670166,
      "learning_rate": 3.369731031972521e-06,
      "loss": 0.5227,
      "step": 189570
    },
    {
      "epoch": 77.6,
      "grad_norm": 2.4159111976623535,
      "learning_rate": 3.369588406460409e-06,
      "loss": 0.5057,
      "step": 189580
    },
    {
      "epoch": 77.61,
      "grad_norm": 1.714181661605835,
      "learning_rate": 3.3694457762490907e-06,
      "loss": 0.4938,
      "step": 189590
    },
    {
      "epoch": 77.61,
      "grad_norm": 2.5833468437194824,
      "learning_rate": 3.3693031413392184e-06,
      "loss": 0.5043,
      "step": 189600
    },
    {
      "epoch": 77.61,
      "grad_norm": 2.4406490325927734,
      "learning_rate": 3.369160501731446e-06,
      "loss": 0.5055,
      "step": 189610
    },
    {
      "epoch": 77.62,
      "grad_norm": 1.9995123147964478,
      "learning_rate": 3.369017857426427e-06,
      "loss": 0.5271,
      "step": 189620
    },
    {
      "epoch": 77.62,
      "grad_norm": 1.8415782451629639,
      "learning_rate": 3.368875208424815e-06,
      "loss": 0.4941,
      "step": 189630
    },
    {
      "epoch": 77.63,
      "grad_norm": 1.9561759233474731,
      "learning_rate": 3.368732554727263e-06,
      "loss": 0.5182,
      "step": 189640
    },
    {
      "epoch": 77.63,
      "grad_norm": 2.0061533451080322,
      "learning_rate": 3.3685898963344247e-06,
      "loss": 0.5257,
      "step": 189650
    },
    {
      "epoch": 77.63,
      "grad_norm": 2.1474053859710693,
      "learning_rate": 3.3684472332469537e-06,
      "loss": 0.5077,
      "step": 189660
    },
    {
      "epoch": 77.64,
      "grad_norm": 2.1666805744171143,
      "learning_rate": 3.368304565465503e-06,
      "loss": 0.5006,
      "step": 189670
    },
    {
      "epoch": 77.64,
      "grad_norm": 1.9359387159347534,
      "learning_rate": 3.3681618929907266e-06,
      "loss": 0.4981,
      "step": 189680
    },
    {
      "epoch": 77.65,
      "grad_norm": 1.4193899631500244,
      "learning_rate": 3.368019215823278e-06,
      "loss": 0.4978,
      "step": 189690
    },
    {
      "epoch": 77.65,
      "grad_norm": 1.9018763303756714,
      "learning_rate": 3.3678765339638103e-06,
      "loss": 0.4924,
      "step": 189700
    },
    {
      "epoch": 77.65,
      "grad_norm": 2.2226510047912598,
      "learning_rate": 3.3677338474129786e-06,
      "loss": 0.4958,
      "step": 189710
    },
    {
      "epoch": 77.66,
      "grad_norm": 1.8822901248931885,
      "learning_rate": 3.3675911561714347e-06,
      "loss": 0.4978,
      "step": 189720
    },
    {
      "epoch": 77.66,
      "grad_norm": 2.2640466690063477,
      "learning_rate": 3.367448460239833e-06,
      "loss": 0.5035,
      "step": 189730
    },
    {
      "epoch": 77.67,
      "grad_norm": 1.7202352285385132,
      "learning_rate": 3.367305759618827e-06,
      "loss": 0.5278,
      "step": 189740
    },
    {
      "epoch": 77.67,
      "grad_norm": 1.457676649093628,
      "learning_rate": 3.3671630543090715e-06,
      "loss": 0.5072,
      "step": 189750
    },
    {
      "epoch": 77.67,
      "grad_norm": 2.0819082260131836,
      "learning_rate": 3.367020344311219e-06,
      "loss": 0.5235,
      "step": 189760
    },
    {
      "epoch": 77.68,
      "grad_norm": 1.6632603406906128,
      "learning_rate": 3.366877629625924e-06,
      "loss": 0.516,
      "step": 189770
    },
    {
      "epoch": 77.68,
      "grad_norm": 1.9063655138015747,
      "learning_rate": 3.3667349102538385e-06,
      "loss": 0.4932,
      "step": 189780
    },
    {
      "epoch": 77.69,
      "grad_norm": 1.8353220224380493,
      "learning_rate": 3.3665921861956184e-06,
      "loss": 0.5028,
      "step": 189790
    },
    {
      "epoch": 77.69,
      "grad_norm": 2.1904067993164062,
      "learning_rate": 3.366449457451916e-06,
      "loss": 0.5079,
      "step": 189800
    },
    {
      "epoch": 77.7,
      "grad_norm": 1.6667275428771973,
      "learning_rate": 3.366306724023386e-06,
      "loss": 0.502,
      "step": 189810
    },
    {
      "epoch": 77.7,
      "grad_norm": 1.7627769708633423,
      "learning_rate": 3.366163985910683e-06,
      "loss": 0.4984,
      "step": 189820
    },
    {
      "epoch": 77.7,
      "grad_norm": 1.632826566696167,
      "learning_rate": 3.3660212431144584e-06,
      "loss": 0.5071,
      "step": 189830
    },
    {
      "epoch": 77.71,
      "grad_norm": 1.7376676797866821,
      "learning_rate": 3.365878495635368e-06,
      "loss": 0.4998,
      "step": 189840
    },
    {
      "epoch": 77.71,
      "grad_norm": 2.06706166267395,
      "learning_rate": 3.3657357434740657e-06,
      "loss": 0.5206,
      "step": 189850
    },
    {
      "epoch": 77.72,
      "grad_norm": 1.782711148262024,
      "learning_rate": 3.3655929866312053e-06,
      "loss": 0.5002,
      "step": 189860
    },
    {
      "epoch": 77.72,
      "grad_norm": 1.846975564956665,
      "learning_rate": 3.3654502251074397e-06,
      "loss": 0.5159,
      "step": 189870
    },
    {
      "epoch": 77.72,
      "grad_norm": 2.2223987579345703,
      "learning_rate": 3.3653074589034237e-06,
      "loss": 0.5208,
      "step": 189880
    },
    {
      "epoch": 77.73,
      "grad_norm": 1.5536302328109741,
      "learning_rate": 3.365164688019811e-06,
      "loss": 0.5252,
      "step": 189890
    },
    {
      "epoch": 77.73,
      "grad_norm": 1.4907927513122559,
      "learning_rate": 3.365021912457256e-06,
      "loss": 0.5241,
      "step": 189900
    },
    {
      "epoch": 77.74,
      "grad_norm": 2.149977922439575,
      "learning_rate": 3.3648791322164125e-06,
      "loss": 0.5047,
      "step": 189910
    },
    {
      "epoch": 77.74,
      "grad_norm": 2.786513328552246,
      "learning_rate": 3.364736347297935e-06,
      "loss": 0.5016,
      "step": 189920
    },
    {
      "epoch": 77.74,
      "grad_norm": 1.9711412191390991,
      "learning_rate": 3.364593557702477e-06,
      "loss": 0.5096,
      "step": 189930
    },
    {
      "epoch": 77.75,
      "grad_norm": 1.4529505968093872,
      "learning_rate": 3.3644507634306925e-06,
      "loss": 0.5334,
      "step": 189940
    },
    {
      "epoch": 77.75,
      "grad_norm": 1.6018569469451904,
      "learning_rate": 3.364307964483236e-06,
      "loss": 0.5027,
      "step": 189950
    },
    {
      "epoch": 77.76,
      "grad_norm": 1.537896990776062,
      "learning_rate": 3.364165160860761e-06,
      "loss": 0.4766,
      "step": 189960
    },
    {
      "epoch": 77.76,
      "grad_norm": 1.8588228225708008,
      "learning_rate": 3.3640223525639237e-06,
      "loss": 0.5094,
      "step": 189970
    },
    {
      "epoch": 77.77,
      "grad_norm": 1.6389868259429932,
      "learning_rate": 3.363879539593376e-06,
      "loss": 0.5065,
      "step": 189980
    },
    {
      "epoch": 77.77,
      "grad_norm": 2.2700624465942383,
      "learning_rate": 3.363736721949772e-06,
      "loss": 0.5075,
      "step": 189990
    },
    {
      "epoch": 77.77,
      "grad_norm": 1.5011115074157715,
      "learning_rate": 3.3635938996337673e-06,
      "loss": 0.4986,
      "step": 190000
    },
    {
      "epoch": 77.78,
      "grad_norm": 2.1617307662963867,
      "learning_rate": 3.363451072646015e-06,
      "loss": 0.5031,
      "step": 190010
    },
    {
      "epoch": 77.78,
      "grad_norm": 1.8658790588378906,
      "learning_rate": 3.363308240987171e-06,
      "loss": 0.4946,
      "step": 190020
    },
    {
      "epoch": 77.79,
      "grad_norm": 1.7199569940567017,
      "learning_rate": 3.3631654046578885e-06,
      "loss": 0.4937,
      "step": 190030
    },
    {
      "epoch": 77.79,
      "grad_norm": 2.012791872024536,
      "learning_rate": 3.3630225636588223e-06,
      "loss": 0.4916,
      "step": 190040
    },
    {
      "epoch": 77.79,
      "grad_norm": 2.063918113708496,
      "learning_rate": 3.3628797179906254e-06,
      "loss": 0.488,
      "step": 190050
    },
    {
      "epoch": 77.8,
      "grad_norm": 1.8464161157608032,
      "learning_rate": 3.362736867653954e-06,
      "loss": 0.5318,
      "step": 190060
    },
    {
      "epoch": 77.8,
      "grad_norm": 1.7725781202316284,
      "learning_rate": 3.362594012649461e-06,
      "loss": 0.4915,
      "step": 190070
    },
    {
      "epoch": 77.81,
      "grad_norm": 1.9317482709884644,
      "learning_rate": 3.362451152977801e-06,
      "loss": 0.5208,
      "step": 190080
    },
    {
      "epoch": 77.81,
      "grad_norm": 1.689655065536499,
      "learning_rate": 3.362308288639629e-06,
      "loss": 0.5083,
      "step": 190090
    },
    {
      "epoch": 77.81,
      "grad_norm": 1.5592416524887085,
      "learning_rate": 3.3621654196355993e-06,
      "loss": 0.5019,
      "step": 190100
    },
    {
      "epoch": 77.82,
      "grad_norm": 1.649262547492981,
      "learning_rate": 3.362022545966367e-06,
      "loss": 0.5287,
      "step": 190110
    },
    {
      "epoch": 77.82,
      "grad_norm": 1.7925037145614624,
      "learning_rate": 3.361879667632585e-06,
      "loss": 0.4909,
      "step": 190120
    },
    {
      "epoch": 77.83,
      "grad_norm": 2.688605308532715,
      "learning_rate": 3.3617367846349096e-06,
      "loss": 0.4995,
      "step": 190130
    },
    {
      "epoch": 77.83,
      "grad_norm": 1.9161101579666138,
      "learning_rate": 3.3615938969739935e-06,
      "loss": 0.5126,
      "step": 190140
    },
    {
      "epoch": 77.83,
      "grad_norm": 2.099846124649048,
      "learning_rate": 3.3614510046504926e-06,
      "loss": 0.5168,
      "step": 190150
    },
    {
      "epoch": 77.84,
      "grad_norm": 1.6737369298934937,
      "learning_rate": 3.3613081076650604e-06,
      "loss": 0.5067,
      "step": 190160
    },
    {
      "epoch": 77.84,
      "grad_norm": 1.7550989389419556,
      "learning_rate": 3.3611652060183526e-06,
      "loss": 0.5132,
      "step": 190170
    },
    {
      "epoch": 77.85,
      "grad_norm": 1.956051230430603,
      "learning_rate": 3.361022299711023e-06,
      "loss": 0.5047,
      "step": 190180
    },
    {
      "epoch": 77.85,
      "grad_norm": 1.6637036800384521,
      "learning_rate": 3.3608793887437268e-06,
      "loss": 0.5174,
      "step": 190190
    },
    {
      "epoch": 77.86,
      "grad_norm": 1.515895962715149,
      "learning_rate": 3.3607364731171183e-06,
      "loss": 0.51,
      "step": 190200
    },
    {
      "epoch": 77.86,
      "grad_norm": 1.6659165620803833,
      "learning_rate": 3.3605935528318524e-06,
      "loss": 0.4958,
      "step": 190210
    },
    {
      "epoch": 77.86,
      "grad_norm": 1.6235555410385132,
      "learning_rate": 3.3604506278885833e-06,
      "loss": 0.5125,
      "step": 190220
    },
    {
      "epoch": 77.87,
      "grad_norm": 1.992089033126831,
      "learning_rate": 3.3603076982879667e-06,
      "loss": 0.5158,
      "step": 190230
    },
    {
      "epoch": 77.87,
      "grad_norm": 1.5666747093200684,
      "learning_rate": 3.3601647640306564e-06,
      "loss": 0.5056,
      "step": 190240
    },
    {
      "epoch": 77.88,
      "grad_norm": 1.4944725036621094,
      "learning_rate": 3.360021825117307e-06,
      "loss": 0.5043,
      "step": 190250
    },
    {
      "epoch": 77.88,
      "grad_norm": 1.7657631635665894,
      "learning_rate": 3.3598788815485747e-06,
      "loss": 0.5163,
      "step": 190260
    },
    {
      "epoch": 77.88,
      "grad_norm": 2.0547337532043457,
      "learning_rate": 3.359735933325113e-06,
      "loss": 0.5069,
      "step": 190270
    },
    {
      "epoch": 77.89,
      "grad_norm": 1.7694056034088135,
      "learning_rate": 3.359592980447577e-06,
      "loss": 0.4997,
      "step": 190280
    },
    {
      "epoch": 77.89,
      "grad_norm": 1.9796292781829834,
      "learning_rate": 3.359450022916621e-06,
      "loss": 0.5089,
      "step": 190290
    },
    {
      "epoch": 77.9,
      "grad_norm": 1.891592025756836,
      "learning_rate": 3.359307060732901e-06,
      "loss": 0.5001,
      "step": 190300
    },
    {
      "epoch": 77.9,
      "grad_norm": 2.042497396469116,
      "learning_rate": 3.3591640938970715e-06,
      "loss": 0.5091,
      "step": 190310
    },
    {
      "epoch": 77.9,
      "grad_norm": 1.818657636642456,
      "learning_rate": 3.3590211224097875e-06,
      "loss": 0.5006,
      "step": 190320
    },
    {
      "epoch": 77.91,
      "grad_norm": 1.5644758939743042,
      "learning_rate": 3.358878146271704e-06,
      "loss": 0.4982,
      "step": 190330
    },
    {
      "epoch": 77.91,
      "grad_norm": 1.5225226879119873,
      "learning_rate": 3.3587351654834747e-06,
      "loss": 0.4949,
      "step": 190340
    },
    {
      "epoch": 77.92,
      "grad_norm": 2.4285688400268555,
      "learning_rate": 3.3585921800457562e-06,
      "loss": 0.5035,
      "step": 190350
    },
    {
      "epoch": 77.92,
      "grad_norm": 2.111618995666504,
      "learning_rate": 3.358449189959202e-06,
      "loss": 0.5023,
      "step": 190360
    },
    {
      "epoch": 77.92,
      "grad_norm": 1.8526992797851562,
      "learning_rate": 3.3583061952244686e-06,
      "loss": 0.5184,
      "step": 190370
    },
    {
      "epoch": 77.93,
      "grad_norm": 2.2725441455841064,
      "learning_rate": 3.3581631958422103e-06,
      "loss": 0.4892,
      "step": 190380
    },
    {
      "epoch": 77.93,
      "grad_norm": 1.7291094064712524,
      "learning_rate": 3.3580201918130825e-06,
      "loss": 0.5196,
      "step": 190390
    },
    {
      "epoch": 77.94,
      "grad_norm": 1.7114496231079102,
      "learning_rate": 3.3578771831377403e-06,
      "loss": 0.5094,
      "step": 190400
    },
    {
      "epoch": 77.94,
      "grad_norm": 1.8095389604568481,
      "learning_rate": 3.3577341698168377e-06,
      "loss": 0.4875,
      "step": 190410
    },
    {
      "epoch": 77.95,
      "grad_norm": 2.1237435340881348,
      "learning_rate": 3.357591151851031e-06,
      "loss": 0.4937,
      "step": 190420
    },
    {
      "epoch": 77.95,
      "grad_norm": 1.8499512672424316,
      "learning_rate": 3.3574481292409756e-06,
      "loss": 0.5122,
      "step": 190430
    },
    {
      "epoch": 77.95,
      "grad_norm": 2.0948336124420166,
      "learning_rate": 3.3573051019873255e-06,
      "loss": 0.4953,
      "step": 190440
    },
    {
      "epoch": 77.96,
      "grad_norm": 1.7607694864273071,
      "learning_rate": 3.3571620700907364e-06,
      "loss": 0.5117,
      "step": 190450
    },
    {
      "epoch": 77.96,
      "grad_norm": 1.9775031805038452,
      "learning_rate": 3.3570190335518633e-06,
      "loss": 0.5084,
      "step": 190460
    },
    {
      "epoch": 77.97,
      "grad_norm": 2.3328049182891846,
      "learning_rate": 3.3568759923713623e-06,
      "loss": 0.4958,
      "step": 190470
    },
    {
      "epoch": 77.97,
      "grad_norm": 1.6638025045394897,
      "learning_rate": 3.3567329465498873e-06,
      "loss": 0.5033,
      "step": 190480
    },
    {
      "epoch": 77.97,
      "grad_norm": 17.326398849487305,
      "learning_rate": 3.356589896088095e-06,
      "loss": 0.5291,
      "step": 190490
    },
    {
      "epoch": 77.98,
      "grad_norm": 1.4590411186218262,
      "learning_rate": 3.3564468409866395e-06,
      "loss": 0.5308,
      "step": 190500
    },
    {
      "epoch": 77.98,
      "grad_norm": 1.9405574798583984,
      "learning_rate": 3.356303781246177e-06,
      "loss": 0.4874,
      "step": 190510
    },
    {
      "epoch": 77.99,
      "grad_norm": 1.8210376501083374,
      "learning_rate": 3.3561607168673625e-06,
      "loss": 0.5087,
      "step": 190520
    },
    {
      "epoch": 77.99,
      "grad_norm": 1.9162311553955078,
      "learning_rate": 3.3560176478508504e-06,
      "loss": 0.5058,
      "step": 190530
    },
    {
      "epoch": 77.99,
      "grad_norm": 2.237902879714966,
      "learning_rate": 3.355874574197298e-06,
      "loss": 0.5069,
      "step": 190540
    },
    {
      "epoch": 78.0,
      "grad_norm": 1.667717695236206,
      "learning_rate": 3.355731495907359e-06,
      "loss": 0.514,
      "step": 190550
    },
    {
      "epoch": 78.0,
      "eval_loss": 0.5071980357170105,
      "eval_runtime": 74.7729,
      "eval_samples_per_second": 46.126,
      "eval_steps_per_second": 5.777,
      "step": 190554
    },
    {
      "epoch": 78.0,
      "grad_norm": 1.8157811164855957,
      "learning_rate": 3.35558841298169e-06,
      "loss": 0.5068,
      "step": 190560
    },
    {
      "epoch": 78.01,
      "grad_norm": 1.5306906700134277,
      "learning_rate": 3.3554453254209462e-06,
      "loss": 0.5088,
      "step": 190570
    },
    {
      "epoch": 78.01,
      "grad_norm": 1.510362982749939,
      "learning_rate": 3.355302233225782e-06,
      "loss": 0.5109,
      "step": 190580
    },
    {
      "epoch": 78.01,
      "grad_norm": 1.3764657974243164,
      "learning_rate": 3.355159136396854e-06,
      "loss": 0.4974,
      "step": 190590
    },
    {
      "epoch": 78.02,
      "grad_norm": 1.9666200876235962,
      "learning_rate": 3.3550160349348173e-06,
      "loss": 0.5191,
      "step": 190600
    },
    {
      "epoch": 78.02,
      "grad_norm": 1.7805094718933105,
      "learning_rate": 3.3548729288403273e-06,
      "loss": 0.4974,
      "step": 190610
    },
    {
      "epoch": 78.03,
      "grad_norm": 1.5128843784332275,
      "learning_rate": 3.354729818114041e-06,
      "loss": 0.5135,
      "step": 190620
    },
    {
      "epoch": 78.03,
      "grad_norm": 2.0142734050750732,
      "learning_rate": 3.354586702756611e-06,
      "loss": 0.5048,
      "step": 190630
    },
    {
      "epoch": 78.04,
      "grad_norm": 2.773242950439453,
      "learning_rate": 3.3544435827686963e-06,
      "loss": 0.4842,
      "step": 190640
    },
    {
      "epoch": 78.04,
      "grad_norm": 1.7775822877883911,
      "learning_rate": 3.3543004581509493e-06,
      "loss": 0.4975,
      "step": 190650
    },
    {
      "epoch": 78.04,
      "grad_norm": 2.344925880432129,
      "learning_rate": 3.3541573289040275e-06,
      "loss": 0.5011,
      "step": 190660
    },
    {
      "epoch": 78.05,
      "grad_norm": 2.377856731414795,
      "learning_rate": 3.3540141950285864e-06,
      "loss": 0.5081,
      "step": 190670
    },
    {
      "epoch": 78.05,
      "grad_norm": 1.7798985242843628,
      "learning_rate": 3.3538710565252814e-06,
      "loss": 0.5419,
      "step": 190680
    },
    {
      "epoch": 78.06,
      "grad_norm": 2.090562105178833,
      "learning_rate": 3.353727913394769e-06,
      "loss": 0.4902,
      "step": 190690
    },
    {
      "epoch": 78.06,
      "grad_norm": 1.500836730003357,
      "learning_rate": 3.353584765637704e-06,
      "loss": 0.4918,
      "step": 190700
    },
    {
      "epoch": 78.06,
      "grad_norm": 2.445895195007324,
      "learning_rate": 3.353441613254742e-06,
      "loss": 0.516,
      "step": 190710
    },
    {
      "epoch": 78.07,
      "grad_norm": 1.8787031173706055,
      "learning_rate": 3.3532984562465387e-06,
      "loss": 0.5148,
      "step": 190720
    },
    {
      "epoch": 78.07,
      "grad_norm": 1.9353179931640625,
      "learning_rate": 3.353155294613751e-06,
      "loss": 0.5127,
      "step": 190730
    },
    {
      "epoch": 78.08,
      "grad_norm": 2.1354000568389893,
      "learning_rate": 3.3530121283570327e-06,
      "loss": 0.5241,
      "step": 190740
    },
    {
      "epoch": 78.08,
      "grad_norm": 1.7571735382080078,
      "learning_rate": 3.352868957477042e-06,
      "loss": 0.5193,
      "step": 190750
    },
    {
      "epoch": 78.08,
      "grad_norm": 2.108395576477051,
      "learning_rate": 3.352725781974434e-06,
      "loss": 0.5326,
      "step": 190760
    },
    {
      "epoch": 78.09,
      "grad_norm": 1.9999918937683105,
      "learning_rate": 3.352582601849863e-06,
      "loss": 0.4875,
      "step": 190770
    },
    {
      "epoch": 78.09,
      "grad_norm": 1.954153299331665,
      "learning_rate": 3.3524394171039873e-06,
      "loss": 0.5108,
      "step": 190780
    },
    {
      "epoch": 78.1,
      "grad_norm": 2.0222108364105225,
      "learning_rate": 3.352296227737461e-06,
      "loss": 0.5402,
      "step": 190790
    },
    {
      "epoch": 78.1,
      "grad_norm": 2.0763556957244873,
      "learning_rate": 3.35215303375094e-06,
      "loss": 0.5254,
      "step": 190800
    },
    {
      "epoch": 78.1,
      "grad_norm": 1.43897545337677,
      "learning_rate": 3.3520098351450815e-06,
      "loss": 0.5011,
      "step": 190810
    },
    {
      "epoch": 78.11,
      "grad_norm": 1.6712580919265747,
      "learning_rate": 3.351866631920541e-06,
      "loss": 0.5051,
      "step": 190820
    },
    {
      "epoch": 78.11,
      "grad_norm": 1.4710025787353516,
      "learning_rate": 3.3517234240779743e-06,
      "loss": 0.512,
      "step": 190830
    },
    {
      "epoch": 78.12,
      "grad_norm": 2.00323486328125,
      "learning_rate": 3.3515802116180373e-06,
      "loss": 0.4929,
      "step": 190840
    },
    {
      "epoch": 78.12,
      "grad_norm": 1.6349917650222778,
      "learning_rate": 3.351436994541386e-06,
      "loss": 0.5082,
      "step": 190850
    },
    {
      "epoch": 78.13,
      "grad_norm": 1.7263600826263428,
      "learning_rate": 3.351293772848677e-06,
      "loss": 0.5133,
      "step": 190860
    },
    {
      "epoch": 78.13,
      "grad_norm": 2.1933515071868896,
      "learning_rate": 3.351150546540566e-06,
      "loss": 0.5116,
      "step": 190870
    },
    {
      "epoch": 78.13,
      "grad_norm": 1.9455486536026,
      "learning_rate": 3.351007315617709e-06,
      "loss": 0.4938,
      "step": 190880
    },
    {
      "epoch": 78.14,
      "grad_norm": 1.710392713546753,
      "learning_rate": 3.350864080080762e-06,
      "loss": 0.5128,
      "step": 190890
    },
    {
      "epoch": 78.14,
      "grad_norm": 1.5085241794586182,
      "learning_rate": 3.3507208399303817e-06,
      "loss": 0.5175,
      "step": 190900
    },
    {
      "epoch": 78.15,
      "grad_norm": 1.4134784936904907,
      "learning_rate": 3.3505775951672245e-06,
      "loss": 0.501,
      "step": 190910
    },
    {
      "epoch": 78.15,
      "grad_norm": 2.2810516357421875,
      "learning_rate": 3.350434345791945e-06,
      "loss": 0.4893,
      "step": 190920
    },
    {
      "epoch": 78.15,
      "grad_norm": 2.517559766769409,
      "learning_rate": 3.3502910918052004e-06,
      "loss": 0.5171,
      "step": 190930
    },
    {
      "epoch": 78.16,
      "grad_norm": 2.197657823562622,
      "learning_rate": 3.350147833207647e-06,
      "loss": 0.508,
      "step": 190940
    },
    {
      "epoch": 78.16,
      "grad_norm": 1.6926711797714233,
      "learning_rate": 3.350004569999941e-06,
      "loss": 0.5133,
      "step": 190950
    },
    {
      "epoch": 78.17,
      "grad_norm": 2.8688158988952637,
      "learning_rate": 3.3498613021827392e-06,
      "loss": 0.5198,
      "step": 190960
    },
    {
      "epoch": 78.17,
      "grad_norm": 1.9872504472732544,
      "learning_rate": 3.349718029756697e-06,
      "loss": 0.502,
      "step": 190970
    },
    {
      "epoch": 78.17,
      "grad_norm": 1.7452775239944458,
      "learning_rate": 3.3495747527224716e-06,
      "loss": 0.5006,
      "step": 190980
    },
    {
      "epoch": 78.18,
      "grad_norm": 1.7086092233657837,
      "learning_rate": 3.349431471080718e-06,
      "loss": 0.5143,
      "step": 190990
    },
    {
      "epoch": 78.18,
      "grad_norm": 2.0111327171325684,
      "learning_rate": 3.349288184832094e-06,
      "loss": 0.4985,
      "step": 191000
    },
    {
      "epoch": 78.19,
      "grad_norm": 1.6367861032485962,
      "learning_rate": 3.349144893977255e-06,
      "loss": 0.513,
      "step": 191010
    },
    {
      "epoch": 78.19,
      "grad_norm": 1.736564040184021,
      "learning_rate": 3.3490015985168575e-06,
      "loss": 0.5258,
      "step": 191020
    },
    {
      "epoch": 78.19,
      "grad_norm": 1.7668070793151855,
      "learning_rate": 3.348858298451558e-06,
      "loss": 0.4938,
      "step": 191030
    },
    {
      "epoch": 78.2,
      "grad_norm": 1.7656699419021606,
      "learning_rate": 3.348714993782013e-06,
      "loss": 0.5138,
      "step": 191040
    },
    {
      "epoch": 78.2,
      "grad_norm": 1.9097143411636353,
      "learning_rate": 3.34857168450888e-06,
      "loss": 0.5236,
      "step": 191050
    },
    {
      "epoch": 78.21,
      "grad_norm": 2.2719216346740723,
      "learning_rate": 3.348428370632814e-06,
      "loss": 0.5292,
      "step": 191060
    },
    {
      "epoch": 78.21,
      "grad_norm": 2.1909689903259277,
      "learning_rate": 3.3482850521544715e-06,
      "loss": 0.4915,
      "step": 191070
    },
    {
      "epoch": 78.22,
      "grad_norm": 2.1572060585021973,
      "learning_rate": 3.3481417290745095e-06,
      "loss": 0.5178,
      "step": 191080
    },
    {
      "epoch": 78.22,
      "grad_norm": 1.4555788040161133,
      "learning_rate": 3.347998401393585e-06,
      "loss": 0.5104,
      "step": 191090
    },
    {
      "epoch": 78.22,
      "grad_norm": 1.991676688194275,
      "learning_rate": 3.347855069112353e-06,
      "loss": 0.5088,
      "step": 191100
    },
    {
      "epoch": 78.23,
      "grad_norm": 1.7504466772079468,
      "learning_rate": 3.3477117322314724e-06,
      "loss": 0.4966,
      "step": 191110
    },
    {
      "epoch": 78.23,
      "grad_norm": 1.4262531995773315,
      "learning_rate": 3.3475683907515984e-06,
      "loss": 0.485,
      "step": 191120
    },
    {
      "epoch": 78.24,
      "grad_norm": 1.8906753063201904,
      "learning_rate": 3.347425044673388e-06,
      "loss": 0.5029,
      "step": 191130
    },
    {
      "epoch": 78.24,
      "grad_norm": 1.9607884883880615,
      "learning_rate": 3.3472816939974974e-06,
      "loss": 0.4907,
      "step": 191140
    },
    {
      "epoch": 78.24,
      "grad_norm": 1.81626296043396,
      "learning_rate": 3.3471383387245834e-06,
      "loss": 0.5111,
      "step": 191150
    },
    {
      "epoch": 78.25,
      "grad_norm": 2.380032539367676,
      "learning_rate": 3.3469949788553026e-06,
      "loss": 0.5087,
      "step": 191160
    },
    {
      "epoch": 78.25,
      "grad_norm": 1.6986027956008911,
      "learning_rate": 3.3468516143903125e-06,
      "loss": 0.512,
      "step": 191170
    },
    {
      "epoch": 78.26,
      "grad_norm": 1.3813769817352295,
      "learning_rate": 3.3467082453302685e-06,
      "loss": 0.5072,
      "step": 191180
    },
    {
      "epoch": 78.26,
      "grad_norm": 1.8214154243469238,
      "learning_rate": 3.346564871675829e-06,
      "loss": 0.535,
      "step": 191190
    },
    {
      "epoch": 78.26,
      "grad_norm": 2.07570219039917,
      "learning_rate": 3.3464214934276493e-06,
      "loss": 0.5056,
      "step": 191200
    },
    {
      "epoch": 78.27,
      "grad_norm": 1.54766845703125,
      "learning_rate": 3.3462781105863866e-06,
      "loss": 0.5245,
      "step": 191210
    },
    {
      "epoch": 78.27,
      "grad_norm": 2.233182907104492,
      "learning_rate": 3.346134723152698e-06,
      "loss": 0.5007,
      "step": 191220
    },
    {
      "epoch": 78.28,
      "grad_norm": 2.727509021759033,
      "learning_rate": 3.3459913311272403e-06,
      "loss": 0.5091,
      "step": 191230
    },
    {
      "epoch": 78.28,
      "grad_norm": 2.2468342781066895,
      "learning_rate": 3.3458479345106705e-06,
      "loss": 0.5148,
      "step": 191240
    },
    {
      "epoch": 78.28,
      "grad_norm": 2.020477294921875,
      "learning_rate": 3.3457045333036455e-06,
      "loss": 0.5086,
      "step": 191250
    },
    {
      "epoch": 78.29,
      "grad_norm": 2.036484479904175,
      "learning_rate": 3.3455611275068212e-06,
      "loss": 0.516,
      "step": 191260
    },
    {
      "epoch": 78.29,
      "grad_norm": 1.558415174484253,
      "learning_rate": 3.3454177171208563e-06,
      "loss": 0.5122,
      "step": 191270
    },
    {
      "epoch": 78.3,
      "grad_norm": 2.1228625774383545,
      "learning_rate": 3.345274302146406e-06,
      "loss": 0.5199,
      "step": 191280
    },
    {
      "epoch": 78.3,
      "grad_norm": 1.4409111738204956,
      "learning_rate": 3.345130882584128e-06,
      "loss": 0.5271,
      "step": 191290
    },
    {
      "epoch": 78.31,
      "grad_norm": 2.0864017009735107,
      "learning_rate": 3.3449874584346793e-06,
      "loss": 0.4875,
      "step": 191300
    },
    {
      "epoch": 78.31,
      "grad_norm": 1.457327127456665,
      "learning_rate": 3.3448440296987167e-06,
      "loss": 0.5308,
      "step": 191310
    },
    {
      "epoch": 78.31,
      "grad_norm": 1.8503813743591309,
      "learning_rate": 3.3447005963768974e-06,
      "loss": 0.5182,
      "step": 191320
    },
    {
      "epoch": 78.32,
      "grad_norm": 2.077493190765381,
      "learning_rate": 3.3445571584698788e-06,
      "loss": 0.4958,
      "step": 191330
    },
    {
      "epoch": 78.32,
      "grad_norm": 2.0341956615448,
      "learning_rate": 3.3444137159783175e-06,
      "loss": 0.5259,
      "step": 191340
    },
    {
      "epoch": 78.33,
      "grad_norm": 1.4852581024169922,
      "learning_rate": 3.3442702689028707e-06,
      "loss": 0.5136,
      "step": 191350
    },
    {
      "epoch": 78.33,
      "grad_norm": 2.2415709495544434,
      "learning_rate": 3.344126817244195e-06,
      "loss": 0.4989,
      "step": 191360
    },
    {
      "epoch": 78.33,
      "grad_norm": 2.8013601303100586,
      "learning_rate": 3.3439833610029483e-06,
      "loss": 0.4806,
      "step": 191370
    },
    {
      "epoch": 78.34,
      "grad_norm": 2.42842435836792,
      "learning_rate": 3.343839900179788e-06,
      "loss": 0.5205,
      "step": 191380
    },
    {
      "epoch": 78.34,
      "grad_norm": 1.7015935182571411,
      "learning_rate": 3.34369643477537e-06,
      "loss": 0.5237,
      "step": 191390
    },
    {
      "epoch": 78.35,
      "grad_norm": 1.82025146484375,
      "learning_rate": 3.343552964790352e-06,
      "loss": 0.498,
      "step": 191400
    },
    {
      "epoch": 78.35,
      "grad_norm": 1.851669192314148,
      "learning_rate": 3.343409490225393e-06,
      "loss": 0.4913,
      "step": 191410
    },
    {
      "epoch": 78.35,
      "grad_norm": 1.5146950483322144,
      "learning_rate": 3.3432660110811477e-06,
      "loss": 0.4967,
      "step": 191420
    },
    {
      "epoch": 78.36,
      "grad_norm": 1.764428734779358,
      "learning_rate": 3.343122527358274e-06,
      "loss": 0.509,
      "step": 191430
    },
    {
      "epoch": 78.36,
      "grad_norm": 1.910283088684082,
      "learning_rate": 3.34297903905743e-06,
      "loss": 0.4986,
      "step": 191440
    },
    {
      "epoch": 78.37,
      "grad_norm": 1.8716442584991455,
      "learning_rate": 3.342835546179273e-06,
      "loss": 0.4999,
      "step": 191450
    },
    {
      "epoch": 78.37,
      "grad_norm": 2.3284103870391846,
      "learning_rate": 3.342692048724459e-06,
      "loss": 0.4972,
      "step": 191460
    },
    {
      "epoch": 78.37,
      "grad_norm": 1.7639074325561523,
      "learning_rate": 3.342548546693646e-06,
      "loss": 0.4956,
      "step": 191470
    },
    {
      "epoch": 78.38,
      "grad_norm": 2.19663405418396,
      "learning_rate": 3.342405040087492e-06,
      "loss": 0.5062,
      "step": 191480
    },
    {
      "epoch": 78.38,
      "grad_norm": 1.2947434186935425,
      "learning_rate": 3.3422615289066543e-06,
      "loss": 0.4921,
      "step": 191490
    },
    {
      "epoch": 78.39,
      "grad_norm": 1.5736273527145386,
      "learning_rate": 3.3421180131517893e-06,
      "loss": 0.4939,
      "step": 191500
    },
    {
      "epoch": 78.39,
      "grad_norm": 1.7676277160644531,
      "learning_rate": 3.3419744928235555e-06,
      "loss": 0.5033,
      "step": 191510
    },
    {
      "epoch": 78.4,
      "grad_norm": 2.4212892055511475,
      "learning_rate": 3.3418309679226096e-06,
      "loss": 0.5222,
      "step": 191520
    },
    {
      "epoch": 78.4,
      "grad_norm": 1.8879104852676392,
      "learning_rate": 3.3416874384496094e-06,
      "loss": 0.4934,
      "step": 191530
    },
    {
      "epoch": 78.4,
      "grad_norm": 3.037449836730957,
      "learning_rate": 3.3415439044052126e-06,
      "loss": 0.5183,
      "step": 191540
    },
    {
      "epoch": 78.41,
      "grad_norm": 1.4749624729156494,
      "learning_rate": 3.341400365790076e-06,
      "loss": 0.5012,
      "step": 191550
    },
    {
      "epoch": 78.41,
      "grad_norm": 1.7616409063339233,
      "learning_rate": 3.341256822604858e-06,
      "loss": 0.4951,
      "step": 191560
    },
    {
      "epoch": 78.42,
      "grad_norm": 2.6534037590026855,
      "learning_rate": 3.341113274850216e-06,
      "loss": 0.5028,
      "step": 191570
    },
    {
      "epoch": 78.42,
      "grad_norm": 2.7772457599639893,
      "learning_rate": 3.340969722526807e-06,
      "loss": 0.513,
      "step": 191580
    },
    {
      "epoch": 78.42,
      "grad_norm": 1.8402377367019653,
      "learning_rate": 3.3408261656352885e-06,
      "loss": 0.4978,
      "step": 191590
    },
    {
      "epoch": 78.43,
      "grad_norm": 2.2413127422332764,
      "learning_rate": 3.3406826041763184e-06,
      "loss": 0.4845,
      "step": 191600
    },
    {
      "epoch": 78.43,
      "grad_norm": 2.644787549972534,
      "learning_rate": 3.340539038150555e-06,
      "loss": 0.497,
      "step": 191610
    },
    {
      "epoch": 78.44,
      "grad_norm": 1.9730271100997925,
      "learning_rate": 3.340395467558655e-06,
      "loss": 0.5213,
      "step": 191620
    },
    {
      "epoch": 78.44,
      "grad_norm": 2.1449480056762695,
      "learning_rate": 3.340251892401277e-06,
      "loss": 0.5041,
      "step": 191630
    },
    {
      "epoch": 78.44,
      "grad_norm": 2.0331482887268066,
      "learning_rate": 3.3401083126790774e-06,
      "loss": 0.5066,
      "step": 191640
    },
    {
      "epoch": 78.45,
      "grad_norm": 1.6105493307113647,
      "learning_rate": 3.3399647283927156e-06,
      "loss": 0.5161,
      "step": 191650
    },
    {
      "epoch": 78.45,
      "grad_norm": 1.534545660018921,
      "learning_rate": 3.339821139542848e-06,
      "loss": 0.5013,
      "step": 191660
    },
    {
      "epoch": 78.46,
      "grad_norm": 2.03940486907959,
      "learning_rate": 3.3396775461301323e-06,
      "loss": 0.5016,
      "step": 191670
    },
    {
      "epoch": 78.46,
      "grad_norm": 1.9389457702636719,
      "learning_rate": 3.339533948155227e-06,
      "loss": 0.481,
      "step": 191680
    },
    {
      "epoch": 78.47,
      "grad_norm": 1.8653101921081543,
      "learning_rate": 3.3393903456187895e-06,
      "loss": 0.5222,
      "step": 191690
    },
    {
      "epoch": 78.47,
      "grad_norm": 2.1891562938690186,
      "learning_rate": 3.3392467385214784e-06,
      "loss": 0.503,
      "step": 191700
    },
    {
      "epoch": 78.47,
      "grad_norm": 2.006058692932129,
      "learning_rate": 3.3391031268639504e-06,
      "loss": 0.5244,
      "step": 191710
    },
    {
      "epoch": 78.48,
      "grad_norm": 1.6495625972747803,
      "learning_rate": 3.3389595106468642e-06,
      "loss": 0.4896,
      "step": 191720
    },
    {
      "epoch": 78.48,
      "grad_norm": 2.3707518577575684,
      "learning_rate": 3.338815889870877e-06,
      "loss": 0.4918,
      "step": 191730
    },
    {
      "epoch": 78.49,
      "grad_norm": 2.1133792400360107,
      "learning_rate": 3.338672264536647e-06,
      "loss": 0.4987,
      "step": 191740
    },
    {
      "epoch": 78.49,
      "grad_norm": 1.6493721008300781,
      "learning_rate": 3.338528634644833e-06,
      "loss": 0.4924,
      "step": 191750
    },
    {
      "epoch": 78.49,
      "grad_norm": 1.944403886795044,
      "learning_rate": 3.338385000196091e-06,
      "loss": 0.5299,
      "step": 191760
    },
    {
      "epoch": 78.5,
      "grad_norm": 2.009845733642578,
      "learning_rate": 3.338241361191081e-06,
      "loss": 0.5128,
      "step": 191770
    },
    {
      "epoch": 78.5,
      "grad_norm": 1.8565707206726074,
      "learning_rate": 3.3380977176304593e-06,
      "loss": 0.4929,
      "step": 191780
    },
    {
      "epoch": 78.51,
      "grad_norm": 1.7273023128509521,
      "learning_rate": 3.3379540695148857e-06,
      "loss": 0.4917,
      "step": 191790
    },
    {
      "epoch": 78.51,
      "grad_norm": 2.0123085975646973,
      "learning_rate": 3.3378104168450167e-06,
      "loss": 0.5297,
      "step": 191800
    },
    {
      "epoch": 78.51,
      "grad_norm": 2.0677571296691895,
      "learning_rate": 3.3376667596215105e-06,
      "loss": 0.5106,
      "step": 191810
    },
    {
      "epoch": 78.52,
      "grad_norm": 1.9792195558547974,
      "learning_rate": 3.337523097845026e-06,
      "loss": 0.5056,
      "step": 191820
    },
    {
      "epoch": 78.52,
      "grad_norm": 1.771776795387268,
      "learning_rate": 3.3373794315162205e-06,
      "loss": 0.4786,
      "step": 191830
    },
    {
      "epoch": 78.53,
      "grad_norm": 1.7211847305297852,
      "learning_rate": 3.337235760635753e-06,
      "loss": 0.5033,
      "step": 191840
    },
    {
      "epoch": 78.53,
      "grad_norm": 2.0668017864227295,
      "learning_rate": 3.3370920852042804e-06,
      "loss": 0.502,
      "step": 191850
    },
    {
      "epoch": 78.53,
      "grad_norm": 1.8906174898147583,
      "learning_rate": 3.3369484052224625e-06,
      "loss": 0.5262,
      "step": 191860
    },
    {
      "epoch": 78.54,
      "grad_norm": 1.626007318496704,
      "learning_rate": 3.3368047206909554e-06,
      "loss": 0.4911,
      "step": 191870
    },
    {
      "epoch": 78.54,
      "grad_norm": 2.4452526569366455,
      "learning_rate": 3.3366610316104193e-06,
      "loss": 0.5112,
      "step": 191880
    },
    {
      "epoch": 78.55,
      "grad_norm": 1.9994535446166992,
      "learning_rate": 3.336517337981511e-06,
      "loss": 0.5017,
      "step": 191890
    },
    {
      "epoch": 78.55,
      "grad_norm": 1.6184512376785278,
      "learning_rate": 3.33637363980489e-06,
      "loss": 0.4969,
      "step": 191900
    },
    {
      "epoch": 78.56,
      "grad_norm": 1.786630630493164,
      "learning_rate": 3.336229937081213e-06,
      "loss": 0.5153,
      "step": 191910
    },
    {
      "epoch": 78.56,
      "grad_norm": 1.697646975517273,
      "learning_rate": 3.3360862298111395e-06,
      "loss": 0.501,
      "step": 191920
    },
    {
      "epoch": 78.56,
      "grad_norm": 1.7941875457763672,
      "learning_rate": 3.3359425179953277e-06,
      "loss": 0.5244,
      "step": 191930
    },
    {
      "epoch": 78.57,
      "grad_norm": 1.9140450954437256,
      "learning_rate": 3.335798801634435e-06,
      "loss": 0.5056,
      "step": 191940
    },
    {
      "epoch": 78.57,
      "grad_norm": 1.6595711708068848,
      "learning_rate": 3.335655080729121e-06,
      "loss": 0.496,
      "step": 191950
    },
    {
      "epoch": 78.58,
      "grad_norm": 1.7804993391036987,
      "learning_rate": 3.335511355280043e-06,
      "loss": 0.5258,
      "step": 191960
    },
    {
      "epoch": 78.58,
      "grad_norm": 2.093642473220825,
      "learning_rate": 3.335367625287859e-06,
      "loss": 0.4986,
      "step": 191970
    },
    {
      "epoch": 78.58,
      "grad_norm": 1.9554202556610107,
      "learning_rate": 3.3352238907532295e-06,
      "loss": 0.5164,
      "step": 191980
    },
    {
      "epoch": 78.59,
      "grad_norm": 1.7616779804229736,
      "learning_rate": 3.335080151676812e-06,
      "loss": 0.4994,
      "step": 191990
    },
    {
      "epoch": 78.59,
      "grad_norm": 2.238802909851074,
      "learning_rate": 3.3349364080592635e-06,
      "loss": 0.5199,
      "step": 192000
    },
    {
      "epoch": 78.6,
      "grad_norm": 1.77321195602417,
      "learning_rate": 3.334792659901244e-06,
      "loss": 0.4945,
      "step": 192010
    },
    {
      "epoch": 78.6,
      "grad_norm": 1.98164963722229,
      "learning_rate": 3.334648907203412e-06,
      "loss": 0.5107,
      "step": 192020
    },
    {
      "epoch": 78.6,
      "grad_norm": 1.9525474309921265,
      "learning_rate": 3.3345051499664247e-06,
      "loss": 0.5133,
      "step": 192030
    },
    {
      "epoch": 78.61,
      "grad_norm": 1.5697546005249023,
      "learning_rate": 3.334361388190942e-06,
      "loss": 0.5301,
      "step": 192040
    },
    {
      "epoch": 78.61,
      "grad_norm": 2.1011109352111816,
      "learning_rate": 3.334217621877622e-06,
      "loss": 0.504,
      "step": 192050
    },
    {
      "epoch": 78.62,
      "grad_norm": 1.7041341066360474,
      "learning_rate": 3.334073851027123e-06,
      "loss": 0.4994,
      "step": 192060
    },
    {
      "epoch": 78.62,
      "grad_norm": 2.01117205619812,
      "learning_rate": 3.333930075640104e-06,
      "loss": 0.4988,
      "step": 192070
    },
    {
      "epoch": 78.62,
      "grad_norm": 1.9730573892593384,
      "learning_rate": 3.3337862957172236e-06,
      "loss": 0.5006,
      "step": 192080
    },
    {
      "epoch": 78.63,
      "grad_norm": 1.994924783706665,
      "learning_rate": 3.33364251125914e-06,
      "loss": 0.493,
      "step": 192090
    },
    {
      "epoch": 78.63,
      "grad_norm": 1.7794901132583618,
      "learning_rate": 3.3334987222665115e-06,
      "loss": 0.4967,
      "step": 192100
    },
    {
      "epoch": 78.64,
      "grad_norm": 1.605815052986145,
      "learning_rate": 3.333354928739998e-06,
      "loss": 0.49,
      "step": 192110
    },
    {
      "epoch": 78.64,
      "grad_norm": 1.5681565999984741,
      "learning_rate": 3.3332111306802577e-06,
      "loss": 0.5364,
      "step": 192120
    },
    {
      "epoch": 78.65,
      "grad_norm": 1.6798170804977417,
      "learning_rate": 3.3330673280879487e-06,
      "loss": 0.5018,
      "step": 192130
    },
    {
      "epoch": 78.65,
      "grad_norm": 2.4364583492279053,
      "learning_rate": 3.332923520963731e-06,
      "loss": 0.4973,
      "step": 192140
    },
    {
      "epoch": 78.65,
      "grad_norm": 1.9807325601577759,
      "learning_rate": 3.332779709308262e-06,
      "loss": 0.5094,
      "step": 192150
    },
    {
      "epoch": 78.66,
      "grad_norm": 1.9913054704666138,
      "learning_rate": 3.332635893122201e-06,
      "loss": 0.5103,
      "step": 192160
    },
    {
      "epoch": 78.66,
      "grad_norm": 2.0056228637695312,
      "learning_rate": 3.3324920724062073e-06,
      "loss": 0.5025,
      "step": 192170
    },
    {
      "epoch": 78.67,
      "grad_norm": 1.8502111434936523,
      "learning_rate": 3.3323482471609393e-06,
      "loss": 0.5055,
      "step": 192180
    },
    {
      "epoch": 78.67,
      "grad_norm": 2.3238184452056885,
      "learning_rate": 3.3322044173870554e-06,
      "loss": 0.5163,
      "step": 192190
    },
    {
      "epoch": 78.67,
      "grad_norm": 1.905821681022644,
      "learning_rate": 3.332060583085215e-06,
      "loss": 0.5088,
      "step": 192200
    },
    {
      "epoch": 78.68,
      "grad_norm": 1.9049803018569946,
      "learning_rate": 3.3319167442560775e-06,
      "loss": 0.5012,
      "step": 192210
    },
    {
      "epoch": 78.68,
      "grad_norm": 2.062594413757324,
      "learning_rate": 3.3317729009003003e-06,
      "loss": 0.4982,
      "step": 192220
    },
    {
      "epoch": 78.69,
      "grad_norm": 2.214603900909424,
      "learning_rate": 3.3316290530185443e-06,
      "loss": 0.5046,
      "step": 192230
    },
    {
      "epoch": 78.69,
      "grad_norm": 2.197824001312256,
      "learning_rate": 3.331485200611466e-06,
      "loss": 0.4967,
      "step": 192240
    },
    {
      "epoch": 78.69,
      "grad_norm": 1.7614223957061768,
      "learning_rate": 3.3313413436797265e-06,
      "loss": 0.5082,
      "step": 192250
    },
    {
      "epoch": 78.7,
      "grad_norm": 1.6529628038406372,
      "learning_rate": 3.331197482223984e-06,
      "loss": 0.498,
      "step": 192260
    },
    {
      "epoch": 78.7,
      "grad_norm": 1.7491099834442139,
      "learning_rate": 3.331053616244897e-06,
      "loss": 0.5057,
      "step": 192270
    },
    {
      "epoch": 78.71,
      "grad_norm": 1.573180079460144,
      "learning_rate": 3.330909745743126e-06,
      "loss": 0.5061,
      "step": 192280
    },
    {
      "epoch": 78.71,
      "grad_norm": 1.5052827596664429,
      "learning_rate": 3.3307658707193283e-06,
      "loss": 0.4944,
      "step": 192290
    },
    {
      "epoch": 78.71,
      "grad_norm": 1.5698143243789673,
      "learning_rate": 3.3306219911741637e-06,
      "loss": 0.5073,
      "step": 192300
    },
    {
      "epoch": 78.72,
      "grad_norm": 1.830125331878662,
      "learning_rate": 3.3304781071082915e-06,
      "loss": 0.5023,
      "step": 192310
    },
    {
      "epoch": 78.72,
      "grad_norm": 1.5867253541946411,
      "learning_rate": 3.330334218522371e-06,
      "loss": 0.4938,
      "step": 192320
    },
    {
      "epoch": 78.73,
      "grad_norm": 1.7881371974945068,
      "learning_rate": 3.3301903254170607e-06,
      "loss": 0.5186,
      "step": 192330
    },
    {
      "epoch": 78.73,
      "grad_norm": 1.645617961883545,
      "learning_rate": 3.3300464277930205e-06,
      "loss": 0.5063,
      "step": 192340
    },
    {
      "epoch": 78.74,
      "grad_norm": 1.550386667251587,
      "learning_rate": 3.3299025256509082e-06,
      "loss": 0.4939,
      "step": 192350
    },
    {
      "epoch": 78.74,
      "grad_norm": 2.1755356788635254,
      "learning_rate": 3.3297586189913846e-06,
      "loss": 0.4827,
      "step": 192360
    },
    {
      "epoch": 78.74,
      "grad_norm": 1.8273392915725708,
      "learning_rate": 3.3296147078151082e-06,
      "loss": 0.4931,
      "step": 192370
    },
    {
      "epoch": 78.75,
      "grad_norm": 1.8535614013671875,
      "learning_rate": 3.3294707921227376e-06,
      "loss": 0.4888,
      "step": 192380
    },
    {
      "epoch": 78.75,
      "grad_norm": 1.7839401960372925,
      "learning_rate": 3.329326871914933e-06,
      "loss": 0.505,
      "step": 192390
    },
    {
      "epoch": 78.76,
      "grad_norm": 1.9305511713027954,
      "learning_rate": 3.329182947192354e-06,
      "loss": 0.5298,
      "step": 192400
    },
    {
      "epoch": 78.76,
      "grad_norm": 2.4651687145233154,
      "learning_rate": 3.3290390179556588e-06,
      "loss": 0.5042,
      "step": 192410
    },
    {
      "epoch": 78.76,
      "grad_norm": 2.0826711654663086,
      "learning_rate": 3.3288950842055075e-06,
      "loss": 0.5048,
      "step": 192420
    },
    {
      "epoch": 78.77,
      "grad_norm": 2.138942003250122,
      "learning_rate": 3.3287511459425587e-06,
      "loss": 0.5256,
      "step": 192430
    },
    {
      "epoch": 78.77,
      "grad_norm": 1.9892807006835938,
      "learning_rate": 3.328607203167472e-06,
      "loss": 0.5298,
      "step": 192440
    },
    {
      "epoch": 78.78,
      "grad_norm": 2.016443967819214,
      "learning_rate": 3.3284632558809077e-06,
      "loss": 0.4898,
      "step": 192450
    },
    {
      "epoch": 78.78,
      "grad_norm": 2.711902379989624,
      "learning_rate": 3.328319304083524e-06,
      "loss": 0.4848,
      "step": 192460
    },
    {
      "epoch": 78.78,
      "grad_norm": 2.4534494876861572,
      "learning_rate": 3.328175347775981e-06,
      "loss": 0.4949,
      "step": 192470
    },
    {
      "epoch": 78.79,
      "grad_norm": 1.653733253479004,
      "learning_rate": 3.3280313869589377e-06,
      "loss": 0.5018,
      "step": 192480
    },
    {
      "epoch": 78.79,
      "grad_norm": 2.299130439758301,
      "learning_rate": 3.3278874216330537e-06,
      "loss": 0.5018,
      "step": 192490
    },
    {
      "epoch": 78.8,
      "grad_norm": 2.2743239402770996,
      "learning_rate": 3.327743451798989e-06,
      "loss": 0.5258,
      "step": 192500
    },
    {
      "epoch": 78.8,
      "grad_norm": 2.126953601837158,
      "learning_rate": 3.3275994774574035e-06,
      "loss": 0.5021,
      "step": 192510
    },
    {
      "epoch": 78.8,
      "grad_norm": 1.8417770862579346,
      "learning_rate": 3.3274554986089545e-06,
      "loss": 0.4953,
      "step": 192520
    },
    {
      "epoch": 78.81,
      "grad_norm": 1.866652250289917,
      "learning_rate": 3.327311515254303e-06,
      "loss": 0.5232,
      "step": 192530
    },
    {
      "epoch": 78.81,
      "grad_norm": 2.842639684677124,
      "learning_rate": 3.3271675273941093e-06,
      "loss": 0.5143,
      "step": 192540
    },
    {
      "epoch": 78.82,
      "grad_norm": 1.660573124885559,
      "learning_rate": 3.3270235350290312e-06,
      "loss": 0.5143,
      "step": 192550
    },
    {
      "epoch": 78.82,
      "grad_norm": 1.5374782085418701,
      "learning_rate": 3.32687953815973e-06,
      "loss": 0.4877,
      "step": 192560
    },
    {
      "epoch": 78.83,
      "grad_norm": 1.9739419221878052,
      "learning_rate": 3.3267355367868657e-06,
      "loss": 0.5117,
      "step": 192570
    },
    {
      "epoch": 78.83,
      "grad_norm": 1.8303065299987793,
      "learning_rate": 3.3265915309110958e-06,
      "loss": 0.5085,
      "step": 192580
    },
    {
      "epoch": 78.83,
      "grad_norm": 1.8362339735031128,
      "learning_rate": 3.3264475205330807e-06,
      "loss": 0.4865,
      "step": 192590
    },
    {
      "epoch": 78.84,
      "grad_norm": 1.8224995136260986,
      "learning_rate": 3.3263035056534806e-06,
      "loss": 0.4951,
      "step": 192600
    },
    {
      "epoch": 78.84,
      "grad_norm": 1.8267689943313599,
      "learning_rate": 3.3261594862729553e-06,
      "loss": 0.5169,
      "step": 192610
    },
    {
      "epoch": 78.85,
      "grad_norm": 1.867283821105957,
      "learning_rate": 3.3260154623921644e-06,
      "loss": 0.5043,
      "step": 192620
    },
    {
      "epoch": 78.85,
      "grad_norm": 2.005582571029663,
      "learning_rate": 3.3258714340117667e-06,
      "loss": 0.5029,
      "step": 192630
    },
    {
      "epoch": 78.85,
      "grad_norm": 1.908034086227417,
      "learning_rate": 3.325727401132424e-06,
      "loss": 0.5198,
      "step": 192640
    },
    {
      "epoch": 78.86,
      "grad_norm": 1.7409318685531616,
      "learning_rate": 3.3255833637547947e-06,
      "loss": 0.5036,
      "step": 192650
    },
    {
      "epoch": 78.86,
      "grad_norm": 1.5836665630340576,
      "learning_rate": 3.3254393218795384e-06,
      "loss": 0.5199,
      "step": 192660
    },
    {
      "epoch": 78.87,
      "grad_norm": 2.003966808319092,
      "learning_rate": 3.3252952755073156e-06,
      "loss": 0.5016,
      "step": 192670
    },
    {
      "epoch": 78.87,
      "grad_norm": 2.141510009765625,
      "learning_rate": 3.3251512246387853e-06,
      "loss": 0.4751,
      "step": 192680
    },
    {
      "epoch": 78.87,
      "grad_norm": 2.331296682357788,
      "learning_rate": 3.325007169274609e-06,
      "loss": 0.4872,
      "step": 192690
    },
    {
      "epoch": 78.88,
      "grad_norm": 1.6923267841339111,
      "learning_rate": 3.3248631094154447e-06,
      "loss": 0.4902,
      "step": 192700
    },
    {
      "epoch": 78.88,
      "grad_norm": 1.8027589321136475,
      "learning_rate": 3.3247190450619533e-06,
      "loss": 0.4674,
      "step": 192710
    },
    {
      "epoch": 78.89,
      "grad_norm": 1.4569486379623413,
      "learning_rate": 3.3245749762147946e-06,
      "loss": 0.5035,
      "step": 192720
    },
    {
      "epoch": 78.89,
      "grad_norm": 1.526768684387207,
      "learning_rate": 3.3244309028746292e-06,
      "loss": 0.4922,
      "step": 192730
    },
    {
      "epoch": 78.89,
      "grad_norm": 1.9440511465072632,
      "learning_rate": 3.324286825042116e-06,
      "loss": 0.5251,
      "step": 192740
    },
    {
      "epoch": 78.9,
      "grad_norm": 1.8013759851455688,
      "learning_rate": 3.324142742717916e-06,
      "loss": 0.4945,
      "step": 192750
    },
    {
      "epoch": 78.9,
      "grad_norm": 1.5115821361541748,
      "learning_rate": 3.323998655902688e-06,
      "loss": 0.5108,
      "step": 192760
    },
    {
      "epoch": 78.91,
      "grad_norm": 1.889654278755188,
      "learning_rate": 3.3238545645970926e-06,
      "loss": 0.496,
      "step": 192770
    },
    {
      "epoch": 78.91,
      "grad_norm": 1.9773041009902954,
      "learning_rate": 3.3237104688017903e-06,
      "loss": 0.5234,
      "step": 192780
    },
    {
      "epoch": 78.92,
      "grad_norm": 1.8703540563583374,
      "learning_rate": 3.3235663685174412e-06,
      "loss": 0.4915,
      "step": 192790
    },
    {
      "epoch": 78.92,
      "grad_norm": 1.6659812927246094,
      "learning_rate": 3.3234222637447052e-06,
      "loss": 0.51,
      "step": 192800
    },
    {
      "epoch": 78.92,
      "grad_norm": 1.5343496799468994,
      "learning_rate": 3.3232781544842417e-06,
      "loss": 0.4891,
      "step": 192810
    },
    {
      "epoch": 78.93,
      "grad_norm": 1.5248454809188843,
      "learning_rate": 3.323134040736711e-06,
      "loss": 0.5118,
      "step": 192820
    },
    {
      "epoch": 78.93,
      "grad_norm": 1.865267038345337,
      "learning_rate": 3.322989922502774e-06,
      "loss": 0.5162,
      "step": 192830
    },
    {
      "epoch": 78.94,
      "grad_norm": 2.5640196800231934,
      "learning_rate": 3.322845799783091e-06,
      "loss": 0.5306,
      "step": 192840
    },
    {
      "epoch": 78.94,
      "grad_norm": 1.71611750125885,
      "learning_rate": 3.3227016725783216e-06,
      "loss": 0.5109,
      "step": 192850
    },
    {
      "epoch": 78.94,
      "grad_norm": 1.957698106765747,
      "learning_rate": 3.322557540889126e-06,
      "loss": 0.4832,
      "step": 192860
    },
    {
      "epoch": 78.95,
      "grad_norm": 1.8439042568206787,
      "learning_rate": 3.3224134047161656e-06,
      "loss": 0.4961,
      "step": 192870
    },
    {
      "epoch": 78.95,
      "grad_norm": 1.6986421346664429,
      "learning_rate": 3.322269264060099e-06,
      "loss": 0.5191,
      "step": 192880
    },
    {
      "epoch": 78.96,
      "grad_norm": 1.8460867404937744,
      "learning_rate": 3.322125118921587e-06,
      "loss": 0.5209,
      "step": 192890
    },
    {
      "epoch": 78.96,
      "grad_norm": 1.9325575828552246,
      "learning_rate": 3.32198096930129e-06,
      "loss": 0.4929,
      "step": 192900
    },
    {
      "epoch": 78.96,
      "grad_norm": 1.8061312437057495,
      "learning_rate": 3.321836815199868e-06,
      "loss": 0.5167,
      "step": 192910
    },
    {
      "epoch": 78.97,
      "grad_norm": 1.9985438585281372,
      "learning_rate": 3.3216926566179825e-06,
      "loss": 0.5191,
      "step": 192920
    },
    {
      "epoch": 78.97,
      "grad_norm": 1.7089507579803467,
      "learning_rate": 3.3215484935562935e-06,
      "loss": 0.5125,
      "step": 192930
    },
    {
      "epoch": 78.98,
      "grad_norm": 2.0952579975128174,
      "learning_rate": 3.3214043260154608e-06,
      "loss": 0.5062,
      "step": 192940
    },
    {
      "epoch": 78.98,
      "grad_norm": 2.4276278018951416,
      "learning_rate": 3.3212601539961447e-06,
      "loss": 0.5128,
      "step": 192950
    },
    {
      "epoch": 78.98,
      "grad_norm": 1.7741620540618896,
      "learning_rate": 3.3211159774990063e-06,
      "loss": 0.5219,
      "step": 192960
    },
    {
      "epoch": 78.99,
      "grad_norm": 1.6235533952713013,
      "learning_rate": 3.3209717965247054e-06,
      "loss": 0.5146,
      "step": 192970
    },
    {
      "epoch": 78.99,
      "grad_norm": 1.8102755546569824,
      "learning_rate": 3.320827611073903e-06,
      "loss": 0.5023,
      "step": 192980
    },
    {
      "epoch": 79.0,
      "grad_norm": 2.2445757389068604,
      "learning_rate": 3.3206834211472594e-06,
      "loss": 0.4803,
      "step": 192990
    },
    {
      "epoch": 79.0,
      "eval_loss": 0.5052504539489746,
      "eval_runtime": 76.9273,
      "eval_samples_per_second": 44.835,
      "eval_steps_per_second": 5.616,
      "step": 192997
    },
    {
      "epoch": 79.0,
      "grad_norm": 2.1665945053100586,
      "learning_rate": 3.3205392267454347e-06,
      "loss": 0.5075,
      "step": 193000
    },
    {
      "epoch": 79.01,
      "grad_norm": 1.7941800355911255,
      "learning_rate": 3.32039502786909e-06,
      "loss": 0.4928,
      "step": 193010
    },
    {
      "epoch": 79.01,
      "grad_norm": 2.1832635402679443,
      "learning_rate": 3.320250824518886e-06,
      "loss": 0.4933,
      "step": 193020
    },
    {
      "epoch": 79.01,
      "grad_norm": 1.8860751390457153,
      "learning_rate": 3.3201066166954827e-06,
      "loss": 0.5102,
      "step": 193030
    },
    {
      "epoch": 79.02,
      "grad_norm": 2.231323480606079,
      "learning_rate": 3.319962404399541e-06,
      "loss": 0.506,
      "step": 193040
    },
    {
      "epoch": 79.02,
      "grad_norm": 1.8336637020111084,
      "learning_rate": 3.3198181876317214e-06,
      "loss": 0.5032,
      "step": 193050
    },
    {
      "epoch": 79.03,
      "grad_norm": 1.841269612312317,
      "learning_rate": 3.3196739663926844e-06,
      "loss": 0.4914,
      "step": 193060
    },
    {
      "epoch": 79.03,
      "grad_norm": 2.1621460914611816,
      "learning_rate": 3.3195297406830915e-06,
      "loss": 0.5198,
      "step": 193070
    },
    {
      "epoch": 79.03,
      "grad_norm": 2.0630154609680176,
      "learning_rate": 3.3193855105036023e-06,
      "loss": 0.5185,
      "step": 193080
    },
    {
      "epoch": 79.04,
      "grad_norm": 2.1289925575256348,
      "learning_rate": 3.3192412758548784e-06,
      "loss": 0.4824,
      "step": 193090
    },
    {
      "epoch": 79.04,
      "grad_norm": 1.6376338005065918,
      "learning_rate": 3.3190970367375793e-06,
      "loss": 0.5155,
      "step": 193100
    },
    {
      "epoch": 79.05,
      "grad_norm": 2.522613525390625,
      "learning_rate": 3.3189527931523665e-06,
      "loss": 0.5122,
      "step": 193110
    },
    {
      "epoch": 79.05,
      "grad_norm": 1.775084376335144,
      "learning_rate": 3.318808545099901e-06,
      "loss": 0.4993,
      "step": 193120
    },
    {
      "epoch": 79.05,
      "grad_norm": 1.940629005432129,
      "learning_rate": 3.3186642925808433e-06,
      "loss": 0.4797,
      "step": 193130
    },
    {
      "epoch": 79.06,
      "grad_norm": 1.491414189338684,
      "learning_rate": 3.3185200355958542e-06,
      "loss": 0.4961,
      "step": 193140
    },
    {
      "epoch": 79.06,
      "grad_norm": 1.7861014604568481,
      "learning_rate": 3.3183757741455947e-06,
      "loss": 0.492,
      "step": 193150
    },
    {
      "epoch": 79.07,
      "grad_norm": 1.632126808166504,
      "learning_rate": 3.3182315082307256e-06,
      "loss": 0.4952,
      "step": 193160
    },
    {
      "epoch": 79.07,
      "grad_norm": 2.5091001987457275,
      "learning_rate": 3.3180872378519074e-06,
      "loss": 0.4976,
      "step": 193170
    },
    {
      "epoch": 79.07,
      "grad_norm": 1.9696158170700073,
      "learning_rate": 3.317942963009801e-06,
      "loss": 0.5021,
      "step": 193180
    },
    {
      "epoch": 79.08,
      "grad_norm": 1.7334703207015991,
      "learning_rate": 3.3177986837050675e-06,
      "loss": 0.496,
      "step": 193190
    },
    {
      "epoch": 79.08,
      "grad_norm": 1.5389268398284912,
      "learning_rate": 3.3176543999383675e-06,
      "loss": 0.5133,
      "step": 193200
    },
    {
      "epoch": 79.09,
      "grad_norm": 1.7856533527374268,
      "learning_rate": 3.317510111710363e-06,
      "loss": 0.4911,
      "step": 193210
    },
    {
      "epoch": 79.09,
      "grad_norm": 1.340343713760376,
      "learning_rate": 3.3173658190217145e-06,
      "loss": 0.5068,
      "step": 193220
    },
    {
      "epoch": 79.1,
      "grad_norm": 1.5563746690750122,
      "learning_rate": 3.317221521873082e-06,
      "loss": 0.5055,
      "step": 193230
    },
    {
      "epoch": 79.1,
      "grad_norm": 1.7536706924438477,
      "learning_rate": 3.3170772202651276e-06,
      "loss": 0.519,
      "step": 193240
    },
    {
      "epoch": 79.1,
      "grad_norm": 1.9679588079452515,
      "learning_rate": 3.316932914198512e-06,
      "loss": 0.5232,
      "step": 193250
    },
    {
      "epoch": 79.11,
      "grad_norm": 2.247570514678955,
      "learning_rate": 3.316788603673896e-06,
      "loss": 0.495,
      "step": 193260
    },
    {
      "epoch": 79.11,
      "grad_norm": 1.8672958612442017,
      "learning_rate": 3.3166442886919407e-06,
      "loss": 0.5269,
      "step": 193270
    },
    {
      "epoch": 79.12,
      "grad_norm": 1.5053178071975708,
      "learning_rate": 3.316499969253308e-06,
      "loss": 0.5062,
      "step": 193280
    },
    {
      "epoch": 79.12,
      "grad_norm": 1.8276222944259644,
      "learning_rate": 3.3163556453586575e-06,
      "loss": 0.5363,
      "step": 193290
    },
    {
      "epoch": 79.12,
      "grad_norm": 2.1893465518951416,
      "learning_rate": 3.316211317008651e-06,
      "loss": 0.5011,
      "step": 193300
    },
    {
      "epoch": 79.13,
      "grad_norm": 1.8522768020629883,
      "learning_rate": 3.3160669842039507e-06,
      "loss": 0.5053,
      "step": 193310
    },
    {
      "epoch": 79.13,
      "grad_norm": 1.7878351211547852,
      "learning_rate": 3.3159226469452164e-06,
      "loss": 0.5005,
      "step": 193320
    },
    {
      "epoch": 79.14,
      "grad_norm": 1.537624716758728,
      "learning_rate": 3.3157783052331095e-06,
      "loss": 0.5182,
      "step": 193330
    },
    {
      "epoch": 79.14,
      "grad_norm": 2.482602596282959,
      "learning_rate": 3.3156339590682917e-06,
      "loss": 0.5015,
      "step": 193340
    },
    {
      "epoch": 79.14,
      "grad_norm": 1.7646743059158325,
      "learning_rate": 3.3154896084514244e-06,
      "loss": 0.5121,
      "step": 193350
    },
    {
      "epoch": 79.15,
      "grad_norm": 1.7697077989578247,
      "learning_rate": 3.315345253383168e-06,
      "loss": 0.5036,
      "step": 193360
    },
    {
      "epoch": 79.15,
      "grad_norm": 1.6303675174713135,
      "learning_rate": 3.3152008938641847e-06,
      "loss": 0.5129,
      "step": 193370
    },
    {
      "epoch": 79.16,
      "grad_norm": 1.8961193561553955,
      "learning_rate": 3.3150565298951347e-06,
      "loss": 0.4933,
      "step": 193380
    },
    {
      "epoch": 79.16,
      "grad_norm": 1.8584285974502563,
      "learning_rate": 3.3149121614766803e-06,
      "loss": 0.5094,
      "step": 193390
    },
    {
      "epoch": 79.16,
      "grad_norm": 1.6689448356628418,
      "learning_rate": 3.314767788609482e-06,
      "loss": 0.5078,
      "step": 193400
    },
    {
      "epoch": 79.17,
      "grad_norm": 1.752148151397705,
      "learning_rate": 3.314623411294202e-06,
      "loss": 0.5102,
      "step": 193410
    },
    {
      "epoch": 79.17,
      "grad_norm": 1.9301596879959106,
      "learning_rate": 3.314479029531501e-06,
      "loss": 0.5115,
      "step": 193420
    },
    {
      "epoch": 79.18,
      "grad_norm": 1.6845968961715698,
      "learning_rate": 3.3143346433220413e-06,
      "loss": 0.5101,
      "step": 193430
    },
    {
      "epoch": 79.18,
      "grad_norm": 1.9974066019058228,
      "learning_rate": 3.314190252666483e-06,
      "loss": 0.5096,
      "step": 193440
    },
    {
      "epoch": 79.19,
      "grad_norm": 2.446307897567749,
      "learning_rate": 3.3140458575654887e-06,
      "loss": 0.5022,
      "step": 193450
    },
    {
      "epoch": 79.19,
      "grad_norm": 1.8574395179748535,
      "learning_rate": 3.3139014580197187e-06,
      "loss": 0.5094,
      "step": 193460
    },
    {
      "epoch": 79.19,
      "grad_norm": 2.025820255279541,
      "learning_rate": 3.3137570540298357e-06,
      "loss": 0.5104,
      "step": 193470
    },
    {
      "epoch": 79.2,
      "grad_norm": 1.5987498760223389,
      "learning_rate": 3.3136126455965e-06,
      "loss": 0.5006,
      "step": 193480
    },
    {
      "epoch": 79.2,
      "grad_norm": 1.7043066024780273,
      "learning_rate": 3.313468232720374e-06,
      "loss": 0.4972,
      "step": 193490
    },
    {
      "epoch": 79.21,
      "grad_norm": 4.45361852645874,
      "learning_rate": 3.313323815402119e-06,
      "loss": 0.4813,
      "step": 193500
    },
    {
      "epoch": 79.21,
      "grad_norm": 2.1232662200927734,
      "learning_rate": 3.3131793936423966e-06,
      "loss": 0.4832,
      "step": 193510
    },
    {
      "epoch": 79.21,
      "grad_norm": 1.3933993577957153,
      "learning_rate": 3.3130349674418685e-06,
      "loss": 0.5074,
      "step": 193520
    },
    {
      "epoch": 79.22,
      "grad_norm": 1.5400142669677734,
      "learning_rate": 3.312890536801196e-06,
      "loss": 0.4736,
      "step": 193530
    },
    {
      "epoch": 79.22,
      "grad_norm": 1.6933202743530273,
      "learning_rate": 3.31274610172104e-06,
      "loss": 0.502,
      "step": 193540
    },
    {
      "epoch": 79.23,
      "grad_norm": 2.009413242340088,
      "learning_rate": 3.312601662202064e-06,
      "loss": 0.5215,
      "step": 193550
    },
    {
      "epoch": 79.23,
      "grad_norm": 1.876840591430664,
      "learning_rate": 3.3124572182449277e-06,
      "loss": 0.5103,
      "step": 193560
    },
    {
      "epoch": 79.23,
      "grad_norm": 2.165508508682251,
      "learning_rate": 3.3123127698502937e-06,
      "loss": 0.5057,
      "step": 193570
    },
    {
      "epoch": 79.24,
      "grad_norm": 2.245621919631958,
      "learning_rate": 3.312168317018825e-06,
      "loss": 0.5167,
      "step": 193580
    },
    {
      "epoch": 79.24,
      "grad_norm": 1.9651635885238647,
      "learning_rate": 3.3120238597511806e-06,
      "loss": 0.506,
      "step": 193590
    },
    {
      "epoch": 79.25,
      "grad_norm": 1.7434810400009155,
      "learning_rate": 3.311879398048024e-06,
      "loss": 0.5042,
      "step": 193600
    },
    {
      "epoch": 79.25,
      "grad_norm": 2.2568047046661377,
      "learning_rate": 3.3117349319100164e-06,
      "loss": 0.4984,
      "step": 193610
    },
    {
      "epoch": 79.26,
      "grad_norm": 1.925199031829834,
      "learning_rate": 3.31159046133782e-06,
      "loss": 0.4998,
      "step": 193620
    },
    {
      "epoch": 79.26,
      "grad_norm": 1.9537428617477417,
      "learning_rate": 3.311445986332096e-06,
      "loss": 0.5193,
      "step": 193630
    },
    {
      "epoch": 79.26,
      "grad_norm": 1.6966111660003662,
      "learning_rate": 3.311301506893507e-06,
      "loss": 0.5211,
      "step": 193640
    },
    {
      "epoch": 79.27,
      "grad_norm": 1.8581089973449707,
      "learning_rate": 3.311157023022714e-06,
      "loss": 0.4833,
      "step": 193650
    },
    {
      "epoch": 79.27,
      "grad_norm": 1.8184475898742676,
      "learning_rate": 3.311012534720379e-06,
      "loss": 0.5038,
      "step": 193660
    },
    {
      "epoch": 79.28,
      "grad_norm": 1.4938321113586426,
      "learning_rate": 3.3108680419871647e-06,
      "loss": 0.5068,
      "step": 193670
    },
    {
      "epoch": 79.28,
      "grad_norm": 1.9586142301559448,
      "learning_rate": 3.3107235448237325e-06,
      "loss": 0.511,
      "step": 193680
    },
    {
      "epoch": 79.28,
      "grad_norm": 1.849596619606018,
      "learning_rate": 3.3105790432307436e-06,
      "loss": 0.5097,
      "step": 193690
    },
    {
      "epoch": 79.29,
      "grad_norm": 2.130476474761963,
      "learning_rate": 3.3104345372088613e-06,
      "loss": 0.5133,
      "step": 193700
    },
    {
      "epoch": 79.29,
      "grad_norm": 2.1926565170288086,
      "learning_rate": 3.3102900267587466e-06,
      "loss": 0.4996,
      "step": 193710
    },
    {
      "epoch": 79.3,
      "grad_norm": 2.1859357357025146,
      "learning_rate": 3.3101455118810615e-06,
      "loss": 0.4909,
      "step": 193720
    },
    {
      "epoch": 79.3,
      "grad_norm": 1.7996389865875244,
      "learning_rate": 3.3100009925764685e-06,
      "loss": 0.5077,
      "step": 193730
    },
    {
      "epoch": 79.3,
      "grad_norm": 1.9768515825271606,
      "learning_rate": 3.3098564688456293e-06,
      "loss": 0.5076,
      "step": 193740
    },
    {
      "epoch": 79.31,
      "grad_norm": 2.0886247158050537,
      "learning_rate": 3.3097119406892065e-06,
      "loss": 0.4942,
      "step": 193750
    },
    {
      "epoch": 79.31,
      "grad_norm": 1.746375560760498,
      "learning_rate": 3.3095674081078603e-06,
      "loss": 0.51,
      "step": 193760
    },
    {
      "epoch": 79.32,
      "grad_norm": 1.8551338911056519,
      "learning_rate": 3.3094228711022548e-06,
      "loss": 0.4988,
      "step": 193770
    },
    {
      "epoch": 79.32,
      "grad_norm": 2.1038899421691895,
      "learning_rate": 3.3092783296730514e-06,
      "loss": 0.4896,
      "step": 193780
    },
    {
      "epoch": 79.32,
      "grad_norm": 2.062802314758301,
      "learning_rate": 3.309133783820912e-06,
      "loss": 0.5047,
      "step": 193790
    },
    {
      "epoch": 79.33,
      "grad_norm": 2.2396533489227295,
      "learning_rate": 3.3089892335464995e-06,
      "loss": 0.49,
      "step": 193800
    },
    {
      "epoch": 79.33,
      "grad_norm": 1.6068254709243774,
      "learning_rate": 3.3088446788504766e-06,
      "loss": 0.4983,
      "step": 193810
    },
    {
      "epoch": 79.34,
      "grad_norm": 1.8391549587249756,
      "learning_rate": 3.3087001197335033e-06,
      "loss": 0.5151,
      "step": 193820
    },
    {
      "epoch": 79.34,
      "grad_norm": 1.8604799509048462,
      "learning_rate": 3.3085555561962424e-06,
      "loss": 0.4956,
      "step": 193830
    },
    {
      "epoch": 79.35,
      "grad_norm": 2.2035133838653564,
      "learning_rate": 3.3084109882393575e-06,
      "loss": 0.5139,
      "step": 193840
    },
    {
      "epoch": 79.35,
      "grad_norm": 2.024966239929199,
      "learning_rate": 3.308266415863509e-06,
      "loss": 0.5058,
      "step": 193850
    },
    {
      "epoch": 79.35,
      "grad_norm": 2.1852550506591797,
      "learning_rate": 3.3081218390693604e-06,
      "loss": 0.5063,
      "step": 193860
    },
    {
      "epoch": 79.36,
      "grad_norm": 1.4412254095077515,
      "learning_rate": 3.307977257857574e-06,
      "loss": 0.4952,
      "step": 193870
    },
    {
      "epoch": 79.36,
      "grad_norm": 1.6274000406265259,
      "learning_rate": 3.307832672228813e-06,
      "loss": 0.5274,
      "step": 193880
    },
    {
      "epoch": 79.37,
      "grad_norm": 1.8330038785934448,
      "learning_rate": 3.3076880821837374e-06,
      "loss": 0.5033,
      "step": 193890
    },
    {
      "epoch": 79.37,
      "grad_norm": 1.9848606586456299,
      "learning_rate": 3.3075434877230116e-06,
      "loss": 0.5111,
      "step": 193900
    },
    {
      "epoch": 79.37,
      "grad_norm": 2.3619680404663086,
      "learning_rate": 3.307398888847296e-06,
      "loss": 0.4863,
      "step": 193910
    },
    {
      "epoch": 79.38,
      "grad_norm": 1.4361591339111328,
      "learning_rate": 3.307254285557255e-06,
      "loss": 0.522,
      "step": 193920
    },
    {
      "epoch": 79.38,
      "grad_norm": 1.6463123559951782,
      "learning_rate": 3.3071096778535495e-06,
      "loss": 0.4885,
      "step": 193930
    },
    {
      "epoch": 79.39,
      "grad_norm": 1.9030754566192627,
      "learning_rate": 3.3069650657368425e-06,
      "loss": 0.4984,
      "step": 193940
    },
    {
      "epoch": 79.39,
      "grad_norm": 1.8643614053726196,
      "learning_rate": 3.3068204492077965e-06,
      "loss": 0.5081,
      "step": 193950
    },
    {
      "epoch": 79.39,
      "grad_norm": 1.6210819482803345,
      "learning_rate": 3.306675828267074e-06,
      "loss": 0.5126,
      "step": 193960
    },
    {
      "epoch": 79.4,
      "grad_norm": 2.7432820796966553,
      "learning_rate": 3.3065312029153376e-06,
      "loss": 0.5107,
      "step": 193970
    },
    {
      "epoch": 79.4,
      "grad_norm": 2.145023822784424,
      "learning_rate": 3.3063865731532496e-06,
      "loss": 0.5006,
      "step": 193980
    },
    {
      "epoch": 79.41,
      "grad_norm": 2.502702474594116,
      "learning_rate": 3.306241938981472e-06,
      "loss": 0.4875,
      "step": 193990
    },
    {
      "epoch": 79.41,
      "grad_norm": 2.040761709213257,
      "learning_rate": 3.3060973004006685e-06,
      "loss": 0.5019,
      "step": 194000
    },
    {
      "epoch": 79.41,
      "grad_norm": 2.2151620388031006,
      "learning_rate": 3.3059526574115008e-06,
      "loss": 0.4973,
      "step": 194010
    },
    {
      "epoch": 79.42,
      "grad_norm": 1.8574347496032715,
      "learning_rate": 3.3058080100146323e-06,
      "loss": 0.4958,
      "step": 194020
    },
    {
      "epoch": 79.42,
      "grad_norm": 1.9144529104232788,
      "learning_rate": 3.3056633582107242e-06,
      "loss": 0.5031,
      "step": 194030
    },
    {
      "epoch": 79.43,
      "grad_norm": 1.8870046138763428,
      "learning_rate": 3.3055187020004403e-06,
      "loss": 0.4747,
      "step": 194040
    },
    {
      "epoch": 79.43,
      "grad_norm": 1.9146560430526733,
      "learning_rate": 3.3053740413844427e-06,
      "loss": 0.5124,
      "step": 194050
    },
    {
      "epoch": 79.44,
      "grad_norm": 1.8228845596313477,
      "learning_rate": 3.305229376363395e-06,
      "loss": 0.5008,
      "step": 194060
    },
    {
      "epoch": 79.44,
      "grad_norm": 2.974900245666504,
      "learning_rate": 3.3050847069379587e-06,
      "loss": 0.5087,
      "step": 194070
    },
    {
      "epoch": 79.44,
      "grad_norm": 2.6887564659118652,
      "learning_rate": 3.304940033108797e-06,
      "loss": 0.4851,
      "step": 194080
    },
    {
      "epoch": 79.45,
      "grad_norm": 1.6367703676223755,
      "learning_rate": 3.3047953548765728e-06,
      "loss": 0.4941,
      "step": 194090
    },
    {
      "epoch": 79.45,
      "grad_norm": 2.3234333992004395,
      "learning_rate": 3.3046506722419487e-06,
      "loss": 0.506,
      "step": 194100
    },
    {
      "epoch": 79.46,
      "grad_norm": 1.87421452999115,
      "learning_rate": 3.304505985205587e-06,
      "loss": 0.4962,
      "step": 194110
    },
    {
      "epoch": 79.46,
      "grad_norm": 1.9526982307434082,
      "learning_rate": 3.304361293768151e-06,
      "loss": 0.4845,
      "step": 194120
    },
    {
      "epoch": 79.46,
      "grad_norm": 1.8093366622924805,
      "learning_rate": 3.3042165979303035e-06,
      "loss": 0.51,
      "step": 194130
    },
    {
      "epoch": 79.47,
      "grad_norm": 1.4843982458114624,
      "learning_rate": 3.304071897692707e-06,
      "loss": 0.5069,
      "step": 194140
    },
    {
      "epoch": 79.47,
      "grad_norm": 1.4158215522766113,
      "learning_rate": 3.303927193056025e-06,
      "loss": 0.4793,
      "step": 194150
    },
    {
      "epoch": 79.48,
      "grad_norm": 1.8545037508010864,
      "learning_rate": 3.30378248402092e-06,
      "loss": 0.5151,
      "step": 194160
    },
    {
      "epoch": 79.48,
      "grad_norm": 2.013465404510498,
      "learning_rate": 3.303637770588055e-06,
      "loss": 0.5273,
      "step": 194170
    },
    {
      "epoch": 79.48,
      "grad_norm": 1.7856175899505615,
      "learning_rate": 3.303493052758093e-06,
      "loss": 0.5022,
      "step": 194180
    },
    {
      "epoch": 79.49,
      "grad_norm": 2.0769639015197754,
      "learning_rate": 3.303348330531696e-06,
      "loss": 0.4966,
      "step": 194190
    },
    {
      "epoch": 79.49,
      "grad_norm": 1.9243113994598389,
      "learning_rate": 3.303203603909528e-06,
      "loss": 0.5092,
      "step": 194200
    },
    {
      "epoch": 79.5,
      "grad_norm": 2.173245906829834,
      "learning_rate": 3.3030588728922515e-06,
      "loss": 0.5009,
      "step": 194210
    },
    {
      "epoch": 79.5,
      "grad_norm": 2.0996291637420654,
      "learning_rate": 3.30291413748053e-06,
      "loss": 0.5103,
      "step": 194220
    },
    {
      "epoch": 79.5,
      "grad_norm": 1.8742302656173706,
      "learning_rate": 3.3027693976750253e-06,
      "loss": 0.4948,
      "step": 194230
    },
    {
      "epoch": 79.51,
      "grad_norm": 1.9441466331481934,
      "learning_rate": 3.302624653476402e-06,
      "loss": 0.5227,
      "step": 194240
    },
    {
      "epoch": 79.51,
      "grad_norm": 1.640830636024475,
      "learning_rate": 3.3024799048853223e-06,
      "loss": 0.5349,
      "step": 194250
    },
    {
      "epoch": 79.52,
      "grad_norm": 1.52910315990448,
      "learning_rate": 3.302335151902449e-06,
      "loss": 0.488,
      "step": 194260
    },
    {
      "epoch": 79.52,
      "grad_norm": 2.0264408588409424,
      "learning_rate": 3.3021903945284465e-06,
      "loss": 0.5239,
      "step": 194270
    },
    {
      "epoch": 79.53,
      "grad_norm": 1.8909971714019775,
      "learning_rate": 3.302045632763976e-06,
      "loss": 0.5327,
      "step": 194280
    },
    {
      "epoch": 79.53,
      "grad_norm": 2.9175822734832764,
      "learning_rate": 3.301900866609702e-06,
      "loss": 0.5094,
      "step": 194290
    },
    {
      "epoch": 79.53,
      "grad_norm": 1.717560887336731,
      "learning_rate": 3.3017560960662874e-06,
      "loss": 0.5018,
      "step": 194300
    },
    {
      "epoch": 79.54,
      "grad_norm": 1.7431411743164062,
      "learning_rate": 3.301611321134395e-06,
      "loss": 0.4881,
      "step": 194310
    },
    {
      "epoch": 79.54,
      "grad_norm": 2.042358875274658,
      "learning_rate": 3.3014665418146887e-06,
      "loss": 0.5111,
      "step": 194320
    },
    {
      "epoch": 79.55,
      "grad_norm": 6.049269199371338,
      "learning_rate": 3.3013217581078303e-06,
      "loss": 0.5231,
      "step": 194330
    },
    {
      "epoch": 79.55,
      "grad_norm": 2.0585904121398926,
      "learning_rate": 3.3011769700144846e-06,
      "loss": 0.5206,
      "step": 194340
    },
    {
      "epoch": 79.55,
      "grad_norm": 1.7162014245986938,
      "learning_rate": 3.301032177535314e-06,
      "loss": 0.5117,
      "step": 194350
    },
    {
      "epoch": 79.56,
      "grad_norm": 1.8120536804199219,
      "learning_rate": 3.3008873806709822e-06,
      "loss": 0.5131,
      "step": 194360
    },
    {
      "epoch": 79.56,
      "grad_norm": 2.1824605464935303,
      "learning_rate": 3.3007425794221523e-06,
      "loss": 0.5232,
      "step": 194370
    },
    {
      "epoch": 79.57,
      "grad_norm": 1.858605980873108,
      "learning_rate": 3.3005977737894873e-06,
      "loss": 0.4855,
      "step": 194380
    },
    {
      "epoch": 79.57,
      "grad_norm": 1.6071549654006958,
      "learning_rate": 3.3004529637736517e-06,
      "loss": 0.5143,
      "step": 194390
    },
    {
      "epoch": 79.57,
      "grad_norm": 2.2257277965545654,
      "learning_rate": 3.3003081493753074e-06,
      "loss": 0.5131,
      "step": 194400
    },
    {
      "epoch": 79.58,
      "grad_norm": 2.869015693664551,
      "learning_rate": 3.300163330595118e-06,
      "loss": 0.4952,
      "step": 194410
    },
    {
      "epoch": 79.58,
      "grad_norm": 1.8901538848876953,
      "learning_rate": 3.3000185074337474e-06,
      "loss": 0.502,
      "step": 194420
    },
    {
      "epoch": 79.59,
      "grad_norm": 1.7876211404800415,
      "learning_rate": 3.299873679891859e-06,
      "loss": 0.5061,
      "step": 194430
    },
    {
      "epoch": 79.59,
      "grad_norm": 1.7710758447647095,
      "learning_rate": 3.2997288479701162e-06,
      "loss": 0.4952,
      "step": 194440
    },
    {
      "epoch": 79.59,
      "grad_norm": 1.677416205406189,
      "learning_rate": 3.299584011669182e-06,
      "loss": 0.4929,
      "step": 194450
    },
    {
      "epoch": 79.6,
      "grad_norm": 1.4912745952606201,
      "learning_rate": 3.2994391709897213e-06,
      "loss": 0.5279,
      "step": 194460
    },
    {
      "epoch": 79.6,
      "grad_norm": 1.6912014484405518,
      "learning_rate": 3.2992943259323956e-06,
      "loss": 0.4906,
      "step": 194470
    },
    {
      "epoch": 79.61,
      "grad_norm": 1.5252325534820557,
      "learning_rate": 3.2991494764978697e-06,
      "loss": 0.5181,
      "step": 194480
    },
    {
      "epoch": 79.61,
      "grad_norm": 1.6222352981567383,
      "learning_rate": 3.2990046226868066e-06,
      "loss": 0.4983,
      "step": 194490
    },
    {
      "epoch": 79.62,
      "grad_norm": 1.4438153505325317,
      "learning_rate": 3.2988597644998697e-06,
      "loss": 0.5115,
      "step": 194500
    },
    {
      "epoch": 79.62,
      "grad_norm": 1.7319108247756958,
      "learning_rate": 3.2987149019377222e-06,
      "loss": 0.5139,
      "step": 194510
    },
    {
      "epoch": 79.62,
      "grad_norm": 1.7398782968521118,
      "learning_rate": 3.29857003500103e-06,
      "loss": 0.5051,
      "step": 194520
    },
    {
      "epoch": 79.63,
      "grad_norm": 1.857312560081482,
      "learning_rate": 3.2984251636904546e-06,
      "loss": 0.5032,
      "step": 194530
    },
    {
      "epoch": 79.63,
      "grad_norm": 1.84479558467865,
      "learning_rate": 3.2982802880066597e-06,
      "loss": 0.5248,
      "step": 194540
    },
    {
      "epoch": 79.64,
      "grad_norm": 1.456096887588501,
      "learning_rate": 3.2981354079503096e-06,
      "loss": 0.5033,
      "step": 194550
    },
    {
      "epoch": 79.64,
      "grad_norm": 2.151388168334961,
      "learning_rate": 3.297990523522068e-06,
      "loss": 0.5062,
      "step": 194560
    },
    {
      "epoch": 79.64,
      "grad_norm": 1.5278806686401367,
      "learning_rate": 3.2978456347225976e-06,
      "loss": 0.4793,
      "step": 194570
    },
    {
      "epoch": 79.65,
      "grad_norm": 2.0729050636291504,
      "learning_rate": 3.297700741552564e-06,
      "loss": 0.5075,
      "step": 194580
    },
    {
      "epoch": 79.65,
      "grad_norm": 1.320015788078308,
      "learning_rate": 3.2975558440126283e-06,
      "loss": 0.4892,
      "step": 194590
    },
    {
      "epoch": 79.66,
      "grad_norm": 1.6459404230117798,
      "learning_rate": 3.297410942103457e-06,
      "loss": 0.4968,
      "step": 194600
    },
    {
      "epoch": 79.66,
      "grad_norm": 1.83749520778656,
      "learning_rate": 3.297266035825712e-06,
      "loss": 0.5053,
      "step": 194610
    },
    {
      "epoch": 79.66,
      "grad_norm": 1.978766918182373,
      "learning_rate": 3.297121125180058e-06,
      "loss": 0.5146,
      "step": 194620
    },
    {
      "epoch": 79.67,
      "grad_norm": 1.9995774030685425,
      "learning_rate": 3.2969762101671584e-06,
      "loss": 0.5156,
      "step": 194630
    },
    {
      "epoch": 79.67,
      "grad_norm": 1.768866777420044,
      "learning_rate": 3.2968312907876773e-06,
      "loss": 0.5009,
      "step": 194640
    },
    {
      "epoch": 79.68,
      "grad_norm": 1.520440697669983,
      "learning_rate": 3.2966863670422783e-06,
      "loss": 0.5082,
      "step": 194650
    },
    {
      "epoch": 79.68,
      "grad_norm": 1.5302114486694336,
      "learning_rate": 3.296541438931625e-06,
      "loss": 0.5074,
      "step": 194660
    },
    {
      "epoch": 79.68,
      "grad_norm": 1.5108064413070679,
      "learning_rate": 3.2963965064563823e-06,
      "loss": 0.4865,
      "step": 194670
    },
    {
      "epoch": 79.69,
      "grad_norm": 1.8146501779556274,
      "learning_rate": 3.296251569617214e-06,
      "loss": 0.514,
      "step": 194680
    },
    {
      "epoch": 79.69,
      "grad_norm": 2.147106885910034,
      "learning_rate": 3.2961066284147823e-06,
      "loss": 0.521,
      "step": 194690
    },
    {
      "epoch": 79.7,
      "grad_norm": 1.9042479991912842,
      "learning_rate": 3.2959616828497528e-06,
      "loss": 0.4928,
      "step": 194700
    },
    {
      "epoch": 79.7,
      "grad_norm": 2.0405797958374023,
      "learning_rate": 3.295816732922789e-06,
      "loss": 0.4963,
      "step": 194710
    },
    {
      "epoch": 79.71,
      "grad_norm": 1.5145894289016724,
      "learning_rate": 3.295671778634555e-06,
      "loss": 0.5136,
      "step": 194720
    },
    {
      "epoch": 79.71,
      "grad_norm": 1.884948492050171,
      "learning_rate": 3.2955268199857147e-06,
      "loss": 0.5096,
      "step": 194730
    },
    {
      "epoch": 79.71,
      "grad_norm": 1.7107070684432983,
      "learning_rate": 3.2953818569769327e-06,
      "loss": 0.5136,
      "step": 194740
    },
    {
      "epoch": 79.72,
      "grad_norm": 1.8509858846664429,
      "learning_rate": 3.2952368896088724e-06,
      "loss": 0.512,
      "step": 194750
    },
    {
      "epoch": 79.72,
      "grad_norm": 2.392852783203125,
      "learning_rate": 3.2950919178821973e-06,
      "loss": 0.5269,
      "step": 194760
    },
    {
      "epoch": 79.73,
      "grad_norm": 2.1294288635253906,
      "learning_rate": 3.2949469417975725e-06,
      "loss": 0.493,
      "step": 194770
    },
    {
      "epoch": 79.73,
      "grad_norm": 1.6963361501693726,
      "learning_rate": 3.2948019613556624e-06,
      "loss": 0.5043,
      "step": 194780
    },
    {
      "epoch": 79.73,
      "grad_norm": 1.3776488304138184,
      "learning_rate": 3.29465697655713e-06,
      "loss": 0.5033,
      "step": 194790
    },
    {
      "epoch": 79.74,
      "grad_norm": 2.7658472061157227,
      "learning_rate": 3.2945119874026396e-06,
      "loss": 0.5001,
      "step": 194800
    },
    {
      "epoch": 79.74,
      "grad_norm": 1.452430009841919,
      "learning_rate": 3.2943669938928563e-06,
      "loss": 0.5171,
      "step": 194810
    },
    {
      "epoch": 79.75,
      "grad_norm": 1.6465368270874023,
      "learning_rate": 3.294221996028444e-06,
      "loss": 0.5021,
      "step": 194820
    },
    {
      "epoch": 79.75,
      "grad_norm": 1.7203550338745117,
      "learning_rate": 3.294076993810067e-06,
      "loss": 0.4984,
      "step": 194830
    },
    {
      "epoch": 79.75,
      "grad_norm": 1.3199408054351807,
      "learning_rate": 3.2939319872383885e-06,
      "loss": 0.5227,
      "step": 194840
    },
    {
      "epoch": 79.76,
      "grad_norm": 1.8839281797409058,
      "learning_rate": 3.2937869763140734e-06,
      "loss": 0.5016,
      "step": 194850
    },
    {
      "epoch": 79.76,
      "grad_norm": 2.0306458473205566,
      "learning_rate": 3.2936419610377867e-06,
      "loss": 0.5189,
      "step": 194860
    },
    {
      "epoch": 79.77,
      "grad_norm": 1.4980319738388062,
      "learning_rate": 3.2934969414101915e-06,
      "loss": 0.507,
      "step": 194870
    },
    {
      "epoch": 79.77,
      "grad_norm": 1.7466115951538086,
      "learning_rate": 3.293351917431953e-06,
      "loss": 0.5185,
      "step": 194880
    },
    {
      "epoch": 79.77,
      "grad_norm": 1.8160804510116577,
      "learning_rate": 3.293206889103735e-06,
      "loss": 0.5025,
      "step": 194890
    },
    {
      "epoch": 79.78,
      "grad_norm": 2.1840851306915283,
      "learning_rate": 3.2930618564262016e-06,
      "loss": 0.5128,
      "step": 194900
    },
    {
      "epoch": 79.78,
      "grad_norm": 1.6933761835098267,
      "learning_rate": 3.292916819400018e-06,
      "loss": 0.4988,
      "step": 194910
    },
    {
      "epoch": 79.79,
      "grad_norm": 2.1273043155670166,
      "learning_rate": 3.292771778025848e-06,
      "loss": 0.4996,
      "step": 194920
    },
    {
      "epoch": 79.79,
      "grad_norm": 1.3014343976974487,
      "learning_rate": 3.2926267323043563e-06,
      "loss": 0.5161,
      "step": 194930
    },
    {
      "epoch": 79.8,
      "grad_norm": 2.1767101287841797,
      "learning_rate": 3.2924816822362075e-06,
      "loss": 0.4978,
      "step": 194940
    },
    {
      "epoch": 79.8,
      "grad_norm": 2.2353198528289795,
      "learning_rate": 3.292336627822066e-06,
      "loss": 0.5146,
      "step": 194950
    },
    {
      "epoch": 79.8,
      "grad_norm": 2.0319955348968506,
      "learning_rate": 3.2921915690625955e-06,
      "loss": 0.5023,
      "step": 194960
    },
    {
      "epoch": 79.81,
      "grad_norm": 1.7579745054244995,
      "learning_rate": 3.2920465059584614e-06,
      "loss": 0.5239,
      "step": 194970
    },
    {
      "epoch": 79.81,
      "grad_norm": 2.5576534271240234,
      "learning_rate": 3.291901438510328e-06,
      "loss": 0.5093,
      "step": 194980
    },
    {
      "epoch": 79.82,
      "grad_norm": 1.5480414628982544,
      "learning_rate": 3.291756366718859e-06,
      "loss": 0.5057,
      "step": 194990
    },
    {
      "epoch": 79.82,
      "grad_norm": 1.9875105619430542,
      "learning_rate": 3.2916112905847204e-06,
      "loss": 0.4938,
      "step": 195000
    },
    {
      "epoch": 79.82,
      "grad_norm": 1.609770655632019,
      "learning_rate": 3.2914662101085754e-06,
      "loss": 0.5195,
      "step": 195010
    },
    {
      "epoch": 79.83,
      "grad_norm": 2.14996337890625,
      "learning_rate": 3.2913211252910897e-06,
      "loss": 0.4997,
      "step": 195020
    },
    {
      "epoch": 79.83,
      "grad_norm": 2.288287401199341,
      "learning_rate": 3.291176036132927e-06,
      "loss": 0.5074,
      "step": 195030
    },
    {
      "epoch": 79.84,
      "grad_norm": 2.020256280899048,
      "learning_rate": 3.2910309426347535e-06,
      "loss": 0.4829,
      "step": 195040
    },
    {
      "epoch": 79.84,
      "grad_norm": 2.6523988246917725,
      "learning_rate": 3.2908858447972312e-06,
      "loss": 0.5019,
      "step": 195050
    },
    {
      "epoch": 79.84,
      "grad_norm": 1.9052574634552002,
      "learning_rate": 3.290740742621027e-06,
      "loss": 0.4822,
      "step": 195060
    },
    {
      "epoch": 79.85,
      "grad_norm": 1.5690563917160034,
      "learning_rate": 3.290595636106805e-06,
      "loss": 0.5184,
      "step": 195070
    },
    {
      "epoch": 79.85,
      "grad_norm": 1.737880825996399,
      "learning_rate": 3.290450525255229e-06,
      "loss": 0.5088,
      "step": 195080
    },
    {
      "epoch": 79.86,
      "grad_norm": 1.65813148021698,
      "learning_rate": 3.290305410066965e-06,
      "loss": 0.5139,
      "step": 195090
    },
    {
      "epoch": 79.86,
      "grad_norm": 2.294184446334839,
      "learning_rate": 3.290160290542677e-06,
      "loss": 0.5014,
      "step": 195100
    },
    {
      "epoch": 79.86,
      "grad_norm": 1.7269842624664307,
      "learning_rate": 3.290015166683031e-06,
      "loss": 0.4983,
      "step": 195110
    },
    {
      "epoch": 79.87,
      "grad_norm": 2.145214796066284,
      "learning_rate": 3.28987003848869e-06,
      "loss": 0.5213,
      "step": 195120
    },
    {
      "epoch": 79.87,
      "grad_norm": 2.0881497859954834,
      "learning_rate": 3.28972490596032e-06,
      "loss": 0.5155,
      "step": 195130
    },
    {
      "epoch": 79.88,
      "grad_norm": 1.5497174263000488,
      "learning_rate": 3.2895797690985853e-06,
      "loss": 0.5109,
      "step": 195140
    },
    {
      "epoch": 79.88,
      "grad_norm": 2.0412209033966064,
      "learning_rate": 3.2894346279041502e-06,
      "loss": 0.5018,
      "step": 195150
    },
    {
      "epoch": 79.89,
      "grad_norm": 1.7787377834320068,
      "learning_rate": 3.289289482377681e-06,
      "loss": 0.514,
      "step": 195160
    },
    {
      "epoch": 79.89,
      "grad_norm": 2.1835641860961914,
      "learning_rate": 3.2891443325198418e-06,
      "loss": 0.5,
      "step": 195170
    },
    {
      "epoch": 79.89,
      "grad_norm": 1.6946160793304443,
      "learning_rate": 3.288999178331298e-06,
      "loss": 0.4942,
      "step": 195180
    },
    {
      "epoch": 79.9,
      "grad_norm": 1.7894905805587769,
      "learning_rate": 3.2888540198127134e-06,
      "loss": 0.5036,
      "step": 195190
    },
    {
      "epoch": 79.9,
      "grad_norm": 2.204522132873535,
      "learning_rate": 3.2887088569647543e-06,
      "loss": 0.4958,
      "step": 195200
    },
    {
      "epoch": 79.91,
      "grad_norm": 1.7377727031707764,
      "learning_rate": 3.288563689788084e-06,
      "loss": 0.5366,
      "step": 195210
    },
    {
      "epoch": 79.91,
      "grad_norm": 1.5691790580749512,
      "learning_rate": 3.2884185182833696e-06,
      "loss": 0.4978,
      "step": 195220
    },
    {
      "epoch": 79.91,
      "grad_norm": 1.5298399925231934,
      "learning_rate": 3.2882733424512746e-06,
      "loss": 0.4933,
      "step": 195230
    },
    {
      "epoch": 79.92,
      "grad_norm": 2.11761474609375,
      "learning_rate": 3.288128162292464e-06,
      "loss": 0.4942,
      "step": 195240
    },
    {
      "epoch": 79.92,
      "grad_norm": 1.9528074264526367,
      "learning_rate": 3.2879829778076037e-06,
      "loss": 0.5043,
      "step": 195250
    },
    {
      "epoch": 79.93,
      "grad_norm": 1.6756591796875,
      "learning_rate": 3.2878377889973592e-06,
      "loss": 0.4985,
      "step": 195260
    },
    {
      "epoch": 79.93,
      "grad_norm": 1.921524167060852,
      "learning_rate": 3.2876925958623937e-06,
      "loss": 0.4937,
      "step": 195270
    },
    {
      "epoch": 79.93,
      "grad_norm": 2.009469509124756,
      "learning_rate": 3.2875473984033732e-06,
      "loss": 0.4918,
      "step": 195280
    },
    {
      "epoch": 79.94,
      "grad_norm": 1.8736331462860107,
      "learning_rate": 3.2874021966209637e-06,
      "loss": 0.4994,
      "step": 195290
    },
    {
      "epoch": 79.94,
      "grad_norm": 2.4412591457366943,
      "learning_rate": 3.287256990515829e-06,
      "loss": 0.5186,
      "step": 195300
    },
    {
      "epoch": 79.95,
      "grad_norm": 2.1318187713623047,
      "learning_rate": 3.287111780088635e-06,
      "loss": 0.5006,
      "step": 195310
    },
    {
      "epoch": 79.95,
      "grad_norm": 1.876722812652588,
      "learning_rate": 3.286966565340047e-06,
      "loss": 0.4941,
      "step": 195320
    },
    {
      "epoch": 79.95,
      "grad_norm": 2.628965139389038,
      "learning_rate": 3.2868213462707305e-06,
      "loss": 0.514,
      "step": 195330
    },
    {
      "epoch": 79.96,
      "grad_norm": 2.160407543182373,
      "learning_rate": 3.2866761228813494e-06,
      "loss": 0.5009,
      "step": 195340
    },
    {
      "epoch": 79.96,
      "grad_norm": 1.5585225820541382,
      "learning_rate": 3.2865308951725704e-06,
      "loss": 0.5284,
      "step": 195350
    },
    {
      "epoch": 79.97,
      "grad_norm": 2.177858352661133,
      "learning_rate": 3.286385663145057e-06,
      "loss": 0.4956,
      "step": 195360
    },
    {
      "epoch": 79.97,
      "grad_norm": 1.9561399221420288,
      "learning_rate": 3.2862404267994766e-06,
      "loss": 0.5166,
      "step": 195370
    },
    {
      "epoch": 79.98,
      "grad_norm": 2.4241647720336914,
      "learning_rate": 3.286095186136493e-06,
      "loss": 0.495,
      "step": 195380
    },
    {
      "epoch": 79.98,
      "grad_norm": 2.0195462703704834,
      "learning_rate": 3.2859499411567726e-06,
      "loss": 0.485,
      "step": 195390
    },
    {
      "epoch": 79.98,
      "grad_norm": 2.85996150970459,
      "learning_rate": 3.2858046918609803e-06,
      "loss": 0.517,
      "step": 195400
    },
    {
      "epoch": 79.99,
      "grad_norm": 1.982799768447876,
      "learning_rate": 3.2856594382497813e-06,
      "loss": 0.5032,
      "step": 195410
    },
    {
      "epoch": 79.99,
      "grad_norm": 2.7855606079101562,
      "learning_rate": 3.2855141803238405e-06,
      "loss": 0.5136,
      "step": 195420
    },
    {
      "epoch": 80.0,
      "grad_norm": 1.6471617221832275,
      "learning_rate": 3.285368918083824e-06,
      "loss": 0.5087,
      "step": 195430
    },
    {
      "epoch": 80.0,
      "grad_norm": 2.0110023021698,
      "learning_rate": 3.285223651530397e-06,
      "loss": 0.4882,
      "step": 195440
    },
    {
      "epoch": 80.0,
      "eval_loss": 0.5074847936630249,
      "eval_runtime": 79.681,
      "eval_samples_per_second": 43.285,
      "eval_steps_per_second": 5.422,
      "step": 195440
    },
    {
      "epoch": 80.0,
      "grad_norm": 1.9916036128997803,
      "learning_rate": 3.2850783806642255e-06,
      "loss": 0.4687,
      "step": 195450
    },
    {
      "epoch": 80.01,
      "grad_norm": 2.3478333950042725,
      "learning_rate": 3.284933105485973e-06,
      "loss": 0.4915,
      "step": 195460
    },
    {
      "epoch": 80.01,
      "grad_norm": 1.8412771224975586,
      "learning_rate": 3.284787825996308e-06,
      "loss": 0.4827,
      "step": 195470
    },
    {
      "epoch": 80.02,
      "grad_norm": 1.8837307691574097,
      "learning_rate": 3.284642542195895e-06,
      "loss": 0.493,
      "step": 195480
    },
    {
      "epoch": 80.02,
      "grad_norm": 1.6359148025512695,
      "learning_rate": 3.2844972540853975e-06,
      "loss": 0.4854,
      "step": 195490
    },
    {
      "epoch": 80.02,
      "grad_norm": 1.8697439432144165,
      "learning_rate": 3.284351961665483e-06,
      "loss": 0.5037,
      "step": 195500
    },
    {
      "epoch": 80.03,
      "grad_norm": 1.8683688640594482,
      "learning_rate": 3.2842066649368166e-06,
      "loss": 0.5044,
      "step": 195510
    },
    {
      "epoch": 80.03,
      "grad_norm": 1.7971259355545044,
      "learning_rate": 3.2840613639000637e-06,
      "loss": 0.4944,
      "step": 195520
    },
    {
      "epoch": 80.04,
      "grad_norm": 2.05256724357605,
      "learning_rate": 3.28391605855589e-06,
      "loss": 0.5267,
      "step": 195530
    },
    {
      "epoch": 80.04,
      "grad_norm": 3.039715528488159,
      "learning_rate": 3.2837707489049615e-06,
      "loss": 0.5279,
      "step": 195540
    },
    {
      "epoch": 80.05,
      "grad_norm": 1.8295835256576538,
      "learning_rate": 3.2836254349479437e-06,
      "loss": 0.48,
      "step": 195550
    },
    {
      "epoch": 80.05,
      "grad_norm": 1.6141400337219238,
      "learning_rate": 3.2834801166855015e-06,
      "loss": 0.4935,
      "step": 195560
    },
    {
      "epoch": 80.05,
      "grad_norm": 1.7996667623519897,
      "learning_rate": 3.2833347941183015e-06,
      "loss": 0.5125,
      "step": 195570
    },
    {
      "epoch": 80.06,
      "grad_norm": 1.5299782752990723,
      "learning_rate": 3.2831894672470094e-06,
      "loss": 0.4926,
      "step": 195580
    },
    {
      "epoch": 80.06,
      "grad_norm": 1.645134687423706,
      "learning_rate": 3.28304413607229e-06,
      "loss": 0.5129,
      "step": 195590
    },
    {
      "epoch": 80.07,
      "grad_norm": 1.9672092199325562,
      "learning_rate": 3.2828988005948097e-06,
      "loss": 0.5148,
      "step": 195600
    },
    {
      "epoch": 80.07,
      "grad_norm": 1.4939478635787964,
      "learning_rate": 3.2827534608152344e-06,
      "loss": 0.4793,
      "step": 195610
    },
    {
      "epoch": 80.07,
      "grad_norm": 1.8912361860275269,
      "learning_rate": 3.2826081167342305e-06,
      "loss": 0.5136,
      "step": 195620
    },
    {
      "epoch": 80.08,
      "grad_norm": 1.8774471282958984,
      "learning_rate": 3.282462768352462e-06,
      "loss": 0.4991,
      "step": 195630
    },
    {
      "epoch": 80.08,
      "grad_norm": 1.810698390007019,
      "learning_rate": 3.2823174156705956e-06,
      "loss": 0.4979,
      "step": 195640
    },
    {
      "epoch": 80.09,
      "grad_norm": 1.5546667575836182,
      "learning_rate": 3.2821720586892974e-06,
      "loss": 0.5261,
      "step": 195650
    },
    {
      "epoch": 80.09,
      "grad_norm": 1.882102131843567,
      "learning_rate": 3.2820266974092332e-06,
      "loss": 0.5228,
      "step": 195660
    },
    {
      "epoch": 80.09,
      "grad_norm": 1.8173819780349731,
      "learning_rate": 3.281881331831068e-06,
      "loss": 0.5195,
      "step": 195670
    },
    {
      "epoch": 80.1,
      "grad_norm": 1.9328718185424805,
      "learning_rate": 3.28173596195547e-06,
      "loss": 0.5264,
      "step": 195680
    },
    {
      "epoch": 80.1,
      "grad_norm": 2.199244499206543,
      "learning_rate": 3.2815905877831026e-06,
      "loss": 0.5125,
      "step": 195690
    },
    {
      "epoch": 80.11,
      "grad_norm": 1.4972869157791138,
      "learning_rate": 3.2814452093146325e-06,
      "loss": 0.4947,
      "step": 195700
    },
    {
      "epoch": 80.11,
      "grad_norm": 1.8653565645217896,
      "learning_rate": 3.2812998265507267e-06,
      "loss": 0.5083,
      "step": 195710
    },
    {
      "epoch": 80.11,
      "grad_norm": 2.1189680099487305,
      "learning_rate": 3.2811544394920495e-06,
      "loss": 0.502,
      "step": 195720
    },
    {
      "epoch": 80.12,
      "grad_norm": 2.025287389755249,
      "learning_rate": 3.281009048139268e-06,
      "loss": 0.5241,
      "step": 195730
    },
    {
      "epoch": 80.12,
      "grad_norm": 1.7580621242523193,
      "learning_rate": 3.280863652493048e-06,
      "loss": 0.5248,
      "step": 195740
    },
    {
      "epoch": 80.13,
      "grad_norm": 1.9463584423065186,
      "learning_rate": 3.280718252554055e-06,
      "loss": 0.4923,
      "step": 195750
    },
    {
      "epoch": 80.13,
      "grad_norm": 1.8303860425949097,
      "learning_rate": 3.280572848322957e-06,
      "loss": 0.4964,
      "step": 195760
    },
    {
      "epoch": 80.14,
      "grad_norm": 1.7054507732391357,
      "learning_rate": 3.280427439800418e-06,
      "loss": 0.4845,
      "step": 195770
    },
    {
      "epoch": 80.14,
      "grad_norm": 2.181410551071167,
      "learning_rate": 3.280282026987104e-06,
      "loss": 0.5139,
      "step": 195780
    },
    {
      "epoch": 80.14,
      "grad_norm": 1.6095404624938965,
      "learning_rate": 3.280136609883682e-06,
      "loss": 0.5207,
      "step": 195790
    },
    {
      "epoch": 80.15,
      "grad_norm": 1.4437869787216187,
      "learning_rate": 3.279991188490818e-06,
      "loss": 0.5243,
      "step": 195800
    },
    {
      "epoch": 80.15,
      "grad_norm": 1.6637805700302124,
      "learning_rate": 3.2798457628091783e-06,
      "loss": 0.5032,
      "step": 195810
    },
    {
      "epoch": 80.16,
      "grad_norm": 1.7378422021865845,
      "learning_rate": 3.2797003328394288e-06,
      "loss": 0.5017,
      "step": 195820
    },
    {
      "epoch": 80.16,
      "grad_norm": 1.8110754489898682,
      "learning_rate": 3.279554898582236e-06,
      "loss": 0.5216,
      "step": 195830
    },
    {
      "epoch": 80.16,
      "grad_norm": 2.1743032932281494,
      "learning_rate": 3.279409460038266e-06,
      "loss": 0.5143,
      "step": 195840
    },
    {
      "epoch": 80.17,
      "grad_norm": 1.86761474609375,
      "learning_rate": 3.279264017208184e-06,
      "loss": 0.4964,
      "step": 195850
    },
    {
      "epoch": 80.17,
      "grad_norm": 1.4796010255813599,
      "learning_rate": 3.279118570092658e-06,
      "loss": 0.5007,
      "step": 195860
    },
    {
      "epoch": 80.18,
      "grad_norm": 1.992116093635559,
      "learning_rate": 3.2789731186923533e-06,
      "loss": 0.5246,
      "step": 195870
    },
    {
      "epoch": 80.18,
      "grad_norm": 1.8357750177383423,
      "learning_rate": 3.2788276630079366e-06,
      "loss": 0.4847,
      "step": 195880
    },
    {
      "epoch": 80.18,
      "grad_norm": 1.4529825448989868,
      "learning_rate": 3.2786822030400735e-06,
      "loss": 0.509,
      "step": 195890
    },
    {
      "epoch": 80.19,
      "grad_norm": 2.0653865337371826,
      "learning_rate": 3.2785367387894306e-06,
      "loss": 0.4859,
      "step": 195900
    },
    {
      "epoch": 80.19,
      "grad_norm": 2.0352330207824707,
      "learning_rate": 3.2783912702566752e-06,
      "loss": 0.494,
      "step": 195910
    },
    {
      "epoch": 80.2,
      "grad_norm": 2.009089231491089,
      "learning_rate": 3.2782457974424725e-06,
      "loss": 0.511,
      "step": 195920
    },
    {
      "epoch": 80.2,
      "grad_norm": 1.6534162759780884,
      "learning_rate": 3.278100320347489e-06,
      "loss": 0.4926,
      "step": 195930
    },
    {
      "epoch": 80.2,
      "grad_norm": 1.8808079957962036,
      "learning_rate": 3.2779548389723914e-06,
      "loss": 0.4856,
      "step": 195940
    },
    {
      "epoch": 80.21,
      "grad_norm": 1.8713088035583496,
      "learning_rate": 3.277809353317846e-06,
      "loss": 0.4918,
      "step": 195950
    },
    {
      "epoch": 80.21,
      "grad_norm": 1.5730514526367188,
      "learning_rate": 3.2776638633845198e-06,
      "loss": 0.5018,
      "step": 195960
    },
    {
      "epoch": 80.22,
      "grad_norm": 1.9787358045578003,
      "learning_rate": 3.2775183691730785e-06,
      "loss": 0.5149,
      "step": 195970
    },
    {
      "epoch": 80.22,
      "grad_norm": 1.7992143630981445,
      "learning_rate": 3.2773728706841894e-06,
      "loss": 0.4907,
      "step": 195980
    },
    {
      "epoch": 80.23,
      "grad_norm": 1.9549269676208496,
      "learning_rate": 3.2772273679185187e-06,
      "loss": 0.4992,
      "step": 195990
    },
    {
      "epoch": 80.23,
      "grad_norm": 1.8237420320510864,
      "learning_rate": 3.2770818608767322e-06,
      "loss": 0.5029,
      "step": 196000
    },
    {
      "epoch": 80.23,
      "grad_norm": 2.2493693828582764,
      "learning_rate": 3.2769363495594966e-06,
      "loss": 0.5236,
      "step": 196010
    },
    {
      "epoch": 80.24,
      "grad_norm": 2.239257335662842,
      "learning_rate": 3.2767908339674787e-06,
      "loss": 0.4859,
      "step": 196020
    },
    {
      "epoch": 80.24,
      "grad_norm": 2.0431087017059326,
      "learning_rate": 3.276645314101346e-06,
      "loss": 0.5152,
      "step": 196030
    },
    {
      "epoch": 80.25,
      "grad_norm": 1.8943977355957031,
      "learning_rate": 3.2764997899617643e-06,
      "loss": 0.5123,
      "step": 196040
    },
    {
      "epoch": 80.25,
      "grad_norm": 1.8707808256149292,
      "learning_rate": 3.2763542615494005e-06,
      "loss": 0.4855,
      "step": 196050
    },
    {
      "epoch": 80.25,
      "grad_norm": 1.8992990255355835,
      "learning_rate": 3.276208728864921e-06,
      "loss": 0.4799,
      "step": 196060
    },
    {
      "epoch": 80.26,
      "grad_norm": 1.7107996940612793,
      "learning_rate": 3.2760631919089916e-06,
      "loss": 0.4942,
      "step": 196070
    },
    {
      "epoch": 80.26,
      "grad_norm": 1.6538194417953491,
      "learning_rate": 3.2759176506822805e-06,
      "loss": 0.5098,
      "step": 196080
    },
    {
      "epoch": 80.27,
      "grad_norm": 1.8880969285964966,
      "learning_rate": 3.275772105185454e-06,
      "loss": 0.5223,
      "step": 196090
    },
    {
      "epoch": 80.27,
      "grad_norm": 1.6257715225219727,
      "learning_rate": 3.275626555419178e-06,
      "loss": 0.5099,
      "step": 196100
    },
    {
      "epoch": 80.27,
      "grad_norm": 1.7659083604812622,
      "learning_rate": 3.27548100138412e-06,
      "loss": 0.5017,
      "step": 196110
    },
    {
      "epoch": 80.28,
      "grad_norm": 2.119612216949463,
      "learning_rate": 3.275335443080947e-06,
      "loss": 0.5112,
      "step": 196120
    },
    {
      "epoch": 80.28,
      "grad_norm": 1.9106417894363403,
      "learning_rate": 3.275189880510324e-06,
      "loss": 0.5016,
      "step": 196130
    },
    {
      "epoch": 80.29,
      "grad_norm": 1.7858015298843384,
      "learning_rate": 3.2750443136729204e-06,
      "loss": 0.4909,
      "step": 196140
    },
    {
      "epoch": 80.29,
      "grad_norm": 1.4783003330230713,
      "learning_rate": 3.2748987425694015e-06,
      "loss": 0.4996,
      "step": 196150
    },
    {
      "epoch": 80.29,
      "grad_norm": 1.8856679201126099,
      "learning_rate": 3.2747531672004343e-06,
      "loss": 0.4916,
      "step": 196160
    },
    {
      "epoch": 80.3,
      "grad_norm": 1.961180329322815,
      "learning_rate": 3.274607587566686e-06,
      "loss": 0.5134,
      "step": 196170
    },
    {
      "epoch": 80.3,
      "grad_norm": 2.0408573150634766,
      "learning_rate": 3.2744620036688232e-06,
      "loss": 0.5003,
      "step": 196180
    },
    {
      "epoch": 80.31,
      "grad_norm": 1.8078049421310425,
      "learning_rate": 3.274316415507513e-06,
      "loss": 0.5134,
      "step": 196190
    },
    {
      "epoch": 80.31,
      "grad_norm": 1.6289318799972534,
      "learning_rate": 3.2741708230834222e-06,
      "loss": 0.4989,
      "step": 196200
    },
    {
      "epoch": 80.32,
      "grad_norm": 2.0522778034210205,
      "learning_rate": 3.2740252263972167e-06,
      "loss": 0.5157,
      "step": 196210
    },
    {
      "epoch": 80.32,
      "grad_norm": 2.0768415927886963,
      "learning_rate": 3.273879625449565e-06,
      "loss": 0.4795,
      "step": 196220
    },
    {
      "epoch": 80.32,
      "grad_norm": 2.438451051712036,
      "learning_rate": 3.273734020241134e-06,
      "loss": 0.5005,
      "step": 196230
    },
    {
      "epoch": 80.33,
      "grad_norm": 2.0875232219696045,
      "learning_rate": 3.27358841077259e-06,
      "loss": 0.4949,
      "step": 196240
    },
    {
      "epoch": 80.33,
      "grad_norm": 1.7464016675949097,
      "learning_rate": 3.2734427970446e-06,
      "loss": 0.4933,
      "step": 196250
    },
    {
      "epoch": 80.34,
      "grad_norm": 1.8969473838806152,
      "learning_rate": 3.273297179057832e-06,
      "loss": 0.5028,
      "step": 196260
    },
    {
      "epoch": 80.34,
      "grad_norm": 1.5600451231002808,
      "learning_rate": 3.273151556812952e-06,
      "loss": 0.4918,
      "step": 196270
    },
    {
      "epoch": 80.34,
      "grad_norm": 1.563336730003357,
      "learning_rate": 3.2730059303106277e-06,
      "loss": 0.497,
      "step": 196280
    },
    {
      "epoch": 80.35,
      "grad_norm": 2.074784755706787,
      "learning_rate": 3.2728602995515253e-06,
      "loss": 0.5147,
      "step": 196290
    },
    {
      "epoch": 80.35,
      "grad_norm": 2.5415146350860596,
      "learning_rate": 3.2727146645363127e-06,
      "loss": 0.5169,
      "step": 196300
    },
    {
      "epoch": 80.36,
      "grad_norm": 1.7153987884521484,
      "learning_rate": 3.2725690252656558e-06,
      "loss": 0.4978,
      "step": 196310
    },
    {
      "epoch": 80.36,
      "grad_norm": 1.5625386238098145,
      "learning_rate": 3.272423381740224e-06,
      "loss": 0.5091,
      "step": 196320
    },
    {
      "epoch": 80.36,
      "grad_norm": 1.8878686428070068,
      "learning_rate": 3.272277733960683e-06,
      "loss": 0.5475,
      "step": 196330
    },
    {
      "epoch": 80.37,
      "grad_norm": 1.9805930852890015,
      "learning_rate": 3.2721320819277008e-06,
      "loss": 0.4907,
      "step": 196340
    },
    {
      "epoch": 80.37,
      "grad_norm": 1.8409032821655273,
      "learning_rate": 3.2719864256419434e-06,
      "loss": 0.4949,
      "step": 196350
    },
    {
      "epoch": 80.38,
      "grad_norm": 1.5501872301101685,
      "learning_rate": 3.2718407651040793e-06,
      "loss": 0.5114,
      "step": 196360
    },
    {
      "epoch": 80.38,
      "grad_norm": 1.847366452217102,
      "learning_rate": 3.2716951003147743e-06,
      "loss": 0.5162,
      "step": 196370
    },
    {
      "epoch": 80.38,
      "grad_norm": 1.9713146686553955,
      "learning_rate": 3.2715494312746972e-06,
      "loss": 0.5095,
      "step": 196380
    },
    {
      "epoch": 80.39,
      "grad_norm": 1.6071285009384155,
      "learning_rate": 3.271403757984514e-06,
      "loss": 0.5035,
      "step": 196390
    },
    {
      "epoch": 80.39,
      "grad_norm": 1.310930848121643,
      "learning_rate": 3.2712580804448925e-06,
      "loss": 0.5209,
      "step": 196400
    },
    {
      "epoch": 80.4,
      "grad_norm": 2.109776020050049,
      "learning_rate": 3.2711123986565007e-06,
      "loss": 0.483,
      "step": 196410
    },
    {
      "epoch": 80.4,
      "grad_norm": 2.281702756881714,
      "learning_rate": 3.2709667126200053e-06,
      "loss": 0.5293,
      "step": 196420
    },
    {
      "epoch": 80.41,
      "grad_norm": 1.6922341585159302,
      "learning_rate": 3.2708210223360738e-06,
      "loss": 0.5223,
      "step": 196430
    },
    {
      "epoch": 80.41,
      "grad_norm": 1.906448245048523,
      "learning_rate": 3.270675327805373e-06,
      "loss": 0.4945,
      "step": 196440
    },
    {
      "epoch": 80.41,
      "grad_norm": 1.9096065759658813,
      "learning_rate": 3.270529629028571e-06,
      "loss": 0.5226,
      "step": 196450
    },
    {
      "epoch": 80.42,
      "grad_norm": 1.6649147272109985,
      "learning_rate": 3.2703839260063354e-06,
      "loss": 0.5,
      "step": 196460
    },
    {
      "epoch": 80.42,
      "grad_norm": 1.759008765220642,
      "learning_rate": 3.270238218739333e-06,
      "loss": 0.4871,
      "step": 196470
    },
    {
      "epoch": 80.43,
      "grad_norm": 1.5957540273666382,
      "learning_rate": 3.270092507228232e-06,
      "loss": 0.5179,
      "step": 196480
    },
    {
      "epoch": 80.43,
      "grad_norm": 1.7478336095809937,
      "learning_rate": 3.269946791473699e-06,
      "loss": 0.4995,
      "step": 196490
    },
    {
      "epoch": 80.43,
      "grad_norm": 1.8926788568496704,
      "learning_rate": 3.2698010714764024e-06,
      "loss": 0.513,
      "step": 196500
    },
    {
      "epoch": 80.44,
      "grad_norm": 1.6795496940612793,
      "learning_rate": 3.2696553472370084e-06,
      "loss": 0.5065,
      "step": 196510
    },
    {
      "epoch": 80.44,
      "grad_norm": 1.464028239250183,
      "learning_rate": 3.269509618756186e-06,
      "loss": 0.4919,
      "step": 196520
    },
    {
      "epoch": 80.45,
      "grad_norm": 1.7704267501831055,
      "learning_rate": 3.2693638860346018e-06,
      "loss": 0.5093,
      "step": 196530
    },
    {
      "epoch": 80.45,
      "grad_norm": 2.110180616378784,
      "learning_rate": 3.269218149072924e-06,
      "loss": 0.5023,
      "step": 196540
    },
    {
      "epoch": 80.45,
      "grad_norm": 1.996406078338623,
      "learning_rate": 3.2690724078718202e-06,
      "loss": 0.4931,
      "step": 196550
    },
    {
      "epoch": 80.46,
      "grad_norm": 2.0950851440429688,
      "learning_rate": 3.268926662431958e-06,
      "loss": 0.5035,
      "step": 196560
    },
    {
      "epoch": 80.46,
      "grad_norm": 2.0685226917266846,
      "learning_rate": 3.2687809127540043e-06,
      "loss": 0.5096,
      "step": 196570
    },
    {
      "epoch": 80.47,
      "grad_norm": 2.1554837226867676,
      "learning_rate": 3.2686351588386268e-06,
      "loss": 0.5046,
      "step": 196580
    },
    {
      "epoch": 80.47,
      "grad_norm": 1.8346630334854126,
      "learning_rate": 3.2684894006864937e-06,
      "loss": 0.5032,
      "step": 196590
    },
    {
      "epoch": 80.47,
      "grad_norm": 1.448906421661377,
      "learning_rate": 3.2683436382982727e-06,
      "loss": 0.5296,
      "step": 196600
    },
    {
      "epoch": 80.48,
      "grad_norm": 2.5439369678497314,
      "learning_rate": 3.268197871674632e-06,
      "loss": 0.5118,
      "step": 196610
    },
    {
      "epoch": 80.48,
      "grad_norm": 1.8185081481933594,
      "learning_rate": 3.2680521008162386e-06,
      "loss": 0.5281,
      "step": 196620
    },
    {
      "epoch": 80.49,
      "grad_norm": 1.9601713418960571,
      "learning_rate": 3.2679063257237605e-06,
      "loss": 0.5036,
      "step": 196630
    },
    {
      "epoch": 80.49,
      "grad_norm": 2.9203999042510986,
      "learning_rate": 3.267760546397865e-06,
      "loss": 0.5064,
      "step": 196640
    },
    {
      "epoch": 80.5,
      "grad_norm": 1.6528987884521484,
      "learning_rate": 3.2676147628392206e-06,
      "loss": 0.4985,
      "step": 196650
    },
    {
      "epoch": 80.5,
      "grad_norm": 1.6694658994674683,
      "learning_rate": 3.2674689750484946e-06,
      "loss": 0.5009,
      "step": 196660
    },
    {
      "epoch": 80.5,
      "grad_norm": 2.387153148651123,
      "learning_rate": 3.267323183026355e-06,
      "loss": 0.4893,
      "step": 196670
    },
    {
      "epoch": 80.51,
      "grad_norm": 1.7173635959625244,
      "learning_rate": 3.267177386773469e-06,
      "loss": 0.5122,
      "step": 196680
    },
    {
      "epoch": 80.51,
      "grad_norm": 1.8761601448059082,
      "learning_rate": 3.2670315862905063e-06,
      "loss": 0.5153,
      "step": 196690
    },
    {
      "epoch": 80.52,
      "grad_norm": 1.530327558517456,
      "learning_rate": 3.266885781578134e-06,
      "loss": 0.5262,
      "step": 196700
    },
    {
      "epoch": 80.52,
      "grad_norm": 1.5921474695205688,
      "learning_rate": 3.2667399726370185e-06,
      "loss": 0.506,
      "step": 196710
    },
    {
      "epoch": 80.52,
      "grad_norm": 1.7338887453079224,
      "learning_rate": 3.2665941594678297e-06,
      "loss": 0.483,
      "step": 196720
    },
    {
      "epoch": 80.53,
      "grad_norm": 1.9365819692611694,
      "learning_rate": 3.2664483420712338e-06,
      "loss": 0.5088,
      "step": 196730
    },
    {
      "epoch": 80.53,
      "grad_norm": 1.5713937282562256,
      "learning_rate": 3.2663025204479002e-06,
      "loss": 0.5018,
      "step": 196740
    },
    {
      "epoch": 80.54,
      "grad_norm": 1.921775460243225,
      "learning_rate": 3.2661566945984965e-06,
      "loss": 0.4901,
      "step": 196750
    },
    {
      "epoch": 80.54,
      "grad_norm": 1.8768473863601685,
      "learning_rate": 3.2660108645236906e-06,
      "loss": 0.5029,
      "step": 196760
    },
    {
      "epoch": 80.54,
      "grad_norm": 1.491135597229004,
      "learning_rate": 3.2658650302241507e-06,
      "loss": 0.4675,
      "step": 196770
    },
    {
      "epoch": 80.55,
      "grad_norm": 2.3171191215515137,
      "learning_rate": 3.2657191917005443e-06,
      "loss": 0.4915,
      "step": 196780
    },
    {
      "epoch": 80.55,
      "grad_norm": 1.4944390058517456,
      "learning_rate": 3.2655733489535393e-06,
      "loss": 0.5173,
      "step": 196790
    },
    {
      "epoch": 80.56,
      "grad_norm": 1.7686887979507446,
      "learning_rate": 3.2654275019838053e-06,
      "loss": 0.506,
      "step": 196800
    },
    {
      "epoch": 80.56,
      "grad_norm": 1.8287560939788818,
      "learning_rate": 3.265281650792009e-06,
      "loss": 0.4962,
      "step": 196810
    },
    {
      "epoch": 80.56,
      "grad_norm": 2.357328176498413,
      "learning_rate": 3.265135795378819e-06,
      "loss": 0.4793,
      "step": 196820
    },
    {
      "epoch": 80.57,
      "grad_norm": 1.706049919128418,
      "learning_rate": 3.2649899357449033e-06,
      "loss": 0.4868,
      "step": 196830
    },
    {
      "epoch": 80.57,
      "grad_norm": 1.7588080167770386,
      "learning_rate": 3.26484407189093e-06,
      "loss": 0.486,
      "step": 196840
    },
    {
      "epoch": 80.58,
      "grad_norm": 1.935858130455017,
      "learning_rate": 3.264698203817567e-06,
      "loss": 0.4875,
      "step": 196850
    },
    {
      "epoch": 80.58,
      "grad_norm": 1.4736528396606445,
      "learning_rate": 3.264552331525484e-06,
      "loss": 0.519,
      "step": 196860
    },
    {
      "epoch": 80.59,
      "grad_norm": 1.4192088842391968,
      "learning_rate": 3.264406455015347e-06,
      "loss": 0.4955,
      "step": 196870
    },
    {
      "epoch": 80.59,
      "grad_norm": 1.762450933456421,
      "learning_rate": 3.264260574287826e-06,
      "loss": 0.5204,
      "step": 196880
    },
    {
      "epoch": 80.59,
      "grad_norm": 2.0488359928131104,
      "learning_rate": 3.264114689343589e-06,
      "loss": 0.5042,
      "step": 196890
    },
    {
      "epoch": 80.6,
      "grad_norm": 1.9137604236602783,
      "learning_rate": 3.263968800183303e-06,
      "loss": 0.4969,
      "step": 196900
    },
    {
      "epoch": 80.6,
      "grad_norm": 2.1133663654327393,
      "learning_rate": 3.2638229068076373e-06,
      "loss": 0.5179,
      "step": 196910
    },
    {
      "epoch": 80.61,
      "grad_norm": 1.9998568296432495,
      "learning_rate": 3.263677009217261e-06,
      "loss": 0.4934,
      "step": 196920
    },
    {
      "epoch": 80.61,
      "grad_norm": 2.0833070278167725,
      "learning_rate": 3.2635311074128414e-06,
      "loss": 0.5127,
      "step": 196930
    },
    {
      "epoch": 80.61,
      "grad_norm": 1.7285690307617188,
      "learning_rate": 3.263385201395046e-06,
      "loss": 0.5021,
      "step": 196940
    },
    {
      "epoch": 80.62,
      "grad_norm": 1.8328598737716675,
      "learning_rate": 3.263239291164545e-06,
      "loss": 0.4999,
      "step": 196950
    },
    {
      "epoch": 80.62,
      "grad_norm": 1.736116647720337,
      "learning_rate": 3.2630933767220052e-06,
      "loss": 0.492,
      "step": 196960
    },
    {
      "epoch": 80.63,
      "grad_norm": 1.870914101600647,
      "learning_rate": 3.262947458068096e-06,
      "loss": 0.5013,
      "step": 196970
    },
    {
      "epoch": 80.63,
      "grad_norm": 1.8202213048934937,
      "learning_rate": 3.262801535203486e-06,
      "loss": 0.5072,
      "step": 196980
    },
    {
      "epoch": 80.63,
      "grad_norm": 2.0019431114196777,
      "learning_rate": 3.262655608128843e-06,
      "loss": 0.5073,
      "step": 196990
    },
    {
      "epoch": 80.64,
      "grad_norm": 1.946282148361206,
      "learning_rate": 3.262509676844836e-06,
      "loss": 0.5006,
      "step": 197000
    },
    {
      "epoch": 80.64,
      "grad_norm": 2.0129661560058594,
      "learning_rate": 3.262363741352133e-06,
      "loss": 0.5025,
      "step": 197010
    },
    {
      "epoch": 80.65,
      "grad_norm": 1.7747586965560913,
      "learning_rate": 3.2622178016514026e-06,
      "loss": 0.4926,
      "step": 197020
    },
    {
      "epoch": 80.65,
      "grad_norm": 1.6791658401489258,
      "learning_rate": 3.2620718577433133e-06,
      "loss": 0.4937,
      "step": 197030
    },
    {
      "epoch": 80.65,
      "grad_norm": 1.6687074899673462,
      "learning_rate": 3.2619259096285344e-06,
      "loss": 0.5051,
      "step": 197040
    },
    {
      "epoch": 80.66,
      "grad_norm": 2.3156661987304688,
      "learning_rate": 3.2617799573077333e-06,
      "loss": 0.5076,
      "step": 197050
    },
    {
      "epoch": 80.66,
      "grad_norm": 1.7745494842529297,
      "learning_rate": 3.2616340007815792e-06,
      "loss": 0.4866,
      "step": 197060
    },
    {
      "epoch": 80.67,
      "grad_norm": 2.072030544281006,
      "learning_rate": 3.2614880400507414e-06,
      "loss": 0.528,
      "step": 197070
    },
    {
      "epoch": 80.67,
      "grad_norm": 1.904900312423706,
      "learning_rate": 3.2613420751158867e-06,
      "loss": 0.498,
      "step": 197080
    },
    {
      "epoch": 80.68,
      "grad_norm": 2.0538523197174072,
      "learning_rate": 3.2611961059776856e-06,
      "loss": 0.521,
      "step": 197090
    },
    {
      "epoch": 80.68,
      "grad_norm": 2.0644614696502686,
      "learning_rate": 3.2610501326368057e-06,
      "loss": 0.5248,
      "step": 197100
    },
    {
      "epoch": 80.68,
      "grad_norm": 2.0611424446105957,
      "learning_rate": 3.2609041550939155e-06,
      "loss": 0.5025,
      "step": 197110
    },
    {
      "epoch": 80.69,
      "grad_norm": 2.0989365577697754,
      "learning_rate": 3.2607581733496844e-06,
      "loss": 0.5147,
      "step": 197120
    },
    {
      "epoch": 80.69,
      "grad_norm": 1.8804073333740234,
      "learning_rate": 3.260612187404781e-06,
      "loss": 0.5265,
      "step": 197130
    },
    {
      "epoch": 80.7,
      "grad_norm": 2.349564790725708,
      "learning_rate": 3.260466197259874e-06,
      "loss": 0.5086,
      "step": 197140
    },
    {
      "epoch": 80.7,
      "grad_norm": 2.0070254802703857,
      "learning_rate": 3.260320202915632e-06,
      "loss": 0.5083,
      "step": 197150
    },
    {
      "epoch": 80.7,
      "grad_norm": 1.947265386581421,
      "learning_rate": 3.260174204372724e-06,
      "loss": 0.5098,
      "step": 197160
    },
    {
      "epoch": 80.71,
      "grad_norm": 1.6739919185638428,
      "learning_rate": 3.2600282016318182e-06,
      "loss": 0.5059,
      "step": 197170
    },
    {
      "epoch": 80.71,
      "grad_norm": 1.645019292831421,
      "learning_rate": 3.2598821946935848e-06,
      "loss": 0.5043,
      "step": 197180
    },
    {
      "epoch": 80.72,
      "grad_norm": 1.8065454959869385,
      "learning_rate": 3.2597361835586906e-06,
      "loss": 0.5026,
      "step": 197190
    },
    {
      "epoch": 80.72,
      "grad_norm": 2.0964393615722656,
      "learning_rate": 3.2595901682278057e-06,
      "loss": 0.4911,
      "step": 197200
    },
    {
      "epoch": 80.72,
      "grad_norm": 2.6185691356658936,
      "learning_rate": 3.2594441487016e-06,
      "loss": 0.5189,
      "step": 197210
    },
    {
      "epoch": 80.73,
      "grad_norm": 1.5919346809387207,
      "learning_rate": 3.2592981249807403e-06,
      "loss": 0.5022,
      "step": 197220
    },
    {
      "epoch": 80.73,
      "grad_norm": 1.7281990051269531,
      "learning_rate": 3.2591520970658965e-06,
      "loss": 0.4844,
      "step": 197230
    },
    {
      "epoch": 80.74,
      "grad_norm": 1.8991841077804565,
      "learning_rate": 3.2590060649577376e-06,
      "loss": 0.5013,
      "step": 197240
    },
    {
      "epoch": 80.74,
      "grad_norm": 1.463986873626709,
      "learning_rate": 3.2588600286569313e-06,
      "loss": 0.492,
      "step": 197250
    },
    {
      "epoch": 80.74,
      "grad_norm": 1.6281050443649292,
      "learning_rate": 3.2587139881641492e-06,
      "loss": 0.5023,
      "step": 197260
    },
    {
      "epoch": 80.75,
      "grad_norm": 1.8192601203918457,
      "learning_rate": 3.258567943480058e-06,
      "loss": 0.4923,
      "step": 197270
    },
    {
      "epoch": 80.75,
      "grad_norm": 1.7722426652908325,
      "learning_rate": 3.2584218946053283e-06,
      "loss": 0.487,
      "step": 197280
    },
    {
      "epoch": 80.76,
      "grad_norm": 1.959876298904419,
      "learning_rate": 3.2582758415406274e-06,
      "loss": 0.5013,
      "step": 197290
    },
    {
      "epoch": 80.76,
      "grad_norm": 2.401749610900879,
      "learning_rate": 3.258129784286625e-06,
      "loss": 0.5118,
      "step": 197300
    },
    {
      "epoch": 80.77,
      "grad_norm": 2.152923345565796,
      "learning_rate": 3.2579837228439917e-06,
      "loss": 0.4998,
      "step": 197310
    },
    {
      "epoch": 80.77,
      "grad_norm": 2.0239617824554443,
      "learning_rate": 3.2578376572133944e-06,
      "loss": 0.4972,
      "step": 197320
    },
    {
      "epoch": 80.77,
      "grad_norm": 1.56205153465271,
      "learning_rate": 3.2576915873955032e-06,
      "loss": 0.5081,
      "step": 197330
    },
    {
      "epoch": 80.78,
      "grad_norm": 1.6832361221313477,
      "learning_rate": 3.257545513390987e-06,
      "loss": 0.5016,
      "step": 197340
    },
    {
      "epoch": 80.78,
      "grad_norm": 1.8375349044799805,
      "learning_rate": 3.2573994352005153e-06,
      "loss": 0.4839,
      "step": 197350
    },
    {
      "epoch": 80.79,
      "grad_norm": 1.686968445777893,
      "learning_rate": 3.2572533528247567e-06,
      "loss": 0.5106,
      "step": 197360
    },
    {
      "epoch": 80.79,
      "grad_norm": 1.9460705518722534,
      "learning_rate": 3.257107266264381e-06,
      "loss": 0.4948,
      "step": 197370
    },
    {
      "epoch": 80.79,
      "grad_norm": 2.222627878189087,
      "learning_rate": 3.2569611755200573e-06,
      "loss": 0.4941,
      "step": 197380
    },
    {
      "epoch": 80.8,
      "grad_norm": 1.7917344570159912,
      "learning_rate": 3.2568150805924546e-06,
      "loss": 0.5097,
      "step": 197390
    },
    {
      "epoch": 80.8,
      "grad_norm": 2.2890937328338623,
      "learning_rate": 3.2566689814822422e-06,
      "loss": 0.5163,
      "step": 197400
    },
    {
      "epoch": 80.81,
      "grad_norm": 2.1471517086029053,
      "learning_rate": 3.256522878190089e-06,
      "loss": 0.5105,
      "step": 197410
    },
    {
      "epoch": 80.81,
      "grad_norm": 1.7601617574691772,
      "learning_rate": 3.256376770716665e-06,
      "loss": 0.5131,
      "step": 197420
    },
    {
      "epoch": 80.81,
      "grad_norm": 1.6622729301452637,
      "learning_rate": 3.2562306590626387e-06,
      "loss": 0.5186,
      "step": 197430
    },
    {
      "epoch": 80.82,
      "grad_norm": 1.8991433382034302,
      "learning_rate": 3.25608454322868e-06,
      "loss": 0.525,
      "step": 197440
    },
    {
      "epoch": 80.82,
      "grad_norm": 2.322190046310425,
      "learning_rate": 3.255938423215458e-06,
      "loss": 0.4995,
      "step": 197450
    },
    {
      "epoch": 80.83,
      "grad_norm": 1.842913269996643,
      "learning_rate": 3.255792299023642e-06,
      "loss": 0.4957,
      "step": 197460
    },
    {
      "epoch": 80.83,
      "grad_norm": 1.8959441184997559,
      "learning_rate": 3.2556461706539016e-06,
      "loss": 0.5067,
      "step": 197470
    },
    {
      "epoch": 80.84,
      "grad_norm": 1.6198021173477173,
      "learning_rate": 3.255500038106906e-06,
      "loss": 0.5302,
      "step": 197480
    },
    {
      "epoch": 80.84,
      "grad_norm": 1.5909911394119263,
      "learning_rate": 3.255353901383325e-06,
      "loss": 0.5023,
      "step": 197490
    },
    {
      "epoch": 80.84,
      "grad_norm": 2.392531633377075,
      "learning_rate": 3.2552077604838277e-06,
      "loss": 0.5174,
      "step": 197500
    },
    {
      "epoch": 80.85,
      "grad_norm": 1.6503537893295288,
      "learning_rate": 3.2550616154090836e-06,
      "loss": 0.501,
      "step": 197510
    },
    {
      "epoch": 80.85,
      "grad_norm": 1.8447520732879639,
      "learning_rate": 3.254915466159762e-06,
      "loss": 0.4792,
      "step": 197520
    },
    {
      "epoch": 80.86,
      "grad_norm": 1.7258821725845337,
      "learning_rate": 3.254769312736532e-06,
      "loss": 0.4844,
      "step": 197530
    },
    {
      "epoch": 80.86,
      "grad_norm": 2.1064932346343994,
      "learning_rate": 3.2546231551400636e-06,
      "loss": 0.5193,
      "step": 197540
    },
    {
      "epoch": 80.86,
      "grad_norm": 1.9171468019485474,
      "learning_rate": 3.254476993371027e-06,
      "loss": 0.5057,
      "step": 197550
    },
    {
      "epoch": 80.87,
      "grad_norm": 1.8261936902999878,
      "learning_rate": 3.2543308274300908e-06,
      "loss": 0.507,
      "step": 197560
    },
    {
      "epoch": 80.87,
      "grad_norm": 2.037834644317627,
      "learning_rate": 3.254184657317925e-06,
      "loss": 0.5099,
      "step": 197570
    },
    {
      "epoch": 80.88,
      "grad_norm": 1.5912175178527832,
      "learning_rate": 3.254038483035199e-06,
      "loss": 0.5058,
      "step": 197580
    },
    {
      "epoch": 80.88,
      "grad_norm": 1.6371794939041138,
      "learning_rate": 3.2538923045825824e-06,
      "loss": 0.5208,
      "step": 197590
    },
    {
      "epoch": 80.88,
      "grad_norm": 1.581292986869812,
      "learning_rate": 3.2537461219607448e-06,
      "loss": 0.5069,
      "step": 197600
    },
    {
      "epoch": 80.89,
      "grad_norm": 1.6035183668136597,
      "learning_rate": 3.2535999351703556e-06,
      "loss": 0.5007,
      "step": 197610
    },
    {
      "epoch": 80.89,
      "grad_norm": 1.762613296508789,
      "learning_rate": 3.2534537442120854e-06,
      "loss": 0.4984,
      "step": 197620
    },
    {
      "epoch": 80.9,
      "grad_norm": 1.7007381916046143,
      "learning_rate": 3.2533075490866024e-06,
      "loss": 0.519,
      "step": 197630
    },
    {
      "epoch": 80.9,
      "grad_norm": 1.555495262145996,
      "learning_rate": 3.253161349794578e-06,
      "loss": 0.5038,
      "step": 197640
    },
    {
      "epoch": 80.9,
      "grad_norm": 1.7745842933654785,
      "learning_rate": 3.2530151463366805e-06,
      "loss": 0.5065,
      "step": 197650
    },
    {
      "epoch": 80.91,
      "grad_norm": 1.6894769668579102,
      "learning_rate": 3.2528689387135804e-06,
      "loss": 0.5069,
      "step": 197660
    },
    {
      "epoch": 80.91,
      "grad_norm": 1.7994698286056519,
      "learning_rate": 3.2527227269259466e-06,
      "loss": 0.5228,
      "step": 197670
    },
    {
      "epoch": 80.92,
      "grad_norm": 1.6886532306671143,
      "learning_rate": 3.2525765109744505e-06,
      "loss": 0.4917,
      "step": 197680
    },
    {
      "epoch": 80.92,
      "grad_norm": 1.6083554029464722,
      "learning_rate": 3.25243029085976e-06,
      "loss": 0.4985,
      "step": 197690
    },
    {
      "epoch": 80.93,
      "grad_norm": 2.045091152191162,
      "learning_rate": 3.2522840665825467e-06,
      "loss": 0.4945,
      "step": 197700
    },
    {
      "epoch": 80.93,
      "grad_norm": 1.8936530351638794,
      "learning_rate": 3.2521378381434793e-06,
      "loss": 0.5051,
      "step": 197710
    },
    {
      "epoch": 80.93,
      "grad_norm": 2.1457722187042236,
      "learning_rate": 3.2519916055432275e-06,
      "loss": 0.5068,
      "step": 197720
    },
    {
      "epoch": 80.94,
      "grad_norm": 1.919988751411438,
      "learning_rate": 3.2518453687824617e-06,
      "loss": 0.4996,
      "step": 197730
    },
    {
      "epoch": 80.94,
      "grad_norm": 2.3439953327178955,
      "learning_rate": 3.251699127861852e-06,
      "loss": 0.5235,
      "step": 197740
    },
    {
      "epoch": 80.95,
      "grad_norm": 1.561431646347046,
      "learning_rate": 3.2515528827820675e-06,
      "loss": 0.5179,
      "step": 197750
    },
    {
      "epoch": 80.95,
      "grad_norm": 1.5444293022155762,
      "learning_rate": 3.2514066335437788e-06,
      "loss": 0.5314,
      "step": 197760
    },
    {
      "epoch": 80.95,
      "grad_norm": 1.55958890914917,
      "learning_rate": 3.2512603801476556e-06,
      "loss": 0.5049,
      "step": 197770
    },
    {
      "epoch": 80.96,
      "grad_norm": 1.5958302021026611,
      "learning_rate": 3.251114122594368e-06,
      "loss": 0.4903,
      "step": 197780
    },
    {
      "epoch": 80.96,
      "grad_norm": 2.042896270751953,
      "learning_rate": 3.250967860884586e-06,
      "loss": 0.4966,
      "step": 197790
    },
    {
      "epoch": 80.97,
      "grad_norm": 1.9159202575683594,
      "learning_rate": 3.2508215950189794e-06,
      "loss": 0.5074,
      "step": 197800
    },
    {
      "epoch": 80.97,
      "grad_norm": 1.7547227144241333,
      "learning_rate": 3.2506753249982185e-06,
      "loss": 0.4926,
      "step": 197810
    },
    {
      "epoch": 80.97,
      "grad_norm": 1.6597070693969727,
      "learning_rate": 3.2505290508229723e-06,
      "loss": 0.5242,
      "step": 197820
    },
    {
      "epoch": 80.98,
      "grad_norm": 1.4627689123153687,
      "learning_rate": 3.250382772493912e-06,
      "loss": 0.5051,
      "step": 197830
    },
    {
      "epoch": 80.98,
      "grad_norm": 2.1884636878967285,
      "learning_rate": 3.250236490011708e-06,
      "loss": 0.5109,
      "step": 197840
    },
    {
      "epoch": 80.99,
      "grad_norm": 1.9753061532974243,
      "learning_rate": 3.250090203377029e-06,
      "loss": 0.511,
      "step": 197850
    },
    {
      "epoch": 80.99,
      "grad_norm": 2.477938652038574,
      "learning_rate": 3.2499439125905463e-06,
      "loss": 0.5142,
      "step": 197860
    },
    {
      "epoch": 80.99,
      "grad_norm": 1.9213787317276,
      "learning_rate": 3.2497976176529298e-06,
      "loss": 0.4993,
      "step": 197870
    },
    {
      "epoch": 81.0,
      "grad_norm": 2.323026418685913,
      "learning_rate": 3.2496513185648492e-06,
      "loss": 0.5037,
      "step": 197880
    },
    {
      "epoch": 81.0,
      "eval_loss": 0.5062646865844727,
      "eval_runtime": 73.6233,
      "eval_samples_per_second": 46.847,
      "eval_steps_per_second": 5.868,
      "step": 197883
    },
    {
      "epoch": 81.0,
      "grad_norm": 2.1888856887817383,
      "learning_rate": 3.249505015326975e-06,
      "loss": 0.516,
      "step": 197890
    },
    {
      "epoch": 81.01,
      "grad_norm": 1.9207074642181396,
      "learning_rate": 3.2493587079399774e-06,
      "loss": 0.5126,
      "step": 197900
    },
    {
      "epoch": 81.01,
      "grad_norm": 1.7620164155960083,
      "learning_rate": 3.2492123964045262e-06,
      "loss": 0.5001,
      "step": 197910
    },
    {
      "epoch": 81.02,
      "grad_norm": 1.9338380098342896,
      "learning_rate": 3.2490660807212923e-06,
      "loss": 0.5038,
      "step": 197920
    },
    {
      "epoch": 81.02,
      "grad_norm": 1.6903856992721558,
      "learning_rate": 3.248919760890946e-06,
      "loss": 0.5209,
      "step": 197930
    },
    {
      "epoch": 81.02,
      "grad_norm": 1.6345356702804565,
      "learning_rate": 3.2487734369141573e-06,
      "loss": 0.5093,
      "step": 197940
    },
    {
      "epoch": 81.03,
      "grad_norm": 1.4205330610275269,
      "learning_rate": 3.2486271087915957e-06,
      "loss": 0.5266,
      "step": 197950
    },
    {
      "epoch": 81.03,
      "grad_norm": 1.7369266748428345,
      "learning_rate": 3.2484807765239327e-06,
      "loss": 0.5178,
      "step": 197960
    },
    {
      "epoch": 81.04,
      "grad_norm": 1.6954246759414673,
      "learning_rate": 3.248334440111838e-06,
      "loss": 0.5265,
      "step": 197970
    },
    {
      "epoch": 81.04,
      "grad_norm": 1.4756749868392944,
      "learning_rate": 3.248188099555982e-06,
      "loss": 0.5091,
      "step": 197980
    },
    {
      "epoch": 81.04,
      "grad_norm": 1.7833858728408813,
      "learning_rate": 3.248041754857035e-06,
      "loss": 0.5229,
      "step": 197990
    },
    {
      "epoch": 81.05,
      "grad_norm": 2.1007368564605713,
      "learning_rate": 3.2478954060156675e-06,
      "loss": 0.5171,
      "step": 198000
    },
    {
      "epoch": 81.05,
      "grad_norm": 1.7618016004562378,
      "learning_rate": 3.24774905303255e-06,
      "loss": 0.4871,
      "step": 198010
    },
    {
      "epoch": 81.06,
      "grad_norm": 1.5142937898635864,
      "learning_rate": 3.247602695908353e-06,
      "loss": 0.508,
      "step": 198020
    },
    {
      "epoch": 81.06,
      "grad_norm": 2.3449928760528564,
      "learning_rate": 3.2474563346437463e-06,
      "loss": 0.5004,
      "step": 198030
    },
    {
      "epoch": 81.06,
      "grad_norm": 1.682025671005249,
      "learning_rate": 3.2473099692394014e-06,
      "loss": 0.4917,
      "step": 198040
    },
    {
      "epoch": 81.07,
      "grad_norm": 2.024888038635254,
      "learning_rate": 3.247163599695988e-06,
      "loss": 0.4954,
      "step": 198050
    },
    {
      "epoch": 81.07,
      "grad_norm": 1.5940442085266113,
      "learning_rate": 3.247017226014177e-06,
      "loss": 0.4998,
      "step": 198060
    },
    {
      "epoch": 81.08,
      "grad_norm": 1.7499401569366455,
      "learning_rate": 3.2468708481946383e-06,
      "loss": 0.4929,
      "step": 198070
    },
    {
      "epoch": 81.08,
      "grad_norm": 2.033757448196411,
      "learning_rate": 3.2467244662380436e-06,
      "loss": 0.5007,
      "step": 198080
    },
    {
      "epoch": 81.08,
      "grad_norm": 1.7813570499420166,
      "learning_rate": 3.2465780801450625e-06,
      "loss": 0.5221,
      "step": 198090
    },
    {
      "epoch": 81.09,
      "grad_norm": 1.5078734159469604,
      "learning_rate": 3.2464316899163647e-06,
      "loss": 0.5117,
      "step": 198100
    },
    {
      "epoch": 81.09,
      "grad_norm": 1.4182484149932861,
      "learning_rate": 3.2462852955526232e-06,
      "loss": 0.4988,
      "step": 198110
    },
    {
      "epoch": 81.1,
      "grad_norm": 2.5961337089538574,
      "learning_rate": 3.2461388970545067e-06,
      "loss": 0.511,
      "step": 198120
    },
    {
      "epoch": 81.1,
      "grad_norm": 2.049222707748413,
      "learning_rate": 3.2459924944226864e-06,
      "loss": 0.4987,
      "step": 198130
    },
    {
      "epoch": 81.11,
      "grad_norm": 1.9581745862960815,
      "learning_rate": 3.245846087657833e-06,
      "loss": 0.5038,
      "step": 198140
    },
    {
      "epoch": 81.11,
      "grad_norm": 1.5981303453445435,
      "learning_rate": 3.245699676760617e-06,
      "loss": 0.4967,
      "step": 198150
    },
    {
      "epoch": 81.11,
      "grad_norm": 2.3638525009155273,
      "learning_rate": 3.24555326173171e-06,
      "loss": 0.5025,
      "step": 198160
    },
    {
      "epoch": 81.12,
      "grad_norm": 1.6263376474380493,
      "learning_rate": 3.245406842571781e-06,
      "loss": 0.4924,
      "step": 198170
    },
    {
      "epoch": 81.12,
      "grad_norm": 2.449491500854492,
      "learning_rate": 3.2452604192815023e-06,
      "loss": 0.5037,
      "step": 198180
    },
    {
      "epoch": 81.13,
      "grad_norm": 1.8690992593765259,
      "learning_rate": 3.245113991861543e-06,
      "loss": 0.5237,
      "step": 198190
    },
    {
      "epoch": 81.13,
      "grad_norm": 2.1728358268737793,
      "learning_rate": 3.2449675603125754e-06,
      "loss": 0.4726,
      "step": 198200
    },
    {
      "epoch": 81.13,
      "grad_norm": 1.724146842956543,
      "learning_rate": 3.24482112463527e-06,
      "loss": 0.5095,
      "step": 198210
    },
    {
      "epoch": 81.14,
      "grad_norm": 2.57200026512146,
      "learning_rate": 3.244674684830298e-06,
      "loss": 0.4921,
      "step": 198220
    },
    {
      "epoch": 81.14,
      "grad_norm": 1.7525593042373657,
      "learning_rate": 3.2445282408983284e-06,
      "loss": 0.4961,
      "step": 198230
    },
    {
      "epoch": 81.15,
      "grad_norm": 2.2308285236358643,
      "learning_rate": 3.2443817928400337e-06,
      "loss": 0.5087,
      "step": 198240
    },
    {
      "epoch": 81.15,
      "grad_norm": 1.9074479341506958,
      "learning_rate": 3.244235340656084e-06,
      "loss": 0.4975,
      "step": 198250
    },
    {
      "epoch": 81.15,
      "grad_norm": 1.657090425491333,
      "learning_rate": 3.2440888843471504e-06,
      "loss": 0.4811,
      "step": 198260
    },
    {
      "epoch": 81.16,
      "grad_norm": 1.8335062265396118,
      "learning_rate": 3.2439424239139035e-06,
      "loss": 0.4929,
      "step": 198270
    },
    {
      "epoch": 81.16,
      "grad_norm": 1.6326688528060913,
      "learning_rate": 3.243795959357015e-06,
      "loss": 0.4879,
      "step": 198280
    },
    {
      "epoch": 81.17,
      "grad_norm": 2.109926462173462,
      "learning_rate": 3.2436494906771553e-06,
      "loss": 0.527,
      "step": 198290
    },
    {
      "epoch": 81.17,
      "grad_norm": 1.8374078273773193,
      "learning_rate": 3.243503017874995e-06,
      "loss": 0.5078,
      "step": 198300
    },
    {
      "epoch": 81.17,
      "grad_norm": 2.0504186153411865,
      "learning_rate": 3.2433565409512058e-06,
      "loss": 0.4935,
      "step": 198310
    },
    {
      "epoch": 81.18,
      "grad_norm": 2.0463171005249023,
      "learning_rate": 3.2432100599064585e-06,
      "loss": 0.4975,
      "step": 198320
    },
    {
      "epoch": 81.18,
      "grad_norm": 2.082882881164551,
      "learning_rate": 3.2430635747414237e-06,
      "loss": 0.4911,
      "step": 198330
    },
    {
      "epoch": 81.19,
      "grad_norm": 1.6037659645080566,
      "learning_rate": 3.2429170854567726e-06,
      "loss": 0.4742,
      "step": 198340
    },
    {
      "epoch": 81.19,
      "grad_norm": 1.9965338706970215,
      "learning_rate": 3.2427705920531765e-06,
      "loss": 0.4891,
      "step": 198350
    },
    {
      "epoch": 81.2,
      "grad_norm": 1.614193081855774,
      "learning_rate": 3.2426240945313065e-06,
      "loss": 0.512,
      "step": 198360
    },
    {
      "epoch": 81.2,
      "grad_norm": 1.8286778926849365,
      "learning_rate": 3.2424775928918333e-06,
      "loss": 0.5064,
      "step": 198370
    },
    {
      "epoch": 81.2,
      "grad_norm": 1.7515355348587036,
      "learning_rate": 3.242331087135428e-06,
      "loss": 0.4928,
      "step": 198380
    },
    {
      "epoch": 81.21,
      "grad_norm": 1.498926043510437,
      "learning_rate": 3.242184577262762e-06,
      "loss": 0.4955,
      "step": 198390
    },
    {
      "epoch": 81.21,
      "grad_norm": 1.9278526306152344,
      "learning_rate": 3.2420380632745066e-06,
      "loss": 0.5048,
      "step": 198400
    },
    {
      "epoch": 81.22,
      "grad_norm": 1.6114000082015991,
      "learning_rate": 3.2418915451713327e-06,
      "loss": 0.5206,
      "step": 198410
    },
    {
      "epoch": 81.22,
      "grad_norm": 1.5157225131988525,
      "learning_rate": 3.241745022953911e-06,
      "loss": 0.5083,
      "step": 198420
    },
    {
      "epoch": 81.22,
      "grad_norm": 1.6599317789077759,
      "learning_rate": 3.2415984966229136e-06,
      "loss": 0.4968,
      "step": 198430
    },
    {
      "epoch": 81.23,
      "grad_norm": 2.172027826309204,
      "learning_rate": 3.241451966179011e-06,
      "loss": 0.5055,
      "step": 198440
    },
    {
      "epoch": 81.23,
      "grad_norm": 1.5905181169509888,
      "learning_rate": 3.2413054316228755e-06,
      "loss": 0.4999,
      "step": 198450
    },
    {
      "epoch": 81.24,
      "grad_norm": 1.8051934242248535,
      "learning_rate": 3.2411588929551766e-06,
      "loss": 0.5124,
      "step": 198460
    },
    {
      "epoch": 81.24,
      "grad_norm": 2.206484317779541,
      "learning_rate": 3.2410123501765874e-06,
      "loss": 0.5074,
      "step": 198470
    },
    {
      "epoch": 81.24,
      "grad_norm": 1.719611406326294,
      "learning_rate": 3.2408658032877772e-06,
      "loss": 0.5007,
      "step": 198480
    },
    {
      "epoch": 81.25,
      "grad_norm": 2.3095955848693848,
      "learning_rate": 3.240719252289419e-06,
      "loss": 0.5099,
      "step": 198490
    },
    {
      "epoch": 81.25,
      "grad_norm": 2.1443071365356445,
      "learning_rate": 3.2405726971821836e-06,
      "loss": 0.5013,
      "step": 198500
    },
    {
      "epoch": 81.26,
      "grad_norm": 2.1464426517486572,
      "learning_rate": 3.2404261379667426e-06,
      "loss": 0.4913,
      "step": 198510
    },
    {
      "epoch": 81.26,
      "grad_norm": 1.9510585069656372,
      "learning_rate": 3.2402795746437674e-06,
      "loss": 0.4941,
      "step": 198520
    },
    {
      "epoch": 81.26,
      "grad_norm": 1.9063829183578491,
      "learning_rate": 3.240133007213928e-06,
      "loss": 0.5105,
      "step": 198530
    },
    {
      "epoch": 81.27,
      "grad_norm": 1.7994283437728882,
      "learning_rate": 3.2399864356778973e-06,
      "loss": 0.5122,
      "step": 198540
    },
    {
      "epoch": 81.27,
      "grad_norm": 1.76238214969635,
      "learning_rate": 3.2398398600363467e-06,
      "loss": 0.5251,
      "step": 198550
    },
    {
      "epoch": 81.28,
      "grad_norm": 2.176753044128418,
      "learning_rate": 3.2396932802899464e-06,
      "loss": 0.4961,
      "step": 198560
    },
    {
      "epoch": 81.28,
      "grad_norm": 2.1267309188842773,
      "learning_rate": 3.239546696439369e-06,
      "loss": 0.4984,
      "step": 198570
    },
    {
      "epoch": 81.29,
      "grad_norm": 1.8941175937652588,
      "learning_rate": 3.239400108485286e-06,
      "loss": 0.4902,
      "step": 198580
    },
    {
      "epoch": 81.29,
      "grad_norm": 2.2295660972595215,
      "learning_rate": 3.2392535164283683e-06,
      "loss": 0.4894,
      "step": 198590
    },
    {
      "epoch": 81.29,
      "grad_norm": 1.6026595830917358,
      "learning_rate": 3.239106920269288e-06,
      "loss": 0.5063,
      "step": 198600
    },
    {
      "epoch": 81.3,
      "grad_norm": 1.9062957763671875,
      "learning_rate": 3.2389603200087163e-06,
      "loss": 0.4822,
      "step": 198610
    },
    {
      "epoch": 81.3,
      "grad_norm": 2.1868574619293213,
      "learning_rate": 3.2388137156473243e-06,
      "loss": 0.5205,
      "step": 198620
    },
    {
      "epoch": 81.31,
      "grad_norm": 1.9654512405395508,
      "learning_rate": 3.238667107185784e-06,
      "loss": 0.5034,
      "step": 198630
    },
    {
      "epoch": 81.31,
      "grad_norm": 1.655477523803711,
      "learning_rate": 3.2385204946247677e-06,
      "loss": 0.4837,
      "step": 198640
    },
    {
      "epoch": 81.31,
      "grad_norm": 1.8168030977249146,
      "learning_rate": 3.238373877964946e-06,
      "loss": 0.4827,
      "step": 198650
    },
    {
      "epoch": 81.32,
      "grad_norm": 2.3158652782440186,
      "learning_rate": 3.2382272572069906e-06,
      "loss": 0.4912,
      "step": 198660
    },
    {
      "epoch": 81.32,
      "grad_norm": 2.212202310562134,
      "learning_rate": 3.238080632351574e-06,
      "loss": 0.4892,
      "step": 198670
    },
    {
      "epoch": 81.33,
      "grad_norm": 1.797276496887207,
      "learning_rate": 3.237934003399367e-06,
      "loss": 0.4966,
      "step": 198680
    },
    {
      "epoch": 81.33,
      "grad_norm": 1.6031768321990967,
      "learning_rate": 3.237787370351041e-06,
      "loss": 0.5015,
      "step": 198690
    },
    {
      "epoch": 81.33,
      "grad_norm": 1.8486605882644653,
      "learning_rate": 3.2376407332072685e-06,
      "loss": 0.5012,
      "step": 198700
    },
    {
      "epoch": 81.34,
      "grad_norm": 1.7465513944625854,
      "learning_rate": 3.2374940919687213e-06,
      "loss": 0.5084,
      "step": 198710
    },
    {
      "epoch": 81.34,
      "grad_norm": 1.9418818950653076,
      "learning_rate": 3.2373474466360708e-06,
      "loss": 0.5174,
      "step": 198720
    },
    {
      "epoch": 81.35,
      "grad_norm": 2.161125421524048,
      "learning_rate": 3.237200797209989e-06,
      "loss": 0.4939,
      "step": 198730
    },
    {
      "epoch": 81.35,
      "grad_norm": 2.4068901538848877,
      "learning_rate": 3.2370541436911476e-06,
      "loss": 0.5123,
      "step": 198740
    },
    {
      "epoch": 81.35,
      "grad_norm": 1.7512575387954712,
      "learning_rate": 3.236907486080218e-06,
      "loss": 0.5026,
      "step": 198750
    },
    {
      "epoch": 81.36,
      "grad_norm": 1.6618777513504028,
      "learning_rate": 3.2367608243778714e-06,
      "loss": 0.5347,
      "step": 198760
    },
    {
      "epoch": 81.36,
      "grad_norm": 1.7117929458618164,
      "learning_rate": 3.2366141585847813e-06,
      "loss": 0.4805,
      "step": 198770
    },
    {
      "epoch": 81.37,
      "grad_norm": 2.4012370109558105,
      "learning_rate": 3.2364674887016186e-06,
      "loss": 0.514,
      "step": 198780
    },
    {
      "epoch": 81.37,
      "grad_norm": 1.965277910232544,
      "learning_rate": 3.2363208147290554e-06,
      "loss": 0.5132,
      "step": 198790
    },
    {
      "epoch": 81.38,
      "grad_norm": 2.019719123840332,
      "learning_rate": 3.2361741366677637e-06,
      "loss": 0.5059,
      "step": 198800
    },
    {
      "epoch": 81.38,
      "grad_norm": 1.8414182662963867,
      "learning_rate": 3.236027454518416e-06,
      "loss": 0.5121,
      "step": 198810
    },
    {
      "epoch": 81.38,
      "grad_norm": 2.6508333683013916,
      "learning_rate": 3.2358807682816824e-06,
      "loss": 0.5042,
      "step": 198820
    },
    {
      "epoch": 81.39,
      "grad_norm": 2.0541937351226807,
      "learning_rate": 3.235734077958236e-06,
      "loss": 0.5054,
      "step": 198830
    },
    {
      "epoch": 81.39,
      "grad_norm": 1.7782870531082153,
      "learning_rate": 3.2355873835487487e-06,
      "loss": 0.4961,
      "step": 198840
    },
    {
      "epoch": 81.4,
      "grad_norm": 2.0838875770568848,
      "learning_rate": 3.235440685053892e-06,
      "loss": 0.5148,
      "step": 198850
    },
    {
      "epoch": 81.4,
      "grad_norm": 1.9465137720108032,
      "learning_rate": 3.235293982474339e-06,
      "loss": 0.5269,
      "step": 198860
    },
    {
      "epoch": 81.4,
      "grad_norm": 1.7686491012573242,
      "learning_rate": 3.2351472758107607e-06,
      "loss": 0.4998,
      "step": 198870
    },
    {
      "epoch": 81.41,
      "grad_norm": 1.8235704898834229,
      "learning_rate": 3.2350005650638305e-06,
      "loss": 0.5166,
      "step": 198880
    },
    {
      "epoch": 81.41,
      "grad_norm": 2.072427749633789,
      "learning_rate": 3.2348538502342192e-06,
      "loss": 0.4962,
      "step": 198890
    },
    {
      "epoch": 81.42,
      "grad_norm": 2.0493690967559814,
      "learning_rate": 3.2347071313225987e-06,
      "loss": 0.5096,
      "step": 198900
    },
    {
      "epoch": 81.42,
      "grad_norm": 3.538564920425415,
      "learning_rate": 3.2345604083296414e-06,
      "loss": 0.5326,
      "step": 198910
    },
    {
      "epoch": 81.42,
      "grad_norm": 1.821941614151001,
      "learning_rate": 3.2344136812560202e-06,
      "loss": 0.5025,
      "step": 198920
    },
    {
      "epoch": 81.43,
      "grad_norm": 1.8914867639541626,
      "learning_rate": 3.2342669501024057e-06,
      "loss": 0.5046,
      "step": 198930
    },
    {
      "epoch": 81.43,
      "grad_norm": 1.6777592897415161,
      "learning_rate": 3.2341202148694716e-06,
      "loss": 0.5182,
      "step": 198940
    },
    {
      "epoch": 81.44,
      "grad_norm": 1.5384379625320435,
      "learning_rate": 3.2339734755578897e-06,
      "loss": 0.497,
      "step": 198950
    },
    {
      "epoch": 81.44,
      "grad_norm": 1.642188310623169,
      "learning_rate": 3.233826732168331e-06,
      "loss": 0.5223,
      "step": 198960
    },
    {
      "epoch": 81.44,
      "grad_norm": 1.937427043914795,
      "learning_rate": 3.2336799847014698e-06,
      "loss": 0.5078,
      "step": 198970
    },
    {
      "epoch": 81.45,
      "grad_norm": 1.6287338733673096,
      "learning_rate": 3.2335332331579765e-06,
      "loss": 0.5229,
      "step": 198980
    },
    {
      "epoch": 81.45,
      "grad_norm": 1.9683934450149536,
      "learning_rate": 3.233386477538524e-06,
      "loss": 0.4889,
      "step": 198990
    },
    {
      "epoch": 81.46,
      "grad_norm": 1.914523959159851,
      "learning_rate": 3.2332397178437844e-06,
      "loss": 0.4873,
      "step": 199000
    },
    {
      "epoch": 81.46,
      "grad_norm": 1.9193753004074097,
      "learning_rate": 3.2330929540744306e-06,
      "loss": 0.5121,
      "step": 199010
    },
    {
      "epoch": 81.47,
      "grad_norm": 2.3327178955078125,
      "learning_rate": 3.232946186231134e-06,
      "loss": 0.4963,
      "step": 199020
    },
    {
      "epoch": 81.47,
      "grad_norm": 2.110518217086792,
      "learning_rate": 3.2327994143145685e-06,
      "loss": 0.5096,
      "step": 199030
    },
    {
      "epoch": 81.47,
      "grad_norm": 2.3429934978485107,
      "learning_rate": 3.2326526383254037e-06,
      "loss": 0.5272,
      "step": 199040
    },
    {
      "epoch": 81.48,
      "grad_norm": 2.127471685409546,
      "learning_rate": 3.2325058582643143e-06,
      "loss": 0.5402,
      "step": 199050
    },
    {
      "epoch": 81.48,
      "grad_norm": 2.684199094772339,
      "learning_rate": 3.2323590741319725e-06,
      "loss": 0.5082,
      "step": 199060
    },
    {
      "epoch": 81.49,
      "grad_norm": 1.3711837530136108,
      "learning_rate": 3.2322122859290496e-06,
      "loss": 0.4946,
      "step": 199070
    },
    {
      "epoch": 81.49,
      "grad_norm": 1.719547152519226,
      "learning_rate": 3.232065493656219e-06,
      "loss": 0.4829,
      "step": 199080
    },
    {
      "epoch": 81.49,
      "grad_norm": 2.4071645736694336,
      "learning_rate": 3.2319186973141517e-06,
      "loss": 0.4964,
      "step": 199090
    },
    {
      "epoch": 81.5,
      "grad_norm": 1.9083523750305176,
      "learning_rate": 3.2317718969035227e-06,
      "loss": 0.4988,
      "step": 199100
    },
    {
      "epoch": 81.5,
      "grad_norm": 1.797005295753479,
      "learning_rate": 3.231625092425002e-06,
      "loss": 0.4833,
      "step": 199110
    },
    {
      "epoch": 81.51,
      "grad_norm": 1.5918008089065552,
      "learning_rate": 3.231478283879262e-06,
      "loss": 0.4759,
      "step": 199120
    },
    {
      "epoch": 81.51,
      "grad_norm": 2.1517162322998047,
      "learning_rate": 3.231331471266978e-06,
      "loss": 0.4931,
      "step": 199130
    },
    {
      "epoch": 81.51,
      "grad_norm": 1.9874709844589233,
      "learning_rate": 3.2311846545888193e-06,
      "loss": 0.5007,
      "step": 199140
    },
    {
      "epoch": 81.52,
      "grad_norm": 1.9398077726364136,
      "learning_rate": 3.2310378338454604e-06,
      "loss": 0.4976,
      "step": 199150
    },
    {
      "epoch": 81.52,
      "grad_norm": 1.9800574779510498,
      "learning_rate": 3.2308910090375737e-06,
      "loss": 0.5011,
      "step": 199160
    },
    {
      "epoch": 81.53,
      "grad_norm": 1.9771082401275635,
      "learning_rate": 3.2307441801658313e-06,
      "loss": 0.5011,
      "step": 199170
    },
    {
      "epoch": 81.53,
      "grad_norm": 1.4817609786987305,
      "learning_rate": 3.230597347230906e-06,
      "loss": 0.5002,
      "step": 199180
    },
    {
      "epoch": 81.53,
      "grad_norm": 2.2280893325805664,
      "learning_rate": 3.2304505102334703e-06,
      "loss": 0.4827,
      "step": 199190
    },
    {
      "epoch": 81.54,
      "grad_norm": 2.3424265384674072,
      "learning_rate": 3.230303669174197e-06,
      "loss": 0.4912,
      "step": 199200
    },
    {
      "epoch": 81.54,
      "grad_norm": 1.9716291427612305,
      "learning_rate": 3.2301568240537583e-06,
      "loss": 0.4934,
      "step": 199210
    },
    {
      "epoch": 81.55,
      "grad_norm": 1.5832213163375854,
      "learning_rate": 3.230009974872827e-06,
      "loss": 0.4997,
      "step": 199220
    },
    {
      "epoch": 81.55,
      "grad_norm": 1.473285436630249,
      "learning_rate": 3.2298631216320765e-06,
      "loss": 0.4977,
      "step": 199230
    },
    {
      "epoch": 81.56,
      "grad_norm": 2.1107659339904785,
      "learning_rate": 3.2297162643321787e-06,
      "loss": 0.4889,
      "step": 199240
    },
    {
      "epoch": 81.56,
      "grad_norm": 1.8502142429351807,
      "learning_rate": 3.229569402973807e-06,
      "loss": 0.4974,
      "step": 199250
    },
    {
      "epoch": 81.56,
      "grad_norm": 1.6066362857818604,
      "learning_rate": 3.2294225375576334e-06,
      "loss": 0.496,
      "step": 199260
    },
    {
      "epoch": 81.57,
      "grad_norm": 2.0398573875427246,
      "learning_rate": 3.229275668084331e-06,
      "loss": 0.4778,
      "step": 199270
    },
    {
      "epoch": 81.57,
      "grad_norm": 1.8145267963409424,
      "learning_rate": 3.229128794554573e-06,
      "loss": 0.5048,
      "step": 199280
    },
    {
      "epoch": 81.58,
      "grad_norm": 2.041921377182007,
      "learning_rate": 3.2289819169690323e-06,
      "loss": 0.4997,
      "step": 199290
    },
    {
      "epoch": 81.58,
      "grad_norm": 1.966853380203247,
      "learning_rate": 3.2288350353283803e-06,
      "loss": 0.4909,
      "step": 199300
    },
    {
      "epoch": 81.58,
      "grad_norm": 1.4358712434768677,
      "learning_rate": 3.2286881496332916e-06,
      "loss": 0.5075,
      "step": 199310
    },
    {
      "epoch": 81.59,
      "grad_norm": 1.9516183137893677,
      "learning_rate": 3.228541259884438e-06,
      "loss": 0.5001,
      "step": 199320
    },
    {
      "epoch": 81.59,
      "grad_norm": 1.937870979309082,
      "learning_rate": 3.2283943660824925e-06,
      "loss": 0.4887,
      "step": 199330
    },
    {
      "epoch": 81.6,
      "grad_norm": 2.395512342453003,
      "learning_rate": 3.2282474682281285e-06,
      "loss": 0.4918,
      "step": 199340
    },
    {
      "epoch": 81.6,
      "grad_norm": 1.7454551458358765,
      "learning_rate": 3.2281005663220184e-06,
      "loss": 0.507,
      "step": 199350
    },
    {
      "epoch": 81.6,
      "grad_norm": 1.8095600605010986,
      "learning_rate": 3.227953660364835e-06,
      "loss": 0.507,
      "step": 199360
    },
    {
      "epoch": 81.61,
      "grad_norm": 2.147231340408325,
      "learning_rate": 3.2278067503572518e-06,
      "loss": 0.5039,
      "step": 199370
    },
    {
      "epoch": 81.61,
      "grad_norm": 1.9807785749435425,
      "learning_rate": 3.227659836299942e-06,
      "loss": 0.5078,
      "step": 199380
    },
    {
      "epoch": 81.62,
      "grad_norm": 1.796728491783142,
      "learning_rate": 3.227512918193578e-06,
      "loss": 0.4848,
      "step": 199390
    },
    {
      "epoch": 81.62,
      "grad_norm": 2.1254544258117676,
      "learning_rate": 3.227365996038833e-06,
      "loss": 0.4936,
      "step": 199400
    },
    {
      "epoch": 81.63,
      "grad_norm": 2.355010509490967,
      "learning_rate": 3.2272190698363796e-06,
      "loss": 0.5035,
      "step": 199410
    },
    {
      "epoch": 81.63,
      "grad_norm": 1.8445104360580444,
      "learning_rate": 3.2270721395868907e-06,
      "loss": 0.5043,
      "step": 199420
    },
    {
      "epoch": 81.63,
      "grad_norm": 2.0091187953948975,
      "learning_rate": 3.2269252052910403e-06,
      "loss": 0.5152,
      "step": 199430
    },
    {
      "epoch": 81.64,
      "grad_norm": 1.6827623844146729,
      "learning_rate": 3.2267782669495014e-06,
      "loss": 0.5241,
      "step": 199440
    },
    {
      "epoch": 81.64,
      "grad_norm": 1.731070637702942,
      "learning_rate": 3.2266313245629464e-06,
      "loss": 0.5117,
      "step": 199450
    },
    {
      "epoch": 81.65,
      "grad_norm": 1.458977460861206,
      "learning_rate": 3.2264843781320497e-06,
      "loss": 0.5174,
      "step": 199460
    },
    {
      "epoch": 81.65,
      "grad_norm": 1.7603847980499268,
      "learning_rate": 3.226337427657483e-06,
      "loss": 0.4988,
      "step": 199470
    },
    {
      "epoch": 81.65,
      "grad_norm": 2.147350311279297,
      "learning_rate": 3.2261904731399194e-06,
      "loss": 0.5149,
      "step": 199480
    },
    {
      "epoch": 81.66,
      "grad_norm": 2.175623655319214,
      "learning_rate": 3.2260435145800327e-06,
      "loss": 0.5097,
      "step": 199490
    },
    {
      "epoch": 81.66,
      "grad_norm": 1.7475205659866333,
      "learning_rate": 3.2258965519784965e-06,
      "loss": 0.5058,
      "step": 199500
    },
    {
      "epoch": 81.67,
      "grad_norm": 1.7093307971954346,
      "learning_rate": 3.225749585335983e-06,
      "loss": 0.5067,
      "step": 199510
    },
    {
      "epoch": 81.67,
      "grad_norm": 1.4539155960083008,
      "learning_rate": 3.2256026146531666e-06,
      "loss": 0.5006,
      "step": 199520
    },
    {
      "epoch": 81.67,
      "grad_norm": 1.5353193283081055,
      "learning_rate": 3.2254556399307202e-06,
      "loss": 0.5207,
      "step": 199530
    },
    {
      "epoch": 81.68,
      "grad_norm": 1.716579794883728,
      "learning_rate": 3.225308661169316e-06,
      "loss": 0.4897,
      "step": 199540
    },
    {
      "epoch": 81.68,
      "grad_norm": 1.9946258068084717,
      "learning_rate": 3.225161678369629e-06,
      "loss": 0.5093,
      "step": 199550
    },
    {
      "epoch": 81.69,
      "grad_norm": 1.5916671752929688,
      "learning_rate": 3.2250146915323306e-06,
      "loss": 0.5046,
      "step": 199560
    },
    {
      "epoch": 81.69,
      "grad_norm": 1.5955617427825928,
      "learning_rate": 3.224867700658096e-06,
      "loss": 0.4982,
      "step": 199570
    },
    {
      "epoch": 81.69,
      "grad_norm": 1.9332115650177002,
      "learning_rate": 3.224720705747597e-06,
      "loss": 0.524,
      "step": 199580
    },
    {
      "epoch": 81.7,
      "grad_norm": 1.8761550188064575,
      "learning_rate": 3.2245737068015084e-06,
      "loss": 0.51,
      "step": 199590
    },
    {
      "epoch": 81.7,
      "grad_norm": 1.7566324472427368,
      "learning_rate": 3.224426703820502e-06,
      "loss": 0.4893,
      "step": 199600
    },
    {
      "epoch": 81.71,
      "grad_norm": 2.1418375968933105,
      "learning_rate": 3.224279696805252e-06,
      "loss": 0.5011,
      "step": 199610
    },
    {
      "epoch": 81.71,
      "grad_norm": 2.018237829208374,
      "learning_rate": 3.2241326857564322e-06,
      "loss": 0.521,
      "step": 199620
    },
    {
      "epoch": 81.72,
      "grad_norm": 2.2740137577056885,
      "learning_rate": 3.2239856706747162e-06,
      "loss": 0.5024,
      "step": 199630
    },
    {
      "epoch": 81.72,
      "grad_norm": 1.6261872053146362,
      "learning_rate": 3.223838651560776e-06,
      "loss": 0.4998,
      "step": 199640
    },
    {
      "epoch": 81.72,
      "grad_norm": 2.029008150100708,
      "learning_rate": 3.2236916284152866e-06,
      "loss": 0.5078,
      "step": 199650
    },
    {
      "epoch": 81.73,
      "grad_norm": 1.821177363395691,
      "learning_rate": 3.2235446012389205e-06,
      "loss": 0.4899,
      "step": 199660
    },
    {
      "epoch": 81.73,
      "grad_norm": 1.6334764957427979,
      "learning_rate": 3.2233975700323518e-06,
      "loss": 0.5133,
      "step": 199670
    },
    {
      "epoch": 81.74,
      "grad_norm": 1.5228570699691772,
      "learning_rate": 3.2232505347962542e-06,
      "loss": 0.4994,
      "step": 199680
    },
    {
      "epoch": 81.74,
      "grad_norm": 1.724438190460205,
      "learning_rate": 3.2231034955313003e-06,
      "loss": 0.494,
      "step": 199690
    },
    {
      "epoch": 81.74,
      "grad_norm": 1.8255475759506226,
      "learning_rate": 3.2229564522381635e-06,
      "loss": 0.5074,
      "step": 199700
    },
    {
      "epoch": 81.75,
      "grad_norm": 2.258760929107666,
      "learning_rate": 3.222809404917519e-06,
      "loss": 0.513,
      "step": 199710
    },
    {
      "epoch": 81.75,
      "grad_norm": 2.264763116836548,
      "learning_rate": 3.22266235357004e-06,
      "loss": 0.5122,
      "step": 199720
    },
    {
      "epoch": 81.76,
      "grad_norm": 2.0721023082733154,
      "learning_rate": 3.2225152981963984e-06,
      "loss": 0.5037,
      "step": 199730
    },
    {
      "epoch": 81.76,
      "grad_norm": 1.849790096282959,
      "learning_rate": 3.22236823879727e-06,
      "loss": 0.5141,
      "step": 199740
    },
    {
      "epoch": 81.76,
      "grad_norm": 1.8012946844100952,
      "learning_rate": 3.2222211753733273e-06,
      "loss": 0.5116,
      "step": 199750
    },
    {
      "epoch": 81.77,
      "grad_norm": 1.6295430660247803,
      "learning_rate": 3.222074107925244e-06,
      "loss": 0.5078,
      "step": 199760
    },
    {
      "epoch": 81.77,
      "grad_norm": 1.5223579406738281,
      "learning_rate": 3.2219270364536944e-06,
      "loss": 0.4962,
      "step": 199770
    },
    {
      "epoch": 81.78,
      "grad_norm": 1.9786155223846436,
      "learning_rate": 3.2217799609593516e-06,
      "loss": 0.511,
      "step": 199780
    },
    {
      "epoch": 81.78,
      "grad_norm": 2.183664560317993,
      "learning_rate": 3.2216328814428893e-06,
      "loss": 0.5122,
      "step": 199790
    },
    {
      "epoch": 81.78,
      "grad_norm": 1.7194371223449707,
      "learning_rate": 3.2214857979049806e-06,
      "loss": 0.4958,
      "step": 199800
    },
    {
      "epoch": 81.79,
      "grad_norm": 1.4404466152191162,
      "learning_rate": 3.2213387103463014e-06,
      "loss": 0.5112,
      "step": 199810
    },
    {
      "epoch": 81.79,
      "grad_norm": 2.483248233795166,
      "learning_rate": 3.2211916187675246e-06,
      "loss": 0.5045,
      "step": 199820
    },
    {
      "epoch": 81.8,
      "grad_norm": 1.7687902450561523,
      "learning_rate": 3.221044523169323e-06,
      "loss": 0.5172,
      "step": 199830
    },
    {
      "epoch": 81.8,
      "grad_norm": 1.4753828048706055,
      "learning_rate": 3.2208974235523705e-06,
      "loss": 0.5024,
      "step": 199840
    },
    {
      "epoch": 81.81,
      "grad_norm": 1.7255908250808716,
      "learning_rate": 3.2207503199173422e-06,
      "loss": 0.5162,
      "step": 199850
    },
    {
      "epoch": 81.81,
      "grad_norm": 1.8354097604751587,
      "learning_rate": 3.2206032122649116e-06,
      "loss": 0.5092,
      "step": 199860
    },
    {
      "epoch": 81.81,
      "grad_norm": 1.5862466096878052,
      "learning_rate": 3.2204561005957512e-06,
      "loss": 0.4886,
      "step": 199870
    },
    {
      "epoch": 81.82,
      "grad_norm": 1.9439060688018799,
      "learning_rate": 3.2203089849105366e-06,
      "loss": 0.4879,
      "step": 199880
    },
    {
      "epoch": 81.82,
      "grad_norm": 2.0002694129943848,
      "learning_rate": 3.2201618652099407e-06,
      "loss": 0.4975,
      "step": 199890
    },
    {
      "epoch": 81.83,
      "grad_norm": 2.0462591648101807,
      "learning_rate": 3.2200147414946385e-06,
      "loss": 0.499,
      "step": 199900
    },
    {
      "epoch": 81.83,
      "grad_norm": 1.8401833772659302,
      "learning_rate": 3.2198676137653023e-06,
      "loss": 0.516,
      "step": 199910
    },
    {
      "epoch": 81.83,
      "grad_norm": 2.092534065246582,
      "learning_rate": 3.219720482022608e-06,
      "loss": 0.5056,
      "step": 199920
    },
    {
      "epoch": 81.84,
      "grad_norm": 2.2963008880615234,
      "learning_rate": 3.219573346267227e-06,
      "loss": 0.526,
      "step": 199930
    },
    {
      "epoch": 81.84,
      "grad_norm": 1.5786030292510986,
      "learning_rate": 3.2194262064998363e-06,
      "loss": 0.4812,
      "step": 199940
    },
    {
      "epoch": 81.85,
      "grad_norm": 2.090419054031372,
      "learning_rate": 3.2192790627211083e-06,
      "loss": 0.5053,
      "step": 199950
    },
    {
      "epoch": 81.85,
      "grad_norm": 1.864263653755188,
      "learning_rate": 3.219131914931717e-06,
      "loss": 0.5121,
      "step": 199960
    },
    {
      "epoch": 81.85,
      "grad_norm": 1.9075278043746948,
      "learning_rate": 3.2189847631323374e-06,
      "loss": 0.5097,
      "step": 199970
    },
    {
      "epoch": 81.86,
      "grad_norm": 2.359178304672241,
      "learning_rate": 3.2188376073236417e-06,
      "loss": 0.5099,
      "step": 199980
    },
    {
      "epoch": 81.86,
      "grad_norm": 1.665751338005066,
      "learning_rate": 3.2186904475063054e-06,
      "loss": 0.4949,
      "step": 199990
    },
    {
      "epoch": 81.87,
      "grad_norm": 1.8932689428329468,
      "learning_rate": 3.218543283681003e-06,
      "loss": 0.5052,
      "step": 200000
    },
    {
      "epoch": 81.87,
      "grad_norm": 2.0207788944244385,
      "learning_rate": 3.2183961158484076e-06,
      "loss": 0.5081,
      "step": 200010
    },
    {
      "epoch": 81.87,
      "grad_norm": 2.0590171813964844,
      "learning_rate": 3.218248944009194e-06,
      "loss": 0.5009,
      "step": 200020
    },
    {
      "epoch": 81.88,
      "grad_norm": 2.144028425216675,
      "learning_rate": 3.218101768164036e-06,
      "loss": 0.5125,
      "step": 200030
    },
    {
      "epoch": 81.88,
      "grad_norm": 1.96964693069458,
      "learning_rate": 3.217954588313609e-06,
      "loss": 0.5128,
      "step": 200040
    },
    {
      "epoch": 81.89,
      "grad_norm": 1.7827732563018799,
      "learning_rate": 3.2178074044585854e-06,
      "loss": 0.4924,
      "step": 200050
    },
    {
      "epoch": 81.89,
      "grad_norm": 2.2413337230682373,
      "learning_rate": 3.2176602165996403e-06,
      "loss": 0.4981,
      "step": 200060
    },
    {
      "epoch": 81.9,
      "grad_norm": 1.8269308805465698,
      "learning_rate": 3.217513024737447e-06,
      "loss": 0.495,
      "step": 200070
    },
    {
      "epoch": 81.9,
      "grad_norm": 2.1147687435150146,
      "learning_rate": 3.217365828872681e-06,
      "loss": 0.5039,
      "step": 200080
    },
    {
      "epoch": 81.9,
      "grad_norm": 1.6587364673614502,
      "learning_rate": 3.2172186290060165e-06,
      "loss": 0.5058,
      "step": 200090
    },
    {
      "epoch": 81.91,
      "grad_norm": 1.9197808504104614,
      "learning_rate": 3.2170714251381274e-06,
      "loss": 0.4988,
      "step": 200100
    },
    {
      "epoch": 81.91,
      "grad_norm": 2.3036422729492188,
      "learning_rate": 3.2169242172696887e-06,
      "loss": 0.5071,
      "step": 200110
    },
    {
      "epoch": 81.92,
      "grad_norm": 1.7308465242385864,
      "learning_rate": 3.2167770054013735e-06,
      "loss": 0.4778,
      "step": 200120
    },
    {
      "epoch": 81.92,
      "grad_norm": 1.5853878259658813,
      "learning_rate": 3.2166297895338567e-06,
      "loss": 0.5205,
      "step": 200130
    },
    {
      "epoch": 81.92,
      "grad_norm": 1.4725925922393799,
      "learning_rate": 3.216482569667813e-06,
      "loss": 0.513,
      "step": 200140
    },
    {
      "epoch": 81.93,
      "grad_norm": 1.9672660827636719,
      "learning_rate": 3.2163353458039167e-06,
      "loss": 0.4924,
      "step": 200150
    },
    {
      "epoch": 81.93,
      "grad_norm": 1.4414076805114746,
      "learning_rate": 3.216188117942842e-06,
      "loss": 0.4848,
      "step": 200160
    },
    {
      "epoch": 81.94,
      "grad_norm": 1.6286734342575073,
      "learning_rate": 3.2160408860852624e-06,
      "loss": 0.496,
      "step": 200170
    },
    {
      "epoch": 81.94,
      "grad_norm": 2.0496206283569336,
      "learning_rate": 3.2158936502318546e-06,
      "loss": 0.5206,
      "step": 200180
    },
    {
      "epoch": 81.94,
      "grad_norm": 1.689645767211914,
      "learning_rate": 3.2157464103832913e-06,
      "loss": 0.4973,
      "step": 200190
    },
    {
      "epoch": 81.95,
      "grad_norm": 2.0066375732421875,
      "learning_rate": 3.2155991665402477e-06,
      "loss": 0.4911,
      "step": 200200
    },
    {
      "epoch": 81.95,
      "grad_norm": 1.7007304430007935,
      "learning_rate": 3.2154519187033984e-06,
      "loss": 0.5093,
      "step": 200210
    },
    {
      "epoch": 81.96,
      "grad_norm": 1.9896658658981323,
      "learning_rate": 3.215304666873417e-06,
      "loss": 0.5133,
      "step": 200220
    },
    {
      "epoch": 81.96,
      "grad_norm": 1.6983661651611328,
      "learning_rate": 3.215157411050979e-06,
      "loss": 0.5248,
      "step": 200230
    },
    {
      "epoch": 81.96,
      "grad_norm": 2.28237247467041,
      "learning_rate": 3.2150101512367578e-06,
      "loss": 0.5107,
      "step": 200240
    },
    {
      "epoch": 81.97,
      "grad_norm": 2.979870557785034,
      "learning_rate": 3.2148628874314295e-06,
      "loss": 0.5254,
      "step": 200250
    },
    {
      "epoch": 81.97,
      "grad_norm": 1.8385038375854492,
      "learning_rate": 3.2147156196356685e-06,
      "loss": 0.4948,
      "step": 200260
    },
    {
      "epoch": 81.98,
      "grad_norm": 1.7757080793380737,
      "learning_rate": 3.2145683478501477e-06,
      "loss": 0.4787,
      "step": 200270
    },
    {
      "epoch": 81.98,
      "grad_norm": 1.6088989973068237,
      "learning_rate": 3.2144210720755434e-06,
      "loss": 0.5119,
      "step": 200280
    },
    {
      "epoch": 81.99,
      "grad_norm": 1.9894888401031494,
      "learning_rate": 3.21427379231253e-06,
      "loss": 0.5124,
      "step": 200290
    },
    {
      "epoch": 81.99,
      "grad_norm": 2.1860527992248535,
      "learning_rate": 3.2141265085617818e-06,
      "loss": 0.4977,
      "step": 200300
    },
    {
      "epoch": 81.99,
      "grad_norm": 1.6630669832229614,
      "learning_rate": 3.2139792208239734e-06,
      "loss": 0.5098,
      "step": 200310
    },
    {
      "epoch": 82.0,
      "grad_norm": 1.648744821548462,
      "learning_rate": 3.21383192909978e-06,
      "loss": 0.5314,
      "step": 200320
    },
    {
      "epoch": 82.0,
      "eval_loss": 0.5027248859405518,
      "eval_runtime": 74.6286,
      "eval_samples_per_second": 46.216,
      "eval_steps_per_second": 5.789,
      "step": 200326
    },
    {
      "epoch": 82.0,
      "grad_norm": 1.5522692203521729,
      "learning_rate": 3.213684633389876e-06,
      "loss": 0.5032,
      "step": 200330
    },
    {
      "epoch": 82.01,
      "grad_norm": 1.890092372894287,
      "learning_rate": 3.2135373336949368e-06,
      "loss": 0.489,
      "step": 200340
    },
    {
      "epoch": 82.01,
      "grad_norm": 1.708625078201294,
      "learning_rate": 3.2133900300156354e-06,
      "loss": 0.5041,
      "step": 200350
    },
    {
      "epoch": 82.01,
      "grad_norm": 1.8713058233261108,
      "learning_rate": 3.2132427223526475e-06,
      "loss": 0.4992,
      "step": 200360
    },
    {
      "epoch": 82.02,
      "grad_norm": 2.2081429958343506,
      "learning_rate": 3.213095410706649e-06,
      "loss": 0.4841,
      "step": 200370
    },
    {
      "epoch": 82.02,
      "grad_norm": 1.875789999961853,
      "learning_rate": 3.212948095078313e-06,
      "loss": 0.5089,
      "step": 200380
    },
    {
      "epoch": 82.03,
      "grad_norm": 1.7393335103988647,
      "learning_rate": 3.2128007754683155e-06,
      "loss": 0.4905,
      "step": 200390
    },
    {
      "epoch": 82.03,
      "grad_norm": 2.1851158142089844,
      "learning_rate": 3.2126534518773317e-06,
      "loss": 0.4958,
      "step": 200400
    },
    {
      "epoch": 82.03,
      "grad_norm": 2.0462405681610107,
      "learning_rate": 3.2125061243060355e-06,
      "loss": 0.4957,
      "step": 200410
    },
    {
      "epoch": 82.04,
      "grad_norm": 2.011638641357422,
      "learning_rate": 3.212358792755102e-06,
      "loss": 0.4975,
      "step": 200420
    },
    {
      "epoch": 82.04,
      "grad_norm": 1.6467875242233276,
      "learning_rate": 3.2122114572252056e-06,
      "loss": 0.4974,
      "step": 200430
    },
    {
      "epoch": 82.05,
      "grad_norm": 1.9713040590286255,
      "learning_rate": 3.2120641177170217e-06,
      "loss": 0.5071,
      "step": 200440
    },
    {
      "epoch": 82.05,
      "grad_norm": 1.8837016820907593,
      "learning_rate": 3.211916774231226e-06,
      "loss": 0.4782,
      "step": 200450
    },
    {
      "epoch": 82.05,
      "grad_norm": 2.121000289916992,
      "learning_rate": 3.211769426768492e-06,
      "loss": 0.5061,
      "step": 200460
    },
    {
      "epoch": 82.06,
      "grad_norm": 1.7405726909637451,
      "learning_rate": 3.2116220753294963e-06,
      "loss": 0.4987,
      "step": 200470
    },
    {
      "epoch": 82.06,
      "grad_norm": 1.8743220567703247,
      "learning_rate": 3.211474719914913e-06,
      "loss": 0.5046,
      "step": 200480
    },
    {
      "epoch": 82.07,
      "grad_norm": 2.0527470111846924,
      "learning_rate": 3.2113273605254164e-06,
      "loss": 0.5008,
      "step": 200490
    },
    {
      "epoch": 82.07,
      "grad_norm": 2.1285409927368164,
      "learning_rate": 3.211179997161682e-06,
      "loss": 0.4915,
      "step": 200500
    },
    {
      "epoch": 82.08,
      "grad_norm": 1.6669341325759888,
      "learning_rate": 3.211032629824386e-06,
      "loss": 0.5066,
      "step": 200510
    },
    {
      "epoch": 82.08,
      "grad_norm": 2.0204477310180664,
      "learning_rate": 3.210885258514202e-06,
      "loss": 0.4922,
      "step": 200520
    },
    {
      "epoch": 82.08,
      "grad_norm": 1.5232306718826294,
      "learning_rate": 3.210737883231806e-06,
      "loss": 0.4971,
      "step": 200530
    },
    {
      "epoch": 82.09,
      "grad_norm": 1.7506420612335205,
      "learning_rate": 3.210590503977873e-06,
      "loss": 0.4776,
      "step": 200540
    },
    {
      "epoch": 82.09,
      "grad_norm": 2.1280484199523926,
      "learning_rate": 3.210443120753078e-06,
      "loss": 0.5074,
      "step": 200550
    },
    {
      "epoch": 82.1,
      "grad_norm": 2.031907796859741,
      "learning_rate": 3.2102957335580953e-06,
      "loss": 0.5198,
      "step": 200560
    },
    {
      "epoch": 82.1,
      "grad_norm": 1.8345330953598022,
      "learning_rate": 3.210148342393601e-06,
      "loss": 0.5194,
      "step": 200570
    },
    {
      "epoch": 82.1,
      "grad_norm": 2.2478814125061035,
      "learning_rate": 3.2100009472602703e-06,
      "loss": 0.5102,
      "step": 200580
    },
    {
      "epoch": 82.11,
      "grad_norm": 1.6640889644622803,
      "learning_rate": 3.209853548158778e-06,
      "loss": 0.477,
      "step": 200590
    },
    {
      "epoch": 82.11,
      "grad_norm": 1.837868094444275,
      "learning_rate": 3.2097061450898e-06,
      "loss": 0.4975,
      "step": 200600
    },
    {
      "epoch": 82.12,
      "grad_norm": 2.3852405548095703,
      "learning_rate": 3.2095587380540104e-06,
      "loss": 0.4987,
      "step": 200610
    },
    {
      "epoch": 82.12,
      "grad_norm": 1.7744649648666382,
      "learning_rate": 3.209411327052086e-06,
      "loss": 0.522,
      "step": 200620
    },
    {
      "epoch": 82.12,
      "grad_norm": 1.8294075727462769,
      "learning_rate": 3.2092639120847e-06,
      "loss": 0.5008,
      "step": 200630
    },
    {
      "epoch": 82.13,
      "grad_norm": 1.3206226825714111,
      "learning_rate": 3.209116493152529e-06,
      "loss": 0.4998,
      "step": 200640
    },
    {
      "epoch": 82.13,
      "grad_norm": 2.001253843307495,
      "learning_rate": 3.208969070256248e-06,
      "loss": 0.5102,
      "step": 200650
    },
    {
      "epoch": 82.14,
      "grad_norm": 2.111741304397583,
      "learning_rate": 3.2088216433965333e-06,
      "loss": 0.5396,
      "step": 200660
    },
    {
      "epoch": 82.14,
      "grad_norm": 2.1603262424468994,
      "learning_rate": 3.2086742125740583e-06,
      "loss": 0.4793,
      "step": 200670
    },
    {
      "epoch": 82.14,
      "grad_norm": 2.097381591796875,
      "learning_rate": 3.2085267777895e-06,
      "loss": 0.5201,
      "step": 200680
    },
    {
      "epoch": 82.15,
      "grad_norm": 2.176893711090088,
      "learning_rate": 3.208379339043534e-06,
      "loss": 0.5173,
      "step": 200690
    },
    {
      "epoch": 82.15,
      "grad_norm": 1.663307547569275,
      "learning_rate": 3.2082318963368334e-06,
      "loss": 0.4999,
      "step": 200700
    },
    {
      "epoch": 82.16,
      "grad_norm": 1.8746767044067383,
      "learning_rate": 3.208084449670076e-06,
      "loss": 0.5012,
      "step": 200710
    },
    {
      "epoch": 82.16,
      "grad_norm": 1.81235933303833,
      "learning_rate": 3.2079369990439363e-06,
      "loss": 0.4857,
      "step": 200720
    },
    {
      "epoch": 82.17,
      "grad_norm": 2.1106648445129395,
      "learning_rate": 3.2077895444590895e-06,
      "loss": 0.4925,
      "step": 200730
    },
    {
      "epoch": 82.17,
      "grad_norm": 1.7428128719329834,
      "learning_rate": 3.207642085916211e-06,
      "loss": 0.5085,
      "step": 200740
    },
    {
      "epoch": 82.17,
      "grad_norm": 1.3028810024261475,
      "learning_rate": 3.2074946234159777e-06,
      "loss": 0.502,
      "step": 200750
    },
    {
      "epoch": 82.18,
      "grad_norm": 2.55497407913208,
      "learning_rate": 3.207347156959064e-06,
      "loss": 0.5032,
      "step": 200760
    },
    {
      "epoch": 82.18,
      "grad_norm": 1.790123462677002,
      "learning_rate": 3.2071996865461443e-06,
      "loss": 0.4998,
      "step": 200770
    },
    {
      "epoch": 82.19,
      "grad_norm": 1.6213219165802002,
      "learning_rate": 3.207052212177896e-06,
      "loss": 0.5004,
      "step": 200780
    },
    {
      "epoch": 82.19,
      "grad_norm": 2.0416674613952637,
      "learning_rate": 3.206904733854994e-06,
      "loss": 0.5024,
      "step": 200790
    },
    {
      "epoch": 82.19,
      "grad_norm": 1.6672911643981934,
      "learning_rate": 3.2067572515781135e-06,
      "loss": 0.5049,
      "step": 200800
    },
    {
      "epoch": 82.2,
      "grad_norm": 1.664401650428772,
      "learning_rate": 3.2066097653479307e-06,
      "loss": 0.4887,
      "step": 200810
    },
    {
      "epoch": 82.2,
      "grad_norm": 1.6931082010269165,
      "learning_rate": 3.2064622751651212e-06,
      "loss": 0.5069,
      "step": 200820
    },
    {
      "epoch": 82.21,
      "grad_norm": 1.6346392631530762,
      "learning_rate": 3.2063147810303605e-06,
      "loss": 0.5188,
      "step": 200830
    },
    {
      "epoch": 82.21,
      "grad_norm": 2.0525033473968506,
      "learning_rate": 3.206167282944323e-06,
      "loss": 0.5017,
      "step": 200840
    },
    {
      "epoch": 82.21,
      "grad_norm": 1.6615649461746216,
      "learning_rate": 3.206019780907687e-06,
      "loss": 0.5201,
      "step": 200850
    },
    {
      "epoch": 82.22,
      "grad_norm": 1.6469391584396362,
      "learning_rate": 3.205872274921125e-06,
      "loss": 0.4993,
      "step": 200860
    },
    {
      "epoch": 82.22,
      "grad_norm": 1.6190080642700195,
      "learning_rate": 3.205724764985316e-06,
      "loss": 0.4982,
      "step": 200870
    },
    {
      "epoch": 82.23,
      "grad_norm": 1.45289146900177,
      "learning_rate": 3.205577251100933e-06,
      "loss": 0.5144,
      "step": 200880
    },
    {
      "epoch": 82.23,
      "grad_norm": 1.856605887413025,
      "learning_rate": 3.2054297332686535e-06,
      "loss": 0.5091,
      "step": 200890
    },
    {
      "epoch": 82.23,
      "grad_norm": 1.6719391345977783,
      "learning_rate": 3.205282211489152e-06,
      "loss": 0.4996,
      "step": 200900
    },
    {
      "epoch": 82.24,
      "grad_norm": 1.556734561920166,
      "learning_rate": 3.2051346857631055e-06,
      "loss": 0.4943,
      "step": 200910
    },
    {
      "epoch": 82.24,
      "grad_norm": 1.5742266178131104,
      "learning_rate": 3.2049871560911893e-06,
      "loss": 0.4967,
      "step": 200920
    },
    {
      "epoch": 82.25,
      "grad_norm": 1.5796006917953491,
      "learning_rate": 3.2048396224740773e-06,
      "loss": 0.4971,
      "step": 200930
    },
    {
      "epoch": 82.25,
      "grad_norm": 1.7246662378311157,
      "learning_rate": 3.204692084912449e-06,
      "loss": 0.4962,
      "step": 200940
    },
    {
      "epoch": 82.26,
      "grad_norm": 1.8966799974441528,
      "learning_rate": 3.2045445434069776e-06,
      "loss": 0.5123,
      "step": 200950
    },
    {
      "epoch": 82.26,
      "grad_norm": 1.9127949476242065,
      "learning_rate": 3.2043969979583396e-06,
      "loss": 0.498,
      "step": 200960
    },
    {
      "epoch": 82.26,
      "grad_norm": 1.977745532989502,
      "learning_rate": 3.2042494485672114e-06,
      "loss": 0.4832,
      "step": 200970
    },
    {
      "epoch": 82.27,
      "grad_norm": 2.000239610671997,
      "learning_rate": 3.2041018952342687e-06,
      "loss": 0.5047,
      "step": 200980
    },
    {
      "epoch": 82.27,
      "grad_norm": 2.609529972076416,
      "learning_rate": 3.2039543379601864e-06,
      "loss": 0.4983,
      "step": 200990
    },
    {
      "epoch": 82.28,
      "grad_norm": 1.5426822900772095,
      "learning_rate": 3.203806776745642e-06,
      "loss": 0.4918,
      "step": 201000
    },
    {
      "epoch": 82.28,
      "grad_norm": 1.6379989385604858,
      "learning_rate": 3.20365921159131e-06,
      "loss": 0.4836,
      "step": 201010
    },
    {
      "epoch": 82.28,
      "grad_norm": 1.8435336351394653,
      "learning_rate": 3.203511642497867e-06,
      "loss": 0.5024,
      "step": 201020
    },
    {
      "epoch": 82.29,
      "grad_norm": 1.9583172798156738,
      "learning_rate": 3.2033640694659898e-06,
      "loss": 0.5013,
      "step": 201030
    },
    {
      "epoch": 82.29,
      "grad_norm": 2.2490241527557373,
      "learning_rate": 3.2032164924963535e-06,
      "loss": 0.5003,
      "step": 201040
    },
    {
      "epoch": 82.3,
      "grad_norm": 2.0772948265075684,
      "learning_rate": 3.2030689115896347e-06,
      "loss": 0.5083,
      "step": 201050
    },
    {
      "epoch": 82.3,
      "grad_norm": 1.9544605016708374,
      "learning_rate": 3.2029213267465083e-06,
      "loss": 0.5132,
      "step": 201060
    },
    {
      "epoch": 82.3,
      "grad_norm": 1.549637794494629,
      "learning_rate": 3.2027737379676513e-06,
      "loss": 0.4897,
      "step": 201070
    },
    {
      "epoch": 82.31,
      "grad_norm": 1.8421841859817505,
      "learning_rate": 3.2026261452537392e-06,
      "loss": 0.5095,
      "step": 201080
    },
    {
      "epoch": 82.31,
      "grad_norm": 1.8820222616195679,
      "learning_rate": 3.2024785486054498e-06,
      "loss": 0.5002,
      "step": 201090
    },
    {
      "epoch": 82.32,
      "grad_norm": 1.8020308017730713,
      "learning_rate": 3.2023309480234563e-06,
      "loss": 0.5195,
      "step": 201100
    },
    {
      "epoch": 82.32,
      "grad_norm": 2.21722149848938,
      "learning_rate": 3.2021833435084373e-06,
      "loss": 0.4996,
      "step": 201110
    },
    {
      "epoch": 82.33,
      "grad_norm": 1.6004242897033691,
      "learning_rate": 3.2020357350610683e-06,
      "loss": 0.4924,
      "step": 201120
    },
    {
      "epoch": 82.33,
      "grad_norm": 1.6464333534240723,
      "learning_rate": 3.2018881226820247e-06,
      "loss": 0.4825,
      "step": 201130
    },
    {
      "epoch": 82.33,
      "grad_norm": 1.6510905027389526,
      "learning_rate": 3.201740506371984e-06,
      "loss": 0.4845,
      "step": 201140
    },
    {
      "epoch": 82.34,
      "grad_norm": 1.5566728115081787,
      "learning_rate": 3.2015928861316207e-06,
      "loss": 0.506,
      "step": 201150
    },
    {
      "epoch": 82.34,
      "grad_norm": 1.4736202955245972,
      "learning_rate": 3.2014452619616125e-06,
      "loss": 0.5023,
      "step": 201160
    },
    {
      "epoch": 82.35,
      "grad_norm": 1.8844844102859497,
      "learning_rate": 3.201297633862635e-06,
      "loss": 0.4951,
      "step": 201170
    },
    {
      "epoch": 82.35,
      "grad_norm": 1.9526596069335938,
      "learning_rate": 3.201150001835365e-06,
      "loss": 0.4924,
      "step": 201180
    },
    {
      "epoch": 82.35,
      "grad_norm": 1.8571332693099976,
      "learning_rate": 3.201002365880478e-06,
      "loss": 0.4825,
      "step": 201190
    },
    {
      "epoch": 82.36,
      "grad_norm": 1.7976138591766357,
      "learning_rate": 3.200854725998651e-06,
      "loss": 0.4912,
      "step": 201200
    },
    {
      "epoch": 82.36,
      "grad_norm": 1.3453662395477295,
      "learning_rate": 3.2007070821905593e-06,
      "loss": 0.5071,
      "step": 201210
    },
    {
      "epoch": 82.37,
      "grad_norm": 2.073488712310791,
      "learning_rate": 3.20055943445688e-06,
      "loss": 0.522,
      "step": 201220
    },
    {
      "epoch": 82.37,
      "grad_norm": 1.9368607997894287,
      "learning_rate": 3.20041178279829e-06,
      "loss": 0.4996,
      "step": 201230
    },
    {
      "epoch": 82.37,
      "grad_norm": 1.7941341400146484,
      "learning_rate": 3.200264127215465e-06,
      "loss": 0.4883,
      "step": 201240
    },
    {
      "epoch": 82.38,
      "grad_norm": 1.9536062479019165,
      "learning_rate": 3.200116467709081e-06,
      "loss": 0.5101,
      "step": 201250
    },
    {
      "epoch": 82.38,
      "grad_norm": 1.92589271068573,
      "learning_rate": 3.1999688042798154e-06,
      "loss": 0.5005,
      "step": 201260
    },
    {
      "epoch": 82.39,
      "grad_norm": 2.274111032485962,
      "learning_rate": 3.199821136928344e-06,
      "loss": 0.4956,
      "step": 201270
    },
    {
      "epoch": 82.39,
      "grad_norm": 1.9438930749893188,
      "learning_rate": 3.199673465655343e-06,
      "loss": 0.488,
      "step": 201280
    },
    {
      "epoch": 82.39,
      "grad_norm": 1.8205482959747314,
      "learning_rate": 3.199525790461489e-06,
      "loss": 0.491,
      "step": 201290
    },
    {
      "epoch": 82.4,
      "grad_norm": 1.8658901453018188,
      "learning_rate": 3.1993781113474586e-06,
      "loss": 0.4896,
      "step": 201300
    },
    {
      "epoch": 82.4,
      "grad_norm": 1.7099676132202148,
      "learning_rate": 3.1992304283139287e-06,
      "loss": 0.4977,
      "step": 201310
    },
    {
      "epoch": 82.41,
      "grad_norm": 1.6071979999542236,
      "learning_rate": 3.199082741361575e-06,
      "loss": 0.5023,
      "step": 201320
    },
    {
      "epoch": 82.41,
      "grad_norm": 2.1570944786071777,
      "learning_rate": 3.1989350504910744e-06,
      "loss": 0.5005,
      "step": 201330
    },
    {
      "epoch": 82.42,
      "grad_norm": 1.4907712936401367,
      "learning_rate": 3.198787355703105e-06,
      "loss": 0.5108,
      "step": 201340
    },
    {
      "epoch": 82.42,
      "grad_norm": 2.3147995471954346,
      "learning_rate": 3.1986396569983406e-06,
      "loss": 0.4933,
      "step": 201350
    },
    {
      "epoch": 82.42,
      "grad_norm": 2.078707695007324,
      "learning_rate": 3.1984919543774593e-06,
      "loss": 0.5078,
      "step": 201360
    },
    {
      "epoch": 82.43,
      "grad_norm": 1.667795181274414,
      "learning_rate": 3.1983442478411372e-06,
      "loss": 0.5008,
      "step": 201370
    },
    {
      "epoch": 82.43,
      "grad_norm": 2.2529656887054443,
      "learning_rate": 3.1981965373900517e-06,
      "loss": 0.5084,
      "step": 201380
    },
    {
      "epoch": 82.44,
      "grad_norm": 1.615356683731079,
      "learning_rate": 3.198048823024879e-06,
      "loss": 0.4894,
      "step": 201390
    },
    {
      "epoch": 82.44,
      "grad_norm": 1.8930513858795166,
      "learning_rate": 3.1979011047462946e-06,
      "loss": 0.5247,
      "step": 201400
    },
    {
      "epoch": 82.44,
      "grad_norm": 2.1493771076202393,
      "learning_rate": 3.197753382554977e-06,
      "loss": 0.5008,
      "step": 201410
    },
    {
      "epoch": 82.45,
      "grad_norm": 2.011687994003296,
      "learning_rate": 3.1976056564516027e-06,
      "loss": 0.5062,
      "step": 201420
    },
    {
      "epoch": 82.45,
      "grad_norm": 1.808881402015686,
      "learning_rate": 3.197457926436847e-06,
      "loss": 0.4981,
      "step": 201430
    },
    {
      "epoch": 82.46,
      "grad_norm": 1.945608139038086,
      "learning_rate": 3.197310192511388e-06,
      "loss": 0.506,
      "step": 201440
    },
    {
      "epoch": 82.46,
      "grad_norm": 1.5267969369888306,
      "learning_rate": 3.197162454675902e-06,
      "loss": 0.5037,
      "step": 201450
    },
    {
      "epoch": 82.46,
      "grad_norm": 4.526052474975586,
      "learning_rate": 3.1970147129310654e-06,
      "loss": 0.5032,
      "step": 201460
    },
    {
      "epoch": 82.47,
      "grad_norm": 1.906490445137024,
      "learning_rate": 3.1968669672775553e-06,
      "loss": 0.5118,
      "step": 201470
    },
    {
      "epoch": 82.47,
      "grad_norm": 2.261103868484497,
      "learning_rate": 3.1967192177160486e-06,
      "loss": 0.5043,
      "step": 201480
    },
    {
      "epoch": 82.48,
      "grad_norm": 2.0860283374786377,
      "learning_rate": 3.196571464247222e-06,
      "loss": 0.4995,
      "step": 201490
    },
    {
      "epoch": 82.48,
      "grad_norm": 1.6467546224594116,
      "learning_rate": 3.196423706871753e-06,
      "loss": 0.4942,
      "step": 201500
    },
    {
      "epoch": 82.48,
      "grad_norm": 2.5915815830230713,
      "learning_rate": 3.196275945590317e-06,
      "loss": 0.4917,
      "step": 201510
    },
    {
      "epoch": 82.49,
      "grad_norm": 1.7641165256500244,
      "learning_rate": 3.1961281804035916e-06,
      "loss": 0.5135,
      "step": 201520
    },
    {
      "epoch": 82.49,
      "grad_norm": 1.7116400003433228,
      "learning_rate": 3.1959804113122542e-06,
      "loss": 0.5026,
      "step": 201530
    },
    {
      "epoch": 82.5,
      "grad_norm": 1.6635023355484009,
      "learning_rate": 3.1958326383169813e-06,
      "loss": 0.5287,
      "step": 201540
    },
    {
      "epoch": 82.5,
      "grad_norm": 2.5577847957611084,
      "learning_rate": 3.1956848614184493e-06,
      "loss": 0.5256,
      "step": 201550
    },
    {
      "epoch": 82.51,
      "grad_norm": 1.8073372840881348,
      "learning_rate": 3.1955370806173367e-06,
      "loss": 0.4999,
      "step": 201560
    },
    {
      "epoch": 82.51,
      "grad_norm": 3.2119319438934326,
      "learning_rate": 3.1953892959143185e-06,
      "loss": 0.5045,
      "step": 201570
    },
    {
      "epoch": 82.51,
      "grad_norm": 1.8919243812561035,
      "learning_rate": 3.1952415073100727e-06,
      "loss": 0.4854,
      "step": 201580
    },
    {
      "epoch": 82.52,
      "grad_norm": 1.5457422733306885,
      "learning_rate": 3.195093714805276e-06,
      "loss": 0.4999,
      "step": 201590
    },
    {
      "epoch": 82.52,
      "grad_norm": 1.6505743265151978,
      "learning_rate": 3.1949459184006055e-06,
      "loss": 0.4966,
      "step": 201600
    },
    {
      "epoch": 82.53,
      "grad_norm": 1.8164719343185425,
      "learning_rate": 3.1947981180967384e-06,
      "loss": 0.5049,
      "step": 201610
    },
    {
      "epoch": 82.53,
      "grad_norm": 1.8064136505126953,
      "learning_rate": 3.194650313894352e-06,
      "loss": 0.4973,
      "step": 201620
    },
    {
      "epoch": 82.53,
      "grad_norm": 1.8661668300628662,
      "learning_rate": 3.1945025057941236e-06,
      "loss": 0.4874,
      "step": 201630
    },
    {
      "epoch": 82.54,
      "grad_norm": 3.028144121170044,
      "learning_rate": 3.1943546937967287e-06,
      "loss": 0.5081,
      "step": 201640
    },
    {
      "epoch": 82.54,
      "grad_norm": 1.817446231842041,
      "learning_rate": 3.1942068779028458e-06,
      "loss": 0.4932,
      "step": 201650
    },
    {
      "epoch": 82.55,
      "grad_norm": 1.857386589050293,
      "learning_rate": 3.1940590581131517e-06,
      "loss": 0.4849,
      "step": 201660
    },
    {
      "epoch": 82.55,
      "grad_norm": 2.11173415184021,
      "learning_rate": 3.1939112344283233e-06,
      "loss": 0.503,
      "step": 201670
    },
    {
      "epoch": 82.55,
      "grad_norm": 1.552860140800476,
      "learning_rate": 3.193763406849038e-06,
      "loss": 0.5106,
      "step": 201680
    },
    {
      "epoch": 82.56,
      "grad_norm": 1.5388020277023315,
      "learning_rate": 3.193615575375973e-06,
      "loss": 0.4951,
      "step": 201690
    },
    {
      "epoch": 82.56,
      "grad_norm": 1.4405826330184937,
      "learning_rate": 3.193467740009806e-06,
      "loss": 0.5106,
      "step": 201700
    },
    {
      "epoch": 82.57,
      "grad_norm": 1.7921916246414185,
      "learning_rate": 3.193319900751213e-06,
      "loss": 0.5031,
      "step": 201710
    },
    {
      "epoch": 82.57,
      "grad_norm": 1.7895148992538452,
      "learning_rate": 3.1931720576008716e-06,
      "loss": 0.5085,
      "step": 201720
    },
    {
      "epoch": 82.57,
      "grad_norm": 1.412128210067749,
      "learning_rate": 3.1930242105594595e-06,
      "loss": 0.4974,
      "step": 201730
    },
    {
      "epoch": 82.58,
      "grad_norm": 1.5243291854858398,
      "learning_rate": 3.192876359627654e-06,
      "loss": 0.4925,
      "step": 201740
    },
    {
      "epoch": 82.58,
      "grad_norm": 2.4348537921905518,
      "learning_rate": 3.192728504806132e-06,
      "loss": 0.5277,
      "step": 201750
    },
    {
      "epoch": 82.59,
      "grad_norm": 2.1905357837677,
      "learning_rate": 3.192580646095571e-06,
      "loss": 0.4885,
      "step": 201760
    },
    {
      "epoch": 82.59,
      "grad_norm": 1.7887763977050781,
      "learning_rate": 3.192432783496648e-06,
      "loss": 0.5099,
      "step": 201770
    },
    {
      "epoch": 82.6,
      "grad_norm": 1.6772370338439941,
      "learning_rate": 3.192284917010041e-06,
      "loss": 0.4745,
      "step": 201780
    },
    {
      "epoch": 82.6,
      "grad_norm": 2.1477322578430176,
      "learning_rate": 3.192137046636427e-06,
      "loss": 0.5046,
      "step": 201790
    },
    {
      "epoch": 82.6,
      "grad_norm": 1.9100701808929443,
      "learning_rate": 3.191989172376483e-06,
      "loss": 0.5213,
      "step": 201800
    },
    {
      "epoch": 82.61,
      "grad_norm": 1.7274667024612427,
      "learning_rate": 3.1918412942308865e-06,
      "loss": 0.51,
      "step": 201810
    },
    {
      "epoch": 82.61,
      "grad_norm": 2.092287302017212,
      "learning_rate": 3.1916934122003158e-06,
      "loss": 0.5087,
      "step": 201820
    },
    {
      "epoch": 82.62,
      "grad_norm": 1.9595190286636353,
      "learning_rate": 3.191545526285447e-06,
      "loss": 0.4706,
      "step": 201830
    },
    {
      "epoch": 82.62,
      "grad_norm": 2.024542808532715,
      "learning_rate": 3.1913976364869588e-06,
      "loss": 0.507,
      "step": 201840
    },
    {
      "epoch": 82.62,
      "grad_norm": 2.709240436553955,
      "learning_rate": 3.191249742805528e-06,
      "loss": 0.5102,
      "step": 201850
    },
    {
      "epoch": 82.63,
      "grad_norm": 2.0260045528411865,
      "learning_rate": 3.191101845241832e-06,
      "loss": 0.5022,
      "step": 201860
    },
    {
      "epoch": 82.63,
      "grad_norm": 1.9453213214874268,
      "learning_rate": 3.190953943796548e-06,
      "loss": 0.499,
      "step": 201870
    },
    {
      "epoch": 82.64,
      "grad_norm": 1.7454994916915894,
      "learning_rate": 3.1908060384703545e-06,
      "loss": 0.4932,
      "step": 201880
    },
    {
      "epoch": 82.64,
      "grad_norm": 1.9846137762069702,
      "learning_rate": 3.190658129263928e-06,
      "loss": 0.5109,
      "step": 201890
    },
    {
      "epoch": 82.64,
      "grad_norm": 2.2662601470947266,
      "learning_rate": 3.190510216177947e-06,
      "loss": 0.515,
      "step": 201900
    },
    {
      "epoch": 82.65,
      "grad_norm": 1.780734896659851,
      "learning_rate": 3.190362299213088e-06,
      "loss": 0.5075,
      "step": 201910
    },
    {
      "epoch": 82.65,
      "grad_norm": 1.5193465948104858,
      "learning_rate": 3.19021437837003e-06,
      "loss": 0.5133,
      "step": 201920
    },
    {
      "epoch": 82.66,
      "grad_norm": 1.9423749446868896,
      "learning_rate": 3.190066453649449e-06,
      "loss": 0.5087,
      "step": 201930
    },
    {
      "epoch": 82.66,
      "grad_norm": 1.8869911432266235,
      "learning_rate": 3.189918525052024e-06,
      "loss": 0.4997,
      "step": 201940
    },
    {
      "epoch": 82.66,
      "grad_norm": 1.7887972593307495,
      "learning_rate": 3.1897705925784316e-06,
      "loss": 0.499,
      "step": 201950
    },
    {
      "epoch": 82.67,
      "grad_norm": 2.2509231567382812,
      "learning_rate": 3.1896226562293494e-06,
      "loss": 0.5044,
      "step": 201960
    },
    {
      "epoch": 82.67,
      "grad_norm": 1.7828108072280884,
      "learning_rate": 3.189474716005456e-06,
      "loss": 0.5077,
      "step": 201970
    },
    {
      "epoch": 82.68,
      "grad_norm": 2.252194404602051,
      "learning_rate": 3.1893267719074293e-06,
      "loss": 0.5026,
      "step": 201980
    },
    {
      "epoch": 82.68,
      "grad_norm": 1.724046230316162,
      "learning_rate": 3.1891788239359463e-06,
      "loss": 0.4894,
      "step": 201990
    },
    {
      "epoch": 82.69,
      "grad_norm": 1.6906400918960571,
      "learning_rate": 3.1890308720916846e-06,
      "loss": 0.4944,
      "step": 202000
    },
    {
      "epoch": 82.69,
      "grad_norm": 1.756110668182373,
      "learning_rate": 3.1888829163753215e-06,
      "loss": 0.4868,
      "step": 202010
    },
    {
      "epoch": 82.69,
      "grad_norm": 2.029863119125366,
      "learning_rate": 3.188734956787536e-06,
      "loss": 0.4953,
      "step": 202020
    },
    {
      "epoch": 82.7,
      "grad_norm": 2.1081252098083496,
      "learning_rate": 3.1885869933290045e-06,
      "loss": 0.501,
      "step": 202030
    },
    {
      "epoch": 82.7,
      "grad_norm": 1.7641702890396118,
      "learning_rate": 3.1884390260004067e-06,
      "loss": 0.5063,
      "step": 202040
    },
    {
      "epoch": 82.71,
      "grad_norm": 1.985927939414978,
      "learning_rate": 3.188291054802419e-06,
      "loss": 0.474,
      "step": 202050
    },
    {
      "epoch": 82.71,
      "grad_norm": 1.6794806718826294,
      "learning_rate": 3.188143079735719e-06,
      "loss": 0.4936,
      "step": 202060
    },
    {
      "epoch": 82.71,
      "grad_norm": 1.944654941558838,
      "learning_rate": 3.1879951008009855e-06,
      "loss": 0.4834,
      "step": 202070
    },
    {
      "epoch": 82.72,
      "grad_norm": 2.0747530460357666,
      "learning_rate": 3.1878471179988957e-06,
      "loss": 0.4943,
      "step": 202080
    },
    {
      "epoch": 82.72,
      "grad_norm": 1.7340683937072754,
      "learning_rate": 3.187699131330128e-06,
      "loss": 0.506,
      "step": 202090
    },
    {
      "epoch": 82.73,
      "grad_norm": 2.017174482345581,
      "learning_rate": 3.1875511407953593e-06,
      "loss": 0.502,
      "step": 202100
    },
    {
      "epoch": 82.73,
      "grad_norm": 1.8231555223464966,
      "learning_rate": 3.1874031463952687e-06,
      "loss": 0.4931,
      "step": 202110
    },
    {
      "epoch": 82.73,
      "grad_norm": 1.8689980506896973,
      "learning_rate": 3.1872551481305336e-06,
      "loss": 0.5097,
      "step": 202120
    },
    {
      "epoch": 82.74,
      "grad_norm": 1.8649104833602905,
      "learning_rate": 3.187107146001833e-06,
      "loss": 0.503,
      "step": 202130
    },
    {
      "epoch": 82.74,
      "grad_norm": 1.8049066066741943,
      "learning_rate": 3.1869591400098427e-06,
      "loss": 0.5066,
      "step": 202140
    },
    {
      "epoch": 82.75,
      "grad_norm": 1.8178529739379883,
      "learning_rate": 3.186811130155242e-06,
      "loss": 0.4903,
      "step": 202150
    },
    {
      "epoch": 82.75,
      "grad_norm": 2.2480430603027344,
      "learning_rate": 3.1866631164387092e-06,
      "loss": 0.4992,
      "step": 202160
    },
    {
      "epoch": 82.75,
      "grad_norm": 1.8833414316177368,
      "learning_rate": 3.186515098860922e-06,
      "loss": 0.4703,
      "step": 202170
    },
    {
      "epoch": 82.76,
      "grad_norm": 1.832951545715332,
      "learning_rate": 3.1863670774225585e-06,
      "loss": 0.4856,
      "step": 202180
    },
    {
      "epoch": 82.76,
      "grad_norm": 1.8352512121200562,
      "learning_rate": 3.1862190521242967e-06,
      "loss": 0.49,
      "step": 202190
    },
    {
      "epoch": 82.77,
      "grad_norm": 1.8699305057525635,
      "learning_rate": 3.1860710229668144e-06,
      "loss": 0.5252,
      "step": 202200
    },
    {
      "epoch": 82.77,
      "grad_norm": 1.8093394041061401,
      "learning_rate": 3.1859229899507904e-06,
      "loss": 0.496,
      "step": 202210
    },
    {
      "epoch": 82.78,
      "grad_norm": 2.051600217819214,
      "learning_rate": 3.1857749530769023e-06,
      "loss": 0.5119,
      "step": 202220
    },
    {
      "epoch": 82.78,
      "grad_norm": 2.4527344703674316,
      "learning_rate": 3.1856269123458277e-06,
      "loss": 0.5133,
      "step": 202230
    },
    {
      "epoch": 82.78,
      "grad_norm": 1.8662841320037842,
      "learning_rate": 3.1854788677582456e-06,
      "loss": 0.5157,
      "step": 202240
    },
    {
      "epoch": 82.79,
      "grad_norm": 1.7421293258666992,
      "learning_rate": 3.1853308193148336e-06,
      "loss": 0.4887,
      "step": 202250
    },
    {
      "epoch": 82.79,
      "grad_norm": 1.9296119213104248,
      "learning_rate": 3.18518276701627e-06,
      "loss": 0.5011,
      "step": 202260
    },
    {
      "epoch": 82.8,
      "grad_norm": 1.8803592920303345,
      "learning_rate": 3.185034710863234e-06,
      "loss": 0.5129,
      "step": 202270
    },
    {
      "epoch": 82.8,
      "grad_norm": 1.6359001398086548,
      "learning_rate": 3.184886650856403e-06,
      "loss": 0.5065,
      "step": 202280
    },
    {
      "epoch": 82.8,
      "grad_norm": 2.440595865249634,
      "learning_rate": 3.1847385869964553e-06,
      "loss": 0.4932,
      "step": 202290
    },
    {
      "epoch": 82.81,
      "grad_norm": 2.087385416030884,
      "learning_rate": 3.184590519284069e-06,
      "loss": 0.5063,
      "step": 202300
    },
    {
      "epoch": 82.81,
      "grad_norm": 1.4857454299926758,
      "learning_rate": 3.184442447719922e-06,
      "loss": 0.4821,
      "step": 202310
    },
    {
      "epoch": 82.82,
      "grad_norm": 1.6024348735809326,
      "learning_rate": 3.1842943723046935e-06,
      "loss": 0.5026,
      "step": 202320
    },
    {
      "epoch": 82.82,
      "grad_norm": 1.7562196254730225,
      "learning_rate": 3.184146293039061e-06,
      "loss": 0.5059,
      "step": 202330
    },
    {
      "epoch": 82.82,
      "grad_norm": 1.5315221548080444,
      "learning_rate": 3.1839982099237035e-06,
      "loss": 0.5053,
      "step": 202340
    },
    {
      "epoch": 82.83,
      "grad_norm": 1.823701024055481,
      "learning_rate": 3.1838501229592994e-06,
      "loss": 0.4981,
      "step": 202350
    },
    {
      "epoch": 82.83,
      "grad_norm": 1.8457562923431396,
      "learning_rate": 3.183702032146527e-06,
      "loss": 0.5027,
      "step": 202360
    },
    {
      "epoch": 82.84,
      "grad_norm": 2.066411018371582,
      "learning_rate": 3.183553937486064e-06,
      "loss": 0.4952,
      "step": 202370
    },
    {
      "epoch": 82.84,
      "grad_norm": 1.6177328824996948,
      "learning_rate": 3.183405838978589e-06,
      "loss": 0.5005,
      "step": 202380
    },
    {
      "epoch": 82.84,
      "grad_norm": 1.5119080543518066,
      "learning_rate": 3.1832577366247805e-06,
      "loss": 0.4917,
      "step": 202390
    },
    {
      "epoch": 82.85,
      "grad_norm": 1.8120752573013306,
      "learning_rate": 3.1831096304253173e-06,
      "loss": 0.5055,
      "step": 202400
    },
    {
      "epoch": 82.85,
      "grad_norm": 1.793960452079773,
      "learning_rate": 3.1829615203808777e-06,
      "loss": 0.5056,
      "step": 202410
    },
    {
      "epoch": 82.86,
      "grad_norm": 2.0979442596435547,
      "learning_rate": 3.18281340649214e-06,
      "loss": 0.5074,
      "step": 202420
    },
    {
      "epoch": 82.86,
      "grad_norm": 1.8361092805862427,
      "learning_rate": 3.182665288759783e-06,
      "loss": 0.5016,
      "step": 202430
    },
    {
      "epoch": 82.87,
      "grad_norm": 1.6344376802444458,
      "learning_rate": 3.1825171671844848e-06,
      "loss": 0.4899,
      "step": 202440
    },
    {
      "epoch": 82.87,
      "grad_norm": 1.5185428857803345,
      "learning_rate": 3.182369041766924e-06,
      "loss": 0.5072,
      "step": 202450
    },
    {
      "epoch": 82.87,
      "grad_norm": 1.857508897781372,
      "learning_rate": 3.182220912507779e-06,
      "loss": 0.4961,
      "step": 202460
    },
    {
      "epoch": 82.88,
      "grad_norm": 2.031048059463501,
      "learning_rate": 3.1820727794077295e-06,
      "loss": 0.4981,
      "step": 202470
    },
    {
      "epoch": 82.88,
      "grad_norm": 2.027851104736328,
      "learning_rate": 3.1819246424674523e-06,
      "loss": 0.4911,
      "step": 202480
    },
    {
      "epoch": 82.89,
      "grad_norm": 1.7095954418182373,
      "learning_rate": 3.1817765016876277e-06,
      "loss": 0.4906,
      "step": 202490
    },
    {
      "epoch": 82.89,
      "grad_norm": 1.836105465888977,
      "learning_rate": 3.181628357068933e-06,
      "loss": 0.5199,
      "step": 202500
    },
    {
      "epoch": 82.89,
      "grad_norm": 1.8273696899414062,
      "learning_rate": 3.181480208612048e-06,
      "loss": 0.507,
      "step": 202510
    },
    {
      "epoch": 82.9,
      "grad_norm": 1.9154726266860962,
      "learning_rate": 3.18133205631765e-06,
      "loss": 0.4961,
      "step": 202520
    },
    {
      "epoch": 82.9,
      "grad_norm": 2.424572229385376,
      "learning_rate": 3.1811839001864177e-06,
      "loss": 0.4939,
      "step": 202530
    },
    {
      "epoch": 82.91,
      "grad_norm": 1.5897648334503174,
      "learning_rate": 3.1810357402190306e-06,
      "loss": 0.4949,
      "step": 202540
    },
    {
      "epoch": 82.91,
      "grad_norm": 1.5618302822113037,
      "learning_rate": 3.1808875764161678e-06,
      "loss": 0.5003,
      "step": 202550
    },
    {
      "epoch": 82.91,
      "grad_norm": 1.764986276626587,
      "learning_rate": 3.1807394087785072e-06,
      "loss": 0.5101,
      "step": 202560
    },
    {
      "epoch": 82.92,
      "grad_norm": 1.3462069034576416,
      "learning_rate": 3.1805912373067284e-06,
      "loss": 0.5159,
      "step": 202570
    },
    {
      "epoch": 82.92,
      "grad_norm": 2.121539831161499,
      "learning_rate": 3.1804430620015084e-06,
      "loss": 0.4872,
      "step": 202580
    },
    {
      "epoch": 82.93,
      "grad_norm": 1.7633121013641357,
      "learning_rate": 3.1802948828635278e-06,
      "loss": 0.5026,
      "step": 202590
    },
    {
      "epoch": 82.93,
      "grad_norm": 1.5721849203109741,
      "learning_rate": 3.1801466998934643e-06,
      "loss": 0.5077,
      "step": 202600
    },
    {
      "epoch": 82.93,
      "grad_norm": 2.007145404815674,
      "learning_rate": 3.179998513091997e-06,
      "loss": 0.4946,
      "step": 202610
    },
    {
      "epoch": 82.94,
      "grad_norm": 1.7906452417373657,
      "learning_rate": 3.1798503224598045e-06,
      "loss": 0.4966,
      "step": 202620
    },
    {
      "epoch": 82.94,
      "grad_norm": 1.6520500183105469,
      "learning_rate": 3.179702127997566e-06,
      "loss": 0.5094,
      "step": 202630
    },
    {
      "epoch": 82.95,
      "grad_norm": 6.835563659667969,
      "learning_rate": 3.1795539297059614e-06,
      "loss": 0.517,
      "step": 202640
    },
    {
      "epoch": 82.95,
      "grad_norm": 1.512961506843567,
      "learning_rate": 3.1794057275856675e-06,
      "loss": 0.4999,
      "step": 202650
    },
    {
      "epoch": 82.96,
      "grad_norm": 1.6534957885742188,
      "learning_rate": 3.179257521637364e-06,
      "loss": 0.4901,
      "step": 202660
    },
    {
      "epoch": 82.96,
      "grad_norm": 1.91008460521698,
      "learning_rate": 3.17910931186173e-06,
      "loss": 0.4878,
      "step": 202670
    },
    {
      "epoch": 82.96,
      "grad_norm": 1.929890513420105,
      "learning_rate": 3.178961098259445e-06,
      "loss": 0.4954,
      "step": 202680
    },
    {
      "epoch": 82.97,
      "grad_norm": 1.7988414764404297,
      "learning_rate": 3.1788128808311867e-06,
      "loss": 0.4962,
      "step": 202690
    },
    {
      "epoch": 82.97,
      "grad_norm": 2.4297657012939453,
      "learning_rate": 3.1786646595776345e-06,
      "loss": 0.4827,
      "step": 202700
    },
    {
      "epoch": 82.98,
      "grad_norm": 1.4636280536651611,
      "learning_rate": 3.178516434499468e-06,
      "loss": 0.5126,
      "step": 202710
    },
    {
      "epoch": 82.98,
      "grad_norm": 1.597193717956543,
      "learning_rate": 3.178368205597366e-06,
      "loss": 0.5029,
      "step": 202720
    },
    {
      "epoch": 82.98,
      "grad_norm": 2.0821871757507324,
      "learning_rate": 3.1782199728720064e-06,
      "loss": 0.4958,
      "step": 202730
    },
    {
      "epoch": 82.99,
      "grad_norm": 2.6138458251953125,
      "learning_rate": 3.17807173632407e-06,
      "loss": 0.5008,
      "step": 202740
    },
    {
      "epoch": 82.99,
      "grad_norm": 1.5240585803985596,
      "learning_rate": 3.177923495954234e-06,
      "loss": 0.4993,
      "step": 202750
    },
    {
      "epoch": 83.0,
      "grad_norm": 1.80186927318573,
      "learning_rate": 3.177775251763179e-06,
      "loss": 0.5181,
      "step": 202760
    },
    {
      "epoch": 83.0,
      "eval_loss": 0.5013130307197571,
      "eval_runtime": 75.0943,
      "eval_samples_per_second": 45.929,
      "eval_steps_per_second": 5.753,
      "step": 202769
    },
    {
      "epoch": 83.0,
      "grad_norm": 1.805198311805725,
      "learning_rate": 3.177627003751583e-06,
      "loss": 0.511,
      "step": 202770
    },
    {
      "epoch": 83.0,
      "grad_norm": 1.8763389587402344,
      "learning_rate": 3.1774787519201257e-06,
      "loss": 0.5037,
      "step": 202780
    },
    {
      "epoch": 83.01,
      "grad_norm": 1.7026870250701904,
      "learning_rate": 3.1773304962694862e-06,
      "loss": 0.5235,
      "step": 202790
    },
    {
      "epoch": 83.01,
      "grad_norm": 1.6038975715637207,
      "learning_rate": 3.1771822368003447e-06,
      "loss": 0.492,
      "step": 202800
    },
    {
      "epoch": 83.02,
      "grad_norm": 1.8509371280670166,
      "learning_rate": 3.177033973513378e-06,
      "loss": 0.496,
      "step": 202810
    },
    {
      "epoch": 83.02,
      "grad_norm": 2.2481632232666016,
      "learning_rate": 3.1768857064092665e-06,
      "loss": 0.5193,
      "step": 202820
    },
    {
      "epoch": 83.02,
      "grad_norm": 1.8956143856048584,
      "learning_rate": 3.1767374354886893e-06,
      "loss": 0.5143,
      "step": 202830
    },
    {
      "epoch": 83.03,
      "grad_norm": 2.105139970779419,
      "learning_rate": 3.1765891607523256e-06,
      "loss": 0.4713,
      "step": 202840
    },
    {
      "epoch": 83.03,
      "grad_norm": 1.8900619745254517,
      "learning_rate": 3.1764408822008553e-06,
      "loss": 0.5132,
      "step": 202850
    },
    {
      "epoch": 83.04,
      "grad_norm": 2.021683931350708,
      "learning_rate": 3.1762925998349567e-06,
      "loss": 0.502,
      "step": 202860
    },
    {
      "epoch": 83.04,
      "grad_norm": 1.5487909317016602,
      "learning_rate": 3.1761443136553097e-06,
      "loss": 0.4954,
      "step": 202870
    },
    {
      "epoch": 83.05,
      "grad_norm": 2.350313663482666,
      "learning_rate": 3.1759960236625926e-06,
      "loss": 0.5008,
      "step": 202880
    },
    {
      "epoch": 83.05,
      "grad_norm": 1.68433678150177,
      "learning_rate": 3.1758477298574866e-06,
      "loss": 0.4972,
      "step": 202890
    },
    {
      "epoch": 83.05,
      "grad_norm": 1.988840937614441,
      "learning_rate": 3.1756994322406687e-06,
      "loss": 0.5134,
      "step": 202900
    },
    {
      "epoch": 83.06,
      "grad_norm": 1.604211688041687,
      "learning_rate": 3.1755511308128184e-06,
      "loss": 0.5012,
      "step": 202910
    },
    {
      "epoch": 83.06,
      "grad_norm": 1.601094365119934,
      "learning_rate": 3.175402825574617e-06,
      "loss": 0.4755,
      "step": 202920
    },
    {
      "epoch": 83.07,
      "grad_norm": 1.8487205505371094,
      "learning_rate": 3.1752545165267436e-06,
      "loss": 0.5086,
      "step": 202930
    },
    {
      "epoch": 83.07,
      "grad_norm": 1.6660206317901611,
      "learning_rate": 3.175106203669876e-06,
      "loss": 0.5015,
      "step": 202940
    },
    {
      "epoch": 83.07,
      "grad_norm": 2.2674217224121094,
      "learning_rate": 3.174957887004694e-06,
      "loss": 0.5074,
      "step": 202950
    },
    {
      "epoch": 83.08,
      "grad_norm": 1.6834760904312134,
      "learning_rate": 3.174809566531878e-06,
      "loss": 0.5065,
      "step": 202960
    },
    {
      "epoch": 83.08,
      "grad_norm": 1.414398193359375,
      "learning_rate": 3.1746612422521067e-06,
      "loss": 0.4944,
      "step": 202970
    },
    {
      "epoch": 83.09,
      "grad_norm": 2.1782233715057373,
      "learning_rate": 3.174512914166059e-06,
      "loss": 0.4977,
      "step": 202980
    },
    {
      "epoch": 83.09,
      "grad_norm": 1.6880584955215454,
      "learning_rate": 3.174364582274416e-06,
      "loss": 0.5035,
      "step": 202990
    },
    {
      "epoch": 83.09,
      "grad_norm": 1.6637897491455078,
      "learning_rate": 3.174216246577856e-06,
      "loss": 0.493,
      "step": 203000
    },
    {
      "epoch": 83.1,
      "grad_norm": 1.5856914520263672,
      "learning_rate": 3.1740679070770584e-06,
      "loss": 0.5119,
      "step": 203010
    },
    {
      "epoch": 83.1,
      "grad_norm": 2.479426145553589,
      "learning_rate": 3.173919563772703e-06,
      "loss": 0.5287,
      "step": 203020
    },
    {
      "epoch": 83.11,
      "grad_norm": 1.9606314897537231,
      "learning_rate": 3.1737712166654695e-06,
      "loss": 0.5068,
      "step": 203030
    },
    {
      "epoch": 83.11,
      "grad_norm": 1.5909615755081177,
      "learning_rate": 3.1736228657560377e-06,
      "loss": 0.4959,
      "step": 203040
    },
    {
      "epoch": 83.12,
      "grad_norm": 1.6686785221099854,
      "learning_rate": 3.173474511045087e-06,
      "loss": 0.4875,
      "step": 203050
    },
    {
      "epoch": 83.12,
      "grad_norm": 2.20559024810791,
      "learning_rate": 3.1733261525332964e-06,
      "loss": 0.5203,
      "step": 203060
    },
    {
      "epoch": 83.12,
      "grad_norm": 1.8856686353683472,
      "learning_rate": 3.173177790221346e-06,
      "loss": 0.4978,
      "step": 203070
    },
    {
      "epoch": 83.13,
      "grad_norm": 1.8680886030197144,
      "learning_rate": 3.1730294241099148e-06,
      "loss": 0.5022,
      "step": 203080
    },
    {
      "epoch": 83.13,
      "grad_norm": 1.7636891603469849,
      "learning_rate": 3.172881054199684e-06,
      "loss": 0.4993,
      "step": 203090
    },
    {
      "epoch": 83.14,
      "grad_norm": 2.0299570560455322,
      "learning_rate": 3.1727326804913317e-06,
      "loss": 0.4804,
      "step": 203100
    },
    {
      "epoch": 83.14,
      "grad_norm": 1.9335271120071411,
      "learning_rate": 3.1725843029855384e-06,
      "loss": 0.4911,
      "step": 203110
    },
    {
      "epoch": 83.14,
      "grad_norm": 1.6176362037658691,
      "learning_rate": 3.172435921682983e-06,
      "loss": 0.49,
      "step": 203120
    },
    {
      "epoch": 83.15,
      "grad_norm": 1.9404659271240234,
      "learning_rate": 3.172287536584346e-06,
      "loss": 0.501,
      "step": 203130
    },
    {
      "epoch": 83.15,
      "grad_norm": 1.617592215538025,
      "learning_rate": 3.1721391476903063e-06,
      "loss": 0.5023,
      "step": 203140
    },
    {
      "epoch": 83.16,
      "grad_norm": 1.5899447202682495,
      "learning_rate": 3.171990755001545e-06,
      "loss": 0.5024,
      "step": 203150
    },
    {
      "epoch": 83.16,
      "grad_norm": 1.802623987197876,
      "learning_rate": 3.1718423585187416e-06,
      "loss": 0.5202,
      "step": 203160
    },
    {
      "epoch": 83.16,
      "grad_norm": 1.8559426069259644,
      "learning_rate": 3.1716939582425745e-06,
      "loss": 0.4977,
      "step": 203170
    },
    {
      "epoch": 83.17,
      "grad_norm": 2.3001527786254883,
      "learning_rate": 3.1715455541737247e-06,
      "loss": 0.4996,
      "step": 203180
    },
    {
      "epoch": 83.17,
      "grad_norm": 1.398279070854187,
      "learning_rate": 3.171397146312871e-06,
      "loss": 0.5043,
      "step": 203190
    },
    {
      "epoch": 83.18,
      "grad_norm": 1.9000341892242432,
      "learning_rate": 3.171248734660694e-06,
      "loss": 0.4768,
      "step": 203200
    },
    {
      "epoch": 83.18,
      "grad_norm": 1.735587477684021,
      "learning_rate": 3.1711003192178742e-06,
      "loss": 0.4816,
      "step": 203210
    },
    {
      "epoch": 83.18,
      "grad_norm": 2.3087332248687744,
      "learning_rate": 3.1709518999850903e-06,
      "loss": 0.5098,
      "step": 203220
    },
    {
      "epoch": 83.19,
      "grad_norm": 2.7978293895721436,
      "learning_rate": 3.1708034769630224e-06,
      "loss": 0.491,
      "step": 203230
    },
    {
      "epoch": 83.19,
      "grad_norm": 1.9434571266174316,
      "learning_rate": 3.170655050152351e-06,
      "loss": 0.479,
      "step": 203240
    },
    {
      "epoch": 83.2,
      "grad_norm": 2.0688507556915283,
      "learning_rate": 3.1705066195537558e-06,
      "loss": 0.497,
      "step": 203250
    },
    {
      "epoch": 83.2,
      "grad_norm": 1.9393852949142456,
      "learning_rate": 3.170358185167916e-06,
      "loss": 0.5125,
      "step": 203260
    },
    {
      "epoch": 83.21,
      "grad_norm": 1.7582495212554932,
      "learning_rate": 3.170209746995512e-06,
      "loss": 0.497,
      "step": 203270
    },
    {
      "epoch": 83.21,
      "grad_norm": 2.526315212249756,
      "learning_rate": 3.170061305037224e-06,
      "loss": 0.5063,
      "step": 203280
    },
    {
      "epoch": 83.21,
      "grad_norm": 1.9573923349380493,
      "learning_rate": 3.1699128592937325e-06,
      "loss": 0.5054,
      "step": 203290
    },
    {
      "epoch": 83.22,
      "grad_norm": 2.1656930446624756,
      "learning_rate": 3.169764409765716e-06,
      "loss": 0.4932,
      "step": 203300
    },
    {
      "epoch": 83.22,
      "grad_norm": 1.847891092300415,
      "learning_rate": 3.1696159564538563e-06,
      "loss": 0.4949,
      "step": 203310
    },
    {
      "epoch": 83.23,
      "grad_norm": 2.303614854812622,
      "learning_rate": 3.1694674993588322e-06,
      "loss": 0.5013,
      "step": 203320
    },
    {
      "epoch": 83.23,
      "grad_norm": 2.3146791458129883,
      "learning_rate": 3.169319038481324e-06,
      "loss": 0.5014,
      "step": 203330
    },
    {
      "epoch": 83.23,
      "grad_norm": 1.8355463743209839,
      "learning_rate": 3.169170573822012e-06,
      "loss": 0.4826,
      "step": 203340
    },
    {
      "epoch": 83.24,
      "grad_norm": 2.1261086463928223,
      "learning_rate": 3.169022105381576e-06,
      "loss": 0.493,
      "step": 203350
    },
    {
      "epoch": 83.24,
      "grad_norm": 2.035468578338623,
      "learning_rate": 3.1688736331606964e-06,
      "loss": 0.4818,
      "step": 203360
    },
    {
      "epoch": 83.25,
      "grad_norm": 1.8298993110656738,
      "learning_rate": 3.168725157160053e-06,
      "loss": 0.5147,
      "step": 203370
    },
    {
      "epoch": 83.25,
      "grad_norm": 2.023313283920288,
      "learning_rate": 3.1685766773803267e-06,
      "loss": 0.4958,
      "step": 203380
    },
    {
      "epoch": 83.25,
      "grad_norm": 1.9704749584197998,
      "learning_rate": 3.1684281938221972e-06,
      "loss": 0.4925,
      "step": 203390
    },
    {
      "epoch": 83.26,
      "grad_norm": 1.794367790222168,
      "learning_rate": 3.168279706486344e-06,
      "loss": 0.4963,
      "step": 203400
    },
    {
      "epoch": 83.26,
      "grad_norm": 1.5823036432266235,
      "learning_rate": 3.1681312153734484e-06,
      "loss": 0.4942,
      "step": 203410
    },
    {
      "epoch": 83.27,
      "grad_norm": 1.6332944631576538,
      "learning_rate": 3.16798272048419e-06,
      "loss": 0.4622,
      "step": 203420
    },
    {
      "epoch": 83.27,
      "grad_norm": 1.6361188888549805,
      "learning_rate": 3.1678342218192485e-06,
      "loss": 0.5085,
      "step": 203430
    },
    {
      "epoch": 83.27,
      "grad_norm": 1.8920702934265137,
      "learning_rate": 3.1676857193793057e-06,
      "loss": 0.5066,
      "step": 203440
    },
    {
      "epoch": 83.28,
      "grad_norm": 1.7711384296417236,
      "learning_rate": 3.1675372131650405e-06,
      "loss": 0.5014,
      "step": 203450
    },
    {
      "epoch": 83.28,
      "grad_norm": 1.8466967344284058,
      "learning_rate": 3.1673887031771346e-06,
      "loss": 0.4837,
      "step": 203460
    },
    {
      "epoch": 83.29,
      "grad_norm": 2.151886463165283,
      "learning_rate": 3.1672401894162654e-06,
      "loss": 0.5061,
      "step": 203470
    },
    {
      "epoch": 83.29,
      "grad_norm": 1.784916877746582,
      "learning_rate": 3.1670916718831166e-06,
      "loss": 0.4796,
      "step": 203480
    },
    {
      "epoch": 83.3,
      "grad_norm": 2.2564122676849365,
      "learning_rate": 3.1669431505783665e-06,
      "loss": 0.5239,
      "step": 203490
    },
    {
      "epoch": 83.3,
      "grad_norm": 2.1174983978271484,
      "learning_rate": 3.1667946255026965e-06,
      "loss": 0.4907,
      "step": 203500
    },
    {
      "epoch": 83.3,
      "grad_norm": 1.6524981260299683,
      "learning_rate": 3.1666460966567866e-06,
      "loss": 0.4882,
      "step": 203510
    },
    {
      "epoch": 83.31,
      "grad_norm": 1.7026363611221313,
      "learning_rate": 3.1664975640413176e-06,
      "loss": 0.5113,
      "step": 203520
    },
    {
      "epoch": 83.31,
      "grad_norm": 1.743910312652588,
      "learning_rate": 3.1663490276569686e-06,
      "loss": 0.5086,
      "step": 203530
    },
    {
      "epoch": 83.32,
      "grad_norm": 2.486585855484009,
      "learning_rate": 3.1662004875044213e-06,
      "loss": 0.4913,
      "step": 203540
    },
    {
      "epoch": 83.32,
      "grad_norm": 1.704598069190979,
      "learning_rate": 3.1660519435843552e-06,
      "loss": 0.4929,
      "step": 203550
    },
    {
      "epoch": 83.32,
      "grad_norm": 1.7994592189788818,
      "learning_rate": 3.1659033958974515e-06,
      "loss": 0.501,
      "step": 203560
    },
    {
      "epoch": 83.33,
      "grad_norm": 1.7572667598724365,
      "learning_rate": 3.16575484444439e-06,
      "loss": 0.5132,
      "step": 203570
    },
    {
      "epoch": 83.33,
      "grad_norm": 1.48175048828125,
      "learning_rate": 3.1656062892258525e-06,
      "loss": 0.5072,
      "step": 203580
    },
    {
      "epoch": 83.34,
      "grad_norm": 1.740403413772583,
      "learning_rate": 3.1654577302425184e-06,
      "loss": 0.4989,
      "step": 203590
    },
    {
      "epoch": 83.34,
      "grad_norm": 1.719957709312439,
      "learning_rate": 3.165309167495068e-06,
      "loss": 0.5086,
      "step": 203600
    },
    {
      "epoch": 83.34,
      "grad_norm": 1.659398078918457,
      "learning_rate": 3.165160600984182e-06,
      "loss": 0.4928,
      "step": 203610
    },
    {
      "epoch": 83.35,
      "grad_norm": 1.9642064571380615,
      "learning_rate": 3.1650120307105416e-06,
      "loss": 0.5034,
      "step": 203620
    },
    {
      "epoch": 83.35,
      "grad_norm": 1.8802454471588135,
      "learning_rate": 3.1648634566748273e-06,
      "loss": 0.4997,
      "step": 203630
    },
    {
      "epoch": 83.36,
      "grad_norm": 2.3952231407165527,
      "learning_rate": 3.1647148788777196e-06,
      "loss": 0.5136,
      "step": 203640
    },
    {
      "epoch": 83.36,
      "grad_norm": 1.9289753437042236,
      "learning_rate": 3.1645662973198983e-06,
      "loss": 0.5024,
      "step": 203650
    },
    {
      "epoch": 83.36,
      "grad_norm": 2.025423049926758,
      "learning_rate": 3.164417712002045e-06,
      "loss": 0.4875,
      "step": 203660
    },
    {
      "epoch": 83.37,
      "grad_norm": 1.9932957887649536,
      "learning_rate": 3.1642691229248396e-06,
      "loss": 0.4995,
      "step": 203670
    },
    {
      "epoch": 83.37,
      "grad_norm": 2.5909652709960938,
      "learning_rate": 3.164120530088964e-06,
      "loss": 0.5009,
      "step": 203680
    },
    {
      "epoch": 83.38,
      "grad_norm": 2.3846023082733154,
      "learning_rate": 3.163971933495097e-06,
      "loss": 0.4901,
      "step": 203690
    },
    {
      "epoch": 83.38,
      "grad_norm": 2.0577638149261475,
      "learning_rate": 3.1638233331439214e-06,
      "loss": 0.5025,
      "step": 203700
    },
    {
      "epoch": 83.39,
      "grad_norm": 1.605882167816162,
      "learning_rate": 3.1636747290361162e-06,
      "loss": 0.5004,
      "step": 203710
    },
    {
      "epoch": 83.39,
      "grad_norm": 1.632849097251892,
      "learning_rate": 3.1635261211723632e-06,
      "loss": 0.5091,
      "step": 203720
    },
    {
      "epoch": 83.39,
      "grad_norm": 2.23576021194458,
      "learning_rate": 3.163377509553342e-06,
      "loss": 0.4889,
      "step": 203730
    },
    {
      "epoch": 83.4,
      "grad_norm": 1.5415852069854736,
      "learning_rate": 3.1632288941797354e-06,
      "loss": 0.4798,
      "step": 203740
    },
    {
      "epoch": 83.4,
      "grad_norm": 1.9420039653778076,
      "learning_rate": 3.1630802750522225e-06,
      "loss": 0.5145,
      "step": 203750
    },
    {
      "epoch": 83.41,
      "grad_norm": 1.8326737880706787,
      "learning_rate": 3.162931652171484e-06,
      "loss": 0.5007,
      "step": 203760
    },
    {
      "epoch": 83.41,
      "grad_norm": 1.8372315168380737,
      "learning_rate": 3.1627830255382007e-06,
      "loss": 0.4994,
      "step": 203770
    },
    {
      "epoch": 83.41,
      "grad_norm": 1.9016417264938354,
      "learning_rate": 3.1626343951530546e-06,
      "loss": 0.5002,
      "step": 203780
    },
    {
      "epoch": 83.42,
      "grad_norm": 1.9718233346939087,
      "learning_rate": 3.162485761016726e-06,
      "loss": 0.4927,
      "step": 203790
    },
    {
      "epoch": 83.42,
      "grad_norm": 2.218667507171631,
      "learning_rate": 3.1623371231298953e-06,
      "loss": 0.4998,
      "step": 203800
    },
    {
      "epoch": 83.43,
      "grad_norm": 1.6577725410461426,
      "learning_rate": 3.1621884814932447e-06,
      "loss": 0.5004,
      "step": 203810
    },
    {
      "epoch": 83.43,
      "grad_norm": 1.787339448928833,
      "learning_rate": 3.1620398361074534e-06,
      "loss": 0.4825,
      "step": 203820
    },
    {
      "epoch": 83.43,
      "grad_norm": 1.849755048751831,
      "learning_rate": 3.1618911869732038e-06,
      "loss": 0.4935,
      "step": 203830
    },
    {
      "epoch": 83.44,
      "grad_norm": 1.8042160272598267,
      "learning_rate": 3.1617425340911747e-06,
      "loss": 0.5186,
      "step": 203840
    },
    {
      "epoch": 83.44,
      "grad_norm": 1.6451481580734253,
      "learning_rate": 3.1615938774620493e-06,
      "loss": 0.4949,
      "step": 203850
    },
    {
      "epoch": 83.45,
      "grad_norm": 2.2485461235046387,
      "learning_rate": 3.161445217086508e-06,
      "loss": 0.4995,
      "step": 203860
    },
    {
      "epoch": 83.45,
      "grad_norm": 1.8459347486495972,
      "learning_rate": 3.1612965529652308e-06,
      "loss": 0.5183,
      "step": 203870
    },
    {
      "epoch": 83.45,
      "grad_norm": 1.892229676246643,
      "learning_rate": 3.161147885098901e-06,
      "loss": 0.4938,
      "step": 203880
    },
    {
      "epoch": 83.46,
      "grad_norm": 2.09104585647583,
      "learning_rate": 3.160999213488197e-06,
      "loss": 0.4877,
      "step": 203890
    },
    {
      "epoch": 83.46,
      "grad_norm": 1.9946893453598022,
      "learning_rate": 3.1608505381338007e-06,
      "loss": 0.4889,
      "step": 203900
    },
    {
      "epoch": 83.47,
      "grad_norm": 1.7090870141983032,
      "learning_rate": 3.1607018590363938e-06,
      "loss": 0.5,
      "step": 203910
    },
    {
      "epoch": 83.47,
      "grad_norm": 1.566109299659729,
      "learning_rate": 3.160553176196657e-06,
      "loss": 0.5032,
      "step": 203920
    },
    {
      "epoch": 83.48,
      "grad_norm": 1.8723456859588623,
      "learning_rate": 3.1604044896152707e-06,
      "loss": 0.4996,
      "step": 203930
    },
    {
      "epoch": 83.48,
      "grad_norm": 1.6126645803451538,
      "learning_rate": 3.160255799292917e-06,
      "loss": 0.4934,
      "step": 203940
    },
    {
      "epoch": 83.48,
      "grad_norm": 1.9112712144851685,
      "learning_rate": 3.1601071052302768e-06,
      "loss": 0.5066,
      "step": 203950
    },
    {
      "epoch": 83.49,
      "grad_norm": 1.882742166519165,
      "learning_rate": 3.1599584074280314e-06,
      "loss": 0.4746,
      "step": 203960
    },
    {
      "epoch": 83.49,
      "grad_norm": 1.7458058595657349,
      "learning_rate": 3.1598097058868615e-06,
      "loss": 0.4764,
      "step": 203970
    },
    {
      "epoch": 83.5,
      "grad_norm": 2.5297160148620605,
      "learning_rate": 3.1596610006074484e-06,
      "loss": 0.5042,
      "step": 203980
    },
    {
      "epoch": 83.5,
      "grad_norm": 1.8223196268081665,
      "learning_rate": 3.1595122915904732e-06,
      "loss": 0.4963,
      "step": 203990
    },
    {
      "epoch": 83.5,
      "grad_norm": 1.7982609272003174,
      "learning_rate": 3.159363578836618e-06,
      "loss": 0.4888,
      "step": 204000
    },
    {
      "epoch": 83.51,
      "grad_norm": 1.968431830406189,
      "learning_rate": 3.1592148623465624e-06,
      "loss": 0.4881,
      "step": 204010
    },
    {
      "epoch": 83.51,
      "grad_norm": 1.894088864326477,
      "learning_rate": 3.1590661421209887e-06,
      "loss": 0.5139,
      "step": 204020
    },
    {
      "epoch": 83.52,
      "grad_norm": 2.292787790298462,
      "learning_rate": 3.1589174181605786e-06,
      "loss": 0.4797,
      "step": 204030
    },
    {
      "epoch": 83.52,
      "grad_norm": 1.682806372642517,
      "learning_rate": 3.158768690466012e-06,
      "loss": 0.4978,
      "step": 204040
    },
    {
      "epoch": 83.52,
      "grad_norm": 1.7235493659973145,
      "learning_rate": 3.158619959037972e-06,
      "loss": 0.5024,
      "step": 204050
    },
    {
      "epoch": 83.53,
      "grad_norm": 1.796776533126831,
      "learning_rate": 3.1584712238771376e-06,
      "loss": 0.4987,
      "step": 204060
    },
    {
      "epoch": 83.53,
      "grad_norm": 1.4335523843765259,
      "learning_rate": 3.1583224849841923e-06,
      "loss": 0.5021,
      "step": 204070
    },
    {
      "epoch": 83.54,
      "grad_norm": 2.6669535636901855,
      "learning_rate": 3.1581737423598166e-06,
      "loss": 0.5126,
      "step": 204080
    },
    {
      "epoch": 83.54,
      "grad_norm": 1.9249187707901,
      "learning_rate": 3.158024996004692e-06,
      "loss": 0.4986,
      "step": 204090
    },
    {
      "epoch": 83.54,
      "grad_norm": 2.5012588500976562,
      "learning_rate": 3.1578762459195e-06,
      "loss": 0.498,
      "step": 204100
    },
    {
      "epoch": 83.55,
      "grad_norm": 1.723355770111084,
      "learning_rate": 3.157727492104921e-06,
      "loss": 0.4938,
      "step": 204110
    },
    {
      "epoch": 83.55,
      "grad_norm": 1.7816076278686523,
      "learning_rate": 3.1575787345616377e-06,
      "loss": 0.4931,
      "step": 204120
    },
    {
      "epoch": 83.56,
      "grad_norm": 2.062750816345215,
      "learning_rate": 3.15742997329033e-06,
      "loss": 0.4961,
      "step": 204130
    },
    {
      "epoch": 83.56,
      "grad_norm": 1.8992129564285278,
      "learning_rate": 3.157281208291681e-06,
      "loss": 0.5057,
      "step": 204140
    },
    {
      "epoch": 83.57,
      "grad_norm": 1.7045449018478394,
      "learning_rate": 3.1571324395663714e-06,
      "loss": 0.5304,
      "step": 204150
    },
    {
      "epoch": 83.57,
      "grad_norm": 1.631683111190796,
      "learning_rate": 3.156983667115083e-06,
      "loss": 0.4811,
      "step": 204160
    },
    {
      "epoch": 83.57,
      "grad_norm": 1.5699416399002075,
      "learning_rate": 3.156834890938498e-06,
      "loss": 0.5004,
      "step": 204170
    },
    {
      "epoch": 83.58,
      "grad_norm": 2.4557087421417236,
      "learning_rate": 3.1566861110372965e-06,
      "loss": 0.5366,
      "step": 204180
    },
    {
      "epoch": 83.58,
      "grad_norm": 1.5014909505844116,
      "learning_rate": 3.15653732741216e-06,
      "loss": 0.4944,
      "step": 204190
    },
    {
      "epoch": 83.59,
      "grad_norm": 1.6372251510620117,
      "learning_rate": 3.156388540063771e-06,
      "loss": 0.5103,
      "step": 204200
    },
    {
      "epoch": 83.59,
      "grad_norm": 1.8342735767364502,
      "learning_rate": 3.1562397489928104e-06,
      "loss": 0.4782,
      "step": 204210
    },
    {
      "epoch": 83.59,
      "grad_norm": 2.169307231903076,
      "learning_rate": 3.1560909541999605e-06,
      "loss": 0.5,
      "step": 204220
    },
    {
      "epoch": 83.6,
      "grad_norm": 2.310957431793213,
      "learning_rate": 3.155942155685902e-06,
      "loss": 0.5142,
      "step": 204230
    },
    {
      "epoch": 83.6,
      "grad_norm": 1.5206985473632812,
      "learning_rate": 3.155793353451317e-06,
      "loss": 0.4875,
      "step": 204240
    },
    {
      "epoch": 83.61,
      "grad_norm": 2.018141031265259,
      "learning_rate": 3.1556445474968876e-06,
      "loss": 0.4895,
      "step": 204250
    },
    {
      "epoch": 83.61,
      "grad_norm": 2.083730459213257,
      "learning_rate": 3.1554957378232946e-06,
      "loss": 0.5017,
      "step": 204260
    },
    {
      "epoch": 83.61,
      "grad_norm": 1.746139645576477,
      "learning_rate": 3.1553469244312207e-06,
      "loss": 0.5075,
      "step": 204270
    },
    {
      "epoch": 83.62,
      "grad_norm": 2.3660848140716553,
      "learning_rate": 3.1551981073213464e-06,
      "loss": 0.5052,
      "step": 204280
    },
    {
      "epoch": 83.62,
      "grad_norm": 2.1109774112701416,
      "learning_rate": 3.1550492864943544e-06,
      "loss": 0.515,
      "step": 204290
    },
    {
      "epoch": 83.63,
      "grad_norm": 1.9844403266906738,
      "learning_rate": 3.1549004619509257e-06,
      "loss": 0.5059,
      "step": 204300
    },
    {
      "epoch": 83.63,
      "grad_norm": 2.217007637023926,
      "learning_rate": 3.1547516336917426e-06,
      "loss": 0.5012,
      "step": 204310
    },
    {
      "epoch": 83.63,
      "grad_norm": 1.7987463474273682,
      "learning_rate": 3.154602801717486e-06,
      "loss": 0.5032,
      "step": 204320
    },
    {
      "epoch": 83.64,
      "grad_norm": 1.6715643405914307,
      "learning_rate": 3.154453966028839e-06,
      "loss": 0.4997,
      "step": 204330
    },
    {
      "epoch": 83.64,
      "grad_norm": 1.8143904209136963,
      "learning_rate": 3.1543051266264827e-06,
      "loss": 0.5009,
      "step": 204340
    },
    {
      "epoch": 83.65,
      "grad_norm": 1.7323566675186157,
      "learning_rate": 3.1541562835110986e-06,
      "loss": 0.5028,
      "step": 204350
    },
    {
      "epoch": 83.65,
      "grad_norm": 1.613097071647644,
      "learning_rate": 3.1540074366833693e-06,
      "loss": 0.4938,
      "step": 204360
    },
    {
      "epoch": 83.66,
      "grad_norm": 1.8493856191635132,
      "learning_rate": 3.1538585861439756e-06,
      "loss": 0.5327,
      "step": 204370
    },
    {
      "epoch": 83.66,
      "grad_norm": 1.4877327680587769,
      "learning_rate": 3.1537097318936e-06,
      "loss": 0.5014,
      "step": 204380
    },
    {
      "epoch": 83.66,
      "grad_norm": 1.7724376916885376,
      "learning_rate": 3.153560873932925e-06,
      "loss": 0.5112,
      "step": 204390
    },
    {
      "epoch": 83.67,
      "grad_norm": 1.7853295803070068,
      "learning_rate": 3.153412012262631e-06,
      "loss": 0.5069,
      "step": 204400
    },
    {
      "epoch": 83.67,
      "grad_norm": 2.0992538928985596,
      "learning_rate": 3.153263146883401e-06,
      "loss": 0.4961,
      "step": 204410
    },
    {
      "epoch": 83.68,
      "grad_norm": 1.6650161743164062,
      "learning_rate": 3.153114277795916e-06,
      "loss": 0.4735,
      "step": 204420
    },
    {
      "epoch": 83.68,
      "grad_norm": 1.8327687978744507,
      "learning_rate": 3.152965405000859e-06,
      "loss": 0.5188,
      "step": 204430
    },
    {
      "epoch": 83.68,
      "grad_norm": 1.784430980682373,
      "learning_rate": 3.1528165284989117e-06,
      "loss": 0.5058,
      "step": 204440
    },
    {
      "epoch": 83.69,
      "grad_norm": 2.0874667167663574,
      "learning_rate": 3.152667648290756e-06,
      "loss": 0.4962,
      "step": 204450
    },
    {
      "epoch": 83.69,
      "grad_norm": 1.5263237953186035,
      "learning_rate": 3.1525187643770743e-06,
      "loss": 0.5038,
      "step": 204460
    },
    {
      "epoch": 83.7,
      "grad_norm": 1.8001827001571655,
      "learning_rate": 3.1523698767585473e-06,
      "loss": 0.5128,
      "step": 204470
    },
    {
      "epoch": 83.7,
      "grad_norm": 2.230647325515747,
      "learning_rate": 3.152220985435859e-06,
      "loss": 0.4797,
      "step": 204480
    },
    {
      "epoch": 83.7,
      "grad_norm": 2.189035177230835,
      "learning_rate": 3.152072090409689e-06,
      "loss": 0.5018,
      "step": 204490
    },
    {
      "epoch": 83.71,
      "grad_norm": 1.9274895191192627,
      "learning_rate": 3.151923191680721e-06,
      "loss": 0.4951,
      "step": 204500
    },
    {
      "epoch": 83.71,
      "grad_norm": 2.424434185028076,
      "learning_rate": 3.151774289249637e-06,
      "loss": 0.5245,
      "step": 204510
    },
    {
      "epoch": 83.72,
      "grad_norm": 2.062089681625366,
      "learning_rate": 3.1516253831171183e-06,
      "loss": 0.5119,
      "step": 204520
    },
    {
      "epoch": 83.72,
      "grad_norm": 1.7715399265289307,
      "learning_rate": 3.1514764732838487e-06,
      "loss": 0.5145,
      "step": 204530
    },
    {
      "epoch": 83.72,
      "grad_norm": 2.011338472366333,
      "learning_rate": 3.151327559750509e-06,
      "loss": 0.5146,
      "step": 204540
    },
    {
      "epoch": 83.73,
      "grad_norm": 1.6166640520095825,
      "learning_rate": 3.151178642517781e-06,
      "loss": 0.5249,
      "step": 204550
    },
    {
      "epoch": 83.73,
      "grad_norm": 1.913821816444397,
      "learning_rate": 3.1510297215863477e-06,
      "loss": 0.4921,
      "step": 204560
    },
    {
      "epoch": 83.74,
      "grad_norm": 2.3722119331359863,
      "learning_rate": 3.150880796956891e-06,
      "loss": 0.5,
      "step": 204570
    },
    {
      "epoch": 83.74,
      "grad_norm": 2.1647307872772217,
      "learning_rate": 3.150731868630093e-06,
      "loss": 0.4868,
      "step": 204580
    },
    {
      "epoch": 83.75,
      "grad_norm": 1.9300670623779297,
      "learning_rate": 3.150582936606636e-06,
      "loss": 0.4939,
      "step": 204590
    },
    {
      "epoch": 83.75,
      "grad_norm": 2.1819868087768555,
      "learning_rate": 3.1504340008872027e-06,
      "loss": 0.4956,
      "step": 204600
    },
    {
      "epoch": 83.75,
      "grad_norm": 2.178616523742676,
      "learning_rate": 3.150285061472474e-06,
      "loss": 0.4957,
      "step": 204610
    },
    {
      "epoch": 83.76,
      "grad_norm": 1.8943229913711548,
      "learning_rate": 3.150136118363134e-06,
      "loss": 0.5188,
      "step": 204620
    },
    {
      "epoch": 83.76,
      "grad_norm": 2.109405994415283,
      "learning_rate": 3.149987171559864e-06,
      "loss": 0.4892,
      "step": 204630
    },
    {
      "epoch": 83.77,
      "grad_norm": 1.4497190713882446,
      "learning_rate": 3.149838221063346e-06,
      "loss": 0.4956,
      "step": 204640
    },
    {
      "epoch": 83.77,
      "grad_norm": 1.751064658164978,
      "learning_rate": 3.149689266874263e-06,
      "loss": 0.4809,
      "step": 204650
    },
    {
      "epoch": 83.77,
      "grad_norm": 1.932133674621582,
      "learning_rate": 3.1495403089932967e-06,
      "loss": 0.4826,
      "step": 204660
    },
    {
      "epoch": 83.78,
      "grad_norm": 2.2314438819885254,
      "learning_rate": 3.14939134742113e-06,
      "loss": 0.4984,
      "step": 204670
    },
    {
      "epoch": 83.78,
      "grad_norm": 2.103951930999756,
      "learning_rate": 3.149242382158445e-06,
      "loss": 0.5132,
      "step": 204680
    },
    {
      "epoch": 83.79,
      "grad_norm": 2.1831986904144287,
      "learning_rate": 3.149093413205925e-06,
      "loss": 0.5324,
      "step": 204690
    },
    {
      "epoch": 83.79,
      "grad_norm": 1.6264851093292236,
      "learning_rate": 3.14894444056425e-06,
      "loss": 0.4931,
      "step": 204700
    },
    {
      "epoch": 83.79,
      "grad_norm": 1.900937557220459,
      "learning_rate": 3.1487954642341045e-06,
      "loss": 0.4913,
      "step": 204710
    },
    {
      "epoch": 83.8,
      "grad_norm": 1.9058254957199097,
      "learning_rate": 3.1486464842161703e-06,
      "loss": 0.4963,
      "step": 204720
    },
    {
      "epoch": 83.8,
      "grad_norm": 2.430647611618042,
      "learning_rate": 3.1484975005111304e-06,
      "loss": 0.5162,
      "step": 204730
    },
    {
      "epoch": 83.81,
      "grad_norm": 2.286897659301758,
      "learning_rate": 3.148348513119667e-06,
      "loss": 0.4908,
      "step": 204740
    },
    {
      "epoch": 83.81,
      "grad_norm": 2.26255464553833,
      "learning_rate": 3.1481995220424626e-06,
      "loss": 0.4983,
      "step": 204750
    },
    {
      "epoch": 83.81,
      "grad_norm": 2.042579412460327,
      "learning_rate": 3.1480505272801987e-06,
      "loss": 0.4881,
      "step": 204760
    },
    {
      "epoch": 83.82,
      "grad_norm": 2.101484775543213,
      "learning_rate": 3.147901528833559e-06,
      "loss": 0.4882,
      "step": 204770
    },
    {
      "epoch": 83.82,
      "grad_norm": 1.743310809135437,
      "learning_rate": 3.147752526703226e-06,
      "loss": 0.5184,
      "step": 204780
    },
    {
      "epoch": 83.83,
      "grad_norm": 1.8014532327651978,
      "learning_rate": 3.1476035208898807e-06,
      "loss": 0.5029,
      "step": 204790
    },
    {
      "epoch": 83.83,
      "grad_norm": 2.0433554649353027,
      "learning_rate": 3.147454511394208e-06,
      "loss": 0.4871,
      "step": 204800
    },
    {
      "epoch": 83.84,
      "grad_norm": 1.930367350578308,
      "learning_rate": 3.1473054982168893e-06,
      "loss": 0.5034,
      "step": 204810
    },
    {
      "epoch": 83.84,
      "grad_norm": 1.907996416091919,
      "learning_rate": 3.1471564813586076e-06,
      "loss": 0.4733,
      "step": 204820
    },
    {
      "epoch": 83.84,
      "grad_norm": 1.5712512731552124,
      "learning_rate": 3.1470074608200444e-06,
      "loss": 0.4927,
      "step": 204830
    },
    {
      "epoch": 83.85,
      "grad_norm": 1.4444596767425537,
      "learning_rate": 3.146858436601884e-06,
      "loss": 0.4926,
      "step": 204840
    },
    {
      "epoch": 83.85,
      "grad_norm": 1.6839135885238647,
      "learning_rate": 3.146709408704808e-06,
      "loss": 0.5135,
      "step": 204850
    },
    {
      "epoch": 83.86,
      "grad_norm": 2.0203733444213867,
      "learning_rate": 3.146560377129499e-06,
      "loss": 0.4881,
      "step": 204860
    },
    {
      "epoch": 83.86,
      "grad_norm": 2.035872459411621,
      "learning_rate": 3.1464113418766406e-06,
      "loss": 0.5206,
      "step": 204870
    },
    {
      "epoch": 83.86,
      "grad_norm": 1.9191030263900757,
      "learning_rate": 3.1462623029469143e-06,
      "loss": 0.5061,
      "step": 204880
    },
    {
      "epoch": 83.87,
      "grad_norm": 2.062438726425171,
      "learning_rate": 3.1461132603410037e-06,
      "loss": 0.4953,
      "step": 204890
    },
    {
      "epoch": 83.87,
      "grad_norm": 1.9076975584030151,
      "learning_rate": 3.145964214059591e-06,
      "loss": 0.5318,
      "step": 204900
    },
    {
      "epoch": 83.88,
      "grad_norm": 1.708517074584961,
      "learning_rate": 3.145815164103359e-06,
      "loss": 0.5142,
      "step": 204910
    },
    {
      "epoch": 83.88,
      "grad_norm": 1.8009346723556519,
      "learning_rate": 3.145666110472991e-06,
      "loss": 0.4847,
      "step": 204920
    },
    {
      "epoch": 83.88,
      "grad_norm": 1.678213119506836,
      "learning_rate": 3.1455170531691695e-06,
      "loss": 0.5111,
      "step": 204930
    },
    {
      "epoch": 83.89,
      "grad_norm": 2.2316133975982666,
      "learning_rate": 3.1453679921925775e-06,
      "loss": 0.5299,
      "step": 204940
    },
    {
      "epoch": 83.89,
      "grad_norm": 1.671946406364441,
      "learning_rate": 3.1452189275438978e-06,
      "loss": 0.521,
      "step": 204950
    },
    {
      "epoch": 83.9,
      "grad_norm": 2.1021671295166016,
      "learning_rate": 3.1450698592238126e-06,
      "loss": 0.4847,
      "step": 204960
    },
    {
      "epoch": 83.9,
      "grad_norm": 1.7305102348327637,
      "learning_rate": 3.144920787233006e-06,
      "loss": 0.5253,
      "step": 204970
    },
    {
      "epoch": 83.91,
      "grad_norm": 2.618969202041626,
      "learning_rate": 3.144771711572159e-06,
      "loss": 0.4897,
      "step": 204980
    },
    {
      "epoch": 83.91,
      "grad_norm": 1.9276448488235474,
      "learning_rate": 3.1446226322419557e-06,
      "loss": 0.5128,
      "step": 204990
    },
    {
      "epoch": 83.91,
      "grad_norm": 2.049548625946045,
      "learning_rate": 3.1444735492430796e-06,
      "loss": 0.5244,
      "step": 205000
    },
    {
      "epoch": 83.92,
      "grad_norm": 2.0243313312530518,
      "learning_rate": 3.1443244625762125e-06,
      "loss": 0.4926,
      "step": 205010
    },
    {
      "epoch": 83.92,
      "grad_norm": 1.5319606065750122,
      "learning_rate": 3.144175372242038e-06,
      "loss": 0.4911,
      "step": 205020
    },
    {
      "epoch": 83.93,
      "grad_norm": 2.603828191757202,
      "learning_rate": 3.1440262782412385e-06,
      "loss": 0.5129,
      "step": 205030
    },
    {
      "epoch": 83.93,
      "grad_norm": 1.9079723358154297,
      "learning_rate": 3.143877180574498e-06,
      "loss": 0.4975,
      "step": 205040
    },
    {
      "epoch": 83.93,
      "grad_norm": 2.0405964851379395,
      "learning_rate": 3.1437280792424983e-06,
      "loss": 0.5019,
      "step": 205050
    },
    {
      "epoch": 83.94,
      "grad_norm": 1.6909737586975098,
      "learning_rate": 3.1435789742459225e-06,
      "loss": 0.5044,
      "step": 205060
    },
    {
      "epoch": 83.94,
      "grad_norm": 1.6422210931777954,
      "learning_rate": 3.143429865585455e-06,
      "loss": 0.5092,
      "step": 205070
    },
    {
      "epoch": 83.95,
      "grad_norm": 1.7227530479431152,
      "learning_rate": 3.143280753261777e-06,
      "loss": 0.5202,
      "step": 205080
    },
    {
      "epoch": 83.95,
      "grad_norm": 2.5520179271698,
      "learning_rate": 3.1431316372755723e-06,
      "loss": 0.5101,
      "step": 205090
    },
    {
      "epoch": 83.95,
      "grad_norm": 1.7396951913833618,
      "learning_rate": 3.1429825176275246e-06,
      "loss": 0.4897,
      "step": 205100
    },
    {
      "epoch": 83.96,
      "grad_norm": 1.6474336385726929,
      "learning_rate": 3.1428333943183175e-06,
      "loss": 0.503,
      "step": 205110
    },
    {
      "epoch": 83.96,
      "grad_norm": 2.19728422164917,
      "learning_rate": 3.1426842673486318e-06,
      "loss": 0.4769,
      "step": 205120
    },
    {
      "epoch": 83.97,
      "grad_norm": 1.7190839052200317,
      "learning_rate": 3.142535136719152e-06,
      "loss": 0.4856,
      "step": 205130
    },
    {
      "epoch": 83.97,
      "grad_norm": 2.0329134464263916,
      "learning_rate": 3.1423860024305617e-06,
      "loss": 0.5298,
      "step": 205140
    },
    {
      "epoch": 83.97,
      "grad_norm": 2.211594820022583,
      "learning_rate": 3.1422368644835432e-06,
      "loss": 0.4904,
      "step": 205150
    },
    {
      "epoch": 83.98,
      "grad_norm": 1.8869675397872925,
      "learning_rate": 3.1420877228787803e-06,
      "loss": 0.4868,
      "step": 205160
    },
    {
      "epoch": 83.98,
      "grad_norm": 1.7767953872680664,
      "learning_rate": 3.141938577616955e-06,
      "loss": 0.5141,
      "step": 205170
    },
    {
      "epoch": 83.99,
      "grad_norm": 1.6205521821975708,
      "learning_rate": 3.1417894286987534e-06,
      "loss": 0.5061,
      "step": 205180
    },
    {
      "epoch": 83.99,
      "grad_norm": 1.9813752174377441,
      "learning_rate": 3.1416402761248553e-06,
      "loss": 0.4885,
      "step": 205190
    },
    {
      "epoch": 84.0,
      "grad_norm": 2.198514461517334,
      "learning_rate": 3.1414911198959457e-06,
      "loss": 0.4962,
      "step": 205200
    },
    {
      "epoch": 84.0,
      "grad_norm": 1.790420651435852,
      "learning_rate": 3.1413419600127077e-06,
      "loss": 0.5191,
      "step": 205210
    },
    {
      "epoch": 84.0,
      "eval_loss": 0.5009229183197021,
      "eval_runtime": 76.139,
      "eval_samples_per_second": 45.299,
      "eval_steps_per_second": 5.674,
      "step": 205212
    },
    {
      "epoch": 84.0,
      "grad_norm": 1.9927703142166138,
      "learning_rate": 3.1411927964758244e-06,
      "loss": 0.4715,
      "step": 205220
    },
    {
      "epoch": 84.01,
      "grad_norm": 1.910972237586975,
      "learning_rate": 3.141043629285979e-06,
      "loss": 0.4937,
      "step": 205230
    },
    {
      "epoch": 84.01,
      "grad_norm": 1.7899640798568726,
      "learning_rate": 3.1408944584438553e-06,
      "loss": 0.5075,
      "step": 205240
    },
    {
      "epoch": 84.02,
      "grad_norm": 2.0638952255249023,
      "learning_rate": 3.1407452839501362e-06,
      "loss": 0.4927,
      "step": 205250
    },
    {
      "epoch": 84.02,
      "grad_norm": 2.3735179901123047,
      "learning_rate": 3.1405961058055053e-06,
      "loss": 0.4946,
      "step": 205260
    },
    {
      "epoch": 84.02,
      "grad_norm": 1.7460289001464844,
      "learning_rate": 3.140446924010645e-06,
      "loss": 0.502,
      "step": 205270
    },
    {
      "epoch": 84.03,
      "grad_norm": 1.894078254699707,
      "learning_rate": 3.1402977385662397e-06,
      "loss": 0.5043,
      "step": 205280
    },
    {
      "epoch": 84.03,
      "grad_norm": 2.0339107513427734,
      "learning_rate": 3.140148549472973e-06,
      "loss": 0.5015,
      "step": 205290
    },
    {
      "epoch": 84.04,
      "grad_norm": 1.9132559299468994,
      "learning_rate": 3.1399993567315275e-06,
      "loss": 0.471,
      "step": 205300
    },
    {
      "epoch": 84.04,
      "grad_norm": 1.59599769115448,
      "learning_rate": 3.1398501603425874e-06,
      "loss": 0.4937,
      "step": 205310
    },
    {
      "epoch": 84.04,
      "grad_norm": 1.7938928604125977,
      "learning_rate": 3.1397009603068354e-06,
      "loss": 0.5045,
      "step": 205320
    },
    {
      "epoch": 84.05,
      "grad_norm": 2.131988286972046,
      "learning_rate": 3.139551756624956e-06,
      "loss": 0.4961,
      "step": 205330
    },
    {
      "epoch": 84.05,
      "grad_norm": 1.7506182193756104,
      "learning_rate": 3.139402549297632e-06,
      "loss": 0.503,
      "step": 205340
    },
    {
      "epoch": 84.06,
      "grad_norm": 1.6138968467712402,
      "learning_rate": 3.1392533383255458e-06,
      "loss": 0.504,
      "step": 205350
    },
    {
      "epoch": 84.06,
      "grad_norm": 2.112734079360962,
      "learning_rate": 3.1391041237093824e-06,
      "loss": 0.4968,
      "step": 205360
    },
    {
      "epoch": 84.06,
      "grad_norm": 1.6554621458053589,
      "learning_rate": 3.138954905449825e-06,
      "loss": 0.5021,
      "step": 205370
    },
    {
      "epoch": 84.07,
      "grad_norm": 2.204376697540283,
      "learning_rate": 3.1388056835475576e-06,
      "loss": 0.5166,
      "step": 205380
    },
    {
      "epoch": 84.07,
      "grad_norm": 1.429982304573059,
      "learning_rate": 3.1386564580032627e-06,
      "loss": 0.5214,
      "step": 205390
    },
    {
      "epoch": 84.08,
      "grad_norm": 1.7553719282150269,
      "learning_rate": 3.138507228817625e-06,
      "loss": 0.5096,
      "step": 205400
    },
    {
      "epoch": 84.08,
      "grad_norm": 2.577751874923706,
      "learning_rate": 3.1383579959913272e-06,
      "loss": 0.5068,
      "step": 205410
    },
    {
      "epoch": 84.09,
      "grad_norm": 2.2673802375793457,
      "learning_rate": 3.138208759525053e-06,
      "loss": 0.5058,
      "step": 205420
    },
    {
      "epoch": 84.09,
      "grad_norm": 1.815915584564209,
      "learning_rate": 3.138059519419486e-06,
      "loss": 0.4826,
      "step": 205430
    },
    {
      "epoch": 84.09,
      "grad_norm": 1.7269541025161743,
      "learning_rate": 3.1379102756753107e-06,
      "loss": 0.5064,
      "step": 205440
    },
    {
      "epoch": 84.1,
      "grad_norm": 1.8213642835617065,
      "learning_rate": 3.1377610282932097e-06,
      "loss": 0.4967,
      "step": 205450
    },
    {
      "epoch": 84.1,
      "grad_norm": 2.1244709491729736,
      "learning_rate": 3.1376117772738674e-06,
      "loss": 0.4845,
      "step": 205460
    },
    {
      "epoch": 84.11,
      "grad_norm": 2.1859819889068604,
      "learning_rate": 3.1374625226179675e-06,
      "loss": 0.53,
      "step": 205470
    },
    {
      "epoch": 84.11,
      "grad_norm": 1.7470126152038574,
      "learning_rate": 3.1373132643261935e-06,
      "loss": 0.4837,
      "step": 205480
    },
    {
      "epoch": 84.11,
      "grad_norm": 1.9129878282546997,
      "learning_rate": 3.1371640023992285e-06,
      "loss": 0.4874,
      "step": 205490
    },
    {
      "epoch": 84.12,
      "grad_norm": 1.7825464010238647,
      "learning_rate": 3.137014736837757e-06,
      "loss": 0.5072,
      "step": 205500
    },
    {
      "epoch": 84.12,
      "grad_norm": 1.8391190767288208,
      "learning_rate": 3.1368654676424632e-06,
      "loss": 0.4958,
      "step": 205510
    },
    {
      "epoch": 84.13,
      "grad_norm": 2.521456718444824,
      "learning_rate": 3.1367161948140296e-06,
      "loss": 0.5097,
      "step": 205520
    },
    {
      "epoch": 84.13,
      "grad_norm": 2.45145320892334,
      "learning_rate": 3.136566918353141e-06,
      "loss": 0.4977,
      "step": 205530
    },
    {
      "epoch": 84.13,
      "grad_norm": 2.1564416885375977,
      "learning_rate": 3.1364176382604804e-06,
      "loss": 0.5017,
      "step": 205540
    },
    {
      "epoch": 84.14,
      "grad_norm": 1.791074275970459,
      "learning_rate": 3.1362683545367327e-06,
      "loss": 0.5009,
      "step": 205550
    },
    {
      "epoch": 84.14,
      "grad_norm": 2.2807319164276123,
      "learning_rate": 3.136119067182581e-06,
      "loss": 0.5057,
      "step": 205560
    },
    {
      "epoch": 84.15,
      "grad_norm": 2.133026361465454,
      "learning_rate": 3.135969776198709e-06,
      "loss": 0.4929,
      "step": 205570
    },
    {
      "epoch": 84.15,
      "grad_norm": 1.8894532918930054,
      "learning_rate": 3.1358204815858014e-06,
      "loss": 0.5011,
      "step": 205580
    },
    {
      "epoch": 84.15,
      "grad_norm": 2.732952833175659,
      "learning_rate": 3.1356711833445413e-06,
      "loss": 0.4873,
      "step": 205590
    },
    {
      "epoch": 84.16,
      "grad_norm": 2.9504988193511963,
      "learning_rate": 3.1355218814756128e-06,
      "loss": 0.4767,
      "step": 205600
    },
    {
      "epoch": 84.16,
      "grad_norm": 1.612570881843567,
      "learning_rate": 3.1353725759796997e-06,
      "loss": 0.5023,
      "step": 205610
    },
    {
      "epoch": 84.17,
      "grad_norm": 1.9782780408859253,
      "learning_rate": 3.1352232668574874e-06,
      "loss": 0.5115,
      "step": 205620
    },
    {
      "epoch": 84.17,
      "grad_norm": 1.9689586162567139,
      "learning_rate": 3.1350739541096573e-06,
      "loss": 0.5185,
      "step": 205630
    },
    {
      "epoch": 84.18,
      "grad_norm": 2.0708394050598145,
      "learning_rate": 3.134924637736895e-06,
      "loss": 0.4743,
      "step": 205640
    },
    {
      "epoch": 84.18,
      "grad_norm": 2.50398325920105,
      "learning_rate": 3.1347753177398843e-06,
      "loss": 0.484,
      "step": 205650
    },
    {
      "epoch": 84.18,
      "grad_norm": 2.3211288452148438,
      "learning_rate": 3.1346259941193093e-06,
      "loss": 0.5109,
      "step": 205660
    },
    {
      "epoch": 84.19,
      "grad_norm": 1.7823801040649414,
      "learning_rate": 3.1344766668758533e-06,
      "loss": 0.5036,
      "step": 205670
    },
    {
      "epoch": 84.19,
      "grad_norm": 2.0356690883636475,
      "learning_rate": 3.134327336010202e-06,
      "loss": 0.4896,
      "step": 205680
    },
    {
      "epoch": 84.2,
      "grad_norm": 2.6644270420074463,
      "learning_rate": 3.1341780015230383e-06,
      "loss": 0.4992,
      "step": 205690
    },
    {
      "epoch": 84.2,
      "grad_norm": 2.3553647994995117,
      "learning_rate": 3.134028663415045e-06,
      "loss": 0.502,
      "step": 205700
    },
    {
      "epoch": 84.2,
      "grad_norm": 1.6723451614379883,
      "learning_rate": 3.1338793216869086e-06,
      "loss": 0.5116,
      "step": 205710
    },
    {
      "epoch": 84.21,
      "grad_norm": 2.0516529083251953,
      "learning_rate": 3.133729976339312e-06,
      "loss": 0.4966,
      "step": 205720
    },
    {
      "epoch": 84.21,
      "grad_norm": 1.9622000455856323,
      "learning_rate": 3.133580627372939e-06,
      "loss": 0.4641,
      "step": 205730
    },
    {
      "epoch": 84.22,
      "grad_norm": 2.201907157897949,
      "learning_rate": 3.1334312747884747e-06,
      "loss": 0.4966,
      "step": 205740
    },
    {
      "epoch": 84.22,
      "grad_norm": 2.0064873695373535,
      "learning_rate": 3.1332819185866025e-06,
      "loss": 0.5013,
      "step": 205750
    },
    {
      "epoch": 84.22,
      "grad_norm": 2.1008758544921875,
      "learning_rate": 3.1331325587680076e-06,
      "loss": 0.5076,
      "step": 205760
    },
    {
      "epoch": 84.23,
      "grad_norm": 1.9368478059768677,
      "learning_rate": 3.1329831953333723e-06,
      "loss": 0.4973,
      "step": 205770
    },
    {
      "epoch": 84.23,
      "grad_norm": 2.172847032546997,
      "learning_rate": 3.1328338282833827e-06,
      "loss": 0.5207,
      "step": 205780
    },
    {
      "epoch": 84.24,
      "grad_norm": 1.73348867893219,
      "learning_rate": 3.1326844576187223e-06,
      "loss": 0.5019,
      "step": 205790
    },
    {
      "epoch": 84.24,
      "grad_norm": 1.8403217792510986,
      "learning_rate": 3.1325350833400752e-06,
      "loss": 0.5154,
      "step": 205800
    },
    {
      "epoch": 84.24,
      "grad_norm": 1.7569259405136108,
      "learning_rate": 3.1323857054481253e-06,
      "loss": 0.4829,
      "step": 205810
    },
    {
      "epoch": 84.25,
      "grad_norm": 1.9077786207199097,
      "learning_rate": 3.132236323943558e-06,
      "loss": 0.4906,
      "step": 205820
    },
    {
      "epoch": 84.25,
      "grad_norm": 1.8565571308135986,
      "learning_rate": 3.1320869388270564e-06,
      "loss": 0.5054,
      "step": 205830
    },
    {
      "epoch": 84.26,
      "grad_norm": 1.6021500825881958,
      "learning_rate": 3.131937550099306e-06,
      "loss": 0.4938,
      "step": 205840
    },
    {
      "epoch": 84.26,
      "grad_norm": 1.914639949798584,
      "learning_rate": 3.13178815776099e-06,
      "loss": 0.5132,
      "step": 205850
    },
    {
      "epoch": 84.27,
      "grad_norm": 2.063286304473877,
      "learning_rate": 3.1316387618127934e-06,
      "loss": 0.4901,
      "step": 205860
    },
    {
      "epoch": 84.27,
      "grad_norm": 1.6178926229476929,
      "learning_rate": 3.131489362255401e-06,
      "loss": 0.4898,
      "step": 205870
    },
    {
      "epoch": 84.27,
      "grad_norm": 1.6149548292160034,
      "learning_rate": 3.1313399590894956e-06,
      "loss": 0.4887,
      "step": 205880
    },
    {
      "epoch": 84.28,
      "grad_norm": 1.732879877090454,
      "learning_rate": 3.1311905523157627e-06,
      "loss": 0.5133,
      "step": 205890
    },
    {
      "epoch": 84.28,
      "grad_norm": 2.0671310424804688,
      "learning_rate": 3.131041141934887e-06,
      "loss": 0.4926,
      "step": 205900
    },
    {
      "epoch": 84.29,
      "grad_norm": 1.615818977355957,
      "learning_rate": 3.1308917279475534e-06,
      "loss": 0.4952,
      "step": 205910
    },
    {
      "epoch": 84.29,
      "grad_norm": 2.738680362701416,
      "learning_rate": 3.1307423103544442e-06,
      "loss": 0.4975,
      "step": 205920
    },
    {
      "epoch": 84.29,
      "grad_norm": 1.9061262607574463,
      "learning_rate": 3.1305928891562446e-06,
      "loss": 0.5138,
      "step": 205930
    },
    {
      "epoch": 84.3,
      "grad_norm": 1.7179142236709595,
      "learning_rate": 3.1304434643536403e-06,
      "loss": 0.4954,
      "step": 205940
    },
    {
      "epoch": 84.3,
      "grad_norm": 1.8452863693237305,
      "learning_rate": 3.1302940359473153e-06,
      "loss": 0.5082,
      "step": 205950
    },
    {
      "epoch": 84.31,
      "grad_norm": 1.6036773920059204,
      "learning_rate": 3.1301446039379535e-06,
      "loss": 0.4875,
      "step": 205960
    },
    {
      "epoch": 84.31,
      "grad_norm": 1.7088948488235474,
      "learning_rate": 3.12999516832624e-06,
      "loss": 0.4889,
      "step": 205970
    },
    {
      "epoch": 84.31,
      "grad_norm": 1.582938551902771,
      "learning_rate": 3.1298457291128598e-06,
      "loss": 0.5064,
      "step": 205980
    },
    {
      "epoch": 84.32,
      "grad_norm": 1.7975093126296997,
      "learning_rate": 3.1296962862984963e-06,
      "loss": 0.5101,
      "step": 205990
    },
    {
      "epoch": 84.32,
      "grad_norm": 1.7933253049850464,
      "learning_rate": 3.129546839883834e-06,
      "loss": 0.4952,
      "step": 206000
    },
    {
      "epoch": 84.33,
      "grad_norm": 1.8425533771514893,
      "learning_rate": 3.1293973898695583e-06,
      "loss": 0.4965,
      "step": 206010
    },
    {
      "epoch": 84.33,
      "grad_norm": 1.8519287109375,
      "learning_rate": 3.1292479362563535e-06,
      "loss": 0.4976,
      "step": 206020
    },
    {
      "epoch": 84.33,
      "grad_norm": 2.0696065425872803,
      "learning_rate": 3.1290984790449045e-06,
      "loss": 0.5009,
      "step": 206030
    },
    {
      "epoch": 84.34,
      "grad_norm": 2.4881279468536377,
      "learning_rate": 3.128949018235896e-06,
      "loss": 0.4966,
      "step": 206040
    },
    {
      "epoch": 84.34,
      "grad_norm": 2.276888847351074,
      "learning_rate": 3.1287995538300126e-06,
      "loss": 0.5075,
      "step": 206050
    },
    {
      "epoch": 84.35,
      "grad_norm": 2.594500780105591,
      "learning_rate": 3.128650085827938e-06,
      "loss": 0.4877,
      "step": 206060
    },
    {
      "epoch": 84.35,
      "grad_norm": 1.7003668546676636,
      "learning_rate": 3.1285006142303583e-06,
      "loss": 0.4803,
      "step": 206070
    },
    {
      "epoch": 84.36,
      "grad_norm": 2.160968065261841,
      "learning_rate": 3.1283511390379575e-06,
      "loss": 0.4777,
      "step": 206080
    },
    {
      "epoch": 84.36,
      "grad_norm": 1.8803291320800781,
      "learning_rate": 3.1282016602514206e-06,
      "loss": 0.5131,
      "step": 206090
    },
    {
      "epoch": 84.36,
      "grad_norm": 2.226893901824951,
      "learning_rate": 3.128052177871431e-06,
      "loss": 0.5237,
      "step": 206100
    },
    {
      "epoch": 84.37,
      "grad_norm": 1.7182477712631226,
      "learning_rate": 3.127902691898675e-06,
      "loss": 0.4811,
      "step": 206110
    },
    {
      "epoch": 84.37,
      "grad_norm": 1.9763225317001343,
      "learning_rate": 3.127753202333838e-06,
      "loss": 0.4848,
      "step": 206120
    },
    {
      "epoch": 84.38,
      "grad_norm": 2.2723889350891113,
      "learning_rate": 3.1276037091776026e-06,
      "loss": 0.5025,
      "step": 206130
    },
    {
      "epoch": 84.38,
      "grad_norm": 2.123363494873047,
      "learning_rate": 3.1274542124306553e-06,
      "loss": 0.5224,
      "step": 206140
    },
    {
      "epoch": 84.38,
      "grad_norm": 2.153562068939209,
      "learning_rate": 3.127304712093681e-06,
      "loss": 0.493,
      "step": 206150
    },
    {
      "epoch": 84.39,
      "grad_norm": 2.128737211227417,
      "learning_rate": 3.1271552081673627e-06,
      "loss": 0.511,
      "step": 206160
    },
    {
      "epoch": 84.39,
      "grad_norm": 1.743121862411499,
      "learning_rate": 3.127005700652387e-06,
      "loss": 0.5006,
      "step": 206170
    },
    {
      "epoch": 84.4,
      "grad_norm": 1.7068299055099487,
      "learning_rate": 3.126856189549438e-06,
      "loss": 0.4943,
      "step": 206180
    },
    {
      "epoch": 84.4,
      "grad_norm": 1.787197232246399,
      "learning_rate": 3.1267066748592013e-06,
      "loss": 0.5377,
      "step": 206190
    },
    {
      "epoch": 84.4,
      "grad_norm": 2.009779214859009,
      "learning_rate": 3.1265571565823613e-06,
      "loss": 0.4931,
      "step": 206200
    },
    {
      "epoch": 84.41,
      "grad_norm": 2.083843946456909,
      "learning_rate": 3.1264076347196026e-06,
      "loss": 0.484,
      "step": 206210
    },
    {
      "epoch": 84.41,
      "grad_norm": 2.0618948936462402,
      "learning_rate": 3.1262581092716104e-06,
      "loss": 0.508,
      "step": 206220
    },
    {
      "epoch": 84.42,
      "grad_norm": 1.4557386636734009,
      "learning_rate": 3.12610858023907e-06,
      "loss": 0.501,
      "step": 206230
    },
    {
      "epoch": 84.42,
      "grad_norm": 1.8684172630310059,
      "learning_rate": 3.1259590476226658e-06,
      "loss": 0.5037,
      "step": 206240
    },
    {
      "epoch": 84.42,
      "grad_norm": 2.184600591659546,
      "learning_rate": 3.1258095114230837e-06,
      "loss": 0.4939,
      "step": 206250
    },
    {
      "epoch": 84.43,
      "grad_norm": 2.0377111434936523,
      "learning_rate": 3.1256599716410076e-06,
      "loss": 0.4874,
      "step": 206260
    },
    {
      "epoch": 84.43,
      "grad_norm": 1.6501493453979492,
      "learning_rate": 3.1255104282771235e-06,
      "loss": 0.5,
      "step": 206270
    },
    {
      "epoch": 84.44,
      "grad_norm": 1.6590046882629395,
      "learning_rate": 3.1253608813321157e-06,
      "loss": 0.4884,
      "step": 206280
    },
    {
      "epoch": 84.44,
      "grad_norm": 2.5539114475250244,
      "learning_rate": 3.125211330806669e-06,
      "loss": 0.4851,
      "step": 206290
    },
    {
      "epoch": 84.45,
      "grad_norm": 1.7450480461120605,
      "learning_rate": 3.125061776701469e-06,
      "loss": 0.5028,
      "step": 206300
    },
    {
      "epoch": 84.45,
      "grad_norm": 1.858580231666565,
      "learning_rate": 3.1249122190172014e-06,
      "loss": 0.4907,
      "step": 206310
    },
    {
      "epoch": 84.45,
      "grad_norm": 1.9905741214752197,
      "learning_rate": 3.12476265775455e-06,
      "loss": 0.5019,
      "step": 206320
    },
    {
      "epoch": 84.46,
      "grad_norm": 1.7283450365066528,
      "learning_rate": 3.124613092914201e-06,
      "loss": 0.482,
      "step": 206330
    },
    {
      "epoch": 84.46,
      "grad_norm": 2.220646381378174,
      "learning_rate": 3.124463524496839e-06,
      "loss": 0.4975,
      "step": 206340
    },
    {
      "epoch": 84.47,
      "grad_norm": 3.483231782913208,
      "learning_rate": 3.12431395250315e-06,
      "loss": 0.4906,
      "step": 206350
    },
    {
      "epoch": 84.47,
      "grad_norm": 2.2768537998199463,
      "learning_rate": 3.1241643769338174e-06,
      "loss": 0.5158,
      "step": 206360
    },
    {
      "epoch": 84.47,
      "grad_norm": 1.9741394519805908,
      "learning_rate": 3.1240147977895277e-06,
      "loss": 0.5062,
      "step": 206370
    },
    {
      "epoch": 84.48,
      "grad_norm": 1.9511711597442627,
      "learning_rate": 3.1238652150709657e-06,
      "loss": 0.5228,
      "step": 206380
    },
    {
      "epoch": 84.48,
      "grad_norm": 1.9210673570632935,
      "learning_rate": 3.123715628778817e-06,
      "loss": 0.4826,
      "step": 206390
    },
    {
      "epoch": 84.49,
      "grad_norm": 2.076403856277466,
      "learning_rate": 3.123566038913766e-06,
      "loss": 0.5108,
      "step": 206400
    },
    {
      "epoch": 84.49,
      "grad_norm": 1.6916636228561401,
      "learning_rate": 3.123416445476499e-06,
      "loss": 0.5063,
      "step": 206410
    },
    {
      "epoch": 84.49,
      "grad_norm": 1.8507803678512573,
      "learning_rate": 3.123266848467701e-06,
      "loss": 0.5157,
      "step": 206420
    },
    {
      "epoch": 84.5,
      "grad_norm": 1.5354102849960327,
      "learning_rate": 3.1231172478880562e-06,
      "loss": 0.5122,
      "step": 206430
    },
    {
      "epoch": 84.5,
      "grad_norm": 1.6223924160003662,
      "learning_rate": 3.1229676437382515e-06,
      "loss": 0.5114,
      "step": 206440
    },
    {
      "epoch": 84.51,
      "grad_norm": 2.3263707160949707,
      "learning_rate": 3.122818036018971e-06,
      "loss": 0.4913,
      "step": 206450
    },
    {
      "epoch": 84.51,
      "grad_norm": 2.154934883117676,
      "learning_rate": 3.122668424730901e-06,
      "loss": 0.5201,
      "step": 206460
    },
    {
      "epoch": 84.51,
      "grad_norm": 1.984674096107483,
      "learning_rate": 3.122518809874726e-06,
      "loss": 0.48,
      "step": 206470
    },
    {
      "epoch": 84.52,
      "grad_norm": 1.4926728010177612,
      "learning_rate": 3.1223691914511316e-06,
      "loss": 0.513,
      "step": 206480
    },
    {
      "epoch": 84.52,
      "grad_norm": 1.8343950510025024,
      "learning_rate": 3.122219569460803e-06,
      "loss": 0.5078,
      "step": 206490
    },
    {
      "epoch": 84.53,
      "grad_norm": 2.057255983352661,
      "learning_rate": 3.1220699439044263e-06,
      "loss": 0.5029,
      "step": 206500
    },
    {
      "epoch": 84.53,
      "grad_norm": 1.6586971282958984,
      "learning_rate": 3.121920314782686e-06,
      "loss": 0.5039,
      "step": 206510
    },
    {
      "epoch": 84.54,
      "grad_norm": 1.8945153951644897,
      "learning_rate": 3.1217706820962683e-06,
      "loss": 0.5053,
      "step": 206520
    },
    {
      "epoch": 84.54,
      "grad_norm": 2.2418596744537354,
      "learning_rate": 3.121621045845859e-06,
      "loss": 0.4724,
      "step": 206530
    },
    {
      "epoch": 84.54,
      "grad_norm": 1.6265268325805664,
      "learning_rate": 3.121471406032142e-06,
      "loss": 0.4957,
      "step": 206540
    },
    {
      "epoch": 84.55,
      "grad_norm": 1.9776132106781006,
      "learning_rate": 3.121321762655804e-06,
      "loss": 0.5173,
      "step": 206550
    },
    {
      "epoch": 84.55,
      "grad_norm": 1.7973943948745728,
      "learning_rate": 3.121172115717531e-06,
      "loss": 0.5267,
      "step": 206560
    },
    {
      "epoch": 84.56,
      "grad_norm": 1.9477070569992065,
      "learning_rate": 3.1210224652180067e-06,
      "loss": 0.5148,
      "step": 206570
    },
    {
      "epoch": 84.56,
      "grad_norm": 2.2326066493988037,
      "learning_rate": 3.1208728111579176e-06,
      "loss": 0.5052,
      "step": 206580
    },
    {
      "epoch": 84.56,
      "grad_norm": 1.7168927192687988,
      "learning_rate": 3.120723153537949e-06,
      "loss": 0.4992,
      "step": 206590
    },
    {
      "epoch": 84.57,
      "grad_norm": 2.1931872367858887,
      "learning_rate": 3.120573492358787e-06,
      "loss": 0.4913,
      "step": 206600
    },
    {
      "epoch": 84.57,
      "grad_norm": 2.2462658882141113,
      "learning_rate": 3.1204238276211173e-06,
      "loss": 0.4882,
      "step": 206610
    },
    {
      "epoch": 84.58,
      "grad_norm": 2.1370723247528076,
      "learning_rate": 3.1202741593256244e-06,
      "loss": 0.4993,
      "step": 206620
    },
    {
      "epoch": 84.58,
      "grad_norm": 1.810910701751709,
      "learning_rate": 3.120124487472996e-06,
      "loss": 0.5118,
      "step": 206630
    },
    {
      "epoch": 84.58,
      "grad_norm": 2.042865514755249,
      "learning_rate": 3.119974812063915e-06,
      "loss": 0.4915,
      "step": 206640
    },
    {
      "epoch": 84.59,
      "grad_norm": 1.535768985748291,
      "learning_rate": 3.1198251330990683e-06,
      "loss": 0.5048,
      "step": 206650
    },
    {
      "epoch": 84.59,
      "grad_norm": 2.105790138244629,
      "learning_rate": 3.1196754505791423e-06,
      "loss": 0.5176,
      "step": 206660
    },
    {
      "epoch": 84.6,
      "grad_norm": 1.5372182130813599,
      "learning_rate": 3.1195257645048217e-06,
      "loss": 0.5083,
      "step": 206670
    },
    {
      "epoch": 84.6,
      "grad_norm": 1.393754005432129,
      "learning_rate": 3.1193760748767917e-06,
      "loss": 0.49,
      "step": 206680
    },
    {
      "epoch": 84.6,
      "grad_norm": 1.7046774625778198,
      "learning_rate": 3.1192263816957397e-06,
      "loss": 0.4776,
      "step": 206690
    },
    {
      "epoch": 84.61,
      "grad_norm": 2.054871082305908,
      "learning_rate": 3.1190766849623506e-06,
      "loss": 0.4946,
      "step": 206700
    },
    {
      "epoch": 84.61,
      "grad_norm": 2.5056259632110596,
      "learning_rate": 3.1189269846773095e-06,
      "loss": 0.5038,
      "step": 206710
    },
    {
      "epoch": 84.62,
      "grad_norm": 1.7064987421035767,
      "learning_rate": 3.118777280841303e-06,
      "loss": 0.5102,
      "step": 206720
    },
    {
      "epoch": 84.62,
      "grad_norm": 2.053480863571167,
      "learning_rate": 3.118627573455017e-06,
      "loss": 0.4957,
      "step": 206730
    },
    {
      "epoch": 84.63,
      "grad_norm": 1.7684009075164795,
      "learning_rate": 3.1184778625191357e-06,
      "loss": 0.4823,
      "step": 206740
    },
    {
      "epoch": 84.63,
      "grad_norm": 1.7156155109405518,
      "learning_rate": 3.1183281480343462e-06,
      "loss": 0.513,
      "step": 206750
    },
    {
      "epoch": 84.63,
      "grad_norm": 1.6024223566055298,
      "learning_rate": 3.118178430001335e-06,
      "loss": 0.5135,
      "step": 206760
    },
    {
      "epoch": 84.64,
      "grad_norm": 1.9380178451538086,
      "learning_rate": 3.1180287084207865e-06,
      "loss": 0.4936,
      "step": 206770
    },
    {
      "epoch": 84.64,
      "grad_norm": 1.8244179487228394,
      "learning_rate": 3.1178789832933876e-06,
      "loss": 0.508,
      "step": 206780
    },
    {
      "epoch": 84.65,
      "grad_norm": 1.8181825876235962,
      "learning_rate": 3.1177292546198234e-06,
      "loss": 0.4767,
      "step": 206790
    },
    {
      "epoch": 84.65,
      "grad_norm": 1.7484809160232544,
      "learning_rate": 3.1175795224007796e-06,
      "loss": 0.4829,
      "step": 206800
    },
    {
      "epoch": 84.65,
      "grad_norm": 1.835483431816101,
      "learning_rate": 3.117429786636944e-06,
      "loss": 0.4936,
      "step": 206810
    },
    {
      "epoch": 84.66,
      "grad_norm": 2.1863553524017334,
      "learning_rate": 3.1172800473290002e-06,
      "loss": 0.5022,
      "step": 206820
    },
    {
      "epoch": 84.66,
      "grad_norm": 2.1891415119171143,
      "learning_rate": 3.1171303044776347e-06,
      "loss": 0.4843,
      "step": 206830
    },
    {
      "epoch": 84.67,
      "grad_norm": 2.2002034187316895,
      "learning_rate": 3.1169805580835344e-06,
      "loss": 0.4814,
      "step": 206840
    },
    {
      "epoch": 84.67,
      "grad_norm": 2.027287244796753,
      "learning_rate": 3.116830808147384e-06,
      "loss": 0.5106,
      "step": 206850
    },
    {
      "epoch": 84.67,
      "grad_norm": 2.7469723224639893,
      "learning_rate": 3.1166810546698716e-06,
      "loss": 0.4763,
      "step": 206860
    },
    {
      "epoch": 84.68,
      "grad_norm": 1.6136877536773682,
      "learning_rate": 3.11653129765168e-06,
      "loss": 0.4956,
      "step": 206870
    },
    {
      "epoch": 84.68,
      "grad_norm": 1.978335976600647,
      "learning_rate": 3.1163815370934982e-06,
      "loss": 0.4875,
      "step": 206880
    },
    {
      "epoch": 84.69,
      "grad_norm": 1.329634666442871,
      "learning_rate": 3.11623177299601e-06,
      "loss": 0.516,
      "step": 206890
    },
    {
      "epoch": 84.69,
      "grad_norm": 1.4901777505874634,
      "learning_rate": 3.116082005359903e-06,
      "loss": 0.5041,
      "step": 206900
    },
    {
      "epoch": 84.7,
      "grad_norm": 1.7041329145431519,
      "learning_rate": 3.1159322341858627e-06,
      "loss": 0.5238,
      "step": 206910
    },
    {
      "epoch": 84.7,
      "grad_norm": 2.218345880508423,
      "learning_rate": 3.115782459474576e-06,
      "loss": 0.4832,
      "step": 206920
    },
    {
      "epoch": 84.7,
      "grad_norm": 1.8286670446395874,
      "learning_rate": 3.1156326812267267e-06,
      "loss": 0.511,
      "step": 206930
    },
    {
      "epoch": 84.71,
      "grad_norm": 2.695624351501465,
      "learning_rate": 3.1154828994430033e-06,
      "loss": 0.5146,
      "step": 206940
    },
    {
      "epoch": 84.71,
      "grad_norm": 1.6202802658081055,
      "learning_rate": 3.1153331141240904e-06,
      "loss": 0.4981,
      "step": 206950
    },
    {
      "epoch": 84.72,
      "grad_norm": 1.707666039466858,
      "learning_rate": 3.1151833252706747e-06,
      "loss": 0.4798,
      "step": 206960
    },
    {
      "epoch": 84.72,
      "grad_norm": 1.7068073749542236,
      "learning_rate": 3.1150335328834426e-06,
      "loss": 0.501,
      "step": 206970
    },
    {
      "epoch": 84.72,
      "grad_norm": 1.6179554462432861,
      "learning_rate": 3.1148837369630806e-06,
      "loss": 0.5214,
      "step": 206980
    },
    {
      "epoch": 84.73,
      "grad_norm": 1.6642354726791382,
      "learning_rate": 3.1147339375102736e-06,
      "loss": 0.4701,
      "step": 206990
    },
    {
      "epoch": 84.73,
      "grad_norm": 1.8481091260910034,
      "learning_rate": 3.1145841345257094e-06,
      "loss": 0.4881,
      "step": 207000
    },
    {
      "epoch": 84.74,
      "grad_norm": 2.294736623764038,
      "learning_rate": 3.1144343280100727e-06,
      "loss": 0.5059,
      "step": 207010
    },
    {
      "epoch": 84.74,
      "grad_norm": 1.9193249940872192,
      "learning_rate": 3.114284517964051e-06,
      "loss": 0.4975,
      "step": 207020
    },
    {
      "epoch": 84.74,
      "grad_norm": 1.804238200187683,
      "learning_rate": 3.1141347043883295e-06,
      "loss": 0.482,
      "step": 207030
    },
    {
      "epoch": 84.75,
      "grad_norm": 2.034353494644165,
      "learning_rate": 3.1139848872835956e-06,
      "loss": 0.5169,
      "step": 207040
    },
    {
      "epoch": 84.75,
      "grad_norm": 2.054429292678833,
      "learning_rate": 3.1138350666505345e-06,
      "loss": 0.4795,
      "step": 207050
    },
    {
      "epoch": 84.76,
      "grad_norm": 2.2052206993103027,
      "learning_rate": 3.1136852424898324e-06,
      "loss": 0.4782,
      "step": 207060
    },
    {
      "epoch": 84.76,
      "grad_norm": 2.487196445465088,
      "learning_rate": 3.113535414802177e-06,
      "loss": 0.491,
      "step": 207070
    },
    {
      "epoch": 84.76,
      "grad_norm": 2.837332248687744,
      "learning_rate": 3.1133855835882534e-06,
      "loss": 0.4865,
      "step": 207080
    },
    {
      "epoch": 84.77,
      "grad_norm": 1.9623627662658691,
      "learning_rate": 3.1132357488487492e-06,
      "loss": 0.5182,
      "step": 207090
    },
    {
      "epoch": 84.77,
      "grad_norm": 1.7632285356521606,
      "learning_rate": 3.113085910584349e-06,
      "loss": 0.5046,
      "step": 207100
    },
    {
      "epoch": 84.78,
      "grad_norm": 2.95528507232666,
      "learning_rate": 3.1129360687957407e-06,
      "loss": 0.4978,
      "step": 207110
    },
    {
      "epoch": 84.78,
      "grad_norm": 1.937597393989563,
      "learning_rate": 3.11278622348361e-06,
      "loss": 0.495,
      "step": 207120
    },
    {
      "epoch": 84.79,
      "grad_norm": 1.748298168182373,
      "learning_rate": 3.1126363746486435e-06,
      "loss": 0.4965,
      "step": 207130
    },
    {
      "epoch": 84.79,
      "grad_norm": 2.087526321411133,
      "learning_rate": 3.112486522291528e-06,
      "loss": 0.4953,
      "step": 207140
    },
    {
      "epoch": 84.79,
      "grad_norm": 2.5446033477783203,
      "learning_rate": 3.1123366664129485e-06,
      "loss": 0.5177,
      "step": 207150
    },
    {
      "epoch": 84.8,
      "grad_norm": 1.8561420440673828,
      "learning_rate": 3.1121868070135933e-06,
      "loss": 0.4959,
      "step": 207160
    },
    {
      "epoch": 84.8,
      "grad_norm": 1.5561585426330566,
      "learning_rate": 3.1120369440941475e-06,
      "loss": 0.5011,
      "step": 207170
    },
    {
      "epoch": 84.81,
      "grad_norm": 2.1303679943084717,
      "learning_rate": 3.111887077655299e-06,
      "loss": 0.5152,
      "step": 207180
    },
    {
      "epoch": 84.81,
      "grad_norm": 1.9817231893539429,
      "learning_rate": 3.1117372076977334e-06,
      "loss": 0.5069,
      "step": 207190
    },
    {
      "epoch": 84.81,
      "grad_norm": 1.8599765300750732,
      "learning_rate": 3.1115873342221365e-06,
      "loss": 0.4933,
      "step": 207200
    },
    {
      "epoch": 84.82,
      "grad_norm": 2.3440866470336914,
      "learning_rate": 3.1114374572291967e-06,
      "loss": 0.51,
      "step": 207210
    },
    {
      "epoch": 84.82,
      "grad_norm": 2.4538490772247314,
      "learning_rate": 3.111287576719599e-06,
      "loss": 0.5039,
      "step": 207220
    },
    {
      "epoch": 84.83,
      "grad_norm": 1.9492650032043457,
      "learning_rate": 3.111137692694031e-06,
      "loss": 0.5112,
      "step": 207230
    },
    {
      "epoch": 84.83,
      "grad_norm": 1.649682641029358,
      "learning_rate": 3.110987805153178e-06,
      "loss": 0.4984,
      "step": 207240
    },
    {
      "epoch": 84.83,
      "grad_norm": 2.5101675987243652,
      "learning_rate": 3.1108379140977277e-06,
      "loss": 0.5148,
      "step": 207250
    },
    {
      "epoch": 84.84,
      "grad_norm": 1.965264081954956,
      "learning_rate": 3.110688019528366e-06,
      "loss": 0.5008,
      "step": 207260
    },
    {
      "epoch": 84.84,
      "grad_norm": 1.7829082012176514,
      "learning_rate": 3.110538121445781e-06,
      "loss": 0.5243,
      "step": 207270
    },
    {
      "epoch": 84.85,
      "grad_norm": 1.653476357460022,
      "learning_rate": 3.1103882198506583e-06,
      "loss": 0.491,
      "step": 207280
    },
    {
      "epoch": 84.85,
      "grad_norm": 2.2139017581939697,
      "learning_rate": 3.1102383147436845e-06,
      "loss": 0.5073,
      "step": 207290
    },
    {
      "epoch": 84.85,
      "grad_norm": 1.7772719860076904,
      "learning_rate": 3.1100884061255465e-06,
      "loss": 0.5137,
      "step": 207300
    },
    {
      "epoch": 84.86,
      "grad_norm": 1.7640526294708252,
      "learning_rate": 3.109938493996931e-06,
      "loss": 0.5043,
      "step": 207310
    },
    {
      "epoch": 84.86,
      "grad_norm": 1.4896690845489502,
      "learning_rate": 3.1097885783585238e-06,
      "loss": 0.5164,
      "step": 207320
    },
    {
      "epoch": 84.87,
      "grad_norm": 1.6970478296279907,
      "learning_rate": 3.1096386592110133e-06,
      "loss": 0.522,
      "step": 207330
    },
    {
      "epoch": 84.87,
      "grad_norm": 1.9820085763931274,
      "learning_rate": 3.1094887365550847e-06,
      "loss": 0.5003,
      "step": 207340
    },
    {
      "epoch": 84.88,
      "grad_norm": 1.7687236070632935,
      "learning_rate": 3.109338810391426e-06,
      "loss": 0.5004,
      "step": 207350
    },
    {
      "epoch": 84.88,
      "grad_norm": 2.1747329235076904,
      "learning_rate": 3.1091888807207244e-06,
      "loss": 0.4912,
      "step": 207360
    },
    {
      "epoch": 84.88,
      "grad_norm": 1.9573497772216797,
      "learning_rate": 3.1090389475436644e-06,
      "loss": 0.4874,
      "step": 207370
    },
    {
      "epoch": 84.89,
      "grad_norm": 2.1168956756591797,
      "learning_rate": 3.1088890108609355e-06,
      "loss": 0.4954,
      "step": 207380
    },
    {
      "epoch": 84.89,
      "grad_norm": 1.9690121412277222,
      "learning_rate": 3.1087390706732227e-06,
      "loss": 0.5153,
      "step": 207390
    },
    {
      "epoch": 84.9,
      "grad_norm": 1.6669822931289673,
      "learning_rate": 3.1085891269812135e-06,
      "loss": 0.4958,
      "step": 207400
    },
    {
      "epoch": 84.9,
      "grad_norm": 1.9572242498397827,
      "learning_rate": 3.1084391797855944e-06,
      "loss": 0.5181,
      "step": 207410
    },
    {
      "epoch": 84.9,
      "grad_norm": 1.8166062831878662,
      "learning_rate": 3.1082892290870527e-06,
      "loss": 0.4783,
      "step": 207420
    },
    {
      "epoch": 84.91,
      "grad_norm": 1.8737766742706299,
      "learning_rate": 3.1081392748862755e-06,
      "loss": 0.4874,
      "step": 207430
    },
    {
      "epoch": 84.91,
      "grad_norm": 1.9172556400299072,
      "learning_rate": 3.107989317183949e-06,
      "loss": 0.519,
      "step": 207440
    },
    {
      "epoch": 84.92,
      "grad_norm": 2.0511975288391113,
      "learning_rate": 3.107839355980761e-06,
      "loss": 0.4874,
      "step": 207450
    },
    {
      "epoch": 84.92,
      "grad_norm": 1.8553187847137451,
      "learning_rate": 3.107689391277398e-06,
      "loss": 0.4923,
      "step": 207460
    },
    {
      "epoch": 84.92,
      "grad_norm": 1.8145867586135864,
      "learning_rate": 3.1075394230745464e-06,
      "loss": 0.4935,
      "step": 207470
    },
    {
      "epoch": 84.93,
      "grad_norm": 1.9147675037384033,
      "learning_rate": 3.1073894513728945e-06,
      "loss": 0.4937,
      "step": 207480
    },
    {
      "epoch": 84.93,
      "grad_norm": 1.8964273929595947,
      "learning_rate": 3.1072394761731277e-06,
      "loss": 0.4961,
      "step": 207490
    },
    {
      "epoch": 84.94,
      "grad_norm": 2.066685676574707,
      "learning_rate": 3.107089497475934e-06,
      "loss": 0.4949,
      "step": 207500
    },
    {
      "epoch": 84.94,
      "grad_norm": 1.8725706338882446,
      "learning_rate": 3.1069395152820016e-06,
      "loss": 0.5064,
      "step": 207510
    },
    {
      "epoch": 84.94,
      "grad_norm": 1.921525001525879,
      "learning_rate": 3.106789529592015e-06,
      "loss": 0.4973,
      "step": 207520
    },
    {
      "epoch": 84.95,
      "grad_norm": 1.5867464542388916,
      "learning_rate": 3.106639540406662e-06,
      "loss": 0.5004,
      "step": 207530
    },
    {
      "epoch": 84.95,
      "grad_norm": 1.8982619047164917,
      "learning_rate": 3.1064895477266307e-06,
      "loss": 0.5036,
      "step": 207540
    },
    {
      "epoch": 84.96,
      "grad_norm": 1.7962312698364258,
      "learning_rate": 3.106339551552608e-06,
      "loss": 0.5082,
      "step": 207550
    },
    {
      "epoch": 84.96,
      "grad_norm": 2.0184433460235596,
      "learning_rate": 3.1061895518852806e-06,
      "loss": 0.4909,
      "step": 207560
    },
    {
      "epoch": 84.97,
      "grad_norm": 2.114755153656006,
      "learning_rate": 3.106039548725336e-06,
      "loss": 0.533,
      "step": 207570
    },
    {
      "epoch": 84.97,
      "grad_norm": 1.5509945154190063,
      "learning_rate": 3.105889542073461e-06,
      "loss": 0.5236,
      "step": 207580
    },
    {
      "epoch": 84.97,
      "grad_norm": 1.611830234527588,
      "learning_rate": 3.105739531930342e-06,
      "loss": 0.5194,
      "step": 207590
    },
    {
      "epoch": 84.98,
      "grad_norm": 1.5708180665969849,
      "learning_rate": 3.105589518296667e-06,
      "loss": 0.4781,
      "step": 207600
    },
    {
      "epoch": 84.98,
      "grad_norm": 1.577624797821045,
      "learning_rate": 3.1054395011731234e-06,
      "loss": 0.5083,
      "step": 207610
    },
    {
      "epoch": 84.99,
      "grad_norm": 2.025177478790283,
      "learning_rate": 3.105289480560398e-06,
      "loss": 0.4924,
      "step": 207620
    },
    {
      "epoch": 84.99,
      "grad_norm": 1.8807600736618042,
      "learning_rate": 3.105139456459179e-06,
      "loss": 0.483,
      "step": 207630
    },
    {
      "epoch": 84.99,
      "grad_norm": 1.5940572023391724,
      "learning_rate": 3.104989428870152e-06,
      "loss": 0.486,
      "step": 207640
    },
    {
      "epoch": 85.0,
      "grad_norm": 1.856583595275879,
      "learning_rate": 3.1048393977940057e-06,
      "loss": 0.503,
      "step": 207650
    },
    {
      "epoch": 85.0,
      "eval_loss": 0.4979647696018219,
      "eval_runtime": 72.829,
      "eval_samples_per_second": 47.357,
      "eval_steps_per_second": 5.932,
      "step": 207655
    },
    {
      "epoch": 85.0,
      "grad_norm": 1.3714736700057983,
      "learning_rate": 3.104689363231426e-06,
      "loss": 0.508,
      "step": 207660
    },
    {
      "epoch": 85.01,
      "grad_norm": 1.8781588077545166,
      "learning_rate": 3.1045393251831018e-06,
      "loss": 0.488,
      "step": 207670
    },
    {
      "epoch": 85.01,
      "grad_norm": 2.0196096897125244,
      "learning_rate": 3.1043892836497184e-06,
      "loss": 0.5052,
      "step": 207680
    },
    {
      "epoch": 85.01,
      "grad_norm": 1.8330262899398804,
      "learning_rate": 3.1042392386319646e-06,
      "loss": 0.4972,
      "step": 207690
    },
    {
      "epoch": 85.02,
      "grad_norm": 2.1277859210968018,
      "learning_rate": 3.104089190130528e-06,
      "loss": 0.4762,
      "step": 207700
    },
    {
      "epoch": 85.02,
      "grad_norm": 2.0593953132629395,
      "learning_rate": 3.103939138146095e-06,
      "loss": 0.5051,
      "step": 207710
    },
    {
      "epoch": 85.03,
      "grad_norm": 2.0204977989196777,
      "learning_rate": 3.103789082679353e-06,
      "loss": 0.4977,
      "step": 207720
    },
    {
      "epoch": 85.03,
      "grad_norm": 1.4215196371078491,
      "learning_rate": 3.10363902373099e-06,
      "loss": 0.492,
      "step": 207730
    },
    {
      "epoch": 85.03,
      "grad_norm": 2.495798349380493,
      "learning_rate": 3.103488961301693e-06,
      "loss": 0.5211,
      "step": 207740
    },
    {
      "epoch": 85.04,
      "grad_norm": 1.9047117233276367,
      "learning_rate": 3.1033388953921493e-06,
      "loss": 0.5199,
      "step": 207750
    },
    {
      "epoch": 85.04,
      "grad_norm": 2.064614772796631,
      "learning_rate": 3.103188826003047e-06,
      "loss": 0.5019,
      "step": 207760
    },
    {
      "epoch": 85.05,
      "grad_norm": 1.572847604751587,
      "learning_rate": 3.1030387531350726e-06,
      "loss": 0.4873,
      "step": 207770
    },
    {
      "epoch": 85.05,
      "grad_norm": 2.0236074924468994,
      "learning_rate": 3.1028886767889138e-06,
      "loss": 0.5123,
      "step": 207780
    },
    {
      "epoch": 85.06,
      "grad_norm": 2.4440810680389404,
      "learning_rate": 3.1027385969652586e-06,
      "loss": 0.5332,
      "step": 207790
    },
    {
      "epoch": 85.06,
      "grad_norm": 1.906814694404602,
      "learning_rate": 3.1025885136647945e-06,
      "loss": 0.4897,
      "step": 207800
    },
    {
      "epoch": 85.06,
      "grad_norm": 1.5429080724716187,
      "learning_rate": 3.102438426888208e-06,
      "loss": 0.5045,
      "step": 207810
    },
    {
      "epoch": 85.07,
      "grad_norm": 1.8770431280136108,
      "learning_rate": 3.1022883366361878e-06,
      "loss": 0.5062,
      "step": 207820
    },
    {
      "epoch": 85.07,
      "grad_norm": 1.9088505506515503,
      "learning_rate": 3.1021382429094207e-06,
      "loss": 0.5093,
      "step": 207830
    },
    {
      "epoch": 85.08,
      "grad_norm": 1.6059461832046509,
      "learning_rate": 3.1019881457085945e-06,
      "loss": 0.5101,
      "step": 207840
    },
    {
      "epoch": 85.08,
      "grad_norm": 2.1607375144958496,
      "learning_rate": 3.101838045034397e-06,
      "loss": 0.496,
      "step": 207850
    },
    {
      "epoch": 85.08,
      "grad_norm": 1.5924158096313477,
      "learning_rate": 3.1016879408875154e-06,
      "loss": 0.4823,
      "step": 207860
    },
    {
      "epoch": 85.09,
      "grad_norm": 1.647412896156311,
      "learning_rate": 3.1015378332686384e-06,
      "loss": 0.4956,
      "step": 207870
    },
    {
      "epoch": 85.09,
      "grad_norm": 1.7046644687652588,
      "learning_rate": 3.101387722178452e-06,
      "loss": 0.4829,
      "step": 207880
    },
    {
      "epoch": 85.1,
      "grad_norm": 1.954651951789856,
      "learning_rate": 3.101237607617644e-06,
      "loss": 0.5069,
      "step": 207890
    },
    {
      "epoch": 85.1,
      "grad_norm": 1.6943867206573486,
      "learning_rate": 3.1010874895869033e-06,
      "loss": 0.5061,
      "step": 207900
    },
    {
      "epoch": 85.1,
      "grad_norm": 1.7583963871002197,
      "learning_rate": 3.100937368086916e-06,
      "loss": 0.4915,
      "step": 207910
    },
    {
      "epoch": 85.11,
      "grad_norm": 2.35483717918396,
      "learning_rate": 3.100787243118371e-06,
      "loss": 0.4979,
      "step": 207920
    },
    {
      "epoch": 85.11,
      "grad_norm": 1.7962062358856201,
      "learning_rate": 3.100637114681956e-06,
      "loss": 0.4875,
      "step": 207930
    },
    {
      "epoch": 85.12,
      "grad_norm": 1.6271313428878784,
      "learning_rate": 3.1004869827783583e-06,
      "loss": 0.4984,
      "step": 207940
    },
    {
      "epoch": 85.12,
      "grad_norm": 1.50118887424469,
      "learning_rate": 3.1003368474082654e-06,
      "loss": 0.5115,
      "step": 207950
    },
    {
      "epoch": 85.12,
      "grad_norm": 2.1063432693481445,
      "learning_rate": 3.1001867085723656e-06,
      "loss": 0.4644,
      "step": 207960
    },
    {
      "epoch": 85.13,
      "grad_norm": 1.8404219150543213,
      "learning_rate": 3.100036566271346e-06,
      "loss": 0.51,
      "step": 207970
    },
    {
      "epoch": 85.13,
      "grad_norm": 1.718652367591858,
      "learning_rate": 3.099886420505895e-06,
      "loss": 0.5095,
      "step": 207980
    },
    {
      "epoch": 85.14,
      "grad_norm": 1.9789273738861084,
      "learning_rate": 3.0997362712767004e-06,
      "loss": 0.4929,
      "step": 207990
    },
    {
      "epoch": 85.14,
      "grad_norm": 1.5107423067092896,
      "learning_rate": 3.0995861185844495e-06,
      "loss": 0.4884,
      "step": 208000
    },
    {
      "epoch": 85.15,
      "grad_norm": 1.7467920780181885,
      "learning_rate": 3.09943596242983e-06,
      "loss": 0.5051,
      "step": 208010
    },
    {
      "epoch": 85.15,
      "grad_norm": 1.642920732498169,
      "learning_rate": 3.0992858028135308e-06,
      "loss": 0.4841,
      "step": 208020
    },
    {
      "epoch": 85.15,
      "grad_norm": 1.7521369457244873,
      "learning_rate": 3.0991356397362392e-06,
      "loss": 0.4868,
      "step": 208030
    },
    {
      "epoch": 85.16,
      "grad_norm": 1.6095153093338013,
      "learning_rate": 3.0989854731986424e-06,
      "loss": 0.4807,
      "step": 208040
    },
    {
      "epoch": 85.16,
      "grad_norm": 2.4238369464874268,
      "learning_rate": 3.098835303201429e-06,
      "loss": 0.4936,
      "step": 208050
    },
    {
      "epoch": 85.17,
      "grad_norm": 2.0929601192474365,
      "learning_rate": 3.0986851297452867e-06,
      "loss": 0.4935,
      "step": 208060
    },
    {
      "epoch": 85.17,
      "grad_norm": 1.876047968864441,
      "learning_rate": 3.098534952830904e-06,
      "loss": 0.5084,
      "step": 208070
    },
    {
      "epoch": 85.17,
      "grad_norm": 1.7525070905685425,
      "learning_rate": 3.0983847724589674e-06,
      "loss": 0.4882,
      "step": 208080
    },
    {
      "epoch": 85.18,
      "grad_norm": 1.717581033706665,
      "learning_rate": 3.0982345886301666e-06,
      "loss": 0.4903,
      "step": 208090
    },
    {
      "epoch": 85.18,
      "grad_norm": 1.6771355867385864,
      "learning_rate": 3.0980844013451885e-06,
      "loss": 0.5082,
      "step": 208100
    },
    {
      "epoch": 85.19,
      "grad_norm": 2.4652230739593506,
      "learning_rate": 3.097934210604721e-06,
      "loss": 0.4837,
      "step": 208110
    },
    {
      "epoch": 85.19,
      "grad_norm": 1.6245245933532715,
      "learning_rate": 3.0977840164094534e-06,
      "loss": 0.4731,
      "step": 208120
    },
    {
      "epoch": 85.19,
      "grad_norm": 1.6382489204406738,
      "learning_rate": 3.097633818760072e-06,
      "loss": 0.4804,
      "step": 208130
    },
    {
      "epoch": 85.2,
      "grad_norm": 2.314695119857788,
      "learning_rate": 3.0974836176572656e-06,
      "loss": 0.5037,
      "step": 208140
    },
    {
      "epoch": 85.2,
      "grad_norm": 1.983289122581482,
      "learning_rate": 3.097333413101722e-06,
      "loss": 0.4974,
      "step": 208150
    },
    {
      "epoch": 85.21,
      "grad_norm": 2.8519041538238525,
      "learning_rate": 3.0971832050941305e-06,
      "loss": 0.526,
      "step": 208160
    },
    {
      "epoch": 85.21,
      "grad_norm": 1.6783792972564697,
      "learning_rate": 3.0970329936351776e-06,
      "loss": 0.5146,
      "step": 208170
    },
    {
      "epoch": 85.21,
      "grad_norm": 1.5115752220153809,
      "learning_rate": 3.0968827787255514e-06,
      "loss": 0.4858,
      "step": 208180
    },
    {
      "epoch": 85.22,
      "grad_norm": 1.9090181589126587,
      "learning_rate": 3.0967325603659406e-06,
      "loss": 0.5127,
      "step": 208190
    },
    {
      "epoch": 85.22,
      "grad_norm": 1.6978435516357422,
      "learning_rate": 3.0965823385570337e-06,
      "loss": 0.4949,
      "step": 208200
    },
    {
      "epoch": 85.23,
      "grad_norm": 2.0079469680786133,
      "learning_rate": 3.096432113299518e-06,
      "loss": 0.5013,
      "step": 208210
    },
    {
      "epoch": 85.23,
      "grad_norm": 2.1471805572509766,
      "learning_rate": 3.0962818845940833e-06,
      "loss": 0.4709,
      "step": 208220
    },
    {
      "epoch": 85.24,
      "grad_norm": 1.63491952419281,
      "learning_rate": 3.096131652441416e-06,
      "loss": 0.5065,
      "step": 208230
    },
    {
      "epoch": 85.24,
      "grad_norm": 1.4372447729110718,
      "learning_rate": 3.0959814168422045e-06,
      "loss": 0.4962,
      "step": 208240
    },
    {
      "epoch": 85.24,
      "grad_norm": 2.028008222579956,
      "learning_rate": 3.095831177797138e-06,
      "loss": 0.4748,
      "step": 208250
    },
    {
      "epoch": 85.25,
      "grad_norm": 1.8389323949813843,
      "learning_rate": 3.0956809353069036e-06,
      "loss": 0.5008,
      "step": 208260
    },
    {
      "epoch": 85.25,
      "grad_norm": 1.8489092588424683,
      "learning_rate": 3.0955306893721898e-06,
      "loss": 0.4788,
      "step": 208270
    },
    {
      "epoch": 85.26,
      "grad_norm": 2.1838181018829346,
      "learning_rate": 3.095380439993685e-06,
      "loss": 0.5238,
      "step": 208280
    },
    {
      "epoch": 85.26,
      "grad_norm": 1.655122995376587,
      "learning_rate": 3.095230187172079e-06,
      "loss": 0.5167,
      "step": 208290
    },
    {
      "epoch": 85.26,
      "grad_norm": 1.8208290338516235,
      "learning_rate": 3.095079930908057e-06,
      "loss": 0.5032,
      "step": 208300
    },
    {
      "epoch": 85.27,
      "grad_norm": 1.6612095832824707,
      "learning_rate": 3.0949296712023095e-06,
      "loss": 0.4871,
      "step": 208310
    },
    {
      "epoch": 85.27,
      "grad_norm": 1.9786075353622437,
      "learning_rate": 3.094779408055525e-06,
      "loss": 0.4896,
      "step": 208320
    },
    {
      "epoch": 85.28,
      "grad_norm": 2.3360965251922607,
      "learning_rate": 3.09462914146839e-06,
      "loss": 0.5121,
      "step": 208330
    },
    {
      "epoch": 85.28,
      "grad_norm": 1.844064712524414,
      "learning_rate": 3.094478871441594e-06,
      "loss": 0.4952,
      "step": 208340
    },
    {
      "epoch": 85.28,
      "grad_norm": 1.5755025148391724,
      "learning_rate": 3.094328597975826e-06,
      "loss": 0.4959,
      "step": 208350
    },
    {
      "epoch": 85.29,
      "grad_norm": 2.079848527908325,
      "learning_rate": 3.0941783210717733e-06,
      "loss": 0.5034,
      "step": 208360
    },
    {
      "epoch": 85.29,
      "grad_norm": 1.963291049003601,
      "learning_rate": 3.0940280407301246e-06,
      "loss": 0.4844,
      "step": 208370
    },
    {
      "epoch": 85.3,
      "grad_norm": 1.516798734664917,
      "learning_rate": 3.0938777569515686e-06,
      "loss": 0.4893,
      "step": 208380
    },
    {
      "epoch": 85.3,
      "grad_norm": 1.6471060514450073,
      "learning_rate": 3.093727469736794e-06,
      "loss": 0.4955,
      "step": 208390
    },
    {
      "epoch": 85.3,
      "grad_norm": 1.8759835958480835,
      "learning_rate": 3.093577179086488e-06,
      "loss": 0.5026,
      "step": 208400
    },
    {
      "epoch": 85.31,
      "grad_norm": 1.5232563018798828,
      "learning_rate": 3.0934268850013395e-06,
      "loss": 0.4915,
      "step": 208410
    },
    {
      "epoch": 85.31,
      "grad_norm": 1.73784339427948,
      "learning_rate": 3.093276587482038e-06,
      "loss": 0.502,
      "step": 208420
    },
    {
      "epoch": 85.32,
      "grad_norm": 2.2482404708862305,
      "learning_rate": 3.093126286529271e-06,
      "loss": 0.4931,
      "step": 208430
    },
    {
      "epoch": 85.32,
      "grad_norm": 1.5258830785751343,
      "learning_rate": 3.0929759821437277e-06,
      "loss": 0.496,
      "step": 208440
    },
    {
      "epoch": 85.33,
      "grad_norm": 1.893968939781189,
      "learning_rate": 3.0928256743260963e-06,
      "loss": 0.496,
      "step": 208450
    },
    {
      "epoch": 85.33,
      "grad_norm": 1.7898858785629272,
      "learning_rate": 3.092675363077065e-06,
      "loss": 0.495,
      "step": 208460
    },
    {
      "epoch": 85.33,
      "grad_norm": 1.8360896110534668,
      "learning_rate": 3.0925250483973214e-06,
      "loss": 0.5025,
      "step": 208470
    },
    {
      "epoch": 85.34,
      "grad_norm": 2.0096089839935303,
      "learning_rate": 3.0923747302875565e-06,
      "loss": 0.5245,
      "step": 208480
    },
    {
      "epoch": 85.34,
      "grad_norm": 2.292418956756592,
      "learning_rate": 3.0922244087484576e-06,
      "loss": 0.4893,
      "step": 208490
    },
    {
      "epoch": 85.35,
      "grad_norm": 2.3900930881500244,
      "learning_rate": 3.092074083780713e-06,
      "loss": 0.4799,
      "step": 208500
    },
    {
      "epoch": 85.35,
      "grad_norm": 1.9666630029678345,
      "learning_rate": 3.0919237553850123e-06,
      "loss": 0.5167,
      "step": 208510
    },
    {
      "epoch": 85.35,
      "grad_norm": 1.870396375656128,
      "learning_rate": 3.091773423562043e-06,
      "loss": 0.5052,
      "step": 208520
    },
    {
      "epoch": 85.36,
      "grad_norm": 2.5738608837127686,
      "learning_rate": 3.0916230883124942e-06,
      "loss": 0.4865,
      "step": 208530
    },
    {
      "epoch": 85.36,
      "grad_norm": 1.801321268081665,
      "learning_rate": 3.091472749637055e-06,
      "loss": 0.4921,
      "step": 208540
    },
    {
      "epoch": 85.37,
      "grad_norm": 1.465925693511963,
      "learning_rate": 3.0913224075364133e-06,
      "loss": 0.5218,
      "step": 208550
    },
    {
      "epoch": 85.37,
      "grad_norm": 1.7905954122543335,
      "learning_rate": 3.091172062011258e-06,
      "loss": 0.5116,
      "step": 208560
    },
    {
      "epoch": 85.37,
      "grad_norm": 2.132720947265625,
      "learning_rate": 3.091021713062278e-06,
      "loss": 0.4904,
      "step": 208570
    },
    {
      "epoch": 85.38,
      "grad_norm": 1.726206660270691,
      "learning_rate": 3.090871360690163e-06,
      "loss": 0.4809,
      "step": 208580
    },
    {
      "epoch": 85.38,
      "grad_norm": 1.4240931272506714,
      "learning_rate": 3.0907210048955996e-06,
      "loss": 0.4733,
      "step": 208590
    },
    {
      "epoch": 85.39,
      "grad_norm": 2.0553953647613525,
      "learning_rate": 3.0905706456792783e-06,
      "loss": 0.4947,
      "step": 208600
    },
    {
      "epoch": 85.39,
      "grad_norm": 1.7418088912963867,
      "learning_rate": 3.090420283041887e-06,
      "loss": 0.4913,
      "step": 208610
    },
    {
      "epoch": 85.4,
      "grad_norm": 2.10862135887146,
      "learning_rate": 3.090269916984115e-06,
      "loss": 0.4807,
      "step": 208620
    },
    {
      "epoch": 85.4,
      "grad_norm": 1.8354345560073853,
      "learning_rate": 3.090119547506651e-06,
      "loss": 0.4944,
      "step": 208630
    },
    {
      "epoch": 85.4,
      "grad_norm": 2.2177956104278564,
      "learning_rate": 3.089969174610183e-06,
      "loss": 0.4956,
      "step": 208640
    },
    {
      "epoch": 85.41,
      "grad_norm": 1.8342726230621338,
      "learning_rate": 3.0898187982954013e-06,
      "loss": 0.4999,
      "step": 208650
    },
    {
      "epoch": 85.41,
      "grad_norm": 2.0634191036224365,
      "learning_rate": 3.0896684185629934e-06,
      "loss": 0.4894,
      "step": 208660
    },
    {
      "epoch": 85.42,
      "grad_norm": 2.450711250305176,
      "learning_rate": 3.089518035413649e-06,
      "loss": 0.4981,
      "step": 208670
    },
    {
      "epoch": 85.42,
      "grad_norm": 2.412381410598755,
      "learning_rate": 3.089367648848057e-06,
      "loss": 0.5151,
      "step": 208680
    },
    {
      "epoch": 85.42,
      "grad_norm": 2.343167543411255,
      "learning_rate": 3.0892172588669057e-06,
      "loss": 0.5142,
      "step": 208690
    },
    {
      "epoch": 85.43,
      "grad_norm": 1.5060157775878906,
      "learning_rate": 3.089066865470884e-06,
      "loss": 0.4775,
      "step": 208700
    },
    {
      "epoch": 85.43,
      "grad_norm": 1.9396439790725708,
      "learning_rate": 3.0889164686606817e-06,
      "loss": 0.4905,
      "step": 208710
    },
    {
      "epoch": 85.44,
      "grad_norm": 1.641998529434204,
      "learning_rate": 3.088766068436987e-06,
      "loss": 0.52,
      "step": 208720
    },
    {
      "epoch": 85.44,
      "grad_norm": 1.6059730052947998,
      "learning_rate": 3.0886156648004894e-06,
      "loss": 0.5021,
      "step": 208730
    },
    {
      "epoch": 85.44,
      "grad_norm": 1.9183777570724487,
      "learning_rate": 3.0884652577518775e-06,
      "loss": 0.5051,
      "step": 208740
    },
    {
      "epoch": 85.45,
      "grad_norm": 1.8658337593078613,
      "learning_rate": 3.0883148472918396e-06,
      "loss": 0.5074,
      "step": 208750
    },
    {
      "epoch": 85.45,
      "grad_norm": 1.7175617218017578,
      "learning_rate": 3.088164433421066e-06,
      "loss": 0.4835,
      "step": 208760
    },
    {
      "epoch": 85.46,
      "grad_norm": 2.1441657543182373,
      "learning_rate": 3.088014016140245e-06,
      "loss": 0.5141,
      "step": 208770
    },
    {
      "epoch": 85.46,
      "grad_norm": 1.9302371740341187,
      "learning_rate": 3.0878635954500658e-06,
      "loss": 0.4967,
      "step": 208780
    },
    {
      "epoch": 85.46,
      "grad_norm": 1.9460899829864502,
      "learning_rate": 3.087713171351218e-06,
      "loss": 0.4802,
      "step": 208790
    },
    {
      "epoch": 85.47,
      "grad_norm": 1.691055178642273,
      "learning_rate": 3.0875627438443893e-06,
      "loss": 0.4872,
      "step": 208800
    },
    {
      "epoch": 85.47,
      "grad_norm": 1.4738857746124268,
      "learning_rate": 3.087412312930271e-06,
      "loss": 0.508,
      "step": 208810
    },
    {
      "epoch": 85.48,
      "grad_norm": 1.9139419794082642,
      "learning_rate": 3.0872618786095495e-06,
      "loss": 0.504,
      "step": 208820
    },
    {
      "epoch": 85.48,
      "grad_norm": 1.833746314048767,
      "learning_rate": 3.087111440882916e-06,
      "loss": 0.487,
      "step": 208830
    },
    {
      "epoch": 85.49,
      "grad_norm": 1.739277720451355,
      "learning_rate": 3.086960999751058e-06,
      "loss": 0.4926,
      "step": 208840
    },
    {
      "epoch": 85.49,
      "grad_norm": 1.7636826038360596,
      "learning_rate": 3.0868105552146658e-06,
      "loss": 0.4865,
      "step": 208850
    },
    {
      "epoch": 85.49,
      "grad_norm": 1.8954977989196777,
      "learning_rate": 3.0866601072744283e-06,
      "loss": 0.5061,
      "step": 208860
    },
    {
      "epoch": 85.5,
      "grad_norm": 1.6495903730392456,
      "learning_rate": 3.086509655931035e-06,
      "loss": 0.4988,
      "step": 208870
    },
    {
      "epoch": 85.5,
      "grad_norm": 2.1349644660949707,
      "learning_rate": 3.086359201185175e-06,
      "loss": 0.4833,
      "step": 208880
    },
    {
      "epoch": 85.51,
      "grad_norm": 1.9444042444229126,
      "learning_rate": 3.0862087430375365e-06,
      "loss": 0.488,
      "step": 208890
    },
    {
      "epoch": 85.51,
      "grad_norm": 1.8031138181686401,
      "learning_rate": 3.0860582814888097e-06,
      "loss": 0.5233,
      "step": 208900
    },
    {
      "epoch": 85.51,
      "grad_norm": 1.3634343147277832,
      "learning_rate": 3.0859078165396836e-06,
      "loss": 0.4801,
      "step": 208910
    },
    {
      "epoch": 85.52,
      "grad_norm": 1.839820384979248,
      "learning_rate": 3.0857573481908476e-06,
      "loss": 0.4999,
      "step": 208920
    },
    {
      "epoch": 85.52,
      "grad_norm": 1.7745397090911865,
      "learning_rate": 3.085606876442991e-06,
      "loss": 0.4705,
      "step": 208930
    },
    {
      "epoch": 85.53,
      "grad_norm": 1.7594019174575806,
      "learning_rate": 3.085456401296803e-06,
      "loss": 0.4807,
      "step": 208940
    },
    {
      "epoch": 85.53,
      "grad_norm": 1.746431827545166,
      "learning_rate": 3.085305922752972e-06,
      "loss": 0.5024,
      "step": 208950
    },
    {
      "epoch": 85.53,
      "grad_norm": 1.820857286453247,
      "learning_rate": 3.085155440812189e-06,
      "loss": 0.4899,
      "step": 208960
    },
    {
      "epoch": 85.54,
      "grad_norm": 1.8560471534729004,
      "learning_rate": 3.085004955475142e-06,
      "loss": 0.4752,
      "step": 208970
    },
    {
      "epoch": 85.54,
      "grad_norm": 1.8660073280334473,
      "learning_rate": 3.0848544667425215e-06,
      "loss": 0.4986,
      "step": 208980
    },
    {
      "epoch": 85.55,
      "grad_norm": 2.3265063762664795,
      "learning_rate": 3.084703974615016e-06,
      "loss": 0.5172,
      "step": 208990
    },
    {
      "epoch": 85.55,
      "grad_norm": 1.6574981212615967,
      "learning_rate": 3.0845534790933142e-06,
      "loss": 0.495,
      "step": 209000
    },
    {
      "epoch": 85.55,
      "grad_norm": 1.752130389213562,
      "learning_rate": 3.0844029801781076e-06,
      "loss": 0.5065,
      "step": 209010
    },
    {
      "epoch": 85.56,
      "grad_norm": 2.014246940612793,
      "learning_rate": 3.084252477870084e-06,
      "loss": 0.4939,
      "step": 209020
    },
    {
      "epoch": 85.56,
      "grad_norm": 2.2874419689178467,
      "learning_rate": 3.0841019721699337e-06,
      "loss": 0.5307,
      "step": 209030
    },
    {
      "epoch": 85.57,
      "grad_norm": 1.9874253273010254,
      "learning_rate": 3.083951463078344e-06,
      "loss": 0.4969,
      "step": 209040
    },
    {
      "epoch": 85.57,
      "grad_norm": 2.118593454360962,
      "learning_rate": 3.0838009505960074e-06,
      "loss": 0.4947,
      "step": 209050
    },
    {
      "epoch": 85.58,
      "grad_norm": 2.2583987712860107,
      "learning_rate": 3.083650434723612e-06,
      "loss": 0.4937,
      "step": 209060
    },
    {
      "epoch": 85.58,
      "grad_norm": 1.950290560722351,
      "learning_rate": 3.083499915461847e-06,
      "loss": 0.4839,
      "step": 209070
    },
    {
      "epoch": 85.58,
      "grad_norm": 2.760160446166992,
      "learning_rate": 3.0833493928114023e-06,
      "loss": 0.4803,
      "step": 209080
    },
    {
      "epoch": 85.59,
      "grad_norm": 1.924529790878296,
      "learning_rate": 3.083198866772967e-06,
      "loss": 0.4971,
      "step": 209090
    },
    {
      "epoch": 85.59,
      "grad_norm": 2.1309149265289307,
      "learning_rate": 3.0830483373472317e-06,
      "loss": 0.4996,
      "step": 209100
    },
    {
      "epoch": 85.6,
      "grad_norm": 1.6794286966323853,
      "learning_rate": 3.082897804534885e-06,
      "loss": 0.506,
      "step": 209110
    },
    {
      "epoch": 85.6,
      "grad_norm": 1.899614930152893,
      "learning_rate": 3.082747268336616e-06,
      "loss": 0.4926,
      "step": 209120
    },
    {
      "epoch": 85.6,
      "grad_norm": 1.9373942613601685,
      "learning_rate": 3.082596728753115e-06,
      "loss": 0.4972,
      "step": 209130
    },
    {
      "epoch": 85.61,
      "grad_norm": 2.271226644515991,
      "learning_rate": 3.082446185785072e-06,
      "loss": 0.5074,
      "step": 209140
    },
    {
      "epoch": 85.61,
      "grad_norm": 2.2179229259490967,
      "learning_rate": 3.0822956394331765e-06,
      "loss": 0.4804,
      "step": 209150
    },
    {
      "epoch": 85.62,
      "grad_norm": 2.375521183013916,
      "learning_rate": 3.0821450896981174e-06,
      "loss": 0.5075,
      "step": 209160
    },
    {
      "epoch": 85.62,
      "grad_norm": 1.8017743825912476,
      "learning_rate": 3.081994536580585e-06,
      "loss": 0.4784,
      "step": 209170
    },
    {
      "epoch": 85.62,
      "grad_norm": 1.9017763137817383,
      "learning_rate": 3.0818439800812687e-06,
      "loss": 0.4954,
      "step": 209180
    },
    {
      "epoch": 85.63,
      "grad_norm": 2.183187961578369,
      "learning_rate": 3.0816934202008577e-06,
      "loss": 0.5167,
      "step": 209190
    },
    {
      "epoch": 85.63,
      "grad_norm": 1.8249620199203491,
      "learning_rate": 3.0815428569400423e-06,
      "loss": 0.4865,
      "step": 209200
    },
    {
      "epoch": 85.64,
      "grad_norm": 1.2738631963729858,
      "learning_rate": 3.0813922902995126e-06,
      "loss": 0.4995,
      "step": 209210
    },
    {
      "epoch": 85.64,
      "grad_norm": 1.8826744556427002,
      "learning_rate": 3.0812417202799575e-06,
      "loss": 0.5084,
      "step": 209220
    },
    {
      "epoch": 85.64,
      "grad_norm": 2.3001599311828613,
      "learning_rate": 3.0810911468820665e-06,
      "loss": 0.4961,
      "step": 209230
    },
    {
      "epoch": 85.65,
      "grad_norm": 2.0185389518737793,
      "learning_rate": 3.0809405701065305e-06,
      "loss": 0.5007,
      "step": 209240
    },
    {
      "epoch": 85.65,
      "grad_norm": 2.1477465629577637,
      "learning_rate": 3.0807899899540384e-06,
      "loss": 0.4964,
      "step": 209250
    },
    {
      "epoch": 85.66,
      "grad_norm": 2.0011425018310547,
      "learning_rate": 3.0806394064252806e-06,
      "loss": 0.4981,
      "step": 209260
    },
    {
      "epoch": 85.66,
      "grad_norm": 2.1595678329467773,
      "learning_rate": 3.0804888195209466e-06,
      "loss": 0.4898,
      "step": 209270
    },
    {
      "epoch": 85.67,
      "grad_norm": 2.788604736328125,
      "learning_rate": 3.0803382292417255e-06,
      "loss": 0.5078,
      "step": 209280
    },
    {
      "epoch": 85.67,
      "grad_norm": 1.7150537967681885,
      "learning_rate": 3.0801876355883083e-06,
      "loss": 0.4927,
      "step": 209290
    },
    {
      "epoch": 85.67,
      "grad_norm": 2.1621625423431396,
      "learning_rate": 3.0800370385613843e-06,
      "loss": 0.5141,
      "step": 209300
    },
    {
      "epoch": 85.68,
      "grad_norm": 2.2701869010925293,
      "learning_rate": 3.0798864381616436e-06,
      "loss": 0.5079,
      "step": 209310
    },
    {
      "epoch": 85.68,
      "grad_norm": 2.5783791542053223,
      "learning_rate": 3.0797358343897754e-06,
      "loss": 0.5013,
      "step": 209320
    },
    {
      "epoch": 85.69,
      "grad_norm": 1.767501950263977,
      "learning_rate": 3.0795852272464705e-06,
      "loss": 0.5073,
      "step": 209330
    },
    {
      "epoch": 85.69,
      "grad_norm": 1.8904658555984497,
      "learning_rate": 3.079434616732418e-06,
      "loss": 0.5179,
      "step": 209340
    },
    {
      "epoch": 85.69,
      "grad_norm": 2.0916502475738525,
      "learning_rate": 3.0792840028483087e-06,
      "loss": 0.4902,
      "step": 209350
    },
    {
      "epoch": 85.7,
      "grad_norm": 2.0186612606048584,
      "learning_rate": 3.079133385594831e-06,
      "loss": 0.4795,
      "step": 209360
    },
    {
      "epoch": 85.7,
      "grad_norm": 1.6860206127166748,
      "learning_rate": 3.078982764972677e-06,
      "loss": 0.4962,
      "step": 209370
    },
    {
      "epoch": 85.71,
      "grad_norm": 1.7719167470932007,
      "learning_rate": 3.0788321409825355e-06,
      "loss": 0.5139,
      "step": 209380
    },
    {
      "epoch": 85.71,
      "grad_norm": 2.0235109329223633,
      "learning_rate": 3.0786815136250964e-06,
      "loss": 0.4944,
      "step": 209390
    },
    {
      "epoch": 85.71,
      "grad_norm": 1.7141834497451782,
      "learning_rate": 3.0785308829010498e-06,
      "loss": 0.4686,
      "step": 209400
    },
    {
      "epoch": 85.72,
      "grad_norm": 1.9188235998153687,
      "learning_rate": 3.0783802488110853e-06,
      "loss": 0.5053,
      "step": 209410
    },
    {
      "epoch": 85.72,
      "grad_norm": 1.890610694885254,
      "learning_rate": 3.0782296113558936e-06,
      "loss": 0.495,
      "step": 209420
    },
    {
      "epoch": 85.73,
      "grad_norm": 1.636287808418274,
      "learning_rate": 3.0780789705361646e-06,
      "loss": 0.5142,
      "step": 209430
    },
    {
      "epoch": 85.73,
      "grad_norm": 2.1844704151153564,
      "learning_rate": 3.0779283263525883e-06,
      "loss": 0.4923,
      "step": 209440
    },
    {
      "epoch": 85.73,
      "grad_norm": 1.7335188388824463,
      "learning_rate": 3.0777776788058554e-06,
      "loss": 0.4932,
      "step": 209450
    },
    {
      "epoch": 85.74,
      "grad_norm": 1.7191309928894043,
      "learning_rate": 3.077627027896655e-06,
      "loss": 0.508,
      "step": 209460
    },
    {
      "epoch": 85.74,
      "grad_norm": 1.9991966485977173,
      "learning_rate": 3.0774763736256773e-06,
      "loss": 0.5025,
      "step": 209470
    },
    {
      "epoch": 85.75,
      "grad_norm": 2.1828787326812744,
      "learning_rate": 3.0773257159936136e-06,
      "loss": 0.4923,
      "step": 209480
    },
    {
      "epoch": 85.75,
      "grad_norm": 1.9914016723632812,
      "learning_rate": 3.077175055001152e-06,
      "loss": 0.4925,
      "step": 209490
    },
    {
      "epoch": 85.76,
      "grad_norm": 1.8997384309768677,
      "learning_rate": 3.077024390648985e-06,
      "loss": 0.4974,
      "step": 209500
    },
    {
      "epoch": 85.76,
      "grad_norm": 2.033452272415161,
      "learning_rate": 3.0768737229378e-06,
      "loss": 0.4962,
      "step": 209510
    },
    {
      "epoch": 85.76,
      "grad_norm": 2.0367326736450195,
      "learning_rate": 3.07672305186829e-06,
      "loss": 0.5011,
      "step": 209520
    },
    {
      "epoch": 85.77,
      "grad_norm": 1.671370267868042,
      "learning_rate": 3.076572377441144e-06,
      "loss": 0.5109,
      "step": 209530
    },
    {
      "epoch": 85.77,
      "grad_norm": 1.9787825345993042,
      "learning_rate": 3.0764216996570514e-06,
      "loss": 0.5079,
      "step": 209540
    },
    {
      "epoch": 85.78,
      "grad_norm": 1.594295620918274,
      "learning_rate": 3.0762710185167034e-06,
      "loss": 0.497,
      "step": 209550
    },
    {
      "epoch": 85.78,
      "grad_norm": 1.5490312576293945,
      "learning_rate": 3.0761203340207904e-06,
      "loss": 0.4961,
      "step": 209560
    },
    {
      "epoch": 85.78,
      "grad_norm": 1.7984894514083862,
      "learning_rate": 3.075969646170002e-06,
      "loss": 0.5163,
      "step": 209570
    },
    {
      "epoch": 85.79,
      "grad_norm": 1.847730278968811,
      "learning_rate": 3.075818954965029e-06,
      "loss": 0.4797,
      "step": 209580
    },
    {
      "epoch": 85.79,
      "grad_norm": 1.8135823011398315,
      "learning_rate": 3.0756682604065615e-06,
      "loss": 0.5059,
      "step": 209590
    },
    {
      "epoch": 85.8,
      "grad_norm": 2.0445525646209717,
      "learning_rate": 3.07551756249529e-06,
      "loss": 0.493,
      "step": 209600
    },
    {
      "epoch": 85.8,
      "grad_norm": 2.453284978866577,
      "learning_rate": 3.0753668612319043e-06,
      "loss": 0.5041,
      "step": 209610
    },
    {
      "epoch": 85.8,
      "grad_norm": 1.655084490776062,
      "learning_rate": 3.0752161566170956e-06,
      "loss": 0.4984,
      "step": 209620
    },
    {
      "epoch": 85.81,
      "grad_norm": 1.9938395023345947,
      "learning_rate": 3.075065448651553e-06,
      "loss": 0.5074,
      "step": 209630
    },
    {
      "epoch": 85.81,
      "grad_norm": 2.1561663150787354,
      "learning_rate": 3.0749147373359675e-06,
      "loss": 0.4857,
      "step": 209640
    },
    {
      "epoch": 85.82,
      "grad_norm": 2.147381544113159,
      "learning_rate": 3.0747640226710304e-06,
      "loss": 0.5121,
      "step": 209650
    },
    {
      "epoch": 85.82,
      "grad_norm": 1.6076598167419434,
      "learning_rate": 3.0746133046574307e-06,
      "loss": 0.5036,
      "step": 209660
    },
    {
      "epoch": 85.82,
      "grad_norm": 1.840585470199585,
      "learning_rate": 3.0744625832958596e-06,
      "loss": 0.5186,
      "step": 209670
    },
    {
      "epoch": 85.83,
      "grad_norm": 3.0821218490600586,
      "learning_rate": 3.0743118585870074e-06,
      "loss": 0.5033,
      "step": 209680
    },
    {
      "epoch": 85.83,
      "grad_norm": 1.8828736543655396,
      "learning_rate": 3.0741611305315644e-06,
      "loss": 0.49,
      "step": 209690
    },
    {
      "epoch": 85.84,
      "grad_norm": 1.7017693519592285,
      "learning_rate": 3.0740103991302206e-06,
      "loss": 0.487,
      "step": 209700
    },
    {
      "epoch": 85.84,
      "grad_norm": 1.63897705078125,
      "learning_rate": 3.073859664383667e-06,
      "loss": 0.497,
      "step": 209710
    },
    {
      "epoch": 85.85,
      "grad_norm": 1.6545230150222778,
      "learning_rate": 3.0737089262925943e-06,
      "loss": 0.4794,
      "step": 209720
    },
    {
      "epoch": 85.85,
      "grad_norm": 1.5750269889831543,
      "learning_rate": 3.0735581848576933e-06,
      "loss": 0.4885,
      "step": 209730
    },
    {
      "epoch": 85.85,
      "grad_norm": 2.200892686843872,
      "learning_rate": 3.073407440079654e-06,
      "loss": 0.4996,
      "step": 209740
    },
    {
      "epoch": 85.86,
      "grad_norm": 1.6077038049697876,
      "learning_rate": 3.0732566919591674e-06,
      "loss": 0.4993,
      "step": 209750
    },
    {
      "epoch": 85.86,
      "grad_norm": 1.9967671632766724,
      "learning_rate": 3.0731059404969227e-06,
      "loss": 0.5049,
      "step": 209760
    },
    {
      "epoch": 85.87,
      "grad_norm": 2.2042922973632812,
      "learning_rate": 3.0729551856936116e-06,
      "loss": 0.4922,
      "step": 209770
    },
    {
      "epoch": 85.87,
      "grad_norm": 1.6948137283325195,
      "learning_rate": 3.0728044275499244e-06,
      "loss": 0.5019,
      "step": 209780
    },
    {
      "epoch": 85.87,
      "grad_norm": 1.800445556640625,
      "learning_rate": 3.0726536660665515e-06,
      "loss": 0.4871,
      "step": 209790
    },
    {
      "epoch": 85.88,
      "grad_norm": 2.432572841644287,
      "learning_rate": 3.072502901244184e-06,
      "loss": 0.5039,
      "step": 209800
    },
    {
      "epoch": 85.88,
      "grad_norm": 2.151639938354492,
      "learning_rate": 3.0723521330835123e-06,
      "loss": 0.5001,
      "step": 209810
    },
    {
      "epoch": 85.89,
      "grad_norm": 1.919063925743103,
      "learning_rate": 3.0722013615852284e-06,
      "loss": 0.4951,
      "step": 209820
    },
    {
      "epoch": 85.89,
      "grad_norm": 2.019542932510376,
      "learning_rate": 3.07205058675002e-06,
      "loss": 0.4845,
      "step": 209830
    },
    {
      "epoch": 85.89,
      "grad_norm": 2.705988883972168,
      "learning_rate": 3.07189980857858e-06,
      "loss": 0.4888,
      "step": 209840
    },
    {
      "epoch": 85.9,
      "grad_norm": 2.1047751903533936,
      "learning_rate": 3.0717490270715987e-06,
      "loss": 0.5014,
      "step": 209850
    },
    {
      "epoch": 85.9,
      "grad_norm": 1.6069862842559814,
      "learning_rate": 3.0715982422297663e-06,
      "loss": 0.4982,
      "step": 209860
    },
    {
      "epoch": 85.91,
      "grad_norm": 1.7836533784866333,
      "learning_rate": 3.071447454053774e-06,
      "loss": 0.477,
      "step": 209870
    },
    {
      "epoch": 85.91,
      "grad_norm": 1.8724454641342163,
      "learning_rate": 3.0712966625443115e-06,
      "loss": 0.5121,
      "step": 209880
    },
    {
      "epoch": 85.91,
      "grad_norm": 1.9396038055419922,
      "learning_rate": 3.0711458677020715e-06,
      "loss": 0.4739,
      "step": 209890
    },
    {
      "epoch": 85.92,
      "grad_norm": 2.2623543739318848,
      "learning_rate": 3.070995069527743e-06,
      "loss": 0.5082,
      "step": 209900
    },
    {
      "epoch": 85.92,
      "grad_norm": 1.9818674325942993,
      "learning_rate": 3.070844268022018e-06,
      "loss": 0.4879,
      "step": 209910
    },
    {
      "epoch": 85.93,
      "grad_norm": 1.5235830545425415,
      "learning_rate": 3.070693463185586e-06,
      "loss": 0.4932,
      "step": 209920
    },
    {
      "epoch": 85.93,
      "grad_norm": 2.258375644683838,
      "learning_rate": 3.07054265501914e-06,
      "loss": 0.4962,
      "step": 209930
    },
    {
      "epoch": 85.94,
      "grad_norm": 1.7864807844161987,
      "learning_rate": 3.070391843523368e-06,
      "loss": 0.5097,
      "step": 209940
    },
    {
      "epoch": 85.94,
      "grad_norm": 2.1835172176361084,
      "learning_rate": 3.070241028698963e-06,
      "loss": 0.4958,
      "step": 209950
    },
    {
      "epoch": 85.94,
      "grad_norm": 1.91349196434021,
      "learning_rate": 3.0700902105466144e-06,
      "loss": 0.4916,
      "step": 209960
    },
    {
      "epoch": 85.95,
      "grad_norm": 1.5284042358398438,
      "learning_rate": 3.069939389067015e-06,
      "loss": 0.5096,
      "step": 209970
    },
    {
      "epoch": 85.95,
      "grad_norm": 1.6266891956329346,
      "learning_rate": 3.0697885642608525e-06,
      "loss": 0.49,
      "step": 209980
    },
    {
      "epoch": 85.96,
      "grad_norm": 2.0586471557617188,
      "learning_rate": 3.0696377361288213e-06,
      "loss": 0.4913,
      "step": 209990
    },
    {
      "epoch": 85.96,
      "grad_norm": 1.760654091835022,
      "learning_rate": 3.069486904671611e-06,
      "loss": 0.4972,
      "step": 210000
    },
    {
      "epoch": 85.96,
      "grad_norm": 2.461639881134033,
      "learning_rate": 3.0693360698899113e-06,
      "loss": 0.4996,
      "step": 210010
    },
    {
      "epoch": 85.97,
      "grad_norm": 1.6318089962005615,
      "learning_rate": 3.0691852317844144e-06,
      "loss": 0.4724,
      "step": 210020
    },
    {
      "epoch": 85.97,
      "grad_norm": 1.5261150598526,
      "learning_rate": 3.0690343903558116e-06,
      "loss": 0.4885,
      "step": 210030
    },
    {
      "epoch": 85.98,
      "grad_norm": 1.8630249500274658,
      "learning_rate": 3.068883545604794e-06,
      "loss": 0.4996,
      "step": 210040
    },
    {
      "epoch": 85.98,
      "grad_norm": 1.7941627502441406,
      "learning_rate": 3.068732697532051e-06,
      "loss": 0.5166,
      "step": 210050
    },
    {
      "epoch": 85.98,
      "grad_norm": 1.5604665279388428,
      "learning_rate": 3.0685818461382745e-06,
      "loss": 0.493,
      "step": 210060
    },
    {
      "epoch": 85.99,
      "grad_norm": 1.8881292343139648,
      "learning_rate": 3.068430991424155e-06,
      "loss": 0.4806,
      "step": 210070
    },
    {
      "epoch": 85.99,
      "grad_norm": 2.020043134689331,
      "learning_rate": 3.068280133390385e-06,
      "loss": 0.4967,
      "step": 210080
    },
    {
      "epoch": 86.0,
      "grad_norm": 2.2215023040771484,
      "learning_rate": 3.0681292720376538e-06,
      "loss": 0.4894,
      "step": 210090
    },
    {
      "epoch": 86.0,
      "eval_loss": 0.4993281066417694,
      "eval_runtime": 74.8266,
      "eval_samples_per_second": 46.093,
      "eval_steps_per_second": 5.773,
      "step": 210098
    },
    {
      "epoch": 86.0,
      "grad_norm": 2.171912908554077,
      "learning_rate": 3.0679784073666545e-06,
      "loss": 0.4999,
      "step": 210100
    },
    {
      "epoch": 86.0,
      "grad_norm": 1.9149689674377441,
      "learning_rate": 3.0678275393780763e-06,
      "loss": 0.508,
      "step": 210110
    },
    {
      "epoch": 86.01,
      "grad_norm": 2.3913986682891846,
      "learning_rate": 3.0676766680726114e-06,
      "loss": 0.4903,
      "step": 210120
    },
    {
      "epoch": 86.01,
      "grad_norm": 1.8224669694900513,
      "learning_rate": 3.0675257934509503e-06,
      "loss": 0.4941,
      "step": 210130
    },
    {
      "epoch": 86.02,
      "grad_norm": 1.5604140758514404,
      "learning_rate": 3.0673749155137846e-06,
      "loss": 0.5109,
      "step": 210140
    },
    {
      "epoch": 86.02,
      "grad_norm": 1.9589166641235352,
      "learning_rate": 3.067224034261805e-06,
      "loss": 0.4779,
      "step": 210150
    },
    {
      "epoch": 86.03,
      "grad_norm": 1.6743550300598145,
      "learning_rate": 3.0670731496957026e-06,
      "loss": 0.4976,
      "step": 210160
    },
    {
      "epoch": 86.03,
      "grad_norm": 1.59425687789917,
      "learning_rate": 3.066922261816169e-06,
      "loss": 0.5095,
      "step": 210170
    },
    {
      "epoch": 86.03,
      "grad_norm": 2.1622872352600098,
      "learning_rate": 3.066771370623896e-06,
      "loss": 0.5042,
      "step": 210180
    },
    {
      "epoch": 86.04,
      "grad_norm": 1.5725423097610474,
      "learning_rate": 3.0666204761195737e-06,
      "loss": 0.508,
      "step": 210190
    },
    {
      "epoch": 86.04,
      "grad_norm": 1.996830701828003,
      "learning_rate": 3.066469578303893e-06,
      "loss": 0.493,
      "step": 210200
    },
    {
      "epoch": 86.05,
      "grad_norm": 1.3119479417800903,
      "learning_rate": 3.0663186771775466e-06,
      "loss": 0.519,
      "step": 210210
    },
    {
      "epoch": 86.05,
      "grad_norm": 1.9202882051467896,
      "learning_rate": 3.066167772741225e-06,
      "loss": 0.482,
      "step": 210220
    },
    {
      "epoch": 86.05,
      "grad_norm": 1.5106074810028076,
      "learning_rate": 3.0660168649956196e-06,
      "loss": 0.4773,
      "step": 210230
    },
    {
      "epoch": 86.06,
      "grad_norm": 1.7032345533370972,
      "learning_rate": 3.0658659539414206e-06,
      "loss": 0.5002,
      "step": 210240
    },
    {
      "epoch": 86.06,
      "grad_norm": 1.7966136932373047,
      "learning_rate": 3.0657150395793207e-06,
      "loss": 0.4937,
      "step": 210250
    },
    {
      "epoch": 86.07,
      "grad_norm": 1.780962586402893,
      "learning_rate": 3.0655641219100116e-06,
      "loss": 0.4806,
      "step": 210260
    },
    {
      "epoch": 86.07,
      "grad_norm": 2.1308250427246094,
      "learning_rate": 3.065413200934183e-06,
      "loss": 0.5009,
      "step": 210270
    },
    {
      "epoch": 86.07,
      "grad_norm": 2.6795332431793213,
      "learning_rate": 3.0652622766525267e-06,
      "loss": 0.4687,
      "step": 210280
    },
    {
      "epoch": 86.08,
      "grad_norm": 1.9660255908966064,
      "learning_rate": 3.065111349065735e-06,
      "loss": 0.4908,
      "step": 210290
    },
    {
      "epoch": 86.08,
      "grad_norm": 1.7118126153945923,
      "learning_rate": 3.064960418174499e-06,
      "loss": 0.4928,
      "step": 210300
    },
    {
      "epoch": 86.09,
      "grad_norm": 1.802420735359192,
      "learning_rate": 3.0648094839795085e-06,
      "loss": 0.4863,
      "step": 210310
    },
    {
      "epoch": 86.09,
      "grad_norm": 2.13236927986145,
      "learning_rate": 3.064658546481457e-06,
      "loss": 0.4913,
      "step": 210320
    },
    {
      "epoch": 86.09,
      "grad_norm": 1.9231187105178833,
      "learning_rate": 3.0645076056810357e-06,
      "loss": 0.5059,
      "step": 210330
    },
    {
      "epoch": 86.1,
      "grad_norm": 1.546688437461853,
      "learning_rate": 3.064356661578935e-06,
      "loss": 0.4967,
      "step": 210340
    },
    {
      "epoch": 86.1,
      "grad_norm": 1.703304409980774,
      "learning_rate": 3.0642057141758466e-06,
      "loss": 0.5039,
      "step": 210350
    },
    {
      "epoch": 86.11,
      "grad_norm": 2.0049960613250732,
      "learning_rate": 3.0640547634724618e-06,
      "loss": 0.4717,
      "step": 210360
    },
    {
      "epoch": 86.11,
      "grad_norm": 2.160067081451416,
      "learning_rate": 3.0639038094694724e-06,
      "loss": 0.4988,
      "step": 210370
    },
    {
      "epoch": 86.12,
      "grad_norm": 1.8546792268753052,
      "learning_rate": 3.0637528521675705e-06,
      "loss": 0.4927,
      "step": 210380
    },
    {
      "epoch": 86.12,
      "grad_norm": 1.7265231609344482,
      "learning_rate": 3.063601891567447e-06,
      "loss": 0.5027,
      "step": 210390
    },
    {
      "epoch": 86.12,
      "grad_norm": 1.6375138759613037,
      "learning_rate": 3.0634509276697934e-06,
      "loss": 0.4853,
      "step": 210400
    },
    {
      "epoch": 86.13,
      "grad_norm": 1.9788336753845215,
      "learning_rate": 3.063299960475301e-06,
      "loss": 0.4907,
      "step": 210410
    },
    {
      "epoch": 86.13,
      "grad_norm": 2.248368740081787,
      "learning_rate": 3.063148989984662e-06,
      "loss": 0.5002,
      "step": 210420
    },
    {
      "epoch": 86.14,
      "grad_norm": 1.6412758827209473,
      "learning_rate": 3.062998016198568e-06,
      "loss": 0.4889,
      "step": 210430
    },
    {
      "epoch": 86.14,
      "grad_norm": 1.5898816585540771,
      "learning_rate": 3.0628470391177094e-06,
      "loss": 0.4926,
      "step": 210440
    },
    {
      "epoch": 86.14,
      "grad_norm": 1.8219712972640991,
      "learning_rate": 3.0626960587427782e-06,
      "loss": 0.5085,
      "step": 210450
    },
    {
      "epoch": 86.15,
      "grad_norm": 1.9720478057861328,
      "learning_rate": 3.0625450750744674e-06,
      "loss": 0.4815,
      "step": 210460
    },
    {
      "epoch": 86.15,
      "grad_norm": 1.9026364088058472,
      "learning_rate": 3.062394088113468e-06,
      "loss": 0.4839,
      "step": 210470
    },
    {
      "epoch": 86.16,
      "grad_norm": 2.131319046020508,
      "learning_rate": 3.0622430978604707e-06,
      "loss": 0.5084,
      "step": 210480
    },
    {
      "epoch": 86.16,
      "grad_norm": 2.0938150882720947,
      "learning_rate": 3.062092104316168e-06,
      "loss": 0.4857,
      "step": 210490
    },
    {
      "epoch": 86.16,
      "grad_norm": 1.839447259902954,
      "learning_rate": 3.0619411074812515e-06,
      "loss": 0.4978,
      "step": 210500
    },
    {
      "epoch": 86.17,
      "grad_norm": 1.342514991760254,
      "learning_rate": 3.0617901073564127e-06,
      "loss": 0.4768,
      "step": 210510
    },
    {
      "epoch": 86.17,
      "grad_norm": 1.8319859504699707,
      "learning_rate": 3.061639103942343e-06,
      "loss": 0.5136,
      "step": 210520
    },
    {
      "epoch": 86.18,
      "grad_norm": 1.6595858335494995,
      "learning_rate": 3.0614880972397346e-06,
      "loss": 0.491,
      "step": 210530
    },
    {
      "epoch": 86.18,
      "grad_norm": 1.5843394994735718,
      "learning_rate": 3.0613370872492795e-06,
      "loss": 0.4847,
      "step": 210540
    },
    {
      "epoch": 86.19,
      "grad_norm": 1.9592335224151611,
      "learning_rate": 3.0611860739716686e-06,
      "loss": 0.4995,
      "step": 210550
    },
    {
      "epoch": 86.19,
      "grad_norm": 1.5008180141448975,
      "learning_rate": 3.0610350574075944e-06,
      "loss": 0.4943,
      "step": 210560
    },
    {
      "epoch": 86.19,
      "grad_norm": 2.1553351879119873,
      "learning_rate": 3.0608840375577485e-06,
      "loss": 0.4903,
      "step": 210570
    },
    {
      "epoch": 86.2,
      "grad_norm": 2.2144720554351807,
      "learning_rate": 3.0607330144228226e-06,
      "loss": 0.5119,
      "step": 210580
    },
    {
      "epoch": 86.2,
      "grad_norm": 1.9437655210494995,
      "learning_rate": 3.0605819880035083e-06,
      "loss": 0.4885,
      "step": 210590
    },
    {
      "epoch": 86.21,
      "grad_norm": 1.602423071861267,
      "learning_rate": 3.060430958300498e-06,
      "loss": 0.518,
      "step": 210600
    },
    {
      "epoch": 86.21,
      "grad_norm": 1.7746601104736328,
      "learning_rate": 3.060279925314483e-06,
      "loss": 0.4974,
      "step": 210610
    },
    {
      "epoch": 86.21,
      "grad_norm": 1.647759199142456,
      "learning_rate": 3.0601288890461563e-06,
      "loss": 0.4994,
      "step": 210620
    },
    {
      "epoch": 86.22,
      "grad_norm": 2.3534798622131348,
      "learning_rate": 3.059977849496208e-06,
      "loss": 0.4879,
      "step": 210630
    },
    {
      "epoch": 86.22,
      "grad_norm": 1.9995949268341064,
      "learning_rate": 3.0598268066653304e-06,
      "loss": 0.5081,
      "step": 210640
    },
    {
      "epoch": 86.23,
      "grad_norm": 1.8226295709609985,
      "learning_rate": 3.0596757605542165e-06,
      "loss": 0.5064,
      "step": 210650
    },
    {
      "epoch": 86.23,
      "grad_norm": 1.926894187927246,
      "learning_rate": 3.059524711163557e-06,
      "loss": 0.512,
      "step": 210660
    },
    {
      "epoch": 86.23,
      "grad_norm": 1.7467080354690552,
      "learning_rate": 3.059373658494045e-06,
      "loss": 0.497,
      "step": 210670
    },
    {
      "epoch": 86.24,
      "grad_norm": 1.8680564165115356,
      "learning_rate": 3.0592226025463717e-06,
      "loss": 0.5244,
      "step": 210680
    },
    {
      "epoch": 86.24,
      "grad_norm": 2.1417834758758545,
      "learning_rate": 3.0590715433212295e-06,
      "loss": 0.4888,
      "step": 210690
    },
    {
      "epoch": 86.25,
      "grad_norm": 1.7858915328979492,
      "learning_rate": 3.0589204808193094e-06,
      "loss": 0.5088,
      "step": 210700
    },
    {
      "epoch": 86.25,
      "grad_norm": 2.419933795928955,
      "learning_rate": 3.058769415041305e-06,
      "loss": 0.4874,
      "step": 210710
    },
    {
      "epoch": 86.25,
      "grad_norm": 1.6622898578643799,
      "learning_rate": 3.0586183459879066e-06,
      "loss": 0.5098,
      "step": 210720
    },
    {
      "epoch": 86.26,
      "grad_norm": 1.7305299043655396,
      "learning_rate": 3.0584672736598067e-06,
      "loss": 0.4787,
      "step": 210730
    },
    {
      "epoch": 86.26,
      "grad_norm": 1.8563923835754395,
      "learning_rate": 3.058316198057698e-06,
      "loss": 0.5342,
      "step": 210740
    },
    {
      "epoch": 86.27,
      "grad_norm": 1.9244149923324585,
      "learning_rate": 3.058165119182272e-06,
      "loss": 0.494,
      "step": 210750
    },
    {
      "epoch": 86.27,
      "grad_norm": 1.8374834060668945,
      "learning_rate": 3.0580140370342224e-06,
      "loss": 0.4982,
      "step": 210760
    },
    {
      "epoch": 86.28,
      "grad_norm": 1.8678205013275146,
      "learning_rate": 3.0578629516142386e-06,
      "loss": 0.491,
      "step": 210770
    },
    {
      "epoch": 86.28,
      "grad_norm": 1.9553518295288086,
      "learning_rate": 3.0577118629230148e-06,
      "loss": 0.5066,
      "step": 210780
    },
    {
      "epoch": 86.28,
      "grad_norm": 2.0943329334259033,
      "learning_rate": 3.0575607709612417e-06,
      "loss": 0.5079,
      "step": 210790
    },
    {
      "epoch": 86.29,
      "grad_norm": 1.5239946842193604,
      "learning_rate": 3.0574096757296115e-06,
      "loss": 0.4872,
      "step": 210800
    },
    {
      "epoch": 86.29,
      "grad_norm": 1.6209759712219238,
      "learning_rate": 3.0572585772288174e-06,
      "loss": 0.49,
      "step": 210810
    },
    {
      "epoch": 86.3,
      "grad_norm": 2.220945119857788,
      "learning_rate": 3.057107475459551e-06,
      "loss": 0.514,
      "step": 210820
    },
    {
      "epoch": 86.3,
      "grad_norm": 2.109764814376831,
      "learning_rate": 3.0569563704225043e-06,
      "loss": 0.4681,
      "step": 210830
    },
    {
      "epoch": 86.3,
      "grad_norm": 1.8904473781585693,
      "learning_rate": 3.0568052621183702e-06,
      "loss": 0.4784,
      "step": 210840
    },
    {
      "epoch": 86.31,
      "grad_norm": 1.8961071968078613,
      "learning_rate": 3.05665415054784e-06,
      "loss": 0.4739,
      "step": 210850
    },
    {
      "epoch": 86.31,
      "grad_norm": 1.5829371213912964,
      "learning_rate": 3.056503035711606e-06,
      "loss": 0.5001,
      "step": 210860
    },
    {
      "epoch": 86.32,
      "grad_norm": 1.9187334775924683,
      "learning_rate": 3.056351917610361e-06,
      "loss": 0.4786,
      "step": 210870
    },
    {
      "epoch": 86.32,
      "grad_norm": 2.11899471282959,
      "learning_rate": 3.056200796244797e-06,
      "loss": 0.497,
      "step": 210880
    },
    {
      "epoch": 86.32,
      "grad_norm": 2.0838141441345215,
      "learning_rate": 3.0560496716156065e-06,
      "loss": 0.5165,
      "step": 210890
    },
    {
      "epoch": 86.33,
      "grad_norm": 1.5741381645202637,
      "learning_rate": 3.055898543723481e-06,
      "loss": 0.4952,
      "step": 210900
    },
    {
      "epoch": 86.33,
      "grad_norm": 1.8673934936523438,
      "learning_rate": 3.055747412569114e-06,
      "loss": 0.508,
      "step": 210910
    },
    {
      "epoch": 86.34,
      "grad_norm": 2.0327017307281494,
      "learning_rate": 3.0555962781531964e-06,
      "loss": 0.4906,
      "step": 210920
    },
    {
      "epoch": 86.34,
      "grad_norm": 1.6458789110183716,
      "learning_rate": 3.0554451404764216e-06,
      "loss": 0.5097,
      "step": 210930
    },
    {
      "epoch": 86.34,
      "grad_norm": 2.075596332550049,
      "learning_rate": 3.0552939995394818e-06,
      "loss": 0.4872,
      "step": 210940
    },
    {
      "epoch": 86.35,
      "grad_norm": 1.953251838684082,
      "learning_rate": 3.0551428553430686e-06,
      "loss": 0.4892,
      "step": 210950
    },
    {
      "epoch": 86.35,
      "grad_norm": 1.7004057168960571,
      "learning_rate": 3.054991707887875e-06,
      "loss": 0.5214,
      "step": 210960
    },
    {
      "epoch": 86.36,
      "grad_norm": 1.6657609939575195,
      "learning_rate": 3.0548405571745934e-06,
      "loss": 0.515,
      "step": 210970
    },
    {
      "epoch": 86.36,
      "grad_norm": 1.6544454097747803,
      "learning_rate": 3.0546894032039168e-06,
      "loss": 0.4916,
      "step": 210980
    },
    {
      "epoch": 86.37,
      "grad_norm": 1.3026374578475952,
      "learning_rate": 3.054538245976536e-06,
      "loss": 0.4837,
      "step": 210990
    },
    {
      "epoch": 86.37,
      "grad_norm": 1.9965628385543823,
      "learning_rate": 3.0543870854931443e-06,
      "loss": 0.4789,
      "step": 211000
    },
    {
      "epoch": 86.37,
      "grad_norm": 1.5053164958953857,
      "learning_rate": 3.0542359217544345e-06,
      "loss": 0.5088,
      "step": 211010
    },
    {
      "epoch": 86.38,
      "grad_norm": 2.198519706726074,
      "learning_rate": 3.054084754761098e-06,
      "loss": 0.5031,
      "step": 211020
    },
    {
      "epoch": 86.38,
      "grad_norm": 2.048670530319214,
      "learning_rate": 3.053933584513829e-06,
      "loss": 0.4773,
      "step": 211030
    },
    {
      "epoch": 86.39,
      "grad_norm": 2.092449903488159,
      "learning_rate": 3.0537824110133185e-06,
      "loss": 0.4868,
      "step": 211040
    },
    {
      "epoch": 86.39,
      "grad_norm": 1.9668033123016357,
      "learning_rate": 3.05363123426026e-06,
      "loss": 0.5042,
      "step": 211050
    },
    {
      "epoch": 86.39,
      "grad_norm": 2.2976789474487305,
      "learning_rate": 3.053480054255345e-06,
      "loss": 0.5033,
      "step": 211060
    },
    {
      "epoch": 86.4,
      "grad_norm": 1.6254076957702637,
      "learning_rate": 3.0533288709992663e-06,
      "loss": 0.4764,
      "step": 211070
    },
    {
      "epoch": 86.4,
      "grad_norm": 1.7096620798110962,
      "learning_rate": 3.0531776844927176e-06,
      "loss": 0.5233,
      "step": 211080
    },
    {
      "epoch": 86.41,
      "grad_norm": 1.7909941673278809,
      "learning_rate": 3.05302649473639e-06,
      "loss": 0.5023,
      "step": 211090
    },
    {
      "epoch": 86.41,
      "grad_norm": 1.6628127098083496,
      "learning_rate": 3.0528753017309764e-06,
      "loss": 0.4933,
      "step": 211100
    },
    {
      "epoch": 86.41,
      "grad_norm": 2.2579588890075684,
      "learning_rate": 3.0527241054771693e-06,
      "loss": 0.5008,
      "step": 211110
    },
    {
      "epoch": 86.42,
      "grad_norm": 1.738999843597412,
      "learning_rate": 3.0525729059756626e-06,
      "loss": 0.4886,
      "step": 211120
    },
    {
      "epoch": 86.42,
      "grad_norm": 2.499889612197876,
      "learning_rate": 3.0524217032271474e-06,
      "loss": 0.4796,
      "step": 211130
    },
    {
      "epoch": 86.43,
      "grad_norm": 2.4019241333007812,
      "learning_rate": 3.052270497232317e-06,
      "loss": 0.5109,
      "step": 211140
    },
    {
      "epoch": 86.43,
      "grad_norm": 2.170078754425049,
      "learning_rate": 3.0521192879918643e-06,
      "loss": 0.4798,
      "step": 211150
    },
    {
      "epoch": 86.43,
      "grad_norm": 2.5260627269744873,
      "learning_rate": 3.051968075506481e-06,
      "loss": 0.4941,
      "step": 211160
    },
    {
      "epoch": 86.44,
      "grad_norm": 1.5054526329040527,
      "learning_rate": 3.05181685977686e-06,
      "loss": 0.5017,
      "step": 211170
    },
    {
      "epoch": 86.44,
      "grad_norm": 1.8248052597045898,
      "learning_rate": 3.0516656408036956e-06,
      "loss": 0.5011,
      "step": 211180
    },
    {
      "epoch": 86.45,
      "grad_norm": 1.7953904867172241,
      "learning_rate": 3.0515144185876786e-06,
      "loss": 0.4863,
      "step": 211190
    },
    {
      "epoch": 86.45,
      "grad_norm": 2.4456679821014404,
      "learning_rate": 3.0513631931295028e-06,
      "loss": 0.5055,
      "step": 211200
    },
    {
      "epoch": 86.46,
      "grad_norm": 2.025522470474243,
      "learning_rate": 3.0512119644298594e-06,
      "loss": 0.4955,
      "step": 211210
    },
    {
      "epoch": 86.46,
      "grad_norm": 1.627648949623108,
      "learning_rate": 3.051060732489443e-06,
      "loss": 0.5021,
      "step": 211220
    },
    {
      "epoch": 86.46,
      "grad_norm": 1.7991352081298828,
      "learning_rate": 3.050909497308946e-06,
      "loss": 0.5022,
      "step": 211230
    },
    {
      "epoch": 86.47,
      "grad_norm": 1.8681550025939941,
      "learning_rate": 3.0507582588890603e-06,
      "loss": 0.4957,
      "step": 211240
    },
    {
      "epoch": 86.47,
      "grad_norm": 2.0742249488830566,
      "learning_rate": 3.0506070172304798e-06,
      "loss": 0.5042,
      "step": 211250
    },
    {
      "epoch": 86.48,
      "grad_norm": 1.8848942518234253,
      "learning_rate": 3.050455772333897e-06,
      "loss": 0.5218,
      "step": 211260
    },
    {
      "epoch": 86.48,
      "grad_norm": 1.7207551002502441,
      "learning_rate": 3.050304524200004e-06,
      "loss": 0.4834,
      "step": 211270
    },
    {
      "epoch": 86.48,
      "grad_norm": 2.2137866020202637,
      "learning_rate": 3.0501532728294942e-06,
      "loss": 0.5007,
      "step": 211280
    },
    {
      "epoch": 86.49,
      "grad_norm": 2.0239338874816895,
      "learning_rate": 3.0500020182230606e-06,
      "loss": 0.4816,
      "step": 211290
    },
    {
      "epoch": 86.49,
      "grad_norm": 1.6474603414535522,
      "learning_rate": 3.049850760381395e-06,
      "loss": 0.5283,
      "step": 211300
    },
    {
      "epoch": 86.5,
      "grad_norm": 1.5971118211746216,
      "learning_rate": 3.0496994993051917e-06,
      "loss": 0.4885,
      "step": 211310
    },
    {
      "epoch": 86.5,
      "grad_norm": 1.8286560773849487,
      "learning_rate": 3.049548234995143e-06,
      "loss": 0.5017,
      "step": 211320
    },
    {
      "epoch": 86.5,
      "grad_norm": 1.5503997802734375,
      "learning_rate": 3.0493969674519426e-06,
      "loss": 0.4988,
      "step": 211330
    },
    {
      "epoch": 86.51,
      "grad_norm": 1.8065112829208374,
      "learning_rate": 3.0492456966762827e-06,
      "loss": 0.4886,
      "step": 211340
    },
    {
      "epoch": 86.51,
      "grad_norm": 1.7146633863449097,
      "learning_rate": 3.049094422668856e-06,
      "loss": 0.5007,
      "step": 211350
    },
    {
      "epoch": 86.52,
      "grad_norm": 1.9221361875534058,
      "learning_rate": 3.0489431454303553e-06,
      "loss": 0.489,
      "step": 211360
    },
    {
      "epoch": 86.52,
      "grad_norm": 2.3311409950256348,
      "learning_rate": 3.0487918649614744e-06,
      "loss": 0.4897,
      "step": 211370
    },
    {
      "epoch": 86.52,
      "grad_norm": 2.655712366104126,
      "learning_rate": 3.048640581262906e-06,
      "loss": 0.4904,
      "step": 211380
    },
    {
      "epoch": 86.53,
      "grad_norm": 1.829483151435852,
      "learning_rate": 3.0484892943353417e-06,
      "loss": 0.4941,
      "step": 211390
    },
    {
      "epoch": 86.53,
      "grad_norm": 1.5738697052001953,
      "learning_rate": 3.048338004179477e-06,
      "loss": 0.4946,
      "step": 211400
    },
    {
      "epoch": 86.54,
      "grad_norm": 1.778984785079956,
      "learning_rate": 3.048186710796004e-06,
      "loss": 0.4894,
      "step": 211410
    },
    {
      "epoch": 86.54,
      "grad_norm": 1.6134016513824463,
      "learning_rate": 3.0480354141856156e-06,
      "loss": 0.5024,
      "step": 211420
    },
    {
      "epoch": 86.55,
      "grad_norm": 1.5880085229873657,
      "learning_rate": 3.0478841143490045e-06,
      "loss": 0.4925,
      "step": 211430
    },
    {
      "epoch": 86.55,
      "grad_norm": 1.9991739988327026,
      "learning_rate": 3.0477328112868636e-06,
      "loss": 0.5002,
      "step": 211440
    },
    {
      "epoch": 86.55,
      "grad_norm": 2.3709144592285156,
      "learning_rate": 3.047581504999887e-06,
      "loss": 0.4918,
      "step": 211450
    },
    {
      "epoch": 86.56,
      "grad_norm": 1.7212183475494385,
      "learning_rate": 3.047430195488767e-06,
      "loss": 0.496,
      "step": 211460
    },
    {
      "epoch": 86.56,
      "grad_norm": 2.3263931274414062,
      "learning_rate": 3.0472788827541976e-06,
      "loss": 0.5021,
      "step": 211470
    },
    {
      "epoch": 86.57,
      "grad_norm": 2.014026403427124,
      "learning_rate": 3.0471275667968707e-06,
      "loss": 0.4685,
      "step": 211480
    },
    {
      "epoch": 86.57,
      "grad_norm": 1.9229931831359863,
      "learning_rate": 3.04697624761748e-06,
      "loss": 0.4825,
      "step": 211490
    },
    {
      "epoch": 86.57,
      "grad_norm": 1.8881937265396118,
      "learning_rate": 3.046824925216719e-06,
      "loss": 0.495,
      "step": 211500
    },
    {
      "epoch": 86.58,
      "grad_norm": 1.8691127300262451,
      "learning_rate": 3.046673599595281e-06,
      "loss": 0.4888,
      "step": 211510
    },
    {
      "epoch": 86.58,
      "grad_norm": 2.319403886795044,
      "learning_rate": 3.046522270753858e-06,
      "loss": 0.5069,
      "step": 211520
    },
    {
      "epoch": 86.59,
      "grad_norm": 1.9386470317840576,
      "learning_rate": 3.046370938693145e-06,
      "loss": 0.5045,
      "step": 211530
    },
    {
      "epoch": 86.59,
      "grad_norm": 1.8110793828964233,
      "learning_rate": 3.0462196034138334e-06,
      "loss": 0.5055,
      "step": 211540
    },
    {
      "epoch": 86.59,
      "grad_norm": 1.5356348752975464,
      "learning_rate": 3.0460682649166184e-06,
      "loss": 0.4919,
      "step": 211550
    },
    {
      "epoch": 86.6,
      "grad_norm": 2.0328850746154785,
      "learning_rate": 3.0459169232021923e-06,
      "loss": 0.489,
      "step": 211560
    },
    {
      "epoch": 86.6,
      "grad_norm": 2.2481558322906494,
      "learning_rate": 3.045765578271247e-06,
      "loss": 0.4911,
      "step": 211570
    },
    {
      "epoch": 86.61,
      "grad_norm": 1.5762252807617188,
      "learning_rate": 3.0456142301244773e-06,
      "loss": 0.518,
      "step": 211580
    },
    {
      "epoch": 86.61,
      "grad_norm": 1.8104352951049805,
      "learning_rate": 3.0454628787625767e-06,
      "loss": 0.4869,
      "step": 211590
    },
    {
      "epoch": 86.61,
      "grad_norm": 2.291383981704712,
      "learning_rate": 3.0453115241862386e-06,
      "loss": 0.5014,
      "step": 211600
    },
    {
      "epoch": 86.62,
      "grad_norm": 1.8819925785064697,
      "learning_rate": 3.045160166396155e-06,
      "loss": 0.4886,
      "step": 211610
    },
    {
      "epoch": 86.62,
      "grad_norm": 2.106943130493164,
      "learning_rate": 3.04500880539302e-06,
      "loss": 0.5008,
      "step": 211620
    },
    {
      "epoch": 86.63,
      "grad_norm": 2.2570960521698,
      "learning_rate": 3.044857441177528e-06,
      "loss": 0.5023,
      "step": 211630
    },
    {
      "epoch": 86.63,
      "grad_norm": 2.103898763656616,
      "learning_rate": 3.0447060737503706e-06,
      "loss": 0.4944,
      "step": 211640
    },
    {
      "epoch": 86.64,
      "grad_norm": 1.859233021736145,
      "learning_rate": 3.0445547031122418e-06,
      "loss": 0.4855,
      "step": 211650
    },
    {
      "epoch": 86.64,
      "grad_norm": 1.6956861019134521,
      "learning_rate": 3.0444033292638352e-06,
      "loss": 0.5035,
      "step": 211660
    },
    {
      "epoch": 86.64,
      "grad_norm": 1.6216167211532593,
      "learning_rate": 3.0442519522058447e-06,
      "loss": 0.4716,
      "step": 211670
    },
    {
      "epoch": 86.65,
      "grad_norm": 1.8880168199539185,
      "learning_rate": 3.0441005719389626e-06,
      "loss": 0.502,
      "step": 211680
    },
    {
      "epoch": 86.65,
      "grad_norm": 1.633652925491333,
      "learning_rate": 3.043949188463883e-06,
      "loss": 0.5156,
      "step": 211690
    },
    {
      "epoch": 86.66,
      "grad_norm": 1.7740199565887451,
      "learning_rate": 3.0437978017813006e-06,
      "loss": 0.4693,
      "step": 211700
    },
    {
      "epoch": 86.66,
      "grad_norm": 1.5230883359909058,
      "learning_rate": 3.043646411891907e-06,
      "loss": 0.4974,
      "step": 211710
    },
    {
      "epoch": 86.66,
      "grad_norm": 2.1677372455596924,
      "learning_rate": 3.0434950187963957e-06,
      "loss": 0.4931,
      "step": 211720
    },
    {
      "epoch": 86.67,
      "grad_norm": 1.8365682363510132,
      "learning_rate": 3.0433436224954615e-06,
      "loss": 0.4919,
      "step": 211730
    },
    {
      "epoch": 86.67,
      "grad_norm": 1.9495795965194702,
      "learning_rate": 3.043192222989797e-06,
      "loss": 0.4876,
      "step": 211740
    },
    {
      "epoch": 86.68,
      "grad_norm": 1.8617595434188843,
      "learning_rate": 3.0430408202800958e-06,
      "loss": 0.5075,
      "step": 211750
    },
    {
      "epoch": 86.68,
      "grad_norm": 2.470937490463257,
      "learning_rate": 3.042889414367052e-06,
      "loss": 0.5108,
      "step": 211760
    },
    {
      "epoch": 86.68,
      "grad_norm": 2.918027877807617,
      "learning_rate": 3.0427380052513585e-06,
      "loss": 0.5036,
      "step": 211770
    },
    {
      "epoch": 86.69,
      "grad_norm": 1.8804576396942139,
      "learning_rate": 3.0425865929337098e-06,
      "loss": 0.4947,
      "step": 211780
    },
    {
      "epoch": 86.69,
      "grad_norm": 1.897076964378357,
      "learning_rate": 3.0424351774147983e-06,
      "loss": 0.49,
      "step": 211790
    },
    {
      "epoch": 86.7,
      "grad_norm": 1.8884800672531128,
      "learning_rate": 3.042283758695318e-06,
      "loss": 0.5143,
      "step": 211800
    },
    {
      "epoch": 86.7,
      "grad_norm": 2.4253652095794678,
      "learning_rate": 3.042132336775963e-06,
      "loss": 0.509,
      "step": 211810
    },
    {
      "epoch": 86.7,
      "grad_norm": 2.240175724029541,
      "learning_rate": 3.041980911657427e-06,
      "loss": 0.5123,
      "step": 211820
    },
    {
      "epoch": 86.71,
      "grad_norm": 1.8951457738876343,
      "learning_rate": 3.0418294833404032e-06,
      "loss": 0.4886,
      "step": 211830
    },
    {
      "epoch": 86.71,
      "grad_norm": 1.8123483657836914,
      "learning_rate": 3.041678051825585e-06,
      "loss": 0.4921,
      "step": 211840
    },
    {
      "epoch": 86.72,
      "grad_norm": 3.252723217010498,
      "learning_rate": 3.0415266171136666e-06,
      "loss": 0.4917,
      "step": 211850
    },
    {
      "epoch": 86.72,
      "grad_norm": 1.9506839513778687,
      "learning_rate": 3.041375179205343e-06,
      "loss": 0.513,
      "step": 211860
    },
    {
      "epoch": 86.73,
      "grad_norm": 2.006018877029419,
      "learning_rate": 3.041223738101305e-06,
      "loss": 0.4923,
      "step": 211870
    },
    {
      "epoch": 86.73,
      "grad_norm": 2.1816964149475098,
      "learning_rate": 3.0410722938022475e-06,
      "loss": 0.4917,
      "step": 211880
    },
    {
      "epoch": 86.73,
      "grad_norm": 2.2108824253082275,
      "learning_rate": 3.0409208463088655e-06,
      "loss": 0.489,
      "step": 211890
    },
    {
      "epoch": 86.74,
      "grad_norm": 1.7288188934326172,
      "learning_rate": 3.040769395621851e-06,
      "loss": 0.4806,
      "step": 211900
    },
    {
      "epoch": 86.74,
      "grad_norm": 2.183964252471924,
      "learning_rate": 3.040617941741899e-06,
      "loss": 0.4826,
      "step": 211910
    },
    {
      "epoch": 86.75,
      "grad_norm": 1.7348603010177612,
      "learning_rate": 3.0404664846697035e-06,
      "loss": 0.4873,
      "step": 211920
    },
    {
      "epoch": 86.75,
      "grad_norm": 1.9713754653930664,
      "learning_rate": 3.0403150244059573e-06,
      "loss": 0.5194,
      "step": 211930
    },
    {
      "epoch": 86.75,
      "grad_norm": 1.6819210052490234,
      "learning_rate": 3.040163560951355e-06,
      "loss": 0.4758,
      "step": 211940
    },
    {
      "epoch": 86.76,
      "grad_norm": 1.8470067977905273,
      "learning_rate": 3.0400120943065892e-06,
      "loss": 0.4963,
      "step": 211950
    },
    {
      "epoch": 86.76,
      "grad_norm": 1.5148965120315552,
      "learning_rate": 3.039860624472354e-06,
      "loss": 0.5066,
      "step": 211960
    },
    {
      "epoch": 86.77,
      "grad_norm": 1.7989730834960938,
      "learning_rate": 3.0397091514493448e-06,
      "loss": 0.4841,
      "step": 211970
    },
    {
      "epoch": 86.77,
      "grad_norm": 1.8400096893310547,
      "learning_rate": 3.0395576752382547e-06,
      "loss": 0.4897,
      "step": 211980
    },
    {
      "epoch": 86.77,
      "grad_norm": 2.045024871826172,
      "learning_rate": 3.039406195839777e-06,
      "loss": 0.4931,
      "step": 211990
    },
    {
      "epoch": 86.78,
      "grad_norm": 1.726096510887146,
      "learning_rate": 3.039254713254606e-06,
      "loss": 0.4861,
      "step": 212000
    },
    {
      "epoch": 86.78,
      "grad_norm": 1.5470590591430664,
      "learning_rate": 3.039103227483436e-06,
      "loss": 0.4803,
      "step": 212010
    },
    {
      "epoch": 86.79,
      "grad_norm": 2.348846197128296,
      "learning_rate": 3.0389517385269605e-06,
      "loss": 0.4885,
      "step": 212020
    },
    {
      "epoch": 86.79,
      "grad_norm": 2.010101318359375,
      "learning_rate": 3.038800246385873e-06,
      "loss": 0.4953,
      "step": 212030
    },
    {
      "epoch": 86.79,
      "grad_norm": 1.8993922472000122,
      "learning_rate": 3.0386487510608686e-06,
      "loss": 0.5144,
      "step": 212040
    },
    {
      "epoch": 86.8,
      "grad_norm": 1.4728089570999146,
      "learning_rate": 3.03849725255264e-06,
      "loss": 0.4888,
      "step": 212050
    },
    {
      "epoch": 86.8,
      "grad_norm": 1.6837254762649536,
      "learning_rate": 3.038345750861882e-06,
      "loss": 0.4749,
      "step": 212060
    },
    {
      "epoch": 86.81,
      "grad_norm": 2.663236618041992,
      "learning_rate": 3.0381942459892885e-06,
      "loss": 0.5067,
      "step": 212070
    },
    {
      "epoch": 86.81,
      "grad_norm": 2.033269166946411,
      "learning_rate": 3.0380427379355537e-06,
      "loss": 0.4938,
      "step": 212080
    },
    {
      "epoch": 86.82,
      "grad_norm": 1.9860659837722778,
      "learning_rate": 3.037891226701371e-06,
      "loss": 0.485,
      "step": 212090
    },
    {
      "epoch": 86.82,
      "grad_norm": 1.530887484550476,
      "learning_rate": 3.0377397122874353e-06,
      "loss": 0.4967,
      "step": 212100
    },
    {
      "epoch": 86.82,
      "grad_norm": 2.287562131881714,
      "learning_rate": 3.0375881946944396e-06,
      "loss": 0.4931,
      "step": 212110
    },
    {
      "epoch": 86.83,
      "grad_norm": 1.9615163803100586,
      "learning_rate": 3.037436673923079e-06,
      "loss": 0.4891,
      "step": 212120
    },
    {
      "epoch": 86.83,
      "grad_norm": 1.9607727527618408,
      "learning_rate": 3.037285149974047e-06,
      "loss": 0.4993,
      "step": 212130
    },
    {
      "epoch": 86.84,
      "grad_norm": 1.8645480871200562,
      "learning_rate": 3.0371336228480383e-06,
      "loss": 0.516,
      "step": 212140
    },
    {
      "epoch": 86.84,
      "grad_norm": 1.6353881359100342,
      "learning_rate": 3.0369820925457464e-06,
      "loss": 0.4929,
      "step": 212150
    },
    {
      "epoch": 86.84,
      "grad_norm": 2.784005880355835,
      "learning_rate": 3.036830559067865e-06,
      "loss": 0.4823,
      "step": 212160
    },
    {
      "epoch": 86.85,
      "grad_norm": 1.662654995918274,
      "learning_rate": 3.03667902241509e-06,
      "loss": 0.5065,
      "step": 212170
    },
    {
      "epoch": 86.85,
      "grad_norm": 1.799904465675354,
      "learning_rate": 3.0365274825881136e-06,
      "loss": 0.4767,
      "step": 212180
    },
    {
      "epoch": 86.86,
      "grad_norm": 1.797394871711731,
      "learning_rate": 3.0363759395876316e-06,
      "loss": 0.4889,
      "step": 212190
    },
    {
      "epoch": 86.86,
      "grad_norm": 1.873124599456787,
      "learning_rate": 3.0362243934143364e-06,
      "loss": 0.4996,
      "step": 212200
    },
    {
      "epoch": 86.86,
      "grad_norm": 2.1409385204315186,
      "learning_rate": 3.0360728440689245e-06,
      "loss": 0.4919,
      "step": 212210
    },
    {
      "epoch": 86.87,
      "grad_norm": 2.4835116863250732,
      "learning_rate": 3.0359212915520887e-06,
      "loss": 0.4812,
      "step": 212220
    },
    {
      "epoch": 86.87,
      "grad_norm": 1.6346977949142456,
      "learning_rate": 3.0357697358645223e-06,
      "loss": 0.5021,
      "step": 212230
    },
    {
      "epoch": 86.88,
      "grad_norm": 1.7830417156219482,
      "learning_rate": 3.0356181770069207e-06,
      "loss": 0.5069,
      "step": 212240
    },
    {
      "epoch": 86.88,
      "grad_norm": 2.225324869155884,
      "learning_rate": 3.035466614979979e-06,
      "loss": 0.4931,
      "step": 212250
    },
    {
      "epoch": 86.88,
      "grad_norm": 1.6075609922409058,
      "learning_rate": 3.0353150497843905e-06,
      "loss": 0.4995,
      "step": 212260
    },
    {
      "epoch": 86.89,
      "grad_norm": 1.8338147401809692,
      "learning_rate": 3.0351634814208493e-06,
      "loss": 0.4812,
      "step": 212270
    },
    {
      "epoch": 86.89,
      "grad_norm": 2.0372257232666016,
      "learning_rate": 3.0350119098900504e-06,
      "loss": 0.5043,
      "step": 212280
    },
    {
      "epoch": 86.9,
      "grad_norm": 1.890772819519043,
      "learning_rate": 3.0348603351926875e-06,
      "loss": 0.5273,
      "step": 212290
    },
    {
      "epoch": 86.9,
      "grad_norm": 1.634902000427246,
      "learning_rate": 3.034708757329455e-06,
      "loss": 0.494,
      "step": 212300
    },
    {
      "epoch": 86.91,
      "grad_norm": 1.8275657892227173,
      "learning_rate": 3.034557176301048e-06,
      "loss": 0.5108,
      "step": 212310
    },
    {
      "epoch": 86.91,
      "grad_norm": 1.8978283405303955,
      "learning_rate": 3.03440559210816e-06,
      "loss": 0.469,
      "step": 212320
    },
    {
      "epoch": 86.91,
      "grad_norm": 2.2949178218841553,
      "learning_rate": 3.0342540047514854e-06,
      "loss": 0.4709,
      "step": 212330
    },
    {
      "epoch": 86.92,
      "grad_norm": 1.9925134181976318,
      "learning_rate": 3.0341024142317185e-06,
      "loss": 0.4908,
      "step": 212340
    },
    {
      "epoch": 86.92,
      "grad_norm": 2.267603635787964,
      "learning_rate": 3.0339508205495554e-06,
      "loss": 0.4994,
      "step": 212350
    },
    {
      "epoch": 86.93,
      "grad_norm": 1.9980483055114746,
      "learning_rate": 3.0337992237056885e-06,
      "loss": 0.4853,
      "step": 212360
    },
    {
      "epoch": 86.93,
      "grad_norm": 1.8060611486434937,
      "learning_rate": 3.0336476237008127e-06,
      "loss": 0.5172,
      "step": 212370
    },
    {
      "epoch": 86.93,
      "grad_norm": 1.8083139657974243,
      "learning_rate": 3.033496020535623e-06,
      "loss": 0.4972,
      "step": 212380
    },
    {
      "epoch": 86.94,
      "grad_norm": 2.020082950592041,
      "learning_rate": 3.033344414210814e-06,
      "loss": 0.4934,
      "step": 212390
    },
    {
      "epoch": 86.94,
      "grad_norm": 2.3054847717285156,
      "learning_rate": 3.0331928047270797e-06,
      "loss": 0.4702,
      "step": 212400
    },
    {
      "epoch": 86.95,
      "grad_norm": 1.9955106973648071,
      "learning_rate": 3.033041192085114e-06,
      "loss": 0.5052,
      "step": 212410
    },
    {
      "epoch": 86.95,
      "grad_norm": 1.7497061491012573,
      "learning_rate": 3.0328895762856124e-06,
      "loss": 0.4826,
      "step": 212420
    },
    {
      "epoch": 86.95,
      "grad_norm": 2.0988306999206543,
      "learning_rate": 3.0327379573292694e-06,
      "loss": 0.4806,
      "step": 212430
    },
    {
      "epoch": 86.96,
      "grad_norm": 1.9036086797714233,
      "learning_rate": 3.032586335216779e-06,
      "loss": 0.4999,
      "step": 212440
    },
    {
      "epoch": 86.96,
      "grad_norm": 2.2569406032562256,
      "learning_rate": 3.0324347099488366e-06,
      "loss": 0.4949,
      "step": 212450
    },
    {
      "epoch": 86.97,
      "grad_norm": 1.5744110345840454,
      "learning_rate": 3.0322830815261355e-06,
      "loss": 0.4874,
      "step": 212460
    },
    {
      "epoch": 86.97,
      "grad_norm": 1.9218196868896484,
      "learning_rate": 3.032131449949371e-06,
      "loss": 0.4894,
      "step": 212470
    },
    {
      "epoch": 86.98,
      "grad_norm": 1.7022266387939453,
      "learning_rate": 3.0319798152192385e-06,
      "loss": 0.5067,
      "step": 212480
    },
    {
      "epoch": 86.98,
      "grad_norm": 1.9432545900344849,
      "learning_rate": 3.0318281773364304e-06,
      "loss": 0.4937,
      "step": 212490
    },
    {
      "epoch": 86.98,
      "grad_norm": 1.427636981010437,
      "learning_rate": 3.0316765363016442e-06,
      "loss": 0.4758,
      "step": 212500
    },
    {
      "epoch": 86.99,
      "grad_norm": 1.5071823596954346,
      "learning_rate": 3.0315248921155723e-06,
      "loss": 0.4909,
      "step": 212510
    },
    {
      "epoch": 86.99,
      "grad_norm": 1.7183135747909546,
      "learning_rate": 3.0313732447789106e-06,
      "loss": 0.489,
      "step": 212520
    },
    {
      "epoch": 87.0,
      "grad_norm": 1.5099287033081055,
      "learning_rate": 3.0312215942923526e-06,
      "loss": 0.5079,
      "step": 212530
    },
    {
      "epoch": 87.0,
      "grad_norm": 2.208876371383667,
      "learning_rate": 3.031069940656594e-06,
      "loss": 0.4801,
      "step": 212540
    },
    {
      "epoch": 87.0,
      "eval_loss": 0.4963769018650055,
      "eval_runtime": 74.0332,
      "eval_samples_per_second": 46.587,
      "eval_steps_per_second": 5.835,
      "step": 212541
    },
    {
      "epoch": 87.0,
      "grad_norm": 1.5078279972076416,
      "learning_rate": 3.030918283872329e-06,
      "loss": 0.4921,
      "step": 212550
    },
    {
      "epoch": 87.01,
      "grad_norm": 2.209437847137451,
      "learning_rate": 3.0307666239402536e-06,
      "loss": 0.4798,
      "step": 212560
    },
    {
      "epoch": 87.01,
      "grad_norm": 1.4650568962097168,
      "learning_rate": 3.0306149608610606e-06,
      "loss": 0.4955,
      "step": 212570
    },
    {
      "epoch": 87.02,
      "grad_norm": 2.0092387199401855,
      "learning_rate": 3.030463294635446e-06,
      "loss": 0.494,
      "step": 212580
    },
    {
      "epoch": 87.02,
      "grad_norm": 1.6784340143203735,
      "learning_rate": 3.0303116252641033e-06,
      "loss": 0.4872,
      "step": 212590
    },
    {
      "epoch": 87.02,
      "grad_norm": 2.0741446018218994,
      "learning_rate": 3.0301599527477287e-06,
      "loss": 0.5064,
      "step": 212600
    },
    {
      "epoch": 87.03,
      "grad_norm": 1.7594941854476929,
      "learning_rate": 3.0300082770870166e-06,
      "loss": 0.5065,
      "step": 212610
    },
    {
      "epoch": 87.03,
      "grad_norm": 2.043226480484009,
      "learning_rate": 3.0298565982826603e-06,
      "loss": 0.4876,
      "step": 212620
    },
    {
      "epoch": 87.04,
      "grad_norm": 1.8324064016342163,
      "learning_rate": 3.0297049163353573e-06,
      "loss": 0.4849,
      "step": 212630
    },
    {
      "epoch": 87.04,
      "grad_norm": 1.761110782623291,
      "learning_rate": 3.029553231245801e-06,
      "loss": 0.4902,
      "step": 212640
    },
    {
      "epoch": 87.04,
      "grad_norm": 2.192192792892456,
      "learning_rate": 3.029401543014686e-06,
      "loss": 0.5059,
      "step": 212650
    },
    {
      "epoch": 87.05,
      "grad_norm": 1.5057884454727173,
      "learning_rate": 3.0292498516427073e-06,
      "loss": 0.4835,
      "step": 212660
    },
    {
      "epoch": 87.05,
      "grad_norm": 1.7848176956176758,
      "learning_rate": 3.0290981571305603e-06,
      "loss": 0.5102,
      "step": 212670
    },
    {
      "epoch": 87.06,
      "grad_norm": 1.6148532629013062,
      "learning_rate": 3.0289464594789395e-06,
      "loss": 0.4843,
      "step": 212680
    },
    {
      "epoch": 87.06,
      "grad_norm": 1.8818837404251099,
      "learning_rate": 3.0287947586885395e-06,
      "loss": 0.4896,
      "step": 212690
    },
    {
      "epoch": 87.07,
      "grad_norm": 1.8714771270751953,
      "learning_rate": 3.0286430547600556e-06,
      "loss": 0.5045,
      "step": 212700
    },
    {
      "epoch": 87.07,
      "grad_norm": 1.8598219156265259,
      "learning_rate": 3.0284913476941833e-06,
      "loss": 0.4893,
      "step": 212710
    },
    {
      "epoch": 87.07,
      "grad_norm": 1.654494047164917,
      "learning_rate": 3.0283396374916158e-06,
      "loss": 0.4996,
      "step": 212720
    },
    {
      "epoch": 87.08,
      "grad_norm": 1.9031999111175537,
      "learning_rate": 3.02818792415305e-06,
      "loss": 0.4819,
      "step": 212730
    },
    {
      "epoch": 87.08,
      "grad_norm": 1.5844676494598389,
      "learning_rate": 3.02803620767918e-06,
      "loss": 0.4982,
      "step": 212740
    },
    {
      "epoch": 87.09,
      "grad_norm": 1.8764293193817139,
      "learning_rate": 3.027884488070701e-06,
      "loss": 0.5071,
      "step": 212750
    },
    {
      "epoch": 87.09,
      "grad_norm": 2.2182178497314453,
      "learning_rate": 3.0277327653283074e-06,
      "loss": 0.4998,
      "step": 212760
    },
    {
      "epoch": 87.09,
      "grad_norm": 1.7650216817855835,
      "learning_rate": 3.0275810394526946e-06,
      "loss": 0.4915,
      "step": 212770
    },
    {
      "epoch": 87.1,
      "grad_norm": 1.9252727031707764,
      "learning_rate": 3.0274293104445583e-06,
      "loss": 0.4817,
      "step": 212780
    },
    {
      "epoch": 87.1,
      "grad_norm": 1.6577550172805786,
      "learning_rate": 3.027277578304593e-06,
      "loss": 0.4816,
      "step": 212790
    },
    {
      "epoch": 87.11,
      "grad_norm": 1.9544109106063843,
      "learning_rate": 3.0271258430334934e-06,
      "loss": 0.4927,
      "step": 212800
    },
    {
      "epoch": 87.11,
      "grad_norm": 1.7801960706710815,
      "learning_rate": 3.0269741046319544e-06,
      "loss": 0.5128,
      "step": 212810
    },
    {
      "epoch": 87.11,
      "grad_norm": 2.201460361480713,
      "learning_rate": 3.0268223631006723e-06,
      "loss": 0.4955,
      "step": 212820
    },
    {
      "epoch": 87.12,
      "grad_norm": 1.6565519571304321,
      "learning_rate": 3.026670618440341e-06,
      "loss": 0.4944,
      "step": 212830
    },
    {
      "epoch": 87.12,
      "grad_norm": 2.2135026454925537,
      "learning_rate": 3.026518870651657e-06,
      "loss": 0.4846,
      "step": 212840
    },
    {
      "epoch": 87.13,
      "grad_norm": 1.6070117950439453,
      "learning_rate": 3.0263671197353137e-06,
      "loss": 0.5091,
      "step": 212850
    },
    {
      "epoch": 87.13,
      "grad_norm": 1.434252381324768,
      "learning_rate": 3.026215365692007e-06,
      "loss": 0.502,
      "step": 212860
    },
    {
      "epoch": 87.13,
      "grad_norm": 1.523860216140747,
      "learning_rate": 3.0260636085224337e-06,
      "loss": 0.4856,
      "step": 212870
    },
    {
      "epoch": 87.14,
      "grad_norm": 1.950290322303772,
      "learning_rate": 3.025911848227286e-06,
      "loss": 0.4796,
      "step": 212880
    },
    {
      "epoch": 87.14,
      "grad_norm": 1.9268237352371216,
      "learning_rate": 3.0257600848072605e-06,
      "loss": 0.4918,
      "step": 212890
    },
    {
      "epoch": 87.15,
      "grad_norm": 1.728816032409668,
      "learning_rate": 3.025608318263052e-06,
      "loss": 0.495,
      "step": 212900
    },
    {
      "epoch": 87.15,
      "grad_norm": 2.1762378215789795,
      "learning_rate": 3.0254565485953566e-06,
      "loss": 0.499,
      "step": 212910
    },
    {
      "epoch": 87.16,
      "grad_norm": 1.619018793106079,
      "learning_rate": 3.02530477580487e-06,
      "loss": 0.4966,
      "step": 212920
    },
    {
      "epoch": 87.16,
      "grad_norm": 1.9644582271575928,
      "learning_rate": 3.025152999892286e-06,
      "loss": 0.4924,
      "step": 212930
    },
    {
      "epoch": 87.16,
      "grad_norm": 1.915185809135437,
      "learning_rate": 3.0250012208583e-06,
      "loss": 0.4896,
      "step": 212940
    },
    {
      "epoch": 87.17,
      "grad_norm": 1.7214945554733276,
      "learning_rate": 3.024849438703608e-06,
      "loss": 0.5024,
      "step": 212950
    },
    {
      "epoch": 87.17,
      "grad_norm": 1.9044313430786133,
      "learning_rate": 3.0246976534289048e-06,
      "loss": 0.4807,
      "step": 212960
    },
    {
      "epoch": 87.18,
      "grad_norm": 1.9599652290344238,
      "learning_rate": 3.0245458650348857e-06,
      "loss": 0.4898,
      "step": 212970
    },
    {
      "epoch": 87.18,
      "grad_norm": 1.5932481288909912,
      "learning_rate": 3.024394073522247e-06,
      "loss": 0.4798,
      "step": 212980
    },
    {
      "epoch": 87.18,
      "grad_norm": 1.6691056489944458,
      "learning_rate": 3.024242278891682e-06,
      "loss": 0.4876,
      "step": 212990
    },
    {
      "epoch": 87.19,
      "grad_norm": 1.8232353925704956,
      "learning_rate": 3.0240904811438877e-06,
      "loss": 0.4831,
      "step": 213000
    },
    {
      "epoch": 87.19,
      "grad_norm": 1.820975661277771,
      "learning_rate": 3.0239386802795595e-06,
      "loss": 0.5074,
      "step": 213010
    },
    {
      "epoch": 87.2,
      "grad_norm": 2.4562950134277344,
      "learning_rate": 3.0237868762993914e-06,
      "loss": 0.4955,
      "step": 213020
    },
    {
      "epoch": 87.2,
      "grad_norm": 1.7774038314819336,
      "learning_rate": 3.0236350692040805e-06,
      "loss": 0.4985,
      "step": 213030
    },
    {
      "epoch": 87.2,
      "grad_norm": 1.8571877479553223,
      "learning_rate": 3.023483258994321e-06,
      "loss": 0.4936,
      "step": 213040
    },
    {
      "epoch": 87.21,
      "grad_norm": 1.4529128074645996,
      "learning_rate": 3.023331445670809e-06,
      "loss": 0.5038,
      "step": 213050
    },
    {
      "epoch": 87.21,
      "grad_norm": 1.6711732149124146,
      "learning_rate": 3.023179629234239e-06,
      "loss": 0.5027,
      "step": 213060
    },
    {
      "epoch": 87.22,
      "grad_norm": 2.600646495819092,
      "learning_rate": 3.0230278096853078e-06,
      "loss": 0.4904,
      "step": 213070
    },
    {
      "epoch": 87.22,
      "grad_norm": 1.7343194484710693,
      "learning_rate": 3.02287598702471e-06,
      "loss": 0.5025,
      "step": 213080
    },
    {
      "epoch": 87.22,
      "grad_norm": 1.4730260372161865,
      "learning_rate": 3.0227241612531406e-06,
      "loss": 0.4872,
      "step": 213090
    },
    {
      "epoch": 87.23,
      "grad_norm": 2.191147804260254,
      "learning_rate": 3.0225723323712966e-06,
      "loss": 0.485,
      "step": 213100
    },
    {
      "epoch": 87.23,
      "grad_norm": 1.9502969980239868,
      "learning_rate": 3.022420500379872e-06,
      "loss": 0.4932,
      "step": 213110
    },
    {
      "epoch": 87.24,
      "grad_norm": 1.6114153861999512,
      "learning_rate": 3.0222686652795636e-06,
      "loss": 0.5082,
      "step": 213120
    },
    {
      "epoch": 87.24,
      "grad_norm": 2.267461061477661,
      "learning_rate": 3.022116827071066e-06,
      "loss": 0.4983,
      "step": 213130
    },
    {
      "epoch": 87.25,
      "grad_norm": 2.128823757171631,
      "learning_rate": 3.0219649857550744e-06,
      "loss": 0.4971,
      "step": 213140
    },
    {
      "epoch": 87.25,
      "grad_norm": 2.3160412311553955,
      "learning_rate": 3.0218131413322857e-06,
      "loss": 0.4965,
      "step": 213150
    },
    {
      "epoch": 87.25,
      "grad_norm": 1.7176486253738403,
      "learning_rate": 3.0216612938033954e-06,
      "loss": 0.5132,
      "step": 213160
    },
    {
      "epoch": 87.26,
      "grad_norm": 1.678257942199707,
      "learning_rate": 3.0215094431690973e-06,
      "loss": 0.4935,
      "step": 213170
    },
    {
      "epoch": 87.26,
      "grad_norm": 1.7495139837265015,
      "learning_rate": 3.0213575894300884e-06,
      "loss": 0.4952,
      "step": 213180
    },
    {
      "epoch": 87.27,
      "grad_norm": 1.377047061920166,
      "learning_rate": 3.0212057325870637e-06,
      "loss": 0.4829,
      "step": 213190
    },
    {
      "epoch": 87.27,
      "grad_norm": 1.9739365577697754,
      "learning_rate": 3.0210538726407196e-06,
      "loss": 0.4894,
      "step": 213200
    },
    {
      "epoch": 87.27,
      "grad_norm": 1.5698401927947998,
      "learning_rate": 3.0209020095917513e-06,
      "loss": 0.5071,
      "step": 213210
    },
    {
      "epoch": 87.28,
      "grad_norm": 1.8036514520645142,
      "learning_rate": 3.0207501434408553e-06,
      "loss": 0.4906,
      "step": 213220
    },
    {
      "epoch": 87.28,
      "grad_norm": 1.7124584913253784,
      "learning_rate": 3.0205982741887255e-06,
      "loss": 0.4847,
      "step": 213230
    },
    {
      "epoch": 87.29,
      "grad_norm": 1.8674356937408447,
      "learning_rate": 3.0204464018360588e-06,
      "loss": 0.5171,
      "step": 213240
    },
    {
      "epoch": 87.29,
      "grad_norm": 2.0404961109161377,
      "learning_rate": 3.020294526383551e-06,
      "loss": 0.505,
      "step": 213250
    },
    {
      "epoch": 87.29,
      "grad_norm": 1.7920950651168823,
      "learning_rate": 3.0201426478318972e-06,
      "loss": 0.5003,
      "step": 213260
    },
    {
      "epoch": 87.3,
      "grad_norm": 2.029953718185425,
      "learning_rate": 3.0199907661817937e-06,
      "loss": 0.4666,
      "step": 213270
    },
    {
      "epoch": 87.3,
      "grad_norm": 2.092141628265381,
      "learning_rate": 3.019838881433935e-06,
      "loss": 0.5108,
      "step": 213280
    },
    {
      "epoch": 87.31,
      "grad_norm": 2.0139195919036865,
      "learning_rate": 3.0196869935890193e-06,
      "loss": 0.489,
      "step": 213290
    },
    {
      "epoch": 87.31,
      "grad_norm": 1.475988507270813,
      "learning_rate": 3.01953510264774e-06,
      "loss": 0.5006,
      "step": 213300
    },
    {
      "epoch": 87.31,
      "grad_norm": 2.3243017196655273,
      "learning_rate": 3.0193832086107945e-06,
      "loss": 0.4858,
      "step": 213310
    },
    {
      "epoch": 87.32,
      "grad_norm": 1.6494910717010498,
      "learning_rate": 3.0192313114788773e-06,
      "loss": 0.4996,
      "step": 213320
    },
    {
      "epoch": 87.32,
      "grad_norm": 1.9830693006515503,
      "learning_rate": 3.019079411252685e-06,
      "loss": 0.4848,
      "step": 213330
    },
    {
      "epoch": 87.33,
      "grad_norm": 1.9391535520553589,
      "learning_rate": 3.018927507932913e-06,
      "loss": 0.5114,
      "step": 213340
    },
    {
      "epoch": 87.33,
      "grad_norm": 1.9771863222122192,
      "learning_rate": 3.018775601520258e-06,
      "loss": 0.4971,
      "step": 213350
    },
    {
      "epoch": 87.34,
      "grad_norm": 1.7605669498443604,
      "learning_rate": 3.0186236920154145e-06,
      "loss": 0.489,
      "step": 213360
    },
    {
      "epoch": 87.34,
      "grad_norm": 1.4872446060180664,
      "learning_rate": 3.0184717794190795e-06,
      "loss": 0.503,
      "step": 213370
    },
    {
      "epoch": 87.34,
      "grad_norm": 2.0471956729888916,
      "learning_rate": 3.018319863731949e-06,
      "loss": 0.5057,
      "step": 213380
    },
    {
      "epoch": 87.35,
      "grad_norm": 1.6724599599838257,
      "learning_rate": 3.018167944954718e-06,
      "loss": 0.5044,
      "step": 213390
    },
    {
      "epoch": 87.35,
      "grad_norm": 2.0144035816192627,
      "learning_rate": 3.0180160230880825e-06,
      "loss": 0.4904,
      "step": 213400
    },
    {
      "epoch": 87.36,
      "grad_norm": 1.9895228147506714,
      "learning_rate": 3.017864098132739e-06,
      "loss": 0.5057,
      "step": 213410
    },
    {
      "epoch": 87.36,
      "grad_norm": 1.5576249361038208,
      "learning_rate": 3.0177121700893834e-06,
      "loss": 0.5095,
      "step": 213420
    },
    {
      "epoch": 87.36,
      "grad_norm": 2.0587856769561768,
      "learning_rate": 3.0175602389587114e-06,
      "loss": 0.4945,
      "step": 213430
    },
    {
      "epoch": 87.37,
      "grad_norm": 1.8171474933624268,
      "learning_rate": 3.0174083047414187e-06,
      "loss": 0.4786,
      "step": 213440
    },
    {
      "epoch": 87.37,
      "grad_norm": 2.1881234645843506,
      "learning_rate": 3.0172563674382024e-06,
      "loss": 0.4807,
      "step": 213450
    },
    {
      "epoch": 87.38,
      "grad_norm": 2.2248318195343018,
      "learning_rate": 3.017104427049757e-06,
      "loss": 0.498,
      "step": 213460
    },
    {
      "epoch": 87.38,
      "grad_norm": 1.6297314167022705,
      "learning_rate": 3.0169524835767792e-06,
      "loss": 0.4877,
      "step": 213470
    },
    {
      "epoch": 87.38,
      "grad_norm": 2.3370399475097656,
      "learning_rate": 3.0168005370199656e-06,
      "loss": 0.4903,
      "step": 213480
    },
    {
      "epoch": 87.39,
      "grad_norm": 1.5934032201766968,
      "learning_rate": 3.016648587380012e-06,
      "loss": 0.4856,
      "step": 213490
    },
    {
      "epoch": 87.39,
      "grad_norm": 2.3076729774475098,
      "learning_rate": 3.016496634657613e-06,
      "loss": 0.5089,
      "step": 213500
    },
    {
      "epoch": 87.4,
      "grad_norm": 2.1000468730926514,
      "learning_rate": 3.016344678853467e-06,
      "loss": 0.5016,
      "step": 213510
    },
    {
      "epoch": 87.4,
      "grad_norm": 1.6966328620910645,
      "learning_rate": 3.0161927199682687e-06,
      "loss": 0.5169,
      "step": 213520
    },
    {
      "epoch": 87.4,
      "grad_norm": 1.8911938667297363,
      "learning_rate": 3.0160407580027146e-06,
      "loss": 0.4909,
      "step": 213530
    },
    {
      "epoch": 87.41,
      "grad_norm": 2.067162275314331,
      "learning_rate": 3.0158887929575002e-06,
      "loss": 0.4965,
      "step": 213540
    },
    {
      "epoch": 87.41,
      "grad_norm": 1.4781930446624756,
      "learning_rate": 3.015736824833322e-06,
      "loss": 0.5068,
      "step": 213550
    },
    {
      "epoch": 87.42,
      "grad_norm": 2.4963181018829346,
      "learning_rate": 3.0155848536308756e-06,
      "loss": 0.4905,
      "step": 213560
    },
    {
      "epoch": 87.42,
      "grad_norm": 2.481557607650757,
      "learning_rate": 3.015432879350859e-06,
      "loss": 0.4748,
      "step": 213570
    },
    {
      "epoch": 87.43,
      "grad_norm": 2.079028606414795,
      "learning_rate": 3.015280901993967e-06,
      "loss": 0.474,
      "step": 213580
    },
    {
      "epoch": 87.43,
      "grad_norm": 2.326037645339966,
      "learning_rate": 3.015128921560896e-06,
      "loss": 0.4946,
      "step": 213590
    },
    {
      "epoch": 87.43,
      "grad_norm": 1.9430112838745117,
      "learning_rate": 3.014976938052342e-06,
      "loss": 0.4948,
      "step": 213600
    },
    {
      "epoch": 87.44,
      "grad_norm": 1.7781703472137451,
      "learning_rate": 3.014824951469001e-06,
      "loss": 0.4827,
      "step": 213610
    },
    {
      "epoch": 87.44,
      "grad_norm": 1.8900383710861206,
      "learning_rate": 3.0146729618115696e-06,
      "loss": 0.4874,
      "step": 213620
    },
    {
      "epoch": 87.45,
      "grad_norm": 2.1328036785125732,
      "learning_rate": 3.014520969080744e-06,
      "loss": 0.4909,
      "step": 213630
    },
    {
      "epoch": 87.45,
      "grad_norm": 2.108109474182129,
      "learning_rate": 3.014368973277221e-06,
      "loss": 0.4993,
      "step": 213640
    },
    {
      "epoch": 87.45,
      "grad_norm": 2.223921060562134,
      "learning_rate": 3.014216974401696e-06,
      "loss": 0.503,
      "step": 213650
    },
    {
      "epoch": 87.46,
      "grad_norm": 1.837090253829956,
      "learning_rate": 3.0140649724548657e-06,
      "loss": 0.4762,
      "step": 213660
    },
    {
      "epoch": 87.46,
      "grad_norm": 1.8067013025283813,
      "learning_rate": 3.013912967437426e-06,
      "loss": 0.4822,
      "step": 213670
    },
    {
      "epoch": 87.47,
      "grad_norm": 2.2071759700775146,
      "learning_rate": 3.0137609593500743e-06,
      "loss": 0.4968,
      "step": 213680
    },
    {
      "epoch": 87.47,
      "grad_norm": 2.0446693897247314,
      "learning_rate": 3.013608948193505e-06,
      "loss": 0.4872,
      "step": 213690
    },
    {
      "epoch": 87.47,
      "grad_norm": 1.9661831855773926,
      "learning_rate": 3.0134569339684166e-06,
      "loss": 0.4853,
      "step": 213700
    },
    {
      "epoch": 87.48,
      "grad_norm": 1.5599747896194458,
      "learning_rate": 3.0133049166755038e-06,
      "loss": 0.4877,
      "step": 213710
    },
    {
      "epoch": 87.48,
      "grad_norm": 1.6578408479690552,
      "learning_rate": 3.013152896315464e-06,
      "loss": 0.498,
      "step": 213720
    },
    {
      "epoch": 87.49,
      "grad_norm": 1.9674354791641235,
      "learning_rate": 3.013000872888993e-06,
      "loss": 0.5093,
      "step": 213730
    },
    {
      "epoch": 87.49,
      "grad_norm": 2.221468448638916,
      "learning_rate": 3.012848846396788e-06,
      "loss": 0.4851,
      "step": 213740
    },
    {
      "epoch": 87.49,
      "grad_norm": 1.8849601745605469,
      "learning_rate": 3.0126968168395438e-06,
      "loss": 0.4717,
      "step": 213750
    },
    {
      "epoch": 87.5,
      "grad_norm": 1.667554497718811,
      "learning_rate": 3.0125447842179587e-06,
      "loss": 0.4855,
      "step": 213760
    },
    {
      "epoch": 87.5,
      "grad_norm": 2.013761520385742,
      "learning_rate": 3.0123927485327275e-06,
      "loss": 0.5031,
      "step": 213770
    },
    {
      "epoch": 87.51,
      "grad_norm": 2.314018964767456,
      "learning_rate": 3.0122407097845477e-06,
      "loss": 0.506,
      "step": 213780
    },
    {
      "epoch": 87.51,
      "grad_norm": 2.050814390182495,
      "learning_rate": 3.012088667974116e-06,
      "loss": 0.5006,
      "step": 213790
    },
    {
      "epoch": 87.52,
      "grad_norm": 2.183286190032959,
      "learning_rate": 3.011936623102128e-06,
      "loss": 0.4769,
      "step": 213800
    },
    {
      "epoch": 87.52,
      "grad_norm": 2.3845717906951904,
      "learning_rate": 3.011784575169281e-06,
      "loss": 0.5072,
      "step": 213810
    },
    {
      "epoch": 87.52,
      "grad_norm": 2.1679131984710693,
      "learning_rate": 3.0116325241762704e-06,
      "loss": 0.4907,
      "step": 213820
    },
    {
      "epoch": 87.53,
      "grad_norm": 2.1619458198547363,
      "learning_rate": 3.011480470123794e-06,
      "loss": 0.4952,
      "step": 213830
    },
    {
      "epoch": 87.53,
      "grad_norm": 1.4358350038528442,
      "learning_rate": 3.0113284130125466e-06,
      "loss": 0.5095,
      "step": 213840
    },
    {
      "epoch": 87.54,
      "grad_norm": 1.4070205688476562,
      "learning_rate": 3.0111763528432266e-06,
      "loss": 0.5137,
      "step": 213850
    },
    {
      "epoch": 87.54,
      "grad_norm": 1.9805834293365479,
      "learning_rate": 3.0110242896165294e-06,
      "loss": 0.4938,
      "step": 213860
    },
    {
      "epoch": 87.54,
      "grad_norm": 1.4242738485336304,
      "learning_rate": 3.0108722233331533e-06,
      "loss": 0.4783,
      "step": 213870
    },
    {
      "epoch": 87.55,
      "grad_norm": 1.9553180932998657,
      "learning_rate": 3.010720153993793e-06,
      "loss": 0.5053,
      "step": 213880
    },
    {
      "epoch": 87.55,
      "grad_norm": 1.6369096040725708,
      "learning_rate": 3.010568081599145e-06,
      "loss": 0.4955,
      "step": 213890
    },
    {
      "epoch": 87.56,
      "grad_norm": 2.020657539367676,
      "learning_rate": 3.0104160061499077e-06,
      "loss": 0.4768,
      "step": 213900
    },
    {
      "epoch": 87.56,
      "grad_norm": 1.5063161849975586,
      "learning_rate": 3.0102639276467756e-06,
      "loss": 0.4963,
      "step": 213910
    },
    {
      "epoch": 87.56,
      "grad_norm": 2.2198026180267334,
      "learning_rate": 3.010111846090447e-06,
      "loss": 0.5057,
      "step": 213920
    },
    {
      "epoch": 87.57,
      "grad_norm": 1.4926975965499878,
      "learning_rate": 3.0099597614816183e-06,
      "loss": 0.4891,
      "step": 213930
    },
    {
      "epoch": 87.57,
      "grad_norm": 1.3945167064666748,
      "learning_rate": 3.009807673820985e-06,
      "loss": 0.4931,
      "step": 213940
    },
    {
      "epoch": 87.58,
      "grad_norm": 1.8859080076217651,
      "learning_rate": 3.0096555831092453e-06,
      "loss": 0.4857,
      "step": 213950
    },
    {
      "epoch": 87.58,
      "grad_norm": 2.0245742797851562,
      "learning_rate": 3.009503489347095e-06,
      "loss": 0.4988,
      "step": 213960
    },
    {
      "epoch": 87.58,
      "grad_norm": 1.8242521286010742,
      "learning_rate": 3.0093513925352314e-06,
      "loss": 0.4903,
      "step": 213970
    },
    {
      "epoch": 87.59,
      "grad_norm": 2.1392059326171875,
      "learning_rate": 3.0091992926743508e-06,
      "loss": 0.5011,
      "step": 213980
    },
    {
      "epoch": 87.59,
      "grad_norm": 1.9736926555633545,
      "learning_rate": 3.0090471897651497e-06,
      "loss": 0.4897,
      "step": 213990
    },
    {
      "epoch": 87.6,
      "grad_norm": 1.816516399383545,
      "learning_rate": 3.0088950838083254e-06,
      "loss": 0.4871,
      "step": 214000
    },
    {
      "epoch": 87.6,
      "grad_norm": 1.7658389806747437,
      "learning_rate": 3.0087429748045745e-06,
      "loss": 0.4923,
      "step": 214010
    },
    {
      "epoch": 87.61,
      "grad_norm": 1.9211434125900269,
      "learning_rate": 3.0085908627545937e-06,
      "loss": 0.4936,
      "step": 214020
    },
    {
      "epoch": 87.61,
      "grad_norm": 2.125392198562622,
      "learning_rate": 3.0084387476590805e-06,
      "loss": 0.4873,
      "step": 214030
    },
    {
      "epoch": 87.61,
      "grad_norm": 1.8080207109451294,
      "learning_rate": 3.00828662951873e-06,
      "loss": 0.4757,
      "step": 214040
    },
    {
      "epoch": 87.62,
      "grad_norm": 1.945029377937317,
      "learning_rate": 3.0081345083342408e-06,
      "loss": 0.5083,
      "step": 214050
    },
    {
      "epoch": 87.62,
      "grad_norm": 1.7386749982833862,
      "learning_rate": 3.0079823841063084e-06,
      "loss": 0.4974,
      "step": 214060
    },
    {
      "epoch": 87.63,
      "grad_norm": 1.93059504032135,
      "learning_rate": 3.007830256835631e-06,
      "loss": 0.4795,
      "step": 214070
    },
    {
      "epoch": 87.63,
      "grad_norm": 2.721818685531616,
      "learning_rate": 3.007678126522905e-06,
      "loss": 0.4849,
      "step": 214080
    },
    {
      "epoch": 87.63,
      "grad_norm": 1.7634111642837524,
      "learning_rate": 3.007525993168826e-06,
      "loss": 0.4937,
      "step": 214090
    },
    {
      "epoch": 87.64,
      "grad_norm": 2.0588338375091553,
      "learning_rate": 3.0073738567740935e-06,
      "loss": 0.4782,
      "step": 214100
    },
    {
      "epoch": 87.64,
      "grad_norm": 1.8484280109405518,
      "learning_rate": 3.007221717339402e-06,
      "loss": 0.4902,
      "step": 214110
    },
    {
      "epoch": 87.65,
      "grad_norm": 1.4443645477294922,
      "learning_rate": 3.0070695748654495e-06,
      "loss": 0.5036,
      "step": 214120
    },
    {
      "epoch": 87.65,
      "grad_norm": 1.597503900527954,
      "learning_rate": 3.0069174293529315e-06,
      "loss": 0.518,
      "step": 214130
    },
    {
      "epoch": 87.65,
      "grad_norm": 2.1752986907958984,
      "learning_rate": 3.0067652808025476e-06,
      "loss": 0.4923,
      "step": 214140
    },
    {
      "epoch": 87.66,
      "grad_norm": 2.0920395851135254,
      "learning_rate": 3.006613129214993e-06,
      "loss": 0.4958,
      "step": 214150
    },
    {
      "epoch": 87.66,
      "grad_norm": 1.950099229812622,
      "learning_rate": 3.0064609745909656e-06,
      "loss": 0.5109,
      "step": 214160
    },
    {
      "epoch": 87.67,
      "grad_norm": 2.4358510971069336,
      "learning_rate": 3.0063088169311617e-06,
      "loss": 0.5275,
      "step": 214170
    },
    {
      "epoch": 87.67,
      "grad_norm": 1.7155089378356934,
      "learning_rate": 3.006156656236278e-06,
      "loss": 0.4941,
      "step": 214180
    },
    {
      "epoch": 87.67,
      "grad_norm": 1.6646794080734253,
      "learning_rate": 3.006004492507012e-06,
      "loss": 0.4866,
      "step": 214190
    },
    {
      "epoch": 87.68,
      "grad_norm": 1.6786381006240845,
      "learning_rate": 3.0058523257440613e-06,
      "loss": 0.4903,
      "step": 214200
    },
    {
      "epoch": 87.68,
      "grad_norm": 2.0449323654174805,
      "learning_rate": 3.0057001559481218e-06,
      "loss": 0.4778,
      "step": 214210
    },
    {
      "epoch": 87.69,
      "grad_norm": 1.8733749389648438,
      "learning_rate": 3.0055479831198907e-06,
      "loss": 0.4984,
      "step": 214220
    },
    {
      "epoch": 87.69,
      "grad_norm": 1.8839596509933472,
      "learning_rate": 3.0053958072600673e-06,
      "loss": 0.5068,
      "step": 214230
    },
    {
      "epoch": 87.7,
      "grad_norm": 1.4467613697052002,
      "learning_rate": 3.0052436283693457e-06,
      "loss": 0.4682,
      "step": 214240
    },
    {
      "epoch": 87.7,
      "grad_norm": 1.8455272912979126,
      "learning_rate": 3.005091446448424e-06,
      "loss": 0.4879,
      "step": 214250
    },
    {
      "epoch": 87.7,
      "grad_norm": 2.796529769897461,
      "learning_rate": 3.004939261498e-06,
      "loss": 0.4943,
      "step": 214260
    },
    {
      "epoch": 87.71,
      "grad_norm": 2.2974069118499756,
      "learning_rate": 3.0047870735187704e-06,
      "loss": 0.5034,
      "step": 214270
    },
    {
      "epoch": 87.71,
      "grad_norm": 1.9738342761993408,
      "learning_rate": 3.004634882511432e-06,
      "loss": 0.5035,
      "step": 214280
    },
    {
      "epoch": 87.72,
      "grad_norm": 2.1939468383789062,
      "learning_rate": 3.0044826884766827e-06,
      "loss": 0.5044,
      "step": 214290
    },
    {
      "epoch": 87.72,
      "grad_norm": 2.2769157886505127,
      "learning_rate": 3.004330491415219e-06,
      "loss": 0.4986,
      "step": 214300
    },
    {
      "epoch": 87.72,
      "grad_norm": 1.7300208806991577,
      "learning_rate": 3.0041782913277383e-06,
      "loss": 0.4947,
      "step": 214310
    },
    {
      "epoch": 87.73,
      "grad_norm": 1.7666577100753784,
      "learning_rate": 3.004026088214938e-06,
      "loss": 0.4985,
      "step": 214320
    },
    {
      "epoch": 87.73,
      "grad_norm": 2.354781150817871,
      "learning_rate": 3.0038738820775154e-06,
      "loss": 0.5126,
      "step": 214330
    },
    {
      "epoch": 87.74,
      "grad_norm": 1.7220797538757324,
      "learning_rate": 3.0037216729161678e-06,
      "loss": 0.5034,
      "step": 214340
    },
    {
      "epoch": 87.74,
      "grad_norm": 1.8838059902191162,
      "learning_rate": 3.0035694607315914e-06,
      "loss": 0.4876,
      "step": 214350
    },
    {
      "epoch": 87.74,
      "grad_norm": 1.7132667303085327,
      "learning_rate": 3.0034172455244846e-06,
      "loss": 0.4988,
      "step": 214360
    },
    {
      "epoch": 87.75,
      "grad_norm": 1.6146296262741089,
      "learning_rate": 3.003265027295544e-06,
      "loss": 0.5107,
      "step": 214370
    },
    {
      "epoch": 87.75,
      "grad_norm": 1.4889774322509766,
      "learning_rate": 3.003112806045468e-06,
      "loss": 0.5108,
      "step": 214380
    },
    {
      "epoch": 87.76,
      "grad_norm": 1.7456231117248535,
      "learning_rate": 3.0029605817749533e-06,
      "loss": 0.4875,
      "step": 214390
    },
    {
      "epoch": 87.76,
      "grad_norm": 2.1188342571258545,
      "learning_rate": 3.002808354484696e-06,
      "loss": 0.5099,
      "step": 214400
    },
    {
      "epoch": 87.77,
      "grad_norm": 1.5692048072814941,
      "learning_rate": 3.002656124175394e-06,
      "loss": 0.5079,
      "step": 214410
    },
    {
      "epoch": 87.77,
      "grad_norm": 2.299227714538574,
      "learning_rate": 3.002503890847746e-06,
      "loss": 0.5002,
      "step": 214420
    },
    {
      "epoch": 87.77,
      "grad_norm": 2.5466740131378174,
      "learning_rate": 3.0023516545024485e-06,
      "loss": 0.5078,
      "step": 214430
    },
    {
      "epoch": 87.78,
      "grad_norm": 1.9138638973236084,
      "learning_rate": 3.0021994151401987e-06,
      "loss": 0.5057,
      "step": 214440
    },
    {
      "epoch": 87.78,
      "grad_norm": 1.6922434568405151,
      "learning_rate": 3.0020471727616938e-06,
      "loss": 0.4831,
      "step": 214450
    },
    {
      "epoch": 87.79,
      "grad_norm": 2.267916202545166,
      "learning_rate": 3.001894927367632e-06,
      "loss": 0.5105,
      "step": 214460
    },
    {
      "epoch": 87.79,
      "grad_norm": 2.121338129043579,
      "learning_rate": 3.0017426789587104e-06,
      "loss": 0.5013,
      "step": 214470
    },
    {
      "epoch": 87.79,
      "grad_norm": 1.5826538801193237,
      "learning_rate": 3.001590427535626e-06,
      "loss": 0.4903,
      "step": 214480
    },
    {
      "epoch": 87.8,
      "grad_norm": 1.8374682664871216,
      "learning_rate": 3.001438173099076e-06,
      "loss": 0.4929,
      "step": 214490
    },
    {
      "epoch": 87.8,
      "grad_norm": 2.0452582836151123,
      "learning_rate": 3.0012859156497588e-06,
      "loss": 0.4833,
      "step": 214500
    },
    {
      "epoch": 87.81,
      "grad_norm": 1.9234994649887085,
      "learning_rate": 3.0011336551883705e-06,
      "loss": 0.4925,
      "step": 214510
    },
    {
      "epoch": 87.81,
      "grad_norm": 1.8308188915252686,
      "learning_rate": 3.0009813917156104e-06,
      "loss": 0.5049,
      "step": 214520
    },
    {
      "epoch": 87.81,
      "grad_norm": 1.6815707683563232,
      "learning_rate": 3.000829125232175e-06,
      "loss": 0.4855,
      "step": 214530
    },
    {
      "epoch": 87.82,
      "grad_norm": 1.932900309562683,
      "learning_rate": 3.0006768557387617e-06,
      "loss": 0.4823,
      "step": 214540
    },
    {
      "epoch": 87.82,
      "grad_norm": 1.8520386219024658,
      "learning_rate": 3.000524583236068e-06,
      "loss": 0.5021,
      "step": 214550
    },
    {
      "epoch": 87.83,
      "grad_norm": 2.1394753456115723,
      "learning_rate": 3.0003723077247922e-06,
      "loss": 0.4918,
      "step": 214560
    },
    {
      "epoch": 87.83,
      "grad_norm": 2.017130136489868,
      "learning_rate": 3.000220029205631e-06,
      "loss": 0.498,
      "step": 214570
    },
    {
      "epoch": 87.83,
      "grad_norm": 2.1356496810913086,
      "learning_rate": 3.0000677476792826e-06,
      "loss": 0.4953,
      "step": 214580
    },
    {
      "epoch": 87.84,
      "grad_norm": 2.3752942085266113,
      "learning_rate": 2.9999154631464433e-06,
      "loss": 0.4938,
      "step": 214590
    },
    {
      "epoch": 87.84,
      "grad_norm": 1.7484403848648071,
      "learning_rate": 2.9997631756078126e-06,
      "loss": 0.4856,
      "step": 214600
    },
    {
      "epoch": 87.85,
      "grad_norm": 1.5025776624679565,
      "learning_rate": 2.9996108850640866e-06,
      "loss": 0.5022,
      "step": 214610
    },
    {
      "epoch": 87.85,
      "grad_norm": 1.9253273010253906,
      "learning_rate": 2.999458591515964e-06,
      "loss": 0.4831,
      "step": 214620
    },
    {
      "epoch": 87.86,
      "grad_norm": 2.0856056213378906,
      "learning_rate": 2.9993062949641416e-06,
      "loss": 0.4882,
      "step": 214630
    },
    {
      "epoch": 87.86,
      "grad_norm": 1.849158525466919,
      "learning_rate": 2.999153995409317e-06,
      "loss": 0.4804,
      "step": 214640
    },
    {
      "epoch": 87.86,
      "grad_norm": 1.8920031785964966,
      "learning_rate": 2.9990016928521886e-06,
      "loss": 0.4728,
      "step": 214650
    },
    {
      "epoch": 87.87,
      "grad_norm": 1.6046795845031738,
      "learning_rate": 2.9988493872934534e-06,
      "loss": 0.4848,
      "step": 214660
    },
    {
      "epoch": 87.87,
      "grad_norm": 2.6299831867218018,
      "learning_rate": 2.998697078733809e-06,
      "loss": 0.4985,
      "step": 214670
    },
    {
      "epoch": 87.88,
      "grad_norm": 1.6449556350708008,
      "learning_rate": 2.998544767173955e-06,
      "loss": 0.4867,
      "step": 214680
    },
    {
      "epoch": 87.88,
      "grad_norm": 1.6226574182510376,
      "learning_rate": 2.998392452614587e-06,
      "loss": 0.4964,
      "step": 214690
    },
    {
      "epoch": 87.88,
      "grad_norm": 2.1756651401519775,
      "learning_rate": 2.9982401350564018e-06,
      "loss": 0.4798,
      "step": 214700
    },
    {
      "epoch": 87.89,
      "grad_norm": 2.411959648132324,
      "learning_rate": 2.9980878145001e-06,
      "loss": 0.5126,
      "step": 214710
    },
    {
      "epoch": 87.89,
      "grad_norm": 2.229527711868286,
      "learning_rate": 2.997935490946378e-06,
      "loss": 0.5028,
      "step": 214720
    },
    {
      "epoch": 87.9,
      "grad_norm": 1.5593454837799072,
      "learning_rate": 2.9977831643959327e-06,
      "loss": 0.5066,
      "step": 214730
    },
    {
      "epoch": 87.9,
      "grad_norm": 1.6947489976882935,
      "learning_rate": 2.9976308348494632e-06,
      "loss": 0.4886,
      "step": 214740
    },
    {
      "epoch": 87.9,
      "grad_norm": 1.7648658752441406,
      "learning_rate": 2.997478502307668e-06,
      "loss": 0.4765,
      "step": 214750
    },
    {
      "epoch": 87.91,
      "grad_norm": 1.7764636278152466,
      "learning_rate": 2.997326166771243e-06,
      "loss": 0.4895,
      "step": 214760
    },
    {
      "epoch": 87.91,
      "grad_norm": 1.832505226135254,
      "learning_rate": 2.9971738282408862e-06,
      "loss": 0.4649,
      "step": 214770
    },
    {
      "epoch": 87.92,
      "grad_norm": 1.7335093021392822,
      "learning_rate": 2.9970214867172964e-06,
      "loss": 0.48,
      "step": 214780
    },
    {
      "epoch": 87.92,
      "grad_norm": 2.361070394515991,
      "learning_rate": 2.996869142201171e-06,
      "loss": 0.4987,
      "step": 214790
    },
    {
      "epoch": 87.92,
      "grad_norm": 1.764478325843811,
      "learning_rate": 2.996716794693208e-06,
      "loss": 0.4874,
      "step": 214800
    },
    {
      "epoch": 87.93,
      "grad_norm": 2.7887072563171387,
      "learning_rate": 2.9965644441941052e-06,
      "loss": 0.4842,
      "step": 214810
    },
    {
      "epoch": 87.93,
      "grad_norm": 2.2340614795684814,
      "learning_rate": 2.9964120907045614e-06,
      "loss": 0.5132,
      "step": 214820
    },
    {
      "epoch": 87.94,
      "grad_norm": 1.375528335571289,
      "learning_rate": 2.9962597342252725e-06,
      "loss": 0.4898,
      "step": 214830
    },
    {
      "epoch": 87.94,
      "grad_norm": 2.271695137023926,
      "learning_rate": 2.996107374756938e-06,
      "loss": 0.4624,
      "step": 214840
    },
    {
      "epoch": 87.95,
      "grad_norm": 1.9885318279266357,
      "learning_rate": 2.9959550123002555e-06,
      "loss": 0.5081,
      "step": 214850
    },
    {
      "epoch": 87.95,
      "grad_norm": 2.0106282234191895,
      "learning_rate": 2.995802646855923e-06,
      "loss": 0.4896,
      "step": 214860
    },
    {
      "epoch": 87.95,
      "grad_norm": 2.360717296600342,
      "learning_rate": 2.9956502784246377e-06,
      "loss": 0.486,
      "step": 214870
    },
    {
      "epoch": 87.96,
      "grad_norm": 1.8112659454345703,
      "learning_rate": 2.9954979070070986e-06,
      "loss": 0.5189,
      "step": 214880
    },
    {
      "epoch": 87.96,
      "grad_norm": 2.212766647338867,
      "learning_rate": 2.995345532604003e-06,
      "loss": 0.481,
      "step": 214890
    },
    {
      "epoch": 87.97,
      "grad_norm": 1.6975971460342407,
      "learning_rate": 2.9951931552160497e-06,
      "loss": 0.5093,
      "step": 214900
    },
    {
      "epoch": 87.97,
      "grad_norm": 1.5023159980773926,
      "learning_rate": 2.9950407748439354e-06,
      "loss": 0.4878,
      "step": 214910
    },
    {
      "epoch": 87.97,
      "grad_norm": 2.236524820327759,
      "learning_rate": 2.9948883914883592e-06,
      "loss": 0.4837,
      "step": 214920
    },
    {
      "epoch": 87.98,
      "grad_norm": 2.3485817909240723,
      "learning_rate": 2.9947360051500192e-06,
      "loss": 0.4722,
      "step": 214930
    },
    {
      "epoch": 87.98,
      "grad_norm": 1.8071773052215576,
      "learning_rate": 2.994583615829613e-06,
      "loss": 0.4832,
      "step": 214940
    },
    {
      "epoch": 87.99,
      "grad_norm": 2.442387819290161,
      "learning_rate": 2.994431223527839e-06,
      "loss": 0.5162,
      "step": 214950
    },
    {
      "epoch": 87.99,
      "grad_norm": 1.76190185546875,
      "learning_rate": 2.9942788282453945e-06,
      "loss": 0.4977,
      "step": 214960
    },
    {
      "epoch": 87.99,
      "grad_norm": 1.816740870475769,
      "learning_rate": 2.994126429982979e-06,
      "loss": 0.507,
      "step": 214970
    },
    {
      "epoch": 88.0,
      "grad_norm": 2.479827404022217,
      "learning_rate": 2.993974028741289e-06,
      "loss": 0.5019,
      "step": 214980
    },
    {
      "epoch": 88.0,
      "eval_loss": 0.4956228733062744,
      "eval_runtime": 76.8216,
      "eval_samples_per_second": 44.896,
      "eval_steps_per_second": 5.623,
      "step": 214984
    },
    {
      "epoch": 88.0,
      "grad_norm": 1.924480676651001,
      "learning_rate": 2.993821624521024e-06,
      "loss": 0.4782,
      "step": 214990
    },
    {
      "epoch": 88.01,
      "grad_norm": 2.248845338821411,
      "learning_rate": 2.9936692173228816e-06,
      "loss": 0.5093,
      "step": 215000
    },
    {
      "epoch": 88.01,
      "grad_norm": 1.7247058153152466,
      "learning_rate": 2.9935168071475594e-06,
      "loss": 0.4929,
      "step": 215010
    },
    {
      "epoch": 88.01,
      "grad_norm": 1.9038293361663818,
      "learning_rate": 2.993364393995756e-06,
      "loss": 0.4831,
      "step": 215020
    },
    {
      "epoch": 88.02,
      "grad_norm": 1.8870152235031128,
      "learning_rate": 2.9932119778681704e-06,
      "loss": 0.4823,
      "step": 215030
    },
    {
      "epoch": 88.02,
      "grad_norm": 2.1129372119903564,
      "learning_rate": 2.9930595587655e-06,
      "loss": 0.5071,
      "step": 215040
    },
    {
      "epoch": 88.03,
      "grad_norm": 1.9183465242385864,
      "learning_rate": 2.9929071366884428e-06,
      "loss": 0.489,
      "step": 215050
    },
    {
      "epoch": 88.03,
      "grad_norm": 1.7957855463027954,
      "learning_rate": 2.9927547116376973e-06,
      "loss": 0.4924,
      "step": 215060
    },
    {
      "epoch": 88.04,
      "grad_norm": 2.390146255493164,
      "learning_rate": 2.992602283613961e-06,
      "loss": 0.4943,
      "step": 215070
    },
    {
      "epoch": 88.04,
      "grad_norm": 1.6096177101135254,
      "learning_rate": 2.9924498526179333e-06,
      "loss": 0.4801,
      "step": 215080
    },
    {
      "epoch": 88.04,
      "grad_norm": 1.665211796760559,
      "learning_rate": 2.992297418650312e-06,
      "loss": 0.4956,
      "step": 215090
    },
    {
      "epoch": 88.05,
      "grad_norm": 1.8945716619491577,
      "learning_rate": 2.9921449817117962e-06,
      "loss": 0.5123,
      "step": 215100
    },
    {
      "epoch": 88.05,
      "grad_norm": 1.8173006772994995,
      "learning_rate": 2.9919925418030832e-06,
      "loss": 0.4832,
      "step": 215110
    },
    {
      "epoch": 88.06,
      "grad_norm": 1.7315187454223633,
      "learning_rate": 2.9918400989248713e-06,
      "loss": 0.4944,
      "step": 215120
    },
    {
      "epoch": 88.06,
      "grad_norm": 1.665888786315918,
      "learning_rate": 2.9916876530778585e-06,
      "loss": 0.5099,
      "step": 215130
    },
    {
      "epoch": 88.06,
      "grad_norm": 1.9799472093582153,
      "learning_rate": 2.9915352042627444e-06,
      "loss": 0.49,
      "step": 215140
    },
    {
      "epoch": 88.07,
      "grad_norm": 1.3309193849563599,
      "learning_rate": 2.9913827524802266e-06,
      "loss": 0.5122,
      "step": 215150
    },
    {
      "epoch": 88.07,
      "grad_norm": 1.9271090030670166,
      "learning_rate": 2.9912302977310025e-06,
      "loss": 0.4813,
      "step": 215160
    },
    {
      "epoch": 88.08,
      "grad_norm": 1.9142615795135498,
      "learning_rate": 2.991077840015772e-06,
      "loss": 0.4947,
      "step": 215170
    },
    {
      "epoch": 88.08,
      "grad_norm": 2.2736053466796875,
      "learning_rate": 2.9909253793352335e-06,
      "loss": 0.4751,
      "step": 215180
    },
    {
      "epoch": 88.08,
      "grad_norm": 2.0875327587127686,
      "learning_rate": 2.990772915690084e-06,
      "loss": 0.4863,
      "step": 215190
    },
    {
      "epoch": 88.09,
      "grad_norm": 2.7641236782073975,
      "learning_rate": 2.990620449081023e-06,
      "loss": 0.4791,
      "step": 215200
    },
    {
      "epoch": 88.09,
      "grad_norm": 1.9564791917800903,
      "learning_rate": 2.9904679795087487e-06,
      "loss": 0.4972,
      "step": 215210
    },
    {
      "epoch": 88.1,
      "grad_norm": 1.9225437641143799,
      "learning_rate": 2.9903155069739597e-06,
      "loss": 0.4981,
      "step": 215220
    },
    {
      "epoch": 88.1,
      "grad_norm": 1.9317725896835327,
      "learning_rate": 2.990163031477354e-06,
      "loss": 0.4792,
      "step": 215230
    },
    {
      "epoch": 88.1,
      "grad_norm": 1.767772912979126,
      "learning_rate": 2.9900105530196302e-06,
      "loss": 0.5005,
      "step": 215240
    },
    {
      "epoch": 88.11,
      "grad_norm": 1.921555757522583,
      "learning_rate": 2.989858071601487e-06,
      "loss": 0.4929,
      "step": 215250
    },
    {
      "epoch": 88.11,
      "grad_norm": 1.6966722011566162,
      "learning_rate": 2.989705587223623e-06,
      "loss": 0.5009,
      "step": 215260
    },
    {
      "epoch": 88.12,
      "grad_norm": 1.870470643043518,
      "learning_rate": 2.9895530998867363e-06,
      "loss": 0.4673,
      "step": 215270
    },
    {
      "epoch": 88.12,
      "grad_norm": 2.1345534324645996,
      "learning_rate": 2.9894006095915256e-06,
      "loss": 0.505,
      "step": 215280
    },
    {
      "epoch": 88.13,
      "grad_norm": 1.892098307609558,
      "learning_rate": 2.9892481163386893e-06,
      "loss": 0.5066,
      "step": 215290
    },
    {
      "epoch": 88.13,
      "grad_norm": 1.7334403991699219,
      "learning_rate": 2.9890956201289266e-06,
      "loss": 0.485,
      "step": 215300
    },
    {
      "epoch": 88.13,
      "grad_norm": 1.6896337270736694,
      "learning_rate": 2.988943120962935e-06,
      "loss": 0.491,
      "step": 215310
    },
    {
      "epoch": 88.14,
      "grad_norm": 1.9082858562469482,
      "learning_rate": 2.988790618841414e-06,
      "loss": 0.5196,
      "step": 215320
    },
    {
      "epoch": 88.14,
      "grad_norm": 1.5385671854019165,
      "learning_rate": 2.988638113765062e-06,
      "loss": 0.5016,
      "step": 215330
    },
    {
      "epoch": 88.15,
      "grad_norm": 1.768977403640747,
      "learning_rate": 2.988485605734577e-06,
      "loss": 0.4997,
      "step": 215340
    },
    {
      "epoch": 88.15,
      "grad_norm": 2.446138381958008,
      "learning_rate": 2.988333094750658e-06,
      "loss": 0.4885,
      "step": 215350
    },
    {
      "epoch": 88.15,
      "grad_norm": 1.8405956029891968,
      "learning_rate": 2.9881805808140032e-06,
      "loss": 0.5117,
      "step": 215360
    },
    {
      "epoch": 88.16,
      "grad_norm": 1.4904931783676147,
      "learning_rate": 2.9880280639253126e-06,
      "loss": 0.4808,
      "step": 215370
    },
    {
      "epoch": 88.16,
      "grad_norm": 2.0842387676239014,
      "learning_rate": 2.9878755440852835e-06,
      "loss": 0.5084,
      "step": 215380
    },
    {
      "epoch": 88.17,
      "grad_norm": 1.902759075164795,
      "learning_rate": 2.9877230212946156e-06,
      "loss": 0.5068,
      "step": 215390
    },
    {
      "epoch": 88.17,
      "grad_norm": 3.4886703491210938,
      "learning_rate": 2.9875704955540068e-06,
      "loss": 0.5069,
      "step": 215400
    },
    {
      "epoch": 88.17,
      "grad_norm": 1.9536314010620117,
      "learning_rate": 2.987417966864156e-06,
      "loss": 0.5082,
      "step": 215410
    },
    {
      "epoch": 88.18,
      "grad_norm": 2.0473451614379883,
      "learning_rate": 2.9872654352257616e-06,
      "loss": 0.4738,
      "step": 215420
    },
    {
      "epoch": 88.18,
      "grad_norm": 1.6495819091796875,
      "learning_rate": 2.9871129006395234e-06,
      "loss": 0.4892,
      "step": 215430
    },
    {
      "epoch": 88.19,
      "grad_norm": 1.806641697883606,
      "learning_rate": 2.9869603631061386e-06,
      "loss": 0.4907,
      "step": 215440
    },
    {
      "epoch": 88.19,
      "grad_norm": 2.2032859325408936,
      "learning_rate": 2.986807822626306e-06,
      "loss": 0.4911,
      "step": 215450
    },
    {
      "epoch": 88.19,
      "grad_norm": 2.28265118598938,
      "learning_rate": 2.9866552792007267e-06,
      "loss": 0.4918,
      "step": 215460
    },
    {
      "epoch": 88.2,
      "grad_norm": 2.258457899093628,
      "learning_rate": 2.9865027328300977e-06,
      "loss": 0.4904,
      "step": 215470
    },
    {
      "epoch": 88.2,
      "grad_norm": 1.877278208732605,
      "learning_rate": 2.9863501835151173e-06,
      "loss": 0.486,
      "step": 215480
    },
    {
      "epoch": 88.21,
      "grad_norm": 1.860904335975647,
      "learning_rate": 2.9861976312564853e-06,
      "loss": 0.4857,
      "step": 215490
    },
    {
      "epoch": 88.21,
      "grad_norm": 2.2451486587524414,
      "learning_rate": 2.9860450760549e-06,
      "loss": 0.4973,
      "step": 215500
    },
    {
      "epoch": 88.22,
      "grad_norm": 1.7237635850906372,
      "learning_rate": 2.9858925179110605e-06,
      "loss": 0.4817,
      "step": 215510
    },
    {
      "epoch": 88.22,
      "grad_norm": 1.9900226593017578,
      "learning_rate": 2.985739956825666e-06,
      "loss": 0.4967,
      "step": 215520
    },
    {
      "epoch": 88.22,
      "grad_norm": 2.529095411300659,
      "learning_rate": 2.985587392799414e-06,
      "loss": 0.4682,
      "step": 215530
    },
    {
      "epoch": 88.23,
      "grad_norm": 2.084627628326416,
      "learning_rate": 2.985434825833005e-06,
      "loss": 0.5015,
      "step": 215540
    },
    {
      "epoch": 88.23,
      "grad_norm": 1.829673171043396,
      "learning_rate": 2.985282255927137e-06,
      "loss": 0.5019,
      "step": 215550
    },
    {
      "epoch": 88.24,
      "grad_norm": 2.0146617889404297,
      "learning_rate": 2.985129683082509e-06,
      "loss": 0.5102,
      "step": 215560
    },
    {
      "epoch": 88.24,
      "grad_norm": 2.506960391998291,
      "learning_rate": 2.9849771072998207e-06,
      "loss": 0.4691,
      "step": 215570
    },
    {
      "epoch": 88.24,
      "grad_norm": 1.6972512006759644,
      "learning_rate": 2.9848245285797693e-06,
      "loss": 0.4835,
      "step": 215580
    },
    {
      "epoch": 88.25,
      "grad_norm": 2.068699836730957,
      "learning_rate": 2.9846719469230555e-06,
      "loss": 0.5049,
      "step": 215590
    },
    {
      "epoch": 88.25,
      "grad_norm": 1.974429965019226,
      "learning_rate": 2.9845193623303773e-06,
      "loss": 0.4957,
      "step": 215600
    },
    {
      "epoch": 88.26,
      "grad_norm": 2.0049517154693604,
      "learning_rate": 2.984366774802434e-06,
      "loss": 0.5074,
      "step": 215610
    },
    {
      "epoch": 88.26,
      "grad_norm": 1.6553078889846802,
      "learning_rate": 2.9842141843399244e-06,
      "loss": 0.5007,
      "step": 215620
    },
    {
      "epoch": 88.26,
      "grad_norm": 2.1218252182006836,
      "learning_rate": 2.9840615909435474e-06,
      "loss": 0.5059,
      "step": 215630
    },
    {
      "epoch": 88.27,
      "grad_norm": 1.6284066438674927,
      "learning_rate": 2.9839089946140016e-06,
      "loss": 0.4977,
      "step": 215640
    },
    {
      "epoch": 88.27,
      "grad_norm": 2.3243680000305176,
      "learning_rate": 2.9837563953519875e-06,
      "loss": 0.5032,
      "step": 215650
    },
    {
      "epoch": 88.28,
      "grad_norm": 2.322075605392456,
      "learning_rate": 2.983603793158203e-06,
      "loss": 0.4795,
      "step": 215660
    },
    {
      "epoch": 88.28,
      "grad_norm": 1.9128227233886719,
      "learning_rate": 2.983451188033347e-06,
      "loss": 0.4911,
      "step": 215670
    },
    {
      "epoch": 88.28,
      "grad_norm": 1.4088952541351318,
      "learning_rate": 2.9832985799781193e-06,
      "loss": 0.4748,
      "step": 215680
    },
    {
      "epoch": 88.29,
      "grad_norm": 1.965980887413025,
      "learning_rate": 2.983145968993219e-06,
      "loss": 0.4811,
      "step": 215690
    },
    {
      "epoch": 88.29,
      "grad_norm": 1.8884345293045044,
      "learning_rate": 2.9829933550793443e-06,
      "loss": 0.5082,
      "step": 215700
    },
    {
      "epoch": 88.3,
      "grad_norm": 2.4152612686157227,
      "learning_rate": 2.982840738237195e-06,
      "loss": 0.5031,
      "step": 215710
    },
    {
      "epoch": 88.3,
      "grad_norm": 1.9530447721481323,
      "learning_rate": 2.9826881184674696e-06,
      "loss": 0.4994,
      "step": 215720
    },
    {
      "epoch": 88.31,
      "grad_norm": 1.8627666234970093,
      "learning_rate": 2.9825354957708677e-06,
      "loss": 0.4961,
      "step": 215730
    },
    {
      "epoch": 88.31,
      "grad_norm": 2.0094501972198486,
      "learning_rate": 2.9823828701480884e-06,
      "loss": 0.4862,
      "step": 215740
    },
    {
      "epoch": 88.31,
      "grad_norm": 1.7559813261032104,
      "learning_rate": 2.982230241599831e-06,
      "loss": 0.4986,
      "step": 215750
    },
    {
      "epoch": 88.32,
      "grad_norm": 1.7916889190673828,
      "learning_rate": 2.982077610126795e-06,
      "loss": 0.4961,
      "step": 215760
    },
    {
      "epoch": 88.32,
      "grad_norm": 1.862672209739685,
      "learning_rate": 2.981924975729678e-06,
      "loss": 0.4993,
      "step": 215770
    },
    {
      "epoch": 88.33,
      "grad_norm": 1.9443267583847046,
      "learning_rate": 2.981772338409181e-06,
      "loss": 0.4749,
      "step": 215780
    },
    {
      "epoch": 88.33,
      "grad_norm": 1.477637529373169,
      "learning_rate": 2.9816196981660024e-06,
      "loss": 0.4845,
      "step": 215790
    },
    {
      "epoch": 88.33,
      "grad_norm": 1.6324598789215088,
      "learning_rate": 2.9814670550008413e-06,
      "loss": 0.4953,
      "step": 215800
    },
    {
      "epoch": 88.34,
      "grad_norm": 1.6866109371185303,
      "learning_rate": 2.981314408914397e-06,
      "loss": 0.4822,
      "step": 215810
    },
    {
      "epoch": 88.34,
      "grad_norm": 2.2344870567321777,
      "learning_rate": 2.981161759907369e-06,
      "loss": 0.4834,
      "step": 215820
    },
    {
      "epoch": 88.35,
      "grad_norm": 2.3226451873779297,
      "learning_rate": 2.981009107980457e-06,
      "loss": 0.4828,
      "step": 215830
    },
    {
      "epoch": 88.35,
      "grad_norm": 1.7386424541473389,
      "learning_rate": 2.980856453134359e-06,
      "loss": 0.4792,
      "step": 215840
    },
    {
      "epoch": 88.35,
      "grad_norm": 2.3416574001312256,
      "learning_rate": 2.980703795369775e-06,
      "loss": 0.4973,
      "step": 215850
    },
    {
      "epoch": 88.36,
      "grad_norm": 1.660832405090332,
      "learning_rate": 2.980551134687405e-06,
      "loss": 0.4804,
      "step": 215860
    },
    {
      "epoch": 88.36,
      "grad_norm": 1.8751707077026367,
      "learning_rate": 2.980398471087947e-06,
      "loss": 0.4924,
      "step": 215870
    },
    {
      "epoch": 88.37,
      "grad_norm": 1.565947413444519,
      "learning_rate": 2.9802458045721013e-06,
      "loss": 0.5106,
      "step": 215880
    },
    {
      "epoch": 88.37,
      "grad_norm": 1.8010538816452026,
      "learning_rate": 2.9800931351405662e-06,
      "loss": 0.4953,
      "step": 215890
    },
    {
      "epoch": 88.37,
      "grad_norm": 1.8629202842712402,
      "learning_rate": 2.9799404627940426e-06,
      "loss": 0.4751,
      "step": 215900
    },
    {
      "epoch": 88.38,
      "grad_norm": 1.9131991863250732,
      "learning_rate": 2.9797877875332294e-06,
      "loss": 0.4944,
      "step": 215910
    },
    {
      "epoch": 88.38,
      "grad_norm": 1.887190580368042,
      "learning_rate": 2.979635109358824e-06,
      "loss": 0.4832,
      "step": 215920
    },
    {
      "epoch": 88.39,
      "grad_norm": 1.787437915802002,
      "learning_rate": 2.9794824282715285e-06,
      "loss": 0.4771,
      "step": 215930
    },
    {
      "epoch": 88.39,
      "grad_norm": 1.9529320001602173,
      "learning_rate": 2.9793297442720406e-06,
      "loss": 0.5274,
      "step": 215940
    },
    {
      "epoch": 88.4,
      "grad_norm": 2.2401063442230225,
      "learning_rate": 2.979177057361061e-06,
      "loss": 0.4931,
      "step": 215950
    },
    {
      "epoch": 88.4,
      "grad_norm": 2.3396239280700684,
      "learning_rate": 2.9790243675392883e-06,
      "loss": 0.4909,
      "step": 215960
    },
    {
      "epoch": 88.4,
      "grad_norm": 1.6443456411361694,
      "learning_rate": 2.978871674807422e-06,
      "loss": 0.4813,
      "step": 215970
    },
    {
      "epoch": 88.41,
      "grad_norm": 2.227402687072754,
      "learning_rate": 2.978718979166162e-06,
      "loss": 0.5198,
      "step": 215980
    },
    {
      "epoch": 88.41,
      "grad_norm": 1.8007911443710327,
      "learning_rate": 2.978566280616207e-06,
      "loss": 0.498,
      "step": 215990
    },
    {
      "epoch": 88.42,
      "grad_norm": 1.7135589122772217,
      "learning_rate": 2.9784135791582566e-06,
      "loss": 0.4624,
      "step": 216000
    },
    {
      "epoch": 88.42,
      "grad_norm": 2.102099657058716,
      "learning_rate": 2.9782608747930114e-06,
      "loss": 0.47,
      "step": 216010
    },
    {
      "epoch": 88.42,
      "grad_norm": 2.1621696949005127,
      "learning_rate": 2.978108167521169e-06,
      "loss": 0.4972,
      "step": 216020
    },
    {
      "epoch": 88.43,
      "grad_norm": 1.715563416481018,
      "learning_rate": 2.9779554573434314e-06,
      "loss": 0.4942,
      "step": 216030
    },
    {
      "epoch": 88.43,
      "grad_norm": 1.7466679811477661,
      "learning_rate": 2.977802744260496e-06,
      "loss": 0.497,
      "step": 216040
    },
    {
      "epoch": 88.44,
      "grad_norm": 2.026975154876709,
      "learning_rate": 2.977650028273064e-06,
      "loss": 0.4924,
      "step": 216050
    },
    {
      "epoch": 88.44,
      "grad_norm": 1.6154792308807373,
      "learning_rate": 2.9774973093818337e-06,
      "loss": 0.4987,
      "step": 216060
    },
    {
      "epoch": 88.44,
      "grad_norm": 1.8288838863372803,
      "learning_rate": 2.9773445875875054e-06,
      "loss": 0.4927,
      "step": 216070
    },
    {
      "epoch": 88.45,
      "grad_norm": 2.2001307010650635,
      "learning_rate": 2.9771918628907775e-06,
      "loss": 0.4974,
      "step": 216080
    },
    {
      "epoch": 88.45,
      "grad_norm": 1.6629410982131958,
      "learning_rate": 2.977039135292351e-06,
      "loss": 0.4779,
      "step": 216090
    },
    {
      "epoch": 88.46,
      "grad_norm": 1.9058085680007935,
      "learning_rate": 2.9768864047929254e-06,
      "loss": 0.4875,
      "step": 216100
    },
    {
      "epoch": 88.46,
      "grad_norm": 1.5155857801437378,
      "learning_rate": 2.976733671393199e-06,
      "loss": 0.4705,
      "step": 216110
    },
    {
      "epoch": 88.47,
      "grad_norm": 2.0418877601623535,
      "learning_rate": 2.9765809350938737e-06,
      "loss": 0.5085,
      "step": 216120
    },
    {
      "epoch": 88.47,
      "grad_norm": 2.476207971572876,
      "learning_rate": 2.9764281958956474e-06,
      "loss": 0.4968,
      "step": 216130
    },
    {
      "epoch": 88.47,
      "grad_norm": 2.434086561203003,
      "learning_rate": 2.9762754537992205e-06,
      "loss": 0.4963,
      "step": 216140
    },
    {
      "epoch": 88.48,
      "grad_norm": 1.8249847888946533,
      "learning_rate": 2.976122708805292e-06,
      "loss": 0.5056,
      "step": 216150
    },
    {
      "epoch": 88.48,
      "grad_norm": 2.1404852867126465,
      "learning_rate": 2.975969960914562e-06,
      "loss": 0.4936,
      "step": 216160
    },
    {
      "epoch": 88.49,
      "grad_norm": 1.4430557489395142,
      "learning_rate": 2.9758172101277308e-06,
      "loss": 0.5094,
      "step": 216170
    },
    {
      "epoch": 88.49,
      "grad_norm": 1.7189396619796753,
      "learning_rate": 2.9756644564454974e-06,
      "loss": 0.4917,
      "step": 216180
    },
    {
      "epoch": 88.49,
      "grad_norm": 2.611104726791382,
      "learning_rate": 2.9755116998685615e-06,
      "loss": 0.4812,
      "step": 216190
    },
    {
      "epoch": 88.5,
      "grad_norm": 1.6772595643997192,
      "learning_rate": 2.975358940397623e-06,
      "loss": 0.5132,
      "step": 216200
    },
    {
      "epoch": 88.5,
      "grad_norm": 1.9536564350128174,
      "learning_rate": 2.975206178033382e-06,
      "loss": 0.4846,
      "step": 216210
    },
    {
      "epoch": 88.51,
      "grad_norm": 1.7324250936508179,
      "learning_rate": 2.9750534127765383e-06,
      "loss": 0.4968,
      "step": 216220
    },
    {
      "epoch": 88.51,
      "grad_norm": 2.012817144393921,
      "learning_rate": 2.9749006446277906e-06,
      "loss": 0.4799,
      "step": 216230
    },
    {
      "epoch": 88.51,
      "grad_norm": 2.01875901222229,
      "learning_rate": 2.9747478735878404e-06,
      "loss": 0.4665,
      "step": 216240
    },
    {
      "epoch": 88.52,
      "grad_norm": 1.629575490951538,
      "learning_rate": 2.974595099657386e-06,
      "loss": 0.4988,
      "step": 216250
    },
    {
      "epoch": 88.52,
      "grad_norm": 2.108245372772217,
      "learning_rate": 2.9744423228371284e-06,
      "loss": 0.5061,
      "step": 216260
    },
    {
      "epoch": 88.53,
      "grad_norm": 1.6036332845687866,
      "learning_rate": 2.9742895431277672e-06,
      "loss": 0.4883,
      "step": 216270
    },
    {
      "epoch": 88.53,
      "grad_norm": 1.7246507406234741,
      "learning_rate": 2.974136760530002e-06,
      "loss": 0.5004,
      "step": 216280
    },
    {
      "epoch": 88.53,
      "grad_norm": 1.7049864530563354,
      "learning_rate": 2.9739839750445317e-06,
      "loss": 0.4754,
      "step": 216290
    },
    {
      "epoch": 88.54,
      "grad_norm": 1.7348737716674805,
      "learning_rate": 2.9738311866720576e-06,
      "loss": 0.5063,
      "step": 216300
    },
    {
      "epoch": 88.54,
      "grad_norm": 1.9451888799667358,
      "learning_rate": 2.9736783954132793e-06,
      "loss": 0.5223,
      "step": 216310
    },
    {
      "epoch": 88.55,
      "grad_norm": 1.9833248853683472,
      "learning_rate": 2.9735256012688965e-06,
      "loss": 0.4996,
      "step": 216320
    },
    {
      "epoch": 88.55,
      "grad_norm": 2.1571671962738037,
      "learning_rate": 2.9733728042396097e-06,
      "loss": 0.5003,
      "step": 216330
    },
    {
      "epoch": 88.56,
      "grad_norm": 2.1346919536590576,
      "learning_rate": 2.973220004326118e-06,
      "loss": 0.5164,
      "step": 216340
    },
    {
      "epoch": 88.56,
      "grad_norm": 2.405374765396118,
      "learning_rate": 2.973067201529122e-06,
      "loss": 0.5047,
      "step": 216350
    },
    {
      "epoch": 88.56,
      "grad_norm": 1.7416698932647705,
      "learning_rate": 2.972914395849321e-06,
      "loss": 0.5027,
      "step": 216360
    },
    {
      "epoch": 88.57,
      "grad_norm": 1.9482413530349731,
      "learning_rate": 2.9727615872874146e-06,
      "loss": 0.4863,
      "step": 216370
    },
    {
      "epoch": 88.57,
      "grad_norm": 1.920525312423706,
      "learning_rate": 2.9726087758441046e-06,
      "loss": 0.4965,
      "step": 216380
    },
    {
      "epoch": 88.58,
      "grad_norm": 1.9301306009292603,
      "learning_rate": 2.972455961520089e-06,
      "loss": 0.5175,
      "step": 216390
    },
    {
      "epoch": 88.58,
      "grad_norm": 1.3366129398345947,
      "learning_rate": 2.9723031443160694e-06,
      "loss": 0.4912,
      "step": 216400
    },
    {
      "epoch": 88.58,
      "grad_norm": 2.0800728797912598,
      "learning_rate": 2.972150324232746e-06,
      "loss": 0.4965,
      "step": 216410
    },
    {
      "epoch": 88.59,
      "grad_norm": 1.9225655794143677,
      "learning_rate": 2.971997501270817e-06,
      "loss": 0.5052,
      "step": 216420
    },
    {
      "epoch": 88.59,
      "grad_norm": 2.217366933822632,
      "learning_rate": 2.971844675430984e-06,
      "loss": 0.4931,
      "step": 216430
    },
    {
      "epoch": 88.6,
      "grad_norm": 1.8936511278152466,
      "learning_rate": 2.971691846713946e-06,
      "loss": 0.5073,
      "step": 216440
    },
    {
      "epoch": 88.6,
      "grad_norm": 2.559394359588623,
      "learning_rate": 2.971539015120404e-06,
      "loss": 0.4785,
      "step": 216450
    },
    {
      "epoch": 88.6,
      "grad_norm": 1.6224249601364136,
      "learning_rate": 2.9713861806510574e-06,
      "loss": 0.5047,
      "step": 216460
    },
    {
      "epoch": 88.61,
      "grad_norm": 2.944056749343872,
      "learning_rate": 2.9712333433066073e-06,
      "loss": 0.4988,
      "step": 216470
    },
    {
      "epoch": 88.61,
      "grad_norm": 1.7704441547393799,
      "learning_rate": 2.9710805030877525e-06,
      "loss": 0.4982,
      "step": 216480
    },
    {
      "epoch": 88.62,
      "grad_norm": 2.0444040298461914,
      "learning_rate": 2.970927659995194e-06,
      "loss": 0.4902,
      "step": 216490
    },
    {
      "epoch": 88.62,
      "grad_norm": 1.966209888458252,
      "learning_rate": 2.9707748140296317e-06,
      "loss": 0.4729,
      "step": 216500
    },
    {
      "epoch": 88.62,
      "grad_norm": 2.4448299407958984,
      "learning_rate": 2.970621965191766e-06,
      "loss": 0.5174,
      "step": 216510
    },
    {
      "epoch": 88.63,
      "grad_norm": 1.9024678468704224,
      "learning_rate": 2.9704691134822974e-06,
      "loss": 0.4973,
      "step": 216520
    },
    {
      "epoch": 88.63,
      "grad_norm": 1.9901628494262695,
      "learning_rate": 2.9703162589019246e-06,
      "loss": 0.4796,
      "step": 216530
    },
    {
      "epoch": 88.64,
      "grad_norm": 2.146178722381592,
      "learning_rate": 2.970163401451349e-06,
      "loss": 0.4791,
      "step": 216540
    },
    {
      "epoch": 88.64,
      "grad_norm": 2.124431848526001,
      "learning_rate": 2.9700105411312714e-06,
      "loss": 0.4853,
      "step": 216550
    },
    {
      "epoch": 88.65,
      "grad_norm": 1.7056472301483154,
      "learning_rate": 2.9698576779423917e-06,
      "loss": 0.5079,
      "step": 216560
    },
    {
      "epoch": 88.65,
      "grad_norm": 1.5229116678237915,
      "learning_rate": 2.9697048118854085e-06,
      "loss": 0.4854,
      "step": 216570
    },
    {
      "epoch": 88.65,
      "grad_norm": 1.7992783784866333,
      "learning_rate": 2.9695519429610227e-06,
      "loss": 0.4843,
      "step": 216580
    },
    {
      "epoch": 88.66,
      "grad_norm": 1.497135877609253,
      "learning_rate": 2.969399071169936e-06,
      "loss": 0.4911,
      "step": 216590
    },
    {
      "epoch": 88.66,
      "grad_norm": 1.9302442073822021,
      "learning_rate": 2.969246196512848e-06,
      "loss": 0.5126,
      "step": 216600
    },
    {
      "epoch": 88.67,
      "grad_norm": 2.1153759956359863,
      "learning_rate": 2.9690933189904587e-06,
      "loss": 0.4904,
      "step": 216610
    },
    {
      "epoch": 88.67,
      "grad_norm": 1.9968035221099854,
      "learning_rate": 2.968940438603468e-06,
      "loss": 0.4718,
      "step": 216620
    },
    {
      "epoch": 88.67,
      "grad_norm": 1.6792685985565186,
      "learning_rate": 2.968787555352578e-06,
      "loss": 0.4997,
      "step": 216630
    },
    {
      "epoch": 88.68,
      "grad_norm": 1.6993393898010254,
      "learning_rate": 2.968634669238487e-06,
      "loss": 0.5021,
      "step": 216640
    },
    {
      "epoch": 88.68,
      "grad_norm": 2.9620909690856934,
      "learning_rate": 2.9684817802618954e-06,
      "loss": 0.5205,
      "step": 216650
    },
    {
      "epoch": 88.69,
      "grad_norm": 2.2332873344421387,
      "learning_rate": 2.9683288884235044e-06,
      "loss": 0.4994,
      "step": 216660
    },
    {
      "epoch": 88.69,
      "grad_norm": 1.8951764106750488,
      "learning_rate": 2.968175993724015e-06,
      "loss": 0.4868,
      "step": 216670
    },
    {
      "epoch": 88.69,
      "grad_norm": 1.82198965549469,
      "learning_rate": 2.968023096164126e-06,
      "loss": 0.4737,
      "step": 216680
    },
    {
      "epoch": 88.7,
      "grad_norm": 1.5554312467575073,
      "learning_rate": 2.967870195744539e-06,
      "loss": 0.5202,
      "step": 216690
    },
    {
      "epoch": 88.7,
      "grad_norm": 1.8380180597305298,
      "learning_rate": 2.9677172924659545e-06,
      "loss": 0.4651,
      "step": 216700
    },
    {
      "epoch": 88.71,
      "grad_norm": 2.205965280532837,
      "learning_rate": 2.9675643863290723e-06,
      "loss": 0.4985,
      "step": 216710
    },
    {
      "epoch": 88.71,
      "grad_norm": 1.6791954040527344,
      "learning_rate": 2.967411477334593e-06,
      "loss": 0.4976,
      "step": 216720
    },
    {
      "epoch": 88.71,
      "grad_norm": 1.9059276580810547,
      "learning_rate": 2.9672585654832165e-06,
      "loss": 0.4862,
      "step": 216730
    },
    {
      "epoch": 88.72,
      "grad_norm": 1.5216883420944214,
      "learning_rate": 2.9671056507756445e-06,
      "loss": 0.4924,
      "step": 216740
    },
    {
      "epoch": 88.72,
      "grad_norm": 1.7094354629516602,
      "learning_rate": 2.9669527332125766e-06,
      "loss": 0.4934,
      "step": 216750
    },
    {
      "epoch": 88.73,
      "grad_norm": 1.7338252067565918,
      "learning_rate": 2.966799812794713e-06,
      "loss": 0.4835,
      "step": 216760
    },
    {
      "epoch": 88.73,
      "grad_norm": 1.774472951889038,
      "learning_rate": 2.966646889522755e-06,
      "loss": 0.4813,
      "step": 216770
    },
    {
      "epoch": 88.74,
      "grad_norm": 1.8369098901748657,
      "learning_rate": 2.966493963397403e-06,
      "loss": 0.4626,
      "step": 216780
    },
    {
      "epoch": 88.74,
      "grad_norm": 1.9893004894256592,
      "learning_rate": 2.9663410344193574e-06,
      "loss": 0.5005,
      "step": 216790
    },
    {
      "epoch": 88.74,
      "grad_norm": 1.7322075366973877,
      "learning_rate": 2.9661881025893184e-06,
      "loss": 0.4895,
      "step": 216800
    },
    {
      "epoch": 88.75,
      "grad_norm": 2.2441108226776123,
      "learning_rate": 2.966035167907987e-06,
      "loss": 0.4772,
      "step": 216810
    },
    {
      "epoch": 88.75,
      "grad_norm": 2.498114585876465,
      "learning_rate": 2.965882230376064e-06,
      "loss": 0.4991,
      "step": 216820
    },
    {
      "epoch": 88.76,
      "grad_norm": 1.8494534492492676,
      "learning_rate": 2.9657292899942486e-06,
      "loss": 0.5186,
      "step": 216830
    },
    {
      "epoch": 88.76,
      "grad_norm": 2.0354511737823486,
      "learning_rate": 2.965576346763243e-06,
      "loss": 0.5028,
      "step": 216840
    },
    {
      "epoch": 88.76,
      "grad_norm": 2.2082369327545166,
      "learning_rate": 2.9654234006837476e-06,
      "loss": 0.488,
      "step": 216850
    },
    {
      "epoch": 88.77,
      "grad_norm": 1.808782696723938,
      "learning_rate": 2.9652704517564617e-06,
      "loss": 0.4935,
      "step": 216860
    },
    {
      "epoch": 88.77,
      "grad_norm": 1.5556813478469849,
      "learning_rate": 2.965117499982088e-06,
      "loss": 0.488,
      "step": 216870
    },
    {
      "epoch": 88.78,
      "grad_norm": 1.7049105167388916,
      "learning_rate": 2.9649645453613252e-06,
      "loss": 0.5091,
      "step": 216880
    },
    {
      "epoch": 88.78,
      "grad_norm": 2.5060832500457764,
      "learning_rate": 2.9648115878948752e-06,
      "loss": 0.4947,
      "step": 216890
    },
    {
      "epoch": 88.78,
      "grad_norm": 1.8991953134536743,
      "learning_rate": 2.9646586275834377e-06,
      "loss": 0.4843,
      "step": 216900
    },
    {
      "epoch": 88.79,
      "grad_norm": 1.9968302249908447,
      "learning_rate": 2.9645056644277143e-06,
      "loss": 0.4939,
      "step": 216910
    },
    {
      "epoch": 88.79,
      "grad_norm": 2.286914825439453,
      "learning_rate": 2.964352698428406e-06,
      "loss": 0.4954,
      "step": 216920
    },
    {
      "epoch": 88.8,
      "grad_norm": 2.1870832443237305,
      "learning_rate": 2.9641997295862126e-06,
      "loss": 0.491,
      "step": 216930
    },
    {
      "epoch": 88.8,
      "grad_norm": 1.8128546476364136,
      "learning_rate": 2.9640467579018343e-06,
      "loss": 0.4802,
      "step": 216940
    },
    {
      "epoch": 88.8,
      "grad_norm": 1.8860158920288086,
      "learning_rate": 2.9638937833759726e-06,
      "loss": 0.4786,
      "step": 216950
    },
    {
      "epoch": 88.81,
      "grad_norm": 1.7121071815490723,
      "learning_rate": 2.963740806009328e-06,
      "loss": 0.4852,
      "step": 216960
    },
    {
      "epoch": 88.81,
      "grad_norm": 1.8903990983963013,
      "learning_rate": 2.9635878258026028e-06,
      "loss": 0.485,
      "step": 216970
    },
    {
      "epoch": 88.82,
      "grad_norm": 1.8023303747177124,
      "learning_rate": 2.963434842756496e-06,
      "loss": 0.4875,
      "step": 216980
    },
    {
      "epoch": 88.82,
      "grad_norm": 1.7390283346176147,
      "learning_rate": 2.9632818568717095e-06,
      "loss": 0.475,
      "step": 216990
    },
    {
      "epoch": 88.83,
      "grad_norm": 1.5540300607681274,
      "learning_rate": 2.9631288681489424e-06,
      "loss": 0.4909,
      "step": 217000
    },
    {
      "epoch": 88.83,
      "grad_norm": 1.9959276914596558,
      "learning_rate": 2.9629758765888967e-06,
      "loss": 0.4879,
      "step": 217010
    },
    {
      "epoch": 88.83,
      "grad_norm": 1.8040771484375,
      "learning_rate": 2.962822882192274e-06,
      "loss": 0.4895,
      "step": 217020
    },
    {
      "epoch": 88.84,
      "grad_norm": 2.193063497543335,
      "learning_rate": 2.9626698849597735e-06,
      "loss": 0.4963,
      "step": 217030
    },
    {
      "epoch": 88.84,
      "grad_norm": 2.0316402912139893,
      "learning_rate": 2.9625168848920975e-06,
      "loss": 0.5004,
      "step": 217040
    },
    {
      "epoch": 88.85,
      "grad_norm": 1.7850279808044434,
      "learning_rate": 2.962363881989945e-06,
      "loss": 0.4968,
      "step": 217050
    },
    {
      "epoch": 88.85,
      "grad_norm": 1.4760198593139648,
      "learning_rate": 2.962210876254019e-06,
      "loss": 0.5,
      "step": 217060
    },
    {
      "epoch": 88.85,
      "grad_norm": 1.731453537940979,
      "learning_rate": 2.96205786768502e-06,
      "loss": 0.4806,
      "step": 217070
    },
    {
      "epoch": 88.86,
      "grad_norm": 1.8849236965179443,
      "learning_rate": 2.961904856283648e-06,
      "loss": 0.496,
      "step": 217080
    },
    {
      "epoch": 88.86,
      "grad_norm": 1.7735546827316284,
      "learning_rate": 2.9617518420506045e-06,
      "loss": 0.4864,
      "step": 217090
    },
    {
      "epoch": 88.87,
      "grad_norm": 1.9637794494628906,
      "learning_rate": 2.96159882498659e-06,
      "loss": 0.4873,
      "step": 217100
    },
    {
      "epoch": 88.87,
      "grad_norm": 1.9762158393859863,
      "learning_rate": 2.961445805092305e-06,
      "loss": 0.509,
      "step": 217110
    },
    {
      "epoch": 88.87,
      "grad_norm": 2.342451572418213,
      "learning_rate": 2.9612927823684523e-06,
      "loss": 0.4926,
      "step": 217120
    },
    {
      "epoch": 88.88,
      "grad_norm": 2.313591480255127,
      "learning_rate": 2.9611397568157314e-06,
      "loss": 0.5091,
      "step": 217130
    },
    {
      "epoch": 88.88,
      "grad_norm": 2.397392749786377,
      "learning_rate": 2.9609867284348437e-06,
      "loss": 0.4875,
      "step": 217140
    },
    {
      "epoch": 88.89,
      "grad_norm": 2.917910099029541,
      "learning_rate": 2.9608336972264897e-06,
      "loss": 0.4788,
      "step": 217150
    },
    {
      "epoch": 88.89,
      "grad_norm": 1.7926403284072876,
      "learning_rate": 2.960680663191371e-06,
      "loss": 0.4906,
      "step": 217160
    },
    {
      "epoch": 88.89,
      "grad_norm": 1.7560386657714844,
      "learning_rate": 2.9605276263301886e-06,
      "loss": 0.4871,
      "step": 217170
    },
    {
      "epoch": 88.9,
      "grad_norm": 2.0900113582611084,
      "learning_rate": 2.960374586643644e-06,
      "loss": 0.4909,
      "step": 217180
    },
    {
      "epoch": 88.9,
      "grad_norm": 2.0639278888702393,
      "learning_rate": 2.9602215441324373e-06,
      "loss": 0.503,
      "step": 217190
    },
    {
      "epoch": 88.91,
      "grad_norm": 1.717184066772461,
      "learning_rate": 2.96006849879727e-06,
      "loss": 0.4913,
      "step": 217200
    },
    {
      "epoch": 88.91,
      "grad_norm": 1.929621934890747,
      "learning_rate": 2.959915450638843e-06,
      "loss": 0.4995,
      "step": 217210
    },
    {
      "epoch": 88.92,
      "grad_norm": 2.0230767726898193,
      "learning_rate": 2.959762399657857e-06,
      "loss": 0.4865,
      "step": 217220
    },
    {
      "epoch": 88.92,
      "grad_norm": 1.9251246452331543,
      "learning_rate": 2.959609345855014e-06,
      "loss": 0.482,
      "step": 217230
    },
    {
      "epoch": 88.92,
      "grad_norm": 1.7693003416061401,
      "learning_rate": 2.9594562892310146e-06,
      "loss": 0.4944,
      "step": 217240
    },
    {
      "epoch": 88.93,
      "grad_norm": 1.8353655338287354,
      "learning_rate": 2.9593032297865602e-06,
      "loss": 0.49,
      "step": 217250
    },
    {
      "epoch": 88.93,
      "grad_norm": 2.5908682346343994,
      "learning_rate": 2.959150167522352e-06,
      "loss": 0.5061,
      "step": 217260
    },
    {
      "epoch": 88.94,
      "grad_norm": 1.5580487251281738,
      "learning_rate": 2.9589971024390903e-06,
      "loss": 0.4945,
      "step": 217270
    },
    {
      "epoch": 88.94,
      "grad_norm": 2.2513630390167236,
      "learning_rate": 2.9588440345374778e-06,
      "loss": 0.4666,
      "step": 217280
    },
    {
      "epoch": 88.94,
      "grad_norm": 2.1214051246643066,
      "learning_rate": 2.9586909638182144e-06,
      "loss": 0.491,
      "step": 217290
    },
    {
      "epoch": 88.95,
      "grad_norm": 1.7793588638305664,
      "learning_rate": 2.958537890282001e-06,
      "loss": 0.4933,
      "step": 217300
    },
    {
      "epoch": 88.95,
      "grad_norm": 2.2276055812835693,
      "learning_rate": 2.95838481392954e-06,
      "loss": 0.5122,
      "step": 217310
    },
    {
      "epoch": 88.96,
      "grad_norm": 1.9963128566741943,
      "learning_rate": 2.9582317347615326e-06,
      "loss": 0.5035,
      "step": 217320
    },
    {
      "epoch": 88.96,
      "grad_norm": 1.7714728116989136,
      "learning_rate": 2.958078652778679e-06,
      "loss": 0.5025,
      "step": 217330
    },
    {
      "epoch": 88.96,
      "grad_norm": 1.9003239870071411,
      "learning_rate": 2.9579255679816805e-06,
      "loss": 0.4872,
      "step": 217340
    },
    {
      "epoch": 88.97,
      "grad_norm": 1.7812209129333496,
      "learning_rate": 2.95777248037124e-06,
      "loss": 0.4839,
      "step": 217350
    },
    {
      "epoch": 88.97,
      "grad_norm": 1.9600502252578735,
      "learning_rate": 2.9576193899480567e-06,
      "loss": 0.5058,
      "step": 217360
    },
    {
      "epoch": 88.98,
      "grad_norm": 1.5831316709518433,
      "learning_rate": 2.9574662967128335e-06,
      "loss": 0.4929,
      "step": 217370
    },
    {
      "epoch": 88.98,
      "grad_norm": 1.4897770881652832,
      "learning_rate": 2.9573132006662703e-06,
      "loss": 0.5138,
      "step": 217380
    },
    {
      "epoch": 88.98,
      "grad_norm": 1.9470555782318115,
      "learning_rate": 2.957160101809069e-06,
      "loss": 0.4951,
      "step": 217390
    },
    {
      "epoch": 88.99,
      "grad_norm": 1.603779911994934,
      "learning_rate": 2.957007000141932e-06,
      "loss": 0.5084,
      "step": 217400
    },
    {
      "epoch": 88.99,
      "grad_norm": 2.01690411567688,
      "learning_rate": 2.956853895665559e-06,
      "loss": 0.4916,
      "step": 217410
    },
    {
      "epoch": 89.0,
      "grad_norm": 1.9918694496154785,
      "learning_rate": 2.956700788380652e-06,
      "loss": 0.5036,
      "step": 217420
    },
    {
      "epoch": 89.0,
      "eval_loss": 0.4926697313785553,
      "eval_runtime": 76.5814,
      "eval_samples_per_second": 45.037,
      "eval_steps_per_second": 5.641,
      "step": 217427
    },
    {
      "epoch": 89.0,
      "grad_norm": 1.8269280195236206,
      "learning_rate": 2.9565476782879124e-06,
      "loss": 0.4987,
      "step": 217430
    },
    {
      "epoch": 89.01,
      "grad_norm": 1.699037790298462,
      "learning_rate": 2.9563945653880417e-06,
      "loss": 0.4928,
      "step": 217440
    },
    {
      "epoch": 89.01,
      "grad_norm": 1.861743688583374,
      "learning_rate": 2.956241449681741e-06,
      "loss": 0.4739,
      "step": 217450
    },
    {
      "epoch": 89.01,
      "grad_norm": 1.5747919082641602,
      "learning_rate": 2.956088331169712e-06,
      "loss": 0.501,
      "step": 217460
    },
    {
      "epoch": 89.02,
      "grad_norm": 2.533867359161377,
      "learning_rate": 2.955935209852655e-06,
      "loss": 0.5061,
      "step": 217470
    },
    {
      "epoch": 89.02,
      "grad_norm": 2.0949270725250244,
      "learning_rate": 2.955782085731274e-06,
      "loss": 0.4826,
      "step": 217480
    },
    {
      "epoch": 89.03,
      "grad_norm": 1.741434097290039,
      "learning_rate": 2.9556289588062673e-06,
      "loss": 0.4834,
      "step": 217490
    },
    {
      "epoch": 89.03,
      "grad_norm": 1.6018469333648682,
      "learning_rate": 2.9554758290783385e-06,
      "loss": 0.4991,
      "step": 217500
    },
    {
      "epoch": 89.03,
      "grad_norm": 1.971185564994812,
      "learning_rate": 2.955322696548189e-06,
      "loss": 0.5189,
      "step": 217510
    },
    {
      "epoch": 89.04,
      "grad_norm": 2.025585651397705,
      "learning_rate": 2.9551695612165187e-06,
      "loss": 0.4752,
      "step": 217520
    },
    {
      "epoch": 89.04,
      "grad_norm": 1.7729369401931763,
      "learning_rate": 2.9550164230840304e-06,
      "loss": 0.4786,
      "step": 217530
    },
    {
      "epoch": 89.05,
      "grad_norm": 1.7326236963272095,
      "learning_rate": 2.9548632821514256e-06,
      "loss": 0.5068,
      "step": 217540
    },
    {
      "epoch": 89.05,
      "grad_norm": 1.7937570810317993,
      "learning_rate": 2.954710138419405e-06,
      "loss": 0.4904,
      "step": 217550
    },
    {
      "epoch": 89.05,
      "grad_norm": 1.960135579109192,
      "learning_rate": 2.954556991888671e-06,
      "loss": 0.4787,
      "step": 217560
    },
    {
      "epoch": 89.06,
      "grad_norm": 1.7629435062408447,
      "learning_rate": 2.954403842559925e-06,
      "loss": 0.4818,
      "step": 217570
    },
    {
      "epoch": 89.06,
      "grad_norm": 1.8737783432006836,
      "learning_rate": 2.954250690433868e-06,
      "loss": 0.4911,
      "step": 217580
    },
    {
      "epoch": 89.07,
      "grad_norm": 1.8133244514465332,
      "learning_rate": 2.9540975355112016e-06,
      "loss": 0.4902,
      "step": 217590
    },
    {
      "epoch": 89.07,
      "grad_norm": 1.8367440700531006,
      "learning_rate": 2.953944377792628e-06,
      "loss": 0.492,
      "step": 217600
    },
    {
      "epoch": 89.07,
      "grad_norm": 1.8808863162994385,
      "learning_rate": 2.9537912172788485e-06,
      "loss": 0.5095,
      "step": 217610
    },
    {
      "epoch": 89.08,
      "grad_norm": 1.8962831497192383,
      "learning_rate": 2.9536380539705637e-06,
      "loss": 0.4822,
      "step": 217620
    },
    {
      "epoch": 89.08,
      "grad_norm": 1.6530534029006958,
      "learning_rate": 2.9534848878684774e-06,
      "loss": 0.4931,
      "step": 217630
    },
    {
      "epoch": 89.09,
      "grad_norm": 2.1683709621429443,
      "learning_rate": 2.95333171897329e-06,
      "loss": 0.5043,
      "step": 217640
    },
    {
      "epoch": 89.09,
      "grad_norm": 1.602908968925476,
      "learning_rate": 2.9531785472857027e-06,
      "loss": 0.5119,
      "step": 217650
    },
    {
      "epoch": 89.1,
      "grad_norm": 2.208709955215454,
      "learning_rate": 2.9530253728064178e-06,
      "loss": 0.475,
      "step": 217660
    },
    {
      "epoch": 89.1,
      "grad_norm": 1.8583710193634033,
      "learning_rate": 2.9528721955361365e-06,
      "loss": 0.5023,
      "step": 217670
    },
    {
      "epoch": 89.1,
      "grad_norm": 1.6646090745925903,
      "learning_rate": 2.952719015475561e-06,
      "loss": 0.4974,
      "step": 217680
    },
    {
      "epoch": 89.11,
      "grad_norm": 1.6686898469924927,
      "learning_rate": 2.9525658326253924e-06,
      "loss": 0.4762,
      "step": 217690
    },
    {
      "epoch": 89.11,
      "grad_norm": 1.955348253250122,
      "learning_rate": 2.9524126469863333e-06,
      "loss": 0.4848,
      "step": 217700
    },
    {
      "epoch": 89.12,
      "grad_norm": 1.8198013305664062,
      "learning_rate": 2.9522594585590843e-06,
      "loss": 0.521,
      "step": 217710
    },
    {
      "epoch": 89.12,
      "grad_norm": 2.6108367443084717,
      "learning_rate": 2.9521062673443477e-06,
      "loss": 0.5086,
      "step": 217720
    },
    {
      "epoch": 89.12,
      "grad_norm": 1.6729480028152466,
      "learning_rate": 2.9519530733428254e-06,
      "loss": 0.4819,
      "step": 217730
    },
    {
      "epoch": 89.13,
      "grad_norm": 1.81737220287323,
      "learning_rate": 2.9517998765552193e-06,
      "loss": 0.5094,
      "step": 217740
    },
    {
      "epoch": 89.13,
      "grad_norm": 1.7154784202575684,
      "learning_rate": 2.951646676982231e-06,
      "loss": 0.5178,
      "step": 217750
    },
    {
      "epoch": 89.14,
      "grad_norm": 1.980812907218933,
      "learning_rate": 2.951493474624562e-06,
      "loss": 0.5024,
      "step": 217760
    },
    {
      "epoch": 89.14,
      "grad_norm": 1.7593046426773071,
      "learning_rate": 2.951340269482914e-06,
      "loss": 0.482,
      "step": 217770
    },
    {
      "epoch": 89.14,
      "grad_norm": 1.5155020952224731,
      "learning_rate": 2.951187061557989e-06,
      "loss": 0.5182,
      "step": 217780
    },
    {
      "epoch": 89.15,
      "grad_norm": 1.7439453601837158,
      "learning_rate": 2.9510338508504892e-06,
      "loss": 0.4984,
      "step": 217790
    },
    {
      "epoch": 89.15,
      "grad_norm": 1.6701838970184326,
      "learning_rate": 2.9508806373611168e-06,
      "loss": 0.4848,
      "step": 217800
    },
    {
      "epoch": 89.16,
      "grad_norm": 1.8394250869750977,
      "learning_rate": 2.9507274210905718e-06,
      "loss": 0.4909,
      "step": 217810
    },
    {
      "epoch": 89.16,
      "grad_norm": 1.8361752033233643,
      "learning_rate": 2.9505742020395577e-06,
      "loss": 0.4913,
      "step": 217820
    },
    {
      "epoch": 89.16,
      "grad_norm": 1.6004136800765991,
      "learning_rate": 2.950420980208776e-06,
      "loss": 0.4928,
      "step": 217830
    },
    {
      "epoch": 89.17,
      "grad_norm": 1.5362225770950317,
      "learning_rate": 2.9502677555989285e-06,
      "loss": 0.4821,
      "step": 217840
    },
    {
      "epoch": 89.17,
      "grad_norm": 1.550026297569275,
      "learning_rate": 2.950114528210717e-06,
      "loss": 0.5072,
      "step": 217850
    },
    {
      "epoch": 89.18,
      "grad_norm": 1.8612887859344482,
      "learning_rate": 2.9499612980448438e-06,
      "loss": 0.4951,
      "step": 217860
    },
    {
      "epoch": 89.18,
      "grad_norm": 1.479553461074829,
      "learning_rate": 2.94980806510201e-06,
      "loss": 0.4834,
      "step": 217870
    },
    {
      "epoch": 89.19,
      "grad_norm": 1.8571134805679321,
      "learning_rate": 2.949654829382918e-06,
      "loss": 0.5099,
      "step": 217880
    },
    {
      "epoch": 89.19,
      "grad_norm": 1.6268062591552734,
      "learning_rate": 2.9495015908882707e-06,
      "loss": 0.4884,
      "step": 217890
    },
    {
      "epoch": 89.19,
      "grad_norm": 1.5363938808441162,
      "learning_rate": 2.9493483496187675e-06,
      "loss": 0.4855,
      "step": 217900
    },
    {
      "epoch": 89.2,
      "grad_norm": 2.036181688308716,
      "learning_rate": 2.9491951055751133e-06,
      "loss": 0.5092,
      "step": 217910
    },
    {
      "epoch": 89.2,
      "grad_norm": 2.037992000579834,
      "learning_rate": 2.9490418587580086e-06,
      "loss": 0.5059,
      "step": 217920
    },
    {
      "epoch": 89.21,
      "grad_norm": 1.2540725469589233,
      "learning_rate": 2.9488886091681563e-06,
      "loss": 0.4794,
      "step": 217930
    },
    {
      "epoch": 89.21,
      "grad_norm": 2.430962562561035,
      "learning_rate": 2.9487353568062574e-06,
      "loss": 0.4933,
      "step": 217940
    },
    {
      "epoch": 89.21,
      "grad_norm": 2.342822551727295,
      "learning_rate": 2.9485821016730136e-06,
      "loss": 0.4734,
      "step": 217950
    },
    {
      "epoch": 89.22,
      "grad_norm": 1.540234923362732,
      "learning_rate": 2.948428843769128e-06,
      "loss": 0.4851,
      "step": 217960
    },
    {
      "epoch": 89.22,
      "grad_norm": 1.8753973245620728,
      "learning_rate": 2.948275583095302e-06,
      "loss": 0.5026,
      "step": 217970
    },
    {
      "epoch": 89.23,
      "grad_norm": 2.4319355487823486,
      "learning_rate": 2.948122319652238e-06,
      "loss": 0.4991,
      "step": 217980
    },
    {
      "epoch": 89.23,
      "grad_norm": 1.8053046464920044,
      "learning_rate": 2.947969053440639e-06,
      "loss": 0.5062,
      "step": 217990
    },
    {
      "epoch": 89.23,
      "grad_norm": 1.6162713766098022,
      "learning_rate": 2.947815784461205e-06,
      "loss": 0.4841,
      "step": 218000
    },
    {
      "epoch": 89.24,
      "grad_norm": 2.3837993144989014,
      "learning_rate": 2.9476625127146393e-06,
      "loss": 0.4838,
      "step": 218010
    },
    {
      "epoch": 89.24,
      "grad_norm": 2.0075907707214355,
      "learning_rate": 2.9475092382016442e-06,
      "loss": 0.4908,
      "step": 218020
    },
    {
      "epoch": 89.25,
      "grad_norm": 1.526584267616272,
      "learning_rate": 2.9473559609229214e-06,
      "loss": 0.4974,
      "step": 218030
    },
    {
      "epoch": 89.25,
      "grad_norm": 2.1099236011505127,
      "learning_rate": 2.9472026808791735e-06,
      "loss": 0.4913,
      "step": 218040
    },
    {
      "epoch": 89.26,
      "grad_norm": 1.7582533359527588,
      "learning_rate": 2.9470493980711022e-06,
      "loss": 0.4845,
      "step": 218050
    },
    {
      "epoch": 89.26,
      "grad_norm": 1.9440582990646362,
      "learning_rate": 2.946896112499409e-06,
      "loss": 0.4996,
      "step": 218060
    },
    {
      "epoch": 89.26,
      "grad_norm": 1.851366400718689,
      "learning_rate": 2.946742824164798e-06,
      "loss": 0.4852,
      "step": 218070
    },
    {
      "epoch": 89.27,
      "grad_norm": 1.6895123720169067,
      "learning_rate": 2.9465895330679697e-06,
      "loss": 0.4898,
      "step": 218080
    },
    {
      "epoch": 89.27,
      "grad_norm": 1.8630532026290894,
      "learning_rate": 2.9464362392096272e-06,
      "loss": 0.471,
      "step": 218090
    },
    {
      "epoch": 89.28,
      "grad_norm": 1.70518159866333,
      "learning_rate": 2.9462829425904716e-06,
      "loss": 0.4956,
      "step": 218100
    },
    {
      "epoch": 89.28,
      "grad_norm": 2.0099308490753174,
      "learning_rate": 2.9461296432112065e-06,
      "loss": 0.4964,
      "step": 218110
    },
    {
      "epoch": 89.28,
      "grad_norm": 1.4588088989257812,
      "learning_rate": 2.9459763410725334e-06,
      "loss": 0.5023,
      "step": 218120
    },
    {
      "epoch": 89.29,
      "grad_norm": 1.9110124111175537,
      "learning_rate": 2.9458230361751545e-06,
      "loss": 0.5015,
      "step": 218130
    },
    {
      "epoch": 89.29,
      "grad_norm": 2.3376801013946533,
      "learning_rate": 2.9456697285197723e-06,
      "loss": 0.5073,
      "step": 218140
    },
    {
      "epoch": 89.3,
      "grad_norm": 2.0560057163238525,
      "learning_rate": 2.9455164181070896e-06,
      "loss": 0.4839,
      "step": 218150
    },
    {
      "epoch": 89.3,
      "grad_norm": 1.8611524105072021,
      "learning_rate": 2.945363104937808e-06,
      "loss": 0.5084,
      "step": 218160
    },
    {
      "epoch": 89.3,
      "grad_norm": 2.685307264328003,
      "learning_rate": 2.94520978901263e-06,
      "loss": 0.4734,
      "step": 218170
    },
    {
      "epoch": 89.31,
      "grad_norm": 1.7762844562530518,
      "learning_rate": 2.9450564703322574e-06,
      "loss": 0.4805,
      "step": 218180
    },
    {
      "epoch": 89.31,
      "grad_norm": 2.097001075744629,
      "learning_rate": 2.9449031488973924e-06,
      "loss": 0.4881,
      "step": 218190
    },
    {
      "epoch": 89.32,
      "grad_norm": 1.7135906219482422,
      "learning_rate": 2.944749824708739e-06,
      "loss": 0.4962,
      "step": 218200
    },
    {
      "epoch": 89.32,
      "grad_norm": 1.7373982667922974,
      "learning_rate": 2.9445964977669985e-06,
      "loss": 0.5103,
      "step": 218210
    },
    {
      "epoch": 89.32,
      "grad_norm": 1.9421533346176147,
      "learning_rate": 2.9444431680728732e-06,
      "loss": 0.4827,
      "step": 218220
    },
    {
      "epoch": 89.33,
      "grad_norm": 1.737513780593872,
      "learning_rate": 2.9442898356270653e-06,
      "loss": 0.4952,
      "step": 218230
    },
    {
      "epoch": 89.33,
      "grad_norm": 1.4238178730010986,
      "learning_rate": 2.944136500430277e-06,
      "loss": 0.468,
      "step": 218240
    },
    {
      "epoch": 89.34,
      "grad_norm": 1.7855415344238281,
      "learning_rate": 2.9439831624832118e-06,
      "loss": 0.5021,
      "step": 218250
    },
    {
      "epoch": 89.34,
      "grad_norm": 1.8292347192764282,
      "learning_rate": 2.943829821786571e-06,
      "loss": 0.4705,
      "step": 218260
    },
    {
      "epoch": 89.35,
      "grad_norm": 1.945142388343811,
      "learning_rate": 2.9436764783410577e-06,
      "loss": 0.4871,
      "step": 218270
    },
    {
      "epoch": 89.35,
      "grad_norm": 2.040040969848633,
      "learning_rate": 2.9435231321473735e-06,
      "loss": 0.5109,
      "step": 218280
    },
    {
      "epoch": 89.35,
      "grad_norm": 1.7237765789031982,
      "learning_rate": 2.9433697832062223e-06,
      "loss": 0.4918,
      "step": 218290
    },
    {
      "epoch": 89.36,
      "grad_norm": 1.8124438524246216,
      "learning_rate": 2.9432164315183054e-06,
      "loss": 0.498,
      "step": 218300
    },
    {
      "epoch": 89.36,
      "grad_norm": 2.036825180053711,
      "learning_rate": 2.9430630770843256e-06,
      "loss": 0.4871,
      "step": 218310
    },
    {
      "epoch": 89.37,
      "grad_norm": 1.6601347923278809,
      "learning_rate": 2.9429097199049853e-06,
      "loss": 0.4898,
      "step": 218320
    },
    {
      "epoch": 89.37,
      "grad_norm": 2.012326717376709,
      "learning_rate": 2.942756359980987e-06,
      "loss": 0.508,
      "step": 218330
    },
    {
      "epoch": 89.37,
      "grad_norm": 1.879050374031067,
      "learning_rate": 2.942602997313034e-06,
      "loss": 0.4916,
      "step": 218340
    },
    {
      "epoch": 89.38,
      "grad_norm": 1.8529468774795532,
      "learning_rate": 2.9424496319018272e-06,
      "loss": 0.4831,
      "step": 218350
    },
    {
      "epoch": 89.38,
      "grad_norm": 1.848387598991394,
      "learning_rate": 2.942296263748071e-06,
      "loss": 0.497,
      "step": 218360
    },
    {
      "epoch": 89.39,
      "grad_norm": 1.993351936340332,
      "learning_rate": 2.9421428928524666e-06,
      "loss": 0.4893,
      "step": 218370
    },
    {
      "epoch": 89.39,
      "grad_norm": 1.8659292459487915,
      "learning_rate": 2.941989519215717e-06,
      "loss": 0.4882,
      "step": 218380
    },
    {
      "epoch": 89.39,
      "grad_norm": 2.7863917350769043,
      "learning_rate": 2.941836142838525e-06,
      "loss": 0.498,
      "step": 218390
    },
    {
      "epoch": 89.4,
      "grad_norm": 2.1164252758026123,
      "learning_rate": 2.9416827637215933e-06,
      "loss": 0.4883,
      "step": 218400
    },
    {
      "epoch": 89.4,
      "grad_norm": 1.9984242916107178,
      "learning_rate": 2.9415293818656235e-06,
      "loss": 0.5031,
      "step": 218410
    },
    {
      "epoch": 89.41,
      "grad_norm": 2.116583824157715,
      "learning_rate": 2.9413759972713195e-06,
      "loss": 0.4773,
      "step": 218420
    },
    {
      "epoch": 89.41,
      "grad_norm": 1.7942965030670166,
      "learning_rate": 2.941222609939383e-06,
      "loss": 0.4926,
      "step": 218430
    },
    {
      "epoch": 89.41,
      "grad_norm": 2.3915915489196777,
      "learning_rate": 2.9410692198705173e-06,
      "loss": 0.4799,
      "step": 218440
    },
    {
      "epoch": 89.42,
      "grad_norm": 1.6844253540039062,
      "learning_rate": 2.940915827065425e-06,
      "loss": 0.4965,
      "step": 218450
    },
    {
      "epoch": 89.42,
      "grad_norm": 1.7538610696792603,
      "learning_rate": 2.9407624315248077e-06,
      "loss": 0.4979,
      "step": 218460
    },
    {
      "epoch": 89.43,
      "grad_norm": 2.295461893081665,
      "learning_rate": 2.940609033249369e-06,
      "loss": 0.4768,
      "step": 218470
    },
    {
      "epoch": 89.43,
      "grad_norm": 1.8163007497787476,
      "learning_rate": 2.940455632239812e-06,
      "loss": 0.4818,
      "step": 218480
    },
    {
      "epoch": 89.44,
      "grad_norm": 1.880385160446167,
      "learning_rate": 2.940302228496839e-06,
      "loss": 0.4653,
      "step": 218490
    },
    {
      "epoch": 89.44,
      "grad_norm": 1.8877581357955933,
      "learning_rate": 2.940148822021152e-06,
      "loss": 0.5191,
      "step": 218500
    },
    {
      "epoch": 89.44,
      "grad_norm": 1.9396061897277832,
      "learning_rate": 2.939995412813455e-06,
      "loss": 0.4916,
      "step": 218510
    },
    {
      "epoch": 89.45,
      "grad_norm": 1.764948844909668,
      "learning_rate": 2.9398420008744502e-06,
      "loss": 0.4816,
      "step": 218520
    },
    {
      "epoch": 89.45,
      "grad_norm": 1.9062135219573975,
      "learning_rate": 2.9396885862048404e-06,
      "loss": 0.4828,
      "step": 218530
    },
    {
      "epoch": 89.46,
      "grad_norm": 2.1845130920410156,
      "learning_rate": 2.9395351688053272e-06,
      "loss": 0.4875,
      "step": 218540
    },
    {
      "epoch": 89.46,
      "grad_norm": 2.3542487621307373,
      "learning_rate": 2.939381748676615e-06,
      "loss": 0.4854,
      "step": 218550
    },
    {
      "epoch": 89.46,
      "grad_norm": 2.0215470790863037,
      "learning_rate": 2.939228325819406e-06,
      "loss": 0.479,
      "step": 218560
    },
    {
      "epoch": 89.47,
      "grad_norm": 2.010265588760376,
      "learning_rate": 2.9390749002344028e-06,
      "loss": 0.4908,
      "step": 218570
    },
    {
      "epoch": 89.47,
      "grad_norm": 2.5484225749969482,
      "learning_rate": 2.9389214719223088e-06,
      "loss": 0.4787,
      "step": 218580
    },
    {
      "epoch": 89.48,
      "grad_norm": 2.9156899452209473,
      "learning_rate": 2.938768040883826e-06,
      "loss": 0.5084,
      "step": 218590
    },
    {
      "epoch": 89.48,
      "grad_norm": 1.8129022121429443,
      "learning_rate": 2.938614607119658e-06,
      "loss": 0.479,
      "step": 218600
    },
    {
      "epoch": 89.48,
      "grad_norm": 1.9694796800613403,
      "learning_rate": 2.9384611706305075e-06,
      "loss": 0.4936,
      "step": 218610
    },
    {
      "epoch": 89.49,
      "grad_norm": 2.094343900680542,
      "learning_rate": 2.9383077314170774e-06,
      "loss": 0.4772,
      "step": 218620
    },
    {
      "epoch": 89.49,
      "grad_norm": 2.0720715522766113,
      "learning_rate": 2.9381542894800697e-06,
      "loss": 0.4755,
      "step": 218630
    },
    {
      "epoch": 89.5,
      "grad_norm": 2.255293607711792,
      "learning_rate": 2.9380008448201885e-06,
      "loss": 0.4888,
      "step": 218640
    },
    {
      "epoch": 89.5,
      "grad_norm": 2.120694160461426,
      "learning_rate": 2.937847397438136e-06,
      "loss": 0.5115,
      "step": 218650
    },
    {
      "epoch": 89.5,
      "grad_norm": 1.6461505889892578,
      "learning_rate": 2.9376939473346153e-06,
      "loss": 0.4866,
      "step": 218660
    },
    {
      "epoch": 89.51,
      "grad_norm": 1.7277977466583252,
      "learning_rate": 2.9375404945103297e-06,
      "loss": 0.4877,
      "step": 218670
    },
    {
      "epoch": 89.51,
      "grad_norm": 1.7267842292785645,
      "learning_rate": 2.9373870389659816e-06,
      "loss": 0.4869,
      "step": 218680
    },
    {
      "epoch": 89.52,
      "grad_norm": 2.4284634590148926,
      "learning_rate": 2.937233580702274e-06,
      "loss": 0.5105,
      "step": 218690
    },
    {
      "epoch": 89.52,
      "grad_norm": 2.189481496810913,
      "learning_rate": 2.9370801197199104e-06,
      "loss": 0.5078,
      "step": 218700
    },
    {
      "epoch": 89.53,
      "grad_norm": 1.7534438371658325,
      "learning_rate": 2.936926656019593e-06,
      "loss": 0.4915,
      "step": 218710
    },
    {
      "epoch": 89.53,
      "grad_norm": 2.203911304473877,
      "learning_rate": 2.9367731896020253e-06,
      "loss": 0.502,
      "step": 218720
    },
    {
      "epoch": 89.53,
      "grad_norm": 1.9743436574935913,
      "learning_rate": 2.9366197204679102e-06,
      "loss": 0.511,
      "step": 218730
    },
    {
      "epoch": 89.54,
      "grad_norm": 1.4921494722366333,
      "learning_rate": 2.9364662486179518e-06,
      "loss": 0.505,
      "step": 218740
    },
    {
      "epoch": 89.54,
      "grad_norm": 1.9418227672576904,
      "learning_rate": 2.9363127740528503e-06,
      "loss": 0.4894,
      "step": 218750
    },
    {
      "epoch": 89.55,
      "grad_norm": 1.6921168565750122,
      "learning_rate": 2.936159296773311e-06,
      "loss": 0.4963,
      "step": 218760
    },
    {
      "epoch": 89.55,
      "grad_norm": 1.7642207145690918,
      "learning_rate": 2.936005816780037e-06,
      "loss": 0.4773,
      "step": 218770
    },
    {
      "epoch": 89.55,
      "grad_norm": 1.9905071258544922,
      "learning_rate": 2.9358523340737306e-06,
      "loss": 0.475,
      "step": 218780
    },
    {
      "epoch": 89.56,
      "grad_norm": 2.562930107116699,
      "learning_rate": 2.9356988486550955e-06,
      "loss": 0.5109,
      "step": 218790
    },
    {
      "epoch": 89.56,
      "grad_norm": 2.112931251525879,
      "learning_rate": 2.9355453605248334e-06,
      "loss": 0.5118,
      "step": 218800
    },
    {
      "epoch": 89.57,
      "grad_norm": 1.9673725366592407,
      "learning_rate": 2.9353918696836495e-06,
      "loss": 0.5029,
      "step": 218810
    },
    {
      "epoch": 89.57,
      "grad_norm": 1.9634215831756592,
      "learning_rate": 2.9352383761322453e-06,
      "loss": 0.4969,
      "step": 218820
    },
    {
      "epoch": 89.57,
      "grad_norm": 1.906097650527954,
      "learning_rate": 2.9350848798713245e-06,
      "loss": 0.4789,
      "step": 218830
    },
    {
      "epoch": 89.58,
      "grad_norm": 1.8666032552719116,
      "learning_rate": 2.93493138090159e-06,
      "loss": 0.4707,
      "step": 218840
    },
    {
      "epoch": 89.58,
      "grad_norm": 2.1300785541534424,
      "learning_rate": 2.9347778792237448e-06,
      "loss": 0.4894,
      "step": 218850
    },
    {
      "epoch": 89.59,
      "grad_norm": 2.066340208053589,
      "learning_rate": 2.934624374838493e-06,
      "loss": 0.4922,
      "step": 218860
    },
    {
      "epoch": 89.59,
      "grad_norm": 1.771743893623352,
      "learning_rate": 2.934470867746538e-06,
      "loss": 0.4832,
      "step": 218870
    },
    {
      "epoch": 89.59,
      "grad_norm": 1.8435559272766113,
      "learning_rate": 2.934317357948581e-06,
      "loss": 0.5067,
      "step": 218880
    },
    {
      "epoch": 89.6,
      "grad_norm": 2.009509563446045,
      "learning_rate": 2.9341638454453265e-06,
      "loss": 0.4813,
      "step": 218890
    },
    {
      "epoch": 89.6,
      "grad_norm": 1.6807972192764282,
      "learning_rate": 2.9340103302374777e-06,
      "loss": 0.487,
      "step": 218900
    },
    {
      "epoch": 89.61,
      "grad_norm": 1.6708344221115112,
      "learning_rate": 2.933856812325738e-06,
      "loss": 0.4903,
      "step": 218910
    },
    {
      "epoch": 89.61,
      "grad_norm": 2.417045831680298,
      "learning_rate": 2.9337032917108103e-06,
      "loss": 0.5,
      "step": 218920
    },
    {
      "epoch": 89.62,
      "grad_norm": 2.0157392024993896,
      "learning_rate": 2.9335497683933977e-06,
      "loss": 0.4854,
      "step": 218930
    },
    {
      "epoch": 89.62,
      "grad_norm": 1.9489405155181885,
      "learning_rate": 2.933396242374204e-06,
      "loss": 0.4785,
      "step": 218940
    },
    {
      "epoch": 89.62,
      "grad_norm": 1.724367380142212,
      "learning_rate": 2.933242713653932e-06,
      "loss": 0.4875,
      "step": 218950
    },
    {
      "epoch": 89.63,
      "grad_norm": 1.7476905584335327,
      "learning_rate": 2.9330891822332854e-06,
      "loss": 0.4787,
      "step": 218960
    },
    {
      "epoch": 89.63,
      "grad_norm": 1.706431269645691,
      "learning_rate": 2.9329356481129667e-06,
      "loss": 0.4818,
      "step": 218970
    },
    {
      "epoch": 89.64,
      "grad_norm": 1.6025283336639404,
      "learning_rate": 2.9327821112936804e-06,
      "loss": 0.5056,
      "step": 218980
    },
    {
      "epoch": 89.64,
      "grad_norm": 2.1383397579193115,
      "learning_rate": 2.932628571776129e-06,
      "loss": 0.5172,
      "step": 218990
    },
    {
      "epoch": 89.64,
      "grad_norm": 1.5136749744415283,
      "learning_rate": 2.9324750295610164e-06,
      "loss": 0.4984,
      "step": 219000
    },
    {
      "epoch": 89.65,
      "grad_norm": 2.043241024017334,
      "learning_rate": 2.932321484649045e-06,
      "loss": 0.4861,
      "step": 219010
    },
    {
      "epoch": 89.65,
      "grad_norm": 1.8704111576080322,
      "learning_rate": 2.932167937040919e-06,
      "loss": 0.4843,
      "step": 219020
    },
    {
      "epoch": 89.66,
      "grad_norm": 1.4317777156829834,
      "learning_rate": 2.9320143867373417e-06,
      "loss": 0.4823,
      "step": 219030
    },
    {
      "epoch": 89.66,
      "grad_norm": 1.8660178184509277,
      "learning_rate": 2.9318608337390163e-06,
      "loss": 0.4932,
      "step": 219040
    },
    {
      "epoch": 89.66,
      "grad_norm": 2.281125783920288,
      "learning_rate": 2.9317072780466467e-06,
      "loss": 0.4684,
      "step": 219050
    },
    {
      "epoch": 89.67,
      "grad_norm": 1.99667489528656,
      "learning_rate": 2.9315537196609347e-06,
      "loss": 0.4844,
      "step": 219060
    },
    {
      "epoch": 89.67,
      "grad_norm": 1.9852648973464966,
      "learning_rate": 2.931400158582586e-06,
      "loss": 0.4903,
      "step": 219070
    },
    {
      "epoch": 89.68,
      "grad_norm": 1.9512382745742798,
      "learning_rate": 2.9312465948123023e-06,
      "loss": 0.4714,
      "step": 219080
    },
    {
      "epoch": 89.68,
      "grad_norm": 2.460225820541382,
      "learning_rate": 2.9310930283507884e-06,
      "loss": 0.4913,
      "step": 219090
    },
    {
      "epoch": 89.68,
      "grad_norm": 1.6103737354278564,
      "learning_rate": 2.9309394591987464e-06,
      "loss": 0.4929,
      "step": 219100
    },
    {
      "epoch": 89.69,
      "grad_norm": 2.678234815597534,
      "learning_rate": 2.930785887356881e-06,
      "loss": 0.498,
      "step": 219110
    },
    {
      "epoch": 89.69,
      "grad_norm": 2.639984130859375,
      "learning_rate": 2.9306323128258947e-06,
      "loss": 0.5043,
      "step": 219120
    },
    {
      "epoch": 89.7,
      "grad_norm": 2.1394670009613037,
      "learning_rate": 2.9304787356064906e-06,
      "loss": 0.4826,
      "step": 219130
    },
    {
      "epoch": 89.7,
      "grad_norm": 1.8850783109664917,
      "learning_rate": 2.930325155699374e-06,
      "loss": 0.4823,
      "step": 219140
    },
    {
      "epoch": 89.71,
      "grad_norm": 1.604967474937439,
      "learning_rate": 2.9301715731052473e-06,
      "loss": 0.5044,
      "step": 219150
    },
    {
      "epoch": 89.71,
      "grad_norm": 1.7153427600860596,
      "learning_rate": 2.9300179878248145e-06,
      "loss": 0.4658,
      "step": 219160
    },
    {
      "epoch": 89.71,
      "grad_norm": 2.4254038333892822,
      "learning_rate": 2.929864399858779e-06,
      "loss": 0.4833,
      "step": 219170
    },
    {
      "epoch": 89.72,
      "grad_norm": 1.9205890893936157,
      "learning_rate": 2.9297108092078435e-06,
      "loss": 0.4982,
      "step": 219180
    },
    {
      "epoch": 89.72,
      "grad_norm": 2.8221771717071533,
      "learning_rate": 2.9295572158727122e-06,
      "loss": 0.4749,
      "step": 219190
    },
    {
      "epoch": 89.73,
      "grad_norm": 1.888783574104309,
      "learning_rate": 2.9294036198540895e-06,
      "loss": 0.4918,
      "step": 219200
    },
    {
      "epoch": 89.73,
      "grad_norm": 2.4342620372772217,
      "learning_rate": 2.9292500211526776e-06,
      "loss": 0.4874,
      "step": 219210
    },
    {
      "epoch": 89.73,
      "grad_norm": 1.6770457029342651,
      "learning_rate": 2.9290964197691807e-06,
      "loss": 0.4779,
      "step": 219220
    },
    {
      "epoch": 89.74,
      "grad_norm": 2.1539580821990967,
      "learning_rate": 2.9289428157043033e-06,
      "loss": 0.4842,
      "step": 219230
    },
    {
      "epoch": 89.74,
      "grad_norm": 1.9777122735977173,
      "learning_rate": 2.9287892089587477e-06,
      "loss": 0.5044,
      "step": 219240
    },
    {
      "epoch": 89.75,
      "grad_norm": 2.2558302879333496,
      "learning_rate": 2.9286355995332177e-06,
      "loss": 0.478,
      "step": 219250
    },
    {
      "epoch": 89.75,
      "grad_norm": 1.8357337713241577,
      "learning_rate": 2.9284819874284182e-06,
      "loss": 0.4967,
      "step": 219260
    },
    {
      "epoch": 89.75,
      "grad_norm": 1.8248674869537354,
      "learning_rate": 2.9283283726450516e-06,
      "loss": 0.4773,
      "step": 219270
    },
    {
      "epoch": 89.76,
      "grad_norm": 2.09466552734375,
      "learning_rate": 2.928174755183822e-06,
      "loss": 0.4891,
      "step": 219280
    },
    {
      "epoch": 89.76,
      "grad_norm": 1.6049350500106812,
      "learning_rate": 2.928021135045433e-06,
      "loss": 0.4656,
      "step": 219290
    },
    {
      "epoch": 89.77,
      "grad_norm": 2.034681797027588,
      "learning_rate": 2.9278675122305885e-06,
      "loss": 0.4772,
      "step": 219300
    },
    {
      "epoch": 89.77,
      "grad_norm": 2.1764066219329834,
      "learning_rate": 2.9277138867399927e-06,
      "loss": 0.5078,
      "step": 219310
    },
    {
      "epoch": 89.77,
      "grad_norm": 1.8740826845169067,
      "learning_rate": 2.927560258574348e-06,
      "loss": 0.4995,
      "step": 219320
    },
    {
      "epoch": 89.78,
      "grad_norm": 1.7798938751220703,
      "learning_rate": 2.9274066277343595e-06,
      "loss": 0.4832,
      "step": 219330
    },
    {
      "epoch": 89.78,
      "grad_norm": 1.5574921369552612,
      "learning_rate": 2.9272529942207303e-06,
      "loss": 0.4855,
      "step": 219340
    },
    {
      "epoch": 89.79,
      "grad_norm": 1.6203473806381226,
      "learning_rate": 2.9270993580341637e-06,
      "loss": 0.4871,
      "step": 219350
    },
    {
      "epoch": 89.79,
      "grad_norm": 2.8262484073638916,
      "learning_rate": 2.926945719175364e-06,
      "loss": 0.4915,
      "step": 219360
    },
    {
      "epoch": 89.8,
      "grad_norm": 1.8028315305709839,
      "learning_rate": 2.9267920776450355e-06,
      "loss": 0.4846,
      "step": 219370
    },
    {
      "epoch": 89.8,
      "grad_norm": 2.025900363922119,
      "learning_rate": 2.9266384334438817e-06,
      "loss": 0.4929,
      "step": 219380
    },
    {
      "epoch": 89.8,
      "grad_norm": 1.8970321416854858,
      "learning_rate": 2.9264847865726065e-06,
      "loss": 0.4913,
      "step": 219390
    },
    {
      "epoch": 89.81,
      "grad_norm": 2.3763179779052734,
      "learning_rate": 2.9263311370319124e-06,
      "loss": 0.4959,
      "step": 219400
    },
    {
      "epoch": 89.81,
      "grad_norm": 1.972562313079834,
      "learning_rate": 2.9261774848225047e-06,
      "loss": 0.4984,
      "step": 219410
    },
    {
      "epoch": 89.82,
      "grad_norm": 2.1011414527893066,
      "learning_rate": 2.926023829945087e-06,
      "loss": 0.5025,
      "step": 219420
    },
    {
      "epoch": 89.82,
      "grad_norm": 1.6702845096588135,
      "learning_rate": 2.925870172400363e-06,
      "loss": 0.4944,
      "step": 219430
    },
    {
      "epoch": 89.82,
      "grad_norm": 1.548804759979248,
      "learning_rate": 2.9257165121890374e-06,
      "loss": 0.4894,
      "step": 219440
    },
    {
      "epoch": 89.83,
      "grad_norm": 2.7885189056396484,
      "learning_rate": 2.9255628493118125e-06,
      "loss": 0.5107,
      "step": 219450
    },
    {
      "epoch": 89.83,
      "grad_norm": 2.8931541442871094,
      "learning_rate": 2.925409183769394e-06,
      "loss": 0.4908,
      "step": 219460
    },
    {
      "epoch": 89.84,
      "grad_norm": 1.878240942955017,
      "learning_rate": 2.9252555155624835e-06,
      "loss": 0.5085,
      "step": 219470
    },
    {
      "epoch": 89.84,
      "grad_norm": 2.244567394256592,
      "learning_rate": 2.9251018446917876e-06,
      "loss": 0.4919,
      "step": 219480
    },
    {
      "epoch": 89.84,
      "grad_norm": 1.5332003831863403,
      "learning_rate": 2.924948171158008e-06,
      "loss": 0.4798,
      "step": 219490
    },
    {
      "epoch": 89.85,
      "grad_norm": 2.1525821685791016,
      "learning_rate": 2.92479449496185e-06,
      "loss": 0.4914,
      "step": 219500
    },
    {
      "epoch": 89.85,
      "grad_norm": 1.7526075839996338,
      "learning_rate": 2.9246408161040167e-06,
      "loss": 0.5089,
      "step": 219510
    },
    {
      "epoch": 89.86,
      "grad_norm": 2.199159860610962,
      "learning_rate": 2.9244871345852123e-06,
      "loss": 0.5149,
      "step": 219520
    },
    {
      "epoch": 89.86,
      "grad_norm": 1.677223801612854,
      "learning_rate": 2.9243334504061422e-06,
      "loss": 0.4996,
      "step": 219530
    },
    {
      "epoch": 89.86,
      "grad_norm": 2.031770706176758,
      "learning_rate": 2.9241797635675085e-06,
      "loss": 0.4777,
      "step": 219540
    },
    {
      "epoch": 89.87,
      "grad_norm": 1.9861407279968262,
      "learning_rate": 2.924026074070016e-06,
      "loss": 0.5023,
      "step": 219550
    },
    {
      "epoch": 89.87,
      "grad_norm": 1.7769804000854492,
      "learning_rate": 2.9238723819143693e-06,
      "loss": 0.474,
      "step": 219560
    },
    {
      "epoch": 89.88,
      "grad_norm": 1.6829336881637573,
      "learning_rate": 2.9237186871012706e-06,
      "loss": 0.4796,
      "step": 219570
    },
    {
      "epoch": 89.88,
      "grad_norm": 1.5244454145431519,
      "learning_rate": 2.9235649896314264e-06,
      "loss": 0.4854,
      "step": 219580
    },
    {
      "epoch": 89.89,
      "grad_norm": 2.247795581817627,
      "learning_rate": 2.9234112895055385e-06,
      "loss": 0.5022,
      "step": 219590
    },
    {
      "epoch": 89.89,
      "grad_norm": 1.8674677610397339,
      "learning_rate": 2.9232575867243124e-06,
      "loss": 0.5154,
      "step": 219600
    },
    {
      "epoch": 89.89,
      "grad_norm": 1.9480113983154297,
      "learning_rate": 2.9231038812884516e-06,
      "loss": 0.5323,
      "step": 219610
    },
    {
      "epoch": 89.9,
      "grad_norm": 1.7134373188018799,
      "learning_rate": 2.9229501731986608e-06,
      "loss": 0.4871,
      "step": 219620
    },
    {
      "epoch": 89.9,
      "grad_norm": 1.6303342580795288,
      "learning_rate": 2.9227964624556435e-06,
      "loss": 0.4909,
      "step": 219630
    },
    {
      "epoch": 89.91,
      "grad_norm": 2.2342891693115234,
      "learning_rate": 2.9226427490601043e-06,
      "loss": 0.506,
      "step": 219640
    },
    {
      "epoch": 89.91,
      "grad_norm": 1.9454625844955444,
      "learning_rate": 2.922489033012747e-06,
      "loss": 0.4997,
      "step": 219650
    },
    {
      "epoch": 89.91,
      "grad_norm": 1.950626015663147,
      "learning_rate": 2.922335314314275e-06,
      "loss": 0.4675,
      "step": 219660
    },
    {
      "epoch": 89.92,
      "grad_norm": 2.006183385848999,
      "learning_rate": 2.9221815929653942e-06,
      "loss": 0.4826,
      "step": 219670
    },
    {
      "epoch": 89.92,
      "grad_norm": 1.635932207107544,
      "learning_rate": 2.922027868966808e-06,
      "loss": 0.4778,
      "step": 219680
    },
    {
      "epoch": 89.93,
      "grad_norm": 2.2973814010620117,
      "learning_rate": 2.9218741423192195e-06,
      "loss": 0.5057,
      "step": 219690
    },
    {
      "epoch": 89.93,
      "grad_norm": 2.182004690170288,
      "learning_rate": 2.9217204130233342e-06,
      "loss": 0.4753,
      "step": 219700
    },
    {
      "epoch": 89.93,
      "grad_norm": 1.8492995500564575,
      "learning_rate": 2.9215666810798564e-06,
      "loss": 0.4831,
      "step": 219710
    },
    {
      "epoch": 89.94,
      "grad_norm": 2.588796615600586,
      "learning_rate": 2.9214129464894896e-06,
      "loss": 0.4774,
      "step": 219720
    },
    {
      "epoch": 89.94,
      "grad_norm": 2.0145211219787598,
      "learning_rate": 2.921259209252938e-06,
      "loss": 0.5097,
      "step": 219730
    },
    {
      "epoch": 89.95,
      "grad_norm": 1.6125484704971313,
      "learning_rate": 2.9211054693709065e-06,
      "loss": 0.484,
      "step": 219740
    },
    {
      "epoch": 89.95,
      "grad_norm": 1.7906242609024048,
      "learning_rate": 2.9209517268440993e-06,
      "loss": 0.494,
      "step": 219750
    },
    {
      "epoch": 89.95,
      "grad_norm": 1.8544068336486816,
      "learning_rate": 2.92079798167322e-06,
      "loss": 0.4854,
      "step": 219760
    },
    {
      "epoch": 89.96,
      "grad_norm": 1.9490488767623901,
      "learning_rate": 2.9206442338589734e-06,
      "loss": 0.4923,
      "step": 219770
    },
    {
      "epoch": 89.96,
      "grad_norm": 2.0485284328460693,
      "learning_rate": 2.920490483402064e-06,
      "loss": 0.4947,
      "step": 219780
    },
    {
      "epoch": 89.97,
      "grad_norm": 1.8737592697143555,
      "learning_rate": 2.9203367303031944e-06,
      "loss": 0.4892,
      "step": 219790
    },
    {
      "epoch": 89.97,
      "grad_norm": 2.316514492034912,
      "learning_rate": 2.920182974563071e-06,
      "loss": 0.4995,
      "step": 219800
    },
    {
      "epoch": 89.98,
      "grad_norm": 1.8979511260986328,
      "learning_rate": 2.9200292161823976e-06,
      "loss": 0.5001,
      "step": 219810
    },
    {
      "epoch": 89.98,
      "grad_norm": 1.7664670944213867,
      "learning_rate": 2.9198754551618792e-06,
      "loss": 0.4822,
      "step": 219820
    },
    {
      "epoch": 89.98,
      "grad_norm": 1.9516355991363525,
      "learning_rate": 2.9197216915022183e-06,
      "loss": 0.4752,
      "step": 219830
    },
    {
      "epoch": 89.99,
      "grad_norm": 1.4608268737792969,
      "learning_rate": 2.9195679252041205e-06,
      "loss": 0.4879,
      "step": 219840
    },
    {
      "epoch": 89.99,
      "grad_norm": 2.2808103561401367,
      "learning_rate": 2.9194141562682897e-06,
      "loss": 0.4747,
      "step": 219850
    },
    {
      "epoch": 90.0,
      "grad_norm": 2.426152467727661,
      "learning_rate": 2.9192603846954306e-06,
      "loss": 0.5021,
      "step": 219860
    },
    {
      "epoch": 90.0,
      "grad_norm": 1.932152271270752,
      "learning_rate": 2.919106610486248e-06,
      "loss": 0.4844,
      "step": 219870
    },
    {
      "epoch": 90.0,
      "eval_loss": 0.49319443106651306,
      "eval_runtime": 74.6095,
      "eval_samples_per_second": 46.227,
      "eval_steps_per_second": 5.79,
      "step": 219870
    },
    {
      "epoch": 90.0,
      "grad_norm": 1.8678662776947021,
      "learning_rate": 2.9189528336414456e-06,
      "loss": 0.5128,
      "step": 219880
    },
    {
      "epoch": 90.01,
      "grad_norm": 1.6234136819839478,
      "learning_rate": 2.918799054161728e-06,
      "loss": 0.4904,
      "step": 219890
    },
    {
      "epoch": 90.01,
      "grad_norm": 1.8026001453399658,
      "learning_rate": 2.9186452720478e-06,
      "loss": 0.4891,
      "step": 219900
    },
    {
      "epoch": 90.02,
      "grad_norm": 2.0705795288085938,
      "learning_rate": 2.9184914873003657e-06,
      "loss": 0.4735,
      "step": 219910
    },
    {
      "epoch": 90.02,
      "grad_norm": 1.8507747650146484,
      "learning_rate": 2.9183376999201295e-06,
      "loss": 0.4879,
      "step": 219920
    },
    {
      "epoch": 90.02,
      "grad_norm": 1.77671217918396,
      "learning_rate": 2.918183909907796e-06,
      "loss": 0.4782,
      "step": 219930
    },
    {
      "epoch": 90.03,
      "grad_norm": 1.8888630867004395,
      "learning_rate": 2.91803011726407e-06,
      "loss": 0.5129,
      "step": 219940
    },
    {
      "epoch": 90.03,
      "grad_norm": 2.176928997039795,
      "learning_rate": 2.917876321989655e-06,
      "loss": 0.4909,
      "step": 219950
    },
    {
      "epoch": 90.04,
      "grad_norm": 1.8177627325057983,
      "learning_rate": 2.9177225240852573e-06,
      "loss": 0.4958,
      "step": 219960
    },
    {
      "epoch": 90.04,
      "grad_norm": 1.982875943183899,
      "learning_rate": 2.91756872355158e-06,
      "loss": 0.4964,
      "step": 219970
    },
    {
      "epoch": 90.05,
      "grad_norm": 1.7163513898849487,
      "learning_rate": 2.917414920389328e-06,
      "loss": 0.5005,
      "step": 219980
    },
    {
      "epoch": 90.05,
      "grad_norm": 1.6587079763412476,
      "learning_rate": 2.917261114599205e-06,
      "loss": 0.4961,
      "step": 219990
    },
    {
      "epoch": 90.05,
      "grad_norm": 2.415513277053833,
      "learning_rate": 2.9171073061819173e-06,
      "loss": 0.4909,
      "step": 220000
    },
    {
      "epoch": 90.06,
      "grad_norm": 1.5275663137435913,
      "learning_rate": 2.9169534951381686e-06,
      "loss": 0.4961,
      "step": 220010
    },
    {
      "epoch": 90.06,
      "grad_norm": 1.8357956409454346,
      "learning_rate": 2.9167996814686627e-06,
      "loss": 0.5024,
      "step": 220020
    },
    {
      "epoch": 90.07,
      "grad_norm": 2.603466749191284,
      "learning_rate": 2.916645865174106e-06,
      "loss": 0.5116,
      "step": 220030
    },
    {
      "epoch": 90.07,
      "grad_norm": 1.8604390621185303,
      "learning_rate": 2.9164920462552022e-06,
      "loss": 0.5091,
      "step": 220040
    },
    {
      "epoch": 90.07,
      "grad_norm": 1.7517112493515015,
      "learning_rate": 2.916338224712655e-06,
      "loss": 0.5088,
      "step": 220050
    },
    {
      "epoch": 90.08,
      "grad_norm": 2.5024945735931396,
      "learning_rate": 2.91618440054717e-06,
      "loss": 0.4888,
      "step": 220060
    },
    {
      "epoch": 90.08,
      "grad_norm": 2.1681811809539795,
      "learning_rate": 2.916030573759451e-06,
      "loss": 0.4914,
      "step": 220070
    },
    {
      "epoch": 90.09,
      "grad_norm": 1.5628526210784912,
      "learning_rate": 2.9158767443502035e-06,
      "loss": 0.4833,
      "step": 220080
    },
    {
      "epoch": 90.09,
      "grad_norm": 2.070803165435791,
      "learning_rate": 2.9157229123201327e-06,
      "loss": 0.4899,
      "step": 220090
    },
    {
      "epoch": 90.09,
      "grad_norm": 2.0624144077301025,
      "learning_rate": 2.9155690776699425e-06,
      "loss": 0.5085,
      "step": 220100
    },
    {
      "epoch": 90.1,
      "grad_norm": 2.2774789333343506,
      "learning_rate": 2.9154152404003377e-06,
      "loss": 0.4812,
      "step": 220110
    },
    {
      "epoch": 90.1,
      "grad_norm": 1.5940407514572144,
      "learning_rate": 2.915261400512023e-06,
      "loss": 0.4674,
      "step": 220120
    },
    {
      "epoch": 90.11,
      "grad_norm": 1.7953649759292603,
      "learning_rate": 2.9151075580057026e-06,
      "loss": 0.4922,
      "step": 220130
    },
    {
      "epoch": 90.11,
      "grad_norm": 2.350782632827759,
      "learning_rate": 2.914953712882082e-06,
      "loss": 0.4812,
      "step": 220140
    },
    {
      "epoch": 90.11,
      "grad_norm": 1.8930577039718628,
      "learning_rate": 2.9147998651418656e-06,
      "loss": 0.4943,
      "step": 220150
    },
    {
      "epoch": 90.12,
      "grad_norm": 1.866775631904602,
      "learning_rate": 2.9146460147857584e-06,
      "loss": 0.481,
      "step": 220160
    },
    {
      "epoch": 90.12,
      "grad_norm": 1.7363649606704712,
      "learning_rate": 2.914492161814464e-06,
      "loss": 0.518,
      "step": 220170
    },
    {
      "epoch": 90.13,
      "grad_norm": 1.4251853227615356,
      "learning_rate": 2.914338306228689e-06,
      "loss": 0.4935,
      "step": 220180
    },
    {
      "epoch": 90.13,
      "grad_norm": 2.296677827835083,
      "learning_rate": 2.914184448029138e-06,
      "loss": 0.4812,
      "step": 220190
    },
    {
      "epoch": 90.14,
      "grad_norm": 1.6805133819580078,
      "learning_rate": 2.914030587216514e-06,
      "loss": 0.4815,
      "step": 220200
    },
    {
      "epoch": 90.14,
      "grad_norm": 2.216740846633911,
      "learning_rate": 2.913876723791523e-06,
      "loss": 0.5046,
      "step": 220210
    },
    {
      "epoch": 90.14,
      "grad_norm": 1.8864802122116089,
      "learning_rate": 2.9137228577548702e-06,
      "loss": 0.5082,
      "step": 220220
    },
    {
      "epoch": 90.15,
      "grad_norm": 2.3417739868164062,
      "learning_rate": 2.91356898910726e-06,
      "loss": 0.4812,
      "step": 220230
    },
    {
      "epoch": 90.15,
      "grad_norm": 2.3071842193603516,
      "learning_rate": 2.9134151178493975e-06,
      "loss": 0.5185,
      "step": 220240
    },
    {
      "epoch": 90.16,
      "grad_norm": 1.8297905921936035,
      "learning_rate": 2.913261243981987e-06,
      "loss": 0.4822,
      "step": 220250
    },
    {
      "epoch": 90.16,
      "grad_norm": 2.244288921356201,
      "learning_rate": 2.913107367505733e-06,
      "loss": 0.5083,
      "step": 220260
    },
    {
      "epoch": 90.16,
      "grad_norm": 1.8451098203659058,
      "learning_rate": 2.9129534884213423e-06,
      "loss": 0.485,
      "step": 220270
    },
    {
      "epoch": 90.17,
      "grad_norm": 1.7726954221725464,
      "learning_rate": 2.912799606729518e-06,
      "loss": 0.4864,
      "step": 220280
    },
    {
      "epoch": 90.17,
      "grad_norm": 1.7717273235321045,
      "learning_rate": 2.912645722430965e-06,
      "loss": 0.5012,
      "step": 220290
    },
    {
      "epoch": 90.18,
      "grad_norm": 2.257798910140991,
      "learning_rate": 2.9124918355263897e-06,
      "loss": 0.4837,
      "step": 220300
    },
    {
      "epoch": 90.18,
      "grad_norm": 1.6754212379455566,
      "learning_rate": 2.912337946016496e-06,
      "loss": 0.5153,
      "step": 220310
    },
    {
      "epoch": 90.18,
      "grad_norm": 1.6000072956085205,
      "learning_rate": 2.912184053901989e-06,
      "loss": 0.4819,
      "step": 220320
    },
    {
      "epoch": 90.19,
      "grad_norm": 1.7474361658096313,
      "learning_rate": 2.9120301591835737e-06,
      "loss": 0.5,
      "step": 220330
    },
    {
      "epoch": 90.19,
      "grad_norm": 1.7435643672943115,
      "learning_rate": 2.9118762618619545e-06,
      "loss": 0.4827,
      "step": 220340
    },
    {
      "epoch": 90.2,
      "grad_norm": 1.9550000429153442,
      "learning_rate": 2.911722361937837e-06,
      "loss": 0.5231,
      "step": 220350
    },
    {
      "epoch": 90.2,
      "grad_norm": 1.8318886756896973,
      "learning_rate": 2.911568459411926e-06,
      "loss": 0.474,
      "step": 220360
    },
    {
      "epoch": 90.2,
      "grad_norm": 1.8052829504013062,
      "learning_rate": 2.911414554284927e-06,
      "loss": 0.4783,
      "step": 220370
    },
    {
      "epoch": 90.21,
      "grad_norm": 1.486293911933899,
      "learning_rate": 2.911260646557544e-06,
      "loss": 0.4919,
      "step": 220380
    },
    {
      "epoch": 90.21,
      "grad_norm": 2.214416027069092,
      "learning_rate": 2.9111067362304834e-06,
      "loss": 0.4904,
      "step": 220390
    },
    {
      "epoch": 90.22,
      "grad_norm": 3.4680263996124268,
      "learning_rate": 2.9109528233044497e-06,
      "loss": 0.5148,
      "step": 220400
    },
    {
      "epoch": 90.22,
      "grad_norm": 1.946290135383606,
      "learning_rate": 2.9107989077801467e-06,
      "loss": 0.4874,
      "step": 220410
    },
    {
      "epoch": 90.23,
      "grad_norm": 2.186960220336914,
      "learning_rate": 2.910644989658281e-06,
      "loss": 0.4867,
      "step": 220420
    },
    {
      "epoch": 90.23,
      "grad_norm": 1.7335667610168457,
      "learning_rate": 2.9104910689395566e-06,
      "loss": 0.4935,
      "step": 220430
    },
    {
      "epoch": 90.23,
      "grad_norm": 2.387897491455078,
      "learning_rate": 2.9103371456246795e-06,
      "loss": 0.4955,
      "step": 220440
    },
    {
      "epoch": 90.24,
      "grad_norm": 1.5880377292633057,
      "learning_rate": 2.9101832197143545e-06,
      "loss": 0.4913,
      "step": 220450
    },
    {
      "epoch": 90.24,
      "grad_norm": 2.2887041568756104,
      "learning_rate": 2.910029291209286e-06,
      "loss": 0.5143,
      "step": 220460
    },
    {
      "epoch": 90.25,
      "grad_norm": 1.465972900390625,
      "learning_rate": 2.9098753601101806e-06,
      "loss": 0.5029,
      "step": 220470
    },
    {
      "epoch": 90.25,
      "grad_norm": 1.824750304222107,
      "learning_rate": 2.9097214264177425e-06,
      "loss": 0.4808,
      "step": 220480
    },
    {
      "epoch": 90.25,
      "grad_norm": 2.0047078132629395,
      "learning_rate": 2.9095674901326768e-06,
      "loss": 0.5167,
      "step": 220490
    },
    {
      "epoch": 90.26,
      "grad_norm": 2.449777364730835,
      "learning_rate": 2.9094135512556894e-06,
      "loss": 0.4891,
      "step": 220500
    },
    {
      "epoch": 90.26,
      "grad_norm": 1.6793417930603027,
      "learning_rate": 2.9092596097874837e-06,
      "loss": 0.4731,
      "step": 220510
    },
    {
      "epoch": 90.27,
      "grad_norm": 1.6866154670715332,
      "learning_rate": 2.9091056657287667e-06,
      "loss": 0.4775,
      "step": 220520
    },
    {
      "epoch": 90.27,
      "grad_norm": 1.8241838216781616,
      "learning_rate": 2.908951719080243e-06,
      "loss": 0.4973,
      "step": 220530
    },
    {
      "epoch": 90.27,
      "grad_norm": 1.8130935430526733,
      "learning_rate": 2.9087977698426176e-06,
      "loss": 0.5084,
      "step": 220540
    },
    {
      "epoch": 90.28,
      "grad_norm": 2.2269692420959473,
      "learning_rate": 2.9086438180165962e-06,
      "loss": 0.4854,
      "step": 220550
    },
    {
      "epoch": 90.28,
      "grad_norm": 2.050975799560547,
      "learning_rate": 2.908489863602883e-06,
      "loss": 0.4919,
      "step": 220560
    },
    {
      "epoch": 90.29,
      "grad_norm": 2.034485101699829,
      "learning_rate": 2.9083359066021843e-06,
      "loss": 0.4897,
      "step": 220570
    },
    {
      "epoch": 90.29,
      "grad_norm": 1.729042410850525,
      "learning_rate": 2.9081819470152057e-06,
      "loss": 0.4834,
      "step": 220580
    },
    {
      "epoch": 90.29,
      "grad_norm": 1.8789256811141968,
      "learning_rate": 2.9080279848426506e-06,
      "loss": 0.4928,
      "step": 220590
    },
    {
      "epoch": 90.3,
      "grad_norm": 1.8536145687103271,
      "learning_rate": 2.9078740200852263e-06,
      "loss": 0.4881,
      "step": 220600
    },
    {
      "epoch": 90.3,
      "grad_norm": 1.8833421468734741,
      "learning_rate": 2.9077200527436365e-06,
      "loss": 0.4965,
      "step": 220610
    },
    {
      "epoch": 90.31,
      "grad_norm": 1.7666988372802734,
      "learning_rate": 2.9075660828185882e-06,
      "loss": 0.4938,
      "step": 220620
    },
    {
      "epoch": 90.31,
      "grad_norm": 2.1910629272460938,
      "learning_rate": 2.9074121103107853e-06,
      "loss": 0.4858,
      "step": 220630
    },
    {
      "epoch": 90.32,
      "grad_norm": 1.8721258640289307,
      "learning_rate": 2.907258135220933e-06,
      "loss": 0.4789,
      "step": 220640
    },
    {
      "epoch": 90.32,
      "grad_norm": 13.702561378479004,
      "learning_rate": 2.9071041575497377e-06,
      "loss": 0.4774,
      "step": 220650
    },
    {
      "epoch": 90.32,
      "grad_norm": 1.6806389093399048,
      "learning_rate": 2.9069501772979043e-06,
      "loss": 0.4903,
      "step": 220660
    },
    {
      "epoch": 90.33,
      "grad_norm": 1.682518720626831,
      "learning_rate": 2.9067961944661377e-06,
      "loss": 0.4659,
      "step": 220670
    },
    {
      "epoch": 90.33,
      "grad_norm": 1.462205171585083,
      "learning_rate": 2.906642209055144e-06,
      "loss": 0.4663,
      "step": 220680
    },
    {
      "epoch": 90.34,
      "grad_norm": 2.3891875743865967,
      "learning_rate": 2.9064882210656285e-06,
      "loss": 0.4827,
      "step": 220690
    },
    {
      "epoch": 90.34,
      "grad_norm": 1.669020414352417,
      "learning_rate": 2.9063342304982962e-06,
      "loss": 0.5022,
      "step": 220700
    },
    {
      "epoch": 90.34,
      "grad_norm": 1.7661207914352417,
      "learning_rate": 2.9061802373538527e-06,
      "loss": 0.4985,
      "step": 220710
    },
    {
      "epoch": 90.35,
      "grad_norm": 1.8513970375061035,
      "learning_rate": 2.9060262416330034e-06,
      "loss": 0.4954,
      "step": 220720
    },
    {
      "epoch": 90.35,
      "grad_norm": 2.086945056915283,
      "learning_rate": 2.905872243336453e-06,
      "loss": 0.4932,
      "step": 220730
    },
    {
      "epoch": 90.36,
      "grad_norm": 1.8012741804122925,
      "learning_rate": 2.9057182424649083e-06,
      "loss": 0.4867,
      "step": 220740
    },
    {
      "epoch": 90.36,
      "grad_norm": 2.354053020477295,
      "learning_rate": 2.905564239019074e-06,
      "loss": 0.4813,
      "step": 220750
    },
    {
      "epoch": 90.36,
      "grad_norm": 1.9001855850219727,
      "learning_rate": 2.905410232999656e-06,
      "loss": 0.5077,
      "step": 220760
    },
    {
      "epoch": 90.37,
      "grad_norm": 1.970440149307251,
      "learning_rate": 2.9052562244073593e-06,
      "loss": 0.4903,
      "step": 220770
    },
    {
      "epoch": 90.37,
      "grad_norm": 2.0649938583374023,
      "learning_rate": 2.9051022132428895e-06,
      "loss": 0.4791,
      "step": 220780
    },
    {
      "epoch": 90.38,
      "grad_norm": 1.6884444952011108,
      "learning_rate": 2.904948199506952e-06,
      "loss": 0.5164,
      "step": 220790
    },
    {
      "epoch": 90.38,
      "grad_norm": 1.5693663358688354,
      "learning_rate": 2.9047941832002524e-06,
      "loss": 0.5078,
      "step": 220800
    },
    {
      "epoch": 90.38,
      "grad_norm": 2.1782095432281494,
      "learning_rate": 2.9046401643234964e-06,
      "loss": 0.4873,
      "step": 220810
    },
    {
      "epoch": 90.39,
      "grad_norm": 1.9966747760772705,
      "learning_rate": 2.9044861428773896e-06,
      "loss": 0.5013,
      "step": 220820
    },
    {
      "epoch": 90.39,
      "grad_norm": 1.5528005361557007,
      "learning_rate": 2.904332118862637e-06,
      "loss": 0.4827,
      "step": 220830
    },
    {
      "epoch": 90.4,
      "grad_norm": 2.147840976715088,
      "learning_rate": 2.904178092279945e-06,
      "loss": 0.5021,
      "step": 220840
    },
    {
      "epoch": 90.4,
      "grad_norm": 1.8215134143829346,
      "learning_rate": 2.9040240631300185e-06,
      "loss": 0.5184,
      "step": 220850
    },
    {
      "epoch": 90.41,
      "grad_norm": 1.9122675657272339,
      "learning_rate": 2.903870031413563e-06,
      "loss": 0.5178,
      "step": 220860
    },
    {
      "epoch": 90.41,
      "grad_norm": 1.7794727087020874,
      "learning_rate": 2.903715997131285e-06,
      "loss": 0.5028,
      "step": 220870
    },
    {
      "epoch": 90.41,
      "grad_norm": 2.035604953765869,
      "learning_rate": 2.903561960283889e-06,
      "loss": 0.5145,
      "step": 220880
    },
    {
      "epoch": 90.42,
      "grad_norm": 1.814833402633667,
      "learning_rate": 2.9034079208720814e-06,
      "loss": 0.5083,
      "step": 220890
    },
    {
      "epoch": 90.42,
      "grad_norm": 1.7983036041259766,
      "learning_rate": 2.903253878896567e-06,
      "loss": 0.4911,
      "step": 220900
    },
    {
      "epoch": 90.43,
      "grad_norm": 2.0382907390594482,
      "learning_rate": 2.9030998343580533e-06,
      "loss": 0.5004,
      "step": 220910
    },
    {
      "epoch": 90.43,
      "grad_norm": 2.4004621505737305,
      "learning_rate": 2.902945787257243e-06,
      "loss": 0.4902,
      "step": 220920
    },
    {
      "epoch": 90.43,
      "grad_norm": 1.4581564664840698,
      "learning_rate": 2.902791737594844e-06,
      "loss": 0.4921,
      "step": 220930
    },
    {
      "epoch": 90.44,
      "grad_norm": 2.186436653137207,
      "learning_rate": 2.9026376853715613e-06,
      "loss": 0.484,
      "step": 220940
    },
    {
      "epoch": 90.44,
      "grad_norm": 1.7385430335998535,
      "learning_rate": 2.9024836305881008e-06,
      "loss": 0.4904,
      "step": 220950
    },
    {
      "epoch": 90.45,
      "grad_norm": 1.8359800577163696,
      "learning_rate": 2.902329573245168e-06,
      "loss": 0.4852,
      "step": 220960
    },
    {
      "epoch": 90.45,
      "grad_norm": 1.9307520389556885,
      "learning_rate": 2.902175513343469e-06,
      "loss": 0.4915,
      "step": 220970
    },
    {
      "epoch": 90.45,
      "grad_norm": 2.287437915802002,
      "learning_rate": 2.9020214508837095e-06,
      "loss": 0.4833,
      "step": 220980
    },
    {
      "epoch": 90.46,
      "grad_norm": 2.416463613510132,
      "learning_rate": 2.901867385866594e-06,
      "loss": 0.4585,
      "step": 220990
    },
    {
      "epoch": 90.46,
      "grad_norm": 1.9281662702560425,
      "learning_rate": 2.90171331829283e-06,
      "loss": 0.5099,
      "step": 221000
    },
    {
      "epoch": 90.47,
      "grad_norm": 1.909458875656128,
      "learning_rate": 2.9015592481631215e-06,
      "loss": 0.4865,
      "step": 221010
    },
    {
      "epoch": 90.47,
      "grad_norm": 1.6596546173095703,
      "learning_rate": 2.9014051754781756e-06,
      "loss": 0.4766,
      "step": 221020
    },
    {
      "epoch": 90.47,
      "grad_norm": 1.7252142429351807,
      "learning_rate": 2.901251100238698e-06,
      "loss": 0.5006,
      "step": 221030
    },
    {
      "epoch": 90.48,
      "grad_norm": 1.5814467668533325,
      "learning_rate": 2.901097022445394e-06,
      "loss": 0.4859,
      "step": 221040
    },
    {
      "epoch": 90.48,
      "grad_norm": 1.7361609935760498,
      "learning_rate": 2.90094294209897e-06,
      "loss": 0.4808,
      "step": 221050
    },
    {
      "epoch": 90.49,
      "grad_norm": 1.8765950202941895,
      "learning_rate": 2.900788859200131e-06,
      "loss": 0.484,
      "step": 221060
    },
    {
      "epoch": 90.49,
      "grad_norm": 1.7117083072662354,
      "learning_rate": 2.9006347737495836e-06,
      "loss": 0.4739,
      "step": 221070
    },
    {
      "epoch": 90.5,
      "grad_norm": 2.0753417015075684,
      "learning_rate": 2.900480685748033e-06,
      "loss": 0.5065,
      "step": 221080
    },
    {
      "epoch": 90.5,
      "grad_norm": 2.2155396938323975,
      "learning_rate": 2.9003265951961856e-06,
      "loss": 0.4786,
      "step": 221090
    },
    {
      "epoch": 90.5,
      "grad_norm": 1.8402578830718994,
      "learning_rate": 2.9001725020947468e-06,
      "loss": 0.5085,
      "step": 221100
    },
    {
      "epoch": 90.51,
      "grad_norm": 1.720808982849121,
      "learning_rate": 2.9000184064444223e-06,
      "loss": 0.4817,
      "step": 221110
    },
    {
      "epoch": 90.51,
      "grad_norm": 1.822629690170288,
      "learning_rate": 2.899864308245919e-06,
      "loss": 0.5032,
      "step": 221120
    },
    {
      "epoch": 90.52,
      "grad_norm": 1.8629871606826782,
      "learning_rate": 2.8997102074999424e-06,
      "loss": 0.4987,
      "step": 221130
    },
    {
      "epoch": 90.52,
      "grad_norm": 1.9454858303070068,
      "learning_rate": 2.8995561042071976e-06,
      "loss": 0.4755,
      "step": 221140
    },
    {
      "epoch": 90.52,
      "grad_norm": 1.7112332582473755,
      "learning_rate": 2.8994019983683915e-06,
      "loss": 0.4629,
      "step": 221150
    },
    {
      "epoch": 90.53,
      "grad_norm": 1.8952339887619019,
      "learning_rate": 2.899247889984229e-06,
      "loss": 0.5184,
      "step": 221160
    },
    {
      "epoch": 90.53,
      "grad_norm": 1.6568613052368164,
      "learning_rate": 2.899093779055417e-06,
      "loss": 0.4642,
      "step": 221170
    },
    {
      "epoch": 90.54,
      "grad_norm": 2.203857898712158,
      "learning_rate": 2.8989396655826614e-06,
      "loss": 0.4851,
      "step": 221180
    },
    {
      "epoch": 90.54,
      "grad_norm": 2.434440851211548,
      "learning_rate": 2.898785549566668e-06,
      "loss": 0.4981,
      "step": 221190
    },
    {
      "epoch": 90.54,
      "grad_norm": 2.479229688644409,
      "learning_rate": 2.8986314310081424e-06,
      "loss": 0.4996,
      "step": 221200
    },
    {
      "epoch": 90.55,
      "grad_norm": 2.275491952896118,
      "learning_rate": 2.898477309907791e-06,
      "loss": 0.4804,
      "step": 221210
    },
    {
      "epoch": 90.55,
      "grad_norm": 2.0167181491851807,
      "learning_rate": 2.89832318626632e-06,
      "loss": 0.5039,
      "step": 221220
    },
    {
      "epoch": 90.56,
      "grad_norm": 2.580178737640381,
      "learning_rate": 2.898169060084435e-06,
      "loss": 0.494,
      "step": 221230
    },
    {
      "epoch": 90.56,
      "grad_norm": 1.710132122039795,
      "learning_rate": 2.898014931362842e-06,
      "loss": 0.4908,
      "step": 221240
    },
    {
      "epoch": 90.56,
      "grad_norm": 1.8578976392745972,
      "learning_rate": 2.8978608001022465e-06,
      "loss": 0.4939,
      "step": 221250
    },
    {
      "epoch": 90.57,
      "grad_norm": 1.8245501518249512,
      "learning_rate": 2.8977066663033564e-06,
      "loss": 0.4871,
      "step": 221260
    },
    {
      "epoch": 90.57,
      "grad_norm": 1.7029370069503784,
      "learning_rate": 2.8975525299668763e-06,
      "loss": 0.4787,
      "step": 221270
    },
    {
      "epoch": 90.58,
      "grad_norm": 1.6264981031417847,
      "learning_rate": 2.8973983910935126e-06,
      "loss": 0.4933,
      "step": 221280
    },
    {
      "epoch": 90.58,
      "grad_norm": 1.5789391994476318,
      "learning_rate": 2.897244249683971e-06,
      "loss": 0.4863,
      "step": 221290
    },
    {
      "epoch": 90.59,
      "grad_norm": 1.7918709516525269,
      "learning_rate": 2.8970901057389577e-06,
      "loss": 0.4565,
      "step": 221300
    },
    {
      "epoch": 90.59,
      "grad_norm": 1.960421085357666,
      "learning_rate": 2.8969359592591792e-06,
      "loss": 0.4837,
      "step": 221310
    },
    {
      "epoch": 90.59,
      "grad_norm": 1.4383516311645508,
      "learning_rate": 2.896781810245342e-06,
      "loss": 0.4815,
      "step": 221320
    },
    {
      "epoch": 90.6,
      "grad_norm": 1.9228620529174805,
      "learning_rate": 2.8966276586981513e-06,
      "loss": 0.4905,
      "step": 221330
    },
    {
      "epoch": 90.6,
      "grad_norm": 2.317676305770874,
      "learning_rate": 2.8964735046183143e-06,
      "loss": 0.4778,
      "step": 221340
    },
    {
      "epoch": 90.61,
      "grad_norm": 2.537705898284912,
      "learning_rate": 2.8963193480065366e-06,
      "loss": 0.4921,
      "step": 221350
    },
    {
      "epoch": 90.61,
      "grad_norm": 1.4449437856674194,
      "learning_rate": 2.8961651888635237e-06,
      "loss": 0.4942,
      "step": 221360
    },
    {
      "epoch": 90.61,
      "grad_norm": 1.9498610496520996,
      "learning_rate": 2.896011027189982e-06,
      "loss": 0.5055,
      "step": 221370
    },
    {
      "epoch": 90.62,
      "grad_norm": 1.906671166419983,
      "learning_rate": 2.8958568629866186e-06,
      "loss": 0.501,
      "step": 221380
    },
    {
      "epoch": 90.62,
      "grad_norm": 2.0610015392303467,
      "learning_rate": 2.895702696254139e-06,
      "loss": 0.4722,
      "step": 221390
    },
    {
      "epoch": 90.63,
      "grad_norm": 1.8315956592559814,
      "learning_rate": 2.895548526993249e-06,
      "loss": 0.5073,
      "step": 221400
    },
    {
      "epoch": 90.63,
      "grad_norm": 2.522331953048706,
      "learning_rate": 2.895394355204657e-06,
      "loss": 0.4818,
      "step": 221410
    },
    {
      "epoch": 90.63,
      "grad_norm": 1.929532766342163,
      "learning_rate": 2.895240180889067e-06,
      "loss": 0.5065,
      "step": 221420
    },
    {
      "epoch": 90.64,
      "grad_norm": 1.7451189756393433,
      "learning_rate": 2.8950860040471853e-06,
      "loss": 0.4865,
      "step": 221430
    },
    {
      "epoch": 90.64,
      "grad_norm": 1.9144256114959717,
      "learning_rate": 2.8949318246797192e-06,
      "loss": 0.4963,
      "step": 221440
    },
    {
      "epoch": 90.65,
      "grad_norm": 2.033208131790161,
      "learning_rate": 2.894777642787374e-06,
      "loss": 0.4945,
      "step": 221450
    },
    {
      "epoch": 90.65,
      "grad_norm": 2.158430576324463,
      "learning_rate": 2.894623458370857e-06,
      "loss": 0.4781,
      "step": 221460
    },
    {
      "epoch": 90.65,
      "grad_norm": 1.9199391603469849,
      "learning_rate": 2.8944692714308742e-06,
      "loss": 0.5025,
      "step": 221470
    },
    {
      "epoch": 90.66,
      "grad_norm": 2.006767749786377,
      "learning_rate": 2.894315081968131e-06,
      "loss": 0.4846,
      "step": 221480
    },
    {
      "epoch": 90.66,
      "grad_norm": 2.2410788536071777,
      "learning_rate": 2.894160889983335e-06,
      "loss": 0.4829,
      "step": 221490
    },
    {
      "epoch": 90.67,
      "grad_norm": 1.8114700317382812,
      "learning_rate": 2.8940066954771917e-06,
      "loss": 0.4787,
      "step": 221500
    },
    {
      "epoch": 90.67,
      "grad_norm": 1.4461232423782349,
      "learning_rate": 2.893852498450408e-06,
      "loss": 0.4984,
      "step": 221510
    },
    {
      "epoch": 90.68,
      "grad_norm": 1.6481032371520996,
      "learning_rate": 2.8936982989036896e-06,
      "loss": 0.509,
      "step": 221520
    },
    {
      "epoch": 90.68,
      "grad_norm": 1.6047275066375732,
      "learning_rate": 2.8935440968377433e-06,
      "loss": 0.5018,
      "step": 221530
    },
    {
      "epoch": 90.68,
      "grad_norm": 2.1121864318847656,
      "learning_rate": 2.8933898922532754e-06,
      "loss": 0.4964,
      "step": 221540
    },
    {
      "epoch": 90.69,
      "grad_norm": 1.5940676927566528,
      "learning_rate": 2.8932356851509924e-06,
      "loss": 0.51,
      "step": 221550
    },
    {
      "epoch": 90.69,
      "grad_norm": 2.741225481033325,
      "learning_rate": 2.893081475531601e-06,
      "loss": 0.4805,
      "step": 221560
    },
    {
      "epoch": 90.7,
      "grad_norm": 2.2476320266723633,
      "learning_rate": 2.8929272633958064e-06,
      "loss": 0.4941,
      "step": 221570
    },
    {
      "epoch": 90.7,
      "grad_norm": 2.534952163696289,
      "learning_rate": 2.8927730487443155e-06,
      "loss": 0.4668,
      "step": 221580
    },
    {
      "epoch": 90.7,
      "grad_norm": 2.1443631649017334,
      "learning_rate": 2.8926188315778356e-06,
      "loss": 0.4661,
      "step": 221590
    },
    {
      "epoch": 90.71,
      "grad_norm": 1.3871060609817505,
      "learning_rate": 2.8924646118970726e-06,
      "loss": 0.48,
      "step": 221600
    },
    {
      "epoch": 90.71,
      "grad_norm": 1.9277409315109253,
      "learning_rate": 2.892310389702733e-06,
      "loss": 0.494,
      "step": 221610
    },
    {
      "epoch": 90.72,
      "grad_norm": 1.8587466478347778,
      "learning_rate": 2.892156164995523e-06,
      "loss": 0.4943,
      "step": 221620
    },
    {
      "epoch": 90.72,
      "grad_norm": 2.0194573402404785,
      "learning_rate": 2.89200193777615e-06,
      "loss": 0.5058,
      "step": 221630
    },
    {
      "epoch": 90.72,
      "grad_norm": 2.198646068572998,
      "learning_rate": 2.891847708045319e-06,
      "loss": 0.4779,
      "step": 221640
    },
    {
      "epoch": 90.73,
      "grad_norm": 2.1429076194763184,
      "learning_rate": 2.891693475803738e-06,
      "loss": 0.4958,
      "step": 221650
    },
    {
      "epoch": 90.73,
      "grad_norm": 2.0244557857513428,
      "learning_rate": 2.891539241052112e-06,
      "loss": 0.4941,
      "step": 221660
    },
    {
      "epoch": 90.74,
      "grad_norm": 1.9069797992706299,
      "learning_rate": 2.891385003791148e-06,
      "loss": 0.5031,
      "step": 221670
    },
    {
      "epoch": 90.74,
      "grad_norm": 1.9989745616912842,
      "learning_rate": 2.8912307640215538e-06,
      "loss": 0.4645,
      "step": 221680
    },
    {
      "epoch": 90.74,
      "grad_norm": 1.606364130973816,
      "learning_rate": 2.8910765217440346e-06,
      "loss": 0.4979,
      "step": 221690
    },
    {
      "epoch": 90.75,
      "grad_norm": 1.8772965669631958,
      "learning_rate": 2.890922276959298e-06,
      "loss": 0.4683,
      "step": 221700
    },
    {
      "epoch": 90.75,
      "grad_norm": 5.868133068084717,
      "learning_rate": 2.8907680296680493e-06,
      "loss": 0.484,
      "step": 221710
    },
    {
      "epoch": 90.76,
      "grad_norm": 1.7592240571975708,
      "learning_rate": 2.890613779870996e-06,
      "loss": 0.4981,
      "step": 221720
    },
    {
      "epoch": 90.76,
      "grad_norm": 2.0640320777893066,
      "learning_rate": 2.890459527568844e-06,
      "loss": 0.4779,
      "step": 221730
    },
    {
      "epoch": 90.77,
      "grad_norm": 1.8505600690841675,
      "learning_rate": 2.8903052727623006e-06,
      "loss": 0.4741,
      "step": 221740
    },
    {
      "epoch": 90.77,
      "grad_norm": 1.6892974376678467,
      "learning_rate": 2.890151015452072e-06,
      "loss": 0.4786,
      "step": 221750
    },
    {
      "epoch": 90.77,
      "grad_norm": 1.978912115097046,
      "learning_rate": 2.889996755638865e-06,
      "loss": 0.4909,
      "step": 221760
    },
    {
      "epoch": 90.78,
      "grad_norm": 1.734512448310852,
      "learning_rate": 2.8898424933233863e-06,
      "loss": 0.4869,
      "step": 221770
    },
    {
      "epoch": 90.78,
      "grad_norm": 1.6386414766311646,
      "learning_rate": 2.889688228506343e-06,
      "loss": 0.4872,
      "step": 221780
    },
    {
      "epoch": 90.79,
      "grad_norm": 1.9793280363082886,
      "learning_rate": 2.8895339611884406e-06,
      "loss": 0.5061,
      "step": 221790
    },
    {
      "epoch": 90.79,
      "grad_norm": 1.6591594219207764,
      "learning_rate": 2.889379691370386e-06,
      "loss": 0.5011,
      "step": 221800
    },
    {
      "epoch": 90.79,
      "grad_norm": 1.7300328016281128,
      "learning_rate": 2.8892254190528873e-06,
      "loss": 0.5105,
      "step": 221810
    },
    {
      "epoch": 90.8,
      "grad_norm": 1.677559733390808,
      "learning_rate": 2.889071144236649e-06,
      "loss": 0.5103,
      "step": 221820
    },
    {
      "epoch": 90.8,
      "grad_norm": 1.7752493619918823,
      "learning_rate": 2.88891686692238e-06,
      "loss": 0.4825,
      "step": 221830
    },
    {
      "epoch": 90.81,
      "grad_norm": 1.6573808193206787,
      "learning_rate": 2.8887625871107857e-06,
      "loss": 0.4789,
      "step": 221840
    },
    {
      "epoch": 90.81,
      "grad_norm": 1.4861769676208496,
      "learning_rate": 2.888608304802573e-06,
      "loss": 0.5063,
      "step": 221850
    },
    {
      "epoch": 90.81,
      "grad_norm": 2.0813560485839844,
      "learning_rate": 2.8884540199984487e-06,
      "loss": 0.4739,
      "step": 221860
    },
    {
      "epoch": 90.82,
      "grad_norm": 2.1158719062805176,
      "learning_rate": 2.8882997326991196e-06,
      "loss": 0.5014,
      "step": 221870
    },
    {
      "epoch": 90.82,
      "grad_norm": 2.074705123901367,
      "learning_rate": 2.8881454429052925e-06,
      "loss": 0.4918,
      "step": 221880
    },
    {
      "epoch": 90.83,
      "grad_norm": 1.9970126152038574,
      "learning_rate": 2.8879911506176746e-06,
      "loss": 0.4908,
      "step": 221890
    },
    {
      "epoch": 90.83,
      "grad_norm": 1.742943525314331,
      "learning_rate": 2.887836855836972e-06,
      "loss": 0.4924,
      "step": 221900
    },
    {
      "epoch": 90.84,
      "grad_norm": 2.0531482696533203,
      "learning_rate": 2.8876825585638922e-06,
      "loss": 0.4984,
      "step": 221910
    },
    {
      "epoch": 90.84,
      "grad_norm": 1.7813198566436768,
      "learning_rate": 2.8875282587991414e-06,
      "loss": 0.4764,
      "step": 221920
    },
    {
      "epoch": 90.84,
      "grad_norm": 1.8275188207626343,
      "learning_rate": 2.8873739565434267e-06,
      "loss": 0.5078,
      "step": 221930
    },
    {
      "epoch": 90.85,
      "grad_norm": 1.7879666090011597,
      "learning_rate": 2.8872196517974543e-06,
      "loss": 0.5018,
      "step": 221940
    },
    {
      "epoch": 90.85,
      "grad_norm": 1.7301576137542725,
      "learning_rate": 2.8870653445619317e-06,
      "loss": 0.4878,
      "step": 221950
    },
    {
      "epoch": 90.86,
      "grad_norm": 1.6445831060409546,
      "learning_rate": 2.886911034837566e-06,
      "loss": 0.4812,
      "step": 221960
    },
    {
      "epoch": 90.86,
      "grad_norm": 1.837497591972351,
      "learning_rate": 2.8867567226250633e-06,
      "loss": 0.4772,
      "step": 221970
    },
    {
      "epoch": 90.86,
      "grad_norm": 2.0485167503356934,
      "learning_rate": 2.886602407925131e-06,
      "loss": 0.4874,
      "step": 221980
    },
    {
      "epoch": 90.87,
      "grad_norm": 1.9292980432510376,
      "learning_rate": 2.8864480907384765e-06,
      "loss": 0.5074,
      "step": 221990
    },
    {
      "epoch": 90.87,
      "grad_norm": 1.8150668144226074,
      "learning_rate": 2.8862937710658057e-06,
      "loss": 0.4834,
      "step": 222000
    },
    {
      "epoch": 90.88,
      "grad_norm": 2.1768248081207275,
      "learning_rate": 2.886139448907826e-06,
      "loss": 0.4839,
      "step": 222010
    },
    {
      "epoch": 90.88,
      "grad_norm": 2.454740524291992,
      "learning_rate": 2.885985124265244e-06,
      "loss": 0.4956,
      "step": 222020
    },
    {
      "epoch": 90.88,
      "grad_norm": 2.029442548751831,
      "learning_rate": 2.8858307971387666e-06,
      "loss": 0.4964,
      "step": 222030
    },
    {
      "epoch": 90.89,
      "grad_norm": 1.7510032653808594,
      "learning_rate": 2.8856764675291017e-06,
      "loss": 0.471,
      "step": 222040
    },
    {
      "epoch": 90.89,
      "grad_norm": 1.9544271230697632,
      "learning_rate": 2.8855221354369547e-06,
      "loss": 0.4859,
      "step": 222050
    },
    {
      "epoch": 90.9,
      "grad_norm": 1.6040393114089966,
      "learning_rate": 2.8853678008630347e-06,
      "loss": 0.4929,
      "step": 222060
    },
    {
      "epoch": 90.9,
      "grad_norm": 2.6898841857910156,
      "learning_rate": 2.8852134638080463e-06,
      "loss": 0.4698,
      "step": 222070
    },
    {
      "epoch": 90.9,
      "grad_norm": 1.6482375860214233,
      "learning_rate": 2.885059124272698e-06,
      "loss": 0.4939,
      "step": 222080
    },
    {
      "epoch": 90.91,
      "grad_norm": 1.4680448770523071,
      "learning_rate": 2.8849047822576965e-06,
      "loss": 0.4722,
      "step": 222090
    },
    {
      "epoch": 90.91,
      "grad_norm": 1.729533314704895,
      "learning_rate": 2.8847504377637493e-06,
      "loss": 0.4838,
      "step": 222100
    },
    {
      "epoch": 90.92,
      "grad_norm": 1.6888474225997925,
      "learning_rate": 2.884596090791562e-06,
      "loss": 0.477,
      "step": 222110
    },
    {
      "epoch": 90.92,
      "grad_norm": 2.1778929233551025,
      "learning_rate": 2.884441741341843e-06,
      "loss": 0.4868,
      "step": 222120
    },
    {
      "epoch": 90.93,
      "grad_norm": 1.4794085025787354,
      "learning_rate": 2.8842873894152985e-06,
      "loss": 0.4773,
      "step": 222130
    },
    {
      "epoch": 90.93,
      "grad_norm": 1.5796934366226196,
      "learning_rate": 2.8841330350126366e-06,
      "loss": 0.4866,
      "step": 222140
    },
    {
      "epoch": 90.93,
      "grad_norm": 1.7611072063446045,
      "learning_rate": 2.883978678134563e-06,
      "loss": 0.4858,
      "step": 222150
    },
    {
      "epoch": 90.94,
      "grad_norm": 2.515528440475464,
      "learning_rate": 2.8838243187817857e-06,
      "loss": 0.4915,
      "step": 222160
    },
    {
      "epoch": 90.94,
      "grad_norm": 2.3584353923797607,
      "learning_rate": 2.8836699569550117e-06,
      "loss": 0.474,
      "step": 222170
    },
    {
      "epoch": 90.95,
      "grad_norm": 1.911502718925476,
      "learning_rate": 2.883515592654948e-06,
      "loss": 0.5052,
      "step": 222180
    },
    {
      "epoch": 90.95,
      "grad_norm": 1.8834691047668457,
      "learning_rate": 2.8833612258823013e-06,
      "loss": 0.5049,
      "step": 222190
    },
    {
      "epoch": 90.95,
      "grad_norm": 1.7959240674972534,
      "learning_rate": 2.88320685663778e-06,
      "loss": 0.499,
      "step": 222200
    },
    {
      "epoch": 90.96,
      "grad_norm": 1.7138396501541138,
      "learning_rate": 2.8830524849220902e-06,
      "loss": 0.5037,
      "step": 222210
    },
    {
      "epoch": 90.96,
      "grad_norm": 3.810673236846924,
      "learning_rate": 2.8828981107359387e-06,
      "loss": 0.4885,
      "step": 222220
    },
    {
      "epoch": 90.97,
      "grad_norm": 2.0911941528320312,
      "learning_rate": 2.882743734080033e-06,
      "loss": 0.499,
      "step": 222230
    },
    {
      "epoch": 90.97,
      "grad_norm": 1.8900704383850098,
      "learning_rate": 2.8825893549550808e-06,
      "loss": 0.4817,
      "step": 222240
    },
    {
      "epoch": 90.97,
      "grad_norm": 1.8325293064117432,
      "learning_rate": 2.882434973361789e-06,
      "loss": 0.4921,
      "step": 222250
    },
    {
      "epoch": 90.98,
      "grad_norm": 1.9924042224884033,
      "learning_rate": 2.882280589300865e-06,
      "loss": 0.4886,
      "step": 222260
    },
    {
      "epoch": 90.98,
      "grad_norm": 1.4783244132995605,
      "learning_rate": 2.8821262027730153e-06,
      "loss": 0.4922,
      "step": 222270
    },
    {
      "epoch": 90.99,
      "grad_norm": 1.7651625871658325,
      "learning_rate": 2.8819718137789487e-06,
      "loss": 0.4973,
      "step": 222280
    },
    {
      "epoch": 90.99,
      "grad_norm": 2.1060385704040527,
      "learning_rate": 2.8818174223193704e-06,
      "loss": 0.4884,
      "step": 222290
    },
    {
      "epoch": 90.99,
      "grad_norm": 1.6031254529953003,
      "learning_rate": 2.8816630283949884e-06,
      "loss": 0.5085,
      "step": 222300
    },
    {
      "epoch": 91.0,
      "grad_norm": 1.9987061023712158,
      "learning_rate": 2.8815086320065105e-06,
      "loss": 0.4656,
      "step": 222310
    },
    {
      "epoch": 91.0,
      "eval_loss": 0.4889759123325348,
      "eval_runtime": 60.9402,
      "eval_samples_per_second": 56.596,
      "eval_steps_per_second": 7.089,
      "step": 222313
    },
    {
      "epoch": 91.0,
      "grad_norm": 1.8473801612854004,
      "learning_rate": 2.881354233154643e-06,
      "loss": 0.5171,
      "step": 222320
    },
    {
      "epoch": 91.01,
      "grad_norm": 1.8387387990951538,
      "learning_rate": 2.8811998318400944e-06,
      "loss": 0.5009,
      "step": 222330
    },
    {
      "epoch": 91.01,
      "grad_norm": 1.772506833076477,
      "learning_rate": 2.881045428063571e-06,
      "loss": 0.4892,
      "step": 222340
    },
    {
      "epoch": 91.02,
      "grad_norm": 1.6539502143859863,
      "learning_rate": 2.880891021825781e-06,
      "loss": 0.5059,
      "step": 222350
    },
    {
      "epoch": 91.02,
      "grad_norm": 1.9545379877090454,
      "learning_rate": 2.880736613127431e-06,
      "loss": 0.4838,
      "step": 222360
    },
    {
      "epoch": 91.02,
      "grad_norm": 2.1521072387695312,
      "learning_rate": 2.8805822019692286e-06,
      "loss": 0.4808,
      "step": 222370
    },
    {
      "epoch": 91.03,
      "grad_norm": 1.8498342037200928,
      "learning_rate": 2.8804277883518808e-06,
      "loss": 0.4896,
      "step": 222380
    },
    {
      "epoch": 91.03,
      "grad_norm": 1.8011530637741089,
      "learning_rate": 2.8802733722760948e-06,
      "loss": 0.4646,
      "step": 222390
    },
    {
      "epoch": 91.04,
      "grad_norm": 7.887609004974365,
      "learning_rate": 2.8801189537425783e-06,
      "loss": 0.486,
      "step": 222400
    },
    {
      "epoch": 91.04,
      "grad_norm": 1.7979615926742554,
      "learning_rate": 2.879964532752039e-06,
      "loss": 0.5012,
      "step": 222410
    },
    {
      "epoch": 91.04,
      "grad_norm": 1.8565388917922974,
      "learning_rate": 2.8798101093051844e-06,
      "loss": 0.4907,
      "step": 222420
    },
    {
      "epoch": 91.05,
      "grad_norm": 6.577983856201172,
      "learning_rate": 2.879655683402721e-06,
      "loss": 0.4866,
      "step": 222430
    },
    {
      "epoch": 91.05,
      "grad_norm": 1.9771661758422852,
      "learning_rate": 2.8795012550453566e-06,
      "loss": 0.4819,
      "step": 222440
    },
    {
      "epoch": 91.06,
      "grad_norm": 2.257852554321289,
      "learning_rate": 2.8793468242337986e-06,
      "loss": 0.4931,
      "step": 222450
    },
    {
      "epoch": 91.06,
      "grad_norm": 1.9293320178985596,
      "learning_rate": 2.879192390968755e-06,
      "loss": 0.4802,
      "step": 222460
    },
    {
      "epoch": 91.06,
      "grad_norm": 1.972396969795227,
      "learning_rate": 2.879037955250932e-06,
      "loss": 0.5167,
      "step": 222470
    },
    {
      "epoch": 91.07,
      "grad_norm": 2.0341274738311768,
      "learning_rate": 2.878883517081038e-06,
      "loss": 0.489,
      "step": 222480
    },
    {
      "epoch": 91.07,
      "grad_norm": 2.1846766471862793,
      "learning_rate": 2.878729076459781e-06,
      "loss": 0.4935,
      "step": 222490
    },
    {
      "epoch": 91.08,
      "grad_norm": 1.547470211982727,
      "learning_rate": 2.8785746333878666e-06,
      "loss": 0.4982,
      "step": 222500
    },
    {
      "epoch": 91.08,
      "grad_norm": 1.7510671615600586,
      "learning_rate": 2.8784201878660047e-06,
      "loss": 0.4803,
      "step": 222510
    },
    {
      "epoch": 91.08,
      "grad_norm": 1.6418193578720093,
      "learning_rate": 2.8782657398949e-06,
      "loss": 0.4957,
      "step": 222520
    },
    {
      "epoch": 91.09,
      "grad_norm": 2.0908427238464355,
      "learning_rate": 2.8781112894752624e-06,
      "loss": 0.4787,
      "step": 222530
    },
    {
      "epoch": 91.09,
      "grad_norm": 2.002873182296753,
      "learning_rate": 2.877956836607798e-06,
      "loss": 0.483,
      "step": 222540
    },
    {
      "epoch": 91.1,
      "grad_norm": 2.19826602935791,
      "learning_rate": 2.877802381293215e-06,
      "loss": 0.4985,
      "step": 222550
    },
    {
      "epoch": 91.1,
      "grad_norm": 2.0295138359069824,
      "learning_rate": 2.877647923532221e-06,
      "loss": 0.4964,
      "step": 222560
    },
    {
      "epoch": 91.11,
      "grad_norm": 1.6572388410568237,
      "learning_rate": 2.8774934633255237e-06,
      "loss": 0.4826,
      "step": 222570
    },
    {
      "epoch": 91.11,
      "grad_norm": 2.133129835128784,
      "learning_rate": 2.8773390006738295e-06,
      "loss": 0.4737,
      "step": 222580
    },
    {
      "epoch": 91.11,
      "grad_norm": 1.97670316696167,
      "learning_rate": 2.877184535577847e-06,
      "loss": 0.4774,
      "step": 222590
    },
    {
      "epoch": 91.12,
      "grad_norm": 1.7999850511550903,
      "learning_rate": 2.8770300680382833e-06,
      "loss": 0.5027,
      "step": 222600
    },
    {
      "epoch": 91.12,
      "grad_norm": 1.769741415977478,
      "learning_rate": 2.8768755980558463e-06,
      "loss": 0.4879,
      "step": 222610
    },
    {
      "epoch": 91.13,
      "grad_norm": 1.619608759880066,
      "learning_rate": 2.876721125631243e-06,
      "loss": 0.5015,
      "step": 222620
    },
    {
      "epoch": 91.13,
      "grad_norm": 1.6451361179351807,
      "learning_rate": 2.876566650765182e-06,
      "loss": 0.4713,
      "step": 222630
    },
    {
      "epoch": 91.13,
      "grad_norm": 1.8643319606781006,
      "learning_rate": 2.876412173458371e-06,
      "loss": 0.4846,
      "step": 222640
    },
    {
      "epoch": 91.14,
      "grad_norm": 2.1377980709075928,
      "learning_rate": 2.876257693711516e-06,
      "loss": 0.5125,
      "step": 222650
    },
    {
      "epoch": 91.14,
      "grad_norm": 1.912173867225647,
      "learning_rate": 2.8761032115253266e-06,
      "loss": 0.5045,
      "step": 222660
    },
    {
      "epoch": 91.15,
      "grad_norm": 1.8425086736679077,
      "learning_rate": 2.8759487269005093e-06,
      "loss": 0.4819,
      "step": 222670
    },
    {
      "epoch": 91.15,
      "grad_norm": 2.447378635406494,
      "learning_rate": 2.875794239837772e-06,
      "loss": 0.4953,
      "step": 222680
    },
    {
      "epoch": 91.15,
      "grad_norm": 1.9489697217941284,
      "learning_rate": 2.8756397503378226e-06,
      "loss": 0.4938,
      "step": 222690
    },
    {
      "epoch": 91.16,
      "grad_norm": 1.9012196063995361,
      "learning_rate": 2.875485258401368e-06,
      "loss": 0.4995,
      "step": 222700
    },
    {
      "epoch": 91.16,
      "grad_norm": 1.7896898984909058,
      "learning_rate": 2.875330764029117e-06,
      "loss": 0.4982,
      "step": 222710
    },
    {
      "epoch": 91.17,
      "grad_norm": 2.0397520065307617,
      "learning_rate": 2.875176267221777e-06,
      "loss": 0.4854,
      "step": 222720
    },
    {
      "epoch": 91.17,
      "grad_norm": 1.5996665954589844,
      "learning_rate": 2.8750217679800555e-06,
      "loss": 0.485,
      "step": 222730
    },
    {
      "epoch": 91.17,
      "grad_norm": 1.6731590032577515,
      "learning_rate": 2.87486726630466e-06,
      "loss": 0.4922,
      "step": 222740
    },
    {
      "epoch": 91.18,
      "grad_norm": 1.6009554862976074,
      "learning_rate": 2.8747127621962983e-06,
      "loss": 0.4893,
      "step": 222750
    },
    {
      "epoch": 91.18,
      "grad_norm": 2.171841859817505,
      "learning_rate": 2.874558255655679e-06,
      "loss": 0.4841,
      "step": 222760
    },
    {
      "epoch": 91.19,
      "grad_norm": 1.8893433809280396,
      "learning_rate": 2.874403746683509e-06,
      "loss": 0.4808,
      "step": 222770
    },
    {
      "epoch": 91.19,
      "grad_norm": 1.619645118713379,
      "learning_rate": 2.874249235280497e-06,
      "loss": 0.4984,
      "step": 222780
    },
    {
      "epoch": 91.2,
      "grad_norm": 1.993173599243164,
      "learning_rate": 2.8740947214473494e-06,
      "loss": 0.4803,
      "step": 222790
    },
    {
      "epoch": 91.2,
      "grad_norm": 2.0158908367156982,
      "learning_rate": 2.8739402051847753e-06,
      "loss": 0.506,
      "step": 222800
    },
    {
      "epoch": 91.2,
      "grad_norm": 1.8754593133926392,
      "learning_rate": 2.8737856864934817e-06,
      "loss": 0.5018,
      "step": 222810
    },
    {
      "epoch": 91.21,
      "grad_norm": 1.9187535047531128,
      "learning_rate": 2.8736311653741766e-06,
      "loss": 0.487,
      "step": 222820
    },
    {
      "epoch": 91.21,
      "grad_norm": 2.0181095600128174,
      "learning_rate": 2.873476641827568e-06,
      "loss": 0.4884,
      "step": 222830
    },
    {
      "epoch": 91.22,
      "grad_norm": 1.7713900804519653,
      "learning_rate": 2.8733221158543635e-06,
      "loss": 0.4692,
      "step": 222840
    },
    {
      "epoch": 91.22,
      "grad_norm": 2.009852647781372,
      "learning_rate": 2.873167587455272e-06,
      "loss": 0.4859,
      "step": 222850
    },
    {
      "epoch": 91.22,
      "grad_norm": 1.7034375667572021,
      "learning_rate": 2.8730130566310004e-06,
      "loss": 0.5109,
      "step": 222860
    },
    {
      "epoch": 91.23,
      "grad_norm": 1.7563236951828003,
      "learning_rate": 2.872858523382256e-06,
      "loss": 0.4851,
      "step": 222870
    },
    {
      "epoch": 91.23,
      "grad_norm": 1.8601142168045044,
      "learning_rate": 2.8727039877097473e-06,
      "loss": 0.4871,
      "step": 222880
    },
    {
      "epoch": 91.24,
      "grad_norm": 1.889863133430481,
      "learning_rate": 2.872549449614183e-06,
      "loss": 0.4753,
      "step": 222890
    },
    {
      "epoch": 91.24,
      "grad_norm": 1.8308558464050293,
      "learning_rate": 2.8723949090962694e-06,
      "loss": 0.5033,
      "step": 222900
    },
    {
      "epoch": 91.24,
      "grad_norm": 1.8624823093414307,
      "learning_rate": 2.8722403661567162e-06,
      "loss": 0.489,
      "step": 222910
    },
    {
      "epoch": 91.25,
      "grad_norm": 1.9021538496017456,
      "learning_rate": 2.8720858207962303e-06,
      "loss": 0.4973,
      "step": 222920
    },
    {
      "epoch": 91.25,
      "grad_norm": 1.8050202131271362,
      "learning_rate": 2.87193127301552e-06,
      "loss": 0.4958,
      "step": 222930
    },
    {
      "epoch": 91.26,
      "grad_norm": 1.5983365774154663,
      "learning_rate": 2.871776722815293e-06,
      "loss": 0.4918,
      "step": 222940
    },
    {
      "epoch": 91.26,
      "grad_norm": 1.5104272365570068,
      "learning_rate": 2.871622170196257e-06,
      "loss": 0.4953,
      "step": 222950
    },
    {
      "epoch": 91.26,
      "grad_norm": 1.8729792833328247,
      "learning_rate": 2.87146761515912e-06,
      "loss": 0.497,
      "step": 222960
    },
    {
      "epoch": 91.27,
      "grad_norm": 2.0059053897857666,
      "learning_rate": 2.871313057704591e-06,
      "loss": 0.4881,
      "step": 222970
    },
    {
      "epoch": 91.27,
      "grad_norm": 1.869825005531311,
      "learning_rate": 2.871158497833377e-06,
      "loss": 0.4838,
      "step": 222980
    },
    {
      "epoch": 91.28,
      "grad_norm": 2.160001754760742,
      "learning_rate": 2.871003935546186e-06,
      "loss": 0.49,
      "step": 222990
    },
    {
      "epoch": 91.28,
      "grad_norm": 1.8416082859039307,
      "learning_rate": 2.8708493708437277e-06,
      "loss": 0.4714,
      "step": 223000
    },
    {
      "epoch": 91.29,
      "grad_norm": 2.269282579421997,
      "learning_rate": 2.8706948037267083e-06,
      "loss": 0.4986,
      "step": 223010
    },
    {
      "epoch": 91.29,
      "grad_norm": 1.930729627609253,
      "learning_rate": 2.8705402341958356e-06,
      "loss": 0.4934,
      "step": 223020
    },
    {
      "epoch": 91.29,
      "grad_norm": 1.7813422679901123,
      "learning_rate": 2.870385662251819e-06,
      "loss": 0.5019,
      "step": 223030
    },
    {
      "epoch": 91.3,
      "grad_norm": 1.7962373495101929,
      "learning_rate": 2.8702310878953658e-06,
      "loss": 0.5043,
      "step": 223040
    },
    {
      "epoch": 91.3,
      "grad_norm": 2.669849395751953,
      "learning_rate": 2.8700765111271844e-06,
      "loss": 0.5002,
      "step": 223050
    },
    {
      "epoch": 91.31,
      "grad_norm": 1.5964614152908325,
      "learning_rate": 2.869921931947983e-06,
      "loss": 0.5045,
      "step": 223060
    },
    {
      "epoch": 91.31,
      "grad_norm": 1.8794975280761719,
      "learning_rate": 2.8697673503584694e-06,
      "loss": 0.507,
      "step": 223070
    },
    {
      "epoch": 91.31,
      "grad_norm": 1.9579139947891235,
      "learning_rate": 2.869612766359351e-06,
      "loss": 0.4796,
      "step": 223080
    },
    {
      "epoch": 91.32,
      "grad_norm": 1.6239442825317383,
      "learning_rate": 2.8694581799513373e-06,
      "loss": 0.4812,
      "step": 223090
    },
    {
      "epoch": 91.32,
      "grad_norm": 1.7604995965957642,
      "learning_rate": 2.869303591135136e-06,
      "loss": 0.493,
      "step": 223100
    },
    {
      "epoch": 91.33,
      "grad_norm": 2.1302647590637207,
      "learning_rate": 2.8691489999114546e-06,
      "loss": 0.5125,
      "step": 223110
    },
    {
      "epoch": 91.33,
      "grad_norm": 2.307140588760376,
      "learning_rate": 2.8689944062810026e-06,
      "loss": 0.4924,
      "step": 223120
    },
    {
      "epoch": 91.33,
      "grad_norm": 3.3616178035736084,
      "learning_rate": 2.8688398102444865e-06,
      "loss": 0.4962,
      "step": 223130
    },
    {
      "epoch": 91.34,
      "grad_norm": 3.107609272003174,
      "learning_rate": 2.868685211802616e-06,
      "loss": 0.4961,
      "step": 223140
    },
    {
      "epoch": 91.34,
      "grad_norm": 1.823943853378296,
      "learning_rate": 2.8685306109560978e-06,
      "loss": 0.5029,
      "step": 223150
    },
    {
      "epoch": 91.35,
      "grad_norm": 2.33716082572937,
      "learning_rate": 2.868376007705642e-06,
      "loss": 0.5264,
      "step": 223160
    },
    {
      "epoch": 91.35,
      "grad_norm": 1.957512378692627,
      "learning_rate": 2.8682214020519543e-06,
      "loss": 0.482,
      "step": 223170
    },
    {
      "epoch": 91.35,
      "grad_norm": 1.6082059144973755,
      "learning_rate": 2.8680667939957447e-06,
      "loss": 0.4971,
      "step": 223180
    },
    {
      "epoch": 91.36,
      "grad_norm": 2.005585193634033,
      "learning_rate": 2.8679121835377214e-06,
      "loss": 0.4857,
      "step": 223190
    },
    {
      "epoch": 91.36,
      "grad_norm": 2.093120574951172,
      "learning_rate": 2.8677575706785923e-06,
      "loss": 0.4916,
      "step": 223200
    },
    {
      "epoch": 91.37,
      "grad_norm": 2.119074821472168,
      "learning_rate": 2.8676029554190654e-06,
      "loss": 0.4742,
      "step": 223210
    },
    {
      "epoch": 91.37,
      "grad_norm": 1.8273144960403442,
      "learning_rate": 2.8674483377598506e-06,
      "loss": 0.4964,
      "step": 223220
    },
    {
      "epoch": 91.38,
      "grad_norm": 1.8804914951324463,
      "learning_rate": 2.8672937177016533e-06,
      "loss": 0.495,
      "step": 223230
    },
    {
      "epoch": 91.38,
      "grad_norm": 2.45516300201416,
      "learning_rate": 2.8671390952451835e-06,
      "loss": 0.4912,
      "step": 223240
    },
    {
      "epoch": 91.38,
      "grad_norm": 2.1272642612457275,
      "learning_rate": 2.8669844703911497e-06,
      "loss": 0.4886,
      "step": 223250
    },
    {
      "epoch": 91.39,
      "grad_norm": 1.5006624460220337,
      "learning_rate": 2.86682984314026e-06,
      "loss": 0.4977,
      "step": 223260
    },
    {
      "epoch": 91.39,
      "grad_norm": 1.9805703163146973,
      "learning_rate": 2.866675213493222e-06,
      "loss": 0.4755,
      "step": 223270
    },
    {
      "epoch": 91.4,
      "grad_norm": 1.5257868766784668,
      "learning_rate": 2.8665205814507444e-06,
      "loss": 0.459,
      "step": 223280
    },
    {
      "epoch": 91.4,
      "grad_norm": 1.6905224323272705,
      "learning_rate": 2.8663659470135365e-06,
      "loss": 0.4881,
      "step": 223290
    },
    {
      "epoch": 91.4,
      "grad_norm": 2.1623430252075195,
      "learning_rate": 2.8662113101823056e-06,
      "loss": 0.5028,
      "step": 223300
    },
    {
      "epoch": 91.41,
      "grad_norm": 2.946261167526245,
      "learning_rate": 2.86605667095776e-06,
      "loss": 0.5056,
      "step": 223310
    },
    {
      "epoch": 91.41,
      "grad_norm": 2.1858718395233154,
      "learning_rate": 2.8659020293406085e-06,
      "loss": 0.4807,
      "step": 223320
    },
    {
      "epoch": 91.42,
      "grad_norm": 2.171977996826172,
      "learning_rate": 2.86574738533156e-06,
      "loss": 0.4777,
      "step": 223330
    },
    {
      "epoch": 91.42,
      "grad_norm": 1.9220951795578003,
      "learning_rate": 2.8655927389313215e-06,
      "loss": 0.4672,
      "step": 223340
    },
    {
      "epoch": 91.42,
      "grad_norm": 1.646532416343689,
      "learning_rate": 2.865438090140603e-06,
      "loss": 0.5018,
      "step": 223350
    },
    {
      "epoch": 91.43,
      "grad_norm": 2.051652431488037,
      "learning_rate": 2.8652834389601116e-06,
      "loss": 0.4899,
      "step": 223360
    },
    {
      "epoch": 91.43,
      "grad_norm": 1.7198317050933838,
      "learning_rate": 2.8651287853905567e-06,
      "loss": 0.4957,
      "step": 223370
    },
    {
      "epoch": 91.44,
      "grad_norm": 1.833676815032959,
      "learning_rate": 2.864974129432646e-06,
      "loss": 0.4891,
      "step": 223380
    },
    {
      "epoch": 91.44,
      "grad_norm": 1.5716694593429565,
      "learning_rate": 2.864819471087088e-06,
      "loss": 0.4862,
      "step": 223390
    },
    {
      "epoch": 91.44,
      "grad_norm": 1.6230015754699707,
      "learning_rate": 2.8646648103545925e-06,
      "loss": 0.5004,
      "step": 223400
    },
    {
      "epoch": 91.45,
      "grad_norm": 2.069667339324951,
      "learning_rate": 2.864510147235866e-06,
      "loss": 0.4717,
      "step": 223410
    },
    {
      "epoch": 91.45,
      "grad_norm": 1.6257611513137817,
      "learning_rate": 2.8643554817316186e-06,
      "loss": 0.4959,
      "step": 223420
    },
    {
      "epoch": 91.46,
      "grad_norm": 2.5938801765441895,
      "learning_rate": 2.8642008138425576e-06,
      "loss": 0.4914,
      "step": 223430
    },
    {
      "epoch": 91.46,
      "grad_norm": 2.1802563667297363,
      "learning_rate": 2.864046143569392e-06,
      "loss": 0.4705,
      "step": 223440
    },
    {
      "epoch": 91.47,
      "grad_norm": 1.9218958616256714,
      "learning_rate": 2.863891470912831e-06,
      "loss": 0.486,
      "step": 223450
    },
    {
      "epoch": 91.47,
      "grad_norm": 1.573327660560608,
      "learning_rate": 2.863736795873582e-06,
      "loss": 0.5072,
      "step": 223460
    },
    {
      "epoch": 91.47,
      "grad_norm": 2.285501480102539,
      "learning_rate": 2.8635821184523534e-06,
      "loss": 0.4955,
      "step": 223470
    },
    {
      "epoch": 91.48,
      "grad_norm": 1.9077962636947632,
      "learning_rate": 2.863427438649855e-06,
      "loss": 0.5039,
      "step": 223480
    },
    {
      "epoch": 91.48,
      "grad_norm": 2.087468385696411,
      "learning_rate": 2.8632727564667944e-06,
      "loss": 0.502,
      "step": 223490
    },
    {
      "epoch": 91.49,
      "grad_norm": 1.723938226699829,
      "learning_rate": 2.8631180719038813e-06,
      "loss": 0.5034,
      "step": 223500
    },
    {
      "epoch": 91.49,
      "grad_norm": 1.9071117639541626,
      "learning_rate": 2.8629633849618225e-06,
      "loss": 0.4944,
      "step": 223510
    },
    {
      "epoch": 91.49,
      "grad_norm": 1.882964015007019,
      "learning_rate": 2.8628086956413287e-06,
      "loss": 0.4967,
      "step": 223520
    },
    {
      "epoch": 91.5,
      "grad_norm": 1.6380633115768433,
      "learning_rate": 2.8626540039431066e-06,
      "loss": 0.4913,
      "step": 223530
    },
    {
      "epoch": 91.5,
      "grad_norm": 2.281130313873291,
      "learning_rate": 2.862499309867866e-06,
      "loss": 0.4999,
      "step": 223540
    },
    {
      "epoch": 91.51,
      "grad_norm": 1.8328698873519897,
      "learning_rate": 2.862344613416315e-06,
      "loss": 0.5055,
      "step": 223550
    },
    {
      "epoch": 91.51,
      "grad_norm": 2.251145601272583,
      "learning_rate": 2.8621899145891617e-06,
      "loss": 0.478,
      "step": 223560
    },
    {
      "epoch": 91.51,
      "grad_norm": 2.1095831394195557,
      "learning_rate": 2.8620352133871165e-06,
      "loss": 0.4711,
      "step": 223570
    },
    {
      "epoch": 91.52,
      "grad_norm": 1.581763744354248,
      "learning_rate": 2.8618805098108867e-06,
      "loss": 0.4857,
      "step": 223580
    },
    {
      "epoch": 91.52,
      "grad_norm": 1.793029546737671,
      "learning_rate": 2.861725803861181e-06,
      "loss": 0.4921,
      "step": 223590
    },
    {
      "epoch": 91.53,
      "grad_norm": 1.7548078298568726,
      "learning_rate": 2.861571095538709e-06,
      "loss": 0.4782,
      "step": 223600
    },
    {
      "epoch": 91.53,
      "grad_norm": 2.097938060760498,
      "learning_rate": 2.861416384844178e-06,
      "loss": 0.4984,
      "step": 223610
    },
    {
      "epoch": 91.53,
      "grad_norm": 1.7098881006240845,
      "learning_rate": 2.8612616717782975e-06,
      "loss": 0.4846,
      "step": 223620
    },
    {
      "epoch": 91.54,
      "grad_norm": 1.7970962524414062,
      "learning_rate": 2.8611069563417764e-06,
      "loss": 0.4962,
      "step": 223630
    },
    {
      "epoch": 91.54,
      "grad_norm": 3.2449588775634766,
      "learning_rate": 2.8609522385353235e-06,
      "loss": 0.485,
      "step": 223640
    },
    {
      "epoch": 91.55,
      "grad_norm": 1.8015730381011963,
      "learning_rate": 2.860797518359647e-06,
      "loss": 0.5157,
      "step": 223650
    },
    {
      "epoch": 91.55,
      "grad_norm": 1.8033007383346558,
      "learning_rate": 2.8606427958154557e-06,
      "loss": 0.4864,
      "step": 223660
    },
    {
      "epoch": 91.56,
      "grad_norm": 1.7330029010772705,
      "learning_rate": 2.860488070903459e-06,
      "loss": 0.4937,
      "step": 223670
    },
    {
      "epoch": 91.56,
      "grad_norm": 2.0358901023864746,
      "learning_rate": 2.860333343624365e-06,
      "loss": 0.4818,
      "step": 223680
    },
    {
      "epoch": 91.56,
      "grad_norm": 2.0132389068603516,
      "learning_rate": 2.8601786139788825e-06,
      "loss": 0.505,
      "step": 223690
    },
    {
      "epoch": 91.57,
      "grad_norm": 1.9058070182800293,
      "learning_rate": 2.8600238819677206e-06,
      "loss": 0.4773,
      "step": 223700
    },
    {
      "epoch": 91.57,
      "grad_norm": 1.7462321519851685,
      "learning_rate": 2.8598691475915876e-06,
      "loss": 0.472,
      "step": 223710
    },
    {
      "epoch": 91.58,
      "grad_norm": 2.0107486248016357,
      "learning_rate": 2.8597144108511935e-06,
      "loss": 0.4968,
      "step": 223720
    },
    {
      "epoch": 91.58,
      "grad_norm": 2.7209017276763916,
      "learning_rate": 2.859559671747246e-06,
      "loss": 0.4729,
      "step": 223730
    },
    {
      "epoch": 91.58,
      "grad_norm": 1.9536442756652832,
      "learning_rate": 2.8594049302804547e-06,
      "loss": 0.5136,
      "step": 223740
    },
    {
      "epoch": 91.59,
      "grad_norm": 1.783316731452942,
      "learning_rate": 2.8592501864515272e-06,
      "loss": 0.4947,
      "step": 223750
    },
    {
      "epoch": 91.59,
      "grad_norm": 2.173941135406494,
      "learning_rate": 2.8590954402611736e-06,
      "loss": 0.4843,
      "step": 223760
    },
    {
      "epoch": 91.6,
      "grad_norm": 1.9724568128585815,
      "learning_rate": 2.858940691710102e-06,
      "loss": 0.5095,
      "step": 223770
    },
    {
      "epoch": 91.6,
      "grad_norm": 2.107269763946533,
      "learning_rate": 2.858785940799022e-06,
      "loss": 0.4904,
      "step": 223780
    },
    {
      "epoch": 91.6,
      "grad_norm": 1.665694236755371,
      "learning_rate": 2.8586311875286424e-06,
      "loss": 0.4911,
      "step": 223790
    },
    {
      "epoch": 91.61,
      "grad_norm": 2.987710475921631,
      "learning_rate": 2.8584764318996714e-06,
      "loss": 0.495,
      "step": 223800
    },
    {
      "epoch": 91.61,
      "grad_norm": 1.8957717418670654,
      "learning_rate": 2.858321673912819e-06,
      "loss": 0.475,
      "step": 223810
    },
    {
      "epoch": 91.62,
      "grad_norm": 1.9077587127685547,
      "learning_rate": 2.858166913568793e-06,
      "loss": 0.5026,
      "step": 223820
    },
    {
      "epoch": 91.62,
      "grad_norm": 1.8016085624694824,
      "learning_rate": 2.8580121508683026e-06,
      "loss": 0.4679,
      "step": 223830
    },
    {
      "epoch": 91.63,
      "grad_norm": 1.7365643978118896,
      "learning_rate": 2.8578573858120565e-06,
      "loss": 0.5065,
      "step": 223840
    },
    {
      "epoch": 91.63,
      "grad_norm": 1.6663486957550049,
      "learning_rate": 2.857702618400765e-06,
      "loss": 0.4776,
      "step": 223850
    },
    {
      "epoch": 91.63,
      "grad_norm": 2.064702272415161,
      "learning_rate": 2.857547848635135e-06,
      "loss": 0.4766,
      "step": 223860
    },
    {
      "epoch": 91.64,
      "grad_norm": 2.2459230422973633,
      "learning_rate": 2.857393076515878e-06,
      "loss": 0.4832,
      "step": 223870
    },
    {
      "epoch": 91.64,
      "grad_norm": 2.338667154312134,
      "learning_rate": 2.8572383020437016e-06,
      "loss": 0.4877,
      "step": 223880
    },
    {
      "epoch": 91.65,
      "grad_norm": 1.7260991334915161,
      "learning_rate": 2.8570835252193138e-06,
      "loss": 0.5021,
      "step": 223890
    },
    {
      "epoch": 91.65,
      "grad_norm": 1.7283847332000732,
      "learning_rate": 2.8569287460434253e-06,
      "loss": 0.4895,
      "step": 223900
    },
    {
      "epoch": 91.65,
      "grad_norm": 2.1129982471466064,
      "learning_rate": 2.856773964516744e-06,
      "loss": 0.4709,
      "step": 223910
    },
    {
      "epoch": 91.66,
      "grad_norm": 2.5203452110290527,
      "learning_rate": 2.8566191806399796e-06,
      "loss": 0.4989,
      "step": 223920
    },
    {
      "epoch": 91.66,
      "grad_norm": 2.102059841156006,
      "learning_rate": 2.8564643944138403e-06,
      "loss": 0.4779,
      "step": 223930
    },
    {
      "epoch": 91.67,
      "grad_norm": 1.5192503929138184,
      "learning_rate": 2.8563096058390366e-06,
      "loss": 0.4526,
      "step": 223940
    },
    {
      "epoch": 91.67,
      "grad_norm": 1.6180208921432495,
      "learning_rate": 2.8561548149162763e-06,
      "loss": 0.5021,
      "step": 223950
    },
    {
      "epoch": 91.67,
      "grad_norm": 1.5836663246154785,
      "learning_rate": 2.856000021646269e-06,
      "loss": 0.4742,
      "step": 223960
    },
    {
      "epoch": 91.68,
      "grad_norm": 1.6854184865951538,
      "learning_rate": 2.855845226029724e-06,
      "loss": 0.4942,
      "step": 223970
    },
    {
      "epoch": 91.68,
      "grad_norm": 1.8169275522232056,
      "learning_rate": 2.8556904280673493e-06,
      "loss": 0.4901,
      "step": 223980
    },
    {
      "epoch": 91.69,
      "grad_norm": 2.176467180252075,
      "learning_rate": 2.8555356277598557e-06,
      "loss": 0.4818,
      "step": 223990
    },
    {
      "epoch": 91.69,
      "grad_norm": 2.452165365219116,
      "learning_rate": 2.8553808251079506e-06,
      "loss": 0.4872,
      "step": 224000
    },
    {
      "epoch": 91.69,
      "grad_norm": 1.9540224075317383,
      "learning_rate": 2.8552260201123447e-06,
      "loss": 0.4862,
      "step": 224010
    },
    {
      "epoch": 91.7,
      "grad_norm": 1.5899549722671509,
      "learning_rate": 2.8550712127737457e-06,
      "loss": 0.4829,
      "step": 224020
    },
    {
      "epoch": 91.7,
      "grad_norm": 2.0063397884368896,
      "learning_rate": 2.854916403092864e-06,
      "loss": 0.4754,
      "step": 224030
    },
    {
      "epoch": 91.71,
      "grad_norm": 1.5306261777877808,
      "learning_rate": 2.854761591070408e-06,
      "loss": 0.4933,
      "step": 224040
    },
    {
      "epoch": 91.71,
      "grad_norm": 1.85953688621521,
      "learning_rate": 2.8546067767070868e-06,
      "loss": 0.4959,
      "step": 224050
    },
    {
      "epoch": 91.72,
      "grad_norm": 1.8237457275390625,
      "learning_rate": 2.85445196000361e-06,
      "loss": 0.523,
      "step": 224060
    },
    {
      "epoch": 91.72,
      "grad_norm": 1.7613205909729004,
      "learning_rate": 2.854297140960686e-06,
      "loss": 0.4875,
      "step": 224070
    },
    {
      "epoch": 91.72,
      "grad_norm": 1.6434916257858276,
      "learning_rate": 2.854142319579025e-06,
      "loss": 0.4614,
      "step": 224080
    },
    {
      "epoch": 91.73,
      "grad_norm": 1.7659406661987305,
      "learning_rate": 2.8539874958593363e-06,
      "loss": 0.4863,
      "step": 224090
    },
    {
      "epoch": 91.73,
      "grad_norm": 2.56854248046875,
      "learning_rate": 2.8538326698023286e-06,
      "loss": 0.502,
      "step": 224100
    },
    {
      "epoch": 91.74,
      "grad_norm": 2.3035213947296143,
      "learning_rate": 2.853677841408711e-06,
      "loss": 0.5126,
      "step": 224110
    },
    {
      "epoch": 91.74,
      "grad_norm": 1.4932620525360107,
      "learning_rate": 2.8535230106791926e-06,
      "loss": 0.4795,
      "step": 224120
    },
    {
      "epoch": 91.74,
      "grad_norm": 3.1312520503997803,
      "learning_rate": 2.853368177614483e-06,
      "loss": 0.4882,
      "step": 224130
    },
    {
      "epoch": 91.75,
      "grad_norm": 1.878493070602417,
      "learning_rate": 2.853213342215291e-06,
      "loss": 0.4736,
      "step": 224140
    },
    {
      "epoch": 91.75,
      "grad_norm": 1.6605829000473022,
      "learning_rate": 2.8530585044823273e-06,
      "loss": 0.4929,
      "step": 224150
    },
    {
      "epoch": 91.76,
      "grad_norm": 2.1040806770324707,
      "learning_rate": 2.8529036644163e-06,
      "loss": 0.4909,
      "step": 224160
    },
    {
      "epoch": 91.76,
      "grad_norm": 1.902101993560791,
      "learning_rate": 2.8527488220179195e-06,
      "loss": 0.4916,
      "step": 224170
    },
    {
      "epoch": 91.76,
      "grad_norm": 1.8115509748458862,
      "learning_rate": 2.852593977287893e-06,
      "loss": 0.4907,
      "step": 224180
    },
    {
      "epoch": 91.77,
      "grad_norm": 2.080451250076294,
      "learning_rate": 2.852439130226931e-06,
      "loss": 0.4902,
      "step": 224190
    },
    {
      "epoch": 91.77,
      "grad_norm": 1.9436583518981934,
      "learning_rate": 2.852284280835743e-06,
      "loss": 0.4748,
      "step": 224200
    },
    {
      "epoch": 91.78,
      "grad_norm": 1.6025607585906982,
      "learning_rate": 2.852129429115039e-06,
      "loss": 0.4856,
      "step": 224210
    },
    {
      "epoch": 91.78,
      "grad_norm": 1.7592058181762695,
      "learning_rate": 2.8519745750655262e-06,
      "loss": 0.472,
      "step": 224220
    },
    {
      "epoch": 91.78,
      "grad_norm": 1.6794005632400513,
      "learning_rate": 2.8518197186879167e-06,
      "loss": 0.5081,
      "step": 224230
    },
    {
      "epoch": 91.79,
      "grad_norm": 2.2927305698394775,
      "learning_rate": 2.851664859982918e-06,
      "loss": 0.4844,
      "step": 224240
    },
    {
      "epoch": 91.79,
      "grad_norm": 2.031865119934082,
      "learning_rate": 2.8515099989512394e-06,
      "loss": 0.4705,
      "step": 224250
    },
    {
      "epoch": 91.8,
      "grad_norm": 1.932271122932434,
      "learning_rate": 2.8513551355935917e-06,
      "loss": 0.4906,
      "step": 224260
    },
    {
      "epoch": 91.8,
      "grad_norm": 1.8339170217514038,
      "learning_rate": 2.851200269910683e-06,
      "loss": 0.4648,
      "step": 224270
    },
    {
      "epoch": 91.81,
      "grad_norm": 2.2600700855255127,
      "learning_rate": 2.8510454019032236e-06,
      "loss": 0.4747,
      "step": 224280
    },
    {
      "epoch": 91.81,
      "grad_norm": 1.9460824728012085,
      "learning_rate": 2.850890531571922e-06,
      "loss": 0.5018,
      "step": 224290
    },
    {
      "epoch": 91.81,
      "grad_norm": 2.1531412601470947,
      "learning_rate": 2.8507356589174888e-06,
      "loss": 0.4859,
      "step": 224300
    },
    {
      "epoch": 91.82,
      "grad_norm": 2.187948703765869,
      "learning_rate": 2.850580783940632e-06,
      "loss": 0.4882,
      "step": 224310
    },
    {
      "epoch": 91.82,
      "grad_norm": 2.287611246109009,
      "learning_rate": 2.850425906642063e-06,
      "loss": 0.4719,
      "step": 224320
    },
    {
      "epoch": 91.83,
      "grad_norm": 2.1621668338775635,
      "learning_rate": 2.8502710270224892e-06,
      "loss": 0.4726,
      "step": 224330
    },
    {
      "epoch": 91.83,
      "grad_norm": 2.2200419902801514,
      "learning_rate": 2.850116145082621e-06,
      "loss": 0.4757,
      "step": 224340
    },
    {
      "epoch": 91.83,
      "grad_norm": 2.373317241668701,
      "learning_rate": 2.849961260823168e-06,
      "loss": 0.4881,
      "step": 224350
    },
    {
      "epoch": 91.84,
      "grad_norm": 1.910203456878662,
      "learning_rate": 2.8498063742448402e-06,
      "loss": 0.4808,
      "step": 224360
    },
    {
      "epoch": 91.84,
      "grad_norm": 1.6985081434249878,
      "learning_rate": 2.849651485348346e-06,
      "loss": 0.5054,
      "step": 224370
    },
    {
      "epoch": 91.85,
      "grad_norm": 1.8270678520202637,
      "learning_rate": 2.8494965941343955e-06,
      "loss": 0.4753,
      "step": 224380
    },
    {
      "epoch": 91.85,
      "grad_norm": 2.5272560119628906,
      "learning_rate": 2.849341700603699e-06,
      "loss": 0.4744,
      "step": 224390
    },
    {
      "epoch": 91.85,
      "grad_norm": 2.3201284408569336,
      "learning_rate": 2.849186804756964e-06,
      "loss": 0.5057,
      "step": 224400
    },
    {
      "epoch": 91.86,
      "grad_norm": 1.912187099456787,
      "learning_rate": 2.8490319065949014e-06,
      "loss": 0.4715,
      "step": 224410
    },
    {
      "epoch": 91.86,
      "grad_norm": 1.830360770225525,
      "learning_rate": 2.848877006118221e-06,
      "loss": 0.4837,
      "step": 224420
    },
    {
      "epoch": 91.87,
      "grad_norm": 1.8929154872894287,
      "learning_rate": 2.8487221033276325e-06,
      "loss": 0.5018,
      "step": 224430
    },
    {
      "epoch": 91.87,
      "grad_norm": 1.513521432876587,
      "learning_rate": 2.848567198223844e-06,
      "loss": 0.4921,
      "step": 224440
    },
    {
      "epoch": 91.87,
      "grad_norm": 1.913759708404541,
      "learning_rate": 2.8484122908075667e-06,
      "loss": 0.4885,
      "step": 224450
    },
    {
      "epoch": 91.88,
      "grad_norm": 2.7761118412017822,
      "learning_rate": 2.8482573810795103e-06,
      "loss": 0.4923,
      "step": 224460
    },
    {
      "epoch": 91.88,
      "grad_norm": 2.325525999069214,
      "learning_rate": 2.8481024690403826e-06,
      "loss": 0.4665,
      "step": 224470
    },
    {
      "epoch": 91.89,
      "grad_norm": 2.3295106887817383,
      "learning_rate": 2.8479475546908945e-06,
      "loss": 0.4762,
      "step": 224480
    },
    {
      "epoch": 91.89,
      "grad_norm": 2.4265809059143066,
      "learning_rate": 2.847792638031756e-06,
      "loss": 0.4906,
      "step": 224490
    },
    {
      "epoch": 91.9,
      "grad_norm": 2.3289804458618164,
      "learning_rate": 2.8476377190636755e-06,
      "loss": 0.4775,
      "step": 224500
    },
    {
      "epoch": 91.9,
      "grad_norm": 1.7054500579833984,
      "learning_rate": 2.8474827977873633e-06,
      "loss": 0.4864,
      "step": 224510
    },
    {
      "epoch": 91.9,
      "grad_norm": 1.5920380353927612,
      "learning_rate": 2.8473278742035297e-06,
      "loss": 0.4723,
      "step": 224520
    },
    {
      "epoch": 91.91,
      "grad_norm": 2.4387612342834473,
      "learning_rate": 2.8471729483128844e-06,
      "loss": 0.5094,
      "step": 224530
    },
    {
      "epoch": 91.91,
      "grad_norm": 1.5743443965911865,
      "learning_rate": 2.847018020116136e-06,
      "loss": 0.486,
      "step": 224540
    },
    {
      "epoch": 91.92,
      "grad_norm": 2.1454062461853027,
      "learning_rate": 2.8468630896139943e-06,
      "loss": 0.4811,
      "step": 224550
    },
    {
      "epoch": 91.92,
      "grad_norm": 1.9407591819763184,
      "learning_rate": 2.8467081568071705e-06,
      "loss": 0.4933,
      "step": 224560
    },
    {
      "epoch": 91.92,
      "grad_norm": 1.8733004331588745,
      "learning_rate": 2.846553221696372e-06,
      "loss": 0.4853,
      "step": 224570
    },
    {
      "epoch": 91.93,
      "grad_norm": 1.7673463821411133,
      "learning_rate": 2.8463982842823103e-06,
      "loss": 0.4833,
      "step": 224580
    },
    {
      "epoch": 91.93,
      "grad_norm": 2.138103485107422,
      "learning_rate": 2.846243344565695e-06,
      "loss": 0.497,
      "step": 224590
    },
    {
      "epoch": 91.94,
      "grad_norm": 1.8075907230377197,
      "learning_rate": 2.8460884025472356e-06,
      "loss": 0.4941,
      "step": 224600
    },
    {
      "epoch": 91.94,
      "grad_norm": 1.535165548324585,
      "learning_rate": 2.845933458227641e-06,
      "loss": 0.5004,
      "step": 224610
    },
    {
      "epoch": 91.94,
      "grad_norm": 1.775275707244873,
      "learning_rate": 2.845778511607622e-06,
      "loss": 0.4964,
      "step": 224620
    },
    {
      "epoch": 91.95,
      "grad_norm": 1.498491883277893,
      "learning_rate": 2.8456235626878886e-06,
      "loss": 0.4812,
      "step": 224630
    },
    {
      "epoch": 91.95,
      "grad_norm": 2.5393097400665283,
      "learning_rate": 2.8454686114691502e-06,
      "loss": 0.4808,
      "step": 224640
    },
    {
      "epoch": 91.96,
      "grad_norm": 2.314121961593628,
      "learning_rate": 2.8453136579521163e-06,
      "loss": 0.4961,
      "step": 224650
    },
    {
      "epoch": 91.96,
      "grad_norm": 1.8795355558395386,
      "learning_rate": 2.845158702137497e-06,
      "loss": 0.4886,
      "step": 224660
    },
    {
      "epoch": 91.96,
      "grad_norm": 2.050276279449463,
      "learning_rate": 2.8450037440260013e-06,
      "loss": 0.4588,
      "step": 224670
    },
    {
      "epoch": 91.97,
      "grad_norm": 1.6495925188064575,
      "learning_rate": 2.844848783618341e-06,
      "loss": 0.4806,
      "step": 224680
    },
    {
      "epoch": 91.97,
      "grad_norm": 1.7708994150161743,
      "learning_rate": 2.8446938209152243e-06,
      "loss": 0.4914,
      "step": 224690
    },
    {
      "epoch": 91.98,
      "grad_norm": 1.9205859899520874,
      "learning_rate": 2.8445388559173613e-06,
      "loss": 0.4946,
      "step": 224700
    },
    {
      "epoch": 91.98,
      "grad_norm": 1.7576193809509277,
      "learning_rate": 2.8443838886254625e-06,
      "loss": 0.4882,
      "step": 224710
    },
    {
      "epoch": 91.99,
      "grad_norm": 2.6926629543304443,
      "learning_rate": 2.844228919040237e-06,
      "loss": 0.491,
      "step": 224720
    },
    {
      "epoch": 91.99,
      "grad_norm": 1.6128544807434082,
      "learning_rate": 2.8440739471623953e-06,
      "loss": 0.5112,
      "step": 224730
    },
    {
      "epoch": 91.99,
      "grad_norm": 1.3618417978286743,
      "learning_rate": 2.8439189729926474e-06,
      "loss": 0.5036,
      "step": 224740
    },
    {
      "epoch": 92.0,
      "grad_norm": 1.7849409580230713,
      "learning_rate": 2.843763996531703e-06,
      "loss": 0.4839,
      "step": 224750
    },
    {
      "epoch": 92.0,
      "eval_loss": 0.48813551664352417,
      "eval_runtime": 52.3041,
      "eval_samples_per_second": 65.941,
      "eval_steps_per_second": 8.259,
      "step": 224756
    },
    {
      "epoch": 92.0,
      "grad_norm": 2.1907358169555664,
      "learning_rate": 2.8436090177802713e-06,
      "loss": 0.4607,
      "step": 224760
    },
    {
      "epoch": 92.01,
      "grad_norm": 1.653171181678772,
      "learning_rate": 2.843454036739063e-06,
      "loss": 0.5076,
      "step": 224770
    },
    {
      "epoch": 92.01,
      "grad_norm": 2.975916624069214,
      "learning_rate": 2.8432990534087876e-06,
      "loss": 0.4685,
      "step": 224780
    },
    {
      "epoch": 92.01,
      "grad_norm": 2.0030508041381836,
      "learning_rate": 2.8431440677901558e-06,
      "loss": 0.4883,
      "step": 224790
    },
    {
      "epoch": 92.02,
      "grad_norm": 1.925081491470337,
      "learning_rate": 2.8429890798838768e-06,
      "loss": 0.5015,
      "step": 224800
    },
    {
      "epoch": 92.02,
      "grad_norm": 2.3448028564453125,
      "learning_rate": 2.8428340896906614e-06,
      "loss": 0.4872,
      "step": 224810
    },
    {
      "epoch": 92.03,
      "grad_norm": 1.6777546405792236,
      "learning_rate": 2.842679097211219e-06,
      "loss": 0.4827,
      "step": 224820
    },
    {
      "epoch": 92.03,
      "grad_norm": 1.6974323987960815,
      "learning_rate": 2.842524102446259e-06,
      "loss": 0.5019,
      "step": 224830
    },
    {
      "epoch": 92.03,
      "grad_norm": 1.9931402206420898,
      "learning_rate": 2.842369105396493e-06,
      "loss": 0.4834,
      "step": 224840
    },
    {
      "epoch": 92.04,
      "grad_norm": 2.376566171646118,
      "learning_rate": 2.842214106062629e-06,
      "loss": 0.4795,
      "step": 224850
    },
    {
      "epoch": 92.04,
      "grad_norm": 2.0900213718414307,
      "learning_rate": 2.8420591044453794e-06,
      "loss": 0.4889,
      "step": 224860
    },
    {
      "epoch": 92.05,
      "grad_norm": 1.925708532333374,
      "learning_rate": 2.8419041005454523e-06,
      "loss": 0.4705,
      "step": 224870
    },
    {
      "epoch": 92.05,
      "grad_norm": 1.890868902206421,
      "learning_rate": 2.8417490943635582e-06,
      "loss": 0.4737,
      "step": 224880
    },
    {
      "epoch": 92.05,
      "grad_norm": 3.5072357654571533,
      "learning_rate": 2.841594085900408e-06,
      "loss": 0.4913,
      "step": 224890
    },
    {
      "epoch": 92.06,
      "grad_norm": 2.1712889671325684,
      "learning_rate": 2.8414390751567104e-06,
      "loss": 0.4918,
      "step": 224900
    },
    {
      "epoch": 92.06,
      "grad_norm": 1.6660531759262085,
      "learning_rate": 2.8412840621331767e-06,
      "loss": 0.4987,
      "step": 224910
    },
    {
      "epoch": 92.07,
      "grad_norm": 2.2062504291534424,
      "learning_rate": 2.8411290468305166e-06,
      "loss": 0.4885,
      "step": 224920
    },
    {
      "epoch": 92.07,
      "grad_norm": 1.8992655277252197,
      "learning_rate": 2.8409740292494395e-06,
      "loss": 0.4916,
      "step": 224930
    },
    {
      "epoch": 92.08,
      "grad_norm": 1.818604826927185,
      "learning_rate": 2.8408190093906565e-06,
      "loss": 0.4548,
      "step": 224940
    },
    {
      "epoch": 92.08,
      "grad_norm": 1.613076090812683,
      "learning_rate": 2.8406639872548774e-06,
      "loss": 0.5036,
      "step": 224950
    },
    {
      "epoch": 92.08,
      "grad_norm": 2.161163806915283,
      "learning_rate": 2.840508962842812e-06,
      "loss": 0.4934,
      "step": 224960
    },
    {
      "epoch": 92.09,
      "grad_norm": 1.4677801132202148,
      "learning_rate": 2.8403539361551715e-06,
      "loss": 0.4863,
      "step": 224970
    },
    {
      "epoch": 92.09,
      "grad_norm": 1.697121024131775,
      "learning_rate": 2.8401989071926643e-06,
      "loss": 0.4845,
      "step": 224980
    },
    {
      "epoch": 92.1,
      "grad_norm": 2.056083917617798,
      "learning_rate": 2.840043875956002e-06,
      "loss": 0.4793,
      "step": 224990
    },
    {
      "epoch": 92.1,
      "grad_norm": 1.9900462627410889,
      "learning_rate": 2.8398888424458947e-06,
      "loss": 0.4834,
      "step": 225000
    },
    {
      "epoch": 92.1,
      "grad_norm": 2.006868362426758,
      "learning_rate": 2.8397338066630516e-06,
      "loss": 0.4697,
      "step": 225010
    },
    {
      "epoch": 92.11,
      "grad_norm": 1.763899326324463,
      "learning_rate": 2.8395787686081836e-06,
      "loss": 0.4774,
      "step": 225020
    },
    {
      "epoch": 92.11,
      "grad_norm": 2.054685354232788,
      "learning_rate": 2.839423728282001e-06,
      "loss": 0.4794,
      "step": 225030
    },
    {
      "epoch": 92.12,
      "grad_norm": 2.18515944480896,
      "learning_rate": 2.8392686856852144e-06,
      "loss": 0.4898,
      "step": 225040
    },
    {
      "epoch": 92.12,
      "grad_norm": 1.8754810094833374,
      "learning_rate": 2.8391136408185324e-06,
      "loss": 0.5211,
      "step": 225050
    },
    {
      "epoch": 92.12,
      "grad_norm": 1.8094420433044434,
      "learning_rate": 2.838958593682667e-06,
      "loss": 0.474,
      "step": 225060
    },
    {
      "epoch": 92.13,
      "grad_norm": 1.6786103248596191,
      "learning_rate": 2.8388035442783265e-06,
      "loss": 0.4946,
      "step": 225070
    },
    {
      "epoch": 92.13,
      "grad_norm": 2.0567119121551514,
      "learning_rate": 2.8386484926062236e-06,
      "loss": 0.4889,
      "step": 225080
    },
    {
      "epoch": 92.14,
      "grad_norm": 2.285534381866455,
      "learning_rate": 2.838493438667067e-06,
      "loss": 0.4878,
      "step": 225090
    },
    {
      "epoch": 92.14,
      "grad_norm": 1.7236480712890625,
      "learning_rate": 2.8383383824615674e-06,
      "loss": 0.4887,
      "step": 225100
    },
    {
      "epoch": 92.14,
      "grad_norm": 2.3916754722595215,
      "learning_rate": 2.838183323990435e-06,
      "loss": 0.5114,
      "step": 225110
    },
    {
      "epoch": 92.15,
      "grad_norm": 2.0537638664245605,
      "learning_rate": 2.8380282632543804e-06,
      "loss": 0.4646,
      "step": 225120
    },
    {
      "epoch": 92.15,
      "grad_norm": 2.076713800430298,
      "learning_rate": 2.837873200254113e-06,
      "loss": 0.4901,
      "step": 225130
    },
    {
      "epoch": 92.16,
      "grad_norm": 1.7013306617736816,
      "learning_rate": 2.837718134990344e-06,
      "loss": 0.509,
      "step": 225140
    },
    {
      "epoch": 92.16,
      "grad_norm": 2.039001941680908,
      "learning_rate": 2.8375630674637837e-06,
      "loss": 0.5198,
      "step": 225150
    },
    {
      "epoch": 92.17,
      "grad_norm": 1.6423414945602417,
      "learning_rate": 2.8374079976751414e-06,
      "loss": 0.4682,
      "step": 225160
    },
    {
      "epoch": 92.17,
      "grad_norm": 2.092740058898926,
      "learning_rate": 2.837252925625129e-06,
      "loss": 0.4821,
      "step": 225170
    },
    {
      "epoch": 92.17,
      "grad_norm": 1.4623297452926636,
      "learning_rate": 2.8370978513144566e-06,
      "loss": 0.4976,
      "step": 225180
    },
    {
      "epoch": 92.18,
      "grad_norm": 1.9353513717651367,
      "learning_rate": 2.8369427747438333e-06,
      "loss": 0.4897,
      "step": 225190
    },
    {
      "epoch": 92.18,
      "grad_norm": 2.014111280441284,
      "learning_rate": 2.8367876959139706e-06,
      "loss": 0.4748,
      "step": 225200
    },
    {
      "epoch": 92.19,
      "grad_norm": 1.8022147417068481,
      "learning_rate": 2.8366326148255785e-06,
      "loss": 0.5035,
      "step": 225210
    },
    {
      "epoch": 92.19,
      "grad_norm": 1.7086610794067383,
      "learning_rate": 2.8364775314793675e-06,
      "loss": 0.4847,
      "step": 225220
    },
    {
      "epoch": 92.19,
      "grad_norm": 2.0876049995422363,
      "learning_rate": 2.836322445876048e-06,
      "loss": 0.4896,
      "step": 225230
    },
    {
      "epoch": 92.2,
      "grad_norm": 1.7671853303909302,
      "learning_rate": 2.8361673580163304e-06,
      "loss": 0.4772,
      "step": 225240
    },
    {
      "epoch": 92.2,
      "grad_norm": 2.158780097961426,
      "learning_rate": 2.8360122679009247e-06,
      "loss": 0.4889,
      "step": 225250
    },
    {
      "epoch": 92.21,
      "grad_norm": 1.4613913297653198,
      "learning_rate": 2.8358571755305425e-06,
      "loss": 0.4866,
      "step": 225260
    },
    {
      "epoch": 92.21,
      "grad_norm": 1.8131089210510254,
      "learning_rate": 2.8357020809058934e-06,
      "loss": 0.4759,
      "step": 225270
    },
    {
      "epoch": 92.21,
      "grad_norm": 1.7028186321258545,
      "learning_rate": 2.8355469840276878e-06,
      "loss": 0.5092,
      "step": 225280
    },
    {
      "epoch": 92.22,
      "grad_norm": 1.7993824481964111,
      "learning_rate": 2.8353918848966362e-06,
      "loss": 0.4896,
      "step": 225290
    },
    {
      "epoch": 92.22,
      "grad_norm": 1.60416579246521,
      "learning_rate": 2.8352367835134495e-06,
      "loss": 0.4789,
      "step": 225300
    },
    {
      "epoch": 92.23,
      "grad_norm": 1.7120743989944458,
      "learning_rate": 2.8350816798788372e-06,
      "loss": 0.4891,
      "step": 225310
    },
    {
      "epoch": 92.23,
      "grad_norm": 1.6870172023773193,
      "learning_rate": 2.834926573993512e-06,
      "loss": 0.4916,
      "step": 225320
    },
    {
      "epoch": 92.23,
      "grad_norm": 1.6723041534423828,
      "learning_rate": 2.8347714658581825e-06,
      "loss": 0.482,
      "step": 225330
    },
    {
      "epoch": 92.24,
      "grad_norm": 2.2349767684936523,
      "learning_rate": 2.8346163554735593e-06,
      "loss": 0.5068,
      "step": 225340
    },
    {
      "epoch": 92.24,
      "grad_norm": 2.402517557144165,
      "learning_rate": 2.8344612428403536e-06,
      "loss": 0.4951,
      "step": 225350
    },
    {
      "epoch": 92.25,
      "grad_norm": 2.1399106979370117,
      "learning_rate": 2.8343061279592754e-06,
      "loss": 0.4851,
      "step": 225360
    },
    {
      "epoch": 92.25,
      "grad_norm": 1.8417755365371704,
      "learning_rate": 2.8341510108310357e-06,
      "loss": 0.4925,
      "step": 225370
    },
    {
      "epoch": 92.26,
      "grad_norm": 2.7191734313964844,
      "learning_rate": 2.833995891456345e-06,
      "loss": 0.4884,
      "step": 225380
    },
    {
      "epoch": 92.26,
      "grad_norm": 1.5476216077804565,
      "learning_rate": 2.833840769835914e-06,
      "loss": 0.4645,
      "step": 225390
    },
    {
      "epoch": 92.26,
      "grad_norm": 1.9884581565856934,
      "learning_rate": 2.833685645970453e-06,
      "loss": 0.5127,
      "step": 225400
    },
    {
      "epoch": 92.27,
      "grad_norm": 1.661111831665039,
      "learning_rate": 2.833530519860672e-06,
      "loss": 0.4605,
      "step": 225410
    },
    {
      "epoch": 92.27,
      "grad_norm": 2.4420599937438965,
      "learning_rate": 2.8333753915072834e-06,
      "loss": 0.4616,
      "step": 225420
    },
    {
      "epoch": 92.28,
      "grad_norm": 1.9022830724716187,
      "learning_rate": 2.833220260910996e-06,
      "loss": 0.491,
      "step": 225430
    },
    {
      "epoch": 92.28,
      "grad_norm": 2.5158863067626953,
      "learning_rate": 2.833065128072521e-06,
      "loss": 0.4968,
      "step": 225440
    },
    {
      "epoch": 92.28,
      "grad_norm": 1.616324782371521,
      "learning_rate": 2.832909992992569e-06,
      "loss": 0.4894,
      "step": 225450
    },
    {
      "epoch": 92.29,
      "grad_norm": 1.625508427619934,
      "learning_rate": 2.8327548556718518e-06,
      "loss": 0.4811,
      "step": 225460
    },
    {
      "epoch": 92.29,
      "grad_norm": 2.303220748901367,
      "learning_rate": 2.8325997161110787e-06,
      "loss": 0.4971,
      "step": 225470
    },
    {
      "epoch": 92.3,
      "grad_norm": 1.8705124855041504,
      "learning_rate": 2.8324445743109605e-06,
      "loss": 0.4659,
      "step": 225480
    },
    {
      "epoch": 92.3,
      "grad_norm": 1.54218327999115,
      "learning_rate": 2.832289430272209e-06,
      "loss": 0.4818,
      "step": 225490
    },
    {
      "epoch": 92.3,
      "grad_norm": 1.700172781944275,
      "learning_rate": 2.832134283995533e-06,
      "loss": 0.4952,
      "step": 225500
    },
    {
      "epoch": 92.31,
      "grad_norm": 1.6539545059204102,
      "learning_rate": 2.831979135481644e-06,
      "loss": 0.4982,
      "step": 225510
    },
    {
      "epoch": 92.31,
      "grad_norm": 1.6750975847244263,
      "learning_rate": 2.8318239847312536e-06,
      "loss": 0.4789,
      "step": 225520
    },
    {
      "epoch": 92.32,
      "grad_norm": 1.6646130084991455,
      "learning_rate": 2.8316688317450722e-06,
      "loss": 0.4922,
      "step": 225530
    },
    {
      "epoch": 92.32,
      "grad_norm": 1.8903217315673828,
      "learning_rate": 2.8315136765238094e-06,
      "loss": 0.4859,
      "step": 225540
    },
    {
      "epoch": 92.33,
      "grad_norm": 1.9707635641098022,
      "learning_rate": 2.831358519068177e-06,
      "loss": 0.4724,
      "step": 225550
    },
    {
      "epoch": 92.33,
      "grad_norm": 2.0234367847442627,
      "learning_rate": 2.831203359378886e-06,
      "loss": 0.4763,
      "step": 225560
    },
    {
      "epoch": 92.33,
      "grad_norm": 2.3854284286499023,
      "learning_rate": 2.8310481974566464e-06,
      "loss": 0.4931,
      "step": 225570
    },
    {
      "epoch": 92.34,
      "grad_norm": 2.0999534130096436,
      "learning_rate": 2.8308930333021687e-06,
      "loss": 0.4996,
      "step": 225580
    },
    {
      "epoch": 92.34,
      "grad_norm": 2.767848491668701,
      "learning_rate": 2.8307378669161646e-06,
      "loss": 0.5068,
      "step": 225590
    },
    {
      "epoch": 92.35,
      "grad_norm": 1.8316619396209717,
      "learning_rate": 2.830582698299344e-06,
      "loss": 0.4653,
      "step": 225600
    },
    {
      "epoch": 92.35,
      "grad_norm": 1.844999074935913,
      "learning_rate": 2.830427527452419e-06,
      "loss": 0.4967,
      "step": 225610
    },
    {
      "epoch": 92.35,
      "grad_norm": 1.8921066522598267,
      "learning_rate": 2.8302723543760997e-06,
      "loss": 0.4988,
      "step": 225620
    },
    {
      "epoch": 92.36,
      "grad_norm": 1.4906284809112549,
      "learning_rate": 2.8301171790710965e-06,
      "loss": 0.5134,
      "step": 225630
    },
    {
      "epoch": 92.36,
      "grad_norm": 2.052656412124634,
      "learning_rate": 2.82996200153812e-06,
      "loss": 0.4913,
      "step": 225640
    },
    {
      "epoch": 92.37,
      "grad_norm": 2.7756729125976562,
      "learning_rate": 2.829806821777882e-06,
      "loss": 0.4879,
      "step": 225650
    },
    {
      "epoch": 92.37,
      "grad_norm": 2.1493027210235596,
      "learning_rate": 2.8296516397910934e-06,
      "loss": 0.4778,
      "step": 225660
    },
    {
      "epoch": 92.37,
      "grad_norm": 1.6230639219284058,
      "learning_rate": 2.8294964555784643e-06,
      "loss": 0.4807,
      "step": 225670
    },
    {
      "epoch": 92.38,
      "grad_norm": 2.2173545360565186,
      "learning_rate": 2.8293412691407056e-06,
      "loss": 0.5049,
      "step": 225680
    },
    {
      "epoch": 92.38,
      "grad_norm": 1.4716187715530396,
      "learning_rate": 2.8291860804785294e-06,
      "loss": 0.4765,
      "step": 225690
    },
    {
      "epoch": 92.39,
      "grad_norm": 1.8836458921432495,
      "learning_rate": 2.8290308895926453e-06,
      "loss": 0.4923,
      "step": 225700
    },
    {
      "epoch": 92.39,
      "grad_norm": 1.6762057542800903,
      "learning_rate": 2.8288756964837643e-06,
      "loss": 0.5106,
      "step": 225710
    },
    {
      "epoch": 92.39,
      "grad_norm": 1.685890793800354,
      "learning_rate": 2.8287205011525975e-06,
      "loss": 0.4939,
      "step": 225720
    },
    {
      "epoch": 92.4,
      "grad_norm": 1.9271042346954346,
      "learning_rate": 2.8285653035998555e-06,
      "loss": 0.4936,
      "step": 225730
    },
    {
      "epoch": 92.4,
      "grad_norm": 1.8760040998458862,
      "learning_rate": 2.828410103826251e-06,
      "loss": 0.4953,
      "step": 225740
    },
    {
      "epoch": 92.41,
      "grad_norm": 1.8715351819992065,
      "learning_rate": 2.8282549018324926e-06,
      "loss": 0.4968,
      "step": 225750
    },
    {
      "epoch": 92.41,
      "grad_norm": 1.8790252208709717,
      "learning_rate": 2.828099697619293e-06,
      "loss": 0.4891,
      "step": 225760
    },
    {
      "epoch": 92.42,
      "grad_norm": 2.000667095184326,
      "learning_rate": 2.8279444911873626e-06,
      "loss": 0.4774,
      "step": 225770
    },
    {
      "epoch": 92.42,
      "grad_norm": 2.510366678237915,
      "learning_rate": 2.8277892825374116e-06,
      "loss": 0.4753,
      "step": 225780
    },
    {
      "epoch": 92.42,
      "grad_norm": 1.9519721269607544,
      "learning_rate": 2.827634071670152e-06,
      "loss": 0.4578,
      "step": 225790
    },
    {
      "epoch": 92.43,
      "grad_norm": 2.3922464847564697,
      "learning_rate": 2.827478858586294e-06,
      "loss": 0.4732,
      "step": 225800
    },
    {
      "epoch": 92.43,
      "grad_norm": 1.8369741439819336,
      "learning_rate": 2.827323643286549e-06,
      "loss": 0.4986,
      "step": 225810
    },
    {
      "epoch": 92.44,
      "grad_norm": 2.440333604812622,
      "learning_rate": 2.827168425771628e-06,
      "loss": 0.4901,
      "step": 225820
    },
    {
      "epoch": 92.44,
      "grad_norm": 2.0574121475219727,
      "learning_rate": 2.8270132060422424e-06,
      "loss": 0.4913,
      "step": 225830
    },
    {
      "epoch": 92.44,
      "grad_norm": 1.793549656867981,
      "learning_rate": 2.826857984099103e-06,
      "loss": 0.473,
      "step": 225840
    },
    {
      "epoch": 92.45,
      "grad_norm": 1.9031245708465576,
      "learning_rate": 2.8267027599429204e-06,
      "loss": 0.4933,
      "step": 225850
    },
    {
      "epoch": 92.45,
      "grad_norm": 2.0557808876037598,
      "learning_rate": 2.826547533574406e-06,
      "loss": 0.4676,
      "step": 225860
    },
    {
      "epoch": 92.46,
      "grad_norm": 2.6717357635498047,
      "learning_rate": 2.8263923049942708e-06,
      "loss": 0.4859,
      "step": 225870
    },
    {
      "epoch": 92.46,
      "grad_norm": 1.927512526512146,
      "learning_rate": 2.8262370742032262e-06,
      "loss": 0.4762,
      "step": 225880
    },
    {
      "epoch": 92.46,
      "grad_norm": 1.9775928258895874,
      "learning_rate": 2.826081841201983e-06,
      "loss": 0.5129,
      "step": 225890
    },
    {
      "epoch": 92.47,
      "grad_norm": 1.9160692691802979,
      "learning_rate": 2.8259266059912523e-06,
      "loss": 0.4703,
      "step": 225900
    },
    {
      "epoch": 92.47,
      "grad_norm": 1.6719598770141602,
      "learning_rate": 2.825771368571746e-06,
      "loss": 0.4947,
      "step": 225910
    },
    {
      "epoch": 92.48,
      "grad_norm": 1.6094361543655396,
      "learning_rate": 2.825616128944173e-06,
      "loss": 0.4801,
      "step": 225920
    },
    {
      "epoch": 92.48,
      "grad_norm": 2.0237183570861816,
      "learning_rate": 2.825460887109247e-06,
      "loss": 0.4975,
      "step": 225930
    },
    {
      "epoch": 92.48,
      "grad_norm": 1.6816977262496948,
      "learning_rate": 2.8253056430676768e-06,
      "loss": 0.4911,
      "step": 225940
    },
    {
      "epoch": 92.49,
      "grad_norm": 1.6882693767547607,
      "learning_rate": 2.8251503968201758e-06,
      "loss": 0.4882,
      "step": 225950
    },
    {
      "epoch": 92.49,
      "grad_norm": 1.9123119115829468,
      "learning_rate": 2.824995148367454e-06,
      "loss": 0.4834,
      "step": 225960
    },
    {
      "epoch": 92.5,
      "grad_norm": 1.6125071048736572,
      "learning_rate": 2.8248398977102225e-06,
      "loss": 0.4974,
      "step": 225970
    },
    {
      "epoch": 92.5,
      "grad_norm": 2.2444980144500732,
      "learning_rate": 2.8246846448491933e-06,
      "loss": 0.4835,
      "step": 225980
    },
    {
      "epoch": 92.51,
      "grad_norm": 1.830837368965149,
      "learning_rate": 2.8245293897850767e-06,
      "loss": 0.4739,
      "step": 225990
    },
    {
      "epoch": 92.51,
      "grad_norm": 2.3003625869750977,
      "learning_rate": 2.824374132518584e-06,
      "loss": 0.4924,
      "step": 226000
    },
    {
      "epoch": 92.51,
      "grad_norm": 1.7492055892944336,
      "learning_rate": 2.824218873050426e-06,
      "loss": 0.482,
      "step": 226010
    },
    {
      "epoch": 92.52,
      "grad_norm": 2.4920613765716553,
      "learning_rate": 2.824063611381315e-06,
      "loss": 0.4937,
      "step": 226020
    },
    {
      "epoch": 92.52,
      "grad_norm": 1.6810822486877441,
      "learning_rate": 2.823908347511962e-06,
      "loss": 0.4983,
      "step": 226030
    },
    {
      "epoch": 92.53,
      "grad_norm": 2.9570469856262207,
      "learning_rate": 2.8237530814430778e-06,
      "loss": 0.4984,
      "step": 226040
    },
    {
      "epoch": 92.53,
      "grad_norm": 2.010327100753784,
      "learning_rate": 2.823597813175374e-06,
      "loss": 0.4925,
      "step": 226050
    },
    {
      "epoch": 92.53,
      "grad_norm": 1.8143558502197266,
      "learning_rate": 2.8234425427095615e-06,
      "loss": 0.4859,
      "step": 226060
    },
    {
      "epoch": 92.54,
      "grad_norm": 1.6011278629302979,
      "learning_rate": 2.8232872700463517e-06,
      "loss": 0.475,
      "step": 226070
    },
    {
      "epoch": 92.54,
      "grad_norm": 1.5818766355514526,
      "learning_rate": 2.8231319951864555e-06,
      "loss": 0.5027,
      "step": 226080
    },
    {
      "epoch": 92.55,
      "grad_norm": 1.7348456382751465,
      "learning_rate": 2.8229767181305857e-06,
      "loss": 0.4805,
      "step": 226090
    },
    {
      "epoch": 92.55,
      "grad_norm": 1.9565510749816895,
      "learning_rate": 2.8228214388794514e-06,
      "loss": 0.4871,
      "step": 226100
    },
    {
      "epoch": 92.55,
      "grad_norm": 2.191180467605591,
      "learning_rate": 2.822666157433765e-06,
      "loss": 0.5038,
      "step": 226110
    },
    {
      "epoch": 92.56,
      "grad_norm": 2.8269150257110596,
      "learning_rate": 2.822510873794239e-06,
      "loss": 0.4949,
      "step": 226120
    },
    {
      "epoch": 92.56,
      "grad_norm": 2.386317729949951,
      "learning_rate": 2.822355587961583e-06,
      "loss": 0.4973,
      "step": 226130
    },
    {
      "epoch": 92.57,
      "grad_norm": 1.7246274948120117,
      "learning_rate": 2.8222002999365086e-06,
      "loss": 0.4743,
      "step": 226140
    },
    {
      "epoch": 92.57,
      "grad_norm": 2.2443442344665527,
      "learning_rate": 2.8220450097197276e-06,
      "loss": 0.4972,
      "step": 226150
    },
    {
      "epoch": 92.57,
      "grad_norm": 2.1865241527557373,
      "learning_rate": 2.8218897173119515e-06,
      "loss": 0.4742,
      "step": 226160
    },
    {
      "epoch": 92.58,
      "grad_norm": 1.9827280044555664,
      "learning_rate": 2.8217344227138914e-06,
      "loss": 0.4922,
      "step": 226170
    },
    {
      "epoch": 92.58,
      "grad_norm": 1.9148980379104614,
      "learning_rate": 2.8215791259262587e-06,
      "loss": 0.5018,
      "step": 226180
    },
    {
      "epoch": 92.59,
      "grad_norm": 1.982655644416809,
      "learning_rate": 2.8214238269497645e-06,
      "loss": 0.4801,
      "step": 226190
    },
    {
      "epoch": 92.59,
      "grad_norm": 2.1460583209991455,
      "learning_rate": 2.821268525785121e-06,
      "loss": 0.504,
      "step": 226200
    },
    {
      "epoch": 92.6,
      "grad_norm": 2.0550947189331055,
      "learning_rate": 2.8211132224330383e-06,
      "loss": 0.4984,
      "step": 226210
    },
    {
      "epoch": 92.6,
      "grad_norm": 2.1790049076080322,
      "learning_rate": 2.820957916894229e-06,
      "loss": 0.4947,
      "step": 226220
    },
    {
      "epoch": 92.6,
      "grad_norm": 1.84318208694458,
      "learning_rate": 2.8208026091694043e-06,
      "loss": 0.5046,
      "step": 226230
    },
    {
      "epoch": 92.61,
      "grad_norm": 1.673215627670288,
      "learning_rate": 2.8206472992592754e-06,
      "loss": 0.4833,
      "step": 226240
    },
    {
      "epoch": 92.61,
      "grad_norm": 1.7920165061950684,
      "learning_rate": 2.8204919871645537e-06,
      "loss": 0.4952,
      "step": 226250
    },
    {
      "epoch": 92.62,
      "grad_norm": 2.0883302688598633,
      "learning_rate": 2.8203366728859507e-06,
      "loss": 0.4973,
      "step": 226260
    },
    {
      "epoch": 92.62,
      "grad_norm": 2.9804303646087646,
      "learning_rate": 2.8201813564241787e-06,
      "loss": 0.4753,
      "step": 226270
    },
    {
      "epoch": 92.62,
      "grad_norm": 2.399306297302246,
      "learning_rate": 2.8200260377799475e-06,
      "loss": 0.4937,
      "step": 226280
    },
    {
      "epoch": 92.63,
      "grad_norm": 1.6294547319412231,
      "learning_rate": 2.8198707169539704e-06,
      "loss": 0.4813,
      "step": 226290
    },
    {
      "epoch": 92.63,
      "grad_norm": 2.0405609607696533,
      "learning_rate": 2.8197153939469567e-06,
      "loss": 0.4684,
      "step": 226300
    },
    {
      "epoch": 92.64,
      "grad_norm": 1.7699089050292969,
      "learning_rate": 2.81956006875962e-06,
      "loss": 0.4793,
      "step": 226310
    },
    {
      "epoch": 92.64,
      "grad_norm": 2.4756836891174316,
      "learning_rate": 2.819404741392671e-06,
      "loss": 0.4803,
      "step": 226320
    },
    {
      "epoch": 92.64,
      "grad_norm": 1.6457997560501099,
      "learning_rate": 2.8192494118468214e-06,
      "loss": 0.4914,
      "step": 226330
    },
    {
      "epoch": 92.65,
      "grad_norm": 2.3198904991149902,
      "learning_rate": 2.8190940801227833e-06,
      "loss": 0.4853,
      "step": 226340
    },
    {
      "epoch": 92.65,
      "grad_norm": 1.8545783758163452,
      "learning_rate": 2.818938746221267e-06,
      "loss": 0.4952,
      "step": 226350
    },
    {
      "epoch": 92.66,
      "grad_norm": 1.3628522157669067,
      "learning_rate": 2.8187834101429844e-06,
      "loss": 0.4637,
      "step": 226360
    },
    {
      "epoch": 92.66,
      "grad_norm": 1.8007981777191162,
      "learning_rate": 2.818628071888648e-06,
      "loss": 0.4613,
      "step": 226370
    },
    {
      "epoch": 92.66,
      "grad_norm": 1.8801084756851196,
      "learning_rate": 2.818472731458968e-06,
      "loss": 0.5052,
      "step": 226380
    },
    {
      "epoch": 92.67,
      "grad_norm": 2.0124764442443848,
      "learning_rate": 2.8183173888546564e-06,
      "loss": 0.5025,
      "step": 226390
    },
    {
      "epoch": 92.67,
      "grad_norm": 1.6703886985778809,
      "learning_rate": 2.818162044076425e-06,
      "loss": 0.4857,
      "step": 226400
    },
    {
      "epoch": 92.68,
      "grad_norm": 2.167145013809204,
      "learning_rate": 2.8180066971249873e-06,
      "loss": 0.4808,
      "step": 226410
    },
    {
      "epoch": 92.68,
      "grad_norm": 1.925742506980896,
      "learning_rate": 2.8178513480010514e-06,
      "loss": 0.4752,
      "step": 226420
    },
    {
      "epoch": 92.69,
      "grad_norm": 1.8997002840042114,
      "learning_rate": 2.817695996705331e-06,
      "loss": 0.4914,
      "step": 226430
    },
    {
      "epoch": 92.69,
      "grad_norm": 2.2410829067230225,
      "learning_rate": 2.817540643238538e-06,
      "loss": 0.5129,
      "step": 226440
    },
    {
      "epoch": 92.69,
      "grad_norm": 1.8294488191604614,
      "learning_rate": 2.8173852876013824e-06,
      "loss": 0.4976,
      "step": 226450
    },
    {
      "epoch": 92.7,
      "grad_norm": 1.6381750106811523,
      "learning_rate": 2.817229929794578e-06,
      "loss": 0.477,
      "step": 226460
    },
    {
      "epoch": 92.7,
      "grad_norm": 1.776374340057373,
      "learning_rate": 2.8170745698188344e-06,
      "loss": 0.4888,
      "step": 226470
    },
    {
      "epoch": 92.71,
      "grad_norm": 1.5953058004379272,
      "learning_rate": 2.8169192076748646e-06,
      "loss": 0.4821,
      "step": 226480
    },
    {
      "epoch": 92.71,
      "grad_norm": 1.903009295463562,
      "learning_rate": 2.8167638433633806e-06,
      "loss": 0.4847,
      "step": 226490
    },
    {
      "epoch": 92.71,
      "grad_norm": 1.6842842102050781,
      "learning_rate": 2.8166084768850928e-06,
      "loss": 0.4753,
      "step": 226500
    },
    {
      "epoch": 92.72,
      "grad_norm": 2.0251107215881348,
      "learning_rate": 2.8164531082407134e-06,
      "loss": 0.485,
      "step": 226510
    },
    {
      "epoch": 92.72,
      "grad_norm": 2.543344736099243,
      "learning_rate": 2.8162977374309545e-06,
      "loss": 0.4744,
      "step": 226520
    },
    {
      "epoch": 92.73,
      "grad_norm": 1.7605886459350586,
      "learning_rate": 2.8161423644565274e-06,
      "loss": 0.4726,
      "step": 226530
    },
    {
      "epoch": 92.73,
      "grad_norm": 1.8224294185638428,
      "learning_rate": 2.8159869893181442e-06,
      "loss": 0.4901,
      "step": 226540
    },
    {
      "epoch": 92.73,
      "grad_norm": 1.731990098953247,
      "learning_rate": 2.8158316120165164e-06,
      "loss": 0.4873,
      "step": 226550
    },
    {
      "epoch": 92.74,
      "grad_norm": 2.4129695892333984,
      "learning_rate": 2.8156762325523567e-06,
      "loss": 0.495,
      "step": 226560
    },
    {
      "epoch": 92.74,
      "grad_norm": 1.8584479093551636,
      "learning_rate": 2.815520850926375e-06,
      "loss": 0.4997,
      "step": 226570
    },
    {
      "epoch": 92.75,
      "grad_norm": 1.9425448179244995,
      "learning_rate": 2.815365467139284e-06,
      "loss": 0.4941,
      "step": 226580
    },
    {
      "epoch": 92.75,
      "grad_norm": 1.7049468755722046,
      "learning_rate": 2.8152100811917956e-06,
      "loss": 0.5058,
      "step": 226590
    },
    {
      "epoch": 92.75,
      "grad_norm": 1.6029046773910522,
      "learning_rate": 2.8150546930846222e-06,
      "loss": 0.4766,
      "step": 226600
    },
    {
      "epoch": 92.76,
      "grad_norm": 1.8039864301681519,
      "learning_rate": 2.8148993028184745e-06,
      "loss": 0.4635,
      "step": 226610
    },
    {
      "epoch": 92.76,
      "grad_norm": 1.8364776372909546,
      "learning_rate": 2.8147439103940653e-06,
      "loss": 0.5159,
      "step": 226620
    },
    {
      "epoch": 92.77,
      "grad_norm": 1.6870687007904053,
      "learning_rate": 2.8145885158121056e-06,
      "loss": 0.4754,
      "step": 226630
    },
    {
      "epoch": 92.77,
      "grad_norm": 2.078688859939575,
      "learning_rate": 2.8144331190733082e-06,
      "loss": 0.4654,
      "step": 226640
    },
    {
      "epoch": 92.78,
      "grad_norm": 1.6381487846374512,
      "learning_rate": 2.814277720178384e-06,
      "loss": 0.478,
      "step": 226650
    },
    {
      "epoch": 92.78,
      "grad_norm": 2.3814141750335693,
      "learning_rate": 2.8141223191280448e-06,
      "loss": 0.5009,
      "step": 226660
    },
    {
      "epoch": 92.78,
      "grad_norm": 1.5289268493652344,
      "learning_rate": 2.8139669159230025e-06,
      "loss": 0.4516,
      "step": 226670
    },
    {
      "epoch": 92.79,
      "grad_norm": 1.7764959335327148,
      "learning_rate": 2.8138115105639695e-06,
      "loss": 0.4647,
      "step": 226680
    },
    {
      "epoch": 92.79,
      "grad_norm": 2.4288382530212402,
      "learning_rate": 2.813656103051658e-06,
      "loss": 0.4731,
      "step": 226690
    },
    {
      "epoch": 92.8,
      "grad_norm": 1.8672385215759277,
      "learning_rate": 2.81350069338678e-06,
      "loss": 0.4838,
      "step": 226700
    },
    {
      "epoch": 92.8,
      "grad_norm": 1.6340579986572266,
      "learning_rate": 2.8133452815700456e-06,
      "loss": 0.4733,
      "step": 226710
    },
    {
      "epoch": 92.8,
      "grad_norm": 1.57486891746521,
      "learning_rate": 2.8131898676021686e-06,
      "loss": 0.4746,
      "step": 226720
    },
    {
      "epoch": 92.81,
      "grad_norm": 2.1525986194610596,
      "learning_rate": 2.8130344514838597e-06,
      "loss": 0.4999,
      "step": 226730
    },
    {
      "epoch": 92.81,
      "grad_norm": 2.079493761062622,
      "learning_rate": 2.8128790332158323e-06,
      "loss": 0.4852,
      "step": 226740
    },
    {
      "epoch": 92.82,
      "grad_norm": 2.518688917160034,
      "learning_rate": 2.812723612798797e-06,
      "loss": 0.4938,
      "step": 226750
    },
    {
      "epoch": 92.82,
      "grad_norm": 2.339798927307129,
      "learning_rate": 2.8125681902334664e-06,
      "loss": 0.4917,
      "step": 226760
    },
    {
      "epoch": 92.82,
      "grad_norm": 1.7708697319030762,
      "learning_rate": 2.8124127655205517e-06,
      "loss": 0.4855,
      "step": 226770
    },
    {
      "epoch": 92.83,
      "grad_norm": 1.7531784772872925,
      "learning_rate": 2.8122573386607655e-06,
      "loss": 0.4904,
      "step": 226780
    },
    {
      "epoch": 92.83,
      "grad_norm": 1.8832197189331055,
      "learning_rate": 2.81210190965482e-06,
      "loss": 0.4894,
      "step": 226790
    },
    {
      "epoch": 92.84,
      "grad_norm": 2.1229026317596436,
      "learning_rate": 2.8119464785034272e-06,
      "loss": 0.4835,
      "step": 226800
    },
    {
      "epoch": 92.84,
      "grad_norm": 2.0465378761291504,
      "learning_rate": 2.8117910452072985e-06,
      "loss": 0.4924,
      "step": 226810
    },
    {
      "epoch": 92.84,
      "grad_norm": 2.010045289993286,
      "learning_rate": 2.811635609767146e-06,
      "loss": 0.4862,
      "step": 226820
    },
    {
      "epoch": 92.85,
      "grad_norm": 1.951422929763794,
      "learning_rate": 2.811480172183683e-06,
      "loss": 0.5021,
      "step": 226830
    },
    {
      "epoch": 92.85,
      "grad_norm": 3.3826630115509033,
      "learning_rate": 2.8113247324576193e-06,
      "loss": 0.4981,
      "step": 226840
    },
    {
      "epoch": 92.86,
      "grad_norm": 1.7519557476043701,
      "learning_rate": 2.8111692905896697e-06,
      "loss": 0.4742,
      "step": 226850
    },
    {
      "epoch": 92.86,
      "grad_norm": 1.9631268978118896,
      "learning_rate": 2.811013846580543e-06,
      "loss": 0.4956,
      "step": 226860
    },
    {
      "epoch": 92.87,
      "grad_norm": 1.685778260231018,
      "learning_rate": 2.810858400430954e-06,
      "loss": 0.4695,
      "step": 226870
    },
    {
      "epoch": 92.87,
      "grad_norm": 2.0626912117004395,
      "learning_rate": 2.8107029521416135e-06,
      "loss": 0.4847,
      "step": 226880
    },
    {
      "epoch": 92.87,
      "grad_norm": 2.410292863845825,
      "learning_rate": 2.8105475017132338e-06,
      "loss": 0.478,
      "step": 226890
    },
    {
      "epoch": 92.88,
      "grad_norm": 1.9866433143615723,
      "learning_rate": 2.8103920491465266e-06,
      "loss": 0.4844,
      "step": 226900
    },
    {
      "epoch": 92.88,
      "grad_norm": 2.633984327316284,
      "learning_rate": 2.8102365944422056e-06,
      "loss": 0.4728,
      "step": 226910
    },
    {
      "epoch": 92.89,
      "grad_norm": 1.8879203796386719,
      "learning_rate": 2.810081137600981e-06,
      "loss": 0.4774,
      "step": 226920
    },
    {
      "epoch": 92.89,
      "grad_norm": 2.2808570861816406,
      "learning_rate": 2.809925678623566e-06,
      "loss": 0.4852,
      "step": 226930
    },
    {
      "epoch": 92.89,
      "grad_norm": 2.2947685718536377,
      "learning_rate": 2.8097702175106724e-06,
      "loss": 0.4865,
      "step": 226940
    },
    {
      "epoch": 92.9,
      "grad_norm": 1.6263490915298462,
      "learning_rate": 2.809614754263012e-06,
      "loss": 0.5123,
      "step": 226950
    },
    {
      "epoch": 92.9,
      "grad_norm": 1.5739446878433228,
      "learning_rate": 2.8094592888812964e-06,
      "loss": 0.5018,
      "step": 226960
    },
    {
      "epoch": 92.91,
      "grad_norm": 1.7250447273254395,
      "learning_rate": 2.80930382136624e-06,
      "loss": 0.4815,
      "step": 226970
    },
    {
      "epoch": 92.91,
      "grad_norm": 1.906983733177185,
      "learning_rate": 2.8091483517185534e-06,
      "loss": 0.4901,
      "step": 226980
    },
    {
      "epoch": 92.91,
      "grad_norm": 2.48596453666687,
      "learning_rate": 2.8089928799389494e-06,
      "loss": 0.4732,
      "step": 226990
    },
    {
      "epoch": 92.92,
      "grad_norm": 1.6557878255844116,
      "learning_rate": 2.8088374060281393e-06,
      "loss": 0.4936,
      "step": 227000
    },
    {
      "epoch": 92.92,
      "grad_norm": 1.7451908588409424,
      "learning_rate": 2.808681929986836e-06,
      "loss": 0.4868,
      "step": 227010
    },
    {
      "epoch": 92.93,
      "grad_norm": 1.7816606760025024,
      "learning_rate": 2.8085264518157514e-06,
      "loss": 0.4959,
      "step": 227020
    },
    {
      "epoch": 92.93,
      "grad_norm": 2.602675199508667,
      "learning_rate": 2.8083709715155978e-06,
      "loss": 0.4824,
      "step": 227030
    },
    {
      "epoch": 92.93,
      "grad_norm": 1.6364214420318604,
      "learning_rate": 2.8082154890870875e-06,
      "loss": 0.4952,
      "step": 227040
    },
    {
      "epoch": 92.94,
      "grad_norm": 1.7978618144989014,
      "learning_rate": 2.8080600045309326e-06,
      "loss": 0.4842,
      "step": 227050
    },
    {
      "epoch": 92.94,
      "grad_norm": 2.1158692836761475,
      "learning_rate": 2.807904517847846e-06,
      "loss": 0.4909,
      "step": 227060
    },
    {
      "epoch": 92.95,
      "grad_norm": 1.6349738836288452,
      "learning_rate": 2.807749029038539e-06,
      "loss": 0.4874,
      "step": 227070
    },
    {
      "epoch": 92.95,
      "grad_norm": 1.7710716724395752,
      "learning_rate": 2.8075935381037248e-06,
      "loss": 0.4995,
      "step": 227080
    },
    {
      "epoch": 92.96,
      "grad_norm": 1.801926612854004,
      "learning_rate": 2.807438045044114e-06,
      "loss": 0.4736,
      "step": 227090
    },
    {
      "epoch": 92.96,
      "grad_norm": 1.6713876724243164,
      "learning_rate": 2.807282549860422e-06,
      "loss": 0.4945,
      "step": 227100
    },
    {
      "epoch": 92.96,
      "grad_norm": 2.0502500534057617,
      "learning_rate": 2.807127052553358e-06,
      "loss": 0.5026,
      "step": 227110
    },
    {
      "epoch": 92.97,
      "grad_norm": 1.6916064023971558,
      "learning_rate": 2.806971553123635e-06,
      "loss": 0.5045,
      "step": 227120
    },
    {
      "epoch": 92.97,
      "grad_norm": 3.0261290073394775,
      "learning_rate": 2.8068160515719665e-06,
      "loss": 0.4893,
      "step": 227130
    },
    {
      "epoch": 92.98,
      "grad_norm": 1.7255905866622925,
      "learning_rate": 2.806660547899064e-06,
      "loss": 0.492,
      "step": 227140
    },
    {
      "epoch": 92.98,
      "grad_norm": 1.9722808599472046,
      "learning_rate": 2.80650504210564e-06,
      "loss": 0.49,
      "step": 227150
    },
    {
      "epoch": 92.98,
      "grad_norm": 2.174996852874756,
      "learning_rate": 2.8063495341924065e-06,
      "loss": 0.471,
      "step": 227160
    },
    {
      "epoch": 92.99,
      "grad_norm": 1.8855807781219482,
      "learning_rate": 2.8061940241600768e-06,
      "loss": 0.4818,
      "step": 227170
    },
    {
      "epoch": 92.99,
      "grad_norm": 2.307969570159912,
      "learning_rate": 2.806038512009362e-06,
      "loss": 0.5176,
      "step": 227180
    },
    {
      "epoch": 93.0,
      "grad_norm": 2.66713547706604,
      "learning_rate": 2.8058829977409757e-06,
      "loss": 0.4955,
      "step": 227190
    },
    {
      "epoch": 93.0,
      "eval_loss": 0.48804131150245667,
      "eval_runtime": 57.6314,
      "eval_samples_per_second": 59.846,
      "eval_steps_per_second": 7.496,
      "step": 227199
    },
    {
      "epoch": 93.0,
      "grad_norm": 2.158867835998535,
      "learning_rate": 2.805727481355629e-06,
      "loss": 0.4724,
      "step": 227200
    },
    {
      "epoch": 93.0,
      "grad_norm": 2.0160701274871826,
      "learning_rate": 2.8055719628540362e-06,
      "loss": 0.4736,
      "step": 227210
    },
    {
      "epoch": 93.01,
      "grad_norm": 2.018536329269409,
      "learning_rate": 2.8054164422369075e-06,
      "loss": 0.4882,
      "step": 227220
    },
    {
      "epoch": 93.01,
      "grad_norm": 1.8788105249404907,
      "learning_rate": 2.805260919504957e-06,
      "loss": 0.4829,
      "step": 227230
    },
    {
      "epoch": 93.02,
      "grad_norm": 2.3004519939422607,
      "learning_rate": 2.805105394658895e-06,
      "loss": 0.509,
      "step": 227240
    },
    {
      "epoch": 93.02,
      "grad_norm": 1.8529857397079468,
      "learning_rate": 2.8049498676994364e-06,
      "loss": 0.4597,
      "step": 227250
    },
    {
      "epoch": 93.02,
      "grad_norm": 1.7299572229385376,
      "learning_rate": 2.804794338627293e-06,
      "loss": 0.5036,
      "step": 227260
    },
    {
      "epoch": 93.03,
      "grad_norm": 1.7611892223358154,
      "learning_rate": 2.8046388074431766e-06,
      "loss": 0.4816,
      "step": 227270
    },
    {
      "epoch": 93.03,
      "grad_norm": 1.864345908164978,
      "learning_rate": 2.8044832741478e-06,
      "loss": 0.4941,
      "step": 227280
    },
    {
      "epoch": 93.04,
      "grad_norm": 2.2332539558410645,
      "learning_rate": 2.804327738741876e-06,
      "loss": 0.4809,
      "step": 227290
    },
    {
      "epoch": 93.04,
      "grad_norm": 1.7069475650787354,
      "learning_rate": 2.8041722012261155e-06,
      "loss": 0.4911,
      "step": 227300
    },
    {
      "epoch": 93.05,
      "grad_norm": 1.9340616464614868,
      "learning_rate": 2.8040166616012332e-06,
      "loss": 0.4962,
      "step": 227310
    },
    {
      "epoch": 93.05,
      "grad_norm": 1.9709885120391846,
      "learning_rate": 2.8038611198679407e-06,
      "loss": 0.4773,
      "step": 227320
    },
    {
      "epoch": 93.05,
      "grad_norm": 1.4857205152511597,
      "learning_rate": 2.8037055760269495e-06,
      "loss": 0.4718,
      "step": 227330
    },
    {
      "epoch": 93.06,
      "grad_norm": 2.081124782562256,
      "learning_rate": 2.8035500300789735e-06,
      "loss": 0.479,
      "step": 227340
    },
    {
      "epoch": 93.06,
      "grad_norm": 2.2864227294921875,
      "learning_rate": 2.803394482024725e-06,
      "loss": 0.4852,
      "step": 227350
    },
    {
      "epoch": 93.07,
      "grad_norm": 1.9001027345657349,
      "learning_rate": 2.8032389318649163e-06,
      "loss": 0.4657,
      "step": 227360
    },
    {
      "epoch": 93.07,
      "grad_norm": 2.2307302951812744,
      "learning_rate": 2.80308337960026e-06,
      "loss": 0.4745,
      "step": 227370
    },
    {
      "epoch": 93.07,
      "grad_norm": 1.6904902458190918,
      "learning_rate": 2.802927825231468e-06,
      "loss": 0.4798,
      "step": 227380
    },
    {
      "epoch": 93.08,
      "grad_norm": 1.7505382299423218,
      "learning_rate": 2.802772268759254e-06,
      "loss": 0.5067,
      "step": 227390
    },
    {
      "epoch": 93.08,
      "grad_norm": 2.6124191284179688,
      "learning_rate": 2.80261671018433e-06,
      "loss": 0.4783,
      "step": 227400
    },
    {
      "epoch": 93.09,
      "grad_norm": 1.7880535125732422,
      "learning_rate": 2.802461149507409e-06,
      "loss": 0.4805,
      "step": 227410
    },
    {
      "epoch": 93.09,
      "grad_norm": 2.1088385581970215,
      "learning_rate": 2.8023055867292026e-06,
      "loss": 0.4768,
      "step": 227420
    },
    {
      "epoch": 93.09,
      "grad_norm": 2.0401649475097656,
      "learning_rate": 2.802150021850425e-06,
      "loss": 0.5182,
      "step": 227430
    },
    {
      "epoch": 93.1,
      "grad_norm": 2.1826038360595703,
      "learning_rate": 2.8019944548717867e-06,
      "loss": 0.4686,
      "step": 227440
    },
    {
      "epoch": 93.1,
      "grad_norm": 2.472379446029663,
      "learning_rate": 2.8018388857940023e-06,
      "loss": 0.4853,
      "step": 227450
    },
    {
      "epoch": 93.11,
      "grad_norm": 2.083574056625366,
      "learning_rate": 2.801683314617783e-06,
      "loss": 0.4892,
      "step": 227460
    },
    {
      "epoch": 93.11,
      "grad_norm": 1.9134351015090942,
      "learning_rate": 2.8015277413438428e-06,
      "loss": 0.5066,
      "step": 227470
    },
    {
      "epoch": 93.12,
      "grad_norm": 1.8911341428756714,
      "learning_rate": 2.801372165972894e-06,
      "loss": 0.487,
      "step": 227480
    },
    {
      "epoch": 93.12,
      "grad_norm": 1.755052924156189,
      "learning_rate": 2.8012165885056477e-06,
      "loss": 0.4723,
      "step": 227490
    },
    {
      "epoch": 93.12,
      "grad_norm": 2.2021825313568115,
      "learning_rate": 2.801061008942819e-06,
      "loss": 0.495,
      "step": 227500
    },
    {
      "epoch": 93.13,
      "grad_norm": 2.0720839500427246,
      "learning_rate": 2.800905427285119e-06,
      "loss": 0.5047,
      "step": 227510
    },
    {
      "epoch": 93.13,
      "grad_norm": 1.4938842058181763,
      "learning_rate": 2.800749843533261e-06,
      "loss": 0.4858,
      "step": 227520
    },
    {
      "epoch": 93.14,
      "grad_norm": 1.7026411294937134,
      "learning_rate": 2.8005942576879565e-06,
      "loss": 0.4874,
      "step": 227530
    },
    {
      "epoch": 93.14,
      "grad_norm": 1.8553448915481567,
      "learning_rate": 2.80043866974992e-06,
      "loss": 0.4892,
      "step": 227540
    },
    {
      "epoch": 93.14,
      "grad_norm": 2.1297762393951416,
      "learning_rate": 2.8002830797198635e-06,
      "loss": 0.479,
      "step": 227550
    },
    {
      "epoch": 93.15,
      "grad_norm": 1.5804903507232666,
      "learning_rate": 2.8001274875984995e-06,
      "loss": 0.4928,
      "step": 227560
    },
    {
      "epoch": 93.15,
      "grad_norm": 2.4965837001800537,
      "learning_rate": 2.799971893386541e-06,
      "loss": 0.4903,
      "step": 227570
    },
    {
      "epoch": 93.16,
      "grad_norm": 2.0010457038879395,
      "learning_rate": 2.7998162970847008e-06,
      "loss": 0.4712,
      "step": 227580
    },
    {
      "epoch": 93.16,
      "grad_norm": 2.1949243545532227,
      "learning_rate": 2.7996606986936908e-06,
      "loss": 0.4814,
      "step": 227590
    },
    {
      "epoch": 93.16,
      "grad_norm": 2.089120864868164,
      "learning_rate": 2.799505098214225e-06,
      "loss": 0.4819,
      "step": 227600
    },
    {
      "epoch": 93.17,
      "grad_norm": 1.7001029253005981,
      "learning_rate": 2.799349495647016e-06,
      "loss": 0.4795,
      "step": 227610
    },
    {
      "epoch": 93.17,
      "grad_norm": 1.7062321901321411,
      "learning_rate": 2.799193890992775e-06,
      "loss": 0.5035,
      "step": 227620
    },
    {
      "epoch": 93.18,
      "grad_norm": 1.4841032028198242,
      "learning_rate": 2.799038284252217e-06,
      "loss": 0.4862,
      "step": 227630
    },
    {
      "epoch": 93.18,
      "grad_norm": 1.6805956363677979,
      "learning_rate": 2.7988826754260546e-06,
      "loss": 0.4788,
      "step": 227640
    },
    {
      "epoch": 93.18,
      "grad_norm": 1.4695048332214355,
      "learning_rate": 2.798727064514999e-06,
      "loss": 0.4722,
      "step": 227650
    },
    {
      "epoch": 93.19,
      "grad_norm": 2.0015299320220947,
      "learning_rate": 2.798571451519764e-06,
      "loss": 0.4813,
      "step": 227660
    },
    {
      "epoch": 93.19,
      "grad_norm": 1.957855224609375,
      "learning_rate": 2.7984158364410628e-06,
      "loss": 0.4804,
      "step": 227670
    },
    {
      "epoch": 93.2,
      "grad_norm": 1.7164688110351562,
      "learning_rate": 2.7982602192796074e-06,
      "loss": 0.5088,
      "step": 227680
    },
    {
      "epoch": 93.2,
      "grad_norm": 2.3392488956451416,
      "learning_rate": 2.798104600036111e-06,
      "loss": 0.4727,
      "step": 227690
    },
    {
      "epoch": 93.21,
      "grad_norm": 1.905256748199463,
      "learning_rate": 2.7979489787112865e-06,
      "loss": 0.4957,
      "step": 227700
    },
    {
      "epoch": 93.21,
      "grad_norm": 1.6401530504226685,
      "learning_rate": 2.7977933553058474e-06,
      "loss": 0.4857,
      "step": 227710
    },
    {
      "epoch": 93.21,
      "grad_norm": 1.5863803625106812,
      "learning_rate": 2.7976377298205056e-06,
      "loss": 0.474,
      "step": 227720
    },
    {
      "epoch": 93.22,
      "grad_norm": 1.8490532636642456,
      "learning_rate": 2.797482102255974e-06,
      "loss": 0.5023,
      "step": 227730
    },
    {
      "epoch": 93.22,
      "grad_norm": 1.7371124029159546,
      "learning_rate": 2.797326472612966e-06,
      "loss": 0.4715,
      "step": 227740
    },
    {
      "epoch": 93.23,
      "grad_norm": 1.445264220237732,
      "learning_rate": 2.7971708408921952e-06,
      "loss": 0.4966,
      "step": 227750
    },
    {
      "epoch": 93.23,
      "grad_norm": 1.5038508176803589,
      "learning_rate": 2.7970152070943733e-06,
      "loss": 0.4828,
      "step": 227760
    },
    {
      "epoch": 93.23,
      "grad_norm": 2.0581653118133545,
      "learning_rate": 2.7968595712202136e-06,
      "loss": 0.4845,
      "step": 227770
    },
    {
      "epoch": 93.24,
      "grad_norm": 1.6771647930145264,
      "learning_rate": 2.7967039332704293e-06,
      "loss": 0.4795,
      "step": 227780
    },
    {
      "epoch": 93.24,
      "grad_norm": 1.5759347677230835,
      "learning_rate": 2.796548293245733e-06,
      "loss": 0.4856,
      "step": 227790
    },
    {
      "epoch": 93.25,
      "grad_norm": 2.54270339012146,
      "learning_rate": 2.796392651146838e-06,
      "loss": 0.4822,
      "step": 227800
    },
    {
      "epoch": 93.25,
      "grad_norm": 2.063039541244507,
      "learning_rate": 2.796237006974457e-06,
      "loss": 0.4615,
      "step": 227810
    },
    {
      "epoch": 93.25,
      "grad_norm": 1.6557461023330688,
      "learning_rate": 2.796081360729303e-06,
      "loss": 0.4897,
      "step": 227820
    },
    {
      "epoch": 93.26,
      "grad_norm": 2.281637191772461,
      "learning_rate": 2.7959257124120894e-06,
      "loss": 0.508,
      "step": 227830
    },
    {
      "epoch": 93.26,
      "grad_norm": 1.9194709062576294,
      "learning_rate": 2.7957700620235285e-06,
      "loss": 0.4893,
      "step": 227840
    },
    {
      "epoch": 93.27,
      "grad_norm": 1.9906315803527832,
      "learning_rate": 2.795614409564334e-06,
      "loss": 0.4994,
      "step": 227850
    },
    {
      "epoch": 93.27,
      "grad_norm": 2.3457095623016357,
      "learning_rate": 2.795458755035219e-06,
      "loss": 0.4844,
      "step": 227860
    },
    {
      "epoch": 93.27,
      "grad_norm": 1.8102506399154663,
      "learning_rate": 2.795303098436896e-06,
      "loss": 0.493,
      "step": 227870
    },
    {
      "epoch": 93.28,
      "grad_norm": 2.730867385864258,
      "learning_rate": 2.795147439770078e-06,
      "loss": 0.5022,
      "step": 227880
    },
    {
      "epoch": 93.28,
      "grad_norm": 2.071580648422241,
      "learning_rate": 2.794991779035478e-06,
      "loss": 0.482,
      "step": 227890
    },
    {
      "epoch": 93.29,
      "grad_norm": 2.046024799346924,
      "learning_rate": 2.7948361162338087e-06,
      "loss": 0.4732,
      "step": 227900
    },
    {
      "epoch": 93.29,
      "grad_norm": 2.614933490753174,
      "learning_rate": 2.794680451365784e-06,
      "loss": 0.4929,
      "step": 227910
    },
    {
      "epoch": 93.3,
      "grad_norm": 1.7139403820037842,
      "learning_rate": 2.794524784432118e-06,
      "loss": 0.4863,
      "step": 227920
    },
    {
      "epoch": 93.3,
      "grad_norm": 1.7738527059555054,
      "learning_rate": 2.794369115433522e-06,
      "loss": 0.4899,
      "step": 227930
    },
    {
      "epoch": 93.3,
      "grad_norm": 2.085624933242798,
      "learning_rate": 2.794213444370709e-06,
      "loss": 0.4796,
      "step": 227940
    },
    {
      "epoch": 93.31,
      "grad_norm": 1.6382606029510498,
      "learning_rate": 2.7940577712443934e-06,
      "loss": 0.4879,
      "step": 227950
    },
    {
      "epoch": 93.31,
      "grad_norm": 2.146913766860962,
      "learning_rate": 2.793902096055287e-06,
      "loss": 0.4985,
      "step": 227960
    },
    {
      "epoch": 93.32,
      "grad_norm": 1.8355052471160889,
      "learning_rate": 2.793746418804104e-06,
      "loss": 0.4865,
      "step": 227970
    },
    {
      "epoch": 93.32,
      "grad_norm": 2.169728994369507,
      "learning_rate": 2.793590739491557e-06,
      "loss": 0.5002,
      "step": 227980
    },
    {
      "epoch": 93.32,
      "grad_norm": 1.9332703351974487,
      "learning_rate": 2.7934350581183586e-06,
      "loss": 0.4814,
      "step": 227990
    },
    {
      "epoch": 93.33,
      "grad_norm": 2.143795967102051,
      "learning_rate": 2.7932793746852238e-06,
      "loss": 0.4601,
      "step": 228000
    },
    {
      "epoch": 93.33,
      "grad_norm": 1.7037413120269775,
      "learning_rate": 2.793123689192864e-06,
      "loss": 0.4804,
      "step": 228010
    },
    {
      "epoch": 93.34,
      "grad_norm": 1.938334345817566,
      "learning_rate": 2.792968001641993e-06,
      "loss": 0.4849,
      "step": 228020
    },
    {
      "epoch": 93.34,
      "grad_norm": 1.815853238105774,
      "learning_rate": 2.7928123120333235e-06,
      "loss": 0.4796,
      "step": 228030
    },
    {
      "epoch": 93.34,
      "grad_norm": 1.949898362159729,
      "learning_rate": 2.7926566203675697e-06,
      "loss": 0.5036,
      "step": 228040
    },
    {
      "epoch": 93.35,
      "grad_norm": 1.8422983884811401,
      "learning_rate": 2.7925009266454435e-06,
      "loss": 0.4871,
      "step": 228050
    },
    {
      "epoch": 93.35,
      "grad_norm": 1.6144511699676514,
      "learning_rate": 2.7923452308676594e-06,
      "loss": 0.4806,
      "step": 228060
    },
    {
      "epoch": 93.36,
      "grad_norm": 1.9856398105621338,
      "learning_rate": 2.7921895330349297e-06,
      "loss": 0.4956,
      "step": 228070
    },
    {
      "epoch": 93.36,
      "grad_norm": 1.9871232509613037,
      "learning_rate": 2.792033833147968e-06,
      "loss": 0.5051,
      "step": 228080
    },
    {
      "epoch": 93.36,
      "grad_norm": 3.2961742877960205,
      "learning_rate": 2.7918781312074874e-06,
      "loss": 0.4844,
      "step": 228090
    },
    {
      "epoch": 93.37,
      "grad_norm": 1.7572238445281982,
      "learning_rate": 2.7917224272142015e-06,
      "loss": 0.4696,
      "step": 228100
    },
    {
      "epoch": 93.37,
      "grad_norm": 2.295046329498291,
      "learning_rate": 2.7915667211688234e-06,
      "loss": 0.5032,
      "step": 228110
    },
    {
      "epoch": 93.38,
      "grad_norm": 1.6615536212921143,
      "learning_rate": 2.7914110130720657e-06,
      "loss": 0.4893,
      "step": 228120
    },
    {
      "epoch": 93.38,
      "grad_norm": 1.8214763402938843,
      "learning_rate": 2.7912553029246425e-06,
      "loss": 0.4652,
      "step": 228130
    },
    {
      "epoch": 93.39,
      "grad_norm": 2.0329363346099854,
      "learning_rate": 2.791099590727267e-06,
      "loss": 0.4825,
      "step": 228140
    },
    {
      "epoch": 93.39,
      "grad_norm": 1.5966312885284424,
      "learning_rate": 2.7909438764806524e-06,
      "loss": 0.4774,
      "step": 228150
    },
    {
      "epoch": 93.39,
      "grad_norm": 1.8866560459136963,
      "learning_rate": 2.790788160185512e-06,
      "loss": 0.4924,
      "step": 228160
    },
    {
      "epoch": 93.4,
      "grad_norm": 2.897831916809082,
      "learning_rate": 2.7906324418425583e-06,
      "loss": 0.4832,
      "step": 228170
    },
    {
      "epoch": 93.4,
      "grad_norm": 1.9923490285873413,
      "learning_rate": 2.7904767214525053e-06,
      "loss": 0.4867,
      "step": 228180
    },
    {
      "epoch": 93.41,
      "grad_norm": 1.8001924753189087,
      "learning_rate": 2.7903209990160668e-06,
      "loss": 0.4905,
      "step": 228190
    },
    {
      "epoch": 93.41,
      "grad_norm": 1.9668279886245728,
      "learning_rate": 2.790165274533956e-06,
      "loss": 0.4983,
      "step": 228200
    },
    {
      "epoch": 93.41,
      "grad_norm": 1.9907596111297607,
      "learning_rate": 2.790009548006886e-06,
      "loss": 0.4825,
      "step": 228210
    },
    {
      "epoch": 93.42,
      "grad_norm": 3.121671199798584,
      "learning_rate": 2.78985381943557e-06,
      "loss": 0.5007,
      "step": 228220
    },
    {
      "epoch": 93.42,
      "grad_norm": 1.9152591228485107,
      "learning_rate": 2.7896980888207207e-06,
      "loss": 0.4952,
      "step": 228230
    },
    {
      "epoch": 93.43,
      "grad_norm": 2.169051170349121,
      "learning_rate": 2.7895423561630535e-06,
      "loss": 0.5047,
      "step": 228240
    },
    {
      "epoch": 93.43,
      "grad_norm": 2.7562859058380127,
      "learning_rate": 2.78938662146328e-06,
      "loss": 0.4785,
      "step": 228250
    },
    {
      "epoch": 93.43,
      "grad_norm": 1.9837526082992554,
      "learning_rate": 2.789230884722114e-06,
      "loss": 0.4668,
      "step": 228260
    },
    {
      "epoch": 93.44,
      "grad_norm": 1.7869081497192383,
      "learning_rate": 2.7890751459402685e-06,
      "loss": 0.4856,
      "step": 228270
    },
    {
      "epoch": 93.44,
      "grad_norm": 2.233893394470215,
      "learning_rate": 2.7889194051184576e-06,
      "loss": 0.4703,
      "step": 228280
    },
    {
      "epoch": 93.45,
      "grad_norm": 2.2071826457977295,
      "learning_rate": 2.788763662257396e-06,
      "loss": 0.4976,
      "step": 228290
    },
    {
      "epoch": 93.45,
      "grad_norm": 1.835456132888794,
      "learning_rate": 2.788607917357795e-06,
      "loss": 0.481,
      "step": 228300
    },
    {
      "epoch": 93.45,
      "grad_norm": 1.9761779308319092,
      "learning_rate": 2.7884521704203685e-06,
      "loss": 0.5104,
      "step": 228310
    },
    {
      "epoch": 93.46,
      "grad_norm": 1.778157114982605,
      "learning_rate": 2.78829642144583e-06,
      "loss": 0.4504,
      "step": 228320
    },
    {
      "epoch": 93.46,
      "grad_norm": 1.8676972389221191,
      "learning_rate": 2.788140670434894e-06,
      "loss": 0.4838,
      "step": 228330
    },
    {
      "epoch": 93.47,
      "grad_norm": 1.998974323272705,
      "learning_rate": 2.7879849173882723e-06,
      "loss": 0.4766,
      "step": 228340
    },
    {
      "epoch": 93.47,
      "grad_norm": 2.2563557624816895,
      "learning_rate": 2.7878291623066796e-06,
      "loss": 0.501,
      "step": 228350
    },
    {
      "epoch": 93.48,
      "grad_norm": 2.0607731342315674,
      "learning_rate": 2.7876734051908293e-06,
      "loss": 0.4885,
      "step": 228360
    },
    {
      "epoch": 93.48,
      "grad_norm": 1.6495226621627808,
      "learning_rate": 2.787517646041434e-06,
      "loss": 0.4835,
      "step": 228370
    },
    {
      "epoch": 93.48,
      "grad_norm": 1.7939891815185547,
      "learning_rate": 2.7873618848592086e-06,
      "loss": 0.4776,
      "step": 228380
    },
    {
      "epoch": 93.49,
      "grad_norm": 1.6831119060516357,
      "learning_rate": 2.7872061216448655e-06,
      "loss": 0.493,
      "step": 228390
    },
    {
      "epoch": 93.49,
      "grad_norm": 1.6913352012634277,
      "learning_rate": 2.7870503563991184e-06,
      "loss": 0.4811,
      "step": 228400
    },
    {
      "epoch": 93.5,
      "grad_norm": 1.9917722940444946,
      "learning_rate": 2.786894589122681e-06,
      "loss": 0.4732,
      "step": 228410
    },
    {
      "epoch": 93.5,
      "grad_norm": 1.9363842010498047,
      "learning_rate": 2.7867388198162673e-06,
      "loss": 0.4845,
      "step": 228420
    },
    {
      "epoch": 93.5,
      "grad_norm": 2.3214826583862305,
      "learning_rate": 2.78658304848059e-06,
      "loss": 0.462,
      "step": 228430
    },
    {
      "epoch": 93.51,
      "grad_norm": 2.667163610458374,
      "learning_rate": 2.7864272751163633e-06,
      "loss": 0.4746,
      "step": 228440
    },
    {
      "epoch": 93.51,
      "grad_norm": 1.6906590461730957,
      "learning_rate": 2.786271499724301e-06,
      "loss": 0.4679,
      "step": 228450
    },
    {
      "epoch": 93.52,
      "grad_norm": 1.9135533571243286,
      "learning_rate": 2.7861157223051156e-06,
      "loss": 0.4899,
      "step": 228460
    },
    {
      "epoch": 93.52,
      "grad_norm": 1.601036548614502,
      "learning_rate": 2.7859599428595214e-06,
      "loss": 0.4828,
      "step": 228470
    },
    {
      "epoch": 93.52,
      "grad_norm": 1.646161675453186,
      "learning_rate": 2.785804161388232e-06,
      "loss": 0.4809,
      "step": 228480
    },
    {
      "epoch": 93.53,
      "grad_norm": 2.1744492053985596,
      "learning_rate": 2.785648377891961e-06,
      "loss": 0.4699,
      "step": 228490
    },
    {
      "epoch": 93.53,
      "grad_norm": 1.7210239171981812,
      "learning_rate": 2.785492592371422e-06,
      "loss": 0.488,
      "step": 228500
    },
    {
      "epoch": 93.54,
      "grad_norm": 1.4793922901153564,
      "learning_rate": 2.7853368048273283e-06,
      "loss": 0.5026,
      "step": 228510
    },
    {
      "epoch": 93.54,
      "grad_norm": 2.0474908351898193,
      "learning_rate": 2.7851810152603947e-06,
      "loss": 0.4747,
      "step": 228520
    },
    {
      "epoch": 93.54,
      "grad_norm": 1.75483238697052,
      "learning_rate": 2.785025223671333e-06,
      "loss": 0.4906,
      "step": 228530
    },
    {
      "epoch": 93.55,
      "grad_norm": 1.5991426706314087,
      "learning_rate": 2.7848694300608585e-06,
      "loss": 0.4624,
      "step": 228540
    },
    {
      "epoch": 93.55,
      "grad_norm": 1.9189679622650146,
      "learning_rate": 2.784713634429684e-06,
      "loss": 0.486,
      "step": 228550
    },
    {
      "epoch": 93.56,
      "grad_norm": 2.1422030925750732,
      "learning_rate": 2.7845578367785227e-06,
      "loss": 0.4935,
      "step": 228560
    },
    {
      "epoch": 93.56,
      "grad_norm": 1.347328543663025,
      "learning_rate": 2.7844020371080895e-06,
      "loss": 0.4706,
      "step": 228570
    },
    {
      "epoch": 93.57,
      "grad_norm": 1.783410668373108,
      "learning_rate": 2.784246235419098e-06,
      "loss": 0.4897,
      "step": 228580
    },
    {
      "epoch": 93.57,
      "grad_norm": 1.8296220302581787,
      "learning_rate": 2.784090431712261e-06,
      "loss": 0.4761,
      "step": 228590
    },
    {
      "epoch": 93.57,
      "grad_norm": 1.80812668800354,
      "learning_rate": 2.7839346259882933e-06,
      "loss": 0.49,
      "step": 228600
    },
    {
      "epoch": 93.58,
      "grad_norm": 1.5264836549758911,
      "learning_rate": 2.783778818247907e-06,
      "loss": 0.4826,
      "step": 228610
    },
    {
      "epoch": 93.58,
      "grad_norm": 1.8357057571411133,
      "learning_rate": 2.7836230084918178e-06,
      "loss": 0.4828,
      "step": 228620
    },
    {
      "epoch": 93.59,
      "grad_norm": 1.9132566452026367,
      "learning_rate": 2.7834671967207375e-06,
      "loss": 0.4865,
      "step": 228630
    },
    {
      "epoch": 93.59,
      "grad_norm": 2.3628122806549072,
      "learning_rate": 2.7833113829353815e-06,
      "loss": 0.4833,
      "step": 228640
    },
    {
      "epoch": 93.59,
      "grad_norm": 1.9056942462921143,
      "learning_rate": 2.7831555671364623e-06,
      "loss": 0.4924,
      "step": 228650
    },
    {
      "epoch": 93.6,
      "grad_norm": 1.4716545343399048,
      "learning_rate": 2.782999749324695e-06,
      "loss": 0.4982,
      "step": 228660
    },
    {
      "epoch": 93.6,
      "grad_norm": 2.195047378540039,
      "learning_rate": 2.782843929500792e-06,
      "loss": 0.4671,
      "step": 228670
    },
    {
      "epoch": 93.61,
      "grad_norm": 2.0690267086029053,
      "learning_rate": 2.782688107665468e-06,
      "loss": 0.4765,
      "step": 228680
    },
    {
      "epoch": 93.61,
      "grad_norm": 2.2206220626831055,
      "learning_rate": 2.7825322838194366e-06,
      "loss": 0.4845,
      "step": 228690
    },
    {
      "epoch": 93.61,
      "grad_norm": 2.0276987552642822,
      "learning_rate": 2.782376457963411e-06,
      "loss": 0.4884,
      "step": 228700
    },
    {
      "epoch": 93.62,
      "grad_norm": 1.4980964660644531,
      "learning_rate": 2.7822206300981058e-06,
      "loss": 0.4789,
      "step": 228710
    },
    {
      "epoch": 93.62,
      "grad_norm": 1.7006324529647827,
      "learning_rate": 2.7820648002242353e-06,
      "loss": 0.4669,
      "step": 228720
    },
    {
      "epoch": 93.63,
      "grad_norm": 1.7622499465942383,
      "learning_rate": 2.7819089683425117e-06,
      "loss": 0.4947,
      "step": 228730
    },
    {
      "epoch": 93.63,
      "grad_norm": 2.0240581035614014,
      "learning_rate": 2.7817531344536503e-06,
      "loss": 0.5043,
      "step": 228740
    },
    {
      "epoch": 93.63,
      "grad_norm": 1.8928678035736084,
      "learning_rate": 2.7815972985583636e-06,
      "loss": 0.4841,
      "step": 228750
    },
    {
      "epoch": 93.64,
      "grad_norm": 1.8389461040496826,
      "learning_rate": 2.781441460657367e-06,
      "loss": 0.4923,
      "step": 228760
    },
    {
      "epoch": 93.64,
      "grad_norm": 2.41580867767334,
      "learning_rate": 2.781285620751373e-06,
      "loss": 0.4781,
      "step": 228770
    },
    {
      "epoch": 93.65,
      "grad_norm": 1.2887903451919556,
      "learning_rate": 2.7811297788410963e-06,
      "loss": 0.4827,
      "step": 228780
    },
    {
      "epoch": 93.65,
      "grad_norm": 1.6962703466415405,
      "learning_rate": 2.7809739349272506e-06,
      "loss": 0.4985,
      "step": 228790
    },
    {
      "epoch": 93.66,
      "grad_norm": 1.885820746421814,
      "learning_rate": 2.7808180890105498e-06,
      "loss": 0.4728,
      "step": 228800
    },
    {
      "epoch": 93.66,
      "grad_norm": 1.6235851049423218,
      "learning_rate": 2.780662241091708e-06,
      "loss": 0.4906,
      "step": 228810
    },
    {
      "epoch": 93.66,
      "grad_norm": 1.919592022895813,
      "learning_rate": 2.7805063911714385e-06,
      "loss": 0.4965,
      "step": 228820
    },
    {
      "epoch": 93.67,
      "grad_norm": 2.417804002761841,
      "learning_rate": 2.7803505392504557e-06,
      "loss": 0.4974,
      "step": 228830
    },
    {
      "epoch": 93.67,
      "grad_norm": 2.0656352043151855,
      "learning_rate": 2.7801946853294735e-06,
      "loss": 0.4805,
      "step": 228840
    },
    {
      "epoch": 93.68,
      "grad_norm": 2.763275623321533,
      "learning_rate": 2.7800388294092056e-06,
      "loss": 0.4796,
      "step": 228850
    },
    {
      "epoch": 93.68,
      "grad_norm": 2.3968801498413086,
      "learning_rate": 2.7798829714903664e-06,
      "loss": 0.4877,
      "step": 228860
    },
    {
      "epoch": 93.68,
      "grad_norm": 2.106273889541626,
      "learning_rate": 2.7797271115736696e-06,
      "loss": 0.4675,
      "step": 228870
    },
    {
      "epoch": 93.69,
      "grad_norm": 1.518265962600708,
      "learning_rate": 2.779571249659829e-06,
      "loss": 0.4912,
      "step": 228880
    },
    {
      "epoch": 93.69,
      "grad_norm": 1.948506236076355,
      "learning_rate": 2.779415385749559e-06,
      "loss": 0.4837,
      "step": 228890
    },
    {
      "epoch": 93.7,
      "grad_norm": 1.9590387344360352,
      "learning_rate": 2.7792595198435737e-06,
      "loss": 0.463,
      "step": 228900
    },
    {
      "epoch": 93.7,
      "grad_norm": 2.108811616897583,
      "learning_rate": 2.779103651942586e-06,
      "loss": 0.4886,
      "step": 228910
    },
    {
      "epoch": 93.7,
      "grad_norm": 2.1423957347869873,
      "learning_rate": 2.7789477820473107e-06,
      "loss": 0.4965,
      "step": 228920
    },
    {
      "epoch": 93.71,
      "grad_norm": 2.3365366458892822,
      "learning_rate": 2.778791910158462e-06,
      "loss": 0.4859,
      "step": 228930
    },
    {
      "epoch": 93.71,
      "grad_norm": 1.6938183307647705,
      "learning_rate": 2.778636036276754e-06,
      "loss": 0.4929,
      "step": 228940
    },
    {
      "epoch": 93.72,
      "grad_norm": 2.1696884632110596,
      "learning_rate": 2.7784801604028995e-06,
      "loss": 0.4957,
      "step": 228950
    },
    {
      "epoch": 93.72,
      "grad_norm": 2.0454673767089844,
      "learning_rate": 2.7783242825376143e-06,
      "loss": 0.4918,
      "step": 228960
    },
    {
      "epoch": 93.72,
      "grad_norm": 2.408799409866333,
      "learning_rate": 2.778168402681611e-06,
      "loss": 0.4895,
      "step": 228970
    },
    {
      "epoch": 93.73,
      "grad_norm": 1.469132423400879,
      "learning_rate": 2.778012520835605e-06,
      "loss": 0.4949,
      "step": 228980
    },
    {
      "epoch": 93.73,
      "grad_norm": 1.5564006567001343,
      "learning_rate": 2.777856637000309e-06,
      "loss": 0.4612,
      "step": 228990
    },
    {
      "epoch": 93.74,
      "grad_norm": 1.4500951766967773,
      "learning_rate": 2.7777007511764378e-06,
      "loss": 0.4569,
      "step": 229000
    },
    {
      "epoch": 93.74,
      "grad_norm": 1.7343710660934448,
      "learning_rate": 2.777544863364706e-06,
      "loss": 0.498,
      "step": 229010
    },
    {
      "epoch": 93.75,
      "grad_norm": 2.2283883094787598,
      "learning_rate": 2.7773889735658265e-06,
      "loss": 0.492,
      "step": 229020
    },
    {
      "epoch": 93.75,
      "grad_norm": 1.9287915229797363,
      "learning_rate": 2.777233081780514e-06,
      "loss": 0.502,
      "step": 229030
    },
    {
      "epoch": 93.75,
      "grad_norm": 1.9478830099105835,
      "learning_rate": 2.777077188009483e-06,
      "loss": 0.4892,
      "step": 229040
    },
    {
      "epoch": 93.76,
      "grad_norm": 1.9122527837753296,
      "learning_rate": 2.7769212922534473e-06,
      "loss": 0.4946,
      "step": 229050
    },
    {
      "epoch": 93.76,
      "grad_norm": 1.65111243724823,
      "learning_rate": 2.7767653945131206e-06,
      "loss": 0.4794,
      "step": 229060
    },
    {
      "epoch": 93.77,
      "grad_norm": 2.1025099754333496,
      "learning_rate": 2.776609494789218e-06,
      "loss": 0.5112,
      "step": 229070
    },
    {
      "epoch": 93.77,
      "grad_norm": 1.7843601703643799,
      "learning_rate": 2.7764535930824524e-06,
      "loss": 0.4926,
      "step": 229080
    },
    {
      "epoch": 93.77,
      "grad_norm": 2.064838409423828,
      "learning_rate": 2.7762976893935387e-06,
      "loss": 0.4755,
      "step": 229090
    },
    {
      "epoch": 93.78,
      "grad_norm": 1.7546970844268799,
      "learning_rate": 2.776141783723192e-06,
      "loss": 0.4648,
      "step": 229100
    },
    {
      "epoch": 93.78,
      "grad_norm": 1.644691824913025,
      "learning_rate": 2.7759858760721246e-06,
      "loss": 0.5115,
      "step": 229110
    },
    {
      "epoch": 93.79,
      "grad_norm": 2.0169241428375244,
      "learning_rate": 2.7758299664410516e-06,
      "loss": 0.4757,
      "step": 229120
    },
    {
      "epoch": 93.79,
      "grad_norm": 2.0775747299194336,
      "learning_rate": 2.775674054830687e-06,
      "loss": 0.4789,
      "step": 229130
    },
    {
      "epoch": 93.79,
      "grad_norm": 1.8915704488754272,
      "learning_rate": 2.775518141241746e-06,
      "loss": 0.4742,
      "step": 229140
    },
    {
      "epoch": 93.8,
      "grad_norm": 1.9538480043411255,
      "learning_rate": 2.7753622256749415e-06,
      "loss": 0.4782,
      "step": 229150
    },
    {
      "epoch": 93.8,
      "grad_norm": 2.8248603343963623,
      "learning_rate": 2.7752063081309883e-06,
      "loss": 0.4889,
      "step": 229160
    },
    {
      "epoch": 93.81,
      "grad_norm": 1.6436686515808105,
      "learning_rate": 2.7750503886106014e-06,
      "loss": 0.4713,
      "step": 229170
    },
    {
      "epoch": 93.81,
      "grad_norm": 2.0276596546173096,
      "learning_rate": 2.7748944671144933e-06,
      "loss": 0.4783,
      "step": 229180
    },
    {
      "epoch": 93.81,
      "grad_norm": 2.1950252056121826,
      "learning_rate": 2.7747385436433787e-06,
      "loss": 0.4756,
      "step": 229190
    },
    {
      "epoch": 93.82,
      "grad_norm": 2.0003597736358643,
      "learning_rate": 2.7745826181979735e-06,
      "loss": 0.487,
      "step": 229200
    },
    {
      "epoch": 93.82,
      "grad_norm": 1.6491549015045166,
      "learning_rate": 2.77442669077899e-06,
      "loss": 0.4776,
      "step": 229210
    },
    {
      "epoch": 93.83,
      "grad_norm": 1.9965358972549438,
      "learning_rate": 2.774270761387143e-06,
      "loss": 0.4789,
      "step": 229220
    },
    {
      "epoch": 93.83,
      "grad_norm": 1.9583590030670166,
      "learning_rate": 2.774114830023148e-06,
      "loss": 0.4833,
      "step": 229230
    },
    {
      "epoch": 93.84,
      "grad_norm": 2.125584602355957,
      "learning_rate": 2.773958896687718e-06,
      "loss": 0.4817,
      "step": 229240
    },
    {
      "epoch": 93.84,
      "grad_norm": 2.090322494506836,
      "learning_rate": 2.7738029613815677e-06,
      "loss": 0.4888,
      "step": 229250
    },
    {
      "epoch": 93.84,
      "grad_norm": 1.8759033679962158,
      "learning_rate": 2.7736470241054115e-06,
      "loss": 0.4824,
      "step": 229260
    },
    {
      "epoch": 93.85,
      "grad_norm": 1.7474685907363892,
      "learning_rate": 2.773491084859963e-06,
      "loss": 0.4724,
      "step": 229270
    },
    {
      "epoch": 93.85,
      "grad_norm": 1.6440030336380005,
      "learning_rate": 2.7733351436459375e-06,
      "loss": 0.4982,
      "step": 229280
    },
    {
      "epoch": 93.86,
      "grad_norm": 1.7581011056900024,
      "learning_rate": 2.773179200464049e-06,
      "loss": 0.4643,
      "step": 229290
    },
    {
      "epoch": 93.86,
      "grad_norm": 1.8108693361282349,
      "learning_rate": 2.7730232553150117e-06,
      "loss": 0.4939,
      "step": 229300
    },
    {
      "epoch": 93.86,
      "grad_norm": 2.0119760036468506,
      "learning_rate": 2.772867308199541e-06,
      "loss": 0.4884,
      "step": 229310
    },
    {
      "epoch": 93.87,
      "grad_norm": 1.9425108432769775,
      "learning_rate": 2.7727113591183495e-06,
      "loss": 0.4645,
      "step": 229320
    },
    {
      "epoch": 93.87,
      "grad_norm": 1.9288221597671509,
      "learning_rate": 2.7725554080721523e-06,
      "loss": 0.5003,
      "step": 229330
    },
    {
      "epoch": 93.88,
      "grad_norm": 1.7261335849761963,
      "learning_rate": 2.772399455061664e-06,
      "loss": 0.4795,
      "step": 229340
    },
    {
      "epoch": 93.88,
      "grad_norm": 1.7722392082214355,
      "learning_rate": 2.7722435000875995e-06,
      "loss": 0.4717,
      "step": 229350
    },
    {
      "epoch": 93.88,
      "grad_norm": 1.8065439462661743,
      "learning_rate": 2.7720875431506723e-06,
      "loss": 0.476,
      "step": 229360
    },
    {
      "epoch": 93.89,
      "grad_norm": 2.0728888511657715,
      "learning_rate": 2.7719315842515966e-06,
      "loss": 0.4933,
      "step": 229370
    },
    {
      "epoch": 93.89,
      "grad_norm": 2.0233747959136963,
      "learning_rate": 2.7717756233910883e-06,
      "loss": 0.4868,
      "step": 229380
    },
    {
      "epoch": 93.9,
      "grad_norm": 1.749098300933838,
      "learning_rate": 2.771619660569861e-06,
      "loss": 0.4899,
      "step": 229390
    },
    {
      "epoch": 93.9,
      "grad_norm": 1.8894476890563965,
      "learning_rate": 2.7714636957886285e-06,
      "loss": 0.5008,
      "step": 229400
    },
    {
      "epoch": 93.91,
      "grad_norm": 1.6983665227890015,
      "learning_rate": 2.7713077290481056e-06,
      "loss": 0.4995,
      "step": 229410
    },
    {
      "epoch": 93.91,
      "grad_norm": 2.0621304512023926,
      "learning_rate": 2.771151760349007e-06,
      "loss": 0.4722,
      "step": 229420
    },
    {
      "epoch": 93.91,
      "grad_norm": 2.106285333633423,
      "learning_rate": 2.770995789692047e-06,
      "loss": 0.488,
      "step": 229430
    },
    {
      "epoch": 93.92,
      "grad_norm": 2.3139419555664062,
      "learning_rate": 2.7708398170779407e-06,
      "loss": 0.5097,
      "step": 229440
    },
    {
      "epoch": 93.92,
      "grad_norm": 1.8717573881149292,
      "learning_rate": 2.770683842507402e-06,
      "loss": 0.4889,
      "step": 229450
    },
    {
      "epoch": 93.93,
      "grad_norm": 1.9672759771347046,
      "learning_rate": 2.770527865981146e-06,
      "loss": 0.4589,
      "step": 229460
    },
    {
      "epoch": 93.93,
      "grad_norm": 1.6504350900650024,
      "learning_rate": 2.770371887499886e-06,
      "loss": 0.4994,
      "step": 229470
    },
    {
      "epoch": 93.93,
      "grad_norm": 2.3173909187316895,
      "learning_rate": 2.770215907064337e-06,
      "loss": 0.4879,
      "step": 229480
    },
    {
      "epoch": 93.94,
      "grad_norm": 1.7194730043411255,
      "learning_rate": 2.7700599246752143e-06,
      "loss": 0.4995,
      "step": 229490
    },
    {
      "epoch": 93.94,
      "grad_norm": 1.5659281015396118,
      "learning_rate": 2.7699039403332315e-06,
      "loss": 0.5015,
      "step": 229500
    },
    {
      "epoch": 93.95,
      "grad_norm": 2.5492336750030518,
      "learning_rate": 2.7697479540391037e-06,
      "loss": 0.4865,
      "step": 229510
    },
    {
      "epoch": 93.95,
      "grad_norm": 2.116950273513794,
      "learning_rate": 2.7695919657935446e-06,
      "loss": 0.4913,
      "step": 229520
    },
    {
      "epoch": 93.95,
      "grad_norm": 1.8866726160049438,
      "learning_rate": 2.7694359755972707e-06,
      "loss": 0.5036,
      "step": 229530
    },
    {
      "epoch": 93.96,
      "grad_norm": 1.8337152004241943,
      "learning_rate": 2.7692799834509945e-06,
      "loss": 0.4725,
      "step": 229540
    },
    {
      "epoch": 93.96,
      "grad_norm": 1.707342505455017,
      "learning_rate": 2.769123989355431e-06,
      "loss": 0.5102,
      "step": 229550
    },
    {
      "epoch": 93.97,
      "grad_norm": 1.8031325340270996,
      "learning_rate": 2.768967993311296e-06,
      "loss": 0.4835,
      "step": 229560
    },
    {
      "epoch": 93.97,
      "grad_norm": 2.6268463134765625,
      "learning_rate": 2.7688119953193023e-06,
      "loss": 0.4992,
      "step": 229570
    },
    {
      "epoch": 93.97,
      "grad_norm": 1.820980429649353,
      "learning_rate": 2.768655995380166e-06,
      "loss": 0.4767,
      "step": 229580
    },
    {
      "epoch": 93.98,
      "grad_norm": 1.8461987972259521,
      "learning_rate": 2.768499993494601e-06,
      "loss": 0.4932,
      "step": 229590
    },
    {
      "epoch": 93.98,
      "grad_norm": 1.9826648235321045,
      "learning_rate": 2.768343989663322e-06,
      "loss": 0.4806,
      "step": 229600
    },
    {
      "epoch": 93.99,
      "grad_norm": 1.8744138479232788,
      "learning_rate": 2.7681879838870436e-06,
      "loss": 0.4682,
      "step": 229610
    },
    {
      "epoch": 93.99,
      "grad_norm": 1.6168311834335327,
      "learning_rate": 2.768031976166481e-06,
      "loss": 0.493,
      "step": 229620
    },
    {
      "epoch": 94.0,
      "grad_norm": 1.5107427835464478,
      "learning_rate": 2.767875966502348e-06,
      "loss": 0.4702,
      "step": 229630
    },
    {
      "epoch": 94.0,
      "grad_norm": 2.322699546813965,
      "learning_rate": 2.76771995489536e-06,
      "loss": 0.4792,
      "step": 229640
    },
    {
      "epoch": 94.0,
      "eval_loss": 0.48769110441207886,
      "eval_runtime": 58.9756,
      "eval_samples_per_second": 58.482,
      "eval_steps_per_second": 7.325,
      "step": 229642
    },
    {
      "epoch": 94.0,
      "grad_norm": 1.9748939275741577,
      "learning_rate": 2.7675639413462306e-06,
      "loss": 0.501,
      "step": 229650
    },
    {
      "epoch": 94.01,
      "grad_norm": 2.1846799850463867,
      "learning_rate": 2.7674079258556755e-06,
      "loss": 0.4878,
      "step": 229660
    },
    {
      "epoch": 94.01,
      "grad_norm": 1.52445650100708,
      "learning_rate": 2.7672519084244095e-06,
      "loss": 0.4779,
      "step": 229670
    },
    {
      "epoch": 94.02,
      "grad_norm": 2.005229949951172,
      "learning_rate": 2.7670958890531466e-06,
      "loss": 0.4927,
      "step": 229680
    },
    {
      "epoch": 94.02,
      "grad_norm": 1.9456419944763184,
      "learning_rate": 2.7669398677426014e-06,
      "loss": 0.4881,
      "step": 229690
    },
    {
      "epoch": 94.02,
      "grad_norm": 1.8043725490570068,
      "learning_rate": 2.766783844493489e-06,
      "loss": 0.4726,
      "step": 229700
    },
    {
      "epoch": 94.03,
      "grad_norm": 1.829648494720459,
      "learning_rate": 2.766627819306524e-06,
      "loss": 0.4782,
      "step": 229710
    },
    {
      "epoch": 94.03,
      "grad_norm": 1.8196223974227905,
      "learning_rate": 2.7664717921824215e-06,
      "loss": 0.4764,
      "step": 229720
    },
    {
      "epoch": 94.04,
      "grad_norm": 1.6055595874786377,
      "learning_rate": 2.766315763121896e-06,
      "loss": 0.4947,
      "step": 229730
    },
    {
      "epoch": 94.04,
      "grad_norm": 2.080641031265259,
      "learning_rate": 2.7661597321256623e-06,
      "loss": 0.4843,
      "step": 229740
    },
    {
      "epoch": 94.04,
      "grad_norm": 1.8206381797790527,
      "learning_rate": 2.766003699194435e-06,
      "loss": 0.4739,
      "step": 229750
    },
    {
      "epoch": 94.05,
      "grad_norm": 2.003638982772827,
      "learning_rate": 2.7658476643289287e-06,
      "loss": 0.4695,
      "step": 229760
    },
    {
      "epoch": 94.05,
      "grad_norm": 1.7368611097335815,
      "learning_rate": 2.765691627529858e-06,
      "loss": 0.4848,
      "step": 229770
    },
    {
      "epoch": 94.06,
      "grad_norm": 2.0856990814208984,
      "learning_rate": 2.765535588797938e-06,
      "loss": 0.4875,
      "step": 229780
    },
    {
      "epoch": 94.06,
      "grad_norm": 1.5348269939422607,
      "learning_rate": 2.7653795481338843e-06,
      "loss": 0.48,
      "step": 229790
    },
    {
      "epoch": 94.06,
      "grad_norm": 2.5873517990112305,
      "learning_rate": 2.7652235055384104e-06,
      "loss": 0.4639,
      "step": 229800
    },
    {
      "epoch": 94.07,
      "grad_norm": 1.669182300567627,
      "learning_rate": 2.765067461012232e-06,
      "loss": 0.4848,
      "step": 229810
    },
    {
      "epoch": 94.07,
      "grad_norm": 2.479536771774292,
      "learning_rate": 2.7649114145560637e-06,
      "loss": 0.513,
      "step": 229820
    },
    {
      "epoch": 94.08,
      "grad_norm": 2.134310483932495,
      "learning_rate": 2.76475536617062e-06,
      "loss": 0.5049,
      "step": 229830
    },
    {
      "epoch": 94.08,
      "grad_norm": 1.7962805032730103,
      "learning_rate": 2.7645993158566158e-06,
      "loss": 0.5073,
      "step": 229840
    },
    {
      "epoch": 94.09,
      "grad_norm": 1.9349671602249146,
      "learning_rate": 2.764443263614766e-06,
      "loss": 0.473,
      "step": 229850
    },
    {
      "epoch": 94.09,
      "grad_norm": 1.620646595954895,
      "learning_rate": 2.7642872094457854e-06,
      "loss": 0.4934,
      "step": 229860
    },
    {
      "epoch": 94.09,
      "grad_norm": 1.9219720363616943,
      "learning_rate": 2.7641311533503893e-06,
      "loss": 0.4875,
      "step": 229870
    },
    {
      "epoch": 94.1,
      "grad_norm": 1.9572325944900513,
      "learning_rate": 2.7639750953292918e-06,
      "loss": 0.4812,
      "step": 229880
    },
    {
      "epoch": 94.1,
      "grad_norm": 1.9384257793426514,
      "learning_rate": 2.763819035383209e-06,
      "loss": 0.4836,
      "step": 229890
    },
    {
      "epoch": 94.11,
      "grad_norm": 1.9301608800888062,
      "learning_rate": 2.763662973512854e-06,
      "loss": 0.493,
      "step": 229900
    },
    {
      "epoch": 94.11,
      "grad_norm": 1.8147048950195312,
      "learning_rate": 2.763506909718943e-06,
      "loss": 0.4846,
      "step": 229910
    },
    {
      "epoch": 94.11,
      "grad_norm": 2.3104398250579834,
      "learning_rate": 2.7633508440021913e-06,
      "loss": 0.4986,
      "step": 229920
    },
    {
      "epoch": 94.12,
      "grad_norm": 2.0213704109191895,
      "learning_rate": 2.763194776363312e-06,
      "loss": 0.4927,
      "step": 229930
    },
    {
      "epoch": 94.12,
      "grad_norm": 2.1493101119995117,
      "learning_rate": 2.7630387068030225e-06,
      "loss": 0.4913,
      "step": 229940
    },
    {
      "epoch": 94.13,
      "grad_norm": 2.2600252628326416,
      "learning_rate": 2.762882635322035e-06,
      "loss": 0.4872,
      "step": 229950
    },
    {
      "epoch": 94.13,
      "grad_norm": 1.9721546173095703,
      "learning_rate": 2.7627265619210665e-06,
      "loss": 0.4889,
      "step": 229960
    },
    {
      "epoch": 94.13,
      "grad_norm": 1.978869915008545,
      "learning_rate": 2.7625704866008317e-06,
      "loss": 0.4831,
      "step": 229970
    },
    {
      "epoch": 94.14,
      "grad_norm": 1.6485304832458496,
      "learning_rate": 2.7624144093620444e-06,
      "loss": 0.4865,
      "step": 229980
    },
    {
      "epoch": 94.14,
      "grad_norm": 1.910736322402954,
      "learning_rate": 2.76225833020542e-06,
      "loss": 0.4934,
      "step": 229990
    },
    {
      "epoch": 94.15,
      "grad_norm": 2.342481851577759,
      "learning_rate": 2.7621022491316744e-06,
      "loss": 0.4717,
      "step": 230000
    },
    {
      "epoch": 94.15,
      "grad_norm": 1.89113450050354,
      "learning_rate": 2.761946166141522e-06,
      "loss": 0.4889,
      "step": 230010
    },
    {
      "epoch": 94.15,
      "grad_norm": 1.6252596378326416,
      "learning_rate": 2.7617900812356775e-06,
      "loss": 0.4877,
      "step": 230020
    },
    {
      "epoch": 94.16,
      "grad_norm": 1.6688438653945923,
      "learning_rate": 2.7616339944148562e-06,
      "loss": 0.4873,
      "step": 230030
    },
    {
      "epoch": 94.16,
      "grad_norm": 2.2933287620544434,
      "learning_rate": 2.761477905679773e-06,
      "loss": 0.4907,
      "step": 230040
    },
    {
      "epoch": 94.17,
      "grad_norm": 2.8458714485168457,
      "learning_rate": 2.7613218150311435e-06,
      "loss": 0.4673,
      "step": 230050
    },
    {
      "epoch": 94.17,
      "grad_norm": 1.657141089439392,
      "learning_rate": 2.7611657224696813e-06,
      "loss": 0.4727,
      "step": 230060
    },
    {
      "epoch": 94.18,
      "grad_norm": 2.0339770317077637,
      "learning_rate": 2.761009627996102e-06,
      "loss": 0.4935,
      "step": 230070
    },
    {
      "epoch": 94.18,
      "grad_norm": 1.8758139610290527,
      "learning_rate": 2.7608535316111218e-06,
      "loss": 0.4993,
      "step": 230080
    },
    {
      "epoch": 94.18,
      "grad_norm": 1.7803986072540283,
      "learning_rate": 2.7606974333154545e-06,
      "loss": 0.4862,
      "step": 230090
    },
    {
      "epoch": 94.19,
      "grad_norm": 2.2746009826660156,
      "learning_rate": 2.760541333109816e-06,
      "loss": 0.4866,
      "step": 230100
    },
    {
      "epoch": 94.19,
      "grad_norm": 1.6123172044754028,
      "learning_rate": 2.760385230994921e-06,
      "loss": 0.4714,
      "step": 230110
    },
    {
      "epoch": 94.2,
      "grad_norm": 1.5242114067077637,
      "learning_rate": 2.7602291269714844e-06,
      "loss": 0.4948,
      "step": 230120
    },
    {
      "epoch": 94.2,
      "grad_norm": 1.548152208328247,
      "learning_rate": 2.7600730210402213e-06,
      "loss": 0.5056,
      "step": 230130
    },
    {
      "epoch": 94.2,
      "grad_norm": 1.6116478443145752,
      "learning_rate": 2.7599169132018466e-06,
      "loss": 0.4787,
      "step": 230140
    },
    {
      "epoch": 94.21,
      "grad_norm": 1.8379979133605957,
      "learning_rate": 2.7597608034570764e-06,
      "loss": 0.4833,
      "step": 230150
    },
    {
      "epoch": 94.21,
      "grad_norm": 2.018498659133911,
      "learning_rate": 2.7596046918066244e-06,
      "loss": 0.5006,
      "step": 230160
    },
    {
      "epoch": 94.22,
      "grad_norm": 3.344801902770996,
      "learning_rate": 2.7594485782512067e-06,
      "loss": 0.4994,
      "step": 230170
    },
    {
      "epoch": 94.22,
      "grad_norm": 1.6414281129837036,
      "learning_rate": 2.7592924627915387e-06,
      "loss": 0.4678,
      "step": 230180
    },
    {
      "epoch": 94.22,
      "grad_norm": 2.4677586555480957,
      "learning_rate": 2.7591363454283344e-06,
      "loss": 0.4882,
      "step": 230190
    },
    {
      "epoch": 94.23,
      "grad_norm": 2.0267529487609863,
      "learning_rate": 2.7589802261623104e-06,
      "loss": 0.4776,
      "step": 230200
    },
    {
      "epoch": 94.23,
      "grad_norm": 1.3630496263504028,
      "learning_rate": 2.7588241049941803e-06,
      "loss": 0.4847,
      "step": 230210
    },
    {
      "epoch": 94.24,
      "grad_norm": 2.5511720180511475,
      "learning_rate": 2.7586679819246603e-06,
      "loss": 0.4848,
      "step": 230220
    },
    {
      "epoch": 94.24,
      "grad_norm": 2.5664260387420654,
      "learning_rate": 2.758511856954465e-06,
      "loss": 0.4938,
      "step": 230230
    },
    {
      "epoch": 94.24,
      "grad_norm": 2.3127353191375732,
      "learning_rate": 2.7583557300843097e-06,
      "loss": 0.4809,
      "step": 230240
    },
    {
      "epoch": 94.25,
      "grad_norm": 1.7795310020446777,
      "learning_rate": 2.75819960131491e-06,
      "loss": 0.454,
      "step": 230250
    },
    {
      "epoch": 94.25,
      "grad_norm": 1.763972520828247,
      "learning_rate": 2.758043470646981e-06,
      "loss": 0.5189,
      "step": 230260
    },
    {
      "epoch": 94.26,
      "grad_norm": 1.7736241817474365,
      "learning_rate": 2.7578873380812384e-06,
      "loss": 0.4883,
      "step": 230270
    },
    {
      "epoch": 94.26,
      "grad_norm": 2.2082719802856445,
      "learning_rate": 2.7577312036183955e-06,
      "loss": 0.4813,
      "step": 230280
    },
    {
      "epoch": 94.27,
      "grad_norm": 1.8098821640014648,
      "learning_rate": 2.7575750672591696e-06,
      "loss": 0.4693,
      "step": 230290
    },
    {
      "epoch": 94.27,
      "grad_norm": 1.8786972761154175,
      "learning_rate": 2.757418929004275e-06,
      "loss": 0.4767,
      "step": 230300
    },
    {
      "epoch": 94.27,
      "grad_norm": 1.7487746477127075,
      "learning_rate": 2.757262788854427e-06,
      "loss": 0.485,
      "step": 230310
    },
    {
      "epoch": 94.28,
      "grad_norm": 1.830737829208374,
      "learning_rate": 2.7571066468103416e-06,
      "loss": 0.5007,
      "step": 230320
    },
    {
      "epoch": 94.28,
      "grad_norm": 1.5414652824401855,
      "learning_rate": 2.756950502872733e-06,
      "loss": 0.4826,
      "step": 230330
    },
    {
      "epoch": 94.29,
      "grad_norm": 1.7857229709625244,
      "learning_rate": 2.7567943570423167e-06,
      "loss": 0.4752,
      "step": 230340
    },
    {
      "epoch": 94.29,
      "grad_norm": 2.0857510566711426,
      "learning_rate": 2.756638209319808e-06,
      "loss": 0.4906,
      "step": 230350
    },
    {
      "epoch": 94.29,
      "grad_norm": 1.945407509803772,
      "learning_rate": 2.7564820597059224e-06,
      "loss": 0.4887,
      "step": 230360
    },
    {
      "epoch": 94.3,
      "grad_norm": 1.7212871313095093,
      "learning_rate": 2.756325908201375e-06,
      "loss": 0.4825,
      "step": 230370
    },
    {
      "epoch": 94.3,
      "grad_norm": 1.9232879877090454,
      "learning_rate": 2.7561697548068816e-06,
      "loss": 0.4667,
      "step": 230380
    },
    {
      "epoch": 94.31,
      "grad_norm": 1.8507843017578125,
      "learning_rate": 2.7560135995231575e-06,
      "loss": 0.4949,
      "step": 230390
    },
    {
      "epoch": 94.31,
      "grad_norm": 1.780519962310791,
      "learning_rate": 2.755857442350918e-06,
      "loss": 0.4693,
      "step": 230400
    },
    {
      "epoch": 94.31,
      "grad_norm": 1.7377128601074219,
      "learning_rate": 2.7557012832908773e-06,
      "loss": 0.4632,
      "step": 230410
    },
    {
      "epoch": 94.32,
      "grad_norm": 1.7154645919799805,
      "learning_rate": 2.755545122343752e-06,
      "loss": 0.477,
      "step": 230420
    },
    {
      "epoch": 94.32,
      "grad_norm": 1.6288232803344727,
      "learning_rate": 2.7553889595102567e-06,
      "loss": 0.4986,
      "step": 230430
    },
    {
      "epoch": 94.33,
      "grad_norm": 2.2360548973083496,
      "learning_rate": 2.7552327947911064e-06,
      "loss": 0.4785,
      "step": 230440
    },
    {
      "epoch": 94.33,
      "grad_norm": 2.0971405506134033,
      "learning_rate": 2.7550766281870177e-06,
      "loss": 0.4898,
      "step": 230450
    },
    {
      "epoch": 94.33,
      "grad_norm": 1.936198115348816,
      "learning_rate": 2.754920459698706e-06,
      "loss": 0.4693,
      "step": 230460
    },
    {
      "epoch": 94.34,
      "grad_norm": 2.164020299911499,
      "learning_rate": 2.754764289326886e-06,
      "loss": 0.4936,
      "step": 230470
    },
    {
      "epoch": 94.34,
      "grad_norm": 2.069481372833252,
      "learning_rate": 2.754608117072273e-06,
      "loss": 0.4713,
      "step": 230480
    },
    {
      "epoch": 94.35,
      "grad_norm": 1.8278112411499023,
      "learning_rate": 2.7544519429355826e-06,
      "loss": 0.4934,
      "step": 230490
    },
    {
      "epoch": 94.35,
      "grad_norm": 1.5960750579833984,
      "learning_rate": 2.7542957669175304e-06,
      "loss": 0.4859,
      "step": 230500
    },
    {
      "epoch": 94.36,
      "grad_norm": 1.7032661437988281,
      "learning_rate": 2.754139589018831e-06,
      "loss": 0.4913,
      "step": 230510
    },
    {
      "epoch": 94.36,
      "grad_norm": 2.167480707168579,
      "learning_rate": 2.753983409240201e-06,
      "loss": 0.4824,
      "step": 230520
    },
    {
      "epoch": 94.36,
      "grad_norm": 1.6863515377044678,
      "learning_rate": 2.753827227582355e-06,
      "loss": 0.4988,
      "step": 230530
    },
    {
      "epoch": 94.37,
      "grad_norm": 1.796333909034729,
      "learning_rate": 2.7536710440460087e-06,
      "loss": 0.494,
      "step": 230540
    },
    {
      "epoch": 94.37,
      "grad_norm": 2.341434955596924,
      "learning_rate": 2.753514858631878e-06,
      "loss": 0.4742,
      "step": 230550
    },
    {
      "epoch": 94.38,
      "grad_norm": 1.877395510673523,
      "learning_rate": 2.7533586713406774e-06,
      "loss": 0.4813,
      "step": 230560
    },
    {
      "epoch": 94.38,
      "grad_norm": 1.5096393823623657,
      "learning_rate": 2.7532024821731226e-06,
      "loss": 0.4714,
      "step": 230570
    },
    {
      "epoch": 94.38,
      "grad_norm": 2.0949816703796387,
      "learning_rate": 2.7530462911299307e-06,
      "loss": 0.4927,
      "step": 230580
    },
    {
      "epoch": 94.39,
      "grad_norm": 1.6508448123931885,
      "learning_rate": 2.7528900982118152e-06,
      "loss": 0.4933,
      "step": 230590
    },
    {
      "epoch": 94.39,
      "grad_norm": 1.6859747171401978,
      "learning_rate": 2.752733903419492e-06,
      "loss": 0.4838,
      "step": 230600
    },
    {
      "epoch": 94.4,
      "grad_norm": 1.9305568933486938,
      "learning_rate": 2.752577706753677e-06,
      "loss": 0.4862,
      "step": 230610
    },
    {
      "epoch": 94.4,
      "grad_norm": 1.6455808877944946,
      "learning_rate": 2.752421508215086e-06,
      "loss": 0.5106,
      "step": 230620
    },
    {
      "epoch": 94.4,
      "grad_norm": 2.9688639640808105,
      "learning_rate": 2.7522653078044335e-06,
      "loss": 0.4877,
      "step": 230630
    },
    {
      "epoch": 94.41,
      "grad_norm": 1.633218765258789,
      "learning_rate": 2.7521091055224357e-06,
      "loss": 0.4918,
      "step": 230640
    },
    {
      "epoch": 94.41,
      "grad_norm": 1.7146984338760376,
      "learning_rate": 2.751952901369808e-06,
      "loss": 0.4846,
      "step": 230650
    },
    {
      "epoch": 94.42,
      "grad_norm": 1.8396954536437988,
      "learning_rate": 2.7517966953472666e-06,
      "loss": 0.4673,
      "step": 230660
    },
    {
      "epoch": 94.42,
      "grad_norm": 2.0846540927886963,
      "learning_rate": 2.7516404874555263e-06,
      "loss": 0.4959,
      "step": 230670
    },
    {
      "epoch": 94.42,
      "grad_norm": 1.8321408033370972,
      "learning_rate": 2.7514842776953025e-06,
      "loss": 0.4873,
      "step": 230680
    },
    {
      "epoch": 94.43,
      "grad_norm": 1.6977307796478271,
      "learning_rate": 2.7513280660673118e-06,
      "loss": 0.4823,
      "step": 230690
    },
    {
      "epoch": 94.43,
      "grad_norm": 1.7664564847946167,
      "learning_rate": 2.7511718525722686e-06,
      "loss": 0.4841,
      "step": 230700
    },
    {
      "epoch": 94.44,
      "grad_norm": 1.7592105865478516,
      "learning_rate": 2.751015637210889e-06,
      "loss": 0.4664,
      "step": 230710
    },
    {
      "epoch": 94.44,
      "grad_norm": 1.41141939163208,
      "learning_rate": 2.7508594199838886e-06,
      "loss": 0.508,
      "step": 230720
    },
    {
      "epoch": 94.45,
      "grad_norm": 1.8965797424316406,
      "learning_rate": 2.750703200891983e-06,
      "loss": 0.4958,
      "step": 230730
    },
    {
      "epoch": 94.45,
      "grad_norm": 2.003638505935669,
      "learning_rate": 2.750546979935888e-06,
      "loss": 0.496,
      "step": 230740
    },
    {
      "epoch": 94.45,
      "grad_norm": 1.8202545642852783,
      "learning_rate": 2.7503907571163186e-06,
      "loss": 0.4852,
      "step": 230750
    },
    {
      "epoch": 94.46,
      "grad_norm": 1.7552491426467896,
      "learning_rate": 2.7502345324339916e-06,
      "loss": 0.487,
      "step": 230760
    },
    {
      "epoch": 94.46,
      "grad_norm": 1.9979124069213867,
      "learning_rate": 2.7500783058896215e-06,
      "loss": 0.4887,
      "step": 230770
    },
    {
      "epoch": 94.47,
      "grad_norm": 2.170626640319824,
      "learning_rate": 2.7499220774839244e-06,
      "loss": 0.4951,
      "step": 230780
    },
    {
      "epoch": 94.47,
      "grad_norm": 2.1754846572875977,
      "learning_rate": 2.7497658472176153e-06,
      "loss": 0.4988,
      "step": 230790
    },
    {
      "epoch": 94.47,
      "grad_norm": 1.8310519456863403,
      "learning_rate": 2.7496096150914115e-06,
      "loss": 0.4814,
      "step": 230800
    },
    {
      "epoch": 94.48,
      "grad_norm": 1.684697151184082,
      "learning_rate": 2.7494533811060272e-06,
      "loss": 0.4703,
      "step": 230810
    },
    {
      "epoch": 94.48,
      "grad_norm": 2.0963423252105713,
      "learning_rate": 2.749297145262178e-06,
      "loss": 0.4869,
      "step": 230820
    },
    {
      "epoch": 94.49,
      "grad_norm": 1.9492433071136475,
      "learning_rate": 2.749140907560581e-06,
      "loss": 0.4741,
      "step": 230830
    },
    {
      "epoch": 94.49,
      "grad_norm": 2.142695903778076,
      "learning_rate": 2.748984668001951e-06,
      "loss": 0.4948,
      "step": 230840
    },
    {
      "epoch": 94.49,
      "grad_norm": 2.1680779457092285,
      "learning_rate": 2.7488284265870034e-06,
      "loss": 0.4909,
      "step": 230850
    },
    {
      "epoch": 94.5,
      "grad_norm": 1.5476813316345215,
      "learning_rate": 2.748672183316454e-06,
      "loss": 0.4703,
      "step": 230860
    },
    {
      "epoch": 94.5,
      "grad_norm": 1.73039710521698,
      "learning_rate": 2.7485159381910195e-06,
      "loss": 0.4972,
      "step": 230870
    },
    {
      "epoch": 94.51,
      "grad_norm": 2.1945550441741943,
      "learning_rate": 2.748359691211414e-06,
      "loss": 0.4829,
      "step": 230880
    },
    {
      "epoch": 94.51,
      "grad_norm": 1.7809149026870728,
      "learning_rate": 2.748203442378355e-06,
      "loss": 0.4807,
      "step": 230890
    },
    {
      "epoch": 94.51,
      "grad_norm": 1.8971350193023682,
      "learning_rate": 2.748047191692557e-06,
      "loss": 0.4771,
      "step": 230900
    },
    {
      "epoch": 94.52,
      "grad_norm": 1.6815392971038818,
      "learning_rate": 2.7478909391547367e-06,
      "loss": 0.5104,
      "step": 230910
    },
    {
      "epoch": 94.52,
      "grad_norm": 1.6829417943954468,
      "learning_rate": 2.7477346847656085e-06,
      "loss": 0.4838,
      "step": 230920
    },
    {
      "epoch": 94.53,
      "grad_norm": 1.9367841482162476,
      "learning_rate": 2.7475784285258896e-06,
      "loss": 0.4801,
      "step": 230930
    },
    {
      "epoch": 94.53,
      "grad_norm": 2.235823631286621,
      "learning_rate": 2.7474221704362947e-06,
      "loss": 0.4705,
      "step": 230940
    },
    {
      "epoch": 94.54,
      "grad_norm": 1.8321553468704224,
      "learning_rate": 2.747265910497541e-06,
      "loss": 0.4984,
      "step": 230950
    },
    {
      "epoch": 94.54,
      "grad_norm": 1.7950669527053833,
      "learning_rate": 2.7471096487103424e-06,
      "loss": 0.4856,
      "step": 230960
    },
    {
      "epoch": 94.54,
      "grad_norm": 2.0553362369537354,
      "learning_rate": 2.746953385075416e-06,
      "loss": 0.5039,
      "step": 230970
    },
    {
      "epoch": 94.55,
      "grad_norm": 2.1716370582580566,
      "learning_rate": 2.7467971195934782e-06,
      "loss": 0.4999,
      "step": 230980
    },
    {
      "epoch": 94.55,
      "grad_norm": 1.9085947275161743,
      "learning_rate": 2.7466408522652433e-06,
      "loss": 0.4894,
      "step": 230990
    },
    {
      "epoch": 94.56,
      "grad_norm": 1.6767181158065796,
      "learning_rate": 2.7464845830914274e-06,
      "loss": 0.4858,
      "step": 231000
    },
    {
      "epoch": 94.56,
      "grad_norm": 1.89971125125885,
      "learning_rate": 2.746328312072747e-06,
      "loss": 0.4803,
      "step": 231010
    },
    {
      "epoch": 94.56,
      "grad_norm": 1.9887144565582275,
      "learning_rate": 2.7461720392099176e-06,
      "loss": 0.4976,
      "step": 231020
    },
    {
      "epoch": 94.57,
      "grad_norm": 2.000223398208618,
      "learning_rate": 2.7460157645036556e-06,
      "loss": 0.4708,
      "step": 231030
    },
    {
      "epoch": 94.57,
      "grad_norm": 2.2232308387756348,
      "learning_rate": 2.745859487954676e-06,
      "loss": 0.4735,
      "step": 231040
    },
    {
      "epoch": 94.58,
      "grad_norm": 2.0050370693206787,
      "learning_rate": 2.7457032095636957e-06,
      "loss": 0.4671,
      "step": 231050
    },
    {
      "epoch": 94.58,
      "grad_norm": 2.155867099761963,
      "learning_rate": 2.7455469293314297e-06,
      "loss": 0.4938,
      "step": 231060
    },
    {
      "epoch": 94.58,
      "grad_norm": 1.7462488412857056,
      "learning_rate": 2.745390647258594e-06,
      "loss": 0.4765,
      "step": 231070
    },
    {
      "epoch": 94.59,
      "grad_norm": 2.316896438598633,
      "learning_rate": 2.7452343633459056e-06,
      "loss": 0.4904,
      "step": 231080
    },
    {
      "epoch": 94.59,
      "grad_norm": 1.8107224702835083,
      "learning_rate": 2.7450780775940786e-06,
      "loss": 0.472,
      "step": 231090
    },
    {
      "epoch": 94.6,
      "grad_norm": 1.5415453910827637,
      "learning_rate": 2.7449217900038293e-06,
      "loss": 0.4986,
      "step": 231100
    },
    {
      "epoch": 94.6,
      "grad_norm": 1.5526777505874634,
      "learning_rate": 2.7447655005758746e-06,
      "loss": 0.4845,
      "step": 231110
    },
    {
      "epoch": 94.6,
      "grad_norm": 1.8121572732925415,
      "learning_rate": 2.7446092093109308e-06,
      "loss": 0.4938,
      "step": 231120
    },
    {
      "epoch": 94.61,
      "grad_norm": 1.7088358402252197,
      "learning_rate": 2.7444529162097123e-06,
      "loss": 0.4765,
      "step": 231130
    },
    {
      "epoch": 94.61,
      "grad_norm": 1.6790727376937866,
      "learning_rate": 2.744296621272936e-06,
      "loss": 0.469,
      "step": 231140
    },
    {
      "epoch": 94.62,
      "grad_norm": 2.2796709537506104,
      "learning_rate": 2.7441403245013176e-06,
      "loss": 0.4923,
      "step": 231150
    },
    {
      "epoch": 94.62,
      "grad_norm": 1.9933587312698364,
      "learning_rate": 2.7439840258955733e-06,
      "loss": 0.4951,
      "step": 231160
    },
    {
      "epoch": 94.63,
      "grad_norm": 1.6508156061172485,
      "learning_rate": 2.743827725456419e-06,
      "loss": 0.4754,
      "step": 231170
    },
    {
      "epoch": 94.63,
      "grad_norm": 2.33497953414917,
      "learning_rate": 2.74367142318457e-06,
      "loss": 0.4764,
      "step": 231180
    },
    {
      "epoch": 94.63,
      "grad_norm": 2.5378780364990234,
      "learning_rate": 2.7435151190807434e-06,
      "loss": 0.4989,
      "step": 231190
    },
    {
      "epoch": 94.64,
      "grad_norm": 1.6926889419555664,
      "learning_rate": 2.7433588131456548e-06,
      "loss": 0.4945,
      "step": 231200
    },
    {
      "epoch": 94.64,
      "grad_norm": 1.9225422143936157,
      "learning_rate": 2.7432025053800196e-06,
      "loss": 0.4711,
      "step": 231210
    },
    {
      "epoch": 94.65,
      "grad_norm": 1.6388734579086304,
      "learning_rate": 2.7430461957845547e-06,
      "loss": 0.4971,
      "step": 231220
    },
    {
      "epoch": 94.65,
      "grad_norm": 1.8639686107635498,
      "learning_rate": 2.7428898843599756e-06,
      "loss": 0.5084,
      "step": 231230
    },
    {
      "epoch": 94.65,
      "grad_norm": 2.0450713634490967,
      "learning_rate": 2.7427335711069984e-06,
      "loss": 0.4746,
      "step": 231240
    },
    {
      "epoch": 94.66,
      "grad_norm": 1.8664236068725586,
      "learning_rate": 2.7425772560263392e-06,
      "loss": 0.4861,
      "step": 231250
    },
    {
      "epoch": 94.66,
      "grad_norm": 2.001983880996704,
      "learning_rate": 2.7424209391187147e-06,
      "loss": 0.4948,
      "step": 231260
    },
    {
      "epoch": 94.67,
      "grad_norm": 1.7619068622589111,
      "learning_rate": 2.7422646203848405e-06,
      "loss": 0.4866,
      "step": 231270
    },
    {
      "epoch": 94.67,
      "grad_norm": 2.059018611907959,
      "learning_rate": 2.742108299825432e-06,
      "loss": 0.5035,
      "step": 231280
    },
    {
      "epoch": 94.67,
      "grad_norm": 1.8798327445983887,
      "learning_rate": 2.7419519774412054e-06,
      "loss": 0.4752,
      "step": 231290
    },
    {
      "epoch": 94.68,
      "grad_norm": 1.8622665405273438,
      "learning_rate": 2.7417956532328773e-06,
      "loss": 0.4952,
      "step": 231300
    },
    {
      "epoch": 94.68,
      "grad_norm": 1.5782039165496826,
      "learning_rate": 2.741639327201164e-06,
      "loss": 0.469,
      "step": 231310
    },
    {
      "epoch": 94.69,
      "grad_norm": 1.8040318489074707,
      "learning_rate": 2.741482999346782e-06,
      "loss": 0.4913,
      "step": 231320
    },
    {
      "epoch": 94.69,
      "grad_norm": 2.264167547225952,
      "learning_rate": 2.741326669670446e-06,
      "loss": 0.5095,
      "step": 231330
    },
    {
      "epoch": 94.7,
      "grad_norm": 1.774714469909668,
      "learning_rate": 2.7411703381728734e-06,
      "loss": 0.4703,
      "step": 231340
    },
    {
      "epoch": 94.7,
      "grad_norm": 1.6954681873321533,
      "learning_rate": 2.741014004854779e-06,
      "loss": 0.4838,
      "step": 231350
    },
    {
      "epoch": 94.7,
      "grad_norm": 2.219348192214966,
      "learning_rate": 2.7408576697168804e-06,
      "loss": 0.4989,
      "step": 231360
    },
    {
      "epoch": 94.71,
      "grad_norm": 1.814380407333374,
      "learning_rate": 2.7407013327598928e-06,
      "loss": 0.4783,
      "step": 231370
    },
    {
      "epoch": 94.71,
      "grad_norm": 2.48801589012146,
      "learning_rate": 2.7405449939845326e-06,
      "loss": 0.4791,
      "step": 231380
    },
    {
      "epoch": 94.72,
      "grad_norm": 1.8138424158096313,
      "learning_rate": 2.7403886533915154e-06,
      "loss": 0.4955,
      "step": 231390
    },
    {
      "epoch": 94.72,
      "grad_norm": 1.5507208108901978,
      "learning_rate": 2.7402323109815584e-06,
      "loss": 0.5066,
      "step": 231400
    },
    {
      "epoch": 94.72,
      "grad_norm": 2.0713388919830322,
      "learning_rate": 2.7400759667553777e-06,
      "loss": 0.48,
      "step": 231410
    },
    {
      "epoch": 94.73,
      "grad_norm": 1.5592739582061768,
      "learning_rate": 2.739919620713689e-06,
      "loss": 0.4865,
      "step": 231420
    },
    {
      "epoch": 94.73,
      "grad_norm": 2.0733461380004883,
      "learning_rate": 2.7397632728572085e-06,
      "loss": 0.5012,
      "step": 231430
    },
    {
      "epoch": 94.74,
      "grad_norm": 2.1362271308898926,
      "learning_rate": 2.7396069231866524e-06,
      "loss": 0.4828,
      "step": 231440
    },
    {
      "epoch": 94.74,
      "grad_norm": 1.6966904401779175,
      "learning_rate": 2.7394505717027372e-06,
      "loss": 0.4735,
      "step": 231450
    },
    {
      "epoch": 94.74,
      "grad_norm": 1.8256443738937378,
      "learning_rate": 2.739294218406179e-06,
      "loss": 0.4838,
      "step": 231460
    },
    {
      "epoch": 94.75,
      "grad_norm": 1.8891017436981201,
      "learning_rate": 2.739137863297694e-06,
      "loss": 0.4814,
      "step": 231470
    },
    {
      "epoch": 94.75,
      "grad_norm": 1.957402229309082,
      "learning_rate": 2.738981506377998e-06,
      "loss": 0.4715,
      "step": 231480
    },
    {
      "epoch": 94.76,
      "grad_norm": 1.8916540145874023,
      "learning_rate": 2.738825147647808e-06,
      "loss": 0.5,
      "step": 231490
    },
    {
      "epoch": 94.76,
      "grad_norm": 2.987456798553467,
      "learning_rate": 2.73866878710784e-06,
      "loss": 0.4574,
      "step": 231500
    },
    {
      "epoch": 94.76,
      "grad_norm": 2.140869379043579,
      "learning_rate": 2.73851242475881e-06,
      "loss": 0.4878,
      "step": 231510
    },
    {
      "epoch": 94.77,
      "grad_norm": 2.334496259689331,
      "learning_rate": 2.738356060601435e-06,
      "loss": 0.4866,
      "step": 231520
    },
    {
      "epoch": 94.77,
      "grad_norm": 2.08013653755188,
      "learning_rate": 2.7381996946364303e-06,
      "loss": 0.4939,
      "step": 231530
    },
    {
      "epoch": 94.78,
      "grad_norm": 1.8051120042800903,
      "learning_rate": 2.7380433268645124e-06,
      "loss": 0.5055,
      "step": 231540
    },
    {
      "epoch": 94.78,
      "grad_norm": 1.9476125240325928,
      "learning_rate": 2.7378869572863984e-06,
      "loss": 0.4774,
      "step": 231550
    },
    {
      "epoch": 94.79,
      "grad_norm": 1.89741051197052,
      "learning_rate": 2.737730585902804e-06,
      "loss": 0.4912,
      "step": 231560
    },
    {
      "epoch": 94.79,
      "grad_norm": 2.2954297065734863,
      "learning_rate": 2.737574212714446e-06,
      "loss": 0.4706,
      "step": 231570
    },
    {
      "epoch": 94.79,
      "grad_norm": 1.8462250232696533,
      "learning_rate": 2.737417837722039e-06,
      "loss": 0.4734,
      "step": 231580
    },
    {
      "epoch": 94.8,
      "grad_norm": 1.8672910928726196,
      "learning_rate": 2.737261460926301e-06,
      "loss": 0.4951,
      "step": 231590
    },
    {
      "epoch": 94.8,
      "grad_norm": 2.413825035095215,
      "learning_rate": 2.737105082327948e-06,
      "loss": 0.4908,
      "step": 231600
    },
    {
      "epoch": 94.81,
      "grad_norm": 2.266526699066162,
      "learning_rate": 2.7369487019276963e-06,
      "loss": 0.4892,
      "step": 231610
    },
    {
      "epoch": 94.81,
      "grad_norm": 1.9805768728256226,
      "learning_rate": 2.736792319726262e-06,
      "loss": 0.4672,
      "step": 231620
    },
    {
      "epoch": 94.81,
      "grad_norm": 2.265366554260254,
      "learning_rate": 2.7366359357243633e-06,
      "loss": 0.4855,
      "step": 231630
    },
    {
      "epoch": 94.82,
      "grad_norm": 2.3773605823516846,
      "learning_rate": 2.7364795499227134e-06,
      "loss": 0.4973,
      "step": 231640
    },
    {
      "epoch": 94.82,
      "grad_norm": 1.787940502166748,
      "learning_rate": 2.736323162322031e-06,
      "loss": 0.5036,
      "step": 231650
    },
    {
      "epoch": 94.83,
      "grad_norm": 1.7007391452789307,
      "learning_rate": 2.736166772923031e-06,
      "loss": 0.4902,
      "step": 231660
    },
    {
      "epoch": 94.83,
      "grad_norm": 2.123412847518921,
      "learning_rate": 2.7360103817264303e-06,
      "loss": 0.4843,
      "step": 231670
    },
    {
      "epoch": 94.83,
      "grad_norm": 1.4683260917663574,
      "learning_rate": 2.735853988732946e-06,
      "loss": 0.4841,
      "step": 231680
    },
    {
      "epoch": 94.84,
      "grad_norm": 1.8165943622589111,
      "learning_rate": 2.7356975939432944e-06,
      "loss": 0.4631,
      "step": 231690
    },
    {
      "epoch": 94.84,
      "grad_norm": 1.9070374965667725,
      "learning_rate": 2.7355411973581913e-06,
      "loss": 0.5051,
      "step": 231700
    },
    {
      "epoch": 94.85,
      "grad_norm": 1.9403386116027832,
      "learning_rate": 2.7353847989783538e-06,
      "loss": 0.4735,
      "step": 231710
    },
    {
      "epoch": 94.85,
      "grad_norm": 2.057917356491089,
      "learning_rate": 2.735228398804498e-06,
      "loss": 0.506,
      "step": 231720
    },
    {
      "epoch": 94.85,
      "grad_norm": 2.07376766204834,
      "learning_rate": 2.735071996837339e-06,
      "loss": 0.4836,
      "step": 231730
    },
    {
      "epoch": 94.86,
      "grad_norm": 1.9067473411560059,
      "learning_rate": 2.734915593077596e-06,
      "loss": 0.5029,
      "step": 231740
    },
    {
      "epoch": 94.86,
      "grad_norm": 2.190519094467163,
      "learning_rate": 2.7347591875259837e-06,
      "loss": 0.4809,
      "step": 231750
    },
    {
      "epoch": 94.87,
      "grad_norm": 1.7871015071868896,
      "learning_rate": 2.734602780183218e-06,
      "loss": 0.4897,
      "step": 231760
    },
    {
      "epoch": 94.87,
      "grad_norm": 1.9697476625442505,
      "learning_rate": 2.7344463710500167e-06,
      "loss": 0.5127,
      "step": 231770
    },
    {
      "epoch": 94.88,
      "grad_norm": 1.9348224401474,
      "learning_rate": 2.734289960127096e-06,
      "loss": 0.4986,
      "step": 231780
    },
    {
      "epoch": 94.88,
      "grad_norm": 2.4191884994506836,
      "learning_rate": 2.7341335474151724e-06,
      "loss": 0.4906,
      "step": 231790
    },
    {
      "epoch": 94.88,
      "grad_norm": 1.6996163129806519,
      "learning_rate": 2.733977132914962e-06,
      "loss": 0.4698,
      "step": 231800
    },
    {
      "epoch": 94.89,
      "grad_norm": 2.0950472354888916,
      "learning_rate": 2.7338207166271815e-06,
      "loss": 0.4776,
      "step": 231810
    },
    {
      "epoch": 94.89,
      "grad_norm": 2.111701726913452,
      "learning_rate": 2.733664298552547e-06,
      "loss": 0.4856,
      "step": 231820
    },
    {
      "epoch": 94.9,
      "grad_norm": 1.8677500486373901,
      "learning_rate": 2.7335078786917766e-06,
      "loss": 0.4981,
      "step": 231830
    },
    {
      "epoch": 94.9,
      "grad_norm": 1.8782950639724731,
      "learning_rate": 2.733351457045585e-06,
      "loss": 0.4859,
      "step": 231840
    },
    {
      "epoch": 94.9,
      "grad_norm": 1.9088397026062012,
      "learning_rate": 2.7331950336146897e-06,
      "loss": 0.4741,
      "step": 231850
    },
    {
      "epoch": 94.91,
      "grad_norm": 1.8333252668380737,
      "learning_rate": 2.7330386083998065e-06,
      "loss": 0.4886,
      "step": 231860
    },
    {
      "epoch": 94.91,
      "grad_norm": 1.5309585332870483,
      "learning_rate": 2.732882181401653e-06,
      "loss": 0.4807,
      "step": 231870
    },
    {
      "epoch": 94.92,
      "grad_norm": 1.7703826427459717,
      "learning_rate": 2.732725752620945e-06,
      "loss": 0.486,
      "step": 231880
    },
    {
      "epoch": 94.92,
      "grad_norm": 1.7095061540603638,
      "learning_rate": 2.7325693220583993e-06,
      "loss": 0.4752,
      "step": 231890
    },
    {
      "epoch": 94.92,
      "grad_norm": 2.045055389404297,
      "learning_rate": 2.7324128897147327e-06,
      "loss": 0.4683,
      "step": 231900
    },
    {
      "epoch": 94.93,
      "grad_norm": 1.721434235572815,
      "learning_rate": 2.7322564555906617e-06,
      "loss": 0.4781,
      "step": 231910
    },
    {
      "epoch": 94.93,
      "grad_norm": 1.9663869142532349,
      "learning_rate": 2.732100019686903e-06,
      "loss": 0.4952,
      "step": 231920
    },
    {
      "epoch": 94.94,
      "grad_norm": 1.9977903366088867,
      "learning_rate": 2.7319435820041724e-06,
      "loss": 0.4912,
      "step": 231930
    },
    {
      "epoch": 94.94,
      "grad_norm": 2.372159481048584,
      "learning_rate": 2.7317871425431874e-06,
      "loss": 0.486,
      "step": 231940
    },
    {
      "epoch": 94.94,
      "grad_norm": 2.2879910469055176,
      "learning_rate": 2.731630701304664e-06,
      "loss": 0.4859,
      "step": 231950
    },
    {
      "epoch": 94.95,
      "grad_norm": 2.0734641551971436,
      "learning_rate": 2.731474258289319e-06,
      "loss": 0.497,
      "step": 231960
    },
    {
      "epoch": 94.95,
      "grad_norm": 1.8914093971252441,
      "learning_rate": 2.73131781349787e-06,
      "loss": 0.4803,
      "step": 231970
    },
    {
      "epoch": 94.96,
      "grad_norm": 2.369844913482666,
      "learning_rate": 2.7311613669310324e-06,
      "loss": 0.4974,
      "step": 231980
    },
    {
      "epoch": 94.96,
      "grad_norm": 1.7800244092941284,
      "learning_rate": 2.731004918589524e-06,
      "loss": 0.4809,
      "step": 231990
    },
    {
      "epoch": 94.97,
      "grad_norm": 1.5415217876434326,
      "learning_rate": 2.73084846847406e-06,
      "loss": 0.4699,
      "step": 232000
    },
    {
      "epoch": 94.97,
      "grad_norm": 1.6847354173660278,
      "learning_rate": 2.730692016585358e-06,
      "loss": 0.494,
      "step": 232010
    },
    {
      "epoch": 94.97,
      "grad_norm": 2.0335824489593506,
      "learning_rate": 2.730535562924135e-06,
      "loss": 0.5024,
      "step": 232020
    },
    {
      "epoch": 94.98,
      "grad_norm": 1.9453872442245483,
      "learning_rate": 2.730379107491107e-06,
      "loss": 0.5026,
      "step": 232030
    },
    {
      "epoch": 94.98,
      "grad_norm": 1.7566189765930176,
      "learning_rate": 2.7302226502869904e-06,
      "loss": 0.4696,
      "step": 232040
    },
    {
      "epoch": 94.99,
      "grad_norm": 1.7024898529052734,
      "learning_rate": 2.730066191312503e-06,
      "loss": 0.4805,
      "step": 232050
    },
    {
      "epoch": 94.99,
      "grad_norm": 2.0898985862731934,
      "learning_rate": 2.729909730568361e-06,
      "loss": 0.4691,
      "step": 232060
    },
    {
      "epoch": 94.99,
      "grad_norm": 2.132427453994751,
      "learning_rate": 2.7297532680552807e-06,
      "loss": 0.5087,
      "step": 232070
    },
    {
      "epoch": 95.0,
      "grad_norm": 2.181823253631592,
      "learning_rate": 2.7295968037739798e-06,
      "loss": 0.4655,
      "step": 232080
    },
    {
      "epoch": 95.0,
      "eval_loss": 0.4833046495914459,
      "eval_runtime": 53.1604,
      "eval_samples_per_second": 64.879,
      "eval_steps_per_second": 8.126,
      "step": 232085
    },
    {
      "epoch": 95.0,
      "grad_norm": 2.3486971855163574,
      "learning_rate": 2.729440337725174e-06,
      "loss": 0.4745,
      "step": 232090
    },
    {
      "epoch": 95.01,
      "grad_norm": 1.832624912261963,
      "learning_rate": 2.7292838699095805e-06,
      "loss": 0.4918,
      "step": 232100
    },
    {
      "epoch": 95.01,
      "grad_norm": 1.6248441934585571,
      "learning_rate": 2.729127400327916e-06,
      "loss": 0.472,
      "step": 232110
    },
    {
      "epoch": 95.01,
      "grad_norm": 1.9741909503936768,
      "learning_rate": 2.7289709289808972e-06,
      "loss": 0.4708,
      "step": 232120
    },
    {
      "epoch": 95.02,
      "grad_norm": 2.34012770652771,
      "learning_rate": 2.7288144558692414e-06,
      "loss": 0.4898,
      "step": 232130
    },
    {
      "epoch": 95.02,
      "grad_norm": 1.8702608346939087,
      "learning_rate": 2.728657980993665e-06,
      "loss": 0.4784,
      "step": 232140
    },
    {
      "epoch": 95.03,
      "grad_norm": 1.4212688207626343,
      "learning_rate": 2.7285015043548845e-06,
      "loss": 0.4802,
      "step": 232150
    },
    {
      "epoch": 95.03,
      "grad_norm": 2.3857688903808594,
      "learning_rate": 2.7283450259536167e-06,
      "loss": 0.4765,
      "step": 232160
    },
    {
      "epoch": 95.03,
      "grad_norm": 1.8848451375961304,
      "learning_rate": 2.7281885457905793e-06,
      "loss": 0.4848,
      "step": 232170
    },
    {
      "epoch": 95.04,
      "grad_norm": 1.9155491590499878,
      "learning_rate": 2.728032063866488e-06,
      "loss": 0.4733,
      "step": 232180
    },
    {
      "epoch": 95.04,
      "grad_norm": 1.7817035913467407,
      "learning_rate": 2.7278755801820597e-06,
      "loss": 0.4778,
      "step": 232190
    },
    {
      "epoch": 95.05,
      "grad_norm": 2.0008604526519775,
      "learning_rate": 2.727719094738012e-06,
      "loss": 0.4807,
      "step": 232200
    },
    {
      "epoch": 95.05,
      "grad_norm": 2.2371695041656494,
      "learning_rate": 2.727562607535062e-06,
      "loss": 0.5044,
      "step": 232210
    },
    {
      "epoch": 95.06,
      "grad_norm": 1.7260199785232544,
      "learning_rate": 2.7274061185739253e-06,
      "loss": 0.4764,
      "step": 232220
    },
    {
      "epoch": 95.06,
      "grad_norm": 2.291003704071045,
      "learning_rate": 2.7272496278553195e-06,
      "loss": 0.4775,
      "step": 232230
    },
    {
      "epoch": 95.06,
      "grad_norm": 1.7343963384628296,
      "learning_rate": 2.7270931353799602e-06,
      "loss": 0.4923,
      "step": 232240
    },
    {
      "epoch": 95.07,
      "grad_norm": 1.7011057138442993,
      "learning_rate": 2.726936641148567e-06,
      "loss": 0.4865,
      "step": 232250
    },
    {
      "epoch": 95.07,
      "grad_norm": 1.8150880336761475,
      "learning_rate": 2.726780145161854e-06,
      "loss": 0.4568,
      "step": 232260
    },
    {
      "epoch": 95.08,
      "grad_norm": 2.074711799621582,
      "learning_rate": 2.7266236474205398e-06,
      "loss": 0.4848,
      "step": 232270
    },
    {
      "epoch": 95.08,
      "grad_norm": 2.720566987991333,
      "learning_rate": 2.7264671479253407e-06,
      "loss": 0.5005,
      "step": 232280
    },
    {
      "epoch": 95.08,
      "grad_norm": 1.737703800201416,
      "learning_rate": 2.726310646676974e-06,
      "loss": 0.4798,
      "step": 232290
    },
    {
      "epoch": 95.09,
      "grad_norm": 1.861838936805725,
      "learning_rate": 2.7261541436761555e-06,
      "loss": 0.4896,
      "step": 232300
    },
    {
      "epoch": 95.09,
      "grad_norm": 2.0903306007385254,
      "learning_rate": 2.725997638923603e-06,
      "loss": 0.4784,
      "step": 232310
    },
    {
      "epoch": 95.1,
      "grad_norm": 1.9633190631866455,
      "learning_rate": 2.7258411324200334e-06,
      "loss": 0.5147,
      "step": 232320
    },
    {
      "epoch": 95.1,
      "grad_norm": 1.8684054613113403,
      "learning_rate": 2.725684624166163e-06,
      "loss": 0.4966,
      "step": 232330
    },
    {
      "epoch": 95.1,
      "grad_norm": 1.7022314071655273,
      "learning_rate": 2.7255281141627092e-06,
      "loss": 0.4712,
      "step": 232340
    },
    {
      "epoch": 95.11,
      "grad_norm": 1.687994122505188,
      "learning_rate": 2.7253716024103903e-06,
      "loss": 0.4917,
      "step": 232350
    },
    {
      "epoch": 95.11,
      "grad_norm": 2.288989305496216,
      "learning_rate": 2.7252150889099208e-06,
      "loss": 0.4998,
      "step": 232360
    },
    {
      "epoch": 95.12,
      "grad_norm": 1.7550644874572754,
      "learning_rate": 2.725058573662019e-06,
      "loss": 0.4917,
      "step": 232370
    },
    {
      "epoch": 95.12,
      "grad_norm": 1.8063311576843262,
      "learning_rate": 2.724902056667402e-06,
      "loss": 0.4879,
      "step": 232380
    },
    {
      "epoch": 95.12,
      "grad_norm": 1.7751320600509644,
      "learning_rate": 2.7247455379267864e-06,
      "loss": 0.4886,
      "step": 232390
    },
    {
      "epoch": 95.13,
      "grad_norm": 1.9955401420593262,
      "learning_rate": 2.7245890174408887e-06,
      "loss": 0.4937,
      "step": 232400
    },
    {
      "epoch": 95.13,
      "grad_norm": 2.1815998554229736,
      "learning_rate": 2.724432495210427e-06,
      "loss": 0.483,
      "step": 232410
    },
    {
      "epoch": 95.14,
      "grad_norm": 1.9312574863433838,
      "learning_rate": 2.7242759712361177e-06,
      "loss": 0.4837,
      "step": 232420
    },
    {
      "epoch": 95.14,
      "grad_norm": 1.7437653541564941,
      "learning_rate": 2.7241194455186777e-06,
      "loss": 0.4933,
      "step": 232430
    },
    {
      "epoch": 95.15,
      "grad_norm": 2.230231523513794,
      "learning_rate": 2.723962918058824e-06,
      "loss": 0.4848,
      "step": 232440
    },
    {
      "epoch": 95.15,
      "grad_norm": 1.9577515125274658,
      "learning_rate": 2.7238063888572735e-06,
      "loss": 0.4962,
      "step": 232450
    },
    {
      "epoch": 95.15,
      "grad_norm": 1.9810693264007568,
      "learning_rate": 2.7236498579147447e-06,
      "loss": 0.4846,
      "step": 232460
    },
    {
      "epoch": 95.16,
      "grad_norm": 1.7141021490097046,
      "learning_rate": 2.7234933252319524e-06,
      "loss": 0.4733,
      "step": 232470
    },
    {
      "epoch": 95.16,
      "grad_norm": 1.913803219795227,
      "learning_rate": 2.723336790809615e-06,
      "loss": 0.4733,
      "step": 232480
    },
    {
      "epoch": 95.17,
      "grad_norm": 1.7325674295425415,
      "learning_rate": 2.72318025464845e-06,
      "loss": 0.4926,
      "step": 232490
    },
    {
      "epoch": 95.17,
      "grad_norm": 1.832532286643982,
      "learning_rate": 2.7230237167491728e-06,
      "loss": 0.4903,
      "step": 232500
    },
    {
      "epoch": 95.17,
      "grad_norm": 2.1109461784362793,
      "learning_rate": 2.7228671771125026e-06,
      "loss": 0.4802,
      "step": 232510
    },
    {
      "epoch": 95.18,
      "grad_norm": 1.851242184638977,
      "learning_rate": 2.722710635739154e-06,
      "loss": 0.4715,
      "step": 232520
    },
    {
      "epoch": 95.18,
      "grad_norm": 1.8404262065887451,
      "learning_rate": 2.722554092629846e-06,
      "loss": 0.5104,
      "step": 232530
    },
    {
      "epoch": 95.19,
      "grad_norm": 1.8359180688858032,
      "learning_rate": 2.722397547785295e-06,
      "loss": 0.4755,
      "step": 232540
    },
    {
      "epoch": 95.19,
      "grad_norm": 1.6212127208709717,
      "learning_rate": 2.7222410012062186e-06,
      "loss": 0.5001,
      "step": 232550
    },
    {
      "epoch": 95.19,
      "grad_norm": 1.8377798795700073,
      "learning_rate": 2.722084452893333e-06,
      "loss": 0.4926,
      "step": 232560
    },
    {
      "epoch": 95.2,
      "grad_norm": 1.5932377576828003,
      "learning_rate": 2.721927902847357e-06,
      "loss": 0.4577,
      "step": 232570
    },
    {
      "epoch": 95.2,
      "grad_norm": 1.5917950868606567,
      "learning_rate": 2.7217713510690055e-06,
      "loss": 0.4925,
      "step": 232580
    },
    {
      "epoch": 95.21,
      "grad_norm": 1.7289972305297852,
      "learning_rate": 2.7216147975589965e-06,
      "loss": 0.4937,
      "step": 232590
    },
    {
      "epoch": 95.21,
      "grad_norm": 1.5712924003601074,
      "learning_rate": 2.7214582423180486e-06,
      "loss": 0.4833,
      "step": 232600
    },
    {
      "epoch": 95.21,
      "grad_norm": 1.5903109312057495,
      "learning_rate": 2.7213016853468763e-06,
      "loss": 0.4884,
      "step": 232610
    },
    {
      "epoch": 95.22,
      "grad_norm": 1.9851350784301758,
      "learning_rate": 2.7211451266461986e-06,
      "loss": 0.4873,
      "step": 232620
    },
    {
      "epoch": 95.22,
      "grad_norm": 1.9836281538009644,
      "learning_rate": 2.7209885662167325e-06,
      "loss": 0.4792,
      "step": 232630
    },
    {
      "epoch": 95.23,
      "grad_norm": 2.0977065563201904,
      "learning_rate": 2.720832004059196e-06,
      "loss": 0.492,
      "step": 232640
    },
    {
      "epoch": 95.23,
      "grad_norm": 1.7980471849441528,
      "learning_rate": 2.720675440174304e-06,
      "loss": 0.4828,
      "step": 232650
    },
    {
      "epoch": 95.24,
      "grad_norm": 1.8441942930221558,
      "learning_rate": 2.720518874562775e-06,
      "loss": 0.4918,
      "step": 232660
    },
    {
      "epoch": 95.24,
      "grad_norm": 1.6232939958572388,
      "learning_rate": 2.7203623072253264e-06,
      "loss": 0.474,
      "step": 232670
    },
    {
      "epoch": 95.24,
      "grad_norm": 1.6306335926055908,
      "learning_rate": 2.720205738162675e-06,
      "loss": 0.4664,
      "step": 232680
    },
    {
      "epoch": 95.25,
      "grad_norm": 2.0285251140594482,
      "learning_rate": 2.7200491673755388e-06,
      "loss": 0.5098,
      "step": 232690
    },
    {
      "epoch": 95.25,
      "grad_norm": 1.5401426553726196,
      "learning_rate": 2.719892594864633e-06,
      "loss": 0.4701,
      "step": 232700
    },
    {
      "epoch": 95.26,
      "grad_norm": 2.42242431640625,
      "learning_rate": 2.7197360206306772e-06,
      "loss": 0.5028,
      "step": 232710
    },
    {
      "epoch": 95.26,
      "grad_norm": 2.012084484100342,
      "learning_rate": 2.719579444674388e-06,
      "loss": 0.4936,
      "step": 232720
    },
    {
      "epoch": 95.26,
      "grad_norm": 2.020986318588257,
      "learning_rate": 2.7194228669964816e-06,
      "loss": 0.4886,
      "step": 232730
    },
    {
      "epoch": 95.27,
      "grad_norm": 1.641363263130188,
      "learning_rate": 2.719266287597676e-06,
      "loss": 0.4799,
      "step": 232740
    },
    {
      "epoch": 95.27,
      "grad_norm": 1.688861608505249,
      "learning_rate": 2.7191097064786886e-06,
      "loss": 0.4795,
      "step": 232750
    },
    {
      "epoch": 95.28,
      "grad_norm": 2.0553603172302246,
      "learning_rate": 2.718953123640237e-06,
      "loss": 0.4991,
      "step": 232760
    },
    {
      "epoch": 95.28,
      "grad_norm": 2.326972484588623,
      "learning_rate": 2.7187965390830376e-06,
      "loss": 0.4817,
      "step": 232770
    },
    {
      "epoch": 95.28,
      "grad_norm": 1.83210027217865,
      "learning_rate": 2.718639952807808e-06,
      "loss": 0.4787,
      "step": 232780
    },
    {
      "epoch": 95.29,
      "grad_norm": 1.7886501550674438,
      "learning_rate": 2.7184833648152655e-06,
      "loss": 0.5092,
      "step": 232790
    },
    {
      "epoch": 95.29,
      "grad_norm": 1.8306183815002441,
      "learning_rate": 2.718326775106128e-06,
      "loss": 0.4915,
      "step": 232800
    },
    {
      "epoch": 95.3,
      "grad_norm": 1.6517155170440674,
      "learning_rate": 2.7181701836811115e-06,
      "loss": 0.4948,
      "step": 232810
    },
    {
      "epoch": 95.3,
      "grad_norm": 1.5212743282318115,
      "learning_rate": 2.7180135905409347e-06,
      "loss": 0.4901,
      "step": 232820
    },
    {
      "epoch": 95.3,
      "grad_norm": 2.3633406162261963,
      "learning_rate": 2.717856995686314e-06,
      "loss": 0.4603,
      "step": 232830
    },
    {
      "epoch": 95.31,
      "grad_norm": 2.7001383304595947,
      "learning_rate": 2.7177003991179672e-06,
      "loss": 0.4772,
      "step": 232840
    },
    {
      "epoch": 95.31,
      "grad_norm": 1.7709941864013672,
      "learning_rate": 2.717543800836612e-06,
      "loss": 0.4858,
      "step": 232850
    },
    {
      "epoch": 95.32,
      "grad_norm": 1.915137767791748,
      "learning_rate": 2.7173872008429657e-06,
      "loss": 0.487,
      "step": 232860
    },
    {
      "epoch": 95.32,
      "grad_norm": 1.8353151082992554,
      "learning_rate": 2.717230599137744e-06,
      "loss": 0.4702,
      "step": 232870
    },
    {
      "epoch": 95.33,
      "grad_norm": 1.7095608711242676,
      "learning_rate": 2.717073995721666e-06,
      "loss": 0.4957,
      "step": 232880
    },
    {
      "epoch": 95.33,
      "grad_norm": 2.0568127632141113,
      "learning_rate": 2.716917390595449e-06,
      "loss": 0.4779,
      "step": 232890
    },
    {
      "epoch": 95.33,
      "grad_norm": 1.7406291961669922,
      "learning_rate": 2.7167607837598086e-06,
      "loss": 0.4774,
      "step": 232900
    },
    {
      "epoch": 95.34,
      "grad_norm": 1.796392560005188,
      "learning_rate": 2.7166041752154644e-06,
      "loss": 0.4714,
      "step": 232910
    },
    {
      "epoch": 95.34,
      "grad_norm": 1.8312040567398071,
      "learning_rate": 2.7164475649631333e-06,
      "loss": 0.4881,
      "step": 232920
    },
    {
      "epoch": 95.35,
      "grad_norm": 1.8812745809555054,
      "learning_rate": 2.7162909530035324e-06,
      "loss": 0.4914,
      "step": 232930
    },
    {
      "epoch": 95.35,
      "grad_norm": 2.1809756755828857,
      "learning_rate": 2.716134339337379e-06,
      "loss": 0.4998,
      "step": 232940
    },
    {
      "epoch": 95.35,
      "grad_norm": 1.9955406188964844,
      "learning_rate": 2.7159777239653907e-06,
      "loss": 0.4961,
      "step": 232950
    },
    {
      "epoch": 95.36,
      "grad_norm": 2.0220131874084473,
      "learning_rate": 2.7158211068882843e-06,
      "loss": 0.4735,
      "step": 232960
    },
    {
      "epoch": 95.36,
      "grad_norm": 1.7524391412734985,
      "learning_rate": 2.7156644881067783e-06,
      "loss": 0.4841,
      "step": 232970
    },
    {
      "epoch": 95.37,
      "grad_norm": 2.365554094314575,
      "learning_rate": 2.715507867621589e-06,
      "loss": 0.4958,
      "step": 232980
    },
    {
      "epoch": 95.37,
      "grad_norm": 2.1689200401306152,
      "learning_rate": 2.715351245433435e-06,
      "loss": 0.47,
      "step": 232990
    },
    {
      "epoch": 95.37,
      "grad_norm": 1.7151316404342651,
      "learning_rate": 2.7151946215430334e-06,
      "loss": 0.4693,
      "step": 233000
    },
    {
      "epoch": 95.38,
      "grad_norm": 1.908267617225647,
      "learning_rate": 2.7150379959511017e-06,
      "loss": 0.4815,
      "step": 233010
    },
    {
      "epoch": 95.38,
      "grad_norm": 1.8715440034866333,
      "learning_rate": 2.7148813686583566e-06,
      "loss": 0.4858,
      "step": 233020
    },
    {
      "epoch": 95.39,
      "grad_norm": 1.6356899738311768,
      "learning_rate": 2.7147247396655164e-06,
      "loss": 0.4899,
      "step": 233030
    },
    {
      "epoch": 95.39,
      "grad_norm": 2.979576826095581,
      "learning_rate": 2.714568108973298e-06,
      "loss": 0.4746,
      "step": 233040
    },
    {
      "epoch": 95.4,
      "grad_norm": 1.9319239854812622,
      "learning_rate": 2.7144114765824198e-06,
      "loss": 0.4909,
      "step": 233050
    },
    {
      "epoch": 95.4,
      "grad_norm": 2.2734639644622803,
      "learning_rate": 2.7142548424935986e-06,
      "loss": 0.502,
      "step": 233060
    },
    {
      "epoch": 95.4,
      "grad_norm": 2.135324239730835,
      "learning_rate": 2.7140982067075516e-06,
      "loss": 0.4675,
      "step": 233070
    },
    {
      "epoch": 95.41,
      "grad_norm": 1.5469801425933838,
      "learning_rate": 2.7139415692249972e-06,
      "loss": 0.4939,
      "step": 233080
    },
    {
      "epoch": 95.41,
      "grad_norm": 1.709675669670105,
      "learning_rate": 2.713784930046653e-06,
      "loss": 0.4672,
      "step": 233090
    },
    {
      "epoch": 95.42,
      "grad_norm": 1.8663396835327148,
      "learning_rate": 2.7136282891732358e-06,
      "loss": 0.4801,
      "step": 233100
    },
    {
      "epoch": 95.42,
      "grad_norm": 1.9208498001098633,
      "learning_rate": 2.713471646605463e-06,
      "loss": 0.4941,
      "step": 233110
    },
    {
      "epoch": 95.42,
      "grad_norm": 1.5665640830993652,
      "learning_rate": 2.7133150023440533e-06,
      "loss": 0.4685,
      "step": 233120
    },
    {
      "epoch": 95.43,
      "grad_norm": 1.7370463609695435,
      "learning_rate": 2.713158356389723e-06,
      "loss": 0.4833,
      "step": 233130
    },
    {
      "epoch": 95.43,
      "grad_norm": 1.970491647720337,
      "learning_rate": 2.7130017087431905e-06,
      "loss": 0.4747,
      "step": 233140
    },
    {
      "epoch": 95.44,
      "grad_norm": 1.8272433280944824,
      "learning_rate": 2.7128450594051733e-06,
      "loss": 0.4678,
      "step": 233150
    },
    {
      "epoch": 95.44,
      "grad_norm": 1.630295991897583,
      "learning_rate": 2.7126884083763886e-06,
      "loss": 0.48,
      "step": 233160
    },
    {
      "epoch": 95.44,
      "grad_norm": 2.3018274307250977,
      "learning_rate": 2.712531755657554e-06,
      "loss": 0.491,
      "step": 233170
    },
    {
      "epoch": 95.45,
      "grad_norm": 2.319201946258545,
      "learning_rate": 2.712375101249387e-06,
      "loss": 0.4917,
      "step": 233180
    },
    {
      "epoch": 95.45,
      "grad_norm": 2.137723922729492,
      "learning_rate": 2.7122184451526065e-06,
      "loss": 0.4693,
      "step": 233190
    },
    {
      "epoch": 95.46,
      "grad_norm": 1.9863638877868652,
      "learning_rate": 2.7120617873679285e-06,
      "loss": 0.4747,
      "step": 233200
    },
    {
      "epoch": 95.46,
      "grad_norm": 1.9534591436386108,
      "learning_rate": 2.711905127896071e-06,
      "loss": 0.488,
      "step": 233210
    },
    {
      "epoch": 95.46,
      "grad_norm": 2.1921567916870117,
      "learning_rate": 2.7117484667377526e-06,
      "loss": 0.4679,
      "step": 233220
    },
    {
      "epoch": 95.47,
      "grad_norm": 1.7790497541427612,
      "learning_rate": 2.7115918038936897e-06,
      "loss": 0.4704,
      "step": 233230
    },
    {
      "epoch": 95.47,
      "grad_norm": 2.6235735416412354,
      "learning_rate": 2.7114351393646003e-06,
      "loss": 0.4736,
      "step": 233240
    },
    {
      "epoch": 95.48,
      "grad_norm": 2.649986982345581,
      "learning_rate": 2.711278473151203e-06,
      "loss": 0.4784,
      "step": 233250
    },
    {
      "epoch": 95.48,
      "grad_norm": 1.7463260889053345,
      "learning_rate": 2.711121805254214e-06,
      "loss": 0.4713,
      "step": 233260
    },
    {
      "epoch": 95.49,
      "grad_norm": 1.8548519611358643,
      "learning_rate": 2.710965135674351e-06,
      "loss": 0.4934,
      "step": 233270
    },
    {
      "epoch": 95.49,
      "grad_norm": 1.934737205505371,
      "learning_rate": 2.7108084644123337e-06,
      "loss": 0.4919,
      "step": 233280
    },
    {
      "epoch": 95.49,
      "grad_norm": 1.9930405616760254,
      "learning_rate": 2.710651791468878e-06,
      "loss": 0.4915,
      "step": 233290
    },
    {
      "epoch": 95.5,
      "grad_norm": 1.741405725479126,
      "learning_rate": 2.7104951168447017e-06,
      "loss": 0.4751,
      "step": 233300
    },
    {
      "epoch": 95.5,
      "grad_norm": 1.9751123189926147,
      "learning_rate": 2.710338440540523e-06,
      "loss": 0.4965,
      "step": 233310
    },
    {
      "epoch": 95.51,
      "grad_norm": 1.5236259698867798,
      "learning_rate": 2.7101817625570597e-06,
      "loss": 0.484,
      "step": 233320
    },
    {
      "epoch": 95.51,
      "grad_norm": 2.412391185760498,
      "learning_rate": 2.710025082895029e-06,
      "loss": 0.4703,
      "step": 233330
    },
    {
      "epoch": 95.51,
      "grad_norm": 1.5450252294540405,
      "learning_rate": 2.709868401555149e-06,
      "loss": 0.4819,
      "step": 233340
    },
    {
      "epoch": 95.52,
      "grad_norm": 2.877851963043213,
      "learning_rate": 2.709711718538137e-06,
      "loss": 0.494,
      "step": 233350
    },
    {
      "epoch": 95.52,
      "grad_norm": 1.7621910572052002,
      "learning_rate": 2.709555033844711e-06,
      "loss": 0.476,
      "step": 233360
    },
    {
      "epoch": 95.53,
      "grad_norm": 1.6957005262374878,
      "learning_rate": 2.709398347475589e-06,
      "loss": 0.4713,
      "step": 233370
    },
    {
      "epoch": 95.53,
      "grad_norm": 1.9155595302581787,
      "learning_rate": 2.7092416594314886e-06,
      "loss": 0.4755,
      "step": 233380
    },
    {
      "epoch": 95.53,
      "grad_norm": 1.7332769632339478,
      "learning_rate": 2.7090849697131272e-06,
      "loss": 0.501,
      "step": 233390
    },
    {
      "epoch": 95.54,
      "grad_norm": 2.1262168884277344,
      "learning_rate": 2.708928278321223e-06,
      "loss": 0.4868,
      "step": 233400
    },
    {
      "epoch": 95.54,
      "grad_norm": 2.0019283294677734,
      "learning_rate": 2.708771585256494e-06,
      "loss": 0.4752,
      "step": 233410
    },
    {
      "epoch": 95.55,
      "grad_norm": 2.12036395072937,
      "learning_rate": 2.7086148905196575e-06,
      "loss": 0.4789,
      "step": 233420
    },
    {
      "epoch": 95.55,
      "grad_norm": 1.954718828201294,
      "learning_rate": 2.7084581941114307e-06,
      "loss": 0.4727,
      "step": 233430
    },
    {
      "epoch": 95.55,
      "grad_norm": 1.6462570428848267,
      "learning_rate": 2.708301496032533e-06,
      "loss": 0.4933,
      "step": 233440
    },
    {
      "epoch": 95.56,
      "grad_norm": 1.8184412717819214,
      "learning_rate": 2.7081447962836817e-06,
      "loss": 0.4937,
      "step": 233450
    },
    {
      "epoch": 95.56,
      "grad_norm": 1.6461175680160522,
      "learning_rate": 2.7079880948655935e-06,
      "loss": 0.487,
      "step": 233460
    },
    {
      "epoch": 95.57,
      "grad_norm": 1.7808804512023926,
      "learning_rate": 2.707831391778987e-06,
      "loss": 0.491,
      "step": 233470
    },
    {
      "epoch": 95.57,
      "grad_norm": 1.8038103580474854,
      "learning_rate": 2.70767468702458e-06,
      "loss": 0.5146,
      "step": 233480
    },
    {
      "epoch": 95.58,
      "grad_norm": 1.9635480642318726,
      "learning_rate": 2.707517980603091e-06,
      "loss": 0.4881,
      "step": 233490
    },
    {
      "epoch": 95.58,
      "grad_norm": 1.5736732482910156,
      "learning_rate": 2.707361272515237e-06,
      "loss": 0.4725,
      "step": 233500
    },
    {
      "epoch": 95.58,
      "grad_norm": 1.7197866439819336,
      "learning_rate": 2.707204562761736e-06,
      "loss": 0.4927,
      "step": 233510
    },
    {
      "epoch": 95.59,
      "grad_norm": 2.2686362266540527,
      "learning_rate": 2.707047851343306e-06,
      "loss": 0.4787,
      "step": 233520
    },
    {
      "epoch": 95.59,
      "grad_norm": 1.8775790929794312,
      "learning_rate": 2.7068911382606648e-06,
      "loss": 0.4678,
      "step": 233530
    },
    {
      "epoch": 95.6,
      "grad_norm": 1.7389981746673584,
      "learning_rate": 2.7067344235145305e-06,
      "loss": 0.4853,
      "step": 233540
    },
    {
      "epoch": 95.6,
      "grad_norm": 1.6628016233444214,
      "learning_rate": 2.70657770710562e-06,
      "loss": 0.4655,
      "step": 233550
    },
    {
      "epoch": 95.6,
      "grad_norm": 2.2879347801208496,
      "learning_rate": 2.706420989034652e-06,
      "loss": 0.4859,
      "step": 233560
    },
    {
      "epoch": 95.61,
      "grad_norm": 1.8149172067642212,
      "learning_rate": 2.7062642693023447e-06,
      "loss": 0.508,
      "step": 233570
    },
    {
      "epoch": 95.61,
      "grad_norm": 1.6313990354537964,
      "learning_rate": 2.706107547909416e-06,
      "loss": 0.4885,
      "step": 233580
    },
    {
      "epoch": 95.62,
      "grad_norm": 1.743862271308899,
      "learning_rate": 2.705950824856583e-06,
      "loss": 0.4782,
      "step": 233590
    },
    {
      "epoch": 95.62,
      "grad_norm": 2.258777618408203,
      "learning_rate": 2.7057941001445646e-06,
      "loss": 0.4693,
      "step": 233600
    },
    {
      "epoch": 95.62,
      "grad_norm": 1.9017735719680786,
      "learning_rate": 2.7056373737740783e-06,
      "loss": 0.4889,
      "step": 233610
    },
    {
      "epoch": 95.63,
      "grad_norm": 1.4909271001815796,
      "learning_rate": 2.7054806457458415e-06,
      "loss": 0.5012,
      "step": 233620
    },
    {
      "epoch": 95.63,
      "grad_norm": 1.9099783897399902,
      "learning_rate": 2.705323916060573e-06,
      "loss": 0.4747,
      "step": 233630
    },
    {
      "epoch": 95.64,
      "grad_norm": 1.9048243761062622,
      "learning_rate": 2.7051671847189903e-06,
      "loss": 0.4826,
      "step": 233640
    },
    {
      "epoch": 95.64,
      "grad_norm": 1.631613850593567,
      "learning_rate": 2.705010451721811e-06,
      "loss": 0.5052,
      "step": 233650
    },
    {
      "epoch": 95.64,
      "grad_norm": 1.5836373567581177,
      "learning_rate": 2.704853717069754e-06,
      "loss": 0.505,
      "step": 233660
    },
    {
      "epoch": 95.65,
      "grad_norm": 2.003059148788452,
      "learning_rate": 2.7046969807635364e-06,
      "loss": 0.4778,
      "step": 233670
    },
    {
      "epoch": 95.65,
      "grad_norm": 1.689616084098816,
      "learning_rate": 2.704540242803877e-06,
      "loss": 0.4809,
      "step": 233680
    },
    {
      "epoch": 95.66,
      "grad_norm": 2.397732734680176,
      "learning_rate": 2.7043835031914933e-06,
      "loss": 0.4919,
      "step": 233690
    },
    {
      "epoch": 95.66,
      "grad_norm": 1.782380223274231,
      "learning_rate": 2.7042267619271033e-06,
      "loss": 0.4781,
      "step": 233700
    },
    {
      "epoch": 95.67,
      "grad_norm": 1.6071261167526245,
      "learning_rate": 2.7040700190114246e-06,
      "loss": 0.482,
      "step": 233710
    },
    {
      "epoch": 95.67,
      "grad_norm": 1.3985170125961304,
      "learning_rate": 2.703913274445176e-06,
      "loss": 0.514,
      "step": 233720
    },
    {
      "epoch": 95.67,
      "grad_norm": 2.0495386123657227,
      "learning_rate": 2.7037565282290755e-06,
      "loss": 0.4721,
      "step": 233730
    },
    {
      "epoch": 95.68,
      "grad_norm": 1.9988406896591187,
      "learning_rate": 2.703599780363841e-06,
      "loss": 0.4809,
      "step": 233740
    },
    {
      "epoch": 95.68,
      "grad_norm": 2.0755279064178467,
      "learning_rate": 2.70344303085019e-06,
      "loss": 0.488,
      "step": 233750
    },
    {
      "epoch": 95.69,
      "grad_norm": 1.9918111562728882,
      "learning_rate": 2.703286279688841e-06,
      "loss": 0.4924,
      "step": 233760
    },
    {
      "epoch": 95.69,
      "grad_norm": 2.1152844429016113,
      "learning_rate": 2.7031295268805117e-06,
      "loss": 0.4637,
      "step": 233770
    },
    {
      "epoch": 95.69,
      "grad_norm": 2.5748109817504883,
      "learning_rate": 2.7029727724259213e-06,
      "loss": 0.4811,
      "step": 233780
    },
    {
      "epoch": 95.7,
      "grad_norm": 1.8366013765335083,
      "learning_rate": 2.7028160163257865e-06,
      "loss": 0.4939,
      "step": 233790
    },
    {
      "epoch": 95.7,
      "grad_norm": 2.1705362796783447,
      "learning_rate": 2.702659258580826e-06,
      "loss": 0.4826,
      "step": 233800
    },
    {
      "epoch": 95.71,
      "grad_norm": 2.2980270385742188,
      "learning_rate": 2.7025024991917575e-06,
      "loss": 0.4877,
      "step": 233810
    },
    {
      "epoch": 95.71,
      "grad_norm": 1.9254851341247559,
      "learning_rate": 2.7023457381592994e-06,
      "loss": 0.4763,
      "step": 233820
    },
    {
      "epoch": 95.71,
      "grad_norm": 1.9887696504592896,
      "learning_rate": 2.70218897548417e-06,
      "loss": 0.4873,
      "step": 233830
    },
    {
      "epoch": 95.72,
      "grad_norm": 1.7546433210372925,
      "learning_rate": 2.7020322111670868e-06,
      "loss": 0.4907,
      "step": 233840
    },
    {
      "epoch": 95.72,
      "grad_norm": 1.568496823310852,
      "learning_rate": 2.7018754452087687e-06,
      "loss": 0.4889,
      "step": 233850
    },
    {
      "epoch": 95.73,
      "grad_norm": 1.7572879791259766,
      "learning_rate": 2.701718677609933e-06,
      "loss": 0.4801,
      "step": 233860
    },
    {
      "epoch": 95.73,
      "grad_norm": 2.0330629348754883,
      "learning_rate": 2.7015619083712986e-06,
      "loss": 0.48,
      "step": 233870
    },
    {
      "epoch": 95.73,
      "grad_norm": 1.746604561805725,
      "learning_rate": 2.7014051374935834e-06,
      "loss": 0.4741,
      "step": 233880
    },
    {
      "epoch": 95.74,
      "grad_norm": 2.152052164077759,
      "learning_rate": 2.701248364977505e-06,
      "loss": 0.4857,
      "step": 233890
    },
    {
      "epoch": 95.74,
      "grad_norm": 1.7431930303573608,
      "learning_rate": 2.701091590823782e-06,
      "loss": 0.4793,
      "step": 233900
    },
    {
      "epoch": 95.75,
      "grad_norm": 1.5170620679855347,
      "learning_rate": 2.7009348150331326e-06,
      "loss": 0.475,
      "step": 233910
    },
    {
      "epoch": 95.75,
      "grad_norm": 1.9258147478103638,
      "learning_rate": 2.700778037606275e-06,
      "loss": 0.4942,
      "step": 233920
    },
    {
      "epoch": 95.76,
      "grad_norm": 2.2759222984313965,
      "learning_rate": 2.7006212585439263e-06,
      "loss": 0.4894,
      "step": 233930
    },
    {
      "epoch": 95.76,
      "grad_norm": 1.844780445098877,
      "learning_rate": 2.700464477846807e-06,
      "loss": 0.4754,
      "step": 233940
    },
    {
      "epoch": 95.76,
      "grad_norm": 1.605544924736023,
      "learning_rate": 2.7003076955156333e-06,
      "loss": 0.4817,
      "step": 233950
    },
    {
      "epoch": 95.77,
      "grad_norm": 1.488604187965393,
      "learning_rate": 2.700150911551124e-06,
      "loss": 0.4858,
      "step": 233960
    },
    {
      "epoch": 95.77,
      "grad_norm": 1.8061549663543701,
      "learning_rate": 2.6999941259539973e-06,
      "loss": 0.4858,
      "step": 233970
    },
    {
      "epoch": 95.78,
      "grad_norm": 2.575291395187378,
      "learning_rate": 2.6998373387249714e-06,
      "loss": 0.488,
      "step": 233980
    },
    {
      "epoch": 95.78,
      "grad_norm": 1.8669991493225098,
      "learning_rate": 2.699680549864765e-06,
      "loss": 0.4627,
      "step": 233990
    },
    {
      "epoch": 95.78,
      "grad_norm": 1.7554155588150024,
      "learning_rate": 2.699523759374095e-06,
      "loss": 0.4777,
      "step": 234000
    },
    {
      "epoch": 95.79,
      "grad_norm": 2.604459524154663,
      "learning_rate": 2.6993669672536814e-06,
      "loss": 0.4899,
      "step": 234010
    },
    {
      "epoch": 95.79,
      "grad_norm": 2.5885696411132812,
      "learning_rate": 2.699210173504241e-06,
      "loss": 0.4776,
      "step": 234020
    },
    {
      "epoch": 95.8,
      "grad_norm": 1.806822657585144,
      "learning_rate": 2.699053378126493e-06,
      "loss": 0.4667,
      "step": 234030
    },
    {
      "epoch": 95.8,
      "grad_norm": 1.825075387954712,
      "learning_rate": 2.6988965811211552e-06,
      "loss": 0.4675,
      "step": 234040
    },
    {
      "epoch": 95.8,
      "grad_norm": 1.850636601448059,
      "learning_rate": 2.6987397824889457e-06,
      "loss": 0.4775,
      "step": 234050
    },
    {
      "epoch": 95.81,
      "grad_norm": 1.9922293424606323,
      "learning_rate": 2.698582982230583e-06,
      "loss": 0.4894,
      "step": 234060
    },
    {
      "epoch": 95.81,
      "grad_norm": 1.7875009775161743,
      "learning_rate": 2.6984261803467854e-06,
      "loss": 0.4922,
      "step": 234070
    },
    {
      "epoch": 95.82,
      "grad_norm": 1.861228585243225,
      "learning_rate": 2.6982693768382708e-06,
      "loss": 0.4754,
      "step": 234080
    },
    {
      "epoch": 95.82,
      "grad_norm": 1.8656110763549805,
      "learning_rate": 2.6981125717057585e-06,
      "loss": 0.4628,
      "step": 234090
    },
    {
      "epoch": 95.82,
      "grad_norm": 2.493995189666748,
      "learning_rate": 2.697955764949966e-06,
      "loss": 0.4902,
      "step": 234100
    },
    {
      "epoch": 95.83,
      "grad_norm": 1.7056788206100464,
      "learning_rate": 2.6977989565716116e-06,
      "loss": 0.5001,
      "step": 234110
    },
    {
      "epoch": 95.83,
      "grad_norm": 2.308791399002075,
      "learning_rate": 2.6976421465714135e-06,
      "loss": 0.4706,
      "step": 234120
    },
    {
      "epoch": 95.84,
      "grad_norm": 1.9280011653900146,
      "learning_rate": 2.6974853349500906e-06,
      "loss": 0.4877,
      "step": 234130
    },
    {
      "epoch": 95.84,
      "grad_norm": 2.273669958114624,
      "learning_rate": 2.6973285217083605e-06,
      "loss": 0.4864,
      "step": 234140
    },
    {
      "epoch": 95.85,
      "grad_norm": 2.360285520553589,
      "learning_rate": 2.697171706846942e-06,
      "loss": 0.475,
      "step": 234150
    },
    {
      "epoch": 95.85,
      "grad_norm": 1.7148809432983398,
      "learning_rate": 2.697014890366554e-06,
      "loss": 0.4556,
      "step": 234160
    },
    {
      "epoch": 95.85,
      "grad_norm": 1.9740182161331177,
      "learning_rate": 2.6968580722679147e-06,
      "loss": 0.4834,
      "step": 234170
    },
    {
      "epoch": 95.86,
      "grad_norm": 1.6882582902908325,
      "learning_rate": 2.696701252551741e-06,
      "loss": 0.475,
      "step": 234180
    },
    {
      "epoch": 95.86,
      "grad_norm": 1.9977679252624512,
      "learning_rate": 2.696544431218752e-06,
      "loss": 0.4515,
      "step": 234190
    },
    {
      "epoch": 95.87,
      "grad_norm": 2.479418992996216,
      "learning_rate": 2.696387608269667e-06,
      "loss": 0.4892,
      "step": 234200
    },
    {
      "epoch": 95.87,
      "grad_norm": 2.518425941467285,
      "learning_rate": 2.696230783705204e-06,
      "loss": 0.4967,
      "step": 234210
    },
    {
      "epoch": 95.87,
      "grad_norm": 1.570151686668396,
      "learning_rate": 2.6960739575260797e-06,
      "loss": 0.4724,
      "step": 234220
    },
    {
      "epoch": 95.88,
      "grad_norm": 1.804864764213562,
      "learning_rate": 2.695917129733016e-06,
      "loss": 0.5087,
      "step": 234230
    },
    {
      "epoch": 95.88,
      "grad_norm": 1.4896576404571533,
      "learning_rate": 2.695760300326728e-06,
      "loss": 0.4656,
      "step": 234240
    },
    {
      "epoch": 95.89,
      "grad_norm": 1.742487907409668,
      "learning_rate": 2.6956034693079353e-06,
      "loss": 0.4855,
      "step": 234250
    },
    {
      "epoch": 95.89,
      "grad_norm": 1.7452452182769775,
      "learning_rate": 2.6954466366773564e-06,
      "loss": 0.4626,
      "step": 234260
    },
    {
      "epoch": 95.89,
      "grad_norm": 2.0133187770843506,
      "learning_rate": 2.69528980243571e-06,
      "loss": 0.4868,
      "step": 234270
    },
    {
      "epoch": 95.9,
      "grad_norm": 2.761737823486328,
      "learning_rate": 2.695132966583714e-06,
      "loss": 0.4715,
      "step": 234280
    },
    {
      "epoch": 95.9,
      "grad_norm": 2.074214458465576,
      "learning_rate": 2.694976129122087e-06,
      "loss": 0.4929,
      "step": 234290
    },
    {
      "epoch": 95.91,
      "grad_norm": 1.9398119449615479,
      "learning_rate": 2.6948192900515474e-06,
      "loss": 0.4784,
      "step": 234300
    },
    {
      "epoch": 95.91,
      "grad_norm": 2.035886764526367,
      "learning_rate": 2.694662449372814e-06,
      "loss": 0.4926,
      "step": 234310
    },
    {
      "epoch": 95.91,
      "grad_norm": 1.8868844509124756,
      "learning_rate": 2.6945056070866048e-06,
      "loss": 0.4826,
      "step": 234320
    },
    {
      "epoch": 95.92,
      "grad_norm": 2.1690590381622314,
      "learning_rate": 2.6943487631936382e-06,
      "loss": 0.4524,
      "step": 234330
    },
    {
      "epoch": 95.92,
      "grad_norm": 2.4318621158599854,
      "learning_rate": 2.6941919176946337e-06,
      "loss": 0.477,
      "step": 234340
    },
    {
      "epoch": 95.93,
      "grad_norm": 1.699318766593933,
      "learning_rate": 2.6940350705903086e-06,
      "loss": 0.483,
      "step": 234350
    },
    {
      "epoch": 95.93,
      "grad_norm": 1.9701374769210815,
      "learning_rate": 2.693878221881382e-06,
      "loss": 0.4959,
      "step": 234360
    },
    {
      "epoch": 95.94,
      "grad_norm": 2.2404086589813232,
      "learning_rate": 2.693721371568572e-06,
      "loss": 0.4954,
      "step": 234370
    },
    {
      "epoch": 95.94,
      "grad_norm": 1.8342632055282593,
      "learning_rate": 2.693564519652598e-06,
      "loss": 0.4814,
      "step": 234380
    },
    {
      "epoch": 95.94,
      "grad_norm": 1.637623906135559,
      "learning_rate": 2.693407666134177e-06,
      "loss": 0.4787,
      "step": 234390
    },
    {
      "epoch": 95.95,
      "grad_norm": 2.3306257724761963,
      "learning_rate": 2.6932508110140295e-06,
      "loss": 0.493,
      "step": 234400
    },
    {
      "epoch": 95.95,
      "grad_norm": 2.271472692489624,
      "learning_rate": 2.6930939542928714e-06,
      "loss": 0.4781,
      "step": 234410
    },
    {
      "epoch": 95.96,
      "grad_norm": 2.0375049114227295,
      "learning_rate": 2.692937095971423e-06,
      "loss": 0.4993,
      "step": 234420
    },
    {
      "epoch": 95.96,
      "grad_norm": 2.149827003479004,
      "learning_rate": 2.6927802360504033e-06,
      "loss": 0.4656,
      "step": 234430
    },
    {
      "epoch": 95.96,
      "grad_norm": 2.4428961277008057,
      "learning_rate": 2.6926233745305303e-06,
      "loss": 0.4926,
      "step": 234440
    },
    {
      "epoch": 95.97,
      "grad_norm": 2.0018882751464844,
      "learning_rate": 2.692466511412522e-06,
      "loss": 0.492,
      "step": 234450
    },
    {
      "epoch": 95.97,
      "grad_norm": 1.7961225509643555,
      "learning_rate": 2.692309646697098e-06,
      "loss": 0.4777,
      "step": 234460
    },
    {
      "epoch": 95.98,
      "grad_norm": 2.42277455329895,
      "learning_rate": 2.6921527803849757e-06,
      "loss": 0.4824,
      "step": 234470
    },
    {
      "epoch": 95.98,
      "grad_norm": 2.1996114253997803,
      "learning_rate": 2.691995912476874e-06,
      "loss": 0.5005,
      "step": 234480
    },
    {
      "epoch": 95.98,
      "grad_norm": 2.2496230602264404,
      "learning_rate": 2.691839042973512e-06,
      "loss": 0.4917,
      "step": 234490
    },
    {
      "epoch": 95.99,
      "grad_norm": 1.5969048738479614,
      "learning_rate": 2.691682171875607e-06,
      "loss": 0.4847,
      "step": 234500
    },
    {
      "epoch": 95.99,
      "grad_norm": 1.7284438610076904,
      "learning_rate": 2.69152529918388e-06,
      "loss": 0.4913,
      "step": 234510
    },
    {
      "epoch": 96.0,
      "grad_norm": 1.8315458297729492,
      "learning_rate": 2.6913684248990475e-06,
      "loss": 0.4811,
      "step": 234520
    },
    {
      "epoch": 96.0,
      "eval_loss": 0.4834519624710083,
      "eval_runtime": 57.7439,
      "eval_samples_per_second": 59.729,
      "eval_steps_per_second": 7.481,
      "step": 234528
    },
    {
      "epoch": 96.0,
      "grad_norm": 2.1060421466827393,
      "learning_rate": 2.6912115490218298e-06,
      "loss": 0.4619,
      "step": 234530
    },
    {
      "epoch": 96.0,
      "grad_norm": 1.7746045589447021,
      "learning_rate": 2.6910546715529435e-06,
      "loss": 0.4749,
      "step": 234540
    },
    {
      "epoch": 96.01,
      "grad_norm": 1.9978982210159302,
      "learning_rate": 2.6908977924931087e-06,
      "loss": 0.4785,
      "step": 234550
    },
    {
      "epoch": 96.01,
      "grad_norm": 1.7240784168243408,
      "learning_rate": 2.690740911843043e-06,
      "loss": 0.4761,
      "step": 234560
    },
    {
      "epoch": 96.02,
      "grad_norm": 1.9477086067199707,
      "learning_rate": 2.690584029603466e-06,
      "loss": 0.4797,
      "step": 234570
    },
    {
      "epoch": 96.02,
      "grad_norm": 1.7226760387420654,
      "learning_rate": 2.6904271457750968e-06,
      "loss": 0.4804,
      "step": 234580
    },
    {
      "epoch": 96.03,
      "grad_norm": 2.276540994644165,
      "learning_rate": 2.6902702603586524e-06,
      "loss": 0.4552,
      "step": 234590
    },
    {
      "epoch": 96.03,
      "grad_norm": 1.9472557306289673,
      "learning_rate": 2.6901133733548526e-06,
      "loss": 0.4831,
      "step": 234600
    },
    {
      "epoch": 96.03,
      "grad_norm": 2.308745861053467,
      "learning_rate": 2.689956484764416e-06,
      "loss": 0.4897,
      "step": 234610
    },
    {
      "epoch": 96.04,
      "grad_norm": 2.166912078857422,
      "learning_rate": 2.6897995945880607e-06,
      "loss": 0.5014,
      "step": 234620
    },
    {
      "epoch": 96.04,
      "grad_norm": 1.9216102361679077,
      "learning_rate": 2.6896427028265065e-06,
      "loss": 0.4976,
      "step": 234630
    },
    {
      "epoch": 96.05,
      "grad_norm": 1.6359915733337402,
      "learning_rate": 2.689485809480471e-06,
      "loss": 0.475,
      "step": 234640
    },
    {
      "epoch": 96.05,
      "grad_norm": 1.7659050226211548,
      "learning_rate": 2.6893289145506734e-06,
      "loss": 0.4822,
      "step": 234650
    },
    {
      "epoch": 96.05,
      "grad_norm": 1.9892244338989258,
      "learning_rate": 2.689172018037832e-06,
      "loss": 0.4961,
      "step": 234660
    },
    {
      "epoch": 96.06,
      "grad_norm": 2.4539129734039307,
      "learning_rate": 2.689015119942666e-06,
      "loss": 0.4965,
      "step": 234670
    },
    {
      "epoch": 96.06,
      "grad_norm": 2.0366954803466797,
      "learning_rate": 2.688858220265895e-06,
      "loss": 0.4861,
      "step": 234680
    },
    {
      "epoch": 96.07,
      "grad_norm": 2.108339309692383,
      "learning_rate": 2.6887013190082344e-06,
      "loss": 0.4913,
      "step": 234690
    },
    {
      "epoch": 96.07,
      "grad_norm": 1.7558597326278687,
      "learning_rate": 2.688544416170407e-06,
      "loss": 0.4785,
      "step": 234700
    },
    {
      "epoch": 96.07,
      "grad_norm": 1.787336826324463,
      "learning_rate": 2.6883875117531293e-06,
      "loss": 0.4755,
      "step": 234710
    },
    {
      "epoch": 96.08,
      "grad_norm": 2.0784175395965576,
      "learning_rate": 2.6882306057571203e-06,
      "loss": 0.4818,
      "step": 234720
    },
    {
      "epoch": 96.08,
      "grad_norm": 2.3802332878112793,
      "learning_rate": 2.688073698183099e-06,
      "loss": 0.4848,
      "step": 234730
    },
    {
      "epoch": 96.09,
      "grad_norm": 1.9340332746505737,
      "learning_rate": 2.687916789031785e-06,
      "loss": 0.512,
      "step": 234740
    },
    {
      "epoch": 96.09,
      "grad_norm": 3.189432382583618,
      "learning_rate": 2.687759878303896e-06,
      "loss": 0.4656,
      "step": 234750
    },
    {
      "epoch": 96.09,
      "grad_norm": 1.9833767414093018,
      "learning_rate": 2.6876029660001498e-06,
      "loss": 0.5065,
      "step": 234760
    },
    {
      "epoch": 96.1,
      "grad_norm": 1.6643706560134888,
      "learning_rate": 2.6874460521212672e-06,
      "loss": 0.4713,
      "step": 234770
    },
    {
      "epoch": 96.1,
      "grad_norm": 2.1034350395202637,
      "learning_rate": 2.687289136667966e-06,
      "loss": 0.4649,
      "step": 234780
    },
    {
      "epoch": 96.11,
      "grad_norm": 2.5201528072357178,
      "learning_rate": 2.6871322196409657e-06,
      "loss": 0.508,
      "step": 234790
    },
    {
      "epoch": 96.11,
      "grad_norm": 2.7911322116851807,
      "learning_rate": 2.686975301040984e-06,
      "loss": 0.4996,
      "step": 234800
    },
    {
      "epoch": 96.12,
      "grad_norm": 1.8804742097854614,
      "learning_rate": 2.686818380868741e-06,
      "loss": 0.4937,
      "step": 234810
    },
    {
      "epoch": 96.12,
      "grad_norm": 2.0037314891815186,
      "learning_rate": 2.6866614591249552e-06,
      "loss": 0.4754,
      "step": 234820
    },
    {
      "epoch": 96.12,
      "grad_norm": 2.055429458618164,
      "learning_rate": 2.6865045358103446e-06,
      "loss": 0.4824,
      "step": 234830
    },
    {
      "epoch": 96.13,
      "grad_norm": 1.7801270484924316,
      "learning_rate": 2.6863476109256284e-06,
      "loss": 0.4745,
      "step": 234840
    },
    {
      "epoch": 96.13,
      "grad_norm": 1.6366591453552246,
      "learning_rate": 2.6861906844715258e-06,
      "loss": 0.472,
      "step": 234850
    },
    {
      "epoch": 96.14,
      "grad_norm": 1.4218307733535767,
      "learning_rate": 2.6860337564487553e-06,
      "loss": 0.4851,
      "step": 234860
    },
    {
      "epoch": 96.14,
      "grad_norm": 1.9957246780395508,
      "learning_rate": 2.685876826858036e-06,
      "loss": 0.4734,
      "step": 234870
    },
    {
      "epoch": 96.14,
      "grad_norm": 1.7363364696502686,
      "learning_rate": 2.6857198957000865e-06,
      "loss": 0.4838,
      "step": 234880
    },
    {
      "epoch": 96.15,
      "grad_norm": 2.0159800052642822,
      "learning_rate": 2.6855629629756264e-06,
      "loss": 0.4792,
      "step": 234890
    },
    {
      "epoch": 96.15,
      "grad_norm": 1.675647258758545,
      "learning_rate": 2.6854060286853737e-06,
      "loss": 0.4723,
      "step": 234900
    },
    {
      "epoch": 96.16,
      "grad_norm": 2.087796449661255,
      "learning_rate": 2.6852490928300477e-06,
      "loss": 0.4941,
      "step": 234910
    },
    {
      "epoch": 96.16,
      "grad_norm": 1.7232890129089355,
      "learning_rate": 2.685092155410367e-06,
      "loss": 0.4751,
      "step": 234920
    },
    {
      "epoch": 96.16,
      "grad_norm": 1.8570945262908936,
      "learning_rate": 2.684935216427051e-06,
      "loss": 0.4807,
      "step": 234930
    },
    {
      "epoch": 96.17,
      "grad_norm": 1.7465598583221436,
      "learning_rate": 2.6847782758808183e-06,
      "loss": 0.4817,
      "step": 234940
    },
    {
      "epoch": 96.17,
      "grad_norm": 1.8422133922576904,
      "learning_rate": 2.6846213337723876e-06,
      "loss": 0.4856,
      "step": 234950
    },
    {
      "epoch": 96.18,
      "grad_norm": 1.875925064086914,
      "learning_rate": 2.684464390102479e-06,
      "loss": 0.4638,
      "step": 234960
    },
    {
      "epoch": 96.18,
      "grad_norm": 1.7506263256072998,
      "learning_rate": 2.6843074448718093e-06,
      "loss": 0.4793,
      "step": 234970
    },
    {
      "epoch": 96.19,
      "grad_norm": 2.026839256286621,
      "learning_rate": 2.6841504980810995e-06,
      "loss": 0.4696,
      "step": 234980
    },
    {
      "epoch": 96.19,
      "grad_norm": 1.9032434225082397,
      "learning_rate": 2.683993549731067e-06,
      "loss": 0.4602,
      "step": 234990
    },
    {
      "epoch": 96.19,
      "grad_norm": 1.9608694314956665,
      "learning_rate": 2.6838365998224325e-06,
      "loss": 0.4656,
      "step": 235000
    },
    {
      "epoch": 96.2,
      "grad_norm": 1.8824409246444702,
      "learning_rate": 2.6836796483559136e-06,
      "loss": 0.4809,
      "step": 235010
    },
    {
      "epoch": 96.2,
      "grad_norm": 2.048530340194702,
      "learning_rate": 2.6835226953322293e-06,
      "loss": 0.4671,
      "step": 235020
    },
    {
      "epoch": 96.21,
      "grad_norm": 2.1440718173980713,
      "learning_rate": 2.6833657407520994e-06,
      "loss": 0.4865,
      "step": 235030
    },
    {
      "epoch": 96.21,
      "grad_norm": 1.3259098529815674,
      "learning_rate": 2.683208784616243e-06,
      "loss": 0.4714,
      "step": 235040
    },
    {
      "epoch": 96.21,
      "grad_norm": 1.9729745388031006,
      "learning_rate": 2.683051826925377e-06,
      "loss": 0.4919,
      "step": 235050
    },
    {
      "epoch": 96.22,
      "grad_norm": 2.080845594406128,
      "learning_rate": 2.682894867680223e-06,
      "loss": 0.478,
      "step": 235060
    },
    {
      "epoch": 96.22,
      "grad_norm": 1.9781169891357422,
      "learning_rate": 2.6827379068814977e-06,
      "loss": 0.4863,
      "step": 235070
    },
    {
      "epoch": 96.23,
      "grad_norm": 2.2221052646636963,
      "learning_rate": 2.6825809445299215e-06,
      "loss": 0.4808,
      "step": 235080
    },
    {
      "epoch": 96.23,
      "grad_norm": 2.3222689628601074,
      "learning_rate": 2.682423980626214e-06,
      "loss": 0.4729,
      "step": 235090
    },
    {
      "epoch": 96.23,
      "grad_norm": 1.7489689588546753,
      "learning_rate": 2.6822670151710926e-06,
      "loss": 0.4803,
      "step": 235100
    },
    {
      "epoch": 96.24,
      "grad_norm": 1.7640280723571777,
      "learning_rate": 2.6821100481652787e-06,
      "loss": 0.4801,
      "step": 235110
    },
    {
      "epoch": 96.24,
      "grad_norm": 2.406682252883911,
      "learning_rate": 2.681953079609489e-06,
      "loss": 0.4655,
      "step": 235120
    },
    {
      "epoch": 96.25,
      "grad_norm": 1.6703203916549683,
      "learning_rate": 2.681796109504443e-06,
      "loss": 0.5032,
      "step": 235130
    },
    {
      "epoch": 96.25,
      "grad_norm": 1.8418818712234497,
      "learning_rate": 2.68163913785086e-06,
      "loss": 0.4698,
      "step": 235140
    },
    {
      "epoch": 96.25,
      "grad_norm": 1.8591831922531128,
      "learning_rate": 2.6814821646494593e-06,
      "loss": 0.4868,
      "step": 235150
    },
    {
      "epoch": 96.26,
      "grad_norm": 2.093357563018799,
      "learning_rate": 2.6813251899009596e-06,
      "loss": 0.4892,
      "step": 235160
    },
    {
      "epoch": 96.26,
      "grad_norm": 1.5742238759994507,
      "learning_rate": 2.6811682136060813e-06,
      "loss": 0.4805,
      "step": 235170
    },
    {
      "epoch": 96.27,
      "grad_norm": 2.027331829071045,
      "learning_rate": 2.681011235765542e-06,
      "loss": 0.4763,
      "step": 235180
    },
    {
      "epoch": 96.27,
      "grad_norm": 2.605250835418701,
      "learning_rate": 2.6808542563800613e-06,
      "loss": 0.4977,
      "step": 235190
    },
    {
      "epoch": 96.28,
      "grad_norm": 1.4985378980636597,
      "learning_rate": 2.6806972754503577e-06,
      "loss": 0.4805,
      "step": 235200
    },
    {
      "epoch": 96.28,
      "grad_norm": 2.033216714859009,
      "learning_rate": 2.680540292977151e-06,
      "loss": 0.4789,
      "step": 235210
    },
    {
      "epoch": 96.28,
      "grad_norm": 1.7701499462127686,
      "learning_rate": 2.6803833089611603e-06,
      "loss": 0.5028,
      "step": 235220
    },
    {
      "epoch": 96.29,
      "grad_norm": 1.8872071504592896,
      "learning_rate": 2.6802263234031043e-06,
      "loss": 0.452,
      "step": 235230
    },
    {
      "epoch": 96.29,
      "grad_norm": 2.105532169342041,
      "learning_rate": 2.6800693363037025e-06,
      "loss": 0.4936,
      "step": 235240
    },
    {
      "epoch": 96.3,
      "grad_norm": 1.776778221130371,
      "learning_rate": 2.679912347663674e-06,
      "loss": 0.4799,
      "step": 235250
    },
    {
      "epoch": 96.3,
      "grad_norm": 1.8753550052642822,
      "learning_rate": 2.6797553574837374e-06,
      "loss": 0.489,
      "step": 235260
    },
    {
      "epoch": 96.3,
      "grad_norm": 2.232107400894165,
      "learning_rate": 2.679598365764613e-06,
      "loss": 0.4685,
      "step": 235270
    },
    {
      "epoch": 96.31,
      "grad_norm": 1.7029346227645874,
      "learning_rate": 2.679441372507019e-06,
      "loss": 0.4821,
      "step": 235280
    },
    {
      "epoch": 96.31,
      "grad_norm": 2.0725302696228027,
      "learning_rate": 2.679284377711675e-06,
      "loss": 0.4715,
      "step": 235290
    },
    {
      "epoch": 96.32,
      "grad_norm": 1.870479702949524,
      "learning_rate": 2.6791273813792992e-06,
      "loss": 0.4754,
      "step": 235300
    },
    {
      "epoch": 96.32,
      "grad_norm": 1.5993502140045166,
      "learning_rate": 2.6789703835106117e-06,
      "loss": 0.484,
      "step": 235310
    },
    {
      "epoch": 96.32,
      "grad_norm": 1.7592195272445679,
      "learning_rate": 2.6788133841063322e-06,
      "loss": 0.4748,
      "step": 235320
    },
    {
      "epoch": 96.33,
      "grad_norm": 2.2425222396850586,
      "learning_rate": 2.6786563831671795e-06,
      "loss": 0.476,
      "step": 235330
    },
    {
      "epoch": 96.33,
      "grad_norm": 2.0002217292785645,
      "learning_rate": 2.678499380693872e-06,
      "loss": 0.4634,
      "step": 235340
    },
    {
      "epoch": 96.34,
      "grad_norm": 2.0319936275482178,
      "learning_rate": 2.6783423766871286e-06,
      "loss": 0.4758,
      "step": 235350
    },
    {
      "epoch": 96.34,
      "grad_norm": 1.7535279989242554,
      "learning_rate": 2.67818537114767e-06,
      "loss": 0.4833,
      "step": 235360
    },
    {
      "epoch": 96.34,
      "grad_norm": 1.688707947731018,
      "learning_rate": 2.6780283640762148e-06,
      "loss": 0.4693,
      "step": 235370
    },
    {
      "epoch": 96.35,
      "grad_norm": 2.223186731338501,
      "learning_rate": 2.6778713554734827e-06,
      "loss": 0.4827,
      "step": 235380
    },
    {
      "epoch": 96.35,
      "grad_norm": 1.632118582725525,
      "learning_rate": 2.6777143453401918e-06,
      "loss": 0.4882,
      "step": 235390
    },
    {
      "epoch": 96.36,
      "grad_norm": 1.9765161275863647,
      "learning_rate": 2.6775573336770628e-06,
      "loss": 0.4795,
      "step": 235400
    },
    {
      "epoch": 96.36,
      "grad_norm": 2.0178916454315186,
      "learning_rate": 2.677400320484813e-06,
      "loss": 0.4844,
      "step": 235410
    },
    {
      "epoch": 96.37,
      "grad_norm": 1.90226149559021,
      "learning_rate": 2.6772433057641633e-06,
      "loss": 0.4859,
      "step": 235420
    },
    {
      "epoch": 96.37,
      "grad_norm": 1.8317193984985352,
      "learning_rate": 2.6770862895158324e-06,
      "loss": 0.479,
      "step": 235430
    },
    {
      "epoch": 96.37,
      "grad_norm": 2.3324389457702637,
      "learning_rate": 2.6769292717405386e-06,
      "loss": 0.4724,
      "step": 235440
    },
    {
      "epoch": 96.38,
      "grad_norm": 1.7721953392028809,
      "learning_rate": 2.6767722524390033e-06,
      "loss": 0.5017,
      "step": 235450
    },
    {
      "epoch": 96.38,
      "grad_norm": 1.926222324371338,
      "learning_rate": 2.6766152316119443e-06,
      "loss": 0.496,
      "step": 235460
    },
    {
      "epoch": 96.39,
      "grad_norm": 2.0917208194732666,
      "learning_rate": 2.6764582092600815e-06,
      "loss": 0.4989,
      "step": 235470
    },
    {
      "epoch": 96.39,
      "grad_norm": 2.907165288925171,
      "learning_rate": 2.6763011853841337e-06,
      "loss": 0.4763,
      "step": 235480
    },
    {
      "epoch": 96.39,
      "grad_norm": 1.6991907358169556,
      "learning_rate": 2.676144159984821e-06,
      "loss": 0.4768,
      "step": 235490
    },
    {
      "epoch": 96.4,
      "grad_norm": 2.386307954788208,
      "learning_rate": 2.6759871330628614e-06,
      "loss": 0.4826,
      "step": 235500
    },
    {
      "epoch": 96.4,
      "grad_norm": 1.6629729270935059,
      "learning_rate": 2.675830104618975e-06,
      "loss": 0.5024,
      "step": 235510
    },
    {
      "epoch": 96.41,
      "grad_norm": 1.7054861783981323,
      "learning_rate": 2.6756730746538815e-06,
      "loss": 0.4772,
      "step": 235520
    },
    {
      "epoch": 96.41,
      "grad_norm": 1.5562187433242798,
      "learning_rate": 2.675516043168299e-06,
      "loss": 0.5026,
      "step": 235530
    },
    {
      "epoch": 96.41,
      "grad_norm": 2.755284070968628,
      "learning_rate": 2.6753590101629484e-06,
      "loss": 0.4774,
      "step": 235540
    },
    {
      "epoch": 96.42,
      "grad_norm": 1.8981488943099976,
      "learning_rate": 2.6752019756385475e-06,
      "loss": 0.4822,
      "step": 235550
    },
    {
      "epoch": 96.42,
      "grad_norm": 1.6872082948684692,
      "learning_rate": 2.6750449395958172e-06,
      "loss": 0.4752,
      "step": 235560
    },
    {
      "epoch": 96.43,
      "grad_norm": 2.0026917457580566,
      "learning_rate": 2.674887902035476e-06,
      "loss": 0.4932,
      "step": 235570
    },
    {
      "epoch": 96.43,
      "grad_norm": 1.9123061895370483,
      "learning_rate": 2.674730862958243e-06,
      "loss": 0.4941,
      "step": 235580
    },
    {
      "epoch": 96.43,
      "grad_norm": 1.9550563097000122,
      "learning_rate": 2.6745738223648386e-06,
      "loss": 0.468,
      "step": 235590
    },
    {
      "epoch": 96.44,
      "grad_norm": 2.3426194190979004,
      "learning_rate": 2.6744167802559806e-06,
      "loss": 0.4721,
      "step": 235600
    },
    {
      "epoch": 96.44,
      "grad_norm": 2.0928235054016113,
      "learning_rate": 2.6742597366323893e-06,
      "loss": 0.4758,
      "step": 235610
    },
    {
      "epoch": 96.45,
      "grad_norm": 2.1827144622802734,
      "learning_rate": 2.674102691494785e-06,
      "loss": 0.4809,
      "step": 235620
    },
    {
      "epoch": 96.45,
      "grad_norm": 1.6203410625457764,
      "learning_rate": 2.6739456448438853e-06,
      "loss": 0.4719,
      "step": 235630
    },
    {
      "epoch": 96.46,
      "grad_norm": 1.659193992614746,
      "learning_rate": 2.673788596680411e-06,
      "loss": 0.4846,
      "step": 235640
    },
    {
      "epoch": 96.46,
      "grad_norm": 2.0440447330474854,
      "learning_rate": 2.673631547005081e-06,
      "loss": 0.4788,
      "step": 235650
    },
    {
      "epoch": 96.46,
      "grad_norm": 1.7709665298461914,
      "learning_rate": 2.6734744958186144e-06,
      "loss": 0.4685,
      "step": 235660
    },
    {
      "epoch": 96.47,
      "grad_norm": 2.2757279872894287,
      "learning_rate": 2.6733174431217307e-06,
      "loss": 0.4876,
      "step": 235670
    },
    {
      "epoch": 96.47,
      "grad_norm": 1.6277583837509155,
      "learning_rate": 2.6731603889151505e-06,
      "loss": 0.4845,
      "step": 235680
    },
    {
      "epoch": 96.48,
      "grad_norm": 2.7012739181518555,
      "learning_rate": 2.673003333199592e-06,
      "loss": 0.5101,
      "step": 235690
    },
    {
      "epoch": 96.48,
      "grad_norm": 1.922408938407898,
      "learning_rate": 2.672846275975775e-06,
      "loss": 0.4899,
      "step": 235700
    },
    {
      "epoch": 96.48,
      "grad_norm": 2.1563920974731445,
      "learning_rate": 2.6726892172444186e-06,
      "loss": 0.4852,
      "step": 235710
    },
    {
      "epoch": 96.49,
      "grad_norm": 1.722435712814331,
      "learning_rate": 2.6725321570062425e-06,
      "loss": 0.4918,
      "step": 235720
    },
    {
      "epoch": 96.49,
      "grad_norm": 2.1412460803985596,
      "learning_rate": 2.672375095261966e-06,
      "loss": 0.4837,
      "step": 235730
    },
    {
      "epoch": 96.5,
      "grad_norm": 1.6852707862854004,
      "learning_rate": 2.6722180320123097e-06,
      "loss": 0.4831,
      "step": 235740
    },
    {
      "epoch": 96.5,
      "grad_norm": 1.8396497964859009,
      "learning_rate": 2.6720609672579916e-06,
      "loss": 0.4827,
      "step": 235750
    },
    {
      "epoch": 96.5,
      "grad_norm": 1.9146817922592163,
      "learning_rate": 2.6719039009997326e-06,
      "loss": 0.4621,
      "step": 235760
    },
    {
      "epoch": 96.51,
      "grad_norm": 1.8623883724212646,
      "learning_rate": 2.6717468332382506e-06,
      "loss": 0.4742,
      "step": 235770
    },
    {
      "epoch": 96.51,
      "grad_norm": 1.833858609199524,
      "learning_rate": 2.6715897639742664e-06,
      "loss": 0.4622,
      "step": 235780
    },
    {
      "epoch": 96.52,
      "grad_norm": 1.5996060371398926,
      "learning_rate": 2.6714326932084987e-06,
      "loss": 0.4776,
      "step": 235790
    },
    {
      "epoch": 96.52,
      "grad_norm": 2.258509635925293,
      "learning_rate": 2.6712756209416675e-06,
      "loss": 0.4855,
      "step": 235800
    },
    {
      "epoch": 96.52,
      "grad_norm": 1.9928548336029053,
      "learning_rate": 2.6711185471744915e-06,
      "loss": 0.4664,
      "step": 235810
    },
    {
      "epoch": 96.53,
      "grad_norm": 1.6795964241027832,
      "learning_rate": 2.6709614719076907e-06,
      "loss": 0.4764,
      "step": 235820
    },
    {
      "epoch": 96.53,
      "grad_norm": 2.0313239097595215,
      "learning_rate": 2.6708043951419863e-06,
      "loss": 0.4967,
      "step": 235830
    },
    {
      "epoch": 96.54,
      "grad_norm": 2.2284843921661377,
      "learning_rate": 2.6706473168780958e-06,
      "loss": 0.4746,
      "step": 235840
    },
    {
      "epoch": 96.54,
      "grad_norm": 2.3162777423858643,
      "learning_rate": 2.6704902371167387e-06,
      "loss": 0.4852,
      "step": 235850
    },
    {
      "epoch": 96.55,
      "grad_norm": 1.3684501647949219,
      "learning_rate": 2.6703331558586358e-06,
      "loss": 0.4698,
      "step": 235860
    },
    {
      "epoch": 96.55,
      "grad_norm": 2.0487470626831055,
      "learning_rate": 2.670176073104506e-06,
      "loss": 0.4677,
      "step": 235870
    },
    {
      "epoch": 96.55,
      "grad_norm": 2.294027090072632,
      "learning_rate": 2.670018988855069e-06,
      "loss": 0.4882,
      "step": 235880
    },
    {
      "epoch": 96.56,
      "grad_norm": 2.052043914794922,
      "learning_rate": 2.669861903111044e-06,
      "loss": 0.4926,
      "step": 235890
    },
    {
      "epoch": 96.56,
      "grad_norm": 1.9263213872909546,
      "learning_rate": 2.66970481587315e-06,
      "loss": 0.4996,
      "step": 235900
    },
    {
      "epoch": 96.57,
      "grad_norm": 1.8603079319000244,
      "learning_rate": 2.669547727142109e-06,
      "loss": 0.4909,
      "step": 235910
    },
    {
      "epoch": 96.57,
      "grad_norm": 1.9235228300094604,
      "learning_rate": 2.6693906369186383e-06,
      "loss": 0.4803,
      "step": 235920
    },
    {
      "epoch": 96.57,
      "grad_norm": 1.8648557662963867,
      "learning_rate": 2.6692335452034584e-06,
      "loss": 0.4798,
      "step": 235930
    },
    {
      "epoch": 96.58,
      "grad_norm": 2.5005931854248047,
      "learning_rate": 2.6690764519972888e-06,
      "loss": 0.5039,
      "step": 235940
    },
    {
      "epoch": 96.58,
      "grad_norm": 1.8865238428115845,
      "learning_rate": 2.6689193573008494e-06,
      "loss": 0.4758,
      "step": 235950
    },
    {
      "epoch": 96.59,
      "grad_norm": 2.3299174308776855,
      "learning_rate": 2.668762261114859e-06,
      "loss": 0.4879,
      "step": 235960
    },
    {
      "epoch": 96.59,
      "grad_norm": 1.7156540155410767,
      "learning_rate": 2.668605163440038e-06,
      "loss": 0.4999,
      "step": 235970
    },
    {
      "epoch": 96.59,
      "grad_norm": 2.003295660018921,
      "learning_rate": 2.668448064277106e-06,
      "loss": 0.4965,
      "step": 235980
    },
    {
      "epoch": 96.6,
      "grad_norm": 2.625640630722046,
      "learning_rate": 2.6682909636267826e-06,
      "loss": 0.4686,
      "step": 235990
    },
    {
      "epoch": 96.6,
      "grad_norm": 1.6028465032577515,
      "learning_rate": 2.668133861489787e-06,
      "loss": 0.4845,
      "step": 236000
    },
    {
      "epoch": 96.61,
      "grad_norm": 2.3106184005737305,
      "learning_rate": 2.6679767578668383e-06,
      "loss": 0.505,
      "step": 236010
    },
    {
      "epoch": 96.61,
      "grad_norm": 2.048518419265747,
      "learning_rate": 2.6678196527586576e-06,
      "loss": 0.4977,
      "step": 236020
    },
    {
      "epoch": 96.61,
      "grad_norm": 1.8282874822616577,
      "learning_rate": 2.6676625461659644e-06,
      "loss": 0.4898,
      "step": 236030
    },
    {
      "epoch": 96.62,
      "grad_norm": 2.1161906719207764,
      "learning_rate": 2.6675054380894783e-06,
      "loss": 0.4819,
      "step": 236040
    },
    {
      "epoch": 96.62,
      "grad_norm": 2.0150680541992188,
      "learning_rate": 2.667348328529919e-06,
      "loss": 0.4852,
      "step": 236050
    },
    {
      "epoch": 96.63,
      "grad_norm": 2.346592426300049,
      "learning_rate": 2.667191217488005e-06,
      "loss": 0.5026,
      "step": 236060
    },
    {
      "epoch": 96.63,
      "grad_norm": 1.8173869848251343,
      "learning_rate": 2.667034104964457e-06,
      "loss": 0.4941,
      "step": 236070
    },
    {
      "epoch": 96.64,
      "grad_norm": 1.5761030912399292,
      "learning_rate": 2.6668769909599945e-06,
      "loss": 0.4831,
      "step": 236080
    },
    {
      "epoch": 96.64,
      "grad_norm": 1.8199565410614014,
      "learning_rate": 2.666719875475337e-06,
      "loss": 0.465,
      "step": 236090
    },
    {
      "epoch": 96.64,
      "grad_norm": 1.9256374835968018,
      "learning_rate": 2.666562758511205e-06,
      "loss": 0.4823,
      "step": 236100
    },
    {
      "epoch": 96.65,
      "grad_norm": 1.660780668258667,
      "learning_rate": 2.6664056400683176e-06,
      "loss": 0.4829,
      "step": 236110
    },
    {
      "epoch": 96.65,
      "grad_norm": 1.9162895679473877,
      "learning_rate": 2.6662485201473946e-06,
      "loss": 0.4793,
      "step": 236120
    },
    {
      "epoch": 96.66,
      "grad_norm": 2.281989336013794,
      "learning_rate": 2.6660913987491563e-06,
      "loss": 0.4708,
      "step": 236130
    },
    {
      "epoch": 96.66,
      "grad_norm": 2.3185946941375732,
      "learning_rate": 2.665934275874322e-06,
      "loss": 0.4633,
      "step": 236140
    },
    {
      "epoch": 96.66,
      "grad_norm": 2.2537286281585693,
      "learning_rate": 2.6657771515236105e-06,
      "loss": 0.4803,
      "step": 236150
    },
    {
      "epoch": 96.67,
      "grad_norm": 2.025484323501587,
      "learning_rate": 2.665620025697743e-06,
      "loss": 0.4921,
      "step": 236160
    },
    {
      "epoch": 96.67,
      "grad_norm": 1.6530499458312988,
      "learning_rate": 2.665462898397439e-06,
      "loss": 0.469,
      "step": 236170
    },
    {
      "epoch": 96.68,
      "grad_norm": 1.8340799808502197,
      "learning_rate": 2.6653057696234177e-06,
      "loss": 0.4889,
      "step": 236180
    },
    {
      "epoch": 96.68,
      "grad_norm": 1.7505030632019043,
      "learning_rate": 2.665148639376399e-06,
      "loss": 0.4901,
      "step": 236190
    },
    {
      "epoch": 96.68,
      "grad_norm": 1.9772802591323853,
      "learning_rate": 2.6649915076571037e-06,
      "loss": 0.4705,
      "step": 236200
    },
    {
      "epoch": 96.69,
      "grad_norm": 1.75623619556427,
      "learning_rate": 2.6648343744662504e-06,
      "loss": 0.4807,
      "step": 236210
    },
    {
      "epoch": 96.69,
      "grad_norm": 2.356048822402954,
      "learning_rate": 2.664677239804559e-06,
      "loss": 0.4747,
      "step": 236220
    },
    {
      "epoch": 96.7,
      "grad_norm": 2.2256314754486084,
      "learning_rate": 2.6645201036727502e-06,
      "loss": 0.4763,
      "step": 236230
    },
    {
      "epoch": 96.7,
      "grad_norm": 1.8615409135818481,
      "learning_rate": 2.664362966071543e-06,
      "loss": 0.4887,
      "step": 236240
    },
    {
      "epoch": 96.7,
      "grad_norm": 1.7909036874771118,
      "learning_rate": 2.664205827001657e-06,
      "loss": 0.4794,
      "step": 236250
    },
    {
      "epoch": 96.71,
      "grad_norm": 2.0739500522613525,
      "learning_rate": 2.6640486864638133e-06,
      "loss": 0.4836,
      "step": 236260
    },
    {
      "epoch": 96.71,
      "grad_norm": 1.9601824283599854,
      "learning_rate": 2.663891544458731e-06,
      "loss": 0.4928,
      "step": 236270
    },
    {
      "epoch": 96.72,
      "grad_norm": 1.8139846324920654,
      "learning_rate": 2.663734400987129e-06,
      "loss": 0.4743,
      "step": 236280
    },
    {
      "epoch": 96.72,
      "grad_norm": 1.3441272974014282,
      "learning_rate": 2.6635772560497275e-06,
      "loss": 0.4695,
      "step": 236290
    },
    {
      "epoch": 96.73,
      "grad_norm": 1.9842935800552368,
      "learning_rate": 2.6634201096472486e-06,
      "loss": 0.4849,
      "step": 236300
    },
    {
      "epoch": 96.73,
      "grad_norm": 2.07045841217041,
      "learning_rate": 2.6632629617804093e-06,
      "loss": 0.4919,
      "step": 236310
    },
    {
      "epoch": 96.73,
      "grad_norm": 2.2799906730651855,
      "learning_rate": 2.663105812449931e-06,
      "loss": 0.469,
      "step": 236320
    },
    {
      "epoch": 96.74,
      "grad_norm": 2.376018762588501,
      "learning_rate": 2.662948661656533e-06,
      "loss": 0.4785,
      "step": 236330
    },
    {
      "epoch": 96.74,
      "grad_norm": 1.8280842304229736,
      "learning_rate": 2.662791509400936e-06,
      "loss": 0.4623,
      "step": 236340
    },
    {
      "epoch": 96.75,
      "grad_norm": 1.8952597379684448,
      "learning_rate": 2.6626343556838583e-06,
      "loss": 0.4759,
      "step": 236350
    },
    {
      "epoch": 96.75,
      "grad_norm": 2.042325019836426,
      "learning_rate": 2.6624772005060215e-06,
      "loss": 0.4978,
      "step": 236360
    },
    {
      "epoch": 96.75,
      "grad_norm": 1.6504026651382446,
      "learning_rate": 2.6623200438681443e-06,
      "loss": 0.4576,
      "step": 236370
    },
    {
      "epoch": 96.76,
      "grad_norm": 1.545957326889038,
      "learning_rate": 2.662162885770947e-06,
      "loss": 0.5018,
      "step": 236380
    },
    {
      "epoch": 96.76,
      "grad_norm": 2.4282002449035645,
      "learning_rate": 2.6620057262151492e-06,
      "loss": 0.4929,
      "step": 236390
    },
    {
      "epoch": 96.77,
      "grad_norm": 1.8226438760757446,
      "learning_rate": 2.661848565201472e-06,
      "loss": 0.47,
      "step": 236400
    },
    {
      "epoch": 96.77,
      "grad_norm": 2.1157548427581787,
      "learning_rate": 2.6616914027306343e-06,
      "loss": 0.4807,
      "step": 236410
    },
    {
      "epoch": 96.77,
      "grad_norm": 2.0648012161254883,
      "learning_rate": 2.6615342388033563e-06,
      "loss": 0.5076,
      "step": 236420
    },
    {
      "epoch": 96.78,
      "grad_norm": 2.002739906311035,
      "learning_rate": 2.6613770734203577e-06,
      "loss": 0.4983,
      "step": 236430
    },
    {
      "epoch": 96.78,
      "grad_norm": 1.7871496677398682,
      "learning_rate": 2.661219906582359e-06,
      "loss": 0.5085,
      "step": 236440
    },
    {
      "epoch": 96.79,
      "grad_norm": 1.5393270254135132,
      "learning_rate": 2.66106273829008e-06,
      "loss": 0.4984,
      "step": 236450
    },
    {
      "epoch": 96.79,
      "grad_norm": 1.9884170293807983,
      "learning_rate": 2.66090556854424e-06,
      "loss": 0.4843,
      "step": 236460
    },
    {
      "epoch": 96.79,
      "grad_norm": 1.729602336883545,
      "learning_rate": 2.6607483973455597e-06,
      "loss": 0.4572,
      "step": 236470
    },
    {
      "epoch": 96.8,
      "grad_norm": 1.9610422849655151,
      "learning_rate": 2.6605912246947584e-06,
      "loss": 0.491,
      "step": 236480
    },
    {
      "epoch": 96.8,
      "grad_norm": 1.8752456903457642,
      "learning_rate": 2.660434050592557e-06,
      "loss": 0.4528,
      "step": 236490
    },
    {
      "epoch": 96.81,
      "grad_norm": 2.4530749320983887,
      "learning_rate": 2.660276875039675e-06,
      "loss": 0.4758,
      "step": 236500
    },
    {
      "epoch": 96.81,
      "grad_norm": 2.673161745071411,
      "learning_rate": 2.660119698036832e-06,
      "loss": 0.4824,
      "step": 236510
    },
    {
      "epoch": 96.82,
      "grad_norm": 1.7224897146224976,
      "learning_rate": 2.659962519584749e-06,
      "loss": 0.4673,
      "step": 236520
    },
    {
      "epoch": 96.82,
      "grad_norm": 1.8581640720367432,
      "learning_rate": 2.6598053396841446e-06,
      "loss": 0.486,
      "step": 236530
    },
    {
      "epoch": 96.82,
      "grad_norm": 1.6220121383666992,
      "learning_rate": 2.659648158335741e-06,
      "loss": 0.4727,
      "step": 236540
    },
    {
      "epoch": 96.83,
      "grad_norm": 1.9041295051574707,
      "learning_rate": 2.6594909755402555e-06,
      "loss": 0.4999,
      "step": 236550
    },
    {
      "epoch": 96.83,
      "grad_norm": 2.0756819248199463,
      "learning_rate": 2.65933379129841e-06,
      "loss": 0.4842,
      "step": 236560
    },
    {
      "epoch": 96.84,
      "grad_norm": 2.1114189624786377,
      "learning_rate": 2.659176605610924e-06,
      "loss": 0.4909,
      "step": 236570
    },
    {
      "epoch": 96.84,
      "grad_norm": 2.038527250289917,
      "learning_rate": 2.6590194184785175e-06,
      "loss": 0.4783,
      "step": 236580
    },
    {
      "epoch": 96.84,
      "grad_norm": 1.5843795537948608,
      "learning_rate": 2.6588622299019107e-06,
      "loss": 0.4807,
      "step": 236590
    },
    {
      "epoch": 96.85,
      "grad_norm": 2.1808934211730957,
      "learning_rate": 2.6587050398818238e-06,
      "loss": 0.4925,
      "step": 236600
    },
    {
      "epoch": 96.85,
      "grad_norm": 1.687784194946289,
      "learning_rate": 2.6585478484189763e-06,
      "loss": 0.4872,
      "step": 236610
    },
    {
      "epoch": 96.86,
      "grad_norm": 2.060487985610962,
      "learning_rate": 2.6583906555140888e-06,
      "loss": 0.4742,
      "step": 236620
    },
    {
      "epoch": 96.86,
      "grad_norm": 1.4660251140594482,
      "learning_rate": 2.6582334611678815e-06,
      "loss": 0.4839,
      "step": 236630
    },
    {
      "epoch": 96.86,
      "grad_norm": 1.5147898197174072,
      "learning_rate": 2.6580762653810736e-06,
      "loss": 0.4742,
      "step": 236640
    },
    {
      "epoch": 96.87,
      "grad_norm": 1.7819514274597168,
      "learning_rate": 2.657919068154386e-06,
      "loss": 0.4681,
      "step": 236650
    },
    {
      "epoch": 96.87,
      "grad_norm": 1.9972128868103027,
      "learning_rate": 2.6577618694885384e-06,
      "loss": 0.4939,
      "step": 236660
    },
    {
      "epoch": 96.88,
      "grad_norm": 2.1974050998687744,
      "learning_rate": 2.657604669384251e-06,
      "loss": 0.4604,
      "step": 236670
    },
    {
      "epoch": 96.88,
      "grad_norm": 2.0455048084259033,
      "learning_rate": 2.657447467842244e-06,
      "loss": 0.4794,
      "step": 236680
    },
    {
      "epoch": 96.88,
      "grad_norm": 1.4836256504058838,
      "learning_rate": 2.6572902648632373e-06,
      "loss": 0.4939,
      "step": 236690
    },
    {
      "epoch": 96.89,
      "grad_norm": 1.6463502645492554,
      "learning_rate": 2.657133060447952e-06,
      "loss": 0.5024,
      "step": 236700
    },
    {
      "epoch": 96.89,
      "grad_norm": 1.9836405515670776,
      "learning_rate": 2.6569758545971074e-06,
      "loss": 0.4699,
      "step": 236710
    },
    {
      "epoch": 96.9,
      "grad_norm": 2.338967800140381,
      "learning_rate": 2.656818647311423e-06,
      "loss": 0.4917,
      "step": 236720
    },
    {
      "epoch": 96.9,
      "grad_norm": 2.487518548965454,
      "learning_rate": 2.6566614385916195e-06,
      "loss": 0.4744,
      "step": 236730
    },
    {
      "epoch": 96.91,
      "grad_norm": 1.9439972639083862,
      "learning_rate": 2.656504228438418e-06,
      "loss": 0.4826,
      "step": 236740
    },
    {
      "epoch": 96.91,
      "grad_norm": 2.0944595336914062,
      "learning_rate": 2.656347016852537e-06,
      "loss": 0.4701,
      "step": 236750
    },
    {
      "epoch": 96.91,
      "grad_norm": 1.7460277080535889,
      "learning_rate": 2.6561898038346972e-06,
      "loss": 0.4797,
      "step": 236760
    },
    {
      "epoch": 96.92,
      "grad_norm": 2.0542805194854736,
      "learning_rate": 2.6560325893856207e-06,
      "loss": 0.485,
      "step": 236770
    },
    {
      "epoch": 96.92,
      "grad_norm": 2.1284847259521484,
      "learning_rate": 2.6558753735060245e-06,
      "loss": 0.5005,
      "step": 236780
    },
    {
      "epoch": 96.93,
      "grad_norm": 1.937230110168457,
      "learning_rate": 2.6557181561966307e-06,
      "loss": 0.4982,
      "step": 236790
    },
    {
      "epoch": 96.93,
      "grad_norm": 1.9393267631530762,
      "learning_rate": 2.6555609374581594e-06,
      "loss": 0.5127,
      "step": 236800
    },
    {
      "epoch": 96.93,
      "grad_norm": 1.8040515184402466,
      "learning_rate": 2.65540371729133e-06,
      "loss": 0.475,
      "step": 236810
    },
    {
      "epoch": 96.94,
      "grad_norm": 1.644855260848999,
      "learning_rate": 2.655246495696863e-06,
      "loss": 0.4715,
      "step": 236820
    },
    {
      "epoch": 96.94,
      "grad_norm": 1.788222312927246,
      "learning_rate": 2.6550892726754794e-06,
      "loss": 0.4771,
      "step": 236830
    },
    {
      "epoch": 96.95,
      "grad_norm": 3.779707670211792,
      "learning_rate": 2.654932048227899e-06,
      "loss": 0.4922,
      "step": 236840
    },
    {
      "epoch": 96.95,
      "grad_norm": 1.8597147464752197,
      "learning_rate": 2.6547748223548415e-06,
      "loss": 0.4801,
      "step": 236850
    },
    {
      "epoch": 96.95,
      "grad_norm": 1.7692036628723145,
      "learning_rate": 2.6546175950570273e-06,
      "loss": 0.4743,
      "step": 236860
    },
    {
      "epoch": 96.96,
      "grad_norm": 1.64989173412323,
      "learning_rate": 2.6544603663351766e-06,
      "loss": 0.4719,
      "step": 236870
    },
    {
      "epoch": 96.96,
      "grad_norm": 1.9216222763061523,
      "learning_rate": 2.6543031361900103e-06,
      "loss": 0.4843,
      "step": 236880
    },
    {
      "epoch": 96.97,
      "grad_norm": 1.9734891653060913,
      "learning_rate": 2.654145904622248e-06,
      "loss": 0.4896,
      "step": 236890
    },
    {
      "epoch": 96.97,
      "grad_norm": 1.741756558418274,
      "learning_rate": 2.65398867163261e-06,
      "loss": 0.4643,
      "step": 236900
    },
    {
      "epoch": 96.98,
      "grad_norm": 2.4185564517974854,
      "learning_rate": 2.653831437221817e-06,
      "loss": 0.4776,
      "step": 236910
    },
    {
      "epoch": 96.98,
      "grad_norm": 1.8055672645568848,
      "learning_rate": 2.653674201390589e-06,
      "loss": 0.4871,
      "step": 236920
    },
    {
      "epoch": 96.98,
      "grad_norm": 2.028345823287964,
      "learning_rate": 2.653516964139646e-06,
      "loss": 0.4831,
      "step": 236930
    },
    {
      "epoch": 96.99,
      "grad_norm": 2.2854578495025635,
      "learning_rate": 2.6533597254697083e-06,
      "loss": 0.4847,
      "step": 236940
    },
    {
      "epoch": 96.99,
      "grad_norm": 1.5388661623001099,
      "learning_rate": 2.6532024853814967e-06,
      "loss": 0.4812,
      "step": 236950
    },
    {
      "epoch": 97.0,
      "grad_norm": 1.6635499000549316,
      "learning_rate": 2.653045243875731e-06,
      "loss": 0.4854,
      "step": 236960
    },
    {
      "epoch": 97.0,
      "grad_norm": 1.847739577293396,
      "learning_rate": 2.6528880009531316e-06,
      "loss": 0.5118,
      "step": 236970
    },
    {
      "epoch": 97.0,
      "eval_loss": 0.4842064678668976,
      "eval_runtime": 52.554,
      "eval_samples_per_second": 65.628,
      "eval_steps_per_second": 8.22,
      "step": 236971
    },
    {
      "epoch": 97.0,
      "grad_norm": 1.8623337745666504,
      "learning_rate": 2.6527307566144193e-06,
      "loss": 0.4715,
      "step": 236980
    },
    {
      "epoch": 97.01,
      "grad_norm": 1.963713526725769,
      "learning_rate": 2.652573510860314e-06,
      "loss": 0.4715,
      "step": 236990
    },
    {
      "epoch": 97.01,
      "grad_norm": 1.8499841690063477,
      "learning_rate": 2.652416263691536e-06,
      "loss": 0.4784,
      "step": 237000
    },
    {
      "epoch": 97.02,
      "grad_norm": 1.9382861852645874,
      "learning_rate": 2.6522590151088053e-06,
      "loss": 0.4534,
      "step": 237010
    },
    {
      "epoch": 97.02,
      "grad_norm": 2.2386345863342285,
      "learning_rate": 2.652101765112843e-06,
      "loss": 0.4907,
      "step": 237020
    },
    {
      "epoch": 97.02,
      "grad_norm": 1.6951115131378174,
      "learning_rate": 2.651944513704369e-06,
      "loss": 0.4838,
      "step": 237030
    },
    {
      "epoch": 97.03,
      "grad_norm": 2.069518566131592,
      "learning_rate": 2.651787260884103e-06,
      "loss": 0.4631,
      "step": 237040
    },
    {
      "epoch": 97.03,
      "grad_norm": 1.6199467182159424,
      "learning_rate": 2.651630006652766e-06,
      "loss": 0.4815,
      "step": 237050
    },
    {
      "epoch": 97.04,
      "grad_norm": 1.9446043968200684,
      "learning_rate": 2.6514727510110795e-06,
      "loss": 0.494,
      "step": 237060
    },
    {
      "epoch": 97.04,
      "grad_norm": 2.1768760681152344,
      "learning_rate": 2.651315493959762e-06,
      "loss": 0.4756,
      "step": 237070
    },
    {
      "epoch": 97.04,
      "grad_norm": 1.922731876373291,
      "learning_rate": 2.6511582354995342e-06,
      "loss": 0.5025,
      "step": 237080
    },
    {
      "epoch": 97.05,
      "grad_norm": 2.1724941730499268,
      "learning_rate": 2.651000975631118e-06,
      "loss": 0.4804,
      "step": 237090
    },
    {
      "epoch": 97.05,
      "grad_norm": 1.7606266736984253,
      "learning_rate": 2.6508437143552317e-06,
      "loss": 0.4867,
      "step": 237100
    },
    {
      "epoch": 97.06,
      "grad_norm": 2.3137450218200684,
      "learning_rate": 2.6506864516725966e-06,
      "loss": 0.499,
      "step": 237110
    },
    {
      "epoch": 97.06,
      "grad_norm": 2.3481242656707764,
      "learning_rate": 2.6505291875839335e-06,
      "loss": 0.4698,
      "step": 237120
    },
    {
      "epoch": 97.07,
      "grad_norm": 1.8172690868377686,
      "learning_rate": 2.6503719220899626e-06,
      "loss": 0.4776,
      "step": 237130
    },
    {
      "epoch": 97.07,
      "grad_norm": 1.8981516361236572,
      "learning_rate": 2.6502146551914044e-06,
      "loss": 0.4813,
      "step": 237140
    },
    {
      "epoch": 97.07,
      "grad_norm": 1.8813012838363647,
      "learning_rate": 2.650057386888978e-06,
      "loss": 0.4716,
      "step": 237150
    },
    {
      "epoch": 97.08,
      "grad_norm": 1.9747796058654785,
      "learning_rate": 2.649900117183406e-06,
      "loss": 0.4624,
      "step": 237160
    },
    {
      "epoch": 97.08,
      "grad_norm": 1.9928197860717773,
      "learning_rate": 2.6497428460754075e-06,
      "loss": 0.4894,
      "step": 237170
    },
    {
      "epoch": 97.09,
      "grad_norm": 1.9432682991027832,
      "learning_rate": 2.649585573565703e-06,
      "loss": 0.4946,
      "step": 237180
    },
    {
      "epoch": 97.09,
      "grad_norm": 1.9556337594985962,
      "learning_rate": 2.649428299655013e-06,
      "loss": 0.5001,
      "step": 237190
    },
    {
      "epoch": 97.09,
      "grad_norm": 2.6436402797698975,
      "learning_rate": 2.649271024344058e-06,
      "loss": 0.4972,
      "step": 237200
    },
    {
      "epoch": 97.1,
      "grad_norm": 2.097748041152954,
      "learning_rate": 2.6491137476335592e-06,
      "loss": 0.4679,
      "step": 237210
    },
    {
      "epoch": 97.1,
      "grad_norm": 1.73480224609375,
      "learning_rate": 2.6489564695242354e-06,
      "loss": 0.4844,
      "step": 237220
    },
    {
      "epoch": 97.11,
      "grad_norm": 2.1405889987945557,
      "learning_rate": 2.6487991900168083e-06,
      "loss": 0.4838,
      "step": 237230
    },
    {
      "epoch": 97.11,
      "grad_norm": 2.282090425491333,
      "learning_rate": 2.6486419091119983e-06,
      "loss": 0.4712,
      "step": 237240
    },
    {
      "epoch": 97.11,
      "grad_norm": 2.0163588523864746,
      "learning_rate": 2.6484846268105255e-06,
      "loss": 0.4824,
      "step": 237250
    },
    {
      "epoch": 97.12,
      "grad_norm": 1.8062607049942017,
      "learning_rate": 2.6483273431131106e-06,
      "loss": 0.493,
      "step": 237260
    },
    {
      "epoch": 97.12,
      "grad_norm": 1.673604965209961,
      "learning_rate": 2.648170058020474e-06,
      "loss": 0.4748,
      "step": 237270
    },
    {
      "epoch": 97.13,
      "grad_norm": 1.8063974380493164,
      "learning_rate": 2.648012771533337e-06,
      "loss": 0.4891,
      "step": 237280
    },
    {
      "epoch": 97.13,
      "grad_norm": 2.2665398120880127,
      "learning_rate": 2.6478554836524187e-06,
      "loss": 0.4792,
      "step": 237290
    },
    {
      "epoch": 97.13,
      "grad_norm": 2.0052969455718994,
      "learning_rate": 2.6476981943784404e-06,
      "loss": 0.5021,
      "step": 237300
    },
    {
      "epoch": 97.14,
      "grad_norm": 1.717759370803833,
      "learning_rate": 2.647540903712122e-06,
      "loss": 0.4589,
      "step": 237310
    },
    {
      "epoch": 97.14,
      "grad_norm": 2.0381641387939453,
      "learning_rate": 2.6473836116541854e-06,
      "loss": 0.48,
      "step": 237320
    },
    {
      "epoch": 97.15,
      "grad_norm": 2.039207935333252,
      "learning_rate": 2.6472263182053494e-06,
      "loss": 0.4977,
      "step": 237330
    },
    {
      "epoch": 97.15,
      "grad_norm": 1.7524104118347168,
      "learning_rate": 2.6470690233663354e-06,
      "loss": 0.4661,
      "step": 237340
    },
    {
      "epoch": 97.16,
      "grad_norm": 2.183382272720337,
      "learning_rate": 2.646911727137865e-06,
      "loss": 0.4997,
      "step": 237350
    },
    {
      "epoch": 97.16,
      "grad_norm": 2.1378417015075684,
      "learning_rate": 2.646754429520657e-06,
      "loss": 0.4796,
      "step": 237360
    },
    {
      "epoch": 97.16,
      "grad_norm": 1.8373373746871948,
      "learning_rate": 2.6465971305154324e-06,
      "loss": 0.474,
      "step": 237370
    },
    {
      "epoch": 97.17,
      "grad_norm": 2.009432554244995,
      "learning_rate": 2.646439830122913e-06,
      "loss": 0.4693,
      "step": 237380
    },
    {
      "epoch": 97.17,
      "grad_norm": 2.0837879180908203,
      "learning_rate": 2.646282528343817e-06,
      "loss": 0.48,
      "step": 237390
    },
    {
      "epoch": 97.18,
      "grad_norm": 1.6937549114227295,
      "learning_rate": 2.6461252251788674e-06,
      "loss": 0.4563,
      "step": 237400
    },
    {
      "epoch": 97.18,
      "grad_norm": 1.853337287902832,
      "learning_rate": 2.645967920628783e-06,
      "loss": 0.4837,
      "step": 237410
    },
    {
      "epoch": 97.18,
      "grad_norm": 2.573885679244995,
      "learning_rate": 2.6458106146942853e-06,
      "loss": 0.4856,
      "step": 237420
    },
    {
      "epoch": 97.19,
      "grad_norm": 1.9108351469039917,
      "learning_rate": 2.645653307376095e-06,
      "loss": 0.4833,
      "step": 237430
    },
    {
      "epoch": 97.19,
      "grad_norm": 1.7042125463485718,
      "learning_rate": 2.645495998674932e-06,
      "loss": 0.4662,
      "step": 237440
    },
    {
      "epoch": 97.2,
      "grad_norm": 2.2354955673217773,
      "learning_rate": 2.6453386885915175e-06,
      "loss": 0.4859,
      "step": 237450
    },
    {
      "epoch": 97.2,
      "grad_norm": 2.4112327098846436,
      "learning_rate": 2.645181377126572e-06,
      "loss": 0.4655,
      "step": 237460
    },
    {
      "epoch": 97.2,
      "grad_norm": 2.2486326694488525,
      "learning_rate": 2.6450240642808157e-06,
      "loss": 0.4785,
      "step": 237470
    },
    {
      "epoch": 97.21,
      "grad_norm": 2.2004425525665283,
      "learning_rate": 2.64486675005497e-06,
      "loss": 0.5131,
      "step": 237480
    },
    {
      "epoch": 97.21,
      "grad_norm": 1.7257786989212036,
      "learning_rate": 2.644709434449755e-06,
      "loss": 0.4846,
      "step": 237490
    },
    {
      "epoch": 97.22,
      "grad_norm": 1.7990319728851318,
      "learning_rate": 2.6445521174658913e-06,
      "loss": 0.4828,
      "step": 237500
    },
    {
      "epoch": 97.22,
      "grad_norm": 1.686508059501648,
      "learning_rate": 2.6443947991041e-06,
      "loss": 0.4715,
      "step": 237510
    },
    {
      "epoch": 97.22,
      "grad_norm": 1.7134675979614258,
      "learning_rate": 2.644237479365101e-06,
      "loss": 0.515,
      "step": 237520
    },
    {
      "epoch": 97.23,
      "grad_norm": 1.9069721698760986,
      "learning_rate": 2.6440801582496156e-06,
      "loss": 0.477,
      "step": 237530
    },
    {
      "epoch": 97.23,
      "grad_norm": 1.5234739780426025,
      "learning_rate": 2.643922835758364e-06,
      "loss": 0.4739,
      "step": 237540
    },
    {
      "epoch": 97.24,
      "grad_norm": 2.182556629180908,
      "learning_rate": 2.643765511892067e-06,
      "loss": 0.4908,
      "step": 237550
    },
    {
      "epoch": 97.24,
      "grad_norm": 3.0026679039001465,
      "learning_rate": 2.643608186651446e-06,
      "loss": 0.4684,
      "step": 237560
    },
    {
      "epoch": 97.25,
      "grad_norm": 1.5644675493240356,
      "learning_rate": 2.6434508600372214e-06,
      "loss": 0.4886,
      "step": 237570
    },
    {
      "epoch": 97.25,
      "grad_norm": 2.196660041809082,
      "learning_rate": 2.643293532050113e-06,
      "loss": 0.4747,
      "step": 237580
    },
    {
      "epoch": 97.25,
      "grad_norm": 1.9365993738174438,
      "learning_rate": 2.643136202690842e-06,
      "loss": 0.4732,
      "step": 237590
    },
    {
      "epoch": 97.26,
      "grad_norm": 1.6511491537094116,
      "learning_rate": 2.6429788719601297e-06,
      "loss": 0.494,
      "step": 237600
    },
    {
      "epoch": 97.26,
      "grad_norm": 1.5459378957748413,
      "learning_rate": 2.642821539858695e-06,
      "loss": 0.4654,
      "step": 237610
    },
    {
      "epoch": 97.27,
      "grad_norm": 2.1303627490997314,
      "learning_rate": 2.6426642063872606e-06,
      "loss": 0.4911,
      "step": 237620
    },
    {
      "epoch": 97.27,
      "grad_norm": 1.521673321723938,
      "learning_rate": 2.642506871546547e-06,
      "loss": 0.484,
      "step": 237630
    },
    {
      "epoch": 97.27,
      "grad_norm": 2.1260595321655273,
      "learning_rate": 2.6423495353372744e-06,
      "loss": 0.4728,
      "step": 237640
    },
    {
      "epoch": 97.28,
      "grad_norm": 1.7576721906661987,
      "learning_rate": 2.642192197760163e-06,
      "loss": 0.4938,
      "step": 237650
    },
    {
      "epoch": 97.28,
      "grad_norm": 1.8632769584655762,
      "learning_rate": 2.642034858815935e-06,
      "loss": 0.487,
      "step": 237660
    },
    {
      "epoch": 97.29,
      "grad_norm": 2.4538800716400146,
      "learning_rate": 2.6418775185053094e-06,
      "loss": 0.4932,
      "step": 237670
    },
    {
      "epoch": 97.29,
      "grad_norm": 1.9981036186218262,
      "learning_rate": 2.6417201768290085e-06,
      "loss": 0.5038,
      "step": 237680
    },
    {
      "epoch": 97.29,
      "grad_norm": 2.3888583183288574,
      "learning_rate": 2.6415628337877512e-06,
      "loss": 0.4658,
      "step": 237690
    },
    {
      "epoch": 97.3,
      "grad_norm": 2.1012303829193115,
      "learning_rate": 2.6414054893822606e-06,
      "loss": 0.4838,
      "step": 237700
    },
    {
      "epoch": 97.3,
      "grad_norm": 2.0468971729278564,
      "learning_rate": 2.6412481436132553e-06,
      "loss": 0.4876,
      "step": 237710
    },
    {
      "epoch": 97.31,
      "grad_norm": 2.0759787559509277,
      "learning_rate": 2.6410907964814583e-06,
      "loss": 0.4827,
      "step": 237720
    },
    {
      "epoch": 97.31,
      "grad_norm": 1.810009241104126,
      "learning_rate": 2.640933447987588e-06,
      "loss": 0.4913,
      "step": 237730
    },
    {
      "epoch": 97.31,
      "grad_norm": 2.156691789627075,
      "learning_rate": 2.6407760981323668e-06,
      "loss": 0.4736,
      "step": 237740
    },
    {
      "epoch": 97.32,
      "grad_norm": 1.6630587577819824,
      "learning_rate": 2.640618746916515e-06,
      "loss": 0.4888,
      "step": 237750
    },
    {
      "epoch": 97.32,
      "grad_norm": 1.7167805433273315,
      "learning_rate": 2.6404613943407536e-06,
      "loss": 0.477,
      "step": 237760
    },
    {
      "epoch": 97.33,
      "grad_norm": 1.414826512336731,
      "learning_rate": 2.6403040404058035e-06,
      "loss": 0.4831,
      "step": 237770
    },
    {
      "epoch": 97.33,
      "grad_norm": 2.291870594024658,
      "learning_rate": 2.6401466851123845e-06,
      "loss": 0.4688,
      "step": 237780
    },
    {
      "epoch": 97.34,
      "grad_norm": 6.795844554901123,
      "learning_rate": 2.6399893284612187e-06,
      "loss": 0.5068,
      "step": 237790
    },
    {
      "epoch": 97.34,
      "grad_norm": 2.1954898834228516,
      "learning_rate": 2.6398319704530265e-06,
      "loss": 0.4905,
      "step": 237800
    },
    {
      "epoch": 97.34,
      "grad_norm": 2.3420684337615967,
      "learning_rate": 2.6396746110885287e-06,
      "loss": 0.494,
      "step": 237810
    },
    {
      "epoch": 97.35,
      "grad_norm": 1.9797495603561401,
      "learning_rate": 2.6395172503684453e-06,
      "loss": 0.4714,
      "step": 237820
    },
    {
      "epoch": 97.35,
      "grad_norm": 2.019634962081909,
      "learning_rate": 2.639359888293499e-06,
      "loss": 0.4783,
      "step": 237830
    },
    {
      "epoch": 97.36,
      "grad_norm": 1.7386459112167358,
      "learning_rate": 2.6392025248644093e-06,
      "loss": 0.458,
      "step": 237840
    },
    {
      "epoch": 97.36,
      "grad_norm": 1.8860920667648315,
      "learning_rate": 2.6390451600818966e-06,
      "loss": 0.4586,
      "step": 237850
    },
    {
      "epoch": 97.36,
      "grad_norm": 1.831924557685852,
      "learning_rate": 2.6388877939466838e-06,
      "loss": 0.4814,
      "step": 237860
    },
    {
      "epoch": 97.37,
      "grad_norm": 3.006844997406006,
      "learning_rate": 2.63873042645949e-06,
      "loss": 0.4936,
      "step": 237870
    },
    {
      "epoch": 97.37,
      "grad_norm": 1.8035705089569092,
      "learning_rate": 2.638573057621036e-06,
      "loss": 0.4964,
      "step": 237880
    },
    {
      "epoch": 97.38,
      "grad_norm": 1.77703058719635,
      "learning_rate": 2.638415687432043e-06,
      "loss": 0.4918,
      "step": 237890
    },
    {
      "epoch": 97.38,
      "grad_norm": 2.260239362716675,
      "learning_rate": 2.638258315893232e-06,
      "loss": 0.4801,
      "step": 237900
    },
    {
      "epoch": 97.38,
      "grad_norm": 1.549397587776184,
      "learning_rate": 2.6381009430053252e-06,
      "loss": 0.4855,
      "step": 237910
    },
    {
      "epoch": 97.39,
      "grad_norm": 2.076714277267456,
      "learning_rate": 2.6379435687690414e-06,
      "loss": 0.4977,
      "step": 237920
    },
    {
      "epoch": 97.39,
      "grad_norm": 1.6622471809387207,
      "learning_rate": 2.6377861931851036e-06,
      "loss": 0.5015,
      "step": 237930
    },
    {
      "epoch": 97.4,
      "grad_norm": 1.9344079494476318,
      "learning_rate": 2.637628816254231e-06,
      "loss": 0.481,
      "step": 237940
    },
    {
      "epoch": 97.4,
      "grad_norm": 1.949749231338501,
      "learning_rate": 2.6374714379771446e-06,
      "loss": 0.4631,
      "step": 237950
    },
    {
      "epoch": 97.4,
      "grad_norm": 2.531304121017456,
      "learning_rate": 2.6373140583545656e-06,
      "loss": 0.4727,
      "step": 237960
    },
    {
      "epoch": 97.41,
      "grad_norm": 1.9333245754241943,
      "learning_rate": 2.637156677387216e-06,
      "loss": 0.4696,
      "step": 237970
    },
    {
      "epoch": 97.41,
      "grad_norm": 1.4410914182662964,
      "learning_rate": 2.636999295075815e-06,
      "loss": 0.4764,
      "step": 237980
    },
    {
      "epoch": 97.42,
      "grad_norm": 1.895638346672058,
      "learning_rate": 2.6368419114210838e-06,
      "loss": 0.4719,
      "step": 237990
    },
    {
      "epoch": 97.42,
      "grad_norm": 2.2456278800964355,
      "learning_rate": 2.6366845264237457e-06,
      "loss": 0.4758,
      "step": 238000
    },
    {
      "epoch": 97.43,
      "grad_norm": 1.7680318355560303,
      "learning_rate": 2.636527140084519e-06,
      "loss": 0.4623,
      "step": 238010
    },
    {
      "epoch": 97.43,
      "grad_norm": 1.932877779006958,
      "learning_rate": 2.6363697524041256e-06,
      "loss": 0.4817,
      "step": 238020
    },
    {
      "epoch": 97.43,
      "grad_norm": 2.6242449283599854,
      "learning_rate": 2.6362123633832866e-06,
      "loss": 0.481,
      "step": 238030
    },
    {
      "epoch": 97.44,
      "grad_norm": 2.1403589248657227,
      "learning_rate": 2.636054973022722e-06,
      "loss": 0.5026,
      "step": 238040
    },
    {
      "epoch": 97.44,
      "grad_norm": 1.9728766679763794,
      "learning_rate": 2.6358975813231546e-06,
      "loss": 0.4678,
      "step": 238050
    },
    {
      "epoch": 97.45,
      "grad_norm": 2.013949394226074,
      "learning_rate": 2.635740188285304e-06,
      "loss": 0.4757,
      "step": 238060
    },
    {
      "epoch": 97.45,
      "grad_norm": 2.1413352489471436,
      "learning_rate": 2.6355827939098916e-06,
      "loss": 0.4633,
      "step": 238070
    },
    {
      "epoch": 97.45,
      "grad_norm": 1.7967151403427124,
      "learning_rate": 2.6354253981976384e-06,
      "loss": 0.4881,
      "step": 238080
    },
    {
      "epoch": 97.46,
      "grad_norm": 2.3325867652893066,
      "learning_rate": 2.6352680011492657e-06,
      "loss": 0.4907,
      "step": 238090
    },
    {
      "epoch": 97.46,
      "grad_norm": 1.8816465139389038,
      "learning_rate": 2.6351106027654934e-06,
      "loss": 0.4671,
      "step": 238100
    },
    {
      "epoch": 97.47,
      "grad_norm": 1.8051210641860962,
      "learning_rate": 2.6349532030470433e-06,
      "loss": 0.4706,
      "step": 238110
    },
    {
      "epoch": 97.47,
      "grad_norm": 1.9345836639404297,
      "learning_rate": 2.634795801994637e-06,
      "loss": 0.4863,
      "step": 238120
    },
    {
      "epoch": 97.47,
      "grad_norm": 2.0557615756988525,
      "learning_rate": 2.634638399608995e-06,
      "loss": 0.4818,
      "step": 238130
    },
    {
      "epoch": 97.48,
      "grad_norm": 1.9576632976531982,
      "learning_rate": 2.634480995890838e-06,
      "loss": 0.464,
      "step": 238140
    },
    {
      "epoch": 97.48,
      "grad_norm": 1.831276297569275,
      "learning_rate": 2.6343235908408875e-06,
      "loss": 0.4684,
      "step": 238150
    },
    {
      "epoch": 97.49,
      "grad_norm": 1.830856442451477,
      "learning_rate": 2.6341661844598646e-06,
      "loss": 0.4843,
      "step": 238160
    },
    {
      "epoch": 97.49,
      "grad_norm": 2.1296141147613525,
      "learning_rate": 2.63400877674849e-06,
      "loss": 0.4842,
      "step": 238170
    },
    {
      "epoch": 97.49,
      "grad_norm": 1.6705493927001953,
      "learning_rate": 2.6338513677074845e-06,
      "loss": 0.486,
      "step": 238180
    },
    {
      "epoch": 97.5,
      "grad_norm": 2.056612491607666,
      "learning_rate": 2.6336939573375696e-06,
      "loss": 0.4729,
      "step": 238190
    },
    {
      "epoch": 97.5,
      "grad_norm": 1.7788723707199097,
      "learning_rate": 2.6335365456394664e-06,
      "loss": 0.4867,
      "step": 238200
    },
    {
      "epoch": 97.51,
      "grad_norm": 1.7153587341308594,
      "learning_rate": 2.633379132613896e-06,
      "loss": 0.466,
      "step": 238210
    },
    {
      "epoch": 97.51,
      "grad_norm": 1.7665053606033325,
      "learning_rate": 2.63322171826158e-06,
      "loss": 0.4997,
      "step": 238220
    },
    {
      "epoch": 97.52,
      "grad_norm": 1.9015568494796753,
      "learning_rate": 2.6330643025832382e-06,
      "loss": 0.5047,
      "step": 238230
    },
    {
      "epoch": 97.52,
      "grad_norm": 2.231614828109741,
      "learning_rate": 2.6329068855795926e-06,
      "loss": 0.4927,
      "step": 238240
    },
    {
      "epoch": 97.52,
      "grad_norm": 2.1092288494110107,
      "learning_rate": 2.632749467251364e-06,
      "loss": 0.4863,
      "step": 238250
    },
    {
      "epoch": 97.53,
      "grad_norm": 1.8024346828460693,
      "learning_rate": 2.6325920475992738e-06,
      "loss": 0.4898,
      "step": 238260
    },
    {
      "epoch": 97.53,
      "grad_norm": 2.271331787109375,
      "learning_rate": 2.632434626624042e-06,
      "loss": 0.4846,
      "step": 238270
    },
    {
      "epoch": 97.54,
      "grad_norm": 1.8764472007751465,
      "learning_rate": 2.632277204326391e-06,
      "loss": 0.4591,
      "step": 238280
    },
    {
      "epoch": 97.54,
      "grad_norm": 1.887591004371643,
      "learning_rate": 2.6321197807070425e-06,
      "loss": 0.4884,
      "step": 238290
    },
    {
      "epoch": 97.54,
      "grad_norm": 1.8944462537765503,
      "learning_rate": 2.6319623557667163e-06,
      "loss": 0.4745,
      "step": 238300
    },
    {
      "epoch": 97.55,
      "grad_norm": 2.1792256832122803,
      "learning_rate": 2.631804929506134e-06,
      "loss": 0.4872,
      "step": 238310
    },
    {
      "epoch": 97.55,
      "grad_norm": 1.8413270711898804,
      "learning_rate": 2.631647501926016e-06,
      "loss": 0.4657,
      "step": 238320
    },
    {
      "epoch": 97.56,
      "grad_norm": 2.0115227699279785,
      "learning_rate": 2.6314900730270843e-06,
      "loss": 0.4969,
      "step": 238330
    },
    {
      "epoch": 97.56,
      "grad_norm": 1.8599828481674194,
      "learning_rate": 2.63133264281006e-06,
      "loss": 0.4709,
      "step": 238340
    },
    {
      "epoch": 97.56,
      "grad_norm": 2.094949960708618,
      "learning_rate": 2.6311752112756646e-06,
      "loss": 0.4715,
      "step": 238350
    },
    {
      "epoch": 97.57,
      "grad_norm": 1.9096357822418213,
      "learning_rate": 2.6310177784246187e-06,
      "loss": 0.4854,
      "step": 238360
    },
    {
      "epoch": 97.57,
      "grad_norm": 1.7503920793533325,
      "learning_rate": 2.6308603442576435e-06,
      "loss": 0.4746,
      "step": 238370
    },
    {
      "epoch": 97.58,
      "grad_norm": 1.7585269212722778,
      "learning_rate": 2.63070290877546e-06,
      "loss": 0.487,
      "step": 238380
    },
    {
      "epoch": 97.58,
      "grad_norm": 2.1814773082733154,
      "learning_rate": 2.6305454719787897e-06,
      "loss": 0.4673,
      "step": 238390
    },
    {
      "epoch": 97.58,
      "grad_norm": 2.0141396522521973,
      "learning_rate": 2.6303880338683536e-06,
      "loss": 0.4905,
      "step": 238400
    },
    {
      "epoch": 97.59,
      "grad_norm": 1.6387799978256226,
      "learning_rate": 2.6302305944448736e-06,
      "loss": 0.4801,
      "step": 238410
    },
    {
      "epoch": 97.59,
      "grad_norm": 1.5249935388565063,
      "learning_rate": 2.63007315370907e-06,
      "loss": 0.467,
      "step": 238420
    },
    {
      "epoch": 97.6,
      "grad_norm": 2.384337902069092,
      "learning_rate": 2.6299157116616644e-06,
      "loss": 0.4714,
      "step": 238430
    },
    {
      "epoch": 97.6,
      "grad_norm": 1.635952115058899,
      "learning_rate": 2.6297582683033784e-06,
      "loss": 0.4844,
      "step": 238440
    },
    {
      "epoch": 97.61,
      "grad_norm": 2.4315593242645264,
      "learning_rate": 2.629600823634933e-06,
      "loss": 0.496,
      "step": 238450
    },
    {
      "epoch": 97.61,
      "grad_norm": 2.8459317684173584,
      "learning_rate": 2.629443377657048e-06,
      "loss": 0.4774,
      "step": 238460
    },
    {
      "epoch": 97.61,
      "grad_norm": 1.9888834953308105,
      "learning_rate": 2.629285930370447e-06,
      "loss": 0.484,
      "step": 238470
    },
    {
      "epoch": 97.62,
      "grad_norm": 2.0275957584381104,
      "learning_rate": 2.6291284817758494e-06,
      "loss": 0.4761,
      "step": 238480
    },
    {
      "epoch": 97.62,
      "grad_norm": 2.2147717475891113,
      "learning_rate": 2.6289710318739777e-06,
      "loss": 0.4654,
      "step": 238490
    },
    {
      "epoch": 97.63,
      "grad_norm": 1.841045618057251,
      "learning_rate": 2.628813580665552e-06,
      "loss": 0.4866,
      "step": 238500
    },
    {
      "epoch": 97.63,
      "grad_norm": 1.8442726135253906,
      "learning_rate": 2.628656128151295e-06,
      "loss": 0.466,
      "step": 238510
    },
    {
      "epoch": 97.63,
      "grad_norm": 1.862318515777588,
      "learning_rate": 2.628498674331927e-06,
      "loss": 0.5024,
      "step": 238520
    },
    {
      "epoch": 97.64,
      "grad_norm": 1.877827525138855,
      "learning_rate": 2.62834121920817e-06,
      "loss": 0.4806,
      "step": 238530
    },
    {
      "epoch": 97.64,
      "grad_norm": 2.188629627227783,
      "learning_rate": 2.6281837627807436e-06,
      "loss": 0.4793,
      "step": 238540
    },
    {
      "epoch": 97.65,
      "grad_norm": 2.245944023132324,
      "learning_rate": 2.6280263050503698e-06,
      "loss": 0.4677,
      "step": 238550
    },
    {
      "epoch": 97.65,
      "grad_norm": 1.6989355087280273,
      "learning_rate": 2.6278688460177713e-06,
      "loss": 0.4909,
      "step": 238560
    },
    {
      "epoch": 97.65,
      "grad_norm": 1.7432655096054077,
      "learning_rate": 2.6277113856836678e-06,
      "loss": 0.4913,
      "step": 238570
    },
    {
      "epoch": 97.66,
      "grad_norm": 1.7891515493392944,
      "learning_rate": 2.6275539240487817e-06,
      "loss": 0.482,
      "step": 238580
    },
    {
      "epoch": 97.66,
      "grad_norm": 1.4044859409332275,
      "learning_rate": 2.6273964611138334e-06,
      "loss": 0.4922,
      "step": 238590
    },
    {
      "epoch": 97.67,
      "grad_norm": 1.8067034482955933,
      "learning_rate": 2.6272389968795446e-06,
      "loss": 0.4901,
      "step": 238600
    },
    {
      "epoch": 97.67,
      "grad_norm": 2.5344629287719727,
      "learning_rate": 2.627081531346637e-06,
      "loss": 0.4682,
      "step": 238610
    },
    {
      "epoch": 97.67,
      "grad_norm": 2.570441484451294,
      "learning_rate": 2.6269240645158315e-06,
      "loss": 0.4748,
      "step": 238620
    },
    {
      "epoch": 97.68,
      "grad_norm": 2.87693452835083,
      "learning_rate": 2.626766596387849e-06,
      "loss": 0.4703,
      "step": 238630
    },
    {
      "epoch": 97.68,
      "grad_norm": 1.9820960760116577,
      "learning_rate": 2.626609126963411e-06,
      "loss": 0.4609,
      "step": 238640
    },
    {
      "epoch": 97.69,
      "grad_norm": 1.6951329708099365,
      "learning_rate": 2.6264516562432396e-06,
      "loss": 0.4949,
      "step": 238650
    },
    {
      "epoch": 97.69,
      "grad_norm": 2.0517570972442627,
      "learning_rate": 2.626294184228056e-06,
      "loss": 0.4726,
      "step": 238660
    },
    {
      "epoch": 97.7,
      "grad_norm": 1.4771279096603394,
      "learning_rate": 2.6261367109185807e-06,
      "loss": 0.4971,
      "step": 238670
    },
    {
      "epoch": 97.7,
      "grad_norm": 2.1002843379974365,
      "learning_rate": 2.625979236315536e-06,
      "loss": 0.4755,
      "step": 238680
    },
    {
      "epoch": 97.7,
      "grad_norm": 1.561632513999939,
      "learning_rate": 2.6258217604196426e-06,
      "loss": 0.49,
      "step": 238690
    },
    {
      "epoch": 97.71,
      "grad_norm": 2.5005223751068115,
      "learning_rate": 2.6256642832316227e-06,
      "loss": 0.4827,
      "step": 238700
    },
    {
      "epoch": 97.71,
      "grad_norm": 1.7215725183486938,
      "learning_rate": 2.6255068047521964e-06,
      "loss": 0.4589,
      "step": 238710
    },
    {
      "epoch": 97.72,
      "grad_norm": 2.008037805557251,
      "learning_rate": 2.625349324982086e-06,
      "loss": 0.4848,
      "step": 238720
    },
    {
      "epoch": 97.72,
      "grad_norm": 1.838653802871704,
      "learning_rate": 2.6251918439220128e-06,
      "loss": 0.4752,
      "step": 238730
    },
    {
      "epoch": 97.72,
      "grad_norm": 1.9600427150726318,
      "learning_rate": 2.6250343615726988e-06,
      "loss": 0.4798,
      "step": 238740
    },
    {
      "epoch": 97.73,
      "grad_norm": 1.924987554550171,
      "learning_rate": 2.6248768779348634e-06,
      "loss": 0.5127,
      "step": 238750
    },
    {
      "epoch": 97.73,
      "grad_norm": 1.8114742040634155,
      "learning_rate": 2.6247193930092305e-06,
      "loss": 0.5028,
      "step": 238760
    },
    {
      "epoch": 97.74,
      "grad_norm": 1.9354722499847412,
      "learning_rate": 2.6245619067965196e-06,
      "loss": 0.4966,
      "step": 238770
    },
    {
      "epoch": 97.74,
      "grad_norm": 1.443560242652893,
      "learning_rate": 2.624404419297453e-06,
      "loss": 0.4595,
      "step": 238780
    },
    {
      "epoch": 97.74,
      "grad_norm": 1.936602234840393,
      "learning_rate": 2.6242469305127522e-06,
      "loss": 0.4847,
      "step": 238790
    },
    {
      "epoch": 97.75,
      "grad_norm": 2.0965733528137207,
      "learning_rate": 2.624089440443138e-06,
      "loss": 0.4651,
      "step": 238800
    },
    {
      "epoch": 97.75,
      "grad_norm": 2.0800580978393555,
      "learning_rate": 2.623931949089333e-06,
      "loss": 0.4864,
      "step": 238810
    },
    {
      "epoch": 97.76,
      "grad_norm": 1.9466708898544312,
      "learning_rate": 2.6237744564520575e-06,
      "loss": 0.4771,
      "step": 238820
    },
    {
      "epoch": 97.76,
      "grad_norm": 2.577420473098755,
      "learning_rate": 2.6236169625320325e-06,
      "loss": 0.4724,
      "step": 238830
    },
    {
      "epoch": 97.77,
      "grad_norm": 2.1066341400146484,
      "learning_rate": 2.6234594673299807e-06,
      "loss": 0.4885,
      "step": 238840
    },
    {
      "epoch": 97.77,
      "grad_norm": 1.8826932907104492,
      "learning_rate": 2.6233019708466235e-06,
      "loss": 0.4572,
      "step": 238850
    },
    {
      "epoch": 97.77,
      "grad_norm": 1.6224064826965332,
      "learning_rate": 2.623144473082682e-06,
      "loss": 0.4732,
      "step": 238860
    },
    {
      "epoch": 97.78,
      "grad_norm": 2.1939942836761475,
      "learning_rate": 2.622986974038878e-06,
      "loss": 0.4776,
      "step": 238870
    },
    {
      "epoch": 97.78,
      "grad_norm": 1.6484788656234741,
      "learning_rate": 2.622829473715932e-06,
      "loss": 0.4859,
      "step": 238880
    },
    {
      "epoch": 97.79,
      "grad_norm": 1.6600139141082764,
      "learning_rate": 2.6226719721145666e-06,
      "loss": 0.4827,
      "step": 238890
    },
    {
      "epoch": 97.79,
      "grad_norm": 2.0396645069122314,
      "learning_rate": 2.622514469235503e-06,
      "loss": 0.4795,
      "step": 238900
    },
    {
      "epoch": 97.79,
      "grad_norm": 1.7940839529037476,
      "learning_rate": 2.6223569650794616e-06,
      "loss": 0.4786,
      "step": 238910
    },
    {
      "epoch": 97.8,
      "grad_norm": 1.7904411554336548,
      "learning_rate": 2.6221994596471657e-06,
      "loss": 0.4805,
      "step": 238920
    },
    {
      "epoch": 97.8,
      "grad_norm": 2.219595193862915,
      "learning_rate": 2.622041952939335e-06,
      "loss": 0.4796,
      "step": 238930
    },
    {
      "epoch": 97.81,
      "grad_norm": 1.7184405326843262,
      "learning_rate": 2.621884444956693e-06,
      "loss": 0.4897,
      "step": 238940
    },
    {
      "epoch": 97.81,
      "grad_norm": 2.2450687885284424,
      "learning_rate": 2.62172693569996e-06,
      "loss": 0.4736,
      "step": 238950
    },
    {
      "epoch": 97.81,
      "grad_norm": 2.0589914321899414,
      "learning_rate": 2.6215694251698576e-06,
      "loss": 0.4728,
      "step": 238960
    },
    {
      "epoch": 97.82,
      "grad_norm": 1.8588337898254395,
      "learning_rate": 2.6214119133671074e-06,
      "loss": 0.482,
      "step": 238970
    },
    {
      "epoch": 97.82,
      "grad_norm": 1.784744143486023,
      "learning_rate": 2.621254400292431e-06,
      "loss": 0.4921,
      "step": 238980
    },
    {
      "epoch": 97.83,
      "grad_norm": 1.8398045301437378,
      "learning_rate": 2.62109688594655e-06,
      "loss": 0.4753,
      "step": 238990
    },
    {
      "epoch": 97.83,
      "grad_norm": 1.7387332916259766,
      "learning_rate": 2.6209393703301854e-06,
      "loss": 0.4925,
      "step": 239000
    },
    {
      "epoch": 97.83,
      "grad_norm": 1.7393872737884521,
      "learning_rate": 2.6207818534440597e-06,
      "loss": 0.4671,
      "step": 239010
    },
    {
      "epoch": 97.84,
      "grad_norm": 1.788279414176941,
      "learning_rate": 2.620624335288894e-06,
      "loss": 0.4828,
      "step": 239020
    },
    {
      "epoch": 97.84,
      "grad_norm": 2.4507060050964355,
      "learning_rate": 2.6204668158654095e-06,
      "loss": 0.4864,
      "step": 239030
    },
    {
      "epoch": 97.85,
      "grad_norm": 1.663749098777771,
      "learning_rate": 2.6203092951743284e-06,
      "loss": 0.4932,
      "step": 239040
    },
    {
      "epoch": 97.85,
      "grad_norm": 2.151852607727051,
      "learning_rate": 2.620151773216372e-06,
      "loss": 0.4876,
      "step": 239050
    },
    {
      "epoch": 97.86,
      "grad_norm": 1.777061939239502,
      "learning_rate": 2.619994249992262e-06,
      "loss": 0.4735,
      "step": 239060
    },
    {
      "epoch": 97.86,
      "grad_norm": 1.8878308534622192,
      "learning_rate": 2.619836725502719e-06,
      "loss": 0.4845,
      "step": 239070
    },
    {
      "epoch": 97.86,
      "grad_norm": 1.8187904357910156,
      "learning_rate": 2.6196791997484664e-06,
      "loss": 0.4853,
      "step": 239080
    },
    {
      "epoch": 97.87,
      "grad_norm": 1.926135540008545,
      "learning_rate": 2.6195216727302245e-06,
      "loss": 0.4853,
      "step": 239090
    },
    {
      "epoch": 97.87,
      "grad_norm": 1.916103720664978,
      "learning_rate": 2.6193641444487164e-06,
      "loss": 0.4791,
      "step": 239100
    },
    {
      "epoch": 97.88,
      "grad_norm": 1.813450574874878,
      "learning_rate": 2.619206614904661e-06,
      "loss": 0.474,
      "step": 239110
    },
    {
      "epoch": 97.88,
      "grad_norm": 2.236538887023926,
      "learning_rate": 2.6190490840987814e-06,
      "loss": 0.4778,
      "step": 239120
    },
    {
      "epoch": 97.88,
      "grad_norm": 1.8902099132537842,
      "learning_rate": 2.6188915520318005e-06,
      "loss": 0.48,
      "step": 239130
    },
    {
      "epoch": 97.89,
      "grad_norm": 2.0077126026153564,
      "learning_rate": 2.6187340187044383e-06,
      "loss": 0.4746,
      "step": 239140
    },
    {
      "epoch": 97.89,
      "grad_norm": 1.9037050008773804,
      "learning_rate": 2.6185764841174166e-06,
      "loss": 0.4748,
      "step": 239150
    },
    {
      "epoch": 97.9,
      "grad_norm": 1.7101619243621826,
      "learning_rate": 2.618418948271458e-06,
      "loss": 0.479,
      "step": 239160
    },
    {
      "epoch": 97.9,
      "grad_norm": 1.5616823434829712,
      "learning_rate": 2.6182614111672834e-06,
      "loss": 0.5031,
      "step": 239170
    },
    {
      "epoch": 97.9,
      "grad_norm": 1.9067986011505127,
      "learning_rate": 2.618103872805614e-06,
      "loss": 0.4596,
      "step": 239180
    },
    {
      "epoch": 97.91,
      "grad_norm": 1.80559241771698,
      "learning_rate": 2.6179463331871725e-06,
      "loss": 0.4633,
      "step": 239190
    },
    {
      "epoch": 97.91,
      "grad_norm": 1.7376631498336792,
      "learning_rate": 2.61778879231268e-06,
      "loss": 0.4661,
      "step": 239200
    },
    {
      "epoch": 97.92,
      "grad_norm": 6.287152290344238,
      "learning_rate": 2.6176312501828577e-06,
      "loss": 0.4907,
      "step": 239210
    },
    {
      "epoch": 97.92,
      "grad_norm": 2.022836685180664,
      "learning_rate": 2.6174737067984277e-06,
      "loss": 0.4995,
      "step": 239220
    },
    {
      "epoch": 97.92,
      "grad_norm": 2.072803497314453,
      "learning_rate": 2.617316162160113e-06,
      "loss": 0.4665,
      "step": 239230
    },
    {
      "epoch": 97.93,
      "grad_norm": 2.3264429569244385,
      "learning_rate": 2.6171586162686326e-06,
      "loss": 0.4679,
      "step": 239240
    },
    {
      "epoch": 97.93,
      "grad_norm": 2.0949935913085938,
      "learning_rate": 2.617001069124711e-06,
      "loss": 0.5084,
      "step": 239250
    },
    {
      "epoch": 97.94,
      "grad_norm": 2.446758985519409,
      "learning_rate": 2.616843520729068e-06,
      "loss": 0.5034,
      "step": 239260
    },
    {
      "epoch": 97.94,
      "grad_norm": 1.5764820575714111,
      "learning_rate": 2.6166859710824255e-06,
      "loss": 0.4822,
      "step": 239270
    },
    {
      "epoch": 97.95,
      "grad_norm": 1.785037636756897,
      "learning_rate": 2.616528420185506e-06,
      "loss": 0.4926,
      "step": 239280
    },
    {
      "epoch": 97.95,
      "grad_norm": 2.336921215057373,
      "learning_rate": 2.6163708680390307e-06,
      "loss": 0.4592,
      "step": 239290
    },
    {
      "epoch": 97.95,
      "grad_norm": 1.9128438234329224,
      "learning_rate": 2.616213314643721e-06,
      "loss": 0.4787,
      "step": 239300
    },
    {
      "epoch": 97.96,
      "grad_norm": 1.9454452991485596,
      "learning_rate": 2.6160557600002997e-06,
      "loss": 0.5008,
      "step": 239310
    },
    {
      "epoch": 97.96,
      "grad_norm": 2.0626440048217773,
      "learning_rate": 2.6158982041094875e-06,
      "loss": 0.4997,
      "step": 239320
    },
    {
      "epoch": 97.97,
      "grad_norm": 1.9735060930252075,
      "learning_rate": 2.6157406469720072e-06,
      "loss": 0.4739,
      "step": 239330
    },
    {
      "epoch": 97.97,
      "grad_norm": 2.0515148639678955,
      "learning_rate": 2.615583088588579e-06,
      "loss": 0.4789,
      "step": 239340
    },
    {
      "epoch": 97.97,
      "grad_norm": 1.835176706314087,
      "learning_rate": 2.615425528959926e-06,
      "loss": 0.4665,
      "step": 239350
    },
    {
      "epoch": 97.98,
      "grad_norm": 1.9810577630996704,
      "learning_rate": 2.6152679680867692e-06,
      "loss": 0.4794,
      "step": 239360
    },
    {
      "epoch": 97.98,
      "grad_norm": 2.024657964706421,
      "learning_rate": 2.615110405969831e-06,
      "loss": 0.4579,
      "step": 239370
    },
    {
      "epoch": 97.99,
      "grad_norm": 1.8128602504730225,
      "learning_rate": 2.614952842609832e-06,
      "loss": 0.4804,
      "step": 239380
    },
    {
      "epoch": 97.99,
      "grad_norm": 2.1037380695343018,
      "learning_rate": 2.6147952780074966e-06,
      "loss": 0.4658,
      "step": 239390
    },
    {
      "epoch": 97.99,
      "grad_norm": 1.8693863153457642,
      "learning_rate": 2.614637712163543e-06,
      "loss": 0.4827,
      "step": 239400
    },
    {
      "epoch": 98.0,
      "grad_norm": 1.60349440574646,
      "learning_rate": 2.614480145078695e-06,
      "loss": 0.479,
      "step": 239410
    },
    {
      "epoch": 98.0,
      "eval_loss": 0.48297253251075745,
      "eval_runtime": 52.2242,
      "eval_samples_per_second": 66.042,
      "eval_steps_per_second": 8.272,
      "step": 239414
    },
    {
      "epoch": 98.0,
      "grad_norm": 1.8003069162368774,
      "learning_rate": 2.6143225767536745e-06,
      "loss": 0.5026,
      "step": 239420
    },
    {
      "epoch": 98.01,
      "grad_norm": 2.0298025608062744,
      "learning_rate": 2.6141650071892027e-06,
      "loss": 0.466,
      "step": 239430
    },
    {
      "epoch": 98.01,
      "grad_norm": 1.6522912979125977,
      "learning_rate": 2.614007436386002e-06,
      "loss": 0.4864,
      "step": 239440
    },
    {
      "epoch": 98.01,
      "grad_norm": 1.8410882949829102,
      "learning_rate": 2.613849864344793e-06,
      "loss": 0.4716,
      "step": 239450
    },
    {
      "epoch": 98.02,
      "grad_norm": 1.7517762184143066,
      "learning_rate": 2.6136922910662995e-06,
      "loss": 0.4776,
      "step": 239460
    },
    {
      "epoch": 98.02,
      "grad_norm": 1.8575401306152344,
      "learning_rate": 2.613534716551242e-06,
      "loss": 0.5007,
      "step": 239470
    },
    {
      "epoch": 98.03,
      "grad_norm": 1.428151249885559,
      "learning_rate": 2.613377140800342e-06,
      "loss": 0.4876,
      "step": 239480
    },
    {
      "epoch": 98.03,
      "grad_norm": 1.7568893432617188,
      "learning_rate": 2.613219563814322e-06,
      "loss": 0.4695,
      "step": 239490
    },
    {
      "epoch": 98.04,
      "grad_norm": 1.6028218269348145,
      "learning_rate": 2.613061985593903e-06,
      "loss": 0.4764,
      "step": 239500
    },
    {
      "epoch": 98.04,
      "grad_norm": 2.0037729740142822,
      "learning_rate": 2.6129044061398084e-06,
      "loss": 0.4827,
      "step": 239510
    },
    {
      "epoch": 98.04,
      "grad_norm": 1.9727946519851685,
      "learning_rate": 2.6127468254527585e-06,
      "loss": 0.4697,
      "step": 239520
    },
    {
      "epoch": 98.05,
      "grad_norm": 2.1509792804718018,
      "learning_rate": 2.6125892435334773e-06,
      "loss": 0.4749,
      "step": 239530
    },
    {
      "epoch": 98.05,
      "grad_norm": 1.7834733724594116,
      "learning_rate": 2.612431660382684e-06,
      "loss": 0.4562,
      "step": 239540
    },
    {
      "epoch": 98.06,
      "grad_norm": 1.6140633821487427,
      "learning_rate": 2.6122740760011015e-06,
      "loss": 0.4768,
      "step": 239550
    },
    {
      "epoch": 98.06,
      "grad_norm": 2.032625198364258,
      "learning_rate": 2.612116490389452e-06,
      "loss": 0.4765,
      "step": 239560
    },
    {
      "epoch": 98.06,
      "grad_norm": 2.57112979888916,
      "learning_rate": 2.6119589035484566e-06,
      "loss": 0.4826,
      "step": 239570
    },
    {
      "epoch": 98.07,
      "grad_norm": 2.0017428398132324,
      "learning_rate": 2.6118013154788387e-06,
      "loss": 0.4901,
      "step": 239580
    },
    {
      "epoch": 98.07,
      "grad_norm": 1.7772612571716309,
      "learning_rate": 2.6116437261813185e-06,
      "loss": 0.4917,
      "step": 239590
    },
    {
      "epoch": 98.08,
      "grad_norm": 3.3000097274780273,
      "learning_rate": 2.6114861356566188e-06,
      "loss": 0.4871,
      "step": 239600
    },
    {
      "epoch": 98.08,
      "grad_norm": 1.6731890439987183,
      "learning_rate": 2.611328543905462e-06,
      "loss": 0.4616,
      "step": 239610
    },
    {
      "epoch": 98.08,
      "grad_norm": 2.1893324851989746,
      "learning_rate": 2.6111709509285687e-06,
      "loss": 0.4592,
      "step": 239620
    },
    {
      "epoch": 98.09,
      "grad_norm": 1.7536616325378418,
      "learning_rate": 2.6110133567266618e-06,
      "loss": 0.4811,
      "step": 239630
    },
    {
      "epoch": 98.09,
      "grad_norm": 2.3059496879577637,
      "learning_rate": 2.610855761300462e-06,
      "loss": 0.4891,
      "step": 239640
    },
    {
      "epoch": 98.1,
      "grad_norm": 1.7595186233520508,
      "learning_rate": 2.6106981646506928e-06,
      "loss": 0.4598,
      "step": 239650
    },
    {
      "epoch": 98.1,
      "grad_norm": 1.8968552350997925,
      "learning_rate": 2.610540566778075e-06,
      "loss": 0.4734,
      "step": 239660
    },
    {
      "epoch": 98.1,
      "grad_norm": 1.5473017692565918,
      "learning_rate": 2.6103829676833313e-06,
      "loss": 0.4691,
      "step": 239670
    },
    {
      "epoch": 98.11,
      "grad_norm": 1.6648873090744019,
      "learning_rate": 2.610225367367184e-06,
      "loss": 0.4863,
      "step": 239680
    },
    {
      "epoch": 98.11,
      "grad_norm": 1.7382837533950806,
      "learning_rate": 2.610067765830353e-06,
      "loss": 0.488,
      "step": 239690
    },
    {
      "epoch": 98.12,
      "grad_norm": 1.946425437927246,
      "learning_rate": 2.6099101630735618e-06,
      "loss": 0.4832,
      "step": 239700
    },
    {
      "epoch": 98.12,
      "grad_norm": 2.126772880554199,
      "learning_rate": 2.6097525590975324e-06,
      "loss": 0.4829,
      "step": 239710
    },
    {
      "epoch": 98.13,
      "grad_norm": 2.28796648979187,
      "learning_rate": 2.609594953902986e-06,
      "loss": 0.4856,
      "step": 239720
    },
    {
      "epoch": 98.13,
      "grad_norm": 2.3409533500671387,
      "learning_rate": 2.609437347490646e-06,
      "loss": 0.5012,
      "step": 239730
    },
    {
      "epoch": 98.13,
      "grad_norm": 1.7944159507751465,
      "learning_rate": 2.6092797398612322e-06,
      "loss": 0.4808,
      "step": 239740
    },
    {
      "epoch": 98.14,
      "grad_norm": 1.6923637390136719,
      "learning_rate": 2.6091221310154694e-06,
      "loss": 0.4675,
      "step": 239750
    },
    {
      "epoch": 98.14,
      "grad_norm": 1.7422538995742798,
      "learning_rate": 2.6089645209540767e-06,
      "loss": 0.4915,
      "step": 239760
    },
    {
      "epoch": 98.15,
      "grad_norm": 1.8107638359069824,
      "learning_rate": 2.608806909677778e-06,
      "loss": 0.4835,
      "step": 239770
    },
    {
      "epoch": 98.15,
      "grad_norm": 2.0331790447235107,
      "learning_rate": 2.6086492971872937e-06,
      "loss": 0.488,
      "step": 239780
    },
    {
      "epoch": 98.15,
      "grad_norm": 1.9970637559890747,
      "learning_rate": 2.608491683483347e-06,
      "loss": 0.4818,
      "step": 239790
    },
    {
      "epoch": 98.16,
      "grad_norm": 2.336214780807495,
      "learning_rate": 2.6083340685666604e-06,
      "loss": 0.4758,
      "step": 239800
    },
    {
      "epoch": 98.16,
      "grad_norm": 1.9598190784454346,
      "learning_rate": 2.608176452437955e-06,
      "loss": 0.4694,
      "step": 239810
    },
    {
      "epoch": 98.17,
      "grad_norm": 1.9870442152023315,
      "learning_rate": 2.608018835097953e-06,
      "loss": 0.48,
      "step": 239820
    },
    {
      "epoch": 98.17,
      "grad_norm": 1.6946868896484375,
      "learning_rate": 2.6078612165473764e-06,
      "loss": 0.4848,
      "step": 239830
    },
    {
      "epoch": 98.17,
      "grad_norm": 1.654133677482605,
      "learning_rate": 2.6077035967869465e-06,
      "loss": 0.4759,
      "step": 239840
    },
    {
      "epoch": 98.18,
      "grad_norm": 2.058929204940796,
      "learning_rate": 2.6075459758173864e-06,
      "loss": 0.495,
      "step": 239850
    },
    {
      "epoch": 98.18,
      "grad_norm": 1.6588364839553833,
      "learning_rate": 2.607388353639418e-06,
      "loss": 0.4735,
      "step": 239860
    },
    {
      "epoch": 98.19,
      "grad_norm": 1.8158701658248901,
      "learning_rate": 2.607230730253763e-06,
      "loss": 0.4634,
      "step": 239870
    },
    {
      "epoch": 98.19,
      "grad_norm": 1.9721102714538574,
      "learning_rate": 2.6070731056611437e-06,
      "loss": 0.4744,
      "step": 239880
    },
    {
      "epoch": 98.19,
      "grad_norm": 1.705596685409546,
      "learning_rate": 2.606915479862282e-06,
      "loss": 0.4725,
      "step": 239890
    },
    {
      "epoch": 98.2,
      "grad_norm": 1.7076810598373413,
      "learning_rate": 2.6067578528579e-06,
      "loss": 0.4954,
      "step": 239900
    },
    {
      "epoch": 98.2,
      "grad_norm": 1.4252184629440308,
      "learning_rate": 2.6066002246487194e-06,
      "loss": 0.4808,
      "step": 239910
    },
    {
      "epoch": 98.21,
      "grad_norm": 2.0079846382141113,
      "learning_rate": 2.606442595235463e-06,
      "loss": 0.4976,
      "step": 239920
    },
    {
      "epoch": 98.21,
      "grad_norm": 1.7796510457992554,
      "learning_rate": 2.6062849646188525e-06,
      "loss": 0.4896,
      "step": 239930
    },
    {
      "epoch": 98.22,
      "grad_norm": 1.6378235816955566,
      "learning_rate": 2.60612733279961e-06,
      "loss": 0.4992,
      "step": 239940
    },
    {
      "epoch": 98.22,
      "grad_norm": 2.4690194129943848,
      "learning_rate": 2.605969699778457e-06,
      "loss": 0.4818,
      "step": 239950
    },
    {
      "epoch": 98.22,
      "grad_norm": 2.3098838329315186,
      "learning_rate": 2.6058120655561167e-06,
      "loss": 0.4785,
      "step": 239960
    },
    {
      "epoch": 98.23,
      "grad_norm": 1.854427695274353,
      "learning_rate": 2.6056544301333107e-06,
      "loss": 0.4604,
      "step": 239970
    },
    {
      "epoch": 98.23,
      "grad_norm": 1.7055087089538574,
      "learning_rate": 2.605496793510761e-06,
      "loss": 0.4669,
      "step": 239980
    },
    {
      "epoch": 98.24,
      "grad_norm": 1.7467706203460693,
      "learning_rate": 2.6053391556891897e-06,
      "loss": 0.4565,
      "step": 239990
    },
    {
      "epoch": 98.24,
      "grad_norm": 2.1571767330169678,
      "learning_rate": 2.605181516669319e-06,
      "loss": 0.4897,
      "step": 240000
    },
    {
      "epoch": 98.24,
      "grad_norm": 1.962561011314392,
      "learning_rate": 2.605023876451871e-06,
      "loss": 0.4735,
      "step": 240010
    },
    {
      "epoch": 98.25,
      "grad_norm": 2.345874309539795,
      "learning_rate": 2.604866235037568e-06,
      "loss": 0.4876,
      "step": 240020
    },
    {
      "epoch": 98.25,
      "grad_norm": 1.9283485412597656,
      "learning_rate": 2.6047085924271317e-06,
      "loss": 0.4664,
      "step": 240030
    },
    {
      "epoch": 98.26,
      "grad_norm": 1.8776415586471558,
      "learning_rate": 2.6045509486212855e-06,
      "loss": 0.4876,
      "step": 240040
    },
    {
      "epoch": 98.26,
      "grad_norm": 2.0797319412231445,
      "learning_rate": 2.6043933036207493e-06,
      "loss": 0.4644,
      "step": 240050
    },
    {
      "epoch": 98.26,
      "grad_norm": 1.6170406341552734,
      "learning_rate": 2.604235657426247e-06,
      "loss": 0.4621,
      "step": 240060
    },
    {
      "epoch": 98.27,
      "grad_norm": 2.1330788135528564,
      "learning_rate": 2.6040780100384996e-06,
      "loss": 0.4898,
      "step": 240070
    },
    {
      "epoch": 98.27,
      "grad_norm": 2.028365135192871,
      "learning_rate": 2.6039203614582302e-06,
      "loss": 0.491,
      "step": 240080
    },
    {
      "epoch": 98.28,
      "grad_norm": 1.8525242805480957,
      "learning_rate": 2.6037627116861614e-06,
      "loss": 0.4863,
      "step": 240090
    },
    {
      "epoch": 98.28,
      "grad_norm": 1.9559687376022339,
      "learning_rate": 2.603605060723014e-06,
      "loss": 0.486,
      "step": 240100
    },
    {
      "epoch": 98.28,
      "grad_norm": 1.994160532951355,
      "learning_rate": 2.6034474085695114e-06,
      "loss": 0.4932,
      "step": 240110
    },
    {
      "epoch": 98.29,
      "grad_norm": 1.832227349281311,
      "learning_rate": 2.6032897552263746e-06,
      "loss": 0.4718,
      "step": 240120
    },
    {
      "epoch": 98.29,
      "grad_norm": 2.130784749984741,
      "learning_rate": 2.6031321006943266e-06,
      "loss": 0.4847,
      "step": 240130
    },
    {
      "epoch": 98.3,
      "grad_norm": 1.9396638870239258,
      "learning_rate": 2.602974444974089e-06,
      "loss": 0.486,
      "step": 240140
    },
    {
      "epoch": 98.3,
      "grad_norm": 2.333561897277832,
      "learning_rate": 2.6028167880663847e-06,
      "loss": 0.481,
      "step": 240150
    },
    {
      "epoch": 98.31,
      "grad_norm": 2.285867691040039,
      "learning_rate": 2.602659129971935e-06,
      "loss": 0.495,
      "step": 240160
    },
    {
      "epoch": 98.31,
      "grad_norm": 1.861824631690979,
      "learning_rate": 2.602501470691463e-06,
      "loss": 0.4722,
      "step": 240170
    },
    {
      "epoch": 98.31,
      "grad_norm": 1.5953154563903809,
      "learning_rate": 2.6023438102256915e-06,
      "loss": 0.4836,
      "step": 240180
    },
    {
      "epoch": 98.32,
      "grad_norm": 1.6940364837646484,
      "learning_rate": 2.602186148575341e-06,
      "loss": 0.4871,
      "step": 240190
    },
    {
      "epoch": 98.32,
      "grad_norm": 1.6442053318023682,
      "learning_rate": 2.602028485741135e-06,
      "loss": 0.4766,
      "step": 240200
    },
    {
      "epoch": 98.33,
      "grad_norm": 1.9289451837539673,
      "learning_rate": 2.6018708217237947e-06,
      "loss": 0.464,
      "step": 240210
    },
    {
      "epoch": 98.33,
      "grad_norm": 1.8834620714187622,
      "learning_rate": 2.601713156524043e-06,
      "loss": 0.4937,
      "step": 240220
    },
    {
      "epoch": 98.33,
      "grad_norm": 1.9844974279403687,
      "learning_rate": 2.6015554901426017e-06,
      "loss": 0.4909,
      "step": 240230
    },
    {
      "epoch": 98.34,
      "grad_norm": 1.8314250707626343,
      "learning_rate": 2.6013978225801944e-06,
      "loss": 0.4749,
      "step": 240240
    },
    {
      "epoch": 98.34,
      "grad_norm": 1.662465214729309,
      "learning_rate": 2.601240153837541e-06,
      "loss": 0.4833,
      "step": 240250
    },
    {
      "epoch": 98.35,
      "grad_norm": 1.599281907081604,
      "learning_rate": 2.601082483915366e-06,
      "loss": 0.4923,
      "step": 240260
    },
    {
      "epoch": 98.35,
      "grad_norm": 1.8645343780517578,
      "learning_rate": 2.6009248128143906e-06,
      "loss": 0.4949,
      "step": 240270
    },
    {
      "epoch": 98.35,
      "grad_norm": 1.8653688430786133,
      "learning_rate": 2.6007671405353375e-06,
      "loss": 0.4884,
      "step": 240280
    },
    {
      "epoch": 98.36,
      "grad_norm": 2.2880730628967285,
      "learning_rate": 2.6006094670789277e-06,
      "loss": 0.5031,
      "step": 240290
    },
    {
      "epoch": 98.36,
      "grad_norm": 2.047416925430298,
      "learning_rate": 2.600451792445885e-06,
      "loss": 0.4841,
      "step": 240300
    },
    {
      "epoch": 98.37,
      "grad_norm": 2.2711918354034424,
      "learning_rate": 2.6002941166369312e-06,
      "loss": 0.4928,
      "step": 240310
    },
    {
      "epoch": 98.37,
      "grad_norm": 2.2916438579559326,
      "learning_rate": 2.6001364396527884e-06,
      "loss": 0.4811,
      "step": 240320
    },
    {
      "epoch": 98.37,
      "grad_norm": 2.06756329536438,
      "learning_rate": 2.59997876149418e-06,
      "loss": 0.4679,
      "step": 240330
    },
    {
      "epoch": 98.38,
      "grad_norm": 1.9261136054992676,
      "learning_rate": 2.5998210821618266e-06,
      "loss": 0.4869,
      "step": 240340
    },
    {
      "epoch": 98.38,
      "grad_norm": 1.5140397548675537,
      "learning_rate": 2.5996634016564505e-06,
      "loss": 0.457,
      "step": 240350
    },
    {
      "epoch": 98.39,
      "grad_norm": 1.9994540214538574,
      "learning_rate": 2.5995057199787757e-06,
      "loss": 0.4922,
      "step": 240360
    },
    {
      "epoch": 98.39,
      "grad_norm": 1.6713887453079224,
      "learning_rate": 2.5993480371295227e-06,
      "loss": 0.4667,
      "step": 240370
    },
    {
      "epoch": 98.4,
      "grad_norm": 1.5003118515014648,
      "learning_rate": 2.5991903531094156e-06,
      "loss": 0.478,
      "step": 240380
    },
    {
      "epoch": 98.4,
      "grad_norm": 2.01967191696167,
      "learning_rate": 2.599032667919175e-06,
      "loss": 0.5046,
      "step": 240390
    },
    {
      "epoch": 98.4,
      "grad_norm": 1.990851879119873,
      "learning_rate": 2.598874981559525e-06,
      "loss": 0.4759,
      "step": 240400
    },
    {
      "epoch": 98.41,
      "grad_norm": 2.059288263320923,
      "learning_rate": 2.5987172940311865e-06,
      "loss": 0.4936,
      "step": 240410
    },
    {
      "epoch": 98.41,
      "grad_norm": 1.9258837699890137,
      "learning_rate": 2.598559605334882e-06,
      "loss": 0.4857,
      "step": 240420
    },
    {
      "epoch": 98.42,
      "grad_norm": 1.8706706762313843,
      "learning_rate": 2.5984019154713343e-06,
      "loss": 0.4632,
      "step": 240430
    },
    {
      "epoch": 98.42,
      "grad_norm": 1.762839674949646,
      "learning_rate": 2.598244224441265e-06,
      "loss": 0.4691,
      "step": 240440
    },
    {
      "epoch": 98.42,
      "grad_norm": 2.4307589530944824,
      "learning_rate": 2.598086532245398e-06,
      "loss": 0.4948,
      "step": 240450
    },
    {
      "epoch": 98.43,
      "grad_norm": 1.9477022886276245,
      "learning_rate": 2.597928838884455e-06,
      "loss": 0.467,
      "step": 240460
    },
    {
      "epoch": 98.43,
      "grad_norm": 2.189546585083008,
      "learning_rate": 2.597771144359158e-06,
      "loss": 0.4918,
      "step": 240470
    },
    {
      "epoch": 98.44,
      "grad_norm": 2.0904898643493652,
      "learning_rate": 2.5976134486702288e-06,
      "loss": 0.4745,
      "step": 240480
    },
    {
      "epoch": 98.44,
      "grad_norm": 1.8436983823776245,
      "learning_rate": 2.5974557518183908e-06,
      "loss": 0.4653,
      "step": 240490
    },
    {
      "epoch": 98.44,
      "grad_norm": 2.126101493835449,
      "learning_rate": 2.597298053804366e-06,
      "loss": 0.476,
      "step": 240500
    },
    {
      "epoch": 98.45,
      "grad_norm": 1.9914499521255493,
      "learning_rate": 2.597140354628877e-06,
      "loss": 0.4841,
      "step": 240510
    },
    {
      "epoch": 98.45,
      "grad_norm": 1.99906587600708,
      "learning_rate": 2.5969826542926455e-06,
      "loss": 0.4888,
      "step": 240520
    },
    {
      "epoch": 98.46,
      "grad_norm": 1.9325453042984009,
      "learning_rate": 2.5968249527963945e-06,
      "loss": 0.4718,
      "step": 240530
    },
    {
      "epoch": 98.46,
      "grad_norm": 1.7365458011627197,
      "learning_rate": 2.596667250140847e-06,
      "loss": 0.48,
      "step": 240540
    },
    {
      "epoch": 98.47,
      "grad_norm": 2.1402599811553955,
      "learning_rate": 2.596509546326724e-06,
      "loss": 0.4692,
      "step": 240550
    },
    {
      "epoch": 98.47,
      "grad_norm": 1.7771995067596436,
      "learning_rate": 2.5963518413547488e-06,
      "loss": 0.5021,
      "step": 240560
    },
    {
      "epoch": 98.47,
      "grad_norm": 1.9463735818862915,
      "learning_rate": 2.5961941352256443e-06,
      "loss": 0.4921,
      "step": 240570
    },
    {
      "epoch": 98.48,
      "grad_norm": 1.7944806814193726,
      "learning_rate": 2.5960364279401315e-06,
      "loss": 0.4993,
      "step": 240580
    },
    {
      "epoch": 98.48,
      "grad_norm": 2.295788049697876,
      "learning_rate": 2.5958787194989346e-06,
      "loss": 0.5002,
      "step": 240590
    },
    {
      "epoch": 98.49,
      "grad_norm": 1.9904292821884155,
      "learning_rate": 2.595721009902774e-06,
      "loss": 0.4749,
      "step": 240600
    },
    {
      "epoch": 98.49,
      "grad_norm": 1.6338133811950684,
      "learning_rate": 2.595563299152374e-06,
      "loss": 0.4813,
      "step": 240610
    },
    {
      "epoch": 98.49,
      "grad_norm": 2.7564008235931396,
      "learning_rate": 2.5954055872484566e-06,
      "loss": 0.4909,
      "step": 240620
    },
    {
      "epoch": 98.5,
      "grad_norm": 2.2903928756713867,
      "learning_rate": 2.595247874191743e-06,
      "loss": 0.4764,
      "step": 240630
    },
    {
      "epoch": 98.5,
      "grad_norm": 2.4634385108947754,
      "learning_rate": 2.5950901599829567e-06,
      "loss": 0.4877,
      "step": 240640
    },
    {
      "epoch": 98.51,
      "grad_norm": 2.0051205158233643,
      "learning_rate": 2.5949324446228202e-06,
      "loss": 0.4924,
      "step": 240650
    },
    {
      "epoch": 98.51,
      "grad_norm": 2.216493606567383,
      "learning_rate": 2.594774728112056e-06,
      "loss": 0.4753,
      "step": 240660
    },
    {
      "epoch": 98.51,
      "grad_norm": 1.776942253112793,
      "learning_rate": 2.5946170104513866e-06,
      "loss": 0.4929,
      "step": 240670
    },
    {
      "epoch": 98.52,
      "grad_norm": 2.5570456981658936,
      "learning_rate": 2.5944592916415336e-06,
      "loss": 0.4641,
      "step": 240680
    },
    {
      "epoch": 98.52,
      "grad_norm": 1.7683204412460327,
      "learning_rate": 2.594301571683221e-06,
      "loss": 0.4925,
      "step": 240690
    },
    {
      "epoch": 98.53,
      "grad_norm": 1.980830430984497,
      "learning_rate": 2.5941438505771703e-06,
      "loss": 0.4787,
      "step": 240700
    },
    {
      "epoch": 98.53,
      "grad_norm": 2.0040910243988037,
      "learning_rate": 2.5939861283241037e-06,
      "loss": 0.4835,
      "step": 240710
    },
    {
      "epoch": 98.53,
      "grad_norm": 1.8100429773330688,
      "learning_rate": 2.5938284049247437e-06,
      "loss": 0.4898,
      "step": 240720
    },
    {
      "epoch": 98.54,
      "grad_norm": 2.057271957397461,
      "learning_rate": 2.5936706803798136e-06,
      "loss": 0.469,
      "step": 240730
    },
    {
      "epoch": 98.54,
      "grad_norm": 2.3477981090545654,
      "learning_rate": 2.5935129546900362e-06,
      "loss": 0.4857,
      "step": 240740
    },
    {
      "epoch": 98.55,
      "grad_norm": 2.032968282699585,
      "learning_rate": 2.593355227856133e-06,
      "loss": 0.4752,
      "step": 240750
    },
    {
      "epoch": 98.55,
      "grad_norm": 1.6666970252990723,
      "learning_rate": 2.5931974998788275e-06,
      "loss": 0.4885,
      "step": 240760
    },
    {
      "epoch": 98.56,
      "grad_norm": 2.0963027477264404,
      "learning_rate": 2.593039770758841e-06,
      "loss": 0.4664,
      "step": 240770
    },
    {
      "epoch": 98.56,
      "grad_norm": 2.3545093536376953,
      "learning_rate": 2.592882040496897e-06,
      "loss": 0.5,
      "step": 240780
    },
    {
      "epoch": 98.56,
      "grad_norm": 1.558821201324463,
      "learning_rate": 2.592724309093717e-06,
      "loss": 0.4469,
      "step": 240790
    },
    {
      "epoch": 98.57,
      "grad_norm": 2.115129232406616,
      "learning_rate": 2.592566576550025e-06,
      "loss": 0.4769,
      "step": 240800
    },
    {
      "epoch": 98.57,
      "grad_norm": 2.1714625358581543,
      "learning_rate": 2.5924088428665423e-06,
      "loss": 0.4654,
      "step": 240810
    },
    {
      "epoch": 98.58,
      "grad_norm": 1.7525516748428345,
      "learning_rate": 2.5922511080439913e-06,
      "loss": 0.4617,
      "step": 240820
    },
    {
      "epoch": 98.58,
      "grad_norm": 1.605749487876892,
      "learning_rate": 2.592093372083097e-06,
      "loss": 0.4925,
      "step": 240830
    },
    {
      "epoch": 98.58,
      "grad_norm": 1.7829962968826294,
      "learning_rate": 2.5919356349845792e-06,
      "loss": 0.4668,
      "step": 240840
    },
    {
      "epoch": 98.59,
      "grad_norm": 2.326253890991211,
      "learning_rate": 2.5917778967491614e-06,
      "loss": 0.4786,
      "step": 240850
    },
    {
      "epoch": 98.59,
      "grad_norm": 1.9649641513824463,
      "learning_rate": 2.5916201573775663e-06,
      "loss": 0.4742,
      "step": 240860
    },
    {
      "epoch": 98.6,
      "grad_norm": 2.5212931632995605,
      "learning_rate": 2.5914624168705162e-06,
      "loss": 0.4805,
      "step": 240870
    },
    {
      "epoch": 98.6,
      "grad_norm": 1.689160943031311,
      "learning_rate": 2.5913046752287342e-06,
      "loss": 0.4967,
      "step": 240880
    },
    {
      "epoch": 98.6,
      "grad_norm": 1.827184796333313,
      "learning_rate": 2.591146932452942e-06,
      "loss": 0.4709,
      "step": 240890
    },
    {
      "epoch": 98.61,
      "grad_norm": 1.9025079011917114,
      "learning_rate": 2.5909891885438633e-06,
      "loss": 0.4689,
      "step": 240900
    },
    {
      "epoch": 98.61,
      "grad_norm": 1.870065689086914,
      "learning_rate": 2.5908314435022203e-06,
      "loss": 0.4912,
      "step": 240910
    },
    {
      "epoch": 98.62,
      "grad_norm": 1.7701648473739624,
      "learning_rate": 2.5906736973287354e-06,
      "loss": 0.4985,
      "step": 240920
    },
    {
      "epoch": 98.62,
      "grad_norm": 1.8511065244674683,
      "learning_rate": 2.5905159500241303e-06,
      "loss": 0.4878,
      "step": 240930
    },
    {
      "epoch": 98.62,
      "grad_norm": 2.2280540466308594,
      "learning_rate": 2.59035820158913e-06,
      "loss": 0.5046,
      "step": 240940
    },
    {
      "epoch": 98.63,
      "grad_norm": 2.365901231765747,
      "learning_rate": 2.590200452024455e-06,
      "loss": 0.4807,
      "step": 240950
    },
    {
      "epoch": 98.63,
      "grad_norm": 1.9789565801620483,
      "learning_rate": 2.590042701330829e-06,
      "loss": 0.4749,
      "step": 240960
    },
    {
      "epoch": 98.64,
      "grad_norm": 2.238443613052368,
      "learning_rate": 2.589884949508974e-06,
      "loss": 0.4813,
      "step": 240970
    },
    {
      "epoch": 98.64,
      "grad_norm": 2.074327230453491,
      "learning_rate": 2.589727196559613e-06,
      "loss": 0.4942,
      "step": 240980
    },
    {
      "epoch": 98.65,
      "grad_norm": 2.1948812007904053,
      "learning_rate": 2.5895694424834687e-06,
      "loss": 0.4912,
      "step": 240990
    },
    {
      "epoch": 98.65,
      "grad_norm": 1.5483145713806152,
      "learning_rate": 2.589411687281264e-06,
      "loss": 0.4944,
      "step": 241000
    },
    {
      "epoch": 98.65,
      "grad_norm": 2.073949098587036,
      "learning_rate": 2.58925393095372e-06,
      "loss": 0.4839,
      "step": 241010
    },
    {
      "epoch": 98.66,
      "grad_norm": 1.8260639905929565,
      "learning_rate": 2.5890961735015607e-06,
      "loss": 0.4721,
      "step": 241020
    },
    {
      "epoch": 98.66,
      "grad_norm": 1.7377058267593384,
      "learning_rate": 2.588938414925509e-06,
      "loss": 0.4892,
      "step": 241030
    },
    {
      "epoch": 98.67,
      "grad_norm": 1.8664432764053345,
      "learning_rate": 2.588780655226287e-06,
      "loss": 0.4778,
      "step": 241040
    },
    {
      "epoch": 98.67,
      "grad_norm": 2.7566521167755127,
      "learning_rate": 2.5886228944046182e-06,
      "loss": 0.4686,
      "step": 241050
    },
    {
      "epoch": 98.67,
      "grad_norm": 1.7792648077011108,
      "learning_rate": 2.5884651324612243e-06,
      "loss": 0.4888,
      "step": 241060
    },
    {
      "epoch": 98.68,
      "grad_norm": 1.6627637147903442,
      "learning_rate": 2.5883073693968275e-06,
      "loss": 0.4676,
      "step": 241070
    },
    {
      "epoch": 98.68,
      "grad_norm": 1.9796826839447021,
      "learning_rate": 2.588149605212152e-06,
      "loss": 0.478,
      "step": 241080
    },
    {
      "epoch": 98.69,
      "grad_norm": 2.3192906379699707,
      "learning_rate": 2.5879918399079196e-06,
      "loss": 0.4655,
      "step": 241090
    },
    {
      "epoch": 98.69,
      "grad_norm": 2.040212392807007,
      "learning_rate": 2.5878340734848525e-06,
      "loss": 0.4581,
      "step": 241100
    },
    {
      "epoch": 98.69,
      "grad_norm": 2.2382142543792725,
      "learning_rate": 2.5876763059436745e-06,
      "loss": 0.4828,
      "step": 241110
    },
    {
      "epoch": 98.7,
      "grad_norm": 1.8313268423080444,
      "learning_rate": 2.5875185372851083e-06,
      "loss": 0.4754,
      "step": 241120
    },
    {
      "epoch": 98.7,
      "grad_norm": 1.9382153749465942,
      "learning_rate": 2.5873607675098762e-06,
      "loss": 0.4907,
      "step": 241130
    },
    {
      "epoch": 98.71,
      "grad_norm": 2.0018200874328613,
      "learning_rate": 2.5872029966187007e-06,
      "loss": 0.481,
      "step": 241140
    },
    {
      "epoch": 98.71,
      "grad_norm": 2.064720869064331,
      "learning_rate": 2.5870452246123046e-06,
      "loss": 0.4867,
      "step": 241150
    },
    {
      "epoch": 98.71,
      "grad_norm": 1.7832838296890259,
      "learning_rate": 2.586887451491411e-06,
      "loss": 0.4752,
      "step": 241160
    },
    {
      "epoch": 98.72,
      "grad_norm": 2.1932766437530518,
      "learning_rate": 2.5867296772567417e-06,
      "loss": 0.4902,
      "step": 241170
    },
    {
      "epoch": 98.72,
      "grad_norm": 1.9190009832382202,
      "learning_rate": 2.5865719019090205e-06,
      "loss": 0.4826,
      "step": 241180
    },
    {
      "epoch": 98.73,
      "grad_norm": 1.6393342018127441,
      "learning_rate": 2.58641412544897e-06,
      "loss": 0.4751,
      "step": 241190
    },
    {
      "epoch": 98.73,
      "grad_norm": 1.787710428237915,
      "learning_rate": 2.5862563478773125e-06,
      "loss": 0.4832,
      "step": 241200
    },
    {
      "epoch": 98.74,
      "grad_norm": 2.079779624938965,
      "learning_rate": 2.5860985691947713e-06,
      "loss": 0.4732,
      "step": 241210
    },
    {
      "epoch": 98.74,
      "grad_norm": 1.9984705448150635,
      "learning_rate": 2.5859407894020685e-06,
      "loss": 0.48,
      "step": 241220
    },
    {
      "epoch": 98.74,
      "grad_norm": 1.9409949779510498,
      "learning_rate": 2.585783008499928e-06,
      "loss": 0.4744,
      "step": 241230
    },
    {
      "epoch": 98.75,
      "grad_norm": 1.7873330116271973,
      "learning_rate": 2.5856252264890706e-06,
      "loss": 0.4786,
      "step": 241240
    },
    {
      "epoch": 98.75,
      "grad_norm": 1.9347825050354004,
      "learning_rate": 2.585467443370221e-06,
      "loss": 0.4756,
      "step": 241250
    },
    {
      "epoch": 98.76,
      "grad_norm": 1.6780879497528076,
      "learning_rate": 2.5853096591441007e-06,
      "loss": 0.4838,
      "step": 241260
    },
    {
      "epoch": 98.76,
      "grad_norm": 2.020700454711914,
      "learning_rate": 2.5851518738114335e-06,
      "loss": 0.4898,
      "step": 241270
    },
    {
      "epoch": 98.76,
      "grad_norm": 1.7160495519638062,
      "learning_rate": 2.5849940873729415e-06,
      "loss": 0.4849,
      "step": 241280
    },
    {
      "epoch": 98.77,
      "grad_norm": 1.9772473573684692,
      "learning_rate": 2.5848362998293468e-06,
      "loss": 0.4659,
      "step": 241290
    },
    {
      "epoch": 98.77,
      "grad_norm": 1.84742271900177,
      "learning_rate": 2.5846785111813744e-06,
      "loss": 0.4694,
      "step": 241300
    },
    {
      "epoch": 98.78,
      "grad_norm": 1.5359487533569336,
      "learning_rate": 2.5845207214297452e-06,
      "loss": 0.4841,
      "step": 241310
    },
    {
      "epoch": 98.78,
      "grad_norm": 1.9687318801879883,
      "learning_rate": 2.584362930575183e-06,
      "loss": 0.4521,
      "step": 241320
    },
    {
      "epoch": 98.78,
      "grad_norm": 2.0581045150756836,
      "learning_rate": 2.58420513861841e-06,
      "loss": 0.4923,
      "step": 241330
    },
    {
      "epoch": 98.79,
      "grad_norm": 2.321524143218994,
      "learning_rate": 2.5840473455601495e-06,
      "loss": 0.469,
      "step": 241340
    },
    {
      "epoch": 98.79,
      "grad_norm": 2.0114943981170654,
      "learning_rate": 2.5838895514011237e-06,
      "loss": 0.5011,
      "step": 241350
    },
    {
      "epoch": 98.8,
      "grad_norm": 1.9437214136123657,
      "learning_rate": 2.5837317561420562e-06,
      "loss": 0.4814,
      "step": 241360
    },
    {
      "epoch": 98.8,
      "grad_norm": 2.1834025382995605,
      "learning_rate": 2.583573959783669e-06,
      "loss": 0.4939,
      "step": 241370
    },
    {
      "epoch": 98.8,
      "grad_norm": 2.2334201335906982,
      "learning_rate": 2.583416162326685e-06,
      "loss": 0.4998,
      "step": 241380
    },
    {
      "epoch": 98.81,
      "grad_norm": 1.9766757488250732,
      "learning_rate": 2.5832583637718284e-06,
      "loss": 0.4676,
      "step": 241390
    },
    {
      "epoch": 98.81,
      "grad_norm": 1.7514469623565674,
      "learning_rate": 2.5831005641198207e-06,
      "loss": 0.4676,
      "step": 241400
    },
    {
      "epoch": 98.82,
      "grad_norm": 2.179014205932617,
      "learning_rate": 2.5829427633713855e-06,
      "loss": 0.4764,
      "step": 241410
    },
    {
      "epoch": 98.82,
      "grad_norm": 1.8226518630981445,
      "learning_rate": 2.5827849615272446e-06,
      "loss": 0.478,
      "step": 241420
    },
    {
      "epoch": 98.83,
      "grad_norm": 1.928520679473877,
      "learning_rate": 2.582627158588122e-06,
      "loss": 0.4903,
      "step": 241430
    },
    {
      "epoch": 98.83,
      "grad_norm": 2.1364974975585938,
      "learning_rate": 2.58246935455474e-06,
      "loss": 0.4983,
      "step": 241440
    },
    {
      "epoch": 98.83,
      "grad_norm": 1.727381944656372,
      "learning_rate": 2.5823115494278216e-06,
      "loss": 0.4722,
      "step": 241450
    },
    {
      "epoch": 98.84,
      "grad_norm": 2.235919237136841,
      "learning_rate": 2.5821537432080898e-06,
      "loss": 0.4663,
      "step": 241460
    },
    {
      "epoch": 98.84,
      "grad_norm": 2.076028823852539,
      "learning_rate": 2.5819959358962674e-06,
      "loss": 0.4729,
      "step": 241470
    },
    {
      "epoch": 98.85,
      "grad_norm": 2.6121015548706055,
      "learning_rate": 2.581838127493077e-06,
      "loss": 0.4677,
      "step": 241480
    },
    {
      "epoch": 98.85,
      "grad_norm": 2.1548409461975098,
      "learning_rate": 2.5816803179992416e-06,
      "loss": 0.489,
      "step": 241490
    },
    {
      "epoch": 98.85,
      "grad_norm": 1.7727792263031006,
      "learning_rate": 2.5815225074154847e-06,
      "loss": 0.508,
      "step": 241500
    },
    {
      "epoch": 98.86,
      "grad_norm": 2.5696780681610107,
      "learning_rate": 2.581364695742529e-06,
      "loss": 0.4766,
      "step": 241510
    },
    {
      "epoch": 98.86,
      "grad_norm": 1.6931298971176147,
      "learning_rate": 2.5812068829810964e-06,
      "loss": 0.4847,
      "step": 241520
    },
    {
      "epoch": 98.87,
      "grad_norm": 2.305638074874878,
      "learning_rate": 2.581049069131911e-06,
      "loss": 0.4579,
      "step": 241530
    },
    {
      "epoch": 98.87,
      "grad_norm": 1.8036167621612549,
      "learning_rate": 2.580891254195695e-06,
      "loss": 0.4786,
      "step": 241540
    },
    {
      "epoch": 98.87,
      "grad_norm": 4.687670707702637,
      "learning_rate": 2.580733438173172e-06,
      "loss": 0.5052,
      "step": 241550
    },
    {
      "epoch": 98.88,
      "grad_norm": 1.6815004348754883,
      "learning_rate": 2.580575621065065e-06,
      "loss": 0.4973,
      "step": 241560
    },
    {
      "epoch": 98.88,
      "grad_norm": 2.2494585514068604,
      "learning_rate": 2.5804178028720957e-06,
      "loss": 0.4837,
      "step": 241570
    },
    {
      "epoch": 98.89,
      "grad_norm": 2.024064302444458,
      "learning_rate": 2.580259983594988e-06,
      "loss": 0.4794,
      "step": 241580
    },
    {
      "epoch": 98.89,
      "grad_norm": 2.1050314903259277,
      "learning_rate": 2.5801021632344646e-06,
      "loss": 0.4617,
      "step": 241590
    },
    {
      "epoch": 98.89,
      "grad_norm": 1.6643530130386353,
      "learning_rate": 2.579944341791249e-06,
      "loss": 0.4951,
      "step": 241600
    },
    {
      "epoch": 98.9,
      "grad_norm": 1.7635822296142578,
      "learning_rate": 2.5797865192660634e-06,
      "loss": 0.4759,
      "step": 241610
    },
    {
      "epoch": 98.9,
      "grad_norm": 2.049927234649658,
      "learning_rate": 2.5796286956596314e-06,
      "loss": 0.4871,
      "step": 241620
    },
    {
      "epoch": 98.91,
      "grad_norm": 1.9816232919692993,
      "learning_rate": 2.579470870972676e-06,
      "loss": 0.4771,
      "step": 241630
    },
    {
      "epoch": 98.91,
      "grad_norm": 1.8765897750854492,
      "learning_rate": 2.579313045205919e-06,
      "loss": 0.4693,
      "step": 241640
    },
    {
      "epoch": 98.92,
      "grad_norm": 1.5059648752212524,
      "learning_rate": 2.5791552183600845e-06,
      "loss": 0.4732,
      "step": 241650
    },
    {
      "epoch": 98.92,
      "grad_norm": 1.7926121950149536,
      "learning_rate": 2.5789973904358953e-06,
      "loss": 0.4846,
      "step": 241660
    },
    {
      "epoch": 98.92,
      "grad_norm": 1.6806527376174927,
      "learning_rate": 2.578839561434073e-06,
      "loss": 0.4432,
      "step": 241670
    },
    {
      "epoch": 98.93,
      "grad_norm": 2.202993631362915,
      "learning_rate": 2.578681731355343e-06,
      "loss": 0.5008,
      "step": 241680
    },
    {
      "epoch": 98.93,
      "grad_norm": 2.191763162612915,
      "learning_rate": 2.5785239002004265e-06,
      "loss": 0.4936,
      "step": 241690
    },
    {
      "epoch": 98.94,
      "grad_norm": 2.0151867866516113,
      "learning_rate": 2.578366067970048e-06,
      "loss": 0.4799,
      "step": 241700
    },
    {
      "epoch": 98.94,
      "grad_norm": 2.292379140853882,
      "learning_rate": 2.578208234664929e-06,
      "loss": 0.4882,
      "step": 241710
    },
    {
      "epoch": 98.94,
      "grad_norm": 1.9717553853988647,
      "learning_rate": 2.5780504002857936e-06,
      "loss": 0.4956,
      "step": 241720
    },
    {
      "epoch": 98.95,
      "grad_norm": 1.793624758720398,
      "learning_rate": 2.5778925648333636e-06,
      "loss": 0.4496,
      "step": 241730
    },
    {
      "epoch": 98.95,
      "grad_norm": 2.0105741024017334,
      "learning_rate": 2.5777347283083633e-06,
      "loss": 0.4823,
      "step": 241740
    },
    {
      "epoch": 98.96,
      "grad_norm": 3.0056421756744385,
      "learning_rate": 2.5775768907115156e-06,
      "loss": 0.4684,
      "step": 241750
    },
    {
      "epoch": 98.96,
      "grad_norm": 2.3066327571868896,
      "learning_rate": 2.577419052043542e-06,
      "loss": 0.491,
      "step": 241760
    },
    {
      "epoch": 98.96,
      "grad_norm": 2.0110902786254883,
      "learning_rate": 2.5772612123051675e-06,
      "loss": 0.4813,
      "step": 241770
    },
    {
      "epoch": 98.97,
      "grad_norm": 2.0805368423461914,
      "learning_rate": 2.5771033714971142e-06,
      "loss": 0.4685,
      "step": 241780
    },
    {
      "epoch": 98.97,
      "grad_norm": 2.0385046005249023,
      "learning_rate": 2.5769455296201053e-06,
      "loss": 0.4807,
      "step": 241790
    },
    {
      "epoch": 98.98,
      "grad_norm": 1.651772141456604,
      "learning_rate": 2.5767876866748634e-06,
      "loss": 0.4754,
      "step": 241800
    },
    {
      "epoch": 98.98,
      "grad_norm": 2.1068780422210693,
      "learning_rate": 2.5766298426621126e-06,
      "loss": 0.4768,
      "step": 241810
    },
    {
      "epoch": 98.98,
      "grad_norm": 1.7100238800048828,
      "learning_rate": 2.576471997582574e-06,
      "loss": 0.4841,
      "step": 241820
    },
    {
      "epoch": 98.99,
      "grad_norm": 1.6780880689620972,
      "learning_rate": 2.5763141514369733e-06,
      "loss": 0.4827,
      "step": 241830
    },
    {
      "epoch": 98.99,
      "grad_norm": 1.8925073146820068,
      "learning_rate": 2.576156304226032e-06,
      "loss": 0.4717,
      "step": 241840
    },
    {
      "epoch": 99.0,
      "grad_norm": 2.0027050971984863,
      "learning_rate": 2.5759984559504734e-06,
      "loss": 0.4693,
      "step": 241850
    },
    {
      "epoch": 99.0,
      "eval_loss": 0.482724130153656,
      "eval_runtime": 52.9794,
      "eval_samples_per_second": 65.101,
      "eval_steps_per_second": 8.154,
      "step": 241857
    },
    {
      "epoch": 99.0,
      "grad_norm": 1.6105660200119019,
      "learning_rate": 2.5758406066110204e-06,
      "loss": 0.4804,
      "step": 241860
    },
    {
      "epoch": 99.01,
      "grad_norm": 2.3362879753112793,
      "learning_rate": 2.575682756208396e-06,
      "loss": 0.4876,
      "step": 241870
    },
    {
      "epoch": 99.01,
      "grad_norm": 1.6273146867752075,
      "learning_rate": 2.575524904743324e-06,
      "loss": 0.4906,
      "step": 241880
    },
    {
      "epoch": 99.01,
      "grad_norm": 2.244968891143799,
      "learning_rate": 2.5753670522165265e-06,
      "loss": 0.4544,
      "step": 241890
    },
    {
      "epoch": 99.02,
      "grad_norm": 2.1235601902008057,
      "learning_rate": 2.5752091986287276e-06,
      "loss": 0.4738,
      "step": 241900
    },
    {
      "epoch": 99.02,
      "grad_norm": 1.8883126974105835,
      "learning_rate": 2.5750513439806498e-06,
      "loss": 0.4685,
      "step": 241910
    },
    {
      "epoch": 99.03,
      "grad_norm": 1.8091256618499756,
      "learning_rate": 2.574893488273017e-06,
      "loss": 0.4765,
      "step": 241920
    },
    {
      "epoch": 99.03,
      "grad_norm": 2.093451738357544,
      "learning_rate": 2.574735631506551e-06,
      "loss": 0.4659,
      "step": 241930
    },
    {
      "epoch": 99.03,
      "grad_norm": 2.7511353492736816,
      "learning_rate": 2.574577773681976e-06,
      "loss": 0.4912,
      "step": 241940
    },
    {
      "epoch": 99.04,
      "grad_norm": 2.0128235816955566,
      "learning_rate": 2.574419914800014e-06,
      "loss": 0.4869,
      "step": 241950
    },
    {
      "epoch": 99.04,
      "grad_norm": 1.6760661602020264,
      "learning_rate": 2.5742620548613896e-06,
      "loss": 0.4808,
      "step": 241960
    },
    {
      "epoch": 99.05,
      "grad_norm": 1.9870866537094116,
      "learning_rate": 2.5741041938668246e-06,
      "loss": 0.464,
      "step": 241970
    },
    {
      "epoch": 99.05,
      "grad_norm": 2.0320262908935547,
      "learning_rate": 2.573946331817043e-06,
      "loss": 0.4915,
      "step": 241980
    },
    {
      "epoch": 99.05,
      "grad_norm": 2.330958604812622,
      "learning_rate": 2.573788468712768e-06,
      "loss": 0.493,
      "step": 241990
    },
    {
      "epoch": 99.06,
      "grad_norm": 2.0416054725646973,
      "learning_rate": 2.573630604554722e-06,
      "loss": 0.4801,
      "step": 242000
    },
    {
      "epoch": 99.06,
      "grad_norm": 2.0405709743499756,
      "learning_rate": 2.5734727393436285e-06,
      "loss": 0.4798,
      "step": 242010
    },
    {
      "epoch": 99.07,
      "grad_norm": 2.5846893787384033,
      "learning_rate": 2.5733148730802107e-06,
      "loss": 0.472,
      "step": 242020
    },
    {
      "epoch": 99.07,
      "grad_norm": 1.899117350578308,
      "learning_rate": 2.5731570057651918e-06,
      "loss": 0.4619,
      "step": 242030
    },
    {
      "epoch": 99.07,
      "grad_norm": 2.103278160095215,
      "learning_rate": 2.5729991373992946e-06,
      "loss": 0.4737,
      "step": 242040
    },
    {
      "epoch": 99.08,
      "grad_norm": 1.630185604095459,
      "learning_rate": 2.572841267983243e-06,
      "loss": 0.4661,
      "step": 242050
    },
    {
      "epoch": 99.08,
      "grad_norm": 2.0279505252838135,
      "learning_rate": 2.5726833975177605e-06,
      "loss": 0.4868,
      "step": 242060
    },
    {
      "epoch": 99.09,
      "grad_norm": 2.348804235458374,
      "learning_rate": 2.572525526003569e-06,
      "loss": 0.4835,
      "step": 242070
    },
    {
      "epoch": 99.09,
      "grad_norm": 1.6964082717895508,
      "learning_rate": 2.572367653441392e-06,
      "loss": 0.4943,
      "step": 242080
    },
    {
      "epoch": 99.1,
      "grad_norm": 1.6390026807785034,
      "learning_rate": 2.572209779831953e-06,
      "loss": 0.5086,
      "step": 242090
    },
    {
      "epoch": 99.1,
      "grad_norm": 1.643422245979309,
      "learning_rate": 2.572051905175975e-06,
      "loss": 0.4898,
      "step": 242100
    },
    {
      "epoch": 99.1,
      "grad_norm": 1.8251533508300781,
      "learning_rate": 2.5718940294741815e-06,
      "loss": 0.4648,
      "step": 242110
    },
    {
      "epoch": 99.11,
      "grad_norm": 2.4333841800689697,
      "learning_rate": 2.5717361527272952e-06,
      "loss": 0.4644,
      "step": 242120
    },
    {
      "epoch": 99.11,
      "grad_norm": 1.748488426208496,
      "learning_rate": 2.5715782749360403e-06,
      "loss": 0.4786,
      "step": 242130
    },
    {
      "epoch": 99.12,
      "grad_norm": 2.5961384773254395,
      "learning_rate": 2.571420396101139e-06,
      "loss": 0.4771,
      "step": 242140
    },
    {
      "epoch": 99.12,
      "grad_norm": 1.9615923166275024,
      "learning_rate": 2.571262516223315e-06,
      "loss": 0.4939,
      "step": 242150
    },
    {
      "epoch": 99.12,
      "grad_norm": 1.7341792583465576,
      "learning_rate": 2.5711046353032914e-06,
      "loss": 0.4889,
      "step": 242160
    },
    {
      "epoch": 99.13,
      "grad_norm": 1.7691508531570435,
      "learning_rate": 2.570946753341791e-06,
      "loss": 0.471,
      "step": 242170
    },
    {
      "epoch": 99.13,
      "grad_norm": 1.5145512819290161,
      "learning_rate": 2.570788870339538e-06,
      "loss": 0.4946,
      "step": 242180
    },
    {
      "epoch": 99.14,
      "grad_norm": 2.3761355876922607,
      "learning_rate": 2.5706309862972547e-06,
      "loss": 0.4796,
      "step": 242190
    },
    {
      "epoch": 99.14,
      "grad_norm": 1.9779309034347534,
      "learning_rate": 2.570473101215665e-06,
      "loss": 0.4674,
      "step": 242200
    },
    {
      "epoch": 99.14,
      "grad_norm": 1.7422208786010742,
      "learning_rate": 2.5703152150954924e-06,
      "loss": 0.4647,
      "step": 242210
    },
    {
      "epoch": 99.15,
      "grad_norm": 1.6837804317474365,
      "learning_rate": 2.570157327937459e-06,
      "loss": 0.4782,
      "step": 242220
    },
    {
      "epoch": 99.15,
      "grad_norm": 3.315812349319458,
      "learning_rate": 2.569999439742288e-06,
      "loss": 0.459,
      "step": 242230
    },
    {
      "epoch": 99.16,
      "grad_norm": 1.940682053565979,
      "learning_rate": 2.569841550510704e-06,
      "loss": 0.4843,
      "step": 242240
    },
    {
      "epoch": 99.16,
      "grad_norm": 2.266249656677246,
      "learning_rate": 2.56968366024343e-06,
      "loss": 0.5081,
      "step": 242250
    },
    {
      "epoch": 99.16,
      "grad_norm": 2.177518844604492,
      "learning_rate": 2.5695257689411883e-06,
      "loss": 0.4904,
      "step": 242260
    },
    {
      "epoch": 99.17,
      "grad_norm": 1.7428460121154785,
      "learning_rate": 2.5693678766047034e-06,
      "loss": 0.4615,
      "step": 242270
    },
    {
      "epoch": 99.17,
      "grad_norm": 2.5417189598083496,
      "learning_rate": 2.5692099832346982e-06,
      "loss": 0.4871,
      "step": 242280
    },
    {
      "epoch": 99.18,
      "grad_norm": 1.789344072341919,
      "learning_rate": 2.569052088831895e-06,
      "loss": 0.4619,
      "step": 242290
    },
    {
      "epoch": 99.18,
      "grad_norm": 1.7420942783355713,
      "learning_rate": 2.568894193397018e-06,
      "loss": 0.4906,
      "step": 242300
    },
    {
      "epoch": 99.19,
      "grad_norm": 1.9583115577697754,
      "learning_rate": 2.5687362969307906e-06,
      "loss": 0.4724,
      "step": 242310
    },
    {
      "epoch": 99.19,
      "grad_norm": 1.878334403038025,
      "learning_rate": 2.5685783994339354e-06,
      "loss": 0.492,
      "step": 242320
    },
    {
      "epoch": 99.19,
      "grad_norm": 2.0939524173736572,
      "learning_rate": 2.5684205009071753e-06,
      "loss": 0.4893,
      "step": 242330
    },
    {
      "epoch": 99.2,
      "grad_norm": 1.9419269561767578,
      "learning_rate": 2.568262601351235e-06,
      "loss": 0.5018,
      "step": 242340
    },
    {
      "epoch": 99.2,
      "grad_norm": 1.552506446838379,
      "learning_rate": 2.5681047007668385e-06,
      "loss": 0.4785,
      "step": 242350
    },
    {
      "epoch": 99.21,
      "grad_norm": 1.6274747848510742,
      "learning_rate": 2.5679467991547072e-06,
      "loss": 0.4645,
      "step": 242360
    },
    {
      "epoch": 99.21,
      "grad_norm": 1.8784847259521484,
      "learning_rate": 2.567788896515565e-06,
      "loss": 0.4851,
      "step": 242370
    },
    {
      "epoch": 99.21,
      "grad_norm": 2.540910482406616,
      "learning_rate": 2.5676309928501348e-06,
      "loss": 0.4633,
      "step": 242380
    },
    {
      "epoch": 99.22,
      "grad_norm": 2.016697883605957,
      "learning_rate": 2.5674730881591407e-06,
      "loss": 0.4768,
      "step": 242390
    },
    {
      "epoch": 99.22,
      "grad_norm": 1.7830219268798828,
      "learning_rate": 2.567315182443306e-06,
      "loss": 0.4808,
      "step": 242400
    },
    {
      "epoch": 99.23,
      "grad_norm": 1.7705910205841064,
      "learning_rate": 2.567157275703354e-06,
      "loss": 0.4857,
      "step": 242410
    },
    {
      "epoch": 99.23,
      "grad_norm": 1.770492434501648,
      "learning_rate": 2.5669993679400076e-06,
      "loss": 0.4926,
      "step": 242420
    },
    {
      "epoch": 99.23,
      "grad_norm": 1.7512166500091553,
      "learning_rate": 2.56684145915399e-06,
      "loss": 0.4846,
      "step": 242430
    },
    {
      "epoch": 99.24,
      "grad_norm": 2.2899153232574463,
      "learning_rate": 2.5666835493460255e-06,
      "loss": 0.4868,
      "step": 242440
    },
    {
      "epoch": 99.24,
      "grad_norm": 1.8727014064788818,
      "learning_rate": 2.566525638516837e-06,
      "loss": 0.4778,
      "step": 242450
    },
    {
      "epoch": 99.25,
      "grad_norm": 1.8519867658615112,
      "learning_rate": 2.5663677266671474e-06,
      "loss": 0.4919,
      "step": 242460
    },
    {
      "epoch": 99.25,
      "grad_norm": 2.11397385597229,
      "learning_rate": 2.5662098137976806e-06,
      "loss": 0.4876,
      "step": 242470
    },
    {
      "epoch": 99.26,
      "grad_norm": 1.6345056295394897,
      "learning_rate": 2.5660518999091603e-06,
      "loss": 0.4855,
      "step": 242480
    },
    {
      "epoch": 99.26,
      "grad_norm": 4.352372646331787,
      "learning_rate": 2.5658939850023085e-06,
      "loss": 0.4814,
      "step": 242490
    },
    {
      "epoch": 99.26,
      "grad_norm": 8.4862642288208,
      "learning_rate": 2.5657360690778505e-06,
      "loss": 0.4659,
      "step": 242500
    },
    {
      "epoch": 99.27,
      "grad_norm": 4.992554187774658,
      "learning_rate": 2.5655781521365087e-06,
      "loss": 0.4763,
      "step": 242510
    },
    {
      "epoch": 99.27,
      "grad_norm": 1.8500902652740479,
      "learning_rate": 2.565420234179005e-06,
      "loss": 0.4622,
      "step": 242520
    },
    {
      "epoch": 99.28,
      "grad_norm": 1.9103797674179077,
      "learning_rate": 2.5652623152060658e-06,
      "loss": 0.4642,
      "step": 242530
    },
    {
      "epoch": 99.28,
      "grad_norm": 1.6358164548873901,
      "learning_rate": 2.5651043952184123e-06,
      "loss": 0.479,
      "step": 242540
    },
    {
      "epoch": 99.28,
      "grad_norm": 1.746584177017212,
      "learning_rate": 2.5649464742167693e-06,
      "loss": 0.4716,
      "step": 242550
    },
    {
      "epoch": 99.29,
      "grad_norm": 1.652821660041809,
      "learning_rate": 2.564788552201859e-06,
      "loss": 0.4778,
      "step": 242560
    },
    {
      "epoch": 99.29,
      "grad_norm": 1.9293473958969116,
      "learning_rate": 2.564630629174406e-06,
      "loss": 0.4892,
      "step": 242570
    },
    {
      "epoch": 99.3,
      "grad_norm": 1.792967677116394,
      "learning_rate": 2.564472705135132e-06,
      "loss": 0.4956,
      "step": 242580
    },
    {
      "epoch": 99.3,
      "grad_norm": 1.7866302728652954,
      "learning_rate": 2.5643147800847617e-06,
      "loss": 0.4793,
      "step": 242590
    },
    {
      "epoch": 99.3,
      "grad_norm": 1.6870055198669434,
      "learning_rate": 2.5641568540240184e-06,
      "loss": 0.4852,
      "step": 242600
    },
    {
      "epoch": 99.31,
      "grad_norm": 2.0844712257385254,
      "learning_rate": 2.5639989269536253e-06,
      "loss": 0.4757,
      "step": 242610
    },
    {
      "epoch": 99.31,
      "grad_norm": 1.7947347164154053,
      "learning_rate": 2.5638409988743063e-06,
      "loss": 0.4803,
      "step": 242620
    },
    {
      "epoch": 99.32,
      "grad_norm": 1.6789804697036743,
      "learning_rate": 2.563683069786784e-06,
      "loss": 0.4686,
      "step": 242630
    },
    {
      "epoch": 99.32,
      "grad_norm": 1.8038179874420166,
      "learning_rate": 2.5635251396917836e-06,
      "loss": 0.4664,
      "step": 242640
    },
    {
      "epoch": 99.32,
      "grad_norm": 2.3566715717315674,
      "learning_rate": 2.563367208590026e-06,
      "loss": 0.4683,
      "step": 242650
    },
    {
      "epoch": 99.33,
      "grad_norm": 1.9057272672653198,
      "learning_rate": 2.5632092764822367e-06,
      "loss": 0.4941,
      "step": 242660
    },
    {
      "epoch": 99.33,
      "grad_norm": 2.1983482837677,
      "learning_rate": 2.5630513433691385e-06,
      "loss": 0.4882,
      "step": 242670
    },
    {
      "epoch": 99.34,
      "grad_norm": 2.1405844688415527,
      "learning_rate": 2.5628934092514546e-06,
      "loss": 0.4767,
      "step": 242680
    },
    {
      "epoch": 99.34,
      "grad_norm": 2.03022837638855,
      "learning_rate": 2.5627354741299084e-06,
      "loss": 0.4768,
      "step": 242690
    },
    {
      "epoch": 99.35,
      "grad_norm": 2.0290451049804688,
      "learning_rate": 2.5625775380052235e-06,
      "loss": 0.4815,
      "step": 242700
    },
    {
      "epoch": 99.35,
      "grad_norm": 1.8258156776428223,
      "learning_rate": 2.562419600878124e-06,
      "loss": 0.478,
      "step": 242710
    },
    {
      "epoch": 99.35,
      "grad_norm": 2.097858428955078,
      "learning_rate": 2.562261662749333e-06,
      "loss": 0.4686,
      "step": 242720
    },
    {
      "epoch": 99.36,
      "grad_norm": 2.2315449714660645,
      "learning_rate": 2.5621037236195737e-06,
      "loss": 0.4891,
      "step": 242730
    },
    {
      "epoch": 99.36,
      "grad_norm": 2.625501871109009,
      "learning_rate": 2.56194578348957e-06,
      "loss": 0.4801,
      "step": 242740
    },
    {
      "epoch": 99.37,
      "grad_norm": 1.7263671159744263,
      "learning_rate": 2.561787842360045e-06,
      "loss": 0.4729,
      "step": 242750
    },
    {
      "epoch": 99.37,
      "grad_norm": 2.154200315475464,
      "learning_rate": 2.5616299002317226e-06,
      "loss": 0.493,
      "step": 242760
    },
    {
      "epoch": 99.37,
      "grad_norm": 2.6121633052825928,
      "learning_rate": 2.5614719571053258e-06,
      "loss": 0.4824,
      "step": 242770
    },
    {
      "epoch": 99.38,
      "grad_norm": 1.8099029064178467,
      "learning_rate": 2.5613140129815786e-06,
      "loss": 0.4696,
      "step": 242780
    },
    {
      "epoch": 99.38,
      "grad_norm": 2.5060434341430664,
      "learning_rate": 2.5611560678612044e-06,
      "loss": 0.5141,
      "step": 242790
    },
    {
      "epoch": 99.39,
      "grad_norm": 1.6552960872650146,
      "learning_rate": 2.5609981217449265e-06,
      "loss": 0.4749,
      "step": 242800
    },
    {
      "epoch": 99.39,
      "grad_norm": 2.151272773742676,
      "learning_rate": 2.5608401746334688e-06,
      "loss": 0.5044,
      "step": 242810
    },
    {
      "epoch": 99.39,
      "grad_norm": 2.0235061645507812,
      "learning_rate": 2.5606822265275546e-06,
      "loss": 0.4894,
      "step": 242820
    },
    {
      "epoch": 99.4,
      "grad_norm": 2.1681032180786133,
      "learning_rate": 2.5605242774279077e-06,
      "loss": 0.4761,
      "step": 242830
    },
    {
      "epoch": 99.4,
      "grad_norm": 1.8475849628448486,
      "learning_rate": 2.560366327335251e-06,
      "loss": 0.4818,
      "step": 242840
    },
    {
      "epoch": 99.41,
      "grad_norm": 1.980482816696167,
      "learning_rate": 2.5602083762503084e-06,
      "loss": 0.4794,
      "step": 242850
    },
    {
      "epoch": 99.41,
      "grad_norm": 1.7423373460769653,
      "learning_rate": 2.560050424173804e-06,
      "loss": 0.4697,
      "step": 242860
    },
    {
      "epoch": 99.41,
      "grad_norm": 1.6619900465011597,
      "learning_rate": 2.5598924711064608e-06,
      "loss": 0.4768,
      "step": 242870
    },
    {
      "epoch": 99.42,
      "grad_norm": 2.0201964378356934,
      "learning_rate": 2.559734517049002e-06,
      "loss": 0.4874,
      "step": 242880
    },
    {
      "epoch": 99.42,
      "grad_norm": 1.8076122999191284,
      "learning_rate": 2.5595765620021514e-06,
      "loss": 0.474,
      "step": 242890
    },
    {
      "epoch": 99.43,
      "grad_norm": 3.4226691722869873,
      "learning_rate": 2.5594186059666328e-06,
      "loss": 0.4457,
      "step": 242900
    },
    {
      "epoch": 99.43,
      "grad_norm": 2.414177656173706,
      "learning_rate": 2.55926064894317e-06,
      "loss": 0.4727,
      "step": 242910
    },
    {
      "epoch": 99.44,
      "grad_norm": 1.5850549936294556,
      "learning_rate": 2.5591026909324867e-06,
      "loss": 0.4696,
      "step": 242920
    },
    {
      "epoch": 99.44,
      "grad_norm": 2.1231496334075928,
      "learning_rate": 2.558944731935306e-06,
      "loss": 0.4945,
      "step": 242930
    },
    {
      "epoch": 99.44,
      "grad_norm": 1.6934633255004883,
      "learning_rate": 2.5587867719523513e-06,
      "loss": 0.467,
      "step": 242940
    },
    {
      "epoch": 99.45,
      "grad_norm": 1.7101255655288696,
      "learning_rate": 2.558628810984346e-06,
      "loss": 0.4635,
      "step": 242950
    },
    {
      "epoch": 99.45,
      "grad_norm": 1.7031625509262085,
      "learning_rate": 2.5584708490320146e-06,
      "loss": 0.4891,
      "step": 242960
    },
    {
      "epoch": 99.46,
      "grad_norm": 1.9767855405807495,
      "learning_rate": 2.5583128860960803e-06,
      "loss": 0.4936,
      "step": 242970
    },
    {
      "epoch": 99.46,
      "grad_norm": 2.3500800132751465,
      "learning_rate": 2.5581549221772667e-06,
      "loss": 0.4733,
      "step": 242980
    },
    {
      "epoch": 99.46,
      "grad_norm": 1.8895764350891113,
      "learning_rate": 2.5579969572762964e-06,
      "loss": 0.4778,
      "step": 242990
    },
    {
      "epoch": 99.47,
      "grad_norm": 3.3470635414123535,
      "learning_rate": 2.5578389913938953e-06,
      "loss": 0.4745,
      "step": 243000
    },
    {
      "epoch": 99.47,
      "grad_norm": 1.6393091678619385,
      "learning_rate": 2.5576810245307854e-06,
      "loss": 0.4663,
      "step": 243010
    },
    {
      "epoch": 99.48,
      "grad_norm": 1.8012709617614746,
      "learning_rate": 2.55752305668769e-06,
      "loss": 0.4865,
      "step": 243020
    },
    {
      "epoch": 99.48,
      "grad_norm": 2.010603189468384,
      "learning_rate": 2.557365087865334e-06,
      "loss": 0.4816,
      "step": 243030
    },
    {
      "epoch": 99.48,
      "grad_norm": 1.9400314092636108,
      "learning_rate": 2.5572071180644398e-06,
      "loss": 0.4743,
      "step": 243040
    },
    {
      "epoch": 99.49,
      "grad_norm": 2.1130290031433105,
      "learning_rate": 2.5570491472857317e-06,
      "loss": 0.487,
      "step": 243050
    },
    {
      "epoch": 99.49,
      "grad_norm": 1.7220728397369385,
      "learning_rate": 2.5568911755299333e-06,
      "loss": 0.499,
      "step": 243060
    },
    {
      "epoch": 99.5,
      "grad_norm": 2.161081075668335,
      "learning_rate": 2.5567332027977686e-06,
      "loss": 0.4832,
      "step": 243070
    },
    {
      "epoch": 99.5,
      "grad_norm": 1.6828819513320923,
      "learning_rate": 2.5565752290899604e-06,
      "loss": 0.4562,
      "step": 243080
    },
    {
      "epoch": 99.5,
      "grad_norm": 2.1890759468078613,
      "learning_rate": 2.556417254407233e-06,
      "loss": 0.4959,
      "step": 243090
    },
    {
      "epoch": 99.51,
      "grad_norm": 1.6718122959136963,
      "learning_rate": 2.5562592787503097e-06,
      "loss": 0.4816,
      "step": 243100
    },
    {
      "epoch": 99.51,
      "grad_norm": 1.9445801973342896,
      "learning_rate": 2.5561013021199143e-06,
      "loss": 0.4781,
      "step": 243110
    },
    {
      "epoch": 99.52,
      "grad_norm": 2.0548744201660156,
      "learning_rate": 2.55594332451677e-06,
      "loss": 0.52,
      "step": 243120
    },
    {
      "epoch": 99.52,
      "grad_norm": 1.9152246713638306,
      "learning_rate": 2.555785345941602e-06,
      "loss": 0.4854,
      "step": 243130
    },
    {
      "epoch": 99.53,
      "grad_norm": 1.613927960395813,
      "learning_rate": 2.5556273663951325e-06,
      "loss": 0.4887,
      "step": 243140
    },
    {
      "epoch": 99.53,
      "grad_norm": 2.3645596504211426,
      "learning_rate": 2.555469385878085e-06,
      "loss": 0.4844,
      "step": 243150
    },
    {
      "epoch": 99.53,
      "grad_norm": 1.8263012170791626,
      "learning_rate": 2.5553114043911846e-06,
      "loss": 0.4914,
      "step": 243160
    },
    {
      "epoch": 99.54,
      "grad_norm": 1.3388665914535522,
      "learning_rate": 2.555153421935154e-06,
      "loss": 0.483,
      "step": 243170
    },
    {
      "epoch": 99.54,
      "grad_norm": 1.7159044742584229,
      "learning_rate": 2.5549954385107163e-06,
      "loss": 0.4702,
      "step": 243180
    },
    {
      "epoch": 99.55,
      "grad_norm": 1.9719372987747192,
      "learning_rate": 2.5548374541185967e-06,
      "loss": 0.4691,
      "step": 243190
    },
    {
      "epoch": 99.55,
      "grad_norm": 1.824832558631897,
      "learning_rate": 2.554679468759518e-06,
      "loss": 0.4785,
      "step": 243200
    },
    {
      "epoch": 99.55,
      "grad_norm": 1.699103593826294,
      "learning_rate": 2.5545214824342044e-06,
      "loss": 0.4771,
      "step": 243210
    },
    {
      "epoch": 99.56,
      "grad_norm": 2.421046018600464,
      "learning_rate": 2.554363495143379e-06,
      "loss": 0.4984,
      "step": 243220
    },
    {
      "epoch": 99.56,
      "grad_norm": 2.091630697250366,
      "learning_rate": 2.5542055068877662e-06,
      "loss": 0.4675,
      "step": 243230
    },
    {
      "epoch": 99.57,
      "grad_norm": 2.08898663520813,
      "learning_rate": 2.5540475176680883e-06,
      "loss": 0.4646,
      "step": 243240
    },
    {
      "epoch": 99.57,
      "grad_norm": 2.1246302127838135,
      "learning_rate": 2.553889527485071e-06,
      "loss": 0.4885,
      "step": 243250
    },
    {
      "epoch": 99.57,
      "grad_norm": 1.800744652748108,
      "learning_rate": 2.5537315363394364e-06,
      "loss": 0.4714,
      "step": 243260
    },
    {
      "epoch": 99.58,
      "grad_norm": 2.217953681945801,
      "learning_rate": 2.5535735442319085e-06,
      "loss": 0.4674,
      "step": 243270
    },
    {
      "epoch": 99.58,
      "grad_norm": 1.8197078704833984,
      "learning_rate": 2.5534155511632124e-06,
      "loss": 0.4719,
      "step": 243280
    },
    {
      "epoch": 99.59,
      "grad_norm": 1.5828973054885864,
      "learning_rate": 2.553257557134071e-06,
      "loss": 0.4698,
      "step": 243290
    },
    {
      "epoch": 99.59,
      "grad_norm": 1.8038041591644287,
      "learning_rate": 2.5530995621452078e-06,
      "loss": 0.4824,
      "step": 243300
    },
    {
      "epoch": 99.59,
      "grad_norm": 2.6880297660827637,
      "learning_rate": 2.552941566197346e-06,
      "loss": 0.488,
      "step": 243310
    },
    {
      "epoch": 99.6,
      "grad_norm": 1.962630033493042,
      "learning_rate": 2.5527835692912107e-06,
      "loss": 0.4895,
      "step": 243320
    },
    {
      "epoch": 99.6,
      "grad_norm": 4.7397027015686035,
      "learning_rate": 2.5526255714275247e-06,
      "loss": 0.4637,
      "step": 243330
    },
    {
      "epoch": 99.61,
      "grad_norm": 1.7412649393081665,
      "learning_rate": 2.5524675726070124e-06,
      "loss": 0.475,
      "step": 243340
    },
    {
      "epoch": 99.61,
      "grad_norm": 1.8409606218338013,
      "learning_rate": 2.5523095728303965e-06,
      "loss": 0.4624,
      "step": 243350
    },
    {
      "epoch": 99.62,
      "grad_norm": 1.976162314414978,
      "learning_rate": 2.5521515720984023e-06,
      "loss": 0.4776,
      "step": 243360
    },
    {
      "epoch": 99.62,
      "grad_norm": 1.8695309162139893,
      "learning_rate": 2.5519935704117517e-06,
      "loss": 0.4791,
      "step": 243370
    },
    {
      "epoch": 99.62,
      "grad_norm": 1.7947248220443726,
      "learning_rate": 2.551835567771171e-06,
      "loss": 0.474,
      "step": 243380
    },
    {
      "epoch": 99.63,
      "grad_norm": 1.7307684421539307,
      "learning_rate": 2.5516775641773814e-06,
      "loss": 0.502,
      "step": 243390
    },
    {
      "epoch": 99.63,
      "grad_norm": 1.9097086191177368,
      "learning_rate": 2.551519559631108e-06,
      "loss": 0.4681,
      "step": 243400
    },
    {
      "epoch": 99.64,
      "grad_norm": 1.9146413803100586,
      "learning_rate": 2.5513615541330747e-06,
      "loss": 0.4716,
      "step": 243410
    },
    {
      "epoch": 99.64,
      "grad_norm": 2.050060987472534,
      "learning_rate": 2.551203547684005e-06,
      "loss": 0.4639,
      "step": 243420
    },
    {
      "epoch": 99.64,
      "grad_norm": 1.604371428489685,
      "learning_rate": 2.551045540284623e-06,
      "loss": 0.4795,
      "step": 243430
    },
    {
      "epoch": 99.65,
      "grad_norm": 2.1006743907928467,
      "learning_rate": 2.550887531935652e-06,
      "loss": 0.4997,
      "step": 243440
    },
    {
      "epoch": 99.65,
      "grad_norm": 2.173577308654785,
      "learning_rate": 2.550729522637816e-06,
      "loss": 0.488,
      "step": 243450
    },
    {
      "epoch": 99.66,
      "grad_norm": 2.3478949069976807,
      "learning_rate": 2.550571512391838e-06,
      "loss": 0.4827,
      "step": 243460
    },
    {
      "epoch": 99.66,
      "grad_norm": 2.1216280460357666,
      "learning_rate": 2.5504135011984436e-06,
      "loss": 0.4792,
      "step": 243470
    },
    {
      "epoch": 99.66,
      "grad_norm": 2.0297720432281494,
      "learning_rate": 2.5502554890583557e-06,
      "loss": 0.4691,
      "step": 243480
    },
    {
      "epoch": 99.67,
      "grad_norm": 2.2230992317199707,
      "learning_rate": 2.550097475972298e-06,
      "loss": 0.4765,
      "step": 243490
    },
    {
      "epoch": 99.67,
      "grad_norm": 1.635117769241333,
      "learning_rate": 2.5499394619409946e-06,
      "loss": 0.4745,
      "step": 243500
    },
    {
      "epoch": 99.68,
      "grad_norm": 1.5324705839157104,
      "learning_rate": 2.5497814469651694e-06,
      "loss": 0.4581,
      "step": 243510
    },
    {
      "epoch": 99.68,
      "grad_norm": 2.0252623558044434,
      "learning_rate": 2.549623431045546e-06,
      "loss": 0.4707,
      "step": 243520
    },
    {
      "epoch": 99.68,
      "grad_norm": 1.6401480436325073,
      "learning_rate": 2.5494654141828477e-06,
      "loss": 0.4724,
      "step": 243530
    },
    {
      "epoch": 99.69,
      "grad_norm": 2.071824312210083,
      "learning_rate": 2.5493073963777996e-06,
      "loss": 0.4846,
      "step": 243540
    },
    {
      "epoch": 99.69,
      "grad_norm": 1.8263356685638428,
      "learning_rate": 2.549149377631124e-06,
      "loss": 0.4787,
      "step": 243550
    },
    {
      "epoch": 99.7,
      "grad_norm": 1.8223109245300293,
      "learning_rate": 2.5489913579435464e-06,
      "loss": 0.48,
      "step": 243560
    },
    {
      "epoch": 99.7,
      "grad_norm": 1.9249190092086792,
      "learning_rate": 2.54883333731579e-06,
      "loss": 0.4726,
      "step": 243570
    },
    {
      "epoch": 99.71,
      "grad_norm": 2.3995461463928223,
      "learning_rate": 2.5486753157485787e-06,
      "loss": 0.4781,
      "step": 243580
    },
    {
      "epoch": 99.71,
      "grad_norm": 2.1047182083129883,
      "learning_rate": 2.5485172932426357e-06,
      "loss": 0.4914,
      "step": 243590
    },
    {
      "epoch": 99.71,
      "grad_norm": 1.7784026861190796,
      "learning_rate": 2.548359269798686e-06,
      "loss": 0.4855,
      "step": 243600
    },
    {
      "epoch": 99.72,
      "grad_norm": 1.9757026433944702,
      "learning_rate": 2.5482012454174528e-06,
      "loss": 0.4821,
      "step": 243610
    },
    {
      "epoch": 99.72,
      "grad_norm": 1.7976899147033691,
      "learning_rate": 2.5480432200996594e-06,
      "loss": 0.4716,
      "step": 243620
    },
    {
      "epoch": 99.73,
      "grad_norm": 2.003687858581543,
      "learning_rate": 2.547885193846031e-06,
      "loss": 0.4555,
      "step": 243630
    },
    {
      "epoch": 99.73,
      "grad_norm": 1.7704492807388306,
      "learning_rate": 2.547727166657291e-06,
      "loss": 0.4902,
      "step": 243640
    },
    {
      "epoch": 99.73,
      "grad_norm": 2.622347593307495,
      "learning_rate": 2.547569138534163e-06,
      "loss": 0.4738,
      "step": 243650
    },
    {
      "epoch": 99.74,
      "grad_norm": 1.6811577081680298,
      "learning_rate": 2.547411109477371e-06,
      "loss": 0.5057,
      "step": 243660
    },
    {
      "epoch": 99.74,
      "grad_norm": 1.6365811824798584,
      "learning_rate": 2.547253079487639e-06,
      "loss": 0.4854,
      "step": 243670
    },
    {
      "epoch": 99.75,
      "grad_norm": 1.5923749208450317,
      "learning_rate": 2.547095048565691e-06,
      "loss": 0.4619,
      "step": 243680
    },
    {
      "epoch": 99.75,
      "grad_norm": 1.9103599786758423,
      "learning_rate": 2.5469370167122507e-06,
      "loss": 0.4778,
      "step": 243690
    },
    {
      "epoch": 99.75,
      "grad_norm": 1.6534243822097778,
      "learning_rate": 2.5467789839280423e-06,
      "loss": 0.4828,
      "step": 243700
    },
    {
      "epoch": 99.76,
      "grad_norm": 1.6651685237884521,
      "learning_rate": 2.5466209502137897e-06,
      "loss": 0.4967,
      "step": 243710
    },
    {
      "epoch": 99.76,
      "grad_norm": 1.870233178138733,
      "learning_rate": 2.5464629155702166e-06,
      "loss": 0.4776,
      "step": 243720
    },
    {
      "epoch": 99.77,
      "grad_norm": 2.0039162635803223,
      "learning_rate": 2.5463048799980467e-06,
      "loss": 0.4646,
      "step": 243730
    },
    {
      "epoch": 99.77,
      "grad_norm": 2.2297842502593994,
      "learning_rate": 2.546146843498005e-06,
      "loss": 0.472,
      "step": 243740
    },
    {
      "epoch": 99.77,
      "grad_norm": 1.8516278266906738,
      "learning_rate": 2.545988806070814e-06,
      "loss": 0.4741,
      "step": 243750
    },
    {
      "epoch": 99.78,
      "grad_norm": 1.8165092468261719,
      "learning_rate": 2.545830767717198e-06,
      "loss": 0.4698,
      "step": 243760
    },
    {
      "epoch": 99.78,
      "grad_norm": 2.7558681964874268,
      "learning_rate": 2.5456727284378824e-06,
      "loss": 0.473,
      "step": 243770
    },
    {
      "epoch": 99.79,
      "grad_norm": 1.8619033098220825,
      "learning_rate": 2.5455146882335896e-06,
      "loss": 0.4585,
      "step": 243780
    },
    {
      "epoch": 99.79,
      "grad_norm": 2.8154945373535156,
      "learning_rate": 2.545356647105044e-06,
      "loss": 0.466,
      "step": 243790
    },
    {
      "epoch": 99.8,
      "grad_norm": 1.722017526626587,
      "learning_rate": 2.5451986050529698e-06,
      "loss": 0.4738,
      "step": 243800
    },
    {
      "epoch": 99.8,
      "grad_norm": 1.8591430187225342,
      "learning_rate": 2.5450405620780907e-06,
      "loss": 0.4746,
      "step": 243810
    },
    {
      "epoch": 99.8,
      "grad_norm": 1.625280737876892,
      "learning_rate": 2.5448825181811307e-06,
      "loss": 0.4748,
      "step": 243820
    },
    {
      "epoch": 99.81,
      "grad_norm": 1.708395004272461,
      "learning_rate": 2.544724473362813e-06,
      "loss": 0.4794,
      "step": 243830
    },
    {
      "epoch": 99.81,
      "grad_norm": 1.8013079166412354,
      "learning_rate": 2.5445664276238622e-06,
      "loss": 0.4592,
      "step": 243840
    },
    {
      "epoch": 99.82,
      "grad_norm": 1.8611749410629272,
      "learning_rate": 2.5444083809650035e-06,
      "loss": 0.4791,
      "step": 243850
    },
    {
      "epoch": 99.82,
      "grad_norm": 1.7377276420593262,
      "learning_rate": 2.5442503333869595e-06,
      "loss": 0.4918,
      "step": 243860
    },
    {
      "epoch": 99.82,
      "grad_norm": 1.7467464208602905,
      "learning_rate": 2.544092284890455e-06,
      "loss": 0.4753,
      "step": 243870
    },
    {
      "epoch": 99.83,
      "grad_norm": 1.88163423538208,
      "learning_rate": 2.543934235476213e-06,
      "loss": 0.4805,
      "step": 243880
    },
    {
      "epoch": 99.83,
      "grad_norm": 1.8339048624038696,
      "learning_rate": 2.5437761851449583e-06,
      "loss": 0.4793,
      "step": 243890
    },
    {
      "epoch": 99.84,
      "grad_norm": 2.0701804161071777,
      "learning_rate": 2.5436181338974146e-06,
      "loss": 0.5135,
      "step": 243900
    },
    {
      "epoch": 99.84,
      "grad_norm": 1.9488736391067505,
      "learning_rate": 2.5434600817343054e-06,
      "loss": 0.4923,
      "step": 243910
    },
    {
      "epoch": 99.84,
      "grad_norm": 1.7167104482650757,
      "learning_rate": 2.5433020286563557e-06,
      "loss": 0.4809,
      "step": 243920
    },
    {
      "epoch": 99.85,
      "grad_norm": 2.0081913471221924,
      "learning_rate": 2.543143974664288e-06,
      "loss": 0.4712,
      "step": 243930
    },
    {
      "epoch": 99.85,
      "grad_norm": 2.087146759033203,
      "learning_rate": 2.542985919758829e-06,
      "loss": 0.4872,
      "step": 243940
    },
    {
      "epoch": 99.86,
      "grad_norm": 2.106447219848633,
      "learning_rate": 2.5428278639407e-06,
      "loss": 0.4576,
      "step": 243950
    },
    {
      "epoch": 99.86,
      "grad_norm": 1.8405072689056396,
      "learning_rate": 2.5426698072106267e-06,
      "loss": 0.4944,
      "step": 243960
    },
    {
      "epoch": 99.86,
      "grad_norm": 2.050875425338745,
      "learning_rate": 2.5425117495693327e-06,
      "loss": 0.4781,
      "step": 243970
    },
    {
      "epoch": 99.87,
      "grad_norm": 2.049785614013672,
      "learning_rate": 2.542353691017542e-06,
      "loss": 0.4681,
      "step": 243980
    },
    {
      "epoch": 99.87,
      "grad_norm": 1.8018476963043213,
      "learning_rate": 2.542195631555978e-06,
      "loss": 0.4762,
      "step": 243990
    },
    {
      "epoch": 99.88,
      "grad_norm": 1.6154791116714478,
      "learning_rate": 2.542037571185365e-06,
      "loss": 0.47,
      "step": 244000
    },
    {
      "epoch": 99.88,
      "grad_norm": 1.9665908813476562,
      "learning_rate": 2.5418795099064276e-06,
      "loss": 0.4898,
      "step": 244010
    },
    {
      "epoch": 99.89,
      "grad_norm": 2.4530961513519287,
      "learning_rate": 2.54172144771989e-06,
      "loss": 0.4688,
      "step": 244020
    },
    {
      "epoch": 99.89,
      "grad_norm": 1.9927847385406494,
      "learning_rate": 2.5415633846264754e-06,
      "loss": 0.4769,
      "step": 244030
    },
    {
      "epoch": 99.89,
      "grad_norm": 1.8582197427749634,
      "learning_rate": 2.5414053206269087e-06,
      "loss": 0.4712,
      "step": 244040
    },
    {
      "epoch": 99.9,
      "grad_norm": 1.8326233625411987,
      "learning_rate": 2.541247255721914e-06,
      "loss": 0.4843,
      "step": 244050
    },
    {
      "epoch": 99.9,
      "grad_norm": 2.0276196002960205,
      "learning_rate": 2.5410891899122144e-06,
      "loss": 0.4842,
      "step": 244060
    },
    {
      "epoch": 99.91,
      "grad_norm": 1.8908222913742065,
      "learning_rate": 2.5409311231985344e-06,
      "loss": 0.4797,
      "step": 244070
    },
    {
      "epoch": 99.91,
      "grad_norm": 1.7707667350769043,
      "learning_rate": 2.5407730555815983e-06,
      "loss": 0.5037,
      "step": 244080
    },
    {
      "epoch": 99.91,
      "grad_norm": 1.7269190549850464,
      "learning_rate": 2.54061498706213e-06,
      "loss": 0.4674,
      "step": 244090
    },
    {
      "epoch": 99.92,
      "grad_norm": 2.7358248233795166,
      "learning_rate": 2.5404569176408545e-06,
      "loss": 0.4864,
      "step": 244100
    },
    {
      "epoch": 99.92,
      "grad_norm": 2.306948661804199,
      "learning_rate": 2.5402988473184943e-06,
      "loss": 0.4663,
      "step": 244110
    },
    {
      "epoch": 99.93,
      "grad_norm": 2.158271074295044,
      "learning_rate": 2.5401407760957737e-06,
      "loss": 0.4616,
      "step": 244120
    },
    {
      "epoch": 99.93,
      "grad_norm": 2.2126927375793457,
      "learning_rate": 2.539982703973418e-06,
      "loss": 0.48,
      "step": 244130
    },
    {
      "epoch": 99.93,
      "grad_norm": 2.05922794342041,
      "learning_rate": 2.539824630952151e-06,
      "loss": 0.4667,
      "step": 244140
    },
    {
      "epoch": 99.94,
      "grad_norm": 1.665948748588562,
      "learning_rate": 2.5396665570326965e-06,
      "loss": 0.4908,
      "step": 244150
    },
    {
      "epoch": 99.94,
      "grad_norm": 1.7914555072784424,
      "learning_rate": 2.539508482215778e-06,
      "loss": 0.4725,
      "step": 244160
    },
    {
      "epoch": 99.95,
      "grad_norm": 1.679227352142334,
      "learning_rate": 2.5393504065021212e-06,
      "loss": 0.4788,
      "step": 244170
    },
    {
      "epoch": 99.95,
      "grad_norm": 2.0248804092407227,
      "learning_rate": 2.5391923298924485e-06,
      "loss": 0.498,
      "step": 244180
    },
    {
      "epoch": 99.95,
      "grad_norm": 1.7013986110687256,
      "learning_rate": 2.539034252387485e-06,
      "loss": 0.4905,
      "step": 244190
    },
    {
      "epoch": 99.96,
      "grad_norm": 2.3054261207580566,
      "learning_rate": 2.538876173987954e-06,
      "loss": 0.4867,
      "step": 244200
    },
    {
      "epoch": 99.96,
      "grad_norm": 1.8919585943222046,
      "learning_rate": 2.5387180946945803e-06,
      "loss": 0.4896,
      "step": 244210
    },
    {
      "epoch": 99.97,
      "grad_norm": 2.1236965656280518,
      "learning_rate": 2.538560014508088e-06,
      "loss": 0.4845,
      "step": 244220
    },
    {
      "epoch": 99.97,
      "grad_norm": 2.401301145553589,
      "learning_rate": 2.5384019334292024e-06,
      "loss": 0.496,
      "step": 244230
    },
    {
      "epoch": 99.98,
      "grad_norm": 1.7985455989837646,
      "learning_rate": 2.5382438514586455e-06,
      "loss": 0.4819,
      "step": 244240
    },
    {
      "epoch": 99.98,
      "grad_norm": 1.9683367013931274,
      "learning_rate": 2.538085768597142e-06,
      "loss": 0.4793,
      "step": 244250
    },
    {
      "epoch": 99.98,
      "grad_norm": 1.6209112405776978,
      "learning_rate": 2.537927684845417e-06,
      "loss": 0.4889,
      "step": 244260
    },
    {
      "epoch": 99.99,
      "grad_norm": 1.697989583015442,
      "learning_rate": 2.5377696002041937e-06,
      "loss": 0.4811,
      "step": 244270
    },
    {
      "epoch": 99.99,
      "grad_norm": 1.9073668718338013,
      "learning_rate": 2.5376115146741973e-06,
      "loss": 0.4762,
      "step": 244280
    },
    {
      "epoch": 100.0,
      "grad_norm": 1.801020860671997,
      "learning_rate": 2.5374534282561503e-06,
      "loss": 0.4878,
      "step": 244290
    },
    {
      "epoch": 100.0,
      "grad_norm": 1.9918807744979858,
      "learning_rate": 2.537295340950779e-06,
      "loss": 0.46,
      "step": 244300
    },
    {
      "epoch": 100.0,
      "eval_loss": 0.4785168170928955,
      "eval_runtime": 52.3233,
      "eval_samples_per_second": 65.917,
      "eval_steps_per_second": 8.256,
      "step": 244300
    },
    {
      "epoch": 100.0,
      "grad_norm": 2.35998797416687,
      "learning_rate": 2.537137252758806e-06,
      "loss": 0.4968,
      "step": 244310
    },
    {
      "epoch": 100.01,
      "grad_norm": 1.6423008441925049,
      "learning_rate": 2.5369791636809555e-06,
      "loss": 0.4671,
      "step": 244320
    },
    {
      "epoch": 100.01,
      "grad_norm": 2.272199869155884,
      "learning_rate": 2.536821073717953e-06,
      "loss": 0.4714,
      "step": 244330
    },
    {
      "epoch": 100.02,
      "grad_norm": 2.0834460258483887,
      "learning_rate": 2.536662982870521e-06,
      "loss": 0.5105,
      "step": 244340
    },
    {
      "epoch": 100.02,
      "grad_norm": 1.8103630542755127,
      "learning_rate": 2.536504891139385e-06,
      "loss": 0.4939,
      "step": 244350
    },
    {
      "epoch": 100.02,
      "grad_norm": 2.178250312805176,
      "learning_rate": 2.5363467985252687e-06,
      "loss": 0.4894,
      "step": 244360
    },
    {
      "epoch": 100.03,
      "grad_norm": 1.9264518022537231,
      "learning_rate": 2.536188705028896e-06,
      "loss": 0.4695,
      "step": 244370
    },
    {
      "epoch": 100.03,
      "grad_norm": 1.8428070545196533,
      "learning_rate": 2.5360306106509912e-06,
      "loss": 0.4765,
      "step": 244380
    },
    {
      "epoch": 100.04,
      "grad_norm": 1.8116496801376343,
      "learning_rate": 2.53587251539228e-06,
      "loss": 0.4835,
      "step": 244390
    },
    {
      "epoch": 100.04,
      "grad_norm": 1.7583504915237427,
      "learning_rate": 2.5357144192534835e-06,
      "loss": 0.4723,
      "step": 244400
    },
    {
      "epoch": 100.05,
      "grad_norm": 1.8340922594070435,
      "learning_rate": 2.5355563222353283e-06,
      "loss": 0.4978,
      "step": 244410
    },
    {
      "epoch": 100.05,
      "grad_norm": 1.868282675743103,
      "learning_rate": 2.5353982243385386e-06,
      "loss": 0.466,
      "step": 244420
    },
    {
      "epoch": 100.05,
      "grad_norm": 2.6887574195861816,
      "learning_rate": 2.5352401255638377e-06,
      "loss": 0.4886,
      "step": 244430
    },
    {
      "epoch": 100.06,
      "grad_norm": 1.9086132049560547,
      "learning_rate": 2.5350820259119504e-06,
      "loss": 0.4752,
      "step": 244440
    },
    {
      "epoch": 100.06,
      "grad_norm": 1.506494402885437,
      "learning_rate": 2.5349239253836004e-06,
      "loss": 0.4522,
      "step": 244450
    },
    {
      "epoch": 100.07,
      "grad_norm": 2.257436990737915,
      "learning_rate": 2.534765823979513e-06,
      "loss": 0.4857,
      "step": 244460
    },
    {
      "epoch": 100.07,
      "grad_norm": 2.181795358657837,
      "learning_rate": 2.5346077217004113e-06,
      "loss": 0.4797,
      "step": 244470
    },
    {
      "epoch": 100.07,
      "grad_norm": 2.0489518642425537,
      "learning_rate": 2.5344496185470194e-06,
      "loss": 0.4794,
      "step": 244480
    },
    {
      "epoch": 100.08,
      "grad_norm": 1.6204830408096313,
      "learning_rate": 2.5342915145200625e-06,
      "loss": 0.4926,
      "step": 244490
    },
    {
      "epoch": 100.08,
      "grad_norm": 1.670851469039917,
      "learning_rate": 2.5341334096202644e-06,
      "loss": 0.46,
      "step": 244500
    },
    {
      "epoch": 100.09,
      "grad_norm": 2.4615156650543213,
      "learning_rate": 2.533975303848349e-06,
      "loss": 0.4931,
      "step": 244510
    },
    {
      "epoch": 100.09,
      "grad_norm": 1.8014508485794067,
      "learning_rate": 2.5338171972050415e-06,
      "loss": 0.5064,
      "step": 244520
    },
    {
      "epoch": 100.09,
      "grad_norm": 2.394627332687378,
      "learning_rate": 2.5336590896910655e-06,
      "loss": 0.4894,
      "step": 244530
    },
    {
      "epoch": 100.1,
      "grad_norm": 2.2761833667755127,
      "learning_rate": 2.5335009813071456e-06,
      "loss": 0.4665,
      "step": 244540
    },
    {
      "epoch": 100.1,
      "grad_norm": 2.126539945602417,
      "learning_rate": 2.5333428720540056e-06,
      "loss": 0.4681,
      "step": 244550
    },
    {
      "epoch": 100.11,
      "grad_norm": 1.9458425045013428,
      "learning_rate": 2.5331847619323697e-06,
      "loss": 0.4618,
      "step": 244560
    },
    {
      "epoch": 100.11,
      "grad_norm": 2.103684186935425,
      "learning_rate": 2.533026650942963e-06,
      "loss": 0.4924,
      "step": 244570
    },
    {
      "epoch": 100.11,
      "grad_norm": 2.2046079635620117,
      "learning_rate": 2.532868539086509e-06,
      "loss": 0.5017,
      "step": 244580
    },
    {
      "epoch": 100.12,
      "grad_norm": 1.9946907758712769,
      "learning_rate": 2.5327104263637323e-06,
      "loss": 0.4856,
      "step": 244590
    },
    {
      "epoch": 100.12,
      "grad_norm": 2.3020739555358887,
      "learning_rate": 2.5325523127753574e-06,
      "loss": 0.4844,
      "step": 244600
    },
    {
      "epoch": 100.13,
      "grad_norm": 1.9754879474639893,
      "learning_rate": 2.5323941983221076e-06,
      "loss": 0.4803,
      "step": 244610
    },
    {
      "epoch": 100.13,
      "grad_norm": 2.167665719985962,
      "learning_rate": 2.5322360830047084e-06,
      "loss": 0.4859,
      "step": 244620
    },
    {
      "epoch": 100.14,
      "grad_norm": 2.058021306991577,
      "learning_rate": 2.5320779668238837e-06,
      "loss": 0.4765,
      "step": 244630
    },
    {
      "epoch": 100.14,
      "grad_norm": 2.142756938934326,
      "learning_rate": 2.5319198497803576e-06,
      "loss": 0.4595,
      "step": 244640
    },
    {
      "epoch": 100.14,
      "grad_norm": 1.4458314180374146,
      "learning_rate": 2.531761731874855e-06,
      "loss": 0.474,
      "step": 244650
    },
    {
      "epoch": 100.15,
      "grad_norm": 1.9185985326766968,
      "learning_rate": 2.531603613108099e-06,
      "loss": 0.4885,
      "step": 244660
    },
    {
      "epoch": 100.15,
      "grad_norm": 2.27054762840271,
      "learning_rate": 2.5314454934808155e-06,
      "loss": 0.4965,
      "step": 244670
    },
    {
      "epoch": 100.16,
      "grad_norm": 1.7413971424102783,
      "learning_rate": 2.5312873729937274e-06,
      "loss": 0.4566,
      "step": 244680
    },
    {
      "epoch": 100.16,
      "grad_norm": 1.735660195350647,
      "learning_rate": 2.53112925164756e-06,
      "loss": 0.4953,
      "step": 244690
    },
    {
      "epoch": 100.16,
      "grad_norm": 1.9494521617889404,
      "learning_rate": 2.530971129443037e-06,
      "loss": 0.5021,
      "step": 244700
    },
    {
      "epoch": 100.17,
      "grad_norm": 1.9377270936965942,
      "learning_rate": 2.530813006380883e-06,
      "loss": 0.4624,
      "step": 244710
    },
    {
      "epoch": 100.17,
      "grad_norm": 2.202565908432007,
      "learning_rate": 2.530654882461823e-06,
      "loss": 0.4796,
      "step": 244720
    },
    {
      "epoch": 100.18,
      "grad_norm": 1.8736635446548462,
      "learning_rate": 2.53049675768658e-06,
      "loss": 0.4666,
      "step": 244730
    },
    {
      "epoch": 100.18,
      "grad_norm": 2.051957845687866,
      "learning_rate": 2.5303386320558793e-06,
      "loss": 0.4723,
      "step": 244740
    },
    {
      "epoch": 100.18,
      "grad_norm": 2.135371208190918,
      "learning_rate": 2.5301805055704448e-06,
      "loss": 0.4829,
      "step": 244750
    },
    {
      "epoch": 100.19,
      "grad_norm": 2.1611595153808594,
      "learning_rate": 2.5300223782310015e-06,
      "loss": 0.495,
      "step": 244760
    },
    {
      "epoch": 100.19,
      "grad_norm": 1.603665828704834,
      "learning_rate": 2.5298642500382726e-06,
      "loss": 0.4856,
      "step": 244770
    },
    {
      "epoch": 100.2,
      "grad_norm": 2.106532335281372,
      "learning_rate": 2.529706120992983e-06,
      "loss": 0.4714,
      "step": 244780
    },
    {
      "epoch": 100.2,
      "grad_norm": 2.571941375732422,
      "learning_rate": 2.5295479910958577e-06,
      "loss": 0.4931,
      "step": 244790
    },
    {
      "epoch": 100.2,
      "grad_norm": 2.197688579559326,
      "learning_rate": 2.529389860347621e-06,
      "loss": 0.4847,
      "step": 244800
    },
    {
      "epoch": 100.21,
      "grad_norm": 1.7648108005523682,
      "learning_rate": 2.529231728748996e-06,
      "loss": 0.4941,
      "step": 244810
    },
    {
      "epoch": 100.21,
      "grad_norm": 1.5258196592330933,
      "learning_rate": 2.5290735963007086e-06,
      "loss": 0.4829,
      "step": 244820
    },
    {
      "epoch": 100.22,
      "grad_norm": 2.228714942932129,
      "learning_rate": 2.528915463003482e-06,
      "loss": 0.5003,
      "step": 244830
    },
    {
      "epoch": 100.22,
      "grad_norm": 1.9991360902786255,
      "learning_rate": 2.5287573288580415e-06,
      "loss": 0.4737,
      "step": 244840
    },
    {
      "epoch": 100.23,
      "grad_norm": 1.803665280342102,
      "learning_rate": 2.528599193865111e-06,
      "loss": 0.473,
      "step": 244850
    },
    {
      "epoch": 100.23,
      "grad_norm": 1.9754197597503662,
      "learning_rate": 2.5284410580254147e-06,
      "loss": 0.4825,
      "step": 244860
    },
    {
      "epoch": 100.23,
      "grad_norm": 2.2423810958862305,
      "learning_rate": 2.528282921339677e-06,
      "loss": 0.4589,
      "step": 244870
    },
    {
      "epoch": 100.24,
      "grad_norm": 2.871042251586914,
      "learning_rate": 2.5281247838086236e-06,
      "loss": 0.4923,
      "step": 244880
    },
    {
      "epoch": 100.24,
      "grad_norm": 1.9213014841079712,
      "learning_rate": 2.527966645432977e-06,
      "loss": 0.4967,
      "step": 244890
    },
    {
      "epoch": 100.25,
      "grad_norm": 1.6015859842300415,
      "learning_rate": 2.527808506213463e-06,
      "loss": 0.4631,
      "step": 244900
    },
    {
      "epoch": 100.25,
      "grad_norm": 1.9506242275238037,
      "learning_rate": 2.527650366150805e-06,
      "loss": 0.474,
      "step": 244910
    },
    {
      "epoch": 100.25,
      "grad_norm": 2.2944133281707764,
      "learning_rate": 2.527492225245728e-06,
      "loss": 0.4632,
      "step": 244920
    },
    {
      "epoch": 100.26,
      "grad_norm": 1.5193523168563843,
      "learning_rate": 2.5273340834989566e-06,
      "loss": 0.4714,
      "step": 244930
    },
    {
      "epoch": 100.26,
      "grad_norm": 1.7517352104187012,
      "learning_rate": 2.527175940911215e-06,
      "loss": 0.4636,
      "step": 244940
    },
    {
      "epoch": 100.27,
      "grad_norm": 1.5585957765579224,
      "learning_rate": 2.527017797483227e-06,
      "loss": 0.4691,
      "step": 244950
    },
    {
      "epoch": 100.27,
      "grad_norm": 1.985449194908142,
      "learning_rate": 2.526859653215718e-06,
      "loss": 0.4843,
      "step": 244960
    },
    {
      "epoch": 100.27,
      "grad_norm": 2.021883726119995,
      "learning_rate": 2.526701508109412e-06,
      "loss": 0.4889,
      "step": 244970
    },
    {
      "epoch": 100.28,
      "grad_norm": 2.086562395095825,
      "learning_rate": 2.526543362165033e-06,
      "loss": 0.4805,
      "step": 244980
    },
    {
      "epoch": 100.28,
      "grad_norm": 2.2463839054107666,
      "learning_rate": 2.526385215383307e-06,
      "loss": 0.471,
      "step": 244990
    },
    {
      "epoch": 100.29,
      "grad_norm": 1.9008829593658447,
      "learning_rate": 2.5262270677649564e-06,
      "loss": 0.4825,
      "step": 245000
    },
    {
      "epoch": 100.29,
      "grad_norm": 2.4062979221343994,
      "learning_rate": 2.5260689193107068e-06,
      "loss": 0.4879,
      "step": 245010
    },
    {
      "epoch": 100.29,
      "grad_norm": 1.7022188901901245,
      "learning_rate": 2.5259107700212827e-06,
      "loss": 0.4758,
      "step": 245020
    },
    {
      "epoch": 100.3,
      "grad_norm": 3.1098203659057617,
      "learning_rate": 2.5257526198974083e-06,
      "loss": 0.482,
      "step": 245030
    },
    {
      "epoch": 100.3,
      "grad_norm": 2.1209537982940674,
      "learning_rate": 2.5255944689398087e-06,
      "loss": 0.4984,
      "step": 245040
    },
    {
      "epoch": 100.31,
      "grad_norm": 1.4842561483383179,
      "learning_rate": 2.525436317149207e-06,
      "loss": 0.4712,
      "step": 245050
    },
    {
      "epoch": 100.31,
      "grad_norm": 1.9253202676773071,
      "learning_rate": 2.525278164526328e-06,
      "loss": 0.4951,
      "step": 245060
    },
    {
      "epoch": 100.32,
      "grad_norm": 2.3263466358184814,
      "learning_rate": 2.525120011071897e-06,
      "loss": 0.4856,
      "step": 245070
    },
    {
      "epoch": 100.32,
      "grad_norm": 2.301656484603882,
      "learning_rate": 2.524961856786638e-06,
      "loss": 0.4768,
      "step": 245080
    },
    {
      "epoch": 100.32,
      "grad_norm": 1.8808120489120483,
      "learning_rate": 2.5248037016712754e-06,
      "loss": 0.4726,
      "step": 245090
    },
    {
      "epoch": 100.33,
      "grad_norm": 2.136157989501953,
      "learning_rate": 2.5246455457265343e-06,
      "loss": 0.4653,
      "step": 245100
    },
    {
      "epoch": 100.33,
      "grad_norm": 1.8906867504119873,
      "learning_rate": 2.524487388953139e-06,
      "loss": 0.4699,
      "step": 245110
    },
    {
      "epoch": 100.34,
      "grad_norm": 1.8828634023666382,
      "learning_rate": 2.524329231351814e-06,
      "loss": 0.4611,
      "step": 245120
    },
    {
      "epoch": 100.34,
      "grad_norm": 2.1021974086761475,
      "learning_rate": 2.5241710729232824e-06,
      "loss": 0.4744,
      "step": 245130
    },
    {
      "epoch": 100.34,
      "grad_norm": 2.041517972946167,
      "learning_rate": 2.52401291366827e-06,
      "loss": 0.4777,
      "step": 245140
    },
    {
      "epoch": 100.35,
      "grad_norm": 1.9573042392730713,
      "learning_rate": 2.5238547535875014e-06,
      "loss": 0.479,
      "step": 245150
    },
    {
      "epoch": 100.35,
      "grad_norm": 2.2204699516296387,
      "learning_rate": 2.5236965926817e-06,
      "loss": 0.4755,
      "step": 245160
    },
    {
      "epoch": 100.36,
      "grad_norm": 1.973583698272705,
      "learning_rate": 2.523538430951592e-06,
      "loss": 0.486,
      "step": 245170
    },
    {
      "epoch": 100.36,
      "grad_norm": 1.9279136657714844,
      "learning_rate": 2.523380268397901e-06,
      "loss": 0.4724,
      "step": 245180
    },
    {
      "epoch": 100.36,
      "grad_norm": 1.8215841054916382,
      "learning_rate": 2.523222105021352e-06,
      "loss": 0.5009,
      "step": 245190
    },
    {
      "epoch": 100.37,
      "grad_norm": 1.7282397747039795,
      "learning_rate": 2.5230639408226686e-06,
      "loss": 0.4696,
      "step": 245200
    },
    {
      "epoch": 100.37,
      "grad_norm": 1.6035199165344238,
      "learning_rate": 2.5229057758025755e-06,
      "loss": 0.4843,
      "step": 245210
    },
    {
      "epoch": 100.38,
      "grad_norm": 2.221954822540283,
      "learning_rate": 2.5227476099617977e-06,
      "loss": 0.4775,
      "step": 245220
    },
    {
      "epoch": 100.38,
      "grad_norm": 1.8325718641281128,
      "learning_rate": 2.5225894433010594e-06,
      "loss": 0.4598,
      "step": 245230
    },
    {
      "epoch": 100.38,
      "grad_norm": 1.6496461629867554,
      "learning_rate": 2.5224312758210856e-06,
      "loss": 0.4565,
      "step": 245240
    },
    {
      "epoch": 100.39,
      "grad_norm": 2.2445249557495117,
      "learning_rate": 2.5222731075226e-06,
      "loss": 0.4614,
      "step": 245250
    },
    {
      "epoch": 100.39,
      "grad_norm": 2.4970128536224365,
      "learning_rate": 2.5221149384063286e-06,
      "loss": 0.4834,
      "step": 245260
    },
    {
      "epoch": 100.4,
      "grad_norm": 2.0870473384857178,
      "learning_rate": 2.5219567684729947e-06,
      "loss": 0.4802,
      "step": 245270
    },
    {
      "epoch": 100.4,
      "grad_norm": 2.0070996284484863,
      "learning_rate": 2.521798597723323e-06,
      "loss": 0.4771,
      "step": 245280
    },
    {
      "epoch": 100.41,
      "grad_norm": 2.2126476764678955,
      "learning_rate": 2.521640426158038e-06,
      "loss": 0.486,
      "step": 245290
    },
    {
      "epoch": 100.41,
      "grad_norm": 2.101107358932495,
      "learning_rate": 2.521482253777865e-06,
      "loss": 0.4802,
      "step": 245300
    },
    {
      "epoch": 100.41,
      "grad_norm": 1.777013897895813,
      "learning_rate": 2.5213240805835278e-06,
      "loss": 0.4742,
      "step": 245310
    },
    {
      "epoch": 100.42,
      "grad_norm": 1.942078948020935,
      "learning_rate": 2.5211659065757515e-06,
      "loss": 0.4678,
      "step": 245320
    },
    {
      "epoch": 100.42,
      "grad_norm": 1.9766125679016113,
      "learning_rate": 2.521007731755261e-06,
      "loss": 0.4828,
      "step": 245330
    },
    {
      "epoch": 100.43,
      "grad_norm": 3.10323166847229,
      "learning_rate": 2.520849556122779e-06,
      "loss": 0.4649,
      "step": 245340
    },
    {
      "epoch": 100.43,
      "grad_norm": 1.7364107370376587,
      "learning_rate": 2.5206913796790316e-06,
      "loss": 0.478,
      "step": 245350
    },
    {
      "epoch": 100.43,
      "grad_norm": 2.0407745838165283,
      "learning_rate": 2.5205332024247436e-06,
      "loss": 0.4854,
      "step": 245360
    },
    {
      "epoch": 100.44,
      "grad_norm": 1.9478254318237305,
      "learning_rate": 2.520375024360639e-06,
      "loss": 0.4813,
      "step": 245370
    },
    {
      "epoch": 100.44,
      "grad_norm": 1.9628303050994873,
      "learning_rate": 2.5202168454874425e-06,
      "loss": 0.4708,
      "step": 245380
    },
    {
      "epoch": 100.45,
      "grad_norm": 1.6284505128860474,
      "learning_rate": 2.520058665805879e-06,
      "loss": 0.4835,
      "step": 245390
    },
    {
      "epoch": 100.45,
      "grad_norm": 1.937934398651123,
      "learning_rate": 2.519900485316673e-06,
      "loss": 0.4657,
      "step": 245400
    },
    {
      "epoch": 100.45,
      "grad_norm": 2.0328619480133057,
      "learning_rate": 2.5197423040205485e-06,
      "loss": 0.4608,
      "step": 245410
    },
    {
      "epoch": 100.46,
      "grad_norm": 2.267575740814209,
      "learning_rate": 2.51958412191823e-06,
      "loss": 0.478,
      "step": 245420
    },
    {
      "epoch": 100.46,
      "grad_norm": 2.321187734603882,
      "learning_rate": 2.5194259390104432e-06,
      "loss": 0.4715,
      "step": 245430
    },
    {
      "epoch": 100.47,
      "grad_norm": 1.7103335857391357,
      "learning_rate": 2.5192677552979114e-06,
      "loss": 0.4708,
      "step": 245440
    },
    {
      "epoch": 100.47,
      "grad_norm": 1.7322031259536743,
      "learning_rate": 2.519109570781361e-06,
      "loss": 0.4757,
      "step": 245450
    },
    {
      "epoch": 100.47,
      "grad_norm": 1.8998907804489136,
      "learning_rate": 2.518951385461515e-06,
      "loss": 0.4791,
      "step": 245460
    },
    {
      "epoch": 100.48,
      "grad_norm": 1.8499276638031006,
      "learning_rate": 2.518793199339099e-06,
      "loss": 0.4881,
      "step": 245470
    },
    {
      "epoch": 100.48,
      "grad_norm": 2.02604603767395,
      "learning_rate": 2.5186350124148368e-06,
      "loss": 0.4652,
      "step": 245480
    },
    {
      "epoch": 100.49,
      "grad_norm": 1.9306880235671997,
      "learning_rate": 2.5184768246894534e-06,
      "loss": 0.4652,
      "step": 245490
    },
    {
      "epoch": 100.49,
      "grad_norm": 1.6565409898757935,
      "learning_rate": 2.5183186361636742e-06,
      "loss": 0.4808,
      "step": 245500
    },
    {
      "epoch": 100.5,
      "grad_norm": 1.7816766500473022,
      "learning_rate": 2.5181604468382223e-06,
      "loss": 0.4611,
      "step": 245510
    },
    {
      "epoch": 100.5,
      "grad_norm": 2.350206136703491,
      "learning_rate": 2.518002256713823e-06,
      "loss": 0.492,
      "step": 245520
    },
    {
      "epoch": 100.5,
      "grad_norm": 2.6323447227478027,
      "learning_rate": 2.5178440657912015e-06,
      "loss": 0.4813,
      "step": 245530
    },
    {
      "epoch": 100.51,
      "grad_norm": 1.580635905265808,
      "learning_rate": 2.517685874071082e-06,
      "loss": 0.4861,
      "step": 245540
    },
    {
      "epoch": 100.51,
      "grad_norm": 1.883622407913208,
      "learning_rate": 2.5175276815541895e-06,
      "loss": 0.4843,
      "step": 245550
    },
    {
      "epoch": 100.52,
      "grad_norm": 1.6473443508148193,
      "learning_rate": 2.517369488241248e-06,
      "loss": 0.48,
      "step": 245560
    },
    {
      "epoch": 100.52,
      "grad_norm": 1.8809046745300293,
      "learning_rate": 2.517211294132983e-06,
      "loss": 0.4678,
      "step": 245570
    },
    {
      "epoch": 100.52,
      "grad_norm": 1.9220679998397827,
      "learning_rate": 2.517053099230118e-06,
      "loss": 0.4858,
      "step": 245580
    },
    {
      "epoch": 100.53,
      "grad_norm": 1.741136074066162,
      "learning_rate": 2.5168949035333786e-06,
      "loss": 0.4749,
      "step": 245590
    },
    {
      "epoch": 100.53,
      "grad_norm": 1.5783133506774902,
      "learning_rate": 2.516736707043489e-06,
      "loss": 0.4871,
      "step": 245600
    },
    {
      "epoch": 100.54,
      "grad_norm": 1.7612080574035645,
      "learning_rate": 2.5165785097611748e-06,
      "loss": 0.487,
      "step": 245610
    },
    {
      "epoch": 100.54,
      "grad_norm": 2.48994779586792,
      "learning_rate": 2.5164203116871595e-06,
      "loss": 0.4644,
      "step": 245620
    },
    {
      "epoch": 100.54,
      "grad_norm": 1.6713587045669556,
      "learning_rate": 2.5162621128221676e-06,
      "loss": 0.489,
      "step": 245630
    },
    {
      "epoch": 100.55,
      "grad_norm": 1.7665529251098633,
      "learning_rate": 2.5161039131669253e-06,
      "loss": 0.4585,
      "step": 245640
    },
    {
      "epoch": 100.55,
      "grad_norm": 1.7097301483154297,
      "learning_rate": 2.515945712722156e-06,
      "loss": 0.4718,
      "step": 245650
    },
    {
      "epoch": 100.56,
      "grad_norm": 1.8833342790603638,
      "learning_rate": 2.515787511488585e-06,
      "loss": 0.4966,
      "step": 245660
    },
    {
      "epoch": 100.56,
      "grad_norm": 1.649537444114685,
      "learning_rate": 2.515629309466937e-06,
      "loss": 0.4528,
      "step": 245670
    },
    {
      "epoch": 100.56,
      "grad_norm": 2.549504518508911,
      "learning_rate": 2.515471106657936e-06,
      "loss": 0.4598,
      "step": 245680
    },
    {
      "epoch": 100.57,
      "grad_norm": 1.930509328842163,
      "learning_rate": 2.515312903062308e-06,
      "loss": 0.4916,
      "step": 245690
    },
    {
      "epoch": 100.57,
      "grad_norm": 2.313148021697998,
      "learning_rate": 2.5151546986807768e-06,
      "loss": 0.4774,
      "step": 245700
    },
    {
      "epoch": 100.58,
      "grad_norm": 2.1778042316436768,
      "learning_rate": 2.5149964935140664e-06,
      "loss": 0.4907,
      "step": 245710
    },
    {
      "epoch": 100.58,
      "grad_norm": 1.9777668714523315,
      "learning_rate": 2.5148382875629022e-06,
      "loss": 0.4861,
      "step": 245720
    },
    {
      "epoch": 100.59,
      "grad_norm": 2.1472244262695312,
      "learning_rate": 2.51468008082801e-06,
      "loss": 0.4857,
      "step": 245730
    },
    {
      "epoch": 100.59,
      "grad_norm": 1.7208425998687744,
      "learning_rate": 2.5145218733101126e-06,
      "loss": 0.5034,
      "step": 245740
    },
    {
      "epoch": 100.59,
      "grad_norm": 1.6584265232086182,
      "learning_rate": 2.5143636650099364e-06,
      "loss": 0.4792,
      "step": 245750
    },
    {
      "epoch": 100.6,
      "grad_norm": 2.056642532348633,
      "learning_rate": 2.514205455928206e-06,
      "loss": 0.4906,
      "step": 245760
    },
    {
      "epoch": 100.6,
      "grad_norm": 1.9817354679107666,
      "learning_rate": 2.5140472460656447e-06,
      "loss": 0.4564,
      "step": 245770
    },
    {
      "epoch": 100.61,
      "grad_norm": 2.042083740234375,
      "learning_rate": 2.513889035422978e-06,
      "loss": 0.4766,
      "step": 245780
    },
    {
      "epoch": 100.61,
      "grad_norm": 1.8907257318496704,
      "learning_rate": 2.5137308240009315e-06,
      "loss": 0.4726,
      "step": 245790
    },
    {
      "epoch": 100.61,
      "grad_norm": 1.8581476211547852,
      "learning_rate": 2.5135726118002287e-06,
      "loss": 0.4772,
      "step": 245800
    },
    {
      "epoch": 100.62,
      "grad_norm": 2.0323026180267334,
      "learning_rate": 2.5134143988215943e-06,
      "loss": 0.4674,
      "step": 245810
    },
    {
      "epoch": 100.62,
      "grad_norm": 2.076594829559326,
      "learning_rate": 2.5132561850657534e-06,
      "loss": 0.475,
      "step": 245820
    },
    {
      "epoch": 100.63,
      "grad_norm": 1.6990365982055664,
      "learning_rate": 2.5130979705334323e-06,
      "loss": 0.4736,
      "step": 245830
    },
    {
      "epoch": 100.63,
      "grad_norm": 2.1426680088043213,
      "learning_rate": 2.5129397552253536e-06,
      "loss": 0.4628,
      "step": 245840
    },
    {
      "epoch": 100.63,
      "grad_norm": 1.7989206314086914,
      "learning_rate": 2.5127815391422426e-06,
      "loss": 0.499,
      "step": 245850
    },
    {
      "epoch": 100.64,
      "grad_norm": 2.2857539653778076,
      "learning_rate": 2.5126233222848246e-06,
      "loss": 0.4863,
      "step": 245860
    },
    {
      "epoch": 100.64,
      "grad_norm": 2.1245200634002686,
      "learning_rate": 2.5124651046538244e-06,
      "loss": 0.4823,
      "step": 245870
    },
    {
      "epoch": 100.65,
      "grad_norm": 1.700179934501648,
      "learning_rate": 2.5123068862499656e-06,
      "loss": 0.4688,
      "step": 245880
    },
    {
      "epoch": 100.65,
      "grad_norm": 1.9678361415863037,
      "learning_rate": 2.512148667073974e-06,
      "loss": 0.4771,
      "step": 245890
    },
    {
      "epoch": 100.65,
      "grad_norm": 1.9819436073303223,
      "learning_rate": 2.5119904471265747e-06,
      "loss": 0.476,
      "step": 245900
    },
    {
      "epoch": 100.66,
      "grad_norm": 1.905182957649231,
      "learning_rate": 2.511832226408492e-06,
      "loss": 0.4759,
      "step": 245910
    },
    {
      "epoch": 100.66,
      "grad_norm": 1.884326457977295,
      "learning_rate": 2.51167400492045e-06,
      "loss": 0.4921,
      "step": 245920
    },
    {
      "epoch": 100.67,
      "grad_norm": 2.223907709121704,
      "learning_rate": 2.5115157826631746e-06,
      "loss": 0.4797,
      "step": 245930
    },
    {
      "epoch": 100.67,
      "grad_norm": 1.7364463806152344,
      "learning_rate": 2.51135755963739e-06,
      "loss": 0.5002,
      "step": 245940
    },
    {
      "epoch": 100.68,
      "grad_norm": 2.203404426574707,
      "learning_rate": 2.511199335843821e-06,
      "loss": 0.4675,
      "step": 245950
    },
    {
      "epoch": 100.68,
      "grad_norm": 4.717592716217041,
      "learning_rate": 2.5110411112831927e-06,
      "loss": 0.4699,
      "step": 245960
    },
    {
      "epoch": 100.68,
      "grad_norm": 1.6779588460922241,
      "learning_rate": 2.51088288595623e-06,
      "loss": 0.4653,
      "step": 245970
    },
    {
      "epoch": 100.69,
      "grad_norm": 2.3377201557159424,
      "learning_rate": 2.5107246598636574e-06,
      "loss": 0.4721,
      "step": 245980
    },
    {
      "epoch": 100.69,
      "grad_norm": 2.409586191177368,
      "learning_rate": 2.5105664330061997e-06,
      "loss": 0.4698,
      "step": 245990
    },
    {
      "epoch": 100.7,
      "grad_norm": 2.6437854766845703,
      "learning_rate": 2.5104082053845806e-06,
      "loss": 0.4907,
      "step": 246000
    },
    {
      "epoch": 100.7,
      "grad_norm": 2.083198070526123,
      "learning_rate": 2.510249976999527e-06,
      "loss": 0.4982,
      "step": 246010
    },
    {
      "epoch": 100.7,
      "grad_norm": 1.7781546115875244,
      "learning_rate": 2.510091747851763e-06,
      "loss": 0.4535,
      "step": 246020
    },
    {
      "epoch": 100.71,
      "grad_norm": 2.054172992706299,
      "learning_rate": 2.5099335179420127e-06,
      "loss": 0.4644,
      "step": 246030
    },
    {
      "epoch": 100.71,
      "grad_norm": 1.8450709581375122,
      "learning_rate": 2.5097752872710018e-06,
      "loss": 0.4979,
      "step": 246040
    },
    {
      "epoch": 100.72,
      "grad_norm": 1.8703153133392334,
      "learning_rate": 2.509617055839455e-06,
      "loss": 0.4694,
      "step": 246050
    },
    {
      "epoch": 100.72,
      "grad_norm": 2.397994041442871,
      "learning_rate": 2.5094588236480968e-06,
      "loss": 0.4719,
      "step": 246060
    },
    {
      "epoch": 100.72,
      "grad_norm": 1.8248002529144287,
      "learning_rate": 2.509300590697652e-06,
      "loss": 0.4671,
      "step": 246070
    },
    {
      "epoch": 100.73,
      "grad_norm": 2.0658257007598877,
      "learning_rate": 2.509142356988846e-06,
      "loss": 0.4868,
      "step": 246080
    },
    {
      "epoch": 100.73,
      "grad_norm": 2.1466379165649414,
      "learning_rate": 2.5089841225224027e-06,
      "loss": 0.477,
      "step": 246090
    },
    {
      "epoch": 100.74,
      "grad_norm": 1.816781997680664,
      "learning_rate": 2.5088258872990467e-06,
      "loss": 0.4741,
      "step": 246100
    },
    {
      "epoch": 100.74,
      "grad_norm": 1.9377731084823608,
      "learning_rate": 2.5086676513195045e-06,
      "loss": 0.4799,
      "step": 246110
    },
    {
      "epoch": 100.74,
      "grad_norm": 1.9065090417861938,
      "learning_rate": 2.508509414584501e-06,
      "loss": 0.49,
      "step": 246120
    },
    {
      "epoch": 100.75,
      "grad_norm": 1.81345534324646,
      "learning_rate": 2.5083511770947587e-06,
      "loss": 0.4665,
      "step": 246130
    },
    {
      "epoch": 100.75,
      "grad_norm": 2.146487236022949,
      "learning_rate": 2.5081929388510042e-06,
      "loss": 0.472,
      "step": 246140
    },
    {
      "epoch": 100.76,
      "grad_norm": 1.7053087949752808,
      "learning_rate": 2.5080346998539627e-06,
      "loss": 0.4765,
      "step": 246150
    },
    {
      "epoch": 100.76,
      "grad_norm": 1.8251779079437256,
      "learning_rate": 2.5078764601043577e-06,
      "loss": 0.4703,
      "step": 246160
    },
    {
      "epoch": 100.77,
      "grad_norm": 1.9862653017044067,
      "learning_rate": 2.5077182196029154e-06,
      "loss": 0.481,
      "step": 246170
    },
    {
      "epoch": 100.77,
      "grad_norm": 1.8472363948822021,
      "learning_rate": 2.5075599783503597e-06,
      "loss": 0.4655,
      "step": 246180
    },
    {
      "epoch": 100.77,
      "grad_norm": 1.752484917640686,
      "learning_rate": 2.507401736347416e-06,
      "loss": 0.4703,
      "step": 246190
    },
    {
      "epoch": 100.78,
      "grad_norm": 1.6161521673202515,
      "learning_rate": 2.5072434935948086e-06,
      "loss": 0.4677,
      "step": 246200
    },
    {
      "epoch": 100.78,
      "grad_norm": 1.6845953464508057,
      "learning_rate": 2.507085250093263e-06,
      "loss": 0.4633,
      "step": 246210
    },
    {
      "epoch": 100.79,
      "grad_norm": 1.8002698421478271,
      "learning_rate": 2.5069270058435047e-06,
      "loss": 0.474,
      "step": 246220
    },
    {
      "epoch": 100.79,
      "grad_norm": 1.774309515953064,
      "learning_rate": 2.506768760846257e-06,
      "loss": 0.4718,
      "step": 246230
    },
    {
      "epoch": 100.79,
      "grad_norm": 1.6675621271133423,
      "learning_rate": 2.5066105151022458e-06,
      "loss": 0.4905,
      "step": 246240
    },
    {
      "epoch": 100.8,
      "grad_norm": 1.701901912689209,
      "learning_rate": 2.5064522686121953e-06,
      "loss": 0.4745,
      "step": 246250
    },
    {
      "epoch": 100.8,
      "grad_norm": 1.9945075511932373,
      "learning_rate": 2.506294021376832e-06,
      "loss": 0.4675,
      "step": 246260
    },
    {
      "epoch": 100.81,
      "grad_norm": 1.969266414642334,
      "learning_rate": 2.5061357733968795e-06,
      "loss": 0.501,
      "step": 246270
    },
    {
      "epoch": 100.81,
      "grad_norm": 2.1853418350219727,
      "learning_rate": 2.5059775246730626e-06,
      "loss": 0.4935,
      "step": 246280
    },
    {
      "epoch": 100.81,
      "grad_norm": 1.7037043571472168,
      "learning_rate": 2.505819275206106e-06,
      "loss": 0.4706,
      "step": 246290
    },
    {
      "epoch": 100.82,
      "grad_norm": 2.08674955368042,
      "learning_rate": 2.5056610249967355e-06,
      "loss": 0.4867,
      "step": 246300
    },
    {
      "epoch": 100.82,
      "grad_norm": 2.3240432739257812,
      "learning_rate": 2.505502774045676e-06,
      "loss": 0.4616,
      "step": 246310
    },
    {
      "epoch": 100.83,
      "grad_norm": 2.1352856159210205,
      "learning_rate": 2.505344522353652e-06,
      "loss": 0.4483,
      "step": 246320
    },
    {
      "epoch": 100.83,
      "grad_norm": 1.8439432382583618,
      "learning_rate": 2.505186269921388e-06,
      "loss": 0.4988,
      "step": 246330
    },
    {
      "epoch": 100.84,
      "grad_norm": 1.999618411064148,
      "learning_rate": 2.5050280167496107e-06,
      "loss": 0.4869,
      "step": 246340
    },
    {
      "epoch": 100.84,
      "grad_norm": 1.80039381980896,
      "learning_rate": 2.504869762839043e-06,
      "loss": 0.4636,
      "step": 246350
    },
    {
      "epoch": 100.84,
      "grad_norm": 1.9554965496063232,
      "learning_rate": 2.5047115081904106e-06,
      "loss": 0.4824,
      "step": 246360
    },
    {
      "epoch": 100.85,
      "grad_norm": 1.635842204093933,
      "learning_rate": 2.504553252804438e-06,
      "loss": 0.5053,
      "step": 246370
    },
    {
      "epoch": 100.85,
      "grad_norm": 1.819010853767395,
      "learning_rate": 2.5043949966818507e-06,
      "loss": 0.4759,
      "step": 246380
    },
    {
      "epoch": 100.86,
      "grad_norm": 1.7595175504684448,
      "learning_rate": 2.504236739823374e-06,
      "loss": 0.4721,
      "step": 246390
    },
    {
      "epoch": 100.86,
      "grad_norm": 1.8846665620803833,
      "learning_rate": 2.5040784822297322e-06,
      "loss": 0.4787,
      "step": 246400
    },
    {
      "epoch": 100.86,
      "grad_norm": 1.965067982673645,
      "learning_rate": 2.503920223901651e-06,
      "loss": 0.4841,
      "step": 246410
    },
    {
      "epoch": 100.87,
      "grad_norm": 1.736371397972107,
      "learning_rate": 2.503761964839854e-06,
      "loss": 0.4678,
      "step": 246420
    },
    {
      "epoch": 100.87,
      "grad_norm": 1.7582638263702393,
      "learning_rate": 2.503603705045068e-06,
      "loss": 0.4773,
      "step": 246430
    },
    {
      "epoch": 100.88,
      "grad_norm": 1.8794554471969604,
      "learning_rate": 2.503445444518016e-06,
      "loss": 0.4941,
      "step": 246440
    },
    {
      "epoch": 100.88,
      "grad_norm": 2.0400490760803223,
      "learning_rate": 2.5032871832594238e-06,
      "loss": 0.4692,
      "step": 246450
    },
    {
      "epoch": 100.88,
      "grad_norm": 2.410630702972412,
      "learning_rate": 2.5031289212700167e-06,
      "loss": 0.4537,
      "step": 246460
    },
    {
      "epoch": 100.89,
      "grad_norm": 1.8295117616653442,
      "learning_rate": 2.50297065855052e-06,
      "loss": 0.4837,
      "step": 246470
    },
    {
      "epoch": 100.89,
      "grad_norm": 2.1238629817962646,
      "learning_rate": 2.5028123951016572e-06,
      "loss": 0.4662,
      "step": 246480
    },
    {
      "epoch": 100.9,
      "grad_norm": 2.40974497795105,
      "learning_rate": 2.502654130924155e-06,
      "loss": 0.4814,
      "step": 246490
    },
    {
      "epoch": 100.9,
      "grad_norm": 2.38448166847229,
      "learning_rate": 2.5024958660187374e-06,
      "loss": 0.474,
      "step": 246500
    },
    {
      "epoch": 100.9,
      "grad_norm": 2.082608461380005,
      "learning_rate": 2.5023376003861295e-06,
      "loss": 0.4662,
      "step": 246510
    },
    {
      "epoch": 100.91,
      "grad_norm": 2.2176754474639893,
      "learning_rate": 2.5021793340270565e-06,
      "loss": 0.4797,
      "step": 246520
    },
    {
      "epoch": 100.91,
      "grad_norm": 1.714903473854065,
      "learning_rate": 2.5020210669422427e-06,
      "loss": 0.462,
      "step": 246530
    },
    {
      "epoch": 100.92,
      "grad_norm": 1.7675845623016357,
      "learning_rate": 2.501862799132414e-06,
      "loss": 0.4677,
      "step": 246540
    },
    {
      "epoch": 100.92,
      "grad_norm": 1.9029953479766846,
      "learning_rate": 2.501704530598295e-06,
      "loss": 0.4744,
      "step": 246550
    },
    {
      "epoch": 100.93,
      "grad_norm": 3.120936870574951,
      "learning_rate": 2.5015462613406115e-06,
      "loss": 0.5052,
      "step": 246560
    },
    {
      "epoch": 100.93,
      "grad_norm": 1.820572018623352,
      "learning_rate": 2.5013879913600865e-06,
      "loss": 0.4884,
      "step": 246570
    },
    {
      "epoch": 100.93,
      "grad_norm": 2.1757118701934814,
      "learning_rate": 2.501229720657447e-06,
      "loss": 0.4803,
      "step": 246580
    },
    {
      "epoch": 100.94,
      "grad_norm": 2.7546770572662354,
      "learning_rate": 2.5010714492334175e-06,
      "loss": 0.4934,
      "step": 246590
    },
    {
      "epoch": 100.94,
      "grad_norm": 2.2844865322113037,
      "learning_rate": 2.500913177088722e-06,
      "loss": 0.4843,
      "step": 246600
    },
    {
      "epoch": 100.95,
      "grad_norm": 2.1887102127075195,
      "learning_rate": 2.5007549042240865e-06,
      "loss": 0.4898,
      "step": 246610
    },
    {
      "epoch": 100.95,
      "grad_norm": 1.3634462356567383,
      "learning_rate": 2.5005966306402363e-06,
      "loss": 0.4733,
      "step": 246620
    },
    {
      "epoch": 100.95,
      "grad_norm": 1.7031854391098022,
      "learning_rate": 2.5004383563378963e-06,
      "loss": 0.4763,
      "step": 246630
    },
    {
      "epoch": 100.96,
      "grad_norm": 2.32655930519104,
      "learning_rate": 2.5002800813177907e-06,
      "loss": 0.4646,
      "step": 246640
    },
    {
      "epoch": 100.96,
      "grad_norm": 1.4705315828323364,
      "learning_rate": 2.500121805580645e-06,
      "loss": 0.4841,
      "step": 246650
    },
    {
      "epoch": 100.97,
      "grad_norm": 1.9904812574386597,
      "learning_rate": 2.499963529127185e-06,
      "loss": 0.4885,
      "step": 246660
    },
    {
      "epoch": 100.97,
      "grad_norm": 1.643922209739685,
      "learning_rate": 2.4998052519581335e-06,
      "loss": 0.4636,
      "step": 246670
    },
    {
      "epoch": 100.97,
      "grad_norm": 1.5878856182098389,
      "learning_rate": 2.499646974074218e-06,
      "loss": 0.4682,
      "step": 246680
    },
    {
      "epoch": 100.98,
      "grad_norm": 1.9779143333435059,
      "learning_rate": 2.4994886954761622e-06,
      "loss": 0.479,
      "step": 246690
    },
    {
      "epoch": 100.98,
      "grad_norm": 1.9263066053390503,
      "learning_rate": 2.499330416164692e-06,
      "loss": 0.4946,
      "step": 246700
    },
    {
      "epoch": 100.99,
      "grad_norm": 1.9872136116027832,
      "learning_rate": 2.4991721361405324e-06,
      "loss": 0.4992,
      "step": 246710
    },
    {
      "epoch": 100.99,
      "grad_norm": 2.0997018814086914,
      "learning_rate": 2.4990138554044075e-06,
      "loss": 0.4757,
      "step": 246720
    },
    {
      "epoch": 100.99,
      "grad_norm": 1.8850353956222534,
      "learning_rate": 2.498855573957043e-06,
      "loss": 0.4924,
      "step": 246730
    },
    {
      "epoch": 101.0,
      "grad_norm": 1.8848614692687988,
      "learning_rate": 2.4986972917991636e-06,
      "loss": 0.479,
      "step": 246740
    },
    {
      "epoch": 101.0,
      "eval_loss": 0.47921693325042725,
      "eval_runtime": 60.5272,
      "eval_samples_per_second": 56.983,
      "eval_steps_per_second": 7.137,
      "step": 246743
    },
    {
      "epoch": 101.0,
      "grad_norm": 1.9411004781723022,
      "learning_rate": 2.498539008931495e-06,
      "loss": 0.4914,
      "step": 246750
    },
    {
      "epoch": 101.01,
      "grad_norm": 2.3114500045776367,
      "learning_rate": 2.498380725354761e-06,
      "loss": 0.4742,
      "step": 246760
    },
    {
      "epoch": 101.01,
      "grad_norm": 1.671054720878601,
      "learning_rate": 2.4982224410696897e-06,
      "loss": 0.4746,
      "step": 246770
    },
    {
      "epoch": 101.02,
      "grad_norm": 1.910792589187622,
      "learning_rate": 2.4980641560770028e-06,
      "loss": 0.4748,
      "step": 246780
    },
    {
      "epoch": 101.02,
      "grad_norm": 2.0550286769866943,
      "learning_rate": 2.497905870377426e-06,
      "loss": 0.4595,
      "step": 246790
    },
    {
      "epoch": 101.02,
      "grad_norm": 1.8404109477996826,
      "learning_rate": 2.4977475839716863e-06,
      "loss": 0.4892,
      "step": 246800
    },
    {
      "epoch": 101.03,
      "grad_norm": 2.066859483718872,
      "learning_rate": 2.4975892968605064e-06,
      "loss": 0.5043,
      "step": 246810
    },
    {
      "epoch": 101.03,
      "grad_norm": 1.8995954990386963,
      "learning_rate": 2.4974310090446134e-06,
      "loss": 0.4783,
      "step": 246820
    },
    {
      "epoch": 101.04,
      "grad_norm": 2.0126171112060547,
      "learning_rate": 2.497272720524731e-06,
      "loss": 0.4857,
      "step": 246830
    },
    {
      "epoch": 101.04,
      "grad_norm": 1.6635489463806152,
      "learning_rate": 2.497114431301585e-06,
      "loss": 0.4643,
      "step": 246840
    },
    {
      "epoch": 101.04,
      "grad_norm": 1.8747225999832153,
      "learning_rate": 2.4969561413759e-06,
      "loss": 0.4748,
      "step": 246850
    },
    {
      "epoch": 101.05,
      "grad_norm": 1.8656283617019653,
      "learning_rate": 2.4967978507484015e-06,
      "loss": 0.4977,
      "step": 246860
    },
    {
      "epoch": 101.05,
      "grad_norm": 1.6455589532852173,
      "learning_rate": 2.4966395594198145e-06,
      "loss": 0.4822,
      "step": 246870
    },
    {
      "epoch": 101.06,
      "grad_norm": 1.975117802619934,
      "learning_rate": 2.4964812673908647e-06,
      "loss": 0.4815,
      "step": 246880
    },
    {
      "epoch": 101.06,
      "grad_norm": 2.036409378051758,
      "learning_rate": 2.4963229746622754e-06,
      "loss": 0.4803,
      "step": 246890
    },
    {
      "epoch": 101.06,
      "grad_norm": 1.9155406951904297,
      "learning_rate": 2.496164681234774e-06,
      "loss": 0.4971,
      "step": 246900
    },
    {
      "epoch": 101.07,
      "grad_norm": 2.1641061305999756,
      "learning_rate": 2.4960063871090838e-06,
      "loss": 0.4654,
      "step": 246910
    },
    {
      "epoch": 101.07,
      "grad_norm": 2.201451301574707,
      "learning_rate": 2.495848092285932e-06,
      "loss": 0.4647,
      "step": 246920
    },
    {
      "epoch": 101.08,
      "grad_norm": 1.7869572639465332,
      "learning_rate": 2.495689796766041e-06,
      "loss": 0.4895,
      "step": 246930
    },
    {
      "epoch": 101.08,
      "grad_norm": 1.9176750183105469,
      "learning_rate": 2.495531500550138e-06,
      "loss": 0.4569,
      "step": 246940
    },
    {
      "epoch": 101.08,
      "grad_norm": 2.1714060306549072,
      "learning_rate": 2.4953732036389463e-06,
      "loss": 0.4721,
      "step": 246950
    },
    {
      "epoch": 101.09,
      "grad_norm": 1.7885701656341553,
      "learning_rate": 2.4952149060331934e-06,
      "loss": 0.4775,
      "step": 246960
    },
    {
      "epoch": 101.09,
      "grad_norm": 1.9825685024261475,
      "learning_rate": 2.495056607733603e-06,
      "loss": 0.4567,
      "step": 246970
    },
    {
      "epoch": 101.1,
      "grad_norm": 3.330096960067749,
      "learning_rate": 2.4948983087409004e-06,
      "loss": 0.4719,
      "step": 246980
    },
    {
      "epoch": 101.1,
      "grad_norm": 2.219358444213867,
      "learning_rate": 2.494740009055811e-06,
      "loss": 0.4865,
      "step": 246990
    },
    {
      "epoch": 101.11,
      "grad_norm": 2.00217342376709,
      "learning_rate": 2.49458170867906e-06,
      "loss": 0.4853,
      "step": 247000
    },
    {
      "epoch": 101.11,
      "grad_norm": 1.9897950887680054,
      "learning_rate": 2.494423407611372e-06,
      "loss": 0.4788,
      "step": 247010
    },
    {
      "epoch": 101.11,
      "grad_norm": 1.7221882343292236,
      "learning_rate": 2.4942651058534716e-06,
      "loss": 0.4866,
      "step": 247020
    },
    {
      "epoch": 101.12,
      "grad_norm": 2.434610366821289,
      "learning_rate": 2.494106803406086e-06,
      "loss": 0.4955,
      "step": 247030
    },
    {
      "epoch": 101.12,
      "grad_norm": 1.8234245777130127,
      "learning_rate": 2.493948500269938e-06,
      "loss": 0.484,
      "step": 247040
    },
    {
      "epoch": 101.13,
      "grad_norm": 1.9464040994644165,
      "learning_rate": 2.4937901964457547e-06,
      "loss": 0.4734,
      "step": 247050
    },
    {
      "epoch": 101.13,
      "grad_norm": 2.027451515197754,
      "learning_rate": 2.4936318919342607e-06,
      "loss": 0.4739,
      "step": 247060
    },
    {
      "epoch": 101.13,
      "grad_norm": 1.5463190078735352,
      "learning_rate": 2.493473586736181e-06,
      "loss": 0.4752,
      "step": 247070
    },
    {
      "epoch": 101.14,
      "grad_norm": 2.1753621101379395,
      "learning_rate": 2.4933152808522404e-06,
      "loss": 0.4743,
      "step": 247080
    },
    {
      "epoch": 101.14,
      "grad_norm": 2.1950182914733887,
      "learning_rate": 2.4931569742831645e-06,
      "loss": 0.4733,
      "step": 247090
    },
    {
      "epoch": 101.15,
      "grad_norm": 1.8601124286651611,
      "learning_rate": 2.492998667029678e-06,
      "loss": 0.4567,
      "step": 247100
    },
    {
      "epoch": 101.15,
      "grad_norm": 1.917860507965088,
      "learning_rate": 2.492840359092507e-06,
      "loss": 0.4614,
      "step": 247110
    },
    {
      "epoch": 101.15,
      "grad_norm": 1.9143325090408325,
      "learning_rate": 2.4926820504723756e-06,
      "loss": 0.4738,
      "step": 247120
    },
    {
      "epoch": 101.16,
      "grad_norm": 1.7460757493972778,
      "learning_rate": 2.49252374117001e-06,
      "loss": 0.4804,
      "step": 247130
    },
    {
      "epoch": 101.16,
      "grad_norm": 1.941692590713501,
      "learning_rate": 2.492365431186135e-06,
      "loss": 0.4706,
      "step": 247140
    },
    {
      "epoch": 101.17,
      "grad_norm": 3.1238913536071777,
      "learning_rate": 2.4922071205214754e-06,
      "loss": 0.4936,
      "step": 247150
    },
    {
      "epoch": 101.17,
      "grad_norm": 1.4915891885757446,
      "learning_rate": 2.492048809176757e-06,
      "loss": 0.4957,
      "step": 247160
    },
    {
      "epoch": 101.17,
      "grad_norm": 1.8249720335006714,
      "learning_rate": 2.4918904971527043e-06,
      "loss": 0.4886,
      "step": 247170
    },
    {
      "epoch": 101.18,
      "grad_norm": 6.184079647064209,
      "learning_rate": 2.491732184450043e-06,
      "loss": 0.4584,
      "step": 247180
    },
    {
      "epoch": 101.18,
      "grad_norm": 2.1811699867248535,
      "learning_rate": 2.491573871069499e-06,
      "loss": 0.4772,
      "step": 247190
    },
    {
      "epoch": 101.19,
      "grad_norm": 1.682705283164978,
      "learning_rate": 2.4914155570117958e-06,
      "loss": 0.4757,
      "step": 247200
    },
    {
      "epoch": 101.19,
      "grad_norm": 2.553286075592041,
      "learning_rate": 2.4912572422776607e-06,
      "loss": 0.4647,
      "step": 247210
    },
    {
      "epoch": 101.2,
      "grad_norm": 1.5335990190505981,
      "learning_rate": 2.491098926867817e-06,
      "loss": 0.4789,
      "step": 247220
    },
    {
      "epoch": 101.2,
      "grad_norm": 1.9576512575149536,
      "learning_rate": 2.4909406107829904e-06,
      "loss": 0.4786,
      "step": 247230
    },
    {
      "epoch": 101.2,
      "grad_norm": 2.2621445655822754,
      "learning_rate": 2.4907822940239065e-06,
      "loss": 0.4879,
      "step": 247240
    },
    {
      "epoch": 101.21,
      "grad_norm": 2.2424302101135254,
      "learning_rate": 2.4906239765912905e-06,
      "loss": 0.492,
      "step": 247250
    },
    {
      "epoch": 101.21,
      "grad_norm": 2.0493953227996826,
      "learning_rate": 2.4904656584858677e-06,
      "loss": 0.4865,
      "step": 247260
    },
    {
      "epoch": 101.22,
      "grad_norm": 1.8592784404754639,
      "learning_rate": 2.4903073397083632e-06,
      "loss": 0.4781,
      "step": 247270
    },
    {
      "epoch": 101.22,
      "grad_norm": 1.79185950756073,
      "learning_rate": 2.490149020259503e-06,
      "loss": 0.497,
      "step": 247280
    },
    {
      "epoch": 101.22,
      "grad_norm": 1.878890037536621,
      "learning_rate": 2.4899907001400103e-06,
      "loss": 0.464,
      "step": 247290
    },
    {
      "epoch": 101.23,
      "grad_norm": 1.8978067636489868,
      "learning_rate": 2.489832379350612e-06,
      "loss": 0.4799,
      "step": 247300
    },
    {
      "epoch": 101.23,
      "grad_norm": 1.9044150114059448,
      "learning_rate": 2.489674057892033e-06,
      "loss": 0.4928,
      "step": 247310
    },
    {
      "epoch": 101.24,
      "grad_norm": 1.906948208808899,
      "learning_rate": 2.4895157357649983e-06,
      "loss": 0.4831,
      "step": 247320
    },
    {
      "epoch": 101.24,
      "grad_norm": 1.6562217473983765,
      "learning_rate": 2.4893574129702334e-06,
      "loss": 0.4727,
      "step": 247330
    },
    {
      "epoch": 101.24,
      "grad_norm": 10.466073036193848,
      "learning_rate": 2.4891990895084635e-06,
      "loss": 0.4838,
      "step": 247340
    },
    {
      "epoch": 101.25,
      "grad_norm": 1.971387505531311,
      "learning_rate": 2.4890407653804145e-06,
      "loss": 0.4767,
      "step": 247350
    },
    {
      "epoch": 101.25,
      "grad_norm": 1.651121973991394,
      "learning_rate": 2.4888824405868103e-06,
      "loss": 0.484,
      "step": 247360
    },
    {
      "epoch": 101.26,
      "grad_norm": 1.7654951810836792,
      "learning_rate": 2.488724115128377e-06,
      "loss": 0.4892,
      "step": 247370
    },
    {
      "epoch": 101.26,
      "grad_norm": 2.3168704509735107,
      "learning_rate": 2.48856578900584e-06,
      "loss": 0.4591,
      "step": 247380
    },
    {
      "epoch": 101.26,
      "grad_norm": 1.8425557613372803,
      "learning_rate": 2.4884074622199244e-06,
      "loss": 0.4814,
      "step": 247390
    },
    {
      "epoch": 101.27,
      "grad_norm": 2.337444543838501,
      "learning_rate": 2.4882491347713547e-06,
      "loss": 0.4885,
      "step": 247400
    },
    {
      "epoch": 101.27,
      "grad_norm": 1.9743003845214844,
      "learning_rate": 2.488090806660857e-06,
      "loss": 0.4889,
      "step": 247410
    },
    {
      "epoch": 101.28,
      "grad_norm": 1.73164701461792,
      "learning_rate": 2.487932477889157e-06,
      "loss": 0.4878,
      "step": 247420
    },
    {
      "epoch": 101.28,
      "grad_norm": 1.8562382459640503,
      "learning_rate": 2.487774148456979e-06,
      "loss": 0.4922,
      "step": 247430
    },
    {
      "epoch": 101.29,
      "grad_norm": 1.768746018409729,
      "learning_rate": 2.4876158183650484e-06,
      "loss": 0.4711,
      "step": 247440
    },
    {
      "epoch": 101.29,
      "grad_norm": 1.8635752201080322,
      "learning_rate": 2.4874574876140912e-06,
      "loss": 0.4855,
      "step": 247450
    },
    {
      "epoch": 101.29,
      "grad_norm": 1.7739979028701782,
      "learning_rate": 2.4872991562048323e-06,
      "loss": 0.4694,
      "step": 247460
    },
    {
      "epoch": 101.3,
      "grad_norm": 1.9294711351394653,
      "learning_rate": 2.4871408241379966e-06,
      "loss": 0.4739,
      "step": 247470
    },
    {
      "epoch": 101.3,
      "grad_norm": 1.9554604291915894,
      "learning_rate": 2.48698249141431e-06,
      "loss": 0.478,
      "step": 247480
    },
    {
      "epoch": 101.31,
      "grad_norm": 2.1523520946502686,
      "learning_rate": 2.4868241580344973e-06,
      "loss": 0.4778,
      "step": 247490
    },
    {
      "epoch": 101.31,
      "grad_norm": 2.382505416870117,
      "learning_rate": 2.4866658239992842e-06,
      "loss": 0.4663,
      "step": 247500
    },
    {
      "epoch": 101.31,
      "grad_norm": 1.866737961769104,
      "learning_rate": 2.486507489309396e-06,
      "loss": 0.4673,
      "step": 247510
    },
    {
      "epoch": 101.32,
      "grad_norm": 2.840686082839966,
      "learning_rate": 2.486349153965557e-06,
      "loss": 0.4789,
      "step": 247520
    },
    {
      "epoch": 101.32,
      "grad_norm": 2.019835948944092,
      "learning_rate": 2.4861908179684946e-06,
      "loss": 0.476,
      "step": 247530
    },
    {
      "epoch": 101.33,
      "grad_norm": 2.181533098220825,
      "learning_rate": 2.486032481318932e-06,
      "loss": 0.4677,
      "step": 247540
    },
    {
      "epoch": 101.33,
      "grad_norm": 1.656097650527954,
      "learning_rate": 2.4858741440175954e-06,
      "loss": 0.4682,
      "step": 247550
    },
    {
      "epoch": 101.33,
      "grad_norm": 2.3369905948638916,
      "learning_rate": 2.4857158060652107e-06,
      "loss": 0.479,
      "step": 247560
    },
    {
      "epoch": 101.34,
      "grad_norm": 2.1908059120178223,
      "learning_rate": 2.485557467462503e-06,
      "loss": 0.4611,
      "step": 247570
    },
    {
      "epoch": 101.34,
      "grad_norm": 1.898417353630066,
      "learning_rate": 2.4853991282101965e-06,
      "loss": 0.4617,
      "step": 247580
    },
    {
      "epoch": 101.35,
      "grad_norm": 1.7982107400894165,
      "learning_rate": 2.4852407883090176e-06,
      "loss": 0.4623,
      "step": 247590
    },
    {
      "epoch": 101.35,
      "grad_norm": 1.7636234760284424,
      "learning_rate": 2.485082447759691e-06,
      "loss": 0.4828,
      "step": 247600
    },
    {
      "epoch": 101.35,
      "grad_norm": 1.6608558893203735,
      "learning_rate": 2.4849241065629416e-06,
      "loss": 0.4656,
      "step": 247610
    },
    {
      "epoch": 101.36,
      "grad_norm": 1.706797480583191,
      "learning_rate": 2.4847657647194963e-06,
      "loss": 0.4662,
      "step": 247620
    },
    {
      "epoch": 101.36,
      "grad_norm": 1.6217527389526367,
      "learning_rate": 2.4846074222300795e-06,
      "loss": 0.495,
      "step": 247630
    },
    {
      "epoch": 101.37,
      "grad_norm": 1.8622385263442993,
      "learning_rate": 2.484449079095417e-06,
      "loss": 0.4584,
      "step": 247640
    },
    {
      "epoch": 101.37,
      "grad_norm": 1.932188630104065,
      "learning_rate": 2.484290735316234e-06,
      "loss": 0.5029,
      "step": 247650
    },
    {
      "epoch": 101.38,
      "grad_norm": 2.393446445465088,
      "learning_rate": 2.484132390893255e-06,
      "loss": 0.4822,
      "step": 247660
    },
    {
      "epoch": 101.38,
      "grad_norm": 2.533108711242676,
      "learning_rate": 2.4839740458272065e-06,
      "loss": 0.4651,
      "step": 247670
    },
    {
      "epoch": 101.38,
      "grad_norm": 1.8007732629776,
      "learning_rate": 2.483815700118813e-06,
      "loss": 0.4569,
      "step": 247680
    },
    {
      "epoch": 101.39,
      "grad_norm": 1.53935706615448,
      "learning_rate": 2.4836573537688004e-06,
      "loss": 0.4694,
      "step": 247690
    },
    {
      "epoch": 101.39,
      "grad_norm": 1.8947049379348755,
      "learning_rate": 2.483499006777893e-06,
      "loss": 0.4642,
      "step": 247700
    },
    {
      "epoch": 101.4,
      "grad_norm": 2.5593841075897217,
      "learning_rate": 2.4833406591468186e-06,
      "loss": 0.4974,
      "step": 247710
    },
    {
      "epoch": 101.4,
      "grad_norm": 1.47588312625885,
      "learning_rate": 2.4831823108763003e-06,
      "loss": 0.4772,
      "step": 247720
    },
    {
      "epoch": 101.4,
      "grad_norm": 2.5446953773498535,
      "learning_rate": 2.483023961967064e-06,
      "loss": 0.4907,
      "step": 247730
    },
    {
      "epoch": 101.41,
      "grad_norm": 1.5965708494186401,
      "learning_rate": 2.482865612419836e-06,
      "loss": 0.4861,
      "step": 247740
    },
    {
      "epoch": 101.41,
      "grad_norm": 2.0152294635772705,
      "learning_rate": 2.4827072622353395e-06,
      "loss": 0.493,
      "step": 247750
    },
    {
      "epoch": 101.42,
      "grad_norm": 1.8828555345535278,
      "learning_rate": 2.482548911414303e-06,
      "loss": 0.4625,
      "step": 247760
    },
    {
      "epoch": 101.42,
      "grad_norm": 1.5490310192108154,
      "learning_rate": 2.482390559957449e-06,
      "loss": 0.4606,
      "step": 247770
    },
    {
      "epoch": 101.42,
      "grad_norm": 1.7133171558380127,
      "learning_rate": 2.482232207865504e-06,
      "loss": 0.4719,
      "step": 247780
    },
    {
      "epoch": 101.43,
      "grad_norm": 2.357454538345337,
      "learning_rate": 2.482073855139194e-06,
      "loss": 0.4604,
      "step": 247790
    },
    {
      "epoch": 101.43,
      "grad_norm": 1.607318639755249,
      "learning_rate": 2.4819155017792438e-06,
      "loss": 0.453,
      "step": 247800
    },
    {
      "epoch": 101.44,
      "grad_norm": 2.141392469406128,
      "learning_rate": 2.4817571477863784e-06,
      "loss": 0.4747,
      "step": 247810
    },
    {
      "epoch": 101.44,
      "grad_norm": 2.282888650894165,
      "learning_rate": 2.481598793161324e-06,
      "loss": 0.4828,
      "step": 247820
    },
    {
      "epoch": 101.44,
      "grad_norm": 1.8924564123153687,
      "learning_rate": 2.4814404379048054e-06,
      "loss": 0.4801,
      "step": 247830
    },
    {
      "epoch": 101.45,
      "grad_norm": 1.7971255779266357,
      "learning_rate": 2.4812820820175487e-06,
      "loss": 0.4542,
      "step": 247840
    },
    {
      "epoch": 101.45,
      "grad_norm": 1.694364309310913,
      "learning_rate": 2.4811237255002787e-06,
      "loss": 0.4797,
      "step": 247850
    },
    {
      "epoch": 101.46,
      "grad_norm": 1.7604680061340332,
      "learning_rate": 2.4809653683537207e-06,
      "loss": 0.4783,
      "step": 247860
    },
    {
      "epoch": 101.46,
      "grad_norm": 2.5620992183685303,
      "learning_rate": 2.480807010578601e-06,
      "loss": 0.4801,
      "step": 247870
    },
    {
      "epoch": 101.47,
      "grad_norm": 2.067014455795288,
      "learning_rate": 2.480648652175643e-06,
      "loss": 0.4865,
      "step": 247880
    },
    {
      "epoch": 101.47,
      "grad_norm": 1.8320046663284302,
      "learning_rate": 2.480490293145574e-06,
      "loss": 0.4989,
      "step": 247890
    },
    {
      "epoch": 101.47,
      "grad_norm": 1.7394578456878662,
      "learning_rate": 2.4803319334891195e-06,
      "loss": 0.4869,
      "step": 247900
    },
    {
      "epoch": 101.48,
      "grad_norm": 1.8730275630950928,
      "learning_rate": 2.4801735732070036e-06,
      "loss": 0.4764,
      "step": 247910
    },
    {
      "epoch": 101.48,
      "grad_norm": 1.947486400604248,
      "learning_rate": 2.480015212299953e-06,
      "loss": 0.4622,
      "step": 247920
    },
    {
      "epoch": 101.49,
      "grad_norm": 2.100320816040039,
      "learning_rate": 2.4798568507686927e-06,
      "loss": 0.4695,
      "step": 247930
    },
    {
      "epoch": 101.49,
      "grad_norm": 2.421860456466675,
      "learning_rate": 2.4796984886139475e-06,
      "loss": 0.4647,
      "step": 247940
    },
    {
      "epoch": 101.49,
      "grad_norm": 2.039463758468628,
      "learning_rate": 2.4795401258364432e-06,
      "loss": 0.4756,
      "step": 247950
    },
    {
      "epoch": 101.5,
      "grad_norm": 1.8117492198944092,
      "learning_rate": 2.4793817624369056e-06,
      "loss": 0.4539,
      "step": 247960
    },
    {
      "epoch": 101.5,
      "grad_norm": 1.5813677310943604,
      "learning_rate": 2.4792233984160603e-06,
      "loss": 0.4426,
      "step": 247970
    },
    {
      "epoch": 101.51,
      "grad_norm": 1.8796488046646118,
      "learning_rate": 2.4790650337746318e-06,
      "loss": 0.4802,
      "step": 247980
    },
    {
      "epoch": 101.51,
      "grad_norm": 2.015658378601074,
      "learning_rate": 2.4789066685133457e-06,
      "loss": 0.5031,
      "step": 247990
    },
    {
      "epoch": 101.51,
      "grad_norm": 1.9294310808181763,
      "learning_rate": 2.478748302632929e-06,
      "loss": 0.4926,
      "step": 248000
    },
    {
      "epoch": 101.52,
      "grad_norm": 2.195995569229126,
      "learning_rate": 2.4785899361341054e-06,
      "loss": 0.4711,
      "step": 248010
    },
    {
      "epoch": 101.52,
      "grad_norm": 1.54714834690094,
      "learning_rate": 2.4784315690176013e-06,
      "loss": 0.4603,
      "step": 248020
    },
    {
      "epoch": 101.53,
      "grad_norm": 1.9540990591049194,
      "learning_rate": 2.478273201284141e-06,
      "loss": 0.4903,
      "step": 248030
    },
    {
      "epoch": 101.53,
      "grad_norm": 1.7103787660598755,
      "learning_rate": 2.478114832934451e-06,
      "loss": 0.4701,
      "step": 248040
    },
    {
      "epoch": 101.53,
      "grad_norm": 1.7375102043151855,
      "learning_rate": 2.477956463969257e-06,
      "loss": 0.4706,
      "step": 248050
    },
    {
      "epoch": 101.54,
      "grad_norm": 2.2468745708465576,
      "learning_rate": 2.477798094389283e-06,
      "loss": 0.4764,
      "step": 248060
    },
    {
      "epoch": 101.54,
      "grad_norm": 2.1629440784454346,
      "learning_rate": 2.4776397241952567e-06,
      "loss": 0.4915,
      "step": 248070
    },
    {
      "epoch": 101.55,
      "grad_norm": 1.763932704925537,
      "learning_rate": 2.4774813533879013e-06,
      "loss": 0.4725,
      "step": 248080
    },
    {
      "epoch": 101.55,
      "grad_norm": 1.6899335384368896,
      "learning_rate": 2.477322981967944e-06,
      "loss": 0.4861,
      "step": 248090
    },
    {
      "epoch": 101.56,
      "grad_norm": 2.497755289077759,
      "learning_rate": 2.477164609936109e-06,
      "loss": 0.4746,
      "step": 248100
    },
    {
      "epoch": 101.56,
      "grad_norm": 1.4526021480560303,
      "learning_rate": 2.4770062372931228e-06,
      "loss": 0.4759,
      "step": 248110
    },
    {
      "epoch": 101.56,
      "grad_norm": 4.082660675048828,
      "learning_rate": 2.47684786403971e-06,
      "loss": 0.4779,
      "step": 248120
    },
    {
      "epoch": 101.57,
      "grad_norm": 2.2144625186920166,
      "learning_rate": 2.476689490176597e-06,
      "loss": 0.4702,
      "step": 248130
    },
    {
      "epoch": 101.57,
      "grad_norm": 1.5892257690429688,
      "learning_rate": 2.476531115704508e-06,
      "loss": 0.4623,
      "step": 248140
    },
    {
      "epoch": 101.58,
      "grad_norm": 2.7077200412750244,
      "learning_rate": 2.47637274062417e-06,
      "loss": 0.4613,
      "step": 248150
    },
    {
      "epoch": 101.58,
      "grad_norm": 2.0271313190460205,
      "learning_rate": 2.476214364936308e-06,
      "loss": 0.473,
      "step": 248160
    },
    {
      "epoch": 101.58,
      "grad_norm": 1.7686933279037476,
      "learning_rate": 2.476055988641646e-06,
      "loss": 0.4784,
      "step": 248170
    },
    {
      "epoch": 101.59,
      "grad_norm": 1.8094984292984009,
      "learning_rate": 2.4758976117409118e-06,
      "loss": 0.4779,
      "step": 248180
    },
    {
      "epoch": 101.59,
      "grad_norm": 1.9044678211212158,
      "learning_rate": 2.475739234234829e-06,
      "loss": 0.4787,
      "step": 248190
    },
    {
      "epoch": 101.6,
      "grad_norm": 1.9731805324554443,
      "learning_rate": 2.4755808561241247e-06,
      "loss": 0.4846,
      "step": 248200
    },
    {
      "epoch": 101.6,
      "grad_norm": 1.5699325799942017,
      "learning_rate": 2.475422477409524e-06,
      "loss": 0.4695,
      "step": 248210
    },
    {
      "epoch": 101.6,
      "grad_norm": 1.8332582712173462,
      "learning_rate": 2.4752640980917517e-06,
      "loss": 0.4527,
      "step": 248220
    },
    {
      "epoch": 101.61,
      "grad_norm": 1.6154582500457764,
      "learning_rate": 2.4751057181715335e-06,
      "loss": 0.4844,
      "step": 248230
    },
    {
      "epoch": 101.61,
      "grad_norm": 2.584867477416992,
      "learning_rate": 2.474947337649595e-06,
      "loss": 0.4804,
      "step": 248240
    },
    {
      "epoch": 101.62,
      "grad_norm": 2.0658457279205322,
      "learning_rate": 2.474788956526662e-06,
      "loss": 0.4877,
      "step": 248250
    },
    {
      "epoch": 101.62,
      "grad_norm": 2.115365982055664,
      "learning_rate": 2.47463057480346e-06,
      "loss": 0.4782,
      "step": 248260
    },
    {
      "epoch": 101.63,
      "grad_norm": 1.9345574378967285,
      "learning_rate": 2.4744721924807138e-06,
      "loss": 0.4797,
      "step": 248270
    },
    {
      "epoch": 101.63,
      "grad_norm": 2.0108931064605713,
      "learning_rate": 2.47431380955915e-06,
      "loss": 0.4883,
      "step": 248280
    },
    {
      "epoch": 101.63,
      "grad_norm": 2.0058603286743164,
      "learning_rate": 2.4741554260394936e-06,
      "loss": 0.4746,
      "step": 248290
    },
    {
      "epoch": 101.64,
      "grad_norm": 2.2767908573150635,
      "learning_rate": 2.47399704192247e-06,
      "loss": 0.4557,
      "step": 248300
    },
    {
      "epoch": 101.64,
      "grad_norm": 1.85377836227417,
      "learning_rate": 2.4738386572088045e-06,
      "loss": 0.4687,
      "step": 248310
    },
    {
      "epoch": 101.65,
      "grad_norm": 1.9124449491500854,
      "learning_rate": 2.4736802718992236e-06,
      "loss": 0.4864,
      "step": 248320
    },
    {
      "epoch": 101.65,
      "grad_norm": 2.0601701736450195,
      "learning_rate": 2.473521885994452e-06,
      "loss": 0.4662,
      "step": 248330
    },
    {
      "epoch": 101.65,
      "grad_norm": 2.0115652084350586,
      "learning_rate": 2.4733634994952154e-06,
      "loss": 0.4733,
      "step": 248340
    },
    {
      "epoch": 101.66,
      "grad_norm": 2.028986930847168,
      "learning_rate": 2.4732051124022396e-06,
      "loss": 0.5005,
      "step": 248350
    },
    {
      "epoch": 101.66,
      "grad_norm": 2.050811529159546,
      "learning_rate": 2.47304672471625e-06,
      "loss": 0.4901,
      "step": 248360
    },
    {
      "epoch": 101.67,
      "grad_norm": 1.7872463464736938,
      "learning_rate": 2.4728883364379716e-06,
      "loss": 0.4794,
      "step": 248370
    },
    {
      "epoch": 101.67,
      "grad_norm": 1.8570092916488647,
      "learning_rate": 2.472729947568131e-06,
      "loss": 0.4683,
      "step": 248380
    },
    {
      "epoch": 101.67,
      "grad_norm": 1.6061698198318481,
      "learning_rate": 2.472571558107453e-06,
      "loss": 0.4684,
      "step": 248390
    },
    {
      "epoch": 101.68,
      "grad_norm": 1.8835188150405884,
      "learning_rate": 2.472413168056664e-06,
      "loss": 0.4747,
      "step": 248400
    },
    {
      "epoch": 101.68,
      "grad_norm": 1.9532164335250854,
      "learning_rate": 2.4722547774164884e-06,
      "loss": 0.473,
      "step": 248410
    },
    {
      "epoch": 101.69,
      "grad_norm": 2.69574236869812,
      "learning_rate": 2.4720963861876523e-06,
      "loss": 0.4769,
      "step": 248420
    },
    {
      "epoch": 101.69,
      "grad_norm": 1.948482632637024,
      "learning_rate": 2.4719379943708814e-06,
      "loss": 0.4767,
      "step": 248430
    },
    {
      "epoch": 101.69,
      "grad_norm": 2.138171434402466,
      "learning_rate": 2.471779601966901e-06,
      "loss": 0.4458,
      "step": 248440
    },
    {
      "epoch": 101.7,
      "grad_norm": 1.7121461629867554,
      "learning_rate": 2.4716212089764376e-06,
      "loss": 0.4619,
      "step": 248450
    },
    {
      "epoch": 101.7,
      "grad_norm": 1.6514047384262085,
      "learning_rate": 2.4714628154002145e-06,
      "loss": 0.4887,
      "step": 248460
    },
    {
      "epoch": 101.71,
      "grad_norm": 1.8493660688400269,
      "learning_rate": 2.4713044212389596e-06,
      "loss": 0.495,
      "step": 248470
    },
    {
      "epoch": 101.71,
      "grad_norm": 1.869495153427124,
      "learning_rate": 2.4711460264933978e-06,
      "loss": 0.4772,
      "step": 248480
    },
    {
      "epoch": 101.72,
      "grad_norm": 2.1107499599456787,
      "learning_rate": 2.4709876311642544e-06,
      "loss": 0.49,
      "step": 248490
    },
    {
      "epoch": 101.72,
      "grad_norm": 1.874770164489746,
      "learning_rate": 2.4708292352522553e-06,
      "loss": 0.4589,
      "step": 248500
    },
    {
      "epoch": 101.72,
      "grad_norm": 1.9767855405807495,
      "learning_rate": 2.4706708387581256e-06,
      "loss": 0.4824,
      "step": 248510
    },
    {
      "epoch": 101.73,
      "grad_norm": 1.7512037754058838,
      "learning_rate": 2.470512441682592e-06,
      "loss": 0.4719,
      "step": 248520
    },
    {
      "epoch": 101.73,
      "grad_norm": 2.004223108291626,
      "learning_rate": 2.470354044026379e-06,
      "loss": 0.4716,
      "step": 248530
    },
    {
      "epoch": 101.74,
      "grad_norm": 2.0356075763702393,
      "learning_rate": 2.4701956457902117e-06,
      "loss": 0.4644,
      "step": 248540
    },
    {
      "epoch": 101.74,
      "grad_norm": 2.0021464824676514,
      "learning_rate": 2.470037246974816e-06,
      "loss": 0.4802,
      "step": 248550
    },
    {
      "epoch": 101.74,
      "grad_norm": 2.2356462478637695,
      "learning_rate": 2.4698788475809192e-06,
      "loss": 0.4668,
      "step": 248560
    },
    {
      "epoch": 101.75,
      "grad_norm": 1.993066430091858,
      "learning_rate": 2.4697204476092454e-06,
      "loss": 0.479,
      "step": 248570
    },
    {
      "epoch": 101.75,
      "grad_norm": 2.182023048400879,
      "learning_rate": 2.469562047060521e-06,
      "loss": 0.4706,
      "step": 248580
    },
    {
      "epoch": 101.76,
      "grad_norm": 1.9886224269866943,
      "learning_rate": 2.4694036459354704e-06,
      "loss": 0.4775,
      "step": 248590
    },
    {
      "epoch": 101.76,
      "grad_norm": 2.0769083499908447,
      "learning_rate": 2.46924524423482e-06,
      "loss": 0.4546,
      "step": 248600
    },
    {
      "epoch": 101.76,
      "grad_norm": 1.9261150360107422,
      "learning_rate": 2.469086841959295e-06,
      "loss": 0.4779,
      "step": 248610
    },
    {
      "epoch": 101.77,
      "grad_norm": 2.0428593158721924,
      "learning_rate": 2.4689284391096223e-06,
      "loss": 0.4798,
      "step": 248620
    },
    {
      "epoch": 101.77,
      "grad_norm": 1.89871084690094,
      "learning_rate": 2.4687700356865255e-06,
      "loss": 0.483,
      "step": 248630
    },
    {
      "epoch": 101.78,
      "grad_norm": 2.021955966949463,
      "learning_rate": 2.468611631690732e-06,
      "loss": 0.4736,
      "step": 248640
    },
    {
      "epoch": 101.78,
      "grad_norm": 2.009272336959839,
      "learning_rate": 2.468453227122966e-06,
      "loss": 0.474,
      "step": 248650
    },
    {
      "epoch": 101.78,
      "grad_norm": 2.6420819759368896,
      "learning_rate": 2.4682948219839544e-06,
      "loss": 0.4626,
      "step": 248660
    },
    {
      "epoch": 101.79,
      "grad_norm": 2.0650057792663574,
      "learning_rate": 2.4681364162744223e-06,
      "loss": 0.4773,
      "step": 248670
    },
    {
      "epoch": 101.79,
      "grad_norm": 2.0789825916290283,
      "learning_rate": 2.467978009995095e-06,
      "loss": 0.4657,
      "step": 248680
    },
    {
      "epoch": 101.8,
      "grad_norm": 1.9918384552001953,
      "learning_rate": 2.4678196031466985e-06,
      "loss": 0.4775,
      "step": 248690
    },
    {
      "epoch": 101.8,
      "grad_norm": 1.9555106163024902,
      "learning_rate": 2.4676611957299587e-06,
      "loss": 0.4662,
      "step": 248700
    },
    {
      "epoch": 101.81,
      "grad_norm": 2.049816846847534,
      "learning_rate": 2.4675027877456e-06,
      "loss": 0.4619,
      "step": 248710
    },
    {
      "epoch": 101.81,
      "grad_norm": 1.8839192390441895,
      "learning_rate": 2.46734437919435e-06,
      "loss": 0.4923,
      "step": 248720
    },
    {
      "epoch": 101.81,
      "grad_norm": 2.2547411918640137,
      "learning_rate": 2.467185970076933e-06,
      "loss": 0.4865,
      "step": 248730
    },
    {
      "epoch": 101.82,
      "grad_norm": 1.7502325773239136,
      "learning_rate": 2.4670275603940743e-06,
      "loss": 0.4949,
      "step": 248740
    },
    {
      "epoch": 101.82,
      "grad_norm": 2.1100287437438965,
      "learning_rate": 2.466869150146501e-06,
      "loss": 0.4763,
      "step": 248750
    },
    {
      "epoch": 101.83,
      "grad_norm": 1.527752161026001,
      "learning_rate": 2.4667107393349373e-06,
      "loss": 0.4808,
      "step": 248760
    },
    {
      "epoch": 101.83,
      "grad_norm": 1.9475046396255493,
      "learning_rate": 2.4665523279601095e-06,
      "loss": 0.467,
      "step": 248770
    },
    {
      "epoch": 101.83,
      "grad_norm": 1.6097098588943481,
      "learning_rate": 2.4663939160227434e-06,
      "loss": 0.4635,
      "step": 248780
    },
    {
      "epoch": 101.84,
      "grad_norm": 1.9568605422973633,
      "learning_rate": 2.466235503523565e-06,
      "loss": 0.4777,
      "step": 248790
    },
    {
      "epoch": 101.84,
      "grad_norm": 2.0146353244781494,
      "learning_rate": 2.466077090463299e-06,
      "loss": 0.4702,
      "step": 248800
    },
    {
      "epoch": 101.85,
      "grad_norm": 2.3165647983551025,
      "learning_rate": 2.4659186768426724e-06,
      "loss": 0.4725,
      "step": 248810
    },
    {
      "epoch": 101.85,
      "grad_norm": 1.8016852140426636,
      "learning_rate": 2.4657602626624092e-06,
      "loss": 0.4816,
      "step": 248820
    },
    {
      "epoch": 101.85,
      "grad_norm": 1.8573676347732544,
      "learning_rate": 2.465601847923235e-06,
      "loss": 0.4859,
      "step": 248830
    },
    {
      "epoch": 101.86,
      "grad_norm": 2.33516526222229,
      "learning_rate": 2.465443432625878e-06,
      "loss": 0.4691,
      "step": 248840
    },
    {
      "epoch": 101.86,
      "grad_norm": 2.084580421447754,
      "learning_rate": 2.4652850167710615e-06,
      "loss": 0.4956,
      "step": 248850
    },
    {
      "epoch": 101.87,
      "grad_norm": 1.8790165185928345,
      "learning_rate": 2.465126600359512e-06,
      "loss": 0.487,
      "step": 248860
    },
    {
      "epoch": 101.87,
      "grad_norm": 2.047022581100464,
      "learning_rate": 2.4649681833919556e-06,
      "loss": 0.4805,
      "step": 248870
    },
    {
      "epoch": 101.87,
      "grad_norm": 2.963355779647827,
      "learning_rate": 2.4648097658691166e-06,
      "loss": 0.4645,
      "step": 248880
    },
    {
      "epoch": 101.88,
      "grad_norm": 1.7124207019805908,
      "learning_rate": 2.464651347791722e-06,
      "loss": 0.4868,
      "step": 248890
    },
    {
      "epoch": 101.88,
      "grad_norm": 1.6931557655334473,
      "learning_rate": 2.464492929160497e-06,
      "loss": 0.462,
      "step": 248900
    },
    {
      "epoch": 101.89,
      "grad_norm": 2.0675206184387207,
      "learning_rate": 2.4643345099761673e-06,
      "loss": 0.4715,
      "step": 248910
    },
    {
      "epoch": 101.89,
      "grad_norm": 1.821105718612671,
      "learning_rate": 2.4641760902394587e-06,
      "loss": 0.4696,
      "step": 248920
    },
    {
      "epoch": 101.9,
      "grad_norm": 2.4179577827453613,
      "learning_rate": 2.464017669951096e-06,
      "loss": 0.4724,
      "step": 248930
    },
    {
      "epoch": 101.9,
      "grad_norm": 1.7608091831207275,
      "learning_rate": 2.463859249111807e-06,
      "loss": 0.4728,
      "step": 248940
    },
    {
      "epoch": 101.9,
      "grad_norm": 2.3325366973876953,
      "learning_rate": 2.4637008277223158e-06,
      "loss": 0.4697,
      "step": 248950
    },
    {
      "epoch": 101.91,
      "grad_norm": 1.9279247522354126,
      "learning_rate": 2.4635424057833485e-06,
      "loss": 0.4888,
      "step": 248960
    },
    {
      "epoch": 101.91,
      "grad_norm": 1.9172676801681519,
      "learning_rate": 2.46338398329563e-06,
      "loss": 0.4865,
      "step": 248970
    },
    {
      "epoch": 101.92,
      "grad_norm": 2.074071168899536,
      "learning_rate": 2.4632255602598874e-06,
      "loss": 0.4905,
      "step": 248980
    },
    {
      "epoch": 101.92,
      "grad_norm": 1.750623345375061,
      "learning_rate": 2.4630671366768454e-06,
      "loss": 0.4883,
      "step": 248990
    },
    {
      "epoch": 101.92,
      "grad_norm": 1.8229776620864868,
      "learning_rate": 2.4629087125472303e-06,
      "loss": 0.4848,
      "step": 249000
    },
    {
      "epoch": 101.93,
      "grad_norm": 1.9037986993789673,
      "learning_rate": 2.4627502878717677e-06,
      "loss": 0.4543,
      "step": 249010
    },
    {
      "epoch": 101.93,
      "grad_norm": 1.6350094079971313,
      "learning_rate": 2.462591862651183e-06,
      "loss": 0.4616,
      "step": 249020
    },
    {
      "epoch": 101.94,
      "grad_norm": 1.5054070949554443,
      "learning_rate": 2.4624334368862024e-06,
      "loss": 0.4644,
      "step": 249030
    },
    {
      "epoch": 101.94,
      "grad_norm": 1.7435485124588013,
      "learning_rate": 2.462275010577551e-06,
      "loss": 0.4635,
      "step": 249040
    },
    {
      "epoch": 101.94,
      "grad_norm": 1.6752920150756836,
      "learning_rate": 2.462116583725955e-06,
      "loss": 0.4726,
      "step": 249050
    },
    {
      "epoch": 101.95,
      "grad_norm": 2.2932865619659424,
      "learning_rate": 2.4619581563321404e-06,
      "loss": 0.4806,
      "step": 249060
    },
    {
      "epoch": 101.95,
      "grad_norm": 1.8923957347869873,
      "learning_rate": 2.461799728396832e-06,
      "loss": 0.468,
      "step": 249070
    },
    {
      "epoch": 101.96,
      "grad_norm": 2.1766886711120605,
      "learning_rate": 2.461641299920756e-06,
      "loss": 0.4753,
      "step": 249080
    },
    {
      "epoch": 101.96,
      "grad_norm": 1.827966570854187,
      "learning_rate": 2.4614828709046384e-06,
      "loss": 0.4713,
      "step": 249090
    },
    {
      "epoch": 101.96,
      "grad_norm": 2.6825077533721924,
      "learning_rate": 2.4613244413492055e-06,
      "loss": 0.4767,
      "step": 249100
    },
    {
      "epoch": 101.97,
      "grad_norm": 2.081129312515259,
      "learning_rate": 2.4611660112551817e-06,
      "loss": 0.4825,
      "step": 249110
    },
    {
      "epoch": 101.97,
      "grad_norm": 1.7602256536483765,
      "learning_rate": 2.461007580623293e-06,
      "loss": 0.4761,
      "step": 249120
    },
    {
      "epoch": 101.98,
      "grad_norm": 2.108175039291382,
      "learning_rate": 2.460849149454266e-06,
      "loss": 0.4958,
      "step": 249130
    },
    {
      "epoch": 101.98,
      "grad_norm": 1.7389756441116333,
      "learning_rate": 2.460690717748825e-06,
      "loss": 0.4783,
      "step": 249140
    },
    {
      "epoch": 101.99,
      "grad_norm": 2.3488962650299072,
      "learning_rate": 2.4605322855076977e-06,
      "loss": 0.4648,
      "step": 249150
    },
    {
      "epoch": 101.99,
      "grad_norm": 1.5856292247772217,
      "learning_rate": 2.460373852731609e-06,
      "loss": 0.4658,
      "step": 249160
    },
    {
      "epoch": 101.99,
      "grad_norm": 1.9123151302337646,
      "learning_rate": 2.460215419421284e-06,
      "loss": 0.4783,
      "step": 249170
    },
    {
      "epoch": 102.0,
      "grad_norm": 1.8436099290847778,
      "learning_rate": 2.460056985577449e-06,
      "loss": 0.4702,
      "step": 249180
    },
    {
      "epoch": 102.0,
      "eval_loss": 0.4793243408203125,
      "eval_runtime": 52.7481,
      "eval_samples_per_second": 65.386,
      "eval_steps_per_second": 8.19,
      "step": 249186
    },
    {
      "epoch": 102.0,
      "grad_norm": 1.52336585521698,
      "learning_rate": 2.4598985512008293e-06,
      "loss": 0.4618,
      "step": 249190
    },
    {
      "epoch": 102.01,
      "grad_norm": 1.8383105993270874,
      "learning_rate": 2.4597401162921516e-06,
      "loss": 0.4741,
      "step": 249200
    },
    {
      "epoch": 102.01,
      "grad_norm": 6.019161701202393,
      "learning_rate": 2.4595816808521407e-06,
      "loss": 0.4844,
      "step": 249210
    },
    {
      "epoch": 102.01,
      "grad_norm": 2.0365169048309326,
      "learning_rate": 2.4594232448815226e-06,
      "loss": 0.4519,
      "step": 249220
    },
    {
      "epoch": 102.02,
      "grad_norm": 1.9905630350112915,
      "learning_rate": 2.4592648083810244e-06,
      "loss": 0.467,
      "step": 249230
    },
    {
      "epoch": 102.02,
      "grad_norm": 1.7700035572052002,
      "learning_rate": 2.4591063713513703e-06,
      "loss": 0.461,
      "step": 249240
    },
    {
      "epoch": 102.03,
      "grad_norm": 2.2523508071899414,
      "learning_rate": 2.4589479337932866e-06,
      "loss": 0.5067,
      "step": 249250
    },
    {
      "epoch": 102.03,
      "grad_norm": 1.9185550212860107,
      "learning_rate": 2.4587894957074984e-06,
      "loss": 0.4864,
      "step": 249260
    },
    {
      "epoch": 102.03,
      "grad_norm": 1.904159665107727,
      "learning_rate": 2.4586310570947325e-06,
      "loss": 0.4783,
      "step": 249270
    },
    {
      "epoch": 102.04,
      "grad_norm": 1.9310886859893799,
      "learning_rate": 2.4584726179557144e-06,
      "loss": 0.4804,
      "step": 249280
    },
    {
      "epoch": 102.04,
      "grad_norm": 1.8999890089035034,
      "learning_rate": 2.4583141782911695e-06,
      "loss": 0.469,
      "step": 249290
    },
    {
      "epoch": 102.05,
      "grad_norm": 2.37245774269104,
      "learning_rate": 2.458155738101824e-06,
      "loss": 0.4998,
      "step": 249300
    },
    {
      "epoch": 102.05,
      "grad_norm": 2.1207385063171387,
      "learning_rate": 2.4579972973884034e-06,
      "loss": 0.462,
      "step": 249310
    },
    {
      "epoch": 102.05,
      "grad_norm": 1.4604190587997437,
      "learning_rate": 2.4578388561516332e-06,
      "loss": 0.4731,
      "step": 249320
    },
    {
      "epoch": 102.06,
      "grad_norm": 1.978458046913147,
      "learning_rate": 2.45768041439224e-06,
      "loss": 0.4715,
      "step": 249330
    },
    {
      "epoch": 102.06,
      "grad_norm": 2.045665979385376,
      "learning_rate": 2.4575219721109497e-06,
      "loss": 0.4773,
      "step": 249340
    },
    {
      "epoch": 102.07,
      "grad_norm": 2.2507691383361816,
      "learning_rate": 2.457363529308487e-06,
      "loss": 0.4789,
      "step": 249350
    },
    {
      "epoch": 102.07,
      "grad_norm": 1.7927682399749756,
      "learning_rate": 2.4572050859855784e-06,
      "loss": 0.4706,
      "step": 249360
    },
    {
      "epoch": 102.08,
      "grad_norm": 1.8426185846328735,
      "learning_rate": 2.4570466421429503e-06,
      "loss": 0.4738,
      "step": 249370
    },
    {
      "epoch": 102.08,
      "grad_norm": 2.0040135383605957,
      "learning_rate": 2.456888197781327e-06,
      "loss": 0.4693,
      "step": 249380
    },
    {
      "epoch": 102.08,
      "grad_norm": 1.622275948524475,
      "learning_rate": 2.456729752901436e-06,
      "loss": 0.4988,
      "step": 249390
    },
    {
      "epoch": 102.09,
      "grad_norm": 1.4899473190307617,
      "learning_rate": 2.456571307504001e-06,
      "loss": 0.4671,
      "step": 249400
    },
    {
      "epoch": 102.09,
      "grad_norm": 1.8804570436477661,
      "learning_rate": 2.4564128615897495e-06,
      "loss": 0.4596,
      "step": 249410
    },
    {
      "epoch": 102.1,
      "grad_norm": 2.2553484439849854,
      "learning_rate": 2.4562544151594074e-06,
      "loss": 0.4629,
      "step": 249420
    },
    {
      "epoch": 102.1,
      "grad_norm": 1.5778453350067139,
      "learning_rate": 2.4560959682136995e-06,
      "loss": 0.483,
      "step": 249430
    },
    {
      "epoch": 102.1,
      "grad_norm": 2.0920872688293457,
      "learning_rate": 2.4559375207533526e-06,
      "loss": 0.4858,
      "step": 249440
    },
    {
      "epoch": 102.11,
      "grad_norm": 2.0317580699920654,
      "learning_rate": 2.4557790727790918e-06,
      "loss": 0.4604,
      "step": 249450
    },
    {
      "epoch": 102.11,
      "grad_norm": 1.8422433137893677,
      "learning_rate": 2.4556206242916433e-06,
      "loss": 0.4796,
      "step": 249460
    },
    {
      "epoch": 102.12,
      "grad_norm": 1.548359751701355,
      "learning_rate": 2.4554621752917327e-06,
      "loss": 0.4887,
      "step": 249470
    },
    {
      "epoch": 102.12,
      "grad_norm": 1.8241328001022339,
      "learning_rate": 2.455303725780086e-06,
      "loss": 0.4683,
      "step": 249480
    },
    {
      "epoch": 102.12,
      "grad_norm": 2.2497644424438477,
      "learning_rate": 2.455145275757428e-06,
      "loss": 0.4883,
      "step": 249490
    },
    {
      "epoch": 102.13,
      "grad_norm": 1.9104387760162354,
      "learning_rate": 2.4549868252244863e-06,
      "loss": 0.4619,
      "step": 249500
    },
    {
      "epoch": 102.13,
      "grad_norm": 1.5927233695983887,
      "learning_rate": 2.454828374181986e-06,
      "loss": 0.48,
      "step": 249510
    },
    {
      "epoch": 102.14,
      "grad_norm": 1.5014575719833374,
      "learning_rate": 2.4546699226306532e-06,
      "loss": 0.4678,
      "step": 249520
    },
    {
      "epoch": 102.14,
      "grad_norm": 2.412367105484009,
      "learning_rate": 2.454511470571213e-06,
      "loss": 0.4764,
      "step": 249530
    },
    {
      "epoch": 102.14,
      "grad_norm": 1.7970904111862183,
      "learning_rate": 2.454353018004392e-06,
      "loss": 0.4799,
      "step": 249540
    },
    {
      "epoch": 102.15,
      "grad_norm": 1.6280784606933594,
      "learning_rate": 2.4541945649309154e-06,
      "loss": 0.4816,
      "step": 249550
    },
    {
      "epoch": 102.15,
      "grad_norm": 1.8859277963638306,
      "learning_rate": 2.4540361113515093e-06,
      "loss": 0.4699,
      "step": 249560
    },
    {
      "epoch": 102.16,
      "grad_norm": 2.0345230102539062,
      "learning_rate": 2.4538776572668997e-06,
      "loss": 0.4753,
      "step": 249570
    },
    {
      "epoch": 102.16,
      "grad_norm": 1.8436411619186401,
      "learning_rate": 2.4537192026778117e-06,
      "loss": 0.4923,
      "step": 249580
    },
    {
      "epoch": 102.17,
      "grad_norm": 2.215902090072632,
      "learning_rate": 2.453560747584973e-06,
      "loss": 0.4614,
      "step": 249590
    },
    {
      "epoch": 102.17,
      "grad_norm": 1.8407565355300903,
      "learning_rate": 2.4534022919891075e-06,
      "loss": 0.476,
      "step": 249600
    },
    {
      "epoch": 102.17,
      "grad_norm": 1.7648038864135742,
      "learning_rate": 2.453243835890942e-06,
      "loss": 0.4888,
      "step": 249610
    },
    {
      "epoch": 102.18,
      "grad_norm": 2.3882551193237305,
      "learning_rate": 2.453085379291202e-06,
      "loss": 0.4701,
      "step": 249620
    },
    {
      "epoch": 102.18,
      "grad_norm": 2.3005337715148926,
      "learning_rate": 2.4529269221906136e-06,
      "loss": 0.4766,
      "step": 249630
    },
    {
      "epoch": 102.19,
      "grad_norm": 1.863562822341919,
      "learning_rate": 2.4527684645899034e-06,
      "loss": 0.4624,
      "step": 249640
    },
    {
      "epoch": 102.19,
      "grad_norm": 1.678847074508667,
      "learning_rate": 2.4526100064897956e-06,
      "loss": 0.4641,
      "step": 249650
    },
    {
      "epoch": 102.19,
      "grad_norm": 1.9894286394119263,
      "learning_rate": 2.452451547891017e-06,
      "loss": 0.4833,
      "step": 249660
    },
    {
      "epoch": 102.2,
      "grad_norm": 2.229522228240967,
      "learning_rate": 2.452293088794294e-06,
      "loss": 0.4725,
      "step": 249670
    },
    {
      "epoch": 102.2,
      "grad_norm": 2.307117223739624,
      "learning_rate": 2.4521346292003516e-06,
      "loss": 0.4897,
      "step": 249680
    },
    {
      "epoch": 102.21,
      "grad_norm": 1.6689265966415405,
      "learning_rate": 2.451976169109916e-06,
      "loss": 0.475,
      "step": 249690
    },
    {
      "epoch": 102.21,
      "grad_norm": 1.8592009544372559,
      "learning_rate": 2.4518177085237127e-06,
      "loss": 0.48,
      "step": 249700
    },
    {
      "epoch": 102.21,
      "grad_norm": 2.031912088394165,
      "learning_rate": 2.4516592474424687e-06,
      "loss": 0.4824,
      "step": 249710
    },
    {
      "epoch": 102.22,
      "grad_norm": 2.037442445755005,
      "learning_rate": 2.451500785866909e-06,
      "loss": 0.4848,
      "step": 249720
    },
    {
      "epoch": 102.22,
      "grad_norm": 1.8982303142547607,
      "learning_rate": 2.4513423237977594e-06,
      "loss": 0.481,
      "step": 249730
    },
    {
      "epoch": 102.23,
      "grad_norm": 1.8668420314788818,
      "learning_rate": 2.451183861235746e-06,
      "loss": 0.4787,
      "step": 249740
    },
    {
      "epoch": 102.23,
      "grad_norm": 1.901068925857544,
      "learning_rate": 2.4510253981815953e-06,
      "loss": 0.4691,
      "step": 249750
    },
    {
      "epoch": 102.23,
      "grad_norm": 2.431727647781372,
      "learning_rate": 2.450866934636032e-06,
      "loss": 0.4775,
      "step": 249760
    },
    {
      "epoch": 102.24,
      "grad_norm": 2.062929630279541,
      "learning_rate": 2.4507084705997825e-06,
      "loss": 0.4648,
      "step": 249770
    },
    {
      "epoch": 102.24,
      "grad_norm": 1.9473820924758911,
      "learning_rate": 2.450550006073573e-06,
      "loss": 0.4689,
      "step": 249780
    },
    {
      "epoch": 102.25,
      "grad_norm": 2.1854054927825928,
      "learning_rate": 2.450391541058129e-06,
      "loss": 0.4635,
      "step": 249790
    },
    {
      "epoch": 102.25,
      "grad_norm": 2.013890266418457,
      "learning_rate": 2.450233075554177e-06,
      "loss": 0.4758,
      "step": 249800
    },
    {
      "epoch": 102.26,
      "grad_norm": 1.8383803367614746,
      "learning_rate": 2.4500746095624423e-06,
      "loss": 0.4537,
      "step": 249810
    },
    {
      "epoch": 102.26,
      "grad_norm": 1.8855892419815063,
      "learning_rate": 2.4499161430836515e-06,
      "loss": 0.4938,
      "step": 249820
    },
    {
      "epoch": 102.26,
      "grad_norm": 1.9243007898330688,
      "learning_rate": 2.44975767611853e-06,
      "loss": 0.478,
      "step": 249830
    },
    {
      "epoch": 102.27,
      "grad_norm": 1.846311330795288,
      "learning_rate": 2.4495992086678032e-06,
      "loss": 0.4915,
      "step": 249840
    },
    {
      "epoch": 102.27,
      "grad_norm": 2.1535446643829346,
      "learning_rate": 2.449440740732198e-06,
      "loss": 0.5035,
      "step": 249850
    },
    {
      "epoch": 102.28,
      "grad_norm": 1.7728617191314697,
      "learning_rate": 2.4492822723124396e-06,
      "loss": 0.4476,
      "step": 249860
    },
    {
      "epoch": 102.28,
      "grad_norm": 1.7453590631484985,
      "learning_rate": 2.4491238034092537e-06,
      "loss": 0.4799,
      "step": 249870
    },
    {
      "epoch": 102.28,
      "grad_norm": 2.0474798679351807,
      "learning_rate": 2.448965334023368e-06,
      "loss": 0.4844,
      "step": 249880
    },
    {
      "epoch": 102.29,
      "grad_norm": 1.7282347679138184,
      "learning_rate": 2.448806864155506e-06,
      "loss": 0.4942,
      "step": 249890
    },
    {
      "epoch": 102.29,
      "grad_norm": 1.6919164657592773,
      "learning_rate": 2.448648393806395e-06,
      "loss": 0.4797,
      "step": 249900
    },
    {
      "epoch": 102.3,
      "grad_norm": 1.968795895576477,
      "learning_rate": 2.448489922976761e-06,
      "loss": 0.4498,
      "step": 249910
    },
    {
      "epoch": 102.3,
      "grad_norm": 1.9733062982559204,
      "learning_rate": 2.4483314516673296e-06,
      "loss": 0.4991,
      "step": 249920
    },
    {
      "epoch": 102.3,
      "grad_norm": 1.7224485874176025,
      "learning_rate": 2.4481729798788272e-06,
      "loss": 0.4737,
      "step": 249930
    },
    {
      "epoch": 102.31,
      "grad_norm": 1.7670882940292358,
      "learning_rate": 2.4480145076119782e-06,
      "loss": 0.4734,
      "step": 249940
    },
    {
      "epoch": 102.31,
      "grad_norm": 2.0167458057403564,
      "learning_rate": 2.4478560348675107e-06,
      "loss": 0.4846,
      "step": 249950
    },
    {
      "epoch": 102.32,
      "grad_norm": 2.1609365940093994,
      "learning_rate": 2.447697561646149e-06,
      "loss": 0.4726,
      "step": 249960
    },
    {
      "epoch": 102.32,
      "grad_norm": 2.150315284729004,
      "learning_rate": 2.44753908794862e-06,
      "loss": 0.5103,
      "step": 249970
    },
    {
      "epoch": 102.33,
      "grad_norm": 2.3594136238098145,
      "learning_rate": 2.4473806137756483e-06,
      "loss": 0.471,
      "step": 249980
    },
    {
      "epoch": 102.33,
      "grad_norm": 1.6726986169815063,
      "learning_rate": 2.447222139127962e-06,
      "loss": 0.4957,
      "step": 249990
    },
    {
      "epoch": 102.33,
      "grad_norm": 1.9403268098831177,
      "learning_rate": 2.447063664006285e-06,
      "loss": 0.4724,
      "step": 250000
    },
    {
      "epoch": 102.34,
      "grad_norm": 1.6541175842285156,
      "learning_rate": 2.4469051884113446e-06,
      "loss": 0.4867,
      "step": 250010
    },
    {
      "epoch": 102.34,
      "grad_norm": 1.8397072553634644,
      "learning_rate": 2.4467467123438664e-06,
      "loss": 0.4692,
      "step": 250020
    },
    {
      "epoch": 102.35,
      "grad_norm": 1.998462200164795,
      "learning_rate": 2.446588235804576e-06,
      "loss": 0.4847,
      "step": 250030
    },
    {
      "epoch": 102.35,
      "grad_norm": 2.325735092163086,
      "learning_rate": 2.4464297587942e-06,
      "loss": 0.47,
      "step": 250040
    },
    {
      "epoch": 102.35,
      "grad_norm": 1.895353078842163,
      "learning_rate": 2.4462712813134637e-06,
      "loss": 0.4671,
      "step": 250050
    },
    {
      "epoch": 102.36,
      "grad_norm": 2.1799519062042236,
      "learning_rate": 2.4461128033630922e-06,
      "loss": 0.4581,
      "step": 250060
    },
    {
      "epoch": 102.36,
      "grad_norm": 1.9820390939712524,
      "learning_rate": 2.445954324943813e-06,
      "loss": 0.4855,
      "step": 250070
    },
    {
      "epoch": 102.37,
      "grad_norm": 2.0778889656066895,
      "learning_rate": 2.4457958460563526e-06,
      "loss": 0.4806,
      "step": 250080
    },
    {
      "epoch": 102.37,
      "grad_norm": 1.6126632690429688,
      "learning_rate": 2.4456373667014353e-06,
      "loss": 0.4871,
      "step": 250090
    },
    {
      "epoch": 102.37,
      "grad_norm": 1.8685674667358398,
      "learning_rate": 2.4454788868797877e-06,
      "loss": 0.4513,
      "step": 250100
    },
    {
      "epoch": 102.38,
      "grad_norm": 1.5781329870224,
      "learning_rate": 2.4453204065921364e-06,
      "loss": 0.4882,
      "step": 250110
    },
    {
      "epoch": 102.38,
      "grad_norm": 1.9864178895950317,
      "learning_rate": 2.4451619258392065e-06,
      "loss": 0.4937,
      "step": 250120
    },
    {
      "epoch": 102.39,
      "grad_norm": 1.791111707687378,
      "learning_rate": 2.445003444621724e-06,
      "loss": 0.4877,
      "step": 250130
    },
    {
      "epoch": 102.39,
      "grad_norm": 2.1466224193573,
      "learning_rate": 2.4448449629404155e-06,
      "loss": 0.4613,
      "step": 250140
    },
    {
      "epoch": 102.39,
      "grad_norm": 2.411208152770996,
      "learning_rate": 2.4446864807960053e-06,
      "loss": 0.4743,
      "step": 250150
    },
    {
      "epoch": 102.4,
      "grad_norm": 1.776105523109436,
      "learning_rate": 2.4445279981892223e-06,
      "loss": 0.4929,
      "step": 250160
    },
    {
      "epoch": 102.4,
      "grad_norm": 1.9015107154846191,
      "learning_rate": 2.4443695151207906e-06,
      "loss": 0.4691,
      "step": 250170
    },
    {
      "epoch": 102.41,
      "grad_norm": 2.4842233657836914,
      "learning_rate": 2.4442110315914367e-06,
      "loss": 0.4783,
      "step": 250180
    },
    {
      "epoch": 102.41,
      "grad_norm": 1.876639485359192,
      "learning_rate": 2.444052547601886e-06,
      "loss": 0.4746,
      "step": 250190
    },
    {
      "epoch": 102.42,
      "grad_norm": 1.978348731994629,
      "learning_rate": 2.4438940631528646e-06,
      "loss": 0.4646,
      "step": 250200
    },
    {
      "epoch": 102.42,
      "grad_norm": 1.6273436546325684,
      "learning_rate": 2.443735578245099e-06,
      "loss": 0.4795,
      "step": 250210
    },
    {
      "epoch": 102.42,
      "grad_norm": 1.8373991250991821,
      "learning_rate": 2.4435770928793148e-06,
      "loss": 0.4858,
      "step": 250220
    },
    {
      "epoch": 102.43,
      "grad_norm": 1.7727872133255005,
      "learning_rate": 2.4434186070562388e-06,
      "loss": 0.4781,
      "step": 250230
    },
    {
      "epoch": 102.43,
      "grad_norm": 1.6150676012039185,
      "learning_rate": 2.4432601207765955e-06,
      "loss": 0.4684,
      "step": 250240
    },
    {
      "epoch": 102.44,
      "grad_norm": 1.639402985572815,
      "learning_rate": 2.443101634041112e-06,
      "loss": 0.4927,
      "step": 250250
    },
    {
      "epoch": 102.44,
      "grad_norm": 2.2429375648498535,
      "learning_rate": 2.4429431468505142e-06,
      "loss": 0.4615,
      "step": 250260
    },
    {
      "epoch": 102.44,
      "grad_norm": 1.7815241813659668,
      "learning_rate": 2.442784659205528e-06,
      "loss": 0.4813,
      "step": 250270
    },
    {
      "epoch": 102.45,
      "grad_norm": 1.839970588684082,
      "learning_rate": 2.4426261711068793e-06,
      "loss": 0.4628,
      "step": 250280
    },
    {
      "epoch": 102.45,
      "grad_norm": 1.815385341644287,
      "learning_rate": 2.4424676825552942e-06,
      "loss": 0.4531,
      "step": 250290
    },
    {
      "epoch": 102.46,
      "grad_norm": 1.8455103635787964,
      "learning_rate": 2.4423091935514986e-06,
      "loss": 0.4858,
      "step": 250300
    },
    {
      "epoch": 102.46,
      "grad_norm": 1.8629372119903564,
      "learning_rate": 2.4421507040962187e-06,
      "loss": 0.4641,
      "step": 250310
    },
    {
      "epoch": 102.46,
      "grad_norm": 1.7784746885299683,
      "learning_rate": 2.4419922141901805e-06,
      "loss": 0.4739,
      "step": 250320
    },
    {
      "epoch": 102.47,
      "grad_norm": 2.371823787689209,
      "learning_rate": 2.44183372383411e-06,
      "loss": 0.4864,
      "step": 250330
    },
    {
      "epoch": 102.47,
      "grad_norm": 2.311647891998291,
      "learning_rate": 2.4416752330287323e-06,
      "loss": 0.5018,
      "step": 250340
    },
    {
      "epoch": 102.48,
      "grad_norm": 1.6924530267715454,
      "learning_rate": 2.441516741774775e-06,
      "loss": 0.481,
      "step": 250350
    },
    {
      "epoch": 102.48,
      "grad_norm": 2.2395966053009033,
      "learning_rate": 2.4413582500729634e-06,
      "loss": 0.4991,
      "step": 250360
    },
    {
      "epoch": 102.48,
      "grad_norm": 1.9661377668380737,
      "learning_rate": 2.4411997579240235e-06,
      "loss": 0.4795,
      "step": 250370
    },
    {
      "epoch": 102.49,
      "grad_norm": 1.9199206829071045,
      "learning_rate": 2.441041265328681e-06,
      "loss": 0.49,
      "step": 250380
    },
    {
      "epoch": 102.49,
      "grad_norm": 1.8912010192871094,
      "learning_rate": 2.4408827722876625e-06,
      "loss": 0.4782,
      "step": 250390
    },
    {
      "epoch": 102.5,
      "grad_norm": 1.9646738767623901,
      "learning_rate": 2.440724278801694e-06,
      "loss": 0.458,
      "step": 250400
    },
    {
      "epoch": 102.5,
      "grad_norm": 1.736262321472168,
      "learning_rate": 2.4405657848715016e-06,
      "loss": 0.4827,
      "step": 250410
    },
    {
      "epoch": 102.51,
      "grad_norm": 1.8966362476348877,
      "learning_rate": 2.440407290497811e-06,
      "loss": 0.4904,
      "step": 250420
    },
    {
      "epoch": 102.51,
      "grad_norm": 1.9220913648605347,
      "learning_rate": 2.4402487956813474e-06,
      "loss": 0.4437,
      "step": 250430
    },
    {
      "epoch": 102.51,
      "grad_norm": 2.5406627655029297,
      "learning_rate": 2.440090300422838e-06,
      "loss": 0.4789,
      "step": 250440
    },
    {
      "epoch": 102.52,
      "grad_norm": 2.039846897125244,
      "learning_rate": 2.439931804723009e-06,
      "loss": 0.4732,
      "step": 250450
    },
    {
      "epoch": 102.52,
      "grad_norm": 1.7395083904266357,
      "learning_rate": 2.439773308582586e-06,
      "loss": 0.4781,
      "step": 250460
    },
    {
      "epoch": 102.53,
      "grad_norm": 2.5725927352905273,
      "learning_rate": 2.4396148120022957e-06,
      "loss": 0.4739,
      "step": 250470
    },
    {
      "epoch": 102.53,
      "grad_norm": 2.0897603034973145,
      "learning_rate": 2.439456314982863e-06,
      "loss": 0.4815,
      "step": 250480
    },
    {
      "epoch": 102.53,
      "grad_norm": 1.5431827306747437,
      "learning_rate": 2.4392978175250144e-06,
      "loss": 0.4659,
      "step": 250490
    },
    {
      "epoch": 102.54,
      "grad_norm": 1.8025398254394531,
      "learning_rate": 2.439139319629476e-06,
      "loss": 0.4815,
      "step": 250500
    },
    {
      "epoch": 102.54,
      "grad_norm": 2.0938854217529297,
      "learning_rate": 2.438980821296974e-06,
      "loss": 0.502,
      "step": 250510
    },
    {
      "epoch": 102.55,
      "grad_norm": 2.127601385116577,
      "learning_rate": 2.438822322528235e-06,
      "loss": 0.4903,
      "step": 250520
    },
    {
      "epoch": 102.55,
      "grad_norm": 1.7106022834777832,
      "learning_rate": 2.438663823323983e-06,
      "loss": 0.4608,
      "step": 250530
    },
    {
      "epoch": 102.55,
      "grad_norm": 2.416025161743164,
      "learning_rate": 2.438505323684947e-06,
      "loss": 0.4789,
      "step": 250540
    },
    {
      "epoch": 102.56,
      "grad_norm": 2.1855809688568115,
      "learning_rate": 2.43834682361185e-06,
      "loss": 0.4659,
      "step": 250550
    },
    {
      "epoch": 102.56,
      "grad_norm": 1.7053022384643555,
      "learning_rate": 2.4381883231054203e-06,
      "loss": 0.4945,
      "step": 250560
    },
    {
      "epoch": 102.57,
      "grad_norm": 1.8545339107513428,
      "learning_rate": 2.4380298221663837e-06,
      "loss": 0.4824,
      "step": 250570
    },
    {
      "epoch": 102.57,
      "grad_norm": 2.0208740234375,
      "learning_rate": 2.4378713207954653e-06,
      "loss": 0.473,
      "step": 250580
    },
    {
      "epoch": 102.57,
      "grad_norm": 1.9082942008972168,
      "learning_rate": 2.437712818993392e-06,
      "loss": 0.4825,
      "step": 250590
    },
    {
      "epoch": 102.58,
      "grad_norm": 2.102494478225708,
      "learning_rate": 2.4375543167608893e-06,
      "loss": 0.482,
      "step": 250600
    },
    {
      "epoch": 102.58,
      "grad_norm": 1.9466447830200195,
      "learning_rate": 2.4373958140986844e-06,
      "loss": 0.4809,
      "step": 250610
    },
    {
      "epoch": 102.59,
      "grad_norm": 1.9301601648330688,
      "learning_rate": 2.437237311007502e-06,
      "loss": 0.4654,
      "step": 250620
    },
    {
      "epoch": 102.59,
      "grad_norm": 1.802851915359497,
      "learning_rate": 2.437078807488068e-06,
      "loss": 0.4845,
      "step": 250630
    },
    {
      "epoch": 102.6,
      "grad_norm": 2.228917360305786,
      "learning_rate": 2.4369203035411096e-06,
      "loss": 0.4718,
      "step": 250640
    },
    {
      "epoch": 102.6,
      "grad_norm": 2.012190580368042,
      "learning_rate": 2.4367617991673527e-06,
      "loss": 0.4832,
      "step": 250650
    },
    {
      "epoch": 102.6,
      "grad_norm": 2.1140224933624268,
      "learning_rate": 2.436603294367523e-06,
      "loss": 0.4744,
      "step": 250660
    },
    {
      "epoch": 102.61,
      "grad_norm": 2.13828182220459,
      "learning_rate": 2.4364447891423467e-06,
      "loss": 0.5085,
      "step": 250670
    },
    {
      "epoch": 102.61,
      "grad_norm": 2.3269784450531006,
      "learning_rate": 2.4362862834925503e-06,
      "loss": 0.4719,
      "step": 250680
    },
    {
      "epoch": 102.62,
      "grad_norm": 1.9625986814498901,
      "learning_rate": 2.4361277774188596e-06,
      "loss": 0.475,
      "step": 250690
    },
    {
      "epoch": 102.62,
      "grad_norm": 2.106264352798462,
      "learning_rate": 2.4359692709220006e-06,
      "loss": 0.4513,
      "step": 250700
    },
    {
      "epoch": 102.62,
      "grad_norm": 1.9441423416137695,
      "learning_rate": 2.4358107640026988e-06,
      "loss": 0.4704,
      "step": 250710
    },
    {
      "epoch": 102.63,
      "grad_norm": 2.2272109985351562,
      "learning_rate": 2.435652256661681e-06,
      "loss": 0.4515,
      "step": 250720
    },
    {
      "epoch": 102.63,
      "grad_norm": 1.6919530630111694,
      "learning_rate": 2.4354937488996727e-06,
      "loss": 0.4813,
      "step": 250730
    },
    {
      "epoch": 102.64,
      "grad_norm": 1.772813320159912,
      "learning_rate": 2.4353352407174015e-06,
      "loss": 0.481,
      "step": 250740
    },
    {
      "epoch": 102.64,
      "grad_norm": 1.7842257022857666,
      "learning_rate": 2.4351767321155923e-06,
      "loss": 0.4794,
      "step": 250750
    },
    {
      "epoch": 102.64,
      "grad_norm": 2.2007713317871094,
      "learning_rate": 2.4350182230949716e-06,
      "loss": 0.4769,
      "step": 250760
    },
    {
      "epoch": 102.65,
      "grad_norm": 2.0298402309417725,
      "learning_rate": 2.4348597136562647e-06,
      "loss": 0.4853,
      "step": 250770
    },
    {
      "epoch": 102.65,
      "grad_norm": 2.007150173187256,
      "learning_rate": 2.4347012038001988e-06,
      "loss": 0.4642,
      "step": 250780
    },
    {
      "epoch": 102.66,
      "grad_norm": 1.7676799297332764,
      "learning_rate": 2.4345426935274994e-06,
      "loss": 0.4841,
      "step": 250790
    },
    {
      "epoch": 102.66,
      "grad_norm": 1.5305769443511963,
      "learning_rate": 2.4343841828388925e-06,
      "loss": 0.4827,
      "step": 250800
    },
    {
      "epoch": 102.66,
      "grad_norm": 1.6545896530151367,
      "learning_rate": 2.434225671735104e-06,
      "loss": 0.4659,
      "step": 250810
    },
    {
      "epoch": 102.67,
      "grad_norm": 2.0085763931274414,
      "learning_rate": 2.434067160216861e-06,
      "loss": 0.4668,
      "step": 250820
    },
    {
      "epoch": 102.67,
      "grad_norm": 1.9601508378982544,
      "learning_rate": 2.4339086482848895e-06,
      "loss": 0.4829,
      "step": 250830
    },
    {
      "epoch": 102.68,
      "grad_norm": 1.8331725597381592,
      "learning_rate": 2.4337501359399147e-06,
      "loss": 0.471,
      "step": 250840
    },
    {
      "epoch": 102.68,
      "grad_norm": 1.6211992502212524,
      "learning_rate": 2.4335916231826636e-06,
      "loss": 0.471,
      "step": 250850
    },
    {
      "epoch": 102.69,
      "grad_norm": 1.7863092422485352,
      "learning_rate": 2.4334331100138616e-06,
      "loss": 0.4631,
      "step": 250860
    },
    {
      "epoch": 102.69,
      "grad_norm": 2.112668514251709,
      "learning_rate": 2.4332745964342346e-06,
      "loss": 0.4695,
      "step": 250870
    },
    {
      "epoch": 102.69,
      "grad_norm": 2.1143603324890137,
      "learning_rate": 2.4331160824445103e-06,
      "loss": 0.4728,
      "step": 250880
    },
    {
      "epoch": 102.7,
      "grad_norm": 2.3610379695892334,
      "learning_rate": 2.4329575680454137e-06,
      "loss": 0.4848,
      "step": 250890
    },
    {
      "epoch": 102.7,
      "grad_norm": 1.7546956539154053,
      "learning_rate": 2.4327990532376703e-06,
      "loss": 0.4863,
      "step": 250900
    },
    {
      "epoch": 102.71,
      "grad_norm": 2.126260995864868,
      "learning_rate": 2.4326405380220077e-06,
      "loss": 0.4622,
      "step": 250910
    },
    {
      "epoch": 102.71,
      "grad_norm": 2.8273744583129883,
      "learning_rate": 2.432482022399151e-06,
      "loss": 0.4683,
      "step": 250920
    },
    {
      "epoch": 102.71,
      "grad_norm": 1.99750554561615,
      "learning_rate": 2.4323235063698268e-06,
      "loss": 0.4745,
      "step": 250930
    },
    {
      "epoch": 102.72,
      "grad_norm": 2.1593616008758545,
      "learning_rate": 2.4321649899347607e-06,
      "loss": 0.4857,
      "step": 250940
    },
    {
      "epoch": 102.72,
      "grad_norm": 1.6840858459472656,
      "learning_rate": 2.4320064730946795e-06,
      "loss": 0.4672,
      "step": 250950
    },
    {
      "epoch": 102.73,
      "grad_norm": 1.5855045318603516,
      "learning_rate": 2.431847955850309e-06,
      "loss": 0.4649,
      "step": 250960
    },
    {
      "epoch": 102.73,
      "grad_norm": 1.9925310611724854,
      "learning_rate": 2.4316894382023756e-06,
      "loss": 0.4652,
      "step": 250970
    },
    {
      "epoch": 102.73,
      "grad_norm": 1.6073600053787231,
      "learning_rate": 2.431530920151605e-06,
      "loss": 0.4604,
      "step": 250980
    },
    {
      "epoch": 102.74,
      "grad_norm": 1.9082103967666626,
      "learning_rate": 2.431372401698724e-06,
      "loss": 0.4673,
      "step": 250990
    },
    {
      "epoch": 102.74,
      "grad_norm": 1.82425057888031,
      "learning_rate": 2.431213882844458e-06,
      "loss": 0.4706,
      "step": 251000
    },
    {
      "epoch": 102.75,
      "grad_norm": 2.0387368202209473,
      "learning_rate": 2.4310553635895335e-06,
      "loss": 0.4611,
      "step": 251010
    },
    {
      "epoch": 102.75,
      "grad_norm": 1.778163194656372,
      "learning_rate": 2.4308968439346767e-06,
      "loss": 0.4787,
      "step": 251020
    },
    {
      "epoch": 102.75,
      "grad_norm": 1.781761646270752,
      "learning_rate": 2.4307383238806132e-06,
      "loss": 0.4776,
      "step": 251030
    },
    {
      "epoch": 102.76,
      "grad_norm": 1.8896644115447998,
      "learning_rate": 2.4305798034280707e-06,
      "loss": 0.4967,
      "step": 251040
    },
    {
      "epoch": 102.76,
      "grad_norm": 1.943121075630188,
      "learning_rate": 2.4304212825777743e-06,
      "loss": 0.4818,
      "step": 251050
    },
    {
      "epoch": 102.77,
      "grad_norm": 1.8370145559310913,
      "learning_rate": 2.4302627613304493e-06,
      "loss": 0.4912,
      "step": 251060
    },
    {
      "epoch": 102.77,
      "grad_norm": 1.7027310132980347,
      "learning_rate": 2.4301042396868235e-06,
      "loss": 0.4765,
      "step": 251070
    },
    {
      "epoch": 102.78,
      "grad_norm": 2.0168204307556152,
      "learning_rate": 2.429945717647622e-06,
      "loss": 0.4627,
      "step": 251080
    },
    {
      "epoch": 102.78,
      "grad_norm": 2.016592025756836,
      "learning_rate": 2.429787195213571e-06,
      "loss": 0.4763,
      "step": 251090
    },
    {
      "epoch": 102.78,
      "grad_norm": 1.589185118675232,
      "learning_rate": 2.4296286723853964e-06,
      "loss": 0.4876,
      "step": 251100
    },
    {
      "epoch": 102.79,
      "grad_norm": 1.8427788019180298,
      "learning_rate": 2.4294701491638257e-06,
      "loss": 0.4609,
      "step": 251110
    },
    {
      "epoch": 102.79,
      "grad_norm": 2.1155481338500977,
      "learning_rate": 2.4293116255495845e-06,
      "loss": 0.4919,
      "step": 251120
    },
    {
      "epoch": 102.8,
      "grad_norm": 2.331481695175171,
      "learning_rate": 2.4291531015433987e-06,
      "loss": 0.4568,
      "step": 251130
    },
    {
      "epoch": 102.8,
      "grad_norm": 1.5926623344421387,
      "learning_rate": 2.4289945771459943e-06,
      "loss": 0.486,
      "step": 251140
    },
    {
      "epoch": 102.8,
      "grad_norm": 2.1310818195343018,
      "learning_rate": 2.428836052358098e-06,
      "loss": 0.4708,
      "step": 251150
    },
    {
      "epoch": 102.81,
      "grad_norm": 1.6830083131790161,
      "learning_rate": 2.428677527180435e-06,
      "loss": 0.4682,
      "step": 251160
    },
    {
      "epoch": 102.81,
      "grad_norm": 1.855389952659607,
      "learning_rate": 2.428519001613733e-06,
      "loss": 0.4671,
      "step": 251170
    },
    {
      "epoch": 102.82,
      "grad_norm": 1.958168625831604,
      "learning_rate": 2.428360475658716e-06,
      "loss": 0.4738,
      "step": 251180
    },
    {
      "epoch": 102.82,
      "grad_norm": 2.8612985610961914,
      "learning_rate": 2.4282019493161128e-06,
      "loss": 0.4676,
      "step": 251190
    },
    {
      "epoch": 102.82,
      "grad_norm": 2.110680341720581,
      "learning_rate": 2.428043422586648e-06,
      "loss": 0.4692,
      "step": 251200
    },
    {
      "epoch": 102.83,
      "grad_norm": 2.036726236343384,
      "learning_rate": 2.4278848954710478e-06,
      "loss": 0.4844,
      "step": 251210
    },
    {
      "epoch": 102.83,
      "grad_norm": 9.193514823913574,
      "learning_rate": 2.4277263679700383e-06,
      "loss": 0.4679,
      "step": 251220
    },
    {
      "epoch": 102.84,
      "grad_norm": 1.9114203453063965,
      "learning_rate": 2.4275678400843468e-06,
      "loss": 0.4868,
      "step": 251230
    },
    {
      "epoch": 102.84,
      "grad_norm": 2.0544044971466064,
      "learning_rate": 2.4274093118146986e-06,
      "loss": 0.4991,
      "step": 251240
    },
    {
      "epoch": 102.84,
      "grad_norm": 2.0911216735839844,
      "learning_rate": 2.4272507831618196e-06,
      "loss": 0.4746,
      "step": 251250
    },
    {
      "epoch": 102.85,
      "grad_norm": 2.0979764461517334,
      "learning_rate": 2.4270922541264367e-06,
      "loss": 0.4867,
      "step": 251260
    },
    {
      "epoch": 102.85,
      "grad_norm": 1.8422086238861084,
      "learning_rate": 2.4269337247092765e-06,
      "loss": 0.4603,
      "step": 251270
    },
    {
      "epoch": 102.86,
      "grad_norm": 2.0096435546875,
      "learning_rate": 2.4267751949110638e-06,
      "loss": 0.4678,
      "step": 251280
    },
    {
      "epoch": 102.86,
      "grad_norm": 2.2147104740142822,
      "learning_rate": 2.426616664732525e-06,
      "loss": 0.466,
      "step": 251290
    },
    {
      "epoch": 102.87,
      "grad_norm": 1.9387789964675903,
      "learning_rate": 2.4264581341743884e-06,
      "loss": 0.4843,
      "step": 251300
    },
    {
      "epoch": 102.87,
      "grad_norm": 1.8650343418121338,
      "learning_rate": 2.4262996032373775e-06,
      "loss": 0.4824,
      "step": 251310
    },
    {
      "epoch": 102.87,
      "grad_norm": 1.937400221824646,
      "learning_rate": 2.4261410719222198e-06,
      "loss": 0.4593,
      "step": 251320
    },
    {
      "epoch": 102.88,
      "grad_norm": 2.327615737915039,
      "learning_rate": 2.425982540229641e-06,
      "loss": 0.488,
      "step": 251330
    },
    {
      "epoch": 102.88,
      "grad_norm": 1.703377366065979,
      "learning_rate": 2.425824008160369e-06,
      "loss": 0.4675,
      "step": 251340
    },
    {
      "epoch": 102.89,
      "grad_norm": 1.6632461547851562,
      "learning_rate": 2.425665475715128e-06,
      "loss": 0.4696,
      "step": 251350
    },
    {
      "epoch": 102.89,
      "grad_norm": 2.1304852962493896,
      "learning_rate": 2.425506942894645e-06,
      "loss": 0.4704,
      "step": 251360
    },
    {
      "epoch": 102.89,
      "grad_norm": 2.285996437072754,
      "learning_rate": 2.425348409699646e-06,
      "loss": 0.4883,
      "step": 251370
    },
    {
      "epoch": 102.9,
      "grad_norm": 9.996339797973633,
      "learning_rate": 2.4251898761308566e-06,
      "loss": 0.4682,
      "step": 251380
    },
    {
      "epoch": 102.9,
      "grad_norm": 2.2246100902557373,
      "learning_rate": 2.4250313421890044e-06,
      "loss": 0.468,
      "step": 251390
    },
    {
      "epoch": 102.91,
      "grad_norm": 2.272644519805908,
      "learning_rate": 2.4248728078748144e-06,
      "loss": 0.4624,
      "step": 251400
    },
    {
      "epoch": 102.91,
      "grad_norm": 2.1377408504486084,
      "learning_rate": 2.4247142731890144e-06,
      "loss": 0.5038,
      "step": 251410
    },
    {
      "epoch": 102.91,
      "grad_norm": 1.7978193759918213,
      "learning_rate": 2.4245557381323288e-06,
      "loss": 0.484,
      "step": 251420
    },
    {
      "epoch": 102.92,
      "grad_norm": 1.7323979139328003,
      "learning_rate": 2.4243972027054855e-06,
      "loss": 0.485,
      "step": 251430
    },
    {
      "epoch": 102.92,
      "grad_norm": 1.6451133489608765,
      "learning_rate": 2.4242386669092093e-06,
      "loss": 0.4521,
      "step": 251440
    },
    {
      "epoch": 102.93,
      "grad_norm": 1.6980518102645874,
      "learning_rate": 2.424080130744227e-06,
      "loss": 0.474,
      "step": 251450
    },
    {
      "epoch": 102.93,
      "grad_norm": 2.0594751834869385,
      "learning_rate": 2.4239215942112644e-06,
      "loss": 0.477,
      "step": 251460
    },
    {
      "epoch": 102.93,
      "grad_norm": 1.643685221672058,
      "learning_rate": 2.423763057311048e-06,
      "loss": 0.4648,
      "step": 251470
    },
    {
      "epoch": 102.94,
      "grad_norm": 1.5819389820098877,
      "learning_rate": 2.423604520044305e-06,
      "loss": 0.4831,
      "step": 251480
    },
    {
      "epoch": 102.94,
      "grad_norm": 1.7330741882324219,
      "learning_rate": 2.42344598241176e-06,
      "loss": 0.4649,
      "step": 251490
    },
    {
      "epoch": 102.95,
      "grad_norm": 1.8885663747787476,
      "learning_rate": 2.423287444414141e-06,
      "loss": 0.4696,
      "step": 251500
    },
    {
      "epoch": 102.95,
      "grad_norm": 1.9761804342269897,
      "learning_rate": 2.4231289060521728e-06,
      "loss": 0.4491,
      "step": 251510
    },
    {
      "epoch": 102.96,
      "grad_norm": 2.298363447189331,
      "learning_rate": 2.422970367326582e-06,
      "loss": 0.475,
      "step": 251520
    },
    {
      "epoch": 102.96,
      "grad_norm": 1.847923994064331,
      "learning_rate": 2.422811828238095e-06,
      "loss": 0.4878,
      "step": 251530
    },
    {
      "epoch": 102.96,
      "grad_norm": 1.8148252964019775,
      "learning_rate": 2.4226532887874383e-06,
      "loss": 0.4777,
      "step": 251540
    },
    {
      "epoch": 102.97,
      "grad_norm": 1.7800238132476807,
      "learning_rate": 2.4224947489753374e-06,
      "loss": 0.4754,
      "step": 251550
    },
    {
      "epoch": 102.97,
      "grad_norm": 1.949209451675415,
      "learning_rate": 2.42233620880252e-06,
      "loss": 0.4842,
      "step": 251560
    },
    {
      "epoch": 102.98,
      "grad_norm": 1.7531949281692505,
      "learning_rate": 2.4221776682697103e-06,
      "loss": 0.4904,
      "step": 251570
    },
    {
      "epoch": 102.98,
      "grad_norm": 1.969339370727539,
      "learning_rate": 2.4220191273776355e-06,
      "loss": 0.4826,
      "step": 251580
    },
    {
      "epoch": 102.98,
      "grad_norm": 1.9543501138687134,
      "learning_rate": 2.4218605861270223e-06,
      "loss": 0.4765,
      "step": 251590
    },
    {
      "epoch": 102.99,
      "grad_norm": 2.1614573001861572,
      "learning_rate": 2.4217020445185966e-06,
      "loss": 0.4836,
      "step": 251600
    },
    {
      "epoch": 102.99,
      "grad_norm": 2.0127429962158203,
      "learning_rate": 2.421543502553085e-06,
      "loss": 0.4562,
      "step": 251610
    },
    {
      "epoch": 103.0,
      "grad_norm": 1.8905842304229736,
      "learning_rate": 2.421384960231213e-06,
      "loss": 0.4683,
      "step": 251620
    },
    {
      "epoch": 103.0,
      "eval_loss": 0.4757035970687866,
      "eval_runtime": 53.1536,
      "eval_samples_per_second": 64.887,
      "eval_steps_per_second": 8.127,
      "step": 251629
    },
    {
      "epoch": 103.0,
      "grad_norm": 1.6533409357070923,
      "learning_rate": 2.421226417553708e-06,
      "loss": 0.4619,
      "step": 251630
    },
    {
      "epoch": 103.0,
      "grad_norm": 1.9176342487335205,
      "learning_rate": 2.421067874521295e-06,
      "loss": 0.462,
      "step": 251640
    },
    {
      "epoch": 103.01,
      "grad_norm": 1.7518576383590698,
      "learning_rate": 2.4209093311347003e-06,
      "loss": 0.4614,
      "step": 251650
    },
    {
      "epoch": 103.01,
      "grad_norm": 1.5509716272354126,
      "learning_rate": 2.420750787394651e-06,
      "loss": 0.4782,
      "step": 251660
    },
    {
      "epoch": 103.02,
      "grad_norm": 1.6494414806365967,
      "learning_rate": 2.420592243301873e-06,
      "loss": 0.4731,
      "step": 251670
    },
    {
      "epoch": 103.02,
      "grad_norm": 1.6247386932373047,
      "learning_rate": 2.420433698857093e-06,
      "loss": 0.4809,
      "step": 251680
    },
    {
      "epoch": 103.02,
      "grad_norm": 2.3544180393218994,
      "learning_rate": 2.420275154061036e-06,
      "loss": 0.499,
      "step": 251690
    },
    {
      "epoch": 103.03,
      "grad_norm": 1.7137631177902222,
      "learning_rate": 2.4201166089144305e-06,
      "loss": 0.4859,
      "step": 251700
    },
    {
      "epoch": 103.03,
      "grad_norm": 1.984848141670227,
      "learning_rate": 2.4199580634180003e-06,
      "loss": 0.4655,
      "step": 251710
    },
    {
      "epoch": 103.04,
      "grad_norm": 1.6302675008773804,
      "learning_rate": 2.4197995175724727e-06,
      "loss": 0.4727,
      "step": 251720
    },
    {
      "epoch": 103.04,
      "grad_norm": 1.878587245941162,
      "learning_rate": 2.4196409713785748e-06,
      "loss": 0.4793,
      "step": 251730
    },
    {
      "epoch": 103.05,
      "grad_norm": 1.7805362939834595,
      "learning_rate": 2.4194824248370313e-06,
      "loss": 0.4673,
      "step": 251740
    },
    {
      "epoch": 103.05,
      "grad_norm": 1.9359174966812134,
      "learning_rate": 2.4193238779485697e-06,
      "loss": 0.4672,
      "step": 251750
    },
    {
      "epoch": 103.05,
      "grad_norm": 2.199097156524658,
      "learning_rate": 2.4191653307139152e-06,
      "loss": 0.4577,
      "step": 251760
    },
    {
      "epoch": 103.06,
      "grad_norm": 1.6949377059936523,
      "learning_rate": 2.419006783133796e-06,
      "loss": 0.4474,
      "step": 251770
    },
    {
      "epoch": 103.06,
      "grad_norm": 1.9907561540603638,
      "learning_rate": 2.4188482352089365e-06,
      "loss": 0.5108,
      "step": 251780
    },
    {
      "epoch": 103.07,
      "grad_norm": 2.1075029373168945,
      "learning_rate": 2.4186896869400633e-06,
      "loss": 0.4718,
      "step": 251790
    },
    {
      "epoch": 103.07,
      "grad_norm": 1.8406912088394165,
      "learning_rate": 2.4185311383279033e-06,
      "loss": 0.4691,
      "step": 251800
    },
    {
      "epoch": 103.07,
      "grad_norm": 1.8911302089691162,
      "learning_rate": 2.4183725893731824e-06,
      "loss": 0.4674,
      "step": 251810
    },
    {
      "epoch": 103.08,
      "grad_norm": 1.6128207445144653,
      "learning_rate": 2.4182140400766274e-06,
      "loss": 0.4713,
      "step": 251820
    },
    {
      "epoch": 103.08,
      "grad_norm": 1.8258000612258911,
      "learning_rate": 2.4180554904389637e-06,
      "loss": 0.4762,
      "step": 251830
    },
    {
      "epoch": 103.09,
      "grad_norm": 1.8327099084854126,
      "learning_rate": 2.417896940460918e-06,
      "loss": 0.4728,
      "step": 251840
    },
    {
      "epoch": 103.09,
      "grad_norm": 2.362931251525879,
      "learning_rate": 2.4177383901432175e-06,
      "loss": 0.4906,
      "step": 251850
    },
    {
      "epoch": 103.09,
      "grad_norm": 1.5406315326690674,
      "learning_rate": 2.4175798394865867e-06,
      "loss": 0.4826,
      "step": 251860
    },
    {
      "epoch": 103.1,
      "grad_norm": 2.0648441314697266,
      "learning_rate": 2.4174212884917527e-06,
      "loss": 0.4664,
      "step": 251870
    },
    {
      "epoch": 103.1,
      "grad_norm": 1.5421799421310425,
      "learning_rate": 2.417262737159443e-06,
      "loss": 0.4818,
      "step": 251880
    },
    {
      "epoch": 103.11,
      "grad_norm": 2.065387010574341,
      "learning_rate": 2.4171041854903816e-06,
      "loss": 0.4715,
      "step": 251890
    },
    {
      "epoch": 103.11,
      "grad_norm": 1.6998828649520874,
      "learning_rate": 2.416945633485297e-06,
      "loss": 0.4705,
      "step": 251900
    },
    {
      "epoch": 103.12,
      "grad_norm": 2.4295153617858887,
      "learning_rate": 2.4167870811449137e-06,
      "loss": 0.4571,
      "step": 251910
    },
    {
      "epoch": 103.12,
      "grad_norm": 1.925532341003418,
      "learning_rate": 2.41662852846996e-06,
      "loss": 0.4711,
      "step": 251920
    },
    {
      "epoch": 103.12,
      "grad_norm": 1.7731355428695679,
      "learning_rate": 2.416469975461161e-06,
      "loss": 0.4521,
      "step": 251930
    },
    {
      "epoch": 103.13,
      "grad_norm": 2.2088639736175537,
      "learning_rate": 2.416311422119242e-06,
      "loss": 0.4612,
      "step": 251940
    },
    {
      "epoch": 103.13,
      "grad_norm": 1.8770817518234253,
      "learning_rate": 2.4161528684449303e-06,
      "loss": 0.4774,
      "step": 251950
    },
    {
      "epoch": 103.14,
      "grad_norm": 1.964169979095459,
      "learning_rate": 2.4159943144389525e-06,
      "loss": 0.486,
      "step": 251960
    },
    {
      "epoch": 103.14,
      "grad_norm": 1.975965976715088,
      "learning_rate": 2.415835760102035e-06,
      "loss": 0.4658,
      "step": 251970
    },
    {
      "epoch": 103.14,
      "grad_norm": 1.8973711729049683,
      "learning_rate": 2.4156772054349038e-06,
      "loss": 0.4934,
      "step": 251980
    },
    {
      "epoch": 103.15,
      "grad_norm": 2.0454888343811035,
      "learning_rate": 2.4155186504382854e-06,
      "loss": 0.4878,
      "step": 251990
    },
    {
      "epoch": 103.15,
      "grad_norm": 2.6337685585021973,
      "learning_rate": 2.4153600951129057e-06,
      "loss": 0.4714,
      "step": 252000
    },
    {
      "epoch": 103.16,
      "grad_norm": 1.576022982597351,
      "learning_rate": 2.415201539459491e-06,
      "loss": 0.4857,
      "step": 252010
    },
    {
      "epoch": 103.16,
      "grad_norm": 1.9593050479888916,
      "learning_rate": 2.4150429834787684e-06,
      "loss": 0.4703,
      "step": 252020
    },
    {
      "epoch": 103.16,
      "grad_norm": 1.4630753993988037,
      "learning_rate": 2.414884427171463e-06,
      "loss": 0.4613,
      "step": 252030
    },
    {
      "epoch": 103.17,
      "grad_norm": 2.4205374717712402,
      "learning_rate": 2.4147258705383017e-06,
      "loss": 0.4848,
      "step": 252040
    },
    {
      "epoch": 103.17,
      "grad_norm": 2.265425443649292,
      "learning_rate": 2.4145673135800118e-06,
      "loss": 0.4843,
      "step": 252050
    },
    {
      "epoch": 103.18,
      "grad_norm": 2.3236334323883057,
      "learning_rate": 2.414408756297319e-06,
      "loss": 0.4584,
      "step": 252060
    },
    {
      "epoch": 103.18,
      "grad_norm": 1.8324021100997925,
      "learning_rate": 2.414250198690948e-06,
      "loss": 0.4829,
      "step": 252070
    },
    {
      "epoch": 103.18,
      "grad_norm": 1.863401174545288,
      "learning_rate": 2.4140916407616275e-06,
      "loss": 0.4751,
      "step": 252080
    },
    {
      "epoch": 103.19,
      "grad_norm": 2.012326717376709,
      "learning_rate": 2.4139330825100826e-06,
      "loss": 0.462,
      "step": 252090
    },
    {
      "epoch": 103.19,
      "grad_norm": 2.0422866344451904,
      "learning_rate": 2.41377452393704e-06,
      "loss": 0.459,
      "step": 252100
    },
    {
      "epoch": 103.2,
      "grad_norm": 2.688615322113037,
      "learning_rate": 2.4136159650432256e-06,
      "loss": 0.4718,
      "step": 252110
    },
    {
      "epoch": 103.2,
      "grad_norm": 1.7259944677352905,
      "learning_rate": 2.413457405829366e-06,
      "loss": 0.4744,
      "step": 252120
    },
    {
      "epoch": 103.21,
      "grad_norm": 2.1395955085754395,
      "learning_rate": 2.4132988462961876e-06,
      "loss": 0.4705,
      "step": 252130
    },
    {
      "epoch": 103.21,
      "grad_norm": 1.5941922664642334,
      "learning_rate": 2.413140286444417e-06,
      "loss": 0.4767,
      "step": 252140
    },
    {
      "epoch": 103.21,
      "grad_norm": 1.9031260013580322,
      "learning_rate": 2.41298172627478e-06,
      "loss": 0.4867,
      "step": 252150
    },
    {
      "epoch": 103.22,
      "grad_norm": 1.8408986330032349,
      "learning_rate": 2.4128231657880033e-06,
      "loss": 0.4777,
      "step": 252160
    },
    {
      "epoch": 103.22,
      "grad_norm": 1.9358389377593994,
      "learning_rate": 2.4126646049848136e-06,
      "loss": 0.4475,
      "step": 252170
    },
    {
      "epoch": 103.23,
      "grad_norm": 1.874059796333313,
      "learning_rate": 2.412506043865936e-06,
      "loss": 0.4691,
      "step": 252180
    },
    {
      "epoch": 103.23,
      "grad_norm": 2.1685733795166016,
      "learning_rate": 2.4123474824320976e-06,
      "loss": 0.4881,
      "step": 252190
    },
    {
      "epoch": 103.23,
      "grad_norm": 2.136596202850342,
      "learning_rate": 2.4121889206840253e-06,
      "loss": 0.4752,
      "step": 252200
    },
    {
      "epoch": 103.24,
      "grad_norm": 1.8693389892578125,
      "learning_rate": 2.412030358622445e-06,
      "loss": 0.4884,
      "step": 252210
    },
    {
      "epoch": 103.24,
      "grad_norm": 2.0093469619750977,
      "learning_rate": 2.4118717962480827e-06,
      "loss": 0.5098,
      "step": 252220
    },
    {
      "epoch": 103.25,
      "grad_norm": 1.795762300491333,
      "learning_rate": 2.4117132335616643e-06,
      "loss": 0.4599,
      "step": 252230
    },
    {
      "epoch": 103.25,
      "grad_norm": 2.2262790203094482,
      "learning_rate": 2.4115546705639173e-06,
      "loss": 0.4719,
      "step": 252240
    },
    {
      "epoch": 103.25,
      "grad_norm": 2.0169732570648193,
      "learning_rate": 2.411396107255568e-06,
      "loss": 0.4635,
      "step": 252250
    },
    {
      "epoch": 103.26,
      "grad_norm": 1.9815579652786255,
      "learning_rate": 2.411237543637342e-06,
      "loss": 0.4629,
      "step": 252260
    },
    {
      "epoch": 103.26,
      "grad_norm": 1.6343826055526733,
      "learning_rate": 2.411078979709966e-06,
      "loss": 0.4911,
      "step": 252270
    },
    {
      "epoch": 103.27,
      "grad_norm": 1.8869669437408447,
      "learning_rate": 2.410920415474167e-06,
      "loss": 0.4899,
      "step": 252280
    },
    {
      "epoch": 103.27,
      "grad_norm": 1.8311070203781128,
      "learning_rate": 2.41076185093067e-06,
      "loss": 0.4608,
      "step": 252290
    },
    {
      "epoch": 103.27,
      "grad_norm": 1.7144501209259033,
      "learning_rate": 2.4106032860802027e-06,
      "loss": 0.4655,
      "step": 252300
    },
    {
      "epoch": 103.28,
      "grad_norm": 2.46459698677063,
      "learning_rate": 2.4104447209234906e-06,
      "loss": 0.4622,
      "step": 252310
    },
    {
      "epoch": 103.28,
      "grad_norm": 1.81931734085083,
      "learning_rate": 2.4102861554612597e-06,
      "loss": 0.4654,
      "step": 252320
    },
    {
      "epoch": 103.29,
      "grad_norm": 2.1336746215820312,
      "learning_rate": 2.410127589694237e-06,
      "loss": 0.4799,
      "step": 252330
    },
    {
      "epoch": 103.29,
      "grad_norm": 1.6519651412963867,
      "learning_rate": 2.4099690236231496e-06,
      "loss": 0.4811,
      "step": 252340
    },
    {
      "epoch": 103.3,
      "grad_norm": 2.408994674682617,
      "learning_rate": 2.409810457248723e-06,
      "loss": 0.4836,
      "step": 252350
    },
    {
      "epoch": 103.3,
      "grad_norm": 1.6102826595306396,
      "learning_rate": 2.409651890571684e-06,
      "loss": 0.4725,
      "step": 252360
    },
    {
      "epoch": 103.3,
      "grad_norm": 1.7373907566070557,
      "learning_rate": 2.409493323592758e-06,
      "loss": 0.4651,
      "step": 252370
    },
    {
      "epoch": 103.31,
      "grad_norm": 2.244976043701172,
      "learning_rate": 2.409334756312672e-06,
      "loss": 0.5038,
      "step": 252380
    },
    {
      "epoch": 103.31,
      "grad_norm": 1.8867214918136597,
      "learning_rate": 2.409176188732152e-06,
      "loss": 0.4573,
      "step": 252390
    },
    {
      "epoch": 103.32,
      "grad_norm": 2.2538256645202637,
      "learning_rate": 2.4090176208519255e-06,
      "loss": 0.4778,
      "step": 252400
    },
    {
      "epoch": 103.32,
      "grad_norm": 1.8622492551803589,
      "learning_rate": 2.4088590526727183e-06,
      "loss": 0.4805,
      "step": 252410
    },
    {
      "epoch": 103.32,
      "grad_norm": 1.8521441221237183,
      "learning_rate": 2.408700484195256e-06,
      "loss": 0.4437,
      "step": 252420
    },
    {
      "epoch": 103.33,
      "grad_norm": 1.5413200855255127,
      "learning_rate": 2.4085419154202656e-06,
      "loss": 0.4744,
      "step": 252430
    },
    {
      "epoch": 103.33,
      "grad_norm": 1.8794727325439453,
      "learning_rate": 2.408383346348474e-06,
      "loss": 0.4436,
      "step": 252440
    },
    {
      "epoch": 103.34,
      "grad_norm": 1.8046566247940063,
      "learning_rate": 2.408224776980607e-06,
      "loss": 0.4573,
      "step": 252450
    },
    {
      "epoch": 103.34,
      "grad_norm": 1.8365970849990845,
      "learning_rate": 2.4080662073173907e-06,
      "loss": 0.4657,
      "step": 252460
    },
    {
      "epoch": 103.34,
      "grad_norm": 1.7314822673797607,
      "learning_rate": 2.407907637359552e-06,
      "loss": 0.4706,
      "step": 252470
    },
    {
      "epoch": 103.35,
      "grad_norm": 2.1141269207000732,
      "learning_rate": 2.4077490671078168e-06,
      "loss": 0.4714,
      "step": 252480
    },
    {
      "epoch": 103.35,
      "grad_norm": 1.8425589799880981,
      "learning_rate": 2.4075904965629117e-06,
      "loss": 0.4957,
      "step": 252490
    },
    {
      "epoch": 103.36,
      "grad_norm": 1.6359307765960693,
      "learning_rate": 2.4074319257255634e-06,
      "loss": 0.4747,
      "step": 252500
    },
    {
      "epoch": 103.36,
      "grad_norm": 2.003788709640503,
      "learning_rate": 2.4072733545964983e-06,
      "loss": 0.4915,
      "step": 252510
    },
    {
      "epoch": 103.36,
      "grad_norm": 2.491732358932495,
      "learning_rate": 2.407114783176442e-06,
      "loss": 0.471,
      "step": 252520
    },
    {
      "epoch": 103.37,
      "grad_norm": 1.9943177700042725,
      "learning_rate": 2.4069562114661223e-06,
      "loss": 0.4819,
      "step": 252530
    },
    {
      "epoch": 103.37,
      "grad_norm": 2.0877838134765625,
      "learning_rate": 2.4067976394662637e-06,
      "loss": 0.4661,
      "step": 252540
    },
    {
      "epoch": 103.38,
      "grad_norm": 2.06564998626709,
      "learning_rate": 2.4066390671775948e-06,
      "loss": 0.4826,
      "step": 252550
    },
    {
      "epoch": 103.38,
      "grad_norm": 1.7211494445800781,
      "learning_rate": 2.4064804946008398e-06,
      "loss": 0.4703,
      "step": 252560
    },
    {
      "epoch": 103.39,
      "grad_norm": 2.0715126991271973,
      "learning_rate": 2.406321921736727e-06,
      "loss": 0.4901,
      "step": 252570
    },
    {
      "epoch": 103.39,
      "grad_norm": 1.95163893699646,
      "learning_rate": 2.4061633485859816e-06,
      "loss": 0.4846,
      "step": 252580
    },
    {
      "epoch": 103.39,
      "grad_norm": 1.6370919942855835,
      "learning_rate": 2.40600477514933e-06,
      "loss": 0.4784,
      "step": 252590
    },
    {
      "epoch": 103.4,
      "grad_norm": 2.0106475353240967,
      "learning_rate": 2.4058462014274993e-06,
      "loss": 0.4737,
      "step": 252600
    },
    {
      "epoch": 103.4,
      "grad_norm": 1.804093837738037,
      "learning_rate": 2.4056876274212145e-06,
      "loss": 0.4893,
      "step": 252610
    },
    {
      "epoch": 103.41,
      "grad_norm": 1.979998230934143,
      "learning_rate": 2.405529053131204e-06,
      "loss": 0.4957,
      "step": 252620
    },
    {
      "epoch": 103.41,
      "grad_norm": 1.982625961303711,
      "learning_rate": 2.4053704785581927e-06,
      "loss": 0.4781,
      "step": 252630
    },
    {
      "epoch": 103.41,
      "grad_norm": 1.5186553001403809,
      "learning_rate": 2.4052119037029085e-06,
      "loss": 0.4638,
      "step": 252640
    },
    {
      "epoch": 103.42,
      "grad_norm": 1.6046222448349,
      "learning_rate": 2.4050533285660765e-06,
      "loss": 0.4772,
      "step": 252650
    },
    {
      "epoch": 103.42,
      "grad_norm": 2.09753155708313,
      "learning_rate": 2.4048947531484226e-06,
      "loss": 0.4895,
      "step": 252660
    },
    {
      "epoch": 103.43,
      "grad_norm": 2.1151814460754395,
      "learning_rate": 2.4047361774506748e-06,
      "loss": 0.4822,
      "step": 252670
    },
    {
      "epoch": 103.43,
      "grad_norm": 1.8418912887573242,
      "learning_rate": 2.4045776014735585e-06,
      "loss": 0.4925,
      "step": 252680
    },
    {
      "epoch": 103.43,
      "grad_norm": 2.113060235977173,
      "learning_rate": 2.4044190252178002e-06,
      "loss": 0.4729,
      "step": 252690
    },
    {
      "epoch": 103.44,
      "grad_norm": 2.407848596572876,
      "learning_rate": 2.404260448684126e-06,
      "loss": 0.4602,
      "step": 252700
    },
    {
      "epoch": 103.44,
      "grad_norm": 1.8282639980316162,
      "learning_rate": 2.404101871873265e-06,
      "loss": 0.4788,
      "step": 252710
    },
    {
      "epoch": 103.45,
      "grad_norm": 1.7809516191482544,
      "learning_rate": 2.40394329478594e-06,
      "loss": 0.4824,
      "step": 252720
    },
    {
      "epoch": 103.45,
      "grad_norm": 1.6044403314590454,
      "learning_rate": 2.4037847174228783e-06,
      "loss": 0.4718,
      "step": 252730
    },
    {
      "epoch": 103.45,
      "grad_norm": 1.9340907335281372,
      "learning_rate": 2.403626139784807e-06,
      "loss": 0.4722,
      "step": 252740
    },
    {
      "epoch": 103.46,
      "grad_norm": 1.8806474208831787,
      "learning_rate": 2.403467561872453e-06,
      "loss": 0.4768,
      "step": 252750
    },
    {
      "epoch": 103.46,
      "grad_norm": 3.00602650642395,
      "learning_rate": 2.4033089836865422e-06,
      "loss": 0.4768,
      "step": 252760
    },
    {
      "epoch": 103.47,
      "grad_norm": 1.812964916229248,
      "learning_rate": 2.4031504052278003e-06,
      "loss": 0.4638,
      "step": 252770
    },
    {
      "epoch": 103.47,
      "grad_norm": 1.7657382488250732,
      "learning_rate": 2.4029918264969546e-06,
      "loss": 0.4714,
      "step": 252780
    },
    {
      "epoch": 103.48,
      "grad_norm": 2.1053335666656494,
      "learning_rate": 2.4028332474947313e-06,
      "loss": 0.5029,
      "step": 252790
    },
    {
      "epoch": 103.48,
      "grad_norm": 1.8804380893707275,
      "learning_rate": 2.4026746682218565e-06,
      "loss": 0.4831,
      "step": 252800
    },
    {
      "epoch": 103.48,
      "grad_norm": 2.360189437866211,
      "learning_rate": 2.402516088679057e-06,
      "loss": 0.4577,
      "step": 252810
    },
    {
      "epoch": 103.49,
      "grad_norm": 1.8705707788467407,
      "learning_rate": 2.40235750886706e-06,
      "loss": 0.4895,
      "step": 252820
    },
    {
      "epoch": 103.49,
      "grad_norm": 1.4643958806991577,
      "learning_rate": 2.4021989287865903e-06,
      "loss": 0.4568,
      "step": 252830
    },
    {
      "epoch": 103.5,
      "grad_norm": 2.469067335128784,
      "learning_rate": 2.4020403484383746e-06,
      "loss": 0.4834,
      "step": 252840
    },
    {
      "epoch": 103.5,
      "grad_norm": 1.8745168447494507,
      "learning_rate": 2.4018817678231405e-06,
      "loss": 0.4489,
      "step": 252850
    },
    {
      "epoch": 103.5,
      "grad_norm": 2.032742500305176,
      "learning_rate": 2.401723186941614e-06,
      "loss": 0.4628,
      "step": 252860
    },
    {
      "epoch": 103.51,
      "grad_norm": 2.164935350418091,
      "learning_rate": 2.4015646057945214e-06,
      "loss": 0.4817,
      "step": 252870
    },
    {
      "epoch": 103.51,
      "grad_norm": 1.8894363641738892,
      "learning_rate": 2.4014060243825885e-06,
      "loss": 0.4694,
      "step": 252880
    },
    {
      "epoch": 103.52,
      "grad_norm": 2.0515849590301514,
      "learning_rate": 2.401247442706542e-06,
      "loss": 0.4653,
      "step": 252890
    },
    {
      "epoch": 103.52,
      "grad_norm": 1.9354583024978638,
      "learning_rate": 2.4010888607671095e-06,
      "loss": 0.482,
      "step": 252900
    },
    {
      "epoch": 103.52,
      "grad_norm": 1.9891899824142456,
      "learning_rate": 2.400930278565016e-06,
      "loss": 0.4652,
      "step": 252910
    },
    {
      "epoch": 103.53,
      "grad_norm": 1.6177831888198853,
      "learning_rate": 2.400771696100988e-06,
      "loss": 0.4682,
      "step": 252920
    },
    {
      "epoch": 103.53,
      "grad_norm": 1.5505033731460571,
      "learning_rate": 2.400613113375754e-06,
      "loss": 0.4905,
      "step": 252930
    },
    {
      "epoch": 103.54,
      "grad_norm": 1.8439041376113892,
      "learning_rate": 2.400454530390038e-06,
      "loss": 0.4803,
      "step": 252940
    },
    {
      "epoch": 103.54,
      "grad_norm": 1.9372715950012207,
      "learning_rate": 2.4002959471445673e-06,
      "loss": 0.4829,
      "step": 252950
    },
    {
      "epoch": 103.54,
      "grad_norm": 1.65068519115448,
      "learning_rate": 2.4001373636400683e-06,
      "loss": 0.4633,
      "step": 252960
    },
    {
      "epoch": 103.55,
      "grad_norm": 2.231973171234131,
      "learning_rate": 2.3999787798772675e-06,
      "loss": 0.469,
      "step": 252970
    },
    {
      "epoch": 103.55,
      "grad_norm": 2.1786017417907715,
      "learning_rate": 2.399820195856891e-06,
      "loss": 0.4709,
      "step": 252980
    },
    {
      "epoch": 103.56,
      "grad_norm": 1.6821141242980957,
      "learning_rate": 2.399661611579666e-06,
      "loss": 0.4825,
      "step": 252990
    },
    {
      "epoch": 103.56,
      "grad_norm": 1.4723947048187256,
      "learning_rate": 2.399503027046319e-06,
      "loss": 0.4941,
      "step": 253000
    },
    {
      "epoch": 103.57,
      "grad_norm": 2.2096619606018066,
      "learning_rate": 2.399344442257576e-06,
      "loss": 0.4676,
      "step": 253010
    },
    {
      "epoch": 103.57,
      "grad_norm": 2.20858097076416,
      "learning_rate": 2.399185857214163e-06,
      "loss": 0.4731,
      "step": 253020
    },
    {
      "epoch": 103.57,
      "grad_norm": 2.327770948410034,
      "learning_rate": 2.399027271916807e-06,
      "loss": 0.4689,
      "step": 253030
    },
    {
      "epoch": 103.58,
      "grad_norm": 2.521704912185669,
      "learning_rate": 2.3988686863662343e-06,
      "loss": 0.4855,
      "step": 253040
    },
    {
      "epoch": 103.58,
      "grad_norm": 1.7248021364212036,
      "learning_rate": 2.3987101005631717e-06,
      "loss": 0.468,
      "step": 253050
    },
    {
      "epoch": 103.59,
      "grad_norm": 1.8174803256988525,
      "learning_rate": 2.3985515145083454e-06,
      "loss": 0.468,
      "step": 253060
    },
    {
      "epoch": 103.59,
      "grad_norm": 2.062893867492676,
      "learning_rate": 2.3983929282024815e-06,
      "loss": 0.4694,
      "step": 253070
    },
    {
      "epoch": 103.59,
      "grad_norm": 1.842484951019287,
      "learning_rate": 2.398234341646307e-06,
      "loss": 0.4772,
      "step": 253080
    },
    {
      "epoch": 103.6,
      "grad_norm": 2.7157604694366455,
      "learning_rate": 2.3980757548405484e-06,
      "loss": 0.476,
      "step": 253090
    },
    {
      "epoch": 103.6,
      "grad_norm": 2.028414487838745,
      "learning_rate": 2.397917167785931e-06,
      "loss": 0.4665,
      "step": 253100
    },
    {
      "epoch": 103.61,
      "grad_norm": 2.1241579055786133,
      "learning_rate": 2.3977585804831834e-06,
      "loss": 0.4651,
      "step": 253110
    },
    {
      "epoch": 103.61,
      "grad_norm": 2.8913660049438477,
      "learning_rate": 2.39759999293303e-06,
      "loss": 0.4539,
      "step": 253120
    },
    {
      "epoch": 103.61,
      "grad_norm": 2.640901803970337,
      "learning_rate": 2.3974414051361985e-06,
      "loss": 0.461,
      "step": 253130
    },
    {
      "epoch": 103.62,
      "grad_norm": 1.8280242681503296,
      "learning_rate": 2.397282817093415e-06,
      "loss": 0.4645,
      "step": 253140
    },
    {
      "epoch": 103.62,
      "grad_norm": 2.1122286319732666,
      "learning_rate": 2.397124228805406e-06,
      "loss": 0.4709,
      "step": 253150
    },
    {
      "epoch": 103.63,
      "grad_norm": 1.8679410219192505,
      "learning_rate": 2.3969656402728983e-06,
      "loss": 0.4912,
      "step": 253160
    },
    {
      "epoch": 103.63,
      "grad_norm": 2.457714796066284,
      "learning_rate": 2.396807051496617e-06,
      "loss": 0.4812,
      "step": 253170
    },
    {
      "epoch": 103.63,
      "grad_norm": 1.8618535995483398,
      "learning_rate": 2.3966484624772902e-06,
      "loss": 0.467,
      "step": 253180
    },
    {
      "epoch": 103.64,
      "grad_norm": 2.186499834060669,
      "learning_rate": 2.396489873215644e-06,
      "loss": 0.4649,
      "step": 253190
    },
    {
      "epoch": 103.64,
      "grad_norm": 1.566524863243103,
      "learning_rate": 2.3963312837124037e-06,
      "loss": 0.4701,
      "step": 253200
    },
    {
      "epoch": 103.65,
      "grad_norm": 8.466752052307129,
      "learning_rate": 2.3961726939682975e-06,
      "loss": 0.4715,
      "step": 253210
    },
    {
      "epoch": 103.65,
      "grad_norm": 1.8398550748825073,
      "learning_rate": 2.396014103984051e-06,
      "loss": 0.4664,
      "step": 253220
    },
    {
      "epoch": 103.66,
      "grad_norm": 1.5487613677978516,
      "learning_rate": 2.395855513760391e-06,
      "loss": 0.4917,
      "step": 253230
    },
    {
      "epoch": 103.66,
      "grad_norm": 1.7053319215774536,
      "learning_rate": 2.395696923298043e-06,
      "loss": 0.4912,
      "step": 253240
    },
    {
      "epoch": 103.66,
      "grad_norm": 1.9784353971481323,
      "learning_rate": 2.3955383325977345e-06,
      "loss": 0.4771,
      "step": 253250
    },
    {
      "epoch": 103.67,
      "grad_norm": 2.153170585632324,
      "learning_rate": 2.395379741660191e-06,
      "loss": 0.4792,
      "step": 253260
    },
    {
      "epoch": 103.67,
      "grad_norm": 1.8326643705368042,
      "learning_rate": 2.3952211504861404e-06,
      "loss": 0.477,
      "step": 253270
    },
    {
      "epoch": 103.68,
      "grad_norm": 2.1722476482391357,
      "learning_rate": 2.3950625590763083e-06,
      "loss": 0.4872,
      "step": 253280
    },
    {
      "epoch": 103.68,
      "grad_norm": 1.6749118566513062,
      "learning_rate": 2.3949039674314212e-06,
      "loss": 0.4833,
      "step": 253290
    },
    {
      "epoch": 103.68,
      "grad_norm": 2.626757860183716,
      "learning_rate": 2.3947453755522063e-06,
      "loss": 0.4716,
      "step": 253300
    },
    {
      "epoch": 103.69,
      "grad_norm": 1.7692407369613647,
      "learning_rate": 2.3945867834393885e-06,
      "loss": 0.4656,
      "step": 253310
    },
    {
      "epoch": 103.69,
      "grad_norm": 1.9012932777404785,
      "learning_rate": 2.394428191093696e-06,
      "loss": 0.4598,
      "step": 253320
    },
    {
      "epoch": 103.7,
      "grad_norm": 1.723008632659912,
      "learning_rate": 2.3942695985158542e-06,
      "loss": 0.47,
      "step": 253330
    },
    {
      "epoch": 103.7,
      "grad_norm": 1.776800274848938,
      "learning_rate": 2.39411100570659e-06,
      "loss": 0.4678,
      "step": 253340
    },
    {
      "epoch": 103.7,
      "grad_norm": 1.973388671875,
      "learning_rate": 2.3939524126666298e-06,
      "loss": 0.5113,
      "step": 253350
    },
    {
      "epoch": 103.71,
      "grad_norm": 2.0173258781433105,
      "learning_rate": 2.3937938193967002e-06,
      "loss": 0.4679,
      "step": 253360
    },
    {
      "epoch": 103.71,
      "grad_norm": 1.7897217273712158,
      "learning_rate": 2.3936352258975276e-06,
      "loss": 0.4545,
      "step": 253370
    },
    {
      "epoch": 103.72,
      "grad_norm": 1.550589680671692,
      "learning_rate": 2.3934766321698386e-06,
      "loss": 0.4966,
      "step": 253380
    },
    {
      "epoch": 103.72,
      "grad_norm": 1.8506594896316528,
      "learning_rate": 2.393318038214359e-06,
      "loss": 0.4964,
      "step": 253390
    },
    {
      "epoch": 103.72,
      "grad_norm": 2.1340491771698,
      "learning_rate": 2.3931594440318167e-06,
      "loss": 0.4794,
      "step": 253400
    },
    {
      "epoch": 103.73,
      "grad_norm": 2.3211212158203125,
      "learning_rate": 2.3930008496229375e-06,
      "loss": 0.5012,
      "step": 253410
    },
    {
      "epoch": 103.73,
      "grad_norm": 2.1787304878234863,
      "learning_rate": 2.392842254988447e-06,
      "loss": 0.4882,
      "step": 253420
    },
    {
      "epoch": 103.74,
      "grad_norm": 1.857528805732727,
      "learning_rate": 2.392683660129073e-06,
      "loss": 0.4625,
      "step": 253430
    },
    {
      "epoch": 103.74,
      "grad_norm": 2.116314649581909,
      "learning_rate": 2.3925250650455413e-06,
      "loss": 0.4693,
      "step": 253440
    },
    {
      "epoch": 103.75,
      "grad_norm": 2.050276756286621,
      "learning_rate": 2.3923664697385792e-06,
      "loss": 0.4858,
      "step": 253450
    },
    {
      "epoch": 103.75,
      "grad_norm": 1.6748379468917847,
      "learning_rate": 2.3922078742089115e-06,
      "loss": 0.4676,
      "step": 253460
    },
    {
      "epoch": 103.75,
      "grad_norm": 1.9292106628417969,
      "learning_rate": 2.3920492784572665e-06,
      "loss": 0.4815,
      "step": 253470
    },
    {
      "epoch": 103.76,
      "grad_norm": 1.8298848867416382,
      "learning_rate": 2.3918906824843692e-06,
      "loss": 0.4721,
      "step": 253480
    },
    {
      "epoch": 103.76,
      "grad_norm": 1.9500410556793213,
      "learning_rate": 2.391732086290948e-06,
      "loss": 0.4782,
      "step": 253490
    },
    {
      "epoch": 103.77,
      "grad_norm": 2.0175395011901855,
      "learning_rate": 2.3915734898777273e-06,
      "loss": 0.4943,
      "step": 253500
    },
    {
      "epoch": 103.77,
      "grad_norm": 2.404534339904785,
      "learning_rate": 2.3914148932454352e-06,
      "loss": 0.4588,
      "step": 253510
    },
    {
      "epoch": 103.77,
      "grad_norm": 7.948938369750977,
      "learning_rate": 2.391256296394798e-06,
      "loss": 0.4736,
      "step": 253520
    },
    {
      "epoch": 103.78,
      "grad_norm": 1.9288907051086426,
      "learning_rate": 2.391097699326541e-06,
      "loss": 0.4607,
      "step": 253530
    },
    {
      "epoch": 103.78,
      "grad_norm": 1.6071149110794067,
      "learning_rate": 2.390939102041392e-06,
      "loss": 0.4649,
      "step": 253540
    },
    {
      "epoch": 103.79,
      "grad_norm": 1.8607360124588013,
      "learning_rate": 2.3907805045400763e-06,
      "loss": 0.4706,
      "step": 253550
    },
    {
      "epoch": 103.79,
      "grad_norm": 2.33974552154541,
      "learning_rate": 2.3906219068233216e-06,
      "loss": 0.472,
      "step": 253560
    },
    {
      "epoch": 103.79,
      "grad_norm": 1.8998383283615112,
      "learning_rate": 2.3904633088918542e-06,
      "loss": 0.4751,
      "step": 253570
    },
    {
      "epoch": 103.8,
      "grad_norm": 2.040776252746582,
      "learning_rate": 2.3903047107464005e-06,
      "loss": 0.4721,
      "step": 253580
    },
    {
      "epoch": 103.8,
      "grad_norm": 1.596014380455017,
      "learning_rate": 2.3901461123876863e-06,
      "loss": 0.4933,
      "step": 253590
    },
    {
      "epoch": 103.81,
      "grad_norm": 1.9256466627120972,
      "learning_rate": 2.389987513816439e-06,
      "loss": 0.4574,
      "step": 253600
    },
    {
      "epoch": 103.81,
      "grad_norm": 1.889988660812378,
      "learning_rate": 2.389828915033385e-06,
      "loss": 0.4775,
      "step": 253610
    },
    {
      "epoch": 103.81,
      "grad_norm": 1.9581149816513062,
      "learning_rate": 2.3896703160392503e-06,
      "loss": 0.4856,
      "step": 253620
    },
    {
      "epoch": 103.82,
      "grad_norm": 1.6770100593566895,
      "learning_rate": 2.3895117168347614e-06,
      "loss": 0.4828,
      "step": 253630
    },
    {
      "epoch": 103.82,
      "grad_norm": 1.6315325498580933,
      "learning_rate": 2.389353117420645e-06,
      "loss": 0.4842,
      "step": 253640
    },
    {
      "epoch": 103.83,
      "grad_norm": 2.3635847568511963,
      "learning_rate": 2.389194517797629e-06,
      "loss": 0.4743,
      "step": 253650
    },
    {
      "epoch": 103.83,
      "grad_norm": 1.7705492973327637,
      "learning_rate": 2.389035917966438e-06,
      "loss": 0.4649,
      "step": 253660
    },
    {
      "epoch": 103.84,
      "grad_norm": 1.8546383380889893,
      "learning_rate": 2.3888773179277995e-06,
      "loss": 0.4696,
      "step": 253670
    },
    {
      "epoch": 103.84,
      "grad_norm": 1.9870299100875854,
      "learning_rate": 2.3887187176824398e-06,
      "loss": 0.4723,
      "step": 253680
    },
    {
      "epoch": 103.84,
      "grad_norm": 2.07413649559021,
      "learning_rate": 2.388560117231085e-06,
      "loss": 0.4758,
      "step": 253690
    },
    {
      "epoch": 103.85,
      "grad_norm": 2.1795902252197266,
      "learning_rate": 2.388401516574462e-06,
      "loss": 0.4835,
      "step": 253700
    },
    {
      "epoch": 103.85,
      "grad_norm": 1.7194989919662476,
      "learning_rate": 2.3882429157132972e-06,
      "loss": 0.5012,
      "step": 253710
    },
    {
      "epoch": 103.86,
      "grad_norm": 1.6870497465133667,
      "learning_rate": 2.3880843146483174e-06,
      "loss": 0.4738,
      "step": 253720
    },
    {
      "epoch": 103.86,
      "grad_norm": 2.287177324295044,
      "learning_rate": 2.3879257133802492e-06,
      "loss": 0.4979,
      "step": 253730
    },
    {
      "epoch": 103.86,
      "grad_norm": 1.9836766719818115,
      "learning_rate": 2.387767111909819e-06,
      "loss": 0.4783,
      "step": 253740
    },
    {
      "epoch": 103.87,
      "grad_norm": 1.8013827800750732,
      "learning_rate": 2.3876085102377532e-06,
      "loss": 0.5108,
      "step": 253750
    },
    {
      "epoch": 103.87,
      "grad_norm": 1.797192096710205,
      "learning_rate": 2.3874499083647777e-06,
      "loss": 0.485,
      "step": 253760
    },
    {
      "epoch": 103.88,
      "grad_norm": 1.913995385169983,
      "learning_rate": 2.38729130629162e-06,
      "loss": 0.4826,
      "step": 253770
    },
    {
      "epoch": 103.88,
      "grad_norm": 2.035875082015991,
      "learning_rate": 2.387132704019007e-06,
      "loss": 0.4691,
      "step": 253780
    },
    {
      "epoch": 103.88,
      "grad_norm": 2.1587235927581787,
      "learning_rate": 2.3869741015476637e-06,
      "loss": 0.4908,
      "step": 253790
    },
    {
      "epoch": 103.89,
      "grad_norm": 2.4000930786132812,
      "learning_rate": 2.386815498878318e-06,
      "loss": 0.4705,
      "step": 253800
    },
    {
      "epoch": 103.89,
      "grad_norm": 1.9051249027252197,
      "learning_rate": 2.386656896011696e-06,
      "loss": 0.4831,
      "step": 253810
    },
    {
      "epoch": 103.9,
      "grad_norm": 1.7800495624542236,
      "learning_rate": 2.3864982929485237e-06,
      "loss": 0.4536,
      "step": 253820
    },
    {
      "epoch": 103.9,
      "grad_norm": 1.8855066299438477,
      "learning_rate": 2.386339689689528e-06,
      "loss": 0.4831,
      "step": 253830
    },
    {
      "epoch": 103.91,
      "grad_norm": 1.599591612815857,
      "learning_rate": 2.3861810862354357e-06,
      "loss": 0.4683,
      "step": 253840
    },
    {
      "epoch": 103.91,
      "grad_norm": 1.8599956035614014,
      "learning_rate": 2.3860224825869732e-06,
      "loss": 0.4954,
      "step": 253850
    },
    {
      "epoch": 103.91,
      "grad_norm": 1.8024771213531494,
      "learning_rate": 2.385863878744867e-06,
      "loss": 0.4727,
      "step": 253860
    },
    {
      "epoch": 103.92,
      "grad_norm": 1.7967439889907837,
      "learning_rate": 2.385705274709844e-06,
      "loss": 0.4888,
      "step": 253870
    },
    {
      "epoch": 103.92,
      "grad_norm": 1.5211701393127441,
      "learning_rate": 2.38554667048263e-06,
      "loss": 0.4516,
      "step": 253880
    },
    {
      "epoch": 103.93,
      "grad_norm": 2.3522303104400635,
      "learning_rate": 2.385388066063952e-06,
      "loss": 0.484,
      "step": 253890
    },
    {
      "epoch": 103.93,
      "grad_norm": 1.992637276649475,
      "learning_rate": 2.3852294614545366e-06,
      "loss": 0.471,
      "step": 253900
    },
    {
      "epoch": 103.93,
      "grad_norm": 2.059152841567993,
      "learning_rate": 2.38507085665511e-06,
      "loss": 0.4584,
      "step": 253910
    },
    {
      "epoch": 103.94,
      "grad_norm": 2.1227715015411377,
      "learning_rate": 2.3849122516663993e-06,
      "loss": 0.4812,
      "step": 253920
    },
    {
      "epoch": 103.94,
      "grad_norm": 1.815448522567749,
      "learning_rate": 2.38475364648913e-06,
      "loss": 0.4873,
      "step": 253930
    },
    {
      "epoch": 103.95,
      "grad_norm": 1.9453314542770386,
      "learning_rate": 2.38459504112403e-06,
      "loss": 0.4525,
      "step": 253940
    },
    {
      "epoch": 103.95,
      "grad_norm": 1.9092978239059448,
      "learning_rate": 2.3844364355718244e-06,
      "loss": 0.4648,
      "step": 253950
    },
    {
      "epoch": 103.95,
      "grad_norm": 1.8230234384536743,
      "learning_rate": 2.3842778298332413e-06,
      "loss": 0.4597,
      "step": 253960
    },
    {
      "epoch": 103.96,
      "grad_norm": 2.465012550354004,
      "learning_rate": 2.3841192239090062e-06,
      "loss": 0.4911,
      "step": 253970
    },
    {
      "epoch": 103.96,
      "grad_norm": 2.4973998069763184,
      "learning_rate": 2.383960617799846e-06,
      "loss": 0.4843,
      "step": 253980
    },
    {
      "epoch": 103.97,
      "grad_norm": 1.8465923070907593,
      "learning_rate": 2.3838020115064873e-06,
      "loss": 0.4685,
      "step": 253990
    },
    {
      "epoch": 103.97,
      "grad_norm": 1.745157241821289,
      "learning_rate": 2.3836434050296565e-06,
      "loss": 0.4878,
      "step": 254000
    },
    {
      "epoch": 103.97,
      "grad_norm": 2.3823676109313965,
      "learning_rate": 2.38348479837008e-06,
      "loss": 0.4614,
      "step": 254010
    },
    {
      "epoch": 103.98,
      "grad_norm": 1.6968916654586792,
      "learning_rate": 2.3833261915284845e-06,
      "loss": 0.4789,
      "step": 254020
    },
    {
      "epoch": 103.98,
      "grad_norm": 2.3530564308166504,
      "learning_rate": 2.383167584505597e-06,
      "loss": 0.4664,
      "step": 254030
    },
    {
      "epoch": 103.99,
      "grad_norm": 1.9389442205429077,
      "learning_rate": 2.3830089773021436e-06,
      "loss": 0.4802,
      "step": 254040
    },
    {
      "epoch": 103.99,
      "grad_norm": 2.387922763824463,
      "learning_rate": 2.3828503699188503e-06,
      "loss": 0.4644,
      "step": 254050
    },
    {
      "epoch": 104.0,
      "grad_norm": 2.490292549133301,
      "learning_rate": 2.382691762356445e-06,
      "loss": 0.4873,
      "step": 254060
    },
    {
      "epoch": 104.0,
      "grad_norm": 1.931312918663025,
      "learning_rate": 2.382533154615653e-06,
      "loss": 0.4682,
      "step": 254070
    },
    {
      "epoch": 104.0,
      "eval_loss": 0.47496721148490906,
      "eval_runtime": 54.5998,
      "eval_samples_per_second": 63.169,
      "eval_steps_per_second": 7.912,
      "step": 254072
    },
    {
      "epoch": 104.0,
      "grad_norm": 2.0244200229644775,
      "learning_rate": 2.382374546697202e-06,
      "loss": 0.4482,
      "step": 254080
    },
    {
      "epoch": 104.01,
      "grad_norm": 1.7728550434112549,
      "learning_rate": 2.3822159386018176e-06,
      "loss": 0.4627,
      "step": 254090
    },
    {
      "epoch": 104.01,
      "grad_norm": 1.741735816001892,
      "learning_rate": 2.382057330330227e-06,
      "loss": 0.4576,
      "step": 254100
    },
    {
      "epoch": 104.02,
      "grad_norm": 1.890296459197998,
      "learning_rate": 2.381898721883156e-06,
      "loss": 0.455,
      "step": 254110
    },
    {
      "epoch": 104.02,
      "grad_norm": 2.193283796310425,
      "learning_rate": 2.381740113261331e-06,
      "loss": 0.4743,
      "step": 254120
    },
    {
      "epoch": 104.02,
      "grad_norm": 1.9703348875045776,
      "learning_rate": 2.3815815044654798e-06,
      "loss": 0.4605,
      "step": 254130
    },
    {
      "epoch": 104.03,
      "grad_norm": 1.8979886770248413,
      "learning_rate": 2.3814228954963286e-06,
      "loss": 0.4798,
      "step": 254140
    },
    {
      "epoch": 104.03,
      "grad_norm": 1.9705911874771118,
      "learning_rate": 2.3812642863546034e-06,
      "loss": 0.4775,
      "step": 254150
    },
    {
      "epoch": 104.04,
      "grad_norm": 1.7969828844070435,
      "learning_rate": 2.3811056770410312e-06,
      "loss": 0.484,
      "step": 254160
    },
    {
      "epoch": 104.04,
      "grad_norm": 1.959833025932312,
      "learning_rate": 2.3809470675563385e-06,
      "loss": 0.4665,
      "step": 254170
    },
    {
      "epoch": 104.04,
      "grad_norm": 1.9898711442947388,
      "learning_rate": 2.3807884579012516e-06,
      "loss": 0.4588,
      "step": 254180
    },
    {
      "epoch": 104.05,
      "grad_norm": 1.8236393928527832,
      "learning_rate": 2.3806298480764976e-06,
      "loss": 0.4817,
      "step": 254190
    },
    {
      "epoch": 104.05,
      "grad_norm": 1.8841310739517212,
      "learning_rate": 2.3804712380828025e-06,
      "loss": 0.4813,
      "step": 254200
    },
    {
      "epoch": 104.06,
      "grad_norm": 2.408989429473877,
      "learning_rate": 2.3803126279208926e-06,
      "loss": 0.4501,
      "step": 254210
    },
    {
      "epoch": 104.06,
      "grad_norm": 1.8435629606246948,
      "learning_rate": 2.3801540175914954e-06,
      "loss": 0.4855,
      "step": 254220
    },
    {
      "epoch": 104.06,
      "grad_norm": 2.143927574157715,
      "learning_rate": 2.379995407095338e-06,
      "loss": 0.4668,
      "step": 254230
    },
    {
      "epoch": 104.07,
      "grad_norm": 1.914303183555603,
      "learning_rate": 2.3798367964331444e-06,
      "loss": 0.4848,
      "step": 254240
    },
    {
      "epoch": 104.07,
      "grad_norm": 1.8472166061401367,
      "learning_rate": 2.3796781856056436e-06,
      "loss": 0.4668,
      "step": 254250
    },
    {
      "epoch": 104.08,
      "grad_norm": 2.2083945274353027,
      "learning_rate": 2.379519574613561e-06,
      "loss": 0.4772,
      "step": 254260
    },
    {
      "epoch": 104.08,
      "grad_norm": 1.8995522260665894,
      "learning_rate": 2.3793609634576235e-06,
      "loss": 0.4937,
      "step": 254270
    },
    {
      "epoch": 104.09,
      "grad_norm": 2.5523488521575928,
      "learning_rate": 2.3792023521385586e-06,
      "loss": 0.4658,
      "step": 254280
    },
    {
      "epoch": 104.09,
      "grad_norm": 2.121542453765869,
      "learning_rate": 2.3790437406570907e-06,
      "loss": 0.475,
      "step": 254290
    },
    {
      "epoch": 104.09,
      "grad_norm": 1.727051854133606,
      "learning_rate": 2.3788851290139485e-06,
      "loss": 0.4724,
      "step": 254300
    },
    {
      "epoch": 104.1,
      "grad_norm": 1.987220048904419,
      "learning_rate": 2.378726517209857e-06,
      "loss": 0.4905,
      "step": 254310
    },
    {
      "epoch": 104.1,
      "grad_norm": 1.701559066772461,
      "learning_rate": 2.378567905245544e-06,
      "loss": 0.4749,
      "step": 254320
    },
    {
      "epoch": 104.11,
      "grad_norm": 2.3232548236846924,
      "learning_rate": 2.3784092931217357e-06,
      "loss": 0.4871,
      "step": 254330
    },
    {
      "epoch": 104.11,
      "grad_norm": 1.7304221391677856,
      "learning_rate": 2.378250680839158e-06,
      "loss": 0.488,
      "step": 254340
    },
    {
      "epoch": 104.11,
      "grad_norm": 2.321058511734009,
      "learning_rate": 2.378092068398538e-06,
      "loss": 0.4991,
      "step": 254350
    },
    {
      "epoch": 104.12,
      "grad_norm": 1.5981625318527222,
      "learning_rate": 2.3779334558006033e-06,
      "loss": 0.4755,
      "step": 254360
    },
    {
      "epoch": 104.12,
      "grad_norm": 2.1451618671417236,
      "learning_rate": 2.3777748430460784e-06,
      "loss": 0.4733,
      "step": 254370
    },
    {
      "epoch": 104.13,
      "grad_norm": 2.1357035636901855,
      "learning_rate": 2.377616230135692e-06,
      "loss": 0.4753,
      "step": 254380
    },
    {
      "epoch": 104.13,
      "grad_norm": 1.9759148359298706,
      "learning_rate": 2.377457617070169e-06,
      "loss": 0.4788,
      "step": 254390
    },
    {
      "epoch": 104.13,
      "grad_norm": 1.3357504606246948,
      "learning_rate": 2.377299003850236e-06,
      "loss": 0.4824,
      "step": 254400
    },
    {
      "epoch": 104.14,
      "grad_norm": 1.928990125656128,
      "learning_rate": 2.377140390476621e-06,
      "loss": 0.4719,
      "step": 254410
    },
    {
      "epoch": 104.14,
      "grad_norm": 1.7429293394088745,
      "learning_rate": 2.3769817769500498e-06,
      "loss": 0.4801,
      "step": 254420
    },
    {
      "epoch": 104.15,
      "grad_norm": 1.9166837930679321,
      "learning_rate": 2.376823163271249e-06,
      "loss": 0.4595,
      "step": 254430
    },
    {
      "epoch": 104.15,
      "grad_norm": 2.2285025119781494,
      "learning_rate": 2.3766645494409447e-06,
      "loss": 0.4779,
      "step": 254440
    },
    {
      "epoch": 104.15,
      "grad_norm": 1.7442457675933838,
      "learning_rate": 2.376505935459864e-06,
      "loss": 0.4883,
      "step": 254450
    },
    {
      "epoch": 104.16,
      "grad_norm": 1.6748589277267456,
      "learning_rate": 2.3763473213287344e-06,
      "loss": 0.4685,
      "step": 254460
    },
    {
      "epoch": 104.16,
      "grad_norm": 1.7581921815872192,
      "learning_rate": 2.376188707048281e-06,
      "loss": 0.4597,
      "step": 254470
    },
    {
      "epoch": 104.17,
      "grad_norm": 2.145554780960083,
      "learning_rate": 2.3760300926192307e-06,
      "loss": 0.4858,
      "step": 254480
    },
    {
      "epoch": 104.17,
      "grad_norm": 1.7883400917053223,
      "learning_rate": 2.3758714780423094e-06,
      "loss": 0.4797,
      "step": 254490
    },
    {
      "epoch": 104.18,
      "grad_norm": 1.9077519178390503,
      "learning_rate": 2.3757128633182457e-06,
      "loss": 0.4948,
      "step": 254500
    },
    {
      "epoch": 104.18,
      "grad_norm": 1.931796908378601,
      "learning_rate": 2.3755542484477647e-06,
      "loss": 0.4597,
      "step": 254510
    },
    {
      "epoch": 104.18,
      "grad_norm": 2.1731319427490234,
      "learning_rate": 2.3753956334315945e-06,
      "loss": 0.4766,
      "step": 254520
    },
    {
      "epoch": 104.19,
      "grad_norm": 2.1676058769226074,
      "learning_rate": 2.3752370182704593e-06,
      "loss": 0.4615,
      "step": 254530
    },
    {
      "epoch": 104.19,
      "grad_norm": 1.941697597503662,
      "learning_rate": 2.3750784029650875e-06,
      "loss": 0.4851,
      "step": 254540
    },
    {
      "epoch": 104.2,
      "grad_norm": 2.018948554992676,
      "learning_rate": 2.374919787516204e-06,
      "loss": 0.4857,
      "step": 254550
    },
    {
      "epoch": 104.2,
      "grad_norm": 1.7769232988357544,
      "learning_rate": 2.374761171924538e-06,
      "loss": 0.4788,
      "step": 254560
    },
    {
      "epoch": 104.2,
      "grad_norm": 1.8036805391311646,
      "learning_rate": 2.374602556190814e-06,
      "loss": 0.4788,
      "step": 254570
    },
    {
      "epoch": 104.21,
      "grad_norm": 1.7561997175216675,
      "learning_rate": 2.3744439403157593e-06,
      "loss": 0.4798,
      "step": 254580
    },
    {
      "epoch": 104.21,
      "grad_norm": 1.9900139570236206,
      "learning_rate": 2.3742853243001e-06,
      "loss": 0.4768,
      "step": 254590
    },
    {
      "epoch": 104.22,
      "grad_norm": 2.11155104637146,
      "learning_rate": 2.3741267081445637e-06,
      "loss": 0.4687,
      "step": 254600
    },
    {
      "epoch": 104.22,
      "grad_norm": 2.485952377319336,
      "learning_rate": 2.373968091849876e-06,
      "loss": 0.4828,
      "step": 254610
    },
    {
      "epoch": 104.22,
      "grad_norm": 1.7105308771133423,
      "learning_rate": 2.373809475416764e-06,
      "loss": 0.4793,
      "step": 254620
    },
    {
      "epoch": 104.23,
      "grad_norm": 1.7470016479492188,
      "learning_rate": 2.3736508588459543e-06,
      "loss": 0.4716,
      "step": 254630
    },
    {
      "epoch": 104.23,
      "grad_norm": 1.5318206548690796,
      "learning_rate": 2.3734922421381733e-06,
      "loss": 0.4865,
      "step": 254640
    },
    {
      "epoch": 104.24,
      "grad_norm": 1.992305874824524,
      "learning_rate": 2.3733336252941475e-06,
      "loss": 0.4853,
      "step": 254650
    },
    {
      "epoch": 104.24,
      "grad_norm": 2.261246681213379,
      "learning_rate": 2.373175008314604e-06,
      "loss": 0.4672,
      "step": 254660
    },
    {
      "epoch": 104.24,
      "grad_norm": 1.7219494581222534,
      "learning_rate": 2.3730163912002693e-06,
      "loss": 0.4554,
      "step": 254670
    },
    {
      "epoch": 104.25,
      "grad_norm": 2.0136611461639404,
      "learning_rate": 2.3728577739518696e-06,
      "loss": 0.482,
      "step": 254680
    },
    {
      "epoch": 104.25,
      "grad_norm": 2.013916492462158,
      "learning_rate": 2.3726991565701316e-06,
      "loss": 0.4863,
      "step": 254690
    },
    {
      "epoch": 104.26,
      "grad_norm": 1.9482074975967407,
      "learning_rate": 2.3725405390557817e-06,
      "loss": 0.4724,
      "step": 254700
    },
    {
      "epoch": 104.26,
      "grad_norm": 1.9360816478729248,
      "learning_rate": 2.372381921409547e-06,
      "loss": 0.4798,
      "step": 254710
    },
    {
      "epoch": 104.27,
      "grad_norm": 2.039376735687256,
      "learning_rate": 2.372223303632154e-06,
      "loss": 0.4843,
      "step": 254720
    },
    {
      "epoch": 104.27,
      "grad_norm": 1.95827054977417,
      "learning_rate": 2.3720646857243295e-06,
      "loss": 0.4568,
      "step": 254730
    },
    {
      "epoch": 104.27,
      "grad_norm": 2.579127073287964,
      "learning_rate": 2.3719060676867995e-06,
      "loss": 0.464,
      "step": 254740
    },
    {
      "epoch": 104.28,
      "grad_norm": 1.5959144830703735,
      "learning_rate": 2.3717474495202914e-06,
      "loss": 0.4721,
      "step": 254750
    },
    {
      "epoch": 104.28,
      "grad_norm": 2.524402618408203,
      "learning_rate": 2.371588831225531e-06,
      "loss": 0.4588,
      "step": 254760
    },
    {
      "epoch": 104.29,
      "grad_norm": 1.5721604824066162,
      "learning_rate": 2.3714302128032453e-06,
      "loss": 0.4492,
      "step": 254770
    },
    {
      "epoch": 104.29,
      "grad_norm": 1.8051024675369263,
      "learning_rate": 2.3712715942541596e-06,
      "loss": 0.4809,
      "step": 254780
    },
    {
      "epoch": 104.29,
      "grad_norm": 2.656575918197632,
      "learning_rate": 2.371112975579003e-06,
      "loss": 0.4706,
      "step": 254790
    },
    {
      "epoch": 104.3,
      "grad_norm": 1.9304444789886475,
      "learning_rate": 2.3709543567785006e-06,
      "loss": 0.4684,
      "step": 254800
    },
    {
      "epoch": 104.3,
      "grad_norm": 2.231177568435669,
      "learning_rate": 2.3707957378533793e-06,
      "loss": 0.4624,
      "step": 254810
    },
    {
      "epoch": 104.31,
      "grad_norm": 1.818852186203003,
      "learning_rate": 2.370637118804366e-06,
      "loss": 0.4824,
      "step": 254820
    },
    {
      "epoch": 104.31,
      "grad_norm": 1.6190910339355469,
      "learning_rate": 2.3704784996321865e-06,
      "loss": 0.461,
      "step": 254830
    },
    {
      "epoch": 104.31,
      "grad_norm": 2.14681077003479,
      "learning_rate": 2.370319880337568e-06,
      "loss": 0.458,
      "step": 254840
    },
    {
      "epoch": 104.32,
      "grad_norm": 1.6659992933273315,
      "learning_rate": 2.3701612609212375e-06,
      "loss": 0.4691,
      "step": 254850
    },
    {
      "epoch": 104.32,
      "grad_norm": 1.901436686515808,
      "learning_rate": 2.37000264138392e-06,
      "loss": 0.5006,
      "step": 254860
    },
    {
      "epoch": 104.33,
      "grad_norm": 1.8555837869644165,
      "learning_rate": 2.3698440217263435e-06,
      "loss": 0.4686,
      "step": 254870
    },
    {
      "epoch": 104.33,
      "grad_norm": 2.0078930854797363,
      "learning_rate": 2.3696854019492353e-06,
      "loss": 0.4818,
      "step": 254880
    },
    {
      "epoch": 104.33,
      "grad_norm": 1.6261202096939087,
      "learning_rate": 2.36952678205332e-06,
      "loss": 0.4801,
      "step": 254890
    },
    {
      "epoch": 104.34,
      "grad_norm": 2.1735055446624756,
      "learning_rate": 2.3693681620393257e-06,
      "loss": 0.4808,
      "step": 254900
    },
    {
      "epoch": 104.34,
      "grad_norm": 2.011950969696045,
      "learning_rate": 2.3692095419079786e-06,
      "loss": 0.4687,
      "step": 254910
    },
    {
      "epoch": 104.35,
      "grad_norm": 1.8956804275512695,
      "learning_rate": 2.369050921660005e-06,
      "loss": 0.4625,
      "step": 254920
    },
    {
      "epoch": 104.35,
      "grad_norm": 1.5556397438049316,
      "learning_rate": 2.3688923012961323e-06,
      "loss": 0.4611,
      "step": 254930
    },
    {
      "epoch": 104.36,
      "grad_norm": 1.611485242843628,
      "learning_rate": 2.3687336808170857e-06,
      "loss": 0.4709,
      "step": 254940
    },
    {
      "epoch": 104.36,
      "grad_norm": 1.669224739074707,
      "learning_rate": 2.3685750602235936e-06,
      "loss": 0.4502,
      "step": 254950
    },
    {
      "epoch": 104.36,
      "grad_norm": 1.9082342386245728,
      "learning_rate": 2.368416439516382e-06,
      "loss": 0.4696,
      "step": 254960
    },
    {
      "epoch": 104.37,
      "grad_norm": 1.863954782485962,
      "learning_rate": 2.368257818696176e-06,
      "loss": 0.4885,
      "step": 254970
    },
    {
      "epoch": 104.37,
      "grad_norm": 1.8949823379516602,
      "learning_rate": 2.368099197763704e-06,
      "loss": 0.4775,
      "step": 254980
    },
    {
      "epoch": 104.38,
      "grad_norm": 1.6628838777542114,
      "learning_rate": 2.3679405767196926e-06,
      "loss": 0.483,
      "step": 254990
    },
    {
      "epoch": 104.38,
      "grad_norm": 1.8667283058166504,
      "learning_rate": 2.3677819555648675e-06,
      "loss": 0.4688,
      "step": 255000
    },
    {
      "epoch": 104.38,
      "grad_norm": 2.301147937774658,
      "learning_rate": 2.367623334299956e-06,
      "loss": 0.4866,
      "step": 255010
    },
    {
      "epoch": 104.39,
      "grad_norm": 1.9405864477157593,
      "learning_rate": 2.3674647129256843e-06,
      "loss": 0.4847,
      "step": 255020
    },
    {
      "epoch": 104.39,
      "grad_norm": 2.2120797634124756,
      "learning_rate": 2.3673060914427792e-06,
      "loss": 0.4709,
      "step": 255030
    },
    {
      "epoch": 104.4,
      "grad_norm": 1.9924010038375854,
      "learning_rate": 2.3671474698519677e-06,
      "loss": 0.4841,
      "step": 255040
    },
    {
      "epoch": 104.4,
      "grad_norm": 1.8151854276657104,
      "learning_rate": 2.3669888481539755e-06,
      "loss": 0.4768,
      "step": 255050
    },
    {
      "epoch": 104.4,
      "grad_norm": 2.1784305572509766,
      "learning_rate": 2.3668302263495295e-06,
      "loss": 0.5094,
      "step": 255060
    },
    {
      "epoch": 104.41,
      "grad_norm": 1.885811448097229,
      "learning_rate": 2.3666716044393565e-06,
      "loss": 0.4599,
      "step": 255070
    },
    {
      "epoch": 104.41,
      "grad_norm": 1.992835283279419,
      "learning_rate": 2.3665129824241835e-06,
      "loss": 0.4648,
      "step": 255080
    },
    {
      "epoch": 104.42,
      "grad_norm": 1.7821955680847168,
      "learning_rate": 2.366354360304737e-06,
      "loss": 0.4978,
      "step": 255090
    },
    {
      "epoch": 104.42,
      "grad_norm": 3.3540170192718506,
      "learning_rate": 2.3661957380817432e-06,
      "loss": 0.4813,
      "step": 255100
    },
    {
      "epoch": 104.42,
      "grad_norm": 1.997564673423767,
      "learning_rate": 2.3660371157559294e-06,
      "loss": 0.4672,
      "step": 255110
    },
    {
      "epoch": 104.43,
      "grad_norm": 2.1822102069854736,
      "learning_rate": 2.365878493328022e-06,
      "loss": 0.4739,
      "step": 255120
    },
    {
      "epoch": 104.43,
      "grad_norm": 1.9986584186553955,
      "learning_rate": 2.365719870798746e-06,
      "loss": 0.4783,
      "step": 255130
    },
    {
      "epoch": 104.44,
      "grad_norm": 1.9906479120254517,
      "learning_rate": 2.3655612481688305e-06,
      "loss": 0.4434,
      "step": 255140
    },
    {
      "epoch": 104.44,
      "grad_norm": 1.9217884540557861,
      "learning_rate": 2.365402625439e-06,
      "loss": 0.4694,
      "step": 255150
    },
    {
      "epoch": 104.45,
      "grad_norm": 2.0106732845306396,
      "learning_rate": 2.365244002609983e-06,
      "loss": 0.4871,
      "step": 255160
    },
    {
      "epoch": 104.45,
      "grad_norm": 1.6345088481903076,
      "learning_rate": 2.3650853796825053e-06,
      "loss": 0.4653,
      "step": 255170
    },
    {
      "epoch": 104.45,
      "grad_norm": 1.6964614391326904,
      "learning_rate": 2.3649267566572943e-06,
      "loss": 0.4589,
      "step": 255180
    },
    {
      "epoch": 104.46,
      "grad_norm": 2.1099071502685547,
      "learning_rate": 2.3647681335350747e-06,
      "loss": 0.4329,
      "step": 255190
    },
    {
      "epoch": 104.46,
      "grad_norm": 1.733962893486023,
      "learning_rate": 2.3646095103165745e-06,
      "loss": 0.4946,
      "step": 255200
    },
    {
      "epoch": 104.47,
      "grad_norm": 1.8477824926376343,
      "learning_rate": 2.3644508870025206e-06,
      "loss": 0.5054,
      "step": 255210
    },
    {
      "epoch": 104.47,
      "grad_norm": 2.2226879596710205,
      "learning_rate": 2.3642922635936388e-06,
      "loss": 0.477,
      "step": 255220
    },
    {
      "epoch": 104.47,
      "grad_norm": 2.2949585914611816,
      "learning_rate": 2.364133640090656e-06,
      "loss": 0.4584,
      "step": 255230
    },
    {
      "epoch": 104.48,
      "grad_norm": 2.3561971187591553,
      "learning_rate": 2.3639750164942985e-06,
      "loss": 0.4818,
      "step": 255240
    },
    {
      "epoch": 104.48,
      "grad_norm": 2.3439137935638428,
      "learning_rate": 2.363816392805294e-06,
      "loss": 0.4921,
      "step": 255250
    },
    {
      "epoch": 104.49,
      "grad_norm": 1.6733111143112183,
      "learning_rate": 2.3636577690243682e-06,
      "loss": 0.4691,
      "step": 255260
    },
    {
      "epoch": 104.49,
      "grad_norm": 1.9612194299697876,
      "learning_rate": 2.3634991451522483e-06,
      "loss": 0.4715,
      "step": 255270
    },
    {
      "epoch": 104.49,
      "grad_norm": 1.8625868558883667,
      "learning_rate": 2.3633405211896602e-06,
      "loss": 0.4469,
      "step": 255280
    },
    {
      "epoch": 104.5,
      "grad_norm": 2.0933632850646973,
      "learning_rate": 2.363181897137331e-06,
      "loss": 0.5056,
      "step": 255290
    },
    {
      "epoch": 104.5,
      "grad_norm": 1.7496343851089478,
      "learning_rate": 2.3630232729959877e-06,
      "loss": 0.4618,
      "step": 255300
    },
    {
      "epoch": 104.51,
      "grad_norm": 1.6471350193023682,
      "learning_rate": 2.362864648766356e-06,
      "loss": 0.4874,
      "step": 255310
    },
    {
      "epoch": 104.51,
      "grad_norm": 1.968794822692871,
      "learning_rate": 2.362706024449163e-06,
      "loss": 0.4789,
      "step": 255320
    },
    {
      "epoch": 104.51,
      "grad_norm": 1.6956835985183716,
      "learning_rate": 2.362547400045136e-06,
      "loss": 0.4718,
      "step": 255330
    },
    {
      "epoch": 104.52,
      "grad_norm": 1.710209846496582,
      "learning_rate": 2.362388775555e-06,
      "loss": 0.4866,
      "step": 255340
    },
    {
      "epoch": 104.52,
      "grad_norm": 1.772800087928772,
      "learning_rate": 2.362230150979484e-06,
      "loss": 0.4863,
      "step": 255350
    },
    {
      "epoch": 104.53,
      "grad_norm": 1.826064944267273,
      "learning_rate": 2.3620715263193124e-06,
      "loss": 0.4822,
      "step": 255360
    },
    {
      "epoch": 104.53,
      "grad_norm": 1.698122501373291,
      "learning_rate": 2.361912901575213e-06,
      "loss": 0.4731,
      "step": 255370
    },
    {
      "epoch": 104.54,
      "grad_norm": 1.9666250944137573,
      "learning_rate": 2.361754276747912e-06,
      "loss": 0.4869,
      "step": 255380
    },
    {
      "epoch": 104.54,
      "grad_norm": 1.62640380859375,
      "learning_rate": 2.3615956518381365e-06,
      "loss": 0.4954,
      "step": 255390
    },
    {
      "epoch": 104.54,
      "grad_norm": 1.7588939666748047,
      "learning_rate": 2.3614370268466126e-06,
      "loss": 0.4916,
      "step": 255400
    },
    {
      "epoch": 104.55,
      "grad_norm": 1.7768008708953857,
      "learning_rate": 2.3612784017740675e-06,
      "loss": 0.4706,
      "step": 255410
    },
    {
      "epoch": 104.55,
      "grad_norm": 1.9471194744110107,
      "learning_rate": 2.3611197766212274e-06,
      "loss": 0.4712,
      "step": 255420
    },
    {
      "epoch": 104.56,
      "grad_norm": 1.860520839691162,
      "learning_rate": 2.3609611513888188e-06,
      "loss": 0.4885,
      "step": 255430
    },
    {
      "epoch": 104.56,
      "grad_norm": 2.199939489364624,
      "learning_rate": 2.360802526077568e-06,
      "loss": 0.4938,
      "step": 255440
    },
    {
      "epoch": 104.56,
      "grad_norm": 1.6585651636123657,
      "learning_rate": 2.360643900688203e-06,
      "loss": 0.4725,
      "step": 255450
    },
    {
      "epoch": 104.57,
      "grad_norm": 2.230799674987793,
      "learning_rate": 2.3604852752214497e-06,
      "loss": 0.4821,
      "step": 255460
    },
    {
      "epoch": 104.57,
      "grad_norm": 1.9733943939208984,
      "learning_rate": 2.360326649678035e-06,
      "loss": 0.4756,
      "step": 255470
    },
    {
      "epoch": 104.58,
      "grad_norm": 2.4455795288085938,
      "learning_rate": 2.3601680240586845e-06,
      "loss": 0.4711,
      "step": 255480
    },
    {
      "epoch": 104.58,
      "grad_norm": 1.913893699645996,
      "learning_rate": 2.360009398364126e-06,
      "loss": 0.4592,
      "step": 255490
    },
    {
      "epoch": 104.58,
      "grad_norm": 2.2805707454681396,
      "learning_rate": 2.3598507725950853e-06,
      "loss": 0.4781,
      "step": 255500
    },
    {
      "epoch": 104.59,
      "grad_norm": 2.439690589904785,
      "learning_rate": 2.35969214675229e-06,
      "loss": 0.4507,
      "step": 255510
    },
    {
      "epoch": 104.59,
      "grad_norm": 2.0175461769104004,
      "learning_rate": 2.359533520836466e-06,
      "loss": 0.4834,
      "step": 255520
    },
    {
      "epoch": 104.6,
      "grad_norm": 2.422553777694702,
      "learning_rate": 2.35937489484834e-06,
      "loss": 0.4922,
      "step": 255530
    },
    {
      "epoch": 104.6,
      "grad_norm": 1.918012261390686,
      "learning_rate": 2.3592162687886384e-06,
      "loss": 0.4929,
      "step": 255540
    },
    {
      "epoch": 104.6,
      "grad_norm": 1.607208251953125,
      "learning_rate": 2.3590576426580886e-06,
      "loss": 0.4744,
      "step": 255550
    },
    {
      "epoch": 104.61,
      "grad_norm": 1.6134169101715088,
      "learning_rate": 2.358899016457417e-06,
      "loss": 0.4782,
      "step": 255560
    },
    {
      "epoch": 104.61,
      "grad_norm": 1.9964807033538818,
      "learning_rate": 2.35874039018735e-06,
      "loss": 0.4567,
      "step": 255570
    },
    {
      "epoch": 104.62,
      "grad_norm": 1.8107370138168335,
      "learning_rate": 2.358581763848614e-06,
      "loss": 0.4555,
      "step": 255580
    },
    {
      "epoch": 104.62,
      "grad_norm": 2.026716947555542,
      "learning_rate": 2.3584231374419366e-06,
      "loss": 0.4734,
      "step": 255590
    },
    {
      "epoch": 104.63,
      "grad_norm": 1.8450322151184082,
      "learning_rate": 2.358264510968043e-06,
      "loss": 0.4784,
      "step": 255600
    },
    {
      "epoch": 104.63,
      "grad_norm": 2.0275580883026123,
      "learning_rate": 2.3581058844276615e-06,
      "loss": 0.4778,
      "step": 255610
    },
    {
      "epoch": 104.63,
      "grad_norm": 6.379750728607178,
      "learning_rate": 2.357947257821518e-06,
      "loss": 0.4832,
      "step": 255620
    },
    {
      "epoch": 104.64,
      "grad_norm": 2.0562329292297363,
      "learning_rate": 2.3577886311503387e-06,
      "loss": 0.4785,
      "step": 255630
    },
    {
      "epoch": 104.64,
      "grad_norm": 1.9706075191497803,
      "learning_rate": 2.3576300044148506e-06,
      "loss": 0.4567,
      "step": 255640
    },
    {
      "epoch": 104.65,
      "grad_norm": 2.574653148651123,
      "learning_rate": 2.3574713776157804e-06,
      "loss": 0.4803,
      "step": 255650
    },
    {
      "epoch": 104.65,
      "grad_norm": 2.171800374984741,
      "learning_rate": 2.3573127507538544e-06,
      "loss": 0.4949,
      "step": 255660
    },
    {
      "epoch": 104.65,
      "grad_norm": 2.613469123840332,
      "learning_rate": 2.3571541238298e-06,
      "loss": 0.4753,
      "step": 255670
    },
    {
      "epoch": 104.66,
      "grad_norm": 1.8595280647277832,
      "learning_rate": 2.356995496844343e-06,
      "loss": 0.4961,
      "step": 255680
    },
    {
      "epoch": 104.66,
      "grad_norm": 2.049229860305786,
      "learning_rate": 2.356836869798211e-06,
      "loss": 0.4874,
      "step": 255690
    },
    {
      "epoch": 104.67,
      "grad_norm": 1.9241106510162354,
      "learning_rate": 2.3566782426921296e-06,
      "loss": 0.479,
      "step": 255700
    },
    {
      "epoch": 104.67,
      "grad_norm": 1.9115852117538452,
      "learning_rate": 2.3565196155268264e-06,
      "loss": 0.4768,
      "step": 255710
    },
    {
      "epoch": 104.67,
      "grad_norm": 1.999487280845642,
      "learning_rate": 2.3563609883030264e-06,
      "loss": 0.469,
      "step": 255720
    },
    {
      "epoch": 104.68,
      "grad_norm": 1.7112641334533691,
      "learning_rate": 2.3562023610214583e-06,
      "loss": 0.4718,
      "step": 255730
    },
    {
      "epoch": 104.68,
      "grad_norm": 1.5996333360671997,
      "learning_rate": 2.3560437336828476e-06,
      "loss": 0.4861,
      "step": 255740
    },
    {
      "epoch": 104.69,
      "grad_norm": 1.6866445541381836,
      "learning_rate": 2.355885106287921e-06,
      "loss": 0.4879,
      "step": 255750
    },
    {
      "epoch": 104.69,
      "grad_norm": 1.8876146078109741,
      "learning_rate": 2.3557264788374067e-06,
      "loss": 0.4598,
      "step": 255760
    },
    {
      "epoch": 104.7,
      "grad_norm": 2.3384299278259277,
      "learning_rate": 2.3555678513320288e-06,
      "loss": 0.4831,
      "step": 255770
    },
    {
      "epoch": 104.7,
      "grad_norm": 2.0156948566436768,
      "learning_rate": 2.3554092237725153e-06,
      "loss": 0.4549,
      "step": 255780
    },
    {
      "epoch": 104.7,
      "grad_norm": 1.8656361103057861,
      "learning_rate": 2.3552505961595925e-06,
      "loss": 0.4789,
      "step": 255790
    },
    {
      "epoch": 104.71,
      "grad_norm": 1.859904408454895,
      "learning_rate": 2.3550919684939877e-06,
      "loss": 0.4644,
      "step": 255800
    },
    {
      "epoch": 104.71,
      "grad_norm": 2.1843152046203613,
      "learning_rate": 2.3549333407764263e-06,
      "loss": 0.4712,
      "step": 255810
    },
    {
      "epoch": 104.72,
      "grad_norm": 2.2006888389587402,
      "learning_rate": 2.354774713007637e-06,
      "loss": 0.4759,
      "step": 255820
    },
    {
      "epoch": 104.72,
      "grad_norm": 1.747287392616272,
      "learning_rate": 2.3546160851883446e-06,
      "loss": 0.4888,
      "step": 255830
    },
    {
      "epoch": 104.72,
      "grad_norm": 1.7476333379745483,
      "learning_rate": 2.3544574573192766e-06,
      "loss": 0.4727,
      "step": 255840
    },
    {
      "epoch": 104.73,
      "grad_norm": 1.9576306343078613,
      "learning_rate": 2.354298829401159e-06,
      "loss": 0.4695,
      "step": 255850
    },
    {
      "epoch": 104.73,
      "grad_norm": 2.019209146499634,
      "learning_rate": 2.354140201434719e-06,
      "loss": 0.4703,
      "step": 255860
    },
    {
      "epoch": 104.74,
      "grad_norm": 1.9153916835784912,
      "learning_rate": 2.353981573420683e-06,
      "loss": 0.4822,
      "step": 255870
    },
    {
      "epoch": 104.74,
      "grad_norm": 1.629132628440857,
      "learning_rate": 2.3538229453597776e-06,
      "loss": 0.4954,
      "step": 255880
    },
    {
      "epoch": 104.74,
      "grad_norm": 1.9042401313781738,
      "learning_rate": 2.35366431725273e-06,
      "loss": 0.4593,
      "step": 255890
    },
    {
      "epoch": 104.75,
      "grad_norm": 2.831636667251587,
      "learning_rate": 2.353505689100266e-06,
      "loss": 0.4638,
      "step": 255900
    },
    {
      "epoch": 104.75,
      "grad_norm": 1.6759569644927979,
      "learning_rate": 2.353347060903113e-06,
      "loss": 0.4553,
      "step": 255910
    },
    {
      "epoch": 104.76,
      "grad_norm": 2.2021126747131348,
      "learning_rate": 2.3531884326619975e-06,
      "loss": 0.4797,
      "step": 255920
    },
    {
      "epoch": 104.76,
      "grad_norm": 2.489461898803711,
      "learning_rate": 2.3530298043776457e-06,
      "loss": 0.4659,
      "step": 255930
    },
    {
      "epoch": 104.76,
      "grad_norm": 1.9336590766906738,
      "learning_rate": 2.3528711760507847e-06,
      "loss": 0.4737,
      "step": 255940
    },
    {
      "epoch": 104.77,
      "grad_norm": 1.821062445640564,
      "learning_rate": 2.352712547682141e-06,
      "loss": 0.4907,
      "step": 255950
    },
    {
      "epoch": 104.77,
      "grad_norm": 1.7831928730010986,
      "learning_rate": 2.352553919272441e-06,
      "loss": 0.4789,
      "step": 255960
    },
    {
      "epoch": 104.78,
      "grad_norm": 1.6901577711105347,
      "learning_rate": 2.352395290822412e-06,
      "loss": 0.4579,
      "step": 255970
    },
    {
      "epoch": 104.78,
      "grad_norm": 2.1576836109161377,
      "learning_rate": 2.35223666233278e-06,
      "loss": 0.47,
      "step": 255980
    },
    {
      "epoch": 104.79,
      "grad_norm": 1.793277621269226,
      "learning_rate": 2.3520780338042723e-06,
      "loss": 0.4791,
      "step": 255990
    },
    {
      "epoch": 104.79,
      "grad_norm": 1.7854584455490112,
      "learning_rate": 2.3519194052376146e-06,
      "loss": 0.4704,
      "step": 256000
    },
    {
      "epoch": 104.79,
      "grad_norm": 1.4707995653152466,
      "learning_rate": 2.351760776633534e-06,
      "loss": 0.4571,
      "step": 256010
    },
    {
      "epoch": 104.8,
      "grad_norm": 1.7284959554672241,
      "learning_rate": 2.3516021479927577e-06,
      "loss": 0.4713,
      "step": 256020
    },
    {
      "epoch": 104.8,
      "grad_norm": 2.0548713207244873,
      "learning_rate": 2.351443519316012e-06,
      "loss": 0.4579,
      "step": 256030
    },
    {
      "epoch": 104.81,
      "grad_norm": 2.078974723815918,
      "learning_rate": 2.3512848906040233e-06,
      "loss": 0.4491,
      "step": 256040
    },
    {
      "epoch": 104.81,
      "grad_norm": 1.8483864068984985,
      "learning_rate": 2.351126261857519e-06,
      "loss": 0.4681,
      "step": 256050
    },
    {
      "epoch": 104.81,
      "grad_norm": 1.9191704988479614,
      "learning_rate": 2.3509676330772247e-06,
      "loss": 0.4671,
      "step": 256060
    },
    {
      "epoch": 104.82,
      "grad_norm": 1.8539248704910278,
      "learning_rate": 2.3508090042638673e-06,
      "loss": 0.475,
      "step": 256070
    },
    {
      "epoch": 104.82,
      "grad_norm": 1.8102176189422607,
      "learning_rate": 2.350650375418174e-06,
      "loss": 0.4723,
      "step": 256080
    },
    {
      "epoch": 104.83,
      "grad_norm": 2.5956883430480957,
      "learning_rate": 2.3504917465408713e-06,
      "loss": 0.4685,
      "step": 256090
    },
    {
      "epoch": 104.83,
      "grad_norm": 1.8382134437561035,
      "learning_rate": 2.3503331176326844e-06,
      "loss": 0.4916,
      "step": 256100
    },
    {
      "epoch": 104.83,
      "grad_norm": 1.5776640176773071,
      "learning_rate": 2.3501744886943423e-06,
      "loss": 0.4791,
      "step": 256110
    },
    {
      "epoch": 104.84,
      "grad_norm": 1.7411929368972778,
      "learning_rate": 2.3500158597265713e-06,
      "loss": 0.4764,
      "step": 256120
    },
    {
      "epoch": 104.84,
      "grad_norm": 2.1844775676727295,
      "learning_rate": 2.349857230730097e-06,
      "loss": 0.4838,
      "step": 256130
    },
    {
      "epoch": 104.85,
      "grad_norm": 2.01741361618042,
      "learning_rate": 2.3496986017056462e-06,
      "loss": 0.4959,
      "step": 256140
    },
    {
      "epoch": 104.85,
      "grad_norm": 1.812957525253296,
      "learning_rate": 2.3495399726539457e-06,
      "loss": 0.4662,
      "step": 256150
    },
    {
      "epoch": 104.85,
      "grad_norm": 1.5053986310958862,
      "learning_rate": 2.349381343575722e-06,
      "loss": 0.4577,
      "step": 256160
    },
    {
      "epoch": 104.86,
      "grad_norm": 2.670781373977661,
      "learning_rate": 2.3492227144717024e-06,
      "loss": 0.4651,
      "step": 256170
    },
    {
      "epoch": 104.86,
      "grad_norm": 2.8655247688293457,
      "learning_rate": 2.3490640853426127e-06,
      "loss": 0.4953,
      "step": 256180
    },
    {
      "epoch": 104.87,
      "grad_norm": 1.8143783807754517,
      "learning_rate": 2.34890545618918e-06,
      "loss": 0.4724,
      "step": 256190
    },
    {
      "epoch": 104.87,
      "grad_norm": 1.8082830905914307,
      "learning_rate": 2.3487468270121314e-06,
      "loss": 0.4589,
      "step": 256200
    },
    {
      "epoch": 104.88,
      "grad_norm": 2.280064105987549,
      "learning_rate": 2.348588197812193e-06,
      "loss": 0.4843,
      "step": 256210
    },
    {
      "epoch": 104.88,
      "grad_norm": 2.080636739730835,
      "learning_rate": 2.3484295685900914e-06,
      "loss": 0.4917,
      "step": 256220
    },
    {
      "epoch": 104.88,
      "grad_norm": 1.9218255281448364,
      "learning_rate": 2.3482709393465535e-06,
      "loss": 0.4652,
      "step": 256230
    },
    {
      "epoch": 104.89,
      "grad_norm": 1.9029544591903687,
      "learning_rate": 2.348112310082306e-06,
      "loss": 0.4769,
      "step": 256240
    },
    {
      "epoch": 104.89,
      "grad_norm": 2.0892903804779053,
      "learning_rate": 2.3479536807980756e-06,
      "loss": 0.4821,
      "step": 256250
    },
    {
      "epoch": 104.9,
      "grad_norm": 1.876455307006836,
      "learning_rate": 2.347795051494588e-06,
      "loss": 0.4641,
      "step": 256260
    },
    {
      "epoch": 104.9,
      "grad_norm": 1.6458475589752197,
      "learning_rate": 2.347636422172572e-06,
      "loss": 0.4717,
      "step": 256270
    },
    {
      "epoch": 104.9,
      "grad_norm": 1.792918086051941,
      "learning_rate": 2.3474777928327514e-06,
      "loss": 0.4673,
      "step": 256280
    },
    {
      "epoch": 104.91,
      "grad_norm": 2.089113235473633,
      "learning_rate": 2.347319163475855e-06,
      "loss": 0.4795,
      "step": 256290
    },
    {
      "epoch": 104.91,
      "grad_norm": 2.141279458999634,
      "learning_rate": 2.3471605341026087e-06,
      "loss": 0.4814,
      "step": 256300
    },
    {
      "epoch": 104.92,
      "grad_norm": 1.8720192909240723,
      "learning_rate": 2.3470019047137395e-06,
      "loss": 0.488,
      "step": 256310
    },
    {
      "epoch": 104.92,
      "grad_norm": 2.069624662399292,
      "learning_rate": 2.346843275309974e-06,
      "loss": 0.4934,
      "step": 256320
    },
    {
      "epoch": 104.92,
      "grad_norm": 1.8321468830108643,
      "learning_rate": 2.3466846458920384e-06,
      "loss": 0.4695,
      "step": 256330
    },
    {
      "epoch": 104.93,
      "grad_norm": 1.772965431213379,
      "learning_rate": 2.34652601646066e-06,
      "loss": 0.4961,
      "step": 256340
    },
    {
      "epoch": 104.93,
      "grad_norm": 2.0132737159729004,
      "learning_rate": 2.3463673870165646e-06,
      "loss": 0.465,
      "step": 256350
    },
    {
      "epoch": 104.94,
      "grad_norm": 1.7690552473068237,
      "learning_rate": 2.3462087575604796e-06,
      "loss": 0.4911,
      "step": 256360
    },
    {
      "epoch": 104.94,
      "grad_norm": 2.2920725345611572,
      "learning_rate": 2.346050128093131e-06,
      "loss": 0.4824,
      "step": 256370
    },
    {
      "epoch": 104.94,
      "grad_norm": 1.8136745691299438,
      "learning_rate": 2.345891498615246e-06,
      "loss": 0.4631,
      "step": 256380
    },
    {
      "epoch": 104.95,
      "grad_norm": 2.1796703338623047,
      "learning_rate": 2.3457328691275513e-06,
      "loss": 0.476,
      "step": 256390
    },
    {
      "epoch": 104.95,
      "grad_norm": 1.8489477634429932,
      "learning_rate": 2.3455742396307734e-06,
      "loss": 0.4678,
      "step": 256400
    },
    {
      "epoch": 104.96,
      "grad_norm": 2.2169315814971924,
      "learning_rate": 2.3454156101256392e-06,
      "loss": 0.4877,
      "step": 256410
    },
    {
      "epoch": 104.96,
      "grad_norm": 2.2876899242401123,
      "learning_rate": 2.345256980612875e-06,
      "loss": 0.498,
      "step": 256420
    },
    {
      "epoch": 104.97,
      "grad_norm": 1.6208381652832031,
      "learning_rate": 2.3450983510932073e-06,
      "loss": 0.4743,
      "step": 256430
    },
    {
      "epoch": 104.97,
      "grad_norm": 2.4137778282165527,
      "learning_rate": 2.344939721567363e-06,
      "loss": 0.4753,
      "step": 256440
    },
    {
      "epoch": 104.97,
      "grad_norm": 1.6636161804199219,
      "learning_rate": 2.344781092036069e-06,
      "loss": 0.4826,
      "step": 256450
    },
    {
      "epoch": 104.98,
      "grad_norm": 1.5260374546051025,
      "learning_rate": 2.3446224625000517e-06,
      "loss": 0.4722,
      "step": 256460
    },
    {
      "epoch": 104.98,
      "grad_norm": 2.2541427612304688,
      "learning_rate": 2.344463832960037e-06,
      "loss": 0.4664,
      "step": 256470
    },
    {
      "epoch": 104.99,
      "grad_norm": 2.0214717388153076,
      "learning_rate": 2.344305203416754e-06,
      "loss": 0.4817,
      "step": 256480
    },
    {
      "epoch": 104.99,
      "grad_norm": 1.8849886655807495,
      "learning_rate": 2.3441465738709265e-06,
      "loss": 0.4965,
      "step": 256490
    },
    {
      "epoch": 104.99,
      "grad_norm": 1.9716321229934692,
      "learning_rate": 2.3439879443232826e-06,
      "loss": 0.4568,
      "step": 256500
    },
    {
      "epoch": 105.0,
      "grad_norm": 1.8314112424850464,
      "learning_rate": 2.3438293147745483e-06,
      "loss": 0.4749,
      "step": 256510
    },
    {
      "epoch": 105.0,
      "eval_loss": 0.474666565656662,
      "eval_runtime": 54.4905,
      "eval_samples_per_second": 63.295,
      "eval_steps_per_second": 7.928,
      "step": 256515
    },
    {
      "epoch": 105.0,
      "grad_norm": 1.636638879776001,
      "learning_rate": 2.343670685225451e-06,
      "loss": 0.4837,
      "step": 256520
    },
    {
      "epoch": 105.01,
      "grad_norm": 2.3581655025482178,
      "learning_rate": 2.343512055676717e-06,
      "loss": 0.4735,
      "step": 256530
    },
    {
      "epoch": 105.01,
      "grad_norm": 1.8897641897201538,
      "learning_rate": 2.3433534261290735e-06,
      "loss": 0.4767,
      "step": 256540
    },
    {
      "epoch": 105.01,
      "grad_norm": 1.5913721323013306,
      "learning_rate": 2.343194796583246e-06,
      "loss": 0.4708,
      "step": 256550
    },
    {
      "epoch": 105.02,
      "grad_norm": 1.806599497795105,
      "learning_rate": 2.343036167039962e-06,
      "loss": 0.4941,
      "step": 256560
    },
    {
      "epoch": 105.02,
      "grad_norm": 2.0856010913848877,
      "learning_rate": 2.3428775374999483e-06,
      "loss": 0.4771,
      "step": 256570
    },
    {
      "epoch": 105.03,
      "grad_norm": 1.564001202583313,
      "learning_rate": 2.342718907963931e-06,
      "loss": 0.4768,
      "step": 256580
    },
    {
      "epoch": 105.03,
      "grad_norm": 1.8436776399612427,
      "learning_rate": 2.342560278432637e-06,
      "loss": 0.4698,
      "step": 256590
    },
    {
      "epoch": 105.03,
      "grad_norm": 1.9107418060302734,
      "learning_rate": 2.3424016489067922e-06,
      "loss": 0.4804,
      "step": 256600
    },
    {
      "epoch": 105.04,
      "grad_norm": 2.334855556488037,
      "learning_rate": 2.342243019387125e-06,
      "loss": 0.4692,
      "step": 256610
    },
    {
      "epoch": 105.04,
      "grad_norm": 2.1196324825286865,
      "learning_rate": 2.3420843898743604e-06,
      "loss": 0.4878,
      "step": 256620
    },
    {
      "epoch": 105.05,
      "grad_norm": 1.5132485628128052,
      "learning_rate": 2.3419257603692266e-06,
      "loss": 0.4621,
      "step": 256630
    },
    {
      "epoch": 105.05,
      "grad_norm": 2.0693843364715576,
      "learning_rate": 2.3417671308724488e-06,
      "loss": 0.4626,
      "step": 256640
    },
    {
      "epoch": 105.06,
      "grad_norm": 2.692431926727295,
      "learning_rate": 2.341608501384754e-06,
      "loss": 0.4677,
      "step": 256650
    },
    {
      "epoch": 105.06,
      "grad_norm": 1.8337976932525635,
      "learning_rate": 2.341449871906869e-06,
      "loss": 0.4588,
      "step": 256660
    },
    {
      "epoch": 105.06,
      "grad_norm": 1.7969974279403687,
      "learning_rate": 2.341291242439521e-06,
      "loss": 0.4683,
      "step": 256670
    },
    {
      "epoch": 105.07,
      "grad_norm": 1.7823710441589355,
      "learning_rate": 2.341132612983436e-06,
      "loss": 0.4659,
      "step": 256680
    },
    {
      "epoch": 105.07,
      "grad_norm": 2.246267318725586,
      "learning_rate": 2.3409739835393405e-06,
      "loss": 0.4586,
      "step": 256690
    },
    {
      "epoch": 105.08,
      "grad_norm": 2.452214479446411,
      "learning_rate": 2.3408153541079617e-06,
      "loss": 0.4691,
      "step": 256700
    },
    {
      "epoch": 105.08,
      "grad_norm": 2.2272982597351074,
      "learning_rate": 2.3406567246900257e-06,
      "loss": 0.4708,
      "step": 256710
    },
    {
      "epoch": 105.08,
      "grad_norm": 1.6647355556488037,
      "learning_rate": 2.3404980952862597e-06,
      "loss": 0.461,
      "step": 256720
    },
    {
      "epoch": 105.09,
      "grad_norm": 2.440209150314331,
      "learning_rate": 2.3403394658973905e-06,
      "loss": 0.4524,
      "step": 256730
    },
    {
      "epoch": 105.09,
      "grad_norm": 1.8867555856704712,
      "learning_rate": 2.3401808365241444e-06,
      "loss": 0.4852,
      "step": 256740
    },
    {
      "epoch": 105.1,
      "grad_norm": 1.6104031801223755,
      "learning_rate": 2.340022207167248e-06,
      "loss": 0.4601,
      "step": 256750
    },
    {
      "epoch": 105.1,
      "grad_norm": 1.7070717811584473,
      "learning_rate": 2.339863577827428e-06,
      "loss": 0.4513,
      "step": 256760
    },
    {
      "epoch": 105.1,
      "grad_norm": 2.5340380668640137,
      "learning_rate": 2.3397049485054114e-06,
      "loss": 0.476,
      "step": 256770
    },
    {
      "epoch": 105.11,
      "grad_norm": 1.6380807161331177,
      "learning_rate": 2.3395463192019244e-06,
      "loss": 0.4471,
      "step": 256780
    },
    {
      "epoch": 105.11,
      "grad_norm": 1.541702389717102,
      "learning_rate": 2.339387689917694e-06,
      "loss": 0.4833,
      "step": 256790
    },
    {
      "epoch": 105.12,
      "grad_norm": 1.6463905572891235,
      "learning_rate": 2.339229060653446e-06,
      "loss": 0.4769,
      "step": 256800
    },
    {
      "epoch": 105.12,
      "grad_norm": 1.7543613910675049,
      "learning_rate": 2.3390704314099082e-06,
      "loss": 0.4828,
      "step": 256810
    },
    {
      "epoch": 105.12,
      "grad_norm": 2.378412961959839,
      "learning_rate": 2.3389118021878067e-06,
      "loss": 0.4741,
      "step": 256820
    },
    {
      "epoch": 105.13,
      "grad_norm": 2.4244091510772705,
      "learning_rate": 2.338753172987868e-06,
      "loss": 0.4809,
      "step": 256830
    },
    {
      "epoch": 105.13,
      "grad_norm": 1.9460779428482056,
      "learning_rate": 2.3385945438108195e-06,
      "loss": 0.4661,
      "step": 256840
    },
    {
      "epoch": 105.14,
      "grad_norm": 1.882467269897461,
      "learning_rate": 2.3384359146573877e-06,
      "loss": 0.4577,
      "step": 256850
    },
    {
      "epoch": 105.14,
      "grad_norm": 2.710869073867798,
      "learning_rate": 2.338277285528298e-06,
      "loss": 0.4641,
      "step": 256860
    },
    {
      "epoch": 105.15,
      "grad_norm": 2.2664365768432617,
      "learning_rate": 2.3381186564242784e-06,
      "loss": 0.49,
      "step": 256870
    },
    {
      "epoch": 105.15,
      "grad_norm": 1.7198442220687866,
      "learning_rate": 2.337960027346055e-06,
      "loss": 0.4739,
      "step": 256880
    },
    {
      "epoch": 105.15,
      "grad_norm": 1.8667997121810913,
      "learning_rate": 2.3378013982943546e-06,
      "loss": 0.4614,
      "step": 256890
    },
    {
      "epoch": 105.16,
      "grad_norm": 1.6711896657943726,
      "learning_rate": 2.3376427692699027e-06,
      "loss": 0.4848,
      "step": 256900
    },
    {
      "epoch": 105.16,
      "grad_norm": 1.5156428813934326,
      "learning_rate": 2.3374841402734283e-06,
      "loss": 0.4818,
      "step": 256910
    },
    {
      "epoch": 105.17,
      "grad_norm": 1.782254934310913,
      "learning_rate": 2.3373255113056564e-06,
      "loss": 0.4732,
      "step": 256920
    },
    {
      "epoch": 105.17,
      "grad_norm": 2.157148838043213,
      "learning_rate": 2.3371668823673148e-06,
      "loss": 0.4832,
      "step": 256930
    },
    {
      "epoch": 105.17,
      "grad_norm": 1.9447565078735352,
      "learning_rate": 2.3370082534591287e-06,
      "loss": 0.4708,
      "step": 256940
    },
    {
      "epoch": 105.18,
      "grad_norm": 1.9569950103759766,
      "learning_rate": 2.336849624581826e-06,
      "loss": 0.4794,
      "step": 256950
    },
    {
      "epoch": 105.18,
      "grad_norm": 1.7183506488800049,
      "learning_rate": 2.3366909957361323e-06,
      "loss": 0.4634,
      "step": 256960
    },
    {
      "epoch": 105.19,
      "grad_norm": 1.8299570083618164,
      "learning_rate": 2.3365323669227754e-06,
      "loss": 0.4697,
      "step": 256970
    },
    {
      "epoch": 105.19,
      "grad_norm": 1.7596631050109863,
      "learning_rate": 2.3363737381424807e-06,
      "loss": 0.4551,
      "step": 256980
    },
    {
      "epoch": 105.19,
      "grad_norm": 3.288116931915283,
      "learning_rate": 2.3362151093959767e-06,
      "loss": 0.4743,
      "step": 256990
    },
    {
      "epoch": 105.2,
      "grad_norm": 1.9522000551223755,
      "learning_rate": 2.336056480683988e-06,
      "loss": 0.4819,
      "step": 257000
    },
    {
      "epoch": 105.2,
      "grad_norm": 1.7048439979553223,
      "learning_rate": 2.3358978520072423e-06,
      "loss": 0.4726,
      "step": 257010
    },
    {
      "epoch": 105.21,
      "grad_norm": 1.870571255683899,
      "learning_rate": 2.335739223366466e-06,
      "loss": 0.4613,
      "step": 257020
    },
    {
      "epoch": 105.21,
      "grad_norm": 1.966184139251709,
      "learning_rate": 2.335580594762386e-06,
      "loss": 0.4805,
      "step": 257030
    },
    {
      "epoch": 105.21,
      "grad_norm": 2.647993564605713,
      "learning_rate": 2.335421966195728e-06,
      "loss": 0.49,
      "step": 257040
    },
    {
      "epoch": 105.22,
      "grad_norm": 2.153383255004883,
      "learning_rate": 2.3352633376672203e-06,
      "loss": 0.4774,
      "step": 257050
    },
    {
      "epoch": 105.22,
      "grad_norm": 2.1431002616882324,
      "learning_rate": 2.335104709177588e-06,
      "loss": 0.4599,
      "step": 257060
    },
    {
      "epoch": 105.23,
      "grad_norm": 1.8907771110534668,
      "learning_rate": 2.334946080727559e-06,
      "loss": 0.4679,
      "step": 257070
    },
    {
      "epoch": 105.23,
      "grad_norm": 2.175342559814453,
      "learning_rate": 2.3347874523178596e-06,
      "loss": 0.4941,
      "step": 257080
    },
    {
      "epoch": 105.24,
      "grad_norm": 1.7115824222564697,
      "learning_rate": 2.334628823949215e-06,
      "loss": 0.4806,
      "step": 257090
    },
    {
      "epoch": 105.24,
      "grad_norm": 1.9221880435943604,
      "learning_rate": 2.334470195622354e-06,
      "loss": 0.4873,
      "step": 257100
    },
    {
      "epoch": 105.24,
      "grad_norm": 1.7972590923309326,
      "learning_rate": 2.334311567338002e-06,
      "loss": 0.4811,
      "step": 257110
    },
    {
      "epoch": 105.25,
      "grad_norm": 2.386852741241455,
      "learning_rate": 2.3341529390968863e-06,
      "loss": 0.4748,
      "step": 257120
    },
    {
      "epoch": 105.25,
      "grad_norm": 2.7982068061828613,
      "learning_rate": 2.333994310899734e-06,
      "loss": 0.4871,
      "step": 257130
    },
    {
      "epoch": 105.26,
      "grad_norm": 1.8984168767929077,
      "learning_rate": 2.33383568274727e-06,
      "loss": 0.4637,
      "step": 257140
    },
    {
      "epoch": 105.26,
      "grad_norm": 2.3599853515625,
      "learning_rate": 2.3336770546402224e-06,
      "loss": 0.4699,
      "step": 257150
    },
    {
      "epoch": 105.26,
      "grad_norm": 2.2178893089294434,
      "learning_rate": 2.333518426579317e-06,
      "loss": 0.4841,
      "step": 257160
    },
    {
      "epoch": 105.27,
      "grad_norm": 1.7704757452011108,
      "learning_rate": 2.333359798565281e-06,
      "loss": 0.4789,
      "step": 257170
    },
    {
      "epoch": 105.27,
      "grad_norm": 2.086531162261963,
      "learning_rate": 2.333201170598841e-06,
      "loss": 0.4545,
      "step": 257180
    },
    {
      "epoch": 105.28,
      "grad_norm": 1.9825022220611572,
      "learning_rate": 2.3330425426807234e-06,
      "loss": 0.4759,
      "step": 257190
    },
    {
      "epoch": 105.28,
      "grad_norm": 1.609188199043274,
      "learning_rate": 2.332883914811655e-06,
      "loss": 0.4802,
      "step": 257200
    },
    {
      "epoch": 105.28,
      "grad_norm": 2.088879108428955,
      "learning_rate": 2.3327252869923636e-06,
      "loss": 0.479,
      "step": 257210
    },
    {
      "epoch": 105.29,
      "grad_norm": 1.9805034399032593,
      "learning_rate": 2.3325666592235733e-06,
      "loss": 0.496,
      "step": 257220
    },
    {
      "epoch": 105.29,
      "grad_norm": 2.931323528289795,
      "learning_rate": 2.3324080315060127e-06,
      "loss": 0.4797,
      "step": 257230
    },
    {
      "epoch": 105.3,
      "grad_norm": 1.9147751331329346,
      "learning_rate": 2.3322494038404075e-06,
      "loss": 0.4599,
      "step": 257240
    },
    {
      "epoch": 105.3,
      "grad_norm": 1.6345996856689453,
      "learning_rate": 2.332090776227485e-06,
      "loss": 0.4691,
      "step": 257250
    },
    {
      "epoch": 105.3,
      "grad_norm": 1.9215683937072754,
      "learning_rate": 2.3319321486679717e-06,
      "loss": 0.4872,
      "step": 257260
    },
    {
      "epoch": 105.31,
      "grad_norm": 1.871092438697815,
      "learning_rate": 2.3317735211625933e-06,
      "loss": 0.4883,
      "step": 257270
    },
    {
      "epoch": 105.31,
      "grad_norm": 2.233325958251953,
      "learning_rate": 2.331614893712078e-06,
      "loss": 0.4469,
      "step": 257280
    },
    {
      "epoch": 105.32,
      "grad_norm": 2.277191638946533,
      "learning_rate": 2.331456266317152e-06,
      "loss": 0.4759,
      "step": 257290
    },
    {
      "epoch": 105.32,
      "grad_norm": 1.728100299835205,
      "learning_rate": 2.3312976389785413e-06,
      "loss": 0.4954,
      "step": 257300
    },
    {
      "epoch": 105.33,
      "grad_norm": 2.1918702125549316,
      "learning_rate": 2.331139011696973e-06,
      "loss": 0.4606,
      "step": 257310
    },
    {
      "epoch": 105.33,
      "grad_norm": 1.583543062210083,
      "learning_rate": 2.330980384473174e-06,
      "loss": 0.4685,
      "step": 257320
    },
    {
      "epoch": 105.33,
      "grad_norm": 1.6466786861419678,
      "learning_rate": 2.33082175730787e-06,
      "loss": 0.4757,
      "step": 257330
    },
    {
      "epoch": 105.34,
      "grad_norm": 1.879111647605896,
      "learning_rate": 2.330663130201789e-06,
      "loss": 0.487,
      "step": 257340
    },
    {
      "epoch": 105.34,
      "grad_norm": 1.8667765855789185,
      "learning_rate": 2.330504503155657e-06,
      "loss": 0.4566,
      "step": 257350
    },
    {
      "epoch": 105.35,
      "grad_norm": 1.8805314302444458,
      "learning_rate": 2.3303458761702e-06,
      "loss": 0.4598,
      "step": 257360
    },
    {
      "epoch": 105.35,
      "grad_norm": 1.6017751693725586,
      "learning_rate": 2.3301872492461456e-06,
      "loss": 0.4617,
      "step": 257370
    },
    {
      "epoch": 105.35,
      "grad_norm": 2.857299327850342,
      "learning_rate": 2.3300286223842196e-06,
      "loss": 0.4609,
      "step": 257380
    },
    {
      "epoch": 105.36,
      "grad_norm": 1.9674731492996216,
      "learning_rate": 2.3298699955851494e-06,
      "loss": 0.4658,
      "step": 257390
    },
    {
      "epoch": 105.36,
      "grad_norm": 1.5021923780441284,
      "learning_rate": 2.3297113688496614e-06,
      "loss": 0.4657,
      "step": 257400
    },
    {
      "epoch": 105.37,
      "grad_norm": 2.415865182876587,
      "learning_rate": 2.329552742178482e-06,
      "loss": 0.4924,
      "step": 257410
    },
    {
      "epoch": 105.37,
      "grad_norm": 1.8568514585494995,
      "learning_rate": 2.329394115572338e-06,
      "loss": 0.4707,
      "step": 257420
    },
    {
      "epoch": 105.37,
      "grad_norm": 2.049478054046631,
      "learning_rate": 2.3292354890319564e-06,
      "loss": 0.4692,
      "step": 257430
    },
    {
      "epoch": 105.38,
      "grad_norm": 1.8417630195617676,
      "learning_rate": 2.329076862558064e-06,
      "loss": 0.4677,
      "step": 257440
    },
    {
      "epoch": 105.38,
      "grad_norm": 2.0783941745758057,
      "learning_rate": 2.3289182361513863e-06,
      "loss": 0.449,
      "step": 257450
    },
    {
      "epoch": 105.39,
      "grad_norm": 2.09924578666687,
      "learning_rate": 2.3287596098126493e-06,
      "loss": 0.5011,
      "step": 257460
    },
    {
      "epoch": 105.39,
      "grad_norm": 2.4424526691436768,
      "learning_rate": 2.328600983542582e-06,
      "loss": 0.4841,
      "step": 257470
    },
    {
      "epoch": 105.4,
      "grad_norm": 2.163699150085449,
      "learning_rate": 2.3284423573419106e-06,
      "loss": 0.4669,
      "step": 257480
    },
    {
      "epoch": 105.4,
      "grad_norm": 1.8110074996948242,
      "learning_rate": 2.328283731211361e-06,
      "loss": 0.4745,
      "step": 257490
    },
    {
      "epoch": 105.4,
      "grad_norm": 1.8214458227157593,
      "learning_rate": 2.32812510515166e-06,
      "loss": 0.4785,
      "step": 257500
    },
    {
      "epoch": 105.41,
      "grad_norm": 1.6832165718078613,
      "learning_rate": 2.327966479163534e-06,
      "loss": 0.4796,
      "step": 257510
    },
    {
      "epoch": 105.41,
      "grad_norm": 2.039156913757324,
      "learning_rate": 2.32780785324771e-06,
      "loss": 0.4581,
      "step": 257520
    },
    {
      "epoch": 105.42,
      "grad_norm": 1.8051674365997314,
      "learning_rate": 2.3276492274049143e-06,
      "loss": 0.4764,
      "step": 257530
    },
    {
      "epoch": 105.42,
      "grad_norm": 1.58460533618927,
      "learning_rate": 2.327490601635874e-06,
      "loss": 0.4679,
      "step": 257540
    },
    {
      "epoch": 105.42,
      "grad_norm": 2.0321950912475586,
      "learning_rate": 2.327331975941315e-06,
      "loss": 0.4822,
      "step": 257550
    },
    {
      "epoch": 105.43,
      "grad_norm": 2.1567583084106445,
      "learning_rate": 2.327173350321965e-06,
      "loss": 0.4744,
      "step": 257560
    },
    {
      "epoch": 105.43,
      "grad_norm": 1.8608288764953613,
      "learning_rate": 2.3270147247785503e-06,
      "loss": 0.4804,
      "step": 257570
    },
    {
      "epoch": 105.44,
      "grad_norm": 1.5467865467071533,
      "learning_rate": 2.326856099311797e-06,
      "loss": 0.4733,
      "step": 257580
    },
    {
      "epoch": 105.44,
      "grad_norm": 1.843495488166809,
      "learning_rate": 2.3266974739224317e-06,
      "loss": 0.4793,
      "step": 257590
    },
    {
      "epoch": 105.44,
      "grad_norm": 2.12143874168396,
      "learning_rate": 2.3265388486111817e-06,
      "loss": 0.455,
      "step": 257600
    },
    {
      "epoch": 105.45,
      "grad_norm": 1.6315034627914429,
      "learning_rate": 2.326380223378773e-06,
      "loss": 0.4847,
      "step": 257610
    },
    {
      "epoch": 105.45,
      "grad_norm": 2.525850534439087,
      "learning_rate": 2.3262215982259325e-06,
      "loss": 0.464,
      "step": 257620
    },
    {
      "epoch": 105.46,
      "grad_norm": 1.4308785200119019,
      "learning_rate": 2.3260629731533874e-06,
      "loss": 0.494,
      "step": 257630
    },
    {
      "epoch": 105.46,
      "grad_norm": 1.8072900772094727,
      "learning_rate": 2.3259043481618635e-06,
      "loss": 0.4743,
      "step": 257640
    },
    {
      "epoch": 105.46,
      "grad_norm": 2.460361957550049,
      "learning_rate": 2.325745723252087e-06,
      "loss": 0.4562,
      "step": 257650
    },
    {
      "epoch": 105.47,
      "grad_norm": 2.0604867935180664,
      "learning_rate": 2.3255870984247864e-06,
      "loss": 0.477,
      "step": 257660
    },
    {
      "epoch": 105.47,
      "grad_norm": 2.0575966835021973,
      "learning_rate": 2.325428473680687e-06,
      "loss": 0.4735,
      "step": 257670
    },
    {
      "epoch": 105.48,
      "grad_norm": 2.3325531482696533,
      "learning_rate": 2.3252698490205158e-06,
      "loss": 0.4774,
      "step": 257680
    },
    {
      "epoch": 105.48,
      "grad_norm": 1.910692811012268,
      "learning_rate": 2.325111224444999e-06,
      "loss": 0.4827,
      "step": 257690
    },
    {
      "epoch": 105.49,
      "grad_norm": 1.7906869649887085,
      "learning_rate": 2.3249525999548635e-06,
      "loss": 0.4572,
      "step": 257700
    },
    {
      "epoch": 105.49,
      "grad_norm": 1.9824323654174805,
      "learning_rate": 2.3247939755508366e-06,
      "loss": 0.4686,
      "step": 257710
    },
    {
      "epoch": 105.49,
      "grad_norm": 2.0043416023254395,
      "learning_rate": 2.324635351233644e-06,
      "loss": 0.4841,
      "step": 257720
    },
    {
      "epoch": 105.5,
      "grad_norm": 1.4491522312164307,
      "learning_rate": 2.3244767270040123e-06,
      "loss": 0.4766,
      "step": 257730
    },
    {
      "epoch": 105.5,
      "grad_norm": 2.101048707962036,
      "learning_rate": 2.324318102862669e-06,
      "loss": 0.4753,
      "step": 257740
    },
    {
      "epoch": 105.51,
      "grad_norm": 2.116612195968628,
      "learning_rate": 2.3241594788103398e-06,
      "loss": 0.4647,
      "step": 257750
    },
    {
      "epoch": 105.51,
      "grad_norm": 1.9305232763290405,
      "learning_rate": 2.324000854847752e-06,
      "loss": 0.4899,
      "step": 257760
    },
    {
      "epoch": 105.51,
      "grad_norm": 1.850730538368225,
      "learning_rate": 2.323842230975632e-06,
      "loss": 0.4721,
      "step": 257770
    },
    {
      "epoch": 105.52,
      "grad_norm": 1.4604861736297607,
      "learning_rate": 2.323683607194706e-06,
      "loss": 0.4553,
      "step": 257780
    },
    {
      "epoch": 105.52,
      "grad_norm": 2.192526340484619,
      "learning_rate": 2.323524983505701e-06,
      "loss": 0.4666,
      "step": 257790
    },
    {
      "epoch": 105.53,
      "grad_norm": 1.9333617687225342,
      "learning_rate": 2.3233663599093446e-06,
      "loss": 0.4685,
      "step": 257800
    },
    {
      "epoch": 105.53,
      "grad_norm": 1.7958450317382812,
      "learning_rate": 2.323207736406362e-06,
      "loss": 0.4545,
      "step": 257810
    },
    {
      "epoch": 105.53,
      "grad_norm": 2.9106478691101074,
      "learning_rate": 2.32304911299748e-06,
      "loss": 0.4879,
      "step": 257820
    },
    {
      "epoch": 105.54,
      "grad_norm": 2.066736936569214,
      "learning_rate": 2.3228904896834255e-06,
      "loss": 0.4776,
      "step": 257830
    },
    {
      "epoch": 105.54,
      "grad_norm": 2.3576505184173584,
      "learning_rate": 2.3227318664649245e-06,
      "loss": 0.4698,
      "step": 257840
    },
    {
      "epoch": 105.55,
      "grad_norm": 1.7627787590026855,
      "learning_rate": 2.3225732433427058e-06,
      "loss": 0.4813,
      "step": 257850
    },
    {
      "epoch": 105.55,
      "grad_norm": 1.8683466911315918,
      "learning_rate": 2.322414620317494e-06,
      "loss": 0.4876,
      "step": 257860
    },
    {
      "epoch": 105.55,
      "grad_norm": 1.9410573244094849,
      "learning_rate": 2.3222559973900165e-06,
      "loss": 0.4716,
      "step": 257870
    },
    {
      "epoch": 105.56,
      "grad_norm": 1.801388144493103,
      "learning_rate": 2.3220973745609994e-06,
      "loss": 0.4856,
      "step": 257880
    },
    {
      "epoch": 105.56,
      "grad_norm": 1.9884051084518433,
      "learning_rate": 2.3219387518311695e-06,
      "loss": 0.4574,
      "step": 257890
    },
    {
      "epoch": 105.57,
      "grad_norm": 2.1841979026794434,
      "learning_rate": 2.3217801292012535e-06,
      "loss": 0.4553,
      "step": 257900
    },
    {
      "epoch": 105.57,
      "grad_norm": 2.1078648567199707,
      "learning_rate": 2.3216215066719786e-06,
      "loss": 0.4713,
      "step": 257910
    },
    {
      "epoch": 105.58,
      "grad_norm": 2.169314384460449,
      "learning_rate": 2.32146288424407e-06,
      "loss": 0.4622,
      "step": 257920
    },
    {
      "epoch": 105.58,
      "grad_norm": 2.3536250591278076,
      "learning_rate": 2.3213042619182568e-06,
      "loss": 0.4778,
      "step": 257930
    },
    {
      "epoch": 105.58,
      "grad_norm": 1.779888391494751,
      "learning_rate": 2.321145639695263e-06,
      "loss": 0.5016,
      "step": 257940
    },
    {
      "epoch": 105.59,
      "grad_norm": 1.928464651107788,
      "learning_rate": 2.320987017575816e-06,
      "loss": 0.4682,
      "step": 257950
    },
    {
      "epoch": 105.59,
      "grad_norm": 2.3793530464172363,
      "learning_rate": 2.3208283955606435e-06,
      "loss": 0.4565,
      "step": 257960
    },
    {
      "epoch": 105.6,
      "grad_norm": 2.290217399597168,
      "learning_rate": 2.320669773650471e-06,
      "loss": 0.4823,
      "step": 257970
    },
    {
      "epoch": 105.6,
      "grad_norm": 2.041818380355835,
      "learning_rate": 2.320511151846025e-06,
      "loss": 0.5042,
      "step": 257980
    },
    {
      "epoch": 105.6,
      "grad_norm": 2.1028122901916504,
      "learning_rate": 2.3203525301480328e-06,
      "loss": 0.4972,
      "step": 257990
    },
    {
      "epoch": 105.61,
      "grad_norm": 1.5249308347702026,
      "learning_rate": 2.320193908557221e-06,
      "loss": 0.4739,
      "step": 258000
    },
    {
      "epoch": 105.61,
      "grad_norm": 2.1351189613342285,
      "learning_rate": 2.3200352870743157e-06,
      "loss": 0.4738,
      "step": 258010
    },
    {
      "epoch": 105.62,
      "grad_norm": 1.8764111995697021,
      "learning_rate": 2.3198766657000447e-06,
      "loss": 0.4776,
      "step": 258020
    },
    {
      "epoch": 105.62,
      "grad_norm": 1.8611122369766235,
      "learning_rate": 2.3197180444351316e-06,
      "loss": 0.4759,
      "step": 258030
    },
    {
      "epoch": 105.62,
      "grad_norm": 1.940020203590393,
      "learning_rate": 2.3195594232803066e-06,
      "loss": 0.4798,
      "step": 258040
    },
    {
      "epoch": 105.63,
      "grad_norm": 1.7962747812271118,
      "learning_rate": 2.319400802236295e-06,
      "loss": 0.4764,
      "step": 258050
    },
    {
      "epoch": 105.63,
      "grad_norm": 2.0032341480255127,
      "learning_rate": 2.3192421813038233e-06,
      "loss": 0.4544,
      "step": 258060
    },
    {
      "epoch": 105.64,
      "grad_norm": 2.200937271118164,
      "learning_rate": 2.3190835604836186e-06,
      "loss": 0.4762,
      "step": 258070
    },
    {
      "epoch": 105.64,
      "grad_norm": 1.4055206775665283,
      "learning_rate": 2.3189249397764064e-06,
      "loss": 0.4727,
      "step": 258080
    },
    {
      "epoch": 105.64,
      "grad_norm": 1.940648078918457,
      "learning_rate": 2.318766319182914e-06,
      "loss": 0.4933,
      "step": 258090
    },
    {
      "epoch": 105.65,
      "grad_norm": 1.827040672302246,
      "learning_rate": 2.3186076987038678e-06,
      "loss": 0.451,
      "step": 258100
    },
    {
      "epoch": 105.65,
      "grad_norm": 2.254284381866455,
      "learning_rate": 2.318449078339995e-06,
      "loss": 0.4658,
      "step": 258110
    },
    {
      "epoch": 105.66,
      "grad_norm": 1.863660454750061,
      "learning_rate": 2.3182904580920215e-06,
      "loss": 0.4595,
      "step": 258120
    },
    {
      "epoch": 105.66,
      "grad_norm": 2.04909348487854,
      "learning_rate": 2.318131837960674e-06,
      "loss": 0.4608,
      "step": 258130
    },
    {
      "epoch": 105.67,
      "grad_norm": 1.8362250328063965,
      "learning_rate": 2.31797321794668e-06,
      "loss": 0.4703,
      "step": 258140
    },
    {
      "epoch": 105.67,
      "grad_norm": 1.8130460977554321,
      "learning_rate": 2.3178145980507647e-06,
      "loss": 0.4931,
      "step": 258150
    },
    {
      "epoch": 105.67,
      "grad_norm": 1.8163303136825562,
      "learning_rate": 2.3176559782736565e-06,
      "loss": 0.4752,
      "step": 258160
    },
    {
      "epoch": 105.68,
      "grad_norm": 1.8798245191574097,
      "learning_rate": 2.3174973586160803e-06,
      "loss": 0.4898,
      "step": 258170
    },
    {
      "epoch": 105.68,
      "grad_norm": 1.9701051712036133,
      "learning_rate": 2.3173387390787633e-06,
      "loss": 0.4821,
      "step": 258180
    },
    {
      "epoch": 105.69,
      "grad_norm": 1.6693840026855469,
      "learning_rate": 2.3171801196624323e-06,
      "loss": 0.5001,
      "step": 258190
    },
    {
      "epoch": 105.69,
      "grad_norm": 2.3092379570007324,
      "learning_rate": 2.3170215003678135e-06,
      "loss": 0.4661,
      "step": 258200
    },
    {
      "epoch": 105.69,
      "grad_norm": 1.862758755683899,
      "learning_rate": 2.3168628811956343e-06,
      "loss": 0.4728,
      "step": 258210
    },
    {
      "epoch": 105.7,
      "grad_norm": 1.851906418800354,
      "learning_rate": 2.3167042621466203e-06,
      "loss": 0.4777,
      "step": 258220
    },
    {
      "epoch": 105.7,
      "grad_norm": 1.9552699327468872,
      "learning_rate": 2.316545643221499e-06,
      "loss": 0.4845,
      "step": 258230
    },
    {
      "epoch": 105.71,
      "grad_norm": 1.6621067523956299,
      "learning_rate": 2.3163870244209965e-06,
      "loss": 0.4656,
      "step": 258240
    },
    {
      "epoch": 105.71,
      "grad_norm": 1.8556320667266846,
      "learning_rate": 2.3162284057458396e-06,
      "loss": 0.4698,
      "step": 258250
    },
    {
      "epoch": 105.71,
      "grad_norm": 2.142801284790039,
      "learning_rate": 2.316069787196755e-06,
      "loss": 0.4842,
      "step": 258260
    },
    {
      "epoch": 105.72,
      "grad_norm": 2.3136837482452393,
      "learning_rate": 2.315911168774469e-06,
      "loss": 0.4886,
      "step": 258270
    },
    {
      "epoch": 105.72,
      "grad_norm": 1.7669645547866821,
      "learning_rate": 2.3157525504797086e-06,
      "loss": 0.4666,
      "step": 258280
    },
    {
      "epoch": 105.73,
      "grad_norm": 1.9956871271133423,
      "learning_rate": 2.3155939323132005e-06,
      "loss": 0.4506,
      "step": 258290
    },
    {
      "epoch": 105.73,
      "grad_norm": 1.8558815717697144,
      "learning_rate": 2.3154353142756706e-06,
      "loss": 0.4782,
      "step": 258300
    },
    {
      "epoch": 105.73,
      "grad_norm": 1.7516508102416992,
      "learning_rate": 2.3152766963678455e-06,
      "loss": 0.4769,
      "step": 258310
    },
    {
      "epoch": 105.74,
      "grad_norm": 1.7437403202056885,
      "learning_rate": 2.3151180785904525e-06,
      "loss": 0.4643,
      "step": 258320
    },
    {
      "epoch": 105.74,
      "grad_norm": 1.8351707458496094,
      "learning_rate": 2.314959460944218e-06,
      "loss": 0.4912,
      "step": 258330
    },
    {
      "epoch": 105.75,
      "grad_norm": 1.989937663078308,
      "learning_rate": 2.3148008434298684e-06,
      "loss": 0.4753,
      "step": 258340
    },
    {
      "epoch": 105.75,
      "grad_norm": 3.095529079437256,
      "learning_rate": 2.3146422260481304e-06,
      "loss": 0.4626,
      "step": 258350
    },
    {
      "epoch": 105.76,
      "grad_norm": 1.59714937210083,
      "learning_rate": 2.3144836087997307e-06,
      "loss": 0.4723,
      "step": 258360
    },
    {
      "epoch": 105.76,
      "grad_norm": 2.043764591217041,
      "learning_rate": 2.314324991685396e-06,
      "loss": 0.4899,
      "step": 258370
    },
    {
      "epoch": 105.76,
      "grad_norm": 1.923672080039978,
      "learning_rate": 2.3141663747058525e-06,
      "loss": 0.4749,
      "step": 258380
    },
    {
      "epoch": 105.77,
      "grad_norm": 2.0063087940216064,
      "learning_rate": 2.314007757861827e-06,
      "loss": 0.4602,
      "step": 258390
    },
    {
      "epoch": 105.77,
      "grad_norm": 1.621700644493103,
      "learning_rate": 2.313849141154046e-06,
      "loss": 0.4739,
      "step": 258400
    },
    {
      "epoch": 105.78,
      "grad_norm": 2.006193161010742,
      "learning_rate": 2.3136905245832355e-06,
      "loss": 0.4819,
      "step": 258410
    },
    {
      "epoch": 105.78,
      "grad_norm": 2.04555606842041,
      "learning_rate": 2.313531908150123e-06,
      "loss": 0.4808,
      "step": 258420
    },
    {
      "epoch": 105.78,
      "grad_norm": 2.1107301712036133,
      "learning_rate": 2.3133732918554363e-06,
      "loss": 0.483,
      "step": 258430
    },
    {
      "epoch": 105.79,
      "grad_norm": 1.6699563264846802,
      "learning_rate": 2.3132146756998996e-06,
      "loss": 0.4824,
      "step": 258440
    },
    {
      "epoch": 105.79,
      "grad_norm": 4.533162593841553,
      "learning_rate": 2.3130560596842407e-06,
      "loss": 0.4625,
      "step": 258450
    },
    {
      "epoch": 105.8,
      "grad_norm": 1.9045498371124268,
      "learning_rate": 2.3128974438091863e-06,
      "loss": 0.4816,
      "step": 258460
    },
    {
      "epoch": 105.8,
      "grad_norm": 1.5451571941375732,
      "learning_rate": 2.3127388280754623e-06,
      "loss": 0.4842,
      "step": 258470
    },
    {
      "epoch": 105.8,
      "grad_norm": 2.11124587059021,
      "learning_rate": 2.3125802124837955e-06,
      "loss": 0.4753,
      "step": 258480
    },
    {
      "epoch": 105.81,
      "grad_norm": 2.0854105949401855,
      "learning_rate": 2.3124215970349126e-06,
      "loss": 0.4817,
      "step": 258490
    },
    {
      "epoch": 105.81,
      "grad_norm": 2.207064628601074,
      "learning_rate": 2.3122629817295407e-06,
      "loss": 0.4602,
      "step": 258500
    },
    {
      "epoch": 105.82,
      "grad_norm": 1.9023303985595703,
      "learning_rate": 2.3121043665684055e-06,
      "loss": 0.4689,
      "step": 258510
    },
    {
      "epoch": 105.82,
      "grad_norm": 1.6345971822738647,
      "learning_rate": 2.311945751552235e-06,
      "loss": 0.4662,
      "step": 258520
    },
    {
      "epoch": 105.82,
      "grad_norm": 2.0433690547943115,
      "learning_rate": 2.311787136681754e-06,
      "loss": 0.4418,
      "step": 258530
    },
    {
      "epoch": 105.83,
      "grad_norm": 1.9859730005264282,
      "learning_rate": 2.3116285219576902e-06,
      "loss": 0.4643,
      "step": 258540
    },
    {
      "epoch": 105.83,
      "grad_norm": 2.240206480026245,
      "learning_rate": 2.3114699073807697e-06,
      "loss": 0.4789,
      "step": 258550
    },
    {
      "epoch": 105.84,
      "grad_norm": 1.7417672872543335,
      "learning_rate": 2.3113112929517195e-06,
      "loss": 0.4862,
      "step": 258560
    },
    {
      "epoch": 105.84,
      "grad_norm": 2.2007696628570557,
      "learning_rate": 2.311152678671266e-06,
      "loss": 0.4516,
      "step": 258570
    },
    {
      "epoch": 105.85,
      "grad_norm": 2.138977527618408,
      "learning_rate": 2.3109940645401356e-06,
      "loss": 0.4503,
      "step": 258580
    },
    {
      "epoch": 105.85,
      "grad_norm": 2.029508113861084,
      "learning_rate": 2.3108354505590545e-06,
      "loss": 0.4789,
      "step": 258590
    },
    {
      "epoch": 105.85,
      "grad_norm": 2.3615095615386963,
      "learning_rate": 2.3106768367287507e-06,
      "loss": 0.4788,
      "step": 258600
    },
    {
      "epoch": 105.86,
      "grad_norm": 1.9446802139282227,
      "learning_rate": 2.3105182230499494e-06,
      "loss": 0.4778,
      "step": 258610
    },
    {
      "epoch": 105.86,
      "grad_norm": 1.693651556968689,
      "learning_rate": 2.310359609523378e-06,
      "loss": 0.4603,
      "step": 258620
    },
    {
      "epoch": 105.87,
      "grad_norm": 2.197108507156372,
      "learning_rate": 2.310200996149763e-06,
      "loss": 0.4525,
      "step": 258630
    },
    {
      "epoch": 105.87,
      "grad_norm": 2.297004222869873,
      "learning_rate": 2.3100423829298305e-06,
      "loss": 0.4534,
      "step": 258640
    },
    {
      "epoch": 105.87,
      "grad_norm": 2.3117575645446777,
      "learning_rate": 2.309883769864308e-06,
      "loss": 0.4736,
      "step": 258650
    },
    {
      "epoch": 105.88,
      "grad_norm": 1.9658678770065308,
      "learning_rate": 2.309725156953921e-06,
      "loss": 0.4583,
      "step": 258660
    },
    {
      "epoch": 105.88,
      "grad_norm": 1.9937453269958496,
      "learning_rate": 2.3095665441993967e-06,
      "loss": 0.4751,
      "step": 258670
    },
    {
      "epoch": 105.89,
      "grad_norm": 1.9649180173873901,
      "learning_rate": 2.3094079316014615e-06,
      "loss": 0.4798,
      "step": 258680
    },
    {
      "epoch": 105.89,
      "grad_norm": 1.7496992349624634,
      "learning_rate": 2.3092493191608415e-06,
      "loss": 0.4639,
      "step": 258690
    },
    {
      "epoch": 105.89,
      "grad_norm": 1.9790157079696655,
      "learning_rate": 2.3090907068782643e-06,
      "loss": 0.4695,
      "step": 258700
    },
    {
      "epoch": 105.9,
      "grad_norm": 1.9463876485824585,
      "learning_rate": 2.308932094754456e-06,
      "loss": 0.4671,
      "step": 258710
    },
    {
      "epoch": 105.9,
      "grad_norm": 1.8857208490371704,
      "learning_rate": 2.308773482790143e-06,
      "loss": 0.4777,
      "step": 258720
    },
    {
      "epoch": 105.91,
      "grad_norm": 1.68930983543396,
      "learning_rate": 2.3086148709860515e-06,
      "loss": 0.4628,
      "step": 258730
    },
    {
      "epoch": 105.91,
      "grad_norm": 1.8662490844726562,
      "learning_rate": 2.3084562593429097e-06,
      "loss": 0.4686,
      "step": 258740
    },
    {
      "epoch": 105.91,
      "grad_norm": 2.2562191486358643,
      "learning_rate": 2.3082976478614423e-06,
      "loss": 0.4635,
      "step": 258750
    },
    {
      "epoch": 105.92,
      "grad_norm": 1.9863197803497314,
      "learning_rate": 2.3081390365423765e-06,
      "loss": 0.4656,
      "step": 258760
    },
    {
      "epoch": 105.92,
      "grad_norm": 2.0025534629821777,
      "learning_rate": 2.307980425386439e-06,
      "loss": 0.4635,
      "step": 258770
    },
    {
      "epoch": 105.93,
      "grad_norm": 1.9102189540863037,
      "learning_rate": 2.307821814394356e-06,
      "loss": 0.4811,
      "step": 258780
    },
    {
      "epoch": 105.93,
      "grad_norm": 2.4156742095947266,
      "learning_rate": 2.307663203566855e-06,
      "loss": 0.4647,
      "step": 258790
    },
    {
      "epoch": 105.94,
      "grad_norm": 1.9376556873321533,
      "learning_rate": 2.3075045929046626e-06,
      "loss": 0.4825,
      "step": 258800
    },
    {
      "epoch": 105.94,
      "grad_norm": 2.066115140914917,
      "learning_rate": 2.307345982408504e-06,
      "loss": 0.4849,
      "step": 258810
    },
    {
      "epoch": 105.94,
      "grad_norm": 2.117374897003174,
      "learning_rate": 2.3071873720791066e-06,
      "loss": 0.4686,
      "step": 258820
    },
    {
      "epoch": 105.95,
      "grad_norm": 2.1147806644439697,
      "learning_rate": 2.3070287619171975e-06,
      "loss": 0.4569,
      "step": 258830
    },
    {
      "epoch": 105.95,
      "grad_norm": 2.034782648086548,
      "learning_rate": 2.306870151923502e-06,
      "loss": 0.4687,
      "step": 258840
    },
    {
      "epoch": 105.96,
      "grad_norm": 2.014310121536255,
      "learning_rate": 2.306711542098748e-06,
      "loss": 0.4612,
      "step": 258850
    },
    {
      "epoch": 105.96,
      "grad_norm": 1.959340214729309,
      "learning_rate": 2.306552932443661e-06,
      "loss": 0.4844,
      "step": 258860
    },
    {
      "epoch": 105.96,
      "grad_norm": 2.124159574508667,
      "learning_rate": 2.3063943229589688e-06,
      "loss": 0.4966,
      "step": 258870
    },
    {
      "epoch": 105.97,
      "grad_norm": 1.7615265846252441,
      "learning_rate": 2.3062357136453966e-06,
      "loss": 0.4558,
      "step": 258880
    },
    {
      "epoch": 105.97,
      "grad_norm": 1.6979583501815796,
      "learning_rate": 2.3060771045036714e-06,
      "loss": 0.4587,
      "step": 258890
    },
    {
      "epoch": 105.98,
      "grad_norm": 1.8915868997573853,
      "learning_rate": 2.3059184955345202e-06,
      "loss": 0.467,
      "step": 258900
    },
    {
      "epoch": 105.98,
      "grad_norm": 2.365281105041504,
      "learning_rate": 2.3057598867386686e-06,
      "loss": 0.4648,
      "step": 258910
    },
    {
      "epoch": 105.98,
      "grad_norm": 2.149824380874634,
      "learning_rate": 2.3056012781168446e-06,
      "loss": 0.4835,
      "step": 258920
    },
    {
      "epoch": 105.99,
      "grad_norm": 2.389014482498169,
      "learning_rate": 2.3054426696697736e-06,
      "loss": 0.4578,
      "step": 258930
    },
    {
      "epoch": 105.99,
      "grad_norm": 1.8205467462539673,
      "learning_rate": 2.305284061398183e-06,
      "loss": 0.465,
      "step": 258940
    },
    {
      "epoch": 106.0,
      "grad_norm": 2.155534267425537,
      "learning_rate": 2.3051254533027983e-06,
      "loss": 0.4915,
      "step": 258950
    },
    {
      "epoch": 106.0,
      "eval_loss": 0.47192227840423584,
      "eval_runtime": 52.468,
      "eval_samples_per_second": 65.735,
      "eval_steps_per_second": 8.234,
      "step": 258958
    },
    {
      "epoch": 106.0,
      "grad_norm": 1.9559242725372314,
      "learning_rate": 2.3049668453843474e-06,
      "loss": 0.5041,
      "step": 258960
    },
    {
      "epoch": 106.0,
      "grad_norm": 1.72059965133667,
      "learning_rate": 2.3048082376435545e-06,
      "loss": 0.464,
      "step": 258970
    },
    {
      "epoch": 106.01,
      "grad_norm": 1.8391587734222412,
      "learning_rate": 2.3046496300811485e-06,
      "loss": 0.5042,
      "step": 258980
    },
    {
      "epoch": 106.01,
      "grad_norm": 1.6230783462524414,
      "learning_rate": 2.304491022697856e-06,
      "loss": 0.4693,
      "step": 258990
    },
    {
      "epoch": 106.02,
      "grad_norm": 1.565252661705017,
      "learning_rate": 2.304332415494402e-06,
      "loss": 0.4565,
      "step": 259000
    },
    {
      "epoch": 106.02,
      "grad_norm": 2.1954314708709717,
      "learning_rate": 2.304173808471515e-06,
      "loss": 0.4743,
      "step": 259010
    },
    {
      "epoch": 106.03,
      "grad_norm": 2.505291700363159,
      "learning_rate": 2.3040152016299196e-06,
      "loss": 0.4708,
      "step": 259020
    },
    {
      "epoch": 106.03,
      "grad_norm": 1.582674264907837,
      "learning_rate": 2.303856594970343e-06,
      "loss": 0.4746,
      "step": 259030
    },
    {
      "epoch": 106.03,
      "grad_norm": 1.9266455173492432,
      "learning_rate": 2.3036979884935123e-06,
      "loss": 0.4534,
      "step": 259040
    },
    {
      "epoch": 106.04,
      "grad_norm": 1.931757926940918,
      "learning_rate": 2.3035393822001534e-06,
      "loss": 0.4917,
      "step": 259050
    },
    {
      "epoch": 106.04,
      "grad_norm": 1.899880290031433,
      "learning_rate": 2.3033807760909933e-06,
      "loss": 0.4616,
      "step": 259060
    },
    {
      "epoch": 106.05,
      "grad_norm": 1.964657187461853,
      "learning_rate": 2.3032221701667583e-06,
      "loss": 0.4789,
      "step": 259070
    },
    {
      "epoch": 106.05,
      "grad_norm": 1.9509336948394775,
      "learning_rate": 2.303063564428175e-06,
      "loss": 0.4718,
      "step": 259080
    },
    {
      "epoch": 106.05,
      "grad_norm": 1.5591496229171753,
      "learning_rate": 2.30290495887597e-06,
      "loss": 0.4815,
      "step": 259090
    },
    {
      "epoch": 106.06,
      "grad_norm": 1.7743593454360962,
      "learning_rate": 2.3027463535108705e-06,
      "loss": 0.4707,
      "step": 259100
    },
    {
      "epoch": 106.06,
      "grad_norm": 1.8137567043304443,
      "learning_rate": 2.3025877483336016e-06,
      "loss": 0.4819,
      "step": 259110
    },
    {
      "epoch": 106.07,
      "grad_norm": 2.0266058444976807,
      "learning_rate": 2.3024291433448903e-06,
      "loss": 0.4779,
      "step": 259120
    },
    {
      "epoch": 106.07,
      "grad_norm": 1.4147759675979614,
      "learning_rate": 2.302270538545464e-06,
      "loss": 0.4726,
      "step": 259130
    },
    {
      "epoch": 106.07,
      "grad_norm": 1.9116424322128296,
      "learning_rate": 2.3021119339360484e-06,
      "loss": 0.4888,
      "step": 259140
    },
    {
      "epoch": 106.08,
      "grad_norm": 1.521730661392212,
      "learning_rate": 2.30195332951737e-06,
      "loss": 0.4573,
      "step": 259150
    },
    {
      "epoch": 106.08,
      "grad_norm": 2.02154541015625,
      "learning_rate": 2.3017947252901557e-06,
      "loss": 0.485,
      "step": 259160
    },
    {
      "epoch": 106.09,
      "grad_norm": 1.8394498825073242,
      "learning_rate": 2.3016361212551324e-06,
      "loss": 0.4592,
      "step": 259170
    },
    {
      "epoch": 106.09,
      "grad_norm": 1.8569620847702026,
      "learning_rate": 2.3014775174130264e-06,
      "loss": 0.4736,
      "step": 259180
    },
    {
      "epoch": 106.09,
      "grad_norm": 1.761931300163269,
      "learning_rate": 2.301318913764564e-06,
      "loss": 0.4845,
      "step": 259190
    },
    {
      "epoch": 106.1,
      "grad_norm": 2.1846675872802734,
      "learning_rate": 2.3011603103104718e-06,
      "loss": 0.4881,
      "step": 259200
    },
    {
      "epoch": 106.1,
      "grad_norm": 1.9716200828552246,
      "learning_rate": 2.3010017070514763e-06,
      "loss": 0.4785,
      "step": 259210
    },
    {
      "epoch": 106.11,
      "grad_norm": 1.7819185256958008,
      "learning_rate": 2.300843103988304e-06,
      "loss": 0.4782,
      "step": 259220
    },
    {
      "epoch": 106.11,
      "grad_norm": 1.8157511949539185,
      "learning_rate": 2.300684501121682e-06,
      "loss": 0.4887,
      "step": 259230
    },
    {
      "epoch": 106.12,
      "grad_norm": 2.0620429515838623,
      "learning_rate": 2.3005258984523363e-06,
      "loss": 0.4721,
      "step": 259240
    },
    {
      "epoch": 106.12,
      "grad_norm": 1.6592848300933838,
      "learning_rate": 2.300367295980993e-06,
      "loss": 0.4734,
      "step": 259250
    },
    {
      "epoch": 106.12,
      "grad_norm": 1.8073971271514893,
      "learning_rate": 2.3002086937083797e-06,
      "loss": 0.4859,
      "step": 259260
    },
    {
      "epoch": 106.13,
      "grad_norm": 1.9901272058486938,
      "learning_rate": 2.300050091635222e-06,
      "loss": 0.4802,
      "step": 259270
    },
    {
      "epoch": 106.13,
      "grad_norm": 1.7151517868041992,
      "learning_rate": 2.2998914897622468e-06,
      "loss": 0.4871,
      "step": 259280
    },
    {
      "epoch": 106.14,
      "grad_norm": 1.7438961267471313,
      "learning_rate": 2.2997328880901807e-06,
      "loss": 0.465,
      "step": 259290
    },
    {
      "epoch": 106.14,
      "grad_norm": 1.6388757228851318,
      "learning_rate": 2.2995742866197504e-06,
      "loss": 0.4663,
      "step": 259300
    },
    {
      "epoch": 106.14,
      "grad_norm": 2.661057710647583,
      "learning_rate": 2.299415685351682e-06,
      "loss": 0.4671,
      "step": 259310
    },
    {
      "epoch": 106.15,
      "grad_norm": 1.5834312438964844,
      "learning_rate": 2.2992570842867028e-06,
      "loss": 0.4497,
      "step": 259320
    },
    {
      "epoch": 106.15,
      "grad_norm": 1.916538953781128,
      "learning_rate": 2.299098483425538e-06,
      "loss": 0.4668,
      "step": 259330
    },
    {
      "epoch": 106.16,
      "grad_norm": 2.183997631072998,
      "learning_rate": 2.298939882768915e-06,
      "loss": 0.4781,
      "step": 259340
    },
    {
      "epoch": 106.16,
      "grad_norm": 2.0685887336730957,
      "learning_rate": 2.2987812823175594e-06,
      "loss": 0.4821,
      "step": 259350
    },
    {
      "epoch": 106.16,
      "grad_norm": 2.0330700874328613,
      "learning_rate": 2.2986226820721996e-06,
      "loss": 0.4624,
      "step": 259360
    },
    {
      "epoch": 106.17,
      "grad_norm": 2.114161491394043,
      "learning_rate": 2.2984640820335616e-06,
      "loss": 0.4711,
      "step": 259370
    },
    {
      "epoch": 106.17,
      "grad_norm": 1.9202790260314941,
      "learning_rate": 2.2983054822023704e-06,
      "loss": 0.4511,
      "step": 259380
    },
    {
      "epoch": 106.18,
      "grad_norm": 1.9632905721664429,
      "learning_rate": 2.2981468825793537e-06,
      "loss": 0.4718,
      "step": 259390
    },
    {
      "epoch": 106.18,
      "grad_norm": 1.795447587966919,
      "learning_rate": 2.297988283165238e-06,
      "loss": 0.4616,
      "step": 259400
    },
    {
      "epoch": 106.19,
      "grad_norm": 1.8150451183319092,
      "learning_rate": 2.2978296839607497e-06,
      "loss": 0.4724,
      "step": 259410
    },
    {
      "epoch": 106.19,
      "grad_norm": 1.8028359413146973,
      "learning_rate": 2.297671084966615e-06,
      "loss": 0.4812,
      "step": 259420
    },
    {
      "epoch": 106.19,
      "grad_norm": 1.8682128190994263,
      "learning_rate": 2.2975124861835607e-06,
      "loss": 0.4694,
      "step": 259430
    },
    {
      "epoch": 106.2,
      "grad_norm": 1.7299995422363281,
      "learning_rate": 2.2973538876123133e-06,
      "loss": 0.4845,
      "step": 259440
    },
    {
      "epoch": 106.2,
      "grad_norm": 2.0964174270629883,
      "learning_rate": 2.2971952892535995e-06,
      "loss": 0.451,
      "step": 259450
    },
    {
      "epoch": 106.21,
      "grad_norm": 2.1410973072052,
      "learning_rate": 2.297036691108146e-06,
      "loss": 0.4694,
      "step": 259460
    },
    {
      "epoch": 106.21,
      "grad_norm": 2.9028117656707764,
      "learning_rate": 2.2968780931766784e-06,
      "loss": 0.4821,
      "step": 259470
    },
    {
      "epoch": 106.21,
      "grad_norm": 2.0657432079315186,
      "learning_rate": 2.2967194954599237e-06,
      "loss": 0.481,
      "step": 259480
    },
    {
      "epoch": 106.22,
      "grad_norm": 1.9435372352600098,
      "learning_rate": 2.2965608979586085e-06,
      "loss": 0.467,
      "step": 259490
    },
    {
      "epoch": 106.22,
      "grad_norm": 1.711596131324768,
      "learning_rate": 2.2964023006734594e-06,
      "loss": 0.446,
      "step": 259500
    },
    {
      "epoch": 106.23,
      "grad_norm": 1.8262238502502441,
      "learning_rate": 2.2962437036052025e-06,
      "loss": 0.4707,
      "step": 259510
    },
    {
      "epoch": 106.23,
      "grad_norm": 2.1393942832946777,
      "learning_rate": 2.296085106754565e-06,
      "loss": 0.4712,
      "step": 259520
    },
    {
      "epoch": 106.23,
      "grad_norm": 1.6518632173538208,
      "learning_rate": 2.2959265101222723e-06,
      "loss": 0.4776,
      "step": 259530
    },
    {
      "epoch": 106.24,
      "grad_norm": 1.8754745721817017,
      "learning_rate": 2.2957679137090518e-06,
      "loss": 0.4655,
      "step": 259540
    },
    {
      "epoch": 106.24,
      "grad_norm": 1.7217599153518677,
      "learning_rate": 2.29560931751563e-06,
      "loss": 0.4672,
      "step": 259550
    },
    {
      "epoch": 106.25,
      "grad_norm": 2.0435574054718018,
      "learning_rate": 2.295450721542733e-06,
      "loss": 0.4795,
      "step": 259560
    },
    {
      "epoch": 106.25,
      "grad_norm": 1.7037636041641235,
      "learning_rate": 2.295292125791088e-06,
      "loss": 0.4701,
      "step": 259570
    },
    {
      "epoch": 106.25,
      "grad_norm": 1.690180778503418,
      "learning_rate": 2.2951335302614208e-06,
      "loss": 0.4817,
      "step": 259580
    },
    {
      "epoch": 106.26,
      "grad_norm": 2.035647392272949,
      "learning_rate": 2.2949749349544587e-06,
      "loss": 0.4647,
      "step": 259590
    },
    {
      "epoch": 106.26,
      "grad_norm": 2.20279860496521,
      "learning_rate": 2.294816339870927e-06,
      "loss": 0.4811,
      "step": 259600
    },
    {
      "epoch": 106.27,
      "grad_norm": 2.1385695934295654,
      "learning_rate": 2.294657745011553e-06,
      "loss": 0.4616,
      "step": 259610
    },
    {
      "epoch": 106.27,
      "grad_norm": 2.508622169494629,
      "learning_rate": 2.2944991503770625e-06,
      "loss": 0.466,
      "step": 259620
    },
    {
      "epoch": 106.28,
      "grad_norm": 1.869224190711975,
      "learning_rate": 2.294340555968183e-06,
      "loss": 0.4771,
      "step": 259630
    },
    {
      "epoch": 106.28,
      "grad_norm": 2.4760172367095947,
      "learning_rate": 2.2941819617856405e-06,
      "loss": 0.4643,
      "step": 259640
    },
    {
      "epoch": 106.28,
      "grad_norm": 2.2139675617218018,
      "learning_rate": 2.2940233678301615e-06,
      "loss": 0.4697,
      "step": 259650
    },
    {
      "epoch": 106.29,
      "grad_norm": 2.90653395652771,
      "learning_rate": 2.293864774102472e-06,
      "loss": 0.4647,
      "step": 259660
    },
    {
      "epoch": 106.29,
      "grad_norm": 1.5968960523605347,
      "learning_rate": 2.2937061806032998e-06,
      "loss": 0.4521,
      "step": 259670
    },
    {
      "epoch": 106.3,
      "grad_norm": 1.9479438066482544,
      "learning_rate": 2.2935475873333707e-06,
      "loss": 0.4698,
      "step": 259680
    },
    {
      "epoch": 106.3,
      "grad_norm": 1.7325111627578735,
      "learning_rate": 2.29338899429341e-06,
      "loss": 0.4799,
      "step": 259690
    },
    {
      "epoch": 106.3,
      "grad_norm": 1.8574168682098389,
      "learning_rate": 2.2932304014841462e-06,
      "loss": 0.4788,
      "step": 259700
    },
    {
      "epoch": 106.31,
      "grad_norm": 2.1672677993774414,
      "learning_rate": 2.2930718089063044e-06,
      "loss": 0.4715,
      "step": 259710
    },
    {
      "epoch": 106.31,
      "grad_norm": 1.6451730728149414,
      "learning_rate": 2.292913216560612e-06,
      "loss": 0.4708,
      "step": 259720
    },
    {
      "epoch": 106.32,
      "grad_norm": 2.261399745941162,
      "learning_rate": 2.2927546244477937e-06,
      "loss": 0.46,
      "step": 259730
    },
    {
      "epoch": 106.32,
      "grad_norm": 1.9180893898010254,
      "learning_rate": 2.2925960325685784e-06,
      "loss": 0.4644,
      "step": 259740
    },
    {
      "epoch": 106.32,
      "grad_norm": 1.7284151315689087,
      "learning_rate": 2.2924374409236913e-06,
      "loss": 0.4821,
      "step": 259750
    },
    {
      "epoch": 106.33,
      "grad_norm": 2.04286789894104,
      "learning_rate": 2.292278849513859e-06,
      "loss": 0.4881,
      "step": 259760
    },
    {
      "epoch": 106.33,
      "grad_norm": 1.9594687223434448,
      "learning_rate": 2.2921202583398085e-06,
      "loss": 0.4686,
      "step": 259770
    },
    {
      "epoch": 106.34,
      "grad_norm": 1.8533610105514526,
      "learning_rate": 2.2919616674022655e-06,
      "loss": 0.4479,
      "step": 259780
    },
    {
      "epoch": 106.34,
      "grad_norm": 6.2758259773254395,
      "learning_rate": 2.291803076701957e-06,
      "loss": 0.4645,
      "step": 259790
    },
    {
      "epoch": 106.34,
      "grad_norm": 1.9695031642913818,
      "learning_rate": 2.291644486239609e-06,
      "loss": 0.4591,
      "step": 259800
    },
    {
      "epoch": 106.35,
      "grad_norm": 1.6833206415176392,
      "learning_rate": 2.291485896015949e-06,
      "loss": 0.4643,
      "step": 259810
    },
    {
      "epoch": 106.35,
      "grad_norm": 2.210780143737793,
      "learning_rate": 2.2913273060317025e-06,
      "loss": 0.4763,
      "step": 259820
    },
    {
      "epoch": 106.36,
      "grad_norm": 1.9922432899475098,
      "learning_rate": 2.291168716287596e-06,
      "loss": 0.4666,
      "step": 259830
    },
    {
      "epoch": 106.36,
      "grad_norm": 1.690829873085022,
      "learning_rate": 2.291010126784356e-06,
      "loss": 0.4726,
      "step": 259840
    },
    {
      "epoch": 106.37,
      "grad_norm": 1.7765614986419678,
      "learning_rate": 2.2908515375227098e-06,
      "loss": 0.4903,
      "step": 259850
    },
    {
      "epoch": 106.37,
      "grad_norm": 2.026632308959961,
      "learning_rate": 2.2906929485033826e-06,
      "loss": 0.476,
      "step": 259860
    },
    {
      "epoch": 106.37,
      "grad_norm": 1.570439338684082,
      "learning_rate": 2.290534359727102e-06,
      "loss": 0.4758,
      "step": 259870
    },
    {
      "epoch": 106.38,
      "grad_norm": 1.693415641784668,
      "learning_rate": 2.2903757711945935e-06,
      "loss": 0.4767,
      "step": 259880
    },
    {
      "epoch": 106.38,
      "grad_norm": 2.840885877609253,
      "learning_rate": 2.290217182906585e-06,
      "loss": 0.4759,
      "step": 259890
    },
    {
      "epoch": 106.39,
      "grad_norm": 2.1554677486419678,
      "learning_rate": 2.2900585948638015e-06,
      "loss": 0.4591,
      "step": 259900
    },
    {
      "epoch": 106.39,
      "grad_norm": 2.552635669708252,
      "learning_rate": 2.289900007066969e-06,
      "loss": 0.4658,
      "step": 259910
    },
    {
      "epoch": 106.39,
      "grad_norm": 1.6684504747390747,
      "learning_rate": 2.289741419516816e-06,
      "loss": 0.4847,
      "step": 259920
    },
    {
      "epoch": 106.4,
      "grad_norm": 1.4827038049697876,
      "learning_rate": 2.2895828322140677e-06,
      "loss": 0.4808,
      "step": 259930
    },
    {
      "epoch": 106.4,
      "grad_norm": 2.3055419921875,
      "learning_rate": 2.2894242451594512e-06,
      "loss": 0.4787,
      "step": 259940
    },
    {
      "epoch": 106.41,
      "grad_norm": 1.9461982250213623,
      "learning_rate": 2.289265658353693e-06,
      "loss": 0.4599,
      "step": 259950
    },
    {
      "epoch": 106.41,
      "grad_norm": 2.1965131759643555,
      "learning_rate": 2.289107071797518e-06,
      "loss": 0.4751,
      "step": 259960
    },
    {
      "epoch": 106.41,
      "grad_norm": 1.7263474464416504,
      "learning_rate": 2.2889484854916546e-06,
      "loss": 0.4686,
      "step": 259970
    },
    {
      "epoch": 106.42,
      "grad_norm": 1.6701287031173706,
      "learning_rate": 2.2887898994368283e-06,
      "loss": 0.4624,
      "step": 259980
    },
    {
      "epoch": 106.42,
      "grad_norm": 2.2415826320648193,
      "learning_rate": 2.2886313136337653e-06,
      "loss": 0.478,
      "step": 259990
    },
    {
      "epoch": 106.43,
      "grad_norm": 2.356550455093384,
      "learning_rate": 2.288472728083193e-06,
      "loss": 0.4925,
      "step": 260000
    },
    {
      "epoch": 106.43,
      "grad_norm": 1.5477879047393799,
      "learning_rate": 2.288314142785837e-06,
      "loss": 0.472,
      "step": 260010
    },
    {
      "epoch": 106.43,
      "grad_norm": 1.5916376113891602,
      "learning_rate": 2.288155557742424e-06,
      "loss": 0.4649,
      "step": 260020
    },
    {
      "epoch": 106.44,
      "grad_norm": 1.9855533838272095,
      "learning_rate": 2.2879969729536807e-06,
      "loss": 0.4626,
      "step": 260030
    },
    {
      "epoch": 106.44,
      "grad_norm": 1.925040364265442,
      "learning_rate": 2.2878383884203336e-06,
      "loss": 0.5064,
      "step": 260040
    },
    {
      "epoch": 106.45,
      "grad_norm": 1.9553864002227783,
      "learning_rate": 2.287679804143109e-06,
      "loss": 0.4733,
      "step": 260050
    },
    {
      "epoch": 106.45,
      "grad_norm": 1.9710191488265991,
      "learning_rate": 2.2875212201227325e-06,
      "loss": 0.4767,
      "step": 260060
    },
    {
      "epoch": 106.46,
      "grad_norm": 2.515498399734497,
      "learning_rate": 2.287362636359932e-06,
      "loss": 0.4697,
      "step": 260070
    },
    {
      "epoch": 106.46,
      "grad_norm": 1.7350636720657349,
      "learning_rate": 2.287204052855433e-06,
      "loss": 0.4694,
      "step": 260080
    },
    {
      "epoch": 106.46,
      "grad_norm": 1.9723515510559082,
      "learning_rate": 2.2870454696099624e-06,
      "loss": 0.4609,
      "step": 260090
    },
    {
      "epoch": 106.47,
      "grad_norm": 2.0734403133392334,
      "learning_rate": 2.286886886624246e-06,
      "loss": 0.4749,
      "step": 260100
    },
    {
      "epoch": 106.47,
      "grad_norm": 2.021973133087158,
      "learning_rate": 2.2867283038990106e-06,
      "loss": 0.4603,
      "step": 260110
    },
    {
      "epoch": 106.48,
      "grad_norm": 1.497931718826294,
      "learning_rate": 2.2865697214349836e-06,
      "loss": 0.4793,
      "step": 260120
    },
    {
      "epoch": 106.48,
      "grad_norm": 1.7696552276611328,
      "learning_rate": 2.28641113923289e-06,
      "loss": 0.4721,
      "step": 260130
    },
    {
      "epoch": 106.48,
      "grad_norm": 1.6967473030090332,
      "learning_rate": 2.286252557293457e-06,
      "loss": 0.4829,
      "step": 260140
    },
    {
      "epoch": 106.49,
      "grad_norm": 1.6672483682632446,
      "learning_rate": 2.286093975617411e-06,
      "loss": 0.4707,
      "step": 260150
    },
    {
      "epoch": 106.49,
      "grad_norm": 1.9498637914657593,
      "learning_rate": 2.285935394205478e-06,
      "loss": 0.479,
      "step": 260160
    },
    {
      "epoch": 106.5,
      "grad_norm": 1.8811002969741821,
      "learning_rate": 2.2857768130583854e-06,
      "loss": 0.4823,
      "step": 260170
    },
    {
      "epoch": 106.5,
      "grad_norm": 2.229505777359009,
      "learning_rate": 2.285618232176859e-06,
      "loss": 0.4815,
      "step": 260180
    },
    {
      "epoch": 106.5,
      "grad_norm": 1.5229861736297607,
      "learning_rate": 2.285459651561625e-06,
      "loss": 0.4759,
      "step": 260190
    },
    {
      "epoch": 106.51,
      "grad_norm": 2.2591819763183594,
      "learning_rate": 2.28530107121341e-06,
      "loss": 0.4762,
      "step": 260200
    },
    {
      "epoch": 106.51,
      "grad_norm": 1.849305510520935,
      "learning_rate": 2.28514249113294e-06,
      "loss": 0.4727,
      "step": 260210
    },
    {
      "epoch": 106.52,
      "grad_norm": 3.4059672355651855,
      "learning_rate": 2.2849839113209424e-06,
      "loss": 0.4614,
      "step": 260220
    },
    {
      "epoch": 106.52,
      "grad_norm": 1.7981088161468506,
      "learning_rate": 2.284825331778143e-06,
      "loss": 0.4739,
      "step": 260230
    },
    {
      "epoch": 106.52,
      "grad_norm": 1.9888197183609009,
      "learning_rate": 2.2846667525052687e-06,
      "loss": 0.4665,
      "step": 260240
    },
    {
      "epoch": 106.53,
      "grad_norm": 1.7875282764434814,
      "learning_rate": 2.2845081735030455e-06,
      "loss": 0.472,
      "step": 260250
    },
    {
      "epoch": 106.53,
      "grad_norm": 2.1368277072906494,
      "learning_rate": 2.2843495947722e-06,
      "loss": 0.4714,
      "step": 260260
    },
    {
      "epoch": 106.54,
      "grad_norm": 1.922943353652954,
      "learning_rate": 2.2841910163134586e-06,
      "loss": 0.5034,
      "step": 260270
    },
    {
      "epoch": 106.54,
      "grad_norm": 1.988716721534729,
      "learning_rate": 2.2840324381275473e-06,
      "loss": 0.4489,
      "step": 260280
    },
    {
      "epoch": 106.55,
      "grad_norm": 2.6450448036193848,
      "learning_rate": 2.283873860215192e-06,
      "loss": 0.4688,
      "step": 260290
    },
    {
      "epoch": 106.55,
      "grad_norm": 1.6759480237960815,
      "learning_rate": 2.283715282577121e-06,
      "loss": 0.4557,
      "step": 260300
    },
    {
      "epoch": 106.55,
      "grad_norm": 2.2262866497039795,
      "learning_rate": 2.2835567052140605e-06,
      "loss": 0.4577,
      "step": 260310
    },
    {
      "epoch": 106.56,
      "grad_norm": 2.3431556224823,
      "learning_rate": 2.283398128126735e-06,
      "loss": 0.4877,
      "step": 260320
    },
    {
      "epoch": 106.56,
      "grad_norm": 2.380141496658325,
      "learning_rate": 2.2832395513158726e-06,
      "loss": 0.4728,
      "step": 260330
    },
    {
      "epoch": 106.57,
      "grad_norm": 1.4969732761383057,
      "learning_rate": 2.2830809747821994e-06,
      "loss": 0.4763,
      "step": 260340
    },
    {
      "epoch": 106.57,
      "grad_norm": 2.4207003116607666,
      "learning_rate": 2.282922398526441e-06,
      "loss": 0.4913,
      "step": 260350
    },
    {
      "epoch": 106.57,
      "grad_norm": 6.833361625671387,
      "learning_rate": 2.2827638225493253e-06,
      "loss": 0.4587,
      "step": 260360
    },
    {
      "epoch": 106.58,
      "grad_norm": 1.585294485092163,
      "learning_rate": 2.282605246851577e-06,
      "loss": 0.4885,
      "step": 260370
    },
    {
      "epoch": 106.58,
      "grad_norm": 1.6786174774169922,
      "learning_rate": 2.2824466714339235e-06,
      "loss": 0.465,
      "step": 260380
    },
    {
      "epoch": 106.59,
      "grad_norm": 2.004762887954712,
      "learning_rate": 2.2822880962970915e-06,
      "loss": 0.469,
      "step": 260390
    },
    {
      "epoch": 106.59,
      "grad_norm": 1.717490553855896,
      "learning_rate": 2.2821295214418073e-06,
      "loss": 0.4455,
      "step": 260400
    },
    {
      "epoch": 106.59,
      "grad_norm": 1.977898120880127,
      "learning_rate": 2.281970946868796e-06,
      "loss": 0.4517,
      "step": 260410
    },
    {
      "epoch": 106.6,
      "grad_norm": 1.7037839889526367,
      "learning_rate": 2.281812372578785e-06,
      "loss": 0.4784,
      "step": 260420
    },
    {
      "epoch": 106.6,
      "grad_norm": 1.943480372428894,
      "learning_rate": 2.281653798572501e-06,
      "loss": 0.4604,
      "step": 260430
    },
    {
      "epoch": 106.61,
      "grad_norm": 1.7456203699111938,
      "learning_rate": 2.2814952248506702e-06,
      "loss": 0.4694,
      "step": 260440
    },
    {
      "epoch": 106.61,
      "grad_norm": 1.569820523262024,
      "learning_rate": 2.281336651414019e-06,
      "loss": 0.4679,
      "step": 260450
    },
    {
      "epoch": 106.61,
      "grad_norm": 2.0955185890197754,
      "learning_rate": 2.2811780782632733e-06,
      "loss": 0.4604,
      "step": 260460
    },
    {
      "epoch": 106.62,
      "grad_norm": 1.8938907384872437,
      "learning_rate": 2.28101950539916e-06,
      "loss": 0.4489,
      "step": 260470
    },
    {
      "epoch": 106.62,
      "grad_norm": 1.8149551153182983,
      "learning_rate": 2.280860932822405e-06,
      "loss": 0.5028,
      "step": 260480
    },
    {
      "epoch": 106.63,
      "grad_norm": 2.2532174587249756,
      "learning_rate": 2.280702360533735e-06,
      "loss": 0.4729,
      "step": 260490
    },
    {
      "epoch": 106.63,
      "grad_norm": 1.5539382696151733,
      "learning_rate": 2.2805437885338773e-06,
      "loss": 0.4649,
      "step": 260500
    },
    {
      "epoch": 106.64,
      "grad_norm": 2.1706347465515137,
      "learning_rate": 2.280385216823557e-06,
      "loss": 0.4765,
      "step": 260510
    },
    {
      "epoch": 106.64,
      "grad_norm": 1.6139127016067505,
      "learning_rate": 2.280226645403501e-06,
      "loss": 0.464,
      "step": 260520
    },
    {
      "epoch": 106.64,
      "grad_norm": 1.712689757347107,
      "learning_rate": 2.280068074274436e-06,
      "loss": 0.4513,
      "step": 260530
    },
    {
      "epoch": 106.65,
      "grad_norm": 1.6364089250564575,
      "learning_rate": 2.279909503437088e-06,
      "loss": 0.473,
      "step": 260540
    },
    {
      "epoch": 106.65,
      "grad_norm": 1.7843629121780396,
      "learning_rate": 2.2797509328921833e-06,
      "loss": 0.4677,
      "step": 260550
    },
    {
      "epoch": 106.66,
      "grad_norm": 2.0776474475860596,
      "learning_rate": 2.279592362640448e-06,
      "loss": 0.4859,
      "step": 260560
    },
    {
      "epoch": 106.66,
      "grad_norm": 1.777159333229065,
      "learning_rate": 2.2794337926826094e-06,
      "loss": 0.4685,
      "step": 260570
    },
    {
      "epoch": 106.66,
      "grad_norm": 1.954276442527771,
      "learning_rate": 2.279275223019393e-06,
      "loss": 0.4568,
      "step": 260580
    },
    {
      "epoch": 106.67,
      "grad_norm": 2.7907211780548096,
      "learning_rate": 2.279116653651526e-06,
      "loss": 0.4709,
      "step": 260590
    },
    {
      "epoch": 106.67,
      "grad_norm": 1.8525619506835938,
      "learning_rate": 2.278958084579734e-06,
      "loss": 0.468,
      "step": 260600
    },
    {
      "epoch": 106.68,
      "grad_norm": 2.5030879974365234,
      "learning_rate": 2.2787995158047437e-06,
      "loss": 0.4734,
      "step": 260610
    },
    {
      "epoch": 106.68,
      "grad_norm": 2.1450958251953125,
      "learning_rate": 2.278640947327282e-06,
      "loss": 0.4583,
      "step": 260620
    },
    {
      "epoch": 106.68,
      "grad_norm": 1.8355050086975098,
      "learning_rate": 2.2784823791480745e-06,
      "loss": 0.4757,
      "step": 260630
    },
    {
      "epoch": 106.69,
      "grad_norm": 1.6877959966659546,
      "learning_rate": 2.2783238112678483e-06,
      "loss": 0.4934,
      "step": 260640
    },
    {
      "epoch": 106.69,
      "grad_norm": 1.9329537153244019,
      "learning_rate": 2.2781652436873286e-06,
      "loss": 0.46,
      "step": 260650
    },
    {
      "epoch": 106.7,
      "grad_norm": 2.0205931663513184,
      "learning_rate": 2.2780066764072425e-06,
      "loss": 0.4679,
      "step": 260660
    },
    {
      "epoch": 106.7,
      "grad_norm": 1.9013170003890991,
      "learning_rate": 2.277848109428316e-06,
      "loss": 0.4697,
      "step": 260670
    },
    {
      "epoch": 106.7,
      "grad_norm": 2.191250801086426,
      "learning_rate": 2.277689542751277e-06,
      "loss": 0.4802,
      "step": 260680
    },
    {
      "epoch": 106.71,
      "grad_norm": 2.190335512161255,
      "learning_rate": 2.27753097637685e-06,
      "loss": 0.467,
      "step": 260690
    },
    {
      "epoch": 106.71,
      "grad_norm": 1.7487801313400269,
      "learning_rate": 2.277372410305762e-06,
      "loss": 0.4582,
      "step": 260700
    },
    {
      "epoch": 106.72,
      "grad_norm": 1.9110227823257446,
      "learning_rate": 2.27721384453874e-06,
      "loss": 0.4967,
      "step": 260710
    },
    {
      "epoch": 106.72,
      "grad_norm": 1.8452200889587402,
      "learning_rate": 2.2770552790765094e-06,
      "loss": 0.4585,
      "step": 260720
    },
    {
      "epoch": 106.73,
      "grad_norm": 2.8510496616363525,
      "learning_rate": 2.2768967139197973e-06,
      "loss": 0.4696,
      "step": 260730
    },
    {
      "epoch": 106.73,
      "grad_norm": 1.9263783693313599,
      "learning_rate": 2.2767381490693296e-06,
      "loss": 0.4669,
      "step": 260740
    },
    {
      "epoch": 106.73,
      "grad_norm": 2.1835246086120605,
      "learning_rate": 2.276579584525833e-06,
      "loss": 0.4527,
      "step": 260750
    },
    {
      "epoch": 106.74,
      "grad_norm": 1.7638256549835205,
      "learning_rate": 2.2764210202900338e-06,
      "loss": 0.4747,
      "step": 260760
    },
    {
      "epoch": 106.74,
      "grad_norm": 2.3688931465148926,
      "learning_rate": 2.276262456362658e-06,
      "loss": 0.4537,
      "step": 260770
    },
    {
      "epoch": 106.75,
      "grad_norm": 2.3609180450439453,
      "learning_rate": 2.276103892744432e-06,
      "loss": 0.4854,
      "step": 260780
    },
    {
      "epoch": 106.75,
      "grad_norm": 2.2489094734191895,
      "learning_rate": 2.2759453294360827e-06,
      "loss": 0.4748,
      "step": 260790
    },
    {
      "epoch": 106.75,
      "grad_norm": 1.7191952466964722,
      "learning_rate": 2.2757867664383357e-06,
      "loss": 0.4848,
      "step": 260800
    },
    {
      "epoch": 106.76,
      "grad_norm": 2.216595411300659,
      "learning_rate": 2.2756282037519177e-06,
      "loss": 0.4603,
      "step": 260810
    },
    {
      "epoch": 106.76,
      "grad_norm": 1.6643378734588623,
      "learning_rate": 2.2754696413775556e-06,
      "loss": 0.4769,
      "step": 260820
    },
    {
      "epoch": 106.77,
      "grad_norm": 1.5753540992736816,
      "learning_rate": 2.2753110793159747e-06,
      "loss": 0.483,
      "step": 260830
    },
    {
      "epoch": 106.77,
      "grad_norm": 1.896177887916565,
      "learning_rate": 2.2751525175679024e-06,
      "loss": 0.4689,
      "step": 260840
    },
    {
      "epoch": 106.77,
      "grad_norm": 1.6501747369766235,
      "learning_rate": 2.2749939561340645e-06,
      "loss": 0.4762,
      "step": 260850
    },
    {
      "epoch": 106.78,
      "grad_norm": 1.9196860790252686,
      "learning_rate": 2.274835395015186e-06,
      "loss": 0.4677,
      "step": 260860
    },
    {
      "epoch": 106.78,
      "grad_norm": 1.8807218074798584,
      "learning_rate": 2.274676834211996e-06,
      "loss": 0.4604,
      "step": 260870
    },
    {
      "epoch": 106.79,
      "grad_norm": 1.90505051612854,
      "learning_rate": 2.274518273725219e-06,
      "loss": 0.4599,
      "step": 260880
    },
    {
      "epoch": 106.79,
      "grad_norm": 2.013223886489868,
      "learning_rate": 2.274359713555583e-06,
      "loss": 0.4705,
      "step": 260890
    },
    {
      "epoch": 106.79,
      "grad_norm": 1.846604824066162,
      "learning_rate": 2.274201153703812e-06,
      "loss": 0.4661,
      "step": 260900
    },
    {
      "epoch": 106.8,
      "grad_norm": 2.176769971847534,
      "learning_rate": 2.2740425941706334e-06,
      "loss": 0.4748,
      "step": 260910
    },
    {
      "epoch": 106.8,
      "grad_norm": 1.8619452714920044,
      "learning_rate": 2.2738840349567744e-06,
      "loss": 0.4659,
      "step": 260920
    },
    {
      "epoch": 106.81,
      "grad_norm": 2.894902467727661,
      "learning_rate": 2.27372547606296e-06,
      "loss": 0.4787,
      "step": 260930
    },
    {
      "epoch": 106.81,
      "grad_norm": 2.2552742958068848,
      "learning_rate": 2.273566917489917e-06,
      "loss": 0.472,
      "step": 260940
    },
    {
      "epoch": 106.82,
      "grad_norm": 1.697380542755127,
      "learning_rate": 2.2734083592383725e-06,
      "loss": 0.4623,
      "step": 260950
    },
    {
      "epoch": 106.82,
      "grad_norm": 1.6932791471481323,
      "learning_rate": 2.2732498013090516e-06,
      "loss": 0.474,
      "step": 260960
    },
    {
      "epoch": 106.82,
      "grad_norm": 1.9327744245529175,
      "learning_rate": 2.273091243702681e-06,
      "loss": 0.4773,
      "step": 260970
    },
    {
      "epoch": 106.83,
      "grad_norm": 1.76393723487854,
      "learning_rate": 2.2729326864199882e-06,
      "loss": 0.4747,
      "step": 260980
    },
    {
      "epoch": 106.83,
      "grad_norm": 1.775872826576233,
      "learning_rate": 2.272774129461698e-06,
      "loss": 0.475,
      "step": 260990
    },
    {
      "epoch": 106.84,
      "grad_norm": 3.0613040924072266,
      "learning_rate": 2.2726155728285374e-06,
      "loss": 0.4687,
      "step": 261000
    },
    {
      "epoch": 106.84,
      "grad_norm": 2.1823246479034424,
      "learning_rate": 2.272457016521232e-06,
      "loss": 0.4601,
      "step": 261010
    },
    {
      "epoch": 106.84,
      "grad_norm": 1.8436193466186523,
      "learning_rate": 2.2722984605405093e-06,
      "loss": 0.4901,
      "step": 261020
    },
    {
      "epoch": 106.85,
      "grad_norm": 2.234297275543213,
      "learning_rate": 2.2721399048870948e-06,
      "loss": 0.4634,
      "step": 261030
    },
    {
      "epoch": 106.85,
      "grad_norm": 1.8876184225082397,
      "learning_rate": 2.271981349561715e-06,
      "loss": 0.4714,
      "step": 261040
    },
    {
      "epoch": 106.86,
      "grad_norm": 1.8963303565979004,
      "learning_rate": 2.271822794565096e-06,
      "loss": 0.4703,
      "step": 261050
    },
    {
      "epoch": 106.86,
      "grad_norm": 1.9049187898635864,
      "learning_rate": 2.2716642398979644e-06,
      "loss": 0.4594,
      "step": 261060
    },
    {
      "epoch": 106.86,
      "grad_norm": 2.1355838775634766,
      "learning_rate": 2.2715056855610466e-06,
      "loss": 0.4698,
      "step": 261070
    },
    {
      "epoch": 106.87,
      "grad_norm": 2.512758493423462,
      "learning_rate": 2.2713471315550693e-06,
      "loss": 0.4794,
      "step": 261080
    },
    {
      "epoch": 106.87,
      "grad_norm": 1.758878469467163,
      "learning_rate": 2.271188577880758e-06,
      "loss": 0.4846,
      "step": 261090
    },
    {
      "epoch": 106.88,
      "grad_norm": 1.894514799118042,
      "learning_rate": 2.2710300245388396e-06,
      "loss": 0.5041,
      "step": 261100
    },
    {
      "epoch": 106.88,
      "grad_norm": 1.7178248167037964,
      "learning_rate": 2.2708714715300394e-06,
      "loss": 0.4641,
      "step": 261110
    },
    {
      "epoch": 106.88,
      "grad_norm": 1.783079743385315,
      "learning_rate": 2.270712918855086e-06,
      "loss": 0.4873,
      "step": 261120
    },
    {
      "epoch": 106.89,
      "grad_norm": 2.1831061840057373,
      "learning_rate": 2.2705543665147027e-06,
      "loss": 0.4494,
      "step": 261130
    },
    {
      "epoch": 106.89,
      "grad_norm": 2.4095354080200195,
      "learning_rate": 2.270395814509618e-06,
      "loss": 0.4821,
      "step": 261140
    },
    {
      "epoch": 106.9,
      "grad_norm": 1.8406051397323608,
      "learning_rate": 2.270237262840557e-06,
      "loss": 0.4753,
      "step": 261150
    },
    {
      "epoch": 106.9,
      "grad_norm": 2.205226182937622,
      "learning_rate": 2.270078711508247e-06,
      "loss": 0.4935,
      "step": 261160
    },
    {
      "epoch": 106.91,
      "grad_norm": 1.824462652206421,
      "learning_rate": 2.2699201605134133e-06,
      "loss": 0.452,
      "step": 261170
    },
    {
      "epoch": 106.91,
      "grad_norm": 1.8184030055999756,
      "learning_rate": 2.269761609856783e-06,
      "loss": 0.4851,
      "step": 261180
    },
    {
      "epoch": 106.91,
      "grad_norm": 2.127875566482544,
      "learning_rate": 2.2696030595390814e-06,
      "loss": 0.4788,
      "step": 261190
    },
    {
      "epoch": 106.92,
      "grad_norm": 1.6506268978118896,
      "learning_rate": 2.2694445095610367e-06,
      "loss": 0.4722,
      "step": 261200
    },
    {
      "epoch": 106.92,
      "grad_norm": 1.7089624404907227,
      "learning_rate": 2.269285959923373e-06,
      "loss": 0.4843,
      "step": 261210
    },
    {
      "epoch": 106.93,
      "grad_norm": 1.8355861902236938,
      "learning_rate": 2.2691274106268176e-06,
      "loss": 0.469,
      "step": 261220
    },
    {
      "epoch": 106.93,
      "grad_norm": 1.747534155845642,
      "learning_rate": 2.268968861672097e-06,
      "loss": 0.461,
      "step": 261230
    },
    {
      "epoch": 106.93,
      "grad_norm": 1.5612977743148804,
      "learning_rate": 2.268810313059936e-06,
      "loss": 0.4646,
      "step": 261240
    },
    {
      "epoch": 106.94,
      "grad_norm": 2.5653786659240723,
      "learning_rate": 2.2686517647910635e-06,
      "loss": 0.4741,
      "step": 261250
    },
    {
      "epoch": 106.94,
      "grad_norm": 1.977332353591919,
      "learning_rate": 2.2684932168662038e-06,
      "loss": 0.4652,
      "step": 261260
    },
    {
      "epoch": 106.95,
      "grad_norm": 1.7156206369400024,
      "learning_rate": 2.268334669286084e-06,
      "loss": 0.4598,
      "step": 261270
    },
    {
      "epoch": 106.95,
      "grad_norm": 1.7278265953063965,
      "learning_rate": 2.2681761220514303e-06,
      "loss": 0.5002,
      "step": 261280
    },
    {
      "epoch": 106.95,
      "grad_norm": 1.9887648820877075,
      "learning_rate": 2.2680175751629683e-06,
      "loss": 0.4799,
      "step": 261290
    },
    {
      "epoch": 106.96,
      "grad_norm": 1.7734440565109253,
      "learning_rate": 2.2678590286214252e-06,
      "loss": 0.4828,
      "step": 261300
    },
    {
      "epoch": 106.96,
      "grad_norm": 2.253990888595581,
      "learning_rate": 2.267700482427527e-06,
      "loss": 0.4689,
      "step": 261310
    },
    {
      "epoch": 106.97,
      "grad_norm": 1.8544708490371704,
      "learning_rate": 2.2675419365819997e-06,
      "loss": 0.4583,
      "step": 261320
    },
    {
      "epoch": 106.97,
      "grad_norm": 1.9672106504440308,
      "learning_rate": 2.26738339108557e-06,
      "loss": 0.481,
      "step": 261330
    },
    {
      "epoch": 106.98,
      "grad_norm": 2.5613083839416504,
      "learning_rate": 2.267224845938964e-06,
      "loss": 0.4624,
      "step": 261340
    },
    {
      "epoch": 106.98,
      "grad_norm": 1.995674967765808,
      "learning_rate": 2.2670663011429075e-06,
      "loss": 0.481,
      "step": 261350
    },
    {
      "epoch": 106.98,
      "grad_norm": 1.912387490272522,
      "learning_rate": 2.2669077566981267e-06,
      "loss": 0.4735,
      "step": 261360
    },
    {
      "epoch": 106.99,
      "grad_norm": 1.5638450384140015,
      "learning_rate": 2.266749212605349e-06,
      "loss": 0.4657,
      "step": 261370
    },
    {
      "epoch": 106.99,
      "grad_norm": 2.2064549922943115,
      "learning_rate": 2.2665906688652997e-06,
      "loss": 0.4774,
      "step": 261380
    },
    {
      "epoch": 107.0,
      "grad_norm": 1.81434166431427,
      "learning_rate": 2.2664321254787057e-06,
      "loss": 0.455,
      "step": 261390
    },
    {
      "epoch": 107.0,
      "grad_norm": 2.3378384113311768,
      "learning_rate": 2.2662735824462922e-06,
      "loss": 0.4832,
      "step": 261400
    },
    {
      "epoch": 107.0,
      "eval_loss": 0.4729343056678772,
      "eval_runtime": 52.8848,
      "eval_samples_per_second": 65.217,
      "eval_steps_per_second": 8.169,
      "step": 261401
    },
    {
      "epoch": 107.0,
      "grad_norm": 1.7820459604263306,
      "learning_rate": 2.266115039768787e-06,
      "loss": 0.4553,
      "step": 261410
    },
    {
      "epoch": 107.01,
      "grad_norm": 1.7524662017822266,
      "learning_rate": 2.2659564974469145e-06,
      "loss": 0.4909,
      "step": 261420
    },
    {
      "epoch": 107.01,
      "grad_norm": 1.7434861660003662,
      "learning_rate": 2.2657979554814026e-06,
      "loss": 0.4888,
      "step": 261430
    },
    {
      "epoch": 107.02,
      "grad_norm": 2.1383047103881836,
      "learning_rate": 2.265639413872977e-06,
      "loss": 0.5092,
      "step": 261440
    },
    {
      "epoch": 107.02,
      "grad_norm": 1.8253309726715088,
      "learning_rate": 2.2654808726223636e-06,
      "loss": 0.4815,
      "step": 261450
    },
    {
      "epoch": 107.02,
      "grad_norm": 1.9304289817810059,
      "learning_rate": 2.2653223317302893e-06,
      "loss": 0.4901,
      "step": 261460
    },
    {
      "epoch": 107.03,
      "grad_norm": 1.915164828300476,
      "learning_rate": 2.2651637911974798e-06,
      "loss": 0.4644,
      "step": 261470
    },
    {
      "epoch": 107.03,
      "grad_norm": 2.0181422233581543,
      "learning_rate": 2.2650052510246622e-06,
      "loss": 0.4787,
      "step": 261480
    },
    {
      "epoch": 107.04,
      "grad_norm": 2.2721686363220215,
      "learning_rate": 2.2648467112125617e-06,
      "loss": 0.4753,
      "step": 261490
    },
    {
      "epoch": 107.04,
      "grad_norm": 1.7851791381835938,
      "learning_rate": 2.2646881717619045e-06,
      "loss": 0.4811,
      "step": 261500
    },
    {
      "epoch": 107.04,
      "grad_norm": 2.144676923751831,
      "learning_rate": 2.2645296326734175e-06,
      "loss": 0.4665,
      "step": 261510
    },
    {
      "epoch": 107.05,
      "grad_norm": 1.8588355779647827,
      "learning_rate": 2.2643710939478273e-06,
      "loss": 0.5011,
      "step": 261520
    },
    {
      "epoch": 107.05,
      "grad_norm": 1.7979284524917603,
      "learning_rate": 2.2642125555858586e-06,
      "loss": 0.4797,
      "step": 261530
    },
    {
      "epoch": 107.06,
      "grad_norm": 2.035942554473877,
      "learning_rate": 2.2640540175882394e-06,
      "loss": 0.4782,
      "step": 261540
    },
    {
      "epoch": 107.06,
      "grad_norm": 1.9134241342544556,
      "learning_rate": 2.263895479955695e-06,
      "loss": 0.4632,
      "step": 261550
    },
    {
      "epoch": 107.07,
      "grad_norm": 1.9773478507995605,
      "learning_rate": 2.263736942688952e-06,
      "loss": 0.4797,
      "step": 261560
    },
    {
      "epoch": 107.07,
      "grad_norm": 2.0995519161224365,
      "learning_rate": 2.263578405788736e-06,
      "loss": 0.4752,
      "step": 261570
    },
    {
      "epoch": 107.07,
      "grad_norm": 1.7749125957489014,
      "learning_rate": 2.2634198692557736e-06,
      "loss": 0.4694,
      "step": 261580
    },
    {
      "epoch": 107.08,
      "grad_norm": 1.9205806255340576,
      "learning_rate": 2.2632613330907916e-06,
      "loss": 0.4716,
      "step": 261590
    },
    {
      "epoch": 107.08,
      "grad_norm": 1.7782893180847168,
      "learning_rate": 2.2631027972945153e-06,
      "loss": 0.4637,
      "step": 261600
    },
    {
      "epoch": 107.09,
      "grad_norm": 1.6104177236557007,
      "learning_rate": 2.262944261867671e-06,
      "loss": 0.4743,
      "step": 261610
    },
    {
      "epoch": 107.09,
      "grad_norm": 1.7025014162063599,
      "learning_rate": 2.262785726810985e-06,
      "loss": 0.4829,
      "step": 261620
    },
    {
      "epoch": 107.09,
      "grad_norm": 2.013441324234009,
      "learning_rate": 2.262627192125185e-06,
      "loss": 0.4794,
      "step": 261630
    },
    {
      "epoch": 107.1,
      "grad_norm": 2.001681327819824,
      "learning_rate": 2.262468657810995e-06,
      "loss": 0.4964,
      "step": 261640
    },
    {
      "epoch": 107.1,
      "grad_norm": 1.7447450160980225,
      "learning_rate": 2.262310123869143e-06,
      "loss": 0.4673,
      "step": 261650
    },
    {
      "epoch": 107.11,
      "grad_norm": 2.1429152488708496,
      "learning_rate": 2.262151590300354e-06,
      "loss": 0.4945,
      "step": 261660
    },
    {
      "epoch": 107.11,
      "grad_norm": 1.8124786615371704,
      "learning_rate": 2.261993057105355e-06,
      "loss": 0.4796,
      "step": 261670
    },
    {
      "epoch": 107.11,
      "grad_norm": 1.8765499591827393,
      "learning_rate": 2.2618345242848716e-06,
      "loss": 0.4766,
      "step": 261680
    },
    {
      "epoch": 107.12,
      "grad_norm": 1.8102638721466064,
      "learning_rate": 2.2616759918396304e-06,
      "loss": 0.4781,
      "step": 261690
    },
    {
      "epoch": 107.12,
      "grad_norm": 1.7601826190948486,
      "learning_rate": 2.2615174597703585e-06,
      "loss": 0.4685,
      "step": 261700
    },
    {
      "epoch": 107.13,
      "grad_norm": 1.9139761924743652,
      "learning_rate": 2.26135892807778e-06,
      "loss": 0.4748,
      "step": 261710
    },
    {
      "epoch": 107.13,
      "grad_norm": 2.3952672481536865,
      "learning_rate": 2.2612003967626225e-06,
      "loss": 0.4646,
      "step": 261720
    },
    {
      "epoch": 107.13,
      "grad_norm": 2.000643253326416,
      "learning_rate": 2.261041865825612e-06,
      "loss": 0.4644,
      "step": 261730
    },
    {
      "epoch": 107.14,
      "grad_norm": 1.7880288362503052,
      "learning_rate": 2.2608833352674744e-06,
      "loss": 0.4602,
      "step": 261740
    },
    {
      "epoch": 107.14,
      "grad_norm": 1.611664891242981,
      "learning_rate": 2.2607248050889362e-06,
      "loss": 0.473,
      "step": 261750
    },
    {
      "epoch": 107.15,
      "grad_norm": 1.6255512237548828,
      "learning_rate": 2.260566275290724e-06,
      "loss": 0.4744,
      "step": 261760
    },
    {
      "epoch": 107.15,
      "grad_norm": 1.8019180297851562,
      "learning_rate": 2.260407745873563e-06,
      "loss": 0.4774,
      "step": 261770
    },
    {
      "epoch": 107.16,
      "grad_norm": 1.712462306022644,
      "learning_rate": 2.2602492168381804e-06,
      "loss": 0.4794,
      "step": 261780
    },
    {
      "epoch": 107.16,
      "grad_norm": 1.867724061012268,
      "learning_rate": 2.2600906881853023e-06,
      "loss": 0.4594,
      "step": 261790
    },
    {
      "epoch": 107.16,
      "grad_norm": 2.0950753688812256,
      "learning_rate": 2.259932159915653e-06,
      "loss": 0.4657,
      "step": 261800
    },
    {
      "epoch": 107.17,
      "grad_norm": 1.7904013395309448,
      "learning_rate": 2.259773632029961e-06,
      "loss": 0.479,
      "step": 261810
    },
    {
      "epoch": 107.17,
      "grad_norm": 2.045748710632324,
      "learning_rate": 2.259615104528952e-06,
      "loss": 0.4705,
      "step": 261820
    },
    {
      "epoch": 107.18,
      "grad_norm": 1.8000065088272095,
      "learning_rate": 2.2594565774133517e-06,
      "loss": 0.4727,
      "step": 261830
    },
    {
      "epoch": 107.18,
      "grad_norm": 1.9160863161087036,
      "learning_rate": 2.2592980506838872e-06,
      "loss": 0.484,
      "step": 261840
    },
    {
      "epoch": 107.18,
      "grad_norm": 1.782368779182434,
      "learning_rate": 2.2591395243412835e-06,
      "loss": 0.4849,
      "step": 261850
    },
    {
      "epoch": 107.19,
      "grad_norm": 1.8878093957901,
      "learning_rate": 2.2589809983862676e-06,
      "loss": 0.4559,
      "step": 261860
    },
    {
      "epoch": 107.19,
      "grad_norm": 1.8052641153335571,
      "learning_rate": 2.258822472819565e-06,
      "loss": 0.452,
      "step": 261870
    },
    {
      "epoch": 107.2,
      "grad_norm": 1.9062588214874268,
      "learning_rate": 2.258663947641902e-06,
      "loss": 0.4673,
      "step": 261880
    },
    {
      "epoch": 107.2,
      "grad_norm": 2.003901720046997,
      "learning_rate": 2.2585054228540058e-06,
      "loss": 0.4729,
      "step": 261890
    },
    {
      "epoch": 107.2,
      "grad_norm": 3.0778915882110596,
      "learning_rate": 2.2583468984566013e-06,
      "loss": 0.4646,
      "step": 261900
    },
    {
      "epoch": 107.21,
      "grad_norm": 1.9918544292449951,
      "learning_rate": 2.258188374450415e-06,
      "loss": 0.4892,
      "step": 261910
    },
    {
      "epoch": 107.21,
      "grad_norm": 1.9002196788787842,
      "learning_rate": 2.2580298508361743e-06,
      "loss": 0.4767,
      "step": 261920
    },
    {
      "epoch": 107.22,
      "grad_norm": 1.9104253053665161,
      "learning_rate": 2.257871327614603e-06,
      "loss": 0.4516,
      "step": 261930
    },
    {
      "epoch": 107.22,
      "grad_norm": 1.589375615119934,
      "learning_rate": 2.2577128047864293e-06,
      "loss": 0.4703,
      "step": 261940
    },
    {
      "epoch": 107.22,
      "grad_norm": 2.067138671875,
      "learning_rate": 2.2575542823523786e-06,
      "loss": 0.4765,
      "step": 261950
    },
    {
      "epoch": 107.23,
      "grad_norm": 2.2992610931396484,
      "learning_rate": 2.257395760313177e-06,
      "loss": 0.4852,
      "step": 261960
    },
    {
      "epoch": 107.23,
      "grad_norm": 1.6431286334991455,
      "learning_rate": 2.2572372386695507e-06,
      "loss": 0.4845,
      "step": 261970
    },
    {
      "epoch": 107.24,
      "grad_norm": 1.9689891338348389,
      "learning_rate": 2.257078717422226e-06,
      "loss": 0.4732,
      "step": 261980
    },
    {
      "epoch": 107.24,
      "grad_norm": 2.0798959732055664,
      "learning_rate": 2.256920196571929e-06,
      "loss": 0.4636,
      "step": 261990
    },
    {
      "epoch": 107.25,
      "grad_norm": 2.375669002532959,
      "learning_rate": 2.2567616761193855e-06,
      "loss": 0.4563,
      "step": 262000
    },
    {
      "epoch": 107.25,
      "grad_norm": 2.2047531604766846,
      "learning_rate": 2.256603156065323e-06,
      "loss": 0.4694,
      "step": 262010
    },
    {
      "epoch": 107.25,
      "grad_norm": 2.11236572265625,
      "learning_rate": 2.256444636410466e-06,
      "loss": 0.4697,
      "step": 262020
    },
    {
      "epoch": 107.26,
      "grad_norm": 2.1080644130706787,
      "learning_rate": 2.2562861171555413e-06,
      "loss": 0.4782,
      "step": 262030
    },
    {
      "epoch": 107.26,
      "grad_norm": 1.8648102283477783,
      "learning_rate": 2.2561275983012755e-06,
      "loss": 0.4674,
      "step": 262040
    },
    {
      "epoch": 107.27,
      "grad_norm": 2.228544235229492,
      "learning_rate": 2.2559690798483944e-06,
      "loss": 0.451,
      "step": 262050
    },
    {
      "epoch": 107.27,
      "grad_norm": 1.895249843597412,
      "learning_rate": 2.2558105617976244e-06,
      "loss": 0.487,
      "step": 262060
    },
    {
      "epoch": 107.27,
      "grad_norm": 1.7074536085128784,
      "learning_rate": 2.2556520441496906e-06,
      "loss": 0.4624,
      "step": 262070
    },
    {
      "epoch": 107.28,
      "grad_norm": 1.8038148880004883,
      "learning_rate": 2.25549352690532e-06,
      "loss": 0.4793,
      "step": 262080
    },
    {
      "epoch": 107.28,
      "grad_norm": 2.463473320007324,
      "learning_rate": 2.255335010065239e-06,
      "loss": 0.4645,
      "step": 262090
    },
    {
      "epoch": 107.29,
      "grad_norm": 1.9676059484481812,
      "learning_rate": 2.2551764936301732e-06,
      "loss": 0.4653,
      "step": 262100
    },
    {
      "epoch": 107.29,
      "grad_norm": 1.8565735816955566,
      "learning_rate": 2.255017977600849e-06,
      "loss": 0.4774,
      "step": 262110
    },
    {
      "epoch": 107.29,
      "grad_norm": 2.018510580062866,
      "learning_rate": 2.2548594619779923e-06,
      "loss": 0.4805,
      "step": 262120
    },
    {
      "epoch": 107.3,
      "grad_norm": 1.6492953300476074,
      "learning_rate": 2.2547009467623293e-06,
      "loss": 0.4484,
      "step": 262130
    },
    {
      "epoch": 107.3,
      "grad_norm": 1.933121681213379,
      "learning_rate": 2.2545424319545867e-06,
      "loss": 0.4783,
      "step": 262140
    },
    {
      "epoch": 107.31,
      "grad_norm": 2.2779202461242676,
      "learning_rate": 2.25438391755549e-06,
      "loss": 0.467,
      "step": 262150
    },
    {
      "epoch": 107.31,
      "grad_norm": 3.547437906265259,
      "learning_rate": 2.2542254035657654e-06,
      "loss": 0.4647,
      "step": 262160
    },
    {
      "epoch": 107.31,
      "grad_norm": 1.6682511568069458,
      "learning_rate": 2.254066889986139e-06,
      "loss": 0.463,
      "step": 262170
    },
    {
      "epoch": 107.32,
      "grad_norm": 2.03505539894104,
      "learning_rate": 2.253908376817336e-06,
      "loss": 0.4503,
      "step": 262180
    },
    {
      "epoch": 107.32,
      "grad_norm": 2.1859161853790283,
      "learning_rate": 2.2537498640600844e-06,
      "loss": 0.4638,
      "step": 262190
    },
    {
      "epoch": 107.33,
      "grad_norm": 2.195391893386841,
      "learning_rate": 2.2535913517151105e-06,
      "loss": 0.4752,
      "step": 262200
    },
    {
      "epoch": 107.33,
      "grad_norm": 1.8479372262954712,
      "learning_rate": 2.2534328397831384e-06,
      "loss": 0.4728,
      "step": 262210
    },
    {
      "epoch": 107.34,
      "grad_norm": 1.8379703760147095,
      "learning_rate": 2.2532743282648957e-06,
      "loss": 0.4831,
      "step": 262220
    },
    {
      "epoch": 107.34,
      "grad_norm": 2.8337645530700684,
      "learning_rate": 2.2531158171611075e-06,
      "loss": 0.4746,
      "step": 262230
    },
    {
      "epoch": 107.34,
      "grad_norm": 1.9646228551864624,
      "learning_rate": 2.2529573064725006e-06,
      "loss": 0.4875,
      "step": 262240
    },
    {
      "epoch": 107.35,
      "grad_norm": 1.6940443515777588,
      "learning_rate": 2.2527987961998012e-06,
      "loss": 0.4726,
      "step": 262250
    },
    {
      "epoch": 107.35,
      "grad_norm": 1.8252816200256348,
      "learning_rate": 2.252640286343735e-06,
      "loss": 0.4655,
      "step": 262260
    },
    {
      "epoch": 107.36,
      "grad_norm": 1.6878893375396729,
      "learning_rate": 2.2524817769050284e-06,
      "loss": 0.4588,
      "step": 262270
    },
    {
      "epoch": 107.36,
      "grad_norm": 2.4452102184295654,
      "learning_rate": 2.2523232678844077e-06,
      "loss": 0.4703,
      "step": 262280
    },
    {
      "epoch": 107.36,
      "grad_norm": 1.830319881439209,
      "learning_rate": 2.2521647592825985e-06,
      "loss": 0.487,
      "step": 262290
    },
    {
      "epoch": 107.37,
      "grad_norm": 1.6804033517837524,
      "learning_rate": 2.252006251100327e-06,
      "loss": 0.4718,
      "step": 262300
    },
    {
      "epoch": 107.37,
      "grad_norm": 2.0868802070617676,
      "learning_rate": 2.251847743338319e-06,
      "loss": 0.4702,
      "step": 262310
    },
    {
      "epoch": 107.38,
      "grad_norm": 2.135648488998413,
      "learning_rate": 2.2516892359973017e-06,
      "loss": 0.4721,
      "step": 262320
    },
    {
      "epoch": 107.38,
      "grad_norm": 2.123051166534424,
      "learning_rate": 2.2515307290780002e-06,
      "loss": 0.4732,
      "step": 262330
    },
    {
      "epoch": 107.38,
      "grad_norm": 2.2195498943328857,
      "learning_rate": 2.251372222581141e-06,
      "loss": 0.4635,
      "step": 262340
    },
    {
      "epoch": 107.39,
      "grad_norm": 2.6246116161346436,
      "learning_rate": 2.2512137165074493e-06,
      "loss": 0.4683,
      "step": 262350
    },
    {
      "epoch": 107.39,
      "grad_norm": 1.8718597888946533,
      "learning_rate": 2.251055210857653e-06,
      "loss": 0.4788,
      "step": 262360
    },
    {
      "epoch": 107.4,
      "grad_norm": 1.8349167108535767,
      "learning_rate": 2.2508967056324762e-06,
      "loss": 0.4634,
      "step": 262370
    },
    {
      "epoch": 107.4,
      "grad_norm": 2.3518226146698,
      "learning_rate": 2.2507382008326465e-06,
      "loss": 0.4762,
      "step": 262380
    },
    {
      "epoch": 107.4,
      "grad_norm": 2.239190101623535,
      "learning_rate": 2.2505796964588896e-06,
      "loss": 0.4702,
      "step": 262390
    },
    {
      "epoch": 107.41,
      "grad_norm": 1.8550926446914673,
      "learning_rate": 2.2504211925119313e-06,
      "loss": 0.4692,
      "step": 262400
    },
    {
      "epoch": 107.41,
      "grad_norm": 2.136139392852783,
      "learning_rate": 2.2502626889924978e-06,
      "loss": 0.4593,
      "step": 262410
    },
    {
      "epoch": 107.42,
      "grad_norm": 1.7707022428512573,
      "learning_rate": 2.250104185901316e-06,
      "loss": 0.4558,
      "step": 262420
    },
    {
      "epoch": 107.42,
      "grad_norm": 1.4473800659179688,
      "learning_rate": 2.2499456832391103e-06,
      "loss": 0.4751,
      "step": 262430
    },
    {
      "epoch": 107.43,
      "grad_norm": 2.2672038078308105,
      "learning_rate": 2.2497871810066077e-06,
      "loss": 0.4669,
      "step": 262440
    },
    {
      "epoch": 107.43,
      "grad_norm": 1.8451546430587769,
      "learning_rate": 2.2496286792045343e-06,
      "loss": 0.4875,
      "step": 262450
    },
    {
      "epoch": 107.43,
      "grad_norm": 1.919621229171753,
      "learning_rate": 2.249470177833616e-06,
      "loss": 0.4539,
      "step": 262460
    },
    {
      "epoch": 107.44,
      "grad_norm": 2.1771340370178223,
      "learning_rate": 2.2493116768945792e-06,
      "loss": 0.4811,
      "step": 262470
    },
    {
      "epoch": 107.44,
      "grad_norm": 1.7620368003845215,
      "learning_rate": 2.2491531763881494e-06,
      "loss": 0.4773,
      "step": 262480
    },
    {
      "epoch": 107.45,
      "grad_norm": 1.8707083463668823,
      "learning_rate": 2.248994676315053e-06,
      "loss": 0.4567,
      "step": 262490
    },
    {
      "epoch": 107.45,
      "grad_norm": 1.969003677368164,
      "learning_rate": 2.2488361766760165e-06,
      "loss": 0.472,
      "step": 262500
    },
    {
      "epoch": 107.45,
      "grad_norm": 2.02445912361145,
      "learning_rate": 2.248677677471766e-06,
      "loss": 0.4463,
      "step": 262510
    },
    {
      "epoch": 107.46,
      "grad_norm": 2.1185359954833984,
      "learning_rate": 2.2485191787030264e-06,
      "loss": 0.464,
      "step": 262520
    },
    {
      "epoch": 107.46,
      "grad_norm": 1.7502260208129883,
      "learning_rate": 2.248360680370524e-06,
      "loss": 0.4607,
      "step": 262530
    },
    {
      "epoch": 107.47,
      "grad_norm": 2.112692356109619,
      "learning_rate": 2.248202182474986e-06,
      "loss": 0.5043,
      "step": 262540
    },
    {
      "epoch": 107.47,
      "grad_norm": 1.9548635482788086,
      "learning_rate": 2.2480436850171375e-06,
      "loss": 0.4673,
      "step": 262550
    },
    {
      "epoch": 107.47,
      "grad_norm": 2.417337656021118,
      "learning_rate": 2.2478851879977043e-06,
      "loss": 0.4946,
      "step": 262560
    },
    {
      "epoch": 107.48,
      "grad_norm": 2.2360920906066895,
      "learning_rate": 2.2477266914174132e-06,
      "loss": 0.4655,
      "step": 262570
    },
    {
      "epoch": 107.48,
      "grad_norm": 1.6885253190994263,
      "learning_rate": 2.24756819527699e-06,
      "loss": 0.4847,
      "step": 262580
    },
    {
      "epoch": 107.49,
      "grad_norm": 1.865900993347168,
      "learning_rate": 2.247409699577161e-06,
      "loss": 0.4781,
      "step": 262590
    },
    {
      "epoch": 107.49,
      "grad_norm": 2.243419647216797,
      "learning_rate": 2.247251204318652e-06,
      "loss": 0.489,
      "step": 262600
    },
    {
      "epoch": 107.49,
      "grad_norm": 1.8122891187667847,
      "learning_rate": 2.247092709502189e-06,
      "loss": 0.4721,
      "step": 262610
    },
    {
      "epoch": 107.5,
      "grad_norm": 1.7575269937515259,
      "learning_rate": 2.2469342151284985e-06,
      "loss": 0.4634,
      "step": 262620
    },
    {
      "epoch": 107.5,
      "grad_norm": 3.4797589778900146,
      "learning_rate": 2.2467757211983054e-06,
      "loss": 0.4634,
      "step": 262630
    },
    {
      "epoch": 107.51,
      "grad_norm": 1.8272013664245605,
      "learning_rate": 2.2466172277123375e-06,
      "loss": 0.5064,
      "step": 262640
    },
    {
      "epoch": 107.51,
      "grad_norm": 1.870749831199646,
      "learning_rate": 2.246458734671319e-06,
      "loss": 0.4686,
      "step": 262650
    },
    {
      "epoch": 107.52,
      "grad_norm": 1.7524070739746094,
      "learning_rate": 2.2463002420759765e-06,
      "loss": 0.4686,
      "step": 262660
    },
    {
      "epoch": 107.52,
      "grad_norm": 2.019407272338867,
      "learning_rate": 2.2461417499270366e-06,
      "loss": 0.4866,
      "step": 262670
    },
    {
      "epoch": 107.52,
      "grad_norm": 1.7905309200286865,
      "learning_rate": 2.245983258225225e-06,
      "loss": 0.4534,
      "step": 262680
    },
    {
      "epoch": 107.53,
      "grad_norm": 1.8840340375900269,
      "learning_rate": 2.2458247669712677e-06,
      "loss": 0.494,
      "step": 262690
    },
    {
      "epoch": 107.53,
      "grad_norm": 2.1109507083892822,
      "learning_rate": 2.2456662761658906e-06,
      "loss": 0.4644,
      "step": 262700
    },
    {
      "epoch": 107.54,
      "grad_norm": 1.6103465557098389,
      "learning_rate": 2.2455077858098195e-06,
      "loss": 0.4713,
      "step": 262710
    },
    {
      "epoch": 107.54,
      "grad_norm": 1.8134443759918213,
      "learning_rate": 2.2453492959037813e-06,
      "loss": 0.4844,
      "step": 262720
    },
    {
      "epoch": 107.54,
      "grad_norm": 1.7188256978988647,
      "learning_rate": 2.245190806448502e-06,
      "loss": 0.4671,
      "step": 262730
    },
    {
      "epoch": 107.55,
      "grad_norm": 1.6495500802993774,
      "learning_rate": 2.2450323174447053e-06,
      "loss": 0.4674,
      "step": 262740
    },
    {
      "epoch": 107.55,
      "grad_norm": 1.7765824794769287,
      "learning_rate": 2.24487382889312e-06,
      "loss": 0.4688,
      "step": 262750
    },
    {
      "epoch": 107.56,
      "grad_norm": 4.51403284072876,
      "learning_rate": 2.244715340794471e-06,
      "loss": 0.4799,
      "step": 262760
    },
    {
      "epoch": 107.56,
      "grad_norm": 2.1033005714416504,
      "learning_rate": 2.244556853149485e-06,
      "loss": 0.4687,
      "step": 262770
    },
    {
      "epoch": 107.56,
      "grad_norm": 1.8832957744598389,
      "learning_rate": 2.2443983659588874e-06,
      "loss": 0.4542,
      "step": 262780
    },
    {
      "epoch": 107.57,
      "grad_norm": 1.4841943979263306,
      "learning_rate": 2.2442398792234045e-06,
      "loss": 0.4693,
      "step": 262790
    },
    {
      "epoch": 107.57,
      "grad_norm": 1.8882840871810913,
      "learning_rate": 2.2440813929437612e-06,
      "loss": 0.5054,
      "step": 262800
    },
    {
      "epoch": 107.58,
      "grad_norm": 2.1579697132110596,
      "learning_rate": 2.243922907120685e-06,
      "loss": 0.4737,
      "step": 262810
    },
    {
      "epoch": 107.58,
      "grad_norm": 1.74062180519104,
      "learning_rate": 2.2437644217549008e-06,
      "loss": 0.4595,
      "step": 262820
    },
    {
      "epoch": 107.58,
      "grad_norm": 1.9086960554122925,
      "learning_rate": 2.2436059368471354e-06,
      "loss": 0.4436,
      "step": 262830
    },
    {
      "epoch": 107.59,
      "grad_norm": 1.9824023246765137,
      "learning_rate": 2.2434474523981137e-06,
      "loss": 0.4845,
      "step": 262840
    },
    {
      "epoch": 107.59,
      "grad_norm": 1.648425817489624,
      "learning_rate": 2.2432889684085634e-06,
      "loss": 0.4779,
      "step": 262850
    },
    {
      "epoch": 107.6,
      "grad_norm": 2.25567626953125,
      "learning_rate": 2.24313048487921e-06,
      "loss": 0.4774,
      "step": 262860
    },
    {
      "epoch": 107.6,
      "grad_norm": 2.1042838096618652,
      "learning_rate": 2.2429720018107777e-06,
      "loss": 0.4692,
      "step": 262870
    },
    {
      "epoch": 107.61,
      "grad_norm": 2.0226988792419434,
      "learning_rate": 2.2428135192039943e-06,
      "loss": 0.4758,
      "step": 262880
    },
    {
      "epoch": 107.61,
      "grad_norm": 2.379967451095581,
      "learning_rate": 2.242655037059585e-06,
      "loss": 0.4465,
      "step": 262890
    },
    {
      "epoch": 107.61,
      "grad_norm": 1.9112492799758911,
      "learning_rate": 2.242496555378276e-06,
      "loss": 0.4615,
      "step": 262900
    },
    {
      "epoch": 107.62,
      "grad_norm": 2.301910400390625,
      "learning_rate": 2.242338074160794e-06,
      "loss": 0.4801,
      "step": 262910
    },
    {
      "epoch": 107.62,
      "grad_norm": 1.5815430879592896,
      "learning_rate": 2.242179593407864e-06,
      "loss": 0.4844,
      "step": 262920
    },
    {
      "epoch": 107.63,
      "grad_norm": 1.8375904560089111,
      "learning_rate": 2.2420211131202115e-06,
      "loss": 0.474,
      "step": 262930
    },
    {
      "epoch": 107.63,
      "grad_norm": 1.9714221954345703,
      "learning_rate": 2.2418626332985642e-06,
      "loss": 0.4604,
      "step": 262940
    },
    {
      "epoch": 107.63,
      "grad_norm": 1.850353479385376,
      "learning_rate": 2.241704153943647e-06,
      "loss": 0.4704,
      "step": 262950
    },
    {
      "epoch": 107.64,
      "grad_norm": 1.7815134525299072,
      "learning_rate": 2.241545675056186e-06,
      "loss": 0.461,
      "step": 262960
    },
    {
      "epoch": 107.64,
      "grad_norm": 1.7239786386489868,
      "learning_rate": 2.2413871966369073e-06,
      "loss": 0.4607,
      "step": 262970
    },
    {
      "epoch": 107.65,
      "grad_norm": 1.968349814414978,
      "learning_rate": 2.2412287186865367e-06,
      "loss": 0.4672,
      "step": 262980
    },
    {
      "epoch": 107.65,
      "grad_norm": 2.2462353706359863,
      "learning_rate": 2.2410702412058e-06,
      "loss": 0.4753,
      "step": 262990
    },
    {
      "epoch": 107.65,
      "grad_norm": 1.9344197511672974,
      "learning_rate": 2.240911764195424e-06,
      "loss": 0.4529,
      "step": 263000
    },
    {
      "epoch": 107.66,
      "grad_norm": 2.2039690017700195,
      "learning_rate": 2.2407532876561336e-06,
      "loss": 0.4855,
      "step": 263010
    },
    {
      "epoch": 107.66,
      "grad_norm": 1.9343153238296509,
      "learning_rate": 2.240594811588655e-06,
      "loss": 0.4893,
      "step": 263020
    },
    {
      "epoch": 107.67,
      "grad_norm": 1.6762160062789917,
      "learning_rate": 2.2404363359937144e-06,
      "loss": 0.4764,
      "step": 263030
    },
    {
      "epoch": 107.67,
      "grad_norm": 2.000764846801758,
      "learning_rate": 2.240277860872038e-06,
      "loss": 0.4878,
      "step": 263040
    },
    {
      "epoch": 107.67,
      "grad_norm": 1.8684287071228027,
      "learning_rate": 2.2401193862243513e-06,
      "loss": 0.4978,
      "step": 263050
    },
    {
      "epoch": 107.68,
      "grad_norm": 1.7585511207580566,
      "learning_rate": 2.2399609120513804e-06,
      "loss": 0.4639,
      "step": 263060
    },
    {
      "epoch": 107.68,
      "grad_norm": 1.9885836839675903,
      "learning_rate": 2.239802438353851e-06,
      "loss": 0.4606,
      "step": 263070
    },
    {
      "epoch": 107.69,
      "grad_norm": 1.8377578258514404,
      "learning_rate": 2.2396439651324893e-06,
      "loss": 0.4607,
      "step": 263080
    },
    {
      "epoch": 107.69,
      "grad_norm": 1.9490939378738403,
      "learning_rate": 2.239485492388022e-06,
      "loss": 0.4769,
      "step": 263090
    },
    {
      "epoch": 107.7,
      "grad_norm": 2.4629385471343994,
      "learning_rate": 2.2393270201211736e-06,
      "loss": 0.4697,
      "step": 263100
    },
    {
      "epoch": 107.7,
      "grad_norm": 1.8546642065048218,
      "learning_rate": 2.239168548332671e-06,
      "loss": 0.4734,
      "step": 263110
    },
    {
      "epoch": 107.7,
      "grad_norm": 2.0458240509033203,
      "learning_rate": 2.2390100770232385e-06,
      "loss": 0.4642,
      "step": 263120
    },
    {
      "epoch": 107.71,
      "grad_norm": 1.96456778049469,
      "learning_rate": 2.238851606193604e-06,
      "loss": 0.4834,
      "step": 263130
    },
    {
      "epoch": 107.71,
      "grad_norm": 2.1200480461120605,
      "learning_rate": 2.2386931358444936e-06,
      "loss": 0.4568,
      "step": 263140
    },
    {
      "epoch": 107.72,
      "grad_norm": 2.3198647499084473,
      "learning_rate": 2.2385346659766325e-06,
      "loss": 0.4604,
      "step": 263150
    },
    {
      "epoch": 107.72,
      "grad_norm": 2.1882708072662354,
      "learning_rate": 2.238376196590746e-06,
      "loss": 0.4619,
      "step": 263160
    },
    {
      "epoch": 107.72,
      "grad_norm": 1.6839946508407593,
      "learning_rate": 2.2382177276875604e-06,
      "loss": 0.482,
      "step": 263170
    },
    {
      "epoch": 107.73,
      "grad_norm": 1.6077888011932373,
      "learning_rate": 2.238059259267802e-06,
      "loss": 0.4595,
      "step": 263180
    },
    {
      "epoch": 107.73,
      "grad_norm": 1.8757469654083252,
      "learning_rate": 2.237900791332197e-06,
      "loss": 0.4797,
      "step": 263190
    },
    {
      "epoch": 107.74,
      "grad_norm": 1.9315171241760254,
      "learning_rate": 2.23774232388147e-06,
      "loss": 0.4665,
      "step": 263200
    },
    {
      "epoch": 107.74,
      "grad_norm": 1.9522103071212769,
      "learning_rate": 2.2375838569163485e-06,
      "loss": 0.4806,
      "step": 263210
    },
    {
      "epoch": 107.74,
      "grad_norm": 2.0398340225219727,
      "learning_rate": 2.2374253904375577e-06,
      "loss": 0.475,
      "step": 263220
    },
    {
      "epoch": 107.75,
      "grad_norm": 1.7914189100265503,
      "learning_rate": 2.2372669244458227e-06,
      "loss": 0.465,
      "step": 263230
    },
    {
      "epoch": 107.75,
      "grad_norm": 2.435189723968506,
      "learning_rate": 2.2371084589418707e-06,
      "loss": 0.4776,
      "step": 263240
    },
    {
      "epoch": 107.76,
      "grad_norm": 4.026710033416748,
      "learning_rate": 2.236949993926427e-06,
      "loss": 0.4435,
      "step": 263250
    },
    {
      "epoch": 107.76,
      "grad_norm": 2.0245790481567383,
      "learning_rate": 2.2367915294002176e-06,
      "loss": 0.4839,
      "step": 263260
    },
    {
      "epoch": 107.77,
      "grad_norm": 2.0485453605651855,
      "learning_rate": 2.2366330653639682e-06,
      "loss": 0.4583,
      "step": 263270
    },
    {
      "epoch": 107.77,
      "grad_norm": 1.821170687675476,
      "learning_rate": 2.236474601818405e-06,
      "loss": 0.4541,
      "step": 263280
    },
    {
      "epoch": 107.77,
      "grad_norm": 4.897507667541504,
      "learning_rate": 2.236316138764254e-06,
      "loss": 0.5013,
      "step": 263290
    },
    {
      "epoch": 107.78,
      "grad_norm": 1.7378571033477783,
      "learning_rate": 2.2361576762022406e-06,
      "loss": 0.4902,
      "step": 263300
    },
    {
      "epoch": 107.78,
      "grad_norm": 1.915221929550171,
      "learning_rate": 2.2359992141330907e-06,
      "loss": 0.4801,
      "step": 263310
    },
    {
      "epoch": 107.79,
      "grad_norm": 1.6609712839126587,
      "learning_rate": 2.2358407525575305e-06,
      "loss": 0.4726,
      "step": 263320
    },
    {
      "epoch": 107.79,
      "grad_norm": 2.196162700653076,
      "learning_rate": 2.235682291476286e-06,
      "loss": 0.4508,
      "step": 263330
    },
    {
      "epoch": 107.79,
      "grad_norm": 2.155768632888794,
      "learning_rate": 2.2355238308900833e-06,
      "loss": 0.4777,
      "step": 263340
    },
    {
      "epoch": 107.8,
      "grad_norm": 1.610711932182312,
      "learning_rate": 2.2353653707996476e-06,
      "loss": 0.467,
      "step": 263350
    },
    {
      "epoch": 107.8,
      "grad_norm": 1.7147353887557983,
      "learning_rate": 2.2352069112057057e-06,
      "loss": 0.4846,
      "step": 263360
    },
    {
      "epoch": 107.81,
      "grad_norm": 2.028947591781616,
      "learning_rate": 2.2350484521089824e-06,
      "loss": 0.4582,
      "step": 263370
    },
    {
      "epoch": 107.81,
      "grad_norm": 2.7271602153778076,
      "learning_rate": 2.2348899935102044e-06,
      "loss": 0.4785,
      "step": 263380
    },
    {
      "epoch": 107.81,
      "grad_norm": 2.0183236598968506,
      "learning_rate": 2.234731535410097e-06,
      "loss": 0.4792,
      "step": 263390
    },
    {
      "epoch": 107.82,
      "grad_norm": 2.273926019668579,
      "learning_rate": 2.234573077809386e-06,
      "loss": 0.4562,
      "step": 263400
    },
    {
      "epoch": 107.82,
      "grad_norm": 2.3228776454925537,
      "learning_rate": 2.234414620708798e-06,
      "loss": 0.456,
      "step": 263410
    },
    {
      "epoch": 107.83,
      "grad_norm": 1.8417924642562866,
      "learning_rate": 2.2342561641090578e-06,
      "loss": 0.4654,
      "step": 263420
    },
    {
      "epoch": 107.83,
      "grad_norm": 1.8715174198150635,
      "learning_rate": 2.2340977080108925e-06,
      "loss": 0.4603,
      "step": 263430
    },
    {
      "epoch": 107.83,
      "grad_norm": 1.829892873764038,
      "learning_rate": 2.233939252415027e-06,
      "loss": 0.5021,
      "step": 263440
    },
    {
      "epoch": 107.84,
      "grad_norm": 1.8116455078125,
      "learning_rate": 2.2337807973221883e-06,
      "loss": 0.479,
      "step": 263450
    },
    {
      "epoch": 107.84,
      "grad_norm": 1.6122605800628662,
      "learning_rate": 2.233622342733101e-06,
      "loss": 0.4536,
      "step": 263460
    },
    {
      "epoch": 107.85,
      "grad_norm": 2.162323236465454,
      "learning_rate": 2.233463888648491e-06,
      "loss": 0.4675,
      "step": 263470
    },
    {
      "epoch": 107.85,
      "grad_norm": 1.983690619468689,
      "learning_rate": 2.233305435069085e-06,
      "loss": 0.4825,
      "step": 263480
    },
    {
      "epoch": 107.86,
      "grad_norm": 1.7558913230895996,
      "learning_rate": 2.2331469819956084e-06,
      "loss": 0.4663,
      "step": 263490
    },
    {
      "epoch": 107.86,
      "grad_norm": 1.993714451789856,
      "learning_rate": 2.2329885294287866e-06,
      "loss": 0.4736,
      "step": 263500
    },
    {
      "epoch": 107.86,
      "grad_norm": 2.7849974632263184,
      "learning_rate": 2.2328300773693464e-06,
      "loss": 0.4797,
      "step": 263510
    },
    {
      "epoch": 107.87,
      "grad_norm": 1.9479608535766602,
      "learning_rate": 2.2326716258180133e-06,
      "loss": 0.4694,
      "step": 263520
    },
    {
      "epoch": 107.87,
      "grad_norm": 1.79987633228302,
      "learning_rate": 2.232513174775513e-06,
      "loss": 0.4686,
      "step": 263530
    },
    {
      "epoch": 107.88,
      "grad_norm": 1.7350561618804932,
      "learning_rate": 2.2323547242425713e-06,
      "loss": 0.4662,
      "step": 263540
    },
    {
      "epoch": 107.88,
      "grad_norm": 2.286961793899536,
      "learning_rate": 2.232196274219914e-06,
      "loss": 0.4771,
      "step": 263550
    },
    {
      "epoch": 107.88,
      "grad_norm": 1.987732172012329,
      "learning_rate": 2.2320378247082673e-06,
      "loss": 0.4808,
      "step": 263560
    },
    {
      "epoch": 107.89,
      "grad_norm": 2.191190719604492,
      "learning_rate": 2.2318793757083563e-06,
      "loss": 0.4528,
      "step": 263570
    },
    {
      "epoch": 107.89,
      "grad_norm": 1.7862321138381958,
      "learning_rate": 2.2317209272209083e-06,
      "loss": 0.4592,
      "step": 263580
    },
    {
      "epoch": 107.9,
      "grad_norm": 1.6914434432983398,
      "learning_rate": 2.2315624792466474e-06,
      "loss": 0.4705,
      "step": 263590
    },
    {
      "epoch": 107.9,
      "grad_norm": 2.18209171295166,
      "learning_rate": 2.2314040317863e-06,
      "loss": 0.4762,
      "step": 263600
    },
    {
      "epoch": 107.9,
      "grad_norm": 1.8430349826812744,
      "learning_rate": 2.2312455848405926e-06,
      "loss": 0.4715,
      "step": 263610
    },
    {
      "epoch": 107.91,
      "grad_norm": 2.055818796157837,
      "learning_rate": 2.23108713841025e-06,
      "loss": 0.4822,
      "step": 263620
    },
    {
      "epoch": 107.91,
      "grad_norm": 1.7722818851470947,
      "learning_rate": 2.2309286924959988e-06,
      "loss": 0.4619,
      "step": 263630
    },
    {
      "epoch": 107.92,
      "grad_norm": 2.0301990509033203,
      "learning_rate": 2.2307702470985646e-06,
      "loss": 0.4641,
      "step": 263640
    },
    {
      "epoch": 107.92,
      "grad_norm": 1.8113504648208618,
      "learning_rate": 2.2306118022186727e-06,
      "loss": 0.4614,
      "step": 263650
    },
    {
      "epoch": 107.92,
      "grad_norm": 1.6969869136810303,
      "learning_rate": 2.23045335785705e-06,
      "loss": 0.4668,
      "step": 263660
    },
    {
      "epoch": 107.93,
      "grad_norm": 1.930025339126587,
      "learning_rate": 2.230294914014422e-06,
      "loss": 0.4619,
      "step": 263670
    },
    {
      "epoch": 107.93,
      "grad_norm": 1.9034631252288818,
      "learning_rate": 2.2301364706915135e-06,
      "loss": 0.4643,
      "step": 263680
    },
    {
      "epoch": 107.94,
      "grad_norm": 2.205261468887329,
      "learning_rate": 2.22997802788905e-06,
      "loss": 0.4663,
      "step": 263690
    },
    {
      "epoch": 107.94,
      "grad_norm": 1.8074458837509155,
      "learning_rate": 2.229819585607759e-06,
      "loss": 0.4639,
      "step": 263700
    },
    {
      "epoch": 107.95,
      "grad_norm": 1.8882708549499512,
      "learning_rate": 2.229661143848366e-06,
      "loss": 0.4681,
      "step": 263710
    },
    {
      "epoch": 107.95,
      "grad_norm": 1.5983389616012573,
      "learning_rate": 2.2295027026115966e-06,
      "loss": 0.4702,
      "step": 263720
    },
    {
      "epoch": 107.95,
      "grad_norm": 1.967511773109436,
      "learning_rate": 2.2293442618981763e-06,
      "loss": 0.4527,
      "step": 263730
    },
    {
      "epoch": 107.96,
      "grad_norm": 1.9893866777420044,
      "learning_rate": 2.2291858217088305e-06,
      "loss": 0.4832,
      "step": 263740
    },
    {
      "epoch": 107.96,
      "grad_norm": 2.639512300491333,
      "learning_rate": 2.2290273820442856e-06,
      "loss": 0.4579,
      "step": 263750
    },
    {
      "epoch": 107.97,
      "grad_norm": 1.8980954885482788,
      "learning_rate": 2.2288689429052675e-06,
      "loss": 0.4485,
      "step": 263760
    },
    {
      "epoch": 107.97,
      "grad_norm": 2.1282529830932617,
      "learning_rate": 2.2287105042925016e-06,
      "loss": 0.4748,
      "step": 263770
    },
    {
      "epoch": 107.97,
      "grad_norm": 2.3628435134887695,
      "learning_rate": 2.228552066206714e-06,
      "loss": 0.4889,
      "step": 263780
    },
    {
      "epoch": 107.98,
      "grad_norm": 1.6025668382644653,
      "learning_rate": 2.2283936286486293e-06,
      "loss": 0.4813,
      "step": 263790
    },
    {
      "epoch": 107.98,
      "grad_norm": 1.7197071313858032,
      "learning_rate": 2.2282351916189752e-06,
      "loss": 0.48,
      "step": 263800
    },
    {
      "epoch": 107.99,
      "grad_norm": 1.8771851062774658,
      "learning_rate": 2.228076755118477e-06,
      "loss": 0.4743,
      "step": 263810
    },
    {
      "epoch": 107.99,
      "grad_norm": 1.5922449827194214,
      "learning_rate": 2.2279183191478593e-06,
      "loss": 0.4694,
      "step": 263820
    },
    {
      "epoch": 107.99,
      "grad_norm": 2.1600685119628906,
      "learning_rate": 2.227759883707849e-06,
      "loss": 0.4609,
      "step": 263830
    },
    {
      "epoch": 108.0,
      "grad_norm": 1.92283296585083,
      "learning_rate": 2.2276014487991707e-06,
      "loss": 0.4371,
      "step": 263840
    },
    {
      "epoch": 108.0,
      "eval_loss": 0.4719949960708618,
      "eval_runtime": 52.7156,
      "eval_samples_per_second": 65.427,
      "eval_steps_per_second": 8.195,
      "step": 263844
    },
    {
      "epoch": 108.0,
      "grad_norm": 2.5797512531280518,
      "learning_rate": 2.2274430144225517e-06,
      "loss": 0.4546,
      "step": 263850
    },
    {
      "epoch": 108.01,
      "grad_norm": 2.6221694946289062,
      "learning_rate": 2.2272845805787164e-06,
      "loss": 0.4838,
      "step": 263860
    },
    {
      "epoch": 108.01,
      "grad_norm": 2.0679941177368164,
      "learning_rate": 2.2271261472683916e-06,
      "loss": 0.4733,
      "step": 263870
    },
    {
      "epoch": 108.01,
      "grad_norm": 1.882468819618225,
      "learning_rate": 2.2269677144923015e-06,
      "loss": 0.4679,
      "step": 263880
    },
    {
      "epoch": 108.02,
      "grad_norm": 2.128775119781494,
      "learning_rate": 2.226809282251174e-06,
      "loss": 0.4681,
      "step": 263890
    },
    {
      "epoch": 108.02,
      "grad_norm": 1.7824323177337646,
      "learning_rate": 2.2266508505457336e-06,
      "loss": 0.4667,
      "step": 263900
    },
    {
      "epoch": 108.03,
      "grad_norm": 1.474369764328003,
      "learning_rate": 2.226492419376707e-06,
      "loss": 0.4611,
      "step": 263910
    },
    {
      "epoch": 108.03,
      "grad_norm": 1.8729352951049805,
      "learning_rate": 2.226333988744818e-06,
      "loss": 0.4642,
      "step": 263920
    },
    {
      "epoch": 108.04,
      "grad_norm": 1.8685849905014038,
      "learning_rate": 2.2261755586507945e-06,
      "loss": 0.4766,
      "step": 263930
    },
    {
      "epoch": 108.04,
      "grad_norm": 1.7652007341384888,
      "learning_rate": 2.2260171290953612e-06,
      "loss": 0.4812,
      "step": 263940
    },
    {
      "epoch": 108.04,
      "grad_norm": 1.7593560218811035,
      "learning_rate": 2.2258587000792436e-06,
      "loss": 0.4695,
      "step": 263950
    },
    {
      "epoch": 108.05,
      "grad_norm": 1.6844632625579834,
      "learning_rate": 2.225700271603168e-06,
      "loss": 0.4688,
      "step": 263960
    },
    {
      "epoch": 108.05,
      "grad_norm": 2.014920949935913,
      "learning_rate": 2.22554184366786e-06,
      "loss": 0.4723,
      "step": 263970
    },
    {
      "epoch": 108.06,
      "grad_norm": 1.5490047931671143,
      "learning_rate": 2.2253834162740447e-06,
      "loss": 0.4795,
      "step": 263980
    },
    {
      "epoch": 108.06,
      "grad_norm": 1.925708293914795,
      "learning_rate": 2.2252249894224486e-06,
      "loss": 0.4764,
      "step": 263990
    },
    {
      "epoch": 108.06,
      "grad_norm": 1.9699532985687256,
      "learning_rate": 2.2250665631137976e-06,
      "loss": 0.4898,
      "step": 264000
    },
    {
      "epoch": 108.07,
      "grad_norm": 2.378998041152954,
      "learning_rate": 2.224908137348817e-06,
      "loss": 0.4721,
      "step": 264010
    },
    {
      "epoch": 108.07,
      "grad_norm": 1.8817341327667236,
      "learning_rate": 2.2247497121282323e-06,
      "loss": 0.4564,
      "step": 264020
    },
    {
      "epoch": 108.08,
      "grad_norm": 1.7890400886535645,
      "learning_rate": 2.2245912874527697e-06,
      "loss": 0.4756,
      "step": 264030
    },
    {
      "epoch": 108.08,
      "grad_norm": 1.8714886903762817,
      "learning_rate": 2.2244328633231546e-06,
      "loss": 0.4731,
      "step": 264040
    },
    {
      "epoch": 108.08,
      "grad_norm": 2.1724941730499268,
      "learning_rate": 2.224274439740113e-06,
      "loss": 0.4544,
      "step": 264050
    },
    {
      "epoch": 108.09,
      "grad_norm": 1.681117057800293,
      "learning_rate": 2.2241160167043694e-06,
      "loss": 0.4692,
      "step": 264060
    },
    {
      "epoch": 108.09,
      "grad_norm": 1.7715520858764648,
      "learning_rate": 2.223957594216651e-06,
      "loss": 0.4837,
      "step": 264070
    },
    {
      "epoch": 108.1,
      "grad_norm": 1.9441967010498047,
      "learning_rate": 2.2237991722776842e-06,
      "loss": 0.4512,
      "step": 264080
    },
    {
      "epoch": 108.1,
      "grad_norm": 1.828689455986023,
      "learning_rate": 2.223640750888193e-06,
      "loss": 0.479,
      "step": 264090
    },
    {
      "epoch": 108.1,
      "grad_norm": 1.9662120342254639,
      "learning_rate": 2.2234823300489035e-06,
      "loss": 0.4846,
      "step": 264100
    },
    {
      "epoch": 108.11,
      "grad_norm": 1.4386883974075317,
      "learning_rate": 2.2233239097605413e-06,
      "loss": 0.4752,
      "step": 264110
    },
    {
      "epoch": 108.11,
      "grad_norm": 1.618241548538208,
      "learning_rate": 2.2231654900238327e-06,
      "loss": 0.4868,
      "step": 264120
    },
    {
      "epoch": 108.12,
      "grad_norm": 1.975590467453003,
      "learning_rate": 2.223007070839503e-06,
      "loss": 0.4696,
      "step": 264130
    },
    {
      "epoch": 108.12,
      "grad_norm": 2.0532569885253906,
      "learning_rate": 2.2228486522082778e-06,
      "loss": 0.4899,
      "step": 264140
    },
    {
      "epoch": 108.13,
      "grad_norm": 2.023365020751953,
      "learning_rate": 2.2226902341308834e-06,
      "loss": 0.4634,
      "step": 264150
    },
    {
      "epoch": 108.13,
      "grad_norm": 1.6739810705184937,
      "learning_rate": 2.2225318166080444e-06,
      "loss": 0.4757,
      "step": 264160
    },
    {
      "epoch": 108.13,
      "grad_norm": 1.7876936197280884,
      "learning_rate": 2.222373399640488e-06,
      "loss": 0.4624,
      "step": 264170
    },
    {
      "epoch": 108.14,
      "grad_norm": 1.8263922929763794,
      "learning_rate": 2.2222149832289385e-06,
      "loss": 0.4823,
      "step": 264180
    },
    {
      "epoch": 108.14,
      "grad_norm": 1.633536696434021,
      "learning_rate": 2.222056567374122e-06,
      "loss": 0.4656,
      "step": 264190
    },
    {
      "epoch": 108.15,
      "grad_norm": 1.8804218769073486,
      "learning_rate": 2.2218981520767644e-06,
      "loss": 0.4818,
      "step": 264200
    },
    {
      "epoch": 108.15,
      "grad_norm": 2.1024324893951416,
      "learning_rate": 2.2217397373375912e-06,
      "loss": 0.4762,
      "step": 264210
    },
    {
      "epoch": 108.15,
      "grad_norm": 2.191147804260254,
      "learning_rate": 2.221581323157328e-06,
      "loss": 0.4897,
      "step": 264220
    },
    {
      "epoch": 108.16,
      "grad_norm": 1.9810622930526733,
      "learning_rate": 2.221422909536701e-06,
      "loss": 0.4592,
      "step": 264230
    },
    {
      "epoch": 108.16,
      "grad_norm": 1.9545278549194336,
      "learning_rate": 2.221264496476435e-06,
      "loss": 0.4554,
      "step": 264240
    },
    {
      "epoch": 108.17,
      "grad_norm": 2.1508679389953613,
      "learning_rate": 2.2211060839772553e-06,
      "loss": 0.476,
      "step": 264250
    },
    {
      "epoch": 108.17,
      "grad_norm": 2.1391334533691406,
      "learning_rate": 2.2209476720398897e-06,
      "loss": 0.4716,
      "step": 264260
    },
    {
      "epoch": 108.17,
      "grad_norm": 1.4701528549194336,
      "learning_rate": 2.2207892606650623e-06,
      "loss": 0.4842,
      "step": 264270
    },
    {
      "epoch": 108.18,
      "grad_norm": 1.874107837677002,
      "learning_rate": 2.2206308498534987e-06,
      "loss": 0.4762,
      "step": 264280
    },
    {
      "epoch": 108.18,
      "grad_norm": 1.853238821029663,
      "learning_rate": 2.2204724396059253e-06,
      "loss": 0.4888,
      "step": 264290
    },
    {
      "epoch": 108.19,
      "grad_norm": 1.8995529413223267,
      "learning_rate": 2.2203140299230674e-06,
      "loss": 0.4904,
      "step": 264300
    },
    {
      "epoch": 108.19,
      "grad_norm": 1.886446475982666,
      "learning_rate": 2.22015562080565e-06,
      "loss": 0.472,
      "step": 264310
    },
    {
      "epoch": 108.19,
      "grad_norm": 2.0811634063720703,
      "learning_rate": 2.2199972122543995e-06,
      "loss": 0.4764,
      "step": 264320
    },
    {
      "epoch": 108.2,
      "grad_norm": 2.1515884399414062,
      "learning_rate": 2.2198388042700414e-06,
      "loss": 0.4707,
      "step": 264330
    },
    {
      "epoch": 108.2,
      "grad_norm": 1.8858888149261475,
      "learning_rate": 2.2196803968533016e-06,
      "loss": 0.4534,
      "step": 264340
    },
    {
      "epoch": 108.21,
      "grad_norm": 2.094381093978882,
      "learning_rate": 2.2195219900049048e-06,
      "loss": 0.4809,
      "step": 264350
    },
    {
      "epoch": 108.21,
      "grad_norm": 2.201569080352783,
      "learning_rate": 2.2193635837255778e-06,
      "loss": 0.4793,
      "step": 264360
    },
    {
      "epoch": 108.22,
      "grad_norm": 2.127124786376953,
      "learning_rate": 2.219205178016045e-06,
      "loss": 0.4561,
      "step": 264370
    },
    {
      "epoch": 108.22,
      "grad_norm": 2.282111644744873,
      "learning_rate": 2.2190467728770337e-06,
      "loss": 0.4696,
      "step": 264380
    },
    {
      "epoch": 108.22,
      "grad_norm": 2.03836989402771,
      "learning_rate": 2.2188883683092685e-06,
      "loss": 0.4617,
      "step": 264390
    },
    {
      "epoch": 108.23,
      "grad_norm": 1.6197261810302734,
      "learning_rate": 2.218729964313475e-06,
      "loss": 0.4531,
      "step": 264400
    },
    {
      "epoch": 108.23,
      "grad_norm": 2.203279972076416,
      "learning_rate": 2.2185715608903786e-06,
      "loss": 0.4743,
      "step": 264410
    },
    {
      "epoch": 108.24,
      "grad_norm": 2.3427512645721436,
      "learning_rate": 2.2184131580407053e-06,
      "loss": 0.4694,
      "step": 264420
    },
    {
      "epoch": 108.24,
      "grad_norm": 2.3681576251983643,
      "learning_rate": 2.2182547557651807e-06,
      "loss": 0.4845,
      "step": 264430
    },
    {
      "epoch": 108.24,
      "grad_norm": 2.0171127319335938,
      "learning_rate": 2.2180963540645296e-06,
      "loss": 0.4675,
      "step": 264440
    },
    {
      "epoch": 108.25,
      "grad_norm": 1.9453195333480835,
      "learning_rate": 2.217937952939479e-06,
      "loss": 0.4656,
      "step": 264450
    },
    {
      "epoch": 108.25,
      "grad_norm": 1.8878750801086426,
      "learning_rate": 2.217779552390754e-06,
      "loss": 0.4652,
      "step": 264460
    },
    {
      "epoch": 108.26,
      "grad_norm": 1.6711534261703491,
      "learning_rate": 2.2176211524190804e-06,
      "loss": 0.4628,
      "step": 264470
    },
    {
      "epoch": 108.26,
      "grad_norm": 1.8394039869308472,
      "learning_rate": 2.217462753025183e-06,
      "loss": 0.4849,
      "step": 264480
    },
    {
      "epoch": 108.26,
      "grad_norm": 2.0926735401153564,
      "learning_rate": 2.2173043542097883e-06,
      "loss": 0.4724,
      "step": 264490
    },
    {
      "epoch": 108.27,
      "grad_norm": 1.9803647994995117,
      "learning_rate": 2.2171459559736215e-06,
      "loss": 0.455,
      "step": 264500
    },
    {
      "epoch": 108.27,
      "grad_norm": 1.9174772500991821,
      "learning_rate": 2.216987558317408e-06,
      "loss": 0.489,
      "step": 264510
    },
    {
      "epoch": 108.28,
      "grad_norm": 1.7697142362594604,
      "learning_rate": 2.216829161241874e-06,
      "loss": 0.4575,
      "step": 264520
    },
    {
      "epoch": 108.28,
      "grad_norm": 1.9961891174316406,
      "learning_rate": 2.2166707647477443e-06,
      "loss": 0.4902,
      "step": 264530
    },
    {
      "epoch": 108.28,
      "grad_norm": 1.645766258239746,
      "learning_rate": 2.2165123688357452e-06,
      "loss": 0.4345,
      "step": 264540
    },
    {
      "epoch": 108.29,
      "grad_norm": 1.8519126176834106,
      "learning_rate": 2.216353973506602e-06,
      "loss": 0.4657,
      "step": 264550
    },
    {
      "epoch": 108.29,
      "grad_norm": 1.656644344329834,
      "learning_rate": 2.21619557876104e-06,
      "loss": 0.4825,
      "step": 264560
    },
    {
      "epoch": 108.3,
      "grad_norm": 1.5012187957763672,
      "learning_rate": 2.216037184599785e-06,
      "loss": 0.4632,
      "step": 264570
    },
    {
      "epoch": 108.3,
      "grad_norm": 2.289595365524292,
      "learning_rate": 2.215878791023563e-06,
      "loss": 0.451,
      "step": 264580
    },
    {
      "epoch": 108.31,
      "grad_norm": 1.8871408700942993,
      "learning_rate": 2.2157203980330986e-06,
      "loss": 0.4717,
      "step": 264590
    },
    {
      "epoch": 108.31,
      "grad_norm": 1.894157886505127,
      "learning_rate": 2.2155620056291186e-06,
      "loss": 0.4984,
      "step": 264600
    },
    {
      "epoch": 108.31,
      "grad_norm": 1.9395111799240112,
      "learning_rate": 2.215403613812348e-06,
      "loss": 0.4576,
      "step": 264610
    },
    {
      "epoch": 108.32,
      "grad_norm": 2.0884644985198975,
      "learning_rate": 2.215245222583512e-06,
      "loss": 0.4426,
      "step": 264620
    },
    {
      "epoch": 108.32,
      "grad_norm": 1.5760829448699951,
      "learning_rate": 2.2150868319433355e-06,
      "loss": 0.4927,
      "step": 264630
    },
    {
      "epoch": 108.33,
      "grad_norm": 1.900572419166565,
      "learning_rate": 2.214928441892546e-06,
      "loss": 0.4702,
      "step": 264640
    },
    {
      "epoch": 108.33,
      "grad_norm": 1.960991382598877,
      "learning_rate": 2.2147700524318684e-06,
      "loss": 0.4825,
      "step": 264650
    },
    {
      "epoch": 108.33,
      "grad_norm": 1.7617230415344238,
      "learning_rate": 2.214611663562028e-06,
      "loss": 0.4757,
      "step": 264660
    },
    {
      "epoch": 108.34,
      "grad_norm": 1.9030927419662476,
      "learning_rate": 2.21445327528375e-06,
      "loss": 0.4445,
      "step": 264670
    },
    {
      "epoch": 108.34,
      "grad_norm": 1.6438716650009155,
      "learning_rate": 2.2142948875977604e-06,
      "loss": 0.465,
      "step": 264680
    },
    {
      "epoch": 108.35,
      "grad_norm": 1.9280592203140259,
      "learning_rate": 2.214136500504784e-06,
      "loss": 0.477,
      "step": 264690
    },
    {
      "epoch": 108.35,
      "grad_norm": 1.548970341682434,
      "learning_rate": 2.213978114005548e-06,
      "loss": 0.4921,
      "step": 264700
    },
    {
      "epoch": 108.35,
      "grad_norm": 1.7895195484161377,
      "learning_rate": 2.213819728100776e-06,
      "loss": 0.4701,
      "step": 264710
    },
    {
      "epoch": 108.36,
      "grad_norm": 2.096912384033203,
      "learning_rate": 2.213661342791195e-06,
      "loss": 0.4632,
      "step": 264720
    },
    {
      "epoch": 108.36,
      "grad_norm": 1.918210506439209,
      "learning_rate": 2.21350295807753e-06,
      "loss": 0.4672,
      "step": 264730
    },
    {
      "epoch": 108.37,
      "grad_norm": 1.9224927425384521,
      "learning_rate": 2.2133445739605064e-06,
      "loss": 0.4493,
      "step": 264740
    },
    {
      "epoch": 108.37,
      "grad_norm": 1.903918981552124,
      "learning_rate": 2.2131861904408505e-06,
      "loss": 0.4591,
      "step": 264750
    },
    {
      "epoch": 108.37,
      "grad_norm": 2.056868553161621,
      "learning_rate": 2.2130278075192863e-06,
      "loss": 0.4486,
      "step": 264760
    },
    {
      "epoch": 108.38,
      "grad_norm": 2.416749954223633,
      "learning_rate": 2.2128694251965404e-06,
      "loss": 0.477,
      "step": 264770
    },
    {
      "epoch": 108.38,
      "grad_norm": 2.0756568908691406,
      "learning_rate": 2.212711043473338e-06,
      "loss": 0.4563,
      "step": 264780
    },
    {
      "epoch": 108.39,
      "grad_norm": 2.2756731510162354,
      "learning_rate": 2.212552662350405e-06,
      "loss": 0.4756,
      "step": 264790
    },
    {
      "epoch": 108.39,
      "grad_norm": 2.158323049545288,
      "learning_rate": 2.212394281828467e-06,
      "loss": 0.47,
      "step": 264800
    },
    {
      "epoch": 108.4,
      "grad_norm": 2.090553045272827,
      "learning_rate": 2.2122359019082487e-06,
      "loss": 0.4604,
      "step": 264810
    },
    {
      "epoch": 108.4,
      "grad_norm": 1.8905572891235352,
      "learning_rate": 2.2120775225904755e-06,
      "loss": 0.4708,
      "step": 264820
    },
    {
      "epoch": 108.4,
      "grad_norm": 2.4787585735321045,
      "learning_rate": 2.2119191438758745e-06,
      "loss": 0.4685,
      "step": 264830
    },
    {
      "epoch": 108.41,
      "grad_norm": 2.2773447036743164,
      "learning_rate": 2.2117607657651698e-06,
      "loss": 0.4753,
      "step": 264840
    },
    {
      "epoch": 108.41,
      "grad_norm": 2.5068984031677246,
      "learning_rate": 2.211602388259088e-06,
      "loss": 0.4591,
      "step": 264850
    },
    {
      "epoch": 108.42,
      "grad_norm": 2.0973334312438965,
      "learning_rate": 2.211444011358353e-06,
      "loss": 0.473,
      "step": 264860
    },
    {
      "epoch": 108.42,
      "grad_norm": 1.661222219467163,
      "learning_rate": 2.2112856350636918e-06,
      "loss": 0.4842,
      "step": 264870
    },
    {
      "epoch": 108.42,
      "grad_norm": 1.9452941417694092,
      "learning_rate": 2.21112725937583e-06,
      "loss": 0.4515,
      "step": 264880
    },
    {
      "epoch": 108.43,
      "grad_norm": 1.7744674682617188,
      "learning_rate": 2.2109688842954916e-06,
      "loss": 0.5002,
      "step": 264890
    },
    {
      "epoch": 108.43,
      "grad_norm": 1.8192037343978882,
      "learning_rate": 2.2108105098234032e-06,
      "loss": 0.4777,
      "step": 264900
    },
    {
      "epoch": 108.44,
      "grad_norm": 1.915348768234253,
      "learning_rate": 2.2106521359602897e-06,
      "loss": 0.4709,
      "step": 264910
    },
    {
      "epoch": 108.44,
      "grad_norm": 1.9193732738494873,
      "learning_rate": 2.2104937627068772e-06,
      "loss": 0.4714,
      "step": 264920
    },
    {
      "epoch": 108.44,
      "grad_norm": 1.6869513988494873,
      "learning_rate": 2.2103353900638905e-06,
      "loss": 0.4701,
      "step": 264930
    },
    {
      "epoch": 108.45,
      "grad_norm": 1.7460709810256958,
      "learning_rate": 2.210177018032056e-06,
      "loss": 0.4702,
      "step": 264940
    },
    {
      "epoch": 108.45,
      "grad_norm": 1.823136806488037,
      "learning_rate": 2.2100186466120983e-06,
      "loss": 0.4595,
      "step": 264950
    },
    {
      "epoch": 108.46,
      "grad_norm": 1.933951497077942,
      "learning_rate": 2.2098602758047434e-06,
      "loss": 0.4738,
      "step": 264960
    },
    {
      "epoch": 108.46,
      "grad_norm": 1.863821029663086,
      "learning_rate": 2.209701905610717e-06,
      "loss": 0.4752,
      "step": 264970
    },
    {
      "epoch": 108.47,
      "grad_norm": 2.002897024154663,
      "learning_rate": 2.2095435360307436e-06,
      "loss": 0.468,
      "step": 264980
    },
    {
      "epoch": 108.47,
      "grad_norm": 1.7867746353149414,
      "learning_rate": 2.209385167065549e-06,
      "loss": 0.4864,
      "step": 264990
    },
    {
      "epoch": 108.47,
      "grad_norm": 2.291205406188965,
      "learning_rate": 2.2092267987158593e-06,
      "loss": 0.4643,
      "step": 265000
    },
    {
      "epoch": 108.48,
      "grad_norm": 2.1922178268432617,
      "learning_rate": 2.2090684309823983e-06,
      "loss": 0.493,
      "step": 265010
    },
    {
      "epoch": 108.48,
      "grad_norm": 1.5894662141799927,
      "learning_rate": 2.208910063865894e-06,
      "loss": 0.469,
      "step": 265020
    },
    {
      "epoch": 108.49,
      "grad_norm": 1.890045166015625,
      "learning_rate": 2.2087516973670706e-06,
      "loss": 0.4766,
      "step": 265030
    },
    {
      "epoch": 108.49,
      "grad_norm": 2.1932997703552246,
      "learning_rate": 2.2085933314866534e-06,
      "loss": 0.4584,
      "step": 265040
    },
    {
      "epoch": 108.49,
      "grad_norm": 1.7219703197479248,
      "learning_rate": 2.208434966225368e-06,
      "loss": 0.4674,
      "step": 265050
    },
    {
      "epoch": 108.5,
      "grad_norm": 1.5680756568908691,
      "learning_rate": 2.2082766015839397e-06,
      "loss": 0.4534,
      "step": 265060
    },
    {
      "epoch": 108.5,
      "grad_norm": 1.5563830137252808,
      "learning_rate": 2.208118237563094e-06,
      "loss": 0.4554,
      "step": 265070
    },
    {
      "epoch": 108.51,
      "grad_norm": 1.6594756841659546,
      "learning_rate": 2.2079598741635563e-06,
      "loss": 0.4707,
      "step": 265080
    },
    {
      "epoch": 108.51,
      "grad_norm": 2.1742186546325684,
      "learning_rate": 2.207801511386052e-06,
      "loss": 0.4624,
      "step": 265090
    },
    {
      "epoch": 108.51,
      "grad_norm": 2.1717922687530518,
      "learning_rate": 2.2076431492313073e-06,
      "loss": 0.4573,
      "step": 265100
    },
    {
      "epoch": 108.52,
      "grad_norm": 1.90884268283844,
      "learning_rate": 2.2074847877000473e-06,
      "loss": 0.4813,
      "step": 265110
    },
    {
      "epoch": 108.52,
      "grad_norm": 1.9552255868911743,
      "learning_rate": 2.2073264267929964e-06,
      "loss": 0.485,
      "step": 265120
    },
    {
      "epoch": 108.53,
      "grad_norm": 1.8631318807601929,
      "learning_rate": 2.207168066510881e-06,
      "loss": 0.4967,
      "step": 265130
    },
    {
      "epoch": 108.53,
      "grad_norm": 1.8960226774215698,
      "learning_rate": 2.207009706854426e-06,
      "loss": 0.4848,
      "step": 265140
    },
    {
      "epoch": 108.53,
      "grad_norm": 1.7950153350830078,
      "learning_rate": 2.206851347824357e-06,
      "loss": 0.4514,
      "step": 265150
    },
    {
      "epoch": 108.54,
      "grad_norm": 1.9893521070480347,
      "learning_rate": 2.2066929894214e-06,
      "loss": 0.4714,
      "step": 265160
    },
    {
      "epoch": 108.54,
      "grad_norm": 2.2915258407592773,
      "learning_rate": 2.2065346316462797e-06,
      "loss": 0.4639,
      "step": 265170
    },
    {
      "epoch": 108.55,
      "grad_norm": 1.7691895961761475,
      "learning_rate": 2.2063762744997214e-06,
      "loss": 0.4738,
      "step": 265180
    },
    {
      "epoch": 108.55,
      "grad_norm": 1.8819926977157593,
      "learning_rate": 2.2062179179824517e-06,
      "loss": 0.4821,
      "step": 265190
    },
    {
      "epoch": 108.56,
      "grad_norm": 1.9531065225601196,
      "learning_rate": 2.206059562095194e-06,
      "loss": 0.4654,
      "step": 265200
    },
    {
      "epoch": 108.56,
      "grad_norm": 1.6201008558273315,
      "learning_rate": 2.2059012068386755e-06,
      "loss": 0.4519,
      "step": 265210
    },
    {
      "epoch": 108.56,
      "grad_norm": 1.98509681224823,
      "learning_rate": 2.2057428522136208e-06,
      "loss": 0.4806,
      "step": 265220
    },
    {
      "epoch": 108.57,
      "grad_norm": 1.8013503551483154,
      "learning_rate": 2.205584498220756e-06,
      "loss": 0.486,
      "step": 265230
    },
    {
      "epoch": 108.57,
      "grad_norm": 2.3649137020111084,
      "learning_rate": 2.2054261448608058e-06,
      "loss": 0.4826,
      "step": 265240
    },
    {
      "epoch": 108.58,
      "grad_norm": 1.9655346870422363,
      "learning_rate": 2.2052677921344956e-06,
      "loss": 0.4604,
      "step": 265250
    },
    {
      "epoch": 108.58,
      "grad_norm": 2.4428088665008545,
      "learning_rate": 2.2051094400425513e-06,
      "loss": 0.4587,
      "step": 265260
    },
    {
      "epoch": 108.58,
      "grad_norm": 2.0760297775268555,
      "learning_rate": 2.2049510885856975e-06,
      "loss": 0.4613,
      "step": 265270
    },
    {
      "epoch": 108.59,
      "grad_norm": 2.6010591983795166,
      "learning_rate": 2.20479273776466e-06,
      "loss": 0.4618,
      "step": 265280
    },
    {
      "epoch": 108.59,
      "grad_norm": 2.286062002182007,
      "learning_rate": 2.2046343875801646e-06,
      "loss": 0.4733,
      "step": 265290
    },
    {
      "epoch": 108.6,
      "grad_norm": 1.715563416481018,
      "learning_rate": 2.2044760380329355e-06,
      "loss": 0.4812,
      "step": 265300
    },
    {
      "epoch": 108.6,
      "grad_norm": 2.4415643215179443,
      "learning_rate": 2.2043176891236998e-06,
      "loss": 0.4845,
      "step": 265310
    },
    {
      "epoch": 108.6,
      "grad_norm": 1.893567442893982,
      "learning_rate": 2.204159340853181e-06,
      "loss": 0.4635,
      "step": 265320
    },
    {
      "epoch": 108.61,
      "grad_norm": 2.4160287380218506,
      "learning_rate": 2.2040009932221065e-06,
      "loss": 0.4539,
      "step": 265330
    },
    {
      "epoch": 108.61,
      "grad_norm": 2.2319223880767822,
      "learning_rate": 2.2038426462312e-06,
      "loss": 0.4696,
      "step": 265340
    },
    {
      "epoch": 108.62,
      "grad_norm": 2.3742899894714355,
      "learning_rate": 2.2036842998811875e-06,
      "loss": 0.4723,
      "step": 265350
    },
    {
      "epoch": 108.62,
      "grad_norm": 2.5811731815338135,
      "learning_rate": 2.203525954172794e-06,
      "loss": 0.4569,
      "step": 265360
    },
    {
      "epoch": 108.62,
      "grad_norm": 1.8949427604675293,
      "learning_rate": 2.203367609106745e-06,
      "loss": 0.4646,
      "step": 265370
    },
    {
      "epoch": 108.63,
      "grad_norm": 1.6489068269729614,
      "learning_rate": 2.2032092646837657e-06,
      "loss": 0.4663,
      "step": 265380
    },
    {
      "epoch": 108.63,
      "grad_norm": 1.8663859367370605,
      "learning_rate": 2.2030509209045825e-06,
      "loss": 0.4682,
      "step": 265390
    },
    {
      "epoch": 108.64,
      "grad_norm": 1.9027161598205566,
      "learning_rate": 2.2028925777699197e-06,
      "loss": 0.4865,
      "step": 265400
    },
    {
      "epoch": 108.64,
      "grad_norm": 2.1835126876831055,
      "learning_rate": 2.202734235280503e-06,
      "loss": 0.4864,
      "step": 265410
    },
    {
      "epoch": 108.65,
      "grad_norm": 1.9423080682754517,
      "learning_rate": 2.2025758934370576e-06,
      "loss": 0.4752,
      "step": 265420
    },
    {
      "epoch": 108.65,
      "grad_norm": 1.9982479810714722,
      "learning_rate": 2.202417552240309e-06,
      "loss": 0.4427,
      "step": 265430
    },
    {
      "epoch": 108.65,
      "grad_norm": 2.195220947265625,
      "learning_rate": 2.2022592116909824e-06,
      "loss": 0.469,
      "step": 265440
    },
    {
      "epoch": 108.66,
      "grad_norm": 1.9553035497665405,
      "learning_rate": 2.2021008717898035e-06,
      "loss": 0.4716,
      "step": 265450
    },
    {
      "epoch": 108.66,
      "grad_norm": 1.7622737884521484,
      "learning_rate": 2.201942532537497e-06,
      "loss": 0.4803,
      "step": 265460
    },
    {
      "epoch": 108.67,
      "grad_norm": 2.0913052558898926,
      "learning_rate": 2.2017841939347893e-06,
      "loss": 0.4534,
      "step": 265470
    },
    {
      "epoch": 108.67,
      "grad_norm": 1.9973795413970947,
      "learning_rate": 2.201625855982404e-06,
      "loss": 0.4681,
      "step": 265480
    },
    {
      "epoch": 108.67,
      "grad_norm": 1.7145726680755615,
      "learning_rate": 2.201467518681068e-06,
      "loss": 0.4676,
      "step": 265490
    },
    {
      "epoch": 108.68,
      "grad_norm": 2.048104763031006,
      "learning_rate": 2.2013091820315055e-06,
      "loss": 0.4761,
      "step": 265500
    },
    {
      "epoch": 108.68,
      "grad_norm": 2.261502504348755,
      "learning_rate": 2.2011508460344425e-06,
      "loss": 0.4732,
      "step": 265510
    },
    {
      "epoch": 108.69,
      "grad_norm": 2.1775991916656494,
      "learning_rate": 2.200992510690604e-06,
      "loss": 0.4787,
      "step": 265520
    },
    {
      "epoch": 108.69,
      "grad_norm": 2.0273053646087646,
      "learning_rate": 2.200834176000716e-06,
      "loss": 0.4884,
      "step": 265530
    },
    {
      "epoch": 108.69,
      "grad_norm": 1.6745151281356812,
      "learning_rate": 2.2006758419655027e-06,
      "loss": 0.4717,
      "step": 265540
    },
    {
      "epoch": 108.7,
      "grad_norm": 2.1098718643188477,
      "learning_rate": 2.2005175085856906e-06,
      "loss": 0.47,
      "step": 265550
    },
    {
      "epoch": 108.7,
      "grad_norm": 1.9634698629379272,
      "learning_rate": 2.200359175862004e-06,
      "loss": 0.4708,
      "step": 265560
    },
    {
      "epoch": 108.71,
      "grad_norm": 2.169196605682373,
      "learning_rate": 2.2002008437951673e-06,
      "loss": 0.4828,
      "step": 265570
    },
    {
      "epoch": 108.71,
      "grad_norm": 2.1294777393341064,
      "learning_rate": 2.2000425123859084e-06,
      "loss": 0.4634,
      "step": 265580
    },
    {
      "epoch": 108.71,
      "grad_norm": 1.8405812978744507,
      "learning_rate": 2.1998841816349507e-06,
      "loss": 0.4806,
      "step": 265590
    },
    {
      "epoch": 108.72,
      "grad_norm": 2.216047763824463,
      "learning_rate": 2.199725851543021e-06,
      "loss": 0.479,
      "step": 265600
    },
    {
      "epoch": 108.72,
      "grad_norm": 1.869205117225647,
      "learning_rate": 2.199567522110843e-06,
      "loss": 0.473,
      "step": 265610
    },
    {
      "epoch": 108.73,
      "grad_norm": 2.0512735843658447,
      "learning_rate": 2.1994091933391425e-06,
      "loss": 0.4695,
      "step": 265620
    },
    {
      "epoch": 108.73,
      "grad_norm": 2.207587480545044,
      "learning_rate": 2.199250865228645e-06,
      "loss": 0.4725,
      "step": 265630
    },
    {
      "epoch": 108.74,
      "grad_norm": 2.0039708614349365,
      "learning_rate": 2.1990925377800756e-06,
      "loss": 0.4532,
      "step": 265640
    },
    {
      "epoch": 108.74,
      "grad_norm": 3.0351433753967285,
      "learning_rate": 2.1989342109941596e-06,
      "loss": 0.484,
      "step": 265650
    },
    {
      "epoch": 108.74,
      "grad_norm": 2.021446466445923,
      "learning_rate": 2.1987758848716225e-06,
      "loss": 0.4859,
      "step": 265660
    },
    {
      "epoch": 108.75,
      "grad_norm": 1.887147068977356,
      "learning_rate": 2.1986175594131893e-06,
      "loss": 0.4569,
      "step": 265670
    },
    {
      "epoch": 108.75,
      "grad_norm": 2.1856296062469482,
      "learning_rate": 2.1984592346195855e-06,
      "loss": 0.4622,
      "step": 265680
    },
    {
      "epoch": 108.76,
      "grad_norm": 2.0155029296875,
      "learning_rate": 2.1983009104915365e-06,
      "loss": 0.4703,
      "step": 265690
    },
    {
      "epoch": 108.76,
      "grad_norm": 1.6388819217681885,
      "learning_rate": 2.1981425870297666e-06,
      "loss": 0.4849,
      "step": 265700
    },
    {
      "epoch": 108.76,
      "grad_norm": 2.154585123062134,
      "learning_rate": 2.1979842642350017e-06,
      "loss": 0.4716,
      "step": 265710
    },
    {
      "epoch": 108.77,
      "grad_norm": 2.1216256618499756,
      "learning_rate": 2.1978259421079673e-06,
      "loss": 0.468,
      "step": 265720
    },
    {
      "epoch": 108.77,
      "grad_norm": 2.272280216217041,
      "learning_rate": 2.1976676206493884e-06,
      "loss": 0.4444,
      "step": 265730
    },
    {
      "epoch": 108.78,
      "grad_norm": 1.8779125213623047,
      "learning_rate": 2.19750929985999e-06,
      "loss": 0.4717,
      "step": 265740
    },
    {
      "epoch": 108.78,
      "grad_norm": 2.0797359943389893,
      "learning_rate": 2.197350979740498e-06,
      "loss": 0.4748,
      "step": 265750
    },
    {
      "epoch": 108.78,
      "grad_norm": 2.5136477947235107,
      "learning_rate": 2.1971926602916364e-06,
      "loss": 0.4911,
      "step": 265760
    },
    {
      "epoch": 108.79,
      "grad_norm": 1.8391940593719482,
      "learning_rate": 2.1970343415141315e-06,
      "loss": 0.4842,
      "step": 265770
    },
    {
      "epoch": 108.79,
      "grad_norm": 2.087584972381592,
      "learning_rate": 2.196876023408709e-06,
      "loss": 0.4527,
      "step": 265780
    },
    {
      "epoch": 108.8,
      "grad_norm": 1.8711243867874146,
      "learning_rate": 2.196717705976093e-06,
      "loss": 0.4729,
      "step": 265790
    },
    {
      "epoch": 108.8,
      "grad_norm": 2.180600643157959,
      "learning_rate": 2.1965593892170092e-06,
      "loss": 0.4827,
      "step": 265800
    },
    {
      "epoch": 108.8,
      "grad_norm": 1.850122332572937,
      "learning_rate": 2.196401073132183e-06,
      "loss": 0.4681,
      "step": 265810
    },
    {
      "epoch": 108.81,
      "grad_norm": 1.5753651857376099,
      "learning_rate": 2.1962427577223393e-06,
      "loss": 0.4699,
      "step": 265820
    },
    {
      "epoch": 108.81,
      "grad_norm": 2.1135523319244385,
      "learning_rate": 2.196084442988204e-06,
      "loss": 0.4633,
      "step": 265830
    },
    {
      "epoch": 108.82,
      "grad_norm": 2.5021743774414062,
      "learning_rate": 2.1959261289305008e-06,
      "loss": 0.4775,
      "step": 265840
    },
    {
      "epoch": 108.82,
      "grad_norm": 1.732230544090271,
      "learning_rate": 2.1957678155499565e-06,
      "loss": 0.4861,
      "step": 265850
    },
    {
      "epoch": 108.83,
      "grad_norm": 1.8050205707550049,
      "learning_rate": 2.1956095028472953e-06,
      "loss": 0.4721,
      "step": 265860
    },
    {
      "epoch": 108.83,
      "grad_norm": 1.5713047981262207,
      "learning_rate": 2.195451190823243e-06,
      "loss": 0.4456,
      "step": 265870
    },
    {
      "epoch": 108.83,
      "grad_norm": 2.1391615867614746,
      "learning_rate": 2.1952928794785246e-06,
      "loss": 0.4696,
      "step": 265880
    },
    {
      "epoch": 108.84,
      "grad_norm": 2.1072728633880615,
      "learning_rate": 2.1951345688138652e-06,
      "loss": 0.4817,
      "step": 265890
    },
    {
      "epoch": 108.84,
      "grad_norm": 1.7552084922790527,
      "learning_rate": 2.1949762588299896e-06,
      "loss": 0.4757,
      "step": 265900
    },
    {
      "epoch": 108.85,
      "grad_norm": 1.8289462327957153,
      "learning_rate": 2.194817949527625e-06,
      "loss": 0.451,
      "step": 265910
    },
    {
      "epoch": 108.85,
      "grad_norm": 2.193767547607422,
      "learning_rate": 2.1946596409074934e-06,
      "loss": 0.4611,
      "step": 265920
    },
    {
      "epoch": 108.85,
      "grad_norm": 2.011859893798828,
      "learning_rate": 2.1945013329703225e-06,
      "loss": 0.4789,
      "step": 265930
    },
    {
      "epoch": 108.86,
      "grad_norm": 1.928481101989746,
      "learning_rate": 2.194343025716836e-06,
      "loss": 0.462,
      "step": 265940
    },
    {
      "epoch": 108.86,
      "grad_norm": 2.3794476985931396,
      "learning_rate": 2.194184719147759e-06,
      "loss": 0.4838,
      "step": 265950
    },
    {
      "epoch": 108.87,
      "grad_norm": 2.009772539138794,
      "learning_rate": 2.194026413263819e-06,
      "loss": 0.4757,
      "step": 265960
    },
    {
      "epoch": 108.87,
      "grad_norm": 2.1598081588745117,
      "learning_rate": 2.193868108065739e-06,
      "loss": 0.4898,
      "step": 265970
    },
    {
      "epoch": 108.87,
      "grad_norm": 2.0272748470306396,
      "learning_rate": 2.1937098035542445e-06,
      "loss": 0.4708,
      "step": 265980
    },
    {
      "epoch": 108.88,
      "grad_norm": 2.2533371448516846,
      "learning_rate": 2.1935514997300615e-06,
      "loss": 0.474,
      "step": 265990
    },
    {
      "epoch": 108.88,
      "grad_norm": 2.061615467071533,
      "learning_rate": 2.193393196593914e-06,
      "loss": 0.4762,
      "step": 266000
    },
    {
      "epoch": 108.89,
      "grad_norm": 2.4748995304107666,
      "learning_rate": 2.193234894146528e-06,
      "loss": 0.4536,
      "step": 266010
    },
    {
      "epoch": 108.89,
      "grad_norm": 1.7977215051651,
      "learning_rate": 2.1930765923886285e-06,
      "loss": 0.453,
      "step": 266020
    },
    {
      "epoch": 108.89,
      "grad_norm": 2.0544323921203613,
      "learning_rate": 2.19291829132094e-06,
      "loss": 0.4661,
      "step": 266030
    },
    {
      "epoch": 108.9,
      "grad_norm": 2.1078288555145264,
      "learning_rate": 2.1927599909441884e-06,
      "loss": 0.4593,
      "step": 266040
    },
    {
      "epoch": 108.9,
      "grad_norm": 1.8510937690734863,
      "learning_rate": 2.1926016912590996e-06,
      "loss": 0.467,
      "step": 266050
    },
    {
      "epoch": 108.91,
      "grad_norm": 2.0456831455230713,
      "learning_rate": 2.1924433922663968e-06,
      "loss": 0.4691,
      "step": 266060
    },
    {
      "epoch": 108.91,
      "grad_norm": 1.429587483406067,
      "learning_rate": 2.192285093966806e-06,
      "loss": 0.487,
      "step": 266070
    },
    {
      "epoch": 108.92,
      "grad_norm": 1.6956336498260498,
      "learning_rate": 2.1921267963610532e-06,
      "loss": 0.472,
      "step": 266080
    },
    {
      "epoch": 108.92,
      "grad_norm": 2.0748164653778076,
      "learning_rate": 2.1919684994498623e-06,
      "loss": 0.4738,
      "step": 266090
    },
    {
      "epoch": 108.92,
      "grad_norm": 1.9632831811904907,
      "learning_rate": 2.191810203233959e-06,
      "loss": 0.4956,
      "step": 266100
    },
    {
      "epoch": 108.93,
      "grad_norm": 1.8136054277420044,
      "learning_rate": 2.1916519077140685e-06,
      "loss": 0.4617,
      "step": 266110
    },
    {
      "epoch": 108.93,
      "grad_norm": 1.7657554149627686,
      "learning_rate": 2.191493612890916e-06,
      "loss": 0.4533,
      "step": 266120
    },
    {
      "epoch": 108.94,
      "grad_norm": 2.088076114654541,
      "learning_rate": 2.191335318765226e-06,
      "loss": 0.4919,
      "step": 266130
    },
    {
      "epoch": 108.94,
      "grad_norm": 1.9746203422546387,
      "learning_rate": 2.1911770253377238e-06,
      "loss": 0.476,
      "step": 266140
    },
    {
      "epoch": 108.94,
      "grad_norm": 1.7546988725662231,
      "learning_rate": 2.191018732609135e-06,
      "loss": 0.4762,
      "step": 266150
    },
    {
      "epoch": 108.95,
      "grad_norm": 1.9444262981414795,
      "learning_rate": 2.1908604405801847e-06,
      "loss": 0.4845,
      "step": 266160
    },
    {
      "epoch": 108.95,
      "grad_norm": 1.9935053586959839,
      "learning_rate": 2.1907021492515977e-06,
      "loss": 0.472,
      "step": 266170
    },
    {
      "epoch": 108.96,
      "grad_norm": 1.6014357805252075,
      "learning_rate": 2.1905438586240995e-06,
      "loss": 0.4627,
      "step": 266180
    },
    {
      "epoch": 108.96,
      "grad_norm": 2.1093766689300537,
      "learning_rate": 2.190385568698415e-06,
      "loss": 0.458,
      "step": 266190
    },
    {
      "epoch": 108.96,
      "grad_norm": 1.8905785083770752,
      "learning_rate": 2.190227279475269e-06,
      "loss": 0.4826,
      "step": 266200
    },
    {
      "epoch": 108.97,
      "grad_norm": 1.932517170906067,
      "learning_rate": 2.1900689909553866e-06,
      "loss": 0.4807,
      "step": 266210
    },
    {
      "epoch": 108.97,
      "grad_norm": 2.196160078048706,
      "learning_rate": 2.189910703139493e-06,
      "loss": 0.458,
      "step": 266220
    },
    {
      "epoch": 108.98,
      "grad_norm": 1.7171883583068848,
      "learning_rate": 2.1897524160283137e-06,
      "loss": 0.4775,
      "step": 266230
    },
    {
      "epoch": 108.98,
      "grad_norm": 2.1825151443481445,
      "learning_rate": 2.1895941296225734e-06,
      "loss": 0.5009,
      "step": 266240
    },
    {
      "epoch": 108.98,
      "grad_norm": 2.178112030029297,
      "learning_rate": 2.1894358439229972e-06,
      "loss": 0.4736,
      "step": 266250
    },
    {
      "epoch": 108.99,
      "grad_norm": 1.8156465291976929,
      "learning_rate": 2.1892775589303107e-06,
      "loss": 0.4671,
      "step": 266260
    },
    {
      "epoch": 108.99,
      "grad_norm": 1.8610668182373047,
      "learning_rate": 2.1891192746452385e-06,
      "loss": 0.4697,
      "step": 266270
    },
    {
      "epoch": 109.0,
      "grad_norm": 1.9566452503204346,
      "learning_rate": 2.1889609910685052e-06,
      "loss": 0.4779,
      "step": 266280
    },
    {
      "epoch": 109.0,
      "eval_loss": 0.47102105617523193,
      "eval_runtime": 56.2864,
      "eval_samples_per_second": 61.276,
      "eval_steps_per_second": 7.675,
      "step": 266287
    },
    {
      "epoch": 109.0,
      "grad_norm": 1.4798086881637573,
      "learning_rate": 2.1888027082008364e-06,
      "loss": 0.4545,
      "step": 266290
    },
    {
      "epoch": 109.01,
      "grad_norm": 1.713863730430603,
      "learning_rate": 2.1886444260429574e-06,
      "loss": 0.4744,
      "step": 266300
    },
    {
      "epoch": 109.01,
      "grad_norm": 2.0415313243865967,
      "learning_rate": 2.188486144595593e-06,
      "loss": 0.4701,
      "step": 266310
    },
    {
      "epoch": 109.01,
      "grad_norm": 1.7433198690414429,
      "learning_rate": 2.188327863859468e-06,
      "loss": 0.46,
      "step": 266320
    },
    {
      "epoch": 109.02,
      "grad_norm": 1.854781150817871,
      "learning_rate": 2.1881695838353074e-06,
      "loss": 0.4323,
      "step": 266330
    },
    {
      "epoch": 109.02,
      "grad_norm": 1.5450466871261597,
      "learning_rate": 2.188011304523837e-06,
      "loss": 0.4605,
      "step": 266340
    },
    {
      "epoch": 109.03,
      "grad_norm": 2.028911590576172,
      "learning_rate": 2.1878530259257816e-06,
      "loss": 0.4785,
      "step": 266350
    },
    {
      "epoch": 109.03,
      "grad_norm": 1.97580087184906,
      "learning_rate": 2.187694748041866e-06,
      "loss": 0.4615,
      "step": 266360
    },
    {
      "epoch": 109.03,
      "grad_norm": 2.1273064613342285,
      "learning_rate": 2.1875364708728155e-06,
      "loss": 0.4656,
      "step": 266370
    },
    {
      "epoch": 109.04,
      "grad_norm": 1.7985130548477173,
      "learning_rate": 2.1873781944193545e-06,
      "loss": 0.4725,
      "step": 266380
    },
    {
      "epoch": 109.04,
      "grad_norm": 2.032179117202759,
      "learning_rate": 2.187219918682209e-06,
      "loss": 0.4727,
      "step": 266390
    },
    {
      "epoch": 109.05,
      "grad_norm": 1.5646815299987793,
      "learning_rate": 2.1870616436621033e-06,
      "loss": 0.47,
      "step": 266400
    },
    {
      "epoch": 109.05,
      "grad_norm": 2.5476632118225098,
      "learning_rate": 2.1869033693597633e-06,
      "loss": 0.4891,
      "step": 266410
    },
    {
      "epoch": 109.05,
      "grad_norm": 1.8906018733978271,
      "learning_rate": 2.186745095775913e-06,
      "loss": 0.4645,
      "step": 266420
    },
    {
      "epoch": 109.06,
      "grad_norm": 1.8974404335021973,
      "learning_rate": 2.1865868229112777e-06,
      "loss": 0.4609,
      "step": 266430
    },
    {
      "epoch": 109.06,
      "grad_norm": 2.0167040824890137,
      "learning_rate": 2.186428550766583e-06,
      "loss": 0.4768,
      "step": 266440
    },
    {
      "epoch": 109.07,
      "grad_norm": 1.8402987718582153,
      "learning_rate": 2.1862702793425533e-06,
      "loss": 0.4673,
      "step": 266450
    },
    {
      "epoch": 109.07,
      "grad_norm": 1.6363842487335205,
      "learning_rate": 2.186112008639913e-06,
      "loss": 0.4647,
      "step": 266460
    },
    {
      "epoch": 109.07,
      "grad_norm": 1.8216806650161743,
      "learning_rate": 2.185953738659389e-06,
      "loss": 0.4588,
      "step": 266470
    },
    {
      "epoch": 109.08,
      "grad_norm": 2.1066107749938965,
      "learning_rate": 2.185795469401705e-06,
      "loss": 0.4684,
      "step": 266480
    },
    {
      "epoch": 109.08,
      "grad_norm": 1.8166600465774536,
      "learning_rate": 2.185637200867586e-06,
      "loss": 0.4916,
      "step": 266490
    },
    {
      "epoch": 109.09,
      "grad_norm": 1.5663203001022339,
      "learning_rate": 2.1854789330577577e-06,
      "loss": 0.4972,
      "step": 266500
    },
    {
      "epoch": 109.09,
      "grad_norm": 1.6763877868652344,
      "learning_rate": 2.1853206659729444e-06,
      "loss": 0.4835,
      "step": 266510
    },
    {
      "epoch": 109.1,
      "grad_norm": 1.630264163017273,
      "learning_rate": 2.18516239961387e-06,
      "loss": 0.4565,
      "step": 266520
    },
    {
      "epoch": 109.1,
      "grad_norm": 1.8056799173355103,
      "learning_rate": 2.185004133981262e-06,
      "loss": 0.475,
      "step": 266530
    },
    {
      "epoch": 109.1,
      "grad_norm": 2.5596866607666016,
      "learning_rate": 2.1848458690758447e-06,
      "loss": 0.4751,
      "step": 266540
    },
    {
      "epoch": 109.11,
      "grad_norm": 1.7720649242401123,
      "learning_rate": 2.1846876048983423e-06,
      "loss": 0.4588,
      "step": 266550
    },
    {
      "epoch": 109.11,
      "grad_norm": 1.391430139541626,
      "learning_rate": 2.1845293414494798e-06,
      "loss": 0.4714,
      "step": 266560
    },
    {
      "epoch": 109.12,
      "grad_norm": 1.8498796224594116,
      "learning_rate": 2.184371078729983e-06,
      "loss": 0.4877,
      "step": 266570
    },
    {
      "epoch": 109.12,
      "grad_norm": 2.0222723484039307,
      "learning_rate": 2.184212816740576e-06,
      "loss": 0.4831,
      "step": 266580
    },
    {
      "epoch": 109.12,
      "grad_norm": 1.5315877199172974,
      "learning_rate": 2.184054555481984e-06,
      "loss": 0.4764,
      "step": 266590
    },
    {
      "epoch": 109.13,
      "grad_norm": 2.1008825302124023,
      "learning_rate": 2.183896294954932e-06,
      "loss": 0.4922,
      "step": 266600
    },
    {
      "epoch": 109.13,
      "grad_norm": 1.6291676759719849,
      "learning_rate": 2.1837380351601455e-06,
      "loss": 0.4819,
      "step": 266610
    },
    {
      "epoch": 109.14,
      "grad_norm": 1.6745405197143555,
      "learning_rate": 2.1835797760983486e-06,
      "loss": 0.4697,
      "step": 266620
    },
    {
      "epoch": 109.14,
      "grad_norm": 1.9371163845062256,
      "learning_rate": 2.1834215177702678e-06,
      "loss": 0.4557,
      "step": 266630
    },
    {
      "epoch": 109.14,
      "grad_norm": 1.7113540172576904,
      "learning_rate": 2.183263260176626e-06,
      "loss": 0.4505,
      "step": 266640
    },
    {
      "epoch": 109.15,
      "grad_norm": 1.887804627418518,
      "learning_rate": 2.1831050033181493e-06,
      "loss": 0.4677,
      "step": 266650
    },
    {
      "epoch": 109.15,
      "grad_norm": 2.030399799346924,
      "learning_rate": 2.1829467471955623e-06,
      "loss": 0.4773,
      "step": 266660
    },
    {
      "epoch": 109.16,
      "grad_norm": 1.929580569267273,
      "learning_rate": 2.1827884918095902e-06,
      "loss": 0.4457,
      "step": 266670
    },
    {
      "epoch": 109.16,
      "grad_norm": 1.5201903581619263,
      "learning_rate": 2.182630237160958e-06,
      "loss": 0.4616,
      "step": 266680
    },
    {
      "epoch": 109.16,
      "grad_norm": 1.6468974351882935,
      "learning_rate": 2.18247198325039e-06,
      "loss": 0.4767,
      "step": 266690
    },
    {
      "epoch": 109.17,
      "grad_norm": 1.909162163734436,
      "learning_rate": 2.1823137300786113e-06,
      "loss": 0.4679,
      "step": 266700
    },
    {
      "epoch": 109.17,
      "grad_norm": 1.8394354581832886,
      "learning_rate": 2.1821554776463478e-06,
      "loss": 0.4552,
      "step": 266710
    },
    {
      "epoch": 109.18,
      "grad_norm": 1.7728477716445923,
      "learning_rate": 2.1819972259543237e-06,
      "loss": 0.4763,
      "step": 266720
    },
    {
      "epoch": 109.18,
      "grad_norm": 2.009129524230957,
      "learning_rate": 2.1818389750032637e-06,
      "loss": 0.4526,
      "step": 266730
    },
    {
      "epoch": 109.19,
      "grad_norm": 2.0505218505859375,
      "learning_rate": 2.1816807247938932e-06,
      "loss": 0.4589,
      "step": 266740
    },
    {
      "epoch": 109.19,
      "grad_norm": 2.0942976474761963,
      "learning_rate": 2.1815224753269375e-06,
      "loss": 0.461,
      "step": 266750
    },
    {
      "epoch": 109.19,
      "grad_norm": 2.0159828662872314,
      "learning_rate": 2.1813642266031205e-06,
      "loss": 0.4708,
      "step": 266760
    },
    {
      "epoch": 109.2,
      "grad_norm": 2.4593453407287598,
      "learning_rate": 2.1812059786231683e-06,
      "loss": 0.4676,
      "step": 266770
    },
    {
      "epoch": 109.2,
      "grad_norm": 1.889042615890503,
      "learning_rate": 2.1810477313878043e-06,
      "loss": 0.4821,
      "step": 266780
    },
    {
      "epoch": 109.21,
      "grad_norm": 1.7841373682022095,
      "learning_rate": 2.1808894848977543e-06,
      "loss": 0.4774,
      "step": 266790
    },
    {
      "epoch": 109.21,
      "grad_norm": 1.6982581615447998,
      "learning_rate": 2.1807312391537433e-06,
      "loss": 0.4715,
      "step": 266800
    },
    {
      "epoch": 109.21,
      "grad_norm": 2.1397147178649902,
      "learning_rate": 2.1805729941564957e-06,
      "loss": 0.4844,
      "step": 266810
    },
    {
      "epoch": 109.22,
      "grad_norm": 3.0105011463165283,
      "learning_rate": 2.1804147499067364e-06,
      "loss": 0.4637,
      "step": 266820
    },
    {
      "epoch": 109.22,
      "grad_norm": 2.6357240676879883,
      "learning_rate": 2.180256506405191e-06,
      "loss": 0.4599,
      "step": 266830
    },
    {
      "epoch": 109.23,
      "grad_norm": 1.945704460144043,
      "learning_rate": 2.180098263652584e-06,
      "loss": 0.4582,
      "step": 266840
    },
    {
      "epoch": 109.23,
      "grad_norm": 1.6802927255630493,
      "learning_rate": 2.179940021649641e-06,
      "loss": 0.4515,
      "step": 266850
    },
    {
      "epoch": 109.23,
      "grad_norm": 2.0338470935821533,
      "learning_rate": 2.1797817803970855e-06,
      "loss": 0.4711,
      "step": 266860
    },
    {
      "epoch": 109.24,
      "grad_norm": 1.8076106309890747,
      "learning_rate": 2.1796235398956427e-06,
      "loss": 0.4918,
      "step": 266870
    },
    {
      "epoch": 109.24,
      "grad_norm": 1.950416922569275,
      "learning_rate": 2.1794653001460378e-06,
      "loss": 0.4497,
      "step": 266880
    },
    {
      "epoch": 109.25,
      "grad_norm": 1.9956380128860474,
      "learning_rate": 2.179307061148995e-06,
      "loss": 0.4673,
      "step": 266890
    },
    {
      "epoch": 109.25,
      "grad_norm": 2.0236704349517822,
      "learning_rate": 2.179148822905241e-06,
      "loss": 0.4554,
      "step": 266900
    },
    {
      "epoch": 109.26,
      "grad_norm": 2.1728270053863525,
      "learning_rate": 2.178990585415499e-06,
      "loss": 0.4592,
      "step": 266910
    },
    {
      "epoch": 109.26,
      "grad_norm": 1.7565404176712036,
      "learning_rate": 2.178832348680495e-06,
      "loss": 0.4799,
      "step": 266920
    },
    {
      "epoch": 109.26,
      "grad_norm": 2.5383543968200684,
      "learning_rate": 2.178674112700953e-06,
      "loss": 0.4751,
      "step": 266930
    },
    {
      "epoch": 109.27,
      "grad_norm": 2.1117382049560547,
      "learning_rate": 2.1785158774775973e-06,
      "loss": 0.4597,
      "step": 266940
    },
    {
      "epoch": 109.27,
      "grad_norm": 1.9587211608886719,
      "learning_rate": 2.178357643011154e-06,
      "loss": 0.4682,
      "step": 266950
    },
    {
      "epoch": 109.28,
      "grad_norm": 2.2170419692993164,
      "learning_rate": 2.1781994093023476e-06,
      "loss": 0.4648,
      "step": 266960
    },
    {
      "epoch": 109.28,
      "grad_norm": 1.8613463640213013,
      "learning_rate": 2.178041176351903e-06,
      "loss": 0.4885,
      "step": 266970
    },
    {
      "epoch": 109.28,
      "grad_norm": 1.723632574081421,
      "learning_rate": 2.1778829441605444e-06,
      "loss": 0.4827,
      "step": 266980
    },
    {
      "epoch": 109.29,
      "grad_norm": 1.9723536968231201,
      "learning_rate": 2.177724712728998e-06,
      "loss": 0.4458,
      "step": 266990
    },
    {
      "epoch": 109.29,
      "grad_norm": 1.9090156555175781,
      "learning_rate": 2.177566482057987e-06,
      "loss": 0.4767,
      "step": 267000
    },
    {
      "epoch": 109.3,
      "grad_norm": 1.8508914709091187,
      "learning_rate": 2.177408252148237e-06,
      "loss": 0.4736,
      "step": 267010
    },
    {
      "epoch": 109.3,
      "grad_norm": 1.5850118398666382,
      "learning_rate": 2.1772500230004726e-06,
      "loss": 0.4722,
      "step": 267020
    },
    {
      "epoch": 109.3,
      "grad_norm": 1.620792269706726,
      "learning_rate": 2.177091794615419e-06,
      "loss": 0.4643,
      "step": 267030
    },
    {
      "epoch": 109.31,
      "grad_norm": 2.0676379203796387,
      "learning_rate": 2.176933566993801e-06,
      "loss": 0.4507,
      "step": 267040
    },
    {
      "epoch": 109.31,
      "grad_norm": 1.7367497682571411,
      "learning_rate": 2.176775340136343e-06,
      "loss": 0.4752,
      "step": 267050
    },
    {
      "epoch": 109.32,
      "grad_norm": 1.9741686582565308,
      "learning_rate": 2.17661711404377e-06,
      "loss": 0.4973,
      "step": 267060
    },
    {
      "epoch": 109.32,
      "grad_norm": 2.7706825733184814,
      "learning_rate": 2.1764588887168073e-06,
      "loss": 0.477,
      "step": 267070
    },
    {
      "epoch": 109.32,
      "grad_norm": 1.9811499118804932,
      "learning_rate": 2.176300664156178e-06,
      "loss": 0.4536,
      "step": 267080
    },
    {
      "epoch": 109.33,
      "grad_norm": 1.7436107397079468,
      "learning_rate": 2.1761424403626097e-06,
      "loss": 0.4886,
      "step": 267090
    },
    {
      "epoch": 109.33,
      "grad_norm": 1.9375020265579224,
      "learning_rate": 2.175984217336825e-06,
      "loss": 0.4531,
      "step": 267100
    },
    {
      "epoch": 109.34,
      "grad_norm": 1.952497124671936,
      "learning_rate": 2.175825995079549e-06,
      "loss": 0.452,
      "step": 267110
    },
    {
      "epoch": 109.34,
      "grad_norm": 1.6683554649353027,
      "learning_rate": 2.1756677735915077e-06,
      "loss": 0.4591,
      "step": 267120
    },
    {
      "epoch": 109.35,
      "grad_norm": 1.8948334455490112,
      "learning_rate": 2.1755095528734253e-06,
      "loss": 0.4902,
      "step": 267130
    },
    {
      "epoch": 109.35,
      "grad_norm": 2.6068027019500732,
      "learning_rate": 2.1753513329260256e-06,
      "loss": 0.4698,
      "step": 267140
    },
    {
      "epoch": 109.35,
      "grad_norm": 1.7678396701812744,
      "learning_rate": 2.1751931137500344e-06,
      "loss": 0.4631,
      "step": 267150
    },
    {
      "epoch": 109.36,
      "grad_norm": 1.867864966392517,
      "learning_rate": 2.175034895346176e-06,
      "loss": 0.4721,
      "step": 267160
    },
    {
      "epoch": 109.36,
      "grad_norm": 1.8288253545761108,
      "learning_rate": 2.1748766777151755e-06,
      "loss": 0.4646,
      "step": 267170
    },
    {
      "epoch": 109.37,
      "grad_norm": 1.7747406959533691,
      "learning_rate": 2.1747184608577574e-06,
      "loss": 0.4746,
      "step": 267180
    },
    {
      "epoch": 109.37,
      "grad_norm": 2.001586437225342,
      "learning_rate": 2.1745602447746465e-06,
      "loss": 0.4807,
      "step": 267190
    },
    {
      "epoch": 109.37,
      "grad_norm": 1.7233916521072388,
      "learning_rate": 2.1744020294665677e-06,
      "loss": 0.4644,
      "step": 267200
    },
    {
      "epoch": 109.38,
      "grad_norm": 2.1268160343170166,
      "learning_rate": 2.174243814934246e-06,
      "loss": 0.4807,
      "step": 267210
    },
    {
      "epoch": 109.38,
      "grad_norm": 1.9930719137191772,
      "learning_rate": 2.1740856011784057e-06,
      "loss": 0.4724,
      "step": 267220
    },
    {
      "epoch": 109.39,
      "grad_norm": 1.7535746097564697,
      "learning_rate": 2.173927388199772e-06,
      "loss": 0.4532,
      "step": 267230
    },
    {
      "epoch": 109.39,
      "grad_norm": 1.8322440385818481,
      "learning_rate": 2.1737691759990694e-06,
      "loss": 0.4834,
      "step": 267240
    },
    {
      "epoch": 109.39,
      "grad_norm": 1.532120704650879,
      "learning_rate": 2.173610964577022e-06,
      "loss": 0.475,
      "step": 267250
    },
    {
      "epoch": 109.4,
      "grad_norm": 1.855931043624878,
      "learning_rate": 2.1734527539343557e-06,
      "loss": 0.4664,
      "step": 267260
    },
    {
      "epoch": 109.4,
      "grad_norm": 2.4842896461486816,
      "learning_rate": 2.1732945440717942e-06,
      "loss": 0.4749,
      "step": 267270
    },
    {
      "epoch": 109.41,
      "grad_norm": 2.589179754257202,
      "learning_rate": 2.173136334990063e-06,
      "loss": 0.464,
      "step": 267280
    },
    {
      "epoch": 109.41,
      "grad_norm": 1.6368072032928467,
      "learning_rate": 2.1729781266898866e-06,
      "loss": 0.4676,
      "step": 267290
    },
    {
      "epoch": 109.41,
      "grad_norm": 1.7114993333816528,
      "learning_rate": 2.1728199191719897e-06,
      "loss": 0.48,
      "step": 267300
    },
    {
      "epoch": 109.42,
      "grad_norm": 2.180769443511963,
      "learning_rate": 2.172661712437097e-06,
      "loss": 0.4782,
      "step": 267310
    },
    {
      "epoch": 109.42,
      "grad_norm": 1.7538905143737793,
      "learning_rate": 2.1725035064859332e-06,
      "loss": 0.4721,
      "step": 267320
    },
    {
      "epoch": 109.43,
      "grad_norm": 2.2924091815948486,
      "learning_rate": 2.1723453013192233e-06,
      "loss": 0.4669,
      "step": 267330
    },
    {
      "epoch": 109.43,
      "grad_norm": 1.8807158470153809,
      "learning_rate": 2.1721870969376916e-06,
      "loss": 0.4815,
      "step": 267340
    },
    {
      "epoch": 109.44,
      "grad_norm": 2.004734992980957,
      "learning_rate": 2.172028893342064e-06,
      "loss": 0.4679,
      "step": 267350
    },
    {
      "epoch": 109.44,
      "grad_norm": 1.9745877981185913,
      "learning_rate": 2.171870690533063e-06,
      "loss": 0.4715,
      "step": 267360
    },
    {
      "epoch": 109.44,
      "grad_norm": 2.0562098026275635,
      "learning_rate": 2.1717124885114147e-06,
      "loss": 0.4807,
      "step": 267370
    },
    {
      "epoch": 109.45,
      "grad_norm": 1.8757810592651367,
      "learning_rate": 2.1715542872778437e-06,
      "loss": 0.4627,
      "step": 267380
    },
    {
      "epoch": 109.45,
      "grad_norm": 2.16092848777771,
      "learning_rate": 2.1713960868330748e-06,
      "loss": 0.4634,
      "step": 267390
    },
    {
      "epoch": 109.46,
      "grad_norm": 1.7489951848983765,
      "learning_rate": 2.171237887177832e-06,
      "loss": 0.4739,
      "step": 267400
    },
    {
      "epoch": 109.46,
      "grad_norm": 1.9371250867843628,
      "learning_rate": 2.1710796883128405e-06,
      "loss": 0.462,
      "step": 267410
    },
    {
      "epoch": 109.46,
      "grad_norm": 1.8542860746383667,
      "learning_rate": 2.1709214902388257e-06,
      "loss": 0.4787,
      "step": 267420
    },
    {
      "epoch": 109.47,
      "grad_norm": 1.7000659704208374,
      "learning_rate": 2.1707632929565105e-06,
      "loss": 0.4607,
      "step": 267430
    },
    {
      "epoch": 109.47,
      "grad_norm": 2.4441640377044678,
      "learning_rate": 2.170605096466622e-06,
      "loss": 0.4726,
      "step": 267440
    },
    {
      "epoch": 109.48,
      "grad_norm": 2.186086416244507,
      "learning_rate": 2.1704469007698825e-06,
      "loss": 0.4507,
      "step": 267450
    },
    {
      "epoch": 109.48,
      "grad_norm": 1.956546664237976,
      "learning_rate": 2.1702887058670168e-06,
      "loss": 0.4859,
      "step": 267460
    },
    {
      "epoch": 109.48,
      "grad_norm": 1.8495076894760132,
      "learning_rate": 2.1701305117587514e-06,
      "loss": 0.4667,
      "step": 267470
    },
    {
      "epoch": 109.49,
      "grad_norm": 2.2206759452819824,
      "learning_rate": 2.16997231844581e-06,
      "loss": 0.4927,
      "step": 267480
    },
    {
      "epoch": 109.49,
      "grad_norm": 2.2834911346435547,
      "learning_rate": 2.1698141259289176e-06,
      "loss": 0.4711,
      "step": 267490
    },
    {
      "epoch": 109.5,
      "grad_norm": 2.175842046737671,
      "learning_rate": 2.169655934208798e-06,
      "loss": 0.4607,
      "step": 267500
    },
    {
      "epoch": 109.5,
      "grad_norm": 2.0168402194976807,
      "learning_rate": 2.1694977432861765e-06,
      "loss": 0.4719,
      "step": 267510
    },
    {
      "epoch": 109.5,
      "grad_norm": 1.8196696043014526,
      "learning_rate": 2.1693395531617777e-06,
      "loss": 0.4718,
      "step": 267520
    },
    {
      "epoch": 109.51,
      "grad_norm": 1.6650179624557495,
      "learning_rate": 2.169181363836326e-06,
      "loss": 0.4884,
      "step": 267530
    },
    {
      "epoch": 109.51,
      "grad_norm": 2.851696491241455,
      "learning_rate": 2.1690231753105462e-06,
      "loss": 0.4814,
      "step": 267540
    },
    {
      "epoch": 109.52,
      "grad_norm": 2.330761671066284,
      "learning_rate": 2.1688649875851632e-06,
      "loss": 0.4793,
      "step": 267550
    },
    {
      "epoch": 109.52,
      "grad_norm": 1.6697888374328613,
      "learning_rate": 2.168706800660901e-06,
      "loss": 0.4768,
      "step": 267560
    },
    {
      "epoch": 109.53,
      "grad_norm": 1.7654094696044922,
      "learning_rate": 2.168548614538485e-06,
      "loss": 0.4632,
      "step": 267570
    },
    {
      "epoch": 109.53,
      "grad_norm": 2.122537136077881,
      "learning_rate": 2.1683904292186393e-06,
      "loss": 0.4603,
      "step": 267580
    },
    {
      "epoch": 109.53,
      "grad_norm": 2.1940512657165527,
      "learning_rate": 2.168232244702088e-06,
      "loss": 0.4836,
      "step": 267590
    },
    {
      "epoch": 109.54,
      "grad_norm": 1.9395112991333008,
      "learning_rate": 2.1680740609895572e-06,
      "loss": 0.4561,
      "step": 267600
    },
    {
      "epoch": 109.54,
      "grad_norm": 1.8712348937988281,
      "learning_rate": 2.1679158780817703e-06,
      "loss": 0.4553,
      "step": 267610
    },
    {
      "epoch": 109.55,
      "grad_norm": 1.6436071395874023,
      "learning_rate": 2.1677576959794523e-06,
      "loss": 0.4684,
      "step": 267620
    },
    {
      "epoch": 109.55,
      "grad_norm": 2.1923530101776123,
      "learning_rate": 2.1675995146833277e-06,
      "loss": 0.4722,
      "step": 267630
    },
    {
      "epoch": 109.55,
      "grad_norm": 1.8900333642959595,
      "learning_rate": 2.1674413341941213e-06,
      "loss": 0.4668,
      "step": 267640
    },
    {
      "epoch": 109.56,
      "grad_norm": 1.9083319902420044,
      "learning_rate": 2.1672831545125567e-06,
      "loss": 0.468,
      "step": 267650
    },
    {
      "epoch": 109.56,
      "grad_norm": 2.06860613822937,
      "learning_rate": 2.1671249756393604e-06,
      "loss": 0.4907,
      "step": 267660
    },
    {
      "epoch": 109.57,
      "grad_norm": 1.9050534963607788,
      "learning_rate": 2.1669667975752556e-06,
      "loss": 0.4561,
      "step": 267670
    },
    {
      "epoch": 109.57,
      "grad_norm": 1.8781940937042236,
      "learning_rate": 2.1668086203209676e-06,
      "loss": 0.485,
      "step": 267680
    },
    {
      "epoch": 109.57,
      "grad_norm": 1.9476505517959595,
      "learning_rate": 2.1666504438772206e-06,
      "loss": 0.4591,
      "step": 267690
    },
    {
      "epoch": 109.58,
      "grad_norm": 2.1434483528137207,
      "learning_rate": 2.166492268244739e-06,
      "loss": 0.4761,
      "step": 267700
    },
    {
      "epoch": 109.58,
      "grad_norm": 1.6295835971832275,
      "learning_rate": 2.166334093424248e-06,
      "loss": 0.4746,
      "step": 267710
    },
    {
      "epoch": 109.59,
      "grad_norm": 1.9682987928390503,
      "learning_rate": 2.166175919416472e-06,
      "loss": 0.4735,
      "step": 267720
    },
    {
      "epoch": 109.59,
      "grad_norm": 2.410006046295166,
      "learning_rate": 2.1660177462221347e-06,
      "loss": 0.4727,
      "step": 267730
    },
    {
      "epoch": 109.59,
      "grad_norm": 2.0755136013031006,
      "learning_rate": 2.1658595738419614e-06,
      "loss": 0.4728,
      "step": 267740
    },
    {
      "epoch": 109.6,
      "grad_norm": 1.9798260927200317,
      "learning_rate": 2.1657014022766766e-06,
      "loss": 0.4733,
      "step": 267750
    },
    {
      "epoch": 109.6,
      "grad_norm": 2.094345808029175,
      "learning_rate": 2.1655432315270053e-06,
      "loss": 0.4813,
      "step": 267760
    },
    {
      "epoch": 109.61,
      "grad_norm": 2.1593093872070312,
      "learning_rate": 2.1653850615936714e-06,
      "loss": 0.4846,
      "step": 267770
    },
    {
      "epoch": 109.61,
      "grad_norm": 1.7555042505264282,
      "learning_rate": 2.1652268924773994e-06,
      "loss": 0.4577,
      "step": 267780
    },
    {
      "epoch": 109.62,
      "grad_norm": 1.8486725091934204,
      "learning_rate": 2.1650687241789144e-06,
      "loss": 0.4806,
      "step": 267790
    },
    {
      "epoch": 109.62,
      "grad_norm": 1.8879162073135376,
      "learning_rate": 2.1649105566989407e-06,
      "loss": 0.4595,
      "step": 267800
    },
    {
      "epoch": 109.62,
      "grad_norm": 1.94065260887146,
      "learning_rate": 2.1647523900382028e-06,
      "loss": 0.4688,
      "step": 267810
    },
    {
      "epoch": 109.63,
      "grad_norm": 2.2836506366729736,
      "learning_rate": 2.1645942241974254e-06,
      "loss": 0.4477,
      "step": 267820
    },
    {
      "epoch": 109.63,
      "grad_norm": 1.8620638847351074,
      "learning_rate": 2.1644360591773323e-06,
      "loss": 0.4677,
      "step": 267830
    },
    {
      "epoch": 109.64,
      "grad_norm": 2.336178779602051,
      "learning_rate": 2.1642778949786477e-06,
      "loss": 0.4776,
      "step": 267840
    },
    {
      "epoch": 109.64,
      "grad_norm": 2.408172130584717,
      "learning_rate": 2.164119731602099e-06,
      "loss": 0.487,
      "step": 267850
    },
    {
      "epoch": 109.64,
      "grad_norm": 2.1180062294006348,
      "learning_rate": 2.1639615690484077e-06,
      "loss": 0.4493,
      "step": 267860
    },
    {
      "epoch": 109.65,
      "grad_norm": 1.8903712034225464,
      "learning_rate": 2.1638034073182994e-06,
      "loss": 0.484,
      "step": 267870
    },
    {
      "epoch": 109.65,
      "grad_norm": 2.1240170001983643,
      "learning_rate": 2.1636452464124986e-06,
      "loss": 0.4645,
      "step": 267880
    },
    {
      "epoch": 109.66,
      "grad_norm": 2.2845895290374756,
      "learning_rate": 2.16348708633173e-06,
      "loss": 0.4722,
      "step": 267890
    },
    {
      "epoch": 109.66,
      "grad_norm": 1.8798177242279053,
      "learning_rate": 2.1633289270767176e-06,
      "loss": 0.4635,
      "step": 267900
    },
    {
      "epoch": 109.66,
      "grad_norm": 1.8042702674865723,
      "learning_rate": 2.1631707686481866e-06,
      "loss": 0.4784,
      "step": 267910
    },
    {
      "epoch": 109.67,
      "grad_norm": 2.0492753982543945,
      "learning_rate": 2.1630126110468604e-06,
      "loss": 0.4841,
      "step": 267920
    },
    {
      "epoch": 109.67,
      "grad_norm": 1.5644378662109375,
      "learning_rate": 2.1628544542734653e-06,
      "loss": 0.461,
      "step": 267930
    },
    {
      "epoch": 109.68,
      "grad_norm": 1.8821040391921997,
      "learning_rate": 2.162696298328724e-06,
      "loss": 0.4766,
      "step": 267940
    },
    {
      "epoch": 109.68,
      "grad_norm": 2.372885227203369,
      "learning_rate": 2.1625381432133617e-06,
      "loss": 0.4841,
      "step": 267950
    },
    {
      "epoch": 109.68,
      "grad_norm": 1.8574045896530151,
      "learning_rate": 2.162379988928103e-06,
      "loss": 0.4729,
      "step": 267960
    },
    {
      "epoch": 109.69,
      "grad_norm": 1.9839059114456177,
      "learning_rate": 2.162221835473672e-06,
      "loss": 0.4535,
      "step": 267970
    },
    {
      "epoch": 109.69,
      "grad_norm": 3.423233985900879,
      "learning_rate": 2.1620636828507935e-06,
      "loss": 0.4673,
      "step": 267980
    },
    {
      "epoch": 109.7,
      "grad_norm": 2.1050734519958496,
      "learning_rate": 2.161905531060192e-06,
      "loss": 0.472,
      "step": 267990
    },
    {
      "epoch": 109.7,
      "grad_norm": 2.1846203804016113,
      "learning_rate": 2.1617473801025917e-06,
      "loss": 0.4745,
      "step": 268000
    },
    {
      "epoch": 109.71,
      "grad_norm": 1.9452074766159058,
      "learning_rate": 2.1615892299787173e-06,
      "loss": 0.4863,
      "step": 268010
    },
    {
      "epoch": 109.71,
      "grad_norm": 1.9610458612442017,
      "learning_rate": 2.1614310806892932e-06,
      "loss": 0.4563,
      "step": 268020
    },
    {
      "epoch": 109.71,
      "grad_norm": 1.7892223596572876,
      "learning_rate": 2.161272932235043e-06,
      "loss": 0.4645,
      "step": 268030
    },
    {
      "epoch": 109.72,
      "grad_norm": 1.765851378440857,
      "learning_rate": 2.1611147846166927e-06,
      "loss": 0.4596,
      "step": 268040
    },
    {
      "epoch": 109.72,
      "grad_norm": 1.695904016494751,
      "learning_rate": 2.160956637834966e-06,
      "loss": 0.4834,
      "step": 268050
    },
    {
      "epoch": 109.73,
      "grad_norm": 1.7950961589813232,
      "learning_rate": 2.1607984918905877e-06,
      "loss": 0.4683,
      "step": 268060
    },
    {
      "epoch": 109.73,
      "grad_norm": 2.1690478324890137,
      "learning_rate": 2.160640346784282e-06,
      "loss": 0.4735,
      "step": 268070
    },
    {
      "epoch": 109.73,
      "grad_norm": 1.7204983234405518,
      "learning_rate": 2.160482202516773e-06,
      "loss": 0.4784,
      "step": 268080
    },
    {
      "epoch": 109.74,
      "grad_norm": 1.9896068572998047,
      "learning_rate": 2.160324059088785e-06,
      "loss": 0.4648,
      "step": 268090
    },
    {
      "epoch": 109.74,
      "grad_norm": 1.8067740201950073,
      "learning_rate": 2.1601659165010434e-06,
      "loss": 0.4675,
      "step": 268100
    },
    {
      "epoch": 109.75,
      "grad_norm": 2.0714879035949707,
      "learning_rate": 2.160007774754272e-06,
      "loss": 0.4864,
      "step": 268110
    },
    {
      "epoch": 109.75,
      "grad_norm": 2.04020619392395,
      "learning_rate": 2.159849633849195e-06,
      "loss": 0.4634,
      "step": 268120
    },
    {
      "epoch": 109.75,
      "grad_norm": 2.52724289894104,
      "learning_rate": 2.1596914937865372e-06,
      "loss": 0.4513,
      "step": 268130
    },
    {
      "epoch": 109.76,
      "grad_norm": 2.4944043159484863,
      "learning_rate": 2.159533354567023e-06,
      "loss": 0.4495,
      "step": 268140
    },
    {
      "epoch": 109.76,
      "grad_norm": 2.1685116291046143,
      "learning_rate": 2.1593752161913764e-06,
      "loss": 0.4796,
      "step": 268150
    },
    {
      "epoch": 109.77,
      "grad_norm": 2.5562310218811035,
      "learning_rate": 2.159217078660323e-06,
      "loss": 0.4745,
      "step": 268160
    },
    {
      "epoch": 109.77,
      "grad_norm": 1.602681040763855,
      "learning_rate": 2.1590589419745858e-06,
      "loss": 0.4792,
      "step": 268170
    },
    {
      "epoch": 109.77,
      "grad_norm": 1.6900547742843628,
      "learning_rate": 2.1589008061348895e-06,
      "loss": 0.4607,
      "step": 268180
    },
    {
      "epoch": 109.78,
      "grad_norm": 1.9580206871032715,
      "learning_rate": 2.158742671141959e-06,
      "loss": 0.4676,
      "step": 268190
    },
    {
      "epoch": 109.78,
      "grad_norm": 2.174712657928467,
      "learning_rate": 2.1585845369965184e-06,
      "loss": 0.4741,
      "step": 268200
    },
    {
      "epoch": 109.79,
      "grad_norm": 2.0467350482940674,
      "learning_rate": 2.158426403699291e-06,
      "loss": 0.4568,
      "step": 268210
    },
    {
      "epoch": 109.79,
      "grad_norm": 2.0414464473724365,
      "learning_rate": 2.1582682712510033e-06,
      "loss": 0.4638,
      "step": 268220
    },
    {
      "epoch": 109.8,
      "grad_norm": 1.927381157875061,
      "learning_rate": 2.1581101396523786e-06,
      "loss": 0.4738,
      "step": 268230
    },
    {
      "epoch": 109.8,
      "grad_norm": 2.1569738388061523,
      "learning_rate": 2.1579520089041415e-06,
      "loss": 0.454,
      "step": 268240
    },
    {
      "epoch": 109.8,
      "grad_norm": 2.0753822326660156,
      "learning_rate": 2.1577938790070162e-06,
      "loss": 0.4521,
      "step": 268250
    },
    {
      "epoch": 109.81,
      "grad_norm": 3.770246744155884,
      "learning_rate": 2.157635749961727e-06,
      "loss": 0.4615,
      "step": 268260
    },
    {
      "epoch": 109.81,
      "grad_norm": 1.8630913496017456,
      "learning_rate": 2.1574776217689985e-06,
      "loss": 0.4714,
      "step": 268270
    },
    {
      "epoch": 109.82,
      "grad_norm": 2.2679340839385986,
      "learning_rate": 2.157319494429555e-06,
      "loss": 0.4573,
      "step": 268280
    },
    {
      "epoch": 109.82,
      "grad_norm": 1.797813057899475,
      "learning_rate": 2.1571613679441207e-06,
      "loss": 0.469,
      "step": 268290
    },
    {
      "epoch": 109.82,
      "grad_norm": 2.4772276878356934,
      "learning_rate": 2.15700324231342e-06,
      "loss": 0.4747,
      "step": 268300
    },
    {
      "epoch": 109.83,
      "grad_norm": 2.140933036804199,
      "learning_rate": 2.156845117538177e-06,
      "loss": 0.4544,
      "step": 268310
    },
    {
      "epoch": 109.83,
      "grad_norm": 1.9179258346557617,
      "learning_rate": 2.1566869936191165e-06,
      "loss": 0.4668,
      "step": 268320
    },
    {
      "epoch": 109.84,
      "grad_norm": 1.6630439758300781,
      "learning_rate": 2.1565288705569626e-06,
      "loss": 0.4821,
      "step": 268330
    },
    {
      "epoch": 109.84,
      "grad_norm": 1.9200695753097534,
      "learning_rate": 2.15637074835244e-06,
      "loss": 0.4878,
      "step": 268340
    },
    {
      "epoch": 109.84,
      "grad_norm": 1.7581090927124023,
      "learning_rate": 2.1562126270062722e-06,
      "loss": 0.4839,
      "step": 268350
    },
    {
      "epoch": 109.85,
      "grad_norm": 2.014852285385132,
      "learning_rate": 2.1560545065191845e-06,
      "loss": 0.4741,
      "step": 268360
    },
    {
      "epoch": 109.85,
      "grad_norm": 2.32820200920105,
      "learning_rate": 2.155896386891901e-06,
      "loss": 0.4759,
      "step": 268370
    },
    {
      "epoch": 109.86,
      "grad_norm": 2.1599581241607666,
      "learning_rate": 2.1557382681251456e-06,
      "loss": 0.4887,
      "step": 268380
    },
    {
      "epoch": 109.86,
      "grad_norm": 2.321995258331299,
      "learning_rate": 2.1555801502196428e-06,
      "loss": 0.4684,
      "step": 268390
    },
    {
      "epoch": 109.86,
      "grad_norm": 2.127347230911255,
      "learning_rate": 2.1554220331761154e-06,
      "loss": 0.4612,
      "step": 268400
    },
    {
      "epoch": 109.87,
      "grad_norm": 1.8868132829666138,
      "learning_rate": 2.155263916995291e-06,
      "loss": 0.4609,
      "step": 268410
    },
    {
      "epoch": 109.87,
      "grad_norm": 2.062422752380371,
      "learning_rate": 2.1551058016778916e-06,
      "loss": 0.4695,
      "step": 268420
    },
    {
      "epoch": 109.88,
      "grad_norm": 2.063833475112915,
      "learning_rate": 2.154947687224643e-06,
      "loss": 0.4785,
      "step": 268430
    },
    {
      "epoch": 109.88,
      "grad_norm": 1.9978435039520264,
      "learning_rate": 2.1547895736362677e-06,
      "loss": 0.4547,
      "step": 268440
    },
    {
      "epoch": 109.89,
      "grad_norm": 1.8190250396728516,
      "learning_rate": 2.154631460913491e-06,
      "loss": 0.4697,
      "step": 268450
    },
    {
      "epoch": 109.89,
      "grad_norm": 1.989530324935913,
      "learning_rate": 2.154473349057037e-06,
      "loss": 0.4984,
      "step": 268460
    },
    {
      "epoch": 109.89,
      "grad_norm": 1.8505678176879883,
      "learning_rate": 2.15431523806763e-06,
      "loss": 0.4791,
      "step": 268470
    },
    {
      "epoch": 109.9,
      "grad_norm": 2.1136648654937744,
      "learning_rate": 2.1541571279459944e-06,
      "loss": 0.4657,
      "step": 268480
    },
    {
      "epoch": 109.9,
      "grad_norm": 1.919113039970398,
      "learning_rate": 2.1539990186928544e-06,
      "loss": 0.4702,
      "step": 268490
    },
    {
      "epoch": 109.91,
      "grad_norm": 1.8471527099609375,
      "learning_rate": 2.153840910308934e-06,
      "loss": 0.4757,
      "step": 268500
    },
    {
      "epoch": 109.91,
      "grad_norm": 1.7678719758987427,
      "learning_rate": 2.1536828027949585e-06,
      "loss": 0.4776,
      "step": 268510
    },
    {
      "epoch": 109.91,
      "grad_norm": 2.0109126567840576,
      "learning_rate": 2.1535246961516507e-06,
      "loss": 0.46,
      "step": 268520
    },
    {
      "epoch": 109.92,
      "grad_norm": 2.293888807296753,
      "learning_rate": 2.153366590379736e-06,
      "loss": 0.4762,
      "step": 268530
    },
    {
      "epoch": 109.92,
      "grad_norm": 1.8187514543533325,
      "learning_rate": 2.1532084854799375e-06,
      "loss": 0.4744,
      "step": 268540
    },
    {
      "epoch": 109.93,
      "grad_norm": 1.6411899328231812,
      "learning_rate": 2.1530503814529806e-06,
      "loss": 0.4772,
      "step": 268550
    },
    {
      "epoch": 109.93,
      "grad_norm": 1.7059499025344849,
      "learning_rate": 2.152892278299589e-06,
      "loss": 0.4561,
      "step": 268560
    },
    {
      "epoch": 109.93,
      "grad_norm": 2.1645426750183105,
      "learning_rate": 2.1527341760204872e-06,
      "loss": 0.4662,
      "step": 268570
    },
    {
      "epoch": 109.94,
      "grad_norm": 2.037428379058838,
      "learning_rate": 2.152576074616399e-06,
      "loss": 0.455,
      "step": 268580
    },
    {
      "epoch": 109.94,
      "grad_norm": 2.037916421890259,
      "learning_rate": 2.152417974088049e-06,
      "loss": 0.4957,
      "step": 268590
    },
    {
      "epoch": 109.95,
      "grad_norm": 1.899736762046814,
      "learning_rate": 2.1522598744361615e-06,
      "loss": 0.4743,
      "step": 268600
    },
    {
      "epoch": 109.95,
      "grad_norm": 1.5563188791275024,
      "learning_rate": 2.1521017756614606e-06,
      "loss": 0.4649,
      "step": 268610
    },
    {
      "epoch": 109.95,
      "grad_norm": 2.006657361984253,
      "learning_rate": 2.1519436777646705e-06,
      "loss": 0.4607,
      "step": 268620
    },
    {
      "epoch": 109.96,
      "grad_norm": 1.746373176574707,
      "learning_rate": 2.151785580746516e-06,
      "loss": 0.4829,
      "step": 268630
    },
    {
      "epoch": 109.96,
      "grad_norm": 2.03776216506958,
      "learning_rate": 2.1516274846077203e-06,
      "loss": 0.4526,
      "step": 268640
    },
    {
      "epoch": 109.97,
      "grad_norm": 2.1588754653930664,
      "learning_rate": 2.1514693893490084e-06,
      "loss": 0.4828,
      "step": 268650
    },
    {
      "epoch": 109.97,
      "grad_norm": 1.7831640243530273,
      "learning_rate": 2.151311294971104e-06,
      "loss": 0.4737,
      "step": 268660
    },
    {
      "epoch": 109.98,
      "grad_norm": 1.9995721578598022,
      "learning_rate": 2.1511532014747313e-06,
      "loss": 0.4603,
      "step": 268670
    },
    {
      "epoch": 109.98,
      "grad_norm": 2.0571649074554443,
      "learning_rate": 2.150995108860615e-06,
      "loss": 0.4754,
      "step": 268680
    },
    {
      "epoch": 109.98,
      "grad_norm": 1.6514225006103516,
      "learning_rate": 2.150837017129479e-06,
      "loss": 0.4675,
      "step": 268690
    },
    {
      "epoch": 109.99,
      "grad_norm": 1.862928867340088,
      "learning_rate": 2.150678926282047e-06,
      "loss": 0.4797,
      "step": 268700
    },
    {
      "epoch": 109.99,
      "grad_norm": 2.1616079807281494,
      "learning_rate": 2.150520836319044e-06,
      "loss": 0.4476,
      "step": 268710
    },
    {
      "epoch": 110.0,
      "grad_norm": 1.7584589719772339,
      "learning_rate": 2.150362747241194e-06,
      "loss": 0.4727,
      "step": 268720
    },
    {
      "epoch": 110.0,
      "grad_norm": 1.8384761810302734,
      "learning_rate": 2.150204659049221e-06,
      "loss": 0.4796,
      "step": 268730
    },
    {
      "epoch": 110.0,
      "eval_loss": 0.4693068861961365,
      "eval_runtime": 54.8248,
      "eval_samples_per_second": 62.91,
      "eval_steps_per_second": 7.88,
      "step": 268730
    },
    {
      "epoch": 110.0,
      "grad_norm": 2.067336320877075,
      "learning_rate": 2.1500465717438497e-06,
      "loss": 0.4568,
      "step": 268740
    },
    {
      "epoch": 110.01,
      "grad_norm": 1.7844619750976562,
      "learning_rate": 2.1498884853258036e-06,
      "loss": 0.4742,
      "step": 268750
    },
    {
      "epoch": 110.01,
      "grad_norm": 1.8063846826553345,
      "learning_rate": 2.1497303997958063e-06,
      "loss": 0.4611,
      "step": 268760
    },
    {
      "epoch": 110.02,
      "grad_norm": 1.8541271686553955,
      "learning_rate": 2.1495723151545833e-06,
      "loss": 0.4765,
      "step": 268770
    },
    {
      "epoch": 110.02,
      "grad_norm": 1.8466148376464844,
      "learning_rate": 2.149414231402857e-06,
      "loss": 0.4894,
      "step": 268780
    },
    {
      "epoch": 110.02,
      "grad_norm": 2.0500710010528564,
      "learning_rate": 2.1492561485413545e-06,
      "loss": 0.4726,
      "step": 268790
    },
    {
      "epoch": 110.03,
      "grad_norm": 2.1796681880950928,
      "learning_rate": 2.149098066570798e-06,
      "loss": 0.4472,
      "step": 268800
    },
    {
      "epoch": 110.03,
      "grad_norm": 1.8897552490234375,
      "learning_rate": 2.148939985491911e-06,
      "loss": 0.4824,
      "step": 268810
    },
    {
      "epoch": 110.04,
      "grad_norm": 2.195887804031372,
      "learning_rate": 2.1487819053054193e-06,
      "loss": 0.462,
      "step": 268820
    },
    {
      "epoch": 110.04,
      "grad_norm": 1.6267515420913696,
      "learning_rate": 2.1486238260120455e-06,
      "loss": 0.4456,
      "step": 268830
    },
    {
      "epoch": 110.05,
      "grad_norm": 2.4072299003601074,
      "learning_rate": 2.148465747612515e-06,
      "loss": 0.4591,
      "step": 268840
    },
    {
      "epoch": 110.05,
      "grad_norm": 1.9675980806350708,
      "learning_rate": 2.1483076701075515e-06,
      "loss": 0.4831,
      "step": 268850
    },
    {
      "epoch": 110.05,
      "grad_norm": 1.9335331916809082,
      "learning_rate": 2.148149593497879e-06,
      "loss": 0.4658,
      "step": 268860
    },
    {
      "epoch": 110.06,
      "grad_norm": 2.1950974464416504,
      "learning_rate": 2.1479915177842217e-06,
      "loss": 0.4868,
      "step": 268870
    },
    {
      "epoch": 110.06,
      "grad_norm": 2.1272926330566406,
      "learning_rate": 2.1478334429673036e-06,
      "loss": 0.4757,
      "step": 268880
    },
    {
      "epoch": 110.07,
      "grad_norm": 1.9266608953475952,
      "learning_rate": 2.147675369047849e-06,
      "loss": 0.4681,
      "step": 268890
    },
    {
      "epoch": 110.07,
      "grad_norm": 1.8931539058685303,
      "learning_rate": 2.1475172960265814e-06,
      "loss": 0.468,
      "step": 268900
    },
    {
      "epoch": 110.07,
      "grad_norm": 2.0881690979003906,
      "learning_rate": 2.147359223904226e-06,
      "loss": 0.4799,
      "step": 268910
    },
    {
      "epoch": 110.08,
      "grad_norm": 1.7686132192611694,
      "learning_rate": 2.147201152681506e-06,
      "loss": 0.4778,
      "step": 268920
    },
    {
      "epoch": 110.08,
      "grad_norm": 2.4337387084960938,
      "learning_rate": 2.147043082359146e-06,
      "loss": 0.5015,
      "step": 268930
    },
    {
      "epoch": 110.09,
      "grad_norm": 1.7519692182540894,
      "learning_rate": 2.1468850129378696e-06,
      "loss": 0.4658,
      "step": 268940
    },
    {
      "epoch": 110.09,
      "grad_norm": 1.7123619318008423,
      "learning_rate": 2.1467269444184017e-06,
      "loss": 0.453,
      "step": 268950
    },
    {
      "epoch": 110.09,
      "grad_norm": 2.49983549118042,
      "learning_rate": 2.146568876801466e-06,
      "loss": 0.487,
      "step": 268960
    },
    {
      "epoch": 110.1,
      "grad_norm": 1.957861304283142,
      "learning_rate": 2.146410810087785e-06,
      "loss": 0.4651,
      "step": 268970
    },
    {
      "epoch": 110.1,
      "grad_norm": 2.0687408447265625,
      "learning_rate": 2.1462527442780858e-06,
      "loss": 0.4608,
      "step": 268980
    },
    {
      "epoch": 110.11,
      "grad_norm": 1.8543270826339722,
      "learning_rate": 2.1460946793730904e-06,
      "loss": 0.4498,
      "step": 268990
    },
    {
      "epoch": 110.11,
      "grad_norm": 1.792284369468689,
      "learning_rate": 2.1459366153735233e-06,
      "loss": 0.4632,
      "step": 269000
    },
    {
      "epoch": 110.11,
      "grad_norm": 1.8961901664733887,
      "learning_rate": 2.14577855228011e-06,
      "loss": 0.4713,
      "step": 269010
    },
    {
      "epoch": 110.12,
      "grad_norm": 1.8825445175170898,
      "learning_rate": 2.145620490093572e-06,
      "loss": 0.4646,
      "step": 269020
    },
    {
      "epoch": 110.12,
      "grad_norm": 1.874479055404663,
      "learning_rate": 2.145462428814635e-06,
      "loss": 0.4503,
      "step": 269030
    },
    {
      "epoch": 110.13,
      "grad_norm": 1.6386364698410034,
      "learning_rate": 2.1453043684440225e-06,
      "loss": 0.4662,
      "step": 269040
    },
    {
      "epoch": 110.13,
      "grad_norm": 1.4310791492462158,
      "learning_rate": 2.1451463089824585e-06,
      "loss": 0.5026,
      "step": 269050
    },
    {
      "epoch": 110.14,
      "grad_norm": 2.238511800765991,
      "learning_rate": 2.1449882504306673e-06,
      "loss": 0.4631,
      "step": 269060
    },
    {
      "epoch": 110.14,
      "grad_norm": 2.4323532581329346,
      "learning_rate": 2.144830192789373e-06,
      "loss": 0.4766,
      "step": 269070
    },
    {
      "epoch": 110.14,
      "grad_norm": 2.0094563961029053,
      "learning_rate": 2.1446721360592996e-06,
      "loss": 0.4595,
      "step": 269080
    },
    {
      "epoch": 110.15,
      "grad_norm": 2.0299010276794434,
      "learning_rate": 2.144514080241171e-06,
      "loss": 0.4605,
      "step": 269090
    },
    {
      "epoch": 110.15,
      "grad_norm": 2.662498712539673,
      "learning_rate": 2.1443560253357116e-06,
      "loss": 0.4865,
      "step": 269100
    },
    {
      "epoch": 110.16,
      "grad_norm": 1.7527223825454712,
      "learning_rate": 2.144197971343645e-06,
      "loss": 0.4779,
      "step": 269110
    },
    {
      "epoch": 110.16,
      "grad_norm": 2.0868325233459473,
      "learning_rate": 2.144039918265695e-06,
      "loss": 0.4669,
      "step": 269120
    },
    {
      "epoch": 110.16,
      "grad_norm": 1.7187918424606323,
      "learning_rate": 2.1438818661025863e-06,
      "loss": 0.4651,
      "step": 269130
    },
    {
      "epoch": 110.17,
      "grad_norm": 2.2948050498962402,
      "learning_rate": 2.143723814855042e-06,
      "loss": 0.4714,
      "step": 269140
    },
    {
      "epoch": 110.17,
      "grad_norm": 2.0390381813049316,
      "learning_rate": 2.1435657645237874e-06,
      "loss": 0.4836,
      "step": 269150
    },
    {
      "epoch": 110.18,
      "grad_norm": 2.0684194564819336,
      "learning_rate": 2.143407715109545e-06,
      "loss": 0.4804,
      "step": 269160
    },
    {
      "epoch": 110.18,
      "grad_norm": 2.2885563373565674,
      "learning_rate": 2.1432496666130397e-06,
      "loss": 0.4732,
      "step": 269170
    },
    {
      "epoch": 110.18,
      "grad_norm": 1.8086297512054443,
      "learning_rate": 2.143091619034996e-06,
      "loss": 0.4565,
      "step": 269180
    },
    {
      "epoch": 110.19,
      "grad_norm": 1.9470415115356445,
      "learning_rate": 2.142933572376137e-06,
      "loss": 0.4845,
      "step": 269190
    },
    {
      "epoch": 110.19,
      "grad_norm": 2.0292468070983887,
      "learning_rate": 2.142775526637187e-06,
      "loss": 0.4584,
      "step": 269200
    },
    {
      "epoch": 110.2,
      "grad_norm": 1.571097731590271,
      "learning_rate": 2.1426174818188698e-06,
      "loss": 0.479,
      "step": 269210
    },
    {
      "epoch": 110.2,
      "grad_norm": 1.9552624225616455,
      "learning_rate": 2.1424594379219093e-06,
      "loss": 0.4762,
      "step": 269220
    },
    {
      "epoch": 110.2,
      "grad_norm": 1.8118396997451782,
      "learning_rate": 2.1423013949470307e-06,
      "loss": 0.4657,
      "step": 269230
    },
    {
      "epoch": 110.21,
      "grad_norm": 2.1152868270874023,
      "learning_rate": 2.142143352894956e-06,
      "loss": 0.4536,
      "step": 269240
    },
    {
      "epoch": 110.21,
      "grad_norm": 1.94746994972229,
      "learning_rate": 2.14198531176641e-06,
      "loss": 0.472,
      "step": 269250
    },
    {
      "epoch": 110.22,
      "grad_norm": 1.9197248220443726,
      "learning_rate": 2.1418272715621176e-06,
      "loss": 0.4759,
      "step": 269260
    },
    {
      "epoch": 110.22,
      "grad_norm": 1.862612009048462,
      "learning_rate": 2.1416692322828014e-06,
      "loss": 0.4735,
      "step": 269270
    },
    {
      "epoch": 110.23,
      "grad_norm": 3.043767213821411,
      "learning_rate": 2.141511193929186e-06,
      "loss": 0.4938,
      "step": 269280
    },
    {
      "epoch": 110.23,
      "grad_norm": 4.729262828826904,
      "learning_rate": 2.141353156501995e-06,
      "loss": 0.4675,
      "step": 269290
    },
    {
      "epoch": 110.23,
      "grad_norm": 1.9958854913711548,
      "learning_rate": 2.141195120001953e-06,
      "loss": 0.4807,
      "step": 269300
    },
    {
      "epoch": 110.24,
      "grad_norm": 1.619850516319275,
      "learning_rate": 2.141037084429783e-06,
      "loss": 0.4798,
      "step": 269310
    },
    {
      "epoch": 110.24,
      "grad_norm": 1.803859829902649,
      "learning_rate": 2.1408790497862107e-06,
      "loss": 0.4656,
      "step": 269320
    },
    {
      "epoch": 110.25,
      "grad_norm": 1.9732394218444824,
      "learning_rate": 2.1407210160719577e-06,
      "loss": 0.463,
      "step": 269330
    },
    {
      "epoch": 110.25,
      "grad_norm": 2.298719644546509,
      "learning_rate": 2.1405629832877493e-06,
      "loss": 0.4644,
      "step": 269340
    },
    {
      "epoch": 110.25,
      "grad_norm": 1.9266464710235596,
      "learning_rate": 2.1404049514343082e-06,
      "loss": 0.4576,
      "step": 269350
    },
    {
      "epoch": 110.26,
      "grad_norm": 1.9318032264709473,
      "learning_rate": 2.14024692051236e-06,
      "loss": 0.4584,
      "step": 269360
    },
    {
      "epoch": 110.26,
      "grad_norm": 1.9971253871917725,
      "learning_rate": 2.140088890522629e-06,
      "loss": 0.4701,
      "step": 269370
    },
    {
      "epoch": 110.27,
      "grad_norm": 1.9857780933380127,
      "learning_rate": 2.139930861465837e-06,
      "loss": 0.4704,
      "step": 269380
    },
    {
      "epoch": 110.27,
      "grad_norm": 1.7893173694610596,
      "learning_rate": 2.139772833342709e-06,
      "loss": 0.4749,
      "step": 269390
    },
    {
      "epoch": 110.27,
      "grad_norm": 1.5598423480987549,
      "learning_rate": 2.1396148061539685e-06,
      "loss": 0.4677,
      "step": 269400
    },
    {
      "epoch": 110.28,
      "grad_norm": 1.8476125001907349,
      "learning_rate": 2.13945677990034e-06,
      "loss": 0.4682,
      "step": 269410
    },
    {
      "epoch": 110.28,
      "grad_norm": 2.51310396194458,
      "learning_rate": 2.1392987545825473e-06,
      "loss": 0.4648,
      "step": 269420
    },
    {
      "epoch": 110.29,
      "grad_norm": 1.6638317108154297,
      "learning_rate": 2.139140730201314e-06,
      "loss": 0.4805,
      "step": 269430
    },
    {
      "epoch": 110.29,
      "grad_norm": 1.6079388856887817,
      "learning_rate": 2.138982706757364e-06,
      "loss": 0.458,
      "step": 269440
    },
    {
      "epoch": 110.29,
      "grad_norm": 2.3052566051483154,
      "learning_rate": 2.1388246842514213e-06,
      "loss": 0.4793,
      "step": 269450
    },
    {
      "epoch": 110.3,
      "grad_norm": 1.7908928394317627,
      "learning_rate": 2.1386666626842103e-06,
      "loss": 0.4578,
      "step": 269460
    },
    {
      "epoch": 110.3,
      "grad_norm": 1.9294158220291138,
      "learning_rate": 2.1385086420564536e-06,
      "loss": 0.4833,
      "step": 269470
    },
    {
      "epoch": 110.31,
      "grad_norm": 1.747304081916809,
      "learning_rate": 2.1383506223688756e-06,
      "loss": 0.4826,
      "step": 269480
    },
    {
      "epoch": 110.31,
      "grad_norm": 2.254020929336548,
      "learning_rate": 2.138192603622201e-06,
      "loss": 0.4682,
      "step": 269490
    },
    {
      "epoch": 110.32,
      "grad_norm": 1.8278357982635498,
      "learning_rate": 2.1380345858171523e-06,
      "loss": 0.4476,
      "step": 269500
    },
    {
      "epoch": 110.32,
      "grad_norm": 2.0180246829986572,
      "learning_rate": 2.1378765689544545e-06,
      "loss": 0.4811,
      "step": 269510
    },
    {
      "epoch": 110.32,
      "grad_norm": 2.2173399925231934,
      "learning_rate": 2.137718553034831e-06,
      "loss": 0.4744,
      "step": 269520
    },
    {
      "epoch": 110.33,
      "grad_norm": 1.9915955066680908,
      "learning_rate": 2.1375605380590046e-06,
      "loss": 0.484,
      "step": 269530
    },
    {
      "epoch": 110.33,
      "grad_norm": 2.441133499145508,
      "learning_rate": 2.1374025240277013e-06,
      "loss": 0.4811,
      "step": 269540
    },
    {
      "epoch": 110.34,
      "grad_norm": 2.2592740058898926,
      "learning_rate": 2.1372445109416435e-06,
      "loss": 0.4603,
      "step": 269550
    },
    {
      "epoch": 110.34,
      "grad_norm": 2.4719107151031494,
      "learning_rate": 2.1370864988015555e-06,
      "loss": 0.4614,
      "step": 269560
    },
    {
      "epoch": 110.34,
      "grad_norm": 1.4583144187927246,
      "learning_rate": 2.136928487608161e-06,
      "loss": 0.4504,
      "step": 269570
    },
    {
      "epoch": 110.35,
      "grad_norm": 2.481999397277832,
      "learning_rate": 2.136770477362184e-06,
      "loss": 0.4747,
      "step": 269580
    },
    {
      "epoch": 110.35,
      "grad_norm": 1.5666253566741943,
      "learning_rate": 2.1366124680643486e-06,
      "loss": 0.4556,
      "step": 269590
    },
    {
      "epoch": 110.36,
      "grad_norm": 2.399545431137085,
      "learning_rate": 2.1364544597153772e-06,
      "loss": 0.4745,
      "step": 269600
    },
    {
      "epoch": 110.36,
      "grad_norm": 1.6153404712677002,
      "learning_rate": 2.136296452315995e-06,
      "loss": 0.4902,
      "step": 269610
    },
    {
      "epoch": 110.36,
      "grad_norm": 1.9668982028961182,
      "learning_rate": 2.136138445866925e-06,
      "loss": 0.4595,
      "step": 269620
    },
    {
      "epoch": 110.37,
      "grad_norm": 1.721540093421936,
      "learning_rate": 2.1359804403688915e-06,
      "loss": 0.469,
      "step": 269630
    },
    {
      "epoch": 110.37,
      "grad_norm": 1.7407883405685425,
      "learning_rate": 2.1358224358226186e-06,
      "loss": 0.4758,
      "step": 269640
    },
    {
      "epoch": 110.38,
      "grad_norm": 1.603459119796753,
      "learning_rate": 2.1356644322288292e-06,
      "loss": 0.46,
      "step": 269650
    },
    {
      "epoch": 110.38,
      "grad_norm": 2.165762424468994,
      "learning_rate": 2.135506429588248e-06,
      "loss": 0.4758,
      "step": 269660
    },
    {
      "epoch": 110.38,
      "grad_norm": 1.590395450592041,
      "learning_rate": 2.135348427901598e-06,
      "loss": 0.4753,
      "step": 269670
    },
    {
      "epoch": 110.39,
      "grad_norm": 2.0004096031188965,
      "learning_rate": 2.135190427169604e-06,
      "loss": 0.4693,
      "step": 269680
    },
    {
      "epoch": 110.39,
      "grad_norm": 2.0217466354370117,
      "learning_rate": 2.1350324273929884e-06,
      "loss": 0.4559,
      "step": 269690
    },
    {
      "epoch": 110.4,
      "grad_norm": 2.1868796348571777,
      "learning_rate": 2.1348744285724758e-06,
      "loss": 0.4732,
      "step": 269700
    },
    {
      "epoch": 110.4,
      "grad_norm": 2.147770404815674,
      "learning_rate": 2.1347164307087897e-06,
      "loss": 0.4847,
      "step": 269710
    },
    {
      "epoch": 110.41,
      "grad_norm": 1.5978648662567139,
      "learning_rate": 2.1345584338026532e-06,
      "loss": 0.4535,
      "step": 269720
    },
    {
      "epoch": 110.41,
      "grad_norm": 2.627537250518799,
      "learning_rate": 2.1344004378547922e-06,
      "loss": 0.4714,
      "step": 269730
    },
    {
      "epoch": 110.41,
      "grad_norm": 2.290329933166504,
      "learning_rate": 2.134242442865929e-06,
      "loss": 0.5107,
      "step": 269740
    },
    {
      "epoch": 110.42,
      "grad_norm": 2.102435350418091,
      "learning_rate": 2.1340844488367868e-06,
      "loss": 0.4681,
      "step": 269750
    },
    {
      "epoch": 110.42,
      "grad_norm": 1.633235216140747,
      "learning_rate": 2.1339264557680907e-06,
      "loss": 0.4761,
      "step": 269760
    },
    {
      "epoch": 110.43,
      "grad_norm": 1.7800209522247314,
      "learning_rate": 2.133768463660563e-06,
      "loss": 0.4799,
      "step": 269770
    },
    {
      "epoch": 110.43,
      "grad_norm": 1.8330919742584229,
      "learning_rate": 2.133610472514929e-06,
      "loss": 0.469,
      "step": 269780
    },
    {
      "epoch": 110.43,
      "grad_norm": 1.6978052854537964,
      "learning_rate": 2.1334524823319113e-06,
      "loss": 0.5131,
      "step": 269790
    },
    {
      "epoch": 110.44,
      "grad_norm": 2.0130615234375,
      "learning_rate": 2.133294493112234e-06,
      "loss": 0.4716,
      "step": 269800
    },
    {
      "epoch": 110.44,
      "grad_norm": 2.3476269245147705,
      "learning_rate": 2.1331365048566208e-06,
      "loss": 0.4612,
      "step": 269810
    },
    {
      "epoch": 110.45,
      "grad_norm": 1.7519948482513428,
      "learning_rate": 2.132978517565796e-06,
      "loss": 0.4809,
      "step": 269820
    },
    {
      "epoch": 110.45,
      "grad_norm": 1.4953213930130005,
      "learning_rate": 2.1328205312404817e-06,
      "loss": 0.4642,
      "step": 269830
    },
    {
      "epoch": 110.45,
      "grad_norm": 1.8651819229125977,
      "learning_rate": 2.132662545881403e-06,
      "loss": 0.4602,
      "step": 269840
    },
    {
      "epoch": 110.46,
      "grad_norm": 2.048125743865967,
      "learning_rate": 2.1325045614892833e-06,
      "loss": 0.4758,
      "step": 269850
    },
    {
      "epoch": 110.46,
      "grad_norm": 1.957505702972412,
      "learning_rate": 2.1323465780648464e-06,
      "loss": 0.4742,
      "step": 269860
    },
    {
      "epoch": 110.47,
      "grad_norm": 1.8983198404312134,
      "learning_rate": 2.1321885956088154e-06,
      "loss": 0.4686,
      "step": 269870
    },
    {
      "epoch": 110.47,
      "grad_norm": 1.8761738538742065,
      "learning_rate": 2.1320306141219147e-06,
      "loss": 0.4607,
      "step": 269880
    },
    {
      "epoch": 110.47,
      "grad_norm": 1.8227125406265259,
      "learning_rate": 2.1318726336048675e-06,
      "loss": 0.4735,
      "step": 269890
    },
    {
      "epoch": 110.48,
      "grad_norm": 1.813890814781189,
      "learning_rate": 2.1317146540583985e-06,
      "loss": 0.473,
      "step": 269900
    },
    {
      "epoch": 110.48,
      "grad_norm": 1.5278929471969604,
      "learning_rate": 2.1315566754832286e-06,
      "loss": 0.4658,
      "step": 269910
    },
    {
      "epoch": 110.49,
      "grad_norm": 1.9704383611679077,
      "learning_rate": 2.1313986978800853e-06,
      "loss": 0.4743,
      "step": 269920
    },
    {
      "epoch": 110.49,
      "grad_norm": 1.9699991941452026,
      "learning_rate": 2.13124072124969e-06,
      "loss": 0.4607,
      "step": 269930
    },
    {
      "epoch": 110.5,
      "grad_norm": 1.7744203805923462,
      "learning_rate": 2.1310827455927666e-06,
      "loss": 0.4618,
      "step": 269940
    },
    {
      "epoch": 110.5,
      "grad_norm": 1.54220712184906,
      "learning_rate": 2.1309247709100396e-06,
      "loss": 0.4571,
      "step": 269950
    },
    {
      "epoch": 110.5,
      "grad_norm": 2.134676694869995,
      "learning_rate": 2.1307667972022314e-06,
      "loss": 0.4707,
      "step": 269960
    },
    {
      "epoch": 110.51,
      "grad_norm": 2.14314866065979,
      "learning_rate": 2.1306088244700662e-06,
      "loss": 0.4694,
      "step": 269970
    },
    {
      "epoch": 110.51,
      "grad_norm": 1.7408554553985596,
      "learning_rate": 2.130450852714268e-06,
      "loss": 0.4621,
      "step": 269980
    },
    {
      "epoch": 110.52,
      "grad_norm": 1.9102132320404053,
      "learning_rate": 2.13029288193556e-06,
      "loss": 0.4581,
      "step": 269990
    },
    {
      "epoch": 110.52,
      "grad_norm": 1.795168161392212,
      "learning_rate": 2.1301349121346657e-06,
      "loss": 0.4677,
      "step": 270000
    },
    {
      "epoch": 110.52,
      "grad_norm": 1.9609925746917725,
      "learning_rate": 2.1299769433123098e-06,
      "loss": 0.4736,
      "step": 270010
    },
    {
      "epoch": 110.53,
      "grad_norm": 1.6895619630813599,
      "learning_rate": 2.1298189754692146e-06,
      "loss": 0.4779,
      "step": 270020
    },
    {
      "epoch": 110.53,
      "grad_norm": 1.8804599046707153,
      "learning_rate": 2.1296610086061043e-06,
      "loss": 0.4799,
      "step": 270030
    },
    {
      "epoch": 110.54,
      "grad_norm": 2.1468636989593506,
      "learning_rate": 2.129503042723703e-06,
      "loss": 0.4711,
      "step": 270040
    },
    {
      "epoch": 110.54,
      "grad_norm": 1.7604167461395264,
      "learning_rate": 2.1293450778227337e-06,
      "loss": 0.4676,
      "step": 270050
    },
    {
      "epoch": 110.54,
      "grad_norm": 2.038275957107544,
      "learning_rate": 2.12918711390392e-06,
      "loss": 0.486,
      "step": 270060
    },
    {
      "epoch": 110.55,
      "grad_norm": 2.126075506210327,
      "learning_rate": 2.1290291509679854e-06,
      "loss": 0.4656,
      "step": 270070
    },
    {
      "epoch": 110.55,
      "grad_norm": 1.7991271018981934,
      "learning_rate": 2.128871189015654e-06,
      "loss": 0.4717,
      "step": 270080
    },
    {
      "epoch": 110.56,
      "grad_norm": 2.29716420173645,
      "learning_rate": 2.128713228047649e-06,
      "loss": 0.4685,
      "step": 270090
    },
    {
      "epoch": 110.56,
      "grad_norm": 1.7646839618682861,
      "learning_rate": 2.1285552680646936e-06,
      "loss": 0.4761,
      "step": 270100
    },
    {
      "epoch": 110.56,
      "grad_norm": 1.7521063089370728,
      "learning_rate": 2.128397309067513e-06,
      "loss": 0.4601,
      "step": 270110
    },
    {
      "epoch": 110.57,
      "grad_norm": 2.781919479370117,
      "learning_rate": 2.128239351056829e-06,
      "loss": 0.4615,
      "step": 270120
    },
    {
      "epoch": 110.57,
      "grad_norm": 2.1922922134399414,
      "learning_rate": 2.1280813940333664e-06,
      "loss": 0.4689,
      "step": 270130
    },
    {
      "epoch": 110.58,
      "grad_norm": 1.9359948635101318,
      "learning_rate": 2.1279234379978478e-06,
      "loss": 0.4815,
      "step": 270140
    },
    {
      "epoch": 110.58,
      "grad_norm": 1.9821722507476807,
      "learning_rate": 2.1277654829509975e-06,
      "loss": 0.4579,
      "step": 270150
    },
    {
      "epoch": 110.59,
      "grad_norm": 1.7971841096878052,
      "learning_rate": 2.127607528893539e-06,
      "loss": 0.441,
      "step": 270160
    },
    {
      "epoch": 110.59,
      "grad_norm": 2.0405712127685547,
      "learning_rate": 2.1274495758261956e-06,
      "loss": 0.4654,
      "step": 270170
    },
    {
      "epoch": 110.59,
      "grad_norm": 2.0897085666656494,
      "learning_rate": 2.1272916237496912e-06,
      "loss": 0.4654,
      "step": 270180
    },
    {
      "epoch": 110.6,
      "grad_norm": 1.9284799098968506,
      "learning_rate": 2.127133672664749e-06,
      "loss": 0.465,
      "step": 270190
    },
    {
      "epoch": 110.6,
      "grad_norm": 2.1714911460876465,
      "learning_rate": 2.1269757225720924e-06,
      "loss": 0.4832,
      "step": 270200
    },
    {
      "epoch": 110.61,
      "grad_norm": 1.8274767398834229,
      "learning_rate": 2.1268177734724454e-06,
      "loss": 0.4719,
      "step": 270210
    },
    {
      "epoch": 110.61,
      "grad_norm": 1.7788805961608887,
      "learning_rate": 2.1266598253665312e-06,
      "loss": 0.4628,
      "step": 270220
    },
    {
      "epoch": 110.61,
      "grad_norm": 7.216081619262695,
      "learning_rate": 2.126501878255073e-06,
      "loss": 0.4768,
      "step": 270230
    },
    {
      "epoch": 110.62,
      "grad_norm": 1.8252134323120117,
      "learning_rate": 2.1263439321387957e-06,
      "loss": 0.4592,
      "step": 270240
    },
    {
      "epoch": 110.62,
      "grad_norm": 2.7376110553741455,
      "learning_rate": 2.1261859870184214e-06,
      "loss": 0.4474,
      "step": 270250
    },
    {
      "epoch": 110.63,
      "grad_norm": 2.4214117527008057,
      "learning_rate": 2.1260280428946747e-06,
      "loss": 0.4709,
      "step": 270260
    },
    {
      "epoch": 110.63,
      "grad_norm": 1.9921839237213135,
      "learning_rate": 2.125870099768278e-06,
      "loss": 0.478,
      "step": 270270
    },
    {
      "epoch": 110.63,
      "grad_norm": 2.213221788406372,
      "learning_rate": 2.1257121576399556e-06,
      "loss": 0.4564,
      "step": 270280
    },
    {
      "epoch": 110.64,
      "grad_norm": 2.0326128005981445,
      "learning_rate": 2.1255542165104296e-06,
      "loss": 0.4652,
      "step": 270290
    },
    {
      "epoch": 110.64,
      "grad_norm": 1.6409071683883667,
      "learning_rate": 2.1253962763804254e-06,
      "loss": 0.4563,
      "step": 270300
    },
    {
      "epoch": 110.65,
      "grad_norm": 2.0452630519866943,
      "learning_rate": 2.1252383372506667e-06,
      "loss": 0.452,
      "step": 270310
    },
    {
      "epoch": 110.65,
      "grad_norm": 1.7274701595306396,
      "learning_rate": 2.1250803991218757e-06,
      "loss": 0.461,
      "step": 270320
    },
    {
      "epoch": 110.65,
      "grad_norm": 2.8656797409057617,
      "learning_rate": 2.124922461994776e-06,
      "loss": 0.4639,
      "step": 270330
    },
    {
      "epoch": 110.66,
      "grad_norm": 1.7199511528015137,
      "learning_rate": 2.1247645258700917e-06,
      "loss": 0.4792,
      "step": 270340
    },
    {
      "epoch": 110.66,
      "grad_norm": 2.2365925312042236,
      "learning_rate": 2.124606590748546e-06,
      "loss": 0.4901,
      "step": 270350
    },
    {
      "epoch": 110.67,
      "grad_norm": 1.8727277517318726,
      "learning_rate": 2.1244486566308615e-06,
      "loss": 0.4621,
      "step": 270360
    },
    {
      "epoch": 110.67,
      "grad_norm": 1.708808183670044,
      "learning_rate": 2.124290723517763e-06,
      "loss": 0.4574,
      "step": 270370
    },
    {
      "epoch": 110.68,
      "grad_norm": 1.7654136419296265,
      "learning_rate": 2.1241327914099734e-06,
      "loss": 0.4794,
      "step": 270380
    },
    {
      "epoch": 110.68,
      "grad_norm": 2.215456485748291,
      "learning_rate": 2.1239748603082164e-06,
      "loss": 0.4796,
      "step": 270390
    },
    {
      "epoch": 110.68,
      "grad_norm": 2.0262184143066406,
      "learning_rate": 2.1238169302132157e-06,
      "loss": 0.4615,
      "step": 270400
    },
    {
      "epoch": 110.69,
      "grad_norm": 1.8379416465759277,
      "learning_rate": 2.123659001125694e-06,
      "loss": 0.4591,
      "step": 270410
    },
    {
      "epoch": 110.69,
      "grad_norm": 1.8575729131698608,
      "learning_rate": 2.1235010730463747e-06,
      "loss": 0.4581,
      "step": 270420
    },
    {
      "epoch": 110.7,
      "grad_norm": 2.223212480545044,
      "learning_rate": 2.1233431459759816e-06,
      "loss": 0.4775,
      "step": 270430
    },
    {
      "epoch": 110.7,
      "grad_norm": 2.0577504634857178,
      "learning_rate": 2.1231852199152383e-06,
      "loss": 0.4666,
      "step": 270440
    },
    {
      "epoch": 110.7,
      "grad_norm": 2.525489568710327,
      "learning_rate": 2.1230272948648684e-06,
      "loss": 0.4503,
      "step": 270450
    },
    {
      "epoch": 110.71,
      "grad_norm": 1.712642788887024,
      "learning_rate": 2.1228693708255946e-06,
      "loss": 0.4539,
      "step": 270460
    },
    {
      "epoch": 110.71,
      "grad_norm": 1.6157678365707397,
      "learning_rate": 2.122711447798141e-06,
      "loss": 0.4584,
      "step": 270470
    },
    {
      "epoch": 110.72,
      "grad_norm": 1.7615647315979004,
      "learning_rate": 2.1225535257832303e-06,
      "loss": 0.4514,
      "step": 270480
    },
    {
      "epoch": 110.72,
      "grad_norm": 2.0641722679138184,
      "learning_rate": 2.122395604781587e-06,
      "loss": 0.4979,
      "step": 270490
    },
    {
      "epoch": 110.72,
      "grad_norm": 2.200185775756836,
      "learning_rate": 2.1222376847939334e-06,
      "loss": 0.4704,
      "step": 270500
    },
    {
      "epoch": 110.73,
      "grad_norm": 1.430079460144043,
      "learning_rate": 2.122079765820994e-06,
      "loss": 0.4655,
      "step": 270510
    },
    {
      "epoch": 110.73,
      "grad_norm": 1.9333230257034302,
      "learning_rate": 2.1219218478634913e-06,
      "loss": 0.4547,
      "step": 270520
    },
    {
      "epoch": 110.74,
      "grad_norm": 1.9237889051437378,
      "learning_rate": 2.1217639309221495e-06,
      "loss": 0.4779,
      "step": 270530
    },
    {
      "epoch": 110.74,
      "grad_norm": 2.0505433082580566,
      "learning_rate": 2.121606014997691e-06,
      "loss": 0.4518,
      "step": 270540
    },
    {
      "epoch": 110.74,
      "grad_norm": 1.677517294883728,
      "learning_rate": 2.1214481000908398e-06,
      "loss": 0.474,
      "step": 270550
    },
    {
      "epoch": 110.75,
      "grad_norm": 1.704588532447815,
      "learning_rate": 2.121290186202319e-06,
      "loss": 0.4558,
      "step": 270560
    },
    {
      "epoch": 110.75,
      "grad_norm": 2.2551140785217285,
      "learning_rate": 2.1211322733328526e-06,
      "loss": 0.487,
      "step": 270570
    },
    {
      "epoch": 110.76,
      "grad_norm": 2.1977903842926025,
      "learning_rate": 2.120974361483163e-06,
      "loss": 0.4619,
      "step": 270580
    },
    {
      "epoch": 110.76,
      "grad_norm": 2.190345048904419,
      "learning_rate": 2.1208164506539745e-06,
      "loss": 0.4724,
      "step": 270590
    },
    {
      "epoch": 110.77,
      "grad_norm": 1.8504188060760498,
      "learning_rate": 2.1206585408460096e-06,
      "loss": 0.4677,
      "step": 270600
    },
    {
      "epoch": 110.77,
      "grad_norm": 2.159522533416748,
      "learning_rate": 2.1205006320599924e-06,
      "loss": 0.4529,
      "step": 270610
    },
    {
      "epoch": 110.77,
      "grad_norm": 1.8343671560287476,
      "learning_rate": 2.120342724296647e-06,
      "loss": 0.4761,
      "step": 270620
    },
    {
      "epoch": 110.78,
      "grad_norm": 1.781779408454895,
      "learning_rate": 2.1201848175566945e-06,
      "loss": 0.4569,
      "step": 270630
    },
    {
      "epoch": 110.78,
      "grad_norm": 2.2600934505462646,
      "learning_rate": 2.1200269118408593e-06,
      "loss": 0.4661,
      "step": 270640
    },
    {
      "epoch": 110.79,
      "grad_norm": 1.7729088068008423,
      "learning_rate": 2.1198690071498657e-06,
      "loss": 0.4587,
      "step": 270650
    },
    {
      "epoch": 110.79,
      "grad_norm": 2.0333316326141357,
      "learning_rate": 2.119711103484436e-06,
      "loss": 0.4736,
      "step": 270660
    },
    {
      "epoch": 110.79,
      "grad_norm": 2.2298903465270996,
      "learning_rate": 2.1195532008452928e-06,
      "loss": 0.4555,
      "step": 270670
    },
    {
      "epoch": 110.8,
      "grad_norm": 1.8368538618087769,
      "learning_rate": 2.119395299233161e-06,
      "loss": 0.4839,
      "step": 270680
    },
    {
      "epoch": 110.8,
      "grad_norm": 2.1319873332977295,
      "learning_rate": 2.119237398648764e-06,
      "loss": 0.4502,
      "step": 270690
    },
    {
      "epoch": 110.81,
      "grad_norm": 1.7802762985229492,
      "learning_rate": 2.119079499092824e-06,
      "loss": 0.4754,
      "step": 270700
    },
    {
      "epoch": 110.81,
      "grad_norm": 1.7045419216156006,
      "learning_rate": 2.1189216005660647e-06,
      "loss": 0.4711,
      "step": 270710
    },
    {
      "epoch": 110.81,
      "grad_norm": 1.6599059104919434,
      "learning_rate": 2.11876370306921e-06,
      "loss": 0.4692,
      "step": 270720
    },
    {
      "epoch": 110.82,
      "grad_norm": 2.0168986320495605,
      "learning_rate": 2.118605806602982e-06,
      "loss": 0.4505,
      "step": 270730
    },
    {
      "epoch": 110.82,
      "grad_norm": 1.779587984085083,
      "learning_rate": 2.118447911168105e-06,
      "loss": 0.4646,
      "step": 270740
    },
    {
      "epoch": 110.83,
      "grad_norm": 2.1148834228515625,
      "learning_rate": 2.118290016765302e-06,
      "loss": 0.4648,
      "step": 270750
    },
    {
      "epoch": 110.83,
      "grad_norm": 2.012869119644165,
      "learning_rate": 2.1181321233952966e-06,
      "loss": 0.4699,
      "step": 270760
    },
    {
      "epoch": 110.84,
      "grad_norm": 1.8265169858932495,
      "learning_rate": 2.1179742310588113e-06,
      "loss": 0.4632,
      "step": 270770
    },
    {
      "epoch": 110.84,
      "grad_norm": 1.6208791732788086,
      "learning_rate": 2.1178163397565698e-06,
      "loss": 0.4601,
      "step": 270780
    },
    {
      "epoch": 110.84,
      "grad_norm": 1.833630919456482,
      "learning_rate": 2.117658449489296e-06,
      "loss": 0.5026,
      "step": 270790
    },
    {
      "epoch": 110.85,
      "grad_norm": 1.8125579357147217,
      "learning_rate": 2.117500560257712e-06,
      "loss": 0.4573,
      "step": 270800
    },
    {
      "epoch": 110.85,
      "grad_norm": 1.5914671421051025,
      "learning_rate": 2.1173426720625414e-06,
      "loss": 0.458,
      "step": 270810
    },
    {
      "epoch": 110.86,
      "grad_norm": 2.09684157371521,
      "learning_rate": 2.117184784904508e-06,
      "loss": 0.4626,
      "step": 270820
    },
    {
      "epoch": 110.86,
      "grad_norm": 2.5340213775634766,
      "learning_rate": 2.117026898784335e-06,
      "loss": 0.4923,
      "step": 270830
    },
    {
      "epoch": 110.86,
      "grad_norm": 1.7358074188232422,
      "learning_rate": 2.1168690137027457e-06,
      "loss": 0.4661,
      "step": 270840
    },
    {
      "epoch": 110.87,
      "grad_norm": 2.6620469093322754,
      "learning_rate": 2.1167111296604617e-06,
      "loss": 0.4859,
      "step": 270850
    },
    {
      "epoch": 110.87,
      "grad_norm": 1.7235838174819946,
      "learning_rate": 2.116553246658208e-06,
      "loss": 0.4597,
      "step": 270860
    },
    {
      "epoch": 110.88,
      "grad_norm": 1.924230694770813,
      "learning_rate": 2.116395364696708e-06,
      "loss": 0.463,
      "step": 270870
    },
    {
      "epoch": 110.88,
      "grad_norm": 1.8326741456985474,
      "learning_rate": 2.116237483776684e-06,
      "loss": 0.4718,
      "step": 270880
    },
    {
      "epoch": 110.88,
      "grad_norm": 1.9831149578094482,
      "learning_rate": 2.116079603898861e-06,
      "loss": 0.4638,
      "step": 270890
    },
    {
      "epoch": 110.89,
      "grad_norm": 2.9222447872161865,
      "learning_rate": 2.1159217250639597e-06,
      "loss": 0.4789,
      "step": 270900
    },
    {
      "epoch": 110.89,
      "grad_norm": 2.445796489715576,
      "learning_rate": 2.1157638472727043e-06,
      "loss": 0.4631,
      "step": 270910
    },
    {
      "epoch": 110.9,
      "grad_norm": 1.8655118942260742,
      "learning_rate": 2.1156059705258185e-06,
      "loss": 0.4677,
      "step": 270920
    },
    {
      "epoch": 110.9,
      "grad_norm": 1.897403597831726,
      "learning_rate": 2.115448094824025e-06,
      "loss": 0.4691,
      "step": 270930
    },
    {
      "epoch": 110.9,
      "grad_norm": 1.743949294090271,
      "learning_rate": 2.115290220168047e-06,
      "loss": 0.459,
      "step": 270940
    },
    {
      "epoch": 110.91,
      "grad_norm": 1.9721672534942627,
      "learning_rate": 2.115132346558608e-06,
      "loss": 0.4549,
      "step": 270950
    },
    {
      "epoch": 110.91,
      "grad_norm": 1.7281179428100586,
      "learning_rate": 2.114974473996431e-06,
      "loss": 0.4607,
      "step": 270960
    },
    {
      "epoch": 110.92,
      "grad_norm": 2.300675392150879,
      "learning_rate": 2.1148166024822395e-06,
      "loss": 0.4601,
      "step": 270970
    },
    {
      "epoch": 110.92,
      "grad_norm": 1.7252320051193237,
      "learning_rate": 2.114658732016757e-06,
      "loss": 0.4624,
      "step": 270980
    },
    {
      "epoch": 110.93,
      "grad_norm": 1.841765284538269,
      "learning_rate": 2.114500862600705e-06,
      "loss": 0.4553,
      "step": 270990
    },
    {
      "epoch": 110.93,
      "grad_norm": 1.5208345651626587,
      "learning_rate": 2.1143429942348083e-06,
      "loss": 0.4772,
      "step": 271000
    },
    {
      "epoch": 110.93,
      "grad_norm": 1.9223358631134033,
      "learning_rate": 2.1141851269197893e-06,
      "loss": 0.4529,
      "step": 271010
    },
    {
      "epoch": 110.94,
      "grad_norm": 2.3888983726501465,
      "learning_rate": 2.114027260656372e-06,
      "loss": 0.4932,
      "step": 271020
    },
    {
      "epoch": 110.94,
      "grad_norm": 2.458498239517212,
      "learning_rate": 2.1138693954452783e-06,
      "loss": 0.4756,
      "step": 271030
    },
    {
      "epoch": 110.95,
      "grad_norm": 2.046226739883423,
      "learning_rate": 2.1137115312872316e-06,
      "loss": 0.4659,
      "step": 271040
    },
    {
      "epoch": 110.95,
      "grad_norm": 1.9206719398498535,
      "learning_rate": 2.1135536681829566e-06,
      "loss": 0.479,
      "step": 271050
    },
    {
      "epoch": 110.95,
      "grad_norm": 2.6191375255584717,
      "learning_rate": 2.113395806133175e-06,
      "loss": 0.501,
      "step": 271060
    },
    {
      "epoch": 110.96,
      "grad_norm": 2.3906476497650146,
      "learning_rate": 2.11323794513861e-06,
      "loss": 0.5028,
      "step": 271070
    },
    {
      "epoch": 110.96,
      "grad_norm": 3.298793077468872,
      "learning_rate": 2.1130800851999855e-06,
      "loss": 0.4889,
      "step": 271080
    },
    {
      "epoch": 110.97,
      "grad_norm": 1.7534102201461792,
      "learning_rate": 2.1129222263180237e-06,
      "loss": 0.4734,
      "step": 271090
    },
    {
      "epoch": 110.97,
      "grad_norm": 1.9510267972946167,
      "learning_rate": 2.1127643684934486e-06,
      "loss": 0.4762,
      "step": 271100
    },
    {
      "epoch": 110.97,
      "grad_norm": 2.3982696533203125,
      "learning_rate": 2.112606511726983e-06,
      "loss": 0.4407,
      "step": 271110
    },
    {
      "epoch": 110.98,
      "grad_norm": 1.6825896501541138,
      "learning_rate": 2.11244865601935e-06,
      "loss": 0.4616,
      "step": 271120
    },
    {
      "epoch": 110.98,
      "grad_norm": 1.8623414039611816,
      "learning_rate": 2.1122908013712724e-06,
      "loss": 0.4656,
      "step": 271130
    },
    {
      "epoch": 110.99,
      "grad_norm": 2.0222342014312744,
      "learning_rate": 2.112132947783473e-06,
      "loss": 0.4729,
      "step": 271140
    },
    {
      "epoch": 110.99,
      "grad_norm": 2.0418176651000977,
      "learning_rate": 2.1119750952566762e-06,
      "loss": 0.4595,
      "step": 271150
    },
    {
      "epoch": 110.99,
      "grad_norm": 1.7923914194107056,
      "learning_rate": 2.111817243791604e-06,
      "loss": 0.4819,
      "step": 271160
    },
    {
      "epoch": 111.0,
      "grad_norm": 1.8301291465759277,
      "learning_rate": 2.11165939338898e-06,
      "loss": 0.463,
      "step": 271170
    },
    {
      "epoch": 111.0,
      "eval_loss": 0.4695966839790344,
      "eval_runtime": 52.5569,
      "eval_samples_per_second": 65.624,
      "eval_steps_per_second": 8.22,
      "step": 271173
    },
    {
      "epoch": 111.0,
      "grad_norm": 2.3141252994537354,
      "learning_rate": 2.111501544049527e-06,
      "loss": 0.4714,
      "step": 271180
    },
    {
      "epoch": 111.01,
      "grad_norm": 1.5893077850341797,
      "learning_rate": 2.111343695773968e-06,
      "loss": 0.469,
      "step": 271190
    },
    {
      "epoch": 111.01,
      "grad_norm": 1.8182543516159058,
      "learning_rate": 2.1111858485630267e-06,
      "loss": 0.4615,
      "step": 271200
    },
    {
      "epoch": 111.02,
      "grad_norm": 2.1129000186920166,
      "learning_rate": 2.111028002417426e-06,
      "loss": 0.4708,
      "step": 271210
    },
    {
      "epoch": 111.02,
      "grad_norm": 2.5147042274475098,
      "learning_rate": 2.1108701573378883e-06,
      "loss": 0.4693,
      "step": 271220
    },
    {
      "epoch": 111.02,
      "grad_norm": 1.807457685470581,
      "learning_rate": 2.1107123133251358e-06,
      "loss": 0.478,
      "step": 271230
    },
    {
      "epoch": 111.03,
      "grad_norm": 1.9948821067810059,
      "learning_rate": 2.1105544703798942e-06,
      "loss": 0.4616,
      "step": 271240
    },
    {
      "epoch": 111.03,
      "grad_norm": 2.3371665477752686,
      "learning_rate": 2.1103966285028858e-06,
      "loss": 0.4663,
      "step": 271250
    },
    {
      "epoch": 111.04,
      "grad_norm": 2.009453535079956,
      "learning_rate": 2.110238787694832e-06,
      "loss": 0.4633,
      "step": 271260
    },
    {
      "epoch": 111.04,
      "grad_norm": 1.976435661315918,
      "learning_rate": 2.1100809479564573e-06,
      "loss": 0.4669,
      "step": 271270
    },
    {
      "epoch": 111.04,
      "grad_norm": 1.7794837951660156,
      "learning_rate": 2.1099231092884844e-06,
      "loss": 0.4684,
      "step": 271280
    },
    {
      "epoch": 111.05,
      "grad_norm": 1.9994524717330933,
      "learning_rate": 2.1097652716916363e-06,
      "loss": 0.479,
      "step": 271290
    },
    {
      "epoch": 111.05,
      "grad_norm": 2.0635011196136475,
      "learning_rate": 2.109607435166636e-06,
      "loss": 0.4738,
      "step": 271300
    },
    {
      "epoch": 111.06,
      "grad_norm": 1.9252660274505615,
      "learning_rate": 2.1094495997142064e-06,
      "loss": 0.4608,
      "step": 271310
    },
    {
      "epoch": 111.06,
      "grad_norm": 1.613724708557129,
      "learning_rate": 2.1092917653350705e-06,
      "loss": 0.4776,
      "step": 271320
    },
    {
      "epoch": 111.06,
      "grad_norm": 1.9644993543624878,
      "learning_rate": 2.1091339320299516e-06,
      "loss": 0.4716,
      "step": 271330
    },
    {
      "epoch": 111.07,
      "grad_norm": 2.2545173168182373,
      "learning_rate": 2.108976099799573e-06,
      "loss": 0.4707,
      "step": 271340
    },
    {
      "epoch": 111.07,
      "grad_norm": 2.2871251106262207,
      "learning_rate": 2.108818268644657e-06,
      "loss": 0.4835,
      "step": 271350
    },
    {
      "epoch": 111.08,
      "grad_norm": 2.1603410243988037,
      "learning_rate": 2.108660438565927e-06,
      "loss": 0.4849,
      "step": 271360
    },
    {
      "epoch": 111.08,
      "grad_norm": 3.871142625808716,
      "learning_rate": 2.1085026095641056e-06,
      "loss": 0.4805,
      "step": 271370
    },
    {
      "epoch": 111.08,
      "grad_norm": 1.9539092779159546,
      "learning_rate": 2.108344781639916e-06,
      "loss": 0.4832,
      "step": 271380
    },
    {
      "epoch": 111.09,
      "grad_norm": 1.9102813005447388,
      "learning_rate": 2.1081869547940814e-06,
      "loss": 0.4529,
      "step": 271390
    },
    {
      "epoch": 111.09,
      "grad_norm": 3.1207873821258545,
      "learning_rate": 2.1080291290273245e-06,
      "loss": 0.4658,
      "step": 271400
    },
    {
      "epoch": 111.1,
      "grad_norm": 2.0198349952697754,
      "learning_rate": 2.1078713043403686e-06,
      "loss": 0.4766,
      "step": 271410
    },
    {
      "epoch": 111.1,
      "grad_norm": 1.8495560884475708,
      "learning_rate": 2.1077134807339358e-06,
      "loss": 0.4798,
      "step": 271420
    },
    {
      "epoch": 111.11,
      "grad_norm": 1.809264063835144,
      "learning_rate": 2.1075556582087502e-06,
      "loss": 0.4813,
      "step": 271430
    },
    {
      "epoch": 111.11,
      "grad_norm": 2.1145784854888916,
      "learning_rate": 2.1073978367655345e-06,
      "loss": 0.4559,
      "step": 271440
    },
    {
      "epoch": 111.11,
      "grad_norm": 1.8181170225143433,
      "learning_rate": 2.107240016405011e-06,
      "loss": 0.4756,
      "step": 271450
    },
    {
      "epoch": 111.12,
      "grad_norm": 1.8035953044891357,
      "learning_rate": 2.107082197127904e-06,
      "loss": 0.4711,
      "step": 271460
    },
    {
      "epoch": 111.12,
      "grad_norm": 1.8026071786880493,
      "learning_rate": 2.106924378934935e-06,
      "loss": 0.4692,
      "step": 271470
    },
    {
      "epoch": 111.13,
      "grad_norm": 6.957906723022461,
      "learning_rate": 2.106766561826828e-06,
      "loss": 0.475,
      "step": 271480
    },
    {
      "epoch": 111.13,
      "grad_norm": 2.0773561000823975,
      "learning_rate": 2.1066087458043045e-06,
      "loss": 0.4776,
      "step": 271490
    },
    {
      "epoch": 111.13,
      "grad_norm": 1.7373183965682983,
      "learning_rate": 2.106450930868089e-06,
      "loss": 0.4805,
      "step": 271500
    },
    {
      "epoch": 111.14,
      "grad_norm": 1.994123935699463,
      "learning_rate": 2.1062931170189036e-06,
      "loss": 0.4642,
      "step": 271510
    },
    {
      "epoch": 111.14,
      "grad_norm": 2.167778253555298,
      "learning_rate": 2.1061353042574715e-06,
      "loss": 0.4672,
      "step": 271520
    },
    {
      "epoch": 111.15,
      "grad_norm": 1.8317588567733765,
      "learning_rate": 2.1059774925845153e-06,
      "loss": 0.4625,
      "step": 271530
    },
    {
      "epoch": 111.15,
      "grad_norm": 1.725647211074829,
      "learning_rate": 2.105819682000758e-06,
      "loss": 0.4573,
      "step": 271540
    },
    {
      "epoch": 111.15,
      "grad_norm": 1.6960463523864746,
      "learning_rate": 2.105661872506923e-06,
      "loss": 0.4624,
      "step": 271550
    },
    {
      "epoch": 111.16,
      "grad_norm": 2.082362174987793,
      "learning_rate": 2.105504064103733e-06,
      "loss": 0.4718,
      "step": 271560
    },
    {
      "epoch": 111.16,
      "grad_norm": 2.897054672241211,
      "learning_rate": 2.1053462567919107e-06,
      "loss": 0.4819,
      "step": 271570
    },
    {
      "epoch": 111.17,
      "grad_norm": 2.0021140575408936,
      "learning_rate": 2.105188450572179e-06,
      "loss": 0.4785,
      "step": 271580
    },
    {
      "epoch": 111.17,
      "grad_norm": 1.6907296180725098,
      "learning_rate": 2.1050306454452605e-06,
      "loss": 0.4557,
      "step": 271590
    },
    {
      "epoch": 111.17,
      "grad_norm": 2.5870249271392822,
      "learning_rate": 2.1048728414118783e-06,
      "loss": 0.4737,
      "step": 271600
    },
    {
      "epoch": 111.18,
      "grad_norm": 2.3341257572174072,
      "learning_rate": 2.104715038472755e-06,
      "loss": 0.471,
      "step": 271610
    },
    {
      "epoch": 111.18,
      "grad_norm": 2.1112239360809326,
      "learning_rate": 2.1045572366286145e-06,
      "loss": 0.4904,
      "step": 271620
    },
    {
      "epoch": 111.19,
      "grad_norm": 1.9830455780029297,
      "learning_rate": 2.104399435880179e-06,
      "loss": 0.4821,
      "step": 271630
    },
    {
      "epoch": 111.19,
      "grad_norm": 2.8456122875213623,
      "learning_rate": 2.104241636228171e-06,
      "loss": 0.4577,
      "step": 271640
    },
    {
      "epoch": 111.2,
      "grad_norm": 2.0539650917053223,
      "learning_rate": 2.104083837673314e-06,
      "loss": 0.4584,
      "step": 271650
    },
    {
      "epoch": 111.2,
      "grad_norm": 2.426666021347046,
      "learning_rate": 2.103926040216331e-06,
      "loss": 0.452,
      "step": 271660
    },
    {
      "epoch": 111.2,
      "grad_norm": 1.7541168928146362,
      "learning_rate": 2.1037682438579438e-06,
      "loss": 0.4609,
      "step": 271670
    },
    {
      "epoch": 111.21,
      "grad_norm": 1.9146349430084229,
      "learning_rate": 2.103610448598876e-06,
      "loss": 0.4769,
      "step": 271680
    },
    {
      "epoch": 111.21,
      "grad_norm": 2.2279117107391357,
      "learning_rate": 2.10345265443985e-06,
      "loss": 0.4939,
      "step": 271690
    },
    {
      "epoch": 111.22,
      "grad_norm": 2.3500332832336426,
      "learning_rate": 2.1032948613815898e-06,
      "loss": 0.4805,
      "step": 271700
    },
    {
      "epoch": 111.22,
      "grad_norm": 2.058695077896118,
      "learning_rate": 2.103137069424817e-06,
      "loss": 0.4946,
      "step": 271710
    },
    {
      "epoch": 111.22,
      "grad_norm": 2.063251495361328,
      "learning_rate": 2.1029792785702544e-06,
      "loss": 0.4746,
      "step": 271720
    },
    {
      "epoch": 111.23,
      "grad_norm": 2.238065481185913,
      "learning_rate": 2.1028214888186256e-06,
      "loss": 0.49,
      "step": 271730
    },
    {
      "epoch": 111.23,
      "grad_norm": 2.053306818008423,
      "learning_rate": 2.102663700170653e-06,
      "loss": 0.4822,
      "step": 271740
    },
    {
      "epoch": 111.24,
      "grad_norm": 1.9712508916854858,
      "learning_rate": 2.102505912627059e-06,
      "loss": 0.4583,
      "step": 271750
    },
    {
      "epoch": 111.24,
      "grad_norm": 1.9742482900619507,
      "learning_rate": 2.102348126188567e-06,
      "loss": 0.4659,
      "step": 271760
    },
    {
      "epoch": 111.24,
      "grad_norm": 1.7125911712646484,
      "learning_rate": 2.1021903408558993e-06,
      "loss": 0.4624,
      "step": 271770
    },
    {
      "epoch": 111.25,
      "grad_norm": 2.095493793487549,
      "learning_rate": 2.1020325566297795e-06,
      "loss": 0.4643,
      "step": 271780
    },
    {
      "epoch": 111.25,
      "grad_norm": 1.8622006177902222,
      "learning_rate": 2.10187477351093e-06,
      "loss": 0.4602,
      "step": 271790
    },
    {
      "epoch": 111.26,
      "grad_norm": 1.9718736410140991,
      "learning_rate": 2.1017169915000718e-06,
      "loss": 0.4576,
      "step": 271800
    },
    {
      "epoch": 111.26,
      "grad_norm": 2.202084541320801,
      "learning_rate": 2.1015592105979307e-06,
      "loss": 0.4492,
      "step": 271810
    },
    {
      "epoch": 111.26,
      "grad_norm": 1.6473971605300903,
      "learning_rate": 2.1014014308052283e-06,
      "loss": 0.4712,
      "step": 271820
    },
    {
      "epoch": 111.27,
      "grad_norm": 2.147481918334961,
      "learning_rate": 2.1012436521226866e-06,
      "loss": 0.4805,
      "step": 271830
    },
    {
      "epoch": 111.27,
      "grad_norm": 1.9384382963180542,
      "learning_rate": 2.1010858745510295e-06,
      "loss": 0.4827,
      "step": 271840
    },
    {
      "epoch": 111.28,
      "grad_norm": 2.2826626300811768,
      "learning_rate": 2.100928098090979e-06,
      "loss": 0.4669,
      "step": 271850
    },
    {
      "epoch": 111.28,
      "grad_norm": 1.4534670114517212,
      "learning_rate": 2.100770322743258e-06,
      "loss": 0.4672,
      "step": 271860
    },
    {
      "epoch": 111.29,
      "grad_norm": 1.81383216381073,
      "learning_rate": 2.100612548508589e-06,
      "loss": 0.4528,
      "step": 271870
    },
    {
      "epoch": 111.29,
      "grad_norm": 2.0529298782348633,
      "learning_rate": 2.100454775387695e-06,
      "loss": 0.4729,
      "step": 271880
    },
    {
      "epoch": 111.29,
      "grad_norm": 2.184572219848633,
      "learning_rate": 2.1002970033812994e-06,
      "loss": 0.4751,
      "step": 271890
    },
    {
      "epoch": 111.3,
      "grad_norm": 1.7353171110153198,
      "learning_rate": 2.1001392324901238e-06,
      "loss": 0.4673,
      "step": 271900
    },
    {
      "epoch": 111.3,
      "grad_norm": 2.0915963649749756,
      "learning_rate": 2.0999814627148913e-06,
      "loss": 0.4785,
      "step": 271910
    },
    {
      "epoch": 111.31,
      "grad_norm": 1.932770848274231,
      "learning_rate": 2.099823694056325e-06,
      "loss": 0.4717,
      "step": 271920
    },
    {
      "epoch": 111.31,
      "grad_norm": 1.7135621309280396,
      "learning_rate": 2.0996659265151475e-06,
      "loss": 0.4797,
      "step": 271930
    },
    {
      "epoch": 111.31,
      "grad_norm": 2.0312578678131104,
      "learning_rate": 2.099508160092081e-06,
      "loss": 0.4545,
      "step": 271940
    },
    {
      "epoch": 111.32,
      "grad_norm": 2.157487392425537,
      "learning_rate": 2.0993503947878482e-06,
      "loss": 0.4592,
      "step": 271950
    },
    {
      "epoch": 111.32,
      "grad_norm": 2.2637107372283936,
      "learning_rate": 2.099192630603173e-06,
      "loss": 0.4623,
      "step": 271960
    },
    {
      "epoch": 111.33,
      "grad_norm": 2.016416311264038,
      "learning_rate": 2.099034867538776e-06,
      "loss": 0.4862,
      "step": 271970
    },
    {
      "epoch": 111.33,
      "grad_norm": 1.8724344968795776,
      "learning_rate": 2.098877105595382e-06,
      "loss": 0.4478,
      "step": 271980
    },
    {
      "epoch": 111.33,
      "grad_norm": 2.048856735229492,
      "learning_rate": 2.0987193447737123e-06,
      "loss": 0.4546,
      "step": 271990
    },
    {
      "epoch": 111.34,
      "grad_norm": 2.1041786670684814,
      "learning_rate": 2.0985615850744906e-06,
      "loss": 0.4772,
      "step": 272000
    },
    {
      "epoch": 111.34,
      "grad_norm": 1.7031054496765137,
      "learning_rate": 2.098403826498439e-06,
      "loss": 0.4574,
      "step": 272010
    },
    {
      "epoch": 111.35,
      "grad_norm": 2.1017231941223145,
      "learning_rate": 2.0982460690462796e-06,
      "loss": 0.4586,
      "step": 272020
    },
    {
      "epoch": 111.35,
      "grad_norm": 1.8414644002914429,
      "learning_rate": 2.0980883127187365e-06,
      "loss": 0.4786,
      "step": 272030
    },
    {
      "epoch": 111.35,
      "grad_norm": 1.8528791666030884,
      "learning_rate": 2.097930557516531e-06,
      "loss": 0.4661,
      "step": 272040
    },
    {
      "epoch": 111.36,
      "grad_norm": 2.0111985206604004,
      "learning_rate": 2.0977728034403866e-06,
      "loss": 0.4673,
      "step": 272050
    },
    {
      "epoch": 111.36,
      "grad_norm": 1.7353670597076416,
      "learning_rate": 2.097615050491026e-06,
      "loss": 0.4807,
      "step": 272060
    },
    {
      "epoch": 111.37,
      "grad_norm": 1.6902028322219849,
      "learning_rate": 2.097457298669171e-06,
      "loss": 0.4784,
      "step": 272070
    },
    {
      "epoch": 111.37,
      "grad_norm": 2.5008342266082764,
      "learning_rate": 2.0972995479755448e-06,
      "loss": 0.4767,
      "step": 272080
    },
    {
      "epoch": 111.38,
      "grad_norm": 2.8476078510284424,
      "learning_rate": 2.0971417984108703e-06,
      "loss": 0.4534,
      "step": 272090
    },
    {
      "epoch": 111.38,
      "grad_norm": 1.9799785614013672,
      "learning_rate": 2.0969840499758693e-06,
      "loss": 0.4726,
      "step": 272100
    },
    {
      "epoch": 111.38,
      "grad_norm": 2.1101090908050537,
      "learning_rate": 2.096826302671265e-06,
      "loss": 0.4697,
      "step": 272110
    },
    {
      "epoch": 111.39,
      "grad_norm": 2.2131221294403076,
      "learning_rate": 2.0966685564977797e-06,
      "loss": 0.4832,
      "step": 272120
    },
    {
      "epoch": 111.39,
      "grad_norm": 2.375494956970215,
      "learning_rate": 2.0965108114561362e-06,
      "loss": 0.4756,
      "step": 272130
    },
    {
      "epoch": 111.4,
      "grad_norm": 1.856069564819336,
      "learning_rate": 2.0963530675470576e-06,
      "loss": 0.4574,
      "step": 272140
    },
    {
      "epoch": 111.4,
      "grad_norm": 2.003990650177002,
      "learning_rate": 2.0961953247712666e-06,
      "loss": 0.4719,
      "step": 272150
    },
    {
      "epoch": 111.4,
      "grad_norm": 1.6443321704864502,
      "learning_rate": 2.0960375831294842e-06,
      "loss": 0.4637,
      "step": 272160
    },
    {
      "epoch": 111.41,
      "grad_norm": 1.749566912651062,
      "learning_rate": 2.0958798426224333e-06,
      "loss": 0.4496,
      "step": 272170
    },
    {
      "epoch": 111.41,
      "grad_norm": 1.7175428867340088,
      "learning_rate": 2.095722103250838e-06,
      "loss": 0.4644,
      "step": 272180
    },
    {
      "epoch": 111.42,
      "grad_norm": 1.77882981300354,
      "learning_rate": 2.0955643650154204e-06,
      "loss": 0.4695,
      "step": 272190
    },
    {
      "epoch": 111.42,
      "grad_norm": 2.325214385986328,
      "learning_rate": 2.0954066279169033e-06,
      "loss": 0.4833,
      "step": 272200
    },
    {
      "epoch": 111.42,
      "grad_norm": 2.192338466644287,
      "learning_rate": 2.095248891956008e-06,
      "loss": 0.4625,
      "step": 272210
    },
    {
      "epoch": 111.43,
      "grad_norm": 1.8302857875823975,
      "learning_rate": 2.0950911571334577e-06,
      "loss": 0.4474,
      "step": 272220
    },
    {
      "epoch": 111.43,
      "grad_norm": 1.7807631492614746,
      "learning_rate": 2.094933423449975e-06,
      "loss": 0.4733,
      "step": 272230
    },
    {
      "epoch": 111.44,
      "grad_norm": 2.4905035495758057,
      "learning_rate": 2.094775690906283e-06,
      "loss": 0.4887,
      "step": 272240
    },
    {
      "epoch": 111.44,
      "grad_norm": 2.454960823059082,
      "learning_rate": 2.094617959503103e-06,
      "loss": 0.4636,
      "step": 272250
    },
    {
      "epoch": 111.44,
      "grad_norm": 2.382181406021118,
      "learning_rate": 2.094460229241159e-06,
      "loss": 0.4767,
      "step": 272260
    },
    {
      "epoch": 111.45,
      "grad_norm": 1.8801530599594116,
      "learning_rate": 2.0943025001211725e-06,
      "loss": 0.4633,
      "step": 272270
    },
    {
      "epoch": 111.45,
      "grad_norm": 1.5339857339859009,
      "learning_rate": 2.094144772143867e-06,
      "loss": 0.4619,
      "step": 272280
    },
    {
      "epoch": 111.46,
      "grad_norm": 1.920682668685913,
      "learning_rate": 2.0939870453099638e-06,
      "loss": 0.4649,
      "step": 272290
    },
    {
      "epoch": 111.46,
      "grad_norm": 2.07470703125,
      "learning_rate": 2.093829319620186e-06,
      "loss": 0.465,
      "step": 272300
    },
    {
      "epoch": 111.47,
      "grad_norm": 2.002636671066284,
      "learning_rate": 2.093671595075256e-06,
      "loss": 0.464,
      "step": 272310
    },
    {
      "epoch": 111.47,
      "grad_norm": 2.1512069702148438,
      "learning_rate": 2.0935138716758967e-06,
      "loss": 0.4576,
      "step": 272320
    },
    {
      "epoch": 111.47,
      "grad_norm": 2.21829891204834,
      "learning_rate": 2.0933561494228306e-06,
      "loss": 0.4654,
      "step": 272330
    },
    {
      "epoch": 111.48,
      "grad_norm": 2.1659975051879883,
      "learning_rate": 2.0931984283167792e-06,
      "loss": 0.4691,
      "step": 272340
    },
    {
      "epoch": 111.48,
      "grad_norm": 2.0815584659576416,
      "learning_rate": 2.0930407083584664e-06,
      "loss": 0.4896,
      "step": 272350
    },
    {
      "epoch": 111.49,
      "grad_norm": 1.7350959777832031,
      "learning_rate": 2.092882989548613e-06,
      "loss": 0.4741,
      "step": 272360
    },
    {
      "epoch": 111.49,
      "grad_norm": 2.1579887866973877,
      "learning_rate": 2.092725271887943e-06,
      "loss": 0.4817,
      "step": 272370
    },
    {
      "epoch": 111.49,
      "grad_norm": 2.5298173427581787,
      "learning_rate": 2.0925675553771794e-06,
      "loss": 0.4792,
      "step": 272380
    },
    {
      "epoch": 111.5,
      "grad_norm": 1.638283610343933,
      "learning_rate": 2.0924098400170425e-06,
      "loss": 0.45,
      "step": 272390
    },
    {
      "epoch": 111.5,
      "grad_norm": 2.0690293312072754,
      "learning_rate": 2.0922521258082568e-06,
      "loss": 0.4818,
      "step": 272400
    },
    {
      "epoch": 111.51,
      "grad_norm": 1.87565279006958,
      "learning_rate": 2.0920944127515434e-06,
      "loss": 0.489,
      "step": 272410
    },
    {
      "epoch": 111.51,
      "grad_norm": 2.0037479400634766,
      "learning_rate": 2.091936700847626e-06,
      "loss": 0.4684,
      "step": 272420
    },
    {
      "epoch": 111.51,
      "grad_norm": 1.8983391523361206,
      "learning_rate": 2.0917789900972257e-06,
      "loss": 0.4736,
      "step": 272430
    },
    {
      "epoch": 111.52,
      "grad_norm": 1.7095891237258911,
      "learning_rate": 2.0916212805010654e-06,
      "loss": 0.467,
      "step": 272440
    },
    {
      "epoch": 111.52,
      "grad_norm": 1.88275945186615,
      "learning_rate": 2.091463572059868e-06,
      "loss": 0.4869,
      "step": 272450
    },
    {
      "epoch": 111.53,
      "grad_norm": 1.5578936338424683,
      "learning_rate": 2.0913058647743557e-06,
      "loss": 0.4852,
      "step": 272460
    },
    {
      "epoch": 111.53,
      "grad_norm": 2.2858474254608154,
      "learning_rate": 2.091148158645251e-06,
      "loss": 0.4451,
      "step": 272470
    },
    {
      "epoch": 111.53,
      "grad_norm": 1.7367316484451294,
      "learning_rate": 2.090990453673276e-06,
      "loss": 0.4568,
      "step": 272480
    },
    {
      "epoch": 111.54,
      "grad_norm": 2.104783296585083,
      "learning_rate": 2.090832749859153e-06,
      "loss": 0.4787,
      "step": 272490
    },
    {
      "epoch": 111.54,
      "grad_norm": 1.9673147201538086,
      "learning_rate": 2.0906750472036055e-06,
      "loss": 0.4732,
      "step": 272500
    },
    {
      "epoch": 111.55,
      "grad_norm": 1.9123799800872803,
      "learning_rate": 2.090517345707355e-06,
      "loss": 0.4729,
      "step": 272510
    },
    {
      "epoch": 111.55,
      "grad_norm": 2.067310094833374,
      "learning_rate": 2.090359645371124e-06,
      "loss": 0.4591,
      "step": 272520
    },
    {
      "epoch": 111.56,
      "grad_norm": 2.113751173019409,
      "learning_rate": 2.0902019461956345e-06,
      "loss": 0.4722,
      "step": 272530
    },
    {
      "epoch": 111.56,
      "grad_norm": 1.6934103965759277,
      "learning_rate": 2.0900442481816097e-06,
      "loss": 0.482,
      "step": 272540
    },
    {
      "epoch": 111.56,
      "grad_norm": 2.3496389389038086,
      "learning_rate": 2.0898865513297712e-06,
      "loss": 0.4556,
      "step": 272550
    },
    {
      "epoch": 111.57,
      "grad_norm": 1.5671502351760864,
      "learning_rate": 2.089728855640842e-06,
      "loss": 0.4525,
      "step": 272560
    },
    {
      "epoch": 111.57,
      "grad_norm": 2.047447919845581,
      "learning_rate": 2.089571161115545e-06,
      "loss": 0.4834,
      "step": 272570
    },
    {
      "epoch": 111.58,
      "grad_norm": 1.9238479137420654,
      "learning_rate": 2.0894134677546014e-06,
      "loss": 0.4538,
      "step": 272580
    },
    {
      "epoch": 111.58,
      "grad_norm": 1.8496044874191284,
      "learning_rate": 2.089255775558734e-06,
      "loss": 0.4685,
      "step": 272590
    },
    {
      "epoch": 111.58,
      "grad_norm": 2.07003116607666,
      "learning_rate": 2.0890980845286653e-06,
      "loss": 0.4783,
      "step": 272600
    },
    {
      "epoch": 111.59,
      "grad_norm": 1.9027200937271118,
      "learning_rate": 2.0889403946651176e-06,
      "loss": 0.4615,
      "step": 272610
    },
    {
      "epoch": 111.59,
      "grad_norm": 1.951607584953308,
      "learning_rate": 2.0887827059688135e-06,
      "loss": 0.4803,
      "step": 272620
    },
    {
      "epoch": 111.6,
      "grad_norm": 1.7176650762557983,
      "learning_rate": 2.0886250184404746e-06,
      "loss": 0.4608,
      "step": 272630
    },
    {
      "epoch": 111.6,
      "grad_norm": 1.8735384941101074,
      "learning_rate": 2.088467332080825e-06,
      "loss": 0.4864,
      "step": 272640
    },
    {
      "epoch": 111.6,
      "grad_norm": 1.6191468238830566,
      "learning_rate": 2.088309646890585e-06,
      "loss": 0.4778,
      "step": 272650
    },
    {
      "epoch": 111.61,
      "grad_norm": 1.7653146982192993,
      "learning_rate": 2.088151962870477e-06,
      "loss": 0.4604,
      "step": 272660
    },
    {
      "epoch": 111.61,
      "grad_norm": 1.7894212007522583,
      "learning_rate": 2.0879942800212247e-06,
      "loss": 0.4503,
      "step": 272670
    },
    {
      "epoch": 111.62,
      "grad_norm": 1.8448868989944458,
      "learning_rate": 2.0878365983435495e-06,
      "loss": 0.4842,
      "step": 272680
    },
    {
      "epoch": 111.62,
      "grad_norm": 2.213979959487915,
      "learning_rate": 2.087678917838174e-06,
      "loss": 0.4469,
      "step": 272690
    },
    {
      "epoch": 111.63,
      "grad_norm": 2.0842325687408447,
      "learning_rate": 2.0875212385058202e-06,
      "loss": 0.4787,
      "step": 272700
    },
    {
      "epoch": 111.63,
      "grad_norm": 1.8545271158218384,
      "learning_rate": 2.087363560347211e-06,
      "loss": 0.4644,
      "step": 272710
    },
    {
      "epoch": 111.63,
      "grad_norm": 2.0874650478363037,
      "learning_rate": 2.0872058833630683e-06,
      "loss": 0.4457,
      "step": 272720
    },
    {
      "epoch": 111.64,
      "grad_norm": 1.7775232791900635,
      "learning_rate": 2.087048207554115e-06,
      "loss": 0.4753,
      "step": 272730
    },
    {
      "epoch": 111.64,
      "grad_norm": 1.865707278251648,
      "learning_rate": 2.0868905329210715e-06,
      "loss": 0.4447,
      "step": 272740
    },
    {
      "epoch": 111.65,
      "grad_norm": 2.2813308238983154,
      "learning_rate": 2.086732859464662e-06,
      "loss": 0.4922,
      "step": 272750
    },
    {
      "epoch": 111.65,
      "grad_norm": 2.031888246536255,
      "learning_rate": 2.086575187185609e-06,
      "loss": 0.4472,
      "step": 272760
    },
    {
      "epoch": 111.65,
      "grad_norm": 2.042616367340088,
      "learning_rate": 2.086417516084633e-06,
      "loss": 0.4798,
      "step": 272770
    },
    {
      "epoch": 111.66,
      "grad_norm": 1.9013530015945435,
      "learning_rate": 2.0862598461624584e-06,
      "loss": 0.4908,
      "step": 272780
    },
    {
      "epoch": 111.66,
      "grad_norm": 2.3706531524658203,
      "learning_rate": 2.086102177419806e-06,
      "loss": 0.4722,
      "step": 272790
    },
    {
      "epoch": 111.67,
      "grad_norm": 1.8400852680206299,
      "learning_rate": 2.085944509857398e-06,
      "loss": 0.4776,
      "step": 272800
    },
    {
      "epoch": 111.67,
      "grad_norm": 1.9083919525146484,
      "learning_rate": 2.085786843475957e-06,
      "loss": 0.4903,
      "step": 272810
    },
    {
      "epoch": 111.67,
      "grad_norm": 1.7513939142227173,
      "learning_rate": 2.0856291782762053e-06,
      "loss": 0.4546,
      "step": 272820
    },
    {
      "epoch": 111.68,
      "grad_norm": 1.9831863641738892,
      "learning_rate": 2.085471514258865e-06,
      "loss": 0.4657,
      "step": 272830
    },
    {
      "epoch": 111.68,
      "grad_norm": 1.9893338680267334,
      "learning_rate": 2.085313851424659e-06,
      "loss": 0.4636,
      "step": 272840
    },
    {
      "epoch": 111.69,
      "grad_norm": 2.426980972290039,
      "learning_rate": 2.085156189774308e-06,
      "loss": 0.4545,
      "step": 272850
    },
    {
      "epoch": 111.69,
      "grad_norm": 1.8353232145309448,
      "learning_rate": 2.0849985293085364e-06,
      "loss": 0.4786,
      "step": 272860
    },
    {
      "epoch": 111.69,
      "grad_norm": 2.1160707473754883,
      "learning_rate": 2.084840870028065e-06,
      "loss": 0.4731,
      "step": 272870
    },
    {
      "epoch": 111.7,
      "grad_norm": 1.9133002758026123,
      "learning_rate": 2.0846832119336153e-06,
      "loss": 0.4711,
      "step": 272880
    },
    {
      "epoch": 111.7,
      "grad_norm": 2.01826548576355,
      "learning_rate": 2.084525555025911e-06,
      "loss": 0.4678,
      "step": 272890
    },
    {
      "epoch": 111.71,
      "grad_norm": 1.7451958656311035,
      "learning_rate": 2.084367899305674e-06,
      "loss": 0.4695,
      "step": 272900
    },
    {
      "epoch": 111.71,
      "grad_norm": 1.6845381259918213,
      "learning_rate": 2.084210244773626e-06,
      "loss": 0.4821,
      "step": 272910
    },
    {
      "epoch": 111.72,
      "grad_norm": 1.7460672855377197,
      "learning_rate": 2.084052591430489e-06,
      "loss": 0.4522,
      "step": 272920
    },
    {
      "epoch": 111.72,
      "grad_norm": 1.8779937028884888,
      "learning_rate": 2.0838949392769853e-06,
      "loss": 0.4669,
      "step": 272930
    },
    {
      "epoch": 111.72,
      "grad_norm": 1.695592999458313,
      "learning_rate": 2.083737288313838e-06,
      "loss": 0.4963,
      "step": 272940
    },
    {
      "epoch": 111.73,
      "grad_norm": 2.096660614013672,
      "learning_rate": 2.083579638541769e-06,
      "loss": 0.472,
      "step": 272950
    },
    {
      "epoch": 111.73,
      "grad_norm": 2.13521146774292,
      "learning_rate": 2.0834219899614996e-06,
      "loss": 0.4738,
      "step": 272960
    },
    {
      "epoch": 111.74,
      "grad_norm": 2.1137049198150635,
      "learning_rate": 2.0832643425737527e-06,
      "loss": 0.4574,
      "step": 272970
    },
    {
      "epoch": 111.74,
      "grad_norm": 2.2351481914520264,
      "learning_rate": 2.0831066963792503e-06,
      "loss": 0.4678,
      "step": 272980
    },
    {
      "epoch": 111.74,
      "grad_norm": 1.971416711807251,
      "learning_rate": 2.0829490513787145e-06,
      "loss": 0.4668,
      "step": 272990
    },
    {
      "epoch": 111.75,
      "grad_norm": 2.021519184112549,
      "learning_rate": 2.082791407572868e-06,
      "loss": 0.4616,
      "step": 273000
    },
    {
      "epoch": 111.75,
      "grad_norm": 2.0104384422302246,
      "learning_rate": 2.0826337649624317e-06,
      "loss": 0.481,
      "step": 273010
    },
    {
      "epoch": 111.76,
      "grad_norm": 2.344419240951538,
      "learning_rate": 2.0824761235481284e-06,
      "loss": 0.4726,
      "step": 273020
    },
    {
      "epoch": 111.76,
      "grad_norm": 1.9674910306930542,
      "learning_rate": 2.0823184833306806e-06,
      "loss": 0.474,
      "step": 273030
    },
    {
      "epoch": 111.76,
      "grad_norm": 2.1652073860168457,
      "learning_rate": 2.0821608443108103e-06,
      "loss": 0.4465,
      "step": 273040
    },
    {
      "epoch": 111.77,
      "grad_norm": 1.7884736061096191,
      "learning_rate": 2.0820032064892387e-06,
      "loss": 0.4483,
      "step": 273050
    },
    {
      "epoch": 111.77,
      "grad_norm": 1.8351961374282837,
      "learning_rate": 2.0818455698666893e-06,
      "loss": 0.4712,
      "step": 273060
    },
    {
      "epoch": 111.78,
      "grad_norm": 2.225853204727173,
      "learning_rate": 2.081687934443883e-06,
      "loss": 0.4516,
      "step": 273070
    },
    {
      "epoch": 111.78,
      "grad_norm": 6.2342209815979,
      "learning_rate": 2.081530300221543e-06,
      "loss": 0.4872,
      "step": 273080
    },
    {
      "epoch": 111.78,
      "grad_norm": 1.7883579730987549,
      "learning_rate": 2.0813726672003908e-06,
      "loss": 0.4729,
      "step": 273090
    },
    {
      "epoch": 111.79,
      "grad_norm": 1.6966215372085571,
      "learning_rate": 2.081215035381148e-06,
      "loss": 0.4763,
      "step": 273100
    },
    {
      "epoch": 111.79,
      "grad_norm": 1.653609037399292,
      "learning_rate": 2.0810574047645374e-06,
      "loss": 0.4591,
      "step": 273110
    },
    {
      "epoch": 111.8,
      "grad_norm": 2.195732831954956,
      "learning_rate": 2.08089977535128e-06,
      "loss": 0.4621,
      "step": 273120
    },
    {
      "epoch": 111.8,
      "grad_norm": 1.743841290473938,
      "learning_rate": 2.0807421471420995e-06,
      "loss": 0.4439,
      "step": 273130
    },
    {
      "epoch": 111.81,
      "grad_norm": 1.5952730178833008,
      "learning_rate": 2.080584520137718e-06,
      "loss": 0.4903,
      "step": 273140
    },
    {
      "epoch": 111.81,
      "grad_norm": 1.8477803468704224,
      "learning_rate": 2.0804268943388563e-06,
      "loss": 0.4605,
      "step": 273150
    },
    {
      "epoch": 111.81,
      "grad_norm": 2.1517555713653564,
      "learning_rate": 2.080269269746237e-06,
      "loss": 0.4859,
      "step": 273160
    },
    {
      "epoch": 111.82,
      "grad_norm": 2.2545201778411865,
      "learning_rate": 2.080111646360582e-06,
      "loss": 0.4544,
      "step": 273170
    },
    {
      "epoch": 111.82,
      "grad_norm": 1.79799485206604,
      "learning_rate": 2.079954024182613e-06,
      "loss": 0.4343,
      "step": 273180
    },
    {
      "epoch": 111.83,
      "grad_norm": 2.19938063621521,
      "learning_rate": 2.079796403213053e-06,
      "loss": 0.4651,
      "step": 273190
    },
    {
      "epoch": 111.83,
      "grad_norm": 1.8623663187026978,
      "learning_rate": 2.079638783452624e-06,
      "loss": 0.4502,
      "step": 273200
    },
    {
      "epoch": 111.83,
      "grad_norm": 1.4988350868225098,
      "learning_rate": 2.079481164902047e-06,
      "loss": 0.4672,
      "step": 273210
    },
    {
      "epoch": 111.84,
      "grad_norm": 1.8169797658920288,
      "learning_rate": 2.0793235475620454e-06,
      "loss": 0.4622,
      "step": 273220
    },
    {
      "epoch": 111.84,
      "grad_norm": 1.8918144702911377,
      "learning_rate": 2.0791659314333396e-06,
      "loss": 0.4636,
      "step": 273230
    },
    {
      "epoch": 111.85,
      "grad_norm": 1.8086100816726685,
      "learning_rate": 2.0790083165166524e-06,
      "loss": 0.4649,
      "step": 273240
    },
    {
      "epoch": 111.85,
      "grad_norm": 1.8293880224227905,
      "learning_rate": 2.0788507028127063e-06,
      "loss": 0.4579,
      "step": 273250
    },
    {
      "epoch": 111.85,
      "grad_norm": 1.7473208904266357,
      "learning_rate": 2.0786930903222226e-06,
      "loss": 0.4555,
      "step": 273260
    },
    {
      "epoch": 111.86,
      "grad_norm": 1.855217695236206,
      "learning_rate": 2.0785354790459233e-06,
      "loss": 0.464,
      "step": 273270
    },
    {
      "epoch": 111.86,
      "grad_norm": 2.4152042865753174,
      "learning_rate": 2.078377868984531e-06,
      "loss": 0.4711,
      "step": 273280
    },
    {
      "epoch": 111.87,
      "grad_norm": 1.8310490846633911,
      "learning_rate": 2.0782202601387674e-06,
      "loss": 0.4864,
      "step": 273290
    },
    {
      "epoch": 111.87,
      "grad_norm": 2.545013427734375,
      "learning_rate": 2.078062652509354e-06,
      "loss": 0.4607,
      "step": 273300
    },
    {
      "epoch": 111.87,
      "grad_norm": 1.8619850873947144,
      "learning_rate": 2.077905046097013e-06,
      "loss": 0.4635,
      "step": 273310
    },
    {
      "epoch": 111.88,
      "grad_norm": 1.823720097541809,
      "learning_rate": 2.0777474409024668e-06,
      "loss": 0.4568,
      "step": 273320
    },
    {
      "epoch": 111.88,
      "grad_norm": 1.9845712184906006,
      "learning_rate": 2.0775898369264374e-06,
      "loss": 0.458,
      "step": 273330
    },
    {
      "epoch": 111.89,
      "grad_norm": 2.000154495239258,
      "learning_rate": 2.0774322341696468e-06,
      "loss": 0.4538,
      "step": 273340
    },
    {
      "epoch": 111.89,
      "grad_norm": 1.7242155075073242,
      "learning_rate": 2.077274632632816e-06,
      "loss": 0.4523,
      "step": 273350
    },
    {
      "epoch": 111.9,
      "grad_norm": 1.8982837200164795,
      "learning_rate": 2.0771170323166687e-06,
      "loss": 0.455,
      "step": 273360
    },
    {
      "epoch": 111.9,
      "grad_norm": 1.7185721397399902,
      "learning_rate": 2.0769594332219246e-06,
      "loss": 0.4753,
      "step": 273370
    },
    {
      "epoch": 111.9,
      "grad_norm": 1.8539217710494995,
      "learning_rate": 2.0768018353493072e-06,
      "loss": 0.4806,
      "step": 273380
    },
    {
      "epoch": 111.91,
      "grad_norm": 1.961765170097351,
      "learning_rate": 2.0766442386995377e-06,
      "loss": 0.4688,
      "step": 273390
    },
    {
      "epoch": 111.91,
      "grad_norm": 1.8017637729644775,
      "learning_rate": 2.0764866432733387e-06,
      "loss": 0.4774,
      "step": 273400
    },
    {
      "epoch": 111.92,
      "grad_norm": 1.567234754562378,
      "learning_rate": 2.0763290490714313e-06,
      "loss": 0.4609,
      "step": 273410
    },
    {
      "epoch": 111.92,
      "grad_norm": 1.517140507698059,
      "learning_rate": 2.076171456094538e-06,
      "loss": 0.4584,
      "step": 273420
    },
    {
      "epoch": 111.92,
      "grad_norm": 2.255927562713623,
      "learning_rate": 2.076013864343381e-06,
      "loss": 0.4712,
      "step": 273430
    },
    {
      "epoch": 111.93,
      "grad_norm": 1.8641681671142578,
      "learning_rate": 2.075856273818681e-06,
      "loss": 0.4463,
      "step": 273440
    },
    {
      "epoch": 111.93,
      "grad_norm": 1.9215456247329712,
      "learning_rate": 2.0756986845211618e-06,
      "loss": 0.4747,
      "step": 273450
    },
    {
      "epoch": 111.94,
      "grad_norm": 1.5373859405517578,
      "learning_rate": 2.0755410964515434e-06,
      "loss": 0.4637,
      "step": 273460
    },
    {
      "epoch": 111.94,
      "grad_norm": 2.0710179805755615,
      "learning_rate": 2.075383509610549e-06,
      "loss": 0.4773,
      "step": 273470
    },
    {
      "epoch": 111.94,
      "grad_norm": 1.9114065170288086,
      "learning_rate": 2.075225923998899e-06,
      "loss": 0.4578,
      "step": 273480
    },
    {
      "epoch": 111.95,
      "grad_norm": 1.8097920417785645,
      "learning_rate": 2.0750683396173158e-06,
      "loss": 0.4647,
      "step": 273490
    },
    {
      "epoch": 111.95,
      "grad_norm": 1.8012748956680298,
      "learning_rate": 2.0749107564665228e-06,
      "loss": 0.4659,
      "step": 273500
    },
    {
      "epoch": 111.96,
      "grad_norm": 1.7404245138168335,
      "learning_rate": 2.0747531745472407e-06,
      "loss": 0.4459,
      "step": 273510
    },
    {
      "epoch": 111.96,
      "grad_norm": 1.9058066606521606,
      "learning_rate": 2.074595593860191e-06,
      "loss": 0.4416,
      "step": 273520
    },
    {
      "epoch": 111.96,
      "grad_norm": 1.9030020236968994,
      "learning_rate": 2.074438014406096e-06,
      "loss": 0.4675,
      "step": 273530
    },
    {
      "epoch": 111.97,
      "grad_norm": 2.018092393875122,
      "learning_rate": 2.0742804361856776e-06,
      "loss": 0.478,
      "step": 273540
    },
    {
      "epoch": 111.97,
      "grad_norm": 1.8706694841384888,
      "learning_rate": 2.0741228591996577e-06,
      "loss": 0.4524,
      "step": 273550
    },
    {
      "epoch": 111.98,
      "grad_norm": 2.2612321376800537,
      "learning_rate": 2.073965283448758e-06,
      "loss": 0.4557,
      "step": 273560
    },
    {
      "epoch": 111.98,
      "grad_norm": 1.7825912237167358,
      "learning_rate": 2.0738077089337e-06,
      "loss": 0.4529,
      "step": 273570
    },
    {
      "epoch": 111.99,
      "grad_norm": 1.6113582849502563,
      "learning_rate": 2.0736501356552065e-06,
      "loss": 0.4613,
      "step": 273580
    },
    {
      "epoch": 111.99,
      "grad_norm": 1.897264003753662,
      "learning_rate": 2.073492563613998e-06,
      "loss": 0.4526,
      "step": 273590
    },
    {
      "epoch": 111.99,
      "grad_norm": 2.3145928382873535,
      "learning_rate": 2.0733349928107973e-06,
      "loss": 0.4675,
      "step": 273600
    },
    {
      "epoch": 112.0,
      "grad_norm": 1.8504477739334106,
      "learning_rate": 2.0731774232463255e-06,
      "loss": 0.4722,
      "step": 273610
    },
    {
      "epoch": 112.0,
      "eval_loss": 0.46790140867233276,
      "eval_runtime": 52.2709,
      "eval_samples_per_second": 65.983,
      "eval_steps_per_second": 8.265,
      "step": 273616
    },
    {
      "epoch": 112.0,
      "grad_norm": 1.5975885391235352,
      "learning_rate": 2.0730198549213047e-06,
      "loss": 0.4572,
      "step": 273620
    },
    {
      "epoch": 112.01,
      "grad_norm": 2.109713554382324,
      "learning_rate": 2.072862287836457e-06,
      "loss": 0.4621,
      "step": 273630
    },
    {
      "epoch": 112.01,
      "grad_norm": 1.9692479372024536,
      "learning_rate": 2.0727047219925042e-06,
      "loss": 0.477,
      "step": 273640
    },
    {
      "epoch": 112.01,
      "grad_norm": 2.4479691982269287,
      "learning_rate": 2.0725471573901674e-06,
      "loss": 0.4394,
      "step": 273650
    },
    {
      "epoch": 112.02,
      "grad_norm": 1.7122434377670288,
      "learning_rate": 2.0723895940301688e-06,
      "loss": 0.4648,
      "step": 273660
    },
    {
      "epoch": 112.02,
      "grad_norm": 2.2174766063690186,
      "learning_rate": 2.072232031913231e-06,
      "loss": 0.4603,
      "step": 273670
    },
    {
      "epoch": 112.03,
      "grad_norm": 2.300632953643799,
      "learning_rate": 2.0720744710400734e-06,
      "loss": 0.4843,
      "step": 273680
    },
    {
      "epoch": 112.03,
      "grad_norm": 2.363739013671875,
      "learning_rate": 2.0719169114114205e-06,
      "loss": 0.4673,
      "step": 273690
    },
    {
      "epoch": 112.03,
      "grad_norm": 1.67686128616333,
      "learning_rate": 2.0717593530279924e-06,
      "loss": 0.4703,
      "step": 273700
    },
    {
      "epoch": 112.04,
      "grad_norm": 1.9401589632034302,
      "learning_rate": 2.0716017958905116e-06,
      "loss": 0.4769,
      "step": 273710
    },
    {
      "epoch": 112.04,
      "grad_norm": 1.8767850399017334,
      "learning_rate": 2.0714442399997003e-06,
      "loss": 0.4615,
      "step": 273720
    },
    {
      "epoch": 112.05,
      "grad_norm": 1.8202553987503052,
      "learning_rate": 2.071286685356279e-06,
      "loss": 0.4799,
      "step": 273730
    },
    {
      "epoch": 112.05,
      "grad_norm": 1.8671525716781616,
      "learning_rate": 2.0711291319609693e-06,
      "loss": 0.4636,
      "step": 273740
    },
    {
      "epoch": 112.05,
      "grad_norm": 1.9140926599502563,
      "learning_rate": 2.0709715798144943e-06,
      "loss": 0.4586,
      "step": 273750
    },
    {
      "epoch": 112.06,
      "grad_norm": 2.046776294708252,
      "learning_rate": 2.0708140289175745e-06,
      "loss": 0.4919,
      "step": 273760
    },
    {
      "epoch": 112.06,
      "grad_norm": 2.346933603286743,
      "learning_rate": 2.0706564792709325e-06,
      "loss": 0.497,
      "step": 273770
    },
    {
      "epoch": 112.07,
      "grad_norm": 2.4634697437286377,
      "learning_rate": 2.070498930875289e-06,
      "loss": 0.469,
      "step": 273780
    },
    {
      "epoch": 112.07,
      "grad_norm": 2.187270164489746,
      "learning_rate": 2.070341383731367e-06,
      "loss": 0.4524,
      "step": 273790
    },
    {
      "epoch": 112.08,
      "grad_norm": 2.2085771560668945,
      "learning_rate": 2.070183837839887e-06,
      "loss": 0.483,
      "step": 273800
    },
    {
      "epoch": 112.08,
      "grad_norm": 1.8852012157440186,
      "learning_rate": 2.0700262932015723e-06,
      "loss": 0.4648,
      "step": 273810
    },
    {
      "epoch": 112.08,
      "grad_norm": 1.6452511548995972,
      "learning_rate": 2.0698687498171427e-06,
      "loss": 0.4512,
      "step": 273820
    },
    {
      "epoch": 112.09,
      "grad_norm": 2.0693886280059814,
      "learning_rate": 2.0697112076873206e-06,
      "loss": 0.468,
      "step": 273830
    },
    {
      "epoch": 112.09,
      "grad_norm": 1.984175682067871,
      "learning_rate": 2.069553666812828e-06,
      "loss": 0.4611,
      "step": 273840
    },
    {
      "epoch": 112.1,
      "grad_norm": 2.1187918186187744,
      "learning_rate": 2.069396127194386e-06,
      "loss": 0.4837,
      "step": 273850
    },
    {
      "epoch": 112.1,
      "grad_norm": 1.8475252389907837,
      "learning_rate": 2.069238588832717e-06,
      "loss": 0.4725,
      "step": 273860
    },
    {
      "epoch": 112.1,
      "grad_norm": 1.7793961763381958,
      "learning_rate": 2.0690810517285417e-06,
      "loss": 0.4432,
      "step": 273870
    },
    {
      "epoch": 112.11,
      "grad_norm": 2.074077606201172,
      "learning_rate": 2.0689235158825826e-06,
      "loss": 0.4828,
      "step": 273880
    },
    {
      "epoch": 112.11,
      "grad_norm": 1.7402478456497192,
      "learning_rate": 2.0687659812955613e-06,
      "loss": 0.4726,
      "step": 273890
    },
    {
      "epoch": 112.12,
      "grad_norm": 1.6633204221725464,
      "learning_rate": 2.068608447968199e-06,
      "loss": 0.4661,
      "step": 273900
    },
    {
      "epoch": 112.12,
      "grad_norm": 1.8933320045471191,
      "learning_rate": 2.0684509159012173e-06,
      "loss": 0.4844,
      "step": 273910
    },
    {
      "epoch": 112.12,
      "grad_norm": 2.006504535675049,
      "learning_rate": 2.068293385095339e-06,
      "loss": 0.4811,
      "step": 273920
    },
    {
      "epoch": 112.13,
      "grad_norm": 2.1890859603881836,
      "learning_rate": 2.068135855551284e-06,
      "loss": 0.4918,
      "step": 273930
    },
    {
      "epoch": 112.13,
      "grad_norm": 1.9887129068374634,
      "learning_rate": 2.067978327269775e-06,
      "loss": 0.4695,
      "step": 273940
    },
    {
      "epoch": 112.14,
      "grad_norm": 1.7781962156295776,
      "learning_rate": 2.0678208002515336e-06,
      "loss": 0.4713,
      "step": 273950
    },
    {
      "epoch": 112.14,
      "grad_norm": 1.7722406387329102,
      "learning_rate": 2.0676632744972804e-06,
      "loss": 0.4771,
      "step": 273960
    },
    {
      "epoch": 112.14,
      "grad_norm": 1.793918490409851,
      "learning_rate": 2.067505750007738e-06,
      "loss": 0.4671,
      "step": 273970
    },
    {
      "epoch": 112.15,
      "grad_norm": 1.5543161630630493,
      "learning_rate": 2.067348226783628e-06,
      "loss": 0.4918,
      "step": 273980
    },
    {
      "epoch": 112.15,
      "grad_norm": 1.811888575553894,
      "learning_rate": 2.0671907048256717e-06,
      "loss": 0.4886,
      "step": 273990
    },
    {
      "epoch": 112.16,
      "grad_norm": 2.0124311447143555,
      "learning_rate": 2.0670331841345905e-06,
      "loss": 0.4369,
      "step": 274000
    },
    {
      "epoch": 112.16,
      "grad_norm": 2.466043710708618,
      "learning_rate": 2.066875664711106e-06,
      "loss": 0.4731,
      "step": 274010
    },
    {
      "epoch": 112.17,
      "grad_norm": 2.18645977973938,
      "learning_rate": 2.06671814655594e-06,
      "loss": 0.4767,
      "step": 274020
    },
    {
      "epoch": 112.17,
      "grad_norm": 2.175297260284424,
      "learning_rate": 2.066560629669815e-06,
      "loss": 0.4433,
      "step": 274030
    },
    {
      "epoch": 112.17,
      "grad_norm": 2.287585735321045,
      "learning_rate": 2.0664031140534506e-06,
      "loss": 0.4671,
      "step": 274040
    },
    {
      "epoch": 112.18,
      "grad_norm": 1.9083935022354126,
      "learning_rate": 2.0662455997075695e-06,
      "loss": 0.4816,
      "step": 274050
    },
    {
      "epoch": 112.18,
      "grad_norm": 2.2516889572143555,
      "learning_rate": 2.0660880866328918e-06,
      "loss": 0.4752,
      "step": 274060
    },
    {
      "epoch": 112.19,
      "grad_norm": 1.8761396408081055,
      "learning_rate": 2.065930574830142e-06,
      "loss": 0.4694,
      "step": 274070
    },
    {
      "epoch": 112.19,
      "grad_norm": 1.9998054504394531,
      "learning_rate": 2.0657730643000397e-06,
      "loss": 0.4844,
      "step": 274080
    },
    {
      "epoch": 112.19,
      "grad_norm": 1.3965373039245605,
      "learning_rate": 2.0656155550433066e-06,
      "loss": 0.4651,
      "step": 274090
    },
    {
      "epoch": 112.2,
      "grad_norm": 1.8924261331558228,
      "learning_rate": 2.065458047060664e-06,
      "loss": 0.4864,
      "step": 274100
    },
    {
      "epoch": 112.2,
      "grad_norm": 1.8089457750320435,
      "learning_rate": 2.0653005403528343e-06,
      "loss": 0.4484,
      "step": 274110
    },
    {
      "epoch": 112.21,
      "grad_norm": 1.76842200756073,
      "learning_rate": 2.065143034920538e-06,
      "loss": 0.4706,
      "step": 274120
    },
    {
      "epoch": 112.21,
      "grad_norm": 3.1487958431243896,
      "learning_rate": 2.064985530764497e-06,
      "loss": 0.464,
      "step": 274130
    },
    {
      "epoch": 112.21,
      "grad_norm": 2.5648293495178223,
      "learning_rate": 2.0648280278854334e-06,
      "loss": 0.4638,
      "step": 274140
    },
    {
      "epoch": 112.22,
      "grad_norm": 1.866781234741211,
      "learning_rate": 2.0646705262840677e-06,
      "loss": 0.4728,
      "step": 274150
    },
    {
      "epoch": 112.22,
      "grad_norm": 2.38358211517334,
      "learning_rate": 2.064513025961122e-06,
      "loss": 0.4464,
      "step": 274160
    },
    {
      "epoch": 112.23,
      "grad_norm": 2.4193034172058105,
      "learning_rate": 2.064355526917318e-06,
      "loss": 0.4757,
      "step": 274170
    },
    {
      "epoch": 112.23,
      "grad_norm": 1.8481563329696655,
      "learning_rate": 2.0641980291533765e-06,
      "loss": 0.4852,
      "step": 274180
    },
    {
      "epoch": 112.23,
      "grad_norm": 2.029491901397705,
      "learning_rate": 2.0640405326700193e-06,
      "loss": 0.4538,
      "step": 274190
    },
    {
      "epoch": 112.24,
      "grad_norm": 1.420746088027954,
      "learning_rate": 2.0638830374679675e-06,
      "loss": 0.4761,
      "step": 274200
    },
    {
      "epoch": 112.24,
      "grad_norm": 1.595875859260559,
      "learning_rate": 2.0637255435479433e-06,
      "loss": 0.4563,
      "step": 274210
    },
    {
      "epoch": 112.25,
      "grad_norm": 1.903302550315857,
      "learning_rate": 2.0635680509106675e-06,
      "loss": 0.4584,
      "step": 274220
    },
    {
      "epoch": 112.25,
      "grad_norm": 2.2775163650512695,
      "learning_rate": 2.063410559556862e-06,
      "loss": 0.47,
      "step": 274230
    },
    {
      "epoch": 112.26,
      "grad_norm": 1.794315218925476,
      "learning_rate": 2.063253069487248e-06,
      "loss": 0.4413,
      "step": 274240
    },
    {
      "epoch": 112.26,
      "grad_norm": 2.1370372772216797,
      "learning_rate": 2.0630955807025463e-06,
      "loss": 0.4619,
      "step": 274250
    },
    {
      "epoch": 112.26,
      "grad_norm": 1.7062479257583618,
      "learning_rate": 2.06293809320348e-06,
      "loss": 0.4679,
      "step": 274260
    },
    {
      "epoch": 112.27,
      "grad_norm": 1.712030053138733,
      "learning_rate": 2.062780606990769e-06,
      "loss": 0.4825,
      "step": 274270
    },
    {
      "epoch": 112.27,
      "grad_norm": 1.972316861152649,
      "learning_rate": 2.0626231220651357e-06,
      "loss": 0.4846,
      "step": 274280
    },
    {
      "epoch": 112.28,
      "grad_norm": 1.9917309284210205,
      "learning_rate": 2.062465638427301e-06,
      "loss": 0.4595,
      "step": 274290
    },
    {
      "epoch": 112.28,
      "grad_norm": 1.7505183219909668,
      "learning_rate": 2.062308156077987e-06,
      "loss": 0.4574,
      "step": 274300
    },
    {
      "epoch": 112.28,
      "grad_norm": 3.2919199466705322,
      "learning_rate": 2.0621506750179137e-06,
      "loss": 0.4925,
      "step": 274310
    },
    {
      "epoch": 112.29,
      "grad_norm": 3.32092022895813,
      "learning_rate": 2.061993195247803e-06,
      "loss": 0.4765,
      "step": 274320
    },
    {
      "epoch": 112.29,
      "grad_norm": 1.6094040870666504,
      "learning_rate": 2.0618357167683773e-06,
      "loss": 0.4623,
      "step": 274330
    },
    {
      "epoch": 112.3,
      "grad_norm": 1.5979353189468384,
      "learning_rate": 2.061678239580357e-06,
      "loss": 0.4671,
      "step": 274340
    },
    {
      "epoch": 112.3,
      "grad_norm": 1.779617190361023,
      "learning_rate": 2.0615207636844637e-06,
      "loss": 0.4532,
      "step": 274350
    },
    {
      "epoch": 112.3,
      "grad_norm": 2.1058828830718994,
      "learning_rate": 2.061363289081419e-06,
      "loss": 0.4667,
      "step": 274360
    },
    {
      "epoch": 112.31,
      "grad_norm": 2.0813403129577637,
      "learning_rate": 2.0612058157719438e-06,
      "loss": 0.4616,
      "step": 274370
    },
    {
      "epoch": 112.31,
      "grad_norm": 2.3349108695983887,
      "learning_rate": 2.06104834375676e-06,
      "loss": 0.469,
      "step": 274380
    },
    {
      "epoch": 112.32,
      "grad_norm": 2.1368672847747803,
      "learning_rate": 2.060890873036589e-06,
      "loss": 0.4571,
      "step": 274390
    },
    {
      "epoch": 112.32,
      "grad_norm": 2.8026392459869385,
      "learning_rate": 2.0607334036121517e-06,
      "loss": 0.4837,
      "step": 274400
    },
    {
      "epoch": 112.33,
      "grad_norm": 2.0332577228546143,
      "learning_rate": 2.0605759354841694e-06,
      "loss": 0.4643,
      "step": 274410
    },
    {
      "epoch": 112.33,
      "grad_norm": 1.5409908294677734,
      "learning_rate": 2.0604184686533634e-06,
      "loss": 0.4524,
      "step": 274420
    },
    {
      "epoch": 112.33,
      "grad_norm": 2.20133900642395,
      "learning_rate": 2.0602610031204554e-06,
      "loss": 0.4856,
      "step": 274430
    },
    {
      "epoch": 112.34,
      "grad_norm": 1.8830872774124146,
      "learning_rate": 2.0601035388861666e-06,
      "loss": 0.4654,
      "step": 274440
    },
    {
      "epoch": 112.34,
      "grad_norm": 1.7012038230895996,
      "learning_rate": 2.0599460759512183e-06,
      "loss": 0.4469,
      "step": 274450
    },
    {
      "epoch": 112.35,
      "grad_norm": 2.1974575519561768,
      "learning_rate": 2.059788614316332e-06,
      "loss": 0.4619,
      "step": 274460
    },
    {
      "epoch": 112.35,
      "grad_norm": 1.6268539428710938,
      "learning_rate": 2.0596311539822283e-06,
      "loss": 0.4759,
      "step": 274470
    },
    {
      "epoch": 112.35,
      "grad_norm": 2.1465952396392822,
      "learning_rate": 2.0594736949496294e-06,
      "loss": 0.4858,
      "step": 274480
    },
    {
      "epoch": 112.36,
      "grad_norm": 2.2229576110839844,
      "learning_rate": 2.0593162372192564e-06,
      "loss": 0.4813,
      "step": 274490
    },
    {
      "epoch": 112.36,
      "grad_norm": 1.6112685203552246,
      "learning_rate": 2.0591587807918302e-06,
      "loss": 0.4686,
      "step": 274500
    },
    {
      "epoch": 112.37,
      "grad_norm": 1.9428155422210693,
      "learning_rate": 2.0590013256680724e-06,
      "loss": 0.4667,
      "step": 274510
    },
    {
      "epoch": 112.37,
      "grad_norm": 1.8561843633651733,
      "learning_rate": 2.058843871848705e-06,
      "loss": 0.4784,
      "step": 274520
    },
    {
      "epoch": 112.37,
      "grad_norm": 1.6092116832733154,
      "learning_rate": 2.0586864193344474e-06,
      "loss": 0.4661,
      "step": 274530
    },
    {
      "epoch": 112.38,
      "grad_norm": 1.6917238235473633,
      "learning_rate": 2.0585289681260223e-06,
      "loss": 0.4362,
      "step": 274540
    },
    {
      "epoch": 112.38,
      "grad_norm": 1.7880078554153442,
      "learning_rate": 2.05837151822415e-06,
      "loss": 0.4672,
      "step": 274550
    },
    {
      "epoch": 112.39,
      "grad_norm": 1.6295578479766846,
      "learning_rate": 2.058214069629553e-06,
      "loss": 0.4586,
      "step": 274560
    },
    {
      "epoch": 112.39,
      "grad_norm": 1.9418225288391113,
      "learning_rate": 2.0580566223429516e-06,
      "loss": 0.4577,
      "step": 274570
    },
    {
      "epoch": 112.39,
      "grad_norm": 1.9987331628799438,
      "learning_rate": 2.0578991763650676e-06,
      "loss": 0.4576,
      "step": 274580
    },
    {
      "epoch": 112.4,
      "grad_norm": 2.0688838958740234,
      "learning_rate": 2.0577417316966216e-06,
      "loss": 0.4813,
      "step": 274590
    },
    {
      "epoch": 112.4,
      "grad_norm": 1.9573227167129517,
      "learning_rate": 2.057584288338335e-06,
      "loss": 0.462,
      "step": 274600
    },
    {
      "epoch": 112.41,
      "grad_norm": 1.9901047945022583,
      "learning_rate": 2.05742684629093e-06,
      "loss": 0.4989,
      "step": 274610
    },
    {
      "epoch": 112.41,
      "grad_norm": 1.8263423442840576,
      "learning_rate": 2.057269405555127e-06,
      "loss": 0.4814,
      "step": 274620
    },
    {
      "epoch": 112.42,
      "grad_norm": 1.631048321723938,
      "learning_rate": 2.057111966131645e-06,
      "loss": 0.4846,
      "step": 274630
    },
    {
      "epoch": 112.42,
      "grad_norm": 2.178847551345825,
      "learning_rate": 2.0569545280212095e-06,
      "loss": 0.4695,
      "step": 274640
    },
    {
      "epoch": 112.42,
      "grad_norm": 1.964589238166809,
      "learning_rate": 2.0567970912245396e-06,
      "loss": 0.4732,
      "step": 274650
    },
    {
      "epoch": 112.43,
      "grad_norm": 1.7728153467178345,
      "learning_rate": 2.0566396557423565e-06,
      "loss": 0.4665,
      "step": 274660
    },
    {
      "epoch": 112.43,
      "grad_norm": 1.9573777914047241,
      "learning_rate": 2.0564822215753813e-06,
      "loss": 0.4846,
      "step": 274670
    },
    {
      "epoch": 112.44,
      "grad_norm": 2.363065242767334,
      "learning_rate": 2.056324788724335e-06,
      "loss": 0.4799,
      "step": 274680
    },
    {
      "epoch": 112.44,
      "grad_norm": 1.9326610565185547,
      "learning_rate": 2.0561673571899395e-06,
      "loss": 0.4589,
      "step": 274690
    },
    {
      "epoch": 112.44,
      "grad_norm": 1.5648870468139648,
      "learning_rate": 2.0560099269729153e-06,
      "loss": 0.4817,
      "step": 274700
    },
    {
      "epoch": 112.45,
      "grad_norm": 1.7778347730636597,
      "learning_rate": 2.055852498073984e-06,
      "loss": 0.4648,
      "step": 274710
    },
    {
      "epoch": 112.45,
      "grad_norm": 2.3098182678222656,
      "learning_rate": 2.055695070493866e-06,
      "loss": 0.4598,
      "step": 274720
    },
    {
      "epoch": 112.46,
      "grad_norm": 2.203850269317627,
      "learning_rate": 2.0555376442332837e-06,
      "loss": 0.4705,
      "step": 274730
    },
    {
      "epoch": 112.46,
      "grad_norm": 1.8635494709014893,
      "learning_rate": 2.055380219292957e-06,
      "loss": 0.4962,
      "step": 274740
    },
    {
      "epoch": 112.46,
      "grad_norm": 1.7869421243667603,
      "learning_rate": 2.0552227956736085e-06,
      "loss": 0.4701,
      "step": 274750
    },
    {
      "epoch": 112.47,
      "grad_norm": 2.0288619995117188,
      "learning_rate": 2.055065373375958e-06,
      "loss": 0.4689,
      "step": 274760
    },
    {
      "epoch": 112.47,
      "grad_norm": 1.803684949874878,
      "learning_rate": 2.054907952400727e-06,
      "loss": 0.4565,
      "step": 274770
    },
    {
      "epoch": 112.48,
      "grad_norm": 2.045732021331787,
      "learning_rate": 2.0547505327486364e-06,
      "loss": 0.4528,
      "step": 274780
    },
    {
      "epoch": 112.48,
      "grad_norm": 1.5385857820510864,
      "learning_rate": 2.054593114420408e-06,
      "loss": 0.4655,
      "step": 274790
    },
    {
      "epoch": 112.48,
      "grad_norm": 2.270665407180786,
      "learning_rate": 2.054435697416762e-06,
      "loss": 0.4701,
      "step": 274800
    },
    {
      "epoch": 112.49,
      "grad_norm": 2.118138313293457,
      "learning_rate": 2.0542782817384207e-06,
      "loss": 0.4691,
      "step": 274810
    },
    {
      "epoch": 112.49,
      "grad_norm": 1.8922628164291382,
      "learning_rate": 2.0541208673861034e-06,
      "loss": 0.4558,
      "step": 274820
    },
    {
      "epoch": 112.5,
      "grad_norm": 1.6789153814315796,
      "learning_rate": 2.053963454360533e-06,
      "loss": 0.4647,
      "step": 274830
    },
    {
      "epoch": 112.5,
      "grad_norm": 1.7903637886047363,
      "learning_rate": 2.05380604266243e-06,
      "loss": 0.4674,
      "step": 274840
    },
    {
      "epoch": 112.51,
      "grad_norm": 1.9304848909378052,
      "learning_rate": 2.053648632292515e-06,
      "loss": 0.457,
      "step": 274850
    },
    {
      "epoch": 112.51,
      "grad_norm": 2.1763174533843994,
      "learning_rate": 2.05349122325151e-06,
      "loss": 0.4525,
      "step": 274860
    },
    {
      "epoch": 112.51,
      "grad_norm": 2.1824872493743896,
      "learning_rate": 2.053333815540135e-06,
      "loss": 0.4666,
      "step": 274870
    },
    {
      "epoch": 112.52,
      "grad_norm": 1.7239766120910645,
      "learning_rate": 2.0531764091591126e-06,
      "loss": 0.4676,
      "step": 274880
    },
    {
      "epoch": 112.52,
      "grad_norm": 1.749533772468567,
      "learning_rate": 2.0530190041091618e-06,
      "loss": 0.49,
      "step": 274890
    },
    {
      "epoch": 112.53,
      "grad_norm": 1.997298240661621,
      "learning_rate": 2.052861600391005e-06,
      "loss": 0.4633,
      "step": 274900
    },
    {
      "epoch": 112.53,
      "grad_norm": 1.7352569103240967,
      "learning_rate": 2.0527041980053626e-06,
      "loss": 0.4358,
      "step": 274910
    },
    {
      "epoch": 112.53,
      "grad_norm": 1.9669559001922607,
      "learning_rate": 2.0525467969529563e-06,
      "loss": 0.4898,
      "step": 274920
    },
    {
      "epoch": 112.54,
      "grad_norm": 2.065906047821045,
      "learning_rate": 2.0523893972345066e-06,
      "loss": 0.4707,
      "step": 274930
    },
    {
      "epoch": 112.54,
      "grad_norm": 2.2650606632232666,
      "learning_rate": 2.0522319988507347e-06,
      "loss": 0.4627,
      "step": 274940
    },
    {
      "epoch": 112.55,
      "grad_norm": 1.5637668371200562,
      "learning_rate": 2.0520746018023616e-06,
      "loss": 0.4706,
      "step": 274950
    },
    {
      "epoch": 112.55,
      "grad_norm": 1.872825264930725,
      "learning_rate": 2.0519172060901084e-06,
      "loss": 0.4736,
      "step": 274960
    },
    {
      "epoch": 112.55,
      "grad_norm": 1.9920532703399658,
      "learning_rate": 2.0517598117146963e-06,
      "loss": 0.4572,
      "step": 274970
    },
    {
      "epoch": 112.56,
      "grad_norm": 2.3539795875549316,
      "learning_rate": 2.051602418676846e-06,
      "loss": 0.4732,
      "step": 274980
    },
    {
      "epoch": 112.56,
      "grad_norm": 1.9752267599105835,
      "learning_rate": 2.051445026977278e-06,
      "loss": 0.4619,
      "step": 274990
    },
    {
      "epoch": 112.57,
      "grad_norm": 1.9432573318481445,
      "learning_rate": 2.051287636616713e-06,
      "loss": 0.4798,
      "step": 275000
    },
    {
      "epoch": 112.57,
      "grad_norm": 2.26840877532959,
      "learning_rate": 2.0511302475958736e-06,
      "loss": 0.4696,
      "step": 275010
    },
    {
      "epoch": 112.57,
      "grad_norm": 1.9546031951904297,
      "learning_rate": 2.0509728599154807e-06,
      "loss": 0.4715,
      "step": 275020
    },
    {
      "epoch": 112.58,
      "grad_norm": 1.7975424528121948,
      "learning_rate": 2.0508154735762543e-06,
      "loss": 0.4727,
      "step": 275030
    },
    {
      "epoch": 112.58,
      "grad_norm": 2.5175602436065674,
      "learning_rate": 2.0506580885789154e-06,
      "loss": 0.4704,
      "step": 275040
    },
    {
      "epoch": 112.59,
      "grad_norm": 1.7833995819091797,
      "learning_rate": 2.050500704924185e-06,
      "loss": 0.4572,
      "step": 275050
    },
    {
      "epoch": 112.59,
      "grad_norm": 2.028027296066284,
      "learning_rate": 2.050343322612784e-06,
      "loss": 0.4672,
      "step": 275060
    },
    {
      "epoch": 112.6,
      "grad_norm": 1.7295589447021484,
      "learning_rate": 2.050185941645434e-06,
      "loss": 0.4517,
      "step": 275070
    },
    {
      "epoch": 112.6,
      "grad_norm": 2.0992956161499023,
      "learning_rate": 2.0500285620228554e-06,
      "loss": 0.4795,
      "step": 275080
    },
    {
      "epoch": 112.6,
      "grad_norm": 2.0030453205108643,
      "learning_rate": 2.049871183745769e-06,
      "loss": 0.4448,
      "step": 275090
    },
    {
      "epoch": 112.61,
      "grad_norm": 1.8667945861816406,
      "learning_rate": 2.0497138068148965e-06,
      "loss": 0.4851,
      "step": 275100
    },
    {
      "epoch": 112.61,
      "grad_norm": 1.9246965646743774,
      "learning_rate": 2.049556431230958e-06,
      "loss": 0.4751,
      "step": 275110
    },
    {
      "epoch": 112.62,
      "grad_norm": 1.8562294244766235,
      "learning_rate": 2.049399056994675e-06,
      "loss": 0.4663,
      "step": 275120
    },
    {
      "epoch": 112.62,
      "grad_norm": 2.6670596599578857,
      "learning_rate": 2.0492416841067675e-06,
      "loss": 0.4536,
      "step": 275130
    },
    {
      "epoch": 112.62,
      "grad_norm": 1.8294769525527954,
      "learning_rate": 2.049084312567957e-06,
      "loss": 0.4573,
      "step": 275140
    },
    {
      "epoch": 112.63,
      "grad_norm": 1.6500622034072876,
      "learning_rate": 2.0489269423789643e-06,
      "loss": 0.4575,
      "step": 275150
    },
    {
      "epoch": 112.63,
      "grad_norm": 1.6046535968780518,
      "learning_rate": 2.048769573540511e-06,
      "loss": 0.4764,
      "step": 275160
    },
    {
      "epoch": 112.64,
      "grad_norm": 1.5525963306427002,
      "learning_rate": 2.0486122060533167e-06,
      "loss": 0.4531,
      "step": 275170
    },
    {
      "epoch": 112.64,
      "grad_norm": 1.9256335496902466,
      "learning_rate": 2.048454839918103e-06,
      "loss": 0.4708,
      "step": 275180
    },
    {
      "epoch": 112.64,
      "grad_norm": 1.982745885848999,
      "learning_rate": 2.0482974751355903e-06,
      "loss": 0.4719,
      "step": 275190
    },
    {
      "epoch": 112.65,
      "grad_norm": 1.8364568948745728,
      "learning_rate": 2.0481401117065007e-06,
      "loss": 0.4726,
      "step": 275200
    },
    {
      "epoch": 112.65,
      "grad_norm": 2.5583901405334473,
      "learning_rate": 2.047982749631554e-06,
      "loss": 0.4793,
      "step": 275210
    },
    {
      "epoch": 112.66,
      "grad_norm": 2.0657708644866943,
      "learning_rate": 2.047825388911471e-06,
      "loss": 0.4687,
      "step": 275220
    },
    {
      "epoch": 112.66,
      "grad_norm": 1.8309458494186401,
      "learning_rate": 2.047668029546973e-06,
      "loss": 0.4352,
      "step": 275230
    },
    {
      "epoch": 112.66,
      "grad_norm": 1.854027271270752,
      "learning_rate": 2.047510671538781e-06,
      "loss": 0.4782,
      "step": 275240
    },
    {
      "epoch": 112.67,
      "grad_norm": 2.293368101119995,
      "learning_rate": 2.047353314887615e-06,
      "loss": 0.4605,
      "step": 275250
    },
    {
      "epoch": 112.67,
      "grad_norm": 1.8771629333496094,
      "learning_rate": 2.047195959594197e-06,
      "loss": 0.4626,
      "step": 275260
    },
    {
      "epoch": 112.68,
      "grad_norm": 1.6980482339859009,
      "learning_rate": 2.047038605659246e-06,
      "loss": 0.4596,
      "step": 275270
    },
    {
      "epoch": 112.68,
      "grad_norm": 1.8107119798660278,
      "learning_rate": 2.0468812530834847e-06,
      "loss": 0.462,
      "step": 275280
    },
    {
      "epoch": 112.69,
      "grad_norm": 2.1196062564849854,
      "learning_rate": 2.0467239018676332e-06,
      "loss": 0.4694,
      "step": 275290
    },
    {
      "epoch": 112.69,
      "grad_norm": 2.2090649604797363,
      "learning_rate": 2.046566552012412e-06,
      "loss": 0.4527,
      "step": 275300
    },
    {
      "epoch": 112.69,
      "grad_norm": 2.36543345451355,
      "learning_rate": 2.046409203518542e-06,
      "loss": 0.4705,
      "step": 275310
    },
    {
      "epoch": 112.7,
      "grad_norm": 1.803168535232544,
      "learning_rate": 2.0462518563867443e-06,
      "loss": 0.4717,
      "step": 275320
    },
    {
      "epoch": 112.7,
      "grad_norm": 2.465050458908081,
      "learning_rate": 2.04609451061774e-06,
      "loss": 0.4916,
      "step": 275330
    },
    {
      "epoch": 112.71,
      "grad_norm": 2.132312059402466,
      "learning_rate": 2.045937166212249e-06,
      "loss": 0.4572,
      "step": 275340
    },
    {
      "epoch": 112.71,
      "grad_norm": 1.7290922403335571,
      "learning_rate": 2.0457798231709924e-06,
      "loss": 0.4529,
      "step": 275350
    },
    {
      "epoch": 112.71,
      "grad_norm": 1.7185404300689697,
      "learning_rate": 2.045622481494691e-06,
      "loss": 0.4674,
      "step": 275360
    },
    {
      "epoch": 112.72,
      "grad_norm": 2.142367362976074,
      "learning_rate": 2.0454651411840656e-06,
      "loss": 0.4418,
      "step": 275370
    },
    {
      "epoch": 112.72,
      "grad_norm": 2.1507434844970703,
      "learning_rate": 2.0453078022398364e-06,
      "loss": 0.4688,
      "step": 275380
    },
    {
      "epoch": 112.73,
      "grad_norm": 2.5449557304382324,
      "learning_rate": 2.045150464662725e-06,
      "loss": 0.4735,
      "step": 275390
    },
    {
      "epoch": 112.73,
      "grad_norm": 2.6735000610351562,
      "learning_rate": 2.0449931284534527e-06,
      "loss": 0.4642,
      "step": 275400
    },
    {
      "epoch": 112.73,
      "grad_norm": 1.994859218597412,
      "learning_rate": 2.0448357936127386e-06,
      "loss": 0.468,
      "step": 275410
    },
    {
      "epoch": 112.74,
      "grad_norm": 1.5598599910736084,
      "learning_rate": 2.0446784601413045e-06,
      "loss": 0.4669,
      "step": 275420
    },
    {
      "epoch": 112.74,
      "grad_norm": 1.8934378623962402,
      "learning_rate": 2.0445211280398703e-06,
      "loss": 0.4782,
      "step": 275430
    },
    {
      "epoch": 112.75,
      "grad_norm": 1.9785209894180298,
      "learning_rate": 2.0443637973091574e-06,
      "loss": 0.4881,
      "step": 275440
    },
    {
      "epoch": 112.75,
      "grad_norm": 2.1154274940490723,
      "learning_rate": 2.0442064679498866e-06,
      "loss": 0.4593,
      "step": 275450
    },
    {
      "epoch": 112.75,
      "grad_norm": 2.318892240524292,
      "learning_rate": 2.044049139962778e-06,
      "loss": 0.4593,
      "step": 275460
    },
    {
      "epoch": 112.76,
      "grad_norm": 1.8995399475097656,
      "learning_rate": 2.0438918133485535e-06,
      "loss": 0.4741,
      "step": 275470
    },
    {
      "epoch": 112.76,
      "grad_norm": 1.9484977722167969,
      "learning_rate": 2.0437344881079325e-06,
      "loss": 0.4871,
      "step": 275480
    },
    {
      "epoch": 112.77,
      "grad_norm": 2.181309461593628,
      "learning_rate": 2.043577164241636e-06,
      "loss": 0.4702,
      "step": 275490
    },
    {
      "epoch": 112.77,
      "grad_norm": 1.6436084508895874,
      "learning_rate": 2.0434198417503844e-06,
      "loss": 0.4679,
      "step": 275500
    },
    {
      "epoch": 112.78,
      "grad_norm": 2.189079999923706,
      "learning_rate": 2.043262520634899e-06,
      "loss": 0.4467,
      "step": 275510
    },
    {
      "epoch": 112.78,
      "grad_norm": 1.6833879947662354,
      "learning_rate": 2.0431052008959e-06,
      "loss": 0.4374,
      "step": 275520
    },
    {
      "epoch": 112.78,
      "grad_norm": 1.9328762292861938,
      "learning_rate": 2.0429478825341087e-06,
      "loss": 0.4828,
      "step": 275530
    },
    {
      "epoch": 112.79,
      "grad_norm": 1.9869394302368164,
      "learning_rate": 2.042790565550245e-06,
      "loss": 0.4717,
      "step": 275540
    },
    {
      "epoch": 112.79,
      "grad_norm": 2.095125913619995,
      "learning_rate": 2.0426332499450304e-06,
      "loss": 0.4852,
      "step": 275550
    },
    {
      "epoch": 112.8,
      "grad_norm": 2.189540147781372,
      "learning_rate": 2.0424759357191847e-06,
      "loss": 0.4768,
      "step": 275560
    },
    {
      "epoch": 112.8,
      "grad_norm": 1.7436580657958984,
      "learning_rate": 2.0423186228734277e-06,
      "loss": 0.4815,
      "step": 275570
    },
    {
      "epoch": 112.8,
      "grad_norm": 1.7917263507843018,
      "learning_rate": 2.042161311408482e-06,
      "loss": 0.4692,
      "step": 275580
    },
    {
      "epoch": 112.81,
      "grad_norm": 2.1005749702453613,
      "learning_rate": 2.0420040013250676e-06,
      "loss": 0.4686,
      "step": 275590
    },
    {
      "epoch": 112.81,
      "grad_norm": 1.8555561304092407,
      "learning_rate": 2.041846692623905e-06,
      "loss": 0.4496,
      "step": 275600
    },
    {
      "epoch": 112.82,
      "grad_norm": 1.6152079105377197,
      "learning_rate": 2.0416893853057143e-06,
      "loss": 0.4409,
      "step": 275610
    },
    {
      "epoch": 112.82,
      "grad_norm": 1.5973442792892456,
      "learning_rate": 2.0415320793712167e-06,
      "loss": 0.4638,
      "step": 275620
    },
    {
      "epoch": 112.82,
      "grad_norm": 2.211796998977661,
      "learning_rate": 2.041374774821133e-06,
      "loss": 0.4664,
      "step": 275630
    },
    {
      "epoch": 112.83,
      "grad_norm": 2.0620789527893066,
      "learning_rate": 2.0412174716561824e-06,
      "loss": 0.4769,
      "step": 275640
    },
    {
      "epoch": 112.83,
      "grad_norm": 2.6889781951904297,
      "learning_rate": 2.0410601698770874e-06,
      "loss": 0.4699,
      "step": 275650
    },
    {
      "epoch": 112.84,
      "grad_norm": 2.750750780105591,
      "learning_rate": 2.040902869484567e-06,
      "loss": 0.4615,
      "step": 275660
    },
    {
      "epoch": 112.84,
      "grad_norm": 1.6403156518936157,
      "learning_rate": 2.0407455704793424e-06,
      "loss": 0.4608,
      "step": 275670
    },
    {
      "epoch": 112.84,
      "grad_norm": 1.659595251083374,
      "learning_rate": 2.0405882728621345e-06,
      "loss": 0.4624,
      "step": 275680
    },
    {
      "epoch": 112.85,
      "grad_norm": 1.6931486129760742,
      "learning_rate": 2.040430976633664e-06,
      "loss": 0.4691,
      "step": 275690
    },
    {
      "epoch": 112.85,
      "grad_norm": 1.7630786895751953,
      "learning_rate": 2.0402736817946506e-06,
      "loss": 0.4643,
      "step": 275700
    },
    {
      "epoch": 112.86,
      "grad_norm": 1.6613754034042358,
      "learning_rate": 2.040116388345815e-06,
      "loss": 0.4503,
      "step": 275710
    },
    {
      "epoch": 112.86,
      "grad_norm": 1.7681864500045776,
      "learning_rate": 2.039959096287878e-06,
      "loss": 0.4595,
      "step": 275720
    },
    {
      "epoch": 112.87,
      "grad_norm": 2.3026514053344727,
      "learning_rate": 2.03980180562156e-06,
      "loss": 0.4443,
      "step": 275730
    },
    {
      "epoch": 112.87,
      "grad_norm": 1.7461835145950317,
      "learning_rate": 2.0396445163475817e-06,
      "loss": 0.4543,
      "step": 275740
    },
    {
      "epoch": 112.87,
      "grad_norm": 2.2658510208129883,
      "learning_rate": 2.039487228466663e-06,
      "loss": 0.4666,
      "step": 275750
    },
    {
      "epoch": 112.88,
      "grad_norm": 1.5329962968826294,
      "learning_rate": 2.039329941979525e-06,
      "loss": 0.4643,
      "step": 275760
    },
    {
      "epoch": 112.88,
      "grad_norm": 1.940185308456421,
      "learning_rate": 2.039172656886889e-06,
      "loss": 0.4387,
      "step": 275770
    },
    {
      "epoch": 112.89,
      "grad_norm": 2.3384830951690674,
      "learning_rate": 2.039015373189474e-06,
      "loss": 0.4621,
      "step": 275780
    },
    {
      "epoch": 112.89,
      "grad_norm": 1.9811186790466309,
      "learning_rate": 2.0388580908880013e-06,
      "loss": 0.4613,
      "step": 275790
    },
    {
      "epoch": 112.89,
      "grad_norm": 2.590656280517578,
      "learning_rate": 2.038700809983191e-06,
      "loss": 0.4729,
      "step": 275800
    },
    {
      "epoch": 112.9,
      "grad_norm": 2.1292500495910645,
      "learning_rate": 2.038543530475764e-06,
      "loss": 0.4693,
      "step": 275810
    },
    {
      "epoch": 112.9,
      "grad_norm": 1.6649494171142578,
      "learning_rate": 2.0383862523664408e-06,
      "loss": 0.4618,
      "step": 275820
    },
    {
      "epoch": 112.91,
      "grad_norm": 3.0204763412475586,
      "learning_rate": 2.038228975655942e-06,
      "loss": 0.4841,
      "step": 275830
    },
    {
      "epoch": 112.91,
      "grad_norm": 1.8416571617126465,
      "learning_rate": 2.038071700344987e-06,
      "loss": 0.4533,
      "step": 275840
    },
    {
      "epoch": 112.91,
      "grad_norm": 1.7573730945587158,
      "learning_rate": 2.037914426434297e-06,
      "loss": 0.4545,
      "step": 275850
    },
    {
      "epoch": 112.92,
      "grad_norm": 2.393156051635742,
      "learning_rate": 2.0377571539245925e-06,
      "loss": 0.4772,
      "step": 275860
    },
    {
      "epoch": 112.92,
      "grad_norm": 1.5060399770736694,
      "learning_rate": 2.037599882816594e-06,
      "loss": 0.4511,
      "step": 275870
    },
    {
      "epoch": 112.93,
      "grad_norm": 2.4518327713012695,
      "learning_rate": 2.0374426131110215e-06,
      "loss": 0.4689,
      "step": 275880
    },
    {
      "epoch": 112.93,
      "grad_norm": 1.6933746337890625,
      "learning_rate": 2.0372853448085956e-06,
      "loss": 0.461,
      "step": 275890
    },
    {
      "epoch": 112.93,
      "grad_norm": 1.7151203155517578,
      "learning_rate": 2.037128077910037e-06,
      "loss": 0.4779,
      "step": 275900
    },
    {
      "epoch": 112.94,
      "grad_norm": 1.7633394002914429,
      "learning_rate": 2.0369708124160665e-06,
      "loss": 0.4646,
      "step": 275910
    },
    {
      "epoch": 112.94,
      "grad_norm": 2.0676629543304443,
      "learning_rate": 2.0368135483274034e-06,
      "loss": 0.4879,
      "step": 275920
    },
    {
      "epoch": 112.95,
      "grad_norm": 2.017216682434082,
      "learning_rate": 2.0366562856447687e-06,
      "loss": 0.4536,
      "step": 275930
    },
    {
      "epoch": 112.95,
      "grad_norm": 1.764351487159729,
      "learning_rate": 2.036499024368883e-06,
      "loss": 0.4486,
      "step": 275940
    },
    {
      "epoch": 112.96,
      "grad_norm": 1.9187393188476562,
      "learning_rate": 2.036341764500465e-06,
      "loss": 0.4667,
      "step": 275950
    },
    {
      "epoch": 112.96,
      "grad_norm": 1.8658567667007446,
      "learning_rate": 2.0361845060402383e-06,
      "loss": 0.4618,
      "step": 275960
    },
    {
      "epoch": 112.96,
      "grad_norm": 2.261958599090576,
      "learning_rate": 2.0360272489889205e-06,
      "loss": 0.479,
      "step": 275970
    },
    {
      "epoch": 112.97,
      "grad_norm": 2.1578729152679443,
      "learning_rate": 2.0358699933472333e-06,
      "loss": 0.4595,
      "step": 275980
    },
    {
      "epoch": 112.97,
      "grad_norm": 2.144848585128784,
      "learning_rate": 2.035712739115897e-06,
      "loss": 0.4649,
      "step": 275990
    },
    {
      "epoch": 112.98,
      "grad_norm": 2.442396640777588,
      "learning_rate": 2.035555486295631e-06,
      "loss": 0.4503,
      "step": 276000
    },
    {
      "epoch": 112.98,
      "grad_norm": 3.0257821083068848,
      "learning_rate": 2.035398234887157e-06,
      "loss": 0.469,
      "step": 276010
    },
    {
      "epoch": 112.98,
      "grad_norm": 1.4243131875991821,
      "learning_rate": 2.0352409848911943e-06,
      "loss": 0.4384,
      "step": 276020
    },
    {
      "epoch": 112.99,
      "grad_norm": 1.9923875331878662,
      "learning_rate": 2.035083736308464e-06,
      "loss": 0.4575,
      "step": 276030
    },
    {
      "epoch": 112.99,
      "grad_norm": 1.5581337213516235,
      "learning_rate": 2.0349264891396855e-06,
      "loss": 0.4429,
      "step": 276040
    },
    {
      "epoch": 113.0,
      "grad_norm": 2.213961601257324,
      "learning_rate": 2.0347692433855807e-06,
      "loss": 0.4689,
      "step": 276050
    },
    {
      "epoch": 113.0,
      "eval_loss": 0.46925821900367737,
      "eval_runtime": 52.5028,
      "eval_samples_per_second": 65.692,
      "eval_steps_per_second": 8.228,
      "step": 276059
    },
    {
      "epoch": 113.0,
      "grad_norm": 1.7815977334976196,
      "learning_rate": 2.034611999046868e-06,
      "loss": 0.4722,
      "step": 276060
    },
    {
      "epoch": 113.0,
      "grad_norm": 1.974336862564087,
      "learning_rate": 2.034454756124269e-06,
      "loss": 0.4558,
      "step": 276070
    },
    {
      "epoch": 113.01,
      "grad_norm": 2.003185510635376,
      "learning_rate": 2.0342975146185037e-06,
      "loss": 0.4562,
      "step": 276080
    },
    {
      "epoch": 113.01,
      "grad_norm": 1.99407958984375,
      "learning_rate": 2.0341402745302917e-06,
      "loss": 0.4575,
      "step": 276090
    },
    {
      "epoch": 113.02,
      "grad_norm": 1.8480570316314697,
      "learning_rate": 2.033983035860354e-06,
      "loss": 0.479,
      "step": 276100
    },
    {
      "epoch": 113.02,
      "grad_norm": 2.4455816745758057,
      "learning_rate": 2.033825798609411e-06,
      "loss": 0.4674,
      "step": 276110
    },
    {
      "epoch": 113.02,
      "grad_norm": 2.49552583694458,
      "learning_rate": 2.033668562778183e-06,
      "loss": 0.4674,
      "step": 276120
    },
    {
      "epoch": 113.03,
      "grad_norm": 1.908345103263855,
      "learning_rate": 2.0335113283673896e-06,
      "loss": 0.4704,
      "step": 276130
    },
    {
      "epoch": 113.03,
      "grad_norm": 2.613630533218384,
      "learning_rate": 2.0333540953777516e-06,
      "loss": 0.4704,
      "step": 276140
    },
    {
      "epoch": 113.04,
      "grad_norm": 1.8243342638015747,
      "learning_rate": 2.0331968638099893e-06,
      "loss": 0.4798,
      "step": 276150
    },
    {
      "epoch": 113.04,
      "grad_norm": 2.163857936859131,
      "learning_rate": 2.0330396336648225e-06,
      "loss": 0.4697,
      "step": 276160
    },
    {
      "epoch": 113.05,
      "grad_norm": 2.2870147228240967,
      "learning_rate": 2.0328824049429723e-06,
      "loss": 0.4812,
      "step": 276170
    },
    {
      "epoch": 113.05,
      "grad_norm": 1.7402437925338745,
      "learning_rate": 2.032725177645158e-06,
      "loss": 0.4587,
      "step": 276180
    },
    {
      "epoch": 113.05,
      "grad_norm": 1.877975344657898,
      "learning_rate": 2.032567951772101e-06,
      "loss": 0.4842,
      "step": 276190
    },
    {
      "epoch": 113.06,
      "grad_norm": 2.2518765926361084,
      "learning_rate": 2.03241072732452e-06,
      "loss": 0.4791,
      "step": 276200
    },
    {
      "epoch": 113.06,
      "grad_norm": 1.7194501161575317,
      "learning_rate": 2.0322535043031365e-06,
      "loss": 0.4629,
      "step": 276210
    },
    {
      "epoch": 113.07,
      "grad_norm": 1.7527213096618652,
      "learning_rate": 2.03209628270867e-06,
      "loss": 0.465,
      "step": 276220
    },
    {
      "epoch": 113.07,
      "grad_norm": 2.5923895835876465,
      "learning_rate": 2.0319390625418406e-06,
      "loss": 0.4559,
      "step": 276230
    },
    {
      "epoch": 113.07,
      "grad_norm": 1.8244388103485107,
      "learning_rate": 2.0317818438033693e-06,
      "loss": 0.4668,
      "step": 276240
    },
    {
      "epoch": 113.08,
      "grad_norm": 1.9919567108154297,
      "learning_rate": 2.031624626493975e-06,
      "loss": 0.4486,
      "step": 276250
    },
    {
      "epoch": 113.08,
      "grad_norm": 1.6668462753295898,
      "learning_rate": 2.0314674106143798e-06,
      "loss": 0.4862,
      "step": 276260
    },
    {
      "epoch": 113.09,
      "grad_norm": 1.7531628608703613,
      "learning_rate": 2.0313101961653024e-06,
      "loss": 0.4555,
      "step": 276270
    },
    {
      "epoch": 113.09,
      "grad_norm": 2.456150770187378,
      "learning_rate": 2.0311529831474633e-06,
      "loss": 0.4602,
      "step": 276280
    },
    {
      "epoch": 113.09,
      "grad_norm": 1.7506874799728394,
      "learning_rate": 2.0309957715615826e-06,
      "loss": 0.4731,
      "step": 276290
    },
    {
      "epoch": 113.1,
      "grad_norm": 2.0436863899230957,
      "learning_rate": 2.0308385614083805e-06,
      "loss": 0.478,
      "step": 276300
    },
    {
      "epoch": 113.1,
      "grad_norm": 3.088205575942993,
      "learning_rate": 2.0306813526885777e-06,
      "loss": 0.4637,
      "step": 276310
    },
    {
      "epoch": 113.11,
      "grad_norm": 2.23223614692688,
      "learning_rate": 2.0305241454028926e-06,
      "loss": 0.4676,
      "step": 276320
    },
    {
      "epoch": 113.11,
      "grad_norm": 1.9798061847686768,
      "learning_rate": 2.0303669395520475e-06,
      "loss": 0.4578,
      "step": 276330
    },
    {
      "epoch": 113.12,
      "grad_norm": 1.6968730688095093,
      "learning_rate": 2.030209735136762e-06,
      "loss": 0.4635,
      "step": 276340
    },
    {
      "epoch": 113.12,
      "grad_norm": 1.7329206466674805,
      "learning_rate": 2.0300525321577553e-06,
      "loss": 0.4844,
      "step": 276350
    },
    {
      "epoch": 113.12,
      "grad_norm": 1.8108694553375244,
      "learning_rate": 2.0298953306157485e-06,
      "loss": 0.487,
      "step": 276360
    },
    {
      "epoch": 113.13,
      "grad_norm": 2.0603764057159424,
      "learning_rate": 2.0297381305114612e-06,
      "loss": 0.4648,
      "step": 276370
    },
    {
      "epoch": 113.13,
      "grad_norm": 1.922683596611023,
      "learning_rate": 2.0295809318456137e-06,
      "loss": 0.4481,
      "step": 276380
    },
    {
      "epoch": 113.14,
      "grad_norm": 1.708048701286316,
      "learning_rate": 2.029423734618926e-06,
      "loss": 0.4616,
      "step": 276390
    },
    {
      "epoch": 113.14,
      "grad_norm": 2.4006094932556152,
      "learning_rate": 2.029266538832118e-06,
      "loss": 0.4835,
      "step": 276400
    },
    {
      "epoch": 113.14,
      "grad_norm": 1.9278799295425415,
      "learning_rate": 2.0291093444859112e-06,
      "loss": 0.4666,
      "step": 276410
    },
    {
      "epoch": 113.15,
      "grad_norm": 1.7303671836853027,
      "learning_rate": 2.0289521515810237e-06,
      "loss": 0.4642,
      "step": 276420
    },
    {
      "epoch": 113.15,
      "grad_norm": 1.772775650024414,
      "learning_rate": 2.0287949601181762e-06,
      "loss": 0.4818,
      "step": 276430
    },
    {
      "epoch": 113.16,
      "grad_norm": 2.035435676574707,
      "learning_rate": 2.0286377700980893e-06,
      "loss": 0.4765,
      "step": 276440
    },
    {
      "epoch": 113.16,
      "grad_norm": 1.729037880897522,
      "learning_rate": 2.028480581521482e-06,
      "loss": 0.472,
      "step": 276450
    },
    {
      "epoch": 113.16,
      "grad_norm": 1.938763976097107,
      "learning_rate": 2.0283233943890758e-06,
      "loss": 0.4605,
      "step": 276460
    },
    {
      "epoch": 113.17,
      "grad_norm": 1.914752721786499,
      "learning_rate": 2.02816620870159e-06,
      "loss": 0.4747,
      "step": 276470
    },
    {
      "epoch": 113.17,
      "grad_norm": 1.982291579246521,
      "learning_rate": 2.0280090244597445e-06,
      "loss": 0.4545,
      "step": 276480
    },
    {
      "epoch": 113.18,
      "grad_norm": 1.6389411687850952,
      "learning_rate": 2.02785184166426e-06,
      "loss": 0.4426,
      "step": 276490
    },
    {
      "epoch": 113.18,
      "grad_norm": 2.5124692916870117,
      "learning_rate": 2.0276946603158554e-06,
      "loss": 0.4368,
      "step": 276500
    },
    {
      "epoch": 113.18,
      "grad_norm": 2.1490440368652344,
      "learning_rate": 2.0275374804152504e-06,
      "loss": 0.4648,
      "step": 276510
    },
    {
      "epoch": 113.19,
      "grad_norm": 3.326167583465576,
      "learning_rate": 2.0273803019631674e-06,
      "loss": 0.4578,
      "step": 276520
    },
    {
      "epoch": 113.19,
      "grad_norm": 3.0534300804138184,
      "learning_rate": 2.0272231249603243e-06,
      "loss": 0.4527,
      "step": 276530
    },
    {
      "epoch": 113.2,
      "grad_norm": 5.035248756408691,
      "learning_rate": 2.027065949407443e-06,
      "loss": 0.4691,
      "step": 276540
    },
    {
      "epoch": 113.2,
      "grad_norm": 1.6439422369003296,
      "learning_rate": 2.0269087753052412e-06,
      "loss": 0.4642,
      "step": 276550
    },
    {
      "epoch": 113.21,
      "grad_norm": 1.969763994216919,
      "learning_rate": 2.0267516026544403e-06,
      "loss": 0.4651,
      "step": 276560
    },
    {
      "epoch": 113.21,
      "grad_norm": 2.0671908855438232,
      "learning_rate": 2.02659443145576e-06,
      "loss": 0.4815,
      "step": 276570
    },
    {
      "epoch": 113.21,
      "grad_norm": 2.125884771347046,
      "learning_rate": 2.02643726170992e-06,
      "loss": 0.4619,
      "step": 276580
    },
    {
      "epoch": 113.22,
      "grad_norm": 1.9561984539031982,
      "learning_rate": 2.0262800934176407e-06,
      "loss": 0.4619,
      "step": 276590
    },
    {
      "epoch": 113.22,
      "grad_norm": 2.061842203140259,
      "learning_rate": 2.026122926579642e-06,
      "loss": 0.4694,
      "step": 276600
    },
    {
      "epoch": 113.23,
      "grad_norm": 1.6048588752746582,
      "learning_rate": 2.0259657611966437e-06,
      "loss": 0.4744,
      "step": 276610
    },
    {
      "epoch": 113.23,
      "grad_norm": 1.8511698246002197,
      "learning_rate": 2.0258085972693653e-06,
      "loss": 0.4644,
      "step": 276620
    },
    {
      "epoch": 113.23,
      "grad_norm": 1.9663201570510864,
      "learning_rate": 2.025651434798528e-06,
      "loss": 0.4698,
      "step": 276630
    },
    {
      "epoch": 113.24,
      "grad_norm": 1.7906217575073242,
      "learning_rate": 2.0254942737848508e-06,
      "loss": 0.4879,
      "step": 276640
    },
    {
      "epoch": 113.24,
      "grad_norm": 2.140193462371826,
      "learning_rate": 2.0253371142290534e-06,
      "loss": 0.4945,
      "step": 276650
    },
    {
      "epoch": 113.25,
      "grad_norm": 2.7154619693756104,
      "learning_rate": 2.025179956131856e-06,
      "loss": 0.4444,
      "step": 276660
    },
    {
      "epoch": 113.25,
      "grad_norm": 1.979183316230774,
      "learning_rate": 2.0250227994939793e-06,
      "loss": 0.4622,
      "step": 276670
    },
    {
      "epoch": 113.25,
      "grad_norm": 1.6074379682540894,
      "learning_rate": 2.024865644316142e-06,
      "loss": 0.458,
      "step": 276680
    },
    {
      "epoch": 113.26,
      "grad_norm": 2.077423334121704,
      "learning_rate": 2.024708490599065e-06,
      "loss": 0.454,
      "step": 276690
    },
    {
      "epoch": 113.26,
      "grad_norm": 2.3302676677703857,
      "learning_rate": 2.0245513383434667e-06,
      "loss": 0.4556,
      "step": 276700
    },
    {
      "epoch": 113.27,
      "grad_norm": 2.3617799282073975,
      "learning_rate": 2.0243941875500685e-06,
      "loss": 0.4653,
      "step": 276710
    },
    {
      "epoch": 113.27,
      "grad_norm": 1.7581238746643066,
      "learning_rate": 2.0242370382195902e-06,
      "loss": 0.449,
      "step": 276720
    },
    {
      "epoch": 113.27,
      "grad_norm": 2.153952121734619,
      "learning_rate": 2.024079890352751e-06,
      "loss": 0.4782,
      "step": 276730
    },
    {
      "epoch": 113.28,
      "grad_norm": 1.9584137201309204,
      "learning_rate": 2.0239227439502713e-06,
      "loss": 0.4277,
      "step": 276740
    },
    {
      "epoch": 113.28,
      "grad_norm": 1.7637711763381958,
      "learning_rate": 2.0237655990128705e-06,
      "loss": 0.4612,
      "step": 276750
    },
    {
      "epoch": 113.29,
      "grad_norm": 1.5134979486465454,
      "learning_rate": 2.023608455541269e-06,
      "loss": 0.4498,
      "step": 276760
    },
    {
      "epoch": 113.29,
      "grad_norm": 2.622344493865967,
      "learning_rate": 2.023451313536187e-06,
      "loss": 0.4826,
      "step": 276770
    },
    {
      "epoch": 113.3,
      "grad_norm": 1.8178355693817139,
      "learning_rate": 2.023294172998343e-06,
      "loss": 0.4595,
      "step": 276780
    },
    {
      "epoch": 113.3,
      "grad_norm": 2.383294105529785,
      "learning_rate": 2.023137033928457e-06,
      "loss": 0.481,
      "step": 276790
    },
    {
      "epoch": 113.3,
      "grad_norm": 2.1171135902404785,
      "learning_rate": 2.0229798963272498e-06,
      "loss": 0.4597,
      "step": 276800
    },
    {
      "epoch": 113.31,
      "grad_norm": 2.0174598693847656,
      "learning_rate": 2.022822760195441e-06,
      "loss": 0.4637,
      "step": 276810
    },
    {
      "epoch": 113.31,
      "grad_norm": 2.1527562141418457,
      "learning_rate": 2.0226656255337496e-06,
      "loss": 0.4741,
      "step": 276820
    },
    {
      "epoch": 113.32,
      "grad_norm": 1.9259682893753052,
      "learning_rate": 2.0225084923428963e-06,
      "loss": 0.4649,
      "step": 276830
    },
    {
      "epoch": 113.32,
      "grad_norm": 1.6810929775238037,
      "learning_rate": 2.0223513606236005e-06,
      "loss": 0.4563,
      "step": 276840
    },
    {
      "epoch": 113.32,
      "grad_norm": 2.375427484512329,
      "learning_rate": 2.0221942303765827e-06,
      "loss": 0.4749,
      "step": 276850
    },
    {
      "epoch": 113.33,
      "grad_norm": 3.0932979583740234,
      "learning_rate": 2.0220371016025615e-06,
      "loss": 0.4775,
      "step": 276860
    },
    {
      "epoch": 113.33,
      "grad_norm": 1.9194507598876953,
      "learning_rate": 2.0218799743022574e-06,
      "loss": 0.4727,
      "step": 276870
    },
    {
      "epoch": 113.34,
      "grad_norm": 2.273785352706909,
      "learning_rate": 2.02172284847639e-06,
      "loss": 0.4791,
      "step": 276880
    },
    {
      "epoch": 113.34,
      "grad_norm": 1.9792755842208862,
      "learning_rate": 2.0215657241256777e-06,
      "loss": 0.4709,
      "step": 276890
    },
    {
      "epoch": 113.34,
      "grad_norm": 1.8212863206863403,
      "learning_rate": 2.0214086012508437e-06,
      "loss": 0.4365,
      "step": 276900
    },
    {
      "epoch": 113.35,
      "grad_norm": 2.0513336658477783,
      "learning_rate": 2.021251479852605e-06,
      "loss": 0.4762,
      "step": 276910
    },
    {
      "epoch": 113.35,
      "grad_norm": 1.8864504098892212,
      "learning_rate": 2.021094359931682e-06,
      "loss": 0.4466,
      "step": 276920
    },
    {
      "epoch": 113.36,
      "grad_norm": 1.7541474103927612,
      "learning_rate": 2.0209372414887946e-06,
      "loss": 0.4847,
      "step": 276930
    },
    {
      "epoch": 113.36,
      "grad_norm": 1.9718800783157349,
      "learning_rate": 2.0207801245246625e-06,
      "loss": 0.4757,
      "step": 276940
    },
    {
      "epoch": 113.36,
      "grad_norm": 2.312263011932373,
      "learning_rate": 2.0206230090400055e-06,
      "loss": 0.4778,
      "step": 276950
    },
    {
      "epoch": 113.37,
      "grad_norm": 2.532538652420044,
      "learning_rate": 2.020465895035543e-06,
      "loss": 0.4669,
      "step": 276960
    },
    {
      "epoch": 113.37,
      "grad_norm": 2.0929901599884033,
      "learning_rate": 2.0203087825119953e-06,
      "loss": 0.4579,
      "step": 276970
    },
    {
      "epoch": 113.38,
      "grad_norm": 2.2139155864715576,
      "learning_rate": 2.020151671470081e-06,
      "loss": 0.455,
      "step": 276980
    },
    {
      "epoch": 113.38,
      "grad_norm": 3.02101993560791,
      "learning_rate": 2.0199945619105217e-06,
      "loss": 0.4923,
      "step": 276990
    },
    {
      "epoch": 113.39,
      "grad_norm": 1.7182368040084839,
      "learning_rate": 2.019837453834035e-06,
      "loss": 0.4873,
      "step": 277000
    },
    {
      "epoch": 113.39,
      "grad_norm": 1.8836196660995483,
      "learning_rate": 2.019680347241342e-06,
      "loss": 0.4509,
      "step": 277010
    },
    {
      "epoch": 113.39,
      "grad_norm": 1.507737636566162,
      "learning_rate": 2.0195232421331613e-06,
      "loss": 0.456,
      "step": 277020
    },
    {
      "epoch": 113.4,
      "grad_norm": 1.9481239318847656,
      "learning_rate": 2.0193661385102136e-06,
      "loss": 0.4642,
      "step": 277030
    },
    {
      "epoch": 113.4,
      "grad_norm": 1.6256916522979736,
      "learning_rate": 2.019209036373218e-06,
      "loss": 0.4584,
      "step": 277040
    },
    {
      "epoch": 113.41,
      "grad_norm": 1.8597337007522583,
      "learning_rate": 2.019051935722894e-06,
      "loss": 0.466,
      "step": 277050
    },
    {
      "epoch": 113.41,
      "grad_norm": 2.2603037357330322,
      "learning_rate": 2.0188948365599622e-06,
      "loss": 0.4822,
      "step": 277060
    },
    {
      "epoch": 113.41,
      "grad_norm": 1.8790301084518433,
      "learning_rate": 2.0187377388851407e-06,
      "loss": 0.4687,
      "step": 277070
    },
    {
      "epoch": 113.42,
      "grad_norm": 1.7097923755645752,
      "learning_rate": 2.0185806426991502e-06,
      "loss": 0.4574,
      "step": 277080
    },
    {
      "epoch": 113.42,
      "grad_norm": 1.674875020980835,
      "learning_rate": 2.0184235480027104e-06,
      "loss": 0.4538,
      "step": 277090
    },
    {
      "epoch": 113.43,
      "grad_norm": 1.6475584506988525,
      "learning_rate": 2.018266454796541e-06,
      "loss": 0.4872,
      "step": 277100
    },
    {
      "epoch": 113.43,
      "grad_norm": 2.161949634552002,
      "learning_rate": 2.0181093630813613e-06,
      "loss": 0.4488,
      "step": 277110
    },
    {
      "epoch": 113.43,
      "grad_norm": 1.6643365621566772,
      "learning_rate": 2.0179522728578903e-06,
      "loss": 0.4481,
      "step": 277120
    },
    {
      "epoch": 113.44,
      "grad_norm": 1.8747416734695435,
      "learning_rate": 2.0177951841268494e-06,
      "loss": 0.4677,
      "step": 277130
    },
    {
      "epoch": 113.44,
      "grad_norm": 2.3346593379974365,
      "learning_rate": 2.0176380968889565e-06,
      "loss": 0.488,
      "step": 277140
    },
    {
      "epoch": 113.45,
      "grad_norm": 2.0548267364501953,
      "learning_rate": 2.017481011144931e-06,
      "loss": 0.4591,
      "step": 277150
    },
    {
      "epoch": 113.45,
      "grad_norm": 1.6974014043807983,
      "learning_rate": 2.017323926895494e-06,
      "loss": 0.4662,
      "step": 277160
    },
    {
      "epoch": 113.45,
      "grad_norm": 2.1124649047851562,
      "learning_rate": 2.0171668441413642e-06,
      "loss": 0.4814,
      "step": 277170
    },
    {
      "epoch": 113.46,
      "grad_norm": 2.3446850776672363,
      "learning_rate": 2.017009762883261e-06,
      "loss": 0.4563,
      "step": 277180
    },
    {
      "epoch": 113.46,
      "grad_norm": 2.127690076828003,
      "learning_rate": 2.0168526831219042e-06,
      "loss": 0.4662,
      "step": 277190
    },
    {
      "epoch": 113.47,
      "grad_norm": 1.8270673751831055,
      "learning_rate": 2.0166956048580137e-06,
      "loss": 0.4537,
      "step": 277200
    },
    {
      "epoch": 113.47,
      "grad_norm": 2.553335189819336,
      "learning_rate": 2.016538528092309e-06,
      "loss": 0.4687,
      "step": 277210
    },
    {
      "epoch": 113.48,
      "grad_norm": 1.8048640489578247,
      "learning_rate": 2.016381452825509e-06,
      "loss": 0.4578,
      "step": 277220
    },
    {
      "epoch": 113.48,
      "grad_norm": 1.6615604162216187,
      "learning_rate": 2.0162243790583333e-06,
      "loss": 0.485,
      "step": 277230
    },
    {
      "epoch": 113.48,
      "grad_norm": 1.5535180568695068,
      "learning_rate": 2.0160673067915017e-06,
      "loss": 0.4705,
      "step": 277240
    },
    {
      "epoch": 113.49,
      "grad_norm": 1.9953207969665527,
      "learning_rate": 2.0159102360257345e-06,
      "loss": 0.4655,
      "step": 277250
    },
    {
      "epoch": 113.49,
      "grad_norm": 2.147387742996216,
      "learning_rate": 2.01575316676175e-06,
      "loss": 0.4762,
      "step": 277260
    },
    {
      "epoch": 113.5,
      "grad_norm": 1.6709685325622559,
      "learning_rate": 2.0155960990002674e-06,
      "loss": 0.4895,
      "step": 277270
    },
    {
      "epoch": 113.5,
      "grad_norm": 1.8064007759094238,
      "learning_rate": 2.015439032742008e-06,
      "loss": 0.4719,
      "step": 277280
    },
    {
      "epoch": 113.5,
      "grad_norm": 1.6645704507827759,
      "learning_rate": 2.01528196798769e-06,
      "loss": 0.4666,
      "step": 277290
    },
    {
      "epoch": 113.51,
      "grad_norm": 1.8642321825027466,
      "learning_rate": 2.0151249047380335e-06,
      "loss": 0.484,
      "step": 277300
    },
    {
      "epoch": 113.51,
      "grad_norm": 1.9320091009140015,
      "learning_rate": 2.014967842993757e-06,
      "loss": 0.4779,
      "step": 277310
    },
    {
      "epoch": 113.52,
      "grad_norm": 2.10380220413208,
      "learning_rate": 2.0148107827555815e-06,
      "loss": 0.4536,
      "step": 277320
    },
    {
      "epoch": 113.52,
      "grad_norm": 1.6996346712112427,
      "learning_rate": 2.014653724024225e-06,
      "loss": 0.4598,
      "step": 277330
    },
    {
      "epoch": 113.52,
      "grad_norm": 1.7852853536605835,
      "learning_rate": 2.014496666800408e-06,
      "loss": 0.4825,
      "step": 277340
    },
    {
      "epoch": 113.53,
      "grad_norm": 2.062882423400879,
      "learning_rate": 2.0143396110848496e-06,
      "loss": 0.4661,
      "step": 277350
    },
    {
      "epoch": 113.53,
      "grad_norm": 2.009572982788086,
      "learning_rate": 2.014182556878269e-06,
      "loss": 0.4602,
      "step": 277360
    },
    {
      "epoch": 113.54,
      "grad_norm": 1.8782521486282349,
      "learning_rate": 2.0140255041813856e-06,
      "loss": 0.4651,
      "step": 277370
    },
    {
      "epoch": 113.54,
      "grad_norm": 1.791369915008545,
      "learning_rate": 2.013868452994919e-06,
      "loss": 0.469,
      "step": 277380
    },
    {
      "epoch": 113.54,
      "grad_norm": 2.4393301010131836,
      "learning_rate": 2.0137114033195888e-06,
      "loss": 0.4511,
      "step": 277390
    },
    {
      "epoch": 113.55,
      "grad_norm": 1.6675057411193848,
      "learning_rate": 2.0135543551561147e-06,
      "loss": 0.4714,
      "step": 277400
    },
    {
      "epoch": 113.55,
      "grad_norm": 2.1908185482025146,
      "learning_rate": 2.013397308505215e-06,
      "loss": 0.465,
      "step": 277410
    },
    {
      "epoch": 113.56,
      "grad_norm": 2.317556381225586,
      "learning_rate": 2.0132402633676103e-06,
      "loss": 0.4617,
      "step": 277420
    },
    {
      "epoch": 113.56,
      "grad_norm": 1.5171149969100952,
      "learning_rate": 2.0130832197440194e-06,
      "loss": 0.464,
      "step": 277430
    },
    {
      "epoch": 113.57,
      "grad_norm": 2.490795135498047,
      "learning_rate": 2.0129261776351623e-06,
      "loss": 0.4902,
      "step": 277440
    },
    {
      "epoch": 113.57,
      "grad_norm": 1.7708991765975952,
      "learning_rate": 2.0127691370417576e-06,
      "loss": 0.4872,
      "step": 277450
    },
    {
      "epoch": 113.57,
      "grad_norm": 1.5062267780303955,
      "learning_rate": 2.0126120979645234e-06,
      "loss": 0.481,
      "step": 277460
    },
    {
      "epoch": 113.58,
      "grad_norm": 1.667868971824646,
      "learning_rate": 2.012455060404182e-06,
      "loss": 0.4547,
      "step": 277470
    },
    {
      "epoch": 113.58,
      "grad_norm": 2.096920967102051,
      "learning_rate": 2.0122980243614516e-06,
      "loss": 0.4619,
      "step": 277480
    },
    {
      "epoch": 113.59,
      "grad_norm": 1.9194103479385376,
      "learning_rate": 2.0121409898370516e-06,
      "loss": 0.4752,
      "step": 277490
    },
    {
      "epoch": 113.59,
      "grad_norm": 1.688124418258667,
      "learning_rate": 2.0119839568317006e-06,
      "loss": 0.4798,
      "step": 277500
    },
    {
      "epoch": 113.59,
      "grad_norm": 1.8905377388000488,
      "learning_rate": 2.0118269253461185e-06,
      "loss": 0.4513,
      "step": 277510
    },
    {
      "epoch": 113.6,
      "grad_norm": 1.7521347999572754,
      "learning_rate": 2.011669895381025e-06,
      "loss": 0.4566,
      "step": 277520
    },
    {
      "epoch": 113.6,
      "grad_norm": 1.869187831878662,
      "learning_rate": 2.0115128669371386e-06,
      "loss": 0.4573,
      "step": 277530
    },
    {
      "epoch": 113.61,
      "grad_norm": 1.6161059141159058,
      "learning_rate": 2.011355840015179e-06,
      "loss": 0.4714,
      "step": 277540
    },
    {
      "epoch": 113.61,
      "grad_norm": 1.8326029777526855,
      "learning_rate": 2.011198814615866e-06,
      "loss": 0.4675,
      "step": 277550
    },
    {
      "epoch": 113.61,
      "grad_norm": 1.844592809677124,
      "learning_rate": 2.011041790739918e-06,
      "loss": 0.4584,
      "step": 277560
    },
    {
      "epoch": 113.62,
      "grad_norm": 1.9214386940002441,
      "learning_rate": 2.0108847683880553e-06,
      "loss": 0.4621,
      "step": 277570
    },
    {
      "epoch": 113.62,
      "grad_norm": 1.93405020236969,
      "learning_rate": 2.0107277475609963e-06,
      "loss": 0.4735,
      "step": 277580
    },
    {
      "epoch": 113.63,
      "grad_norm": 2.1825790405273438,
      "learning_rate": 2.010570728259461e-06,
      "loss": 0.466,
      "step": 277590
    },
    {
      "epoch": 113.63,
      "grad_norm": 2.070873975753784,
      "learning_rate": 2.010413710484168e-06,
      "loss": 0.4659,
      "step": 277600
    },
    {
      "epoch": 113.63,
      "grad_norm": 1.676142692565918,
      "learning_rate": 2.010256694235837e-06,
      "loss": 0.4482,
      "step": 277610
    },
    {
      "epoch": 113.64,
      "grad_norm": 1.8646981716156006,
      "learning_rate": 2.0100996795151873e-06,
      "loss": 0.4852,
      "step": 277620
    },
    {
      "epoch": 113.64,
      "grad_norm": 1.8092995882034302,
      "learning_rate": 2.009942666322938e-06,
      "loss": 0.4776,
      "step": 277630
    },
    {
      "epoch": 113.65,
      "grad_norm": 2.329561233520508,
      "learning_rate": 2.0097856546598074e-06,
      "loss": 0.461,
      "step": 277640
    },
    {
      "epoch": 113.65,
      "grad_norm": 2.0294158458709717,
      "learning_rate": 2.009628644526517e-06,
      "loss": 0.4627,
      "step": 277650
    },
    {
      "epoch": 113.66,
      "grad_norm": 1.9226016998291016,
      "learning_rate": 2.0094716359237844e-06,
      "loss": 0.4836,
      "step": 277660
    },
    {
      "epoch": 113.66,
      "grad_norm": 1.657662272453308,
      "learning_rate": 2.009314628852329e-06,
      "loss": 0.4649,
      "step": 277670
    },
    {
      "epoch": 113.66,
      "grad_norm": 2.0217604637145996,
      "learning_rate": 2.0091576233128706e-06,
      "loss": 0.4518,
      "step": 277680
    },
    {
      "epoch": 113.67,
      "grad_norm": 1.9974873065948486,
      "learning_rate": 2.0090006193061283e-06,
      "loss": 0.4812,
      "step": 277690
    },
    {
      "epoch": 113.67,
      "grad_norm": 2.0580904483795166,
      "learning_rate": 2.0088436168328206e-06,
      "loss": 0.4698,
      "step": 277700
    },
    {
      "epoch": 113.68,
      "grad_norm": 2.52900767326355,
      "learning_rate": 2.008686615893668e-06,
      "loss": 0.4724,
      "step": 277710
    },
    {
      "epoch": 113.68,
      "grad_norm": 3.504136323928833,
      "learning_rate": 2.008529616489388e-06,
      "loss": 0.4876,
      "step": 277720
    },
    {
      "epoch": 113.68,
      "grad_norm": 2.0033376216888428,
      "learning_rate": 2.0083726186207008e-06,
      "loss": 0.4939,
      "step": 277730
    },
    {
      "epoch": 113.69,
      "grad_norm": 1.958678960800171,
      "learning_rate": 2.0082156222883256e-06,
      "loss": 0.4688,
      "step": 277740
    },
    {
      "epoch": 113.69,
      "grad_norm": 2.823936939239502,
      "learning_rate": 2.008058627492981e-06,
      "loss": 0.4575,
      "step": 277750
    },
    {
      "epoch": 113.7,
      "grad_norm": 2.4816057682037354,
      "learning_rate": 2.007901634235387e-06,
      "loss": 0.465,
      "step": 277760
    },
    {
      "epoch": 113.7,
      "grad_norm": 2.7461771965026855,
      "learning_rate": 2.007744642516262e-06,
      "loss": 0.4732,
      "step": 277770
    },
    {
      "epoch": 113.7,
      "grad_norm": 2.261634111404419,
      "learning_rate": 2.007587652336326e-06,
      "loss": 0.4684,
      "step": 277780
    },
    {
      "epoch": 113.71,
      "grad_norm": 1.6572400331497192,
      "learning_rate": 2.007430663696297e-06,
      "loss": 0.4602,
      "step": 277790
    },
    {
      "epoch": 113.71,
      "grad_norm": 1.6859352588653564,
      "learning_rate": 2.007273676596896e-06,
      "loss": 0.4777,
      "step": 277800
    },
    {
      "epoch": 113.72,
      "grad_norm": 2.4235100746154785,
      "learning_rate": 2.00711669103884e-06,
      "loss": 0.4617,
      "step": 277810
    },
    {
      "epoch": 113.72,
      "grad_norm": 1.8731725215911865,
      "learning_rate": 2.006959707022849e-06,
      "loss": 0.4685,
      "step": 277820
    },
    {
      "epoch": 113.72,
      "grad_norm": 1.7757292985916138,
      "learning_rate": 2.0068027245496415e-06,
      "loss": 0.4765,
      "step": 277830
    },
    {
      "epoch": 113.73,
      "grad_norm": 1.9366153478622437,
      "learning_rate": 2.0066457436199383e-06,
      "loss": 0.4833,
      "step": 277840
    },
    {
      "epoch": 113.73,
      "grad_norm": 1.7660115957260132,
      "learning_rate": 2.0064887642344575e-06,
      "loss": 0.473,
      "step": 277850
    },
    {
      "epoch": 113.74,
      "grad_norm": 2.0863659381866455,
      "learning_rate": 2.0063317863939182e-06,
      "loss": 0.4776,
      "step": 277860
    },
    {
      "epoch": 113.74,
      "grad_norm": 2.3213319778442383,
      "learning_rate": 2.0061748100990396e-06,
      "loss": 0.4594,
      "step": 277870
    },
    {
      "epoch": 113.75,
      "grad_norm": 1.5488249063491821,
      "learning_rate": 2.0060178353505403e-06,
      "loss": 0.4568,
      "step": 277880
    },
    {
      "epoch": 113.75,
      "grad_norm": 1.7763280868530273,
      "learning_rate": 2.0058608621491395e-06,
      "loss": 0.4789,
      "step": 277890
    },
    {
      "epoch": 113.75,
      "grad_norm": 2.0573782920837402,
      "learning_rate": 2.005703890495557e-06,
      "loss": 0.4795,
      "step": 277900
    },
    {
      "epoch": 113.76,
      "grad_norm": 1.74599289894104,
      "learning_rate": 2.0055469203905115e-06,
      "loss": 0.4544,
      "step": 277910
    },
    {
      "epoch": 113.76,
      "grad_norm": 1.9125802516937256,
      "learning_rate": 2.0053899518347213e-06,
      "loss": 0.4588,
      "step": 277920
    },
    {
      "epoch": 113.77,
      "grad_norm": 1.8174399137496948,
      "learning_rate": 2.005232984828907e-06,
      "loss": 0.4491,
      "step": 277930
    },
    {
      "epoch": 113.77,
      "grad_norm": 1.880868911743164,
      "learning_rate": 2.0050760193737863e-06,
      "loss": 0.467,
      "step": 277940
    },
    {
      "epoch": 113.77,
      "grad_norm": 1.8889721632003784,
      "learning_rate": 2.004919055470078e-06,
      "loss": 0.4764,
      "step": 277950
    },
    {
      "epoch": 113.78,
      "grad_norm": 1.8090901374816895,
      "learning_rate": 2.0047620931185023e-06,
      "loss": 0.4781,
      "step": 277960
    },
    {
      "epoch": 113.78,
      "grad_norm": 1.682279109954834,
      "learning_rate": 2.0046051323197776e-06,
      "loss": 0.4473,
      "step": 277970
    },
    {
      "epoch": 113.79,
      "grad_norm": 1.8964879512786865,
      "learning_rate": 2.004448173074623e-06,
      "loss": 0.4624,
      "step": 277980
    },
    {
      "epoch": 113.79,
      "grad_norm": 4.114267349243164,
      "learning_rate": 2.004291215383758e-06,
      "loss": 0.4685,
      "step": 277990
    },
    {
      "epoch": 113.79,
      "grad_norm": 1.984410285949707,
      "learning_rate": 2.0041342592479006e-06,
      "loss": 0.4739,
      "step": 278000
    },
    {
      "epoch": 113.8,
      "grad_norm": 2.1747453212738037,
      "learning_rate": 2.0039773046677703e-06,
      "loss": 0.4465,
      "step": 278010
    },
    {
      "epoch": 113.8,
      "grad_norm": 2.570786952972412,
      "learning_rate": 2.003820351644086e-06,
      "loss": 0.5016,
      "step": 278020
    },
    {
      "epoch": 113.81,
      "grad_norm": 6.636737823486328,
      "learning_rate": 2.003663400177567e-06,
      "loss": 0.4481,
      "step": 278030
    },
    {
      "epoch": 113.81,
      "grad_norm": 2.0200273990631104,
      "learning_rate": 2.0035064502689315e-06,
      "loss": 0.473,
      "step": 278040
    },
    {
      "epoch": 113.81,
      "grad_norm": 1.7148501873016357,
      "learning_rate": 2.0033495019188997e-06,
      "loss": 0.4713,
      "step": 278050
    },
    {
      "epoch": 113.82,
      "grad_norm": 2.558739185333252,
      "learning_rate": 2.00319255512819e-06,
      "loss": 0.4616,
      "step": 278060
    },
    {
      "epoch": 113.82,
      "grad_norm": 2.245631694793701,
      "learning_rate": 2.0030356098975212e-06,
      "loss": 0.4427,
      "step": 278070
    },
    {
      "epoch": 113.83,
      "grad_norm": 1.8486658334732056,
      "learning_rate": 2.002878666227612e-06,
      "loss": 0.4502,
      "step": 278080
    },
    {
      "epoch": 113.83,
      "grad_norm": 1.7474370002746582,
      "learning_rate": 2.0027217241191817e-06,
      "loss": 0.4768,
      "step": 278090
    },
    {
      "epoch": 113.84,
      "grad_norm": 1.6452215909957886,
      "learning_rate": 2.002564783572949e-06,
      "loss": 0.4783,
      "step": 278100
    },
    {
      "epoch": 113.84,
      "grad_norm": 1.6306167840957642,
      "learning_rate": 2.002407844589633e-06,
      "loss": 0.4667,
      "step": 278110
    },
    {
      "epoch": 113.84,
      "grad_norm": 1.753720998764038,
      "learning_rate": 2.0022509071699527e-06,
      "loss": 0.471,
      "step": 278120
    },
    {
      "epoch": 113.85,
      "grad_norm": 2.023409128189087,
      "learning_rate": 2.0020939713146263e-06,
      "loss": 0.4686,
      "step": 278130
    },
    {
      "epoch": 113.85,
      "grad_norm": 2.127223491668701,
      "learning_rate": 2.0019370370243736e-06,
      "loss": 0.4559,
      "step": 278140
    },
    {
      "epoch": 113.86,
      "grad_norm": 2.3134539127349854,
      "learning_rate": 2.0017801042999135e-06,
      "loss": 0.4829,
      "step": 278150
    },
    {
      "epoch": 113.86,
      "grad_norm": 2.016228199005127,
      "learning_rate": 2.001623173141964e-06,
      "loss": 0.4818,
      "step": 278160
    },
    {
      "epoch": 113.86,
      "grad_norm": 1.929287075996399,
      "learning_rate": 2.001466243551245e-06,
      "loss": 0.4738,
      "step": 278170
    },
    {
      "epoch": 113.87,
      "grad_norm": 1.8683595657348633,
      "learning_rate": 2.0013093155284747e-06,
      "loss": 0.4741,
      "step": 278180
    },
    {
      "epoch": 113.87,
      "grad_norm": 2.2586140632629395,
      "learning_rate": 2.001152389074372e-06,
      "loss": 0.4549,
      "step": 278190
    },
    {
      "epoch": 113.88,
      "grad_norm": 2.42147159576416,
      "learning_rate": 2.000995464189656e-06,
      "loss": 0.4588,
      "step": 278200
    },
    {
      "epoch": 113.88,
      "grad_norm": 2.457688570022583,
      "learning_rate": 2.0008385408750448e-06,
      "loss": 0.4816,
      "step": 278210
    },
    {
      "epoch": 113.88,
      "grad_norm": 1.8462846279144287,
      "learning_rate": 2.0006816191312584e-06,
      "loss": 0.4591,
      "step": 278220
    },
    {
      "epoch": 113.89,
      "grad_norm": 1.7617263793945312,
      "learning_rate": 2.000524698959015e-06,
      "loss": 0.4658,
      "step": 278230
    },
    {
      "epoch": 113.89,
      "grad_norm": 2.0426557064056396,
      "learning_rate": 2.000367780359034e-06,
      "loss": 0.4743,
      "step": 278240
    },
    {
      "epoch": 113.9,
      "grad_norm": 1.7437995672225952,
      "learning_rate": 2.000210863332033e-06,
      "loss": 0.4517,
      "step": 278250
    },
    {
      "epoch": 113.9,
      "grad_norm": 1.7441314458847046,
      "learning_rate": 2.0000539478787324e-06,
      "loss": 0.4471,
      "step": 278260
    },
    {
      "epoch": 113.91,
      "grad_norm": 1.6682188510894775,
      "learning_rate": 1.99989703399985e-06,
      "loss": 0.4757,
      "step": 278270
    },
    {
      "epoch": 113.91,
      "grad_norm": 2.1633896827697754,
      "learning_rate": 1.9997401216961042e-06,
      "loss": 0.4643,
      "step": 278280
    },
    {
      "epoch": 113.91,
      "grad_norm": 2.359722375869751,
      "learning_rate": 1.9995832109682156e-06,
      "loss": 0.4608,
      "step": 278290
    },
    {
      "epoch": 113.92,
      "grad_norm": 2.139211654663086,
      "learning_rate": 1.9994263018169005e-06,
      "loss": 0.4467,
      "step": 278300
    },
    {
      "epoch": 113.92,
      "grad_norm": 1.8553799390792847,
      "learning_rate": 1.9992693942428797e-06,
      "loss": 0.4818,
      "step": 278310
    },
    {
      "epoch": 113.93,
      "grad_norm": 1.7039166688919067,
      "learning_rate": 1.9991124882468708e-06,
      "loss": 0.4724,
      "step": 278320
    },
    {
      "epoch": 113.93,
      "grad_norm": 1.9183951616287231,
      "learning_rate": 1.998955583829593e-06,
      "loss": 0.4794,
      "step": 278330
    },
    {
      "epoch": 113.93,
      "grad_norm": 1.885759711265564,
      "learning_rate": 1.998798680991765e-06,
      "loss": 0.4486,
      "step": 278340
    },
    {
      "epoch": 113.94,
      "grad_norm": 1.686321496963501,
      "learning_rate": 1.998641779734106e-06,
      "loss": 0.4836,
      "step": 278350
    },
    {
      "epoch": 113.94,
      "grad_norm": 2.0278103351593018,
      "learning_rate": 1.9984848800573338e-06,
      "loss": 0.4774,
      "step": 278360
    },
    {
      "epoch": 113.95,
      "grad_norm": 1.6733338832855225,
      "learning_rate": 1.9983279819621678e-06,
      "loss": 0.4738,
      "step": 278370
    },
    {
      "epoch": 113.95,
      "grad_norm": 2.0776875019073486,
      "learning_rate": 1.998171085449327e-06,
      "loss": 0.4559,
      "step": 278380
    },
    {
      "epoch": 113.95,
      "grad_norm": 2.2499563694000244,
      "learning_rate": 1.9980141905195296e-06,
      "loss": 0.4434,
      "step": 278390
    },
    {
      "epoch": 113.96,
      "grad_norm": 1.9538532495498657,
      "learning_rate": 1.9978572971734926e-06,
      "loss": 0.4687,
      "step": 278400
    },
    {
      "epoch": 113.96,
      "grad_norm": 1.5646131038665771,
      "learning_rate": 1.9977004054119384e-06,
      "loss": 0.4574,
      "step": 278410
    },
    {
      "epoch": 113.97,
      "grad_norm": 1.9614261388778687,
      "learning_rate": 1.997543515235583e-06,
      "loss": 0.4797,
      "step": 278420
    },
    {
      "epoch": 113.97,
      "grad_norm": 1.8283004760742188,
      "learning_rate": 1.997386626645147e-06,
      "loss": 0.4687,
      "step": 278430
    },
    {
      "epoch": 113.97,
      "grad_norm": 1.787290096282959,
      "learning_rate": 1.997229739641347e-06,
      "loss": 0.4752,
      "step": 278440
    },
    {
      "epoch": 113.98,
      "grad_norm": 2.0383987426757812,
      "learning_rate": 1.9970728542249032e-06,
      "loss": 0.4515,
      "step": 278450
    },
    {
      "epoch": 113.98,
      "grad_norm": 2.1664068698883057,
      "learning_rate": 1.9969159703965334e-06,
      "loss": 0.4496,
      "step": 278460
    },
    {
      "epoch": 113.99,
      "grad_norm": 1.756830096244812,
      "learning_rate": 1.9967590881569564e-06,
      "loss": 0.4673,
      "step": 278470
    },
    {
      "epoch": 113.99,
      "grad_norm": 1.7963825464248657,
      "learning_rate": 1.9966022075068913e-06,
      "loss": 0.4911,
      "step": 278480
    },
    {
      "epoch": 114.0,
      "grad_norm": 1.7097867727279663,
      "learning_rate": 1.9964453284470565e-06,
      "loss": 0.4445,
      "step": 278490
    },
    {
      "epoch": 114.0,
      "grad_norm": 1.7017509937286377,
      "learning_rate": 1.9962884509781703e-06,
      "loss": 0.4644,
      "step": 278500
    },
    {
      "epoch": 114.0,
      "eval_loss": 0.466499000787735,
      "eval_runtime": 56.5054,
      "eval_samples_per_second": 61.038,
      "eval_steps_per_second": 7.645,
      "step": 278502
    },
    {
      "epoch": 114.0,
      "grad_norm": 1.6774855852127075,
      "learning_rate": 1.9961315751009525e-06,
      "loss": 0.4608,
      "step": 278510
    },
    {
      "epoch": 114.01,
      "grad_norm": 2.6634957790374756,
      "learning_rate": 1.9959747008161202e-06,
      "loss": 0.4575,
      "step": 278520
    },
    {
      "epoch": 114.01,
      "grad_norm": 1.9350855350494385,
      "learning_rate": 1.9958178281243926e-06,
      "loss": 0.4808,
      "step": 278530
    },
    {
      "epoch": 114.02,
      "grad_norm": 2.119206190109253,
      "learning_rate": 1.9956609570264884e-06,
      "loss": 0.4617,
      "step": 278540
    },
    {
      "epoch": 114.02,
      "grad_norm": 2.1495273113250732,
      "learning_rate": 1.995504087523126e-06,
      "loss": 0.4682,
      "step": 278550
    },
    {
      "epoch": 114.02,
      "grad_norm": 1.847203254699707,
      "learning_rate": 1.9953472196150248e-06,
      "loss": 0.4512,
      "step": 278560
    },
    {
      "epoch": 114.03,
      "grad_norm": 1.7229630947113037,
      "learning_rate": 1.9951903533029024e-06,
      "loss": 0.4639,
      "step": 278570
    },
    {
      "epoch": 114.03,
      "grad_norm": 1.793109655380249,
      "learning_rate": 1.995033488587478e-06,
      "loss": 0.4613,
      "step": 278580
    },
    {
      "epoch": 114.04,
      "grad_norm": 2.4588451385498047,
      "learning_rate": 1.9948766254694692e-06,
      "loss": 0.48,
      "step": 278590
    },
    {
      "epoch": 114.04,
      "grad_norm": 1.9529248476028442,
      "learning_rate": 1.994719763949596e-06,
      "loss": 0.4949,
      "step": 278600
    },
    {
      "epoch": 114.04,
      "grad_norm": 1.9234644174575806,
      "learning_rate": 1.994562904028576e-06,
      "loss": 0.4708,
      "step": 278610
    },
    {
      "epoch": 114.05,
      "grad_norm": 1.909073829650879,
      "learning_rate": 1.9944060457071282e-06,
      "loss": 0.4566,
      "step": 278620
    },
    {
      "epoch": 114.05,
      "grad_norm": 1.754990816116333,
      "learning_rate": 1.9942491889859705e-06,
      "loss": 0.4663,
      "step": 278630
    },
    {
      "epoch": 114.06,
      "grad_norm": 1.5143803358078003,
      "learning_rate": 1.9940923338658224e-06,
      "loss": 0.4502,
      "step": 278640
    },
    {
      "epoch": 114.06,
      "grad_norm": 2.059495449066162,
      "learning_rate": 1.993935480347402e-06,
      "loss": 0.4713,
      "step": 278650
    },
    {
      "epoch": 114.06,
      "grad_norm": 1.8439733982086182,
      "learning_rate": 1.993778628431428e-06,
      "loss": 0.4579,
      "step": 278660
    },
    {
      "epoch": 114.07,
      "grad_norm": 2.1951098442077637,
      "learning_rate": 1.993621778118618e-06,
      "loss": 0.4499,
      "step": 278670
    },
    {
      "epoch": 114.07,
      "grad_norm": 2.0966460704803467,
      "learning_rate": 1.9934649294096914e-06,
      "loss": 0.4756,
      "step": 278680
    },
    {
      "epoch": 114.08,
      "grad_norm": 1.9465469121932983,
      "learning_rate": 1.9933080823053663e-06,
      "loss": 0.4769,
      "step": 278690
    },
    {
      "epoch": 114.08,
      "grad_norm": 2.019749879837036,
      "learning_rate": 1.9931512368063614e-06,
      "loss": 0.4721,
      "step": 278700
    },
    {
      "epoch": 114.09,
      "grad_norm": 1.9269967079162598,
      "learning_rate": 1.9929943929133952e-06,
      "loss": 0.4818,
      "step": 278710
    },
    {
      "epoch": 114.09,
      "grad_norm": 1.9803632497787476,
      "learning_rate": 1.9928375506271862e-06,
      "loss": 0.4716,
      "step": 278720
    },
    {
      "epoch": 114.09,
      "grad_norm": 2.404637575149536,
      "learning_rate": 1.9926807099484527e-06,
      "loss": 0.453,
      "step": 278730
    },
    {
      "epoch": 114.1,
      "grad_norm": 1.9583927392959595,
      "learning_rate": 1.9925238708779136e-06,
      "loss": 0.4681,
      "step": 278740
    },
    {
      "epoch": 114.1,
      "grad_norm": 2.0435004234313965,
      "learning_rate": 1.9923670334162865e-06,
      "loss": 0.4654,
      "step": 278750
    },
    {
      "epoch": 114.11,
      "grad_norm": 1.758815884590149,
      "learning_rate": 1.9922101975642906e-06,
      "loss": 0.4889,
      "step": 278760
    },
    {
      "epoch": 114.11,
      "grad_norm": 2.195641040802002,
      "learning_rate": 1.992053363322644e-06,
      "loss": 0.4527,
      "step": 278770
    },
    {
      "epoch": 114.11,
      "grad_norm": 1.9798120260238647,
      "learning_rate": 1.9918965306920643e-06,
      "loss": 0.4403,
      "step": 278780
    },
    {
      "epoch": 114.12,
      "grad_norm": 1.6760060787200928,
      "learning_rate": 1.9917396996732722e-06,
      "loss": 0.4737,
      "step": 278790
    },
    {
      "epoch": 114.12,
      "grad_norm": 2.017127513885498,
      "learning_rate": 1.9915828702669844e-06,
      "loss": 0.4423,
      "step": 278800
    },
    {
      "epoch": 114.13,
      "grad_norm": 5.466413974761963,
      "learning_rate": 1.9914260424739195e-06,
      "loss": 0.4801,
      "step": 278810
    },
    {
      "epoch": 114.13,
      "grad_norm": 1.704932451248169,
      "learning_rate": 1.9912692162947963e-06,
      "loss": 0.4622,
      "step": 278820
    },
    {
      "epoch": 114.13,
      "grad_norm": 2.1910932064056396,
      "learning_rate": 1.9911123917303326e-06,
      "loss": 0.4615,
      "step": 278830
    },
    {
      "epoch": 114.14,
      "grad_norm": 2.1620054244995117,
      "learning_rate": 1.9909555687812475e-06,
      "loss": 0.4705,
      "step": 278840
    },
    {
      "epoch": 114.14,
      "grad_norm": 2.0225305557250977,
      "learning_rate": 1.990798747448259e-06,
      "loss": 0.4622,
      "step": 278850
    },
    {
      "epoch": 114.15,
      "grad_norm": 1.743857502937317,
      "learning_rate": 1.9906419277320857e-06,
      "loss": 0.447,
      "step": 278860
    },
    {
      "epoch": 114.15,
      "grad_norm": 1.9620329141616821,
      "learning_rate": 1.990485109633446e-06,
      "loss": 0.4792,
      "step": 278870
    },
    {
      "epoch": 114.15,
      "grad_norm": 2.1991233825683594,
      "learning_rate": 1.9903282931530574e-06,
      "loss": 0.4614,
      "step": 278880
    },
    {
      "epoch": 114.16,
      "grad_norm": 2.2146170139312744,
      "learning_rate": 1.9901714782916395e-06,
      "loss": 0.4738,
      "step": 278890
    },
    {
      "epoch": 114.16,
      "grad_norm": 2.060753345489502,
      "learning_rate": 1.9900146650499095e-06,
      "loss": 0.4715,
      "step": 278900
    },
    {
      "epoch": 114.17,
      "grad_norm": 2.310164451599121,
      "learning_rate": 1.9898578534285865e-06,
      "loss": 0.4648,
      "step": 278910
    },
    {
      "epoch": 114.17,
      "grad_norm": 1.9123173952102661,
      "learning_rate": 1.989701043428389e-06,
      "loss": 0.455,
      "step": 278920
    },
    {
      "epoch": 114.18,
      "grad_norm": 1.736014723777771,
      "learning_rate": 1.989544235050034e-06,
      "loss": 0.4665,
      "step": 278930
    },
    {
      "epoch": 114.18,
      "grad_norm": 1.6054226160049438,
      "learning_rate": 1.9893874282942416e-06,
      "loss": 0.4712,
      "step": 278940
    },
    {
      "epoch": 114.18,
      "grad_norm": 1.8323689699172974,
      "learning_rate": 1.989230623161729e-06,
      "loss": 0.4708,
      "step": 278950
    },
    {
      "epoch": 114.19,
      "grad_norm": 1.4293559789657593,
      "learning_rate": 1.989073819653214e-06,
      "loss": 0.4764,
      "step": 278960
    },
    {
      "epoch": 114.19,
      "grad_norm": 1.8169810771942139,
      "learning_rate": 1.9889170177694167e-06,
      "loss": 0.4665,
      "step": 278970
    },
    {
      "epoch": 114.2,
      "grad_norm": 1.8681367635726929,
      "learning_rate": 1.988760217511054e-06,
      "loss": 0.4569,
      "step": 278980
    },
    {
      "epoch": 114.2,
      "grad_norm": 2.0571296215057373,
      "learning_rate": 1.9886034188788443e-06,
      "loss": 0.4693,
      "step": 278990
    },
    {
      "epoch": 114.2,
      "grad_norm": 2.0065274238586426,
      "learning_rate": 1.9884466218735065e-06,
      "loss": 0.4734,
      "step": 279000
    },
    {
      "epoch": 114.21,
      "grad_norm": 1.6078473329544067,
      "learning_rate": 1.9882898264957585e-06,
      "loss": 0.4517,
      "step": 279010
    },
    {
      "epoch": 114.21,
      "grad_norm": 1.8557639122009277,
      "learning_rate": 1.9881330327463182e-06,
      "loss": 0.4732,
      "step": 279020
    },
    {
      "epoch": 114.22,
      "grad_norm": 2.0890750885009766,
      "learning_rate": 1.9879762406259045e-06,
      "loss": 0.4491,
      "step": 279030
    },
    {
      "epoch": 114.22,
      "grad_norm": 2.0360610485076904,
      "learning_rate": 1.987819450135235e-06,
      "loss": 0.4702,
      "step": 279040
    },
    {
      "epoch": 114.22,
      "grad_norm": 1.6608059406280518,
      "learning_rate": 1.9876626612750282e-06,
      "loss": 0.4351,
      "step": 279050
    },
    {
      "epoch": 114.23,
      "grad_norm": 1.7722266912460327,
      "learning_rate": 1.9875058740460023e-06,
      "loss": 0.4705,
      "step": 279060
    },
    {
      "epoch": 114.23,
      "grad_norm": 1.53203547000885,
      "learning_rate": 1.9873490884488756e-06,
      "loss": 0.4469,
      "step": 279070
    },
    {
      "epoch": 114.24,
      "grad_norm": 1.954808235168457,
      "learning_rate": 1.9871923044843667e-06,
      "loss": 0.4535,
      "step": 279080
    },
    {
      "epoch": 114.24,
      "grad_norm": 2.1810550689697266,
      "learning_rate": 1.9870355221531928e-06,
      "loss": 0.4558,
      "step": 279090
    },
    {
      "epoch": 114.24,
      "grad_norm": 1.6006245613098145,
      "learning_rate": 1.9868787414560733e-06,
      "loss": 0.4624,
      "step": 279100
    },
    {
      "epoch": 114.25,
      "grad_norm": 1.8522052764892578,
      "learning_rate": 1.9867219623937257e-06,
      "loss": 0.4421,
      "step": 279110
    },
    {
      "epoch": 114.25,
      "grad_norm": 2.6071832180023193,
      "learning_rate": 1.986565184966868e-06,
      "loss": 0.4391,
      "step": 279120
    },
    {
      "epoch": 114.26,
      "grad_norm": 1.8001506328582764,
      "learning_rate": 1.9864084091762185e-06,
      "loss": 0.4834,
      "step": 279130
    },
    {
      "epoch": 114.26,
      "grad_norm": 2.2092702388763428,
      "learning_rate": 1.9862516350224954e-06,
      "loss": 0.4611,
      "step": 279140
    },
    {
      "epoch": 114.27,
      "grad_norm": 1.859744906425476,
      "learning_rate": 1.9860948625064166e-06,
      "loss": 0.4621,
      "step": 279150
    },
    {
      "epoch": 114.27,
      "grad_norm": 1.9415795803070068,
      "learning_rate": 1.985938091628701e-06,
      "loss": 0.4493,
      "step": 279160
    },
    {
      "epoch": 114.27,
      "grad_norm": 2.0224673748016357,
      "learning_rate": 1.9857813223900666e-06,
      "loss": 0.431,
      "step": 279170
    },
    {
      "epoch": 114.28,
      "grad_norm": 2.2892141342163086,
      "learning_rate": 1.985624554791231e-06,
      "loss": 0.4797,
      "step": 279180
    },
    {
      "epoch": 114.28,
      "grad_norm": 1.9527980089187622,
      "learning_rate": 1.985467788832913e-06,
      "loss": 0.4893,
      "step": 279190
    },
    {
      "epoch": 114.29,
      "grad_norm": 2.518392324447632,
      "learning_rate": 1.9853110245158296e-06,
      "loss": 0.4479,
      "step": 279200
    },
    {
      "epoch": 114.29,
      "grad_norm": 1.8325529098510742,
      "learning_rate": 1.9851542618407e-06,
      "loss": 0.4711,
      "step": 279210
    },
    {
      "epoch": 114.29,
      "grad_norm": 1.9320898056030273,
      "learning_rate": 1.984997500808242e-06,
      "loss": 0.4775,
      "step": 279220
    },
    {
      "epoch": 114.3,
      "grad_norm": 1.7847857475280762,
      "learning_rate": 1.984840741419174e-06,
      "loss": 0.4595,
      "step": 279230
    },
    {
      "epoch": 114.3,
      "grad_norm": 2.2779078483581543,
      "learning_rate": 1.9846839836742135e-06,
      "loss": 0.4733,
      "step": 279240
    },
    {
      "epoch": 114.31,
      "grad_norm": 1.7344690561294556,
      "learning_rate": 1.9845272275740787e-06,
      "loss": 0.4741,
      "step": 279250
    },
    {
      "epoch": 114.31,
      "grad_norm": 2.1225833892822266,
      "learning_rate": 1.984370473119488e-06,
      "loss": 0.4549,
      "step": 279260
    },
    {
      "epoch": 114.31,
      "grad_norm": 1.6340121030807495,
      "learning_rate": 1.984213720311159e-06,
      "loss": 0.4592,
      "step": 279270
    },
    {
      "epoch": 114.32,
      "grad_norm": 1.849982500076294,
      "learning_rate": 1.98405696914981e-06,
      "loss": 0.4694,
      "step": 279280
    },
    {
      "epoch": 114.32,
      "grad_norm": 1.719970941543579,
      "learning_rate": 1.983900219636159e-06,
      "loss": 0.4796,
      "step": 279290
    },
    {
      "epoch": 114.33,
      "grad_norm": 2.4680917263031006,
      "learning_rate": 1.983743471770924e-06,
      "loss": 0.4614,
      "step": 279300
    },
    {
      "epoch": 114.33,
      "grad_norm": 1.8344742059707642,
      "learning_rate": 1.9835867255548235e-06,
      "loss": 0.4794,
      "step": 279310
    },
    {
      "epoch": 114.33,
      "grad_norm": 1.9222174882888794,
      "learning_rate": 1.9834299809885754e-06,
      "loss": 0.4508,
      "step": 279320
    },
    {
      "epoch": 114.34,
      "grad_norm": 2.350173234939575,
      "learning_rate": 1.9832732380728976e-06,
      "loss": 0.4587,
      "step": 279330
    },
    {
      "epoch": 114.34,
      "grad_norm": 2.3435442447662354,
      "learning_rate": 1.9831164968085063e-06,
      "loss": 0.4693,
      "step": 279340
    },
    {
      "epoch": 114.35,
      "grad_norm": 2.114921808242798,
      "learning_rate": 1.9829597571961224e-06,
      "loss": 0.4661,
      "step": 279350
    },
    {
      "epoch": 114.35,
      "grad_norm": 2.0971200466156006,
      "learning_rate": 1.9828030192364627e-06,
      "loss": 0.4544,
      "step": 279360
    },
    {
      "epoch": 114.36,
      "grad_norm": 2.0279979705810547,
      "learning_rate": 1.982646282930246e-06,
      "loss": 0.4702,
      "step": 279370
    },
    {
      "epoch": 114.36,
      "grad_norm": 2.0484445095062256,
      "learning_rate": 1.9824895482781888e-06,
      "loss": 0.4711,
      "step": 279380
    },
    {
      "epoch": 114.36,
      "grad_norm": 1.9832799434661865,
      "learning_rate": 1.9823328152810097e-06,
      "loss": 0.4464,
      "step": 279390
    },
    {
      "epoch": 114.37,
      "grad_norm": 2.0716893672943115,
      "learning_rate": 1.9821760839394273e-06,
      "loss": 0.4701,
      "step": 279400
    },
    {
      "epoch": 114.37,
      "grad_norm": 1.7248433828353882,
      "learning_rate": 1.9820193542541585e-06,
      "loss": 0.4649,
      "step": 279410
    },
    {
      "epoch": 114.38,
      "grad_norm": 1.918617606163025,
      "learning_rate": 1.9818626262259217e-06,
      "loss": 0.4551,
      "step": 279420
    },
    {
      "epoch": 114.38,
      "grad_norm": 2.0015861988067627,
      "learning_rate": 1.981705899855435e-06,
      "loss": 0.4803,
      "step": 279430
    },
    {
      "epoch": 114.38,
      "grad_norm": 1.9586360454559326,
      "learning_rate": 1.9815491751434164e-06,
      "loss": 0.4646,
      "step": 279440
    },
    {
      "epoch": 114.39,
      "grad_norm": 2.128755807876587,
      "learning_rate": 1.9813924520905838e-06,
      "loss": 0.4563,
      "step": 279450
    },
    {
      "epoch": 114.39,
      "grad_norm": 2.209998846054077,
      "learning_rate": 1.9812357306976553e-06,
      "loss": 0.4623,
      "step": 279460
    },
    {
      "epoch": 114.4,
      "grad_norm": 1.8960151672363281,
      "learning_rate": 1.981079010965348e-06,
      "loss": 0.4827,
      "step": 279470
    },
    {
      "epoch": 114.4,
      "grad_norm": 2.0050008296966553,
      "learning_rate": 1.9809222928943803e-06,
      "loss": 0.4751,
      "step": 279480
    },
    {
      "epoch": 114.4,
      "grad_norm": 2.1794190406799316,
      "learning_rate": 1.9807655764854704e-06,
      "loss": 0.4472,
      "step": 279490
    },
    {
      "epoch": 114.41,
      "grad_norm": 1.6490899324417114,
      "learning_rate": 1.9806088617393357e-06,
      "loss": 0.4989,
      "step": 279500
    },
    {
      "epoch": 114.41,
      "grad_norm": 1.7869160175323486,
      "learning_rate": 1.980452148656694e-06,
      "loss": 0.4661,
      "step": 279510
    },
    {
      "epoch": 114.42,
      "grad_norm": 1.7562981843948364,
      "learning_rate": 1.9802954372382638e-06,
      "loss": 0.4537,
      "step": 279520
    },
    {
      "epoch": 114.42,
      "grad_norm": 2.124569892883301,
      "learning_rate": 1.9801387274847628e-06,
      "loss": 0.4921,
      "step": 279530
    },
    {
      "epoch": 114.42,
      "grad_norm": 1.8696597814559937,
      "learning_rate": 1.9799820193969084e-06,
      "loss": 0.4731,
      "step": 279540
    },
    {
      "epoch": 114.43,
      "grad_norm": 1.7484322786331177,
      "learning_rate": 1.979825312975419e-06,
      "loss": 0.4866,
      "step": 279550
    },
    {
      "epoch": 114.43,
      "grad_norm": 1.874707818031311,
      "learning_rate": 1.9796686082210124e-06,
      "loss": 0.4477,
      "step": 279560
    },
    {
      "epoch": 114.44,
      "grad_norm": 1.853238821029663,
      "learning_rate": 1.9795119051344056e-06,
      "loss": 0.4668,
      "step": 279570
    },
    {
      "epoch": 114.44,
      "grad_norm": 1.713163137435913,
      "learning_rate": 1.979355203716318e-06,
      "loss": 0.4427,
      "step": 279580
    },
    {
      "epoch": 114.45,
      "grad_norm": 1.9323347806930542,
      "learning_rate": 1.9791985039674664e-06,
      "loss": 0.4921,
      "step": 279590
    },
    {
      "epoch": 114.45,
      "grad_norm": 1.88395357131958,
      "learning_rate": 1.979041805888569e-06,
      "loss": 0.4718,
      "step": 279600
    },
    {
      "epoch": 114.45,
      "grad_norm": 1.7212742567062378,
      "learning_rate": 1.9788851094803425e-06,
      "loss": 0.4845,
      "step": 279610
    },
    {
      "epoch": 114.46,
      "grad_norm": 3.34328031539917,
      "learning_rate": 1.978728414743506e-06,
      "loss": 0.4591,
      "step": 279620
    },
    {
      "epoch": 114.46,
      "grad_norm": 1.993181586265564,
      "learning_rate": 1.9785717216787764e-06,
      "loss": 0.4808,
      "step": 279630
    },
    {
      "epoch": 114.47,
      "grad_norm": 1.8784643411636353,
      "learning_rate": 1.9784150302868728e-06,
      "loss": 0.4434,
      "step": 279640
    },
    {
      "epoch": 114.47,
      "grad_norm": 2.3219656944274902,
      "learning_rate": 1.9782583405685114e-06,
      "loss": 0.4742,
      "step": 279650
    },
    {
      "epoch": 114.47,
      "grad_norm": 1.8466236591339111,
      "learning_rate": 1.978101652524411e-06,
      "loss": 0.478,
      "step": 279660
    },
    {
      "epoch": 114.48,
      "grad_norm": 2.031655788421631,
      "learning_rate": 1.977944966155289e-06,
      "loss": 0.4671,
      "step": 279670
    },
    {
      "epoch": 114.48,
      "grad_norm": 1.6964679956436157,
      "learning_rate": 1.9777882814618634e-06,
      "loss": 0.4721,
      "step": 279680
    },
    {
      "epoch": 114.49,
      "grad_norm": 1.6228200197219849,
      "learning_rate": 1.977631598444852e-06,
      "loss": 0.478,
      "step": 279690
    },
    {
      "epoch": 114.49,
      "grad_norm": 1.9558804035186768,
      "learning_rate": 1.9774749171049716e-06,
      "loss": 0.4664,
      "step": 279700
    },
    {
      "epoch": 114.49,
      "grad_norm": 2.489842414855957,
      "learning_rate": 1.977318237442941e-06,
      "loss": 0.4693,
      "step": 279710
    },
    {
      "epoch": 114.5,
      "grad_norm": 1.9649006128311157,
      "learning_rate": 1.9771615594594763e-06,
      "loss": 0.4625,
      "step": 279720
    },
    {
      "epoch": 114.5,
      "grad_norm": 1.8731908798217773,
      "learning_rate": 1.977004883155298e-06,
      "loss": 0.4587,
      "step": 279730
    },
    {
      "epoch": 114.51,
      "grad_norm": 1.7966138124465942,
      "learning_rate": 1.9768482085311217e-06,
      "loss": 0.4701,
      "step": 279740
    },
    {
      "epoch": 114.51,
      "grad_norm": 1.8463486433029175,
      "learning_rate": 1.976691535587666e-06,
      "loss": 0.4545,
      "step": 279750
    },
    {
      "epoch": 114.51,
      "grad_norm": 2.0307743549346924,
      "learning_rate": 1.976534864325648e-06,
      "loss": 0.5075,
      "step": 279760
    },
    {
      "epoch": 114.52,
      "grad_norm": 2.011167049407959,
      "learning_rate": 1.976378194745786e-06,
      "loss": 0.4485,
      "step": 279770
    },
    {
      "epoch": 114.52,
      "grad_norm": 2.283128023147583,
      "learning_rate": 1.976221526848797e-06,
      "loss": 0.469,
      "step": 279780
    },
    {
      "epoch": 114.53,
      "grad_norm": 2.0608580112457275,
      "learning_rate": 1.9760648606353993e-06,
      "loss": 0.454,
      "step": 279790
    },
    {
      "epoch": 114.53,
      "grad_norm": 2.013906717300415,
      "learning_rate": 1.97590819610631e-06,
      "loss": 0.4688,
      "step": 279800
    },
    {
      "epoch": 114.54,
      "grad_norm": 2.5610103607177734,
      "learning_rate": 1.9757515332622474e-06,
      "loss": 0.4734,
      "step": 279810
    },
    {
      "epoch": 114.54,
      "grad_norm": 1.913374662399292,
      "learning_rate": 1.975594872103929e-06,
      "loss": 0.4809,
      "step": 279820
    },
    {
      "epoch": 114.54,
      "grad_norm": 2.060168981552124,
      "learning_rate": 1.9754382126320715e-06,
      "loss": 0.4375,
      "step": 279830
    },
    {
      "epoch": 114.55,
      "grad_norm": 1.9635894298553467,
      "learning_rate": 1.975281554847394e-06,
      "loss": 0.4658,
      "step": 279840
    },
    {
      "epoch": 114.55,
      "grad_norm": 2.2721526622772217,
      "learning_rate": 1.9751248987506125e-06,
      "loss": 0.4815,
      "step": 279850
    },
    {
      "epoch": 114.56,
      "grad_norm": 1.9595321416854858,
      "learning_rate": 1.9749682443424463e-06,
      "loss": 0.4808,
      "step": 279860
    },
    {
      "epoch": 114.56,
      "grad_norm": 2.1517350673675537,
      "learning_rate": 1.9748115916236114e-06,
      "loss": 0.4545,
      "step": 279870
    },
    {
      "epoch": 114.56,
      "grad_norm": 1.8555501699447632,
      "learning_rate": 1.9746549405948267e-06,
      "loss": 0.4558,
      "step": 279880
    },
    {
      "epoch": 114.57,
      "grad_norm": 1.9651521444320679,
      "learning_rate": 1.9744982912568095e-06,
      "loss": 0.4856,
      "step": 279890
    },
    {
      "epoch": 114.57,
      "grad_norm": 1.7147222757339478,
      "learning_rate": 1.9743416436102774e-06,
      "loss": 0.4486,
      "step": 279900
    },
    {
      "epoch": 114.58,
      "grad_norm": 1.9852054119110107,
      "learning_rate": 1.9741849976559463e-06,
      "loss": 0.4292,
      "step": 279910
    },
    {
      "epoch": 114.58,
      "grad_norm": 2.1183066368103027,
      "learning_rate": 1.974028353394536e-06,
      "loss": 0.4923,
      "step": 279920
    },
    {
      "epoch": 114.58,
      "grad_norm": 1.948196291923523,
      "learning_rate": 1.973871710826764e-06,
      "loss": 0.4558,
      "step": 279930
    },
    {
      "epoch": 114.59,
      "grad_norm": 1.9871963262557983,
      "learning_rate": 1.9737150699533464e-06,
      "loss": 0.4871,
      "step": 279940
    },
    {
      "epoch": 114.59,
      "grad_norm": 3.1973137855529785,
      "learning_rate": 1.9735584307750023e-06,
      "loss": 0.4869,
      "step": 279950
    },
    {
      "epoch": 114.6,
      "grad_norm": 2.0400145053863525,
      "learning_rate": 1.973401793292448e-06,
      "loss": 0.4756,
      "step": 279960
    },
    {
      "epoch": 114.6,
      "grad_norm": 1.9516247510910034,
      "learning_rate": 1.9732451575064014e-06,
      "loss": 0.4554,
      "step": 279970
    },
    {
      "epoch": 114.6,
      "grad_norm": 2.2665486335754395,
      "learning_rate": 1.9730885234175802e-06,
      "loss": 0.4714,
      "step": 279980
    },
    {
      "epoch": 114.61,
      "grad_norm": 2.026179552078247,
      "learning_rate": 1.972931891026702e-06,
      "loss": 0.4683,
      "step": 279990
    },
    {
      "epoch": 114.61,
      "grad_norm": 2.4453139305114746,
      "learning_rate": 1.9727752603344836e-06,
      "loss": 0.4811,
      "step": 280000
    },
    {
      "epoch": 114.62,
      "grad_norm": 2.063210964202881,
      "learning_rate": 1.9726186313416434e-06,
      "loss": 0.4545,
      "step": 280010
    },
    {
      "epoch": 114.62,
      "grad_norm": 1.9304804801940918,
      "learning_rate": 1.9724620040488984e-06,
      "loss": 0.473,
      "step": 280020
    },
    {
      "epoch": 114.63,
      "grad_norm": 2.200728416442871,
      "learning_rate": 1.9723053784569666e-06,
      "loss": 0.465,
      "step": 280030
    },
    {
      "epoch": 114.63,
      "grad_norm": 2.028134822845459,
      "learning_rate": 1.9721487545665653e-06,
      "loss": 0.4488,
      "step": 280040
    },
    {
      "epoch": 114.63,
      "grad_norm": 1.8922783136367798,
      "learning_rate": 1.971992132378411e-06,
      "loss": 0.4504,
      "step": 280050
    },
    {
      "epoch": 114.64,
      "grad_norm": 2.027888536453247,
      "learning_rate": 1.971835511893222e-06,
      "loss": 0.4692,
      "step": 280060
    },
    {
      "epoch": 114.64,
      "grad_norm": 2.0114946365356445,
      "learning_rate": 1.9716788931117157e-06,
      "loss": 0.4662,
      "step": 280070
    },
    {
      "epoch": 114.65,
      "grad_norm": 1.9900819063186646,
      "learning_rate": 1.9715222760346098e-06,
      "loss": 0.4743,
      "step": 280080
    },
    {
      "epoch": 114.65,
      "grad_norm": 2.0157766342163086,
      "learning_rate": 1.9713656606626214e-06,
      "loss": 0.4542,
      "step": 280090
    },
    {
      "epoch": 114.65,
      "grad_norm": 2.173365592956543,
      "learning_rate": 1.971209046996467e-06,
      "loss": 0.4808,
      "step": 280100
    },
    {
      "epoch": 114.66,
      "grad_norm": 2.361712694168091,
      "learning_rate": 1.971052435036866e-06,
      "loss": 0.4564,
      "step": 280110
    },
    {
      "epoch": 114.66,
      "grad_norm": 1.942796230316162,
      "learning_rate": 1.9708958247845348e-06,
      "loss": 0.4627,
      "step": 280120
    },
    {
      "epoch": 114.67,
      "grad_norm": 1.7231637239456177,
      "learning_rate": 1.9707392162401906e-06,
      "loss": 0.4616,
      "step": 280130
    },
    {
      "epoch": 114.67,
      "grad_norm": 1.9049524068832397,
      "learning_rate": 1.9705826094045516e-06,
      "loss": 0.4671,
      "step": 280140
    },
    {
      "epoch": 114.67,
      "grad_norm": 3.032320022583008,
      "learning_rate": 1.970426004278334e-06,
      "loss": 0.4592,
      "step": 280150
    },
    {
      "epoch": 114.68,
      "grad_norm": 2.060467004776001,
      "learning_rate": 1.970269400862256e-06,
      "loss": 0.4486,
      "step": 280160
    },
    {
      "epoch": 114.68,
      "grad_norm": 2.2423315048217773,
      "learning_rate": 1.9701127991570343e-06,
      "loss": 0.493,
      "step": 280170
    },
    {
      "epoch": 114.69,
      "grad_norm": 4.504210472106934,
      "learning_rate": 1.969956199163388e-06,
      "loss": 0.4652,
      "step": 280180
    },
    {
      "epoch": 114.69,
      "grad_norm": 1.7526984214782715,
      "learning_rate": 1.9697996008820323e-06,
      "loss": 0.4735,
      "step": 280190
    },
    {
      "epoch": 114.7,
      "grad_norm": 2.1676154136657715,
      "learning_rate": 1.969643004313686e-06,
      "loss": 0.451,
      "step": 280200
    },
    {
      "epoch": 114.7,
      "grad_norm": 1.670623540878296,
      "learning_rate": 1.9694864094590653e-06,
      "loss": 0.4806,
      "step": 280210
    },
    {
      "epoch": 114.7,
      "grad_norm": 2.4308860301971436,
      "learning_rate": 1.969329816318888e-06,
      "loss": 0.4595,
      "step": 280220
    },
    {
      "epoch": 114.71,
      "grad_norm": 1.775749921798706,
      "learning_rate": 1.9691732248938726e-06,
      "loss": 0.444,
      "step": 280230
    },
    {
      "epoch": 114.71,
      "grad_norm": 2.184044361114502,
      "learning_rate": 1.9690166351847345e-06,
      "loss": 0.4668,
      "step": 280240
    },
    {
      "epoch": 114.72,
      "grad_norm": 1.9465265274047852,
      "learning_rate": 1.968860047192192e-06,
      "loss": 0.455,
      "step": 280250
    },
    {
      "epoch": 114.72,
      "grad_norm": 1.9264637231826782,
      "learning_rate": 1.968703460916963e-06,
      "loss": 0.4558,
      "step": 280260
    },
    {
      "epoch": 114.72,
      "grad_norm": 1.8324893712997437,
      "learning_rate": 1.9685468763597638e-06,
      "loss": 0.465,
      "step": 280270
    },
    {
      "epoch": 114.73,
      "grad_norm": 1.8480547666549683,
      "learning_rate": 1.9683902935213106e-06,
      "loss": 0.4499,
      "step": 280280
    },
    {
      "epoch": 114.73,
      "grad_norm": 2.2660653591156006,
      "learning_rate": 1.9682337124023232e-06,
      "loss": 0.4428,
      "step": 280290
    },
    {
      "epoch": 114.74,
      "grad_norm": 1.7329750061035156,
      "learning_rate": 1.9680771330035175e-06,
      "loss": 0.4901,
      "step": 280300
    },
    {
      "epoch": 114.74,
      "grad_norm": 1.896342396736145,
      "learning_rate": 1.967920555325612e-06,
      "loss": 0.4776,
      "step": 280310
    },
    {
      "epoch": 114.74,
      "grad_norm": 1.746045708656311,
      "learning_rate": 1.9677639793693224e-06,
      "loss": 0.4579,
      "step": 280320
    },
    {
      "epoch": 114.75,
      "grad_norm": 1.6877176761627197,
      "learning_rate": 1.9676074051353665e-06,
      "loss": 0.4852,
      "step": 280330
    },
    {
      "epoch": 114.75,
      "grad_norm": 2.0378963947296143,
      "learning_rate": 1.9674508326244617e-06,
      "loss": 0.4585,
      "step": 280340
    },
    {
      "epoch": 114.76,
      "grad_norm": 2.3132598400115967,
      "learning_rate": 1.9672942618373247e-06,
      "loss": 0.468,
      "step": 280350
    },
    {
      "epoch": 114.76,
      "grad_norm": 2.1036629676818848,
      "learning_rate": 1.967137692774673e-06,
      "loss": 0.4754,
      "step": 280360
    },
    {
      "epoch": 114.76,
      "grad_norm": 1.6840250492095947,
      "learning_rate": 1.966981125437225e-06,
      "loss": 0.4351,
      "step": 280370
    },
    {
      "epoch": 114.77,
      "grad_norm": 1.6573768854141235,
      "learning_rate": 1.966824559825696e-06,
      "loss": 0.4702,
      "step": 280380
    },
    {
      "epoch": 114.77,
      "grad_norm": 1.987458348274231,
      "learning_rate": 1.9666679959408044e-06,
      "loss": 0.463,
      "step": 280390
    },
    {
      "epoch": 114.78,
      "grad_norm": 1.6358466148376465,
      "learning_rate": 1.966511433783267e-06,
      "loss": 0.473,
      "step": 280400
    },
    {
      "epoch": 114.78,
      "grad_norm": 1.9343839883804321,
      "learning_rate": 1.966354873353801e-06,
      "loss": 0.4488,
      "step": 280410
    },
    {
      "epoch": 114.79,
      "grad_norm": 1.886836051940918,
      "learning_rate": 1.9661983146531237e-06,
      "loss": 0.4643,
      "step": 280420
    },
    {
      "epoch": 114.79,
      "grad_norm": 2.066743850708008,
      "learning_rate": 1.9660417576819522e-06,
      "loss": 0.4523,
      "step": 280430
    },
    {
      "epoch": 114.79,
      "grad_norm": 2.1214561462402344,
      "learning_rate": 1.9658852024410035e-06,
      "loss": 0.475,
      "step": 280440
    },
    {
      "epoch": 114.8,
      "grad_norm": 2.154759645462036,
      "learning_rate": 1.965728648930995e-06,
      "loss": 0.4653,
      "step": 280450
    },
    {
      "epoch": 114.8,
      "grad_norm": 1.967767596244812,
      "learning_rate": 1.9655720971526436e-06,
      "loss": 0.4531,
      "step": 280460
    },
    {
      "epoch": 114.81,
      "grad_norm": 1.7189453840255737,
      "learning_rate": 1.9654155471066665e-06,
      "loss": 0.4765,
      "step": 280470
    },
    {
      "epoch": 114.81,
      "grad_norm": 1.9457504749298096,
      "learning_rate": 1.965258998793781e-06,
      "loss": 0.4908,
      "step": 280480
    },
    {
      "epoch": 114.81,
      "grad_norm": 2.0085721015930176,
      "learning_rate": 1.9651024522147043e-06,
      "loss": 0.4562,
      "step": 280490
    },
    {
      "epoch": 114.82,
      "grad_norm": 2.083474636077881,
      "learning_rate": 1.9649459073701536e-06,
      "loss": 0.4627,
      "step": 280500
    },
    {
      "epoch": 114.82,
      "grad_norm": 1.9106427431106567,
      "learning_rate": 1.9647893642608453e-06,
      "loss": 0.4692,
      "step": 280510
    },
    {
      "epoch": 114.83,
      "grad_norm": 1.673085331916809,
      "learning_rate": 1.964632822887497e-06,
      "loss": 0.4803,
      "step": 280520
    },
    {
      "epoch": 114.83,
      "grad_norm": 2.2385435104370117,
      "learning_rate": 1.964476283250827e-06,
      "loss": 0.4771,
      "step": 280530
    },
    {
      "epoch": 114.83,
      "grad_norm": 1.878014087677002,
      "learning_rate": 1.96431974535155e-06,
      "loss": 0.4579,
      "step": 280540
    },
    {
      "epoch": 114.84,
      "grad_norm": 1.8382487297058105,
      "learning_rate": 1.9641632091903847e-06,
      "loss": 0.4685,
      "step": 280550
    },
    {
      "epoch": 114.84,
      "grad_norm": 2.2769477367401123,
      "learning_rate": 1.964006674768047e-06,
      "loss": 0.469,
      "step": 280560
    },
    {
      "epoch": 114.85,
      "grad_norm": 1.8987759351730347,
      "learning_rate": 1.9638501420852557e-06,
      "loss": 0.4634,
      "step": 280570
    },
    {
      "epoch": 114.85,
      "grad_norm": 2.053385019302368,
      "learning_rate": 1.963693611142726e-06,
      "loss": 0.477,
      "step": 280580
    },
    {
      "epoch": 114.85,
      "grad_norm": 2.164339303970337,
      "learning_rate": 1.963537081941176e-06,
      "loss": 0.4675,
      "step": 280590
    },
    {
      "epoch": 114.86,
      "grad_norm": 1.6949541568756104,
      "learning_rate": 1.9633805544813223e-06,
      "loss": 0.4502,
      "step": 280600
    },
    {
      "epoch": 114.86,
      "grad_norm": 2.2849411964416504,
      "learning_rate": 1.9632240287638823e-06,
      "loss": 0.4513,
      "step": 280610
    },
    {
      "epoch": 114.87,
      "grad_norm": 1.8383610248565674,
      "learning_rate": 1.9630675047895734e-06,
      "loss": 0.454,
      "step": 280620
    },
    {
      "epoch": 114.87,
      "grad_norm": 2.0395116806030273,
      "learning_rate": 1.9629109825591113e-06,
      "loss": 0.4788,
      "step": 280630
    },
    {
      "epoch": 114.88,
      "grad_norm": 1.983638048171997,
      "learning_rate": 1.9627544620732144e-06,
      "loss": 0.4852,
      "step": 280640
    },
    {
      "epoch": 114.88,
      "grad_norm": 2.1310360431671143,
      "learning_rate": 1.9625979433325984e-06,
      "loss": 0.4688,
      "step": 280650
    },
    {
      "epoch": 114.88,
      "grad_norm": 1.7028465270996094,
      "learning_rate": 1.96244142633798e-06,
      "loss": 0.4602,
      "step": 280660
    },
    {
      "epoch": 114.89,
      "grad_norm": 2.3388631343841553,
      "learning_rate": 1.962284911090079e-06,
      "loss": 0.4519,
      "step": 280670
    },
    {
      "epoch": 114.89,
      "grad_norm": 2.1966841220855713,
      "learning_rate": 1.9621283975896097e-06,
      "loss": 0.45,
      "step": 280680
    },
    {
      "epoch": 114.9,
      "grad_norm": 1.6037437915802002,
      "learning_rate": 1.96197188583729e-06,
      "loss": 0.4741,
      "step": 280690
    },
    {
      "epoch": 114.9,
      "grad_norm": 1.687865138053894,
      "learning_rate": 1.9618153758338366e-06,
      "loss": 0.4585,
      "step": 280700
    },
    {
      "epoch": 114.9,
      "grad_norm": 2.3588521480560303,
      "learning_rate": 1.9616588675799666e-06,
      "loss": 0.488,
      "step": 280710
    },
    {
      "epoch": 114.91,
      "grad_norm": 2.2588043212890625,
      "learning_rate": 1.9615023610763966e-06,
      "loss": 0.4554,
      "step": 280720
    },
    {
      "epoch": 114.91,
      "grad_norm": 1.875145435333252,
      "learning_rate": 1.9613458563238445e-06,
      "loss": 0.4614,
      "step": 280730
    },
    {
      "epoch": 114.92,
      "grad_norm": 1.947180986404419,
      "learning_rate": 1.9611893533230264e-06,
      "loss": 0.4566,
      "step": 280740
    },
    {
      "epoch": 114.92,
      "grad_norm": 1.8064134120941162,
      "learning_rate": 1.961032852074659e-06,
      "loss": 0.4987,
      "step": 280750
    },
    {
      "epoch": 114.92,
      "grad_norm": 2.145360231399536,
      "learning_rate": 1.9608763525794602e-06,
      "loss": 0.4623,
      "step": 280760
    },
    {
      "epoch": 114.93,
      "grad_norm": 2.097130060195923,
      "learning_rate": 1.960719854838146e-06,
      "loss": 0.4429,
      "step": 280770
    },
    {
      "epoch": 114.93,
      "grad_norm": 2.104461431503296,
      "learning_rate": 1.960563358851433e-06,
      "loss": 0.4684,
      "step": 280780
    },
    {
      "epoch": 114.94,
      "grad_norm": 1.6333917379379272,
      "learning_rate": 1.9604068646200394e-06,
      "loss": 0.4654,
      "step": 280790
    },
    {
      "epoch": 114.94,
      "grad_norm": 2.317389726638794,
      "learning_rate": 1.960250372144681e-06,
      "loss": 0.4542,
      "step": 280800
    },
    {
      "epoch": 114.94,
      "grad_norm": 1.8709698915481567,
      "learning_rate": 1.960093881426075e-06,
      "loss": 0.4757,
      "step": 280810
    },
    {
      "epoch": 114.95,
      "grad_norm": 1.9562816619873047,
      "learning_rate": 1.9599373924649383e-06,
      "loss": 0.4801,
      "step": 280820
    },
    {
      "epoch": 114.95,
      "grad_norm": 2.882140636444092,
      "learning_rate": 1.9597809052619877e-06,
      "loss": 0.452,
      "step": 280830
    },
    {
      "epoch": 114.96,
      "grad_norm": 1.8406695127487183,
      "learning_rate": 1.9596244198179403e-06,
      "loss": 0.4757,
      "step": 280840
    },
    {
      "epoch": 114.96,
      "grad_norm": 1.6779136657714844,
      "learning_rate": 1.9594679361335116e-06,
      "loss": 0.4564,
      "step": 280850
    },
    {
      "epoch": 114.97,
      "grad_norm": 2.056488037109375,
      "learning_rate": 1.9593114542094203e-06,
      "loss": 0.4846,
      "step": 280860
    },
    {
      "epoch": 114.97,
      "grad_norm": 1.9614263772964478,
      "learning_rate": 1.959154974046383e-06,
      "loss": 0.4659,
      "step": 280870
    },
    {
      "epoch": 114.97,
      "grad_norm": 2.381875514984131,
      "learning_rate": 1.958998495645115e-06,
      "loss": 0.4787,
      "step": 280880
    },
    {
      "epoch": 114.98,
      "grad_norm": 1.5602339506149292,
      "learning_rate": 1.9588420190063352e-06,
      "loss": 0.4723,
      "step": 280890
    },
    {
      "epoch": 114.98,
      "grad_norm": 2.0227153301239014,
      "learning_rate": 1.9586855441307582e-06,
      "loss": 0.4766,
      "step": 280900
    },
    {
      "epoch": 114.99,
      "grad_norm": 2.0396406650543213,
      "learning_rate": 1.9585290710191024e-06,
      "loss": 0.4579,
      "step": 280910
    },
    {
      "epoch": 114.99,
      "grad_norm": 1.7361170053482056,
      "learning_rate": 1.9583725996720838e-06,
      "loss": 0.456,
      "step": 280920
    },
    {
      "epoch": 114.99,
      "grad_norm": 1.9281355142593384,
      "learning_rate": 1.958216130090419e-06,
      "loss": 0.4831,
      "step": 280930
    },
    {
      "epoch": 115.0,
      "grad_norm": 2.2371182441711426,
      "learning_rate": 1.958059662274826e-06,
      "loss": 0.4688,
      "step": 280940
    },
    {
      "epoch": 115.0,
      "eval_loss": 0.46742546558380127,
      "eval_runtime": 54.7129,
      "eval_samples_per_second": 63.038,
      "eval_steps_per_second": 7.896,
      "step": 280945
    },
    {
      "epoch": 115.0,
      "grad_norm": 1.8635140657424927,
      "learning_rate": 1.9579031962260202e-06,
      "loss": 0.4726,
      "step": 280950
    },
    {
      "epoch": 115.01,
      "grad_norm": 1.9070249795913696,
      "learning_rate": 1.957746731944719e-06,
      "loss": 0.4487,
      "step": 280960
    },
    {
      "epoch": 115.01,
      "grad_norm": 1.6892098188400269,
      "learning_rate": 1.957590269431639e-06,
      "loss": 0.4664,
      "step": 280970
    },
    {
      "epoch": 115.01,
      "grad_norm": 2.5008301734924316,
      "learning_rate": 1.957433808687497e-06,
      "loss": 0.4661,
      "step": 280980
    },
    {
      "epoch": 115.02,
      "grad_norm": 1.9254543781280518,
      "learning_rate": 1.9572773497130097e-06,
      "loss": 0.4825,
      "step": 280990
    },
    {
      "epoch": 115.02,
      "grad_norm": 1.8120231628417969,
      "learning_rate": 1.9571208925088934e-06,
      "loss": 0.4665,
      "step": 281000
    },
    {
      "epoch": 115.03,
      "grad_norm": 2.7864744663238525,
      "learning_rate": 1.9569644370758655e-06,
      "loss": 0.4639,
      "step": 281010
    },
    {
      "epoch": 115.03,
      "grad_norm": 1.8807692527770996,
      "learning_rate": 1.956807983414642e-06,
      "loss": 0.453,
      "step": 281020
    },
    {
      "epoch": 115.03,
      "grad_norm": 1.9607855081558228,
      "learning_rate": 1.9566515315259404e-06,
      "loss": 0.4688,
      "step": 281030
    },
    {
      "epoch": 115.04,
      "grad_norm": 1.7619858980178833,
      "learning_rate": 1.956495081410476e-06,
      "loss": 0.454,
      "step": 281040
    },
    {
      "epoch": 115.04,
      "grad_norm": 1.837013840675354,
      "learning_rate": 1.9563386330689668e-06,
      "loss": 0.4389,
      "step": 281050
    },
    {
      "epoch": 115.05,
      "grad_norm": 1.7992033958435059,
      "learning_rate": 1.9561821865021296e-06,
      "loss": 0.4888,
      "step": 281060
    },
    {
      "epoch": 115.05,
      "grad_norm": 1.9383095502853394,
      "learning_rate": 1.95602574171068e-06,
      "loss": 0.4637,
      "step": 281070
    },
    {
      "epoch": 115.06,
      "grad_norm": 1.6191076040267944,
      "learning_rate": 1.9558692986953357e-06,
      "loss": 0.4864,
      "step": 281080
    },
    {
      "epoch": 115.06,
      "grad_norm": 1.9292621612548828,
      "learning_rate": 1.9557128574568126e-06,
      "loss": 0.4549,
      "step": 281090
    },
    {
      "epoch": 115.06,
      "grad_norm": 2.0192208290100098,
      "learning_rate": 1.955556417995827e-06,
      "loss": 0.4652,
      "step": 281100
    },
    {
      "epoch": 115.07,
      "grad_norm": 1.7011840343475342,
      "learning_rate": 1.955399980313097e-06,
      "loss": 0.4526,
      "step": 281110
    },
    {
      "epoch": 115.07,
      "grad_norm": 1.6963701248168945,
      "learning_rate": 1.9552435444093383e-06,
      "loss": 0.457,
      "step": 281120
    },
    {
      "epoch": 115.08,
      "grad_norm": 2.0497498512268066,
      "learning_rate": 1.955087110285267e-06,
      "loss": 0.4832,
      "step": 281130
    },
    {
      "epoch": 115.08,
      "grad_norm": 1.9937855005264282,
      "learning_rate": 1.9549306779416003e-06,
      "loss": 0.4821,
      "step": 281140
    },
    {
      "epoch": 115.08,
      "grad_norm": 2.6095385551452637,
      "learning_rate": 1.9547742473790546e-06,
      "loss": 0.4525,
      "step": 281150
    },
    {
      "epoch": 115.09,
      "grad_norm": 2.451373815536499,
      "learning_rate": 1.954617818598347e-06,
      "loss": 0.4852,
      "step": 281160
    },
    {
      "epoch": 115.09,
      "grad_norm": 2.0941450595855713,
      "learning_rate": 1.954461391600193e-06,
      "loss": 0.4546,
      "step": 281170
    },
    {
      "epoch": 115.1,
      "grad_norm": 2.144834280014038,
      "learning_rate": 1.9543049663853107e-06,
      "loss": 0.4826,
      "step": 281180
    },
    {
      "epoch": 115.1,
      "grad_norm": 1.7248163223266602,
      "learning_rate": 1.9541485429544153e-06,
      "loss": 0.4589,
      "step": 281190
    },
    {
      "epoch": 115.1,
      "grad_norm": 1.9354802370071411,
      "learning_rate": 1.9539921213082242e-06,
      "loss": 0.4547,
      "step": 281200
    },
    {
      "epoch": 115.11,
      "grad_norm": 2.1649434566497803,
      "learning_rate": 1.953835701447453e-06,
      "loss": 0.4566,
      "step": 281210
    },
    {
      "epoch": 115.11,
      "grad_norm": 1.906307578086853,
      "learning_rate": 1.953679283372819e-06,
      "loss": 0.4699,
      "step": 281220
    },
    {
      "epoch": 115.12,
      "grad_norm": 2.696871519088745,
      "learning_rate": 1.9535228670850375e-06,
      "loss": 0.455,
      "step": 281230
    },
    {
      "epoch": 115.12,
      "grad_norm": 1.813210368156433,
      "learning_rate": 1.9533664525848268e-06,
      "loss": 0.4564,
      "step": 281240
    },
    {
      "epoch": 115.12,
      "grad_norm": 2.079271078109741,
      "learning_rate": 1.9532100398729038e-06,
      "loss": 0.4738,
      "step": 281250
    },
    {
      "epoch": 115.13,
      "grad_norm": 1.976494550704956,
      "learning_rate": 1.953053628949983e-06,
      "loss": 0.4891,
      "step": 281260
    },
    {
      "epoch": 115.13,
      "grad_norm": 2.1840505599975586,
      "learning_rate": 1.9528972198167817e-06,
      "loss": 0.4866,
      "step": 281270
    },
    {
      "epoch": 115.14,
      "grad_norm": 1.8355083465576172,
      "learning_rate": 1.9527408124740168e-06,
      "loss": 0.459,
      "step": 281280
    },
    {
      "epoch": 115.14,
      "grad_norm": 1.8966013193130493,
      "learning_rate": 1.9525844069224038e-06,
      "loss": 0.4659,
      "step": 281290
    },
    {
      "epoch": 115.15,
      "grad_norm": 2.0688424110412598,
      "learning_rate": 1.9524280031626606e-06,
      "loss": 0.5001,
      "step": 281300
    },
    {
      "epoch": 115.15,
      "grad_norm": 1.5197490453720093,
      "learning_rate": 1.952271601195502e-06,
      "loss": 0.4616,
      "step": 281310
    },
    {
      "epoch": 115.15,
      "grad_norm": 2.0261662006378174,
      "learning_rate": 1.9521152010216462e-06,
      "loss": 0.463,
      "step": 281320
    },
    {
      "epoch": 115.16,
      "grad_norm": 1.9353491067886353,
      "learning_rate": 1.9519588026418083e-06,
      "loss": 0.4778,
      "step": 281330
    },
    {
      "epoch": 115.16,
      "grad_norm": 1.9145169258117676,
      "learning_rate": 1.9518024060567056e-06,
      "loss": 0.4662,
      "step": 281340
    },
    {
      "epoch": 115.17,
      "grad_norm": 2.089200019836426,
      "learning_rate": 1.9516460112670536e-06,
      "loss": 0.4689,
      "step": 281350
    },
    {
      "epoch": 115.17,
      "grad_norm": 1.942764163017273,
      "learning_rate": 1.9514896182735697e-06,
      "loss": 0.4727,
      "step": 281360
    },
    {
      "epoch": 115.17,
      "grad_norm": 1.813370704650879,
      "learning_rate": 1.9513332270769692e-06,
      "loss": 0.4507,
      "step": 281370
    },
    {
      "epoch": 115.18,
      "grad_norm": 1.859184980392456,
      "learning_rate": 1.9511768376779696e-06,
      "loss": 0.4601,
      "step": 281380
    },
    {
      "epoch": 115.18,
      "grad_norm": 1.7848553657531738,
      "learning_rate": 1.951020450077287e-06,
      "loss": 0.4718,
      "step": 281390
    },
    {
      "epoch": 115.19,
      "grad_norm": 2.0239620208740234,
      "learning_rate": 1.9508640642756376e-06,
      "loss": 0.4697,
      "step": 281400
    },
    {
      "epoch": 115.19,
      "grad_norm": 1.8370270729064941,
      "learning_rate": 1.9507076802737375e-06,
      "loss": 0.469,
      "step": 281410
    },
    {
      "epoch": 115.19,
      "grad_norm": 2.155496120452881,
      "learning_rate": 1.9505512980723033e-06,
      "loss": 0.4432,
      "step": 281420
    },
    {
      "epoch": 115.2,
      "grad_norm": 2.6097567081451416,
      "learning_rate": 1.9503949176720513e-06,
      "loss": 0.466,
      "step": 281430
    },
    {
      "epoch": 115.2,
      "grad_norm": 2.01039457321167,
      "learning_rate": 1.9502385390736988e-06,
      "loss": 0.4636,
      "step": 281440
    },
    {
      "epoch": 115.21,
      "grad_norm": 1.9364734888076782,
      "learning_rate": 1.9500821622779607e-06,
      "loss": 0.4718,
      "step": 281450
    },
    {
      "epoch": 115.21,
      "grad_norm": 1.7433784008026123,
      "learning_rate": 1.9499257872855545e-06,
      "loss": 0.457,
      "step": 281460
    },
    {
      "epoch": 115.21,
      "grad_norm": 2.194530963897705,
      "learning_rate": 1.9497694140971955e-06,
      "loss": 0.4627,
      "step": 281470
    },
    {
      "epoch": 115.22,
      "grad_norm": 1.9943710565567017,
      "learning_rate": 1.9496130427136016e-06,
      "loss": 0.4605,
      "step": 281480
    },
    {
      "epoch": 115.22,
      "grad_norm": 1.824617624282837,
      "learning_rate": 1.949456673135487e-06,
      "loss": 0.4563,
      "step": 281490
    },
    {
      "epoch": 115.23,
      "grad_norm": 1.6414759159088135,
      "learning_rate": 1.9493003053635697e-06,
      "loss": 0.4763,
      "step": 281500
    },
    {
      "epoch": 115.23,
      "grad_norm": 1.8419486284255981,
      "learning_rate": 1.949143939398565e-06,
      "loss": 0.474,
      "step": 281510
    },
    {
      "epoch": 115.24,
      "grad_norm": 1.9967622756958008,
      "learning_rate": 1.9489875752411895e-06,
      "loss": 0.473,
      "step": 281520
    },
    {
      "epoch": 115.24,
      "grad_norm": 2.575922727584839,
      "learning_rate": 1.94883121289216e-06,
      "loss": 0.4963,
      "step": 281530
    },
    {
      "epoch": 115.24,
      "grad_norm": 1.6773041486740112,
      "learning_rate": 1.9486748523521916e-06,
      "loss": 0.4678,
      "step": 281540
    },
    {
      "epoch": 115.25,
      "grad_norm": 1.9386063814163208,
      "learning_rate": 1.948518493622002e-06,
      "loss": 0.4639,
      "step": 281550
    },
    {
      "epoch": 115.25,
      "grad_norm": 2.2748055458068848,
      "learning_rate": 1.9483621367023065e-06,
      "loss": 0.4354,
      "step": 281560
    },
    {
      "epoch": 115.26,
      "grad_norm": 1.6489003896713257,
      "learning_rate": 1.9482057815938213e-06,
      "loss": 0.4768,
      "step": 281570
    },
    {
      "epoch": 115.26,
      "grad_norm": 1.917704463005066,
      "learning_rate": 1.9480494282972632e-06,
      "loss": 0.4508,
      "step": 281580
    },
    {
      "epoch": 115.26,
      "grad_norm": 1.7414960861206055,
      "learning_rate": 1.947893076813348e-06,
      "loss": 0.4604,
      "step": 281590
    },
    {
      "epoch": 115.27,
      "grad_norm": 2.1556577682495117,
      "learning_rate": 1.947736727142792e-06,
      "loss": 0.5037,
      "step": 281600
    },
    {
      "epoch": 115.27,
      "grad_norm": 1.857589840888977,
      "learning_rate": 1.947580379286311e-06,
      "loss": 0.4545,
      "step": 281610
    },
    {
      "epoch": 115.28,
      "grad_norm": 3.905963182449341,
      "learning_rate": 1.9474240332446223e-06,
      "loss": 0.4666,
      "step": 281620
    },
    {
      "epoch": 115.28,
      "grad_norm": 2.051882266998291,
      "learning_rate": 1.947267689018441e-06,
      "loss": 0.4725,
      "step": 281630
    },
    {
      "epoch": 115.28,
      "grad_norm": 1.8269318342208862,
      "learning_rate": 1.947111346608484e-06,
      "loss": 0.4674,
      "step": 281640
    },
    {
      "epoch": 115.29,
      "grad_norm": 1.8668428659439087,
      "learning_rate": 1.9469550060154675e-06,
      "loss": 0.4776,
      "step": 281650
    },
    {
      "epoch": 115.29,
      "grad_norm": 2.120598554611206,
      "learning_rate": 1.9467986672401072e-06,
      "loss": 0.4651,
      "step": 281660
    },
    {
      "epoch": 115.3,
      "grad_norm": 2.0607545375823975,
      "learning_rate": 1.9466423302831197e-06,
      "loss": 0.471,
      "step": 281670
    },
    {
      "epoch": 115.3,
      "grad_norm": 1.9831006526947021,
      "learning_rate": 1.946485995145221e-06,
      "loss": 0.4593,
      "step": 281680
    },
    {
      "epoch": 115.3,
      "grad_norm": 2.1796998977661133,
      "learning_rate": 1.9463296618271266e-06,
      "loss": 0.4626,
      "step": 281690
    },
    {
      "epoch": 115.31,
      "grad_norm": 2.0145227909088135,
      "learning_rate": 1.946173330329554e-06,
      "loss": 0.4687,
      "step": 281700
    },
    {
      "epoch": 115.31,
      "grad_norm": 1.9648202657699585,
      "learning_rate": 1.946017000653218e-06,
      "loss": 0.4915,
      "step": 281710
    },
    {
      "epoch": 115.32,
      "grad_norm": 1.6481562852859497,
      "learning_rate": 1.9458606727988354e-06,
      "loss": 0.4623,
      "step": 281720
    },
    {
      "epoch": 115.32,
      "grad_norm": 2.0398738384246826,
      "learning_rate": 1.9457043467671223e-06,
      "loss": 0.4711,
      "step": 281730
    },
    {
      "epoch": 115.33,
      "grad_norm": 1.8184523582458496,
      "learning_rate": 1.9455480225587946e-06,
      "loss": 0.4552,
      "step": 281740
    },
    {
      "epoch": 115.33,
      "grad_norm": 2.4846384525299072,
      "learning_rate": 1.9453917001745684e-06,
      "loss": 0.4627,
      "step": 281750
    },
    {
      "epoch": 115.33,
      "grad_norm": 2.256143569946289,
      "learning_rate": 1.94523537961516e-06,
      "loss": 0.4828,
      "step": 281760
    },
    {
      "epoch": 115.34,
      "grad_norm": 2.949507713317871,
      "learning_rate": 1.9450790608812853e-06,
      "loss": 0.469,
      "step": 281770
    },
    {
      "epoch": 115.34,
      "grad_norm": 2.351954698562622,
      "learning_rate": 1.9449227439736604e-06,
      "loss": 0.4821,
      "step": 281780
    },
    {
      "epoch": 115.35,
      "grad_norm": 1.6639927625656128,
      "learning_rate": 1.9447664288930008e-06,
      "loss": 0.4785,
      "step": 281790
    },
    {
      "epoch": 115.35,
      "grad_norm": 1.9254955053329468,
      "learning_rate": 1.944610115640024e-06,
      "loss": 0.4719,
      "step": 281800
    },
    {
      "epoch": 115.35,
      "grad_norm": 1.6124634742736816,
      "learning_rate": 1.944453804215445e-06,
      "loss": 0.4583,
      "step": 281810
    },
    {
      "epoch": 115.36,
      "grad_norm": 2.363720417022705,
      "learning_rate": 1.94429749461998e-06,
      "loss": 0.4707,
      "step": 281820
    },
    {
      "epoch": 115.36,
      "grad_norm": 1.9673702716827393,
      "learning_rate": 1.944141186854345e-06,
      "loss": 0.4632,
      "step": 281830
    },
    {
      "epoch": 115.37,
      "grad_norm": 2.0339572429656982,
      "learning_rate": 1.9439848809192566e-06,
      "loss": 0.4789,
      "step": 281840
    },
    {
      "epoch": 115.37,
      "grad_norm": 1.9597371816635132,
      "learning_rate": 1.94382857681543e-06,
      "loss": 0.4406,
      "step": 281850
    },
    {
      "epoch": 115.37,
      "grad_norm": 1.6055972576141357,
      "learning_rate": 1.9436722745435814e-06,
      "loss": 0.4709,
      "step": 281860
    },
    {
      "epoch": 115.38,
      "grad_norm": 2.156984329223633,
      "learning_rate": 1.9435159741044267e-06,
      "loss": 0.4753,
      "step": 281870
    },
    {
      "epoch": 115.38,
      "grad_norm": 2.009470224380493,
      "learning_rate": 1.943359675498682e-06,
      "loss": 0.4448,
      "step": 281880
    },
    {
      "epoch": 115.39,
      "grad_norm": 1.8584271669387817,
      "learning_rate": 1.943203378727064e-06,
      "loss": 0.471,
      "step": 281890
    },
    {
      "epoch": 115.39,
      "grad_norm": 1.6390949487686157,
      "learning_rate": 1.9430470837902873e-06,
      "loss": 0.4734,
      "step": 281900
    },
    {
      "epoch": 115.4,
      "grad_norm": 1.9292629957199097,
      "learning_rate": 1.942890790689069e-06,
      "loss": 0.4435,
      "step": 281910
    },
    {
      "epoch": 115.4,
      "grad_norm": 1.769851803779602,
      "learning_rate": 1.942734499424125e-06,
      "loss": 0.4524,
      "step": 281920
    },
    {
      "epoch": 115.4,
      "grad_norm": 1.9060518741607666,
      "learning_rate": 1.9425782099961708e-06,
      "loss": 0.4537,
      "step": 281930
    },
    {
      "epoch": 115.41,
      "grad_norm": 1.6054377555847168,
      "learning_rate": 1.9424219224059223e-06,
      "loss": 0.478,
      "step": 281940
    },
    {
      "epoch": 115.41,
      "grad_norm": 1.7045376300811768,
      "learning_rate": 1.9422656366540952e-06,
      "loss": 0.4649,
      "step": 281950
    },
    {
      "epoch": 115.42,
      "grad_norm": 2.196545362472534,
      "learning_rate": 1.942109352741406e-06,
      "loss": 0.4721,
      "step": 281960
    },
    {
      "epoch": 115.42,
      "grad_norm": 2.665027618408203,
      "learning_rate": 1.9419530706685703e-06,
      "loss": 0.4767,
      "step": 281970
    },
    {
      "epoch": 115.42,
      "grad_norm": 2.0693838596343994,
      "learning_rate": 1.941796790436304e-06,
      "loss": 0.4827,
      "step": 281980
    },
    {
      "epoch": 115.43,
      "grad_norm": 1.8064566850662231,
      "learning_rate": 1.9416405120453232e-06,
      "loss": 0.4723,
      "step": 281990
    },
    {
      "epoch": 115.43,
      "grad_norm": 2.054664373397827,
      "learning_rate": 1.941484235496344e-06,
      "loss": 0.4488,
      "step": 282000
    },
    {
      "epoch": 115.44,
      "grad_norm": 1.7305982112884521,
      "learning_rate": 1.9413279607900816e-06,
      "loss": 0.4543,
      "step": 282010
    },
    {
      "epoch": 115.44,
      "grad_norm": 1.8420075178146362,
      "learning_rate": 1.9411716879272525e-06,
      "loss": 0.4851,
      "step": 282020
    },
    {
      "epoch": 115.44,
      "grad_norm": 2.472883701324463,
      "learning_rate": 1.941015416908572e-06,
      "loss": 0.456,
      "step": 282030
    },
    {
      "epoch": 115.45,
      "grad_norm": 1.6551507711410522,
      "learning_rate": 1.9408591477347567e-06,
      "loss": 0.4741,
      "step": 282040
    },
    {
      "epoch": 115.45,
      "grad_norm": 1.7583833932876587,
      "learning_rate": 1.940702880406522e-06,
      "loss": 0.4892,
      "step": 282050
    },
    {
      "epoch": 115.46,
      "grad_norm": 2.024765968322754,
      "learning_rate": 1.9405466149245837e-06,
      "loss": 0.4764,
      "step": 282060
    },
    {
      "epoch": 115.46,
      "grad_norm": 2.1076996326446533,
      "learning_rate": 1.9403903512896577e-06,
      "loss": 0.4627,
      "step": 282070
    },
    {
      "epoch": 115.46,
      "grad_norm": 1.834853172302246,
      "learning_rate": 1.9402340895024595e-06,
      "loss": 0.487,
      "step": 282080
    },
    {
      "epoch": 115.47,
      "grad_norm": 2.1443259716033936,
      "learning_rate": 1.9400778295637053e-06,
      "loss": 0.4394,
      "step": 282090
    },
    {
      "epoch": 115.47,
      "grad_norm": 2.446218729019165,
      "learning_rate": 1.9399215714741104e-06,
      "loss": 0.4889,
      "step": 282100
    },
    {
      "epoch": 115.48,
      "grad_norm": 2.2006723880767822,
      "learning_rate": 1.9397653152343915e-06,
      "loss": 0.4905,
      "step": 282110
    },
    {
      "epoch": 115.48,
      "grad_norm": 1.7765108346939087,
      "learning_rate": 1.9396090608452638e-06,
      "loss": 0.4595,
      "step": 282120
    },
    {
      "epoch": 115.49,
      "grad_norm": 2.1890690326690674,
      "learning_rate": 1.939452808307443e-06,
      "loss": 0.4518,
      "step": 282130
    },
    {
      "epoch": 115.49,
      "grad_norm": 2.3183579444885254,
      "learning_rate": 1.939296557621645e-06,
      "loss": 0.4484,
      "step": 282140
    },
    {
      "epoch": 115.49,
      "grad_norm": 1.9897257089614868,
      "learning_rate": 1.9391403087885863e-06,
      "loss": 0.4698,
      "step": 282150
    },
    {
      "epoch": 115.5,
      "grad_norm": 2.0829615592956543,
      "learning_rate": 1.9389840618089813e-06,
      "loss": 0.4598,
      "step": 282160
    },
    {
      "epoch": 115.5,
      "grad_norm": 2.5015172958374023,
      "learning_rate": 1.938827816683545e-06,
      "loss": 0.4425,
      "step": 282170
    },
    {
      "epoch": 115.51,
      "grad_norm": 2.1758925914764404,
      "learning_rate": 1.938671573412996e-06,
      "loss": 0.4741,
      "step": 282180
    },
    {
      "epoch": 115.51,
      "grad_norm": 1.5781196355819702,
      "learning_rate": 1.938515331998049e-06,
      "loss": 0.453,
      "step": 282190
    },
    {
      "epoch": 115.51,
      "grad_norm": 1.9869661331176758,
      "learning_rate": 1.938359092439419e-06,
      "loss": 0.4517,
      "step": 282200
    },
    {
      "epoch": 115.52,
      "grad_norm": 1.7642772197723389,
      "learning_rate": 1.9382028547378215e-06,
      "loss": 0.4628,
      "step": 282210
    },
    {
      "epoch": 115.52,
      "grad_norm": 1.9845927953720093,
      "learning_rate": 1.938046618893973e-06,
      "loss": 0.4708,
      "step": 282220
    },
    {
      "epoch": 115.53,
      "grad_norm": 2.271162509918213,
      "learning_rate": 1.9378903849085885e-06,
      "loss": 0.4578,
      "step": 282230
    },
    {
      "epoch": 115.53,
      "grad_norm": 1.7244048118591309,
      "learning_rate": 1.9377341527823843e-06,
      "loss": 0.4409,
      "step": 282240
    },
    {
      "epoch": 115.53,
      "grad_norm": 1.8781001567840576,
      "learning_rate": 1.9375779225160756e-06,
      "loss": 0.4542,
      "step": 282250
    },
    {
      "epoch": 115.54,
      "grad_norm": 1.8942818641662598,
      "learning_rate": 1.937421694110378e-06,
      "loss": 0.4705,
      "step": 282260
    },
    {
      "epoch": 115.54,
      "grad_norm": 2.1384241580963135,
      "learning_rate": 1.937265467566008e-06,
      "loss": 0.4478,
      "step": 282270
    },
    {
      "epoch": 115.55,
      "grad_norm": 1.449794054031372,
      "learning_rate": 1.9371092428836814e-06,
      "loss": 0.4587,
      "step": 282280
    },
    {
      "epoch": 115.55,
      "grad_norm": 2.662196159362793,
      "learning_rate": 1.9369530200641124e-06,
      "loss": 0.469,
      "step": 282290
    },
    {
      "epoch": 115.55,
      "grad_norm": 1.7537742853164673,
      "learning_rate": 1.936796799108017e-06,
      "loss": 0.4458,
      "step": 282300
    },
    {
      "epoch": 115.56,
      "grad_norm": 1.963060975074768,
      "learning_rate": 1.9366405800161115e-06,
      "loss": 0.468,
      "step": 282310
    },
    {
      "epoch": 115.56,
      "grad_norm": 1.9577341079711914,
      "learning_rate": 1.936484362789111e-06,
      "loss": 0.4684,
      "step": 282320
    },
    {
      "epoch": 115.57,
      "grad_norm": 1.7661150693893433,
      "learning_rate": 1.9363281474277314e-06,
      "loss": 0.4656,
      "step": 282330
    },
    {
      "epoch": 115.57,
      "grad_norm": 2.1632285118103027,
      "learning_rate": 1.9361719339326882e-06,
      "loss": 0.4611,
      "step": 282340
    },
    {
      "epoch": 115.58,
      "grad_norm": 1.6291292905807495,
      "learning_rate": 1.9360157223046975e-06,
      "loss": 0.4558,
      "step": 282350
    },
    {
      "epoch": 115.58,
      "grad_norm": 1.760357141494751,
      "learning_rate": 1.9358595125444733e-06,
      "loss": 0.4604,
      "step": 282360
    },
    {
      "epoch": 115.58,
      "grad_norm": 2.1045355796813965,
      "learning_rate": 1.935703304652733e-06,
      "loss": 0.464,
      "step": 282370
    },
    {
      "epoch": 115.59,
      "grad_norm": 1.6458861827850342,
      "learning_rate": 1.935547098630191e-06,
      "loss": 0.4511,
      "step": 282380
    },
    {
      "epoch": 115.59,
      "grad_norm": 1.8441784381866455,
      "learning_rate": 1.935390894477564e-06,
      "loss": 0.4565,
      "step": 282390
    },
    {
      "epoch": 115.6,
      "grad_norm": 1.6825571060180664,
      "learning_rate": 1.935234692195566e-06,
      "loss": 0.4788,
      "step": 282400
    },
    {
      "epoch": 115.6,
      "grad_norm": 2.1404342651367188,
      "learning_rate": 1.935078491784914e-06,
      "loss": 0.4891,
      "step": 282410
    },
    {
      "epoch": 115.6,
      "grad_norm": 1.912022352218628,
      "learning_rate": 1.9349222932463233e-06,
      "loss": 0.4655,
      "step": 282420
    },
    {
      "epoch": 115.61,
      "grad_norm": 2.2992045879364014,
      "learning_rate": 1.934766096580508e-06,
      "loss": 0.476,
      "step": 282430
    },
    {
      "epoch": 115.61,
      "grad_norm": 1.686625599861145,
      "learning_rate": 1.9346099017881852e-06,
      "loss": 0.4885,
      "step": 282440
    },
    {
      "epoch": 115.62,
      "grad_norm": 2.031447172164917,
      "learning_rate": 1.9344537088700698e-06,
      "loss": 0.4753,
      "step": 282450
    },
    {
      "epoch": 115.62,
      "grad_norm": 1.7699172496795654,
      "learning_rate": 1.934297517826877e-06,
      "loss": 0.4678,
      "step": 282460
    },
    {
      "epoch": 115.62,
      "grad_norm": 2.1238067150115967,
      "learning_rate": 1.9341413286593226e-06,
      "loss": 0.4741,
      "step": 282470
    },
    {
      "epoch": 115.63,
      "grad_norm": 1.796559453010559,
      "learning_rate": 1.9339851413681224e-06,
      "loss": 0.4668,
      "step": 282480
    },
    {
      "epoch": 115.63,
      "grad_norm": 2.2603368759155273,
      "learning_rate": 1.9338289559539913e-06,
      "loss": 0.476,
      "step": 282490
    },
    {
      "epoch": 115.64,
      "grad_norm": 1.8106048107147217,
      "learning_rate": 1.9336727724176454e-06,
      "loss": 0.4722,
      "step": 282500
    },
    {
      "epoch": 115.64,
      "grad_norm": 1.9345329999923706,
      "learning_rate": 1.9335165907597996e-06,
      "loss": 0.4769,
      "step": 282510
    },
    {
      "epoch": 115.64,
      "grad_norm": 1.996145248413086,
      "learning_rate": 1.9333604109811692e-06,
      "loss": 0.4798,
      "step": 282520
    },
    {
      "epoch": 115.65,
      "grad_norm": 1.646915316581726,
      "learning_rate": 1.9332042330824705e-06,
      "loss": 0.4654,
      "step": 282530
    },
    {
      "epoch": 115.65,
      "grad_norm": 1.79738450050354,
      "learning_rate": 1.933048057064418e-06,
      "loss": 0.4706,
      "step": 282540
    },
    {
      "epoch": 115.66,
      "grad_norm": 2.129307270050049,
      "learning_rate": 1.932891882927727e-06,
      "loss": 0.4514,
      "step": 282550
    },
    {
      "epoch": 115.66,
      "grad_norm": 1.8612396717071533,
      "learning_rate": 1.9327357106731137e-06,
      "loss": 0.463,
      "step": 282560
    },
    {
      "epoch": 115.67,
      "grad_norm": 1.7781175374984741,
      "learning_rate": 1.9325795403012933e-06,
      "loss": 0.4644,
      "step": 282570
    },
    {
      "epoch": 115.67,
      "grad_norm": 1.6652617454528809,
      "learning_rate": 1.932423371812981e-06,
      "loss": 0.4386,
      "step": 282580
    },
    {
      "epoch": 115.67,
      "grad_norm": 1.7339955568313599,
      "learning_rate": 1.9322672052088928e-06,
      "loss": 0.4625,
      "step": 282590
    },
    {
      "epoch": 115.68,
      "grad_norm": 1.9354751110076904,
      "learning_rate": 1.9321110404897434e-06,
      "loss": 0.4711,
      "step": 282600
    },
    {
      "epoch": 115.68,
      "grad_norm": 2.0187113285064697,
      "learning_rate": 1.931954877656248e-06,
      "loss": 0.471,
      "step": 282610
    },
    {
      "epoch": 115.69,
      "grad_norm": 1.7206802368164062,
      "learning_rate": 1.9317987167091227e-06,
      "loss": 0.4604,
      "step": 282620
    },
    {
      "epoch": 115.69,
      "grad_norm": 1.9208433628082275,
      "learning_rate": 1.931642557649082e-06,
      "loss": 0.4591,
      "step": 282630
    },
    {
      "epoch": 115.69,
      "grad_norm": 1.837584376335144,
      "learning_rate": 1.9314864004768426e-06,
      "loss": 0.4697,
      "step": 282640
    },
    {
      "epoch": 115.7,
      "grad_norm": 2.6059751510620117,
      "learning_rate": 1.931330245193118e-06,
      "loss": 0.479,
      "step": 282650
    },
    {
      "epoch": 115.7,
      "grad_norm": 2.4332621097564697,
      "learning_rate": 1.9311740917986246e-06,
      "loss": 0.4691,
      "step": 282660
    },
    {
      "epoch": 115.71,
      "grad_norm": 1.715997576713562,
      "learning_rate": 1.9310179402940776e-06,
      "loss": 0.4611,
      "step": 282670
    },
    {
      "epoch": 115.71,
      "grad_norm": 1.759912133216858,
      "learning_rate": 1.930861790680192e-06,
      "loss": 0.465,
      "step": 282680
    },
    {
      "epoch": 115.71,
      "grad_norm": 1.6812107563018799,
      "learning_rate": 1.9307056429576837e-06,
      "loss": 0.4548,
      "step": 282690
    },
    {
      "epoch": 115.72,
      "grad_norm": 1.8239481449127197,
      "learning_rate": 1.9305494971272674e-06,
      "loss": 0.4607,
      "step": 282700
    },
    {
      "epoch": 115.72,
      "grad_norm": 1.986669898033142,
      "learning_rate": 1.930393353189659e-06,
      "loss": 0.4686,
      "step": 282710
    },
    {
      "epoch": 115.73,
      "grad_norm": 2.3171942234039307,
      "learning_rate": 1.9302372111455725e-06,
      "loss": 0.4442,
      "step": 282720
    },
    {
      "epoch": 115.73,
      "grad_norm": 1.837497591972351,
      "learning_rate": 1.9300810709957255e-06,
      "loss": 0.4503,
      "step": 282730
    },
    {
      "epoch": 115.73,
      "grad_norm": 2.365983009338379,
      "learning_rate": 1.9299249327408296e-06,
      "loss": 0.4536,
      "step": 282740
    },
    {
      "epoch": 115.74,
      "grad_norm": 2.1672441959381104,
      "learning_rate": 1.9297687963816036e-06,
      "loss": 0.4885,
      "step": 282750
    },
    {
      "epoch": 115.74,
      "grad_norm": 2.05957293510437,
      "learning_rate": 1.929612661918761e-06,
      "loss": 0.4702,
      "step": 282760
    },
    {
      "epoch": 115.75,
      "grad_norm": 2.031050443649292,
      "learning_rate": 1.929456529353018e-06,
      "loss": 0.4421,
      "step": 282770
    },
    {
      "epoch": 115.75,
      "grad_norm": 1.7478948831558228,
      "learning_rate": 1.9293003986850894e-06,
      "loss": 0.4308,
      "step": 282780
    },
    {
      "epoch": 115.76,
      "grad_norm": 1.79916512966156,
      "learning_rate": 1.92914426991569e-06,
      "loss": 0.4343,
      "step": 282790
    },
    {
      "epoch": 115.76,
      "grad_norm": 1.8667535781860352,
      "learning_rate": 1.9289881430455347e-06,
      "loss": 0.4522,
      "step": 282800
    },
    {
      "epoch": 115.76,
      "grad_norm": 1.591637372970581,
      "learning_rate": 1.9288320180753397e-06,
      "loss": 0.4505,
      "step": 282810
    },
    {
      "epoch": 115.77,
      "grad_norm": 2.277233362197876,
      "learning_rate": 1.9286758950058197e-06,
      "loss": 0.4624,
      "step": 282820
    },
    {
      "epoch": 115.77,
      "grad_norm": 1.9549344778060913,
      "learning_rate": 1.9285197738376896e-06,
      "loss": 0.4613,
      "step": 282830
    },
    {
      "epoch": 115.78,
      "grad_norm": 2.1362295150756836,
      "learning_rate": 1.928363654571665e-06,
      "loss": 0.4784,
      "step": 282840
    },
    {
      "epoch": 115.78,
      "grad_norm": 1.538912057876587,
      "learning_rate": 1.9282075372084613e-06,
      "loss": 0.4701,
      "step": 282850
    },
    {
      "epoch": 115.78,
      "grad_norm": 2.0317513942718506,
      "learning_rate": 1.9280514217487933e-06,
      "loss": 0.4491,
      "step": 282860
    },
    {
      "epoch": 115.79,
      "grad_norm": 2.245591878890991,
      "learning_rate": 1.9278953081933756e-06,
      "loss": 0.4728,
      "step": 282870
    },
    {
      "epoch": 115.79,
      "grad_norm": 2.297034740447998,
      "learning_rate": 1.927739196542924e-06,
      "loss": 0.4629,
      "step": 282880
    },
    {
      "epoch": 115.8,
      "grad_norm": 2.129889965057373,
      "learning_rate": 1.9275830867981535e-06,
      "loss": 0.4526,
      "step": 282890
    },
    {
      "epoch": 115.8,
      "grad_norm": 1.6915004253387451,
      "learning_rate": 1.927426978959779e-06,
      "loss": 0.49,
      "step": 282900
    },
    {
      "epoch": 115.8,
      "grad_norm": 2.2252814769744873,
      "learning_rate": 1.9272708730285156e-06,
      "loss": 0.4558,
      "step": 282910
    },
    {
      "epoch": 115.81,
      "grad_norm": 1.6754074096679688,
      "learning_rate": 1.927114769005079e-06,
      "loss": 0.4533,
      "step": 282920
    },
    {
      "epoch": 115.81,
      "grad_norm": 1.8061589002609253,
      "learning_rate": 1.9269586668901835e-06,
      "loss": 0.4781,
      "step": 282930
    },
    {
      "epoch": 115.82,
      "grad_norm": 1.9493815898895264,
      "learning_rate": 1.9268025666845446e-06,
      "loss": 0.4608,
      "step": 282940
    },
    {
      "epoch": 115.82,
      "grad_norm": 1.9820044040679932,
      "learning_rate": 1.926646468388878e-06,
      "loss": 0.4709,
      "step": 282950
    },
    {
      "epoch": 115.82,
      "grad_norm": 2.126466751098633,
      "learning_rate": 1.926490372003897e-06,
      "loss": 0.4668,
      "step": 282960
    },
    {
      "epoch": 115.83,
      "grad_norm": 1.9772974252700806,
      "learning_rate": 1.9263342775303187e-06,
      "loss": 0.4511,
      "step": 282970
    },
    {
      "epoch": 115.83,
      "grad_norm": 1.8193259239196777,
      "learning_rate": 1.9261781849688565e-06,
      "loss": 0.4562,
      "step": 282980
    },
    {
      "epoch": 115.84,
      "grad_norm": 1.888975977897644,
      "learning_rate": 1.9260220943202265e-06,
      "loss": 0.4746,
      "step": 282990
    },
    {
      "epoch": 115.84,
      "grad_norm": 1.8670967817306519,
      "learning_rate": 1.925866005585144e-06,
      "loss": 0.4695,
      "step": 283000
    },
    {
      "epoch": 115.85,
      "grad_norm": 2.1723859310150146,
      "learning_rate": 1.9257099187643225e-06,
      "loss": 0.4823,
      "step": 283010
    },
    {
      "epoch": 115.85,
      "grad_norm": 2.187528610229492,
      "learning_rate": 1.925553833858478e-06,
      "loss": 0.4694,
      "step": 283020
    },
    {
      "epoch": 115.85,
      "grad_norm": 2.114126205444336,
      "learning_rate": 1.925397750868325e-06,
      "loss": 0.4522,
      "step": 283030
    },
    {
      "epoch": 115.86,
      "grad_norm": 1.8048008680343628,
      "learning_rate": 1.9252416697945797e-06,
      "loss": 0.4615,
      "step": 283040
    },
    {
      "epoch": 115.86,
      "grad_norm": 1.5832023620605469,
      "learning_rate": 1.9250855906379556e-06,
      "loss": 0.4673,
      "step": 283050
    },
    {
      "epoch": 115.87,
      "grad_norm": 2.2735605239868164,
      "learning_rate": 1.9249295133991687e-06,
      "loss": 0.4533,
      "step": 283060
    },
    {
      "epoch": 115.87,
      "grad_norm": 1.6734168529510498,
      "learning_rate": 1.924773438078933e-06,
      "loss": 0.4797,
      "step": 283070
    },
    {
      "epoch": 115.87,
      "grad_norm": 2.19891095161438,
      "learning_rate": 1.924617364677965e-06,
      "loss": 0.4548,
      "step": 283080
    },
    {
      "epoch": 115.88,
      "grad_norm": 1.9819867610931396,
      "learning_rate": 1.9244612931969784e-06,
      "loss": 0.4538,
      "step": 283090
    },
    {
      "epoch": 115.88,
      "grad_norm": 1.9332172870635986,
      "learning_rate": 1.924305223636688e-06,
      "loss": 0.4592,
      "step": 283100
    },
    {
      "epoch": 115.89,
      "grad_norm": 1.8547190427780151,
      "learning_rate": 1.9241491559978082e-06,
      "loss": 0.4693,
      "step": 283110
    },
    {
      "epoch": 115.89,
      "grad_norm": 1.8707940578460693,
      "learning_rate": 1.923993090281056e-06,
      "loss": 0.4864,
      "step": 283120
    },
    {
      "epoch": 115.89,
      "grad_norm": 1.9901454448699951,
      "learning_rate": 1.923837026487145e-06,
      "loss": 0.4539,
      "step": 283130
    },
    {
      "epoch": 115.9,
      "grad_norm": 1.9602441787719727,
      "learning_rate": 1.9236809646167915e-06,
      "loss": 0.4517,
      "step": 283140
    },
    {
      "epoch": 115.9,
      "grad_norm": 1.7490700483322144,
      "learning_rate": 1.923524904670708e-06,
      "loss": 0.4739,
      "step": 283150
    },
    {
      "epoch": 115.91,
      "grad_norm": 2.0636258125305176,
      "learning_rate": 1.9233688466496107e-06,
      "loss": 0.4812,
      "step": 283160
    },
    {
      "epoch": 115.91,
      "grad_norm": 1.6633081436157227,
      "learning_rate": 1.9232127905542146e-06,
      "loss": 0.4613,
      "step": 283170
    },
    {
      "epoch": 115.91,
      "grad_norm": 2.0588693618774414,
      "learning_rate": 1.923056736385234e-06,
      "loss": 0.4675,
      "step": 283180
    },
    {
      "epoch": 115.92,
      "grad_norm": 2.1957430839538574,
      "learning_rate": 1.9229006841433842e-06,
      "loss": 0.4809,
      "step": 283190
    },
    {
      "epoch": 115.92,
      "grad_norm": 1.502087950706482,
      "learning_rate": 1.92274463382938e-06,
      "loss": 0.473,
      "step": 283200
    },
    {
      "epoch": 115.93,
      "grad_norm": 1.8164480924606323,
      "learning_rate": 1.9225885854439363e-06,
      "loss": 0.4536,
      "step": 283210
    },
    {
      "epoch": 115.93,
      "grad_norm": 2.0342960357666016,
      "learning_rate": 1.922432538987768e-06,
      "loss": 0.4661,
      "step": 283220
    },
    {
      "epoch": 115.94,
      "grad_norm": 2.035386800765991,
      "learning_rate": 1.9222764944615896e-06,
      "loss": 0.4664,
      "step": 283230
    },
    {
      "epoch": 115.94,
      "grad_norm": 2.066148042678833,
      "learning_rate": 1.9221204518661157e-06,
      "loss": 0.4688,
      "step": 283240
    },
    {
      "epoch": 115.94,
      "grad_norm": 2.168454885482788,
      "learning_rate": 1.9219644112020616e-06,
      "loss": 0.4544,
      "step": 283250
    },
    {
      "epoch": 115.95,
      "grad_norm": 2.2740976810455322,
      "learning_rate": 1.921808372470142e-06,
      "loss": 0.46,
      "step": 283260
    },
    {
      "epoch": 115.95,
      "grad_norm": 1.9854285717010498,
      "learning_rate": 1.9216523356710718e-06,
      "loss": 0.453,
      "step": 283270
    },
    {
      "epoch": 115.96,
      "grad_norm": 1.9415682554244995,
      "learning_rate": 1.9214963008055655e-06,
      "loss": 0.4641,
      "step": 283280
    },
    {
      "epoch": 115.96,
      "grad_norm": 2.5229544639587402,
      "learning_rate": 1.9213402678743377e-06,
      "loss": 0.4627,
      "step": 283290
    },
    {
      "epoch": 115.96,
      "grad_norm": 1.90238618850708,
      "learning_rate": 1.9211842368781037e-06,
      "loss": 0.4629,
      "step": 283300
    },
    {
      "epoch": 115.97,
      "grad_norm": 1.9413881301879883,
      "learning_rate": 1.9210282078175776e-06,
      "loss": 0.46,
      "step": 283310
    },
    {
      "epoch": 115.97,
      "grad_norm": 1.7988265752792358,
      "learning_rate": 1.920872180693475e-06,
      "loss": 0.4485,
      "step": 283320
    },
    {
      "epoch": 115.98,
      "grad_norm": 2.2191808223724365,
      "learning_rate": 1.9207161555065104e-06,
      "loss": 0.4567,
      "step": 283330
    },
    {
      "epoch": 115.98,
      "grad_norm": 1.7544457912445068,
      "learning_rate": 1.920560132257398e-06,
      "loss": 0.4396,
      "step": 283340
    },
    {
      "epoch": 115.98,
      "grad_norm": 2.1756672859191895,
      "learning_rate": 1.920404110946853e-06,
      "loss": 0.4704,
      "step": 283350
    },
    {
      "epoch": 115.99,
      "grad_norm": 1.84477961063385,
      "learning_rate": 1.9202480915755906e-06,
      "loss": 0.4651,
      "step": 283360
    },
    {
      "epoch": 115.99,
      "grad_norm": 1.6595708131790161,
      "learning_rate": 1.920092074144324e-06,
      "loss": 0.4473,
      "step": 283370
    },
    {
      "epoch": 116.0,
      "grad_norm": 1.9620895385742188,
      "learning_rate": 1.919936058653769e-06,
      "loss": 0.4619,
      "step": 283380
    },
    {
      "epoch": 116.0,
      "eval_loss": 0.46444717049598694,
      "eval_runtime": 52.3843,
      "eval_samples_per_second": 65.84,
      "eval_steps_per_second": 8.247,
      "step": 283388
    },
    {
      "epoch": 116.0,
      "grad_norm": 1.6310746669769287,
      "learning_rate": 1.91978004510464e-06,
      "loss": 0.4825,
      "step": 283390
    },
    {
      "epoch": 116.0,
      "grad_norm": 1.6603084802627563,
      "learning_rate": 1.919624033497652e-06,
      "loss": 0.4534,
      "step": 283400
    },
    {
      "epoch": 116.01,
      "grad_norm": 1.5608652830123901,
      "learning_rate": 1.919468023833519e-06,
      "loss": 0.4716,
      "step": 283410
    },
    {
      "epoch": 116.01,
      "grad_norm": 2.1431713104248047,
      "learning_rate": 1.919312016112956e-06,
      "loss": 0.4731,
      "step": 283420
    },
    {
      "epoch": 116.02,
      "grad_norm": 1.8456215858459473,
      "learning_rate": 1.919156010336678e-06,
      "loss": 0.4661,
      "step": 283430
    },
    {
      "epoch": 116.02,
      "grad_norm": 1.7541669607162476,
      "learning_rate": 1.9190000065053987e-06,
      "loss": 0.4932,
      "step": 283440
    },
    {
      "epoch": 116.03,
      "grad_norm": 1.8726929426193237,
      "learning_rate": 1.9188440046198343e-06,
      "loss": 0.453,
      "step": 283450
    },
    {
      "epoch": 116.03,
      "grad_norm": 1.8579916954040527,
      "learning_rate": 1.9186880046806977e-06,
      "loss": 0.4668,
      "step": 283460
    },
    {
      "epoch": 116.03,
      "grad_norm": 1.9063998460769653,
      "learning_rate": 1.9185320066887045e-06,
      "loss": 0.4546,
      "step": 283470
    },
    {
      "epoch": 116.04,
      "grad_norm": 1.666621446609497,
      "learning_rate": 1.918376010644569e-06,
      "loss": 0.4586,
      "step": 283480
    },
    {
      "epoch": 116.04,
      "grad_norm": 1.5866366624832153,
      "learning_rate": 1.918220016549005e-06,
      "loss": 0.4522,
      "step": 283490
    },
    {
      "epoch": 116.05,
      "grad_norm": 2.1080503463745117,
      "learning_rate": 1.9180640244027293e-06,
      "loss": 0.4554,
      "step": 283500
    },
    {
      "epoch": 116.05,
      "grad_norm": 2.5794074535369873,
      "learning_rate": 1.9179080342064546e-06,
      "loss": 0.456,
      "step": 283510
    },
    {
      "epoch": 116.05,
      "grad_norm": 1.5479966402053833,
      "learning_rate": 1.917752045960896e-06,
      "loss": 0.4648,
      "step": 283520
    },
    {
      "epoch": 116.06,
      "grad_norm": 1.9337215423583984,
      "learning_rate": 1.917596059666768e-06,
      "loss": 0.4711,
      "step": 283530
    },
    {
      "epoch": 116.06,
      "grad_norm": 2.077633857727051,
      "learning_rate": 1.9174400753247853e-06,
      "loss": 0.4626,
      "step": 283540
    },
    {
      "epoch": 116.07,
      "grad_norm": 1.9198153018951416,
      "learning_rate": 1.9172840929356624e-06,
      "loss": 0.4425,
      "step": 283550
    },
    {
      "epoch": 116.07,
      "grad_norm": 1.7115974426269531,
      "learning_rate": 1.917128112500114e-06,
      "loss": 0.4873,
      "step": 283560
    },
    {
      "epoch": 116.07,
      "grad_norm": 1.8850834369659424,
      "learning_rate": 1.9169721340188536e-06,
      "loss": 0.4523,
      "step": 283570
    },
    {
      "epoch": 116.08,
      "grad_norm": 1.6246939897537231,
      "learning_rate": 1.916816157492598e-06,
      "loss": 0.4665,
      "step": 283580
    },
    {
      "epoch": 116.08,
      "grad_norm": 1.8626883029937744,
      "learning_rate": 1.916660182922059e-06,
      "loss": 0.4422,
      "step": 283590
    },
    {
      "epoch": 116.09,
      "grad_norm": 1.788319706916809,
      "learning_rate": 1.916504210307953e-06,
      "loss": 0.4701,
      "step": 283600
    },
    {
      "epoch": 116.09,
      "grad_norm": 2.0141115188598633,
      "learning_rate": 1.9163482396509926e-06,
      "loss": 0.4426,
      "step": 283610
    },
    {
      "epoch": 116.09,
      "grad_norm": 2.209338665008545,
      "learning_rate": 1.916192270951895e-06,
      "loss": 0.4358,
      "step": 283620
    },
    {
      "epoch": 116.1,
      "grad_norm": 1.7292932271957397,
      "learning_rate": 1.916036304211372e-06,
      "loss": 0.4831,
      "step": 283630
    },
    {
      "epoch": 116.1,
      "grad_norm": 1.7574462890625,
      "learning_rate": 1.9158803394301395e-06,
      "loss": 0.4623,
      "step": 283640
    },
    {
      "epoch": 116.11,
      "grad_norm": 1.8005409240722656,
      "learning_rate": 1.9157243766089117e-06,
      "loss": 0.4809,
      "step": 283650
    },
    {
      "epoch": 116.11,
      "grad_norm": 2.1695780754089355,
      "learning_rate": 1.915568415748403e-06,
      "loss": 0.4707,
      "step": 283660
    },
    {
      "epoch": 116.12,
      "grad_norm": 2.181427001953125,
      "learning_rate": 1.915412456849328e-06,
      "loss": 0.4726,
      "step": 283670
    },
    {
      "epoch": 116.12,
      "grad_norm": 2.204272747039795,
      "learning_rate": 1.9152564999124e-06,
      "loss": 0.4505,
      "step": 283680
    },
    {
      "epoch": 116.12,
      "grad_norm": 2.4153668880462646,
      "learning_rate": 1.9151005449383354e-06,
      "loss": 0.4577,
      "step": 283690
    },
    {
      "epoch": 116.13,
      "grad_norm": 1.7776429653167725,
      "learning_rate": 1.9149445919278472e-06,
      "loss": 0.4559,
      "step": 283700
    },
    {
      "epoch": 116.13,
      "grad_norm": 1.6219335794448853,
      "learning_rate": 1.91478864088165e-06,
      "loss": 0.4362,
      "step": 283710
    },
    {
      "epoch": 116.14,
      "grad_norm": 2.2348577976226807,
      "learning_rate": 1.9146326918004592e-06,
      "loss": 0.4726,
      "step": 283720
    },
    {
      "epoch": 116.14,
      "grad_norm": 1.6939818859100342,
      "learning_rate": 1.914476744684988e-06,
      "loss": 0.4627,
      "step": 283730
    },
    {
      "epoch": 116.14,
      "grad_norm": 1.753075361251831,
      "learning_rate": 1.914320799535951e-06,
      "loss": 0.434,
      "step": 283740
    },
    {
      "epoch": 116.15,
      "grad_norm": 2.741022825241089,
      "learning_rate": 1.914164856354062e-06,
      "loss": 0.4714,
      "step": 283750
    },
    {
      "epoch": 116.15,
      "grad_norm": 2.195101261138916,
      "learning_rate": 1.9140089151400366e-06,
      "loss": 0.4395,
      "step": 283760
    },
    {
      "epoch": 116.16,
      "grad_norm": 2.777714967727661,
      "learning_rate": 1.9138529758945885e-06,
      "loss": 0.4817,
      "step": 283770
    },
    {
      "epoch": 116.16,
      "grad_norm": 1.653227686882019,
      "learning_rate": 1.9136970386184324e-06,
      "loss": 0.4599,
      "step": 283780
    },
    {
      "epoch": 116.16,
      "grad_norm": 2.2422895431518555,
      "learning_rate": 1.9135411033122817e-06,
      "loss": 0.4559,
      "step": 283790
    },
    {
      "epoch": 116.17,
      "grad_norm": 2.006007194519043,
      "learning_rate": 1.913385169976852e-06,
      "loss": 0.4368,
      "step": 283800
    },
    {
      "epoch": 116.17,
      "grad_norm": 2.2315855026245117,
      "learning_rate": 1.913229238612857e-06,
      "loss": 0.4483,
      "step": 283810
    },
    {
      "epoch": 116.18,
      "grad_norm": 2.284982204437256,
      "learning_rate": 1.9130733092210104e-06,
      "loss": 0.4865,
      "step": 283820
    },
    {
      "epoch": 116.18,
      "grad_norm": 1.926327109336853,
      "learning_rate": 1.912917381802027e-06,
      "loss": 0.463,
      "step": 283830
    },
    {
      "epoch": 116.19,
      "grad_norm": 1.9480562210083008,
      "learning_rate": 1.9127614563566213e-06,
      "loss": 0.4454,
      "step": 283840
    },
    {
      "epoch": 116.19,
      "grad_norm": 2.2612650394439697,
      "learning_rate": 1.912605532885507e-06,
      "loss": 0.4797,
      "step": 283850
    },
    {
      "epoch": 116.19,
      "grad_norm": 2.017015218734741,
      "learning_rate": 1.9124496113893994e-06,
      "loss": 0.4564,
      "step": 283860
    },
    {
      "epoch": 116.2,
      "grad_norm": 2.0377745628356934,
      "learning_rate": 1.912293691869011e-06,
      "loss": 0.4548,
      "step": 283870
    },
    {
      "epoch": 116.2,
      "grad_norm": 1.8427079916000366,
      "learning_rate": 1.9121377743250577e-06,
      "loss": 0.4729,
      "step": 283880
    },
    {
      "epoch": 116.21,
      "grad_norm": 1.9633097648620605,
      "learning_rate": 1.9119818587582534e-06,
      "loss": 0.4724,
      "step": 283890
    },
    {
      "epoch": 116.21,
      "grad_norm": 1.9899846315383911,
      "learning_rate": 1.9118259451693118e-06,
      "loss": 0.4753,
      "step": 283900
    },
    {
      "epoch": 116.21,
      "grad_norm": 2.3687374591827393,
      "learning_rate": 1.9116700335589476e-06,
      "loss": 0.4677,
      "step": 283910
    },
    {
      "epoch": 116.22,
      "grad_norm": 2.12898588180542,
      "learning_rate": 1.911514123927875e-06,
      "loss": 0.471,
      "step": 283920
    },
    {
      "epoch": 116.22,
      "grad_norm": 1.8857228755950928,
      "learning_rate": 1.911358216276808e-06,
      "loss": 0.4572,
      "step": 283930
    },
    {
      "epoch": 116.23,
      "grad_norm": 2.0040042400360107,
      "learning_rate": 1.911202310606461e-06,
      "loss": 0.4501,
      "step": 283940
    },
    {
      "epoch": 116.23,
      "grad_norm": 1.7125890254974365,
      "learning_rate": 1.911046406917547e-06,
      "loss": 0.4636,
      "step": 283950
    },
    {
      "epoch": 116.23,
      "grad_norm": 1.796841025352478,
      "learning_rate": 1.9108905052107824e-06,
      "loss": 0.4865,
      "step": 283960
    },
    {
      "epoch": 116.24,
      "grad_norm": 2.162135362625122,
      "learning_rate": 1.910734605486879e-06,
      "loss": 0.4746,
      "step": 283970
    },
    {
      "epoch": 116.24,
      "grad_norm": 2.0517702102661133,
      "learning_rate": 1.9105787077465527e-06,
      "loss": 0.4746,
      "step": 283980
    },
    {
      "epoch": 116.25,
      "grad_norm": 2.166512966156006,
      "learning_rate": 1.9104228119905167e-06,
      "loss": 0.4554,
      "step": 283990
    },
    {
      "epoch": 116.25,
      "grad_norm": 1.7563132047653198,
      "learning_rate": 1.9102669182194855e-06,
      "loss": 0.4629,
      "step": 284000
    },
    {
      "epoch": 116.25,
      "grad_norm": 2.090672016143799,
      "learning_rate": 1.9101110264341735e-06,
      "loss": 0.4554,
      "step": 284010
    },
    {
      "epoch": 116.26,
      "grad_norm": 1.8262622356414795,
      "learning_rate": 1.909955136635294e-06,
      "loss": 0.4644,
      "step": 284020
    },
    {
      "epoch": 116.26,
      "grad_norm": 2.261549472808838,
      "learning_rate": 1.9097992488235622e-06,
      "loss": 0.4708,
      "step": 284030
    },
    {
      "epoch": 116.27,
      "grad_norm": 1.999348521232605,
      "learning_rate": 1.9096433629996914e-06,
      "loss": 0.4582,
      "step": 284040
    },
    {
      "epoch": 116.27,
      "grad_norm": 2.815067768096924,
      "learning_rate": 1.9094874791643956e-06,
      "loss": 0.4559,
      "step": 284050
    },
    {
      "epoch": 116.28,
      "grad_norm": 1.9562597274780273,
      "learning_rate": 1.909331597318388e-06,
      "loss": 0.4672,
      "step": 284060
    },
    {
      "epoch": 116.28,
      "grad_norm": 1.7244606018066406,
      "learning_rate": 1.909175717462385e-06,
      "loss": 0.456,
      "step": 284070
    },
    {
      "epoch": 116.28,
      "grad_norm": 1.9627519845962524,
      "learning_rate": 1.9090198395971e-06,
      "loss": 0.4522,
      "step": 284080
    },
    {
      "epoch": 116.29,
      "grad_norm": 2.111539125442505,
      "learning_rate": 1.908863963723246e-06,
      "loss": 0.462,
      "step": 284090
    },
    {
      "epoch": 116.29,
      "grad_norm": 1.6559455394744873,
      "learning_rate": 1.908708089841538e-06,
      "loss": 0.4473,
      "step": 284100
    },
    {
      "epoch": 116.3,
      "grad_norm": 1.8144282102584839,
      "learning_rate": 1.908552217952689e-06,
      "loss": 0.4599,
      "step": 284110
    },
    {
      "epoch": 116.3,
      "grad_norm": 2.2802417278289795,
      "learning_rate": 1.9083963480574137e-06,
      "loss": 0.462,
      "step": 284120
    },
    {
      "epoch": 116.3,
      "grad_norm": 1.7273552417755127,
      "learning_rate": 1.9082404801564263e-06,
      "loss": 0.462,
      "step": 284130
    },
    {
      "epoch": 116.31,
      "grad_norm": 1.7418676614761353,
      "learning_rate": 1.9080846142504404e-06,
      "loss": 0.4726,
      "step": 284140
    },
    {
      "epoch": 116.31,
      "grad_norm": 1.9323502779006958,
      "learning_rate": 1.9079287503401704e-06,
      "loss": 0.4603,
      "step": 284150
    },
    {
      "epoch": 116.32,
      "grad_norm": 2.1231961250305176,
      "learning_rate": 1.90777288842633e-06,
      "loss": 0.4667,
      "step": 284160
    },
    {
      "epoch": 116.32,
      "grad_norm": 2.0455029010772705,
      "learning_rate": 1.9076170285096336e-06,
      "loss": 0.4565,
      "step": 284170
    },
    {
      "epoch": 116.32,
      "grad_norm": 1.680962324142456,
      "learning_rate": 1.9074611705907944e-06,
      "loss": 0.4659,
      "step": 284180
    },
    {
      "epoch": 116.33,
      "grad_norm": 2.184494733810425,
      "learning_rate": 1.9073053146705265e-06,
      "loss": 0.4644,
      "step": 284190
    },
    {
      "epoch": 116.33,
      "grad_norm": 2.040362596511841,
      "learning_rate": 1.9071494607495443e-06,
      "loss": 0.4773,
      "step": 284200
    },
    {
      "epoch": 116.34,
      "grad_norm": 2.551676034927368,
      "learning_rate": 1.9069936088285617e-06,
      "loss": 0.4575,
      "step": 284210
    },
    {
      "epoch": 116.34,
      "grad_norm": 1.8838551044464111,
      "learning_rate": 1.9068377589082921e-06,
      "loss": 0.4762,
      "step": 284220
    },
    {
      "epoch": 116.34,
      "grad_norm": 1.7589064836502075,
      "learning_rate": 1.9066819109894502e-06,
      "loss": 0.4628,
      "step": 284230
    },
    {
      "epoch": 116.35,
      "grad_norm": 1.8915917873382568,
      "learning_rate": 1.9065260650727494e-06,
      "loss": 0.4675,
      "step": 284240
    },
    {
      "epoch": 116.35,
      "grad_norm": 1.7928169965744019,
      "learning_rate": 1.9063702211589031e-06,
      "loss": 0.4705,
      "step": 284250
    },
    {
      "epoch": 116.36,
      "grad_norm": 1.9724392890930176,
      "learning_rate": 1.9062143792486266e-06,
      "loss": 0.4709,
      "step": 284260
    },
    {
      "epoch": 116.36,
      "grad_norm": 1.700074553489685,
      "learning_rate": 1.9060585393426328e-06,
      "loss": 0.4588,
      "step": 284270
    },
    {
      "epoch": 116.37,
      "grad_norm": 1.8127793073654175,
      "learning_rate": 1.905902701441636e-06,
      "loss": 0.472,
      "step": 284280
    },
    {
      "epoch": 116.37,
      "grad_norm": 1.864396095275879,
      "learning_rate": 1.9057468655463495e-06,
      "loss": 0.4963,
      "step": 284290
    },
    {
      "epoch": 116.37,
      "grad_norm": 1.5359423160552979,
      "learning_rate": 1.9055910316574881e-06,
      "loss": 0.4529,
      "step": 284300
    },
    {
      "epoch": 116.38,
      "grad_norm": 2.47520112991333,
      "learning_rate": 1.905435199775765e-06,
      "loss": 0.4508,
      "step": 284310
    },
    {
      "epoch": 116.38,
      "grad_norm": 1.8318018913269043,
      "learning_rate": 1.9052793699018936e-06,
      "loss": 0.4787,
      "step": 284320
    },
    {
      "epoch": 116.39,
      "grad_norm": 2.1922152042388916,
      "learning_rate": 1.9051235420365887e-06,
      "loss": 0.4649,
      "step": 284330
    },
    {
      "epoch": 116.39,
      "grad_norm": 1.773115634918213,
      "learning_rate": 1.9049677161805635e-06,
      "loss": 0.4586,
      "step": 284340
    },
    {
      "epoch": 116.39,
      "grad_norm": 2.30438494682312,
      "learning_rate": 1.904811892334532e-06,
      "loss": 0.4621,
      "step": 284350
    },
    {
      "epoch": 116.4,
      "grad_norm": 1.818778157234192,
      "learning_rate": 1.9046560704992078e-06,
      "loss": 0.4854,
      "step": 284360
    },
    {
      "epoch": 116.4,
      "grad_norm": 1.655056118965149,
      "learning_rate": 1.904500250675305e-06,
      "loss": 0.4467,
      "step": 284370
    },
    {
      "epoch": 116.41,
      "grad_norm": 1.7434507608413696,
      "learning_rate": 1.9043444328635371e-06,
      "loss": 0.4533,
      "step": 284380
    },
    {
      "epoch": 116.41,
      "grad_norm": 2.5834081172943115,
      "learning_rate": 1.904188617064619e-06,
      "loss": 0.4782,
      "step": 284390
    },
    {
      "epoch": 116.41,
      "grad_norm": 2.351792812347412,
      "learning_rate": 1.9040328032792627e-06,
      "loss": 0.453,
      "step": 284400
    },
    {
      "epoch": 116.42,
      "grad_norm": 2.0832159519195557,
      "learning_rate": 1.903876991508183e-06,
      "loss": 0.4656,
      "step": 284410
    },
    {
      "epoch": 116.42,
      "grad_norm": 1.8416101932525635,
      "learning_rate": 1.9037211817520932e-06,
      "loss": 0.4697,
      "step": 284420
    },
    {
      "epoch": 116.43,
      "grad_norm": 1.9317457675933838,
      "learning_rate": 1.9035653740117065e-06,
      "loss": 0.4639,
      "step": 284430
    },
    {
      "epoch": 116.43,
      "grad_norm": 1.8467583656311035,
      "learning_rate": 1.9034095682877386e-06,
      "loss": 0.4551,
      "step": 284440
    },
    {
      "epoch": 116.43,
      "grad_norm": 1.928604245185852,
      "learning_rate": 1.9032537645809016e-06,
      "loss": 0.4548,
      "step": 284450
    },
    {
      "epoch": 116.44,
      "grad_norm": 1.9770958423614502,
      "learning_rate": 1.9030979628919099e-06,
      "loss": 0.4627,
      "step": 284460
    },
    {
      "epoch": 116.44,
      "grad_norm": 1.8462685346603394,
      "learning_rate": 1.9029421632214766e-06,
      "loss": 0.4561,
      "step": 284470
    },
    {
      "epoch": 116.45,
      "grad_norm": 1.8425273895263672,
      "learning_rate": 1.9027863655703162e-06,
      "loss": 0.4648,
      "step": 284480
    },
    {
      "epoch": 116.45,
      "grad_norm": 1.835839033126831,
      "learning_rate": 1.9026305699391415e-06,
      "loss": 0.4548,
      "step": 284490
    },
    {
      "epoch": 116.46,
      "grad_norm": 2.988133192062378,
      "learning_rate": 1.9024747763286665e-06,
      "loss": 0.4686,
      "step": 284500
    },
    {
      "epoch": 116.46,
      "grad_norm": 1.7968604564666748,
      "learning_rate": 1.9023189847396053e-06,
      "loss": 0.4562,
      "step": 284510
    },
    {
      "epoch": 116.46,
      "grad_norm": 1.8932169675827026,
      "learning_rate": 1.9021631951726717e-06,
      "loss": 0.4607,
      "step": 284520
    },
    {
      "epoch": 116.47,
      "grad_norm": 1.6068854331970215,
      "learning_rate": 1.9020074076285781e-06,
      "loss": 0.4594,
      "step": 284530
    },
    {
      "epoch": 116.47,
      "grad_norm": 2.2531044483184814,
      "learning_rate": 1.901851622108039e-06,
      "loss": 0.4646,
      "step": 284540
    },
    {
      "epoch": 116.48,
      "grad_norm": 2.0839390754699707,
      "learning_rate": 1.9016958386117681e-06,
      "loss": 0.4569,
      "step": 284550
    },
    {
      "epoch": 116.48,
      "grad_norm": 1.632704496383667,
      "learning_rate": 1.9015400571404788e-06,
      "loss": 0.4476,
      "step": 284560
    },
    {
      "epoch": 116.48,
      "grad_norm": 2.034663200378418,
      "learning_rate": 1.9013842776948846e-06,
      "loss": 0.4464,
      "step": 284570
    },
    {
      "epoch": 116.49,
      "grad_norm": 1.9613736867904663,
      "learning_rate": 1.9012285002756992e-06,
      "loss": 0.4532,
      "step": 284580
    },
    {
      "epoch": 116.49,
      "grad_norm": 1.8885549306869507,
      "learning_rate": 1.9010727248836367e-06,
      "loss": 0.4569,
      "step": 284590
    },
    {
      "epoch": 116.5,
      "grad_norm": 1.768513798713684,
      "learning_rate": 1.9009169515194098e-06,
      "loss": 0.4661,
      "step": 284600
    },
    {
      "epoch": 116.5,
      "grad_norm": 1.8398683071136475,
      "learning_rate": 1.9007611801837332e-06,
      "loss": 0.4834,
      "step": 284610
    },
    {
      "epoch": 116.5,
      "grad_norm": 1.9742852449417114,
      "learning_rate": 1.9006054108773182e-06,
      "loss": 0.4622,
      "step": 284620
    },
    {
      "epoch": 116.51,
      "grad_norm": 2.161987543106079,
      "learning_rate": 1.9004496436008812e-06,
      "loss": 0.4655,
      "step": 284630
    },
    {
      "epoch": 116.51,
      "grad_norm": 1.8777399063110352,
      "learning_rate": 1.900293878355134e-06,
      "loss": 0.4727,
      "step": 284640
    },
    {
      "epoch": 116.52,
      "grad_norm": 2.3283395767211914,
      "learning_rate": 1.900138115140791e-06,
      "loss": 0.4604,
      "step": 284650
    },
    {
      "epoch": 116.52,
      "grad_norm": 4.4593892097473145,
      "learning_rate": 1.8999823539585656e-06,
      "loss": 0.4724,
      "step": 284660
    },
    {
      "epoch": 116.52,
      "grad_norm": 1.722902536392212,
      "learning_rate": 1.8998265948091707e-06,
      "loss": 0.4717,
      "step": 284670
    },
    {
      "epoch": 116.53,
      "grad_norm": 2.083434581756592,
      "learning_rate": 1.8996708376933202e-06,
      "loss": 0.4619,
      "step": 284680
    },
    {
      "epoch": 116.53,
      "grad_norm": 1.8112224340438843,
      "learning_rate": 1.8995150826117275e-06,
      "loss": 0.4675,
      "step": 284690
    },
    {
      "epoch": 116.54,
      "grad_norm": 2.1146974563598633,
      "learning_rate": 1.8993593295651062e-06,
      "loss": 0.4764,
      "step": 284700
    },
    {
      "epoch": 116.54,
      "grad_norm": 2.665992021560669,
      "learning_rate": 1.8992035785541698e-06,
      "loss": 0.4604,
      "step": 284710
    },
    {
      "epoch": 116.55,
      "grad_norm": 1.9827815294265747,
      "learning_rate": 1.8990478295796315e-06,
      "loss": 0.4736,
      "step": 284720
    },
    {
      "epoch": 116.55,
      "grad_norm": 1.7907263040542603,
      "learning_rate": 1.8988920826422051e-06,
      "loss": 0.4697,
      "step": 284730
    },
    {
      "epoch": 116.55,
      "grad_norm": 1.9544508457183838,
      "learning_rate": 1.8987363377426039e-06,
      "loss": 0.4525,
      "step": 284740
    },
    {
      "epoch": 116.56,
      "grad_norm": 2.050447463989258,
      "learning_rate": 1.8985805948815417e-06,
      "loss": 0.4803,
      "step": 284750
    },
    {
      "epoch": 116.56,
      "grad_norm": 1.9887498617172241,
      "learning_rate": 1.8984248540597316e-06,
      "loss": 0.4703,
      "step": 284760
    },
    {
      "epoch": 116.57,
      "grad_norm": 2.291656970977783,
      "learning_rate": 1.8982691152778867e-06,
      "loss": 0.4444,
      "step": 284770
    },
    {
      "epoch": 116.57,
      "grad_norm": 1.6393929719924927,
      "learning_rate": 1.8981133785367209e-06,
      "loss": 0.4627,
      "step": 284780
    },
    {
      "epoch": 116.57,
      "grad_norm": 1.6749531030654907,
      "learning_rate": 1.8979576438369472e-06,
      "loss": 0.4605,
      "step": 284790
    },
    {
      "epoch": 116.58,
      "grad_norm": 1.7412794828414917,
      "learning_rate": 1.8978019111792793e-06,
      "loss": 0.4824,
      "step": 284800
    },
    {
      "epoch": 116.58,
      "grad_norm": 2.1591453552246094,
      "learning_rate": 1.89764618056443e-06,
      "loss": 0.4629,
      "step": 284810
    },
    {
      "epoch": 116.59,
      "grad_norm": 1.9498822689056396,
      "learning_rate": 1.897490451993114e-06,
      "loss": 0.4693,
      "step": 284820
    },
    {
      "epoch": 116.59,
      "grad_norm": 2.6832096576690674,
      "learning_rate": 1.8973347254660436e-06,
      "loss": 0.4748,
      "step": 284830
    },
    {
      "epoch": 116.59,
      "grad_norm": 2.0447235107421875,
      "learning_rate": 1.8971790009839324e-06,
      "loss": 0.4781,
      "step": 284840
    },
    {
      "epoch": 116.6,
      "grad_norm": 1.8880151510238647,
      "learning_rate": 1.8970232785474939e-06,
      "loss": 0.4695,
      "step": 284850
    },
    {
      "epoch": 116.6,
      "grad_norm": 2.0975441932678223,
      "learning_rate": 1.8968675581574412e-06,
      "loss": 0.4585,
      "step": 284860
    },
    {
      "epoch": 116.61,
      "grad_norm": 1.789251446723938,
      "learning_rate": 1.896711839814488e-06,
      "loss": 0.4884,
      "step": 284870
    },
    {
      "epoch": 116.61,
      "grad_norm": 1.844185709953308,
      "learning_rate": 1.8965561235193478e-06,
      "loss": 0.4654,
      "step": 284880
    },
    {
      "epoch": 116.61,
      "grad_norm": 1.8229198455810547,
      "learning_rate": 1.896400409272733e-06,
      "loss": 0.4812,
      "step": 284890
    },
    {
      "epoch": 116.62,
      "grad_norm": 1.9509367942810059,
      "learning_rate": 1.8962446970753573e-06,
      "loss": 0.4756,
      "step": 284900
    },
    {
      "epoch": 116.62,
      "grad_norm": 1.8413180112838745,
      "learning_rate": 1.8960889869279341e-06,
      "loss": 0.4413,
      "step": 284910
    },
    {
      "epoch": 116.63,
      "grad_norm": 1.911026954650879,
      "learning_rate": 1.8959332788311766e-06,
      "loss": 0.4535,
      "step": 284920
    },
    {
      "epoch": 116.63,
      "grad_norm": 1.8847174644470215,
      "learning_rate": 1.8957775727857983e-06,
      "loss": 0.4651,
      "step": 284930
    },
    {
      "epoch": 116.64,
      "grad_norm": 1.730829119682312,
      "learning_rate": 1.8956218687925122e-06,
      "loss": 0.4574,
      "step": 284940
    },
    {
      "epoch": 116.64,
      "grad_norm": 1.8302621841430664,
      "learning_rate": 1.895466166852032e-06,
      "loss": 0.4762,
      "step": 284950
    },
    {
      "epoch": 116.64,
      "grad_norm": 1.704469919204712,
      "learning_rate": 1.89531046696507e-06,
      "loss": 0.466,
      "step": 284960
    },
    {
      "epoch": 116.65,
      "grad_norm": 2.1063618659973145,
      "learning_rate": 1.895154769132341e-06,
      "loss": 0.4471,
      "step": 284970
    },
    {
      "epoch": 116.65,
      "grad_norm": 2.7266454696655273,
      "learning_rate": 1.8949990733545565e-06,
      "loss": 0.4486,
      "step": 284980
    },
    {
      "epoch": 116.66,
      "grad_norm": 2.112494707107544,
      "learning_rate": 1.894843379632431e-06,
      "loss": 0.4827,
      "step": 284990
    },
    {
      "epoch": 116.66,
      "grad_norm": 2.203768014907837,
      "learning_rate": 1.8946876879666757e-06,
      "loss": 0.4511,
      "step": 285000
    },
    {
      "epoch": 116.66,
      "grad_norm": 2.2891576290130615,
      "learning_rate": 1.8945319983580064e-06,
      "loss": 0.464,
      "step": 285010
    },
    {
      "epoch": 116.67,
      "grad_norm": 2.122488021850586,
      "learning_rate": 1.8943763108071359e-06,
      "loss": 0.4724,
      "step": 285020
    },
    {
      "epoch": 116.67,
      "grad_norm": 1.7622047662734985,
      "learning_rate": 1.894220625314776e-06,
      "loss": 0.4737,
      "step": 285030
    },
    {
      "epoch": 116.68,
      "grad_norm": 2.493684768676758,
      "learning_rate": 1.8940649418816406e-06,
      "loss": 0.4699,
      "step": 285040
    },
    {
      "epoch": 116.68,
      "grad_norm": 1.6762092113494873,
      "learning_rate": 1.893909260508443e-06,
      "loss": 0.4609,
      "step": 285050
    },
    {
      "epoch": 116.68,
      "grad_norm": 1.7714505195617676,
      "learning_rate": 1.8937535811958957e-06,
      "loss": 0.4617,
      "step": 285060
    },
    {
      "epoch": 116.69,
      "grad_norm": 1.7728984355926514,
      "learning_rate": 1.8935979039447127e-06,
      "loss": 0.4577,
      "step": 285070
    },
    {
      "epoch": 116.69,
      "grad_norm": 2.2939541339874268,
      "learning_rate": 1.8934422287556066e-06,
      "loss": 0.4765,
      "step": 285080
    },
    {
      "epoch": 116.7,
      "grad_norm": 1.527514100074768,
      "learning_rate": 1.8932865556292906e-06,
      "loss": 0.4506,
      "step": 285090
    },
    {
      "epoch": 116.7,
      "grad_norm": 2.617708683013916,
      "learning_rate": 1.8931308845664778e-06,
      "loss": 0.4639,
      "step": 285100
    },
    {
      "epoch": 116.7,
      "grad_norm": 1.5621395111083984,
      "learning_rate": 1.8929752155678821e-06,
      "loss": 0.4766,
      "step": 285110
    },
    {
      "epoch": 116.71,
      "grad_norm": 2.5944342613220215,
      "learning_rate": 1.8928195486342154e-06,
      "loss": 0.4808,
      "step": 285120
    },
    {
      "epoch": 116.71,
      "grad_norm": 1.8435537815093994,
      "learning_rate": 1.8926638837661911e-06,
      "loss": 0.4601,
      "step": 285130
    },
    {
      "epoch": 116.72,
      "grad_norm": 2.523956537246704,
      "learning_rate": 1.8925082209645226e-06,
      "loss": 0.4777,
      "step": 285140
    },
    {
      "epoch": 116.72,
      "grad_norm": 2.157386302947998,
      "learning_rate": 1.8923525602299228e-06,
      "loss": 0.4636,
      "step": 285150
    },
    {
      "epoch": 116.73,
      "grad_norm": 1.6677736043930054,
      "learning_rate": 1.8921969015631047e-06,
      "loss": 0.4406,
      "step": 285160
    },
    {
      "epoch": 116.73,
      "grad_norm": 1.6188082695007324,
      "learning_rate": 1.8920412449647813e-06,
      "loss": 0.4614,
      "step": 285170
    },
    {
      "epoch": 116.73,
      "grad_norm": 1.9978833198547363,
      "learning_rate": 1.8918855904356658e-06,
      "loss": 0.4671,
      "step": 285180
    },
    {
      "epoch": 116.74,
      "grad_norm": 1.781262993812561,
      "learning_rate": 1.8917299379764707e-06,
      "loss": 0.4559,
      "step": 285190
    },
    {
      "epoch": 116.74,
      "grad_norm": 2.448885202407837,
      "learning_rate": 1.8915742875879101e-06,
      "loss": 0.4599,
      "step": 285200
    },
    {
      "epoch": 116.75,
      "grad_norm": 1.867293119430542,
      "learning_rate": 1.8914186392706962e-06,
      "loss": 0.445,
      "step": 285210
    },
    {
      "epoch": 116.75,
      "grad_norm": 1.987981915473938,
      "learning_rate": 1.8912629930255424e-06,
      "loss": 0.4575,
      "step": 285220
    },
    {
      "epoch": 116.75,
      "grad_norm": 2.0523576736450195,
      "learning_rate": 1.8911073488531614e-06,
      "loss": 0.4825,
      "step": 285230
    },
    {
      "epoch": 116.76,
      "grad_norm": 2.1473493576049805,
      "learning_rate": 1.8909517067542667e-06,
      "loss": 0.4428,
      "step": 285240
    },
    {
      "epoch": 116.76,
      "grad_norm": 1.9056004285812378,
      "learning_rate": 1.8907960667295705e-06,
      "loss": 0.4592,
      "step": 285250
    },
    {
      "epoch": 116.77,
      "grad_norm": 1.868451476097107,
      "learning_rate": 1.8906404287797864e-06,
      "loss": 0.4494,
      "step": 285260
    },
    {
      "epoch": 116.77,
      "grad_norm": 1.8905402421951294,
      "learning_rate": 1.8904847929056267e-06,
      "loss": 0.4571,
      "step": 285270
    },
    {
      "epoch": 116.77,
      "grad_norm": 1.767064094543457,
      "learning_rate": 1.8903291591078046e-06,
      "loss": 0.481,
      "step": 285280
    },
    {
      "epoch": 116.78,
      "grad_norm": 2.328387498855591,
      "learning_rate": 1.8901735273870334e-06,
      "loss": 0.4744,
      "step": 285290
    },
    {
      "epoch": 116.78,
      "grad_norm": 2.646026849746704,
      "learning_rate": 1.8900178977440258e-06,
      "loss": 0.4478,
      "step": 285300
    },
    {
      "epoch": 116.79,
      "grad_norm": 2.233367443084717,
      "learning_rate": 1.8898622701794946e-06,
      "loss": 0.464,
      "step": 285310
    },
    {
      "epoch": 116.79,
      "grad_norm": 1.9059717655181885,
      "learning_rate": 1.8897066446941529e-06,
      "loss": 0.4724,
      "step": 285320
    },
    {
      "epoch": 116.79,
      "grad_norm": 1.729966163635254,
      "learning_rate": 1.8895510212887137e-06,
      "loss": 0.4703,
      "step": 285330
    },
    {
      "epoch": 116.8,
      "grad_norm": 2.465348958969116,
      "learning_rate": 1.8893953999638894e-06,
      "loss": 0.5004,
      "step": 285340
    },
    {
      "epoch": 116.8,
      "grad_norm": 2.677800416946411,
      "learning_rate": 1.8892397807203932e-06,
      "loss": 0.4707,
      "step": 285350
    },
    {
      "epoch": 116.81,
      "grad_norm": 1.8250648975372314,
      "learning_rate": 1.8890841635589379e-06,
      "loss": 0.4529,
      "step": 285360
    },
    {
      "epoch": 116.81,
      "grad_norm": 5.930434226989746,
      "learning_rate": 1.8889285484802362e-06,
      "loss": 0.4628,
      "step": 285370
    },
    {
      "epoch": 116.82,
      "grad_norm": 1.931890606880188,
      "learning_rate": 1.8887729354850007e-06,
      "loss": 0.4765,
      "step": 285380
    },
    {
      "epoch": 116.82,
      "grad_norm": 1.7385038137435913,
      "learning_rate": 1.8886173245739453e-06,
      "loss": 0.4614,
      "step": 285390
    },
    {
      "epoch": 116.82,
      "grad_norm": 1.7692029476165771,
      "learning_rate": 1.8884617157477822e-06,
      "loss": 0.471,
      "step": 285400
    },
    {
      "epoch": 116.83,
      "grad_norm": 1.9042348861694336,
      "learning_rate": 1.8883061090072242e-06,
      "loss": 0.4533,
      "step": 285410
    },
    {
      "epoch": 116.83,
      "grad_norm": 1.8746812343597412,
      "learning_rate": 1.888150504352984e-06,
      "loss": 0.479,
      "step": 285420
    },
    {
      "epoch": 116.84,
      "grad_norm": 1.9397884607315063,
      "learning_rate": 1.8879949017857747e-06,
      "loss": 0.4613,
      "step": 285430
    },
    {
      "epoch": 116.84,
      "grad_norm": 1.871070384979248,
      "learning_rate": 1.8878393013063088e-06,
      "loss": 0.4625,
      "step": 285440
    },
    {
      "epoch": 116.84,
      "grad_norm": 1.7795859575271606,
      "learning_rate": 1.8876837029152993e-06,
      "loss": 0.4812,
      "step": 285450
    },
    {
      "epoch": 116.85,
      "grad_norm": 1.8508728742599487,
      "learning_rate": 1.8875281066134586e-06,
      "loss": 0.4868,
      "step": 285460
    },
    {
      "epoch": 116.85,
      "grad_norm": 1.5800584554672241,
      "learning_rate": 1.8873725124015008e-06,
      "loss": 0.4792,
      "step": 285470
    },
    {
      "epoch": 116.86,
      "grad_norm": 1.999944806098938,
      "learning_rate": 1.8872169202801367e-06,
      "loss": 0.4582,
      "step": 285480
    },
    {
      "epoch": 116.86,
      "grad_norm": 1.7680151462554932,
      "learning_rate": 1.88706133025008e-06,
      "loss": 0.4658,
      "step": 285490
    },
    {
      "epoch": 116.86,
      "grad_norm": 2.3029472827911377,
      "learning_rate": 1.8869057423120433e-06,
      "loss": 0.4781,
      "step": 285500
    },
    {
      "epoch": 116.87,
      "grad_norm": 2.374924898147583,
      "learning_rate": 1.8867501564667394e-06,
      "loss": 0.4804,
      "step": 285510
    },
    {
      "epoch": 116.87,
      "grad_norm": 1.9704599380493164,
      "learning_rate": 1.886594572714881e-06,
      "loss": 0.4723,
      "step": 285520
    },
    {
      "epoch": 116.88,
      "grad_norm": 2.501967430114746,
      "learning_rate": 1.886438991057181e-06,
      "loss": 0.4582,
      "step": 285530
    },
    {
      "epoch": 116.88,
      "grad_norm": 2.2231340408325195,
      "learning_rate": 1.8862834114943516e-06,
      "loss": 0.4909,
      "step": 285540
    },
    {
      "epoch": 116.88,
      "grad_norm": 2.530654191970825,
      "learning_rate": 1.8861278340271067e-06,
      "loss": 0.4562,
      "step": 285550
    },
    {
      "epoch": 116.89,
      "grad_norm": 2.095470428466797,
      "learning_rate": 1.8859722586561572e-06,
      "loss": 0.4483,
      "step": 285560
    },
    {
      "epoch": 116.89,
      "grad_norm": 1.8848265409469604,
      "learning_rate": 1.8858166853822159e-06,
      "loss": 0.4487,
      "step": 285570
    },
    {
      "epoch": 116.9,
      "grad_norm": 2.1246070861816406,
      "learning_rate": 1.8856611142059971e-06,
      "loss": 0.4618,
      "step": 285580
    },
    {
      "epoch": 116.9,
      "grad_norm": 2.010179042816162,
      "learning_rate": 1.8855055451282123e-06,
      "loss": 0.4663,
      "step": 285590
    },
    {
      "epoch": 116.91,
      "grad_norm": 1.8282933235168457,
      "learning_rate": 1.8853499781495752e-06,
      "loss": 0.4732,
      "step": 285600
    },
    {
      "epoch": 116.91,
      "grad_norm": 1.6707216501235962,
      "learning_rate": 1.885194413270797e-06,
      "loss": 0.4582,
      "step": 285610
    },
    {
      "epoch": 116.91,
      "grad_norm": 1.701484203338623,
      "learning_rate": 1.885038850492591e-06,
      "loss": 0.4353,
      "step": 285620
    },
    {
      "epoch": 116.92,
      "grad_norm": 2.0412778854370117,
      "learning_rate": 1.8848832898156696e-06,
      "loss": 0.4538,
      "step": 285630
    },
    {
      "epoch": 116.92,
      "grad_norm": 1.64910089969635,
      "learning_rate": 1.8847277312407457e-06,
      "loss": 0.4554,
      "step": 285640
    },
    {
      "epoch": 116.93,
      "grad_norm": 2.178694009780884,
      "learning_rate": 1.8845721747685318e-06,
      "loss": 0.4655,
      "step": 285650
    },
    {
      "epoch": 116.93,
      "grad_norm": 1.9230443239212036,
      "learning_rate": 1.88441662039974e-06,
      "loss": 0.4665,
      "step": 285660
    },
    {
      "epoch": 116.93,
      "grad_norm": 1.8137269020080566,
      "learning_rate": 1.8842610681350837e-06,
      "loss": 0.4492,
      "step": 285670
    },
    {
      "epoch": 116.94,
      "grad_norm": 2.0788936614990234,
      "learning_rate": 1.8841055179752747e-06,
      "loss": 0.4568,
      "step": 285680
    },
    {
      "epoch": 116.94,
      "grad_norm": 1.894661784172058,
      "learning_rate": 1.8839499699210268e-06,
      "loss": 0.4501,
      "step": 285690
    },
    {
      "epoch": 116.95,
      "grad_norm": 2.215542793273926,
      "learning_rate": 1.8837944239730508e-06,
      "loss": 0.4621,
      "step": 285700
    },
    {
      "epoch": 116.95,
      "grad_norm": 1.7691465616226196,
      "learning_rate": 1.8836388801320601e-06,
      "loss": 0.474,
      "step": 285710
    },
    {
      "epoch": 116.95,
      "grad_norm": 1.8406864404678345,
      "learning_rate": 1.8834833383987672e-06,
      "loss": 0.462,
      "step": 285720
    },
    {
      "epoch": 116.96,
      "grad_norm": 1.6520881652832031,
      "learning_rate": 1.8833277987738845e-06,
      "loss": 0.4584,
      "step": 285730
    },
    {
      "epoch": 116.96,
      "grad_norm": 2.1150054931640625,
      "learning_rate": 1.883172261258125e-06,
      "loss": 0.4718,
      "step": 285740
    },
    {
      "epoch": 116.97,
      "grad_norm": 2.8892579078674316,
      "learning_rate": 1.8830167258521997e-06,
      "loss": 0.4743,
      "step": 285750
    },
    {
      "epoch": 116.97,
      "grad_norm": 1.9805713891983032,
      "learning_rate": 1.882861192556823e-06,
      "loss": 0.4527,
      "step": 285760
    },
    {
      "epoch": 116.98,
      "grad_norm": 2.0361740589141846,
      "learning_rate": 1.8827056613727067e-06,
      "loss": 0.4741,
      "step": 285770
    },
    {
      "epoch": 116.98,
      "grad_norm": 1.963733196258545,
      "learning_rate": 1.8825501323005626e-06,
      "loss": 0.474,
      "step": 285780
    },
    {
      "epoch": 116.98,
      "grad_norm": 2.1524102687835693,
      "learning_rate": 1.882394605341104e-06,
      "loss": 0.4514,
      "step": 285790
    },
    {
      "epoch": 116.99,
      "grad_norm": 2.0884673595428467,
      "learning_rate": 1.8822390804950432e-06,
      "loss": 0.4658,
      "step": 285800
    },
    {
      "epoch": 116.99,
      "grad_norm": 2.236830949783325,
      "learning_rate": 1.8820835577630923e-06,
      "loss": 0.4735,
      "step": 285810
    },
    {
      "epoch": 117.0,
      "grad_norm": 1.9785258769989014,
      "learning_rate": 1.8819280371459638e-06,
      "loss": 0.461,
      "step": 285820
    },
    {
      "epoch": 117.0,
      "grad_norm": 1.7911698818206787,
      "learning_rate": 1.8817725186443708e-06,
      "loss": 0.4533,
      "step": 285830
    },
    {
      "epoch": 117.0,
      "eval_loss": 0.46632975339889526,
      "eval_runtime": 52.4193,
      "eval_samples_per_second": 65.796,
      "eval_steps_per_second": 8.241,
      "step": 285831
    },
    {
      "epoch": 117.0,
      "grad_norm": 2.603872299194336,
      "learning_rate": 1.8816170022590243e-06,
      "loss": 0.4754,
      "step": 285840
    },
    {
      "epoch": 117.01,
      "grad_norm": 1.8178330659866333,
      "learning_rate": 1.881461487990638e-06,
      "loss": 0.4692,
      "step": 285850
    },
    {
      "epoch": 117.01,
      "grad_norm": 1.90498685836792,
      "learning_rate": 1.8813059758399234e-06,
      "loss": 0.4679,
      "step": 285860
    },
    {
      "epoch": 117.02,
      "grad_norm": 1.8030142784118652,
      "learning_rate": 1.881150465807593e-06,
      "loss": 0.4845,
      "step": 285870
    },
    {
      "epoch": 117.02,
      "grad_norm": 1.864724040031433,
      "learning_rate": 1.88099495789436e-06,
      "loss": 0.4651,
      "step": 285880
    },
    {
      "epoch": 117.02,
      "grad_norm": 2.029883861541748,
      "learning_rate": 1.880839452100936e-06,
      "loss": 0.4442,
      "step": 285890
    },
    {
      "epoch": 117.03,
      "grad_norm": 2.0935275554656982,
      "learning_rate": 1.8806839484280335e-06,
      "loss": 0.4765,
      "step": 285900
    },
    {
      "epoch": 117.03,
      "grad_norm": 1.6470648050308228,
      "learning_rate": 1.8805284468763652e-06,
      "loss": 0.4593,
      "step": 285910
    },
    {
      "epoch": 117.04,
      "grad_norm": 1.8354544639587402,
      "learning_rate": 1.8803729474466428e-06,
      "loss": 0.4629,
      "step": 285920
    },
    {
      "epoch": 117.04,
      "grad_norm": 1.8594657182693481,
      "learning_rate": 1.880217450139579e-06,
      "loss": 0.4857,
      "step": 285930
    },
    {
      "epoch": 117.04,
      "grad_norm": 2.2629222869873047,
      "learning_rate": 1.8800619549558848e-06,
      "loss": 0.4734,
      "step": 285940
    },
    {
      "epoch": 117.05,
      "grad_norm": 2.193952798843384,
      "learning_rate": 1.8799064618962748e-06,
      "loss": 0.4611,
      "step": 285950
    },
    {
      "epoch": 117.05,
      "grad_norm": 2.031083583831787,
      "learning_rate": 1.8797509709614605e-06,
      "loss": 0.4691,
      "step": 285960
    },
    {
      "epoch": 117.06,
      "grad_norm": 1.9581489562988281,
      "learning_rate": 1.879595482152154e-06,
      "loss": 0.4849,
      "step": 285970
    },
    {
      "epoch": 117.06,
      "grad_norm": 1.7015305757522583,
      "learning_rate": 1.879439995469067e-06,
      "loss": 0.4505,
      "step": 285980
    },
    {
      "epoch": 117.07,
      "grad_norm": 1.895269751548767,
      "learning_rate": 1.8792845109129123e-06,
      "loss": 0.4629,
      "step": 285990
    },
    {
      "epoch": 117.07,
      "grad_norm": 1.9813392162322998,
      "learning_rate": 1.879129028484402e-06,
      "loss": 0.4529,
      "step": 286000
    },
    {
      "epoch": 117.07,
      "grad_norm": 2.3340394496917725,
      "learning_rate": 1.8789735481842486e-06,
      "loss": 0.4631,
      "step": 286010
    },
    {
      "epoch": 117.08,
      "grad_norm": 2.1131231784820557,
      "learning_rate": 1.878818070013164e-06,
      "loss": 0.4826,
      "step": 286020
    },
    {
      "epoch": 117.08,
      "grad_norm": 1.7273072004318237,
      "learning_rate": 1.8786625939718605e-06,
      "loss": 0.4313,
      "step": 286030
    },
    {
      "epoch": 117.09,
      "grad_norm": 1.6687451601028442,
      "learning_rate": 1.8785071200610506e-06,
      "loss": 0.4536,
      "step": 286040
    },
    {
      "epoch": 117.09,
      "grad_norm": 1.760472059249878,
      "learning_rate": 1.8783516482814466e-06,
      "loss": 0.4643,
      "step": 286050
    },
    {
      "epoch": 117.09,
      "grad_norm": 1.9429270029067993,
      "learning_rate": 1.8781961786337598e-06,
      "loss": 0.4634,
      "step": 286060
    },
    {
      "epoch": 117.1,
      "grad_norm": 2.4809186458587646,
      "learning_rate": 1.878040711118703e-06,
      "loss": 0.4644,
      "step": 286070
    },
    {
      "epoch": 117.1,
      "grad_norm": 2.578230381011963,
      "learning_rate": 1.8778852457369885e-06,
      "loss": 0.4621,
      "step": 286080
    },
    {
      "epoch": 117.11,
      "grad_norm": 2.1882829666137695,
      "learning_rate": 1.8777297824893285e-06,
      "loss": 0.4389,
      "step": 286090
    },
    {
      "epoch": 117.11,
      "grad_norm": 1.9836487770080566,
      "learning_rate": 1.8775743213764345e-06,
      "loss": 0.4471,
      "step": 286100
    },
    {
      "epoch": 117.11,
      "grad_norm": 2.442335844039917,
      "learning_rate": 1.8774188623990193e-06,
      "loss": 0.4805,
      "step": 286110
    },
    {
      "epoch": 117.12,
      "grad_norm": 1.9922568798065186,
      "learning_rate": 1.8772634055577946e-06,
      "loss": 0.4594,
      "step": 286120
    },
    {
      "epoch": 117.12,
      "grad_norm": 1.9934393167495728,
      "learning_rate": 1.8771079508534723e-06,
      "loss": 0.4751,
      "step": 286130
    },
    {
      "epoch": 117.13,
      "grad_norm": 2.0576183795928955,
      "learning_rate": 1.8769524982867658e-06,
      "loss": 0.4624,
      "step": 286140
    },
    {
      "epoch": 117.13,
      "grad_norm": 2.1576642990112305,
      "learning_rate": 1.876797047858386e-06,
      "loss": 0.4774,
      "step": 286150
    },
    {
      "epoch": 117.13,
      "grad_norm": 2.091082811355591,
      "learning_rate": 1.8766415995690456e-06,
      "loss": 0.4764,
      "step": 286160
    },
    {
      "epoch": 117.14,
      "grad_norm": 2.381704092025757,
      "learning_rate": 1.8764861534194562e-06,
      "loss": 0.4873,
      "step": 286170
    },
    {
      "epoch": 117.14,
      "grad_norm": 2.0342531204223633,
      "learning_rate": 1.8763307094103303e-06,
      "loss": 0.4779,
      "step": 286180
    },
    {
      "epoch": 117.15,
      "grad_norm": 1.9292962551116943,
      "learning_rate": 1.87617526754238e-06,
      "loss": 0.4943,
      "step": 286190
    },
    {
      "epoch": 117.15,
      "grad_norm": 1.7617911100387573,
      "learning_rate": 1.8760198278163171e-06,
      "loss": 0.4824,
      "step": 286200
    },
    {
      "epoch": 117.16,
      "grad_norm": 1.4956717491149902,
      "learning_rate": 1.8758643902328533e-06,
      "loss": 0.4674,
      "step": 286210
    },
    {
      "epoch": 117.16,
      "grad_norm": 2.083829879760742,
      "learning_rate": 1.8757089547927013e-06,
      "loss": 0.4647,
      "step": 286220
    },
    {
      "epoch": 117.16,
      "grad_norm": 1.7864742279052734,
      "learning_rate": 1.8755535214965728e-06,
      "loss": 0.4523,
      "step": 286230
    },
    {
      "epoch": 117.17,
      "grad_norm": 1.765805959701538,
      "learning_rate": 1.8753980903451796e-06,
      "loss": 0.4726,
      "step": 286240
    },
    {
      "epoch": 117.17,
      "grad_norm": 1.6394201517105103,
      "learning_rate": 1.8752426613392343e-06,
      "loss": 0.4537,
      "step": 286250
    },
    {
      "epoch": 117.18,
      "grad_norm": 2.3659775257110596,
      "learning_rate": 1.8750872344794483e-06,
      "loss": 0.4617,
      "step": 286260
    },
    {
      "epoch": 117.18,
      "grad_norm": 2.1638176441192627,
      "learning_rate": 1.8749318097665344e-06,
      "loss": 0.4834,
      "step": 286270
    },
    {
      "epoch": 117.18,
      "grad_norm": 2.2435853481292725,
      "learning_rate": 1.8747763872012036e-06,
      "loss": 0.4731,
      "step": 286280
    },
    {
      "epoch": 117.19,
      "grad_norm": 2.5501420497894287,
      "learning_rate": 1.8746209667841681e-06,
      "loss": 0.4564,
      "step": 286290
    },
    {
      "epoch": 117.19,
      "grad_norm": 2.0000438690185547,
      "learning_rate": 1.8744655485161403e-06,
      "loss": 0.453,
      "step": 286300
    },
    {
      "epoch": 117.2,
      "grad_norm": 1.831697940826416,
      "learning_rate": 1.8743101323978316e-06,
      "loss": 0.4729,
      "step": 286310
    },
    {
      "epoch": 117.2,
      "grad_norm": 2.3782873153686523,
      "learning_rate": 1.8741547184299542e-06,
      "loss": 0.464,
      "step": 286320
    },
    {
      "epoch": 117.2,
      "grad_norm": 2.400371789932251,
      "learning_rate": 1.87399930661322e-06,
      "loss": 0.4694,
      "step": 286330
    },
    {
      "epoch": 117.21,
      "grad_norm": 2.2458081245422363,
      "learning_rate": 1.8738438969483414e-06,
      "loss": 0.4579,
      "step": 286340
    },
    {
      "epoch": 117.21,
      "grad_norm": 2.1175975799560547,
      "learning_rate": 1.8736884894360297e-06,
      "loss": 0.4504,
      "step": 286350
    },
    {
      "epoch": 117.22,
      "grad_norm": 2.113063097000122,
      "learning_rate": 1.8735330840769973e-06,
      "loss": 0.4474,
      "step": 286360
    },
    {
      "epoch": 117.22,
      "grad_norm": 1.9169567823410034,
      "learning_rate": 1.8733776808719555e-06,
      "loss": 0.4656,
      "step": 286370
    },
    {
      "epoch": 117.22,
      "grad_norm": 1.977126121520996,
      "learning_rate": 1.8732222798216162e-06,
      "loss": 0.4619,
      "step": 286380
    },
    {
      "epoch": 117.23,
      "grad_norm": 1.9173978567123413,
      "learning_rate": 1.8730668809266918e-06,
      "loss": 0.4694,
      "step": 286390
    },
    {
      "epoch": 117.23,
      "grad_norm": 1.9359935522079468,
      "learning_rate": 1.8729114841878938e-06,
      "loss": 0.4625,
      "step": 286400
    },
    {
      "epoch": 117.24,
      "grad_norm": 2.301356315612793,
      "learning_rate": 1.8727560896059347e-06,
      "loss": 0.4635,
      "step": 286410
    },
    {
      "epoch": 117.24,
      "grad_norm": 1.6435208320617676,
      "learning_rate": 1.872600697181525e-06,
      "loss": 0.4598,
      "step": 286420
    },
    {
      "epoch": 117.25,
      "grad_norm": 1.5114802122116089,
      "learning_rate": 1.8724453069153778e-06,
      "loss": 0.4737,
      "step": 286430
    },
    {
      "epoch": 117.25,
      "grad_norm": 2.5379526615142822,
      "learning_rate": 1.8722899188082038e-06,
      "loss": 0.4757,
      "step": 286440
    },
    {
      "epoch": 117.25,
      "grad_norm": 1.9880778789520264,
      "learning_rate": 1.872134532860716e-06,
      "loss": 0.4622,
      "step": 286450
    },
    {
      "epoch": 117.26,
      "grad_norm": 1.8555128574371338,
      "learning_rate": 1.8719791490736253e-06,
      "loss": 0.4615,
      "step": 286460
    },
    {
      "epoch": 117.26,
      "grad_norm": 1.900902271270752,
      "learning_rate": 1.8718237674476437e-06,
      "loss": 0.4519,
      "step": 286470
    },
    {
      "epoch": 117.27,
      "grad_norm": 1.8352802991867065,
      "learning_rate": 1.8716683879834834e-06,
      "loss": 0.4608,
      "step": 286480
    },
    {
      "epoch": 117.27,
      "grad_norm": 1.692973017692566,
      "learning_rate": 1.8715130106818562e-06,
      "loss": 0.4682,
      "step": 286490
    },
    {
      "epoch": 117.27,
      "grad_norm": 2.06520414352417,
      "learning_rate": 1.871357635543473e-06,
      "loss": 0.4514,
      "step": 286500
    },
    {
      "epoch": 117.28,
      "grad_norm": 2.604933738708496,
      "learning_rate": 1.871202262569045e-06,
      "loss": 0.4874,
      "step": 286510
    },
    {
      "epoch": 117.28,
      "grad_norm": 1.6689012050628662,
      "learning_rate": 1.8710468917592862e-06,
      "loss": 0.4732,
      "step": 286520
    },
    {
      "epoch": 117.29,
      "grad_norm": 2.093886137008667,
      "learning_rate": 1.870891523114907e-06,
      "loss": 0.4592,
      "step": 286530
    },
    {
      "epoch": 117.29,
      "grad_norm": 3.2623634338378906,
      "learning_rate": 1.8707361566366196e-06,
      "loss": 0.4664,
      "step": 286540
    },
    {
      "epoch": 117.29,
      "grad_norm": 2.2111783027648926,
      "learning_rate": 1.870580792325135e-06,
      "loss": 0.437,
      "step": 286550
    },
    {
      "epoch": 117.3,
      "grad_norm": 2.2174932956695557,
      "learning_rate": 1.8704254301811651e-06,
      "loss": 0.4605,
      "step": 286560
    },
    {
      "epoch": 117.3,
      "grad_norm": 1.7038788795471191,
      "learning_rate": 1.8702700702054222e-06,
      "loss": 0.473,
      "step": 286570
    },
    {
      "epoch": 117.31,
      "grad_norm": 1.776686191558838,
      "learning_rate": 1.8701147123986172e-06,
      "loss": 0.4672,
      "step": 286580
    },
    {
      "epoch": 117.31,
      "grad_norm": 1.8853330612182617,
      "learning_rate": 1.8699593567614623e-06,
      "loss": 0.4609,
      "step": 286590
    },
    {
      "epoch": 117.31,
      "grad_norm": 2.0406711101531982,
      "learning_rate": 1.8698040032946685e-06,
      "loss": 0.4768,
      "step": 286600
    },
    {
      "epoch": 117.32,
      "grad_norm": 1.8039464950561523,
      "learning_rate": 1.8696486519989484e-06,
      "loss": 0.4628,
      "step": 286610
    },
    {
      "epoch": 117.32,
      "grad_norm": 2.0792152881622314,
      "learning_rate": 1.8694933028750129e-06,
      "loss": 0.4732,
      "step": 286620
    },
    {
      "epoch": 117.33,
      "grad_norm": 1.8876080513000488,
      "learning_rate": 1.8693379559235744e-06,
      "loss": 0.455,
      "step": 286630
    },
    {
      "epoch": 117.33,
      "grad_norm": 1.819008231163025,
      "learning_rate": 1.8691826111453436e-06,
      "loss": 0.4847,
      "step": 286640
    },
    {
      "epoch": 117.34,
      "grad_norm": 1.9445689916610718,
      "learning_rate": 1.8690272685410326e-06,
      "loss": 0.4461,
      "step": 286650
    },
    {
      "epoch": 117.34,
      "grad_norm": 1.6989961862564087,
      "learning_rate": 1.8688719281113527e-06,
      "loss": 0.4721,
      "step": 286660
    },
    {
      "epoch": 117.34,
      "grad_norm": 1.930219054222107,
      "learning_rate": 1.8687165898570158e-06,
      "loss": 0.4571,
      "step": 286670
    },
    {
      "epoch": 117.35,
      "grad_norm": 1.6626083850860596,
      "learning_rate": 1.8685612537787334e-06,
      "loss": 0.4684,
      "step": 286680
    },
    {
      "epoch": 117.35,
      "grad_norm": 2.3312957286834717,
      "learning_rate": 1.868405919877217e-06,
      "loss": 0.4511,
      "step": 286690
    },
    {
      "epoch": 117.36,
      "grad_norm": 1.8041138648986816,
      "learning_rate": 1.8682505881531777e-06,
      "loss": 0.4534,
      "step": 286700
    },
    {
      "epoch": 117.36,
      "grad_norm": 1.7215181589126587,
      "learning_rate": 1.868095258607328e-06,
      "loss": 0.4719,
      "step": 286710
    },
    {
      "epoch": 117.36,
      "grad_norm": 1.7310080528259277,
      "learning_rate": 1.8679399312403793e-06,
      "loss": 0.461,
      "step": 286720
    },
    {
      "epoch": 117.37,
      "grad_norm": 2.3226304054260254,
      "learning_rate": 1.8677846060530425e-06,
      "loss": 0.4764,
      "step": 286730
    },
    {
      "epoch": 117.37,
      "grad_norm": 2.314919948577881,
      "learning_rate": 1.8676292830460298e-06,
      "loss": 0.4695,
      "step": 286740
    },
    {
      "epoch": 117.38,
      "grad_norm": 1.7376717329025269,
      "learning_rate": 1.8674739622200519e-06,
      "loss": 0.4569,
      "step": 286750
    },
    {
      "epoch": 117.38,
      "grad_norm": 1.8828030824661255,
      "learning_rate": 1.8673186435758211e-06,
      "loss": 0.4685,
      "step": 286760
    },
    {
      "epoch": 117.38,
      "grad_norm": 1.9413542747497559,
      "learning_rate": 1.8671633271140491e-06,
      "loss": 0.4731,
      "step": 286770
    },
    {
      "epoch": 117.39,
      "grad_norm": 1.9532474279403687,
      "learning_rate": 1.8670080128354463e-06,
      "loss": 0.4703,
      "step": 286780
    },
    {
      "epoch": 117.39,
      "grad_norm": 2.0176706314086914,
      "learning_rate": 1.8668527007407248e-06,
      "loss": 0.4719,
      "step": 286790
    },
    {
      "epoch": 117.4,
      "grad_norm": 2.153855562210083,
      "learning_rate": 1.8666973908305955e-06,
      "loss": 0.4524,
      "step": 286800
    },
    {
      "epoch": 117.4,
      "grad_norm": 1.849298119544983,
      "learning_rate": 1.8665420831057708e-06,
      "loss": 0.4753,
      "step": 286810
    },
    {
      "epoch": 117.4,
      "grad_norm": 2.0676512718200684,
      "learning_rate": 1.8663867775669615e-06,
      "loss": 0.4411,
      "step": 286820
    },
    {
      "epoch": 117.41,
      "grad_norm": 1.7385368347167969,
      "learning_rate": 1.8662314742148794e-06,
      "loss": 0.4645,
      "step": 286830
    },
    {
      "epoch": 117.41,
      "grad_norm": 2.614344358444214,
      "learning_rate": 1.8660761730502353e-06,
      "loss": 0.473,
      "step": 286840
    },
    {
      "epoch": 117.42,
      "grad_norm": 2.1907124519348145,
      "learning_rate": 1.8659208740737418e-06,
      "loss": 0.4752,
      "step": 286850
    },
    {
      "epoch": 117.42,
      "grad_norm": 2.1440606117248535,
      "learning_rate": 1.8657655772861088e-06,
      "loss": 0.4693,
      "step": 286860
    },
    {
      "epoch": 117.43,
      "grad_norm": 1.705463171005249,
      "learning_rate": 1.8656102826880485e-06,
      "loss": 0.4569,
      "step": 286870
    },
    {
      "epoch": 117.43,
      "grad_norm": 1.9393142461776733,
      "learning_rate": 1.8654549902802724e-06,
      "loss": 0.4592,
      "step": 286880
    },
    {
      "epoch": 117.43,
      "grad_norm": 2.0277395248413086,
      "learning_rate": 1.8652997000634905e-06,
      "loss": 0.483,
      "step": 286890
    },
    {
      "epoch": 117.44,
      "grad_norm": 2.0359904766082764,
      "learning_rate": 1.8651444120384169e-06,
      "loss": 0.4818,
      "step": 286900
    },
    {
      "epoch": 117.44,
      "grad_norm": 2.112766742706299,
      "learning_rate": 1.8649891262057611e-06,
      "loss": 0.4613,
      "step": 286910
    },
    {
      "epoch": 117.45,
      "grad_norm": 2.653517007827759,
      "learning_rate": 1.8648338425662342e-06,
      "loss": 0.4663,
      "step": 286920
    },
    {
      "epoch": 117.45,
      "grad_norm": 1.6550054550170898,
      "learning_rate": 1.8646785611205482e-06,
      "loss": 0.4634,
      "step": 286930
    },
    {
      "epoch": 117.45,
      "grad_norm": 1.7715468406677246,
      "learning_rate": 1.8645232818694145e-06,
      "loss": 0.4481,
      "step": 286940
    },
    {
      "epoch": 117.46,
      "grad_norm": 1.6739881038665771,
      "learning_rate": 1.8643680048135439e-06,
      "loss": 0.461,
      "step": 286950
    },
    {
      "epoch": 117.46,
      "grad_norm": 2.0232372283935547,
      "learning_rate": 1.8642127299536483e-06,
      "loss": 0.4435,
      "step": 286960
    },
    {
      "epoch": 117.47,
      "grad_norm": 2.0113444328308105,
      "learning_rate": 1.8640574572904385e-06,
      "loss": 0.4815,
      "step": 286970
    },
    {
      "epoch": 117.47,
      "grad_norm": 1.9320486783981323,
      "learning_rate": 1.863902186824626e-06,
      "loss": 0.4419,
      "step": 286980
    },
    {
      "epoch": 117.47,
      "grad_norm": 2.137821912765503,
      "learning_rate": 1.8637469185569225e-06,
      "loss": 0.4588,
      "step": 286990
    },
    {
      "epoch": 117.48,
      "grad_norm": 2.6171038150787354,
      "learning_rate": 1.8635916524880382e-06,
      "loss": 0.4623,
      "step": 287000
    },
    {
      "epoch": 117.48,
      "grad_norm": 2.191483497619629,
      "learning_rate": 1.863436388618685e-06,
      "loss": 0.4697,
      "step": 287010
    },
    {
      "epoch": 117.49,
      "grad_norm": 2.2220921516418457,
      "learning_rate": 1.863281126949574e-06,
      "loss": 0.4704,
      "step": 287020
    },
    {
      "epoch": 117.49,
      "grad_norm": 1.6575511693954468,
      "learning_rate": 1.8631258674814166e-06,
      "loss": 0.4701,
      "step": 287030
    },
    {
      "epoch": 117.49,
      "grad_norm": 1.8490822315216064,
      "learning_rate": 1.8629706102149237e-06,
      "loss": 0.4446,
      "step": 287040
    },
    {
      "epoch": 117.5,
      "grad_norm": 1.8550007343292236,
      "learning_rate": 1.862815355150807e-06,
      "loss": 0.4502,
      "step": 287050
    },
    {
      "epoch": 117.5,
      "grad_norm": 1.761885166168213,
      "learning_rate": 1.8626601022897773e-06,
      "loss": 0.4558,
      "step": 287060
    },
    {
      "epoch": 117.51,
      "grad_norm": 1.8671706914901733,
      "learning_rate": 1.8625048516325454e-06,
      "loss": 0.4721,
      "step": 287070
    },
    {
      "epoch": 117.51,
      "grad_norm": 1.9387791156768799,
      "learning_rate": 1.8623496031798234e-06,
      "loss": 0.4721,
      "step": 287080
    },
    {
      "epoch": 117.52,
      "grad_norm": 1.9589632749557495,
      "learning_rate": 1.8621943569323222e-06,
      "loss": 0.4549,
      "step": 287090
    },
    {
      "epoch": 117.52,
      "grad_norm": 1.903169870376587,
      "learning_rate": 1.862039112890753e-06,
      "loss": 0.4734,
      "step": 287100
    },
    {
      "epoch": 117.52,
      "grad_norm": 2.367734670639038,
      "learning_rate": 1.8618838710558264e-06,
      "loss": 0.4509,
      "step": 287110
    },
    {
      "epoch": 117.53,
      "grad_norm": 1.9703813791275024,
      "learning_rate": 1.861728631428254e-06,
      "loss": 0.4833,
      "step": 287120
    },
    {
      "epoch": 117.53,
      "grad_norm": 2.2301182746887207,
      "learning_rate": 1.8615733940087475e-06,
      "loss": 0.4897,
      "step": 287130
    },
    {
      "epoch": 117.54,
      "grad_norm": 2.622302293777466,
      "learning_rate": 1.861418158798017e-06,
      "loss": 0.4653,
      "step": 287140
    },
    {
      "epoch": 117.54,
      "grad_norm": 1.7851325273513794,
      "learning_rate": 1.8612629257967733e-06,
      "loss": 0.4681,
      "step": 287150
    },
    {
      "epoch": 117.54,
      "grad_norm": 2.369391679763794,
      "learning_rate": 1.861107695005729e-06,
      "loss": 0.4595,
      "step": 287160
    },
    {
      "epoch": 117.55,
      "grad_norm": 2.581796169281006,
      "learning_rate": 1.8609524664255938e-06,
      "loss": 0.4583,
      "step": 287170
    },
    {
      "epoch": 117.55,
      "grad_norm": 2.48622465133667,
      "learning_rate": 1.8607972400570794e-06,
      "loss": 0.4567,
      "step": 287180
    },
    {
      "epoch": 117.56,
      "grad_norm": 1.9302574396133423,
      "learning_rate": 1.8606420159008971e-06,
      "loss": 0.4586,
      "step": 287190
    },
    {
      "epoch": 117.56,
      "grad_norm": 1.5724656581878662,
      "learning_rate": 1.8604867939577574e-06,
      "loss": 0.4648,
      "step": 287200
    },
    {
      "epoch": 117.56,
      "grad_norm": 1.8104991912841797,
      "learning_rate": 1.8603315742283723e-06,
      "loss": 0.4526,
      "step": 287210
    },
    {
      "epoch": 117.57,
      "grad_norm": 2.0651626586914062,
      "learning_rate": 1.8601763567134516e-06,
      "loss": 0.4706,
      "step": 287220
    },
    {
      "epoch": 117.57,
      "grad_norm": 2.0860395431518555,
      "learning_rate": 1.8600211414137065e-06,
      "loss": 0.4673,
      "step": 287230
    },
    {
      "epoch": 117.58,
      "grad_norm": 2.0165812969207764,
      "learning_rate": 1.8598659283298486e-06,
      "loss": 0.4608,
      "step": 287240
    },
    {
      "epoch": 117.58,
      "grad_norm": 2.1020638942718506,
      "learning_rate": 1.8597107174625888e-06,
      "loss": 0.4569,
      "step": 287250
    },
    {
      "epoch": 117.58,
      "grad_norm": 2.565355062484741,
      "learning_rate": 1.8595555088126374e-06,
      "loss": 0.4576,
      "step": 287260
    },
    {
      "epoch": 117.59,
      "grad_norm": 2.06833815574646,
      "learning_rate": 1.8594003023807066e-06,
      "loss": 0.4626,
      "step": 287270
    },
    {
      "epoch": 117.59,
      "grad_norm": 1.832568883895874,
      "learning_rate": 1.8592450981675067e-06,
      "loss": 0.4937,
      "step": 287280
    },
    {
      "epoch": 117.6,
      "grad_norm": 3.480487585067749,
      "learning_rate": 1.8590898961737484e-06,
      "loss": 0.4495,
      "step": 287290
    },
    {
      "epoch": 117.6,
      "grad_norm": 1.708568811416626,
      "learning_rate": 1.8589346964001434e-06,
      "loss": 0.4665,
      "step": 287300
    },
    {
      "epoch": 117.61,
      "grad_norm": 1.985546588897705,
      "learning_rate": 1.858779498847402e-06,
      "loss": 0.4552,
      "step": 287310
    },
    {
      "epoch": 117.61,
      "grad_norm": 2.0642964839935303,
      "learning_rate": 1.8586243035162355e-06,
      "loss": 0.4491,
      "step": 287320
    },
    {
      "epoch": 117.61,
      "grad_norm": 2.4408504962921143,
      "learning_rate": 1.858469110407355e-06,
      "loss": 0.4699,
      "step": 287330
    },
    {
      "epoch": 117.62,
      "grad_norm": 2.249300956726074,
      "learning_rate": 1.8583139195214706e-06,
      "loss": 0.4644,
      "step": 287340
    },
    {
      "epoch": 117.62,
      "grad_norm": 1.9999020099639893,
      "learning_rate": 1.8581587308592942e-06,
      "loss": 0.4458,
      "step": 287350
    },
    {
      "epoch": 117.63,
      "grad_norm": 2.1251912117004395,
      "learning_rate": 1.8580035444215359e-06,
      "loss": 0.4637,
      "step": 287360
    },
    {
      "epoch": 117.63,
      "grad_norm": 1.716715931892395,
      "learning_rate": 1.8578483602089068e-06,
      "loss": 0.4674,
      "step": 287370
    },
    {
      "epoch": 117.63,
      "grad_norm": 1.9937479496002197,
      "learning_rate": 1.857693178222118e-06,
      "loss": 0.4572,
      "step": 287380
    },
    {
      "epoch": 117.64,
      "grad_norm": 2.324843406677246,
      "learning_rate": 1.8575379984618799e-06,
      "loss": 0.4678,
      "step": 287390
    },
    {
      "epoch": 117.64,
      "grad_norm": 3.3660550117492676,
      "learning_rate": 1.857382820928904e-06,
      "loss": 0.4657,
      "step": 287400
    },
    {
      "epoch": 117.65,
      "grad_norm": 2.0370101928710938,
      "learning_rate": 1.8572276456239005e-06,
      "loss": 0.4686,
      "step": 287410
    },
    {
      "epoch": 117.65,
      "grad_norm": 1.8511313199996948,
      "learning_rate": 1.857072472547581e-06,
      "loss": 0.4699,
      "step": 287420
    },
    {
      "epoch": 117.65,
      "grad_norm": 1.7877916097640991,
      "learning_rate": 1.8569173017006554e-06,
      "loss": 0.4427,
      "step": 287430
    },
    {
      "epoch": 117.66,
      "grad_norm": 1.7017567157745361,
      "learning_rate": 1.8567621330838358e-06,
      "loss": 0.4533,
      "step": 287440
    },
    {
      "epoch": 117.66,
      "grad_norm": 1.8590846061706543,
      "learning_rate": 1.8566069666978309e-06,
      "loss": 0.4679,
      "step": 287450
    },
    {
      "epoch": 117.67,
      "grad_norm": 2.1053080558776855,
      "learning_rate": 1.8564518025433534e-06,
      "loss": 0.4572,
      "step": 287460
    },
    {
      "epoch": 117.67,
      "grad_norm": 1.9417099952697754,
      "learning_rate": 1.8562966406211136e-06,
      "loss": 0.449,
      "step": 287470
    },
    {
      "epoch": 117.67,
      "grad_norm": 1.71290123462677,
      "learning_rate": 1.856141480931822e-06,
      "loss": 0.4544,
      "step": 287480
    },
    {
      "epoch": 117.68,
      "grad_norm": 2.4728493690490723,
      "learning_rate": 1.8559863234761902e-06,
      "loss": 0.4675,
      "step": 287490
    },
    {
      "epoch": 117.68,
      "grad_norm": 1.9016376733779907,
      "learning_rate": 1.8558311682549278e-06,
      "loss": 0.4761,
      "step": 287500
    },
    {
      "epoch": 117.69,
      "grad_norm": 1.6844813823699951,
      "learning_rate": 1.8556760152687458e-06,
      "loss": 0.4464,
      "step": 287510
    },
    {
      "epoch": 117.69,
      "grad_norm": 2.0298595428466797,
      "learning_rate": 1.8555208645183555e-06,
      "loss": 0.4741,
      "step": 287520
    },
    {
      "epoch": 117.7,
      "grad_norm": 2.328853130340576,
      "learning_rate": 1.8553657160044672e-06,
      "loss": 0.4548,
      "step": 287530
    },
    {
      "epoch": 117.7,
      "grad_norm": 1.7903271913528442,
      "learning_rate": 1.8552105697277914e-06,
      "loss": 0.4784,
      "step": 287540
    },
    {
      "epoch": 117.7,
      "grad_norm": 1.914235234260559,
      "learning_rate": 1.8550554256890389e-06,
      "loss": 0.4603,
      "step": 287550
    },
    {
      "epoch": 117.71,
      "grad_norm": 2.251058340072632,
      "learning_rate": 1.854900283888921e-06,
      "loss": 0.441,
      "step": 287560
    },
    {
      "epoch": 117.71,
      "grad_norm": 2.0066561698913574,
      "learning_rate": 1.8547451443281485e-06,
      "loss": 0.4789,
      "step": 287570
    },
    {
      "epoch": 117.72,
      "grad_norm": 1.8036847114562988,
      "learning_rate": 1.8545900070074304e-06,
      "loss": 0.4905,
      "step": 287580
    },
    {
      "epoch": 117.72,
      "grad_norm": 1.9864481687545776,
      "learning_rate": 1.854434871927479e-06,
      "loss": 0.4701,
      "step": 287590
    },
    {
      "epoch": 117.72,
      "grad_norm": 7.397694110870361,
      "learning_rate": 1.8542797390890042e-06,
      "loss": 0.4621,
      "step": 287600
    },
    {
      "epoch": 117.73,
      "grad_norm": 2.3682756423950195,
      "learning_rate": 1.854124608492717e-06,
      "loss": 0.4585,
      "step": 287610
    },
    {
      "epoch": 117.73,
      "grad_norm": 2.049722194671631,
      "learning_rate": 1.8539694801393278e-06,
      "loss": 0.4683,
      "step": 287620
    },
    {
      "epoch": 117.74,
      "grad_norm": 1.9793223142623901,
      "learning_rate": 1.8538143540295474e-06,
      "loss": 0.474,
      "step": 287630
    },
    {
      "epoch": 117.74,
      "grad_norm": 1.3378008604049683,
      "learning_rate": 1.8536592301640856e-06,
      "loss": 0.4506,
      "step": 287640
    },
    {
      "epoch": 117.74,
      "grad_norm": 1.917102575302124,
      "learning_rate": 1.8535041085436543e-06,
      "loss": 0.4671,
      "step": 287650
    },
    {
      "epoch": 117.75,
      "grad_norm": 1.8971755504608154,
      "learning_rate": 1.8533489891689639e-06,
      "loss": 0.4514,
      "step": 287660
    },
    {
      "epoch": 117.75,
      "grad_norm": 1.490273356437683,
      "learning_rate": 1.8531938720407242e-06,
      "loss": 0.479,
      "step": 287670
    },
    {
      "epoch": 117.76,
      "grad_norm": 1.7605799436569214,
      "learning_rate": 1.8530387571596462e-06,
      "loss": 0.4406,
      "step": 287680
    },
    {
      "epoch": 117.76,
      "grad_norm": 1.9724843502044678,
      "learning_rate": 1.8528836445264403e-06,
      "loss": 0.4584,
      "step": 287690
    },
    {
      "epoch": 117.77,
      "grad_norm": 2.005959987640381,
      "learning_rate": 1.8527285341418173e-06,
      "loss": 0.4646,
      "step": 287700
    },
    {
      "epoch": 117.77,
      "grad_norm": 1.9308916330337524,
      "learning_rate": 1.852573426006488e-06,
      "loss": 0.47,
      "step": 287710
    },
    {
      "epoch": 117.77,
      "grad_norm": 1.892818808555603,
      "learning_rate": 1.8524183201211621e-06,
      "loss": 0.4548,
      "step": 287720
    },
    {
      "epoch": 117.78,
      "grad_norm": 3.029209613800049,
      "learning_rate": 1.8522632164865508e-06,
      "loss": 0.4426,
      "step": 287730
    },
    {
      "epoch": 117.78,
      "grad_norm": 1.5840940475463867,
      "learning_rate": 1.852108115103364e-06,
      "loss": 0.4732,
      "step": 287740
    },
    {
      "epoch": 117.79,
      "grad_norm": 2.0806918144226074,
      "learning_rate": 1.8519530159723124e-06,
      "loss": 0.4513,
      "step": 287750
    },
    {
      "epoch": 117.79,
      "grad_norm": 1.8314648866653442,
      "learning_rate": 1.8517979190941068e-06,
      "loss": 0.4619,
      "step": 287760
    },
    {
      "epoch": 117.79,
      "grad_norm": 2.5012924671173096,
      "learning_rate": 1.8516428244694575e-06,
      "loss": 0.4486,
      "step": 287770
    },
    {
      "epoch": 117.8,
      "grad_norm": 2.1987035274505615,
      "learning_rate": 1.851487732099075e-06,
      "loss": 0.4807,
      "step": 287780
    },
    {
      "epoch": 117.8,
      "grad_norm": 1.7191928625106812,
      "learning_rate": 1.8513326419836696e-06,
      "loss": 0.4675,
      "step": 287790
    },
    {
      "epoch": 117.81,
      "grad_norm": 2.578171491622925,
      "learning_rate": 1.8511775541239525e-06,
      "loss": 0.4486,
      "step": 287800
    },
    {
      "epoch": 117.81,
      "grad_norm": 1.7781705856323242,
      "learning_rate": 1.851022468520633e-06,
      "loss": 0.4504,
      "step": 287810
    },
    {
      "epoch": 117.81,
      "grad_norm": 1.9381729364395142,
      "learning_rate": 1.850867385174422e-06,
      "loss": 0.4704,
      "step": 287820
    },
    {
      "epoch": 117.82,
      "grad_norm": 2.058694362640381,
      "learning_rate": 1.8507123040860288e-06,
      "loss": 0.4619,
      "step": 287830
    },
    {
      "epoch": 117.82,
      "grad_norm": 1.8164036273956299,
      "learning_rate": 1.850557225256166e-06,
      "loss": 0.452,
      "step": 287840
    },
    {
      "epoch": 117.83,
      "grad_norm": 2.613802671432495,
      "learning_rate": 1.8504021486855434e-06,
      "loss": 0.459,
      "step": 287850
    },
    {
      "epoch": 117.83,
      "grad_norm": 1.8643933534622192,
      "learning_rate": 1.8502470743748703e-06,
      "loss": 0.4616,
      "step": 287860
    },
    {
      "epoch": 117.83,
      "grad_norm": 2.027740716934204,
      "learning_rate": 1.8500920023248578e-06,
      "loss": 0.4644,
      "step": 287870
    },
    {
      "epoch": 117.84,
      "grad_norm": 1.9006320238113403,
      "learning_rate": 1.8499369325362161e-06,
      "loss": 0.4476,
      "step": 287880
    },
    {
      "epoch": 117.84,
      "grad_norm": 1.786208987236023,
      "learning_rate": 1.8497818650096556e-06,
      "loss": 0.4761,
      "step": 287890
    },
    {
      "epoch": 117.85,
      "grad_norm": 1.6064705848693848,
      "learning_rate": 1.8496267997458866e-06,
      "loss": 0.4741,
      "step": 287900
    },
    {
      "epoch": 117.85,
      "grad_norm": 2.0301172733306885,
      "learning_rate": 1.8494717367456196e-06,
      "loss": 0.4723,
      "step": 287910
    },
    {
      "epoch": 117.86,
      "grad_norm": 2.0799617767333984,
      "learning_rate": 1.8493166760095644e-06,
      "loss": 0.4646,
      "step": 287920
    },
    {
      "epoch": 117.86,
      "grad_norm": 1.8839001655578613,
      "learning_rate": 1.8491616175384328e-06,
      "loss": 0.4949,
      "step": 287930
    },
    {
      "epoch": 117.86,
      "grad_norm": 2.151533603668213,
      "learning_rate": 1.849006561332933e-06,
      "loss": 0.4616,
      "step": 287940
    },
    {
      "epoch": 117.87,
      "grad_norm": 2.0344555377960205,
      "learning_rate": 1.8488515073937765e-06,
      "loss": 0.4623,
      "step": 287950
    },
    {
      "epoch": 117.87,
      "grad_norm": 2.253899574279785,
      "learning_rate": 1.8486964557216733e-06,
      "loss": 0.4697,
      "step": 287960
    },
    {
      "epoch": 117.88,
      "grad_norm": 1.8521896600723267,
      "learning_rate": 1.8485414063173335e-06,
      "loss": 0.4737,
      "step": 287970
    },
    {
      "epoch": 117.88,
      "grad_norm": 1.697960376739502,
      "learning_rate": 1.8483863591814676e-06,
      "loss": 0.4591,
      "step": 287980
    },
    {
      "epoch": 117.88,
      "grad_norm": 2.026700735092163,
      "learning_rate": 1.848231314314786e-06,
      "loss": 0.4535,
      "step": 287990
    },
    {
      "epoch": 117.89,
      "grad_norm": 2.1399285793304443,
      "learning_rate": 1.8480762717179986e-06,
      "loss": 0.4622,
      "step": 288000
    },
    {
      "epoch": 117.89,
      "grad_norm": 2.0831058025360107,
      "learning_rate": 1.847921231391816e-06,
      "loss": 0.4729,
      "step": 288010
    },
    {
      "epoch": 117.9,
      "grad_norm": 1.9634888172149658,
      "learning_rate": 1.8477661933369478e-06,
      "loss": 0.4384,
      "step": 288020
    },
    {
      "epoch": 117.9,
      "grad_norm": 1.9152079820632935,
      "learning_rate": 1.8476111575541048e-06,
      "loss": 0.4591,
      "step": 288030
    },
    {
      "epoch": 117.9,
      "grad_norm": 2.3904075622558594,
      "learning_rate": 1.847456124043997e-06,
      "loss": 0.4686,
      "step": 288040
    },
    {
      "epoch": 117.91,
      "grad_norm": 1.730953335762024,
      "learning_rate": 1.8473010928073347e-06,
      "loss": 0.4879,
      "step": 288050
    },
    {
      "epoch": 117.91,
      "grad_norm": 1.93695068359375,
      "learning_rate": 1.8471460638448283e-06,
      "loss": 0.4572,
      "step": 288060
    },
    {
      "epoch": 117.92,
      "grad_norm": 2.386669397354126,
      "learning_rate": 1.8469910371571876e-06,
      "loss": 0.4384,
      "step": 288070
    },
    {
      "epoch": 117.92,
      "grad_norm": 1.6755616664886475,
      "learning_rate": 1.8468360127451226e-06,
      "loss": 0.4664,
      "step": 288080
    },
    {
      "epoch": 117.92,
      "grad_norm": 2.188100814819336,
      "learning_rate": 1.8466809906093433e-06,
      "loss": 0.4387,
      "step": 288090
    },
    {
      "epoch": 117.93,
      "grad_norm": 2.3465592861175537,
      "learning_rate": 1.8465259707505603e-06,
      "loss": 0.4706,
      "step": 288100
    },
    {
      "epoch": 117.93,
      "grad_norm": 2.090588331222534,
      "learning_rate": 1.8463709531694837e-06,
      "loss": 0.483,
      "step": 288110
    },
    {
      "epoch": 117.94,
      "grad_norm": 1.6285406351089478,
      "learning_rate": 1.8462159378668236e-06,
      "loss": 0.4558,
      "step": 288120
    },
    {
      "epoch": 117.94,
      "grad_norm": 2.1823086738586426,
      "learning_rate": 1.8460609248432896e-06,
      "loss": 0.4518,
      "step": 288130
    },
    {
      "epoch": 117.95,
      "grad_norm": 1.7463514804840088,
      "learning_rate": 1.8459059140995923e-06,
      "loss": 0.451,
      "step": 288140
    },
    {
      "epoch": 117.95,
      "grad_norm": 1.7924028635025024,
      "learning_rate": 1.8457509056364418e-06,
      "loss": 0.4651,
      "step": 288150
    },
    {
      "epoch": 117.95,
      "grad_norm": 2.0315582752227783,
      "learning_rate": 1.8455958994545484e-06,
      "loss": 0.4564,
      "step": 288160
    },
    {
      "epoch": 117.96,
      "grad_norm": 3.341107130050659,
      "learning_rate": 1.8454408955546212e-06,
      "loss": 0.4632,
      "step": 288170
    },
    {
      "epoch": 117.96,
      "grad_norm": 2.249620199203491,
      "learning_rate": 1.845285893937371e-06,
      "loss": 0.4512,
      "step": 288180
    },
    {
      "epoch": 117.97,
      "grad_norm": 1.9824578762054443,
      "learning_rate": 1.8451308946035076e-06,
      "loss": 0.46,
      "step": 288190
    },
    {
      "epoch": 117.97,
      "grad_norm": 1.8538943529129028,
      "learning_rate": 1.844975897553741e-06,
      "loss": 0.4739,
      "step": 288200
    },
    {
      "epoch": 117.97,
      "grad_norm": 1.910713791847229,
      "learning_rate": 1.8448209027887811e-06,
      "loss": 0.4703,
      "step": 288210
    },
    {
      "epoch": 117.98,
      "grad_norm": 1.6284351348876953,
      "learning_rate": 1.8446659103093384e-06,
      "loss": 0.4637,
      "step": 288220
    },
    {
      "epoch": 117.98,
      "grad_norm": 1.9220070838928223,
      "learning_rate": 1.8445109201161228e-06,
      "loss": 0.4709,
      "step": 288230
    },
    {
      "epoch": 117.99,
      "grad_norm": 2.219312906265259,
      "learning_rate": 1.844355932209844e-06,
      "loss": 0.4761,
      "step": 288240
    },
    {
      "epoch": 117.99,
      "grad_norm": 2.011298656463623,
      "learning_rate": 1.844200946591212e-06,
      "loss": 0.4728,
      "step": 288250
    },
    {
      "epoch": 117.99,
      "grad_norm": 2.1683950424194336,
      "learning_rate": 1.844045963260937e-06,
      "loss": 0.4533,
      "step": 288260
    },
    {
      "epoch": 118.0,
      "grad_norm": 1.9582200050354004,
      "learning_rate": 1.8438909822197288e-06,
      "loss": 0.4604,
      "step": 288270
    },
    {
      "epoch": 118.0,
      "eval_loss": 0.46343156695365906,
      "eval_runtime": 52.7866,
      "eval_samples_per_second": 65.339,
      "eval_steps_per_second": 8.184,
      "step": 288274
    },
    {
      "epoch": 118.0,
      "grad_norm": 1.9305678606033325,
      "learning_rate": 1.843736003468297e-06,
      "loss": 0.462,
      "step": 288280
    },
    {
      "epoch": 118.01,
      "grad_norm": 2.5012576580047607,
      "learning_rate": 1.8435810270073528e-06,
      "loss": 0.4503,
      "step": 288290
    },
    {
      "epoch": 118.01,
      "grad_norm": 1.650377869606018,
      "learning_rate": 1.8434260528376047e-06,
      "loss": 0.4603,
      "step": 288300
    },
    {
      "epoch": 118.01,
      "grad_norm": 2.1694798469543457,
      "learning_rate": 1.8432710809597628e-06,
      "loss": 0.4642,
      "step": 288310
    },
    {
      "epoch": 118.02,
      "grad_norm": 1.8285518884658813,
      "learning_rate": 1.8431161113745375e-06,
      "loss": 0.4606,
      "step": 288320
    },
    {
      "epoch": 118.02,
      "grad_norm": 1.6089813709259033,
      "learning_rate": 1.8429611440826385e-06,
      "loss": 0.4599,
      "step": 288330
    },
    {
      "epoch": 118.03,
      "grad_norm": 2.0816335678100586,
      "learning_rate": 1.8428061790847757e-06,
      "loss": 0.4497,
      "step": 288340
    },
    {
      "epoch": 118.03,
      "grad_norm": 2.3657007217407227,
      "learning_rate": 1.8426512163816589e-06,
      "loss": 0.4762,
      "step": 288350
    },
    {
      "epoch": 118.04,
      "grad_norm": 1.6851115226745605,
      "learning_rate": 1.8424962559739981e-06,
      "loss": 0.4526,
      "step": 288360
    },
    {
      "epoch": 118.04,
      "grad_norm": 2.5941057205200195,
      "learning_rate": 1.8423412978625033e-06,
      "loss": 0.4661,
      "step": 288370
    },
    {
      "epoch": 118.04,
      "grad_norm": 1.9180415868759155,
      "learning_rate": 1.8421863420478842e-06,
      "loss": 0.4407,
      "step": 288380
    },
    {
      "epoch": 118.05,
      "grad_norm": 2.005216121673584,
      "learning_rate": 1.8420313885308502e-06,
      "loss": 0.4573,
      "step": 288390
    },
    {
      "epoch": 118.05,
      "grad_norm": 1.8446482419967651,
      "learning_rate": 1.8418764373121103e-06,
      "loss": 0.4678,
      "step": 288400
    },
    {
      "epoch": 118.06,
      "grad_norm": 2.0276927947998047,
      "learning_rate": 1.841721488392377e-06,
      "loss": 0.4576,
      "step": 288410
    },
    {
      "epoch": 118.06,
      "grad_norm": 1.9205436706542969,
      "learning_rate": 1.8415665417723582e-06,
      "loss": 0.4725,
      "step": 288420
    },
    {
      "epoch": 118.06,
      "grad_norm": 1.8806787729263306,
      "learning_rate": 1.8414115974527644e-06,
      "loss": 0.4649,
      "step": 288430
    },
    {
      "epoch": 118.07,
      "grad_norm": 2.1397323608398438,
      "learning_rate": 1.8412566554343048e-06,
      "loss": 0.4628,
      "step": 288440
    },
    {
      "epoch": 118.07,
      "grad_norm": 2.2192862033843994,
      "learning_rate": 1.8411017157176891e-06,
      "loss": 0.4778,
      "step": 288450
    },
    {
      "epoch": 118.08,
      "grad_norm": 2.030195951461792,
      "learning_rate": 1.8409467783036275e-06,
      "loss": 0.4602,
      "step": 288460
    },
    {
      "epoch": 118.08,
      "grad_norm": 2.14992094039917,
      "learning_rate": 1.8407918431928298e-06,
      "loss": 0.4447,
      "step": 288470
    },
    {
      "epoch": 118.08,
      "grad_norm": 2.2766737937927246,
      "learning_rate": 1.840636910386005e-06,
      "loss": 0.4853,
      "step": 288480
    },
    {
      "epoch": 118.09,
      "grad_norm": 1.8104838132858276,
      "learning_rate": 1.840481979883864e-06,
      "loss": 0.4619,
      "step": 288490
    },
    {
      "epoch": 118.09,
      "grad_norm": 1.9234529733657837,
      "learning_rate": 1.8403270516871152e-06,
      "loss": 0.4723,
      "step": 288500
    },
    {
      "epoch": 118.1,
      "grad_norm": 2.2853310108184814,
      "learning_rate": 1.8401721257964701e-06,
      "loss": 0.4712,
      "step": 288510
    },
    {
      "epoch": 118.1,
      "grad_norm": 2.0864241123199463,
      "learning_rate": 1.8400172022126363e-06,
      "loss": 0.4545,
      "step": 288520
    },
    {
      "epoch": 118.1,
      "grad_norm": 2.269432544708252,
      "learning_rate": 1.8398622809363247e-06,
      "loss": 0.4644,
      "step": 288530
    },
    {
      "epoch": 118.11,
      "grad_norm": 2.2854461669921875,
      "learning_rate": 1.8397073619682445e-06,
      "loss": 0.4583,
      "step": 288540
    },
    {
      "epoch": 118.11,
      "grad_norm": 1.8751745223999023,
      "learning_rate": 1.8395524453091057e-06,
      "loss": 0.4563,
      "step": 288550
    },
    {
      "epoch": 118.12,
      "grad_norm": 2.2435619831085205,
      "learning_rate": 1.8393975309596179e-06,
      "loss": 0.4589,
      "step": 288560
    },
    {
      "epoch": 118.12,
      "grad_norm": 2.134716272354126,
      "learning_rate": 1.8392426189204903e-06,
      "loss": 0.4589,
      "step": 288570
    },
    {
      "epoch": 118.13,
      "grad_norm": 1.4528121948242188,
      "learning_rate": 1.8390877091924328e-06,
      "loss": 0.4634,
      "step": 288580
    },
    {
      "epoch": 118.13,
      "grad_norm": 1.9822205305099487,
      "learning_rate": 1.838932801776155e-06,
      "loss": 0.4605,
      "step": 288590
    },
    {
      "epoch": 118.13,
      "grad_norm": 1.6475560665130615,
      "learning_rate": 1.8387778966723673e-06,
      "loss": 0.459,
      "step": 288600
    },
    {
      "epoch": 118.14,
      "grad_norm": 1.9725463390350342,
      "learning_rate": 1.8386229938817782e-06,
      "loss": 0.4637,
      "step": 288610
    },
    {
      "epoch": 118.14,
      "grad_norm": 2.020267963409424,
      "learning_rate": 1.8384680934050978e-06,
      "loss": 0.4692,
      "step": 288620
    },
    {
      "epoch": 118.15,
      "grad_norm": 2.4665777683258057,
      "learning_rate": 1.8383131952430355e-06,
      "loss": 0.4579,
      "step": 288630
    },
    {
      "epoch": 118.15,
      "grad_norm": 1.649262547492981,
      "learning_rate": 1.8381582993963012e-06,
      "loss": 0.4576,
      "step": 288640
    },
    {
      "epoch": 118.15,
      "grad_norm": 1.8002519607543945,
      "learning_rate": 1.8380034058656041e-06,
      "loss": 0.4695,
      "step": 288650
    },
    {
      "epoch": 118.16,
      "grad_norm": 1.7646821737289429,
      "learning_rate": 1.8378485146516538e-06,
      "loss": 0.4696,
      "step": 288660
    },
    {
      "epoch": 118.16,
      "grad_norm": 1.8656437397003174,
      "learning_rate": 1.8376936257551598e-06,
      "loss": 0.4595,
      "step": 288670
    },
    {
      "epoch": 118.17,
      "grad_norm": 1.9221094846725464,
      "learning_rate": 1.8375387391768317e-06,
      "loss": 0.4861,
      "step": 288680
    },
    {
      "epoch": 118.17,
      "grad_norm": 1.9727165699005127,
      "learning_rate": 1.8373838549173787e-06,
      "loss": 0.4616,
      "step": 288690
    },
    {
      "epoch": 118.17,
      "grad_norm": 1.6934815645217896,
      "learning_rate": 1.837228972977511e-06,
      "loss": 0.4822,
      "step": 288700
    },
    {
      "epoch": 118.18,
      "grad_norm": 1.6195627450942993,
      "learning_rate": 1.8370740933579374e-06,
      "loss": 0.4691,
      "step": 288710
    },
    {
      "epoch": 118.18,
      "grad_norm": 1.9604672193527222,
      "learning_rate": 1.8369192160593675e-06,
      "loss": 0.4632,
      "step": 288720
    },
    {
      "epoch": 118.19,
      "grad_norm": 2.0557141304016113,
      "learning_rate": 1.8367643410825112e-06,
      "loss": 0.4658,
      "step": 288730
    },
    {
      "epoch": 118.19,
      "grad_norm": 1.9696186780929565,
      "learning_rate": 1.8366094684280781e-06,
      "loss": 0.448,
      "step": 288740
    },
    {
      "epoch": 118.19,
      "grad_norm": 2.3754849433898926,
      "learning_rate": 1.8364545980967769e-06,
      "loss": 0.4793,
      "step": 288750
    },
    {
      "epoch": 118.2,
      "grad_norm": 2.3314483165740967,
      "learning_rate": 1.8362997300893175e-06,
      "loss": 0.4838,
      "step": 288760
    },
    {
      "epoch": 118.2,
      "grad_norm": 1.8533451557159424,
      "learning_rate": 1.8361448644064075e-06,
      "loss": 0.4642,
      "step": 288770
    },
    {
      "epoch": 118.21,
      "grad_norm": 2.012864351272583,
      "learning_rate": 1.8359900010487596e-06,
      "loss": 0.4646,
      "step": 288780
    },
    {
      "epoch": 118.21,
      "grad_norm": 1.6958037614822388,
      "learning_rate": 1.835835140017082e-06,
      "loss": 0.4777,
      "step": 288790
    },
    {
      "epoch": 118.22,
      "grad_norm": 1.6772749423980713,
      "learning_rate": 1.8356802813120834e-06,
      "loss": 0.4372,
      "step": 288800
    },
    {
      "epoch": 118.22,
      "grad_norm": 1.962998390197754,
      "learning_rate": 1.8355254249344732e-06,
      "loss": 0.452,
      "step": 288810
    },
    {
      "epoch": 118.22,
      "grad_norm": 1.8370790481567383,
      "learning_rate": 1.835370570884961e-06,
      "loss": 0.4503,
      "step": 288820
    },
    {
      "epoch": 118.23,
      "grad_norm": 1.6493630409240723,
      "learning_rate": 1.8352157191642564e-06,
      "loss": 0.4593,
      "step": 288830
    },
    {
      "epoch": 118.23,
      "grad_norm": 2.102132558822632,
      "learning_rate": 1.8350608697730686e-06,
      "loss": 0.4709,
      "step": 288840
    },
    {
      "epoch": 118.24,
      "grad_norm": 1.8666139841079712,
      "learning_rate": 1.834906022712107e-06,
      "loss": 0.4687,
      "step": 288850
    },
    {
      "epoch": 118.24,
      "grad_norm": 2.338963031768799,
      "learning_rate": 1.8347511779820807e-06,
      "loss": 0.4569,
      "step": 288860
    },
    {
      "epoch": 118.24,
      "grad_norm": 2.0912578105926514,
      "learning_rate": 1.8345963355836996e-06,
      "loss": 0.479,
      "step": 288870
    },
    {
      "epoch": 118.25,
      "grad_norm": 2.0130608081817627,
      "learning_rate": 1.8344414955176723e-06,
      "loss": 0.4639,
      "step": 288880
    },
    {
      "epoch": 118.25,
      "grad_norm": 1.844659447669983,
      "learning_rate": 1.8342866577847082e-06,
      "loss": 0.4613,
      "step": 288890
    },
    {
      "epoch": 118.26,
      "grad_norm": 1.6800463199615479,
      "learning_rate": 1.834131822385517e-06,
      "loss": 0.4697,
      "step": 288900
    },
    {
      "epoch": 118.26,
      "grad_norm": 1.9776585102081299,
      "learning_rate": 1.8339769893208076e-06,
      "loss": 0.4802,
      "step": 288910
    },
    {
      "epoch": 118.26,
      "grad_norm": 1.997554898262024,
      "learning_rate": 1.8338221585912894e-06,
      "loss": 0.4658,
      "step": 288920
    },
    {
      "epoch": 118.27,
      "grad_norm": 2.095531940460205,
      "learning_rate": 1.8336673301976716e-06,
      "loss": 0.4534,
      "step": 288930
    },
    {
      "epoch": 118.27,
      "grad_norm": 1.9088654518127441,
      "learning_rate": 1.8335125041406637e-06,
      "loss": 0.4754,
      "step": 288940
    },
    {
      "epoch": 118.28,
      "grad_norm": 2.0323574542999268,
      "learning_rate": 1.8333576804209748e-06,
      "loss": 0.4541,
      "step": 288950
    },
    {
      "epoch": 118.28,
      "grad_norm": 1.6120802164077759,
      "learning_rate": 1.8332028590393133e-06,
      "loss": 0.4844,
      "step": 288960
    },
    {
      "epoch": 118.28,
      "grad_norm": 2.4450724124908447,
      "learning_rate": 1.8330480399963898e-06,
      "loss": 0.4716,
      "step": 288970
    },
    {
      "epoch": 118.29,
      "grad_norm": 2.072070837020874,
      "learning_rate": 1.832893223292913e-06,
      "loss": 0.4614,
      "step": 288980
    },
    {
      "epoch": 118.29,
      "grad_norm": 1.88472318649292,
      "learning_rate": 1.8327384089295918e-06,
      "loss": 0.4823,
      "step": 288990
    },
    {
      "epoch": 118.3,
      "grad_norm": 1.8124595880508423,
      "learning_rate": 1.8325835969071357e-06,
      "loss": 0.4479,
      "step": 289000
    },
    {
      "epoch": 118.3,
      "grad_norm": 1.7384483814239502,
      "learning_rate": 1.832428787226254e-06,
      "loss": 0.4596,
      "step": 289010
    },
    {
      "epoch": 118.31,
      "grad_norm": 2.3583099842071533,
      "learning_rate": 1.8322739798876553e-06,
      "loss": 0.4681,
      "step": 289020
    },
    {
      "epoch": 118.31,
      "grad_norm": 1.9989290237426758,
      "learning_rate": 1.8321191748920492e-06,
      "loss": 0.4496,
      "step": 289030
    },
    {
      "epoch": 118.31,
      "grad_norm": 2.3966872692108154,
      "learning_rate": 1.8319643722401444e-06,
      "loss": 0.4838,
      "step": 289040
    },
    {
      "epoch": 118.32,
      "grad_norm": 1.970295786857605,
      "learning_rate": 1.8318095719326503e-06,
      "loss": 0.4728,
      "step": 289050
    },
    {
      "epoch": 118.32,
      "grad_norm": 1.7221916913986206,
      "learning_rate": 1.8316547739702762e-06,
      "loss": 0.4566,
      "step": 289060
    },
    {
      "epoch": 118.33,
      "grad_norm": 2.0609090328216553,
      "learning_rate": 1.8314999783537308e-06,
      "loss": 0.4566,
      "step": 289070
    },
    {
      "epoch": 118.33,
      "grad_norm": 1.8061960935592651,
      "learning_rate": 1.8313451850837233e-06,
      "loss": 0.4599,
      "step": 289080
    },
    {
      "epoch": 118.33,
      "grad_norm": 1.8994371891021729,
      "learning_rate": 1.8311903941609632e-06,
      "loss": 0.4551,
      "step": 289090
    },
    {
      "epoch": 118.34,
      "grad_norm": 2.366783857345581,
      "learning_rate": 1.8310356055861597e-06,
      "loss": 0.4809,
      "step": 289100
    },
    {
      "epoch": 118.34,
      "grad_norm": 2.0552730560302734,
      "learning_rate": 1.830880819360021e-06,
      "loss": 0.4468,
      "step": 289110
    },
    {
      "epoch": 118.35,
      "grad_norm": 1.9849944114685059,
      "learning_rate": 1.8307260354832564e-06,
      "loss": 0.4692,
      "step": 289120
    },
    {
      "epoch": 118.35,
      "grad_norm": 2.7791733741760254,
      "learning_rate": 1.8305712539565756e-06,
      "loss": 0.4823,
      "step": 289130
    },
    {
      "epoch": 118.35,
      "grad_norm": 1.8298522233963013,
      "learning_rate": 1.8304164747806865e-06,
      "loss": 0.4703,
      "step": 289140
    },
    {
      "epoch": 118.36,
      "grad_norm": 2.004770040512085,
      "learning_rate": 1.8302616979562987e-06,
      "loss": 0.4483,
      "step": 289150
    },
    {
      "epoch": 118.36,
      "grad_norm": 2.1066219806671143,
      "learning_rate": 1.8301069234841216e-06,
      "loss": 0.4813,
      "step": 289160
    },
    {
      "epoch": 118.37,
      "grad_norm": 1.9332129955291748,
      "learning_rate": 1.8299521513648641e-06,
      "loss": 0.4484,
      "step": 289170
    },
    {
      "epoch": 118.37,
      "grad_norm": 1.7157881259918213,
      "learning_rate": 1.829797381599235e-06,
      "loss": 0.4681,
      "step": 289180
    },
    {
      "epoch": 118.37,
      "grad_norm": 1.9831981658935547,
      "learning_rate": 1.8296426141879429e-06,
      "loss": 0.4588,
      "step": 289190
    },
    {
      "epoch": 118.38,
      "grad_norm": 2.1490137577056885,
      "learning_rate": 1.8294878491316974e-06,
      "loss": 0.4708,
      "step": 289200
    },
    {
      "epoch": 118.38,
      "grad_norm": 1.992804765701294,
      "learning_rate": 1.829333086431207e-06,
      "loss": 0.4421,
      "step": 289210
    },
    {
      "epoch": 118.39,
      "grad_norm": 1.5672800540924072,
      "learning_rate": 1.8291783260871808e-06,
      "loss": 0.4689,
      "step": 289220
    },
    {
      "epoch": 118.39,
      "grad_norm": 1.9700775146484375,
      "learning_rate": 1.8290235681003284e-06,
      "loss": 0.4524,
      "step": 289230
    },
    {
      "epoch": 118.4,
      "grad_norm": 1.8795108795166016,
      "learning_rate": 1.8288688124713577e-06,
      "loss": 0.4614,
      "step": 289240
    },
    {
      "epoch": 118.4,
      "grad_norm": 1.810710072517395,
      "learning_rate": 1.8287140592009774e-06,
      "loss": 0.4796,
      "step": 289250
    },
    {
      "epoch": 118.4,
      "grad_norm": 2.0972957611083984,
      "learning_rate": 1.8285593082898974e-06,
      "loss": 0.4547,
      "step": 289260
    },
    {
      "epoch": 118.41,
      "grad_norm": 1.8596254587173462,
      "learning_rate": 1.8284045597388262e-06,
      "loss": 0.4564,
      "step": 289270
    },
    {
      "epoch": 118.41,
      "grad_norm": 1.9236600399017334,
      "learning_rate": 1.8282498135484726e-06,
      "loss": 0.4438,
      "step": 289280
    },
    {
      "epoch": 118.42,
      "grad_norm": 2.258924961090088,
      "learning_rate": 1.8280950697195455e-06,
      "loss": 0.4575,
      "step": 289290
    },
    {
      "epoch": 118.42,
      "grad_norm": 2.821467638015747,
      "learning_rate": 1.8279403282527539e-06,
      "loss": 0.4652,
      "step": 289300
    },
    {
      "epoch": 118.42,
      "grad_norm": 1.9931085109710693,
      "learning_rate": 1.8277855891488063e-06,
      "loss": 0.4593,
      "step": 289310
    },
    {
      "epoch": 118.43,
      "grad_norm": 1.825201153755188,
      "learning_rate": 1.8276308524084124e-06,
      "loss": 0.445,
      "step": 289320
    },
    {
      "epoch": 118.43,
      "grad_norm": 1.8368152379989624,
      "learning_rate": 1.8274761180322798e-06,
      "loss": 0.4652,
      "step": 289330
    },
    {
      "epoch": 118.44,
      "grad_norm": 1.7300939559936523,
      "learning_rate": 1.8273213860211169e-06,
      "loss": 0.4655,
      "step": 289340
    },
    {
      "epoch": 118.44,
      "grad_norm": 2.013671875,
      "learning_rate": 1.8271666563756348e-06,
      "loss": 0.4695,
      "step": 289350
    },
    {
      "epoch": 118.44,
      "grad_norm": 2.415520668029785,
      "learning_rate": 1.8270119290965406e-06,
      "loss": 0.4684,
      "step": 289360
    },
    {
      "epoch": 118.45,
      "grad_norm": 1.978000283241272,
      "learning_rate": 1.8268572041845441e-06,
      "loss": 0.4621,
      "step": 289370
    },
    {
      "epoch": 118.45,
      "grad_norm": 1.789322853088379,
      "learning_rate": 1.826702481640353e-06,
      "loss": 0.4698,
      "step": 289380
    },
    {
      "epoch": 118.46,
      "grad_norm": 1.9607125520706177,
      "learning_rate": 1.8265477614646763e-06,
      "loss": 0.4867,
      "step": 289390
    },
    {
      "epoch": 118.46,
      "grad_norm": 2.050215482711792,
      "learning_rate": 1.8263930436582232e-06,
      "loss": 0.4755,
      "step": 289400
    },
    {
      "epoch": 118.47,
      "grad_norm": 2.2187249660491943,
      "learning_rate": 1.826238328221702e-06,
      "loss": 0.4552,
      "step": 289410
    },
    {
      "epoch": 118.47,
      "grad_norm": 1.984148383140564,
      "learning_rate": 1.8260836151558218e-06,
      "loss": 0.4596,
      "step": 289420
    },
    {
      "epoch": 118.47,
      "grad_norm": 1.7516803741455078,
      "learning_rate": 1.8259289044612912e-06,
      "loss": 0.4654,
      "step": 289430
    },
    {
      "epoch": 118.48,
      "grad_norm": 2.241737127304077,
      "learning_rate": 1.825774196138819e-06,
      "loss": 0.486,
      "step": 289440
    },
    {
      "epoch": 118.48,
      "grad_norm": 2.00034236907959,
      "learning_rate": 1.8256194901891131e-06,
      "loss": 0.4597,
      "step": 289450
    },
    {
      "epoch": 118.49,
      "grad_norm": 1.8103073835372925,
      "learning_rate": 1.8254647866128837e-06,
      "loss": 0.4597,
      "step": 289460
    },
    {
      "epoch": 118.49,
      "grad_norm": 2.0521724224090576,
      "learning_rate": 1.8253100854108383e-06,
      "loss": 0.4573,
      "step": 289470
    },
    {
      "epoch": 118.49,
      "grad_norm": 1.888200283050537,
      "learning_rate": 1.8251553865836854e-06,
      "loss": 0.4638,
      "step": 289480
    },
    {
      "epoch": 118.5,
      "grad_norm": 1.9309412240982056,
      "learning_rate": 1.8250006901321345e-06,
      "loss": 0.4686,
      "step": 289490
    },
    {
      "epoch": 118.5,
      "grad_norm": 1.738764762878418,
      "learning_rate": 1.8248459960568936e-06,
      "loss": 0.4758,
      "step": 289500
    },
    {
      "epoch": 118.51,
      "grad_norm": 2.299832344055176,
      "learning_rate": 1.8246913043586715e-06,
      "loss": 0.4742,
      "step": 289510
    },
    {
      "epoch": 118.51,
      "grad_norm": 1.6002559661865234,
      "learning_rate": 1.8245366150381773e-06,
      "loss": 0.4672,
      "step": 289520
    },
    {
      "epoch": 118.51,
      "grad_norm": 1.561826467514038,
      "learning_rate": 1.8243819280961185e-06,
      "loss": 0.4625,
      "step": 289530
    },
    {
      "epoch": 118.52,
      "grad_norm": 2.0837416648864746,
      "learning_rate": 1.8242272435332048e-06,
      "loss": 0.4529,
      "step": 289540
    },
    {
      "epoch": 118.52,
      "grad_norm": 1.9772688150405884,
      "learning_rate": 1.8240725613501445e-06,
      "loss": 0.4664,
      "step": 289550
    },
    {
      "epoch": 118.53,
      "grad_norm": 2.7466814517974854,
      "learning_rate": 1.823917881547646e-06,
      "loss": 0.4673,
      "step": 289560
    },
    {
      "epoch": 118.53,
      "grad_norm": 1.7388637065887451,
      "learning_rate": 1.8237632041264181e-06,
      "loss": 0.4673,
      "step": 289570
    },
    {
      "epoch": 118.53,
      "grad_norm": 1.992824912071228,
      "learning_rate": 1.8236085290871691e-06,
      "loss": 0.445,
      "step": 289580
    },
    {
      "epoch": 118.54,
      "grad_norm": 1.8503177165985107,
      "learning_rate": 1.823453856430608e-06,
      "loss": 0.4645,
      "step": 289590
    },
    {
      "epoch": 118.54,
      "grad_norm": 1.9236377477645874,
      "learning_rate": 1.8232991861574424e-06,
      "loss": 0.4728,
      "step": 289600
    },
    {
      "epoch": 118.55,
      "grad_norm": 1.7736163139343262,
      "learning_rate": 1.8231445182683816e-06,
      "loss": 0.4455,
      "step": 289610
    },
    {
      "epoch": 118.55,
      "grad_norm": 2.064932346343994,
      "learning_rate": 1.8229898527641338e-06,
      "loss": 0.4532,
      "step": 289620
    },
    {
      "epoch": 118.56,
      "grad_norm": 1.713407278060913,
      "learning_rate": 1.8228351896454076e-06,
      "loss": 0.4618,
      "step": 289630
    },
    {
      "epoch": 118.56,
      "grad_norm": 1.9013848304748535,
      "learning_rate": 1.8226805289129114e-06,
      "loss": 0.4665,
      "step": 289640
    },
    {
      "epoch": 118.56,
      "grad_norm": 1.791128158569336,
      "learning_rate": 1.8225258705673538e-06,
      "loss": 0.4582,
      "step": 289650
    },
    {
      "epoch": 118.57,
      "grad_norm": 2.097146987915039,
      "learning_rate": 1.8223712146094433e-06,
      "loss": 0.4819,
      "step": 289660
    },
    {
      "epoch": 118.57,
      "grad_norm": 1.8341752290725708,
      "learning_rate": 1.8222165610398882e-06,
      "loss": 0.4739,
      "step": 289670
    },
    {
      "epoch": 118.58,
      "grad_norm": 1.7585512399673462,
      "learning_rate": 1.8220619098593974e-06,
      "loss": 0.451,
      "step": 289680
    },
    {
      "epoch": 118.58,
      "grad_norm": 2.2192773818969727,
      "learning_rate": 1.8219072610686785e-06,
      "loss": 0.4811,
      "step": 289690
    },
    {
      "epoch": 118.58,
      "grad_norm": 1.8497326374053955,
      "learning_rate": 1.8217526146684404e-06,
      "loss": 0.4518,
      "step": 289700
    },
    {
      "epoch": 118.59,
      "grad_norm": 1.7666999101638794,
      "learning_rate": 1.8215979706593915e-06,
      "loss": 0.4484,
      "step": 289710
    },
    {
      "epoch": 118.59,
      "grad_norm": 1.8880668878555298,
      "learning_rate": 1.8214433290422394e-06,
      "loss": 0.4597,
      "step": 289720
    },
    {
      "epoch": 118.6,
      "grad_norm": 1.5466068983078003,
      "learning_rate": 1.8212886898176945e-06,
      "loss": 0.4607,
      "step": 289730
    },
    {
      "epoch": 118.6,
      "grad_norm": 1.707501769065857,
      "learning_rate": 1.8211340529864633e-06,
      "loss": 0.461,
      "step": 289740
    },
    {
      "epoch": 118.6,
      "grad_norm": 2.1121809482574463,
      "learning_rate": 1.820979418549255e-06,
      "loss": 0.468,
      "step": 289750
    },
    {
      "epoch": 118.61,
      "grad_norm": 1.724889874458313,
      "learning_rate": 1.820824786506778e-06,
      "loss": 0.4738,
      "step": 289760
    },
    {
      "epoch": 118.61,
      "grad_norm": 1.675010085105896,
      "learning_rate": 1.8206701568597403e-06,
      "loss": 0.4474,
      "step": 289770
    },
    {
      "epoch": 118.62,
      "grad_norm": 1.8971104621887207,
      "learning_rate": 1.8205155296088501e-06,
      "loss": 0.4694,
      "step": 289780
    },
    {
      "epoch": 118.62,
      "grad_norm": 1.7056361436843872,
      "learning_rate": 1.820360904754816e-06,
      "loss": 0.4687,
      "step": 289790
    },
    {
      "epoch": 118.62,
      "grad_norm": 1.770524501800537,
      "learning_rate": 1.8202062822983465e-06,
      "loss": 0.469,
      "step": 289800
    },
    {
      "epoch": 118.63,
      "grad_norm": 2.3614981174468994,
      "learning_rate": 1.8200516622401497e-06,
      "loss": 0.457,
      "step": 289810
    },
    {
      "epoch": 118.63,
      "grad_norm": 1.7004027366638184,
      "learning_rate": 1.8198970445809342e-06,
      "loss": 0.4782,
      "step": 289820
    },
    {
      "epoch": 118.64,
      "grad_norm": 1.9935871362686157,
      "learning_rate": 1.8197424293214075e-06,
      "loss": 0.4682,
      "step": 289830
    },
    {
      "epoch": 118.64,
      "grad_norm": 2.091191291809082,
      "learning_rate": 1.8195878164622784e-06,
      "loss": 0.474,
      "step": 289840
    },
    {
      "epoch": 118.65,
      "grad_norm": 1.625766396522522,
      "learning_rate": 1.8194332060042551e-06,
      "loss": 0.4559,
      "step": 289850
    },
    {
      "epoch": 118.65,
      "grad_norm": 1.6750092506408691,
      "learning_rate": 1.8192785979480457e-06,
      "loss": 0.453,
      "step": 289860
    },
    {
      "epoch": 118.65,
      "grad_norm": 2.034031391143799,
      "learning_rate": 1.8191239922943586e-06,
      "loss": 0.4535,
      "step": 289870
    },
    {
      "epoch": 118.66,
      "grad_norm": 1.893689513206482,
      "learning_rate": 1.8189693890439022e-06,
      "loss": 0.4495,
      "step": 289880
    },
    {
      "epoch": 118.66,
      "grad_norm": 2.788576602935791,
      "learning_rate": 1.8188147881973844e-06,
      "loss": 0.4649,
      "step": 289890
    },
    {
      "epoch": 118.67,
      "grad_norm": 2.0381600856781006,
      "learning_rate": 1.8186601897555127e-06,
      "loss": 0.4832,
      "step": 289900
    },
    {
      "epoch": 118.67,
      "grad_norm": 1.8832453489303589,
      "learning_rate": 1.8185055937189972e-06,
      "loss": 0.4527,
      "step": 289910
    },
    {
      "epoch": 118.67,
      "grad_norm": 2.031122922897339,
      "learning_rate": 1.8183510000885445e-06,
      "loss": 0.4613,
      "step": 289920
    },
    {
      "epoch": 118.68,
      "grad_norm": 1.9894887208938599,
      "learning_rate": 1.8181964088648633e-06,
      "loss": 0.4639,
      "step": 289930
    },
    {
      "epoch": 118.68,
      "grad_norm": 1.9384363889694214,
      "learning_rate": 1.818041820048662e-06,
      "loss": 0.454,
      "step": 289940
    },
    {
      "epoch": 118.69,
      "grad_norm": 2.5513217449188232,
      "learning_rate": 1.8178872336406489e-06,
      "loss": 0.462,
      "step": 289950
    },
    {
      "epoch": 118.69,
      "grad_norm": 1.8432910442352295,
      "learning_rate": 1.8177326496415308e-06,
      "loss": 0.4712,
      "step": 289960
    },
    {
      "epoch": 118.69,
      "grad_norm": 1.9338783025741577,
      "learning_rate": 1.8175780680520172e-06,
      "loss": 0.4559,
      "step": 289970
    },
    {
      "epoch": 118.7,
      "grad_norm": 1.935902714729309,
      "learning_rate": 1.8174234888728154e-06,
      "loss": 0.4653,
      "step": 289980
    },
    {
      "epoch": 118.7,
      "grad_norm": 1.9287984371185303,
      "learning_rate": 1.8172689121046338e-06,
      "loss": 0.4668,
      "step": 289990
    },
    {
      "epoch": 118.71,
      "grad_norm": 2.8472626209259033,
      "learning_rate": 1.8171143377481808e-06,
      "loss": 0.4748,
      "step": 290000
    },
    {
      "epoch": 118.71,
      "grad_norm": 2.213273525238037,
      "learning_rate": 1.816959765804164e-06,
      "loss": 0.4619,
      "step": 290010
    },
    {
      "epoch": 118.71,
      "grad_norm": 1.5184952020645142,
      "learning_rate": 1.816805196273292e-06,
      "loss": 0.4663,
      "step": 290020
    },
    {
      "epoch": 118.72,
      "grad_norm": 1.9784302711486816,
      "learning_rate": 1.8166506291562723e-06,
      "loss": 0.4673,
      "step": 290030
    },
    {
      "epoch": 118.72,
      "grad_norm": 1.6760660409927368,
      "learning_rate": 1.8164960644538136e-06,
      "loss": 0.4621,
      "step": 290040
    },
    {
      "epoch": 118.73,
      "grad_norm": 1.9424817562103271,
      "learning_rate": 1.816341502166623e-06,
      "loss": 0.4427,
      "step": 290050
    },
    {
      "epoch": 118.73,
      "grad_norm": 2.2020344734191895,
      "learning_rate": 1.8161869422954092e-06,
      "loss": 0.4455,
      "step": 290060
    },
    {
      "epoch": 118.74,
      "grad_norm": 1.9047306776046753,
      "learning_rate": 1.81603238484088e-06,
      "loss": 0.4847,
      "step": 290070
    },
    {
      "epoch": 118.74,
      "grad_norm": 2.045727014541626,
      "learning_rate": 1.8158778298037434e-06,
      "loss": 0.4615,
      "step": 290080
    },
    {
      "epoch": 118.74,
      "grad_norm": 2.0771243572235107,
      "learning_rate": 1.815723277184707e-06,
      "loss": 0.4875,
      "step": 290090
    },
    {
      "epoch": 118.75,
      "grad_norm": 1.9766831398010254,
      "learning_rate": 1.81556872698448e-06,
      "loss": 0.4792,
      "step": 290100
    },
    {
      "epoch": 118.75,
      "grad_norm": 2.076425075531006,
      "learning_rate": 1.8154141792037695e-06,
      "loss": 0.4747,
      "step": 290110
    },
    {
      "epoch": 118.76,
      "grad_norm": 1.6454370021820068,
      "learning_rate": 1.8152596338432834e-06,
      "loss": 0.4777,
      "step": 290120
    },
    {
      "epoch": 118.76,
      "grad_norm": 1.9093798398971558,
      "learning_rate": 1.8151050909037298e-06,
      "loss": 0.4497,
      "step": 290130
    },
    {
      "epoch": 118.76,
      "grad_norm": 1.9046058654785156,
      "learning_rate": 1.8149505503858167e-06,
      "loss": 0.4701,
      "step": 290140
    },
    {
      "epoch": 118.77,
      "grad_norm": 1.7236019372940063,
      "learning_rate": 1.814796012290252e-06,
      "loss": 0.4608,
      "step": 290150
    },
    {
      "epoch": 118.77,
      "grad_norm": 2.0485923290252686,
      "learning_rate": 1.8146414766177437e-06,
      "loss": 0.4533,
      "step": 290160
    },
    {
      "epoch": 118.78,
      "grad_norm": 1.9669677019119263,
      "learning_rate": 1.8144869433689997e-06,
      "loss": 0.4492,
      "step": 290170
    },
    {
      "epoch": 118.78,
      "grad_norm": 1.7399482727050781,
      "learning_rate": 1.8143324125447281e-06,
      "loss": 0.467,
      "step": 290180
    },
    {
      "epoch": 118.78,
      "grad_norm": 1.7775561809539795,
      "learning_rate": 1.8141778841456358e-06,
      "loss": 0.4757,
      "step": 290190
    },
    {
      "epoch": 118.79,
      "grad_norm": 2.0619752407073975,
      "learning_rate": 1.8140233581724317e-06,
      "loss": 0.4804,
      "step": 290200
    },
    {
      "epoch": 118.79,
      "grad_norm": 2.1443545818328857,
      "learning_rate": 1.8138688346258232e-06,
      "loss": 0.4694,
      "step": 290210
    },
    {
      "epoch": 118.8,
      "grad_norm": 1.8600635528564453,
      "learning_rate": 1.8137143135065183e-06,
      "loss": 0.4541,
      "step": 290220
    },
    {
      "epoch": 118.8,
      "grad_norm": 2.256983518600464,
      "learning_rate": 1.8135597948152247e-06,
      "loss": 0.4635,
      "step": 290230
    },
    {
      "epoch": 118.8,
      "grad_norm": 1.8282513618469238,
      "learning_rate": 1.8134052785526504e-06,
      "loss": 0.4612,
      "step": 290240
    },
    {
      "epoch": 118.81,
      "grad_norm": 2.1412551403045654,
      "learning_rate": 1.8132507647195032e-06,
      "loss": 0.4634,
      "step": 290250
    },
    {
      "epoch": 118.81,
      "grad_norm": 1.8678672313690186,
      "learning_rate": 1.8130962533164908e-06,
      "loss": 0.4854,
      "step": 290260
    },
    {
      "epoch": 118.82,
      "grad_norm": 1.8527193069458008,
      "learning_rate": 1.812941744344321e-06,
      "loss": 0.4755,
      "step": 290270
    },
    {
      "epoch": 118.82,
      "grad_norm": 1.7358345985412598,
      "learning_rate": 1.8127872378037007e-06,
      "loss": 0.4521,
      "step": 290280
    },
    {
      "epoch": 118.83,
      "grad_norm": 1.8744590282440186,
      "learning_rate": 1.8126327336953394e-06,
      "loss": 0.4554,
      "step": 290290
    },
    {
      "epoch": 118.83,
      "grad_norm": 2.100950241088867,
      "learning_rate": 1.812478232019944e-06,
      "loss": 0.4527,
      "step": 290300
    },
    {
      "epoch": 118.83,
      "grad_norm": 2.4720983505249023,
      "learning_rate": 1.812323732778223e-06,
      "loss": 0.4703,
      "step": 290310
    },
    {
      "epoch": 118.84,
      "grad_norm": 1.9711047410964966,
      "learning_rate": 1.8121692359708828e-06,
      "loss": 0.4531,
      "step": 290320
    },
    {
      "epoch": 118.84,
      "grad_norm": 2.0219430923461914,
      "learning_rate": 1.8120147415986318e-06,
      "loss": 0.4804,
      "step": 290330
    },
    {
      "epoch": 118.85,
      "grad_norm": 1.915237307548523,
      "learning_rate": 1.8118602496621776e-06,
      "loss": 0.4518,
      "step": 290340
    },
    {
      "epoch": 118.85,
      "grad_norm": 1.9775606393814087,
      "learning_rate": 1.811705760162228e-06,
      "loss": 0.4604,
      "step": 290350
    },
    {
      "epoch": 118.85,
      "grad_norm": 2.095754861831665,
      "learning_rate": 1.8115512730994907e-06,
      "loss": 0.4478,
      "step": 290360
    },
    {
      "epoch": 118.86,
      "grad_norm": 1.811706781387329,
      "learning_rate": 1.8113967884746734e-06,
      "loss": 0.4553,
      "step": 290370
    },
    {
      "epoch": 118.86,
      "grad_norm": 1.863632321357727,
      "learning_rate": 1.8112423062884834e-06,
      "loss": 0.4628,
      "step": 290380
    },
    {
      "epoch": 118.87,
      "grad_norm": 2.008502244949341,
      "learning_rate": 1.811087826541629e-06,
      "loss": 0.462,
      "step": 290390
    },
    {
      "epoch": 118.87,
      "grad_norm": 2.43649959564209,
      "learning_rate": 1.810933349234818e-06,
      "loss": 0.462,
      "step": 290400
    },
    {
      "epoch": 118.87,
      "grad_norm": 3.001000165939331,
      "learning_rate": 1.810778874368757e-06,
      "loss": 0.4788,
      "step": 290410
    },
    {
      "epoch": 118.88,
      "grad_norm": 2.4924728870391846,
      "learning_rate": 1.810624401944154e-06,
      "loss": 0.4736,
      "step": 290420
    },
    {
      "epoch": 118.88,
      "grad_norm": 2.5871224403381348,
      "learning_rate": 1.8104699319617171e-06,
      "loss": 0.4576,
      "step": 290430
    },
    {
      "epoch": 118.89,
      "grad_norm": 2.6561222076416016,
      "learning_rate": 1.8103154644221534e-06,
      "loss": 0.4658,
      "step": 290440
    },
    {
      "epoch": 118.89,
      "grad_norm": 1.8241859674453735,
      "learning_rate": 1.8101609993261707e-06,
      "loss": 0.4593,
      "step": 290450
    },
    {
      "epoch": 118.89,
      "grad_norm": 1.8578087091445923,
      "learning_rate": 1.8100065366744767e-06,
      "loss": 0.4556,
      "step": 290460
    },
    {
      "epoch": 118.9,
      "grad_norm": 2.1672909259796143,
      "learning_rate": 1.8098520764677784e-06,
      "loss": 0.4559,
      "step": 290470
    },
    {
      "epoch": 118.9,
      "grad_norm": 1.8324079513549805,
      "learning_rate": 1.8096976187067842e-06,
      "loss": 0.4729,
      "step": 290480
    },
    {
      "epoch": 118.91,
      "grad_norm": 1.9921748638153076,
      "learning_rate": 1.8095431633922012e-06,
      "loss": 0.4787,
      "step": 290490
    },
    {
      "epoch": 118.91,
      "grad_norm": 2.225238800048828,
      "learning_rate": 1.809388710524737e-06,
      "loss": 0.4663,
      "step": 290500
    },
    {
      "epoch": 118.92,
      "grad_norm": 1.7958457469940186,
      "learning_rate": 1.8092342601050992e-06,
      "loss": 0.4644,
      "step": 290510
    },
    {
      "epoch": 118.92,
      "grad_norm": 1.8256584405899048,
      "learning_rate": 1.8090798121339951e-06,
      "loss": 0.4705,
      "step": 290520
    },
    {
      "epoch": 118.92,
      "grad_norm": 2.0160434246063232,
      "learning_rate": 1.8089253666121328e-06,
      "loss": 0.4543,
      "step": 290530
    },
    {
      "epoch": 118.93,
      "grad_norm": 1.8986186981201172,
      "learning_rate": 1.8087709235402192e-06,
      "loss": 0.4613,
      "step": 290540
    },
    {
      "epoch": 118.93,
      "grad_norm": 2.0626144409179688,
      "learning_rate": 1.8086164829189613e-06,
      "loss": 0.4536,
      "step": 290550
    },
    {
      "epoch": 118.94,
      "grad_norm": 2.0119175910949707,
      "learning_rate": 1.8084620447490675e-06,
      "loss": 0.4599,
      "step": 290560
    },
    {
      "epoch": 118.94,
      "grad_norm": 1.7901458740234375,
      "learning_rate": 1.808307609031245e-06,
      "loss": 0.4731,
      "step": 290570
    },
    {
      "epoch": 118.94,
      "grad_norm": 1.9120947122573853,
      "learning_rate": 1.8081531757662012e-06,
      "loss": 0.4705,
      "step": 290580
    },
    {
      "epoch": 118.95,
      "grad_norm": 2.1703531742095947,
      "learning_rate": 1.8079987449546434e-06,
      "loss": 0.4457,
      "step": 290590
    },
    {
      "epoch": 118.95,
      "grad_norm": 1.7146968841552734,
      "learning_rate": 1.807844316597279e-06,
      "loss": 0.4509,
      "step": 290600
    },
    {
      "epoch": 118.96,
      "grad_norm": 1.7884256839752197,
      "learning_rate": 1.8076898906948157e-06,
      "loss": 0.4718,
      "step": 290610
    },
    {
      "epoch": 118.96,
      "grad_norm": 2.016793966293335,
      "learning_rate": 1.8075354672479611e-06,
      "loss": 0.4468,
      "step": 290620
    },
    {
      "epoch": 118.96,
      "grad_norm": 1.8077671527862549,
      "learning_rate": 1.807381046257422e-06,
      "loss": 0.4552,
      "step": 290630
    },
    {
      "epoch": 118.97,
      "grad_norm": 1.87704598903656,
      "learning_rate": 1.8072266277239057e-06,
      "loss": 0.4666,
      "step": 290640
    },
    {
      "epoch": 118.97,
      "grad_norm": 1.8629075288772583,
      "learning_rate": 1.8070722116481199e-06,
      "loss": 0.454,
      "step": 290650
    },
    {
      "epoch": 118.98,
      "grad_norm": 2.1842756271362305,
      "learning_rate": 1.806917798030771e-06,
      "loss": 0.4533,
      "step": 290660
    },
    {
      "epoch": 118.98,
      "grad_norm": 1.9236533641815186,
      "learning_rate": 1.8067633868725688e-06,
      "loss": 0.4602,
      "step": 290670
    },
    {
      "epoch": 118.98,
      "grad_norm": 2.2499887943267822,
      "learning_rate": 1.8066089781742185e-06,
      "loss": 0.4675,
      "step": 290680
    },
    {
      "epoch": 118.99,
      "grad_norm": 2.0645482540130615,
      "learning_rate": 1.8064545719364284e-06,
      "loss": 0.4638,
      "step": 290690
    },
    {
      "epoch": 118.99,
      "grad_norm": 2.1657848358154297,
      "learning_rate": 1.8063001681599052e-06,
      "loss": 0.4514,
      "step": 290700
    },
    {
      "epoch": 119.0,
      "grad_norm": 1.9531617164611816,
      "learning_rate": 1.8061457668453562e-06,
      "loss": 0.4722,
      "step": 290710
    },
    {
      "epoch": 119.0,
      "eval_loss": 0.46370527148246765,
      "eval_runtime": 60.2813,
      "eval_samples_per_second": 57.215,
      "eval_steps_per_second": 7.166,
      "step": 290717
    },
    {
      "epoch": 119.0,
      "grad_norm": 1.9374051094055176,
      "learning_rate": 1.8059913679934893e-06,
      "loss": 0.456,
      "step": 290720
    },
    {
      "epoch": 119.01,
      "grad_norm": 2.112835168838501,
      "learning_rate": 1.8058369716050114e-06,
      "loss": 0.4587,
      "step": 290730
    },
    {
      "epoch": 119.01,
      "grad_norm": 2.227628231048584,
      "learning_rate": 1.8056825776806297e-06,
      "loss": 0.466,
      "step": 290740
    },
    {
      "epoch": 119.01,
      "grad_norm": 1.8879470825195312,
      "learning_rate": 1.8055281862210513e-06,
      "loss": 0.4356,
      "step": 290750
    },
    {
      "epoch": 119.02,
      "grad_norm": 1.779397964477539,
      "learning_rate": 1.8053737972269843e-06,
      "loss": 0.4605,
      "step": 290760
    },
    {
      "epoch": 119.02,
      "grad_norm": 1.9043099880218506,
      "learning_rate": 1.8052194106991353e-06,
      "loss": 0.4727,
      "step": 290770
    },
    {
      "epoch": 119.03,
      "grad_norm": 1.7885931730270386,
      "learning_rate": 1.8050650266382108e-06,
      "loss": 0.4686,
      "step": 290780
    },
    {
      "epoch": 119.03,
      "grad_norm": 5.215573787689209,
      "learning_rate": 1.8049106450449193e-06,
      "loss": 0.4693,
      "step": 290790
    },
    {
      "epoch": 119.03,
      "grad_norm": 2.2246599197387695,
      "learning_rate": 1.804756265919967e-06,
      "loss": 0.4544,
      "step": 290800
    },
    {
      "epoch": 119.04,
      "grad_norm": 1.9476282596588135,
      "learning_rate": 1.8046018892640617e-06,
      "loss": 0.477,
      "step": 290810
    },
    {
      "epoch": 119.04,
      "grad_norm": 1.6444275379180908,
      "learning_rate": 1.8044475150779106e-06,
      "loss": 0.4669,
      "step": 290820
    },
    {
      "epoch": 119.05,
      "grad_norm": 2.315225839614868,
      "learning_rate": 1.8042931433622203e-06,
      "loss": 0.453,
      "step": 290830
    },
    {
      "epoch": 119.05,
      "grad_norm": 1.5558826923370361,
      "learning_rate": 1.8041387741176989e-06,
      "loss": 0.4484,
      "step": 290840
    },
    {
      "epoch": 119.05,
      "grad_norm": 2.146280527114868,
      "learning_rate": 1.8039844073450517e-06,
      "loss": 0.4437,
      "step": 290850
    },
    {
      "epoch": 119.06,
      "grad_norm": 1.5112581253051758,
      "learning_rate": 1.8038300430449878e-06,
      "loss": 0.465,
      "step": 290860
    },
    {
      "epoch": 119.06,
      "grad_norm": 6.86295223236084,
      "learning_rate": 1.8036756812182139e-06,
      "loss": 0.4507,
      "step": 290870
    },
    {
      "epoch": 119.07,
      "grad_norm": 1.8052802085876465,
      "learning_rate": 1.8035213218654363e-06,
      "loss": 0.454,
      "step": 290880
    },
    {
      "epoch": 119.07,
      "grad_norm": 1.9119524955749512,
      "learning_rate": 1.8033669649873636e-06,
      "loss": 0.4649,
      "step": 290890
    },
    {
      "epoch": 119.07,
      "grad_norm": 1.8649495840072632,
      "learning_rate": 1.8032126105847013e-06,
      "loss": 0.4548,
      "step": 290900
    },
    {
      "epoch": 119.08,
      "grad_norm": 2.2975575923919678,
      "learning_rate": 1.803058258658157e-06,
      "loss": 0.4693,
      "step": 290910
    },
    {
      "epoch": 119.08,
      "grad_norm": 1.740953803062439,
      "learning_rate": 1.8029039092084377e-06,
      "loss": 0.4527,
      "step": 290920
    },
    {
      "epoch": 119.09,
      "grad_norm": 2.121927499771118,
      "learning_rate": 1.802749562236251e-06,
      "loss": 0.4631,
      "step": 290930
    },
    {
      "epoch": 119.09,
      "grad_norm": 1.7485586404800415,
      "learning_rate": 1.8025952177423033e-06,
      "loss": 0.4568,
      "step": 290940
    },
    {
      "epoch": 119.1,
      "grad_norm": 2.30413556098938,
      "learning_rate": 1.8024408757273015e-06,
      "loss": 0.473,
      "step": 290950
    },
    {
      "epoch": 119.1,
      "grad_norm": 1.9646140336990356,
      "learning_rate": 1.8022865361919535e-06,
      "loss": 0.4437,
      "step": 290960
    },
    {
      "epoch": 119.1,
      "grad_norm": 2.5422191619873047,
      "learning_rate": 1.8021321991369655e-06,
      "loss": 0.4699,
      "step": 290970
    },
    {
      "epoch": 119.11,
      "grad_norm": 1.8408868312835693,
      "learning_rate": 1.8019778645630453e-06,
      "loss": 0.4508,
      "step": 290980
    },
    {
      "epoch": 119.11,
      "grad_norm": 1.855508804321289,
      "learning_rate": 1.8018235324708985e-06,
      "loss": 0.4685,
      "step": 290990
    },
    {
      "epoch": 119.12,
      "grad_norm": 1.8719487190246582,
      "learning_rate": 1.8016692028612334e-06,
      "loss": 0.4672,
      "step": 291000
    },
    {
      "epoch": 119.12,
      "grad_norm": 2.0642669200897217,
      "learning_rate": 1.8015148757347563e-06,
      "loss": 0.485,
      "step": 291010
    },
    {
      "epoch": 119.12,
      "grad_norm": 2.0603976249694824,
      "learning_rate": 1.8013605510921747e-06,
      "loss": 0.4704,
      "step": 291020
    },
    {
      "epoch": 119.13,
      "grad_norm": 1.5607551336288452,
      "learning_rate": 1.8012062289341945e-06,
      "loss": 0.4706,
      "step": 291030
    },
    {
      "epoch": 119.13,
      "grad_norm": 1.6427909135818481,
      "learning_rate": 1.8010519092615233e-06,
      "loss": 0.4602,
      "step": 291040
    },
    {
      "epoch": 119.14,
      "grad_norm": 1.9562175273895264,
      "learning_rate": 1.8008975920748683e-06,
      "loss": 0.4444,
      "step": 291050
    },
    {
      "epoch": 119.14,
      "grad_norm": 2.177185297012329,
      "learning_rate": 1.8007432773749361e-06,
      "loss": 0.4821,
      "step": 291060
    },
    {
      "epoch": 119.14,
      "grad_norm": 1.7626243829727173,
      "learning_rate": 1.8005889651624339e-06,
      "loss": 0.4727,
      "step": 291070
    },
    {
      "epoch": 119.15,
      "grad_norm": 2.1188228130340576,
      "learning_rate": 1.8004346554380678e-06,
      "loss": 0.4619,
      "step": 291080
    },
    {
      "epoch": 119.15,
      "grad_norm": 1.8244872093200684,
      "learning_rate": 1.8002803482025455e-06,
      "loss": 0.4635,
      "step": 291090
    },
    {
      "epoch": 119.16,
      "grad_norm": 2.2009987831115723,
      "learning_rate": 1.8001260434565735e-06,
      "loss": 0.4626,
      "step": 291100
    },
    {
      "epoch": 119.16,
      "grad_norm": 1.8266050815582275,
      "learning_rate": 1.7999717412008582e-06,
      "loss": 0.4553,
      "step": 291110
    },
    {
      "epoch": 119.16,
      "grad_norm": 1.7531307935714722,
      "learning_rate": 1.799817441436108e-06,
      "loss": 0.466,
      "step": 291120
    },
    {
      "epoch": 119.17,
      "grad_norm": 1.8362092971801758,
      "learning_rate": 1.7996631441630276e-06,
      "loss": 0.4387,
      "step": 291130
    },
    {
      "epoch": 119.17,
      "grad_norm": 2.1818411350250244,
      "learning_rate": 1.7995088493823252e-06,
      "loss": 0.4547,
      "step": 291140
    },
    {
      "epoch": 119.18,
      "grad_norm": 1.7895169258117676,
      "learning_rate": 1.799354557094707e-06,
      "loss": 0.4382,
      "step": 291150
    },
    {
      "epoch": 119.18,
      "grad_norm": 1.803667664527893,
      "learning_rate": 1.7992002673008804e-06,
      "loss": 0.4552,
      "step": 291160
    },
    {
      "epoch": 119.19,
      "grad_norm": 2.0012545585632324,
      "learning_rate": 1.7990459800015513e-06,
      "loss": 0.4702,
      "step": 291170
    },
    {
      "epoch": 119.19,
      "grad_norm": 2.4543397426605225,
      "learning_rate": 1.7988916951974272e-06,
      "loss": 0.4526,
      "step": 291180
    },
    {
      "epoch": 119.19,
      "grad_norm": 2.0257108211517334,
      "learning_rate": 1.7987374128892146e-06,
      "loss": 0.4784,
      "step": 291190
    },
    {
      "epoch": 119.2,
      "grad_norm": 2.281257390975952,
      "learning_rate": 1.7985831330776208e-06,
      "loss": 0.4597,
      "step": 291200
    },
    {
      "epoch": 119.2,
      "grad_norm": 2.0276150703430176,
      "learning_rate": 1.7984288557633511e-06,
      "loss": 0.4539,
      "step": 291210
    },
    {
      "epoch": 119.21,
      "grad_norm": 2.269315481185913,
      "learning_rate": 1.7982745809471125e-06,
      "loss": 0.4679,
      "step": 291220
    },
    {
      "epoch": 119.21,
      "grad_norm": 1.6626297235488892,
      "learning_rate": 1.7981203086296133e-06,
      "loss": 0.4727,
      "step": 291230
    },
    {
      "epoch": 119.21,
      "grad_norm": 2.1001877784729004,
      "learning_rate": 1.7979660388115592e-06,
      "loss": 0.4644,
      "step": 291240
    },
    {
      "epoch": 119.22,
      "grad_norm": 1.9002028703689575,
      "learning_rate": 1.7978117714936574e-06,
      "loss": 0.4524,
      "step": 291250
    },
    {
      "epoch": 119.22,
      "grad_norm": 2.427243709564209,
      "learning_rate": 1.7976575066766135e-06,
      "loss": 0.4721,
      "step": 291260
    },
    {
      "epoch": 119.23,
      "grad_norm": 1.8042999505996704,
      "learning_rate": 1.7975032443611347e-06,
      "loss": 0.461,
      "step": 291270
    },
    {
      "epoch": 119.23,
      "grad_norm": 2.1384949684143066,
      "learning_rate": 1.7973489845479278e-06,
      "loss": 0.4528,
      "step": 291280
    },
    {
      "epoch": 119.23,
      "grad_norm": 2.5589144229888916,
      "learning_rate": 1.7971947272376992e-06,
      "loss": 0.4528,
      "step": 291290
    },
    {
      "epoch": 119.24,
      "grad_norm": 1.8939802646636963,
      "learning_rate": 1.797040472431156e-06,
      "loss": 0.4452,
      "step": 291300
    },
    {
      "epoch": 119.24,
      "grad_norm": 2.1125030517578125,
      "learning_rate": 1.796886220129004e-06,
      "loss": 0.455,
      "step": 291310
    },
    {
      "epoch": 119.25,
      "grad_norm": 2.3751704692840576,
      "learning_rate": 1.7967319703319505e-06,
      "loss": 0.4583,
      "step": 291320
    },
    {
      "epoch": 119.25,
      "grad_norm": 1.7483246326446533,
      "learning_rate": 1.796577723040702e-06,
      "loss": 0.4621,
      "step": 291330
    },
    {
      "epoch": 119.26,
      "grad_norm": 1.9016374349594116,
      "learning_rate": 1.7964234782559654e-06,
      "loss": 0.4675,
      "step": 291340
    },
    {
      "epoch": 119.26,
      "grad_norm": 1.8710564374923706,
      "learning_rate": 1.796269235978446e-06,
      "loss": 0.4668,
      "step": 291350
    },
    {
      "epoch": 119.26,
      "grad_norm": 2.293937921524048,
      "learning_rate": 1.7961149962088517e-06,
      "loss": 0.4848,
      "step": 291360
    },
    {
      "epoch": 119.27,
      "grad_norm": 2.3585922718048096,
      "learning_rate": 1.7959607589478883e-06,
      "loss": 0.4526,
      "step": 291370
    },
    {
      "epoch": 119.27,
      "grad_norm": 1.9422632455825806,
      "learning_rate": 1.7958065241962626e-06,
      "loss": 0.474,
      "step": 291380
    },
    {
      "epoch": 119.28,
      "grad_norm": 2.790891408920288,
      "learning_rate": 1.795652291954681e-06,
      "loss": 0.4527,
      "step": 291390
    },
    {
      "epoch": 119.28,
      "grad_norm": 2.0876262187957764,
      "learning_rate": 1.79549806222385e-06,
      "loss": 0.4599,
      "step": 291400
    },
    {
      "epoch": 119.28,
      "grad_norm": 2.0354158878326416,
      "learning_rate": 1.7953438350044762e-06,
      "loss": 0.4767,
      "step": 291410
    },
    {
      "epoch": 119.29,
      "grad_norm": 2.631143093109131,
      "learning_rate": 1.7951896102972662e-06,
      "loss": 0.4573,
      "step": 291420
    },
    {
      "epoch": 119.29,
      "grad_norm": 2.40830397605896,
      "learning_rate": 1.7950353881029266e-06,
      "loss": 0.4532,
      "step": 291430
    },
    {
      "epoch": 119.3,
      "grad_norm": 1.7181696891784668,
      "learning_rate": 1.7948811684221636e-06,
      "loss": 0.4473,
      "step": 291440
    },
    {
      "epoch": 119.3,
      "grad_norm": 2.1046690940856934,
      "learning_rate": 1.7947269512556837e-06,
      "loss": 0.4703,
      "step": 291450
    },
    {
      "epoch": 119.3,
      "grad_norm": 1.9314756393432617,
      "learning_rate": 1.7945727366041934e-06,
      "loss": 0.4802,
      "step": 291460
    },
    {
      "epoch": 119.31,
      "grad_norm": 1.750720500946045,
      "learning_rate": 1.7944185244683988e-06,
      "loss": 0.4538,
      "step": 291470
    },
    {
      "epoch": 119.31,
      "grad_norm": 2.0769853591918945,
      "learning_rate": 1.7942643148490077e-06,
      "loss": 0.4654,
      "step": 291480
    },
    {
      "epoch": 119.32,
      "grad_norm": 1.6728122234344482,
      "learning_rate": 1.7941101077467246e-06,
      "loss": 0.4612,
      "step": 291490
    },
    {
      "epoch": 119.32,
      "grad_norm": 1.912512183189392,
      "learning_rate": 1.7939559031622565e-06,
      "loss": 0.4694,
      "step": 291500
    },
    {
      "epoch": 119.32,
      "grad_norm": 2.23877215385437,
      "learning_rate": 1.7938017010963104e-06,
      "loss": 0.4845,
      "step": 291510
    },
    {
      "epoch": 119.33,
      "grad_norm": 1.8941092491149902,
      "learning_rate": 1.7936475015495922e-06,
      "loss": 0.4744,
      "step": 291520
    },
    {
      "epoch": 119.33,
      "grad_norm": 1.8163751363754272,
      "learning_rate": 1.793493304522808e-06,
      "loss": 0.4756,
      "step": 291530
    },
    {
      "epoch": 119.34,
      "grad_norm": 1.7628604173660278,
      "learning_rate": 1.793339110016665e-06,
      "loss": 0.453,
      "step": 291540
    },
    {
      "epoch": 119.34,
      "grad_norm": 1.5831236839294434,
      "learning_rate": 1.7931849180318685e-06,
      "loss": 0.4609,
      "step": 291550
    },
    {
      "epoch": 119.35,
      "grad_norm": 2.0266153812408447,
      "learning_rate": 1.7930307285691262e-06,
      "loss": 0.452,
      "step": 291560
    },
    {
      "epoch": 119.35,
      "grad_norm": 2.2413530349731445,
      "learning_rate": 1.7928765416291432e-06,
      "loss": 0.4498,
      "step": 291570
    },
    {
      "epoch": 119.35,
      "grad_norm": 2.0379903316497803,
      "learning_rate": 1.7927223572126261e-06,
      "loss": 0.4631,
      "step": 291580
    },
    {
      "epoch": 119.36,
      "grad_norm": 1.771504282951355,
      "learning_rate": 1.7925681753202812e-06,
      "loss": 0.4567,
      "step": 291590
    },
    {
      "epoch": 119.36,
      "grad_norm": 1.9912647008895874,
      "learning_rate": 1.792413995952814e-06,
      "loss": 0.4479,
      "step": 291600
    },
    {
      "epoch": 119.37,
      "grad_norm": 2.132817268371582,
      "learning_rate": 1.7922598191109334e-06,
      "loss": 0.4721,
      "step": 291610
    },
    {
      "epoch": 119.37,
      "grad_norm": 2.2539331912994385,
      "learning_rate": 1.7921056447953429e-06,
      "loss": 0.4732,
      "step": 291620
    },
    {
      "epoch": 119.37,
      "grad_norm": 2.048506498336792,
      "learning_rate": 1.79195147300675e-06,
      "loss": 0.4653,
      "step": 291630
    },
    {
      "epoch": 119.38,
      "grad_norm": 1.9180853366851807,
      "learning_rate": 1.7917973037458608e-06,
      "loss": 0.462,
      "step": 291640
    },
    {
      "epoch": 119.38,
      "grad_norm": 1.8361319303512573,
      "learning_rate": 1.791643137013381e-06,
      "loss": 0.4436,
      "step": 291650
    },
    {
      "epoch": 119.39,
      "grad_norm": 3.9481077194213867,
      "learning_rate": 1.7914889728100178e-06,
      "loss": 0.4753,
      "step": 291660
    },
    {
      "epoch": 119.39,
      "grad_norm": 1.8744378089904785,
      "learning_rate": 1.7913348111364764e-06,
      "loss": 0.486,
      "step": 291670
    },
    {
      "epoch": 119.39,
      "grad_norm": 2.2069509029388428,
      "learning_rate": 1.7911806519934637e-06,
      "loss": 0.4386,
      "step": 291680
    },
    {
      "epoch": 119.4,
      "grad_norm": 1.9149632453918457,
      "learning_rate": 1.7910264953816853e-06,
      "loss": 0.4766,
      "step": 291690
    },
    {
      "epoch": 119.4,
      "grad_norm": 2.6740643978118896,
      "learning_rate": 1.7908723413018485e-06,
      "loss": 0.462,
      "step": 291700
    },
    {
      "epoch": 119.41,
      "grad_norm": 1.8583670854568481,
      "learning_rate": 1.790718189754658e-06,
      "loss": 0.4665,
      "step": 291710
    },
    {
      "epoch": 119.41,
      "grad_norm": 2.0807480812072754,
      "learning_rate": 1.7905640407408206e-06,
      "loss": 0.4743,
      "step": 291720
    },
    {
      "epoch": 119.41,
      "grad_norm": 1.9525985717773438,
      "learning_rate": 1.7904098942610424e-06,
      "loss": 0.4342,
      "step": 291730
    },
    {
      "epoch": 119.42,
      "grad_norm": 1.9696751832962036,
      "learning_rate": 1.7902557503160293e-06,
      "loss": 0.4535,
      "step": 291740
    },
    {
      "epoch": 119.42,
      "grad_norm": 1.8197449445724487,
      "learning_rate": 1.7901016089064878e-06,
      "loss": 0.4337,
      "step": 291750
    },
    {
      "epoch": 119.43,
      "grad_norm": 1.6086610555648804,
      "learning_rate": 1.7899474700331241e-06,
      "loss": 0.4614,
      "step": 291760
    },
    {
      "epoch": 119.43,
      "grad_norm": 1.6545995473861694,
      "learning_rate": 1.7897933336966438e-06,
      "loss": 0.4465,
      "step": 291770
    },
    {
      "epoch": 119.44,
      "grad_norm": 1.907640814781189,
      "learning_rate": 1.789639199897753e-06,
      "loss": 0.4404,
      "step": 291780
    },
    {
      "epoch": 119.44,
      "grad_norm": 1.9110512733459473,
      "learning_rate": 1.789485068637158e-06,
      "loss": 0.4482,
      "step": 291790
    },
    {
      "epoch": 119.44,
      "grad_norm": 2.163454532623291,
      "learning_rate": 1.789330939915565e-06,
      "loss": 0.4661,
      "step": 291800
    },
    {
      "epoch": 119.45,
      "grad_norm": 1.747219204902649,
      "learning_rate": 1.7891768137336796e-06,
      "loss": 0.4606,
      "step": 291810
    },
    {
      "epoch": 119.45,
      "grad_norm": 1.7175023555755615,
      "learning_rate": 1.7890226900922083e-06,
      "loss": 0.466,
      "step": 291820
    },
    {
      "epoch": 119.46,
      "grad_norm": 1.652146577835083,
      "learning_rate": 1.788868568991857e-06,
      "loss": 0.4747,
      "step": 291830
    },
    {
      "epoch": 119.46,
      "grad_norm": 2.4043092727661133,
      "learning_rate": 1.7887144504333321e-06,
      "loss": 0.4683,
      "step": 291840
    },
    {
      "epoch": 119.46,
      "grad_norm": 2.0977413654327393,
      "learning_rate": 1.7885603344173382e-06,
      "loss": 0.4788,
      "step": 291850
    },
    {
      "epoch": 119.47,
      "grad_norm": 1.778066873550415,
      "learning_rate": 1.7884062209445826e-06,
      "loss": 0.4901,
      "step": 291860
    },
    {
      "epoch": 119.47,
      "grad_norm": 1.8316432237625122,
      "learning_rate": 1.788252110015771e-06,
      "loss": 0.4678,
      "step": 291870
    },
    {
      "epoch": 119.48,
      "grad_norm": 1.9414550065994263,
      "learning_rate": 1.7880980016316087e-06,
      "loss": 0.4773,
      "step": 291880
    },
    {
      "epoch": 119.48,
      "grad_norm": 2.082260847091675,
      "learning_rate": 1.7879438957928025e-06,
      "loss": 0.4508,
      "step": 291890
    },
    {
      "epoch": 119.48,
      "grad_norm": 2.020639181137085,
      "learning_rate": 1.7877897925000578e-06,
      "loss": 0.4702,
      "step": 291900
    },
    {
      "epoch": 119.49,
      "grad_norm": 2.285343647003174,
      "learning_rate": 1.7876356917540804e-06,
      "loss": 0.47,
      "step": 291910
    },
    {
      "epoch": 119.49,
      "grad_norm": 1.350475549697876,
      "learning_rate": 1.7874815935555777e-06,
      "loss": 0.4586,
      "step": 291920
    },
    {
      "epoch": 119.5,
      "grad_norm": 1.8394144773483276,
      "learning_rate": 1.7873274979052537e-06,
      "loss": 0.4817,
      "step": 291930
    },
    {
      "epoch": 119.5,
      "grad_norm": 2.1429617404937744,
      "learning_rate": 1.7871734048038147e-06,
      "loss": 0.462,
      "step": 291940
    },
    {
      "epoch": 119.5,
      "grad_norm": 2.4902327060699463,
      "learning_rate": 1.787019314251967e-06,
      "loss": 0.4631,
      "step": 291950
    },
    {
      "epoch": 119.51,
      "grad_norm": 2.409696578979492,
      "learning_rate": 1.7868652262504167e-06,
      "loss": 0.4396,
      "step": 291960
    },
    {
      "epoch": 119.51,
      "grad_norm": 1.9141639471054077,
      "learning_rate": 1.786711140799869e-06,
      "loss": 0.4635,
      "step": 291970
    },
    {
      "epoch": 119.52,
      "grad_norm": 2.0629220008850098,
      "learning_rate": 1.7865570579010295e-06,
      "loss": 0.4687,
      "step": 291980
    },
    {
      "epoch": 119.52,
      "grad_norm": 2.307865858078003,
      "learning_rate": 1.7864029775546055e-06,
      "loss": 0.4936,
      "step": 291990
    },
    {
      "epoch": 119.53,
      "grad_norm": 1.571708083152771,
      "learning_rate": 1.7862488997613015e-06,
      "loss": 0.4512,
      "step": 292000
    },
    {
      "epoch": 119.53,
      "grad_norm": 2.6052205562591553,
      "learning_rate": 1.7860948245218238e-06,
      "loss": 0.4639,
      "step": 292010
    },
    {
      "epoch": 119.53,
      "grad_norm": 1.915576696395874,
      "learning_rate": 1.7859407518368781e-06,
      "loss": 0.4508,
      "step": 292020
    },
    {
      "epoch": 119.54,
      "grad_norm": 1.9498107433319092,
      "learning_rate": 1.7857866817071702e-06,
      "loss": 0.4535,
      "step": 292030
    },
    {
      "epoch": 119.54,
      "grad_norm": 2.280663251876831,
      "learning_rate": 1.7856326141334056e-06,
      "loss": 0.4485,
      "step": 292040
    },
    {
      "epoch": 119.55,
      "grad_norm": 1.9918919801712036,
      "learning_rate": 1.7854785491162907e-06,
      "loss": 0.4559,
      "step": 292050
    },
    {
      "epoch": 119.55,
      "grad_norm": 1.8274478912353516,
      "learning_rate": 1.7853244866565312e-06,
      "loss": 0.4333,
      "step": 292060
    },
    {
      "epoch": 119.55,
      "grad_norm": 1.8703924417495728,
      "learning_rate": 1.7851704267548317e-06,
      "loss": 0.4693,
      "step": 292070
    },
    {
      "epoch": 119.56,
      "grad_norm": 2.4021525382995605,
      "learning_rate": 1.785016369411899e-06,
      "loss": 0.4781,
      "step": 292080
    },
    {
      "epoch": 119.56,
      "grad_norm": 1.9103704690933228,
      "learning_rate": 1.7848623146284387e-06,
      "loss": 0.4615,
      "step": 292090
    },
    {
      "epoch": 119.57,
      "grad_norm": 1.6468039751052856,
      "learning_rate": 1.7847082624051561e-06,
      "loss": 0.4644,
      "step": 292100
    },
    {
      "epoch": 119.57,
      "grad_norm": 1.9338923692703247,
      "learning_rate": 1.784554212742757e-06,
      "loss": 0.4596,
      "step": 292110
    },
    {
      "epoch": 119.57,
      "grad_norm": 1.9379699230194092,
      "learning_rate": 1.7844001656419475e-06,
      "loss": 0.4453,
      "step": 292120
    },
    {
      "epoch": 119.58,
      "grad_norm": 2.291567087173462,
      "learning_rate": 1.7842461211034328e-06,
      "loss": 0.4746,
      "step": 292130
    },
    {
      "epoch": 119.58,
      "grad_norm": 1.8890303373336792,
      "learning_rate": 1.7840920791279187e-06,
      "loss": 0.4577,
      "step": 292140
    },
    {
      "epoch": 119.59,
      "grad_norm": 2.251993179321289,
      "learning_rate": 1.7839380397161115e-06,
      "loss": 0.4749,
      "step": 292150
    },
    {
      "epoch": 119.59,
      "grad_norm": 1.9634708166122437,
      "learning_rate": 1.7837840028687157e-06,
      "loss": 0.4833,
      "step": 292160
    },
    {
      "epoch": 119.59,
      "grad_norm": 1.7687751054763794,
      "learning_rate": 1.783629968586436e-06,
      "loss": 0.4694,
      "step": 292170
    },
    {
      "epoch": 119.6,
      "grad_norm": 2.103070020675659,
      "learning_rate": 1.783475936869981e-06,
      "loss": 0.4569,
      "step": 292180
    },
    {
      "epoch": 119.6,
      "grad_norm": 1.5373973846435547,
      "learning_rate": 1.783321907720055e-06,
      "loss": 0.4845,
      "step": 292190
    },
    {
      "epoch": 119.61,
      "grad_norm": 1.7659447193145752,
      "learning_rate": 1.7831678811373625e-06,
      "loss": 0.4473,
      "step": 292200
    },
    {
      "epoch": 119.61,
      "grad_norm": 1.5712822675704956,
      "learning_rate": 1.7830138571226102e-06,
      "loss": 0.4425,
      "step": 292210
    },
    {
      "epoch": 119.62,
      "grad_norm": 2.2909207344055176,
      "learning_rate": 1.7828598356765032e-06,
      "loss": 0.4629,
      "step": 292220
    },
    {
      "epoch": 119.62,
      "grad_norm": 1.7680718898773193,
      "learning_rate": 1.7827058167997471e-06,
      "loss": 0.4555,
      "step": 292230
    },
    {
      "epoch": 119.62,
      "grad_norm": 2.098653554916382,
      "learning_rate": 1.7825518004930479e-06,
      "loss": 0.4438,
      "step": 292240
    },
    {
      "epoch": 119.63,
      "grad_norm": 2.1886863708496094,
      "learning_rate": 1.7823977867571103e-06,
      "loss": 0.4535,
      "step": 292250
    },
    {
      "epoch": 119.63,
      "grad_norm": 1.8705084323883057,
      "learning_rate": 1.7822437755926405e-06,
      "loss": 0.4616,
      "step": 292260
    },
    {
      "epoch": 119.64,
      "grad_norm": 2.3253536224365234,
      "learning_rate": 1.7820897670003438e-06,
      "loss": 0.4587,
      "step": 292270
    },
    {
      "epoch": 119.64,
      "grad_norm": 1.8329113721847534,
      "learning_rate": 1.781935760980926e-06,
      "loss": 0.465,
      "step": 292280
    },
    {
      "epoch": 119.64,
      "grad_norm": 2.0347094535827637,
      "learning_rate": 1.7817817575350917e-06,
      "loss": 0.4647,
      "step": 292290
    },
    {
      "epoch": 119.65,
      "grad_norm": 1.64849054813385,
      "learning_rate": 1.7816277566635469e-06,
      "loss": 0.4662,
      "step": 292300
    },
    {
      "epoch": 119.65,
      "grad_norm": 2.0026121139526367,
      "learning_rate": 1.7814737583669973e-06,
      "loss": 0.4438,
      "step": 292310
    },
    {
      "epoch": 119.66,
      "grad_norm": 2.741283655166626,
      "learning_rate": 1.7813197626461478e-06,
      "loss": 0.4779,
      "step": 292320
    },
    {
      "epoch": 119.66,
      "grad_norm": 1.4785094261169434,
      "learning_rate": 1.7811657695017038e-06,
      "loss": 0.4552,
      "step": 292330
    },
    {
      "epoch": 119.66,
      "grad_norm": 1.8869720697402954,
      "learning_rate": 1.7810117789343717e-06,
      "loss": 0.4843,
      "step": 292340
    },
    {
      "epoch": 119.67,
      "grad_norm": 2.204982042312622,
      "learning_rate": 1.7808577909448559e-06,
      "loss": 0.4734,
      "step": 292350
    },
    {
      "epoch": 119.67,
      "grad_norm": 2.235952377319336,
      "learning_rate": 1.7807038055338617e-06,
      "loss": 0.4596,
      "step": 292360
    },
    {
      "epoch": 119.68,
      "grad_norm": 2.2357521057128906,
      "learning_rate": 1.7805498227020955e-06,
      "loss": 0.4728,
      "step": 292370
    },
    {
      "epoch": 119.68,
      "grad_norm": 1.8819384574890137,
      "learning_rate": 1.7803958424502618e-06,
      "loss": 0.4845,
      "step": 292380
    },
    {
      "epoch": 119.68,
      "grad_norm": 1.8460781574249268,
      "learning_rate": 1.7802418647790665e-06,
      "loss": 0.4388,
      "step": 292390
    },
    {
      "epoch": 119.69,
      "grad_norm": 1.6290004253387451,
      "learning_rate": 1.7800878896892145e-06,
      "loss": 0.4643,
      "step": 292400
    },
    {
      "epoch": 119.69,
      "grad_norm": 1.6631793975830078,
      "learning_rate": 1.7799339171814114e-06,
      "loss": 0.4643,
      "step": 292410
    },
    {
      "epoch": 119.7,
      "grad_norm": 2.276703357696533,
      "learning_rate": 1.7797799472563633e-06,
      "loss": 0.4571,
      "step": 292420
    },
    {
      "epoch": 119.7,
      "grad_norm": 2.314619779586792,
      "learning_rate": 1.779625979914774e-06,
      "loss": 0.469,
      "step": 292430
    },
    {
      "epoch": 119.71,
      "grad_norm": 2.0405166149139404,
      "learning_rate": 1.7794720151573492e-06,
      "loss": 0.4576,
      "step": 292440
    },
    {
      "epoch": 119.71,
      "grad_norm": 1.9396278858184814,
      "learning_rate": 1.7793180529847945e-06,
      "loss": 0.4594,
      "step": 292450
    },
    {
      "epoch": 119.71,
      "grad_norm": 1.6260391473770142,
      "learning_rate": 1.7791640933978153e-06,
      "loss": 0.4562,
      "step": 292460
    },
    {
      "epoch": 119.72,
      "grad_norm": 2.020530939102173,
      "learning_rate": 1.7790101363971167e-06,
      "loss": 0.4506,
      "step": 292470
    },
    {
      "epoch": 119.72,
      "grad_norm": 2.1161513328552246,
      "learning_rate": 1.778856181983404e-06,
      "loss": 0.458,
      "step": 292480
    },
    {
      "epoch": 119.73,
      "grad_norm": 1.9716283082962036,
      "learning_rate": 1.7787022301573822e-06,
      "loss": 0.4548,
      "step": 292490
    },
    {
      "epoch": 119.73,
      "grad_norm": 1.6790162324905396,
      "learning_rate": 1.7785482809197573e-06,
      "loss": 0.4615,
      "step": 292500
    },
    {
      "epoch": 119.73,
      "grad_norm": 1.7197463512420654,
      "learning_rate": 1.7783943342712335e-06,
      "loss": 0.4543,
      "step": 292510
    },
    {
      "epoch": 119.74,
      "grad_norm": 1.9463902711868286,
      "learning_rate": 1.7782403902125165e-06,
      "loss": 0.464,
      "step": 292520
    },
    {
      "epoch": 119.74,
      "grad_norm": 1.6882444620132446,
      "learning_rate": 1.7780864487443115e-06,
      "loss": 0.4597,
      "step": 292530
    },
    {
      "epoch": 119.75,
      "grad_norm": 1.7326709032058716,
      "learning_rate": 1.7779325098673224e-06,
      "loss": 0.4546,
      "step": 292540
    },
    {
      "epoch": 119.75,
      "grad_norm": 2.1214542388916016,
      "learning_rate": 1.777778573582257e-06,
      "loss": 0.4636,
      "step": 292550
    },
    {
      "epoch": 119.75,
      "grad_norm": 1.874449372291565,
      "learning_rate": 1.7776246398898188e-06,
      "loss": 0.4587,
      "step": 292560
    },
    {
      "epoch": 119.76,
      "grad_norm": 1.7880948781967163,
      "learning_rate": 1.7774707087907133e-06,
      "loss": 0.4677,
      "step": 292570
    },
    {
      "epoch": 119.76,
      "grad_norm": 1.8519541025161743,
      "learning_rate": 1.7773167802856453e-06,
      "loss": 0.4624,
      "step": 292580
    },
    {
      "epoch": 119.77,
      "grad_norm": 1.8867651224136353,
      "learning_rate": 1.7771628543753201e-06,
      "loss": 0.4474,
      "step": 292590
    },
    {
      "epoch": 119.77,
      "grad_norm": 2.29862642288208,
      "learning_rate": 1.7770089310604432e-06,
      "loss": 0.4681,
      "step": 292600
    },
    {
      "epoch": 119.77,
      "grad_norm": 2.342679977416992,
      "learning_rate": 1.776855010341719e-06,
      "loss": 0.4698,
      "step": 292610
    },
    {
      "epoch": 119.78,
      "grad_norm": 1.6830577850341797,
      "learning_rate": 1.7767010922198531e-06,
      "loss": 0.4328,
      "step": 292620
    },
    {
      "epoch": 119.78,
      "grad_norm": 2.548283100128174,
      "learning_rate": 1.7765471766955505e-06,
      "loss": 0.4693,
      "step": 292630
    },
    {
      "epoch": 119.79,
      "grad_norm": 1.9695645570755005,
      "learning_rate": 1.7763932637695168e-06,
      "loss": 0.446,
      "step": 292640
    },
    {
      "epoch": 119.79,
      "grad_norm": 1.6667125225067139,
      "learning_rate": 1.7762393534424557e-06,
      "loss": 0.4591,
      "step": 292650
    },
    {
      "epoch": 119.8,
      "grad_norm": 1.8812365531921387,
      "learning_rate": 1.7760854457150731e-06,
      "loss": 0.4559,
      "step": 292660
    },
    {
      "epoch": 119.8,
      "grad_norm": 1.7823973894119263,
      "learning_rate": 1.7759315405880737e-06,
      "loss": 0.4495,
      "step": 292670
    },
    {
      "epoch": 119.8,
      "grad_norm": 1.9535093307495117,
      "learning_rate": 1.775777638062163e-06,
      "loss": 0.4603,
      "step": 292680
    },
    {
      "epoch": 119.81,
      "grad_norm": 2.191925525665283,
      "learning_rate": 1.7756237381380455e-06,
      "loss": 0.4757,
      "step": 292690
    },
    {
      "epoch": 119.81,
      "grad_norm": 1.9223273992538452,
      "learning_rate": 1.7754698408164267e-06,
      "loss": 0.4694,
      "step": 292700
    },
    {
      "epoch": 119.82,
      "grad_norm": 2.1400232315063477,
      "learning_rate": 1.775315946098011e-06,
      "loss": 0.4926,
      "step": 292710
    },
    {
      "epoch": 119.82,
      "grad_norm": 2.480605125427246,
      "learning_rate": 1.7751620539835038e-06,
      "loss": 0.4869,
      "step": 292720
    },
    {
      "epoch": 119.82,
      "grad_norm": 8.261990547180176,
      "learning_rate": 1.7750081644736095e-06,
      "loss": 0.4516,
      "step": 292730
    },
    {
      "epoch": 119.83,
      "grad_norm": 1.9833495616912842,
      "learning_rate": 1.7748542775690338e-06,
      "loss": 0.4539,
      "step": 292740
    },
    {
      "epoch": 119.83,
      "grad_norm": 1.8114632368087769,
      "learning_rate": 1.7747003932704815e-06,
      "loss": 0.4673,
      "step": 292750
    },
    {
      "epoch": 119.84,
      "grad_norm": 1.9773786067962646,
      "learning_rate": 1.774546511578657e-06,
      "loss": 0.451,
      "step": 292760
    },
    {
      "epoch": 119.84,
      "grad_norm": 1.89958918094635,
      "learning_rate": 1.7743926324942658e-06,
      "loss": 0.4477,
      "step": 292770
    },
    {
      "epoch": 119.84,
      "grad_norm": 2.114279270172119,
      "learning_rate": 1.774238756018013e-06,
      "loss": 0.4741,
      "step": 292780
    },
    {
      "epoch": 119.85,
      "grad_norm": 1.8447670936584473,
      "learning_rate": 1.7740848821506026e-06,
      "loss": 0.4884,
      "step": 292790
    },
    {
      "epoch": 119.85,
      "grad_norm": 2.156367778778076,
      "learning_rate": 1.7739310108927398e-06,
      "loss": 0.4517,
      "step": 292800
    },
    {
      "epoch": 119.86,
      "grad_norm": 1.9938833713531494,
      "learning_rate": 1.7737771422451296e-06,
      "loss": 0.4559,
      "step": 292810
    },
    {
      "epoch": 119.86,
      "grad_norm": 2.297144651412964,
      "learning_rate": 1.7736232762084766e-06,
      "loss": 0.4488,
      "step": 292820
    },
    {
      "epoch": 119.86,
      "grad_norm": 2.179982900619507,
      "learning_rate": 1.7734694127834858e-06,
      "loss": 0.467,
      "step": 292830
    },
    {
      "epoch": 119.87,
      "grad_norm": 1.9740610122680664,
      "learning_rate": 1.7733155519708624e-06,
      "loss": 0.4608,
      "step": 292840
    },
    {
      "epoch": 119.87,
      "grad_norm": 1.6741695404052734,
      "learning_rate": 1.7731616937713104e-06,
      "loss": 0.4652,
      "step": 292850
    },
    {
      "epoch": 119.88,
      "grad_norm": 1.944794774055481,
      "learning_rate": 1.773007838185536e-06,
      "loss": 0.4535,
      "step": 292860
    },
    {
      "epoch": 119.88,
      "grad_norm": 2.012798547744751,
      "learning_rate": 1.7728539852142422e-06,
      "loss": 0.4633,
      "step": 292870
    },
    {
      "epoch": 119.89,
      "grad_norm": 1.536675214767456,
      "learning_rate": 1.7727001348581346e-06,
      "loss": 0.4689,
      "step": 292880
    },
    {
      "epoch": 119.89,
      "grad_norm": 2.1583917140960693,
      "learning_rate": 1.7725462871179185e-06,
      "loss": 0.4499,
      "step": 292890
    },
    {
      "epoch": 119.89,
      "grad_norm": 1.9884400367736816,
      "learning_rate": 1.7723924419942978e-06,
      "loss": 0.4607,
      "step": 292900
    },
    {
      "epoch": 119.9,
      "grad_norm": 1.924034595489502,
      "learning_rate": 1.7722385994879775e-06,
      "loss": 0.4569,
      "step": 292910
    },
    {
      "epoch": 119.9,
      "grad_norm": 2.0492281913757324,
      "learning_rate": 1.772084759599662e-06,
      "loss": 0.4683,
      "step": 292920
    },
    {
      "epoch": 119.91,
      "grad_norm": 2.0384202003479004,
      "learning_rate": 1.7719309223300573e-06,
      "loss": 0.4793,
      "step": 292930
    },
    {
      "epoch": 119.91,
      "grad_norm": 1.8291977643966675,
      "learning_rate": 1.7717770876798667e-06,
      "loss": 0.4518,
      "step": 292940
    },
    {
      "epoch": 119.91,
      "grad_norm": 1.8626290559768677,
      "learning_rate": 1.7716232556497957e-06,
      "loss": 0.4452,
      "step": 292950
    },
    {
      "epoch": 119.92,
      "grad_norm": 1.8620457649230957,
      "learning_rate": 1.7714694262405484e-06,
      "loss": 0.4574,
      "step": 292960
    },
    {
      "epoch": 119.92,
      "grad_norm": 2.0565829277038574,
      "learning_rate": 1.7713155994528302e-06,
      "loss": 0.4828,
      "step": 292970
    },
    {
      "epoch": 119.93,
      "grad_norm": 2.036835193634033,
      "learning_rate": 1.7711617752873451e-06,
      "loss": 0.4608,
      "step": 292980
    },
    {
      "epoch": 119.93,
      "grad_norm": 1.776193618774414,
      "learning_rate": 1.771007953744798e-06,
      "loss": 0.4467,
      "step": 292990
    },
    {
      "epoch": 119.93,
      "grad_norm": 2.1406538486480713,
      "learning_rate": 1.770854134825894e-06,
      "loss": 0.4745,
      "step": 293000
    },
    {
      "epoch": 119.94,
      "grad_norm": 1.9701614379882812,
      "learning_rate": 1.7707003185313367e-06,
      "loss": 0.4691,
      "step": 293010
    },
    {
      "epoch": 119.94,
      "grad_norm": 2.005432367324829,
      "learning_rate": 1.7705465048618314e-06,
      "loss": 0.4483,
      "step": 293020
    },
    {
      "epoch": 119.95,
      "grad_norm": 1.559082269668579,
      "learning_rate": 1.7703926938180825e-06,
      "loss": 0.4539,
      "step": 293030
    },
    {
      "epoch": 119.95,
      "grad_norm": 2.0843145847320557,
      "learning_rate": 1.7702388854007945e-06,
      "loss": 0.4378,
      "step": 293040
    },
    {
      "epoch": 119.95,
      "grad_norm": 1.9843029975891113,
      "learning_rate": 1.7700850796106722e-06,
      "loss": 0.4684,
      "step": 293050
    },
    {
      "epoch": 119.96,
      "grad_norm": 2.0455524921417236,
      "learning_rate": 1.76993127644842e-06,
      "loss": 0.4617,
      "step": 293060
    },
    {
      "epoch": 119.96,
      "grad_norm": 1.898970127105713,
      "learning_rate": 1.7697774759147428e-06,
      "loss": 0.4642,
      "step": 293070
    },
    {
      "epoch": 119.97,
      "grad_norm": 1.785555124282837,
      "learning_rate": 1.7696236780103445e-06,
      "loss": 0.4701,
      "step": 293080
    },
    {
      "epoch": 119.97,
      "grad_norm": 2.4708240032196045,
      "learning_rate": 1.7694698827359306e-06,
      "loss": 0.4829,
      "step": 293090
    },
    {
      "epoch": 119.98,
      "grad_norm": 2.087974786758423,
      "learning_rate": 1.7693160900922045e-06,
      "loss": 0.4459,
      "step": 293100
    },
    {
      "epoch": 119.98,
      "grad_norm": 1.928651213645935,
      "learning_rate": 1.7691623000798699e-06,
      "loss": 0.4561,
      "step": 293110
    },
    {
      "epoch": 119.98,
      "grad_norm": 2.2060396671295166,
      "learning_rate": 1.769008512699634e-06,
      "loss": 0.4795,
      "step": 293120
    },
    {
      "epoch": 119.99,
      "grad_norm": 1.583176612854004,
      "learning_rate": 1.7688547279521994e-06,
      "loss": 0.4652,
      "step": 293130
    },
    {
      "epoch": 119.99,
      "grad_norm": 1.8936814069747925,
      "learning_rate": 1.7687009458382717e-06,
      "loss": 0.4487,
      "step": 293140
    },
    {
      "epoch": 120.0,
      "grad_norm": 1.8449660539627075,
      "learning_rate": 1.7685471663585544e-06,
      "loss": 0.4646,
      "step": 293150
    },
    {
      "epoch": 120.0,
      "grad_norm": 2.0251617431640625,
      "learning_rate": 1.7683933895137519e-06,
      "loss": 0.4622,
      "step": 293160
    },
    {
      "epoch": 120.0,
      "eval_loss": 0.46338173747062683,
      "eval_runtime": 52.2192,
      "eval_samples_per_second": 66.049,
      "eval_steps_per_second": 8.273,
      "step": 293160
    },
    {
      "epoch": 120.0,
      "grad_norm": 1.9686614274978638,
      "learning_rate": 1.7682396153045692e-06,
      "loss": 0.4477,
      "step": 293170
    },
    {
      "epoch": 120.01,
      "grad_norm": 2.2357301712036133,
      "learning_rate": 1.7680858437317101e-06,
      "loss": 0.4653,
      "step": 293180
    },
    {
      "epoch": 120.01,
      "grad_norm": 2.116075277328491,
      "learning_rate": 1.7679320747958795e-06,
      "loss": 0.4532,
      "step": 293190
    },
    {
      "epoch": 120.02,
      "grad_norm": 2.7900261878967285,
      "learning_rate": 1.7677783084977817e-06,
      "loss": 0.4601,
      "step": 293200
    },
    {
      "epoch": 120.02,
      "grad_norm": 1.8247371912002563,
      "learning_rate": 1.7676245448381208e-06,
      "loss": 0.4711,
      "step": 293210
    },
    {
      "epoch": 120.02,
      "grad_norm": 2.074151039123535,
      "learning_rate": 1.7674707838176022e-06,
      "loss": 0.455,
      "step": 293220
    },
    {
      "epoch": 120.03,
      "grad_norm": 2.256507635116577,
      "learning_rate": 1.7673170254369286e-06,
      "loss": 0.4658,
      "step": 293230
    },
    {
      "epoch": 120.03,
      "grad_norm": 1.8943334817886353,
      "learning_rate": 1.7671632696968056e-06,
      "loss": 0.4689,
      "step": 293240
    },
    {
      "epoch": 120.04,
      "grad_norm": 2.0158791542053223,
      "learning_rate": 1.767009516597937e-06,
      "loss": 0.443,
      "step": 293250
    },
    {
      "epoch": 120.04,
      "grad_norm": 1.8896310329437256,
      "learning_rate": 1.766855766141027e-06,
      "loss": 0.4553,
      "step": 293260
    },
    {
      "epoch": 120.05,
      "grad_norm": 2.6207945346832275,
      "learning_rate": 1.7667020183267803e-06,
      "loss": 0.4516,
      "step": 293270
    },
    {
      "epoch": 120.05,
      "grad_norm": 2.019421100616455,
      "learning_rate": 1.7665482731559011e-06,
      "loss": 0.4658,
      "step": 293280
    },
    {
      "epoch": 120.05,
      "grad_norm": 2.898770809173584,
      "learning_rate": 1.7663945306290935e-06,
      "loss": 0.4428,
      "step": 293290
    },
    {
      "epoch": 120.06,
      "grad_norm": 2.2011289596557617,
      "learning_rate": 1.7662407907470613e-06,
      "loss": 0.4709,
      "step": 293300
    },
    {
      "epoch": 120.06,
      "grad_norm": 2.152208089828491,
      "learning_rate": 1.76608705351051e-06,
      "loss": 0.4436,
      "step": 293310
    },
    {
      "epoch": 120.07,
      "grad_norm": 2.1295435428619385,
      "learning_rate": 1.7659333189201432e-06,
      "loss": 0.4431,
      "step": 293320
    },
    {
      "epoch": 120.07,
      "grad_norm": 1.6084004640579224,
      "learning_rate": 1.7657795869766652e-06,
      "loss": 0.4483,
      "step": 293330
    },
    {
      "epoch": 120.07,
      "grad_norm": 1.759088158607483,
      "learning_rate": 1.7656258576807798e-06,
      "loss": 0.4604,
      "step": 293340
    },
    {
      "epoch": 120.08,
      "grad_norm": 1.9984662532806396,
      "learning_rate": 1.765472131033192e-06,
      "loss": 0.4637,
      "step": 293350
    },
    {
      "epoch": 120.08,
      "grad_norm": 1.945088267326355,
      "learning_rate": 1.7653184070346058e-06,
      "loss": 0.4529,
      "step": 293360
    },
    {
      "epoch": 120.09,
      "grad_norm": 1.8114709854125977,
      "learning_rate": 1.7651646856857246e-06,
      "loss": 0.4472,
      "step": 293370
    },
    {
      "epoch": 120.09,
      "grad_norm": 1.8396071195602417,
      "learning_rate": 1.7650109669872533e-06,
      "loss": 0.4539,
      "step": 293380
    },
    {
      "epoch": 120.09,
      "grad_norm": 2.38088059425354,
      "learning_rate": 1.7648572509398959e-06,
      "loss": 0.4666,
      "step": 293390
    },
    {
      "epoch": 120.1,
      "grad_norm": 2.1419947147369385,
      "learning_rate": 1.7647035375443561e-06,
      "loss": 0.4709,
      "step": 293400
    },
    {
      "epoch": 120.1,
      "grad_norm": 1.7773494720458984,
      "learning_rate": 1.764549826801339e-06,
      "loss": 0.4728,
      "step": 293410
    },
    {
      "epoch": 120.11,
      "grad_norm": 1.7516151666641235,
      "learning_rate": 1.764396118711548e-06,
      "loss": 0.4722,
      "step": 293420
    },
    {
      "epoch": 120.11,
      "grad_norm": 18.980165481567383,
      "learning_rate": 1.7642424132756876e-06,
      "loss": 0.4753,
      "step": 293430
    },
    {
      "epoch": 120.11,
      "grad_norm": 1.4983853101730347,
      "learning_rate": 1.7640887104944615e-06,
      "loss": 0.4701,
      "step": 293440
    },
    {
      "epoch": 120.12,
      "grad_norm": 1.9354923963546753,
      "learning_rate": 1.7639350103685744e-06,
      "loss": 0.4569,
      "step": 293450
    },
    {
      "epoch": 120.12,
      "grad_norm": 2.138779640197754,
      "learning_rate": 1.7637813128987296e-06,
      "loss": 0.45,
      "step": 293460
    },
    {
      "epoch": 120.13,
      "grad_norm": 1.759704828262329,
      "learning_rate": 1.7636276180856316e-06,
      "loss": 0.4454,
      "step": 293470
    },
    {
      "epoch": 120.13,
      "grad_norm": 2.0986452102661133,
      "learning_rate": 1.7634739259299843e-06,
      "loss": 0.4425,
      "step": 293480
    },
    {
      "epoch": 120.14,
      "grad_norm": 1.8014249801635742,
      "learning_rate": 1.7633202364324906e-06,
      "loss": 0.4823,
      "step": 293490
    },
    {
      "epoch": 120.14,
      "grad_norm": 1.7260152101516724,
      "learning_rate": 1.7631665495938578e-06,
      "loss": 0.4393,
      "step": 293500
    },
    {
      "epoch": 120.14,
      "grad_norm": 1.9694912433624268,
      "learning_rate": 1.7630128654147868e-06,
      "loss": 0.4687,
      "step": 293510
    },
    {
      "epoch": 120.15,
      "grad_norm": 2.5319015979766846,
      "learning_rate": 1.7628591838959831e-06,
      "loss": 0.4661,
      "step": 293520
    },
    {
      "epoch": 120.15,
      "grad_norm": 1.868556022644043,
      "learning_rate": 1.76270550503815e-06,
      "loss": 0.4856,
      "step": 293530
    },
    {
      "epoch": 120.16,
      "grad_norm": 1.8061168193817139,
      "learning_rate": 1.762551828841992e-06,
      "loss": 0.4619,
      "step": 293540
    },
    {
      "epoch": 120.16,
      "grad_norm": 2.0645816326141357,
      "learning_rate": 1.7623981553082127e-06,
      "loss": 0.428,
      "step": 293550
    },
    {
      "epoch": 120.16,
      "grad_norm": 2.076442003250122,
      "learning_rate": 1.7622444844375159e-06,
      "loss": 0.4535,
      "step": 293560
    },
    {
      "epoch": 120.17,
      "grad_norm": 1.8195463418960571,
      "learning_rate": 1.7620908162306062e-06,
      "loss": 0.4544,
      "step": 293570
    },
    {
      "epoch": 120.17,
      "grad_norm": 1.9503552913665771,
      "learning_rate": 1.7619371506881875e-06,
      "loss": 0.4603,
      "step": 293580
    },
    {
      "epoch": 120.18,
      "grad_norm": 2.1349081993103027,
      "learning_rate": 1.7617834878109626e-06,
      "loss": 0.4533,
      "step": 293590
    },
    {
      "epoch": 120.18,
      "grad_norm": 2.111964464187622,
      "learning_rate": 1.7616298275996365e-06,
      "loss": 0.4493,
      "step": 293600
    },
    {
      "epoch": 120.18,
      "grad_norm": 1.881105661392212,
      "learning_rate": 1.7614761700549129e-06,
      "loss": 0.4671,
      "step": 293610
    },
    {
      "epoch": 120.19,
      "grad_norm": 2.633467197418213,
      "learning_rate": 1.761322515177495e-06,
      "loss": 0.4628,
      "step": 293620
    },
    {
      "epoch": 120.19,
      "grad_norm": 2.22471022605896,
      "learning_rate": 1.7611688629680876e-06,
      "loss": 0.4667,
      "step": 293630
    },
    {
      "epoch": 120.2,
      "grad_norm": 2.369250774383545,
      "learning_rate": 1.7610152134273942e-06,
      "loss": 0.455,
      "step": 293640
    },
    {
      "epoch": 120.2,
      "grad_norm": 1.6764723062515259,
      "learning_rate": 1.7608615665561183e-06,
      "loss": 0.4732,
      "step": 293650
    },
    {
      "epoch": 120.2,
      "grad_norm": 1.996880292892456,
      "learning_rate": 1.7607079223549643e-06,
      "loss": 0.4617,
      "step": 293660
    },
    {
      "epoch": 120.21,
      "grad_norm": 1.9039726257324219,
      "learning_rate": 1.7605542808246362e-06,
      "loss": 0.4623,
      "step": 293670
    },
    {
      "epoch": 120.21,
      "grad_norm": 2.4650471210479736,
      "learning_rate": 1.7604006419658359e-06,
      "loss": 0.4772,
      "step": 293680
    },
    {
      "epoch": 120.22,
      "grad_norm": 1.6656723022460938,
      "learning_rate": 1.7602470057792695e-06,
      "loss": 0.4639,
      "step": 293690
    },
    {
      "epoch": 120.22,
      "grad_norm": 2.0596957206726074,
      "learning_rate": 1.76009337226564e-06,
      "loss": 0.4672,
      "step": 293700
    },
    {
      "epoch": 120.23,
      "grad_norm": 2.2221858501434326,
      "learning_rate": 1.7599397414256513e-06,
      "loss": 0.4705,
      "step": 293710
    },
    {
      "epoch": 120.23,
      "grad_norm": 1.766485571861267,
      "learning_rate": 1.7597861132600076e-06,
      "loss": 0.4635,
      "step": 293720
    },
    {
      "epoch": 120.23,
      "grad_norm": 2.658720016479492,
      "learning_rate": 1.7596324877694111e-06,
      "loss": 0.4742,
      "step": 293730
    },
    {
      "epoch": 120.24,
      "grad_norm": 2.048490047454834,
      "learning_rate": 1.7594788649545666e-06,
      "loss": 0.4534,
      "step": 293740
    },
    {
      "epoch": 120.24,
      "grad_norm": 1.7241302728652954,
      "learning_rate": 1.7593252448161778e-06,
      "loss": 0.4551,
      "step": 293750
    },
    {
      "epoch": 120.25,
      "grad_norm": 1.6977472305297852,
      "learning_rate": 1.759171627354948e-06,
      "loss": 0.4666,
      "step": 293760
    },
    {
      "epoch": 120.25,
      "grad_norm": 1.5770025253295898,
      "learning_rate": 1.7590180125715816e-06,
      "loss": 0.4637,
      "step": 293770
    },
    {
      "epoch": 120.25,
      "grad_norm": 1.8210850954055786,
      "learning_rate": 1.7588644004667817e-06,
      "loss": 0.4631,
      "step": 293780
    },
    {
      "epoch": 120.26,
      "grad_norm": 1.9155426025390625,
      "learning_rate": 1.7587107910412523e-06,
      "loss": 0.4847,
      "step": 293790
    },
    {
      "epoch": 120.26,
      "grad_norm": 1.7620525360107422,
      "learning_rate": 1.7585571842956968e-06,
      "loss": 0.4502,
      "step": 293800
    },
    {
      "epoch": 120.27,
      "grad_norm": 1.8305503129959106,
      "learning_rate": 1.7584035802308193e-06,
      "loss": 0.4596,
      "step": 293810
    },
    {
      "epoch": 120.27,
      "grad_norm": 2.1909701824188232,
      "learning_rate": 1.7582499788473226e-06,
      "loss": 0.4471,
      "step": 293820
    },
    {
      "epoch": 120.27,
      "grad_norm": 1.5503413677215576,
      "learning_rate": 1.7580963801459111e-06,
      "loss": 0.4663,
      "step": 293830
    },
    {
      "epoch": 120.28,
      "grad_norm": 1.9305899143218994,
      "learning_rate": 1.757942784127288e-06,
      "loss": 0.4581,
      "step": 293840
    },
    {
      "epoch": 120.28,
      "grad_norm": 1.8962938785552979,
      "learning_rate": 1.7577891907921567e-06,
      "loss": 0.4476,
      "step": 293850
    },
    {
      "epoch": 120.29,
      "grad_norm": 1.7567294836044312,
      "learning_rate": 1.7576356001412212e-06,
      "loss": 0.4545,
      "step": 293860
    },
    {
      "epoch": 120.29,
      "grad_norm": 1.8452869653701782,
      "learning_rate": 1.7574820121751853e-06,
      "loss": 0.4586,
      "step": 293870
    },
    {
      "epoch": 120.29,
      "grad_norm": 2.1722288131713867,
      "learning_rate": 1.7573284268947521e-06,
      "loss": 0.446,
      "step": 293880
    },
    {
      "epoch": 120.3,
      "grad_norm": 2.0533688068389893,
      "learning_rate": 1.7571748443006254e-06,
      "loss": 0.4579,
      "step": 293890
    },
    {
      "epoch": 120.3,
      "grad_norm": 1.7994602918624878,
      "learning_rate": 1.7570212643935088e-06,
      "loss": 0.4489,
      "step": 293900
    },
    {
      "epoch": 120.31,
      "grad_norm": 1.8733347654342651,
      "learning_rate": 1.7568676871741051e-06,
      "loss": 0.4489,
      "step": 293910
    },
    {
      "epoch": 120.31,
      "grad_norm": 1.894155740737915,
      "learning_rate": 1.756714112643119e-06,
      "loss": 0.4564,
      "step": 293920
    },
    {
      "epoch": 120.32,
      "grad_norm": 1.9487457275390625,
      "learning_rate": 1.756560540801253e-06,
      "loss": 0.456,
      "step": 293930
    },
    {
      "epoch": 120.32,
      "grad_norm": 2.7797749042510986,
      "learning_rate": 1.7564069716492118e-06,
      "loss": 0.4463,
      "step": 293940
    },
    {
      "epoch": 120.32,
      "grad_norm": 2.2086904048919678,
      "learning_rate": 1.7562534051876975e-06,
      "loss": 0.4714,
      "step": 293950
    },
    {
      "epoch": 120.33,
      "grad_norm": 1.827489972114563,
      "learning_rate": 1.756099841417414e-06,
      "loss": 0.4774,
      "step": 293960
    },
    {
      "epoch": 120.33,
      "grad_norm": 2.1018636226654053,
      "learning_rate": 1.7559462803390647e-06,
      "loss": 0.4711,
      "step": 293970
    },
    {
      "epoch": 120.34,
      "grad_norm": 1.8978344202041626,
      "learning_rate": 1.7557927219533535e-06,
      "loss": 0.4382,
      "step": 293980
    },
    {
      "epoch": 120.34,
      "grad_norm": 2.050375461578369,
      "learning_rate": 1.7556391662609837e-06,
      "loss": 0.4739,
      "step": 293990
    },
    {
      "epoch": 120.34,
      "grad_norm": 1.7135052680969238,
      "learning_rate": 1.755485613262658e-06,
      "loss": 0.4667,
      "step": 294000
    },
    {
      "epoch": 120.35,
      "grad_norm": 1.7877775430679321,
      "learning_rate": 1.7553320629590806e-06,
      "loss": 0.4658,
      "step": 294010
    },
    {
      "epoch": 120.35,
      "grad_norm": 1.6237047910690308,
      "learning_rate": 1.7551785153509548e-06,
      "loss": 0.456,
      "step": 294020
    },
    {
      "epoch": 120.36,
      "grad_norm": 2.022284507751465,
      "learning_rate": 1.7550249704389842e-06,
      "loss": 0.4751,
      "step": 294030
    },
    {
      "epoch": 120.36,
      "grad_norm": 1.8935654163360596,
      "learning_rate": 1.7548714282238713e-06,
      "loss": 0.4707,
      "step": 294040
    },
    {
      "epoch": 120.36,
      "grad_norm": 1.8995089530944824,
      "learning_rate": 1.7547178887063187e-06,
      "loss": 0.4623,
      "step": 294050
    },
    {
      "epoch": 120.37,
      "grad_norm": 2.0128109455108643,
      "learning_rate": 1.7545643518870322e-06,
      "loss": 0.4733,
      "step": 294060
    },
    {
      "epoch": 120.37,
      "grad_norm": 1.8193385601043701,
      "learning_rate": 1.754410817766714e-06,
      "loss": 0.4659,
      "step": 294070
    },
    {
      "epoch": 120.38,
      "grad_norm": 1.8870911598205566,
      "learning_rate": 1.7542572863460675e-06,
      "loss": 0.4694,
      "step": 294080
    },
    {
      "epoch": 120.38,
      "grad_norm": 1.7875826358795166,
      "learning_rate": 1.7541037576257958e-06,
      "loss": 0.4577,
      "step": 294090
    },
    {
      "epoch": 120.38,
      "grad_norm": 2.2470152378082275,
      "learning_rate": 1.7539502316066018e-06,
      "loss": 0.4642,
      "step": 294100
    },
    {
      "epoch": 120.39,
      "grad_norm": 1.9563783407211304,
      "learning_rate": 1.7537967082891893e-06,
      "loss": 0.4556,
      "step": 294110
    },
    {
      "epoch": 120.39,
      "grad_norm": 1.7456464767456055,
      "learning_rate": 1.7536431876742616e-06,
      "loss": 0.4498,
      "step": 294120
    },
    {
      "epoch": 120.4,
      "grad_norm": 2.353372812271118,
      "learning_rate": 1.7534896697625219e-06,
      "loss": 0.4591,
      "step": 294130
    },
    {
      "epoch": 120.4,
      "grad_norm": 2.865617036819458,
      "learning_rate": 1.753336154554673e-06,
      "loss": 0.4561,
      "step": 294140
    },
    {
      "epoch": 120.41,
      "grad_norm": 2.5500969886779785,
      "learning_rate": 1.7531826420514189e-06,
      "loss": 0.4581,
      "step": 294150
    },
    {
      "epoch": 120.41,
      "grad_norm": 1.7881397008895874,
      "learning_rate": 1.7530291322534623e-06,
      "loss": 0.4772,
      "step": 294160
    },
    {
      "epoch": 120.41,
      "grad_norm": 2.225015640258789,
      "learning_rate": 1.7528756251615067e-06,
      "loss": 0.4768,
      "step": 294170
    },
    {
      "epoch": 120.42,
      "grad_norm": 2.00740385055542,
      "learning_rate": 1.7527221207762548e-06,
      "loss": 0.4772,
      "step": 294180
    },
    {
      "epoch": 120.42,
      "grad_norm": 1.7911062240600586,
      "learning_rate": 1.7525686190984101e-06,
      "loss": 0.449,
      "step": 294190
    },
    {
      "epoch": 120.43,
      "grad_norm": 1.9023072719573975,
      "learning_rate": 1.7524151201286757e-06,
      "loss": 0.4503,
      "step": 294200
    },
    {
      "epoch": 120.43,
      "grad_norm": 1.591179609298706,
      "learning_rate": 1.752261623867755e-06,
      "loss": 0.4649,
      "step": 294210
    },
    {
      "epoch": 120.43,
      "grad_norm": 2.154480218887329,
      "learning_rate": 1.7521081303163505e-06,
      "loss": 0.4613,
      "step": 294220
    },
    {
      "epoch": 120.44,
      "grad_norm": 2.222134590148926,
      "learning_rate": 1.7519546394751664e-06,
      "loss": 0.4745,
      "step": 294230
    },
    {
      "epoch": 120.44,
      "grad_norm": 2.604940176010132,
      "learning_rate": 1.7518011513449041e-06,
      "loss": 0.4553,
      "step": 294240
    },
    {
      "epoch": 120.45,
      "grad_norm": 2.1767547130584717,
      "learning_rate": 1.7516476659262685e-06,
      "loss": 0.4531,
      "step": 294250
    },
    {
      "epoch": 120.45,
      "grad_norm": 1.716939091682434,
      "learning_rate": 1.7514941832199622e-06,
      "loss": 0.4724,
      "step": 294260
    },
    {
      "epoch": 120.45,
      "grad_norm": 2.0875773429870605,
      "learning_rate": 1.7513407032266878e-06,
      "loss": 0.4744,
      "step": 294270
    },
    {
      "epoch": 120.46,
      "grad_norm": 1.8141719102859497,
      "learning_rate": 1.7511872259471488e-06,
      "loss": 0.4598,
      "step": 294280
    },
    {
      "epoch": 120.46,
      "grad_norm": 1.6961476802825928,
      "learning_rate": 1.7510337513820483e-06,
      "loss": 0.4623,
      "step": 294290
    },
    {
      "epoch": 120.47,
      "grad_norm": 2.143362283706665,
      "learning_rate": 1.7508802795320893e-06,
      "loss": 0.4392,
      "step": 294300
    },
    {
      "epoch": 120.47,
      "grad_norm": 1.7882156372070312,
      "learning_rate": 1.7507268103979743e-06,
      "loss": 0.4607,
      "step": 294310
    },
    {
      "epoch": 120.47,
      "grad_norm": 2.3942484855651855,
      "learning_rate": 1.7505733439804067e-06,
      "loss": 0.4546,
      "step": 294320
    },
    {
      "epoch": 120.48,
      "grad_norm": 2.0992393493652344,
      "learning_rate": 1.7504198802800898e-06,
      "loss": 0.4476,
      "step": 294330
    },
    {
      "epoch": 120.48,
      "grad_norm": 1.638899326324463,
      "learning_rate": 1.7502664192977258e-06,
      "loss": 0.4408,
      "step": 294340
    },
    {
      "epoch": 120.49,
      "grad_norm": 2.0314228534698486,
      "learning_rate": 1.7501129610340182e-06,
      "loss": 0.4635,
      "step": 294350
    },
    {
      "epoch": 120.49,
      "grad_norm": 3.618753671646118,
      "learning_rate": 1.7499595054896703e-06,
      "loss": 0.4589,
      "step": 294360
    },
    {
      "epoch": 120.5,
      "grad_norm": 1.9377986192703247,
      "learning_rate": 1.7498060526653843e-06,
      "loss": 0.4528,
      "step": 294370
    },
    {
      "epoch": 120.5,
      "grad_norm": 2.4406774044036865,
      "learning_rate": 1.7496526025618636e-06,
      "loss": 0.4482,
      "step": 294380
    },
    {
      "epoch": 120.5,
      "grad_norm": 1.7489076852798462,
      "learning_rate": 1.7494991551798117e-06,
      "loss": 0.4883,
      "step": 294390
    },
    {
      "epoch": 120.51,
      "grad_norm": 1.8419959545135498,
      "learning_rate": 1.7493457105199306e-06,
      "loss": 0.4658,
      "step": 294400
    },
    {
      "epoch": 120.51,
      "grad_norm": 2.091170072555542,
      "learning_rate": 1.7491922685829232e-06,
      "loss": 0.4494,
      "step": 294410
    },
    {
      "epoch": 120.52,
      "grad_norm": 2.186210870742798,
      "learning_rate": 1.7490388293694927e-06,
      "loss": 0.4541,
      "step": 294420
    },
    {
      "epoch": 120.52,
      "grad_norm": 1.7304749488830566,
      "learning_rate": 1.748885392880341e-06,
      "loss": 0.4488,
      "step": 294430
    },
    {
      "epoch": 120.52,
      "grad_norm": 1.8653677701950073,
      "learning_rate": 1.7487319591161734e-06,
      "loss": 0.4436,
      "step": 294440
    },
    {
      "epoch": 120.53,
      "grad_norm": 2.0555648803710938,
      "learning_rate": 1.748578528077691e-06,
      "loss": 0.4755,
      "step": 294450
    },
    {
      "epoch": 120.53,
      "grad_norm": 1.9672443866729736,
      "learning_rate": 1.7484250997655968e-06,
      "loss": 0.4596,
      "step": 294460
    },
    {
      "epoch": 120.54,
      "grad_norm": 1.793111801147461,
      "learning_rate": 1.7482716741805938e-06,
      "loss": 0.4901,
      "step": 294470
    },
    {
      "epoch": 120.54,
      "grad_norm": 1.6786656379699707,
      "learning_rate": 1.7481182513233848e-06,
      "loss": 0.4557,
      "step": 294480
    },
    {
      "epoch": 120.54,
      "grad_norm": 1.716835379600525,
      "learning_rate": 1.7479648311946726e-06,
      "loss": 0.464,
      "step": 294490
    },
    {
      "epoch": 120.55,
      "grad_norm": 2.678413152694702,
      "learning_rate": 1.7478114137951598e-06,
      "loss": 0.4644,
      "step": 294500
    },
    {
      "epoch": 120.55,
      "grad_norm": 1.9599684476852417,
      "learning_rate": 1.7476579991255496e-06,
      "loss": 0.483,
      "step": 294510
    },
    {
      "epoch": 120.56,
      "grad_norm": 2.251603603363037,
      "learning_rate": 1.7475045871865448e-06,
      "loss": 0.479,
      "step": 294520
    },
    {
      "epoch": 120.56,
      "grad_norm": 2.1100518703460693,
      "learning_rate": 1.7473511779788476e-06,
      "loss": 0.4641,
      "step": 294530
    },
    {
      "epoch": 120.56,
      "grad_norm": 1.7425495386123657,
      "learning_rate": 1.7471977715031611e-06,
      "loss": 0.4643,
      "step": 294540
    },
    {
      "epoch": 120.57,
      "grad_norm": 1.7155736684799194,
      "learning_rate": 1.7470443677601878e-06,
      "loss": 0.4534,
      "step": 294550
    },
    {
      "epoch": 120.57,
      "grad_norm": 2.123203754425049,
      "learning_rate": 1.7468909667506304e-06,
      "loss": 0.4463,
      "step": 294560
    },
    {
      "epoch": 120.58,
      "grad_norm": 2.169546604156494,
      "learning_rate": 1.7467375684751923e-06,
      "loss": 0.464,
      "step": 294570
    },
    {
      "epoch": 120.58,
      "grad_norm": 1.9990488290786743,
      "learning_rate": 1.7465841729345754e-06,
      "loss": 0.4779,
      "step": 294580
    },
    {
      "epoch": 120.59,
      "grad_norm": 1.9558402299880981,
      "learning_rate": 1.7464307801294827e-06,
      "loss": 0.4655,
      "step": 294590
    },
    {
      "epoch": 120.59,
      "grad_norm": 4.8895792961120605,
      "learning_rate": 1.7462773900606167e-06,
      "loss": 0.4561,
      "step": 294600
    },
    {
      "epoch": 120.59,
      "grad_norm": 2.184314489364624,
      "learning_rate": 1.7461240027286812e-06,
      "loss": 0.4363,
      "step": 294610
    },
    {
      "epoch": 120.6,
      "grad_norm": 2.1789393424987793,
      "learning_rate": 1.7459706181343757e-06,
      "loss": 0.4555,
      "step": 294620
    },
    {
      "epoch": 120.6,
      "grad_norm": 1.699623465538025,
      "learning_rate": 1.7458172362784063e-06,
      "loss": 0.4591,
      "step": 294630
    },
    {
      "epoch": 120.61,
      "grad_norm": 2.0103025436401367,
      "learning_rate": 1.7456638571614743e-06,
      "loss": 0.4534,
      "step": 294640
    },
    {
      "epoch": 120.61,
      "grad_norm": 2.4577531814575195,
      "learning_rate": 1.7455104807842822e-06,
      "loss": 0.4804,
      "step": 294650
    },
    {
      "epoch": 120.61,
      "grad_norm": 1.9800456762313843,
      "learning_rate": 1.7453571071475332e-06,
      "loss": 0.452,
      "step": 294660
    },
    {
      "epoch": 120.62,
      "grad_norm": 1.9812252521514893,
      "learning_rate": 1.7452037362519288e-06,
      "loss": 0.4456,
      "step": 294670
    },
    {
      "epoch": 120.62,
      "grad_norm": 1.849839210510254,
      "learning_rate": 1.7450503680981722e-06,
      "loss": 0.4692,
      "step": 294680
    },
    {
      "epoch": 120.63,
      "grad_norm": 1.6787923574447632,
      "learning_rate": 1.744897002686966e-06,
      "loss": 0.4478,
      "step": 294690
    },
    {
      "epoch": 120.63,
      "grad_norm": 2.319322109222412,
      "learning_rate": 1.7447436400190124e-06,
      "loss": 0.4732,
      "step": 294700
    },
    {
      "epoch": 120.63,
      "grad_norm": 1.6072016954421997,
      "learning_rate": 1.7445902800950143e-06,
      "loss": 0.4581,
      "step": 294710
    },
    {
      "epoch": 120.64,
      "grad_norm": 2.223646640777588,
      "learning_rate": 1.7444369229156745e-06,
      "loss": 0.4666,
      "step": 294720
    },
    {
      "epoch": 120.64,
      "grad_norm": 2.0396993160247803,
      "learning_rate": 1.7442835684816946e-06,
      "loss": 0.465,
      "step": 294730
    },
    {
      "epoch": 120.65,
      "grad_norm": 1.9886289834976196,
      "learning_rate": 1.7441302167937777e-06,
      "loss": 0.4622,
      "step": 294740
    },
    {
      "epoch": 120.65,
      "grad_norm": 2.012803554534912,
      "learning_rate": 1.7439768678526267e-06,
      "loss": 0.4643,
      "step": 294750
    },
    {
      "epoch": 120.65,
      "grad_norm": 1.958315372467041,
      "learning_rate": 1.743823521658943e-06,
      "loss": 0.454,
      "step": 294760
    },
    {
      "epoch": 120.66,
      "grad_norm": 2.1583824157714844,
      "learning_rate": 1.7436701782134294e-06,
      "loss": 0.4566,
      "step": 294770
    },
    {
      "epoch": 120.66,
      "grad_norm": 2.046135187149048,
      "learning_rate": 1.7435168375167889e-06,
      "loss": 0.4637,
      "step": 294780
    },
    {
      "epoch": 120.67,
      "grad_norm": 2.283891201019287,
      "learning_rate": 1.7433634995697232e-06,
      "loss": 0.4688,
      "step": 294790
    },
    {
      "epoch": 120.67,
      "grad_norm": 1.9539159536361694,
      "learning_rate": 1.7432101643729352e-06,
      "loss": 0.4606,
      "step": 294800
    },
    {
      "epoch": 120.68,
      "grad_norm": 2.356865644454956,
      "learning_rate": 1.7430568319271268e-06,
      "loss": 0.4627,
      "step": 294810
    },
    {
      "epoch": 120.68,
      "grad_norm": 1.8203152418136597,
      "learning_rate": 1.7429035022330013e-06,
      "loss": 0.4753,
      "step": 294820
    },
    {
      "epoch": 120.68,
      "grad_norm": 1.6972277164459229,
      "learning_rate": 1.7427501752912602e-06,
      "loss": 0.4627,
      "step": 294830
    },
    {
      "epoch": 120.69,
      "grad_norm": 2.0869641304016113,
      "learning_rate": 1.7425968511026066e-06,
      "loss": 0.4425,
      "step": 294840
    },
    {
      "epoch": 120.69,
      "grad_norm": 1.8710846900939941,
      "learning_rate": 1.7424435296677422e-06,
      "loss": 0.4643,
      "step": 294850
    },
    {
      "epoch": 120.7,
      "grad_norm": 1.610496163368225,
      "learning_rate": 1.7422902109873699e-06,
      "loss": 0.4694,
      "step": 294860
    },
    {
      "epoch": 120.7,
      "grad_norm": 1.7526830434799194,
      "learning_rate": 1.7421368950621916e-06,
      "loss": 0.4456,
      "step": 294870
    },
    {
      "epoch": 120.7,
      "grad_norm": 2.4287171363830566,
      "learning_rate": 1.7419835818929104e-06,
      "loss": 0.4633,
      "step": 294880
    },
    {
      "epoch": 120.71,
      "grad_norm": 1.9606813192367554,
      "learning_rate": 1.7418302714802272e-06,
      "loss": 0.4609,
      "step": 294890
    },
    {
      "epoch": 120.71,
      "grad_norm": 1.8942228555679321,
      "learning_rate": 1.741676963824845e-06,
      "loss": 0.4696,
      "step": 294900
    },
    {
      "epoch": 120.72,
      "grad_norm": 1.76605224609375,
      "learning_rate": 1.7415236589274666e-06,
      "loss": 0.4574,
      "step": 294910
    },
    {
      "epoch": 120.72,
      "grad_norm": 1.6400043964385986,
      "learning_rate": 1.7413703567887935e-06,
      "loss": 0.4548,
      "step": 294920
    },
    {
      "epoch": 120.72,
      "grad_norm": 1.7209482192993164,
      "learning_rate": 1.7412170574095282e-06,
      "loss": 0.4681,
      "step": 294930
    },
    {
      "epoch": 120.73,
      "grad_norm": 2.1503515243530273,
      "learning_rate": 1.741063760790373e-06,
      "loss": 0.474,
      "step": 294940
    },
    {
      "epoch": 120.73,
      "grad_norm": 1.6449835300445557,
      "learning_rate": 1.7409104669320304e-06,
      "loss": 0.4473,
      "step": 294950
    },
    {
      "epoch": 120.74,
      "grad_norm": 1.9915556907653809,
      "learning_rate": 1.7407571758352021e-06,
      "loss": 0.451,
      "step": 294960
    },
    {
      "epoch": 120.74,
      "grad_norm": 2.134638786315918,
      "learning_rate": 1.740603887500591e-06,
      "loss": 0.4442,
      "step": 294970
    },
    {
      "epoch": 120.74,
      "grad_norm": 1.9369131326675415,
      "learning_rate": 1.7404506019288986e-06,
      "loss": 0.4635,
      "step": 294980
    },
    {
      "epoch": 120.75,
      "grad_norm": 1.9126416444778442,
      "learning_rate": 1.740297319120827e-06,
      "loss": 0.4579,
      "step": 294990
    },
    {
      "epoch": 120.75,
      "grad_norm": 1.8600918054580688,
      "learning_rate": 1.740144039077078e-06,
      "loss": 0.4565,
      "step": 295000
    },
    {
      "epoch": 120.76,
      "grad_norm": 2.005518913269043,
      "learning_rate": 1.739990761798355e-06,
      "loss": 0.4544,
      "step": 295010
    },
    {
      "epoch": 120.76,
      "grad_norm": 1.8151293992996216,
      "learning_rate": 1.73983748728536e-06,
      "loss": 0.4608,
      "step": 295020
    },
    {
      "epoch": 120.77,
      "grad_norm": 2.2371466159820557,
      "learning_rate": 1.7396842155387948e-06,
      "loss": 0.4814,
      "step": 295030
    },
    {
      "epoch": 120.77,
      "grad_norm": 1.9505128860473633,
      "learning_rate": 1.739530946559361e-06,
      "loss": 0.4658,
      "step": 295040
    },
    {
      "epoch": 120.77,
      "grad_norm": 1.7922974824905396,
      "learning_rate": 1.7393776803477612e-06,
      "loss": 0.4616,
      "step": 295050
    },
    {
      "epoch": 120.78,
      "grad_norm": 2.144681215286255,
      "learning_rate": 1.7392244169046973e-06,
      "loss": 0.4571,
      "step": 295060
    },
    {
      "epoch": 120.78,
      "grad_norm": 2.01554799079895,
      "learning_rate": 1.7390711562308717e-06,
      "loss": 0.4579,
      "step": 295070
    },
    {
      "epoch": 120.79,
      "grad_norm": 1.8818352222442627,
      "learning_rate": 1.7389178983269862e-06,
      "loss": 0.4463,
      "step": 295080
    },
    {
      "epoch": 120.79,
      "grad_norm": 1.9185729026794434,
      "learning_rate": 1.7387646431937428e-06,
      "loss": 0.4564,
      "step": 295090
    },
    {
      "epoch": 120.79,
      "grad_norm": 2.2687978744506836,
      "learning_rate": 1.7386113908318435e-06,
      "loss": 0.4685,
      "step": 295100
    },
    {
      "epoch": 120.8,
      "grad_norm": 1.8822473287582397,
      "learning_rate": 1.738458141241991e-06,
      "loss": 0.4803,
      "step": 295110
    },
    {
      "epoch": 120.8,
      "grad_norm": 1.6726480722427368,
      "learning_rate": 1.7383048944248867e-06,
      "loss": 0.4587,
      "step": 295120
    },
    {
      "epoch": 120.81,
      "grad_norm": 2.5911471843719482,
      "learning_rate": 1.738151650381232e-06,
      "loss": 0.4605,
      "step": 295130
    },
    {
      "epoch": 120.81,
      "grad_norm": 2.3071706295013428,
      "learning_rate": 1.7379984091117302e-06,
      "loss": 0.5017,
      "step": 295140
    },
    {
      "epoch": 120.81,
      "grad_norm": 1.822914481163025,
      "learning_rate": 1.737845170617082e-06,
      "loss": 0.4326,
      "step": 295150
    },
    {
      "epoch": 120.82,
      "grad_norm": 1.541848063468933,
      "learning_rate": 1.73769193489799e-06,
      "loss": 0.457,
      "step": 295160
    },
    {
      "epoch": 120.82,
      "grad_norm": 2.014514446258545,
      "learning_rate": 1.737538701955157e-06,
      "loss": 0.4603,
      "step": 295170
    },
    {
      "epoch": 120.83,
      "grad_norm": 1.9699902534484863,
      "learning_rate": 1.7373854717892833e-06,
      "loss": 0.4704,
      "step": 295180
    },
    {
      "epoch": 120.83,
      "grad_norm": 2.210106134414673,
      "learning_rate": 1.7372322444010713e-06,
      "loss": 0.4518,
      "step": 295190
    },
    {
      "epoch": 120.84,
      "grad_norm": 2.192924976348877,
      "learning_rate": 1.7370790197912235e-06,
      "loss": 0.4559,
      "step": 295200
    },
    {
      "epoch": 120.84,
      "grad_norm": 1.8484128713607788,
      "learning_rate": 1.7369257979604417e-06,
      "loss": 0.4636,
      "step": 295210
    },
    {
      "epoch": 120.84,
      "grad_norm": 1.8864243030548096,
      "learning_rate": 1.7367725789094276e-06,
      "loss": 0.4621,
      "step": 295220
    },
    {
      "epoch": 120.85,
      "grad_norm": 2.484971284866333,
      "learning_rate": 1.736619362638883e-06,
      "loss": 0.4674,
      "step": 295230
    },
    {
      "epoch": 120.85,
      "grad_norm": 1.9539073705673218,
      "learning_rate": 1.7364661491495104e-06,
      "loss": 0.4481,
      "step": 295240
    },
    {
      "epoch": 120.86,
      "grad_norm": 2.0354721546173096,
      "learning_rate": 1.7363129384420105e-06,
      "loss": 0.4764,
      "step": 295250
    },
    {
      "epoch": 120.86,
      "grad_norm": 1.8251686096191406,
      "learning_rate": 1.7361597305170857e-06,
      "loss": 0.4513,
      "step": 295260
    },
    {
      "epoch": 120.86,
      "grad_norm": 2.1751394271850586,
      "learning_rate": 1.736006525375438e-06,
      "loss": 0.4596,
      "step": 295270
    },
    {
      "epoch": 120.87,
      "grad_norm": 2.282078266143799,
      "learning_rate": 1.7358533230177686e-06,
      "loss": 0.487,
      "step": 295280
    },
    {
      "epoch": 120.87,
      "grad_norm": 2.386946201324463,
      "learning_rate": 1.7357001234447803e-06,
      "loss": 0.4618,
      "step": 295290
    },
    {
      "epoch": 120.88,
      "grad_norm": 1.8450608253479004,
      "learning_rate": 1.735546926657174e-06,
      "loss": 0.479,
      "step": 295300
    },
    {
      "epoch": 120.88,
      "grad_norm": 2.485553026199341,
      "learning_rate": 1.7353937326556519e-06,
      "loss": 0.4588,
      "step": 295310
    },
    {
      "epoch": 120.88,
      "grad_norm": 1.958103895187378,
      "learning_rate": 1.7352405414409155e-06,
      "loss": 0.4691,
      "step": 295320
    },
    {
      "epoch": 120.89,
      "grad_norm": 1.944962739944458,
      "learning_rate": 1.7350873530136676e-06,
      "loss": 0.4655,
      "step": 295330
    },
    {
      "epoch": 120.89,
      "grad_norm": 1.5821093320846558,
      "learning_rate": 1.734934167374608e-06,
      "loss": 0.4616,
      "step": 295340
    },
    {
      "epoch": 120.9,
      "grad_norm": 2.17606258392334,
      "learning_rate": 1.7347809845244395e-06,
      "loss": 0.4625,
      "step": 295350
    },
    {
      "epoch": 120.9,
      "grad_norm": 2.150973081588745,
      "learning_rate": 1.7346278044638642e-06,
      "loss": 0.456,
      "step": 295360
    },
    {
      "epoch": 120.9,
      "grad_norm": 1.7863867282867432,
      "learning_rate": 1.7344746271935818e-06,
      "loss": 0.4814,
      "step": 295370
    },
    {
      "epoch": 120.91,
      "grad_norm": 2.190098524093628,
      "learning_rate": 1.7343214527142975e-06,
      "loss": 0.4633,
      "step": 295380
    },
    {
      "epoch": 120.91,
      "grad_norm": 1.9673875570297241,
      "learning_rate": 1.73416828102671e-06,
      "loss": 0.4555,
      "step": 295390
    },
    {
      "epoch": 120.92,
      "grad_norm": 1.7793407440185547,
      "learning_rate": 1.7340151121315222e-06,
      "loss": 0.4563,
      "step": 295400
    },
    {
      "epoch": 120.92,
      "grad_norm": 2.09137225151062,
      "learning_rate": 1.7338619460294355e-06,
      "loss": 0.4465,
      "step": 295410
    },
    {
      "epoch": 120.93,
      "grad_norm": 2.016692638397217,
      "learning_rate": 1.7337087827211517e-06,
      "loss": 0.4465,
      "step": 295420
    },
    {
      "epoch": 120.93,
      "grad_norm": 1.9378687143325806,
      "learning_rate": 1.7335556222073718e-06,
      "loss": 0.4437,
      "step": 295430
    },
    {
      "epoch": 120.93,
      "grad_norm": 1.5893453359603882,
      "learning_rate": 1.7334024644887984e-06,
      "loss": 0.4773,
      "step": 295440
    },
    {
      "epoch": 120.94,
      "grad_norm": 2.046286106109619,
      "learning_rate": 1.7332493095661319e-06,
      "loss": 0.4629,
      "step": 295450
    },
    {
      "epoch": 120.94,
      "grad_norm": 2.8900656700134277,
      "learning_rate": 1.7330961574400748e-06,
      "loss": 0.4584,
      "step": 295460
    },
    {
      "epoch": 120.95,
      "grad_norm": 1.7515785694122314,
      "learning_rate": 1.732943008111329e-06,
      "loss": 0.464,
      "step": 295470
    },
    {
      "epoch": 120.95,
      "grad_norm": 1.9356938600540161,
      "learning_rate": 1.7327898615805948e-06,
      "loss": 0.4729,
      "step": 295480
    },
    {
      "epoch": 120.95,
      "grad_norm": 2.015941858291626,
      "learning_rate": 1.7326367178485746e-06,
      "loss": 0.4574,
      "step": 295490
    },
    {
      "epoch": 120.96,
      "grad_norm": 1.7969156503677368,
      "learning_rate": 1.7324835769159696e-06,
      "loss": 0.4644,
      "step": 295500
    },
    {
      "epoch": 120.96,
      "grad_norm": 1.8423631191253662,
      "learning_rate": 1.7323304387834813e-06,
      "loss": 0.4664,
      "step": 295510
    },
    {
      "epoch": 120.97,
      "grad_norm": 2.7325119972229004,
      "learning_rate": 1.7321773034518115e-06,
      "loss": 0.4777,
      "step": 295520
    },
    {
      "epoch": 120.97,
      "grad_norm": 2.005824565887451,
      "learning_rate": 1.7320241709216611e-06,
      "loss": 0.4837,
      "step": 295530
    },
    {
      "epoch": 120.97,
      "grad_norm": 1.9661768674850464,
      "learning_rate": 1.7318710411937325e-06,
      "loss": 0.4701,
      "step": 295540
    },
    {
      "epoch": 120.98,
      "grad_norm": 1.8262909650802612,
      "learning_rate": 1.731717914268727e-06,
      "loss": 0.4801,
      "step": 295550
    },
    {
      "epoch": 120.98,
      "grad_norm": 2.121000051498413,
      "learning_rate": 1.7315647901473438e-06,
      "loss": 0.4705,
      "step": 295560
    },
    {
      "epoch": 120.99,
      "grad_norm": 1.759798526763916,
      "learning_rate": 1.7314116688302878e-06,
      "loss": 0.4714,
      "step": 295570
    },
    {
      "epoch": 120.99,
      "grad_norm": 1.6512730121612549,
      "learning_rate": 1.7312585503182585e-06,
      "loss": 0.463,
      "step": 295580
    },
    {
      "epoch": 120.99,
      "grad_norm": 1.7020816802978516,
      "learning_rate": 1.7311054346119579e-06,
      "loss": 0.4544,
      "step": 295590
    },
    {
      "epoch": 121.0,
      "grad_norm": 1.8553985357284546,
      "learning_rate": 1.7309523217120877e-06,
      "loss": 0.4575,
      "step": 295600
    },
    {
      "epoch": 121.0,
      "eval_loss": 0.462830126285553,
      "eval_runtime": 58.9694,
      "eval_samples_per_second": 58.488,
      "eval_steps_per_second": 7.326,
      "step": 295603
    },
    {
      "epoch": 121.0,
      "grad_norm": 1.879035472869873,
      "learning_rate": 1.7307992116193477e-06,
      "loss": 0.4282,
      "step": 295610
    },
    {
      "epoch": 121.01,
      "grad_norm": 2.1642420291900635,
      "learning_rate": 1.730646104334441e-06,
      "loss": 0.4512,
      "step": 295620
    },
    {
      "epoch": 121.01,
      "grad_norm": 2.1307485103607178,
      "learning_rate": 1.7304929998580682e-06,
      "loss": 0.4716,
      "step": 295630
    },
    {
      "epoch": 121.02,
      "grad_norm": 1.6873630285263062,
      "learning_rate": 1.7303398981909304e-06,
      "loss": 0.4692,
      "step": 295640
    },
    {
      "epoch": 121.02,
      "grad_norm": 1.9126859903335571,
      "learning_rate": 1.7301867993337298e-06,
      "loss": 0.4529,
      "step": 295650
    },
    {
      "epoch": 121.02,
      "grad_norm": 1.9760732650756836,
      "learning_rate": 1.7300337032871665e-06,
      "loss": 0.4804,
      "step": 295660
    },
    {
      "epoch": 121.03,
      "grad_norm": 1.911428689956665,
      "learning_rate": 1.7298806100519431e-06,
      "loss": 0.4488,
      "step": 295670
    },
    {
      "epoch": 121.03,
      "grad_norm": 1.9481415748596191,
      "learning_rate": 1.72972751962876e-06,
      "loss": 0.457,
      "step": 295680
    },
    {
      "epoch": 121.04,
      "grad_norm": 2.270406484603882,
      "learning_rate": 1.7295744320183193e-06,
      "loss": 0.47,
      "step": 295690
    },
    {
      "epoch": 121.04,
      "grad_norm": 1.6161632537841797,
      "learning_rate": 1.7294213472213216e-06,
      "loss": 0.452,
      "step": 295700
    },
    {
      "epoch": 121.04,
      "grad_norm": 2.0934882164001465,
      "learning_rate": 1.7292682652384678e-06,
      "loss": 0.467,
      "step": 295710
    },
    {
      "epoch": 121.05,
      "grad_norm": 2.019261360168457,
      "learning_rate": 1.72911518607046e-06,
      "loss": 0.472,
      "step": 295720
    },
    {
      "epoch": 121.05,
      "grad_norm": 2.867903232574463,
      "learning_rate": 1.7289621097179989e-06,
      "loss": 0.4372,
      "step": 295730
    },
    {
      "epoch": 121.06,
      "grad_norm": 2.284233808517456,
      "learning_rate": 1.7288090361817862e-06,
      "loss": 0.4762,
      "step": 295740
    },
    {
      "epoch": 121.06,
      "grad_norm": 1.7845391035079956,
      "learning_rate": 1.7286559654625222e-06,
      "loss": 0.4678,
      "step": 295750
    },
    {
      "epoch": 121.06,
      "grad_norm": 2.0676984786987305,
      "learning_rate": 1.7285028975609093e-06,
      "loss": 0.4713,
      "step": 295760
    },
    {
      "epoch": 121.07,
      "grad_norm": 1.7310792207717896,
      "learning_rate": 1.7283498324776478e-06,
      "loss": 0.4601,
      "step": 295770
    },
    {
      "epoch": 121.07,
      "grad_norm": 1.960540771484375,
      "learning_rate": 1.7281967702134391e-06,
      "loss": 0.4657,
      "step": 295780
    },
    {
      "epoch": 121.08,
      "grad_norm": 2.1021642684936523,
      "learning_rate": 1.7280437107689848e-06,
      "loss": 0.462,
      "step": 295790
    },
    {
      "epoch": 121.08,
      "grad_norm": 2.036123037338257,
      "learning_rate": 1.7278906541449853e-06,
      "loss": 0.4722,
      "step": 295800
    },
    {
      "epoch": 121.08,
      "grad_norm": 1.7217128276824951,
      "learning_rate": 1.7277376003421425e-06,
      "loss": 0.4664,
      "step": 295810
    },
    {
      "epoch": 121.09,
      "grad_norm": 1.9363892078399658,
      "learning_rate": 1.727584549361157e-06,
      "loss": 0.4645,
      "step": 295820
    },
    {
      "epoch": 121.09,
      "grad_norm": 2.057281494140625,
      "learning_rate": 1.7274315012027303e-06,
      "loss": 0.4726,
      "step": 295830
    },
    {
      "epoch": 121.1,
      "grad_norm": 2.166130542755127,
      "learning_rate": 1.7272784558675627e-06,
      "loss": 0.4546,
      "step": 295840
    },
    {
      "epoch": 121.1,
      "grad_norm": 1.745804786682129,
      "learning_rate": 1.7271254133563558e-06,
      "loss": 0.472,
      "step": 295850
    },
    {
      "epoch": 121.11,
      "grad_norm": 2.1030938625335693,
      "learning_rate": 1.7269723736698108e-06,
      "loss": 0.4609,
      "step": 295860
    },
    {
      "epoch": 121.11,
      "grad_norm": 2.0861353874206543,
      "learning_rate": 1.7268193368086282e-06,
      "loss": 0.4895,
      "step": 295870
    },
    {
      "epoch": 121.11,
      "grad_norm": 1.9164587259292603,
      "learning_rate": 1.7266663027735097e-06,
      "loss": 0.4522,
      "step": 295880
    },
    {
      "epoch": 121.12,
      "grad_norm": 1.8833853006362915,
      "learning_rate": 1.7265132715651563e-06,
      "loss": 0.4627,
      "step": 295890
    },
    {
      "epoch": 121.12,
      "grad_norm": 1.821662425994873,
      "learning_rate": 1.7263602431842686e-06,
      "loss": 0.4703,
      "step": 295900
    },
    {
      "epoch": 121.13,
      "grad_norm": 1.9600579738616943,
      "learning_rate": 1.726207217631548e-06,
      "loss": 0.4773,
      "step": 295910
    },
    {
      "epoch": 121.13,
      "grad_norm": 2.0378799438476562,
      "learning_rate": 1.726054194907695e-06,
      "loss": 0.445,
      "step": 295920
    },
    {
      "epoch": 121.13,
      "grad_norm": 1.890614628791809,
      "learning_rate": 1.7259011750134106e-06,
      "loss": 0.4733,
      "step": 295930
    },
    {
      "epoch": 121.14,
      "grad_norm": 2.039555311203003,
      "learning_rate": 1.7257481579493953e-06,
      "loss": 0.4504,
      "step": 295940
    },
    {
      "epoch": 121.14,
      "grad_norm": 1.7805016040802002,
      "learning_rate": 1.7255951437163513e-06,
      "loss": 0.4763,
      "step": 295950
    },
    {
      "epoch": 121.15,
      "grad_norm": 1.4781426191329956,
      "learning_rate": 1.7254421323149798e-06,
      "loss": 0.4807,
      "step": 295960
    },
    {
      "epoch": 121.15,
      "grad_norm": 2.0431768894195557,
      "learning_rate": 1.7252891237459803e-06,
      "loss": 0.4593,
      "step": 295970
    },
    {
      "epoch": 121.15,
      "grad_norm": 1.7759462594985962,
      "learning_rate": 1.7251361180100544e-06,
      "loss": 0.4895,
      "step": 295980
    },
    {
      "epoch": 121.16,
      "grad_norm": 2.1373178958892822,
      "learning_rate": 1.7249831151079027e-06,
      "loss": 0.4657,
      "step": 295990
    },
    {
      "epoch": 121.16,
      "grad_norm": 1.6394329071044922,
      "learning_rate": 1.7248301150402263e-06,
      "loss": 0.4481,
      "step": 296000
    },
    {
      "epoch": 121.17,
      "grad_norm": 2.0316214561462402,
      "learning_rate": 1.7246771178077263e-06,
      "loss": 0.4693,
      "step": 296010
    },
    {
      "epoch": 121.17,
      "grad_norm": 2.0107126235961914,
      "learning_rate": 1.7245241234111027e-06,
      "loss": 0.4751,
      "step": 296020
    },
    {
      "epoch": 121.17,
      "grad_norm": 1.968392252922058,
      "learning_rate": 1.7243711318510576e-06,
      "loss": 0.4816,
      "step": 296030
    },
    {
      "epoch": 121.18,
      "grad_norm": 2.203002691268921,
      "learning_rate": 1.7242181431282908e-06,
      "loss": 0.475,
      "step": 296040
    },
    {
      "epoch": 121.18,
      "grad_norm": 2.1242966651916504,
      "learning_rate": 1.724065157243504e-06,
      "loss": 0.4552,
      "step": 296050
    },
    {
      "epoch": 121.19,
      "grad_norm": 1.5198875665664673,
      "learning_rate": 1.7239121741973973e-06,
      "loss": 0.4684,
      "step": 296060
    },
    {
      "epoch": 121.19,
      "grad_norm": 1.7323909997940063,
      "learning_rate": 1.7237591939906713e-06,
      "loss": 0.4761,
      "step": 296070
    },
    {
      "epoch": 121.2,
      "grad_norm": 2.276304244995117,
      "learning_rate": 1.7236062166240274e-06,
      "loss": 0.446,
      "step": 296080
    },
    {
      "epoch": 121.2,
      "grad_norm": 2.0176849365234375,
      "learning_rate": 1.7234532420981661e-06,
      "loss": 0.4694,
      "step": 296090
    },
    {
      "epoch": 121.2,
      "grad_norm": 1.9966357946395874,
      "learning_rate": 1.7233002704137883e-06,
      "loss": 0.4788,
      "step": 296100
    },
    {
      "epoch": 121.21,
      "grad_norm": 1.9681336879730225,
      "learning_rate": 1.7231473015715942e-06,
      "loss": 0.4724,
      "step": 296110
    },
    {
      "epoch": 121.21,
      "grad_norm": 2.1263256072998047,
      "learning_rate": 1.7229943355722853e-06,
      "loss": 0.4503,
      "step": 296120
    },
    {
      "epoch": 121.22,
      "grad_norm": 1.77780020236969,
      "learning_rate": 1.7228413724165613e-06,
      "loss": 0.4603,
      "step": 296130
    },
    {
      "epoch": 121.22,
      "grad_norm": 2.0329103469848633,
      "learning_rate": 1.7226884121051243e-06,
      "loss": 0.4822,
      "step": 296140
    },
    {
      "epoch": 121.22,
      "grad_norm": 1.8846696615219116,
      "learning_rate": 1.722535454638674e-06,
      "loss": 0.4742,
      "step": 296150
    },
    {
      "epoch": 121.23,
      "grad_norm": 2.206763982772827,
      "learning_rate": 1.7223825000179114e-06,
      "loss": 0.4707,
      "step": 296160
    },
    {
      "epoch": 121.23,
      "grad_norm": 1.8997056484222412,
      "learning_rate": 1.7222295482435373e-06,
      "loss": 0.4412,
      "step": 296170
    },
    {
      "epoch": 121.24,
      "grad_norm": 2.1115570068359375,
      "learning_rate": 1.7220765993162524e-06,
      "loss": 0.454,
      "step": 296180
    },
    {
      "epoch": 121.24,
      "grad_norm": 2.1139025688171387,
      "learning_rate": 1.7219236532367564e-06,
      "loss": 0.4754,
      "step": 296190
    },
    {
      "epoch": 121.24,
      "grad_norm": 1.808789610862732,
      "learning_rate": 1.7217707100057507e-06,
      "loss": 0.4481,
      "step": 296200
    },
    {
      "epoch": 121.25,
      "grad_norm": 1.9266939163208008,
      "learning_rate": 1.721617769623936e-06,
      "loss": 0.4672,
      "step": 296210
    },
    {
      "epoch": 121.25,
      "grad_norm": 2.3931186199188232,
      "learning_rate": 1.7214648320920128e-06,
      "loss": 0.4532,
      "step": 296220
    },
    {
      "epoch": 121.26,
      "grad_norm": 2.0883140563964844,
      "learning_rate": 1.7213118974106814e-06,
      "loss": 0.4672,
      "step": 296230
    },
    {
      "epoch": 121.26,
      "grad_norm": 2.0646324157714844,
      "learning_rate": 1.7211589655806422e-06,
      "loss": 0.4383,
      "step": 296240
    },
    {
      "epoch": 121.26,
      "grad_norm": 2.1465444564819336,
      "learning_rate": 1.7210060366025966e-06,
      "loss": 0.4929,
      "step": 296250
    },
    {
      "epoch": 121.27,
      "grad_norm": 1.9044770002365112,
      "learning_rate": 1.7208531104772445e-06,
      "loss": 0.4653,
      "step": 296260
    },
    {
      "epoch": 121.27,
      "grad_norm": 1.8385626077651978,
      "learning_rate": 1.7207001872052872e-06,
      "loss": 0.4586,
      "step": 296270
    },
    {
      "epoch": 121.28,
      "grad_norm": 5.652352333068848,
      "learning_rate": 1.7205472667874239e-06,
      "loss": 0.4753,
      "step": 296280
    },
    {
      "epoch": 121.28,
      "grad_norm": 1.7531169652938843,
      "learning_rate": 1.7203943492243557e-06,
      "loss": 0.4736,
      "step": 296290
    },
    {
      "epoch": 121.29,
      "grad_norm": 1.9388219118118286,
      "learning_rate": 1.7202414345167835e-06,
      "loss": 0.4715,
      "step": 296300
    },
    {
      "epoch": 121.29,
      "grad_norm": 2.0058493614196777,
      "learning_rate": 1.7200885226654076e-06,
      "loss": 0.4527,
      "step": 296310
    },
    {
      "epoch": 121.29,
      "grad_norm": 2.0754261016845703,
      "learning_rate": 1.7199356136709273e-06,
      "loss": 0.4655,
      "step": 296320
    },
    {
      "epoch": 121.3,
      "grad_norm": 2.0361547470092773,
      "learning_rate": 1.719782707534045e-06,
      "loss": 0.4492,
      "step": 296330
    },
    {
      "epoch": 121.3,
      "grad_norm": 2.3857150077819824,
      "learning_rate": 1.7196298042554602e-06,
      "loss": 0.4611,
      "step": 296340
    },
    {
      "epoch": 121.31,
      "grad_norm": 1.600616216659546,
      "learning_rate": 1.7194769038358733e-06,
      "loss": 0.4462,
      "step": 296350
    },
    {
      "epoch": 121.31,
      "grad_norm": 1.765601396560669,
      "learning_rate": 1.7193240062759846e-06,
      "loss": 0.4715,
      "step": 296360
    },
    {
      "epoch": 121.31,
      "grad_norm": 1.6320923566818237,
      "learning_rate": 1.719171111576495e-06,
      "loss": 0.4671,
      "step": 296370
    },
    {
      "epoch": 121.32,
      "grad_norm": 1.7886476516723633,
      "learning_rate": 1.7190182197381042e-06,
      "loss": 0.4483,
      "step": 296380
    },
    {
      "epoch": 121.32,
      "grad_norm": 1.7048358917236328,
      "learning_rate": 1.7188653307615133e-06,
      "loss": 0.4615,
      "step": 296390
    },
    {
      "epoch": 121.33,
      "grad_norm": 2.207571506500244,
      "learning_rate": 1.7187124446474222e-06,
      "loss": 0.4539,
      "step": 296400
    },
    {
      "epoch": 121.33,
      "grad_norm": 2.1445400714874268,
      "learning_rate": 1.7185595613965316e-06,
      "loss": 0.4692,
      "step": 296410
    },
    {
      "epoch": 121.33,
      "grad_norm": 1.8697417974472046,
      "learning_rate": 1.7184066810095413e-06,
      "loss": 0.4494,
      "step": 296420
    },
    {
      "epoch": 121.34,
      "grad_norm": 2.3914358615875244,
      "learning_rate": 1.7182538034871519e-06,
      "loss": 0.4863,
      "step": 296430
    },
    {
      "epoch": 121.34,
      "grad_norm": 1.7517681121826172,
      "learning_rate": 1.7181009288300636e-06,
      "loss": 0.4736,
      "step": 296440
    },
    {
      "epoch": 121.35,
      "grad_norm": 2.253964900970459,
      "learning_rate": 1.7179480570389767e-06,
      "loss": 0.4632,
      "step": 296450
    },
    {
      "epoch": 121.35,
      "grad_norm": 1.886069655418396,
      "learning_rate": 1.7177951881145921e-06,
      "loss": 0.4418,
      "step": 296460
    },
    {
      "epoch": 121.35,
      "grad_norm": 1.987581491470337,
      "learning_rate": 1.7176423220576091e-06,
      "loss": 0.4556,
      "step": 296470
    },
    {
      "epoch": 121.36,
      "grad_norm": 2.1693739891052246,
      "learning_rate": 1.7174894588687284e-06,
      "loss": 0.452,
      "step": 296480
    },
    {
      "epoch": 121.36,
      "grad_norm": 1.8053739070892334,
      "learning_rate": 1.7173365985486508e-06,
      "loss": 0.4695,
      "step": 296490
    },
    {
      "epoch": 121.37,
      "grad_norm": 1.975265622138977,
      "learning_rate": 1.7171837410980754e-06,
      "loss": 0.4936,
      "step": 296500
    },
    {
      "epoch": 121.37,
      "grad_norm": 1.8768585920333862,
      "learning_rate": 1.7170308865177024e-06,
      "loss": 0.467,
      "step": 296510
    },
    {
      "epoch": 121.38,
      "grad_norm": 2.2606287002563477,
      "learning_rate": 1.716878034808233e-06,
      "loss": 0.4619,
      "step": 296520
    },
    {
      "epoch": 121.38,
      "grad_norm": 1.79515540599823,
      "learning_rate": 1.7167251859703673e-06,
      "loss": 0.4596,
      "step": 296530
    },
    {
      "epoch": 121.38,
      "grad_norm": 1.9412710666656494,
      "learning_rate": 1.716572340004806e-06,
      "loss": 0.4663,
      "step": 296540
    },
    {
      "epoch": 121.39,
      "grad_norm": 1.7325682640075684,
      "learning_rate": 1.7164194969122477e-06,
      "loss": 0.4514,
      "step": 296550
    },
    {
      "epoch": 121.39,
      "grad_norm": 2.0079641342163086,
      "learning_rate": 1.716266656693393e-06,
      "loss": 0.4536,
      "step": 296560
    },
    {
      "epoch": 121.4,
      "grad_norm": 1.8998451232910156,
      "learning_rate": 1.7161138193489424e-06,
      "loss": 0.4611,
      "step": 296570
    },
    {
      "epoch": 121.4,
      "grad_norm": 2.3573291301727295,
      "learning_rate": 1.715960984879596e-06,
      "loss": 0.4778,
      "step": 296580
    },
    {
      "epoch": 121.4,
      "grad_norm": 2.0205626487731934,
      "learning_rate": 1.715808153286054e-06,
      "loss": 0.4554,
      "step": 296590
    },
    {
      "epoch": 121.41,
      "grad_norm": 1.98911714553833,
      "learning_rate": 1.715655324569016e-06,
      "loss": 0.4496,
      "step": 296600
    },
    {
      "epoch": 121.41,
      "grad_norm": 1.9432693719863892,
      "learning_rate": 1.715502498729183e-06,
      "loss": 0.4554,
      "step": 296610
    },
    {
      "epoch": 121.42,
      "grad_norm": 2.1334073543548584,
      "learning_rate": 1.715349675767254e-06,
      "loss": 0.4806,
      "step": 296620
    },
    {
      "epoch": 121.42,
      "grad_norm": 2.3827028274536133,
      "learning_rate": 1.7151968556839302e-06,
      "loss": 0.4474,
      "step": 296630
    },
    {
      "epoch": 121.42,
      "grad_norm": 2.002434492111206,
      "learning_rate": 1.7150440384799106e-06,
      "loss": 0.4476,
      "step": 296640
    },
    {
      "epoch": 121.43,
      "grad_norm": 2.0614044666290283,
      "learning_rate": 1.7148912241558956e-06,
      "loss": 0.4598,
      "step": 296650
    },
    {
      "epoch": 121.43,
      "grad_norm": 2.0569381713867188,
      "learning_rate": 1.7147384127125854e-06,
      "loss": 0.4555,
      "step": 296660
    },
    {
      "epoch": 121.44,
      "grad_norm": 1.9196804761886597,
      "learning_rate": 1.7145856041506797e-06,
      "loss": 0.4722,
      "step": 296670
    },
    {
      "epoch": 121.44,
      "grad_norm": 1.9849731922149658,
      "learning_rate": 1.7144327984708786e-06,
      "loss": 0.4369,
      "step": 296680
    },
    {
      "epoch": 121.44,
      "grad_norm": 1.8132221698760986,
      "learning_rate": 1.7142799956738818e-06,
      "loss": 0.4522,
      "step": 296690
    },
    {
      "epoch": 121.45,
      "grad_norm": 2.1979098320007324,
      "learning_rate": 1.71412719576039e-06,
      "loss": 0.4647,
      "step": 296700
    },
    {
      "epoch": 121.45,
      "grad_norm": 1.7120410203933716,
      "learning_rate": 1.7139743987311028e-06,
      "loss": 0.4676,
      "step": 296710
    },
    {
      "epoch": 121.46,
      "grad_norm": 2.066183090209961,
      "learning_rate": 1.7138216045867198e-06,
      "loss": 0.4473,
      "step": 296720
    },
    {
      "epoch": 121.46,
      "grad_norm": 1.7054904699325562,
      "learning_rate": 1.7136688133279418e-06,
      "loss": 0.4554,
      "step": 296730
    },
    {
      "epoch": 121.47,
      "grad_norm": 1.956514596939087,
      "learning_rate": 1.7135160249554677e-06,
      "loss": 0.4513,
      "step": 296740
    },
    {
      "epoch": 121.47,
      "grad_norm": 2.270034074783325,
      "learning_rate": 1.7133632394699982e-06,
      "loss": 0.4619,
      "step": 296750
    },
    {
      "epoch": 121.47,
      "grad_norm": 1.939261794090271,
      "learning_rate": 1.7132104568722324e-06,
      "loss": 0.4532,
      "step": 296760
    },
    {
      "epoch": 121.48,
      "grad_norm": 1.8406869173049927,
      "learning_rate": 1.7130576771628714e-06,
      "loss": 0.478,
      "step": 296770
    },
    {
      "epoch": 121.48,
      "grad_norm": 2.095123529434204,
      "learning_rate": 1.7129049003426135e-06,
      "loss": 0.4899,
      "step": 296780
    },
    {
      "epoch": 121.49,
      "grad_norm": 1.8135989904403687,
      "learning_rate": 1.7127521264121596e-06,
      "loss": 0.4621,
      "step": 296790
    },
    {
      "epoch": 121.49,
      "grad_norm": 2.0901079177856445,
      "learning_rate": 1.7125993553722092e-06,
      "loss": 0.4775,
      "step": 296800
    },
    {
      "epoch": 121.49,
      "grad_norm": 1.851880431175232,
      "learning_rate": 1.712446587223462e-06,
      "loss": 0.4604,
      "step": 296810
    },
    {
      "epoch": 121.5,
      "grad_norm": 1.8206897974014282,
      "learning_rate": 1.712293821966618e-06,
      "loss": 0.4603,
      "step": 296820
    },
    {
      "epoch": 121.5,
      "grad_norm": 1.7869547605514526,
      "learning_rate": 1.7121410596023769e-06,
      "loss": 0.4548,
      "step": 296830
    },
    {
      "epoch": 121.51,
      "grad_norm": 2.07403302192688,
      "learning_rate": 1.7119883001314385e-06,
      "loss": 0.4975,
      "step": 296840
    },
    {
      "epoch": 121.51,
      "grad_norm": 1.7820889949798584,
      "learning_rate": 1.7118355435545035e-06,
      "loss": 0.4669,
      "step": 296850
    },
    {
      "epoch": 121.51,
      "grad_norm": 2.401355028152466,
      "learning_rate": 1.7116827898722696e-06,
      "loss": 0.4639,
      "step": 296860
    },
    {
      "epoch": 121.52,
      "grad_norm": 2.1549224853515625,
      "learning_rate": 1.7115300390854382e-06,
      "loss": 0.4505,
      "step": 296870
    },
    {
      "epoch": 121.52,
      "grad_norm": 1.9258116483688354,
      "learning_rate": 1.7113772911947074e-06,
      "loss": 0.4652,
      "step": 296880
    },
    {
      "epoch": 121.53,
      "grad_norm": 2.3104286193847656,
      "learning_rate": 1.7112245462007793e-06,
      "loss": 0.4584,
      "step": 296890
    },
    {
      "epoch": 121.53,
      "grad_norm": 2.6800806522369385,
      "learning_rate": 1.7110718041043524e-06,
      "loss": 0.4496,
      "step": 296900
    },
    {
      "epoch": 121.53,
      "grad_norm": 1.890513300895691,
      "learning_rate": 1.710919064906126e-06,
      "loss": 0.4573,
      "step": 296910
    },
    {
      "epoch": 121.54,
      "grad_norm": 1.805769920349121,
      "learning_rate": 1.7107663286068002e-06,
      "loss": 0.4738,
      "step": 296920
    },
    {
      "epoch": 121.54,
      "grad_norm": 2.013005018234253,
      "learning_rate": 1.7106135952070746e-06,
      "loss": 0.459,
      "step": 296930
    },
    {
      "epoch": 121.55,
      "grad_norm": 1.7309166193008423,
      "learning_rate": 1.7104608647076485e-06,
      "loss": 0.4743,
      "step": 296940
    },
    {
      "epoch": 121.55,
      "grad_norm": 2.433746576309204,
      "learning_rate": 1.710308137109222e-06,
      "loss": 0.4706,
      "step": 296950
    },
    {
      "epoch": 121.56,
      "grad_norm": 2.131335973739624,
      "learning_rate": 1.7101554124124949e-06,
      "loss": 0.4549,
      "step": 296960
    },
    {
      "epoch": 121.56,
      "grad_norm": 2.450536012649536,
      "learning_rate": 1.7100026906181663e-06,
      "loss": 0.4666,
      "step": 296970
    },
    {
      "epoch": 121.56,
      "grad_norm": 1.9041118621826172,
      "learning_rate": 1.7098499717269356e-06,
      "loss": 0.4626,
      "step": 296980
    },
    {
      "epoch": 121.57,
      "grad_norm": 1.9876220226287842,
      "learning_rate": 1.7096972557395036e-06,
      "loss": 0.4699,
      "step": 296990
    },
    {
      "epoch": 121.57,
      "grad_norm": 1.8339312076568604,
      "learning_rate": 1.7095445426565686e-06,
      "loss": 0.4647,
      "step": 297000
    },
    {
      "epoch": 121.58,
      "grad_norm": 2.2608416080474854,
      "learning_rate": 1.7093918324788306e-06,
      "loss": 0.4652,
      "step": 297010
    },
    {
      "epoch": 121.58,
      "grad_norm": 1.726500391960144,
      "learning_rate": 1.7092391252069888e-06,
      "loss": 0.4724,
      "step": 297020
    },
    {
      "epoch": 121.58,
      "grad_norm": 1.838866949081421,
      "learning_rate": 1.7090864208417434e-06,
      "loss": 0.4314,
      "step": 297030
    },
    {
      "epoch": 121.59,
      "grad_norm": 2.25338077545166,
      "learning_rate": 1.7089337193837935e-06,
      "loss": 0.4427,
      "step": 297040
    },
    {
      "epoch": 121.59,
      "grad_norm": 1.817523717880249,
      "learning_rate": 1.7087810208338385e-06,
      "loss": 0.4609,
      "step": 297050
    },
    {
      "epoch": 121.6,
      "grad_norm": 2.259075403213501,
      "learning_rate": 1.708628325192578e-06,
      "loss": 0.4673,
      "step": 297060
    },
    {
      "epoch": 121.6,
      "grad_norm": 2.1459357738494873,
      "learning_rate": 1.7084756324607113e-06,
      "loss": 0.4505,
      "step": 297070
    },
    {
      "epoch": 121.6,
      "grad_norm": 1.7952138185501099,
      "learning_rate": 1.7083229426389381e-06,
      "loss": 0.4563,
      "step": 297080
    },
    {
      "epoch": 121.61,
      "grad_norm": 1.8469021320343018,
      "learning_rate": 1.7081702557279585e-06,
      "loss": 0.4598,
      "step": 297090
    },
    {
      "epoch": 121.61,
      "grad_norm": 1.7763727903366089,
      "learning_rate": 1.708017571728471e-06,
      "loss": 0.477,
      "step": 297100
    },
    {
      "epoch": 121.62,
      "grad_norm": 1.710018277168274,
      "learning_rate": 1.7078648906411752e-06,
      "loss": 0.4673,
      "step": 297110
    },
    {
      "epoch": 121.62,
      "grad_norm": 1.9848909378051758,
      "learning_rate": 1.7077122124667706e-06,
      "loss": 0.4516,
      "step": 297120
    },
    {
      "epoch": 121.63,
      "grad_norm": 2.431079864501953,
      "learning_rate": 1.7075595372059572e-06,
      "loss": 0.457,
      "step": 297130
    },
    {
      "epoch": 121.63,
      "grad_norm": 2.104782819747925,
      "learning_rate": 1.7074068648594332e-06,
      "loss": 0.458,
      "step": 297140
    },
    {
      "epoch": 121.63,
      "grad_norm": 1.7738069295883179,
      "learning_rate": 1.707254195427899e-06,
      "loss": 0.4626,
      "step": 297150
    },
    {
      "epoch": 121.64,
      "grad_norm": 1.8061292171478271,
      "learning_rate": 1.7071015289120529e-06,
      "loss": 0.474,
      "step": 297160
    },
    {
      "epoch": 121.64,
      "grad_norm": 1.8159120082855225,
      "learning_rate": 1.706948865312595e-06,
      "loss": 0.467,
      "step": 297170
    },
    {
      "epoch": 121.65,
      "grad_norm": 1.7286648750305176,
      "learning_rate": 1.7067962046302246e-06,
      "loss": 0.4659,
      "step": 297180
    },
    {
      "epoch": 121.65,
      "grad_norm": 1.930419683456421,
      "learning_rate": 1.7066435468656409e-06,
      "loss": 0.4755,
      "step": 297190
    },
    {
      "epoch": 121.65,
      "grad_norm": 1.921876311302185,
      "learning_rate": 1.706490892019543e-06,
      "loss": 0.4809,
      "step": 297200
    },
    {
      "epoch": 121.66,
      "grad_norm": 1.5420589447021484,
      "learning_rate": 1.706338240092631e-06,
      "loss": 0.4604,
      "step": 297210
    },
    {
      "epoch": 121.66,
      "grad_norm": 1.874422311782837,
      "learning_rate": 1.706185591085603e-06,
      "loss": 0.4447,
      "step": 297220
    },
    {
      "epoch": 121.67,
      "grad_norm": 2.5246167182922363,
      "learning_rate": 1.7060329449991591e-06,
      "loss": 0.4555,
      "step": 297230
    },
    {
      "epoch": 121.67,
      "grad_norm": 2.0005664825439453,
      "learning_rate": 1.705880301833998e-06,
      "loss": 0.4717,
      "step": 297240
    },
    {
      "epoch": 121.67,
      "grad_norm": 2.0332257747650146,
      "learning_rate": 1.7057276615908194e-06,
      "loss": 0.4531,
      "step": 297250
    },
    {
      "epoch": 121.68,
      "grad_norm": 1.9640395641326904,
      "learning_rate": 1.7055750242703215e-06,
      "loss": 0.4524,
      "step": 297260
    },
    {
      "epoch": 121.68,
      "grad_norm": 1.5840566158294678,
      "learning_rate": 1.705422389873205e-06,
      "loss": 0.4633,
      "step": 297270
    },
    {
      "epoch": 121.69,
      "grad_norm": 1.7773462533950806,
      "learning_rate": 1.7052697584001686e-06,
      "loss": 0.4507,
      "step": 297280
    },
    {
      "epoch": 121.69,
      "grad_norm": 1.7766687870025635,
      "learning_rate": 1.7051171298519112e-06,
      "loss": 0.4525,
      "step": 297290
    },
    {
      "epoch": 121.69,
      "grad_norm": 2.2028541564941406,
      "learning_rate": 1.7049645042291321e-06,
      "loss": 0.4741,
      "step": 297300
    },
    {
      "epoch": 121.7,
      "grad_norm": 1.9561620950698853,
      "learning_rate": 1.7048118815325304e-06,
      "loss": 0.4321,
      "step": 297310
    },
    {
      "epoch": 121.7,
      "grad_norm": 1.7688326835632324,
      "learning_rate": 1.7046592617628051e-06,
      "loss": 0.4501,
      "step": 297320
    },
    {
      "epoch": 121.71,
      "grad_norm": 2.2576048374176025,
      "learning_rate": 1.7045066449206555e-06,
      "loss": 0.4598,
      "step": 297330
    },
    {
      "epoch": 121.71,
      "grad_norm": 1.7166212797164917,
      "learning_rate": 1.7043540310067807e-06,
      "loss": 0.4749,
      "step": 297340
    },
    {
      "epoch": 121.72,
      "grad_norm": 2.5743987560272217,
      "learning_rate": 1.7042014200218805e-06,
      "loss": 0.4521,
      "step": 297350
    },
    {
      "epoch": 121.72,
      "grad_norm": 2.816488265991211,
      "learning_rate": 1.7040488119666527e-06,
      "loss": 0.4666,
      "step": 297360
    },
    {
      "epoch": 121.72,
      "grad_norm": 2.268866777420044,
      "learning_rate": 1.7038962068417971e-06,
      "loss": 0.4477,
      "step": 297370
    },
    {
      "epoch": 121.73,
      "grad_norm": 1.9957457780838013,
      "learning_rate": 1.7037436046480125e-06,
      "loss": 0.4361,
      "step": 297380
    },
    {
      "epoch": 121.73,
      "grad_norm": 2.5050058364868164,
      "learning_rate": 1.7035910053859982e-06,
      "loss": 0.4622,
      "step": 297390
    },
    {
      "epoch": 121.74,
      "grad_norm": 2.329641819000244,
      "learning_rate": 1.7034384090564528e-06,
      "loss": 0.4744,
      "step": 297400
    },
    {
      "epoch": 121.74,
      "grad_norm": 5.048967361450195,
      "learning_rate": 1.703285815660076e-06,
      "loss": 0.4688,
      "step": 297410
    },
    {
      "epoch": 121.74,
      "grad_norm": 1.91050386428833,
      "learning_rate": 1.7031332251975664e-06,
      "loss": 0.4615,
      "step": 297420
    },
    {
      "epoch": 121.75,
      "grad_norm": 1.7986993789672852,
      "learning_rate": 1.7029806376696227e-06,
      "loss": 0.4316,
      "step": 297430
    },
    {
      "epoch": 121.75,
      "grad_norm": 1.9003247022628784,
      "learning_rate": 1.702828053076945e-06,
      "loss": 0.4758,
      "step": 297440
    },
    {
      "epoch": 121.76,
      "grad_norm": 1.8071753978729248,
      "learning_rate": 1.70267547142023e-06,
      "loss": 0.4616,
      "step": 297450
    },
    {
      "epoch": 121.76,
      "grad_norm": 2.224061965942383,
      "learning_rate": 1.702522892700179e-06,
      "loss": 0.4436,
      "step": 297460
    },
    {
      "epoch": 121.76,
      "grad_norm": 4.851157188415527,
      "learning_rate": 1.70237031691749e-06,
      "loss": 0.451,
      "step": 297470
    },
    {
      "epoch": 121.77,
      "grad_norm": 1.7972477674484253,
      "learning_rate": 1.7022177440728622e-06,
      "loss": 0.4807,
      "step": 297480
    },
    {
      "epoch": 121.77,
      "grad_norm": 1.7056467533111572,
      "learning_rate": 1.7020651741669947e-06,
      "loss": 0.4445,
      "step": 297490
    },
    {
      "epoch": 121.78,
      "grad_norm": 2.063014030456543,
      "learning_rate": 1.7019126072005857e-06,
      "loss": 0.4546,
      "step": 297500
    },
    {
      "epoch": 121.78,
      "grad_norm": 1.9214308261871338,
      "learning_rate": 1.7017600431743344e-06,
      "loss": 0.4603,
      "step": 297510
    },
    {
      "epoch": 121.78,
      "grad_norm": 2.352841854095459,
      "learning_rate": 1.7016074820889395e-06,
      "loss": 0.4614,
      "step": 297520
    },
    {
      "epoch": 121.79,
      "grad_norm": 1.9789421558380127,
      "learning_rate": 1.7014549239450999e-06,
      "loss": 0.4383,
      "step": 297530
    },
    {
      "epoch": 121.79,
      "grad_norm": 2.2790024280548096,
      "learning_rate": 1.7013023687435148e-06,
      "loss": 0.4712,
      "step": 297540
    },
    {
      "epoch": 121.8,
      "grad_norm": 1.809808373451233,
      "learning_rate": 1.7011498164848827e-06,
      "loss": 0.4593,
      "step": 297550
    },
    {
      "epoch": 121.8,
      "grad_norm": 1.9043140411376953,
      "learning_rate": 1.7009972671699024e-06,
      "loss": 0.4621,
      "step": 297560
    },
    {
      "epoch": 121.81,
      "grad_norm": 1.6483826637268066,
      "learning_rate": 1.7008447207992735e-06,
      "loss": 0.4526,
      "step": 297570
    },
    {
      "epoch": 121.81,
      "grad_norm": 1.6964889764785767,
      "learning_rate": 1.7006921773736935e-06,
      "loss": 0.4601,
      "step": 297580
    },
    {
      "epoch": 121.81,
      "grad_norm": 2.040353298187256,
      "learning_rate": 1.700539636893862e-06,
      "loss": 0.4606,
      "step": 297590
    },
    {
      "epoch": 121.82,
      "grad_norm": 1.8211519718170166,
      "learning_rate": 1.7003870993604775e-06,
      "loss": 0.4495,
      "step": 297600
    },
    {
      "epoch": 121.82,
      "grad_norm": 2.4935286045074463,
      "learning_rate": 1.7002345647742384e-06,
      "loss": 0.4582,
      "step": 297610
    },
    {
      "epoch": 121.83,
      "grad_norm": 1.7479963302612305,
      "learning_rate": 1.7000820331358446e-06,
      "loss": 0.4438,
      "step": 297620
    },
    {
      "epoch": 121.83,
      "grad_norm": 1.8707886934280396,
      "learning_rate": 1.6999295044459934e-06,
      "loss": 0.4391,
      "step": 297630
    },
    {
      "epoch": 121.83,
      "grad_norm": 1.951764702796936,
      "learning_rate": 1.6997769787053842e-06,
      "loss": 0.437,
      "step": 297640
    },
    {
      "epoch": 121.84,
      "grad_norm": 2.2551369667053223,
      "learning_rate": 1.6996244559147158e-06,
      "loss": 0.477,
      "step": 297650
    },
    {
      "epoch": 121.84,
      "grad_norm": 1.3968732357025146,
      "learning_rate": 1.6994719360746868e-06,
      "loss": 0.4713,
      "step": 297660
    },
    {
      "epoch": 121.85,
      "grad_norm": 1.8916651010513306,
      "learning_rate": 1.699319419185996e-06,
      "loss": 0.4591,
      "step": 297670
    },
    {
      "epoch": 121.85,
      "grad_norm": 3.257728099822998,
      "learning_rate": 1.6991669052493418e-06,
      "loss": 0.4707,
      "step": 297680
    },
    {
      "epoch": 121.85,
      "grad_norm": 1.619553565979004,
      "learning_rate": 1.699014394265423e-06,
      "loss": 0.4668,
      "step": 297690
    },
    {
      "epoch": 121.86,
      "grad_norm": 2.3932290077209473,
      "learning_rate": 1.6988618862349381e-06,
      "loss": 0.4519,
      "step": 297700
    },
    {
      "epoch": 121.86,
      "grad_norm": 2.179610013961792,
      "learning_rate": 1.6987093811585864e-06,
      "loss": 0.466,
      "step": 297710
    },
    {
      "epoch": 121.87,
      "grad_norm": 1.7051122188568115,
      "learning_rate": 1.698556879037065e-06,
      "loss": 0.452,
      "step": 297720
    },
    {
      "epoch": 121.87,
      "grad_norm": 2.0614993572235107,
      "learning_rate": 1.6984043798710737e-06,
      "loss": 0.4482,
      "step": 297730
    },
    {
      "epoch": 121.87,
      "grad_norm": 2.2052104473114014,
      "learning_rate": 1.6982518836613108e-06,
      "loss": 0.4605,
      "step": 297740
    },
    {
      "epoch": 121.88,
      "grad_norm": 3.0611929893493652,
      "learning_rate": 1.6980993904084744e-06,
      "loss": 0.4545,
      "step": 297750
    },
    {
      "epoch": 121.88,
      "grad_norm": 2.0291967391967773,
      "learning_rate": 1.6979469001132637e-06,
      "loss": 0.4625,
      "step": 297760
    },
    {
      "epoch": 121.89,
      "grad_norm": 2.351612091064453,
      "learning_rate": 1.6977944127763772e-06,
      "loss": 0.4678,
      "step": 297770
    },
    {
      "epoch": 121.89,
      "grad_norm": 1.717281699180603,
      "learning_rate": 1.6976419283985126e-06,
      "loss": 0.4644,
      "step": 297780
    },
    {
      "epoch": 121.9,
      "grad_norm": 2.425243377685547,
      "learning_rate": 1.6974894469803698e-06,
      "loss": 0.4691,
      "step": 297790
    },
    {
      "epoch": 121.9,
      "grad_norm": 1.9674588441848755,
      "learning_rate": 1.6973369685226467e-06,
      "loss": 0.4553,
      "step": 297800
    },
    {
      "epoch": 121.9,
      "grad_norm": 1.9992471933364868,
      "learning_rate": 1.697184493026041e-06,
      "loss": 0.4407,
      "step": 297810
    },
    {
      "epoch": 121.91,
      "grad_norm": 1.787460446357727,
      "learning_rate": 1.6970320204912515e-06,
      "loss": 0.4639,
      "step": 297820
    },
    {
      "epoch": 121.91,
      "grad_norm": 1.9842238426208496,
      "learning_rate": 1.6968795509189763e-06,
      "loss": 0.4592,
      "step": 297830
    },
    {
      "epoch": 121.92,
      "grad_norm": 1.6901880502700806,
      "learning_rate": 1.6967270843099152e-06,
      "loss": 0.4582,
      "step": 297840
    },
    {
      "epoch": 121.92,
      "grad_norm": 2.2672228813171387,
      "learning_rate": 1.6965746206647665e-06,
      "loss": 0.482,
      "step": 297850
    },
    {
      "epoch": 121.92,
      "grad_norm": 2.1548099517822266,
      "learning_rate": 1.6964221599842275e-06,
      "loss": 0.4624,
      "step": 297860
    },
    {
      "epoch": 121.93,
      "grad_norm": 1.9379700422286987,
      "learning_rate": 1.696269702268997e-06,
      "loss": 0.47,
      "step": 297870
    },
    {
      "epoch": 121.93,
      "grad_norm": 1.7904105186462402,
      "learning_rate": 1.6961172475197737e-06,
      "loss": 0.4625,
      "step": 297880
    },
    {
      "epoch": 121.94,
      "grad_norm": 2.126014232635498,
      "learning_rate": 1.6959647957372556e-06,
      "loss": 0.4453,
      "step": 297890
    },
    {
      "epoch": 121.94,
      "grad_norm": 1.7768751382827759,
      "learning_rate": 1.6958123469221409e-06,
      "loss": 0.4614,
      "step": 297900
    },
    {
      "epoch": 121.94,
      "grad_norm": 2.2678675651550293,
      "learning_rate": 1.695659901075129e-06,
      "loss": 0.4519,
      "step": 297910
    },
    {
      "epoch": 121.95,
      "grad_norm": 2.127825975418091,
      "learning_rate": 1.6955074581969164e-06,
      "loss": 0.45,
      "step": 297920
    },
    {
      "epoch": 121.95,
      "grad_norm": 2.3639893531799316,
      "learning_rate": 1.6953550182882038e-06,
      "loss": 0.4524,
      "step": 297930
    },
    {
      "epoch": 121.96,
      "grad_norm": 1.8564062118530273,
      "learning_rate": 1.6952025813496874e-06,
      "loss": 0.4762,
      "step": 297940
    },
    {
      "epoch": 121.96,
      "grad_norm": 2.073712110519409,
      "learning_rate": 1.6950501473820665e-06,
      "loss": 0.457,
      "step": 297950
    },
    {
      "epoch": 121.96,
      "grad_norm": 1.9572581052780151,
      "learning_rate": 1.6948977163860391e-06,
      "loss": 0.4493,
      "step": 297960
    },
    {
      "epoch": 121.97,
      "grad_norm": 2.2012455463409424,
      "learning_rate": 1.6947452883623033e-06,
      "loss": 0.4467,
      "step": 297970
    },
    {
      "epoch": 121.97,
      "grad_norm": 1.9543565511703491,
      "learning_rate": 1.6945928633115577e-06,
      "loss": 0.4496,
      "step": 297980
    },
    {
      "epoch": 121.98,
      "grad_norm": 2.3393843173980713,
      "learning_rate": 1.6944404412345005e-06,
      "loss": 0.4677,
      "step": 297990
    },
    {
      "epoch": 121.98,
      "grad_norm": 2.39017915725708,
      "learning_rate": 1.6942880221318299e-06,
      "loss": 0.4544,
      "step": 298000
    },
    {
      "epoch": 121.99,
      "grad_norm": 1.7605501413345337,
      "learning_rate": 1.694135606004243e-06,
      "loss": 0.4463,
      "step": 298010
    },
    {
      "epoch": 121.99,
      "grad_norm": 1.5356727838516235,
      "learning_rate": 1.69398319285244e-06,
      "loss": 0.471,
      "step": 298020
    },
    {
      "epoch": 121.99,
      "grad_norm": 1.7065430879592896,
      "learning_rate": 1.6938307826771182e-06,
      "loss": 0.4559,
      "step": 298030
    },
    {
      "epoch": 122.0,
      "grad_norm": 2.703421115875244,
      "learning_rate": 1.6936783754789755e-06,
      "loss": 0.4824,
      "step": 298040
    },
    {
      "epoch": 122.0,
      "eval_loss": 0.46311235427856445,
      "eval_runtime": 53.0195,
      "eval_samples_per_second": 65.052,
      "eval_steps_per_second": 8.148,
      "step": 298046
    },
    {
      "epoch": 122.0,
      "grad_norm": 1.9038079977035522,
      "learning_rate": 1.6935259712587104e-06,
      "loss": 0.4731,
      "step": 298050
    },
    {
      "epoch": 122.01,
      "grad_norm": 1.926439642906189,
      "learning_rate": 1.6933735700170205e-06,
      "loss": 0.4614,
      "step": 298060
    },
    {
      "epoch": 122.01,
      "grad_norm": 1.6771634817123413,
      "learning_rate": 1.693221171754605e-06,
      "loss": 0.4672,
      "step": 298070
    },
    {
      "epoch": 122.01,
      "grad_norm": 1.8999086618423462,
      "learning_rate": 1.6930687764721611e-06,
      "loss": 0.4867,
      "step": 298080
    },
    {
      "epoch": 122.02,
      "grad_norm": 2.2350103855133057,
      "learning_rate": 1.6929163841703865e-06,
      "loss": 0.4585,
      "step": 298090
    },
    {
      "epoch": 122.02,
      "grad_norm": 1.5744658708572388,
      "learning_rate": 1.6927639948499806e-06,
      "loss": 0.4617,
      "step": 298100
    },
    {
      "epoch": 122.03,
      "grad_norm": 1.593062400817871,
      "learning_rate": 1.6926116085116402e-06,
      "loss": 0.4416,
      "step": 298110
    },
    {
      "epoch": 122.03,
      "grad_norm": 2.5840158462524414,
      "learning_rate": 1.692459225156064e-06,
      "loss": 0.4375,
      "step": 298120
    },
    {
      "epoch": 122.03,
      "grad_norm": 6.960775375366211,
      "learning_rate": 1.6923068447839505e-06,
      "loss": 0.4698,
      "step": 298130
    },
    {
      "epoch": 122.04,
      "grad_norm": 2.029592990875244,
      "learning_rate": 1.6921544673959966e-06,
      "loss": 0.4571,
      "step": 298140
    },
    {
      "epoch": 122.04,
      "grad_norm": 2.392728567123413,
      "learning_rate": 1.6920020929929016e-06,
      "loss": 0.4607,
      "step": 298150
    },
    {
      "epoch": 122.05,
      "grad_norm": 2.0163846015930176,
      "learning_rate": 1.6918497215753625e-06,
      "loss": 0.4536,
      "step": 298160
    },
    {
      "epoch": 122.05,
      "grad_norm": 1.8652795553207397,
      "learning_rate": 1.6916973531440776e-06,
      "loss": 0.4758,
      "step": 298170
    },
    {
      "epoch": 122.05,
      "grad_norm": 1.8495471477508545,
      "learning_rate": 1.691544987699745e-06,
      "loss": 0.4641,
      "step": 298180
    },
    {
      "epoch": 122.06,
      "grad_norm": 2.28817081451416,
      "learning_rate": 1.691392625243062e-06,
      "loss": 0.4576,
      "step": 298190
    },
    {
      "epoch": 122.06,
      "grad_norm": 1.8526430130004883,
      "learning_rate": 1.691240265774727e-06,
      "loss": 0.4394,
      "step": 298200
    },
    {
      "epoch": 122.07,
      "grad_norm": 1.9228917360305786,
      "learning_rate": 1.6910879092954388e-06,
      "loss": 0.4665,
      "step": 298210
    },
    {
      "epoch": 122.07,
      "grad_norm": 2.083723783493042,
      "learning_rate": 1.6909355558058941e-06,
      "loss": 0.4738,
      "step": 298220
    },
    {
      "epoch": 122.08,
      "grad_norm": 2.745088815689087,
      "learning_rate": 1.6907832053067915e-06,
      "loss": 0.4683,
      "step": 298230
    },
    {
      "epoch": 122.08,
      "grad_norm": 2.0141518115997314,
      "learning_rate": 1.6906308577988285e-06,
      "loss": 0.4731,
      "step": 298240
    },
    {
      "epoch": 122.08,
      "grad_norm": 1.8855057954788208,
      "learning_rate": 1.6904785132827032e-06,
      "loss": 0.4449,
      "step": 298250
    },
    {
      "epoch": 122.09,
      "grad_norm": 2.030233144760132,
      "learning_rate": 1.6903261717591134e-06,
      "loss": 0.4797,
      "step": 298260
    },
    {
      "epoch": 122.09,
      "grad_norm": 1.8301728963851929,
      "learning_rate": 1.6901738332287571e-06,
      "loss": 0.4407,
      "step": 298270
    },
    {
      "epoch": 122.1,
      "grad_norm": 1.6934159994125366,
      "learning_rate": 1.690021497692332e-06,
      "loss": 0.4566,
      "step": 298280
    },
    {
      "epoch": 122.1,
      "grad_norm": 2.076155662536621,
      "learning_rate": 1.6898691651505361e-06,
      "loss": 0.4405,
      "step": 298290
    },
    {
      "epoch": 122.1,
      "grad_norm": 1.8824056386947632,
      "learning_rate": 1.689716835604067e-06,
      "loss": 0.4675,
      "step": 298300
    },
    {
      "epoch": 122.11,
      "grad_norm": 2.157930612564087,
      "learning_rate": 1.6895645090536223e-06,
      "loss": 0.4663,
      "step": 298310
    },
    {
      "epoch": 122.11,
      "grad_norm": 1.7343789339065552,
      "learning_rate": 1.6894121854999e-06,
      "loss": 0.4803,
      "step": 298320
    },
    {
      "epoch": 122.12,
      "grad_norm": 1.6246589422225952,
      "learning_rate": 1.6892598649435978e-06,
      "loss": 0.4625,
      "step": 298330
    },
    {
      "epoch": 122.12,
      "grad_norm": 1.8347558975219727,
      "learning_rate": 1.6891075473854138e-06,
      "loss": 0.4644,
      "step": 298340
    },
    {
      "epoch": 122.12,
      "grad_norm": 1.653975009918213,
      "learning_rate": 1.6889552328260454e-06,
      "loss": 0.475,
      "step": 298350
    },
    {
      "epoch": 122.13,
      "grad_norm": 1.945919394493103,
      "learning_rate": 1.6888029212661902e-06,
      "loss": 0.4756,
      "step": 298360
    },
    {
      "epoch": 122.13,
      "grad_norm": 2.3256800174713135,
      "learning_rate": 1.6886506127065464e-06,
      "loss": 0.4488,
      "step": 298370
    },
    {
      "epoch": 122.14,
      "grad_norm": 1.8789501190185547,
      "learning_rate": 1.688498307147812e-06,
      "loss": 0.4721,
      "step": 298380
    },
    {
      "epoch": 122.14,
      "grad_norm": 2.085120439529419,
      "learning_rate": 1.6883460045906825e-06,
      "loss": 0.477,
      "step": 298390
    },
    {
      "epoch": 122.14,
      "grad_norm": 1.8095667362213135,
      "learning_rate": 1.688193705035858e-06,
      "loss": 0.4785,
      "step": 298400
    },
    {
      "epoch": 122.15,
      "grad_norm": 1.64146089553833,
      "learning_rate": 1.6880414084840353e-06,
      "loss": 0.4659,
      "step": 298410
    },
    {
      "epoch": 122.15,
      "grad_norm": 1.7040014266967773,
      "learning_rate": 1.6878891149359124e-06,
      "loss": 0.4866,
      "step": 298420
    },
    {
      "epoch": 122.16,
      "grad_norm": 2.5994319915771484,
      "learning_rate": 1.6877368243921872e-06,
      "loss": 0.4548,
      "step": 298430
    },
    {
      "epoch": 122.16,
      "grad_norm": 2.021165609359741,
      "learning_rate": 1.6875845368535561e-06,
      "loss": 0.4452,
      "step": 298440
    },
    {
      "epoch": 122.17,
      "grad_norm": 2.3650100231170654,
      "learning_rate": 1.6874322523207176e-06,
      "loss": 0.4763,
      "step": 298450
    },
    {
      "epoch": 122.17,
      "grad_norm": 2.3821825981140137,
      "learning_rate": 1.687279970794369e-06,
      "loss": 0.4716,
      "step": 298460
    },
    {
      "epoch": 122.17,
      "grad_norm": 1.8431121110916138,
      "learning_rate": 1.6871276922752076e-06,
      "loss": 0.4674,
      "step": 298470
    },
    {
      "epoch": 122.18,
      "grad_norm": 2.2607178688049316,
      "learning_rate": 1.6869754167639315e-06,
      "loss": 0.4896,
      "step": 298480
    },
    {
      "epoch": 122.18,
      "grad_norm": 1.838201880455017,
      "learning_rate": 1.686823144261238e-06,
      "loss": 0.4659,
      "step": 298490
    },
    {
      "epoch": 122.19,
      "grad_norm": 2.1971983909606934,
      "learning_rate": 1.6866708747678249e-06,
      "loss": 0.4548,
      "step": 298500
    },
    {
      "epoch": 122.19,
      "grad_norm": 2.2822346687316895,
      "learning_rate": 1.6865186082843899e-06,
      "loss": 0.4506,
      "step": 298510
    },
    {
      "epoch": 122.19,
      "grad_norm": 1.7850425243377686,
      "learning_rate": 1.6863663448116295e-06,
      "loss": 0.4448,
      "step": 298520
    },
    {
      "epoch": 122.2,
      "grad_norm": 1.879440426826477,
      "learning_rate": 1.686214084350242e-06,
      "loss": 0.4537,
      "step": 298530
    },
    {
      "epoch": 122.2,
      "grad_norm": 2.0278303623199463,
      "learning_rate": 1.6860618269009242e-06,
      "loss": 0.4589,
      "step": 298540
    },
    {
      "epoch": 122.21,
      "grad_norm": 1.9656076431274414,
      "learning_rate": 1.6859095724643748e-06,
      "loss": 0.4484,
      "step": 298550
    },
    {
      "epoch": 122.21,
      "grad_norm": 2.212189197540283,
      "learning_rate": 1.6857573210412902e-06,
      "loss": 0.4383,
      "step": 298560
    },
    {
      "epoch": 122.21,
      "grad_norm": 2.1155972480773926,
      "learning_rate": 1.6856050726323679e-06,
      "loss": 0.4608,
      "step": 298570
    },
    {
      "epoch": 122.22,
      "grad_norm": 2.232520818710327,
      "learning_rate": 1.6854528272383054e-06,
      "loss": 0.4462,
      "step": 298580
    },
    {
      "epoch": 122.22,
      "grad_norm": 2.126525640487671,
      "learning_rate": 1.685300584859801e-06,
      "loss": 0.4603,
      "step": 298590
    },
    {
      "epoch": 122.23,
      "grad_norm": 2.5577127933502197,
      "learning_rate": 1.685148345497551e-06,
      "loss": 0.456,
      "step": 298600
    },
    {
      "epoch": 122.23,
      "grad_norm": 1.4927014112472534,
      "learning_rate": 1.6849961091522533e-06,
      "loss": 0.4575,
      "step": 298610
    },
    {
      "epoch": 122.23,
      "grad_norm": 1.8007839918136597,
      "learning_rate": 1.684843875824605e-06,
      "loss": 0.4554,
      "step": 298620
    },
    {
      "epoch": 122.24,
      "grad_norm": 2.0425148010253906,
      "learning_rate": 1.684691645515304e-06,
      "loss": 0.4673,
      "step": 298630
    },
    {
      "epoch": 122.24,
      "grad_norm": 1.6197187900543213,
      "learning_rate": 1.6845394182250469e-06,
      "loss": 0.4599,
      "step": 298640
    },
    {
      "epoch": 122.25,
      "grad_norm": 38.02088165283203,
      "learning_rate": 1.6843871939545322e-06,
      "loss": 0.4713,
      "step": 298650
    },
    {
      "epoch": 122.25,
      "grad_norm": 1.872488260269165,
      "learning_rate": 1.6842349727044554e-06,
      "loss": 0.4651,
      "step": 298660
    },
    {
      "epoch": 122.26,
      "grad_norm": 1.6980353593826294,
      "learning_rate": 1.6840827544755152e-06,
      "loss": 0.4472,
      "step": 298670
    },
    {
      "epoch": 122.26,
      "grad_norm": 2.3164143562316895,
      "learning_rate": 1.6839305392684082e-06,
      "loss": 0.4743,
      "step": 298680
    },
    {
      "epoch": 122.26,
      "grad_norm": 2.0036532878875732,
      "learning_rate": 1.6837783270838325e-06,
      "loss": 0.453,
      "step": 298690
    },
    {
      "epoch": 122.27,
      "grad_norm": 1.9797903299331665,
      "learning_rate": 1.6836261179224841e-06,
      "loss": 0.4658,
      "step": 298700
    },
    {
      "epoch": 122.27,
      "grad_norm": 1.9175387620925903,
      "learning_rate": 1.6834739117850619e-06,
      "loss": 0.468,
      "step": 298710
    },
    {
      "epoch": 122.28,
      "grad_norm": 2.11189603805542,
      "learning_rate": 1.6833217086722615e-06,
      "loss": 0.4581,
      "step": 298720
    },
    {
      "epoch": 122.28,
      "grad_norm": 1.9930915832519531,
      "learning_rate": 1.6831695085847807e-06,
      "loss": 0.4555,
      "step": 298730
    },
    {
      "epoch": 122.28,
      "grad_norm": 2.3645100593566895,
      "learning_rate": 1.683017311523318e-06,
      "loss": 0.4733,
      "step": 298740
    },
    {
      "epoch": 122.29,
      "grad_norm": 1.5004621744155884,
      "learning_rate": 1.6828651174885684e-06,
      "loss": 0.4367,
      "step": 298750
    },
    {
      "epoch": 122.29,
      "grad_norm": 1.9070104360580444,
      "learning_rate": 1.6827129264812302e-06,
      "loss": 0.4613,
      "step": 298760
    },
    {
      "epoch": 122.3,
      "grad_norm": 1.960275411605835,
      "learning_rate": 1.6825607385019995e-06,
      "loss": 0.47,
      "step": 298770
    },
    {
      "epoch": 122.3,
      "grad_norm": 2.3009181022644043,
      "learning_rate": 1.6824085535515753e-06,
      "loss": 0.4773,
      "step": 298780
    },
    {
      "epoch": 122.3,
      "grad_norm": 2.006239891052246,
      "learning_rate": 1.6822563716306546e-06,
      "loss": 0.4565,
      "step": 298790
    },
    {
      "epoch": 122.31,
      "grad_norm": 1.9586310386657715,
      "learning_rate": 1.682104192739933e-06,
      "loss": 0.4676,
      "step": 298800
    },
    {
      "epoch": 122.31,
      "grad_norm": 1.8628848791122437,
      "learning_rate": 1.6819520168801083e-06,
      "loss": 0.4715,
      "step": 298810
    },
    {
      "epoch": 122.32,
      "grad_norm": 2.2657017707824707,
      "learning_rate": 1.681799844051878e-06,
      "loss": 0.4619,
      "step": 298820
    },
    {
      "epoch": 122.32,
      "grad_norm": 2.0836408138275146,
      "learning_rate": 1.6816476742559387e-06,
      "loss": 0.4498,
      "step": 298830
    },
    {
      "epoch": 122.33,
      "grad_norm": 1.747094988822937,
      "learning_rate": 1.6814955074929876e-06,
      "loss": 0.4644,
      "step": 298840
    },
    {
      "epoch": 122.33,
      "grad_norm": 2.023252010345459,
      "learning_rate": 1.6813433437637216e-06,
      "loss": 0.4793,
      "step": 298850
    },
    {
      "epoch": 122.33,
      "grad_norm": 2.0521464347839355,
      "learning_rate": 1.681191183068838e-06,
      "loss": 0.4661,
      "step": 298860
    },
    {
      "epoch": 122.34,
      "grad_norm": 1.978488564491272,
      "learning_rate": 1.6810390254090344e-06,
      "loss": 0.4641,
      "step": 298870
    },
    {
      "epoch": 122.34,
      "grad_norm": 2.1943199634552,
      "learning_rate": 1.6808868707850067e-06,
      "loss": 0.4477,
      "step": 298880
    },
    {
      "epoch": 122.35,
      "grad_norm": 1.908868670463562,
      "learning_rate": 1.680734719197452e-06,
      "loss": 0.4632,
      "step": 298890
    },
    {
      "epoch": 122.35,
      "grad_norm": 2.495718479156494,
      "learning_rate": 1.6805825706470679e-06,
      "loss": 0.4598,
      "step": 298900
    },
    {
      "epoch": 122.35,
      "grad_norm": 1.8385955095291138,
      "learning_rate": 1.6804304251345512e-06,
      "loss": 0.4426,
      "step": 298910
    },
    {
      "epoch": 122.36,
      "grad_norm": 5.553786754608154,
      "learning_rate": 1.6802782826605983e-06,
      "loss": 0.5023,
      "step": 298920
    },
    {
      "epoch": 122.36,
      "grad_norm": 1.8726271390914917,
      "learning_rate": 1.680126143225907e-06,
      "loss": 0.4732,
      "step": 298930
    },
    {
      "epoch": 122.37,
      "grad_norm": 2.505431652069092,
      "learning_rate": 1.6799740068311735e-06,
      "loss": 0.4773,
      "step": 298940
    },
    {
      "epoch": 122.37,
      "grad_norm": 1.6894139051437378,
      "learning_rate": 1.6798218734770952e-06,
      "loss": 0.4632,
      "step": 298950
    },
    {
      "epoch": 122.37,
      "grad_norm": 3.2785990238189697,
      "learning_rate": 1.6796697431643683e-06,
      "loss": 0.4537,
      "step": 298960
    },
    {
      "epoch": 122.38,
      "grad_norm": 2.196693181991577,
      "learning_rate": 1.6795176158936905e-06,
      "loss": 0.4719,
      "step": 298970
    },
    {
      "epoch": 122.38,
      "grad_norm": 2.275592088699341,
      "learning_rate": 1.6793654916657588e-06,
      "loss": 0.4611,
      "step": 298980
    },
    {
      "epoch": 122.39,
      "grad_norm": 2.1526713371276855,
      "learning_rate": 1.6792133704812695e-06,
      "loss": 0.4638,
      "step": 298990
    },
    {
      "epoch": 122.39,
      "grad_norm": 2.4981489181518555,
      "learning_rate": 1.6790612523409195e-06,
      "loss": 0.4642,
      "step": 299000
    },
    {
      "epoch": 122.39,
      "grad_norm": 1.8931788206100464,
      "learning_rate": 1.678909137245406e-06,
      "loss": 0.4421,
      "step": 299010
    },
    {
      "epoch": 122.4,
      "grad_norm": 2.2225234508514404,
      "learning_rate": 1.6787570251954253e-06,
      "loss": 0.4658,
      "step": 299020
    },
    {
      "epoch": 122.4,
      "grad_norm": 1.8050647974014282,
      "learning_rate": 1.6786049161916746e-06,
      "loss": 0.4443,
      "step": 299030
    },
    {
      "epoch": 122.41,
      "grad_norm": 1.8323007822036743,
      "learning_rate": 1.6784528102348503e-06,
      "loss": 0.4507,
      "step": 299040
    },
    {
      "epoch": 122.41,
      "grad_norm": 2.3372068405151367,
      "learning_rate": 1.6783007073256495e-06,
      "loss": 0.4633,
      "step": 299050
    },
    {
      "epoch": 122.42,
      "grad_norm": 2.344778537750244,
      "learning_rate": 1.6781486074647686e-06,
      "loss": 0.4616,
      "step": 299060
    },
    {
      "epoch": 122.42,
      "grad_norm": 2.0169336795806885,
      "learning_rate": 1.6779965106529045e-06,
      "loss": 0.4685,
      "step": 299070
    },
    {
      "epoch": 122.42,
      "grad_norm": 1.8105366230010986,
      "learning_rate": 1.6778444168907547e-06,
      "loss": 0.4578,
      "step": 299080
    },
    {
      "epoch": 122.43,
      "grad_norm": 1.7774113416671753,
      "learning_rate": 1.6776923261790144e-06,
      "loss": 0.479,
      "step": 299090
    },
    {
      "epoch": 122.43,
      "grad_norm": 2.5444042682647705,
      "learning_rate": 1.6775402385183824e-06,
      "loss": 0.4561,
      "step": 299100
    },
    {
      "epoch": 122.44,
      "grad_norm": 3.5474836826324463,
      "learning_rate": 1.6773881539095532e-06,
      "loss": 0.4579,
      "step": 299110
    },
    {
      "epoch": 122.44,
      "grad_norm": 1.9637888669967651,
      "learning_rate": 1.6772360723532244e-06,
      "loss": 0.4838,
      "step": 299120
    },
    {
      "epoch": 122.44,
      "grad_norm": 2.374161958694458,
      "learning_rate": 1.677083993850093e-06,
      "loss": 0.4653,
      "step": 299130
    },
    {
      "epoch": 122.45,
      "grad_norm": 1.9814649820327759,
      "learning_rate": 1.6769319184008549e-06,
      "loss": 0.4686,
      "step": 299140
    },
    {
      "epoch": 122.45,
      "grad_norm": 1.5117857456207275,
      "learning_rate": 1.676779846006207e-06,
      "loss": 0.4323,
      "step": 299150
    },
    {
      "epoch": 122.46,
      "grad_norm": 2.1495249271392822,
      "learning_rate": 1.6766277766668465e-06,
      "loss": 0.4591,
      "step": 299160
    },
    {
      "epoch": 122.46,
      "grad_norm": 1.8396661281585693,
      "learning_rate": 1.6764757103834695e-06,
      "loss": 0.4654,
      "step": 299170
    },
    {
      "epoch": 122.46,
      "grad_norm": 1.8551818132400513,
      "learning_rate": 1.6763236471567727e-06,
      "loss": 0.4487,
      "step": 299180
    },
    {
      "epoch": 122.47,
      "grad_norm": 1.6365748643875122,
      "learning_rate": 1.6761715869874528e-06,
      "loss": 0.4781,
      "step": 299190
    },
    {
      "epoch": 122.47,
      "grad_norm": 2.2342803478240967,
      "learning_rate": 1.6760195298762062e-06,
      "loss": 0.4613,
      "step": 299200
    },
    {
      "epoch": 122.48,
      "grad_norm": 1.760133981704712,
      "learning_rate": 1.6758674758237296e-06,
      "loss": 0.4799,
      "step": 299210
    },
    {
      "epoch": 122.48,
      "grad_norm": 1.8933428525924683,
      "learning_rate": 1.675715424830719e-06,
      "loss": 0.4452,
      "step": 299220
    },
    {
      "epoch": 122.48,
      "grad_norm": 2.258178472518921,
      "learning_rate": 1.675563376897872e-06,
      "loss": 0.457,
      "step": 299230
    },
    {
      "epoch": 122.49,
      "grad_norm": 1.9667190313339233,
      "learning_rate": 1.6754113320258842e-06,
      "loss": 0.4653,
      "step": 299240
    },
    {
      "epoch": 122.49,
      "grad_norm": 2.3070037364959717,
      "learning_rate": 1.6752592902154517e-06,
      "loss": 0.4518,
      "step": 299250
    },
    {
      "epoch": 122.5,
      "grad_norm": 2.40226149559021,
      "learning_rate": 1.6751072514672723e-06,
      "loss": 0.4593,
      "step": 299260
    },
    {
      "epoch": 122.5,
      "grad_norm": 1.9775441884994507,
      "learning_rate": 1.6749552157820415e-06,
      "loss": 0.4571,
      "step": 299270
    },
    {
      "epoch": 122.51,
      "grad_norm": 2.2732975482940674,
      "learning_rate": 1.6748031831604558e-06,
      "loss": 0.4796,
      "step": 299280
    },
    {
      "epoch": 122.51,
      "grad_norm": 1.7594300508499146,
      "learning_rate": 1.6746511536032124e-06,
      "loss": 0.4614,
      "step": 299290
    },
    {
      "epoch": 122.51,
      "grad_norm": 2.2364611625671387,
      "learning_rate": 1.6744991271110068e-06,
      "loss": 0.4599,
      "step": 299300
    },
    {
      "epoch": 122.52,
      "grad_norm": 2.2917299270629883,
      "learning_rate": 1.674347103684536e-06,
      "loss": 0.4691,
      "step": 299310
    },
    {
      "epoch": 122.52,
      "grad_norm": 1.9638622999191284,
      "learning_rate": 1.6741950833244965e-06,
      "loss": 0.4445,
      "step": 299320
    },
    {
      "epoch": 122.53,
      "grad_norm": 1.800027847290039,
      "learning_rate": 1.6740430660315828e-06,
      "loss": 0.4548,
      "step": 299330
    },
    {
      "epoch": 122.53,
      "grad_norm": 2.126046895980835,
      "learning_rate": 1.6738910518064942e-06,
      "loss": 0.4484,
      "step": 299340
    },
    {
      "epoch": 122.53,
      "grad_norm": 1.9735268354415894,
      "learning_rate": 1.6737390406499253e-06,
      "loss": 0.4726,
      "step": 299350
    },
    {
      "epoch": 122.54,
      "grad_norm": 1.71373450756073,
      "learning_rate": 1.6735870325625732e-06,
      "loss": 0.4813,
      "step": 299360
    },
    {
      "epoch": 122.54,
      "grad_norm": 1.917568564414978,
      "learning_rate": 1.6734350275451344e-06,
      "loss": 0.4383,
      "step": 299370
    },
    {
      "epoch": 122.55,
      "grad_norm": 1.573307752609253,
      "learning_rate": 1.673283025598304e-06,
      "loss": 0.4388,
      "step": 299380
    },
    {
      "epoch": 122.55,
      "grad_norm": 2.1786046028137207,
      "learning_rate": 1.673131026722779e-06,
      "loss": 0.4733,
      "step": 299390
    },
    {
      "epoch": 122.55,
      "grad_norm": 2.228881359100342,
      "learning_rate": 1.6729790309192555e-06,
      "loss": 0.4368,
      "step": 299400
    },
    {
      "epoch": 122.56,
      "grad_norm": 1.8348108530044556,
      "learning_rate": 1.67282703818843e-06,
      "loss": 0.4415,
      "step": 299410
    },
    {
      "epoch": 122.56,
      "grad_norm": 1.8155964612960815,
      "learning_rate": 1.6726750485309988e-06,
      "loss": 0.4582,
      "step": 299420
    },
    {
      "epoch": 122.57,
      "grad_norm": 1.728218913078308,
      "learning_rate": 1.6725230619476581e-06,
      "loss": 0.4568,
      "step": 299430
    },
    {
      "epoch": 122.57,
      "grad_norm": 1.969225287437439,
      "learning_rate": 1.6723710784391038e-06,
      "loss": 0.4699,
      "step": 299440
    },
    {
      "epoch": 122.57,
      "grad_norm": 2.1336019039154053,
      "learning_rate": 1.6722190980060324e-06,
      "loss": 0.473,
      "step": 299450
    },
    {
      "epoch": 122.58,
      "grad_norm": 2.2432124614715576,
      "learning_rate": 1.6720671206491408e-06,
      "loss": 0.4673,
      "step": 299460
    },
    {
      "epoch": 122.58,
      "grad_norm": 2.140167236328125,
      "learning_rate": 1.6719151463691237e-06,
      "loss": 0.4548,
      "step": 299470
    },
    {
      "epoch": 122.59,
      "grad_norm": 2.0332889556884766,
      "learning_rate": 1.671763175166678e-06,
      "loss": 0.4544,
      "step": 299480
    },
    {
      "epoch": 122.59,
      "grad_norm": 2.4403812885284424,
      "learning_rate": 1.6716112070425002e-06,
      "loss": 0.4418,
      "step": 299490
    },
    {
      "epoch": 122.6,
      "grad_norm": 2.027791976928711,
      "learning_rate": 1.671459241997286e-06,
      "loss": 0.4704,
      "step": 299500
    },
    {
      "epoch": 122.6,
      "grad_norm": 1.9525552988052368,
      "learning_rate": 1.6713072800317313e-06,
      "loss": 0.4539,
      "step": 299510
    },
    {
      "epoch": 122.6,
      "grad_norm": 2.3189783096313477,
      "learning_rate": 1.6711553211465323e-06,
      "loss": 0.4693,
      "step": 299520
    },
    {
      "epoch": 122.61,
      "grad_norm": 1.6620457172393799,
      "learning_rate": 1.671003365342386e-06,
      "loss": 0.445,
      "step": 299530
    },
    {
      "epoch": 122.61,
      "grad_norm": 2.43670916557312,
      "learning_rate": 1.670851412619988e-06,
      "loss": 0.4795,
      "step": 299540
    },
    {
      "epoch": 122.62,
      "grad_norm": 1.773576259613037,
      "learning_rate": 1.6706994629800337e-06,
      "loss": 0.4471,
      "step": 299550
    },
    {
      "epoch": 122.62,
      "grad_norm": 1.989525318145752,
      "learning_rate": 1.67054751642322e-06,
      "loss": 0.461,
      "step": 299560
    },
    {
      "epoch": 122.62,
      "grad_norm": 2.3782989978790283,
      "learning_rate": 1.6703955729502424e-06,
      "loss": 0.46,
      "step": 299570
    },
    {
      "epoch": 122.63,
      "grad_norm": 1.973230242729187,
      "learning_rate": 1.6702436325617976e-06,
      "loss": 0.4347,
      "step": 299580
    },
    {
      "epoch": 122.63,
      "grad_norm": 2.041858673095703,
      "learning_rate": 1.6700916952585811e-06,
      "loss": 0.474,
      "step": 299590
    },
    {
      "epoch": 122.64,
      "grad_norm": 2.2485709190368652,
      "learning_rate": 1.6699397610412888e-06,
      "loss": 0.4736,
      "step": 299600
    },
    {
      "epoch": 122.64,
      "grad_norm": 1.8352534770965576,
      "learning_rate": 1.6697878299106166e-06,
      "loss": 0.4488,
      "step": 299610
    },
    {
      "epoch": 122.64,
      "grad_norm": 2.9982645511627197,
      "learning_rate": 1.6696359018672607e-06,
      "loss": 0.4643,
      "step": 299620
    },
    {
      "epoch": 122.65,
      "grad_norm": 1.9039274454116821,
      "learning_rate": 1.6694839769119175e-06,
      "loss": 0.4589,
      "step": 299630
    },
    {
      "epoch": 122.65,
      "grad_norm": 2.1151955127716064,
      "learning_rate": 1.6693320550452824e-06,
      "loss": 0.4695,
      "step": 299640
    },
    {
      "epoch": 122.66,
      "grad_norm": 1.6503548622131348,
      "learning_rate": 1.6691801362680514e-06,
      "loss": 0.4581,
      "step": 299650
    },
    {
      "epoch": 122.66,
      "grad_norm": 1.8655000925064087,
      "learning_rate": 1.6690282205809203e-06,
      "loss": 0.4704,
      "step": 299660
    },
    {
      "epoch": 122.66,
      "grad_norm": 1.8646329641342163,
      "learning_rate": 1.6688763079845853e-06,
      "loss": 0.4541,
      "step": 299670
    },
    {
      "epoch": 122.67,
      "grad_norm": 1.8437273502349854,
      "learning_rate": 1.6687243984797425e-06,
      "loss": 0.4639,
      "step": 299680
    },
    {
      "epoch": 122.67,
      "grad_norm": 2.149294376373291,
      "learning_rate": 1.6685724920670873e-06,
      "loss": 0.4649,
      "step": 299690
    },
    {
      "epoch": 122.68,
      "grad_norm": 2.0588319301605225,
      "learning_rate": 1.6684205887473157e-06,
      "loss": 0.473,
      "step": 299700
    },
    {
      "epoch": 122.68,
      "grad_norm": 2.2044074535369873,
      "learning_rate": 1.668268688521122e-06,
      "loss": 0.4692,
      "step": 299710
    },
    {
      "epoch": 122.69,
      "grad_norm": 1.6625134944915771,
      "learning_rate": 1.668116791389205e-06,
      "loss": 0.4493,
      "step": 299720
    },
    {
      "epoch": 122.69,
      "grad_norm": 2.1414377689361572,
      "learning_rate": 1.6679648973522594e-06,
      "loss": 0.48,
      "step": 299730
    },
    {
      "epoch": 122.69,
      "grad_norm": 1.8866268396377563,
      "learning_rate": 1.6678130064109805e-06,
      "loss": 0.4678,
      "step": 299740
    },
    {
      "epoch": 122.7,
      "grad_norm": 1.5873650312423706,
      "learning_rate": 1.6676611185660642e-06,
      "loss": 0.4456,
      "step": 299750
    },
    {
      "epoch": 122.7,
      "grad_norm": 1.7319027185440063,
      "learning_rate": 1.6675092338182064e-06,
      "loss": 0.4733,
      "step": 299760
    },
    {
      "epoch": 122.71,
      "grad_norm": 2.1156325340270996,
      "learning_rate": 1.6673573521681024e-06,
      "loss": 0.4775,
      "step": 299770
    },
    {
      "epoch": 122.71,
      "grad_norm": 1.8674439191818237,
      "learning_rate": 1.6672054736164487e-06,
      "loss": 0.4723,
      "step": 299780
    },
    {
      "epoch": 122.71,
      "grad_norm": 2.298614978790283,
      "learning_rate": 1.6670535981639408e-06,
      "loss": 0.4701,
      "step": 299790
    },
    {
      "epoch": 122.72,
      "grad_norm": 1.716017246246338,
      "learning_rate": 1.6669017258112739e-06,
      "loss": 0.4855,
      "step": 299800
    },
    {
      "epoch": 122.72,
      "grad_norm": 1.622233510017395,
      "learning_rate": 1.6667498565591447e-06,
      "loss": 0.4364,
      "step": 299810
    },
    {
      "epoch": 122.73,
      "grad_norm": 2.0824193954467773,
      "learning_rate": 1.6665979904082485e-06,
      "loss": 0.4772,
      "step": 299820
    },
    {
      "epoch": 122.73,
      "grad_norm": 2.0895836353302,
      "learning_rate": 1.6664461273592804e-06,
      "loss": 0.471,
      "step": 299830
    },
    {
      "epoch": 122.73,
      "grad_norm": 1.6222752332687378,
      "learning_rate": 1.666294267412936e-06,
      "loss": 0.4678,
      "step": 299840
    },
    {
      "epoch": 122.74,
      "grad_norm": 2.4916059970855713,
      "learning_rate": 1.6661424105699118e-06,
      "loss": 0.4686,
      "step": 299850
    },
    {
      "epoch": 122.74,
      "grad_norm": 1.6036858558654785,
      "learning_rate": 1.665990556830903e-06,
      "loss": 0.4478,
      "step": 299860
    },
    {
      "epoch": 122.75,
      "grad_norm": 1.8241616487503052,
      "learning_rate": 1.6658387061966053e-06,
      "loss": 0.4676,
      "step": 299870
    },
    {
      "epoch": 122.75,
      "grad_norm": 1.7732466459274292,
      "learning_rate": 1.6656868586677141e-06,
      "loss": 0.4545,
      "step": 299880
    },
    {
      "epoch": 122.75,
      "grad_norm": 1.7027288675308228,
      "learning_rate": 1.6655350142449252e-06,
      "loss": 0.4534,
      "step": 299890
    },
    {
      "epoch": 122.76,
      "grad_norm": 1.738939881324768,
      "learning_rate": 1.6653831729289337e-06,
      "loss": 0.4589,
      "step": 299900
    },
    {
      "epoch": 122.76,
      "grad_norm": 1.6502156257629395,
      "learning_rate": 1.665231334720436e-06,
      "loss": 0.4758,
      "step": 299910
    },
    {
      "epoch": 122.77,
      "grad_norm": 1.6082651615142822,
      "learning_rate": 1.665079499620127e-06,
      "loss": 0.4533,
      "step": 299920
    },
    {
      "epoch": 122.77,
      "grad_norm": 1.616927981376648,
      "learning_rate": 1.6649276676287026e-06,
      "loss": 0.4496,
      "step": 299930
    },
    {
      "epoch": 122.78,
      "grad_norm": 2.113987922668457,
      "learning_rate": 1.6647758387468583e-06,
      "loss": 0.4607,
      "step": 299940
    },
    {
      "epoch": 122.78,
      "grad_norm": 2.462031126022339,
      "learning_rate": 1.66462401297529e-06,
      "loss": 0.4474,
      "step": 299950
    },
    {
      "epoch": 122.78,
      "grad_norm": 1.5866857767105103,
      "learning_rate": 1.664472190314692e-06,
      "loss": 0.4878,
      "step": 299960
    },
    {
      "epoch": 122.79,
      "grad_norm": 1.7715531587600708,
      "learning_rate": 1.6643203707657606e-06,
      "loss": 0.4588,
      "step": 299970
    },
    {
      "epoch": 122.79,
      "grad_norm": 2.356916904449463,
      "learning_rate": 1.664168554329191e-06,
      "loss": 0.4507,
      "step": 299980
    },
    {
      "epoch": 122.8,
      "grad_norm": 2.2367074489593506,
      "learning_rate": 1.664016741005679e-06,
      "loss": 0.4657,
      "step": 299990
    },
    {
      "epoch": 122.8,
      "grad_norm": 1.839851975440979,
      "learning_rate": 1.6638649307959195e-06,
      "loss": 0.4791,
      "step": 300000
    },
    {
      "epoch": 122.8,
      "grad_norm": 2.1918318271636963,
      "learning_rate": 1.6637131237006082e-06,
      "loss": 0.4288,
      "step": 300010
    },
    {
      "epoch": 122.81,
      "grad_norm": 2.1318976879119873,
      "learning_rate": 1.6635613197204407e-06,
      "loss": 0.4632,
      "step": 300020
    },
    {
      "epoch": 122.81,
      "grad_norm": 2.139134645462036,
      "learning_rate": 1.663409518856112e-06,
      "loss": 0.442,
      "step": 300030
    },
    {
      "epoch": 122.82,
      "grad_norm": 3.1253206729888916,
      "learning_rate": 1.6632577211083184e-06,
      "loss": 0.4423,
      "step": 300040
    },
    {
      "epoch": 122.82,
      "grad_norm": 2.2588412761688232,
      "learning_rate": 1.6631059264777538e-06,
      "loss": 0.4688,
      "step": 300050
    },
    {
      "epoch": 122.82,
      "grad_norm": 1.748039960861206,
      "learning_rate": 1.6629541349651145e-06,
      "loss": 0.4316,
      "step": 300060
    },
    {
      "epoch": 122.83,
      "grad_norm": 2.1158432960510254,
      "learning_rate": 1.6628023465710955e-06,
      "loss": 0.4586,
      "step": 300070
    },
    {
      "epoch": 122.83,
      "grad_norm": 2.13535737991333,
      "learning_rate": 1.6626505612963922e-06,
      "loss": 0.4582,
      "step": 300080
    },
    {
      "epoch": 122.84,
      "grad_norm": 1.899895429611206,
      "learning_rate": 1.6624987791416998e-06,
      "loss": 0.4751,
      "step": 300090
    },
    {
      "epoch": 122.84,
      "grad_norm": 1.8071283102035522,
      "learning_rate": 1.662347000107714e-06,
      "loss": 0.4283,
      "step": 300100
    },
    {
      "epoch": 122.84,
      "grad_norm": 1.8685098886489868,
      "learning_rate": 1.66219522419513e-06,
      "loss": 0.4625,
      "step": 300110
    },
    {
      "epoch": 122.85,
      "grad_norm": 1.6811949014663696,
      "learning_rate": 1.6620434514046424e-06,
      "loss": 0.4507,
      "step": 300120
    },
    {
      "epoch": 122.85,
      "grad_norm": 1.7584787607192993,
      "learning_rate": 1.661891681736947e-06,
      "loss": 0.462,
      "step": 300130
    },
    {
      "epoch": 122.86,
      "grad_norm": 2.0021395683288574,
      "learning_rate": 1.6617399151927393e-06,
      "loss": 0.4419,
      "step": 300140
    },
    {
      "epoch": 122.86,
      "grad_norm": 1.7337247133255005,
      "learning_rate": 1.661588151772714e-06,
      "loss": 0.4539,
      "step": 300150
    },
    {
      "epoch": 122.87,
      "grad_norm": 1.8001234531402588,
      "learning_rate": 1.6614363914775666e-06,
      "loss": 0.465,
      "step": 300160
    },
    {
      "epoch": 122.87,
      "grad_norm": 1.7313793897628784,
      "learning_rate": 1.6612846343079924e-06,
      "loss": 0.4626,
      "step": 300170
    },
    {
      "epoch": 122.87,
      "grad_norm": 1.7753273248672485,
      "learning_rate": 1.661132880264686e-06,
      "loss": 0.4712,
      "step": 300180
    },
    {
      "epoch": 122.88,
      "grad_norm": 2.184061050415039,
      "learning_rate": 1.660981129348343e-06,
      "loss": 0.4533,
      "step": 300190
    },
    {
      "epoch": 122.88,
      "grad_norm": 1.6894738674163818,
      "learning_rate": 1.6608293815596584e-06,
      "loss": 0.4652,
      "step": 300200
    },
    {
      "epoch": 122.89,
      "grad_norm": 1.756176471710205,
      "learning_rate": 1.6606776368993273e-06,
      "loss": 0.446,
      "step": 300210
    },
    {
      "epoch": 122.89,
      "grad_norm": 1.701833963394165,
      "learning_rate": 1.660525895368045e-06,
      "loss": 0.4752,
      "step": 300220
    },
    {
      "epoch": 122.89,
      "grad_norm": 2.0335941314697266,
      "learning_rate": 1.6603741569665066e-06,
      "loss": 0.4839,
      "step": 300230
    },
    {
      "epoch": 122.9,
      "grad_norm": 1.6710764169692993,
      "learning_rate": 1.6602224216954072e-06,
      "loss": 0.4544,
      "step": 300240
    },
    {
      "epoch": 122.9,
      "grad_norm": 1.644408106803894,
      "learning_rate": 1.6600706895554415e-06,
      "loss": 0.4748,
      "step": 300250
    },
    {
      "epoch": 122.91,
      "grad_norm": 1.984668254852295,
      "learning_rate": 1.6599189605473054e-06,
      "loss": 0.4328,
      "step": 300260
    },
    {
      "epoch": 122.91,
      "grad_norm": 2.0872983932495117,
      "learning_rate": 1.659767234671693e-06,
      "loss": 0.4591,
      "step": 300270
    },
    {
      "epoch": 122.91,
      "grad_norm": 1.740607738494873,
      "learning_rate": 1.6596155119292987e-06,
      "loss": 0.4561,
      "step": 300280
    },
    {
      "epoch": 122.92,
      "grad_norm": 1.9103885889053345,
      "learning_rate": 1.6594637923208191e-06,
      "loss": 0.45,
      "step": 300290
    },
    {
      "epoch": 122.92,
      "grad_norm": 2.115034580230713,
      "learning_rate": 1.659312075846949e-06,
      "loss": 0.458,
      "step": 300300
    },
    {
      "epoch": 122.93,
      "grad_norm": 2.1995034217834473,
      "learning_rate": 1.6591603625083838e-06,
      "loss": 0.4902,
      "step": 300310
    },
    {
      "epoch": 122.93,
      "grad_norm": 2.178083658218384,
      "learning_rate": 1.659008652305817e-06,
      "loss": 0.475,
      "step": 300320
    },
    {
      "epoch": 122.93,
      "grad_norm": 2.219984769821167,
      "learning_rate": 1.6588569452399442e-06,
      "loss": 0.4706,
      "step": 300330
    },
    {
      "epoch": 122.94,
      "grad_norm": 1.9639400243759155,
      "learning_rate": 1.6587052413114605e-06,
      "loss": 0.4891,
      "step": 300340
    },
    {
      "epoch": 122.94,
      "grad_norm": 1.833795189857483,
      "learning_rate": 1.6585535405210607e-06,
      "loss": 0.4638,
      "step": 300350
    },
    {
      "epoch": 122.95,
      "grad_norm": 1.6840274333953857,
      "learning_rate": 1.6584018428694397e-06,
      "loss": 0.452,
      "step": 300360
    },
    {
      "epoch": 122.95,
      "grad_norm": 2.4645259380340576,
      "learning_rate": 1.6582501483572923e-06,
      "loss": 0.4627,
      "step": 300370
    },
    {
      "epoch": 122.96,
      "grad_norm": 2.1455442905426025,
      "learning_rate": 1.6580984569853138e-06,
      "loss": 0.4626,
      "step": 300380
    },
    {
      "epoch": 122.96,
      "grad_norm": 1.8817728757858276,
      "learning_rate": 1.6579467687541988e-06,
      "loss": 0.4768,
      "step": 300390
    },
    {
      "epoch": 122.96,
      "grad_norm": 2.0051417350769043,
      "learning_rate": 1.6577950836646427e-06,
      "loss": 0.4657,
      "step": 300400
    },
    {
      "epoch": 122.97,
      "grad_norm": 2.241368055343628,
      "learning_rate": 1.6576434017173393e-06,
      "loss": 0.4577,
      "step": 300410
    },
    {
      "epoch": 122.97,
      "grad_norm": 1.8939549922943115,
      "learning_rate": 1.6574917229129843e-06,
      "loss": 0.4677,
      "step": 300420
    },
    {
      "epoch": 122.98,
      "grad_norm": 3.3174474239349365,
      "learning_rate": 1.6573400472522717e-06,
      "loss": 0.4599,
      "step": 300430
    },
    {
      "epoch": 122.98,
      "grad_norm": 2.0675973892211914,
      "learning_rate": 1.657188374735897e-06,
      "loss": 0.4583,
      "step": 300440
    },
    {
      "epoch": 122.98,
      "grad_norm": 1.578279972076416,
      "learning_rate": 1.657036705364555e-06,
      "loss": 0.4548,
      "step": 300450
    },
    {
      "epoch": 122.99,
      "grad_norm": 2.3763277530670166,
      "learning_rate": 1.6568850391389397e-06,
      "loss": 0.4504,
      "step": 300460
    },
    {
      "epoch": 122.99,
      "grad_norm": 1.722421407699585,
      "learning_rate": 1.6567333760597462e-06,
      "loss": 0.4582,
      "step": 300470
    },
    {
      "epoch": 123.0,
      "grad_norm": 2.2857441902160645,
      "learning_rate": 1.6565817161276698e-06,
      "loss": 0.4757,
      "step": 300480
    },
    {
      "epoch": 123.0,
      "eval_loss": 0.46198371052742004,
      "eval_runtime": 52.2467,
      "eval_samples_per_second": 66.014,
      "eval_steps_per_second": 8.268,
      "step": 300489
    },
    {
      "epoch": 123.0,
      "grad_norm": 2.046290159225464,
      "learning_rate": 1.6564300593434052e-06,
      "loss": 0.4616,
      "step": 300490
    },
    {
      "epoch": 123.0,
      "grad_norm": 2.127769708633423,
      "learning_rate": 1.6562784057076465e-06,
      "loss": 0.4569,
      "step": 300500
    },
    {
      "epoch": 123.01,
      "grad_norm": 2.091614007949829,
      "learning_rate": 1.656126755221089e-06,
      "loss": 0.4481,
      "step": 300510
    },
    {
      "epoch": 123.01,
      "grad_norm": 1.783374547958374,
      "learning_rate": 1.655975107884427e-06,
      "loss": 0.466,
      "step": 300520
    },
    {
      "epoch": 123.02,
      "grad_norm": 2.544826030731201,
      "learning_rate": 1.6558234636983558e-06,
      "loss": 0.4789,
      "step": 300530
    },
    {
      "epoch": 123.02,
      "grad_norm": 2.141570568084717,
      "learning_rate": 1.655671822663569e-06,
      "loss": 0.442,
      "step": 300540
    },
    {
      "epoch": 123.02,
      "grad_norm": 2.184236526489258,
      "learning_rate": 1.6555201847807617e-06,
      "loss": 0.454,
      "step": 300550
    },
    {
      "epoch": 123.03,
      "grad_norm": 1.7342805862426758,
      "learning_rate": 1.6553685500506285e-06,
      "loss": 0.4454,
      "step": 300560
    },
    {
      "epoch": 123.03,
      "grad_norm": 1.7245374917984009,
      "learning_rate": 1.6552169184738643e-06,
      "loss": 0.4523,
      "step": 300570
    },
    {
      "epoch": 123.04,
      "grad_norm": 1.9643906354904175,
      "learning_rate": 1.6550652900511637e-06,
      "loss": 0.4787,
      "step": 300580
    },
    {
      "epoch": 123.04,
      "grad_norm": 2.177687883377075,
      "learning_rate": 1.6549136647832207e-06,
      "loss": 0.4604,
      "step": 300590
    },
    {
      "epoch": 123.05,
      "grad_norm": 2.491856575012207,
      "learning_rate": 1.6547620426707304e-06,
      "loss": 0.48,
      "step": 300600
    },
    {
      "epoch": 123.05,
      "grad_norm": 1.9332836866378784,
      "learning_rate": 1.6546104237143872e-06,
      "loss": 0.4969,
      "step": 300610
    },
    {
      "epoch": 123.05,
      "grad_norm": 1.7970128059387207,
      "learning_rate": 1.6544588079148864e-06,
      "loss": 0.4514,
      "step": 300620
    },
    {
      "epoch": 123.06,
      "grad_norm": 1.7721223831176758,
      "learning_rate": 1.654307195272921e-06,
      "loss": 0.4736,
      "step": 300630
    },
    {
      "epoch": 123.06,
      "grad_norm": 2.2278287410736084,
      "learning_rate": 1.6541555857891864e-06,
      "loss": 0.4513,
      "step": 300640
    },
    {
      "epoch": 123.07,
      "grad_norm": 1.7446441650390625,
      "learning_rate": 1.6540039794643762e-06,
      "loss": 0.4575,
      "step": 300650
    },
    {
      "epoch": 123.07,
      "grad_norm": 2.0746655464172363,
      "learning_rate": 1.6538523762991864e-06,
      "loss": 0.4435,
      "step": 300660
    },
    {
      "epoch": 123.07,
      "grad_norm": 2.2480616569519043,
      "learning_rate": 1.6537007762943113e-06,
      "loss": 0.4532,
      "step": 300670
    },
    {
      "epoch": 123.08,
      "grad_norm": 2.222278594970703,
      "learning_rate": 1.6535491794504446e-06,
      "loss": 0.4546,
      "step": 300680
    },
    {
      "epoch": 123.08,
      "grad_norm": 1.9031226634979248,
      "learning_rate": 1.6533975857682806e-06,
      "loss": 0.4719,
      "step": 300690
    },
    {
      "epoch": 123.09,
      "grad_norm": 2.0813658237457275,
      "learning_rate": 1.6532459952485146e-06,
      "loss": 0.4577,
      "step": 300700
    },
    {
      "epoch": 123.09,
      "grad_norm": 2.2393758296966553,
      "learning_rate": 1.6530944078918402e-06,
      "loss": 0.4624,
      "step": 300710
    },
    {
      "epoch": 123.09,
      "grad_norm": 1.8962945938110352,
      "learning_rate": 1.652942823698952e-06,
      "loss": 0.4748,
      "step": 300720
    },
    {
      "epoch": 123.1,
      "grad_norm": 1.9761220216751099,
      "learning_rate": 1.6527912426705444e-06,
      "loss": 0.4541,
      "step": 300730
    },
    {
      "epoch": 123.1,
      "grad_norm": 2.0579028129577637,
      "learning_rate": 1.6526396648073123e-06,
      "loss": 0.4463,
      "step": 300740
    },
    {
      "epoch": 123.11,
      "grad_norm": 1.9619574546813965,
      "learning_rate": 1.6524880901099492e-06,
      "loss": 0.4577,
      "step": 300750
    },
    {
      "epoch": 123.11,
      "grad_norm": 1.7615234851837158,
      "learning_rate": 1.652336518579151e-06,
      "loss": 0.4379,
      "step": 300760
    },
    {
      "epoch": 123.12,
      "grad_norm": 1.9973677396774292,
      "learning_rate": 1.6521849502156096e-06,
      "loss": 0.4405,
      "step": 300770
    },
    {
      "epoch": 123.12,
      "grad_norm": 2.1281023025512695,
      "learning_rate": 1.652033385020021e-06,
      "loss": 0.4565,
      "step": 300780
    },
    {
      "epoch": 123.12,
      "grad_norm": 2.35461688041687,
      "learning_rate": 1.651881822993079e-06,
      "loss": 0.4594,
      "step": 300790
    },
    {
      "epoch": 123.13,
      "grad_norm": 2.4767191410064697,
      "learning_rate": 1.651730264135478e-06,
      "loss": 0.4674,
      "step": 300800
    },
    {
      "epoch": 123.13,
      "grad_norm": 1.786179542541504,
      "learning_rate": 1.6515787084479121e-06,
      "loss": 0.4509,
      "step": 300810
    },
    {
      "epoch": 123.14,
      "grad_norm": 1.8610795736312866,
      "learning_rate": 1.651427155931076e-06,
      "loss": 0.4601,
      "step": 300820
    },
    {
      "epoch": 123.14,
      "grad_norm": 1.980469822883606,
      "learning_rate": 1.6512756065856631e-06,
      "loss": 0.4524,
      "step": 300830
    },
    {
      "epoch": 123.14,
      "grad_norm": 1.7374697923660278,
      "learning_rate": 1.6511240604123682e-06,
      "loss": 0.4647,
      "step": 300840
    },
    {
      "epoch": 123.15,
      "grad_norm": 2.190737009048462,
      "learning_rate": 1.6509725174118857e-06,
      "loss": 0.4406,
      "step": 300850
    },
    {
      "epoch": 123.15,
      "grad_norm": 1.7185909748077393,
      "learning_rate": 1.6508209775849094e-06,
      "loss": 0.4764,
      "step": 300860
    },
    {
      "epoch": 123.16,
      "grad_norm": 1.7191989421844482,
      "learning_rate": 1.650669440932134e-06,
      "loss": 0.4615,
      "step": 300870
    },
    {
      "epoch": 123.16,
      "grad_norm": 2.7281124591827393,
      "learning_rate": 1.6505179074542531e-06,
      "loss": 0.4439,
      "step": 300880
    },
    {
      "epoch": 123.16,
      "grad_norm": 2.189706563949585,
      "learning_rate": 1.6503663771519613e-06,
      "loss": 0.4728,
      "step": 300890
    },
    {
      "epoch": 123.17,
      "grad_norm": 1.6485743522644043,
      "learning_rate": 1.6502148500259526e-06,
      "loss": 0.471,
      "step": 300900
    },
    {
      "epoch": 123.17,
      "grad_norm": 1.921901822090149,
      "learning_rate": 1.6500633260769206e-06,
      "loss": 0.4846,
      "step": 300910
    },
    {
      "epoch": 123.18,
      "grad_norm": 1.847488522529602,
      "learning_rate": 1.64991180530556e-06,
      "loss": 0.4541,
      "step": 300920
    },
    {
      "epoch": 123.18,
      "grad_norm": 2.272841453552246,
      "learning_rate": 1.6497602877125647e-06,
      "loss": 0.4379,
      "step": 300930
    },
    {
      "epoch": 123.18,
      "grad_norm": 2.1194875240325928,
      "learning_rate": 1.6496087732986289e-06,
      "loss": 0.4794,
      "step": 300940
    },
    {
      "epoch": 123.19,
      "grad_norm": 2.611480474472046,
      "learning_rate": 1.649457262064446e-06,
      "loss": 0.4733,
      "step": 300950
    },
    {
      "epoch": 123.19,
      "grad_norm": 1.7072663307189941,
      "learning_rate": 1.649305754010711e-06,
      "loss": 0.4657,
      "step": 300960
    },
    {
      "epoch": 123.2,
      "grad_norm": 2.001155138015747,
      "learning_rate": 1.6491542491381176e-06,
      "loss": 0.4859,
      "step": 300970
    },
    {
      "epoch": 123.2,
      "grad_norm": 1.955368995666504,
      "learning_rate": 1.6490027474473602e-06,
      "loss": 0.4537,
      "step": 300980
    },
    {
      "epoch": 123.21,
      "grad_norm": 1.9404778480529785,
      "learning_rate": 1.648851248939132e-06,
      "loss": 0.464,
      "step": 300990
    },
    {
      "epoch": 123.21,
      "grad_norm": 1.990305781364441,
      "learning_rate": 1.648699753614127e-06,
      "loss": 0.4616,
      "step": 301000
    },
    {
      "epoch": 123.21,
      "grad_norm": 2.302579402923584,
      "learning_rate": 1.6485482614730402e-06,
      "loss": 0.444,
      "step": 301010
    },
    {
      "epoch": 123.22,
      "grad_norm": 1.6590253114700317,
      "learning_rate": 1.6483967725165642e-06,
      "loss": 0.4549,
      "step": 301020
    },
    {
      "epoch": 123.22,
      "grad_norm": 2.263031244277954,
      "learning_rate": 1.6482452867453933e-06,
      "loss": 0.4783,
      "step": 301030
    },
    {
      "epoch": 123.23,
      "grad_norm": 2.1604368686676025,
      "learning_rate": 1.6480938041602223e-06,
      "loss": 0.4693,
      "step": 301040
    },
    {
      "epoch": 123.23,
      "grad_norm": 3.243943691253662,
      "learning_rate": 1.6479423247617451e-06,
      "loss": 0.474,
      "step": 301050
    },
    {
      "epoch": 123.23,
      "grad_norm": 1.99216628074646,
      "learning_rate": 1.6477908485506546e-06,
      "loss": 0.4662,
      "step": 301060
    },
    {
      "epoch": 123.24,
      "grad_norm": 1.9964247941970825,
      "learning_rate": 1.647639375527645e-06,
      "loss": 0.4597,
      "step": 301070
    },
    {
      "epoch": 123.24,
      "grad_norm": 1.8801504373550415,
      "learning_rate": 1.647487905693411e-06,
      "loss": 0.4662,
      "step": 301080
    },
    {
      "epoch": 123.25,
      "grad_norm": 2.0354363918304443,
      "learning_rate": 1.6473364390486453e-06,
      "loss": 0.4611,
      "step": 301090
    },
    {
      "epoch": 123.25,
      "grad_norm": 1.4554965496063232,
      "learning_rate": 1.6471849755940423e-06,
      "loss": 0.4639,
      "step": 301100
    },
    {
      "epoch": 123.25,
      "grad_norm": 1.8020637035369873,
      "learning_rate": 1.647033515330296e-06,
      "loss": 0.4895,
      "step": 301110
    },
    {
      "epoch": 123.26,
      "grad_norm": 1.7027555704116821,
      "learning_rate": 1.6468820582581005e-06,
      "loss": 0.4561,
      "step": 301120
    },
    {
      "epoch": 123.26,
      "grad_norm": 1.7332772016525269,
      "learning_rate": 1.6467306043781484e-06,
      "loss": 0.4575,
      "step": 301130
    },
    {
      "epoch": 123.27,
      "grad_norm": 2.0383098125457764,
      "learning_rate": 1.6465791536911345e-06,
      "loss": 0.4538,
      "step": 301140
    },
    {
      "epoch": 123.27,
      "grad_norm": 1.7896757125854492,
      "learning_rate": 1.6464277061977523e-06,
      "loss": 0.4693,
      "step": 301150
    },
    {
      "epoch": 123.27,
      "grad_norm": 2.200061082839966,
      "learning_rate": 1.6462762618986954e-06,
      "loss": 0.4733,
      "step": 301160
    },
    {
      "epoch": 123.28,
      "grad_norm": 1.9139090776443481,
      "learning_rate": 1.6461248207946576e-06,
      "loss": 0.4374,
      "step": 301170
    },
    {
      "epoch": 123.28,
      "grad_norm": 2.19018292427063,
      "learning_rate": 1.6459733828863327e-06,
      "loss": 0.4722,
      "step": 301180
    },
    {
      "epoch": 123.29,
      "grad_norm": 1.4333804845809937,
      "learning_rate": 1.6458219481744146e-06,
      "loss": 0.455,
      "step": 301190
    },
    {
      "epoch": 123.29,
      "grad_norm": 1.864156723022461,
      "learning_rate": 1.6456705166595974e-06,
      "loss": 0.4517,
      "step": 301200
    },
    {
      "epoch": 123.3,
      "grad_norm": 1.8840975761413574,
      "learning_rate": 1.6455190883425731e-06,
      "loss": 0.4552,
      "step": 301210
    },
    {
      "epoch": 123.3,
      "grad_norm": 1.8215649127960205,
      "learning_rate": 1.645367663224036e-06,
      "loss": 0.4644,
      "step": 301220
    },
    {
      "epoch": 123.3,
      "grad_norm": 1.8480337858200073,
      "learning_rate": 1.645216241304681e-06,
      "loss": 0.4686,
      "step": 301230
    },
    {
      "epoch": 123.31,
      "grad_norm": 1.6655789613723755,
      "learning_rate": 1.6450648225852015e-06,
      "loss": 0.443,
      "step": 301240
    },
    {
      "epoch": 123.31,
      "grad_norm": 1.723686933517456,
      "learning_rate": 1.6449134070662904e-06,
      "loss": 0.4702,
      "step": 301250
    },
    {
      "epoch": 123.32,
      "grad_norm": 1.8710678815841675,
      "learning_rate": 1.6447619947486415e-06,
      "loss": 0.4494,
      "step": 301260
    },
    {
      "epoch": 123.32,
      "grad_norm": 1.7742975950241089,
      "learning_rate": 1.644610585632948e-06,
      "loss": 0.449,
      "step": 301270
    },
    {
      "epoch": 123.32,
      "grad_norm": 1.9688951969146729,
      "learning_rate": 1.644459179719904e-06,
      "loss": 0.4468,
      "step": 301280
    },
    {
      "epoch": 123.33,
      "grad_norm": 2.0581161975860596,
      "learning_rate": 1.644307777010203e-06,
      "loss": 0.4618,
      "step": 301290
    },
    {
      "epoch": 123.33,
      "grad_norm": 2.121504545211792,
      "learning_rate": 1.6441563775045387e-06,
      "loss": 0.4711,
      "step": 301300
    },
    {
      "epoch": 123.34,
      "grad_norm": 1.8805599212646484,
      "learning_rate": 1.644004981203604e-06,
      "loss": 0.4778,
      "step": 301310
    },
    {
      "epoch": 123.34,
      "grad_norm": 2.040764331817627,
      "learning_rate": 1.6438535881080933e-06,
      "loss": 0.4663,
      "step": 301320
    },
    {
      "epoch": 123.34,
      "grad_norm": 1.808441162109375,
      "learning_rate": 1.6437021982186994e-06,
      "loss": 0.4738,
      "step": 301330
    },
    {
      "epoch": 123.35,
      "grad_norm": 1.971651315689087,
      "learning_rate": 1.6435508115361166e-06,
      "loss": 0.4598,
      "step": 301340
    },
    {
      "epoch": 123.35,
      "grad_norm": 1.7611204385757446,
      "learning_rate": 1.6433994280610372e-06,
      "loss": 0.4704,
      "step": 301350
    },
    {
      "epoch": 123.36,
      "grad_norm": 2.5446035861968994,
      "learning_rate": 1.6432480477941558e-06,
      "loss": 0.4606,
      "step": 301360
    },
    {
      "epoch": 123.36,
      "grad_norm": 2.188595771789551,
      "learning_rate": 1.6430966707361648e-06,
      "loss": 0.4639,
      "step": 301370
    },
    {
      "epoch": 123.36,
      "grad_norm": 1.8119678497314453,
      "learning_rate": 1.6429452968877587e-06,
      "loss": 0.4526,
      "step": 301380
    },
    {
      "epoch": 123.37,
      "grad_norm": 1.7743037939071655,
      "learning_rate": 1.64279392624963e-06,
      "loss": 0.455,
      "step": 301390
    },
    {
      "epoch": 123.37,
      "grad_norm": 2.121610641479492,
      "learning_rate": 1.6426425588224728e-06,
      "loss": 0.4443,
      "step": 301400
    },
    {
      "epoch": 123.38,
      "grad_norm": 1.975408911705017,
      "learning_rate": 1.6424911946069793e-06,
      "loss": 0.4446,
      "step": 301410
    },
    {
      "epoch": 123.38,
      "grad_norm": 1.6778430938720703,
      "learning_rate": 1.6423398336038448e-06,
      "loss": 0.4463,
      "step": 301420
    },
    {
      "epoch": 123.39,
      "grad_norm": 1.987163782119751,
      "learning_rate": 1.6421884758137612e-06,
      "loss": 0.4446,
      "step": 301430
    },
    {
      "epoch": 123.39,
      "grad_norm": 1.8856751918792725,
      "learning_rate": 1.6420371212374225e-06,
      "loss": 0.4439,
      "step": 301440
    },
    {
      "epoch": 123.39,
      "grad_norm": 1.913891077041626,
      "learning_rate": 1.6418857698755217e-06,
      "loss": 0.4646,
      "step": 301450
    },
    {
      "epoch": 123.4,
      "grad_norm": 1.9101065397262573,
      "learning_rate": 1.6417344217287523e-06,
      "loss": 0.4267,
      "step": 301460
    },
    {
      "epoch": 123.4,
      "grad_norm": 1.889959454536438,
      "learning_rate": 1.641583076797808e-06,
      "loss": 0.4539,
      "step": 301470
    },
    {
      "epoch": 123.41,
      "grad_norm": 1.616686463356018,
      "learning_rate": 1.6414317350833816e-06,
      "loss": 0.4758,
      "step": 301480
    },
    {
      "epoch": 123.41,
      "grad_norm": 1.5792319774627686,
      "learning_rate": 1.641280396586166e-06,
      "loss": 0.4493,
      "step": 301490
    },
    {
      "epoch": 123.41,
      "grad_norm": 1.6781032085418701,
      "learning_rate": 1.6411290613068547e-06,
      "loss": 0.4682,
      "step": 301500
    },
    {
      "epoch": 123.42,
      "grad_norm": 2.3145627975463867,
      "learning_rate": 1.6409777292461413e-06,
      "loss": 0.4544,
      "step": 301510
    },
    {
      "epoch": 123.42,
      "grad_norm": 1.8127710819244385,
      "learning_rate": 1.6408264004047188e-06,
      "loss": 0.4536,
      "step": 301520
    },
    {
      "epoch": 123.43,
      "grad_norm": 2.4514827728271484,
      "learning_rate": 1.6406750747832806e-06,
      "loss": 0.4553,
      "step": 301530
    },
    {
      "epoch": 123.43,
      "grad_norm": 3.314687728881836,
      "learning_rate": 1.6405237523825196e-06,
      "loss": 0.4569,
      "step": 301540
    },
    {
      "epoch": 123.43,
      "grad_norm": 1.6103832721710205,
      "learning_rate": 1.6403724332031294e-06,
      "loss": 0.4624,
      "step": 301550
    },
    {
      "epoch": 123.44,
      "grad_norm": 1.8967286348342896,
      "learning_rate": 1.6402211172458033e-06,
      "loss": 0.4572,
      "step": 301560
    },
    {
      "epoch": 123.44,
      "grad_norm": 1.9448068141937256,
      "learning_rate": 1.640069804511233e-06,
      "loss": 0.4562,
      "step": 301570
    },
    {
      "epoch": 123.45,
      "grad_norm": 2.3140318393707275,
      "learning_rate": 1.6399184950001133e-06,
      "loss": 0.4432,
      "step": 301580
    },
    {
      "epoch": 123.45,
      "grad_norm": 2.2443084716796875,
      "learning_rate": 1.6397671887131365e-06,
      "loss": 0.4587,
      "step": 301590
    },
    {
      "epoch": 123.45,
      "grad_norm": 2.276242733001709,
      "learning_rate": 1.639615885650995e-06,
      "loss": 0.4625,
      "step": 301600
    },
    {
      "epoch": 123.46,
      "grad_norm": 2.745335817337036,
      "learning_rate": 1.6394645858143844e-06,
      "loss": 0.477,
      "step": 301610
    },
    {
      "epoch": 123.46,
      "grad_norm": 1.6956202983856201,
      "learning_rate": 1.6393132892039953e-06,
      "loss": 0.4499,
      "step": 301620
    },
    {
      "epoch": 123.47,
      "grad_norm": 1.9296938180923462,
      "learning_rate": 1.6391619958205223e-06,
      "loss": 0.4649,
      "step": 301630
    },
    {
      "epoch": 123.47,
      "grad_norm": 2.073681354522705,
      "learning_rate": 1.6390107056646575e-06,
      "loss": 0.4646,
      "step": 301640
    },
    {
      "epoch": 123.48,
      "grad_norm": 1.9169161319732666,
      "learning_rate": 1.6388594187370943e-06,
      "loss": 0.4675,
      "step": 301650
    },
    {
      "epoch": 123.48,
      "grad_norm": 2.025656223297119,
      "learning_rate": 1.6387081350385257e-06,
      "loss": 0.4492,
      "step": 301660
    },
    {
      "epoch": 123.48,
      "grad_norm": 2.230564832687378,
      "learning_rate": 1.6385568545696445e-06,
      "loss": 0.4845,
      "step": 301670
    },
    {
      "epoch": 123.49,
      "grad_norm": 1.9559465646743774,
      "learning_rate": 1.6384055773311442e-06,
      "loss": 0.4598,
      "step": 301680
    },
    {
      "epoch": 123.49,
      "grad_norm": 1.8542593717575073,
      "learning_rate": 1.6382543033237171e-06,
      "loss": 0.4668,
      "step": 301690
    },
    {
      "epoch": 123.5,
      "grad_norm": 1.9150031805038452,
      "learning_rate": 1.6381030325480574e-06,
      "loss": 0.4842,
      "step": 301700
    },
    {
      "epoch": 123.5,
      "grad_norm": 1.5379616022109985,
      "learning_rate": 1.6379517650048563e-06,
      "loss": 0.4532,
      "step": 301710
    },
    {
      "epoch": 123.5,
      "grad_norm": 2.0834667682647705,
      "learning_rate": 1.637800500694808e-06,
      "loss": 0.4758,
      "step": 301720
    },
    {
      "epoch": 123.51,
      "grad_norm": 1.8315917253494263,
      "learning_rate": 1.6376492396186049e-06,
      "loss": 0.4693,
      "step": 301730
    },
    {
      "epoch": 123.51,
      "grad_norm": 2.260653257369995,
      "learning_rate": 1.63749798177694e-06,
      "loss": 0.4593,
      "step": 301740
    },
    {
      "epoch": 123.52,
      "grad_norm": 2.077453851699829,
      "learning_rate": 1.637346727170506e-06,
      "loss": 0.4445,
      "step": 301750
    },
    {
      "epoch": 123.52,
      "grad_norm": 1.7656352519989014,
      "learning_rate": 1.6371954757999963e-06,
      "loss": 0.4838,
      "step": 301760
    },
    {
      "epoch": 123.52,
      "grad_norm": 1.645835280418396,
      "learning_rate": 1.6370442276661033e-06,
      "loss": 0.4696,
      "step": 301770
    },
    {
      "epoch": 123.53,
      "grad_norm": 1.7564195394515991,
      "learning_rate": 1.63689298276952e-06,
      "loss": 0.4795,
      "step": 301780
    },
    {
      "epoch": 123.53,
      "grad_norm": 1.9449448585510254,
      "learning_rate": 1.636741741110939e-06,
      "loss": 0.4474,
      "step": 301790
    },
    {
      "epoch": 123.54,
      "grad_norm": 1.9380041360855103,
      "learning_rate": 1.6365905026910537e-06,
      "loss": 0.4802,
      "step": 301800
    },
    {
      "epoch": 123.54,
      "grad_norm": 2.262449026107788,
      "learning_rate": 1.6364392675105563e-06,
      "loss": 0.4442,
      "step": 301810
    },
    {
      "epoch": 123.54,
      "grad_norm": 1.477100133895874,
      "learning_rate": 1.63628803557014e-06,
      "loss": 0.4796,
      "step": 301820
    },
    {
      "epoch": 123.55,
      "grad_norm": 1.77494215965271,
      "learning_rate": 1.6361368068704974e-06,
      "loss": 0.465,
      "step": 301830
    },
    {
      "epoch": 123.55,
      "grad_norm": 1.7726818323135376,
      "learning_rate": 1.6359855814123216e-06,
      "loss": 0.4525,
      "step": 301840
    },
    {
      "epoch": 123.56,
      "grad_norm": 2.0922858715057373,
      "learning_rate": 1.6358343591963046e-06,
      "loss": 0.4674,
      "step": 301850
    },
    {
      "epoch": 123.56,
      "grad_norm": 2.089960813522339,
      "learning_rate": 1.6356831402231395e-06,
      "loss": 0.4385,
      "step": 301860
    },
    {
      "epoch": 123.57,
      "grad_norm": 2.1575639247894287,
      "learning_rate": 1.6355319244935191e-06,
      "loss": 0.4561,
      "step": 301870
    },
    {
      "epoch": 123.57,
      "grad_norm": 1.9580743312835693,
      "learning_rate": 1.635380712008136e-06,
      "loss": 0.4723,
      "step": 301880
    },
    {
      "epoch": 123.57,
      "grad_norm": 2.166247844696045,
      "learning_rate": 1.6352295027676828e-06,
      "loss": 0.4447,
      "step": 301890
    },
    {
      "epoch": 123.58,
      "grad_norm": 1.9410609006881714,
      "learning_rate": 1.6350782967728522e-06,
      "loss": 0.4506,
      "step": 301900
    },
    {
      "epoch": 123.58,
      "grad_norm": 2.421455144882202,
      "learning_rate": 1.634927094024337e-06,
      "loss": 0.4753,
      "step": 301910
    },
    {
      "epoch": 123.59,
      "grad_norm": 2.1273863315582275,
      "learning_rate": 1.6347758945228305e-06,
      "loss": 0.4491,
      "step": 301920
    },
    {
      "epoch": 123.59,
      "grad_norm": 2.0720882415771484,
      "learning_rate": 1.6346246982690238e-06,
      "loss": 0.4663,
      "step": 301930
    },
    {
      "epoch": 123.59,
      "grad_norm": 1.8474609851837158,
      "learning_rate": 1.6344735052636105e-06,
      "loss": 0.4629,
      "step": 301940
    },
    {
      "epoch": 123.6,
      "grad_norm": 2.060910701751709,
      "learning_rate": 1.634322315507283e-06,
      "loss": 0.4686,
      "step": 301950
    },
    {
      "epoch": 123.6,
      "grad_norm": 1.6847264766693115,
      "learning_rate": 1.6341711290007337e-06,
      "loss": 0.4578,
      "step": 301960
    },
    {
      "epoch": 123.61,
      "grad_norm": 1.8279855251312256,
      "learning_rate": 1.6340199457446552e-06,
      "loss": 0.465,
      "step": 301970
    },
    {
      "epoch": 123.61,
      "grad_norm": 1.7619109153747559,
      "learning_rate": 1.6338687657397397e-06,
      "loss": 0.4403,
      "step": 301980
    },
    {
      "epoch": 123.61,
      "grad_norm": 2.1444408893585205,
      "learning_rate": 1.633717588986681e-06,
      "loss": 0.477,
      "step": 301990
    },
    {
      "epoch": 123.62,
      "grad_norm": 2.115809679031372,
      "learning_rate": 1.6335664154861706e-06,
      "loss": 0.4599,
      "step": 302000
    },
    {
      "epoch": 123.62,
      "grad_norm": 2.1950573921203613,
      "learning_rate": 1.633415245238901e-06,
      "loss": 0.4519,
      "step": 302010
    },
    {
      "epoch": 123.63,
      "grad_norm": 1.8385447263717651,
      "learning_rate": 1.6332640782455653e-06,
      "loss": 0.47,
      "step": 302020
    },
    {
      "epoch": 123.63,
      "grad_norm": 2.316199541091919,
      "learning_rate": 1.6331129145068553e-06,
      "loss": 0.4638,
      "step": 302030
    },
    {
      "epoch": 123.63,
      "grad_norm": 2.148394823074341,
      "learning_rate": 1.632961754023464e-06,
      "loss": 0.4858,
      "step": 302040
    },
    {
      "epoch": 123.64,
      "grad_norm": 1.7181529998779297,
      "learning_rate": 1.6328105967960832e-06,
      "loss": 0.4562,
      "step": 302050
    },
    {
      "epoch": 123.64,
      "grad_norm": 1.923336386680603,
      "learning_rate": 1.6326594428254066e-06,
      "loss": 0.4622,
      "step": 302060
    },
    {
      "epoch": 123.65,
      "grad_norm": 1.7566298246383667,
      "learning_rate": 1.632508292112125e-06,
      "loss": 0.4749,
      "step": 302070
    },
    {
      "epoch": 123.65,
      "grad_norm": 1.7548664808273315,
      "learning_rate": 1.6323571446569314e-06,
      "loss": 0.4642,
      "step": 302080
    },
    {
      "epoch": 123.66,
      "grad_norm": 1.5922331809997559,
      "learning_rate": 1.6322060004605184e-06,
      "loss": 0.4808,
      "step": 302090
    },
    {
      "epoch": 123.66,
      "grad_norm": 2.1370737552642822,
      "learning_rate": 1.6320548595235784e-06,
      "loss": 0.4763,
      "step": 302100
    },
    {
      "epoch": 123.66,
      "grad_norm": 1.7444967031478882,
      "learning_rate": 1.6319037218468034e-06,
      "loss": 0.4513,
      "step": 302110
    },
    {
      "epoch": 123.67,
      "grad_norm": 2.1025397777557373,
      "learning_rate": 1.6317525874308861e-06,
      "loss": 0.4662,
      "step": 302120
    },
    {
      "epoch": 123.67,
      "grad_norm": 2.4154889583587646,
      "learning_rate": 1.6316014562765187e-06,
      "loss": 0.454,
      "step": 302130
    },
    {
      "epoch": 123.68,
      "grad_norm": 2.220707416534424,
      "learning_rate": 1.6314503283843935e-06,
      "loss": 0.45,
      "step": 302140
    },
    {
      "epoch": 123.68,
      "grad_norm": 2.2592880725860596,
      "learning_rate": 1.6312992037552033e-06,
      "loss": 0.4553,
      "step": 302150
    },
    {
      "epoch": 123.68,
      "grad_norm": 1.7566132545471191,
      "learning_rate": 1.6311480823896382e-06,
      "loss": 0.4529,
      "step": 302160
    },
    {
      "epoch": 123.69,
      "grad_norm": 1.6241254806518555,
      "learning_rate": 1.630996964288393e-06,
      "loss": 0.4551,
      "step": 302170
    },
    {
      "epoch": 123.69,
      "grad_norm": 1.796096682548523,
      "learning_rate": 1.6308458494521594e-06,
      "loss": 0.4479,
      "step": 302180
    },
    {
      "epoch": 123.7,
      "grad_norm": 1.8650585412979126,
      "learning_rate": 1.6306947378816298e-06,
      "loss": 0.4452,
      "step": 302190
    },
    {
      "epoch": 123.7,
      "grad_norm": 2.0581653118133545,
      "learning_rate": 1.630543629577495e-06,
      "loss": 0.4723,
      "step": 302200
    },
    {
      "epoch": 123.7,
      "grad_norm": 2.006006956100464,
      "learning_rate": 1.6303925245404488e-06,
      "loss": 0.478,
      "step": 302210
    },
    {
      "epoch": 123.71,
      "grad_norm": 1.9405698776245117,
      "learning_rate": 1.6302414227711822e-06,
      "loss": 0.4528,
      "step": 302220
    },
    {
      "epoch": 123.71,
      "grad_norm": 2.095705270767212,
      "learning_rate": 1.630090324270388e-06,
      "loss": 0.4759,
      "step": 302230
    },
    {
      "epoch": 123.72,
      "grad_norm": 1.732060194015503,
      "learning_rate": 1.6299392290387585e-06,
      "loss": 0.4564,
      "step": 302240
    },
    {
      "epoch": 123.72,
      "grad_norm": 2.051243543624878,
      "learning_rate": 1.6297881370769852e-06,
      "loss": 0.4515,
      "step": 302250
    },
    {
      "epoch": 123.72,
      "grad_norm": 1.8017725944519043,
      "learning_rate": 1.629637048385761e-06,
      "loss": 0.4385,
      "step": 302260
    },
    {
      "epoch": 123.73,
      "grad_norm": 1.6760753393173218,
      "learning_rate": 1.6294859629657772e-06,
      "loss": 0.4658,
      "step": 302270
    },
    {
      "epoch": 123.73,
      "grad_norm": 2.03653621673584,
      "learning_rate": 1.6293348808177272e-06,
      "loss": 0.4572,
      "step": 302280
    },
    {
      "epoch": 123.74,
      "grad_norm": 1.679695725440979,
      "learning_rate": 1.6291838019423017e-06,
      "loss": 0.4603,
      "step": 302290
    },
    {
      "epoch": 123.74,
      "grad_norm": 1.7691360712051392,
      "learning_rate": 1.6290327263401931e-06,
      "loss": 0.4563,
      "step": 302300
    },
    {
      "epoch": 123.75,
      "grad_norm": 1.853297233581543,
      "learning_rate": 1.6288816540120936e-06,
      "loss": 0.4519,
      "step": 302310
    },
    {
      "epoch": 123.75,
      "grad_norm": 2.186413049697876,
      "learning_rate": 1.6287305849586958e-06,
      "loss": 0.4791,
      "step": 302320
    },
    {
      "epoch": 123.75,
      "grad_norm": 2.1235928535461426,
      "learning_rate": 1.6285795191806906e-06,
      "loss": 0.4543,
      "step": 302330
    },
    {
      "epoch": 123.76,
      "grad_norm": 1.9071159362792969,
      "learning_rate": 1.628428456678771e-06,
      "loss": 0.4798,
      "step": 302340
    },
    {
      "epoch": 123.76,
      "grad_norm": 1.4932860136032104,
      "learning_rate": 1.628277397453628e-06,
      "loss": 0.4567,
      "step": 302350
    },
    {
      "epoch": 123.77,
      "grad_norm": 1.5815949440002441,
      "learning_rate": 1.6281263415059547e-06,
      "loss": 0.4601,
      "step": 302360
    },
    {
      "epoch": 123.77,
      "grad_norm": 1.9909342527389526,
      "learning_rate": 1.6279752888364421e-06,
      "loss": 0.4749,
      "step": 302370
    },
    {
      "epoch": 123.77,
      "grad_norm": 1.8382107019424438,
      "learning_rate": 1.6278242394457833e-06,
      "loss": 0.4695,
      "step": 302380
    },
    {
      "epoch": 123.78,
      "grad_norm": 2.104236364364624,
      "learning_rate": 1.6276731933346692e-06,
      "loss": 0.4525,
      "step": 302390
    },
    {
      "epoch": 123.78,
      "grad_norm": 1.9020980596542358,
      "learning_rate": 1.6275221505037923e-06,
      "loss": 0.4634,
      "step": 302400
    },
    {
      "epoch": 123.79,
      "grad_norm": 1.7421938180923462,
      "learning_rate": 1.627371110953844e-06,
      "loss": 0.4449,
      "step": 302410
    },
    {
      "epoch": 123.79,
      "grad_norm": 2.410210371017456,
      "learning_rate": 1.627220074685517e-06,
      "loss": 0.4766,
      "step": 302420
    },
    {
      "epoch": 123.79,
      "grad_norm": 2.242943525314331,
      "learning_rate": 1.6270690416995017e-06,
      "loss": 0.4526,
      "step": 302430
    },
    {
      "epoch": 123.8,
      "grad_norm": 1.9744728803634644,
      "learning_rate": 1.6269180119964915e-06,
      "loss": 0.4478,
      "step": 302440
    },
    {
      "epoch": 123.8,
      "grad_norm": 1.8173189163208008,
      "learning_rate": 1.6267669855771776e-06,
      "loss": 0.4435,
      "step": 302450
    },
    {
      "epoch": 123.81,
      "grad_norm": 2.2570531368255615,
      "learning_rate": 1.6266159624422517e-06,
      "loss": 0.462,
      "step": 302460
    },
    {
      "epoch": 123.81,
      "grad_norm": 2.1897850036621094,
      "learning_rate": 1.6264649425924054e-06,
      "loss": 0.4605,
      "step": 302470
    },
    {
      "epoch": 123.81,
      "grad_norm": 2.081113576889038,
      "learning_rate": 1.6263139260283313e-06,
      "loss": 0.469,
      "step": 302480
    },
    {
      "epoch": 123.82,
      "grad_norm": 2.058323621749878,
      "learning_rate": 1.6261629127507206e-06,
      "loss": 0.4395,
      "step": 302490
    },
    {
      "epoch": 123.82,
      "grad_norm": 2.016632556915283,
      "learning_rate": 1.6260119027602656e-06,
      "loss": 0.4591,
      "step": 302500
    },
    {
      "epoch": 123.83,
      "grad_norm": 2.2047948837280273,
      "learning_rate": 1.6258608960576573e-06,
      "loss": 0.4651,
      "step": 302510
    },
    {
      "epoch": 123.83,
      "grad_norm": 1.6977813243865967,
      "learning_rate": 1.6257098926435882e-06,
      "loss": 0.4501,
      "step": 302520
    },
    {
      "epoch": 123.84,
      "grad_norm": 1.8834893703460693,
      "learning_rate": 1.625558892518749e-06,
      "loss": 0.4497,
      "step": 302530
    },
    {
      "epoch": 123.84,
      "grad_norm": 1.9181791543960571,
      "learning_rate": 1.6254078956838314e-06,
      "loss": 0.4624,
      "step": 302540
    },
    {
      "epoch": 123.84,
      "grad_norm": 2.2481703758239746,
      "learning_rate": 1.6252569021395288e-06,
      "loss": 0.4561,
      "step": 302550
    },
    {
      "epoch": 123.85,
      "grad_norm": 2.1887760162353516,
      "learning_rate": 1.6251059118865316e-06,
      "loss": 0.4476,
      "step": 302560
    },
    {
      "epoch": 123.85,
      "grad_norm": 1.7693438529968262,
      "learning_rate": 1.624954924925532e-06,
      "loss": 0.4555,
      "step": 302570
    },
    {
      "epoch": 123.86,
      "grad_norm": 2.007418394088745,
      "learning_rate": 1.624803941257221e-06,
      "loss": 0.4949,
      "step": 302580
    },
    {
      "epoch": 123.86,
      "grad_norm": 1.7385481595993042,
      "learning_rate": 1.6246529608822904e-06,
      "loss": 0.4504,
      "step": 302590
    },
    {
      "epoch": 123.86,
      "grad_norm": 1.9635292291641235,
      "learning_rate": 1.6245019838014323e-06,
      "loss": 0.471,
      "step": 302600
    },
    {
      "epoch": 123.87,
      "grad_norm": 1.8273497819900513,
      "learning_rate": 1.6243510100153377e-06,
      "loss": 0.4741,
      "step": 302610
    },
    {
      "epoch": 123.87,
      "grad_norm": 1.7498515844345093,
      "learning_rate": 1.6242000395246987e-06,
      "loss": 0.4629,
      "step": 302620
    },
    {
      "epoch": 123.88,
      "grad_norm": 1.8322100639343262,
      "learning_rate": 1.6240490723302062e-06,
      "loss": 0.4736,
      "step": 302630
    },
    {
      "epoch": 123.88,
      "grad_norm": 1.9605872631072998,
      "learning_rate": 1.6238981084325532e-06,
      "loss": 0.4649,
      "step": 302640
    },
    {
      "epoch": 123.88,
      "grad_norm": 1.7996894121170044,
      "learning_rate": 1.6237471478324295e-06,
      "loss": 0.4545,
      "step": 302650
    },
    {
      "epoch": 123.89,
      "grad_norm": 2.2372303009033203,
      "learning_rate": 1.6235961905305272e-06,
      "loss": 0.4775,
      "step": 302660
    },
    {
      "epoch": 123.89,
      "grad_norm": 2.611037254333496,
      "learning_rate": 1.6234452365275384e-06,
      "loss": 0.4657,
      "step": 302670
    },
    {
      "epoch": 123.9,
      "grad_norm": 2.2614316940307617,
      "learning_rate": 1.6232942858241539e-06,
      "loss": 0.4623,
      "step": 302680
    },
    {
      "epoch": 123.9,
      "grad_norm": 2.176694393157959,
      "learning_rate": 1.6231433384210654e-06,
      "loss": 0.4407,
      "step": 302690
    },
    {
      "epoch": 123.91,
      "grad_norm": 1.9620147943496704,
      "learning_rate": 1.6229923943189643e-06,
      "loss": 0.4634,
      "step": 302700
    },
    {
      "epoch": 123.91,
      "grad_norm": 2.437407970428467,
      "learning_rate": 1.6228414535185424e-06,
      "loss": 0.4775,
      "step": 302710
    },
    {
      "epoch": 123.91,
      "grad_norm": 2.47768497467041,
      "learning_rate": 1.622690516020491e-06,
      "loss": 0.4651,
      "step": 302720
    },
    {
      "epoch": 123.92,
      "grad_norm": 1.859898328781128,
      "learning_rate": 1.6225395818255008e-06,
      "loss": 0.4468,
      "step": 302730
    },
    {
      "epoch": 123.92,
      "grad_norm": 2.0892865657806396,
      "learning_rate": 1.622388650934264e-06,
      "loss": 0.4555,
      "step": 302740
    },
    {
      "epoch": 123.93,
      "grad_norm": 2.6982314586639404,
      "learning_rate": 1.6222377233474724e-06,
      "loss": 0.4396,
      "step": 302750
    },
    {
      "epoch": 123.93,
      "grad_norm": 2.090433120727539,
      "learning_rate": 1.6220867990658168e-06,
      "loss": 0.4523,
      "step": 302760
    },
    {
      "epoch": 123.93,
      "grad_norm": 2.0105702877044678,
      "learning_rate": 1.6219358780899882e-06,
      "loss": 0.4748,
      "step": 302770
    },
    {
      "epoch": 123.94,
      "grad_norm": 2.0989363193511963,
      "learning_rate": 1.6217849604206786e-06,
      "loss": 0.4492,
      "step": 302780
    },
    {
      "epoch": 123.94,
      "grad_norm": 1.735154151916504,
      "learning_rate": 1.621634046058579e-06,
      "loss": 0.4591,
      "step": 302790
    },
    {
      "epoch": 123.95,
      "grad_norm": 2.283881902694702,
      "learning_rate": 1.6214831350043806e-06,
      "loss": 0.461,
      "step": 302800
    },
    {
      "epoch": 123.95,
      "grad_norm": 1.8689308166503906,
      "learning_rate": 1.6213322272587747e-06,
      "loss": 0.4517,
      "step": 302810
    },
    {
      "epoch": 123.95,
      "grad_norm": 1.962170124053955,
      "learning_rate": 1.621181322822453e-06,
      "loss": 0.4659,
      "step": 302820
    },
    {
      "epoch": 123.96,
      "grad_norm": 1.7724838256835938,
      "learning_rate": 1.6210304216961062e-06,
      "loss": 0.4622,
      "step": 302830
    },
    {
      "epoch": 123.96,
      "grad_norm": 1.9571354389190674,
      "learning_rate": 1.6208795238804261e-06,
      "loss": 0.4555,
      "step": 302840
    },
    {
      "epoch": 123.97,
      "grad_norm": 1.7160396575927734,
      "learning_rate": 1.6207286293761036e-06,
      "loss": 0.4547,
      "step": 302850
    },
    {
      "epoch": 123.97,
      "grad_norm": 1.54378080368042,
      "learning_rate": 1.6205777381838308e-06,
      "loss": 0.4602,
      "step": 302860
    },
    {
      "epoch": 123.97,
      "grad_norm": 2.0574557781219482,
      "learning_rate": 1.6204268503042974e-06,
      "loss": 0.4577,
      "step": 302870
    },
    {
      "epoch": 123.98,
      "grad_norm": 1.7981081008911133,
      "learning_rate": 1.6202759657381953e-06,
      "loss": 0.4462,
      "step": 302880
    },
    {
      "epoch": 123.98,
      "grad_norm": 2.055634021759033,
      "learning_rate": 1.6201250844862156e-06,
      "loss": 0.4385,
      "step": 302890
    },
    {
      "epoch": 123.99,
      "grad_norm": 2.1986308097839355,
      "learning_rate": 1.61997420654905e-06,
      "loss": 0.4755,
      "step": 302900
    },
    {
      "epoch": 123.99,
      "grad_norm": 1.8219655752182007,
      "learning_rate": 1.6198233319273886e-06,
      "loss": 0.4474,
      "step": 302910
    },
    {
      "epoch": 124.0,
      "grad_norm": 2.0474064350128174,
      "learning_rate": 1.6196724606219233e-06,
      "loss": 0.4503,
      "step": 302920
    },
    {
      "epoch": 124.0,
      "grad_norm": 1.8414394855499268,
      "learning_rate": 1.6195215926333453e-06,
      "loss": 0.4457,
      "step": 302930
    },
    {
      "epoch": 124.0,
      "eval_loss": 0.46196743845939636,
      "eval_runtime": 52.4296,
      "eval_samples_per_second": 65.783,
      "eval_steps_per_second": 8.24,
      "step": 302932
    },
    {
      "epoch": 124.0,
      "grad_norm": 1.9098894596099854,
      "learning_rate": 1.6193707279623452e-06,
      "loss": 0.4582,
      "step": 302940
    },
    {
      "epoch": 124.01,
      "grad_norm": 1.9318127632141113,
      "learning_rate": 1.6192198666096146e-06,
      "loss": 0.4624,
      "step": 302950
    },
    {
      "epoch": 124.01,
      "grad_norm": 1.8371038436889648,
      "learning_rate": 1.6190690085758445e-06,
      "loss": 0.4511,
      "step": 302960
    },
    {
      "epoch": 124.02,
      "grad_norm": 1.9071542024612427,
      "learning_rate": 1.6189181538617255e-06,
      "loss": 0.4574,
      "step": 302970
    },
    {
      "epoch": 124.02,
      "grad_norm": 1.9502748250961304,
      "learning_rate": 1.6187673024679491e-06,
      "loss": 0.4647,
      "step": 302980
    },
    {
      "epoch": 124.02,
      "grad_norm": 1.57663094997406,
      "learning_rate": 1.6186164543952061e-06,
      "loss": 0.4627,
      "step": 302990
    },
    {
      "epoch": 124.03,
      "grad_norm": 1.9405171871185303,
      "learning_rate": 1.618465609644188e-06,
      "loss": 0.4782,
      "step": 303000
    },
    {
      "epoch": 124.03,
      "grad_norm": 2.5960421562194824,
      "learning_rate": 1.618314768215585e-06,
      "loss": 0.4543,
      "step": 303010
    },
    {
      "epoch": 124.04,
      "grad_norm": 2.375823736190796,
      "learning_rate": 1.6181639301100883e-06,
      "loss": 0.4599,
      "step": 303020
    },
    {
      "epoch": 124.04,
      "grad_norm": 2.0083794593811035,
      "learning_rate": 1.6180130953283891e-06,
      "loss": 0.465,
      "step": 303030
    },
    {
      "epoch": 124.04,
      "grad_norm": 2.0815489292144775,
      "learning_rate": 1.6178622638711783e-06,
      "loss": 0.4423,
      "step": 303040
    },
    {
      "epoch": 124.05,
      "grad_norm": 2.0660948753356934,
      "learning_rate": 1.6177114357391469e-06,
      "loss": 0.4439,
      "step": 303050
    },
    {
      "epoch": 124.05,
      "grad_norm": 2.088712453842163,
      "learning_rate": 1.6175606109329852e-06,
      "loss": 0.4171,
      "step": 303060
    },
    {
      "epoch": 124.06,
      "grad_norm": 2.234915256500244,
      "learning_rate": 1.6174097894533852e-06,
      "loss": 0.4471,
      "step": 303070
    },
    {
      "epoch": 124.06,
      "grad_norm": 1.8084375858306885,
      "learning_rate": 1.617258971301037e-06,
      "loss": 0.4494,
      "step": 303080
    },
    {
      "epoch": 124.06,
      "grad_norm": 2.67210054397583,
      "learning_rate": 1.6171081564766323e-06,
      "loss": 0.459,
      "step": 303090
    },
    {
      "epoch": 124.07,
      "grad_norm": 1.7390426397323608,
      "learning_rate": 1.616957344980861e-06,
      "loss": 0.4747,
      "step": 303100
    },
    {
      "epoch": 124.07,
      "grad_norm": 2.381545305252075,
      "learning_rate": 1.616806536814413e-06,
      "loss": 0.4634,
      "step": 303110
    },
    {
      "epoch": 124.08,
      "grad_norm": 2.220961570739746,
      "learning_rate": 1.6166557319779814e-06,
      "loss": 0.4678,
      "step": 303120
    },
    {
      "epoch": 124.08,
      "grad_norm": 1.6199615001678467,
      "learning_rate": 1.6165049304722562e-06,
      "loss": 0.4665,
      "step": 303130
    },
    {
      "epoch": 124.09,
      "grad_norm": 2.5276355743408203,
      "learning_rate": 1.6163541322979285e-06,
      "loss": 0.4627,
      "step": 303140
    },
    {
      "epoch": 124.09,
      "grad_norm": 1.9392225742340088,
      "learning_rate": 1.616203337455688e-06,
      "loss": 0.4438,
      "step": 303150
    },
    {
      "epoch": 124.09,
      "grad_norm": 2.29056715965271,
      "learning_rate": 1.6160525459462262e-06,
      "loss": 0.4498,
      "step": 303160
    },
    {
      "epoch": 124.1,
      "grad_norm": 2.040325164794922,
      "learning_rate": 1.615901757770234e-06,
      "loss": 0.4566,
      "step": 303170
    },
    {
      "epoch": 124.1,
      "grad_norm": 2.024102210998535,
      "learning_rate": 1.6157509729284013e-06,
      "loss": 0.4733,
      "step": 303180
    },
    {
      "epoch": 124.11,
      "grad_norm": 2.4689056873321533,
      "learning_rate": 1.6156001914214198e-06,
      "loss": 0.4626,
      "step": 303190
    },
    {
      "epoch": 124.11,
      "grad_norm": 1.9599210023880005,
      "learning_rate": 1.6154494132499796e-06,
      "loss": 0.4495,
      "step": 303200
    },
    {
      "epoch": 124.11,
      "grad_norm": 1.87204909324646,
      "learning_rate": 1.6152986384147719e-06,
      "loss": 0.4469,
      "step": 303210
    },
    {
      "epoch": 124.12,
      "grad_norm": 1.9606107473373413,
      "learning_rate": 1.615147866916487e-06,
      "loss": 0.4529,
      "step": 303220
    },
    {
      "epoch": 124.12,
      "grad_norm": 2.58927845954895,
      "learning_rate": 1.6149970987558156e-06,
      "loss": 0.4666,
      "step": 303230
    },
    {
      "epoch": 124.13,
      "grad_norm": 1.6490942239761353,
      "learning_rate": 1.6148463339334485e-06,
      "loss": 0.4704,
      "step": 303240
    },
    {
      "epoch": 124.13,
      "grad_norm": 2.030900239944458,
      "learning_rate": 1.6146955724500758e-06,
      "loss": 0.4626,
      "step": 303250
    },
    {
      "epoch": 124.13,
      "grad_norm": 1.8767528533935547,
      "learning_rate": 1.6145448143063889e-06,
      "loss": 0.4624,
      "step": 303260
    },
    {
      "epoch": 124.14,
      "grad_norm": 1.9969520568847656,
      "learning_rate": 1.6143940595030777e-06,
      "loss": 0.464,
      "step": 303270
    },
    {
      "epoch": 124.14,
      "grad_norm": 2.0962016582489014,
      "learning_rate": 1.6142433080408333e-06,
      "loss": 0.4528,
      "step": 303280
    },
    {
      "epoch": 124.15,
      "grad_norm": 2.1022818088531494,
      "learning_rate": 1.614092559920346e-06,
      "loss": 0.4705,
      "step": 303290
    },
    {
      "epoch": 124.15,
      "grad_norm": 1.6999006271362305,
      "learning_rate": 1.6139418151423059e-06,
      "loss": 0.4502,
      "step": 303300
    },
    {
      "epoch": 124.15,
      "grad_norm": 2.0910491943359375,
      "learning_rate": 1.6137910737074047e-06,
      "loss": 0.4553,
      "step": 303310
    },
    {
      "epoch": 124.16,
      "grad_norm": 2.046107769012451,
      "learning_rate": 1.613640335616332e-06,
      "loss": 0.471,
      "step": 303320
    },
    {
      "epoch": 124.16,
      "grad_norm": 2.161914825439453,
      "learning_rate": 1.613489600869779e-06,
      "loss": 0.4617,
      "step": 303330
    },
    {
      "epoch": 124.17,
      "grad_norm": 1.9115054607391357,
      "learning_rate": 1.6133388694684358e-06,
      "loss": 0.4748,
      "step": 303340
    },
    {
      "epoch": 124.17,
      "grad_norm": 2.2037582397460938,
      "learning_rate": 1.6131881414129927e-06,
      "loss": 0.4718,
      "step": 303350
    },
    {
      "epoch": 124.18,
      "grad_norm": 2.0046966075897217,
      "learning_rate": 1.6130374167041407e-06,
      "loss": 0.4696,
      "step": 303360
    },
    {
      "epoch": 124.18,
      "grad_norm": 1.659983515739441,
      "learning_rate": 1.6128866953425695e-06,
      "loss": 0.4653,
      "step": 303370
    },
    {
      "epoch": 124.18,
      "grad_norm": 1.9967671632766724,
      "learning_rate": 1.6127359773289698e-06,
      "loss": 0.4389,
      "step": 303380
    },
    {
      "epoch": 124.19,
      "grad_norm": 2.679021120071411,
      "learning_rate": 1.6125852626640323e-06,
      "loss": 0.4304,
      "step": 303390
    },
    {
      "epoch": 124.19,
      "grad_norm": 1.733561635017395,
      "learning_rate": 1.612434551348447e-06,
      "loss": 0.4748,
      "step": 303400
    },
    {
      "epoch": 124.2,
      "grad_norm": 1.9129228591918945,
      "learning_rate": 1.6122838433829046e-06,
      "loss": 0.4508,
      "step": 303410
    },
    {
      "epoch": 124.2,
      "grad_norm": 2.326561450958252,
      "learning_rate": 1.6121331387680955e-06,
      "loss": 0.4535,
      "step": 303420
    },
    {
      "epoch": 124.2,
      "grad_norm": 2.1148245334625244,
      "learning_rate": 1.6119824375047098e-06,
      "loss": 0.4533,
      "step": 303430
    },
    {
      "epoch": 124.21,
      "grad_norm": 1.79720139503479,
      "learning_rate": 1.611831739593438e-06,
      "loss": 0.4653,
      "step": 303440
    },
    {
      "epoch": 124.21,
      "grad_norm": 2.1092171669006348,
      "learning_rate": 1.611681045034971e-06,
      "loss": 0.4957,
      "step": 303450
    },
    {
      "epoch": 124.22,
      "grad_norm": 1.9644911289215088,
      "learning_rate": 1.611530353829998e-06,
      "loss": 0.4485,
      "step": 303460
    },
    {
      "epoch": 124.22,
      "grad_norm": 1.8303368091583252,
      "learning_rate": 1.6113796659792099e-06,
      "loss": 0.4662,
      "step": 303470
    },
    {
      "epoch": 124.22,
      "grad_norm": 1.8196762800216675,
      "learning_rate": 1.6112289814832957e-06,
      "loss": 0.4616,
      "step": 303480
    },
    {
      "epoch": 124.23,
      "grad_norm": 2.1260111331939697,
      "learning_rate": 1.6110783003429477e-06,
      "loss": 0.4727,
      "step": 303490
    },
    {
      "epoch": 124.23,
      "grad_norm": 1.8352042436599731,
      "learning_rate": 1.6109276225588562e-06,
      "loss": 0.4549,
      "step": 303500
    },
    {
      "epoch": 124.24,
      "grad_norm": 2.2928056716918945,
      "learning_rate": 1.6107769481317097e-06,
      "loss": 0.4532,
      "step": 303510
    },
    {
      "epoch": 124.24,
      "grad_norm": 1.8795541524887085,
      "learning_rate": 1.6106262770621993e-06,
      "loss": 0.4436,
      "step": 303520
    },
    {
      "epoch": 124.24,
      "grad_norm": 1.7773540019989014,
      "learning_rate": 1.6104756093510153e-06,
      "loss": 0.4833,
      "step": 303530
    },
    {
      "epoch": 124.25,
      "grad_norm": 1.9851891994476318,
      "learning_rate": 1.6103249449988476e-06,
      "loss": 0.4846,
      "step": 303540
    },
    {
      "epoch": 124.25,
      "grad_norm": 1.7785906791687012,
      "learning_rate": 1.6101742840063864e-06,
      "loss": 0.4606,
      "step": 303550
    },
    {
      "epoch": 124.26,
      "grad_norm": 1.6360034942626953,
      "learning_rate": 1.610023626374322e-06,
      "loss": 0.4644,
      "step": 303560
    },
    {
      "epoch": 124.26,
      "grad_norm": 1.73979914188385,
      "learning_rate": 1.6098729721033444e-06,
      "loss": 0.4596,
      "step": 303570
    },
    {
      "epoch": 124.27,
      "grad_norm": 1.9287357330322266,
      "learning_rate": 1.6097223211941446e-06,
      "loss": 0.4521,
      "step": 303580
    },
    {
      "epoch": 124.27,
      "grad_norm": 1.6892766952514648,
      "learning_rate": 1.6095716736474113e-06,
      "loss": 0.4493,
      "step": 303590
    },
    {
      "epoch": 124.27,
      "grad_norm": 1.7451109886169434,
      "learning_rate": 1.6094210294638352e-06,
      "loss": 0.4715,
      "step": 303600
    },
    {
      "epoch": 124.28,
      "grad_norm": 2.389434337615967,
      "learning_rate": 1.6092703886441062e-06,
      "loss": 0.4879,
      "step": 303610
    },
    {
      "epoch": 124.28,
      "grad_norm": 1.9509612321853638,
      "learning_rate": 1.6091197511889148e-06,
      "loss": 0.4554,
      "step": 303620
    },
    {
      "epoch": 124.29,
      "grad_norm": 1.7733678817749023,
      "learning_rate": 1.6089691170989507e-06,
      "loss": 0.4785,
      "step": 303630
    },
    {
      "epoch": 124.29,
      "grad_norm": 2.027244806289673,
      "learning_rate": 1.6088184863749039e-06,
      "loss": 0.4652,
      "step": 303640
    },
    {
      "epoch": 124.29,
      "grad_norm": 1.704089641571045,
      "learning_rate": 1.6086678590174645e-06,
      "loss": 0.4421,
      "step": 303650
    },
    {
      "epoch": 124.3,
      "grad_norm": 1.8208523988723755,
      "learning_rate": 1.6085172350273228e-06,
      "loss": 0.4695,
      "step": 303660
    },
    {
      "epoch": 124.3,
      "grad_norm": 1.748902678489685,
      "learning_rate": 1.608366614405168e-06,
      "loss": 0.4859,
      "step": 303670
    },
    {
      "epoch": 124.31,
      "grad_norm": 2.356072187423706,
      "learning_rate": 1.6082159971516911e-06,
      "loss": 0.4564,
      "step": 303680
    },
    {
      "epoch": 124.31,
      "grad_norm": 2.231689929962158,
      "learning_rate": 1.6080653832675813e-06,
      "loss": 0.4678,
      "step": 303690
    },
    {
      "epoch": 124.31,
      "grad_norm": 1.8607215881347656,
      "learning_rate": 1.6079147727535293e-06,
      "loss": 0.4663,
      "step": 303700
    },
    {
      "epoch": 124.32,
      "grad_norm": 2.3962886333465576,
      "learning_rate": 1.607764165610224e-06,
      "loss": 0.4418,
      "step": 303710
    },
    {
      "epoch": 124.32,
      "grad_norm": 1.5864962339401245,
      "learning_rate": 1.6076135618383566e-06,
      "loss": 0.4682,
      "step": 303720
    },
    {
      "epoch": 124.33,
      "grad_norm": 1.4760500192642212,
      "learning_rate": 1.6074629614386158e-06,
      "loss": 0.4684,
      "step": 303730
    },
    {
      "epoch": 124.33,
      "grad_norm": 2.2426652908325195,
      "learning_rate": 1.6073123644116915e-06,
      "loss": 0.4421,
      "step": 303740
    },
    {
      "epoch": 124.33,
      "grad_norm": 1.9166620969772339,
      "learning_rate": 1.6071617707582741e-06,
      "loss": 0.4486,
      "step": 303750
    },
    {
      "epoch": 124.34,
      "grad_norm": 1.7114548683166504,
      "learning_rate": 1.6070111804790536e-06,
      "loss": 0.4715,
      "step": 303760
    },
    {
      "epoch": 124.34,
      "grad_norm": 1.8090331554412842,
      "learning_rate": 1.6068605935747192e-06,
      "loss": 0.4785,
      "step": 303770
    },
    {
      "epoch": 124.35,
      "grad_norm": 1.8672200441360474,
      "learning_rate": 1.6067100100459612e-06,
      "loss": 0.4834,
      "step": 303780
    },
    {
      "epoch": 124.35,
      "grad_norm": 1.9696764945983887,
      "learning_rate": 1.6065594298934693e-06,
      "loss": 0.477,
      "step": 303790
    },
    {
      "epoch": 124.36,
      "grad_norm": 2.319920301437378,
      "learning_rate": 1.606408853117933e-06,
      "loss": 0.4635,
      "step": 303800
    },
    {
      "epoch": 124.36,
      "grad_norm": 2.138054847717285,
      "learning_rate": 1.6062582797200434e-06,
      "loss": 0.4621,
      "step": 303810
    },
    {
      "epoch": 124.36,
      "grad_norm": 1.995039463043213,
      "learning_rate": 1.6061077097004879e-06,
      "loss": 0.4438,
      "step": 303820
    },
    {
      "epoch": 124.37,
      "grad_norm": 1.7780508995056152,
      "learning_rate": 1.6059571430599579e-06,
      "loss": 0.457,
      "step": 303830
    },
    {
      "epoch": 124.37,
      "grad_norm": 2.1517934799194336,
      "learning_rate": 1.6058065797991425e-06,
      "loss": 0.4573,
      "step": 303840
    },
    {
      "epoch": 124.38,
      "grad_norm": 1.7998876571655273,
      "learning_rate": 1.6056560199187321e-06,
      "loss": 0.4443,
      "step": 303850
    },
    {
      "epoch": 124.38,
      "grad_norm": 1.888208031654358,
      "learning_rate": 1.605505463419415e-06,
      "loss": 0.4948,
      "step": 303860
    },
    {
      "epoch": 124.38,
      "grad_norm": 1.865093469619751,
      "learning_rate": 1.6053549103018824e-06,
      "loss": 0.4382,
      "step": 303870
    },
    {
      "epoch": 124.39,
      "grad_norm": 1.813049554824829,
      "learning_rate": 1.605204360566823e-06,
      "loss": 0.4489,
      "step": 303880
    },
    {
      "epoch": 124.39,
      "grad_norm": 2.437063694000244,
      "learning_rate": 1.6050538142149273e-06,
      "loss": 0.4638,
      "step": 303890
    },
    {
      "epoch": 124.4,
      "grad_norm": 1.7050344944000244,
      "learning_rate": 1.6049032712468845e-06,
      "loss": 0.4635,
      "step": 303900
    },
    {
      "epoch": 124.4,
      "grad_norm": 2.3507015705108643,
      "learning_rate": 1.6047527316633835e-06,
      "loss": 0.4457,
      "step": 303910
    },
    {
      "epoch": 124.4,
      "grad_norm": 1.9236435890197754,
      "learning_rate": 1.604602195465115e-06,
      "loss": 0.4643,
      "step": 303920
    },
    {
      "epoch": 124.41,
      "grad_norm": 1.62380850315094,
      "learning_rate": 1.6044516626527683e-06,
      "loss": 0.4549,
      "step": 303930
    },
    {
      "epoch": 124.41,
      "grad_norm": 1.8433459997177124,
      "learning_rate": 1.6043011332270328e-06,
      "loss": 0.4675,
      "step": 303940
    },
    {
      "epoch": 124.42,
      "grad_norm": 1.795459508895874,
      "learning_rate": 1.6041506071885977e-06,
      "loss": 0.4718,
      "step": 303950
    },
    {
      "epoch": 124.42,
      "grad_norm": 2.0755393505096436,
      "learning_rate": 1.604000084538153e-06,
      "loss": 0.4742,
      "step": 303960
    },
    {
      "epoch": 124.42,
      "grad_norm": 1.6980208158493042,
      "learning_rate": 1.603849565276388e-06,
      "loss": 0.4582,
      "step": 303970
    },
    {
      "epoch": 124.43,
      "grad_norm": 2.5042617321014404,
      "learning_rate": 1.6036990494039924e-06,
      "loss": 0.4504,
      "step": 303980
    },
    {
      "epoch": 124.43,
      "grad_norm": 1.6604586839675903,
      "learning_rate": 1.6035485369216554e-06,
      "loss": 0.4506,
      "step": 303990
    },
    {
      "epoch": 124.44,
      "grad_norm": 1.9520163536071777,
      "learning_rate": 1.603398027830067e-06,
      "loss": 0.4523,
      "step": 304000
    },
    {
      "epoch": 124.44,
      "grad_norm": 2.2067601680755615,
      "learning_rate": 1.603247522129916e-06,
      "loss": 0.4764,
      "step": 304010
    },
    {
      "epoch": 124.45,
      "grad_norm": 1.7846252918243408,
      "learning_rate": 1.6030970198218924e-06,
      "loss": 0.4599,
      "step": 304020
    },
    {
      "epoch": 124.45,
      "grad_norm": 2.220903158187866,
      "learning_rate": 1.6029465209066858e-06,
      "loss": 0.4703,
      "step": 304030
    },
    {
      "epoch": 124.45,
      "grad_norm": 2.4549286365509033,
      "learning_rate": 1.602796025384985e-06,
      "loss": 0.4557,
      "step": 304040
    },
    {
      "epoch": 124.46,
      "grad_norm": 2.4599111080169678,
      "learning_rate": 1.602645533257478e-06,
      "loss": 0.4651,
      "step": 304050
    },
    {
      "epoch": 124.46,
      "grad_norm": 1.6928341388702393,
      "learning_rate": 1.6024950445248571e-06,
      "loss": 0.446,
      "step": 304060
    },
    {
      "epoch": 124.47,
      "grad_norm": 2.2087390422821045,
      "learning_rate": 1.6023445591878103e-06,
      "loss": 0.4435,
      "step": 304070
    },
    {
      "epoch": 124.47,
      "grad_norm": 2.1404004096984863,
      "learning_rate": 1.6021940772470275e-06,
      "loss": 0.4714,
      "step": 304080
    },
    {
      "epoch": 124.47,
      "grad_norm": 2.006227493286133,
      "learning_rate": 1.6020435987031973e-06,
      "loss": 0.4554,
      "step": 304090
    },
    {
      "epoch": 124.48,
      "grad_norm": 1.9963035583496094,
      "learning_rate": 1.6018931235570089e-06,
      "loss": 0.4799,
      "step": 304100
    },
    {
      "epoch": 124.48,
      "grad_norm": 2.7750110626220703,
      "learning_rate": 1.601742651809152e-06,
      "loss": 0.4582,
      "step": 304110
    },
    {
      "epoch": 124.49,
      "grad_norm": 2.3360540866851807,
      "learning_rate": 1.6015921834603162e-06,
      "loss": 0.4579,
      "step": 304120
    },
    {
      "epoch": 124.49,
      "grad_norm": 1.955053448677063,
      "learning_rate": 1.60144171851119e-06,
      "loss": 0.4476,
      "step": 304130
    },
    {
      "epoch": 124.49,
      "grad_norm": 1.720155119895935,
      "learning_rate": 1.6012912569624635e-06,
      "loss": 0.4628,
      "step": 304140
    },
    {
      "epoch": 124.5,
      "grad_norm": 1.996586799621582,
      "learning_rate": 1.6011407988148254e-06,
      "loss": 0.458,
      "step": 304150
    },
    {
      "epoch": 124.5,
      "grad_norm": 1.565326452255249,
      "learning_rate": 1.6009903440689651e-06,
      "loss": 0.4489,
      "step": 304160
    },
    {
      "epoch": 124.51,
      "grad_norm": 1.942138910293579,
      "learning_rate": 1.6008398927255717e-06,
      "loss": 0.4631,
      "step": 304170
    },
    {
      "epoch": 124.51,
      "grad_norm": 2.0893850326538086,
      "learning_rate": 1.6006894447853342e-06,
      "loss": 0.4531,
      "step": 304180
    },
    {
      "epoch": 124.51,
      "grad_norm": 2.1540756225585938,
      "learning_rate": 1.6005390002489423e-06,
      "loss": 0.4442,
      "step": 304190
    },
    {
      "epoch": 124.52,
      "grad_norm": 1.8345869779586792,
      "learning_rate": 1.6003885591170848e-06,
      "loss": 0.4734,
      "step": 304200
    },
    {
      "epoch": 124.52,
      "grad_norm": 1.6965420246124268,
      "learning_rate": 1.6002381213904507e-06,
      "loss": 0.4675,
      "step": 304210
    },
    {
      "epoch": 124.53,
      "grad_norm": 2.1699001789093018,
      "learning_rate": 1.6000876870697295e-06,
      "loss": 0.4525,
      "step": 304220
    },
    {
      "epoch": 124.53,
      "grad_norm": 1.6954096555709839,
      "learning_rate": 1.59993725615561e-06,
      "loss": 0.4905,
      "step": 304230
    },
    {
      "epoch": 124.54,
      "grad_norm": 1.9773197174072266,
      "learning_rate": 1.5997868286487817e-06,
      "loss": 0.4781,
      "step": 304240
    },
    {
      "epoch": 124.54,
      "grad_norm": 2.08318829536438,
      "learning_rate": 1.5996364045499332e-06,
      "loss": 0.4609,
      "step": 304250
    },
    {
      "epoch": 124.54,
      "grad_norm": 2.3439781665802,
      "learning_rate": 1.599485983859754e-06,
      "loss": 0.4756,
      "step": 304260
    },
    {
      "epoch": 124.55,
      "grad_norm": 1.9961590766906738,
      "learning_rate": 1.5993355665789334e-06,
      "loss": 0.46,
      "step": 304270
    },
    {
      "epoch": 124.55,
      "grad_norm": 1.8950214385986328,
      "learning_rate": 1.5991851527081598e-06,
      "loss": 0.4431,
      "step": 304280
    },
    {
      "epoch": 124.56,
      "grad_norm": 1.817517638206482,
      "learning_rate": 1.5990347422481225e-06,
      "loss": 0.4606,
      "step": 304290
    },
    {
      "epoch": 124.56,
      "grad_norm": 1.9146322011947632,
      "learning_rate": 1.5988843351995109e-06,
      "loss": 0.4554,
      "step": 304300
    },
    {
      "epoch": 124.56,
      "grad_norm": 1.8045318126678467,
      "learning_rate": 1.598733931563013e-06,
      "loss": 0.4579,
      "step": 304310
    },
    {
      "epoch": 124.57,
      "grad_norm": 1.6937415599822998,
      "learning_rate": 1.598583531339318e-06,
      "loss": 0.4425,
      "step": 304320
    },
    {
      "epoch": 124.57,
      "grad_norm": 1.9409979581832886,
      "learning_rate": 1.5984331345291158e-06,
      "loss": 0.4915,
      "step": 304330
    },
    {
      "epoch": 124.58,
      "grad_norm": 2.1466546058654785,
      "learning_rate": 1.5982827411330943e-06,
      "loss": 0.446,
      "step": 304340
    },
    {
      "epoch": 124.58,
      "grad_norm": 2.5909624099731445,
      "learning_rate": 1.5981323511519432e-06,
      "loss": 0.4461,
      "step": 304350
    },
    {
      "epoch": 124.58,
      "grad_norm": 2.0262563228607178,
      "learning_rate": 1.597981964586351e-06,
      "loss": 0.464,
      "step": 304360
    },
    {
      "epoch": 124.59,
      "grad_norm": 2.351527214050293,
      "learning_rate": 1.5978315814370062e-06,
      "loss": 0.4439,
      "step": 304370
    },
    {
      "epoch": 124.59,
      "grad_norm": 2.1135153770446777,
      "learning_rate": 1.5976812017045987e-06,
      "loss": 0.468,
      "step": 304380
    },
    {
      "epoch": 124.6,
      "grad_norm": 2.2213385105133057,
      "learning_rate": 1.5975308253898171e-06,
      "loss": 0.4669,
      "step": 304390
    },
    {
      "epoch": 124.6,
      "grad_norm": 1.8711786270141602,
      "learning_rate": 1.5973804524933497e-06,
      "loss": 0.4622,
      "step": 304400
    },
    {
      "epoch": 124.6,
      "grad_norm": 1.862514853477478,
      "learning_rate": 1.5972300830158851e-06,
      "loss": 0.4378,
      "step": 304410
    },
    {
      "epoch": 124.61,
      "grad_norm": 1.9203025102615356,
      "learning_rate": 1.597079716958113e-06,
      "loss": 0.4755,
      "step": 304420
    },
    {
      "epoch": 124.61,
      "grad_norm": 2.3145527839660645,
      "learning_rate": 1.596929354320721e-06,
      "loss": 0.4618,
      "step": 304430
    },
    {
      "epoch": 124.62,
      "grad_norm": 1.7354378700256348,
      "learning_rate": 1.5967789951043998e-06,
      "loss": 0.4409,
      "step": 304440
    },
    {
      "epoch": 124.62,
      "grad_norm": 2.0182762145996094,
      "learning_rate": 1.5966286393098371e-06,
      "loss": 0.471,
      "step": 304450
    },
    {
      "epoch": 124.63,
      "grad_norm": 1.8687388896942139,
      "learning_rate": 1.596478286937721e-06,
      "loss": 0.4625,
      "step": 304460
    },
    {
      "epoch": 124.63,
      "grad_norm": 2.1212656497955322,
      "learning_rate": 1.5963279379887414e-06,
      "loss": 0.4541,
      "step": 304470
    },
    {
      "epoch": 124.63,
      "grad_norm": 1.613118290901184,
      "learning_rate": 1.5961775924635863e-06,
      "loss": 0.4808,
      "step": 304480
    },
    {
      "epoch": 124.64,
      "grad_norm": 2.478698253631592,
      "learning_rate": 1.596027250362945e-06,
      "loss": 0.4489,
      "step": 304490
    },
    {
      "epoch": 124.64,
      "grad_norm": 2.457261323928833,
      "learning_rate": 1.5958769116875053e-06,
      "loss": 0.4571,
      "step": 304500
    },
    {
      "epoch": 124.65,
      "grad_norm": 2.430814743041992,
      "learning_rate": 1.5957265764379568e-06,
      "loss": 0.4636,
      "step": 304510
    },
    {
      "epoch": 124.65,
      "grad_norm": 2.4357049465179443,
      "learning_rate": 1.5955762446149879e-06,
      "loss": 0.4571,
      "step": 304520
    },
    {
      "epoch": 124.65,
      "grad_norm": 3.351663827896118,
      "learning_rate": 1.5954259162192869e-06,
      "loss": 0.4617,
      "step": 304530
    },
    {
      "epoch": 124.66,
      "grad_norm": 1.556084394454956,
      "learning_rate": 1.5952755912515422e-06,
      "loss": 0.4694,
      "step": 304540
    },
    {
      "epoch": 124.66,
      "grad_norm": 2.082174301147461,
      "learning_rate": 1.5951252697124435e-06,
      "loss": 0.4536,
      "step": 304550
    },
    {
      "epoch": 124.67,
      "grad_norm": 1.5641281604766846,
      "learning_rate": 1.5949749516026782e-06,
      "loss": 0.4494,
      "step": 304560
    },
    {
      "epoch": 124.67,
      "grad_norm": 1.7650020122528076,
      "learning_rate": 1.5948246369229355e-06,
      "loss": 0.4514,
      "step": 304570
    },
    {
      "epoch": 124.67,
      "grad_norm": 2.077080249786377,
      "learning_rate": 1.594674325673904e-06,
      "loss": 0.4549,
      "step": 304580
    },
    {
      "epoch": 124.68,
      "grad_norm": 1.943991780281067,
      "learning_rate": 1.5945240178562723e-06,
      "loss": 0.4525,
      "step": 304590
    },
    {
      "epoch": 124.68,
      "grad_norm": 1.884523630142212,
      "learning_rate": 1.5943737134707286e-06,
      "loss": 0.4659,
      "step": 304600
    },
    {
      "epoch": 124.69,
      "grad_norm": 2.033825397491455,
      "learning_rate": 1.5942234125179623e-06,
      "loss": 0.4698,
      "step": 304610
    },
    {
      "epoch": 124.69,
      "grad_norm": 2.1760571002960205,
      "learning_rate": 1.5940731149986595e-06,
      "loss": 0.4671,
      "step": 304620
    },
    {
      "epoch": 124.7,
      "grad_norm": 1.9487061500549316,
      "learning_rate": 1.5939228209135114e-06,
      "loss": 0.4745,
      "step": 304630
    },
    {
      "epoch": 124.7,
      "grad_norm": 2.1879866123199463,
      "learning_rate": 1.593772530263206e-06,
      "loss": 0.4535,
      "step": 304640
    },
    {
      "epoch": 124.7,
      "grad_norm": 1.8461675643920898,
      "learning_rate": 1.5936222430484306e-06,
      "loss": 0.4503,
      "step": 304650
    },
    {
      "epoch": 124.71,
      "grad_norm": 1.9823241233825684,
      "learning_rate": 1.593471959269875e-06,
      "loss": 0.454,
      "step": 304660
    },
    {
      "epoch": 124.71,
      "grad_norm": 1.8316446542739868,
      "learning_rate": 1.5933216789282263e-06,
      "loss": 0.4499,
      "step": 304670
    },
    {
      "epoch": 124.72,
      "grad_norm": 1.8775073289871216,
      "learning_rate": 1.593171402024174e-06,
      "loss": 0.4676,
      "step": 304680
    },
    {
      "epoch": 124.72,
      "grad_norm": 1.9537330865859985,
      "learning_rate": 1.5930211285584054e-06,
      "loss": 0.4887,
      "step": 304690
    },
    {
      "epoch": 124.72,
      "grad_norm": 1.6180325746536255,
      "learning_rate": 1.5928708585316095e-06,
      "loss": 0.4663,
      "step": 304700
    },
    {
      "epoch": 124.73,
      "grad_norm": 2.1078615188598633,
      "learning_rate": 1.5927205919444753e-06,
      "loss": 0.4585,
      "step": 304710
    },
    {
      "epoch": 124.73,
      "grad_norm": 1.835328221321106,
      "learning_rate": 1.59257032879769e-06,
      "loss": 0.4495,
      "step": 304720
    },
    {
      "epoch": 124.74,
      "grad_norm": 1.9102131128311157,
      "learning_rate": 1.5924200690919423e-06,
      "loss": 0.4718,
      "step": 304730
    },
    {
      "epoch": 124.74,
      "grad_norm": 1.6381080150604248,
      "learning_rate": 1.5922698128279213e-06,
      "loss": 0.4538,
      "step": 304740
    },
    {
      "epoch": 124.74,
      "grad_norm": 1.9636280536651611,
      "learning_rate": 1.5921195600063148e-06,
      "loss": 0.4531,
      "step": 304750
    },
    {
      "epoch": 124.75,
      "grad_norm": 1.80527925491333,
      "learning_rate": 1.5919693106278103e-06,
      "loss": 0.4684,
      "step": 304760
    },
    {
      "epoch": 124.75,
      "grad_norm": 2.3369641304016113,
      "learning_rate": 1.591819064693097e-06,
      "loss": 0.4688,
      "step": 304770
    },
    {
      "epoch": 124.76,
      "grad_norm": 1.807136058807373,
      "learning_rate": 1.5916688222028626e-06,
      "loss": 0.4625,
      "step": 304780
    },
    {
      "epoch": 124.76,
      "grad_norm": 2.129969835281372,
      "learning_rate": 1.5915185831577954e-06,
      "loss": 0.4608,
      "step": 304790
    },
    {
      "epoch": 124.76,
      "grad_norm": 1.796234130859375,
      "learning_rate": 1.591368347558584e-06,
      "loss": 0.4495,
      "step": 304800
    },
    {
      "epoch": 124.77,
      "grad_norm": 1.9459446668624878,
      "learning_rate": 1.5912181154059163e-06,
      "loss": 0.4715,
      "step": 304810
    },
    {
      "epoch": 124.77,
      "grad_norm": 1.6750627756118774,
      "learning_rate": 1.5910678867004809e-06,
      "loss": 0.4548,
      "step": 304820
    },
    {
      "epoch": 124.78,
      "grad_norm": 1.7392364740371704,
      "learning_rate": 1.5909176614429657e-06,
      "loss": 0.4439,
      "step": 304830
    },
    {
      "epoch": 124.78,
      "grad_norm": 2.0066895484924316,
      "learning_rate": 1.5907674396340586e-06,
      "loss": 0.4515,
      "step": 304840
    },
    {
      "epoch": 124.79,
      "grad_norm": 1.6700550317764282,
      "learning_rate": 1.5906172212744484e-06,
      "loss": 0.4619,
      "step": 304850
    },
    {
      "epoch": 124.79,
      "grad_norm": 3.5638937950134277,
      "learning_rate": 1.5904670063648226e-06,
      "loss": 0.4601,
      "step": 304860
    },
    {
      "epoch": 124.79,
      "grad_norm": 1.7659341096878052,
      "learning_rate": 1.5903167949058695e-06,
      "loss": 0.4571,
      "step": 304870
    },
    {
      "epoch": 124.8,
      "grad_norm": 1.949293851852417,
      "learning_rate": 1.5901665868982778e-06,
      "loss": 0.4726,
      "step": 304880
    },
    {
      "epoch": 124.8,
      "grad_norm": 1.971866488456726,
      "learning_rate": 1.5900163823427347e-06,
      "loss": 0.4561,
      "step": 304890
    },
    {
      "epoch": 124.81,
      "grad_norm": 1.8362793922424316,
      "learning_rate": 1.589866181239928e-06,
      "loss": 0.4614,
      "step": 304900
    },
    {
      "epoch": 124.81,
      "grad_norm": 1.65325129032135,
      "learning_rate": 1.5897159835905468e-06,
      "loss": 0.4503,
      "step": 304910
    },
    {
      "epoch": 124.81,
      "grad_norm": 2.327749490737915,
      "learning_rate": 1.5895657893952784e-06,
      "loss": 0.4654,
      "step": 304920
    },
    {
      "epoch": 124.82,
      "grad_norm": 2.134944200515747,
      "learning_rate": 1.5894155986548113e-06,
      "loss": 0.4895,
      "step": 304930
    },
    {
      "epoch": 124.82,
      "grad_norm": 2.0399491786956787,
      "learning_rate": 1.5892654113698334e-06,
      "loss": 0.4547,
      "step": 304940
    },
    {
      "epoch": 124.83,
      "grad_norm": 2.125582218170166,
      "learning_rate": 1.589115227541032e-06,
      "loss": 0.4476,
      "step": 304950
    },
    {
      "epoch": 124.83,
      "grad_norm": 1.6382815837860107,
      "learning_rate": 1.5889650471690963e-06,
      "loss": 0.4588,
      "step": 304960
    },
    {
      "epoch": 124.83,
      "grad_norm": 1.8368117809295654,
      "learning_rate": 1.5888148702547135e-06,
      "loss": 0.4536,
      "step": 304970
    },
    {
      "epoch": 124.84,
      "grad_norm": 2.5271079540252686,
      "learning_rate": 1.5886646967985714e-06,
      "loss": 0.4509,
      "step": 304980
    },
    {
      "epoch": 124.84,
      "grad_norm": 1.861318588256836,
      "learning_rate": 1.588514526801357e-06,
      "loss": 0.4545,
      "step": 304990
    },
    {
      "epoch": 124.85,
      "grad_norm": 1.8868982791900635,
      "learning_rate": 1.5883643602637603e-06,
      "loss": 0.4674,
      "step": 305000
    },
    {
      "epoch": 124.85,
      "grad_norm": 1.885766625404358,
      "learning_rate": 1.5882141971864686e-06,
      "loss": 0.477,
      "step": 305010
    },
    {
      "epoch": 124.85,
      "grad_norm": 1.9066892862319946,
      "learning_rate": 1.5880640375701696e-06,
      "loss": 0.4725,
      "step": 305020
    },
    {
      "epoch": 124.86,
      "grad_norm": 1.998471975326538,
      "learning_rate": 1.5879138814155505e-06,
      "loss": 0.4693,
      "step": 305030
    },
    {
      "epoch": 124.86,
      "grad_norm": 2.082658529281616,
      "learning_rate": 1.5877637287232996e-06,
      "loss": 0.4734,
      "step": 305040
    },
    {
      "epoch": 124.87,
      "grad_norm": 2.1803483963012695,
      "learning_rate": 1.5876135794941047e-06,
      "loss": 0.4505,
      "step": 305050
    },
    {
      "epoch": 124.87,
      "grad_norm": 2.2979705333709717,
      "learning_rate": 1.5874634337286536e-06,
      "loss": 0.4481,
      "step": 305060
    },
    {
      "epoch": 124.88,
      "grad_norm": 1.6916471719741821,
      "learning_rate": 1.5873132914276344e-06,
      "loss": 0.4765,
      "step": 305070
    },
    {
      "epoch": 124.88,
      "grad_norm": 2.4609885215759277,
      "learning_rate": 1.5871631525917344e-06,
      "loss": 0.4536,
      "step": 305080
    },
    {
      "epoch": 124.88,
      "grad_norm": 2.1086466312408447,
      "learning_rate": 1.5870130172216415e-06,
      "loss": 0.466,
      "step": 305090
    },
    {
      "epoch": 124.89,
      "grad_norm": 1.9452935457229614,
      "learning_rate": 1.5868628853180437e-06,
      "loss": 0.4534,
      "step": 305100
    },
    {
      "epoch": 124.89,
      "grad_norm": 1.6023198366165161,
      "learning_rate": 1.586712756881629e-06,
      "loss": 0.4393,
      "step": 305110
    },
    {
      "epoch": 124.9,
      "grad_norm": 2.031221866607666,
      "learning_rate": 1.586562631913084e-06,
      "loss": 0.4515,
      "step": 305120
    },
    {
      "epoch": 124.9,
      "grad_norm": 1.6703975200653076,
      "learning_rate": 1.5864125104130974e-06,
      "loss": 0.4333,
      "step": 305130
    },
    {
      "epoch": 124.9,
      "grad_norm": 1.8047300577163696,
      "learning_rate": 1.5862623923823563e-06,
      "loss": 0.465,
      "step": 305140
    },
    {
      "epoch": 124.91,
      "grad_norm": 1.819018006324768,
      "learning_rate": 1.5861122778215486e-06,
      "loss": 0.4633,
      "step": 305150
    },
    {
      "epoch": 124.91,
      "grad_norm": 1.7843782901763916,
      "learning_rate": 1.5859621667313623e-06,
      "loss": 0.455,
      "step": 305160
    },
    {
      "epoch": 124.92,
      "grad_norm": 2.2117910385131836,
      "learning_rate": 1.5858120591124842e-06,
      "loss": 0.4657,
      "step": 305170
    },
    {
      "epoch": 124.92,
      "grad_norm": 1.8311982154846191,
      "learning_rate": 1.5856619549656023e-06,
      "loss": 0.4594,
      "step": 305180
    },
    {
      "epoch": 124.92,
      "grad_norm": 2.0395865440368652,
      "learning_rate": 1.5855118542914044e-06,
      "loss": 0.4447,
      "step": 305190
    },
    {
      "epoch": 124.93,
      "grad_norm": 2.0065414905548096,
      "learning_rate": 1.5853617570905787e-06,
      "loss": 0.4751,
      "step": 305200
    },
    {
      "epoch": 124.93,
      "grad_norm": 1.6401315927505493,
      "learning_rate": 1.5852116633638118e-06,
      "loss": 0.4521,
      "step": 305210
    },
    {
      "epoch": 124.94,
      "grad_norm": 1.973631501197815,
      "learning_rate": 1.5850615731117911e-06,
      "loss": 0.4556,
      "step": 305220
    },
    {
      "epoch": 124.94,
      "grad_norm": 1.6441181898117065,
      "learning_rate": 1.584911486335205e-06,
      "loss": 0.4546,
      "step": 305230
    },
    {
      "epoch": 124.94,
      "grad_norm": 2.6307601928710938,
      "learning_rate": 1.5847614030347412e-06,
      "loss": 0.4563,
      "step": 305240
    },
    {
      "epoch": 124.95,
      "grad_norm": 2.005014181137085,
      "learning_rate": 1.5846113232110858e-06,
      "loss": 0.4614,
      "step": 305250
    },
    {
      "epoch": 124.95,
      "grad_norm": 1.9558192491531372,
      "learning_rate": 1.5844612468649276e-06,
      "loss": 0.4717,
      "step": 305260
    },
    {
      "epoch": 124.96,
      "grad_norm": 1.7035410404205322,
      "learning_rate": 1.5843111739969533e-06,
      "loss": 0.4737,
      "step": 305270
    },
    {
      "epoch": 124.96,
      "grad_norm": 2.02095627784729,
      "learning_rate": 1.5841611046078505e-06,
      "loss": 0.4596,
      "step": 305280
    },
    {
      "epoch": 124.97,
      "grad_norm": 2.3091163635253906,
      "learning_rate": 1.5840110386983069e-06,
      "loss": 0.4217,
      "step": 305290
    },
    {
      "epoch": 124.97,
      "grad_norm": 1.8758294582366943,
      "learning_rate": 1.5838609762690101e-06,
      "loss": 0.4525,
      "step": 305300
    },
    {
      "epoch": 124.97,
      "grad_norm": 2.2182681560516357,
      "learning_rate": 1.5837109173206467e-06,
      "loss": 0.4743,
      "step": 305310
    },
    {
      "epoch": 124.98,
      "grad_norm": 2.1583707332611084,
      "learning_rate": 1.5835608618539051e-06,
      "loss": 0.4766,
      "step": 305320
    },
    {
      "epoch": 124.98,
      "grad_norm": 1.9596418142318726,
      "learning_rate": 1.5834108098694722e-06,
      "loss": 0.4722,
      "step": 305330
    },
    {
      "epoch": 124.99,
      "grad_norm": 1.5310941934585571,
      "learning_rate": 1.5832607613680354e-06,
      "loss": 0.457,
      "step": 305340
    },
    {
      "epoch": 124.99,
      "grad_norm": 1.9616481065750122,
      "learning_rate": 1.5831107163502818e-06,
      "loss": 0.4642,
      "step": 305350
    },
    {
      "epoch": 124.99,
      "grad_norm": 2.149092435836792,
      "learning_rate": 1.5829606748168991e-06,
      "loss": 0.484,
      "step": 305360
    },
    {
      "epoch": 125.0,
      "grad_norm": 1.8342081308364868,
      "learning_rate": 1.5828106367685735e-06,
      "loss": 0.4471,
      "step": 305370
    },
    {
      "epoch": 125.0,
      "eval_loss": 0.45987436175346375,
      "eval_runtime": 52.9717,
      "eval_samples_per_second": 65.11,
      "eval_steps_per_second": 8.155,
      "step": 305375
    },
    {
      "epoch": 125.0,
      "grad_norm": 2.2745397090911865,
      "learning_rate": 1.5826606022059945e-06,
      "loss": 0.4569,
      "step": 305380
    },
    {
      "epoch": 125.01,
      "grad_norm": 2.5329155921936035,
      "learning_rate": 1.5825105711298476e-06,
      "loss": 0.4566,
      "step": 305390
    },
    {
      "epoch": 125.01,
      "grad_norm": 2.4783318042755127,
      "learning_rate": 1.582360543540821e-06,
      "loss": 0.441,
      "step": 305400
    },
    {
      "epoch": 125.01,
      "grad_norm": 2.1487910747528076,
      "learning_rate": 1.5822105194396012e-06,
      "loss": 0.4645,
      "step": 305410
    },
    {
      "epoch": 125.02,
      "grad_norm": 1.7239350080490112,
      "learning_rate": 1.582060498826876e-06,
      "loss": 0.4695,
      "step": 305420
    },
    {
      "epoch": 125.02,
      "grad_norm": 2.0933761596679688,
      "learning_rate": 1.5819104817033328e-06,
      "loss": 0.453,
      "step": 305430
    },
    {
      "epoch": 125.03,
      "grad_norm": 1.749402642250061,
      "learning_rate": 1.581760468069658e-06,
      "loss": 0.4731,
      "step": 305440
    },
    {
      "epoch": 125.03,
      "grad_norm": 1.8208541870117188,
      "learning_rate": 1.5816104579265394e-06,
      "loss": 0.4548,
      "step": 305450
    },
    {
      "epoch": 125.03,
      "grad_norm": 2.143751621246338,
      "learning_rate": 1.5814604512746638e-06,
      "loss": 0.4679,
      "step": 305460
    },
    {
      "epoch": 125.04,
      "grad_norm": 1.8811999559402466,
      "learning_rate": 1.5813104481147196e-06,
      "loss": 0.4675,
      "step": 305470
    },
    {
      "epoch": 125.04,
      "grad_norm": 2.1061065196990967,
      "learning_rate": 1.5811604484473919e-06,
      "loss": 0.4379,
      "step": 305480
    },
    {
      "epoch": 125.05,
      "grad_norm": 2.083479404449463,
      "learning_rate": 1.5810104522733689e-06,
      "loss": 0.4619,
      "step": 305490
    },
    {
      "epoch": 125.05,
      "grad_norm": 2.189286708831787,
      "learning_rate": 1.5808604595933377e-06,
      "loss": 0.4685,
      "step": 305500
    },
    {
      "epoch": 125.06,
      "grad_norm": 1.960045576095581,
      "learning_rate": 1.5807104704079853e-06,
      "loss": 0.4469,
      "step": 305510
    },
    {
      "epoch": 125.06,
      "grad_norm": 2.204901933670044,
      "learning_rate": 1.580560484717999e-06,
      "loss": 0.4687,
      "step": 305520
    },
    {
      "epoch": 125.06,
      "grad_norm": 1.9338736534118652,
      "learning_rate": 1.5804105025240653e-06,
      "loss": 0.4503,
      "step": 305530
    },
    {
      "epoch": 125.07,
      "grad_norm": 2.134901762008667,
      "learning_rate": 1.5802605238268723e-06,
      "loss": 0.4405,
      "step": 305540
    },
    {
      "epoch": 125.07,
      "grad_norm": 2.3059377670288086,
      "learning_rate": 1.5801105486271062e-06,
      "loss": 0.4819,
      "step": 305550
    },
    {
      "epoch": 125.08,
      "grad_norm": 2.142587661743164,
      "learning_rate": 1.5799605769254528e-06,
      "loss": 0.446,
      "step": 305560
    },
    {
      "epoch": 125.08,
      "grad_norm": 2.526357650756836,
      "learning_rate": 1.5798106087226017e-06,
      "loss": 0.4583,
      "step": 305570
    },
    {
      "epoch": 125.08,
      "grad_norm": 2.0698394775390625,
      "learning_rate": 1.5796606440192385e-06,
      "loss": 0.4669,
      "step": 305580
    },
    {
      "epoch": 125.09,
      "grad_norm": 1.8138622045516968,
      "learning_rate": 1.57951068281605e-06,
      "loss": 0.4422,
      "step": 305590
    },
    {
      "epoch": 125.09,
      "grad_norm": 3.0772998332977295,
      "learning_rate": 1.5793607251137243e-06,
      "loss": 0.4731,
      "step": 305600
    },
    {
      "epoch": 125.1,
      "grad_norm": 1.558874249458313,
      "learning_rate": 1.5792107709129473e-06,
      "loss": 0.4444,
      "step": 305610
    },
    {
      "epoch": 125.1,
      "grad_norm": 2.190398931503296,
      "learning_rate": 1.5790608202144052e-06,
      "loss": 0.4481,
      "step": 305620
    },
    {
      "epoch": 125.1,
      "grad_norm": 1.9851770401000977,
      "learning_rate": 1.5789108730187865e-06,
      "loss": 0.4685,
      "step": 305630
    },
    {
      "epoch": 125.11,
      "grad_norm": 2.2138214111328125,
      "learning_rate": 1.5787609293267773e-06,
      "loss": 0.4448,
      "step": 305640
    },
    {
      "epoch": 125.11,
      "grad_norm": 1.8306500911712646,
      "learning_rate": 1.5786109891390646e-06,
      "loss": 0.4594,
      "step": 305650
    },
    {
      "epoch": 125.12,
      "grad_norm": 3.2358922958374023,
      "learning_rate": 1.5784610524563352e-06,
      "loss": 0.4648,
      "step": 305660
    },
    {
      "epoch": 125.12,
      "grad_norm": 1.7318788766860962,
      "learning_rate": 1.5783111192792759e-06,
      "loss": 0.4597,
      "step": 305670
    },
    {
      "epoch": 125.12,
      "grad_norm": 2.2479162216186523,
      "learning_rate": 1.5781611896085734e-06,
      "loss": 0.4425,
      "step": 305680
    },
    {
      "epoch": 125.13,
      "grad_norm": 2.0273079872131348,
      "learning_rate": 1.5780112634449151e-06,
      "loss": 0.4501,
      "step": 305690
    },
    {
      "epoch": 125.13,
      "grad_norm": 1.6569101810455322,
      "learning_rate": 1.577861340788987e-06,
      "loss": 0.4555,
      "step": 305700
    },
    {
      "epoch": 125.14,
      "grad_norm": 1.9004713296890259,
      "learning_rate": 1.5777114216414765e-06,
      "loss": 0.4668,
      "step": 305710
    },
    {
      "epoch": 125.14,
      "grad_norm": 2.001089334487915,
      "learning_rate": 1.5775615060030697e-06,
      "loss": 0.4537,
      "step": 305720
    },
    {
      "epoch": 125.15,
      "grad_norm": 2.0118160247802734,
      "learning_rate": 1.5774115938744539e-06,
      "loss": 0.4544,
      "step": 305730
    },
    {
      "epoch": 125.15,
      "grad_norm": 2.0733320713043213,
      "learning_rate": 1.5772616852563157e-06,
      "loss": 0.4773,
      "step": 305740
    },
    {
      "epoch": 125.15,
      "grad_norm": 1.7959306240081787,
      "learning_rate": 1.577111780149341e-06,
      "loss": 0.4475,
      "step": 305750
    },
    {
      "epoch": 125.16,
      "grad_norm": 1.9132384061813354,
      "learning_rate": 1.5769618785542184e-06,
      "loss": 0.4421,
      "step": 305760
    },
    {
      "epoch": 125.16,
      "grad_norm": 4.614712715148926,
      "learning_rate": 1.576811980471633e-06,
      "loss": 0.4653,
      "step": 305770
    },
    {
      "epoch": 125.17,
      "grad_norm": 1.7101285457611084,
      "learning_rate": 1.576662085902272e-06,
      "loss": 0.4556,
      "step": 305780
    },
    {
      "epoch": 125.17,
      "grad_norm": 2.0349104404449463,
      "learning_rate": 1.5765121948468217e-06,
      "loss": 0.4664,
      "step": 305790
    },
    {
      "epoch": 125.17,
      "grad_norm": 1.9964325428009033,
      "learning_rate": 1.576362307305969e-06,
      "loss": 0.4478,
      "step": 305800
    },
    {
      "epoch": 125.18,
      "grad_norm": 2.4545958042144775,
      "learning_rate": 1.5762124232804006e-06,
      "loss": 0.4594,
      "step": 305810
    },
    {
      "epoch": 125.18,
      "grad_norm": 1.666007161140442,
      "learning_rate": 1.5760625427708028e-06,
      "loss": 0.4737,
      "step": 305820
    },
    {
      "epoch": 125.19,
      "grad_norm": 2.1214988231658936,
      "learning_rate": 1.575912665777863e-06,
      "loss": 0.452,
      "step": 305830
    },
    {
      "epoch": 125.19,
      "grad_norm": 2.3437438011169434,
      "learning_rate": 1.5757627923022668e-06,
      "loss": 0.4562,
      "step": 305840
    },
    {
      "epoch": 125.19,
      "grad_norm": 1.8605659008026123,
      "learning_rate": 1.5756129223447008e-06,
      "loss": 0.4793,
      "step": 305850
    },
    {
      "epoch": 125.2,
      "grad_norm": 2.1412322521209717,
      "learning_rate": 1.5754630559058519e-06,
      "loss": 0.4708,
      "step": 305860
    },
    {
      "epoch": 125.2,
      "grad_norm": 1.9953842163085938,
      "learning_rate": 1.5753131929864065e-06,
      "loss": 0.4395,
      "step": 305870
    },
    {
      "epoch": 125.21,
      "grad_norm": 2.3690335750579834,
      "learning_rate": 1.5751633335870513e-06,
      "loss": 0.4639,
      "step": 305880
    },
    {
      "epoch": 125.21,
      "grad_norm": 2.1452219486236572,
      "learning_rate": 1.5750134777084724e-06,
      "loss": 0.4475,
      "step": 305890
    },
    {
      "epoch": 125.21,
      "grad_norm": 1.934064507484436,
      "learning_rate": 1.574863625351356e-06,
      "loss": 0.4504,
      "step": 305900
    },
    {
      "epoch": 125.22,
      "grad_norm": 2.0804872512817383,
      "learning_rate": 1.5747137765163902e-06,
      "loss": 0.4549,
      "step": 305910
    },
    {
      "epoch": 125.22,
      "grad_norm": 1.9230420589447021,
      "learning_rate": 1.5745639312042595e-06,
      "loss": 0.457,
      "step": 305920
    },
    {
      "epoch": 125.23,
      "grad_norm": 1.9582288265228271,
      "learning_rate": 1.574414089415651e-06,
      "loss": 0.4603,
      "step": 305930
    },
    {
      "epoch": 125.23,
      "grad_norm": 1.8499507904052734,
      "learning_rate": 1.5742642511512502e-06,
      "loss": 0.4522,
      "step": 305940
    },
    {
      "epoch": 125.24,
      "grad_norm": 1.8378639221191406,
      "learning_rate": 1.5741144164117456e-06,
      "loss": 0.4658,
      "step": 305950
    },
    {
      "epoch": 125.24,
      "grad_norm": 2.2687430381774902,
      "learning_rate": 1.5739645851978227e-06,
      "loss": 0.4818,
      "step": 305960
    },
    {
      "epoch": 125.24,
      "grad_norm": 2.1196939945220947,
      "learning_rate": 1.573814757510167e-06,
      "loss": 0.4565,
      "step": 305970
    },
    {
      "epoch": 125.25,
      "grad_norm": 2.0623738765716553,
      "learning_rate": 1.5736649333494657e-06,
      "loss": 0.4598,
      "step": 305980
    },
    {
      "epoch": 125.25,
      "grad_norm": 2.4604597091674805,
      "learning_rate": 1.5735151127164046e-06,
      "loss": 0.4691,
      "step": 305990
    },
    {
      "epoch": 125.26,
      "grad_norm": 1.9659883975982666,
      "learning_rate": 1.57336529561167e-06,
      "loss": 0.4873,
      "step": 306000
    },
    {
      "epoch": 125.26,
      "grad_norm": 2.3868446350097656,
      "learning_rate": 1.573215482035949e-06,
      "loss": 0.4668,
      "step": 306010
    },
    {
      "epoch": 125.26,
      "grad_norm": 3.166820526123047,
      "learning_rate": 1.5730656719899269e-06,
      "loss": 0.4483,
      "step": 306020
    },
    {
      "epoch": 125.27,
      "grad_norm": 1.8647814989089966,
      "learning_rate": 1.5729158654742904e-06,
      "loss": 0.4636,
      "step": 306030
    },
    {
      "epoch": 125.27,
      "grad_norm": 1.912459373474121,
      "learning_rate": 1.5727660624897258e-06,
      "loss": 0.4567,
      "step": 306040
    },
    {
      "epoch": 125.28,
      "grad_norm": 1.825597882270813,
      "learning_rate": 1.5726162630369196e-06,
      "loss": 0.4581,
      "step": 306050
    },
    {
      "epoch": 125.28,
      "grad_norm": 1.9210761785507202,
      "learning_rate": 1.5724664671165572e-06,
      "loss": 0.4607,
      "step": 306060
    },
    {
      "epoch": 125.28,
      "grad_norm": 2.311941623687744,
      "learning_rate": 1.5723166747293253e-06,
      "loss": 0.4639,
      "step": 306070
    },
    {
      "epoch": 125.29,
      "grad_norm": 2.0395069122314453,
      "learning_rate": 1.57216688587591e-06,
      "loss": 0.4626,
      "step": 306080
    },
    {
      "epoch": 125.29,
      "grad_norm": 1.705077886581421,
      "learning_rate": 1.5720171005569974e-06,
      "loss": 0.4446,
      "step": 306090
    },
    {
      "epoch": 125.3,
      "grad_norm": 1.6352125406265259,
      "learning_rate": 1.5718673187732735e-06,
      "loss": 0.4552,
      "step": 306100
    },
    {
      "epoch": 125.3,
      "grad_norm": 1.7710222005844116,
      "learning_rate": 1.571717540525425e-06,
      "loss": 0.47,
      "step": 306110
    },
    {
      "epoch": 125.3,
      "grad_norm": 1.97135329246521,
      "learning_rate": 1.5715677658141367e-06,
      "loss": 0.4584,
      "step": 306120
    },
    {
      "epoch": 125.31,
      "grad_norm": 2.3508355617523193,
      "learning_rate": 1.5714179946400962e-06,
      "loss": 0.4467,
      "step": 306130
    },
    {
      "epoch": 125.31,
      "grad_norm": 2.498969793319702,
      "learning_rate": 1.571268227003989e-06,
      "loss": 0.4468,
      "step": 306140
    },
    {
      "epoch": 125.32,
      "grad_norm": 1.9352434873580933,
      "learning_rate": 1.5711184629065016e-06,
      "loss": 0.4633,
      "step": 306150
    },
    {
      "epoch": 125.32,
      "grad_norm": 2.0718181133270264,
      "learning_rate": 1.5709687023483193e-06,
      "loss": 0.4613,
      "step": 306160
    },
    {
      "epoch": 125.33,
      "grad_norm": 2.247173309326172,
      "learning_rate": 1.5708189453301284e-06,
      "loss": 0.4485,
      "step": 306170
    },
    {
      "epoch": 125.33,
      "grad_norm": 2.1243443489074707,
      "learning_rate": 1.5706691918526155e-06,
      "loss": 0.4459,
      "step": 306180
    },
    {
      "epoch": 125.33,
      "grad_norm": 2.391618490219116,
      "learning_rate": 1.5705194419164657e-06,
      "loss": 0.4715,
      "step": 306190
    },
    {
      "epoch": 125.34,
      "grad_norm": 1.6544878482818604,
      "learning_rate": 1.570369695522365e-06,
      "loss": 0.4625,
      "step": 306200
    },
    {
      "epoch": 125.34,
      "grad_norm": 1.7806735038757324,
      "learning_rate": 1.570219952671e-06,
      "loss": 0.4482,
      "step": 306210
    },
    {
      "epoch": 125.35,
      "grad_norm": 1.9508745670318604,
      "learning_rate": 1.5700702133630562e-06,
      "loss": 0.4748,
      "step": 306220
    },
    {
      "epoch": 125.35,
      "grad_norm": 2.165440559387207,
      "learning_rate": 1.5699204775992197e-06,
      "loss": 0.4672,
      "step": 306230
    },
    {
      "epoch": 125.35,
      "grad_norm": 1.7618952989578247,
      "learning_rate": 1.5697707453801764e-06,
      "loss": 0.4593,
      "step": 306240
    },
    {
      "epoch": 125.36,
      "grad_norm": 1.7757060527801514,
      "learning_rate": 1.5696210167066122e-06,
      "loss": 0.452,
      "step": 306250
    },
    {
      "epoch": 125.36,
      "grad_norm": 2.065850257873535,
      "learning_rate": 1.569471291579213e-06,
      "loss": 0.4475,
      "step": 306260
    },
    {
      "epoch": 125.37,
      "grad_norm": 1.6597845554351807,
      "learning_rate": 1.5693215699986653e-06,
      "loss": 0.4463,
      "step": 306270
    },
    {
      "epoch": 125.37,
      "grad_norm": 1.988208293914795,
      "learning_rate": 1.5691718519656538e-06,
      "loss": 0.458,
      "step": 306280
    },
    {
      "epoch": 125.37,
      "grad_norm": 2.055997371673584,
      "learning_rate": 1.5690221374808645e-06,
      "loss": 0.4494,
      "step": 306290
    },
    {
      "epoch": 125.38,
      "grad_norm": 2.0373690128326416,
      "learning_rate": 1.5688724265449838e-06,
      "loss": 0.4492,
      "step": 306300
    },
    {
      "epoch": 125.38,
      "grad_norm": 1.9894912242889404,
      "learning_rate": 1.5687227191586961e-06,
      "loss": 0.4679,
      "step": 306310
    },
    {
      "epoch": 125.39,
      "grad_norm": 2.012784242630005,
      "learning_rate": 1.56857301532269e-06,
      "loss": 0.4739,
      "step": 306320
    },
    {
      "epoch": 125.39,
      "grad_norm": 1.9716033935546875,
      "learning_rate": 1.5684233150376494e-06,
      "loss": 0.4624,
      "step": 306330
    },
    {
      "epoch": 125.4,
      "grad_norm": 1.6164309978485107,
      "learning_rate": 1.56827361830426e-06,
      "loss": 0.4679,
      "step": 306340
    },
    {
      "epoch": 125.4,
      "grad_norm": 1.9670758247375488,
      "learning_rate": 1.5681239251232077e-06,
      "loss": 0.4759,
      "step": 306350
    },
    {
      "epoch": 125.4,
      "grad_norm": 1.506751298904419,
      "learning_rate": 1.5679742354951783e-06,
      "loss": 0.4414,
      "step": 306360
    },
    {
      "epoch": 125.41,
      "grad_norm": 2.802361011505127,
      "learning_rate": 1.5678245494208577e-06,
      "loss": 0.4545,
      "step": 306370
    },
    {
      "epoch": 125.41,
      "grad_norm": 1.9106550216674805,
      "learning_rate": 1.5676748669009313e-06,
      "loss": 0.4688,
      "step": 306380
    },
    {
      "epoch": 125.42,
      "grad_norm": 2.2278175354003906,
      "learning_rate": 1.5675251879360847e-06,
      "loss": 0.4637,
      "step": 306390
    },
    {
      "epoch": 125.42,
      "grad_norm": 1.5868598222732544,
      "learning_rate": 1.5673755125270043e-06,
      "loss": 0.4557,
      "step": 306400
    },
    {
      "epoch": 125.42,
      "grad_norm": 1.9743605852127075,
      "learning_rate": 1.5672258406743752e-06,
      "loss": 0.474,
      "step": 306410
    },
    {
      "epoch": 125.43,
      "grad_norm": 1.6219089031219482,
      "learning_rate": 1.5670761723788829e-06,
      "loss": 0.4505,
      "step": 306420
    },
    {
      "epoch": 125.43,
      "grad_norm": 2.173344612121582,
      "learning_rate": 1.5669265076412127e-06,
      "loss": 0.4628,
      "step": 306430
    },
    {
      "epoch": 125.44,
      "grad_norm": 2.2526237964630127,
      "learning_rate": 1.566776846462051e-06,
      "loss": 0.4712,
      "step": 306440
    },
    {
      "epoch": 125.44,
      "grad_norm": 1.9718469381332397,
      "learning_rate": 1.5666271888420826e-06,
      "loss": 0.4571,
      "step": 306450
    },
    {
      "epoch": 125.44,
      "grad_norm": 1.92162024974823,
      "learning_rate": 1.5664775347819937e-06,
      "loss": 0.4641,
      "step": 306460
    },
    {
      "epoch": 125.45,
      "grad_norm": 1.8483622074127197,
      "learning_rate": 1.56632788428247e-06,
      "loss": 0.4438,
      "step": 306470
    },
    {
      "epoch": 125.45,
      "grad_norm": 1.8694429397583008,
      "learning_rate": 1.566178237344196e-06,
      "loss": 0.449,
      "step": 306480
    },
    {
      "epoch": 125.46,
      "grad_norm": 1.7815431356430054,
      "learning_rate": 1.5660285939678584e-06,
      "loss": 0.4608,
      "step": 306490
    },
    {
      "epoch": 125.46,
      "grad_norm": 2.0436854362487793,
      "learning_rate": 1.5658789541541405e-06,
      "loss": 0.4621,
      "step": 306500
    },
    {
      "epoch": 125.46,
      "grad_norm": 2.1305196285247803,
      "learning_rate": 1.5657293179037308e-06,
      "loss": 0.4608,
      "step": 306510
    },
    {
      "epoch": 125.47,
      "grad_norm": 1.8734287023544312,
      "learning_rate": 1.565579685217313e-06,
      "loss": 0.4573,
      "step": 306520
    },
    {
      "epoch": 125.47,
      "grad_norm": 1.7413264513015747,
      "learning_rate": 1.5654300560955733e-06,
      "loss": 0.4522,
      "step": 306530
    },
    {
      "epoch": 125.48,
      "grad_norm": 2.468026638031006,
      "learning_rate": 1.5652804305391966e-06,
      "loss": 0.464,
      "step": 306540
    },
    {
      "epoch": 125.48,
      "grad_norm": 2.1666500568389893,
      "learning_rate": 1.5651308085488685e-06,
      "loss": 0.4646,
      "step": 306550
    },
    {
      "epoch": 125.49,
      "grad_norm": 1.9369035959243774,
      "learning_rate": 1.5649811901252743e-06,
      "loss": 0.4653,
      "step": 306560
    },
    {
      "epoch": 125.49,
      "grad_norm": 1.662052035331726,
      "learning_rate": 1.5648315752690993e-06,
      "loss": 0.4583,
      "step": 306570
    },
    {
      "epoch": 125.49,
      "grad_norm": 2.1373374462127686,
      "learning_rate": 1.5646819639810286e-06,
      "loss": 0.4625,
      "step": 306580
    },
    {
      "epoch": 125.5,
      "grad_norm": 3.5829429626464844,
      "learning_rate": 1.5645323562617485e-06,
      "loss": 0.4607,
      "step": 306590
    },
    {
      "epoch": 125.5,
      "grad_norm": 2.2278714179992676,
      "learning_rate": 1.5643827521119434e-06,
      "loss": 0.4745,
      "step": 306600
    },
    {
      "epoch": 125.51,
      "grad_norm": 1.8110439777374268,
      "learning_rate": 1.5642331515322989e-06,
      "loss": 0.4453,
      "step": 306610
    },
    {
      "epoch": 125.51,
      "grad_norm": 2.19089674949646,
      "learning_rate": 1.5640835545235005e-06,
      "loss": 0.4625,
      "step": 306620
    },
    {
      "epoch": 125.51,
      "grad_norm": 2.070324182510376,
      "learning_rate": 1.563933961086234e-06,
      "loss": 0.4716,
      "step": 306630
    },
    {
      "epoch": 125.52,
      "grad_norm": 2.1946399211883545,
      "learning_rate": 1.563784371221183e-06,
      "loss": 0.4762,
      "step": 306640
    },
    {
      "epoch": 125.52,
      "grad_norm": 2.173793315887451,
      "learning_rate": 1.5636347849290343e-06,
      "loss": 0.4566,
      "step": 306650
    },
    {
      "epoch": 125.53,
      "grad_norm": 2.118452548980713,
      "learning_rate": 1.5634852022104725e-06,
      "loss": 0.4551,
      "step": 306660
    },
    {
      "epoch": 125.53,
      "grad_norm": 2.44051194190979,
      "learning_rate": 1.5633356230661828e-06,
      "loss": 0.4419,
      "step": 306670
    },
    {
      "epoch": 125.53,
      "grad_norm": 2.5146985054016113,
      "learning_rate": 1.5631860474968505e-06,
      "loss": 0.4509,
      "step": 306680
    },
    {
      "epoch": 125.54,
      "grad_norm": 2.021540880203247,
      "learning_rate": 1.5630364755031602e-06,
      "loss": 0.438,
      "step": 306690
    },
    {
      "epoch": 125.54,
      "grad_norm": 1.750830888748169,
      "learning_rate": 1.5628869070857984e-06,
      "loss": 0.4586,
      "step": 306700
    },
    {
      "epoch": 125.55,
      "grad_norm": 2.642261028289795,
      "learning_rate": 1.5627373422454494e-06,
      "loss": 0.448,
      "step": 306710
    },
    {
      "epoch": 125.55,
      "grad_norm": 1.8494584560394287,
      "learning_rate": 1.5625877809827984e-06,
      "loss": 0.4616,
      "step": 306720
    },
    {
      "epoch": 125.55,
      "grad_norm": 1.8625074625015259,
      "learning_rate": 1.5624382232985303e-06,
      "loss": 0.4732,
      "step": 306730
    },
    {
      "epoch": 125.56,
      "grad_norm": 2.0020251274108887,
      "learning_rate": 1.5622886691933308e-06,
      "loss": 0.4465,
      "step": 306740
    },
    {
      "epoch": 125.56,
      "grad_norm": 1.921181082725525,
      "learning_rate": 1.5621391186678845e-06,
      "loss": 0.4711,
      "step": 306750
    },
    {
      "epoch": 125.57,
      "grad_norm": 2.0057454109191895,
      "learning_rate": 1.5619895717228763e-06,
      "loss": 0.4578,
      "step": 306760
    },
    {
      "epoch": 125.57,
      "grad_norm": 1.7330206632614136,
      "learning_rate": 1.5618400283589926e-06,
      "loss": 0.4706,
      "step": 306770
    },
    {
      "epoch": 125.58,
      "grad_norm": 2.1959304809570312,
      "learning_rate": 1.5616904885769165e-06,
      "loss": 0.465,
      "step": 306780
    },
    {
      "epoch": 125.58,
      "grad_norm": 2.297640085220337,
      "learning_rate": 1.5615409523773338e-06,
      "loss": 0.4746,
      "step": 306790
    },
    {
      "epoch": 125.58,
      "grad_norm": 1.9785106182098389,
      "learning_rate": 1.56139141976093e-06,
      "loss": 0.4848,
      "step": 306800
    },
    {
      "epoch": 125.59,
      "grad_norm": 1.966832160949707,
      "learning_rate": 1.5612418907283894e-06,
      "loss": 0.4607,
      "step": 306810
    },
    {
      "epoch": 125.59,
      "grad_norm": 1.8435797691345215,
      "learning_rate": 1.5610923652803975e-06,
      "loss": 0.4587,
      "step": 306820
    },
    {
      "epoch": 125.6,
      "grad_norm": 1.7699289321899414,
      "learning_rate": 1.5609428434176387e-06,
      "loss": 0.4507,
      "step": 306830
    },
    {
      "epoch": 125.6,
      "grad_norm": 1.6269190311431885,
      "learning_rate": 1.5607933251407985e-06,
      "loss": 0.4462,
      "step": 306840
    },
    {
      "epoch": 125.6,
      "grad_norm": 2.0006701946258545,
      "learning_rate": 1.560643810450562e-06,
      "loss": 0.4634,
      "step": 306850
    },
    {
      "epoch": 125.61,
      "grad_norm": 2.126990556716919,
      "learning_rate": 1.5604942993476133e-06,
      "loss": 0.4457,
      "step": 306860
    },
    {
      "epoch": 125.61,
      "grad_norm": 1.7078930139541626,
      "learning_rate": 1.5603447918326375e-06,
      "loss": 0.4424,
      "step": 306870
    },
    {
      "epoch": 125.62,
      "grad_norm": 1.9593193531036377,
      "learning_rate": 1.5601952879063188e-06,
      "loss": 0.462,
      "step": 306880
    },
    {
      "epoch": 125.62,
      "grad_norm": 1.5669822692871094,
      "learning_rate": 1.5600457875693436e-06,
      "loss": 0.4596,
      "step": 306890
    },
    {
      "epoch": 125.62,
      "grad_norm": 2.061335563659668,
      "learning_rate": 1.5598962908223968e-06,
      "loss": 0.4339,
      "step": 306900
    },
    {
      "epoch": 125.63,
      "grad_norm": 1.6569405794143677,
      "learning_rate": 1.559746797666162e-06,
      "loss": 0.4571,
      "step": 306910
    },
    {
      "epoch": 125.63,
      "grad_norm": 3.4213855266571045,
      "learning_rate": 1.5595973081013242e-06,
      "loss": 0.4773,
      "step": 306920
    },
    {
      "epoch": 125.64,
      "grad_norm": 1.7890037298202515,
      "learning_rate": 1.5594478221285685e-06,
      "loss": 0.4656,
      "step": 306930
    },
    {
      "epoch": 125.64,
      "grad_norm": 2.3517110347747803,
      "learning_rate": 1.5592983397485796e-06,
      "loss": 0.4581,
      "step": 306940
    },
    {
      "epoch": 125.64,
      "grad_norm": 2.016227960586548,
      "learning_rate": 1.5591488609620425e-06,
      "loss": 0.4703,
      "step": 306950
    },
    {
      "epoch": 125.65,
      "grad_norm": 1.977903962135315,
      "learning_rate": 1.5589993857696413e-06,
      "loss": 0.4617,
      "step": 306960
    },
    {
      "epoch": 125.65,
      "grad_norm": 2.1453118324279785,
      "learning_rate": 1.5588499141720614e-06,
      "loss": 0.4552,
      "step": 306970
    },
    {
      "epoch": 125.66,
      "grad_norm": 1.9055684804916382,
      "learning_rate": 1.5587004461699872e-06,
      "loss": 0.444,
      "step": 306980
    },
    {
      "epoch": 125.66,
      "grad_norm": 2.4991862773895264,
      "learning_rate": 1.558550981764104e-06,
      "loss": 0.4714,
      "step": 306990
    },
    {
      "epoch": 125.67,
      "grad_norm": 1.9323614835739136,
      "learning_rate": 1.558401520955095e-06,
      "loss": 0.4633,
      "step": 307000
    },
    {
      "epoch": 125.67,
      "grad_norm": 1.8589081764221191,
      "learning_rate": 1.5582520637436463e-06,
      "loss": 0.4373,
      "step": 307010
    },
    {
      "epoch": 125.67,
      "grad_norm": 2.1619873046875,
      "learning_rate": 1.5581026101304417e-06,
      "loss": 0.4598,
      "step": 307020
    },
    {
      "epoch": 125.68,
      "grad_norm": 2.3327701091766357,
      "learning_rate": 1.5579531601161662e-06,
      "loss": 0.4704,
      "step": 307030
    },
    {
      "epoch": 125.68,
      "grad_norm": 1.9458717107772827,
      "learning_rate": 1.5578037137015046e-06,
      "loss": 0.4538,
      "step": 307040
    },
    {
      "epoch": 125.69,
      "grad_norm": 1.888075351715088,
      "learning_rate": 1.5576542708871409e-06,
      "loss": 0.4858,
      "step": 307050
    },
    {
      "epoch": 125.69,
      "grad_norm": 1.7900192737579346,
      "learning_rate": 1.5575048316737598e-06,
      "loss": 0.4645,
      "step": 307060
    },
    {
      "epoch": 125.69,
      "grad_norm": 1.8740718364715576,
      "learning_rate": 1.557355396062046e-06,
      "loss": 0.4561,
      "step": 307070
    },
    {
      "epoch": 125.7,
      "grad_norm": 2.709268093109131,
      "learning_rate": 1.5572059640526843e-06,
      "loss": 0.474,
      "step": 307080
    },
    {
      "epoch": 125.7,
      "grad_norm": 2.1797149181365967,
      "learning_rate": 1.557056535646359e-06,
      "loss": 0.4566,
      "step": 307090
    },
    {
      "epoch": 125.71,
      "grad_norm": 2.483840227127075,
      "learning_rate": 1.5569071108437548e-06,
      "loss": 0.4702,
      "step": 307100
    },
    {
      "epoch": 125.71,
      "grad_norm": 1.6906880140304565,
      "learning_rate": 1.5567576896455558e-06,
      "loss": 0.4479,
      "step": 307110
    },
    {
      "epoch": 125.71,
      "grad_norm": 1.853206753730774,
      "learning_rate": 1.5566082720524468e-06,
      "loss": 0.4641,
      "step": 307120
    },
    {
      "epoch": 125.72,
      "grad_norm": 1.732997179031372,
      "learning_rate": 1.5564588580651128e-06,
      "loss": 0.4625,
      "step": 307130
    },
    {
      "epoch": 125.72,
      "grad_norm": 1.8799512386322021,
      "learning_rate": 1.556309447684237e-06,
      "loss": 0.4536,
      "step": 307140
    },
    {
      "epoch": 125.73,
      "grad_norm": 1.8576174974441528,
      "learning_rate": 1.5561600409105044e-06,
      "loss": 0.4517,
      "step": 307150
    },
    {
      "epoch": 125.73,
      "grad_norm": 2.153082847595215,
      "learning_rate": 1.5560106377445993e-06,
      "loss": 0.4631,
      "step": 307160
    },
    {
      "epoch": 125.73,
      "grad_norm": 2.0306336879730225,
      "learning_rate": 1.5558612381872064e-06,
      "loss": 0.4615,
      "step": 307170
    },
    {
      "epoch": 125.74,
      "grad_norm": 1.771373987197876,
      "learning_rate": 1.5557118422390098e-06,
      "loss": 0.4692,
      "step": 307180
    },
    {
      "epoch": 125.74,
      "grad_norm": 1.7294468879699707,
      "learning_rate": 1.555562449900694e-06,
      "loss": 0.481,
      "step": 307190
    },
    {
      "epoch": 125.75,
      "grad_norm": 1.6256929636001587,
      "learning_rate": 1.5554130611729432e-06,
      "loss": 0.4499,
      "step": 307200
    },
    {
      "epoch": 125.75,
      "grad_norm": 2.0386314392089844,
      "learning_rate": 1.5552636760564424e-06,
      "loss": 0.4714,
      "step": 307210
    },
    {
      "epoch": 125.76,
      "grad_norm": 1.7434319257736206,
      "learning_rate": 1.555114294551875e-06,
      "loss": 0.463,
      "step": 307220
    },
    {
      "epoch": 125.76,
      "grad_norm": 1.6092877388000488,
      "learning_rate": 1.5549649166599256e-06,
      "loss": 0.4659,
      "step": 307230
    },
    {
      "epoch": 125.76,
      "grad_norm": 2.0322988033294678,
      "learning_rate": 1.5548155423812783e-06,
      "loss": 0.4539,
      "step": 307240
    },
    {
      "epoch": 125.77,
      "grad_norm": 1.9361224174499512,
      "learning_rate": 1.5546661717166177e-06,
      "loss": 0.4653,
      "step": 307250
    },
    {
      "epoch": 125.77,
      "grad_norm": 1.7751339673995972,
      "learning_rate": 1.554516804666627e-06,
      "loss": 0.4619,
      "step": 307260
    },
    {
      "epoch": 125.78,
      "grad_norm": 2.017545700073242,
      "learning_rate": 1.5543674412319926e-06,
      "loss": 0.4672,
      "step": 307270
    },
    {
      "epoch": 125.78,
      "grad_norm": 1.361693263053894,
      "learning_rate": 1.5542180814133971e-06,
      "loss": 0.4759,
      "step": 307280
    },
    {
      "epoch": 125.78,
      "grad_norm": 1.9781326055526733,
      "learning_rate": 1.554068725211525e-06,
      "loss": 0.4492,
      "step": 307290
    },
    {
      "epoch": 125.79,
      "grad_norm": 1.8537639379501343,
      "learning_rate": 1.5539193726270604e-06,
      "loss": 0.4644,
      "step": 307300
    },
    {
      "epoch": 125.79,
      "grad_norm": 2.0141148567199707,
      "learning_rate": 1.5537700236606879e-06,
      "loss": 0.4431,
      "step": 307310
    },
    {
      "epoch": 125.8,
      "grad_norm": 1.7491817474365234,
      "learning_rate": 1.5536206783130908e-06,
      "loss": 0.4605,
      "step": 307320
    },
    {
      "epoch": 125.8,
      "grad_norm": 1.7961400747299194,
      "learning_rate": 1.5534713365849542e-06,
      "loss": 0.459,
      "step": 307330
    },
    {
      "epoch": 125.8,
      "grad_norm": 1.9873456954956055,
      "learning_rate": 1.553321998476962e-06,
      "loss": 0.4626,
      "step": 307340
    },
    {
      "epoch": 125.81,
      "grad_norm": 2.098508596420288,
      "learning_rate": 1.5531726639897979e-06,
      "loss": 0.4322,
      "step": 307350
    },
    {
      "epoch": 125.81,
      "grad_norm": 1.9945411682128906,
      "learning_rate": 1.553023333124146e-06,
      "loss": 0.4461,
      "step": 307360
    },
    {
      "epoch": 125.82,
      "grad_norm": 1.8465783596038818,
      "learning_rate": 1.5528740058806907e-06,
      "loss": 0.4805,
      "step": 307370
    },
    {
      "epoch": 125.82,
      "grad_norm": 2.1527419090270996,
      "learning_rate": 1.5527246822601155e-06,
      "loss": 0.4614,
      "step": 307380
    },
    {
      "epoch": 125.82,
      "grad_norm": 1.8967453241348267,
      "learning_rate": 1.552575362263105e-06,
      "loss": 0.4706,
      "step": 307390
    },
    {
      "epoch": 125.83,
      "grad_norm": 2.278367519378662,
      "learning_rate": 1.5524260458903427e-06,
      "loss": 0.4564,
      "step": 307400
    },
    {
      "epoch": 125.83,
      "grad_norm": 2.065819025039673,
      "learning_rate": 1.5522767331425132e-06,
      "loss": 0.459,
      "step": 307410
    },
    {
      "epoch": 125.84,
      "grad_norm": 1.963172197341919,
      "learning_rate": 1.5521274240202999e-06,
      "loss": 0.4478,
      "step": 307420
    },
    {
      "epoch": 125.84,
      "grad_norm": 1.9498467445373535,
      "learning_rate": 1.5519781185243873e-06,
      "loss": 0.4625,
      "step": 307430
    },
    {
      "epoch": 125.85,
      "grad_norm": 2.0555903911590576,
      "learning_rate": 1.5518288166554583e-06,
      "loss": 0.4793,
      "step": 307440
    },
    {
      "epoch": 125.85,
      "grad_norm": 5.30534553527832,
      "learning_rate": 1.5516795184141984e-06,
      "loss": 0.448,
      "step": 307450
    },
    {
      "epoch": 125.85,
      "grad_norm": 1.9252417087554932,
      "learning_rate": 1.5515302238012901e-06,
      "loss": 0.4145,
      "step": 307460
    },
    {
      "epoch": 125.86,
      "grad_norm": 2.1971793174743652,
      "learning_rate": 1.5513809328174185e-06,
      "loss": 0.4528,
      "step": 307470
    },
    {
      "epoch": 125.86,
      "grad_norm": 2.3665013313293457,
      "learning_rate": 1.5512316454632669e-06,
      "loss": 0.4637,
      "step": 307480
    },
    {
      "epoch": 125.87,
      "grad_norm": 2.2171287536621094,
      "learning_rate": 1.5510823617395192e-06,
      "loss": 0.4588,
      "step": 307490
    },
    {
      "epoch": 125.87,
      "grad_norm": 1.8739594221115112,
      "learning_rate": 1.550933081646859e-06,
      "loss": 0.4643,
      "step": 307500
    },
    {
      "epoch": 125.87,
      "grad_norm": 3.1040709018707275,
      "learning_rate": 1.5507838051859704e-06,
      "loss": 0.4484,
      "step": 307510
    },
    {
      "epoch": 125.88,
      "grad_norm": 2.185370445251465,
      "learning_rate": 1.550634532357537e-06,
      "loss": 0.4626,
      "step": 307520
    },
    {
      "epoch": 125.88,
      "grad_norm": 1.979230523109436,
      "learning_rate": 1.5504852631622426e-06,
      "loss": 0.4652,
      "step": 307530
    },
    {
      "epoch": 125.89,
      "grad_norm": 2.135932683944702,
      "learning_rate": 1.5503359976007713e-06,
      "loss": 0.442,
      "step": 307540
    },
    {
      "epoch": 125.89,
      "grad_norm": 1.9983168840408325,
      "learning_rate": 1.5501867356738065e-06,
      "loss": 0.466,
      "step": 307550
    },
    {
      "epoch": 125.89,
      "grad_norm": 1.551722764968872,
      "learning_rate": 1.5500374773820323e-06,
      "loss": 0.4508,
      "step": 307560
    },
    {
      "epoch": 125.9,
      "grad_norm": 2.294163465499878,
      "learning_rate": 1.5498882227261326e-06,
      "loss": 0.4634,
      "step": 307570
    },
    {
      "epoch": 125.9,
      "grad_norm": 1.7920218706130981,
      "learning_rate": 1.5497389717067903e-06,
      "loss": 0.4702,
      "step": 307580
    },
    {
      "epoch": 125.91,
      "grad_norm": 1.6789807081222534,
      "learning_rate": 1.5495897243246895e-06,
      "loss": 0.4361,
      "step": 307590
    },
    {
      "epoch": 125.91,
      "grad_norm": 1.6178607940673828,
      "learning_rate": 1.549440480580514e-06,
      "loss": 0.4789,
      "step": 307600
    },
    {
      "epoch": 125.91,
      "grad_norm": 2.0623421669006348,
      "learning_rate": 1.5492912404749474e-06,
      "loss": 0.4668,
      "step": 307610
    },
    {
      "epoch": 125.92,
      "grad_norm": 2.019280433654785,
      "learning_rate": 1.5491420040086734e-06,
      "loss": 0.4642,
      "step": 307620
    },
    {
      "epoch": 125.92,
      "grad_norm": 2.420708417892456,
      "learning_rate": 1.548992771182375e-06,
      "loss": 0.4677,
      "step": 307630
    },
    {
      "epoch": 125.93,
      "grad_norm": 2.0707132816314697,
      "learning_rate": 1.548843541996737e-06,
      "loss": 0.4686,
      "step": 307640
    },
    {
      "epoch": 125.93,
      "grad_norm": 2.0664024353027344,
      "learning_rate": 1.5486943164524422e-06,
      "loss": 0.4616,
      "step": 307650
    },
    {
      "epoch": 125.94,
      "grad_norm": 2.1777408123016357,
      "learning_rate": 1.5485450945501743e-06,
      "loss": 0.4434,
      "step": 307660
    },
    {
      "epoch": 125.94,
      "grad_norm": 1.9204117059707642,
      "learning_rate": 1.5483958762906172e-06,
      "loss": 0.4652,
      "step": 307670
    },
    {
      "epoch": 125.94,
      "grad_norm": 1.8501352071762085,
      "learning_rate": 1.5482466616744538e-06,
      "loss": 0.4499,
      "step": 307680
    },
    {
      "epoch": 125.95,
      "grad_norm": 1.8366734981536865,
      "learning_rate": 1.5480974507023683e-06,
      "loss": 0.4585,
      "step": 307690
    },
    {
      "epoch": 125.95,
      "grad_norm": 1.62773859500885,
      "learning_rate": 1.5479482433750436e-06,
      "loss": 0.4765,
      "step": 307700
    },
    {
      "epoch": 125.96,
      "grad_norm": 2.207272529602051,
      "learning_rate": 1.5477990396931644e-06,
      "loss": 0.4498,
      "step": 307710
    },
    {
      "epoch": 125.96,
      "grad_norm": 1.8985671997070312,
      "learning_rate": 1.5476498396574126e-06,
      "loss": 0.4499,
      "step": 307720
    },
    {
      "epoch": 125.96,
      "grad_norm": 1.7774204015731812,
      "learning_rate": 1.5475006432684725e-06,
      "loss": 0.4699,
      "step": 307730
    },
    {
      "epoch": 125.97,
      "grad_norm": 2.1413698196411133,
      "learning_rate": 1.5473514505270271e-06,
      "loss": 0.4607,
      "step": 307740
    },
    {
      "epoch": 125.97,
      "grad_norm": 2.152015209197998,
      "learning_rate": 1.54720226143376e-06,
      "loss": 0.4272,
      "step": 307750
    },
    {
      "epoch": 125.98,
      "grad_norm": 1.7878726720809937,
      "learning_rate": 1.5470530759893551e-06,
      "loss": 0.4773,
      "step": 307760
    },
    {
      "epoch": 125.98,
      "grad_norm": 2.5078439712524414,
      "learning_rate": 1.5469038941944954e-06,
      "loss": 0.4449,
      "step": 307770
    },
    {
      "epoch": 125.98,
      "grad_norm": 1.956695318222046,
      "learning_rate": 1.546754716049864e-06,
      "loss": 0.4661,
      "step": 307780
    },
    {
      "epoch": 125.99,
      "grad_norm": 1.8627713918685913,
      "learning_rate": 1.546605541556145e-06,
      "loss": 0.4641,
      "step": 307790
    },
    {
      "epoch": 125.99,
      "grad_norm": 1.8258230686187744,
      "learning_rate": 1.5464563707140215e-06,
      "loss": 0.4592,
      "step": 307800
    },
    {
      "epoch": 126.0,
      "grad_norm": 1.7448952198028564,
      "learning_rate": 1.546307203524176e-06,
      "loss": 0.444,
      "step": 307810
    },
    {
      "epoch": 126.0,
      "eval_loss": 0.4575129449367523,
      "eval_runtime": 56.8339,
      "eval_samples_per_second": 60.686,
      "eval_steps_per_second": 7.601,
      "step": 307818
    },
    {
      "epoch": 126.0,
      "grad_norm": 2.065725564956665,
      "learning_rate": 1.546158039987292e-06,
      "loss": 0.4767,
      "step": 307820
    },
    {
      "epoch": 126.0,
      "grad_norm": 2.4823217391967773,
      "learning_rate": 1.5460088801040539e-06,
      "loss": 0.4701,
      "step": 307830
    },
    {
      "epoch": 126.01,
      "grad_norm": 2.1532797813415527,
      "learning_rate": 1.5458597238751443e-06,
      "loss": 0.4619,
      "step": 307840
    },
    {
      "epoch": 126.01,
      "grad_norm": 1.704652190208435,
      "learning_rate": 1.5457105713012468e-06,
      "loss": 0.4458,
      "step": 307850
    },
    {
      "epoch": 126.02,
      "grad_norm": 1.623517394065857,
      "learning_rate": 1.5455614223830439e-06,
      "loss": 0.4695,
      "step": 307860
    },
    {
      "epoch": 126.02,
      "grad_norm": 2.6232783794403076,
      "learning_rate": 1.5454122771212195e-06,
      "loss": 0.4623,
      "step": 307870
    },
    {
      "epoch": 126.03,
      "grad_norm": 1.9542557001113892,
      "learning_rate": 1.5452631355164566e-06,
      "loss": 0.4606,
      "step": 307880
    },
    {
      "epoch": 126.03,
      "grad_norm": 2.3121092319488525,
      "learning_rate": 1.5451139975694383e-06,
      "loss": 0.4385,
      "step": 307890
    },
    {
      "epoch": 126.03,
      "grad_norm": 2.610698699951172,
      "learning_rate": 1.5449648632808475e-06,
      "loss": 0.454,
      "step": 307900
    },
    {
      "epoch": 126.04,
      "grad_norm": 1.8103431463241577,
      "learning_rate": 1.5448157326513682e-06,
      "loss": 0.4563,
      "step": 307910
    },
    {
      "epoch": 126.04,
      "grad_norm": 1.9319329261779785,
      "learning_rate": 1.5446666056816827e-06,
      "loss": 0.4638,
      "step": 307920
    },
    {
      "epoch": 126.05,
      "grad_norm": 2.567723035812378,
      "learning_rate": 1.5445174823724752e-06,
      "loss": 0.459,
      "step": 307930
    },
    {
      "epoch": 126.05,
      "grad_norm": 2.103774309158325,
      "learning_rate": 1.5443683627244277e-06,
      "loss": 0.4517,
      "step": 307940
    },
    {
      "epoch": 126.05,
      "grad_norm": 1.8149627447128296,
      "learning_rate": 1.5442192467382232e-06,
      "loss": 0.4492,
      "step": 307950
    },
    {
      "epoch": 126.06,
      "grad_norm": 1.944224238395691,
      "learning_rate": 1.5440701344145456e-06,
      "loss": 0.4516,
      "step": 307960
    },
    {
      "epoch": 126.06,
      "grad_norm": 1.9501391649246216,
      "learning_rate": 1.5439210257540775e-06,
      "loss": 0.4588,
      "step": 307970
    },
    {
      "epoch": 126.07,
      "grad_norm": 2.1457228660583496,
      "learning_rate": 1.543771920757502e-06,
      "loss": 0.4733,
      "step": 307980
    },
    {
      "epoch": 126.07,
      "grad_norm": 1.8797575235366821,
      "learning_rate": 1.5436228194255024e-06,
      "loss": 0.4609,
      "step": 307990
    },
    {
      "epoch": 126.07,
      "grad_norm": 2.1834309101104736,
      "learning_rate": 1.5434737217587615e-06,
      "loss": 0.4627,
      "step": 308000
    },
    {
      "epoch": 126.08,
      "grad_norm": 2.2234206199645996,
      "learning_rate": 1.5433246277579617e-06,
      "loss": 0.4576,
      "step": 308010
    },
    {
      "epoch": 126.08,
      "grad_norm": 2.293097734451294,
      "learning_rate": 1.543175537423787e-06,
      "loss": 0.4505,
      "step": 308020
    },
    {
      "epoch": 126.09,
      "grad_norm": 2.119356155395508,
      "learning_rate": 1.54302645075692e-06,
      "loss": 0.4349,
      "step": 308030
    },
    {
      "epoch": 126.09,
      "grad_norm": 2.2322211265563965,
      "learning_rate": 1.5428773677580435e-06,
      "loss": 0.4764,
      "step": 308040
    },
    {
      "epoch": 126.09,
      "grad_norm": 1.9759788513183594,
      "learning_rate": 1.5427282884278406e-06,
      "loss": 0.4508,
      "step": 308050
    },
    {
      "epoch": 126.1,
      "grad_norm": 2.2654194831848145,
      "learning_rate": 1.542579212766994e-06,
      "loss": 0.4435,
      "step": 308060
    },
    {
      "epoch": 126.1,
      "grad_norm": 2.2297511100769043,
      "learning_rate": 1.542430140776187e-06,
      "loss": 0.43,
      "step": 308070
    },
    {
      "epoch": 126.11,
      "grad_norm": 2.377063751220703,
      "learning_rate": 1.5422810724561022e-06,
      "loss": 0.451,
      "step": 308080
    },
    {
      "epoch": 126.11,
      "grad_norm": 1.4004422426223755,
      "learning_rate": 1.542132007807422e-06,
      "loss": 0.461,
      "step": 308090
    },
    {
      "epoch": 126.12,
      "grad_norm": 1.9908264875411987,
      "learning_rate": 1.54198294683083e-06,
      "loss": 0.4397,
      "step": 308100
    },
    {
      "epoch": 126.12,
      "grad_norm": 2.171591281890869,
      "learning_rate": 1.5418338895270089e-06,
      "loss": 0.4549,
      "step": 308110
    },
    {
      "epoch": 126.12,
      "grad_norm": 1.8740177154541016,
      "learning_rate": 1.5416848358966408e-06,
      "loss": 0.4672,
      "step": 308120
    },
    {
      "epoch": 126.13,
      "grad_norm": 2.384247303009033,
      "learning_rate": 1.541535785940409e-06,
      "loss": 0.4665,
      "step": 308130
    },
    {
      "epoch": 126.13,
      "grad_norm": 2.3912296295166016,
      "learning_rate": 1.5413867396589965e-06,
      "loss": 0.4411,
      "step": 308140
    },
    {
      "epoch": 126.14,
      "grad_norm": 1.8121620416641235,
      "learning_rate": 1.541237697053086e-06,
      "loss": 0.4642,
      "step": 308150
    },
    {
      "epoch": 126.14,
      "grad_norm": 2.128312349319458,
      "learning_rate": 1.5410886581233603e-06,
      "loss": 0.4448,
      "step": 308160
    },
    {
      "epoch": 126.14,
      "grad_norm": 1.9543207883834839,
      "learning_rate": 1.5409396228705011e-06,
      "loss": 0.4475,
      "step": 308170
    },
    {
      "epoch": 126.15,
      "grad_norm": 1.9514132738113403,
      "learning_rate": 1.5407905912951925e-06,
      "loss": 0.4489,
      "step": 308180
    },
    {
      "epoch": 126.15,
      "grad_norm": 1.8364149332046509,
      "learning_rate": 1.540641563398116e-06,
      "loss": 0.4184,
      "step": 308190
    },
    {
      "epoch": 126.16,
      "grad_norm": 1.9937784671783447,
      "learning_rate": 1.540492539179955e-06,
      "loss": 0.4694,
      "step": 308200
    },
    {
      "epoch": 126.16,
      "grad_norm": 2.1254637241363525,
      "learning_rate": 1.5403435186413924e-06,
      "loss": 0.4683,
      "step": 308210
    },
    {
      "epoch": 126.16,
      "grad_norm": 1.9107131958007812,
      "learning_rate": 1.5401945017831103e-06,
      "loss": 0.449,
      "step": 308220
    },
    {
      "epoch": 126.17,
      "grad_norm": 1.802452802658081,
      "learning_rate": 1.5400454886057913e-06,
      "loss": 0.4497,
      "step": 308230
    },
    {
      "epoch": 126.17,
      "grad_norm": 2.1791515350341797,
      "learning_rate": 1.5398964791101184e-06,
      "loss": 0.4652,
      "step": 308240
    },
    {
      "epoch": 126.18,
      "grad_norm": 2.7086470127105713,
      "learning_rate": 1.539747473296774e-06,
      "loss": 0.451,
      "step": 308250
    },
    {
      "epoch": 126.18,
      "grad_norm": 2.8328981399536133,
      "learning_rate": 1.5395984711664408e-06,
      "loss": 0.4531,
      "step": 308260
    },
    {
      "epoch": 126.19,
      "grad_norm": 1.7996538877487183,
      "learning_rate": 1.539449472719801e-06,
      "loss": 0.4646,
      "step": 308270
    },
    {
      "epoch": 126.19,
      "grad_norm": 1.705946683883667,
      "learning_rate": 1.5393004779575375e-06,
      "loss": 0.4611,
      "step": 308280
    },
    {
      "epoch": 126.19,
      "grad_norm": 1.8068363666534424,
      "learning_rate": 1.539151486880333e-06,
      "loss": 0.4541,
      "step": 308290
    },
    {
      "epoch": 126.2,
      "grad_norm": 2.0942108631134033,
      "learning_rate": 1.5390024994888692e-06,
      "loss": 0.4556,
      "step": 308300
    },
    {
      "epoch": 126.2,
      "grad_norm": 2.7169830799102783,
      "learning_rate": 1.538853515783829e-06,
      "loss": 0.4347,
      "step": 308310
    },
    {
      "epoch": 126.21,
      "grad_norm": 2.2539587020874023,
      "learning_rate": 1.5387045357658953e-06,
      "loss": 0.4505,
      "step": 308320
    },
    {
      "epoch": 126.21,
      "grad_norm": 2.1543338298797607,
      "learning_rate": 1.53855555943575e-06,
      "loss": 0.4399,
      "step": 308330
    },
    {
      "epoch": 126.21,
      "grad_norm": 2.5586349964141846,
      "learning_rate": 1.538406586794076e-06,
      "loss": 0.4432,
      "step": 308340
    },
    {
      "epoch": 126.22,
      "grad_norm": 1.8755675554275513,
      "learning_rate": 1.5382576178415548e-06,
      "loss": 0.4408,
      "step": 308350
    },
    {
      "epoch": 126.22,
      "grad_norm": 2.0787107944488525,
      "learning_rate": 1.5381086525788697e-06,
      "loss": 0.4712,
      "step": 308360
    },
    {
      "epoch": 126.23,
      "grad_norm": 2.206843852996826,
      "learning_rate": 1.5379596910067031e-06,
      "loss": 0.4705,
      "step": 308370
    },
    {
      "epoch": 126.23,
      "grad_norm": 1.6485276222229004,
      "learning_rate": 1.5378107331257376e-06,
      "loss": 0.4628,
      "step": 308380
    },
    {
      "epoch": 126.23,
      "grad_norm": 2.1833760738372803,
      "learning_rate": 1.5376617789366534e-06,
      "loss": 0.4594,
      "step": 308390
    },
    {
      "epoch": 126.24,
      "grad_norm": 1.7892166376113892,
      "learning_rate": 1.5375128284401353e-06,
      "loss": 0.4707,
      "step": 308400
    },
    {
      "epoch": 126.24,
      "grad_norm": 2.0011963844299316,
      "learning_rate": 1.5373638816368653e-06,
      "loss": 0.4665,
      "step": 308410
    },
    {
      "epoch": 126.25,
      "grad_norm": 1.689646601676941,
      "learning_rate": 1.537214938527525e-06,
      "loss": 0.4602,
      "step": 308420
    },
    {
      "epoch": 126.25,
      "grad_norm": 1.9067039489746094,
      "learning_rate": 1.5370659991127973e-06,
      "loss": 0.4653,
      "step": 308430
    },
    {
      "epoch": 126.25,
      "grad_norm": 2.2673587799072266,
      "learning_rate": 1.5369170633933638e-06,
      "loss": 0.4544,
      "step": 308440
    },
    {
      "epoch": 126.26,
      "grad_norm": 1.9894486665725708,
      "learning_rate": 1.536768131369907e-06,
      "loss": 0.456,
      "step": 308450
    },
    {
      "epoch": 126.26,
      "grad_norm": 2.0650782585144043,
      "learning_rate": 1.536619203043109e-06,
      "loss": 0.4874,
      "step": 308460
    },
    {
      "epoch": 126.27,
      "grad_norm": 1.951249361038208,
      "learning_rate": 1.5364702784136521e-06,
      "loss": 0.4665,
      "step": 308470
    },
    {
      "epoch": 126.27,
      "grad_norm": 2.0805091857910156,
      "learning_rate": 1.5363213574822187e-06,
      "loss": 0.4648,
      "step": 308480
    },
    {
      "epoch": 126.28,
      "grad_norm": 2.2964258193969727,
      "learning_rate": 1.536172440249491e-06,
      "loss": 0.4546,
      "step": 308490
    },
    {
      "epoch": 126.28,
      "grad_norm": 2.0729804039001465,
      "learning_rate": 1.536023526716151e-06,
      "loss": 0.4624,
      "step": 308500
    },
    {
      "epoch": 126.28,
      "grad_norm": 2.060258626937866,
      "learning_rate": 1.535874616882881e-06,
      "loss": 0.456,
      "step": 308510
    },
    {
      "epoch": 126.29,
      "grad_norm": 1.9485996961593628,
      "learning_rate": 1.535725710750363e-06,
      "loss": 0.4434,
      "step": 308520
    },
    {
      "epoch": 126.29,
      "grad_norm": 2.509765625,
      "learning_rate": 1.5355768083192792e-06,
      "loss": 0.4652,
      "step": 308530
    },
    {
      "epoch": 126.3,
      "grad_norm": 2.265824317932129,
      "learning_rate": 1.5354279095903115e-06,
      "loss": 0.4571,
      "step": 308540
    },
    {
      "epoch": 126.3,
      "grad_norm": 1.821580410003662,
      "learning_rate": 1.535279014564142e-06,
      "loss": 0.455,
      "step": 308550
    },
    {
      "epoch": 126.3,
      "grad_norm": 2.0894596576690674,
      "learning_rate": 1.5351301232414527e-06,
      "loss": 0.4589,
      "step": 308560
    },
    {
      "epoch": 126.31,
      "grad_norm": 1.7742220163345337,
      "learning_rate": 1.534981235622926e-06,
      "loss": 0.4507,
      "step": 308570
    },
    {
      "epoch": 126.31,
      "grad_norm": 2.249115467071533,
      "learning_rate": 1.5348323517092434e-06,
      "loss": 0.4553,
      "step": 308580
    },
    {
      "epoch": 126.32,
      "grad_norm": 2.2178635597229004,
      "learning_rate": 1.5346834715010873e-06,
      "loss": 0.4504,
      "step": 308590
    },
    {
      "epoch": 126.32,
      "grad_norm": 2.107635974884033,
      "learning_rate": 1.5345345949991402e-06,
      "loss": 0.469,
      "step": 308600
    },
    {
      "epoch": 126.32,
      "grad_norm": 1.8581538200378418,
      "learning_rate": 1.5343857222040834e-06,
      "loss": 0.4711,
      "step": 308610
    },
    {
      "epoch": 126.33,
      "grad_norm": 1.6377063989639282,
      "learning_rate": 1.534236853116599e-06,
      "loss": 0.4479,
      "step": 308620
    },
    {
      "epoch": 126.33,
      "grad_norm": 2.017911672592163,
      "learning_rate": 1.5340879877373686e-06,
      "loss": 0.4653,
      "step": 308630
    },
    {
      "epoch": 126.34,
      "grad_norm": 2.739809513092041,
      "learning_rate": 1.533939126067075e-06,
      "loss": 0.4623,
      "step": 308640
    },
    {
      "epoch": 126.34,
      "grad_norm": 2.1478047370910645,
      "learning_rate": 1.5337902681064e-06,
      "loss": 0.4586,
      "step": 308650
    },
    {
      "epoch": 126.34,
      "grad_norm": 1.8039778470993042,
      "learning_rate": 1.5336414138560244e-06,
      "loss": 0.4494,
      "step": 308660
    },
    {
      "epoch": 126.35,
      "grad_norm": 2.1358253955841064,
      "learning_rate": 1.533492563316631e-06,
      "loss": 0.4489,
      "step": 308670
    },
    {
      "epoch": 126.35,
      "grad_norm": 2.083371877670288,
      "learning_rate": 1.5333437164889008e-06,
      "loss": 0.4512,
      "step": 308680
    },
    {
      "epoch": 126.36,
      "grad_norm": 2.3686389923095703,
      "learning_rate": 1.533194873373517e-06,
      "loss": 0.4578,
      "step": 308690
    },
    {
      "epoch": 126.36,
      "grad_norm": 2.001739501953125,
      "learning_rate": 1.5330460339711603e-06,
      "loss": 0.4551,
      "step": 308700
    },
    {
      "epoch": 126.37,
      "grad_norm": 1.9994169473648071,
      "learning_rate": 1.5328971982825132e-06,
      "loss": 0.4873,
      "step": 308710
    },
    {
      "epoch": 126.37,
      "grad_norm": 1.9229414463043213,
      "learning_rate": 1.5327483663082572e-06,
      "loss": 0.4616,
      "step": 308720
    },
    {
      "epoch": 126.37,
      "grad_norm": 2.2115066051483154,
      "learning_rate": 1.5325995380490743e-06,
      "loss": 0.4269,
      "step": 308730
    },
    {
      "epoch": 126.38,
      "grad_norm": 1.6511051654815674,
      "learning_rate": 1.532450713505646e-06,
      "loss": 0.4559,
      "step": 308740
    },
    {
      "epoch": 126.38,
      "grad_norm": 1.8881700038909912,
      "learning_rate": 1.5323018926786538e-06,
      "loss": 0.4481,
      "step": 308750
    },
    {
      "epoch": 126.39,
      "grad_norm": 1.9157466888427734,
      "learning_rate": 1.5321530755687795e-06,
      "loss": 0.4349,
      "step": 308760
    },
    {
      "epoch": 126.39,
      "grad_norm": 1.9739896059036255,
      "learning_rate": 1.5320042621767043e-06,
      "loss": 0.4578,
      "step": 308770
    },
    {
      "epoch": 126.39,
      "grad_norm": 1.968955397605896,
      "learning_rate": 1.5318554525031116e-06,
      "loss": 0.4689,
      "step": 308780
    },
    {
      "epoch": 126.4,
      "grad_norm": 2.2524633407592773,
      "learning_rate": 1.5317066465486826e-06,
      "loss": 0.4529,
      "step": 308790
    },
    {
      "epoch": 126.4,
      "grad_norm": 1.8001282215118408,
      "learning_rate": 1.5315578443140979e-06,
      "loss": 0.4755,
      "step": 308800
    },
    {
      "epoch": 126.41,
      "grad_norm": 2.3752079010009766,
      "learning_rate": 1.5314090458000397e-06,
      "loss": 0.4467,
      "step": 308810
    },
    {
      "epoch": 126.41,
      "grad_norm": 2.609524965286255,
      "learning_rate": 1.5312602510071896e-06,
      "loss": 0.4657,
      "step": 308820
    },
    {
      "epoch": 126.41,
      "grad_norm": 2.1197104454040527,
      "learning_rate": 1.5311114599362291e-06,
      "loss": 0.4658,
      "step": 308830
    },
    {
      "epoch": 126.42,
      "grad_norm": 2.351501226425171,
      "learning_rate": 1.5309626725878402e-06,
      "loss": 0.4744,
      "step": 308840
    },
    {
      "epoch": 126.42,
      "grad_norm": 2.076754331588745,
      "learning_rate": 1.5308138889627038e-06,
      "loss": 0.451,
      "step": 308850
    },
    {
      "epoch": 126.43,
      "grad_norm": 2.088357925415039,
      "learning_rate": 1.530665109061502e-06,
      "loss": 0.459,
      "step": 308860
    },
    {
      "epoch": 126.43,
      "grad_norm": 1.7737997770309448,
      "learning_rate": 1.5305163328849164e-06,
      "loss": 0.4608,
      "step": 308870
    },
    {
      "epoch": 126.43,
      "grad_norm": 2.0101702213287354,
      "learning_rate": 1.530367560433628e-06,
      "loss": 0.4533,
      "step": 308880
    },
    {
      "epoch": 126.44,
      "grad_norm": 2.0633177757263184,
      "learning_rate": 1.5302187917083189e-06,
      "loss": 0.4542,
      "step": 308890
    },
    {
      "epoch": 126.44,
      "grad_norm": 1.6702673435211182,
      "learning_rate": 1.53007002670967e-06,
      "loss": 0.4553,
      "step": 308900
    },
    {
      "epoch": 126.45,
      "grad_norm": 1.747534155845642,
      "learning_rate": 1.529921265438363e-06,
      "loss": 0.4617,
      "step": 308910
    },
    {
      "epoch": 126.45,
      "grad_norm": 1.7499381303787231,
      "learning_rate": 1.5297725078950795e-06,
      "loss": 0.4764,
      "step": 308920
    },
    {
      "epoch": 126.46,
      "grad_norm": 1.8820010423660278,
      "learning_rate": 1.5296237540805004e-06,
      "loss": 0.4558,
      "step": 308930
    },
    {
      "epoch": 126.46,
      "grad_norm": 2.1526994705200195,
      "learning_rate": 1.5294750039953082e-06,
      "loss": 0.4693,
      "step": 308940
    },
    {
      "epoch": 126.46,
      "grad_norm": 2.3536088466644287,
      "learning_rate": 1.5293262576401826e-06,
      "loss": 0.4501,
      "step": 308950
    },
    {
      "epoch": 126.47,
      "grad_norm": 1.6984822750091553,
      "learning_rate": 1.5291775150158067e-06,
      "loss": 0.4497,
      "step": 308960
    },
    {
      "epoch": 126.47,
      "grad_norm": 2.137565851211548,
      "learning_rate": 1.5290287761228616e-06,
      "loss": 0.4582,
      "step": 308970
    },
    {
      "epoch": 126.48,
      "grad_norm": 1.6145899295806885,
      "learning_rate": 1.528880040962028e-06,
      "loss": 0.4599,
      "step": 308980
    },
    {
      "epoch": 126.48,
      "grad_norm": 2.05802845954895,
      "learning_rate": 1.5287313095339872e-06,
      "loss": 0.4338,
      "step": 308990
    },
    {
      "epoch": 126.48,
      "grad_norm": 1.9691026210784912,
      "learning_rate": 1.528582581839421e-06,
      "loss": 0.4565,
      "step": 309000
    },
    {
      "epoch": 126.49,
      "grad_norm": 2.272883176803589,
      "learning_rate": 1.528433857879011e-06,
      "loss": 0.4439,
      "step": 309010
    },
    {
      "epoch": 126.49,
      "grad_norm": 1.8428597450256348,
      "learning_rate": 1.5282851376534377e-06,
      "loss": 0.4729,
      "step": 309020
    },
    {
      "epoch": 126.5,
      "grad_norm": 1.7719659805297852,
      "learning_rate": 1.5281364211633824e-06,
      "loss": 0.4483,
      "step": 309030
    },
    {
      "epoch": 126.5,
      "grad_norm": 1.8393590450286865,
      "learning_rate": 1.5279877084095266e-06,
      "loss": 0.4485,
      "step": 309040
    },
    {
      "epoch": 126.5,
      "grad_norm": 1.8979295492172241,
      "learning_rate": 1.5278389993925516e-06,
      "loss": 0.4658,
      "step": 309050
    },
    {
      "epoch": 126.51,
      "grad_norm": 2.1080877780914307,
      "learning_rate": 1.5276902941131383e-06,
      "loss": 0.4834,
      "step": 309060
    },
    {
      "epoch": 126.51,
      "grad_norm": 2.0720231533050537,
      "learning_rate": 1.5275415925719686e-06,
      "loss": 0.4665,
      "step": 309070
    },
    {
      "epoch": 126.52,
      "grad_norm": 1.9894770383834839,
      "learning_rate": 1.5273928947697228e-06,
      "loss": 0.4637,
      "step": 309080
    },
    {
      "epoch": 126.52,
      "grad_norm": 1.7171138525009155,
      "learning_rate": 1.5272442007070824e-06,
      "loss": 0.446,
      "step": 309090
    },
    {
      "epoch": 126.52,
      "grad_norm": 2.3921666145324707,
      "learning_rate": 1.5270955103847293e-06,
      "loss": 0.4621,
      "step": 309100
    },
    {
      "epoch": 126.53,
      "grad_norm": 1.7096326351165771,
      "learning_rate": 1.5269468238033437e-06,
      "loss": 0.4566,
      "step": 309110
    },
    {
      "epoch": 126.53,
      "grad_norm": 2.0302233695983887,
      "learning_rate": 1.5267981409636067e-06,
      "loss": 0.4262,
      "step": 309120
    },
    {
      "epoch": 126.54,
      "grad_norm": 2.292846918106079,
      "learning_rate": 1.5266494618661995e-06,
      "loss": 0.4585,
      "step": 309130
    },
    {
      "epoch": 126.54,
      "grad_norm": 1.9585156440734863,
      "learning_rate": 1.5265007865118025e-06,
      "loss": 0.4444,
      "step": 309140
    },
    {
      "epoch": 126.55,
      "grad_norm": 1.6630277633666992,
      "learning_rate": 1.5263521149010991e-06,
      "loss": 0.4789,
      "step": 309150
    },
    {
      "epoch": 126.55,
      "grad_norm": 1.7178491353988647,
      "learning_rate": 1.5262034470347682e-06,
      "loss": 0.4598,
      "step": 309160
    },
    {
      "epoch": 126.55,
      "grad_norm": 1.9217976331710815,
      "learning_rate": 1.5260547829134915e-06,
      "loss": 0.4613,
      "step": 309170
    },
    {
      "epoch": 126.56,
      "grad_norm": 1.90190589427948,
      "learning_rate": 1.5259061225379498e-06,
      "loss": 0.4541,
      "step": 309180
    },
    {
      "epoch": 126.56,
      "grad_norm": 1.9333250522613525,
      "learning_rate": 1.5257574659088247e-06,
      "loss": 0.4761,
      "step": 309190
    },
    {
      "epoch": 126.57,
      "grad_norm": 2.516875982284546,
      "learning_rate": 1.5256088130267965e-06,
      "loss": 0.4647,
      "step": 309200
    },
    {
      "epoch": 126.57,
      "grad_norm": 2.322430372238159,
      "learning_rate": 1.5254601638925462e-06,
      "loss": 0.4789,
      "step": 309210
    },
    {
      "epoch": 126.57,
      "grad_norm": 2.0658745765686035,
      "learning_rate": 1.525311518506755e-06,
      "loss": 0.4522,
      "step": 309220
    },
    {
      "epoch": 126.58,
      "grad_norm": 1.9159983396530151,
      "learning_rate": 1.5251628768701043e-06,
      "loss": 0.4807,
      "step": 309230
    },
    {
      "epoch": 126.58,
      "grad_norm": 1.9580730199813843,
      "learning_rate": 1.525014238983274e-06,
      "loss": 0.4842,
      "step": 309240
    },
    {
      "epoch": 126.59,
      "grad_norm": 1.7708145380020142,
      "learning_rate": 1.5248656048469452e-06,
      "loss": 0.4832,
      "step": 309250
    },
    {
      "epoch": 126.59,
      "grad_norm": 2.8531527519226074,
      "learning_rate": 1.5247169744617991e-06,
      "loss": 0.4623,
      "step": 309260
    },
    {
      "epoch": 126.59,
      "grad_norm": 2.0543506145477295,
      "learning_rate": 1.5245683478285164e-06,
      "loss": 0.4455,
      "step": 309270
    },
    {
      "epoch": 126.6,
      "grad_norm": 2.3772811889648438,
      "learning_rate": 1.5244197249477781e-06,
      "loss": 0.4719,
      "step": 309280
    },
    {
      "epoch": 126.6,
      "grad_norm": 2.181413173675537,
      "learning_rate": 1.5242711058202647e-06,
      "loss": 0.4609,
      "step": 309290
    },
    {
      "epoch": 126.61,
      "grad_norm": 2.2343740463256836,
      "learning_rate": 1.5241224904466576e-06,
      "loss": 0.4443,
      "step": 309300
    },
    {
      "epoch": 126.61,
      "grad_norm": 2.1920206546783447,
      "learning_rate": 1.5239738788276368e-06,
      "loss": 0.4744,
      "step": 309310
    },
    {
      "epoch": 126.61,
      "grad_norm": 1.7934130430221558,
      "learning_rate": 1.523825270963884e-06,
      "loss": 0.4662,
      "step": 309320
    },
    {
      "epoch": 126.62,
      "grad_norm": 1.7153657674789429,
      "learning_rate": 1.5236766668560782e-06,
      "loss": 0.4531,
      "step": 309330
    },
    {
      "epoch": 126.62,
      "grad_norm": 1.5110597610473633,
      "learning_rate": 1.523528066504902e-06,
      "loss": 0.4635,
      "step": 309340
    },
    {
      "epoch": 126.63,
      "grad_norm": 2.043644905090332,
      "learning_rate": 1.5233794699110357e-06,
      "loss": 0.4746,
      "step": 309350
    },
    {
      "epoch": 126.63,
      "grad_norm": 1.7898998260498047,
      "learning_rate": 1.5232308770751595e-06,
      "loss": 0.4525,
      "step": 309360
    },
    {
      "epoch": 126.64,
      "grad_norm": 2.260913372039795,
      "learning_rate": 1.5230822879979548e-06,
      "loss": 0.4745,
      "step": 309370
    },
    {
      "epoch": 126.64,
      "grad_norm": 1.9064509868621826,
      "learning_rate": 1.5229337026801013e-06,
      "loss": 0.4508,
      "step": 309380
    },
    {
      "epoch": 126.64,
      "grad_norm": 1.6056886911392212,
      "learning_rate": 1.5227851211222806e-06,
      "loss": 0.4455,
      "step": 309390
    },
    {
      "epoch": 126.65,
      "grad_norm": 1.8682782649993896,
      "learning_rate": 1.5226365433251723e-06,
      "loss": 0.4462,
      "step": 309400
    },
    {
      "epoch": 126.65,
      "grad_norm": 1.971606731414795,
      "learning_rate": 1.522487969289458e-06,
      "loss": 0.454,
      "step": 309410
    },
    {
      "epoch": 126.66,
      "grad_norm": 1.7689363956451416,
      "learning_rate": 1.5223393990158178e-06,
      "loss": 0.4635,
      "step": 309420
    },
    {
      "epoch": 126.66,
      "grad_norm": 1.912955641746521,
      "learning_rate": 1.5221908325049322e-06,
      "loss": 0.4442,
      "step": 309430
    },
    {
      "epoch": 126.66,
      "grad_norm": 1.8288646936416626,
      "learning_rate": 1.5220422697574818e-06,
      "loss": 0.4594,
      "step": 309440
    },
    {
      "epoch": 126.67,
      "grad_norm": 2.413064479827881,
      "learning_rate": 1.5218937107741475e-06,
      "loss": 0.458,
      "step": 309450
    },
    {
      "epoch": 126.67,
      "grad_norm": 13.964244842529297,
      "learning_rate": 1.5217451555556102e-06,
      "loss": 0.4568,
      "step": 309460
    },
    {
      "epoch": 126.68,
      "grad_norm": 2.184250593185425,
      "learning_rate": 1.521596604102549e-06,
      "loss": 0.4769,
      "step": 309470
    },
    {
      "epoch": 126.68,
      "grad_norm": 1.9313610792160034,
      "learning_rate": 1.521448056415645e-06,
      "loss": 0.4385,
      "step": 309480
    },
    {
      "epoch": 126.68,
      "grad_norm": 1.8643336296081543,
      "learning_rate": 1.5212995124955792e-06,
      "loss": 0.4319,
      "step": 309490
    },
    {
      "epoch": 126.69,
      "grad_norm": 16.318796157836914,
      "learning_rate": 1.5211509723430316e-06,
      "loss": 0.4776,
      "step": 309500
    },
    {
      "epoch": 126.69,
      "grad_norm": 2.026301622390747,
      "learning_rate": 1.521002435958683e-06,
      "loss": 0.4692,
      "step": 309510
    },
    {
      "epoch": 126.7,
      "grad_norm": 2.0232911109924316,
      "learning_rate": 1.5208539033432125e-06,
      "loss": 0.457,
      "step": 309520
    },
    {
      "epoch": 126.7,
      "grad_norm": 2.0779225826263428,
      "learning_rate": 1.5207053744973025e-06,
      "loss": 0.464,
      "step": 309530
    },
    {
      "epoch": 126.7,
      "grad_norm": 1.831556797027588,
      "learning_rate": 1.5205568494216324e-06,
      "loss": 0.4578,
      "step": 309540
    },
    {
      "epoch": 126.71,
      "grad_norm": 1.9860777854919434,
      "learning_rate": 1.520408328116883e-06,
      "loss": 0.4595,
      "step": 309550
    },
    {
      "epoch": 126.71,
      "grad_norm": 1.9359205961227417,
      "learning_rate": 1.5202598105837336e-06,
      "loss": 0.4432,
      "step": 309560
    },
    {
      "epoch": 126.72,
      "grad_norm": 1.905264139175415,
      "learning_rate": 1.5201112968228656e-06,
      "loss": 0.4463,
      "step": 309570
    },
    {
      "epoch": 126.72,
      "grad_norm": 2.123159170150757,
      "learning_rate": 1.5199627868349589e-06,
      "loss": 0.4563,
      "step": 309580
    },
    {
      "epoch": 126.73,
      "grad_norm": 2.260698080062866,
      "learning_rate": 1.5198142806206943e-06,
      "loss": 0.4781,
      "step": 309590
    },
    {
      "epoch": 126.73,
      "grad_norm": 1.8028773069381714,
      "learning_rate": 1.519665778180751e-06,
      "loss": 0.4587,
      "step": 309600
    },
    {
      "epoch": 126.73,
      "grad_norm": 1.6658457517623901,
      "learning_rate": 1.5195172795158104e-06,
      "loss": 0.4653,
      "step": 309610
    },
    {
      "epoch": 126.74,
      "grad_norm": 2.138227701187134,
      "learning_rate": 1.5193687846265516e-06,
      "loss": 0.4714,
      "step": 309620
    },
    {
      "epoch": 126.74,
      "grad_norm": 2.458690881729126,
      "learning_rate": 1.5192202935136558e-06,
      "loss": 0.4335,
      "step": 309630
    },
    {
      "epoch": 126.75,
      "grad_norm": 2.3407046794891357,
      "learning_rate": 1.519071806177803e-06,
      "loss": 0.4748,
      "step": 309640
    },
    {
      "epoch": 126.75,
      "grad_norm": 2.4052786827087402,
      "learning_rate": 1.518923322619673e-06,
      "loss": 0.4561,
      "step": 309650
    },
    {
      "epoch": 126.75,
      "grad_norm": 1.924241065979004,
      "learning_rate": 1.5187748428399462e-06,
      "loss": 0.4464,
      "step": 309660
    },
    {
      "epoch": 126.76,
      "grad_norm": 1.6130139827728271,
      "learning_rate": 1.5186263668393034e-06,
      "loss": 0.452,
      "step": 309670
    },
    {
      "epoch": 126.76,
      "grad_norm": 1.6585582494735718,
      "learning_rate": 1.5184778946184243e-06,
      "loss": 0.4582,
      "step": 309680
    },
    {
      "epoch": 126.77,
      "grad_norm": 2.6281256675720215,
      "learning_rate": 1.5183294261779883e-06,
      "loss": 0.4727,
      "step": 309690
    },
    {
      "epoch": 126.77,
      "grad_norm": 1.7020176649093628,
      "learning_rate": 1.5181809615186764e-06,
      "loss": 0.446,
      "step": 309700
    },
    {
      "epoch": 126.77,
      "grad_norm": 1.7350424528121948,
      "learning_rate": 1.5180325006411674e-06,
      "loss": 0.48,
      "step": 309710
    },
    {
      "epoch": 126.78,
      "grad_norm": 1.8661396503448486,
      "learning_rate": 1.5178840435461428e-06,
      "loss": 0.4633,
      "step": 309720
    },
    {
      "epoch": 126.78,
      "grad_norm": 2.1396238803863525,
      "learning_rate": 1.5177355902342832e-06,
      "loss": 0.4556,
      "step": 309730
    },
    {
      "epoch": 126.79,
      "grad_norm": 2.0256454944610596,
      "learning_rate": 1.5175871407062673e-06,
      "loss": 0.4614,
      "step": 309740
    },
    {
      "epoch": 126.79,
      "grad_norm": 2.050621509552002,
      "learning_rate": 1.5174386949627756e-06,
      "loss": 0.4473,
      "step": 309750
    },
    {
      "epoch": 126.79,
      "grad_norm": 1.8947638273239136,
      "learning_rate": 1.5172902530044878e-06,
      "loss": 0.4615,
      "step": 309760
    },
    {
      "epoch": 126.8,
      "grad_norm": 1.7318283319473267,
      "learning_rate": 1.517141814832084e-06,
      "loss": 0.4547,
      "step": 309770
    },
    {
      "epoch": 126.8,
      "grad_norm": 1.5797749757766724,
      "learning_rate": 1.5169933804462444e-06,
      "loss": 0.4533,
      "step": 309780
    },
    {
      "epoch": 126.81,
      "grad_norm": 1.8711031675338745,
      "learning_rate": 1.5168449498476487e-06,
      "loss": 0.4635,
      "step": 309790
    },
    {
      "epoch": 126.81,
      "grad_norm": 9.413233757019043,
      "learning_rate": 1.5166965230369772e-06,
      "loss": 0.4576,
      "step": 309800
    },
    {
      "epoch": 126.82,
      "grad_norm": 3.4066174030303955,
      "learning_rate": 1.5165481000149097e-06,
      "loss": 0.4567,
      "step": 309810
    },
    {
      "epoch": 126.82,
      "grad_norm": 1.4304075241088867,
      "learning_rate": 1.516399680782126e-06,
      "loss": 0.461,
      "step": 309820
    },
    {
      "epoch": 126.82,
      "grad_norm": 1.7578926086425781,
      "learning_rate": 1.5162512653393056e-06,
      "loss": 0.4405,
      "step": 309830
    },
    {
      "epoch": 126.83,
      "grad_norm": 1.9816161394119263,
      "learning_rate": 1.516102853687129e-06,
      "loss": 0.4555,
      "step": 309840
    },
    {
      "epoch": 126.83,
      "grad_norm": 2.114501953125,
      "learning_rate": 1.515954445826276e-06,
      "loss": 0.4385,
      "step": 309850
    },
    {
      "epoch": 126.84,
      "grad_norm": 1.8356021642684937,
      "learning_rate": 1.515806041757426e-06,
      "loss": 0.438,
      "step": 309860
    },
    {
      "epoch": 126.84,
      "grad_norm": 1.6671267747879028,
      "learning_rate": 1.5156576414812588e-06,
      "loss": 0.4519,
      "step": 309870
    },
    {
      "epoch": 126.84,
      "grad_norm": 2.05700421333313,
      "learning_rate": 1.5155092449984545e-06,
      "loss": 0.4898,
      "step": 309880
    },
    {
      "epoch": 126.85,
      "grad_norm": 1.9253475666046143,
      "learning_rate": 1.5153608523096933e-06,
      "loss": 0.449,
      "step": 309890
    },
    {
      "epoch": 126.85,
      "grad_norm": 1.7814024686813354,
      "learning_rate": 1.5152124634156536e-06,
      "loss": 0.4531,
      "step": 309900
    },
    {
      "epoch": 126.86,
      "grad_norm": 1.8759530782699585,
      "learning_rate": 1.5150640783170165e-06,
      "loss": 0.4401,
      "step": 309910
    },
    {
      "epoch": 126.86,
      "grad_norm": 2.2973101139068604,
      "learning_rate": 1.5149156970144614e-06,
      "loss": 0.4688,
      "step": 309920
    },
    {
      "epoch": 126.86,
      "grad_norm": 2.222790479660034,
      "learning_rate": 1.5147673195086679e-06,
      "loss": 0.4599,
      "step": 309930
    },
    {
      "epoch": 126.87,
      "grad_norm": 1.6475601196289062,
      "learning_rate": 1.5146189458003156e-06,
      "loss": 0.4574,
      "step": 309940
    },
    {
      "epoch": 126.87,
      "grad_norm": 1.7115520238876343,
      "learning_rate": 1.5144705758900846e-06,
      "loss": 0.4441,
      "step": 309950
    },
    {
      "epoch": 126.88,
      "grad_norm": 1.962109923362732,
      "learning_rate": 1.514322209778654e-06,
      "loss": 0.4492,
      "step": 309960
    },
    {
      "epoch": 126.88,
      "grad_norm": 2.0418248176574707,
      "learning_rate": 1.5141738474667037e-06,
      "loss": 0.4514,
      "step": 309970
    },
    {
      "epoch": 126.88,
      "grad_norm": 2.117145299911499,
      "learning_rate": 1.5140254889549131e-06,
      "loss": 0.4474,
      "step": 309980
    },
    {
      "epoch": 126.89,
      "grad_norm": 2.0080766677856445,
      "learning_rate": 1.513877134243962e-06,
      "loss": 0.465,
      "step": 309990
    },
    {
      "epoch": 126.89,
      "grad_norm": 1.628705382347107,
      "learning_rate": 1.5137287833345299e-06,
      "loss": 0.4482,
      "step": 310000
    },
    {
      "epoch": 126.9,
      "grad_norm": 2.1085243225097656,
      "learning_rate": 1.5135804362272966e-06,
      "loss": 0.4661,
      "step": 310010
    },
    {
      "epoch": 126.9,
      "grad_norm": 1.6411254405975342,
      "learning_rate": 1.5134320929229414e-06,
      "loss": 0.4559,
      "step": 310020
    },
    {
      "epoch": 126.91,
      "grad_norm": 1.9983912706375122,
      "learning_rate": 1.513283753422144e-06,
      "loss": 0.4331,
      "step": 310030
    },
    {
      "epoch": 126.91,
      "grad_norm": 1.8586349487304688,
      "learning_rate": 1.5131354177255845e-06,
      "loss": 0.4771,
      "step": 310040
    },
    {
      "epoch": 126.91,
      "grad_norm": 1.925065040588379,
      "learning_rate": 1.512987085833941e-06,
      "loss": 0.4556,
      "step": 310050
    },
    {
      "epoch": 126.92,
      "grad_norm": 2.1793458461761475,
      "learning_rate": 1.512838757747894e-06,
      "loss": 0.4635,
      "step": 310060
    },
    {
      "epoch": 126.92,
      "grad_norm": 2.146162748336792,
      "learning_rate": 1.5126904334681226e-06,
      "loss": 0.4573,
      "step": 310070
    },
    {
      "epoch": 126.93,
      "grad_norm": 2.1823017597198486,
      "learning_rate": 1.5125421129953063e-06,
      "loss": 0.4913,
      "step": 310080
    },
    {
      "epoch": 126.93,
      "grad_norm": 1.9538006782531738,
      "learning_rate": 1.5123937963301242e-06,
      "loss": 0.4914,
      "step": 310090
    },
    {
      "epoch": 126.93,
      "grad_norm": 1.651009202003479,
      "learning_rate": 1.5122454834732566e-06,
      "loss": 0.4423,
      "step": 310100
    },
    {
      "epoch": 126.94,
      "grad_norm": 1.7713890075683594,
      "learning_rate": 1.512097174425382e-06,
      "loss": 0.4645,
      "step": 310110
    },
    {
      "epoch": 126.94,
      "grad_norm": 2.9835164546966553,
      "learning_rate": 1.5119488691871807e-06,
      "loss": 0.4773,
      "step": 310120
    },
    {
      "epoch": 126.95,
      "grad_norm": 2.0560693740844727,
      "learning_rate": 1.5118005677593315e-06,
      "loss": 0.4774,
      "step": 310130
    },
    {
      "epoch": 126.95,
      "grad_norm": 2.6061301231384277,
      "learning_rate": 1.5116522701425137e-06,
      "loss": 0.4696,
      "step": 310140
    },
    {
      "epoch": 126.95,
      "grad_norm": 1.8398261070251465,
      "learning_rate": 1.5115039763374068e-06,
      "loss": 0.4406,
      "step": 310150
    },
    {
      "epoch": 126.96,
      "grad_norm": 2.2943594455718994,
      "learning_rate": 1.5113556863446901e-06,
      "loss": 0.4726,
      "step": 310160
    },
    {
      "epoch": 126.96,
      "grad_norm": 1.952833652496338,
      "learning_rate": 1.5112074001650433e-06,
      "loss": 0.4604,
      "step": 310170
    },
    {
      "epoch": 126.97,
      "grad_norm": 1.879429578781128,
      "learning_rate": 1.5110591177991445e-06,
      "loss": 0.4736,
      "step": 310180
    },
    {
      "epoch": 126.97,
      "grad_norm": 1.9822853803634644,
      "learning_rate": 1.510910839247674e-06,
      "loss": 0.4554,
      "step": 310190
    },
    {
      "epoch": 126.98,
      "grad_norm": 2.297699451446533,
      "learning_rate": 1.5107625645113103e-06,
      "loss": 0.4457,
      "step": 310200
    },
    {
      "epoch": 126.98,
      "grad_norm": 2.0867607593536377,
      "learning_rate": 1.5106142935907335e-06,
      "loss": 0.4646,
      "step": 310210
    },
    {
      "epoch": 126.98,
      "grad_norm": 1.8494224548339844,
      "learning_rate": 1.510466026486622e-06,
      "loss": 0.4398,
      "step": 310220
    },
    {
      "epoch": 126.99,
      "grad_norm": 1.764202356338501,
      "learning_rate": 1.5103177631996555e-06,
      "loss": 0.4626,
      "step": 310230
    },
    {
      "epoch": 126.99,
      "grad_norm": 1.9224414825439453,
      "learning_rate": 1.5101695037305132e-06,
      "loss": 0.4558,
      "step": 310240
    },
    {
      "epoch": 127.0,
      "grad_norm": 1.8089107275009155,
      "learning_rate": 1.5100212480798739e-06,
      "loss": 0.4522,
      "step": 310250
    },
    {
      "epoch": 127.0,
      "grad_norm": 2.038267135620117,
      "learning_rate": 1.5098729962484172e-06,
      "loss": 0.4521,
      "step": 310260
    },
    {
      "epoch": 127.0,
      "eval_loss": 0.4598701596260071,
      "eval_runtime": 52.4239,
      "eval_samples_per_second": 65.791,
      "eval_steps_per_second": 8.241,
      "step": 310261
    },
    {
      "epoch": 127.0,
      "grad_norm": 2.3043360710144043,
      "learning_rate": 1.5097247482368206e-06,
      "loss": 0.451,
      "step": 310270
    },
    {
      "epoch": 127.01,
      "grad_norm": 1.8093082904815674,
      "learning_rate": 1.5095765040457656e-06,
      "loss": 0.4312,
      "step": 310280
    },
    {
      "epoch": 127.01,
      "grad_norm": 1.7040867805480957,
      "learning_rate": 1.50942826367593e-06,
      "loss": 0.4336,
      "step": 310290
    },
    {
      "epoch": 127.02,
      "grad_norm": 1.7513656616210938,
      "learning_rate": 1.509280027127993e-06,
      "loss": 0.443,
      "step": 310300
    },
    {
      "epoch": 127.02,
      "grad_norm": 2.78096342086792,
      "learning_rate": 1.5091317944026344e-06,
      "loss": 0.4595,
      "step": 310310
    },
    {
      "epoch": 127.02,
      "grad_norm": 1.9594777822494507,
      "learning_rate": 1.5089835655005317e-06,
      "loss": 0.4531,
      "step": 310320
    },
    {
      "epoch": 127.03,
      "grad_norm": 6.878755569458008,
      "learning_rate": 1.508835340422365e-06,
      "loss": 0.4515,
      "step": 310330
    },
    {
      "epoch": 127.03,
      "grad_norm": 1.9264658689498901,
      "learning_rate": 1.5086871191688135e-06,
      "loss": 0.4697,
      "step": 310340
    },
    {
      "epoch": 127.04,
      "grad_norm": 2.102602481842041,
      "learning_rate": 1.508538901740555e-06,
      "loss": 0.4444,
      "step": 310350
    },
    {
      "epoch": 127.04,
      "grad_norm": 2.127531051635742,
      "learning_rate": 1.5083906881382696e-06,
      "loss": 0.454,
      "step": 310360
    },
    {
      "epoch": 127.04,
      "grad_norm": 1.9515037536621094,
      "learning_rate": 1.5082424783626355e-06,
      "loss": 0.4334,
      "step": 310370
    },
    {
      "epoch": 127.05,
      "grad_norm": 2.1188018321990967,
      "learning_rate": 1.5080942724143325e-06,
      "loss": 0.4663,
      "step": 310380
    },
    {
      "epoch": 127.05,
      "grad_norm": 1.811955213546753,
      "learning_rate": 1.5079460702940386e-06,
      "loss": 0.4466,
      "step": 310390
    },
    {
      "epoch": 127.06,
      "grad_norm": 2.0542333126068115,
      "learning_rate": 1.5077978720024334e-06,
      "loss": 0.4699,
      "step": 310400
    },
    {
      "epoch": 127.06,
      "grad_norm": 1.846053957939148,
      "learning_rate": 1.5076496775401953e-06,
      "loss": 0.4321,
      "step": 310410
    },
    {
      "epoch": 127.07,
      "grad_norm": 2.199542999267578,
      "learning_rate": 1.5075014869080033e-06,
      "loss": 0.4524,
      "step": 310420
    },
    {
      "epoch": 127.07,
      "grad_norm": 2.341341495513916,
      "learning_rate": 1.5073533001065362e-06,
      "loss": 0.4614,
      "step": 310430
    },
    {
      "epoch": 127.07,
      "grad_norm": 1.7929514646530151,
      "learning_rate": 1.5072051171364724e-06,
      "loss": 0.4487,
      "step": 310440
    },
    {
      "epoch": 127.08,
      "grad_norm": 1.8607208728790283,
      "learning_rate": 1.5070569379984919e-06,
      "loss": 0.4535,
      "step": 310450
    },
    {
      "epoch": 127.08,
      "grad_norm": 1.9353759288787842,
      "learning_rate": 1.5069087626932718e-06,
      "loss": 0.4561,
      "step": 310460
    },
    {
      "epoch": 127.09,
      "grad_norm": 2.1307382583618164,
      "learning_rate": 1.5067605912214921e-06,
      "loss": 0.4432,
      "step": 310470
    },
    {
      "epoch": 127.09,
      "grad_norm": 2.196059226989746,
      "learning_rate": 1.5066124235838314e-06,
      "loss": 0.4374,
      "step": 310480
    },
    {
      "epoch": 127.09,
      "grad_norm": 1.9237977266311646,
      "learning_rate": 1.5064642597809686e-06,
      "loss": 0.4513,
      "step": 310490
    },
    {
      "epoch": 127.1,
      "grad_norm": 1.9044647216796875,
      "learning_rate": 1.506316099813582e-06,
      "loss": 0.4724,
      "step": 310500
    },
    {
      "epoch": 127.1,
      "grad_norm": 1.8220405578613281,
      "learning_rate": 1.5061679436823502e-06,
      "loss": 0.4559,
      "step": 310510
    },
    {
      "epoch": 127.11,
      "grad_norm": 1.7228034734725952,
      "learning_rate": 1.506019791387952e-06,
      "loss": 0.4625,
      "step": 310520
    },
    {
      "epoch": 127.11,
      "grad_norm": 2.20379638671875,
      "learning_rate": 1.505871642931067e-06,
      "loss": 0.4474,
      "step": 310530
    },
    {
      "epoch": 127.11,
      "grad_norm": 2.1636695861816406,
      "learning_rate": 1.5057234983123721e-06,
      "loss": 0.4851,
      "step": 310540
    },
    {
      "epoch": 127.12,
      "grad_norm": 1.7965195178985596,
      "learning_rate": 1.505575357532547e-06,
      "loss": 0.4618,
      "step": 310550
    },
    {
      "epoch": 127.12,
      "grad_norm": 2.0804331302642822,
      "learning_rate": 1.5054272205922705e-06,
      "loss": 0.4557,
      "step": 310560
    },
    {
      "epoch": 127.13,
      "grad_norm": 2.080070972442627,
      "learning_rate": 1.5052790874922202e-06,
      "loss": 0.4663,
      "step": 310570
    },
    {
      "epoch": 127.13,
      "grad_norm": 2.020862340927124,
      "learning_rate": 1.5051309582330756e-06,
      "loss": 0.4702,
      "step": 310580
    },
    {
      "epoch": 127.13,
      "grad_norm": 1.6767436265945435,
      "learning_rate": 1.5049828328155152e-06,
      "loss": 0.479,
      "step": 310590
    },
    {
      "epoch": 127.14,
      "grad_norm": 2.1234312057495117,
      "learning_rate": 1.5048347112402168e-06,
      "loss": 0.4379,
      "step": 310600
    },
    {
      "epoch": 127.14,
      "grad_norm": 1.8457410335540771,
      "learning_rate": 1.50468659350786e-06,
      "loss": 0.4374,
      "step": 310610
    },
    {
      "epoch": 127.15,
      "grad_norm": 2.366607189178467,
      "learning_rate": 1.5045384796191227e-06,
      "loss": 0.4589,
      "step": 310620
    },
    {
      "epoch": 127.15,
      "grad_norm": 2.1718597412109375,
      "learning_rate": 1.504390369574683e-06,
      "loss": 0.4406,
      "step": 310630
    },
    {
      "epoch": 127.16,
      "grad_norm": 2.6373820304870605,
      "learning_rate": 1.50424226337522e-06,
      "loss": 0.4778,
      "step": 310640
    },
    {
      "epoch": 127.16,
      "grad_norm": 2.0209262371063232,
      "learning_rate": 1.5040941610214105e-06,
      "loss": 0.4665,
      "step": 310650
    },
    {
      "epoch": 127.16,
      "grad_norm": 1.981645941734314,
      "learning_rate": 1.5039460625139359e-06,
      "loss": 0.4514,
      "step": 310660
    },
    {
      "epoch": 127.17,
      "grad_norm": 2.2340142726898193,
      "learning_rate": 1.5037979678534734e-06,
      "loss": 0.4804,
      "step": 310670
    },
    {
      "epoch": 127.17,
      "grad_norm": 1.892026424407959,
      "learning_rate": 1.5036498770407002e-06,
      "loss": 0.4619,
      "step": 310680
    },
    {
      "epoch": 127.18,
      "grad_norm": 2.0299410820007324,
      "learning_rate": 1.5035017900762959e-06,
      "loss": 0.4595,
      "step": 310690
    },
    {
      "epoch": 127.18,
      "grad_norm": 4.707353115081787,
      "learning_rate": 1.5033537069609384e-06,
      "loss": 0.4626,
      "step": 310700
    },
    {
      "epoch": 127.18,
      "grad_norm": 2.043889045715332,
      "learning_rate": 1.5032056276953065e-06,
      "loss": 0.4607,
      "step": 310710
    },
    {
      "epoch": 127.19,
      "grad_norm": 1.9273768663406372,
      "learning_rate": 1.5030575522800776e-06,
      "loss": 0.4815,
      "step": 310720
    },
    {
      "epoch": 127.19,
      "grad_norm": 1.813235878944397,
      "learning_rate": 1.5029094807159312e-06,
      "loss": 0.4593,
      "step": 310730
    },
    {
      "epoch": 127.2,
      "grad_norm": 1.5238600969314575,
      "learning_rate": 1.5027614130035448e-06,
      "loss": 0.4606,
      "step": 310740
    },
    {
      "epoch": 127.2,
      "grad_norm": 1.753402829170227,
      "learning_rate": 1.5026133491435965e-06,
      "loss": 0.4463,
      "step": 310750
    },
    {
      "epoch": 127.2,
      "grad_norm": 1.6765613555908203,
      "learning_rate": 1.502465289136766e-06,
      "loss": 0.4545,
      "step": 310760
    },
    {
      "epoch": 127.21,
      "grad_norm": 2.645169496536255,
      "learning_rate": 1.5023172329837297e-06,
      "loss": 0.4643,
      "step": 310770
    },
    {
      "epoch": 127.21,
      "grad_norm": 2.2714338302612305,
      "learning_rate": 1.5021691806851666e-06,
      "loss": 0.4429,
      "step": 310780
    },
    {
      "epoch": 127.22,
      "grad_norm": 1.6648006439208984,
      "learning_rate": 1.5020211322417548e-06,
      "loss": 0.4501,
      "step": 310790
    },
    {
      "epoch": 127.22,
      "grad_norm": 1.7771008014678955,
      "learning_rate": 1.5018730876541727e-06,
      "loss": 0.4687,
      "step": 310800
    },
    {
      "epoch": 127.22,
      "grad_norm": 1.7361946105957031,
      "learning_rate": 1.5017250469230986e-06,
      "loss": 0.4665,
      "step": 310810
    },
    {
      "epoch": 127.23,
      "grad_norm": 2.7104547023773193,
      "learning_rate": 1.50157701004921e-06,
      "loss": 0.4725,
      "step": 310820
    },
    {
      "epoch": 127.23,
      "grad_norm": 2.001352548599243,
      "learning_rate": 1.5014289770331856e-06,
      "loss": 0.4599,
      "step": 310830
    },
    {
      "epoch": 127.24,
      "grad_norm": 1.8982490301132202,
      "learning_rate": 1.5012809478757028e-06,
      "loss": 0.4609,
      "step": 310840
    },
    {
      "epoch": 127.24,
      "grad_norm": 2.0491034984588623,
      "learning_rate": 1.5011329225774407e-06,
      "loss": 0.4678,
      "step": 310850
    },
    {
      "epoch": 127.25,
      "grad_norm": 1.937106966972351,
      "learning_rate": 1.500984901139077e-06,
      "loss": 0.4494,
      "step": 310860
    },
    {
      "epoch": 127.25,
      "grad_norm": 1.8424149751663208,
      "learning_rate": 1.50083688356129e-06,
      "loss": 0.4648,
      "step": 310870
    },
    {
      "epoch": 127.25,
      "grad_norm": 1.9676644802093506,
      "learning_rate": 1.5006888698447568e-06,
      "loss": 0.4587,
      "step": 310880
    },
    {
      "epoch": 127.26,
      "grad_norm": 2.3042778968811035,
      "learning_rate": 1.5005408599901573e-06,
      "loss": 0.4684,
      "step": 310890
    },
    {
      "epoch": 127.26,
      "grad_norm": 1.7018964290618896,
      "learning_rate": 1.5003928539981674e-06,
      "loss": 0.4639,
      "step": 310900
    },
    {
      "epoch": 127.27,
      "grad_norm": 1.9428046941757202,
      "learning_rate": 1.5002448518694658e-06,
      "loss": 0.4534,
      "step": 310910
    },
    {
      "epoch": 127.27,
      "grad_norm": 2.3288636207580566,
      "learning_rate": 1.500096853604731e-06,
      "loss": 0.4556,
      "step": 310920
    },
    {
      "epoch": 127.27,
      "grad_norm": 1.9019138813018799,
      "learning_rate": 1.4999488592046405e-06,
      "loss": 0.4499,
      "step": 310930
    },
    {
      "epoch": 127.28,
      "grad_norm": 1.5628718137741089,
      "learning_rate": 1.4998008686698722e-06,
      "loss": 0.4662,
      "step": 310940
    },
    {
      "epoch": 127.28,
      "grad_norm": 2.3349742889404297,
      "learning_rate": 1.4996528820011043e-06,
      "loss": 0.4561,
      "step": 310950
    },
    {
      "epoch": 127.29,
      "grad_norm": 1.878446102142334,
      "learning_rate": 1.4995048991990147e-06,
      "loss": 0.4705,
      "step": 310960
    },
    {
      "epoch": 127.29,
      "grad_norm": 2.552497625350952,
      "learning_rate": 1.499356920264281e-06,
      "loss": 0.4507,
      "step": 310970
    },
    {
      "epoch": 127.29,
      "grad_norm": 1.9968101978302002,
      "learning_rate": 1.4992089451975818e-06,
      "loss": 0.46,
      "step": 310980
    },
    {
      "epoch": 127.3,
      "grad_norm": 2.0098330974578857,
      "learning_rate": 1.4990609739995937e-06,
      "loss": 0.4472,
      "step": 310990
    },
    {
      "epoch": 127.3,
      "grad_norm": 2.6231095790863037,
      "learning_rate": 1.4989130066709953e-06,
      "loss": 0.4541,
      "step": 311000
    },
    {
      "epoch": 127.31,
      "grad_norm": 1.9406667947769165,
      "learning_rate": 1.4987650432124645e-06,
      "loss": 0.4476,
      "step": 311010
    },
    {
      "epoch": 127.31,
      "grad_norm": 1.9695221185684204,
      "learning_rate": 1.4986170836246785e-06,
      "loss": 0.4487,
      "step": 311020
    },
    {
      "epoch": 127.31,
      "grad_norm": 1.779040813446045,
      "learning_rate": 1.4984691279083156e-06,
      "loss": 0.466,
      "step": 311030
    },
    {
      "epoch": 127.32,
      "grad_norm": 1.976582646369934,
      "learning_rate": 1.4983211760640538e-06,
      "loss": 0.4544,
      "step": 311040
    },
    {
      "epoch": 127.32,
      "grad_norm": 1.7557682991027832,
      "learning_rate": 1.49817322809257e-06,
      "loss": 0.4572,
      "step": 311050
    },
    {
      "epoch": 127.33,
      "grad_norm": 1.7959033250808716,
      "learning_rate": 1.498025283994543e-06,
      "loss": 0.4682,
      "step": 311060
    },
    {
      "epoch": 127.33,
      "grad_norm": 1.7836740016937256,
      "learning_rate": 1.4978773437706498e-06,
      "loss": 0.4532,
      "step": 311070
    },
    {
      "epoch": 127.34,
      "grad_norm": 2.4602444171905518,
      "learning_rate": 1.497729407421568e-06,
      "loss": 0.4518,
      "step": 311080
    },
    {
      "epoch": 127.34,
      "grad_norm": 1.9413628578186035,
      "learning_rate": 1.4975814749479758e-06,
      "loss": 0.4509,
      "step": 311090
    },
    {
      "epoch": 127.34,
      "grad_norm": 2.3074538707733154,
      "learning_rate": 1.4974335463505505e-06,
      "loss": 0.4448,
      "step": 311100
    },
    {
      "epoch": 127.35,
      "grad_norm": 2.1683168411254883,
      "learning_rate": 1.4972856216299698e-06,
      "loss": 0.4615,
      "step": 311110
    },
    {
      "epoch": 127.35,
      "grad_norm": 1.8115249872207642,
      "learning_rate": 1.4971377007869118e-06,
      "loss": 0.4722,
      "step": 311120
    },
    {
      "epoch": 127.36,
      "grad_norm": 1.9908692836761475,
      "learning_rate": 1.496989783822053e-06,
      "loss": 0.4428,
      "step": 311130
    },
    {
      "epoch": 127.36,
      "grad_norm": 1.7750144004821777,
      "learning_rate": 1.496841870736072e-06,
      "loss": 0.4657,
      "step": 311140
    },
    {
      "epoch": 127.36,
      "grad_norm": 2.505496025085449,
      "learning_rate": 1.4966939615296455e-06,
      "loss": 0.4189,
      "step": 311150
    },
    {
      "epoch": 127.37,
      "grad_norm": 1.7594164609909058,
      "learning_rate": 1.4965460562034516e-06,
      "loss": 0.4839,
      "step": 311160
    },
    {
      "epoch": 127.37,
      "grad_norm": 14.312342643737793,
      "learning_rate": 1.496398154758168e-06,
      "loss": 0.4655,
      "step": 311170
    },
    {
      "epoch": 127.38,
      "grad_norm": 1.9694254398345947,
      "learning_rate": 1.496250257194472e-06,
      "loss": 0.4409,
      "step": 311180
    },
    {
      "epoch": 127.38,
      "grad_norm": 1.784943699836731,
      "learning_rate": 1.4961023635130408e-06,
      "loss": 0.4765,
      "step": 311190
    },
    {
      "epoch": 127.38,
      "grad_norm": 1.5951296091079712,
      "learning_rate": 1.495954473714553e-06,
      "loss": 0.4306,
      "step": 311200
    },
    {
      "epoch": 127.39,
      "grad_norm": 1.9686973094940186,
      "learning_rate": 1.4958065877996844e-06,
      "loss": 0.4617,
      "step": 311210
    },
    {
      "epoch": 127.39,
      "grad_norm": 1.8898605108261108,
      "learning_rate": 1.4956587057691124e-06,
      "loss": 0.4527,
      "step": 311220
    },
    {
      "epoch": 127.4,
      "grad_norm": 1.8233308792114258,
      "learning_rate": 1.4955108276235163e-06,
      "loss": 0.4456,
      "step": 311230
    },
    {
      "epoch": 127.4,
      "grad_norm": 2.180166244506836,
      "learning_rate": 1.4953629533635727e-06,
      "loss": 0.4493,
      "step": 311240
    },
    {
      "epoch": 127.4,
      "grad_norm": 2.019200086593628,
      "learning_rate": 1.4952150829899587e-06,
      "loss": 0.4406,
      "step": 311250
    },
    {
      "epoch": 127.41,
      "grad_norm": 2.07454252243042,
      "learning_rate": 1.4950672165033514e-06,
      "loss": 0.457,
      "step": 311260
    },
    {
      "epoch": 127.41,
      "grad_norm": 2.2418174743652344,
      "learning_rate": 1.4949193539044288e-06,
      "loss": 0.4543,
      "step": 311270
    },
    {
      "epoch": 127.42,
      "grad_norm": 2.2362213134765625,
      "learning_rate": 1.4947714951938677e-06,
      "loss": 0.4736,
      "step": 311280
    },
    {
      "epoch": 127.42,
      "grad_norm": 1.4851863384246826,
      "learning_rate": 1.4946236403723456e-06,
      "loss": 0.4589,
      "step": 311290
    },
    {
      "epoch": 127.43,
      "grad_norm": 2.110908031463623,
      "learning_rate": 1.4944757894405401e-06,
      "loss": 0.4569,
      "step": 311300
    },
    {
      "epoch": 127.43,
      "grad_norm": 2.0430498123168945,
      "learning_rate": 1.494327942399128e-06,
      "loss": 0.4569,
      "step": 311310
    },
    {
      "epoch": 127.43,
      "grad_norm": 2.2511539459228516,
      "learning_rate": 1.494180099248787e-06,
      "loss": 0.4444,
      "step": 311320
    },
    {
      "epoch": 127.44,
      "grad_norm": 4.74553918838501,
      "learning_rate": 1.4940322599901943e-06,
      "loss": 0.4683,
      "step": 311330
    },
    {
      "epoch": 127.44,
      "grad_norm": 1.8680051565170288,
      "learning_rate": 1.4938844246240272e-06,
      "loss": 0.458,
      "step": 311340
    },
    {
      "epoch": 127.45,
      "grad_norm": 1.7791892290115356,
      "learning_rate": 1.493736593150962e-06,
      "loss": 0.4645,
      "step": 311350
    },
    {
      "epoch": 127.45,
      "grad_norm": 2.2115530967712402,
      "learning_rate": 1.4935887655716767e-06,
      "loss": 0.4578,
      "step": 311360
    },
    {
      "epoch": 127.45,
      "grad_norm": 1.6811916828155518,
      "learning_rate": 1.4934409418868483e-06,
      "loss": 0.4614,
      "step": 311370
    },
    {
      "epoch": 127.46,
      "grad_norm": 1.9807549715042114,
      "learning_rate": 1.4932931220971543e-06,
      "loss": 0.4506,
      "step": 311380
    },
    {
      "epoch": 127.46,
      "grad_norm": 1.7184159755706787,
      "learning_rate": 1.4931453062032714e-06,
      "loss": 0.4494,
      "step": 311390
    },
    {
      "epoch": 127.47,
      "grad_norm": 1.5321686267852783,
      "learning_rate": 1.4929974942058768e-06,
      "loss": 0.4559,
      "step": 311400
    },
    {
      "epoch": 127.47,
      "grad_norm": 1.9134036302566528,
      "learning_rate": 1.4928496861056472e-06,
      "loss": 0.4545,
      "step": 311410
    },
    {
      "epoch": 127.47,
      "grad_norm": 1.5619237422943115,
      "learning_rate": 1.4927018819032605e-06,
      "loss": 0.4481,
      "step": 311420
    },
    {
      "epoch": 127.48,
      "grad_norm": 1.9268627166748047,
      "learning_rate": 1.4925540815993939e-06,
      "loss": 0.4528,
      "step": 311430
    },
    {
      "epoch": 127.48,
      "grad_norm": 2.2672126293182373,
      "learning_rate": 1.4924062851947235e-06,
      "loss": 0.4635,
      "step": 311440
    },
    {
      "epoch": 127.49,
      "grad_norm": 1.872809648513794,
      "learning_rate": 1.4922584926899273e-06,
      "loss": 0.4727,
      "step": 311450
    },
    {
      "epoch": 127.49,
      "grad_norm": 2.2472879886627197,
      "learning_rate": 1.4921107040856814e-06,
      "loss": 0.4436,
      "step": 311460
    },
    {
      "epoch": 127.49,
      "grad_norm": 2.4056525230407715,
      "learning_rate": 1.4919629193826631e-06,
      "loss": 0.4476,
      "step": 311470
    },
    {
      "epoch": 127.5,
      "grad_norm": 2.285677433013916,
      "learning_rate": 1.4918151385815503e-06,
      "loss": 0.474,
      "step": 311480
    },
    {
      "epoch": 127.5,
      "grad_norm": 7.9717698097229,
      "learning_rate": 1.4916673616830187e-06,
      "loss": 0.4503,
      "step": 311490
    },
    {
      "epoch": 127.51,
      "grad_norm": 2.3106818199157715,
      "learning_rate": 1.4915195886877454e-06,
      "loss": 0.466,
      "step": 311500
    },
    {
      "epoch": 127.51,
      "grad_norm": 1.9681543111801147,
      "learning_rate": 1.4913718195964077e-06,
      "loss": 0.4487,
      "step": 311510
    },
    {
      "epoch": 127.52,
      "grad_norm": 1.924831509590149,
      "learning_rate": 1.4912240544096828e-06,
      "loss": 0.4488,
      "step": 311520
    },
    {
      "epoch": 127.52,
      "grad_norm": 2.1914048194885254,
      "learning_rate": 1.491076293128247e-06,
      "loss": 0.4585,
      "step": 311530
    },
    {
      "epoch": 127.52,
      "grad_norm": 1.9088813066482544,
      "learning_rate": 1.4909285357527775e-06,
      "loss": 0.4681,
      "step": 311540
    },
    {
      "epoch": 127.53,
      "grad_norm": 1.9175817966461182,
      "learning_rate": 1.490780782283951e-06,
      "loss": 0.4403,
      "step": 311550
    },
    {
      "epoch": 127.53,
      "grad_norm": 1.983750820159912,
      "learning_rate": 1.4906330327224447e-06,
      "loss": 0.4762,
      "step": 311560
    },
    {
      "epoch": 127.54,
      "grad_norm": 2.9121735095977783,
      "learning_rate": 1.4904852870689348e-06,
      "loss": 0.4626,
      "step": 311570
    },
    {
      "epoch": 127.54,
      "grad_norm": 1.9889614582061768,
      "learning_rate": 1.4903375453240982e-06,
      "loss": 0.439,
      "step": 311580
    },
    {
      "epoch": 127.54,
      "grad_norm": 1.7344926595687866,
      "learning_rate": 1.490189807488611e-06,
      "loss": 0.4544,
      "step": 311590
    },
    {
      "epoch": 127.55,
      "grad_norm": 2.16367244720459,
      "learning_rate": 1.490042073563152e-06,
      "loss": 0.4565,
      "step": 311600
    },
    {
      "epoch": 127.55,
      "grad_norm": 2.016702651977539,
      "learning_rate": 1.4898943435483973e-06,
      "loss": 0.4633,
      "step": 311610
    },
    {
      "epoch": 127.56,
      "grad_norm": 2.435870885848999,
      "learning_rate": 1.4897466174450223e-06,
      "loss": 0.4604,
      "step": 311620
    },
    {
      "epoch": 127.56,
      "grad_norm": 1.8126052618026733,
      "learning_rate": 1.489598895253705e-06,
      "loss": 0.4855,
      "step": 311630
    },
    {
      "epoch": 127.56,
      "grad_norm": 2.028162717819214,
      "learning_rate": 1.4894511769751213e-06,
      "loss": 0.4616,
      "step": 311640
    },
    {
      "epoch": 127.57,
      "grad_norm": 1.4375602006912231,
      "learning_rate": 1.4893034626099481e-06,
      "loss": 0.4642,
      "step": 311650
    },
    {
      "epoch": 127.57,
      "grad_norm": 1.5991214513778687,
      "learning_rate": 1.4891557521588624e-06,
      "loss": 0.449,
      "step": 311660
    },
    {
      "epoch": 127.58,
      "grad_norm": 1.9418314695358276,
      "learning_rate": 1.4890080456225405e-06,
      "loss": 0.4844,
      "step": 311670
    },
    {
      "epoch": 127.58,
      "grad_norm": 1.896793007850647,
      "learning_rate": 1.488860343001659e-06,
      "loss": 0.4637,
      "step": 311680
    },
    {
      "epoch": 127.58,
      "grad_norm": 1.7480319738388062,
      "learning_rate": 1.4887126442968952e-06,
      "loss": 0.4754,
      "step": 311690
    },
    {
      "epoch": 127.59,
      "grad_norm": 2.273564577102661,
      "learning_rate": 1.488564949508925e-06,
      "loss": 0.478,
      "step": 311700
    },
    {
      "epoch": 127.59,
      "grad_norm": 1.6714097261428833,
      "learning_rate": 1.4884172586384248e-06,
      "loss": 0.4657,
      "step": 311710
    },
    {
      "epoch": 127.6,
      "grad_norm": 2.381582498550415,
      "learning_rate": 1.4882695716860715e-06,
      "loss": 0.4625,
      "step": 311720
    },
    {
      "epoch": 127.6,
      "grad_norm": 1.992492437362671,
      "learning_rate": 1.4881218886525412e-06,
      "loss": 0.435,
      "step": 311730
    },
    {
      "epoch": 127.61,
      "grad_norm": 1.9319140911102295,
      "learning_rate": 1.487974209538511e-06,
      "loss": 0.4564,
      "step": 311740
    },
    {
      "epoch": 127.61,
      "grad_norm": 1.9367289543151855,
      "learning_rate": 1.4878265343446572e-06,
      "loss": 0.4354,
      "step": 311750
    },
    {
      "epoch": 127.61,
      "grad_norm": 2.006338119506836,
      "learning_rate": 1.4876788630716563e-06,
      "loss": 0.4601,
      "step": 311760
    },
    {
      "epoch": 127.62,
      "grad_norm": 2.005220890045166,
      "learning_rate": 1.4875311957201846e-06,
      "loss": 0.4483,
      "step": 311770
    },
    {
      "epoch": 127.62,
      "grad_norm": 2.390037775039673,
      "learning_rate": 1.4873835322909183e-06,
      "loss": 0.46,
      "step": 311780
    },
    {
      "epoch": 127.63,
      "grad_norm": 1.581436038017273,
      "learning_rate": 1.4872358727845342e-06,
      "loss": 0.4313,
      "step": 311790
    },
    {
      "epoch": 127.63,
      "grad_norm": 2.1905601024627686,
      "learning_rate": 1.487088217201709e-06,
      "loss": 0.4526,
      "step": 311800
    },
    {
      "epoch": 127.63,
      "grad_norm": 1.7587828636169434,
      "learning_rate": 1.4869405655431188e-06,
      "loss": 0.4519,
      "step": 311810
    },
    {
      "epoch": 127.64,
      "grad_norm": 2.0019187927246094,
      "learning_rate": 1.4867929178094399e-06,
      "loss": 0.4554,
      "step": 311820
    },
    {
      "epoch": 127.64,
      "grad_norm": 2.8071177005767822,
      "learning_rate": 1.4866452740013485e-06,
      "loss": 0.4786,
      "step": 311830
    },
    {
      "epoch": 127.65,
      "grad_norm": 1.915418028831482,
      "learning_rate": 1.486497634119522e-06,
      "loss": 0.4316,
      "step": 311840
    },
    {
      "epoch": 127.65,
      "grad_norm": 2.2423593997955322,
      "learning_rate": 1.4863499981646348e-06,
      "loss": 0.4611,
      "step": 311850
    },
    {
      "epoch": 127.65,
      "grad_norm": 2.616708278656006,
      "learning_rate": 1.4862023661373646e-06,
      "loss": 0.4515,
      "step": 311860
    },
    {
      "epoch": 127.66,
      "grad_norm": 1.800749659538269,
      "learning_rate": 1.4860547380383875e-06,
      "loss": 0.4641,
      "step": 311870
    },
    {
      "epoch": 127.66,
      "grad_norm": 1.7712225914001465,
      "learning_rate": 1.485907113868379e-06,
      "loss": 0.4533,
      "step": 311880
    },
    {
      "epoch": 127.67,
      "grad_norm": 1.8200808763504028,
      "learning_rate": 1.4857594936280162e-06,
      "loss": 0.4528,
      "step": 311890
    },
    {
      "epoch": 127.67,
      "grad_norm": 1.9804584980010986,
      "learning_rate": 1.4856118773179751e-06,
      "loss": 0.4379,
      "step": 311900
    },
    {
      "epoch": 127.67,
      "grad_norm": 2.539486885070801,
      "learning_rate": 1.4854642649389318e-06,
      "loss": 0.4786,
      "step": 311910
    },
    {
      "epoch": 127.68,
      "grad_norm": 2.148625373840332,
      "learning_rate": 1.4853166564915627e-06,
      "loss": 0.4616,
      "step": 311920
    },
    {
      "epoch": 127.68,
      "grad_norm": 2.059237003326416,
      "learning_rate": 1.4851690519765437e-06,
      "loss": 0.4598,
      "step": 311930
    },
    {
      "epoch": 127.69,
      "grad_norm": 2.043919324874878,
      "learning_rate": 1.485021451394551e-06,
      "loss": 0.4578,
      "step": 311940
    },
    {
      "epoch": 127.69,
      "grad_norm": 1.5673233270645142,
      "learning_rate": 1.4848738547462608e-06,
      "loss": 0.4618,
      "step": 311950
    },
    {
      "epoch": 127.7,
      "grad_norm": 1.901090145111084,
      "learning_rate": 1.484726262032349e-06,
      "loss": 0.477,
      "step": 311960
    },
    {
      "epoch": 127.7,
      "grad_norm": 2.441053867340088,
      "learning_rate": 1.4845786732534917e-06,
      "loss": 0.4573,
      "step": 311970
    },
    {
      "epoch": 127.7,
      "grad_norm": 1.7783606052398682,
      "learning_rate": 1.4844310884103655e-06,
      "loss": 0.4479,
      "step": 311980
    },
    {
      "epoch": 127.71,
      "grad_norm": 2.610389232635498,
      "learning_rate": 1.484283507503646e-06,
      "loss": 0.4557,
      "step": 311990
    },
    {
      "epoch": 127.71,
      "grad_norm": 2.4437689781188965,
      "learning_rate": 1.4841359305340094e-06,
      "loss": 0.4639,
      "step": 312000
    },
    {
      "epoch": 127.72,
      "grad_norm": 1.8484889268875122,
      "learning_rate": 1.483988357502132e-06,
      "loss": 0.4351,
      "step": 312010
    },
    {
      "epoch": 127.72,
      "grad_norm": 1.7398093938827515,
      "learning_rate": 1.4838407884086894e-06,
      "loss": 0.4641,
      "step": 312020
    },
    {
      "epoch": 127.72,
      "grad_norm": 2.0616490840911865,
      "learning_rate": 1.4836932232543578e-06,
      "loss": 0.4385,
      "step": 312030
    },
    {
      "epoch": 127.73,
      "grad_norm": 1.8628145456314087,
      "learning_rate": 1.4835456620398132e-06,
      "loss": 0.472,
      "step": 312040
    },
    {
      "epoch": 127.73,
      "grad_norm": 2.026634931564331,
      "learning_rate": 1.4833981047657313e-06,
      "loss": 0.4784,
      "step": 312050
    },
    {
      "epoch": 127.74,
      "grad_norm": 2.731719970703125,
      "learning_rate": 1.4832505514327887e-06,
      "loss": 0.4696,
      "step": 312060
    },
    {
      "epoch": 127.74,
      "grad_norm": 1.879328727722168,
      "learning_rate": 1.4831030020416602e-06,
      "loss": 0.4428,
      "step": 312070
    },
    {
      "epoch": 127.74,
      "grad_norm": 1.8922001123428345,
      "learning_rate": 1.4829554565930224e-06,
      "loss": 0.4673,
      "step": 312080
    },
    {
      "epoch": 127.75,
      "grad_norm": 1.9536503553390503,
      "learning_rate": 1.4828079150875513e-06,
      "loss": 0.4473,
      "step": 312090
    },
    {
      "epoch": 127.75,
      "grad_norm": 2.2993967533111572,
      "learning_rate": 1.482660377525922e-06,
      "loss": 0.4454,
      "step": 312100
    },
    {
      "epoch": 127.76,
      "grad_norm": 1.7316778898239136,
      "learning_rate": 1.4825128439088116e-06,
      "loss": 0.4433,
      "step": 312110
    },
    {
      "epoch": 127.76,
      "grad_norm": 1.9809263944625854,
      "learning_rate": 1.4823653142368945e-06,
      "loss": 0.4553,
      "step": 312120
    },
    {
      "epoch": 127.77,
      "grad_norm": 1.9067103862762451,
      "learning_rate": 1.4822177885108478e-06,
      "loss": 0.4551,
      "step": 312130
    },
    {
      "epoch": 127.77,
      "grad_norm": 2.0999350547790527,
      "learning_rate": 1.4820702667313468e-06,
      "loss": 0.4544,
      "step": 312140
    },
    {
      "epoch": 127.77,
      "grad_norm": 2.077230930328369,
      "learning_rate": 1.4819227488990675e-06,
      "loss": 0.4649,
      "step": 312150
    },
    {
      "epoch": 127.78,
      "grad_norm": 2.0366435050964355,
      "learning_rate": 1.481775235014684e-06,
      "loss": 0.468,
      "step": 312160
    },
    {
      "epoch": 127.78,
      "grad_norm": 1.6972535848617554,
      "learning_rate": 1.481627725078874e-06,
      "loss": 0.4709,
      "step": 312170
    },
    {
      "epoch": 127.79,
      "grad_norm": 2.355646848678589,
      "learning_rate": 1.4814802190923129e-06,
      "loss": 0.4629,
      "step": 312180
    },
    {
      "epoch": 127.79,
      "grad_norm": 1.7056212425231934,
      "learning_rate": 1.4813327170556764e-06,
      "loss": 0.4747,
      "step": 312190
    },
    {
      "epoch": 127.79,
      "grad_norm": 1.9339892864227295,
      "learning_rate": 1.4811852189696398e-06,
      "loss": 0.4312,
      "step": 312200
    },
    {
      "epoch": 127.8,
      "grad_norm": 2.139392137527466,
      "learning_rate": 1.4810377248348786e-06,
      "loss": 0.4555,
      "step": 312210
    },
    {
      "epoch": 127.8,
      "grad_norm": 1.8426015377044678,
      "learning_rate": 1.4808902346520689e-06,
      "loss": 0.4447,
      "step": 312220
    },
    {
      "epoch": 127.81,
      "grad_norm": 2.423576593399048,
      "learning_rate": 1.480742748421886e-06,
      "loss": 0.4587,
      "step": 312230
    },
    {
      "epoch": 127.81,
      "grad_norm": 1.9162263870239258,
      "learning_rate": 1.480595266145006e-06,
      "loss": 0.441,
      "step": 312240
    },
    {
      "epoch": 127.81,
      "grad_norm": 2.0355849266052246,
      "learning_rate": 1.4804477878221036e-06,
      "loss": 0.4752,
      "step": 312250
    },
    {
      "epoch": 127.82,
      "grad_norm": 1.927755355834961,
      "learning_rate": 1.4803003134538553e-06,
      "loss": 0.4585,
      "step": 312260
    },
    {
      "epoch": 127.82,
      "grad_norm": 1.6835273504257202,
      "learning_rate": 1.4801528430409364e-06,
      "loss": 0.4515,
      "step": 312270
    },
    {
      "epoch": 127.83,
      "grad_norm": 3.657008647918701,
      "learning_rate": 1.4800053765840225e-06,
      "loss": 0.4629,
      "step": 312280
    },
    {
      "epoch": 127.83,
      "grad_norm": 2.02654767036438,
      "learning_rate": 1.4798579140837887e-06,
      "loss": 0.4856,
      "step": 312290
    },
    {
      "epoch": 127.83,
      "grad_norm": 2.203169107437134,
      "learning_rate": 1.4797104555409107e-06,
      "loss": 0.4586,
      "step": 312300
    },
    {
      "epoch": 127.84,
      "grad_norm": 1.8772791624069214,
      "learning_rate": 1.4795630009560641e-06,
      "loss": 0.4505,
      "step": 312310
    },
    {
      "epoch": 127.84,
      "grad_norm": 2.0752739906311035,
      "learning_rate": 1.4794155503299243e-06,
      "loss": 0.4529,
      "step": 312320
    },
    {
      "epoch": 127.85,
      "grad_norm": 2.288295030593872,
      "learning_rate": 1.4792681036631666e-06,
      "loss": 0.469,
      "step": 312330
    },
    {
      "epoch": 127.85,
      "grad_norm": 1.9084895849227905,
      "learning_rate": 1.4791206609564664e-06,
      "loss": 0.4654,
      "step": 312340
    },
    {
      "epoch": 127.86,
      "grad_norm": 2.008626699447632,
      "learning_rate": 1.4789732222104991e-06,
      "loss": 0.4605,
      "step": 312350
    },
    {
      "epoch": 127.86,
      "grad_norm": 1.7097506523132324,
      "learning_rate": 1.4788257874259409e-06,
      "loss": 0.4509,
      "step": 312360
    },
    {
      "epoch": 127.86,
      "grad_norm": 1.9003996849060059,
      "learning_rate": 1.4786783566034663e-06,
      "loss": 0.4436,
      "step": 312370
    },
    {
      "epoch": 127.87,
      "grad_norm": 1.733910322189331,
      "learning_rate": 1.4785309297437512e-06,
      "loss": 0.4746,
      "step": 312380
    },
    {
      "epoch": 127.87,
      "grad_norm": 1.9401236772537231,
      "learning_rate": 1.4783835068474704e-06,
      "loss": 0.485,
      "step": 312390
    },
    {
      "epoch": 127.88,
      "grad_norm": 1.7113887071609497,
      "learning_rate": 1.4782360879152996e-06,
      "loss": 0.4598,
      "step": 312400
    },
    {
      "epoch": 127.88,
      "grad_norm": 1.9827773571014404,
      "learning_rate": 1.478088672947914e-06,
      "loss": 0.4422,
      "step": 312410
    },
    {
      "epoch": 127.88,
      "grad_norm": 1.9241597652435303,
      "learning_rate": 1.4779412619459894e-06,
      "loss": 0.4689,
      "step": 312420
    },
    {
      "epoch": 127.89,
      "grad_norm": 2.042289972305298,
      "learning_rate": 1.4777938549101998e-06,
      "loss": 0.4716,
      "step": 312430
    },
    {
      "epoch": 127.89,
      "grad_norm": 2.035839080810547,
      "learning_rate": 1.4776464518412213e-06,
      "loss": 0.4594,
      "step": 312440
    },
    {
      "epoch": 127.9,
      "grad_norm": 1.6001758575439453,
      "learning_rate": 1.4774990527397293e-06,
      "loss": 0.4624,
      "step": 312450
    },
    {
      "epoch": 127.9,
      "grad_norm": 2.1406662464141846,
      "learning_rate": 1.4773516576063986e-06,
      "loss": 0.4638,
      "step": 312460
    },
    {
      "epoch": 127.9,
      "grad_norm": 1.989480972290039,
      "learning_rate": 1.4772042664419045e-06,
      "loss": 0.4403,
      "step": 312470
    },
    {
      "epoch": 127.91,
      "grad_norm": 1.9872833490371704,
      "learning_rate": 1.4770568792469221e-06,
      "loss": 0.4413,
      "step": 312480
    },
    {
      "epoch": 127.91,
      "grad_norm": 1.7736371755599976,
      "learning_rate": 1.476909496022127e-06,
      "loss": 0.4507,
      "step": 312490
    },
    {
      "epoch": 127.92,
      "grad_norm": 2.114583969116211,
      "learning_rate": 1.4767621167681939e-06,
      "loss": 0.4495,
      "step": 312500
    },
    {
      "epoch": 127.92,
      "grad_norm": 2.1769399642944336,
      "learning_rate": 1.476614741485798e-06,
      "loss": 0.4845,
      "step": 312510
    },
    {
      "epoch": 127.92,
      "grad_norm": 1.8688323497772217,
      "learning_rate": 1.476467370175614e-06,
      "loss": 0.4472,
      "step": 312520
    },
    {
      "epoch": 127.93,
      "grad_norm": 2.0336685180664062,
      "learning_rate": 1.476320002838318e-06,
      "loss": 0.4651,
      "step": 312530
    },
    {
      "epoch": 127.93,
      "grad_norm": 2.3772006034851074,
      "learning_rate": 1.4761726394745834e-06,
      "loss": 0.4556,
      "step": 312540
    },
    {
      "epoch": 127.94,
      "grad_norm": 1.7835357189178467,
      "learning_rate": 1.4760252800850874e-06,
      "loss": 0.4256,
      "step": 312550
    },
    {
      "epoch": 127.94,
      "grad_norm": 1.8595402240753174,
      "learning_rate": 1.4758779246705035e-06,
      "loss": 0.4678,
      "step": 312560
    },
    {
      "epoch": 127.95,
      "grad_norm": 1.7840425968170166,
      "learning_rate": 1.4757305732315074e-06,
      "loss": 0.4724,
      "step": 312570
    },
    {
      "epoch": 127.95,
      "grad_norm": 1.9441732168197632,
      "learning_rate": 1.475583225768774e-06,
      "loss": 0.4554,
      "step": 312580
    },
    {
      "epoch": 127.95,
      "grad_norm": 2.1207611560821533,
      "learning_rate": 1.4754358822829776e-06,
      "loss": 0.455,
      "step": 312590
    },
    {
      "epoch": 127.96,
      "grad_norm": 2.5676217079162598,
      "learning_rate": 1.475288542774794e-06,
      "loss": 0.4506,
      "step": 312600
    },
    {
      "epoch": 127.96,
      "grad_norm": 2.219935894012451,
      "learning_rate": 1.4751412072448981e-06,
      "loss": 0.4456,
      "step": 312610
    },
    {
      "epoch": 127.97,
      "grad_norm": 2.2816669940948486,
      "learning_rate": 1.4749938756939643e-06,
      "loss": 0.451,
      "step": 312620
    },
    {
      "epoch": 127.97,
      "grad_norm": 1.6679034233093262,
      "learning_rate": 1.4748465481226677e-06,
      "loss": 0.4545,
      "step": 312630
    },
    {
      "epoch": 127.97,
      "grad_norm": 1.8139259815216064,
      "learning_rate": 1.4746992245316838e-06,
      "loss": 0.4569,
      "step": 312640
    },
    {
      "epoch": 127.98,
      "grad_norm": 1.5277200937271118,
      "learning_rate": 1.4745519049216865e-06,
      "loss": 0.4595,
      "step": 312650
    },
    {
      "epoch": 127.98,
      "grad_norm": 1.660220980644226,
      "learning_rate": 1.474404589293351e-06,
      "loss": 0.4402,
      "step": 312660
    },
    {
      "epoch": 127.99,
      "grad_norm": 2.373453140258789,
      "learning_rate": 1.4742572776473523e-06,
      "loss": 0.4719,
      "step": 312670
    },
    {
      "epoch": 127.99,
      "grad_norm": 1.7415623664855957,
      "learning_rate": 1.4741099699843652e-06,
      "loss": 0.4622,
      "step": 312680
    },
    {
      "epoch": 127.99,
      "grad_norm": 2.3271827697753906,
      "learning_rate": 1.4739626663050639e-06,
      "loss": 0.4735,
      "step": 312690
    },
    {
      "epoch": 128.0,
      "grad_norm": 2.0514066219329834,
      "learning_rate": 1.473815366610124e-06,
      "loss": 0.4441,
      "step": 312700
    },
    {
      "epoch": 128.0,
      "eval_loss": 0.4588308036327362,
      "eval_runtime": 53.7675,
      "eval_samples_per_second": 64.147,
      "eval_steps_per_second": 8.035,
      "step": 312704
    },
    {
      "epoch": 128.0,
      "grad_norm": 2.166200876235962,
      "learning_rate": 1.47366807090022e-06,
      "loss": 0.4763,
      "step": 312710
    },
    {
      "epoch": 128.01,
      "grad_norm": 2.687467336654663,
      "learning_rate": 1.4735207791760266e-06,
      "loss": 0.4624,
      "step": 312720
    },
    {
      "epoch": 128.01,
      "grad_norm": 2.124342918395996,
      "learning_rate": 1.4733734914382178e-06,
      "loss": 0.4412,
      "step": 312730
    },
    {
      "epoch": 128.01,
      "grad_norm": 1.9229137897491455,
      "learning_rate": 1.4732262076874695e-06,
      "loss": 0.4488,
      "step": 312740
    },
    {
      "epoch": 128.02,
      "grad_norm": 2.2252891063690186,
      "learning_rate": 1.4730789279244558e-06,
      "loss": 0.4773,
      "step": 312750
    },
    {
      "epoch": 128.02,
      "grad_norm": 2.124274969100952,
      "learning_rate": 1.4729316521498517e-06,
      "loss": 0.4366,
      "step": 312760
    },
    {
      "epoch": 128.03,
      "grad_norm": 2.0904598236083984,
      "learning_rate": 1.4727843803643313e-06,
      "loss": 0.4827,
      "step": 312770
    },
    {
      "epoch": 128.03,
      "grad_norm": 2.095735788345337,
      "learning_rate": 1.4726371125685705e-06,
      "loss": 0.4395,
      "step": 312780
    },
    {
      "epoch": 128.04,
      "grad_norm": 1.5236380100250244,
      "learning_rate": 1.4724898487632418e-06,
      "loss": 0.4473,
      "step": 312790
    },
    {
      "epoch": 128.04,
      "grad_norm": 6.536238193511963,
      "learning_rate": 1.4723425889490212e-06,
      "loss": 0.4435,
      "step": 312800
    },
    {
      "epoch": 128.04,
      "grad_norm": 4.526971817016602,
      "learning_rate": 1.4721953331265831e-06,
      "loss": 0.4554,
      "step": 312810
    },
    {
      "epoch": 128.05,
      "grad_norm": 2.2532856464385986,
      "learning_rate": 1.4720480812966018e-06,
      "loss": 0.4475,
      "step": 312820
    },
    {
      "epoch": 128.05,
      "grad_norm": 1.9680432081222534,
      "learning_rate": 1.4719008334597523e-06,
      "loss": 0.4404,
      "step": 312830
    },
    {
      "epoch": 128.06,
      "grad_norm": 3.41121244430542,
      "learning_rate": 1.4717535896167083e-06,
      "loss": 0.4587,
      "step": 312840
    },
    {
      "epoch": 128.06,
      "grad_norm": 2.131505012512207,
      "learning_rate": 1.4716063497681452e-06,
      "loss": 0.4641,
      "step": 312850
    },
    {
      "epoch": 128.06,
      "grad_norm": 1.6310385465621948,
      "learning_rate": 1.4714591139147376e-06,
      "loss": 0.4984,
      "step": 312860
    },
    {
      "epoch": 128.07,
      "grad_norm": 1.9741812944412231,
      "learning_rate": 1.4713118820571587e-06,
      "loss": 0.4504,
      "step": 312870
    },
    {
      "epoch": 128.07,
      "grad_norm": 2.036665439605713,
      "learning_rate": 1.471164654196084e-06,
      "loss": 0.4573,
      "step": 312880
    },
    {
      "epoch": 128.08,
      "grad_norm": 1.9603965282440186,
      "learning_rate": 1.4710174303321874e-06,
      "loss": 0.4531,
      "step": 312890
    },
    {
      "epoch": 128.08,
      "grad_norm": 1.9114422798156738,
      "learning_rate": 1.4708702104661435e-06,
      "loss": 0.4502,
      "step": 312900
    },
    {
      "epoch": 128.08,
      "grad_norm": 2.0152676105499268,
      "learning_rate": 1.4707229945986263e-06,
      "loss": 0.4634,
      "step": 312910
    },
    {
      "epoch": 128.09,
      "grad_norm": 2.2841033935546875,
      "learning_rate": 1.4705757827303114e-06,
      "loss": 0.4469,
      "step": 312920
    },
    {
      "epoch": 128.09,
      "grad_norm": 1.8167264461517334,
      "learning_rate": 1.4704285748618722e-06,
      "loss": 0.4623,
      "step": 312930
    },
    {
      "epoch": 128.1,
      "grad_norm": 1.9811880588531494,
      "learning_rate": 1.470281370993983e-06,
      "loss": 0.4661,
      "step": 312940
    },
    {
      "epoch": 128.1,
      "grad_norm": 2.173203229904175,
      "learning_rate": 1.470134171127318e-06,
      "loss": 0.4776,
      "step": 312950
    },
    {
      "epoch": 128.1,
      "grad_norm": 2.650752067565918,
      "learning_rate": 1.4699869752625525e-06,
      "loss": 0.4409,
      "step": 312960
    },
    {
      "epoch": 128.11,
      "grad_norm": 2.065037488937378,
      "learning_rate": 1.46983978340036e-06,
      "loss": 0.4632,
      "step": 312970
    },
    {
      "epoch": 128.11,
      "grad_norm": 2.5398385524749756,
      "learning_rate": 1.4696925955414148e-06,
      "loss": 0.4481,
      "step": 312980
    },
    {
      "epoch": 128.12,
      "grad_norm": 2.081876277923584,
      "learning_rate": 1.469545411686391e-06,
      "loss": 0.442,
      "step": 312990
    },
    {
      "epoch": 128.12,
      "grad_norm": 2.0161192417144775,
      "learning_rate": 1.4693982318359634e-06,
      "loss": 0.4384,
      "step": 313000
    },
    {
      "epoch": 128.13,
      "grad_norm": 1.9687713384628296,
      "learning_rate": 1.4692510559908059e-06,
      "loss": 0.458,
      "step": 313010
    },
    {
      "epoch": 128.13,
      "grad_norm": 1.794577717781067,
      "learning_rate": 1.469103884151592e-06,
      "loss": 0.4377,
      "step": 313020
    },
    {
      "epoch": 128.13,
      "grad_norm": 1.9161251783370972,
      "learning_rate": 1.4689567163189968e-06,
      "loss": 0.4625,
      "step": 313030
    },
    {
      "epoch": 128.14,
      "grad_norm": 2.486745595932007,
      "learning_rate": 1.4688095524936942e-06,
      "loss": 0.4462,
      "step": 313040
    },
    {
      "epoch": 128.14,
      "grad_norm": 2.244882106781006,
      "learning_rate": 1.4686623926763583e-06,
      "loss": 0.4742,
      "step": 313050
    },
    {
      "epoch": 128.15,
      "grad_norm": 1.7330399751663208,
      "learning_rate": 1.4685152368676633e-06,
      "loss": 0.4631,
      "step": 313060
    },
    {
      "epoch": 128.15,
      "grad_norm": 1.9285081624984741,
      "learning_rate": 1.4683680850682829e-06,
      "loss": 0.4595,
      "step": 313070
    },
    {
      "epoch": 128.15,
      "grad_norm": 2.32856822013855,
      "learning_rate": 1.4682209372788915e-06,
      "loss": 0.4694,
      "step": 313080
    },
    {
      "epoch": 128.16,
      "grad_norm": 1.9352339506149292,
      "learning_rate": 1.4680737935001637e-06,
      "loss": 0.4762,
      "step": 313090
    },
    {
      "epoch": 128.16,
      "grad_norm": 2.2723546028137207,
      "learning_rate": 1.4679266537327718e-06,
      "loss": 0.4569,
      "step": 313100
    },
    {
      "epoch": 128.17,
      "grad_norm": 1.8614825010299683,
      "learning_rate": 1.4677795179773916e-06,
      "loss": 0.4555,
      "step": 313110
    },
    {
      "epoch": 128.17,
      "grad_norm": 2.304433822631836,
      "learning_rate": 1.4676323862346966e-06,
      "loss": 0.451,
      "step": 313120
    },
    {
      "epoch": 128.17,
      "grad_norm": 2.0360825061798096,
      "learning_rate": 1.467485258505361e-06,
      "loss": 0.4488,
      "step": 313130
    },
    {
      "epoch": 128.18,
      "grad_norm": 2.047667980194092,
      "learning_rate": 1.4673381347900587e-06,
      "loss": 0.4559,
      "step": 313140
    },
    {
      "epoch": 128.18,
      "grad_norm": 1.8446286916732788,
      "learning_rate": 1.4671910150894632e-06,
      "loss": 0.4475,
      "step": 313150
    },
    {
      "epoch": 128.19,
      "grad_norm": 1.806957721710205,
      "learning_rate": 1.4670438994042482e-06,
      "loss": 0.4493,
      "step": 313160
    },
    {
      "epoch": 128.19,
      "grad_norm": 2.249134063720703,
      "learning_rate": 1.4668967877350886e-06,
      "loss": 0.4584,
      "step": 313170
    },
    {
      "epoch": 128.19,
      "grad_norm": 1.9857763051986694,
      "learning_rate": 1.4667496800826572e-06,
      "loss": 0.4364,
      "step": 313180
    },
    {
      "epoch": 128.2,
      "grad_norm": 1.9611713886260986,
      "learning_rate": 1.4666025764476288e-06,
      "loss": 0.459,
      "step": 313190
    },
    {
      "epoch": 128.2,
      "grad_norm": 1.4745899438858032,
      "learning_rate": 1.466455476830677e-06,
      "loss": 0.4575,
      "step": 313200
    },
    {
      "epoch": 128.21,
      "grad_norm": 2.531290054321289,
      "learning_rate": 1.4663083812324752e-06,
      "loss": 0.4487,
      "step": 313210
    },
    {
      "epoch": 128.21,
      "grad_norm": 1.8242089748382568,
      "learning_rate": 1.4661612896536984e-06,
      "loss": 0.4632,
      "step": 313220
    },
    {
      "epoch": 128.22,
      "grad_norm": 1.6353576183319092,
      "learning_rate": 1.466014202095019e-06,
      "loss": 0.4482,
      "step": 313230
    },
    {
      "epoch": 128.22,
      "grad_norm": 2.150022029876709,
      "learning_rate": 1.465867118557111e-06,
      "loss": 0.4782,
      "step": 313240
    },
    {
      "epoch": 128.22,
      "grad_norm": 1.8522789478302002,
      "learning_rate": 1.465720039040649e-06,
      "loss": 0.4451,
      "step": 313250
    },
    {
      "epoch": 128.23,
      "grad_norm": 1.8451341390609741,
      "learning_rate": 1.465572963546306e-06,
      "loss": 0.4505,
      "step": 313260
    },
    {
      "epoch": 128.23,
      "grad_norm": 2.7596099376678467,
      "learning_rate": 1.4654258920747563e-06,
      "loss": 0.4507,
      "step": 313270
    },
    {
      "epoch": 128.24,
      "grad_norm": 2.3830277919769287,
      "learning_rate": 1.465278824626673e-06,
      "loss": 0.4602,
      "step": 313280
    },
    {
      "epoch": 128.24,
      "grad_norm": 1.9459470510482788,
      "learning_rate": 1.4651317612027296e-06,
      "loss": 0.4579,
      "step": 313290
    },
    {
      "epoch": 128.24,
      "grad_norm": 1.732257604598999,
      "learning_rate": 1.4649847018036012e-06,
      "loss": 0.4567,
      "step": 313300
    },
    {
      "epoch": 128.25,
      "grad_norm": 1.8315162658691406,
      "learning_rate": 1.4648376464299598e-06,
      "loss": 0.4511,
      "step": 313310
    },
    {
      "epoch": 128.25,
      "grad_norm": 1.8001642227172852,
      "learning_rate": 1.4646905950824803e-06,
      "loss": 0.4876,
      "step": 313320
    },
    {
      "epoch": 128.26,
      "grad_norm": 1.8537098169326782,
      "learning_rate": 1.4645435477618358e-06,
      "loss": 0.4625,
      "step": 313330
    },
    {
      "epoch": 128.26,
      "grad_norm": 1.5875320434570312,
      "learning_rate": 1.4643965044686997e-06,
      "loss": 0.4505,
      "step": 313340
    },
    {
      "epoch": 128.26,
      "grad_norm": 1.941834568977356,
      "learning_rate": 1.464249465203746e-06,
      "loss": 0.4548,
      "step": 313350
    },
    {
      "epoch": 128.27,
      "grad_norm": 1.9620616436004639,
      "learning_rate": 1.4641024299676482e-06,
      "loss": 0.4695,
      "step": 313360
    },
    {
      "epoch": 128.27,
      "grad_norm": 1.9338715076446533,
      "learning_rate": 1.4639553987610795e-06,
      "loss": 0.4526,
      "step": 313370
    },
    {
      "epoch": 128.28,
      "grad_norm": 1.8453038930892944,
      "learning_rate": 1.4638083715847135e-06,
      "loss": 0.4828,
      "step": 313380
    },
    {
      "epoch": 128.28,
      "grad_norm": 2.44503116607666,
      "learning_rate": 1.4636613484392239e-06,
      "loss": 0.4814,
      "step": 313390
    },
    {
      "epoch": 128.28,
      "grad_norm": 1.9009172916412354,
      "learning_rate": 1.463514329325284e-06,
      "loss": 0.45,
      "step": 313400
    },
    {
      "epoch": 128.29,
      "grad_norm": 2.1706910133361816,
      "learning_rate": 1.4633673142435673e-06,
      "loss": 0.4347,
      "step": 313410
    },
    {
      "epoch": 128.29,
      "grad_norm": 1.9479739665985107,
      "learning_rate": 1.4632203031947474e-06,
      "loss": 0.4576,
      "step": 313420
    },
    {
      "epoch": 128.3,
      "grad_norm": 2.4967410564422607,
      "learning_rate": 1.463073296179498e-06,
      "loss": 0.4679,
      "step": 313430
    },
    {
      "epoch": 128.3,
      "grad_norm": 2.057121992111206,
      "learning_rate": 1.4629262931984918e-06,
      "loss": 0.4671,
      "step": 313440
    },
    {
      "epoch": 128.31,
      "grad_norm": 1.6389000415802002,
      "learning_rate": 1.4627792942524032e-06,
      "loss": 0.4674,
      "step": 313450
    },
    {
      "epoch": 128.31,
      "grad_norm": 2.081742525100708,
      "learning_rate": 1.4626322993419042e-06,
      "loss": 0.4688,
      "step": 313460
    },
    {
      "epoch": 128.31,
      "grad_norm": 2.073402166366577,
      "learning_rate": 1.4624853084676694e-06,
      "loss": 0.4576,
      "step": 313470
    },
    {
      "epoch": 128.32,
      "grad_norm": 2.045564889907837,
      "learning_rate": 1.4623383216303705e-06,
      "loss": 0.4814,
      "step": 313480
    },
    {
      "epoch": 128.32,
      "grad_norm": 1.8919440507888794,
      "learning_rate": 1.4621913388306832e-06,
      "loss": 0.4569,
      "step": 313490
    },
    {
      "epoch": 128.33,
      "grad_norm": 1.7990241050720215,
      "learning_rate": 1.4620443600692798e-06,
      "loss": 0.4516,
      "step": 313500
    },
    {
      "epoch": 128.33,
      "grad_norm": 2.0047504901885986,
      "learning_rate": 1.461897385346833e-06,
      "loss": 0.4479,
      "step": 313510
    },
    {
      "epoch": 128.33,
      "grad_norm": 2.5771992206573486,
      "learning_rate": 1.4617504146640163e-06,
      "loss": 0.4497,
      "step": 313520
    },
    {
      "epoch": 128.34,
      "grad_norm": 1.9642889499664307,
      "learning_rate": 1.4616034480215031e-06,
      "loss": 0.4386,
      "step": 313530
    },
    {
      "epoch": 128.34,
      "grad_norm": 1.8364177942276,
      "learning_rate": 1.461456485419967e-06,
      "loss": 0.4509,
      "step": 313540
    },
    {
      "epoch": 128.35,
      "grad_norm": 2.1198766231536865,
      "learning_rate": 1.4613095268600806e-06,
      "loss": 0.4657,
      "step": 313550
    },
    {
      "epoch": 128.35,
      "grad_norm": 2.176375150680542,
      "learning_rate": 1.4611625723425173e-06,
      "loss": 0.4624,
      "step": 313560
    },
    {
      "epoch": 128.35,
      "grad_norm": 2.025269031524658,
      "learning_rate": 1.4610156218679505e-06,
      "loss": 0.451,
      "step": 313570
    },
    {
      "epoch": 128.36,
      "grad_norm": 2.2129907608032227,
      "learning_rate": 1.4608686754370534e-06,
      "loss": 0.4612,
      "step": 313580
    },
    {
      "epoch": 128.36,
      "grad_norm": 1.8357070684432983,
      "learning_rate": 1.4607217330504984e-06,
      "loss": 0.4445,
      "step": 313590
    },
    {
      "epoch": 128.37,
      "grad_norm": 4.053933620452881,
      "learning_rate": 1.4605747947089595e-06,
      "loss": 0.4495,
      "step": 313600
    },
    {
      "epoch": 128.37,
      "grad_norm": 1.9587210416793823,
      "learning_rate": 1.4604278604131093e-06,
      "loss": 0.4665,
      "step": 313610
    },
    {
      "epoch": 128.37,
      "grad_norm": 1.9626561403274536,
      "learning_rate": 1.4602809301636208e-06,
      "loss": 0.4738,
      "step": 313620
    },
    {
      "epoch": 128.38,
      "grad_norm": 2.0498905181884766,
      "learning_rate": 1.4601340039611675e-06,
      "loss": 0.4593,
      "step": 313630
    },
    {
      "epoch": 128.38,
      "grad_norm": 2.105289936065674,
      "learning_rate": 1.459987081806422e-06,
      "loss": 0.4734,
      "step": 313640
    },
    {
      "epoch": 128.39,
      "grad_norm": 1.820876121520996,
      "learning_rate": 1.459840163700058e-06,
      "loss": 0.4382,
      "step": 313650
    },
    {
      "epoch": 128.39,
      "grad_norm": 2.4931225776672363,
      "learning_rate": 1.4596932496427478e-06,
      "loss": 0.4644,
      "step": 313660
    },
    {
      "epoch": 128.4,
      "grad_norm": 1.9938311576843262,
      "learning_rate": 1.4595463396351642e-06,
      "loss": 0.4497,
      "step": 313670
    },
    {
      "epoch": 128.4,
      "grad_norm": 2.4138500690460205,
      "learning_rate": 1.4593994336779807e-06,
      "loss": 0.4407,
      "step": 313680
    },
    {
      "epoch": 128.4,
      "grad_norm": 2.1399669647216797,
      "learning_rate": 1.459252531771871e-06,
      "loss": 0.4683,
      "step": 313690
    },
    {
      "epoch": 128.41,
      "grad_norm": 2.200915575027466,
      "learning_rate": 1.4591056339175069e-06,
      "loss": 0.4514,
      "step": 313700
    },
    {
      "epoch": 128.41,
      "grad_norm": 1.8417638540267944,
      "learning_rate": 1.4589587401155613e-06,
      "loss": 0.4771,
      "step": 313710
    },
    {
      "epoch": 128.42,
      "grad_norm": 1.9927598237991333,
      "learning_rate": 1.4588118503667082e-06,
      "loss": 0.4538,
      "step": 313720
    },
    {
      "epoch": 128.42,
      "grad_norm": 2.173825979232788,
      "learning_rate": 1.458664964671619e-06,
      "loss": 0.4517,
      "step": 313730
    },
    {
      "epoch": 128.42,
      "grad_norm": 2.25764799118042,
      "learning_rate": 1.4585180830309679e-06,
      "loss": 0.4494,
      "step": 313740
    },
    {
      "epoch": 128.43,
      "grad_norm": 1.7518296241760254,
      "learning_rate": 1.4583712054454264e-06,
      "loss": 0.4507,
      "step": 313750
    },
    {
      "epoch": 128.43,
      "grad_norm": 2.079390287399292,
      "learning_rate": 1.4582243319156684e-06,
      "loss": 0.4752,
      "step": 313760
    },
    {
      "epoch": 128.44,
      "grad_norm": 1.52434504032135,
      "learning_rate": 1.4580774624423664e-06,
      "loss": 0.4487,
      "step": 313770
    },
    {
      "epoch": 128.44,
      "grad_norm": 2.1057803630828857,
      "learning_rate": 1.4579305970261928e-06,
      "loss": 0.4648,
      "step": 313780
    },
    {
      "epoch": 128.44,
      "grad_norm": 1.8500727415084839,
      "learning_rate": 1.457783735667821e-06,
      "loss": 0.4507,
      "step": 313790
    },
    {
      "epoch": 128.45,
      "grad_norm": 1.6961017847061157,
      "learning_rate": 1.4576368783679233e-06,
      "loss": 0.4604,
      "step": 313800
    },
    {
      "epoch": 128.45,
      "grad_norm": 1.7993470430374146,
      "learning_rate": 1.4574900251271733e-06,
      "loss": 0.4467,
      "step": 313810
    },
    {
      "epoch": 128.46,
      "grad_norm": 1.8957619667053223,
      "learning_rate": 1.4573431759462423e-06,
      "loss": 0.45,
      "step": 313820
    },
    {
      "epoch": 128.46,
      "grad_norm": 1.7679325342178345,
      "learning_rate": 1.4571963308258036e-06,
      "loss": 0.4613,
      "step": 313830
    },
    {
      "epoch": 128.47,
      "grad_norm": 1.7320362329483032,
      "learning_rate": 1.4570494897665301e-06,
      "loss": 0.455,
      "step": 313840
    },
    {
      "epoch": 128.47,
      "grad_norm": 1.8761470317840576,
      "learning_rate": 1.4569026527690943e-06,
      "loss": 0.4768,
      "step": 313850
    },
    {
      "epoch": 128.47,
      "grad_norm": 1.9243980646133423,
      "learning_rate": 1.4567558198341681e-06,
      "loss": 0.4581,
      "step": 313860
    },
    {
      "epoch": 128.48,
      "grad_norm": 2.24356746673584,
      "learning_rate": 1.4566089909624259e-06,
      "loss": 0.4478,
      "step": 313870
    },
    {
      "epoch": 128.48,
      "grad_norm": 1.9744229316711426,
      "learning_rate": 1.456462166154539e-06,
      "loss": 0.4587,
      "step": 313880
    },
    {
      "epoch": 128.49,
      "grad_norm": 1.8603565692901611,
      "learning_rate": 1.45631534541118e-06,
      "loss": 0.4398,
      "step": 313890
    },
    {
      "epoch": 128.49,
      "grad_norm": 1.6646933555603027,
      "learning_rate": 1.4561685287330222e-06,
      "loss": 0.4463,
      "step": 313900
    },
    {
      "epoch": 128.49,
      "grad_norm": 2.068211078643799,
      "learning_rate": 1.4560217161207373e-06,
      "loss": 0.4441,
      "step": 313910
    },
    {
      "epoch": 128.5,
      "grad_norm": 1.6378403902053833,
      "learning_rate": 1.4558749075749984e-06,
      "loss": 0.4759,
      "step": 313920
    },
    {
      "epoch": 128.5,
      "grad_norm": 4.53336238861084,
      "learning_rate": 1.4557281030964776e-06,
      "loss": 0.4619,
      "step": 313930
    },
    {
      "epoch": 128.51,
      "grad_norm": 1.9940158128738403,
      "learning_rate": 1.4555813026858481e-06,
      "loss": 0.4501,
      "step": 313940
    },
    {
      "epoch": 128.51,
      "grad_norm": 1.7773467302322388,
      "learning_rate": 1.4554345063437813e-06,
      "loss": 0.4424,
      "step": 313950
    },
    {
      "epoch": 128.51,
      "grad_norm": 1.897045373916626,
      "learning_rate": 1.4552877140709504e-06,
      "loss": 0.4297,
      "step": 313960
    },
    {
      "epoch": 128.52,
      "grad_norm": 1.7713450193405151,
      "learning_rate": 1.4551409258680277e-06,
      "loss": 0.4564,
      "step": 313970
    },
    {
      "epoch": 128.52,
      "grad_norm": 2.0324912071228027,
      "learning_rate": 1.454994141735685e-06,
      "loss": 0.4598,
      "step": 313980
    },
    {
      "epoch": 128.53,
      "grad_norm": 1.858005404472351,
      "learning_rate": 1.4548473616745959e-06,
      "loss": 0.4857,
      "step": 313990
    },
    {
      "epoch": 128.53,
      "grad_norm": 2.1016955375671387,
      "learning_rate": 1.454700585685432e-06,
      "loss": 0.4692,
      "step": 314000
    },
    {
      "epoch": 128.53,
      "grad_norm": 1.9565287828445435,
      "learning_rate": 1.4545538137688654e-06,
      "loss": 0.4683,
      "step": 314010
    },
    {
      "epoch": 128.54,
      "grad_norm": 1.8555585145950317,
      "learning_rate": 1.454407045925569e-06,
      "loss": 0.4619,
      "step": 314020
    },
    {
      "epoch": 128.54,
      "grad_norm": 1.6501256227493286,
      "learning_rate": 1.4542602821562157e-06,
      "loss": 0.4392,
      "step": 314030
    },
    {
      "epoch": 128.55,
      "grad_norm": 1.6897149085998535,
      "learning_rate": 1.4541135224614762e-06,
      "loss": 0.4689,
      "step": 314040
    },
    {
      "epoch": 128.55,
      "grad_norm": 2.1157169342041016,
      "learning_rate": 1.453966766842023e-06,
      "loss": 0.4493,
      "step": 314050
    },
    {
      "epoch": 128.56,
      "grad_norm": 1.6984518766403198,
      "learning_rate": 1.4538200152985298e-06,
      "loss": 0.4717,
      "step": 314060
    },
    {
      "epoch": 128.56,
      "grad_norm": 1.7902852296829224,
      "learning_rate": 1.4536732678316678e-06,
      "loss": 0.4606,
      "step": 314070
    },
    {
      "epoch": 128.56,
      "grad_norm": 1.9599573612213135,
      "learning_rate": 1.4535265244421103e-06,
      "loss": 0.4558,
      "step": 314080
    },
    {
      "epoch": 128.57,
      "grad_norm": 1.7999578714370728,
      "learning_rate": 1.453379785130528e-06,
      "loss": 0.4203,
      "step": 314090
    },
    {
      "epoch": 128.57,
      "grad_norm": 2.285162925720215,
      "learning_rate": 1.4532330498975939e-06,
      "loss": 0.4498,
      "step": 314100
    },
    {
      "epoch": 128.58,
      "grad_norm": 2.1644797325134277,
      "learning_rate": 1.45308631874398e-06,
      "loss": 0.4696,
      "step": 314110
    },
    {
      "epoch": 128.58,
      "grad_norm": 2.3896162509918213,
      "learning_rate": 1.4529395916703582e-06,
      "loss": 0.4607,
      "step": 314120
    },
    {
      "epoch": 128.58,
      "grad_norm": 1.749963402748108,
      "learning_rate": 1.4527928686774013e-06,
      "loss": 0.4605,
      "step": 314130
    },
    {
      "epoch": 128.59,
      "grad_norm": 1.5880261659622192,
      "learning_rate": 1.452646149765781e-06,
      "loss": 0.465,
      "step": 314140
    },
    {
      "epoch": 128.59,
      "grad_norm": 2.072308301925659,
      "learning_rate": 1.4524994349361691e-06,
      "loss": 0.4769,
      "step": 314150
    },
    {
      "epoch": 128.6,
      "grad_norm": 1.965497374534607,
      "learning_rate": 1.452352724189239e-06,
      "loss": 0.4687,
      "step": 314160
    },
    {
      "epoch": 128.6,
      "grad_norm": 2.407855272293091,
      "learning_rate": 1.452206017525661e-06,
      "loss": 0.4735,
      "step": 314170
    },
    {
      "epoch": 128.6,
      "grad_norm": 1.6203081607818604,
      "learning_rate": 1.4520593149461077e-06,
      "loss": 0.4618,
      "step": 314180
    },
    {
      "epoch": 128.61,
      "grad_norm": 2.0197274684906006,
      "learning_rate": 1.4519126164512516e-06,
      "loss": 0.4325,
      "step": 314190
    },
    {
      "epoch": 128.61,
      "grad_norm": 1.9147982597351074,
      "learning_rate": 1.451765922041764e-06,
      "loss": 0.4409,
      "step": 314200
    },
    {
      "epoch": 128.62,
      "grad_norm": 1.6792229413986206,
      "learning_rate": 1.4516192317183182e-06,
      "loss": 0.4435,
      "step": 314210
    },
    {
      "epoch": 128.62,
      "grad_norm": 1.7870625257492065,
      "learning_rate": 1.4514725454815846e-06,
      "loss": 0.4353,
      "step": 314220
    },
    {
      "epoch": 128.62,
      "grad_norm": 2.0099151134490967,
      "learning_rate": 1.4513258633322356e-06,
      "loss": 0.4571,
      "step": 314230
    },
    {
      "epoch": 128.63,
      "grad_norm": 2.204389810562134,
      "learning_rate": 1.4511791852709436e-06,
      "loss": 0.4663,
      "step": 314240
    },
    {
      "epoch": 128.63,
      "grad_norm": 1.6526331901550293,
      "learning_rate": 1.4510325112983807e-06,
      "loss": 0.4502,
      "step": 314250
    },
    {
      "epoch": 128.64,
      "grad_norm": 2.326122522354126,
      "learning_rate": 1.450885841415218e-06,
      "loss": 0.4656,
      "step": 314260
    },
    {
      "epoch": 128.64,
      "grad_norm": 2.4041223526000977,
      "learning_rate": 1.450739175622128e-06,
      "loss": 0.4586,
      "step": 314270
    },
    {
      "epoch": 128.65,
      "grad_norm": 2.1103806495666504,
      "learning_rate": 1.450592513919782e-06,
      "loss": 0.4423,
      "step": 314280
    },
    {
      "epoch": 128.65,
      "grad_norm": 1.936008095741272,
      "learning_rate": 1.4504458563088524e-06,
      "loss": 0.4718,
      "step": 314290
    },
    {
      "epoch": 128.65,
      "grad_norm": 2.0504894256591797,
      "learning_rate": 1.450299202790011e-06,
      "loss": 0.439,
      "step": 314300
    },
    {
      "epoch": 128.66,
      "grad_norm": 1.9794683456420898,
      "learning_rate": 1.4501525533639288e-06,
      "loss": 0.4521,
      "step": 314310
    },
    {
      "epoch": 128.66,
      "grad_norm": 2.032919406890869,
      "learning_rate": 1.4500059080312783e-06,
      "loss": 0.4478,
      "step": 314320
    },
    {
      "epoch": 128.67,
      "grad_norm": 2.0937070846557617,
      "learning_rate": 1.4498592667927309e-06,
      "loss": 0.45,
      "step": 314330
    },
    {
      "epoch": 128.67,
      "grad_norm": 1.8873653411865234,
      "learning_rate": 1.4497126296489586e-06,
      "loss": 0.4455,
      "step": 314340
    },
    {
      "epoch": 128.67,
      "grad_norm": 2.170370101928711,
      "learning_rate": 1.4495659966006331e-06,
      "loss": 0.486,
      "step": 314350
    },
    {
      "epoch": 128.68,
      "grad_norm": 1.6359823942184448,
      "learning_rate": 1.449419367648426e-06,
      "loss": 0.4672,
      "step": 314360
    },
    {
      "epoch": 128.68,
      "grad_norm": 2.0702786445617676,
      "learning_rate": 1.449272742793009e-06,
      "loss": 0.4484,
      "step": 314370
    },
    {
      "epoch": 128.69,
      "grad_norm": 1.8440649509429932,
      "learning_rate": 1.449126122035054e-06,
      "loss": 0.452,
      "step": 314380
    },
    {
      "epoch": 128.69,
      "grad_norm": 1.8526811599731445,
      "learning_rate": 1.4489795053752325e-06,
      "loss": 0.4596,
      "step": 314390
    },
    {
      "epoch": 128.69,
      "grad_norm": 1.8522051572799683,
      "learning_rate": 1.4488328928142158e-06,
      "loss": 0.4583,
      "step": 314400
    },
    {
      "epoch": 128.7,
      "grad_norm": 1.7488731145858765,
      "learning_rate": 1.4486862843526757e-06,
      "loss": 0.4359,
      "step": 314410
    },
    {
      "epoch": 128.7,
      "grad_norm": 2.026364803314209,
      "learning_rate": 1.4485396799912831e-06,
      "loss": 0.4776,
      "step": 314420
    },
    {
      "epoch": 128.71,
      "grad_norm": 1.820672631263733,
      "learning_rate": 1.4483930797307111e-06,
      "loss": 0.4664,
      "step": 314430
    },
    {
      "epoch": 128.71,
      "grad_norm": 2.1611056327819824,
      "learning_rate": 1.448246483571631e-06,
      "loss": 0.4607,
      "step": 314440
    },
    {
      "epoch": 128.71,
      "grad_norm": 2.263852834701538,
      "learning_rate": 1.4480998915147134e-06,
      "loss": 0.464,
      "step": 314450
    },
    {
      "epoch": 128.72,
      "grad_norm": 2.2850120067596436,
      "learning_rate": 1.4479533035606304e-06,
      "loss": 0.4578,
      "step": 314460
    },
    {
      "epoch": 128.72,
      "grad_norm": 1.9955337047576904,
      "learning_rate": 1.447806719710053e-06,
      "loss": 0.4511,
      "step": 314470
    },
    {
      "epoch": 128.73,
      "grad_norm": 1.9709501266479492,
      "learning_rate": 1.4476601399636536e-06,
      "loss": 0.4434,
      "step": 314480
    },
    {
      "epoch": 128.73,
      "grad_norm": 1.905052661895752,
      "learning_rate": 1.4475135643221023e-06,
      "loss": 0.4539,
      "step": 314490
    },
    {
      "epoch": 128.74,
      "grad_norm": 2.16815447807312,
      "learning_rate": 1.4473669927860716e-06,
      "loss": 0.4533,
      "step": 314500
    },
    {
      "epoch": 128.74,
      "grad_norm": 1.7143516540527344,
      "learning_rate": 1.4472204253562328e-06,
      "loss": 0.4511,
      "step": 314510
    },
    {
      "epoch": 128.74,
      "grad_norm": 2.0893938541412354,
      "learning_rate": 1.4470738620332574e-06,
      "loss": 0.4491,
      "step": 314520
    },
    {
      "epoch": 128.75,
      "grad_norm": 2.087613344192505,
      "learning_rate": 1.446927302817816e-06,
      "loss": 0.4456,
      "step": 314530
    },
    {
      "epoch": 128.75,
      "grad_norm": 1.7206737995147705,
      "learning_rate": 1.4467807477105808e-06,
      "loss": 0.4632,
      "step": 314540
    },
    {
      "epoch": 128.76,
      "grad_norm": 2.586899757385254,
      "learning_rate": 1.4466341967122228e-06,
      "loss": 0.4414,
      "step": 314550
    },
    {
      "epoch": 128.76,
      "grad_norm": 1.9733169078826904,
      "learning_rate": 1.4464876498234132e-06,
      "loss": 0.4795,
      "step": 314560
    },
    {
      "epoch": 128.76,
      "grad_norm": 2.332749605178833,
      "learning_rate": 1.4463411070448232e-06,
      "loss": 0.4758,
      "step": 314570
    },
    {
      "epoch": 128.77,
      "grad_norm": 1.7534971237182617,
      "learning_rate": 1.4461945683771247e-06,
      "loss": 0.4631,
      "step": 314580
    },
    {
      "epoch": 128.77,
      "grad_norm": 1.849547266960144,
      "learning_rate": 1.4460480338209887e-06,
      "loss": 0.4718,
      "step": 314590
    },
    {
      "epoch": 128.78,
      "grad_norm": 2.466599225997925,
      "learning_rate": 1.4459015033770862e-06,
      "loss": 0.4532,
      "step": 314600
    },
    {
      "epoch": 128.78,
      "grad_norm": 2.229703664779663,
      "learning_rate": 1.4457549770460882e-06,
      "loss": 0.4768,
      "step": 314610
    },
    {
      "epoch": 128.78,
      "grad_norm": 1.8020716905593872,
      "learning_rate": 1.4456084548286669e-06,
      "loss": 0.4542,
      "step": 314620
    },
    {
      "epoch": 128.79,
      "grad_norm": 1.9552490711212158,
      "learning_rate": 1.4454619367254928e-06,
      "loss": 0.453,
      "step": 314630
    },
    {
      "epoch": 128.79,
      "grad_norm": 2.084498643875122,
      "learning_rate": 1.4453154227372371e-06,
      "loss": 0.4552,
      "step": 314640
    },
    {
      "epoch": 128.8,
      "grad_norm": 1.7713031768798828,
      "learning_rate": 1.4451689128645714e-06,
      "loss": 0.455,
      "step": 314650
    },
    {
      "epoch": 128.8,
      "grad_norm": 2.032721519470215,
      "learning_rate": 1.4450224071081667e-06,
      "loss": 0.4613,
      "step": 314660
    },
    {
      "epoch": 128.8,
      "grad_norm": 2.0122389793395996,
      "learning_rate": 1.4448759054686935e-06,
      "loss": 0.4646,
      "step": 314670
    },
    {
      "epoch": 128.81,
      "grad_norm": 2.0618395805358887,
      "learning_rate": 1.4447294079468231e-06,
      "loss": 0.4541,
      "step": 314680
    },
    {
      "epoch": 128.81,
      "grad_norm": 1.9505809545516968,
      "learning_rate": 1.4445829145432272e-06,
      "loss": 0.4692,
      "step": 314690
    },
    {
      "epoch": 128.82,
      "grad_norm": 2.433690071105957,
      "learning_rate": 1.4444364252585763e-06,
      "loss": 0.455,
      "step": 314700
    },
    {
      "epoch": 128.82,
      "grad_norm": 1.8024944067001343,
      "learning_rate": 1.4442899400935417e-06,
      "loss": 0.4687,
      "step": 314710
    },
    {
      "epoch": 128.83,
      "grad_norm": 1.796586036682129,
      "learning_rate": 1.4441434590487942e-06,
      "loss": 0.4573,
      "step": 314720
    },
    {
      "epoch": 128.83,
      "grad_norm": 1.8707599639892578,
      "learning_rate": 1.443996982125005e-06,
      "loss": 0.4801,
      "step": 314730
    },
    {
      "epoch": 128.83,
      "grad_norm": 1.9093199968338013,
      "learning_rate": 1.443850509322845e-06,
      "loss": 0.4465,
      "step": 314740
    },
    {
      "epoch": 128.84,
      "grad_norm": 1.7351435422897339,
      "learning_rate": 1.4437040406429856e-06,
      "loss": 0.4565,
      "step": 314750
    },
    {
      "epoch": 128.84,
      "grad_norm": 1.994268536567688,
      "learning_rate": 1.4435575760860968e-06,
      "loss": 0.4678,
      "step": 314760
    },
    {
      "epoch": 128.85,
      "grad_norm": 1.7905066013336182,
      "learning_rate": 1.4434111156528504e-06,
      "loss": 0.488,
      "step": 314770
    },
    {
      "epoch": 128.85,
      "grad_norm": 1.5567563772201538,
      "learning_rate": 1.4432646593439168e-06,
      "loss": 0.4375,
      "step": 314780
    },
    {
      "epoch": 128.85,
      "grad_norm": 1.6728707551956177,
      "learning_rate": 1.443118207159967e-06,
      "loss": 0.4566,
      "step": 314790
    },
    {
      "epoch": 128.86,
      "grad_norm": 1.6204568147659302,
      "learning_rate": 1.4429717591016716e-06,
      "loss": 0.4568,
      "step": 314800
    },
    {
      "epoch": 128.86,
      "grad_norm": 1.9278578758239746,
      "learning_rate": 1.4428253151697022e-06,
      "loss": 0.4786,
      "step": 314810
    },
    {
      "epoch": 128.87,
      "grad_norm": 2.323413372039795,
      "learning_rate": 1.4426788753647293e-06,
      "loss": 0.4575,
      "step": 314820
    },
    {
      "epoch": 128.87,
      "grad_norm": 2.2205886840820312,
      "learning_rate": 1.4425324396874238e-06,
      "loss": 0.4711,
      "step": 314830
    },
    {
      "epoch": 128.87,
      "grad_norm": 1.8239425420761108,
      "learning_rate": 1.4423860081384564e-06,
      "loss": 0.4565,
      "step": 314840
    },
    {
      "epoch": 128.88,
      "grad_norm": 1.8851064443588257,
      "learning_rate": 1.4422395807184975e-06,
      "loss": 0.4601,
      "step": 314850
    },
    {
      "epoch": 128.88,
      "grad_norm": 1.7910809516906738,
      "learning_rate": 1.4420931574282185e-06,
      "loss": 0.4786,
      "step": 314860
    },
    {
      "epoch": 128.89,
      "grad_norm": 2.692305326461792,
      "learning_rate": 1.44194673826829e-06,
      "loss": 0.4589,
      "step": 314870
    },
    {
      "epoch": 128.89,
      "grad_norm": 2.038853883743286,
      "learning_rate": 1.4418003232393829e-06,
      "loss": 0.4487,
      "step": 314880
    },
    {
      "epoch": 128.89,
      "grad_norm": 2.194715976715088,
      "learning_rate": 1.441653912342167e-06,
      "loss": 0.4657,
      "step": 314890
    },
    {
      "epoch": 128.9,
      "grad_norm": 1.6859484910964966,
      "learning_rate": 1.4415075055773138e-06,
      "loss": 0.4644,
      "step": 314900
    },
    {
      "epoch": 128.9,
      "grad_norm": 2.031909942626953,
      "learning_rate": 1.4413611029454936e-06,
      "loss": 0.4491,
      "step": 314910
    },
    {
      "epoch": 128.91,
      "grad_norm": 1.939711332321167,
      "learning_rate": 1.441214704447377e-06,
      "loss": 0.4447,
      "step": 314920
    },
    {
      "epoch": 128.91,
      "grad_norm": 1.6355745792388916,
      "learning_rate": 1.441068310083635e-06,
      "loss": 0.4697,
      "step": 314930
    },
    {
      "epoch": 128.92,
      "grad_norm": 2.223545551300049,
      "learning_rate": 1.4409219198549377e-06,
      "loss": 0.4656,
      "step": 314940
    },
    {
      "epoch": 128.92,
      "grad_norm": 2.149245500564575,
      "learning_rate": 1.4407755337619567e-06,
      "loss": 0.4682,
      "step": 314950
    },
    {
      "epoch": 128.92,
      "grad_norm": 1.8910561800003052,
      "learning_rate": 1.4406291518053615e-06,
      "loss": 0.4478,
      "step": 314960
    },
    {
      "epoch": 128.93,
      "grad_norm": 2.139082908630371,
      "learning_rate": 1.4404827739858236e-06,
      "loss": 0.4554,
      "step": 314970
    },
    {
      "epoch": 128.93,
      "grad_norm": 1.9728491306304932,
      "learning_rate": 1.4403364003040125e-06,
      "loss": 0.478,
      "step": 314980
    },
    {
      "epoch": 128.94,
      "grad_norm": 2.188354015350342,
      "learning_rate": 1.440190030760598e-06,
      "loss": 0.4326,
      "step": 314990
    },
    {
      "epoch": 128.94,
      "grad_norm": 2.3053646087646484,
      "learning_rate": 1.4400436653562526e-06,
      "loss": 0.4494,
      "step": 315000
    },
    {
      "epoch": 128.94,
      "grad_norm": 1.978379249572754,
      "learning_rate": 1.4398973040916466e-06,
      "loss": 0.4714,
      "step": 315010
    },
    {
      "epoch": 128.95,
      "grad_norm": 1.6627681255340576,
      "learning_rate": 1.4397509469674498e-06,
      "loss": 0.4625,
      "step": 315020
    },
    {
      "epoch": 128.95,
      "grad_norm": 1.9418367147445679,
      "learning_rate": 1.4396045939843323e-06,
      "loss": 0.4432,
      "step": 315030
    },
    {
      "epoch": 128.96,
      "grad_norm": 1.5976730585098267,
      "learning_rate": 1.439458245142965e-06,
      "loss": 0.4438,
      "step": 315040
    },
    {
      "epoch": 128.96,
      "grad_norm": 2.3629634380340576,
      "learning_rate": 1.4393119004440182e-06,
      "loss": 0.4394,
      "step": 315050
    },
    {
      "epoch": 128.96,
      "grad_norm": 2.006380319595337,
      "learning_rate": 1.4391655598881621e-06,
      "loss": 0.4437,
      "step": 315060
    },
    {
      "epoch": 128.97,
      "grad_norm": 1.9977705478668213,
      "learning_rate": 1.4390192234760675e-06,
      "loss": 0.4535,
      "step": 315070
    },
    {
      "epoch": 128.97,
      "grad_norm": 1.7570617198944092,
      "learning_rate": 1.438872891208404e-06,
      "loss": 0.4707,
      "step": 315080
    },
    {
      "epoch": 128.98,
      "grad_norm": 2.104337215423584,
      "learning_rate": 1.4387265630858428e-06,
      "loss": 0.4521,
      "step": 315090
    },
    {
      "epoch": 128.98,
      "grad_norm": 1.798870325088501,
      "learning_rate": 1.4385802391090535e-06,
      "loss": 0.4378,
      "step": 315100
    },
    {
      "epoch": 128.98,
      "grad_norm": 1.7251996994018555,
      "learning_rate": 1.4384339192787073e-06,
      "loss": 0.4381,
      "step": 315110
    },
    {
      "epoch": 128.99,
      "grad_norm": 1.8827217817306519,
      "learning_rate": 1.4382876035954734e-06,
      "loss": 0.4663,
      "step": 315120
    },
    {
      "epoch": 128.99,
      "grad_norm": 2.2516634464263916,
      "learning_rate": 1.4381412920600226e-06,
      "loss": 0.487,
      "step": 315130
    },
    {
      "epoch": 129.0,
      "grad_norm": 2.080775737762451,
      "learning_rate": 1.4379949846730251e-06,
      "loss": 0.4432,
      "step": 315140
    },
    {
      "epoch": 129.0,
      "eval_loss": 0.45963943004608154,
      "eval_runtime": 52.9504,
      "eval_samples_per_second": 65.136,
      "eval_steps_per_second": 8.159,
      "step": 315147
    },
    {
      "epoch": 129.0,
      "grad_norm": 2.0799598693847656,
      "learning_rate": 1.4378486814351508e-06,
      "loss": 0.4543,
      "step": 315150
    },
    {
      "epoch": 129.01,
      "grad_norm": 1.8187742233276367,
      "learning_rate": 1.4377023823470702e-06,
      "loss": 0.4725,
      "step": 315160
    },
    {
      "epoch": 129.01,
      "grad_norm": 1.64678955078125,
      "learning_rate": 1.4375560874094537e-06,
      "loss": 0.4588,
      "step": 315170
    },
    {
      "epoch": 129.01,
      "grad_norm": 1.8101072311401367,
      "learning_rate": 1.4374097966229705e-06,
      "loss": 0.4582,
      "step": 315180
    },
    {
      "epoch": 129.02,
      "grad_norm": 2.7926201820373535,
      "learning_rate": 1.4372635099882919e-06,
      "loss": 0.4647,
      "step": 315190
    },
    {
      "epoch": 129.02,
      "grad_norm": 1.5411163568496704,
      "learning_rate": 1.4371172275060875e-06,
      "loss": 0.4461,
      "step": 315200
    },
    {
      "epoch": 129.03,
      "grad_norm": 1.8856792449951172,
      "learning_rate": 1.4369709491770275e-06,
      "loss": 0.4758,
      "step": 315210
    },
    {
      "epoch": 129.03,
      "grad_norm": 2.2307231426239014,
      "learning_rate": 1.4368246750017817e-06,
      "loss": 0.4432,
      "step": 315220
    },
    {
      "epoch": 129.03,
      "grad_norm": 1.8099077939987183,
      "learning_rate": 1.4366784049810204e-06,
      "loss": 0.468,
      "step": 315230
    },
    {
      "epoch": 129.04,
      "grad_norm": 2.0979838371276855,
      "learning_rate": 1.436532139115414e-06,
      "loss": 0.4483,
      "step": 315240
    },
    {
      "epoch": 129.04,
      "grad_norm": 2.382164478302002,
      "learning_rate": 1.4363858774056318e-06,
      "loss": 0.4591,
      "step": 315250
    },
    {
      "epoch": 129.05,
      "grad_norm": 3.0024266242980957,
      "learning_rate": 1.4362396198523442e-06,
      "loss": 0.4525,
      "step": 315260
    },
    {
      "epoch": 129.05,
      "grad_norm": 1.9339776039123535,
      "learning_rate": 1.4360933664562208e-06,
      "loss": 0.4472,
      "step": 315270
    },
    {
      "epoch": 129.05,
      "grad_norm": 2.680267095565796,
      "learning_rate": 1.4359471172179323e-06,
      "loss": 0.4619,
      "step": 315280
    },
    {
      "epoch": 129.06,
      "grad_norm": 1.9465601444244385,
      "learning_rate": 1.4358008721381482e-06,
      "loss": 0.4568,
      "step": 315290
    },
    {
      "epoch": 129.06,
      "grad_norm": 2.236196517944336,
      "learning_rate": 1.435654631217538e-06,
      "loss": 0.454,
      "step": 315300
    },
    {
      "epoch": 129.07,
      "grad_norm": 1.9065102338790894,
      "learning_rate": 1.4355083944567723e-06,
      "loss": 0.4392,
      "step": 315310
    },
    {
      "epoch": 129.07,
      "grad_norm": 2.0636110305786133,
      "learning_rate": 1.435362161856521e-06,
      "loss": 0.4707,
      "step": 315320
    },
    {
      "epoch": 129.07,
      "grad_norm": 2.711196184158325,
      "learning_rate": 1.4352159334174535e-06,
      "loss": 0.4642,
      "step": 315330
    },
    {
      "epoch": 129.08,
      "grad_norm": 1.907831072807312,
      "learning_rate": 1.4350697091402398e-06,
      "loss": 0.4759,
      "step": 315340
    },
    {
      "epoch": 129.08,
      "grad_norm": 2.0994107723236084,
      "learning_rate": 1.4349234890255497e-06,
      "loss": 0.4471,
      "step": 315350
    },
    {
      "epoch": 129.09,
      "grad_norm": 1.880454182624817,
      "learning_rate": 1.4347772730740532e-06,
      "loss": 0.4642,
      "step": 315360
    },
    {
      "epoch": 129.09,
      "grad_norm": 2.362182140350342,
      "learning_rate": 1.4346310612864192e-06,
      "loss": 0.456,
      "step": 315370
    },
    {
      "epoch": 129.1,
      "grad_norm": 2.0353798866271973,
      "learning_rate": 1.4344848536633195e-06,
      "loss": 0.4621,
      "step": 315380
    },
    {
      "epoch": 129.1,
      "grad_norm": 2.113590955734253,
      "learning_rate": 1.4343386502054222e-06,
      "loss": 0.4575,
      "step": 315390
    },
    {
      "epoch": 129.1,
      "grad_norm": 2.066204071044922,
      "learning_rate": 1.4341924509133974e-06,
      "loss": 0.4498,
      "step": 315400
    },
    {
      "epoch": 129.11,
      "grad_norm": 1.9282143115997314,
      "learning_rate": 1.4340462557879148e-06,
      "loss": 0.4429,
      "step": 315410
    },
    {
      "epoch": 129.11,
      "grad_norm": 1.9382413625717163,
      "learning_rate": 1.433900064829644e-06,
      "loss": 0.4365,
      "step": 315420
    },
    {
      "epoch": 129.12,
      "grad_norm": 2.1211915016174316,
      "learning_rate": 1.4337538780392552e-06,
      "loss": 0.4538,
      "step": 315430
    },
    {
      "epoch": 129.12,
      "grad_norm": 1.9005675315856934,
      "learning_rate": 1.4336076954174176e-06,
      "loss": 0.4491,
      "step": 315440
    },
    {
      "epoch": 129.12,
      "grad_norm": 1.8291537761688232,
      "learning_rate": 1.4334615169648009e-06,
      "loss": 0.4563,
      "step": 315450
    },
    {
      "epoch": 129.13,
      "grad_norm": 1.927630066871643,
      "learning_rate": 1.4333153426820748e-06,
      "loss": 0.4415,
      "step": 315460
    },
    {
      "epoch": 129.13,
      "grad_norm": 2.3356308937072754,
      "learning_rate": 1.4331691725699095e-06,
      "loss": 0.4592,
      "step": 315470
    },
    {
      "epoch": 129.14,
      "grad_norm": 2.232487201690674,
      "learning_rate": 1.4330230066289733e-06,
      "loss": 0.4602,
      "step": 315480
    },
    {
      "epoch": 129.14,
      "grad_norm": 1.7819629907608032,
      "learning_rate": 1.4328768448599364e-06,
      "loss": 0.463,
      "step": 315490
    },
    {
      "epoch": 129.14,
      "grad_norm": 2.2612626552581787,
      "learning_rate": 1.4327306872634682e-06,
      "loss": 0.4429,
      "step": 315500
    },
    {
      "epoch": 129.15,
      "grad_norm": 2.17988920211792,
      "learning_rate": 1.4325845338402384e-06,
      "loss": 0.4565,
      "step": 315510
    },
    {
      "epoch": 129.15,
      "grad_norm": 1.80097234249115,
      "learning_rate": 1.4324383845909168e-06,
      "loss": 0.4648,
      "step": 315520
    },
    {
      "epoch": 129.16,
      "grad_norm": 2.1839776039123535,
      "learning_rate": 1.4322922395161725e-06,
      "loss": 0.4283,
      "step": 315530
    },
    {
      "epoch": 129.16,
      "grad_norm": 1.6272114515304565,
      "learning_rate": 1.432146098616675e-06,
      "loss": 0.4495,
      "step": 315540
    },
    {
      "epoch": 129.16,
      "grad_norm": 2.2729454040527344,
      "learning_rate": 1.431999961893094e-06,
      "loss": 0.4427,
      "step": 315550
    },
    {
      "epoch": 129.17,
      "grad_norm": 1.863911509513855,
      "learning_rate": 1.4318538293460976e-06,
      "loss": 0.4663,
      "step": 315560
    },
    {
      "epoch": 129.17,
      "grad_norm": 1.6794911623001099,
      "learning_rate": 1.4317077009763573e-06,
      "loss": 0.4569,
      "step": 315570
    },
    {
      "epoch": 129.18,
      "grad_norm": 1.8579005002975464,
      "learning_rate": 1.4315615767845413e-06,
      "loss": 0.4453,
      "step": 315580
    },
    {
      "epoch": 129.18,
      "grad_norm": 1.6939350366592407,
      "learning_rate": 1.4314154567713196e-06,
      "loss": 0.454,
      "step": 315590
    },
    {
      "epoch": 129.19,
      "grad_norm": 1.787008285522461,
      "learning_rate": 1.4312693409373613e-06,
      "loss": 0.4579,
      "step": 315600
    },
    {
      "epoch": 129.19,
      "grad_norm": 1.8634088039398193,
      "learning_rate": 1.431123229283335e-06,
      "loss": 0.4436,
      "step": 315610
    },
    {
      "epoch": 129.19,
      "grad_norm": 1.9655715227127075,
      "learning_rate": 1.430977121809911e-06,
      "loss": 0.4347,
      "step": 315620
    },
    {
      "epoch": 129.2,
      "grad_norm": 2.41377854347229,
      "learning_rate": 1.4308310185177578e-06,
      "loss": 0.4672,
      "step": 315630
    },
    {
      "epoch": 129.2,
      "grad_norm": 1.819984793663025,
      "learning_rate": 1.430684919407545e-06,
      "loss": 0.452,
      "step": 315640
    },
    {
      "epoch": 129.21,
      "grad_norm": 1.9600070714950562,
      "learning_rate": 1.4305388244799423e-06,
      "loss": 0.482,
      "step": 315650
    },
    {
      "epoch": 129.21,
      "grad_norm": 1.7210723161697388,
      "learning_rate": 1.4303927337356184e-06,
      "loss": 0.4628,
      "step": 315660
    },
    {
      "epoch": 129.21,
      "grad_norm": 1.7155369520187378,
      "learning_rate": 1.4302466471752426e-06,
      "loss": 0.454,
      "step": 315670
    },
    {
      "epoch": 129.22,
      "grad_norm": 2.4375078678131104,
      "learning_rate": 1.4301005647994845e-06,
      "loss": 0.4444,
      "step": 315680
    },
    {
      "epoch": 129.22,
      "grad_norm": 2.096984624862671,
      "learning_rate": 1.4299544866090132e-06,
      "loss": 0.447,
      "step": 315690
    },
    {
      "epoch": 129.23,
      "grad_norm": 1.9315145015716553,
      "learning_rate": 1.429808412604497e-06,
      "loss": 0.443,
      "step": 315700
    },
    {
      "epoch": 129.23,
      "grad_norm": 1.8624732494354248,
      "learning_rate": 1.4296623427866059e-06,
      "loss": 0.4696,
      "step": 315710
    },
    {
      "epoch": 129.23,
      "grad_norm": 2.1354994773864746,
      "learning_rate": 1.429516277156009e-06,
      "loss": 0.4768,
      "step": 315720
    },
    {
      "epoch": 129.24,
      "grad_norm": 2.119399070739746,
      "learning_rate": 1.4293702157133748e-06,
      "loss": 0.4528,
      "step": 315730
    },
    {
      "epoch": 129.24,
      "grad_norm": 3.4217028617858887,
      "learning_rate": 1.4292241584593724e-06,
      "loss": 0.4511,
      "step": 315740
    },
    {
      "epoch": 129.25,
      "grad_norm": 2.1189963817596436,
      "learning_rate": 1.429078105394672e-06,
      "loss": 0.4618,
      "step": 315750
    },
    {
      "epoch": 129.25,
      "grad_norm": 1.667564034461975,
      "learning_rate": 1.4289320565199415e-06,
      "loss": 0.4535,
      "step": 315760
    },
    {
      "epoch": 129.26,
      "grad_norm": 1.8750067949295044,
      "learning_rate": 1.4287860118358506e-06,
      "loss": 0.461,
      "step": 315770
    },
    {
      "epoch": 129.26,
      "grad_norm": 2.394935369491577,
      "learning_rate": 1.4286399713430678e-06,
      "loss": 0.4604,
      "step": 315780
    },
    {
      "epoch": 129.26,
      "grad_norm": 2.1159439086914062,
      "learning_rate": 1.4284939350422626e-06,
      "loss": 0.4679,
      "step": 315790
    },
    {
      "epoch": 129.27,
      "grad_norm": 2.0249438285827637,
      "learning_rate": 1.4283479029341036e-06,
      "loss": 0.4565,
      "step": 315800
    },
    {
      "epoch": 129.27,
      "grad_norm": 1.7259411811828613,
      "learning_rate": 1.4282018750192595e-06,
      "loss": 0.4639,
      "step": 315810
    },
    {
      "epoch": 129.28,
      "grad_norm": 2.068309783935547,
      "learning_rate": 1.4280558512984e-06,
      "loss": 0.4745,
      "step": 315820
    },
    {
      "epoch": 129.28,
      "grad_norm": 1.9350526332855225,
      "learning_rate": 1.4279098317721939e-06,
      "loss": 0.4619,
      "step": 315830
    },
    {
      "epoch": 129.28,
      "grad_norm": 1.9899377822875977,
      "learning_rate": 1.4277638164413092e-06,
      "loss": 0.4814,
      "step": 315840
    },
    {
      "epoch": 129.29,
      "grad_norm": 2.2517457008361816,
      "learning_rate": 1.4276178053064154e-06,
      "loss": 0.4764,
      "step": 315850
    },
    {
      "epoch": 129.29,
      "grad_norm": 2.8020341396331787,
      "learning_rate": 1.4274717983681814e-06,
      "loss": 0.4644,
      "step": 315860
    },
    {
      "epoch": 129.3,
      "grad_norm": 1.8948297500610352,
      "learning_rate": 1.427325795627276e-06,
      "loss": 0.4535,
      "step": 315870
    },
    {
      "epoch": 129.3,
      "grad_norm": 1.594655156135559,
      "learning_rate": 1.427179797084368e-06,
      "loss": 0.4464,
      "step": 315880
    },
    {
      "epoch": 129.3,
      "grad_norm": 1.7917553186416626,
      "learning_rate": 1.427033802740126e-06,
      "loss": 0.4448,
      "step": 315890
    },
    {
      "epoch": 129.31,
      "grad_norm": 1.738669991493225,
      "learning_rate": 1.4268878125952186e-06,
      "loss": 0.4644,
      "step": 315900
    },
    {
      "epoch": 129.31,
      "grad_norm": 1.8529627323150635,
      "learning_rate": 1.4267418266503158e-06,
      "loss": 0.4603,
      "step": 315910
    },
    {
      "epoch": 129.32,
      "grad_norm": 1.8412805795669556,
      "learning_rate": 1.4265958449060847e-06,
      "loss": 0.4597,
      "step": 315920
    },
    {
      "epoch": 129.32,
      "grad_norm": 1.7628581523895264,
      "learning_rate": 1.4264498673631941e-06,
      "loss": 0.4784,
      "step": 315930
    },
    {
      "epoch": 129.32,
      "grad_norm": 2.4119303226470947,
      "learning_rate": 1.4263038940223142e-06,
      "loss": 0.4613,
      "step": 315940
    },
    {
      "epoch": 129.33,
      "grad_norm": 1.9419022798538208,
      "learning_rate": 1.4261579248841125e-06,
      "loss": 0.4701,
      "step": 315950
    },
    {
      "epoch": 129.33,
      "grad_norm": 2.496933698654175,
      "learning_rate": 1.4260119599492589e-06,
      "loss": 0.4379,
      "step": 315960
    },
    {
      "epoch": 129.34,
      "grad_norm": 2.1121132373809814,
      "learning_rate": 1.4258659992184204e-06,
      "loss": 0.4456,
      "step": 315970
    },
    {
      "epoch": 129.34,
      "grad_norm": 1.9718607664108276,
      "learning_rate": 1.4257200426922663e-06,
      "loss": 0.4431,
      "step": 315980
    },
    {
      "epoch": 129.35,
      "grad_norm": 2.2493200302124023,
      "learning_rate": 1.4255740903714654e-06,
      "loss": 0.4483,
      "step": 315990
    },
    {
      "epoch": 129.35,
      "grad_norm": 2.100860595703125,
      "learning_rate": 1.4254281422566863e-06,
      "loss": 0.4647,
      "step": 316000
    },
    {
      "epoch": 129.35,
      "grad_norm": 2.4121415615081787,
      "learning_rate": 1.4252821983485972e-06,
      "loss": 0.4573,
      "step": 316010
    },
    {
      "epoch": 129.36,
      "grad_norm": 1.8024417161941528,
      "learning_rate": 1.425136258647867e-06,
      "loss": 0.4556,
      "step": 316020
    },
    {
      "epoch": 129.36,
      "grad_norm": 1.8433386087417603,
      "learning_rate": 1.4249903231551638e-06,
      "loss": 0.4597,
      "step": 316030
    },
    {
      "epoch": 129.37,
      "grad_norm": 2.3171446323394775,
      "learning_rate": 1.4248443918711565e-06,
      "loss": 0.4599,
      "step": 316040
    },
    {
      "epoch": 129.37,
      "grad_norm": 1.7004475593566895,
      "learning_rate": 1.424698464796514e-06,
      "loss": 0.4468,
      "step": 316050
    },
    {
      "epoch": 129.37,
      "grad_norm": 2.016627788543701,
      "learning_rate": 1.424552541931904e-06,
      "loss": 0.4711,
      "step": 316060
    },
    {
      "epoch": 129.38,
      "grad_norm": 1.7489033937454224,
      "learning_rate": 1.4244066232779948e-06,
      "loss": 0.4683,
      "step": 316070
    },
    {
      "epoch": 129.38,
      "grad_norm": 1.7736889123916626,
      "learning_rate": 1.4242607088354558e-06,
      "loss": 0.4645,
      "step": 316080
    },
    {
      "epoch": 129.39,
      "grad_norm": 2.0181474685668945,
      "learning_rate": 1.4241147986049543e-06,
      "loss": 0.4382,
      "step": 316090
    },
    {
      "epoch": 129.39,
      "grad_norm": 1.9782906770706177,
      "learning_rate": 1.4239688925871593e-06,
      "loss": 0.4536,
      "step": 316100
    },
    {
      "epoch": 129.39,
      "grad_norm": 2.6748597621917725,
      "learning_rate": 1.4238229907827393e-06,
      "loss": 0.4416,
      "step": 316110
    },
    {
      "epoch": 129.4,
      "grad_norm": 1.7737112045288086,
      "learning_rate": 1.4236770931923619e-06,
      "loss": 0.4609,
      "step": 316120
    },
    {
      "epoch": 129.4,
      "grad_norm": 1.8485522270202637,
      "learning_rate": 1.4235311998166965e-06,
      "loss": 0.4469,
      "step": 316130
    },
    {
      "epoch": 129.41,
      "grad_norm": 2.260051965713501,
      "learning_rate": 1.4233853106564108e-06,
      "loss": 0.4653,
      "step": 316140
    },
    {
      "epoch": 129.41,
      "grad_norm": 1.7288521528244019,
      "learning_rate": 1.4232394257121734e-06,
      "loss": 0.4389,
      "step": 316150
    },
    {
      "epoch": 129.41,
      "grad_norm": 1.9149361848831177,
      "learning_rate": 1.4230935449846521e-06,
      "loss": 0.46,
      "step": 316160
    },
    {
      "epoch": 129.42,
      "grad_norm": 1.741156816482544,
      "learning_rate": 1.4229476684745156e-06,
      "loss": 0.4561,
      "step": 316170
    },
    {
      "epoch": 129.42,
      "grad_norm": 1.7683340311050415,
      "learning_rate": 1.4228017961824326e-06,
      "loss": 0.4401,
      "step": 316180
    },
    {
      "epoch": 129.43,
      "grad_norm": 2.065100908279419,
      "learning_rate": 1.42265592810907e-06,
      "loss": 0.4658,
      "step": 316190
    },
    {
      "epoch": 129.43,
      "grad_norm": 2.397664785385132,
      "learning_rate": 1.422510064255097e-06,
      "loss": 0.4403,
      "step": 316200
    },
    {
      "epoch": 129.44,
      "grad_norm": 2.6111464500427246,
      "learning_rate": 1.422364204621181e-06,
      "loss": 0.4609,
      "step": 316210
    },
    {
      "epoch": 129.44,
      "grad_norm": 2.1101741790771484,
      "learning_rate": 1.4222183492079911e-06,
      "loss": 0.449,
      "step": 316220
    },
    {
      "epoch": 129.44,
      "grad_norm": 1.7827240228652954,
      "learning_rate": 1.4220724980161947e-06,
      "loss": 0.4626,
      "step": 316230
    },
    {
      "epoch": 129.45,
      "grad_norm": 1.7404154539108276,
      "learning_rate": 1.42192665104646e-06,
      "loss": 0.4533,
      "step": 316240
    },
    {
      "epoch": 129.45,
      "grad_norm": 2.085275411605835,
      "learning_rate": 1.4217808082994557e-06,
      "loss": 0.4686,
      "step": 316250
    },
    {
      "epoch": 129.46,
      "grad_norm": 1.7685240507125854,
      "learning_rate": 1.4216349697758495e-06,
      "loss": 0.4616,
      "step": 316260
    },
    {
      "epoch": 129.46,
      "grad_norm": 2.2435662746429443,
      "learning_rate": 1.4214891354763094e-06,
      "loss": 0.4814,
      "step": 316270
    },
    {
      "epoch": 129.46,
      "grad_norm": 1.7794179916381836,
      "learning_rate": 1.4213433054015035e-06,
      "loss": 0.4709,
      "step": 316280
    },
    {
      "epoch": 129.47,
      "grad_norm": 1.6929465532302856,
      "learning_rate": 1.4211974795521e-06,
      "loss": 0.4533,
      "step": 316290
    },
    {
      "epoch": 129.47,
      "grad_norm": 1.919176459312439,
      "learning_rate": 1.4210516579287662e-06,
      "loss": 0.4421,
      "step": 316300
    },
    {
      "epoch": 129.48,
      "grad_norm": 1.9633833169937134,
      "learning_rate": 1.4209058405321699e-06,
      "loss": 0.4652,
      "step": 316310
    },
    {
      "epoch": 129.48,
      "grad_norm": 2.150566816329956,
      "learning_rate": 1.4207600273629811e-06,
      "loss": 0.451,
      "step": 316320
    },
    {
      "epoch": 129.48,
      "grad_norm": 1.7163337469100952,
      "learning_rate": 1.4206142184218662e-06,
      "loss": 0.4228,
      "step": 316330
    },
    {
      "epoch": 129.49,
      "grad_norm": 2.133676290512085,
      "learning_rate": 1.420468413709493e-06,
      "loss": 0.4643,
      "step": 316340
    },
    {
      "epoch": 129.49,
      "grad_norm": 1.961190938949585,
      "learning_rate": 1.4203226132265298e-06,
      "loss": 0.4337,
      "step": 316350
    },
    {
      "epoch": 129.5,
      "grad_norm": 1.8102847337722778,
      "learning_rate": 1.4201768169736446e-06,
      "loss": 0.4704,
      "step": 316360
    },
    {
      "epoch": 129.5,
      "grad_norm": 2.2828381061553955,
      "learning_rate": 1.4200310249515052e-06,
      "loss": 0.4516,
      "step": 316370
    },
    {
      "epoch": 129.5,
      "grad_norm": 2.036201000213623,
      "learning_rate": 1.4198852371607794e-06,
      "loss": 0.4535,
      "step": 316380
    },
    {
      "epoch": 129.51,
      "grad_norm": 1.9244085550308228,
      "learning_rate": 1.4197394536021346e-06,
      "loss": 0.4669,
      "step": 316390
    },
    {
      "epoch": 129.51,
      "grad_norm": 1.996009111404419,
      "learning_rate": 1.4195936742762395e-06,
      "loss": 0.4508,
      "step": 316400
    },
    {
      "epoch": 129.52,
      "grad_norm": 3.1581485271453857,
      "learning_rate": 1.4194478991837616e-06,
      "loss": 0.4467,
      "step": 316410
    },
    {
      "epoch": 129.52,
      "grad_norm": 1.885116696357727,
      "learning_rate": 1.4193021283253678e-06,
      "loss": 0.4587,
      "step": 316420
    },
    {
      "epoch": 129.53,
      "grad_norm": 1.891495704650879,
      "learning_rate": 1.4191563617017269e-06,
      "loss": 0.4586,
      "step": 316430
    },
    {
      "epoch": 129.53,
      "grad_norm": 1.8660041093826294,
      "learning_rate": 1.419010599313506e-06,
      "loss": 0.4357,
      "step": 316440
    },
    {
      "epoch": 129.53,
      "grad_norm": 3.3661882877349854,
      "learning_rate": 1.4188648411613735e-06,
      "loss": 0.4568,
      "step": 316450
    },
    {
      "epoch": 129.54,
      "grad_norm": 2.1317989826202393,
      "learning_rate": 1.4187190872459963e-06,
      "loss": 0.4696,
      "step": 316460
    },
    {
      "epoch": 129.54,
      "grad_norm": 2.221855878829956,
      "learning_rate": 1.4185733375680425e-06,
      "loss": 0.4706,
      "step": 316470
    },
    {
      "epoch": 129.55,
      "grad_norm": 2.1500725746154785,
      "learning_rate": 1.41842759212818e-06,
      "loss": 0.4599,
      "step": 316480
    },
    {
      "epoch": 129.55,
      "grad_norm": 2.1208443641662598,
      "learning_rate": 1.4182818509270761e-06,
      "loss": 0.4678,
      "step": 316490
    },
    {
      "epoch": 129.55,
      "grad_norm": 1.6822856664657593,
      "learning_rate": 1.4181361139653974e-06,
      "loss": 0.4683,
      "step": 316500
    },
    {
      "epoch": 129.56,
      "grad_norm": 2.0375828742980957,
      "learning_rate": 1.4179903812438135e-06,
      "loss": 0.4803,
      "step": 316510
    },
    {
      "epoch": 129.56,
      "grad_norm": 2.0830163955688477,
      "learning_rate": 1.417844652762991e-06,
      "loss": 0.4507,
      "step": 316520
    },
    {
      "epoch": 129.57,
      "grad_norm": 1.595754623413086,
      "learning_rate": 1.4176989285235974e-06,
      "loss": 0.4494,
      "step": 316530
    },
    {
      "epoch": 129.57,
      "grad_norm": 1.9751945734024048,
      "learning_rate": 1.417553208526301e-06,
      "loss": 0.4471,
      "step": 316540
    },
    {
      "epoch": 129.57,
      "grad_norm": 2.361387252807617,
      "learning_rate": 1.4174074927717682e-06,
      "loss": 0.4723,
      "step": 316550
    },
    {
      "epoch": 129.58,
      "grad_norm": 1.8724204301834106,
      "learning_rate": 1.417261781260667e-06,
      "loss": 0.4543,
      "step": 316560
    },
    {
      "epoch": 129.58,
      "grad_norm": 1.8907617330551147,
      "learning_rate": 1.4171160739936644e-06,
      "loss": 0.4549,
      "step": 316570
    },
    {
      "epoch": 129.59,
      "grad_norm": 1.8992557525634766,
      "learning_rate": 1.4169703709714285e-06,
      "loss": 0.4739,
      "step": 316580
    },
    {
      "epoch": 129.59,
      "grad_norm": 1.955938458442688,
      "learning_rate": 1.4168246721946268e-06,
      "loss": 0.4517,
      "step": 316590
    },
    {
      "epoch": 129.59,
      "grad_norm": 2.7033205032348633,
      "learning_rate": 1.4166789776639262e-06,
      "loss": 0.4766,
      "step": 316600
    },
    {
      "epoch": 129.6,
      "grad_norm": 2.2337474822998047,
      "learning_rate": 1.4165332873799947e-06,
      "loss": 0.4422,
      "step": 316610
    },
    {
      "epoch": 129.6,
      "grad_norm": 2.1840100288391113,
      "learning_rate": 1.4163876013434989e-06,
      "loss": 0.44,
      "step": 316620
    },
    {
      "epoch": 129.61,
      "grad_norm": 11.24560260772705,
      "learning_rate": 1.4162419195551077e-06,
      "loss": 0.4656,
      "step": 316630
    },
    {
      "epoch": 129.61,
      "grad_norm": 2.0883212089538574,
      "learning_rate": 1.416096242015486e-06,
      "loss": 0.4732,
      "step": 316640
    },
    {
      "epoch": 129.62,
      "grad_norm": 2.1108574867248535,
      "learning_rate": 1.4159505687253036e-06,
      "loss": 0.4565,
      "step": 316650
    },
    {
      "epoch": 129.62,
      "grad_norm": 2.047447681427002,
      "learning_rate": 1.4158048996852259e-06,
      "loss": 0.4546,
      "step": 316660
    },
    {
      "epoch": 129.62,
      "grad_norm": 1.423812747001648,
      "learning_rate": 1.4156592348959213e-06,
      "loss": 0.468,
      "step": 316670
    },
    {
      "epoch": 129.63,
      "grad_norm": 1.7415099143981934,
      "learning_rate": 1.4155135743580567e-06,
      "loss": 0.4505,
      "step": 316680
    },
    {
      "epoch": 129.63,
      "grad_norm": 2.1533210277557373,
      "learning_rate": 1.415367918072299e-06,
      "loss": 0.4729,
      "step": 316690
    },
    {
      "epoch": 129.64,
      "grad_norm": 1.8758702278137207,
      "learning_rate": 1.4152222660393164e-06,
      "loss": 0.4635,
      "step": 316700
    },
    {
      "epoch": 129.64,
      "grad_norm": 1.9874212741851807,
      "learning_rate": 1.4150766182597754e-06,
      "loss": 0.4808,
      "step": 316710
    },
    {
      "epoch": 129.64,
      "grad_norm": 1.7926276922225952,
      "learning_rate": 1.4149309747343434e-06,
      "loss": 0.4694,
      "step": 316720
    },
    {
      "epoch": 129.65,
      "grad_norm": 2.2415006160736084,
      "learning_rate": 1.4147853354636875e-06,
      "loss": 0.4467,
      "step": 316730
    },
    {
      "epoch": 129.65,
      "grad_norm": 1.7589735984802246,
      "learning_rate": 1.4146397004484747e-06,
      "loss": 0.4596,
      "step": 316740
    },
    {
      "epoch": 129.66,
      "grad_norm": 2.630089044570923,
      "learning_rate": 1.4144940696893725e-06,
      "loss": 0.4608,
      "step": 316750
    },
    {
      "epoch": 129.66,
      "grad_norm": 2.021932601928711,
      "learning_rate": 1.4143484431870478e-06,
      "loss": 0.4689,
      "step": 316760
    },
    {
      "epoch": 129.66,
      "grad_norm": 1.8521671295166016,
      "learning_rate": 1.414202820942168e-06,
      "loss": 0.4438,
      "step": 316770
    },
    {
      "epoch": 129.67,
      "grad_norm": 2.185697078704834,
      "learning_rate": 1.4140572029553995e-06,
      "loss": 0.4678,
      "step": 316780
    },
    {
      "epoch": 129.67,
      "grad_norm": 2.1707725524902344,
      "learning_rate": 1.4139115892274096e-06,
      "loss": 0.4388,
      "step": 316790
    },
    {
      "epoch": 129.68,
      "grad_norm": 1.6667723655700684,
      "learning_rate": 1.4137659797588656e-06,
      "loss": 0.4459,
      "step": 316800
    },
    {
      "epoch": 129.68,
      "grad_norm": 2.0546462535858154,
      "learning_rate": 1.4136203745504345e-06,
      "loss": 0.4378,
      "step": 316810
    },
    {
      "epoch": 129.68,
      "grad_norm": 1.9877116680145264,
      "learning_rate": 1.413474773602783e-06,
      "loss": 0.4526,
      "step": 316820
    },
    {
      "epoch": 129.69,
      "grad_norm": 1.966096043586731,
      "learning_rate": 1.4133291769165782e-06,
      "loss": 0.4448,
      "step": 316830
    },
    {
      "epoch": 129.69,
      "grad_norm": 2.044431447982788,
      "learning_rate": 1.4131835844924872e-06,
      "loss": 0.4446,
      "step": 316840
    },
    {
      "epoch": 129.7,
      "grad_norm": 2.2095773220062256,
      "learning_rate": 1.4130379963311772e-06,
      "loss": 0.4863,
      "step": 316850
    },
    {
      "epoch": 129.7,
      "grad_norm": 2.1404521465301514,
      "learning_rate": 1.4128924124333142e-06,
      "loss": 0.4709,
      "step": 316860
    },
    {
      "epoch": 129.71,
      "grad_norm": 2.0972604751586914,
      "learning_rate": 1.4127468327995657e-06,
      "loss": 0.4556,
      "step": 316870
    },
    {
      "epoch": 129.71,
      "grad_norm": 1.632980465888977,
      "learning_rate": 1.412601257430598e-06,
      "loss": 0.4417,
      "step": 316880
    },
    {
      "epoch": 129.71,
      "grad_norm": 2.456552267074585,
      "learning_rate": 1.412455686327079e-06,
      "loss": 0.4617,
      "step": 316890
    },
    {
      "epoch": 129.72,
      "grad_norm": 2.063173294067383,
      "learning_rate": 1.4123101194896754e-06,
      "loss": 0.4714,
      "step": 316900
    },
    {
      "epoch": 129.72,
      "grad_norm": 2.1869192123413086,
      "learning_rate": 1.4121645569190534e-06,
      "loss": 0.4604,
      "step": 316910
    },
    {
      "epoch": 129.73,
      "grad_norm": 1.8624529838562012,
      "learning_rate": 1.41201899861588e-06,
      "loss": 0.4406,
      "step": 316920
    },
    {
      "epoch": 129.73,
      "grad_norm": 2.247856378555298,
      "learning_rate": 1.4118734445808221e-06,
      "loss": 0.4747,
      "step": 316930
    },
    {
      "epoch": 129.73,
      "grad_norm": 2.077920913696289,
      "learning_rate": 1.4117278948145462e-06,
      "loss": 0.4561,
      "step": 316940
    },
    {
      "epoch": 129.74,
      "grad_norm": 2.2134757041931152,
      "learning_rate": 1.411582349317719e-06,
      "loss": 0.4858,
      "step": 316950
    },
    {
      "epoch": 129.74,
      "grad_norm": 1.7775979042053223,
      "learning_rate": 1.4114368080910082e-06,
      "loss": 0.4594,
      "step": 316960
    },
    {
      "epoch": 129.75,
      "grad_norm": 1.6924481391906738,
      "learning_rate": 1.4112912711350792e-06,
      "loss": 0.4611,
      "step": 316970
    },
    {
      "epoch": 129.75,
      "grad_norm": 1.9303661584854126,
      "learning_rate": 1.4111457384505993e-06,
      "loss": 0.4652,
      "step": 316980
    },
    {
      "epoch": 129.75,
      "grad_norm": 2.294468402862549,
      "learning_rate": 1.4110002100382357e-06,
      "loss": 0.4502,
      "step": 316990
    },
    {
      "epoch": 129.76,
      "grad_norm": 1.799873948097229,
      "learning_rate": 1.4108546858986538e-06,
      "loss": 0.4529,
      "step": 317000
    },
    {
      "epoch": 129.76,
      "grad_norm": 1.7515541315078735,
      "learning_rate": 1.4107091660325209e-06,
      "loss": 0.4529,
      "step": 317010
    },
    {
      "epoch": 129.77,
      "grad_norm": 1.754589319229126,
      "learning_rate": 1.4105636504405035e-06,
      "loss": 0.4641,
      "step": 317020
    },
    {
      "epoch": 129.77,
      "grad_norm": 2.1813488006591797,
      "learning_rate": 1.4104181391232684e-06,
      "loss": 0.4485,
      "step": 317030
    },
    {
      "epoch": 129.77,
      "grad_norm": 2.0236167907714844,
      "learning_rate": 1.4102726320814821e-06,
      "loss": 0.4577,
      "step": 317040
    },
    {
      "epoch": 129.78,
      "grad_norm": 1.925236463546753,
      "learning_rate": 1.4101271293158106e-06,
      "loss": 0.4379,
      "step": 317050
    },
    {
      "epoch": 129.78,
      "grad_norm": 2.1234452724456787,
      "learning_rate": 1.4099816308269207e-06,
      "loss": 0.4565,
      "step": 317060
    },
    {
      "epoch": 129.79,
      "grad_norm": 2.0071356296539307,
      "learning_rate": 1.4098361366154794e-06,
      "loss": 0.4787,
      "step": 317070
    },
    {
      "epoch": 129.79,
      "grad_norm": 2.120807647705078,
      "learning_rate": 1.409690646682153e-06,
      "loss": 0.4664,
      "step": 317080
    },
    {
      "epoch": 129.8,
      "grad_norm": 2.482246160507202,
      "learning_rate": 1.4095451610276082e-06,
      "loss": 0.475,
      "step": 317090
    },
    {
      "epoch": 129.8,
      "grad_norm": 1.9643157720565796,
      "learning_rate": 1.4093996796525108e-06,
      "loss": 0.4416,
      "step": 317100
    },
    {
      "epoch": 129.8,
      "grad_norm": 2.041426658630371,
      "learning_rate": 1.4092542025575276e-06,
      "loss": 0.4646,
      "step": 317110
    },
    {
      "epoch": 129.81,
      "grad_norm": 2.1216344833374023,
      "learning_rate": 1.4091087297433243e-06,
      "loss": 0.4803,
      "step": 317120
    },
    {
      "epoch": 129.81,
      "grad_norm": 1.8324569463729858,
      "learning_rate": 1.4089632612105692e-06,
      "loss": 0.4695,
      "step": 317130
    },
    {
      "epoch": 129.82,
      "grad_norm": 1.8929227590560913,
      "learning_rate": 1.4088177969599267e-06,
      "loss": 0.4587,
      "step": 317140
    },
    {
      "epoch": 129.82,
      "grad_norm": 1.9527112245559692,
      "learning_rate": 1.4086723369920636e-06,
      "loss": 0.4779,
      "step": 317150
    },
    {
      "epoch": 129.82,
      "grad_norm": 1.9371882677078247,
      "learning_rate": 1.4085268813076465e-06,
      "loss": 0.4467,
      "step": 317160
    },
    {
      "epoch": 129.83,
      "grad_norm": 1.7618743181228638,
      "learning_rate": 1.4083814299073417e-06,
      "loss": 0.4587,
      "step": 317170
    },
    {
      "epoch": 129.83,
      "grad_norm": 2.1169095039367676,
      "learning_rate": 1.4082359827918153e-06,
      "loss": 0.4354,
      "step": 317180
    },
    {
      "epoch": 129.84,
      "grad_norm": 2.107452154159546,
      "learning_rate": 1.4080905399617343e-06,
      "loss": 0.4553,
      "step": 317190
    },
    {
      "epoch": 129.84,
      "grad_norm": 1.8169835805892944,
      "learning_rate": 1.4079451014177638e-06,
      "loss": 0.4479,
      "step": 317200
    },
    {
      "epoch": 129.84,
      "grad_norm": 1.8235321044921875,
      "learning_rate": 1.4077996671605712e-06,
      "loss": 0.4479,
      "step": 317210
    },
    {
      "epoch": 129.85,
      "grad_norm": 2.1002514362335205,
      "learning_rate": 1.4076542371908217e-06,
      "loss": 0.4499,
      "step": 317220
    },
    {
      "epoch": 129.85,
      "grad_norm": 1.9101312160491943,
      "learning_rate": 1.4075088115091822e-06,
      "loss": 0.4615,
      "step": 317230
    },
    {
      "epoch": 129.86,
      "grad_norm": 2.084153175354004,
      "learning_rate": 1.4073633901163182e-06,
      "loss": 0.4631,
      "step": 317240
    },
    {
      "epoch": 129.86,
      "grad_norm": 1.8766957521438599,
      "learning_rate": 1.4072179730128956e-06,
      "loss": 0.4636,
      "step": 317250
    },
    {
      "epoch": 129.86,
      "grad_norm": 1.7698718309402466,
      "learning_rate": 1.4070725601995824e-06,
      "loss": 0.4883,
      "step": 317260
    },
    {
      "epoch": 129.87,
      "grad_norm": 1.7187827825546265,
      "learning_rate": 1.4069271516770431e-06,
      "loss": 0.4528,
      "step": 317270
    },
    {
      "epoch": 129.87,
      "grad_norm": 2.0145552158355713,
      "learning_rate": 1.4067817474459442e-06,
      "loss": 0.4723,
      "step": 317280
    },
    {
      "epoch": 129.88,
      "grad_norm": 2.0407943725585938,
      "learning_rate": 1.4066363475069518e-06,
      "loss": 0.4481,
      "step": 317290
    },
    {
      "epoch": 129.88,
      "grad_norm": 2.139521598815918,
      "learning_rate": 1.4064909518607317e-06,
      "loss": 0.4518,
      "step": 317300
    },
    {
      "epoch": 129.89,
      "grad_norm": 2.150871515274048,
      "learning_rate": 1.4063455605079503e-06,
      "loss": 0.4597,
      "step": 317310
    },
    {
      "epoch": 129.89,
      "grad_norm": 2.356527328491211,
      "learning_rate": 1.4062001734492733e-06,
      "loss": 0.4567,
      "step": 317320
    },
    {
      "epoch": 129.89,
      "grad_norm": 1.5592923164367676,
      "learning_rate": 1.4060547906853668e-06,
      "loss": 0.456,
      "step": 317330
    },
    {
      "epoch": 129.9,
      "grad_norm": 1.9620418548583984,
      "learning_rate": 1.4059094122168972e-06,
      "loss": 0.4627,
      "step": 317340
    },
    {
      "epoch": 129.9,
      "grad_norm": 1.8044513463974,
      "learning_rate": 1.4057640380445304e-06,
      "loss": 0.4463,
      "step": 317350
    },
    {
      "epoch": 129.91,
      "grad_norm": 4.074211120605469,
      "learning_rate": 1.4056186681689313e-06,
      "loss": 0.4501,
      "step": 317360
    },
    {
      "epoch": 129.91,
      "grad_norm": 2.185451030731201,
      "learning_rate": 1.405473302590767e-06,
      "loss": 0.4539,
      "step": 317370
    },
    {
      "epoch": 129.91,
      "grad_norm": 1.9085359573364258,
      "learning_rate": 1.4053279413107026e-06,
      "loss": 0.4408,
      "step": 317380
    },
    {
      "epoch": 129.92,
      "grad_norm": 1.9695334434509277,
      "learning_rate": 1.4051825843294046e-06,
      "loss": 0.4525,
      "step": 317390
    },
    {
      "epoch": 129.92,
      "grad_norm": 1.5777474641799927,
      "learning_rate": 1.4050372316475382e-06,
      "loss": 0.4252,
      "step": 317400
    },
    {
      "epoch": 129.93,
      "grad_norm": 2.892507553100586,
      "learning_rate": 1.40489188326577e-06,
      "loss": 0.4703,
      "step": 317410
    },
    {
      "epoch": 129.93,
      "grad_norm": 2.239182710647583,
      "learning_rate": 1.4047465391847652e-06,
      "loss": 0.4424,
      "step": 317420
    },
    {
      "epoch": 129.93,
      "grad_norm": 1.8275141716003418,
      "learning_rate": 1.4046011994051897e-06,
      "loss": 0.4506,
      "step": 317430
    },
    {
      "epoch": 129.94,
      "grad_norm": 1.9520673751831055,
      "learning_rate": 1.4044558639277094e-06,
      "loss": 0.4653,
      "step": 317440
    },
    {
      "epoch": 129.94,
      "grad_norm": 1.8732985258102417,
      "learning_rate": 1.4043105327529904e-06,
      "loss": 0.4664,
      "step": 317450
    },
    {
      "epoch": 129.95,
      "grad_norm": 1.9682698249816895,
      "learning_rate": 1.4041652058816977e-06,
      "loss": 0.4468,
      "step": 317460
    },
    {
      "epoch": 129.95,
      "grad_norm": 2.582279920578003,
      "learning_rate": 1.4040198833144979e-06,
      "loss": 0.4617,
      "step": 317470
    },
    {
      "epoch": 129.95,
      "grad_norm": 2.5904574394226074,
      "learning_rate": 1.4038745650520559e-06,
      "loss": 0.4693,
      "step": 317480
    },
    {
      "epoch": 129.96,
      "grad_norm": 2.174729108810425,
      "learning_rate": 1.4037292510950383e-06,
      "loss": 0.4554,
      "step": 317490
    },
    {
      "epoch": 129.96,
      "grad_norm": 2.0058603286743164,
      "learning_rate": 1.4035839414441096e-06,
      "loss": 0.4473,
      "step": 317500
    },
    {
      "epoch": 129.97,
      "grad_norm": 1.9710516929626465,
      "learning_rate": 1.403438636099936e-06,
      "loss": 0.4633,
      "step": 317510
    },
    {
      "epoch": 129.97,
      "grad_norm": 1.7475157976150513,
      "learning_rate": 1.4032933350631834e-06,
      "loss": 0.4437,
      "step": 317520
    },
    {
      "epoch": 129.98,
      "grad_norm": 1.9006998538970947,
      "learning_rate": 1.4031480383345172e-06,
      "loss": 0.4418,
      "step": 317530
    },
    {
      "epoch": 129.98,
      "grad_norm": 2.377502202987671,
      "learning_rate": 1.4030027459146025e-06,
      "loss": 0.4547,
      "step": 317540
    },
    {
      "epoch": 129.98,
      "grad_norm": 2.0231235027313232,
      "learning_rate": 1.4028574578041054e-06,
      "loss": 0.4875,
      "step": 317550
    },
    {
      "epoch": 129.99,
      "grad_norm": 1.7839385271072388,
      "learning_rate": 1.4027121740036914e-06,
      "loss": 0.4612,
      "step": 317560
    },
    {
      "epoch": 129.99,
      "grad_norm": 1.8401228189468384,
      "learning_rate": 1.4025668945140265e-06,
      "loss": 0.4682,
      "step": 317570
    },
    {
      "epoch": 130.0,
      "grad_norm": 2.181584358215332,
      "learning_rate": 1.4024216193357754e-06,
      "loss": 0.4505,
      "step": 317580
    },
    {
      "epoch": 130.0,
      "grad_norm": 2.3567962646484375,
      "learning_rate": 1.4022763484696032e-06,
      "loss": 0.4518,
      "step": 317590
    },
    {
      "epoch": 130.0,
      "eval_loss": 0.45495495200157166,
      "eval_runtime": 52.3766,
      "eval_samples_per_second": 65.85,
      "eval_steps_per_second": 8.248,
      "step": 317590
    },
    {
      "epoch": 130.0,
      "grad_norm": 2.1592302322387695,
      "learning_rate": 1.4021310819161763e-06,
      "loss": 0.4646,
      "step": 317600
    },
    {
      "epoch": 130.01,
      "grad_norm": 1.8462228775024414,
      "learning_rate": 1.40198581967616e-06,
      "loss": 0.458,
      "step": 317610
    },
    {
      "epoch": 130.01,
      "grad_norm": 2.0257153511047363,
      "learning_rate": 1.4018405617502193e-06,
      "loss": 0.4635,
      "step": 317620
    },
    {
      "epoch": 130.02,
      "grad_norm": 2.218430519104004,
      "learning_rate": 1.4016953081390195e-06,
      "loss": 0.4815,
      "step": 317630
    },
    {
      "epoch": 130.02,
      "grad_norm": 2.1956729888916016,
      "learning_rate": 1.4015500588432268e-06,
      "loss": 0.4414,
      "step": 317640
    },
    {
      "epoch": 130.02,
      "grad_norm": 2.0303993225097656,
      "learning_rate": 1.401404813863506e-06,
      "loss": 0.462,
      "step": 317650
    },
    {
      "epoch": 130.03,
      "grad_norm": 2.066497564315796,
      "learning_rate": 1.4012595732005228e-06,
      "loss": 0.4736,
      "step": 317660
    },
    {
      "epoch": 130.03,
      "grad_norm": 2.373900890350342,
      "learning_rate": 1.4011143368549423e-06,
      "loss": 0.4606,
      "step": 317670
    },
    {
      "epoch": 130.04,
      "grad_norm": 2.299638271331787,
      "learning_rate": 1.4009691048274296e-06,
      "loss": 0.4647,
      "step": 317680
    },
    {
      "epoch": 130.04,
      "grad_norm": 1.7460185289382935,
      "learning_rate": 1.4008238771186502e-06,
      "loss": 0.4547,
      "step": 317690
    },
    {
      "epoch": 130.05,
      "grad_norm": 1.8293150663375854,
      "learning_rate": 1.4006786537292693e-06,
      "loss": 0.4425,
      "step": 317700
    },
    {
      "epoch": 130.05,
      "grad_norm": 5.703585147857666,
      "learning_rate": 1.4005334346599527e-06,
      "loss": 0.4735,
      "step": 317710
    },
    {
      "epoch": 130.05,
      "grad_norm": 2.002490282058716,
      "learning_rate": 1.4003882199113646e-06,
      "loss": 0.4425,
      "step": 317720
    },
    {
      "epoch": 130.06,
      "grad_norm": 2.4922595024108887,
      "learning_rate": 1.4002430094841708e-06,
      "loss": 0.4601,
      "step": 317730
    },
    {
      "epoch": 130.06,
      "grad_norm": 1.9178669452667236,
      "learning_rate": 1.4000978033790364e-06,
      "loss": 0.4514,
      "step": 317740
    },
    {
      "epoch": 130.07,
      "grad_norm": 1.9101896286010742,
      "learning_rate": 1.3999526015966268e-06,
      "loss": 0.4587,
      "step": 317750
    },
    {
      "epoch": 130.07,
      "grad_norm": 1.758703351020813,
      "learning_rate": 1.3998074041376065e-06,
      "loss": 0.4666,
      "step": 317760
    },
    {
      "epoch": 130.07,
      "grad_norm": 2.1650450229644775,
      "learning_rate": 1.399662211002641e-06,
      "loss": 0.4774,
      "step": 317770
    },
    {
      "epoch": 130.08,
      "grad_norm": 1.978934645652771,
      "learning_rate": 1.399517022192396e-06,
      "loss": 0.4487,
      "step": 317780
    },
    {
      "epoch": 130.08,
      "grad_norm": 3.459207057952881,
      "learning_rate": 1.3993718377075354e-06,
      "loss": 0.4584,
      "step": 317790
    },
    {
      "epoch": 130.09,
      "grad_norm": 2.0671162605285645,
      "learning_rate": 1.3992266575487258e-06,
      "loss": 0.45,
      "step": 317800
    },
    {
      "epoch": 130.09,
      "grad_norm": 1.9535956382751465,
      "learning_rate": 1.3990814817166307e-06,
      "loss": 0.4418,
      "step": 317810
    },
    {
      "epoch": 130.09,
      "grad_norm": 1.8086280822753906,
      "learning_rate": 1.3989363102119148e-06,
      "loss": 0.454,
      "step": 317820
    },
    {
      "epoch": 130.1,
      "grad_norm": 1.8750581741333008,
      "learning_rate": 1.3987911430352453e-06,
      "loss": 0.4593,
      "step": 317830
    },
    {
      "epoch": 130.1,
      "grad_norm": 2.5152013301849365,
      "learning_rate": 1.398645980187286e-06,
      "loss": 0.4642,
      "step": 317840
    },
    {
      "epoch": 130.11,
      "grad_norm": 1.9994983673095703,
      "learning_rate": 1.3985008216687022e-06,
      "loss": 0.4418,
      "step": 317850
    },
    {
      "epoch": 130.11,
      "grad_norm": 2.219647169113159,
      "learning_rate": 1.3983556674801578e-06,
      "loss": 0.4641,
      "step": 317860
    },
    {
      "epoch": 130.11,
      "grad_norm": 2.5320487022399902,
      "learning_rate": 1.3982105176223184e-06,
      "loss": 0.4613,
      "step": 317870
    },
    {
      "epoch": 130.12,
      "grad_norm": 2.2997827529907227,
      "learning_rate": 1.3980653720958492e-06,
      "loss": 0.4715,
      "step": 317880
    },
    {
      "epoch": 130.12,
      "grad_norm": 1.856750249862671,
      "learning_rate": 1.3979202309014147e-06,
      "loss": 0.458,
      "step": 317890
    },
    {
      "epoch": 130.13,
      "grad_norm": 1.6890429258346558,
      "learning_rate": 1.3977750940396802e-06,
      "loss": 0.4628,
      "step": 317900
    },
    {
      "epoch": 130.13,
      "grad_norm": 1.6388370990753174,
      "learning_rate": 1.3976299615113099e-06,
      "loss": 0.4283,
      "step": 317910
    },
    {
      "epoch": 130.14,
      "grad_norm": 2.0168039798736572,
      "learning_rate": 1.397484833316969e-06,
      "loss": 0.4398,
      "step": 317920
    },
    {
      "epoch": 130.14,
      "grad_norm": 1.8292547464370728,
      "learning_rate": 1.3973397094573225e-06,
      "loss": 0.4556,
      "step": 317930
    },
    {
      "epoch": 130.14,
      "grad_norm": 1.6373810768127441,
      "learning_rate": 1.397194589933035e-06,
      "loss": 0.4506,
      "step": 317940
    },
    {
      "epoch": 130.15,
      "grad_norm": 2.2108187675476074,
      "learning_rate": 1.3970494747447711e-06,
      "loss": 0.4598,
      "step": 317950
    },
    {
      "epoch": 130.15,
      "grad_norm": 2.2681474685668945,
      "learning_rate": 1.3969043638931958e-06,
      "loss": 0.4538,
      "step": 317960
    },
    {
      "epoch": 130.16,
      "grad_norm": 1.9581518173217773,
      "learning_rate": 1.3967592573789732e-06,
      "loss": 0.4573,
      "step": 317970
    },
    {
      "epoch": 130.16,
      "grad_norm": 1.988695502281189,
      "learning_rate": 1.396614155202769e-06,
      "loss": 0.4502,
      "step": 317980
    },
    {
      "epoch": 130.16,
      "grad_norm": 2.10599422454834,
      "learning_rate": 1.3964690573652474e-06,
      "loss": 0.445,
      "step": 317990
    },
    {
      "epoch": 130.17,
      "grad_norm": 1.8091922998428345,
      "learning_rate": 1.3963239638670728e-06,
      "loss": 0.4547,
      "step": 318000
    },
    {
      "epoch": 130.17,
      "grad_norm": 2.270073890686035,
      "learning_rate": 1.3961788747089101e-06,
      "loss": 0.4775,
      "step": 318010
    },
    {
      "epoch": 130.18,
      "grad_norm": 2.0436110496520996,
      "learning_rate": 1.396033789891424e-06,
      "loss": 0.4487,
      "step": 318020
    },
    {
      "epoch": 130.18,
      "grad_norm": 1.8395533561706543,
      "learning_rate": 1.3958887094152794e-06,
      "loss": 0.4462,
      "step": 318030
    },
    {
      "epoch": 130.18,
      "grad_norm": 2.0585389137268066,
      "learning_rate": 1.3957436332811404e-06,
      "loss": 0.4651,
      "step": 318040
    },
    {
      "epoch": 130.19,
      "grad_norm": 1.7624021768569946,
      "learning_rate": 1.3955985614896719e-06,
      "loss": 0.4774,
      "step": 318050
    },
    {
      "epoch": 130.19,
      "grad_norm": 2.3412811756134033,
      "learning_rate": 1.3954534940415382e-06,
      "loss": 0.4788,
      "step": 318060
    },
    {
      "epoch": 130.2,
      "grad_norm": 1.8107001781463623,
      "learning_rate": 1.3953084309374045e-06,
      "loss": 0.4555,
      "step": 318070
    },
    {
      "epoch": 130.2,
      "grad_norm": 1.826362133026123,
      "learning_rate": 1.3951633721779342e-06,
      "loss": 0.4463,
      "step": 318080
    },
    {
      "epoch": 130.2,
      "grad_norm": 2.4473657608032227,
      "learning_rate": 1.395018317763792e-06,
      "loss": 0.4552,
      "step": 318090
    },
    {
      "epoch": 130.21,
      "grad_norm": 2.0621426105499268,
      "learning_rate": 1.3948732676956433e-06,
      "loss": 0.4523,
      "step": 318100
    },
    {
      "epoch": 130.21,
      "grad_norm": 1.6978216171264648,
      "learning_rate": 1.3947282219741516e-06,
      "loss": 0.4735,
      "step": 318110
    },
    {
      "epoch": 130.22,
      "grad_norm": 1.828719139099121,
      "learning_rate": 1.394583180599982e-06,
      "loss": 0.4576,
      "step": 318120
    },
    {
      "epoch": 130.22,
      "grad_norm": 1.811614751815796,
      "learning_rate": 1.394438143573798e-06,
      "loss": 0.4674,
      "step": 318130
    },
    {
      "epoch": 130.23,
      "grad_norm": 2.1741371154785156,
      "learning_rate": 1.3942931108962652e-06,
      "loss": 0.4486,
      "step": 318140
    },
    {
      "epoch": 130.23,
      "grad_norm": 2.112496852874756,
      "learning_rate": 1.3941480825680477e-06,
      "loss": 0.4896,
      "step": 318150
    },
    {
      "epoch": 130.23,
      "grad_norm": 1.5934571027755737,
      "learning_rate": 1.394003058589809e-06,
      "loss": 0.4539,
      "step": 318160
    },
    {
      "epoch": 130.24,
      "grad_norm": 2.015587329864502,
      "learning_rate": 1.393858038962214e-06,
      "loss": 0.449,
      "step": 318170
    },
    {
      "epoch": 130.24,
      "grad_norm": 2.280665874481201,
      "learning_rate": 1.3937130236859269e-06,
      "loss": 0.4433,
      "step": 318180
    },
    {
      "epoch": 130.25,
      "grad_norm": 2.1401991844177246,
      "learning_rate": 1.393568012761612e-06,
      "loss": 0.4441,
      "step": 318190
    },
    {
      "epoch": 130.25,
      "grad_norm": 2.010998487472534,
      "learning_rate": 1.3934230061899333e-06,
      "loss": 0.4611,
      "step": 318200
    },
    {
      "epoch": 130.25,
      "grad_norm": 1.7364616394042969,
      "learning_rate": 1.3932780039715555e-06,
      "loss": 0.4583,
      "step": 318210
    },
    {
      "epoch": 130.26,
      "grad_norm": 2.2128608226776123,
      "learning_rate": 1.3931330061071428e-06,
      "loss": 0.4551,
      "step": 318220
    },
    {
      "epoch": 130.26,
      "grad_norm": 2.0097057819366455,
      "learning_rate": 1.3929880125973596e-06,
      "loss": 0.4688,
      "step": 318230
    },
    {
      "epoch": 130.27,
      "grad_norm": 1.9252948760986328,
      "learning_rate": 1.3928430234428698e-06,
      "loss": 0.4525,
      "step": 318240
    },
    {
      "epoch": 130.27,
      "grad_norm": 1.7972681522369385,
      "learning_rate": 1.3926980386443376e-06,
      "loss": 0.464,
      "step": 318250
    },
    {
      "epoch": 130.27,
      "grad_norm": 2.29892635345459,
      "learning_rate": 1.3925530582024269e-06,
      "loss": 0.4524,
      "step": 318260
    },
    {
      "epoch": 130.28,
      "grad_norm": 1.6384936571121216,
      "learning_rate": 1.3924080821178023e-06,
      "loss": 0.4862,
      "step": 318270
    },
    {
      "epoch": 130.28,
      "grad_norm": 1.9260118007659912,
      "learning_rate": 1.3922631103911276e-06,
      "loss": 0.4508,
      "step": 318280
    },
    {
      "epoch": 130.29,
      "grad_norm": 2.0353574752807617,
      "learning_rate": 1.3921181430230675e-06,
      "loss": 0.4556,
      "step": 318290
    },
    {
      "epoch": 130.29,
      "grad_norm": 2.338684558868408,
      "learning_rate": 1.3919731800142853e-06,
      "loss": 0.4673,
      "step": 318300
    },
    {
      "epoch": 130.29,
      "grad_norm": 2.0908963680267334,
      "learning_rate": 1.391828221365445e-06,
      "loss": 0.4427,
      "step": 318310
    },
    {
      "epoch": 130.3,
      "grad_norm": 2.499368190765381,
      "learning_rate": 1.391683267077211e-06,
      "loss": 0.4484,
      "step": 318320
    },
    {
      "epoch": 130.3,
      "grad_norm": 2.0886449813842773,
      "learning_rate": 1.3915383171502472e-06,
      "loss": 0.4597,
      "step": 318330
    },
    {
      "epoch": 130.31,
      "grad_norm": 2.11716890335083,
      "learning_rate": 1.3913933715852177e-06,
      "loss": 0.4651,
      "step": 318340
    },
    {
      "epoch": 130.31,
      "grad_norm": 1.5952081680297852,
      "learning_rate": 1.3912484303827863e-06,
      "loss": 0.4574,
      "step": 318350
    },
    {
      "epoch": 130.32,
      "grad_norm": 1.838307499885559,
      "learning_rate": 1.3911034935436175e-06,
      "loss": 0.445,
      "step": 318360
    },
    {
      "epoch": 130.32,
      "grad_norm": 1.8121062517166138,
      "learning_rate": 1.3909585610683744e-06,
      "loss": 0.4481,
      "step": 318370
    },
    {
      "epoch": 130.32,
      "grad_norm": 1.860685110092163,
      "learning_rate": 1.390813632957721e-06,
      "loss": 0.4503,
      "step": 318380
    },
    {
      "epoch": 130.33,
      "grad_norm": 2.2308847904205322,
      "learning_rate": 1.3906687092123221e-06,
      "loss": 0.4577,
      "step": 318390
    },
    {
      "epoch": 130.33,
      "grad_norm": 2.3914992809295654,
      "learning_rate": 1.3905237898328408e-06,
      "loss": 0.4761,
      "step": 318400
    },
    {
      "epoch": 130.34,
      "grad_norm": 1.8993314504623413,
      "learning_rate": 1.3903788748199411e-06,
      "loss": 0.4529,
      "step": 318410
    },
    {
      "epoch": 130.34,
      "grad_norm": 2.0136094093322754,
      "learning_rate": 1.3902339641742872e-06,
      "loss": 0.4714,
      "step": 318420
    },
    {
      "epoch": 130.34,
      "grad_norm": 1.8453165292739868,
      "learning_rate": 1.390089057896543e-06,
      "loss": 0.4619,
      "step": 318430
    },
    {
      "epoch": 130.35,
      "grad_norm": 2.0797276496887207,
      "learning_rate": 1.389944155987371e-06,
      "loss": 0.4652,
      "step": 318440
    },
    {
      "epoch": 130.35,
      "grad_norm": 2.185687303543091,
      "learning_rate": 1.3897992584474363e-06,
      "loss": 0.457,
      "step": 318450
    },
    {
      "epoch": 130.36,
      "grad_norm": 2.0717151165008545,
      "learning_rate": 1.3896543652774018e-06,
      "loss": 0.4668,
      "step": 318460
    },
    {
      "epoch": 130.36,
      "grad_norm": 2.1331543922424316,
      "learning_rate": 1.3895094764779323e-06,
      "loss": 0.4598,
      "step": 318470
    },
    {
      "epoch": 130.36,
      "grad_norm": 2.108149766921997,
      "learning_rate": 1.3893645920496902e-06,
      "loss": 0.4494,
      "step": 318480
    },
    {
      "epoch": 130.37,
      "grad_norm": 1.8174169063568115,
      "learning_rate": 1.3892197119933401e-06,
      "loss": 0.4629,
      "step": 318490
    },
    {
      "epoch": 130.37,
      "grad_norm": 1.6602203845977783,
      "learning_rate": 1.3890748363095454e-06,
      "loss": 0.4512,
      "step": 318500
    },
    {
      "epoch": 130.38,
      "grad_norm": 1.6654876470565796,
      "learning_rate": 1.3889299649989701e-06,
      "loss": 0.4577,
      "step": 318510
    },
    {
      "epoch": 130.38,
      "grad_norm": 1.6609139442443848,
      "learning_rate": 1.3887850980622773e-06,
      "loss": 0.4638,
      "step": 318520
    },
    {
      "epoch": 130.38,
      "grad_norm": 1.6430556774139404,
      "learning_rate": 1.3886402355001307e-06,
      "loss": 0.4565,
      "step": 318530
    },
    {
      "epoch": 130.39,
      "grad_norm": 2.0946335792541504,
      "learning_rate": 1.388495377313194e-06,
      "loss": 0.4441,
      "step": 318540
    },
    {
      "epoch": 130.39,
      "grad_norm": 2.0320637226104736,
      "learning_rate": 1.388350523502131e-06,
      "loss": 0.442,
      "step": 318550
    },
    {
      "epoch": 130.4,
      "grad_norm": 2.1740012168884277,
      "learning_rate": 1.3882056740676046e-06,
      "loss": 0.4606,
      "step": 318560
    },
    {
      "epoch": 130.4,
      "grad_norm": 1.9876816272735596,
      "learning_rate": 1.3880608290102785e-06,
      "loss": 0.4401,
      "step": 318570
    },
    {
      "epoch": 130.41,
      "grad_norm": 1.9142992496490479,
      "learning_rate": 1.387915988330817e-06,
      "loss": 0.4513,
      "step": 318580
    },
    {
      "epoch": 130.41,
      "grad_norm": 1.9016896486282349,
      "learning_rate": 1.3877711520298832e-06,
      "loss": 0.451,
      "step": 318590
    },
    {
      "epoch": 130.41,
      "grad_norm": 2.265475034713745,
      "learning_rate": 1.3876263201081403e-06,
      "loss": 0.4578,
      "step": 318600
    },
    {
      "epoch": 130.42,
      "grad_norm": 2.1794629096984863,
      "learning_rate": 1.387481492566252e-06,
      "loss": 0.4572,
      "step": 318610
    },
    {
      "epoch": 130.42,
      "grad_norm": 1.7676024436950684,
      "learning_rate": 1.3873366694048815e-06,
      "loss": 0.4641,
      "step": 318620
    },
    {
      "epoch": 130.43,
      "grad_norm": 1.9301809072494507,
      "learning_rate": 1.3871918506246924e-06,
      "loss": 0.4663,
      "step": 318630
    },
    {
      "epoch": 130.43,
      "grad_norm": 1.7094675302505493,
      "learning_rate": 1.3870470362263481e-06,
      "loss": 0.4417,
      "step": 318640
    },
    {
      "epoch": 130.43,
      "grad_norm": 2.008166551589966,
      "learning_rate": 1.3869022262105123e-06,
      "loss": 0.4343,
      "step": 318650
    },
    {
      "epoch": 130.44,
      "grad_norm": 2.0723209381103516,
      "learning_rate": 1.3867574205778477e-06,
      "loss": 0.4545,
      "step": 318660
    },
    {
      "epoch": 130.44,
      "grad_norm": 1.9904159307479858,
      "learning_rate": 1.3866126193290176e-06,
      "loss": 0.4531,
      "step": 318670
    },
    {
      "epoch": 130.45,
      "grad_norm": 1.9028366804122925,
      "learning_rate": 1.3864678224646858e-06,
      "loss": 0.454,
      "step": 318680
    },
    {
      "epoch": 130.45,
      "grad_norm": 2.7763381004333496,
      "learning_rate": 1.3863230299855154e-06,
      "loss": 0.4537,
      "step": 318690
    },
    {
      "epoch": 130.45,
      "grad_norm": 2.5557668209075928,
      "learning_rate": 1.3861782418921693e-06,
      "loss": 0.456,
      "step": 318700
    },
    {
      "epoch": 130.46,
      "grad_norm": 1.8402436971664429,
      "learning_rate": 1.3860334581853115e-06,
      "loss": 0.4547,
      "step": 318710
    },
    {
      "epoch": 130.46,
      "grad_norm": 1.8485280275344849,
      "learning_rate": 1.3858886788656052e-06,
      "loss": 0.4588,
      "step": 318720
    },
    {
      "epoch": 130.47,
      "grad_norm": 2.10473370552063,
      "learning_rate": 1.3857439039337126e-06,
      "loss": 0.4503,
      "step": 318730
    },
    {
      "epoch": 130.47,
      "grad_norm": 1.6285028457641602,
      "learning_rate": 1.3855991333902982e-06,
      "loss": 0.4531,
      "step": 318740
    },
    {
      "epoch": 130.47,
      "grad_norm": 1.952045202255249,
      "learning_rate": 1.3854543672360243e-06,
      "loss": 0.4758,
      "step": 318750
    },
    {
      "epoch": 130.48,
      "grad_norm": 11.870698928833008,
      "learning_rate": 1.385309605471553e-06,
      "loss": 0.4388,
      "step": 318760
    },
    {
      "epoch": 130.48,
      "grad_norm": 2.3016037940979004,
      "learning_rate": 1.38516484809755e-06,
      "loss": 0.4628,
      "step": 318770
    },
    {
      "epoch": 130.49,
      "grad_norm": 2.124094247817993,
      "learning_rate": 1.385020095114677e-06,
      "loss": 0.4587,
      "step": 318780
    },
    {
      "epoch": 130.49,
      "grad_norm": 1.884822964668274,
      "learning_rate": 1.3848753465235979e-06,
      "loss": 0.4598,
      "step": 318790
    },
    {
      "epoch": 130.5,
      "grad_norm": 1.8995790481567383,
      "learning_rate": 1.384730602324974e-06,
      "loss": 0.4498,
      "step": 318800
    },
    {
      "epoch": 130.5,
      "grad_norm": 1.78218674659729,
      "learning_rate": 1.38458586251947e-06,
      "loss": 0.4561,
      "step": 318810
    },
    {
      "epoch": 130.5,
      "grad_norm": 2.149223566055298,
      "learning_rate": 1.3844411271077483e-06,
      "loss": 0.4652,
      "step": 318820
    },
    {
      "epoch": 130.51,
      "grad_norm": 2.0767970085144043,
      "learning_rate": 1.3842963960904717e-06,
      "loss": 0.4758,
      "step": 318830
    },
    {
      "epoch": 130.51,
      "grad_norm": 2.4307901859283447,
      "learning_rate": 1.384151669468304e-06,
      "loss": 0.4536,
      "step": 318840
    },
    {
      "epoch": 130.52,
      "grad_norm": 3.2915871143341064,
      "learning_rate": 1.384006947241907e-06,
      "loss": 0.461,
      "step": 318850
    },
    {
      "epoch": 130.52,
      "grad_norm": 1.8238528966903687,
      "learning_rate": 1.3838622294119444e-06,
      "loss": 0.4417,
      "step": 318860
    },
    {
      "epoch": 130.52,
      "grad_norm": 2.285041570663452,
      "learning_rate": 1.38371751597908e-06,
      "loss": 0.4649,
      "step": 318870
    },
    {
      "epoch": 130.53,
      "grad_norm": 2.0501718521118164,
      "learning_rate": 1.3835728069439746e-06,
      "loss": 0.48,
      "step": 318880
    },
    {
      "epoch": 130.53,
      "grad_norm": 1.772481083869934,
      "learning_rate": 1.3834281023072926e-06,
      "loss": 0.4512,
      "step": 318890
    },
    {
      "epoch": 130.54,
      "grad_norm": 2.2926976680755615,
      "learning_rate": 1.3832834020696965e-06,
      "loss": 0.4521,
      "step": 318900
    },
    {
      "epoch": 130.54,
      "grad_norm": 1.6266096830368042,
      "learning_rate": 1.383138706231849e-06,
      "loss": 0.4721,
      "step": 318910
    },
    {
      "epoch": 130.54,
      "grad_norm": 2.603694438934326,
      "learning_rate": 1.3829940147944134e-06,
      "loss": 0.4431,
      "step": 318920
    },
    {
      "epoch": 130.55,
      "grad_norm": 2.264841318130493,
      "learning_rate": 1.3828493277580517e-06,
      "loss": 0.4488,
      "step": 318930
    },
    {
      "epoch": 130.55,
      "grad_norm": 2.0462896823883057,
      "learning_rate": 1.3827046451234274e-06,
      "loss": 0.4526,
      "step": 318940
    },
    {
      "epoch": 130.56,
      "grad_norm": 2.0080995559692383,
      "learning_rate": 1.3825599668912027e-06,
      "loss": 0.458,
      "step": 318950
    },
    {
      "epoch": 130.56,
      "grad_norm": 1.7280267477035522,
      "learning_rate": 1.382415293062041e-06,
      "loss": 0.4611,
      "step": 318960
    },
    {
      "epoch": 130.56,
      "grad_norm": 1.756719708442688,
      "learning_rate": 1.3822706236366047e-06,
      "loss": 0.4602,
      "step": 318970
    },
    {
      "epoch": 130.57,
      "grad_norm": 2.2361462116241455,
      "learning_rate": 1.3821259586155564e-06,
      "loss": 0.4543,
      "step": 318980
    },
    {
      "epoch": 130.57,
      "grad_norm": 1.8938701152801514,
      "learning_rate": 1.381981297999559e-06,
      "loss": 0.4478,
      "step": 318990
    },
    {
      "epoch": 130.58,
      "grad_norm": 1.960091471672058,
      "learning_rate": 1.3818366417892753e-06,
      "loss": 0.4527,
      "step": 319000
    },
    {
      "epoch": 130.58,
      "grad_norm": 2.044633150100708,
      "learning_rate": 1.381691989985368e-06,
      "loss": 0.4617,
      "step": 319010
    },
    {
      "epoch": 130.59,
      "grad_norm": 2.1744742393493652,
      "learning_rate": 1.3815473425884988e-06,
      "loss": 0.4527,
      "step": 319020
    },
    {
      "epoch": 130.59,
      "grad_norm": 1.9398326873779297,
      "learning_rate": 1.3814026995993313e-06,
      "loss": 0.4387,
      "step": 319030
    },
    {
      "epoch": 130.59,
      "grad_norm": 2.083702564239502,
      "learning_rate": 1.3812580610185274e-06,
      "loss": 0.4575,
      "step": 319040
    },
    {
      "epoch": 130.6,
      "grad_norm": 2.285834550857544,
      "learning_rate": 1.3811134268467502e-06,
      "loss": 0.4507,
      "step": 319050
    },
    {
      "epoch": 130.6,
      "grad_norm": 1.6090136766433716,
      "learning_rate": 1.3809687970846622e-06,
      "loss": 0.4654,
      "step": 319060
    },
    {
      "epoch": 130.61,
      "grad_norm": 2.1002519130706787,
      "learning_rate": 1.3808241717329254e-06,
      "loss": 0.4571,
      "step": 319070
    },
    {
      "epoch": 130.61,
      "grad_norm": 1.7151087522506714,
      "learning_rate": 1.380679550792203e-06,
      "loss": 0.4679,
      "step": 319080
    },
    {
      "epoch": 130.61,
      "grad_norm": 2.225661039352417,
      "learning_rate": 1.3805349342631573e-06,
      "loss": 0.4492,
      "step": 319090
    },
    {
      "epoch": 130.62,
      "grad_norm": 2.2781763076782227,
      "learning_rate": 1.3803903221464507e-06,
      "loss": 0.4783,
      "step": 319100
    },
    {
      "epoch": 130.62,
      "grad_norm": 2.555852174758911,
      "learning_rate": 1.3802457144427456e-06,
      "loss": 0.47,
      "step": 319110
    },
    {
      "epoch": 130.63,
      "grad_norm": 1.9940129518508911,
      "learning_rate": 1.380101111152704e-06,
      "loss": 0.4599,
      "step": 319120
    },
    {
      "epoch": 130.63,
      "grad_norm": 2.0802974700927734,
      "learning_rate": 1.379956512276989e-06,
      "loss": 0.4492,
      "step": 319130
    },
    {
      "epoch": 130.63,
      "grad_norm": 2.118864059448242,
      "learning_rate": 1.3798119178162615e-06,
      "loss": 0.4673,
      "step": 319140
    },
    {
      "epoch": 130.64,
      "grad_norm": 1.7939400672912598,
      "learning_rate": 1.3796673277711865e-06,
      "loss": 0.4559,
      "step": 319150
    },
    {
      "epoch": 130.64,
      "grad_norm": 2.0351970195770264,
      "learning_rate": 1.3795227421424248e-06,
      "loss": 0.4601,
      "step": 319160
    },
    {
      "epoch": 130.65,
      "grad_norm": 2.137603998184204,
      "learning_rate": 1.3793781609306385e-06,
      "loss": 0.4531,
      "step": 319170
    },
    {
      "epoch": 130.65,
      "grad_norm": 2.5614912509918213,
      "learning_rate": 1.3792335841364904e-06,
      "loss": 0.4652,
      "step": 319180
    },
    {
      "epoch": 130.65,
      "grad_norm": 2.4350836277008057,
      "learning_rate": 1.3790890117606425e-06,
      "loss": 0.5006,
      "step": 319190
    },
    {
      "epoch": 130.66,
      "grad_norm": 2.112297534942627,
      "learning_rate": 1.3789444438037572e-06,
      "loss": 0.4255,
      "step": 319200
    },
    {
      "epoch": 130.66,
      "grad_norm": 2.3651866912841797,
      "learning_rate": 1.378799880266497e-06,
      "loss": 0.4523,
      "step": 319210
    },
    {
      "epoch": 130.67,
      "grad_norm": 2.1191723346710205,
      "learning_rate": 1.3786553211495236e-06,
      "loss": 0.4417,
      "step": 319220
    },
    {
      "epoch": 130.67,
      "grad_norm": 1.775649070739746,
      "learning_rate": 1.3785107664535e-06,
      "loss": 0.4593,
      "step": 319230
    },
    {
      "epoch": 130.68,
      "grad_norm": 1.4508538246154785,
      "learning_rate": 1.3783662161790874e-06,
      "loss": 0.451,
      "step": 319240
    },
    {
      "epoch": 130.68,
      "grad_norm": 2.430171251296997,
      "learning_rate": 1.3782216703269482e-06,
      "loss": 0.4589,
      "step": 319250
    },
    {
      "epoch": 130.68,
      "grad_norm": 2.028486490249634,
      "learning_rate": 1.3780771288977448e-06,
      "loss": 0.443,
      "step": 319260
    },
    {
      "epoch": 130.69,
      "grad_norm": 1.6289697885513306,
      "learning_rate": 1.3779325918921397e-06,
      "loss": 0.4703,
      "step": 319270
    },
    {
      "epoch": 130.69,
      "grad_norm": 2.0078608989715576,
      "learning_rate": 1.3777880593107943e-06,
      "loss": 0.4798,
      "step": 319280
    },
    {
      "epoch": 130.7,
      "grad_norm": 1.8774585723876953,
      "learning_rate": 1.377643531154371e-06,
      "loss": 0.4673,
      "step": 319290
    },
    {
      "epoch": 130.7,
      "grad_norm": 2.1817967891693115,
      "learning_rate": 1.3774990074235316e-06,
      "loss": 0.4766,
      "step": 319300
    },
    {
      "epoch": 130.7,
      "grad_norm": 2.1595547199249268,
      "learning_rate": 1.3773544881189385e-06,
      "loss": 0.4573,
      "step": 319310
    },
    {
      "epoch": 130.71,
      "grad_norm": 2.4046249389648438,
      "learning_rate": 1.377209973241254e-06,
      "loss": 0.4589,
      "step": 319320
    },
    {
      "epoch": 130.71,
      "grad_norm": 1.7365866899490356,
      "learning_rate": 1.377065462791138e-06,
      "loss": 0.4598,
      "step": 319330
    },
    {
      "epoch": 130.72,
      "grad_norm": 1.7261178493499756,
      "learning_rate": 1.3769209567692555e-06,
      "loss": 0.4392,
      "step": 319340
    },
    {
      "epoch": 130.72,
      "grad_norm": 2.268568992614746,
      "learning_rate": 1.376776455176267e-06,
      "loss": 0.4553,
      "step": 319350
    },
    {
      "epoch": 130.72,
      "grad_norm": 1.9514998197555542,
      "learning_rate": 1.3766319580128344e-06,
      "loss": 0.4464,
      "step": 319360
    },
    {
      "epoch": 130.73,
      "grad_norm": 2.3061015605926514,
      "learning_rate": 1.3764874652796204e-06,
      "loss": 0.431,
      "step": 319370
    },
    {
      "epoch": 130.73,
      "grad_norm": 1.9904669523239136,
      "learning_rate": 1.3763429769772858e-06,
      "loss": 0.4435,
      "step": 319380
    },
    {
      "epoch": 130.74,
      "grad_norm": 1.6665464639663696,
      "learning_rate": 1.3761984931064932e-06,
      "loss": 0.4594,
      "step": 319390
    },
    {
      "epoch": 130.74,
      "grad_norm": 2.277106523513794,
      "learning_rate": 1.3760540136679037e-06,
      "loss": 0.4492,
      "step": 319400
    },
    {
      "epoch": 130.74,
      "grad_norm": 1.8328806161880493,
      "learning_rate": 1.3759095386621802e-06,
      "loss": 0.4693,
      "step": 319410
    },
    {
      "epoch": 130.75,
      "grad_norm": 1.977952480316162,
      "learning_rate": 1.3757650680899834e-06,
      "loss": 0.4397,
      "step": 319420
    },
    {
      "epoch": 130.75,
      "grad_norm": 1.9285107851028442,
      "learning_rate": 1.3756206019519762e-06,
      "loss": 0.4659,
      "step": 319430
    },
    {
      "epoch": 130.76,
      "grad_norm": 1.7740695476531982,
      "learning_rate": 1.3754761402488196e-06,
      "loss": 0.464,
      "step": 319440
    },
    {
      "epoch": 130.76,
      "grad_norm": 1.9600247144699097,
      "learning_rate": 1.3753316829811754e-06,
      "loss": 0.4694,
      "step": 319450
    },
    {
      "epoch": 130.77,
      "grad_norm": 2.134276866912842,
      "learning_rate": 1.375187230149706e-06,
      "loss": 0.4547,
      "step": 319460
    },
    {
      "epoch": 130.77,
      "grad_norm": 2.0725653171539307,
      "learning_rate": 1.3750427817550723e-06,
      "loss": 0.4603,
      "step": 319470
    },
    {
      "epoch": 130.77,
      "grad_norm": 2.1167330741882324,
      "learning_rate": 1.3748983377979365e-06,
      "loss": 0.4643,
      "step": 319480
    },
    {
      "epoch": 130.78,
      "grad_norm": 2.089015007019043,
      "learning_rate": 1.37475389827896e-06,
      "loss": 0.4758,
      "step": 319490
    },
    {
      "epoch": 130.78,
      "grad_norm": 2.0578835010528564,
      "learning_rate": 1.3746094631988044e-06,
      "loss": 0.4355,
      "step": 319500
    },
    {
      "epoch": 130.79,
      "grad_norm": 1.9638400077819824,
      "learning_rate": 1.3744650325581317e-06,
      "loss": 0.4703,
      "step": 319510
    },
    {
      "epoch": 130.79,
      "grad_norm": 2.6555233001708984,
      "learning_rate": 1.374320606357603e-06,
      "loss": 0.4721,
      "step": 319520
    },
    {
      "epoch": 130.79,
      "grad_norm": 1.8092254400253296,
      "learning_rate": 1.3741761845978802e-06,
      "loss": 0.4447,
      "step": 319530
    },
    {
      "epoch": 130.8,
      "grad_norm": 2.0061333179473877,
      "learning_rate": 1.3740317672796255e-06,
      "loss": 0.4463,
      "step": 319540
    },
    {
      "epoch": 130.8,
      "grad_norm": 1.8386198282241821,
      "learning_rate": 1.3738873544034996e-06,
      "loss": 0.4502,
      "step": 319550
    },
    {
      "epoch": 130.81,
      "grad_norm": 2.638392210006714,
      "learning_rate": 1.3737429459701643e-06,
      "loss": 0.4445,
      "step": 319560
    },
    {
      "epoch": 130.81,
      "grad_norm": 2.0583581924438477,
      "learning_rate": 1.373598541980281e-06,
      "loss": 0.4761,
      "step": 319570
    },
    {
      "epoch": 130.81,
      "grad_norm": 2.365424871444702,
      "learning_rate": 1.3734541424345113e-06,
      "loss": 0.4569,
      "step": 319580
    },
    {
      "epoch": 130.82,
      "grad_norm": 1.7384101152420044,
      "learning_rate": 1.373309747333517e-06,
      "loss": 0.4595,
      "step": 319590
    },
    {
      "epoch": 130.82,
      "grad_norm": 1.799208164215088,
      "learning_rate": 1.373165356677959e-06,
      "loss": 0.4383,
      "step": 319600
    },
    {
      "epoch": 130.83,
      "grad_norm": 1.9926469326019287,
      "learning_rate": 1.3730209704684987e-06,
      "loss": 0.433,
      "step": 319610
    },
    {
      "epoch": 130.83,
      "grad_norm": 6.906030654907227,
      "learning_rate": 1.3728765887057978e-06,
      "loss": 0.4659,
      "step": 319620
    },
    {
      "epoch": 130.84,
      "grad_norm": 2.1240804195404053,
      "learning_rate": 1.372732211390518e-06,
      "loss": 0.4747,
      "step": 319630
    },
    {
      "epoch": 130.84,
      "grad_norm": 2.2876598834991455,
      "learning_rate": 1.3725878385233197e-06,
      "loss": 0.468,
      "step": 319640
    },
    {
      "epoch": 130.84,
      "grad_norm": 1.8537696599960327,
      "learning_rate": 1.3724434701048651e-06,
      "loss": 0.4504,
      "step": 319650
    },
    {
      "epoch": 130.85,
      "grad_norm": 1.8030080795288086,
      "learning_rate": 1.3722991061358155e-06,
      "loss": 0.4539,
      "step": 319660
    },
    {
      "epoch": 130.85,
      "grad_norm": 1.6448911428451538,
      "learning_rate": 1.3721547466168318e-06,
      "loss": 0.4383,
      "step": 319670
    },
    {
      "epoch": 130.86,
      "grad_norm": 1.9166957139968872,
      "learning_rate": 1.372010391548576e-06,
      "loss": 0.4732,
      "step": 319680
    },
    {
      "epoch": 130.86,
      "grad_norm": 2.0117173194885254,
      "learning_rate": 1.3718660409317083e-06,
      "loss": 0.4584,
      "step": 319690
    },
    {
      "epoch": 130.86,
      "grad_norm": 1.6886188983917236,
      "learning_rate": 1.3717216947668897e-06,
      "loss": 0.4488,
      "step": 319700
    },
    {
      "epoch": 130.87,
      "grad_norm": 2.5066475868225098,
      "learning_rate": 1.3715773530547832e-06,
      "loss": 0.4427,
      "step": 319710
    },
    {
      "epoch": 130.87,
      "grad_norm": 1.828673005104065,
      "learning_rate": 1.3714330157960489e-06,
      "loss": 0.4498,
      "step": 319720
    },
    {
      "epoch": 130.88,
      "grad_norm": 1.9959299564361572,
      "learning_rate": 1.3712886829913484e-06,
      "loss": 0.4407,
      "step": 319730
    },
    {
      "epoch": 130.88,
      "grad_norm": 2.4241604804992676,
      "learning_rate": 1.3711443546413427e-06,
      "loss": 0.468,
      "step": 319740
    },
    {
      "epoch": 130.88,
      "grad_norm": 1.9892915487289429,
      "learning_rate": 1.3710000307466923e-06,
      "loss": 0.4705,
      "step": 319750
    },
    {
      "epoch": 130.89,
      "grad_norm": 1.9755644798278809,
      "learning_rate": 1.3708557113080591e-06,
      "loss": 0.4587,
      "step": 319760
    },
    {
      "epoch": 130.89,
      "grad_norm": 2.1446523666381836,
      "learning_rate": 1.370711396326104e-06,
      "loss": 0.4697,
      "step": 319770
    },
    {
      "epoch": 130.9,
      "grad_norm": 2.0015132427215576,
      "learning_rate": 1.370567085801488e-06,
      "loss": 0.4707,
      "step": 319780
    },
    {
      "epoch": 130.9,
      "grad_norm": 1.8475595712661743,
      "learning_rate": 1.370422779734872e-06,
      "loss": 0.4573,
      "step": 319790
    },
    {
      "epoch": 130.9,
      "grad_norm": 1.8782577514648438,
      "learning_rate": 1.3702784781269176e-06,
      "loss": 0.449,
      "step": 319800
    },
    {
      "epoch": 130.91,
      "grad_norm": 1.774057388305664,
      "learning_rate": 1.370134180978285e-06,
      "loss": 0.4631,
      "step": 319810
    },
    {
      "epoch": 130.91,
      "grad_norm": 2.000617504119873,
      "learning_rate": 1.3699898882896365e-06,
      "loss": 0.4581,
      "step": 319820
    },
    {
      "epoch": 130.92,
      "grad_norm": 1.9484882354736328,
      "learning_rate": 1.369845600061632e-06,
      "loss": 0.4517,
      "step": 319830
    },
    {
      "epoch": 130.92,
      "grad_norm": 1.9915577173233032,
      "learning_rate": 1.3697013162949325e-06,
      "loss": 0.4473,
      "step": 319840
    },
    {
      "epoch": 130.93,
      "grad_norm": 2.0277366638183594,
      "learning_rate": 1.369557036990199e-06,
      "loss": 0.4827,
      "step": 319850
    },
    {
      "epoch": 130.93,
      "grad_norm": 1.8914830684661865,
      "learning_rate": 1.3694127621480928e-06,
      "loss": 0.4502,
      "step": 319860
    },
    {
      "epoch": 130.93,
      "grad_norm": 2.036954641342163,
      "learning_rate": 1.3692684917692746e-06,
      "loss": 0.4684,
      "step": 319870
    },
    {
      "epoch": 130.94,
      "grad_norm": 1.6993141174316406,
      "learning_rate": 1.3691242258544054e-06,
      "loss": 0.4344,
      "step": 319880
    },
    {
      "epoch": 130.94,
      "grad_norm": 2.396704912185669,
      "learning_rate": 1.3689799644041451e-06,
      "loss": 0.4458,
      "step": 319890
    },
    {
      "epoch": 130.95,
      "grad_norm": 1.824302077293396,
      "learning_rate": 1.3688357074191561e-06,
      "loss": 0.4472,
      "step": 319900
    },
    {
      "epoch": 130.95,
      "grad_norm": 2.335240125656128,
      "learning_rate": 1.3686914549000981e-06,
      "loss": 0.4722,
      "step": 319910
    },
    {
      "epoch": 130.95,
      "grad_norm": 1.7054812908172607,
      "learning_rate": 1.368547206847633e-06,
      "loss": 0.4495,
      "step": 319920
    },
    {
      "epoch": 130.96,
      "grad_norm": 2.0861761569976807,
      "learning_rate": 1.3684029632624205e-06,
      "loss": 0.4412,
      "step": 319930
    },
    {
      "epoch": 130.96,
      "grad_norm": 1.805222988128662,
      "learning_rate": 1.3682587241451216e-06,
      "loss": 0.4562,
      "step": 319940
    },
    {
      "epoch": 130.97,
      "grad_norm": 2.962845802307129,
      "learning_rate": 1.3681144894963973e-06,
      "loss": 0.4494,
      "step": 319950
    },
    {
      "epoch": 130.97,
      "grad_norm": 2.2551429271698,
      "learning_rate": 1.3679702593169081e-06,
      "loss": 0.4363,
      "step": 319960
    },
    {
      "epoch": 130.97,
      "grad_norm": 2.100344181060791,
      "learning_rate": 1.3678260336073147e-06,
      "loss": 0.4512,
      "step": 319970
    },
    {
      "epoch": 130.98,
      "grad_norm": 1.9218398332595825,
      "learning_rate": 1.367681812368278e-06,
      "loss": 0.4561,
      "step": 319980
    },
    {
      "epoch": 130.98,
      "grad_norm": 1.668343186378479,
      "learning_rate": 1.3675375956004584e-06,
      "loss": 0.4601,
      "step": 319990
    },
    {
      "epoch": 130.99,
      "grad_norm": 1.7001104354858398,
      "learning_rate": 1.3673933833045175e-06,
      "loss": 0.4504,
      "step": 320000
    },
    {
      "epoch": 130.99,
      "grad_norm": 2.127041816711426,
      "learning_rate": 1.3672491754811145e-06,
      "loss": 0.4564,
      "step": 320010
    },
    {
      "epoch": 130.99,
      "grad_norm": 2.2969014644622803,
      "learning_rate": 1.36710497213091e-06,
      "loss": 0.4767,
      "step": 320020
    },
    {
      "epoch": 131.0,
      "grad_norm": 1.962369441986084,
      "learning_rate": 1.3669607732545657e-06,
      "loss": 0.4457,
      "step": 320030
    },
    {
      "epoch": 131.0,
      "eval_loss": 0.45777833461761475,
      "eval_runtime": 56.5169,
      "eval_samples_per_second": 61.026,
      "eval_steps_per_second": 7.644,
      "step": 320033
    },
    {
      "epoch": 131.0,
      "grad_norm": 1.9324928522109985,
      "learning_rate": 1.3668165788527413e-06,
      "loss": 0.4734,
      "step": 320040
    },
    {
      "epoch": 131.01,
      "grad_norm": 4.3801751136779785,
      "learning_rate": 1.3666723889260975e-06,
      "loss": 0.4595,
      "step": 320050
    },
    {
      "epoch": 131.01,
      "grad_norm": 1.7388793230056763,
      "learning_rate": 1.3665282034752948e-06,
      "loss": 0.4541,
      "step": 320060
    },
    {
      "epoch": 131.02,
      "grad_norm": 2.085611343383789,
      "learning_rate": 1.3663840225009944e-06,
      "loss": 0.45,
      "step": 320070
    },
    {
      "epoch": 131.02,
      "grad_norm": 2.0877492427825928,
      "learning_rate": 1.3662398460038547e-06,
      "loss": 0.4426,
      "step": 320080
    },
    {
      "epoch": 131.02,
      "grad_norm": 2.103907346725464,
      "learning_rate": 1.3660956739845388e-06,
      "loss": 0.4653,
      "step": 320090
    },
    {
      "epoch": 131.03,
      "grad_norm": 2.15118408203125,
      "learning_rate": 1.3659515064437057e-06,
      "loss": 0.4436,
      "step": 320100
    },
    {
      "epoch": 131.03,
      "grad_norm": 1.9031912088394165,
      "learning_rate": 1.3658073433820165e-06,
      "loss": 0.4381,
      "step": 320110
    },
    {
      "epoch": 131.04,
      "grad_norm": 2.0357587337493896,
      "learning_rate": 1.3656631848001306e-06,
      "loss": 0.4393,
      "step": 320120
    },
    {
      "epoch": 131.04,
      "grad_norm": 1.9069874286651611,
      "learning_rate": 1.3655190306987102e-06,
      "loss": 0.4538,
      "step": 320130
    },
    {
      "epoch": 131.04,
      "grad_norm": 1.8428218364715576,
      "learning_rate": 1.3653748810784138e-06,
      "loss": 0.4834,
      "step": 320140
    },
    {
      "epoch": 131.05,
      "grad_norm": 2.180798292160034,
      "learning_rate": 1.3652307359399017e-06,
      "loss": 0.4602,
      "step": 320150
    },
    {
      "epoch": 131.05,
      "grad_norm": 2.4246127605438232,
      "learning_rate": 1.3650865952838349e-06,
      "loss": 0.4451,
      "step": 320160
    },
    {
      "epoch": 131.06,
      "grad_norm": 2.055447816848755,
      "learning_rate": 1.3649424591108737e-06,
      "loss": 0.4534,
      "step": 320170
    },
    {
      "epoch": 131.06,
      "grad_norm": 1.9350417852401733,
      "learning_rate": 1.3647983274216784e-06,
      "loss": 0.4622,
      "step": 320180
    },
    {
      "epoch": 131.06,
      "grad_norm": 2.4437265396118164,
      "learning_rate": 1.364654200216909e-06,
      "loss": 0.483,
      "step": 320190
    },
    {
      "epoch": 131.07,
      "grad_norm": 2.2422568798065186,
      "learning_rate": 1.3645100774972255e-06,
      "loss": 0.4578,
      "step": 320200
    },
    {
      "epoch": 131.07,
      "grad_norm": 1.895614504814148,
      "learning_rate": 1.3643659592632886e-06,
      "loss": 0.4522,
      "step": 320210
    },
    {
      "epoch": 131.08,
      "grad_norm": 1.8773618936538696,
      "learning_rate": 1.364221845515759e-06,
      "loss": 0.4514,
      "step": 320220
    },
    {
      "epoch": 131.08,
      "grad_norm": 1.988429069519043,
      "learning_rate": 1.3640777362552952e-06,
      "loss": 0.4526,
      "step": 320230
    },
    {
      "epoch": 131.08,
      "grad_norm": 2.2700283527374268,
      "learning_rate": 1.363933631482559e-06,
      "loss": 0.4644,
      "step": 320240
    },
    {
      "epoch": 131.09,
      "grad_norm": 1.4300607442855835,
      "learning_rate": 1.3637895311982093e-06,
      "loss": 0.4638,
      "step": 320250
    },
    {
      "epoch": 131.09,
      "grad_norm": 2.649122953414917,
      "learning_rate": 1.3636454354029072e-06,
      "loss": 0.4447,
      "step": 320260
    },
    {
      "epoch": 131.1,
      "grad_norm": 1.813625693321228,
      "learning_rate": 1.3635013440973122e-06,
      "loss": 0.4602,
      "step": 320270
    },
    {
      "epoch": 131.1,
      "grad_norm": 1.896142840385437,
      "learning_rate": 1.3633572572820842e-06,
      "loss": 0.4569,
      "step": 320280
    },
    {
      "epoch": 131.11,
      "grad_norm": 1.9338676929473877,
      "learning_rate": 1.3632131749578838e-06,
      "loss": 0.4639,
      "step": 320290
    },
    {
      "epoch": 131.11,
      "grad_norm": 1.9786460399627686,
      "learning_rate": 1.3630690971253708e-06,
      "loss": 0.4517,
      "step": 320300
    },
    {
      "epoch": 131.11,
      "grad_norm": 2.505844831466675,
      "learning_rate": 1.3629250237852048e-06,
      "loss": 0.4488,
      "step": 320310
    },
    {
      "epoch": 131.12,
      "grad_norm": 2.2194180488586426,
      "learning_rate": 1.3627809549380465e-06,
      "loss": 0.4565,
      "step": 320320
    },
    {
      "epoch": 131.12,
      "grad_norm": 1.9238386154174805,
      "learning_rate": 1.3626368905845553e-06,
      "loss": 0.4459,
      "step": 320330
    },
    {
      "epoch": 131.13,
      "grad_norm": 2.2751612663269043,
      "learning_rate": 1.3624928307253913e-06,
      "loss": 0.448,
      "step": 320340
    },
    {
      "epoch": 131.13,
      "grad_norm": 2.122292995452881,
      "learning_rate": 1.3623487753612143e-06,
      "loss": 0.4564,
      "step": 320350
    },
    {
      "epoch": 131.13,
      "grad_norm": 2.489931344985962,
      "learning_rate": 1.3622047244926846e-06,
      "loss": 0.4482,
      "step": 320360
    },
    {
      "epoch": 131.14,
      "grad_norm": 1.741986632347107,
      "learning_rate": 1.3620606781204616e-06,
      "loss": 0.4464,
      "step": 320370
    },
    {
      "epoch": 131.14,
      "grad_norm": 1.7846232652664185,
      "learning_rate": 1.3619166362452055e-06,
      "loss": 0.4459,
      "step": 320380
    },
    {
      "epoch": 131.15,
      "grad_norm": 1.7701619863510132,
      "learning_rate": 1.3617725988675758e-06,
      "loss": 0.4575,
      "step": 320390
    },
    {
      "epoch": 131.15,
      "grad_norm": 2.0084826946258545,
      "learning_rate": 1.3616285659882325e-06,
      "loss": 0.4749,
      "step": 320400
    },
    {
      "epoch": 131.15,
      "grad_norm": 1.8612128496170044,
      "learning_rate": 1.3614845376078354e-06,
      "loss": 0.477,
      "step": 320410
    },
    {
      "epoch": 131.16,
      "grad_norm": 2.0337841510772705,
      "learning_rate": 1.3613405137270447e-06,
      "loss": 0.4426,
      "step": 320420
    },
    {
      "epoch": 131.16,
      "grad_norm": 2.0625243186950684,
      "learning_rate": 1.361196494346519e-06,
      "loss": 0.4573,
      "step": 320430
    },
    {
      "epoch": 131.17,
      "grad_norm": 2.110848903656006,
      "learning_rate": 1.3610524794669202e-06,
      "loss": 0.4606,
      "step": 320440
    },
    {
      "epoch": 131.17,
      "grad_norm": 1.8857572078704834,
      "learning_rate": 1.3609084690889055e-06,
      "loss": 0.4521,
      "step": 320450
    },
    {
      "epoch": 131.17,
      "grad_norm": 1.751198649406433,
      "learning_rate": 1.3607644632131346e-06,
      "loss": 0.4534,
      "step": 320460
    },
    {
      "epoch": 131.18,
      "grad_norm": 1.7876474857330322,
      "learning_rate": 1.3606204618402691e-06,
      "loss": 0.4431,
      "step": 320470
    },
    {
      "epoch": 131.18,
      "grad_norm": 1.9728230237960815,
      "learning_rate": 1.3604764649709677e-06,
      "loss": 0.4629,
      "step": 320480
    },
    {
      "epoch": 131.19,
      "grad_norm": 2.147658348083496,
      "learning_rate": 1.3603324726058905e-06,
      "loss": 0.4471,
      "step": 320490
    },
    {
      "epoch": 131.19,
      "grad_norm": 2.0648715496063232,
      "learning_rate": 1.3601884847456963e-06,
      "loss": 0.4604,
      "step": 320500
    },
    {
      "epoch": 131.2,
      "grad_norm": 2.3883378505706787,
      "learning_rate": 1.3600445013910451e-06,
      "loss": 0.448,
      "step": 320510
    },
    {
      "epoch": 131.2,
      "grad_norm": 2.1579644680023193,
      "learning_rate": 1.3599005225425967e-06,
      "loss": 0.4398,
      "step": 320520
    },
    {
      "epoch": 131.2,
      "grad_norm": 1.6350728273391724,
      "learning_rate": 1.3597565482010103e-06,
      "loss": 0.4352,
      "step": 320530
    },
    {
      "epoch": 131.21,
      "grad_norm": 2.1441071033477783,
      "learning_rate": 1.3596125783669452e-06,
      "loss": 0.4719,
      "step": 320540
    },
    {
      "epoch": 131.21,
      "grad_norm": 1.946716070175171,
      "learning_rate": 1.3594686130410617e-06,
      "loss": 0.4548,
      "step": 320550
    },
    {
      "epoch": 131.22,
      "grad_norm": 2.122528314590454,
      "learning_rate": 1.3593246522240183e-06,
      "loss": 0.448,
      "step": 320560
    },
    {
      "epoch": 131.22,
      "grad_norm": 1.925788164138794,
      "learning_rate": 1.3591806959164755e-06,
      "loss": 0.4653,
      "step": 320570
    },
    {
      "epoch": 131.22,
      "grad_norm": 1.9768640995025635,
      "learning_rate": 1.359036744119093e-06,
      "loss": 0.4675,
      "step": 320580
    },
    {
      "epoch": 131.23,
      "grad_norm": 2.1738572120666504,
      "learning_rate": 1.358892796832528e-06,
      "loss": 0.4654,
      "step": 320590
    },
    {
      "epoch": 131.23,
      "grad_norm": 1.9374961853027344,
      "learning_rate": 1.3587488540574417e-06,
      "loss": 0.4816,
      "step": 320600
    },
    {
      "epoch": 131.24,
      "grad_norm": 1.923326015472412,
      "learning_rate": 1.3586049157944933e-06,
      "loss": 0.4656,
      "step": 320610
    },
    {
      "epoch": 131.24,
      "grad_norm": 1.7137656211853027,
      "learning_rate": 1.3584609820443415e-06,
      "loss": 0.4312,
      "step": 320620
    },
    {
      "epoch": 131.24,
      "grad_norm": 2.2453548908233643,
      "learning_rate": 1.3583170528076463e-06,
      "loss": 0.4966,
      "step": 320630
    },
    {
      "epoch": 131.25,
      "grad_norm": 2.2156143188476562,
      "learning_rate": 1.358173128085067e-06,
      "loss": 0.4491,
      "step": 320640
    },
    {
      "epoch": 131.25,
      "grad_norm": 2.2083749771118164,
      "learning_rate": 1.3580292078772616e-06,
      "loss": 0.4434,
      "step": 320650
    },
    {
      "epoch": 131.26,
      "grad_norm": 2.4297003746032715,
      "learning_rate": 1.3578852921848914e-06,
      "loss": 0.452,
      "step": 320660
    },
    {
      "epoch": 131.26,
      "grad_norm": 1.9673312902450562,
      "learning_rate": 1.357741381008615e-06,
      "loss": 0.4651,
      "step": 320670
    },
    {
      "epoch": 131.26,
      "grad_norm": 1.776674747467041,
      "learning_rate": 1.357597474349091e-06,
      "loss": 0.4396,
      "step": 320680
    },
    {
      "epoch": 131.27,
      "grad_norm": 2.4956443309783936,
      "learning_rate": 1.3574535722069791e-06,
      "loss": 0.4419,
      "step": 320690
    },
    {
      "epoch": 131.27,
      "grad_norm": 2.4395880699157715,
      "learning_rate": 1.3573096745829385e-06,
      "loss": 0.4643,
      "step": 320700
    },
    {
      "epoch": 131.28,
      "grad_norm": 1.6022007465362549,
      "learning_rate": 1.3571657814776291e-06,
      "loss": 0.4475,
      "step": 320710
    },
    {
      "epoch": 131.28,
      "grad_norm": 1.9109052419662476,
      "learning_rate": 1.3570218928917085e-06,
      "loss": 0.4592,
      "step": 320720
    },
    {
      "epoch": 131.29,
      "grad_norm": 1.8303266763687134,
      "learning_rate": 1.3568780088258363e-06,
      "loss": 0.4478,
      "step": 320730
    },
    {
      "epoch": 131.29,
      "grad_norm": 2.1075515747070312,
      "learning_rate": 1.3567341292806719e-06,
      "loss": 0.4677,
      "step": 320740
    },
    {
      "epoch": 131.29,
      "grad_norm": 2.0548481941223145,
      "learning_rate": 1.3565902542568743e-06,
      "loss": 0.4469,
      "step": 320750
    },
    {
      "epoch": 131.3,
      "grad_norm": 1.7737594842910767,
      "learning_rate": 1.356446383755103e-06,
      "loss": 0.447,
      "step": 320760
    },
    {
      "epoch": 131.3,
      "grad_norm": 1.5279680490493774,
      "learning_rate": 1.3563025177760164e-06,
      "loss": 0.4527,
      "step": 320770
    },
    {
      "epoch": 131.31,
      "grad_norm": 1.7691048383712769,
      "learning_rate": 1.3561586563202735e-06,
      "loss": 0.4514,
      "step": 320780
    },
    {
      "epoch": 131.31,
      "grad_norm": 2.323201894760132,
      "learning_rate": 1.356014799388534e-06,
      "loss": 0.4417,
      "step": 320790
    },
    {
      "epoch": 131.31,
      "grad_norm": 1.9410518407821655,
      "learning_rate": 1.3558709469814563e-06,
      "loss": 0.4773,
      "step": 320800
    },
    {
      "epoch": 131.32,
      "grad_norm": 2.2036783695220947,
      "learning_rate": 1.3557270990996997e-06,
      "loss": 0.4614,
      "step": 320810
    },
    {
      "epoch": 131.32,
      "grad_norm": 1.7645094394683838,
      "learning_rate": 1.3555832557439225e-06,
      "loss": 0.4567,
      "step": 320820
    },
    {
      "epoch": 131.33,
      "grad_norm": 2.622471809387207,
      "learning_rate": 1.3554394169147848e-06,
      "loss": 0.4494,
      "step": 320830
    },
    {
      "epoch": 131.33,
      "grad_norm": 2.342574119567871,
      "learning_rate": 1.3552955826129436e-06,
      "loss": 0.4514,
      "step": 320840
    },
    {
      "epoch": 131.33,
      "grad_norm": 1.8716212511062622,
      "learning_rate": 1.3551517528390607e-06,
      "loss": 0.4647,
      "step": 320850
    },
    {
      "epoch": 131.34,
      "grad_norm": 2.189391613006592,
      "learning_rate": 1.3550079275937925e-06,
      "loss": 0.4586,
      "step": 320860
    },
    {
      "epoch": 131.34,
      "grad_norm": 2.2878806591033936,
      "learning_rate": 1.3548641068777986e-06,
      "loss": 0.4759,
      "step": 320870
    },
    {
      "epoch": 131.35,
      "grad_norm": 2.098140239715576,
      "learning_rate": 1.354720290691738e-06,
      "loss": 0.4572,
      "step": 320880
    },
    {
      "epoch": 131.35,
      "grad_norm": 2.0966861248016357,
      "learning_rate": 1.354576479036269e-06,
      "loss": 0.4491,
      "step": 320890
    },
    {
      "epoch": 131.35,
      "grad_norm": 1.5244005918502808,
      "learning_rate": 1.3544326719120507e-06,
      "loss": 0.4639,
      "step": 320900
    },
    {
      "epoch": 131.36,
      "grad_norm": 1.9602265357971191,
      "learning_rate": 1.3542888693197423e-06,
      "loss": 0.4596,
      "step": 320910
    },
    {
      "epoch": 131.36,
      "grad_norm": 1.7376184463500977,
      "learning_rate": 1.3541450712600018e-06,
      "loss": 0.4553,
      "step": 320920
    },
    {
      "epoch": 131.37,
      "grad_norm": 2.1910531520843506,
      "learning_rate": 1.3540012777334883e-06,
      "loss": 0.4915,
      "step": 320930
    },
    {
      "epoch": 131.37,
      "grad_norm": 1.6704062223434448,
      "learning_rate": 1.3538574887408604e-06,
      "loss": 0.4526,
      "step": 320940
    },
    {
      "epoch": 131.38,
      "grad_norm": 2.1519553661346436,
      "learning_rate": 1.3537137042827767e-06,
      "loss": 0.4427,
      "step": 320950
    },
    {
      "epoch": 131.38,
      "grad_norm": 1.7545260190963745,
      "learning_rate": 1.353569924359896e-06,
      "loss": 0.4492,
      "step": 320960
    },
    {
      "epoch": 131.38,
      "grad_norm": 1.6043699979782104,
      "learning_rate": 1.3534261489728767e-06,
      "loss": 0.4708,
      "step": 320970
    },
    {
      "epoch": 131.39,
      "grad_norm": 1.805348515510559,
      "learning_rate": 1.353282378122378e-06,
      "loss": 0.4668,
      "step": 320980
    },
    {
      "epoch": 131.39,
      "grad_norm": 1.9836214780807495,
      "learning_rate": 1.3531386118090577e-06,
      "loss": 0.4545,
      "step": 320990
    },
    {
      "epoch": 131.4,
      "grad_norm": 1.9680155515670776,
      "learning_rate": 1.3529948500335747e-06,
      "loss": 0.4748,
      "step": 321000
    },
    {
      "epoch": 131.4,
      "grad_norm": 1.8128383159637451,
      "learning_rate": 1.352851092796588e-06,
      "loss": 0.4714,
      "step": 321010
    },
    {
      "epoch": 131.4,
      "grad_norm": 2.173089027404785,
      "learning_rate": 1.3527073400987555e-06,
      "loss": 0.4709,
      "step": 321020
    },
    {
      "epoch": 131.41,
      "grad_norm": 1.8858754634857178,
      "learning_rate": 1.352563591940736e-06,
      "loss": 0.4634,
      "step": 321030
    },
    {
      "epoch": 131.41,
      "grad_norm": 1.9414105415344238,
      "learning_rate": 1.352419848323188e-06,
      "loss": 0.4578,
      "step": 321040
    },
    {
      "epoch": 131.42,
      "grad_norm": 2.0856645107269287,
      "learning_rate": 1.3522761092467697e-06,
      "loss": 0.4581,
      "step": 321050
    },
    {
      "epoch": 131.42,
      "grad_norm": 2.1425867080688477,
      "learning_rate": 1.3521323747121398e-06,
      "loss": 0.4436,
      "step": 321060
    },
    {
      "epoch": 131.42,
      "grad_norm": 2.1687982082366943,
      "learning_rate": 1.3519886447199569e-06,
      "loss": 0.4523,
      "step": 321070
    },
    {
      "epoch": 131.43,
      "grad_norm": 1.8226913213729858,
      "learning_rate": 1.351844919270879e-06,
      "loss": 0.4187,
      "step": 321080
    },
    {
      "epoch": 131.43,
      "grad_norm": 2.296705484390259,
      "learning_rate": 1.3517011983655644e-06,
      "loss": 0.4591,
      "step": 321090
    },
    {
      "epoch": 131.44,
      "grad_norm": 2.027545690536499,
      "learning_rate": 1.3515574820046723e-06,
      "loss": 0.4485,
      "step": 321100
    },
    {
      "epoch": 131.44,
      "grad_norm": 1.8350661993026733,
      "learning_rate": 1.35141377018886e-06,
      "loss": 0.4617,
      "step": 321110
    },
    {
      "epoch": 131.44,
      "grad_norm": 1.8841395378112793,
      "learning_rate": 1.3512700629187867e-06,
      "loss": 0.4492,
      "step": 321120
    },
    {
      "epoch": 131.45,
      "grad_norm": 1.8282350301742554,
      "learning_rate": 1.3511263601951098e-06,
      "loss": 0.4752,
      "step": 321130
    },
    {
      "epoch": 131.45,
      "grad_norm": 1.9742289781570435,
      "learning_rate": 1.3509826620184885e-06,
      "loss": 0.469,
      "step": 321140
    },
    {
      "epoch": 131.46,
      "grad_norm": 1.555196762084961,
      "learning_rate": 1.3508389683895803e-06,
      "loss": 0.4549,
      "step": 321150
    },
    {
      "epoch": 131.46,
      "grad_norm": 1.8479154109954834,
      "learning_rate": 1.3506952793090448e-06,
      "loss": 0.4729,
      "step": 321160
    },
    {
      "epoch": 131.47,
      "grad_norm": 2.1844613552093506,
      "learning_rate": 1.3505515947775381e-06,
      "loss": 0.4747,
      "step": 321170
    },
    {
      "epoch": 131.47,
      "grad_norm": 1.9311639070510864,
      "learning_rate": 1.3504079147957196e-06,
      "loss": 0.4574,
      "step": 321180
    },
    {
      "epoch": 131.47,
      "grad_norm": 2.0557234287261963,
      "learning_rate": 1.3502642393642475e-06,
      "loss": 0.4534,
      "step": 321190
    },
    {
      "epoch": 131.48,
      "grad_norm": 1.9105345010757446,
      "learning_rate": 1.3501205684837798e-06,
      "loss": 0.4404,
      "step": 321200
    },
    {
      "epoch": 131.48,
      "grad_norm": 1.8885691165924072,
      "learning_rate": 1.3499769021549732e-06,
      "loss": 0.454,
      "step": 321210
    },
    {
      "epoch": 131.49,
      "grad_norm": 2.0541086196899414,
      "learning_rate": 1.349833240378489e-06,
      "loss": 0.4623,
      "step": 321220
    },
    {
      "epoch": 131.49,
      "grad_norm": 2.049978733062744,
      "learning_rate": 1.3496895831549829e-06,
      "loss": 0.4468,
      "step": 321230
    },
    {
      "epoch": 131.49,
      "grad_norm": 2.0511279106140137,
      "learning_rate": 1.349545930485114e-06,
      "loss": 0.4702,
      "step": 321240
    },
    {
      "epoch": 131.5,
      "grad_norm": 2.524791717529297,
      "learning_rate": 1.3494022823695397e-06,
      "loss": 0.4738,
      "step": 321250
    },
    {
      "epoch": 131.5,
      "grad_norm": 2.08878755569458,
      "learning_rate": 1.3492586388089186e-06,
      "loss": 0.4395,
      "step": 321260
    },
    {
      "epoch": 131.51,
      "grad_norm": 1.5729434490203857,
      "learning_rate": 1.3491149998039082e-06,
      "loss": 0.4637,
      "step": 321270
    },
    {
      "epoch": 131.51,
      "grad_norm": 2.10571551322937,
      "learning_rate": 1.3489713653551668e-06,
      "loss": 0.4559,
      "step": 321280
    },
    {
      "epoch": 131.51,
      "grad_norm": 1.7558993101119995,
      "learning_rate": 1.3488277354633523e-06,
      "loss": 0.4587,
      "step": 321290
    },
    {
      "epoch": 131.52,
      "grad_norm": 2.262761116027832,
      "learning_rate": 1.3486841101291232e-06,
      "loss": 0.4469,
      "step": 321300
    },
    {
      "epoch": 131.52,
      "grad_norm": 2.352030038833618,
      "learning_rate": 1.3485404893531362e-06,
      "loss": 0.4493,
      "step": 321310
    },
    {
      "epoch": 131.53,
      "grad_norm": 2.299644708633423,
      "learning_rate": 1.3483968731360499e-06,
      "loss": 0.4744,
      "step": 321320
    },
    {
      "epoch": 131.53,
      "grad_norm": 1.985864281654358,
      "learning_rate": 1.3482532614785222e-06,
      "loss": 0.4509,
      "step": 321330
    },
    {
      "epoch": 131.53,
      "grad_norm": 2.525996446609497,
      "learning_rate": 1.3481096543812105e-06,
      "loss": 0.4535,
      "step": 321340
    },
    {
      "epoch": 131.54,
      "grad_norm": 1.962923526763916,
      "learning_rate": 1.3479660518447733e-06,
      "loss": 0.4493,
      "step": 321350
    },
    {
      "epoch": 131.54,
      "grad_norm": 2.337719202041626,
      "learning_rate": 1.3478224538698682e-06,
      "loss": 0.4517,
      "step": 321360
    },
    {
      "epoch": 131.55,
      "grad_norm": 2.0619359016418457,
      "learning_rate": 1.3476788604571528e-06,
      "loss": 0.4582,
      "step": 321370
    },
    {
      "epoch": 131.55,
      "grad_norm": 1.8146235942840576,
      "learning_rate": 1.3475352716072849e-06,
      "loss": 0.4514,
      "step": 321380
    },
    {
      "epoch": 131.56,
      "grad_norm": 2.143258810043335,
      "learning_rate": 1.3473916873209224e-06,
      "loss": 0.4683,
      "step": 321390
    },
    {
      "epoch": 131.56,
      "grad_norm": 2.2679567337036133,
      "learning_rate": 1.3472481075987224e-06,
      "loss": 0.4425,
      "step": 321400
    },
    {
      "epoch": 131.56,
      "grad_norm": 1.70207679271698,
      "learning_rate": 1.3471045324413443e-06,
      "loss": 0.4542,
      "step": 321410
    },
    {
      "epoch": 131.57,
      "grad_norm": 1.8528074026107788,
      "learning_rate": 1.3469609618494442e-06,
      "loss": 0.4668,
      "step": 321420
    },
    {
      "epoch": 131.57,
      "grad_norm": 1.6151257753372192,
      "learning_rate": 1.3468173958236814e-06,
      "loss": 0.466,
      "step": 321430
    },
    {
      "epoch": 131.58,
      "grad_norm": 2.089684247970581,
      "learning_rate": 1.3466738343647118e-06,
      "loss": 0.464,
      "step": 321440
    },
    {
      "epoch": 131.58,
      "grad_norm": 1.6658705472946167,
      "learning_rate": 1.3465302774731934e-06,
      "loss": 0.4526,
      "step": 321450
    },
    {
      "epoch": 131.58,
      "grad_norm": 2.3782882690429688,
      "learning_rate": 1.3463867251497842e-06,
      "loss": 0.4431,
      "step": 321460
    },
    {
      "epoch": 131.59,
      "grad_norm": 1.732114315032959,
      "learning_rate": 1.3462431773951419e-06,
      "loss": 0.4766,
      "step": 321470
    },
    {
      "epoch": 131.59,
      "grad_norm": 1.5144908428192139,
      "learning_rate": 1.3460996342099236e-06,
      "loss": 0.4606,
      "step": 321480
    },
    {
      "epoch": 131.6,
      "grad_norm": 1.9448144435882568,
      "learning_rate": 1.3459560955947874e-06,
      "loss": 0.4691,
      "step": 321490
    },
    {
      "epoch": 131.6,
      "grad_norm": 1.8925387859344482,
      "learning_rate": 1.3458125615503902e-06,
      "loss": 0.4657,
      "step": 321500
    },
    {
      "epoch": 131.6,
      "grad_norm": 2.033620834350586,
      "learning_rate": 1.34566903207739e-06,
      "loss": 0.4515,
      "step": 321510
    },
    {
      "epoch": 131.61,
      "grad_norm": 1.6082227230072021,
      "learning_rate": 1.3455255071764441e-06,
      "loss": 0.4382,
      "step": 321520
    },
    {
      "epoch": 131.61,
      "grad_norm": 1.6820504665374756,
      "learning_rate": 1.3453819868482103e-06,
      "loss": 0.4498,
      "step": 321530
    },
    {
      "epoch": 131.62,
      "grad_norm": 2.171020030975342,
      "learning_rate": 1.3452384710933455e-06,
      "loss": 0.4577,
      "step": 321540
    },
    {
      "epoch": 131.62,
      "grad_norm": 1.962045669555664,
      "learning_rate": 1.3450949599125074e-06,
      "loss": 0.4416,
      "step": 321550
    },
    {
      "epoch": 131.63,
      "grad_norm": 1.9675699472427368,
      "learning_rate": 1.3449514533063534e-06,
      "loss": 0.4558,
      "step": 321560
    },
    {
      "epoch": 131.63,
      "grad_norm": 1.8161872625350952,
      "learning_rate": 1.3448079512755408e-06,
      "loss": 0.4704,
      "step": 321570
    },
    {
      "epoch": 131.63,
      "grad_norm": 1.885606288909912,
      "learning_rate": 1.3446644538207274e-06,
      "loss": 0.4433,
      "step": 321580
    },
    {
      "epoch": 131.64,
      "grad_norm": 2.1240715980529785,
      "learning_rate": 1.3445209609425696e-06,
      "loss": 0.4678,
      "step": 321590
    },
    {
      "epoch": 131.64,
      "grad_norm": 2.118802785873413,
      "learning_rate": 1.3443774726417253e-06,
      "loss": 0.4544,
      "step": 321600
    },
    {
      "epoch": 131.65,
      "grad_norm": 1.980224847793579,
      "learning_rate": 1.3442339889188523e-06,
      "loss": 0.4668,
      "step": 321610
    },
    {
      "epoch": 131.65,
      "grad_norm": 1.8698785305023193,
      "learning_rate": 1.344090509774607e-06,
      "loss": 0.4405,
      "step": 321620
    },
    {
      "epoch": 131.65,
      "grad_norm": 2.0253615379333496,
      "learning_rate": 1.343947035209647e-06,
      "loss": 0.4738,
      "step": 321630
    },
    {
      "epoch": 131.66,
      "grad_norm": 2.167541980743408,
      "learning_rate": 1.3438035652246294e-06,
      "loss": 0.4655,
      "step": 321640
    },
    {
      "epoch": 131.66,
      "grad_norm": 1.6391973495483398,
      "learning_rate": 1.3436600998202118e-06,
      "loss": 0.4792,
      "step": 321650
    },
    {
      "epoch": 131.67,
      "grad_norm": 1.9450992345809937,
      "learning_rate": 1.343516638997051e-06,
      "loss": 0.4651,
      "step": 321660
    },
    {
      "epoch": 131.67,
      "grad_norm": 1.9138047695159912,
      "learning_rate": 1.3433731827558046e-06,
      "loss": 0.4412,
      "step": 321670
    },
    {
      "epoch": 131.67,
      "grad_norm": 1.822512149810791,
      "learning_rate": 1.3432297310971293e-06,
      "loss": 0.4461,
      "step": 321680
    },
    {
      "epoch": 131.68,
      "grad_norm": 2.0983810424804688,
      "learning_rate": 1.3430862840216823e-06,
      "loss": 0.4448,
      "step": 321690
    },
    {
      "epoch": 131.68,
      "grad_norm": 2.1381940841674805,
      "learning_rate": 1.3429428415301208e-06,
      "loss": 0.4498,
      "step": 321700
    },
    {
      "epoch": 131.69,
      "grad_norm": 2.0788114070892334,
      "learning_rate": 1.3427994036231024e-06,
      "loss": 0.4369,
      "step": 321710
    },
    {
      "epoch": 131.69,
      "grad_norm": 2.1540210247039795,
      "learning_rate": 1.342655970301283e-06,
      "loss": 0.4438,
      "step": 321720
    },
    {
      "epoch": 131.69,
      "grad_norm": 2.439009189605713,
      "learning_rate": 1.3425125415653205e-06,
      "loss": 0.4278,
      "step": 321730
    },
    {
      "epoch": 131.7,
      "grad_norm": 2.5387160778045654,
      "learning_rate": 1.342369117415873e-06,
      "loss": 0.4594,
      "step": 321740
    },
    {
      "epoch": 131.7,
      "grad_norm": 1.8298591375350952,
      "learning_rate": 1.3422256978535949e-06,
      "loss": 0.4774,
      "step": 321750
    },
    {
      "epoch": 131.71,
      "grad_norm": 2.587285280227661,
      "learning_rate": 1.3420822828791445e-06,
      "loss": 0.4597,
      "step": 321760
    },
    {
      "epoch": 131.71,
      "grad_norm": 2.4666194915771484,
      "learning_rate": 1.341938872493179e-06,
      "loss": 0.4554,
      "step": 321770
    },
    {
      "epoch": 131.72,
      "grad_norm": 1.5672669410705566,
      "learning_rate": 1.3417954666963543e-06,
      "loss": 0.4656,
      "step": 321780
    },
    {
      "epoch": 131.72,
      "grad_norm": 1.959385871887207,
      "learning_rate": 1.3416520654893287e-06,
      "loss": 0.4744,
      "step": 321790
    },
    {
      "epoch": 131.72,
      "grad_norm": 1.9169588088989258,
      "learning_rate": 1.3415086688727587e-06,
      "loss": 0.4526,
      "step": 321800
    },
    {
      "epoch": 131.73,
      "grad_norm": 2.280416250228882,
      "learning_rate": 1.341365276847301e-06,
      "loss": 0.4529,
      "step": 321810
    },
    {
      "epoch": 131.73,
      "grad_norm": 2.323843240737915,
      "learning_rate": 1.3412218894136126e-06,
      "loss": 0.4702,
      "step": 321820
    },
    {
      "epoch": 131.74,
      "grad_norm": 1.9819148778915405,
      "learning_rate": 1.3410785065723505e-06,
      "loss": 0.4314,
      "step": 321830
    },
    {
      "epoch": 131.74,
      "grad_norm": 1.9820164442062378,
      "learning_rate": 1.3409351283241708e-06,
      "loss": 0.4451,
      "step": 321840
    },
    {
      "epoch": 131.74,
      "grad_norm": 2.1258492469787598,
      "learning_rate": 1.340791754669731e-06,
      "loss": 0.447,
      "step": 321850
    },
    {
      "epoch": 131.75,
      "grad_norm": 1.88982093334198,
      "learning_rate": 1.3406483856096873e-06,
      "loss": 0.4773,
      "step": 321860
    },
    {
      "epoch": 131.75,
      "grad_norm": 2.0796663761138916,
      "learning_rate": 1.340505021144697e-06,
      "loss": 0.4596,
      "step": 321870
    },
    {
      "epoch": 131.76,
      "grad_norm": 1.8660557270050049,
      "learning_rate": 1.3403616612754173e-06,
      "loss": 0.4592,
      "step": 321880
    },
    {
      "epoch": 131.76,
      "grad_norm": 2.9289000034332275,
      "learning_rate": 1.3402183060025034e-06,
      "loss": 0.4542,
      "step": 321890
    },
    {
      "epoch": 131.76,
      "grad_norm": 6.6493659019470215,
      "learning_rate": 1.3400749553266125e-06,
      "loss": 0.4382,
      "step": 321900
    },
    {
      "epoch": 131.77,
      "grad_norm": 1.9050074815750122,
      "learning_rate": 1.3399316092484018e-06,
      "loss": 0.4546,
      "step": 321910
    },
    {
      "epoch": 131.77,
      "grad_norm": 2.0804123878479004,
      "learning_rate": 1.3397882677685274e-06,
      "loss": 0.4517,
      "step": 321920
    },
    {
      "epoch": 131.78,
      "grad_norm": 2.046703815460205,
      "learning_rate": 1.3396449308876469e-06,
      "loss": 0.4619,
      "step": 321930
    },
    {
      "epoch": 131.78,
      "grad_norm": 1.689092993736267,
      "learning_rate": 1.3395015986064154e-06,
      "loss": 0.4694,
      "step": 321940
    },
    {
      "epoch": 131.78,
      "grad_norm": 1.9556304216384888,
      "learning_rate": 1.3393582709254907e-06,
      "loss": 0.4696,
      "step": 321950
    },
    {
      "epoch": 131.79,
      "grad_norm": 2.157521963119507,
      "learning_rate": 1.339214947845529e-06,
      "loss": 0.4489,
      "step": 321960
    },
    {
      "epoch": 131.79,
      "grad_norm": 2.4337644577026367,
      "learning_rate": 1.339071629367186e-06,
      "loss": 0.4323,
      "step": 321970
    },
    {
      "epoch": 131.8,
      "grad_norm": 2.184177875518799,
      "learning_rate": 1.3389283154911194e-06,
      "loss": 0.4508,
      "step": 321980
    },
    {
      "epoch": 131.8,
      "grad_norm": 2.491339921951294,
      "learning_rate": 1.3387850062179859e-06,
      "loss": 0.4378,
      "step": 321990
    },
    {
      "epoch": 131.81,
      "grad_norm": 2.1591527462005615,
      "learning_rate": 1.338641701548441e-06,
      "loss": 0.4518,
      "step": 322000
    },
    {
      "epoch": 131.81,
      "grad_norm": 1.7660720348358154,
      "learning_rate": 1.3384984014831427e-06,
      "loss": 0.4555,
      "step": 322010
    },
    {
      "epoch": 131.81,
      "grad_norm": 2.088796377182007,
      "learning_rate": 1.3383551060227453e-06,
      "loss": 0.4499,
      "step": 322020
    },
    {
      "epoch": 131.82,
      "grad_norm": 1.973292350769043,
      "learning_rate": 1.3382118151679064e-06,
      "loss": 0.465,
      "step": 322030
    },
    {
      "epoch": 131.82,
      "grad_norm": 2.2369258403778076,
      "learning_rate": 1.338068528919282e-06,
      "loss": 0.4226,
      "step": 322040
    },
    {
      "epoch": 131.83,
      "grad_norm": 2.2586302757263184,
      "learning_rate": 1.3379252472775286e-06,
      "loss": 0.4439,
      "step": 322050
    },
    {
      "epoch": 131.83,
      "grad_norm": 1.7128208875656128,
      "learning_rate": 1.3377819702433025e-06,
      "loss": 0.4399,
      "step": 322060
    },
    {
      "epoch": 131.83,
      "grad_norm": 1.6947457790374756,
      "learning_rate": 1.3376386978172604e-06,
      "loss": 0.4389,
      "step": 322070
    },
    {
      "epoch": 131.84,
      "grad_norm": 2.198072671890259,
      "learning_rate": 1.3374954300000583e-06,
      "loss": 0.482,
      "step": 322080
    },
    {
      "epoch": 131.84,
      "grad_norm": 1.96156907081604,
      "learning_rate": 1.3373521667923525e-06,
      "loss": 0.4363,
      "step": 322090
    },
    {
      "epoch": 131.85,
      "grad_norm": 1.8768810033798218,
      "learning_rate": 1.3372089081947996e-06,
      "loss": 0.4463,
      "step": 322100
    },
    {
      "epoch": 131.85,
      "grad_norm": 1.758017659187317,
      "learning_rate": 1.3370656542080552e-06,
      "loss": 0.4724,
      "step": 322110
    },
    {
      "epoch": 131.85,
      "grad_norm": 1.8588016033172607,
      "learning_rate": 1.3369224048327761e-06,
      "loss": 0.4499,
      "step": 322120
    },
    {
      "epoch": 131.86,
      "grad_norm": 1.5329089164733887,
      "learning_rate": 1.3367791600696181e-06,
      "loss": 0.4442,
      "step": 322130
    },
    {
      "epoch": 131.86,
      "grad_norm": 1.6987699270248413,
      "learning_rate": 1.3366359199192376e-06,
      "loss": 0.4691,
      "step": 322140
    },
    {
      "epoch": 131.87,
      "grad_norm": 1.5986812114715576,
      "learning_rate": 1.3364926843822911e-06,
      "loss": 0.4534,
      "step": 322150
    },
    {
      "epoch": 131.87,
      "grad_norm": 1.947564721107483,
      "learning_rate": 1.336349453459434e-06,
      "loss": 0.4597,
      "step": 322160
    },
    {
      "epoch": 131.87,
      "grad_norm": 1.7736515998840332,
      "learning_rate": 1.3362062271513227e-06,
      "loss": 0.447,
      "step": 322170
    },
    {
      "epoch": 131.88,
      "grad_norm": 1.698217511177063,
      "learning_rate": 1.3360630054586137e-06,
      "loss": 0.4473,
      "step": 322180
    },
    {
      "epoch": 131.88,
      "grad_norm": 2.0487778186798096,
      "learning_rate": 1.3359197883819625e-06,
      "loss": 0.4484,
      "step": 322190
    },
    {
      "epoch": 131.89,
      "grad_norm": 1.5672904253005981,
      "learning_rate": 1.3357765759220255e-06,
      "loss": 0.4308,
      "step": 322200
    },
    {
      "epoch": 131.89,
      "grad_norm": 2.1420984268188477,
      "learning_rate": 1.3356333680794586e-06,
      "loss": 0.4764,
      "step": 322210
    },
    {
      "epoch": 131.9,
      "grad_norm": 1.9327418804168701,
      "learning_rate": 1.335490164854918e-06,
      "loss": 0.4536,
      "step": 322220
    },
    {
      "epoch": 131.9,
      "grad_norm": 1.8291517496109009,
      "learning_rate": 1.3353469662490595e-06,
      "loss": 0.4674,
      "step": 322230
    },
    {
      "epoch": 131.9,
      "grad_norm": 2.282162666320801,
      "learning_rate": 1.3352037722625392e-06,
      "loss": 0.4422,
      "step": 322240
    },
    {
      "epoch": 131.91,
      "grad_norm": 2.451502799987793,
      "learning_rate": 1.335060582896013e-06,
      "loss": 0.4652,
      "step": 322250
    },
    {
      "epoch": 131.91,
      "grad_norm": 1.6728378534317017,
      "learning_rate": 1.3349173981501363e-06,
      "loss": 0.4651,
      "step": 322260
    },
    {
      "epoch": 131.92,
      "grad_norm": 1.8451454639434814,
      "learning_rate": 1.3347742180255662e-06,
      "loss": 0.4563,
      "step": 322270
    },
    {
      "epoch": 131.92,
      "grad_norm": 2.2230563163757324,
      "learning_rate": 1.3346310425229576e-06,
      "loss": 0.4574,
      "step": 322280
    },
    {
      "epoch": 131.92,
      "grad_norm": 1.8138127326965332,
      "learning_rate": 1.3344878716429667e-06,
      "loss": 0.4587,
      "step": 322290
    },
    {
      "epoch": 131.93,
      "grad_norm": 2.0144617557525635,
      "learning_rate": 1.3343447053862493e-06,
      "loss": 0.48,
      "step": 322300
    },
    {
      "epoch": 131.93,
      "grad_norm": 2.0818300247192383,
      "learning_rate": 1.3342015437534613e-06,
      "loss": 0.4637,
      "step": 322310
    },
    {
      "epoch": 131.94,
      "grad_norm": 2.285841464996338,
      "learning_rate": 1.3340583867452593e-06,
      "loss": 0.4478,
      "step": 322320
    },
    {
      "epoch": 131.94,
      "grad_norm": 1.9454256296157837,
      "learning_rate": 1.333915234362297e-06,
      "loss": 0.4618,
      "step": 322330
    },
    {
      "epoch": 131.94,
      "grad_norm": 1.722348690032959,
      "learning_rate": 1.3337720866052317e-06,
      "loss": 0.4509,
      "step": 322340
    },
    {
      "epoch": 131.95,
      "grad_norm": 1.85185968875885,
      "learning_rate": 1.3336289434747176e-06,
      "loss": 0.4579,
      "step": 322350
    },
    {
      "epoch": 131.95,
      "grad_norm": 2.2323832511901855,
      "learning_rate": 1.3334858049714127e-06,
      "loss": 0.4581,
      "step": 322360
    },
    {
      "epoch": 131.96,
      "grad_norm": 1.9927538633346558,
      "learning_rate": 1.333342671095972e-06,
      "loss": 0.4458,
      "step": 322370
    },
    {
      "epoch": 131.96,
      "grad_norm": 2.302565813064575,
      "learning_rate": 1.3331995418490505e-06,
      "loss": 0.4567,
      "step": 322380
    },
    {
      "epoch": 131.96,
      "grad_norm": 1.9579285383224487,
      "learning_rate": 1.333056417231304e-06,
      "loss": 0.4727,
      "step": 322390
    },
    {
      "epoch": 131.97,
      "grad_norm": 2.503154993057251,
      "learning_rate": 1.3329132972433882e-06,
      "loss": 0.4642,
      "step": 322400
    },
    {
      "epoch": 131.97,
      "grad_norm": 1.9725136756896973,
      "learning_rate": 1.332770181885959e-06,
      "loss": 0.4352,
      "step": 322410
    },
    {
      "epoch": 131.98,
      "grad_norm": 1.870768666267395,
      "learning_rate": 1.3326270711596719e-06,
      "loss": 0.4459,
      "step": 322420
    },
    {
      "epoch": 131.98,
      "grad_norm": 1.7751178741455078,
      "learning_rate": 1.3324839650651818e-06,
      "loss": 0.4474,
      "step": 322430
    },
    {
      "epoch": 131.99,
      "grad_norm": 2.025879144668579,
      "learning_rate": 1.3323408636031455e-06,
      "loss": 0.4594,
      "step": 322440
    },
    {
      "epoch": 131.99,
      "grad_norm": 2.4114575386047363,
      "learning_rate": 1.3321977667742174e-06,
      "loss": 0.4446,
      "step": 322450
    },
    {
      "epoch": 131.99,
      "grad_norm": 1.787190556526184,
      "learning_rate": 1.3320546745790546e-06,
      "loss": 0.4679,
      "step": 322460
    },
    {
      "epoch": 132.0,
      "grad_norm": 2.1378655433654785,
      "learning_rate": 1.3319115870183102e-06,
      "loss": 0.4529,
      "step": 322470
    },
    {
      "epoch": 132.0,
      "eval_loss": 0.45427632331848145,
      "eval_runtime": 52.9242,
      "eval_samples_per_second": 65.169,
      "eval_steps_per_second": 8.163,
      "step": 322476
    },
    {
      "epoch": 132.0,
      "grad_norm": 1.7409993410110474,
      "learning_rate": 1.331768504092641e-06,
      "loss": 0.4567,
      "step": 322480
    },
    {
      "epoch": 132.01,
      "grad_norm": 1.8172224760055542,
      "learning_rate": 1.3316254258027023e-06,
      "loss": 0.4527,
      "step": 322490
    },
    {
      "epoch": 132.01,
      "grad_norm": 2.3063855171203613,
      "learning_rate": 1.3314823521491496e-06,
      "loss": 0.4563,
      "step": 322500
    },
    {
      "epoch": 132.01,
      "grad_norm": 2.091731548309326,
      "learning_rate": 1.3313392831326381e-06,
      "loss": 0.4618,
      "step": 322510
    },
    {
      "epoch": 132.02,
      "grad_norm": 1.7486811876296997,
      "learning_rate": 1.3311962187538232e-06,
      "loss": 0.4496,
      "step": 322520
    },
    {
      "epoch": 132.02,
      "grad_norm": 1.943135380744934,
      "learning_rate": 1.3310531590133597e-06,
      "loss": 0.4574,
      "step": 322530
    },
    {
      "epoch": 132.03,
      "grad_norm": 1.7884725332260132,
      "learning_rate": 1.3309101039119046e-06,
      "loss": 0.4243,
      "step": 322540
    },
    {
      "epoch": 132.03,
      "grad_norm": 2.0460455417633057,
      "learning_rate": 1.3307670534501119e-06,
      "loss": 0.4748,
      "step": 322550
    },
    {
      "epoch": 132.03,
      "grad_norm": 2.659698486328125,
      "learning_rate": 1.3306240076286373e-06,
      "loss": 0.4442,
      "step": 322560
    },
    {
      "epoch": 132.04,
      "grad_norm": 1.9639372825622559,
      "learning_rate": 1.3304809664481361e-06,
      "loss": 0.462,
      "step": 322570
    },
    {
      "epoch": 132.04,
      "grad_norm": 1.563143253326416,
      "learning_rate": 1.330337929909263e-06,
      "loss": 0.4435,
      "step": 322580
    },
    {
      "epoch": 132.05,
      "grad_norm": 4.843698978424072,
      "learning_rate": 1.3301948980126743e-06,
      "loss": 0.4783,
      "step": 322590
    },
    {
      "epoch": 132.05,
      "grad_norm": 1.835569977760315,
      "learning_rate": 1.3300518707590248e-06,
      "loss": 0.4627,
      "step": 322600
    },
    {
      "epoch": 132.05,
      "grad_norm": 2.159259080886841,
      "learning_rate": 1.329908848148969e-06,
      "loss": 0.4311,
      "step": 322610
    },
    {
      "epoch": 132.06,
      "grad_norm": 2.7261598110198975,
      "learning_rate": 1.3297658301831623e-06,
      "loss": 0.4443,
      "step": 322620
    },
    {
      "epoch": 132.06,
      "grad_norm": 1.8530852794647217,
      "learning_rate": 1.32962281686226e-06,
      "loss": 0.4635,
      "step": 322630
    },
    {
      "epoch": 132.07,
      "grad_norm": 2.087573528289795,
      "learning_rate": 1.3294798081869175e-06,
      "loss": 0.4434,
      "step": 322640
    },
    {
      "epoch": 132.07,
      "grad_norm": 2.232945203781128,
      "learning_rate": 1.3293368041577893e-06,
      "loss": 0.4352,
      "step": 322650
    },
    {
      "epoch": 132.08,
      "grad_norm": 2.0610265731811523,
      "learning_rate": 1.3291938047755312e-06,
      "loss": 0.442,
      "step": 322660
    },
    {
      "epoch": 132.08,
      "grad_norm": 1.772530436515808,
      "learning_rate": 1.3290508100407977e-06,
      "loss": 0.4437,
      "step": 322670
    },
    {
      "epoch": 132.08,
      "grad_norm": 1.670032262802124,
      "learning_rate": 1.3289078199542442e-06,
      "loss": 0.4436,
      "step": 322680
    },
    {
      "epoch": 132.09,
      "grad_norm": 1.934924840927124,
      "learning_rate": 1.3287648345165254e-06,
      "loss": 0.4432,
      "step": 322690
    },
    {
      "epoch": 132.09,
      "grad_norm": 1.694975733757019,
      "learning_rate": 1.3286218537282963e-06,
      "loss": 0.4666,
      "step": 322700
    },
    {
      "epoch": 132.1,
      "grad_norm": 3.9841790199279785,
      "learning_rate": 1.3284788775902126e-06,
      "loss": 0.4654,
      "step": 322710
    },
    {
      "epoch": 132.1,
      "grad_norm": 2.382681131362915,
      "learning_rate": 1.3283359061029273e-06,
      "loss": 0.4405,
      "step": 322720
    },
    {
      "epoch": 132.1,
      "grad_norm": 2.3944342136383057,
      "learning_rate": 1.3281929392670987e-06,
      "loss": 0.4427,
      "step": 322730
    },
    {
      "epoch": 132.11,
      "grad_norm": 2.1788079738616943,
      "learning_rate": 1.328049977083379e-06,
      "loss": 0.4718,
      "step": 322740
    },
    {
      "epoch": 132.11,
      "grad_norm": 2.1484756469726562,
      "learning_rate": 1.3279070195524234e-06,
      "loss": 0.45,
      "step": 322750
    },
    {
      "epoch": 132.12,
      "grad_norm": 3.322725296020508,
      "learning_rate": 1.3277640666748874e-06,
      "loss": 0.4511,
      "step": 322760
    },
    {
      "epoch": 132.12,
      "grad_norm": 1.974674940109253,
      "learning_rate": 1.3276211184514255e-06,
      "loss": 0.4494,
      "step": 322770
    },
    {
      "epoch": 132.12,
      "grad_norm": 1.8647189140319824,
      "learning_rate": 1.3274781748826928e-06,
      "loss": 0.46,
      "step": 322780
    },
    {
      "epoch": 132.13,
      "grad_norm": 1.820827841758728,
      "learning_rate": 1.3273352359693437e-06,
      "loss": 0.4469,
      "step": 322790
    },
    {
      "epoch": 132.13,
      "grad_norm": 2.329594850540161,
      "learning_rate": 1.3271923017120332e-06,
      "loss": 0.4376,
      "step": 322800
    },
    {
      "epoch": 132.14,
      "grad_norm": 1.8478320837020874,
      "learning_rate": 1.327049372111416e-06,
      "loss": 0.4498,
      "step": 322810
    },
    {
      "epoch": 132.14,
      "grad_norm": 2.0033788681030273,
      "learning_rate": 1.3269064471681476e-06,
      "loss": 0.4553,
      "step": 322820
    },
    {
      "epoch": 132.14,
      "grad_norm": 1.6980141401290894,
      "learning_rate": 1.3267635268828815e-06,
      "loss": 0.4492,
      "step": 322830
    },
    {
      "epoch": 132.15,
      "grad_norm": 2.1304688453674316,
      "learning_rate": 1.326620611256273e-06,
      "loss": 0.4632,
      "step": 322840
    },
    {
      "epoch": 132.15,
      "grad_norm": 2.3024017810821533,
      "learning_rate": 1.3264777002889766e-06,
      "loss": 0.4633,
      "step": 322850
    },
    {
      "epoch": 132.16,
      "grad_norm": 2.2772693634033203,
      "learning_rate": 1.3263347939816472e-06,
      "loss": 0.4783,
      "step": 322860
    },
    {
      "epoch": 132.16,
      "grad_norm": 2.0075297355651855,
      "learning_rate": 1.3261918923349394e-06,
      "loss": 0.4645,
      "step": 322870
    },
    {
      "epoch": 132.17,
      "grad_norm": 2.4679388999938965,
      "learning_rate": 1.3260489953495077e-06,
      "loss": 0.4688,
      "step": 322880
    },
    {
      "epoch": 132.17,
      "grad_norm": 2.59780216217041,
      "learning_rate": 1.3259061030260065e-06,
      "loss": 0.4646,
      "step": 322890
    },
    {
      "epoch": 132.17,
      "grad_norm": 1.9682351350784302,
      "learning_rate": 1.3257632153650906e-06,
      "loss": 0.488,
      "step": 322900
    },
    {
      "epoch": 132.18,
      "grad_norm": 2.1663336753845215,
      "learning_rate": 1.3256203323674144e-06,
      "loss": 0.4675,
      "step": 322910
    },
    {
      "epoch": 132.18,
      "grad_norm": 1.8215863704681396,
      "learning_rate": 1.3254774540336327e-06,
      "loss": 0.4857,
      "step": 322920
    },
    {
      "epoch": 132.19,
      "grad_norm": 1.8992822170257568,
      "learning_rate": 1.3253345803644e-06,
      "loss": 0.4661,
      "step": 322930
    },
    {
      "epoch": 132.19,
      "grad_norm": 1.7513130903244019,
      "learning_rate": 1.32519171136037e-06,
      "loss": 0.4429,
      "step": 322940
    },
    {
      "epoch": 132.19,
      "grad_norm": 2.2197606563568115,
      "learning_rate": 1.3250488470221987e-06,
      "loss": 0.4571,
      "step": 322950
    },
    {
      "epoch": 132.2,
      "grad_norm": 1.6703001260757446,
      "learning_rate": 1.324905987350539e-06,
      "loss": 0.4594,
      "step": 322960
    },
    {
      "epoch": 132.2,
      "grad_norm": 2.1973586082458496,
      "learning_rate": 1.324763132346046e-06,
      "loss": 0.4676,
      "step": 322970
    },
    {
      "epoch": 132.21,
      "grad_norm": 1.6556223630905151,
      "learning_rate": 1.324620282009374e-06,
      "loss": 0.4769,
      "step": 322980
    },
    {
      "epoch": 132.21,
      "grad_norm": 2.16349196434021,
      "learning_rate": 1.3244774363411773e-06,
      "loss": 0.4625,
      "step": 322990
    },
    {
      "epoch": 132.21,
      "grad_norm": 1.7052230834960938,
      "learning_rate": 1.324334595342111e-06,
      "loss": 0.4555,
      "step": 323000
    },
    {
      "epoch": 132.22,
      "grad_norm": 1.8865035772323608,
      "learning_rate": 1.3241917590128281e-06,
      "loss": 0.4568,
      "step": 323010
    },
    {
      "epoch": 132.22,
      "grad_norm": 2.353827714920044,
      "learning_rate": 1.3240489273539838e-06,
      "loss": 0.4496,
      "step": 323020
    },
    {
      "epoch": 132.23,
      "grad_norm": 1.9926621913909912,
      "learning_rate": 1.3239061003662325e-06,
      "loss": 0.4642,
      "step": 323030
    },
    {
      "epoch": 132.23,
      "grad_norm": 2.0073347091674805,
      "learning_rate": 1.3237632780502288e-06,
      "loss": 0.4651,
      "step": 323040
    },
    {
      "epoch": 132.23,
      "grad_norm": 2.040163516998291,
      "learning_rate": 1.3236204604066254e-06,
      "loss": 0.4471,
      "step": 323050
    },
    {
      "epoch": 132.24,
      "grad_norm": 2.0913543701171875,
      "learning_rate": 1.3234776474360774e-06,
      "loss": 0.4589,
      "step": 323060
    },
    {
      "epoch": 132.24,
      "grad_norm": 2.1301629543304443,
      "learning_rate": 1.323334839139239e-06,
      "loss": 0.4511,
      "step": 323070
    },
    {
      "epoch": 132.25,
      "grad_norm": 2.1992077827453613,
      "learning_rate": 1.3231920355167645e-06,
      "loss": 0.4762,
      "step": 323080
    },
    {
      "epoch": 132.25,
      "grad_norm": 1.930487871170044,
      "learning_rate": 1.323049236569308e-06,
      "loss": 0.4512,
      "step": 323090
    },
    {
      "epoch": 132.26,
      "grad_norm": 2.439603805541992,
      "learning_rate": 1.3229064422975228e-06,
      "loss": 0.4481,
      "step": 323100
    },
    {
      "epoch": 132.26,
      "grad_norm": 2.0519862174987793,
      "learning_rate": 1.3227636527020645e-06,
      "loss": 0.4646,
      "step": 323110
    },
    {
      "epoch": 132.26,
      "grad_norm": 1.9784821271896362,
      "learning_rate": 1.3226208677835866e-06,
      "loss": 0.4359,
      "step": 323120
    },
    {
      "epoch": 132.27,
      "grad_norm": 2.0143964290618896,
      "learning_rate": 1.3224780875427434e-06,
      "loss": 0.4596,
      "step": 323130
    },
    {
      "epoch": 132.27,
      "grad_norm": 2.3451786041259766,
      "learning_rate": 1.3223353119801885e-06,
      "loss": 0.473,
      "step": 323140
    },
    {
      "epoch": 132.28,
      "grad_norm": 2.43619441986084,
      "learning_rate": 1.322192541096576e-06,
      "loss": 0.46,
      "step": 323150
    },
    {
      "epoch": 132.28,
      "grad_norm": 1.8198432922363281,
      "learning_rate": 1.32204977489256e-06,
      "loss": 0.4478,
      "step": 323160
    },
    {
      "epoch": 132.28,
      "grad_norm": 2.62735652923584,
      "learning_rate": 1.3219070133687945e-06,
      "loss": 0.4693,
      "step": 323170
    },
    {
      "epoch": 132.29,
      "grad_norm": 1.5703802108764648,
      "learning_rate": 1.3217642565259345e-06,
      "loss": 0.4556,
      "step": 323180
    },
    {
      "epoch": 132.29,
      "grad_norm": 2.033806324005127,
      "learning_rate": 1.3216215043646319e-06,
      "loss": 0.4514,
      "step": 323190
    },
    {
      "epoch": 132.3,
      "grad_norm": 2.1486711502075195,
      "learning_rate": 1.3214787568855418e-06,
      "loss": 0.4654,
      "step": 323200
    },
    {
      "epoch": 132.3,
      "grad_norm": 2.262638568878174,
      "learning_rate": 1.3213360140893179e-06,
      "loss": 0.4421,
      "step": 323210
    },
    {
      "epoch": 132.3,
      "grad_norm": 1.647993803024292,
      "learning_rate": 1.321193275976614e-06,
      "loss": 0.4461,
      "step": 323220
    },
    {
      "epoch": 132.31,
      "grad_norm": 1.6162760257720947,
      "learning_rate": 1.3210505425480842e-06,
      "loss": 0.4389,
      "step": 323230
    },
    {
      "epoch": 132.31,
      "grad_norm": 2.4429497718811035,
      "learning_rate": 1.3209078138043823e-06,
      "loss": 0.4667,
      "step": 323240
    },
    {
      "epoch": 132.32,
      "grad_norm": 1.7132785320281982,
      "learning_rate": 1.320765089746162e-06,
      "loss": 0.4633,
      "step": 323250
    },
    {
      "epoch": 132.32,
      "grad_norm": 1.8131908178329468,
      "learning_rate": 1.3206223703740768e-06,
      "loss": 0.4715,
      "step": 323260
    },
    {
      "epoch": 132.33,
      "grad_norm": 2.014932870864868,
      "learning_rate": 1.320479655688781e-06,
      "loss": 0.4488,
      "step": 323270
    },
    {
      "epoch": 132.33,
      "grad_norm": 2.0172762870788574,
      "learning_rate": 1.3203369456909285e-06,
      "loss": 0.4669,
      "step": 323280
    },
    {
      "epoch": 132.33,
      "grad_norm": 1.9299966096878052,
      "learning_rate": 1.3201942403811713e-06,
      "loss": 0.444,
      "step": 323290
    },
    {
      "epoch": 132.34,
      "grad_norm": 1.975936770439148,
      "learning_rate": 1.3200515397601658e-06,
      "loss": 0.4615,
      "step": 323300
    },
    {
      "epoch": 132.34,
      "grad_norm": 1.7129385471343994,
      "learning_rate": 1.3199088438285645e-06,
      "loss": 0.4343,
      "step": 323310
    },
    {
      "epoch": 132.35,
      "grad_norm": 1.9884605407714844,
      "learning_rate": 1.3197661525870217e-06,
      "loss": 0.4602,
      "step": 323320
    },
    {
      "epoch": 132.35,
      "grad_norm": 1.7956680059432983,
      "learning_rate": 1.319623466036189e-06,
      "loss": 0.4664,
      "step": 323330
    },
    {
      "epoch": 132.35,
      "grad_norm": 1.9207947254180908,
      "learning_rate": 1.3194807841767222e-06,
      "loss": 0.4632,
      "step": 323340
    },
    {
      "epoch": 132.36,
      "grad_norm": 2.414546489715576,
      "learning_rate": 1.3193381070092734e-06,
      "loss": 0.4665,
      "step": 323350
    },
    {
      "epoch": 132.36,
      "grad_norm": 2.1371049880981445,
      "learning_rate": 1.3191954345344972e-06,
      "loss": 0.4469,
      "step": 323360
    },
    {
      "epoch": 132.37,
      "grad_norm": 1.984086036682129,
      "learning_rate": 1.3190527667530466e-06,
      "loss": 0.4353,
      "step": 323370
    },
    {
      "epoch": 132.37,
      "grad_norm": 2.074130058288574,
      "learning_rate": 1.3189101036655753e-06,
      "loss": 0.4492,
      "step": 323380
    },
    {
      "epoch": 132.37,
      "grad_norm": 2.2022128105163574,
      "learning_rate": 1.318767445272737e-06,
      "loss": 0.4567,
      "step": 323390
    },
    {
      "epoch": 132.38,
      "grad_norm": 2.1257474422454834,
      "learning_rate": 1.318624791575185e-06,
      "loss": 0.4482,
      "step": 323400
    },
    {
      "epoch": 132.38,
      "grad_norm": 1.763360619544983,
      "learning_rate": 1.3184821425735726e-06,
      "loss": 0.4388,
      "step": 323410
    },
    {
      "epoch": 132.39,
      "grad_norm": 1.8807588815689087,
      "learning_rate": 1.3183394982685538e-06,
      "loss": 0.4703,
      "step": 323420
    },
    {
      "epoch": 132.39,
      "grad_norm": 2.4506142139434814,
      "learning_rate": 1.3181968586607816e-06,
      "loss": 0.4648,
      "step": 323430
    },
    {
      "epoch": 132.39,
      "grad_norm": 2.0582199096679688,
      "learning_rate": 1.3180542237509093e-06,
      "loss": 0.471,
      "step": 323440
    },
    {
      "epoch": 132.4,
      "grad_norm": 2.0238289833068848,
      "learning_rate": 1.3179115935395906e-06,
      "loss": 0.4378,
      "step": 323450
    },
    {
      "epoch": 132.4,
      "grad_norm": 1.930624008178711,
      "learning_rate": 1.317768968027479e-06,
      "loss": 0.4464,
      "step": 323460
    },
    {
      "epoch": 132.41,
      "grad_norm": 1.970714807510376,
      "learning_rate": 1.3176263472152271e-06,
      "loss": 0.446,
      "step": 323470
    },
    {
      "epoch": 132.41,
      "grad_norm": 1.6530638933181763,
      "learning_rate": 1.3174837311034893e-06,
      "loss": 0.4555,
      "step": 323480
    },
    {
      "epoch": 132.42,
      "grad_norm": 2.567113161087036,
      "learning_rate": 1.317341119692918e-06,
      "loss": 0.4486,
      "step": 323490
    },
    {
      "epoch": 132.42,
      "grad_norm": 1.8090075254440308,
      "learning_rate": 1.3171985129841671e-06,
      "loss": 0.4541,
      "step": 323500
    },
    {
      "epoch": 132.42,
      "grad_norm": 1.897689700126648,
      "learning_rate": 1.3170559109778896e-06,
      "loss": 0.4547,
      "step": 323510
    },
    {
      "epoch": 132.43,
      "grad_norm": 2.1319642066955566,
      "learning_rate": 1.3169133136747384e-06,
      "loss": 0.4558,
      "step": 323520
    },
    {
      "epoch": 132.43,
      "grad_norm": 1.593375325202942,
      "learning_rate": 1.3167707210753673e-06,
      "loss": 0.4323,
      "step": 323530
    },
    {
      "epoch": 132.44,
      "grad_norm": 2.1995787620544434,
      "learning_rate": 1.316628133180429e-06,
      "loss": 0.4723,
      "step": 323540
    },
    {
      "epoch": 132.44,
      "grad_norm": 1.8271979093551636,
      "learning_rate": 1.3164855499905769e-06,
      "loss": 0.4748,
      "step": 323550
    },
    {
      "epoch": 132.44,
      "grad_norm": 1.868923544883728,
      "learning_rate": 1.3163429715064644e-06,
      "loss": 0.4307,
      "step": 323560
    },
    {
      "epoch": 132.45,
      "grad_norm": 2.0301482677459717,
      "learning_rate": 1.316200397728744e-06,
      "loss": 0.429,
      "step": 323570
    },
    {
      "epoch": 132.45,
      "grad_norm": 1.9074653387069702,
      "learning_rate": 1.3160578286580695e-06,
      "loss": 0.4439,
      "step": 323580
    },
    {
      "epoch": 132.46,
      "grad_norm": 2.6251449584960938,
      "learning_rate": 1.3159152642950936e-06,
      "loss": 0.4613,
      "step": 323590
    },
    {
      "epoch": 132.46,
      "grad_norm": 1.8926806449890137,
      "learning_rate": 1.3157727046404694e-06,
      "loss": 0.4481,
      "step": 323600
    },
    {
      "epoch": 132.46,
      "grad_norm": 2.1198196411132812,
      "learning_rate": 1.31563014969485e-06,
      "loss": 0.4513,
      "step": 323610
    },
    {
      "epoch": 132.47,
      "grad_norm": 1.7280813455581665,
      "learning_rate": 1.3154875994588887e-06,
      "loss": 0.4562,
      "step": 323620
    },
    {
      "epoch": 132.47,
      "grad_norm": 1.9750350713729858,
      "learning_rate": 1.3153450539332389e-06,
      "loss": 0.4489,
      "step": 323630
    },
    {
      "epoch": 132.48,
      "grad_norm": 1.9136521816253662,
      "learning_rate": 1.3152025131185518e-06,
      "loss": 0.4458,
      "step": 323640
    },
    {
      "epoch": 132.48,
      "grad_norm": 1.7279802560806274,
      "learning_rate": 1.3150599770154817e-06,
      "loss": 0.4318,
      "step": 323650
    },
    {
      "epoch": 132.48,
      "grad_norm": 2.053973436355591,
      "learning_rate": 1.314917445624681e-06,
      "loss": 0.4385,
      "step": 323660
    },
    {
      "epoch": 132.49,
      "grad_norm": 2.9408366680145264,
      "learning_rate": 1.3147749189468024e-06,
      "loss": 0.4562,
      "step": 323670
    },
    {
      "epoch": 132.49,
      "grad_norm": 2.0797388553619385,
      "learning_rate": 1.3146323969824998e-06,
      "loss": 0.4655,
      "step": 323680
    },
    {
      "epoch": 132.5,
      "grad_norm": 1.7740020751953125,
      "learning_rate": 1.3144898797324259e-06,
      "loss": 0.4655,
      "step": 323690
    },
    {
      "epoch": 132.5,
      "grad_norm": 2.2897584438323975,
      "learning_rate": 1.314347367197233e-06,
      "loss": 0.4589,
      "step": 323700
    },
    {
      "epoch": 132.51,
      "grad_norm": 2.20649790763855,
      "learning_rate": 1.314204859377574e-06,
      "loss": 0.4676,
      "step": 323710
    },
    {
      "epoch": 132.51,
      "grad_norm": 1.6220210790634155,
      "learning_rate": 1.3140623562741023e-06,
      "loss": 0.461,
      "step": 323720
    },
    {
      "epoch": 132.51,
      "grad_norm": 2.22499418258667,
      "learning_rate": 1.3139198578874698e-06,
      "loss": 0.4621,
      "step": 323730
    },
    {
      "epoch": 132.52,
      "grad_norm": 2.2466957569122314,
      "learning_rate": 1.31377736421833e-06,
      "loss": 0.459,
      "step": 323740
    },
    {
      "epoch": 132.52,
      "grad_norm": 2.0738234519958496,
      "learning_rate": 1.3136348752673352e-06,
      "loss": 0.4539,
      "step": 323750
    },
    {
      "epoch": 132.53,
      "grad_norm": 2.1855452060699463,
      "learning_rate": 1.313492391035139e-06,
      "loss": 0.4675,
      "step": 323760
    },
    {
      "epoch": 132.53,
      "grad_norm": 1.9653396606445312,
      "learning_rate": 1.3133499115223926e-06,
      "loss": 0.4584,
      "step": 323770
    },
    {
      "epoch": 132.53,
      "grad_norm": 1.7735984325408936,
      "learning_rate": 1.3132074367297494e-06,
      "loss": 0.4615,
      "step": 323780
    },
    {
      "epoch": 132.54,
      "grad_norm": 1.9126218557357788,
      "learning_rate": 1.313064966657862e-06,
      "loss": 0.4532,
      "step": 323790
    },
    {
      "epoch": 132.54,
      "grad_norm": 2.138300895690918,
      "learning_rate": 1.3129225013073831e-06,
      "loss": 0.4661,
      "step": 323800
    },
    {
      "epoch": 132.55,
      "grad_norm": 1.9977179765701294,
      "learning_rate": 1.312780040678966e-06,
      "loss": 0.4346,
      "step": 323810
    },
    {
      "epoch": 132.55,
      "grad_norm": 2.438015937805176,
      "learning_rate": 1.3126375847732618e-06,
      "loss": 0.4462,
      "step": 323820
    },
    {
      "epoch": 132.55,
      "grad_norm": 2.244915723800659,
      "learning_rate": 1.312495133590924e-06,
      "loss": 0.4646,
      "step": 323830
    },
    {
      "epoch": 132.56,
      "grad_norm": 2.1552672386169434,
      "learning_rate": 1.3123526871326053e-06,
      "loss": 0.4669,
      "step": 323840
    },
    {
      "epoch": 132.56,
      "grad_norm": 2.0774245262145996,
      "learning_rate": 1.3122102453989569e-06,
      "loss": 0.455,
      "step": 323850
    },
    {
      "epoch": 132.57,
      "grad_norm": 1.892822027206421,
      "learning_rate": 1.312067808390633e-06,
      "loss": 0.4385,
      "step": 323860
    },
    {
      "epoch": 132.57,
      "grad_norm": 2.0504202842712402,
      "learning_rate": 1.311925376108286e-06,
      "loss": 0.4415,
      "step": 323870
    },
    {
      "epoch": 132.57,
      "grad_norm": 2.050211191177368,
      "learning_rate": 1.3117829485525673e-06,
      "loss": 0.454,
      "step": 323880
    },
    {
      "epoch": 132.58,
      "grad_norm": 1.6346702575683594,
      "learning_rate": 1.3116405257241304e-06,
      "loss": 0.4505,
      "step": 323890
    },
    {
      "epoch": 132.58,
      "grad_norm": 1.9245306253433228,
      "learning_rate": 1.3114981076236277e-06,
      "loss": 0.4482,
      "step": 323900
    },
    {
      "epoch": 132.59,
      "grad_norm": 2.2710609436035156,
      "learning_rate": 1.3113556942517103e-06,
      "loss": 0.4431,
      "step": 323910
    },
    {
      "epoch": 132.59,
      "grad_norm": 2.0884759426116943,
      "learning_rate": 1.3112132856090311e-06,
      "loss": 0.4586,
      "step": 323920
    },
    {
      "epoch": 132.6,
      "grad_norm": 2.2738161087036133,
      "learning_rate": 1.3110708816962428e-06,
      "loss": 0.469,
      "step": 323930
    },
    {
      "epoch": 132.6,
      "grad_norm": 1.7150051593780518,
      "learning_rate": 1.3109284825139976e-06,
      "loss": 0.4617,
      "step": 323940
    },
    {
      "epoch": 132.6,
      "grad_norm": 1.7920085191726685,
      "learning_rate": 1.3107860880629478e-06,
      "loss": 0.4551,
      "step": 323950
    },
    {
      "epoch": 132.61,
      "grad_norm": 2.9749386310577393,
      "learning_rate": 1.310643698343746e-06,
      "loss": 0.4828,
      "step": 323960
    },
    {
      "epoch": 132.61,
      "grad_norm": 1.9892265796661377,
      "learning_rate": 1.3105013133570439e-06,
      "loss": 0.4494,
      "step": 323970
    },
    {
      "epoch": 132.62,
      "grad_norm": 1.847267985343933,
      "learning_rate": 1.3103589331034936e-06,
      "loss": 0.4639,
      "step": 323980
    },
    {
      "epoch": 132.62,
      "grad_norm": 1.9120327234268188,
      "learning_rate": 1.3102165575837484e-06,
      "loss": 0.4588,
      "step": 323990
    },
    {
      "epoch": 132.62,
      "grad_norm": 1.7139161825180054,
      "learning_rate": 1.3100741867984595e-06,
      "loss": 0.4434,
      "step": 324000
    },
    {
      "epoch": 132.63,
      "grad_norm": 2.1243810653686523,
      "learning_rate": 1.3099318207482797e-06,
      "loss": 0.4425,
      "step": 324010
    },
    {
      "epoch": 132.63,
      "grad_norm": 1.8206846714019775,
      "learning_rate": 1.3097894594338608e-06,
      "loss": 0.4442,
      "step": 324020
    },
    {
      "epoch": 132.64,
      "grad_norm": 2.2605390548706055,
      "learning_rate": 1.309647102855855e-06,
      "loss": 0.4662,
      "step": 324030
    },
    {
      "epoch": 132.64,
      "grad_norm": 1.7631326913833618,
      "learning_rate": 1.3095047510149138e-06,
      "loss": 0.4422,
      "step": 324040
    },
    {
      "epoch": 132.64,
      "grad_norm": 1.822770595550537,
      "learning_rate": 1.3093624039116908e-06,
      "loss": 0.4817,
      "step": 324050
    },
    {
      "epoch": 132.65,
      "grad_norm": 2.2954726219177246,
      "learning_rate": 1.3092200615468366e-06,
      "loss": 0.4411,
      "step": 324060
    },
    {
      "epoch": 132.65,
      "grad_norm": 2.3208186626434326,
      "learning_rate": 1.3090777239210042e-06,
      "loss": 0.4437,
      "step": 324070
    },
    {
      "epoch": 132.66,
      "grad_norm": 2.000959873199463,
      "learning_rate": 1.308935391034845e-06,
      "loss": 0.4547,
      "step": 324080
    },
    {
      "epoch": 132.66,
      "grad_norm": 2.161681890487671,
      "learning_rate": 1.3087930628890114e-06,
      "loss": 0.4529,
      "step": 324090
    },
    {
      "epoch": 132.66,
      "grad_norm": 1.8935803174972534,
      "learning_rate": 1.3086507394841551e-06,
      "loss": 0.4308,
      "step": 324100
    },
    {
      "epoch": 132.67,
      "grad_norm": 2.165410280227661,
      "learning_rate": 1.3085084208209283e-06,
      "loss": 0.4571,
      "step": 324110
    },
    {
      "epoch": 132.67,
      "grad_norm": 2.01918888092041,
      "learning_rate": 1.308366106899983e-06,
      "loss": 0.4605,
      "step": 324120
    },
    {
      "epoch": 132.68,
      "grad_norm": 1.965787410736084,
      "learning_rate": 1.3082237977219707e-06,
      "loss": 0.4561,
      "step": 324130
    },
    {
      "epoch": 132.68,
      "grad_norm": 1.8184819221496582,
      "learning_rate": 1.3080814932875439e-06,
      "loss": 0.4437,
      "step": 324140
    },
    {
      "epoch": 132.69,
      "grad_norm": 1.660237431526184,
      "learning_rate": 1.3079391935973539e-06,
      "loss": 0.4339,
      "step": 324150
    },
    {
      "epoch": 132.69,
      "grad_norm": 1.953795313835144,
      "learning_rate": 1.3077968986520527e-06,
      "loss": 0.4309,
      "step": 324160
    },
    {
      "epoch": 132.69,
      "grad_norm": 2.361196279525757,
      "learning_rate": 1.3076546084522924e-06,
      "loss": 0.4461,
      "step": 324170
    },
    {
      "epoch": 132.7,
      "grad_norm": 2.8638486862182617,
      "learning_rate": 1.3075123229987247e-06,
      "loss": 0.4431,
      "step": 324180
    },
    {
      "epoch": 132.7,
      "grad_norm": 2.279334545135498,
      "learning_rate": 1.307370042292001e-06,
      "loss": 0.4363,
      "step": 324190
    },
    {
      "epoch": 132.71,
      "grad_norm": 2.0605101585388184,
      "learning_rate": 1.3072277663327737e-06,
      "loss": 0.4658,
      "step": 324200
    },
    {
      "epoch": 132.71,
      "grad_norm": 2.981639862060547,
      "learning_rate": 1.307085495121695e-06,
      "loss": 0.4553,
      "step": 324210
    },
    {
      "epoch": 132.71,
      "grad_norm": 2.0050511360168457,
      "learning_rate": 1.3069432286594148e-06,
      "loss": 0.4392,
      "step": 324220
    },
    {
      "epoch": 132.72,
      "grad_norm": 1.9410114288330078,
      "learning_rate": 1.306800966946585e-06,
      "loss": 0.4498,
      "step": 324230
    },
    {
      "epoch": 132.72,
      "grad_norm": 2.1159355640411377,
      "learning_rate": 1.306658709983859e-06,
      "loss": 0.4562,
      "step": 324240
    },
    {
      "epoch": 132.73,
      "grad_norm": 1.9322748184204102,
      "learning_rate": 1.3065164577718878e-06,
      "loss": 0.4581,
      "step": 324250
    },
    {
      "epoch": 132.73,
      "grad_norm": 2.0931684970855713,
      "learning_rate": 1.306374210311323e-06,
      "loss": 0.4566,
      "step": 324260
    },
    {
      "epoch": 132.73,
      "grad_norm": 1.9810831546783447,
      "learning_rate": 1.3062319676028157e-06,
      "loss": 0.4433,
      "step": 324270
    },
    {
      "epoch": 132.74,
      "grad_norm": 2.0522079467773438,
      "learning_rate": 1.306089729647018e-06,
      "loss": 0.4432,
      "step": 324280
    },
    {
      "epoch": 132.74,
      "grad_norm": 2.48872971534729,
      "learning_rate": 1.3059474964445811e-06,
      "loss": 0.4604,
      "step": 324290
    },
    {
      "epoch": 132.75,
      "grad_norm": 1.9666671752929688,
      "learning_rate": 1.3058052679961567e-06,
      "loss": 0.4461,
      "step": 324300
    },
    {
      "epoch": 132.75,
      "grad_norm": 2.1910927295684814,
      "learning_rate": 1.3056630443023967e-06,
      "loss": 0.4643,
      "step": 324310
    },
    {
      "epoch": 132.75,
      "grad_norm": 2.1702065467834473,
      "learning_rate": 1.305520825363952e-06,
      "loss": 0.4599,
      "step": 324320
    },
    {
      "epoch": 132.76,
      "grad_norm": 2.058877468109131,
      "learning_rate": 1.3053786111814743e-06,
      "loss": 0.48,
      "step": 324330
    },
    {
      "epoch": 132.76,
      "grad_norm": 1.9221584796905518,
      "learning_rate": 1.3052364017556163e-06,
      "loss": 0.4753,
      "step": 324340
    },
    {
      "epoch": 132.77,
      "grad_norm": 2.0366194248199463,
      "learning_rate": 1.3050941970870272e-06,
      "loss": 0.4555,
      "step": 324350
    },
    {
      "epoch": 132.77,
      "grad_norm": 1.9091367721557617,
      "learning_rate": 1.3049519971763593e-06,
      "loss": 0.4754,
      "step": 324360
    },
    {
      "epoch": 132.78,
      "grad_norm": 1.9706761837005615,
      "learning_rate": 1.3048098020242645e-06,
      "loss": 0.4675,
      "step": 324370
    },
    {
      "epoch": 132.78,
      "grad_norm": 1.8440361022949219,
      "learning_rate": 1.3046676116313937e-06,
      "loss": 0.4616,
      "step": 324380
    },
    {
      "epoch": 132.78,
      "grad_norm": 2.3544018268585205,
      "learning_rate": 1.3045254259983986e-06,
      "loss": 0.466,
      "step": 324390
    },
    {
      "epoch": 132.79,
      "grad_norm": 2.1347568035125732,
      "learning_rate": 1.3043832451259297e-06,
      "loss": 0.4374,
      "step": 324400
    },
    {
      "epoch": 132.79,
      "grad_norm": 1.897146224975586,
      "learning_rate": 1.3042410690146394e-06,
      "loss": 0.4643,
      "step": 324410
    },
    {
      "epoch": 132.8,
      "grad_norm": 1.8854950666427612,
      "learning_rate": 1.3040988976651778e-06,
      "loss": 0.4535,
      "step": 324420
    },
    {
      "epoch": 132.8,
      "grad_norm": 2.248041868209839,
      "learning_rate": 1.3039567310781976e-06,
      "loss": 0.4468,
      "step": 324430
    },
    {
      "epoch": 132.8,
      "grad_norm": 2.3983569145202637,
      "learning_rate": 1.3038145692543495e-06,
      "loss": 0.4566,
      "step": 324440
    },
    {
      "epoch": 132.81,
      "grad_norm": 2.344038486480713,
      "learning_rate": 1.3036724121942844e-06,
      "loss": 0.4433,
      "step": 324450
    },
    {
      "epoch": 132.81,
      "grad_norm": 1.8703415393829346,
      "learning_rate": 1.3035302598986533e-06,
      "loss": 0.477,
      "step": 324460
    },
    {
      "epoch": 132.82,
      "grad_norm": 2.302725076675415,
      "learning_rate": 1.303388112368108e-06,
      "loss": 0.4607,
      "step": 324470
    },
    {
      "epoch": 132.82,
      "grad_norm": 2.1227638721466064,
      "learning_rate": 1.3032459696033006e-06,
      "loss": 0.4599,
      "step": 324480
    },
    {
      "epoch": 132.82,
      "grad_norm": 1.7514156103134155,
      "learning_rate": 1.3031038316048799e-06,
      "loss": 0.4764,
      "step": 324490
    },
    {
      "epoch": 132.83,
      "grad_norm": 2.0549697875976562,
      "learning_rate": 1.302961698373498e-06,
      "loss": 0.4581,
      "step": 324500
    },
    {
      "epoch": 132.83,
      "grad_norm": 2.40995454788208,
      "learning_rate": 1.3028195699098061e-06,
      "loss": 0.4454,
      "step": 324510
    },
    {
      "epoch": 132.84,
      "grad_norm": 2.0997514724731445,
      "learning_rate": 1.3026774462144556e-06,
      "loss": 0.4389,
      "step": 324520
    },
    {
      "epoch": 132.84,
      "grad_norm": 2.6964282989501953,
      "learning_rate": 1.302535327288097e-06,
      "loss": 0.4417,
      "step": 324530
    },
    {
      "epoch": 132.84,
      "grad_norm": 1.9142035245895386,
      "learning_rate": 1.3023932131313814e-06,
      "loss": 0.4566,
      "step": 324540
    },
    {
      "epoch": 132.85,
      "grad_norm": 2.118872880935669,
      "learning_rate": 1.3022511037449606e-06,
      "loss": 0.449,
      "step": 324550
    },
    {
      "epoch": 132.85,
      "grad_norm": 1.8597159385681152,
      "learning_rate": 1.3021089991294842e-06,
      "loss": 0.4559,
      "step": 324560
    },
    {
      "epoch": 132.86,
      "grad_norm": 1.843093752861023,
      "learning_rate": 1.3019668992856043e-06,
      "loss": 0.4444,
      "step": 324570
    },
    {
      "epoch": 132.86,
      "grad_norm": 2.1490681171417236,
      "learning_rate": 1.3018248042139715e-06,
      "loss": 0.4601,
      "step": 324580
    },
    {
      "epoch": 132.87,
      "grad_norm": 1.5651342868804932,
      "learning_rate": 1.3016827139152365e-06,
      "loss": 0.4525,
      "step": 324590
    },
    {
      "epoch": 132.87,
      "grad_norm": 2.41829514503479,
      "learning_rate": 1.3015406283900504e-06,
      "loss": 0.4629,
      "step": 324600
    },
    {
      "epoch": 132.87,
      "grad_norm": 1.8537169694900513,
      "learning_rate": 1.3013985476390631e-06,
      "loss": 0.4577,
      "step": 324610
    },
    {
      "epoch": 132.88,
      "grad_norm": 2.001542568206787,
      "learning_rate": 1.3012564716629281e-06,
      "loss": 0.4571,
      "step": 324620
    },
    {
      "epoch": 132.88,
      "grad_norm": 1.8111772537231445,
      "learning_rate": 1.3011144004622937e-06,
      "loss": 0.4726,
      "step": 324630
    },
    {
      "epoch": 132.89,
      "grad_norm": 2.1974775791168213,
      "learning_rate": 1.3009723340378117e-06,
      "loss": 0.4641,
      "step": 324640
    },
    {
      "epoch": 132.89,
      "grad_norm": 1.9130970239639282,
      "learning_rate": 1.3008302723901324e-06,
      "loss": 0.4488,
      "step": 324650
    },
    {
      "epoch": 132.89,
      "grad_norm": 1.8809767961502075,
      "learning_rate": 1.300688215519907e-06,
      "loss": 0.4667,
      "step": 324660
    },
    {
      "epoch": 132.9,
      "grad_norm": 1.9205161333084106,
      "learning_rate": 1.300546163427786e-06,
      "loss": 0.4795,
      "step": 324670
    },
    {
      "epoch": 132.9,
      "grad_norm": 2.3523342609405518,
      "learning_rate": 1.3004041161144201e-06,
      "loss": 0.4657,
      "step": 324680
    },
    {
      "epoch": 132.91,
      "grad_norm": 2.0893940925598145,
      "learning_rate": 1.3002620735804603e-06,
      "loss": 0.4423,
      "step": 324690
    },
    {
      "epoch": 132.91,
      "grad_norm": 2.127458333969116,
      "learning_rate": 1.3001200358265572e-06,
      "loss": 0.4733,
      "step": 324700
    },
    {
      "epoch": 132.91,
      "grad_norm": 1.9406328201293945,
      "learning_rate": 1.2999780028533612e-06,
      "loss": 0.4452,
      "step": 324710
    },
    {
      "epoch": 132.92,
      "grad_norm": 1.6360481977462769,
      "learning_rate": 1.299835974661523e-06,
      "loss": 0.4606,
      "step": 324720
    },
    {
      "epoch": 132.92,
      "grad_norm": 2.050006866455078,
      "learning_rate": 1.2996939512516937e-06,
      "loss": 0.4597,
      "step": 324730
    },
    {
      "epoch": 132.93,
      "grad_norm": 1.785683035850525,
      "learning_rate": 1.2995519326245228e-06,
      "loss": 0.4425,
      "step": 324740
    },
    {
      "epoch": 132.93,
      "grad_norm": 1.73183012008667,
      "learning_rate": 1.299409918780662e-06,
      "loss": 0.4611,
      "step": 324750
    },
    {
      "epoch": 132.93,
      "grad_norm": 1.714629054069519,
      "learning_rate": 1.2992679097207612e-06,
      "loss": 0.4265,
      "step": 324760
    },
    {
      "epoch": 132.94,
      "grad_norm": 2.688105821609497,
      "learning_rate": 1.299125905445471e-06,
      "loss": 0.4547,
      "step": 324770
    },
    {
      "epoch": 132.94,
      "grad_norm": 1.8965044021606445,
      "learning_rate": 1.298983905955442e-06,
      "loss": 0.4536,
      "step": 324780
    },
    {
      "epoch": 132.95,
      "grad_norm": 2.0011284351348877,
      "learning_rate": 1.298841911251326e-06,
      "loss": 0.4488,
      "step": 324790
    },
    {
      "epoch": 132.95,
      "grad_norm": 2.0684759616851807,
      "learning_rate": 1.29869992133377e-06,
      "loss": 0.4363,
      "step": 324800
    },
    {
      "epoch": 132.96,
      "grad_norm": 2.6170272827148438,
      "learning_rate": 1.2985579362034273e-06,
      "loss": 0.4522,
      "step": 324810
    },
    {
      "epoch": 132.96,
      "grad_norm": 1.7790783643722534,
      "learning_rate": 1.298415955860948e-06,
      "loss": 0.452,
      "step": 324820
    },
    {
      "epoch": 132.96,
      "grad_norm": 2.249814748764038,
      "learning_rate": 1.2982739803069817e-06,
      "loss": 0.4498,
      "step": 324830
    },
    {
      "epoch": 132.97,
      "grad_norm": 2.159548282623291,
      "learning_rate": 1.2981320095421793e-06,
      "loss": 0.4577,
      "step": 324840
    },
    {
      "epoch": 132.97,
      "grad_norm": 2.044229030609131,
      "learning_rate": 1.2979900435671907e-06,
      "loss": 0.4719,
      "step": 324850
    },
    {
      "epoch": 132.98,
      "grad_norm": 1.569928765296936,
      "learning_rate": 1.2978480823826668e-06,
      "loss": 0.454,
      "step": 324860
    },
    {
      "epoch": 132.98,
      "grad_norm": 1.8997101783752441,
      "learning_rate": 1.2977061259892578e-06,
      "loss": 0.4698,
      "step": 324870
    },
    {
      "epoch": 132.98,
      "grad_norm": 2.060133934020996,
      "learning_rate": 1.2975641743876134e-06,
      "loss": 0.454,
      "step": 324880
    },
    {
      "epoch": 132.99,
      "grad_norm": 2.459075927734375,
      "learning_rate": 1.2974222275783845e-06,
      "loss": 0.4513,
      "step": 324890
    },
    {
      "epoch": 132.99,
      "grad_norm": 2.1810951232910156,
      "learning_rate": 1.297280285562221e-06,
      "loss": 0.4375,
      "step": 324900
    },
    {
      "epoch": 133.0,
      "grad_norm": 1.9703141450881958,
      "learning_rate": 1.2971383483397733e-06,
      "loss": 0.4871,
      "step": 324910
    },
    {
      "epoch": 133.0,
      "eval_loss": 0.4560396671295166,
      "eval_runtime": 52.3363,
      "eval_samples_per_second": 65.901,
      "eval_steps_per_second": 8.254,
      "step": 324919
    },
    {
      "epoch": 133.0,
      "grad_norm": 2.2303786277770996,
      "learning_rate": 1.2969964159116913e-06,
      "loss": 0.4521,
      "step": 324920
    },
    {
      "epoch": 133.0,
      "grad_norm": 1.815033197402954,
      "learning_rate": 1.296854488278626e-06,
      "loss": 0.4596,
      "step": 324930
    },
    {
      "epoch": 133.01,
      "grad_norm": 1.9949060678482056,
      "learning_rate": 1.2967125654412268e-06,
      "loss": 0.4425,
      "step": 324940
    },
    {
      "epoch": 133.01,
      "grad_norm": 1.6768543720245361,
      "learning_rate": 1.2965706474001434e-06,
      "loss": 0.4694,
      "step": 324950
    },
    {
      "epoch": 133.02,
      "grad_norm": 2.21893310546875,
      "learning_rate": 1.2964287341560263e-06,
      "loss": 0.453,
      "step": 324960
    },
    {
      "epoch": 133.02,
      "grad_norm": 1.686044454574585,
      "learning_rate": 1.2962868257095258e-06,
      "loss": 0.4486,
      "step": 324970
    },
    {
      "epoch": 133.02,
      "grad_norm": 2.142627000808716,
      "learning_rate": 1.2961449220612922e-06,
      "loss": 0.4681,
      "step": 324980
    },
    {
      "epoch": 133.03,
      "grad_norm": 1.9257397651672363,
      "learning_rate": 1.2960030232119742e-06,
      "loss": 0.4461,
      "step": 324990
    },
    {
      "epoch": 133.03,
      "grad_norm": 2.0626513957977295,
      "learning_rate": 1.2958611291622236e-06,
      "loss": 0.4621,
      "step": 325000
    },
    {
      "epoch": 133.04,
      "grad_norm": 1.8816817998886108,
      "learning_rate": 1.2957192399126894e-06,
      "loss": 0.4558,
      "step": 325010
    },
    {
      "epoch": 133.04,
      "grad_norm": 2.0825235843658447,
      "learning_rate": 1.2955773554640223e-06,
      "loss": 0.4499,
      "step": 325020
    },
    {
      "epoch": 133.05,
      "grad_norm": 2.2859199047088623,
      "learning_rate": 1.2954354758168715e-06,
      "loss": 0.4683,
      "step": 325030
    },
    {
      "epoch": 133.05,
      "grad_norm": 1.759725570678711,
      "learning_rate": 1.295293600971887e-06,
      "loss": 0.44,
      "step": 325040
    },
    {
      "epoch": 133.05,
      "grad_norm": 2.425173759460449,
      "learning_rate": 1.2951517309297194e-06,
      "loss": 0.4321,
      "step": 325050
    },
    {
      "epoch": 133.06,
      "grad_norm": 2.6524431705474854,
      "learning_rate": 1.2950098656910186e-06,
      "loss": 0.4506,
      "step": 325060
    },
    {
      "epoch": 133.06,
      "grad_norm": 2.9454681873321533,
      "learning_rate": 1.2948680052564333e-06,
      "loss": 0.4563,
      "step": 325070
    },
    {
      "epoch": 133.07,
      "grad_norm": 1.975528359413147,
      "learning_rate": 1.2947261496266135e-06,
      "loss": 0.4513,
      "step": 325080
    },
    {
      "epoch": 133.07,
      "grad_norm": 1.8848471641540527,
      "learning_rate": 1.2945842988022097e-06,
      "loss": 0.4713,
      "step": 325090
    },
    {
      "epoch": 133.07,
      "grad_norm": 2.385676622390747,
      "learning_rate": 1.294442452783872e-06,
      "loss": 0.4625,
      "step": 325100
    },
    {
      "epoch": 133.08,
      "grad_norm": 1.96924889087677,
      "learning_rate": 1.294300611572249e-06,
      "loss": 0.4705,
      "step": 325110
    },
    {
      "epoch": 133.08,
      "grad_norm": 2.032925605773926,
      "learning_rate": 1.2941587751679915e-06,
      "loss": 0.4431,
      "step": 325120
    },
    {
      "epoch": 133.09,
      "grad_norm": 1.976453185081482,
      "learning_rate": 1.294016943571749e-06,
      "loss": 0.4532,
      "step": 325130
    },
    {
      "epoch": 133.09,
      "grad_norm": 1.7963718175888062,
      "learning_rate": 1.2938751167841706e-06,
      "loss": 0.4782,
      "step": 325140
    },
    {
      "epoch": 133.09,
      "grad_norm": 2.417478084564209,
      "learning_rate": 1.2937332948059066e-06,
      "loss": 0.4743,
      "step": 325150
    },
    {
      "epoch": 133.1,
      "grad_norm": 1.9520392417907715,
      "learning_rate": 1.2935914776376065e-06,
      "loss": 0.461,
      "step": 325160
    },
    {
      "epoch": 133.1,
      "grad_norm": 2.4397716522216797,
      "learning_rate": 1.2934496652799192e-06,
      "loss": 0.4579,
      "step": 325170
    },
    {
      "epoch": 133.11,
      "grad_norm": 12.804348945617676,
      "learning_rate": 1.2933078577334958e-06,
      "loss": 0.453,
      "step": 325180
    },
    {
      "epoch": 133.11,
      "grad_norm": 2.5257458686828613,
      "learning_rate": 1.2931660549989854e-06,
      "loss": 0.4572,
      "step": 325190
    },
    {
      "epoch": 133.12,
      "grad_norm": 2.2616021633148193,
      "learning_rate": 1.293024257077038e-06,
      "loss": 0.4749,
      "step": 325200
    },
    {
      "epoch": 133.12,
      "grad_norm": 2.1698710918426514,
      "learning_rate": 1.2928824639683018e-06,
      "loss": 0.4633,
      "step": 325210
    },
    {
      "epoch": 133.12,
      "grad_norm": 1.930013656616211,
      "learning_rate": 1.2927406756734267e-06,
      "loss": 0.4412,
      "step": 325220
    },
    {
      "epoch": 133.13,
      "grad_norm": 2.0870978832244873,
      "learning_rate": 1.2925988921930627e-06,
      "loss": 0.4737,
      "step": 325230
    },
    {
      "epoch": 133.13,
      "grad_norm": 2.541797399520874,
      "learning_rate": 1.2924571135278594e-06,
      "loss": 0.4675,
      "step": 325240
    },
    {
      "epoch": 133.14,
      "grad_norm": 1.8792359828948975,
      "learning_rate": 1.2923153396784657e-06,
      "loss": 0.4676,
      "step": 325250
    },
    {
      "epoch": 133.14,
      "grad_norm": 2.1818482875823975,
      "learning_rate": 1.2921735706455316e-06,
      "loss": 0.4463,
      "step": 325260
    },
    {
      "epoch": 133.14,
      "grad_norm": 2.0779800415039062,
      "learning_rate": 1.2920318064297062e-06,
      "loss": 0.4705,
      "step": 325270
    },
    {
      "epoch": 133.15,
      "grad_norm": 1.830719232559204,
      "learning_rate": 1.2918900470316389e-06,
      "loss": 0.4647,
      "step": 325280
    },
    {
      "epoch": 133.15,
      "grad_norm": 2.232002019882202,
      "learning_rate": 1.2917482924519792e-06,
      "loss": 0.4643,
      "step": 325290
    },
    {
      "epoch": 133.16,
      "grad_norm": 2.1501574516296387,
      "learning_rate": 1.291606542691376e-06,
      "loss": 0.4579,
      "step": 325300
    },
    {
      "epoch": 133.16,
      "grad_norm": 2.1824169158935547,
      "learning_rate": 1.2914647977504796e-06,
      "loss": 0.4826,
      "step": 325310
    },
    {
      "epoch": 133.16,
      "grad_norm": 1.908113718032837,
      "learning_rate": 1.2913230576299383e-06,
      "loss": 0.4468,
      "step": 325320
    },
    {
      "epoch": 133.17,
      "grad_norm": 2.1254971027374268,
      "learning_rate": 1.2911813223304022e-06,
      "loss": 0.4389,
      "step": 325330
    },
    {
      "epoch": 133.17,
      "grad_norm": 1.9209729433059692,
      "learning_rate": 1.2910395918525202e-06,
      "loss": 0.4497,
      "step": 325340
    },
    {
      "epoch": 133.18,
      "grad_norm": 1.7321892976760864,
      "learning_rate": 1.2908978661969412e-06,
      "loss": 0.4288,
      "step": 325350
    },
    {
      "epoch": 133.18,
      "grad_norm": 1.935577630996704,
      "learning_rate": 1.2907561453643148e-06,
      "loss": 0.4532,
      "step": 325360
    },
    {
      "epoch": 133.18,
      "grad_norm": 2.0765902996063232,
      "learning_rate": 1.2906144293552906e-06,
      "loss": 0.4752,
      "step": 325370
    },
    {
      "epoch": 133.19,
      "grad_norm": 1.9241832494735718,
      "learning_rate": 1.290472718170517e-06,
      "loss": 0.461,
      "step": 325380
    },
    {
      "epoch": 133.19,
      "grad_norm": 2.074516773223877,
      "learning_rate": 1.2903310118106437e-06,
      "loss": 0.4622,
      "step": 325390
    },
    {
      "epoch": 133.2,
      "grad_norm": 2.1826250553131104,
      "learning_rate": 1.2901893102763196e-06,
      "loss": 0.4761,
      "step": 325400
    },
    {
      "epoch": 133.2,
      "grad_norm": 2.0162203311920166,
      "learning_rate": 1.2900476135681937e-06,
      "loss": 0.4712,
      "step": 325410
    },
    {
      "epoch": 133.21,
      "grad_norm": 2.436004161834717,
      "learning_rate": 1.2899059216869153e-06,
      "loss": 0.4731,
      "step": 325420
    },
    {
      "epoch": 133.21,
      "grad_norm": 1.8955397605895996,
      "learning_rate": 1.2897642346331334e-06,
      "loss": 0.4646,
      "step": 325430
    },
    {
      "epoch": 133.21,
      "grad_norm": 1.8332589864730835,
      "learning_rate": 1.2896225524074971e-06,
      "loss": 0.4595,
      "step": 325440
    },
    {
      "epoch": 133.22,
      "grad_norm": 2.3426663875579834,
      "learning_rate": 1.2894808750106555e-06,
      "loss": 0.4476,
      "step": 325450
    },
    {
      "epoch": 133.22,
      "grad_norm": 2.0225906372070312,
      "learning_rate": 1.2893392024432574e-06,
      "loss": 0.4523,
      "step": 325460
    },
    {
      "epoch": 133.23,
      "grad_norm": 1.8722490072250366,
      "learning_rate": 1.2891975347059518e-06,
      "loss": 0.4765,
      "step": 325470
    },
    {
      "epoch": 133.23,
      "grad_norm": 1.8918229341506958,
      "learning_rate": 1.2890558717993877e-06,
      "loss": 0.4577,
      "step": 325480
    },
    {
      "epoch": 133.23,
      "grad_norm": 1.8668426275253296,
      "learning_rate": 1.2889142137242143e-06,
      "loss": 0.4688,
      "step": 325490
    },
    {
      "epoch": 133.24,
      "grad_norm": 2.18210506439209,
      "learning_rate": 1.2887725604810802e-06,
      "loss": 0.4523,
      "step": 325500
    },
    {
      "epoch": 133.24,
      "grad_norm": 2.0475234985351562,
      "learning_rate": 1.2886309120706349e-06,
      "loss": 0.4654,
      "step": 325510
    },
    {
      "epoch": 133.25,
      "grad_norm": 2.398071765899658,
      "learning_rate": 1.2884892684935264e-06,
      "loss": 0.4704,
      "step": 325520
    },
    {
      "epoch": 133.25,
      "grad_norm": 1.8312219381332397,
      "learning_rate": 1.2883476297504035e-06,
      "loss": 0.4532,
      "step": 325530
    },
    {
      "epoch": 133.25,
      "grad_norm": 1.9076803922653198,
      "learning_rate": 1.2882059958419152e-06,
      "loss": 0.4596,
      "step": 325540
    },
    {
      "epoch": 133.26,
      "grad_norm": 2.2846455574035645,
      "learning_rate": 1.28806436676871e-06,
      "loss": 0.4431,
      "step": 325550
    },
    {
      "epoch": 133.26,
      "grad_norm": 1.9913427829742432,
      "learning_rate": 1.2879227425314381e-06,
      "loss": 0.4508,
      "step": 325560
    },
    {
      "epoch": 133.27,
      "grad_norm": 1.907257080078125,
      "learning_rate": 1.2877811231307474e-06,
      "loss": 0.4342,
      "step": 325570
    },
    {
      "epoch": 133.27,
      "grad_norm": 2.2648613452911377,
      "learning_rate": 1.2876395085672867e-06,
      "loss": 0.4596,
      "step": 325580
    },
    {
      "epoch": 133.27,
      "grad_norm": 2.2028915882110596,
      "learning_rate": 1.2874978988417042e-06,
      "loss": 0.4601,
      "step": 325590
    },
    {
      "epoch": 133.28,
      "grad_norm": 1.984734058380127,
      "learning_rate": 1.2873562939546492e-06,
      "loss": 0.4647,
      "step": 325600
    },
    {
      "epoch": 133.28,
      "grad_norm": 1.8981740474700928,
      "learning_rate": 1.28721469390677e-06,
      "loss": 0.4612,
      "step": 325610
    },
    {
      "epoch": 133.29,
      "grad_norm": 2.242966890335083,
      "learning_rate": 1.2870730986987155e-06,
      "loss": 0.4516,
      "step": 325620
    },
    {
      "epoch": 133.29,
      "grad_norm": 1.809346079826355,
      "learning_rate": 1.2869315083311345e-06,
      "loss": 0.4395,
      "step": 325630
    },
    {
      "epoch": 133.3,
      "grad_norm": 1.712445616722107,
      "learning_rate": 1.286789922804675e-06,
      "loss": 0.4563,
      "step": 325640
    },
    {
      "epoch": 133.3,
      "grad_norm": 1.799453854560852,
      "learning_rate": 1.2866483421199868e-06,
      "loss": 0.4474,
      "step": 325650
    },
    {
      "epoch": 133.3,
      "grad_norm": 1.8434913158416748,
      "learning_rate": 1.2865067662777166e-06,
      "loss": 0.4503,
      "step": 325660
    },
    {
      "epoch": 133.31,
      "grad_norm": 2.0528290271759033,
      "learning_rate": 1.2863651952785143e-06,
      "loss": 0.4637,
      "step": 325670
    },
    {
      "epoch": 133.31,
      "grad_norm": 2.282937526702881,
      "learning_rate": 1.2862236291230279e-06,
      "loss": 0.4735,
      "step": 325680
    },
    {
      "epoch": 133.32,
      "grad_norm": 2.6295835971832275,
      "learning_rate": 1.2860820678119058e-06,
      "loss": 0.4427,
      "step": 325690
    },
    {
      "epoch": 133.32,
      "grad_norm": 2.3451714515686035,
      "learning_rate": 1.2859405113457971e-06,
      "loss": 0.4499,
      "step": 325700
    },
    {
      "epoch": 133.32,
      "grad_norm": 2.447033166885376,
      "learning_rate": 1.2857989597253496e-06,
      "loss": 0.4737,
      "step": 325710
    },
    {
      "epoch": 133.33,
      "grad_norm": 2.3444035053253174,
      "learning_rate": 1.285657412951212e-06,
      "loss": 0.446,
      "step": 325720
    },
    {
      "epoch": 133.33,
      "grad_norm": 1.7937517166137695,
      "learning_rate": 1.2855158710240325e-06,
      "loss": 0.4796,
      "step": 325730
    },
    {
      "epoch": 133.34,
      "grad_norm": 1.9893189668655396,
      "learning_rate": 1.285374333944459e-06,
      "loss": 0.4347,
      "step": 325740
    },
    {
      "epoch": 133.34,
      "grad_norm": 2.2595856189727783,
      "learning_rate": 1.2852328017131413e-06,
      "loss": 0.4631,
      "step": 325750
    },
    {
      "epoch": 133.34,
      "grad_norm": 2.113325357437134,
      "learning_rate": 1.2850912743307273e-06,
      "loss": 0.4663,
      "step": 325760
    },
    {
      "epoch": 133.35,
      "grad_norm": 1.9049745798110962,
      "learning_rate": 1.2849497517978645e-06,
      "loss": 0.4693,
      "step": 325770
    },
    {
      "epoch": 133.35,
      "grad_norm": 2.420693874359131,
      "learning_rate": 1.2848082341152028e-06,
      "loss": 0.45,
      "step": 325780
    },
    {
      "epoch": 133.36,
      "grad_norm": 2.167584180831909,
      "learning_rate": 1.2846667212833883e-06,
      "loss": 0.4505,
      "step": 325790
    },
    {
      "epoch": 133.36,
      "grad_norm": 1.783603310585022,
      "learning_rate": 1.2845252133030703e-06,
      "loss": 0.4694,
      "step": 325800
    },
    {
      "epoch": 133.36,
      "grad_norm": 2.0301074981689453,
      "learning_rate": 1.284383710174897e-06,
      "loss": 0.45,
      "step": 325810
    },
    {
      "epoch": 133.37,
      "grad_norm": 1.8668348789215088,
      "learning_rate": 1.2842422118995168e-06,
      "loss": 0.4453,
      "step": 325820
    },
    {
      "epoch": 133.37,
      "grad_norm": 1.9026111364364624,
      "learning_rate": 1.2841007184775775e-06,
      "loss": 0.4728,
      "step": 325830
    },
    {
      "epoch": 133.38,
      "grad_norm": 1.8272508382797241,
      "learning_rate": 1.2839592299097275e-06,
      "loss": 0.4507,
      "step": 325840
    },
    {
      "epoch": 133.38,
      "grad_norm": 2.441845417022705,
      "learning_rate": 1.283817746196615e-06,
      "loss": 0.4575,
      "step": 325850
    },
    {
      "epoch": 133.39,
      "grad_norm": 1.8816404342651367,
      "learning_rate": 1.283676267338888e-06,
      "loss": 0.4552,
      "step": 325860
    },
    {
      "epoch": 133.39,
      "grad_norm": 2.024312973022461,
      "learning_rate": 1.283534793337194e-06,
      "loss": 0.4522,
      "step": 325870
    },
    {
      "epoch": 133.39,
      "grad_norm": 2.1154839992523193,
      "learning_rate": 1.2833933241921827e-06,
      "loss": 0.4628,
      "step": 325880
    },
    {
      "epoch": 133.4,
      "grad_norm": 1.7826727628707886,
      "learning_rate": 1.2832518599045003e-06,
      "loss": 0.4567,
      "step": 325890
    },
    {
      "epoch": 133.4,
      "grad_norm": 2.021925687789917,
      "learning_rate": 1.283110400474796e-06,
      "loss": 0.4718,
      "step": 325900
    },
    {
      "epoch": 133.41,
      "grad_norm": 1.8845407962799072,
      "learning_rate": 1.2829689459037174e-06,
      "loss": 0.4449,
      "step": 325910
    },
    {
      "epoch": 133.41,
      "grad_norm": 2.0888872146606445,
      "learning_rate": 1.2828274961919123e-06,
      "loss": 0.4516,
      "step": 325920
    },
    {
      "epoch": 133.41,
      "grad_norm": 2.01625394821167,
      "learning_rate": 1.2826860513400292e-06,
      "loss": 0.4352,
      "step": 325930
    },
    {
      "epoch": 133.42,
      "grad_norm": 2.092499256134033,
      "learning_rate": 1.2825446113487156e-06,
      "loss": 0.4465,
      "step": 325940
    },
    {
      "epoch": 133.42,
      "grad_norm": 1.8968020677566528,
      "learning_rate": 1.2824031762186197e-06,
      "loss": 0.4532,
      "step": 325950
    },
    {
      "epoch": 133.43,
      "grad_norm": 1.7162017822265625,
      "learning_rate": 1.282261745950389e-06,
      "loss": 0.4458,
      "step": 325960
    },
    {
      "epoch": 133.43,
      "grad_norm": 2.3556628227233887,
      "learning_rate": 1.2821203205446714e-06,
      "loss": 0.4688,
      "step": 325970
    },
    {
      "epoch": 133.43,
      "grad_norm": 2.2018415927886963,
      "learning_rate": 1.2819789000021151e-06,
      "loss": 0.456,
      "step": 325980
    },
    {
      "epoch": 133.44,
      "grad_norm": 1.8374582529067993,
      "learning_rate": 1.2818374843233681e-06,
      "loss": 0.4594,
      "step": 325990
    },
    {
      "epoch": 133.44,
      "grad_norm": 2.124715566635132,
      "learning_rate": 1.2816960735090778e-06,
      "loss": 0.4533,
      "step": 326000
    },
    {
      "epoch": 133.45,
      "grad_norm": 1.8594390153884888,
      "learning_rate": 1.2815546675598919e-06,
      "loss": 0.4534,
      "step": 326010
    },
    {
      "epoch": 133.45,
      "grad_norm": 1.6807502508163452,
      "learning_rate": 1.2814132664764583e-06,
      "loss": 0.4602,
      "step": 326020
    },
    {
      "epoch": 133.45,
      "grad_norm": 2.799467086791992,
      "learning_rate": 1.2812718702594245e-06,
      "loss": 0.4564,
      "step": 326030
    },
    {
      "epoch": 133.46,
      "grad_norm": 1.6573123931884766,
      "learning_rate": 1.2811304789094389e-06,
      "loss": 0.4618,
      "step": 326040
    },
    {
      "epoch": 133.46,
      "grad_norm": 1.9493987560272217,
      "learning_rate": 1.2809890924271485e-06,
      "loss": 0.4714,
      "step": 326050
    },
    {
      "epoch": 133.47,
      "grad_norm": 2.1473865509033203,
      "learning_rate": 1.2808477108132018e-06,
      "loss": 0.4514,
      "step": 326060
    },
    {
      "epoch": 133.47,
      "grad_norm": 1.9923549890518188,
      "learning_rate": 1.2807063340682452e-06,
      "loss": 0.4474,
      "step": 326070
    },
    {
      "epoch": 133.48,
      "grad_norm": 2.0980582237243652,
      "learning_rate": 1.2805649621929272e-06,
      "loss": 0.4364,
      "step": 326080
    },
    {
      "epoch": 133.48,
      "grad_norm": 1.9658254384994507,
      "learning_rate": 1.2804235951878958e-06,
      "loss": 0.4495,
      "step": 326090
    },
    {
      "epoch": 133.48,
      "grad_norm": 1.9985991716384888,
      "learning_rate": 1.2802822330537977e-06,
      "loss": 0.4484,
      "step": 326100
    },
    {
      "epoch": 133.49,
      "grad_norm": 2.433720588684082,
      "learning_rate": 1.2801408757912804e-06,
      "loss": 0.433,
      "step": 326110
    },
    {
      "epoch": 133.49,
      "grad_norm": 2.3532707691192627,
      "learning_rate": 1.2799995234009908e-06,
      "loss": 0.4595,
      "step": 326120
    },
    {
      "epoch": 133.5,
      "grad_norm": 1.8320744037628174,
      "learning_rate": 1.2798581758835786e-06,
      "loss": 0.4362,
      "step": 326130
    },
    {
      "epoch": 133.5,
      "grad_norm": 2.2752931118011475,
      "learning_rate": 1.2797168332396898e-06,
      "loss": 0.4729,
      "step": 326140
    },
    {
      "epoch": 133.5,
      "grad_norm": 1.8605141639709473,
      "learning_rate": 1.2795754954699721e-06,
      "loss": 0.4603,
      "step": 326150
    },
    {
      "epoch": 133.51,
      "grad_norm": 2.07676100730896,
      "learning_rate": 1.2794341625750734e-06,
      "loss": 0.4645,
      "step": 326160
    },
    {
      "epoch": 133.51,
      "grad_norm": 2.311345338821411,
      "learning_rate": 1.2792928345556403e-06,
      "loss": 0.4966,
      "step": 326170
    },
    {
      "epoch": 133.52,
      "grad_norm": 1.946726679801941,
      "learning_rate": 1.279151511412321e-06,
      "loss": 0.4563,
      "step": 326180
    },
    {
      "epoch": 133.52,
      "grad_norm": 1.981999397277832,
      "learning_rate": 1.279010193145762e-06,
      "loss": 0.4708,
      "step": 326190
    },
    {
      "epoch": 133.52,
      "grad_norm": 1.9307234287261963,
      "learning_rate": 1.2788688797566118e-06,
      "loss": 0.4593,
      "step": 326200
    },
    {
      "epoch": 133.53,
      "grad_norm": 2.016178846359253,
      "learning_rate": 1.2787275712455164e-06,
      "loss": 0.4356,
      "step": 326210
    },
    {
      "epoch": 133.53,
      "grad_norm": 1.9262720346450806,
      "learning_rate": 1.2785862676131243e-06,
      "loss": 0.4362,
      "step": 326220
    },
    {
      "epoch": 133.54,
      "grad_norm": 1.792181134223938,
      "learning_rate": 1.2784449688600832e-06,
      "loss": 0.4371,
      "step": 326230
    },
    {
      "epoch": 133.54,
      "grad_norm": 2.2386343479156494,
      "learning_rate": 1.2783036749870383e-06,
      "loss": 0.4675,
      "step": 326240
    },
    {
      "epoch": 133.54,
      "grad_norm": 1.7048252820968628,
      "learning_rate": 1.278162385994638e-06,
      "loss": 0.4448,
      "step": 326250
    },
    {
      "epoch": 133.55,
      "grad_norm": 1.7950149774551392,
      "learning_rate": 1.2780211018835298e-06,
      "loss": 0.4661,
      "step": 326260
    },
    {
      "epoch": 133.55,
      "grad_norm": 2.167146682739258,
      "learning_rate": 1.2778798226543602e-06,
      "loss": 0.4642,
      "step": 326270
    },
    {
      "epoch": 133.56,
      "grad_norm": 2.3064913749694824,
      "learning_rate": 1.2777385483077772e-06,
      "loss": 0.4392,
      "step": 326280
    },
    {
      "epoch": 133.56,
      "grad_norm": 1.743450403213501,
      "learning_rate": 1.2775972788444274e-06,
      "loss": 0.4327,
      "step": 326290
    },
    {
      "epoch": 133.57,
      "grad_norm": 1.9465361833572388,
      "learning_rate": 1.2774560142649584e-06,
      "loss": 0.4556,
      "step": 326300
    },
    {
      "epoch": 133.57,
      "grad_norm": 1.848559856414795,
      "learning_rate": 1.2773147545700157e-06,
      "loss": 0.4464,
      "step": 326310
    },
    {
      "epoch": 133.57,
      "grad_norm": 2.2289955615997314,
      "learning_rate": 1.2771734997602488e-06,
      "loss": 0.4434,
      "step": 326320
    },
    {
      "epoch": 133.58,
      "grad_norm": 2.0156798362731934,
      "learning_rate": 1.2770322498363036e-06,
      "loss": 0.4679,
      "step": 326330
    },
    {
      "epoch": 133.58,
      "grad_norm": 2.1402876377105713,
      "learning_rate": 1.2768910047988272e-06,
      "loss": 0.4583,
      "step": 326340
    },
    {
      "epoch": 133.59,
      "grad_norm": 1.7643957138061523,
      "learning_rate": 1.276749764648467e-06,
      "loss": 0.4632,
      "step": 326350
    },
    {
      "epoch": 133.59,
      "grad_norm": 2.0846030712127686,
      "learning_rate": 1.27660852938587e-06,
      "loss": 0.4405,
      "step": 326360
    },
    {
      "epoch": 133.59,
      "grad_norm": 2.4440720081329346,
      "learning_rate": 1.2764672990116821e-06,
      "loss": 0.4405,
      "step": 326370
    },
    {
      "epoch": 133.6,
      "grad_norm": 1.9606494903564453,
      "learning_rate": 1.2763260735265512e-06,
      "loss": 0.4591,
      "step": 326380
    },
    {
      "epoch": 133.6,
      "grad_norm": 2.3137459754943848,
      "learning_rate": 1.276184852931124e-06,
      "loss": 0.4837,
      "step": 326390
    },
    {
      "epoch": 133.61,
      "grad_norm": 1.7406353950500488,
      "learning_rate": 1.276043637226047e-06,
      "loss": 0.4523,
      "step": 326400
    },
    {
      "epoch": 133.61,
      "grad_norm": 1.791192650794983,
      "learning_rate": 1.2759024264119674e-06,
      "loss": 0.4567,
      "step": 326410
    },
    {
      "epoch": 133.61,
      "grad_norm": 1.8027111291885376,
      "learning_rate": 1.2757612204895326e-06,
      "loss": 0.456,
      "step": 326420
    },
    {
      "epoch": 133.62,
      "grad_norm": 1.8641273975372314,
      "learning_rate": 1.2756200194593887e-06,
      "loss": 0.4495,
      "step": 326430
    },
    {
      "epoch": 133.62,
      "grad_norm": 1.6426186561584473,
      "learning_rate": 1.275478823322183e-06,
      "loss": 0.4511,
      "step": 326440
    },
    {
      "epoch": 133.63,
      "grad_norm": 1.8812633752822876,
      "learning_rate": 1.2753376320785621e-06,
      "loss": 0.4587,
      "step": 326450
    },
    {
      "epoch": 133.63,
      "grad_norm": 2.6223371028900146,
      "learning_rate": 1.275196445729173e-06,
      "loss": 0.4634,
      "step": 326460
    },
    {
      "epoch": 133.63,
      "grad_norm": 1.6648725271224976,
      "learning_rate": 1.2750552642746619e-06,
      "loss": 0.4421,
      "step": 326470
    },
    {
      "epoch": 133.64,
      "grad_norm": 2.3869409561157227,
      "learning_rate": 1.2749140877156757e-06,
      "loss": 0.4464,
      "step": 326480
    },
    {
      "epoch": 133.64,
      "grad_norm": 1.9098865985870361,
      "learning_rate": 1.2747729160528609e-06,
      "loss": 0.4755,
      "step": 326490
    },
    {
      "epoch": 133.65,
      "grad_norm": 1.9983258247375488,
      "learning_rate": 1.2746317492868663e-06,
      "loss": 0.4555,
      "step": 326500
    },
    {
      "epoch": 133.65,
      "grad_norm": 1.7796627283096313,
      "learning_rate": 1.2744905874183357e-06,
      "loss": 0.4588,
      "step": 326510
    },
    {
      "epoch": 133.66,
      "grad_norm": 2.1965179443359375,
      "learning_rate": 1.2743494304479171e-06,
      "loss": 0.4643,
      "step": 326520
    },
    {
      "epoch": 133.66,
      "grad_norm": 1.7024989128112793,
      "learning_rate": 1.2742082783762567e-06,
      "loss": 0.4454,
      "step": 326530
    },
    {
      "epoch": 133.66,
      "grad_norm": 1.8590408563613892,
      "learning_rate": 1.2740671312040012e-06,
      "loss": 0.473,
      "step": 326540
    },
    {
      "epoch": 133.67,
      "grad_norm": 2.0069687366485596,
      "learning_rate": 1.2739259889317975e-06,
      "loss": 0.4747,
      "step": 326550
    },
    {
      "epoch": 133.67,
      "grad_norm": 1.9431143999099731,
      "learning_rate": 1.273784851560292e-06,
      "loss": 0.4557,
      "step": 326560
    },
    {
      "epoch": 133.68,
      "grad_norm": 2.134946346282959,
      "learning_rate": 1.273643719090131e-06,
      "loss": 0.4701,
      "step": 326570
    },
    {
      "epoch": 133.68,
      "grad_norm": 2.138921022415161,
      "learning_rate": 1.273502591521961e-06,
      "loss": 0.4632,
      "step": 326580
    },
    {
      "epoch": 133.68,
      "grad_norm": 1.7893171310424805,
      "learning_rate": 1.2733614688564286e-06,
      "loss": 0.4546,
      "step": 326590
    },
    {
      "epoch": 133.69,
      "grad_norm": 2.12670636177063,
      "learning_rate": 1.2732203510941805e-06,
      "loss": 0.4339,
      "step": 326600
    },
    {
      "epoch": 133.69,
      "grad_norm": 2.3999106884002686,
      "learning_rate": 1.2730792382358627e-06,
      "loss": 0.4379,
      "step": 326610
    },
    {
      "epoch": 133.7,
      "grad_norm": 1.898551344871521,
      "learning_rate": 1.272938130282122e-06,
      "loss": 0.4529,
      "step": 326620
    },
    {
      "epoch": 133.7,
      "grad_norm": 2.281315803527832,
      "learning_rate": 1.2727970272336045e-06,
      "loss": 0.4367,
      "step": 326630
    },
    {
      "epoch": 133.7,
      "grad_norm": 1.8108278512954712,
      "learning_rate": 1.2726559290909572e-06,
      "loss": 0.4582,
      "step": 326640
    },
    {
      "epoch": 133.71,
      "grad_norm": 1.864548921585083,
      "learning_rate": 1.2725148358548255e-06,
      "loss": 0.4408,
      "step": 326650
    },
    {
      "epoch": 133.71,
      "grad_norm": 2.3921186923980713,
      "learning_rate": 1.2723737475258563e-06,
      "loss": 0.4657,
      "step": 326660
    },
    {
      "epoch": 133.72,
      "grad_norm": 1.7745857238769531,
      "learning_rate": 1.2722326641046964e-06,
      "loss": 0.4697,
      "step": 326670
    },
    {
      "epoch": 133.72,
      "grad_norm": 1.8807053565979004,
      "learning_rate": 1.27209158559199e-06,
      "loss": 0.4509,
      "step": 326680
    },
    {
      "epoch": 133.72,
      "grad_norm": 2.681293249130249,
      "learning_rate": 1.271950511988386e-06,
      "loss": 0.4555,
      "step": 326690
    },
    {
      "epoch": 133.73,
      "grad_norm": 2.0612308979034424,
      "learning_rate": 1.271809443294529e-06,
      "loss": 0.4739,
      "step": 326700
    },
    {
      "epoch": 133.73,
      "grad_norm": 1.9603512287139893,
      "learning_rate": 1.2716683795110658e-06,
      "loss": 0.4604,
      "step": 326710
    },
    {
      "epoch": 133.74,
      "grad_norm": 1.8993258476257324,
      "learning_rate": 1.2715273206386425e-06,
      "loss": 0.4458,
      "step": 326720
    },
    {
      "epoch": 133.74,
      "grad_norm": 2.262238025665283,
      "learning_rate": 1.2713862666779052e-06,
      "loss": 0.4632,
      "step": 326730
    },
    {
      "epoch": 133.75,
      "grad_norm": 2.349079132080078,
      "learning_rate": 1.2712452176295003e-06,
      "loss": 0.4453,
      "step": 326740
    },
    {
      "epoch": 133.75,
      "grad_norm": 1.8450000286102295,
      "learning_rate": 1.2711041734940735e-06,
      "loss": 0.4532,
      "step": 326750
    },
    {
      "epoch": 133.75,
      "grad_norm": 1.912932276725769,
      "learning_rate": 1.270963134272271e-06,
      "loss": 0.4466,
      "step": 326760
    },
    {
      "epoch": 133.76,
      "grad_norm": 1.7875354290008545,
      "learning_rate": 1.270822099964739e-06,
      "loss": 0.4492,
      "step": 326770
    },
    {
      "epoch": 133.76,
      "grad_norm": 1.7463735342025757,
      "learning_rate": 1.2706810705721239e-06,
      "loss": 0.4339,
      "step": 326780
    },
    {
      "epoch": 133.77,
      "grad_norm": 1.8775428533554077,
      "learning_rate": 1.270540046095071e-06,
      "loss": 0.4466,
      "step": 326790
    },
    {
      "epoch": 133.77,
      "grad_norm": 1.924136757850647,
      "learning_rate": 1.2703990265342269e-06,
      "loss": 0.4414,
      "step": 326800
    },
    {
      "epoch": 133.77,
      "grad_norm": 2.322490930557251,
      "learning_rate": 1.2702580118902384e-06,
      "loss": 0.4691,
      "step": 326810
    },
    {
      "epoch": 133.78,
      "grad_norm": 2.0484602451324463,
      "learning_rate": 1.2701170021637492e-06,
      "loss": 0.4606,
      "step": 326820
    },
    {
      "epoch": 133.78,
      "grad_norm": 1.9658540487289429,
      "learning_rate": 1.2699759973554067e-06,
      "loss": 0.441,
      "step": 326830
    },
    {
      "epoch": 133.79,
      "grad_norm": 2.3806464672088623,
      "learning_rate": 1.2698349974658566e-06,
      "loss": 0.459,
      "step": 326840
    },
    {
      "epoch": 133.79,
      "grad_norm": 1.5606900453567505,
      "learning_rate": 1.2696940024957446e-06,
      "loss": 0.459,
      "step": 326850
    },
    {
      "epoch": 133.79,
      "grad_norm": 1.8982386589050293,
      "learning_rate": 1.2695530124457169e-06,
      "loss": 0.4609,
      "step": 326860
    },
    {
      "epoch": 133.8,
      "grad_norm": 1.6615477800369263,
      "learning_rate": 1.2694120273164185e-06,
      "loss": 0.444,
      "step": 326870
    },
    {
      "epoch": 133.8,
      "grad_norm": 1.6268593072891235,
      "learning_rate": 1.269271047108497e-06,
      "loss": 0.456,
      "step": 326880
    },
    {
      "epoch": 133.81,
      "grad_norm": 2.375661849975586,
      "learning_rate": 1.2691300718225969e-06,
      "loss": 0.4645,
      "step": 326890
    },
    {
      "epoch": 133.81,
      "grad_norm": 1.672064185142517,
      "learning_rate": 1.2689891014593642e-06,
      "loss": 0.4426,
      "step": 326900
    },
    {
      "epoch": 133.81,
      "grad_norm": 2.079123020172119,
      "learning_rate": 1.2688481360194449e-06,
      "loss": 0.4313,
      "step": 326910
    },
    {
      "epoch": 133.82,
      "grad_norm": 2.036283493041992,
      "learning_rate": 1.2687071755034845e-06,
      "loss": 0.4477,
      "step": 326920
    },
    {
      "epoch": 133.82,
      "grad_norm": 2.2276251316070557,
      "learning_rate": 1.2685662199121289e-06,
      "loss": 0.465,
      "step": 326930
    },
    {
      "epoch": 133.83,
      "grad_norm": 1.8249174356460571,
      "learning_rate": 1.2684252692460233e-06,
      "loss": 0.4543,
      "step": 326940
    },
    {
      "epoch": 133.83,
      "grad_norm": 1.9845001697540283,
      "learning_rate": 1.2682843235058152e-06,
      "loss": 0.4653,
      "step": 326950
    },
    {
      "epoch": 133.84,
      "grad_norm": 1.689119815826416,
      "learning_rate": 1.2681433826921477e-06,
      "loss": 0.4391,
      "step": 326960
    },
    {
      "epoch": 133.84,
      "grad_norm": 1.9199227094650269,
      "learning_rate": 1.2680024468056674e-06,
      "loss": 0.4729,
      "step": 326970
    },
    {
      "epoch": 133.84,
      "grad_norm": 2.241748094558716,
      "learning_rate": 1.26786151584702e-06,
      "loss": 0.4563,
      "step": 326980
    },
    {
      "epoch": 133.85,
      "grad_norm": 2.2552356719970703,
      "learning_rate": 1.2677205898168515e-06,
      "loss": 0.4399,
      "step": 326990
    },
    {
      "epoch": 133.85,
      "grad_norm": 1.9305776357650757,
      "learning_rate": 1.2675796687158065e-06,
      "loss": 0.4477,
      "step": 327000
    },
    {
      "epoch": 133.86,
      "grad_norm": 2.2002053260803223,
      "learning_rate": 1.2674387525445315e-06,
      "loss": 0.4702,
      "step": 327010
    },
    {
      "epoch": 133.86,
      "grad_norm": 2.127288579940796,
      "learning_rate": 1.2672978413036717e-06,
      "loss": 0.434,
      "step": 327020
    },
    {
      "epoch": 133.86,
      "grad_norm": 2.1013922691345215,
      "learning_rate": 1.2671569349938724e-06,
      "loss": 0.4309,
      "step": 327030
    },
    {
      "epoch": 133.87,
      "grad_norm": 1.6398051977157593,
      "learning_rate": 1.2670160336157795e-06,
      "loss": 0.4569,
      "step": 327040
    },
    {
      "epoch": 133.87,
      "grad_norm": 2.073401689529419,
      "learning_rate": 1.2668751371700376e-06,
      "loss": 0.4686,
      "step": 327050
    },
    {
      "epoch": 133.88,
      "grad_norm": 2.1130268573760986,
      "learning_rate": 1.266734245657292e-06,
      "loss": 0.441,
      "step": 327060
    },
    {
      "epoch": 133.88,
      "grad_norm": 2.2090017795562744,
      "learning_rate": 1.2665933590781898e-06,
      "loss": 0.4342,
      "step": 327070
    },
    {
      "epoch": 133.88,
      "grad_norm": 2.0015804767608643,
      "learning_rate": 1.266452477433376e-06,
      "loss": 0.4526,
      "step": 327080
    },
    {
      "epoch": 133.89,
      "grad_norm": 1.9015265703201294,
      "learning_rate": 1.2663116007234949e-06,
      "loss": 0.4717,
      "step": 327090
    },
    {
      "epoch": 133.89,
      "grad_norm": 1.9125016927719116,
      "learning_rate": 1.2661707289491919e-06,
      "loss": 0.4585,
      "step": 327100
    },
    {
      "epoch": 133.9,
      "grad_norm": 1.8918768167495728,
      "learning_rate": 1.2660298621111127e-06,
      "loss": 0.4452,
      "step": 327110
    },
    {
      "epoch": 133.9,
      "grad_norm": 1.80948805809021,
      "learning_rate": 1.2658890002099023e-06,
      "loss": 0.4606,
      "step": 327120
    },
    {
      "epoch": 133.91,
      "grad_norm": 2.0051169395446777,
      "learning_rate": 1.2657481432462065e-06,
      "loss": 0.4258,
      "step": 327130
    },
    {
      "epoch": 133.91,
      "grad_norm": 2.301448106765747,
      "learning_rate": 1.26560729122067e-06,
      "loss": 0.4901,
      "step": 327140
    },
    {
      "epoch": 133.91,
      "grad_norm": 1.7765957117080688,
      "learning_rate": 1.2654664441339388e-06,
      "loss": 0.4375,
      "step": 327150
    },
    {
      "epoch": 133.92,
      "grad_norm": 2.051299810409546,
      "learning_rate": 1.265325601986657e-06,
      "loss": 0.4699,
      "step": 327160
    },
    {
      "epoch": 133.92,
      "grad_norm": 1.5415992736816406,
      "learning_rate": 1.2651847647794708e-06,
      "loss": 0.4548,
      "step": 327170
    },
    {
      "epoch": 133.93,
      "grad_norm": 2.233232021331787,
      "learning_rate": 1.2650439325130246e-06,
      "loss": 0.4492,
      "step": 327180
    },
    {
      "epoch": 133.93,
      "grad_norm": 2.1619560718536377,
      "learning_rate": 1.2649031051879641e-06,
      "loss": 0.4475,
      "step": 327190
    },
    {
      "epoch": 133.93,
      "grad_norm": 1.7436671257019043,
      "learning_rate": 1.264762282804934e-06,
      "loss": 0.4591,
      "step": 327200
    },
    {
      "epoch": 133.94,
      "grad_norm": 2.436936140060425,
      "learning_rate": 1.2646214653645795e-06,
      "loss": 0.4526,
      "step": 327210
    },
    {
      "epoch": 133.94,
      "grad_norm": 2.2910449504852295,
      "learning_rate": 1.2644806528675458e-06,
      "loss": 0.4488,
      "step": 327220
    },
    {
      "epoch": 133.95,
      "grad_norm": 2.1364922523498535,
      "learning_rate": 1.2643398453144776e-06,
      "loss": 0.4478,
      "step": 327230
    },
    {
      "epoch": 133.95,
      "grad_norm": 1.8073482513427734,
      "learning_rate": 1.2641990427060204e-06,
      "loss": 0.44,
      "step": 327240
    },
    {
      "epoch": 133.95,
      "grad_norm": 1.790016531944275,
      "learning_rate": 1.2640582450428187e-06,
      "loss": 0.4442,
      "step": 327250
    },
    {
      "epoch": 133.96,
      "grad_norm": 1.863982915878296,
      "learning_rate": 1.2639174523255178e-06,
      "loss": 0.4539,
      "step": 327260
    },
    {
      "epoch": 133.96,
      "grad_norm": 1.9939591884613037,
      "learning_rate": 1.2637766645547629e-06,
      "loss": 0.4519,
      "step": 327270
    },
    {
      "epoch": 133.97,
      "grad_norm": 1.9043668508529663,
      "learning_rate": 1.2636358817311983e-06,
      "loss": 0.4216,
      "step": 327280
    },
    {
      "epoch": 133.97,
      "grad_norm": 1.9776756763458252,
      "learning_rate": 1.2634951038554692e-06,
      "loss": 0.477,
      "step": 327290
    },
    {
      "epoch": 133.97,
      "grad_norm": 1.6934117078781128,
      "learning_rate": 1.2633543309282204e-06,
      "loss": 0.4488,
      "step": 327300
    },
    {
      "epoch": 133.98,
      "grad_norm": 1.7741893529891968,
      "learning_rate": 1.2632135629500968e-06,
      "loss": 0.4547,
      "step": 327310
    },
    {
      "epoch": 133.98,
      "grad_norm": 2.683695077896118,
      "learning_rate": 1.2630727999217435e-06,
      "loss": 0.475,
      "step": 327320
    },
    {
      "epoch": 133.99,
      "grad_norm": 1.672523021697998,
      "learning_rate": 1.262932041843805e-06,
      "loss": 0.4646,
      "step": 327330
    },
    {
      "epoch": 133.99,
      "grad_norm": 2.441046953201294,
      "learning_rate": 1.2627912887169263e-06,
      "loss": 0.4777,
      "step": 327340
    },
    {
      "epoch": 134.0,
      "grad_norm": 2.183504819869995,
      "learning_rate": 1.2626505405417518e-06,
      "loss": 0.4519,
      "step": 327350
    },
    {
      "epoch": 134.0,
      "grad_norm": 1.6505898237228394,
      "learning_rate": 1.2625097973189268e-06,
      "loss": 0.4482,
      "step": 327360
    },
    {
      "epoch": 134.0,
      "eval_loss": 0.4545886516571045,
      "eval_runtime": 52.2789,
      "eval_samples_per_second": 65.973,
      "eval_steps_per_second": 8.263,
      "step": 327362
    },
    {
      "epoch": 134.0,
      "grad_norm": 2.103847026824951,
      "learning_rate": 1.2623690590490954e-06,
      "loss": 0.4599,
      "step": 327370
    },
    {
      "epoch": 134.01,
      "grad_norm": 1.945349931716919,
      "learning_rate": 1.262228325732903e-06,
      "loss": 0.4439,
      "step": 327380
    },
    {
      "epoch": 134.01,
      "grad_norm": 2.1662468910217285,
      "learning_rate": 1.2620875973709945e-06,
      "loss": 0.4634,
      "step": 327390
    },
    {
      "epoch": 134.02,
      "grad_norm": 2.4385263919830322,
      "learning_rate": 1.261946873964013e-06,
      "loss": 0.4662,
      "step": 327400
    },
    {
      "epoch": 134.02,
      "grad_norm": 1.987863540649414,
      "learning_rate": 1.2618061555126041e-06,
      "loss": 0.4517,
      "step": 327410
    },
    {
      "epoch": 134.02,
      "grad_norm": 1.6426506042480469,
      "learning_rate": 1.2616654420174126e-06,
      "loss": 0.4532,
      "step": 327420
    },
    {
      "epoch": 134.03,
      "grad_norm": 2.056103229522705,
      "learning_rate": 1.2615247334790828e-06,
      "loss": 0.4431,
      "step": 327430
    },
    {
      "epoch": 134.03,
      "grad_norm": 1.8571544885635376,
      "learning_rate": 1.2613840298982586e-06,
      "loss": 0.4748,
      "step": 327440
    },
    {
      "epoch": 134.04,
      "grad_norm": 2.139634847640991,
      "learning_rate": 1.2612433312755856e-06,
      "loss": 0.4666,
      "step": 327450
    },
    {
      "epoch": 134.04,
      "grad_norm": 2.124333620071411,
      "learning_rate": 1.2611026376117086e-06,
      "loss": 0.4732,
      "step": 327460
    },
    {
      "epoch": 134.04,
      "grad_norm": 2.0845096111297607,
      "learning_rate": 1.260961948907271e-06,
      "loss": 0.4621,
      "step": 327470
    },
    {
      "epoch": 134.05,
      "grad_norm": 1.938145399093628,
      "learning_rate": 1.2608212651629183e-06,
      "loss": 0.4507,
      "step": 327480
    },
    {
      "epoch": 134.05,
      "grad_norm": 1.79999840259552,
      "learning_rate": 1.2606805863792943e-06,
      "loss": 0.4557,
      "step": 327490
    },
    {
      "epoch": 134.06,
      "grad_norm": 1.972790002822876,
      "learning_rate": 1.2605399125570434e-06,
      "loss": 0.4638,
      "step": 327500
    },
    {
      "epoch": 134.06,
      "grad_norm": 2.418264865875244,
      "learning_rate": 1.26039924369681e-06,
      "loss": 0.457,
      "step": 327510
    },
    {
      "epoch": 134.06,
      "grad_norm": 1.9349693059921265,
      "learning_rate": 1.260258579799239e-06,
      "loss": 0.4539,
      "step": 327520
    },
    {
      "epoch": 134.07,
      "grad_norm": 3.0655810832977295,
      "learning_rate": 1.260117920864975e-06,
      "loss": 0.4656,
      "step": 327530
    },
    {
      "epoch": 134.07,
      "grad_norm": 2.073991537094116,
      "learning_rate": 1.259977266894661e-06,
      "loss": 0.4669,
      "step": 327540
    },
    {
      "epoch": 134.08,
      "grad_norm": 2.169762134552002,
      "learning_rate": 1.2598366178889422e-06,
      "loss": 0.4636,
      "step": 327550
    },
    {
      "epoch": 134.08,
      "grad_norm": 2.0592517852783203,
      "learning_rate": 1.2596959738484624e-06,
      "loss": 0.4468,
      "step": 327560
    },
    {
      "epoch": 134.09,
      "grad_norm": 1.9273979663848877,
      "learning_rate": 1.2595553347738664e-06,
      "loss": 0.4545,
      "step": 327570
    },
    {
      "epoch": 134.09,
      "grad_norm": 1.874996304512024,
      "learning_rate": 1.2594147006657983e-06,
      "loss": 0.4629,
      "step": 327580
    },
    {
      "epoch": 134.09,
      "grad_norm": 1.877960443496704,
      "learning_rate": 1.2592740715249019e-06,
      "loss": 0.4534,
      "step": 327590
    },
    {
      "epoch": 134.1,
      "grad_norm": 2.494605541229248,
      "learning_rate": 1.2591334473518224e-06,
      "loss": 0.4476,
      "step": 327600
    },
    {
      "epoch": 134.1,
      "grad_norm": 1.7136715650558472,
      "learning_rate": 1.2589928281472026e-06,
      "loss": 0.4571,
      "step": 327610
    },
    {
      "epoch": 134.11,
      "grad_norm": 2.2119035720825195,
      "learning_rate": 1.258852213911688e-06,
      "loss": 0.4411,
      "step": 327620
    },
    {
      "epoch": 134.11,
      "grad_norm": 1.8969297409057617,
      "learning_rate": 1.258711604645921e-06,
      "loss": 0.4571,
      "step": 327630
    },
    {
      "epoch": 134.11,
      "grad_norm": 1.9436999559402466,
      "learning_rate": 1.258571000350548e-06,
      "loss": 0.4597,
      "step": 327640
    },
    {
      "epoch": 134.12,
      "grad_norm": 1.9117200374603271,
      "learning_rate": 1.2584304010262114e-06,
      "loss": 0.4417,
      "step": 327650
    },
    {
      "epoch": 134.12,
      "grad_norm": 2.6175670623779297,
      "learning_rate": 1.2582898066735563e-06,
      "loss": 0.4627,
      "step": 327660
    },
    {
      "epoch": 134.13,
      "grad_norm": 2.121460199356079,
      "learning_rate": 1.2581492172932267e-06,
      "loss": 0.4687,
      "step": 327670
    },
    {
      "epoch": 134.13,
      "grad_norm": 1.7934240102767944,
      "learning_rate": 1.2580086328858658e-06,
      "loss": 0.4416,
      "step": 327680
    },
    {
      "epoch": 134.13,
      "grad_norm": 1.481239676475525,
      "learning_rate": 1.2578680534521177e-06,
      "loss": 0.462,
      "step": 327690
    },
    {
      "epoch": 134.14,
      "grad_norm": 1.5345808267593384,
      "learning_rate": 1.2577274789926267e-06,
      "loss": 0.44,
      "step": 327700
    },
    {
      "epoch": 134.14,
      "grad_norm": 1.955188512802124,
      "learning_rate": 1.2575869095080365e-06,
      "loss": 0.4454,
      "step": 327710
    },
    {
      "epoch": 134.15,
      "grad_norm": 2.0524210929870605,
      "learning_rate": 1.2574463449989913e-06,
      "loss": 0.4708,
      "step": 327720
    },
    {
      "epoch": 134.15,
      "grad_norm": 2.152078866958618,
      "learning_rate": 1.2573057854661349e-06,
      "loss": 0.4583,
      "step": 327730
    },
    {
      "epoch": 134.15,
      "grad_norm": 2.1839511394500732,
      "learning_rate": 1.2571652309101112e-06,
      "loss": 0.4546,
      "step": 327740
    },
    {
      "epoch": 134.16,
      "grad_norm": 1.9413888454437256,
      "learning_rate": 1.2570246813315641e-06,
      "loss": 0.466,
      "step": 327750
    },
    {
      "epoch": 134.16,
      "grad_norm": 2.3473148345947266,
      "learning_rate": 1.2568841367311373e-06,
      "loss": 0.4472,
      "step": 327760
    },
    {
      "epoch": 134.17,
      "grad_norm": 2.353257179260254,
      "learning_rate": 1.2567435971094746e-06,
      "loss": 0.4542,
      "step": 327770
    },
    {
      "epoch": 134.17,
      "grad_norm": 2.336810350418091,
      "learning_rate": 1.25660306246722e-06,
      "loss": 0.44,
      "step": 327780
    },
    {
      "epoch": 134.18,
      "grad_norm": 3.7930524349212646,
      "learning_rate": 1.2564625328050173e-06,
      "loss": 0.461,
      "step": 327790
    },
    {
      "epoch": 134.18,
      "grad_norm": 1.8461487293243408,
      "learning_rate": 1.25632200812351e-06,
      "loss": 0.4659,
      "step": 327800
    },
    {
      "epoch": 134.18,
      "grad_norm": 1.9760559797286987,
      "learning_rate": 1.2561814884233417e-06,
      "loss": 0.4599,
      "step": 327810
    },
    {
      "epoch": 134.19,
      "grad_norm": 2.430863380432129,
      "learning_rate": 1.2560409737051568e-06,
      "loss": 0.4505,
      "step": 327820
    },
    {
      "epoch": 134.19,
      "grad_norm": 1.672921061515808,
      "learning_rate": 1.2559004639695982e-06,
      "loss": 0.4461,
      "step": 327830
    },
    {
      "epoch": 134.2,
      "grad_norm": 1.84255051612854,
      "learning_rate": 1.2557599592173097e-06,
      "loss": 0.447,
      "step": 327840
    },
    {
      "epoch": 134.2,
      "grad_norm": 1.9063974618911743,
      "learning_rate": 1.2556194594489352e-06,
      "loss": 0.4606,
      "step": 327850
    },
    {
      "epoch": 134.2,
      "grad_norm": 1.9261025190353394,
      "learning_rate": 1.2554789646651185e-06,
      "loss": 0.4347,
      "step": 327860
    },
    {
      "epoch": 134.21,
      "grad_norm": 2.2475152015686035,
      "learning_rate": 1.2553384748665025e-06,
      "loss": 0.4259,
      "step": 327870
    },
    {
      "epoch": 134.21,
      "grad_norm": 1.983824372291565,
      "learning_rate": 1.255197990053731e-06,
      "loss": 0.4472,
      "step": 327880
    },
    {
      "epoch": 134.22,
      "grad_norm": 1.795580267906189,
      "learning_rate": 1.2550575102274481e-06,
      "loss": 0.4452,
      "step": 327890
    },
    {
      "epoch": 134.22,
      "grad_norm": 2.186328887939453,
      "learning_rate": 1.2549170353882965e-06,
      "loss": 0.4464,
      "step": 327900
    },
    {
      "epoch": 134.22,
      "grad_norm": 1.9469741582870483,
      "learning_rate": 1.2547765655369203e-06,
      "loss": 0.4463,
      "step": 327910
    },
    {
      "epoch": 134.23,
      "grad_norm": 1.6693803071975708,
      "learning_rate": 1.2546361006739628e-06,
      "loss": 0.4563,
      "step": 327920
    },
    {
      "epoch": 134.23,
      "grad_norm": 1.9014956951141357,
      "learning_rate": 1.2544956408000674e-06,
      "loss": 0.4771,
      "step": 327930
    },
    {
      "epoch": 134.24,
      "grad_norm": 2.173672676086426,
      "learning_rate": 1.2543551859158774e-06,
      "loss": 0.4341,
      "step": 327940
    },
    {
      "epoch": 134.24,
      "grad_norm": 2.0266776084899902,
      "learning_rate": 1.2542147360220367e-06,
      "loss": 0.4689,
      "step": 327950
    },
    {
      "epoch": 134.24,
      "grad_norm": 2.1972625255584717,
      "learning_rate": 1.2540742911191877e-06,
      "loss": 0.4668,
      "step": 327960
    },
    {
      "epoch": 134.25,
      "grad_norm": 1.9308817386627197,
      "learning_rate": 1.2539338512079746e-06,
      "loss": 0.4825,
      "step": 327970
    },
    {
      "epoch": 134.25,
      "grad_norm": 2.0613930225372314,
      "learning_rate": 1.2537934162890417e-06,
      "loss": 0.446,
      "step": 327980
    },
    {
      "epoch": 134.26,
      "grad_norm": 2.0606372356414795,
      "learning_rate": 1.2536529863630302e-06,
      "loss": 0.4451,
      "step": 327990
    },
    {
      "epoch": 134.26,
      "grad_norm": 2.1921699047088623,
      "learning_rate": 1.253512561430583e-06,
      "loss": 0.4604,
      "step": 328000
    },
    {
      "epoch": 134.27,
      "grad_norm": 1.607568621635437,
      "learning_rate": 1.2533721414923455e-06,
      "loss": 0.4415,
      "step": 328010
    },
    {
      "epoch": 134.27,
      "grad_norm": 2.284527540206909,
      "learning_rate": 1.2532317265489606e-06,
      "loss": 0.4761,
      "step": 328020
    },
    {
      "epoch": 134.27,
      "grad_norm": 2.0514090061187744,
      "learning_rate": 1.2530913166010706e-06,
      "loss": 0.462,
      "step": 328030
    },
    {
      "epoch": 134.28,
      "grad_norm": 2.0496020317077637,
      "learning_rate": 1.2529509116493194e-06,
      "loss": 0.4558,
      "step": 328040
    },
    {
      "epoch": 134.28,
      "grad_norm": 1.8666554689407349,
      "learning_rate": 1.2528105116943495e-06,
      "loss": 0.4854,
      "step": 328050
    },
    {
      "epoch": 134.29,
      "grad_norm": 2.479006290435791,
      "learning_rate": 1.252670116736805e-06,
      "loss": 0.4612,
      "step": 328060
    },
    {
      "epoch": 134.29,
      "grad_norm": 2.0612025260925293,
      "learning_rate": 1.2525297267773282e-06,
      "loss": 0.4353,
      "step": 328070
    },
    {
      "epoch": 134.29,
      "grad_norm": 1.8739802837371826,
      "learning_rate": 1.2523893418165623e-06,
      "loss": 0.4612,
      "step": 328080
    },
    {
      "epoch": 134.3,
      "grad_norm": 1.8894157409667969,
      "learning_rate": 1.2522489618551508e-06,
      "loss": 0.4516,
      "step": 328090
    },
    {
      "epoch": 134.3,
      "grad_norm": 1.9451009035110474,
      "learning_rate": 1.2521085868937363e-06,
      "loss": 0.4496,
      "step": 328100
    },
    {
      "epoch": 134.31,
      "grad_norm": 1.8422200679779053,
      "learning_rate": 1.2519682169329631e-06,
      "loss": 0.4536,
      "step": 328110
    },
    {
      "epoch": 134.31,
      "grad_norm": 2.203486204147339,
      "learning_rate": 1.2518278519734722e-06,
      "loss": 0.4619,
      "step": 328120
    },
    {
      "epoch": 134.31,
      "grad_norm": 2.1399929523468018,
      "learning_rate": 1.2516874920159077e-06,
      "loss": 0.4367,
      "step": 328130
    },
    {
      "epoch": 134.32,
      "grad_norm": 1.901123285293579,
      "learning_rate": 1.2515471370609125e-06,
      "loss": 0.4579,
      "step": 328140
    },
    {
      "epoch": 134.32,
      "grad_norm": 2.1363627910614014,
      "learning_rate": 1.2514067871091291e-06,
      "loss": 0.458,
      "step": 328150
    },
    {
      "epoch": 134.33,
      "grad_norm": 2.0178048610687256,
      "learning_rate": 1.251266442161201e-06,
      "loss": 0.4768,
      "step": 328160
    },
    {
      "epoch": 134.33,
      "grad_norm": 1.9936703443527222,
      "learning_rate": 1.251126102217771e-06,
      "loss": 0.4589,
      "step": 328170
    },
    {
      "epoch": 134.33,
      "grad_norm": 4.201420307159424,
      "learning_rate": 1.2509857672794816e-06,
      "loss": 0.4444,
      "step": 328180
    },
    {
      "epoch": 134.34,
      "grad_norm": 2.0766828060150146,
      "learning_rate": 1.2508454373469751e-06,
      "loss": 0.4494,
      "step": 328190
    },
    {
      "epoch": 134.34,
      "grad_norm": 1.7312030792236328,
      "learning_rate": 1.2507051124208962e-06,
      "loss": 0.4681,
      "step": 328200
    },
    {
      "epoch": 134.35,
      "grad_norm": 1.4697126150131226,
      "learning_rate": 1.2505647925018864e-06,
      "loss": 0.4512,
      "step": 328210
    },
    {
      "epoch": 134.35,
      "grad_norm": 2.0015456676483154,
      "learning_rate": 1.250424477590589e-06,
      "loss": 0.4413,
      "step": 328220
    },
    {
      "epoch": 134.36,
      "grad_norm": 1.8986471891403198,
      "learning_rate": 1.2502841676876463e-06,
      "loss": 0.4739,
      "step": 328230
    },
    {
      "epoch": 134.36,
      "grad_norm": 2.036668300628662,
      "learning_rate": 1.2501438627937014e-06,
      "loss": 0.4714,
      "step": 328240
    },
    {
      "epoch": 134.36,
      "grad_norm": 2.1768898963928223,
      "learning_rate": 1.2500035629093975e-06,
      "loss": 0.4605,
      "step": 328250
    },
    {
      "epoch": 134.37,
      "grad_norm": 2.068309783935547,
      "learning_rate": 1.2498632680353762e-06,
      "loss": 0.4433,
      "step": 328260
    },
    {
      "epoch": 134.37,
      "grad_norm": 1.9934134483337402,
      "learning_rate": 1.2497229781722801e-06,
      "loss": 0.4618,
      "step": 328270
    },
    {
      "epoch": 134.38,
      "grad_norm": 1.5637526512145996,
      "learning_rate": 1.2495826933207524e-06,
      "loss": 0.4342,
      "step": 328280
    },
    {
      "epoch": 134.38,
      "grad_norm": 2.0286295413970947,
      "learning_rate": 1.2494424134814358e-06,
      "loss": 0.4409,
      "step": 328290
    },
    {
      "epoch": 134.38,
      "grad_norm": 2.0991852283477783,
      "learning_rate": 1.249302138654973e-06,
      "loss": 0.4542,
      "step": 328300
    },
    {
      "epoch": 134.39,
      "grad_norm": 2.1020469665527344,
      "learning_rate": 1.249161868842006e-06,
      "loss": 0.4705,
      "step": 328310
    },
    {
      "epoch": 134.39,
      "grad_norm": 1.6787968873977661,
      "learning_rate": 1.2490216040431782e-06,
      "loss": 0.4583,
      "step": 328320
    },
    {
      "epoch": 134.4,
      "grad_norm": 1.663995623588562,
      "learning_rate": 1.2488813442591312e-06,
      "loss": 0.4515,
      "step": 328330
    },
    {
      "epoch": 134.4,
      "grad_norm": 2.9531443119049072,
      "learning_rate": 1.248741089490508e-06,
      "loss": 0.4543,
      "step": 328340
    },
    {
      "epoch": 134.4,
      "grad_norm": 1.9995814561843872,
      "learning_rate": 1.2486008397379512e-06,
      "loss": 0.4662,
      "step": 328350
    },
    {
      "epoch": 134.41,
      "grad_norm": 3.1203434467315674,
      "learning_rate": 1.248460595002103e-06,
      "loss": 0.4444,
      "step": 328360
    },
    {
      "epoch": 134.41,
      "grad_norm": 1.9364354610443115,
      "learning_rate": 1.248320355283606e-06,
      "loss": 0.4674,
      "step": 328370
    },
    {
      "epoch": 134.42,
      "grad_norm": 1.9735676050186157,
      "learning_rate": 1.2481801205831024e-06,
      "loss": 0.4662,
      "step": 328380
    },
    {
      "epoch": 134.42,
      "grad_norm": 1.8769054412841797,
      "learning_rate": 1.2480398909012348e-06,
      "loss": 0.4565,
      "step": 328390
    },
    {
      "epoch": 134.42,
      "grad_norm": 2.1390957832336426,
      "learning_rate": 1.2478996662386456e-06,
      "loss": 0.4682,
      "step": 328400
    },
    {
      "epoch": 134.43,
      "grad_norm": 2.077263355255127,
      "learning_rate": 1.247759446595977e-06,
      "loss": 0.4414,
      "step": 328410
    },
    {
      "epoch": 134.43,
      "grad_norm": 1.8441122770309448,
      "learning_rate": 1.2476192319738716e-06,
      "loss": 0.4366,
      "step": 328420
    },
    {
      "epoch": 134.44,
      "grad_norm": 2.3205957412719727,
      "learning_rate": 1.247479022372971e-06,
      "loss": 0.4357,
      "step": 328430
    },
    {
      "epoch": 134.44,
      "grad_norm": 1.9919120073318481,
      "learning_rate": 1.2473388177939183e-06,
      "loss": 0.4576,
      "step": 328440
    },
    {
      "epoch": 134.45,
      "grad_norm": 2.1847357749938965,
      "learning_rate": 1.2471986182373555e-06,
      "loss": 0.4685,
      "step": 328450
    },
    {
      "epoch": 134.45,
      "grad_norm": 1.8707736730575562,
      "learning_rate": 1.2470584237039246e-06,
      "loss": 0.4554,
      "step": 328460
    },
    {
      "epoch": 134.45,
      "grad_norm": 2.089951515197754,
      "learning_rate": 1.246918234194268e-06,
      "loss": 0.4412,
      "step": 328470
    },
    {
      "epoch": 134.46,
      "grad_norm": 1.776017189025879,
      "learning_rate": 1.246778049709028e-06,
      "loss": 0.4494,
      "step": 328480
    },
    {
      "epoch": 134.46,
      "grad_norm": 2.206252336502075,
      "learning_rate": 1.2466378702488464e-06,
      "loss": 0.4687,
      "step": 328490
    },
    {
      "epoch": 134.47,
      "grad_norm": 2.2984559535980225,
      "learning_rate": 1.2464976958143657e-06,
      "loss": 0.4568,
      "step": 328500
    },
    {
      "epoch": 134.47,
      "grad_norm": 2.06888747215271,
      "learning_rate": 1.2463575264062278e-06,
      "loss": 0.4877,
      "step": 328510
    },
    {
      "epoch": 134.47,
      "grad_norm": 1.7451680898666382,
      "learning_rate": 1.2462173620250746e-06,
      "loss": 0.4564,
      "step": 328520
    },
    {
      "epoch": 134.48,
      "grad_norm": 2.4564502239227295,
      "learning_rate": 1.246077202671549e-06,
      "loss": 0.4528,
      "step": 328530
    },
    {
      "epoch": 134.48,
      "grad_norm": 1.7106902599334717,
      "learning_rate": 1.2459370483462924e-06,
      "loss": 0.4324,
      "step": 328540
    },
    {
      "epoch": 134.49,
      "grad_norm": 2.1354634761810303,
      "learning_rate": 1.2457968990499466e-06,
      "loss": 0.4625,
      "step": 328550
    },
    {
      "epoch": 134.49,
      "grad_norm": 1.9963451623916626,
      "learning_rate": 1.2456567547831553e-06,
      "loss": 0.4594,
      "step": 328560
    },
    {
      "epoch": 134.49,
      "grad_norm": 1.9659947156906128,
      "learning_rate": 1.2455166155465567e-06,
      "loss": 0.4612,
      "step": 328570
    },
    {
      "epoch": 134.5,
      "grad_norm": 2.257385730743408,
      "learning_rate": 1.2453764813407964e-06,
      "loss": 0.4436,
      "step": 328580
    },
    {
      "epoch": 134.5,
      "grad_norm": 2.1731486320495605,
      "learning_rate": 1.2452363521665147e-06,
      "loss": 0.4705,
      "step": 328590
    },
    {
      "epoch": 134.51,
      "grad_norm": 1.913733720779419,
      "learning_rate": 1.2450962280243541e-06,
      "loss": 0.4485,
      "step": 328600
    },
    {
      "epoch": 134.51,
      "grad_norm": 1.6070082187652588,
      "learning_rate": 1.2449561089149567e-06,
      "loss": 0.4589,
      "step": 328610
    },
    {
      "epoch": 134.51,
      "grad_norm": 1.904042363166809,
      "learning_rate": 1.2448159948389636e-06,
      "loss": 0.4544,
      "step": 328620
    },
    {
      "epoch": 134.52,
      "grad_norm": 1.7052290439605713,
      "learning_rate": 1.244675885797017e-06,
      "loss": 0.4655,
      "step": 328630
    },
    {
      "epoch": 134.52,
      "grad_norm": 1.9414787292480469,
      "learning_rate": 1.2445357817897585e-06,
      "loss": 0.4468,
      "step": 328640
    },
    {
      "epoch": 134.53,
      "grad_norm": 1.9703335762023926,
      "learning_rate": 1.2443956828178305e-06,
      "loss": 0.4469,
      "step": 328650
    },
    {
      "epoch": 134.53,
      "grad_norm": 1.7705003023147583,
      "learning_rate": 1.2442555888818739e-06,
      "loss": 0.4392,
      "step": 328660
    },
    {
      "epoch": 134.54,
      "grad_norm": 2.6392273902893066,
      "learning_rate": 1.2441154999825312e-06,
      "loss": 0.4763,
      "step": 328670
    },
    {
      "epoch": 134.54,
      "grad_norm": 1.6835825443267822,
      "learning_rate": 1.243975416120444e-06,
      "loss": 0.464,
      "step": 328680
    },
    {
      "epoch": 134.54,
      "grad_norm": 2.053335666656494,
      "learning_rate": 1.2438353372962545e-06,
      "loss": 0.432,
      "step": 328690
    },
    {
      "epoch": 134.55,
      "grad_norm": 1.9677265882492065,
      "learning_rate": 1.243695263510603e-06,
      "loss": 0.4573,
      "step": 328700
    },
    {
      "epoch": 134.55,
      "grad_norm": 1.7427302598953247,
      "learning_rate": 1.2435551947641319e-06,
      "loss": 0.4551,
      "step": 328710
    },
    {
      "epoch": 134.56,
      "grad_norm": 2.0042972564697266,
      "learning_rate": 1.2434151310574826e-06,
      "loss": 0.456,
      "step": 328720
    },
    {
      "epoch": 134.56,
      "grad_norm": 2.0331883430480957,
      "learning_rate": 1.243275072391297e-06,
      "loss": 0.4469,
      "step": 328730
    },
    {
      "epoch": 134.56,
      "grad_norm": 1.6973692178726196,
      "learning_rate": 1.2431350187662165e-06,
      "loss": 0.4534,
      "step": 328740
    },
    {
      "epoch": 134.57,
      "grad_norm": 1.9827102422714233,
      "learning_rate": 1.242994970182883e-06,
      "loss": 0.4659,
      "step": 328750
    },
    {
      "epoch": 134.57,
      "grad_norm": 2.3145346641540527,
      "learning_rate": 1.2428549266419368e-06,
      "loss": 0.4514,
      "step": 328760
    },
    {
      "epoch": 134.58,
      "grad_norm": 2.156743288040161,
      "learning_rate": 1.2427148881440216e-06,
      "loss": 0.4645,
      "step": 328770
    },
    {
      "epoch": 134.58,
      "grad_norm": 2.2970917224884033,
      "learning_rate": 1.2425748546897773e-06,
      "loss": 0.4458,
      "step": 328780
    },
    {
      "epoch": 134.58,
      "grad_norm": 2.2236595153808594,
      "learning_rate": 1.242434826279846e-06,
      "loss": 0.4523,
      "step": 328790
    },
    {
      "epoch": 134.59,
      "grad_norm": 2.1367006301879883,
      "learning_rate": 1.242294802914869e-06,
      "loss": 0.4552,
      "step": 328800
    },
    {
      "epoch": 134.59,
      "grad_norm": 1.7108341455459595,
      "learning_rate": 1.2421547845954875e-06,
      "loss": 0.4475,
      "step": 328810
    },
    {
      "epoch": 134.6,
      "grad_norm": 1.7737905979156494,
      "learning_rate": 1.242014771322343e-06,
      "loss": 0.4531,
      "step": 328820
    },
    {
      "epoch": 134.6,
      "grad_norm": 2.0829215049743652,
      "learning_rate": 1.2418747630960785e-06,
      "loss": 0.4662,
      "step": 328830
    },
    {
      "epoch": 134.6,
      "grad_norm": 2.149371862411499,
      "learning_rate": 1.2417347599173325e-06,
      "loss": 0.4478,
      "step": 328840
    },
    {
      "epoch": 134.61,
      "grad_norm": 1.8145489692687988,
      "learning_rate": 1.2415947617867473e-06,
      "loss": 0.4726,
      "step": 328850
    },
    {
      "epoch": 134.61,
      "grad_norm": 1.825721263885498,
      "learning_rate": 1.2414547687049651e-06,
      "loss": 0.4672,
      "step": 328860
    },
    {
      "epoch": 134.62,
      "grad_norm": 2.175896167755127,
      "learning_rate": 1.2413147806726264e-06,
      "loss": 0.4394,
      "step": 328870
    },
    {
      "epoch": 134.62,
      "grad_norm": 1.9064171314239502,
      "learning_rate": 1.2411747976903725e-06,
      "loss": 0.4439,
      "step": 328880
    },
    {
      "epoch": 134.63,
      "grad_norm": 1.8725594282150269,
      "learning_rate": 1.2410348197588454e-06,
      "loss": 0.4342,
      "step": 328890
    },
    {
      "epoch": 134.63,
      "grad_norm": 3.2032053470611572,
      "learning_rate": 1.2408948468786855e-06,
      "loss": 0.4622,
      "step": 328900
    },
    {
      "epoch": 134.63,
      "grad_norm": 2.1128461360931396,
      "learning_rate": 1.2407548790505342e-06,
      "loss": 0.4377,
      "step": 328910
    },
    {
      "epoch": 134.64,
      "grad_norm": 2.0062062740325928,
      "learning_rate": 1.2406149162750326e-06,
      "loss": 0.4287,
      "step": 328920
    },
    {
      "epoch": 134.64,
      "grad_norm": 1.9697329998016357,
      "learning_rate": 1.240474958552822e-06,
      "loss": 0.4652,
      "step": 328930
    },
    {
      "epoch": 134.65,
      "grad_norm": 2.162187337875366,
      "learning_rate": 1.2403350058845436e-06,
      "loss": 0.4687,
      "step": 328940
    },
    {
      "epoch": 134.65,
      "grad_norm": 2.1007184982299805,
      "learning_rate": 1.2401950582708378e-06,
      "loss": 0.4473,
      "step": 328950
    },
    {
      "epoch": 134.65,
      "grad_norm": 1.9935699701309204,
      "learning_rate": 1.2400551157123468e-06,
      "loss": 0.4555,
      "step": 328960
    },
    {
      "epoch": 134.66,
      "grad_norm": 1.6820316314697266,
      "learning_rate": 1.239915178209712e-06,
      "loss": 0.4561,
      "step": 328970
    },
    {
      "epoch": 134.66,
      "grad_norm": 2.0582568645477295,
      "learning_rate": 1.239775245763573e-06,
      "loss": 0.4557,
      "step": 328980
    },
    {
      "epoch": 134.67,
      "grad_norm": 2.516448974609375,
      "learning_rate": 1.239635318374571e-06,
      "loss": 0.4643,
      "step": 328990
    },
    {
      "epoch": 134.67,
      "grad_norm": 2.011594533920288,
      "learning_rate": 1.2394953960433477e-06,
      "loss": 0.4608,
      "step": 329000
    },
    {
      "epoch": 134.67,
      "grad_norm": 2.376434564590454,
      "learning_rate": 1.2393554787705433e-06,
      "loss": 0.4531,
      "step": 329010
    },
    {
      "epoch": 134.68,
      "grad_norm": 1.9964077472686768,
      "learning_rate": 1.2392155665567995e-06,
      "loss": 0.4671,
      "step": 329020
    },
    {
      "epoch": 134.68,
      "grad_norm": 1.7327308654785156,
      "learning_rate": 1.2390756594027568e-06,
      "loss": 0.434,
      "step": 329030
    },
    {
      "epoch": 134.69,
      "grad_norm": 1.7287781238555908,
      "learning_rate": 1.2389357573090564e-06,
      "loss": 0.429,
      "step": 329040
    },
    {
      "epoch": 134.69,
      "grad_norm": 1.9904533624649048,
      "learning_rate": 1.2387958602763387e-06,
      "loss": 0.4574,
      "step": 329050
    },
    {
      "epoch": 134.7,
      "grad_norm": 2.1886661052703857,
      "learning_rate": 1.2386559683052447e-06,
      "loss": 0.4635,
      "step": 329060
    },
    {
      "epoch": 134.7,
      "grad_norm": 2.240772247314453,
      "learning_rate": 1.2385160813964153e-06,
      "loss": 0.4674,
      "step": 329070
    },
    {
      "epoch": 134.7,
      "grad_norm": 2.1352202892303467,
      "learning_rate": 1.2383761995504916e-06,
      "loss": 0.4762,
      "step": 329080
    },
    {
      "epoch": 134.71,
      "grad_norm": 2.329773426055908,
      "learning_rate": 1.2382363227681141e-06,
      "loss": 0.4329,
      "step": 329090
    },
    {
      "epoch": 134.71,
      "grad_norm": 2.2573304176330566,
      "learning_rate": 1.2380964510499233e-06,
      "loss": 0.4746,
      "step": 329100
    },
    {
      "epoch": 134.72,
      "grad_norm": 2.4593825340270996,
      "learning_rate": 1.2379565843965602e-06,
      "loss": 0.4804,
      "step": 329110
    },
    {
      "epoch": 134.72,
      "grad_norm": 2.0571084022521973,
      "learning_rate": 1.2378167228086657e-06,
      "loss": 0.4563,
      "step": 329120
    },
    {
      "epoch": 134.72,
      "grad_norm": 1.8582395315170288,
      "learning_rate": 1.2376768662868799e-06,
      "loss": 0.4548,
      "step": 329130
    },
    {
      "epoch": 134.73,
      "grad_norm": 1.9575731754302979,
      "learning_rate": 1.237537014831844e-06,
      "loss": 0.4558,
      "step": 329140
    },
    {
      "epoch": 134.73,
      "grad_norm": 1.8447424173355103,
      "learning_rate": 1.2373971684441984e-06,
      "loss": 0.4549,
      "step": 329150
    },
    {
      "epoch": 134.74,
      "grad_norm": 2.192592144012451,
      "learning_rate": 1.237257327124584e-06,
      "loss": 0.4584,
      "step": 329160
    },
    {
      "epoch": 134.74,
      "grad_norm": 2.1774673461914062,
      "learning_rate": 1.2371174908736411e-06,
      "loss": 0.4571,
      "step": 329170
    },
    {
      "epoch": 134.74,
      "grad_norm": 1.6616932153701782,
      "learning_rate": 1.2369776596920104e-06,
      "loss": 0.4545,
      "step": 329180
    },
    {
      "epoch": 134.75,
      "grad_norm": 1.9138840436935425,
      "learning_rate": 1.236837833580332e-06,
      "loss": 0.4595,
      "step": 329190
    },
    {
      "epoch": 134.75,
      "grad_norm": 2.2668943405151367,
      "learning_rate": 1.2366980125392474e-06,
      "loss": 0.4481,
      "step": 329200
    },
    {
      "epoch": 134.76,
      "grad_norm": 1.9942055940628052,
      "learning_rate": 1.236558196569396e-06,
      "loss": 0.4622,
      "step": 329210
    },
    {
      "epoch": 134.76,
      "grad_norm": 2.312803268432617,
      "learning_rate": 1.236418385671419e-06,
      "loss": 0.4411,
      "step": 329220
    },
    {
      "epoch": 134.76,
      "grad_norm": 2.077087879180908,
      "learning_rate": 1.2362785798459566e-06,
      "loss": 0.4738,
      "step": 329230
    },
    {
      "epoch": 134.77,
      "grad_norm": 1.9045482873916626,
      "learning_rate": 1.2361387790936493e-06,
      "loss": 0.4416,
      "step": 329240
    },
    {
      "epoch": 134.77,
      "grad_norm": 1.5539146661758423,
      "learning_rate": 1.2359989834151373e-06,
      "loss": 0.4538,
      "step": 329250
    },
    {
      "epoch": 134.78,
      "grad_norm": 1.71812903881073,
      "learning_rate": 1.2358591928110615e-06,
      "loss": 0.4569,
      "step": 329260
    },
    {
      "epoch": 134.78,
      "grad_norm": 2.6500279903411865,
      "learning_rate": 1.2357194072820616e-06,
      "loss": 0.4367,
      "step": 329270
    },
    {
      "epoch": 134.79,
      "grad_norm": 2.2001843452453613,
      "learning_rate": 1.2355796268287795e-06,
      "loss": 0.4616,
      "step": 329280
    },
    {
      "epoch": 134.79,
      "grad_norm": 1.7657785415649414,
      "learning_rate": 1.235439851451853e-06,
      "loss": 0.4452,
      "step": 329290
    },
    {
      "epoch": 134.79,
      "grad_norm": 2.1261332035064697,
      "learning_rate": 1.2353000811519237e-06,
      "loss": 0.4454,
      "step": 329300
    },
    {
      "epoch": 134.8,
      "grad_norm": 1.7132501602172852,
      "learning_rate": 1.2351603159296318e-06,
      "loss": 0.4534,
      "step": 329310
    },
    {
      "epoch": 134.8,
      "grad_norm": 1.715079426765442,
      "learning_rate": 1.2350205557856167e-06,
      "loss": 0.4584,
      "step": 329320
    },
    {
      "epoch": 134.81,
      "grad_norm": 2.387324810028076,
      "learning_rate": 1.2348808007205206e-06,
      "loss": 0.4438,
      "step": 329330
    },
    {
      "epoch": 134.81,
      "grad_norm": 1.9848076105117798,
      "learning_rate": 1.234741050734982e-06,
      "loss": 0.465,
      "step": 329340
    },
    {
      "epoch": 134.81,
      "grad_norm": 2.3709301948547363,
      "learning_rate": 1.2346013058296423e-06,
      "loss": 0.4453,
      "step": 329350
    },
    {
      "epoch": 134.82,
      "grad_norm": 1.8585894107818604,
      "learning_rate": 1.2344615660051406e-06,
      "loss": 0.4602,
      "step": 329360
    },
    {
      "epoch": 134.82,
      "grad_norm": 2.15053391456604,
      "learning_rate": 1.2343218312621176e-06,
      "loss": 0.4517,
      "step": 329370
    },
    {
      "epoch": 134.83,
      "grad_norm": 2.0603253841400146,
      "learning_rate": 1.234182101601213e-06,
      "loss": 0.4374,
      "step": 329380
    },
    {
      "epoch": 134.83,
      "grad_norm": 2.522956132888794,
      "learning_rate": 1.234042377023067e-06,
      "loss": 0.4424,
      "step": 329390
    },
    {
      "epoch": 134.83,
      "grad_norm": 2.236191987991333,
      "learning_rate": 1.2339026575283202e-06,
      "loss": 0.4601,
      "step": 329400
    },
    {
      "epoch": 134.84,
      "grad_norm": 1.949751377105713,
      "learning_rate": 1.233762943117613e-06,
      "loss": 0.4628,
      "step": 329410
    },
    {
      "epoch": 134.84,
      "grad_norm": 2.129395008087158,
      "learning_rate": 1.2336232337915837e-06,
      "loss": 0.451,
      "step": 329420
    },
    {
      "epoch": 134.85,
      "grad_norm": 1.4564268589019775,
      "learning_rate": 1.2334835295508732e-06,
      "loss": 0.4492,
      "step": 329430
    },
    {
      "epoch": 134.85,
      "grad_norm": 1.9602960348129272,
      "learning_rate": 1.2333438303961214e-06,
      "loss": 0.4489,
      "step": 329440
    },
    {
      "epoch": 134.85,
      "grad_norm": 1.7993566989898682,
      "learning_rate": 1.233204136327968e-06,
      "loss": 0.4473,
      "step": 329450
    },
    {
      "epoch": 134.86,
      "grad_norm": 1.9581185579299927,
      "learning_rate": 1.2330644473470535e-06,
      "loss": 0.4253,
      "step": 329460
    },
    {
      "epoch": 134.86,
      "grad_norm": 2.033782720565796,
      "learning_rate": 1.2329247634540175e-06,
      "loss": 0.4559,
      "step": 329470
    },
    {
      "epoch": 134.87,
      "grad_norm": 2.1689915657043457,
      "learning_rate": 1.2327850846494998e-06,
      "loss": 0.4615,
      "step": 329480
    },
    {
      "epoch": 134.87,
      "grad_norm": 2.2225372791290283,
      "learning_rate": 1.2326454109341406e-06,
      "loss": 0.4573,
      "step": 329490
    },
    {
      "epoch": 134.88,
      "grad_norm": 2.1894631385803223,
      "learning_rate": 1.2325057423085791e-06,
      "loss": 0.467,
      "step": 329500
    },
    {
      "epoch": 134.88,
      "grad_norm": 2.02498197555542,
      "learning_rate": 1.2323660787734546e-06,
      "loss": 0.4594,
      "step": 329510
    },
    {
      "epoch": 134.88,
      "grad_norm": 1.9246230125427246,
      "learning_rate": 1.2322264203294087e-06,
      "loss": 0.4512,
      "step": 329520
    },
    {
      "epoch": 134.89,
      "grad_norm": 2.2711293697357178,
      "learning_rate": 1.2320867669770801e-06,
      "loss": 0.4648,
      "step": 329530
    },
    {
      "epoch": 134.89,
      "grad_norm": 1.935734748840332,
      "learning_rate": 1.2319471187171087e-06,
      "loss": 0.4415,
      "step": 329540
    },
    {
      "epoch": 134.9,
      "grad_norm": 2.2505829334259033,
      "learning_rate": 1.231807475550135e-06,
      "loss": 0.4706,
      "step": 329550
    },
    {
      "epoch": 134.9,
      "grad_norm": 2.03275203704834,
      "learning_rate": 1.231667837476797e-06,
      "loss": 0.4509,
      "step": 329560
    },
    {
      "epoch": 134.9,
      "grad_norm": 2.144735097885132,
      "learning_rate": 1.231528204497735e-06,
      "loss": 0.4511,
      "step": 329570
    },
    {
      "epoch": 134.91,
      "grad_norm": 2.414816379547119,
      "learning_rate": 1.231388576613589e-06,
      "loss": 0.4688,
      "step": 329580
    },
    {
      "epoch": 134.91,
      "grad_norm": 1.8388432264328003,
      "learning_rate": 1.231248953824998e-06,
      "loss": 0.4351,
      "step": 329590
    },
    {
      "epoch": 134.92,
      "grad_norm": 1.827016830444336,
      "learning_rate": 1.2311093361326023e-06,
      "loss": 0.453,
      "step": 329600
    },
    {
      "epoch": 134.92,
      "grad_norm": 2.1612610816955566,
      "learning_rate": 1.2309697235370412e-06,
      "loss": 0.4574,
      "step": 329610
    },
    {
      "epoch": 134.92,
      "grad_norm": 1.8339285850524902,
      "learning_rate": 1.230830116038954e-06,
      "loss": 0.4557,
      "step": 329620
    },
    {
      "epoch": 134.93,
      "grad_norm": 2.764155626296997,
      "learning_rate": 1.2306905136389806e-06,
      "loss": 0.4507,
      "step": 329630
    },
    {
      "epoch": 134.93,
      "grad_norm": 1.637237548828125,
      "learning_rate": 1.2305509163377605e-06,
      "loss": 0.466,
      "step": 329640
    },
    {
      "epoch": 134.94,
      "grad_norm": 1.9775924682617188,
      "learning_rate": 1.2304113241359326e-06,
      "loss": 0.4516,
      "step": 329650
    },
    {
      "epoch": 134.94,
      "grad_norm": 2.1884806156158447,
      "learning_rate": 1.230271737034137e-06,
      "loss": 0.4575,
      "step": 329660
    },
    {
      "epoch": 134.94,
      "grad_norm": 2.2628211975097656,
      "learning_rate": 1.2301321550330126e-06,
      "loss": 0.445,
      "step": 329670
    },
    {
      "epoch": 134.95,
      "grad_norm": 1.8299072980880737,
      "learning_rate": 1.2299925781331994e-06,
      "loss": 0.4471,
      "step": 329680
    },
    {
      "epoch": 134.95,
      "grad_norm": 2.0594351291656494,
      "learning_rate": 1.2298530063353366e-06,
      "loss": 0.4336,
      "step": 329690
    },
    {
      "epoch": 134.96,
      "grad_norm": 1.823537826538086,
      "learning_rate": 1.229713439640063e-06,
      "loss": 0.4386,
      "step": 329700
    },
    {
      "epoch": 134.96,
      "grad_norm": 2.3753061294555664,
      "learning_rate": 1.2295738780480187e-06,
      "loss": 0.4439,
      "step": 329710
    },
    {
      "epoch": 134.97,
      "grad_norm": 2.194068431854248,
      "learning_rate": 1.2294343215598423e-06,
      "loss": 0.4611,
      "step": 329720
    },
    {
      "epoch": 134.97,
      "grad_norm": 1.7429981231689453,
      "learning_rate": 1.2292947701761737e-06,
      "loss": 0.4658,
      "step": 329730
    },
    {
      "epoch": 134.97,
      "grad_norm": 1.7228562831878662,
      "learning_rate": 1.229155223897652e-06,
      "loss": 0.4295,
      "step": 329740
    },
    {
      "epoch": 134.98,
      "grad_norm": 1.9011245965957642,
      "learning_rate": 1.2290156827249161e-06,
      "loss": 0.4555,
      "step": 329750
    },
    {
      "epoch": 134.98,
      "grad_norm": 2.6021056175231934,
      "learning_rate": 1.2288761466586057e-06,
      "loss": 0.4479,
      "step": 329760
    },
    {
      "epoch": 134.99,
      "grad_norm": 1.9838496446609497,
      "learning_rate": 1.2287366156993597e-06,
      "loss": 0.4508,
      "step": 329770
    },
    {
      "epoch": 134.99,
      "grad_norm": 1.8847160339355469,
      "learning_rate": 1.228597089847817e-06,
      "loss": 0.432,
      "step": 329780
    },
    {
      "epoch": 134.99,
      "grad_norm": 1.9553385972976685,
      "learning_rate": 1.2284575691046176e-06,
      "loss": 0.4421,
      "step": 329790
    },
    {
      "epoch": 135.0,
      "grad_norm": 2.018740177154541,
      "learning_rate": 1.2283180534703999e-06,
      "loss": 0.4648,
      "step": 329800
    },
    {
      "epoch": 135.0,
      "eval_loss": 0.45739197731018066,
      "eval_runtime": 52.1928,
      "eval_samples_per_second": 66.082,
      "eval_steps_per_second": 8.277,
      "step": 329805
    },
    {
      "epoch": 135.0,
      "grad_norm": 1.5861973762512207,
      "learning_rate": 1.228178542945803e-06,
      "loss": 0.4379,
      "step": 329810
    },
    {
      "epoch": 135.01,
      "grad_norm": 2.2599573135375977,
      "learning_rate": 1.2280390375314664e-06,
      "loss": 0.4565,
      "step": 329820
    },
    {
      "epoch": 135.01,
      "grad_norm": 1.9164960384368896,
      "learning_rate": 1.227899537228029e-06,
      "loss": 0.451,
      "step": 329830
    },
    {
      "epoch": 135.01,
      "grad_norm": 1.7604774236679077,
      "learning_rate": 1.2277600420361294e-06,
      "loss": 0.4689,
      "step": 329840
    },
    {
      "epoch": 135.02,
      "grad_norm": 1.8595548868179321,
      "learning_rate": 1.2276205519564072e-06,
      "loss": 0.4448,
      "step": 329850
    },
    {
      "epoch": 135.02,
      "grad_norm": 1.9712235927581787,
      "learning_rate": 1.227481066989502e-06,
      "loss": 0.4533,
      "step": 329860
    },
    {
      "epoch": 135.03,
      "grad_norm": 2.248464345932007,
      "learning_rate": 1.2273415871360511e-06,
      "loss": 0.4643,
      "step": 329870
    },
    {
      "epoch": 135.03,
      "grad_norm": 2.0741195678710938,
      "learning_rate": 1.2272021123966942e-06,
      "loss": 0.4527,
      "step": 329880
    },
    {
      "epoch": 135.03,
      "grad_norm": 1.9520856142044067,
      "learning_rate": 1.2270626427720693e-06,
      "loss": 0.4539,
      "step": 329890
    },
    {
      "epoch": 135.04,
      "grad_norm": 1.9622067213058472,
      "learning_rate": 1.2269231782628172e-06,
      "loss": 0.4445,
      "step": 329900
    },
    {
      "epoch": 135.04,
      "grad_norm": 3.6335062980651855,
      "learning_rate": 1.226783718869576e-06,
      "loss": 0.4605,
      "step": 329910
    },
    {
      "epoch": 135.05,
      "grad_norm": 2.0216832160949707,
      "learning_rate": 1.226644264592984e-06,
      "loss": 0.4639,
      "step": 329920
    },
    {
      "epoch": 135.05,
      "grad_norm": 2.218735456466675,
      "learning_rate": 1.2265048154336805e-06,
      "loss": 0.4534,
      "step": 329930
    },
    {
      "epoch": 135.06,
      "grad_norm": 2.3221662044525146,
      "learning_rate": 1.2263653713923042e-06,
      "loss": 0.4586,
      "step": 329940
    },
    {
      "epoch": 135.06,
      "grad_norm": 2.4718799591064453,
      "learning_rate": 1.2262259324694937e-06,
      "loss": 0.4314,
      "step": 329950
    },
    {
      "epoch": 135.06,
      "grad_norm": 1.7527109384536743,
      "learning_rate": 1.226086498665888e-06,
      "loss": 0.4288,
      "step": 329960
    },
    {
      "epoch": 135.07,
      "grad_norm": 1.7027993202209473,
      "learning_rate": 1.2259470699821258e-06,
      "loss": 0.4333,
      "step": 329970
    },
    {
      "epoch": 135.07,
      "grad_norm": 2.2525808811187744,
      "learning_rate": 1.2258076464188455e-06,
      "loss": 0.4373,
      "step": 329980
    },
    {
      "epoch": 135.08,
      "grad_norm": 2.096454381942749,
      "learning_rate": 1.225668227976687e-06,
      "loss": 0.4534,
      "step": 329990
    },
    {
      "epoch": 135.08,
      "grad_norm": 2.0219850540161133,
      "learning_rate": 1.225528814656287e-06,
      "loss": 0.4422,
      "step": 330000
    },
    {
      "epoch": 135.08,
      "grad_norm": 1.9044536352157593,
      "learning_rate": 1.2253894064582851e-06,
      "loss": 0.4575,
      "step": 330010
    },
    {
      "epoch": 135.09,
      "grad_norm": 2.338524103164673,
      "learning_rate": 1.22525000338332e-06,
      "loss": 0.4682,
      "step": 330020
    },
    {
      "epoch": 135.09,
      "grad_norm": 2.458709955215454,
      "learning_rate": 1.22511060543203e-06,
      "loss": 0.4575,
      "step": 330030
    },
    {
      "epoch": 135.1,
      "grad_norm": 1.9980075359344482,
      "learning_rate": 1.2249712126050537e-06,
      "loss": 0.4469,
      "step": 330040
    },
    {
      "epoch": 135.1,
      "grad_norm": 1.903943657875061,
      "learning_rate": 1.22483182490303e-06,
      "loss": 0.4553,
      "step": 330050
    },
    {
      "epoch": 135.1,
      "grad_norm": 2.0540120601654053,
      "learning_rate": 1.2246924423265973e-06,
      "loss": 0.4526,
      "step": 330060
    },
    {
      "epoch": 135.11,
      "grad_norm": 2.1113953590393066,
      "learning_rate": 1.2245530648763937e-06,
      "loss": 0.4627,
      "step": 330070
    },
    {
      "epoch": 135.11,
      "grad_norm": 1.6465836763381958,
      "learning_rate": 1.224413692553057e-06,
      "loss": 0.442,
      "step": 330080
    },
    {
      "epoch": 135.12,
      "grad_norm": 1.6213384866714478,
      "learning_rate": 1.2242743253572275e-06,
      "loss": 0.4545,
      "step": 330090
    },
    {
      "epoch": 135.12,
      "grad_norm": 1.9948536157608032,
      "learning_rate": 1.2241349632895427e-06,
      "loss": 0.4596,
      "step": 330100
    },
    {
      "epoch": 135.12,
      "grad_norm": 1.9915895462036133,
      "learning_rate": 1.2239956063506411e-06,
      "loss": 0.4745,
      "step": 330110
    },
    {
      "epoch": 135.13,
      "grad_norm": 2.1029398441314697,
      "learning_rate": 1.2238562545411608e-06,
      "loss": 0.462,
      "step": 330120
    },
    {
      "epoch": 135.13,
      "grad_norm": 1.6994389295578003,
      "learning_rate": 1.2237169078617415e-06,
      "loss": 0.4565,
      "step": 330130
    },
    {
      "epoch": 135.14,
      "grad_norm": 1.7526217699050903,
      "learning_rate": 1.2235775663130193e-06,
      "loss": 0.4516,
      "step": 330140
    },
    {
      "epoch": 135.14,
      "grad_norm": 1.6596550941467285,
      "learning_rate": 1.2234382298956335e-06,
      "loss": 0.4486,
      "step": 330150
    },
    {
      "epoch": 135.15,
      "grad_norm": 1.9219752550125122,
      "learning_rate": 1.2232988986102225e-06,
      "loss": 0.4633,
      "step": 330160
    },
    {
      "epoch": 135.15,
      "grad_norm": 1.7254996299743652,
      "learning_rate": 1.2231595724574246e-06,
      "loss": 0.4465,
      "step": 330170
    },
    {
      "epoch": 135.15,
      "grad_norm": 1.724682092666626,
      "learning_rate": 1.2230202514378773e-06,
      "loss": 0.4487,
      "step": 330180
    },
    {
      "epoch": 135.16,
      "grad_norm": 2.26261305809021,
      "learning_rate": 1.22288093555222e-06,
      "loss": 0.4666,
      "step": 330190
    },
    {
      "epoch": 135.16,
      "grad_norm": 1.905497431755066,
      "learning_rate": 1.2227416248010898e-06,
      "loss": 0.4581,
      "step": 330200
    },
    {
      "epoch": 135.17,
      "grad_norm": 1.7018505334854126,
      "learning_rate": 1.2226023191851256e-06,
      "loss": 0.4545,
      "step": 330210
    },
    {
      "epoch": 135.17,
      "grad_norm": 1.9555376768112183,
      "learning_rate": 1.2224630187049654e-06,
      "loss": 0.4498,
      "step": 330220
    },
    {
      "epoch": 135.17,
      "grad_norm": 1.9351989030838013,
      "learning_rate": 1.2223237233612476e-06,
      "loss": 0.4575,
      "step": 330230
    },
    {
      "epoch": 135.18,
      "grad_norm": 2.41501784324646,
      "learning_rate": 1.2221844331546092e-06,
      "loss": 0.4442,
      "step": 330240
    },
    {
      "epoch": 135.18,
      "grad_norm": 2.1424787044525146,
      "learning_rate": 1.2220451480856892e-06,
      "loss": 0.4606,
      "step": 330250
    },
    {
      "epoch": 135.19,
      "grad_norm": 2.4881389141082764,
      "learning_rate": 1.2219058681551254e-06,
      "loss": 0.4555,
      "step": 330260
    },
    {
      "epoch": 135.19,
      "grad_norm": 2.177943706512451,
      "learning_rate": 1.221766593363556e-06,
      "loss": 0.4584,
      "step": 330270
    },
    {
      "epoch": 135.19,
      "grad_norm": 2.4030418395996094,
      "learning_rate": 1.2216273237116184e-06,
      "loss": 0.4389,
      "step": 330280
    },
    {
      "epoch": 135.2,
      "grad_norm": 1.859053611755371,
      "learning_rate": 1.2214880591999515e-06,
      "loss": 0.4638,
      "step": 330290
    },
    {
      "epoch": 135.2,
      "grad_norm": 2.00294828414917,
      "learning_rate": 1.2213487998291925e-06,
      "loss": 0.467,
      "step": 330300
    },
    {
      "epoch": 135.21,
      "grad_norm": 2.1813056468963623,
      "learning_rate": 1.2212095455999798e-06,
      "loss": 0.4534,
      "step": 330310
    },
    {
      "epoch": 135.21,
      "grad_norm": 2.2740750312805176,
      "learning_rate": 1.2210702965129505e-06,
      "loss": 0.4459,
      "step": 330320
    },
    {
      "epoch": 135.21,
      "grad_norm": 1.7403355836868286,
      "learning_rate": 1.2209310525687437e-06,
      "loss": 0.4475,
      "step": 330330
    },
    {
      "epoch": 135.22,
      "grad_norm": 2.390998363494873,
      "learning_rate": 1.2207918137679962e-06,
      "loss": 0.4554,
      "step": 330340
    },
    {
      "epoch": 135.22,
      "grad_norm": 2.049905300140381,
      "learning_rate": 1.2206525801113465e-06,
      "loss": 0.453,
      "step": 330350
    },
    {
      "epoch": 135.23,
      "grad_norm": 2.8049235343933105,
      "learning_rate": 1.2205133515994317e-06,
      "loss": 0.4506,
      "step": 330360
    },
    {
      "epoch": 135.23,
      "grad_norm": 2.013423204421997,
      "learning_rate": 1.2203741282328906e-06,
      "loss": 0.4618,
      "step": 330370
    },
    {
      "epoch": 135.24,
      "grad_norm": 1.5608539581298828,
      "learning_rate": 1.2202349100123603e-06,
      "loss": 0.4444,
      "step": 330380
    },
    {
      "epoch": 135.24,
      "grad_norm": 1.9927692413330078,
      "learning_rate": 1.2200956969384782e-06,
      "loss": 0.4458,
      "step": 330390
    },
    {
      "epoch": 135.24,
      "grad_norm": 1.7125320434570312,
      "learning_rate": 1.2199564890118826e-06,
      "loss": 0.4655,
      "step": 330400
    },
    {
      "epoch": 135.25,
      "grad_norm": 2.212684392929077,
      "learning_rate": 1.2198172862332114e-06,
      "loss": 0.4507,
      "step": 330410
    },
    {
      "epoch": 135.25,
      "grad_norm": 1.6285898685455322,
      "learning_rate": 1.2196780886031015e-06,
      "loss": 0.4626,
      "step": 330420
    },
    {
      "epoch": 135.26,
      "grad_norm": 2.830371141433716,
      "learning_rate": 1.2195388961221913e-06,
      "loss": 0.4539,
      "step": 330430
    },
    {
      "epoch": 135.26,
      "grad_norm": 2.0113182067871094,
      "learning_rate": 1.2193997087911187e-06,
      "loss": 0.4477,
      "step": 330440
    },
    {
      "epoch": 135.26,
      "grad_norm": 2.239936113357544,
      "learning_rate": 1.2192605266105198e-06,
      "loss": 0.4573,
      "step": 330450
    },
    {
      "epoch": 135.27,
      "grad_norm": 1.7589340209960938,
      "learning_rate": 1.2191213495810321e-06,
      "loss": 0.4573,
      "step": 330460
    },
    {
      "epoch": 135.27,
      "grad_norm": 1.6785051822662354,
      "learning_rate": 1.218982177703295e-06,
      "loss": 0.4626,
      "step": 330470
    },
    {
      "epoch": 135.28,
      "grad_norm": 2.1865811347961426,
      "learning_rate": 1.2188430109779451e-06,
      "loss": 0.4698,
      "step": 330480
    },
    {
      "epoch": 135.28,
      "grad_norm": 1.9992780685424805,
      "learning_rate": 1.2187038494056201e-06,
      "loss": 0.4627,
      "step": 330490
    },
    {
      "epoch": 135.28,
      "grad_norm": 2.312365770339966,
      "learning_rate": 1.218564692986957e-06,
      "loss": 0.4814,
      "step": 330500
    },
    {
      "epoch": 135.29,
      "grad_norm": 2.7441930770874023,
      "learning_rate": 1.2184255417225937e-06,
      "loss": 0.4628,
      "step": 330510
    },
    {
      "epoch": 135.29,
      "grad_norm": 2.093881845474243,
      "learning_rate": 1.2182863956131673e-06,
      "loss": 0.464,
      "step": 330520
    },
    {
      "epoch": 135.3,
      "grad_norm": 1.8304126262664795,
      "learning_rate": 1.2181472546593153e-06,
      "loss": 0.4664,
      "step": 330530
    },
    {
      "epoch": 135.3,
      "grad_norm": 2.8675029277801514,
      "learning_rate": 1.2180081188616754e-06,
      "loss": 0.478,
      "step": 330540
    },
    {
      "epoch": 135.3,
      "grad_norm": 2.056215286254883,
      "learning_rate": 1.2178689882208848e-06,
      "loss": 0.4464,
      "step": 330550
    },
    {
      "epoch": 135.31,
      "grad_norm": 1.8996018171310425,
      "learning_rate": 1.2177298627375805e-06,
      "loss": 0.4611,
      "step": 330560
    },
    {
      "epoch": 135.31,
      "grad_norm": 2.053924560546875,
      "learning_rate": 1.2175907424124e-06,
      "loss": 0.4314,
      "step": 330570
    },
    {
      "epoch": 135.32,
      "grad_norm": 1.933092713356018,
      "learning_rate": 1.2174516272459817e-06,
      "loss": 0.4512,
      "step": 330580
    },
    {
      "epoch": 135.32,
      "grad_norm": 1.9219683408737183,
      "learning_rate": 1.217312517238961e-06,
      "loss": 0.4498,
      "step": 330590
    },
    {
      "epoch": 135.33,
      "grad_norm": 2.167039155960083,
      "learning_rate": 1.2171734123919758e-06,
      "loss": 0.4502,
      "step": 330600
    },
    {
      "epoch": 135.33,
      "grad_norm": 1.868690848350525,
      "learning_rate": 1.2170343127056637e-06,
      "loss": 0.4478,
      "step": 330610
    },
    {
      "epoch": 135.33,
      "grad_norm": 1.6676102876663208,
      "learning_rate": 1.2168952181806615e-06,
      "loss": 0.4553,
      "step": 330620
    },
    {
      "epoch": 135.34,
      "grad_norm": 2.220094919204712,
      "learning_rate": 1.2167561288176064e-06,
      "loss": 0.4623,
      "step": 330630
    },
    {
      "epoch": 135.34,
      "grad_norm": 1.8338161706924438,
      "learning_rate": 1.2166170446171352e-06,
      "loss": 0.4605,
      "step": 330640
    },
    {
      "epoch": 135.35,
      "grad_norm": 1.9389978647232056,
      "learning_rate": 1.216477965579886e-06,
      "loss": 0.4556,
      "step": 330650
    },
    {
      "epoch": 135.35,
      "grad_norm": 1.7616533041000366,
      "learning_rate": 1.2163388917064956e-06,
      "loss": 0.4398,
      "step": 330660
    },
    {
      "epoch": 135.35,
      "grad_norm": 2.415513515472412,
      "learning_rate": 1.2161998229976008e-06,
      "loss": 0.4619,
      "step": 330670
    },
    {
      "epoch": 135.36,
      "grad_norm": 2.020557403564453,
      "learning_rate": 1.2160607594538387e-06,
      "loss": 0.4479,
      "step": 330680
    },
    {
      "epoch": 135.36,
      "grad_norm": 1.9830131530761719,
      "learning_rate": 1.2159217010758464e-06,
      "loss": 0.4386,
      "step": 330690
    },
    {
      "epoch": 135.37,
      "grad_norm": 2.0323197841644287,
      "learning_rate": 1.215782647864261e-06,
      "loss": 0.4529,
      "step": 330700
    },
    {
      "epoch": 135.37,
      "grad_norm": 1.956207513809204,
      "learning_rate": 1.21564359981972e-06,
      "loss": 0.449,
      "step": 330710
    },
    {
      "epoch": 135.37,
      "grad_norm": 1.7711204290390015,
      "learning_rate": 1.2155045569428593e-06,
      "loss": 0.4615,
      "step": 330720
    },
    {
      "epoch": 135.38,
      "grad_norm": 1.8035691976547241,
      "learning_rate": 1.215365519234316e-06,
      "loss": 0.4467,
      "step": 330730
    },
    {
      "epoch": 135.38,
      "grad_norm": 2.174257755279541,
      "learning_rate": 1.2152264866947273e-06,
      "loss": 0.4458,
      "step": 330740
    },
    {
      "epoch": 135.39,
      "grad_norm": 1.9987636804580688,
      "learning_rate": 1.2150874593247297e-06,
      "loss": 0.4399,
      "step": 330750
    },
    {
      "epoch": 135.39,
      "grad_norm": 2.532160520553589,
      "learning_rate": 1.214948437124961e-06,
      "loss": 0.4396,
      "step": 330760
    },
    {
      "epoch": 135.4,
      "grad_norm": 2.5357613563537598,
      "learning_rate": 1.2148094200960568e-06,
      "loss": 0.4543,
      "step": 330770
    },
    {
      "epoch": 135.4,
      "grad_norm": 1.9618738889694214,
      "learning_rate": 1.214670408238655e-06,
      "loss": 0.4509,
      "step": 330780
    },
    {
      "epoch": 135.4,
      "grad_norm": 1.9279639720916748,
      "learning_rate": 1.2145314015533917e-06,
      "loss": 0.4444,
      "step": 330790
    },
    {
      "epoch": 135.41,
      "grad_norm": 1.7289892435073853,
      "learning_rate": 1.2143924000409038e-06,
      "loss": 0.4532,
      "step": 330800
    },
    {
      "epoch": 135.41,
      "grad_norm": 2.2618677616119385,
      "learning_rate": 1.2142534037018282e-06,
      "loss": 0.4243,
      "step": 330810
    },
    {
      "epoch": 135.42,
      "grad_norm": 2.2651641368865967,
      "learning_rate": 1.214114412536802e-06,
      "loss": 0.4416,
      "step": 330820
    },
    {
      "epoch": 135.42,
      "grad_norm": 1.5956838130950928,
      "learning_rate": 1.2139754265464603e-06,
      "loss": 0.434,
      "step": 330830
    },
    {
      "epoch": 135.42,
      "grad_norm": 1.8090150356292725,
      "learning_rate": 1.2138364457314418e-06,
      "loss": 0.4607,
      "step": 330840
    },
    {
      "epoch": 135.43,
      "grad_norm": 1.9394216537475586,
      "learning_rate": 1.2136974700923836e-06,
      "loss": 0.4414,
      "step": 330850
    },
    {
      "epoch": 135.43,
      "grad_norm": 1.6655185222625732,
      "learning_rate": 1.2135584996299196e-06,
      "loss": 0.4655,
      "step": 330860
    },
    {
      "epoch": 135.44,
      "grad_norm": 1.8896600008010864,
      "learning_rate": 1.2134195343446882e-06,
      "loss": 0.4562,
      "step": 330870
    },
    {
      "epoch": 135.44,
      "grad_norm": 2.1507012844085693,
      "learning_rate": 1.2132805742373252e-06,
      "loss": 0.4411,
      "step": 330880
    },
    {
      "epoch": 135.44,
      "grad_norm": 1.881531000137329,
      "learning_rate": 1.213141619308468e-06,
      "loss": 0.4569,
      "step": 330890
    },
    {
      "epoch": 135.45,
      "grad_norm": 2.037283182144165,
      "learning_rate": 1.2130026695587524e-06,
      "loss": 0.4596,
      "step": 330900
    },
    {
      "epoch": 135.45,
      "grad_norm": 3.3157448768615723,
      "learning_rate": 1.2128637249888154e-06,
      "loss": 0.4383,
      "step": 330910
    },
    {
      "epoch": 135.46,
      "grad_norm": 2.063223123550415,
      "learning_rate": 1.2127247855992932e-06,
      "loss": 0.4553,
      "step": 330920
    },
    {
      "epoch": 135.46,
      "grad_norm": 1.8291977643966675,
      "learning_rate": 1.2125858513908223e-06,
      "loss": 0.4728,
      "step": 330930
    },
    {
      "epoch": 135.46,
      "grad_norm": 4.885100364685059,
      "learning_rate": 1.2124469223640393e-06,
      "loss": 0.4643,
      "step": 330940
    },
    {
      "epoch": 135.47,
      "grad_norm": 1.7996629476547241,
      "learning_rate": 1.2123079985195802e-06,
      "loss": 0.4406,
      "step": 330950
    },
    {
      "epoch": 135.47,
      "grad_norm": 2.222602367401123,
      "learning_rate": 1.212169079858082e-06,
      "loss": 0.4525,
      "step": 330960
    },
    {
      "epoch": 135.48,
      "grad_norm": 2.2358691692352295,
      "learning_rate": 1.2120301663801807e-06,
      "loss": 0.454,
      "step": 330970
    },
    {
      "epoch": 135.48,
      "grad_norm": 1.9535497426986694,
      "learning_rate": 1.2118912580865126e-06,
      "loss": 0.4417,
      "step": 330980
    },
    {
      "epoch": 135.49,
      "grad_norm": 2.335196018218994,
      "learning_rate": 1.2117523549777142e-06,
      "loss": 0.4534,
      "step": 330990
    },
    {
      "epoch": 135.49,
      "grad_norm": 2.5365688800811768,
      "learning_rate": 1.2116134570544216e-06,
      "loss": 0.46,
      "step": 331000
    },
    {
      "epoch": 135.49,
      "grad_norm": 1.9236044883728027,
      "learning_rate": 1.2114745643172712e-06,
      "loss": 0.4681,
      "step": 331010
    },
    {
      "epoch": 135.5,
      "grad_norm": 1.9036821126937866,
      "learning_rate": 1.2113356767668994e-06,
      "loss": 0.4429,
      "step": 331020
    },
    {
      "epoch": 135.5,
      "grad_norm": 2.2689623832702637,
      "learning_rate": 1.211196794403942e-06,
      "loss": 0.4446,
      "step": 331030
    },
    {
      "epoch": 135.51,
      "grad_norm": 1.9550377130508423,
      "learning_rate": 1.2110579172290352e-06,
      "loss": 0.4671,
      "step": 331040
    },
    {
      "epoch": 135.51,
      "grad_norm": 3.1539273262023926,
      "learning_rate": 1.2109190452428158e-06,
      "loss": 0.4653,
      "step": 331050
    },
    {
      "epoch": 135.51,
      "grad_norm": 1.870300531387329,
      "learning_rate": 1.2107801784459197e-06,
      "loss": 0.4504,
      "step": 331060
    },
    {
      "epoch": 135.52,
      "grad_norm": 2.284294605255127,
      "learning_rate": 1.2106413168389826e-06,
      "loss": 0.4523,
      "step": 331070
    },
    {
      "epoch": 135.52,
      "grad_norm": 1.9754170179367065,
      "learning_rate": 1.210502460422641e-06,
      "loss": 0.4416,
      "step": 331080
    },
    {
      "epoch": 135.53,
      "grad_norm": 1.477150559425354,
      "learning_rate": 1.2103636091975307e-06,
      "loss": 0.4498,
      "step": 331090
    },
    {
      "epoch": 135.53,
      "grad_norm": 1.8752809762954712,
      "learning_rate": 1.2102247631642882e-06,
      "loss": 0.4571,
      "step": 331100
    },
    {
      "epoch": 135.53,
      "grad_norm": 1.813093662261963,
      "learning_rate": 1.2100859223235493e-06,
      "loss": 0.4663,
      "step": 331110
    },
    {
      "epoch": 135.54,
      "grad_norm": 1.9210857152938843,
      "learning_rate": 1.2099470866759494e-06,
      "loss": 0.4594,
      "step": 331120
    },
    {
      "epoch": 135.54,
      "grad_norm": 2.0039608478546143,
      "learning_rate": 1.2098082562221258e-06,
      "loss": 0.4474,
      "step": 331130
    },
    {
      "epoch": 135.55,
      "grad_norm": 2.493835210800171,
      "learning_rate": 1.2096694309627132e-06,
      "loss": 0.4685,
      "step": 331140
    },
    {
      "epoch": 135.55,
      "grad_norm": 2.1606600284576416,
      "learning_rate": 1.209530610898348e-06,
      "loss": 0.4373,
      "step": 331150
    },
    {
      "epoch": 135.55,
      "grad_norm": 2.3573436737060547,
      "learning_rate": 1.2093917960296674e-06,
      "loss": 0.4383,
      "step": 331160
    },
    {
      "epoch": 135.56,
      "grad_norm": 2.246760606765747,
      "learning_rate": 1.2092529863573053e-06,
      "loss": 0.4697,
      "step": 331170
    },
    {
      "epoch": 135.56,
      "grad_norm": 1.8808619976043701,
      "learning_rate": 1.209114181881898e-06,
      "loss": 0.4521,
      "step": 331180
    },
    {
      "epoch": 135.57,
      "grad_norm": 2.252845048904419,
      "learning_rate": 1.2089753826040817e-06,
      "loss": 0.4672,
      "step": 331190
    },
    {
      "epoch": 135.57,
      "grad_norm": 2.0219008922576904,
      "learning_rate": 1.2088365885244923e-06,
      "loss": 0.4584,
      "step": 331200
    },
    {
      "epoch": 135.58,
      "grad_norm": 2.1281211376190186,
      "learning_rate": 1.2086977996437644e-06,
      "loss": 0.4596,
      "step": 331210
    },
    {
      "epoch": 135.58,
      "grad_norm": 1.9407635927200317,
      "learning_rate": 1.2085590159625359e-06,
      "loss": 0.4414,
      "step": 331220
    },
    {
      "epoch": 135.58,
      "grad_norm": 1.9197256565093994,
      "learning_rate": 1.2084202374814412e-06,
      "loss": 0.4415,
      "step": 331230
    },
    {
      "epoch": 135.59,
      "grad_norm": 1.9977965354919434,
      "learning_rate": 1.2082814642011168e-06,
      "loss": 0.4617,
      "step": 331240
    },
    {
      "epoch": 135.59,
      "grad_norm": 1.9937876462936401,
      "learning_rate": 1.2081426961221974e-06,
      "loss": 0.481,
      "step": 331250
    },
    {
      "epoch": 135.6,
      "grad_norm": 2.071845293045044,
      "learning_rate": 1.2080039332453193e-06,
      "loss": 0.4569,
      "step": 331260
    },
    {
      "epoch": 135.6,
      "grad_norm": 1.9340885877609253,
      "learning_rate": 1.2078651755711184e-06,
      "loss": 0.4658,
      "step": 331270
    },
    {
      "epoch": 135.6,
      "grad_norm": 2.570645332336426,
      "learning_rate": 1.2077264231002297e-06,
      "loss": 0.4593,
      "step": 331280
    },
    {
      "epoch": 135.61,
      "grad_norm": 2.10148024559021,
      "learning_rate": 1.207587675833289e-06,
      "loss": 0.4879,
      "step": 331290
    },
    {
      "epoch": 135.61,
      "grad_norm": 1.9202773571014404,
      "learning_rate": 1.207448933770933e-06,
      "loss": 0.4695,
      "step": 331300
    },
    {
      "epoch": 135.62,
      "grad_norm": 2.1507418155670166,
      "learning_rate": 1.2073101969137954e-06,
      "loss": 0.451,
      "step": 331310
    },
    {
      "epoch": 135.62,
      "grad_norm": 2.294980525970459,
      "learning_rate": 1.2071714652625124e-06,
      "loss": 0.4652,
      "step": 331320
    },
    {
      "epoch": 135.62,
      "grad_norm": 1.828047752380371,
      "learning_rate": 1.2070327388177195e-06,
      "loss": 0.4462,
      "step": 331330
    },
    {
      "epoch": 135.63,
      "grad_norm": 2.210681676864624,
      "learning_rate": 1.2068940175800527e-06,
      "loss": 0.4589,
      "step": 331340
    },
    {
      "epoch": 135.63,
      "grad_norm": 1.748677134513855,
      "learning_rate": 1.2067553015501466e-06,
      "loss": 0.4465,
      "step": 331350
    },
    {
      "epoch": 135.64,
      "grad_norm": 1.7334657907485962,
      "learning_rate": 1.2066165907286375e-06,
      "loss": 0.4405,
      "step": 331360
    },
    {
      "epoch": 135.64,
      "grad_norm": 1.8803366422653198,
      "learning_rate": 1.2064778851161606e-06,
      "loss": 0.4368,
      "step": 331370
    },
    {
      "epoch": 135.64,
      "grad_norm": 2.0299057960510254,
      "learning_rate": 1.2063391847133507e-06,
      "loss": 0.4562,
      "step": 331380
    },
    {
      "epoch": 135.65,
      "grad_norm": 2.4935250282287598,
      "learning_rate": 1.2062004895208437e-06,
      "loss": 0.4607,
      "step": 331390
    },
    {
      "epoch": 135.65,
      "grad_norm": 2.0035197734832764,
      "learning_rate": 1.206061799539274e-06,
      "loss": 0.4612,
      "step": 331400
    },
    {
      "epoch": 135.66,
      "grad_norm": 2.3344521522521973,
      "learning_rate": 1.2059231147692787e-06,
      "loss": 0.4544,
      "step": 331410
    },
    {
      "epoch": 135.66,
      "grad_norm": 2.069134473800659,
      "learning_rate": 1.205784435211492e-06,
      "loss": 0.4461,
      "step": 331420
    },
    {
      "epoch": 135.67,
      "grad_norm": 2.0275185108184814,
      "learning_rate": 1.2056457608665502e-06,
      "loss": 0.4615,
      "step": 331430
    },
    {
      "epoch": 135.67,
      "grad_norm": 1.9757113456726074,
      "learning_rate": 1.2055070917350873e-06,
      "loss": 0.467,
      "step": 331440
    },
    {
      "epoch": 135.67,
      "grad_norm": 1.9772820472717285,
      "learning_rate": 1.2053684278177384e-06,
      "loss": 0.4489,
      "step": 331450
    },
    {
      "epoch": 135.68,
      "grad_norm": 1.8668707609176636,
      "learning_rate": 1.205229769115139e-06,
      "loss": 0.4634,
      "step": 331460
    },
    {
      "epoch": 135.68,
      "grad_norm": 1.3899916410446167,
      "learning_rate": 1.2050911156279248e-06,
      "loss": 0.4593,
      "step": 331470
    },
    {
      "epoch": 135.69,
      "grad_norm": 2.3524107933044434,
      "learning_rate": 1.2049524673567306e-06,
      "loss": 0.4459,
      "step": 331480
    },
    {
      "epoch": 135.69,
      "grad_norm": 2.580352306365967,
      "learning_rate": 1.2048138243021913e-06,
      "loss": 0.4403,
      "step": 331490
    },
    {
      "epoch": 135.69,
      "grad_norm": 1.816367506980896,
      "learning_rate": 1.2046751864649422e-06,
      "loss": 0.4384,
      "step": 331500
    },
    {
      "epoch": 135.7,
      "grad_norm": 1.946410894393921,
      "learning_rate": 1.2045365538456188e-06,
      "loss": 0.4432,
      "step": 331510
    },
    {
      "epoch": 135.7,
      "grad_norm": 2.0257208347320557,
      "learning_rate": 1.2043979264448555e-06,
      "loss": 0.476,
      "step": 331520
    },
    {
      "epoch": 135.71,
      "grad_norm": 1.6290744543075562,
      "learning_rate": 1.2042593042632875e-06,
      "loss": 0.4507,
      "step": 331530
    },
    {
      "epoch": 135.71,
      "grad_norm": 2.0792391300201416,
      "learning_rate": 1.2041206873015503e-06,
      "loss": 0.4284,
      "step": 331540
    },
    {
      "epoch": 135.71,
      "grad_norm": 2.5578322410583496,
      "learning_rate": 1.2039820755602781e-06,
      "loss": 0.4526,
      "step": 331550
    },
    {
      "epoch": 135.72,
      "grad_norm": 2.028658628463745,
      "learning_rate": 1.2038434690401064e-06,
      "loss": 0.4554,
      "step": 331560
    },
    {
      "epoch": 135.72,
      "grad_norm": 1.9579496383666992,
      "learning_rate": 1.2037048677416699e-06,
      "loss": 0.4747,
      "step": 331570
    },
    {
      "epoch": 135.73,
      "grad_norm": 1.8625407218933105,
      "learning_rate": 1.2035662716656037e-06,
      "loss": 0.4308,
      "step": 331580
    },
    {
      "epoch": 135.73,
      "grad_norm": 2.142829418182373,
      "learning_rate": 1.2034276808125426e-06,
      "loss": 0.4657,
      "step": 331590
    },
    {
      "epoch": 135.73,
      "grad_norm": 2.7376468181610107,
      "learning_rate": 1.2032890951831215e-06,
      "loss": 0.4688,
      "step": 331600
    },
    {
      "epoch": 135.74,
      "grad_norm": 1.8810986280441284,
      "learning_rate": 1.2031505147779748e-06,
      "loss": 0.4499,
      "step": 331610
    },
    {
      "epoch": 135.74,
      "grad_norm": 2.0714597702026367,
      "learning_rate": 1.2030119395977382e-06,
      "loss": 0.4591,
      "step": 331620
    },
    {
      "epoch": 135.75,
      "grad_norm": 2.180269956588745,
      "learning_rate": 1.2028733696430456e-06,
      "loss": 0.4541,
      "step": 331630
    },
    {
      "epoch": 135.75,
      "grad_norm": 1.7780921459197998,
      "learning_rate": 1.2027348049145321e-06,
      "loss": 0.4284,
      "step": 331640
    },
    {
      "epoch": 135.76,
      "grad_norm": 2.056806802749634,
      "learning_rate": 1.2025962454128328e-06,
      "loss": 0.462,
      "step": 331650
    },
    {
      "epoch": 135.76,
      "grad_norm": 2.5105185508728027,
      "learning_rate": 1.2024576911385823e-06,
      "loss": 0.4514,
      "step": 331660
    },
    {
      "epoch": 135.76,
      "grad_norm": 1.9938164949417114,
      "learning_rate": 1.2023191420924146e-06,
      "loss": 0.4628,
      "step": 331670
    },
    {
      "epoch": 135.77,
      "grad_norm": 2.241698980331421,
      "learning_rate": 1.2021805982749656e-06,
      "loss": 0.466,
      "step": 331680
    },
    {
      "epoch": 135.77,
      "grad_norm": 1.8053914308547974,
      "learning_rate": 1.2020420596868684e-06,
      "loss": 0.4536,
      "step": 331690
    },
    {
      "epoch": 135.78,
      "grad_norm": 1.783410906791687,
      "learning_rate": 1.2019035263287589e-06,
      "loss": 0.4719,
      "step": 331700
    },
    {
      "epoch": 135.78,
      "grad_norm": 1.7439563274383545,
      "learning_rate": 1.2017649982012713e-06,
      "loss": 0.4702,
      "step": 331710
    },
    {
      "epoch": 135.78,
      "grad_norm": 1.6967884302139282,
      "learning_rate": 1.2016264753050397e-06,
      "loss": 0.4622,
      "step": 331720
    },
    {
      "epoch": 135.79,
      "grad_norm": 2.0156936645507812,
      "learning_rate": 1.2014879576406994e-06,
      "loss": 0.4485,
      "step": 331730
    },
    {
      "epoch": 135.79,
      "grad_norm": 2.03983211517334,
      "learning_rate": 1.2013494452088857e-06,
      "loss": 0.4597,
      "step": 331740
    },
    {
      "epoch": 135.8,
      "grad_norm": 2.225553035736084,
      "learning_rate": 1.201210938010231e-06,
      "loss": 0.4555,
      "step": 331750
    },
    {
      "epoch": 135.8,
      "grad_norm": 1.544830083847046,
      "learning_rate": 1.2010724360453707e-06,
      "loss": 0.4332,
      "step": 331760
    },
    {
      "epoch": 135.8,
      "grad_norm": 2.057314395904541,
      "learning_rate": 1.2009339393149395e-06,
      "loss": 0.4473,
      "step": 331770
    },
    {
      "epoch": 135.81,
      "grad_norm": 2.013155937194824,
      "learning_rate": 1.2007954478195706e-06,
      "loss": 0.442,
      "step": 331780
    },
    {
      "epoch": 135.81,
      "grad_norm": 1.9431874752044678,
      "learning_rate": 1.2006569615599007e-06,
      "loss": 0.4654,
      "step": 331790
    },
    {
      "epoch": 135.82,
      "grad_norm": 2.8980050086975098,
      "learning_rate": 1.2005184805365627e-06,
      "loss": 0.4512,
      "step": 331800
    },
    {
      "epoch": 135.82,
      "grad_norm": 1.8362561464309692,
      "learning_rate": 1.2003800047501911e-06,
      "loss": 0.4557,
      "step": 331810
    },
    {
      "epoch": 135.82,
      "grad_norm": 1.9837225675582886,
      "learning_rate": 1.2002415342014207e-06,
      "loss": 0.4439,
      "step": 331820
    },
    {
      "epoch": 135.83,
      "grad_norm": 1.9120261669158936,
      "learning_rate": 1.2001030688908852e-06,
      "loss": 0.4561,
      "step": 331830
    },
    {
      "epoch": 135.83,
      "grad_norm": 1.9582222700119019,
      "learning_rate": 1.1999646088192194e-06,
      "loss": 0.4625,
      "step": 331840
    },
    {
      "epoch": 135.84,
      "grad_norm": 2.287759304046631,
      "learning_rate": 1.1998261539870575e-06,
      "loss": 0.463,
      "step": 331850
    },
    {
      "epoch": 135.84,
      "grad_norm": 2.0423476696014404,
      "learning_rate": 1.199687704395033e-06,
      "loss": 0.447,
      "step": 331860
    },
    {
      "epoch": 135.85,
      "grad_norm": 2.101487159729004,
      "learning_rate": 1.1995492600437813e-06,
      "loss": 0.438,
      "step": 331870
    },
    {
      "epoch": 135.85,
      "grad_norm": 1.8292714357376099,
      "learning_rate": 1.1994108209339366e-06,
      "loss": 0.4745,
      "step": 331880
    },
    {
      "epoch": 135.85,
      "grad_norm": 2.0956897735595703,
      "learning_rate": 1.1992723870661312e-06,
      "loss": 0.4602,
      "step": 331890
    },
    {
      "epoch": 135.86,
      "grad_norm": 2.1850905418395996,
      "learning_rate": 1.1991339584410012e-06,
      "loss": 0.4644,
      "step": 331900
    },
    {
      "epoch": 135.86,
      "grad_norm": 2.0207972526550293,
      "learning_rate": 1.1989955350591797e-06,
      "loss": 0.4358,
      "step": 331910
    },
    {
      "epoch": 135.87,
      "grad_norm": 1.5725654363632202,
      "learning_rate": 1.1988571169213012e-06,
      "loss": 0.4435,
      "step": 331920
    },
    {
      "epoch": 135.87,
      "grad_norm": 1.992220401763916,
      "learning_rate": 1.1987187040279997e-06,
      "loss": 0.4614,
      "step": 331930
    },
    {
      "epoch": 135.87,
      "grad_norm": 2.297839641571045,
      "learning_rate": 1.1985802963799088e-06,
      "loss": 0.4455,
      "step": 331940
    },
    {
      "epoch": 135.88,
      "grad_norm": 2.166490316390991,
      "learning_rate": 1.1984418939776635e-06,
      "loss": 0.4647,
      "step": 331950
    },
    {
      "epoch": 135.88,
      "grad_norm": 1.8039159774780273,
      "learning_rate": 1.1983034968218962e-06,
      "loss": 0.4551,
      "step": 331960
    },
    {
      "epoch": 135.89,
      "grad_norm": 1.9381624460220337,
      "learning_rate": 1.1981651049132426e-06,
      "loss": 0.4643,
      "step": 331970
    },
    {
      "epoch": 135.89,
      "grad_norm": 1.9728713035583496,
      "learning_rate": 1.1980267182523364e-06,
      "loss": 0.4458,
      "step": 331980
    },
    {
      "epoch": 135.89,
      "grad_norm": 2.0884573459625244,
      "learning_rate": 1.1978883368398108e-06,
      "loss": 0.4625,
      "step": 331990
    },
    {
      "epoch": 135.9,
      "grad_norm": 2.7208945751190186,
      "learning_rate": 1.1977499606763003e-06,
      "loss": 0.4508,
      "step": 332000
    },
    {
      "epoch": 135.9,
      "grad_norm": 1.9538601636886597,
      "learning_rate": 1.197611589762439e-06,
      "loss": 0.4631,
      "step": 332010
    },
    {
      "epoch": 135.91,
      "grad_norm": 2.0744717121124268,
      "learning_rate": 1.1974732240988596e-06,
      "loss": 0.474,
      "step": 332020
    },
    {
      "epoch": 135.91,
      "grad_norm": 2.139909505844116,
      "learning_rate": 1.1973348636861967e-06,
      "loss": 0.4432,
      "step": 332030
    },
    {
      "epoch": 135.91,
      "grad_norm": 1.6910802125930786,
      "learning_rate": 1.197196508525084e-06,
      "loss": 0.4454,
      "step": 332040
    },
    {
      "epoch": 135.92,
      "grad_norm": 2.0408778190612793,
      "learning_rate": 1.197058158616155e-06,
      "loss": 0.4568,
      "step": 332050
    },
    {
      "epoch": 135.92,
      "grad_norm": 1.9092988967895508,
      "learning_rate": 1.196919813960044e-06,
      "loss": 0.4489,
      "step": 332060
    },
    {
      "epoch": 135.93,
      "grad_norm": 1.8948251008987427,
      "learning_rate": 1.1967814745573844e-06,
      "loss": 0.4644,
      "step": 332070
    },
    {
      "epoch": 135.93,
      "grad_norm": 1.928468108177185,
      "learning_rate": 1.1966431404088096e-06,
      "loss": 0.4328,
      "step": 332080
    },
    {
      "epoch": 135.94,
      "grad_norm": 1.6280150413513184,
      "learning_rate": 1.196504811514954e-06,
      "loss": 0.4532,
      "step": 332090
    },
    {
      "epoch": 135.94,
      "grad_norm": 1.9462872743606567,
      "learning_rate": 1.1963664878764512e-06,
      "loss": 0.4764,
      "step": 332100
    },
    {
      "epoch": 135.94,
      "grad_norm": 1.9343007802963257,
      "learning_rate": 1.1962281694939344e-06,
      "loss": 0.4613,
      "step": 332110
    },
    {
      "epoch": 135.95,
      "grad_norm": 1.655476450920105,
      "learning_rate": 1.1960898563680372e-06,
      "loss": 0.4391,
      "step": 332120
    },
    {
      "epoch": 135.95,
      "grad_norm": 2.4459762573242188,
      "learning_rate": 1.1959515484993933e-06,
      "loss": 0.4308,
      "step": 332130
    },
    {
      "epoch": 135.96,
      "grad_norm": 2.1832025051116943,
      "learning_rate": 1.1958132458886368e-06,
      "loss": 0.4605,
      "step": 332140
    },
    {
      "epoch": 135.96,
      "grad_norm": 1.999192476272583,
      "learning_rate": 1.1956749485364006e-06,
      "loss": 0.4644,
      "step": 332150
    },
    {
      "epoch": 135.96,
      "grad_norm": 1.8841897249221802,
      "learning_rate": 1.1955366564433184e-06,
      "loss": 0.4262,
      "step": 332160
    },
    {
      "epoch": 135.97,
      "grad_norm": 2.0648083686828613,
      "learning_rate": 1.1953983696100235e-06,
      "loss": 0.4592,
      "step": 332170
    },
    {
      "epoch": 135.97,
      "grad_norm": 1.9348094463348389,
      "learning_rate": 1.1952600880371497e-06,
      "loss": 0.444,
      "step": 332180
    },
    {
      "epoch": 135.98,
      "grad_norm": 1.5980350971221924,
      "learning_rate": 1.1951218117253304e-06,
      "loss": 0.4456,
      "step": 332190
    },
    {
      "epoch": 135.98,
      "grad_norm": 1.8832165002822876,
      "learning_rate": 1.194983540675199e-06,
      "loss": 0.4446,
      "step": 332200
    },
    {
      "epoch": 135.98,
      "grad_norm": 1.7573719024658203,
      "learning_rate": 1.1948452748873887e-06,
      "loss": 0.465,
      "step": 332210
    },
    {
      "epoch": 135.99,
      "grad_norm": 1.8900712728500366,
      "learning_rate": 1.1947070143625329e-06,
      "loss": 0.4567,
      "step": 332220
    },
    {
      "epoch": 135.99,
      "grad_norm": 2.1287457942962646,
      "learning_rate": 1.1945687591012652e-06,
      "loss": 0.443,
      "step": 332230
    },
    {
      "epoch": 136.0,
      "grad_norm": 1.7671414613723755,
      "learning_rate": 1.194430509104219e-06,
      "loss": 0.4372,
      "step": 332240
    },
    {
      "epoch": 136.0,
      "eval_loss": 0.45464131236076355,
      "eval_runtime": 58.1726,
      "eval_samples_per_second": 59.289,
      "eval_steps_per_second": 7.426,
      "step": 332248
    },
    {
      "epoch": 136.0,
      "grad_norm": 1.8403856754302979,
      "learning_rate": 1.1942922643720269e-06,
      "loss": 0.467,
      "step": 332250
    },
    {
      "epoch": 136.0,
      "grad_norm": 1.9207762479782104,
      "learning_rate": 1.1941540249053228e-06,
      "loss": 0.4414,
      "step": 332260
    },
    {
      "epoch": 136.01,
      "grad_norm": 3.3188366889953613,
      "learning_rate": 1.1940157907047403e-06,
      "loss": 0.4745,
      "step": 332270
    },
    {
      "epoch": 136.01,
      "grad_norm": 2.0232431888580322,
      "learning_rate": 1.193877561770912e-06,
      "loss": 0.4549,
      "step": 332280
    },
    {
      "epoch": 136.02,
      "grad_norm": 2.818937301635742,
      "learning_rate": 1.1937393381044708e-06,
      "loss": 0.4685,
      "step": 332290
    },
    {
      "epoch": 136.02,
      "grad_norm": 2.3022727966308594,
      "learning_rate": 1.1936011197060507e-06,
      "loss": 0.4548,
      "step": 332300
    },
    {
      "epoch": 136.03,
      "grad_norm": 2.6231470108032227,
      "learning_rate": 1.1934629065762846e-06,
      "loss": 0.4755,
      "step": 332310
    },
    {
      "epoch": 136.03,
      "grad_norm": 1.8030333518981934,
      "learning_rate": 1.193324698715806e-06,
      "loss": 0.4558,
      "step": 332320
    },
    {
      "epoch": 136.03,
      "grad_norm": 2.439406394958496,
      "learning_rate": 1.193186496125247e-06,
      "loss": 0.4566,
      "step": 332330
    },
    {
      "epoch": 136.04,
      "grad_norm": 1.6689311265945435,
      "learning_rate": 1.1930482988052403e-06,
      "loss": 0.444,
      "step": 332340
    },
    {
      "epoch": 136.04,
      "grad_norm": 1.833862543106079,
      "learning_rate": 1.1929101067564208e-06,
      "loss": 0.448,
      "step": 332350
    },
    {
      "epoch": 136.05,
      "grad_norm": 2.0095152854919434,
      "learning_rate": 1.1927719199794205e-06,
      "loss": 0.4551,
      "step": 332360
    },
    {
      "epoch": 136.05,
      "grad_norm": 1.8887401819229126,
      "learning_rate": 1.1926337384748723e-06,
      "loss": 0.4725,
      "step": 332370
    },
    {
      "epoch": 136.05,
      "grad_norm": 2.157104969024658,
      "learning_rate": 1.19249556224341e-06,
      "loss": 0.4711,
      "step": 332380
    },
    {
      "epoch": 136.06,
      "grad_norm": 1.6992861032485962,
      "learning_rate": 1.1923573912856654e-06,
      "loss": 0.4632,
      "step": 332390
    },
    {
      "epoch": 136.06,
      "grad_norm": 1.9478245973587036,
      "learning_rate": 1.1922192256022724e-06,
      "loss": 0.4487,
      "step": 332400
    },
    {
      "epoch": 136.07,
      "grad_norm": 1.7658448219299316,
      "learning_rate": 1.1920810651938634e-06,
      "loss": 0.4635,
      "step": 332410
    },
    {
      "epoch": 136.07,
      "grad_norm": 1.5798017978668213,
      "learning_rate": 1.1919429100610717e-06,
      "loss": 0.454,
      "step": 332420
    },
    {
      "epoch": 136.07,
      "grad_norm": 1.8579630851745605,
      "learning_rate": 1.1918047602045295e-06,
      "loss": 0.4263,
      "step": 332430
    },
    {
      "epoch": 136.08,
      "grad_norm": 2.196039915084839,
      "learning_rate": 1.1916666156248703e-06,
      "loss": 0.4538,
      "step": 332440
    },
    {
      "epoch": 136.08,
      "grad_norm": 1.7513588666915894,
      "learning_rate": 1.1915284763227262e-06,
      "loss": 0.4683,
      "step": 332450
    },
    {
      "epoch": 136.09,
      "grad_norm": 2.498168706893921,
      "learning_rate": 1.191390342298732e-06,
      "loss": 0.4561,
      "step": 332460
    },
    {
      "epoch": 136.09,
      "grad_norm": 2.078435182571411,
      "learning_rate": 1.191252213553518e-06,
      "loss": 0.4527,
      "step": 332470
    },
    {
      "epoch": 136.09,
      "grad_norm": 2.5854249000549316,
      "learning_rate": 1.1911140900877175e-06,
      "loss": 0.4698,
      "step": 332480
    },
    {
      "epoch": 136.1,
      "grad_norm": 2.105656147003174,
      "learning_rate": 1.1909759719019635e-06,
      "loss": 0.4676,
      "step": 332490
    },
    {
      "epoch": 136.1,
      "grad_norm": 2.7140324115753174,
      "learning_rate": 1.1908378589968892e-06,
      "loss": 0.4587,
      "step": 332500
    },
    {
      "epoch": 136.11,
      "grad_norm": 2.114666700363159,
      "learning_rate": 1.1906997513731266e-06,
      "loss": 0.4452,
      "step": 332510
    },
    {
      "epoch": 136.11,
      "grad_norm": 1.9037621021270752,
      "learning_rate": 1.1905616490313087e-06,
      "loss": 0.4574,
      "step": 332520
    },
    {
      "epoch": 136.12,
      "grad_norm": 2.5991153717041016,
      "learning_rate": 1.1904235519720671e-06,
      "loss": 0.4621,
      "step": 332530
    },
    {
      "epoch": 136.12,
      "grad_norm": 2.432504892349243,
      "learning_rate": 1.1902854601960363e-06,
      "loss": 0.4598,
      "step": 332540
    },
    {
      "epoch": 136.12,
      "grad_norm": 1.5709526538848877,
      "learning_rate": 1.190147373703848e-06,
      "loss": 0.4371,
      "step": 332550
    },
    {
      "epoch": 136.13,
      "grad_norm": 2.0905606746673584,
      "learning_rate": 1.1900092924961346e-06,
      "loss": 0.4493,
      "step": 332560
    },
    {
      "epoch": 136.13,
      "grad_norm": 2.0975241661071777,
      "learning_rate": 1.1898712165735286e-06,
      "loss": 0.4473,
      "step": 332570
    },
    {
      "epoch": 136.14,
      "grad_norm": 1.958740234375,
      "learning_rate": 1.1897331459366627e-06,
      "loss": 0.4681,
      "step": 332580
    },
    {
      "epoch": 136.14,
      "grad_norm": 1.86924147605896,
      "learning_rate": 1.1895950805861692e-06,
      "loss": 0.4653,
      "step": 332590
    },
    {
      "epoch": 136.14,
      "grad_norm": 2.1792263984680176,
      "learning_rate": 1.1894570205226817e-06,
      "loss": 0.4736,
      "step": 332600
    },
    {
      "epoch": 136.15,
      "grad_norm": 1.9242907762527466,
      "learning_rate": 1.1893189657468307e-06,
      "loss": 0.4605,
      "step": 332610
    },
    {
      "epoch": 136.15,
      "grad_norm": 1.9721962213516235,
      "learning_rate": 1.1891809162592495e-06,
      "loss": 0.4395,
      "step": 332620
    },
    {
      "epoch": 136.16,
      "grad_norm": 1.7991162538528442,
      "learning_rate": 1.1890428720605701e-06,
      "loss": 0.4509,
      "step": 332630
    },
    {
      "epoch": 136.16,
      "grad_norm": 1.8332750797271729,
      "learning_rate": 1.1889048331514258e-06,
      "loss": 0.4503,
      "step": 332640
    },
    {
      "epoch": 136.16,
      "grad_norm": 2.07965087890625,
      "learning_rate": 1.1887667995324478e-06,
      "loss": 0.4418,
      "step": 332650
    },
    {
      "epoch": 136.17,
      "grad_norm": 1.8919317722320557,
      "learning_rate": 1.188628771204269e-06,
      "loss": 0.4548,
      "step": 332660
    },
    {
      "epoch": 136.17,
      "grad_norm": 1.9427831172943115,
      "learning_rate": 1.188490748167522e-06,
      "loss": 0.4468,
      "step": 332670
    },
    {
      "epoch": 136.18,
      "grad_norm": 2.1501221656799316,
      "learning_rate": 1.1883527304228388e-06,
      "loss": 0.4443,
      "step": 332680
    },
    {
      "epoch": 136.18,
      "grad_norm": 2.444727659225464,
      "learning_rate": 1.1882147179708512e-06,
      "loss": 0.4443,
      "step": 332690
    },
    {
      "epoch": 136.19,
      "grad_norm": 1.7226496934890747,
      "learning_rate": 1.1880767108121918e-06,
      "loss": 0.4462,
      "step": 332700
    },
    {
      "epoch": 136.19,
      "grad_norm": 1.8598507642745972,
      "learning_rate": 1.1879387089474929e-06,
      "loss": 0.4345,
      "step": 332710
    },
    {
      "epoch": 136.19,
      "grad_norm": 1.9272726774215698,
      "learning_rate": 1.1878007123773855e-06,
      "loss": 0.4426,
      "step": 332720
    },
    {
      "epoch": 136.2,
      "grad_norm": 1.8499436378479004,
      "learning_rate": 1.1876627211025046e-06,
      "loss": 0.4409,
      "step": 332730
    },
    {
      "epoch": 136.2,
      "grad_norm": 1.7417073249816895,
      "learning_rate": 1.1875247351234797e-06,
      "loss": 0.477,
      "step": 332740
    },
    {
      "epoch": 136.21,
      "grad_norm": 2.7855348587036133,
      "learning_rate": 1.1873867544409434e-06,
      "loss": 0.4776,
      "step": 332750
    },
    {
      "epoch": 136.21,
      "grad_norm": 1.7315037250518799,
      "learning_rate": 1.1872487790555286e-06,
      "loss": 0.4396,
      "step": 332760
    },
    {
      "epoch": 136.21,
      "grad_norm": 1.6676627397537231,
      "learning_rate": 1.187110808967866e-06,
      "loss": 0.4369,
      "step": 332770
    },
    {
      "epoch": 136.22,
      "grad_norm": 2.5824615955352783,
      "learning_rate": 1.1869728441785893e-06,
      "loss": 0.4543,
      "step": 332780
    },
    {
      "epoch": 136.22,
      "grad_norm": 1.6174942255020142,
      "learning_rate": 1.186834884688329e-06,
      "loss": 0.4429,
      "step": 332790
    },
    {
      "epoch": 136.23,
      "grad_norm": 1.8736088275909424,
      "learning_rate": 1.1866969304977177e-06,
      "loss": 0.4572,
      "step": 332800
    },
    {
      "epoch": 136.23,
      "grad_norm": 2.2557363510131836,
      "learning_rate": 1.1865589816073874e-06,
      "loss": 0.4536,
      "step": 332810
    },
    {
      "epoch": 136.23,
      "grad_norm": 1.8831195831298828,
      "learning_rate": 1.1864210380179699e-06,
      "loss": 0.461,
      "step": 332820
    },
    {
      "epoch": 136.24,
      "grad_norm": 1.831557035446167,
      "learning_rate": 1.1862830997300973e-06,
      "loss": 0.4402,
      "step": 332830
    },
    {
      "epoch": 136.24,
      "grad_norm": 1.8815752267837524,
      "learning_rate": 1.1861451667444013e-06,
      "loss": 0.4626,
      "step": 332840
    },
    {
      "epoch": 136.25,
      "grad_norm": 1.889703392982483,
      "learning_rate": 1.1860072390615135e-06,
      "loss": 0.4717,
      "step": 332850
    },
    {
      "epoch": 136.25,
      "grad_norm": 2.031216621398926,
      "learning_rate": 1.1858693166820663e-06,
      "loss": 0.4614,
      "step": 332860
    },
    {
      "epoch": 136.25,
      "grad_norm": 1.7929102182388306,
      "learning_rate": 1.185731399606691e-06,
      "loss": 0.4571,
      "step": 332870
    },
    {
      "epoch": 136.26,
      "grad_norm": 1.9115737676620483,
      "learning_rate": 1.1855934878360197e-06,
      "loss": 0.4635,
      "step": 332880
    },
    {
      "epoch": 136.26,
      "grad_norm": 1.9911996126174927,
      "learning_rate": 1.1854555813706839e-06,
      "loss": 0.4776,
      "step": 332890
    },
    {
      "epoch": 136.27,
      "grad_norm": 1.653083324432373,
      "learning_rate": 1.1853176802113155e-06,
      "loss": 0.4495,
      "step": 332900
    },
    {
      "epoch": 136.27,
      "grad_norm": 2.0461173057556152,
      "learning_rate": 1.1851797843585463e-06,
      "loss": 0.4385,
      "step": 332910
    },
    {
      "epoch": 136.28,
      "grad_norm": 2.891749858856201,
      "learning_rate": 1.1850418938130078e-06,
      "loss": 0.4261,
      "step": 332920
    },
    {
      "epoch": 136.28,
      "grad_norm": 1.9682327508926392,
      "learning_rate": 1.1849040085753315e-06,
      "loss": 0.4726,
      "step": 332930
    },
    {
      "epoch": 136.28,
      "grad_norm": 1.660668969154358,
      "learning_rate": 1.1847661286461494e-06,
      "loss": 0.4674,
      "step": 332940
    },
    {
      "epoch": 136.29,
      "grad_norm": 1.8333524465560913,
      "learning_rate": 1.1846282540260925e-06,
      "loss": 0.465,
      "step": 332950
    },
    {
      "epoch": 136.29,
      "grad_norm": 1.6767158508300781,
      "learning_rate": 1.184490384715793e-06,
      "loss": 0.448,
      "step": 332960
    },
    {
      "epoch": 136.3,
      "grad_norm": 2.0122129917144775,
      "learning_rate": 1.1843525207158826e-06,
      "loss": 0.4493,
      "step": 332970
    },
    {
      "epoch": 136.3,
      "grad_norm": 1.740458369255066,
      "learning_rate": 1.184214662026992e-06,
      "loss": 0.4622,
      "step": 332980
    },
    {
      "epoch": 136.3,
      "grad_norm": 2.1070549488067627,
      "learning_rate": 1.1840768086497536e-06,
      "loss": 0.4412,
      "step": 332990
    },
    {
      "epoch": 136.31,
      "grad_norm": 2.282406806945801,
      "learning_rate": 1.1839389605847984e-06,
      "loss": 0.4428,
      "step": 333000
    },
    {
      "epoch": 136.31,
      "grad_norm": 2.629453182220459,
      "learning_rate": 1.183801117832758e-06,
      "loss": 0.4665,
      "step": 333010
    },
    {
      "epoch": 136.32,
      "grad_norm": 1.9275766611099243,
      "learning_rate": 1.1836632803942638e-06,
      "loss": 0.4574,
      "step": 333020
    },
    {
      "epoch": 136.32,
      "grad_norm": 2.3241939544677734,
      "learning_rate": 1.183525448269947e-06,
      "loss": 0.4507,
      "step": 333030
    },
    {
      "epoch": 136.32,
      "grad_norm": 2.1691319942474365,
      "learning_rate": 1.1833876214604402e-06,
      "loss": 0.4692,
      "step": 333040
    },
    {
      "epoch": 136.33,
      "grad_norm": 1.9666236639022827,
      "learning_rate": 1.1832497999663728e-06,
      "loss": 0.4561,
      "step": 333050
    },
    {
      "epoch": 136.33,
      "grad_norm": 1.854581356048584,
      "learning_rate": 1.183111983788377e-06,
      "loss": 0.4699,
      "step": 333060
    },
    {
      "epoch": 136.34,
      "grad_norm": 2.323786973953247,
      "learning_rate": 1.1829741729270846e-06,
      "loss": 0.4434,
      "step": 333070
    },
    {
      "epoch": 136.34,
      "grad_norm": 1.535233497619629,
      "learning_rate": 1.1828363673831264e-06,
      "loss": 0.4498,
      "step": 333080
    },
    {
      "epoch": 136.34,
      "grad_norm": 1.4962642192840576,
      "learning_rate": 1.1826985671571334e-06,
      "loss": 0.4599,
      "step": 333090
    },
    {
      "epoch": 136.35,
      "grad_norm": 2.234217882156372,
      "learning_rate": 1.1825607722497366e-06,
      "loss": 0.4775,
      "step": 333100
    },
    {
      "epoch": 136.35,
      "grad_norm": 2.030649185180664,
      "learning_rate": 1.1824229826615686e-06,
      "loss": 0.4539,
      "step": 333110
    },
    {
      "epoch": 136.36,
      "grad_norm": 1.8882914781570435,
      "learning_rate": 1.1822851983932596e-06,
      "loss": 0.4542,
      "step": 333120
    },
    {
      "epoch": 136.36,
      "grad_norm": 1.8482633829116821,
      "learning_rate": 1.1821474194454415e-06,
      "loss": 0.4429,
      "step": 333130
    },
    {
      "epoch": 136.37,
      "grad_norm": 1.560530662536621,
      "learning_rate": 1.1820096458187444e-06,
      "loss": 0.4553,
      "step": 333140
    },
    {
      "epoch": 136.37,
      "grad_norm": 1.9821140766143799,
      "learning_rate": 1.1818718775138002e-06,
      "loss": 0.4412,
      "step": 333150
    },
    {
      "epoch": 136.37,
      "grad_norm": 1.8786851167678833,
      "learning_rate": 1.18173411453124e-06,
      "loss": 0.464,
      "step": 333160
    },
    {
      "epoch": 136.38,
      "grad_norm": 2.2165653705596924,
      "learning_rate": 1.181596356871694e-06,
      "loss": 0.4331,
      "step": 333170
    },
    {
      "epoch": 136.38,
      "grad_norm": 2.519789457321167,
      "learning_rate": 1.1814586045357954e-06,
      "loss": 0.4661,
      "step": 333180
    },
    {
      "epoch": 136.39,
      "grad_norm": 1.6875159740447998,
      "learning_rate": 1.1813208575241726e-06,
      "loss": 0.4345,
      "step": 333190
    },
    {
      "epoch": 136.39,
      "grad_norm": 1.8993499279022217,
      "learning_rate": 1.1811831158374573e-06,
      "loss": 0.4704,
      "step": 333200
    },
    {
      "epoch": 136.39,
      "grad_norm": 1.8638806343078613,
      "learning_rate": 1.1810453794762813e-06,
      "loss": 0.4588,
      "step": 333210
    },
    {
      "epoch": 136.4,
      "grad_norm": 2.308197498321533,
      "learning_rate": 1.1809076484412747e-06,
      "loss": 0.4745,
      "step": 333220
    },
    {
      "epoch": 136.4,
      "grad_norm": 1.707992434501648,
      "learning_rate": 1.180769922733069e-06,
      "loss": 0.4511,
      "step": 333230
    },
    {
      "epoch": 136.41,
      "grad_norm": 1.9020410776138306,
      "learning_rate": 1.180632202352295e-06,
      "loss": 0.4438,
      "step": 333240
    },
    {
      "epoch": 136.41,
      "grad_norm": 2.096540927886963,
      "learning_rate": 1.1804944872995835e-06,
      "loss": 0.435,
      "step": 333250
    },
    {
      "epoch": 136.41,
      "grad_norm": 1.6962677240371704,
      "learning_rate": 1.1803567775755654e-06,
      "loss": 0.4704,
      "step": 333260
    },
    {
      "epoch": 136.42,
      "grad_norm": 2.074221134185791,
      "learning_rate": 1.180219073180871e-06,
      "loss": 0.4496,
      "step": 333270
    },
    {
      "epoch": 136.42,
      "grad_norm": 2.1391794681549072,
      "learning_rate": 1.1800813741161313e-06,
      "loss": 0.4738,
      "step": 333280
    },
    {
      "epoch": 136.43,
      "grad_norm": 2.1145031452178955,
      "learning_rate": 1.179943680381978e-06,
      "loss": 0.4472,
      "step": 333290
    },
    {
      "epoch": 136.43,
      "grad_norm": 1.8137121200561523,
      "learning_rate": 1.1798059919790413e-06,
      "loss": 0.4522,
      "step": 333300
    },
    {
      "epoch": 136.43,
      "grad_norm": 2.1480555534362793,
      "learning_rate": 1.1796683089079517e-06,
      "loss": 0.4656,
      "step": 333310
    },
    {
      "epoch": 136.44,
      "grad_norm": 1.9890868663787842,
      "learning_rate": 1.1795306311693409e-06,
      "loss": 0.4572,
      "step": 333320
    },
    {
      "epoch": 136.44,
      "grad_norm": 2.7274422645568848,
      "learning_rate": 1.1793929587638382e-06,
      "loss": 0.4476,
      "step": 333330
    },
    {
      "epoch": 136.45,
      "grad_norm": 2.1836326122283936,
      "learning_rate": 1.1792552916920744e-06,
      "loss": 0.4539,
      "step": 333340
    },
    {
      "epoch": 136.45,
      "grad_norm": 2.0608224868774414,
      "learning_rate": 1.1791176299546806e-06,
      "loss": 0.445,
      "step": 333350
    },
    {
      "epoch": 136.46,
      "grad_norm": 1.8384008407592773,
      "learning_rate": 1.1789799735522874e-06,
      "loss": 0.4394,
      "step": 333360
    },
    {
      "epoch": 136.46,
      "grad_norm": 2.098129987716675,
      "learning_rate": 1.1788423224855253e-06,
      "loss": 0.4837,
      "step": 333370
    },
    {
      "epoch": 136.46,
      "grad_norm": 2.5370936393737793,
      "learning_rate": 1.178704676755025e-06,
      "loss": 0.463,
      "step": 333380
    },
    {
      "epoch": 136.47,
      "grad_norm": 1.9671568870544434,
      "learning_rate": 1.1785670363614164e-06,
      "loss": 0.4466,
      "step": 333390
    },
    {
      "epoch": 136.47,
      "grad_norm": 2.0109379291534424,
      "learning_rate": 1.1784294013053309e-06,
      "loss": 0.4561,
      "step": 333400
    },
    {
      "epoch": 136.48,
      "grad_norm": 2.2374608516693115,
      "learning_rate": 1.1782917715873984e-06,
      "loss": 0.4597,
      "step": 333410
    },
    {
      "epoch": 136.48,
      "grad_norm": 2.2060205936431885,
      "learning_rate": 1.1781541472082498e-06,
      "loss": 0.4508,
      "step": 333420
    },
    {
      "epoch": 136.48,
      "grad_norm": 1.9763190746307373,
      "learning_rate": 1.178016528168515e-06,
      "loss": 0.453,
      "step": 333430
    },
    {
      "epoch": 136.49,
      "grad_norm": 2.7928643226623535,
      "learning_rate": 1.1778789144688248e-06,
      "loss": 0.436,
      "step": 333440
    },
    {
      "epoch": 136.49,
      "grad_norm": 1.9216859340667725,
      "learning_rate": 1.1777413061098099e-06,
      "loss": 0.4572,
      "step": 333450
    },
    {
      "epoch": 136.5,
      "grad_norm": 1.8749085664749146,
      "learning_rate": 1.1776037030920995e-06,
      "loss": 0.4487,
      "step": 333460
    },
    {
      "epoch": 136.5,
      "grad_norm": 2.388921022415161,
      "learning_rate": 1.177466105416325e-06,
      "loss": 0.469,
      "step": 333470
    },
    {
      "epoch": 136.5,
      "grad_norm": 1.9884644746780396,
      "learning_rate": 1.1773285130831167e-06,
      "loss": 0.4613,
      "step": 333480
    },
    {
      "epoch": 136.51,
      "grad_norm": 1.922812819480896,
      "learning_rate": 1.1771909260931044e-06,
      "loss": 0.4518,
      "step": 333490
    },
    {
      "epoch": 136.51,
      "grad_norm": 2.081449508666992,
      "learning_rate": 1.1770533444469186e-06,
      "loss": 0.4438,
      "step": 333500
    },
    {
      "epoch": 136.52,
      "grad_norm": 1.8649977445602417,
      "learning_rate": 1.1769157681451892e-06,
      "loss": 0.4612,
      "step": 333510
    },
    {
      "epoch": 136.52,
      "grad_norm": 1.6543322801589966,
      "learning_rate": 1.1767781971885469e-06,
      "loss": 0.4528,
      "step": 333520
    },
    {
      "epoch": 136.52,
      "grad_norm": 1.8664569854736328,
      "learning_rate": 1.1766406315776216e-06,
      "loss": 0.4244,
      "step": 333530
    },
    {
      "epoch": 136.53,
      "grad_norm": 2.1009609699249268,
      "learning_rate": 1.176503071313044e-06,
      "loss": 0.4462,
      "step": 333540
    },
    {
      "epoch": 136.53,
      "grad_norm": 2.0747087001800537,
      "learning_rate": 1.1763655163954434e-06,
      "loss": 0.4651,
      "step": 333550
    },
    {
      "epoch": 136.54,
      "grad_norm": 1.7437633275985718,
      "learning_rate": 1.1762279668254504e-06,
      "loss": 0.4369,
      "step": 333560
    },
    {
      "epoch": 136.54,
      "grad_norm": 1.7151771783828735,
      "learning_rate": 1.176090422603695e-06,
      "loss": 0.4504,
      "step": 333570
    },
    {
      "epoch": 136.55,
      "grad_norm": 2.006009578704834,
      "learning_rate": 1.1759528837308074e-06,
      "loss": 0.4534,
      "step": 333580
    },
    {
      "epoch": 136.55,
      "grad_norm": 1.8316121101379395,
      "learning_rate": 1.1758153502074174e-06,
      "loss": 0.4414,
      "step": 333590
    },
    {
      "epoch": 136.55,
      "grad_norm": 1.8777199983596802,
      "learning_rate": 1.175677822034155e-06,
      "loss": 0.4359,
      "step": 333600
    },
    {
      "epoch": 136.56,
      "grad_norm": 1.7561732530593872,
      "learning_rate": 1.1755402992116509e-06,
      "loss": 0.4506,
      "step": 333610
    },
    {
      "epoch": 136.56,
      "grad_norm": 2.08821177482605,
      "learning_rate": 1.175402781740534e-06,
      "loss": 0.4495,
      "step": 333620
    },
    {
      "epoch": 136.57,
      "grad_norm": 3.0367658138275146,
      "learning_rate": 1.175265269621436e-06,
      "loss": 0.4509,
      "step": 333630
    },
    {
      "epoch": 136.57,
      "grad_norm": 2.2314226627349854,
      "learning_rate": 1.1751277628549841e-06,
      "loss": 0.4445,
      "step": 333640
    },
    {
      "epoch": 136.57,
      "grad_norm": 1.6279191970825195,
      "learning_rate": 1.1749902614418106e-06,
      "loss": 0.4357,
      "step": 333650
    },
    {
      "epoch": 136.58,
      "grad_norm": 1.6305761337280273,
      "learning_rate": 1.174852765382543e-06,
      "loss": 0.4434,
      "step": 333660
    },
    {
      "epoch": 136.58,
      "grad_norm": 2.1982884407043457,
      "learning_rate": 1.1747152746778134e-06,
      "loss": 0.4581,
      "step": 333670
    },
    {
      "epoch": 136.59,
      "grad_norm": 2.2846462726593018,
      "learning_rate": 1.1745777893282513e-06,
      "loss": 0.4597,
      "step": 333680
    },
    {
      "epoch": 136.59,
      "grad_norm": 1.8940222263336182,
      "learning_rate": 1.1744403093344854e-06,
      "loss": 0.4597,
      "step": 333690
    },
    {
      "epoch": 136.59,
      "grad_norm": 2.632493019104004,
      "learning_rate": 1.1743028346971467e-06,
      "loss": 0.438,
      "step": 333700
    },
    {
      "epoch": 136.6,
      "grad_norm": 1.7174991369247437,
      "learning_rate": 1.1741653654168638e-06,
      "loss": 0.449,
      "step": 333710
    },
    {
      "epoch": 136.6,
      "grad_norm": 2.528036117553711,
      "learning_rate": 1.1740279014942674e-06,
      "loss": 0.4647,
      "step": 333720
    },
    {
      "epoch": 136.61,
      "grad_norm": 1.8878413438796997,
      "learning_rate": 1.1738904429299866e-06,
      "loss": 0.4659,
      "step": 333730
    },
    {
      "epoch": 136.61,
      "grad_norm": 2.2041547298431396,
      "learning_rate": 1.1737529897246514e-06,
      "loss": 0.4392,
      "step": 333740
    },
    {
      "epoch": 136.61,
      "grad_norm": 2.331926107406616,
      "learning_rate": 1.1736155418788911e-06,
      "loss": 0.4275,
      "step": 333750
    },
    {
      "epoch": 136.62,
      "grad_norm": 2.3093490600585938,
      "learning_rate": 1.1734780993933366e-06,
      "loss": 0.4501,
      "step": 333760
    },
    {
      "epoch": 136.62,
      "grad_norm": 2.1476542949676514,
      "learning_rate": 1.1733406622686155e-06,
      "loss": 0.4612,
      "step": 333770
    },
    {
      "epoch": 136.63,
      "grad_norm": 2.3008310794830322,
      "learning_rate": 1.1732032305053584e-06,
      "loss": 0.4506,
      "step": 333780
    },
    {
      "epoch": 136.63,
      "grad_norm": 2.1394357681274414,
      "learning_rate": 1.1730658041041945e-06,
      "loss": 0.439,
      "step": 333790
    },
    {
      "epoch": 136.64,
      "grad_norm": 1.75174880027771,
      "learning_rate": 1.172928383065754e-06,
      "loss": 0.4551,
      "step": 333800
    },
    {
      "epoch": 136.64,
      "grad_norm": 2.4698452949523926,
      "learning_rate": 1.1727909673906658e-06,
      "loss": 0.4375,
      "step": 333810
    },
    {
      "epoch": 136.64,
      "grad_norm": 1.983784794807434,
      "learning_rate": 1.1726535570795595e-06,
      "loss": 0.4474,
      "step": 333820
    },
    {
      "epoch": 136.65,
      "grad_norm": 1.647242784500122,
      "learning_rate": 1.1725161521330646e-06,
      "loss": 0.443,
      "step": 333830
    },
    {
      "epoch": 136.65,
      "grad_norm": 2.303490161895752,
      "learning_rate": 1.1723787525518104e-06,
      "loss": 0.4624,
      "step": 333840
    },
    {
      "epoch": 136.66,
      "grad_norm": 2.06058931350708,
      "learning_rate": 1.1722413583364259e-06,
      "loss": 0.4326,
      "step": 333850
    },
    {
      "epoch": 136.66,
      "grad_norm": 2.5875394344329834,
      "learning_rate": 1.1721039694875419e-06,
      "loss": 0.4411,
      "step": 333860
    },
    {
      "epoch": 136.66,
      "grad_norm": 1.9824236631393433,
      "learning_rate": 1.171966586005787e-06,
      "loss": 0.4513,
      "step": 333870
    },
    {
      "epoch": 136.67,
      "grad_norm": 2.017775774002075,
      "learning_rate": 1.1718292078917905e-06,
      "loss": 0.4432,
      "step": 333880
    },
    {
      "epoch": 136.67,
      "grad_norm": 2.0271170139312744,
      "learning_rate": 1.1716918351461812e-06,
      "loss": 0.4503,
      "step": 333890
    },
    {
      "epoch": 136.68,
      "grad_norm": 3.43337345123291,
      "learning_rate": 1.1715544677695902e-06,
      "loss": 0.4497,
      "step": 333900
    },
    {
      "epoch": 136.68,
      "grad_norm": 1.854660987854004,
      "learning_rate": 1.1714171057626441e-06,
      "loss": 0.4462,
      "step": 333910
    },
    {
      "epoch": 136.68,
      "grad_norm": 1.6874418258666992,
      "learning_rate": 1.1712797491259737e-06,
      "loss": 0.4696,
      "step": 333920
    },
    {
      "epoch": 136.69,
      "grad_norm": 1.7894504070281982,
      "learning_rate": 1.1711423978602082e-06,
      "loss": 0.4558,
      "step": 333930
    },
    {
      "epoch": 136.69,
      "grad_norm": 1.8312236070632935,
      "learning_rate": 1.1710050519659762e-06,
      "loss": 0.4555,
      "step": 333940
    },
    {
      "epoch": 136.7,
      "grad_norm": 1.9904755353927612,
      "learning_rate": 1.170867711443907e-06,
      "loss": 0.4683,
      "step": 333950
    },
    {
      "epoch": 136.7,
      "grad_norm": 2.0120036602020264,
      "learning_rate": 1.1707303762946302e-06,
      "loss": 0.4519,
      "step": 333960
    },
    {
      "epoch": 136.7,
      "grad_norm": 1.928595781326294,
      "learning_rate": 1.1705930465187747e-06,
      "loss": 0.4491,
      "step": 333970
    },
    {
      "epoch": 136.71,
      "grad_norm": 2.0098612308502197,
      "learning_rate": 1.1704557221169693e-06,
      "loss": 0.4555,
      "step": 333980
    },
    {
      "epoch": 136.71,
      "grad_norm": 1.81475830078125,
      "learning_rate": 1.1703184030898433e-06,
      "loss": 0.4377,
      "step": 333990
    },
    {
      "epoch": 136.72,
      "grad_norm": 2.2646231651306152,
      "learning_rate": 1.1701810894380259e-06,
      "loss": 0.4415,
      "step": 334000
    },
    {
      "epoch": 136.72,
      "grad_norm": 2.336305618286133,
      "learning_rate": 1.1700437811621457e-06,
      "loss": 0.4499,
      "step": 334010
    },
    {
      "epoch": 136.73,
      "grad_norm": 2.1496410369873047,
      "learning_rate": 1.169906478262832e-06,
      "loss": 0.4655,
      "step": 334020
    },
    {
      "epoch": 136.73,
      "grad_norm": 2.0299930572509766,
      "learning_rate": 1.1697691807407138e-06,
      "loss": 0.4495,
      "step": 334030
    },
    {
      "epoch": 136.73,
      "grad_norm": 1.5928751230239868,
      "learning_rate": 1.1696318885964199e-06,
      "loss": 0.466,
      "step": 334040
    },
    {
      "epoch": 136.74,
      "grad_norm": 1.9790804386138916,
      "learning_rate": 1.1694946018305791e-06,
      "loss": 0.4432,
      "step": 334050
    },
    {
      "epoch": 136.74,
      "grad_norm": 1.8376240730285645,
      "learning_rate": 1.1693573204438204e-06,
      "loss": 0.43,
      "step": 334060
    },
    {
      "epoch": 136.75,
      "grad_norm": 1.699092984199524,
      "learning_rate": 1.1692200444367728e-06,
      "loss": 0.4584,
      "step": 334070
    },
    {
      "epoch": 136.75,
      "grad_norm": 1.827439546585083,
      "learning_rate": 1.169082773810065e-06,
      "loss": 0.4482,
      "step": 334080
    },
    {
      "epoch": 136.75,
      "grad_norm": 1.7001912593841553,
      "learning_rate": 1.168945508564326e-06,
      "loss": 0.4449,
      "step": 334090
    },
    {
      "epoch": 136.76,
      "grad_norm": 1.971984624862671,
      "learning_rate": 1.1688082487001841e-06,
      "loss": 0.4427,
      "step": 334100
    },
    {
      "epoch": 136.76,
      "grad_norm": 2.5344929695129395,
      "learning_rate": 1.1686709942182687e-06,
      "loss": 0.4728,
      "step": 334110
    },
    {
      "epoch": 136.77,
      "grad_norm": 2.0183820724487305,
      "learning_rate": 1.168533745119208e-06,
      "loss": 0.4478,
      "step": 334120
    },
    {
      "epoch": 136.77,
      "grad_norm": 2.0361926555633545,
      "learning_rate": 1.1683965014036311e-06,
      "loss": 0.4755,
      "step": 334130
    },
    {
      "epoch": 136.77,
      "grad_norm": 1.764716625213623,
      "learning_rate": 1.1682592630721664e-06,
      "loss": 0.4656,
      "step": 334140
    },
    {
      "epoch": 136.78,
      "grad_norm": 1.6407028436660767,
      "learning_rate": 1.1681220301254428e-06,
      "loss": 0.4392,
      "step": 334150
    },
    {
      "epoch": 136.78,
      "grad_norm": 2.369405746459961,
      "learning_rate": 1.167984802564089e-06,
      "loss": 0.4769,
      "step": 334160
    },
    {
      "epoch": 136.79,
      "grad_norm": 2.1331965923309326,
      "learning_rate": 1.1678475803887334e-06,
      "loss": 0.4626,
      "step": 334170
    },
    {
      "epoch": 136.79,
      "grad_norm": 1.8727084398269653,
      "learning_rate": 1.1677103636000047e-06,
      "loss": 0.4519,
      "step": 334180
    },
    {
      "epoch": 136.79,
      "grad_norm": 1.6863445043563843,
      "learning_rate": 1.1675731521985313e-06,
      "loss": 0.4386,
      "step": 334190
    },
    {
      "epoch": 136.8,
      "grad_norm": 2.0591516494750977,
      "learning_rate": 1.167435946184942e-06,
      "loss": 0.4458,
      "step": 334200
    },
    {
      "epoch": 136.8,
      "grad_norm": 1.9438652992248535,
      "learning_rate": 1.1672987455598662e-06,
      "loss": 0.4497,
      "step": 334210
    },
    {
      "epoch": 136.81,
      "grad_norm": 2.090885877609253,
      "learning_rate": 1.1671615503239303e-06,
      "loss": 0.4665,
      "step": 334220
    },
    {
      "epoch": 136.81,
      "grad_norm": 1.9394272565841675,
      "learning_rate": 1.167024360477763e-06,
      "loss": 0.4561,
      "step": 334230
    },
    {
      "epoch": 136.82,
      "grad_norm": 1.864185094833374,
      "learning_rate": 1.1668871760219948e-06,
      "loss": 0.4327,
      "step": 334240
    },
    {
      "epoch": 136.82,
      "grad_norm": 1.825736403465271,
      "learning_rate": 1.1667499969572526e-06,
      "loss": 0.4582,
      "step": 334250
    },
    {
      "epoch": 136.82,
      "grad_norm": 2.0783143043518066,
      "learning_rate": 1.1666128232841652e-06,
      "loss": 0.4814,
      "step": 334260
    },
    {
      "epoch": 136.83,
      "grad_norm": 2.218337059020996,
      "learning_rate": 1.1664756550033609e-06,
      "loss": 0.4697,
      "step": 334270
    },
    {
      "epoch": 136.83,
      "grad_norm": 1.7296512126922607,
      "learning_rate": 1.1663384921154682e-06,
      "loss": 0.4648,
      "step": 334280
    },
    {
      "epoch": 136.84,
      "grad_norm": 2.3663926124572754,
      "learning_rate": 1.1662013346211152e-06,
      "loss": 0.4887,
      "step": 334290
    },
    {
      "epoch": 136.84,
      "grad_norm": 2.2599172592163086,
      "learning_rate": 1.1660641825209302e-06,
      "loss": 0.4626,
      "step": 334300
    },
    {
      "epoch": 136.84,
      "grad_norm": 2.2124695777893066,
      "learning_rate": 1.1659270358155416e-06,
      "loss": 0.4406,
      "step": 334310
    },
    {
      "epoch": 136.85,
      "grad_norm": 2.177802324295044,
      "learning_rate": 1.1657898945055776e-06,
      "loss": 0.445,
      "step": 334320
    },
    {
      "epoch": 136.85,
      "grad_norm": 2.0019145011901855,
      "learning_rate": 1.1656527585916661e-06,
      "loss": 0.4597,
      "step": 334330
    },
    {
      "epoch": 136.86,
      "grad_norm": 1.970631718635559,
      "learning_rate": 1.1655156280744369e-06,
      "loss": 0.464,
      "step": 334340
    },
    {
      "epoch": 136.86,
      "grad_norm": 2.2044618129730225,
      "learning_rate": 1.165378502954516e-06,
      "loss": 0.4508,
      "step": 334350
    },
    {
      "epoch": 136.86,
      "grad_norm": 2.460233688354492,
      "learning_rate": 1.165241383232532e-06,
      "loss": 0.46,
      "step": 334360
    },
    {
      "epoch": 136.87,
      "grad_norm": 2.2640130519866943,
      "learning_rate": 1.165104268909114e-06,
      "loss": 0.4668,
      "step": 334370
    },
    {
      "epoch": 136.87,
      "grad_norm": 1.7640736103057861,
      "learning_rate": 1.1649671599848895e-06,
      "loss": 0.4351,
      "step": 334380
    },
    {
      "epoch": 136.88,
      "grad_norm": 1.9505940675735474,
      "learning_rate": 1.1648300564604863e-06,
      "loss": 0.4498,
      "step": 334390
    },
    {
      "epoch": 136.88,
      "grad_norm": 2.0466363430023193,
      "learning_rate": 1.1646929583365329e-06,
      "loss": 0.451,
      "step": 334400
    },
    {
      "epoch": 136.88,
      "grad_norm": 2.0548079013824463,
      "learning_rate": 1.1645558656136573e-06,
      "loss": 0.4539,
      "step": 334410
    },
    {
      "epoch": 136.89,
      "grad_norm": 2.33674693107605,
      "learning_rate": 1.1644187782924864e-06,
      "loss": 0.454,
      "step": 334420
    },
    {
      "epoch": 136.89,
      "grad_norm": 2.0739572048187256,
      "learning_rate": 1.1642816963736505e-06,
      "loss": 0.4406,
      "step": 334430
    },
    {
      "epoch": 136.9,
      "grad_norm": 2.168041944503784,
      "learning_rate": 1.1641446198577756e-06,
      "loss": 0.4332,
      "step": 334440
    },
    {
      "epoch": 136.9,
      "grad_norm": 2.24143123626709,
      "learning_rate": 1.1640075487454905e-06,
      "loss": 0.4623,
      "step": 334450
    },
    {
      "epoch": 136.91,
      "grad_norm": 2.0526773929595947,
      "learning_rate": 1.163870483037423e-06,
      "loss": 0.4442,
      "step": 334460
    },
    {
      "epoch": 136.91,
      "grad_norm": 2.2008230686187744,
      "learning_rate": 1.1637334227342006e-06,
      "loss": 0.4529,
      "step": 334470
    },
    {
      "epoch": 136.91,
      "grad_norm": 2.0591413974761963,
      "learning_rate": 1.1635963678364527e-06,
      "loss": 0.4617,
      "step": 334480
    },
    {
      "epoch": 136.92,
      "grad_norm": 1.942078709602356,
      "learning_rate": 1.1634593183448046e-06,
      "loss": 0.4442,
      "step": 334490
    },
    {
      "epoch": 136.92,
      "grad_norm": 1.9980549812316895,
      "learning_rate": 1.1633222742598852e-06,
      "loss": 0.4537,
      "step": 334500
    },
    {
      "epoch": 136.93,
      "grad_norm": 2.4276044368743896,
      "learning_rate": 1.1631852355823227e-06,
      "loss": 0.4708,
      "step": 334510
    },
    {
      "epoch": 136.93,
      "grad_norm": 1.5989758968353271,
      "learning_rate": 1.1630482023127445e-06,
      "loss": 0.4564,
      "step": 334520
    },
    {
      "epoch": 136.93,
      "grad_norm": 1.9326226711273193,
      "learning_rate": 1.1629111744517784e-06,
      "loss": 0.4658,
      "step": 334530
    },
    {
      "epoch": 136.94,
      "grad_norm": 1.8253377676010132,
      "learning_rate": 1.162774152000052e-06,
      "loss": 0.4481,
      "step": 334540
    },
    {
      "epoch": 136.94,
      "grad_norm": 1.915787696838379,
      "learning_rate": 1.162637134958193e-06,
      "loss": 0.4706,
      "step": 334550
    },
    {
      "epoch": 136.95,
      "grad_norm": 2.2739100456237793,
      "learning_rate": 1.1625001233268292e-06,
      "loss": 0.463,
      "step": 334560
    },
    {
      "epoch": 136.95,
      "grad_norm": 1.8539386987686157,
      "learning_rate": 1.1623631171065879e-06,
      "loss": 0.4541,
      "step": 334570
    },
    {
      "epoch": 136.95,
      "grad_norm": 2.454695463180542,
      "learning_rate": 1.162226116298097e-06,
      "loss": 0.4361,
      "step": 334580
    },
    {
      "epoch": 136.96,
      "grad_norm": 1.925950527191162,
      "learning_rate": 1.1620891209019844e-06,
      "loss": 0.4519,
      "step": 334590
    },
    {
      "epoch": 136.96,
      "grad_norm": 1.9857144355773926,
      "learning_rate": 1.1619521309188762e-06,
      "loss": 0.4491,
      "step": 334600
    },
    {
      "epoch": 136.97,
      "grad_norm": 2.3206286430358887,
      "learning_rate": 1.1618151463494019e-06,
      "loss": 0.4602,
      "step": 334610
    },
    {
      "epoch": 136.97,
      "grad_norm": 2.054978132247925,
      "learning_rate": 1.161678167194189e-06,
      "loss": 0.4367,
      "step": 334620
    },
    {
      "epoch": 136.98,
      "grad_norm": 2.169440269470215,
      "learning_rate": 1.1615411934538632e-06,
      "loss": 0.4739,
      "step": 334630
    },
    {
      "epoch": 136.98,
      "grad_norm": 1.9467114210128784,
      "learning_rate": 1.161404225129053e-06,
      "loss": 0.4495,
      "step": 334640
    },
    {
      "epoch": 136.98,
      "grad_norm": 1.7119791507720947,
      "learning_rate": 1.1612672622203855e-06,
      "loss": 0.4798,
      "step": 334650
    },
    {
      "epoch": 136.99,
      "grad_norm": 1.755968451499939,
      "learning_rate": 1.1611303047284882e-06,
      "loss": 0.4637,
      "step": 334660
    },
    {
      "epoch": 136.99,
      "grad_norm": 1.96506667137146,
      "learning_rate": 1.1609933526539888e-06,
      "loss": 0.421,
      "step": 334670
    },
    {
      "epoch": 137.0,
      "grad_norm": 1.9938656091690063,
      "learning_rate": 1.1608564059975146e-06,
      "loss": 0.4576,
      "step": 334680
    },
    {
      "epoch": 137.0,
      "grad_norm": 1.9717609882354736,
      "learning_rate": 1.1607194647596923e-06,
      "loss": 0.4353,
      "step": 334690
    },
    {
      "epoch": 137.0,
      "eval_loss": 0.4531143009662628,
      "eval_runtime": 52.4201,
      "eval_samples_per_second": 65.795,
      "eval_steps_per_second": 8.241,
      "step": 334691
    },
    {
      "epoch": 137.0,
      "grad_norm": 2.1515965461730957,
      "learning_rate": 1.1605825289411498e-06,
      "loss": 0.4721,
      "step": 334700
    },
    {
      "epoch": 137.01,
      "grad_norm": 2.3623383045196533,
      "learning_rate": 1.1604455985425144e-06,
      "loss": 0.4374,
      "step": 334710
    },
    {
      "epoch": 137.01,
      "grad_norm": 1.8349981307983398,
      "learning_rate": 1.1603086735644134e-06,
      "loss": 0.4406,
      "step": 334720
    },
    {
      "epoch": 137.02,
      "grad_norm": 2.0726099014282227,
      "learning_rate": 1.1601717540074733e-06,
      "loss": 0.4558,
      "step": 334730
    },
    {
      "epoch": 137.02,
      "grad_norm": 1.7223254442214966,
      "learning_rate": 1.1600348398723221e-06,
      "loss": 0.4425,
      "step": 334740
    },
    {
      "epoch": 137.02,
      "grad_norm": 1.9438966512680054,
      "learning_rate": 1.1598979311595866e-06,
      "loss": 0.4751,
      "step": 334750
    },
    {
      "epoch": 137.03,
      "grad_norm": 2.3533849716186523,
      "learning_rate": 1.1597610278698944e-06,
      "loss": 0.4405,
      "step": 334760
    },
    {
      "epoch": 137.03,
      "grad_norm": 1.914304494857788,
      "learning_rate": 1.159624130003872e-06,
      "loss": 0.4493,
      "step": 334770
    },
    {
      "epoch": 137.04,
      "grad_norm": 2.3070871829986572,
      "learning_rate": 1.1594872375621472e-06,
      "loss": 0.4584,
      "step": 334780
    },
    {
      "epoch": 137.04,
      "grad_norm": 1.794335961341858,
      "learning_rate": 1.1593503505453462e-06,
      "loss": 0.4608,
      "step": 334790
    },
    {
      "epoch": 137.04,
      "grad_norm": 1.8713352680206299,
      "learning_rate": 1.159213468954097e-06,
      "loss": 0.456,
      "step": 334800
    },
    {
      "epoch": 137.05,
      "grad_norm": 1.9306825399398804,
      "learning_rate": 1.159076592789026e-06,
      "loss": 0.4538,
      "step": 334810
    },
    {
      "epoch": 137.05,
      "grad_norm": 2.7098124027252197,
      "learning_rate": 1.1589397220507605e-06,
      "loss": 0.4449,
      "step": 334820
    },
    {
      "epoch": 137.06,
      "grad_norm": 1.9823179244995117,
      "learning_rate": 1.1588028567399273e-06,
      "loss": 0.4498,
      "step": 334830
    },
    {
      "epoch": 137.06,
      "grad_norm": 1.794911503791809,
      "learning_rate": 1.1586659968571536e-06,
      "loss": 0.4486,
      "step": 334840
    },
    {
      "epoch": 137.07,
      "grad_norm": 1.853804111480713,
      "learning_rate": 1.158529142403066e-06,
      "loss": 0.4461,
      "step": 334850
    },
    {
      "epoch": 137.07,
      "grad_norm": 1.5936815738677979,
      "learning_rate": 1.1583922933782918e-06,
      "loss": 0.4844,
      "step": 334860
    },
    {
      "epoch": 137.07,
      "grad_norm": 1.9413726329803467,
      "learning_rate": 1.1582554497834575e-06,
      "loss": 0.4572,
      "step": 334870
    },
    {
      "epoch": 137.08,
      "grad_norm": 2.1465985774993896,
      "learning_rate": 1.15811861161919e-06,
      "loss": 0.46,
      "step": 334880
    },
    {
      "epoch": 137.08,
      "grad_norm": 1.813629388809204,
      "learning_rate": 1.1579817788861165e-06,
      "loss": 0.4325,
      "step": 334890
    },
    {
      "epoch": 137.09,
      "grad_norm": 1.9225549697875977,
      "learning_rate": 1.1578449515848638e-06,
      "loss": 0.457,
      "step": 334900
    },
    {
      "epoch": 137.09,
      "grad_norm": 1.813565731048584,
      "learning_rate": 1.1577081297160578e-06,
      "loss": 0.4591,
      "step": 334910
    },
    {
      "epoch": 137.09,
      "grad_norm": 1.9301822185516357,
      "learning_rate": 1.1575713132803264e-06,
      "loss": 0.4679,
      "step": 334920
    },
    {
      "epoch": 137.1,
      "grad_norm": 2.321733236312866,
      "learning_rate": 1.1574345022782968e-06,
      "loss": 0.4309,
      "step": 334930
    },
    {
      "epoch": 137.1,
      "grad_norm": 2.179738759994507,
      "learning_rate": 1.1572976967105938e-06,
      "loss": 0.4436,
      "step": 334940
    },
    {
      "epoch": 137.11,
      "grad_norm": 1.7850146293640137,
      "learning_rate": 1.1571608965778452e-06,
      "loss": 0.4591,
      "step": 334950
    },
    {
      "epoch": 137.11,
      "grad_norm": 2.0479812622070312,
      "learning_rate": 1.1570241018806773e-06,
      "loss": 0.4389,
      "step": 334960
    },
    {
      "epoch": 137.11,
      "grad_norm": 1.854752540588379,
      "learning_rate": 1.1568873126197169e-06,
      "loss": 0.45,
      "step": 334970
    },
    {
      "epoch": 137.12,
      "grad_norm": 1.9211763143539429,
      "learning_rate": 1.1567505287955899e-06,
      "loss": 0.4624,
      "step": 334980
    },
    {
      "epoch": 137.12,
      "grad_norm": 1.7631635665893555,
      "learning_rate": 1.1566137504089244e-06,
      "loss": 0.4545,
      "step": 334990
    },
    {
      "epoch": 137.13,
      "grad_norm": 2.0959901809692383,
      "learning_rate": 1.1564769774603462e-06,
      "loss": 0.443,
      "step": 335000
    },
    {
      "epoch": 137.13,
      "grad_norm": 2.1111137866973877,
      "learning_rate": 1.156340209950482e-06,
      "loss": 0.4286,
      "step": 335010
    },
    {
      "epoch": 137.13,
      "grad_norm": 1.9869115352630615,
      "learning_rate": 1.1562034478799581e-06,
      "loss": 0.4332,
      "step": 335020
    },
    {
      "epoch": 137.14,
      "grad_norm": 1.873052716255188,
      "learning_rate": 1.156066691249401e-06,
      "loss": 0.4495,
      "step": 335030
    },
    {
      "epoch": 137.14,
      "grad_norm": 2.317983388900757,
      "learning_rate": 1.1559299400594376e-06,
      "loss": 0.4618,
      "step": 335040
    },
    {
      "epoch": 137.15,
      "grad_norm": 2.081461191177368,
      "learning_rate": 1.1557931943106933e-06,
      "loss": 0.4436,
      "step": 335050
    },
    {
      "epoch": 137.15,
      "grad_norm": 1.8926382064819336,
      "learning_rate": 1.1556564540037964e-06,
      "loss": 0.4481,
      "step": 335060
    },
    {
      "epoch": 137.16,
      "grad_norm": 2.263859510421753,
      "learning_rate": 1.1555197191393709e-06,
      "loss": 0.4361,
      "step": 335070
    },
    {
      "epoch": 137.16,
      "grad_norm": 1.8603944778442383,
      "learning_rate": 1.1553829897180444e-06,
      "loss": 0.4589,
      "step": 335080
    },
    {
      "epoch": 137.16,
      "grad_norm": 1.8528087139129639,
      "learning_rate": 1.1552462657404436e-06,
      "loss": 0.4404,
      "step": 335090
    },
    {
      "epoch": 137.17,
      "grad_norm": 1.7361786365509033,
      "learning_rate": 1.1551095472071938e-06,
      "loss": 0.4556,
      "step": 335100
    },
    {
      "epoch": 137.17,
      "grad_norm": 1.9164949655532837,
      "learning_rate": 1.154972834118922e-06,
      "loss": 0.4466,
      "step": 335110
    },
    {
      "epoch": 137.18,
      "grad_norm": 1.8827027082443237,
      "learning_rate": 1.1548361264762541e-06,
      "loss": 0.4623,
      "step": 335120
    },
    {
      "epoch": 137.18,
      "grad_norm": 2.4541099071502686,
      "learning_rate": 1.1546994242798169e-06,
      "loss": 0.4482,
      "step": 335130
    },
    {
      "epoch": 137.18,
      "grad_norm": 1.8551822900772095,
      "learning_rate": 1.154562727530236e-06,
      "loss": 0.4622,
      "step": 335140
    },
    {
      "epoch": 137.19,
      "grad_norm": 2.061281442642212,
      "learning_rate": 1.1544260362281379e-06,
      "loss": 0.4584,
      "step": 335150
    },
    {
      "epoch": 137.19,
      "grad_norm": 2.053359031677246,
      "learning_rate": 1.1542893503741484e-06,
      "loss": 0.4665,
      "step": 335160
    },
    {
      "epoch": 137.2,
      "grad_norm": 1.942893624305725,
      "learning_rate": 1.1541526699688935e-06,
      "loss": 0.4446,
      "step": 335170
    },
    {
      "epoch": 137.2,
      "grad_norm": 1.9189577102661133,
      "learning_rate": 1.1540159950130002e-06,
      "loss": 0.4776,
      "step": 335180
    },
    {
      "epoch": 137.2,
      "grad_norm": 2.0193324089050293,
      "learning_rate": 1.1538793255070946e-06,
      "loss": 0.4688,
      "step": 335190
    },
    {
      "epoch": 137.21,
      "grad_norm": 2.0662405490875244,
      "learning_rate": 1.1537426614518028e-06,
      "loss": 0.4506,
      "step": 335200
    },
    {
      "epoch": 137.21,
      "grad_norm": 1.5764998197555542,
      "learning_rate": 1.1536060028477496e-06,
      "loss": 0.4406,
      "step": 335210
    },
    {
      "epoch": 137.22,
      "grad_norm": 1.6916100978851318,
      "learning_rate": 1.1534693496955616e-06,
      "loss": 0.443,
      "step": 335220
    },
    {
      "epoch": 137.22,
      "grad_norm": 2.097532272338867,
      "learning_rate": 1.153332701995865e-06,
      "loss": 0.4523,
      "step": 335230
    },
    {
      "epoch": 137.22,
      "grad_norm": 2.105011463165283,
      "learning_rate": 1.1531960597492858e-06,
      "loss": 0.4485,
      "step": 335240
    },
    {
      "epoch": 137.23,
      "grad_norm": 2.106156349182129,
      "learning_rate": 1.1530594229564495e-06,
      "loss": 0.4585,
      "step": 335250
    },
    {
      "epoch": 137.23,
      "grad_norm": 2.0991711616516113,
      "learning_rate": 1.1529227916179823e-06,
      "loss": 0.4388,
      "step": 335260
    },
    {
      "epoch": 137.24,
      "grad_norm": 1.778006672859192,
      "learning_rate": 1.1527861657345107e-06,
      "loss": 0.4498,
      "step": 335270
    },
    {
      "epoch": 137.24,
      "grad_norm": 1.6934384107589722,
      "learning_rate": 1.1526495453066596e-06,
      "loss": 0.4512,
      "step": 335280
    },
    {
      "epoch": 137.25,
      "grad_norm": 2.1838338375091553,
      "learning_rate": 1.1525129303350552e-06,
      "loss": 0.4497,
      "step": 335290
    },
    {
      "epoch": 137.25,
      "grad_norm": 2.0791261196136475,
      "learning_rate": 1.1523763208203233e-06,
      "loss": 0.4415,
      "step": 335300
    },
    {
      "epoch": 137.25,
      "grad_norm": 2.0570156574249268,
      "learning_rate": 1.1522397167630897e-06,
      "loss": 0.4539,
      "step": 335310
    },
    {
      "epoch": 137.26,
      "grad_norm": 2.3799612522125244,
      "learning_rate": 1.1521031181639803e-06,
      "loss": 0.4483,
      "step": 335320
    },
    {
      "epoch": 137.26,
      "grad_norm": 2.09308123588562,
      "learning_rate": 1.1519665250236207e-06,
      "loss": 0.4592,
      "step": 335330
    },
    {
      "epoch": 137.27,
      "grad_norm": 1.966192603111267,
      "learning_rate": 1.1518299373426367e-06,
      "loss": 0.4369,
      "step": 335340
    },
    {
      "epoch": 137.27,
      "grad_norm": 1.9866092205047607,
      "learning_rate": 1.1516933551216537e-06,
      "loss": 0.4508,
      "step": 335350
    },
    {
      "epoch": 137.27,
      "grad_norm": 2.2142202854156494,
      "learning_rate": 1.151556778361298e-06,
      "loss": 0.4456,
      "step": 335360
    },
    {
      "epoch": 137.28,
      "grad_norm": 2.4356038570404053,
      "learning_rate": 1.1514202070621943e-06,
      "loss": 0.4493,
      "step": 335370
    },
    {
      "epoch": 137.28,
      "grad_norm": 2.214398145675659,
      "learning_rate": 1.151283641224969e-06,
      "loss": 0.4386,
      "step": 335380
    },
    {
      "epoch": 137.29,
      "grad_norm": 1.827229619026184,
      "learning_rate": 1.1511470808502472e-06,
      "loss": 0.4384,
      "step": 335390
    },
    {
      "epoch": 137.29,
      "grad_norm": 1.5659925937652588,
      "learning_rate": 1.151010525938655e-06,
      "loss": 0.4381,
      "step": 335400
    },
    {
      "epoch": 137.29,
      "grad_norm": 1.786442756652832,
      "learning_rate": 1.1508739764908174e-06,
      "loss": 0.4534,
      "step": 335410
    },
    {
      "epoch": 137.3,
      "grad_norm": 1.8740533590316772,
      "learning_rate": 1.1507374325073601e-06,
      "loss": 0.449,
      "step": 335420
    },
    {
      "epoch": 137.3,
      "grad_norm": 2.0701372623443604,
      "learning_rate": 1.1506008939889086e-06,
      "loss": 0.4544,
      "step": 335430
    },
    {
      "epoch": 137.31,
      "grad_norm": 2.1578359603881836,
      "learning_rate": 1.1504643609360886e-06,
      "loss": 0.4345,
      "step": 335440
    },
    {
      "epoch": 137.31,
      "grad_norm": 1.7486917972564697,
      "learning_rate": 1.1503278333495253e-06,
      "loss": 0.4693,
      "step": 335450
    },
    {
      "epoch": 137.31,
      "grad_norm": 2.1340439319610596,
      "learning_rate": 1.1501913112298438e-06,
      "loss": 0.4478,
      "step": 335460
    },
    {
      "epoch": 137.32,
      "grad_norm": 1.9240424633026123,
      "learning_rate": 1.15005479457767e-06,
      "loss": 0.4523,
      "step": 335470
    },
    {
      "epoch": 137.32,
      "grad_norm": 2.0058560371398926,
      "learning_rate": 1.149918283393629e-06,
      "loss": 0.4548,
      "step": 335480
    },
    {
      "epoch": 137.33,
      "grad_norm": 1.816243052482605,
      "learning_rate": 1.1497817776783464e-06,
      "loss": 0.4542,
      "step": 335490
    },
    {
      "epoch": 137.33,
      "grad_norm": 2.4719526767730713,
      "learning_rate": 1.1496452774324472e-06,
      "loss": 0.4664,
      "step": 335500
    },
    {
      "epoch": 137.34,
      "grad_norm": 2.2292683124542236,
      "learning_rate": 1.1495087826565575e-06,
      "loss": 0.4471,
      "step": 335510
    },
    {
      "epoch": 137.34,
      "grad_norm": 2.1113226413726807,
      "learning_rate": 1.1493722933513013e-06,
      "loss": 0.4645,
      "step": 335520
    },
    {
      "epoch": 137.34,
      "grad_norm": 1.8789629936218262,
      "learning_rate": 1.1492358095173043e-06,
      "loss": 0.4492,
      "step": 335530
    },
    {
      "epoch": 137.35,
      "grad_norm": 1.7555537223815918,
      "learning_rate": 1.1490993311551916e-06,
      "loss": 0.4775,
      "step": 335540
    },
    {
      "epoch": 137.35,
      "grad_norm": 1.754663348197937,
      "learning_rate": 1.148962858265588e-06,
      "loss": 0.4625,
      "step": 335550
    },
    {
      "epoch": 137.36,
      "grad_norm": 2.358428955078125,
      "learning_rate": 1.14882639084912e-06,
      "loss": 0.4726,
      "step": 335560
    },
    {
      "epoch": 137.36,
      "grad_norm": 2.3094139099121094,
      "learning_rate": 1.1486899289064119e-06,
      "loss": 0.439,
      "step": 335570
    },
    {
      "epoch": 137.36,
      "grad_norm": 1.9065098762512207,
      "learning_rate": 1.1485534724380891e-06,
      "loss": 0.4507,
      "step": 335580
    },
    {
      "epoch": 137.37,
      "grad_norm": 2.0120692253112793,
      "learning_rate": 1.1484170214447765e-06,
      "loss": 0.4719,
      "step": 335590
    },
    {
      "epoch": 137.37,
      "grad_norm": 2.04976487159729,
      "learning_rate": 1.148280575927099e-06,
      "loss": 0.4431,
      "step": 335600
    },
    {
      "epoch": 137.38,
      "grad_norm": 1.8515863418579102,
      "learning_rate": 1.148144135885682e-06,
      "loss": 0.4411,
      "step": 335610
    },
    {
      "epoch": 137.38,
      "grad_norm": 2.1433136463165283,
      "learning_rate": 1.14800770132115e-06,
      "loss": 0.457,
      "step": 335620
    },
    {
      "epoch": 137.38,
      "grad_norm": 2.136812925338745,
      "learning_rate": 1.1478712722341286e-06,
      "loss": 0.4626,
      "step": 335630
    },
    {
      "epoch": 137.39,
      "grad_norm": 2.7270123958587646,
      "learning_rate": 1.147734848625242e-06,
      "loss": 0.475,
      "step": 335640
    },
    {
      "epoch": 137.39,
      "grad_norm": 1.8741700649261475,
      "learning_rate": 1.1475984304951169e-06,
      "loss": 0.4424,
      "step": 335650
    },
    {
      "epoch": 137.4,
      "grad_norm": 1.964728593826294,
      "learning_rate": 1.147462017844376e-06,
      "loss": 0.4478,
      "step": 335660
    },
    {
      "epoch": 137.4,
      "grad_norm": 1.8805320262908936,
      "learning_rate": 1.147325610673645e-06,
      "loss": 0.441,
      "step": 335670
    },
    {
      "epoch": 137.4,
      "grad_norm": 2.1994943618774414,
      "learning_rate": 1.1471892089835486e-06,
      "loss": 0.4569,
      "step": 335680
    },
    {
      "epoch": 137.41,
      "grad_norm": 2.0495765209198,
      "learning_rate": 1.1470528127747121e-06,
      "loss": 0.4504,
      "step": 335690
    },
    {
      "epoch": 137.41,
      "grad_norm": 2.171015501022339,
      "learning_rate": 1.14691642204776e-06,
      "loss": 0.4665,
      "step": 335700
    },
    {
      "epoch": 137.42,
      "grad_norm": 2.207720994949341,
      "learning_rate": 1.1467800368033173e-06,
      "loss": 0.4614,
      "step": 335710
    },
    {
      "epoch": 137.42,
      "grad_norm": 2.2070300579071045,
      "learning_rate": 1.1466436570420084e-06,
      "loss": 0.4486,
      "step": 335720
    },
    {
      "epoch": 137.43,
      "grad_norm": 1.7671525478363037,
      "learning_rate": 1.1465072827644586e-06,
      "loss": 0.4607,
      "step": 335730
    },
    {
      "epoch": 137.43,
      "grad_norm": 2.334714889526367,
      "learning_rate": 1.146370913971291e-06,
      "loss": 0.4703,
      "step": 335740
    },
    {
      "epoch": 137.43,
      "grad_norm": 2.128333806991577,
      "learning_rate": 1.1462345506631327e-06,
      "loss": 0.4593,
      "step": 335750
    },
    {
      "epoch": 137.44,
      "grad_norm": 1.6917998790740967,
      "learning_rate": 1.1460981928406073e-06,
      "loss": 0.4555,
      "step": 335760
    },
    {
      "epoch": 137.44,
      "grad_norm": 2.078674793243408,
      "learning_rate": 1.145961840504339e-06,
      "loss": 0.4387,
      "step": 335770
    },
    {
      "epoch": 137.45,
      "grad_norm": 1.8511404991149902,
      "learning_rate": 1.145825493654954e-06,
      "loss": 0.4374,
      "step": 335780
    },
    {
      "epoch": 137.45,
      "grad_norm": 1.8387491703033447,
      "learning_rate": 1.1456891522930746e-06,
      "loss": 0.4412,
      "step": 335790
    },
    {
      "epoch": 137.45,
      "grad_norm": 1.9243812561035156,
      "learning_rate": 1.1455528164193262e-06,
      "loss": 0.4473,
      "step": 335800
    },
    {
      "epoch": 137.46,
      "grad_norm": 2.0756943225860596,
      "learning_rate": 1.1454164860343338e-06,
      "loss": 0.4557,
      "step": 335810
    },
    {
      "epoch": 137.46,
      "grad_norm": 1.798231840133667,
      "learning_rate": 1.1452801611387215e-06,
      "loss": 0.4546,
      "step": 335820
    },
    {
      "epoch": 137.47,
      "grad_norm": 1.9242910146713257,
      "learning_rate": 1.1451438417331142e-06,
      "loss": 0.4371,
      "step": 335830
    },
    {
      "epoch": 137.47,
      "grad_norm": 1.897844910621643,
      "learning_rate": 1.1450075278181357e-06,
      "loss": 0.4404,
      "step": 335840
    },
    {
      "epoch": 137.47,
      "grad_norm": 1.8113728761672974,
      "learning_rate": 1.1448712193944109e-06,
      "loss": 0.4364,
      "step": 335850
    },
    {
      "epoch": 137.48,
      "grad_norm": 2.0621001720428467,
      "learning_rate": 1.1447349164625641e-06,
      "loss": 0.457,
      "step": 335860
    },
    {
      "epoch": 137.48,
      "grad_norm": 2.082278251647949,
      "learning_rate": 1.14459861902322e-06,
      "loss": 0.45,
      "step": 335870
    },
    {
      "epoch": 137.49,
      "grad_norm": 1.6181162595748901,
      "learning_rate": 1.1444623270770025e-06,
      "loss": 0.4561,
      "step": 335880
    },
    {
      "epoch": 137.49,
      "grad_norm": 1.9144142866134644,
      "learning_rate": 1.144326040624536e-06,
      "loss": 0.4639,
      "step": 335890
    },
    {
      "epoch": 137.49,
      "grad_norm": 2.255403995513916,
      "learning_rate": 1.1441897596664449e-06,
      "loss": 0.4591,
      "step": 335900
    },
    {
      "epoch": 137.5,
      "grad_norm": 1.8804899454116821,
      "learning_rate": 1.1440534842033538e-06,
      "loss": 0.4606,
      "step": 335910
    },
    {
      "epoch": 137.5,
      "grad_norm": 1.6704977750778198,
      "learning_rate": 1.1439172142358865e-06,
      "loss": 0.4562,
      "step": 335920
    },
    {
      "epoch": 137.51,
      "grad_norm": 1.8638403415679932,
      "learning_rate": 1.143780949764667e-06,
      "loss": 0.4753,
      "step": 335930
    },
    {
      "epoch": 137.51,
      "grad_norm": 2.0184409618377686,
      "learning_rate": 1.1436446907903203e-06,
      "loss": 0.461,
      "step": 335940
    },
    {
      "epoch": 137.52,
      "grad_norm": 2.2010886669158936,
      "learning_rate": 1.1435084373134701e-06,
      "loss": 0.4544,
      "step": 335950
    },
    {
      "epoch": 137.52,
      "grad_norm": 1.6657954454421997,
      "learning_rate": 1.1433721893347406e-06,
      "loss": 0.4468,
      "step": 335960
    },
    {
      "epoch": 137.52,
      "grad_norm": 4.951493740081787,
      "learning_rate": 1.1432359468547558e-06,
      "loss": 0.4675,
      "step": 335970
    },
    {
      "epoch": 137.53,
      "grad_norm": 2.056976318359375,
      "learning_rate": 1.1430997098741401e-06,
      "loss": 0.4458,
      "step": 335980
    },
    {
      "epoch": 137.53,
      "grad_norm": 2.29892897605896,
      "learning_rate": 1.1429634783935173e-06,
      "loss": 0.4584,
      "step": 335990
    },
    {
      "epoch": 137.54,
      "grad_norm": 2.9131581783294678,
      "learning_rate": 1.1428272524135118e-06,
      "loss": 0.4492,
      "step": 336000
    },
    {
      "epoch": 137.54,
      "grad_norm": 2.0614724159240723,
      "learning_rate": 1.1426910319347473e-06,
      "loss": 0.4418,
      "step": 336010
    },
    {
      "epoch": 137.54,
      "grad_norm": 2.346975326538086,
      "learning_rate": 1.142554816957848e-06,
      "loss": 0.4786,
      "step": 336020
    },
    {
      "epoch": 137.55,
      "grad_norm": 2.0465190410614014,
      "learning_rate": 1.1424186074834374e-06,
      "loss": 0.4417,
      "step": 336030
    },
    {
      "epoch": 137.55,
      "grad_norm": 1.9428298473358154,
      "learning_rate": 1.1422824035121403e-06,
      "loss": 0.4643,
      "step": 336040
    },
    {
      "epoch": 137.56,
      "grad_norm": 1.8517974615097046,
      "learning_rate": 1.14214620504458e-06,
      "loss": 0.4536,
      "step": 336050
    },
    {
      "epoch": 137.56,
      "grad_norm": 2.0986642837524414,
      "learning_rate": 1.1420100120813807e-06,
      "loss": 0.466,
      "step": 336060
    },
    {
      "epoch": 137.56,
      "grad_norm": 1.7033905982971191,
      "learning_rate": 1.141873824623166e-06,
      "loss": 0.443,
      "step": 336070
    },
    {
      "epoch": 137.57,
      "grad_norm": 1.880423903465271,
      "learning_rate": 1.1417376426705604e-06,
      "loss": 0.4511,
      "step": 336080
    },
    {
      "epoch": 137.57,
      "grad_norm": 1.7360066175460815,
      "learning_rate": 1.1416014662241878e-06,
      "loss": 0.4657,
      "step": 336090
    },
    {
      "epoch": 137.58,
      "grad_norm": 2.0016720294952393,
      "learning_rate": 1.1414652952846706e-06,
      "loss": 0.4493,
      "step": 336100
    },
    {
      "epoch": 137.58,
      "grad_norm": 2.294597625732422,
      "learning_rate": 1.1413291298526326e-06,
      "loss": 0.4574,
      "step": 336110
    },
    {
      "epoch": 137.58,
      "grad_norm": 1.797745943069458,
      "learning_rate": 1.141192969928699e-06,
      "loss": 0.469,
      "step": 336120
    },
    {
      "epoch": 137.59,
      "grad_norm": 2.086920976638794,
      "learning_rate": 1.141056815513493e-06,
      "loss": 0.4851,
      "step": 336130
    },
    {
      "epoch": 137.59,
      "grad_norm": 1.6509796380996704,
      "learning_rate": 1.1409206666076383e-06,
      "loss": 0.4555,
      "step": 336140
    },
    {
      "epoch": 137.6,
      "grad_norm": 2.571835994720459,
      "learning_rate": 1.1407845232117587e-06,
      "loss": 0.4541,
      "step": 336150
    },
    {
      "epoch": 137.6,
      "grad_norm": 1.8751713037490845,
      "learning_rate": 1.1406483853264776e-06,
      "loss": 0.4382,
      "step": 336160
    },
    {
      "epoch": 137.61,
      "grad_norm": 1.97611665725708,
      "learning_rate": 1.1405122529524186e-06,
      "loss": 0.4786,
      "step": 336170
    },
    {
      "epoch": 137.61,
      "grad_norm": 1.5754057168960571,
      "learning_rate": 1.1403761260902051e-06,
      "loss": 0.446,
      "step": 336180
    },
    {
      "epoch": 137.61,
      "grad_norm": 1.7379974126815796,
      "learning_rate": 1.1402400047404613e-06,
      "loss": 0.4644,
      "step": 336190
    },
    {
      "epoch": 137.62,
      "grad_norm": 1.8399615287780762,
      "learning_rate": 1.1401038889038102e-06,
      "loss": 0.4646,
      "step": 336200
    },
    {
      "epoch": 137.62,
      "grad_norm": 2.3238086700439453,
      "learning_rate": 1.1399677785808756e-06,
      "loss": 0.4558,
      "step": 336210
    },
    {
      "epoch": 137.63,
      "grad_norm": 2.399264335632324,
      "learning_rate": 1.139831673772281e-06,
      "loss": 0.4668,
      "step": 336220
    },
    {
      "epoch": 137.63,
      "grad_norm": 1.829632043838501,
      "learning_rate": 1.1396955744786506e-06,
      "loss": 0.4385,
      "step": 336230
    },
    {
      "epoch": 137.63,
      "grad_norm": 2.2552034854888916,
      "learning_rate": 1.1395594807006062e-06,
      "loss": 0.4464,
      "step": 336240
    },
    {
      "epoch": 137.64,
      "grad_norm": 2.0796303749084473,
      "learning_rate": 1.139423392438772e-06,
      "loss": 0.4533,
      "step": 336250
    },
    {
      "epoch": 137.64,
      "grad_norm": 2.856076240539551,
      "learning_rate": 1.1392873096937715e-06,
      "loss": 0.4642,
      "step": 336260
    },
    {
      "epoch": 137.65,
      "grad_norm": 1.895997405052185,
      "learning_rate": 1.139151232466228e-06,
      "loss": 0.4518,
      "step": 336270
    },
    {
      "epoch": 137.65,
      "grad_norm": 1.853812575340271,
      "learning_rate": 1.139015160756765e-06,
      "loss": 0.4407,
      "step": 336280
    },
    {
      "epoch": 137.65,
      "grad_norm": 1.9774271249771118,
      "learning_rate": 1.1388790945660058e-06,
      "loss": 0.4539,
      "step": 336290
    },
    {
      "epoch": 137.66,
      "grad_norm": 1.5540308952331543,
      "learning_rate": 1.1387430338945726e-06,
      "loss": 0.4409,
      "step": 336300
    },
    {
      "epoch": 137.66,
      "grad_norm": 1.6940257549285889,
      "learning_rate": 1.1386069787430907e-06,
      "loss": 0.4606,
      "step": 336310
    },
    {
      "epoch": 137.67,
      "grad_norm": 1.7265650033950806,
      "learning_rate": 1.1384709291121822e-06,
      "loss": 0.4459,
      "step": 336320
    },
    {
      "epoch": 137.67,
      "grad_norm": 2.0957322120666504,
      "learning_rate": 1.1383348850024706e-06,
      "loss": 0.4557,
      "step": 336330
    },
    {
      "epoch": 137.67,
      "grad_norm": 2.011324644088745,
      "learning_rate": 1.1381988464145788e-06,
      "loss": 0.4488,
      "step": 336340
    },
    {
      "epoch": 137.68,
      "grad_norm": 1.8961986303329468,
      "learning_rate": 1.1380628133491303e-06,
      "loss": 0.4472,
      "step": 336350
    },
    {
      "epoch": 137.68,
      "grad_norm": 2.140150547027588,
      "learning_rate": 1.1379267858067487e-06,
      "loss": 0.4573,
      "step": 336360
    },
    {
      "epoch": 137.69,
      "grad_norm": 2.266986846923828,
      "learning_rate": 1.137790763788056e-06,
      "loss": 0.4516,
      "step": 336370
    },
    {
      "epoch": 137.69,
      "grad_norm": 2.1653060913085938,
      "learning_rate": 1.1376547472936756e-06,
      "loss": 0.4578,
      "step": 336380
    },
    {
      "epoch": 137.7,
      "grad_norm": 2.0217816829681396,
      "learning_rate": 1.1375187363242308e-06,
      "loss": 0.4476,
      "step": 336390
    },
    {
      "epoch": 137.7,
      "grad_norm": 1.9000331163406372,
      "learning_rate": 1.1373827308803446e-06,
      "loss": 0.4686,
      "step": 336400
    },
    {
      "epoch": 137.7,
      "grad_norm": 1.9220945835113525,
      "learning_rate": 1.1372467309626398e-06,
      "loss": 0.4745,
      "step": 336410
    },
    {
      "epoch": 137.71,
      "grad_norm": 2.239729881286621,
      "learning_rate": 1.13711073657174e-06,
      "loss": 0.4588,
      "step": 336420
    },
    {
      "epoch": 137.71,
      "grad_norm": 1.8502593040466309,
      "learning_rate": 1.1369747477082674e-06,
      "loss": 0.4608,
      "step": 336430
    },
    {
      "epoch": 137.72,
      "grad_norm": 1.909279704093933,
      "learning_rate": 1.1368387643728456e-06,
      "loss": 0.4458,
      "step": 336440
    },
    {
      "epoch": 137.72,
      "grad_norm": 2.431509494781494,
      "learning_rate": 1.1367027865660973e-06,
      "loss": 0.4515,
      "step": 336450
    },
    {
      "epoch": 137.72,
      "grad_norm": 2.104890823364258,
      "learning_rate": 1.1365668142886453e-06,
      "loss": 0.4525,
      "step": 336460
    },
    {
      "epoch": 137.73,
      "grad_norm": 1.8218053579330444,
      "learning_rate": 1.1364308475411122e-06,
      "loss": 0.4464,
      "step": 336470
    },
    {
      "epoch": 137.73,
      "grad_norm": 2.2097063064575195,
      "learning_rate": 1.1362948863241215e-06,
      "loss": 0.4563,
      "step": 336480
    },
    {
      "epoch": 137.74,
      "grad_norm": 1.9703752994537354,
      "learning_rate": 1.1361589306382947e-06,
      "loss": 0.4452,
      "step": 336490
    },
    {
      "epoch": 137.74,
      "grad_norm": 1.845258116722107,
      "learning_rate": 1.1360229804842574e-06,
      "loss": 0.4451,
      "step": 336500
    },
    {
      "epoch": 137.74,
      "grad_norm": 1.8122178316116333,
      "learning_rate": 1.1358870358626295e-06,
      "loss": 0.4675,
      "step": 336510
    },
    {
      "epoch": 137.75,
      "grad_norm": 1.9722434282302856,
      "learning_rate": 1.1357510967740348e-06,
      "loss": 0.4561,
      "step": 336520
    },
    {
      "epoch": 137.75,
      "grad_norm": 1.970859169960022,
      "learning_rate": 1.1356151632190963e-06,
      "loss": 0.4419,
      "step": 336530
    },
    {
      "epoch": 137.76,
      "grad_norm": 1.9724311828613281,
      "learning_rate": 1.135479235198436e-06,
      "loss": 0.469,
      "step": 336540
    },
    {
      "epoch": 137.76,
      "grad_norm": 1.9220144748687744,
      "learning_rate": 1.1353433127126772e-06,
      "loss": 0.4637,
      "step": 336550
    },
    {
      "epoch": 137.77,
      "grad_norm": 2.1177189350128174,
      "learning_rate": 1.1352073957624423e-06,
      "loss": 0.4662,
      "step": 336560
    },
    {
      "epoch": 137.77,
      "grad_norm": 1.8797235488891602,
      "learning_rate": 1.1350714843483538e-06,
      "loss": 0.4634,
      "step": 336570
    },
    {
      "epoch": 137.77,
      "grad_norm": 1.6751137971878052,
      "learning_rate": 1.1349355784710346e-06,
      "loss": 0.4501,
      "step": 336580
    },
    {
      "epoch": 137.78,
      "grad_norm": 1.934690237045288,
      "learning_rate": 1.1347996781311067e-06,
      "loss": 0.458,
      "step": 336590
    },
    {
      "epoch": 137.78,
      "grad_norm": 2.752512216567993,
      "learning_rate": 1.1346637833291931e-06,
      "loss": 0.4627,
      "step": 336600
    },
    {
      "epoch": 137.79,
      "grad_norm": 2.1314163208007812,
      "learning_rate": 1.1345278940659164e-06,
      "loss": 0.4623,
      "step": 336610
    },
    {
      "epoch": 137.79,
      "grad_norm": 1.9916046857833862,
      "learning_rate": 1.1343920103418986e-06,
      "loss": 0.4484,
      "step": 336620
    },
    {
      "epoch": 137.79,
      "grad_norm": 2.1058120727539062,
      "learning_rate": 1.1342561321577626e-06,
      "loss": 0.4472,
      "step": 336630
    },
    {
      "epoch": 137.8,
      "grad_norm": 1.8496496677398682,
      "learning_rate": 1.1341202595141308e-06,
      "loss": 0.4634,
      "step": 336640
    },
    {
      "epoch": 137.8,
      "grad_norm": 1.9052870273590088,
      "learning_rate": 1.1339843924116254e-06,
      "loss": 0.4485,
      "step": 336650
    },
    {
      "epoch": 137.81,
      "grad_norm": 2.1891674995422363,
      "learning_rate": 1.1338485308508689e-06,
      "loss": 0.4253,
      "step": 336660
    },
    {
      "epoch": 137.81,
      "grad_norm": 2.0565130710601807,
      "learning_rate": 1.1337126748324846e-06,
      "loss": 0.4621,
      "step": 336670
    },
    {
      "epoch": 137.81,
      "grad_norm": 2.0441980361938477,
      "learning_rate": 1.1335768243570922e-06,
      "loss": 0.4612,
      "step": 336680
    },
    {
      "epoch": 137.82,
      "grad_norm": 2.1615657806396484,
      "learning_rate": 1.1334409794253163e-06,
      "loss": 0.4714,
      "step": 336690
    },
    {
      "epoch": 137.82,
      "grad_norm": 1.695982813835144,
      "learning_rate": 1.1333051400377788e-06,
      "loss": 0.4375,
      "step": 336700
    },
    {
      "epoch": 137.83,
      "grad_norm": 2.2079384326934814,
      "learning_rate": 1.1331693061951016e-06,
      "loss": 0.4482,
      "step": 336710
    },
    {
      "epoch": 137.83,
      "grad_norm": 1.8136836290359497,
      "learning_rate": 1.1330334778979072e-06,
      "loss": 0.4544,
      "step": 336720
    },
    {
      "epoch": 137.83,
      "grad_norm": 1.8925580978393555,
      "learning_rate": 1.1328976551468175e-06,
      "loss": 0.4278,
      "step": 336730
    },
    {
      "epoch": 137.84,
      "grad_norm": 2.0745105743408203,
      "learning_rate": 1.1327618379424552e-06,
      "loss": 0.4646,
      "step": 336740
    },
    {
      "epoch": 137.84,
      "grad_norm": 1.9995832443237305,
      "learning_rate": 1.132626026285442e-06,
      "loss": 0.4291,
      "step": 336750
    },
    {
      "epoch": 137.85,
      "grad_norm": 2.034186363220215,
      "learning_rate": 1.1324902201764e-06,
      "loss": 0.4679,
      "step": 336760
    },
    {
      "epoch": 137.85,
      "grad_norm": 1.8181442022323608,
      "learning_rate": 1.1323544196159514e-06,
      "loss": 0.4732,
      "step": 336770
    },
    {
      "epoch": 137.86,
      "grad_norm": 2.2884387969970703,
      "learning_rate": 1.1322186246047186e-06,
      "loss": 0.4361,
      "step": 336780
    },
    {
      "epoch": 137.86,
      "grad_norm": 1.9967907667160034,
      "learning_rate": 1.1320828351433233e-06,
      "loss": 0.4514,
      "step": 336790
    },
    {
      "epoch": 137.86,
      "grad_norm": 2.1237213611602783,
      "learning_rate": 1.1319470512323875e-06,
      "loss": 0.447,
      "step": 336800
    },
    {
      "epoch": 137.87,
      "grad_norm": 1.938093662261963,
      "learning_rate": 1.1318112728725345e-06,
      "loss": 0.4632,
      "step": 336810
    },
    {
      "epoch": 137.87,
      "grad_norm": 2.1435792446136475,
      "learning_rate": 1.1316755000643842e-06,
      "loss": 0.4498,
      "step": 336820
    },
    {
      "epoch": 137.88,
      "grad_norm": 2.0862507820129395,
      "learning_rate": 1.1315397328085592e-06,
      "loss": 0.4519,
      "step": 336830
    },
    {
      "epoch": 137.88,
      "grad_norm": 1.9645296335220337,
      "learning_rate": 1.131403971105682e-06,
      "loss": 0.4522,
      "step": 336840
    },
    {
      "epoch": 137.88,
      "grad_norm": 1.8936842679977417,
      "learning_rate": 1.1312682149563739e-06,
      "loss": 0.4477,
      "step": 336850
    },
    {
      "epoch": 137.89,
      "grad_norm": 2.102529764175415,
      "learning_rate": 1.1311324643612573e-06,
      "loss": 0.4503,
      "step": 336860
    },
    {
      "epoch": 137.89,
      "grad_norm": 2.1239426136016846,
      "learning_rate": 1.130996719320953e-06,
      "loss": 0.4708,
      "step": 336870
    },
    {
      "epoch": 137.9,
      "grad_norm": 1.9604309797286987,
      "learning_rate": 1.1308609798360843e-06,
      "loss": 0.4452,
      "step": 336880
    },
    {
      "epoch": 137.9,
      "grad_norm": 2.015638828277588,
      "learning_rate": 1.1307252459072727e-06,
      "loss": 0.4506,
      "step": 336890
    },
    {
      "epoch": 137.9,
      "grad_norm": 1.9139283895492554,
      "learning_rate": 1.1305895175351391e-06,
      "loss": 0.4733,
      "step": 336900
    },
    {
      "epoch": 137.91,
      "grad_norm": 1.9461473226547241,
      "learning_rate": 1.1304537947203061e-06,
      "loss": 0.4427,
      "step": 336910
    },
    {
      "epoch": 137.91,
      "grad_norm": 3.213855504989624,
      "learning_rate": 1.1303180774633955e-06,
      "loss": 0.4459,
      "step": 336920
    },
    {
      "epoch": 137.92,
      "grad_norm": 1.6737308502197266,
      "learning_rate": 1.1301823657650282e-06,
      "loss": 0.4497,
      "step": 336930
    },
    {
      "epoch": 137.92,
      "grad_norm": 1.8167628049850464,
      "learning_rate": 1.1300466596258262e-06,
      "loss": 0.4749,
      "step": 336940
    },
    {
      "epoch": 137.92,
      "grad_norm": 1.9096347093582153,
      "learning_rate": 1.1299109590464121e-06,
      "loss": 0.478,
      "step": 336950
    },
    {
      "epoch": 137.93,
      "grad_norm": 2.759136438369751,
      "learning_rate": 1.129775264027406e-06,
      "loss": 0.4681,
      "step": 336960
    },
    {
      "epoch": 137.93,
      "grad_norm": 2.342789649963379,
      "learning_rate": 1.1296395745694301e-06,
      "loss": 0.4608,
      "step": 336970
    },
    {
      "epoch": 137.94,
      "grad_norm": 2.0390396118164062,
      "learning_rate": 1.1295038906731062e-06,
      "loss": 0.4682,
      "step": 336980
    },
    {
      "epoch": 137.94,
      "grad_norm": 2.2085249423980713,
      "learning_rate": 1.1293682123390555e-06,
      "loss": 0.4574,
      "step": 336990
    },
    {
      "epoch": 137.95,
      "grad_norm": 2.0516104698181152,
      "learning_rate": 1.1292325395678997e-06,
      "loss": 0.4738,
      "step": 337000
    },
    {
      "epoch": 137.95,
      "grad_norm": 1.8397440910339355,
      "learning_rate": 1.1290968723602603e-06,
      "loss": 0.464,
      "step": 337010
    },
    {
      "epoch": 137.95,
      "grad_norm": 2.075608730316162,
      "learning_rate": 1.1289612107167587e-06,
      "loss": 0.4455,
      "step": 337020
    },
    {
      "epoch": 137.96,
      "grad_norm": 1.8823190927505493,
      "learning_rate": 1.1288255546380166e-06,
      "loss": 0.4734,
      "step": 337030
    },
    {
      "epoch": 137.96,
      "grad_norm": 2.1272435188293457,
      "learning_rate": 1.128689904124655e-06,
      "loss": 0.4621,
      "step": 337040
    },
    {
      "epoch": 137.97,
      "grad_norm": 2.4794962406158447,
      "learning_rate": 1.1285542591772955e-06,
      "loss": 0.4437,
      "step": 337050
    },
    {
      "epoch": 137.97,
      "grad_norm": 2.365091323852539,
      "learning_rate": 1.1284186197965586e-06,
      "loss": 0.4301,
      "step": 337060
    },
    {
      "epoch": 137.97,
      "grad_norm": 1.7942917346954346,
      "learning_rate": 1.1282829859830678e-06,
      "loss": 0.4424,
      "step": 337070
    },
    {
      "epoch": 137.98,
      "grad_norm": 2.025653123855591,
      "learning_rate": 1.1281473577374438e-06,
      "loss": 0.451,
      "step": 337080
    },
    {
      "epoch": 137.98,
      "grad_norm": 2.1026744842529297,
      "learning_rate": 1.1280117350603064e-06,
      "loss": 0.4533,
      "step": 337090
    },
    {
      "epoch": 137.99,
      "grad_norm": 2.2446556091308594,
      "learning_rate": 1.1278761179522775e-06,
      "loss": 0.4506,
      "step": 337100
    },
    {
      "epoch": 137.99,
      "grad_norm": 2.2892720699310303,
      "learning_rate": 1.1277405064139787e-06,
      "loss": 0.428,
      "step": 337110
    },
    {
      "epoch": 137.99,
      "grad_norm": 1.9070971012115479,
      "learning_rate": 1.1276049004460312e-06,
      "loss": 0.4423,
      "step": 337120
    },
    {
      "epoch": 138.0,
      "grad_norm": 2.1400959491729736,
      "learning_rate": 1.127469300049056e-06,
      "loss": 0.4446,
      "step": 337130
    },
    {
      "epoch": 138.0,
      "eval_loss": 0.45392659306526184,
      "eval_runtime": 52.2404,
      "eval_samples_per_second": 66.022,
      "eval_steps_per_second": 8.269,
      "step": 337134
    },
    {
      "epoch": 138.0,
      "grad_norm": 1.7636231184005737,
      "learning_rate": 1.127333705223674e-06,
      "loss": 0.486,
      "step": 337140
    },
    {
      "epoch": 138.01,
      "grad_norm": 1.782181739807129,
      "learning_rate": 1.1271981159705069e-06,
      "loss": 0.4454,
      "step": 337150
    },
    {
      "epoch": 138.01,
      "grad_norm": 2.089400291442871,
      "learning_rate": 1.127062532290176e-06,
      "loss": 0.4547,
      "step": 337160
    },
    {
      "epoch": 138.01,
      "grad_norm": 1.7827272415161133,
      "learning_rate": 1.1269269541833012e-06,
      "loss": 0.4452,
      "step": 337170
    },
    {
      "epoch": 138.02,
      "grad_norm": 1.9588950872421265,
      "learning_rate": 1.126791381650505e-06,
      "loss": 0.4449,
      "step": 337180
    },
    {
      "epoch": 138.02,
      "grad_norm": 1.9806817770004272,
      "learning_rate": 1.1266558146924073e-06,
      "loss": 0.4509,
      "step": 337190
    },
    {
      "epoch": 138.03,
      "grad_norm": 1.8053510189056396,
      "learning_rate": 1.1265202533096297e-06,
      "loss": 0.4623,
      "step": 337200
    },
    {
      "epoch": 138.03,
      "grad_norm": 1.9507081508636475,
      "learning_rate": 1.126384697502793e-06,
      "loss": 0.447,
      "step": 337210
    },
    {
      "epoch": 138.04,
      "grad_norm": 1.8979181051254272,
      "learning_rate": 1.1262491472725184e-06,
      "loss": 0.4629,
      "step": 337220
    },
    {
      "epoch": 138.04,
      "grad_norm": 2.2115137577056885,
      "learning_rate": 1.1261136026194264e-06,
      "loss": 0.4682,
      "step": 337230
    },
    {
      "epoch": 138.04,
      "grad_norm": 1.9681698083877563,
      "learning_rate": 1.1259780635441382e-06,
      "loss": 0.4504,
      "step": 337240
    },
    {
      "epoch": 138.05,
      "grad_norm": 2.4512157440185547,
      "learning_rate": 1.1258425300472748e-06,
      "loss": 0.4431,
      "step": 337250
    },
    {
      "epoch": 138.05,
      "grad_norm": 1.8337358236312866,
      "learning_rate": 1.1257070021294568e-06,
      "loss": 0.4524,
      "step": 337260
    },
    {
      "epoch": 138.06,
      "grad_norm": 1.638283610343933,
      "learning_rate": 1.1255714797913052e-06,
      "loss": 0.4533,
      "step": 337270
    },
    {
      "epoch": 138.06,
      "grad_norm": 2.2858076095581055,
      "learning_rate": 1.1254359630334408e-06,
      "loss": 0.4507,
      "step": 337280
    },
    {
      "epoch": 138.06,
      "grad_norm": 1.9337875843048096,
      "learning_rate": 1.1253004518564842e-06,
      "loss": 0.4655,
      "step": 337290
    },
    {
      "epoch": 138.07,
      "grad_norm": 2.4350969791412354,
      "learning_rate": 1.1251649462610563e-06,
      "loss": 0.4556,
      "step": 337300
    },
    {
      "epoch": 138.07,
      "grad_norm": 2.0122110843658447,
      "learning_rate": 1.125029446247778e-06,
      "loss": 0.4675,
      "step": 337310
    },
    {
      "epoch": 138.08,
      "grad_norm": 2.218174934387207,
      "learning_rate": 1.1248939518172695e-06,
      "loss": 0.458,
      "step": 337320
    },
    {
      "epoch": 138.08,
      "grad_norm": 1.9593151807785034,
      "learning_rate": 1.124758462970152e-06,
      "loss": 0.4463,
      "step": 337330
    },
    {
      "epoch": 138.08,
      "grad_norm": 2.2637922763824463,
      "learning_rate": 1.124622979707046e-06,
      "loss": 0.4412,
      "step": 337340
    },
    {
      "epoch": 138.09,
      "grad_norm": 2.1586642265319824,
      "learning_rate": 1.1244875020285719e-06,
      "loss": 0.4543,
      "step": 337350
    },
    {
      "epoch": 138.09,
      "grad_norm": 1.9884718656539917,
      "learning_rate": 1.1243520299353507e-06,
      "loss": 0.4657,
      "step": 337360
    },
    {
      "epoch": 138.1,
      "grad_norm": 1.9313383102416992,
      "learning_rate": 1.1242165634280026e-06,
      "loss": 0.4565,
      "step": 337370
    },
    {
      "epoch": 138.1,
      "grad_norm": 2.5819530487060547,
      "learning_rate": 1.1240811025071483e-06,
      "loss": 0.4469,
      "step": 337380
    },
    {
      "epoch": 138.1,
      "grad_norm": 3.1679861545562744,
      "learning_rate": 1.123945647173409e-06,
      "loss": 0.4851,
      "step": 337390
    },
    {
      "epoch": 138.11,
      "grad_norm": 1.8233598470687866,
      "learning_rate": 1.1238101974274041e-06,
      "loss": 0.4432,
      "step": 337400
    },
    {
      "epoch": 138.11,
      "grad_norm": 1.889447569847107,
      "learning_rate": 1.123674753269754e-06,
      "loss": 0.456,
      "step": 337410
    },
    {
      "epoch": 138.12,
      "grad_norm": 1.5170975923538208,
      "learning_rate": 1.12353931470108e-06,
      "loss": 0.4293,
      "step": 337420
    },
    {
      "epoch": 138.12,
      "grad_norm": 2.7309412956237793,
      "learning_rate": 1.123403881722001e-06,
      "loss": 0.4534,
      "step": 337430
    },
    {
      "epoch": 138.13,
      "grad_norm": 1.6280081272125244,
      "learning_rate": 1.1232684543331398e-06,
      "loss": 0.4528,
      "step": 337440
    },
    {
      "epoch": 138.13,
      "grad_norm": 2.20479416847229,
      "learning_rate": 1.1231330325351155e-06,
      "loss": 0.4581,
      "step": 337450
    },
    {
      "epoch": 138.13,
      "grad_norm": 2.0371813774108887,
      "learning_rate": 1.1229976163285483e-06,
      "loss": 0.4461,
      "step": 337460
    },
    {
      "epoch": 138.14,
      "grad_norm": 1.7661941051483154,
      "learning_rate": 1.1228622057140588e-06,
      "loss": 0.472,
      "step": 337470
    },
    {
      "epoch": 138.14,
      "grad_norm": 2.0070364475250244,
      "learning_rate": 1.1227268006922672e-06,
      "loss": 0.4472,
      "step": 337480
    },
    {
      "epoch": 138.15,
      "grad_norm": 1.8986718654632568,
      "learning_rate": 1.122591401263794e-06,
      "loss": 0.4523,
      "step": 337490
    },
    {
      "epoch": 138.15,
      "grad_norm": 1.9271929264068604,
      "learning_rate": 1.122456007429259e-06,
      "loss": 0.4759,
      "step": 337500
    },
    {
      "epoch": 138.15,
      "grad_norm": 2.2254419326782227,
      "learning_rate": 1.1223206191892826e-06,
      "loss": 0.4602,
      "step": 337510
    },
    {
      "epoch": 138.16,
      "grad_norm": 2.111085891723633,
      "learning_rate": 1.122185236544485e-06,
      "loss": 0.4615,
      "step": 337520
    },
    {
      "epoch": 138.16,
      "grad_norm": 1.6159734725952148,
      "learning_rate": 1.1220498594954875e-06,
      "loss": 0.4479,
      "step": 337530
    },
    {
      "epoch": 138.17,
      "grad_norm": 2.408147096633911,
      "learning_rate": 1.1219144880429082e-06,
      "loss": 0.446,
      "step": 337540
    },
    {
      "epoch": 138.17,
      "grad_norm": 2.159029960632324,
      "learning_rate": 1.121779122187368e-06,
      "loss": 0.4569,
      "step": 337550
    },
    {
      "epoch": 138.17,
      "grad_norm": 2.351039171218872,
      "learning_rate": 1.1216437619294872e-06,
      "loss": 0.4467,
      "step": 337560
    },
    {
      "epoch": 138.18,
      "grad_norm": 2.161741018295288,
      "learning_rate": 1.1215084072698857e-06,
      "loss": 0.4541,
      "step": 337570
    },
    {
      "epoch": 138.18,
      "grad_norm": 1.9732269048690796,
      "learning_rate": 1.121373058209184e-06,
      "loss": 0.4602,
      "step": 337580
    },
    {
      "epoch": 138.19,
      "grad_norm": 1.8216749429702759,
      "learning_rate": 1.1212377147480014e-06,
      "loss": 0.4734,
      "step": 337590
    },
    {
      "epoch": 138.19,
      "grad_norm": 2.0254592895507812,
      "learning_rate": 1.1211023768869586e-06,
      "loss": 0.4326,
      "step": 337600
    },
    {
      "epoch": 138.19,
      "grad_norm": 3.405848741531372,
      "learning_rate": 1.1209670446266748e-06,
      "loss": 0.451,
      "step": 337610
    },
    {
      "epoch": 138.2,
      "grad_norm": 2.7233495712280273,
      "learning_rate": 1.1208317179677698e-06,
      "loss": 0.4604,
      "step": 337620
    },
    {
      "epoch": 138.2,
      "grad_norm": 2.0037903785705566,
      "learning_rate": 1.1206963969108647e-06,
      "loss": 0.4685,
      "step": 337630
    },
    {
      "epoch": 138.21,
      "grad_norm": 2.1033196449279785,
      "learning_rate": 1.120561081456579e-06,
      "loss": 0.4701,
      "step": 337640
    },
    {
      "epoch": 138.21,
      "grad_norm": 1.9655518531799316,
      "learning_rate": 1.120425771605532e-06,
      "loss": 0.4463,
      "step": 337650
    },
    {
      "epoch": 138.22,
      "grad_norm": 2.4289398193359375,
      "learning_rate": 1.120290467358344e-06,
      "loss": 0.4579,
      "step": 337660
    },
    {
      "epoch": 138.22,
      "grad_norm": 1.681402564048767,
      "learning_rate": 1.1201551687156356e-06,
      "loss": 0.4486,
      "step": 337670
    },
    {
      "epoch": 138.22,
      "grad_norm": 2.189206600189209,
      "learning_rate": 1.1200198756780247e-06,
      "loss": 0.4662,
      "step": 337680
    },
    {
      "epoch": 138.23,
      "grad_norm": 2.07464599609375,
      "learning_rate": 1.1198845882461317e-06,
      "loss": 0.4637,
      "step": 337690
    },
    {
      "epoch": 138.23,
      "grad_norm": 2.112682580947876,
      "learning_rate": 1.119749306420577e-06,
      "loss": 0.436,
      "step": 337700
    },
    {
      "epoch": 138.24,
      "grad_norm": 2.0035219192504883,
      "learning_rate": 1.1196140302019798e-06,
      "loss": 0.4573,
      "step": 337710
    },
    {
      "epoch": 138.24,
      "grad_norm": 1.8621435165405273,
      "learning_rate": 1.1194787595909598e-06,
      "loss": 0.4477,
      "step": 337720
    },
    {
      "epoch": 138.24,
      "grad_norm": 2.3014843463897705,
      "learning_rate": 1.119343494588137e-06,
      "loss": 0.4449,
      "step": 337730
    },
    {
      "epoch": 138.25,
      "grad_norm": 1.8012017011642456,
      "learning_rate": 1.1192082351941306e-06,
      "loss": 0.4432,
      "step": 337740
    },
    {
      "epoch": 138.25,
      "grad_norm": 1.929805040359497,
      "learning_rate": 1.1190729814095603e-06,
      "loss": 0.472,
      "step": 337750
    },
    {
      "epoch": 138.26,
      "grad_norm": 2.005934953689575,
      "learning_rate": 1.1189377332350459e-06,
      "loss": 0.4563,
      "step": 337760
    },
    {
      "epoch": 138.26,
      "grad_norm": 2.0835299491882324,
      "learning_rate": 1.1188024906712066e-06,
      "loss": 0.4233,
      "step": 337770
    },
    {
      "epoch": 138.26,
      "grad_norm": 1.8750133514404297,
      "learning_rate": 1.1186672537186622e-06,
      "loss": 0.4479,
      "step": 337780
    },
    {
      "epoch": 138.27,
      "grad_norm": 1.7972601652145386,
      "learning_rate": 1.1185320223780324e-06,
      "loss": 0.4401,
      "step": 337790
    },
    {
      "epoch": 138.27,
      "grad_norm": 2.1068499088287354,
      "learning_rate": 1.1183967966499362e-06,
      "loss": 0.4595,
      "step": 337800
    },
    {
      "epoch": 138.28,
      "grad_norm": 2.488969087600708,
      "learning_rate": 1.1182615765349933e-06,
      "loss": 0.4817,
      "step": 337810
    },
    {
      "epoch": 138.28,
      "grad_norm": 2.209918737411499,
      "learning_rate": 1.1181263620338227e-06,
      "loss": 0.4528,
      "step": 337820
    },
    {
      "epoch": 138.28,
      "grad_norm": 1.6164135932922363,
      "learning_rate": 1.1179911531470446e-06,
      "loss": 0.4453,
      "step": 337830
    },
    {
      "epoch": 138.29,
      "grad_norm": 2.1188182830810547,
      "learning_rate": 1.117855949875278e-06,
      "loss": 0.4569,
      "step": 337840
    },
    {
      "epoch": 138.29,
      "grad_norm": 2.748105525970459,
      "learning_rate": 1.117720752219142e-06,
      "loss": 0.4688,
      "step": 337850
    },
    {
      "epoch": 138.3,
      "grad_norm": 1.7149527072906494,
      "learning_rate": 1.1175855601792563e-06,
      "loss": 0.4647,
      "step": 337860
    },
    {
      "epoch": 138.3,
      "grad_norm": 2.147125005722046,
      "learning_rate": 1.1174503737562397e-06,
      "loss": 0.4514,
      "step": 337870
    },
    {
      "epoch": 138.31,
      "grad_norm": 1.837262749671936,
      "learning_rate": 1.1173151929507118e-06,
      "loss": 0.4617,
      "step": 337880
    },
    {
      "epoch": 138.31,
      "grad_norm": 2.287273406982422,
      "learning_rate": 1.1171800177632922e-06,
      "loss": 0.453,
      "step": 337890
    },
    {
      "epoch": 138.31,
      "grad_norm": 1.987849235534668,
      "learning_rate": 1.1170448481945994e-06,
      "loss": 0.4486,
      "step": 337900
    },
    {
      "epoch": 138.32,
      "grad_norm": 1.8645625114440918,
      "learning_rate": 1.116909684245253e-06,
      "loss": 0.4632,
      "step": 337910
    },
    {
      "epoch": 138.32,
      "grad_norm": 1.8366100788116455,
      "learning_rate": 1.116774525915872e-06,
      "loss": 0.434,
      "step": 337920
    },
    {
      "epoch": 138.33,
      "grad_norm": 2.249641180038452,
      "learning_rate": 1.1166393732070757e-06,
      "loss": 0.461,
      "step": 337930
    },
    {
      "epoch": 138.33,
      "grad_norm": 2.3827407360076904,
      "learning_rate": 1.1165042261194832e-06,
      "loss": 0.4327,
      "step": 337940
    },
    {
      "epoch": 138.33,
      "grad_norm": 1.9054514169692993,
      "learning_rate": 1.1163690846537137e-06,
      "loss": 0.454,
      "step": 337950
    },
    {
      "epoch": 138.34,
      "grad_norm": 1.8315197229385376,
      "learning_rate": 1.1162339488103855e-06,
      "loss": 0.4544,
      "step": 337960
    },
    {
      "epoch": 138.34,
      "grad_norm": 1.900780200958252,
      "learning_rate": 1.1160988185901195e-06,
      "loss": 0.4618,
      "step": 337970
    },
    {
      "epoch": 138.35,
      "grad_norm": 1.7612979412078857,
      "learning_rate": 1.1159636939935327e-06,
      "loss": 0.4553,
      "step": 337980
    },
    {
      "epoch": 138.35,
      "grad_norm": 2.2586612701416016,
      "learning_rate": 1.1158285750212446e-06,
      "loss": 0.4326,
      "step": 337990
    },
    {
      "epoch": 138.35,
      "grad_norm": 2.361941337585449,
      "learning_rate": 1.1156934616738736e-06,
      "loss": 0.4755,
      "step": 338000
    },
    {
      "epoch": 138.36,
      "grad_norm": 2.0104668140411377,
      "learning_rate": 1.1155583539520402e-06,
      "loss": 0.4802,
      "step": 338010
    },
    {
      "epoch": 138.36,
      "grad_norm": 2.2857468128204346,
      "learning_rate": 1.1154232518563627e-06,
      "loss": 0.4553,
      "step": 338020
    },
    {
      "epoch": 138.37,
      "grad_norm": 1.9636355638504028,
      "learning_rate": 1.1152881553874596e-06,
      "loss": 0.4453,
      "step": 338030
    },
    {
      "epoch": 138.37,
      "grad_norm": 1.6433004140853882,
      "learning_rate": 1.11515306454595e-06,
      "loss": 0.4295,
      "step": 338040
    },
    {
      "epoch": 138.37,
      "grad_norm": 1.83650803565979,
      "learning_rate": 1.1150179793324526e-06,
      "loss": 0.443,
      "step": 338050
    },
    {
      "epoch": 138.38,
      "grad_norm": 1.9077939987182617,
      "learning_rate": 1.1148828997475864e-06,
      "loss": 0.4498,
      "step": 338060
    },
    {
      "epoch": 138.38,
      "grad_norm": 3.8655197620391846,
      "learning_rate": 1.1147478257919701e-06,
      "loss": 0.4575,
      "step": 338070
    },
    {
      "epoch": 138.39,
      "grad_norm": 2.0260300636291504,
      "learning_rate": 1.1146127574662225e-06,
      "loss": 0.4541,
      "step": 338080
    },
    {
      "epoch": 138.39,
      "grad_norm": 1.714692234992981,
      "learning_rate": 1.1144776947709623e-06,
      "loss": 0.4613,
      "step": 338090
    },
    {
      "epoch": 138.4,
      "grad_norm": 1.7195441722869873,
      "learning_rate": 1.114342637706808e-06,
      "loss": 0.4486,
      "step": 338100
    },
    {
      "epoch": 138.4,
      "grad_norm": 1.9539711475372314,
      "learning_rate": 1.1142075862743792e-06,
      "loss": 0.4495,
      "step": 338110
    },
    {
      "epoch": 138.4,
      "grad_norm": 2.3268423080444336,
      "learning_rate": 1.1140725404742931e-06,
      "loss": 0.4429,
      "step": 338120
    },
    {
      "epoch": 138.41,
      "grad_norm": 2.2943501472473145,
      "learning_rate": 1.1139375003071688e-06,
      "loss": 0.4403,
      "step": 338130
    },
    {
      "epoch": 138.41,
      "grad_norm": 2.226534843444824,
      "learning_rate": 1.1138024657736254e-06,
      "loss": 0.4717,
      "step": 338140
    },
    {
      "epoch": 138.42,
      "grad_norm": 2.1142640113830566,
      "learning_rate": 1.1136674368742808e-06,
      "loss": 0.4499,
      "step": 338150
    },
    {
      "epoch": 138.42,
      "grad_norm": 1.9460625648498535,
      "learning_rate": 1.1135324136097544e-06,
      "loss": 0.459,
      "step": 338160
    },
    {
      "epoch": 138.42,
      "grad_norm": 2.3415944576263428,
      "learning_rate": 1.1133973959806636e-06,
      "loss": 0.4266,
      "step": 338170
    },
    {
      "epoch": 138.43,
      "grad_norm": 2.2582664489746094,
      "learning_rate": 1.113262383987628e-06,
      "loss": 0.4537,
      "step": 338180
    },
    {
      "epoch": 138.43,
      "grad_norm": 2.205554962158203,
      "learning_rate": 1.1131273776312647e-06,
      "loss": 0.4523,
      "step": 338190
    },
    {
      "epoch": 138.44,
      "grad_norm": 1.8107664585113525,
      "learning_rate": 1.1129923769121938e-06,
      "loss": 0.4413,
      "step": 338200
    },
    {
      "epoch": 138.44,
      "grad_norm": 1.9563112258911133,
      "learning_rate": 1.1128573818310332e-06,
      "loss": 0.4411,
      "step": 338210
    },
    {
      "epoch": 138.44,
      "grad_norm": 1.661742925643921,
      "learning_rate": 1.1127223923884007e-06,
      "loss": 0.4445,
      "step": 338220
    },
    {
      "epoch": 138.45,
      "grad_norm": 1.7467902898788452,
      "learning_rate": 1.112587408584915e-06,
      "loss": 0.452,
      "step": 338230
    },
    {
      "epoch": 138.45,
      "grad_norm": 1.7459778785705566,
      "learning_rate": 1.1124524304211947e-06,
      "loss": 0.4588,
      "step": 338240
    },
    {
      "epoch": 138.46,
      "grad_norm": 1.8715417385101318,
      "learning_rate": 1.112317457897859e-06,
      "loss": 0.444,
      "step": 338250
    },
    {
      "epoch": 138.46,
      "grad_norm": 1.7593032121658325,
      "learning_rate": 1.1121824910155238e-06,
      "loss": 0.4486,
      "step": 338260
    },
    {
      "epoch": 138.47,
      "grad_norm": 1.8707159757614136,
      "learning_rate": 1.1120475297748086e-06,
      "loss": 0.4586,
      "step": 338270
    },
    {
      "epoch": 138.47,
      "grad_norm": 1.6442582607269287,
      "learning_rate": 1.1119125741763317e-06,
      "loss": 0.4746,
      "step": 338280
    },
    {
      "epoch": 138.47,
      "grad_norm": 1.9683330059051514,
      "learning_rate": 1.1117776242207114e-06,
      "loss": 0.4783,
      "step": 338290
    },
    {
      "epoch": 138.48,
      "grad_norm": 1.836466908454895,
      "learning_rate": 1.1116426799085657e-06,
      "loss": 0.426,
      "step": 338300
    },
    {
      "epoch": 138.48,
      "grad_norm": 1.9479129314422607,
      "learning_rate": 1.1115077412405128e-06,
      "loss": 0.4346,
      "step": 338310
    },
    {
      "epoch": 138.49,
      "grad_norm": 2.1838226318359375,
      "learning_rate": 1.1113728082171709e-06,
      "loss": 0.4559,
      "step": 338320
    },
    {
      "epoch": 138.49,
      "grad_norm": 2.146620750427246,
      "learning_rate": 1.1112378808391581e-06,
      "loss": 0.4805,
      "step": 338330
    },
    {
      "epoch": 138.49,
      "grad_norm": 1.8749333620071411,
      "learning_rate": 1.1111029591070924e-06,
      "loss": 0.466,
      "step": 338340
    },
    {
      "epoch": 138.5,
      "grad_norm": 1.80422043800354,
      "learning_rate": 1.110968043021592e-06,
      "loss": 0.44,
      "step": 338350
    },
    {
      "epoch": 138.5,
      "grad_norm": 2.1052465438842773,
      "learning_rate": 1.1108331325832744e-06,
      "loss": 0.4629,
      "step": 338360
    },
    {
      "epoch": 138.51,
      "grad_norm": 2.108851671218872,
      "learning_rate": 1.1106982277927583e-06,
      "loss": 0.4454,
      "step": 338370
    },
    {
      "epoch": 138.51,
      "grad_norm": 1.806915283203125,
      "learning_rate": 1.1105633286506617e-06,
      "loss": 0.4647,
      "step": 338380
    },
    {
      "epoch": 138.51,
      "grad_norm": 2.137967586517334,
      "learning_rate": 1.1104284351576018e-06,
      "loss": 0.4393,
      "step": 338390
    },
    {
      "epoch": 138.52,
      "grad_norm": 2.6503560543060303,
      "learning_rate": 1.110293547314197e-06,
      "loss": 0.4736,
      "step": 338400
    },
    {
      "epoch": 138.52,
      "grad_norm": 1.862855076789856,
      "learning_rate": 1.1101586651210656e-06,
      "loss": 0.4455,
      "step": 338410
    },
    {
      "epoch": 138.53,
      "grad_norm": 1.7643619775772095,
      "learning_rate": 1.1100237885788247e-06,
      "loss": 0.4657,
      "step": 338420
    },
    {
      "epoch": 138.53,
      "grad_norm": 2.048360824584961,
      "learning_rate": 1.1098889176880928e-06,
      "loss": 0.4507,
      "step": 338430
    },
    {
      "epoch": 138.53,
      "grad_norm": 2.184098243713379,
      "learning_rate": 1.109754052449487e-06,
      "loss": 0.4624,
      "step": 338440
    },
    {
      "epoch": 138.54,
      "grad_norm": 2.0374231338500977,
      "learning_rate": 1.109619192863626e-06,
      "loss": 0.4493,
      "step": 338450
    },
    {
      "epoch": 138.54,
      "grad_norm": 1.8997979164123535,
      "learning_rate": 1.1094843389311269e-06,
      "loss": 0.4349,
      "step": 338460
    },
    {
      "epoch": 138.55,
      "grad_norm": 1.6289175748825073,
      "learning_rate": 1.1093494906526076e-06,
      "loss": 0.4561,
      "step": 338470
    },
    {
      "epoch": 138.55,
      "grad_norm": 2.5880656242370605,
      "learning_rate": 1.109214648028686e-06,
      "loss": 0.4831,
      "step": 338480
    },
    {
      "epoch": 138.56,
      "grad_norm": 1.9523826837539673,
      "learning_rate": 1.1090798110599793e-06,
      "loss": 0.4677,
      "step": 338490
    },
    {
      "epoch": 138.56,
      "grad_norm": 1.9367390871047974,
      "learning_rate": 1.1089449797471058e-06,
      "loss": 0.4397,
      "step": 338500
    },
    {
      "epoch": 138.56,
      "grad_norm": 1.979373812675476,
      "learning_rate": 1.108810154090683e-06,
      "loss": 0.4389,
      "step": 338510
    },
    {
      "epoch": 138.57,
      "grad_norm": 1.9315950870513916,
      "learning_rate": 1.1086753340913281e-06,
      "loss": 0.4185,
      "step": 338520
    },
    {
      "epoch": 138.57,
      "grad_norm": 2.1821610927581787,
      "learning_rate": 1.108540519749659e-06,
      "loss": 0.4499,
      "step": 338530
    },
    {
      "epoch": 138.58,
      "grad_norm": 2.306502103805542,
      "learning_rate": 1.1084057110662934e-06,
      "loss": 0.4521,
      "step": 338540
    },
    {
      "epoch": 138.58,
      "grad_norm": 1.8345049619674683,
      "learning_rate": 1.1082709080418485e-06,
      "loss": 0.4532,
      "step": 338550
    },
    {
      "epoch": 138.58,
      "grad_norm": 1.7899891138076782,
      "learning_rate": 1.108136110676943e-06,
      "loss": 0.4448,
      "step": 338560
    },
    {
      "epoch": 138.59,
      "grad_norm": 1.9711410999298096,
      "learning_rate": 1.1080013189721915e-06,
      "loss": 0.4493,
      "step": 338570
    },
    {
      "epoch": 138.59,
      "grad_norm": 1.9706754684448242,
      "learning_rate": 1.1078665329282142e-06,
      "loss": 0.4756,
      "step": 338580
    },
    {
      "epoch": 138.6,
      "grad_norm": 1.8517898321151733,
      "learning_rate": 1.1077317525456277e-06,
      "loss": 0.4659,
      "step": 338590
    },
    {
      "epoch": 138.6,
      "grad_norm": 2.2211148738861084,
      "learning_rate": 1.1075969778250494e-06,
      "loss": 0.4443,
      "step": 338600
    },
    {
      "epoch": 138.6,
      "grad_norm": 2.379673957824707,
      "learning_rate": 1.1074622087670969e-06,
      "loss": 0.4532,
      "step": 338610
    },
    {
      "epoch": 138.61,
      "grad_norm": 2.052959442138672,
      "learning_rate": 1.107327445372387e-06,
      "loss": 0.4542,
      "step": 338620
    },
    {
      "epoch": 138.61,
      "grad_norm": 2.0255372524261475,
      "learning_rate": 1.1071926876415373e-06,
      "loss": 0.448,
      "step": 338630
    },
    {
      "epoch": 138.62,
      "grad_norm": 2.1241655349731445,
      "learning_rate": 1.1070579355751652e-06,
      "loss": 0.4525,
      "step": 338640
    },
    {
      "epoch": 138.62,
      "grad_norm": 2.0125880241394043,
      "learning_rate": 1.1069231891738881e-06,
      "loss": 0.451,
      "step": 338650
    },
    {
      "epoch": 138.62,
      "grad_norm": 1.8832035064697266,
      "learning_rate": 1.1067884484383228e-06,
      "loss": 0.4441,
      "step": 338660
    },
    {
      "epoch": 138.63,
      "grad_norm": 1.9731124639511108,
      "learning_rate": 1.106653713369087e-06,
      "loss": 0.4605,
      "step": 338670
    },
    {
      "epoch": 138.63,
      "grad_norm": 1.8695282936096191,
      "learning_rate": 1.1065189839667975e-06,
      "loss": 0.4529,
      "step": 338680
    },
    {
      "epoch": 138.64,
      "grad_norm": 2.566676139831543,
      "learning_rate": 1.1063842602320726e-06,
      "loss": 0.4671,
      "step": 338690
    },
    {
      "epoch": 138.64,
      "grad_norm": 2.128678560256958,
      "learning_rate": 1.1062495421655278e-06,
      "loss": 0.4723,
      "step": 338700
    },
    {
      "epoch": 138.65,
      "grad_norm": 1.8633744716644287,
      "learning_rate": 1.106114829767781e-06,
      "loss": 0.443,
      "step": 338710
    },
    {
      "epoch": 138.65,
      "grad_norm": 2.1111319065093994,
      "learning_rate": 1.1059801230394492e-06,
      "loss": 0.4566,
      "step": 338720
    },
    {
      "epoch": 138.65,
      "grad_norm": 2.2108566761016846,
      "learning_rate": 1.1058454219811494e-06,
      "loss": 0.4474,
      "step": 338730
    },
    {
      "epoch": 138.66,
      "grad_norm": 2.431433916091919,
      "learning_rate": 1.1057107265934986e-06,
      "loss": 0.4674,
      "step": 338740
    },
    {
      "epoch": 138.66,
      "grad_norm": 1.9218780994415283,
      "learning_rate": 1.1055760368771134e-06,
      "loss": 0.4589,
      "step": 338750
    },
    {
      "epoch": 138.67,
      "grad_norm": 1.9246882200241089,
      "learning_rate": 1.1054413528326121e-06,
      "loss": 0.4443,
      "step": 338760
    },
    {
      "epoch": 138.67,
      "grad_norm": 1.956512451171875,
      "learning_rate": 1.1053066744606111e-06,
      "loss": 0.4541,
      "step": 338770
    },
    {
      "epoch": 138.67,
      "grad_norm": 2.0707836151123047,
      "learning_rate": 1.105172001761727e-06,
      "loss": 0.4527,
      "step": 338780
    },
    {
      "epoch": 138.68,
      "grad_norm": 2.3560707569122314,
      "learning_rate": 1.1050373347365768e-06,
      "loss": 0.4467,
      "step": 338790
    },
    {
      "epoch": 138.68,
      "grad_norm": 1.9242080450057983,
      "learning_rate": 1.1049026733857776e-06,
      "loss": 0.4648,
      "step": 338800
    },
    {
      "epoch": 138.69,
      "grad_norm": 2.0941715240478516,
      "learning_rate": 1.1047680177099463e-06,
      "loss": 0.4172,
      "step": 338810
    },
    {
      "epoch": 138.69,
      "grad_norm": 1.8420761823654175,
      "learning_rate": 1.1046333677096995e-06,
      "loss": 0.4471,
      "step": 338820
    },
    {
      "epoch": 138.69,
      "grad_norm": 1.8647871017456055,
      "learning_rate": 1.1044987233856548e-06,
      "loss": 0.4592,
      "step": 338830
    },
    {
      "epoch": 138.7,
      "grad_norm": 1.820592999458313,
      "learning_rate": 1.1043640847384276e-06,
      "loss": 0.462,
      "step": 338840
    },
    {
      "epoch": 138.7,
      "grad_norm": 2.0128324031829834,
      "learning_rate": 1.1042294517686351e-06,
      "loss": 0.4622,
      "step": 338850
    },
    {
      "epoch": 138.71,
      "grad_norm": 1.48738694190979,
      "learning_rate": 1.1040948244768944e-06,
      "loss": 0.4596,
      "step": 338860
    },
    {
      "epoch": 138.71,
      "grad_norm": 1.7887663841247559,
      "learning_rate": 1.1039602028638222e-06,
      "loss": 0.4309,
      "step": 338870
    },
    {
      "epoch": 138.71,
      "grad_norm": 1.6974563598632812,
      "learning_rate": 1.103825586930035e-06,
      "loss": 0.4321,
      "step": 338880
    },
    {
      "epoch": 138.72,
      "grad_norm": 1.704140305519104,
      "learning_rate": 1.1036909766761493e-06,
      "loss": 0.448,
      "step": 338890
    },
    {
      "epoch": 138.72,
      "grad_norm": 2.084270477294922,
      "learning_rate": 1.1035563721027822e-06,
      "loss": 0.4405,
      "step": 338900
    },
    {
      "epoch": 138.73,
      "grad_norm": 2.5049540996551514,
      "learning_rate": 1.10342177321055e-06,
      "loss": 0.446,
      "step": 338910
    },
    {
      "epoch": 138.73,
      "grad_norm": 1.9244941473007202,
      "learning_rate": 1.103287180000069e-06,
      "loss": 0.4593,
      "step": 338920
    },
    {
      "epoch": 138.74,
      "grad_norm": 2.349186897277832,
      "learning_rate": 1.1031525924719564e-06,
      "loss": 0.4548,
      "step": 338930
    },
    {
      "epoch": 138.74,
      "grad_norm": 2.2144594192504883,
      "learning_rate": 1.1030180106268276e-06,
      "loss": 0.4435,
      "step": 338940
    },
    {
      "epoch": 138.74,
      "grad_norm": 2.4032084941864014,
      "learning_rate": 1.1028834344653005e-06,
      "loss": 0.4551,
      "step": 338950
    },
    {
      "epoch": 138.75,
      "grad_norm": 2.1246469020843506,
      "learning_rate": 1.102748863987991e-06,
      "loss": 0.4544,
      "step": 338960
    },
    {
      "epoch": 138.75,
      "grad_norm": 2.115649461746216,
      "learning_rate": 1.1026142991955167e-06,
      "loss": 0.4598,
      "step": 338970
    },
    {
      "epoch": 138.76,
      "grad_norm": 1.8769841194152832,
      "learning_rate": 1.1024797400884917e-06,
      "loss": 0.4537,
      "step": 338980
    },
    {
      "epoch": 138.76,
      "grad_norm": 1.8999801874160767,
      "learning_rate": 1.1023451866675334e-06,
      "loss": 0.4518,
      "step": 338990
    },
    {
      "epoch": 138.76,
      "grad_norm": 1.7842882871627808,
      "learning_rate": 1.1022106389332582e-06,
      "loss": 0.4475,
      "step": 339000
    },
    {
      "epoch": 138.77,
      "grad_norm": 1.6304428577423096,
      "learning_rate": 1.102076096886283e-06,
      "loss": 0.4826,
      "step": 339010
    },
    {
      "epoch": 138.77,
      "grad_norm": 1.882683515548706,
      "learning_rate": 1.1019415605272233e-06,
      "loss": 0.4692,
      "step": 339020
    },
    {
      "epoch": 138.78,
      "grad_norm": 1.91483473777771,
      "learning_rate": 1.101807029856696e-06,
      "loss": 0.4688,
      "step": 339030
    },
    {
      "epoch": 138.78,
      "grad_norm": 2.2825417518615723,
      "learning_rate": 1.101672504875317e-06,
      "loss": 0.4529,
      "step": 339040
    },
    {
      "epoch": 138.78,
      "grad_norm": 1.8900030851364136,
      "learning_rate": 1.1015379855837027e-06,
      "loss": 0.4494,
      "step": 339050
    },
    {
      "epoch": 138.79,
      "grad_norm": 2.4213874340057373,
      "learning_rate": 1.101403471982469e-06,
      "loss": 0.448,
      "step": 339060
    },
    {
      "epoch": 138.79,
      "grad_norm": 2.317556381225586,
      "learning_rate": 1.1012689640722324e-06,
      "loss": 0.4429,
      "step": 339070
    },
    {
      "epoch": 138.8,
      "grad_norm": 1.8402862548828125,
      "learning_rate": 1.1011344618536092e-06,
      "loss": 0.4696,
      "step": 339080
    },
    {
      "epoch": 138.8,
      "grad_norm": 2.0352563858032227,
      "learning_rate": 1.1009999653272153e-06,
      "loss": 0.4376,
      "step": 339090
    },
    {
      "epoch": 138.8,
      "grad_norm": 1.7954825162887573,
      "learning_rate": 1.100865474493667e-06,
      "loss": 0.4398,
      "step": 339100
    },
    {
      "epoch": 138.81,
      "grad_norm": 1.960567831993103,
      "learning_rate": 1.10073098935358e-06,
      "loss": 0.4636,
      "step": 339110
    },
    {
      "epoch": 138.81,
      "grad_norm": 2.3258204460144043,
      "learning_rate": 1.1005965099075704e-06,
      "loss": 0.4439,
      "step": 339120
    },
    {
      "epoch": 138.82,
      "grad_norm": 1.933892011642456,
      "learning_rate": 1.1004620361562547e-06,
      "loss": 0.445,
      "step": 339130
    },
    {
      "epoch": 138.82,
      "grad_norm": 2.0500197410583496,
      "learning_rate": 1.1003275681002486e-06,
      "loss": 0.4673,
      "step": 339140
    },
    {
      "epoch": 138.83,
      "grad_norm": 2.038954257965088,
      "learning_rate": 1.1001931057401678e-06,
      "loss": 0.4497,
      "step": 339150
    },
    {
      "epoch": 138.83,
      "grad_norm": 1.7848570346832275,
      "learning_rate": 1.1000586490766288e-06,
      "loss": 0.4367,
      "step": 339160
    },
    {
      "epoch": 138.83,
      "grad_norm": 1.878868818283081,
      "learning_rate": 1.0999241981102474e-06,
      "loss": 0.449,
      "step": 339170
    },
    {
      "epoch": 138.84,
      "grad_norm": 2.002833604812622,
      "learning_rate": 1.0997897528416392e-06,
      "loss": 0.4626,
      "step": 339180
    },
    {
      "epoch": 138.84,
      "grad_norm": 2.2204856872558594,
      "learning_rate": 1.0996553132714203e-06,
      "loss": 0.4577,
      "step": 339190
    },
    {
      "epoch": 138.85,
      "grad_norm": 2.1301612854003906,
      "learning_rate": 1.0995208794002064e-06,
      "loss": 0.4452,
      "step": 339200
    },
    {
      "epoch": 138.85,
      "grad_norm": 1.8970909118652344,
      "learning_rate": 1.0993864512286134e-06,
      "loss": 0.4364,
      "step": 339210
    },
    {
      "epoch": 138.85,
      "grad_norm": 1.8235489130020142,
      "learning_rate": 1.099252028757257e-06,
      "loss": 0.4593,
      "step": 339220
    },
    {
      "epoch": 138.86,
      "grad_norm": 1.9519776105880737,
      "learning_rate": 1.0991176119867532e-06,
      "loss": 0.4644,
      "step": 339230
    },
    {
      "epoch": 138.86,
      "grad_norm": 1.905135989189148,
      "learning_rate": 1.0989832009177177e-06,
      "loss": 0.4473,
      "step": 339240
    },
    {
      "epoch": 138.87,
      "grad_norm": 1.9224491119384766,
      "learning_rate": 1.0988487955507663e-06,
      "loss": 0.4639,
      "step": 339250
    },
    {
      "epoch": 138.87,
      "grad_norm": 2.2522335052490234,
      "learning_rate": 1.098714395886514e-06,
      "loss": 0.4563,
      "step": 339260
    },
    {
      "epoch": 138.87,
      "grad_norm": 2.0499203205108643,
      "learning_rate": 1.0985800019255776e-06,
      "loss": 0.4465,
      "step": 339270
    },
    {
      "epoch": 138.88,
      "grad_norm": 1.9019757509231567,
      "learning_rate": 1.0984456136685725e-06,
      "loss": 0.4411,
      "step": 339280
    },
    {
      "epoch": 138.88,
      "grad_norm": 2.1763947010040283,
      "learning_rate": 1.098311231116113e-06,
      "loss": 0.4361,
      "step": 339290
    },
    {
      "epoch": 138.89,
      "grad_norm": 1.6023517847061157,
      "learning_rate": 1.098176854268816e-06,
      "loss": 0.4438,
      "step": 339300
    },
    {
      "epoch": 138.89,
      "grad_norm": 1.8739372491836548,
      "learning_rate": 1.0980424831272967e-06,
      "loss": 0.4341,
      "step": 339310
    },
    {
      "epoch": 138.89,
      "grad_norm": 1.8498954772949219,
      "learning_rate": 1.0979081176921695e-06,
      "loss": 0.4634,
      "step": 339320
    },
    {
      "epoch": 138.9,
      "grad_norm": 1.9760818481445312,
      "learning_rate": 1.097773757964052e-06,
      "loss": 0.4641,
      "step": 339330
    },
    {
      "epoch": 138.9,
      "grad_norm": 3.1188361644744873,
      "learning_rate": 1.0976394039435585e-06,
      "loss": 0.4479,
      "step": 339340
    },
    {
      "epoch": 138.91,
      "grad_norm": 1.8600438833236694,
      "learning_rate": 1.0975050556313048e-06,
      "loss": 0.4612,
      "step": 339350
    },
    {
      "epoch": 138.91,
      "grad_norm": 2.2294769287109375,
      "learning_rate": 1.0973707130279062e-06,
      "loss": 0.4617,
      "step": 339360
    },
    {
      "epoch": 138.92,
      "grad_norm": 1.9610981941223145,
      "learning_rate": 1.0972363761339783e-06,
      "loss": 0.4544,
      "step": 339370
    },
    {
      "epoch": 138.92,
      "grad_norm": 2.112417221069336,
      "learning_rate": 1.0971020449501358e-06,
      "loss": 0.4484,
      "step": 339380
    },
    {
      "epoch": 138.92,
      "grad_norm": 1.822422742843628,
      "learning_rate": 1.096967719476995e-06,
      "loss": 0.4406,
      "step": 339390
    },
    {
      "epoch": 138.93,
      "grad_norm": 2.0252745151519775,
      "learning_rate": 1.0968333997151704e-06,
      "loss": 0.4415,
      "step": 339400
    },
    {
      "epoch": 138.93,
      "grad_norm": 2.389498472213745,
      "learning_rate": 1.0966990856652787e-06,
      "loss": 0.4594,
      "step": 339410
    },
    {
      "epoch": 138.94,
      "grad_norm": 1.7981966733932495,
      "learning_rate": 1.096564777327933e-06,
      "loss": 0.4499,
      "step": 339420
    },
    {
      "epoch": 138.94,
      "grad_norm": 1.8443095684051514,
      "learning_rate": 1.09643047470375e-06,
      "loss": 0.4449,
      "step": 339430
    },
    {
      "epoch": 138.94,
      "grad_norm": 2.2627968788146973,
      "learning_rate": 1.0962961777933442e-06,
      "loss": 0.4679,
      "step": 339440
    },
    {
      "epoch": 138.95,
      "grad_norm": 2.520516872406006,
      "learning_rate": 1.0961618865973314e-06,
      "loss": 0.4723,
      "step": 339450
    },
    {
      "epoch": 138.95,
      "grad_norm": 1.9904260635375977,
      "learning_rate": 1.0960276011163265e-06,
      "loss": 0.4475,
      "step": 339460
    },
    {
      "epoch": 138.96,
      "grad_norm": 2.3830411434173584,
      "learning_rate": 1.0958933213509446e-06,
      "loss": 0.444,
      "step": 339470
    },
    {
      "epoch": 138.96,
      "grad_norm": 5.018158912658691,
      "learning_rate": 1.095759047301801e-06,
      "loss": 0.4571,
      "step": 339480
    },
    {
      "epoch": 138.96,
      "grad_norm": 2.140380382537842,
      "learning_rate": 1.0956247789695106e-06,
      "loss": 0.4505,
      "step": 339490
    },
    {
      "epoch": 138.97,
      "grad_norm": 2.2747085094451904,
      "learning_rate": 1.0954905163546885e-06,
      "loss": 0.4607,
      "step": 339500
    },
    {
      "epoch": 138.97,
      "grad_norm": 2.025646686553955,
      "learning_rate": 1.095356259457949e-06,
      "loss": 0.455,
      "step": 339510
    },
    {
      "epoch": 138.98,
      "grad_norm": 2.188101053237915,
      "learning_rate": 1.0952220082799087e-06,
      "loss": 0.4487,
      "step": 339520
    },
    {
      "epoch": 138.98,
      "grad_norm": 1.8796089887619019,
      "learning_rate": 1.0950877628211817e-06,
      "loss": 0.4829,
      "step": 339530
    },
    {
      "epoch": 138.98,
      "grad_norm": 2.7556886672973633,
      "learning_rate": 1.0949535230823832e-06,
      "loss": 0.4354,
      "step": 339540
    },
    {
      "epoch": 138.99,
      "grad_norm": 2.1221163272857666,
      "learning_rate": 1.0948192890641287e-06,
      "loss": 0.4658,
      "step": 339550
    },
    {
      "epoch": 138.99,
      "grad_norm": 2.3757128715515137,
      "learning_rate": 1.0946850607670316e-06,
      "loss": 0.4483,
      "step": 339560
    },
    {
      "epoch": 139.0,
      "grad_norm": 1.989820957183838,
      "learning_rate": 1.0945508381917075e-06,
      "loss": 0.4666,
      "step": 339570
    },
    {
      "epoch": 139.0,
      "eval_loss": 0.4518420100212097,
      "eval_runtime": 52.5203,
      "eval_samples_per_second": 65.67,
      "eval_steps_per_second": 8.225,
      "step": 339577
    },
    {
      "epoch": 139.0,
      "grad_norm": 1.8148168325424194,
      "learning_rate": 1.0944166213387714e-06,
      "loss": 0.4504,
      "step": 339580
    },
    {
      "epoch": 139.01,
      "grad_norm": 2.043752908706665,
      "learning_rate": 1.0942824102088379e-06,
      "loss": 0.4548,
      "step": 339590
    },
    {
      "epoch": 139.01,
      "grad_norm": 1.8562003374099731,
      "learning_rate": 1.094148204802522e-06,
      "loss": 0.4352,
      "step": 339600
    },
    {
      "epoch": 139.01,
      "grad_norm": 1.9630831480026245,
      "learning_rate": 1.0940140051204385e-06,
      "loss": 0.4331,
      "step": 339610
    },
    {
      "epoch": 139.02,
      "grad_norm": 1.9726289510726929,
      "learning_rate": 1.0938798111632016e-06,
      "loss": 0.4706,
      "step": 339620
    },
    {
      "epoch": 139.02,
      "grad_norm": 1.6054824590682983,
      "learning_rate": 1.0937456229314268e-06,
      "loss": 0.4347,
      "step": 339630
    },
    {
      "epoch": 139.03,
      "grad_norm": 2.063889980316162,
      "learning_rate": 1.0936114404257285e-06,
      "loss": 0.4604,
      "step": 339640
    },
    {
      "epoch": 139.03,
      "grad_norm": 1.8480497598648071,
      "learning_rate": 1.093477263646721e-06,
      "loss": 0.452,
      "step": 339650
    },
    {
      "epoch": 139.03,
      "grad_norm": 2.321979522705078,
      "learning_rate": 1.0933430925950197e-06,
      "loss": 0.4422,
      "step": 339660
    },
    {
      "epoch": 139.04,
      "grad_norm": 1.8804017305374146,
      "learning_rate": 1.0932089272712386e-06,
      "loss": 0.4539,
      "step": 339670
    },
    {
      "epoch": 139.04,
      "grad_norm": 2.0725231170654297,
      "learning_rate": 1.0930747676759927e-06,
      "loss": 0.4809,
      "step": 339680
    },
    {
      "epoch": 139.05,
      "grad_norm": 2.030675172805786,
      "learning_rate": 1.0929406138098958e-06,
      "loss": 0.45,
      "step": 339690
    },
    {
      "epoch": 139.05,
      "grad_norm": 2.08158016204834,
      "learning_rate": 1.0928064656735636e-06,
      "loss": 0.4498,
      "step": 339700
    },
    {
      "epoch": 139.05,
      "grad_norm": 2.317701816558838,
      "learning_rate": 1.0926723232676097e-06,
      "loss": 0.4543,
      "step": 339710
    },
    {
      "epoch": 139.06,
      "grad_norm": 1.7173124551773071,
      "learning_rate": 1.092538186592649e-06,
      "loss": 0.4166,
      "step": 339720
    },
    {
      "epoch": 139.06,
      "grad_norm": 1.6133427619934082,
      "learning_rate": 1.0924040556492958e-06,
      "loss": 0.4458,
      "step": 339730
    },
    {
      "epoch": 139.07,
      "grad_norm": 1.996190071105957,
      "learning_rate": 1.0922699304381643e-06,
      "loss": 0.4737,
      "step": 339740
    },
    {
      "epoch": 139.07,
      "grad_norm": 2.0097861289978027,
      "learning_rate": 1.0921358109598694e-06,
      "loss": 0.459,
      "step": 339750
    },
    {
      "epoch": 139.07,
      "grad_norm": 1.8184300661087036,
      "learning_rate": 1.0920016972150255e-06,
      "loss": 0.4485,
      "step": 339760
    },
    {
      "epoch": 139.08,
      "grad_norm": 2.154393434524536,
      "learning_rate": 1.0918675892042463e-06,
      "loss": 0.4561,
      "step": 339770
    },
    {
      "epoch": 139.08,
      "grad_norm": 1.945727825164795,
      "learning_rate": 1.0917334869281467e-06,
      "loss": 0.4374,
      "step": 339780
    },
    {
      "epoch": 139.09,
      "grad_norm": 2.117476463317871,
      "learning_rate": 1.091599390387341e-06,
      "loss": 0.4656,
      "step": 339790
    },
    {
      "epoch": 139.09,
      "grad_norm": 1.8476539850234985,
      "learning_rate": 1.0914652995824432e-06,
      "loss": 0.4592,
      "step": 339800
    },
    {
      "epoch": 139.1,
      "grad_norm": 1.9136501550674438,
      "learning_rate": 1.0913312145140677e-06,
      "loss": 0.4518,
      "step": 339810
    },
    {
      "epoch": 139.1,
      "grad_norm": 1.952131748199463,
      "learning_rate": 1.0911971351828288e-06,
      "loss": 0.4595,
      "step": 339820
    },
    {
      "epoch": 139.1,
      "grad_norm": 2.2367358207702637,
      "learning_rate": 1.0910630615893405e-06,
      "loss": 0.4387,
      "step": 339830
    },
    {
      "epoch": 139.11,
      "grad_norm": 1.7226955890655518,
      "learning_rate": 1.0909289937342169e-06,
      "loss": 0.4312,
      "step": 339840
    },
    {
      "epoch": 139.11,
      "grad_norm": 2.0630383491516113,
      "learning_rate": 1.0907949316180727e-06,
      "loss": 0.4487,
      "step": 339850
    },
    {
      "epoch": 139.12,
      "grad_norm": 1.8157395124435425,
      "learning_rate": 1.090660875241522e-06,
      "loss": 0.4496,
      "step": 339860
    },
    {
      "epoch": 139.12,
      "grad_norm": 2.317725896835327,
      "learning_rate": 1.090526824605178e-06,
      "loss": 0.4606,
      "step": 339870
    },
    {
      "epoch": 139.12,
      "grad_norm": 2.5549867153167725,
      "learning_rate": 1.090392779709655e-06,
      "loss": 0.4467,
      "step": 339880
    },
    {
      "epoch": 139.13,
      "grad_norm": 2.6413767337799072,
      "learning_rate": 1.0902587405555668e-06,
      "loss": 0.4559,
      "step": 339890
    },
    {
      "epoch": 139.13,
      "grad_norm": 1.6715948581695557,
      "learning_rate": 1.0901247071435286e-06,
      "loss": 0.4563,
      "step": 339900
    },
    {
      "epoch": 139.14,
      "grad_norm": 1.8453121185302734,
      "learning_rate": 1.0899906794741537e-06,
      "loss": 0.4691,
      "step": 339910
    },
    {
      "epoch": 139.14,
      "grad_norm": 1.9612314701080322,
      "learning_rate": 1.089856657548056e-06,
      "loss": 0.4683,
      "step": 339920
    },
    {
      "epoch": 139.14,
      "grad_norm": 1.8419954776763916,
      "learning_rate": 1.0897226413658496e-06,
      "loss": 0.4605,
      "step": 339930
    },
    {
      "epoch": 139.15,
      "grad_norm": 2.437795639038086,
      "learning_rate": 1.0895886309281484e-06,
      "loss": 0.4466,
      "step": 339940
    },
    {
      "epoch": 139.15,
      "grad_norm": 2.3399407863616943,
      "learning_rate": 1.0894546262355663e-06,
      "loss": 0.4368,
      "step": 339950
    },
    {
      "epoch": 139.16,
      "grad_norm": 1.8912638425827026,
      "learning_rate": 1.0893206272887168e-06,
      "loss": 0.4405,
      "step": 339960
    },
    {
      "epoch": 139.16,
      "grad_norm": 1.8006365299224854,
      "learning_rate": 1.089186634088214e-06,
      "loss": 0.4343,
      "step": 339970
    },
    {
      "epoch": 139.16,
      "grad_norm": 1.6966263055801392,
      "learning_rate": 1.0890526466346719e-06,
      "loss": 0.449,
      "step": 339980
    },
    {
      "epoch": 139.17,
      "grad_norm": 2.3332602977752686,
      "learning_rate": 1.0889186649287047e-06,
      "loss": 0.443,
      "step": 339990
    },
    {
      "epoch": 139.17,
      "grad_norm": 2.7388975620269775,
      "learning_rate": 1.0887846889709246e-06,
      "loss": 0.4395,
      "step": 340000
    },
    {
      "epoch": 139.18,
      "grad_norm": 1.9034745693206787,
      "learning_rate": 1.0886507187619466e-06,
      "loss": 0.431,
      "step": 340010
    },
    {
      "epoch": 139.18,
      "grad_norm": 2.1778883934020996,
      "learning_rate": 1.0885167543023836e-06,
      "loss": 0.4511,
      "step": 340020
    },
    {
      "epoch": 139.19,
      "grad_norm": 2.402796745300293,
      "learning_rate": 1.08838279559285e-06,
      "loss": 0.4509,
      "step": 340030
    },
    {
      "epoch": 139.19,
      "grad_norm": 1.9456757307052612,
      "learning_rate": 1.088248842633959e-06,
      "loss": 0.4615,
      "step": 340040
    },
    {
      "epoch": 139.19,
      "grad_norm": 1.9567464590072632,
      "learning_rate": 1.0881148954263245e-06,
      "loss": 0.4643,
      "step": 340050
    },
    {
      "epoch": 139.2,
      "grad_norm": 1.9263591766357422,
      "learning_rate": 1.08798095397056e-06,
      "loss": 0.4368,
      "step": 340060
    },
    {
      "epoch": 139.2,
      "grad_norm": 1.9328325986862183,
      "learning_rate": 1.087847018267278e-06,
      "loss": 0.4643,
      "step": 340070
    },
    {
      "epoch": 139.21,
      "grad_norm": 1.9860777854919434,
      "learning_rate": 1.0877130883170938e-06,
      "loss": 0.4355,
      "step": 340080
    },
    {
      "epoch": 139.21,
      "grad_norm": 2.3693840503692627,
      "learning_rate": 1.0875791641206204e-06,
      "loss": 0.4463,
      "step": 340090
    },
    {
      "epoch": 139.21,
      "grad_norm": 2.1950342655181885,
      "learning_rate": 1.087445245678471e-06,
      "loss": 0.4569,
      "step": 340100
    },
    {
      "epoch": 139.22,
      "grad_norm": 1.9614990949630737,
      "learning_rate": 1.087311332991259e-06,
      "loss": 0.4285,
      "step": 340110
    },
    {
      "epoch": 139.22,
      "grad_norm": 1.6883355379104614,
      "learning_rate": 1.0871774260595979e-06,
      "loss": 0.4481,
      "step": 340120
    },
    {
      "epoch": 139.23,
      "grad_norm": 2.194380283355713,
      "learning_rate": 1.0870435248841023e-06,
      "loss": 0.4529,
      "step": 340130
    },
    {
      "epoch": 139.23,
      "grad_norm": 1.8027900457382202,
      "learning_rate": 1.0869096294653834e-06,
      "loss": 0.4708,
      "step": 340140
    },
    {
      "epoch": 139.23,
      "grad_norm": 2.1843061447143555,
      "learning_rate": 1.0867757398040557e-06,
      "loss": 0.4359,
      "step": 340150
    },
    {
      "epoch": 139.24,
      "grad_norm": 2.0961668491363525,
      "learning_rate": 1.0866418559007323e-06,
      "loss": 0.4702,
      "step": 340160
    },
    {
      "epoch": 139.24,
      "grad_norm": 1.9450218677520752,
      "learning_rate": 1.0865079777560266e-06,
      "loss": 0.4563,
      "step": 340170
    },
    {
      "epoch": 139.25,
      "grad_norm": 1.8706936836242676,
      "learning_rate": 1.0863741053705518e-06,
      "loss": 0.4674,
      "step": 340180
    },
    {
      "epoch": 139.25,
      "grad_norm": 1.7354966402053833,
      "learning_rate": 1.0862402387449216e-06,
      "loss": 0.444,
      "step": 340190
    },
    {
      "epoch": 139.26,
      "grad_norm": 2.1568901538848877,
      "learning_rate": 1.0861063778797486e-06,
      "loss": 0.4635,
      "step": 340200
    },
    {
      "epoch": 139.26,
      "grad_norm": 2.0067992210388184,
      "learning_rate": 1.0859725227756466e-06,
      "loss": 0.4692,
      "step": 340210
    },
    {
      "epoch": 139.26,
      "grad_norm": 1.9835896492004395,
      "learning_rate": 1.085838673433228e-06,
      "loss": 0.4489,
      "step": 340220
    },
    {
      "epoch": 139.27,
      "grad_norm": 2.378516912460327,
      "learning_rate": 1.0857048298531065e-06,
      "loss": 0.4416,
      "step": 340230
    },
    {
      "epoch": 139.27,
      "grad_norm": 1.8066613674163818,
      "learning_rate": 1.0855709920358952e-06,
      "loss": 0.4446,
      "step": 340240
    },
    {
      "epoch": 139.28,
      "grad_norm": 2.4259846210479736,
      "learning_rate": 1.085437159982207e-06,
      "loss": 0.4576,
      "step": 340250
    },
    {
      "epoch": 139.28,
      "grad_norm": 1.933562159538269,
      "learning_rate": 1.0853033336926541e-06,
      "loss": 0.4498,
      "step": 340260
    },
    {
      "epoch": 139.28,
      "grad_norm": 1.7912647724151611,
      "learning_rate": 1.0851695131678523e-06,
      "loss": 0.4426,
      "step": 340270
    },
    {
      "epoch": 139.29,
      "grad_norm": 2.2327919006347656,
      "learning_rate": 1.085035698408412e-06,
      "loss": 0.4392,
      "step": 340280
    },
    {
      "epoch": 139.29,
      "grad_norm": 2.1705031394958496,
      "learning_rate": 1.0849018894149468e-06,
      "loss": 0.4457,
      "step": 340290
    },
    {
      "epoch": 139.3,
      "grad_norm": 2.253622531890869,
      "learning_rate": 1.0847680861880702e-06,
      "loss": 0.47,
      "step": 340300
    },
    {
      "epoch": 139.3,
      "grad_norm": 2.321354866027832,
      "learning_rate": 1.0846342887283942e-06,
      "loss": 0.4667,
      "step": 340310
    },
    {
      "epoch": 139.3,
      "grad_norm": 2.0032901763916016,
      "learning_rate": 1.0845004970365327e-06,
      "loss": 0.4521,
      "step": 340320
    },
    {
      "epoch": 139.31,
      "grad_norm": 1.729109764099121,
      "learning_rate": 1.0843667111130982e-06,
      "loss": 0.4459,
      "step": 340330
    },
    {
      "epoch": 139.31,
      "grad_norm": 2.033642053604126,
      "learning_rate": 1.0842329309587034e-06,
      "loss": 0.4484,
      "step": 340340
    },
    {
      "epoch": 139.32,
      "grad_norm": 2.156494379043579,
      "learning_rate": 1.084099156573961e-06,
      "loss": 0.4605,
      "step": 340350
    },
    {
      "epoch": 139.32,
      "grad_norm": 1.6827771663665771,
      "learning_rate": 1.0839653879594844e-06,
      "loss": 0.4497,
      "step": 340360
    },
    {
      "epoch": 139.32,
      "grad_norm": 2.3208868503570557,
      "learning_rate": 1.0838316251158857e-06,
      "loss": 0.4477,
      "step": 340370
    },
    {
      "epoch": 139.33,
      "grad_norm": 1.8946610689163208,
      "learning_rate": 1.083697868043778e-06,
      "loss": 0.4471,
      "step": 340380
    },
    {
      "epoch": 139.33,
      "grad_norm": 2.115712881088257,
      "learning_rate": 1.083564116743774e-06,
      "loss": 0.4644,
      "step": 340390
    },
    {
      "epoch": 139.34,
      "grad_norm": 2.0408339500427246,
      "learning_rate": 1.0834303712164866e-06,
      "loss": 0.4387,
      "step": 340400
    },
    {
      "epoch": 139.34,
      "grad_norm": 2.0082383155822754,
      "learning_rate": 1.0832966314625279e-06,
      "loss": 0.4353,
      "step": 340410
    },
    {
      "epoch": 139.35,
      "grad_norm": 2.466923475265503,
      "learning_rate": 1.083162897482511e-06,
      "loss": 0.4633,
      "step": 340420
    },
    {
      "epoch": 139.35,
      "grad_norm": 1.8965728282928467,
      "learning_rate": 1.0830291692770485e-06,
      "loss": 0.4553,
      "step": 340430
    },
    {
      "epoch": 139.35,
      "grad_norm": 1.9425292015075684,
      "learning_rate": 1.0828954468467537e-06,
      "loss": 0.4539,
      "step": 340440
    },
    {
      "epoch": 139.36,
      "grad_norm": 1.957456350326538,
      "learning_rate": 1.0827617301922366e-06,
      "loss": 0.47,
      "step": 340450
    },
    {
      "epoch": 139.36,
      "grad_norm": 1.9354223012924194,
      "learning_rate": 1.0826280193141121e-06,
      "loss": 0.4448,
      "step": 340460
    },
    {
      "epoch": 139.37,
      "grad_norm": 2.673917531967163,
      "learning_rate": 1.0824943142129925e-06,
      "loss": 0.4533,
      "step": 340470
    },
    {
      "epoch": 139.37,
      "grad_norm": 2.5031585693359375,
      "learning_rate": 1.0823606148894897e-06,
      "loss": 0.4358,
      "step": 340480
    },
    {
      "epoch": 139.37,
      "grad_norm": 2.2731640338897705,
      "learning_rate": 1.0822269213442161e-06,
      "loss": 0.4552,
      "step": 340490
    },
    {
      "epoch": 139.38,
      "grad_norm": 2.1101038455963135,
      "learning_rate": 1.0820932335777846e-06,
      "loss": 0.4533,
      "step": 340500
    },
    {
      "epoch": 139.38,
      "grad_norm": 2.048973560333252,
      "learning_rate": 1.0819595515908073e-06,
      "loss": 0.4613,
      "step": 340510
    },
    {
      "epoch": 139.39,
      "grad_norm": 1.8356679677963257,
      "learning_rate": 1.0818258753838966e-06,
      "loss": 0.4642,
      "step": 340520
    },
    {
      "epoch": 139.39,
      "grad_norm": 2.080781936645508,
      "learning_rate": 1.0816922049576648e-06,
      "loss": 0.4536,
      "step": 340530
    },
    {
      "epoch": 139.39,
      "grad_norm": 2.110184907913208,
      "learning_rate": 1.0815585403127245e-06,
      "loss": 0.4619,
      "step": 340540
    },
    {
      "epoch": 139.4,
      "grad_norm": 2.3154733180999756,
      "learning_rate": 1.0814248814496878e-06,
      "loss": 0.4551,
      "step": 340550
    },
    {
      "epoch": 139.4,
      "grad_norm": 2.2196404933929443,
      "learning_rate": 1.0812912283691669e-06,
      "loss": 0.4483,
      "step": 340560
    },
    {
      "epoch": 139.41,
      "grad_norm": 1.8305363655090332,
      "learning_rate": 1.081157581071774e-06,
      "loss": 0.4311,
      "step": 340570
    },
    {
      "epoch": 139.41,
      "grad_norm": 2.0495424270629883,
      "learning_rate": 1.0810239395581225e-06,
      "loss": 0.4698,
      "step": 340580
    },
    {
      "epoch": 139.41,
      "grad_norm": 1.8711522817611694,
      "learning_rate": 1.0808903038288226e-06,
      "loss": 0.4521,
      "step": 340590
    },
    {
      "epoch": 139.42,
      "grad_norm": 2.2504684925079346,
      "learning_rate": 1.0807566738844874e-06,
      "loss": 0.4584,
      "step": 340600
    },
    {
      "epoch": 139.42,
      "grad_norm": 1.917131781578064,
      "learning_rate": 1.0806230497257289e-06,
      "loss": 0.4573,
      "step": 340610
    },
    {
      "epoch": 139.43,
      "grad_norm": 2.16444993019104,
      "learning_rate": 1.0804894313531595e-06,
      "loss": 0.4773,
      "step": 340620
    },
    {
      "epoch": 139.43,
      "grad_norm": 2.1667075157165527,
      "learning_rate": 1.080355818767391e-06,
      "loss": 0.4433,
      "step": 340630
    },
    {
      "epoch": 139.44,
      "grad_norm": 2.1101114749908447,
      "learning_rate": 1.0802222119690345e-06,
      "loss": 0.4545,
      "step": 340640
    },
    {
      "epoch": 139.44,
      "grad_norm": 1.9321379661560059,
      "learning_rate": 1.0800886109587043e-06,
      "loss": 0.4487,
      "step": 340650
    },
    {
      "epoch": 139.44,
      "grad_norm": 1.860996127128601,
      "learning_rate": 1.0799550157370112e-06,
      "loss": 0.4435,
      "step": 340660
    },
    {
      "epoch": 139.45,
      "grad_norm": 2.11787486076355,
      "learning_rate": 1.0798214263045672e-06,
      "loss": 0.4605,
      "step": 340670
    },
    {
      "epoch": 139.45,
      "grad_norm": 2.0387110710144043,
      "learning_rate": 1.0796878426619844e-06,
      "loss": 0.4385,
      "step": 340680
    },
    {
      "epoch": 139.46,
      "grad_norm": 1.9074137210845947,
      "learning_rate": 1.0795542648098742e-06,
      "loss": 0.4569,
      "step": 340690
    },
    {
      "epoch": 139.46,
      "grad_norm": 2.2719268798828125,
      "learning_rate": 1.0794206927488492e-06,
      "loss": 0.4719,
      "step": 340700
    },
    {
      "epoch": 139.46,
      "grad_norm": 1.9321258068084717,
      "learning_rate": 1.0792871264795215e-06,
      "loss": 0.4652,
      "step": 340710
    },
    {
      "epoch": 139.47,
      "grad_norm": 2.208050489425659,
      "learning_rate": 1.0791535660025018e-06,
      "loss": 0.4682,
      "step": 340720
    },
    {
      "epoch": 139.47,
      "grad_norm": 1.7535556554794312,
      "learning_rate": 1.0790200113184025e-06,
      "loss": 0.4406,
      "step": 340730
    },
    {
      "epoch": 139.48,
      "grad_norm": 2.579451084136963,
      "learning_rate": 1.0788864624278353e-06,
      "loss": 0.469,
      "step": 340740
    },
    {
      "epoch": 139.48,
      "grad_norm": 2.013381004333496,
      "learning_rate": 1.078752919331412e-06,
      "loss": 0.4575,
      "step": 340750
    },
    {
      "epoch": 139.48,
      "grad_norm": 1.7436162233352661,
      "learning_rate": 1.0786193820297446e-06,
      "loss": 0.4611,
      "step": 340760
    },
    {
      "epoch": 139.49,
      "grad_norm": 2.2347123622894287,
      "learning_rate": 1.0784858505234446e-06,
      "loss": 0.4531,
      "step": 340770
    },
    {
      "epoch": 139.49,
      "grad_norm": 1.7826685905456543,
      "learning_rate": 1.0783523248131239e-06,
      "loss": 0.4619,
      "step": 340780
    },
    {
      "epoch": 139.5,
      "grad_norm": 1.8133569955825806,
      "learning_rate": 1.0782188048993935e-06,
      "loss": 0.4544,
      "step": 340790
    },
    {
      "epoch": 139.5,
      "grad_norm": 1.5513001680374146,
      "learning_rate": 1.0780852907828659e-06,
      "loss": 0.4634,
      "step": 340800
    },
    {
      "epoch": 139.5,
      "grad_norm": 1.837146520614624,
      "learning_rate": 1.0779517824641521e-06,
      "loss": 0.4409,
      "step": 340810
    },
    {
      "epoch": 139.51,
      "grad_norm": 1.9576990604400635,
      "learning_rate": 1.077818279943864e-06,
      "loss": 0.4232,
      "step": 340820
    },
    {
      "epoch": 139.51,
      "grad_norm": 1.8420782089233398,
      "learning_rate": 1.0776847832226122e-06,
      "loss": 0.4439,
      "step": 340830
    },
    {
      "epoch": 139.52,
      "grad_norm": 1.7717336416244507,
      "learning_rate": 1.07755129230101e-06,
      "loss": 0.4751,
      "step": 340840
    },
    {
      "epoch": 139.52,
      "grad_norm": 1.67030930519104,
      "learning_rate": 1.0774178071796688e-06,
      "loss": 0.4355,
      "step": 340850
    },
    {
      "epoch": 139.53,
      "grad_norm": 2.0754427909851074,
      "learning_rate": 1.0772843278591983e-06,
      "loss": 0.4439,
      "step": 340860
    },
    {
      "epoch": 139.53,
      "grad_norm": 1.6857078075408936,
      "learning_rate": 1.077150854340211e-06,
      "loss": 0.4118,
      "step": 340870
    },
    {
      "epoch": 139.53,
      "grad_norm": 1.9089447259902954,
      "learning_rate": 1.0770173866233182e-06,
      "loss": 0.4498,
      "step": 340880
    },
    {
      "epoch": 139.54,
      "grad_norm": 2.283557415008545,
      "learning_rate": 1.0768839247091313e-06,
      "loss": 0.4423,
      "step": 340890
    },
    {
      "epoch": 139.54,
      "grad_norm": 2.290452480316162,
      "learning_rate": 1.0767504685982616e-06,
      "loss": 0.449,
      "step": 340900
    },
    {
      "epoch": 139.55,
      "grad_norm": 1.8645451068878174,
      "learning_rate": 1.0766170182913208e-06,
      "loss": 0.4527,
      "step": 340910
    },
    {
      "epoch": 139.55,
      "grad_norm": 1.8655208349227905,
      "learning_rate": 1.07648357378892e-06,
      "loss": 0.4807,
      "step": 340920
    },
    {
      "epoch": 139.55,
      "grad_norm": 2.21309494972229,
      "learning_rate": 1.0763501350916702e-06,
      "loss": 0.4698,
      "step": 340930
    },
    {
      "epoch": 139.56,
      "grad_norm": 2.348314046859741,
      "learning_rate": 1.0762167022001829e-06,
      "loss": 0.4472,
      "step": 340940
    },
    {
      "epoch": 139.56,
      "grad_norm": 2.2852466106414795,
      "learning_rate": 1.0760832751150692e-06,
      "loss": 0.4723,
      "step": 340950
    },
    {
      "epoch": 139.57,
      "grad_norm": 2.1128506660461426,
      "learning_rate": 1.0759498538369408e-06,
      "loss": 0.4473,
      "step": 340960
    },
    {
      "epoch": 139.57,
      "grad_norm": 1.823195457458496,
      "learning_rate": 1.0758164383664083e-06,
      "loss": 0.4461,
      "step": 340970
    },
    {
      "epoch": 139.57,
      "grad_norm": 1.851147174835205,
      "learning_rate": 1.0756830287040832e-06,
      "loss": 0.4437,
      "step": 340980
    },
    {
      "epoch": 139.58,
      "grad_norm": 2.115999937057495,
      "learning_rate": 1.0755496248505764e-06,
      "loss": 0.4315,
      "step": 340990
    },
    {
      "epoch": 139.58,
      "grad_norm": 1.7394658327102661,
      "learning_rate": 1.0754162268064991e-06,
      "loss": 0.4623,
      "step": 341000
    },
    {
      "epoch": 139.59,
      "grad_norm": 2.385129451751709,
      "learning_rate": 1.0752828345724624e-06,
      "loss": 0.437,
      "step": 341010
    },
    {
      "epoch": 139.59,
      "grad_norm": 1.8596831560134888,
      "learning_rate": 1.0751494481490775e-06,
      "loss": 0.4436,
      "step": 341020
    },
    {
      "epoch": 139.59,
      "grad_norm": 1.717071771621704,
      "learning_rate": 1.0750160675369552e-06,
      "loss": 0.4424,
      "step": 341030
    },
    {
      "epoch": 139.6,
      "grad_norm": 2.118654489517212,
      "learning_rate": 1.0748826927367065e-06,
      "loss": 0.4397,
      "step": 341040
    },
    {
      "epoch": 139.6,
      "grad_norm": 2.37302303314209,
      "learning_rate": 1.0747493237489422e-06,
      "loss": 0.4658,
      "step": 341050
    },
    {
      "epoch": 139.61,
      "grad_norm": 2.0767531394958496,
      "learning_rate": 1.0746159605742739e-06,
      "loss": 0.4648,
      "step": 341060
    },
    {
      "epoch": 139.61,
      "grad_norm": 2.069775342941284,
      "learning_rate": 1.074482603213312e-06,
      "loss": 0.4538,
      "step": 341070
    },
    {
      "epoch": 139.62,
      "grad_norm": 2.104597568511963,
      "learning_rate": 1.0743492516666673e-06,
      "loss": 0.4454,
      "step": 341080
    },
    {
      "epoch": 139.62,
      "grad_norm": 2.0185117721557617,
      "learning_rate": 1.074215905934951e-06,
      "loss": 0.4628,
      "step": 341090
    },
    {
      "epoch": 139.62,
      "grad_norm": 1.8031134605407715,
      "learning_rate": 1.0740825660187735e-06,
      "loss": 0.4425,
      "step": 341100
    },
    {
      "epoch": 139.63,
      "grad_norm": 2.2113492488861084,
      "learning_rate": 1.0739492319187461e-06,
      "loss": 0.46,
      "step": 341110
    },
    {
      "epoch": 139.63,
      "grad_norm": 2.2313709259033203,
      "learning_rate": 1.0738159036354795e-06,
      "loss": 0.436,
      "step": 341120
    },
    {
      "epoch": 139.64,
      "grad_norm": 1.8381785154342651,
      "learning_rate": 1.073682581169584e-06,
      "loss": 0.4442,
      "step": 341130
    },
    {
      "epoch": 139.64,
      "grad_norm": 2.0195136070251465,
      "learning_rate": 1.0735492645216706e-06,
      "loss": 0.473,
      "step": 341140
    },
    {
      "epoch": 139.64,
      "grad_norm": 1.7793532609939575,
      "learning_rate": 1.0734159536923502e-06,
      "loss": 0.4521,
      "step": 341150
    },
    {
      "epoch": 139.65,
      "grad_norm": 1.6940637826919556,
      "learning_rate": 1.0732826486822345e-06,
      "loss": 0.4391,
      "step": 341160
    },
    {
      "epoch": 139.65,
      "grad_norm": 2.0135836601257324,
      "learning_rate": 1.0731493494919317e-06,
      "loss": 0.4454,
      "step": 341170
    },
    {
      "epoch": 139.66,
      "grad_norm": 1.9959909915924072,
      "learning_rate": 1.0730160561220538e-06,
      "loss": 0.445,
      "step": 341180
    },
    {
      "epoch": 139.66,
      "grad_norm": 2.399831533432007,
      "learning_rate": 1.0728827685732112e-06,
      "loss": 0.4402,
      "step": 341190
    },
    {
      "epoch": 139.66,
      "grad_norm": 1.9200252294540405,
      "learning_rate": 1.0727494868460149e-06,
      "loss": 0.4247,
      "step": 341200
    },
    {
      "epoch": 139.67,
      "grad_norm": 2.32143497467041,
      "learning_rate": 1.072616210941074e-06,
      "loss": 0.4547,
      "step": 341210
    },
    {
      "epoch": 139.67,
      "grad_norm": 2.453325033187866,
      "learning_rate": 1.0724829408590012e-06,
      "loss": 0.468,
      "step": 341220
    },
    {
      "epoch": 139.68,
      "grad_norm": 2.6582095623016357,
      "learning_rate": 1.0723496766004056e-06,
      "loss": 0.4404,
      "step": 341230
    },
    {
      "epoch": 139.68,
      "grad_norm": 1.6782299280166626,
      "learning_rate": 1.0722164181658983e-06,
      "loss": 0.4447,
      "step": 341240
    },
    {
      "epoch": 139.68,
      "grad_norm": 1.6340230703353882,
      "learning_rate": 1.0720831655560893e-06,
      "loss": 0.4594,
      "step": 341250
    },
    {
      "epoch": 139.69,
      "grad_norm": 1.622393012046814,
      "learning_rate": 1.071949918771589e-06,
      "loss": 0.4365,
      "step": 341260
    },
    {
      "epoch": 139.69,
      "grad_norm": 1.9085819721221924,
      "learning_rate": 1.0718166778130082e-06,
      "loss": 0.4689,
      "step": 341270
    },
    {
      "epoch": 139.7,
      "grad_norm": 1.851787805557251,
      "learning_rate": 1.0716834426809566e-06,
      "loss": 0.4645,
      "step": 341280
    },
    {
      "epoch": 139.7,
      "grad_norm": 2.0773799419403076,
      "learning_rate": 1.0715502133760451e-06,
      "loss": 0.4632,
      "step": 341290
    },
    {
      "epoch": 139.71,
      "grad_norm": 1.846023440361023,
      "learning_rate": 1.0714169898988846e-06,
      "loss": 0.4604,
      "step": 341300
    },
    {
      "epoch": 139.71,
      "grad_norm": 1.7051632404327393,
      "learning_rate": 1.0712837722500837e-06,
      "loss": 0.4653,
      "step": 341310
    },
    {
      "epoch": 139.71,
      "grad_norm": 1.7404632568359375,
      "learning_rate": 1.0711505604302536e-06,
      "loss": 0.4609,
      "step": 341320
    },
    {
      "epoch": 139.72,
      "grad_norm": 2.5503323078155518,
      "learning_rate": 1.0710173544400047e-06,
      "loss": 0.4306,
      "step": 341330
    },
    {
      "epoch": 139.72,
      "grad_norm": 2.116182327270508,
      "learning_rate": 1.0708841542799466e-06,
      "loss": 0.4531,
      "step": 341340
    },
    {
      "epoch": 139.73,
      "grad_norm": 2.056103467941284,
      "learning_rate": 1.0707509599506897e-06,
      "loss": 0.4423,
      "step": 341350
    },
    {
      "epoch": 139.73,
      "grad_norm": 1.7924549579620361,
      "learning_rate": 1.0706177714528445e-06,
      "loss": 0.4768,
      "step": 341360
    },
    {
      "epoch": 139.73,
      "grad_norm": 2.4499220848083496,
      "learning_rate": 1.0704845887870206e-06,
      "loss": 0.4449,
      "step": 341370
    },
    {
      "epoch": 139.74,
      "grad_norm": 1.8527063131332397,
      "learning_rate": 1.0703514119538285e-06,
      "loss": 0.4414,
      "step": 341380
    },
    {
      "epoch": 139.74,
      "grad_norm": 1.90541410446167,
      "learning_rate": 1.0702182409538773e-06,
      "loss": 0.4758,
      "step": 341390
    },
    {
      "epoch": 139.75,
      "grad_norm": 1.9221218824386597,
      "learning_rate": 1.0700850757877789e-06,
      "loss": 0.4308,
      "step": 341400
    },
    {
      "epoch": 139.75,
      "grad_norm": 2.508869171142578,
      "learning_rate": 1.0699519164561418e-06,
      "loss": 0.4597,
      "step": 341410
    },
    {
      "epoch": 139.75,
      "grad_norm": 2.459171772003174,
      "learning_rate": 1.069818762959577e-06,
      "loss": 0.441,
      "step": 341420
    },
    {
      "epoch": 139.76,
      "grad_norm": 1.8475497961044312,
      "learning_rate": 1.0696856152986944e-06,
      "loss": 0.4633,
      "step": 341430
    },
    {
      "epoch": 139.76,
      "grad_norm": 1.790122628211975,
      "learning_rate": 1.0695524734741027e-06,
      "loss": 0.4498,
      "step": 341440
    },
    {
      "epoch": 139.77,
      "grad_norm": 1.863321304321289,
      "learning_rate": 1.0694193374864125e-06,
      "loss": 0.4371,
      "step": 341450
    },
    {
      "epoch": 139.77,
      "grad_norm": 2.1451528072357178,
      "learning_rate": 1.0692862073362339e-06,
      "loss": 0.4611,
      "step": 341460
    },
    {
      "epoch": 139.77,
      "grad_norm": 2.3434536457061768,
      "learning_rate": 1.0691530830241763e-06,
      "loss": 0.4572,
      "step": 341470
    },
    {
      "epoch": 139.78,
      "grad_norm": 1.707011103630066,
      "learning_rate": 1.06901996455085e-06,
      "loss": 0.4679,
      "step": 341480
    },
    {
      "epoch": 139.78,
      "grad_norm": 2.4269115924835205,
      "learning_rate": 1.0688868519168641e-06,
      "loss": 0.4358,
      "step": 341490
    },
    {
      "epoch": 139.79,
      "grad_norm": 1.7294358015060425,
      "learning_rate": 1.0687537451228294e-06,
      "loss": 0.4459,
      "step": 341500
    },
    {
      "epoch": 139.79,
      "grad_norm": 2.0948023796081543,
      "learning_rate": 1.0686206441693547e-06,
      "loss": 0.4541,
      "step": 341510
    },
    {
      "epoch": 139.8,
      "grad_norm": 1.9637805223464966,
      "learning_rate": 1.0684875490570505e-06,
      "loss": 0.468,
      "step": 341520
    },
    {
      "epoch": 139.8,
      "grad_norm": 2.328601598739624,
      "learning_rate": 1.0683544597865257e-06,
      "loss": 0.4456,
      "step": 341530
    },
    {
      "epoch": 139.8,
      "grad_norm": 2.0373458862304688,
      "learning_rate": 1.0682213763583905e-06,
      "loss": 0.4722,
      "step": 341540
    },
    {
      "epoch": 139.81,
      "grad_norm": 1.9047905206680298,
      "learning_rate": 1.0680882987732544e-06,
      "loss": 0.4635,
      "step": 341550
    },
    {
      "epoch": 139.81,
      "grad_norm": 2.0474560260772705,
      "learning_rate": 1.0679552270317268e-06,
      "loss": 0.4378,
      "step": 341560
    },
    {
      "epoch": 139.82,
      "grad_norm": 2.152082920074463,
      "learning_rate": 1.067822161134418e-06,
      "loss": 0.4513,
      "step": 341570
    },
    {
      "epoch": 139.82,
      "grad_norm": 1.7457400560379028,
      "learning_rate": 1.0676891010819364e-06,
      "loss": 0.4598,
      "step": 341580
    },
    {
      "epoch": 139.82,
      "grad_norm": 1.7019497156143188,
      "learning_rate": 1.0675560468748925e-06,
      "loss": 0.4187,
      "step": 341590
    },
    {
      "epoch": 139.83,
      "grad_norm": 1.900172233581543,
      "learning_rate": 1.0674229985138955e-06,
      "loss": 0.4321,
      "step": 341600
    },
    {
      "epoch": 139.83,
      "grad_norm": 2.163635492324829,
      "learning_rate": 1.0672899559995546e-06,
      "loss": 0.454,
      "step": 341610
    },
    {
      "epoch": 139.84,
      "grad_norm": 2.0113959312438965,
      "learning_rate": 1.0671569193324795e-06,
      "loss": 0.4559,
      "step": 341620
    },
    {
      "epoch": 139.84,
      "grad_norm": 2.0007669925689697,
      "learning_rate": 1.0670238885132799e-06,
      "loss": 0.4343,
      "step": 341630
    },
    {
      "epoch": 139.84,
      "grad_norm": 2.016103506088257,
      "learning_rate": 1.0668908635425645e-06,
      "loss": 0.4506,
      "step": 341640
    },
    {
      "epoch": 139.85,
      "grad_norm": 2.4375369548797607,
      "learning_rate": 1.0667578444209433e-06,
      "loss": 0.4334,
      "step": 341650
    },
    {
      "epoch": 139.85,
      "grad_norm": 2.328259229660034,
      "learning_rate": 1.0666248311490253e-06,
      "loss": 0.4436,
      "step": 341660
    },
    {
      "epoch": 139.86,
      "grad_norm": 2.440420627593994,
      "learning_rate": 1.06649182372742e-06,
      "loss": 0.4347,
      "step": 341670
    },
    {
      "epoch": 139.86,
      "grad_norm": 1.938281774520874,
      "learning_rate": 1.0663588221567367e-06,
      "loss": 0.4321,
      "step": 341680
    },
    {
      "epoch": 139.86,
      "grad_norm": 2.595432996749878,
      "learning_rate": 1.0662258264375841e-06,
      "loss": 0.4814,
      "step": 341690
    },
    {
      "epoch": 139.87,
      "grad_norm": 2.306973695755005,
      "learning_rate": 1.0660928365705725e-06,
      "loss": 0.4465,
      "step": 341700
    },
    {
      "epoch": 139.87,
      "grad_norm": 1.9586386680603027,
      "learning_rate": 1.06595985255631e-06,
      "loss": 0.4762,
      "step": 341710
    },
    {
      "epoch": 139.88,
      "grad_norm": 2.125767230987549,
      "learning_rate": 1.0658268743954065e-06,
      "loss": 0.4586,
      "step": 341720
    },
    {
      "epoch": 139.88,
      "grad_norm": 1.9754221439361572,
      "learning_rate": 1.0656939020884708e-06,
      "loss": 0.4531,
      "step": 341730
    },
    {
      "epoch": 139.89,
      "grad_norm": 1.6207146644592285,
      "learning_rate": 1.0655609356361131e-06,
      "loss": 0.4514,
      "step": 341740
    },
    {
      "epoch": 139.89,
      "grad_norm": 1.7297471761703491,
      "learning_rate": 1.0654279750389408e-06,
      "loss": 0.4797,
      "step": 341750
    },
    {
      "epoch": 139.89,
      "grad_norm": 2.1044671535491943,
      "learning_rate": 1.0652950202975637e-06,
      "loss": 0.4644,
      "step": 341760
    },
    {
      "epoch": 139.9,
      "grad_norm": 1.8407418727874756,
      "learning_rate": 1.0651620714125899e-06,
      "loss": 0.4593,
      "step": 341770
    },
    {
      "epoch": 139.9,
      "grad_norm": 1.8903790712356567,
      "learning_rate": 1.0650291283846304e-06,
      "loss": 0.4648,
      "step": 341780
    },
    {
      "epoch": 139.91,
      "grad_norm": 1.8408759832382202,
      "learning_rate": 1.064896191214293e-06,
      "loss": 0.4687,
      "step": 341790
    },
    {
      "epoch": 139.91,
      "grad_norm": 2.0694055557250977,
      "learning_rate": 1.0647632599021868e-06,
      "loss": 0.4569,
      "step": 341800
    },
    {
      "epoch": 139.91,
      "grad_norm": 2.257648229598999,
      "learning_rate": 1.0646303344489208e-06,
      "loss": 0.4479,
      "step": 341810
    },
    {
      "epoch": 139.92,
      "grad_norm": 1.834916353225708,
      "learning_rate": 1.064497414855104e-06,
      "loss": 0.4325,
      "step": 341820
    },
    {
      "epoch": 139.92,
      "grad_norm": 2.0691137313842773,
      "learning_rate": 1.064364501121345e-06,
      "loss": 0.4263,
      "step": 341830
    },
    {
      "epoch": 139.93,
      "grad_norm": 2.0520987510681152,
      "learning_rate": 1.064231593248253e-06,
      "loss": 0.4469,
      "step": 341840
    },
    {
      "epoch": 139.93,
      "grad_norm": 2.248880386352539,
      "learning_rate": 1.0640986912364367e-06,
      "loss": 0.4396,
      "step": 341850
    },
    {
      "epoch": 139.93,
      "grad_norm": 1.886549472808838,
      "learning_rate": 1.0639657950865047e-06,
      "loss": 0.4408,
      "step": 341860
    },
    {
      "epoch": 139.94,
      "grad_norm": 2.138641357421875,
      "learning_rate": 1.0638329047990663e-06,
      "loss": 0.4571,
      "step": 341870
    },
    {
      "epoch": 139.94,
      "grad_norm": 2.2487072944641113,
      "learning_rate": 1.0637000203747305e-06,
      "loss": 0.4642,
      "step": 341880
    },
    {
      "epoch": 139.95,
      "grad_norm": 1.7559057474136353,
      "learning_rate": 1.0635671418141043e-06,
      "loss": 0.4536,
      "step": 341890
    },
    {
      "epoch": 139.95,
      "grad_norm": 2.444207191467285,
      "learning_rate": 1.063434269117798e-06,
      "loss": 0.4526,
      "step": 341900
    },
    {
      "epoch": 139.95,
      "grad_norm": 2.015786647796631,
      "learning_rate": 1.0633014022864194e-06,
      "loss": 0.4559,
      "step": 341910
    },
    {
      "epoch": 139.96,
      "grad_norm": 2.1016845703125,
      "learning_rate": 1.0631685413205776e-06,
      "loss": 0.4486,
      "step": 341920
    },
    {
      "epoch": 139.96,
      "grad_norm": 5.201946258544922,
      "learning_rate": 1.0630356862208813e-06,
      "loss": 0.4556,
      "step": 341930
    },
    {
      "epoch": 139.97,
      "grad_norm": 1.6213181018829346,
      "learning_rate": 1.0629028369879387e-06,
      "loss": 0.4388,
      "step": 341940
    },
    {
      "epoch": 139.97,
      "grad_norm": 3.491244316101074,
      "learning_rate": 1.0627699936223588e-06,
      "loss": 0.4592,
      "step": 341950
    },
    {
      "epoch": 139.98,
      "grad_norm": 2.293675184249878,
      "learning_rate": 1.062637156124749e-06,
      "loss": 0.4795,
      "step": 341960
    },
    {
      "epoch": 139.98,
      "grad_norm": 2.181091070175171,
      "learning_rate": 1.0625043244957195e-06,
      "loss": 0.4664,
      "step": 341970
    },
    {
      "epoch": 139.98,
      "grad_norm": 1.9888309240341187,
      "learning_rate": 1.0623714987358784e-06,
      "loss": 0.4403,
      "step": 341980
    },
    {
      "epoch": 139.99,
      "grad_norm": 2.006443500518799,
      "learning_rate": 1.0622386788458334e-06,
      "loss": 0.4664,
      "step": 341990
    },
    {
      "epoch": 139.99,
      "grad_norm": 1.7816731929779053,
      "learning_rate": 1.0621058648261932e-06,
      "loss": 0.4549,
      "step": 342000
    },
    {
      "epoch": 140.0,
      "grad_norm": 2.0129504203796387,
      "learning_rate": 1.0619730566775673e-06,
      "loss": 0.4695,
      "step": 342010
    },
    {
      "epoch": 140.0,
      "grad_norm": 1.9259334802627563,
      "learning_rate": 1.0618402544005624e-06,
      "loss": 0.4734,
      "step": 342020
    },
    {
      "epoch": 140.0,
      "eval_loss": 0.45281776785850525,
      "eval_runtime": 53.1634,
      "eval_samples_per_second": 64.876,
      "eval_steps_per_second": 8.126,
      "step": 342020
    },
    {
      "epoch": 140.0,
      "grad_norm": 1.969216227531433,
      "learning_rate": 1.0617074579957874e-06,
      "loss": 0.4391,
      "step": 342030
    },
    {
      "epoch": 140.01,
      "grad_norm": 2.375797986984253,
      "learning_rate": 1.0615746674638505e-06,
      "loss": 0.4455,
      "step": 342040
    },
    {
      "epoch": 140.01,
      "grad_norm": 2.0582075119018555,
      "learning_rate": 1.0614418828053606e-06,
      "loss": 0.4555,
      "step": 342050
    },
    {
      "epoch": 140.02,
      "grad_norm": 2.5751497745513916,
      "learning_rate": 1.0613091040209252e-06,
      "loss": 0.4626,
      "step": 342060
    },
    {
      "epoch": 140.02,
      "grad_norm": 2.7463412284851074,
      "learning_rate": 1.061176331111153e-06,
      "loss": 0.4559,
      "step": 342070
    },
    {
      "epoch": 140.02,
      "grad_norm": 2.344888210296631,
      "learning_rate": 1.0610435640766526e-06,
      "loss": 0.4367,
      "step": 342080
    },
    {
      "epoch": 140.03,
      "grad_norm": 1.9559110403060913,
      "learning_rate": 1.0609108029180313e-06,
      "loss": 0.4631,
      "step": 342090
    },
    {
      "epoch": 140.03,
      "grad_norm": 2.2385499477386475,
      "learning_rate": 1.0607780476358979e-06,
      "loss": 0.4495,
      "step": 342100
    },
    {
      "epoch": 140.04,
      "grad_norm": 2.2067480087280273,
      "learning_rate": 1.0606452982308602e-06,
      "loss": 0.4572,
      "step": 342110
    },
    {
      "epoch": 140.04,
      "grad_norm": 1.719775676727295,
      "learning_rate": 1.0605125547035264e-06,
      "loss": 0.4382,
      "step": 342120
    },
    {
      "epoch": 140.05,
      "grad_norm": 2.109680652618408,
      "learning_rate": 1.0603798170545047e-06,
      "loss": 0.4521,
      "step": 342130
    },
    {
      "epoch": 140.05,
      "grad_norm": 2.1484060287475586,
      "learning_rate": 1.0602470852844028e-06,
      "loss": 0.4396,
      "step": 342140
    },
    {
      "epoch": 140.05,
      "grad_norm": 1.845878005027771,
      "learning_rate": 1.0601143593938294e-06,
      "loss": 0.4449,
      "step": 342150
    },
    {
      "epoch": 140.06,
      "grad_norm": 2.2593417167663574,
      "learning_rate": 1.0599816393833917e-06,
      "loss": 0.4599,
      "step": 342160
    },
    {
      "epoch": 140.06,
      "grad_norm": 1.712956428527832,
      "learning_rate": 1.0598489252536984e-06,
      "loss": 0.4562,
      "step": 342170
    },
    {
      "epoch": 140.07,
      "grad_norm": 1.7511614561080933,
      "learning_rate": 1.0597162170053567e-06,
      "loss": 0.4405,
      "step": 342180
    },
    {
      "epoch": 140.07,
      "grad_norm": 1.945633053779602,
      "learning_rate": 1.059583514638975e-06,
      "loss": 0.4514,
      "step": 342190
    },
    {
      "epoch": 140.07,
      "grad_norm": 2.0845284461975098,
      "learning_rate": 1.0594508181551613e-06,
      "loss": 0.4485,
      "step": 342200
    },
    {
      "epoch": 140.08,
      "grad_norm": 2.1017704010009766,
      "learning_rate": 1.059318127554523e-06,
      "loss": 0.455,
      "step": 342210
    },
    {
      "epoch": 140.08,
      "grad_norm": 2.0623509883880615,
      "learning_rate": 1.0591854428376684e-06,
      "loss": 0.4472,
      "step": 342220
    },
    {
      "epoch": 140.09,
      "grad_norm": 2.395568370819092,
      "learning_rate": 1.0590527640052051e-06,
      "loss": 0.4567,
      "step": 342230
    },
    {
      "epoch": 140.09,
      "grad_norm": 1.8926843404769897,
      "learning_rate": 1.0589200910577406e-06,
      "loss": 0.4534,
      "step": 342240
    },
    {
      "epoch": 140.09,
      "grad_norm": 2.354843854904175,
      "learning_rate": 1.0587874239958835e-06,
      "loss": 0.4657,
      "step": 342250
    },
    {
      "epoch": 140.1,
      "grad_norm": 1.754119634628296,
      "learning_rate": 1.0586547628202402e-06,
      "loss": 0.4461,
      "step": 342260
    },
    {
      "epoch": 140.1,
      "grad_norm": 2.216646432876587,
      "learning_rate": 1.0585221075314197e-06,
      "loss": 0.4597,
      "step": 342270
    },
    {
      "epoch": 140.11,
      "grad_norm": 2.1359386444091797,
      "learning_rate": 1.0583894581300289e-06,
      "loss": 0.4633,
      "step": 342280
    },
    {
      "epoch": 140.11,
      "grad_norm": 2.2318921089172363,
      "learning_rate": 1.058256814616676e-06,
      "loss": 0.4579,
      "step": 342290
    },
    {
      "epoch": 140.11,
      "grad_norm": 2.0473878383636475,
      "learning_rate": 1.0581241769919678e-06,
      "loss": 0.4637,
      "step": 342300
    },
    {
      "epoch": 140.12,
      "grad_norm": 2.1120645999908447,
      "learning_rate": 1.057991545256513e-06,
      "loss": 0.4413,
      "step": 342310
    },
    {
      "epoch": 140.12,
      "grad_norm": 1.92539381980896,
      "learning_rate": 1.0578589194109188e-06,
      "loss": 0.4534,
      "step": 342320
    },
    {
      "epoch": 140.13,
      "grad_norm": 1.9722381830215454,
      "learning_rate": 1.0577262994557919e-06,
      "loss": 0.457,
      "step": 342330
    },
    {
      "epoch": 140.13,
      "grad_norm": 1.9855433702468872,
      "learning_rate": 1.0575936853917395e-06,
      "loss": 0.493,
      "step": 342340
    },
    {
      "epoch": 140.14,
      "grad_norm": 2.237802028656006,
      "learning_rate": 1.057461077219371e-06,
      "loss": 0.4605,
      "step": 342350
    },
    {
      "epoch": 140.14,
      "grad_norm": 2.077017068862915,
      "learning_rate": 1.0573284749392926e-06,
      "loss": 0.4463,
      "step": 342360
    },
    {
      "epoch": 140.14,
      "grad_norm": 2.4039149284362793,
      "learning_rate": 1.057195878552112e-06,
      "loss": 0.4359,
      "step": 342370
    },
    {
      "epoch": 140.15,
      "grad_norm": 1.732655644416809,
      "learning_rate": 1.0570632880584368e-06,
      "loss": 0.4321,
      "step": 342380
    },
    {
      "epoch": 140.15,
      "grad_norm": 1.7227401733398438,
      "learning_rate": 1.0569307034588743e-06,
      "loss": 0.4514,
      "step": 342390
    },
    {
      "epoch": 140.16,
      "grad_norm": 1.9328696727752686,
      "learning_rate": 1.0567981247540314e-06,
      "loss": 0.4437,
      "step": 342400
    },
    {
      "epoch": 140.16,
      "grad_norm": 1.784227967262268,
      "learning_rate": 1.0566655519445157e-06,
      "loss": 0.4461,
      "step": 342410
    },
    {
      "epoch": 140.16,
      "grad_norm": 1.9685713052749634,
      "learning_rate": 1.0565329850309348e-06,
      "loss": 0.4377,
      "step": 342420
    },
    {
      "epoch": 140.17,
      "grad_norm": 2.07470965385437,
      "learning_rate": 1.0564004240138954e-06,
      "loss": 0.459,
      "step": 342430
    },
    {
      "epoch": 140.17,
      "grad_norm": 1.7338989973068237,
      "learning_rate": 1.056267868894005e-06,
      "loss": 0.443,
      "step": 342440
    },
    {
      "epoch": 140.18,
      "grad_norm": 1.7216060161590576,
      "learning_rate": 1.056135319671871e-06,
      "loss": 0.4598,
      "step": 342450
    },
    {
      "epoch": 140.18,
      "grad_norm": 2.0180628299713135,
      "learning_rate": 1.0560027763481016e-06,
      "loss": 0.4741,
      "step": 342460
    },
    {
      "epoch": 140.18,
      "grad_norm": 2.4108383655548096,
      "learning_rate": 1.0558702389233014e-06,
      "loss": 0.4289,
      "step": 342470
    },
    {
      "epoch": 140.19,
      "grad_norm": 2.165847063064575,
      "learning_rate": 1.0557377073980792e-06,
      "loss": 0.4383,
      "step": 342480
    },
    {
      "epoch": 140.19,
      "grad_norm": 2.0704433917999268,
      "learning_rate": 1.0556051817730418e-06,
      "loss": 0.4514,
      "step": 342490
    },
    {
      "epoch": 140.2,
      "grad_norm": 1.956260085105896,
      "learning_rate": 1.0554726620487963e-06,
      "loss": 0.4728,
      "step": 342500
    },
    {
      "epoch": 140.2,
      "grad_norm": 1.653033971786499,
      "learning_rate": 1.0553401482259499e-06,
      "loss": 0.4625,
      "step": 342510
    },
    {
      "epoch": 140.2,
      "grad_norm": 2.1087920665740967,
      "learning_rate": 1.0552076403051094e-06,
      "loss": 0.4483,
      "step": 342520
    },
    {
      "epoch": 140.21,
      "grad_norm": 2.0855979919433594,
      "learning_rate": 1.055075138286881e-06,
      "loss": 0.4514,
      "step": 342530
    },
    {
      "epoch": 140.21,
      "grad_norm": 1.7706927061080933,
      "learning_rate": 1.0549426421718735e-06,
      "loss": 0.4517,
      "step": 342540
    },
    {
      "epoch": 140.22,
      "grad_norm": 1.839592456817627,
      "learning_rate": 1.0548101519606928e-06,
      "loss": 0.4529,
      "step": 342550
    },
    {
      "epoch": 140.22,
      "grad_norm": 2.279858112335205,
      "learning_rate": 1.054677667653946e-06,
      "loss": 0.4398,
      "step": 342560
    },
    {
      "epoch": 140.23,
      "grad_norm": 2.2252514362335205,
      "learning_rate": 1.05454518925224e-06,
      "loss": 0.4667,
      "step": 342570
    },
    {
      "epoch": 140.23,
      "grad_norm": 1.834399938583374,
      "learning_rate": 1.0544127167561814e-06,
      "loss": 0.4472,
      "step": 342580
    },
    {
      "epoch": 140.23,
      "grad_norm": 1.7172619104385376,
      "learning_rate": 1.0542802501663773e-06,
      "loss": 0.468,
      "step": 342590
    },
    {
      "epoch": 140.24,
      "grad_norm": 2.6813764572143555,
      "learning_rate": 1.0541477894834352e-06,
      "loss": 0.4512,
      "step": 342600
    },
    {
      "epoch": 140.24,
      "grad_norm": 2.8151440620422363,
      "learning_rate": 1.0540153347079604e-06,
      "loss": 0.4569,
      "step": 342610
    },
    {
      "epoch": 140.25,
      "grad_norm": 2.059905767440796,
      "learning_rate": 1.0538828858405604e-06,
      "loss": 0.4426,
      "step": 342620
    },
    {
      "epoch": 140.25,
      "grad_norm": 1.8431742191314697,
      "learning_rate": 1.0537504428818418e-06,
      "loss": 0.4454,
      "step": 342630
    },
    {
      "epoch": 140.25,
      "grad_norm": 1.9202314615249634,
      "learning_rate": 1.0536180058324112e-06,
      "loss": 0.4593,
      "step": 342640
    },
    {
      "epoch": 140.26,
      "grad_norm": 2.443819761276245,
      "learning_rate": 1.0534855746928758e-06,
      "loss": 0.4602,
      "step": 342650
    },
    {
      "epoch": 140.26,
      "grad_norm": 2.0779881477355957,
      "learning_rate": 1.0533531494638417e-06,
      "loss": 0.446,
      "step": 342660
    },
    {
      "epoch": 140.27,
      "grad_norm": 2.089036226272583,
      "learning_rate": 1.0532207301459158e-06,
      "loss": 0.4811,
      "step": 342670
    },
    {
      "epoch": 140.27,
      "grad_norm": 1.901281714439392,
      "learning_rate": 1.0530883167397047e-06,
      "loss": 0.4432,
      "step": 342680
    },
    {
      "epoch": 140.27,
      "grad_norm": 1.893951416015625,
      "learning_rate": 1.0529559092458148e-06,
      "loss": 0.4459,
      "step": 342690
    },
    {
      "epoch": 140.28,
      "grad_norm": 1.8236647844314575,
      "learning_rate": 1.0528235076648526e-06,
      "loss": 0.4363,
      "step": 342700
    },
    {
      "epoch": 140.28,
      "grad_norm": 2.0987112522125244,
      "learning_rate": 1.052691111997424e-06,
      "loss": 0.461,
      "step": 342710
    },
    {
      "epoch": 140.29,
      "grad_norm": 2.256908655166626,
      "learning_rate": 1.0525587222441372e-06,
      "loss": 0.4625,
      "step": 342720
    },
    {
      "epoch": 140.29,
      "grad_norm": 2.0045347213745117,
      "learning_rate": 1.0524263384055982e-06,
      "loss": 0.4534,
      "step": 342730
    },
    {
      "epoch": 140.29,
      "grad_norm": 2.1740145683288574,
      "learning_rate": 1.0522939604824123e-06,
      "loss": 0.4649,
      "step": 342740
    },
    {
      "epoch": 140.3,
      "grad_norm": 1.7882747650146484,
      "learning_rate": 1.0521615884751866e-06,
      "loss": 0.451,
      "step": 342750
    },
    {
      "epoch": 140.3,
      "grad_norm": 1.9584457874298096,
      "learning_rate": 1.052029222384527e-06,
      "loss": 0.4383,
      "step": 342760
    },
    {
      "epoch": 140.31,
      "grad_norm": 2.086541175842285,
      "learning_rate": 1.0518968622110405e-06,
      "loss": 0.4479,
      "step": 342770
    },
    {
      "epoch": 140.31,
      "grad_norm": 2.3380143642425537,
      "learning_rate": 1.0517645079553332e-06,
      "loss": 0.4377,
      "step": 342780
    },
    {
      "epoch": 140.32,
      "grad_norm": 1.9000868797302246,
      "learning_rate": 1.0516321596180112e-06,
      "loss": 0.4499,
      "step": 342790
    },
    {
      "epoch": 140.32,
      "grad_norm": 1.969462513923645,
      "learning_rate": 1.0514998171996811e-06,
      "loss": 0.4366,
      "step": 342800
    },
    {
      "epoch": 140.32,
      "grad_norm": 1.8309375047683716,
      "learning_rate": 1.0513674807009488e-06,
      "loss": 0.4349,
      "step": 342810
    },
    {
      "epoch": 140.33,
      "grad_norm": 2.363710641860962,
      "learning_rate": 1.0512351501224208e-06,
      "loss": 0.4768,
      "step": 342820
    },
    {
      "epoch": 140.33,
      "grad_norm": 1.8536330461502075,
      "learning_rate": 1.0511028254647032e-06,
      "loss": 0.4377,
      "step": 342830
    },
    {
      "epoch": 140.34,
      "grad_norm": 1.9801750183105469,
      "learning_rate": 1.0509705067284019e-06,
      "loss": 0.46,
      "step": 342840
    },
    {
      "epoch": 140.34,
      "grad_norm": 2.0958213806152344,
      "learning_rate": 1.0508381939141234e-06,
      "loss": 0.4302,
      "step": 342850
    },
    {
      "epoch": 140.34,
      "grad_norm": 1.9198625087738037,
      "learning_rate": 1.0507058870224737e-06,
      "loss": 0.4458,
      "step": 342860
    },
    {
      "epoch": 140.35,
      "grad_norm": 2.1736550331115723,
      "learning_rate": 1.0505735860540588e-06,
      "loss": 0.462,
      "step": 342870
    },
    {
      "epoch": 140.35,
      "grad_norm": 1.8903419971466064,
      "learning_rate": 1.0504412910094845e-06,
      "loss": 0.4437,
      "step": 342880
    },
    {
      "epoch": 140.36,
      "grad_norm": 1.796486735343933,
      "learning_rate": 1.0503090018893576e-06,
      "loss": 0.444,
      "step": 342890
    },
    {
      "epoch": 140.36,
      "grad_norm": 2.138357639312744,
      "learning_rate": 1.050176718694283e-06,
      "loss": 0.4528,
      "step": 342900
    },
    {
      "epoch": 140.36,
      "grad_norm": 1.844152569770813,
      "learning_rate": 1.0500444414248676e-06,
      "loss": 0.4397,
      "step": 342910
    },
    {
      "epoch": 140.37,
      "grad_norm": 1.7967612743377686,
      "learning_rate": 1.049912170081717e-06,
      "loss": 0.459,
      "step": 342920
    },
    {
      "epoch": 140.37,
      "grad_norm": 2.4239678382873535,
      "learning_rate": 1.0497799046654372e-06,
      "loss": 0.4459,
      "step": 342930
    },
    {
      "epoch": 140.38,
      "grad_norm": 2.116396427154541,
      "learning_rate": 1.049647645176634e-06,
      "loss": 0.4386,
      "step": 342940
    },
    {
      "epoch": 140.38,
      "grad_norm": 2.109034776687622,
      "learning_rate": 1.049515391615913e-06,
      "loss": 0.4411,
      "step": 342950
    },
    {
      "epoch": 140.38,
      "grad_norm": 2.0410163402557373,
      "learning_rate": 1.0493831439838807e-06,
      "loss": 0.453,
      "step": 342960
    },
    {
      "epoch": 140.39,
      "grad_norm": 1.8086234331130981,
      "learning_rate": 1.0492509022811421e-06,
      "loss": 0.4512,
      "step": 342970
    },
    {
      "epoch": 140.39,
      "grad_norm": 1.864909291267395,
      "learning_rate": 1.0491186665083036e-06,
      "loss": 0.4424,
      "step": 342980
    },
    {
      "epoch": 140.4,
      "grad_norm": 1.8730851411819458,
      "learning_rate": 1.0489864366659707e-06,
      "loss": 0.4473,
      "step": 342990
    },
    {
      "epoch": 140.4,
      "grad_norm": 1.965395450592041,
      "learning_rate": 1.0488542127547493e-06,
      "loss": 0.4521,
      "step": 343000
    },
    {
      "epoch": 140.41,
      "grad_norm": 2.2320258617401123,
      "learning_rate": 1.0487219947752448e-06,
      "loss": 0.4615,
      "step": 343010
    },
    {
      "epoch": 140.41,
      "grad_norm": 2.4855704307556152,
      "learning_rate": 1.0485897827280631e-06,
      "loss": 0.4496,
      "step": 343020
    },
    {
      "epoch": 140.41,
      "grad_norm": 1.3510535955429077,
      "learning_rate": 1.0484575766138097e-06,
      "loss": 0.4499,
      "step": 343030
    },
    {
      "epoch": 140.42,
      "grad_norm": 2.512352228164673,
      "learning_rate": 1.0483253764330913e-06,
      "loss": 0.4423,
      "step": 343040
    },
    {
      "epoch": 140.42,
      "grad_norm": 2.2081894874572754,
      "learning_rate": 1.0481931821865116e-06,
      "loss": 0.4412,
      "step": 343050
    },
    {
      "epoch": 140.43,
      "grad_norm": 2.029470205307007,
      "learning_rate": 1.048060993874677e-06,
      "loss": 0.4649,
      "step": 343060
    },
    {
      "epoch": 140.43,
      "grad_norm": 2.1501030921936035,
      "learning_rate": 1.047928811498193e-06,
      "loss": 0.4522,
      "step": 343070
    },
    {
      "epoch": 140.43,
      "grad_norm": 2.4730191230773926,
      "learning_rate": 1.0477966350576653e-06,
      "loss": 0.4563,
      "step": 343080
    },
    {
      "epoch": 140.44,
      "grad_norm": 1.8575917482376099,
      "learning_rate": 1.0476644645536982e-06,
      "loss": 0.4383,
      "step": 343090
    },
    {
      "epoch": 140.44,
      "grad_norm": 2.277362585067749,
      "learning_rate": 1.0475322999868994e-06,
      "loss": 0.4565,
      "step": 343100
    },
    {
      "epoch": 140.45,
      "grad_norm": 2.2379772663116455,
      "learning_rate": 1.0474001413578726e-06,
      "loss": 0.4472,
      "step": 343110
    },
    {
      "epoch": 140.45,
      "grad_norm": 2.2754619121551514,
      "learning_rate": 1.0472679886672242e-06,
      "loss": 0.4722,
      "step": 343120
    },
    {
      "epoch": 140.45,
      "grad_norm": 7.964319229125977,
      "learning_rate": 1.047135841915559e-06,
      "loss": 0.4453,
      "step": 343130
    },
    {
      "epoch": 140.46,
      "grad_norm": 2.0467164516448975,
      "learning_rate": 1.0470037011034825e-06,
      "loss": 0.4623,
      "step": 343140
    },
    {
      "epoch": 140.46,
      "grad_norm": 2.14597487449646,
      "learning_rate": 1.0468715662315998e-06,
      "loss": 0.434,
      "step": 343150
    },
    {
      "epoch": 140.47,
      "grad_norm": 2.031019926071167,
      "learning_rate": 1.0467394373005166e-06,
      "loss": 0.4486,
      "step": 343160
    },
    {
      "epoch": 140.47,
      "grad_norm": 2.0652146339416504,
      "learning_rate": 1.0466073143108377e-06,
      "loss": 0.4636,
      "step": 343170
    },
    {
      "epoch": 140.47,
      "grad_norm": 2.012529134750366,
      "learning_rate": 1.0464751972631695e-06,
      "loss": 0.4365,
      "step": 343180
    },
    {
      "epoch": 140.48,
      "grad_norm": 1.8124275207519531,
      "learning_rate": 1.0463430861581155e-06,
      "loss": 0.4553,
      "step": 343190
    },
    {
      "epoch": 140.48,
      "grad_norm": 2.055799961090088,
      "learning_rate": 1.0462109809962815e-06,
      "loss": 0.436,
      "step": 343200
    },
    {
      "epoch": 140.49,
      "grad_norm": 2.407447338104248,
      "learning_rate": 1.0460788817782727e-06,
      "loss": 0.4703,
      "step": 343210
    },
    {
      "epoch": 140.49,
      "grad_norm": 1.7388125658035278,
      "learning_rate": 1.0459467885046945e-06,
      "loss": 0.4744,
      "step": 343220
    },
    {
      "epoch": 140.5,
      "grad_norm": 1.8446879386901855,
      "learning_rate": 1.0458147011761518e-06,
      "loss": 0.454,
      "step": 343230
    },
    {
      "epoch": 140.5,
      "grad_norm": 2.1921298503875732,
      "learning_rate": 1.0456826197932497e-06,
      "loss": 0.4467,
      "step": 343240
    },
    {
      "epoch": 140.5,
      "grad_norm": 1.7759474515914917,
      "learning_rate": 1.045550544356593e-06,
      "loss": 0.4729,
      "step": 343250
    },
    {
      "epoch": 140.51,
      "grad_norm": 2.5501744747161865,
      "learning_rate": 1.0454184748667875e-06,
      "loss": 0.4326,
      "step": 343260
    },
    {
      "epoch": 140.51,
      "grad_norm": 2.3686559200286865,
      "learning_rate": 1.0452864113244372e-06,
      "loss": 0.4722,
      "step": 343270
    },
    {
      "epoch": 140.52,
      "grad_norm": 3.2624926567077637,
      "learning_rate": 1.0451543537301467e-06,
      "loss": 0.4392,
      "step": 343280
    },
    {
      "epoch": 140.52,
      "grad_norm": 2.1739397048950195,
      "learning_rate": 1.0450223020845226e-06,
      "loss": 0.4347,
      "step": 343290
    },
    {
      "epoch": 140.52,
      "grad_norm": 2.0487396717071533,
      "learning_rate": 1.0448902563881693e-06,
      "loss": 0.4367,
      "step": 343300
    },
    {
      "epoch": 140.53,
      "grad_norm": 1.885673999786377,
      "learning_rate": 1.044758216641691e-06,
      "loss": 0.4407,
      "step": 343310
    },
    {
      "epoch": 140.53,
      "grad_norm": 1.8498586416244507,
      "learning_rate": 1.0446261828456938e-06,
      "loss": 0.4653,
      "step": 343320
    },
    {
      "epoch": 140.54,
      "grad_norm": 3.6658504009246826,
      "learning_rate": 1.0444941550007807e-06,
      "loss": 0.4414,
      "step": 343330
    },
    {
      "epoch": 140.54,
      "grad_norm": 1.7966666221618652,
      "learning_rate": 1.0443621331075576e-06,
      "loss": 0.4283,
      "step": 343340
    },
    {
      "epoch": 140.54,
      "grad_norm": 1.9811416864395142,
      "learning_rate": 1.0442301171666287e-06,
      "loss": 0.4389,
      "step": 343350
    },
    {
      "epoch": 140.55,
      "grad_norm": 2.121915102005005,
      "learning_rate": 1.0440981071785993e-06,
      "loss": 0.4567,
      "step": 343360
    },
    {
      "epoch": 140.55,
      "grad_norm": 2.3205745220184326,
      "learning_rate": 1.0439661031440742e-06,
      "loss": 0.4682,
      "step": 343370
    },
    {
      "epoch": 140.56,
      "grad_norm": 4.3777971267700195,
      "learning_rate": 1.0438341050636575e-06,
      "loss": 0.4676,
      "step": 343380
    },
    {
      "epoch": 140.56,
      "grad_norm": 1.968711018562317,
      "learning_rate": 1.0437021129379546e-06,
      "loss": 0.4568,
      "step": 343390
    },
    {
      "epoch": 140.56,
      "grad_norm": 2.2562055587768555,
      "learning_rate": 1.0435701267675693e-06,
      "loss": 0.4539,
      "step": 343400
    },
    {
      "epoch": 140.57,
      "grad_norm": 2.2244315147399902,
      "learning_rate": 1.043438146553107e-06,
      "loss": 0.4567,
      "step": 343410
    },
    {
      "epoch": 140.57,
      "grad_norm": 1.9847413301467896,
      "learning_rate": 1.0433061722951717e-06,
      "loss": 0.4655,
      "step": 343420
    },
    {
      "epoch": 140.58,
      "grad_norm": 2.659836769104004,
      "learning_rate": 1.0431742039943681e-06,
      "loss": 0.4414,
      "step": 343430
    },
    {
      "epoch": 140.58,
      "grad_norm": 1.8393280506134033,
      "learning_rate": 1.0430422416513008e-06,
      "loss": 0.438,
      "step": 343440
    },
    {
      "epoch": 140.59,
      "grad_norm": 1.9027942419052124,
      "learning_rate": 1.0429102852665743e-06,
      "loss": 0.4493,
      "step": 343450
    },
    {
      "epoch": 140.59,
      "grad_norm": 2.2420592308044434,
      "learning_rate": 1.0427783348407935e-06,
      "loss": 0.4552,
      "step": 343460
    },
    {
      "epoch": 140.59,
      "grad_norm": 1.9251991510391235,
      "learning_rate": 1.0426463903745621e-06,
      "loss": 0.4309,
      "step": 343470
    },
    {
      "epoch": 140.6,
      "grad_norm": 1.7917253971099854,
      "learning_rate": 1.0425144518684849e-06,
      "loss": 0.4597,
      "step": 343480
    },
    {
      "epoch": 140.6,
      "grad_norm": 1.9722964763641357,
      "learning_rate": 1.042382519323166e-06,
      "loss": 0.4589,
      "step": 343490
    },
    {
      "epoch": 140.61,
      "grad_norm": 1.939304232597351,
      "learning_rate": 1.0422505927392102e-06,
      "loss": 0.4588,
      "step": 343500
    },
    {
      "epoch": 140.61,
      "grad_norm": 1.6721282005310059,
      "learning_rate": 1.0421186721172216e-06,
      "loss": 0.4556,
      "step": 343510
    },
    {
      "epoch": 140.61,
      "grad_norm": 2.08746600151062,
      "learning_rate": 1.0419867574578046e-06,
      "loss": 0.4508,
      "step": 343520
    },
    {
      "epoch": 140.62,
      "grad_norm": 2.074396848678589,
      "learning_rate": 1.0418548487615631e-06,
      "loss": 0.4687,
      "step": 343530
    },
    {
      "epoch": 140.62,
      "grad_norm": 2.256579637527466,
      "learning_rate": 1.041722946029102e-06,
      "loss": 0.4699,
      "step": 343540
    },
    {
      "epoch": 140.63,
      "grad_norm": 2.220541000366211,
      "learning_rate": 1.041591049261025e-06,
      "loss": 0.4562,
      "step": 343550
    },
    {
      "epoch": 140.63,
      "grad_norm": 2.738520860671997,
      "learning_rate": 1.0414591584579367e-06,
      "loss": 0.4556,
      "step": 343560
    },
    {
      "epoch": 140.63,
      "grad_norm": 2.727663040161133,
      "learning_rate": 1.0413272736204409e-06,
      "loss": 0.4466,
      "step": 343570
    },
    {
      "epoch": 140.64,
      "grad_norm": 2.264115333557129,
      "learning_rate": 1.0411953947491417e-06,
      "loss": 0.4562,
      "step": 343580
    },
    {
      "epoch": 140.64,
      "grad_norm": 1.678835391998291,
      "learning_rate": 1.0410635218446439e-06,
      "loss": 0.4569,
      "step": 343590
    },
    {
      "epoch": 140.65,
      "grad_norm": 1.9402356147766113,
      "learning_rate": 1.0409316549075508e-06,
      "loss": 0.4333,
      "step": 343600
    },
    {
      "epoch": 140.65,
      "grad_norm": 2.1306052207946777,
      "learning_rate": 1.0407997939384668e-06,
      "loss": 0.4433,
      "step": 343610
    },
    {
      "epoch": 140.65,
      "grad_norm": 2.202805280685425,
      "learning_rate": 1.0406679389379958e-06,
      "loss": 0.457,
      "step": 343620
    },
    {
      "epoch": 140.66,
      "grad_norm": 1.8785626888275146,
      "learning_rate": 1.040536089906743e-06,
      "loss": 0.4594,
      "step": 343630
    },
    {
      "epoch": 140.66,
      "grad_norm": 1.6726583242416382,
      "learning_rate": 1.0404042468453105e-06,
      "loss": 0.4612,
      "step": 343640
    },
    {
      "epoch": 140.67,
      "grad_norm": 1.781782627105713,
      "learning_rate": 1.0402724097543027e-06,
      "loss": 0.4657,
      "step": 343650
    },
    {
      "epoch": 140.67,
      "grad_norm": 1.9237865209579468,
      "learning_rate": 1.0401405786343233e-06,
      "loss": 0.442,
      "step": 343660
    },
    {
      "epoch": 140.68,
      "grad_norm": 2.342212200164795,
      "learning_rate": 1.0400087534859776e-06,
      "loss": 0.4357,
      "step": 343670
    },
    {
      "epoch": 140.68,
      "grad_norm": 1.8111095428466797,
      "learning_rate": 1.0398769343098685e-06,
      "loss": 0.448,
      "step": 343680
    },
    {
      "epoch": 140.68,
      "grad_norm": 1.759551763534546,
      "learning_rate": 1.0397451211066e-06,
      "loss": 0.438,
      "step": 343690
    },
    {
      "epoch": 140.69,
      "grad_norm": 2.185609817504883,
      "learning_rate": 1.039613313876776e-06,
      "loss": 0.4337,
      "step": 343700
    },
    {
      "epoch": 140.69,
      "grad_norm": 2.0144505500793457,
      "learning_rate": 1.0394815126210001e-06,
      "loss": 0.4514,
      "step": 343710
    },
    {
      "epoch": 140.7,
      "grad_norm": 1.766699194908142,
      "learning_rate": 1.0393497173398762e-06,
      "loss": 0.4386,
      "step": 343720
    },
    {
      "epoch": 140.7,
      "grad_norm": 2.0424768924713135,
      "learning_rate": 1.0392179280340081e-06,
      "loss": 0.4447,
      "step": 343730
    },
    {
      "epoch": 140.7,
      "grad_norm": 2.0877702236175537,
      "learning_rate": 1.0390861447039992e-06,
      "loss": 0.4541,
      "step": 343740
    },
    {
      "epoch": 140.71,
      "grad_norm": 2.241814613342285,
      "learning_rate": 1.0389543673504536e-06,
      "loss": 0.4355,
      "step": 343750
    },
    {
      "epoch": 140.71,
      "grad_norm": 1.788671612739563,
      "learning_rate": 1.0388225959739753e-06,
      "loss": 0.4441,
      "step": 343760
    },
    {
      "epoch": 140.72,
      "grad_norm": 2.440321922302246,
      "learning_rate": 1.0386908305751666e-06,
      "loss": 0.4461,
      "step": 343770
    },
    {
      "epoch": 140.72,
      "grad_norm": 2.258348226547241,
      "learning_rate": 1.0385590711546318e-06,
      "loss": 0.4297,
      "step": 343780
    },
    {
      "epoch": 140.72,
      "grad_norm": 1.8552160263061523,
      "learning_rate": 1.0384273177129745e-06,
      "loss": 0.463,
      "step": 343790
    },
    {
      "epoch": 140.73,
      "grad_norm": 2.019402027130127,
      "learning_rate": 1.0382955702507983e-06,
      "loss": 0.4556,
      "step": 343800
    },
    {
      "epoch": 140.73,
      "grad_norm": 2.096637487411499,
      "learning_rate": 1.0381638287687067e-06,
      "loss": 0.4378,
      "step": 343810
    },
    {
      "epoch": 140.74,
      "grad_norm": 2.02051043510437,
      "learning_rate": 1.0380320932673032e-06,
      "loss": 0.4574,
      "step": 343820
    },
    {
      "epoch": 140.74,
      "grad_norm": 2.1596853733062744,
      "learning_rate": 1.037900363747191e-06,
      "loss": 0.4474,
      "step": 343830
    },
    {
      "epoch": 140.74,
      "grad_norm": 2.126793622970581,
      "learning_rate": 1.037768640208974e-06,
      "loss": 0.4631,
      "step": 343840
    },
    {
      "epoch": 140.75,
      "grad_norm": 1.7632410526275635,
      "learning_rate": 1.0376369226532544e-06,
      "loss": 0.4401,
      "step": 343850
    },
    {
      "epoch": 140.75,
      "grad_norm": 2.012968063354492,
      "learning_rate": 1.0375052110806373e-06,
      "loss": 0.4597,
      "step": 343860
    },
    {
      "epoch": 140.76,
      "grad_norm": 2.281341791152954,
      "learning_rate": 1.0373735054917254e-06,
      "loss": 0.4546,
      "step": 343870
    },
    {
      "epoch": 140.76,
      "grad_norm": 1.8202083110809326,
      "learning_rate": 1.0372418058871217e-06,
      "loss": 0.4462,
      "step": 343880
    },
    {
      "epoch": 140.77,
      "grad_norm": 1.857593297958374,
      "learning_rate": 1.03711011226743e-06,
      "loss": 0.459,
      "step": 343890
    },
    {
      "epoch": 140.77,
      "grad_norm": 2.079228162765503,
      "learning_rate": 1.036978424633254e-06,
      "loss": 0.431,
      "step": 343900
    },
    {
      "epoch": 140.77,
      "grad_norm": 3.3954827785491943,
      "learning_rate": 1.0368467429851954e-06,
      "loss": 0.4446,
      "step": 343910
    },
    {
      "epoch": 140.78,
      "grad_norm": 2.4260377883911133,
      "learning_rate": 1.036715067323858e-06,
      "loss": 0.4497,
      "step": 343920
    },
    {
      "epoch": 140.78,
      "grad_norm": 1.969482660293579,
      "learning_rate": 1.0365833976498456e-06,
      "loss": 0.4697,
      "step": 343930
    },
    {
      "epoch": 140.79,
      "grad_norm": 1.933713436126709,
      "learning_rate": 1.0364517339637605e-06,
      "loss": 0.4335,
      "step": 343940
    },
    {
      "epoch": 140.79,
      "grad_norm": 2.3568122386932373,
      "learning_rate": 1.0363200762662065e-06,
      "loss": 0.4531,
      "step": 343950
    },
    {
      "epoch": 140.79,
      "grad_norm": 1.7559044361114502,
      "learning_rate": 1.0361884245577865e-06,
      "loss": 0.4539,
      "step": 343960
    },
    {
      "epoch": 140.8,
      "grad_norm": 1.90569007396698,
      "learning_rate": 1.0360567788391037e-06,
      "loss": 0.4554,
      "step": 343970
    },
    {
      "epoch": 140.8,
      "grad_norm": 2.6616227626800537,
      "learning_rate": 1.0359251391107607e-06,
      "loss": 0.4626,
      "step": 343980
    },
    {
      "epoch": 140.81,
      "grad_norm": 2.050879955291748,
      "learning_rate": 1.0357935053733613e-06,
      "loss": 0.4651,
      "step": 343990
    },
    {
      "epoch": 140.81,
      "grad_norm": 1.825881838798523,
      "learning_rate": 1.0356618776275078e-06,
      "loss": 0.4593,
      "step": 344000
    },
    {
      "epoch": 140.81,
      "grad_norm": 2.032552480697632,
      "learning_rate": 1.0355302558738032e-06,
      "loss": 0.4648,
      "step": 344010
    },
    {
      "epoch": 140.82,
      "grad_norm": 1.848752498626709,
      "learning_rate": 1.0353986401128508e-06,
      "loss": 0.4441,
      "step": 344020
    },
    {
      "epoch": 140.82,
      "grad_norm": 1.8242145776748657,
      "learning_rate": 1.0352670303452535e-06,
      "loss": 0.4681,
      "step": 344030
    },
    {
      "epoch": 140.83,
      "grad_norm": 2.22442626953125,
      "learning_rate": 1.035135426571614e-06,
      "loss": 0.454,
      "step": 344040
    },
    {
      "epoch": 140.83,
      "grad_norm": 1.9450610876083374,
      "learning_rate": 1.0350038287925352e-06,
      "loss": 0.4753,
      "step": 344050
    },
    {
      "epoch": 140.84,
      "grad_norm": 1.8435617685317993,
      "learning_rate": 1.0348722370086198e-06,
      "loss": 0.4704,
      "step": 344060
    },
    {
      "epoch": 140.84,
      "grad_norm": 2.203321933746338,
      "learning_rate": 1.0347406512204708e-06,
      "loss": 0.4485,
      "step": 344070
    },
    {
      "epoch": 140.84,
      "grad_norm": 1.6922334432601929,
      "learning_rate": 1.034609071428691e-06,
      "loss": 0.45,
      "step": 344080
    },
    {
      "epoch": 140.85,
      "grad_norm": 1.8845391273498535,
      "learning_rate": 1.0344774976338828e-06,
      "loss": 0.4706,
      "step": 344090
    },
    {
      "epoch": 140.85,
      "grad_norm": 1.7173476219177246,
      "learning_rate": 1.0343459298366494e-06,
      "loss": 0.4372,
      "step": 344100
    },
    {
      "epoch": 140.86,
      "grad_norm": 1.9398834705352783,
      "learning_rate": 1.0342143680375932e-06,
      "loss": 0.4358,
      "step": 344110
    },
    {
      "epoch": 140.86,
      "grad_norm": 2.4669744968414307,
      "learning_rate": 1.034082812237317e-06,
      "loss": 0.4616,
      "step": 344120
    },
    {
      "epoch": 140.86,
      "grad_norm": 1.8517887592315674,
      "learning_rate": 1.0339512624364233e-06,
      "loss": 0.4315,
      "step": 344130
    },
    {
      "epoch": 140.87,
      "grad_norm": 2.1028361320495605,
      "learning_rate": 1.0338197186355145e-06,
      "loss": 0.4773,
      "step": 344140
    },
    {
      "epoch": 140.87,
      "grad_norm": 2.3116555213928223,
      "learning_rate": 1.0336881808351938e-06,
      "loss": 0.4294,
      "step": 344150
    },
    {
      "epoch": 140.88,
      "grad_norm": 1.9777711629867554,
      "learning_rate": 1.0335566490360634e-06,
      "loss": 0.4564,
      "step": 344160
    },
    {
      "epoch": 140.88,
      "grad_norm": 2.207092523574829,
      "learning_rate": 1.0334251232387255e-06,
      "loss": 0.4531,
      "step": 344170
    },
    {
      "epoch": 140.88,
      "grad_norm": 2.8811795711517334,
      "learning_rate": 1.0332936034437831e-06,
      "loss": 0.4638,
      "step": 344180
    },
    {
      "epoch": 140.89,
      "grad_norm": 1.910286784172058,
      "learning_rate": 1.0331620896518388e-06,
      "loss": 0.4621,
      "step": 344190
    },
    {
      "epoch": 140.89,
      "grad_norm": 2.1725711822509766,
      "learning_rate": 1.0330305818634942e-06,
      "loss": 0.4495,
      "step": 344200
    },
    {
      "epoch": 140.9,
      "grad_norm": 2.1282193660736084,
      "learning_rate": 1.0328990800793534e-06,
      "loss": 0.4546,
      "step": 344210
    },
    {
      "epoch": 140.9,
      "grad_norm": 1.9946174621582031,
      "learning_rate": 1.0327675843000161e-06,
      "loss": 0.4605,
      "step": 344220
    },
    {
      "epoch": 140.9,
      "grad_norm": 1.9608982801437378,
      "learning_rate": 1.0326360945260868e-06,
      "loss": 0.4346,
      "step": 344230
    },
    {
      "epoch": 140.91,
      "grad_norm": 1.9035485982894897,
      "learning_rate": 1.0325046107581674e-06,
      "loss": 0.477,
      "step": 344240
    },
    {
      "epoch": 140.91,
      "grad_norm": 2.1025807857513428,
      "learning_rate": 1.03237313299686e-06,
      "loss": 0.453,
      "step": 344250
    },
    {
      "epoch": 140.92,
      "grad_norm": 1.8707621097564697,
      "learning_rate": 1.0322416612427671e-06,
      "loss": 0.4526,
      "step": 344260
    },
    {
      "epoch": 140.92,
      "grad_norm": 1.6332648992538452,
      "learning_rate": 1.0321101954964905e-06,
      "loss": 0.4653,
      "step": 344270
    },
    {
      "epoch": 140.93,
      "grad_norm": 2.3802566528320312,
      "learning_rate": 1.031978735758633e-06,
      "loss": 0.456,
      "step": 344280
    },
    {
      "epoch": 140.93,
      "grad_norm": 2.531635046005249,
      "learning_rate": 1.0318472820297963e-06,
      "loss": 0.4617,
      "step": 344290
    },
    {
      "epoch": 140.93,
      "grad_norm": 2.0860724449157715,
      "learning_rate": 1.031715834310583e-06,
      "loss": 0.4457,
      "step": 344300
    },
    {
      "epoch": 140.94,
      "grad_norm": 2.015385150909424,
      "learning_rate": 1.031584392601595e-06,
      "loss": 0.4349,
      "step": 344310
    },
    {
      "epoch": 140.94,
      "grad_norm": 1.6643304824829102,
      "learning_rate": 1.031452956903434e-06,
      "loss": 0.4475,
      "step": 344320
    },
    {
      "epoch": 140.95,
      "grad_norm": 2.2668020725250244,
      "learning_rate": 1.031321527216703e-06,
      "loss": 0.4451,
      "step": 344330
    },
    {
      "epoch": 140.95,
      "grad_norm": 1.9127424955368042,
      "learning_rate": 1.0311901035420043e-06,
      "loss": 0.4477,
      "step": 344340
    },
    {
      "epoch": 140.95,
      "grad_norm": 2.547612190246582,
      "learning_rate": 1.0310586858799382e-06,
      "loss": 0.4374,
      "step": 344350
    },
    {
      "epoch": 140.96,
      "grad_norm": 1.730203628540039,
      "learning_rate": 1.0309272742311078e-06,
      "loss": 0.4557,
      "step": 344360
    },
    {
      "epoch": 140.96,
      "grad_norm": 2.179680824279785,
      "learning_rate": 1.0307958685961149e-06,
      "loss": 0.4451,
      "step": 344370
    },
    {
      "epoch": 140.97,
      "grad_norm": 1.8940342664718628,
      "learning_rate": 1.0306644689755617e-06,
      "loss": 0.4409,
      "step": 344380
    },
    {
      "epoch": 140.97,
      "grad_norm": 1.7114659547805786,
      "learning_rate": 1.03053307537005e-06,
      "loss": 0.435,
      "step": 344390
    },
    {
      "epoch": 140.97,
      "grad_norm": 2.865102529525757,
      "learning_rate": 1.0304016877801817e-06,
      "loss": 0.4426,
      "step": 344400
    },
    {
      "epoch": 140.98,
      "grad_norm": 1.760073184967041,
      "learning_rate": 1.0302703062065574e-06,
      "loss": 0.4454,
      "step": 344410
    },
    {
      "epoch": 140.98,
      "grad_norm": 2.29005765914917,
      "learning_rate": 1.0301389306497813e-06,
      "loss": 0.4337,
      "step": 344420
    },
    {
      "epoch": 140.99,
      "grad_norm": 1.9713019132614136,
      "learning_rate": 1.030007561110454e-06,
      "loss": 0.4596,
      "step": 344430
    },
    {
      "epoch": 140.99,
      "grad_norm": 2.320150136947632,
      "learning_rate": 1.0298761975891775e-06,
      "loss": 0.4516,
      "step": 344440
    },
    {
      "epoch": 140.99,
      "grad_norm": 2.0582261085510254,
      "learning_rate": 1.0297448400865533e-06,
      "loss": 0.4466,
      "step": 344450
    },
    {
      "epoch": 141.0,
      "grad_norm": 1.996932029724121,
      "learning_rate": 1.029613488603183e-06,
      "loss": 0.4601,
      "step": 344460
    },
    {
      "epoch": 141.0,
      "eval_loss": 0.45402786135673523,
      "eval_runtime": 57.8341,
      "eval_samples_per_second": 59.636,
      "eval_steps_per_second": 7.47,
      "step": 344463
    },
    {
      "epoch": 141.0,
      "grad_norm": 2.1639842987060547,
      "learning_rate": 1.029482143139669e-06,
      "loss": 0.4301,
      "step": 344470
    },
    {
      "epoch": 141.01,
      "grad_norm": 2.287893295288086,
      "learning_rate": 1.029350803696613e-06,
      "loss": 0.4514,
      "step": 344480
    },
    {
      "epoch": 141.01,
      "grad_norm": 2.304304599761963,
      "learning_rate": 1.0292194702746154e-06,
      "loss": 0.4602,
      "step": 344490
    },
    {
      "epoch": 141.02,
      "grad_norm": 4.215489387512207,
      "learning_rate": 1.0290881428742786e-06,
      "loss": 0.4573,
      "step": 344500
    },
    {
      "epoch": 141.02,
      "grad_norm": 1.9634872674942017,
      "learning_rate": 1.0289568214962044e-06,
      "loss": 0.4634,
      "step": 344510
    },
    {
      "epoch": 141.02,
      "grad_norm": 2.1588196754455566,
      "learning_rate": 1.028825506140994e-06,
      "loss": 0.4389,
      "step": 344520
    },
    {
      "epoch": 141.03,
      "grad_norm": 1.6591312885284424,
      "learning_rate": 1.0286941968092486e-06,
      "loss": 0.4255,
      "step": 344530
    },
    {
      "epoch": 141.03,
      "grad_norm": 2.294708013534546,
      "learning_rate": 1.0285628935015707e-06,
      "loss": 0.4478,
      "step": 344540
    },
    {
      "epoch": 141.04,
      "grad_norm": 1.760398268699646,
      "learning_rate": 1.0284315962185612e-06,
      "loss": 0.4362,
      "step": 344550
    },
    {
      "epoch": 141.04,
      "grad_norm": 1.9877742528915405,
      "learning_rate": 1.0283003049608212e-06,
      "loss": 0.4411,
      "step": 344560
    },
    {
      "epoch": 141.04,
      "grad_norm": 1.8138959407806396,
      "learning_rate": 1.028169019728953e-06,
      "loss": 0.4607,
      "step": 344570
    },
    {
      "epoch": 141.05,
      "grad_norm": 1.7581311464309692,
      "learning_rate": 1.0280377405235575e-06,
      "loss": 0.4571,
      "step": 344580
    },
    {
      "epoch": 141.05,
      "grad_norm": 1.7570006847381592,
      "learning_rate": 1.0279064673452358e-06,
      "loss": 0.4513,
      "step": 344590
    },
    {
      "epoch": 141.06,
      "grad_norm": 2.100673198699951,
      "learning_rate": 1.0277752001945888e-06,
      "loss": 0.4415,
      "step": 344600
    },
    {
      "epoch": 141.06,
      "grad_norm": 1.9467713832855225,
      "learning_rate": 1.0276439390722193e-06,
      "loss": 0.4282,
      "step": 344610
    },
    {
      "epoch": 141.06,
      "grad_norm": 1.7385786771774292,
      "learning_rate": 1.027512683978729e-06,
      "loss": 0.4329,
      "step": 344620
    },
    {
      "epoch": 141.07,
      "grad_norm": 1.7830830812454224,
      "learning_rate": 1.0273814349147171e-06,
      "loss": 0.4386,
      "step": 344630
    },
    {
      "epoch": 141.07,
      "grad_norm": 2.009932518005371,
      "learning_rate": 1.0272501918807857e-06,
      "loss": 0.4362,
      "step": 344640
    },
    {
      "epoch": 141.08,
      "grad_norm": 2.2264840602874756,
      "learning_rate": 1.0271189548775357e-06,
      "loss": 0.4361,
      "step": 344650
    },
    {
      "epoch": 141.08,
      "grad_norm": 1.9651061296463013,
      "learning_rate": 1.026987723905569e-06,
      "loss": 0.4621,
      "step": 344660
    },
    {
      "epoch": 141.08,
      "grad_norm": 1.7079098224639893,
      "learning_rate": 1.0268564989654863e-06,
      "loss": 0.4551,
      "step": 344670
    },
    {
      "epoch": 141.09,
      "grad_norm": 2.337630271911621,
      "learning_rate": 1.0267252800578884e-06,
      "loss": 0.442,
      "step": 344680
    },
    {
      "epoch": 141.09,
      "grad_norm": 2.0617516040802,
      "learning_rate": 1.0265940671833771e-06,
      "loss": 0.4596,
      "step": 344690
    },
    {
      "epoch": 141.1,
      "grad_norm": 1.751145362854004,
      "learning_rate": 1.0264628603425532e-06,
      "loss": 0.4541,
      "step": 344700
    },
    {
      "epoch": 141.1,
      "grad_norm": 2.2365591526031494,
      "learning_rate": 1.0263316595360177e-06,
      "loss": 0.4459,
      "step": 344710
    },
    {
      "epoch": 141.11,
      "grad_norm": 2.5812020301818848,
      "learning_rate": 1.0262004647643714e-06,
      "loss": 0.4512,
      "step": 344720
    },
    {
      "epoch": 141.11,
      "grad_norm": 2.630464792251587,
      "learning_rate": 1.0260692760282155e-06,
      "loss": 0.4442,
      "step": 344730
    },
    {
      "epoch": 141.11,
      "grad_norm": 2.0363833904266357,
      "learning_rate": 1.025938093328151e-06,
      "loss": 0.4547,
      "step": 344740
    },
    {
      "epoch": 141.12,
      "grad_norm": 1.9309386014938354,
      "learning_rate": 1.0258069166647789e-06,
      "loss": 0.4308,
      "step": 344750
    },
    {
      "epoch": 141.12,
      "grad_norm": 2.0045320987701416,
      "learning_rate": 1.0256757460386998e-06,
      "loss": 0.455,
      "step": 344760
    },
    {
      "epoch": 141.13,
      "grad_norm": 1.7856568098068237,
      "learning_rate": 1.0255445814505146e-06,
      "loss": 0.4663,
      "step": 344770
    },
    {
      "epoch": 141.13,
      "grad_norm": 1.9571080207824707,
      "learning_rate": 1.0254134229008246e-06,
      "loss": 0.4395,
      "step": 344780
    },
    {
      "epoch": 141.13,
      "grad_norm": 2.1365420818328857,
      "learning_rate": 1.02528227039023e-06,
      "loss": 0.4358,
      "step": 344790
    },
    {
      "epoch": 141.14,
      "grad_norm": 2.13486647605896,
      "learning_rate": 1.0251511239193319e-06,
      "loss": 0.4559,
      "step": 344800
    },
    {
      "epoch": 141.14,
      "grad_norm": 1.7445148229599,
      "learning_rate": 1.0250199834887311e-06,
      "loss": 0.4472,
      "step": 344810
    },
    {
      "epoch": 141.15,
      "grad_norm": 1.988139271736145,
      "learning_rate": 1.0248888490990283e-06,
      "loss": 0.4315,
      "step": 344820
    },
    {
      "epoch": 141.15,
      "grad_norm": 2.1531450748443604,
      "learning_rate": 1.0247577207508243e-06,
      "loss": 0.457,
      "step": 344830
    },
    {
      "epoch": 141.15,
      "grad_norm": 1.9922895431518555,
      "learning_rate": 1.0246265984447194e-06,
      "loss": 0.4405,
      "step": 344840
    },
    {
      "epoch": 141.16,
      "grad_norm": 2.125654697418213,
      "learning_rate": 1.0244954821813148e-06,
      "loss": 0.464,
      "step": 344850
    },
    {
      "epoch": 141.16,
      "grad_norm": 1.939876675605774,
      "learning_rate": 1.0243643719612106e-06,
      "loss": 0.4431,
      "step": 344860
    },
    {
      "epoch": 141.17,
      "grad_norm": 1.6687233448028564,
      "learning_rate": 1.0242332677850076e-06,
      "loss": 0.4365,
      "step": 344870
    },
    {
      "epoch": 141.17,
      "grad_norm": 2.5231523513793945,
      "learning_rate": 1.0241021696533066e-06,
      "loss": 0.4489,
      "step": 344880
    },
    {
      "epoch": 141.17,
      "grad_norm": 2.220242977142334,
      "learning_rate": 1.0239710775667079e-06,
      "loss": 0.4557,
      "step": 344890
    },
    {
      "epoch": 141.18,
      "grad_norm": 2.007193088531494,
      "learning_rate": 1.0238399915258118e-06,
      "loss": 0.4624,
      "step": 344900
    },
    {
      "epoch": 141.18,
      "grad_norm": 2.2204744815826416,
      "learning_rate": 1.0237089115312193e-06,
      "loss": 0.4361,
      "step": 344910
    },
    {
      "epoch": 141.19,
      "grad_norm": 2.4794459342956543,
      "learning_rate": 1.0235778375835305e-06,
      "loss": 0.4504,
      "step": 344920
    },
    {
      "epoch": 141.19,
      "grad_norm": 1.8588746786117554,
      "learning_rate": 1.0234467696833467e-06,
      "loss": 0.4593,
      "step": 344930
    },
    {
      "epoch": 141.2,
      "grad_norm": 1.7664074897766113,
      "learning_rate": 1.0233157078312668e-06,
      "loss": 0.4448,
      "step": 344940
    },
    {
      "epoch": 141.2,
      "grad_norm": 2.0470569133758545,
      "learning_rate": 1.023184652027892e-06,
      "loss": 0.4466,
      "step": 344950
    },
    {
      "epoch": 141.2,
      "grad_norm": 2.4620606899261475,
      "learning_rate": 1.0230536022738223e-06,
      "loss": 0.4593,
      "step": 344960
    },
    {
      "epoch": 141.21,
      "grad_norm": 1.9057388305664062,
      "learning_rate": 1.0229225585696583e-06,
      "loss": 0.4377,
      "step": 344970
    },
    {
      "epoch": 141.21,
      "grad_norm": 1.9717142581939697,
      "learning_rate": 1.0227915209159996e-06,
      "loss": 0.4468,
      "step": 344980
    },
    {
      "epoch": 141.22,
      "grad_norm": 1.9614179134368896,
      "learning_rate": 1.0226604893134478e-06,
      "loss": 0.4495,
      "step": 344990
    },
    {
      "epoch": 141.22,
      "grad_norm": 2.3777270317077637,
      "learning_rate": 1.0225294637626024e-06,
      "loss": 0.4605,
      "step": 345000
    },
    {
      "epoch": 141.22,
      "grad_norm": 1.7489006519317627,
      "learning_rate": 1.0223984442640638e-06,
      "loss": 0.4627,
      "step": 345010
    },
    {
      "epoch": 141.23,
      "grad_norm": 1.9628618955612183,
      "learning_rate": 1.022267430818432e-06,
      "loss": 0.4741,
      "step": 345020
    },
    {
      "epoch": 141.23,
      "grad_norm": 2.019036293029785,
      "learning_rate": 1.022136423426307e-06,
      "loss": 0.4645,
      "step": 345030
    },
    {
      "epoch": 141.24,
      "grad_norm": 2.1328930854797363,
      "learning_rate": 1.0220054220882893e-06,
      "loss": 0.4691,
      "step": 345040
    },
    {
      "epoch": 141.24,
      "grad_norm": 1.7583065032958984,
      "learning_rate": 1.021874426804979e-06,
      "loss": 0.456,
      "step": 345050
    },
    {
      "epoch": 141.24,
      "grad_norm": 1.8735878467559814,
      "learning_rate": 1.0217434375769764e-06,
      "loss": 0.4478,
      "step": 345060
    },
    {
      "epoch": 141.25,
      "grad_norm": 2.3452179431915283,
      "learning_rate": 1.0216124544048806e-06,
      "loss": 0.4593,
      "step": 345070
    },
    {
      "epoch": 141.25,
      "grad_norm": 2.1881022453308105,
      "learning_rate": 1.0214814772892922e-06,
      "loss": 0.4419,
      "step": 345080
    },
    {
      "epoch": 141.26,
      "grad_norm": 1.9944247007369995,
      "learning_rate": 1.0213505062308109e-06,
      "loss": 0.4438,
      "step": 345090
    },
    {
      "epoch": 141.26,
      "grad_norm": 2.130805015563965,
      "learning_rate": 1.0212195412300371e-06,
      "loss": 0.4728,
      "step": 345100
    },
    {
      "epoch": 141.26,
      "grad_norm": 1.9076260328292847,
      "learning_rate": 1.0210885822875704e-06,
      "loss": 0.4476,
      "step": 345110
    },
    {
      "epoch": 141.27,
      "grad_norm": 2.0361342430114746,
      "learning_rate": 1.0209576294040108e-06,
      "loss": 0.4694,
      "step": 345120
    },
    {
      "epoch": 141.27,
      "grad_norm": 2.147150993347168,
      "learning_rate": 1.020826682579958e-06,
      "loss": 0.4407,
      "step": 345130
    },
    {
      "epoch": 141.28,
      "grad_norm": 2.0292000770568848,
      "learning_rate": 1.0206957418160122e-06,
      "loss": 0.4491,
      "step": 345140
    },
    {
      "epoch": 141.28,
      "grad_norm": 1.7428557872772217,
      "learning_rate": 1.020564807112773e-06,
      "loss": 0.4324,
      "step": 345150
    },
    {
      "epoch": 141.29,
      "grad_norm": 2.104524612426758,
      "learning_rate": 1.0204338784708403e-06,
      "loss": 0.4207,
      "step": 345160
    },
    {
      "epoch": 141.29,
      "grad_norm": 2.069688320159912,
      "learning_rate": 1.0203029558908131e-06,
      "loss": 0.4686,
      "step": 345170
    },
    {
      "epoch": 141.29,
      "grad_norm": 2.0599257946014404,
      "learning_rate": 1.0201720393732926e-06,
      "loss": 0.4461,
      "step": 345180
    },
    {
      "epoch": 141.3,
      "grad_norm": 2.6082003116607666,
      "learning_rate": 1.0200411289188777e-06,
      "loss": 0.4554,
      "step": 345190
    },
    {
      "epoch": 141.3,
      "grad_norm": 2.435640573501587,
      "learning_rate": 1.0199102245281689e-06,
      "loss": 0.4189,
      "step": 345200
    },
    {
      "epoch": 141.31,
      "grad_norm": 2.0196621417999268,
      "learning_rate": 1.0197793262017647e-06,
      "loss": 0.4524,
      "step": 345210
    },
    {
      "epoch": 141.31,
      "grad_norm": 1.8811553716659546,
      "learning_rate": 1.0196484339402644e-06,
      "loss": 0.4569,
      "step": 345220
    },
    {
      "epoch": 141.31,
      "grad_norm": 1.7891709804534912,
      "learning_rate": 1.0195175477442687e-06,
      "loss": 0.4495,
      "step": 345230
    },
    {
      "epoch": 141.32,
      "grad_norm": 1.8993967771530151,
      "learning_rate": 1.0193866676143769e-06,
      "loss": 0.4569,
      "step": 345240
    },
    {
      "epoch": 141.32,
      "grad_norm": 2.065845251083374,
      "learning_rate": 1.0192557935511882e-06,
      "loss": 0.4547,
      "step": 345250
    },
    {
      "epoch": 141.33,
      "grad_norm": 1.7352991104125977,
      "learning_rate": 1.0191249255553025e-06,
      "loss": 0.4538,
      "step": 345260
    },
    {
      "epoch": 141.33,
      "grad_norm": 1.7040237188339233,
      "learning_rate": 1.018994063627319e-06,
      "loss": 0.4565,
      "step": 345270
    },
    {
      "epoch": 141.33,
      "grad_norm": 2.4778828620910645,
      "learning_rate": 1.0188632077678372e-06,
      "loss": 0.4556,
      "step": 345280
    },
    {
      "epoch": 141.34,
      "grad_norm": 1.7473266124725342,
      "learning_rate": 1.0187323579774568e-06,
      "loss": 0.4424,
      "step": 345290
    },
    {
      "epoch": 141.34,
      "grad_norm": 1.717322587966919,
      "learning_rate": 1.018601514256777e-06,
      "loss": 0.4672,
      "step": 345300
    },
    {
      "epoch": 141.35,
      "grad_norm": 1.7866922616958618,
      "learning_rate": 1.0184706766063972e-06,
      "loss": 0.4505,
      "step": 345310
    },
    {
      "epoch": 141.35,
      "grad_norm": 1.8684035539627075,
      "learning_rate": 1.0183398450269167e-06,
      "loss": 0.4408,
      "step": 345320
    },
    {
      "epoch": 141.35,
      "grad_norm": 1.9086153507232666,
      "learning_rate": 1.0182090195189346e-06,
      "loss": 0.4388,
      "step": 345330
    },
    {
      "epoch": 141.36,
      "grad_norm": 2.1938438415527344,
      "learning_rate": 1.018078200083051e-06,
      "loss": 0.4481,
      "step": 345340
    },
    {
      "epoch": 141.36,
      "grad_norm": 2.0704169273376465,
      "learning_rate": 1.0179473867198638e-06,
      "loss": 0.462,
      "step": 345350
    },
    {
      "epoch": 141.37,
      "grad_norm": 2.998502016067505,
      "learning_rate": 1.0178165794299737e-06,
      "loss": 0.4567,
      "step": 345360
    },
    {
      "epoch": 141.37,
      "grad_norm": 2.0901737213134766,
      "learning_rate": 1.017685778213979e-06,
      "loss": 0.4419,
      "step": 345370
    },
    {
      "epoch": 141.38,
      "grad_norm": 1.874985933303833,
      "learning_rate": 1.0175549830724794e-06,
      "loss": 0.4589,
      "step": 345380
    },
    {
      "epoch": 141.38,
      "grad_norm": 2.3588552474975586,
      "learning_rate": 1.0174241940060733e-06,
      "loss": 0.4531,
      "step": 345390
    },
    {
      "epoch": 141.38,
      "grad_norm": 2.1310677528381348,
      "learning_rate": 1.0172934110153606e-06,
      "loss": 0.4507,
      "step": 345400
    },
    {
      "epoch": 141.39,
      "grad_norm": 1.6676899194717407,
      "learning_rate": 1.0171626341009404e-06,
      "loss": 0.4534,
      "step": 345410
    },
    {
      "epoch": 141.39,
      "grad_norm": 2.0310590267181396,
      "learning_rate": 1.0170318632634112e-06,
      "loss": 0.4333,
      "step": 345420
    },
    {
      "epoch": 141.4,
      "grad_norm": 2.119772434234619,
      "learning_rate": 1.0169010985033723e-06,
      "loss": 0.4691,
      "step": 345430
    },
    {
      "epoch": 141.4,
      "grad_norm": 2.154430389404297,
      "learning_rate": 1.0167703398214227e-06,
      "loss": 0.4764,
      "step": 345440
    },
    {
      "epoch": 141.4,
      "grad_norm": 1.9651650190353394,
      "learning_rate": 1.0166395872181612e-06,
      "loss": 0.4529,
      "step": 345450
    },
    {
      "epoch": 141.41,
      "grad_norm": 1.8617455959320068,
      "learning_rate": 1.0165088406941873e-06,
      "loss": 0.4492,
      "step": 345460
    },
    {
      "epoch": 141.41,
      "grad_norm": 1.7431219816207886,
      "learning_rate": 1.0163781002500995e-06,
      "loss": 0.4588,
      "step": 345470
    },
    {
      "epoch": 141.42,
      "grad_norm": 1.794708251953125,
      "learning_rate": 1.016247365886497e-06,
      "loss": 0.4571,
      "step": 345480
    },
    {
      "epoch": 141.42,
      "grad_norm": 1.7896934747695923,
      "learning_rate": 1.0161166376039781e-06,
      "loss": 0.4631,
      "step": 345490
    },
    {
      "epoch": 141.42,
      "grad_norm": 2.27411150932312,
      "learning_rate": 1.0159859154031423e-06,
      "loss": 0.4434,
      "step": 345500
    },
    {
      "epoch": 141.43,
      "grad_norm": 2.271925687789917,
      "learning_rate": 1.015855199284589e-06,
      "loss": 0.452,
      "step": 345510
    },
    {
      "epoch": 141.43,
      "grad_norm": 2.023440361022949,
      "learning_rate": 1.0157244892489153e-06,
      "loss": 0.4494,
      "step": 345520
    },
    {
      "epoch": 141.44,
      "grad_norm": 2.2235589027404785,
      "learning_rate": 1.0155937852967206e-06,
      "loss": 0.4603,
      "step": 345530
    },
    {
      "epoch": 141.44,
      "grad_norm": 1.851305365562439,
      "learning_rate": 1.015463087428603e-06,
      "loss": 0.4536,
      "step": 345540
    },
    {
      "epoch": 141.44,
      "grad_norm": 1.9664201736450195,
      "learning_rate": 1.015332395645163e-06,
      "loss": 0.4596,
      "step": 345550
    },
    {
      "epoch": 141.45,
      "grad_norm": 3.017329692840576,
      "learning_rate": 1.0152017099469983e-06,
      "loss": 0.457,
      "step": 345560
    },
    {
      "epoch": 141.45,
      "grad_norm": 1.7853164672851562,
      "learning_rate": 1.0150710303347076e-06,
      "loss": 0.4543,
      "step": 345570
    },
    {
      "epoch": 141.46,
      "grad_norm": 1.768250584602356,
      "learning_rate": 1.0149403568088892e-06,
      "loss": 0.4526,
      "step": 345580
    },
    {
      "epoch": 141.46,
      "grad_norm": 2.040015935897827,
      "learning_rate": 1.0148096893701423e-06,
      "loss": 0.459,
      "step": 345590
    },
    {
      "epoch": 141.47,
      "grad_norm": 2.1652002334594727,
      "learning_rate": 1.014679028019065e-06,
      "loss": 0.4661,
      "step": 345600
    },
    {
      "epoch": 141.47,
      "grad_norm": 2.0874156951904297,
      "learning_rate": 1.0145483727562561e-06,
      "loss": 0.4272,
      "step": 345610
    },
    {
      "epoch": 141.47,
      "grad_norm": 2.1135244369506836,
      "learning_rate": 1.014417723582314e-06,
      "loss": 0.466,
      "step": 345620
    },
    {
      "epoch": 141.48,
      "grad_norm": 2.1714463233947754,
      "learning_rate": 1.0142870804978372e-06,
      "loss": 0.4376,
      "step": 345630
    },
    {
      "epoch": 141.48,
      "grad_norm": 2.4054791927337646,
      "learning_rate": 1.014156443503424e-06,
      "loss": 0.4392,
      "step": 345640
    },
    {
      "epoch": 141.49,
      "grad_norm": 2.2458815574645996,
      "learning_rate": 1.014025812599674e-06,
      "loss": 0.4598,
      "step": 345650
    },
    {
      "epoch": 141.49,
      "grad_norm": 1.631602168083191,
      "learning_rate": 1.0138951877871835e-06,
      "loss": 0.4532,
      "step": 345660
    },
    {
      "epoch": 141.49,
      "grad_norm": 1.989241600036621,
      "learning_rate": 1.013764569066552e-06,
      "loss": 0.4467,
      "step": 345670
    },
    {
      "epoch": 141.5,
      "grad_norm": 1.7706962823867798,
      "learning_rate": 1.0136339564383777e-06,
      "loss": 0.4574,
      "step": 345680
    },
    {
      "epoch": 141.5,
      "grad_norm": 1.7445261478424072,
      "learning_rate": 1.0135033499032593e-06,
      "loss": 0.4492,
      "step": 345690
    },
    {
      "epoch": 141.51,
      "grad_norm": 2.2043864727020264,
      "learning_rate": 1.0133727494617946e-06,
      "loss": 0.4271,
      "step": 345700
    },
    {
      "epoch": 141.51,
      "grad_norm": 1.809051275253296,
      "learning_rate": 1.013242155114582e-06,
      "loss": 0.4516,
      "step": 345710
    },
    {
      "epoch": 141.51,
      "grad_norm": 2.13923716545105,
      "learning_rate": 1.01311156686222e-06,
      "loss": 0.4601,
      "step": 345720
    },
    {
      "epoch": 141.52,
      "grad_norm": 2.502784013748169,
      "learning_rate": 1.0129809847053054e-06,
      "loss": 0.4462,
      "step": 345730
    },
    {
      "epoch": 141.52,
      "grad_norm": 2.2065157890319824,
      "learning_rate": 1.0128504086444387e-06,
      "loss": 0.4466,
      "step": 345740
    },
    {
      "epoch": 141.53,
      "grad_norm": 1.993985891342163,
      "learning_rate": 1.0127198386802166e-06,
      "loss": 0.4671,
      "step": 345750
    },
    {
      "epoch": 141.53,
      "grad_norm": 1.9774638414382935,
      "learning_rate": 1.0125892748132381e-06,
      "loss": 0.4268,
      "step": 345760
    },
    {
      "epoch": 141.53,
      "grad_norm": 1.8984509706497192,
      "learning_rate": 1.0124587170441002e-06,
      "loss": 0.4549,
      "step": 345770
    },
    {
      "epoch": 141.54,
      "grad_norm": 2.178332567214966,
      "learning_rate": 1.0123281653734024e-06,
      "loss": 0.4609,
      "step": 345780
    },
    {
      "epoch": 141.54,
      "grad_norm": 1.604487419128418,
      "learning_rate": 1.0121976198017412e-06,
      "loss": 0.4429,
      "step": 345790
    },
    {
      "epoch": 141.55,
      "grad_norm": 1.8359969854354858,
      "learning_rate": 1.0120670803297152e-06,
      "loss": 0.4666,
      "step": 345800
    },
    {
      "epoch": 141.55,
      "grad_norm": 1.9197955131530762,
      "learning_rate": 1.0119365469579225e-06,
      "loss": 0.4684,
      "step": 345810
    },
    {
      "epoch": 141.56,
      "grad_norm": 2.176386833190918,
      "learning_rate": 1.0118060196869606e-06,
      "loss": 0.4601,
      "step": 345820
    },
    {
      "epoch": 141.56,
      "grad_norm": 2.1303904056549072,
      "learning_rate": 1.0116754985174281e-06,
      "loss": 0.4548,
      "step": 345830
    },
    {
      "epoch": 141.56,
      "grad_norm": 2.0496859550476074,
      "learning_rate": 1.0115449834499225e-06,
      "loss": 0.4382,
      "step": 345840
    },
    {
      "epoch": 141.57,
      "grad_norm": 1.8215742111206055,
      "learning_rate": 1.011414474485042e-06,
      "loss": 0.4655,
      "step": 345850
    },
    {
      "epoch": 141.57,
      "grad_norm": 2.1729397773742676,
      "learning_rate": 1.0112839716233838e-06,
      "loss": 0.4523,
      "step": 345860
    },
    {
      "epoch": 141.58,
      "grad_norm": 1.8933489322662354,
      "learning_rate": 1.0111534748655463e-06,
      "loss": 0.4563,
      "step": 345870
    },
    {
      "epoch": 141.58,
      "grad_norm": 2.060328245162964,
      "learning_rate": 1.0110229842121272e-06,
      "loss": 0.447,
      "step": 345880
    },
    {
      "epoch": 141.58,
      "grad_norm": 2.5382368564605713,
      "learning_rate": 1.010892499663724e-06,
      "loss": 0.4466,
      "step": 345890
    },
    {
      "epoch": 141.59,
      "grad_norm": 1.8718534708023071,
      "learning_rate": 1.0107620212209347e-06,
      "loss": 0.4531,
      "step": 345900
    },
    {
      "epoch": 141.59,
      "grad_norm": 1.8449159860610962,
      "learning_rate": 1.0106315488843566e-06,
      "loss": 0.4603,
      "step": 345910
    },
    {
      "epoch": 141.6,
      "grad_norm": 2.3119089603424072,
      "learning_rate": 1.0105010826545881e-06,
      "loss": 0.4383,
      "step": 345920
    },
    {
      "epoch": 141.6,
      "grad_norm": 2.064553737640381,
      "learning_rate": 1.0103706225322262e-06,
      "loss": 0.4768,
      "step": 345930
    },
    {
      "epoch": 141.6,
      "grad_norm": 1.9655117988586426,
      "learning_rate": 1.0102401685178686e-06,
      "loss": 0.4337,
      "step": 345940
    },
    {
      "epoch": 141.61,
      "grad_norm": 1.8706600666046143,
      "learning_rate": 1.0101097206121132e-06,
      "loss": 0.4553,
      "step": 345950
    },
    {
      "epoch": 141.61,
      "grad_norm": 2.2167670726776123,
      "learning_rate": 1.0099792788155573e-06,
      "loss": 0.4445,
      "step": 345960
    },
    {
      "epoch": 141.62,
      "grad_norm": 2.0790417194366455,
      "learning_rate": 1.0098488431287987e-06,
      "loss": 0.4616,
      "step": 345970
    },
    {
      "epoch": 141.62,
      "grad_norm": 2.8256497383117676,
      "learning_rate": 1.0097184135524347e-06,
      "loss": 0.4565,
      "step": 345980
    },
    {
      "epoch": 141.63,
      "grad_norm": 2.504920721054077,
      "learning_rate": 1.0095879900870626e-06,
      "loss": 0.47,
      "step": 345990
    },
    {
      "epoch": 141.63,
      "grad_norm": 1.8500241041183472,
      "learning_rate": 1.0094575727332799e-06,
      "loss": 0.4412,
      "step": 346000
    },
    {
      "epoch": 141.63,
      "grad_norm": 2.0453178882598877,
      "learning_rate": 1.0093271614916844e-06,
      "loss": 0.4359,
      "step": 346010
    },
    {
      "epoch": 141.64,
      "grad_norm": 2.104151964187622,
      "learning_rate": 1.0091967563628733e-06,
      "loss": 0.4663,
      "step": 346020
    },
    {
      "epoch": 141.64,
      "grad_norm": 2.3314409255981445,
      "learning_rate": 1.0090663573474439e-06,
      "loss": 0.4737,
      "step": 346030
    },
    {
      "epoch": 141.65,
      "grad_norm": 1.677374005317688,
      "learning_rate": 1.0089359644459936e-06,
      "loss": 0.4591,
      "step": 346040
    },
    {
      "epoch": 141.65,
      "grad_norm": 2.1148509979248047,
      "learning_rate": 1.0088055776591195e-06,
      "loss": 0.453,
      "step": 346050
    },
    {
      "epoch": 141.65,
      "grad_norm": 2.054265022277832,
      "learning_rate": 1.0086751969874191e-06,
      "loss": 0.4457,
      "step": 346060
    },
    {
      "epoch": 141.66,
      "grad_norm": 1.9151917695999146,
      "learning_rate": 1.00854482243149e-06,
      "loss": 0.4642,
      "step": 346070
    },
    {
      "epoch": 141.66,
      "grad_norm": 1.8815388679504395,
      "learning_rate": 1.0084144539919285e-06,
      "loss": 0.4746,
      "step": 346080
    },
    {
      "epoch": 141.67,
      "grad_norm": 2.068800210952759,
      "learning_rate": 1.0082840916693334e-06,
      "loss": 0.4545,
      "step": 346090
    },
    {
      "epoch": 141.67,
      "grad_norm": 1.8386638164520264,
      "learning_rate": 1.0081537354643e-06,
      "loss": 0.4283,
      "step": 346100
    },
    {
      "epoch": 141.67,
      "grad_norm": 2.1522610187530518,
      "learning_rate": 1.0080233853774253e-06,
      "loss": 0.4448,
      "step": 346110
    },
    {
      "epoch": 141.68,
      "grad_norm": 1.8911211490631104,
      "learning_rate": 1.0078930414093083e-06,
      "loss": 0.4395,
      "step": 346120
    },
    {
      "epoch": 141.68,
      "grad_norm": 2.1822969913482666,
      "learning_rate": 1.007762703560545e-06,
      "loss": 0.4416,
      "step": 346130
    },
    {
      "epoch": 141.69,
      "grad_norm": 1.5403481721878052,
      "learning_rate": 1.0076323718317329e-06,
      "loss": 0.4205,
      "step": 346140
    },
    {
      "epoch": 141.69,
      "grad_norm": 2.1317801475524902,
      "learning_rate": 1.0075020462234686e-06,
      "loss": 0.4707,
      "step": 346150
    },
    {
      "epoch": 141.69,
      "grad_norm": 1.837567925453186,
      "learning_rate": 1.007371726736349e-06,
      "loss": 0.455,
      "step": 346160
    },
    {
      "epoch": 141.7,
      "grad_norm": 1.6878585815429688,
      "learning_rate": 1.0072414133709717e-06,
      "loss": 0.4532,
      "step": 346170
    },
    {
      "epoch": 141.7,
      "grad_norm": 2.117354393005371,
      "learning_rate": 1.0071111061279328e-06,
      "loss": 0.441,
      "step": 346180
    },
    {
      "epoch": 141.71,
      "grad_norm": 1.8485584259033203,
      "learning_rate": 1.00698080500783e-06,
      "loss": 0.4776,
      "step": 346190
    },
    {
      "epoch": 141.71,
      "grad_norm": 1.6767308712005615,
      "learning_rate": 1.0068505100112597e-06,
      "loss": 0.447,
      "step": 346200
    },
    {
      "epoch": 141.72,
      "grad_norm": 1.987928867340088,
      "learning_rate": 1.006720221138819e-06,
      "loss": 0.4416,
      "step": 346210
    },
    {
      "epoch": 141.72,
      "grad_norm": 1.668091893196106,
      "learning_rate": 1.0065899383911047e-06,
      "loss": 0.452,
      "step": 346220
    },
    {
      "epoch": 141.72,
      "grad_norm": 2.120204448699951,
      "learning_rate": 1.0064596617687143e-06,
      "loss": 0.4485,
      "step": 346230
    },
    {
      "epoch": 141.73,
      "grad_norm": 2.4189000129699707,
      "learning_rate": 1.006329391272243e-06,
      "loss": 0.4443,
      "step": 346240
    },
    {
      "epoch": 141.73,
      "grad_norm": 1.673171043395996,
      "learning_rate": 1.0061991269022883e-06,
      "loss": 0.4392,
      "step": 346250
    },
    {
      "epoch": 141.74,
      "grad_norm": 4.516591548919678,
      "learning_rate": 1.0060688686594473e-06,
      "loss": 0.4536,
      "step": 346260
    },
    {
      "epoch": 141.74,
      "grad_norm": 1.7932361364364624,
      "learning_rate": 1.0059386165443162e-06,
      "loss": 0.4362,
      "step": 346270
    },
    {
      "epoch": 141.74,
      "grad_norm": 1.7953875064849854,
      "learning_rate": 1.0058083705574917e-06,
      "loss": 0.4375,
      "step": 346280
    },
    {
      "epoch": 141.75,
      "grad_norm": 2.0854227542877197,
      "learning_rate": 1.0056781306995707e-06,
      "loss": 0.4246,
      "step": 346290
    },
    {
      "epoch": 141.75,
      "grad_norm": 1.982331395149231,
      "learning_rate": 1.005547896971149e-06,
      "loss": 0.4336,
      "step": 346300
    },
    {
      "epoch": 141.76,
      "grad_norm": 1.8190679550170898,
      "learning_rate": 1.0054176693728245e-06,
      "loss": 0.458,
      "step": 346310
    },
    {
      "epoch": 141.76,
      "grad_norm": 1.929532766342163,
      "learning_rate": 1.0052874479051934e-06,
      "loss": 0.4495,
      "step": 346320
    },
    {
      "epoch": 141.76,
      "grad_norm": 2.180126667022705,
      "learning_rate": 1.0051572325688519e-06,
      "loss": 0.4205,
      "step": 346330
    },
    {
      "epoch": 141.77,
      "grad_norm": 2.194437265396118,
      "learning_rate": 1.0050270233643965e-06,
      "loss": 0.4704,
      "step": 346340
    },
    {
      "epoch": 141.77,
      "grad_norm": 2.237349271774292,
      "learning_rate": 1.0048968202924237e-06,
      "loss": 0.4556,
      "step": 346350
    },
    {
      "epoch": 141.78,
      "grad_norm": 2.002850294113159,
      "learning_rate": 1.0047666233535308e-06,
      "loss": 0.43,
      "step": 346360
    },
    {
      "epoch": 141.78,
      "grad_norm": 2.1036863327026367,
      "learning_rate": 1.0046364325483127e-06,
      "loss": 0.4408,
      "step": 346370
    },
    {
      "epoch": 141.78,
      "grad_norm": 2.2853288650512695,
      "learning_rate": 1.0045062478773664e-06,
      "loss": 0.4568,
      "step": 346380
    },
    {
      "epoch": 141.79,
      "grad_norm": 1.757301926612854,
      "learning_rate": 1.0043760693412884e-06,
      "loss": 0.453,
      "step": 346390
    },
    {
      "epoch": 141.79,
      "grad_norm": 2.1343953609466553,
      "learning_rate": 1.0042458969406747e-06,
      "loss": 0.4457,
      "step": 346400
    },
    {
      "epoch": 141.8,
      "grad_norm": 1.9397242069244385,
      "learning_rate": 1.0041157306761218e-06,
      "loss": 0.4566,
      "step": 346410
    },
    {
      "epoch": 141.8,
      "grad_norm": 1.828514814376831,
      "learning_rate": 1.0039855705482262e-06,
      "loss": 0.4641,
      "step": 346420
    },
    {
      "epoch": 141.81,
      "grad_norm": 2.1466314792633057,
      "learning_rate": 1.003855416557584e-06,
      "loss": 0.4383,
      "step": 346430
    },
    {
      "epoch": 141.81,
      "grad_norm": 1.9524692296981812,
      "learning_rate": 1.0037252687047913e-06,
      "loss": 0.4487,
      "step": 346440
    },
    {
      "epoch": 141.81,
      "grad_norm": 2.084493637084961,
      "learning_rate": 1.0035951269904446e-06,
      "loss": 0.4467,
      "step": 346450
    },
    {
      "epoch": 141.82,
      "grad_norm": 1.862625241279602,
      "learning_rate": 1.0034649914151395e-06,
      "loss": 0.4609,
      "step": 346460
    },
    {
      "epoch": 141.82,
      "grad_norm": 2.158643960952759,
      "learning_rate": 1.0033348619794726e-06,
      "loss": 0.4482,
      "step": 346470
    },
    {
      "epoch": 141.83,
      "grad_norm": 1.7604981660842896,
      "learning_rate": 1.0032047386840397e-06,
      "loss": 0.4418,
      "step": 346480
    },
    {
      "epoch": 141.83,
      "grad_norm": 1.7434011697769165,
      "learning_rate": 1.0030746215294366e-06,
      "loss": 0.4537,
      "step": 346490
    },
    {
      "epoch": 141.83,
      "grad_norm": 2.17233943939209,
      "learning_rate": 1.0029445105162611e-06,
      "loss": 0.4565,
      "step": 346500
    },
    {
      "epoch": 141.84,
      "grad_norm": 2.0474905967712402,
      "learning_rate": 1.002814405645107e-06,
      "loss": 0.4564,
      "step": 346510
    },
    {
      "epoch": 141.84,
      "grad_norm": 1.8773595094680786,
      "learning_rate": 1.0026843069165714e-06,
      "loss": 0.4643,
      "step": 346520
    },
    {
      "epoch": 141.85,
      "grad_norm": 2.1788198947906494,
      "learning_rate": 1.0025542143312499e-06,
      "loss": 0.4564,
      "step": 346530
    },
    {
      "epoch": 141.85,
      "grad_norm": 2.1049299240112305,
      "learning_rate": 1.0024241278897387e-06,
      "loss": 0.4275,
      "step": 346540
    },
    {
      "epoch": 141.85,
      "grad_norm": 1.6815590858459473,
      "learning_rate": 1.0022940475926338e-06,
      "loss": 0.4577,
      "step": 346550
    },
    {
      "epoch": 141.86,
      "grad_norm": 2.616929769515991,
      "learning_rate": 1.0021639734405306e-06,
      "loss": 0.4644,
      "step": 346560
    },
    {
      "epoch": 141.86,
      "grad_norm": 2.1579370498657227,
      "learning_rate": 1.0020339054340251e-06,
      "loss": 0.4427,
      "step": 346570
    },
    {
      "epoch": 141.87,
      "grad_norm": 2.1126153469085693,
      "learning_rate": 1.0019038435737136e-06,
      "loss": 0.4452,
      "step": 346580
    },
    {
      "epoch": 141.87,
      "grad_norm": 2.0864434242248535,
      "learning_rate": 1.0017737878601912e-06,
      "loss": 0.4346,
      "step": 346590
    },
    {
      "epoch": 141.87,
      "grad_norm": 1.662367582321167,
      "learning_rate": 1.001643738294054e-06,
      "loss": 0.4551,
      "step": 346600
    },
    {
      "epoch": 141.88,
      "grad_norm": 1.9209682941436768,
      "learning_rate": 1.001513694875898e-06,
      "loss": 0.4632,
      "step": 346610
    },
    {
      "epoch": 141.88,
      "grad_norm": 2.0704500675201416,
      "learning_rate": 1.0013836576063182e-06,
      "loss": 0.4416,
      "step": 346620
    },
    {
      "epoch": 141.89,
      "grad_norm": 1.9580650329589844,
      "learning_rate": 1.001253626485911e-06,
      "loss": 0.4539,
      "step": 346630
    },
    {
      "epoch": 141.89,
      "grad_norm": 2.274914026260376,
      "learning_rate": 1.001123601515272e-06,
      "loss": 0.4433,
      "step": 346640
    },
    {
      "epoch": 141.9,
      "grad_norm": 2.1414241790771484,
      "learning_rate": 1.0009935826949963e-06,
      "loss": 0.442,
      "step": 346650
    },
    {
      "epoch": 141.9,
      "grad_norm": 2.0516743659973145,
      "learning_rate": 1.00086357002568e-06,
      "loss": 0.4773,
      "step": 346660
    },
    {
      "epoch": 141.9,
      "grad_norm": 2.100872755050659,
      "learning_rate": 1.0007335635079192e-06,
      "loss": 0.4389,
      "step": 346670
    },
    {
      "epoch": 141.91,
      "grad_norm": 1.9673421382904053,
      "learning_rate": 1.000603563142307e-06,
      "loss": 0.469,
      "step": 346680
    },
    {
      "epoch": 141.91,
      "grad_norm": 2.2224228382110596,
      "learning_rate": 1.0004735689294412e-06,
      "loss": 0.4351,
      "step": 346690
    },
    {
      "epoch": 141.92,
      "grad_norm": 2.268991470336914,
      "learning_rate": 1.0003435808699171e-06,
      "loss": 0.4557,
      "step": 346700
    },
    {
      "epoch": 141.92,
      "grad_norm": 2.2479169368743896,
      "learning_rate": 1.0002135989643296e-06,
      "loss": 0.4657,
      "step": 346710
    },
    {
      "epoch": 141.92,
      "grad_norm": 2.601006507873535,
      "learning_rate": 1.0000836232132744e-06,
      "loss": 0.4453,
      "step": 346720
    },
    {
      "epoch": 141.93,
      "grad_norm": 2.62028169631958,
      "learning_rate": 9.999536536173465e-07,
      "loss": 0.4616,
      "step": 346730
    },
    {
      "epoch": 141.93,
      "grad_norm": 2.405163526535034,
      "learning_rate": 9.998236901771418e-07,
      "loss": 0.4658,
      "step": 346740
    },
    {
      "epoch": 141.94,
      "grad_norm": 2.5015454292297363,
      "learning_rate": 9.996937328932554e-07,
      "loss": 0.4478,
      "step": 346750
    },
    {
      "epoch": 141.94,
      "grad_norm": 1.767068862915039,
      "learning_rate": 9.995637817662823e-07,
      "loss": 0.4326,
      "step": 346760
    },
    {
      "epoch": 141.94,
      "grad_norm": 2.371732473373413,
      "learning_rate": 9.994338367968183e-07,
      "loss": 0.4386,
      "step": 346770
    },
    {
      "epoch": 141.95,
      "grad_norm": 1.8240044116973877,
      "learning_rate": 9.993038979854584e-07,
      "loss": 0.4531,
      "step": 346780
    },
    {
      "epoch": 141.95,
      "grad_norm": 2.3395891189575195,
      "learning_rate": 9.991739653327982e-07,
      "loss": 0.4444,
      "step": 346790
    },
    {
      "epoch": 141.96,
      "grad_norm": 1.8683661222457886,
      "learning_rate": 9.990440388394323e-07,
      "loss": 0.4475,
      "step": 346800
    },
    {
      "epoch": 141.96,
      "grad_norm": 2.0481178760528564,
      "learning_rate": 9.98914118505957e-07,
      "loss": 0.4571,
      "step": 346810
    },
    {
      "epoch": 141.96,
      "grad_norm": 2.437979221343994,
      "learning_rate": 9.987842043329656e-07,
      "loss": 0.4502,
      "step": 346820
    },
    {
      "epoch": 141.97,
      "grad_norm": 1.967130422592163,
      "learning_rate": 9.986542963210544e-07,
      "loss": 0.4489,
      "step": 346830
    },
    {
      "epoch": 141.97,
      "grad_norm": 1.7686636447906494,
      "learning_rate": 9.985243944708182e-07,
      "loss": 0.4582,
      "step": 346840
    },
    {
      "epoch": 141.98,
      "grad_norm": 1.8840278387069702,
      "learning_rate": 9.983944987828526e-07,
      "loss": 0.4434,
      "step": 346850
    },
    {
      "epoch": 141.98,
      "grad_norm": 2.573045015335083,
      "learning_rate": 9.982646092577507e-07,
      "loss": 0.4569,
      "step": 346860
    },
    {
      "epoch": 141.99,
      "grad_norm": 1.9995249509811401,
      "learning_rate": 9.981347258961103e-07,
      "loss": 0.4523,
      "step": 346870
    },
    {
      "epoch": 141.99,
      "grad_norm": 2.00464129447937,
      "learning_rate": 9.98004848698525e-07,
      "loss": 0.4464,
      "step": 346880
    },
    {
      "epoch": 141.99,
      "grad_norm": 2.2984561920166016,
      "learning_rate": 9.978749776655896e-07,
      "loss": 0.4519,
      "step": 346890
    },
    {
      "epoch": 142.0,
      "grad_norm": 2.086238145828247,
      "learning_rate": 9.977451127978995e-07,
      "loss": 0.4415,
      "step": 346900
    },
    {
      "epoch": 142.0,
      "eval_loss": 0.45275160670280457,
      "eval_runtime": 52.3697,
      "eval_samples_per_second": 65.859,
      "eval_steps_per_second": 8.249,
      "step": 346906
    },
    {
      "epoch": 142.0,
      "grad_norm": 2.061068296432495,
      "learning_rate": 9.976152540960491e-07,
      "loss": 0.4624,
      "step": 346910
    },
    {
      "epoch": 142.01,
      "grad_norm": 2.031708002090454,
      "learning_rate": 9.974854015606336e-07,
      "loss": 0.4755,
      "step": 346920
    },
    {
      "epoch": 142.01,
      "grad_norm": 2.3683085441589355,
      "learning_rate": 9.973555551922479e-07,
      "loss": 0.4428,
      "step": 346930
    },
    {
      "epoch": 142.01,
      "grad_norm": 1.6579651832580566,
      "learning_rate": 9.972257149914864e-07,
      "loss": 0.4388,
      "step": 346940
    },
    {
      "epoch": 142.02,
      "grad_norm": 1.9232505559921265,
      "learning_rate": 9.97095880958945e-07,
      "loss": 0.4687,
      "step": 346950
    },
    {
      "epoch": 142.02,
      "grad_norm": 2.282125234603882,
      "learning_rate": 9.96966053095217e-07,
      "loss": 0.4513,
      "step": 346960
    },
    {
      "epoch": 142.03,
      "grad_norm": 2.378599166870117,
      "learning_rate": 9.968362314008973e-07,
      "loss": 0.4476,
      "step": 346970
    },
    {
      "epoch": 142.03,
      "grad_norm": 1.8534096479415894,
      "learning_rate": 9.967064158765813e-07,
      "loss": 0.4507,
      "step": 346980
    },
    {
      "epoch": 142.03,
      "grad_norm": 2.252687692642212,
      "learning_rate": 9.96576606522863e-07,
      "loss": 0.461,
      "step": 346990
    },
    {
      "epoch": 142.04,
      "grad_norm": 1.9877703189849854,
      "learning_rate": 9.964468033403376e-07,
      "loss": 0.4402,
      "step": 347000
    },
    {
      "epoch": 142.04,
      "grad_norm": 2.2479708194732666,
      "learning_rate": 9.963170063295995e-07,
      "loss": 0.4597,
      "step": 347010
    },
    {
      "epoch": 142.05,
      "grad_norm": 1.5943483114242554,
      "learning_rate": 9.96187215491243e-07,
      "loss": 0.4365,
      "step": 347020
    },
    {
      "epoch": 142.05,
      "grad_norm": 1.9068833589553833,
      "learning_rate": 9.960574308258633e-07,
      "loss": 0.4471,
      "step": 347030
    },
    {
      "epoch": 142.05,
      "grad_norm": 2.3105082511901855,
      "learning_rate": 9.959276523340543e-07,
      "loss": 0.4409,
      "step": 347040
    },
    {
      "epoch": 142.06,
      "grad_norm": 1.991595983505249,
      "learning_rate": 9.9579788001641e-07,
      "loss": 0.4663,
      "step": 347050
    },
    {
      "epoch": 142.06,
      "grad_norm": 2.3789639472961426,
      "learning_rate": 9.956681138735266e-07,
      "loss": 0.4505,
      "step": 347060
    },
    {
      "epoch": 142.07,
      "grad_norm": 2.110294818878174,
      "learning_rate": 9.955383539059974e-07,
      "loss": 0.4392,
      "step": 347070
    },
    {
      "epoch": 142.07,
      "grad_norm": 2.327660322189331,
      "learning_rate": 9.954086001144175e-07,
      "loss": 0.4474,
      "step": 347080
    },
    {
      "epoch": 142.08,
      "grad_norm": 1.840844988822937,
      "learning_rate": 9.952788524993803e-07,
      "loss": 0.4428,
      "step": 347090
    },
    {
      "epoch": 142.08,
      "grad_norm": 2.029465675354004,
      "learning_rate": 9.951491110614804e-07,
      "loss": 0.4588,
      "step": 347100
    },
    {
      "epoch": 142.08,
      "grad_norm": 2.312002658843994,
      "learning_rate": 9.950193758013123e-07,
      "loss": 0.4408,
      "step": 347110
    },
    {
      "epoch": 142.09,
      "grad_norm": 1.7868942022323608,
      "learning_rate": 9.948896467194703e-07,
      "loss": 0.4548,
      "step": 347120
    },
    {
      "epoch": 142.09,
      "grad_norm": 2.3461191654205322,
      "learning_rate": 9.947599238165488e-07,
      "loss": 0.4486,
      "step": 347130
    },
    {
      "epoch": 142.1,
      "grad_norm": 1.8805887699127197,
      "learning_rate": 9.946302070931418e-07,
      "loss": 0.4692,
      "step": 347140
    },
    {
      "epoch": 142.1,
      "grad_norm": 2.309832811355591,
      "learning_rate": 9.945004965498437e-07,
      "loss": 0.4776,
      "step": 347150
    },
    {
      "epoch": 142.1,
      "grad_norm": 2.0604753494262695,
      "learning_rate": 9.943707921872488e-07,
      "loss": 0.4474,
      "step": 347160
    },
    {
      "epoch": 142.11,
      "grad_norm": 2.4216394424438477,
      "learning_rate": 9.94241094005951e-07,
      "loss": 0.4424,
      "step": 347170
    },
    {
      "epoch": 142.11,
      "grad_norm": 1.7511173486709595,
      "learning_rate": 9.941114020065442e-07,
      "loss": 0.4431,
      "step": 347180
    },
    {
      "epoch": 142.12,
      "grad_norm": 2.1414005756378174,
      "learning_rate": 9.939817161896228e-07,
      "loss": 0.455,
      "step": 347190
    },
    {
      "epoch": 142.12,
      "grad_norm": 1.7401913404464722,
      "learning_rate": 9.938520365557813e-07,
      "loss": 0.4629,
      "step": 347200
    },
    {
      "epoch": 142.12,
      "grad_norm": 1.7153338193893433,
      "learning_rate": 9.93722363105613e-07,
      "loss": 0.4409,
      "step": 347210
    },
    {
      "epoch": 142.13,
      "grad_norm": 1.8119460344314575,
      "learning_rate": 9.935926958397122e-07,
      "loss": 0.461,
      "step": 347220
    },
    {
      "epoch": 142.13,
      "grad_norm": 2.1397078037261963,
      "learning_rate": 9.934630347586733e-07,
      "loss": 0.4426,
      "step": 347230
    },
    {
      "epoch": 142.14,
      "grad_norm": 2.0208358764648438,
      "learning_rate": 9.933333798630897e-07,
      "loss": 0.4526,
      "step": 347240
    },
    {
      "epoch": 142.14,
      "grad_norm": 2.365561008453369,
      "learning_rate": 9.93203731153555e-07,
      "loss": 0.4517,
      "step": 347250
    },
    {
      "epoch": 142.14,
      "grad_norm": 1.895226001739502,
      "learning_rate": 9.930740886306642e-07,
      "loss": 0.44,
      "step": 347260
    },
    {
      "epoch": 142.15,
      "grad_norm": 1.9392141103744507,
      "learning_rate": 9.9294445229501e-07,
      "loss": 0.4556,
      "step": 347270
    },
    {
      "epoch": 142.15,
      "grad_norm": 1.9149222373962402,
      "learning_rate": 9.928148221471875e-07,
      "loss": 0.4517,
      "step": 347280
    },
    {
      "epoch": 142.16,
      "grad_norm": 2.0396475791931152,
      "learning_rate": 9.926851981877895e-07,
      "loss": 0.4459,
      "step": 347290
    },
    {
      "epoch": 142.16,
      "grad_norm": 2.1969852447509766,
      "learning_rate": 9.9255558041741e-07,
      "loss": 0.4432,
      "step": 347300
    },
    {
      "epoch": 142.17,
      "grad_norm": 2.851313352584839,
      "learning_rate": 9.92425968836643e-07,
      "loss": 0.4621,
      "step": 347310
    },
    {
      "epoch": 142.17,
      "grad_norm": 2.0256972312927246,
      "learning_rate": 9.922963634460821e-07,
      "loss": 0.45,
      "step": 347320
    },
    {
      "epoch": 142.17,
      "grad_norm": 1.9675637483596802,
      "learning_rate": 9.92166764246321e-07,
      "loss": 0.4364,
      "step": 347330
    },
    {
      "epoch": 142.18,
      "grad_norm": 1.8493833541870117,
      "learning_rate": 9.920371712379536e-07,
      "loss": 0.4374,
      "step": 347340
    },
    {
      "epoch": 142.18,
      "grad_norm": 1.6609561443328857,
      "learning_rate": 9.91907584421573e-07,
      "loss": 0.4664,
      "step": 347350
    },
    {
      "epoch": 142.19,
      "grad_norm": 1.9883387088775635,
      "learning_rate": 9.917780037977732e-07,
      "loss": 0.467,
      "step": 347360
    },
    {
      "epoch": 142.19,
      "grad_norm": 2.121304988861084,
      "learning_rate": 9.91648429367148e-07,
      "loss": 0.4542,
      "step": 347370
    },
    {
      "epoch": 142.19,
      "grad_norm": 2.1064980030059814,
      "learning_rate": 9.915188611302905e-07,
      "loss": 0.4603,
      "step": 347380
    },
    {
      "epoch": 142.2,
      "grad_norm": 2.3073062896728516,
      "learning_rate": 9.913892990877952e-07,
      "loss": 0.471,
      "step": 347390
    },
    {
      "epoch": 142.2,
      "grad_norm": 1.8613146543502808,
      "learning_rate": 9.91259743240254e-07,
      "loss": 0.439,
      "step": 347400
    },
    {
      "epoch": 142.21,
      "grad_norm": 2.0007262229919434,
      "learning_rate": 9.911301935882613e-07,
      "loss": 0.4299,
      "step": 347410
    },
    {
      "epoch": 142.21,
      "grad_norm": 1.8772430419921875,
      "learning_rate": 9.910006501324105e-07,
      "loss": 0.4595,
      "step": 347420
    },
    {
      "epoch": 142.21,
      "grad_norm": 1.9164884090423584,
      "learning_rate": 9.90871112873294e-07,
      "loss": 0.4565,
      "step": 347430
    },
    {
      "epoch": 142.22,
      "grad_norm": 2.106227159500122,
      "learning_rate": 9.90741581811507e-07,
      "loss": 0.4616,
      "step": 347440
    },
    {
      "epoch": 142.22,
      "grad_norm": 2.1039555072784424,
      "learning_rate": 9.90612056947642e-07,
      "loss": 0.4401,
      "step": 347450
    },
    {
      "epoch": 142.23,
      "grad_norm": 2.272519826889038,
      "learning_rate": 9.904825382822926e-07,
      "loss": 0.444,
      "step": 347460
    },
    {
      "epoch": 142.23,
      "grad_norm": 1.982784390449524,
      "learning_rate": 9.903530258160515e-07,
      "loss": 0.4537,
      "step": 347470
    },
    {
      "epoch": 142.23,
      "grad_norm": 2.2258670330047607,
      "learning_rate": 9.902235195495128e-07,
      "loss": 0.4553,
      "step": 347480
    },
    {
      "epoch": 142.24,
      "grad_norm": 2.1735851764678955,
      "learning_rate": 9.90094019483269e-07,
      "loss": 0.4592,
      "step": 347490
    },
    {
      "epoch": 142.24,
      "grad_norm": 1.9400471448898315,
      "learning_rate": 9.899645256179137e-07,
      "loss": 0.4416,
      "step": 347500
    },
    {
      "epoch": 142.25,
      "grad_norm": 2.2104594707489014,
      "learning_rate": 9.898350379540399e-07,
      "loss": 0.4454,
      "step": 347510
    },
    {
      "epoch": 142.25,
      "grad_norm": 2.2252280712127686,
      "learning_rate": 9.897055564922408e-07,
      "loss": 0.4638,
      "step": 347520
    },
    {
      "epoch": 142.26,
      "grad_norm": 1.9923441410064697,
      "learning_rate": 9.895760812331104e-07,
      "loss": 0.4765,
      "step": 347530
    },
    {
      "epoch": 142.26,
      "grad_norm": 1.9485957622528076,
      "learning_rate": 9.894466121772405e-07,
      "loss": 0.4661,
      "step": 347540
    },
    {
      "epoch": 142.26,
      "grad_norm": 2.1592249870300293,
      "learning_rate": 9.893171493252243e-07,
      "loss": 0.4405,
      "step": 347550
    },
    {
      "epoch": 142.27,
      "grad_norm": 2.1559317111968994,
      "learning_rate": 9.891876926776554e-07,
      "loss": 0.4703,
      "step": 347560
    },
    {
      "epoch": 142.27,
      "grad_norm": 2.0110301971435547,
      "learning_rate": 9.890582422351267e-07,
      "loss": 0.4562,
      "step": 347570
    },
    {
      "epoch": 142.28,
      "grad_norm": 2.068540334701538,
      "learning_rate": 9.88928797998231e-07,
      "loss": 0.4434,
      "step": 347580
    },
    {
      "epoch": 142.28,
      "grad_norm": 2.7849955558776855,
      "learning_rate": 9.887993599675618e-07,
      "loss": 0.4623,
      "step": 347590
    },
    {
      "epoch": 142.28,
      "grad_norm": 1.851396083831787,
      "learning_rate": 9.886699281437111e-07,
      "loss": 0.4578,
      "step": 347600
    },
    {
      "epoch": 142.29,
      "grad_norm": 2.074734687805176,
      "learning_rate": 9.885405025272726e-07,
      "loss": 0.4761,
      "step": 347610
    },
    {
      "epoch": 142.29,
      "grad_norm": 2.1099159717559814,
      "learning_rate": 9.88411083118838e-07,
      "loss": 0.4649,
      "step": 347620
    },
    {
      "epoch": 142.3,
      "grad_norm": 1.8264858722686768,
      "learning_rate": 9.88281669919002e-07,
      "loss": 0.4465,
      "step": 347630
    },
    {
      "epoch": 142.3,
      "grad_norm": 1.8269789218902588,
      "learning_rate": 9.881522629283565e-07,
      "loss": 0.447,
      "step": 347640
    },
    {
      "epoch": 142.3,
      "grad_norm": 2.314171552658081,
      "learning_rate": 9.880228621474945e-07,
      "loss": 0.4521,
      "step": 347650
    },
    {
      "epoch": 142.31,
      "grad_norm": 1.7418596744537354,
      "learning_rate": 9.878934675770084e-07,
      "loss": 0.4503,
      "step": 347660
    },
    {
      "epoch": 142.31,
      "grad_norm": 1.9093698263168335,
      "learning_rate": 9.877640792174915e-07,
      "loss": 0.4336,
      "step": 347670
    },
    {
      "epoch": 142.32,
      "grad_norm": 2.6431467533111572,
      "learning_rate": 9.87634697069536e-07,
      "loss": 0.4588,
      "step": 347680
    },
    {
      "epoch": 142.32,
      "grad_norm": 2.028193473815918,
      "learning_rate": 9.87505321133734e-07,
      "loss": 0.4433,
      "step": 347690
    },
    {
      "epoch": 142.33,
      "grad_norm": 2.1517767906188965,
      "learning_rate": 9.873759514106792e-07,
      "loss": 0.4405,
      "step": 347700
    },
    {
      "epoch": 142.33,
      "grad_norm": 1.9433751106262207,
      "learning_rate": 9.87246587900964e-07,
      "loss": 0.4653,
      "step": 347710
    },
    {
      "epoch": 142.33,
      "grad_norm": 2.1756913661956787,
      "learning_rate": 9.871172306051806e-07,
      "loss": 0.4448,
      "step": 347720
    },
    {
      "epoch": 142.34,
      "grad_norm": 2.1699633598327637,
      "learning_rate": 9.869878795239219e-07,
      "loss": 0.4449,
      "step": 347730
    },
    {
      "epoch": 142.34,
      "grad_norm": 1.6780331134796143,
      "learning_rate": 9.868585346577803e-07,
      "loss": 0.4683,
      "step": 347740
    },
    {
      "epoch": 142.35,
      "grad_norm": 1.6141117811203003,
      "learning_rate": 9.867291960073485e-07,
      "loss": 0.4538,
      "step": 347750
    },
    {
      "epoch": 142.35,
      "grad_norm": 1.7033034563064575,
      "learning_rate": 9.865998635732188e-07,
      "loss": 0.441,
      "step": 347760
    },
    {
      "epoch": 142.35,
      "grad_norm": 2.006150007247925,
      "learning_rate": 9.864705373559836e-07,
      "loss": 0.4379,
      "step": 347770
    },
    {
      "epoch": 142.36,
      "grad_norm": 16.460630416870117,
      "learning_rate": 9.863412173562355e-07,
      "loss": 0.4335,
      "step": 347780
    },
    {
      "epoch": 142.36,
      "grad_norm": 1.7884092330932617,
      "learning_rate": 9.862119035745664e-07,
      "loss": 0.4445,
      "step": 347790
    },
    {
      "epoch": 142.37,
      "grad_norm": 2.352707862854004,
      "learning_rate": 9.860825960115696e-07,
      "loss": 0.4663,
      "step": 347800
    },
    {
      "epoch": 142.37,
      "grad_norm": 1.9338387250900269,
      "learning_rate": 9.859532946678367e-07,
      "loss": 0.4884,
      "step": 347810
    },
    {
      "epoch": 142.37,
      "grad_norm": 1.9906624555587769,
      "learning_rate": 9.858239995439604e-07,
      "loss": 0.4496,
      "step": 347820
    },
    {
      "epoch": 142.38,
      "grad_norm": 2.086937189102173,
      "learning_rate": 9.856947106405327e-07,
      "loss": 0.4392,
      "step": 347830
    },
    {
      "epoch": 142.38,
      "grad_norm": 2.3177330493927,
      "learning_rate": 9.855654279581457e-07,
      "loss": 0.4667,
      "step": 347840
    },
    {
      "epoch": 142.39,
      "grad_norm": 1.9418079853057861,
      "learning_rate": 9.854361514973922e-07,
      "loss": 0.4486,
      "step": 347850
    },
    {
      "epoch": 142.39,
      "grad_norm": 1.6970738172531128,
      "learning_rate": 9.85306881258864e-07,
      "loss": 0.454,
      "step": 347860
    },
    {
      "epoch": 142.39,
      "grad_norm": 1.684491753578186,
      "learning_rate": 9.851776172431534e-07,
      "loss": 0.4461,
      "step": 347870
    },
    {
      "epoch": 142.4,
      "grad_norm": 1.8155933618545532,
      "learning_rate": 9.850483594508524e-07,
      "loss": 0.441,
      "step": 347880
    },
    {
      "epoch": 142.4,
      "grad_norm": 2.0758378505706787,
      "learning_rate": 9.849191078825533e-07,
      "loss": 0.4497,
      "step": 347890
    },
    {
      "epoch": 142.41,
      "grad_norm": 1.8506556749343872,
      "learning_rate": 9.84789862538848e-07,
      "loss": 0.4478,
      "step": 347900
    },
    {
      "epoch": 142.41,
      "grad_norm": 2.276841402053833,
      "learning_rate": 9.846606234203285e-07,
      "loss": 0.4539,
      "step": 347910
    },
    {
      "epoch": 142.42,
      "grad_norm": 1.7683132886886597,
      "learning_rate": 9.845313905275872e-07,
      "loss": 0.4246,
      "step": 347920
    },
    {
      "epoch": 142.42,
      "grad_norm": 2.047144889831543,
      "learning_rate": 9.844021638612156e-07,
      "loss": 0.4391,
      "step": 347930
    },
    {
      "epoch": 142.42,
      "grad_norm": 1.9745205640792847,
      "learning_rate": 9.842729434218061e-07,
      "loss": 0.447,
      "step": 347940
    },
    {
      "epoch": 142.43,
      "grad_norm": 2.5909886360168457,
      "learning_rate": 9.841437292099504e-07,
      "loss": 0.4492,
      "step": 347950
    },
    {
      "epoch": 142.43,
      "grad_norm": 2.3643853664398193,
      "learning_rate": 9.840145212262407e-07,
      "loss": 0.4583,
      "step": 347960
    },
    {
      "epoch": 142.44,
      "grad_norm": 1.8918756246566772,
      "learning_rate": 9.83885319471269e-07,
      "loss": 0.4191,
      "step": 347970
    },
    {
      "epoch": 142.44,
      "grad_norm": 1.9845731258392334,
      "learning_rate": 9.837561239456263e-07,
      "loss": 0.4773,
      "step": 347980
    },
    {
      "epoch": 142.44,
      "grad_norm": 1.900743007659912,
      "learning_rate": 9.83626934649905e-07,
      "loss": 0.4442,
      "step": 347990
    },
    {
      "epoch": 142.45,
      "grad_norm": 1.9770076274871826,
      "learning_rate": 9.834977515846956e-07,
      "loss": 0.4497,
      "step": 348000
    },
    {
      "epoch": 142.45,
      "grad_norm": 1.848157525062561,
      "learning_rate": 9.833685747505922e-07,
      "loss": 0.4508,
      "step": 348010
    },
    {
      "epoch": 142.46,
      "grad_norm": 2.3156118392944336,
      "learning_rate": 9.832394041481854e-07,
      "loss": 0.4566,
      "step": 348020
    },
    {
      "epoch": 142.46,
      "grad_norm": 1.9988449811935425,
      "learning_rate": 9.831102397780668e-07,
      "loss": 0.4623,
      "step": 348030
    },
    {
      "epoch": 142.46,
      "grad_norm": 1.99440598487854,
      "learning_rate": 9.829810816408285e-07,
      "loss": 0.4468,
      "step": 348040
    },
    {
      "epoch": 142.47,
      "grad_norm": 1.8438116312026978,
      "learning_rate": 9.828519297370615e-07,
      "loss": 0.4416,
      "step": 348050
    },
    {
      "epoch": 142.47,
      "grad_norm": 2.0146801471710205,
      "learning_rate": 9.827227840673582e-07,
      "loss": 0.4741,
      "step": 348060
    },
    {
      "epoch": 142.48,
      "grad_norm": 1.8992074728012085,
      "learning_rate": 9.825936446323094e-07,
      "loss": 0.453,
      "step": 348070
    },
    {
      "epoch": 142.48,
      "grad_norm": 2.004957914352417,
      "learning_rate": 9.824645114325075e-07,
      "loss": 0.468,
      "step": 348080
    },
    {
      "epoch": 142.48,
      "grad_norm": 2.2469115257263184,
      "learning_rate": 9.823353844685434e-07,
      "loss": 0.462,
      "step": 348090
    },
    {
      "epoch": 142.49,
      "grad_norm": 1.8490393161773682,
      "learning_rate": 9.822062637410086e-07,
      "loss": 0.4537,
      "step": 348100
    },
    {
      "epoch": 142.49,
      "grad_norm": 1.6639480590820312,
      "learning_rate": 9.82077149250496e-07,
      "loss": 0.4445,
      "step": 348110
    },
    {
      "epoch": 142.5,
      "grad_norm": 2.4905741214752197,
      "learning_rate": 9.819480409975948e-07,
      "loss": 0.4639,
      "step": 348120
    },
    {
      "epoch": 142.5,
      "grad_norm": 1.979591965675354,
      "learning_rate": 9.818189389828972e-07,
      "loss": 0.461,
      "step": 348130
    },
    {
      "epoch": 142.51,
      "grad_norm": 2.2234249114990234,
      "learning_rate": 9.816898432069951e-07,
      "loss": 0.4552,
      "step": 348140
    },
    {
      "epoch": 142.51,
      "grad_norm": 2.1917459964752197,
      "learning_rate": 9.8156075367048e-07,
      "loss": 0.4648,
      "step": 348150
    },
    {
      "epoch": 142.51,
      "grad_norm": 2.491506814956665,
      "learning_rate": 9.814316703739423e-07,
      "loss": 0.4634,
      "step": 348160
    },
    {
      "epoch": 142.52,
      "grad_norm": 2.116971731185913,
      "learning_rate": 9.813025933179743e-07,
      "loss": 0.4678,
      "step": 348170
    },
    {
      "epoch": 142.52,
      "grad_norm": 1.9530161619186401,
      "learning_rate": 9.811735225031658e-07,
      "loss": 0.4651,
      "step": 348180
    },
    {
      "epoch": 142.53,
      "grad_norm": 1.931705355644226,
      "learning_rate": 9.810444579301102e-07,
      "loss": 0.4628,
      "step": 348190
    },
    {
      "epoch": 142.53,
      "grad_norm": 2.129192352294922,
      "learning_rate": 9.809153995993974e-07,
      "loss": 0.48,
      "step": 348200
    },
    {
      "epoch": 142.53,
      "grad_norm": 2.281430721282959,
      "learning_rate": 9.80786347511619e-07,
      "loss": 0.4518,
      "step": 348210
    },
    {
      "epoch": 142.54,
      "grad_norm": 2.4099133014678955,
      "learning_rate": 9.806573016673658e-07,
      "loss": 0.4436,
      "step": 348220
    },
    {
      "epoch": 142.54,
      "grad_norm": 2.0516390800476074,
      "learning_rate": 9.805282620672294e-07,
      "loss": 0.4542,
      "step": 348230
    },
    {
      "epoch": 142.55,
      "grad_norm": 2.1539106369018555,
      "learning_rate": 9.803992287118005e-07,
      "loss": 0.4454,
      "step": 348240
    },
    {
      "epoch": 142.55,
      "grad_norm": 1.7770904302597046,
      "learning_rate": 9.802702016016712e-07,
      "loss": 0.4656,
      "step": 348250
    },
    {
      "epoch": 142.55,
      "grad_norm": 2.2114365100860596,
      "learning_rate": 9.80141180737431e-07,
      "loss": 0.4395,
      "step": 348260
    },
    {
      "epoch": 142.56,
      "grad_norm": 2.030322551727295,
      "learning_rate": 9.800121661196716e-07,
      "loss": 0.4463,
      "step": 348270
    },
    {
      "epoch": 142.56,
      "grad_norm": 1.9341880083084106,
      "learning_rate": 9.798831577489837e-07,
      "loss": 0.4563,
      "step": 348280
    },
    {
      "epoch": 142.57,
      "grad_norm": 2.099931001663208,
      "learning_rate": 9.797541556259588e-07,
      "loss": 0.4422,
      "step": 348290
    },
    {
      "epoch": 142.57,
      "grad_norm": 2.649958372116089,
      "learning_rate": 9.79625159751188e-07,
      "loss": 0.4522,
      "step": 348300
    },
    {
      "epoch": 142.57,
      "grad_norm": 1.985465168952942,
      "learning_rate": 9.794961701252613e-07,
      "loss": 0.4392,
      "step": 348310
    },
    {
      "epoch": 142.58,
      "grad_norm": 1.680808186531067,
      "learning_rate": 9.793671867487702e-07,
      "loss": 0.4451,
      "step": 348320
    },
    {
      "epoch": 142.58,
      "grad_norm": 2.140368938446045,
      "learning_rate": 9.792382096223055e-07,
      "loss": 0.4447,
      "step": 348330
    },
    {
      "epoch": 142.59,
      "grad_norm": 2.0336573123931885,
      "learning_rate": 9.79109238746458e-07,
      "loss": 0.4576,
      "step": 348340
    },
    {
      "epoch": 142.59,
      "grad_norm": 1.9490089416503906,
      "learning_rate": 9.789802741218187e-07,
      "loss": 0.4454,
      "step": 348350
    },
    {
      "epoch": 142.6,
      "grad_norm": 1.839054822921753,
      "learning_rate": 9.78851315748978e-07,
      "loss": 0.4733,
      "step": 348360
    },
    {
      "epoch": 142.6,
      "grad_norm": 2.0812599658966064,
      "learning_rate": 9.787223636285257e-07,
      "loss": 0.4471,
      "step": 348370
    },
    {
      "epoch": 142.6,
      "grad_norm": 1.7973685264587402,
      "learning_rate": 9.785934177610557e-07,
      "loss": 0.4478,
      "step": 348380
    },
    {
      "epoch": 142.61,
      "grad_norm": 2.155604362487793,
      "learning_rate": 9.784644781471554e-07,
      "loss": 0.4409,
      "step": 348390
    },
    {
      "epoch": 142.61,
      "grad_norm": 1.919248104095459,
      "learning_rate": 9.78335544787417e-07,
      "loss": 0.4409,
      "step": 348400
    },
    {
      "epoch": 142.62,
      "grad_norm": 2.0050885677337646,
      "learning_rate": 9.782066176824305e-07,
      "loss": 0.4468,
      "step": 348410
    },
    {
      "epoch": 142.62,
      "grad_norm": 1.9764823913574219,
      "learning_rate": 9.780776968327867e-07,
      "loss": 0.4468,
      "step": 348420
    },
    {
      "epoch": 142.62,
      "grad_norm": 2.005141496658325,
      "learning_rate": 9.779487822390762e-07,
      "loss": 0.4567,
      "step": 348430
    },
    {
      "epoch": 142.63,
      "grad_norm": 2.3476150035858154,
      "learning_rate": 9.778198739018897e-07,
      "loss": 0.4514,
      "step": 348440
    },
    {
      "epoch": 142.63,
      "grad_norm": 2.2301108837127686,
      "learning_rate": 9.776909718218178e-07,
      "loss": 0.457,
      "step": 348450
    },
    {
      "epoch": 142.64,
      "grad_norm": 2.1257641315460205,
      "learning_rate": 9.775620759994503e-07,
      "loss": 0.4288,
      "step": 348460
    },
    {
      "epoch": 142.64,
      "grad_norm": 1.9745670557022095,
      "learning_rate": 9.774331864353784e-07,
      "loss": 0.4492,
      "step": 348470
    },
    {
      "epoch": 142.64,
      "grad_norm": 1.7964321374893188,
      "learning_rate": 9.773043031301921e-07,
      "loss": 0.46,
      "step": 348480
    },
    {
      "epoch": 142.65,
      "grad_norm": 1.8986058235168457,
      "learning_rate": 9.77175426084482e-07,
      "loss": 0.455,
      "step": 348490
    },
    {
      "epoch": 142.65,
      "grad_norm": 1.6452651023864746,
      "learning_rate": 9.770465552988382e-07,
      "loss": 0.4536,
      "step": 348500
    },
    {
      "epoch": 142.66,
      "grad_norm": 2.689945697784424,
      "learning_rate": 9.769176907738511e-07,
      "loss": 0.4369,
      "step": 348510
    },
    {
      "epoch": 142.66,
      "grad_norm": 1.6514875888824463,
      "learning_rate": 9.767888325101117e-07,
      "loss": 0.4578,
      "step": 348520
    },
    {
      "epoch": 142.66,
      "grad_norm": 3.19810152053833,
      "learning_rate": 9.766599805082094e-07,
      "loss": 0.4605,
      "step": 348530
    },
    {
      "epoch": 142.67,
      "grad_norm": 2.148542642593384,
      "learning_rate": 9.765311347687344e-07,
      "loss": 0.4605,
      "step": 348540
    },
    {
      "epoch": 142.67,
      "grad_norm": 1.9103559255599976,
      "learning_rate": 9.764022952922779e-07,
      "loss": 0.4318,
      "step": 348550
    },
    {
      "epoch": 142.68,
      "grad_norm": 2.404646873474121,
      "learning_rate": 9.762734620794292e-07,
      "loss": 0.4447,
      "step": 348560
    },
    {
      "epoch": 142.68,
      "grad_norm": 1.6446774005889893,
      "learning_rate": 9.761446351307784e-07,
      "loss": 0.4636,
      "step": 348570
    },
    {
      "epoch": 142.69,
      "grad_norm": 2.0283117294311523,
      "learning_rate": 9.760158144469164e-07,
      "loss": 0.4359,
      "step": 348580
    },
    {
      "epoch": 142.69,
      "grad_norm": 1.6614104509353638,
      "learning_rate": 9.758870000284329e-07,
      "loss": 0.4591,
      "step": 348590
    },
    {
      "epoch": 142.69,
      "grad_norm": 2.4600930213928223,
      "learning_rate": 9.757581918759177e-07,
      "loss": 0.4424,
      "step": 348600
    },
    {
      "epoch": 142.7,
      "grad_norm": 2.047912836074829,
      "learning_rate": 9.756293899899613e-07,
      "loss": 0.4704,
      "step": 348610
    },
    {
      "epoch": 142.7,
      "grad_norm": 1.6961554288864136,
      "learning_rate": 9.755005943711536e-07,
      "loss": 0.4564,
      "step": 348620
    },
    {
      "epoch": 142.71,
      "grad_norm": 2.308239221572876,
      "learning_rate": 9.753718050200843e-07,
      "loss": 0.4632,
      "step": 348630
    },
    {
      "epoch": 142.71,
      "grad_norm": 2.17899227142334,
      "learning_rate": 9.752430219373437e-07,
      "loss": 0.4622,
      "step": 348640
    },
    {
      "epoch": 142.71,
      "grad_norm": 2.4136884212493896,
      "learning_rate": 9.751142451235216e-07,
      "loss": 0.4305,
      "step": 348650
    },
    {
      "epoch": 142.72,
      "grad_norm": 2.094529867172241,
      "learning_rate": 9.749854745792081e-07,
      "loss": 0.4528,
      "step": 348660
    },
    {
      "epoch": 142.72,
      "grad_norm": 1.5125555992126465,
      "learning_rate": 9.748567103049927e-07,
      "loss": 0.4475,
      "step": 348670
    },
    {
      "epoch": 142.73,
      "grad_norm": 2.0918285846710205,
      "learning_rate": 9.747279523014655e-07,
      "loss": 0.4553,
      "step": 348680
    },
    {
      "epoch": 142.73,
      "grad_norm": 2.241086959838867,
      "learning_rate": 9.745992005692172e-07,
      "loss": 0.4407,
      "step": 348690
    },
    {
      "epoch": 142.73,
      "grad_norm": 2.147109270095825,
      "learning_rate": 9.744704551088357e-07,
      "loss": 0.448,
      "step": 348700
    },
    {
      "epoch": 142.74,
      "grad_norm": 2.1419193744659424,
      "learning_rate": 9.74341715920912e-07,
      "loss": 0.4499,
      "step": 348710
    },
    {
      "epoch": 142.74,
      "grad_norm": 2.249892234802246,
      "learning_rate": 9.742129830060352e-07,
      "loss": 0.4414,
      "step": 348720
    },
    {
      "epoch": 142.75,
      "grad_norm": 1.968926191329956,
      "learning_rate": 9.740842563647957e-07,
      "loss": 0.4599,
      "step": 348730
    },
    {
      "epoch": 142.75,
      "grad_norm": 2.173645496368408,
      "learning_rate": 9.739555359977827e-07,
      "loss": 0.4387,
      "step": 348740
    },
    {
      "epoch": 142.75,
      "grad_norm": 1.7214653491973877,
      "learning_rate": 9.738268219055853e-07,
      "loss": 0.4437,
      "step": 348750
    },
    {
      "epoch": 142.76,
      "grad_norm": 1.7988526821136475,
      "learning_rate": 9.736981140887943e-07,
      "loss": 0.4633,
      "step": 348760
    },
    {
      "epoch": 142.76,
      "grad_norm": 2.5775842666625977,
      "learning_rate": 9.73569412547999e-07,
      "loss": 0.4447,
      "step": 348770
    },
    {
      "epoch": 142.77,
      "grad_norm": 2.065310478210449,
      "learning_rate": 9.73440717283789e-07,
      "loss": 0.435,
      "step": 348780
    },
    {
      "epoch": 142.77,
      "grad_norm": 1.9492441415786743,
      "learning_rate": 9.733120282967534e-07,
      "loss": 0.4456,
      "step": 348790
    },
    {
      "epoch": 142.78,
      "grad_norm": 2.3857340812683105,
      "learning_rate": 9.73183345587482e-07,
      "loss": 0.4215,
      "step": 348800
    },
    {
      "epoch": 142.78,
      "grad_norm": 2.2286999225616455,
      "learning_rate": 9.730546691565641e-07,
      "loss": 0.4518,
      "step": 348810
    },
    {
      "epoch": 142.78,
      "grad_norm": 5.822037696838379,
      "learning_rate": 9.729259990045892e-07,
      "loss": 0.4558,
      "step": 348820
    },
    {
      "epoch": 142.79,
      "grad_norm": 1.9081887006759644,
      "learning_rate": 9.727973351321477e-07,
      "loss": 0.4521,
      "step": 348830
    },
    {
      "epoch": 142.79,
      "grad_norm": 1.608760952949524,
      "learning_rate": 9.726686775398272e-07,
      "loss": 0.4391,
      "step": 348840
    },
    {
      "epoch": 142.8,
      "grad_norm": 1.9859930276870728,
      "learning_rate": 9.725400262282177e-07,
      "loss": 0.4447,
      "step": 348850
    },
    {
      "epoch": 142.8,
      "grad_norm": 1.955098271369934,
      "learning_rate": 9.72411381197909e-07,
      "loss": 0.4332,
      "step": 348860
    },
    {
      "epoch": 142.8,
      "grad_norm": 1.7435353994369507,
      "learning_rate": 9.7228274244949e-07,
      "loss": 0.4335,
      "step": 348870
    },
    {
      "epoch": 142.81,
      "grad_norm": 1.9211448431015015,
      "learning_rate": 9.721541099835502e-07,
      "loss": 0.4536,
      "step": 348880
    },
    {
      "epoch": 142.81,
      "grad_norm": 2.086138963699341,
      "learning_rate": 9.720254838006784e-07,
      "loss": 0.4588,
      "step": 348890
    },
    {
      "epoch": 142.82,
      "grad_norm": 1.9405813217163086,
      "learning_rate": 9.718968639014647e-07,
      "loss": 0.4461,
      "step": 348900
    },
    {
      "epoch": 142.82,
      "grad_norm": 1.9142043590545654,
      "learning_rate": 9.717682502864974e-07,
      "loss": 0.4564,
      "step": 348910
    },
    {
      "epoch": 142.82,
      "grad_norm": 2.378873586654663,
      "learning_rate": 9.716396429563662e-07,
      "loss": 0.4452,
      "step": 348920
    },
    {
      "epoch": 142.83,
      "grad_norm": 2.032862424850464,
      "learning_rate": 9.715110419116597e-07,
      "loss": 0.4311,
      "step": 348930
    },
    {
      "epoch": 142.83,
      "grad_norm": 2.297776937484741,
      "learning_rate": 9.713824471529668e-07,
      "loss": 0.4384,
      "step": 348940
    },
    {
      "epoch": 142.84,
      "grad_norm": 2.182664394378662,
      "learning_rate": 9.712538586808782e-07,
      "loss": 0.4527,
      "step": 348950
    },
    {
      "epoch": 142.84,
      "grad_norm": 2.0706138610839844,
      "learning_rate": 9.711252764959815e-07,
      "loss": 0.472,
      "step": 348960
    },
    {
      "epoch": 142.84,
      "grad_norm": 1.8203421831130981,
      "learning_rate": 9.709967005988671e-07,
      "loss": 0.4331,
      "step": 348970
    },
    {
      "epoch": 142.85,
      "grad_norm": 2.2139792442321777,
      "learning_rate": 9.70868130990122e-07,
      "loss": 0.4623,
      "step": 348980
    },
    {
      "epoch": 142.85,
      "grad_norm": 1.8321561813354492,
      "learning_rate": 9.70739567670336e-07,
      "loss": 0.4504,
      "step": 348990
    },
    {
      "epoch": 142.86,
      "grad_norm": 1.6741780042648315,
      "learning_rate": 9.706110106400986e-07,
      "loss": 0.4385,
      "step": 349000
    },
    {
      "epoch": 142.86,
      "grad_norm": 2.3677093982696533,
      "learning_rate": 9.704824598999978e-07,
      "loss": 0.442,
      "step": 349010
    },
    {
      "epoch": 142.87,
      "grad_norm": 1.703094482421875,
      "learning_rate": 9.70353915450623e-07,
      "loss": 0.4539,
      "step": 349020
    },
    {
      "epoch": 142.87,
      "grad_norm": 1.6566814184188843,
      "learning_rate": 9.702253772925632e-07,
      "loss": 0.4394,
      "step": 349030
    },
    {
      "epoch": 142.87,
      "grad_norm": 1.7400606870651245,
      "learning_rate": 9.700968454264067e-07,
      "loss": 0.4684,
      "step": 349040
    },
    {
      "epoch": 142.88,
      "grad_norm": 1.974274754524231,
      "learning_rate": 9.69968319852743e-07,
      "loss": 0.458,
      "step": 349050
    },
    {
      "epoch": 142.88,
      "grad_norm": 2.218754768371582,
      "learning_rate": 9.698398005721601e-07,
      "loss": 0.4773,
      "step": 349060
    },
    {
      "epoch": 142.89,
      "grad_norm": 2.4794490337371826,
      "learning_rate": 9.69711287585247e-07,
      "loss": 0.4593,
      "step": 349070
    },
    {
      "epoch": 142.89,
      "grad_norm": 2.76751971244812,
      "learning_rate": 9.695827808925925e-07,
      "loss": 0.4561,
      "step": 349080
    },
    {
      "epoch": 142.89,
      "grad_norm": 1.876145362854004,
      "learning_rate": 9.694542804947851e-07,
      "loss": 0.4588,
      "step": 349090
    },
    {
      "epoch": 142.9,
      "grad_norm": 2.017470359802246,
      "learning_rate": 9.693257863924138e-07,
      "loss": 0.456,
      "step": 349100
    },
    {
      "epoch": 142.9,
      "grad_norm": 2.057530164718628,
      "learning_rate": 9.691972985860669e-07,
      "loss": 0.4194,
      "step": 349110
    },
    {
      "epoch": 142.91,
      "grad_norm": 2.1254053115844727,
      "learning_rate": 9.69068817076333e-07,
      "loss": 0.4474,
      "step": 349120
    },
    {
      "epoch": 142.91,
      "grad_norm": 1.8163808584213257,
      "learning_rate": 9.689403418638007e-07,
      "loss": 0.4447,
      "step": 349130
    },
    {
      "epoch": 142.91,
      "grad_norm": 2.3135898113250732,
      "learning_rate": 9.688118729490585e-07,
      "loss": 0.4309,
      "step": 349140
    },
    {
      "epoch": 142.92,
      "grad_norm": 2.321291208267212,
      "learning_rate": 9.68683410332695e-07,
      "loss": 0.4502,
      "step": 349150
    },
    {
      "epoch": 142.92,
      "grad_norm": 1.8073408603668213,
      "learning_rate": 9.685549540152985e-07,
      "loss": 0.432,
      "step": 349160
    },
    {
      "epoch": 142.93,
      "grad_norm": 1.9285962581634521,
      "learning_rate": 9.684265039974575e-07,
      "loss": 0.4504,
      "step": 349170
    },
    {
      "epoch": 142.93,
      "grad_norm": 2.05255126953125,
      "learning_rate": 9.682980602797606e-07,
      "loss": 0.4353,
      "step": 349180
    },
    {
      "epoch": 142.93,
      "grad_norm": 1.8237684965133667,
      "learning_rate": 9.68169622862796e-07,
      "loss": 0.438,
      "step": 349190
    },
    {
      "epoch": 142.94,
      "grad_norm": 2.040066957473755,
      "learning_rate": 9.68041191747152e-07,
      "loss": 0.4614,
      "step": 349200
    },
    {
      "epoch": 142.94,
      "grad_norm": 1.8929376602172852,
      "learning_rate": 9.67912766933417e-07,
      "loss": 0.4471,
      "step": 349210
    },
    {
      "epoch": 142.95,
      "grad_norm": 2.3358957767486572,
      "learning_rate": 9.677843484221793e-07,
      "loss": 0.4409,
      "step": 349220
    },
    {
      "epoch": 142.95,
      "grad_norm": 1.8464863300323486,
      "learning_rate": 9.676559362140273e-07,
      "loss": 0.4482,
      "step": 349230
    },
    {
      "epoch": 142.96,
      "grad_norm": 2.1012327671051025,
      "learning_rate": 9.675275303095488e-07,
      "loss": 0.4544,
      "step": 349240
    },
    {
      "epoch": 142.96,
      "grad_norm": 1.6853625774383545,
      "learning_rate": 9.673991307093326e-07,
      "loss": 0.4698,
      "step": 349250
    },
    {
      "epoch": 142.96,
      "grad_norm": 1.8337355852127075,
      "learning_rate": 9.672707374139663e-07,
      "loss": 0.4345,
      "step": 349260
    },
    {
      "epoch": 142.97,
      "grad_norm": 1.9183584451675415,
      "learning_rate": 9.671423504240386e-07,
      "loss": 0.449,
      "step": 349270
    },
    {
      "epoch": 142.97,
      "grad_norm": 2.6585168838500977,
      "learning_rate": 9.67013969740138e-07,
      "loss": 0.4325,
      "step": 349280
    },
    {
      "epoch": 142.98,
      "grad_norm": 16.26297378540039,
      "learning_rate": 9.66885595362851e-07,
      "loss": 0.4612,
      "step": 349290
    },
    {
      "epoch": 142.98,
      "grad_norm": 1.7347614765167236,
      "learning_rate": 9.667572272927667e-07,
      "loss": 0.4587,
      "step": 349300
    },
    {
      "epoch": 142.98,
      "grad_norm": 2.0247414112091064,
      "learning_rate": 9.66628865530473e-07,
      "loss": 0.4417,
      "step": 349310
    },
    {
      "epoch": 142.99,
      "grad_norm": 2.3803458213806152,
      "learning_rate": 9.665005100765573e-07,
      "loss": 0.4684,
      "step": 349320
    },
    {
      "epoch": 142.99,
      "grad_norm": 1.8267402648925781,
      "learning_rate": 9.66372160931609e-07,
      "loss": 0.4561,
      "step": 349330
    },
    {
      "epoch": 143.0,
      "grad_norm": 2.1333556175231934,
      "learning_rate": 9.662438180962152e-07,
      "loss": 0.459,
      "step": 349340
    },
    {
      "epoch": 143.0,
      "eval_loss": 0.4505051076412201,
      "eval_runtime": 53.0147,
      "eval_samples_per_second": 65.057,
      "eval_steps_per_second": 8.149,
      "step": 349349
    },
    {
      "epoch": 143.0,
      "grad_norm": 2.20904278755188,
      "learning_rate": 9.661154815709638e-07,
      "loss": 0.4749,
      "step": 349350
    },
    {
      "epoch": 143.0,
      "grad_norm": 2.0885467529296875,
      "learning_rate": 9.65987151356443e-07,
      "loss": 0.4449,
      "step": 349360
    },
    {
      "epoch": 143.01,
      "grad_norm": 1.811445951461792,
      "learning_rate": 9.658588274532403e-07,
      "loss": 0.4547,
      "step": 349370
    },
    {
      "epoch": 143.01,
      "grad_norm": 2.2006211280822754,
      "learning_rate": 9.657305098619436e-07,
      "loss": 0.4771,
      "step": 349380
    },
    {
      "epoch": 143.02,
      "grad_norm": 1.698685884475708,
      "learning_rate": 9.65602198583141e-07,
      "loss": 0.4349,
      "step": 349390
    },
    {
      "epoch": 143.02,
      "grad_norm": 2.113260269165039,
      "learning_rate": 9.654738936174199e-07,
      "loss": 0.4608,
      "step": 349400
    },
    {
      "epoch": 143.02,
      "grad_norm": 2.306551933288574,
      "learning_rate": 9.65345594965369e-07,
      "loss": 0.4396,
      "step": 349410
    },
    {
      "epoch": 143.03,
      "grad_norm": 1.8063749074935913,
      "learning_rate": 9.652173026275743e-07,
      "loss": 0.4416,
      "step": 349420
    },
    {
      "epoch": 143.03,
      "grad_norm": 1.9531854391098022,
      "learning_rate": 9.650890166046245e-07,
      "loss": 0.4402,
      "step": 349430
    },
    {
      "epoch": 143.04,
      "grad_norm": 2.054774522781372,
      "learning_rate": 9.649607368971072e-07,
      "loss": 0.435,
      "step": 349440
    },
    {
      "epoch": 143.04,
      "grad_norm": 2.4300334453582764,
      "learning_rate": 9.648324635056096e-07,
      "loss": 0.464,
      "step": 349450
    },
    {
      "epoch": 143.05,
      "grad_norm": 2.2448248863220215,
      "learning_rate": 9.6470419643072e-07,
      "loss": 0.4664,
      "step": 349460
    },
    {
      "epoch": 143.05,
      "grad_norm": 1.8621702194213867,
      "learning_rate": 9.645759356730258e-07,
      "loss": 0.4727,
      "step": 349470
    },
    {
      "epoch": 143.05,
      "grad_norm": 1.6109840869903564,
      "learning_rate": 9.644476812331144e-07,
      "loss": 0.4482,
      "step": 349480
    },
    {
      "epoch": 143.06,
      "grad_norm": 2.069821834564209,
      "learning_rate": 9.643194331115732e-07,
      "loss": 0.4402,
      "step": 349490
    },
    {
      "epoch": 143.06,
      "grad_norm": 1.9758353233337402,
      "learning_rate": 9.641911913089889e-07,
      "loss": 0.4322,
      "step": 349500
    },
    {
      "epoch": 143.07,
      "grad_norm": 1.7798734903335571,
      "learning_rate": 9.640629558259505e-07,
      "loss": 0.4358,
      "step": 349510
    },
    {
      "epoch": 143.07,
      "grad_norm": 2.234295129776001,
      "learning_rate": 9.63934726663045e-07,
      "loss": 0.4498,
      "step": 349520
    },
    {
      "epoch": 143.07,
      "grad_norm": 1.9185986518859863,
      "learning_rate": 9.638065038208596e-07,
      "loss": 0.4449,
      "step": 349530
    },
    {
      "epoch": 143.08,
      "grad_norm": 2.184892416000366,
      "learning_rate": 9.636782872999815e-07,
      "loss": 0.4362,
      "step": 349540
    },
    {
      "epoch": 143.08,
      "grad_norm": 2.18725323677063,
      "learning_rate": 9.635500771009994e-07,
      "loss": 0.4432,
      "step": 349550
    },
    {
      "epoch": 143.09,
      "grad_norm": 1.6563925743103027,
      "learning_rate": 9.634218732244982e-07,
      "loss": 0.4505,
      "step": 349560
    },
    {
      "epoch": 143.09,
      "grad_norm": 2.011550188064575,
      "learning_rate": 9.632936756710665e-07,
      "loss": 0.4603,
      "step": 349570
    },
    {
      "epoch": 143.09,
      "grad_norm": 1.9960615634918213,
      "learning_rate": 9.631654844412913e-07,
      "loss": 0.44,
      "step": 349580
    },
    {
      "epoch": 143.1,
      "grad_norm": 2.1079394817352295,
      "learning_rate": 9.6303729953576e-07,
      "loss": 0.4579,
      "step": 349590
    },
    {
      "epoch": 143.1,
      "grad_norm": 1.6428089141845703,
      "learning_rate": 9.629091209550597e-07,
      "loss": 0.4328,
      "step": 349600
    },
    {
      "epoch": 143.11,
      "grad_norm": 1.9291876554489136,
      "learning_rate": 9.627809486997778e-07,
      "loss": 0.4368,
      "step": 349610
    },
    {
      "epoch": 143.11,
      "grad_norm": 1.8564282655715942,
      "learning_rate": 9.626527827705013e-07,
      "loss": 0.4365,
      "step": 349620
    },
    {
      "epoch": 143.12,
      "grad_norm": 1.913425087928772,
      "learning_rate": 9.62524623167817e-07,
      "loss": 0.4445,
      "step": 349630
    },
    {
      "epoch": 143.12,
      "grad_norm": 2.1812350749969482,
      "learning_rate": 9.623964698923125e-07,
      "loss": 0.4556,
      "step": 349640
    },
    {
      "epoch": 143.12,
      "grad_norm": 1.8621536493301392,
      "learning_rate": 9.622683229445745e-07,
      "loss": 0.4573,
      "step": 349650
    },
    {
      "epoch": 143.13,
      "grad_norm": 2.11798357963562,
      "learning_rate": 9.6214018232519e-07,
      "loss": 0.4527,
      "step": 349660
    },
    {
      "epoch": 143.13,
      "grad_norm": 2.0973031520843506,
      "learning_rate": 9.62012048034746e-07,
      "loss": 0.4472,
      "step": 349670
    },
    {
      "epoch": 143.14,
      "grad_norm": 1.9934417009353638,
      "learning_rate": 9.618839200738299e-07,
      "loss": 0.4435,
      "step": 349680
    },
    {
      "epoch": 143.14,
      "grad_norm": 2.0249574184417725,
      "learning_rate": 9.61755798443028e-07,
      "loss": 0.4372,
      "step": 349690
    },
    {
      "epoch": 143.14,
      "grad_norm": 1.8273992538452148,
      "learning_rate": 9.616276831429279e-07,
      "loss": 0.4314,
      "step": 349700
    },
    {
      "epoch": 143.15,
      "grad_norm": 3.283411741256714,
      "learning_rate": 9.614995741741158e-07,
      "loss": 0.4433,
      "step": 349710
    },
    {
      "epoch": 143.15,
      "grad_norm": 2.140362024307251,
      "learning_rate": 9.61371471537179e-07,
      "loss": 0.4448,
      "step": 349720
    },
    {
      "epoch": 143.16,
      "grad_norm": 1.993369221687317,
      "learning_rate": 9.61243375232704e-07,
      "loss": 0.4499,
      "step": 349730
    },
    {
      "epoch": 143.16,
      "grad_norm": 2.1635730266571045,
      "learning_rate": 9.611152852612778e-07,
      "loss": 0.4647,
      "step": 349740
    },
    {
      "epoch": 143.16,
      "grad_norm": 2.29941463470459,
      "learning_rate": 9.609872016234872e-07,
      "loss": 0.4658,
      "step": 349750
    },
    {
      "epoch": 143.17,
      "grad_norm": 2.5098507404327393,
      "learning_rate": 9.608591243199186e-07,
      "loss": 0.444,
      "step": 349760
    },
    {
      "epoch": 143.17,
      "grad_norm": 2.0352604389190674,
      "learning_rate": 9.607310533511592e-07,
      "loss": 0.4527,
      "step": 349770
    },
    {
      "epoch": 143.18,
      "grad_norm": 1.9190804958343506,
      "learning_rate": 9.60602988717795e-07,
      "loss": 0.4548,
      "step": 349780
    },
    {
      "epoch": 143.18,
      "grad_norm": 2.091369867324829,
      "learning_rate": 9.604749304204132e-07,
      "loss": 0.4365,
      "step": 349790
    },
    {
      "epoch": 143.18,
      "grad_norm": 1.7117973566055298,
      "learning_rate": 9.603468784596004e-07,
      "loss": 0.4454,
      "step": 349800
    },
    {
      "epoch": 143.19,
      "grad_norm": 1.6682430505752563,
      "learning_rate": 9.60218832835943e-07,
      "loss": 0.4505,
      "step": 349810
    },
    {
      "epoch": 143.19,
      "grad_norm": 1.9610369205474854,
      "learning_rate": 9.600907935500275e-07,
      "loss": 0.4693,
      "step": 349820
    },
    {
      "epoch": 143.2,
      "grad_norm": 2.2580442428588867,
      "learning_rate": 9.599627606024409e-07,
      "loss": 0.4385,
      "step": 349830
    },
    {
      "epoch": 143.2,
      "grad_norm": 2.0589540004730225,
      "learning_rate": 9.598347339937691e-07,
      "loss": 0.4483,
      "step": 349840
    },
    {
      "epoch": 143.21,
      "grad_norm": 1.9870673418045044,
      "learning_rate": 9.597067137245988e-07,
      "loss": 0.4404,
      "step": 349850
    },
    {
      "epoch": 143.21,
      "grad_norm": 2.974027395248413,
      "learning_rate": 9.595786997955173e-07,
      "loss": 0.4254,
      "step": 349860
    },
    {
      "epoch": 143.21,
      "grad_norm": 2.0571088790893555,
      "learning_rate": 9.594506922071092e-07,
      "loss": 0.4403,
      "step": 349870
    },
    {
      "epoch": 143.22,
      "grad_norm": 2.0271193981170654,
      "learning_rate": 9.593226909599613e-07,
      "loss": 0.4435,
      "step": 349880
    },
    {
      "epoch": 143.22,
      "grad_norm": 2.063173770904541,
      "learning_rate": 9.59194696054661e-07,
      "loss": 0.4677,
      "step": 349890
    },
    {
      "epoch": 143.23,
      "grad_norm": 2.5155959129333496,
      "learning_rate": 9.590667074917944e-07,
      "loss": 0.4295,
      "step": 349900
    },
    {
      "epoch": 143.23,
      "grad_norm": 2.329129695892334,
      "learning_rate": 9.589387252719474e-07,
      "loss": 0.4638,
      "step": 349910
    },
    {
      "epoch": 143.23,
      "grad_norm": 2.115978956222534,
      "learning_rate": 9.588107493957062e-07,
      "loss": 0.4652,
      "step": 349920
    },
    {
      "epoch": 143.24,
      "grad_norm": 1.838451862335205,
      "learning_rate": 9.586827798636576e-07,
      "loss": 0.4566,
      "step": 349930
    },
    {
      "epoch": 143.24,
      "grad_norm": 1.9970991611480713,
      "learning_rate": 9.585548166763874e-07,
      "loss": 0.4487,
      "step": 349940
    },
    {
      "epoch": 143.25,
      "grad_norm": 2.183809757232666,
      "learning_rate": 9.584268598344815e-07,
      "loss": 0.446,
      "step": 349950
    },
    {
      "epoch": 143.25,
      "grad_norm": 2.1598854064941406,
      "learning_rate": 9.582989093385264e-07,
      "loss": 0.4568,
      "step": 349960
    },
    {
      "epoch": 143.25,
      "grad_norm": 4.274662494659424,
      "learning_rate": 9.581709651891084e-07,
      "loss": 0.4532,
      "step": 349970
    },
    {
      "epoch": 143.26,
      "grad_norm": 2.341062307357788,
      "learning_rate": 9.580430273868131e-07,
      "loss": 0.4527,
      "step": 349980
    },
    {
      "epoch": 143.26,
      "grad_norm": 2.028096914291382,
      "learning_rate": 9.57915095932228e-07,
      "loss": 0.4803,
      "step": 349990
    },
    {
      "epoch": 143.27,
      "grad_norm": 2.140760898590088,
      "learning_rate": 9.57787170825937e-07,
      "loss": 0.4503,
      "step": 350000
    },
    {
      "epoch": 143.27,
      "grad_norm": 2.0591561794281006,
      "learning_rate": 9.57659252068527e-07,
      "loss": 0.4525,
      "step": 350010
    },
    {
      "epoch": 143.27,
      "grad_norm": 2.124927282333374,
      "learning_rate": 9.575313396605843e-07,
      "loss": 0.4738,
      "step": 350020
    },
    {
      "epoch": 143.28,
      "grad_norm": 1.7177155017852783,
      "learning_rate": 9.574034336026942e-07,
      "loss": 0.4592,
      "step": 350030
    },
    {
      "epoch": 143.28,
      "grad_norm": 2.461658477783203,
      "learning_rate": 9.57275533895443e-07,
      "loss": 0.4709,
      "step": 350040
    },
    {
      "epoch": 143.29,
      "grad_norm": 2.1394460201263428,
      "learning_rate": 9.57147640539417e-07,
      "loss": 0.4628,
      "step": 350050
    },
    {
      "epoch": 143.29,
      "grad_norm": 2.2663373947143555,
      "learning_rate": 9.570197535352015e-07,
      "loss": 0.4472,
      "step": 350060
    },
    {
      "epoch": 143.3,
      "grad_norm": 2.2639241218566895,
      "learning_rate": 9.568918728833817e-07,
      "loss": 0.4379,
      "step": 350070
    },
    {
      "epoch": 143.3,
      "grad_norm": 1.9300363063812256,
      "learning_rate": 9.56763998584545e-07,
      "loss": 0.4545,
      "step": 350080
    },
    {
      "epoch": 143.3,
      "grad_norm": 2.4533448219299316,
      "learning_rate": 9.566361306392762e-07,
      "loss": 0.4602,
      "step": 350090
    },
    {
      "epoch": 143.31,
      "grad_norm": 1.860477328300476,
      "learning_rate": 9.565082690481616e-07,
      "loss": 0.4509,
      "step": 350100
    },
    {
      "epoch": 143.31,
      "grad_norm": 2.4030275344848633,
      "learning_rate": 9.56380413811786e-07,
      "loss": 0.4537,
      "step": 350110
    },
    {
      "epoch": 143.32,
      "grad_norm": 2.3038456439971924,
      "learning_rate": 9.562525649307363e-07,
      "loss": 0.4409,
      "step": 350120
    },
    {
      "epoch": 143.32,
      "grad_norm": 2.3355166912078857,
      "learning_rate": 9.561247224055979e-07,
      "loss": 0.4588,
      "step": 350130
    },
    {
      "epoch": 143.32,
      "grad_norm": 1.7304097414016724,
      "learning_rate": 9.55996886236955e-07,
      "loss": 0.4551,
      "step": 350140
    },
    {
      "epoch": 143.33,
      "grad_norm": 1.7989505529403687,
      "learning_rate": 9.558690564253946e-07,
      "loss": 0.4571,
      "step": 350150
    },
    {
      "epoch": 143.33,
      "grad_norm": 1.9502880573272705,
      "learning_rate": 9.557412329715016e-07,
      "loss": 0.4461,
      "step": 350160
    },
    {
      "epoch": 143.34,
      "grad_norm": 2.185299873352051,
      "learning_rate": 9.556134158758618e-07,
      "loss": 0.4505,
      "step": 350170
    },
    {
      "epoch": 143.34,
      "grad_norm": 2.2483794689178467,
      "learning_rate": 9.554856051390606e-07,
      "loss": 0.4467,
      "step": 350180
    },
    {
      "epoch": 143.34,
      "grad_norm": 2.0321009159088135,
      "learning_rate": 9.553578007616838e-07,
      "loss": 0.4542,
      "step": 350190
    },
    {
      "epoch": 143.35,
      "grad_norm": 2.533933401107788,
      "learning_rate": 9.552300027443166e-07,
      "loss": 0.4452,
      "step": 350200
    },
    {
      "epoch": 143.35,
      "grad_norm": 1.714327096939087,
      "learning_rate": 9.551022110875443e-07,
      "loss": 0.459,
      "step": 350210
    },
    {
      "epoch": 143.36,
      "grad_norm": 2.127124786376953,
      "learning_rate": 9.549744257919528e-07,
      "loss": 0.4279,
      "step": 350220
    },
    {
      "epoch": 143.36,
      "grad_norm": 2.0948939323425293,
      "learning_rate": 9.54846646858127e-07,
      "loss": 0.444,
      "step": 350230
    },
    {
      "epoch": 143.36,
      "grad_norm": 2.036707878112793,
      "learning_rate": 9.547188742866523e-07,
      "loss": 0.476,
      "step": 350240
    },
    {
      "epoch": 143.37,
      "grad_norm": 1.9223133325576782,
      "learning_rate": 9.54591108078114e-07,
      "loss": 0.4324,
      "step": 350250
    },
    {
      "epoch": 143.37,
      "grad_norm": 2.1660959720611572,
      "learning_rate": 9.544633482330966e-07,
      "loss": 0.4569,
      "step": 350260
    },
    {
      "epoch": 143.38,
      "grad_norm": 2.0989794731140137,
      "learning_rate": 9.54335594752188e-07,
      "loss": 0.4715,
      "step": 350270
    },
    {
      "epoch": 143.38,
      "grad_norm": 2.297762632369995,
      "learning_rate": 9.542078476359706e-07,
      "loss": 0.4474,
      "step": 350280
    },
    {
      "epoch": 143.39,
      "grad_norm": 2.187014579772949,
      "learning_rate": 9.54080106885031e-07,
      "loss": 0.4334,
      "step": 350290
    },
    {
      "epoch": 143.39,
      "grad_norm": 2.102877616882324,
      "learning_rate": 9.539523724999537e-07,
      "loss": 0.4381,
      "step": 350300
    },
    {
      "epoch": 143.39,
      "grad_norm": 1.8942523002624512,
      "learning_rate": 9.53824644481324e-07,
      "loss": 0.4429,
      "step": 350310
    },
    {
      "epoch": 143.4,
      "grad_norm": 2.2316153049468994,
      "learning_rate": 9.536969228297274e-07,
      "loss": 0.4416,
      "step": 350320
    },
    {
      "epoch": 143.4,
      "grad_norm": 1.8707114458084106,
      "learning_rate": 9.535692075457484e-07,
      "loss": 0.4596,
      "step": 350330
    },
    {
      "epoch": 143.41,
      "grad_norm": 1.9491682052612305,
      "learning_rate": 9.534414986299722e-07,
      "loss": 0.4737,
      "step": 350340
    },
    {
      "epoch": 143.41,
      "grad_norm": 2.072728395462036,
      "learning_rate": 9.533137960829841e-07,
      "loss": 0.4567,
      "step": 350350
    },
    {
      "epoch": 143.41,
      "grad_norm": 2.1624560356140137,
      "learning_rate": 9.531860999053689e-07,
      "loss": 0.4746,
      "step": 350360
    },
    {
      "epoch": 143.42,
      "grad_norm": 2.210536003112793,
      "learning_rate": 9.530584100977114e-07,
      "loss": 0.4513,
      "step": 350370
    },
    {
      "epoch": 143.42,
      "grad_norm": 2.4183919429779053,
      "learning_rate": 9.52930726660597e-07,
      "loss": 0.4414,
      "step": 350380
    },
    {
      "epoch": 143.43,
      "grad_norm": 1.868918538093567,
      "learning_rate": 9.528030495946103e-07,
      "loss": 0.4504,
      "step": 350390
    },
    {
      "epoch": 143.43,
      "grad_norm": 2.046736478805542,
      "learning_rate": 9.52675378900336e-07,
      "loss": 0.4524,
      "step": 350400
    },
    {
      "epoch": 143.43,
      "grad_norm": 2.310225009918213,
      "learning_rate": 9.525477145783591e-07,
      "loss": 0.4455,
      "step": 350410
    },
    {
      "epoch": 143.44,
      "grad_norm": 2.4733643531799316,
      "learning_rate": 9.524200566292644e-07,
      "loss": 0.4537,
      "step": 350420
    },
    {
      "epoch": 143.44,
      "grad_norm": 2.2075092792510986,
      "learning_rate": 9.522924050536368e-07,
      "loss": 0.4382,
      "step": 350430
    },
    {
      "epoch": 143.45,
      "grad_norm": 2.0813584327697754,
      "learning_rate": 9.521647598520618e-07,
      "loss": 0.4381,
      "step": 350440
    },
    {
      "epoch": 143.45,
      "grad_norm": 1.903384804725647,
      "learning_rate": 9.520371210251214e-07,
      "loss": 0.4569,
      "step": 350450
    },
    {
      "epoch": 143.45,
      "grad_norm": 1.9769654273986816,
      "learning_rate": 9.519094885734033e-07,
      "loss": 0.4443,
      "step": 350460
    },
    {
      "epoch": 143.46,
      "grad_norm": 2.013918161392212,
      "learning_rate": 9.517818624974908e-07,
      "loss": 0.4525,
      "step": 350470
    },
    {
      "epoch": 143.46,
      "grad_norm": 1.6436715126037598,
      "learning_rate": 9.516542427979689e-07,
      "loss": 0.455,
      "step": 350480
    },
    {
      "epoch": 143.47,
      "grad_norm": 2.318626880645752,
      "learning_rate": 9.51526629475422e-07,
      "loss": 0.4494,
      "step": 350490
    },
    {
      "epoch": 143.47,
      "grad_norm": 2.1731326580047607,
      "learning_rate": 9.513990225304348e-07,
      "loss": 0.4488,
      "step": 350500
    },
    {
      "epoch": 143.48,
      "grad_norm": 2.193690299987793,
      "learning_rate": 9.512714219635916e-07,
      "loss": 0.4553,
      "step": 350510
    },
    {
      "epoch": 143.48,
      "grad_norm": 2.213714361190796,
      "learning_rate": 9.511438277754772e-07,
      "loss": 0.4463,
      "step": 350520
    },
    {
      "epoch": 143.48,
      "grad_norm": 1.8955610990524292,
      "learning_rate": 9.51016239966676e-07,
      "loss": 0.4348,
      "step": 350530
    },
    {
      "epoch": 143.49,
      "grad_norm": 1.8670282363891602,
      "learning_rate": 9.508886585377726e-07,
      "loss": 0.4679,
      "step": 350540
    },
    {
      "epoch": 143.49,
      "grad_norm": 1.9568719863891602,
      "learning_rate": 9.507610834893512e-07,
      "loss": 0.4555,
      "step": 350550
    },
    {
      "epoch": 143.5,
      "grad_norm": 2.538219928741455,
      "learning_rate": 9.506335148219963e-07,
      "loss": 0.455,
      "step": 350560
    },
    {
      "epoch": 143.5,
      "grad_norm": 2.1444091796875,
      "learning_rate": 9.505059525362923e-07,
      "loss": 0.425,
      "step": 350570
    },
    {
      "epoch": 143.5,
      "grad_norm": 2.197526216506958,
      "learning_rate": 9.503783966328245e-07,
      "loss": 0.4355,
      "step": 350580
    },
    {
      "epoch": 143.51,
      "grad_norm": 1.9846088886260986,
      "learning_rate": 9.502508471121751e-07,
      "loss": 0.4498,
      "step": 350590
    },
    {
      "epoch": 143.51,
      "grad_norm": 2.37267804145813,
      "learning_rate": 9.501233039749296e-07,
      "loss": 0.4427,
      "step": 350600
    },
    {
      "epoch": 143.52,
      "grad_norm": 2.0457639694213867,
      "learning_rate": 9.499957672216723e-07,
      "loss": 0.4538,
      "step": 350610
    },
    {
      "epoch": 143.52,
      "grad_norm": 2.1241233348846436,
      "learning_rate": 9.498682368529871e-07,
      "loss": 0.4579,
      "step": 350620
    },
    {
      "epoch": 143.52,
      "grad_norm": 1.626554250717163,
      "learning_rate": 9.497407128694584e-07,
      "loss": 0.4432,
      "step": 350630
    },
    {
      "epoch": 143.53,
      "grad_norm": 2.435349941253662,
      "learning_rate": 9.496131952716695e-07,
      "loss": 0.4689,
      "step": 350640
    },
    {
      "epoch": 143.53,
      "grad_norm": 1.7206491231918335,
      "learning_rate": 9.494856840602065e-07,
      "loss": 0.4434,
      "step": 350650
    },
    {
      "epoch": 143.54,
      "grad_norm": 2.088226079940796,
      "learning_rate": 9.493581792356522e-07,
      "loss": 0.4692,
      "step": 350660
    },
    {
      "epoch": 143.54,
      "grad_norm": 1.8793041706085205,
      "learning_rate": 9.49230680798591e-07,
      "loss": 0.437,
      "step": 350670
    },
    {
      "epoch": 143.54,
      "grad_norm": 1.7205983400344849,
      "learning_rate": 9.491031887496065e-07,
      "loss": 0.4636,
      "step": 350680
    },
    {
      "epoch": 143.55,
      "grad_norm": 2.2757506370544434,
      "learning_rate": 9.489757030892834e-07,
      "loss": 0.4589,
      "step": 350690
    },
    {
      "epoch": 143.55,
      "grad_norm": 1.800246000289917,
      "learning_rate": 9.488482238182051e-07,
      "loss": 0.4572,
      "step": 350700
    },
    {
      "epoch": 143.56,
      "grad_norm": 1.9245268106460571,
      "learning_rate": 9.487207509369568e-07,
      "loss": 0.451,
      "step": 350710
    },
    {
      "epoch": 143.56,
      "grad_norm": 2.0180552005767822,
      "learning_rate": 9.485932844461207e-07,
      "loss": 0.4419,
      "step": 350720
    },
    {
      "epoch": 143.57,
      "grad_norm": 2.062983751296997,
      "learning_rate": 9.484658243462813e-07,
      "loss": 0.4333,
      "step": 350730
    },
    {
      "epoch": 143.57,
      "grad_norm": 1.93019700050354,
      "learning_rate": 9.483383706380227e-07,
      "loss": 0.4509,
      "step": 350740
    },
    {
      "epoch": 143.57,
      "grad_norm": 1.7145559787750244,
      "learning_rate": 9.482109233219286e-07,
      "loss": 0.4633,
      "step": 350750
    },
    {
      "epoch": 143.58,
      "grad_norm": 3.207258462905884,
      "learning_rate": 9.480834823985829e-07,
      "loss": 0.4459,
      "step": 350760
    },
    {
      "epoch": 143.58,
      "grad_norm": 1.9528257846832275,
      "learning_rate": 9.479560478685694e-07,
      "loss": 0.4599,
      "step": 350770
    },
    {
      "epoch": 143.59,
      "grad_norm": 1.5893036127090454,
      "learning_rate": 9.478286197324716e-07,
      "loss": 0.4584,
      "step": 350780
    },
    {
      "epoch": 143.59,
      "grad_norm": 2.075249671936035,
      "learning_rate": 9.477011979908737e-07,
      "loss": 0.4637,
      "step": 350790
    },
    {
      "epoch": 143.59,
      "grad_norm": 1.7577683925628662,
      "learning_rate": 9.475737826443593e-07,
      "loss": 0.4769,
      "step": 350800
    },
    {
      "epoch": 143.6,
      "grad_norm": 2.3022563457489014,
      "learning_rate": 9.474463736935115e-07,
      "loss": 0.4501,
      "step": 350810
    },
    {
      "epoch": 143.6,
      "grad_norm": 1.6547476053237915,
      "learning_rate": 9.473189711389148e-07,
      "loss": 0.4565,
      "step": 350820
    },
    {
      "epoch": 143.61,
      "grad_norm": 2.0370147228240967,
      "learning_rate": 9.471915749811514e-07,
      "loss": 0.4479,
      "step": 350830
    },
    {
      "epoch": 143.61,
      "grad_norm": 2.5820822715759277,
      "learning_rate": 9.470641852208066e-07,
      "loss": 0.4562,
      "step": 350840
    },
    {
      "epoch": 143.61,
      "grad_norm": 1.940312147140503,
      "learning_rate": 9.469368018584642e-07,
      "loss": 0.4565,
      "step": 350850
    },
    {
      "epoch": 143.62,
      "grad_norm": 1.9786791801452637,
      "learning_rate": 9.468094248947057e-07,
      "loss": 0.4427,
      "step": 350860
    },
    {
      "epoch": 143.62,
      "grad_norm": 2.2903671264648438,
      "learning_rate": 9.466820543301157e-07,
      "loss": 0.441,
      "step": 350870
    },
    {
      "epoch": 143.63,
      "grad_norm": 2.087878942489624,
      "learning_rate": 9.465546901652776e-07,
      "loss": 0.4694,
      "step": 350880
    },
    {
      "epoch": 143.63,
      "grad_norm": 1.6264123916625977,
      "learning_rate": 9.464273324007749e-07,
      "loss": 0.4315,
      "step": 350890
    },
    {
      "epoch": 143.63,
      "grad_norm": 2.1685736179351807,
      "learning_rate": 9.462999810371909e-07,
      "loss": 0.4382,
      "step": 350900
    },
    {
      "epoch": 143.64,
      "grad_norm": 2.0100603103637695,
      "learning_rate": 9.461726360751089e-07,
      "loss": 0.4617,
      "step": 350910
    },
    {
      "epoch": 143.64,
      "grad_norm": 2.2157089710235596,
      "learning_rate": 9.460452975151125e-07,
      "loss": 0.47,
      "step": 350920
    },
    {
      "epoch": 143.65,
      "grad_norm": 1.853658676147461,
      "learning_rate": 9.459179653577848e-07,
      "loss": 0.4621,
      "step": 350930
    },
    {
      "epoch": 143.65,
      "grad_norm": 1.8084149360656738,
      "learning_rate": 9.457906396037092e-07,
      "loss": 0.4352,
      "step": 350940
    },
    {
      "epoch": 143.66,
      "grad_norm": 2.2579033374786377,
      "learning_rate": 9.456633202534689e-07,
      "loss": 0.4326,
      "step": 350950
    },
    {
      "epoch": 143.66,
      "grad_norm": 2.0381946563720703,
      "learning_rate": 9.455360073076473e-07,
      "loss": 0.4464,
      "step": 350960
    },
    {
      "epoch": 143.66,
      "grad_norm": 1.7258458137512207,
      "learning_rate": 9.454087007668275e-07,
      "loss": 0.4604,
      "step": 350970
    },
    {
      "epoch": 143.67,
      "grad_norm": 2.0199317932128906,
      "learning_rate": 9.452814006315923e-07,
      "loss": 0.4677,
      "step": 350980
    },
    {
      "epoch": 143.67,
      "grad_norm": 2.365626573562622,
      "learning_rate": 9.451541069025254e-07,
      "loss": 0.4381,
      "step": 350990
    },
    {
      "epoch": 143.68,
      "grad_norm": 1.949442982673645,
      "learning_rate": 9.450268195802097e-07,
      "loss": 0.4455,
      "step": 351000
    },
    {
      "epoch": 143.68,
      "grad_norm": 2.085193157196045,
      "learning_rate": 9.448995386652283e-07,
      "loss": 0.4678,
      "step": 351010
    },
    {
      "epoch": 143.68,
      "grad_norm": 2.0642473697662354,
      "learning_rate": 9.44772264158164e-07,
      "loss": 0.4457,
      "step": 351020
    },
    {
      "epoch": 143.69,
      "grad_norm": 1.902079463005066,
      "learning_rate": 9.446449960596004e-07,
      "loss": 0.4521,
      "step": 351030
    },
    {
      "epoch": 143.69,
      "grad_norm": 2.1080775260925293,
      "learning_rate": 9.445177343701198e-07,
      "loss": 0.4604,
      "step": 351040
    },
    {
      "epoch": 143.7,
      "grad_norm": 2.574064254760742,
      "learning_rate": 9.443904790903059e-07,
      "loss": 0.4335,
      "step": 351050
    },
    {
      "epoch": 143.7,
      "grad_norm": 2.4661593437194824,
      "learning_rate": 9.442632302207407e-07,
      "loss": 0.4743,
      "step": 351060
    },
    {
      "epoch": 143.7,
      "grad_norm": 1.880002737045288,
      "learning_rate": 9.44135987762008e-07,
      "loss": 0.4522,
      "step": 351070
    },
    {
      "epoch": 143.71,
      "grad_norm": 1.7788597345352173,
      "learning_rate": 9.440087517146904e-07,
      "loss": 0.4615,
      "step": 351080
    },
    {
      "epoch": 143.71,
      "grad_norm": 1.8329434394836426,
      "learning_rate": 9.438815220793707e-07,
      "loss": 0.4686,
      "step": 351090
    },
    {
      "epoch": 143.72,
      "grad_norm": 2.144768238067627,
      "learning_rate": 9.437542988566314e-07,
      "loss": 0.4496,
      "step": 351100
    },
    {
      "epoch": 143.72,
      "grad_norm": 2.1980626583099365,
      "learning_rate": 9.436270820470559e-07,
      "loss": 0.4584,
      "step": 351110
    },
    {
      "epoch": 143.72,
      "grad_norm": 2.044711112976074,
      "learning_rate": 9.434998716512265e-07,
      "loss": 0.4469,
      "step": 351120
    },
    {
      "epoch": 143.73,
      "grad_norm": 1.8649637699127197,
      "learning_rate": 9.43372667669726e-07,
      "loss": 0.4535,
      "step": 351130
    },
    {
      "epoch": 143.73,
      "grad_norm": 2.084048271179199,
      "learning_rate": 9.432454701031374e-07,
      "loss": 0.418,
      "step": 351140
    },
    {
      "epoch": 143.74,
      "grad_norm": 1.7381807565689087,
      "learning_rate": 9.431182789520428e-07,
      "loss": 0.4701,
      "step": 351150
    },
    {
      "epoch": 143.74,
      "grad_norm": 1.610509991645813,
      "learning_rate": 9.429910942170265e-07,
      "loss": 0.4583,
      "step": 351160
    },
    {
      "epoch": 143.75,
      "grad_norm": 2.512763500213623,
      "learning_rate": 9.428639158986688e-07,
      "loss": 0.4427,
      "step": 351170
    },
    {
      "epoch": 143.75,
      "grad_norm": 2.0581467151641846,
      "learning_rate": 9.42736743997553e-07,
      "loss": 0.4415,
      "step": 351180
    },
    {
      "epoch": 143.75,
      "grad_norm": 2.0505568981170654,
      "learning_rate": 9.426095785142624e-07,
      "loss": 0.4459,
      "step": 351190
    },
    {
      "epoch": 143.76,
      "grad_norm": 2.070030927658081,
      "learning_rate": 9.424824194493779e-07,
      "loss": 0.4459,
      "step": 351200
    },
    {
      "epoch": 143.76,
      "grad_norm": 1.8888208866119385,
      "learning_rate": 9.423552668034842e-07,
      "loss": 0.4254,
      "step": 351210
    },
    {
      "epoch": 143.77,
      "grad_norm": 1.9660693407058716,
      "learning_rate": 9.422281205771625e-07,
      "loss": 0.4574,
      "step": 351220
    },
    {
      "epoch": 143.77,
      "grad_norm": 1.819281816482544,
      "learning_rate": 9.421009807709959e-07,
      "loss": 0.4481,
      "step": 351230
    },
    {
      "epoch": 143.77,
      "grad_norm": 2.295633316040039,
      "learning_rate": 9.41973847385566e-07,
      "loss": 0.4525,
      "step": 351240
    },
    {
      "epoch": 143.78,
      "grad_norm": 1.820194959640503,
      "learning_rate": 9.418467204214557e-07,
      "loss": 0.4403,
      "step": 351250
    },
    {
      "epoch": 143.78,
      "grad_norm": 1.9617046117782593,
      "learning_rate": 9.417195998792472e-07,
      "loss": 0.4635,
      "step": 351260
    },
    {
      "epoch": 143.79,
      "grad_norm": 2.464768886566162,
      "learning_rate": 9.415924857595229e-07,
      "loss": 0.4546,
      "step": 351270
    },
    {
      "epoch": 143.79,
      "grad_norm": 1.736964225769043,
      "learning_rate": 9.41465378062865e-07,
      "loss": 0.4396,
      "step": 351280
    },
    {
      "epoch": 143.79,
      "grad_norm": 2.320011854171753,
      "learning_rate": 9.413382767898557e-07,
      "loss": 0.4439,
      "step": 351290
    },
    {
      "epoch": 143.8,
      "grad_norm": 2.9210338592529297,
      "learning_rate": 9.412111819410784e-07,
      "loss": 0.4496,
      "step": 351300
    },
    {
      "epoch": 143.8,
      "grad_norm": 2.031334638595581,
      "learning_rate": 9.410840935171131e-07,
      "loss": 0.4583,
      "step": 351310
    },
    {
      "epoch": 143.81,
      "grad_norm": 1.8424599170684814,
      "learning_rate": 9.409570115185432e-07,
      "loss": 0.4546,
      "step": 351320
    },
    {
      "epoch": 143.81,
      "grad_norm": 2.020777463912964,
      "learning_rate": 9.408299359459506e-07,
      "loss": 0.4289,
      "step": 351330
    },
    {
      "epoch": 143.81,
      "grad_norm": 2.4768099784851074,
      "learning_rate": 9.407028667999178e-07,
      "loss": 0.4561,
      "step": 351340
    },
    {
      "epoch": 143.82,
      "grad_norm": 1.9321452379226685,
      "learning_rate": 9.405758040810264e-07,
      "loss": 0.4554,
      "step": 351350
    },
    {
      "epoch": 143.82,
      "grad_norm": 1.7943426370620728,
      "learning_rate": 9.404487477898586e-07,
      "loss": 0.4672,
      "step": 351360
    },
    {
      "epoch": 143.83,
      "grad_norm": 1.9671746492385864,
      "learning_rate": 9.403216979269968e-07,
      "loss": 0.4356,
      "step": 351370
    },
    {
      "epoch": 143.83,
      "grad_norm": 1.6579842567443848,
      "learning_rate": 9.401946544930226e-07,
      "loss": 0.4431,
      "step": 351380
    },
    {
      "epoch": 143.84,
      "grad_norm": 1.9873065948486328,
      "learning_rate": 9.400676174885172e-07,
      "loss": 0.4626,
      "step": 351390
    },
    {
      "epoch": 143.84,
      "grad_norm": 1.9007954597473145,
      "learning_rate": 9.399405869140643e-07,
      "loss": 0.4621,
      "step": 351400
    },
    {
      "epoch": 143.84,
      "grad_norm": 2.280594825744629,
      "learning_rate": 9.398135627702448e-07,
      "loss": 0.4446,
      "step": 351410
    },
    {
      "epoch": 143.85,
      "grad_norm": 2.5739991664886475,
      "learning_rate": 9.396865450576407e-07,
      "loss": 0.4529,
      "step": 351420
    },
    {
      "epoch": 143.85,
      "grad_norm": 1.9577181339263916,
      "learning_rate": 9.395595337768347e-07,
      "loss": 0.4441,
      "step": 351430
    },
    {
      "epoch": 143.86,
      "grad_norm": 2.6823463439941406,
      "learning_rate": 9.39432528928407e-07,
      "loss": 0.456,
      "step": 351440
    },
    {
      "epoch": 143.86,
      "grad_norm": 2.042872190475464,
      "learning_rate": 9.3930553051294e-07,
      "loss": 0.4467,
      "step": 351450
    },
    {
      "epoch": 143.86,
      "grad_norm": 2.219216823577881,
      "learning_rate": 9.391785385310158e-07,
      "loss": 0.4496,
      "step": 351460
    },
    {
      "epoch": 143.87,
      "grad_norm": 1.674669861793518,
      "learning_rate": 9.390515529832159e-07,
      "loss": 0.4378,
      "step": 351470
    },
    {
      "epoch": 143.87,
      "grad_norm": 1.8319522142410278,
      "learning_rate": 9.389245738701218e-07,
      "loss": 0.4536,
      "step": 351480
    },
    {
      "epoch": 143.88,
      "grad_norm": 1.9220486879348755,
      "learning_rate": 9.387976011923158e-07,
      "loss": 0.4279,
      "step": 351490
    },
    {
      "epoch": 143.88,
      "grad_norm": 1.8359612226486206,
      "learning_rate": 9.386706349503788e-07,
      "loss": 0.4561,
      "step": 351500
    },
    {
      "epoch": 143.88,
      "grad_norm": 2.2832069396972656,
      "learning_rate": 9.385436751448928e-07,
      "loss": 0.4859,
      "step": 351510
    },
    {
      "epoch": 143.89,
      "grad_norm": 1.8644840717315674,
      "learning_rate": 9.384167217764396e-07,
      "loss": 0.4567,
      "step": 351520
    },
    {
      "epoch": 143.89,
      "grad_norm": 2.0810368061065674,
      "learning_rate": 9.382897748456004e-07,
      "loss": 0.4452,
      "step": 351530
    },
    {
      "epoch": 143.9,
      "grad_norm": 2.1599810123443604,
      "learning_rate": 9.381628343529568e-07,
      "loss": 0.4352,
      "step": 351540
    },
    {
      "epoch": 143.9,
      "grad_norm": 1.7478171586990356,
      "learning_rate": 9.380359002990904e-07,
      "loss": 0.4636,
      "step": 351550
    },
    {
      "epoch": 143.91,
      "grad_norm": 2.0636560916900635,
      "learning_rate": 9.379089726845825e-07,
      "loss": 0.4604,
      "step": 351560
    },
    {
      "epoch": 143.91,
      "grad_norm": 2.70296573638916,
      "learning_rate": 9.377820515100146e-07,
      "loss": 0.4505,
      "step": 351570
    },
    {
      "epoch": 143.91,
      "grad_norm": 2.1255574226379395,
      "learning_rate": 9.376551367759683e-07,
      "loss": 0.4493,
      "step": 351580
    },
    {
      "epoch": 143.92,
      "grad_norm": 2.082987070083618,
      "learning_rate": 9.375282284830245e-07,
      "loss": 0.4665,
      "step": 351590
    },
    {
      "epoch": 143.92,
      "grad_norm": 2.2536215782165527,
      "learning_rate": 9.374013266317652e-07,
      "loss": 0.4622,
      "step": 351600
    },
    {
      "epoch": 143.93,
      "grad_norm": 1.660948634147644,
      "learning_rate": 9.37274431222771e-07,
      "loss": 0.4587,
      "step": 351610
    },
    {
      "epoch": 143.93,
      "grad_norm": 2.0463836193084717,
      "learning_rate": 9.37147542256624e-07,
      "loss": 0.4537,
      "step": 351620
    },
    {
      "epoch": 143.93,
      "grad_norm": 2.246873617172241,
      "learning_rate": 9.370206597339047e-07,
      "loss": 0.428,
      "step": 351630
    },
    {
      "epoch": 143.94,
      "grad_norm": 2.5325927734375,
      "learning_rate": 9.368937836551948e-07,
      "loss": 0.4703,
      "step": 351640
    },
    {
      "epoch": 143.94,
      "grad_norm": 1.849779725074768,
      "learning_rate": 9.367669140210752e-07,
      "loss": 0.4428,
      "step": 351650
    },
    {
      "epoch": 143.95,
      "grad_norm": 1.7894283533096313,
      "learning_rate": 9.366400508321272e-07,
      "loss": 0.4558,
      "step": 351660
    },
    {
      "epoch": 143.95,
      "grad_norm": 1.7850018739700317,
      "learning_rate": 9.365131940889323e-07,
      "loss": 0.4683,
      "step": 351670
    },
    {
      "epoch": 143.95,
      "grad_norm": 2.237065553665161,
      "learning_rate": 9.363863437920709e-07,
      "loss": 0.4408,
      "step": 351680
    },
    {
      "epoch": 143.96,
      "grad_norm": 2.098177909851074,
      "learning_rate": 9.362594999421244e-07,
      "loss": 0.4474,
      "step": 351690
    },
    {
      "epoch": 143.96,
      "grad_norm": 1.9210151433944702,
      "learning_rate": 9.361326625396742e-07,
      "loss": 0.4588,
      "step": 351700
    },
    {
      "epoch": 143.97,
      "grad_norm": 2.0347752571105957,
      "learning_rate": 9.360058315853009e-07,
      "loss": 0.4377,
      "step": 351710
    },
    {
      "epoch": 143.97,
      "grad_norm": 1.9843007326126099,
      "learning_rate": 9.358790070795856e-07,
      "loss": 0.4537,
      "step": 351720
    },
    {
      "epoch": 143.97,
      "grad_norm": 2.042264699935913,
      "learning_rate": 9.357521890231094e-07,
      "loss": 0.4434,
      "step": 351730
    },
    {
      "epoch": 143.98,
      "grad_norm": 2.2262349128723145,
      "learning_rate": 9.356253774164537e-07,
      "loss": 0.4547,
      "step": 351740
    },
    {
      "epoch": 143.98,
      "grad_norm": 2.0179383754730225,
      "learning_rate": 9.354985722601982e-07,
      "loss": 0.4408,
      "step": 351750
    },
    {
      "epoch": 143.99,
      "grad_norm": 2.405679702758789,
      "learning_rate": 9.353717735549244e-07,
      "loss": 0.4544,
      "step": 351760
    },
    {
      "epoch": 143.99,
      "grad_norm": 2.438011646270752,
      "learning_rate": 9.352449813012123e-07,
      "loss": 0.4569,
      "step": 351770
    },
    {
      "epoch": 144.0,
      "grad_norm": 1.8824362754821777,
      "learning_rate": 9.351181954996444e-07,
      "loss": 0.4522,
      "step": 351780
    },
    {
      "epoch": 144.0,
      "grad_norm": 1.9266612529754639,
      "learning_rate": 9.349914161508006e-07,
      "loss": 0.454,
      "step": 351790
    },
    {
      "epoch": 144.0,
      "eval_loss": 0.4514060914516449,
      "eval_runtime": 52.2663,
      "eval_samples_per_second": 65.989,
      "eval_steps_per_second": 8.265,
      "step": 351792
    },
    {
      "epoch": 144.0,
      "grad_norm": 2.148789882659912,
      "learning_rate": 9.348646432552615e-07,
      "loss": 0.4723,
      "step": 351800
    },
    {
      "epoch": 144.01,
      "grad_norm": 2.004689931869507,
      "learning_rate": 9.347378768136084e-07,
      "loss": 0.4441,
      "step": 351810
    },
    {
      "epoch": 144.01,
      "grad_norm": 1.9415034055709839,
      "learning_rate": 9.346111168264214e-07,
      "loss": 0.4409,
      "step": 351820
    },
    {
      "epoch": 144.02,
      "grad_norm": 2.019843816757202,
      "learning_rate": 9.344843632942813e-07,
      "loss": 0.452,
      "step": 351830
    },
    {
      "epoch": 144.02,
      "grad_norm": 2.309053897857666,
      "learning_rate": 9.343576162177689e-07,
      "loss": 0.4535,
      "step": 351840
    },
    {
      "epoch": 144.02,
      "grad_norm": 3.3692219257354736,
      "learning_rate": 9.34230875597465e-07,
      "loss": 0.4528,
      "step": 351850
    },
    {
      "epoch": 144.03,
      "grad_norm": 1.9530795812606812,
      "learning_rate": 9.341041414339496e-07,
      "loss": 0.4433,
      "step": 351860
    },
    {
      "epoch": 144.03,
      "grad_norm": 1.800819993019104,
      "learning_rate": 9.339774137278036e-07,
      "loss": 0.4527,
      "step": 351870
    },
    {
      "epoch": 144.04,
      "grad_norm": 3.4229111671447754,
      "learning_rate": 9.338506924796086e-07,
      "loss": 0.4575,
      "step": 351880
    },
    {
      "epoch": 144.04,
      "grad_norm": 2.5247199535369873,
      "learning_rate": 9.337239776899429e-07,
      "loss": 0.4625,
      "step": 351890
    },
    {
      "epoch": 144.04,
      "grad_norm": 1.9673517942428589,
      "learning_rate": 9.335972693593879e-07,
      "loss": 0.432,
      "step": 351900
    },
    {
      "epoch": 144.05,
      "grad_norm": 1.8276760578155518,
      "learning_rate": 9.334705674885243e-07,
      "loss": 0.4477,
      "step": 351910
    },
    {
      "epoch": 144.05,
      "grad_norm": 2.0390100479125977,
      "learning_rate": 9.333438720779325e-07,
      "loss": 0.4532,
      "step": 351920
    },
    {
      "epoch": 144.06,
      "grad_norm": 2.097458839416504,
      "learning_rate": 9.332171831281927e-07,
      "loss": 0.4547,
      "step": 351930
    },
    {
      "epoch": 144.06,
      "grad_norm": 2.066331624984741,
      "learning_rate": 9.330905006398852e-07,
      "loss": 0.4448,
      "step": 351940
    },
    {
      "epoch": 144.06,
      "grad_norm": 1.8677430152893066,
      "learning_rate": 9.329638246135905e-07,
      "loss": 0.4533,
      "step": 351950
    },
    {
      "epoch": 144.07,
      "grad_norm": 2.086984395980835,
      "learning_rate": 9.328371550498879e-07,
      "loss": 0.4163,
      "step": 351960
    },
    {
      "epoch": 144.07,
      "grad_norm": 2.2936761379241943,
      "learning_rate": 9.327104919493594e-07,
      "loss": 0.4355,
      "step": 351970
    },
    {
      "epoch": 144.08,
      "grad_norm": 2.015836477279663,
      "learning_rate": 9.325838353125843e-07,
      "loss": 0.4332,
      "step": 351980
    },
    {
      "epoch": 144.08,
      "grad_norm": 7.498299598693848,
      "learning_rate": 9.324571851401427e-07,
      "loss": 0.4605,
      "step": 351990
    },
    {
      "epoch": 144.09,
      "grad_norm": 1.7955714464187622,
      "learning_rate": 9.323305414326151e-07,
      "loss": 0.4509,
      "step": 352000
    },
    {
      "epoch": 144.09,
      "grad_norm": 2.197174072265625,
      "learning_rate": 9.322039041905825e-07,
      "loss": 0.4481,
      "step": 352010
    },
    {
      "epoch": 144.09,
      "grad_norm": 2.7310521602630615,
      "learning_rate": 9.320772734146229e-07,
      "loss": 0.4377,
      "step": 352020
    },
    {
      "epoch": 144.1,
      "grad_norm": 2.1357147693634033,
      "learning_rate": 9.319506491053173e-07,
      "loss": 0.4575,
      "step": 352030
    },
    {
      "epoch": 144.1,
      "grad_norm": 2.0595905780792236,
      "learning_rate": 9.31824031263246e-07,
      "loss": 0.4285,
      "step": 352040
    },
    {
      "epoch": 144.11,
      "grad_norm": 1.9659582376480103,
      "learning_rate": 9.31697419888989e-07,
      "loss": 0.4509,
      "step": 352050
    },
    {
      "epoch": 144.11,
      "grad_norm": 2.414280652999878,
      "learning_rate": 9.315708149831261e-07,
      "loss": 0.4572,
      "step": 352060
    },
    {
      "epoch": 144.11,
      "grad_norm": 1.960120677947998,
      "learning_rate": 9.314442165462374e-07,
      "loss": 0.451,
      "step": 352070
    },
    {
      "epoch": 144.12,
      "grad_norm": 2.0102782249450684,
      "learning_rate": 9.313176245789026e-07,
      "loss": 0.4513,
      "step": 352080
    },
    {
      "epoch": 144.12,
      "grad_norm": 1.9827507734298706,
      "learning_rate": 9.311910390817019e-07,
      "loss": 0.4467,
      "step": 352090
    },
    {
      "epoch": 144.13,
      "grad_norm": 1.7880464792251587,
      "learning_rate": 9.310644600552152e-07,
      "loss": 0.4305,
      "step": 352100
    },
    {
      "epoch": 144.13,
      "grad_norm": 1.8297929763793945,
      "learning_rate": 9.309378875000218e-07,
      "loss": 0.4539,
      "step": 352110
    },
    {
      "epoch": 144.13,
      "grad_norm": 2.1222078800201416,
      "learning_rate": 9.308113214167022e-07,
      "loss": 0.446,
      "step": 352120
    },
    {
      "epoch": 144.14,
      "grad_norm": 2.0234296321868896,
      "learning_rate": 9.306847618058357e-07,
      "loss": 0.4568,
      "step": 352130
    },
    {
      "epoch": 144.14,
      "grad_norm": 2.1815426349639893,
      "learning_rate": 9.305582086680022e-07,
      "loss": 0.4461,
      "step": 352140
    },
    {
      "epoch": 144.15,
      "grad_norm": 1.9503506422042847,
      "learning_rate": 9.304316620037816e-07,
      "loss": 0.4223,
      "step": 352150
    },
    {
      "epoch": 144.15,
      "grad_norm": 2.041785955429077,
      "learning_rate": 9.303051218137534e-07,
      "loss": 0.4294,
      "step": 352160
    },
    {
      "epoch": 144.15,
      "grad_norm": 2.102916717529297,
      "learning_rate": 9.301785880984973e-07,
      "loss": 0.4478,
      "step": 352170
    },
    {
      "epoch": 144.16,
      "grad_norm": 2.3659777641296387,
      "learning_rate": 9.300520608585929e-07,
      "loss": 0.4776,
      "step": 352180
    },
    {
      "epoch": 144.16,
      "grad_norm": 2.140693426132202,
      "learning_rate": 9.299255400946195e-07,
      "loss": 0.4497,
      "step": 352190
    },
    {
      "epoch": 144.17,
      "grad_norm": 1.7519162893295288,
      "learning_rate": 9.297990258071573e-07,
      "loss": 0.4507,
      "step": 352200
    },
    {
      "epoch": 144.17,
      "grad_norm": 1.574331521987915,
      "learning_rate": 9.296725179967854e-07,
      "loss": 0.4459,
      "step": 352210
    },
    {
      "epoch": 144.18,
      "grad_norm": 2.038605213165283,
      "learning_rate": 9.295460166640838e-07,
      "loss": 0.4506,
      "step": 352220
    },
    {
      "epoch": 144.18,
      "grad_norm": 1.827890396118164,
      "learning_rate": 9.294195218096313e-07,
      "loss": 0.4586,
      "step": 352230
    },
    {
      "epoch": 144.18,
      "grad_norm": 2.1289560794830322,
      "learning_rate": 9.292930334340077e-07,
      "loss": 0.477,
      "step": 352240
    },
    {
      "epoch": 144.19,
      "grad_norm": 1.9754990339279175,
      "learning_rate": 9.291665515377926e-07,
      "loss": 0.4525,
      "step": 352250
    },
    {
      "epoch": 144.19,
      "grad_norm": 1.8796764612197876,
      "learning_rate": 9.29040076121565e-07,
      "loss": 0.45,
      "step": 352260
    },
    {
      "epoch": 144.2,
      "grad_norm": 1.9192389249801636,
      "learning_rate": 9.289136071859044e-07,
      "loss": 0.4442,
      "step": 352270
    },
    {
      "epoch": 144.2,
      "grad_norm": 1.8323590755462646,
      "learning_rate": 9.287871447313904e-07,
      "loss": 0.4579,
      "step": 352280
    },
    {
      "epoch": 144.2,
      "grad_norm": 1.977321743965149,
      "learning_rate": 9.286606887586019e-07,
      "loss": 0.4594,
      "step": 352290
    },
    {
      "epoch": 144.21,
      "grad_norm": 2.153493642807007,
      "learning_rate": 9.285342392681187e-07,
      "loss": 0.4475,
      "step": 352300
    },
    {
      "epoch": 144.21,
      "grad_norm": 3.0315520763397217,
      "learning_rate": 9.284077962605194e-07,
      "loss": 0.4427,
      "step": 352310
    },
    {
      "epoch": 144.22,
      "grad_norm": 2.3084943294525146,
      "learning_rate": 9.282813597363845e-07,
      "loss": 0.4606,
      "step": 352320
    },
    {
      "epoch": 144.22,
      "grad_norm": 1.9784029722213745,
      "learning_rate": 9.281549296962906e-07,
      "loss": 0.4501,
      "step": 352330
    },
    {
      "epoch": 144.22,
      "grad_norm": 1.8803739547729492,
      "learning_rate": 9.280285061408193e-07,
      "loss": 0.4413,
      "step": 352340
    },
    {
      "epoch": 144.23,
      "grad_norm": 2.4293510913848877,
      "learning_rate": 9.279020890705488e-07,
      "loss": 0.4546,
      "step": 352350
    },
    {
      "epoch": 144.23,
      "grad_norm": 2.182431936264038,
      "learning_rate": 9.277756784860584e-07,
      "loss": 0.4561,
      "step": 352360
    },
    {
      "epoch": 144.24,
      "grad_norm": 2.140289306640625,
      "learning_rate": 9.276492743879272e-07,
      "loss": 0.437,
      "step": 352370
    },
    {
      "epoch": 144.24,
      "grad_norm": 2.0034232139587402,
      "learning_rate": 9.27522876776734e-07,
      "loss": 0.4712,
      "step": 352380
    },
    {
      "epoch": 144.24,
      "grad_norm": 2.0357258319854736,
      "learning_rate": 9.27396485653058e-07,
      "loss": 0.4534,
      "step": 352390
    },
    {
      "epoch": 144.25,
      "grad_norm": 2.0978479385375977,
      "learning_rate": 9.272701010174781e-07,
      "loss": 0.4368,
      "step": 352400
    },
    {
      "epoch": 144.25,
      "grad_norm": 2.1715750694274902,
      "learning_rate": 9.271437228705731e-07,
      "loss": 0.4677,
      "step": 352410
    },
    {
      "epoch": 144.26,
      "grad_norm": 2.067840099334717,
      "learning_rate": 9.270173512129221e-07,
      "loss": 0.4713,
      "step": 352420
    },
    {
      "epoch": 144.26,
      "grad_norm": 1.6731473207473755,
      "learning_rate": 9.26890986045104e-07,
      "loss": 0.465,
      "step": 352430
    },
    {
      "epoch": 144.27,
      "grad_norm": 2.0382063388824463,
      "learning_rate": 9.267646273676976e-07,
      "loss": 0.4532,
      "step": 352440
    },
    {
      "epoch": 144.27,
      "grad_norm": 2.160054922103882,
      "learning_rate": 9.266382751812817e-07,
      "loss": 0.446,
      "step": 352450
    },
    {
      "epoch": 144.27,
      "grad_norm": 2.2675559520721436,
      "learning_rate": 9.265119294864362e-07,
      "loss": 0.4658,
      "step": 352460
    },
    {
      "epoch": 144.28,
      "grad_norm": 1.954651951789856,
      "learning_rate": 9.263855902837377e-07,
      "loss": 0.4495,
      "step": 352470
    },
    {
      "epoch": 144.28,
      "grad_norm": 2.3270928859710693,
      "learning_rate": 9.262592575737661e-07,
      "loss": 0.4511,
      "step": 352480
    },
    {
      "epoch": 144.29,
      "grad_norm": 1.9799647331237793,
      "learning_rate": 9.261329313571001e-07,
      "loss": 0.4423,
      "step": 352490
    },
    {
      "epoch": 144.29,
      "grad_norm": 2.1613662242889404,
      "learning_rate": 9.260066116343185e-07,
      "loss": 0.4301,
      "step": 352500
    },
    {
      "epoch": 144.29,
      "grad_norm": 2.139970302581787,
      "learning_rate": 9.258802984059997e-07,
      "loss": 0.447,
      "step": 352510
    },
    {
      "epoch": 144.3,
      "grad_norm": 1.8003644943237305,
      "learning_rate": 9.257539916727216e-07,
      "loss": 0.4589,
      "step": 352520
    },
    {
      "epoch": 144.3,
      "grad_norm": 2.3540308475494385,
      "learning_rate": 9.256276914350643e-07,
      "loss": 0.4569,
      "step": 352530
    },
    {
      "epoch": 144.31,
      "grad_norm": 2.374483346939087,
      "learning_rate": 9.255013976936058e-07,
      "loss": 0.4602,
      "step": 352540
    },
    {
      "epoch": 144.31,
      "grad_norm": 2.1704812049865723,
      "learning_rate": 9.253751104489244e-07,
      "loss": 0.4578,
      "step": 352550
    },
    {
      "epoch": 144.31,
      "grad_norm": 1.7644370794296265,
      "learning_rate": 9.252488297015987e-07,
      "loss": 0.4408,
      "step": 352560
    },
    {
      "epoch": 144.32,
      "grad_norm": 1.8095524311065674,
      "learning_rate": 9.251225554522073e-07,
      "loss": 0.4342,
      "step": 352570
    },
    {
      "epoch": 144.32,
      "grad_norm": 1.9874581098556519,
      "learning_rate": 9.249962877013285e-07,
      "loss": 0.4909,
      "step": 352580
    },
    {
      "epoch": 144.33,
      "grad_norm": 1.987499713897705,
      "learning_rate": 9.248700264495408e-07,
      "loss": 0.4537,
      "step": 352590
    },
    {
      "epoch": 144.33,
      "grad_norm": 2.3379509449005127,
      "learning_rate": 9.247437716974232e-07,
      "loss": 0.4335,
      "step": 352600
    },
    {
      "epoch": 144.33,
      "grad_norm": 2.360158681869507,
      "learning_rate": 9.246175234455525e-07,
      "loss": 0.469,
      "step": 352610
    },
    {
      "epoch": 144.34,
      "grad_norm": 1.866182565689087,
      "learning_rate": 9.244912816945081e-07,
      "loss": 0.4659,
      "step": 352620
    },
    {
      "epoch": 144.34,
      "grad_norm": 1.8169304132461548,
      "learning_rate": 9.243650464448679e-07,
      "loss": 0.449,
      "step": 352630
    },
    {
      "epoch": 144.35,
      "grad_norm": 1.9236292839050293,
      "learning_rate": 9.242388176972106e-07,
      "loss": 0.4197,
      "step": 352640
    },
    {
      "epoch": 144.35,
      "grad_norm": 1.9627070426940918,
      "learning_rate": 9.24112595452114e-07,
      "loss": 0.4587,
      "step": 352650
    },
    {
      "epoch": 144.36,
      "grad_norm": 1.7802904844284058,
      "learning_rate": 9.239863797101563e-07,
      "loss": 0.4689,
      "step": 352660
    },
    {
      "epoch": 144.36,
      "grad_norm": 1.8126119375228882,
      "learning_rate": 9.238601704719164e-07,
      "loss": 0.4659,
      "step": 352670
    },
    {
      "epoch": 144.36,
      "grad_norm": 1.8504730463027954,
      "learning_rate": 9.237339677379715e-07,
      "loss": 0.4471,
      "step": 352680
    },
    {
      "epoch": 144.37,
      "grad_norm": 2.367847204208374,
      "learning_rate": 9.236077715089002e-07,
      "loss": 0.4683,
      "step": 352690
    },
    {
      "epoch": 144.37,
      "grad_norm": 1.7975655794143677,
      "learning_rate": 9.234815817852806e-07,
      "loss": 0.4335,
      "step": 352700
    },
    {
      "epoch": 144.38,
      "grad_norm": 1.9106119871139526,
      "learning_rate": 9.233553985676898e-07,
      "loss": 0.4476,
      "step": 352710
    },
    {
      "epoch": 144.38,
      "grad_norm": 1.7158256769180298,
      "learning_rate": 9.232292218567074e-07,
      "loss": 0.4421,
      "step": 352720
    },
    {
      "epoch": 144.38,
      "grad_norm": 2.047140121459961,
      "learning_rate": 9.231030516529115e-07,
      "loss": 0.4566,
      "step": 352730
    },
    {
      "epoch": 144.39,
      "grad_norm": 2.1830952167510986,
      "learning_rate": 9.229768879568787e-07,
      "loss": 0.446,
      "step": 352740
    },
    {
      "epoch": 144.39,
      "grad_norm": 1.7997105121612549,
      "learning_rate": 9.228507307691871e-07,
      "loss": 0.4341,
      "step": 352750
    },
    {
      "epoch": 144.4,
      "grad_norm": 1.9617456197738647,
      "learning_rate": 9.227245800904154e-07,
      "loss": 0.4475,
      "step": 352760
    },
    {
      "epoch": 144.4,
      "grad_norm": 2.186288595199585,
      "learning_rate": 9.225984359211408e-07,
      "loss": 0.4743,
      "step": 352770
    },
    {
      "epoch": 144.4,
      "grad_norm": 1.9241453409194946,
      "learning_rate": 9.224722982619417e-07,
      "loss": 0.4603,
      "step": 352780
    },
    {
      "epoch": 144.41,
      "grad_norm": 1.794630527496338,
      "learning_rate": 9.223461671133955e-07,
      "loss": 0.435,
      "step": 352790
    },
    {
      "epoch": 144.41,
      "grad_norm": 2.292638063430786,
      "learning_rate": 9.222200424760801e-07,
      "loss": 0.4555,
      "step": 352800
    },
    {
      "epoch": 144.42,
      "grad_norm": 2.0625159740448,
      "learning_rate": 9.220939243505731e-07,
      "loss": 0.4718,
      "step": 352810
    },
    {
      "epoch": 144.42,
      "grad_norm": 1.8787579536437988,
      "learning_rate": 9.219678127374527e-07,
      "loss": 0.4373,
      "step": 352820
    },
    {
      "epoch": 144.42,
      "grad_norm": 1.7180242538452148,
      "learning_rate": 9.218417076372963e-07,
      "loss": 0.4667,
      "step": 352830
    },
    {
      "epoch": 144.43,
      "grad_norm": 2.5677483081817627,
      "learning_rate": 9.217156090506812e-07,
      "loss": 0.4712,
      "step": 352840
    },
    {
      "epoch": 144.43,
      "grad_norm": 1.952950358390808,
      "learning_rate": 9.215895169781858e-07,
      "loss": 0.4302,
      "step": 352850
    },
    {
      "epoch": 144.44,
      "grad_norm": 2.197857141494751,
      "learning_rate": 9.214634314203871e-07,
      "loss": 0.4553,
      "step": 352860
    },
    {
      "epoch": 144.44,
      "grad_norm": 1.7070499658584595,
      "learning_rate": 9.21337352377863e-07,
      "loss": 0.4611,
      "step": 352870
    },
    {
      "epoch": 144.45,
      "grad_norm": 1.9617844820022583,
      "learning_rate": 9.21211279851191e-07,
      "loss": 0.4548,
      "step": 352880
    },
    {
      "epoch": 144.45,
      "grad_norm": 2.3082504272460938,
      "learning_rate": 9.210852138409487e-07,
      "loss": 0.4513,
      "step": 352890
    },
    {
      "epoch": 144.45,
      "grad_norm": 2.516569137573242,
      "learning_rate": 9.209591543477134e-07,
      "loss": 0.4654,
      "step": 352900
    },
    {
      "epoch": 144.46,
      "grad_norm": 1.9141134023666382,
      "learning_rate": 9.208331013720623e-07,
      "loss": 0.4476,
      "step": 352910
    },
    {
      "epoch": 144.46,
      "grad_norm": 1.894066333770752,
      "learning_rate": 9.207070549145734e-07,
      "loss": 0.4319,
      "step": 352920
    },
    {
      "epoch": 144.47,
      "grad_norm": 2.1634936332702637,
      "learning_rate": 9.205810149758238e-07,
      "loss": 0.4627,
      "step": 352930
    },
    {
      "epoch": 144.47,
      "grad_norm": 1.8368127346038818,
      "learning_rate": 9.204549815563908e-07,
      "loss": 0.4684,
      "step": 352940
    },
    {
      "epoch": 144.47,
      "grad_norm": 1.8481457233428955,
      "learning_rate": 9.203289546568522e-07,
      "loss": 0.4497,
      "step": 352950
    },
    {
      "epoch": 144.48,
      "grad_norm": 2.1799070835113525,
      "learning_rate": 9.202029342777847e-07,
      "loss": 0.4382,
      "step": 352960
    },
    {
      "epoch": 144.48,
      "grad_norm": 1.5922584533691406,
      "learning_rate": 9.200769204197658e-07,
      "loss": 0.4189,
      "step": 352970
    },
    {
      "epoch": 144.49,
      "grad_norm": 2.0183537006378174,
      "learning_rate": 9.199509130833728e-07,
      "loss": 0.4508,
      "step": 352980
    },
    {
      "epoch": 144.49,
      "grad_norm": 2.03884220123291,
      "learning_rate": 9.198249122691832e-07,
      "loss": 0.4705,
      "step": 352990
    },
    {
      "epoch": 144.49,
      "grad_norm": 1.9867371320724487,
      "learning_rate": 9.196989179777735e-07,
      "loss": 0.4517,
      "step": 353000
    },
    {
      "epoch": 144.5,
      "grad_norm": 2.1315317153930664,
      "learning_rate": 9.195729302097216e-07,
      "loss": 0.4603,
      "step": 353010
    },
    {
      "epoch": 144.5,
      "grad_norm": 2.377781391143799,
      "learning_rate": 9.19446948965604e-07,
      "loss": 0.4411,
      "step": 353020
    },
    {
      "epoch": 144.51,
      "grad_norm": 1.831443190574646,
      "learning_rate": 9.193209742459984e-07,
      "loss": 0.4588,
      "step": 353030
    },
    {
      "epoch": 144.51,
      "grad_norm": 2.2132058143615723,
      "learning_rate": 9.191950060514821e-07,
      "loss": 0.4609,
      "step": 353040
    },
    {
      "epoch": 144.51,
      "grad_norm": 2.0150225162506104,
      "learning_rate": 9.19069044382631e-07,
      "loss": 0.449,
      "step": 353050
    },
    {
      "epoch": 144.52,
      "grad_norm": 1.982189416885376,
      "learning_rate": 9.189430892400225e-07,
      "loss": 0.4473,
      "step": 353060
    },
    {
      "epoch": 144.52,
      "grad_norm": 2.0846750736236572,
      "learning_rate": 9.18817140624234e-07,
      "loss": 0.4525,
      "step": 353070
    },
    {
      "epoch": 144.53,
      "grad_norm": 1.778314232826233,
      "learning_rate": 9.186911985358423e-07,
      "loss": 0.455,
      "step": 353080
    },
    {
      "epoch": 144.53,
      "grad_norm": 1.8266524076461792,
      "learning_rate": 9.185652629754232e-07,
      "loss": 0.4484,
      "step": 353090
    },
    {
      "epoch": 144.54,
      "grad_norm": 2.2389721870422363,
      "learning_rate": 9.184393339435556e-07,
      "loss": 0.4722,
      "step": 353100
    },
    {
      "epoch": 144.54,
      "grad_norm": 1.6095290184020996,
      "learning_rate": 9.183134114408154e-07,
      "loss": 0.4578,
      "step": 353110
    },
    {
      "epoch": 144.54,
      "grad_norm": 1.8138970136642456,
      "learning_rate": 9.181874954677798e-07,
      "loss": 0.4573,
      "step": 353120
    },
    {
      "epoch": 144.55,
      "grad_norm": 2.3325395584106445,
      "learning_rate": 9.180615860250248e-07,
      "loss": 0.4502,
      "step": 353130
    },
    {
      "epoch": 144.55,
      "grad_norm": 1.8978729248046875,
      "learning_rate": 9.179356831131279e-07,
      "loss": 0.436,
      "step": 353140
    },
    {
      "epoch": 144.56,
      "grad_norm": 2.3003902435302734,
      "learning_rate": 9.178097867326653e-07,
      "loss": 0.4745,
      "step": 353150
    },
    {
      "epoch": 144.56,
      "grad_norm": 1.9107335805892944,
      "learning_rate": 9.176838968842144e-07,
      "loss": 0.4353,
      "step": 353160
    },
    {
      "epoch": 144.56,
      "grad_norm": 1.8979631662368774,
      "learning_rate": 9.175580135683511e-07,
      "loss": 0.4459,
      "step": 353170
    },
    {
      "epoch": 144.57,
      "grad_norm": 2.0653085708618164,
      "learning_rate": 9.174321367856533e-07,
      "loss": 0.4579,
      "step": 353180
    },
    {
      "epoch": 144.57,
      "grad_norm": 1.6634424924850464,
      "learning_rate": 9.17306266536696e-07,
      "loss": 0.4557,
      "step": 353190
    },
    {
      "epoch": 144.58,
      "grad_norm": 2.219733953475952,
      "learning_rate": 9.171804028220563e-07,
      "loss": 0.4451,
      "step": 353200
    },
    {
      "epoch": 144.58,
      "grad_norm": 2.8862931728363037,
      "learning_rate": 9.170545456423114e-07,
      "loss": 0.4412,
      "step": 353210
    },
    {
      "epoch": 144.58,
      "grad_norm": 2.0155680179595947,
      "learning_rate": 9.169286949980371e-07,
      "loss": 0.4478,
      "step": 353220
    },
    {
      "epoch": 144.59,
      "grad_norm": 1.7166414260864258,
      "learning_rate": 9.168028508898102e-07,
      "loss": 0.4487,
      "step": 353230
    },
    {
      "epoch": 144.59,
      "grad_norm": 2.068941354751587,
      "learning_rate": 9.166770133182075e-07,
      "loss": 0.438,
      "step": 353240
    },
    {
      "epoch": 144.6,
      "grad_norm": 2.0983541011810303,
      "learning_rate": 9.165511822838049e-07,
      "loss": 0.4446,
      "step": 353250
    },
    {
      "epoch": 144.6,
      "grad_norm": 2.274075746536255,
      "learning_rate": 9.164253577871793e-07,
      "loss": 0.4578,
      "step": 353260
    },
    {
      "epoch": 144.6,
      "grad_norm": 2.0110814571380615,
      "learning_rate": 9.162995398289065e-07,
      "loss": 0.4552,
      "step": 353270
    },
    {
      "epoch": 144.61,
      "grad_norm": 2.3465166091918945,
      "learning_rate": 9.161737284095626e-07,
      "loss": 0.4495,
      "step": 353280
    },
    {
      "epoch": 144.61,
      "grad_norm": 1.9822310209274292,
      "learning_rate": 9.160479235297252e-07,
      "loss": 0.4472,
      "step": 353290
    },
    {
      "epoch": 144.62,
      "grad_norm": 2.1660544872283936,
      "learning_rate": 9.159221251899699e-07,
      "loss": 0.4205,
      "step": 353300
    },
    {
      "epoch": 144.62,
      "grad_norm": 1.71401047706604,
      "learning_rate": 9.15796333390873e-07,
      "loss": 0.468,
      "step": 353310
    },
    {
      "epoch": 144.63,
      "grad_norm": 2.249596118927002,
      "learning_rate": 9.156705481330117e-07,
      "loss": 0.4342,
      "step": 353320
    },
    {
      "epoch": 144.63,
      "grad_norm": 1.8916727304458618,
      "learning_rate": 9.155447694169604e-07,
      "loss": 0.4469,
      "step": 353330
    },
    {
      "epoch": 144.63,
      "grad_norm": 2.2710959911346436,
      "learning_rate": 9.154189972432958e-07,
      "loss": 0.4438,
      "step": 353340
    },
    {
      "epoch": 144.64,
      "grad_norm": 1.8392329216003418,
      "learning_rate": 9.152932316125944e-07,
      "loss": 0.4476,
      "step": 353350
    },
    {
      "epoch": 144.64,
      "grad_norm": 1.9358363151550293,
      "learning_rate": 9.151674725254322e-07,
      "loss": 0.4675,
      "step": 353360
    },
    {
      "epoch": 144.65,
      "grad_norm": 1.8948979377746582,
      "learning_rate": 9.150417199823856e-07,
      "loss": 0.4656,
      "step": 353370
    },
    {
      "epoch": 144.65,
      "grad_norm": 2.7121803760528564,
      "learning_rate": 9.1491597398403e-07,
      "loss": 0.4434,
      "step": 353380
    },
    {
      "epoch": 144.65,
      "grad_norm": 2.8060495853424072,
      "learning_rate": 9.147902345309419e-07,
      "loss": 0.4549,
      "step": 353390
    },
    {
      "epoch": 144.66,
      "grad_norm": 1.7387301921844482,
      "learning_rate": 9.146645016236973e-07,
      "loss": 0.4647,
      "step": 353400
    },
    {
      "epoch": 144.66,
      "grad_norm": 1.894777536392212,
      "learning_rate": 9.145387752628718e-07,
      "loss": 0.4499,
      "step": 353410
    },
    {
      "epoch": 144.67,
      "grad_norm": 2.3325068950653076,
      "learning_rate": 9.144130554490419e-07,
      "loss": 0.4374,
      "step": 353420
    },
    {
      "epoch": 144.67,
      "grad_norm": 2.2066566944122314,
      "learning_rate": 9.14287342182783e-07,
      "loss": 0.4395,
      "step": 353430
    },
    {
      "epoch": 144.67,
      "grad_norm": 1.912521481513977,
      "learning_rate": 9.141616354646712e-07,
      "loss": 0.4455,
      "step": 353440
    },
    {
      "epoch": 144.68,
      "grad_norm": 2.06712007522583,
      "learning_rate": 9.140359352952822e-07,
      "loss": 0.4619,
      "step": 353450
    },
    {
      "epoch": 144.68,
      "grad_norm": 1.950911045074463,
      "learning_rate": 9.139102416751921e-07,
      "loss": 0.4576,
      "step": 353460
    },
    {
      "epoch": 144.69,
      "grad_norm": 2.2995975017547607,
      "learning_rate": 9.137845546049763e-07,
      "loss": 0.4599,
      "step": 353470
    },
    {
      "epoch": 144.69,
      "grad_norm": 2.1091318130493164,
      "learning_rate": 9.13658874085211e-07,
      "loss": 0.4479,
      "step": 353480
    },
    {
      "epoch": 144.7,
      "grad_norm": 1.7031725645065308,
      "learning_rate": 9.135332001164715e-07,
      "loss": 0.4554,
      "step": 353490
    },
    {
      "epoch": 144.7,
      "grad_norm": 1.9632391929626465,
      "learning_rate": 9.134075326993335e-07,
      "loss": 0.4507,
      "step": 353500
    },
    {
      "epoch": 144.7,
      "grad_norm": 1.855505108833313,
      "learning_rate": 9.132818718343731e-07,
      "loss": 0.4513,
      "step": 353510
    },
    {
      "epoch": 144.71,
      "grad_norm": 1.9326921701431274,
      "learning_rate": 9.131562175221655e-07,
      "loss": 0.4521,
      "step": 353520
    },
    {
      "epoch": 144.71,
      "grad_norm": 1.9559770822525024,
      "learning_rate": 9.130305697632866e-07,
      "loss": 0.4488,
      "step": 353530
    },
    {
      "epoch": 144.72,
      "grad_norm": 2.8663933277130127,
      "learning_rate": 9.129049285583118e-07,
      "loss": 0.4521,
      "step": 353540
    },
    {
      "epoch": 144.72,
      "grad_norm": 1.7979012727737427,
      "learning_rate": 9.127792939078166e-07,
      "loss": 0.4679,
      "step": 353550
    },
    {
      "epoch": 144.72,
      "grad_norm": 2.3780272006988525,
      "learning_rate": 9.126536658123766e-07,
      "loss": 0.4549,
      "step": 353560
    },
    {
      "epoch": 144.73,
      "grad_norm": 2.320371627807617,
      "learning_rate": 9.125280442725673e-07,
      "loss": 0.4421,
      "step": 353570
    },
    {
      "epoch": 144.73,
      "grad_norm": 1.817383050918579,
      "learning_rate": 9.124024292889642e-07,
      "loss": 0.4161,
      "step": 353580
    },
    {
      "epoch": 144.74,
      "grad_norm": 2.0630812644958496,
      "learning_rate": 9.122768208621424e-07,
      "loss": 0.4694,
      "step": 353590
    },
    {
      "epoch": 144.74,
      "grad_norm": 2.188490867614746,
      "learning_rate": 9.12151218992678e-07,
      "loss": 0.4539,
      "step": 353600
    },
    {
      "epoch": 144.74,
      "grad_norm": 2.120832681655884,
      "learning_rate": 9.120256236811456e-07,
      "loss": 0.4523,
      "step": 353610
    },
    {
      "epoch": 144.75,
      "grad_norm": 1.7868988513946533,
      "learning_rate": 9.119000349281208e-07,
      "loss": 0.4624,
      "step": 353620
    },
    {
      "epoch": 144.75,
      "grad_norm": 1.9701718091964722,
      "learning_rate": 9.117744527341799e-07,
      "loss": 0.4579,
      "step": 353630
    },
    {
      "epoch": 144.76,
      "grad_norm": 1.9195053577423096,
      "learning_rate": 9.116488770998964e-07,
      "loss": 0.4585,
      "step": 353640
    },
    {
      "epoch": 144.76,
      "grad_norm": 1.7740193605422974,
      "learning_rate": 9.115233080258451e-07,
      "loss": 0.4622,
      "step": 353650
    },
    {
      "epoch": 144.76,
      "grad_norm": 2.0873286724090576,
      "learning_rate": 9.11397745512604e-07,
      "loss": 0.461,
      "step": 353660
    },
    {
      "epoch": 144.77,
      "grad_norm": 1.6284520626068115,
      "learning_rate": 9.112721895607462e-07,
      "loss": 0.4512,
      "step": 353670
    },
    {
      "epoch": 144.77,
      "grad_norm": 2.1647348403930664,
      "learning_rate": 9.111466401708476e-07,
      "loss": 0.4661,
      "step": 353680
    },
    {
      "epoch": 144.78,
      "grad_norm": 1.8290839195251465,
      "learning_rate": 9.110210973434831e-07,
      "loss": 0.4487,
      "step": 353690
    },
    {
      "epoch": 144.78,
      "grad_norm": 1.7711793184280396,
      "learning_rate": 9.108955610792279e-07,
      "loss": 0.4575,
      "step": 353700
    },
    {
      "epoch": 144.79,
      "grad_norm": 2.1838390827178955,
      "learning_rate": 9.107700313786569e-07,
      "loss": 0.4629,
      "step": 353710
    },
    {
      "epoch": 144.79,
      "grad_norm": 1.7451120615005493,
      "learning_rate": 9.10644508242345e-07,
      "loss": 0.4509,
      "step": 353720
    },
    {
      "epoch": 144.79,
      "grad_norm": 1.9143753051757812,
      "learning_rate": 9.105189916708677e-07,
      "loss": 0.4528,
      "step": 353730
    },
    {
      "epoch": 144.8,
      "grad_norm": 2.2932090759277344,
      "learning_rate": 9.103934816647996e-07,
      "loss": 0.4643,
      "step": 353740
    },
    {
      "epoch": 144.8,
      "grad_norm": 2.118339776992798,
      "learning_rate": 9.102679782247159e-07,
      "loss": 0.4555,
      "step": 353750
    },
    {
      "epoch": 144.81,
      "grad_norm": 1.7634354829788208,
      "learning_rate": 9.101424813511921e-07,
      "loss": 0.4527,
      "step": 353760
    },
    {
      "epoch": 144.81,
      "grad_norm": 2.1354494094848633,
      "learning_rate": 9.100169910448012e-07,
      "loss": 0.448,
      "step": 353770
    },
    {
      "epoch": 144.81,
      "grad_norm": 1.7360910177230835,
      "learning_rate": 9.098915073061195e-07,
      "loss": 0.4414,
      "step": 353780
    },
    {
      "epoch": 144.82,
      "grad_norm": 2.020193576812744,
      "learning_rate": 9.097660301357212e-07,
      "loss": 0.4432,
      "step": 353790
    },
    {
      "epoch": 144.82,
      "grad_norm": 2.5860581398010254,
      "learning_rate": 9.096405595341816e-07,
      "loss": 0.4618,
      "step": 353800
    },
    {
      "epoch": 144.83,
      "grad_norm": 2.013813018798828,
      "learning_rate": 9.095150955020754e-07,
      "loss": 0.4451,
      "step": 353810
    },
    {
      "epoch": 144.83,
      "grad_norm": 1.9289990663528442,
      "learning_rate": 9.093896380399769e-07,
      "loss": 0.4659,
      "step": 353820
    },
    {
      "epoch": 144.83,
      "grad_norm": 2.050137996673584,
      "learning_rate": 9.092641871484612e-07,
      "loss": 0.4545,
      "step": 353830
    },
    {
      "epoch": 144.84,
      "grad_norm": 2.3184821605682373,
      "learning_rate": 9.091387428281019e-07,
      "loss": 0.4502,
      "step": 353840
    },
    {
      "epoch": 144.84,
      "grad_norm": 2.1381442546844482,
      "learning_rate": 9.090133050794755e-07,
      "loss": 0.4536,
      "step": 353850
    },
    {
      "epoch": 144.85,
      "grad_norm": 2.6035854816436768,
      "learning_rate": 9.088878739031558e-07,
      "loss": 0.435,
      "step": 353860
    },
    {
      "epoch": 144.85,
      "grad_norm": 2.4505019187927246,
      "learning_rate": 9.087624492997171e-07,
      "loss": 0.4413,
      "step": 353870
    },
    {
      "epoch": 144.85,
      "grad_norm": 1.7268985509872437,
      "learning_rate": 9.086370312697344e-07,
      "loss": 0.4479,
      "step": 353880
    },
    {
      "epoch": 144.86,
      "grad_norm": 1.8892701864242554,
      "learning_rate": 9.085116198137817e-07,
      "loss": 0.456,
      "step": 353890
    },
    {
      "epoch": 144.86,
      "grad_norm": 1.8986753225326538,
      "learning_rate": 9.083862149324348e-07,
      "loss": 0.4556,
      "step": 353900
    },
    {
      "epoch": 144.87,
      "grad_norm": 2.219761610031128,
      "learning_rate": 9.082608166262659e-07,
      "loss": 0.4586,
      "step": 353910
    },
    {
      "epoch": 144.87,
      "grad_norm": 2.065584182739258,
      "learning_rate": 9.081354248958508e-07,
      "loss": 0.4516,
      "step": 353920
    },
    {
      "epoch": 144.88,
      "grad_norm": 2.0519158840179443,
      "learning_rate": 9.080100397417638e-07,
      "loss": 0.4406,
      "step": 353930
    },
    {
      "epoch": 144.88,
      "grad_norm": 1.9869860410690308,
      "learning_rate": 9.078846611645793e-07,
      "loss": 0.4499,
      "step": 353940
    },
    {
      "epoch": 144.88,
      "grad_norm": 1.9796619415283203,
      "learning_rate": 9.077592891648714e-07,
      "loss": 0.4831,
      "step": 353950
    },
    {
      "epoch": 144.89,
      "grad_norm": 2.192863941192627,
      "learning_rate": 9.076339237432146e-07,
      "loss": 0.4536,
      "step": 353960
    },
    {
      "epoch": 144.89,
      "grad_norm": 2.3868515491485596,
      "learning_rate": 9.075085649001832e-07,
      "loss": 0.4404,
      "step": 353970
    },
    {
      "epoch": 144.9,
      "grad_norm": 2.1240880489349365,
      "learning_rate": 9.073832126363513e-07,
      "loss": 0.4328,
      "step": 353980
    },
    {
      "epoch": 144.9,
      "grad_norm": 1.9366345405578613,
      "learning_rate": 9.072578669522933e-07,
      "loss": 0.4591,
      "step": 353990
    },
    {
      "epoch": 144.9,
      "grad_norm": 2.0172853469848633,
      "learning_rate": 9.07132527848583e-07,
      "loss": 0.4522,
      "step": 354000
    },
    {
      "epoch": 144.91,
      "grad_norm": 1.7123926877975464,
      "learning_rate": 9.070071953257954e-07,
      "loss": 0.4317,
      "step": 354010
    },
    {
      "epoch": 144.91,
      "grad_norm": 2.2452821731567383,
      "learning_rate": 9.068818693845037e-07,
      "loss": 0.4801,
      "step": 354020
    },
    {
      "epoch": 144.92,
      "grad_norm": 1.8168643712997437,
      "learning_rate": 9.067565500252822e-07,
      "loss": 0.4502,
      "step": 354030
    },
    {
      "epoch": 144.92,
      "grad_norm": 2.2947869300842285,
      "learning_rate": 9.066312372487054e-07,
      "loss": 0.4791,
      "step": 354040
    },
    {
      "epoch": 144.92,
      "grad_norm": 1.894626259803772,
      "learning_rate": 9.065059310553472e-07,
      "loss": 0.461,
      "step": 354050
    },
    {
      "epoch": 144.93,
      "grad_norm": 2.118633270263672,
      "learning_rate": 9.063806314457814e-07,
      "loss": 0.4556,
      "step": 354060
    },
    {
      "epoch": 144.93,
      "grad_norm": 1.7945129871368408,
      "learning_rate": 9.062553384205819e-07,
      "loss": 0.4658,
      "step": 354070
    },
    {
      "epoch": 144.94,
      "grad_norm": 1.768172264099121,
      "learning_rate": 9.061300519803232e-07,
      "loss": 0.4458,
      "step": 354080
    },
    {
      "epoch": 144.94,
      "grad_norm": 2.557990312576294,
      "learning_rate": 9.060047721255785e-07,
      "loss": 0.4703,
      "step": 354090
    },
    {
      "epoch": 144.94,
      "grad_norm": 2.05669903755188,
      "learning_rate": 9.058794988569221e-07,
      "loss": 0.4439,
      "step": 354100
    },
    {
      "epoch": 144.95,
      "grad_norm": 2.2398622035980225,
      "learning_rate": 9.057542321749278e-07,
      "loss": 0.4376,
      "step": 354110
    },
    {
      "epoch": 144.95,
      "grad_norm": 2.0007574558258057,
      "learning_rate": 9.056289720801694e-07,
      "loss": 0.4373,
      "step": 354120
    },
    {
      "epoch": 144.96,
      "grad_norm": 2.1296119689941406,
      "learning_rate": 9.055037185732207e-07,
      "loss": 0.43,
      "step": 354130
    },
    {
      "epoch": 144.96,
      "grad_norm": 1.5771851539611816,
      "learning_rate": 9.053784716546554e-07,
      "loss": 0.4529,
      "step": 354140
    },
    {
      "epoch": 144.97,
      "grad_norm": 1.7475107908248901,
      "learning_rate": 9.052532313250476e-07,
      "loss": 0.4428,
      "step": 354150
    },
    {
      "epoch": 144.97,
      "grad_norm": 1.9032596349716187,
      "learning_rate": 9.051279975849705e-07,
      "loss": 0.4754,
      "step": 354160
    },
    {
      "epoch": 144.97,
      "grad_norm": 2.5062215328216553,
      "learning_rate": 9.05002770434998e-07,
      "loss": 0.4433,
      "step": 354170
    },
    {
      "epoch": 144.98,
      "grad_norm": 1.8636828660964966,
      "learning_rate": 9.048775498757037e-07,
      "loss": 0.4662,
      "step": 354180
    },
    {
      "epoch": 144.98,
      "grad_norm": 1.840585470199585,
      "learning_rate": 9.047523359076614e-07,
      "loss": 0.4827,
      "step": 354190
    },
    {
      "epoch": 144.99,
      "grad_norm": 2.099552869796753,
      "learning_rate": 9.046271285314444e-07,
      "loss": 0.4595,
      "step": 354200
    },
    {
      "epoch": 144.99,
      "grad_norm": 1.7554713487625122,
      "learning_rate": 9.045019277476275e-07,
      "loss": 0.4243,
      "step": 354210
    },
    {
      "epoch": 144.99,
      "grad_norm": 1.6029385328292847,
      "learning_rate": 9.043767335567812e-07,
      "loss": 0.4572,
      "step": 354220
    },
    {
      "epoch": 145.0,
      "grad_norm": 1.7872095108032227,
      "learning_rate": 9.042515459594817e-07,
      "loss": 0.4606,
      "step": 354230
    },
    {
      "epoch": 145.0,
      "eval_loss": 0.45112481713294983,
      "eval_runtime": 52.2633,
      "eval_samples_per_second": 65.993,
      "eval_steps_per_second": 8.266,
      "step": 354235
    },
    {
      "epoch": 145.0,
      "grad_norm": 2.130167007446289,
      "learning_rate": 9.041263649563018e-07,
      "loss": 0.4614,
      "step": 354240
    },
    {
      "epoch": 145.01,
      "grad_norm": 1.8785300254821777,
      "learning_rate": 9.040011905478149e-07,
      "loss": 0.4475,
      "step": 354250
    },
    {
      "epoch": 145.01,
      "grad_norm": 2.0917763710021973,
      "learning_rate": 9.038760227345943e-07,
      "loss": 0.4388,
      "step": 354260
    },
    {
      "epoch": 145.01,
      "grad_norm": 1.9249012470245361,
      "learning_rate": 9.037508615172131e-07,
      "loss": 0.4579,
      "step": 354270
    },
    {
      "epoch": 145.02,
      "grad_norm": 1.7981257438659668,
      "learning_rate": 9.036257068962452e-07,
      "loss": 0.4547,
      "step": 354280
    },
    {
      "epoch": 145.02,
      "grad_norm": 2.3660366535186768,
      "learning_rate": 9.035005588722635e-07,
      "loss": 0.4558,
      "step": 354290
    },
    {
      "epoch": 145.03,
      "grad_norm": 2.1291351318359375,
      "learning_rate": 9.033754174458416e-07,
      "loss": 0.4664,
      "step": 354300
    },
    {
      "epoch": 145.03,
      "grad_norm": 1.688310146331787,
      "learning_rate": 9.032502826175527e-07,
      "loss": 0.438,
      "step": 354310
    },
    {
      "epoch": 145.03,
      "grad_norm": 2.117584705352783,
      "learning_rate": 9.031251543879697e-07,
      "loss": 0.4374,
      "step": 354320
    },
    {
      "epoch": 145.04,
      "grad_norm": 2.2487523555755615,
      "learning_rate": 9.030000327576662e-07,
      "loss": 0.4548,
      "step": 354330
    },
    {
      "epoch": 145.04,
      "grad_norm": 2.2851755619049072,
      "learning_rate": 9.028749177272158e-07,
      "loss": 0.4812,
      "step": 354340
    },
    {
      "epoch": 145.05,
      "grad_norm": 2.0828118324279785,
      "learning_rate": 9.027498092971902e-07,
      "loss": 0.4386,
      "step": 354350
    },
    {
      "epoch": 145.05,
      "grad_norm": 2.0094404220581055,
      "learning_rate": 9.026247074681635e-07,
      "loss": 0.4579,
      "step": 354360
    },
    {
      "epoch": 145.06,
      "grad_norm": 1.9345946311950684,
      "learning_rate": 9.024996122407084e-07,
      "loss": 0.4507,
      "step": 354370
    },
    {
      "epoch": 145.06,
      "grad_norm": 1.871707558631897,
      "learning_rate": 9.023745236153981e-07,
      "loss": 0.4322,
      "step": 354380
    },
    {
      "epoch": 145.06,
      "grad_norm": 2.08966326713562,
      "learning_rate": 9.022494415928056e-07,
      "loss": 0.4409,
      "step": 354390
    },
    {
      "epoch": 145.07,
      "grad_norm": 2.1715376377105713,
      "learning_rate": 9.021243661735041e-07,
      "loss": 0.4614,
      "step": 354400
    },
    {
      "epoch": 145.07,
      "grad_norm": 1.9451918601989746,
      "learning_rate": 9.019992973580653e-07,
      "loss": 0.4706,
      "step": 354410
    },
    {
      "epoch": 145.08,
      "grad_norm": 2.090853691101074,
      "learning_rate": 9.018742351470643e-07,
      "loss": 0.4495,
      "step": 354420
    },
    {
      "epoch": 145.08,
      "grad_norm": 2.12902569770813,
      "learning_rate": 9.017491795410725e-07,
      "loss": 0.4461,
      "step": 354430
    },
    {
      "epoch": 145.08,
      "grad_norm": 2.360405921936035,
      "learning_rate": 9.016241305406636e-07,
      "loss": 0.4624,
      "step": 354440
    },
    {
      "epoch": 145.09,
      "grad_norm": 2.349842071533203,
      "learning_rate": 9.014990881464096e-07,
      "loss": 0.4515,
      "step": 354450
    },
    {
      "epoch": 145.09,
      "grad_norm": 2.1191697120666504,
      "learning_rate": 9.01374052358884e-07,
      "loss": 0.4478,
      "step": 354460
    },
    {
      "epoch": 145.1,
      "grad_norm": 1.7802714109420776,
      "learning_rate": 9.01249023178659e-07,
      "loss": 0.449,
      "step": 354470
    },
    {
      "epoch": 145.1,
      "grad_norm": 1.9880667924880981,
      "learning_rate": 9.011240006063083e-07,
      "loss": 0.4552,
      "step": 354480
    },
    {
      "epoch": 145.1,
      "grad_norm": 1.9239972829818726,
      "learning_rate": 9.009989846424033e-07,
      "loss": 0.4471,
      "step": 354490
    },
    {
      "epoch": 145.11,
      "grad_norm": 1.8761322498321533,
      "learning_rate": 9.008739752875171e-07,
      "loss": 0.4565,
      "step": 354500
    },
    {
      "epoch": 145.11,
      "grad_norm": 2.235063314437866,
      "learning_rate": 9.007489725422227e-07,
      "loss": 0.4459,
      "step": 354510
    },
    {
      "epoch": 145.12,
      "grad_norm": 1.9927270412445068,
      "learning_rate": 9.006239764070923e-07,
      "loss": 0.4663,
      "step": 354520
    },
    {
      "epoch": 145.12,
      "grad_norm": 1.9026049375534058,
      "learning_rate": 9.004989868826989e-07,
      "loss": 0.4426,
      "step": 354530
    },
    {
      "epoch": 145.12,
      "grad_norm": 2.584414482116699,
      "learning_rate": 9.003740039696148e-07,
      "loss": 0.4559,
      "step": 354540
    },
    {
      "epoch": 145.13,
      "grad_norm": 2.2244114875793457,
      "learning_rate": 9.002490276684126e-07,
      "loss": 0.4463,
      "step": 354550
    },
    {
      "epoch": 145.13,
      "grad_norm": 4.236281394958496,
      "learning_rate": 9.001240579796648e-07,
      "loss": 0.4496,
      "step": 354560
    },
    {
      "epoch": 145.14,
      "grad_norm": 2.093548536300659,
      "learning_rate": 8.999990949039438e-07,
      "loss": 0.459,
      "step": 354570
    },
    {
      "epoch": 145.14,
      "grad_norm": 1.886271595954895,
      "learning_rate": 8.998741384418223e-07,
      "loss": 0.4416,
      "step": 354580
    },
    {
      "epoch": 145.15,
      "grad_norm": 2.006437063217163,
      "learning_rate": 8.997491885938721e-07,
      "loss": 0.4535,
      "step": 354590
    },
    {
      "epoch": 145.15,
      "grad_norm": 2.276977300643921,
      "learning_rate": 8.996242453606655e-07,
      "loss": 0.4559,
      "step": 354600
    },
    {
      "epoch": 145.15,
      "grad_norm": 2.3374524116516113,
      "learning_rate": 8.994993087427762e-07,
      "loss": 0.4566,
      "step": 354610
    },
    {
      "epoch": 145.16,
      "grad_norm": 2.009890079498291,
      "learning_rate": 8.993743787407763e-07,
      "loss": 0.4555,
      "step": 354620
    },
    {
      "epoch": 145.16,
      "grad_norm": 2.28755521774292,
      "learning_rate": 8.992494553552369e-07,
      "loss": 0.4748,
      "step": 354630
    },
    {
      "epoch": 145.17,
      "grad_norm": 2.25506854057312,
      "learning_rate": 8.991245385867304e-07,
      "loss": 0.4382,
      "step": 354640
    },
    {
      "epoch": 145.17,
      "grad_norm": 2.330862045288086,
      "learning_rate": 8.989996284358296e-07,
      "loss": 0.4652,
      "step": 354650
    },
    {
      "epoch": 145.17,
      "grad_norm": 2.159205198287964,
      "learning_rate": 8.988747249031064e-07,
      "loss": 0.4414,
      "step": 354660
    },
    {
      "epoch": 145.18,
      "grad_norm": 2.4388427734375,
      "learning_rate": 8.987498279891331e-07,
      "loss": 0.4557,
      "step": 354670
    },
    {
      "epoch": 145.18,
      "grad_norm": 1.8384532928466797,
      "learning_rate": 8.98624937694482e-07,
      "loss": 0.4719,
      "step": 354680
    },
    {
      "epoch": 145.19,
      "grad_norm": 2.120042562484741,
      "learning_rate": 8.985000540197246e-07,
      "loss": 0.4415,
      "step": 354690
    },
    {
      "epoch": 145.19,
      "grad_norm": 2.1700453758239746,
      "learning_rate": 8.983751769654336e-07,
      "loss": 0.4444,
      "step": 354700
    },
    {
      "epoch": 145.19,
      "grad_norm": 1.8830300569534302,
      "learning_rate": 8.98250306532181e-07,
      "loss": 0.4434,
      "step": 354710
    },
    {
      "epoch": 145.2,
      "grad_norm": 2.02160906791687,
      "learning_rate": 8.981254427205385e-07,
      "loss": 0.475,
      "step": 354720
    },
    {
      "epoch": 145.2,
      "grad_norm": 2.465182304382324,
      "learning_rate": 8.980005855310781e-07,
      "loss": 0.4579,
      "step": 354730
    },
    {
      "epoch": 145.21,
      "grad_norm": 1.9694894552230835,
      "learning_rate": 8.97875734964372e-07,
      "loss": 0.4622,
      "step": 354740
    },
    {
      "epoch": 145.21,
      "grad_norm": 2.047452688217163,
      "learning_rate": 8.97750891020992e-07,
      "loss": 0.4661,
      "step": 354750
    },
    {
      "epoch": 145.21,
      "grad_norm": 1.9113301038742065,
      "learning_rate": 8.976260537015099e-07,
      "loss": 0.4455,
      "step": 354760
    },
    {
      "epoch": 145.22,
      "grad_norm": 1.9505162239074707,
      "learning_rate": 8.975012230064976e-07,
      "loss": 0.4623,
      "step": 354770
    },
    {
      "epoch": 145.22,
      "grad_norm": 2.1730644702911377,
      "learning_rate": 8.973763989365272e-07,
      "loss": 0.4802,
      "step": 354780
    },
    {
      "epoch": 145.23,
      "grad_norm": 1.8501406908035278,
      "learning_rate": 8.9725158149217e-07,
      "loss": 0.4286,
      "step": 354790
    },
    {
      "epoch": 145.23,
      "grad_norm": 3.470301389694214,
      "learning_rate": 8.971267706739982e-07,
      "loss": 0.4474,
      "step": 354800
    },
    {
      "epoch": 145.24,
      "grad_norm": 2.0611114501953125,
      "learning_rate": 8.970019664825833e-07,
      "loss": 0.4799,
      "step": 354810
    },
    {
      "epoch": 145.24,
      "grad_norm": 1.8074321746826172,
      "learning_rate": 8.968771689184969e-07,
      "loss": 0.4616,
      "step": 354820
    },
    {
      "epoch": 145.24,
      "grad_norm": 2.2183284759521484,
      "learning_rate": 8.967523779823111e-07,
      "loss": 0.4356,
      "step": 354830
    },
    {
      "epoch": 145.25,
      "grad_norm": 2.05265474319458,
      "learning_rate": 8.966275936745974e-07,
      "loss": 0.4409,
      "step": 354840
    },
    {
      "epoch": 145.25,
      "grad_norm": 2.2827916145324707,
      "learning_rate": 8.965028159959272e-07,
      "loss": 0.4407,
      "step": 354850
    },
    {
      "epoch": 145.26,
      "grad_norm": 1.6509270668029785,
      "learning_rate": 8.963780449468721e-07,
      "loss": 0.439,
      "step": 354860
    },
    {
      "epoch": 145.26,
      "grad_norm": 1.943477749824524,
      "learning_rate": 8.962532805280037e-07,
      "loss": 0.4705,
      "step": 354870
    },
    {
      "epoch": 145.26,
      "grad_norm": 2.0644948482513428,
      "learning_rate": 8.961285227398939e-07,
      "loss": 0.4459,
      "step": 354880
    },
    {
      "epoch": 145.27,
      "grad_norm": 2.3622829914093018,
      "learning_rate": 8.960037715831138e-07,
      "loss": 0.4497,
      "step": 354890
    },
    {
      "epoch": 145.27,
      "grad_norm": 1.8775538206100464,
      "learning_rate": 8.95879027058235e-07,
      "loss": 0.4581,
      "step": 354900
    },
    {
      "epoch": 145.28,
      "grad_norm": 1.8889451026916504,
      "learning_rate": 8.957542891658287e-07,
      "loss": 0.4407,
      "step": 354910
    },
    {
      "epoch": 145.28,
      "grad_norm": 2.1265292167663574,
      "learning_rate": 8.956295579064667e-07,
      "loss": 0.4627,
      "step": 354920
    },
    {
      "epoch": 145.28,
      "grad_norm": 2.303727388381958,
      "learning_rate": 8.955048332807211e-07,
      "loss": 0.4608,
      "step": 354930
    },
    {
      "epoch": 145.29,
      "grad_norm": 2.186995029449463,
      "learning_rate": 8.953801152891611e-07,
      "loss": 0.4405,
      "step": 354940
    },
    {
      "epoch": 145.29,
      "grad_norm": 2.0459697246551514,
      "learning_rate": 8.952554039323595e-07,
      "loss": 0.4568,
      "step": 354950
    },
    {
      "epoch": 145.3,
      "grad_norm": 2.2280523777008057,
      "learning_rate": 8.951306992108874e-07,
      "loss": 0.4613,
      "step": 354960
    },
    {
      "epoch": 145.3,
      "grad_norm": 2.1842548847198486,
      "learning_rate": 8.950060011253149e-07,
      "loss": 0.4551,
      "step": 354970
    },
    {
      "epoch": 145.3,
      "grad_norm": 1.812367558479309,
      "learning_rate": 8.948813096762153e-07,
      "loss": 0.4768,
      "step": 354980
    },
    {
      "epoch": 145.31,
      "grad_norm": 1.929112434387207,
      "learning_rate": 8.947566248641587e-07,
      "loss": 0.4356,
      "step": 354990
    },
    {
      "epoch": 145.31,
      "grad_norm": 1.9963572025299072,
      "learning_rate": 8.946319466897163e-07,
      "loss": 0.4633,
      "step": 355000
    },
    {
      "epoch": 145.32,
      "grad_norm": 2.0114431381225586,
      "learning_rate": 8.945072751534594e-07,
      "loss": 0.4556,
      "step": 355010
    },
    {
      "epoch": 145.32,
      "grad_norm": 2.2016377449035645,
      "learning_rate": 8.94382610255959e-07,
      "loss": 0.4515,
      "step": 355020
    },
    {
      "epoch": 145.33,
      "grad_norm": 1.7871778011322021,
      "learning_rate": 8.942579519977861e-07,
      "loss": 0.4623,
      "step": 355030
    },
    {
      "epoch": 145.33,
      "grad_norm": 2.3903818130493164,
      "learning_rate": 8.941333003795118e-07,
      "loss": 0.4426,
      "step": 355040
    },
    {
      "epoch": 145.33,
      "grad_norm": 1.9782590866088867,
      "learning_rate": 8.94008655401707e-07,
      "loss": 0.4571,
      "step": 355050
    },
    {
      "epoch": 145.34,
      "grad_norm": 1.8745427131652832,
      "learning_rate": 8.938840170649436e-07,
      "loss": 0.4414,
      "step": 355060
    },
    {
      "epoch": 145.34,
      "grad_norm": 1.702192783355713,
      "learning_rate": 8.93759385369791e-07,
      "loss": 0.435,
      "step": 355070
    },
    {
      "epoch": 145.35,
      "grad_norm": 2.196988105773926,
      "learning_rate": 8.936347603168207e-07,
      "loss": 0.4463,
      "step": 355080
    },
    {
      "epoch": 145.35,
      "grad_norm": 1.9114214181900024,
      "learning_rate": 8.935101419066039e-07,
      "loss": 0.4455,
      "step": 355090
    },
    {
      "epoch": 145.35,
      "grad_norm": 1.7228686809539795,
      "learning_rate": 8.933855301397112e-07,
      "loss": 0.4408,
      "step": 355100
    },
    {
      "epoch": 145.36,
      "grad_norm": 1.8562836647033691,
      "learning_rate": 8.932609250167137e-07,
      "loss": 0.4438,
      "step": 355110
    },
    {
      "epoch": 145.36,
      "grad_norm": 2.0710151195526123,
      "learning_rate": 8.931363265381818e-07,
      "loss": 0.4348,
      "step": 355120
    },
    {
      "epoch": 145.37,
      "grad_norm": 2.0387418270111084,
      "learning_rate": 8.930117347046865e-07,
      "loss": 0.4377,
      "step": 355130
    },
    {
      "epoch": 145.37,
      "grad_norm": 2.1489651203155518,
      "learning_rate": 8.928871495167988e-07,
      "loss": 0.4629,
      "step": 355140
    },
    {
      "epoch": 145.37,
      "grad_norm": 2.1560142040252686,
      "learning_rate": 8.927625709750888e-07,
      "loss": 0.4514,
      "step": 355150
    },
    {
      "epoch": 145.38,
      "grad_norm": 1.8751418590545654,
      "learning_rate": 8.926379990801267e-07,
      "loss": 0.4493,
      "step": 355160
    },
    {
      "epoch": 145.38,
      "grad_norm": 1.9660041332244873,
      "learning_rate": 8.92513433832485e-07,
      "loss": 0.4542,
      "step": 355170
    },
    {
      "epoch": 145.39,
      "grad_norm": 1.8482844829559326,
      "learning_rate": 8.92388875232733e-07,
      "loss": 0.4391,
      "step": 355180
    },
    {
      "epoch": 145.39,
      "grad_norm": 1.9724217653274536,
      "learning_rate": 8.922643232814418e-07,
      "loss": 0.4707,
      "step": 355190
    },
    {
      "epoch": 145.4,
      "grad_norm": 1.6319531202316284,
      "learning_rate": 8.921397779791821e-07,
      "loss": 0.4396,
      "step": 355200
    },
    {
      "epoch": 145.4,
      "grad_norm": 1.9887698888778687,
      "learning_rate": 8.920152393265237e-07,
      "loss": 0.4459,
      "step": 355210
    },
    {
      "epoch": 145.4,
      "grad_norm": 2.2003910541534424,
      "learning_rate": 8.918907073240371e-07,
      "loss": 0.4483,
      "step": 355220
    },
    {
      "epoch": 145.41,
      "grad_norm": 2.050499200820923,
      "learning_rate": 8.917661819722931e-07,
      "loss": 0.456,
      "step": 355230
    },
    {
      "epoch": 145.41,
      "grad_norm": 1.815368413925171,
      "learning_rate": 8.91641663271862e-07,
      "loss": 0.4528,
      "step": 355240
    },
    {
      "epoch": 145.42,
      "grad_norm": 2.071681261062622,
      "learning_rate": 8.915171512233148e-07,
      "loss": 0.4895,
      "step": 355250
    },
    {
      "epoch": 145.42,
      "grad_norm": 2.09934401512146,
      "learning_rate": 8.913926458272207e-07,
      "loss": 0.4343,
      "step": 355260
    },
    {
      "epoch": 145.42,
      "grad_norm": 1.7219525575637817,
      "learning_rate": 8.912681470841511e-07,
      "loss": 0.4574,
      "step": 355270
    },
    {
      "epoch": 145.43,
      "grad_norm": 3.0218698978424072,
      "learning_rate": 8.91143654994676e-07,
      "loss": 0.4522,
      "step": 355280
    },
    {
      "epoch": 145.43,
      "grad_norm": 1.9362993240356445,
      "learning_rate": 8.910191695593654e-07,
      "loss": 0.4726,
      "step": 355290
    },
    {
      "epoch": 145.44,
      "grad_norm": 1.9872088432312012,
      "learning_rate": 8.908946907787898e-07,
      "loss": 0.4552,
      "step": 355300
    },
    {
      "epoch": 145.44,
      "grad_norm": 1.8574974536895752,
      "learning_rate": 8.907702186535193e-07,
      "loss": 0.4488,
      "step": 355310
    },
    {
      "epoch": 145.44,
      "grad_norm": 2.2238917350769043,
      "learning_rate": 8.906457531841243e-07,
      "loss": 0.4636,
      "step": 355320
    },
    {
      "epoch": 145.45,
      "grad_norm": 2.046052932739258,
      "learning_rate": 8.905212943711746e-07,
      "loss": 0.4681,
      "step": 355330
    },
    {
      "epoch": 145.45,
      "grad_norm": 1.7855268716812134,
      "learning_rate": 8.903968422152408e-07,
      "loss": 0.4537,
      "step": 355340
    },
    {
      "epoch": 145.46,
      "grad_norm": 2.306008815765381,
      "learning_rate": 8.902723967168926e-07,
      "loss": 0.4489,
      "step": 355350
    },
    {
      "epoch": 145.46,
      "grad_norm": 1.7401316165924072,
      "learning_rate": 8.901479578767002e-07,
      "loss": 0.4311,
      "step": 355360
    },
    {
      "epoch": 145.46,
      "grad_norm": 1.8835206031799316,
      "learning_rate": 8.900235256952336e-07,
      "loss": 0.4563,
      "step": 355370
    },
    {
      "epoch": 145.47,
      "grad_norm": 1.8546689748764038,
      "learning_rate": 8.898991001730629e-07,
      "loss": 0.4466,
      "step": 355380
    },
    {
      "epoch": 145.47,
      "grad_norm": 2.7616078853607178,
      "learning_rate": 8.897746813107581e-07,
      "loss": 0.4347,
      "step": 355390
    },
    {
      "epoch": 145.48,
      "grad_norm": 1.729905128479004,
      "learning_rate": 8.896502691088888e-07,
      "loss": 0.4528,
      "step": 355400
    },
    {
      "epoch": 145.48,
      "grad_norm": 2.0498034954071045,
      "learning_rate": 8.895258635680254e-07,
      "loss": 0.458,
      "step": 355410
    },
    {
      "epoch": 145.49,
      "grad_norm": 2.054140567779541,
      "learning_rate": 8.894014646887377e-07,
      "loss": 0.4504,
      "step": 355420
    },
    {
      "epoch": 145.49,
      "grad_norm": 1.9128108024597168,
      "learning_rate": 8.89277072471595e-07,
      "loss": 0.4543,
      "step": 355430
    },
    {
      "epoch": 145.49,
      "grad_norm": 2.4833145141601562,
      "learning_rate": 8.891526869171678e-07,
      "loss": 0.4638,
      "step": 355440
    },
    {
      "epoch": 145.5,
      "grad_norm": 2.4754250049591064,
      "learning_rate": 8.890283080260256e-07,
      "loss": 0.4466,
      "step": 355450
    },
    {
      "epoch": 145.5,
      "grad_norm": 1.857104778289795,
      "learning_rate": 8.889039357987379e-07,
      "loss": 0.4553,
      "step": 355460
    },
    {
      "epoch": 145.51,
      "grad_norm": 1.8090593814849854,
      "learning_rate": 8.887795702358751e-07,
      "loss": 0.4549,
      "step": 355470
    },
    {
      "epoch": 145.51,
      "grad_norm": 1.8652292490005493,
      "learning_rate": 8.886552113380064e-07,
      "loss": 0.4463,
      "step": 355480
    },
    {
      "epoch": 145.51,
      "grad_norm": 1.918087124824524,
      "learning_rate": 8.885308591057015e-07,
      "loss": 0.4472,
      "step": 355490
    },
    {
      "epoch": 145.52,
      "grad_norm": 2.2182579040527344,
      "learning_rate": 8.8840651353953e-07,
      "loss": 0.4568,
      "step": 355500
    },
    {
      "epoch": 145.52,
      "grad_norm": 1.9647510051727295,
      "learning_rate": 8.882821746400626e-07,
      "loss": 0.4539,
      "step": 355510
    },
    {
      "epoch": 145.53,
      "grad_norm": 2.0869126319885254,
      "learning_rate": 8.881578424078672e-07,
      "loss": 0.4529,
      "step": 355520
    },
    {
      "epoch": 145.53,
      "grad_norm": 1.9386283159255981,
      "learning_rate": 8.880335168435138e-07,
      "loss": 0.4564,
      "step": 355530
    },
    {
      "epoch": 145.53,
      "grad_norm": 2.2359471321105957,
      "learning_rate": 8.879091979475715e-07,
      "loss": 0.4585,
      "step": 355540
    },
    {
      "epoch": 145.54,
      "grad_norm": 2.2513303756713867,
      "learning_rate": 8.877848857206113e-07,
      "loss": 0.4602,
      "step": 355550
    },
    {
      "epoch": 145.54,
      "grad_norm": 2.0725419521331787,
      "learning_rate": 8.876605801632017e-07,
      "loss": 0.4223,
      "step": 355560
    },
    {
      "epoch": 145.55,
      "grad_norm": 2.343987464904785,
      "learning_rate": 8.875362812759123e-07,
      "loss": 0.4427,
      "step": 355570
    },
    {
      "epoch": 145.55,
      "grad_norm": 2.091015100479126,
      "learning_rate": 8.874119890593124e-07,
      "loss": 0.4668,
      "step": 355580
    },
    {
      "epoch": 145.55,
      "grad_norm": 1.8742926120758057,
      "learning_rate": 8.872877035139713e-07,
      "loss": 0.4431,
      "step": 355590
    },
    {
      "epoch": 145.56,
      "grad_norm": 1.7044107913970947,
      "learning_rate": 8.871634246404584e-07,
      "loss": 0.4502,
      "step": 355600
    },
    {
      "epoch": 145.56,
      "grad_norm": 1.8762893676757812,
      "learning_rate": 8.87039152439343e-07,
      "loss": 0.4519,
      "step": 355610
    },
    {
      "epoch": 145.57,
      "grad_norm": 1.8286848068237305,
      "learning_rate": 8.869148869111946e-07,
      "loss": 0.4307,
      "step": 355620
    },
    {
      "epoch": 145.57,
      "grad_norm": 2.4063732624053955,
      "learning_rate": 8.86790628056582e-07,
      "loss": 0.4448,
      "step": 355630
    },
    {
      "epoch": 145.58,
      "grad_norm": 1.8402767181396484,
      "learning_rate": 8.866663758760749e-07,
      "loss": 0.4516,
      "step": 355640
    },
    {
      "epoch": 145.58,
      "grad_norm": 1.9259206056594849,
      "learning_rate": 8.86542130370243e-07,
      "loss": 0.456,
      "step": 355650
    },
    {
      "epoch": 145.58,
      "grad_norm": 1.8727009296417236,
      "learning_rate": 8.864178915396538e-07,
      "loss": 0.4603,
      "step": 355660
    },
    {
      "epoch": 145.59,
      "grad_norm": 2.072348117828369,
      "learning_rate": 8.86293659384877e-07,
      "loss": 0.4213,
      "step": 355670
    },
    {
      "epoch": 145.59,
      "grad_norm": 2.1564979553222656,
      "learning_rate": 8.861694339064824e-07,
      "loss": 0.4336,
      "step": 355680
    },
    {
      "epoch": 145.6,
      "grad_norm": 1.9847124814987183,
      "learning_rate": 8.860452151050385e-07,
      "loss": 0.466,
      "step": 355690
    },
    {
      "epoch": 145.6,
      "grad_norm": 1.9053373336791992,
      "learning_rate": 8.859210029811144e-07,
      "loss": 0.4672,
      "step": 355700
    },
    {
      "epoch": 145.6,
      "grad_norm": 1.7963746786117554,
      "learning_rate": 8.857967975352792e-07,
      "loss": 0.438,
      "step": 355710
    },
    {
      "epoch": 145.61,
      "grad_norm": 2.202547788619995,
      "learning_rate": 8.856725987681017e-07,
      "loss": 0.4486,
      "step": 355720
    },
    {
      "epoch": 145.61,
      "grad_norm": 1.9557640552520752,
      "learning_rate": 8.855484066801504e-07,
      "loss": 0.4577,
      "step": 355730
    },
    {
      "epoch": 145.62,
      "grad_norm": 1.9082032442092896,
      "learning_rate": 8.854242212719954e-07,
      "loss": 0.4702,
      "step": 355740
    },
    {
      "epoch": 145.62,
      "grad_norm": 2.0689260959625244,
      "learning_rate": 8.853000425442052e-07,
      "loss": 0.4537,
      "step": 355750
    },
    {
      "epoch": 145.62,
      "grad_norm": 2.03415846824646,
      "learning_rate": 8.851758704973482e-07,
      "loss": 0.447,
      "step": 355760
    },
    {
      "epoch": 145.63,
      "grad_norm": 2.1512582302093506,
      "learning_rate": 8.850517051319934e-07,
      "loss": 0.4431,
      "step": 355770
    },
    {
      "epoch": 145.63,
      "grad_norm": 2.478299617767334,
      "learning_rate": 8.849275464487102e-07,
      "loss": 0.4594,
      "step": 355780
    },
    {
      "epoch": 145.64,
      "grad_norm": 1.8516706228256226,
      "learning_rate": 8.848033944480663e-07,
      "loss": 0.4677,
      "step": 355790
    },
    {
      "epoch": 145.64,
      "grad_norm": 2.1487627029418945,
      "learning_rate": 8.846792491306306e-07,
      "loss": 0.4405,
      "step": 355800
    },
    {
      "epoch": 145.64,
      "grad_norm": 2.026672601699829,
      "learning_rate": 8.845551104969721e-07,
      "loss": 0.4359,
      "step": 355810
    },
    {
      "epoch": 145.65,
      "grad_norm": 1.9630659818649292,
      "learning_rate": 8.844309785476593e-07,
      "loss": 0.4511,
      "step": 355820
    },
    {
      "epoch": 145.65,
      "grad_norm": 2.4793825149536133,
      "learning_rate": 8.84306853283261e-07,
      "loss": 0.455,
      "step": 355830
    },
    {
      "epoch": 145.66,
      "grad_norm": 2.2993037700653076,
      "learning_rate": 8.841827347043458e-07,
      "loss": 0.4705,
      "step": 355840
    },
    {
      "epoch": 145.66,
      "grad_norm": 2.071505069732666,
      "learning_rate": 8.840586228114822e-07,
      "loss": 0.4452,
      "step": 355850
    },
    {
      "epoch": 145.67,
      "grad_norm": 2.1855406761169434,
      "learning_rate": 8.839345176052387e-07,
      "loss": 0.4353,
      "step": 355860
    },
    {
      "epoch": 145.67,
      "grad_norm": 2.4909849166870117,
      "learning_rate": 8.838104190861838e-07,
      "loss": 0.4455,
      "step": 355870
    },
    {
      "epoch": 145.67,
      "grad_norm": 1.630976915359497,
      "learning_rate": 8.836863272548863e-07,
      "loss": 0.4585,
      "step": 355880
    },
    {
      "epoch": 145.68,
      "grad_norm": 1.8453800678253174,
      "learning_rate": 8.835622421119141e-07,
      "loss": 0.455,
      "step": 355890
    },
    {
      "epoch": 145.68,
      "grad_norm": 2.5354909896850586,
      "learning_rate": 8.834381636578359e-07,
      "loss": 0.4389,
      "step": 355900
    },
    {
      "epoch": 145.69,
      "grad_norm": 2.3187661170959473,
      "learning_rate": 8.8331409189322e-07,
      "loss": 0.4401,
      "step": 355910
    },
    {
      "epoch": 145.69,
      "grad_norm": 1.7029330730438232,
      "learning_rate": 8.831900268186349e-07,
      "loss": 0.4573,
      "step": 355920
    },
    {
      "epoch": 145.69,
      "grad_norm": 1.7517991065979004,
      "learning_rate": 8.83065968434649e-07,
      "loss": 0.4425,
      "step": 355930
    },
    {
      "epoch": 145.7,
      "grad_norm": 2.2346701622009277,
      "learning_rate": 8.829419167418301e-07,
      "loss": 0.4717,
      "step": 355940
    },
    {
      "epoch": 145.7,
      "grad_norm": 2.049503803253174,
      "learning_rate": 8.82817871740747e-07,
      "loss": 0.4457,
      "step": 355950
    },
    {
      "epoch": 145.71,
      "grad_norm": 2.3985595703125,
      "learning_rate": 8.826938334319676e-07,
      "loss": 0.4494,
      "step": 355960
    },
    {
      "epoch": 145.71,
      "grad_norm": 2.4658937454223633,
      "learning_rate": 8.825698018160606e-07,
      "loss": 0.4582,
      "step": 355970
    },
    {
      "epoch": 145.71,
      "grad_norm": 1.9921472072601318,
      "learning_rate": 8.824457768935933e-07,
      "loss": 0.4461,
      "step": 355980
    },
    {
      "epoch": 145.72,
      "grad_norm": 2.200187921524048,
      "learning_rate": 8.823217586651347e-07,
      "loss": 0.4627,
      "step": 355990
    },
    {
      "epoch": 145.72,
      "grad_norm": 1.866531491279602,
      "learning_rate": 8.821977471312522e-07,
      "loss": 0.4543,
      "step": 356000
    },
    {
      "epoch": 145.73,
      "grad_norm": 1.6218472719192505,
      "learning_rate": 8.820737422925144e-07,
      "loss": 0.4527,
      "step": 356010
    },
    {
      "epoch": 145.73,
      "grad_norm": 1.7496938705444336,
      "learning_rate": 8.819497441494894e-07,
      "loss": 0.4502,
      "step": 356020
    },
    {
      "epoch": 145.73,
      "grad_norm": 1.897889494895935,
      "learning_rate": 8.818257527027448e-07,
      "loss": 0.4666,
      "step": 356030
    },
    {
      "epoch": 145.74,
      "grad_norm": 2.043680429458618,
      "learning_rate": 8.817017679528486e-07,
      "loss": 0.4561,
      "step": 356040
    },
    {
      "epoch": 145.74,
      "grad_norm": 2.91104793548584,
      "learning_rate": 8.815777899003691e-07,
      "loss": 0.4389,
      "step": 356050
    },
    {
      "epoch": 145.75,
      "grad_norm": 1.888481855392456,
      "learning_rate": 8.814538185458741e-07,
      "loss": 0.462,
      "step": 356060
    },
    {
      "epoch": 145.75,
      "grad_norm": 2.116586923599243,
      "learning_rate": 8.813298538899312e-07,
      "loss": 0.4389,
      "step": 356070
    },
    {
      "epoch": 145.76,
      "grad_norm": 2.1509463787078857,
      "learning_rate": 8.812058959331087e-07,
      "loss": 0.4299,
      "step": 356080
    },
    {
      "epoch": 145.76,
      "grad_norm": 1.9273183345794678,
      "learning_rate": 8.810819446759751e-07,
      "loss": 0.4533,
      "step": 356090
    },
    {
      "epoch": 145.76,
      "grad_norm": 2.073493003845215,
      "learning_rate": 8.809580001190962e-07,
      "loss": 0.4449,
      "step": 356100
    },
    {
      "epoch": 145.77,
      "grad_norm": 1.9670040607452393,
      "learning_rate": 8.808340622630406e-07,
      "loss": 0.4702,
      "step": 356110
    },
    {
      "epoch": 145.77,
      "grad_norm": 2.232696533203125,
      "learning_rate": 8.807101311083768e-07,
      "loss": 0.4279,
      "step": 356120
    },
    {
      "epoch": 145.78,
      "grad_norm": 1.9012242555618286,
      "learning_rate": 8.805862066556722e-07,
      "loss": 0.4523,
      "step": 356130
    },
    {
      "epoch": 145.78,
      "grad_norm": 1.9662836790084839,
      "learning_rate": 8.804622889054943e-07,
      "loss": 0.4353,
      "step": 356140
    },
    {
      "epoch": 145.78,
      "grad_norm": 2.077876329421997,
      "learning_rate": 8.803383778584109e-07,
      "loss": 0.4642,
      "step": 356150
    },
    {
      "epoch": 145.79,
      "grad_norm": 2.0338189601898193,
      "learning_rate": 8.802144735149893e-07,
      "loss": 0.4442,
      "step": 356160
    },
    {
      "epoch": 145.79,
      "grad_norm": 3.60003662109375,
      "learning_rate": 8.800905758757974e-07,
      "loss": 0.4597,
      "step": 356170
    },
    {
      "epoch": 145.8,
      "grad_norm": 1.737282633781433,
      "learning_rate": 8.799666849414026e-07,
      "loss": 0.4388,
      "step": 356180
    },
    {
      "epoch": 145.8,
      "grad_norm": 2.0355236530303955,
      "learning_rate": 8.798428007123725e-07,
      "loss": 0.4276,
      "step": 356190
    },
    {
      "epoch": 145.8,
      "grad_norm": 1.5681012868881226,
      "learning_rate": 8.797189231892744e-07,
      "loss": 0.434,
      "step": 356200
    },
    {
      "epoch": 145.81,
      "grad_norm": 2.466378927230835,
      "learning_rate": 8.795950523726761e-07,
      "loss": 0.4441,
      "step": 356210
    },
    {
      "epoch": 145.81,
      "grad_norm": 2.132495880126953,
      "learning_rate": 8.794711882631448e-07,
      "loss": 0.4161,
      "step": 356220
    },
    {
      "epoch": 145.82,
      "grad_norm": 2.222625255584717,
      "learning_rate": 8.793473308612489e-07,
      "loss": 0.4518,
      "step": 356230
    },
    {
      "epoch": 145.82,
      "grad_norm": 2.5305726528167725,
      "learning_rate": 8.792234801675538e-07,
      "loss": 0.4627,
      "step": 356240
    },
    {
      "epoch": 145.82,
      "grad_norm": 2.29819393157959,
      "learning_rate": 8.790996361826282e-07,
      "loss": 0.4741,
      "step": 356250
    },
    {
      "epoch": 145.83,
      "grad_norm": 1.5889112949371338,
      "learning_rate": 8.789757989070388e-07,
      "loss": 0.4306,
      "step": 356260
    },
    {
      "epoch": 145.83,
      "grad_norm": 2.34169340133667,
      "learning_rate": 8.788519683413533e-07,
      "loss": 0.449,
      "step": 356270
    },
    {
      "epoch": 145.84,
      "grad_norm": 2.0480358600616455,
      "learning_rate": 8.787281444861388e-07,
      "loss": 0.4361,
      "step": 356280
    },
    {
      "epoch": 145.84,
      "grad_norm": 2.21927547454834,
      "learning_rate": 8.786043273419617e-07,
      "loss": 0.4628,
      "step": 356290
    },
    {
      "epoch": 145.85,
      "grad_norm": 1.9885591268539429,
      "learning_rate": 8.784805169093908e-07,
      "loss": 0.4437,
      "step": 356300
    },
    {
      "epoch": 145.85,
      "grad_norm": 2.4685137271881104,
      "learning_rate": 8.783567131889927e-07,
      "loss": 0.4508,
      "step": 356310
    },
    {
      "epoch": 145.85,
      "grad_norm": 1.9633769989013672,
      "learning_rate": 8.782329161813341e-07,
      "loss": 0.453,
      "step": 356320
    },
    {
      "epoch": 145.86,
      "grad_norm": 1.727181315422058,
      "learning_rate": 8.781091258869822e-07,
      "loss": 0.4374,
      "step": 356330
    },
    {
      "epoch": 145.86,
      "grad_norm": 1.7020602226257324,
      "learning_rate": 8.779853423065043e-07,
      "loss": 0.4449,
      "step": 356340
    },
    {
      "epoch": 145.87,
      "grad_norm": 2.0040054321289062,
      "learning_rate": 8.778615654404674e-07,
      "loss": 0.445,
      "step": 356350
    },
    {
      "epoch": 145.87,
      "grad_norm": 2.4579362869262695,
      "learning_rate": 8.77737795289439e-07,
      "loss": 0.4557,
      "step": 356360
    },
    {
      "epoch": 145.87,
      "grad_norm": 2.103121042251587,
      "learning_rate": 8.776140318539847e-07,
      "loss": 0.4496,
      "step": 356370
    },
    {
      "epoch": 145.88,
      "grad_norm": 1.8622901439666748,
      "learning_rate": 8.774902751346723e-07,
      "loss": 0.4458,
      "step": 356380
    },
    {
      "epoch": 145.88,
      "grad_norm": 1.9686282873153687,
      "learning_rate": 8.773665251320684e-07,
      "loss": 0.4731,
      "step": 356390
    },
    {
      "epoch": 145.89,
      "grad_norm": 2.2587838172912598,
      "learning_rate": 8.772427818467403e-07,
      "loss": 0.4393,
      "step": 356400
    },
    {
      "epoch": 145.89,
      "grad_norm": 1.9869483709335327,
      "learning_rate": 8.771190452792548e-07,
      "loss": 0.4449,
      "step": 356410
    },
    {
      "epoch": 145.89,
      "grad_norm": 1.6196562051773071,
      "learning_rate": 8.769953154301783e-07,
      "loss": 0.423,
      "step": 356420
    },
    {
      "epoch": 145.9,
      "grad_norm": 1.7492035627365112,
      "learning_rate": 8.768715923000779e-07,
      "loss": 0.4613,
      "step": 356430
    },
    {
      "epoch": 145.9,
      "grad_norm": 1.939105749130249,
      "learning_rate": 8.767478758895204e-07,
      "loss": 0.4539,
      "step": 356440
    },
    {
      "epoch": 145.91,
      "grad_norm": 1.7897297143936157,
      "learning_rate": 8.766241661990723e-07,
      "loss": 0.4403,
      "step": 356450
    },
    {
      "epoch": 145.91,
      "grad_norm": 1.5858213901519775,
      "learning_rate": 8.765004632293005e-07,
      "loss": 0.4344,
      "step": 356460
    },
    {
      "epoch": 145.91,
      "grad_norm": 2.29179310798645,
      "learning_rate": 8.763767669807716e-07,
      "loss": 0.4537,
      "step": 356470
    },
    {
      "epoch": 145.92,
      "grad_norm": 2.665493965148926,
      "learning_rate": 8.762530774540514e-07,
      "loss": 0.4467,
      "step": 356480
    },
    {
      "epoch": 145.92,
      "grad_norm": 1.8415240049362183,
      "learning_rate": 8.761293946497083e-07,
      "loss": 0.4365,
      "step": 356490
    },
    {
      "epoch": 145.93,
      "grad_norm": 1.7714476585388184,
      "learning_rate": 8.760057185683087e-07,
      "loss": 0.4506,
      "step": 356500
    },
    {
      "epoch": 145.93,
      "grad_norm": 2.098839044570923,
      "learning_rate": 8.758820492104173e-07,
      "loss": 0.4543,
      "step": 356510
    },
    {
      "epoch": 145.94,
      "grad_norm": 1.991899847984314,
      "learning_rate": 8.757583865766018e-07,
      "loss": 0.433,
      "step": 356520
    },
    {
      "epoch": 145.94,
      "grad_norm": 1.6837666034698486,
      "learning_rate": 8.756347306674286e-07,
      "loss": 0.4494,
      "step": 356530
    },
    {
      "epoch": 145.94,
      "grad_norm": 1.795183539390564,
      "learning_rate": 8.755110814834639e-07,
      "loss": 0.4648,
      "step": 356540
    },
    {
      "epoch": 145.95,
      "grad_norm": 2.239797353744507,
      "learning_rate": 8.753874390252744e-07,
      "loss": 0.4425,
      "step": 356550
    },
    {
      "epoch": 145.95,
      "grad_norm": 1.891608476638794,
      "learning_rate": 8.752638032934264e-07,
      "loss": 0.4518,
      "step": 356560
    },
    {
      "epoch": 145.96,
      "grad_norm": 2.140275716781616,
      "learning_rate": 8.751401742884863e-07,
      "loss": 0.4456,
      "step": 356570
    },
    {
      "epoch": 145.96,
      "grad_norm": 1.8643091917037964,
      "learning_rate": 8.750165520110202e-07,
      "loss": 0.4532,
      "step": 356580
    },
    {
      "epoch": 145.96,
      "grad_norm": 1.6895949840545654,
      "learning_rate": 8.748929364615947e-07,
      "loss": 0.4795,
      "step": 356590
    },
    {
      "epoch": 145.97,
      "grad_norm": 1.8824588060379028,
      "learning_rate": 8.747693276407759e-07,
      "loss": 0.4652,
      "step": 356600
    },
    {
      "epoch": 145.97,
      "grad_norm": 1.8678644895553589,
      "learning_rate": 8.7464572554913e-07,
      "loss": 0.4458,
      "step": 356610
    },
    {
      "epoch": 145.98,
      "grad_norm": 1.925203800201416,
      "learning_rate": 8.745221301872233e-07,
      "loss": 0.4576,
      "step": 356620
    },
    {
      "epoch": 145.98,
      "grad_norm": 2.189481735229492,
      "learning_rate": 8.74398541555622e-07,
      "loss": 0.4741,
      "step": 356630
    },
    {
      "epoch": 145.98,
      "grad_norm": 2.254594326019287,
      "learning_rate": 8.74274959654892e-07,
      "loss": 0.4569,
      "step": 356640
    },
    {
      "epoch": 145.99,
      "grad_norm": 1.8863431215286255,
      "learning_rate": 8.741513844855998e-07,
      "loss": 0.4299,
      "step": 356650
    },
    {
      "epoch": 145.99,
      "grad_norm": 1.8630752563476562,
      "learning_rate": 8.740278160483111e-07,
      "loss": 0.4471,
      "step": 356660
    },
    {
      "epoch": 146.0,
      "grad_norm": 1.9259541034698486,
      "learning_rate": 8.739042543435921e-07,
      "loss": 0.4315,
      "step": 356670
    },
    {
      "epoch": 146.0,
      "eval_loss": 0.451404333114624,
      "eval_runtime": 59.9157,
      "eval_samples_per_second": 57.564,
      "eval_steps_per_second": 7.21,
      "step": 356678
    },
    {
      "epoch": 146.0,
      "grad_norm": 1.7859948873519897,
      "learning_rate": 8.73780699372009e-07,
      "loss": 0.4578,
      "step": 356680
    },
    {
      "epoch": 146.0,
      "grad_norm": 1.770837664604187,
      "learning_rate": 8.736571511341277e-07,
      "loss": 0.4484,
      "step": 356690
    },
    {
      "epoch": 146.01,
      "grad_norm": 2.0047030448913574,
      "learning_rate": 8.735336096305138e-07,
      "loss": 0.4562,
      "step": 356700
    },
    {
      "epoch": 146.01,
      "grad_norm": 1.681038737297058,
      "learning_rate": 8.734100748617334e-07,
      "loss": 0.4487,
      "step": 356710
    },
    {
      "epoch": 146.02,
      "grad_norm": 2.169559955596924,
      "learning_rate": 8.732865468283528e-07,
      "loss": 0.4492,
      "step": 356720
    },
    {
      "epoch": 146.02,
      "grad_norm": 2.3367037773132324,
      "learning_rate": 8.731630255309374e-07,
      "loss": 0.4676,
      "step": 356730
    },
    {
      "epoch": 146.03,
      "grad_norm": 2.156412363052368,
      "learning_rate": 8.730395109700531e-07,
      "loss": 0.4147,
      "step": 356740
    },
    {
      "epoch": 146.03,
      "grad_norm": 1.9333664178848267,
      "learning_rate": 8.72916003146266e-07,
      "loss": 0.4663,
      "step": 356750
    },
    {
      "epoch": 146.03,
      "grad_norm": 1.7424695491790771,
      "learning_rate": 8.727925020601413e-07,
      "loss": 0.4489,
      "step": 356760
    },
    {
      "epoch": 146.04,
      "grad_norm": 2.416499376296997,
      "learning_rate": 8.726690077122456e-07,
      "loss": 0.4519,
      "step": 356770
    },
    {
      "epoch": 146.04,
      "grad_norm": 2.0288822650909424,
      "learning_rate": 8.725455201031437e-07,
      "loss": 0.4566,
      "step": 356780
    },
    {
      "epoch": 146.05,
      "grad_norm": 2.0050504207611084,
      "learning_rate": 8.724220392334017e-07,
      "loss": 0.4199,
      "step": 356790
    },
    {
      "epoch": 146.05,
      "grad_norm": 2.753242254257202,
      "learning_rate": 8.722985651035853e-07,
      "loss": 0.4379,
      "step": 356800
    },
    {
      "epoch": 146.05,
      "grad_norm": 1.9377087354660034,
      "learning_rate": 8.721750977142608e-07,
      "loss": 0.4417,
      "step": 356810
    },
    {
      "epoch": 146.06,
      "grad_norm": 2.497657537460327,
      "learning_rate": 8.720516370659923e-07,
      "loss": 0.4438,
      "step": 356820
    },
    {
      "epoch": 146.06,
      "grad_norm": 1.9656111001968384,
      "learning_rate": 8.719281831593462e-07,
      "loss": 0.4615,
      "step": 356830
    },
    {
      "epoch": 146.07,
      "grad_norm": 2.0199356079101562,
      "learning_rate": 8.718047359948877e-07,
      "loss": 0.4867,
      "step": 356840
    },
    {
      "epoch": 146.07,
      "grad_norm": 2.155446767807007,
      "learning_rate": 8.716812955731826e-07,
      "loss": 0.4492,
      "step": 356850
    },
    {
      "epoch": 146.07,
      "grad_norm": 1.9106379747390747,
      "learning_rate": 8.715578618947954e-07,
      "loss": 0.4533,
      "step": 356860
    },
    {
      "epoch": 146.08,
      "grad_norm": 2.0107297897338867,
      "learning_rate": 8.714344349602932e-07,
      "loss": 0.4538,
      "step": 356870
    },
    {
      "epoch": 146.08,
      "grad_norm": 1.8401870727539062,
      "learning_rate": 8.713110147702407e-07,
      "loss": 0.4362,
      "step": 356880
    },
    {
      "epoch": 146.09,
      "grad_norm": 2.110588312149048,
      "learning_rate": 8.71187601325203e-07,
      "loss": 0.4494,
      "step": 356890
    },
    {
      "epoch": 146.09,
      "grad_norm": 2.4863877296447754,
      "learning_rate": 8.710641946257455e-07,
      "loss": 0.4575,
      "step": 356900
    },
    {
      "epoch": 146.09,
      "grad_norm": 2.0026357173919678,
      "learning_rate": 8.709407946724339e-07,
      "loss": 0.4421,
      "step": 356910
    },
    {
      "epoch": 146.1,
      "grad_norm": 2.1050705909729004,
      "learning_rate": 8.70817401465833e-07,
      "loss": 0.4531,
      "step": 356920
    },
    {
      "epoch": 146.1,
      "grad_norm": 2.185058116912842,
      "learning_rate": 8.706940150065082e-07,
      "loss": 0.4483,
      "step": 356930
    },
    {
      "epoch": 146.11,
      "grad_norm": 1.8781996965408325,
      "learning_rate": 8.705706352950248e-07,
      "loss": 0.4353,
      "step": 356940
    },
    {
      "epoch": 146.11,
      "grad_norm": 2.4293317794799805,
      "learning_rate": 8.704472623319487e-07,
      "loss": 0.4434,
      "step": 356950
    },
    {
      "epoch": 146.12,
      "grad_norm": 1.8382846117019653,
      "learning_rate": 8.703238961178433e-07,
      "loss": 0.4532,
      "step": 356960
    },
    {
      "epoch": 146.12,
      "grad_norm": 1.95906662940979,
      "learning_rate": 8.702005366532749e-07,
      "loss": 0.4407,
      "step": 356970
    },
    {
      "epoch": 146.12,
      "grad_norm": 2.2696943283081055,
      "learning_rate": 8.700771839388084e-07,
      "loss": 0.4395,
      "step": 356980
    },
    {
      "epoch": 146.13,
      "grad_norm": 1.8590648174285889,
      "learning_rate": 8.699538379750088e-07,
      "loss": 0.4507,
      "step": 356990
    },
    {
      "epoch": 146.13,
      "grad_norm": 1.6924798488616943,
      "learning_rate": 8.69830498762441e-07,
      "loss": 0.4674,
      "step": 357000
    },
    {
      "epoch": 146.14,
      "grad_norm": 1.726819396018982,
      "learning_rate": 8.697071663016704e-07,
      "loss": 0.4741,
      "step": 357010
    },
    {
      "epoch": 146.14,
      "grad_norm": 1.9066952466964722,
      "learning_rate": 8.695838405932617e-07,
      "loss": 0.4515,
      "step": 357020
    },
    {
      "epoch": 146.14,
      "grad_norm": 2.2170770168304443,
      "learning_rate": 8.694605216377797e-07,
      "loss": 0.4647,
      "step": 357030
    },
    {
      "epoch": 146.15,
      "grad_norm": 2.3121519088745117,
      "learning_rate": 8.693372094357897e-07,
      "loss": 0.4341,
      "step": 357040
    },
    {
      "epoch": 146.15,
      "grad_norm": 1.8182048797607422,
      "learning_rate": 8.692139039878556e-07,
      "loss": 0.449,
      "step": 357050
    },
    {
      "epoch": 146.16,
      "grad_norm": 1.8993885517120361,
      "learning_rate": 8.690906052945434e-07,
      "loss": 0.4506,
      "step": 357060
    },
    {
      "epoch": 146.16,
      "grad_norm": 2.1658828258514404,
      "learning_rate": 8.689673133564179e-07,
      "loss": 0.4325,
      "step": 357070
    },
    {
      "epoch": 146.16,
      "grad_norm": 1.9420808553695679,
      "learning_rate": 8.688440281740441e-07,
      "loss": 0.4499,
      "step": 357080
    },
    {
      "epoch": 146.17,
      "grad_norm": 1.7905935049057007,
      "learning_rate": 8.687207497479855e-07,
      "loss": 0.4533,
      "step": 357090
    },
    {
      "epoch": 146.17,
      "grad_norm": 1.9739115238189697,
      "learning_rate": 8.685974780788074e-07,
      "loss": 0.4148,
      "step": 357100
    },
    {
      "epoch": 146.18,
      "grad_norm": 1.8671455383300781,
      "learning_rate": 8.684742131670745e-07,
      "loss": 0.4578,
      "step": 357110
    },
    {
      "epoch": 146.18,
      "grad_norm": 1.978820562362671,
      "learning_rate": 8.683509550133516e-07,
      "loss": 0.4381,
      "step": 357120
    },
    {
      "epoch": 146.19,
      "grad_norm": 2.1121201515197754,
      "learning_rate": 8.68227703618203e-07,
      "loss": 0.4389,
      "step": 357130
    },
    {
      "epoch": 146.19,
      "grad_norm": 1.7881412506103516,
      "learning_rate": 8.681044589821935e-07,
      "loss": 0.4427,
      "step": 357140
    },
    {
      "epoch": 146.19,
      "grad_norm": 2.0464017391204834,
      "learning_rate": 8.67981221105888e-07,
      "loss": 0.4754,
      "step": 357150
    },
    {
      "epoch": 146.2,
      "grad_norm": 3.4139480590820312,
      "learning_rate": 8.678579899898506e-07,
      "loss": 0.4593,
      "step": 357160
    },
    {
      "epoch": 146.2,
      "grad_norm": 1.6800826787948608,
      "learning_rate": 8.677347656346459e-07,
      "loss": 0.4246,
      "step": 357170
    },
    {
      "epoch": 146.21,
      "grad_norm": 2.167282819747925,
      "learning_rate": 8.676115480408382e-07,
      "loss": 0.4435,
      "step": 357180
    },
    {
      "epoch": 146.21,
      "grad_norm": 1.7923535108566284,
      "learning_rate": 8.674883372089925e-07,
      "loss": 0.4398,
      "step": 357190
    },
    {
      "epoch": 146.21,
      "grad_norm": 1.8977726697921753,
      "learning_rate": 8.673651331396727e-07,
      "loss": 0.4456,
      "step": 357200
    },
    {
      "epoch": 146.22,
      "grad_norm": 1.9949718713760376,
      "learning_rate": 8.672419358334434e-07,
      "loss": 0.4478,
      "step": 357210
    },
    {
      "epoch": 146.22,
      "grad_norm": 2.1714608669281006,
      "learning_rate": 8.671187452908686e-07,
      "loss": 0.4594,
      "step": 357220
    },
    {
      "epoch": 146.23,
      "grad_norm": 2.4280829429626465,
      "learning_rate": 8.669955615125131e-07,
      "loss": 0.4717,
      "step": 357230
    },
    {
      "epoch": 146.23,
      "grad_norm": 2.3826723098754883,
      "learning_rate": 8.668723844989409e-07,
      "loss": 0.4459,
      "step": 357240
    },
    {
      "epoch": 146.23,
      "grad_norm": 2.031952381134033,
      "learning_rate": 8.667492142507166e-07,
      "loss": 0.451,
      "step": 357250
    },
    {
      "epoch": 146.24,
      "grad_norm": 1.7228447198867798,
      "learning_rate": 8.66626050768404e-07,
      "loss": 0.4724,
      "step": 357260
    },
    {
      "epoch": 146.24,
      "grad_norm": 1.8972303867340088,
      "learning_rate": 8.665028940525672e-07,
      "loss": 0.4526,
      "step": 357270
    },
    {
      "epoch": 146.25,
      "grad_norm": 1.8538013696670532,
      "learning_rate": 8.663797441037708e-07,
      "loss": 0.4518,
      "step": 357280
    },
    {
      "epoch": 146.25,
      "grad_norm": 1.8305208683013916,
      "learning_rate": 8.662566009225788e-07,
      "loss": 0.4587,
      "step": 357290
    },
    {
      "epoch": 146.25,
      "grad_norm": 1.9698477983474731,
      "learning_rate": 8.661334645095552e-07,
      "loss": 0.4483,
      "step": 357300
    },
    {
      "epoch": 146.26,
      "grad_norm": 2.20920467376709,
      "learning_rate": 8.660103348652643e-07,
      "loss": 0.4663,
      "step": 357310
    },
    {
      "epoch": 146.26,
      "grad_norm": 1.965902328491211,
      "learning_rate": 8.658872119902697e-07,
      "loss": 0.44,
      "step": 357320
    },
    {
      "epoch": 146.27,
      "grad_norm": 2.24700927734375,
      "learning_rate": 8.657640958851356e-07,
      "loss": 0.4403,
      "step": 357330
    },
    {
      "epoch": 146.27,
      "grad_norm": 2.146332263946533,
      "learning_rate": 8.656409865504262e-07,
      "loss": 0.4427,
      "step": 357340
    },
    {
      "epoch": 146.28,
      "grad_norm": 2.2424967288970947,
      "learning_rate": 8.655178839867052e-07,
      "loss": 0.4701,
      "step": 357350
    },
    {
      "epoch": 146.28,
      "grad_norm": 1.8161609172821045,
      "learning_rate": 8.653947881945366e-07,
      "loss": 0.451,
      "step": 357360
    },
    {
      "epoch": 146.28,
      "grad_norm": 2.430846691131592,
      "learning_rate": 8.652716991744843e-07,
      "loss": 0.4647,
      "step": 357370
    },
    {
      "epoch": 146.29,
      "grad_norm": 2.4342010021209717,
      "learning_rate": 8.65148616927112e-07,
      "loss": 0.4569,
      "step": 357380
    },
    {
      "epoch": 146.29,
      "grad_norm": 2.235708475112915,
      "learning_rate": 8.650255414529848e-07,
      "loss": 0.4369,
      "step": 357390
    },
    {
      "epoch": 146.3,
      "grad_norm": 2.161574363708496,
      "learning_rate": 8.649024727526643e-07,
      "loss": 0.4343,
      "step": 357400
    },
    {
      "epoch": 146.3,
      "grad_norm": 2.1192681789398193,
      "learning_rate": 8.647794108267153e-07,
      "loss": 0.4614,
      "step": 357410
    },
    {
      "epoch": 146.3,
      "grad_norm": 1.942102074623108,
      "learning_rate": 8.646563556757015e-07,
      "loss": 0.4414,
      "step": 357420
    },
    {
      "epoch": 146.31,
      "grad_norm": 2.083747386932373,
      "learning_rate": 8.645333073001859e-07,
      "loss": 0.4608,
      "step": 357430
    },
    {
      "epoch": 146.31,
      "grad_norm": 1.6804423332214355,
      "learning_rate": 8.644102657007335e-07,
      "loss": 0.4633,
      "step": 357440
    },
    {
      "epoch": 146.32,
      "grad_norm": 1.731111764907837,
      "learning_rate": 8.642872308779074e-07,
      "loss": 0.43,
      "step": 357450
    },
    {
      "epoch": 146.32,
      "grad_norm": 2.336085796356201,
      "learning_rate": 8.641642028322713e-07,
      "loss": 0.4583,
      "step": 357460
    },
    {
      "epoch": 146.32,
      "grad_norm": 2.0600621700286865,
      "learning_rate": 8.640411815643884e-07,
      "loss": 0.4629,
      "step": 357470
    },
    {
      "epoch": 146.33,
      "grad_norm": 2.250603437423706,
      "learning_rate": 8.639181670748223e-07,
      "loss": 0.4419,
      "step": 357480
    },
    {
      "epoch": 146.33,
      "grad_norm": 22.20632553100586,
      "learning_rate": 8.63795159364137e-07,
      "loss": 0.4539,
      "step": 357490
    },
    {
      "epoch": 146.34,
      "grad_norm": 2.125622510910034,
      "learning_rate": 8.636721584328955e-07,
      "loss": 0.4355,
      "step": 357500
    },
    {
      "epoch": 146.34,
      "grad_norm": 2.082463264465332,
      "learning_rate": 8.635491642816611e-07,
      "loss": 0.4643,
      "step": 357510
    },
    {
      "epoch": 146.34,
      "grad_norm": 1.7875678539276123,
      "learning_rate": 8.63426176910998e-07,
      "loss": 0.4466,
      "step": 357520
    },
    {
      "epoch": 146.35,
      "grad_norm": 1.8431979417800903,
      "learning_rate": 8.633031963214693e-07,
      "loss": 0.4535,
      "step": 357530
    },
    {
      "epoch": 146.35,
      "grad_norm": 2.085211753845215,
      "learning_rate": 8.631802225136378e-07,
      "loss": 0.4631,
      "step": 357540
    },
    {
      "epoch": 146.36,
      "grad_norm": 1.8222779035568237,
      "learning_rate": 8.630572554880668e-07,
      "loss": 0.4433,
      "step": 357550
    },
    {
      "epoch": 146.36,
      "grad_norm": 1.957972764968872,
      "learning_rate": 8.629342952453204e-07,
      "loss": 0.4491,
      "step": 357560
    },
    {
      "epoch": 146.37,
      "grad_norm": 2.2751998901367188,
      "learning_rate": 8.628113417859611e-07,
      "loss": 0.4322,
      "step": 357570
    },
    {
      "epoch": 146.37,
      "grad_norm": 2.142026901245117,
      "learning_rate": 8.626883951105525e-07,
      "loss": 0.4737,
      "step": 357580
    },
    {
      "epoch": 146.37,
      "grad_norm": 2.0466625690460205,
      "learning_rate": 8.625654552196576e-07,
      "loss": 0.458,
      "step": 357590
    },
    {
      "epoch": 146.38,
      "grad_norm": 1.9795968532562256,
      "learning_rate": 8.624425221138399e-07,
      "loss": 0.4367,
      "step": 357600
    },
    {
      "epoch": 146.38,
      "grad_norm": 2.2422101497650146,
      "learning_rate": 8.623195957936623e-07,
      "loss": 0.4549,
      "step": 357610
    },
    {
      "epoch": 146.39,
      "grad_norm": 2.205453395843506,
      "learning_rate": 8.62196676259687e-07,
      "loss": 0.4739,
      "step": 357620
    },
    {
      "epoch": 146.39,
      "grad_norm": 1.9675655364990234,
      "learning_rate": 8.620737635124792e-07,
      "loss": 0.4691,
      "step": 357630
    },
    {
      "epoch": 146.39,
      "grad_norm": 2.5692131519317627,
      "learning_rate": 8.619508575526004e-07,
      "loss": 0.4339,
      "step": 357640
    },
    {
      "epoch": 146.4,
      "grad_norm": 1.9247902631759644,
      "learning_rate": 8.618279583806142e-07,
      "loss": 0.4441,
      "step": 357650
    },
    {
      "epoch": 146.4,
      "grad_norm": 2.45123028755188,
      "learning_rate": 8.617050659970841e-07,
      "loss": 0.4567,
      "step": 357660
    },
    {
      "epoch": 146.41,
      "grad_norm": 2.4451191425323486,
      "learning_rate": 8.615821804025714e-07,
      "loss": 0.4589,
      "step": 357670
    },
    {
      "epoch": 146.41,
      "grad_norm": 1.6775448322296143,
      "learning_rate": 8.6145930159764e-07,
      "loss": 0.436,
      "step": 357680
    },
    {
      "epoch": 146.41,
      "grad_norm": 2.418633460998535,
      "learning_rate": 8.613364295828529e-07,
      "loss": 0.4445,
      "step": 357690
    },
    {
      "epoch": 146.42,
      "grad_norm": 1.911267638206482,
      "learning_rate": 8.612135643587724e-07,
      "loss": 0.4713,
      "step": 357700
    },
    {
      "epoch": 146.42,
      "grad_norm": 1.8902016878128052,
      "learning_rate": 8.610907059259622e-07,
      "loss": 0.4469,
      "step": 357710
    },
    {
      "epoch": 146.43,
      "grad_norm": 2.190598487854004,
      "learning_rate": 8.609678542849843e-07,
      "loss": 0.4212,
      "step": 357720
    },
    {
      "epoch": 146.43,
      "grad_norm": 1.8306444883346558,
      "learning_rate": 8.608450094364017e-07,
      "loss": 0.4653,
      "step": 357730
    },
    {
      "epoch": 146.43,
      "grad_norm": 1.8203446865081787,
      "learning_rate": 8.607221713807776e-07,
      "loss": 0.4484,
      "step": 357740
    },
    {
      "epoch": 146.44,
      "grad_norm": 1.891171932220459,
      "learning_rate": 8.605993401186741e-07,
      "loss": 0.4474,
      "step": 357750
    },
    {
      "epoch": 146.44,
      "grad_norm": 1.771702766418457,
      "learning_rate": 8.604765156506539e-07,
      "loss": 0.4525,
      "step": 357760
    },
    {
      "epoch": 146.45,
      "grad_norm": 2.160186290740967,
      "learning_rate": 8.603536979772801e-07,
      "loss": 0.4429,
      "step": 357770
    },
    {
      "epoch": 146.45,
      "grad_norm": 1.9100384712219238,
      "learning_rate": 8.602308870991149e-07,
      "loss": 0.4621,
      "step": 357780
    },
    {
      "epoch": 146.46,
      "grad_norm": 2.4438562393188477,
      "learning_rate": 8.601080830167209e-07,
      "loss": 0.4577,
      "step": 357790
    },
    {
      "epoch": 146.46,
      "grad_norm": 1.9851256608963013,
      "learning_rate": 8.599852857306608e-07,
      "loss": 0.4681,
      "step": 357800
    },
    {
      "epoch": 146.46,
      "grad_norm": 2.162609100341797,
      "learning_rate": 8.59862495241497e-07,
      "loss": 0.4519,
      "step": 357810
    },
    {
      "epoch": 146.47,
      "grad_norm": 1.8674317598342896,
      "learning_rate": 8.597397115497922e-07,
      "loss": 0.4512,
      "step": 357820
    },
    {
      "epoch": 146.47,
      "grad_norm": 1.6425832509994507,
      "learning_rate": 8.596169346561087e-07,
      "loss": 0.4512,
      "step": 357830
    },
    {
      "epoch": 146.48,
      "grad_norm": 2.2155401706695557,
      "learning_rate": 8.594941645610089e-07,
      "loss": 0.4518,
      "step": 357840
    },
    {
      "epoch": 146.48,
      "grad_norm": 2.0701587200164795,
      "learning_rate": 8.593714012650553e-07,
      "loss": 0.446,
      "step": 357850
    },
    {
      "epoch": 146.48,
      "grad_norm": 1.7498382329940796,
      "learning_rate": 8.592486447688099e-07,
      "loss": 0.4431,
      "step": 357860
    },
    {
      "epoch": 146.49,
      "grad_norm": 2.1498265266418457,
      "learning_rate": 8.591258950728356e-07,
      "loss": 0.4542,
      "step": 357870
    },
    {
      "epoch": 146.49,
      "grad_norm": 2.1887869834899902,
      "learning_rate": 8.590031521776942e-07,
      "loss": 0.4413,
      "step": 357880
    },
    {
      "epoch": 146.5,
      "grad_norm": 1.8023544549942017,
      "learning_rate": 8.588804160839483e-07,
      "loss": 0.4459,
      "step": 357890
    },
    {
      "epoch": 146.5,
      "grad_norm": 2.4786899089813232,
      "learning_rate": 8.587576867921599e-07,
      "loss": 0.4426,
      "step": 357900
    },
    {
      "epoch": 146.5,
      "grad_norm": 1.7702938318252563,
      "learning_rate": 8.586349643028914e-07,
      "loss": 0.4529,
      "step": 357910
    },
    {
      "epoch": 146.51,
      "grad_norm": 2.121610403060913,
      "learning_rate": 8.58512248616705e-07,
      "loss": 0.4379,
      "step": 357920
    },
    {
      "epoch": 146.51,
      "grad_norm": 2.065509080886841,
      "learning_rate": 8.583895397341624e-07,
      "loss": 0.4576,
      "step": 357930
    },
    {
      "epoch": 146.52,
      "grad_norm": 1.8713632822036743,
      "learning_rate": 8.582668376558263e-07,
      "loss": 0.4498,
      "step": 357940
    },
    {
      "epoch": 146.52,
      "grad_norm": 1.7952083349227905,
      "learning_rate": 8.581441423822584e-07,
      "loss": 0.4712,
      "step": 357950
    },
    {
      "epoch": 146.52,
      "grad_norm": 2.1902880668640137,
      "learning_rate": 8.58021453914021e-07,
      "loss": 0.4576,
      "step": 357960
    },
    {
      "epoch": 146.53,
      "grad_norm": 1.8912326097488403,
      "learning_rate": 8.578987722516765e-07,
      "loss": 0.4345,
      "step": 357970
    },
    {
      "epoch": 146.53,
      "grad_norm": 1.9147948026657104,
      "learning_rate": 8.577760973957855e-07,
      "loss": 0.4542,
      "step": 357980
    },
    {
      "epoch": 146.54,
      "grad_norm": 2.2689931392669678,
      "learning_rate": 8.576534293469104e-07,
      "loss": 0.4588,
      "step": 357990
    },
    {
      "epoch": 146.54,
      "grad_norm": 2.068059206008911,
      "learning_rate": 8.575307681056141e-07,
      "loss": 0.4649,
      "step": 358000
    },
    {
      "epoch": 146.55,
      "grad_norm": 2.2500905990600586,
      "learning_rate": 8.574081136724581e-07,
      "loss": 0.4391,
      "step": 358010
    },
    {
      "epoch": 146.55,
      "grad_norm": 1.9379338026046753,
      "learning_rate": 8.57285466048004e-07,
      "loss": 0.4505,
      "step": 358020
    },
    {
      "epoch": 146.55,
      "grad_norm": 1.8786298036575317,
      "learning_rate": 8.571628252328137e-07,
      "loss": 0.4501,
      "step": 358030
    },
    {
      "epoch": 146.56,
      "grad_norm": 2.1836397647857666,
      "learning_rate": 8.57040191227449e-07,
      "loss": 0.4416,
      "step": 358040
    },
    {
      "epoch": 146.56,
      "grad_norm": 2.2588231563568115,
      "learning_rate": 8.569175640324715e-07,
      "loss": 0.4303,
      "step": 358050
    },
    {
      "epoch": 146.57,
      "grad_norm": 1.914281964302063,
      "learning_rate": 8.567949436484435e-07,
      "loss": 0.4738,
      "step": 358060
    },
    {
      "epoch": 146.57,
      "grad_norm": 2.109951972961426,
      "learning_rate": 8.566723300759261e-07,
      "loss": 0.4637,
      "step": 358070
    },
    {
      "epoch": 146.57,
      "grad_norm": 2.1134002208709717,
      "learning_rate": 8.565497233154814e-07,
      "loss": 0.4587,
      "step": 358080
    },
    {
      "epoch": 146.58,
      "grad_norm": 2.1800413131713867,
      "learning_rate": 8.564271233676705e-07,
      "loss": 0.4574,
      "step": 358090
    },
    {
      "epoch": 146.58,
      "grad_norm": 1.817684531211853,
      "learning_rate": 8.563045302330557e-07,
      "loss": 0.4524,
      "step": 358100
    },
    {
      "epoch": 146.59,
      "grad_norm": 2.2021374702453613,
      "learning_rate": 8.561819439121989e-07,
      "loss": 0.4518,
      "step": 358110
    },
    {
      "epoch": 146.59,
      "grad_norm": 2.194017171859741,
      "learning_rate": 8.560593644056602e-07,
      "loss": 0.4689,
      "step": 358120
    },
    {
      "epoch": 146.59,
      "grad_norm": 1.8455724716186523,
      "learning_rate": 8.559367917140017e-07,
      "loss": 0.437,
      "step": 358130
    },
    {
      "epoch": 146.6,
      "grad_norm": 2.1311209201812744,
      "learning_rate": 8.558142258377855e-07,
      "loss": 0.4615,
      "step": 358140
    },
    {
      "epoch": 146.6,
      "grad_norm": 2.409994602203369,
      "learning_rate": 8.556916667775725e-07,
      "loss": 0.442,
      "step": 358150
    },
    {
      "epoch": 146.61,
      "grad_norm": 2.019174337387085,
      "learning_rate": 8.555691145339242e-07,
      "loss": 0.4493,
      "step": 358160
    },
    {
      "epoch": 146.61,
      "grad_norm": 1.6434450149536133,
      "learning_rate": 8.55446569107402e-07,
      "loss": 0.4636,
      "step": 358170
    },
    {
      "epoch": 146.61,
      "grad_norm": 2.2562313079833984,
      "learning_rate": 8.553240304985666e-07,
      "loss": 0.4489,
      "step": 358180
    },
    {
      "epoch": 146.62,
      "grad_norm": 2.5739779472351074,
      "learning_rate": 8.552014987079808e-07,
      "loss": 0.4497,
      "step": 358190
    },
    {
      "epoch": 146.62,
      "grad_norm": 2.4915454387664795,
      "learning_rate": 8.550789737362052e-07,
      "loss": 0.4604,
      "step": 358200
    },
    {
      "epoch": 146.63,
      "grad_norm": 2.112841844558716,
      "learning_rate": 8.549564555838011e-07,
      "loss": 0.4433,
      "step": 358210
    },
    {
      "epoch": 146.63,
      "grad_norm": 1.9192845821380615,
      "learning_rate": 8.548339442513293e-07,
      "loss": 0.4723,
      "step": 358220
    },
    {
      "epoch": 146.64,
      "grad_norm": 1.6344337463378906,
      "learning_rate": 8.547114397393518e-07,
      "loss": 0.4355,
      "step": 358230
    },
    {
      "epoch": 146.64,
      "grad_norm": 1.8138560056686401,
      "learning_rate": 8.545889420484292e-07,
      "loss": 0.4511,
      "step": 358240
    },
    {
      "epoch": 146.64,
      "grad_norm": 2.3343849182128906,
      "learning_rate": 8.544664511791235e-07,
      "loss": 0.4374,
      "step": 358250
    },
    {
      "epoch": 146.65,
      "grad_norm": 1.8962666988372803,
      "learning_rate": 8.543439671319945e-07,
      "loss": 0.442,
      "step": 358260
    },
    {
      "epoch": 146.65,
      "grad_norm": 1.8621386289596558,
      "learning_rate": 8.542214899076037e-07,
      "loss": 0.4607,
      "step": 358270
    },
    {
      "epoch": 146.66,
      "grad_norm": 1.718118667602539,
      "learning_rate": 8.540990195065124e-07,
      "loss": 0.4527,
      "step": 358280
    },
    {
      "epoch": 146.66,
      "grad_norm": 1.7217696905136108,
      "learning_rate": 8.539765559292813e-07,
      "loss": 0.4499,
      "step": 358290
    },
    {
      "epoch": 146.66,
      "grad_norm": 2.1331686973571777,
      "learning_rate": 8.538540991764719e-07,
      "loss": 0.4465,
      "step": 358300
    },
    {
      "epoch": 146.67,
      "grad_norm": 1.9229779243469238,
      "learning_rate": 8.537316492486449e-07,
      "loss": 0.4572,
      "step": 358310
    },
    {
      "epoch": 146.67,
      "grad_norm": 2.4000158309936523,
      "learning_rate": 8.536092061463611e-07,
      "loss": 0.4109,
      "step": 358320
    },
    {
      "epoch": 146.68,
      "grad_norm": 1.9032752513885498,
      "learning_rate": 8.534867698701815e-07,
      "loss": 0.431,
      "step": 358330
    },
    {
      "epoch": 146.68,
      "grad_norm": 2.0117697715759277,
      "learning_rate": 8.533643404206669e-07,
      "loss": 0.4431,
      "step": 358340
    },
    {
      "epoch": 146.68,
      "grad_norm": 1.7430603504180908,
      "learning_rate": 8.532419177983782e-07,
      "loss": 0.4343,
      "step": 358350
    },
    {
      "epoch": 146.69,
      "grad_norm": 1.8518105745315552,
      "learning_rate": 8.531195020038762e-07,
      "loss": 0.4503,
      "step": 358360
    },
    {
      "epoch": 146.69,
      "grad_norm": 1.933232069015503,
      "learning_rate": 8.529970930377209e-07,
      "loss": 0.4372,
      "step": 358370
    },
    {
      "epoch": 146.7,
      "grad_norm": 2.013105869293213,
      "learning_rate": 8.528746909004754e-07,
      "loss": 0.4429,
      "step": 358380
    },
    {
      "epoch": 146.7,
      "grad_norm": 2.0300979614257812,
      "learning_rate": 8.52752295592698e-07,
      "loss": 0.439,
      "step": 358390
    },
    {
      "epoch": 146.7,
      "grad_norm": 1.8586094379425049,
      "learning_rate": 8.526299071149504e-07,
      "loss": 0.4608,
      "step": 358400
    },
    {
      "epoch": 146.71,
      "grad_norm": 2.0954530239105225,
      "learning_rate": 8.525075254677927e-07,
      "loss": 0.4436,
      "step": 358410
    },
    {
      "epoch": 146.71,
      "grad_norm": 1.816195011138916,
      "learning_rate": 8.523851506517857e-07,
      "loss": 0.4582,
      "step": 358420
    },
    {
      "epoch": 146.72,
      "grad_norm": 1.6378053426742554,
      "learning_rate": 8.522627826674905e-07,
      "loss": 0.4512,
      "step": 358430
    },
    {
      "epoch": 146.72,
      "grad_norm": 2.1605379581451416,
      "learning_rate": 8.521404215154671e-07,
      "loss": 0.4301,
      "step": 358440
    },
    {
      "epoch": 146.73,
      "grad_norm": 2.270005941390991,
      "learning_rate": 8.520180671962761e-07,
      "loss": 0.4424,
      "step": 358450
    },
    {
      "epoch": 146.73,
      "grad_norm": 2.062739849090576,
      "learning_rate": 8.518957197104781e-07,
      "loss": 0.4615,
      "step": 358460
    },
    {
      "epoch": 146.73,
      "grad_norm": 2.0588676929473877,
      "learning_rate": 8.517733790586336e-07,
      "loss": 0.4307,
      "step": 358470
    },
    {
      "epoch": 146.74,
      "grad_norm": 2.1473681926727295,
      "learning_rate": 8.516510452413027e-07,
      "loss": 0.453,
      "step": 358480
    },
    {
      "epoch": 146.74,
      "grad_norm": 2.9080679416656494,
      "learning_rate": 8.515287182590463e-07,
      "loss": 0.4473,
      "step": 358490
    },
    {
      "epoch": 146.75,
      "grad_norm": 1.8366597890853882,
      "learning_rate": 8.514063981124243e-07,
      "loss": 0.4548,
      "step": 358500
    },
    {
      "epoch": 146.75,
      "grad_norm": 1.9207777976989746,
      "learning_rate": 8.512840848019973e-07,
      "loss": 0.4589,
      "step": 358510
    },
    {
      "epoch": 146.75,
      "grad_norm": 1.983272910118103,
      "learning_rate": 8.511617783283256e-07,
      "loss": 0.4475,
      "step": 358520
    },
    {
      "epoch": 146.76,
      "grad_norm": 1.9019689559936523,
      "learning_rate": 8.510394786919694e-07,
      "loss": 0.4743,
      "step": 358530
    },
    {
      "epoch": 146.76,
      "grad_norm": 1.8041713237762451,
      "learning_rate": 8.509171858934888e-07,
      "loss": 0.4406,
      "step": 358540
    },
    {
      "epoch": 146.77,
      "grad_norm": 2.055950403213501,
      "learning_rate": 8.507948999334442e-07,
      "loss": 0.4484,
      "step": 358550
    },
    {
      "epoch": 146.77,
      "grad_norm": 2.0020978450775146,
      "learning_rate": 8.506726208123956e-07,
      "loss": 0.4535,
      "step": 358560
    },
    {
      "epoch": 146.77,
      "grad_norm": 2.2743120193481445,
      "learning_rate": 8.505503485309035e-07,
      "loss": 0.4653,
      "step": 358570
    },
    {
      "epoch": 146.78,
      "grad_norm": 2.1836159229278564,
      "learning_rate": 8.504280830895275e-07,
      "loss": 0.4583,
      "step": 358580
    },
    {
      "epoch": 146.78,
      "grad_norm": 2.315840244293213,
      "learning_rate": 8.50305824488828e-07,
      "loss": 0.4281,
      "step": 358590
    },
    {
      "epoch": 146.79,
      "grad_norm": 2.0009608268737793,
      "learning_rate": 8.501835727293649e-07,
      "loss": 0.4367,
      "step": 358600
    },
    {
      "epoch": 146.79,
      "grad_norm": 2.004485607147217,
      "learning_rate": 8.500613278116983e-07,
      "loss": 0.4704,
      "step": 358610
    },
    {
      "epoch": 146.79,
      "grad_norm": 2.0893940925598145,
      "learning_rate": 8.499390897363882e-07,
      "loss": 0.4623,
      "step": 358620
    },
    {
      "epoch": 146.8,
      "grad_norm": 2.006765127182007,
      "learning_rate": 8.498168585039947e-07,
      "loss": 0.454,
      "step": 358630
    },
    {
      "epoch": 146.8,
      "grad_norm": 1.844290852546692,
      "learning_rate": 8.496946341150774e-07,
      "loss": 0.444,
      "step": 358640
    },
    {
      "epoch": 146.81,
      "grad_norm": 13.414600372314453,
      "learning_rate": 8.495724165701965e-07,
      "loss": 0.4432,
      "step": 358650
    },
    {
      "epoch": 146.81,
      "grad_norm": 2.002739429473877,
      "learning_rate": 8.494502058699116e-07,
      "loss": 0.45,
      "step": 358660
    },
    {
      "epoch": 146.82,
      "grad_norm": 2.306105852127075,
      "learning_rate": 8.493280020147824e-07,
      "loss": 0.485,
      "step": 358670
    },
    {
      "epoch": 146.82,
      "grad_norm": 1.8914215564727783,
      "learning_rate": 8.492058050053694e-07,
      "loss": 0.464,
      "step": 358680
    },
    {
      "epoch": 146.82,
      "grad_norm": 2.1785387992858887,
      "learning_rate": 8.490836148422323e-07,
      "loss": 0.4488,
      "step": 358690
    },
    {
      "epoch": 146.83,
      "grad_norm": 2.1480634212493896,
      "learning_rate": 8.489614315259297e-07,
      "loss": 0.437,
      "step": 358700
    },
    {
      "epoch": 146.83,
      "grad_norm": 1.8632590770721436,
      "learning_rate": 8.488392550570223e-07,
      "loss": 0.4568,
      "step": 358710
    },
    {
      "epoch": 146.84,
      "grad_norm": 2.0286576747894287,
      "learning_rate": 8.487170854360692e-07,
      "loss": 0.438,
      "step": 358720
    },
    {
      "epoch": 146.84,
      "grad_norm": 1.9068799018859863,
      "learning_rate": 8.485949226636306e-07,
      "loss": 0.4451,
      "step": 358730
    },
    {
      "epoch": 146.84,
      "grad_norm": 1.9211597442626953,
      "learning_rate": 8.484727667402657e-07,
      "loss": 0.4441,
      "step": 358740
    },
    {
      "epoch": 146.85,
      "grad_norm": 2.2284860610961914,
      "learning_rate": 8.483506176665334e-07,
      "loss": 0.4596,
      "step": 358750
    },
    {
      "epoch": 146.85,
      "grad_norm": 2.3815157413482666,
      "learning_rate": 8.48228475442995e-07,
      "loss": 0.4621,
      "step": 358760
    },
    {
      "epoch": 146.86,
      "grad_norm": 2.061600685119629,
      "learning_rate": 8.48106340070209e-07,
      "loss": 0.4338,
      "step": 358770
    },
    {
      "epoch": 146.86,
      "grad_norm": 1.9671381711959839,
      "learning_rate": 8.47984211548735e-07,
      "loss": 0.4645,
      "step": 358780
    },
    {
      "epoch": 146.86,
      "grad_norm": 2.236729383468628,
      "learning_rate": 8.478620898791324e-07,
      "loss": 0.4471,
      "step": 358790
    },
    {
      "epoch": 146.87,
      "grad_norm": 1.9649584293365479,
      "learning_rate": 8.477399750619608e-07,
      "loss": 0.4411,
      "step": 358800
    },
    {
      "epoch": 146.87,
      "grad_norm": 2.376594066619873,
      "learning_rate": 8.476178670977793e-07,
      "loss": 0.4717,
      "step": 358810
    },
    {
      "epoch": 146.88,
      "grad_norm": 2.100553035736084,
      "learning_rate": 8.474957659871475e-07,
      "loss": 0.4522,
      "step": 358820
    },
    {
      "epoch": 146.88,
      "grad_norm": 2.1142737865448,
      "learning_rate": 8.473736717306252e-07,
      "loss": 0.4714,
      "step": 358830
    },
    {
      "epoch": 146.88,
      "grad_norm": 2.0871052742004395,
      "learning_rate": 8.472515843287707e-07,
      "loss": 0.4565,
      "step": 358840
    },
    {
      "epoch": 146.89,
      "grad_norm": 2.139420747756958,
      "learning_rate": 8.471295037821434e-07,
      "loss": 0.4448,
      "step": 358850
    },
    {
      "epoch": 146.89,
      "grad_norm": 2.109499216079712,
      "learning_rate": 8.470074300913027e-07,
      "loss": 0.4633,
      "step": 358860
    },
    {
      "epoch": 146.9,
      "grad_norm": 2.0811684131622314,
      "learning_rate": 8.468853632568083e-07,
      "loss": 0.4572,
      "step": 358870
    },
    {
      "epoch": 146.9,
      "grad_norm": 2.072448492050171,
      "learning_rate": 8.467633032792187e-07,
      "loss": 0.4644,
      "step": 358880
    },
    {
      "epoch": 146.91,
      "grad_norm": 2.4087557792663574,
      "learning_rate": 8.466412501590933e-07,
      "loss": 0.4436,
      "step": 358890
    },
    {
      "epoch": 146.91,
      "grad_norm": 1.9085885286331177,
      "learning_rate": 8.465192038969913e-07,
      "loss": 0.43,
      "step": 358900
    },
    {
      "epoch": 146.91,
      "grad_norm": 2.3094801902770996,
      "learning_rate": 8.463971644934718e-07,
      "loss": 0.4691,
      "step": 358910
    },
    {
      "epoch": 146.92,
      "grad_norm": 2.0120420455932617,
      "learning_rate": 8.462751319490936e-07,
      "loss": 0.4452,
      "step": 358920
    },
    {
      "epoch": 146.92,
      "grad_norm": 2.2785472869873047,
      "learning_rate": 8.461531062644159e-07,
      "loss": 0.4623,
      "step": 358930
    },
    {
      "epoch": 146.93,
      "grad_norm": 1.9677789211273193,
      "learning_rate": 8.460310874399968e-07,
      "loss": 0.4468,
      "step": 358940
    },
    {
      "epoch": 146.93,
      "grad_norm": 1.879237413406372,
      "learning_rate": 8.459090754763971e-07,
      "loss": 0.4276,
      "step": 358950
    },
    {
      "epoch": 146.93,
      "grad_norm": 2.038119077682495,
      "learning_rate": 8.457870703741745e-07,
      "loss": 0.4521,
      "step": 358960
    },
    {
      "epoch": 146.94,
      "grad_norm": 1.9886494874954224,
      "learning_rate": 8.45665072133889e-07,
      "loss": 0.4304,
      "step": 358970
    },
    {
      "epoch": 146.94,
      "grad_norm": 2.2701261043548584,
      "learning_rate": 8.455430807560976e-07,
      "loss": 0.4482,
      "step": 358980
    },
    {
      "epoch": 146.95,
      "grad_norm": 2.076174736022949,
      "learning_rate": 8.454210962413601e-07,
      "loss": 0.4503,
      "step": 358990
    },
    {
      "epoch": 146.95,
      "grad_norm": 1.8571594953536987,
      "learning_rate": 8.452991185902354e-07,
      "loss": 0.4758,
      "step": 359000
    },
    {
      "epoch": 146.95,
      "grad_norm": 2.157590627670288,
      "learning_rate": 8.451771478032819e-07,
      "loss": 0.4547,
      "step": 359010
    },
    {
      "epoch": 146.96,
      "grad_norm": 2.2636654376983643,
      "learning_rate": 8.450551838810588e-07,
      "loss": 0.4395,
      "step": 359020
    },
    {
      "epoch": 146.96,
      "grad_norm": 1.7184385061264038,
      "learning_rate": 8.449332268241245e-07,
      "loss": 0.4449,
      "step": 359030
    },
    {
      "epoch": 146.97,
      "grad_norm": 1.7313034534454346,
      "learning_rate": 8.448112766330377e-07,
      "loss": 0.4474,
      "step": 359040
    },
    {
      "epoch": 146.97,
      "grad_norm": 2.358966827392578,
      "learning_rate": 8.446893333083567e-07,
      "loss": 0.4315,
      "step": 359050
    },
    {
      "epoch": 146.98,
      "grad_norm": 2.3842689990997314,
      "learning_rate": 8.445673968506409e-07,
      "loss": 0.4466,
      "step": 359060
    },
    {
      "epoch": 146.98,
      "grad_norm": 2.2901134490966797,
      "learning_rate": 8.444454672604482e-07,
      "loss": 0.442,
      "step": 359070
    },
    {
      "epoch": 146.98,
      "grad_norm": 2.2571160793304443,
      "learning_rate": 8.443235445383375e-07,
      "loss": 0.453,
      "step": 359080
    },
    {
      "epoch": 146.99,
      "grad_norm": 1.96663498878479,
      "learning_rate": 8.442016286848673e-07,
      "loss": 0.4427,
      "step": 359090
    },
    {
      "epoch": 146.99,
      "grad_norm": 2.276270866394043,
      "learning_rate": 8.440797197005958e-07,
      "loss": 0.4722,
      "step": 359100
    },
    {
      "epoch": 147.0,
      "grad_norm": 1.895693302154541,
      "learning_rate": 8.439578175860813e-07,
      "loss": 0.4445,
      "step": 359110
    },
    {
      "epoch": 147.0,
      "grad_norm": 1.850555658340454,
      "learning_rate": 8.43835922341883e-07,
      "loss": 0.4583,
      "step": 359120
    },
    {
      "epoch": 147.0,
      "eval_loss": 0.4520080089569092,
      "eval_runtime": 53.1619,
      "eval_samples_per_second": 64.877,
      "eval_steps_per_second": 8.126,
      "step": 359121
    },
    {
      "epoch": 147.0,
      "grad_norm": 1.9800610542297363,
      "learning_rate": 8.437140339685585e-07,
      "loss": 0.4461,
      "step": 359130
    },
    {
      "epoch": 147.01,
      "grad_norm": 2.2005460262298584,
      "learning_rate": 8.435921524666667e-07,
      "loss": 0.4638,
      "step": 359140
    },
    {
      "epoch": 147.01,
      "grad_norm": 2.1621391773223877,
      "learning_rate": 8.434702778367656e-07,
      "loss": 0.45,
      "step": 359150
    },
    {
      "epoch": 147.02,
      "grad_norm": 2.12453556060791,
      "learning_rate": 8.433484100794134e-07,
      "loss": 0.4532,
      "step": 359160
    },
    {
      "epoch": 147.02,
      "grad_norm": 2.36242413520813,
      "learning_rate": 8.432265491951687e-07,
      "loss": 0.4484,
      "step": 359170
    },
    {
      "epoch": 147.02,
      "grad_norm": 1.9807920455932617,
      "learning_rate": 8.431046951845895e-07,
      "loss": 0.4485,
      "step": 359180
    },
    {
      "epoch": 147.03,
      "grad_norm": 1.8558772802352905,
      "learning_rate": 8.429828480482337e-07,
      "loss": 0.4597,
      "step": 359190
    },
    {
      "epoch": 147.03,
      "grad_norm": 2.130384922027588,
      "learning_rate": 8.428610077866601e-07,
      "loss": 0.4568,
      "step": 359200
    },
    {
      "epoch": 147.04,
      "grad_norm": 1.9903463125228882,
      "learning_rate": 8.427391744004266e-07,
      "loss": 0.4286,
      "step": 359210
    },
    {
      "epoch": 147.04,
      "grad_norm": 2.094633102416992,
      "learning_rate": 8.42617347890091e-07,
      "loss": 0.447,
      "step": 359220
    },
    {
      "epoch": 147.04,
      "grad_norm": 2.0902483463287354,
      "learning_rate": 8.424955282562118e-07,
      "loss": 0.4545,
      "step": 359230
    },
    {
      "epoch": 147.05,
      "grad_norm": 2.11916184425354,
      "learning_rate": 8.423737154993468e-07,
      "loss": 0.4695,
      "step": 359240
    },
    {
      "epoch": 147.05,
      "grad_norm": 1.9323627948760986,
      "learning_rate": 8.42251909620054e-07,
      "loss": 0.43,
      "step": 359250
    },
    {
      "epoch": 147.06,
      "grad_norm": 1.969778299331665,
      "learning_rate": 8.421301106188915e-07,
      "loss": 0.4457,
      "step": 359260
    },
    {
      "epoch": 147.06,
      "grad_norm": 1.894108533859253,
      "learning_rate": 8.42008318496417e-07,
      "loss": 0.4374,
      "step": 359270
    },
    {
      "epoch": 147.07,
      "grad_norm": 1.7620819807052612,
      "learning_rate": 8.418865332531896e-07,
      "loss": 0.4493,
      "step": 359280
    },
    {
      "epoch": 147.07,
      "grad_norm": 1.9704574346542358,
      "learning_rate": 8.417647548897653e-07,
      "loss": 0.4843,
      "step": 359290
    },
    {
      "epoch": 147.07,
      "grad_norm": 2.5884625911712646,
      "learning_rate": 8.416429834067026e-07,
      "loss": 0.4257,
      "step": 359300
    },
    {
      "epoch": 147.08,
      "grad_norm": 2.3365492820739746,
      "learning_rate": 8.415212188045587e-07,
      "loss": 0.449,
      "step": 359310
    },
    {
      "epoch": 147.08,
      "grad_norm": 2.1480698585510254,
      "learning_rate": 8.413994610838933e-07,
      "loss": 0.4526,
      "step": 359320
    },
    {
      "epoch": 147.09,
      "grad_norm": 2.117467164993286,
      "learning_rate": 8.412777102452629e-07,
      "loss": 0.4518,
      "step": 359330
    },
    {
      "epoch": 147.09,
      "grad_norm": 1.8370180130004883,
      "learning_rate": 8.411559662892251e-07,
      "loss": 0.4604,
      "step": 359340
    },
    {
      "epoch": 147.09,
      "grad_norm": 1.5517842769622803,
      "learning_rate": 8.410342292163383e-07,
      "loss": 0.4433,
      "step": 359350
    },
    {
      "epoch": 147.1,
      "grad_norm": 2.204552173614502,
      "learning_rate": 8.409124990271596e-07,
      "loss": 0.4319,
      "step": 359360
    },
    {
      "epoch": 147.1,
      "grad_norm": 1.817307949066162,
      "learning_rate": 8.407907757222466e-07,
      "loss": 0.4618,
      "step": 359370
    },
    {
      "epoch": 147.11,
      "grad_norm": 1.6413134336471558,
      "learning_rate": 8.406690593021571e-07,
      "loss": 0.4499,
      "step": 359380
    },
    {
      "epoch": 147.11,
      "grad_norm": 1.9294596910476685,
      "learning_rate": 8.405473497674485e-07,
      "loss": 0.4692,
      "step": 359390
    },
    {
      "epoch": 147.11,
      "grad_norm": 1.7406748533248901,
      "learning_rate": 8.404256471186785e-07,
      "loss": 0.4414,
      "step": 359400
    },
    {
      "epoch": 147.12,
      "grad_norm": 2.467088460922241,
      "learning_rate": 8.403039513564055e-07,
      "loss": 0.4449,
      "step": 359410
    },
    {
      "epoch": 147.12,
      "grad_norm": 2.004844903945923,
      "learning_rate": 8.401822624811853e-07,
      "loss": 0.4433,
      "step": 359420
    },
    {
      "epoch": 147.13,
      "grad_norm": 1.9276511669158936,
      "learning_rate": 8.400605804935758e-07,
      "loss": 0.4623,
      "step": 359430
    },
    {
      "epoch": 147.13,
      "grad_norm": 2.2116336822509766,
      "learning_rate": 8.399389053941346e-07,
      "loss": 0.4389,
      "step": 359440
    },
    {
      "epoch": 147.13,
      "grad_norm": 1.8843401670455933,
      "learning_rate": 8.398172371834193e-07,
      "loss": 0.4432,
      "step": 359450
    },
    {
      "epoch": 147.14,
      "grad_norm": 1.9554449319839478,
      "learning_rate": 8.39695575861987e-07,
      "loss": 0.4378,
      "step": 359460
    },
    {
      "epoch": 147.14,
      "grad_norm": 2.1537599563598633,
      "learning_rate": 8.395739214303953e-07,
      "loss": 0.4608,
      "step": 359470
    },
    {
      "epoch": 147.15,
      "grad_norm": 1.7178378105163574,
      "learning_rate": 8.394522738892008e-07,
      "loss": 0.4537,
      "step": 359480
    },
    {
      "epoch": 147.15,
      "grad_norm": 2.1483218669891357,
      "learning_rate": 8.393306332389617e-07,
      "loss": 0.4413,
      "step": 359490
    },
    {
      "epoch": 147.16,
      "grad_norm": 1.939093828201294,
      "learning_rate": 8.392089994802336e-07,
      "loss": 0.4762,
      "step": 359500
    },
    {
      "epoch": 147.16,
      "grad_norm": 1.9503309726715088,
      "learning_rate": 8.390873726135757e-07,
      "loss": 0.4429,
      "step": 359510
    },
    {
      "epoch": 147.16,
      "grad_norm": 1.741204023361206,
      "learning_rate": 8.389657526395444e-07,
      "loss": 0.4474,
      "step": 359520
    },
    {
      "epoch": 147.17,
      "grad_norm": 2.6191885471343994,
      "learning_rate": 8.388441395586965e-07,
      "loss": 0.4666,
      "step": 359530
    },
    {
      "epoch": 147.17,
      "grad_norm": 2.1651530265808105,
      "learning_rate": 8.387225333715894e-07,
      "loss": 0.4451,
      "step": 359540
    },
    {
      "epoch": 147.18,
      "grad_norm": 1.8966761827468872,
      "learning_rate": 8.386009340787807e-07,
      "loss": 0.4517,
      "step": 359550
    },
    {
      "epoch": 147.18,
      "grad_norm": 1.8112481832504272,
      "learning_rate": 8.384793416808264e-07,
      "loss": 0.4449,
      "step": 359560
    },
    {
      "epoch": 147.18,
      "grad_norm": 1.6663192510604858,
      "learning_rate": 8.383577561782835e-07,
      "loss": 0.4438,
      "step": 359570
    },
    {
      "epoch": 147.19,
      "grad_norm": 2.319850444793701,
      "learning_rate": 8.382361775717093e-07,
      "loss": 0.471,
      "step": 359580
    },
    {
      "epoch": 147.19,
      "grad_norm": 1.8812851905822754,
      "learning_rate": 8.381146058616609e-07,
      "loss": 0.4399,
      "step": 359590
    },
    {
      "epoch": 147.2,
      "grad_norm": 2.3320562839508057,
      "learning_rate": 8.379930410486949e-07,
      "loss": 0.4667,
      "step": 359600
    },
    {
      "epoch": 147.2,
      "grad_norm": 1.8768868446350098,
      "learning_rate": 8.378714831333684e-07,
      "loss": 0.4677,
      "step": 359610
    },
    {
      "epoch": 147.2,
      "grad_norm": 1.8452308177947998,
      "learning_rate": 8.377499321162382e-07,
      "loss": 0.4509,
      "step": 359620
    },
    {
      "epoch": 147.21,
      "grad_norm": 1.6787539720535278,
      "learning_rate": 8.376283879978612e-07,
      "loss": 0.4693,
      "step": 359630
    },
    {
      "epoch": 147.21,
      "grad_norm": 1.9930235147476196,
      "learning_rate": 8.375068507787941e-07,
      "loss": 0.4519,
      "step": 359640
    },
    {
      "epoch": 147.22,
      "grad_norm": 2.19535493850708,
      "learning_rate": 8.373853204595932e-07,
      "loss": 0.4476,
      "step": 359650
    },
    {
      "epoch": 147.22,
      "grad_norm": 2.1508381366729736,
      "learning_rate": 8.372637970408161e-07,
      "loss": 0.46,
      "step": 359660
    },
    {
      "epoch": 147.22,
      "grad_norm": 1.8082324266433716,
      "learning_rate": 8.371422805230188e-07,
      "loss": 0.4548,
      "step": 359670
    },
    {
      "epoch": 147.23,
      "grad_norm": 1.8477604389190674,
      "learning_rate": 8.370207709067577e-07,
      "loss": 0.4267,
      "step": 359680
    },
    {
      "epoch": 147.23,
      "grad_norm": 2.0221939086914062,
      "learning_rate": 8.368992681925905e-07,
      "loss": 0.4457,
      "step": 359690
    },
    {
      "epoch": 147.24,
      "grad_norm": 2.2120444774627686,
      "learning_rate": 8.367777723810728e-07,
      "loss": 0.4671,
      "step": 359700
    },
    {
      "epoch": 147.24,
      "grad_norm": 2.0603506565093994,
      "learning_rate": 8.366562834727616e-07,
      "loss": 0.4513,
      "step": 359710
    },
    {
      "epoch": 147.25,
      "grad_norm": 1.6976397037506104,
      "learning_rate": 8.365348014682132e-07,
      "loss": 0.4444,
      "step": 359720
    },
    {
      "epoch": 147.25,
      "grad_norm": 1.7099905014038086,
      "learning_rate": 8.364133263679843e-07,
      "loss": 0.4414,
      "step": 359730
    },
    {
      "epoch": 147.25,
      "grad_norm": 1.9932103157043457,
      "learning_rate": 8.362918581726311e-07,
      "loss": 0.424,
      "step": 359740
    },
    {
      "epoch": 147.26,
      "grad_norm": 1.9643769264221191,
      "learning_rate": 8.361703968827104e-07,
      "loss": 0.431,
      "step": 359750
    },
    {
      "epoch": 147.26,
      "grad_norm": 2.2392356395721436,
      "learning_rate": 8.360489424987783e-07,
      "loss": 0.4553,
      "step": 359760
    },
    {
      "epoch": 147.27,
      "grad_norm": 1.8129336833953857,
      "learning_rate": 8.359274950213914e-07,
      "loss": 0.4616,
      "step": 359770
    },
    {
      "epoch": 147.27,
      "grad_norm": 2.256927013397217,
      "learning_rate": 8.358060544511056e-07,
      "loss": 0.4489,
      "step": 359780
    },
    {
      "epoch": 147.27,
      "grad_norm": 2.187368869781494,
      "learning_rate": 8.356846207884776e-07,
      "loss": 0.4549,
      "step": 359790
    },
    {
      "epoch": 147.28,
      "grad_norm": 2.0267794132232666,
      "learning_rate": 8.355631940340634e-07,
      "loss": 0.4486,
      "step": 359800
    },
    {
      "epoch": 147.28,
      "grad_norm": 2.8098325729370117,
      "learning_rate": 8.354417741884197e-07,
      "loss": 0.4444,
      "step": 359810
    },
    {
      "epoch": 147.29,
      "grad_norm": 1.9588234424591064,
      "learning_rate": 8.353203612521022e-07,
      "loss": 0.455,
      "step": 359820
    },
    {
      "epoch": 147.29,
      "grad_norm": 1.9930065870285034,
      "learning_rate": 8.351989552256672e-07,
      "loss": 0.4388,
      "step": 359830
    },
    {
      "epoch": 147.29,
      "grad_norm": 1.6877167224884033,
      "learning_rate": 8.350775561096712e-07,
      "loss": 0.4526,
      "step": 359840
    },
    {
      "epoch": 147.3,
      "grad_norm": 2.4361863136291504,
      "learning_rate": 8.349561639046696e-07,
      "loss": 0.4304,
      "step": 359850
    },
    {
      "epoch": 147.3,
      "grad_norm": 1.9008342027664185,
      "learning_rate": 8.3483477861122e-07,
      "loss": 0.4647,
      "step": 359860
    },
    {
      "epoch": 147.31,
      "grad_norm": 2.2537858486175537,
      "learning_rate": 8.347134002298765e-07,
      "loss": 0.4593,
      "step": 359870
    },
    {
      "epoch": 147.31,
      "grad_norm": 2.1879031658172607,
      "learning_rate": 8.345920287611952e-07,
      "loss": 0.4538,
      "step": 359880
    },
    {
      "epoch": 147.31,
      "grad_norm": 1.813749074935913,
      "learning_rate": 8.344706642057335e-07,
      "loss": 0.4492,
      "step": 359890
    },
    {
      "epoch": 147.32,
      "grad_norm": 1.977092981338501,
      "learning_rate": 8.343493065640469e-07,
      "loss": 0.4489,
      "step": 359900
    },
    {
      "epoch": 147.32,
      "grad_norm": 2.076887607574463,
      "learning_rate": 8.342279558366907e-07,
      "loss": 0.4453,
      "step": 359910
    },
    {
      "epoch": 147.33,
      "grad_norm": 1.8441178798675537,
      "learning_rate": 8.341066120242216e-07,
      "loss": 0.4292,
      "step": 359920
    },
    {
      "epoch": 147.33,
      "grad_norm": 2.073683023452759,
      "learning_rate": 8.339852751271952e-07,
      "loss": 0.4527,
      "step": 359930
    },
    {
      "epoch": 147.34,
      "grad_norm": 1.9429789781570435,
      "learning_rate": 8.338639451461669e-07,
      "loss": 0.4715,
      "step": 359940
    },
    {
      "epoch": 147.34,
      "grad_norm": 2.051461696624756,
      "learning_rate": 8.337426220816927e-07,
      "loss": 0.4555,
      "step": 359950
    },
    {
      "epoch": 147.34,
      "grad_norm": 1.8877936601638794,
      "learning_rate": 8.336213059343286e-07,
      "loss": 0.4478,
      "step": 359960
    },
    {
      "epoch": 147.35,
      "grad_norm": 1.8557242155075073,
      "learning_rate": 8.334999967046302e-07,
      "loss": 0.461,
      "step": 359970
    },
    {
      "epoch": 147.35,
      "grad_norm": 1.953287124633789,
      "learning_rate": 8.333786943931529e-07,
      "loss": 0.4482,
      "step": 359980
    },
    {
      "epoch": 147.36,
      "grad_norm": 1.8377634286880493,
      "learning_rate": 8.332573990004538e-07,
      "loss": 0.4381,
      "step": 359990
    },
    {
      "epoch": 147.36,
      "grad_norm": 1.712363362312317,
      "learning_rate": 8.331361105270864e-07,
      "loss": 0.4651,
      "step": 360000
    },
    {
      "epoch": 147.36,
      "grad_norm": 1.9320365190505981,
      "learning_rate": 8.330148289736074e-07,
      "loss": 0.4414,
      "step": 360010
    },
    {
      "epoch": 147.37,
      "grad_norm": 1.7438637018203735,
      "learning_rate": 8.32893554340572e-07,
      "loss": 0.4748,
      "step": 360020
    },
    {
      "epoch": 147.37,
      "grad_norm": 2.3553009033203125,
      "learning_rate": 8.327722866285361e-07,
      "loss": 0.4482,
      "step": 360030
    },
    {
      "epoch": 147.38,
      "grad_norm": 1.8239216804504395,
      "learning_rate": 8.326510258380551e-07,
      "loss": 0.4444,
      "step": 360040
    },
    {
      "epoch": 147.38,
      "grad_norm": 1.7585532665252686,
      "learning_rate": 8.325297719696847e-07,
      "loss": 0.4533,
      "step": 360050
    },
    {
      "epoch": 147.38,
      "grad_norm": 2.0166923999786377,
      "learning_rate": 8.324085250239799e-07,
      "loss": 0.4559,
      "step": 360060
    },
    {
      "epoch": 147.39,
      "grad_norm": 2.324183225631714,
      "learning_rate": 8.322872850014956e-07,
      "loss": 0.4654,
      "step": 360070
    },
    {
      "epoch": 147.39,
      "grad_norm": 2.040073871612549,
      "learning_rate": 8.321660519027888e-07,
      "loss": 0.4422,
      "step": 360080
    },
    {
      "epoch": 147.4,
      "grad_norm": 1.5992172956466675,
      "learning_rate": 8.320448257284137e-07,
      "loss": 0.4446,
      "step": 360090
    },
    {
      "epoch": 147.4,
      "grad_norm": 1.7858803272247314,
      "learning_rate": 8.319236064789261e-07,
      "loss": 0.4527,
      "step": 360100
    },
    {
      "epoch": 147.4,
      "grad_norm": 1.9849594831466675,
      "learning_rate": 8.318023941548812e-07,
      "loss": 0.464,
      "step": 360110
    },
    {
      "epoch": 147.41,
      "grad_norm": 1.8637481927871704,
      "learning_rate": 8.316811887568339e-07,
      "loss": 0.462,
      "step": 360120
    },
    {
      "epoch": 147.41,
      "grad_norm": 1.690961480140686,
      "learning_rate": 8.315599902853404e-07,
      "loss": 0.4621,
      "step": 360130
    },
    {
      "epoch": 147.42,
      "grad_norm": 1.814247727394104,
      "learning_rate": 8.314387987409545e-07,
      "loss": 0.4637,
      "step": 360140
    },
    {
      "epoch": 147.42,
      "grad_norm": 2.3772661685943604,
      "learning_rate": 8.31317614124232e-07,
      "loss": 0.4323,
      "step": 360150
    },
    {
      "epoch": 147.43,
      "grad_norm": 2.0485103130340576,
      "learning_rate": 8.311964364357279e-07,
      "loss": 0.4521,
      "step": 360160
    },
    {
      "epoch": 147.43,
      "grad_norm": 2.589003801345825,
      "learning_rate": 8.310752656759975e-07,
      "loss": 0.4729,
      "step": 360170
    },
    {
      "epoch": 147.43,
      "grad_norm": 1.6931103467941284,
      "learning_rate": 8.30954101845596e-07,
      "loss": 0.4309,
      "step": 360180
    },
    {
      "epoch": 147.44,
      "grad_norm": 2.1648590564727783,
      "learning_rate": 8.30832944945078e-07,
      "loss": 0.4688,
      "step": 360190
    },
    {
      "epoch": 147.44,
      "grad_norm": 1.9712646007537842,
      "learning_rate": 8.307117949749986e-07,
      "loss": 0.4449,
      "step": 360200
    },
    {
      "epoch": 147.45,
      "grad_norm": 2.232450008392334,
      "learning_rate": 8.305906519359131e-07,
      "loss": 0.4252,
      "step": 360210
    },
    {
      "epoch": 147.45,
      "grad_norm": 1.7900913953781128,
      "learning_rate": 8.304695158283763e-07,
      "loss": 0.4287,
      "step": 360220
    },
    {
      "epoch": 147.45,
      "grad_norm": 2.0306928157806396,
      "learning_rate": 8.303483866529429e-07,
      "loss": 0.4396,
      "step": 360230
    },
    {
      "epoch": 147.46,
      "grad_norm": 2.062002658843994,
      "learning_rate": 8.302272644101678e-07,
      "loss": 0.4602,
      "step": 360240
    },
    {
      "epoch": 147.46,
      "grad_norm": 2.320669651031494,
      "learning_rate": 8.301061491006061e-07,
      "loss": 0.4461,
      "step": 360250
    },
    {
      "epoch": 147.47,
      "grad_norm": 1.669280767440796,
      "learning_rate": 8.299850407248117e-07,
      "loss": 0.4355,
      "step": 360260
    },
    {
      "epoch": 147.47,
      "grad_norm": 1.8756015300750732,
      "learning_rate": 8.298639392833416e-07,
      "loss": 0.4436,
      "step": 360270
    },
    {
      "epoch": 147.47,
      "grad_norm": 1.8776659965515137,
      "learning_rate": 8.297428447767483e-07,
      "loss": 0.4493,
      "step": 360280
    },
    {
      "epoch": 147.48,
      "grad_norm": 1.8218469619750977,
      "learning_rate": 8.296217572055876e-07,
      "loss": 0.4466,
      "step": 360290
    },
    {
      "epoch": 147.48,
      "grad_norm": 2.59588623046875,
      "learning_rate": 8.295006765704135e-07,
      "loss": 0.4514,
      "step": 360300
    },
    {
      "epoch": 147.49,
      "grad_norm": 1.9777296781539917,
      "learning_rate": 8.293796028717813e-07,
      "loss": 0.4729,
      "step": 360310
    },
    {
      "epoch": 147.49,
      "grad_norm": 1.931021809577942,
      "learning_rate": 8.292585361102453e-07,
      "loss": 0.4585,
      "step": 360320
    },
    {
      "epoch": 147.49,
      "grad_norm": 1.9839794635772705,
      "learning_rate": 8.291374762863602e-07,
      "loss": 0.4594,
      "step": 360330
    },
    {
      "epoch": 147.5,
      "grad_norm": 2.0982303619384766,
      "learning_rate": 8.290164234006803e-07,
      "loss": 0.4638,
      "step": 360340
    },
    {
      "epoch": 147.5,
      "grad_norm": 2.5322279930114746,
      "learning_rate": 8.288953774537603e-07,
      "loss": 0.441,
      "step": 360350
    },
    {
      "epoch": 147.51,
      "grad_norm": 2.157975435256958,
      "learning_rate": 8.287743384461551e-07,
      "loss": 0.4579,
      "step": 360360
    },
    {
      "epoch": 147.51,
      "grad_norm": 2.344029426574707,
      "learning_rate": 8.286533063784186e-07,
      "loss": 0.4572,
      "step": 360370
    },
    {
      "epoch": 147.52,
      "grad_norm": 1.9363008737564087,
      "learning_rate": 8.285322812511052e-07,
      "loss": 0.4402,
      "step": 360380
    },
    {
      "epoch": 147.52,
      "grad_norm": 1.7371739149093628,
      "learning_rate": 8.284112630647696e-07,
      "loss": 0.4612,
      "step": 360390
    },
    {
      "epoch": 147.52,
      "grad_norm": 3.395711898803711,
      "learning_rate": 8.282902518199663e-07,
      "loss": 0.4528,
      "step": 360400
    },
    {
      "epoch": 147.53,
      "grad_norm": 1.8542510271072388,
      "learning_rate": 8.281692475172493e-07,
      "loss": 0.4451,
      "step": 360410
    },
    {
      "epoch": 147.53,
      "grad_norm": 1.8646780252456665,
      "learning_rate": 8.280482501571729e-07,
      "loss": 0.4577,
      "step": 360420
    },
    {
      "epoch": 147.54,
      "grad_norm": 1.7037758827209473,
      "learning_rate": 8.279272597402915e-07,
      "loss": 0.4516,
      "step": 360430
    },
    {
      "epoch": 147.54,
      "grad_norm": 1.8363312482833862,
      "learning_rate": 8.278062762671593e-07,
      "loss": 0.439,
      "step": 360440
    },
    {
      "epoch": 147.54,
      "grad_norm": 2.0717718601226807,
      "learning_rate": 8.276852997383305e-07,
      "loss": 0.4707,
      "step": 360450
    },
    {
      "epoch": 147.55,
      "grad_norm": 2.0087242126464844,
      "learning_rate": 8.275643301543596e-07,
      "loss": 0.4573,
      "step": 360460
    },
    {
      "epoch": 147.55,
      "grad_norm": 2.5051767826080322,
      "learning_rate": 8.274433675158001e-07,
      "loss": 0.4628,
      "step": 360470
    },
    {
      "epoch": 147.56,
      "grad_norm": 1.9402930736541748,
      "learning_rate": 8.273224118232066e-07,
      "loss": 0.4376,
      "step": 360480
    },
    {
      "epoch": 147.56,
      "grad_norm": 2.1763696670532227,
      "learning_rate": 8.272014630771329e-07,
      "loss": 0.4679,
      "step": 360490
    },
    {
      "epoch": 147.56,
      "grad_norm": 1.8211699724197388,
      "learning_rate": 8.270805212781335e-07,
      "loss": 0.4265,
      "step": 360500
    },
    {
      "epoch": 147.57,
      "grad_norm": 2.0820634365081787,
      "learning_rate": 8.269595864267618e-07,
      "loss": 0.4334,
      "step": 360510
    },
    {
      "epoch": 147.57,
      "grad_norm": 2.1148886680603027,
      "learning_rate": 8.268386585235723e-07,
      "loss": 0.446,
      "step": 360520
    },
    {
      "epoch": 147.58,
      "grad_norm": 2.547105550765991,
      "learning_rate": 8.267177375691184e-07,
      "loss": 0.4492,
      "step": 360530
    },
    {
      "epoch": 147.58,
      "grad_norm": 2.2675061225891113,
      "learning_rate": 8.265968235639546e-07,
      "loss": 0.4658,
      "step": 360540
    },
    {
      "epoch": 147.58,
      "grad_norm": 1.8884471654891968,
      "learning_rate": 8.264759165086345e-07,
      "loss": 0.4398,
      "step": 360550
    },
    {
      "epoch": 147.59,
      "grad_norm": 2.1906559467315674,
      "learning_rate": 8.263550164037119e-07,
      "loss": 0.4352,
      "step": 360560
    },
    {
      "epoch": 147.59,
      "grad_norm": 2.0799789428710938,
      "learning_rate": 8.26234123249741e-07,
      "loss": 0.435,
      "step": 360570
    },
    {
      "epoch": 147.6,
      "grad_norm": 2.376948118209839,
      "learning_rate": 8.261132370472758e-07,
      "loss": 0.4349,
      "step": 360580
    },
    {
      "epoch": 147.6,
      "grad_norm": 2.573140859603882,
      "learning_rate": 8.259923577968689e-07,
      "loss": 0.4491,
      "step": 360590
    },
    {
      "epoch": 147.61,
      "grad_norm": 1.9460653066635132,
      "learning_rate": 8.258714854990747e-07,
      "loss": 0.4526,
      "step": 360600
    },
    {
      "epoch": 147.61,
      "grad_norm": 2.2484705448150635,
      "learning_rate": 8.257506201544469e-07,
      "loss": 0.4436,
      "step": 360610
    },
    {
      "epoch": 147.61,
      "grad_norm": 1.846574306488037,
      "learning_rate": 8.256297617635393e-07,
      "loss": 0.449,
      "step": 360620
    },
    {
      "epoch": 147.62,
      "grad_norm": 1.968519926071167,
      "learning_rate": 8.255089103269044e-07,
      "loss": 0.4394,
      "step": 360630
    },
    {
      "epoch": 147.62,
      "grad_norm": 2.2148234844207764,
      "learning_rate": 8.253880658450978e-07,
      "loss": 0.4536,
      "step": 360640
    },
    {
      "epoch": 147.63,
      "grad_norm": 2.171649217605591,
      "learning_rate": 8.252672283186719e-07,
      "loss": 0.4361,
      "step": 360650
    },
    {
      "epoch": 147.63,
      "grad_norm": 2.0194129943847656,
      "learning_rate": 8.251463977481805e-07,
      "loss": 0.4365,
      "step": 360660
    },
    {
      "epoch": 147.63,
      "grad_norm": 1.8455902338027954,
      "learning_rate": 8.25025574134177e-07,
      "loss": 0.4368,
      "step": 360670
    },
    {
      "epoch": 147.64,
      "grad_norm": 1.9395370483398438,
      "learning_rate": 8.249047574772147e-07,
      "loss": 0.4574,
      "step": 360680
    },
    {
      "epoch": 147.64,
      "grad_norm": 2.0529136657714844,
      "learning_rate": 8.247839477778474e-07,
      "loss": 0.4544,
      "step": 360690
    },
    {
      "epoch": 147.65,
      "grad_norm": 2.1552631855010986,
      "learning_rate": 8.246631450366283e-07,
      "loss": 0.4349,
      "step": 360700
    },
    {
      "epoch": 147.65,
      "grad_norm": 1.9291061162948608,
      "learning_rate": 8.245423492541118e-07,
      "loss": 0.4435,
      "step": 360710
    },
    {
      "epoch": 147.65,
      "grad_norm": 1.9646730422973633,
      "learning_rate": 8.244215604308493e-07,
      "loss": 0.4454,
      "step": 360720
    },
    {
      "epoch": 147.66,
      "grad_norm": 1.678312063217163,
      "learning_rate": 8.24300778567395e-07,
      "loss": 0.4517,
      "step": 360730
    },
    {
      "epoch": 147.66,
      "grad_norm": 1.7949652671813965,
      "learning_rate": 8.241800036643025e-07,
      "loss": 0.4661,
      "step": 360740
    },
    {
      "epoch": 147.67,
      "grad_norm": 1.9652172327041626,
      "learning_rate": 8.240592357221246e-07,
      "loss": 0.4557,
      "step": 360750
    },
    {
      "epoch": 147.67,
      "grad_norm": 1.7608642578125,
      "learning_rate": 8.239384747414149e-07,
      "loss": 0.4376,
      "step": 360760
    },
    {
      "epoch": 147.67,
      "grad_norm": 2.1147735118865967,
      "learning_rate": 8.238177207227265e-07,
      "loss": 0.4295,
      "step": 360770
    },
    {
      "epoch": 147.68,
      "grad_norm": 2.0750973224639893,
      "learning_rate": 8.236969736666123e-07,
      "loss": 0.4489,
      "step": 360780
    },
    {
      "epoch": 147.68,
      "grad_norm": 2.4117817878723145,
      "learning_rate": 8.235762335736255e-07,
      "loss": 0.4444,
      "step": 360790
    },
    {
      "epoch": 147.69,
      "grad_norm": 1.975666880607605,
      "learning_rate": 8.234555004443196e-07,
      "loss": 0.4434,
      "step": 360800
    },
    {
      "epoch": 147.69,
      "grad_norm": 2.114581346511841,
      "learning_rate": 8.233347742792471e-07,
      "loss": 0.4292,
      "step": 360810
    },
    {
      "epoch": 147.7,
      "grad_norm": 2.0174686908721924,
      "learning_rate": 8.232140550789607e-07,
      "loss": 0.4616,
      "step": 360820
    },
    {
      "epoch": 147.7,
      "grad_norm": 1.8216640949249268,
      "learning_rate": 8.230933428440146e-07,
      "loss": 0.4499,
      "step": 360830
    },
    {
      "epoch": 147.7,
      "grad_norm": 2.427227258682251,
      "learning_rate": 8.229726375749613e-07,
      "loss": 0.4554,
      "step": 360840
    },
    {
      "epoch": 147.71,
      "grad_norm": 2.437866687774658,
      "learning_rate": 8.228519392723544e-07,
      "loss": 0.4619,
      "step": 360850
    },
    {
      "epoch": 147.71,
      "grad_norm": 2.1208109855651855,
      "learning_rate": 8.227312479367449e-07,
      "loss": 0.4197,
      "step": 360860
    },
    {
      "epoch": 147.72,
      "grad_norm": 2.068805456161499,
      "learning_rate": 8.226105635686871e-07,
      "loss": 0.4479,
      "step": 360870
    },
    {
      "epoch": 147.72,
      "grad_norm": 2.6437020301818848,
      "learning_rate": 8.224898861687331e-07,
      "loss": 0.4747,
      "step": 360880
    },
    {
      "epoch": 147.72,
      "grad_norm": 1.8210655450820923,
      "learning_rate": 8.223692157374364e-07,
      "loss": 0.4559,
      "step": 360890
    },
    {
      "epoch": 147.73,
      "grad_norm": 2.118309736251831,
      "learning_rate": 8.222485522753492e-07,
      "loss": 0.4386,
      "step": 360900
    },
    {
      "epoch": 147.73,
      "grad_norm": 2.0651938915252686,
      "learning_rate": 8.221278957830246e-07,
      "loss": 0.4391,
      "step": 360910
    },
    {
      "epoch": 147.74,
      "grad_norm": 2.1671366691589355,
      "learning_rate": 8.220072462610152e-07,
      "loss": 0.4784,
      "step": 360920
    },
    {
      "epoch": 147.74,
      "grad_norm": 2.250349521636963,
      "learning_rate": 8.218866037098739e-07,
      "loss": 0.4368,
      "step": 360930
    },
    {
      "epoch": 147.74,
      "grad_norm": 1.8476698398590088,
      "learning_rate": 8.21765968130153e-07,
      "loss": 0.4557,
      "step": 360940
    },
    {
      "epoch": 147.75,
      "grad_norm": 2.4753201007843018,
      "learning_rate": 8.216453395224051e-07,
      "loss": 0.4455,
      "step": 360950
    },
    {
      "epoch": 147.75,
      "grad_norm": 2.1359987258911133,
      "learning_rate": 8.21524717887183e-07,
      "loss": 0.4613,
      "step": 360960
    },
    {
      "epoch": 147.76,
      "grad_norm": 2.7859723567962646,
      "learning_rate": 8.214041032250391e-07,
      "loss": 0.456,
      "step": 360970
    },
    {
      "epoch": 147.76,
      "grad_norm": 2.3935577869415283,
      "learning_rate": 8.212834955365259e-07,
      "loss": 0.4379,
      "step": 360980
    },
    {
      "epoch": 147.77,
      "grad_norm": 1.8830814361572266,
      "learning_rate": 8.211628948221963e-07,
      "loss": 0.4614,
      "step": 360990
    },
    {
      "epoch": 147.77,
      "grad_norm": 1.8053559064865112,
      "learning_rate": 8.210423010826023e-07,
      "loss": 0.4534,
      "step": 361000
    },
    {
      "epoch": 147.77,
      "grad_norm": 2.4923267364501953,
      "learning_rate": 8.209217143182964e-07,
      "loss": 0.4352,
      "step": 361010
    },
    {
      "epoch": 147.78,
      "grad_norm": 2.248847723007202,
      "learning_rate": 8.208011345298311e-07,
      "loss": 0.4596,
      "step": 361020
    },
    {
      "epoch": 147.78,
      "grad_norm": 1.755035638809204,
      "learning_rate": 8.206805617177585e-07,
      "loss": 0.4464,
      "step": 361030
    },
    {
      "epoch": 147.79,
      "grad_norm": 1.9354678392410278,
      "learning_rate": 8.205599958826313e-07,
      "loss": 0.4572,
      "step": 361040
    },
    {
      "epoch": 147.79,
      "grad_norm": 1.8330339193344116,
      "learning_rate": 8.204394370250018e-07,
      "loss": 0.4472,
      "step": 361050
    },
    {
      "epoch": 147.79,
      "grad_norm": 1.7780201435089111,
      "learning_rate": 8.203188851454217e-07,
      "loss": 0.4759,
      "step": 361060
    },
    {
      "epoch": 147.8,
      "grad_norm": 1.9929059743881226,
      "learning_rate": 8.201983402444439e-07,
      "loss": 0.4527,
      "step": 361070
    },
    {
      "epoch": 147.8,
      "grad_norm": 2.057992458343506,
      "learning_rate": 8.200778023226201e-07,
      "loss": 0.4576,
      "step": 361080
    },
    {
      "epoch": 147.81,
      "grad_norm": 2.4342868328094482,
      "learning_rate": 8.199572713805027e-07,
      "loss": 0.4713,
      "step": 361090
    },
    {
      "epoch": 147.81,
      "grad_norm": 1.590551495552063,
      "learning_rate": 8.198367474186438e-07,
      "loss": 0.4291,
      "step": 361100
    },
    {
      "epoch": 147.81,
      "grad_norm": 1.9545981884002686,
      "learning_rate": 8.197162304375956e-07,
      "loss": 0.4793,
      "step": 361110
    },
    {
      "epoch": 147.82,
      "grad_norm": 1.8655931949615479,
      "learning_rate": 8.195957204379098e-07,
      "loss": 0.4352,
      "step": 361120
    },
    {
      "epoch": 147.82,
      "grad_norm": 2.0632784366607666,
      "learning_rate": 8.194752174201391e-07,
      "loss": 0.4468,
      "step": 361130
    },
    {
      "epoch": 147.83,
      "grad_norm": 1.9750598669052124,
      "learning_rate": 8.193547213848348e-07,
      "loss": 0.4456,
      "step": 361140
    },
    {
      "epoch": 147.83,
      "grad_norm": 1.942164421081543,
      "learning_rate": 8.192342323325494e-07,
      "loss": 0.462,
      "step": 361150
    },
    {
      "epoch": 147.83,
      "grad_norm": 2.155876874923706,
      "learning_rate": 8.191137502638354e-07,
      "loss": 0.4285,
      "step": 361160
    },
    {
      "epoch": 147.84,
      "grad_norm": 2.405557155609131,
      "learning_rate": 8.18993275179243e-07,
      "loss": 0.4574,
      "step": 361170
    },
    {
      "epoch": 147.84,
      "grad_norm": 2.396733522415161,
      "learning_rate": 8.188728070793251e-07,
      "loss": 0.4427,
      "step": 361180
    },
    {
      "epoch": 147.85,
      "grad_norm": 2.8203694820404053,
      "learning_rate": 8.187523459646336e-07,
      "loss": 0.4544,
      "step": 361190
    },
    {
      "epoch": 147.85,
      "grad_norm": 2.358814239501953,
      "learning_rate": 8.186318918357191e-07,
      "loss": 0.449,
      "step": 361200
    },
    {
      "epoch": 147.86,
      "grad_norm": 1.970137357711792,
      "learning_rate": 8.185114446931353e-07,
      "loss": 0.4455,
      "step": 361210
    },
    {
      "epoch": 147.86,
      "grad_norm": 2.0049712657928467,
      "learning_rate": 8.183910045374332e-07,
      "loss": 0.4725,
      "step": 361220
    },
    {
      "epoch": 147.86,
      "grad_norm": 1.8174920082092285,
      "learning_rate": 8.182705713691645e-07,
      "loss": 0.4544,
      "step": 361230
    },
    {
      "epoch": 147.87,
      "grad_norm": 2.6167542934417725,
      "learning_rate": 8.18150145188881e-07,
      "loss": 0.455,
      "step": 361240
    },
    {
      "epoch": 147.87,
      "grad_norm": 1.9183822870254517,
      "learning_rate": 8.180297259971336e-07,
      "loss": 0.4411,
      "step": 361250
    },
    {
      "epoch": 147.88,
      "grad_norm": 2.0716893672943115,
      "learning_rate": 8.179093137944749e-07,
      "loss": 0.4433,
      "step": 361260
    },
    {
      "epoch": 147.88,
      "grad_norm": 1.7007707357406616,
      "learning_rate": 8.17788908581456e-07,
      "loss": 0.4314,
      "step": 361270
    },
    {
      "epoch": 147.88,
      "grad_norm": 2.066695213317871,
      "learning_rate": 8.176685103586285e-07,
      "loss": 0.4594,
      "step": 361280
    },
    {
      "epoch": 147.89,
      "grad_norm": 1.9490044116973877,
      "learning_rate": 8.175481191265439e-07,
      "loss": 0.4555,
      "step": 361290
    },
    {
      "epoch": 147.89,
      "grad_norm": 1.8150948286056519,
      "learning_rate": 8.174277348857547e-07,
      "loss": 0.4561,
      "step": 361300
    },
    {
      "epoch": 147.9,
      "grad_norm": 2.1738500595092773,
      "learning_rate": 8.173073576368106e-07,
      "loss": 0.4492,
      "step": 361310
    },
    {
      "epoch": 147.9,
      "grad_norm": 1.8912980556488037,
      "learning_rate": 8.171869873802637e-07,
      "loss": 0.4585,
      "step": 361320
    },
    {
      "epoch": 147.9,
      "grad_norm": 2.108323335647583,
      "learning_rate": 8.170666241166658e-07,
      "loss": 0.4468,
      "step": 361330
    },
    {
      "epoch": 147.91,
      "grad_norm": 2.0790395736694336,
      "learning_rate": 8.169462678465677e-07,
      "loss": 0.4468,
      "step": 361340
    },
    {
      "epoch": 147.91,
      "grad_norm": 1.7904731035232544,
      "learning_rate": 8.168259185705212e-07,
      "loss": 0.4541,
      "step": 361350
    },
    {
      "epoch": 147.92,
      "grad_norm": 1.8990315198898315,
      "learning_rate": 8.167055762890776e-07,
      "loss": 0.4163,
      "step": 361360
    },
    {
      "epoch": 147.92,
      "grad_norm": 2.238149404525757,
      "learning_rate": 8.165852410027877e-07,
      "loss": 0.4559,
      "step": 361370
    },
    {
      "epoch": 147.92,
      "grad_norm": 2.0398831367492676,
      "learning_rate": 8.164649127122033e-07,
      "loss": 0.459,
      "step": 361380
    },
    {
      "epoch": 147.93,
      "grad_norm": 2.212169647216797,
      "learning_rate": 8.163445914178744e-07,
      "loss": 0.459,
      "step": 361390
    },
    {
      "epoch": 147.93,
      "grad_norm": 1.7814441919326782,
      "learning_rate": 8.16224277120354e-07,
      "loss": 0.4594,
      "step": 361400
    },
    {
      "epoch": 147.94,
      "grad_norm": 2.2312090396881104,
      "learning_rate": 8.161039698201923e-07,
      "loss": 0.4523,
      "step": 361410
    },
    {
      "epoch": 147.94,
      "grad_norm": 1.9710719585418701,
      "learning_rate": 8.159836695179407e-07,
      "loss": 0.4502,
      "step": 361420
    },
    {
      "epoch": 147.95,
      "grad_norm": 2.0442352294921875,
      "learning_rate": 8.158633762141505e-07,
      "loss": 0.4675,
      "step": 361430
    },
    {
      "epoch": 147.95,
      "grad_norm": 1.881179690361023,
      "learning_rate": 8.157430899093718e-07,
      "loss": 0.4583,
      "step": 361440
    },
    {
      "epoch": 147.95,
      "grad_norm": 1.7619283199310303,
      "learning_rate": 8.156228106041558e-07,
      "loss": 0.448,
      "step": 361450
    },
    {
      "epoch": 147.96,
      "grad_norm": 2.2496867179870605,
      "learning_rate": 8.155025382990542e-07,
      "loss": 0.4582,
      "step": 361460
    },
    {
      "epoch": 147.96,
      "grad_norm": 1.9240453243255615,
      "learning_rate": 8.153822729946173e-07,
      "loss": 0.4444,
      "step": 361470
    },
    {
      "epoch": 147.97,
      "grad_norm": 2.321918487548828,
      "learning_rate": 8.152620146913963e-07,
      "loss": 0.4534,
      "step": 361480
    },
    {
      "epoch": 147.97,
      "grad_norm": 2.9198684692382812,
      "learning_rate": 8.15141763389942e-07,
      "loss": 0.4372,
      "step": 361490
    },
    {
      "epoch": 147.97,
      "grad_norm": 2.034686326980591,
      "learning_rate": 8.150215190908053e-07,
      "loss": 0.4468,
      "step": 361500
    },
    {
      "epoch": 147.98,
      "grad_norm": 1.8506677150726318,
      "learning_rate": 8.149012817945369e-07,
      "loss": 0.451,
      "step": 361510
    },
    {
      "epoch": 147.98,
      "grad_norm": 1.7845087051391602,
      "learning_rate": 8.14781051501688e-07,
      "loss": 0.4495,
      "step": 361520
    },
    {
      "epoch": 147.99,
      "grad_norm": 1.9461289644241333,
      "learning_rate": 8.146608282128089e-07,
      "loss": 0.4336,
      "step": 361530
    },
    {
      "epoch": 147.99,
      "grad_norm": 1.9631510972976685,
      "learning_rate": 8.145406119284504e-07,
      "loss": 0.4645,
      "step": 361540
    },
    {
      "epoch": 147.99,
      "grad_norm": 2.358123540878296,
      "learning_rate": 8.144204026491635e-07,
      "loss": 0.4397,
      "step": 361550
    },
    {
      "epoch": 148.0,
      "grad_norm": 2.1099324226379395,
      "learning_rate": 8.143002003754985e-07,
      "loss": 0.452,
      "step": 361560
    },
    {
      "epoch": 148.0,
      "eval_loss": 0.44952747225761414,
      "eval_runtime": 52.4176,
      "eval_samples_per_second": 65.799,
      "eval_steps_per_second": 8.242,
      "step": 361564
    },
    {
      "epoch": 148.0,
      "grad_norm": 1.819818139076233,
      "learning_rate": 8.141800051080061e-07,
      "loss": 0.4339,
      "step": 361570
    },
    {
      "epoch": 148.01,
      "grad_norm": 1.7530057430267334,
      "learning_rate": 8.140598168472371e-07,
      "loss": 0.4589,
      "step": 361580
    },
    {
      "epoch": 148.01,
      "grad_norm": 2.418884515762329,
      "learning_rate": 8.13939635593742e-07,
      "loss": 0.4239,
      "step": 361590
    },
    {
      "epoch": 148.01,
      "grad_norm": 2.3122053146362305,
      "learning_rate": 8.138194613480712e-07,
      "loss": 0.4428,
      "step": 361600
    },
    {
      "epoch": 148.02,
      "grad_norm": 1.5256305932998657,
      "learning_rate": 8.136992941107752e-07,
      "loss": 0.4467,
      "step": 361610
    },
    {
      "epoch": 148.02,
      "grad_norm": 1.8510841131210327,
      "learning_rate": 8.135791338824044e-07,
      "loss": 0.4451,
      "step": 361620
    },
    {
      "epoch": 148.03,
      "grad_norm": 1.9427285194396973,
      "learning_rate": 8.134589806635095e-07,
      "loss": 0.454,
      "step": 361630
    },
    {
      "epoch": 148.03,
      "grad_norm": 1.9695643186569214,
      "learning_rate": 8.133388344546408e-07,
      "loss": 0.4444,
      "step": 361640
    },
    {
      "epoch": 148.04,
      "grad_norm": 2.401144027709961,
      "learning_rate": 8.132186952563484e-07,
      "loss": 0.4597,
      "step": 361650
    },
    {
      "epoch": 148.04,
      "grad_norm": 1.6754838228225708,
      "learning_rate": 8.130985630691833e-07,
      "loss": 0.4541,
      "step": 361660
    },
    {
      "epoch": 148.04,
      "grad_norm": 2.169907569885254,
      "learning_rate": 8.129784378936949e-07,
      "loss": 0.4821,
      "step": 361670
    },
    {
      "epoch": 148.05,
      "grad_norm": 1.7776130437850952,
      "learning_rate": 8.128583197304342e-07,
      "loss": 0.4466,
      "step": 361680
    },
    {
      "epoch": 148.05,
      "grad_norm": 2.1243550777435303,
      "learning_rate": 8.127382085799514e-07,
      "loss": 0.4242,
      "step": 361690
    },
    {
      "epoch": 148.06,
      "grad_norm": 2.050746202468872,
      "learning_rate": 8.126181044427961e-07,
      "loss": 0.4392,
      "step": 361700
    },
    {
      "epoch": 148.06,
      "grad_norm": 2.418992757797241,
      "learning_rate": 8.124980073195192e-07,
      "loss": 0.4467,
      "step": 361710
    },
    {
      "epoch": 148.06,
      "grad_norm": 2.023592233657837,
      "learning_rate": 8.123779172106703e-07,
      "loss": 0.4431,
      "step": 361720
    },
    {
      "epoch": 148.07,
      "grad_norm": 2.0553977489471436,
      "learning_rate": 8.122578341168e-07,
      "loss": 0.4428,
      "step": 361730
    },
    {
      "epoch": 148.07,
      "grad_norm": 2.2008039951324463,
      "learning_rate": 8.121377580384589e-07,
      "loss": 0.4498,
      "step": 361740
    },
    {
      "epoch": 148.08,
      "grad_norm": 1.9137308597564697,
      "learning_rate": 8.120176889761953e-07,
      "loss": 0.4524,
      "step": 361750
    },
    {
      "epoch": 148.08,
      "grad_norm": 2.0710175037384033,
      "learning_rate": 8.118976269305595e-07,
      "loss": 0.4698,
      "step": 361760
    },
    {
      "epoch": 148.08,
      "grad_norm": 2.2354979515075684,
      "learning_rate": 8.117775719021029e-07,
      "loss": 0.4438,
      "step": 361770
    },
    {
      "epoch": 148.09,
      "grad_norm": 1.8553816080093384,
      "learning_rate": 8.116575238913749e-07,
      "loss": 0.4527,
      "step": 361780
    },
    {
      "epoch": 148.09,
      "grad_norm": 2.5725746154785156,
      "learning_rate": 8.115374828989249e-07,
      "loss": 0.4562,
      "step": 361790
    },
    {
      "epoch": 148.1,
      "grad_norm": 2.4784653186798096,
      "learning_rate": 8.114174489253036e-07,
      "loss": 0.4493,
      "step": 361800
    },
    {
      "epoch": 148.1,
      "grad_norm": 2.082963705062866,
      "learning_rate": 8.112974219710602e-07,
      "loss": 0.4412,
      "step": 361810
    },
    {
      "epoch": 148.1,
      "grad_norm": 2.2037770748138428,
      "learning_rate": 8.111774020367448e-07,
      "loss": 0.4494,
      "step": 361820
    },
    {
      "epoch": 148.11,
      "grad_norm": 1.9963099956512451,
      "learning_rate": 8.11057389122907e-07,
      "loss": 0.4621,
      "step": 361830
    },
    {
      "epoch": 148.11,
      "grad_norm": 1.8549948930740356,
      "learning_rate": 8.109373832300967e-07,
      "loss": 0.4685,
      "step": 361840
    },
    {
      "epoch": 148.12,
      "grad_norm": 1.7561399936676025,
      "learning_rate": 8.108173843588638e-07,
      "loss": 0.4324,
      "step": 361850
    },
    {
      "epoch": 148.12,
      "grad_norm": 2.0360190868377686,
      "learning_rate": 8.106973925097575e-07,
      "loss": 0.4505,
      "step": 361860
    },
    {
      "epoch": 148.13,
      "grad_norm": 2.0068259239196777,
      "learning_rate": 8.10577407683328e-07,
      "loss": 0.4392,
      "step": 361870
    },
    {
      "epoch": 148.13,
      "grad_norm": 1.9481805562973022,
      "learning_rate": 8.104574298801253e-07,
      "loss": 0.4446,
      "step": 361880
    },
    {
      "epoch": 148.13,
      "grad_norm": 2.1525754928588867,
      "learning_rate": 8.103374591006979e-07,
      "loss": 0.4436,
      "step": 361890
    },
    {
      "epoch": 148.14,
      "grad_norm": 2.043515205383301,
      "learning_rate": 8.102174953455958e-07,
      "loss": 0.4393,
      "step": 361900
    },
    {
      "epoch": 148.14,
      "grad_norm": 2.1592657566070557,
      "learning_rate": 8.100975386153683e-07,
      "loss": 0.4513,
      "step": 361910
    },
    {
      "epoch": 148.15,
      "grad_norm": 1.9507148265838623,
      "learning_rate": 8.099775889105655e-07,
      "loss": 0.4471,
      "step": 361920
    },
    {
      "epoch": 148.15,
      "grad_norm": 1.7101415395736694,
      "learning_rate": 8.098576462317366e-07,
      "loss": 0.4596,
      "step": 361930
    },
    {
      "epoch": 148.15,
      "grad_norm": 2.0511651039123535,
      "learning_rate": 8.097377105794309e-07,
      "loss": 0.4569,
      "step": 361940
    },
    {
      "epoch": 148.16,
      "grad_norm": 2.0935587882995605,
      "learning_rate": 8.096177819541971e-07,
      "loss": 0.4536,
      "step": 361950
    },
    {
      "epoch": 148.16,
      "grad_norm": 1.9118822813034058,
      "learning_rate": 8.094978603565864e-07,
      "loss": 0.4723,
      "step": 361960
    },
    {
      "epoch": 148.17,
      "grad_norm": 1.6736518144607544,
      "learning_rate": 8.09377945787147e-07,
      "loss": 0.4463,
      "step": 361970
    },
    {
      "epoch": 148.17,
      "grad_norm": 2.196817636489868,
      "learning_rate": 8.092580382464284e-07,
      "loss": 0.4426,
      "step": 361980
    },
    {
      "epoch": 148.17,
      "grad_norm": 1.8549222946166992,
      "learning_rate": 8.091381377349798e-07,
      "loss": 0.4432,
      "step": 361990
    },
    {
      "epoch": 148.18,
      "grad_norm": 2.5139434337615967,
      "learning_rate": 8.090182442533504e-07,
      "loss": 0.4541,
      "step": 362000
    },
    {
      "epoch": 148.18,
      "grad_norm": 2.139939308166504,
      "learning_rate": 8.088983578020903e-07,
      "loss": 0.4582,
      "step": 362010
    },
    {
      "epoch": 148.19,
      "grad_norm": 1.7616904973983765,
      "learning_rate": 8.087784783817472e-07,
      "loss": 0.4644,
      "step": 362020
    },
    {
      "epoch": 148.19,
      "grad_norm": 1.9504538774490356,
      "learning_rate": 8.086586059928709e-07,
      "loss": 0.4441,
      "step": 362030
    },
    {
      "epoch": 148.19,
      "grad_norm": 1.9640504121780396,
      "learning_rate": 8.085387406360104e-07,
      "loss": 0.4639,
      "step": 362040
    },
    {
      "epoch": 148.2,
      "grad_norm": 2.3766512870788574,
      "learning_rate": 8.084188823117151e-07,
      "loss": 0.4678,
      "step": 362050
    },
    {
      "epoch": 148.2,
      "grad_norm": 1.964476466178894,
      "learning_rate": 8.082990310205337e-07,
      "loss": 0.4358,
      "step": 362060
    },
    {
      "epoch": 148.21,
      "grad_norm": 2.108700752258301,
      "learning_rate": 8.081791867630156e-07,
      "loss": 0.4518,
      "step": 362070
    },
    {
      "epoch": 148.21,
      "grad_norm": 2.13362193107605,
      "learning_rate": 8.080593495397092e-07,
      "loss": 0.4496,
      "step": 362080
    },
    {
      "epoch": 148.22,
      "grad_norm": 2.247083902359009,
      "learning_rate": 8.079395193511641e-07,
      "loss": 0.475,
      "step": 362090
    },
    {
      "epoch": 148.22,
      "grad_norm": 2.003993511199951,
      "learning_rate": 8.078196961979292e-07,
      "loss": 0.449,
      "step": 362100
    },
    {
      "epoch": 148.22,
      "grad_norm": 1.9433765411376953,
      "learning_rate": 8.076998800805528e-07,
      "loss": 0.4733,
      "step": 362110
    },
    {
      "epoch": 148.23,
      "grad_norm": 2.082728862762451,
      "learning_rate": 8.075800709995841e-07,
      "loss": 0.4517,
      "step": 362120
    },
    {
      "epoch": 148.23,
      "grad_norm": 1.7770038843154907,
      "learning_rate": 8.074602689555721e-07,
      "loss": 0.4439,
      "step": 362130
    },
    {
      "epoch": 148.24,
      "grad_norm": 1.9499187469482422,
      "learning_rate": 8.073404739490648e-07,
      "loss": 0.4443,
      "step": 362140
    },
    {
      "epoch": 148.24,
      "grad_norm": 2.3327689170837402,
      "learning_rate": 8.072206859806127e-07,
      "loss": 0.4443,
      "step": 362150
    },
    {
      "epoch": 148.24,
      "grad_norm": 2.197072982788086,
      "learning_rate": 8.07100905050763e-07,
      "loss": 0.4527,
      "step": 362160
    },
    {
      "epoch": 148.25,
      "grad_norm": 1.823506474494934,
      "learning_rate": 8.069811311600648e-07,
      "loss": 0.4459,
      "step": 362170
    },
    {
      "epoch": 148.25,
      "grad_norm": 1.9082190990447998,
      "learning_rate": 8.068613643090666e-07,
      "loss": 0.4573,
      "step": 362180
    },
    {
      "epoch": 148.26,
      "grad_norm": 2.1212856769561768,
      "learning_rate": 8.067416044983176e-07,
      "loss": 0.452,
      "step": 362190
    },
    {
      "epoch": 148.26,
      "grad_norm": 2.6108078956604004,
      "learning_rate": 8.066218517283657e-07,
      "loss": 0.4568,
      "step": 362200
    },
    {
      "epoch": 148.26,
      "grad_norm": 1.812669277191162,
      "learning_rate": 8.065021059997599e-07,
      "loss": 0.4709,
      "step": 362210
    },
    {
      "epoch": 148.27,
      "grad_norm": 1.8380879163742065,
      "learning_rate": 8.063823673130486e-07,
      "loss": 0.446,
      "step": 362220
    },
    {
      "epoch": 148.27,
      "grad_norm": 1.9317736625671387,
      "learning_rate": 8.062626356687802e-07,
      "loss": 0.4458,
      "step": 362230
    },
    {
      "epoch": 148.28,
      "grad_norm": 2.1762566566467285,
      "learning_rate": 8.061429110675035e-07,
      "loss": 0.4532,
      "step": 362240
    },
    {
      "epoch": 148.28,
      "grad_norm": 2.046900749206543,
      "learning_rate": 8.060231935097667e-07,
      "loss": 0.44,
      "step": 362250
    },
    {
      "epoch": 148.28,
      "grad_norm": 1.9585814476013184,
      "learning_rate": 8.059034829961182e-07,
      "loss": 0.4547,
      "step": 362260
    },
    {
      "epoch": 148.29,
      "grad_norm": 2.2646238803863525,
      "learning_rate": 8.057837795271065e-07,
      "loss": 0.4495,
      "step": 362270
    },
    {
      "epoch": 148.29,
      "grad_norm": 1.9707578420639038,
      "learning_rate": 8.056640831032799e-07,
      "loss": 0.4381,
      "step": 362280
    },
    {
      "epoch": 148.3,
      "grad_norm": 1.9618462324142456,
      "learning_rate": 8.055443937251865e-07,
      "loss": 0.4738,
      "step": 362290
    },
    {
      "epoch": 148.3,
      "grad_norm": 2.113884687423706,
      "learning_rate": 8.05424711393375e-07,
      "loss": 0.4535,
      "step": 362300
    },
    {
      "epoch": 148.31,
      "grad_norm": 1.9892058372497559,
      "learning_rate": 8.053050361083931e-07,
      "loss": 0.4257,
      "step": 362310
    },
    {
      "epoch": 148.31,
      "grad_norm": 1.9638450145721436,
      "learning_rate": 8.051853678707905e-07,
      "loss": 0.4476,
      "step": 362320
    },
    {
      "epoch": 148.31,
      "grad_norm": 2.020813465118408,
      "learning_rate": 8.050657066811125e-07,
      "loss": 0.4601,
      "step": 362330
    },
    {
      "epoch": 148.32,
      "grad_norm": 1.8345969915390015,
      "learning_rate": 8.049460525399095e-07,
      "loss": 0.4241,
      "step": 362340
    },
    {
      "epoch": 148.32,
      "grad_norm": 2.1086924076080322,
      "learning_rate": 8.048264054477293e-07,
      "loss": 0.4456,
      "step": 362350
    },
    {
      "epoch": 148.33,
      "grad_norm": 2.0351009368896484,
      "learning_rate": 8.047067654051196e-07,
      "loss": 0.4527,
      "step": 362360
    },
    {
      "epoch": 148.33,
      "grad_norm": 1.9835045337677002,
      "learning_rate": 8.045871324126286e-07,
      "loss": 0.4745,
      "step": 362370
    },
    {
      "epoch": 148.33,
      "grad_norm": 1.954014539718628,
      "learning_rate": 8.044675064708047e-07,
      "loss": 0.4448,
      "step": 362380
    },
    {
      "epoch": 148.34,
      "grad_norm": 1.7787344455718994,
      "learning_rate": 8.043478875801953e-07,
      "loss": 0.4571,
      "step": 362390
    },
    {
      "epoch": 148.34,
      "grad_norm": 1.8561042547225952,
      "learning_rate": 8.042282757413484e-07,
      "loss": 0.4513,
      "step": 362400
    },
    {
      "epoch": 148.35,
      "grad_norm": 1.9266560077667236,
      "learning_rate": 8.041086709548121e-07,
      "loss": 0.4474,
      "step": 362410
    },
    {
      "epoch": 148.35,
      "grad_norm": 2.409539222717285,
      "learning_rate": 8.039890732211344e-07,
      "loss": 0.4505,
      "step": 362420
    },
    {
      "epoch": 148.35,
      "grad_norm": 1.9982831478118896,
      "learning_rate": 8.038694825408629e-07,
      "loss": 0.4513,
      "step": 362430
    },
    {
      "epoch": 148.36,
      "grad_norm": 1.8429179191589355,
      "learning_rate": 8.037498989145458e-07,
      "loss": 0.438,
      "step": 362440
    },
    {
      "epoch": 148.36,
      "grad_norm": 1.8399072885513306,
      "learning_rate": 8.036303223427307e-07,
      "loss": 0.4492,
      "step": 362450
    },
    {
      "epoch": 148.37,
      "grad_norm": 2.282573699951172,
      "learning_rate": 8.035107528259659e-07,
      "loss": 0.4484,
      "step": 362460
    },
    {
      "epoch": 148.37,
      "grad_norm": 1.7976914644241333,
      "learning_rate": 8.033911903647978e-07,
      "loss": 0.447,
      "step": 362470
    },
    {
      "epoch": 148.37,
      "grad_norm": 1.7573766708374023,
      "learning_rate": 8.03271634959775e-07,
      "loss": 0.4508,
      "step": 362480
    },
    {
      "epoch": 148.38,
      "grad_norm": 2.8013298511505127,
      "learning_rate": 8.031520866114449e-07,
      "loss": 0.4467,
      "step": 362490
    },
    {
      "epoch": 148.38,
      "grad_norm": 1.9968689680099487,
      "learning_rate": 8.030325453203553e-07,
      "loss": 0.452,
      "step": 362500
    },
    {
      "epoch": 148.39,
      "grad_norm": 1.7323307991027832,
      "learning_rate": 8.029130110870535e-07,
      "loss": 0.4673,
      "step": 362510
    },
    {
      "epoch": 148.39,
      "grad_norm": 2.3015787601470947,
      "learning_rate": 8.027934839120868e-07,
      "loss": 0.4363,
      "step": 362520
    },
    {
      "epoch": 148.4,
      "grad_norm": 2.246549606323242,
      "learning_rate": 8.026739637960042e-07,
      "loss": 0.4432,
      "step": 362530
    },
    {
      "epoch": 148.4,
      "grad_norm": 2.0101726055145264,
      "learning_rate": 8.025544507393521e-07,
      "loss": 0.4537,
      "step": 362540
    },
    {
      "epoch": 148.4,
      "grad_norm": 2.165351629257202,
      "learning_rate": 8.02434944742678e-07,
      "loss": 0.4628,
      "step": 362550
    },
    {
      "epoch": 148.41,
      "grad_norm": 1.9187896251678467,
      "learning_rate": 8.023154458065296e-07,
      "loss": 0.4233,
      "step": 362560
    },
    {
      "epoch": 148.41,
      "grad_norm": 1.9925756454467773,
      "learning_rate": 8.021959539314539e-07,
      "loss": 0.451,
      "step": 362570
    },
    {
      "epoch": 148.42,
      "grad_norm": 1.9033353328704834,
      "learning_rate": 8.020764691179988e-07,
      "loss": 0.4577,
      "step": 362580
    },
    {
      "epoch": 148.42,
      "grad_norm": 2.280696153640747,
      "learning_rate": 8.019569913667112e-07,
      "loss": 0.4261,
      "step": 362590
    },
    {
      "epoch": 148.42,
      "grad_norm": 2.0637457370758057,
      "learning_rate": 8.018375206781397e-07,
      "loss": 0.4212,
      "step": 362600
    },
    {
      "epoch": 148.43,
      "grad_norm": 2.2579236030578613,
      "learning_rate": 8.017180570528294e-07,
      "loss": 0.4678,
      "step": 362610
    },
    {
      "epoch": 148.43,
      "grad_norm": 2.050873279571533,
      "learning_rate": 8.015986004913287e-07,
      "loss": 0.4641,
      "step": 362620
    },
    {
      "epoch": 148.44,
      "grad_norm": 2.0253171920776367,
      "learning_rate": 8.01479150994185e-07,
      "loss": 0.441,
      "step": 362630
    },
    {
      "epoch": 148.44,
      "grad_norm": 2.079984426498413,
      "learning_rate": 8.013597085619449e-07,
      "loss": 0.4369,
      "step": 362640
    },
    {
      "epoch": 148.44,
      "grad_norm": 1.7675358057022095,
      "learning_rate": 8.012402731951558e-07,
      "loss": 0.4462,
      "step": 362650
    },
    {
      "epoch": 148.45,
      "grad_norm": 2.105537176132202,
      "learning_rate": 8.01120844894365e-07,
      "loss": 0.4386,
      "step": 362660
    },
    {
      "epoch": 148.45,
      "grad_norm": 2.9262304306030273,
      "learning_rate": 8.010014236601196e-07,
      "loss": 0.4434,
      "step": 362670
    },
    {
      "epoch": 148.46,
      "grad_norm": 2.226101875305176,
      "learning_rate": 8.008820094929663e-07,
      "loss": 0.4674,
      "step": 362680
    },
    {
      "epoch": 148.46,
      "grad_norm": 2.0408103466033936,
      "learning_rate": 8.007626023934522e-07,
      "loss": 0.4733,
      "step": 362690
    },
    {
      "epoch": 148.47,
      "grad_norm": 1.765121340751648,
      "learning_rate": 8.006432023621246e-07,
      "loss": 0.4426,
      "step": 362700
    },
    {
      "epoch": 148.47,
      "grad_norm": 1.579260230064392,
      "learning_rate": 8.005238093995295e-07,
      "loss": 0.4241,
      "step": 362710
    },
    {
      "epoch": 148.47,
      "grad_norm": 2.039595365524292,
      "learning_rate": 8.004044235062152e-07,
      "loss": 0.4544,
      "step": 362720
    },
    {
      "epoch": 148.48,
      "grad_norm": 2.3520402908325195,
      "learning_rate": 8.00285044682729e-07,
      "loss": 0.4408,
      "step": 362730
    },
    {
      "epoch": 148.48,
      "grad_norm": 2.278374195098877,
      "learning_rate": 8.001656729296158e-07,
      "loss": 0.4519,
      "step": 362740
    },
    {
      "epoch": 148.49,
      "grad_norm": 2.960221767425537,
      "learning_rate": 8.000463082474234e-07,
      "loss": 0.4497,
      "step": 362750
    },
    {
      "epoch": 148.49,
      "grad_norm": 1.8159539699554443,
      "learning_rate": 7.999269506366985e-07,
      "loss": 0.445,
      "step": 362760
    },
    {
      "epoch": 148.49,
      "grad_norm": 2.1520707607269287,
      "learning_rate": 7.998076000979878e-07,
      "loss": 0.4416,
      "step": 362770
    },
    {
      "epoch": 148.5,
      "grad_norm": 2.345829486846924,
      "learning_rate": 7.996882566318382e-07,
      "loss": 0.4739,
      "step": 362780
    },
    {
      "epoch": 148.5,
      "grad_norm": 2.159376382827759,
      "learning_rate": 7.995689202387963e-07,
      "loss": 0.4404,
      "step": 362790
    },
    {
      "epoch": 148.51,
      "grad_norm": 2.1800456047058105,
      "learning_rate": 7.994495909194089e-07,
      "loss": 0.4731,
      "step": 362800
    },
    {
      "epoch": 148.51,
      "grad_norm": 2.2068092823028564,
      "learning_rate": 7.993302686742226e-07,
      "loss": 0.4411,
      "step": 362810
    },
    {
      "epoch": 148.51,
      "grad_norm": 2.1681861877441406,
      "learning_rate": 7.992109535037835e-07,
      "loss": 0.47,
      "step": 362820
    },
    {
      "epoch": 148.52,
      "grad_norm": 1.96420156955719,
      "learning_rate": 7.99091645408639e-07,
      "loss": 0.4341,
      "step": 362830
    },
    {
      "epoch": 148.52,
      "grad_norm": 2.4436628818511963,
      "learning_rate": 7.989723443893349e-07,
      "loss": 0.4596,
      "step": 362840
    },
    {
      "epoch": 148.53,
      "grad_norm": 2.1538288593292236,
      "learning_rate": 7.988530504464183e-07,
      "loss": 0.459,
      "step": 362850
    },
    {
      "epoch": 148.53,
      "grad_norm": 1.7797755002975464,
      "learning_rate": 7.987337635804351e-07,
      "loss": 0.4456,
      "step": 362860
    },
    {
      "epoch": 148.53,
      "grad_norm": 2.0238444805145264,
      "learning_rate": 7.98614483791932e-07,
      "loss": 0.4562,
      "step": 362870
    },
    {
      "epoch": 148.54,
      "grad_norm": 2.163950204849243,
      "learning_rate": 7.984952110814556e-07,
      "loss": 0.465,
      "step": 362880
    },
    {
      "epoch": 148.54,
      "grad_norm": 1.7912346124649048,
      "learning_rate": 7.983759454495518e-07,
      "loss": 0.4457,
      "step": 362890
    },
    {
      "epoch": 148.55,
      "grad_norm": 1.9839900732040405,
      "learning_rate": 7.982566868967673e-07,
      "loss": 0.4469,
      "step": 362900
    },
    {
      "epoch": 148.55,
      "grad_norm": 1.9348520040512085,
      "learning_rate": 7.981374354236484e-07,
      "loss": 0.4568,
      "step": 362910
    },
    {
      "epoch": 148.56,
      "grad_norm": 1.9259169101715088,
      "learning_rate": 7.980181910307413e-07,
      "loss": 0.4485,
      "step": 362920
    },
    {
      "epoch": 148.56,
      "grad_norm": 2.0258824825286865,
      "learning_rate": 7.978989537185922e-07,
      "loss": 0.4785,
      "step": 362930
    },
    {
      "epoch": 148.56,
      "grad_norm": 1.624900221824646,
      "learning_rate": 7.977797234877472e-07,
      "loss": 0.4444,
      "step": 362940
    },
    {
      "epoch": 148.57,
      "grad_norm": 1.8201524019241333,
      "learning_rate": 7.976605003387529e-07,
      "loss": 0.4474,
      "step": 362950
    },
    {
      "epoch": 148.57,
      "grad_norm": 1.9560720920562744,
      "learning_rate": 7.975412842721551e-07,
      "loss": 0.4398,
      "step": 362960
    },
    {
      "epoch": 148.58,
      "grad_norm": 2.094665050506592,
      "learning_rate": 7.974220752884999e-07,
      "loss": 0.4646,
      "step": 362970
    },
    {
      "epoch": 148.58,
      "grad_norm": 2.090167284011841,
      "learning_rate": 7.973028733883335e-07,
      "loss": 0.4421,
      "step": 362980
    },
    {
      "epoch": 148.58,
      "grad_norm": 2.051379442214966,
      "learning_rate": 7.971836785722016e-07,
      "loss": 0.4529,
      "step": 362990
    },
    {
      "epoch": 148.59,
      "grad_norm": 1.804979681968689,
      "learning_rate": 7.970644908406507e-07,
      "loss": 0.4557,
      "step": 363000
    },
    {
      "epoch": 148.59,
      "grad_norm": 2.5089731216430664,
      "learning_rate": 7.969453101942268e-07,
      "loss": 0.4564,
      "step": 363010
    },
    {
      "epoch": 148.6,
      "grad_norm": 1.8300750255584717,
      "learning_rate": 7.968261366334753e-07,
      "loss": 0.4482,
      "step": 363020
    },
    {
      "epoch": 148.6,
      "grad_norm": 2.1503190994262695,
      "learning_rate": 7.967069701589426e-07,
      "loss": 0.4472,
      "step": 363030
    },
    {
      "epoch": 148.6,
      "grad_norm": 2.4821226596832275,
      "learning_rate": 7.965878107711752e-07,
      "loss": 0.4707,
      "step": 363040
    },
    {
      "epoch": 148.61,
      "grad_norm": 1.951722502708435,
      "learning_rate": 7.964686584707174e-07,
      "loss": 0.4343,
      "step": 363050
    },
    {
      "epoch": 148.61,
      "grad_norm": 2.132545232772827,
      "learning_rate": 7.963495132581159e-07,
      "loss": 0.444,
      "step": 363060
    },
    {
      "epoch": 148.62,
      "grad_norm": 2.078599214553833,
      "learning_rate": 7.962303751339162e-07,
      "loss": 0.4554,
      "step": 363070
    },
    {
      "epoch": 148.62,
      "grad_norm": 1.905712604522705,
      "learning_rate": 7.961112440986634e-07,
      "loss": 0.4485,
      "step": 363080
    },
    {
      "epoch": 148.62,
      "grad_norm": 2.424043655395508,
      "learning_rate": 7.95992120152905e-07,
      "loss": 0.4398,
      "step": 363090
    },
    {
      "epoch": 148.63,
      "grad_norm": 1.884817123413086,
      "learning_rate": 7.958730032971858e-07,
      "loss": 0.4531,
      "step": 363100
    },
    {
      "epoch": 148.63,
      "grad_norm": 2.112128257751465,
      "learning_rate": 7.957538935320513e-07,
      "loss": 0.449,
      "step": 363110
    },
    {
      "epoch": 148.64,
      "grad_norm": 2.55587100982666,
      "learning_rate": 7.956347908580472e-07,
      "loss": 0.4328,
      "step": 363120
    },
    {
      "epoch": 148.64,
      "grad_norm": 2.3578851222991943,
      "learning_rate": 7.955156952757192e-07,
      "loss": 0.4589,
      "step": 363130
    },
    {
      "epoch": 148.65,
      "grad_norm": 1.722034215927124,
      "learning_rate": 7.953966067856127e-07,
      "loss": 0.4494,
      "step": 363140
    },
    {
      "epoch": 148.65,
      "grad_norm": 2.025174140930176,
      "learning_rate": 7.952775253882733e-07,
      "loss": 0.4309,
      "step": 363150
    },
    {
      "epoch": 148.65,
      "grad_norm": 1.8145203590393066,
      "learning_rate": 7.951584510842467e-07,
      "loss": 0.4309,
      "step": 363160
    },
    {
      "epoch": 148.66,
      "grad_norm": 2.1211280822753906,
      "learning_rate": 7.950393838740779e-07,
      "loss": 0.4416,
      "step": 363170
    },
    {
      "epoch": 148.66,
      "grad_norm": 1.8729631900787354,
      "learning_rate": 7.949203237583134e-07,
      "loss": 0.4449,
      "step": 363180
    },
    {
      "epoch": 148.67,
      "grad_norm": 1.7867425680160522,
      "learning_rate": 7.948012707374972e-07,
      "loss": 0.4374,
      "step": 363190
    },
    {
      "epoch": 148.67,
      "grad_norm": 2.393146276473999,
      "learning_rate": 7.94682224812175e-07,
      "loss": 0.4836,
      "step": 363200
    },
    {
      "epoch": 148.67,
      "grad_norm": 2.4977331161499023,
      "learning_rate": 7.945631859828924e-07,
      "loss": 0.457,
      "step": 363210
    },
    {
      "epoch": 148.68,
      "grad_norm": 2.166289806365967,
      "learning_rate": 7.944441542501949e-07,
      "loss": 0.461,
      "step": 363220
    },
    {
      "epoch": 148.68,
      "grad_norm": 2.1527040004730225,
      "learning_rate": 7.943251296146273e-07,
      "loss": 0.4422,
      "step": 363230
    },
    {
      "epoch": 148.69,
      "grad_norm": 2.0399160385131836,
      "learning_rate": 7.942061120767352e-07,
      "loss": 0.433,
      "step": 363240
    },
    {
      "epoch": 148.69,
      "grad_norm": 1.8806109428405762,
      "learning_rate": 7.940871016370636e-07,
      "loss": 0.4391,
      "step": 363250
    },
    {
      "epoch": 148.69,
      "grad_norm": 1.6331483125686646,
      "learning_rate": 7.939680982961579e-07,
      "loss": 0.4538,
      "step": 363260
    },
    {
      "epoch": 148.7,
      "grad_norm": 1.848122000694275,
      "learning_rate": 7.938491020545621e-07,
      "loss": 0.4633,
      "step": 363270
    },
    {
      "epoch": 148.7,
      "grad_norm": 2.194141149520874,
      "learning_rate": 7.937301129128231e-07,
      "loss": 0.456,
      "step": 363280
    },
    {
      "epoch": 148.71,
      "grad_norm": 1.6725958585739136,
      "learning_rate": 7.936111308714852e-07,
      "loss": 0.4558,
      "step": 363290
    },
    {
      "epoch": 148.71,
      "grad_norm": 3.0346109867095947,
      "learning_rate": 7.934921559310932e-07,
      "loss": 0.4456,
      "step": 363300
    },
    {
      "epoch": 148.71,
      "grad_norm": 2.4447121620178223,
      "learning_rate": 7.933731880921927e-07,
      "loss": 0.4559,
      "step": 363310
    },
    {
      "epoch": 148.72,
      "grad_norm": 2.030376672744751,
      "learning_rate": 7.932542273553286e-07,
      "loss": 0.4435,
      "step": 363320
    },
    {
      "epoch": 148.72,
      "grad_norm": 2.3326432704925537,
      "learning_rate": 7.93135273721045e-07,
      "loss": 0.434,
      "step": 363330
    },
    {
      "epoch": 148.73,
      "grad_norm": 2.0457141399383545,
      "learning_rate": 7.930163271898873e-07,
      "loss": 0.4301,
      "step": 363340
    },
    {
      "epoch": 148.73,
      "grad_norm": 2.004131555557251,
      "learning_rate": 7.928973877624004e-07,
      "loss": 0.462,
      "step": 363350
    },
    {
      "epoch": 148.74,
      "grad_norm": 1.9227502346038818,
      "learning_rate": 7.927784554391291e-07,
      "loss": 0.4627,
      "step": 363360
    },
    {
      "epoch": 148.74,
      "grad_norm": 2.1198089122772217,
      "learning_rate": 7.926595302206184e-07,
      "loss": 0.4344,
      "step": 363370
    },
    {
      "epoch": 148.74,
      "grad_norm": 2.232929229736328,
      "learning_rate": 7.925406121074125e-07,
      "loss": 0.4795,
      "step": 363380
    },
    {
      "epoch": 148.75,
      "grad_norm": 1.7731331586837769,
      "learning_rate": 7.924217011000569e-07,
      "loss": 0.4258,
      "step": 363390
    },
    {
      "epoch": 148.75,
      "grad_norm": 2.2043063640594482,
      "learning_rate": 7.92302797199096e-07,
      "loss": 0.4596,
      "step": 363400
    },
    {
      "epoch": 148.76,
      "grad_norm": 1.8736652135849,
      "learning_rate": 7.921839004050743e-07,
      "loss": 0.4579,
      "step": 363410
    },
    {
      "epoch": 148.76,
      "grad_norm": 2.1882426738739014,
      "learning_rate": 7.920650107185368e-07,
      "loss": 0.4381,
      "step": 363420
    },
    {
      "epoch": 148.76,
      "grad_norm": 1.8126633167266846,
      "learning_rate": 7.919461281400277e-07,
      "loss": 0.451,
      "step": 363430
    },
    {
      "epoch": 148.77,
      "grad_norm": 1.8302335739135742,
      "learning_rate": 7.918272526700921e-07,
      "loss": 0.4497,
      "step": 363440
    },
    {
      "epoch": 148.77,
      "grad_norm": 2.761753797531128,
      "learning_rate": 7.917083843092741e-07,
      "loss": 0.4402,
      "step": 363450
    },
    {
      "epoch": 148.78,
      "grad_norm": 2.249886989593506,
      "learning_rate": 7.915895230581185e-07,
      "loss": 0.4647,
      "step": 363460
    },
    {
      "epoch": 148.78,
      "grad_norm": 2.1837215423583984,
      "learning_rate": 7.914706689171693e-07,
      "loss": 0.4452,
      "step": 363470
    },
    {
      "epoch": 148.78,
      "grad_norm": 2.0286402702331543,
      "learning_rate": 7.913518218869716e-07,
      "loss": 0.4799,
      "step": 363480
    },
    {
      "epoch": 148.79,
      "grad_norm": 1.9692556858062744,
      "learning_rate": 7.912329819680694e-07,
      "loss": 0.4455,
      "step": 363490
    },
    {
      "epoch": 148.79,
      "grad_norm": 1.7768861055374146,
      "learning_rate": 7.911141491610075e-07,
      "loss": 0.4261,
      "step": 363500
    },
    {
      "epoch": 148.8,
      "grad_norm": 2.1784894466400146,
      "learning_rate": 7.909953234663297e-07,
      "loss": 0.4511,
      "step": 363510
    },
    {
      "epoch": 148.8,
      "grad_norm": 1.8702236413955688,
      "learning_rate": 7.908765048845806e-07,
      "loss": 0.4536,
      "step": 363520
    },
    {
      "epoch": 148.8,
      "grad_norm": 2.2181665897369385,
      "learning_rate": 7.907576934163046e-07,
      "loss": 0.4676,
      "step": 363530
    },
    {
      "epoch": 148.81,
      "grad_norm": 1.7473113536834717,
      "learning_rate": 7.906388890620458e-07,
      "loss": 0.4559,
      "step": 363540
    },
    {
      "epoch": 148.81,
      "grad_norm": 2.1451938152313232,
      "learning_rate": 7.905200918223485e-07,
      "loss": 0.453,
      "step": 363550
    },
    {
      "epoch": 148.82,
      "grad_norm": 2.290912628173828,
      "learning_rate": 7.90401301697757e-07,
      "loss": 0.4474,
      "step": 363560
    },
    {
      "epoch": 148.82,
      "grad_norm": 1.5143818855285645,
      "learning_rate": 7.90282518688815e-07,
      "loss": 0.4438,
      "step": 363570
    },
    {
      "epoch": 148.83,
      "grad_norm": 1.9028898477554321,
      "learning_rate": 7.901637427960672e-07,
      "loss": 0.4472,
      "step": 363580
    },
    {
      "epoch": 148.83,
      "grad_norm": 2.4862282276153564,
      "learning_rate": 7.900449740200574e-07,
      "loss": 0.4479,
      "step": 363590
    },
    {
      "epoch": 148.83,
      "grad_norm": 2.570413827896118,
      "learning_rate": 7.899262123613296e-07,
      "loss": 0.4688,
      "step": 363600
    },
    {
      "epoch": 148.84,
      "grad_norm": 2.0795748233795166,
      "learning_rate": 7.898074578204282e-07,
      "loss": 0.4413,
      "step": 363610
    },
    {
      "epoch": 148.84,
      "grad_norm": 1.9688886404037476,
      "learning_rate": 7.896887103978966e-07,
      "loss": 0.4493,
      "step": 363620
    },
    {
      "epoch": 148.85,
      "grad_norm": 1.8316770792007446,
      "learning_rate": 7.8956997009428e-07,
      "loss": 0.4356,
      "step": 363630
    },
    {
      "epoch": 148.85,
      "grad_norm": 2.4508256912231445,
      "learning_rate": 7.894512369101207e-07,
      "loss": 0.4463,
      "step": 363640
    },
    {
      "epoch": 148.85,
      "grad_norm": 2.024901866912842,
      "learning_rate": 7.893325108459623e-07,
      "loss": 0.4602,
      "step": 363650
    },
    {
      "epoch": 148.86,
      "grad_norm": 2.27411150932312,
      "learning_rate": 7.892137919023508e-07,
      "loss": 0.4451,
      "step": 363660
    },
    {
      "epoch": 148.86,
      "grad_norm": 2.287660837173462,
      "learning_rate": 7.890950800798287e-07,
      "loss": 0.4468,
      "step": 363670
    },
    {
      "epoch": 148.87,
      "grad_norm": 1.9838323593139648,
      "learning_rate": 7.889763753789401e-07,
      "loss": 0.4614,
      "step": 363680
    },
    {
      "epoch": 148.87,
      "grad_norm": 2.0512373447418213,
      "learning_rate": 7.888576778002285e-07,
      "loss": 0.452,
      "step": 363690
    },
    {
      "epoch": 148.87,
      "grad_norm": 2.2494773864746094,
      "learning_rate": 7.887389873442381e-07,
      "loss": 0.4468,
      "step": 363700
    },
    {
      "epoch": 148.88,
      "grad_norm": 2.1420538425445557,
      "learning_rate": 7.886203040115123e-07,
      "loss": 0.4422,
      "step": 363710
    },
    {
      "epoch": 148.88,
      "grad_norm": 2.2527170181274414,
      "learning_rate": 7.885016278025947e-07,
      "loss": 0.462,
      "step": 363720
    },
    {
      "epoch": 148.89,
      "grad_norm": 2.060821771621704,
      "learning_rate": 7.88382958718029e-07,
      "loss": 0.4648,
      "step": 363730
    },
    {
      "epoch": 148.89,
      "grad_norm": 2.252566337585449,
      "learning_rate": 7.88264296758359e-07,
      "loss": 0.4609,
      "step": 363740
    },
    {
      "epoch": 148.89,
      "grad_norm": 2.0011043548583984,
      "learning_rate": 7.881456419241279e-07,
      "loss": 0.4389,
      "step": 363750
    },
    {
      "epoch": 148.9,
      "grad_norm": 2.545447587966919,
      "learning_rate": 7.880269942158803e-07,
      "loss": 0.4443,
      "step": 363760
    },
    {
      "epoch": 148.9,
      "grad_norm": 2.206622838973999,
      "learning_rate": 7.879083536341581e-07,
      "loss": 0.4319,
      "step": 363770
    },
    {
      "epoch": 148.91,
      "grad_norm": 2.1086816787719727,
      "learning_rate": 7.877897201795055e-07,
      "loss": 0.4583,
      "step": 363780
    },
    {
      "epoch": 148.91,
      "grad_norm": 2.1003081798553467,
      "learning_rate": 7.876710938524661e-07,
      "loss": 0.4666,
      "step": 363790
    },
    {
      "epoch": 148.92,
      "grad_norm": 1.9962475299835205,
      "learning_rate": 7.875524746535829e-07,
      "loss": 0.4616,
      "step": 363800
    },
    {
      "epoch": 148.92,
      "grad_norm": 2.417550802230835,
      "learning_rate": 7.874338625833998e-07,
      "loss": 0.444,
      "step": 363810
    },
    {
      "epoch": 148.92,
      "grad_norm": 2.0994393825531006,
      "learning_rate": 7.873152576424599e-07,
      "loss": 0.4506,
      "step": 363820
    },
    {
      "epoch": 148.93,
      "grad_norm": 1.9455398321151733,
      "learning_rate": 7.871966598313063e-07,
      "loss": 0.4402,
      "step": 363830
    },
    {
      "epoch": 148.93,
      "grad_norm": 13.059037208557129,
      "learning_rate": 7.87078069150482e-07,
      "loss": 0.4294,
      "step": 363840
    },
    {
      "epoch": 148.94,
      "grad_norm": 1.790541648864746,
      "learning_rate": 7.869594856005314e-07,
      "loss": 0.4599,
      "step": 363850
    },
    {
      "epoch": 148.94,
      "grad_norm": 2.0683913230895996,
      "learning_rate": 7.86840909181997e-07,
      "loss": 0.4512,
      "step": 363860
    },
    {
      "epoch": 148.94,
      "grad_norm": 1.8398398160934448,
      "learning_rate": 7.86722339895422e-07,
      "loss": 0.4308,
      "step": 363870
    },
    {
      "epoch": 148.95,
      "grad_norm": 2.540766716003418,
      "learning_rate": 7.866037777413497e-07,
      "loss": 0.4466,
      "step": 363880
    },
    {
      "epoch": 148.95,
      "grad_norm": 1.8805938959121704,
      "learning_rate": 7.864852227203228e-07,
      "loss": 0.451,
      "step": 363890
    },
    {
      "epoch": 148.96,
      "grad_norm": 2.1054601669311523,
      "learning_rate": 7.863666748328857e-07,
      "loss": 0.4333,
      "step": 363900
    },
    {
      "epoch": 148.96,
      "grad_norm": 2.420793294906616,
      "learning_rate": 7.862481340795797e-07,
      "loss": 0.4249,
      "step": 363910
    },
    {
      "epoch": 148.96,
      "grad_norm": 2.0987131595611572,
      "learning_rate": 7.861296004609483e-07,
      "loss": 0.4395,
      "step": 363920
    },
    {
      "epoch": 148.97,
      "grad_norm": 1.839942455291748,
      "learning_rate": 7.860110739775349e-07,
      "loss": 0.4418,
      "step": 363930
    },
    {
      "epoch": 148.97,
      "grad_norm": 1.855191707611084,
      "learning_rate": 7.858925546298821e-07,
      "loss": 0.4327,
      "step": 363940
    },
    {
      "epoch": 148.98,
      "grad_norm": 1.9294177293777466,
      "learning_rate": 7.857740424185332e-07,
      "loss": 0.4428,
      "step": 363950
    },
    {
      "epoch": 148.98,
      "grad_norm": 2.1442108154296875,
      "learning_rate": 7.856555373440309e-07,
      "loss": 0.4262,
      "step": 363960
    },
    {
      "epoch": 148.98,
      "grad_norm": 2.017918586730957,
      "learning_rate": 7.855370394069178e-07,
      "loss": 0.458,
      "step": 363970
    },
    {
      "epoch": 148.99,
      "grad_norm": 1.9413461685180664,
      "learning_rate": 7.854185486077371e-07,
      "loss": 0.4463,
      "step": 363980
    },
    {
      "epoch": 148.99,
      "grad_norm": 1.790766954421997,
      "learning_rate": 7.853000649470315e-07,
      "loss": 0.4357,
      "step": 363990
    },
    {
      "epoch": 149.0,
      "grad_norm": 1.9319112300872803,
      "learning_rate": 7.851815884253437e-07,
      "loss": 0.4449,
      "step": 364000
    },
    {
      "epoch": 149.0,
      "eval_loss": 0.4508100152015686,
      "eval_runtime": 52.3895,
      "eval_samples_per_second": 65.834,
      "eval_steps_per_second": 8.246,
      "step": 364007
    },
    {
      "epoch": 149.0,
      "grad_norm": 1.8197331428527832,
      "learning_rate": 7.850631190432162e-07,
      "loss": 0.4397,
      "step": 364010
    },
    {
      "epoch": 149.01,
      "grad_norm": 2.0267529487609863,
      "learning_rate": 7.849446568011923e-07,
      "loss": 0.4338,
      "step": 364020
    },
    {
      "epoch": 149.01,
      "grad_norm": 1.839205026626587,
      "learning_rate": 7.84826201699814e-07,
      "loss": 0.4314,
      "step": 364030
    },
    {
      "epoch": 149.01,
      "grad_norm": 1.863008737564087,
      "learning_rate": 7.847077537396241e-07,
      "loss": 0.4512,
      "step": 364040
    },
    {
      "epoch": 149.02,
      "grad_norm": 2.358541965484619,
      "learning_rate": 7.845893129211652e-07,
      "loss": 0.4651,
      "step": 364050
    },
    {
      "epoch": 149.02,
      "grad_norm": 1.658252477645874,
      "learning_rate": 7.844708792449803e-07,
      "loss": 0.4372,
      "step": 364060
    },
    {
      "epoch": 149.03,
      "grad_norm": 1.9345893859863281,
      "learning_rate": 7.843524527116115e-07,
      "loss": 0.4278,
      "step": 364070
    },
    {
      "epoch": 149.03,
      "grad_norm": 1.6338015794754028,
      "learning_rate": 7.842340333216012e-07,
      "loss": 0.4358,
      "step": 364080
    },
    {
      "epoch": 149.03,
      "grad_norm": 2.0314502716064453,
      "learning_rate": 7.84115621075492e-07,
      "loss": 0.4551,
      "step": 364090
    },
    {
      "epoch": 149.04,
      "grad_norm": 1.787065863609314,
      "learning_rate": 7.839972159738265e-07,
      "loss": 0.4572,
      "step": 364100
    },
    {
      "epoch": 149.04,
      "grad_norm": 2.1954901218414307,
      "learning_rate": 7.838788180171468e-07,
      "loss": 0.4391,
      "step": 364110
    },
    {
      "epoch": 149.05,
      "grad_norm": 1.9217054843902588,
      "learning_rate": 7.837604272059955e-07,
      "loss": 0.429,
      "step": 364120
    },
    {
      "epoch": 149.05,
      "grad_norm": 2.1080198287963867,
      "learning_rate": 7.836420435409148e-07,
      "loss": 0.4495,
      "step": 364130
    },
    {
      "epoch": 149.05,
      "grad_norm": 2.5224030017852783,
      "learning_rate": 7.835236670224472e-07,
      "loss": 0.4734,
      "step": 364140
    },
    {
      "epoch": 149.06,
      "grad_norm": 1.7937817573547363,
      "learning_rate": 7.834052976511347e-07,
      "loss": 0.4304,
      "step": 364150
    },
    {
      "epoch": 149.06,
      "grad_norm": 2.0794472694396973,
      "learning_rate": 7.832869354275195e-07,
      "loss": 0.4298,
      "step": 364160
    },
    {
      "epoch": 149.07,
      "grad_norm": 1.8677141666412354,
      "learning_rate": 7.83168580352144e-07,
      "loss": 0.4441,
      "step": 364170
    },
    {
      "epoch": 149.07,
      "grad_norm": 1.9978584051132202,
      "learning_rate": 7.830502324255504e-07,
      "loss": 0.463,
      "step": 364180
    },
    {
      "epoch": 149.07,
      "grad_norm": 2.031250238418579,
      "learning_rate": 7.829318916482807e-07,
      "loss": 0.4363,
      "step": 364190
    },
    {
      "epoch": 149.08,
      "grad_norm": 1.82892906665802,
      "learning_rate": 7.828135580208767e-07,
      "loss": 0.4542,
      "step": 364200
    },
    {
      "epoch": 149.08,
      "grad_norm": 2.246152639389038,
      "learning_rate": 7.82695231543882e-07,
      "loss": 0.4502,
      "step": 364210
    },
    {
      "epoch": 149.09,
      "grad_norm": 1.7980974912643433,
      "learning_rate": 7.825769122178358e-07,
      "loss": 0.4708,
      "step": 364220
    },
    {
      "epoch": 149.09,
      "grad_norm": 2.562913417816162,
      "learning_rate": 7.824586000432824e-07,
      "loss": 0.4577,
      "step": 364230
    },
    {
      "epoch": 149.1,
      "grad_norm": 2.0088775157928467,
      "learning_rate": 7.823402950207631e-07,
      "loss": 0.4552,
      "step": 364240
    },
    {
      "epoch": 149.1,
      "grad_norm": 1.8461142778396606,
      "learning_rate": 7.8222199715082e-07,
      "loss": 0.477,
      "step": 364250
    },
    {
      "epoch": 149.1,
      "grad_norm": 2.354092836380005,
      "learning_rate": 7.821037064339948e-07,
      "loss": 0.45,
      "step": 364260
    },
    {
      "epoch": 149.11,
      "grad_norm": 1.934409499168396,
      "learning_rate": 7.819854228708293e-07,
      "loss": 0.4743,
      "step": 364270
    },
    {
      "epoch": 149.11,
      "grad_norm": 1.9586424827575684,
      "learning_rate": 7.818671464618656e-07,
      "loss": 0.4492,
      "step": 364280
    },
    {
      "epoch": 149.12,
      "grad_norm": 2.3848533630371094,
      "learning_rate": 7.817488772076453e-07,
      "loss": 0.4308,
      "step": 364290
    },
    {
      "epoch": 149.12,
      "grad_norm": 2.073310136795044,
      "learning_rate": 7.816306151087102e-07,
      "loss": 0.4493,
      "step": 364300
    },
    {
      "epoch": 149.12,
      "grad_norm": 2.0506815910339355,
      "learning_rate": 7.81512360165602e-07,
      "loss": 0.4563,
      "step": 364310
    },
    {
      "epoch": 149.13,
      "grad_norm": 2.243286371231079,
      "learning_rate": 7.813941123788627e-07,
      "loss": 0.475,
      "step": 364320
    },
    {
      "epoch": 149.13,
      "grad_norm": 2.1813933849334717,
      "learning_rate": 7.812758717490335e-07,
      "loss": 0.4514,
      "step": 364330
    },
    {
      "epoch": 149.14,
      "grad_norm": 1.6672463417053223,
      "learning_rate": 7.811576382766574e-07,
      "loss": 0.4508,
      "step": 364340
    },
    {
      "epoch": 149.14,
      "grad_norm": 2.143768787384033,
      "learning_rate": 7.810394119622739e-07,
      "loss": 0.4484,
      "step": 364350
    },
    {
      "epoch": 149.14,
      "grad_norm": 2.4176132678985596,
      "learning_rate": 7.809211928064257e-07,
      "loss": 0.4429,
      "step": 364360
    },
    {
      "epoch": 149.15,
      "grad_norm": 2.128063917160034,
      "learning_rate": 7.808029808096541e-07,
      "loss": 0.4591,
      "step": 364370
    },
    {
      "epoch": 149.15,
      "grad_norm": 1.9648061990737915,
      "learning_rate": 7.806847759725008e-07,
      "loss": 0.4287,
      "step": 364380
    },
    {
      "epoch": 149.16,
      "grad_norm": 2.3148915767669678,
      "learning_rate": 7.805665782955072e-07,
      "loss": 0.4539,
      "step": 364390
    },
    {
      "epoch": 149.16,
      "grad_norm": 2.4004948139190674,
      "learning_rate": 7.804483877792148e-07,
      "loss": 0.4544,
      "step": 364400
    },
    {
      "epoch": 149.16,
      "grad_norm": 1.980599284172058,
      "learning_rate": 7.803302044241643e-07,
      "loss": 0.432,
      "step": 364410
    },
    {
      "epoch": 149.17,
      "grad_norm": 2.0037384033203125,
      "learning_rate": 7.802120282308985e-07,
      "loss": 0.4681,
      "step": 364420
    },
    {
      "epoch": 149.17,
      "grad_norm": 2.0204765796661377,
      "learning_rate": 7.800938591999579e-07,
      "loss": 0.4352,
      "step": 364430
    },
    {
      "epoch": 149.18,
      "grad_norm": 1.8452067375183105,
      "learning_rate": 7.799756973318839e-07,
      "loss": 0.4593,
      "step": 364440
    },
    {
      "epoch": 149.18,
      "grad_norm": 2.168294906616211,
      "learning_rate": 7.798575426272178e-07,
      "loss": 0.4559,
      "step": 364450
    },
    {
      "epoch": 149.19,
      "grad_norm": 1.8836987018585205,
      "learning_rate": 7.797393950865009e-07,
      "loss": 0.4609,
      "step": 364460
    },
    {
      "epoch": 149.19,
      "grad_norm": 1.8963680267333984,
      "learning_rate": 7.796212547102745e-07,
      "loss": 0.4286,
      "step": 364470
    },
    {
      "epoch": 149.19,
      "grad_norm": 1.9834786653518677,
      "learning_rate": 7.795031214990803e-07,
      "loss": 0.4479,
      "step": 364480
    },
    {
      "epoch": 149.2,
      "grad_norm": 2.2226309776306152,
      "learning_rate": 7.793849954534581e-07,
      "loss": 0.4607,
      "step": 364490
    },
    {
      "epoch": 149.2,
      "grad_norm": 1.784119725227356,
      "learning_rate": 7.792668765739494e-07,
      "loss": 0.458,
      "step": 364500
    },
    {
      "epoch": 149.21,
      "grad_norm": 1.734973430633545,
      "learning_rate": 7.791487648610961e-07,
      "loss": 0.4715,
      "step": 364510
    },
    {
      "epoch": 149.21,
      "grad_norm": 1.936335802078247,
      "learning_rate": 7.790306603154385e-07,
      "loss": 0.4382,
      "step": 364520
    },
    {
      "epoch": 149.21,
      "grad_norm": 2.2755112648010254,
      "learning_rate": 7.789125629375178e-07,
      "loss": 0.4557,
      "step": 364530
    },
    {
      "epoch": 149.22,
      "grad_norm": 2.1442370414733887,
      "learning_rate": 7.787944727278751e-07,
      "loss": 0.4624,
      "step": 364540
    },
    {
      "epoch": 149.22,
      "grad_norm": 2.0710813999176025,
      "learning_rate": 7.786763896870514e-07,
      "loss": 0.4375,
      "step": 364550
    },
    {
      "epoch": 149.23,
      "grad_norm": 2.172773599624634,
      "learning_rate": 7.785583138155872e-07,
      "loss": 0.4616,
      "step": 364560
    },
    {
      "epoch": 149.23,
      "grad_norm": 1.7749096155166626,
      "learning_rate": 7.784402451140239e-07,
      "loss": 0.4418,
      "step": 364570
    },
    {
      "epoch": 149.23,
      "grad_norm": 2.100069522857666,
      "learning_rate": 7.783221835829023e-07,
      "loss": 0.4379,
      "step": 364580
    },
    {
      "epoch": 149.24,
      "grad_norm": 1.8331767320632935,
      "learning_rate": 7.782041292227623e-07,
      "loss": 0.4426,
      "step": 364590
    },
    {
      "epoch": 149.24,
      "grad_norm": 1.760276198387146,
      "learning_rate": 7.780860820341463e-07,
      "loss": 0.4421,
      "step": 364600
    },
    {
      "epoch": 149.25,
      "grad_norm": 2.3891608715057373,
      "learning_rate": 7.779680420175941e-07,
      "loss": 0.4384,
      "step": 364610
    },
    {
      "epoch": 149.25,
      "grad_norm": 2.0259084701538086,
      "learning_rate": 7.778500091736472e-07,
      "loss": 0.4377,
      "step": 364620
    },
    {
      "epoch": 149.26,
      "grad_norm": 2.049633026123047,
      "learning_rate": 7.777319835028452e-07,
      "loss": 0.461,
      "step": 364630
    },
    {
      "epoch": 149.26,
      "grad_norm": 2.1875736713409424,
      "learning_rate": 7.77613965005729e-07,
      "loss": 0.4174,
      "step": 364640
    },
    {
      "epoch": 149.26,
      "grad_norm": 1.773838758468628,
      "learning_rate": 7.774959536828394e-07,
      "loss": 0.4268,
      "step": 364650
    },
    {
      "epoch": 149.27,
      "grad_norm": 2.0171422958374023,
      "learning_rate": 7.773779495347171e-07,
      "loss": 0.4644,
      "step": 364660
    },
    {
      "epoch": 149.27,
      "grad_norm": 2.069632053375244,
      "learning_rate": 7.772599525619025e-07,
      "loss": 0.4391,
      "step": 364670
    },
    {
      "epoch": 149.28,
      "grad_norm": 2.1050174236297607,
      "learning_rate": 7.771419627649362e-07,
      "loss": 0.4689,
      "step": 364680
    },
    {
      "epoch": 149.28,
      "grad_norm": 2.076730489730835,
      "learning_rate": 7.770239801443588e-07,
      "loss": 0.4435,
      "step": 364690
    },
    {
      "epoch": 149.28,
      "grad_norm": 1.7531415224075317,
      "learning_rate": 7.769060047007108e-07,
      "loss": 0.4758,
      "step": 364700
    },
    {
      "epoch": 149.29,
      "grad_norm": 1.7310832738876343,
      "learning_rate": 7.767880364345322e-07,
      "loss": 0.4559,
      "step": 364710
    },
    {
      "epoch": 149.29,
      "grad_norm": 2.2598319053649902,
      "learning_rate": 7.766700753463637e-07,
      "loss": 0.4288,
      "step": 364720
    },
    {
      "epoch": 149.3,
      "grad_norm": 2.0641143321990967,
      "learning_rate": 7.765521214367459e-07,
      "loss": 0.4444,
      "step": 364730
    },
    {
      "epoch": 149.3,
      "grad_norm": 1.800668478012085,
      "learning_rate": 7.764341747062188e-07,
      "loss": 0.4534,
      "step": 364740
    },
    {
      "epoch": 149.3,
      "grad_norm": 1.5513490438461304,
      "learning_rate": 7.763162351553225e-07,
      "loss": 0.4361,
      "step": 364750
    },
    {
      "epoch": 149.31,
      "grad_norm": 1.8723887205123901,
      "learning_rate": 7.761983027845978e-07,
      "loss": 0.4489,
      "step": 364760
    },
    {
      "epoch": 149.31,
      "grad_norm": 2.115121841430664,
      "learning_rate": 7.760803775945846e-07,
      "loss": 0.4464,
      "step": 364770
    },
    {
      "epoch": 149.32,
      "grad_norm": 2.426663875579834,
      "learning_rate": 7.75962459585823e-07,
      "loss": 0.4489,
      "step": 364780
    },
    {
      "epoch": 149.32,
      "grad_norm": 1.7748584747314453,
      "learning_rate": 7.758445487588533e-07,
      "loss": 0.4452,
      "step": 364790
    },
    {
      "epoch": 149.32,
      "grad_norm": 1.95224130153656,
      "learning_rate": 7.75726645114216e-07,
      "loss": 0.4722,
      "step": 364800
    },
    {
      "epoch": 149.33,
      "grad_norm": 2.075248956680298,
      "learning_rate": 7.756087486524506e-07,
      "loss": 0.4616,
      "step": 364810
    },
    {
      "epoch": 149.33,
      "grad_norm": 2.2228035926818848,
      "learning_rate": 7.754908593740975e-07,
      "loss": 0.4568,
      "step": 364820
    },
    {
      "epoch": 149.34,
      "grad_norm": 1.9860684871673584,
      "learning_rate": 7.753729772796966e-07,
      "loss": 0.4634,
      "step": 364830
    },
    {
      "epoch": 149.34,
      "grad_norm": 2.2778642177581787,
      "learning_rate": 7.752551023697879e-07,
      "loss": 0.4364,
      "step": 364840
    },
    {
      "epoch": 149.35,
      "grad_norm": 2.522287130355835,
      "learning_rate": 7.751372346449115e-07,
      "loss": 0.416,
      "step": 364850
    },
    {
      "epoch": 149.35,
      "grad_norm": 2.379779815673828,
      "learning_rate": 7.750193741056073e-07,
      "loss": 0.4555,
      "step": 364860
    },
    {
      "epoch": 149.35,
      "grad_norm": 2.181755304336548,
      "learning_rate": 7.749015207524149e-07,
      "loss": 0.4445,
      "step": 364870
    },
    {
      "epoch": 149.36,
      "grad_norm": 1.8607428073883057,
      "learning_rate": 7.747836745858745e-07,
      "loss": 0.474,
      "step": 364880
    },
    {
      "epoch": 149.36,
      "grad_norm": 2.099440813064575,
      "learning_rate": 7.74665835606526e-07,
      "loss": 0.4473,
      "step": 364890
    },
    {
      "epoch": 149.37,
      "grad_norm": 1.9575449228286743,
      "learning_rate": 7.74548003814909e-07,
      "loss": 0.4411,
      "step": 364900
    },
    {
      "epoch": 149.37,
      "grad_norm": 1.941193699836731,
      "learning_rate": 7.744301792115633e-07,
      "loss": 0.4489,
      "step": 364910
    },
    {
      "epoch": 149.37,
      "grad_norm": 1.9644449949264526,
      "learning_rate": 7.743123617970287e-07,
      "loss": 0.437,
      "step": 364920
    },
    {
      "epoch": 149.38,
      "grad_norm": 2.284590244293213,
      "learning_rate": 7.741945515718456e-07,
      "loss": 0.4569,
      "step": 364930
    },
    {
      "epoch": 149.38,
      "grad_norm": 2.1890106201171875,
      "learning_rate": 7.740767485365522e-07,
      "loss": 0.4673,
      "step": 364940
    },
    {
      "epoch": 149.39,
      "grad_norm": 2.1064348220825195,
      "learning_rate": 7.739589526916887e-07,
      "loss": 0.4536,
      "step": 364950
    },
    {
      "epoch": 149.39,
      "grad_norm": 1.9118711948394775,
      "learning_rate": 7.738411640377951e-07,
      "loss": 0.4459,
      "step": 364960
    },
    {
      "epoch": 149.39,
      "grad_norm": 1.8271011114120483,
      "learning_rate": 7.737233825754098e-07,
      "loss": 0.4488,
      "step": 364970
    },
    {
      "epoch": 149.4,
      "grad_norm": 2.190051317214966,
      "learning_rate": 7.736056083050742e-07,
      "loss": 0.4666,
      "step": 364980
    },
    {
      "epoch": 149.4,
      "grad_norm": 2.607717275619507,
      "learning_rate": 7.734878412273266e-07,
      "loss": 0.4592,
      "step": 364990
    },
    {
      "epoch": 149.41,
      "grad_norm": 2.212254047393799,
      "learning_rate": 7.733700813427073e-07,
      "loss": 0.4566,
      "step": 365000
    },
    {
      "epoch": 149.41,
      "grad_norm": 2.002082347869873,
      "learning_rate": 7.732523286517546e-07,
      "loss": 0.4524,
      "step": 365010
    },
    {
      "epoch": 149.41,
      "grad_norm": 2.045236587524414,
      "learning_rate": 7.73134583155009e-07,
      "loss": 0.4363,
      "step": 365020
    },
    {
      "epoch": 149.42,
      "grad_norm": 1.7865554094314575,
      "learning_rate": 7.730168448530091e-07,
      "loss": 0.4465,
      "step": 365030
    },
    {
      "epoch": 149.42,
      "grad_norm": 2.673086404800415,
      "learning_rate": 7.728991137462946e-07,
      "loss": 0.4462,
      "step": 365040
    },
    {
      "epoch": 149.43,
      "grad_norm": 1.8766764402389526,
      "learning_rate": 7.72781389835405e-07,
      "loss": 0.4441,
      "step": 365050
    },
    {
      "epoch": 149.43,
      "grad_norm": 2.095836639404297,
      "learning_rate": 7.726636731208796e-07,
      "loss": 0.4387,
      "step": 365060
    },
    {
      "epoch": 149.44,
      "grad_norm": 1.9324182271957397,
      "learning_rate": 7.725459636032567e-07,
      "loss": 0.4473,
      "step": 365070
    },
    {
      "epoch": 149.44,
      "grad_norm": 2.2969319820404053,
      "learning_rate": 7.724282612830765e-07,
      "loss": 0.4673,
      "step": 365080
    },
    {
      "epoch": 149.44,
      "grad_norm": 2.0321385860443115,
      "learning_rate": 7.723105661608776e-07,
      "loss": 0.4449,
      "step": 365090
    },
    {
      "epoch": 149.45,
      "grad_norm": 2.1737232208251953,
      "learning_rate": 7.721928782371993e-07,
      "loss": 0.4716,
      "step": 365100
    },
    {
      "epoch": 149.45,
      "grad_norm": 1.9139602184295654,
      "learning_rate": 7.720751975125809e-07,
      "loss": 0.4539,
      "step": 365110
    },
    {
      "epoch": 149.46,
      "grad_norm": 2.389585018157959,
      "learning_rate": 7.719575239875612e-07,
      "loss": 0.4583,
      "step": 365120
    },
    {
      "epoch": 149.46,
      "grad_norm": 1.7411108016967773,
      "learning_rate": 7.718398576626796e-07,
      "loss": 0.4446,
      "step": 365130
    },
    {
      "epoch": 149.46,
      "grad_norm": 2.1257076263427734,
      "learning_rate": 7.717221985384748e-07,
      "loss": 0.4593,
      "step": 365140
    },
    {
      "epoch": 149.47,
      "grad_norm": 1.7683830261230469,
      "learning_rate": 7.71604546615486e-07,
      "loss": 0.4386,
      "step": 365150
    },
    {
      "epoch": 149.47,
      "grad_norm": 1.9970160722732544,
      "learning_rate": 7.714869018942512e-07,
      "loss": 0.469,
      "step": 365160
    },
    {
      "epoch": 149.48,
      "grad_norm": 2.519916296005249,
      "learning_rate": 7.713692643753111e-07,
      "loss": 0.4467,
      "step": 365170
    },
    {
      "epoch": 149.48,
      "grad_norm": 1.910280466079712,
      "learning_rate": 7.712516340592034e-07,
      "loss": 0.4645,
      "step": 365180
    },
    {
      "epoch": 149.48,
      "grad_norm": 1.9580612182617188,
      "learning_rate": 7.711340109464675e-07,
      "loss": 0.4549,
      "step": 365190
    },
    {
      "epoch": 149.49,
      "grad_norm": 2.2132487297058105,
      "learning_rate": 7.710163950376423e-07,
      "loss": 0.4654,
      "step": 365200
    },
    {
      "epoch": 149.49,
      "grad_norm": 2.104160785675049,
      "learning_rate": 7.708987863332655e-07,
      "loss": 0.4412,
      "step": 365210
    },
    {
      "epoch": 149.5,
      "grad_norm": 1.745873212814331,
      "learning_rate": 7.707811848338765e-07,
      "loss": 0.4537,
      "step": 365220
    },
    {
      "epoch": 149.5,
      "grad_norm": 1.8569148778915405,
      "learning_rate": 7.706635905400143e-07,
      "loss": 0.4506,
      "step": 365230
    },
    {
      "epoch": 149.5,
      "grad_norm": 2.0770058631896973,
      "learning_rate": 7.70546003452217e-07,
      "loss": 0.4482,
      "step": 365240
    },
    {
      "epoch": 149.51,
      "grad_norm": 2.3909809589385986,
      "learning_rate": 7.704284235710238e-07,
      "loss": 0.4476,
      "step": 365250
    },
    {
      "epoch": 149.51,
      "grad_norm": 1.9473570585250854,
      "learning_rate": 7.703108508969729e-07,
      "loss": 0.4336,
      "step": 365260
    },
    {
      "epoch": 149.52,
      "grad_norm": 1.939418911933899,
      "learning_rate": 7.701932854306032e-07,
      "loss": 0.4525,
      "step": 365270
    },
    {
      "epoch": 149.52,
      "grad_norm": 2.4105300903320312,
      "learning_rate": 7.70075727172453e-07,
      "loss": 0.4558,
      "step": 365280
    },
    {
      "epoch": 149.53,
      "grad_norm": 2.069780111312866,
      "learning_rate": 7.699581761230611e-07,
      "loss": 0.4488,
      "step": 365290
    },
    {
      "epoch": 149.53,
      "grad_norm": 2.4998955726623535,
      "learning_rate": 7.698406322829657e-07,
      "loss": 0.4437,
      "step": 365300
    },
    {
      "epoch": 149.53,
      "grad_norm": 1.849923849105835,
      "learning_rate": 7.697230956527053e-07,
      "loss": 0.4512,
      "step": 365310
    },
    {
      "epoch": 149.54,
      "grad_norm": 1.7868611812591553,
      "learning_rate": 7.696055662328184e-07,
      "loss": 0.4458,
      "step": 365320
    },
    {
      "epoch": 149.54,
      "grad_norm": 1.670163631439209,
      "learning_rate": 7.694880440238436e-07,
      "loss": 0.4551,
      "step": 365330
    },
    {
      "epoch": 149.55,
      "grad_norm": 1.7453045845031738,
      "learning_rate": 7.693705290263186e-07,
      "loss": 0.462,
      "step": 365340
    },
    {
      "epoch": 149.55,
      "grad_norm": 2.2283823490142822,
      "learning_rate": 7.692530212407825e-07,
      "loss": 0.4629,
      "step": 365350
    },
    {
      "epoch": 149.55,
      "grad_norm": 2.0082316398620605,
      "learning_rate": 7.691355206677731e-07,
      "loss": 0.4504,
      "step": 365360
    },
    {
      "epoch": 149.56,
      "grad_norm": 1.864285945892334,
      "learning_rate": 7.690180273078287e-07,
      "loss": 0.465,
      "step": 365370
    },
    {
      "epoch": 149.56,
      "grad_norm": 2.3668594360351562,
      "learning_rate": 7.689005411614877e-07,
      "loss": 0.4369,
      "step": 365380
    },
    {
      "epoch": 149.57,
      "grad_norm": 1.9017738103866577,
      "learning_rate": 7.68783062229288e-07,
      "loss": 0.4541,
      "step": 365390
    },
    {
      "epoch": 149.57,
      "grad_norm": 1.9307917356491089,
      "learning_rate": 7.686655905117683e-07,
      "loss": 0.448,
      "step": 365400
    },
    {
      "epoch": 149.57,
      "grad_norm": 2.161320686340332,
      "learning_rate": 7.685481260094659e-07,
      "loss": 0.4643,
      "step": 365410
    },
    {
      "epoch": 149.58,
      "grad_norm": 3.6458985805511475,
      "learning_rate": 7.684306687229194e-07,
      "loss": 0.437,
      "step": 365420
    },
    {
      "epoch": 149.58,
      "grad_norm": 1.834590196609497,
      "learning_rate": 7.683132186526669e-07,
      "loss": 0.4504,
      "step": 365430
    },
    {
      "epoch": 149.59,
      "grad_norm": 2.10528302192688,
      "learning_rate": 7.681957757992465e-07,
      "loss": 0.4651,
      "step": 365440
    },
    {
      "epoch": 149.59,
      "grad_norm": 1.9944697618484497,
      "learning_rate": 7.680783401631957e-07,
      "loss": 0.4638,
      "step": 365450
    },
    {
      "epoch": 149.59,
      "grad_norm": 2.469749689102173,
      "learning_rate": 7.67960911745053e-07,
      "loss": 0.4522,
      "step": 365460
    },
    {
      "epoch": 149.6,
      "grad_norm": 1.8317394256591797,
      "learning_rate": 7.678434905453559e-07,
      "loss": 0.4286,
      "step": 365470
    },
    {
      "epoch": 149.6,
      "grad_norm": 1.837226390838623,
      "learning_rate": 7.677260765646423e-07,
      "loss": 0.457,
      "step": 365480
    },
    {
      "epoch": 149.61,
      "grad_norm": 2.619553565979004,
      "learning_rate": 7.676086698034505e-07,
      "loss": 0.4463,
      "step": 365490
    },
    {
      "epoch": 149.61,
      "grad_norm": 1.7858366966247559,
      "learning_rate": 7.67491270262318e-07,
      "loss": 0.4432,
      "step": 365500
    },
    {
      "epoch": 149.62,
      "grad_norm": 1.9233496189117432,
      "learning_rate": 7.673738779417833e-07,
      "loss": 0.4656,
      "step": 365510
    },
    {
      "epoch": 149.62,
      "grad_norm": 2.070765256881714,
      "learning_rate": 7.672564928423826e-07,
      "loss": 0.4526,
      "step": 365520
    },
    {
      "epoch": 149.62,
      "grad_norm": 1.856167197227478,
      "learning_rate": 7.671391149646547e-07,
      "loss": 0.4471,
      "step": 365530
    },
    {
      "epoch": 149.63,
      "grad_norm": 1.8568907976150513,
      "learning_rate": 7.670217443091364e-07,
      "loss": 0.438,
      "step": 365540
    },
    {
      "epoch": 149.63,
      "grad_norm": 2.0041146278381348,
      "learning_rate": 7.669043808763668e-07,
      "loss": 0.4666,
      "step": 365550
    },
    {
      "epoch": 149.64,
      "grad_norm": 1.586266040802002,
      "learning_rate": 7.667870246668826e-07,
      "loss": 0.4455,
      "step": 365560
    },
    {
      "epoch": 149.64,
      "grad_norm": 1.7076996564865112,
      "learning_rate": 7.666696756812216e-07,
      "loss": 0.4386,
      "step": 365570
    },
    {
      "epoch": 149.64,
      "grad_norm": 1.7615879774093628,
      "learning_rate": 7.665523339199215e-07,
      "loss": 0.4471,
      "step": 365580
    },
    {
      "epoch": 149.65,
      "grad_norm": 2.0885586738586426,
      "learning_rate": 7.664349993835195e-07,
      "loss": 0.4284,
      "step": 365590
    },
    {
      "epoch": 149.65,
      "grad_norm": 2.0845937728881836,
      "learning_rate": 7.663176720725533e-07,
      "loss": 0.4483,
      "step": 365600
    },
    {
      "epoch": 149.66,
      "grad_norm": 2.7667181491851807,
      "learning_rate": 7.662003519875601e-07,
      "loss": 0.4389,
      "step": 365610
    },
    {
      "epoch": 149.66,
      "grad_norm": 2.0815935134887695,
      "learning_rate": 7.660830391290776e-07,
      "loss": 0.4398,
      "step": 365620
    },
    {
      "epoch": 149.66,
      "grad_norm": 1.8651095628738403,
      "learning_rate": 7.659657334976433e-07,
      "loss": 0.4408,
      "step": 365630
    },
    {
      "epoch": 149.67,
      "grad_norm": 2.0209996700286865,
      "learning_rate": 7.65848435093795e-07,
      "loss": 0.4403,
      "step": 365640
    },
    {
      "epoch": 149.67,
      "grad_norm": 1.9461368322372437,
      "learning_rate": 7.657311439180687e-07,
      "loss": 0.4427,
      "step": 365650
    },
    {
      "epoch": 149.68,
      "grad_norm": 2.2571935653686523,
      "learning_rate": 7.656138599710025e-07,
      "loss": 0.4429,
      "step": 365660
    },
    {
      "epoch": 149.68,
      "grad_norm": 1.879290223121643,
      "learning_rate": 7.654965832531335e-07,
      "loss": 0.4502,
      "step": 365670
    },
    {
      "epoch": 149.68,
      "grad_norm": 2.170224905014038,
      "learning_rate": 7.653793137649987e-07,
      "loss": 0.4667,
      "step": 365680
    },
    {
      "epoch": 149.69,
      "grad_norm": 2.0240890979766846,
      "learning_rate": 7.652620515071358e-07,
      "loss": 0.4439,
      "step": 365690
    },
    {
      "epoch": 149.69,
      "grad_norm": 2.0473012924194336,
      "learning_rate": 7.651447964800818e-07,
      "loss": 0.4332,
      "step": 365700
    },
    {
      "epoch": 149.7,
      "grad_norm": 1.9502885341644287,
      "learning_rate": 7.650275486843736e-07,
      "loss": 0.4526,
      "step": 365710
    },
    {
      "epoch": 149.7,
      "grad_norm": 1.964368462562561,
      "learning_rate": 7.649103081205487e-07,
      "loss": 0.444,
      "step": 365720
    },
    {
      "epoch": 149.71,
      "grad_norm": 1.6680248975753784,
      "learning_rate": 7.647930747891427e-07,
      "loss": 0.4503,
      "step": 365730
    },
    {
      "epoch": 149.71,
      "grad_norm": 1.9518243074417114,
      "learning_rate": 7.646758486906949e-07,
      "loss": 0.4312,
      "step": 365740
    },
    {
      "epoch": 149.71,
      "grad_norm": 1.7682933807373047,
      "learning_rate": 7.645586298257413e-07,
      "loss": 0.4304,
      "step": 365750
    },
    {
      "epoch": 149.72,
      "grad_norm": 1.6432472467422485,
      "learning_rate": 7.644414181948186e-07,
      "loss": 0.4517,
      "step": 365760
    },
    {
      "epoch": 149.72,
      "grad_norm": 1.898133397102356,
      "learning_rate": 7.643242137984642e-07,
      "loss": 0.4554,
      "step": 365770
    },
    {
      "epoch": 149.73,
      "grad_norm": 1.8972853422164917,
      "learning_rate": 7.642070166372153e-07,
      "loss": 0.4343,
      "step": 365780
    },
    {
      "epoch": 149.73,
      "grad_norm": 1.8138244152069092,
      "learning_rate": 7.640898267116075e-07,
      "loss": 0.4252,
      "step": 365790
    },
    {
      "epoch": 149.73,
      "grad_norm": 1.933611512184143,
      "learning_rate": 7.639726440221782e-07,
      "loss": 0.4408,
      "step": 365800
    },
    {
      "epoch": 149.74,
      "grad_norm": 2.1256604194641113,
      "learning_rate": 7.638554685694642e-07,
      "loss": 0.4643,
      "step": 365810
    },
    {
      "epoch": 149.74,
      "grad_norm": 2.3147342205047607,
      "learning_rate": 7.637383003540025e-07,
      "loss": 0.4719,
      "step": 365820
    },
    {
      "epoch": 149.75,
      "grad_norm": 2.2112488746643066,
      "learning_rate": 7.636211393763299e-07,
      "loss": 0.438,
      "step": 365830
    },
    {
      "epoch": 149.75,
      "grad_norm": 2.2749998569488525,
      "learning_rate": 7.635039856369828e-07,
      "loss": 0.436,
      "step": 365840
    },
    {
      "epoch": 149.75,
      "grad_norm": 1.9514648914337158,
      "learning_rate": 7.633868391364977e-07,
      "loss": 0.4634,
      "step": 365850
    },
    {
      "epoch": 149.76,
      "grad_norm": 4.522523880004883,
      "learning_rate": 7.632696998754117e-07,
      "loss": 0.4526,
      "step": 365860
    },
    {
      "epoch": 149.76,
      "grad_norm": 2.1051619052886963,
      "learning_rate": 7.631525678542613e-07,
      "loss": 0.4317,
      "step": 365870
    },
    {
      "epoch": 149.77,
      "grad_norm": 2.0449750423431396,
      "learning_rate": 7.630354430735827e-07,
      "loss": 0.4629,
      "step": 365880
    },
    {
      "epoch": 149.77,
      "grad_norm": 2.3279778957366943,
      "learning_rate": 7.629183255339127e-07,
      "loss": 0.4374,
      "step": 365890
    },
    {
      "epoch": 149.77,
      "grad_norm": 1.6736252307891846,
      "learning_rate": 7.628012152357881e-07,
      "loss": 0.4517,
      "step": 365900
    },
    {
      "epoch": 149.78,
      "grad_norm": 2.876746892929077,
      "learning_rate": 7.626841121797439e-07,
      "loss": 0.4775,
      "step": 365910
    },
    {
      "epoch": 149.78,
      "grad_norm": 1.7476714849472046,
      "learning_rate": 7.625670163663194e-07,
      "loss": 0.4406,
      "step": 365920
    },
    {
      "epoch": 149.79,
      "grad_norm": 2.1293249130249023,
      "learning_rate": 7.624499277960484e-07,
      "loss": 0.453,
      "step": 365930
    },
    {
      "epoch": 149.79,
      "grad_norm": 2.2642524242401123,
      "learning_rate": 7.623328464694681e-07,
      "loss": 0.4461,
      "step": 365940
    },
    {
      "epoch": 149.8,
      "grad_norm": 2.301298141479492,
      "learning_rate": 7.62215772387115e-07,
      "loss": 0.4616,
      "step": 365950
    },
    {
      "epoch": 149.8,
      "grad_norm": 1.8234647512435913,
      "learning_rate": 7.620987055495254e-07,
      "loss": 0.4476,
      "step": 365960
    },
    {
      "epoch": 149.8,
      "grad_norm": 1.9564462900161743,
      "learning_rate": 7.619816459572351e-07,
      "loss": 0.4668,
      "step": 365970
    },
    {
      "epoch": 149.81,
      "grad_norm": 2.025263786315918,
      "learning_rate": 7.61864593610781e-07,
      "loss": 0.4463,
      "step": 365980
    },
    {
      "epoch": 149.81,
      "grad_norm": 2.145392656326294,
      "learning_rate": 7.617475485106987e-07,
      "loss": 0.4538,
      "step": 365990
    },
    {
      "epoch": 149.82,
      "grad_norm": 1.8060601949691772,
      "learning_rate": 7.616305106575245e-07,
      "loss": 0.4412,
      "step": 366000
    },
    {
      "epoch": 149.82,
      "grad_norm": 1.7645280361175537,
      "learning_rate": 7.61513480051795e-07,
      "loss": 0.4288,
      "step": 366010
    },
    {
      "epoch": 149.82,
      "grad_norm": 2.2403862476348877,
      "learning_rate": 7.613964566940459e-07,
      "loss": 0.4479,
      "step": 366020
    },
    {
      "epoch": 149.83,
      "grad_norm": 2.1748239994049072,
      "learning_rate": 7.612794405848131e-07,
      "loss": 0.4447,
      "step": 366030
    },
    {
      "epoch": 149.83,
      "grad_norm": 1.7027071714401245,
      "learning_rate": 7.61162431724633e-07,
      "loss": 0.4184,
      "step": 366040
    },
    {
      "epoch": 149.84,
      "grad_norm": 2.105170726776123,
      "learning_rate": 7.610454301140416e-07,
      "loss": 0.4694,
      "step": 366050
    },
    {
      "epoch": 149.84,
      "grad_norm": 1.9317753314971924,
      "learning_rate": 7.609284357535746e-07,
      "loss": 0.4534,
      "step": 366060
    },
    {
      "epoch": 149.84,
      "grad_norm": 2.0706465244293213,
      "learning_rate": 7.608114486437681e-07,
      "loss": 0.4566,
      "step": 366070
    },
    {
      "epoch": 149.85,
      "grad_norm": 1.9037737846374512,
      "learning_rate": 7.606944687851579e-07,
      "loss": 0.4536,
      "step": 366080
    },
    {
      "epoch": 149.85,
      "grad_norm": 1.872745156288147,
      "learning_rate": 7.605774961782805e-07,
      "loss": 0.4708,
      "step": 366090
    },
    {
      "epoch": 149.86,
      "grad_norm": 1.9957934617996216,
      "learning_rate": 7.604605308236698e-07,
      "loss": 0.431,
      "step": 366100
    },
    {
      "epoch": 149.86,
      "grad_norm": 2.8431856632232666,
      "learning_rate": 7.603435727218637e-07,
      "loss": 0.4571,
      "step": 366110
    },
    {
      "epoch": 149.86,
      "grad_norm": 1.9944276809692383,
      "learning_rate": 7.602266218733973e-07,
      "loss": 0.4672,
      "step": 366120
    },
    {
      "epoch": 149.87,
      "grad_norm": 1.8084639310836792,
      "learning_rate": 7.601096782788062e-07,
      "loss": 0.4471,
      "step": 366130
    },
    {
      "epoch": 149.87,
      "grad_norm": 2.4700944423675537,
      "learning_rate": 7.599927419386262e-07,
      "loss": 0.4632,
      "step": 366140
    },
    {
      "epoch": 149.88,
      "grad_norm": 2.1611008644104004,
      "learning_rate": 7.59875812853393e-07,
      "loss": 0.4459,
      "step": 366150
    },
    {
      "epoch": 149.88,
      "grad_norm": 2.3524253368377686,
      "learning_rate": 7.59758891023642e-07,
      "loss": 0.4457,
      "step": 366160
    },
    {
      "epoch": 149.89,
      "grad_norm": 1.9439847469329834,
      "learning_rate": 7.59641976449909e-07,
      "loss": 0.4282,
      "step": 366170
    },
    {
      "epoch": 149.89,
      "grad_norm": 2.000765562057495,
      "learning_rate": 7.595250691327295e-07,
      "loss": 0.446,
      "step": 366180
    },
    {
      "epoch": 149.89,
      "grad_norm": 2.1705665588378906,
      "learning_rate": 7.594081690726391e-07,
      "loss": 0.4446,
      "step": 366190
    },
    {
      "epoch": 149.9,
      "grad_norm": 1.9844040870666504,
      "learning_rate": 7.592912762701732e-07,
      "loss": 0.4332,
      "step": 366200
    },
    {
      "epoch": 149.9,
      "grad_norm": 2.254283905029297,
      "learning_rate": 7.591743907258672e-07,
      "loss": 0.4565,
      "step": 366210
    },
    {
      "epoch": 149.91,
      "grad_norm": 2.029111385345459,
      "learning_rate": 7.59057512440257e-07,
      "loss": 0.4565,
      "step": 366220
    },
    {
      "epoch": 149.91,
      "grad_norm": 1.8447059392929077,
      "learning_rate": 7.589406414138783e-07,
      "loss": 0.453,
      "step": 366230
    },
    {
      "epoch": 149.91,
      "grad_norm": 2.370623826980591,
      "learning_rate": 7.588237776472651e-07,
      "loss": 0.4401,
      "step": 366240
    },
    {
      "epoch": 149.92,
      "grad_norm": 2.9466137886047363,
      "learning_rate": 7.587069211409532e-07,
      "loss": 0.4501,
      "step": 366250
    },
    {
      "epoch": 149.92,
      "grad_norm": 1.915565848350525,
      "learning_rate": 7.585900718954785e-07,
      "loss": 0.4629,
      "step": 366260
    },
    {
      "epoch": 149.93,
      "grad_norm": 1.7764419317245483,
      "learning_rate": 7.584732299113759e-07,
      "loss": 0.4421,
      "step": 366270
    },
    {
      "epoch": 149.93,
      "grad_norm": 1.9025064706802368,
      "learning_rate": 7.583563951891805e-07,
      "loss": 0.4539,
      "step": 366280
    },
    {
      "epoch": 149.93,
      "grad_norm": 2.0335726737976074,
      "learning_rate": 7.58239567729427e-07,
      "loss": 0.4374,
      "step": 366290
    },
    {
      "epoch": 149.94,
      "grad_norm": 2.4509634971618652,
      "learning_rate": 7.581227475326519e-07,
      "loss": 0.4421,
      "step": 366300
    },
    {
      "epoch": 149.94,
      "grad_norm": 1.7956745624542236,
      "learning_rate": 7.5800593459939e-07,
      "loss": 0.4403,
      "step": 366310
    },
    {
      "epoch": 149.95,
      "grad_norm": 2.350604772567749,
      "learning_rate": 7.578891289301757e-07,
      "loss": 0.4475,
      "step": 366320
    },
    {
      "epoch": 149.95,
      "grad_norm": 2.1502723693847656,
      "learning_rate": 7.577723305255446e-07,
      "loss": 0.4274,
      "step": 366330
    },
    {
      "epoch": 149.95,
      "grad_norm": 1.7058032751083374,
      "learning_rate": 7.576555393860316e-07,
      "loss": 0.4282,
      "step": 366340
    },
    {
      "epoch": 149.96,
      "grad_norm": 1.9214248657226562,
      "learning_rate": 7.575387555121716e-07,
      "loss": 0.4309,
      "step": 366350
    },
    {
      "epoch": 149.96,
      "grad_norm": 2.5339903831481934,
      "learning_rate": 7.574219789045007e-07,
      "loss": 0.435,
      "step": 366360
    },
    {
      "epoch": 149.97,
      "grad_norm": 2.3460686206817627,
      "learning_rate": 7.573052095635519e-07,
      "loss": 0.4374,
      "step": 366370
    },
    {
      "epoch": 149.97,
      "grad_norm": 2.5117387771606445,
      "learning_rate": 7.571884474898606e-07,
      "loss": 0.4443,
      "step": 366380
    },
    {
      "epoch": 149.98,
      "grad_norm": 2.3425190448760986,
      "learning_rate": 7.570716926839625e-07,
      "loss": 0.4475,
      "step": 366390
    },
    {
      "epoch": 149.98,
      "grad_norm": 2.179905652999878,
      "learning_rate": 7.569549451463918e-07,
      "loss": 0.4637,
      "step": 366400
    },
    {
      "epoch": 149.98,
      "grad_norm": 2.2381584644317627,
      "learning_rate": 7.568382048776837e-07,
      "loss": 0.4423,
      "step": 366410
    },
    {
      "epoch": 149.99,
      "grad_norm": 2.071577548980713,
      "learning_rate": 7.567214718783726e-07,
      "loss": 0.4632,
      "step": 366420
    },
    {
      "epoch": 149.99,
      "grad_norm": 2.0922818183898926,
      "learning_rate": 7.566047461489936e-07,
      "loss": 0.458,
      "step": 366430
    },
    {
      "epoch": 150.0,
      "grad_norm": 3.751713752746582,
      "learning_rate": 7.56488027690081e-07,
      "loss": 0.4403,
      "step": 366440
    },
    {
      "epoch": 150.0,
      "grad_norm": 1.8690534830093384,
      "learning_rate": 7.563713165021697e-07,
      "loss": 0.4272,
      "step": 366450
    },
    {
      "epoch": 150.0,
      "eval_loss": 0.44885411858558655,
      "eval_runtime": 52.4106,
      "eval_samples_per_second": 65.807,
      "eval_steps_per_second": 8.243,
      "step": 366450
    },
    {
      "epoch": 150.0,
      "grad_norm": 2.2534143924713135,
      "learning_rate": 7.562546125857942e-07,
      "loss": 0.4504,
      "step": 366460
    },
    {
      "epoch": 150.01,
      "grad_norm": 1.863433837890625,
      "learning_rate": 7.561379159414894e-07,
      "loss": 0.4568,
      "step": 366470
    },
    {
      "epoch": 150.01,
      "grad_norm": 2.1220943927764893,
      "learning_rate": 7.56021226569789e-07,
      "loss": 0.4577,
      "step": 366480
    },
    {
      "epoch": 150.02,
      "grad_norm": 2.121410846710205,
      "learning_rate": 7.559045444712289e-07,
      "loss": 0.4527,
      "step": 366490
    },
    {
      "epoch": 150.02,
      "grad_norm": 2.0860188007354736,
      "learning_rate": 7.557878696463435e-07,
      "loss": 0.4357,
      "step": 366500
    },
    {
      "epoch": 150.02,
      "grad_norm": 1.8248093128204346,
      "learning_rate": 7.556712020956662e-07,
      "loss": 0.4677,
      "step": 366510
    },
    {
      "epoch": 150.03,
      "grad_norm": 1.691516637802124,
      "learning_rate": 7.555545418197317e-07,
      "loss": 0.4574,
      "step": 366520
    },
    {
      "epoch": 150.03,
      "grad_norm": 2.2244787216186523,
      "learning_rate": 7.554378888190749e-07,
      "loss": 0.4449,
      "step": 366530
    },
    {
      "epoch": 150.04,
      "grad_norm": 2.3332581520080566,
      "learning_rate": 7.553212430942296e-07,
      "loss": 0.4475,
      "step": 366540
    },
    {
      "epoch": 150.04,
      "grad_norm": 2.263936996459961,
      "learning_rate": 7.552046046457308e-07,
      "loss": 0.4506,
      "step": 366550
    },
    {
      "epoch": 150.05,
      "grad_norm": 1.9687330722808838,
      "learning_rate": 7.550879734741121e-07,
      "loss": 0.4348,
      "step": 366560
    },
    {
      "epoch": 150.05,
      "grad_norm": 1.9392122030258179,
      "learning_rate": 7.549713495799082e-07,
      "loss": 0.4715,
      "step": 366570
    },
    {
      "epoch": 150.05,
      "grad_norm": 1.9191553592681885,
      "learning_rate": 7.548547329636532e-07,
      "loss": 0.4535,
      "step": 366580
    },
    {
      "epoch": 150.06,
      "grad_norm": 2.0777151584625244,
      "learning_rate": 7.547381236258815e-07,
      "loss": 0.4623,
      "step": 366590
    },
    {
      "epoch": 150.06,
      "grad_norm": 2.2454097270965576,
      "learning_rate": 7.54621521567127e-07,
      "loss": 0.4563,
      "step": 366600
    },
    {
      "epoch": 150.07,
      "grad_norm": 2.3186750411987305,
      "learning_rate": 7.545049267879239e-07,
      "loss": 0.4443,
      "step": 366610
    },
    {
      "epoch": 150.07,
      "grad_norm": 2.1314282417297363,
      "learning_rate": 7.543883392888064e-07,
      "loss": 0.4468,
      "step": 366620
    },
    {
      "epoch": 150.07,
      "grad_norm": 1.8150498867034912,
      "learning_rate": 7.542717590703087e-07,
      "loss": 0.4519,
      "step": 366630
    },
    {
      "epoch": 150.08,
      "grad_norm": 1.8945832252502441,
      "learning_rate": 7.541551861329647e-07,
      "loss": 0.4589,
      "step": 366640
    },
    {
      "epoch": 150.08,
      "grad_norm": 2.424065351486206,
      "learning_rate": 7.540386204773082e-07,
      "loss": 0.4185,
      "step": 366650
    },
    {
      "epoch": 150.09,
      "grad_norm": 1.9705605506896973,
      "learning_rate": 7.53922062103873e-07,
      "loss": 0.4474,
      "step": 366660
    },
    {
      "epoch": 150.09,
      "grad_norm": 2.045975923538208,
      "learning_rate": 7.538055110131939e-07,
      "loss": 0.4523,
      "step": 366670
    },
    {
      "epoch": 150.09,
      "grad_norm": 1.7883332967758179,
      "learning_rate": 7.53688967205804e-07,
      "loss": 0.4558,
      "step": 366680
    },
    {
      "epoch": 150.1,
      "grad_norm": 1.6860188245773315,
      "learning_rate": 7.535724306822375e-07,
      "loss": 0.432,
      "step": 366690
    },
    {
      "epoch": 150.1,
      "grad_norm": 2.2858052253723145,
      "learning_rate": 7.534559014430282e-07,
      "loss": 0.4605,
      "step": 366700
    },
    {
      "epoch": 150.11,
      "grad_norm": 2.1853673458099365,
      "learning_rate": 7.533393794887097e-07,
      "loss": 0.4426,
      "step": 366710
    },
    {
      "epoch": 150.11,
      "grad_norm": 2.0335986614227295,
      "learning_rate": 7.532228648198162e-07,
      "loss": 0.4641,
      "step": 366720
    },
    {
      "epoch": 150.11,
      "grad_norm": 1.4994120597839355,
      "learning_rate": 7.53106357436881e-07,
      "loss": 0.4452,
      "step": 366730
    },
    {
      "epoch": 150.12,
      "grad_norm": 1.8687057495117188,
      "learning_rate": 7.52989857340438e-07,
      "loss": 0.4176,
      "step": 366740
    },
    {
      "epoch": 150.12,
      "grad_norm": 2.003546714782715,
      "learning_rate": 7.528733645310209e-07,
      "loss": 0.4623,
      "step": 366750
    },
    {
      "epoch": 150.13,
      "grad_norm": 2.0912327766418457,
      "learning_rate": 7.527568790091635e-07,
      "loss": 0.4447,
      "step": 366760
    },
    {
      "epoch": 150.13,
      "grad_norm": 1.8792924880981445,
      "learning_rate": 7.526404007753989e-07,
      "loss": 0.4755,
      "step": 366770
    },
    {
      "epoch": 150.14,
      "grad_norm": 1.7595903873443604,
      "learning_rate": 7.525239298302611e-07,
      "loss": 0.4557,
      "step": 366780
    },
    {
      "epoch": 150.14,
      "grad_norm": 1.8259015083312988,
      "learning_rate": 7.524074661742833e-07,
      "loss": 0.4358,
      "step": 366790
    },
    {
      "epoch": 150.14,
      "grad_norm": 1.9572348594665527,
      "learning_rate": 7.522910098079994e-07,
      "loss": 0.4375,
      "step": 366800
    },
    {
      "epoch": 150.15,
      "grad_norm": 2.1565604209899902,
      "learning_rate": 7.521745607319433e-07,
      "loss": 0.4366,
      "step": 366810
    },
    {
      "epoch": 150.15,
      "grad_norm": 1.7954788208007812,
      "learning_rate": 7.520581189466472e-07,
      "loss": 0.4346,
      "step": 366820
    },
    {
      "epoch": 150.16,
      "grad_norm": 2.094691276550293,
      "learning_rate": 7.51941684452645e-07,
      "loss": 0.4477,
      "step": 366830
    },
    {
      "epoch": 150.16,
      "grad_norm": 1.9730390310287476,
      "learning_rate": 7.518252572504703e-07,
      "loss": 0.4424,
      "step": 366840
    },
    {
      "epoch": 150.16,
      "grad_norm": 2.052851915359497,
      "learning_rate": 7.517088373406562e-07,
      "loss": 0.4607,
      "step": 366850
    },
    {
      "epoch": 150.17,
      "grad_norm": 2.5555665493011475,
      "learning_rate": 7.515924247237353e-07,
      "loss": 0.4372,
      "step": 366860
    },
    {
      "epoch": 150.17,
      "grad_norm": 1.9415645599365234,
      "learning_rate": 7.514760194002425e-07,
      "loss": 0.4521,
      "step": 366870
    },
    {
      "epoch": 150.18,
      "grad_norm": 1.9971381425857544,
      "learning_rate": 7.513596213707102e-07,
      "loss": 0.4528,
      "step": 366880
    },
    {
      "epoch": 150.18,
      "grad_norm": 2.2461018562316895,
      "learning_rate": 7.512432306356715e-07,
      "loss": 0.4575,
      "step": 366890
    },
    {
      "epoch": 150.18,
      "grad_norm": 2.008018970489502,
      "learning_rate": 7.511268471956599e-07,
      "loss": 0.4574,
      "step": 366900
    },
    {
      "epoch": 150.19,
      "grad_norm": 3.1092488765716553,
      "learning_rate": 7.510104710512083e-07,
      "loss": 0.4359,
      "step": 366910
    },
    {
      "epoch": 150.19,
      "grad_norm": 2.147796392440796,
      "learning_rate": 7.508941022028495e-07,
      "loss": 0.4576,
      "step": 366920
    },
    {
      "epoch": 150.2,
      "grad_norm": 2.1246750354766846,
      "learning_rate": 7.507777406511171e-07,
      "loss": 0.4486,
      "step": 366930
    },
    {
      "epoch": 150.2,
      "grad_norm": 1.8284647464752197,
      "learning_rate": 7.50661386396544e-07,
      "loss": 0.4694,
      "step": 366940
    },
    {
      "epoch": 150.2,
      "grad_norm": 1.7250491380691528,
      "learning_rate": 7.505450394396637e-07,
      "loss": 0.4482,
      "step": 366950
    },
    {
      "epoch": 150.21,
      "grad_norm": 2.276740550994873,
      "learning_rate": 7.50428699781008e-07,
      "loss": 0.4475,
      "step": 366960
    },
    {
      "epoch": 150.21,
      "grad_norm": 3.440793514251709,
      "learning_rate": 7.503123674211103e-07,
      "loss": 0.4549,
      "step": 366970
    },
    {
      "epoch": 150.22,
      "grad_norm": 2.0068490505218506,
      "learning_rate": 7.501960423605037e-07,
      "loss": 0.4489,
      "step": 366980
    },
    {
      "epoch": 150.22,
      "grad_norm": 2.063906192779541,
      "learning_rate": 7.50079724599721e-07,
      "loss": 0.4508,
      "step": 366990
    },
    {
      "epoch": 150.23,
      "grad_norm": 2.2584798336029053,
      "learning_rate": 7.499634141392947e-07,
      "loss": 0.4377,
      "step": 367000
    },
    {
      "epoch": 150.23,
      "grad_norm": 1.8319976329803467,
      "learning_rate": 7.498471109797584e-07,
      "loss": 0.4461,
      "step": 367010
    },
    {
      "epoch": 150.23,
      "grad_norm": 2.02832293510437,
      "learning_rate": 7.49730815121644e-07,
      "loss": 0.4498,
      "step": 367020
    },
    {
      "epoch": 150.24,
      "grad_norm": 2.185253381729126,
      "learning_rate": 7.496145265654848e-07,
      "loss": 0.4478,
      "step": 367030
    },
    {
      "epoch": 150.24,
      "grad_norm": 2.1172029972076416,
      "learning_rate": 7.494982453118132e-07,
      "loss": 0.444,
      "step": 367040
    },
    {
      "epoch": 150.25,
      "grad_norm": 1.887435793876648,
      "learning_rate": 7.493819713611614e-07,
      "loss": 0.4484,
      "step": 367050
    },
    {
      "epoch": 150.25,
      "grad_norm": 2.026069164276123,
      "learning_rate": 7.492657047140635e-07,
      "loss": 0.4505,
      "step": 367060
    },
    {
      "epoch": 150.25,
      "grad_norm": 1.7970991134643555,
      "learning_rate": 7.491494453710509e-07,
      "loss": 0.4555,
      "step": 367070
    },
    {
      "epoch": 150.26,
      "grad_norm": 1.9707098007202148,
      "learning_rate": 7.490331933326575e-07,
      "loss": 0.4386,
      "step": 367080
    },
    {
      "epoch": 150.26,
      "grad_norm": 2.273077964782715,
      "learning_rate": 7.489169485994139e-07,
      "loss": 0.4524,
      "step": 367090
    },
    {
      "epoch": 150.27,
      "grad_norm": 1.9664446115493774,
      "learning_rate": 7.488007111718536e-07,
      "loss": 0.4552,
      "step": 367100
    },
    {
      "epoch": 150.27,
      "grad_norm": 1.9927903413772583,
      "learning_rate": 7.486844810505088e-07,
      "loss": 0.457,
      "step": 367110
    },
    {
      "epoch": 150.27,
      "grad_norm": 2.225290536880493,
      "learning_rate": 7.485682582359123e-07,
      "loss": 0.4369,
      "step": 367120
    },
    {
      "epoch": 150.28,
      "grad_norm": 2.3442742824554443,
      "learning_rate": 7.484520427285964e-07,
      "loss": 0.4437,
      "step": 367130
    },
    {
      "epoch": 150.28,
      "grad_norm": 2.751577615737915,
      "learning_rate": 7.483358345290931e-07,
      "loss": 0.4516,
      "step": 367140
    },
    {
      "epoch": 150.29,
      "grad_norm": 1.7624480724334717,
      "learning_rate": 7.482196336379353e-07,
      "loss": 0.4368,
      "step": 367150
    },
    {
      "epoch": 150.29,
      "grad_norm": 2.0256710052490234,
      "learning_rate": 7.481034400556548e-07,
      "loss": 0.4401,
      "step": 367160
    },
    {
      "epoch": 150.29,
      "grad_norm": 2.2831854820251465,
      "learning_rate": 7.479872537827841e-07,
      "loss": 0.4647,
      "step": 367170
    },
    {
      "epoch": 150.3,
      "grad_norm": 2.0007565021514893,
      "learning_rate": 7.478710748198555e-07,
      "loss": 0.4425,
      "step": 367180
    },
    {
      "epoch": 150.3,
      "grad_norm": 1.84617018699646,
      "learning_rate": 7.477549031674009e-07,
      "loss": 0.4304,
      "step": 367190
    },
    {
      "epoch": 150.31,
      "grad_norm": 2.377065896987915,
      "learning_rate": 7.476387388259531e-07,
      "loss": 0.4626,
      "step": 367200
    },
    {
      "epoch": 150.31,
      "grad_norm": 1.632927417755127,
      "learning_rate": 7.475225817960434e-07,
      "loss": 0.4321,
      "step": 367210
    },
    {
      "epoch": 150.32,
      "grad_norm": 1.8113579750061035,
      "learning_rate": 7.474064320782043e-07,
      "loss": 0.446,
      "step": 367220
    },
    {
      "epoch": 150.32,
      "grad_norm": 2.0765163898468018,
      "learning_rate": 7.47290289672968e-07,
      "loss": 0.4466,
      "step": 367230
    },
    {
      "epoch": 150.32,
      "grad_norm": 2.2797012329101562,
      "learning_rate": 7.471741545808666e-07,
      "loss": 0.4607,
      "step": 367240
    },
    {
      "epoch": 150.33,
      "grad_norm": 1.9412590265274048,
      "learning_rate": 7.470580268024315e-07,
      "loss": 0.4392,
      "step": 367250
    },
    {
      "epoch": 150.33,
      "grad_norm": 2.183372735977173,
      "learning_rate": 7.469419063381951e-07,
      "loss": 0.4548,
      "step": 367260
    },
    {
      "epoch": 150.34,
      "grad_norm": 1.952893614768982,
      "learning_rate": 7.468257931886894e-07,
      "loss": 0.4455,
      "step": 367270
    },
    {
      "epoch": 150.34,
      "grad_norm": 2.041327714920044,
      "learning_rate": 7.467096873544461e-07,
      "loss": 0.4415,
      "step": 367280
    },
    {
      "epoch": 150.34,
      "grad_norm": 1.9165481328964233,
      "learning_rate": 7.465935888359974e-07,
      "loss": 0.4722,
      "step": 367290
    },
    {
      "epoch": 150.35,
      "grad_norm": 2.2471768856048584,
      "learning_rate": 7.464774976338746e-07,
      "loss": 0.4681,
      "step": 367300
    },
    {
      "epoch": 150.35,
      "grad_norm": 2.269055128097534,
      "learning_rate": 7.463614137486099e-07,
      "loss": 0.4428,
      "step": 367310
    },
    {
      "epoch": 150.36,
      "grad_norm": 1.9946424961090088,
      "learning_rate": 7.462453371807349e-07,
      "loss": 0.446,
      "step": 367320
    },
    {
      "epoch": 150.36,
      "grad_norm": 1.995008945465088,
      "learning_rate": 7.461292679307811e-07,
      "loss": 0.452,
      "step": 367330
    },
    {
      "epoch": 150.36,
      "grad_norm": 1.8510648012161255,
      "learning_rate": 7.460132059992807e-07,
      "loss": 0.4387,
      "step": 367340
    },
    {
      "epoch": 150.37,
      "grad_norm": 1.834519386291504,
      "learning_rate": 7.458971513867651e-07,
      "loss": 0.4393,
      "step": 367350
    },
    {
      "epoch": 150.37,
      "grad_norm": 1.9413803815841675,
      "learning_rate": 7.457811040937661e-07,
      "loss": 0.4552,
      "step": 367360
    },
    {
      "epoch": 150.38,
      "grad_norm": 1.9493589401245117,
      "learning_rate": 7.456650641208147e-07,
      "loss": 0.4515,
      "step": 367370
    },
    {
      "epoch": 150.38,
      "grad_norm": 2.0184695720672607,
      "learning_rate": 7.455490314684432e-07,
      "loss": 0.4178,
      "step": 367380
    },
    {
      "epoch": 150.38,
      "grad_norm": 2.1881799697875977,
      "learning_rate": 7.454330061371836e-07,
      "loss": 0.4461,
      "step": 367390
    },
    {
      "epoch": 150.39,
      "grad_norm": 1.698768138885498,
      "learning_rate": 7.453169881275658e-07,
      "loss": 0.4426,
      "step": 367400
    },
    {
      "epoch": 150.39,
      "grad_norm": 1.9360334873199463,
      "learning_rate": 7.45200977440122e-07,
      "loss": 0.4567,
      "step": 367410
    },
    {
      "epoch": 150.4,
      "grad_norm": 1.8398442268371582,
      "learning_rate": 7.45084974075383e-07,
      "loss": 0.4341,
      "step": 367420
    },
    {
      "epoch": 150.4,
      "grad_norm": 2.1805732250213623,
      "learning_rate": 7.449689780338818e-07,
      "loss": 0.4601,
      "step": 367430
    },
    {
      "epoch": 150.41,
      "grad_norm": 1.9983582496643066,
      "learning_rate": 7.448529893161488e-07,
      "loss": 0.4621,
      "step": 367440
    },
    {
      "epoch": 150.41,
      "grad_norm": 2.2843210697174072,
      "learning_rate": 7.447370079227152e-07,
      "loss": 0.4494,
      "step": 367450
    },
    {
      "epoch": 150.41,
      "grad_norm": 1.9382036924362183,
      "learning_rate": 7.446210338541125e-07,
      "loss": 0.4464,
      "step": 367460
    },
    {
      "epoch": 150.42,
      "grad_norm": 1.723590612411499,
      "learning_rate": 7.445050671108723e-07,
      "loss": 0.4415,
      "step": 367470
    },
    {
      "epoch": 150.42,
      "grad_norm": 3.621368408203125,
      "learning_rate": 7.443891076935251e-07,
      "loss": 0.4606,
      "step": 367480
    },
    {
      "epoch": 150.43,
      "grad_norm": 2.231674909591675,
      "learning_rate": 7.442731556026025e-07,
      "loss": 0.439,
      "step": 367490
    },
    {
      "epoch": 150.43,
      "grad_norm": 2.02339243888855,
      "learning_rate": 7.441572108386358e-07,
      "loss": 0.4415,
      "step": 367500
    },
    {
      "epoch": 150.43,
      "grad_norm": 1.9906260967254639,
      "learning_rate": 7.440412734021557e-07,
      "loss": 0.4566,
      "step": 367510
    },
    {
      "epoch": 150.44,
      "grad_norm": 1.8864257335662842,
      "learning_rate": 7.439253432936938e-07,
      "loss": 0.4415,
      "step": 367520
    },
    {
      "epoch": 150.44,
      "grad_norm": 2.099837303161621,
      "learning_rate": 7.438094205137812e-07,
      "loss": 0.4483,
      "step": 367530
    },
    {
      "epoch": 150.45,
      "grad_norm": 2.0593135356903076,
      "learning_rate": 7.436935050629483e-07,
      "loss": 0.4442,
      "step": 367540
    },
    {
      "epoch": 150.45,
      "grad_norm": 1.8731290102005005,
      "learning_rate": 7.435775969417263e-07,
      "loss": 0.4569,
      "step": 367550
    },
    {
      "epoch": 150.45,
      "grad_norm": 2.129404306411743,
      "learning_rate": 7.434616961506461e-07,
      "loss": 0.4397,
      "step": 367560
    },
    {
      "epoch": 150.46,
      "grad_norm": 2.0528416633605957,
      "learning_rate": 7.433458026902389e-07,
      "loss": 0.4376,
      "step": 367570
    },
    {
      "epoch": 150.46,
      "grad_norm": 3.062283515930176,
      "learning_rate": 7.432299165610354e-07,
      "loss": 0.4569,
      "step": 367580
    },
    {
      "epoch": 150.47,
      "grad_norm": 2.0087485313415527,
      "learning_rate": 7.431140377635667e-07,
      "loss": 0.4341,
      "step": 367590
    },
    {
      "epoch": 150.47,
      "grad_norm": 1.7984484434127808,
      "learning_rate": 7.429981662983634e-07,
      "loss": 0.4402,
      "step": 367600
    },
    {
      "epoch": 150.47,
      "grad_norm": 1.8348147869110107,
      "learning_rate": 7.428823021659554e-07,
      "loss": 0.452,
      "step": 367610
    },
    {
      "epoch": 150.48,
      "grad_norm": 2.0618526935577393,
      "learning_rate": 7.427664453668752e-07,
      "loss": 0.4559,
      "step": 367620
    },
    {
      "epoch": 150.48,
      "grad_norm": 2.3381001949310303,
      "learning_rate": 7.426505959016531e-07,
      "loss": 0.4251,
      "step": 367630
    },
    {
      "epoch": 150.49,
      "grad_norm": 2.1064417362213135,
      "learning_rate": 7.425347537708189e-07,
      "loss": 0.4607,
      "step": 367640
    },
    {
      "epoch": 150.49,
      "grad_norm": 1.8865892887115479,
      "learning_rate": 7.424189189749041e-07,
      "loss": 0.4447,
      "step": 367650
    },
    {
      "epoch": 150.5,
      "grad_norm": 1.9918441772460938,
      "learning_rate": 7.423030915144395e-07,
      "loss": 0.4485,
      "step": 367660
    },
    {
      "epoch": 150.5,
      "grad_norm": 2.027085781097412,
      "learning_rate": 7.421872713899547e-07,
      "loss": 0.4278,
      "step": 367670
    },
    {
      "epoch": 150.5,
      "grad_norm": 3.3666322231292725,
      "learning_rate": 7.420714586019805e-07,
      "loss": 0.4377,
      "step": 367680
    },
    {
      "epoch": 150.51,
      "grad_norm": 2.1275956630706787,
      "learning_rate": 7.419556531510478e-07,
      "loss": 0.4304,
      "step": 367690
    },
    {
      "epoch": 150.51,
      "grad_norm": 2.232039213180542,
      "learning_rate": 7.418398550376866e-07,
      "loss": 0.4361,
      "step": 367700
    },
    {
      "epoch": 150.52,
      "grad_norm": 3.5527825355529785,
      "learning_rate": 7.417240642624279e-07,
      "loss": 0.4441,
      "step": 367710
    },
    {
      "epoch": 150.52,
      "grad_norm": 1.9703037738800049,
      "learning_rate": 7.416082808258022e-07,
      "loss": 0.4618,
      "step": 367720
    },
    {
      "epoch": 150.52,
      "grad_norm": 2.316434144973755,
      "learning_rate": 7.414925047283392e-07,
      "loss": 0.4598,
      "step": 367730
    },
    {
      "epoch": 150.53,
      "grad_norm": 2.165876865386963,
      "learning_rate": 7.4137673597057e-07,
      "loss": 0.4289,
      "step": 367740
    },
    {
      "epoch": 150.53,
      "grad_norm": 1.8852347135543823,
      "learning_rate": 7.412609745530242e-07,
      "loss": 0.442,
      "step": 367750
    },
    {
      "epoch": 150.54,
      "grad_norm": 2.059398651123047,
      "learning_rate": 7.411452204762324e-07,
      "loss": 0.4506,
      "step": 367760
    },
    {
      "epoch": 150.54,
      "grad_norm": 2.166815757751465,
      "learning_rate": 7.410294737407253e-07,
      "loss": 0.466,
      "step": 367770
    },
    {
      "epoch": 150.54,
      "grad_norm": 2.1598060131073,
      "learning_rate": 7.409137343470323e-07,
      "loss": 0.46,
      "step": 367780
    },
    {
      "epoch": 150.55,
      "grad_norm": 2.33768367767334,
      "learning_rate": 7.407980022956843e-07,
      "loss": 0.4357,
      "step": 367790
    },
    {
      "epoch": 150.55,
      "grad_norm": 2.0947628021240234,
      "learning_rate": 7.40682277587211e-07,
      "loss": 0.4652,
      "step": 367800
    },
    {
      "epoch": 150.56,
      "grad_norm": 3.798344135284424,
      "learning_rate": 7.405665602221428e-07,
      "loss": 0.449,
      "step": 367810
    },
    {
      "epoch": 150.56,
      "grad_norm": 1.6480348110198975,
      "learning_rate": 7.404508502010097e-07,
      "loss": 0.4457,
      "step": 367820
    },
    {
      "epoch": 150.56,
      "grad_norm": 2.1326894760131836,
      "learning_rate": 7.403351475243414e-07,
      "loss": 0.4451,
      "step": 367830
    },
    {
      "epoch": 150.57,
      "grad_norm": 1.880342721939087,
      "learning_rate": 7.402194521926686e-07,
      "loss": 0.4633,
      "step": 367840
    },
    {
      "epoch": 150.57,
      "grad_norm": 1.9247157573699951,
      "learning_rate": 7.401037642065207e-07,
      "loss": 0.4436,
      "step": 367850
    },
    {
      "epoch": 150.58,
      "grad_norm": 1.9448567628860474,
      "learning_rate": 7.399880835664276e-07,
      "loss": 0.4509,
      "step": 367860
    },
    {
      "epoch": 150.58,
      "grad_norm": 2.0128777027130127,
      "learning_rate": 7.398724102729199e-07,
      "loss": 0.4408,
      "step": 367870
    },
    {
      "epoch": 150.59,
      "grad_norm": 1.738150715827942,
      "learning_rate": 7.39756744326527e-07,
      "loss": 0.4214,
      "step": 367880
    },
    {
      "epoch": 150.59,
      "grad_norm": 1.9037436246871948,
      "learning_rate": 7.396410857277785e-07,
      "loss": 0.4293,
      "step": 367890
    },
    {
      "epoch": 150.59,
      "grad_norm": 1.5902190208435059,
      "learning_rate": 7.395254344772048e-07,
      "loss": 0.4616,
      "step": 367900
    },
    {
      "epoch": 150.6,
      "grad_norm": 2.1577835083007812,
      "learning_rate": 7.394097905753352e-07,
      "loss": 0.4351,
      "step": 367910
    },
    {
      "epoch": 150.6,
      "grad_norm": 2.0557847023010254,
      "learning_rate": 7.392941540226997e-07,
      "loss": 0.441,
      "step": 367920
    },
    {
      "epoch": 150.61,
      "grad_norm": 6.265913486480713,
      "learning_rate": 7.39178524819828e-07,
      "loss": 0.447,
      "step": 367930
    },
    {
      "epoch": 150.61,
      "grad_norm": 2.29459547996521,
      "learning_rate": 7.390629029672496e-07,
      "loss": 0.4609,
      "step": 367940
    },
    {
      "epoch": 150.61,
      "grad_norm": 2.477365493774414,
      "learning_rate": 7.389472884654943e-07,
      "loss": 0.4788,
      "step": 367950
    },
    {
      "epoch": 150.62,
      "grad_norm": 2.012425422668457,
      "learning_rate": 7.388316813150919e-07,
      "loss": 0.4641,
      "step": 367960
    },
    {
      "epoch": 150.62,
      "grad_norm": 1.8310999870300293,
      "learning_rate": 7.387160815165724e-07,
      "loss": 0.4478,
      "step": 367970
    },
    {
      "epoch": 150.63,
      "grad_norm": 2.007465124130249,
      "learning_rate": 7.386004890704639e-07,
      "loss": 0.4625,
      "step": 367980
    },
    {
      "epoch": 150.63,
      "grad_norm": 1.9187878370285034,
      "learning_rate": 7.384849039772959e-07,
      "loss": 0.445,
      "step": 367990
    },
    {
      "epoch": 150.63,
      "grad_norm": 2.009063720703125,
      "learning_rate": 7.383693262375996e-07,
      "loss": 0.4375,
      "step": 368000
    },
    {
      "epoch": 150.64,
      "grad_norm": 1.892037034034729,
      "learning_rate": 7.382537558519033e-07,
      "loss": 0.4513,
      "step": 368010
    },
    {
      "epoch": 150.64,
      "grad_norm": 2.0930800437927246,
      "learning_rate": 7.38138192820737e-07,
      "loss": 0.4348,
      "step": 368020
    },
    {
      "epoch": 150.65,
      "grad_norm": 2.362534284591675,
      "learning_rate": 7.380226371446294e-07,
      "loss": 0.4541,
      "step": 368030
    },
    {
      "epoch": 150.65,
      "grad_norm": 3.1084961891174316,
      "learning_rate": 7.379070888241104e-07,
      "loss": 0.456,
      "step": 368040
    },
    {
      "epoch": 150.65,
      "grad_norm": 2.0805552005767822,
      "learning_rate": 7.377915478597091e-07,
      "loss": 0.475,
      "step": 368050
    },
    {
      "epoch": 150.66,
      "grad_norm": 1.9131898880004883,
      "learning_rate": 7.376760142519547e-07,
      "loss": 0.4262,
      "step": 368060
    },
    {
      "epoch": 150.66,
      "grad_norm": 1.8677172660827637,
      "learning_rate": 7.375604880013764e-07,
      "loss": 0.4626,
      "step": 368070
    },
    {
      "epoch": 150.67,
      "grad_norm": 1.8668562173843384,
      "learning_rate": 7.374449691085035e-07,
      "loss": 0.4534,
      "step": 368080
    },
    {
      "epoch": 150.67,
      "grad_norm": 2.2513959407806396,
      "learning_rate": 7.373294575738654e-07,
      "loss": 0.434,
      "step": 368090
    },
    {
      "epoch": 150.68,
      "grad_norm": 2.151198625564575,
      "learning_rate": 7.372139533979908e-07,
      "loss": 0.4383,
      "step": 368100
    },
    {
      "epoch": 150.68,
      "grad_norm": 2.284538745880127,
      "learning_rate": 7.370984565814099e-07,
      "loss": 0.4293,
      "step": 368110
    },
    {
      "epoch": 150.68,
      "grad_norm": 2.4009411334991455,
      "learning_rate": 7.3698296712465e-07,
      "loss": 0.427,
      "step": 368120
    },
    {
      "epoch": 150.69,
      "grad_norm": 2.072502851486206,
      "learning_rate": 7.36867485028241e-07,
      "loss": 0.4441,
      "step": 368130
    },
    {
      "epoch": 150.69,
      "grad_norm": 1.8749122619628906,
      "learning_rate": 7.367520102927122e-07,
      "loss": 0.455,
      "step": 368140
    },
    {
      "epoch": 150.7,
      "grad_norm": 2.016030788421631,
      "learning_rate": 7.366365429185922e-07,
      "loss": 0.4357,
      "step": 368150
    },
    {
      "epoch": 150.7,
      "grad_norm": 3.2038843631744385,
      "learning_rate": 7.365210829064102e-07,
      "loss": 0.4434,
      "step": 368160
    },
    {
      "epoch": 150.7,
      "grad_norm": 1.9629877805709839,
      "learning_rate": 7.364056302566946e-07,
      "loss": 0.4415,
      "step": 368170
    },
    {
      "epoch": 150.71,
      "grad_norm": 2.0040814876556396,
      "learning_rate": 7.362901849699741e-07,
      "loss": 0.4623,
      "step": 368180
    },
    {
      "epoch": 150.71,
      "grad_norm": 2.0381314754486084,
      "learning_rate": 7.361747470467788e-07,
      "loss": 0.4467,
      "step": 368190
    },
    {
      "epoch": 150.72,
      "grad_norm": 2.0936784744262695,
      "learning_rate": 7.360593164876367e-07,
      "loss": 0.455,
      "step": 368200
    },
    {
      "epoch": 150.72,
      "grad_norm": 1.7062187194824219,
      "learning_rate": 7.359438932930768e-07,
      "loss": 0.4492,
      "step": 368210
    },
    {
      "epoch": 150.72,
      "grad_norm": 2.3100264072418213,
      "learning_rate": 7.358284774636275e-07,
      "loss": 0.4617,
      "step": 368220
    },
    {
      "epoch": 150.73,
      "grad_norm": 2.2641875743865967,
      "learning_rate": 7.357130689998177e-07,
      "loss": 0.458,
      "step": 368230
    },
    {
      "epoch": 150.73,
      "grad_norm": 2.0889248847961426,
      "learning_rate": 7.355976679021763e-07,
      "loss": 0.4441,
      "step": 368240
    },
    {
      "epoch": 150.74,
      "grad_norm": 2.3743996620178223,
      "learning_rate": 7.354822741712322e-07,
      "loss": 0.4517,
      "step": 368250
    },
    {
      "epoch": 150.74,
      "grad_norm": 1.6993136405944824,
      "learning_rate": 7.353668878075128e-07,
      "loss": 0.419,
      "step": 368260
    },
    {
      "epoch": 150.74,
      "grad_norm": 1.9117261171340942,
      "learning_rate": 7.352515088115473e-07,
      "loss": 0.4457,
      "step": 368270
    },
    {
      "epoch": 150.75,
      "grad_norm": 1.9869087934494019,
      "learning_rate": 7.351361371838641e-07,
      "loss": 0.4526,
      "step": 368280
    },
    {
      "epoch": 150.75,
      "grad_norm": 1.9858582019805908,
      "learning_rate": 7.350207729249921e-07,
      "loss": 0.4629,
      "step": 368290
    },
    {
      "epoch": 150.76,
      "grad_norm": 2.320737838745117,
      "learning_rate": 7.349054160354597e-07,
      "loss": 0.456,
      "step": 368300
    },
    {
      "epoch": 150.76,
      "grad_norm": 2.036303997039795,
      "learning_rate": 7.347900665157948e-07,
      "loss": 0.473,
      "step": 368310
    },
    {
      "epoch": 150.77,
      "grad_norm": 1.729739785194397,
      "learning_rate": 7.346747243665264e-07,
      "loss": 0.4672,
      "step": 368320
    },
    {
      "epoch": 150.77,
      "grad_norm": 2.0942280292510986,
      "learning_rate": 7.345593895881827e-07,
      "loss": 0.4605,
      "step": 368330
    },
    {
      "epoch": 150.77,
      "grad_norm": 2.9372501373291016,
      "learning_rate": 7.344440621812917e-07,
      "loss": 0.4472,
      "step": 368340
    },
    {
      "epoch": 150.78,
      "grad_norm": 2.019665002822876,
      "learning_rate": 7.343287421463824e-07,
      "loss": 0.462,
      "step": 368350
    },
    {
      "epoch": 150.78,
      "grad_norm": 2.2491455078125,
      "learning_rate": 7.342134294839821e-07,
      "loss": 0.443,
      "step": 368360
    },
    {
      "epoch": 150.79,
      "grad_norm": 2.118765354156494,
      "learning_rate": 7.340981241946192e-07,
      "loss": 0.4469,
      "step": 368370
    },
    {
      "epoch": 150.79,
      "grad_norm": 2.0711464881896973,
      "learning_rate": 7.339828262788237e-07,
      "loss": 0.4372,
      "step": 368380
    },
    {
      "epoch": 150.79,
      "grad_norm": 2.3104536533355713,
      "learning_rate": 7.338675357371215e-07,
      "loss": 0.4571,
      "step": 368390
    },
    {
      "epoch": 150.8,
      "grad_norm": 2.5154006481170654,
      "learning_rate": 7.337522525700415e-07,
      "loss": 0.4472,
      "step": 368400
    },
    {
      "epoch": 150.8,
      "grad_norm": 2.2560269832611084,
      "learning_rate": 7.33636976778112e-07,
      "loss": 0.454,
      "step": 368410
    },
    {
      "epoch": 150.81,
      "grad_norm": 1.8709534406661987,
      "learning_rate": 7.335217083618606e-07,
      "loss": 0.4286,
      "step": 368420
    },
    {
      "epoch": 150.81,
      "grad_norm": 2.0381805896759033,
      "learning_rate": 7.33406447321816e-07,
      "loss": 0.4562,
      "step": 368430
    },
    {
      "epoch": 150.81,
      "grad_norm": 2.0262463092803955,
      "learning_rate": 7.332911936585058e-07,
      "loss": 0.4529,
      "step": 368440
    },
    {
      "epoch": 150.82,
      "grad_norm": 1.8330073356628418,
      "learning_rate": 7.331759473724577e-07,
      "loss": 0.4442,
      "step": 368450
    },
    {
      "epoch": 150.82,
      "grad_norm": 2.6297242641448975,
      "learning_rate": 7.330607084642e-07,
      "loss": 0.4191,
      "step": 368460
    },
    {
      "epoch": 150.83,
      "grad_norm": 1.6896657943725586,
      "learning_rate": 7.329454769342607e-07,
      "loss": 0.4263,
      "step": 368470
    },
    {
      "epoch": 150.83,
      "grad_norm": 1.5822092294692993,
      "learning_rate": 7.328302527831673e-07,
      "loss": 0.4498,
      "step": 368480
    },
    {
      "epoch": 150.84,
      "grad_norm": 1.744203805923462,
      "learning_rate": 7.327150360114477e-07,
      "loss": 0.4354,
      "step": 368490
    },
    {
      "epoch": 150.84,
      "grad_norm": 2.1907174587249756,
      "learning_rate": 7.325998266196299e-07,
      "loss": 0.4256,
      "step": 368500
    },
    {
      "epoch": 150.84,
      "grad_norm": 1.8169561624526978,
      "learning_rate": 7.324846246082414e-07,
      "loss": 0.429,
      "step": 368510
    },
    {
      "epoch": 150.85,
      "grad_norm": 1.917022466659546,
      "learning_rate": 7.323694299778103e-07,
      "loss": 0.4423,
      "step": 368520
    },
    {
      "epoch": 150.85,
      "grad_norm": 1.9384297132492065,
      "learning_rate": 7.322542427288637e-07,
      "loss": 0.4685,
      "step": 368530
    },
    {
      "epoch": 150.86,
      "grad_norm": 2.0314109325408936,
      "learning_rate": 7.321390628619297e-07,
      "loss": 0.4578,
      "step": 368540
    },
    {
      "epoch": 150.86,
      "grad_norm": 2.0737318992614746,
      "learning_rate": 7.320238903775359e-07,
      "loss": 0.4536,
      "step": 368550
    },
    {
      "epoch": 150.86,
      "grad_norm": 1.7560786008834839,
      "learning_rate": 7.3190872527621e-07,
      "loss": 0.4325,
      "step": 368560
    },
    {
      "epoch": 150.87,
      "grad_norm": 1.5518581867218018,
      "learning_rate": 7.317935675584791e-07,
      "loss": 0.4341,
      "step": 368570
    },
    {
      "epoch": 150.87,
      "grad_norm": 2.1754038333892822,
      "learning_rate": 7.316784172248709e-07,
      "loss": 0.4607,
      "step": 368580
    },
    {
      "epoch": 150.88,
      "grad_norm": 1.8121180534362793,
      "learning_rate": 7.315632742759129e-07,
      "loss": 0.4294,
      "step": 368590
    },
    {
      "epoch": 150.88,
      "grad_norm": 2.057421922683716,
      "learning_rate": 7.314481387121328e-07,
      "loss": 0.455,
      "step": 368600
    },
    {
      "epoch": 150.88,
      "grad_norm": 2.3006463050842285,
      "learning_rate": 7.313330105340579e-07,
      "loss": 0.4377,
      "step": 368610
    },
    {
      "epoch": 150.89,
      "grad_norm": 1.9928041696548462,
      "learning_rate": 7.312178897422153e-07,
      "loss": 0.4385,
      "step": 368620
    },
    {
      "epoch": 150.89,
      "grad_norm": 1.9278154373168945,
      "learning_rate": 7.311027763371328e-07,
      "loss": 0.4686,
      "step": 368630
    },
    {
      "epoch": 150.9,
      "grad_norm": 1.9414507150650024,
      "learning_rate": 7.309876703193374e-07,
      "loss": 0.4358,
      "step": 368640
    },
    {
      "epoch": 150.9,
      "grad_norm": 1.9567487239837646,
      "learning_rate": 7.308725716893565e-07,
      "loss": 0.4539,
      "step": 368650
    },
    {
      "epoch": 150.9,
      "grad_norm": 1.9460790157318115,
      "learning_rate": 7.307574804477171e-07,
      "loss": 0.4802,
      "step": 368660
    },
    {
      "epoch": 150.91,
      "grad_norm": 2.0469751358032227,
      "learning_rate": 7.306423965949467e-07,
      "loss": 0.4579,
      "step": 368670
    },
    {
      "epoch": 150.91,
      "grad_norm": 2.4460396766662598,
      "learning_rate": 7.305273201315725e-07,
      "loss": 0.4259,
      "step": 368680
    },
    {
      "epoch": 150.92,
      "grad_norm": 1.6819183826446533,
      "learning_rate": 7.304122510581223e-07,
      "loss": 0.4534,
      "step": 368690
    },
    {
      "epoch": 150.92,
      "grad_norm": 1.636765718460083,
      "learning_rate": 7.302971893751219e-07,
      "loss": 0.4313,
      "step": 368700
    },
    {
      "epoch": 150.93,
      "grad_norm": 2.297076940536499,
      "learning_rate": 7.301821350830988e-07,
      "loss": 0.4456,
      "step": 368710
    },
    {
      "epoch": 150.93,
      "grad_norm": 2.0064351558685303,
      "learning_rate": 7.3006708818258e-07,
      "loss": 0.4526,
      "step": 368720
    },
    {
      "epoch": 150.93,
      "grad_norm": 1.857789158821106,
      "learning_rate": 7.299520486740929e-07,
      "loss": 0.4537,
      "step": 368730
    },
    {
      "epoch": 150.94,
      "grad_norm": 2.408332347869873,
      "learning_rate": 7.298370165581636e-07,
      "loss": 0.4766,
      "step": 368740
    },
    {
      "epoch": 150.94,
      "grad_norm": 2.2620601654052734,
      "learning_rate": 7.297219918353202e-07,
      "loss": 0.4573,
      "step": 368750
    },
    {
      "epoch": 150.95,
      "grad_norm": 2.0954980850219727,
      "learning_rate": 7.296069745060893e-07,
      "loss": 0.4525,
      "step": 368760
    },
    {
      "epoch": 150.95,
      "grad_norm": 2.2336313724517822,
      "learning_rate": 7.294919645709976e-07,
      "loss": 0.4536,
      "step": 368770
    },
    {
      "epoch": 150.95,
      "grad_norm": 2.2107512950897217,
      "learning_rate": 7.293769620305721e-07,
      "loss": 0.4502,
      "step": 368780
    },
    {
      "epoch": 150.96,
      "grad_norm": 1.8922728300094604,
      "learning_rate": 7.292619668853394e-07,
      "loss": 0.4507,
      "step": 368790
    },
    {
      "epoch": 150.96,
      "grad_norm": 1.9700613021850586,
      "learning_rate": 7.291469791358264e-07,
      "loss": 0.446,
      "step": 368800
    },
    {
      "epoch": 150.97,
      "grad_norm": 1.9772785902023315,
      "learning_rate": 7.290319987825597e-07,
      "loss": 0.4365,
      "step": 368810
    },
    {
      "epoch": 150.97,
      "grad_norm": 2.148648262023926,
      "learning_rate": 7.289170258260662e-07,
      "loss": 0.4652,
      "step": 368820
    },
    {
      "epoch": 150.97,
      "grad_norm": 1.9667599201202393,
      "learning_rate": 7.28802060266873e-07,
      "loss": 0.4389,
      "step": 368830
    },
    {
      "epoch": 150.98,
      "grad_norm": 1.9340258836746216,
      "learning_rate": 7.286871021055055e-07,
      "loss": 0.4515,
      "step": 368840
    },
    {
      "epoch": 150.98,
      "grad_norm": 2.1100473403930664,
      "learning_rate": 7.285721513424912e-07,
      "loss": 0.4437,
      "step": 368850
    },
    {
      "epoch": 150.99,
      "grad_norm": 2.014394521713257,
      "learning_rate": 7.284572079783564e-07,
      "loss": 0.4378,
      "step": 368860
    },
    {
      "epoch": 150.99,
      "grad_norm": 2.2984743118286133,
      "learning_rate": 7.283422720136277e-07,
      "loss": 0.462,
      "step": 368870
    },
    {
      "epoch": 150.99,
      "grad_norm": 1.9300740957260132,
      "learning_rate": 7.282273434488314e-07,
      "loss": 0.4223,
      "step": 368880
    },
    {
      "epoch": 151.0,
      "grad_norm": 2.5201210975646973,
      "learning_rate": 7.281124222844946e-07,
      "loss": 0.439,
      "step": 368890
    },
    {
      "epoch": 151.0,
      "eval_loss": 0.45039981603622437,
      "eval_runtime": 58.4856,
      "eval_samples_per_second": 58.972,
      "eval_steps_per_second": 7.386,
      "step": 368893
    },
    {
      "epoch": 151.0,
      "grad_norm": 1.905393362045288,
      "learning_rate": 7.279975085211431e-07,
      "loss": 0.4537,
      "step": 368900
    },
    {
      "epoch": 151.01,
      "grad_norm": 1.8804429769515991,
      "learning_rate": 7.278826021593034e-07,
      "loss": 0.4474,
      "step": 368910
    },
    {
      "epoch": 151.01,
      "grad_norm": 2.658839464187622,
      "learning_rate": 7.277677031995022e-07,
      "loss": 0.4759,
      "step": 368920
    },
    {
      "epoch": 151.02,
      "grad_norm": 2.3666019439697266,
      "learning_rate": 7.276528116422649e-07,
      "loss": 0.4469,
      "step": 368930
    },
    {
      "epoch": 151.02,
      "grad_norm": 2.546642780303955,
      "learning_rate": 7.275379274881191e-07,
      "loss": 0.4544,
      "step": 368940
    },
    {
      "epoch": 151.02,
      "grad_norm": 1.8340986967086792,
      "learning_rate": 7.274230507375909e-07,
      "loss": 0.4626,
      "step": 368950
    },
    {
      "epoch": 151.03,
      "grad_norm": 2.193497657775879,
      "learning_rate": 7.273081813912058e-07,
      "loss": 0.4437,
      "step": 368960
    },
    {
      "epoch": 151.03,
      "grad_norm": 1.9128422737121582,
      "learning_rate": 7.271933194494911e-07,
      "loss": 0.4459,
      "step": 368970
    },
    {
      "epoch": 151.04,
      "grad_norm": 2.3073883056640625,
      "learning_rate": 7.270784649129715e-07,
      "loss": 0.4351,
      "step": 368980
    },
    {
      "epoch": 151.04,
      "grad_norm": 2.4385175704956055,
      "learning_rate": 7.269636177821739e-07,
      "loss": 0.4558,
      "step": 368990
    },
    {
      "epoch": 151.04,
      "grad_norm": 1.902502179145813,
      "learning_rate": 7.268487780576243e-07,
      "loss": 0.4479,
      "step": 369000
    },
    {
      "epoch": 151.05,
      "grad_norm": 2.3584482669830322,
      "learning_rate": 7.267339457398487e-07,
      "loss": 0.4346,
      "step": 369010
    },
    {
      "epoch": 151.05,
      "grad_norm": 1.6335288286209106,
      "learning_rate": 7.266191208293733e-07,
      "loss": 0.4317,
      "step": 369020
    },
    {
      "epoch": 151.06,
      "grad_norm": 1.9362094402313232,
      "learning_rate": 7.26504303326724e-07,
      "loss": 0.4507,
      "step": 369030
    },
    {
      "epoch": 151.06,
      "grad_norm": 2.1944358348846436,
      "learning_rate": 7.263894932324268e-07,
      "loss": 0.4555,
      "step": 369040
    },
    {
      "epoch": 151.06,
      "grad_norm": 1.7740814685821533,
      "learning_rate": 7.262746905470075e-07,
      "loss": 0.4306,
      "step": 369050
    },
    {
      "epoch": 151.07,
      "grad_norm": 2.205799102783203,
      "learning_rate": 7.261598952709921e-07,
      "loss": 0.4216,
      "step": 369060
    },
    {
      "epoch": 151.07,
      "grad_norm": 2.738668203353882,
      "learning_rate": 7.260451074049066e-07,
      "loss": 0.458,
      "step": 369070
    },
    {
      "epoch": 151.08,
      "grad_norm": 2.266916036605835,
      "learning_rate": 7.259303269492764e-07,
      "loss": 0.4566,
      "step": 369080
    },
    {
      "epoch": 151.08,
      "grad_norm": 2.1131629943847656,
      "learning_rate": 7.25815553904628e-07,
      "loss": 0.471,
      "step": 369090
    },
    {
      "epoch": 151.08,
      "grad_norm": 3.5887112617492676,
      "learning_rate": 7.257007882714863e-07,
      "loss": 0.4499,
      "step": 369100
    },
    {
      "epoch": 151.09,
      "grad_norm": 2.0767714977264404,
      "learning_rate": 7.255860300503777e-07,
      "loss": 0.4553,
      "step": 369110
    },
    {
      "epoch": 151.09,
      "grad_norm": 1.7963826656341553,
      "learning_rate": 7.254712792418276e-07,
      "loss": 0.4343,
      "step": 369120
    },
    {
      "epoch": 151.1,
      "grad_norm": 1.8872435092926025,
      "learning_rate": 7.253565358463617e-07,
      "loss": 0.4401,
      "step": 369130
    },
    {
      "epoch": 151.1,
      "grad_norm": 1.839564561843872,
      "learning_rate": 7.252417998645055e-07,
      "loss": 0.4602,
      "step": 369140
    },
    {
      "epoch": 151.11,
      "grad_norm": 2.3725531101226807,
      "learning_rate": 7.251270712967849e-07,
      "loss": 0.454,
      "step": 369150
    },
    {
      "epoch": 151.11,
      "grad_norm": 2.0547051429748535,
      "learning_rate": 7.250123501437251e-07,
      "loss": 0.4582,
      "step": 369160
    },
    {
      "epoch": 151.11,
      "grad_norm": 3.632615566253662,
      "learning_rate": 7.248976364058521e-07,
      "loss": 0.4597,
      "step": 369170
    },
    {
      "epoch": 151.12,
      "grad_norm": 2.075984477996826,
      "learning_rate": 7.247829300836907e-07,
      "loss": 0.4462,
      "step": 369180
    },
    {
      "epoch": 151.12,
      "grad_norm": 1.9079362154006958,
      "learning_rate": 7.246682311777669e-07,
      "loss": 0.4485,
      "step": 369190
    },
    {
      "epoch": 151.13,
      "grad_norm": 2.5185680389404297,
      "learning_rate": 7.245535396886058e-07,
      "loss": 0.4418,
      "step": 369200
    },
    {
      "epoch": 151.13,
      "grad_norm": 2.1031999588012695,
      "learning_rate": 7.244388556167329e-07,
      "loss": 0.4585,
      "step": 369210
    },
    {
      "epoch": 151.13,
      "grad_norm": 2.040005683898926,
      "learning_rate": 7.243241789626736e-07,
      "loss": 0.4332,
      "step": 369220
    },
    {
      "epoch": 151.14,
      "grad_norm": 1.795915126800537,
      "learning_rate": 7.242095097269534e-07,
      "loss": 0.4419,
      "step": 369230
    },
    {
      "epoch": 151.14,
      "grad_norm": 2.051563024520874,
      "learning_rate": 7.240948479100972e-07,
      "loss": 0.4574,
      "step": 369240
    },
    {
      "epoch": 151.15,
      "grad_norm": 2.5513315200805664,
      "learning_rate": 7.239801935126305e-07,
      "loss": 0.4586,
      "step": 369250
    },
    {
      "epoch": 151.15,
      "grad_norm": 1.969773530960083,
      "learning_rate": 7.238655465350784e-07,
      "loss": 0.4424,
      "step": 369260
    },
    {
      "epoch": 151.15,
      "grad_norm": 2.1242780685424805,
      "learning_rate": 7.23750906977966e-07,
      "loss": 0.4403,
      "step": 369270
    },
    {
      "epoch": 151.16,
      "grad_norm": 2.081287145614624,
      "learning_rate": 7.236362748418194e-07,
      "loss": 0.4571,
      "step": 369280
    },
    {
      "epoch": 151.16,
      "grad_norm": 2.1221649646759033,
      "learning_rate": 7.235216501271619e-07,
      "loss": 0.4563,
      "step": 369290
    },
    {
      "epoch": 151.17,
      "grad_norm": 2.1462957859039307,
      "learning_rate": 7.234070328345198e-07,
      "loss": 0.4552,
      "step": 369300
    },
    {
      "epoch": 151.17,
      "grad_norm": 1.9335399866104126,
      "learning_rate": 7.232924229644171e-07,
      "loss": 0.4611,
      "step": 369310
    },
    {
      "epoch": 151.17,
      "grad_norm": 2.0154736042022705,
      "learning_rate": 7.231778205173805e-07,
      "loss": 0.4474,
      "step": 369320
    },
    {
      "epoch": 151.18,
      "grad_norm": 1.9349169731140137,
      "learning_rate": 7.230632254939337e-07,
      "loss": 0.4673,
      "step": 369330
    },
    {
      "epoch": 151.18,
      "grad_norm": 2.2344491481781006,
      "learning_rate": 7.229486378946025e-07,
      "loss": 0.4354,
      "step": 369340
    },
    {
      "epoch": 151.19,
      "grad_norm": 1.9982333183288574,
      "learning_rate": 7.22834057719911e-07,
      "loss": 0.4528,
      "step": 369350
    },
    {
      "epoch": 151.19,
      "grad_norm": 1.780041217803955,
      "learning_rate": 7.227194849703844e-07,
      "loss": 0.4413,
      "step": 369360
    },
    {
      "epoch": 151.2,
      "grad_norm": 2.3745977878570557,
      "learning_rate": 7.226049196465476e-07,
      "loss": 0.4539,
      "step": 369370
    },
    {
      "epoch": 151.2,
      "grad_norm": 2.3615260124206543,
      "learning_rate": 7.224903617489254e-07,
      "loss": 0.4351,
      "step": 369380
    },
    {
      "epoch": 151.2,
      "grad_norm": 2.0045013427734375,
      "learning_rate": 7.223758112780424e-07,
      "loss": 0.4236,
      "step": 369390
    },
    {
      "epoch": 151.21,
      "grad_norm": 1.678086280822754,
      "learning_rate": 7.222612682344236e-07,
      "loss": 0.4386,
      "step": 369400
    },
    {
      "epoch": 151.21,
      "grad_norm": 2.074937105178833,
      "learning_rate": 7.221467326185941e-07,
      "loss": 0.4506,
      "step": 369410
    },
    {
      "epoch": 151.22,
      "grad_norm": 1.9704493284225464,
      "learning_rate": 7.220322044310775e-07,
      "loss": 0.445,
      "step": 369420
    },
    {
      "epoch": 151.22,
      "grad_norm": 1.910884141921997,
      "learning_rate": 7.21917683672399e-07,
      "loss": 0.4396,
      "step": 369430
    },
    {
      "epoch": 151.22,
      "grad_norm": 1.7793852090835571,
      "learning_rate": 7.21803170343083e-07,
      "loss": 0.4437,
      "step": 369440
    },
    {
      "epoch": 151.23,
      "grad_norm": 1.9196377992630005,
      "learning_rate": 7.216886644436544e-07,
      "loss": 0.4337,
      "step": 369450
    },
    {
      "epoch": 151.23,
      "grad_norm": 2.840571165084839,
      "learning_rate": 7.215741659746374e-07,
      "loss": 0.4484,
      "step": 369460
    },
    {
      "epoch": 151.24,
      "grad_norm": 2.2442212104797363,
      "learning_rate": 7.214596749365565e-07,
      "loss": 0.441,
      "step": 369470
    },
    {
      "epoch": 151.24,
      "grad_norm": 2.053114175796509,
      "learning_rate": 7.213451913299369e-07,
      "loss": 0.4464,
      "step": 369480
    },
    {
      "epoch": 151.24,
      "grad_norm": 1.9709510803222656,
      "learning_rate": 7.212307151553019e-07,
      "loss": 0.4458,
      "step": 369490
    },
    {
      "epoch": 151.25,
      "grad_norm": 2.8092763423919678,
      "learning_rate": 7.211162464131759e-07,
      "loss": 0.4733,
      "step": 369500
    },
    {
      "epoch": 151.25,
      "grad_norm": 1.8232227563858032,
      "learning_rate": 7.210017851040847e-07,
      "loss": 0.4401,
      "step": 369510
    },
    {
      "epoch": 151.26,
      "grad_norm": 1.757698655128479,
      "learning_rate": 7.208873312285518e-07,
      "loss": 0.4299,
      "step": 369520
    },
    {
      "epoch": 151.26,
      "grad_norm": 1.5475391149520874,
      "learning_rate": 7.207728847871013e-07,
      "loss": 0.4402,
      "step": 369530
    },
    {
      "epoch": 151.26,
      "grad_norm": 1.8944385051727295,
      "learning_rate": 7.206584457802577e-07,
      "loss": 0.4712,
      "step": 369540
    },
    {
      "epoch": 151.27,
      "grad_norm": 2.1316981315612793,
      "learning_rate": 7.205440142085457e-07,
      "loss": 0.4496,
      "step": 369550
    },
    {
      "epoch": 151.27,
      "grad_norm": 2.0098531246185303,
      "learning_rate": 7.204295900724883e-07,
      "loss": 0.4432,
      "step": 369560
    },
    {
      "epoch": 151.28,
      "grad_norm": 2.0311243534088135,
      "learning_rate": 7.203151733726105e-07,
      "loss": 0.4643,
      "step": 369570
    },
    {
      "epoch": 151.28,
      "grad_norm": 1.8921337127685547,
      "learning_rate": 7.20200764109436e-07,
      "loss": 0.4398,
      "step": 369580
    },
    {
      "epoch": 151.29,
      "grad_norm": 2.044491767883301,
      "learning_rate": 7.200863622834892e-07,
      "loss": 0.4551,
      "step": 369590
    },
    {
      "epoch": 151.29,
      "grad_norm": 1.997753620147705,
      "learning_rate": 7.199719678952941e-07,
      "loss": 0.4535,
      "step": 369600
    },
    {
      "epoch": 151.29,
      "grad_norm": 2.04309344291687,
      "learning_rate": 7.198575809453745e-07,
      "loss": 0.4513,
      "step": 369610
    },
    {
      "epoch": 151.3,
      "grad_norm": 1.5928512811660767,
      "learning_rate": 7.197432014342547e-07,
      "loss": 0.4333,
      "step": 369620
    },
    {
      "epoch": 151.3,
      "grad_norm": 2.010477066040039,
      "learning_rate": 7.196288293624586e-07,
      "loss": 0.4379,
      "step": 369630
    },
    {
      "epoch": 151.31,
      "grad_norm": 2.0102412700653076,
      "learning_rate": 7.195144647305101e-07,
      "loss": 0.461,
      "step": 369640
    },
    {
      "epoch": 151.31,
      "grad_norm": 2.312225103378296,
      "learning_rate": 7.194001075389329e-07,
      "loss": 0.4309,
      "step": 369650
    },
    {
      "epoch": 151.31,
      "grad_norm": 1.919005274772644,
      "learning_rate": 7.19285757788251e-07,
      "loss": 0.453,
      "step": 369660
    },
    {
      "epoch": 151.32,
      "grad_norm": 1.5723085403442383,
      "learning_rate": 7.191714154789881e-07,
      "loss": 0.4246,
      "step": 369670
    },
    {
      "epoch": 151.32,
      "grad_norm": 2.1668570041656494,
      "learning_rate": 7.190570806116683e-07,
      "loss": 0.4407,
      "step": 369680
    },
    {
      "epoch": 151.33,
      "grad_norm": 2.586827516555786,
      "learning_rate": 7.18942753186815e-07,
      "loss": 0.4431,
      "step": 369690
    },
    {
      "epoch": 151.33,
      "grad_norm": 1.9013086557388306,
      "learning_rate": 7.188284332049523e-07,
      "loss": 0.4513,
      "step": 369700
    },
    {
      "epoch": 151.33,
      "grad_norm": 2.4097182750701904,
      "learning_rate": 7.187141206666037e-07,
      "loss": 0.4643,
      "step": 369710
    },
    {
      "epoch": 151.34,
      "grad_norm": 2.148259162902832,
      "learning_rate": 7.185998155722925e-07,
      "loss": 0.4448,
      "step": 369720
    },
    {
      "epoch": 151.34,
      "grad_norm": 2.2236127853393555,
      "learning_rate": 7.184855179225425e-07,
      "loss": 0.4646,
      "step": 369730
    },
    {
      "epoch": 151.35,
      "grad_norm": 1.9783620834350586,
      "learning_rate": 7.183712277178777e-07,
      "loss": 0.4575,
      "step": 369740
    },
    {
      "epoch": 151.35,
      "grad_norm": 2.2033329010009766,
      "learning_rate": 7.182569449588213e-07,
      "loss": 0.4417,
      "step": 369750
    },
    {
      "epoch": 151.35,
      "grad_norm": 1.9551435708999634,
      "learning_rate": 7.181426696458967e-07,
      "loss": 0.4441,
      "step": 369760
    },
    {
      "epoch": 151.36,
      "grad_norm": 2.022484302520752,
      "learning_rate": 7.180284017796275e-07,
      "loss": 0.4422,
      "step": 369770
    },
    {
      "epoch": 151.36,
      "grad_norm": 2.1309757232666016,
      "learning_rate": 7.179141413605375e-07,
      "loss": 0.4536,
      "step": 369780
    },
    {
      "epoch": 151.37,
      "grad_norm": 1.8263732194900513,
      "learning_rate": 7.177998883891495e-07,
      "loss": 0.4476,
      "step": 369790
    },
    {
      "epoch": 151.37,
      "grad_norm": 2.1948423385620117,
      "learning_rate": 7.176856428659873e-07,
      "loss": 0.4495,
      "step": 369800
    },
    {
      "epoch": 151.38,
      "grad_norm": 2.4568493366241455,
      "learning_rate": 7.175714047915738e-07,
      "loss": 0.4447,
      "step": 369810
    },
    {
      "epoch": 151.38,
      "grad_norm": 2.0158233642578125,
      "learning_rate": 7.174571741664329e-07,
      "loss": 0.4407,
      "step": 369820
    },
    {
      "epoch": 151.38,
      "grad_norm": 1.8900749683380127,
      "learning_rate": 7.173429509910874e-07,
      "loss": 0.4686,
      "step": 369830
    },
    {
      "epoch": 151.39,
      "grad_norm": 1.9529615640640259,
      "learning_rate": 7.172287352660606e-07,
      "loss": 0.4395,
      "step": 369840
    },
    {
      "epoch": 151.39,
      "grad_norm": 1.9683984518051147,
      "learning_rate": 7.171145269918762e-07,
      "loss": 0.4548,
      "step": 369850
    },
    {
      "epoch": 151.4,
      "grad_norm": 1.8205124139785767,
      "learning_rate": 7.170003261690575e-07,
      "loss": 0.4459,
      "step": 369860
    },
    {
      "epoch": 151.4,
      "grad_norm": 2.043795347213745,
      "learning_rate": 7.168861327981256e-07,
      "loss": 0.4359,
      "step": 369870
    },
    {
      "epoch": 151.4,
      "grad_norm": 2.2223737239837646,
      "learning_rate": 7.167719468796059e-07,
      "loss": 0.4566,
      "step": 369880
    },
    {
      "epoch": 151.41,
      "grad_norm": 1.8565975427627563,
      "learning_rate": 7.166577684140208e-07,
      "loss": 0.4434,
      "step": 369890
    },
    {
      "epoch": 151.41,
      "grad_norm": 2.1845593452453613,
      "learning_rate": 7.16543597401893e-07,
      "loss": 0.444,
      "step": 369900
    },
    {
      "epoch": 151.42,
      "grad_norm": 2.258835554122925,
      "learning_rate": 7.16429433843746e-07,
      "loss": 0.4449,
      "step": 369910
    },
    {
      "epoch": 151.42,
      "grad_norm": 1.7906852960586548,
      "learning_rate": 7.163152777401023e-07,
      "loss": 0.449,
      "step": 369920
    },
    {
      "epoch": 151.42,
      "grad_norm": 2.5046730041503906,
      "learning_rate": 7.16201129091485e-07,
      "loss": 0.4679,
      "step": 369930
    },
    {
      "epoch": 151.43,
      "grad_norm": 1.979251742362976,
      "learning_rate": 7.160869878984171e-07,
      "loss": 0.4538,
      "step": 369940
    },
    {
      "epoch": 151.43,
      "grad_norm": 2.320502996444702,
      "learning_rate": 7.159728541614213e-07,
      "loss": 0.4529,
      "step": 369950
    },
    {
      "epoch": 151.44,
      "grad_norm": 1.777407169342041,
      "learning_rate": 7.158587278810207e-07,
      "loss": 0.4471,
      "step": 369960
    },
    {
      "epoch": 151.44,
      "grad_norm": 1.7714102268218994,
      "learning_rate": 7.157446090577379e-07,
      "loss": 0.45,
      "step": 369970
    },
    {
      "epoch": 151.44,
      "grad_norm": 1.9715917110443115,
      "learning_rate": 7.156304976920955e-07,
      "loss": 0.4767,
      "step": 369980
    },
    {
      "epoch": 151.45,
      "grad_norm": 1.809482216835022,
      "learning_rate": 7.155163937846172e-07,
      "loss": 0.4451,
      "step": 369990
    },
    {
      "epoch": 151.45,
      "grad_norm": 5.481706142425537,
      "learning_rate": 7.154022973358241e-07,
      "loss": 0.4739,
      "step": 370000
    },
    {
      "epoch": 151.46,
      "grad_norm": 2.3123929500579834,
      "learning_rate": 7.152882083462396e-07,
      "loss": 0.4571,
      "step": 370010
    },
    {
      "epoch": 151.46,
      "grad_norm": 1.905117154121399,
      "learning_rate": 7.151741268163866e-07,
      "loss": 0.4621,
      "step": 370020
    },
    {
      "epoch": 151.47,
      "grad_norm": 2.196096181869507,
      "learning_rate": 7.150600527467873e-07,
      "loss": 0.4472,
      "step": 370030
    },
    {
      "epoch": 151.47,
      "grad_norm": 1.9757670164108276,
      "learning_rate": 7.149459861379644e-07,
      "loss": 0.4434,
      "step": 370040
    },
    {
      "epoch": 151.47,
      "grad_norm": 2.243074655532837,
      "learning_rate": 7.148319269904404e-07,
      "loss": 0.465,
      "step": 370050
    },
    {
      "epoch": 151.48,
      "grad_norm": 2.033477544784546,
      "learning_rate": 7.147178753047372e-07,
      "loss": 0.4484,
      "step": 370060
    },
    {
      "epoch": 151.48,
      "grad_norm": 2.038128614425659,
      "learning_rate": 7.146038310813785e-07,
      "loss": 0.4489,
      "step": 370070
    },
    {
      "epoch": 151.49,
      "grad_norm": 2.478694200515747,
      "learning_rate": 7.144897943208861e-07,
      "loss": 0.4376,
      "step": 370080
    },
    {
      "epoch": 151.49,
      "grad_norm": 1.9701733589172363,
      "learning_rate": 7.143757650237824e-07,
      "loss": 0.4547,
      "step": 370090
    },
    {
      "epoch": 151.49,
      "grad_norm": 2.132093906402588,
      "learning_rate": 7.142617431905898e-07,
      "loss": 0.4374,
      "step": 370100
    },
    {
      "epoch": 151.5,
      "grad_norm": 1.7118932008743286,
      "learning_rate": 7.141477288218306e-07,
      "loss": 0.4485,
      "step": 370110
    },
    {
      "epoch": 151.5,
      "grad_norm": 2.1482951641082764,
      "learning_rate": 7.14033721918027e-07,
      "loss": 0.4523,
      "step": 370120
    },
    {
      "epoch": 151.51,
      "grad_norm": 2.0058906078338623,
      "learning_rate": 7.139197224797018e-07,
      "loss": 0.4493,
      "step": 370130
    },
    {
      "epoch": 151.51,
      "grad_norm": 2.4918200969696045,
      "learning_rate": 7.138057305073763e-07,
      "loss": 0.4421,
      "step": 370140
    },
    {
      "epoch": 151.51,
      "grad_norm": 2.3853867053985596,
      "learning_rate": 7.136917460015728e-07,
      "loss": 0.4427,
      "step": 370150
    },
    {
      "epoch": 151.52,
      "grad_norm": 1.6261347532272339,
      "learning_rate": 7.135777689628138e-07,
      "loss": 0.4442,
      "step": 370160
    },
    {
      "epoch": 151.52,
      "grad_norm": 2.077335834503174,
      "learning_rate": 7.134637993916211e-07,
      "loss": 0.4252,
      "step": 370170
    },
    {
      "epoch": 151.53,
      "grad_norm": 2.1409904956817627,
      "learning_rate": 7.133498372885174e-07,
      "loss": 0.465,
      "step": 370180
    },
    {
      "epoch": 151.53,
      "grad_norm": 2.1400575637817383,
      "learning_rate": 7.132358826540241e-07,
      "loss": 0.4358,
      "step": 370190
    },
    {
      "epoch": 151.53,
      "grad_norm": 1.913163661956787,
      "learning_rate": 7.131219354886636e-07,
      "loss": 0.4463,
      "step": 370200
    },
    {
      "epoch": 151.54,
      "grad_norm": 1.9109209775924683,
      "learning_rate": 7.130079957929576e-07,
      "loss": 0.4552,
      "step": 370210
    },
    {
      "epoch": 151.54,
      "grad_norm": 1.8706554174423218,
      "learning_rate": 7.128940635674282e-07,
      "loss": 0.4386,
      "step": 370220
    },
    {
      "epoch": 151.55,
      "grad_norm": 1.8325374126434326,
      "learning_rate": 7.127801388125973e-07,
      "loss": 0.451,
      "step": 370230
    },
    {
      "epoch": 151.55,
      "grad_norm": 1.747881531715393,
      "learning_rate": 7.126662215289869e-07,
      "loss": 0.4359,
      "step": 370240
    },
    {
      "epoch": 151.56,
      "grad_norm": 2.24819016456604,
      "learning_rate": 7.125523117171178e-07,
      "loss": 0.4393,
      "step": 370250
    },
    {
      "epoch": 151.56,
      "grad_norm": 1.8646255731582642,
      "learning_rate": 7.124384093775133e-07,
      "loss": 0.4434,
      "step": 370260
    },
    {
      "epoch": 151.56,
      "grad_norm": 2.1035103797912598,
      "learning_rate": 7.123245145106956e-07,
      "loss": 0.4501,
      "step": 370270
    },
    {
      "epoch": 151.57,
      "grad_norm": 1.7102093696594238,
      "learning_rate": 7.122106271171844e-07,
      "loss": 0.4605,
      "step": 370280
    },
    {
      "epoch": 151.57,
      "grad_norm": 1.8768095970153809,
      "learning_rate": 7.120967471975024e-07,
      "loss": 0.4541,
      "step": 370290
    },
    {
      "epoch": 151.58,
      "grad_norm": 2.2148237228393555,
      "learning_rate": 7.119828747521712e-07,
      "loss": 0.4541,
      "step": 370300
    },
    {
      "epoch": 151.58,
      "grad_norm": 1.8816756010055542,
      "learning_rate": 7.118690097817126e-07,
      "loss": 0.4439,
      "step": 370310
    },
    {
      "epoch": 151.58,
      "grad_norm": 2.1344947814941406,
      "learning_rate": 7.117551522866482e-07,
      "loss": 0.4413,
      "step": 370320
    },
    {
      "epoch": 151.59,
      "grad_norm": 2.2380614280700684,
      "learning_rate": 7.116413022674992e-07,
      "loss": 0.4465,
      "step": 370330
    },
    {
      "epoch": 151.59,
      "grad_norm": 2.0350968837738037,
      "learning_rate": 7.115274597247875e-07,
      "loss": 0.4531,
      "step": 370340
    },
    {
      "epoch": 151.6,
      "grad_norm": 2.0499370098114014,
      "learning_rate": 7.114136246590347e-07,
      "loss": 0.455,
      "step": 370350
    },
    {
      "epoch": 151.6,
      "grad_norm": 2.3589248657226562,
      "learning_rate": 7.112997970707616e-07,
      "loss": 0.4357,
      "step": 370360
    },
    {
      "epoch": 151.6,
      "grad_norm": 1.8913649320602417,
      "learning_rate": 7.111859769604904e-07,
      "loss": 0.4625,
      "step": 370370
    },
    {
      "epoch": 151.61,
      "grad_norm": 2.1525990962982178,
      "learning_rate": 7.110721643287423e-07,
      "loss": 0.4223,
      "step": 370380
    },
    {
      "epoch": 151.61,
      "grad_norm": 1.8016395568847656,
      "learning_rate": 7.10958359176038e-07,
      "loss": 0.463,
      "step": 370390
    },
    {
      "epoch": 151.62,
      "grad_norm": 1.9093663692474365,
      "learning_rate": 7.108445615028997e-07,
      "loss": 0.4396,
      "step": 370400
    },
    {
      "epoch": 151.62,
      "grad_norm": 1.916041374206543,
      "learning_rate": 7.107307713098482e-07,
      "loss": 0.4589,
      "step": 370410
    },
    {
      "epoch": 151.63,
      "grad_norm": 2.2203760147094727,
      "learning_rate": 7.106169885974047e-07,
      "loss": 0.4482,
      "step": 370420
    },
    {
      "epoch": 151.63,
      "grad_norm": 2.1669728755950928,
      "learning_rate": 7.105032133660908e-07,
      "loss": 0.4385,
      "step": 370430
    },
    {
      "epoch": 151.63,
      "grad_norm": 2.2597782611846924,
      "learning_rate": 7.103894456164277e-07,
      "loss": 0.4748,
      "step": 370440
    },
    {
      "epoch": 151.64,
      "grad_norm": 1.6999186277389526,
      "learning_rate": 7.102756853489362e-07,
      "loss": 0.4371,
      "step": 370450
    },
    {
      "epoch": 151.64,
      "grad_norm": 1.9646809101104736,
      "learning_rate": 7.101619325641376e-07,
      "loss": 0.448,
      "step": 370460
    },
    {
      "epoch": 151.65,
      "grad_norm": 2.044987678527832,
      "learning_rate": 7.100481872625527e-07,
      "loss": 0.4522,
      "step": 370470
    },
    {
      "epoch": 151.65,
      "grad_norm": 2.054457426071167,
      "learning_rate": 7.09934449444703e-07,
      "loss": 0.457,
      "step": 370480
    },
    {
      "epoch": 151.65,
      "grad_norm": 1.9817579984664917,
      "learning_rate": 7.098207191111093e-07,
      "loss": 0.4579,
      "step": 370490
    },
    {
      "epoch": 151.66,
      "grad_norm": 2.3105642795562744,
      "learning_rate": 7.097069962622925e-07,
      "loss": 0.4682,
      "step": 370500
    },
    {
      "epoch": 151.66,
      "grad_norm": 1.8429603576660156,
      "learning_rate": 7.095932808987737e-07,
      "loss": 0.4468,
      "step": 370510
    },
    {
      "epoch": 151.67,
      "grad_norm": 1.673962950706482,
      "learning_rate": 7.094795730210736e-07,
      "loss": 0.4653,
      "step": 370520
    },
    {
      "epoch": 151.67,
      "grad_norm": 2.110384464263916,
      "learning_rate": 7.093658726297134e-07,
      "loss": 0.4579,
      "step": 370530
    },
    {
      "epoch": 151.67,
      "grad_norm": 1.7058875560760498,
      "learning_rate": 7.092521797252137e-07,
      "loss": 0.4602,
      "step": 370540
    },
    {
      "epoch": 151.68,
      "grad_norm": 2.5085482597351074,
      "learning_rate": 7.091384943080955e-07,
      "loss": 0.4359,
      "step": 370550
    },
    {
      "epoch": 151.68,
      "grad_norm": 2.164266347885132,
      "learning_rate": 7.090248163788792e-07,
      "loss": 0.4309,
      "step": 370560
    },
    {
      "epoch": 151.69,
      "grad_norm": 1.7648892402648926,
      "learning_rate": 7.089111459380861e-07,
      "loss": 0.4521,
      "step": 370570
    },
    {
      "epoch": 151.69,
      "grad_norm": 1.619242787361145,
      "learning_rate": 7.087974829862372e-07,
      "loss": 0.4497,
      "step": 370580
    },
    {
      "epoch": 151.69,
      "grad_norm": 1.8455333709716797,
      "learning_rate": 7.086838275238518e-07,
      "loss": 0.4581,
      "step": 370590
    },
    {
      "epoch": 151.7,
      "grad_norm": 2.1844394207000732,
      "learning_rate": 7.085701795514513e-07,
      "loss": 0.4698,
      "step": 370600
    },
    {
      "epoch": 151.7,
      "grad_norm": 2.1274006366729736,
      "learning_rate": 7.084565390695563e-07,
      "loss": 0.4253,
      "step": 370610
    },
    {
      "epoch": 151.71,
      "grad_norm": 1.95566987991333,
      "learning_rate": 7.083429060786876e-07,
      "loss": 0.4512,
      "step": 370620
    },
    {
      "epoch": 151.71,
      "grad_norm": 1.7987340688705444,
      "learning_rate": 7.082292805793646e-07,
      "loss": 0.4518,
      "step": 370630
    },
    {
      "epoch": 151.72,
      "grad_norm": 1.9475553035736084,
      "learning_rate": 7.081156625721095e-07,
      "loss": 0.4555,
      "step": 370640
    },
    {
      "epoch": 151.72,
      "grad_norm": 2.4011569023132324,
      "learning_rate": 7.080020520574419e-07,
      "loss": 0.4327,
      "step": 370650
    },
    {
      "epoch": 151.72,
      "grad_norm": 3.0104830265045166,
      "learning_rate": 7.078884490358826e-07,
      "loss": 0.4725,
      "step": 370660
    },
    {
      "epoch": 151.73,
      "grad_norm": 2.1365246772766113,
      "learning_rate": 7.077748535079515e-07,
      "loss": 0.4581,
      "step": 370670
    },
    {
      "epoch": 151.73,
      "grad_norm": 2.094517707824707,
      "learning_rate": 7.076612654741692e-07,
      "loss": 0.4524,
      "step": 370680
    },
    {
      "epoch": 151.74,
      "grad_norm": 1.8945714235305786,
      "learning_rate": 7.075476849350562e-07,
      "loss": 0.4729,
      "step": 370690
    },
    {
      "epoch": 151.74,
      "grad_norm": 1.9555602073669434,
      "learning_rate": 7.074341118911323e-07,
      "loss": 0.4373,
      "step": 370700
    },
    {
      "epoch": 151.74,
      "grad_norm": 2.2211875915527344,
      "learning_rate": 7.073205463429189e-07,
      "loss": 0.4456,
      "step": 370710
    },
    {
      "epoch": 151.75,
      "grad_norm": 2.0383734703063965,
      "learning_rate": 7.072069882909348e-07,
      "loss": 0.4499,
      "step": 370720
    },
    {
      "epoch": 151.75,
      "grad_norm": 1.995016098022461,
      "learning_rate": 7.070934377357007e-07,
      "loss": 0.4514,
      "step": 370730
    },
    {
      "epoch": 151.76,
      "grad_norm": 2.0412752628326416,
      "learning_rate": 7.069798946777367e-07,
      "loss": 0.4643,
      "step": 370740
    },
    {
      "epoch": 151.76,
      "grad_norm": 1.8987646102905273,
      "learning_rate": 7.068663591175632e-07,
      "loss": 0.4496,
      "step": 370750
    },
    {
      "epoch": 151.76,
      "grad_norm": 1.9602500200271606,
      "learning_rate": 7.067528310557001e-07,
      "loss": 0.4614,
      "step": 370760
    },
    {
      "epoch": 151.77,
      "grad_norm": 2.1368727684020996,
      "learning_rate": 7.066393104926675e-07,
      "loss": 0.4328,
      "step": 370770
    },
    {
      "epoch": 151.77,
      "grad_norm": 2.3856024742126465,
      "learning_rate": 7.065257974289858e-07,
      "loss": 0.4539,
      "step": 370780
    },
    {
      "epoch": 151.78,
      "grad_norm": 2.170510768890381,
      "learning_rate": 7.064122918651742e-07,
      "loss": 0.4565,
      "step": 370790
    },
    {
      "epoch": 151.78,
      "grad_norm": 2.1030595302581787,
      "learning_rate": 7.062987938017532e-07,
      "loss": 0.4453,
      "step": 370800
    },
    {
      "epoch": 151.78,
      "grad_norm": 2.0775699615478516,
      "learning_rate": 7.061853032392424e-07,
      "loss": 0.4461,
      "step": 370810
    },
    {
      "epoch": 151.79,
      "grad_norm": 1.9683810472488403,
      "learning_rate": 7.060718201781613e-07,
      "loss": 0.4513,
      "step": 370820
    },
    {
      "epoch": 151.79,
      "grad_norm": 2.2250616550445557,
      "learning_rate": 7.059583446190311e-07,
      "loss": 0.4353,
      "step": 370830
    },
    {
      "epoch": 151.8,
      "grad_norm": 2.9536969661712646,
      "learning_rate": 7.058448765623709e-07,
      "loss": 0.4644,
      "step": 370840
    },
    {
      "epoch": 151.8,
      "grad_norm": 2.173513650894165,
      "learning_rate": 7.057314160087008e-07,
      "loss": 0.4539,
      "step": 370850
    },
    {
      "epoch": 151.81,
      "grad_norm": 2.1385438442230225,
      "learning_rate": 7.056179629585398e-07,
      "loss": 0.4551,
      "step": 370860
    },
    {
      "epoch": 151.81,
      "grad_norm": 3.3266005516052246,
      "learning_rate": 7.055045174124077e-07,
      "loss": 0.4523,
      "step": 370870
    },
    {
      "epoch": 151.81,
      "grad_norm": 1.8930522203445435,
      "learning_rate": 7.053910793708247e-07,
      "loss": 0.452,
      "step": 370880
    },
    {
      "epoch": 151.82,
      "grad_norm": 2.124279499053955,
      "learning_rate": 7.0527764883431e-07,
      "loss": 0.4648,
      "step": 370890
    },
    {
      "epoch": 151.82,
      "grad_norm": 2.480114221572876,
      "learning_rate": 7.051642258033834e-07,
      "loss": 0.4336,
      "step": 370900
    },
    {
      "epoch": 151.83,
      "grad_norm": 1.8162645101547241,
      "learning_rate": 7.050508102785644e-07,
      "loss": 0.4459,
      "step": 370910
    },
    {
      "epoch": 151.83,
      "grad_norm": 1.702362060546875,
      "learning_rate": 7.049374022603726e-07,
      "loss": 0.4476,
      "step": 370920
    },
    {
      "epoch": 151.83,
      "grad_norm": 2.288306951522827,
      "learning_rate": 7.048240017493276e-07,
      "loss": 0.442,
      "step": 370930
    },
    {
      "epoch": 151.84,
      "grad_norm": 1.7752248048782349,
      "learning_rate": 7.047106087459489e-07,
      "loss": 0.4438,
      "step": 370940
    },
    {
      "epoch": 151.84,
      "grad_norm": 1.9305689334869385,
      "learning_rate": 7.045972232507555e-07,
      "loss": 0.459,
      "step": 370950
    },
    {
      "epoch": 151.85,
      "grad_norm": 1.764163613319397,
      "learning_rate": 7.044838452642674e-07,
      "loss": 0.4573,
      "step": 370960
    },
    {
      "epoch": 151.85,
      "grad_norm": 2.196645736694336,
      "learning_rate": 7.043704747870036e-07,
      "loss": 0.4507,
      "step": 370970
    },
    {
      "epoch": 151.85,
      "grad_norm": 1.9195281267166138,
      "learning_rate": 7.042571118194834e-07,
      "loss": 0.4334,
      "step": 370980
    },
    {
      "epoch": 151.86,
      "grad_norm": 2.027416944503784,
      "learning_rate": 7.041437563622265e-07,
      "loss": 0.4458,
      "step": 370990
    },
    {
      "epoch": 151.86,
      "grad_norm": 1.9263259172439575,
      "learning_rate": 7.040304084157517e-07,
      "loss": 0.4422,
      "step": 371000
    },
    {
      "epoch": 151.87,
      "grad_norm": 2.1486897468566895,
      "learning_rate": 7.039170679805782e-07,
      "loss": 0.4527,
      "step": 371010
    },
    {
      "epoch": 151.87,
      "grad_norm": 2.5404069423675537,
      "learning_rate": 7.038037350572257e-07,
      "loss": 0.452,
      "step": 371020
    },
    {
      "epoch": 151.87,
      "grad_norm": 1.942455530166626,
      "learning_rate": 7.036904096462132e-07,
      "loss": 0.4564,
      "step": 371030
    },
    {
      "epoch": 151.88,
      "grad_norm": 1.8858247995376587,
      "learning_rate": 7.035770917480591e-07,
      "loss": 0.4328,
      "step": 371040
    },
    {
      "epoch": 151.88,
      "grad_norm": 2.190751075744629,
      "learning_rate": 7.034637813632836e-07,
      "loss": 0.4433,
      "step": 371050
    },
    {
      "epoch": 151.89,
      "grad_norm": 1.9118746519088745,
      "learning_rate": 7.033504784924051e-07,
      "loss": 0.456,
      "step": 371060
    },
    {
      "epoch": 151.89,
      "grad_norm": 1.8139984607696533,
      "learning_rate": 7.032371831359427e-07,
      "loss": 0.4477,
      "step": 371070
    },
    {
      "epoch": 151.9,
      "grad_norm": 2.17842960357666,
      "learning_rate": 7.031238952944155e-07,
      "loss": 0.4481,
      "step": 371080
    },
    {
      "epoch": 151.9,
      "grad_norm": 1.877610206604004,
      "learning_rate": 7.030106149683423e-07,
      "loss": 0.4576,
      "step": 371090
    },
    {
      "epoch": 151.9,
      "grad_norm": 1.98589026927948,
      "learning_rate": 7.028973421582422e-07,
      "loss": 0.4571,
      "step": 371100
    },
    {
      "epoch": 151.91,
      "grad_norm": 1.8461434841156006,
      "learning_rate": 7.027840768646339e-07,
      "loss": 0.4622,
      "step": 371110
    },
    {
      "epoch": 151.91,
      "grad_norm": 2.3104519844055176,
      "learning_rate": 7.026708190880366e-07,
      "loss": 0.4344,
      "step": 371120
    },
    {
      "epoch": 151.92,
      "grad_norm": 1.8297579288482666,
      "learning_rate": 7.025575688289686e-07,
      "loss": 0.4457,
      "step": 371130
    },
    {
      "epoch": 151.92,
      "grad_norm": 1.719476342201233,
      "learning_rate": 7.024443260879492e-07,
      "loss": 0.4333,
      "step": 371140
    },
    {
      "epoch": 151.92,
      "grad_norm": 2.476536512374878,
      "learning_rate": 7.023310908654967e-07,
      "loss": 0.4576,
      "step": 371150
    },
    {
      "epoch": 151.93,
      "grad_norm": 2.0512020587921143,
      "learning_rate": 7.02217863162131e-07,
      "loss": 0.4579,
      "step": 371160
    },
    {
      "epoch": 151.93,
      "grad_norm": 1.916243076324463,
      "learning_rate": 7.021046429783688e-07,
      "loss": 0.4521,
      "step": 371170
    },
    {
      "epoch": 151.94,
      "grad_norm": 2.0779612064361572,
      "learning_rate": 7.0199143031473e-07,
      "loss": 0.4699,
      "step": 371180
    },
    {
      "epoch": 151.94,
      "grad_norm": 2.2853448390960693,
      "learning_rate": 7.018782251717328e-07,
      "loss": 0.4408,
      "step": 371190
    },
    {
      "epoch": 151.94,
      "grad_norm": 2.359288215637207,
      "learning_rate": 7.017650275498952e-07,
      "loss": 0.4528,
      "step": 371200
    },
    {
      "epoch": 151.95,
      "grad_norm": 1.9671250581741333,
      "learning_rate": 7.016518374497371e-07,
      "loss": 0.4617,
      "step": 371210
    },
    {
      "epoch": 151.95,
      "grad_norm": 2.0432591438293457,
      "learning_rate": 7.015386548717765e-07,
      "loss": 0.4549,
      "step": 371220
    },
    {
      "epoch": 151.96,
      "grad_norm": 2.1610143184661865,
      "learning_rate": 7.014254798165318e-07,
      "loss": 0.4569,
      "step": 371230
    },
    {
      "epoch": 151.96,
      "grad_norm": 1.7567938566207886,
      "learning_rate": 7.013123122845213e-07,
      "loss": 0.4483,
      "step": 371240
    },
    {
      "epoch": 151.96,
      "grad_norm": 1.798613429069519,
      "learning_rate": 7.011991522762632e-07,
      "loss": 0.4715,
      "step": 371250
    },
    {
      "epoch": 151.97,
      "grad_norm": 2.2340402603149414,
      "learning_rate": 7.010859997922765e-07,
      "loss": 0.4539,
      "step": 371260
    },
    {
      "epoch": 151.97,
      "grad_norm": 1.9486515522003174,
      "learning_rate": 7.009728548330792e-07,
      "loss": 0.4274,
      "step": 371270
    },
    {
      "epoch": 151.98,
      "grad_norm": 1.7734622955322266,
      "learning_rate": 7.008597173991895e-07,
      "loss": 0.43,
      "step": 371280
    },
    {
      "epoch": 151.98,
      "grad_norm": 2.514761209487915,
      "learning_rate": 7.007465874911256e-07,
      "loss": 0.4372,
      "step": 371290
    },
    {
      "epoch": 151.99,
      "grad_norm": 2.7887423038482666,
      "learning_rate": 7.00633465109407e-07,
      "loss": 0.4551,
      "step": 371300
    },
    {
      "epoch": 151.99,
      "grad_norm": 2.4487991333007812,
      "learning_rate": 7.005203502545497e-07,
      "loss": 0.4554,
      "step": 371310
    },
    {
      "epoch": 151.99,
      "grad_norm": 2.3809502124786377,
      "learning_rate": 7.00407242927073e-07,
      "loss": 0.4646,
      "step": 371320
    },
    {
      "epoch": 152.0,
      "grad_norm": 2.0611038208007812,
      "learning_rate": 7.002941431274952e-07,
      "loss": 0.4586,
      "step": 371330
    },
    {
      "epoch": 152.0,
      "eval_loss": 0.4503302574157715,
      "eval_runtime": 52.5236,
      "eval_samples_per_second": 65.666,
      "eval_steps_per_second": 8.225,
      "step": 371336
    },
    {
      "epoch": 152.0,
      "grad_norm": 2.4804956912994385,
      "learning_rate": 7.001810508563339e-07,
      "loss": 0.4326,
      "step": 371340
    },
    {
      "epoch": 152.01,
      "grad_norm": 1.943852424621582,
      "learning_rate": 7.000679661141074e-07,
      "loss": 0.4504,
      "step": 371350
    },
    {
      "epoch": 152.01,
      "grad_norm": 2.1888489723205566,
      "learning_rate": 6.999548889013339e-07,
      "loss": 0.4601,
      "step": 371360
    },
    {
      "epoch": 152.01,
      "grad_norm": 2.6006436347961426,
      "learning_rate": 6.99841819218531e-07,
      "loss": 0.4416,
      "step": 371370
    },
    {
      "epoch": 152.02,
      "grad_norm": 2.2344138622283936,
      "learning_rate": 6.997287570662165e-07,
      "loss": 0.4348,
      "step": 371380
    },
    {
      "epoch": 152.02,
      "grad_norm": 2.249948024749756,
      "learning_rate": 6.996157024449091e-07,
      "loss": 0.4355,
      "step": 371390
    },
    {
      "epoch": 152.03,
      "grad_norm": 2.508369207382202,
      "learning_rate": 6.995026553551266e-07,
      "loss": 0.4583,
      "step": 371400
    },
    {
      "epoch": 152.03,
      "grad_norm": 2.237433910369873,
      "learning_rate": 6.993896157973864e-07,
      "loss": 0.4602,
      "step": 371410
    },
    {
      "epoch": 152.03,
      "grad_norm": 1.7252528667449951,
      "learning_rate": 6.992765837722064e-07,
      "loss": 0.4508,
      "step": 371420
    },
    {
      "epoch": 152.04,
      "grad_norm": 2.5188283920288086,
      "learning_rate": 6.991635592801052e-07,
      "loss": 0.4369,
      "step": 371430
    },
    {
      "epoch": 152.04,
      "grad_norm": 2.0928871631622314,
      "learning_rate": 6.990505423215991e-07,
      "loss": 0.4554,
      "step": 371440
    },
    {
      "epoch": 152.05,
      "grad_norm": 2.2887284755706787,
      "learning_rate": 6.989375328972065e-07,
      "loss": 0.459,
      "step": 371450
    },
    {
      "epoch": 152.05,
      "grad_norm": 2.1384284496307373,
      "learning_rate": 6.988245310074449e-07,
      "loss": 0.4649,
      "step": 371460
    },
    {
      "epoch": 152.05,
      "grad_norm": 1.8894734382629395,
      "learning_rate": 6.987115366528323e-07,
      "loss": 0.4443,
      "step": 371470
    },
    {
      "epoch": 152.06,
      "grad_norm": 2.922373056411743,
      "learning_rate": 6.985985498338861e-07,
      "loss": 0.439,
      "step": 371480
    },
    {
      "epoch": 152.06,
      "grad_norm": 1.8750388622283936,
      "learning_rate": 6.98485570551124e-07,
      "loss": 0.429,
      "step": 371490
    },
    {
      "epoch": 152.07,
      "grad_norm": 1.9875785112380981,
      "learning_rate": 6.983725988050635e-07,
      "loss": 0.4553,
      "step": 371500
    },
    {
      "epoch": 152.07,
      "grad_norm": 1.9751899242401123,
      "learning_rate": 6.982596345962217e-07,
      "loss": 0.4416,
      "step": 371510
    },
    {
      "epoch": 152.08,
      "grad_norm": 2.231855869293213,
      "learning_rate": 6.981466779251167e-07,
      "loss": 0.4354,
      "step": 371520
    },
    {
      "epoch": 152.08,
      "grad_norm": 1.8972288370132446,
      "learning_rate": 6.980337287922657e-07,
      "loss": 0.457,
      "step": 371530
    },
    {
      "epoch": 152.08,
      "grad_norm": 2.098889112472534,
      "learning_rate": 6.979207871981861e-07,
      "loss": 0.4302,
      "step": 371540
    },
    {
      "epoch": 152.09,
      "grad_norm": 1.9282031059265137,
      "learning_rate": 6.978078531433952e-07,
      "loss": 0.4377,
      "step": 371550
    },
    {
      "epoch": 152.09,
      "grad_norm": 2.2805557250976562,
      "learning_rate": 6.976949266284103e-07,
      "loss": 0.443,
      "step": 371560
    },
    {
      "epoch": 152.1,
      "grad_norm": 2.2685306072235107,
      "learning_rate": 6.975820076537487e-07,
      "loss": 0.4601,
      "step": 371570
    },
    {
      "epoch": 152.1,
      "grad_norm": 1.6814239025115967,
      "learning_rate": 6.974690962199277e-07,
      "loss": 0.4402,
      "step": 371580
    },
    {
      "epoch": 152.1,
      "grad_norm": 1.7349799871444702,
      "learning_rate": 6.973561923274646e-07,
      "loss": 0.4773,
      "step": 371590
    },
    {
      "epoch": 152.11,
      "grad_norm": 1.9324263334274292,
      "learning_rate": 6.972432959768767e-07,
      "loss": 0.443,
      "step": 371600
    },
    {
      "epoch": 152.11,
      "grad_norm": 1.9703916311264038,
      "learning_rate": 6.971304071686808e-07,
      "loss": 0.4778,
      "step": 371610
    },
    {
      "epoch": 152.12,
      "grad_norm": 2.079176664352417,
      "learning_rate": 6.970175259033944e-07,
      "loss": 0.4391,
      "step": 371620
    },
    {
      "epoch": 152.12,
      "grad_norm": 2.266411304473877,
      "learning_rate": 6.969046521815343e-07,
      "loss": 0.4411,
      "step": 371630
    },
    {
      "epoch": 152.12,
      "grad_norm": 2.340834617614746,
      "learning_rate": 6.967917860036179e-07,
      "loss": 0.4631,
      "step": 371640
    },
    {
      "epoch": 152.13,
      "grad_norm": 1.741107702255249,
      "learning_rate": 6.96678927370162e-07,
      "loss": 0.441,
      "step": 371650
    },
    {
      "epoch": 152.13,
      "grad_norm": 2.1364905834198,
      "learning_rate": 6.965660762816836e-07,
      "loss": 0.444,
      "step": 371660
    },
    {
      "epoch": 152.14,
      "grad_norm": 2.622144937515259,
      "learning_rate": 6.964532327386995e-07,
      "loss": 0.4519,
      "step": 371670
    },
    {
      "epoch": 152.14,
      "grad_norm": 2.0745112895965576,
      "learning_rate": 6.963403967417267e-07,
      "loss": 0.4348,
      "step": 371680
    },
    {
      "epoch": 152.14,
      "grad_norm": 1.5473397970199585,
      "learning_rate": 6.962275682912824e-07,
      "loss": 0.4511,
      "step": 371690
    },
    {
      "epoch": 152.15,
      "grad_norm": 2.4337973594665527,
      "learning_rate": 6.961147473878831e-07,
      "loss": 0.4382,
      "step": 371700
    },
    {
      "epoch": 152.15,
      "grad_norm": 2.1036667823791504,
      "learning_rate": 6.960019340320457e-07,
      "loss": 0.452,
      "step": 371710
    },
    {
      "epoch": 152.16,
      "grad_norm": 1.9421045780181885,
      "learning_rate": 6.95889128224287e-07,
      "loss": 0.4638,
      "step": 371720
    },
    {
      "epoch": 152.16,
      "grad_norm": 2.256728410720825,
      "learning_rate": 6.957763299651239e-07,
      "loss": 0.4394,
      "step": 371730
    },
    {
      "epoch": 152.17,
      "grad_norm": 2.573330879211426,
      "learning_rate": 6.956635392550735e-07,
      "loss": 0.4738,
      "step": 371740
    },
    {
      "epoch": 152.17,
      "grad_norm": 1.591392993927002,
      "learning_rate": 6.955507560946515e-07,
      "loss": 0.4529,
      "step": 371750
    },
    {
      "epoch": 152.17,
      "grad_norm": 2.0570478439331055,
      "learning_rate": 6.954379804843738e-07,
      "loss": 0.4446,
      "step": 371760
    },
    {
      "epoch": 152.18,
      "grad_norm": 1.9909065961837769,
      "learning_rate": 6.953252124247593e-07,
      "loss": 0.4608,
      "step": 371770
    },
    {
      "epoch": 152.18,
      "grad_norm": 1.9146729707717896,
      "learning_rate": 6.952124519163233e-07,
      "loss": 0.4584,
      "step": 371780
    },
    {
      "epoch": 152.19,
      "grad_norm": 1.704956293106079,
      "learning_rate": 6.950996989595824e-07,
      "loss": 0.4368,
      "step": 371790
    },
    {
      "epoch": 152.19,
      "grad_norm": 1.8128163814544678,
      "learning_rate": 6.949869535550534e-07,
      "loss": 0.4432,
      "step": 371800
    },
    {
      "epoch": 152.19,
      "grad_norm": 3.7727160453796387,
      "learning_rate": 6.948742157032526e-07,
      "loss": 0.4458,
      "step": 371810
    },
    {
      "epoch": 152.2,
      "grad_norm": 2.018785238265991,
      "learning_rate": 6.947614854046962e-07,
      "loss": 0.4348,
      "step": 371820
    },
    {
      "epoch": 152.2,
      "grad_norm": 2.7398927211761475,
      "learning_rate": 6.94648762659901e-07,
      "loss": 0.4554,
      "step": 371830
    },
    {
      "epoch": 152.21,
      "grad_norm": 2.644901990890503,
      "learning_rate": 6.945360474693832e-07,
      "loss": 0.4199,
      "step": 371840
    },
    {
      "epoch": 152.21,
      "grad_norm": 2.1451621055603027,
      "learning_rate": 6.944233398336589e-07,
      "loss": 0.4416,
      "step": 371850
    },
    {
      "epoch": 152.21,
      "grad_norm": 2.2127621173858643,
      "learning_rate": 6.943106397532447e-07,
      "loss": 0.4706,
      "step": 371860
    },
    {
      "epoch": 152.22,
      "grad_norm": 1.998587727546692,
      "learning_rate": 6.941979472286567e-07,
      "loss": 0.4444,
      "step": 371870
    },
    {
      "epoch": 152.22,
      "grad_norm": 2.148228168487549,
      "learning_rate": 6.94085262260412e-07,
      "loss": 0.4352,
      "step": 371880
    },
    {
      "epoch": 152.23,
      "grad_norm": 1.851068139076233,
      "learning_rate": 6.939725848490252e-07,
      "loss": 0.4591,
      "step": 371890
    },
    {
      "epoch": 152.23,
      "grad_norm": 2.184830665588379,
      "learning_rate": 6.938599149950133e-07,
      "loss": 0.4534,
      "step": 371900
    },
    {
      "epoch": 152.23,
      "grad_norm": 2.1046736240386963,
      "learning_rate": 6.937472526988923e-07,
      "loss": 0.4554,
      "step": 371910
    },
    {
      "epoch": 152.24,
      "grad_norm": 2.351081132888794,
      "learning_rate": 6.936345979611784e-07,
      "loss": 0.4436,
      "step": 371920
    },
    {
      "epoch": 152.24,
      "grad_norm": 2.769662618637085,
      "learning_rate": 6.935219507823875e-07,
      "loss": 0.4507,
      "step": 371930
    },
    {
      "epoch": 152.25,
      "grad_norm": 2.0613701343536377,
      "learning_rate": 6.934093111630356e-07,
      "loss": 0.441,
      "step": 371940
    },
    {
      "epoch": 152.25,
      "grad_norm": 2.042928457260132,
      "learning_rate": 6.932966791036385e-07,
      "loss": 0.4398,
      "step": 371950
    },
    {
      "epoch": 152.26,
      "grad_norm": 1.9613807201385498,
      "learning_rate": 6.93184054604713e-07,
      "loss": 0.4602,
      "step": 371960
    },
    {
      "epoch": 152.26,
      "grad_norm": 1.7549768686294556,
      "learning_rate": 6.930714376667743e-07,
      "loss": 0.424,
      "step": 371970
    },
    {
      "epoch": 152.26,
      "grad_norm": 2.4656968116760254,
      "learning_rate": 6.929588282903387e-07,
      "loss": 0.4467,
      "step": 371980
    },
    {
      "epoch": 152.27,
      "grad_norm": 2.514275074005127,
      "learning_rate": 6.928462264759216e-07,
      "loss": 0.4408,
      "step": 371990
    },
    {
      "epoch": 152.27,
      "grad_norm": 2.0232632160186768,
      "learning_rate": 6.927336322240391e-07,
      "loss": 0.4638,
      "step": 372000
    },
    {
      "epoch": 152.28,
      "grad_norm": 2.0729122161865234,
      "learning_rate": 6.926210455352076e-07,
      "loss": 0.4332,
      "step": 372010
    },
    {
      "epoch": 152.28,
      "grad_norm": 1.9285310506820679,
      "learning_rate": 6.925084664099415e-07,
      "loss": 0.4617,
      "step": 372020
    },
    {
      "epoch": 152.28,
      "grad_norm": 2.3423993587493896,
      "learning_rate": 6.92395894848757e-07,
      "loss": 0.433,
      "step": 372030
    },
    {
      "epoch": 152.29,
      "grad_norm": 2.161050796508789,
      "learning_rate": 6.9228333085217e-07,
      "loss": 0.4508,
      "step": 372040
    },
    {
      "epoch": 152.29,
      "grad_norm": 2.5760157108306885,
      "learning_rate": 6.921707744206962e-07,
      "loss": 0.4325,
      "step": 372050
    },
    {
      "epoch": 152.3,
      "grad_norm": 2.2336409091949463,
      "learning_rate": 6.920582255548508e-07,
      "loss": 0.4513,
      "step": 372060
    },
    {
      "epoch": 152.3,
      "grad_norm": 2.036395311355591,
      "learning_rate": 6.919456842551497e-07,
      "loss": 0.4548,
      "step": 372070
    },
    {
      "epoch": 152.3,
      "grad_norm": 1.984437108039856,
      "learning_rate": 6.918331505221083e-07,
      "loss": 0.4676,
      "step": 372080
    },
    {
      "epoch": 152.31,
      "grad_norm": 2.6594090461730957,
      "learning_rate": 6.917206243562423e-07,
      "loss": 0.4547,
      "step": 372090
    },
    {
      "epoch": 152.31,
      "grad_norm": 1.8323495388031006,
      "learning_rate": 6.916081057580669e-07,
      "loss": 0.4376,
      "step": 372100
    },
    {
      "epoch": 152.32,
      "grad_norm": 1.9404569864273071,
      "learning_rate": 6.914955947280976e-07,
      "loss": 0.4321,
      "step": 372110
    },
    {
      "epoch": 152.32,
      "grad_norm": 2.219663619995117,
      "learning_rate": 6.913830912668498e-07,
      "loss": 0.4354,
      "step": 372120
    },
    {
      "epoch": 152.33,
      "grad_norm": 2.0065412521362305,
      "learning_rate": 6.912705953748392e-07,
      "loss": 0.4643,
      "step": 372130
    },
    {
      "epoch": 152.33,
      "grad_norm": 2.035254955291748,
      "learning_rate": 6.911581070525799e-07,
      "loss": 0.469,
      "step": 372140
    },
    {
      "epoch": 152.33,
      "grad_norm": 1.7019915580749512,
      "learning_rate": 6.910456263005897e-07,
      "loss": 0.4642,
      "step": 372150
    },
    {
      "epoch": 152.34,
      "grad_norm": 2.592000722885132,
      "learning_rate": 6.909331531193814e-07,
      "loss": 0.4477,
      "step": 372160
    },
    {
      "epoch": 152.34,
      "grad_norm": 2.0906565189361572,
      "learning_rate": 6.908206875094713e-07,
      "loss": 0.4543,
      "step": 372170
    },
    {
      "epoch": 152.35,
      "grad_norm": 1.8600938320159912,
      "learning_rate": 6.907082294713744e-07,
      "loss": 0.4364,
      "step": 372180
    },
    {
      "epoch": 152.35,
      "grad_norm": 2.231099843978882,
      "learning_rate": 6.905957790056056e-07,
      "loss": 0.4654,
      "step": 372190
    },
    {
      "epoch": 152.35,
      "grad_norm": 2.2525932788848877,
      "learning_rate": 6.904833361126803e-07,
      "loss": 0.4615,
      "step": 372200
    },
    {
      "epoch": 152.36,
      "grad_norm": 1.754240870475769,
      "learning_rate": 6.903709007931136e-07,
      "loss": 0.4564,
      "step": 372210
    },
    {
      "epoch": 152.36,
      "grad_norm": 2.1145143508911133,
      "learning_rate": 6.902584730474205e-07,
      "loss": 0.4461,
      "step": 372220
    },
    {
      "epoch": 152.37,
      "grad_norm": 2.120021343231201,
      "learning_rate": 6.901460528761161e-07,
      "loss": 0.4542,
      "step": 372230
    },
    {
      "epoch": 152.37,
      "grad_norm": 2.152639627456665,
      "learning_rate": 6.900336402797152e-07,
      "loss": 0.4297,
      "step": 372240
    },
    {
      "epoch": 152.37,
      "grad_norm": 1.772063136100769,
      "learning_rate": 6.89921235258733e-07,
      "loss": 0.4279,
      "step": 372250
    },
    {
      "epoch": 152.38,
      "grad_norm": 2.03066086769104,
      "learning_rate": 6.898088378136841e-07,
      "loss": 0.4529,
      "step": 372260
    },
    {
      "epoch": 152.38,
      "grad_norm": 2.1084210872650146,
      "learning_rate": 6.896964479450835e-07,
      "loss": 0.4626,
      "step": 372270
    },
    {
      "epoch": 152.39,
      "grad_norm": 2.103346586227417,
      "learning_rate": 6.89584065653446e-07,
      "loss": 0.4467,
      "step": 372280
    },
    {
      "epoch": 152.39,
      "grad_norm": 2.207442045211792,
      "learning_rate": 6.894716909392864e-07,
      "loss": 0.4421,
      "step": 372290
    },
    {
      "epoch": 152.39,
      "grad_norm": 2.092090606689453,
      "learning_rate": 6.893593238031199e-07,
      "loss": 0.4479,
      "step": 372300
    },
    {
      "epoch": 152.4,
      "grad_norm": 1.872719645500183,
      "learning_rate": 6.892469642454605e-07,
      "loss": 0.4445,
      "step": 372310
    },
    {
      "epoch": 152.4,
      "grad_norm": 1.859693169593811,
      "learning_rate": 6.891346122668244e-07,
      "loss": 0.4452,
      "step": 372320
    },
    {
      "epoch": 152.41,
      "grad_norm": 2.5395703315734863,
      "learning_rate": 6.890222678677233e-07,
      "loss": 0.4525,
      "step": 372330
    },
    {
      "epoch": 152.41,
      "grad_norm": 1.7933528423309326,
      "learning_rate": 6.889099310486745e-07,
      "loss": 0.4387,
      "step": 372340
    },
    {
      "epoch": 152.42,
      "grad_norm": 1.902351975440979,
      "learning_rate": 6.887976018101918e-07,
      "loss": 0.4339,
      "step": 372350
    },
    {
      "epoch": 152.42,
      "grad_norm": 1.8045235872268677,
      "learning_rate": 6.8868528015279e-07,
      "loss": 0.4442,
      "step": 372360
    },
    {
      "epoch": 152.42,
      "grad_norm": 2.515446662902832,
      "learning_rate": 6.885729660769831e-07,
      "loss": 0.4578,
      "step": 372370
    },
    {
      "epoch": 152.43,
      "grad_norm": 2.1498794555664062,
      "learning_rate": 6.884606595832856e-07,
      "loss": 0.4459,
      "step": 372380
    },
    {
      "epoch": 152.43,
      "grad_norm": 2.0831198692321777,
      "learning_rate": 6.883483606722125e-07,
      "loss": 0.4406,
      "step": 372390
    },
    {
      "epoch": 152.44,
      "grad_norm": 2.129204273223877,
      "learning_rate": 6.882360693442779e-07,
      "loss": 0.4414,
      "step": 372400
    },
    {
      "epoch": 152.44,
      "grad_norm": 2.3933560848236084,
      "learning_rate": 6.881237855999964e-07,
      "loss": 0.4282,
      "step": 372410
    },
    {
      "epoch": 152.44,
      "grad_norm": 2.2316434383392334,
      "learning_rate": 6.880115094398819e-07,
      "loss": 0.4444,
      "step": 372420
    },
    {
      "epoch": 152.45,
      "grad_norm": 2.090043067932129,
      "learning_rate": 6.878992408644491e-07,
      "loss": 0.4491,
      "step": 372430
    },
    {
      "epoch": 152.45,
      "grad_norm": 1.9252309799194336,
      "learning_rate": 6.877869798742123e-07,
      "loss": 0.4533,
      "step": 372440
    },
    {
      "epoch": 152.46,
      "grad_norm": 2.2804317474365234,
      "learning_rate": 6.876747264696857e-07,
      "loss": 0.4457,
      "step": 372450
    },
    {
      "epoch": 152.46,
      "grad_norm": 2.1947498321533203,
      "learning_rate": 6.875624806513838e-07,
      "loss": 0.4462,
      "step": 372460
    },
    {
      "epoch": 152.46,
      "grad_norm": 1.7030545473098755,
      "learning_rate": 6.874502424198201e-07,
      "loss": 0.4328,
      "step": 372470
    },
    {
      "epoch": 152.47,
      "grad_norm": 2.307889699935913,
      "learning_rate": 6.873380117755091e-07,
      "loss": 0.4468,
      "step": 372480
    },
    {
      "epoch": 152.47,
      "grad_norm": 2.4278464317321777,
      "learning_rate": 6.872257887189647e-07,
      "loss": 0.4396,
      "step": 372490
    },
    {
      "epoch": 152.48,
      "grad_norm": 2.2323153018951416,
      "learning_rate": 6.87113573250701e-07,
      "loss": 0.4471,
      "step": 372500
    },
    {
      "epoch": 152.48,
      "grad_norm": 2.5202383995056152,
      "learning_rate": 6.870013653712325e-07,
      "loss": 0.4691,
      "step": 372510
    },
    {
      "epoch": 152.48,
      "grad_norm": 1.6934880018234253,
      "learning_rate": 6.868891650810719e-07,
      "loss": 0.4437,
      "step": 372520
    },
    {
      "epoch": 152.49,
      "grad_norm": 2.1714818477630615,
      "learning_rate": 6.867769723807351e-07,
      "loss": 0.4342,
      "step": 372530
    },
    {
      "epoch": 152.49,
      "grad_norm": 1.9610109329223633,
      "learning_rate": 6.866647872707353e-07,
      "loss": 0.4532,
      "step": 372540
    },
    {
      "epoch": 152.5,
      "grad_norm": 2.1126320362091064,
      "learning_rate": 6.86552609751586e-07,
      "loss": 0.448,
      "step": 372550
    },
    {
      "epoch": 152.5,
      "grad_norm": 1.9757604598999023,
      "learning_rate": 6.864404398238013e-07,
      "loss": 0.4449,
      "step": 372560
    },
    {
      "epoch": 152.51,
      "grad_norm": 1.9841556549072266,
      "learning_rate": 6.863282774878951e-07,
      "loss": 0.4468,
      "step": 372570
    },
    {
      "epoch": 152.51,
      "grad_norm": 2.521005868911743,
      "learning_rate": 6.86216122744381e-07,
      "loss": 0.44,
      "step": 372580
    },
    {
      "epoch": 152.51,
      "grad_norm": 1.9753316640853882,
      "learning_rate": 6.861039755937732e-07,
      "loss": 0.4382,
      "step": 372590
    },
    {
      "epoch": 152.52,
      "grad_norm": 1.957302212715149,
      "learning_rate": 6.859918360365856e-07,
      "loss": 0.4566,
      "step": 372600
    },
    {
      "epoch": 152.52,
      "grad_norm": 1.8977267742156982,
      "learning_rate": 6.858797040733305e-07,
      "loss": 0.4452,
      "step": 372610
    },
    {
      "epoch": 152.53,
      "grad_norm": 2.320809841156006,
      "learning_rate": 6.857675797045227e-07,
      "loss": 0.4523,
      "step": 372620
    },
    {
      "epoch": 152.53,
      "grad_norm": 2.488999366760254,
      "learning_rate": 6.856554629306758e-07,
      "loss": 0.4614,
      "step": 372630
    },
    {
      "epoch": 152.53,
      "grad_norm": 1.9186818599700928,
      "learning_rate": 6.855433537523029e-07,
      "loss": 0.4516,
      "step": 372640
    },
    {
      "epoch": 152.54,
      "grad_norm": 1.9048844575881958,
      "learning_rate": 6.85431252169918e-07,
      "loss": 0.4585,
      "step": 372650
    },
    {
      "epoch": 152.54,
      "grad_norm": 2.2932701110839844,
      "learning_rate": 6.853191581840343e-07,
      "loss": 0.444,
      "step": 372660
    },
    {
      "epoch": 152.55,
      "grad_norm": 1.90254545211792,
      "learning_rate": 6.852070717951655e-07,
      "loss": 0.4553,
      "step": 372670
    },
    {
      "epoch": 152.55,
      "grad_norm": 1.7471007108688354,
      "learning_rate": 6.850949930038249e-07,
      "loss": 0.4529,
      "step": 372680
    },
    {
      "epoch": 152.55,
      "grad_norm": 1.97896146774292,
      "learning_rate": 6.849829218105261e-07,
      "loss": 0.4434,
      "step": 372690
    },
    {
      "epoch": 152.56,
      "grad_norm": 2.4306399822235107,
      "learning_rate": 6.84870858215782e-07,
      "loss": 0.4404,
      "step": 372700
    },
    {
      "epoch": 152.56,
      "grad_norm": 1.9391441345214844,
      "learning_rate": 6.847588022201065e-07,
      "loss": 0.4352,
      "step": 372710
    },
    {
      "epoch": 152.57,
      "grad_norm": 2.061717987060547,
      "learning_rate": 6.846467538240131e-07,
      "loss": 0.4364,
      "step": 372720
    },
    {
      "epoch": 152.57,
      "grad_norm": 1.612840175628662,
      "learning_rate": 6.845347130280154e-07,
      "loss": 0.4498,
      "step": 372730
    },
    {
      "epoch": 152.57,
      "grad_norm": 2.1954267024993896,
      "learning_rate": 6.844226798326251e-07,
      "loss": 0.4472,
      "step": 372740
    },
    {
      "epoch": 152.58,
      "grad_norm": 1.8379857540130615,
      "learning_rate": 6.843106542383566e-07,
      "loss": 0.4498,
      "step": 372750
    },
    {
      "epoch": 152.58,
      "grad_norm": 1.9191431999206543,
      "learning_rate": 6.841986362457223e-07,
      "loss": 0.4493,
      "step": 372760
    },
    {
      "epoch": 152.59,
      "grad_norm": 1.8604872226715088,
      "learning_rate": 6.840866258552361e-07,
      "loss": 0.4421,
      "step": 372770
    },
    {
      "epoch": 152.59,
      "grad_norm": 1.601985216140747,
      "learning_rate": 6.839746230674108e-07,
      "loss": 0.4503,
      "step": 372780
    },
    {
      "epoch": 152.6,
      "grad_norm": 1.8731896877288818,
      "learning_rate": 6.838626278827593e-07,
      "loss": 0.4384,
      "step": 372790
    },
    {
      "epoch": 152.6,
      "grad_norm": 1.672429084777832,
      "learning_rate": 6.837506403017948e-07,
      "loss": 0.4627,
      "step": 372800
    },
    {
      "epoch": 152.6,
      "grad_norm": 1.9328572750091553,
      "learning_rate": 6.836386603250303e-07,
      "loss": 0.4393,
      "step": 372810
    },
    {
      "epoch": 152.61,
      "grad_norm": 2.81919002532959,
      "learning_rate": 6.835266879529788e-07,
      "loss": 0.4589,
      "step": 372820
    },
    {
      "epoch": 152.61,
      "grad_norm": 2.2739460468292236,
      "learning_rate": 6.834147231861532e-07,
      "loss": 0.4789,
      "step": 372830
    },
    {
      "epoch": 152.62,
      "grad_norm": 2.143709421157837,
      "learning_rate": 6.833027660250662e-07,
      "loss": 0.4536,
      "step": 372840
    },
    {
      "epoch": 152.62,
      "grad_norm": 1.7355464696884155,
      "learning_rate": 6.831908164702311e-07,
      "loss": 0.429,
      "step": 372850
    },
    {
      "epoch": 152.62,
      "grad_norm": 2.251363515853882,
      "learning_rate": 6.830788745221601e-07,
      "loss": 0.4366,
      "step": 372860
    },
    {
      "epoch": 152.63,
      "grad_norm": 1.9146300554275513,
      "learning_rate": 6.829669401813667e-07,
      "loss": 0.4452,
      "step": 372870
    },
    {
      "epoch": 152.63,
      "grad_norm": 2.0206189155578613,
      "learning_rate": 6.828550134483629e-07,
      "loss": 0.4437,
      "step": 372880
    },
    {
      "epoch": 152.64,
      "grad_norm": 1.824819564819336,
      "learning_rate": 6.82743094323662e-07,
      "loss": 0.4524,
      "step": 372890
    },
    {
      "epoch": 152.64,
      "grad_norm": 3.3967576026916504,
      "learning_rate": 6.826311828077767e-07,
      "loss": 0.4505,
      "step": 372900
    },
    {
      "epoch": 152.64,
      "grad_norm": 1.752135157585144,
      "learning_rate": 6.825192789012192e-07,
      "loss": 0.4525,
      "step": 372910
    },
    {
      "epoch": 152.65,
      "grad_norm": 2.2206532955169678,
      "learning_rate": 6.824073826045025e-07,
      "loss": 0.4376,
      "step": 372920
    },
    {
      "epoch": 152.65,
      "grad_norm": 2.2838962078094482,
      "learning_rate": 6.822954939181388e-07,
      "loss": 0.4397,
      "step": 372930
    },
    {
      "epoch": 152.66,
      "grad_norm": 2.2640206813812256,
      "learning_rate": 6.821836128426413e-07,
      "loss": 0.4458,
      "step": 372940
    },
    {
      "epoch": 152.66,
      "grad_norm": 1.7327485084533691,
      "learning_rate": 6.820717393785216e-07,
      "loss": 0.4311,
      "step": 372950
    },
    {
      "epoch": 152.66,
      "grad_norm": 1.987070918083191,
      "learning_rate": 6.81959873526293e-07,
      "loss": 0.4533,
      "step": 372960
    },
    {
      "epoch": 152.67,
      "grad_norm": 1.8288962841033936,
      "learning_rate": 6.818480152864676e-07,
      "loss": 0.4618,
      "step": 372970
    },
    {
      "epoch": 152.67,
      "grad_norm": 1.9844475984573364,
      "learning_rate": 6.817361646595577e-07,
      "loss": 0.461,
      "step": 372980
    },
    {
      "epoch": 152.68,
      "grad_norm": 2.2458033561706543,
      "learning_rate": 6.816243216460759e-07,
      "loss": 0.4477,
      "step": 372990
    },
    {
      "epoch": 152.68,
      "grad_norm": 1.8968021869659424,
      "learning_rate": 6.815124862465343e-07,
      "loss": 0.4501,
      "step": 373000
    },
    {
      "epoch": 152.69,
      "grad_norm": 2.2202436923980713,
      "learning_rate": 6.814006584614452e-07,
      "loss": 0.4535,
      "step": 373010
    },
    {
      "epoch": 152.69,
      "grad_norm": 1.9800658226013184,
      "learning_rate": 6.812888382913211e-07,
      "loss": 0.4455,
      "step": 373020
    },
    {
      "epoch": 152.69,
      "grad_norm": 2.401289939880371,
      "learning_rate": 6.811770257366743e-07,
      "loss": 0.4312,
      "step": 373030
    },
    {
      "epoch": 152.7,
      "grad_norm": 2.496103048324585,
      "learning_rate": 6.810652207980172e-07,
      "loss": 0.4506,
      "step": 373040
    },
    {
      "epoch": 152.7,
      "grad_norm": 1.8988456726074219,
      "learning_rate": 6.809534234758613e-07,
      "loss": 0.4341,
      "step": 373050
    },
    {
      "epoch": 152.71,
      "grad_norm": 2.4400603771209717,
      "learning_rate": 6.808416337707186e-07,
      "loss": 0.4595,
      "step": 373060
    },
    {
      "epoch": 152.71,
      "grad_norm": 1.7365838289260864,
      "learning_rate": 6.807298516831018e-07,
      "loss": 0.4497,
      "step": 373070
    },
    {
      "epoch": 152.71,
      "grad_norm": 2.0901334285736084,
      "learning_rate": 6.80618077213522e-07,
      "loss": 0.4454,
      "step": 373080
    },
    {
      "epoch": 152.72,
      "grad_norm": 1.7383536100387573,
      "learning_rate": 6.805063103624926e-07,
      "loss": 0.4598,
      "step": 373090
    },
    {
      "epoch": 152.72,
      "grad_norm": 1.760277509689331,
      "learning_rate": 6.80394551130525e-07,
      "loss": 0.4728,
      "step": 373100
    },
    {
      "epoch": 152.73,
      "grad_norm": 2.0520496368408203,
      "learning_rate": 6.802827995181313e-07,
      "loss": 0.4467,
      "step": 373110
    },
    {
      "epoch": 152.73,
      "grad_norm": 2.0277483463287354,
      "learning_rate": 6.80171055525823e-07,
      "loss": 0.4464,
      "step": 373120
    },
    {
      "epoch": 152.73,
      "grad_norm": 2.437147617340088,
      "learning_rate": 6.800593191541121e-07,
      "loss": 0.4531,
      "step": 373130
    },
    {
      "epoch": 152.74,
      "grad_norm": 2.349815845489502,
      "learning_rate": 6.799475904035107e-07,
      "loss": 0.4435,
      "step": 373140
    },
    {
      "epoch": 152.74,
      "grad_norm": 2.304212808609009,
      "learning_rate": 6.798358692745307e-07,
      "loss": 0.4411,
      "step": 373150
    },
    {
      "epoch": 152.75,
      "grad_norm": 1.8299373388290405,
      "learning_rate": 6.797241557676832e-07,
      "loss": 0.4623,
      "step": 373160
    },
    {
      "epoch": 152.75,
      "grad_norm": 2.231196641921997,
      "learning_rate": 6.796124498834806e-07,
      "loss": 0.4382,
      "step": 373170
    },
    {
      "epoch": 152.75,
      "grad_norm": 2.0677988529205322,
      "learning_rate": 6.795007516224353e-07,
      "loss": 0.4564,
      "step": 373180
    },
    {
      "epoch": 152.76,
      "grad_norm": 2.2184133529663086,
      "learning_rate": 6.793890609850571e-07,
      "loss": 0.4422,
      "step": 373190
    },
    {
      "epoch": 152.76,
      "grad_norm": 2.2408370971679688,
      "learning_rate": 6.792773779718586e-07,
      "loss": 0.445,
      "step": 373200
    },
    {
      "epoch": 152.77,
      "grad_norm": 1.8312885761260986,
      "learning_rate": 6.791657025833517e-07,
      "loss": 0.4313,
      "step": 373210
    },
    {
      "epoch": 152.77,
      "grad_norm": 1.8369251489639282,
      "learning_rate": 6.790540348200472e-07,
      "loss": 0.4559,
      "step": 373220
    },
    {
      "epoch": 152.78,
      "grad_norm": 2.23702335357666,
      "learning_rate": 6.789423746824575e-07,
      "loss": 0.4544,
      "step": 373230
    },
    {
      "epoch": 152.78,
      "grad_norm": 2.3628036975860596,
      "learning_rate": 6.788307221710934e-07,
      "loss": 0.4671,
      "step": 373240
    },
    {
      "epoch": 152.78,
      "grad_norm": 2.4100067615509033,
      "learning_rate": 6.787190772864669e-07,
      "loss": 0.4533,
      "step": 373250
    },
    {
      "epoch": 152.79,
      "grad_norm": 1.726356029510498,
      "learning_rate": 6.786074400290892e-07,
      "loss": 0.447,
      "step": 373260
    },
    {
      "epoch": 152.79,
      "grad_norm": 2.202202320098877,
      "learning_rate": 6.784958103994708e-07,
      "loss": 0.4569,
      "step": 373270
    },
    {
      "epoch": 152.8,
      "grad_norm": 1.8304872512817383,
      "learning_rate": 6.783841883981246e-07,
      "loss": 0.4412,
      "step": 373280
    },
    {
      "epoch": 152.8,
      "grad_norm": 1.946906328201294,
      "learning_rate": 6.782725740255615e-07,
      "loss": 0.4437,
      "step": 373290
    },
    {
      "epoch": 152.8,
      "grad_norm": 2.157167911529541,
      "learning_rate": 6.781609672822926e-07,
      "loss": 0.4417,
      "step": 373300
    },
    {
      "epoch": 152.81,
      "grad_norm": 1.954145908355713,
      "learning_rate": 6.780493681688289e-07,
      "loss": 0.4627,
      "step": 373310
    },
    {
      "epoch": 152.81,
      "grad_norm": 2.175377368927002,
      "learning_rate": 6.779377766856827e-07,
      "loss": 0.4442,
      "step": 373320
    },
    {
      "epoch": 152.82,
      "grad_norm": 1.9648797512054443,
      "learning_rate": 6.778261928333635e-07,
      "loss": 0.4337,
      "step": 373330
    },
    {
      "epoch": 152.82,
      "grad_norm": 1.7552306652069092,
      "learning_rate": 6.777146166123833e-07,
      "loss": 0.4516,
      "step": 373340
    },
    {
      "epoch": 152.82,
      "grad_norm": 2.303279399871826,
      "learning_rate": 6.776030480232533e-07,
      "loss": 0.4503,
      "step": 373350
    },
    {
      "epoch": 152.83,
      "grad_norm": 2.122535228729248,
      "learning_rate": 6.77491487066484e-07,
      "loss": 0.4476,
      "step": 373360
    },
    {
      "epoch": 152.83,
      "grad_norm": 1.7518792152404785,
      "learning_rate": 6.773799337425873e-07,
      "loss": 0.438,
      "step": 373370
    },
    {
      "epoch": 152.84,
      "grad_norm": 1.6954355239868164,
      "learning_rate": 6.772683880520738e-07,
      "loss": 0.4486,
      "step": 373380
    },
    {
      "epoch": 152.84,
      "grad_norm": 1.626821517944336,
      "learning_rate": 6.771568499954544e-07,
      "loss": 0.4489,
      "step": 373390
    },
    {
      "epoch": 152.84,
      "grad_norm": 1.9981273412704468,
      "learning_rate": 6.770453195732401e-07,
      "loss": 0.4414,
      "step": 373400
    },
    {
      "epoch": 152.85,
      "grad_norm": 1.8497041463851929,
      "learning_rate": 6.769337967859419e-07,
      "loss": 0.4501,
      "step": 373410
    },
    {
      "epoch": 152.85,
      "grad_norm": 2.0558547973632812,
      "learning_rate": 6.768222816340704e-07,
      "loss": 0.4473,
      "step": 373420
    },
    {
      "epoch": 152.86,
      "grad_norm": 2.0698275566101074,
      "learning_rate": 6.767107741181368e-07,
      "loss": 0.4383,
      "step": 373430
    },
    {
      "epoch": 152.86,
      "grad_norm": 1.857245683670044,
      "learning_rate": 6.765992742386515e-07,
      "loss": 0.4565,
      "step": 373440
    },
    {
      "epoch": 152.87,
      "grad_norm": 1.668405532836914,
      "learning_rate": 6.764877819961257e-07,
      "loss": 0.463,
      "step": 373450
    },
    {
      "epoch": 152.87,
      "grad_norm": 2.0640549659729004,
      "learning_rate": 6.763762973910698e-07,
      "loss": 0.4542,
      "step": 373460
    },
    {
      "epoch": 152.87,
      "grad_norm": 2.1357617378234863,
      "learning_rate": 6.762648204239947e-07,
      "loss": 0.4277,
      "step": 373470
    },
    {
      "epoch": 152.88,
      "grad_norm": 1.8923355340957642,
      "learning_rate": 6.76153351095411e-07,
      "loss": 0.4467,
      "step": 373480
    },
    {
      "epoch": 152.88,
      "grad_norm": 1.701991319656372,
      "learning_rate": 6.760418894058292e-07,
      "loss": 0.4523,
      "step": 373490
    },
    {
      "epoch": 152.89,
      "grad_norm": 2.4849138259887695,
      "learning_rate": 6.7593043535576e-07,
      "loss": 0.4121,
      "step": 373500
    },
    {
      "epoch": 152.89,
      "grad_norm": 1.7722320556640625,
      "learning_rate": 6.75818988945714e-07,
      "loss": 0.4406,
      "step": 373510
    },
    {
      "epoch": 152.89,
      "grad_norm": 3.5862042903900146,
      "learning_rate": 6.757075501762015e-07,
      "loss": 0.4474,
      "step": 373520
    },
    {
      "epoch": 152.9,
      "grad_norm": 1.938171148300171,
      "learning_rate": 6.755961190477332e-07,
      "loss": 0.4553,
      "step": 373530
    },
    {
      "epoch": 152.9,
      "grad_norm": 1.8355196714401245,
      "learning_rate": 6.754846955608193e-07,
      "loss": 0.4496,
      "step": 373540
    },
    {
      "epoch": 152.91,
      "grad_norm": 2.524775981903076,
      "learning_rate": 6.753732797159706e-07,
      "loss": 0.4642,
      "step": 373550
    },
    {
      "epoch": 152.91,
      "grad_norm": 2.0698082447052,
      "learning_rate": 6.752618715136974e-07,
      "loss": 0.4491,
      "step": 373560
    },
    {
      "epoch": 152.91,
      "grad_norm": 2.1016433238983154,
      "learning_rate": 6.751504709545097e-07,
      "loss": 0.464,
      "step": 373570
    },
    {
      "epoch": 152.92,
      "grad_norm": 2.1964030265808105,
      "learning_rate": 6.750390780389181e-07,
      "loss": 0.4623,
      "step": 373580
    },
    {
      "epoch": 152.92,
      "grad_norm": 2.149892568588257,
      "learning_rate": 6.74927692767433e-07,
      "loss": 0.4505,
      "step": 373590
    },
    {
      "epoch": 152.93,
      "grad_norm": 2.1812331676483154,
      "learning_rate": 6.748163151405639e-07,
      "loss": 0.4598,
      "step": 373600
    },
    {
      "epoch": 152.93,
      "grad_norm": 2.0587563514709473,
      "learning_rate": 6.74704945158822e-07,
      "loss": 0.4598,
      "step": 373610
    },
    {
      "epoch": 152.93,
      "grad_norm": 2.2011759281158447,
      "learning_rate": 6.74593582822717e-07,
      "loss": 0.4496,
      "step": 373620
    },
    {
      "epoch": 152.94,
      "grad_norm": 2.8270633220672607,
      "learning_rate": 6.744822281327596e-07,
      "loss": 0.4305,
      "step": 373630
    },
    {
      "epoch": 152.94,
      "grad_norm": 1.6755377054214478,
      "learning_rate": 6.743708810894588e-07,
      "loss": 0.4571,
      "step": 373640
    },
    {
      "epoch": 152.95,
      "grad_norm": 1.8459550142288208,
      "learning_rate": 6.742595416933243e-07,
      "loss": 0.4709,
      "step": 373650
    },
    {
      "epoch": 152.95,
      "grad_norm": 2.5644257068634033,
      "learning_rate": 6.741482099448679e-07,
      "loss": 0.4472,
      "step": 373660
    },
    {
      "epoch": 152.96,
      "grad_norm": 1.8914196491241455,
      "learning_rate": 6.740368858445985e-07,
      "loss": 0.4414,
      "step": 373670
    },
    {
      "epoch": 152.96,
      "grad_norm": 2.6534457206726074,
      "learning_rate": 6.739255693930264e-07,
      "loss": 0.4396,
      "step": 373680
    },
    {
      "epoch": 152.96,
      "grad_norm": 2.3922033309936523,
      "learning_rate": 6.738142605906616e-07,
      "loss": 0.4286,
      "step": 373690
    },
    {
      "epoch": 152.97,
      "grad_norm": 2.1056435108184814,
      "learning_rate": 6.737029594380136e-07,
      "loss": 0.4422,
      "step": 373700
    },
    {
      "epoch": 152.97,
      "grad_norm": 2.2637698650360107,
      "learning_rate": 6.735916659355924e-07,
      "loss": 0.4653,
      "step": 373710
    },
    {
      "epoch": 152.98,
      "grad_norm": 1.8081384897232056,
      "learning_rate": 6.73480380083908e-07,
      "loss": 0.4401,
      "step": 373720
    },
    {
      "epoch": 152.98,
      "grad_norm": 2.228541135787964,
      "learning_rate": 6.733691018834702e-07,
      "loss": 0.4404,
      "step": 373730
    },
    {
      "epoch": 152.98,
      "grad_norm": 2.5996718406677246,
      "learning_rate": 6.732578313347885e-07,
      "loss": 0.4522,
      "step": 373740
    },
    {
      "epoch": 152.99,
      "grad_norm": 2.0325820446014404,
      "learning_rate": 6.731465684383725e-07,
      "loss": 0.4245,
      "step": 373750
    },
    {
      "epoch": 152.99,
      "grad_norm": 1.8720053434371948,
      "learning_rate": 6.730353131947331e-07,
      "loss": 0.4575,
      "step": 373760
    },
    {
      "epoch": 153.0,
      "grad_norm": 1.9271851778030396,
      "learning_rate": 6.72924065604378e-07,
      "loss": 0.4559,
      "step": 373770
    },
    {
      "epoch": 153.0,
      "eval_loss": 0.4499657452106476,
      "eval_runtime": 52.2301,
      "eval_samples_per_second": 66.035,
      "eval_steps_per_second": 8.271,
      "step": 373779
    },
    {
      "epoch": 153.0,
      "grad_norm": 1.8665496110916138,
      "learning_rate": 6.72812825667818e-07,
      "loss": 0.4497,
      "step": 373780
    },
    {
      "epoch": 153.0,
      "grad_norm": 2.2635486125946045,
      "learning_rate": 6.727015933855622e-07,
      "loss": 0.4638,
      "step": 373790
    },
    {
      "epoch": 153.01,
      "grad_norm": 2.236748218536377,
      "learning_rate": 6.725903687581205e-07,
      "loss": 0.4451,
      "step": 373800
    },
    {
      "epoch": 153.01,
      "grad_norm": 1.7113916873931885,
      "learning_rate": 6.724791517860019e-07,
      "loss": 0.4396,
      "step": 373810
    },
    {
      "epoch": 153.02,
      "grad_norm": 1.824873685836792,
      "learning_rate": 6.723679424697166e-07,
      "loss": 0.4632,
      "step": 373820
    },
    {
      "epoch": 153.02,
      "grad_norm": 2.521662473678589,
      "learning_rate": 6.722567408097737e-07,
      "loss": 0.4529,
      "step": 373830
    },
    {
      "epoch": 153.02,
      "grad_norm": 2.457751512527466,
      "learning_rate": 6.721455468066815e-07,
      "loss": 0.4419,
      "step": 373840
    },
    {
      "epoch": 153.03,
      "grad_norm": 2.1916377544403076,
      "learning_rate": 6.720343604609514e-07,
      "loss": 0.4389,
      "step": 373850
    },
    {
      "epoch": 153.03,
      "grad_norm": 2.529609203338623,
      "learning_rate": 6.719231817730917e-07,
      "loss": 0.4326,
      "step": 373860
    },
    {
      "epoch": 153.04,
      "grad_norm": 2.146064281463623,
      "learning_rate": 6.718120107436117e-07,
      "loss": 0.4429,
      "step": 373870
    },
    {
      "epoch": 153.04,
      "grad_norm": 1.9808322191238403,
      "learning_rate": 6.717008473730208e-07,
      "loss": 0.4404,
      "step": 373880
    },
    {
      "epoch": 153.05,
      "grad_norm": 1.9732379913330078,
      "learning_rate": 6.715896916618281e-07,
      "loss": 0.4647,
      "step": 373890
    },
    {
      "epoch": 153.05,
      "grad_norm": 1.9962760210037231,
      "learning_rate": 6.714785436105434e-07,
      "loss": 0.4667,
      "step": 373900
    },
    {
      "epoch": 153.05,
      "grad_norm": 1.5965509414672852,
      "learning_rate": 6.713674032196746e-07,
      "loss": 0.4221,
      "step": 373910
    },
    {
      "epoch": 153.06,
      "grad_norm": 2.0087833404541016,
      "learning_rate": 6.712562704897313e-07,
      "loss": 0.4624,
      "step": 373920
    },
    {
      "epoch": 153.06,
      "grad_norm": 1.7835232019424438,
      "learning_rate": 6.71145145421223e-07,
      "loss": 0.4477,
      "step": 373930
    },
    {
      "epoch": 153.07,
      "grad_norm": 1.6490107774734497,
      "learning_rate": 6.710340280146584e-07,
      "loss": 0.4492,
      "step": 373940
    },
    {
      "epoch": 153.07,
      "grad_norm": 2.1646859645843506,
      "learning_rate": 6.709229182705466e-07,
      "loss": 0.451,
      "step": 373950
    },
    {
      "epoch": 153.07,
      "grad_norm": 1.8868521451950073,
      "learning_rate": 6.708118161893968e-07,
      "loss": 0.4486,
      "step": 373960
    },
    {
      "epoch": 153.08,
      "grad_norm": 2.003615379333496,
      "learning_rate": 6.707007217717175e-07,
      "loss": 0.4492,
      "step": 373970
    },
    {
      "epoch": 153.08,
      "grad_norm": 1.9948385953903198,
      "learning_rate": 6.705896350180181e-07,
      "loss": 0.451,
      "step": 373980
    },
    {
      "epoch": 153.09,
      "grad_norm": 2.046689987182617,
      "learning_rate": 6.70478555928807e-07,
      "loss": 0.4588,
      "step": 373990
    },
    {
      "epoch": 153.09,
      "grad_norm": 1.718151569366455,
      "learning_rate": 6.703674845045932e-07,
      "loss": 0.4206,
      "step": 374000
    },
    {
      "epoch": 153.09,
      "grad_norm": 2.3638195991516113,
      "learning_rate": 6.702564207458856e-07,
      "loss": 0.445,
      "step": 374010
    },
    {
      "epoch": 153.1,
      "grad_norm": 2.155341863632202,
      "learning_rate": 6.701453646531923e-07,
      "loss": 0.4642,
      "step": 374020
    },
    {
      "epoch": 153.1,
      "grad_norm": 1.8487573862075806,
      "learning_rate": 6.700343162270238e-07,
      "loss": 0.4272,
      "step": 374030
    },
    {
      "epoch": 153.11,
      "grad_norm": 2.118398427963257,
      "learning_rate": 6.699232754678873e-07,
      "loss": 0.433,
      "step": 374040
    },
    {
      "epoch": 153.11,
      "grad_norm": 1.9124640226364136,
      "learning_rate": 6.698122423762916e-07,
      "loss": 0.4646,
      "step": 374050
    },
    {
      "epoch": 153.12,
      "grad_norm": 2.047353744506836,
      "learning_rate": 6.697012169527454e-07,
      "loss": 0.4467,
      "step": 374060
    },
    {
      "epoch": 153.12,
      "grad_norm": 2.115772247314453,
      "learning_rate": 6.695901991977577e-07,
      "loss": 0.4553,
      "step": 374070
    },
    {
      "epoch": 153.12,
      "grad_norm": 2.1782588958740234,
      "learning_rate": 6.694791891118366e-07,
      "loss": 0.4424,
      "step": 374080
    },
    {
      "epoch": 153.13,
      "grad_norm": 2.119030475616455,
      "learning_rate": 6.693681866954907e-07,
      "loss": 0.4559,
      "step": 374090
    },
    {
      "epoch": 153.13,
      "grad_norm": 1.8282337188720703,
      "learning_rate": 6.692571919492287e-07,
      "loss": 0.4563,
      "step": 374100
    },
    {
      "epoch": 153.14,
      "grad_norm": 1.9145488739013672,
      "learning_rate": 6.691462048735589e-07,
      "loss": 0.4243,
      "step": 374110
    },
    {
      "epoch": 153.14,
      "grad_norm": 1.9666348695755005,
      "learning_rate": 6.690352254689899e-07,
      "loss": 0.4369,
      "step": 374120
    },
    {
      "epoch": 153.14,
      "grad_norm": 2.1248321533203125,
      "learning_rate": 6.689242537360298e-07,
      "loss": 0.4382,
      "step": 374130
    },
    {
      "epoch": 153.15,
      "grad_norm": 1.8319828510284424,
      "learning_rate": 6.68813289675187e-07,
      "loss": 0.4682,
      "step": 374140
    },
    {
      "epoch": 153.15,
      "grad_norm": 2.0237185955047607,
      "learning_rate": 6.687023332869698e-07,
      "loss": 0.4608,
      "step": 374150
    },
    {
      "epoch": 153.16,
      "grad_norm": 2.096651077270508,
      "learning_rate": 6.685913845718867e-07,
      "loss": 0.4512,
      "step": 374160
    },
    {
      "epoch": 153.16,
      "grad_norm": 2.0884299278259277,
      "learning_rate": 6.684804435304456e-07,
      "loss": 0.4479,
      "step": 374170
    },
    {
      "epoch": 153.16,
      "grad_norm": 1.668726921081543,
      "learning_rate": 6.683695101631553e-07,
      "loss": 0.4516,
      "step": 374180
    },
    {
      "epoch": 153.17,
      "grad_norm": 1.7916969060897827,
      "learning_rate": 6.682585844705232e-07,
      "loss": 0.4376,
      "step": 374190
    },
    {
      "epoch": 153.17,
      "grad_norm": 2.047468662261963,
      "learning_rate": 6.681476664530579e-07,
      "loss": 0.4575,
      "step": 374200
    },
    {
      "epoch": 153.18,
      "grad_norm": 2.03971529006958,
      "learning_rate": 6.680367561112673e-07,
      "loss": 0.4595,
      "step": 374210
    },
    {
      "epoch": 153.18,
      "grad_norm": 2.100170850753784,
      "learning_rate": 6.679258534456599e-07,
      "loss": 0.4737,
      "step": 374220
    },
    {
      "epoch": 153.18,
      "grad_norm": 2.3133461475372314,
      "learning_rate": 6.678149584567429e-07,
      "loss": 0.4478,
      "step": 374230
    },
    {
      "epoch": 153.19,
      "grad_norm": 1.8853483200073242,
      "learning_rate": 6.677040711450252e-07,
      "loss": 0.4626,
      "step": 374240
    },
    {
      "epoch": 153.19,
      "grad_norm": 1.8643019199371338,
      "learning_rate": 6.675931915110141e-07,
      "loss": 0.4378,
      "step": 374250
    },
    {
      "epoch": 153.2,
      "grad_norm": 2.062960624694824,
      "learning_rate": 6.674823195552176e-07,
      "loss": 0.4827,
      "step": 374260
    },
    {
      "epoch": 153.2,
      "grad_norm": 2.338874578475952,
      "learning_rate": 6.673714552781439e-07,
      "loss": 0.4536,
      "step": 374270
    },
    {
      "epoch": 153.21,
      "grad_norm": 1.7769745588302612,
      "learning_rate": 6.672605986803008e-07,
      "loss": 0.4417,
      "step": 374280
    },
    {
      "epoch": 153.21,
      "grad_norm": 2.204500675201416,
      "learning_rate": 6.671497497621959e-07,
      "loss": 0.4358,
      "step": 374290
    },
    {
      "epoch": 153.21,
      "grad_norm": 2.2963976860046387,
      "learning_rate": 6.670389085243372e-07,
      "loss": 0.4668,
      "step": 374300
    },
    {
      "epoch": 153.22,
      "grad_norm": 2.015315532684326,
      "learning_rate": 6.669280749672322e-07,
      "loss": 0.4311,
      "step": 374310
    },
    {
      "epoch": 153.22,
      "grad_norm": 1.8014857769012451,
      "learning_rate": 6.668172490913888e-07,
      "loss": 0.448,
      "step": 374320
    },
    {
      "epoch": 153.23,
      "grad_norm": 1.7738415002822876,
      "learning_rate": 6.667064308973146e-07,
      "loss": 0.4702,
      "step": 374330
    },
    {
      "epoch": 153.23,
      "grad_norm": 2.1791369915008545,
      "learning_rate": 6.665956203855181e-07,
      "loss": 0.4228,
      "step": 374340
    },
    {
      "epoch": 153.23,
      "grad_norm": 1.9570766687393188,
      "learning_rate": 6.664848175565052e-07,
      "loss": 0.4458,
      "step": 374350
    },
    {
      "epoch": 153.24,
      "grad_norm": 1.9924901723861694,
      "learning_rate": 6.663740224107844e-07,
      "loss": 0.4265,
      "step": 374360
    },
    {
      "epoch": 153.24,
      "grad_norm": 2.2699105739593506,
      "learning_rate": 6.662632349488634e-07,
      "loss": 0.4322,
      "step": 374370
    },
    {
      "epoch": 153.25,
      "grad_norm": 1.712255835533142,
      "learning_rate": 6.661524551712492e-07,
      "loss": 0.4528,
      "step": 374380
    },
    {
      "epoch": 153.25,
      "grad_norm": 1.8841875791549683,
      "learning_rate": 6.660416830784497e-07,
      "loss": 0.4428,
      "step": 374390
    },
    {
      "epoch": 153.25,
      "grad_norm": 1.864357352256775,
      "learning_rate": 6.659309186709711e-07,
      "loss": 0.4467,
      "step": 374400
    },
    {
      "epoch": 153.26,
      "grad_norm": 2.055532693862915,
      "learning_rate": 6.658201619493229e-07,
      "loss": 0.4443,
      "step": 374410
    },
    {
      "epoch": 153.26,
      "grad_norm": 1.8128039836883545,
      "learning_rate": 6.657094129140113e-07,
      "loss": 0.4237,
      "step": 374420
    },
    {
      "epoch": 153.27,
      "grad_norm": 1.854194164276123,
      "learning_rate": 6.65598671565544e-07,
      "loss": 0.4468,
      "step": 374430
    },
    {
      "epoch": 153.27,
      "grad_norm": 1.9426440000534058,
      "learning_rate": 6.654879379044279e-07,
      "loss": 0.4646,
      "step": 374440
    },
    {
      "epoch": 153.27,
      "grad_norm": 2.116142511367798,
      "learning_rate": 6.653772119311705e-07,
      "loss": 0.452,
      "step": 374450
    },
    {
      "epoch": 153.28,
      "grad_norm": 1.8645439147949219,
      "learning_rate": 6.652664936462786e-07,
      "loss": 0.4466,
      "step": 374460
    },
    {
      "epoch": 153.28,
      "grad_norm": 1.630582332611084,
      "learning_rate": 6.6515578305026e-07,
      "loss": 0.4678,
      "step": 374470
    },
    {
      "epoch": 153.29,
      "grad_norm": 2.622225761413574,
      "learning_rate": 6.65045080143622e-07,
      "loss": 0.4613,
      "step": 374480
    },
    {
      "epoch": 153.29,
      "grad_norm": 2.106987714767456,
      "learning_rate": 6.649343849268708e-07,
      "loss": 0.4335,
      "step": 374490
    },
    {
      "epoch": 153.3,
      "grad_norm": 1.9610824584960938,
      "learning_rate": 6.648236974005138e-07,
      "loss": 0.461,
      "step": 374500
    },
    {
      "epoch": 153.3,
      "grad_norm": 1.8126925230026245,
      "learning_rate": 6.647130175650582e-07,
      "loss": 0.4486,
      "step": 374510
    },
    {
      "epoch": 153.3,
      "grad_norm": 2.1613123416900635,
      "learning_rate": 6.646023454210109e-07,
      "loss": 0.436,
      "step": 374520
    },
    {
      "epoch": 153.31,
      "grad_norm": 2.589736223220825,
      "learning_rate": 6.644916809688792e-07,
      "loss": 0.4567,
      "step": 374530
    },
    {
      "epoch": 153.31,
      "grad_norm": 2.006442070007324,
      "learning_rate": 6.643810242091696e-07,
      "loss": 0.4372,
      "step": 374540
    },
    {
      "epoch": 153.32,
      "grad_norm": 1.8949679136276245,
      "learning_rate": 6.642703751423892e-07,
      "loss": 0.4256,
      "step": 374550
    },
    {
      "epoch": 153.32,
      "grad_norm": 2.8929738998413086,
      "learning_rate": 6.641597337690449e-07,
      "loss": 0.454,
      "step": 374560
    },
    {
      "epoch": 153.32,
      "grad_norm": 1.9595617055892944,
      "learning_rate": 6.640491000896437e-07,
      "loss": 0.4512,
      "step": 374570
    },
    {
      "epoch": 153.33,
      "grad_norm": 1.7798304557800293,
      "learning_rate": 6.639384741046918e-07,
      "loss": 0.4387,
      "step": 374580
    },
    {
      "epoch": 153.33,
      "grad_norm": 1.912387728691101,
      "learning_rate": 6.638278558146957e-07,
      "loss": 0.4588,
      "step": 374590
    },
    {
      "epoch": 153.34,
      "grad_norm": 2.104599714279175,
      "learning_rate": 6.637172452201636e-07,
      "loss": 0.431,
      "step": 374600
    },
    {
      "epoch": 153.34,
      "grad_norm": 2.0749170780181885,
      "learning_rate": 6.636066423216014e-07,
      "loss": 0.4655,
      "step": 374610
    },
    {
      "epoch": 153.34,
      "grad_norm": 2.019562005996704,
      "learning_rate": 6.634960471195165e-07,
      "loss": 0.4545,
      "step": 374620
    },
    {
      "epoch": 153.35,
      "grad_norm": 1.9541170597076416,
      "learning_rate": 6.633854596144137e-07,
      "loss": 0.4343,
      "step": 374630
    },
    {
      "epoch": 153.35,
      "grad_norm": 1.9276123046875,
      "learning_rate": 6.63274879806801e-07,
      "loss": 0.4647,
      "step": 374640
    },
    {
      "epoch": 153.36,
      "grad_norm": 2.363931894302368,
      "learning_rate": 6.631643076971844e-07,
      "loss": 0.4709,
      "step": 374650
    },
    {
      "epoch": 153.36,
      "grad_norm": 1.8681154251098633,
      "learning_rate": 6.630537432860706e-07,
      "loss": 0.4378,
      "step": 374660
    },
    {
      "epoch": 153.36,
      "grad_norm": 2.299085855484009,
      "learning_rate": 6.629431865739659e-07,
      "loss": 0.4595,
      "step": 374670
    },
    {
      "epoch": 153.37,
      "grad_norm": 2.663698196411133,
      "learning_rate": 6.628326375613774e-07,
      "loss": 0.4354,
      "step": 374680
    },
    {
      "epoch": 153.37,
      "grad_norm": 1.973667860031128,
      "learning_rate": 6.627220962488106e-07,
      "loss": 0.4465,
      "step": 374690
    },
    {
      "epoch": 153.38,
      "grad_norm": 1.836869239807129,
      "learning_rate": 6.626115626367723e-07,
      "loss": 0.4425,
      "step": 374700
    },
    {
      "epoch": 153.38,
      "grad_norm": 2.3761343955993652,
      "learning_rate": 6.625010367257689e-07,
      "loss": 0.4401,
      "step": 374710
    },
    {
      "epoch": 153.39,
      "grad_norm": 2.210793972015381,
      "learning_rate": 6.623905185163067e-07,
      "loss": 0.442,
      "step": 374720
    },
    {
      "epoch": 153.39,
      "grad_norm": 1.9082640409469604,
      "learning_rate": 6.622800080088915e-07,
      "loss": 0.4417,
      "step": 374730
    },
    {
      "epoch": 153.39,
      "grad_norm": 1.7847844362258911,
      "learning_rate": 6.621695052040304e-07,
      "loss": 0.4443,
      "step": 374740
    },
    {
      "epoch": 153.4,
      "grad_norm": 2.246464490890503,
      "learning_rate": 6.620590101022291e-07,
      "loss": 0.46,
      "step": 374750
    },
    {
      "epoch": 153.4,
      "grad_norm": 2.146260976791382,
      "learning_rate": 6.619485227039937e-07,
      "loss": 0.4443,
      "step": 374760
    },
    {
      "epoch": 153.41,
      "grad_norm": 2.0207512378692627,
      "learning_rate": 6.618380430098302e-07,
      "loss": 0.4478,
      "step": 374770
    },
    {
      "epoch": 153.41,
      "grad_norm": 2.324305295944214,
      "learning_rate": 6.617275710202448e-07,
      "loss": 0.4303,
      "step": 374780
    },
    {
      "epoch": 153.41,
      "grad_norm": 1.9711833000183105,
      "learning_rate": 6.616171067357441e-07,
      "loss": 0.4597,
      "step": 374790
    },
    {
      "epoch": 153.42,
      "grad_norm": 1.840889811515808,
      "learning_rate": 6.615066501568336e-07,
      "loss": 0.4484,
      "step": 374800
    },
    {
      "epoch": 153.42,
      "grad_norm": 1.923166036605835,
      "learning_rate": 6.613962012840191e-07,
      "loss": 0.4443,
      "step": 374810
    },
    {
      "epoch": 153.43,
      "grad_norm": 2.091566324234009,
      "learning_rate": 6.61285760117807e-07,
      "loss": 0.4465,
      "step": 374820
    },
    {
      "epoch": 153.43,
      "grad_norm": 2.361069917678833,
      "learning_rate": 6.611753266587029e-07,
      "loss": 0.4546,
      "step": 374830
    },
    {
      "epoch": 153.43,
      "grad_norm": 2.024810791015625,
      "learning_rate": 6.610649009072129e-07,
      "loss": 0.4501,
      "step": 374840
    },
    {
      "epoch": 153.44,
      "grad_norm": 2.4798214435577393,
      "learning_rate": 6.609544828638426e-07,
      "loss": 0.4622,
      "step": 374850
    },
    {
      "epoch": 153.44,
      "grad_norm": 2.1660690307617188,
      "learning_rate": 6.608440725290982e-07,
      "loss": 0.4524,
      "step": 374860
    },
    {
      "epoch": 153.45,
      "grad_norm": 1.9312018156051636,
      "learning_rate": 6.60733669903485e-07,
      "loss": 0.4501,
      "step": 374870
    },
    {
      "epoch": 153.45,
      "grad_norm": 2.3301165103912354,
      "learning_rate": 6.606232749875091e-07,
      "loss": 0.4368,
      "step": 374880
    },
    {
      "epoch": 153.45,
      "grad_norm": 2.444288730621338,
      "learning_rate": 6.60512887781676e-07,
      "loss": 0.442,
      "step": 374890
    },
    {
      "epoch": 153.46,
      "grad_norm": 1.8308424949645996,
      "learning_rate": 6.604025082864914e-07,
      "loss": 0.4434,
      "step": 374900
    },
    {
      "epoch": 153.46,
      "grad_norm": 1.7770119905471802,
      "learning_rate": 6.602921365024608e-07,
      "loss": 0.4306,
      "step": 374910
    },
    {
      "epoch": 153.47,
      "grad_norm": 2.3045060634613037,
      "learning_rate": 6.601817724300903e-07,
      "loss": 0.4309,
      "step": 374920
    },
    {
      "epoch": 153.47,
      "grad_norm": 2.8327395915985107,
      "learning_rate": 6.600714160698855e-07,
      "loss": 0.46,
      "step": 374930
    },
    {
      "epoch": 153.48,
      "grad_norm": 2.1240625381469727,
      "learning_rate": 6.599610674223511e-07,
      "loss": 0.4582,
      "step": 374940
    },
    {
      "epoch": 153.48,
      "grad_norm": 1.5776125192642212,
      "learning_rate": 6.59850726487993e-07,
      "loss": 0.4444,
      "step": 374950
    },
    {
      "epoch": 153.48,
      "grad_norm": 2.0359036922454834,
      "learning_rate": 6.597403932673167e-07,
      "loss": 0.4542,
      "step": 374960
    },
    {
      "epoch": 153.49,
      "grad_norm": 1.8971627950668335,
      "learning_rate": 6.596300677608266e-07,
      "loss": 0.4243,
      "step": 374970
    },
    {
      "epoch": 153.49,
      "grad_norm": 2.0557916164398193,
      "learning_rate": 6.595197499690302e-07,
      "loss": 0.4423,
      "step": 374980
    },
    {
      "epoch": 153.5,
      "grad_norm": 2.1406209468841553,
      "learning_rate": 6.594094398924316e-07,
      "loss": 0.4481,
      "step": 374990
    },
    {
      "epoch": 153.5,
      "grad_norm": 2.25728440284729,
      "learning_rate": 6.59299137531536e-07,
      "loss": 0.4614,
      "step": 375000
    },
    {
      "epoch": 153.5,
      "grad_norm": 2.0653412342071533,
      "learning_rate": 6.591888428868492e-07,
      "loss": 0.4504,
      "step": 375010
    },
    {
      "epoch": 153.51,
      "grad_norm": 2.011481523513794,
      "learning_rate": 6.590785559588762e-07,
      "loss": 0.4592,
      "step": 375020
    },
    {
      "epoch": 153.51,
      "grad_norm": 1.828759789466858,
      "learning_rate": 6.58968276748122e-07,
      "loss": 0.4375,
      "step": 375030
    },
    {
      "epoch": 153.52,
      "grad_norm": 1.901202917098999,
      "learning_rate": 6.588580052550921e-07,
      "loss": 0.464,
      "step": 375040
    },
    {
      "epoch": 153.52,
      "grad_norm": 1.71305251121521,
      "learning_rate": 6.587477414802915e-07,
      "loss": 0.4379,
      "step": 375050
    },
    {
      "epoch": 153.52,
      "grad_norm": 2.2005486488342285,
      "learning_rate": 6.586374854242259e-07,
      "loss": 0.4467,
      "step": 375060
    },
    {
      "epoch": 153.53,
      "grad_norm": 2.1238348484039307,
      "learning_rate": 6.585272370873989e-07,
      "loss": 0.4426,
      "step": 375070
    },
    {
      "epoch": 153.53,
      "grad_norm": 2.8826630115509033,
      "learning_rate": 6.584169964703168e-07,
      "loss": 0.4399,
      "step": 375080
    },
    {
      "epoch": 153.54,
      "grad_norm": 1.7187412977218628,
      "learning_rate": 6.583067635734838e-07,
      "loss": 0.461,
      "step": 375090
    },
    {
      "epoch": 153.54,
      "grad_norm": 1.5493711233139038,
      "learning_rate": 6.581965383974055e-07,
      "loss": 0.4437,
      "step": 375100
    },
    {
      "epoch": 153.54,
      "grad_norm": 2.346844434738159,
      "learning_rate": 6.580863209425863e-07,
      "loss": 0.4311,
      "step": 375110
    },
    {
      "epoch": 153.55,
      "grad_norm": 1.7473171949386597,
      "learning_rate": 6.579761112095315e-07,
      "loss": 0.4324,
      "step": 375120
    },
    {
      "epoch": 153.55,
      "grad_norm": 2.0298962593078613,
      "learning_rate": 6.57865909198746e-07,
      "loss": 0.4681,
      "step": 375130
    },
    {
      "epoch": 153.56,
      "grad_norm": 1.8683885335922241,
      "learning_rate": 6.577557149107342e-07,
      "loss": 0.4341,
      "step": 375140
    },
    {
      "epoch": 153.56,
      "grad_norm": 2.413625478744507,
      "learning_rate": 6.576455283460011e-07,
      "loss": 0.4589,
      "step": 375150
    },
    {
      "epoch": 153.57,
      "grad_norm": 2.1587982177734375,
      "learning_rate": 6.575353495050509e-07,
      "loss": 0.4409,
      "step": 375160
    },
    {
      "epoch": 153.57,
      "grad_norm": 1.8703621625900269,
      "learning_rate": 6.574251783883895e-07,
      "loss": 0.4539,
      "step": 375170
    },
    {
      "epoch": 153.57,
      "grad_norm": 2.432311773300171,
      "learning_rate": 6.573150149965209e-07,
      "loss": 0.4404,
      "step": 375180
    },
    {
      "epoch": 153.58,
      "grad_norm": 1.9891703128814697,
      "learning_rate": 6.5720485932995e-07,
      "loss": 0.4606,
      "step": 375190
    },
    {
      "epoch": 153.58,
      "grad_norm": 1.9332938194274902,
      "learning_rate": 6.570947113891816e-07,
      "loss": 0.48,
      "step": 375200
    },
    {
      "epoch": 153.59,
      "grad_norm": 2.0123019218444824,
      "learning_rate": 6.569845711747193e-07,
      "loss": 0.4378,
      "step": 375210
    },
    {
      "epoch": 153.59,
      "grad_norm": 1.8631303310394287,
      "learning_rate": 6.568744386870682e-07,
      "loss": 0.4239,
      "step": 375220
    },
    {
      "epoch": 153.59,
      "grad_norm": 1.8488754034042358,
      "learning_rate": 6.567643139267329e-07,
      "loss": 0.4719,
      "step": 375230
    },
    {
      "epoch": 153.6,
      "grad_norm": 1.9149426221847534,
      "learning_rate": 6.566541968942175e-07,
      "loss": 0.4312,
      "step": 375240
    },
    {
      "epoch": 153.6,
      "grad_norm": 2.281435489654541,
      "learning_rate": 6.56544087590027e-07,
      "loss": 0.4499,
      "step": 375250
    },
    {
      "epoch": 153.61,
      "grad_norm": 2.0516459941864014,
      "learning_rate": 6.564339860146652e-07,
      "loss": 0.4245,
      "step": 375260
    },
    {
      "epoch": 153.61,
      "grad_norm": 1.805364966392517,
      "learning_rate": 6.56323892168637e-07,
      "loss": 0.4402,
      "step": 375270
    },
    {
      "epoch": 153.61,
      "grad_norm": 2.1563796997070312,
      "learning_rate": 6.562138060524463e-07,
      "loss": 0.4812,
      "step": 375280
    },
    {
      "epoch": 153.62,
      "grad_norm": 1.9856075048446655,
      "learning_rate": 6.561037276665977e-07,
      "loss": 0.4481,
      "step": 375290
    },
    {
      "epoch": 153.62,
      "grad_norm": 1.8662835359573364,
      "learning_rate": 6.55993657011595e-07,
      "loss": 0.4338,
      "step": 375300
    },
    {
      "epoch": 153.63,
      "grad_norm": 2.2555582523345947,
      "learning_rate": 6.558835940879432e-07,
      "loss": 0.4529,
      "step": 375310
    },
    {
      "epoch": 153.63,
      "grad_norm": 1.602670431137085,
      "learning_rate": 6.557735388961457e-07,
      "loss": 0.4467,
      "step": 375320
    },
    {
      "epoch": 153.63,
      "grad_norm": 2.4721696376800537,
      "learning_rate": 6.55663491436707e-07,
      "loss": 0.4465,
      "step": 375330
    },
    {
      "epoch": 153.64,
      "grad_norm": 2.0364716053009033,
      "learning_rate": 6.555534517101313e-07,
      "loss": 0.4392,
      "step": 375340
    },
    {
      "epoch": 153.64,
      "grad_norm": 2.068061113357544,
      "learning_rate": 6.554434197169223e-07,
      "loss": 0.4301,
      "step": 375350
    },
    {
      "epoch": 153.65,
      "grad_norm": 2.051229953765869,
      "learning_rate": 6.553333954575844e-07,
      "loss": 0.4651,
      "step": 375360
    },
    {
      "epoch": 153.65,
      "grad_norm": 2.0915169715881348,
      "learning_rate": 6.552233789326214e-07,
      "loss": 0.4465,
      "step": 375370
    },
    {
      "epoch": 153.66,
      "grad_norm": 2.3204431533813477,
      "learning_rate": 6.551133701425374e-07,
      "loss": 0.4646,
      "step": 375380
    },
    {
      "epoch": 153.66,
      "grad_norm": 1.8016859292984009,
      "learning_rate": 6.550033690878365e-07,
      "loss": 0.4425,
      "step": 375390
    },
    {
      "epoch": 153.66,
      "grad_norm": 2.1077656745910645,
      "learning_rate": 6.548933757690221e-07,
      "loss": 0.4472,
      "step": 375400
    },
    {
      "epoch": 153.67,
      "grad_norm": 1.8836232423782349,
      "learning_rate": 6.547833901865982e-07,
      "loss": 0.4457,
      "step": 375410
    },
    {
      "epoch": 153.67,
      "grad_norm": 2.1971569061279297,
      "learning_rate": 6.546734123410691e-07,
      "loss": 0.433,
      "step": 375420
    },
    {
      "epoch": 153.68,
      "grad_norm": 2.12111496925354,
      "learning_rate": 6.545634422329382e-07,
      "loss": 0.4585,
      "step": 375430
    },
    {
      "epoch": 153.68,
      "grad_norm": 1.9537776708602905,
      "learning_rate": 6.544534798627093e-07,
      "loss": 0.4532,
      "step": 375440
    },
    {
      "epoch": 153.68,
      "grad_norm": 1.9069972038269043,
      "learning_rate": 6.54343525230886e-07,
      "loss": 0.4496,
      "step": 375450
    },
    {
      "epoch": 153.69,
      "grad_norm": 1.881554126739502,
      "learning_rate": 6.542335783379724e-07,
      "loss": 0.452,
      "step": 375460
    },
    {
      "epoch": 153.69,
      "grad_norm": 2.3339955806732178,
      "learning_rate": 6.541236391844716e-07,
      "loss": 0.4464,
      "step": 375470
    },
    {
      "epoch": 153.7,
      "grad_norm": 2.3266541957855225,
      "learning_rate": 6.540137077708877e-07,
      "loss": 0.4583,
      "step": 375480
    },
    {
      "epoch": 153.7,
      "grad_norm": 1.915343165397644,
      "learning_rate": 6.539037840977238e-07,
      "loss": 0.4526,
      "step": 375490
    },
    {
      "epoch": 153.7,
      "grad_norm": 2.1602718830108643,
      "learning_rate": 6.537938681654841e-07,
      "loss": 0.4507,
      "step": 375500
    },
    {
      "epoch": 153.71,
      "grad_norm": 2.327552556991577,
      "learning_rate": 6.536839599746721e-07,
      "loss": 0.4501,
      "step": 375510
    },
    {
      "epoch": 153.71,
      "grad_norm": 1.8702495098114014,
      "learning_rate": 6.535740595257903e-07,
      "loss": 0.4565,
      "step": 375520
    },
    {
      "epoch": 153.72,
      "grad_norm": 1.6697126626968384,
      "learning_rate": 6.534641668193419e-07,
      "loss": 0.4522,
      "step": 375530
    },
    {
      "epoch": 153.72,
      "grad_norm": 1.9023793935775757,
      "learning_rate": 6.53354281855832e-07,
      "loss": 0.443,
      "step": 375540
    },
    {
      "epoch": 153.72,
      "grad_norm": 2.776609182357788,
      "learning_rate": 6.532444046357631e-07,
      "loss": 0.4394,
      "step": 375550
    },
    {
      "epoch": 153.73,
      "grad_norm": 2.3333306312561035,
      "learning_rate": 6.531345351596382e-07,
      "loss": 0.4376,
      "step": 375560
    },
    {
      "epoch": 153.73,
      "grad_norm": 2.0218591690063477,
      "learning_rate": 6.530246734279612e-07,
      "loss": 0.4667,
      "step": 375570
    },
    {
      "epoch": 153.74,
      "grad_norm": 2.7218074798583984,
      "learning_rate": 6.529148194412352e-07,
      "loss": 0.4342,
      "step": 375580
    },
    {
      "epoch": 153.74,
      "grad_norm": 2.6039977073669434,
      "learning_rate": 6.528049731999633e-07,
      "loss": 0.4405,
      "step": 375590
    },
    {
      "epoch": 153.75,
      "grad_norm": 2.146773338317871,
      "learning_rate": 6.526951347046486e-07,
      "loss": 0.4423,
      "step": 375600
    },
    {
      "epoch": 153.75,
      "grad_norm": 2.2253165245056152,
      "learning_rate": 6.52585303955794e-07,
      "loss": 0.4222,
      "step": 375610
    },
    {
      "epoch": 153.75,
      "grad_norm": 2.2107455730438232,
      "learning_rate": 6.524754809539033e-07,
      "loss": 0.4415,
      "step": 375620
    },
    {
      "epoch": 153.76,
      "grad_norm": 2.164517402648926,
      "learning_rate": 6.523656656994793e-07,
      "loss": 0.4416,
      "step": 375630
    },
    {
      "epoch": 153.76,
      "grad_norm": 2.6800239086151123,
      "learning_rate": 6.522558581930255e-07,
      "loss": 0.4435,
      "step": 375640
    },
    {
      "epoch": 153.77,
      "grad_norm": 2.6555099487304688,
      "learning_rate": 6.521460584350439e-07,
      "loss": 0.4536,
      "step": 375650
    },
    {
      "epoch": 153.77,
      "grad_norm": 1.907224178314209,
      "learning_rate": 6.520362664260378e-07,
      "loss": 0.4544,
      "step": 375660
    },
    {
      "epoch": 153.77,
      "grad_norm": 1.844131350517273,
      "learning_rate": 6.519264821665103e-07,
      "loss": 0.4448,
      "step": 375670
    },
    {
      "epoch": 153.78,
      "grad_norm": 2.0936384201049805,
      "learning_rate": 6.518167056569645e-07,
      "loss": 0.4448,
      "step": 375680
    },
    {
      "epoch": 153.78,
      "grad_norm": 2.0385398864746094,
      "learning_rate": 6.517069368979029e-07,
      "loss": 0.4573,
      "step": 375690
    },
    {
      "epoch": 153.79,
      "grad_norm": 1.8415979146957397,
      "learning_rate": 6.515971758898288e-07,
      "loss": 0.4657,
      "step": 375700
    },
    {
      "epoch": 153.79,
      "grad_norm": 2.886249542236328,
      "learning_rate": 6.514874226332445e-07,
      "loss": 0.4612,
      "step": 375710
    },
    {
      "epoch": 153.79,
      "grad_norm": 1.9081525802612305,
      "learning_rate": 6.513776771286524e-07,
      "loss": 0.4622,
      "step": 375720
    },
    {
      "epoch": 153.8,
      "grad_norm": 1.934982419013977,
      "learning_rate": 6.512679393765564e-07,
      "loss": 0.4755,
      "step": 375730
    },
    {
      "epoch": 153.8,
      "grad_norm": 4.478989124298096,
      "learning_rate": 6.511582093774587e-07,
      "loss": 0.4577,
      "step": 375740
    },
    {
      "epoch": 153.81,
      "grad_norm": 1.9088468551635742,
      "learning_rate": 6.510484871318619e-07,
      "loss": 0.4455,
      "step": 375750
    },
    {
      "epoch": 153.81,
      "grad_norm": 2.1065609455108643,
      "learning_rate": 6.509387726402685e-07,
      "loss": 0.4308,
      "step": 375760
    },
    {
      "epoch": 153.81,
      "grad_norm": 2.070605516433716,
      "learning_rate": 6.50829065903181e-07,
      "loss": 0.4561,
      "step": 375770
    },
    {
      "epoch": 153.82,
      "grad_norm": 1.9050532579421997,
      "learning_rate": 6.507193669211032e-07,
      "loss": 0.4307,
      "step": 375780
    },
    {
      "epoch": 153.82,
      "grad_norm": 2.1330931186676025,
      "learning_rate": 6.506096756945354e-07,
      "loss": 0.4257,
      "step": 375790
    },
    {
      "epoch": 153.83,
      "grad_norm": 1.8953217267990112,
      "learning_rate": 6.504999922239815e-07,
      "loss": 0.4404,
      "step": 375800
    },
    {
      "epoch": 153.83,
      "grad_norm": 1.8569833040237427,
      "learning_rate": 6.503903165099437e-07,
      "loss": 0.4727,
      "step": 375810
    },
    {
      "epoch": 153.84,
      "grad_norm": 1.8796323537826538,
      "learning_rate": 6.502806485529242e-07,
      "loss": 0.4397,
      "step": 375820
    },
    {
      "epoch": 153.84,
      "grad_norm": 2.388120412826538,
      "learning_rate": 6.501709883534256e-07,
      "loss": 0.4776,
      "step": 375830
    },
    {
      "epoch": 153.84,
      "grad_norm": 1.9432045221328735,
      "learning_rate": 6.500613359119501e-07,
      "loss": 0.4448,
      "step": 375840
    },
    {
      "epoch": 153.85,
      "grad_norm": 2.2655248641967773,
      "learning_rate": 6.49951691229e-07,
      "loss": 0.4474,
      "step": 375850
    },
    {
      "epoch": 153.85,
      "grad_norm": 2.42852783203125,
      "learning_rate": 6.49842054305078e-07,
      "loss": 0.4442,
      "step": 375860
    },
    {
      "epoch": 153.86,
      "grad_norm": 2.040731906890869,
      "learning_rate": 6.497324251406856e-07,
      "loss": 0.45,
      "step": 375870
    },
    {
      "epoch": 153.86,
      "grad_norm": 2.1153604984283447,
      "learning_rate": 6.496228037363254e-07,
      "loss": 0.4522,
      "step": 375880
    },
    {
      "epoch": 153.86,
      "grad_norm": 2.2890782356262207,
      "learning_rate": 6.495131900924998e-07,
      "loss": 0.4626,
      "step": 375890
    },
    {
      "epoch": 153.87,
      "grad_norm": 1.9229217767715454,
      "learning_rate": 6.494035842097104e-07,
      "loss": 0.4533,
      "step": 375900
    },
    {
      "epoch": 153.87,
      "grad_norm": 2.1171088218688965,
      "learning_rate": 6.492939860884588e-07,
      "loss": 0.4299,
      "step": 375910
    },
    {
      "epoch": 153.88,
      "grad_norm": 2.293984889984131,
      "learning_rate": 6.491843957292492e-07,
      "loss": 0.4255,
      "step": 375920
    },
    {
      "epoch": 153.88,
      "grad_norm": 1.9206078052520752,
      "learning_rate": 6.490748131325815e-07,
      "loss": 0.4705,
      "step": 375930
    },
    {
      "epoch": 153.88,
      "grad_norm": 2.108323097229004,
      "learning_rate": 6.489652382989585e-07,
      "loss": 0.4282,
      "step": 375940
    },
    {
      "epoch": 153.89,
      "grad_norm": 2.0701000690460205,
      "learning_rate": 6.488556712288818e-07,
      "loss": 0.4321,
      "step": 375950
    },
    {
      "epoch": 153.89,
      "grad_norm": 1.94231379032135,
      "learning_rate": 6.487461119228538e-07,
      "loss": 0.4392,
      "step": 375960
    },
    {
      "epoch": 153.9,
      "grad_norm": 2.367892265319824,
      "learning_rate": 6.486365603813759e-07,
      "loss": 0.4551,
      "step": 375970
    },
    {
      "epoch": 153.9,
      "grad_norm": 1.9731918573379517,
      "learning_rate": 6.485270166049503e-07,
      "loss": 0.4394,
      "step": 375980
    },
    {
      "epoch": 153.91,
      "grad_norm": 1.9505503177642822,
      "learning_rate": 6.484174805940785e-07,
      "loss": 0.4748,
      "step": 375990
    },
    {
      "epoch": 153.91,
      "grad_norm": 2.17404842376709,
      "learning_rate": 6.483079523492625e-07,
      "loss": 0.4434,
      "step": 376000
    },
    {
      "epoch": 153.91,
      "grad_norm": 2.0314254760742188,
      "learning_rate": 6.48198431871004e-07,
      "loss": 0.456,
      "step": 376010
    },
    {
      "epoch": 153.92,
      "grad_norm": 2.122908353805542,
      "learning_rate": 6.480889191598046e-07,
      "loss": 0.4636,
      "step": 376020
    },
    {
      "epoch": 153.92,
      "grad_norm": 2.0794286727905273,
      "learning_rate": 6.47979414216166e-07,
      "loss": 0.4492,
      "step": 376030
    },
    {
      "epoch": 153.93,
      "grad_norm": 2.1028802394866943,
      "learning_rate": 6.478699170405899e-07,
      "loss": 0.4658,
      "step": 376040
    },
    {
      "epoch": 153.93,
      "grad_norm": 1.8531877994537354,
      "learning_rate": 6.477604276335777e-07,
      "loss": 0.4373,
      "step": 376050
    },
    {
      "epoch": 153.93,
      "grad_norm": 2.11198091506958,
      "learning_rate": 6.476509459956311e-07,
      "loss": 0.4319,
      "step": 376060
    },
    {
      "epoch": 153.94,
      "grad_norm": 1.8233721256256104,
      "learning_rate": 6.475414721272516e-07,
      "loss": 0.4493,
      "step": 376070
    },
    {
      "epoch": 153.94,
      "grad_norm": 2.014138698577881,
      "learning_rate": 6.474320060289407e-07,
      "loss": 0.4455,
      "step": 376080
    },
    {
      "epoch": 153.95,
      "grad_norm": 2.1088201999664307,
      "learning_rate": 6.473225477012005e-07,
      "loss": 0.4541,
      "step": 376090
    },
    {
      "epoch": 153.95,
      "grad_norm": 1.738234043121338,
      "learning_rate": 6.472130971445303e-07,
      "loss": 0.468,
      "step": 376100
    },
    {
      "epoch": 153.95,
      "grad_norm": 1.919681191444397,
      "learning_rate": 6.471036543594335e-07,
      "loss": 0.4525,
      "step": 376110
    },
    {
      "epoch": 153.96,
      "grad_norm": 2.009317398071289,
      "learning_rate": 6.469942193464112e-07,
      "loss": 0.4539,
      "step": 376120
    },
    {
      "epoch": 153.96,
      "grad_norm": 2.3909099102020264,
      "learning_rate": 6.468847921059641e-07,
      "loss": 0.4433,
      "step": 376130
    },
    {
      "epoch": 153.97,
      "grad_norm": 1.8438411951065063,
      "learning_rate": 6.467753726385937e-07,
      "loss": 0.451,
      "step": 376140
    },
    {
      "epoch": 153.97,
      "grad_norm": 1.8299319744110107,
      "learning_rate": 6.466659609448013e-07,
      "loss": 0.4331,
      "step": 376150
    },
    {
      "epoch": 153.97,
      "grad_norm": 1.9431636333465576,
      "learning_rate": 6.465565570250881e-07,
      "loss": 0.4273,
      "step": 376160
    },
    {
      "epoch": 153.98,
      "grad_norm": 1.8976755142211914,
      "learning_rate": 6.464471608799554e-07,
      "loss": 0.4263,
      "step": 376170
    },
    {
      "epoch": 153.98,
      "grad_norm": 1.8175758123397827,
      "learning_rate": 6.463377725099037e-07,
      "loss": 0.4346,
      "step": 376180
    },
    {
      "epoch": 153.99,
      "grad_norm": 2.2953267097473145,
      "learning_rate": 6.462283919154349e-07,
      "loss": 0.4464,
      "step": 376190
    },
    {
      "epoch": 153.99,
      "grad_norm": 2.4142720699310303,
      "learning_rate": 6.461190190970495e-07,
      "loss": 0.4688,
      "step": 376200
    },
    {
      "epoch": 154.0,
      "grad_norm": 2.1126410961151123,
      "learning_rate": 6.46009654055249e-07,
      "loss": 0.4747,
      "step": 376210
    },
    {
      "epoch": 154.0,
      "grad_norm": 2.0448050498962402,
      "learning_rate": 6.459002967905339e-07,
      "loss": 0.4527,
      "step": 376220
    },
    {
      "epoch": 154.0,
      "eval_loss": 0.4491780400276184,
      "eval_runtime": 52.8986,
      "eval_samples_per_second": 65.2,
      "eval_steps_per_second": 8.167,
      "step": 376222
    },
    {
      "epoch": 154.0,
      "grad_norm": 1.9025347232818604,
      "learning_rate": 6.457909473034059e-07,
      "loss": 0.4465,
      "step": 376230
    },
    {
      "epoch": 154.01,
      "grad_norm": 2.4409425258636475,
      "learning_rate": 6.456816055943648e-07,
      "loss": 0.452,
      "step": 376240
    },
    {
      "epoch": 154.01,
      "grad_norm": 2.2582430839538574,
      "learning_rate": 6.455722716639122e-07,
      "loss": 0.4501,
      "step": 376250
    },
    {
      "epoch": 154.02,
      "grad_norm": 2.2325479984283447,
      "learning_rate": 6.454629455125485e-07,
      "loss": 0.4165,
      "step": 376260
    },
    {
      "epoch": 154.02,
      "grad_norm": 1.9003336429595947,
      "learning_rate": 6.453536271407749e-07,
      "loss": 0.4357,
      "step": 376270
    },
    {
      "epoch": 154.02,
      "grad_norm": 1.9433332681655884,
      "learning_rate": 6.45244316549092e-07,
      "loss": 0.4456,
      "step": 376280
    },
    {
      "epoch": 154.03,
      "grad_norm": 2.104677200317383,
      "learning_rate": 6.451350137379999e-07,
      "loss": 0.45,
      "step": 376290
    },
    {
      "epoch": 154.03,
      "grad_norm": 1.9800498485565186,
      "learning_rate": 6.450257187080007e-07,
      "loss": 0.433,
      "step": 376300
    },
    {
      "epoch": 154.04,
      "grad_norm": 2.43532395362854,
      "learning_rate": 6.449164314595945e-07,
      "loss": 0.4555,
      "step": 376310
    },
    {
      "epoch": 154.04,
      "grad_norm": 2.194171667098999,
      "learning_rate": 6.448071519932815e-07,
      "loss": 0.4585,
      "step": 376320
    },
    {
      "epoch": 154.04,
      "grad_norm": 2.1159746646881104,
      "learning_rate": 6.446978803095629e-07,
      "loss": 0.4497,
      "step": 376330
    },
    {
      "epoch": 154.05,
      "grad_norm": 2.0916085243225098,
      "learning_rate": 6.445886164089388e-07,
      "loss": 0.4731,
      "step": 376340
    },
    {
      "epoch": 154.05,
      "grad_norm": 2.2081050872802734,
      "learning_rate": 6.444793602919098e-07,
      "loss": 0.4444,
      "step": 376350
    },
    {
      "epoch": 154.06,
      "grad_norm": 1.7707093954086304,
      "learning_rate": 6.443701119589771e-07,
      "loss": 0.4366,
      "step": 376360
    },
    {
      "epoch": 154.06,
      "grad_norm": 2.130308151245117,
      "learning_rate": 6.442608714106399e-07,
      "loss": 0.4591,
      "step": 376370
    },
    {
      "epoch": 154.06,
      "grad_norm": 1.8272651433944702,
      "learning_rate": 6.441516386473991e-07,
      "loss": 0.4604,
      "step": 376380
    },
    {
      "epoch": 154.07,
      "grad_norm": 2.5044734477996826,
      "learning_rate": 6.440424136697551e-07,
      "loss": 0.422,
      "step": 376390
    },
    {
      "epoch": 154.07,
      "grad_norm": 2.0169501304626465,
      "learning_rate": 6.439331964782085e-07,
      "loss": 0.4547,
      "step": 376400
    },
    {
      "epoch": 154.08,
      "grad_norm": 2.444256067276001,
      "learning_rate": 6.438239870732592e-07,
      "loss": 0.4392,
      "step": 376410
    },
    {
      "epoch": 154.08,
      "grad_norm": 2.4315590858459473,
      "learning_rate": 6.43714785455408e-07,
      "loss": 0.442,
      "step": 376420
    },
    {
      "epoch": 154.09,
      "grad_norm": 2.0110983848571777,
      "learning_rate": 6.436055916251548e-07,
      "loss": 0.4521,
      "step": 376430
    },
    {
      "epoch": 154.09,
      "grad_norm": 1.937748670578003,
      "learning_rate": 6.434964055829995e-07,
      "loss": 0.4325,
      "step": 376440
    },
    {
      "epoch": 154.09,
      "grad_norm": 2.077460289001465,
      "learning_rate": 6.43387227329443e-07,
      "loss": 0.441,
      "step": 376450
    },
    {
      "epoch": 154.1,
      "grad_norm": 1.9629192352294922,
      "learning_rate": 6.432780568649849e-07,
      "loss": 0.4626,
      "step": 376460
    },
    {
      "epoch": 154.1,
      "grad_norm": 2.0809690952301025,
      "learning_rate": 6.431688941901252e-07,
      "loss": 0.445,
      "step": 376470
    },
    {
      "epoch": 154.11,
      "grad_norm": 1.9397963285446167,
      "learning_rate": 6.430597393053635e-07,
      "loss": 0.4399,
      "step": 376480
    },
    {
      "epoch": 154.11,
      "grad_norm": 2.0531818866729736,
      "learning_rate": 6.429505922112016e-07,
      "loss": 0.4622,
      "step": 376490
    },
    {
      "epoch": 154.11,
      "grad_norm": 1.9904859066009521,
      "learning_rate": 6.428414529081389e-07,
      "loss": 0.4515,
      "step": 376500
    },
    {
      "epoch": 154.12,
      "grad_norm": 2.4560279846191406,
      "learning_rate": 6.42732321396674e-07,
      "loss": 0.4579,
      "step": 376510
    },
    {
      "epoch": 154.12,
      "grad_norm": 1.8041889667510986,
      "learning_rate": 6.426231976773076e-07,
      "loss": 0.4411,
      "step": 376520
    },
    {
      "epoch": 154.13,
      "grad_norm": 2.0318315029144287,
      "learning_rate": 6.425140817505398e-07,
      "loss": 0.4385,
      "step": 376530
    },
    {
      "epoch": 154.13,
      "grad_norm": 2.3383169174194336,
      "learning_rate": 6.424049736168702e-07,
      "loss": 0.4323,
      "step": 376540
    },
    {
      "epoch": 154.13,
      "grad_norm": 2.0938222408294678,
      "learning_rate": 6.422958732767987e-07,
      "loss": 0.4313,
      "step": 376550
    },
    {
      "epoch": 154.14,
      "grad_norm": 2.1370368003845215,
      "learning_rate": 6.421867807308251e-07,
      "loss": 0.4539,
      "step": 376560
    },
    {
      "epoch": 154.14,
      "grad_norm": 1.8814797401428223,
      "learning_rate": 6.420776959794489e-07,
      "loss": 0.4452,
      "step": 376570
    },
    {
      "epoch": 154.15,
      "grad_norm": 2.9389865398406982,
      "learning_rate": 6.419686190231702e-07,
      "loss": 0.4492,
      "step": 376580
    },
    {
      "epoch": 154.15,
      "grad_norm": 2.368624210357666,
      "learning_rate": 6.418595498624885e-07,
      "loss": 0.4446,
      "step": 376590
    },
    {
      "epoch": 154.15,
      "grad_norm": 2.1899495124816895,
      "learning_rate": 6.41750488497903e-07,
      "loss": 0.4614,
      "step": 376600
    },
    {
      "epoch": 154.16,
      "grad_norm": 2.1017119884490967,
      "learning_rate": 6.41641434929914e-07,
      "loss": 0.4453,
      "step": 376610
    },
    {
      "epoch": 154.16,
      "grad_norm": 2.471064329147339,
      "learning_rate": 6.415323891590206e-07,
      "loss": 0.4403,
      "step": 376620
    },
    {
      "epoch": 154.17,
      "grad_norm": 1.8569239377975464,
      "learning_rate": 6.414233511857226e-07,
      "loss": 0.4441,
      "step": 376630
    },
    {
      "epoch": 154.17,
      "grad_norm": 2.0986831188201904,
      "learning_rate": 6.413143210105192e-07,
      "loss": 0.4469,
      "step": 376640
    },
    {
      "epoch": 154.18,
      "grad_norm": 1.927160382270813,
      "learning_rate": 6.412052986339102e-07,
      "loss": 0.423,
      "step": 376650
    },
    {
      "epoch": 154.18,
      "grad_norm": 1.9334032535552979,
      "learning_rate": 6.410962840563946e-07,
      "loss": 0.4469,
      "step": 376660
    },
    {
      "epoch": 154.18,
      "grad_norm": 1.8461862802505493,
      "learning_rate": 6.409872772784719e-07,
      "loss": 0.4482,
      "step": 376670
    },
    {
      "epoch": 154.19,
      "grad_norm": 2.0316107273101807,
      "learning_rate": 6.408782783006419e-07,
      "loss": 0.4391,
      "step": 376680
    },
    {
      "epoch": 154.19,
      "grad_norm": 2.124256134033203,
      "learning_rate": 6.407692871234032e-07,
      "loss": 0.4498,
      "step": 376690
    },
    {
      "epoch": 154.2,
      "grad_norm": 2.3609373569488525,
      "learning_rate": 6.406603037472554e-07,
      "loss": 0.4444,
      "step": 376700
    },
    {
      "epoch": 154.2,
      "grad_norm": 2.133486747741699,
      "learning_rate": 6.40551328172698e-07,
      "loss": 0.4379,
      "step": 376710
    },
    {
      "epoch": 154.2,
      "grad_norm": 1.8444602489471436,
      "learning_rate": 6.404423604002297e-07,
      "loss": 0.4434,
      "step": 376720
    },
    {
      "epoch": 154.21,
      "grad_norm": 2.0767102241516113,
      "learning_rate": 6.403334004303499e-07,
      "loss": 0.4479,
      "step": 376730
    },
    {
      "epoch": 154.21,
      "grad_norm": 1.7688885927200317,
      "learning_rate": 6.40224448263558e-07,
      "loss": 0.4509,
      "step": 376740
    },
    {
      "epoch": 154.22,
      "grad_norm": 2.0052270889282227,
      "learning_rate": 6.401155039003525e-07,
      "loss": 0.4461,
      "step": 376750
    },
    {
      "epoch": 154.22,
      "grad_norm": 2.4340109825134277,
      "learning_rate": 6.400065673412331e-07,
      "loss": 0.4592,
      "step": 376760
    },
    {
      "epoch": 154.22,
      "grad_norm": 2.4714157581329346,
      "learning_rate": 6.398976385866981e-07,
      "loss": 0.4458,
      "step": 376770
    },
    {
      "epoch": 154.23,
      "grad_norm": 2.0593199729919434,
      "learning_rate": 6.397887176372472e-07,
      "loss": 0.4435,
      "step": 376780
    },
    {
      "epoch": 154.23,
      "grad_norm": 1.9528673887252808,
      "learning_rate": 6.39679804493379e-07,
      "loss": 0.423,
      "step": 376790
    },
    {
      "epoch": 154.24,
      "grad_norm": 1.9701682329177856,
      "learning_rate": 6.395708991555926e-07,
      "loss": 0.4398,
      "step": 376800
    },
    {
      "epoch": 154.24,
      "grad_norm": 2.03122615814209,
      "learning_rate": 6.394620016243871e-07,
      "loss": 0.4347,
      "step": 376810
    },
    {
      "epoch": 154.24,
      "grad_norm": 2.109945058822632,
      "learning_rate": 6.393531119002606e-07,
      "loss": 0.4552,
      "step": 376820
    },
    {
      "epoch": 154.25,
      "grad_norm": 1.8696061372756958,
      "learning_rate": 6.392442299837121e-07,
      "loss": 0.435,
      "step": 376830
    },
    {
      "epoch": 154.25,
      "grad_norm": 1.5906533002853394,
      "learning_rate": 6.391353558752406e-07,
      "loss": 0.4462,
      "step": 376840
    },
    {
      "epoch": 154.26,
      "grad_norm": 1.9288897514343262,
      "learning_rate": 6.390264895753441e-07,
      "loss": 0.4472,
      "step": 376850
    },
    {
      "epoch": 154.26,
      "grad_norm": 2.1050689220428467,
      "learning_rate": 6.389176310845226e-07,
      "loss": 0.4723,
      "step": 376860
    },
    {
      "epoch": 154.27,
      "grad_norm": 1.9607291221618652,
      "learning_rate": 6.388087804032743e-07,
      "loss": 0.4433,
      "step": 376870
    },
    {
      "epoch": 154.27,
      "grad_norm": 2.234999656677246,
      "learning_rate": 6.386999375320976e-07,
      "loss": 0.4191,
      "step": 376880
    },
    {
      "epoch": 154.27,
      "grad_norm": 2.097727060317993,
      "learning_rate": 6.385911024714912e-07,
      "loss": 0.4416,
      "step": 376890
    },
    {
      "epoch": 154.28,
      "grad_norm": 2.0045807361602783,
      "learning_rate": 6.384822752219537e-07,
      "loss": 0.4542,
      "step": 376900
    },
    {
      "epoch": 154.28,
      "grad_norm": 2.371396064758301,
      "learning_rate": 6.383734557839835e-07,
      "loss": 0.4355,
      "step": 376910
    },
    {
      "epoch": 154.29,
      "grad_norm": 2.4137892723083496,
      "learning_rate": 6.382646441580791e-07,
      "loss": 0.4387,
      "step": 376920
    },
    {
      "epoch": 154.29,
      "grad_norm": 2.2913496494293213,
      "learning_rate": 6.381558403447388e-07,
      "loss": 0.4412,
      "step": 376930
    },
    {
      "epoch": 154.29,
      "grad_norm": 2.2621283531188965,
      "learning_rate": 6.380470443444616e-07,
      "loss": 0.4384,
      "step": 376940
    },
    {
      "epoch": 154.3,
      "grad_norm": 2.0072457790374756,
      "learning_rate": 6.379382561577458e-07,
      "loss": 0.4414,
      "step": 376950
    },
    {
      "epoch": 154.3,
      "grad_norm": 2.1704163551330566,
      "learning_rate": 6.37829475785089e-07,
      "loss": 0.4784,
      "step": 376960
    },
    {
      "epoch": 154.31,
      "grad_norm": 1.8828125,
      "learning_rate": 6.377207032269899e-07,
      "loss": 0.4733,
      "step": 376970
    },
    {
      "epoch": 154.31,
      "grad_norm": 1.9943293333053589,
      "learning_rate": 6.376119384839467e-07,
      "loss": 0.4732,
      "step": 376980
    },
    {
      "epoch": 154.31,
      "grad_norm": 1.7963558435440063,
      "learning_rate": 6.375031815564579e-07,
      "loss": 0.4451,
      "step": 376990
    },
    {
      "epoch": 154.32,
      "grad_norm": 2.26934552192688,
      "learning_rate": 6.373944324450213e-07,
      "loss": 0.4538,
      "step": 377000
    },
    {
      "epoch": 154.32,
      "grad_norm": 1.9236096143722534,
      "learning_rate": 6.372856911501353e-07,
      "loss": 0.4487,
      "step": 377010
    },
    {
      "epoch": 154.33,
      "grad_norm": 1.8156908750534058,
      "learning_rate": 6.371769576722981e-07,
      "loss": 0.4265,
      "step": 377020
    },
    {
      "epoch": 154.33,
      "grad_norm": 1.7036404609680176,
      "learning_rate": 6.370682320120077e-07,
      "loss": 0.436,
      "step": 377030
    },
    {
      "epoch": 154.33,
      "grad_norm": 1.7376086711883545,
      "learning_rate": 6.369595141697615e-07,
      "loss": 0.44,
      "step": 377040
    },
    {
      "epoch": 154.34,
      "grad_norm": 2.064727306365967,
      "learning_rate": 6.368508041460588e-07,
      "loss": 0.4474,
      "step": 377050
    },
    {
      "epoch": 154.34,
      "grad_norm": 2.1033918857574463,
      "learning_rate": 6.367421019413971e-07,
      "loss": 0.4419,
      "step": 377060
    },
    {
      "epoch": 154.35,
      "grad_norm": 1.8261537551879883,
      "learning_rate": 6.366334075562741e-07,
      "loss": 0.4483,
      "step": 377070
    },
    {
      "epoch": 154.35,
      "grad_norm": 2.237337112426758,
      "learning_rate": 6.365247209911887e-07,
      "loss": 0.4666,
      "step": 377080
    },
    {
      "epoch": 154.36,
      "grad_norm": 1.9589170217514038,
      "learning_rate": 6.364160422466371e-07,
      "loss": 0.449,
      "step": 377090
    },
    {
      "epoch": 154.36,
      "grad_norm": 2.4788358211517334,
      "learning_rate": 6.363073713231179e-07,
      "loss": 0.4376,
      "step": 377100
    },
    {
      "epoch": 154.36,
      "grad_norm": 1.9066832065582275,
      "learning_rate": 6.361987082211291e-07,
      "loss": 0.4503,
      "step": 377110
    },
    {
      "epoch": 154.37,
      "grad_norm": 1.6999750137329102,
      "learning_rate": 6.360900529411681e-07,
      "loss": 0.443,
      "step": 377120
    },
    {
      "epoch": 154.37,
      "grad_norm": 2.2008354663848877,
      "learning_rate": 6.359814054837329e-07,
      "loss": 0.4769,
      "step": 377130
    },
    {
      "epoch": 154.38,
      "grad_norm": 1.9432770013809204,
      "learning_rate": 6.358727658493212e-07,
      "loss": 0.4566,
      "step": 377140
    },
    {
      "epoch": 154.38,
      "grad_norm": 1.8263908624649048,
      "learning_rate": 6.357641340384306e-07,
      "loss": 0.4588,
      "step": 377150
    },
    {
      "epoch": 154.38,
      "grad_norm": 1.7904826402664185,
      "learning_rate": 6.356555100515588e-07,
      "loss": 0.4479,
      "step": 377160
    },
    {
      "epoch": 154.39,
      "grad_norm": 2.071479082107544,
      "learning_rate": 6.355468938892032e-07,
      "loss": 0.4503,
      "step": 377170
    },
    {
      "epoch": 154.39,
      "grad_norm": 1.9703549146652222,
      "learning_rate": 6.354382855518616e-07,
      "loss": 0.4574,
      "step": 377180
    },
    {
      "epoch": 154.4,
      "grad_norm": 2.0646955966949463,
      "learning_rate": 6.353296850400314e-07,
      "loss": 0.4284,
      "step": 377190
    },
    {
      "epoch": 154.4,
      "grad_norm": 2.0906293392181396,
      "learning_rate": 6.352210923542099e-07,
      "loss": 0.4334,
      "step": 377200
    },
    {
      "epoch": 154.4,
      "grad_norm": 2.3829092979431152,
      "learning_rate": 6.351125074948949e-07,
      "loss": 0.4373,
      "step": 377210
    },
    {
      "epoch": 154.41,
      "grad_norm": 2.0522613525390625,
      "learning_rate": 6.350039304625838e-07,
      "loss": 0.4353,
      "step": 377220
    },
    {
      "epoch": 154.41,
      "grad_norm": 1.9404385089874268,
      "learning_rate": 6.348953612577736e-07,
      "loss": 0.4555,
      "step": 377230
    },
    {
      "epoch": 154.42,
      "grad_norm": 1.9555200338363647,
      "learning_rate": 6.347867998809617e-07,
      "loss": 0.4476,
      "step": 377240
    },
    {
      "epoch": 154.42,
      "grad_norm": 1.9813350439071655,
      "learning_rate": 6.346782463326458e-07,
      "loss": 0.4347,
      "step": 377250
    },
    {
      "epoch": 154.42,
      "grad_norm": 2.2222886085510254,
      "learning_rate": 6.345697006133228e-07,
      "loss": 0.4571,
      "step": 377260
    },
    {
      "epoch": 154.43,
      "grad_norm": 2.071031093597412,
      "learning_rate": 6.344611627234902e-07,
      "loss": 0.4462,
      "step": 377270
    },
    {
      "epoch": 154.43,
      "grad_norm": 2.0189242362976074,
      "learning_rate": 6.343526326636448e-07,
      "loss": 0.4319,
      "step": 377280
    },
    {
      "epoch": 154.44,
      "grad_norm": 4.2561869621276855,
      "learning_rate": 6.342441104342842e-07,
      "loss": 0.461,
      "step": 377290
    },
    {
      "epoch": 154.44,
      "grad_norm": 2.0815205574035645,
      "learning_rate": 6.341355960359052e-07,
      "loss": 0.4448,
      "step": 377300
    },
    {
      "epoch": 154.45,
      "grad_norm": 2.1997053623199463,
      "learning_rate": 6.340270894690052e-07,
      "loss": 0.4406,
      "step": 377310
    },
    {
      "epoch": 154.45,
      "grad_norm": 2.586250066757202,
      "learning_rate": 6.339185907340811e-07,
      "loss": 0.4318,
      "step": 377320
    },
    {
      "epoch": 154.45,
      "grad_norm": 1.9051246643066406,
      "learning_rate": 6.338100998316297e-07,
      "loss": 0.434,
      "step": 377330
    },
    {
      "epoch": 154.46,
      "grad_norm": 2.0592143535614014,
      "learning_rate": 6.337016167621481e-07,
      "loss": 0.4579,
      "step": 377340
    },
    {
      "epoch": 154.46,
      "grad_norm": 1.8095109462738037,
      "learning_rate": 6.335931415261335e-07,
      "loss": 0.442,
      "step": 377350
    },
    {
      "epoch": 154.47,
      "grad_norm": 1.8586393594741821,
      "learning_rate": 6.334846741240825e-07,
      "loss": 0.456,
      "step": 377360
    },
    {
      "epoch": 154.47,
      "grad_norm": 2.0260090827941895,
      "learning_rate": 6.333762145564921e-07,
      "loss": 0.4586,
      "step": 377370
    },
    {
      "epoch": 154.47,
      "grad_norm": 1.7055096626281738,
      "learning_rate": 6.332677628238592e-07,
      "loss": 0.4597,
      "step": 377380
    },
    {
      "epoch": 154.48,
      "grad_norm": 2.138355255126953,
      "learning_rate": 6.331593189266812e-07,
      "loss": 0.4513,
      "step": 377390
    },
    {
      "epoch": 154.48,
      "grad_norm": 1.6203035116195679,
      "learning_rate": 6.330508828654535e-07,
      "loss": 0.4562,
      "step": 377400
    },
    {
      "epoch": 154.49,
      "grad_norm": 2.511889934539795,
      "learning_rate": 6.329424546406735e-07,
      "loss": 0.4528,
      "step": 377410
    },
    {
      "epoch": 154.49,
      "grad_norm": 1.8553893566131592,
      "learning_rate": 6.328340342528371e-07,
      "loss": 0.4408,
      "step": 377420
    },
    {
      "epoch": 154.49,
      "grad_norm": 2.2386531829833984,
      "learning_rate": 6.327256217024426e-07,
      "loss": 0.4593,
      "step": 377430
    },
    {
      "epoch": 154.5,
      "grad_norm": 1.8246402740478516,
      "learning_rate": 6.326172169899857e-07,
      "loss": 0.4553,
      "step": 377440
    },
    {
      "epoch": 154.5,
      "grad_norm": 1.9079692363739014,
      "learning_rate": 6.325088201159634e-07,
      "loss": 0.4615,
      "step": 377450
    },
    {
      "epoch": 154.51,
      "grad_norm": 2.264359951019287,
      "learning_rate": 6.324004310808716e-07,
      "loss": 0.4353,
      "step": 377460
    },
    {
      "epoch": 154.51,
      "grad_norm": 2.445155620574951,
      "learning_rate": 6.322920498852072e-07,
      "loss": 0.4525,
      "step": 377470
    },
    {
      "epoch": 154.51,
      "grad_norm": 1.8311179876327515,
      "learning_rate": 6.321836765294667e-07,
      "loss": 0.458,
      "step": 377480
    },
    {
      "epoch": 154.52,
      "grad_norm": 1.8617645502090454,
      "learning_rate": 6.320753110141464e-07,
      "loss": 0.4497,
      "step": 377490
    },
    {
      "epoch": 154.52,
      "grad_norm": 2.1153883934020996,
      "learning_rate": 6.319669533397426e-07,
      "loss": 0.4332,
      "step": 377500
    },
    {
      "epoch": 154.53,
      "grad_norm": 2.023374080657959,
      "learning_rate": 6.318586035067521e-07,
      "loss": 0.4394,
      "step": 377510
    },
    {
      "epoch": 154.53,
      "grad_norm": 2.059767723083496,
      "learning_rate": 6.317502615156709e-07,
      "loss": 0.4536,
      "step": 377520
    },
    {
      "epoch": 154.54,
      "grad_norm": 2.05444598197937,
      "learning_rate": 6.316419273669958e-07,
      "loss": 0.4676,
      "step": 377530
    },
    {
      "epoch": 154.54,
      "grad_norm": 2.1673736572265625,
      "learning_rate": 6.315336010612222e-07,
      "loss": 0.4483,
      "step": 377540
    },
    {
      "epoch": 154.54,
      "grad_norm": 1.976122498512268,
      "learning_rate": 6.314252825988468e-07,
      "loss": 0.4556,
      "step": 377550
    },
    {
      "epoch": 154.55,
      "grad_norm": 2.0104501247406006,
      "learning_rate": 6.313169719803655e-07,
      "loss": 0.4325,
      "step": 377560
    },
    {
      "epoch": 154.55,
      "grad_norm": 1.915125846862793,
      "learning_rate": 6.31208669206275e-07,
      "loss": 0.4471,
      "step": 377570
    },
    {
      "epoch": 154.56,
      "grad_norm": 3.0472891330718994,
      "learning_rate": 6.311003742770708e-07,
      "loss": 0.4445,
      "step": 377580
    },
    {
      "epoch": 154.56,
      "grad_norm": 2.0330557823181152,
      "learning_rate": 6.309920871932496e-07,
      "loss": 0.4455,
      "step": 377590
    },
    {
      "epoch": 154.56,
      "grad_norm": 2.0595004558563232,
      "learning_rate": 6.30883807955307e-07,
      "loss": 0.4367,
      "step": 377600
    },
    {
      "epoch": 154.57,
      "grad_norm": 1.8496953248977661,
      "learning_rate": 6.307755365637384e-07,
      "loss": 0.4518,
      "step": 377610
    },
    {
      "epoch": 154.57,
      "grad_norm": 2.0696749687194824,
      "learning_rate": 6.306672730190412e-07,
      "loss": 0.4676,
      "step": 377620
    },
    {
      "epoch": 154.58,
      "grad_norm": 2.209712028503418,
      "learning_rate": 6.305590173217109e-07,
      "loss": 0.4563,
      "step": 377630
    },
    {
      "epoch": 154.58,
      "grad_norm": 1.8735198974609375,
      "learning_rate": 6.30450769472243e-07,
      "loss": 0.4452,
      "step": 377640
    },
    {
      "epoch": 154.58,
      "grad_norm": 2.409764289855957,
      "learning_rate": 6.303425294711334e-07,
      "loss": 0.4323,
      "step": 377650
    },
    {
      "epoch": 154.59,
      "grad_norm": 1.9508782625198364,
      "learning_rate": 6.302342973188791e-07,
      "loss": 0.447,
      "step": 377660
    },
    {
      "epoch": 154.59,
      "grad_norm": 1.7517884969711304,
      "learning_rate": 6.30126073015974e-07,
      "loss": 0.459,
      "step": 377670
    },
    {
      "epoch": 154.6,
      "grad_norm": 2.010547399520874,
      "learning_rate": 6.300178565629148e-07,
      "loss": 0.4426,
      "step": 377680
    },
    {
      "epoch": 154.6,
      "grad_norm": 1.7141941785812378,
      "learning_rate": 6.29909647960197e-07,
      "loss": 0.4458,
      "step": 377690
    },
    {
      "epoch": 154.6,
      "grad_norm": 1.9877647161483765,
      "learning_rate": 6.298014472083164e-07,
      "loss": 0.459,
      "step": 377700
    },
    {
      "epoch": 154.61,
      "grad_norm": 2.010309934616089,
      "learning_rate": 6.296932543077689e-07,
      "loss": 0.4413,
      "step": 377710
    },
    {
      "epoch": 154.61,
      "grad_norm": 2.3034660816192627,
      "learning_rate": 6.295850692590497e-07,
      "loss": 0.4428,
      "step": 377720
    },
    {
      "epoch": 154.62,
      "grad_norm": 2.1442039012908936,
      "learning_rate": 6.294768920626547e-07,
      "loss": 0.4185,
      "step": 377730
    },
    {
      "epoch": 154.62,
      "grad_norm": 2.126910448074341,
      "learning_rate": 6.293687227190793e-07,
      "loss": 0.4651,
      "step": 377740
    },
    {
      "epoch": 154.63,
      "grad_norm": 2.2490952014923096,
      "learning_rate": 6.292605612288189e-07,
      "loss": 0.4505,
      "step": 377750
    },
    {
      "epoch": 154.63,
      "grad_norm": 1.8857991695404053,
      "learning_rate": 6.291524075923691e-07,
      "loss": 0.4535,
      "step": 377760
    },
    {
      "epoch": 154.63,
      "grad_norm": 1.829542636871338,
      "learning_rate": 6.290442618102255e-07,
      "loss": 0.459,
      "step": 377770
    },
    {
      "epoch": 154.64,
      "grad_norm": 2.1185662746429443,
      "learning_rate": 6.289361238828832e-07,
      "loss": 0.4613,
      "step": 377780
    },
    {
      "epoch": 154.64,
      "grad_norm": 2.4054715633392334,
      "learning_rate": 6.288279938108378e-07,
      "loss": 0.4584,
      "step": 377790
    },
    {
      "epoch": 154.65,
      "grad_norm": 2.0229897499084473,
      "learning_rate": 6.287198715945844e-07,
      "loss": 0.4579,
      "step": 377800
    },
    {
      "epoch": 154.65,
      "grad_norm": 1.9577980041503906,
      "learning_rate": 6.286117572346185e-07,
      "loss": 0.4472,
      "step": 377810
    },
    {
      "epoch": 154.65,
      "grad_norm": 1.527147889137268,
      "learning_rate": 6.285036507314352e-07,
      "loss": 0.4334,
      "step": 377820
    },
    {
      "epoch": 154.66,
      "grad_norm": 1.8032712936401367,
      "learning_rate": 6.283955520855298e-07,
      "loss": 0.4481,
      "step": 377830
    },
    {
      "epoch": 154.66,
      "grad_norm": 1.7680307626724243,
      "learning_rate": 6.282874612973977e-07,
      "loss": 0.444,
      "step": 377840
    },
    {
      "epoch": 154.67,
      "grad_norm": 1.7728124856948853,
      "learning_rate": 6.281793783675335e-07,
      "loss": 0.462,
      "step": 377850
    },
    {
      "epoch": 154.67,
      "grad_norm": 2.0337860584259033,
      "learning_rate": 6.280713032964328e-07,
      "loss": 0.4564,
      "step": 377860
    },
    {
      "epoch": 154.67,
      "grad_norm": 1.8782776594161987,
      "learning_rate": 6.279632360845903e-07,
      "loss": 0.4528,
      "step": 377870
    },
    {
      "epoch": 154.68,
      "grad_norm": 2.0432701110839844,
      "learning_rate": 6.278551767325016e-07,
      "loss": 0.4437,
      "step": 377880
    },
    {
      "epoch": 154.68,
      "grad_norm": 1.7599749565124512,
      "learning_rate": 6.27747125240661e-07,
      "loss": 0.4553,
      "step": 377890
    },
    {
      "epoch": 154.69,
      "grad_norm": 1.7571507692337036,
      "learning_rate": 6.276390816095639e-07,
      "loss": 0.444,
      "step": 377900
    },
    {
      "epoch": 154.69,
      "grad_norm": 1.9145554304122925,
      "learning_rate": 6.275310458397054e-07,
      "loss": 0.4447,
      "step": 377910
    },
    {
      "epoch": 154.7,
      "grad_norm": 1.9514870643615723,
      "learning_rate": 6.274230179315799e-07,
      "loss": 0.4651,
      "step": 377920
    },
    {
      "epoch": 154.7,
      "grad_norm": 1.8309900760650635,
      "learning_rate": 6.273149978856825e-07,
      "loss": 0.4601,
      "step": 377930
    },
    {
      "epoch": 154.7,
      "grad_norm": 2.159111738204956,
      "learning_rate": 6.272069857025081e-07,
      "loss": 0.4498,
      "step": 377940
    },
    {
      "epoch": 154.71,
      "grad_norm": 2.01596999168396,
      "learning_rate": 6.270989813825513e-07,
      "loss": 0.4216,
      "step": 377950
    },
    {
      "epoch": 154.71,
      "grad_norm": 2.8649933338165283,
      "learning_rate": 6.269909849263069e-07,
      "loss": 0.4326,
      "step": 377960
    },
    {
      "epoch": 154.72,
      "grad_norm": 2.064286470413208,
      "learning_rate": 6.268829963342707e-07,
      "loss": 0.4719,
      "step": 377970
    },
    {
      "epoch": 154.72,
      "grad_norm": 1.8661036491394043,
      "learning_rate": 6.267750156069355e-07,
      "loss": 0.4186,
      "step": 377980
    },
    {
      "epoch": 154.72,
      "grad_norm": 2.1062331199645996,
      "learning_rate": 6.266670427447958e-07,
      "loss": 0.4709,
      "step": 377990
    },
    {
      "epoch": 154.73,
      "grad_norm": 1.8943333625793457,
      "learning_rate": 6.265590777483483e-07,
      "loss": 0.4392,
      "step": 378000
    },
    {
      "epoch": 154.73,
      "grad_norm": 1.9938788414001465,
      "learning_rate": 6.264511206180864e-07,
      "loss": 0.4745,
      "step": 378010
    },
    {
      "epoch": 154.74,
      "grad_norm": 1.8260526657104492,
      "learning_rate": 6.263431713545044e-07,
      "loss": 0.4622,
      "step": 378020
    },
    {
      "epoch": 154.74,
      "grad_norm": 2.071640968322754,
      "learning_rate": 6.262352299580974e-07,
      "loss": 0.4231,
      "step": 378030
    },
    {
      "epoch": 154.74,
      "grad_norm": 1.9127435684204102,
      "learning_rate": 6.261272964293596e-07,
      "loss": 0.445,
      "step": 378040
    },
    {
      "epoch": 154.75,
      "grad_norm": 1.933678150177002,
      "learning_rate": 6.260193707687852e-07,
      "loss": 0.4671,
      "step": 378050
    },
    {
      "epoch": 154.75,
      "grad_norm": 2.256911039352417,
      "learning_rate": 6.259114529768689e-07,
      "loss": 0.4411,
      "step": 378060
    },
    {
      "epoch": 154.76,
      "grad_norm": 2.0089707374572754,
      "learning_rate": 6.25803543054105e-07,
      "loss": 0.4444,
      "step": 378070
    },
    {
      "epoch": 154.76,
      "grad_norm": 2.1419382095336914,
      "learning_rate": 6.256956410009878e-07,
      "loss": 0.4472,
      "step": 378080
    },
    {
      "epoch": 154.76,
      "grad_norm": 2.4061906337738037,
      "learning_rate": 6.255877468180113e-07,
      "loss": 0.4443,
      "step": 378090
    },
    {
      "epoch": 154.77,
      "grad_norm": 2.021137237548828,
      "learning_rate": 6.254798605056702e-07,
      "loss": 0.4359,
      "step": 378100
    },
    {
      "epoch": 154.77,
      "grad_norm": 1.9517782926559448,
      "learning_rate": 6.253719820644591e-07,
      "loss": 0.4515,
      "step": 378110
    },
    {
      "epoch": 154.78,
      "grad_norm": 2.2103445529937744,
      "learning_rate": 6.25264111494871e-07,
      "loss": 0.4537,
      "step": 378120
    },
    {
      "epoch": 154.78,
      "grad_norm": 2.073288917541504,
      "learning_rate": 6.251562487974006e-07,
      "loss": 0.4457,
      "step": 378130
    },
    {
      "epoch": 154.79,
      "grad_norm": 1.949010968208313,
      "learning_rate": 6.250483939725419e-07,
      "loss": 0.4594,
      "step": 378140
    },
    {
      "epoch": 154.79,
      "grad_norm": 1.7314817905426025,
      "learning_rate": 6.24940547020789e-07,
      "loss": 0.449,
      "step": 378150
    },
    {
      "epoch": 154.79,
      "grad_norm": 2.614475965499878,
      "learning_rate": 6.248327079426363e-07,
      "loss": 0.4487,
      "step": 378160
    },
    {
      "epoch": 154.8,
      "grad_norm": 2.148922920227051,
      "learning_rate": 6.247248767385764e-07,
      "loss": 0.4435,
      "step": 378170
    },
    {
      "epoch": 154.8,
      "grad_norm": 2.1836860179901123,
      "learning_rate": 6.246170534091055e-07,
      "loss": 0.4585,
      "step": 378180
    },
    {
      "epoch": 154.81,
      "grad_norm": 1.8585429191589355,
      "learning_rate": 6.245092379547162e-07,
      "loss": 0.4688,
      "step": 378190
    },
    {
      "epoch": 154.81,
      "grad_norm": 2.21026873588562,
      "learning_rate": 6.244014303759026e-07,
      "loss": 0.457,
      "step": 378200
    },
    {
      "epoch": 154.81,
      "grad_norm": 2.255547523498535,
      "learning_rate": 6.242936306731583e-07,
      "loss": 0.4188,
      "step": 378210
    },
    {
      "epoch": 154.82,
      "grad_norm": 4.388155460357666,
      "learning_rate": 6.241858388469775e-07,
      "loss": 0.4533,
      "step": 378220
    },
    {
      "epoch": 154.82,
      "grad_norm": 2.0987772941589355,
      "learning_rate": 6.240780548978541e-07,
      "loss": 0.4267,
      "step": 378230
    },
    {
      "epoch": 154.83,
      "grad_norm": 2.0748708248138428,
      "learning_rate": 6.239702788262813e-07,
      "loss": 0.4432,
      "step": 378240
    },
    {
      "epoch": 154.83,
      "grad_norm": 1.8634743690490723,
      "learning_rate": 6.238625106327535e-07,
      "loss": 0.4564,
      "step": 378250
    },
    {
      "epoch": 154.83,
      "grad_norm": 1.7539904117584229,
      "learning_rate": 6.237547503177636e-07,
      "loss": 0.4462,
      "step": 378260
    },
    {
      "epoch": 154.84,
      "grad_norm": 1.7287214994430542,
      "learning_rate": 6.236469978818056e-07,
      "loss": 0.4332,
      "step": 378270
    },
    {
      "epoch": 154.84,
      "grad_norm": 1.7598881721496582,
      "learning_rate": 6.235392533253728e-07,
      "loss": 0.466,
      "step": 378280
    },
    {
      "epoch": 154.85,
      "grad_norm": 1.8274171352386475,
      "learning_rate": 6.234315166489593e-07,
      "loss": 0.4269,
      "step": 378290
    },
    {
      "epoch": 154.85,
      "grad_norm": 1.7770048379898071,
      "learning_rate": 6.233237878530582e-07,
      "loss": 0.4623,
      "step": 378300
    },
    {
      "epoch": 154.85,
      "grad_norm": 1.805090069770813,
      "learning_rate": 6.232160669381631e-07,
      "loss": 0.4433,
      "step": 378310
    },
    {
      "epoch": 154.86,
      "grad_norm": 1.890956163406372,
      "learning_rate": 6.231083539047673e-07,
      "loss": 0.4389,
      "step": 378320
    },
    {
      "epoch": 154.86,
      "grad_norm": 1.7332839965820312,
      "learning_rate": 6.230006487533648e-07,
      "loss": 0.4424,
      "step": 378330
    },
    {
      "epoch": 154.87,
      "grad_norm": 2.004377603530884,
      "learning_rate": 6.228929514844484e-07,
      "loss": 0.4431,
      "step": 378340
    },
    {
      "epoch": 154.87,
      "grad_norm": 2.1139230728149414,
      "learning_rate": 6.227852620985116e-07,
      "loss": 0.4372,
      "step": 378350
    },
    {
      "epoch": 154.88,
      "grad_norm": 2.9660282135009766,
      "learning_rate": 6.226775805960471e-07,
      "loss": 0.461,
      "step": 378360
    },
    {
      "epoch": 154.88,
      "grad_norm": 1.7601678371429443,
      "learning_rate": 6.225699069775495e-07,
      "loss": 0.4295,
      "step": 378370
    },
    {
      "epoch": 154.88,
      "grad_norm": 1.762805461883545,
      "learning_rate": 6.224622412435119e-07,
      "loss": 0.4554,
      "step": 378380
    },
    {
      "epoch": 154.89,
      "grad_norm": 2.0282981395721436,
      "learning_rate": 6.223545833944265e-07,
      "loss": 0.4483,
      "step": 378390
    },
    {
      "epoch": 154.89,
      "grad_norm": 2.2905654907226562,
      "learning_rate": 6.222469334307868e-07,
      "loss": 0.452,
      "step": 378400
    },
    {
      "epoch": 154.9,
      "grad_norm": 1.651811122894287,
      "learning_rate": 6.221392913530861e-07,
      "loss": 0.4348,
      "step": 378410
    },
    {
      "epoch": 154.9,
      "grad_norm": 1.9418565034866333,
      "learning_rate": 6.220316571618172e-07,
      "loss": 0.4441,
      "step": 378420
    },
    {
      "epoch": 154.9,
      "grad_norm": 2.189612627029419,
      "learning_rate": 6.219240308574737e-07,
      "loss": 0.4561,
      "step": 378430
    },
    {
      "epoch": 154.91,
      "grad_norm": 2.0718138217926025,
      "learning_rate": 6.218164124405481e-07,
      "loss": 0.4424,
      "step": 378440
    },
    {
      "epoch": 154.91,
      "grad_norm": 1.8466699123382568,
      "learning_rate": 6.217088019115336e-07,
      "loss": 0.4426,
      "step": 378450
    },
    {
      "epoch": 154.92,
      "grad_norm": 2.3954238891601562,
      "learning_rate": 6.216011992709232e-07,
      "loss": 0.4651,
      "step": 378460
    },
    {
      "epoch": 154.92,
      "grad_norm": 2.232112407684326,
      "learning_rate": 6.214936045192098e-07,
      "loss": 0.4462,
      "step": 378470
    },
    {
      "epoch": 154.92,
      "grad_norm": 1.91325044631958,
      "learning_rate": 6.21386017656886e-07,
      "loss": 0.4359,
      "step": 378480
    },
    {
      "epoch": 154.93,
      "grad_norm": 2.014719009399414,
      "learning_rate": 6.212784386844451e-07,
      "loss": 0.4545,
      "step": 378490
    },
    {
      "epoch": 154.93,
      "grad_norm": 1.963886022567749,
      "learning_rate": 6.211708676023797e-07,
      "loss": 0.4473,
      "step": 378500
    },
    {
      "epoch": 154.94,
      "grad_norm": 2.1333298683166504,
      "learning_rate": 6.210633044111822e-07,
      "loss": 0.4436,
      "step": 378510
    },
    {
      "epoch": 154.94,
      "grad_norm": 1.8485668897628784,
      "learning_rate": 6.20955749111346e-07,
      "loss": 0.4516,
      "step": 378520
    },
    {
      "epoch": 154.94,
      "grad_norm": 1.7810736894607544,
      "learning_rate": 6.208482017033634e-07,
      "loss": 0.4643,
      "step": 378530
    },
    {
      "epoch": 154.95,
      "grad_norm": 2.232273578643799,
      "learning_rate": 6.207406621877269e-07,
      "loss": 0.4571,
      "step": 378540
    },
    {
      "epoch": 154.95,
      "grad_norm": 1.8515727519989014,
      "learning_rate": 6.206331305649297e-07,
      "loss": 0.4322,
      "step": 378550
    },
    {
      "epoch": 154.96,
      "grad_norm": 2.2680156230926514,
      "learning_rate": 6.205256068354637e-07,
      "loss": 0.453,
      "step": 378560
    },
    {
      "epoch": 154.96,
      "grad_norm": 1.9552243947982788,
      "learning_rate": 6.204180909998219e-07,
      "loss": 0.4435,
      "step": 378570
    },
    {
      "epoch": 154.97,
      "grad_norm": 2.072408437728882,
      "learning_rate": 6.203105830584966e-07,
      "loss": 0.4525,
      "step": 378580
    },
    {
      "epoch": 154.97,
      "grad_norm": 1.8559281826019287,
      "learning_rate": 6.202030830119803e-07,
      "loss": 0.4497,
      "step": 378590
    },
    {
      "epoch": 154.97,
      "grad_norm": 1.9666706323623657,
      "learning_rate": 6.200955908607659e-07,
      "loss": 0.4501,
      "step": 378600
    },
    {
      "epoch": 154.98,
      "grad_norm": 2.4243571758270264,
      "learning_rate": 6.199881066053451e-07,
      "loss": 0.4382,
      "step": 378610
    },
    {
      "epoch": 154.98,
      "grad_norm": 1.8590320348739624,
      "learning_rate": 6.198806302462105e-07,
      "loss": 0.461,
      "step": 378620
    },
    {
      "epoch": 154.99,
      "grad_norm": 1.8925042152404785,
      "learning_rate": 6.197731617838544e-07,
      "loss": 0.4421,
      "step": 378630
    },
    {
      "epoch": 154.99,
      "grad_norm": 2.478827476501465,
      "learning_rate": 6.196657012187695e-07,
      "loss": 0.4822,
      "step": 378640
    },
    {
      "epoch": 154.99,
      "grad_norm": 1.8597065210342407,
      "learning_rate": 6.195582485514478e-07,
      "loss": 0.441,
      "step": 378650
    },
    {
      "epoch": 155.0,
      "grad_norm": 2.048715591430664,
      "learning_rate": 6.194508037823813e-07,
      "loss": 0.4511,
      "step": 378660
    },
    {
      "epoch": 155.0,
      "eval_loss": 0.4491494596004486,
      "eval_runtime": 52.3593,
      "eval_samples_per_second": 65.872,
      "eval_steps_per_second": 8.251,
      "step": 378665
    },
    {
      "epoch": 155.0,
      "grad_norm": 2.2445268630981445,
      "learning_rate": 6.193433669120625e-07,
      "loss": 0.4504,
      "step": 378670
    },
    {
      "epoch": 155.01,
      "grad_norm": 2.036545991897583,
      "learning_rate": 6.192359379409835e-07,
      "loss": 0.449,
      "step": 378680
    },
    {
      "epoch": 155.01,
      "grad_norm": 2.6456031799316406,
      "learning_rate": 6.191285168696368e-07,
      "loss": 0.4467,
      "step": 378690
    },
    {
      "epoch": 155.01,
      "grad_norm": 2.3124842643737793,
      "learning_rate": 6.190211036985133e-07,
      "loss": 0.4392,
      "step": 378700
    },
    {
      "epoch": 155.02,
      "grad_norm": 2.156418561935425,
      "learning_rate": 6.189136984281061e-07,
      "loss": 0.4563,
      "step": 378710
    },
    {
      "epoch": 155.02,
      "grad_norm": 1.8941501379013062,
      "learning_rate": 6.188063010589066e-07,
      "loss": 0.4739,
      "step": 378720
    },
    {
      "epoch": 155.03,
      "grad_norm": 2.1346869468688965,
      "learning_rate": 6.186989115914071e-07,
      "loss": 0.4438,
      "step": 378730
    },
    {
      "epoch": 155.03,
      "grad_norm": 2.2611148357391357,
      "learning_rate": 6.185915300260986e-07,
      "loss": 0.4495,
      "step": 378740
    },
    {
      "epoch": 155.03,
      "grad_norm": 1.8715846538543701,
      "learning_rate": 6.184841563634751e-07,
      "loss": 0.4363,
      "step": 378750
    },
    {
      "epoch": 155.04,
      "grad_norm": 2.5730748176574707,
      "learning_rate": 6.183767906040267e-07,
      "loss": 0.4402,
      "step": 378760
    },
    {
      "epoch": 155.04,
      "grad_norm": 1.7803584337234497,
      "learning_rate": 6.182694327482461e-07,
      "loss": 0.4616,
      "step": 378770
    },
    {
      "epoch": 155.05,
      "grad_norm": 1.990177035331726,
      "learning_rate": 6.181620827966246e-07,
      "loss": 0.439,
      "step": 378780
    },
    {
      "epoch": 155.05,
      "grad_norm": 2.203716278076172,
      "learning_rate": 6.18054740749654e-07,
      "loss": 0.4297,
      "step": 378790
    },
    {
      "epoch": 155.06,
      "grad_norm": 2.1442759037017822,
      "learning_rate": 6.179474066078264e-07,
      "loss": 0.4594,
      "step": 378800
    },
    {
      "epoch": 155.06,
      "grad_norm": 2.0682547092437744,
      "learning_rate": 6.17840080371633e-07,
      "loss": 0.4578,
      "step": 378810
    },
    {
      "epoch": 155.06,
      "grad_norm": 1.772497534751892,
      "learning_rate": 6.177327620415658e-07,
      "loss": 0.4553,
      "step": 378820
    },
    {
      "epoch": 155.07,
      "grad_norm": 2.23116135597229,
      "learning_rate": 6.176254516181167e-07,
      "loss": 0.4369,
      "step": 378830
    },
    {
      "epoch": 155.07,
      "grad_norm": 1.7295740842819214,
      "learning_rate": 6.175181491017765e-07,
      "loss": 0.4631,
      "step": 378840
    },
    {
      "epoch": 155.08,
      "grad_norm": 2.296320915222168,
      "learning_rate": 6.174108544930369e-07,
      "loss": 0.4629,
      "step": 378850
    },
    {
      "epoch": 155.08,
      "grad_norm": 2.1943676471710205,
      "learning_rate": 6.173035677923894e-07,
      "loss": 0.4538,
      "step": 378860
    },
    {
      "epoch": 155.08,
      "grad_norm": 2.159196138381958,
      "learning_rate": 6.171962890003259e-07,
      "loss": 0.4474,
      "step": 378870
    },
    {
      "epoch": 155.09,
      "grad_norm": 1.8709640502929688,
      "learning_rate": 6.170890181173373e-07,
      "loss": 0.4553,
      "step": 378880
    },
    {
      "epoch": 155.09,
      "grad_norm": 2.062213659286499,
      "learning_rate": 6.169817551439154e-07,
      "loss": 0.4706,
      "step": 378890
    },
    {
      "epoch": 155.1,
      "grad_norm": 1.9567769765853882,
      "learning_rate": 6.168745000805514e-07,
      "loss": 0.4361,
      "step": 378900
    },
    {
      "epoch": 155.1,
      "grad_norm": 1.7169194221496582,
      "learning_rate": 6.167672529277366e-07,
      "loss": 0.4235,
      "step": 378910
    },
    {
      "epoch": 155.1,
      "grad_norm": 1.9656957387924194,
      "learning_rate": 6.166600136859624e-07,
      "loss": 0.4166,
      "step": 378920
    },
    {
      "epoch": 155.11,
      "grad_norm": 1.9916787147521973,
      "learning_rate": 6.165527823557193e-07,
      "loss": 0.4412,
      "step": 378930
    },
    {
      "epoch": 155.11,
      "grad_norm": 2.1273787021636963,
      "learning_rate": 6.164455589374996e-07,
      "loss": 0.4477,
      "step": 378940
    },
    {
      "epoch": 155.12,
      "grad_norm": 1.9184250831604004,
      "learning_rate": 6.163383434317942e-07,
      "loss": 0.4326,
      "step": 378950
    },
    {
      "epoch": 155.12,
      "grad_norm": 2.256986141204834,
      "learning_rate": 6.16231135839094e-07,
      "loss": 0.4393,
      "step": 378960
    },
    {
      "epoch": 155.12,
      "grad_norm": 2.1015589237213135,
      "learning_rate": 6.161239361598909e-07,
      "loss": 0.4592,
      "step": 378970
    },
    {
      "epoch": 155.13,
      "grad_norm": 2.645150661468506,
      "learning_rate": 6.160167443946747e-07,
      "loss": 0.4646,
      "step": 378980
    },
    {
      "epoch": 155.13,
      "grad_norm": 2.2763726711273193,
      "learning_rate": 6.15909560543937e-07,
      "loss": 0.4443,
      "step": 378990
    },
    {
      "epoch": 155.14,
      "grad_norm": 2.157252073287964,
      "learning_rate": 6.158023846081686e-07,
      "loss": 0.4338,
      "step": 379000
    },
    {
      "epoch": 155.14,
      "grad_norm": 2.2918779850006104,
      "learning_rate": 6.156952165878606e-07,
      "loss": 0.4572,
      "step": 379010
    },
    {
      "epoch": 155.15,
      "grad_norm": 2.165012836456299,
      "learning_rate": 6.155880564835038e-07,
      "loss": 0.4583,
      "step": 379020
    },
    {
      "epoch": 155.15,
      "grad_norm": 1.8056803941726685,
      "learning_rate": 6.154809042955896e-07,
      "loss": 0.4441,
      "step": 379030
    },
    {
      "epoch": 155.15,
      "grad_norm": 2.0050222873687744,
      "learning_rate": 6.153737600246081e-07,
      "loss": 0.441,
      "step": 379040
    },
    {
      "epoch": 155.16,
      "grad_norm": 2.0930495262145996,
      "learning_rate": 6.152666236710506e-07,
      "loss": 0.46,
      "step": 379050
    },
    {
      "epoch": 155.16,
      "grad_norm": 2.6175215244293213,
      "learning_rate": 6.151594952354079e-07,
      "loss": 0.429,
      "step": 379060
    },
    {
      "epoch": 155.17,
      "grad_norm": 2.0732624530792236,
      "learning_rate": 6.150523747181704e-07,
      "loss": 0.4573,
      "step": 379070
    },
    {
      "epoch": 155.17,
      "grad_norm": 1.9561794996261597,
      "learning_rate": 6.149452621198291e-07,
      "loss": 0.437,
      "step": 379080
    },
    {
      "epoch": 155.17,
      "grad_norm": 1.766849398612976,
      "learning_rate": 6.148381574408747e-07,
      "loss": 0.4478,
      "step": 379090
    },
    {
      "epoch": 155.18,
      "grad_norm": 2.087164878845215,
      "learning_rate": 6.147310606817975e-07,
      "loss": 0.4581,
      "step": 379100
    },
    {
      "epoch": 155.18,
      "grad_norm": 2.3728883266448975,
      "learning_rate": 6.146239718430884e-07,
      "loss": 0.4422,
      "step": 379110
    },
    {
      "epoch": 155.19,
      "grad_norm": 1.7767689228057861,
      "learning_rate": 6.145168909252379e-07,
      "loss": 0.4563,
      "step": 379120
    },
    {
      "epoch": 155.19,
      "grad_norm": 2.185155153274536,
      "learning_rate": 6.144098179287362e-07,
      "loss": 0.4582,
      "step": 379130
    },
    {
      "epoch": 155.19,
      "grad_norm": 2.585867404937744,
      "learning_rate": 6.143027528540742e-07,
      "loss": 0.4782,
      "step": 379140
    },
    {
      "epoch": 155.2,
      "grad_norm": 5.04133939743042,
      "learning_rate": 6.141956957017423e-07,
      "loss": 0.4346,
      "step": 379150
    },
    {
      "epoch": 155.2,
      "grad_norm": 1.777671456336975,
      "learning_rate": 6.140886464722308e-07,
      "loss": 0.4349,
      "step": 379160
    },
    {
      "epoch": 155.21,
      "grad_norm": 2.382985830307007,
      "learning_rate": 6.139816051660299e-07,
      "loss": 0.4382,
      "step": 379170
    },
    {
      "epoch": 155.21,
      "grad_norm": 1.7865252494812012,
      "learning_rate": 6.138745717836303e-07,
      "loss": 0.453,
      "step": 379180
    },
    {
      "epoch": 155.21,
      "grad_norm": 1.7723156213760376,
      "learning_rate": 6.137675463255219e-07,
      "loss": 0.4537,
      "step": 379190
    },
    {
      "epoch": 155.22,
      "grad_norm": 2.096210241317749,
      "learning_rate": 6.136605287921955e-07,
      "loss": 0.4651,
      "step": 379200
    },
    {
      "epoch": 155.22,
      "grad_norm": 1.6890006065368652,
      "learning_rate": 6.135535191841408e-07,
      "loss": 0.4505,
      "step": 379210
    },
    {
      "epoch": 155.23,
      "grad_norm": 2.176429510116577,
      "learning_rate": 6.134465175018484e-07,
      "loss": 0.4291,
      "step": 379220
    },
    {
      "epoch": 155.23,
      "grad_norm": 2.5165133476257324,
      "learning_rate": 6.133395237458083e-07,
      "loss": 0.4509,
      "step": 379230
    },
    {
      "epoch": 155.24,
      "grad_norm": 1.860386848449707,
      "learning_rate": 6.132325379165105e-07,
      "loss": 0.4331,
      "step": 379240
    },
    {
      "epoch": 155.24,
      "grad_norm": 2.2257378101348877,
      "learning_rate": 6.131255600144453e-07,
      "loss": 0.4537,
      "step": 379250
    },
    {
      "epoch": 155.24,
      "grad_norm": 2.0927648544311523,
      "learning_rate": 6.130185900401023e-07,
      "loss": 0.4447,
      "step": 379260
    },
    {
      "epoch": 155.25,
      "grad_norm": 1.820086121559143,
      "learning_rate": 6.129116279939722e-07,
      "loss": 0.4451,
      "step": 379270
    },
    {
      "epoch": 155.25,
      "grad_norm": 2.0602242946624756,
      "learning_rate": 6.128046738765454e-07,
      "loss": 0.436,
      "step": 379280
    },
    {
      "epoch": 155.26,
      "grad_norm": 1.7888994216918945,
      "learning_rate": 6.126977276883102e-07,
      "loss": 0.466,
      "step": 379290
    },
    {
      "epoch": 155.26,
      "grad_norm": 2.555882453918457,
      "learning_rate": 6.125907894297573e-07,
      "loss": 0.445,
      "step": 379300
    },
    {
      "epoch": 155.26,
      "grad_norm": 2.228684186935425,
      "learning_rate": 6.124838591013759e-07,
      "loss": 0.4609,
      "step": 379310
    },
    {
      "epoch": 155.27,
      "grad_norm": 1.9064888954162598,
      "learning_rate": 6.123769367036574e-07,
      "loss": 0.4563,
      "step": 379320
    },
    {
      "epoch": 155.27,
      "grad_norm": 1.6762429475784302,
      "learning_rate": 6.122700222370908e-07,
      "loss": 0.4549,
      "step": 379330
    },
    {
      "epoch": 155.28,
      "grad_norm": 2.517948865890503,
      "learning_rate": 6.121631157021656e-07,
      "loss": 0.4449,
      "step": 379340
    },
    {
      "epoch": 155.28,
      "grad_norm": 1.999890685081482,
      "learning_rate": 6.12056217099372e-07,
      "loss": 0.4437,
      "step": 379350
    },
    {
      "epoch": 155.28,
      "grad_norm": 1.8346024751663208,
      "learning_rate": 6.119493264291993e-07,
      "loss": 0.4387,
      "step": 379360
    },
    {
      "epoch": 155.29,
      "grad_norm": 2.3633856773376465,
      "learning_rate": 6.118424436921373e-07,
      "loss": 0.4455,
      "step": 379370
    },
    {
      "epoch": 155.29,
      "grad_norm": 1.784955382347107,
      "learning_rate": 6.117355688886758e-07,
      "loss": 0.4505,
      "step": 379380
    },
    {
      "epoch": 155.3,
      "grad_norm": 2.0760955810546875,
      "learning_rate": 6.11628702019304e-07,
      "loss": 0.4366,
      "step": 379390
    },
    {
      "epoch": 155.3,
      "grad_norm": 2.003383159637451,
      "learning_rate": 6.115218430845116e-07,
      "loss": 0.4645,
      "step": 379400
    },
    {
      "epoch": 155.3,
      "grad_norm": 2.123476266860962,
      "learning_rate": 6.114149920847889e-07,
      "loss": 0.459,
      "step": 379410
    },
    {
      "epoch": 155.31,
      "grad_norm": 1.9919426441192627,
      "learning_rate": 6.113081490206237e-07,
      "loss": 0.4347,
      "step": 379420
    },
    {
      "epoch": 155.31,
      "grad_norm": 2.171534538269043,
      "learning_rate": 6.112013138925066e-07,
      "loss": 0.4407,
      "step": 379430
    },
    {
      "epoch": 155.32,
      "grad_norm": 2.128134250640869,
      "learning_rate": 6.110944867009263e-07,
      "loss": 0.429,
      "step": 379440
    },
    {
      "epoch": 155.32,
      "grad_norm": 1.9267815351486206,
      "learning_rate": 6.109876674463729e-07,
      "loss": 0.4571,
      "step": 379450
    },
    {
      "epoch": 155.33,
      "grad_norm": 2.0538065433502197,
      "learning_rate": 6.108808561293352e-07,
      "loss": 0.4284,
      "step": 379460
    },
    {
      "epoch": 155.33,
      "grad_norm": 1.907663106918335,
      "learning_rate": 6.107740527503028e-07,
      "loss": 0.446,
      "step": 379470
    },
    {
      "epoch": 155.33,
      "grad_norm": 1.6452727317810059,
      "learning_rate": 6.10667257309765e-07,
      "loss": 0.4364,
      "step": 379480
    },
    {
      "epoch": 155.34,
      "grad_norm": 1.8679920434951782,
      "learning_rate": 6.1056046980821e-07,
      "loss": 0.4249,
      "step": 379490
    },
    {
      "epoch": 155.34,
      "grad_norm": 1.9793187379837036,
      "learning_rate": 6.104536902461282e-07,
      "loss": 0.4347,
      "step": 379500
    },
    {
      "epoch": 155.35,
      "grad_norm": 1.909948706626892,
      "learning_rate": 6.103469186240088e-07,
      "loss": 0.4366,
      "step": 379510
    },
    {
      "epoch": 155.35,
      "grad_norm": 1.8964170217514038,
      "learning_rate": 6.102401549423402e-07,
      "loss": 0.4631,
      "step": 379520
    },
    {
      "epoch": 155.35,
      "grad_norm": 2.2604877948760986,
      "learning_rate": 6.101333992016121e-07,
      "loss": 0.4446,
      "step": 379530
    },
    {
      "epoch": 155.36,
      "grad_norm": 1.762642502784729,
      "learning_rate": 6.100266514023128e-07,
      "loss": 0.4584,
      "step": 379540
    },
    {
      "epoch": 155.36,
      "grad_norm": 1.7944227457046509,
      "learning_rate": 6.099199115449324e-07,
      "loss": 0.4443,
      "step": 379550
    },
    {
      "epoch": 155.37,
      "grad_norm": 1.8710490465164185,
      "learning_rate": 6.098131796299585e-07,
      "loss": 0.448,
      "step": 379560
    },
    {
      "epoch": 155.37,
      "grad_norm": 2.2327699661254883,
      "learning_rate": 6.097064556578805e-07,
      "loss": 0.4283,
      "step": 379570
    },
    {
      "epoch": 155.37,
      "grad_norm": 1.8272972106933594,
      "learning_rate": 6.095997396291876e-07,
      "loss": 0.4634,
      "step": 379580
    },
    {
      "epoch": 155.38,
      "grad_norm": 2.003176689147949,
      "learning_rate": 6.094930315443685e-07,
      "loss": 0.4548,
      "step": 379590
    },
    {
      "epoch": 155.38,
      "grad_norm": 1.9596081972122192,
      "learning_rate": 6.093863314039118e-07,
      "loss": 0.4423,
      "step": 379600
    },
    {
      "epoch": 155.39,
      "grad_norm": 2.308093547821045,
      "learning_rate": 6.092796392083067e-07,
      "loss": 0.4358,
      "step": 379610
    },
    {
      "epoch": 155.39,
      "grad_norm": 3.0653531551361084,
      "learning_rate": 6.091729549580414e-07,
      "loss": 0.4275,
      "step": 379620
    },
    {
      "epoch": 155.4,
      "grad_norm": 1.864478588104248,
      "learning_rate": 6.090662786536051e-07,
      "loss": 0.4278,
      "step": 379630
    },
    {
      "epoch": 155.4,
      "grad_norm": 2.154404878616333,
      "learning_rate": 6.089596102954864e-07,
      "loss": 0.4409,
      "step": 379640
    },
    {
      "epoch": 155.4,
      "grad_norm": 2.0477283000946045,
      "learning_rate": 6.088529498841737e-07,
      "loss": 0.4664,
      "step": 379650
    },
    {
      "epoch": 155.41,
      "grad_norm": 2.0386452674865723,
      "learning_rate": 6.087462974201558e-07,
      "loss": 0.4491,
      "step": 379660
    },
    {
      "epoch": 155.41,
      "grad_norm": 2.0796709060668945,
      "learning_rate": 6.08639652903921e-07,
      "loss": 0.4444,
      "step": 379670
    },
    {
      "epoch": 155.42,
      "grad_norm": 2.260855197906494,
      "learning_rate": 6.085330163359578e-07,
      "loss": 0.4301,
      "step": 379680
    },
    {
      "epoch": 155.42,
      "grad_norm": 1.9454092979431152,
      "learning_rate": 6.084263877167552e-07,
      "loss": 0.4539,
      "step": 379690
    },
    {
      "epoch": 155.42,
      "grad_norm": 1.878501534461975,
      "learning_rate": 6.083197670468013e-07,
      "loss": 0.4538,
      "step": 379700
    },
    {
      "epoch": 155.43,
      "grad_norm": 2.185392141342163,
      "learning_rate": 6.082131543265844e-07,
      "loss": 0.4433,
      "step": 379710
    },
    {
      "epoch": 155.43,
      "grad_norm": 2.0263125896453857,
      "learning_rate": 6.081065495565932e-07,
      "loss": 0.4409,
      "step": 379720
    },
    {
      "epoch": 155.44,
      "grad_norm": 1.750823736190796,
      "learning_rate": 6.079999527373157e-07,
      "loss": 0.4463,
      "step": 379730
    },
    {
      "epoch": 155.44,
      "grad_norm": 2.1008005142211914,
      "learning_rate": 6.078933638692401e-07,
      "loss": 0.4441,
      "step": 379740
    },
    {
      "epoch": 155.44,
      "grad_norm": 2.941107749938965,
      "learning_rate": 6.077867829528551e-07,
      "loss": 0.461,
      "step": 379750
    },
    {
      "epoch": 155.45,
      "grad_norm": 2.1068453788757324,
      "learning_rate": 6.07680209988649e-07,
      "loss": 0.4568,
      "step": 379760
    },
    {
      "epoch": 155.45,
      "grad_norm": 1.84490168094635,
      "learning_rate": 6.075736449771095e-07,
      "loss": 0.451,
      "step": 379770
    },
    {
      "epoch": 155.46,
      "grad_norm": 2.287095308303833,
      "learning_rate": 6.074670879187252e-07,
      "loss": 0.4542,
      "step": 379780
    },
    {
      "epoch": 155.46,
      "grad_norm": 1.7784390449523926,
      "learning_rate": 6.073605388139837e-07,
      "loss": 0.4419,
      "step": 379790
    },
    {
      "epoch": 155.46,
      "grad_norm": 2.152777671813965,
      "learning_rate": 6.072539976633738e-07,
      "loss": 0.4457,
      "step": 379800
    },
    {
      "epoch": 155.47,
      "grad_norm": 2.417572021484375,
      "learning_rate": 6.071474644673829e-07,
      "loss": 0.4426,
      "step": 379810
    },
    {
      "epoch": 155.47,
      "grad_norm": 1.9100124835968018,
      "learning_rate": 6.070409392264995e-07,
      "loss": 0.4599,
      "step": 379820
    },
    {
      "epoch": 155.48,
      "grad_norm": 2.1396431922912598,
      "learning_rate": 6.06934421941211e-07,
      "loss": 0.4425,
      "step": 379830
    },
    {
      "epoch": 155.48,
      "grad_norm": 1.989003300666809,
      "learning_rate": 6.068279126120057e-07,
      "loss": 0.4484,
      "step": 379840
    },
    {
      "epoch": 155.49,
      "grad_norm": 1.9571292400360107,
      "learning_rate": 6.067214112393718e-07,
      "loss": 0.4455,
      "step": 379850
    },
    {
      "epoch": 155.49,
      "grad_norm": 1.967932939529419,
      "learning_rate": 6.066149178237974e-07,
      "loss": 0.4402,
      "step": 379860
    },
    {
      "epoch": 155.49,
      "grad_norm": 1.9541085958480835,
      "learning_rate": 6.065084323657682e-07,
      "loss": 0.4334,
      "step": 379870
    },
    {
      "epoch": 155.5,
      "grad_norm": 1.6336768865585327,
      "learning_rate": 6.064019548657746e-07,
      "loss": 0.439,
      "step": 379880
    },
    {
      "epoch": 155.5,
      "grad_norm": 2.690384864807129,
      "learning_rate": 6.062954853243031e-07,
      "loss": 0.4591,
      "step": 379890
    },
    {
      "epoch": 155.51,
      "grad_norm": 2.2108523845672607,
      "learning_rate": 6.061890237418419e-07,
      "loss": 0.4469,
      "step": 379900
    },
    {
      "epoch": 155.51,
      "grad_norm": 1.9313877820968628,
      "learning_rate": 6.060825701188782e-07,
      "loss": 0.4642,
      "step": 379910
    },
    {
      "epoch": 155.51,
      "grad_norm": 2.411649227142334,
      "learning_rate": 6.059761244559001e-07,
      "loss": 0.4662,
      "step": 379920
    },
    {
      "epoch": 155.52,
      "grad_norm": 2.227013349533081,
      "learning_rate": 6.058696867533948e-07,
      "loss": 0.4578,
      "step": 379930
    },
    {
      "epoch": 155.52,
      "grad_norm": 1.7009679079055786,
      "learning_rate": 6.057632570118502e-07,
      "loss": 0.4536,
      "step": 379940
    },
    {
      "epoch": 155.53,
      "grad_norm": 2.0883233547210693,
      "learning_rate": 6.056568352317535e-07,
      "loss": 0.44,
      "step": 379950
    },
    {
      "epoch": 155.53,
      "grad_norm": 1.9206420183181763,
      "learning_rate": 6.055504214135926e-07,
      "loss": 0.4486,
      "step": 379960
    },
    {
      "epoch": 155.53,
      "grad_norm": 2.2732059955596924,
      "learning_rate": 6.054440155578547e-07,
      "loss": 0.4248,
      "step": 379970
    },
    {
      "epoch": 155.54,
      "grad_norm": 2.119412422180176,
      "learning_rate": 6.053376176650272e-07,
      "loss": 0.4367,
      "step": 379980
    },
    {
      "epoch": 155.54,
      "grad_norm": 2.255902051925659,
      "learning_rate": 6.052312277355985e-07,
      "loss": 0.4513,
      "step": 379990
    },
    {
      "epoch": 155.55,
      "grad_norm": 2.2691709995269775,
      "learning_rate": 6.051248457700542e-07,
      "loss": 0.4521,
      "step": 380000
    },
    {
      "epoch": 155.55,
      "grad_norm": 2.054831027984619,
      "learning_rate": 6.050184717688826e-07,
      "loss": 0.4575,
      "step": 380010
    },
    {
      "epoch": 155.55,
      "grad_norm": 2.0046138763427734,
      "learning_rate": 6.049121057325708e-07,
      "loss": 0.4513,
      "step": 380020
    },
    {
      "epoch": 155.56,
      "grad_norm": 1.6981902122497559,
      "learning_rate": 6.048057476616061e-07,
      "loss": 0.4438,
      "step": 380030
    },
    {
      "epoch": 155.56,
      "grad_norm": 1.9993902444839478,
      "learning_rate": 6.046993975564757e-07,
      "loss": 0.4497,
      "step": 380040
    },
    {
      "epoch": 155.57,
      "grad_norm": 1.674667477607727,
      "learning_rate": 6.045930554176668e-07,
      "loss": 0.4733,
      "step": 380050
    },
    {
      "epoch": 155.57,
      "grad_norm": 2.2381298542022705,
      "learning_rate": 6.044867212456659e-07,
      "loss": 0.4518,
      "step": 380060
    },
    {
      "epoch": 155.58,
      "grad_norm": 1.8425508737564087,
      "learning_rate": 6.043803950409614e-07,
      "loss": 0.4462,
      "step": 380070
    },
    {
      "epoch": 155.58,
      "grad_norm": 2.2876639366149902,
      "learning_rate": 6.042740768040395e-07,
      "loss": 0.4553,
      "step": 380080
    },
    {
      "epoch": 155.58,
      "grad_norm": 1.7667933702468872,
      "learning_rate": 6.041677665353874e-07,
      "loss": 0.4373,
      "step": 380090
    },
    {
      "epoch": 155.59,
      "grad_norm": 1.9836881160736084,
      "learning_rate": 6.040614642354922e-07,
      "loss": 0.4269,
      "step": 380100
    },
    {
      "epoch": 155.59,
      "grad_norm": 2.104987144470215,
      "learning_rate": 6.039551699048409e-07,
      "loss": 0.4569,
      "step": 380110
    },
    {
      "epoch": 155.6,
      "grad_norm": 2.102593421936035,
      "learning_rate": 6.038488835439203e-07,
      "loss": 0.428,
      "step": 380120
    },
    {
      "epoch": 155.6,
      "grad_norm": 2.012753963470459,
      "learning_rate": 6.037426051532178e-07,
      "loss": 0.4591,
      "step": 380130
    },
    {
      "epoch": 155.6,
      "grad_norm": 1.691979169845581,
      "learning_rate": 6.036363347332189e-07,
      "loss": 0.4324,
      "step": 380140
    },
    {
      "epoch": 155.61,
      "grad_norm": 2.295950174331665,
      "learning_rate": 6.035300722844114e-07,
      "loss": 0.4451,
      "step": 380150
    },
    {
      "epoch": 155.61,
      "grad_norm": 1.9376710653305054,
      "learning_rate": 6.03423817807282e-07,
      "loss": 0.4191,
      "step": 380160
    },
    {
      "epoch": 155.62,
      "grad_norm": 1.8827908039093018,
      "learning_rate": 6.03317571302317e-07,
      "loss": 0.4405,
      "step": 380170
    },
    {
      "epoch": 155.62,
      "grad_norm": 2.250098943710327,
      "learning_rate": 6.032113327700036e-07,
      "loss": 0.4588,
      "step": 380180
    },
    {
      "epoch": 155.62,
      "grad_norm": 2.2096023559570312,
      "learning_rate": 6.031051022108281e-07,
      "loss": 0.4505,
      "step": 380190
    },
    {
      "epoch": 155.63,
      "grad_norm": 3.027524471282959,
      "learning_rate": 6.029988796252774e-07,
      "loss": 0.4632,
      "step": 380200
    },
    {
      "epoch": 155.63,
      "grad_norm": 2.653616189956665,
      "learning_rate": 6.028926650138382e-07,
      "loss": 0.4487,
      "step": 380210
    },
    {
      "epoch": 155.64,
      "grad_norm": 2.0035078525543213,
      "learning_rate": 6.027864583769966e-07,
      "loss": 0.443,
      "step": 380220
    },
    {
      "epoch": 155.64,
      "grad_norm": 1.8104145526885986,
      "learning_rate": 6.026802597152393e-07,
      "loss": 0.4622,
      "step": 380230
    },
    {
      "epoch": 155.64,
      "grad_norm": 1.8355083465576172,
      "learning_rate": 6.02574069029053e-07,
      "loss": 0.4357,
      "step": 380240
    },
    {
      "epoch": 155.65,
      "grad_norm": 1.8361666202545166,
      "learning_rate": 6.024678863189234e-07,
      "loss": 0.4666,
      "step": 380250
    },
    {
      "epoch": 155.65,
      "grad_norm": 2.0269782543182373,
      "learning_rate": 6.023617115853381e-07,
      "loss": 0.4425,
      "step": 380260
    },
    {
      "epoch": 155.66,
      "grad_norm": 2.126488447189331,
      "learning_rate": 6.022555448287834e-07,
      "loss": 0.4269,
      "step": 380270
    },
    {
      "epoch": 155.66,
      "grad_norm": 2.1423885822296143,
      "learning_rate": 6.021493860497445e-07,
      "loss": 0.4716,
      "step": 380280
    },
    {
      "epoch": 155.67,
      "grad_norm": 2.2921013832092285,
      "learning_rate": 6.020432352487083e-07,
      "loss": 0.4552,
      "step": 380290
    },
    {
      "epoch": 155.67,
      "grad_norm": 2.6361083984375,
      "learning_rate": 6.019370924261611e-07,
      "loss": 0.4341,
      "step": 380300
    },
    {
      "epoch": 155.67,
      "grad_norm": 2.0904228687286377,
      "learning_rate": 6.01830957582589e-07,
      "loss": 0.485,
      "step": 380310
    },
    {
      "epoch": 155.68,
      "grad_norm": 2.0030629634857178,
      "learning_rate": 6.017248307184783e-07,
      "loss": 0.455,
      "step": 380320
    },
    {
      "epoch": 155.68,
      "grad_norm": 2.9979336261749268,
      "learning_rate": 6.016187118343152e-07,
      "loss": 0.4554,
      "step": 380330
    },
    {
      "epoch": 155.69,
      "grad_norm": 1.8673691749572754,
      "learning_rate": 6.015126009305856e-07,
      "loss": 0.4379,
      "step": 380340
    },
    {
      "epoch": 155.69,
      "grad_norm": 1.8956046104431152,
      "learning_rate": 6.014064980077758e-07,
      "loss": 0.4686,
      "step": 380350
    },
    {
      "epoch": 155.69,
      "grad_norm": 2.128692626953125,
      "learning_rate": 6.013004030663718e-07,
      "loss": 0.4674,
      "step": 380360
    },
    {
      "epoch": 155.7,
      "grad_norm": 2.0874531269073486,
      "learning_rate": 6.011943161068593e-07,
      "loss": 0.4567,
      "step": 380370
    },
    {
      "epoch": 155.7,
      "grad_norm": 2.0862457752227783,
      "learning_rate": 6.010882371297247e-07,
      "loss": 0.4384,
      "step": 380380
    },
    {
      "epoch": 155.71,
      "grad_norm": 2.011049509048462,
      "learning_rate": 6.009821661354536e-07,
      "loss": 0.4354,
      "step": 380390
    },
    {
      "epoch": 155.71,
      "grad_norm": 2.182265520095825,
      "learning_rate": 6.008761031245323e-07,
      "loss": 0.4514,
      "step": 380400
    },
    {
      "epoch": 155.71,
      "grad_norm": 2.2977287769317627,
      "learning_rate": 6.007700480974462e-07,
      "loss": 0.4603,
      "step": 380410
    },
    {
      "epoch": 155.72,
      "grad_norm": 2.2432804107666016,
      "learning_rate": 6.006640010546813e-07,
      "loss": 0.4451,
      "step": 380420
    },
    {
      "epoch": 155.72,
      "grad_norm": 2.2999472618103027,
      "learning_rate": 6.005579619967235e-07,
      "loss": 0.4583,
      "step": 380430
    },
    {
      "epoch": 155.73,
      "grad_norm": 1.837660312652588,
      "learning_rate": 6.004519309240585e-07,
      "loss": 0.4543,
      "step": 380440
    },
    {
      "epoch": 155.73,
      "grad_norm": 2.103970766067505,
      "learning_rate": 6.003459078371719e-07,
      "loss": 0.4464,
      "step": 380450
    },
    {
      "epoch": 155.73,
      "grad_norm": 1.7914435863494873,
      "learning_rate": 6.002398927365494e-07,
      "loss": 0.4454,
      "step": 380460
    },
    {
      "epoch": 155.74,
      "grad_norm": 2.184861183166504,
      "learning_rate": 6.001338856226767e-07,
      "loss": 0.443,
      "step": 380470
    },
    {
      "epoch": 155.74,
      "grad_norm": 2.2363595962524414,
      "learning_rate": 6.000278864960397e-07,
      "loss": 0.4635,
      "step": 380480
    },
    {
      "epoch": 155.75,
      "grad_norm": 1.8033764362335205,
      "learning_rate": 5.999218953571233e-07,
      "loss": 0.4603,
      "step": 380490
    },
    {
      "epoch": 155.75,
      "grad_norm": 1.8001939058303833,
      "learning_rate": 5.998159122064135e-07,
      "loss": 0.4588,
      "step": 380500
    },
    {
      "epoch": 155.76,
      "grad_norm": 2.11126446723938,
      "learning_rate": 5.997099370443956e-07,
      "loss": 0.4493,
      "step": 380510
    },
    {
      "epoch": 155.76,
      "grad_norm": 2.0039491653442383,
      "learning_rate": 5.996039698715551e-07,
      "loss": 0.438,
      "step": 380520
    },
    {
      "epoch": 155.76,
      "grad_norm": 2.148477792739868,
      "learning_rate": 5.994980106883776e-07,
      "loss": 0.4292,
      "step": 380530
    },
    {
      "epoch": 155.77,
      "grad_norm": 2.2944180965423584,
      "learning_rate": 5.993920594953482e-07,
      "loss": 0.4552,
      "step": 380540
    },
    {
      "epoch": 155.77,
      "grad_norm": 2.0097341537475586,
      "learning_rate": 5.992861162929524e-07,
      "loss": 0.4465,
      "step": 380550
    },
    {
      "epoch": 155.78,
      "grad_norm": 2.0239436626434326,
      "learning_rate": 5.991801810816756e-07,
      "loss": 0.4361,
      "step": 380560
    },
    {
      "epoch": 155.78,
      "grad_norm": 1.7583523988723755,
      "learning_rate": 5.990742538620028e-07,
      "loss": 0.4599,
      "step": 380570
    },
    {
      "epoch": 155.78,
      "grad_norm": 2.076359987258911,
      "learning_rate": 5.989683346344203e-07,
      "loss": 0.4396,
      "step": 380580
    },
    {
      "epoch": 155.79,
      "grad_norm": 1.9750456809997559,
      "learning_rate": 5.988624233994116e-07,
      "loss": 0.4487,
      "step": 380590
    },
    {
      "epoch": 155.79,
      "grad_norm": 2.2842745780944824,
      "learning_rate": 5.987565201574625e-07,
      "loss": 0.44,
      "step": 380600
    },
    {
      "epoch": 155.8,
      "grad_norm": 2.19901180267334,
      "learning_rate": 5.986506249090585e-07,
      "loss": 0.4493,
      "step": 380610
    },
    {
      "epoch": 155.8,
      "grad_norm": 2.1836862564086914,
      "learning_rate": 5.985447376546845e-07,
      "loss": 0.4415,
      "step": 380620
    },
    {
      "epoch": 155.8,
      "grad_norm": 2.0837268829345703,
      "learning_rate": 5.984388583948246e-07,
      "loss": 0.4496,
      "step": 380630
    },
    {
      "epoch": 155.81,
      "grad_norm": 2.5443966388702393,
      "learning_rate": 5.983329871299655e-07,
      "loss": 0.4665,
      "step": 380640
    },
    {
      "epoch": 155.81,
      "grad_norm": 1.9351624250411987,
      "learning_rate": 5.982271238605917e-07,
      "loss": 0.4246,
      "step": 380650
    },
    {
      "epoch": 155.82,
      "grad_norm": 1.999884009361267,
      "learning_rate": 5.981212685871876e-07,
      "loss": 0.4401,
      "step": 380660
    },
    {
      "epoch": 155.82,
      "grad_norm": 3.055149793624878,
      "learning_rate": 5.980154213102383e-07,
      "loss": 0.4497,
      "step": 380670
    },
    {
      "epoch": 155.82,
      "grad_norm": 2.4542295932769775,
      "learning_rate": 5.979095820302289e-07,
      "loss": 0.45,
      "step": 380680
    },
    {
      "epoch": 155.83,
      "grad_norm": 2.0054099559783936,
      "learning_rate": 5.97803750747644e-07,
      "loss": 0.4513,
      "step": 380690
    },
    {
      "epoch": 155.83,
      "grad_norm": 1.8719642162322998,
      "learning_rate": 5.976979274629685e-07,
      "loss": 0.4466,
      "step": 380700
    },
    {
      "epoch": 155.84,
      "grad_norm": 1.9439573287963867,
      "learning_rate": 5.97592112176688e-07,
      "loss": 0.439,
      "step": 380710
    },
    {
      "epoch": 155.84,
      "grad_norm": 2.1433820724487305,
      "learning_rate": 5.974863048892855e-07,
      "loss": 0.4567,
      "step": 380720
    },
    {
      "epoch": 155.85,
      "grad_norm": 2.0417730808258057,
      "learning_rate": 5.973805056012464e-07,
      "loss": 0.4371,
      "step": 380730
    },
    {
      "epoch": 155.85,
      "grad_norm": 2.329045534133911,
      "learning_rate": 5.972747143130558e-07,
      "loss": 0.4479,
      "step": 380740
    },
    {
      "epoch": 155.85,
      "grad_norm": 2.0865700244903564,
      "learning_rate": 5.97168931025198e-07,
      "loss": 0.4418,
      "step": 380750
    },
    {
      "epoch": 155.86,
      "grad_norm": 1.8744888305664062,
      "learning_rate": 5.970631557381573e-07,
      "loss": 0.4252,
      "step": 380760
    },
    {
      "epoch": 155.86,
      "grad_norm": 2.1849334239959717,
      "learning_rate": 5.969573884524188e-07,
      "loss": 0.4498,
      "step": 380770
    },
    {
      "epoch": 155.87,
      "grad_norm": 2.0836942195892334,
      "learning_rate": 5.968516291684665e-07,
      "loss": 0.4308,
      "step": 380780
    },
    {
      "epoch": 155.87,
      "grad_norm": 2.2785394191741943,
      "learning_rate": 5.967458778867852e-07,
      "loss": 0.4417,
      "step": 380790
    },
    {
      "epoch": 155.87,
      "grad_norm": 1.9421769380569458,
      "learning_rate": 5.966401346078591e-07,
      "loss": 0.4485,
      "step": 380800
    },
    {
      "epoch": 155.88,
      "grad_norm": 2.067903518676758,
      "learning_rate": 5.965343993321723e-07,
      "loss": 0.4376,
      "step": 380810
    },
    {
      "epoch": 155.88,
      "grad_norm": 1.8879388570785522,
      "learning_rate": 5.964286720602102e-07,
      "loss": 0.4496,
      "step": 380820
    },
    {
      "epoch": 155.89,
      "grad_norm": 2.5053629875183105,
      "learning_rate": 5.963229527924563e-07,
      "loss": 0.4468,
      "step": 380830
    },
    {
      "epoch": 155.89,
      "grad_norm": 1.6517016887664795,
      "learning_rate": 5.962172415293952e-07,
      "loss": 0.4457,
      "step": 380840
    },
    {
      "epoch": 155.89,
      "grad_norm": 1.8936351537704468,
      "learning_rate": 5.961115382715117e-07,
      "loss": 0.4563,
      "step": 380850
    },
    {
      "epoch": 155.9,
      "grad_norm": 2.162144660949707,
      "learning_rate": 5.960058430192887e-07,
      "loss": 0.4331,
      "step": 380860
    },
    {
      "epoch": 155.9,
      "grad_norm": 2.2010905742645264,
      "learning_rate": 5.959001557732109e-07,
      "loss": 0.4273,
      "step": 380870
    },
    {
      "epoch": 155.91,
      "grad_norm": 2.1226229667663574,
      "learning_rate": 5.957944765337626e-07,
      "loss": 0.4477,
      "step": 380880
    },
    {
      "epoch": 155.91,
      "grad_norm": 1.9266655445098877,
      "learning_rate": 5.956888053014279e-07,
      "loss": 0.438,
      "step": 380890
    },
    {
      "epoch": 155.91,
      "grad_norm": 1.663449764251709,
      "learning_rate": 5.955831420766907e-07,
      "loss": 0.4447,
      "step": 380900
    },
    {
      "epoch": 155.92,
      "grad_norm": 2.378596782684326,
      "learning_rate": 5.954774868600353e-07,
      "loss": 0.4397,
      "step": 380910
    },
    {
      "epoch": 155.92,
      "grad_norm": 1.9636321067810059,
      "learning_rate": 5.953718396519454e-07,
      "loss": 0.4691,
      "step": 380920
    },
    {
      "epoch": 155.93,
      "grad_norm": 1.8204946517944336,
      "learning_rate": 5.952662004529053e-07,
      "loss": 0.4705,
      "step": 380930
    },
    {
      "epoch": 155.93,
      "grad_norm": 2.3090763092041016,
      "learning_rate": 5.951605692633985e-07,
      "loss": 0.4799,
      "step": 380940
    },
    {
      "epoch": 155.94,
      "grad_norm": 1.9456251859664917,
      "learning_rate": 5.950549460839091e-07,
      "loss": 0.4604,
      "step": 380950
    },
    {
      "epoch": 155.94,
      "grad_norm": 2.254655599594116,
      "learning_rate": 5.949493309149208e-07,
      "loss": 0.4479,
      "step": 380960
    },
    {
      "epoch": 155.94,
      "grad_norm": 1.6991245746612549,
      "learning_rate": 5.948437237569179e-07,
      "loss": 0.4508,
      "step": 380970
    },
    {
      "epoch": 155.95,
      "grad_norm": 2.1063239574432373,
      "learning_rate": 5.947381246103835e-07,
      "loss": 0.4458,
      "step": 380980
    },
    {
      "epoch": 155.95,
      "grad_norm": 2.0291664600372314,
      "learning_rate": 5.946325334758016e-07,
      "loss": 0.4406,
      "step": 380990
    },
    {
      "epoch": 155.96,
      "grad_norm": 1.9343795776367188,
      "learning_rate": 5.945269503536559e-07,
      "loss": 0.4482,
      "step": 381000
    },
    {
      "epoch": 155.96,
      "grad_norm": 1.8104809522628784,
      "learning_rate": 5.944213752444303e-07,
      "loss": 0.428,
      "step": 381010
    },
    {
      "epoch": 155.96,
      "grad_norm": 2.1030712127685547,
      "learning_rate": 5.94315808148608e-07,
      "loss": 0.443,
      "step": 381020
    },
    {
      "epoch": 155.97,
      "grad_norm": 1.6727708578109741,
      "learning_rate": 5.94210249066673e-07,
      "loss": 0.4369,
      "step": 381030
    },
    {
      "epoch": 155.97,
      "grad_norm": 1.9878345727920532,
      "learning_rate": 5.941046979991087e-07,
      "loss": 0.4445,
      "step": 381040
    },
    {
      "epoch": 155.98,
      "grad_norm": 1.9223453998565674,
      "learning_rate": 5.939991549463983e-07,
      "loss": 0.4202,
      "step": 381050
    },
    {
      "epoch": 155.98,
      "grad_norm": 1.8301175832748413,
      "learning_rate": 5.938936199090258e-07,
      "loss": 0.4358,
      "step": 381060
    },
    {
      "epoch": 155.98,
      "grad_norm": 2.110363483428955,
      "learning_rate": 5.937880928874742e-07,
      "loss": 0.4591,
      "step": 381070
    },
    {
      "epoch": 155.99,
      "grad_norm": 1.9898849725723267,
      "learning_rate": 5.936825738822272e-07,
      "loss": 0.4461,
      "step": 381080
    },
    {
      "epoch": 155.99,
      "grad_norm": 2.0427746772766113,
      "learning_rate": 5.935770628937679e-07,
      "loss": 0.4544,
      "step": 381090
    },
    {
      "epoch": 156.0,
      "grad_norm": 2.731483221054077,
      "learning_rate": 5.934715599225799e-07,
      "loss": 0.4405,
      "step": 381100
    },
    {
      "epoch": 156.0,
      "eval_loss": 0.4487847685813904,
      "eval_runtime": 56.8705,
      "eval_samples_per_second": 60.647,
      "eval_steps_per_second": 7.596,
      "step": 381108
    },
    {
      "epoch": 156.0,
      "grad_norm": 2.2313199043273926,
      "learning_rate": 5.933660649691462e-07,
      "loss": 0.449,
      "step": 381110
    },
    {
      "epoch": 156.0,
      "grad_norm": 1.9882502555847168,
      "learning_rate": 5.932605780339503e-07,
      "loss": 0.4632,
      "step": 381120
    },
    {
      "epoch": 156.01,
      "grad_norm": 1.7834521532058716,
      "learning_rate": 5.931550991174755e-07,
      "loss": 0.4324,
      "step": 381130
    },
    {
      "epoch": 156.01,
      "grad_norm": 2.2431423664093018,
      "learning_rate": 5.930496282202046e-07,
      "loss": 0.4528,
      "step": 381140
    },
    {
      "epoch": 156.02,
      "grad_norm": 1.8057948350906372,
      "learning_rate": 5.929441653426211e-07,
      "loss": 0.4466,
      "step": 381150
    },
    {
      "epoch": 156.02,
      "grad_norm": 1.7109581232070923,
      "learning_rate": 5.928387104852087e-07,
      "loss": 0.4416,
      "step": 381160
    },
    {
      "epoch": 156.03,
      "grad_norm": 2.0279276371002197,
      "learning_rate": 5.927332636484491e-07,
      "loss": 0.4487,
      "step": 381170
    },
    {
      "epoch": 156.03,
      "grad_norm": 2.3224551677703857,
      "learning_rate": 5.926278248328257e-07,
      "loss": 0.4399,
      "step": 381180
    },
    {
      "epoch": 156.03,
      "grad_norm": 2.1537764072418213,
      "learning_rate": 5.925223940388215e-07,
      "loss": 0.456,
      "step": 381190
    },
    {
      "epoch": 156.04,
      "grad_norm": 1.82966148853302,
      "learning_rate": 5.924169712669203e-07,
      "loss": 0.4408,
      "step": 381200
    },
    {
      "epoch": 156.04,
      "grad_norm": 2.436647891998291,
      "learning_rate": 5.923115565176045e-07,
      "loss": 0.4725,
      "step": 381210
    },
    {
      "epoch": 156.05,
      "grad_norm": 2.0015289783477783,
      "learning_rate": 5.922061497913568e-07,
      "loss": 0.4445,
      "step": 381220
    },
    {
      "epoch": 156.05,
      "grad_norm": 2.168663263320923,
      "learning_rate": 5.921007510886603e-07,
      "loss": 0.4433,
      "step": 381230
    },
    {
      "epoch": 156.05,
      "grad_norm": 1.9819846153259277,
      "learning_rate": 5.919953604099978e-07,
      "loss": 0.4502,
      "step": 381240
    },
    {
      "epoch": 156.06,
      "grad_norm": 1.6822346448898315,
      "learning_rate": 5.918899777558518e-07,
      "loss": 0.4346,
      "step": 381250
    },
    {
      "epoch": 156.06,
      "grad_norm": 1.7227637767791748,
      "learning_rate": 5.917846031267055e-07,
      "loss": 0.4372,
      "step": 381260
    },
    {
      "epoch": 156.07,
      "grad_norm": 2.3599953651428223,
      "learning_rate": 5.916792365230412e-07,
      "loss": 0.429,
      "step": 381270
    },
    {
      "epoch": 156.07,
      "grad_norm": 2.159135580062866,
      "learning_rate": 5.91573877945342e-07,
      "loss": 0.4512,
      "step": 381280
    },
    {
      "epoch": 156.07,
      "grad_norm": 2.06477952003479,
      "learning_rate": 5.9146852739409e-07,
      "loss": 0.4608,
      "step": 381290
    },
    {
      "epoch": 156.08,
      "grad_norm": 2.1733665466308594,
      "learning_rate": 5.913631848697687e-07,
      "loss": 0.4383,
      "step": 381300
    },
    {
      "epoch": 156.08,
      "grad_norm": 2.35857892036438,
      "learning_rate": 5.912578503728595e-07,
      "loss": 0.469,
      "step": 381310
    },
    {
      "epoch": 156.09,
      "grad_norm": 2.415229320526123,
      "learning_rate": 5.911525239038452e-07,
      "loss": 0.4616,
      "step": 381320
    },
    {
      "epoch": 156.09,
      "grad_norm": 2.4618568420410156,
      "learning_rate": 5.910472054632087e-07,
      "loss": 0.4774,
      "step": 381330
    },
    {
      "epoch": 156.09,
      "grad_norm": 2.1986985206604004,
      "learning_rate": 5.909418950514323e-07,
      "loss": 0.4315,
      "step": 381340
    },
    {
      "epoch": 156.1,
      "grad_norm": 1.9297868013381958,
      "learning_rate": 5.908365926689983e-07,
      "loss": 0.4525,
      "step": 381350
    },
    {
      "epoch": 156.1,
      "grad_norm": 2.4855456352233887,
      "learning_rate": 5.907312983163892e-07,
      "loss": 0.4369,
      "step": 381360
    },
    {
      "epoch": 156.11,
      "grad_norm": 2.074244260787964,
      "learning_rate": 5.906260119940872e-07,
      "loss": 0.4443,
      "step": 381370
    },
    {
      "epoch": 156.11,
      "grad_norm": 2.077179193496704,
      "learning_rate": 5.905207337025737e-07,
      "loss": 0.4478,
      "step": 381380
    },
    {
      "epoch": 156.12,
      "grad_norm": 1.8257766962051392,
      "learning_rate": 5.904154634423331e-07,
      "loss": 0.4174,
      "step": 381390
    },
    {
      "epoch": 156.12,
      "grad_norm": 1.8593037128448486,
      "learning_rate": 5.903102012138463e-07,
      "loss": 0.4407,
      "step": 381400
    },
    {
      "epoch": 156.12,
      "grad_norm": 1.952722191810608,
      "learning_rate": 5.902049470175957e-07,
      "loss": 0.4362,
      "step": 381410
    },
    {
      "epoch": 156.13,
      "grad_norm": 1.9890552759170532,
      "learning_rate": 5.900997008540636e-07,
      "loss": 0.4306,
      "step": 381420
    },
    {
      "epoch": 156.13,
      "grad_norm": 2.4287219047546387,
      "learning_rate": 5.899944627237323e-07,
      "loss": 0.4465,
      "step": 381430
    },
    {
      "epoch": 156.14,
      "grad_norm": 2.15934681892395,
      "learning_rate": 5.898892326270829e-07,
      "loss": 0.4319,
      "step": 381440
    },
    {
      "epoch": 156.14,
      "grad_norm": 2.041130781173706,
      "learning_rate": 5.897840105645981e-07,
      "loss": 0.4565,
      "step": 381450
    },
    {
      "epoch": 156.14,
      "grad_norm": 2.250511646270752,
      "learning_rate": 5.896787965367598e-07,
      "loss": 0.4222,
      "step": 381460
    },
    {
      "epoch": 156.15,
      "grad_norm": 1.9479074478149414,
      "learning_rate": 5.895735905440501e-07,
      "loss": 0.4263,
      "step": 381470
    },
    {
      "epoch": 156.15,
      "grad_norm": 2.484132766723633,
      "learning_rate": 5.89468392586951e-07,
      "loss": 0.4359,
      "step": 381480
    },
    {
      "epoch": 156.16,
      "grad_norm": 1.8804980516433716,
      "learning_rate": 5.893632026659441e-07,
      "loss": 0.4454,
      "step": 381490
    },
    {
      "epoch": 156.16,
      "grad_norm": 1.8712552785873413,
      "learning_rate": 5.892580207815114e-07,
      "loss": 0.4633,
      "step": 381500
    },
    {
      "epoch": 156.16,
      "grad_norm": 1.7879875898361206,
      "learning_rate": 5.891528469341347e-07,
      "loss": 0.4424,
      "step": 381510
    },
    {
      "epoch": 156.17,
      "grad_norm": 2.5150351524353027,
      "learning_rate": 5.89047681124296e-07,
      "loss": 0.4427,
      "step": 381520
    },
    {
      "epoch": 156.17,
      "grad_norm": 1.8367573022842407,
      "learning_rate": 5.889425233524767e-07,
      "loss": 0.4487,
      "step": 381530
    },
    {
      "epoch": 156.18,
      "grad_norm": 2.166956901550293,
      "learning_rate": 5.888373736191587e-07,
      "loss": 0.4431,
      "step": 381540
    },
    {
      "epoch": 156.18,
      "grad_norm": 2.6361091136932373,
      "learning_rate": 5.887322319248239e-07,
      "loss": 0.4398,
      "step": 381550
    },
    {
      "epoch": 156.19,
      "grad_norm": 1.8966023921966553,
      "learning_rate": 5.886270982699534e-07,
      "loss": 0.463,
      "step": 381560
    },
    {
      "epoch": 156.19,
      "grad_norm": 1.9544025659561157,
      "learning_rate": 5.885219726550291e-07,
      "loss": 0.4435,
      "step": 381570
    },
    {
      "epoch": 156.19,
      "grad_norm": 2.054589033126831,
      "learning_rate": 5.884168550805327e-07,
      "loss": 0.437,
      "step": 381580
    },
    {
      "epoch": 156.2,
      "grad_norm": 1.7396442890167236,
      "learning_rate": 5.883117455469457e-07,
      "loss": 0.4427,
      "step": 381590
    },
    {
      "epoch": 156.2,
      "grad_norm": 2.1822643280029297,
      "learning_rate": 5.882066440547494e-07,
      "loss": 0.4593,
      "step": 381600
    },
    {
      "epoch": 156.21,
      "grad_norm": 2.108795166015625,
      "learning_rate": 5.881015506044252e-07,
      "loss": 0.4254,
      "step": 381610
    },
    {
      "epoch": 156.21,
      "grad_norm": 1.9658429622650146,
      "learning_rate": 5.879964651964548e-07,
      "loss": 0.4551,
      "step": 381620
    },
    {
      "epoch": 156.21,
      "grad_norm": 2.561866283416748,
      "learning_rate": 5.878913878313192e-07,
      "loss": 0.4516,
      "step": 381630
    },
    {
      "epoch": 156.22,
      "grad_norm": 1.9355900287628174,
      "learning_rate": 5.877863185095002e-07,
      "loss": 0.46,
      "step": 381640
    },
    {
      "epoch": 156.22,
      "grad_norm": 1.7164627313613892,
      "learning_rate": 5.876812572314788e-07,
      "loss": 0.4421,
      "step": 381650
    },
    {
      "epoch": 156.23,
      "grad_norm": 1.9195998907089233,
      "learning_rate": 5.875762039977365e-07,
      "loss": 0.4412,
      "step": 381660
    },
    {
      "epoch": 156.23,
      "grad_norm": 1.9751046895980835,
      "learning_rate": 5.874711588087541e-07,
      "loss": 0.4471,
      "step": 381670
    },
    {
      "epoch": 156.23,
      "grad_norm": 2.517195224761963,
      "learning_rate": 5.873661216650133e-07,
      "loss": 0.4534,
      "step": 381680
    },
    {
      "epoch": 156.24,
      "grad_norm": 2.252410411834717,
      "learning_rate": 5.872610925669948e-07,
      "loss": 0.4721,
      "step": 381690
    },
    {
      "epoch": 156.24,
      "grad_norm": 2.3834025859832764,
      "learning_rate": 5.871560715151802e-07,
      "loss": 0.4325,
      "step": 381700
    },
    {
      "epoch": 156.25,
      "grad_norm": 2.1118264198303223,
      "learning_rate": 5.870510585100503e-07,
      "loss": 0.4571,
      "step": 381710
    },
    {
      "epoch": 156.25,
      "grad_norm": 2.239468574523926,
      "learning_rate": 5.869460535520861e-07,
      "loss": 0.4524,
      "step": 381720
    },
    {
      "epoch": 156.25,
      "grad_norm": 1.9701998233795166,
      "learning_rate": 5.868410566417689e-07,
      "loss": 0.4271,
      "step": 381730
    },
    {
      "epoch": 156.26,
      "grad_norm": 2.068619966506958,
      "learning_rate": 5.8673606777958e-07,
      "loss": 0.4271,
      "step": 381740
    },
    {
      "epoch": 156.26,
      "grad_norm": 2.071634531021118,
      "learning_rate": 5.86631086965999e-07,
      "loss": 0.4406,
      "step": 381750
    },
    {
      "epoch": 156.27,
      "grad_norm": 2.2092437744140625,
      "learning_rate": 5.865261142015072e-07,
      "loss": 0.4526,
      "step": 381760
    },
    {
      "epoch": 156.27,
      "grad_norm": 2.576369524002075,
      "learning_rate": 5.864211494865863e-07,
      "loss": 0.4416,
      "step": 381770
    },
    {
      "epoch": 156.28,
      "grad_norm": 2.5110301971435547,
      "learning_rate": 5.86316192821717e-07,
      "loss": 0.451,
      "step": 381780
    },
    {
      "epoch": 156.28,
      "grad_norm": 2.0937538146972656,
      "learning_rate": 5.862112442073794e-07,
      "loss": 0.4496,
      "step": 381790
    },
    {
      "epoch": 156.28,
      "grad_norm": 1.751162052154541,
      "learning_rate": 5.861063036440552e-07,
      "loss": 0.445,
      "step": 381800
    },
    {
      "epoch": 156.29,
      "grad_norm": 2.6664862632751465,
      "learning_rate": 5.860013711322243e-07,
      "loss": 0.4391,
      "step": 381810
    },
    {
      "epoch": 156.29,
      "grad_norm": 2.2723660469055176,
      "learning_rate": 5.858964466723676e-07,
      "loss": 0.4363,
      "step": 381820
    },
    {
      "epoch": 156.3,
      "grad_norm": 1.951540470123291,
      "learning_rate": 5.857915302649656e-07,
      "loss": 0.458,
      "step": 381830
    },
    {
      "epoch": 156.3,
      "grad_norm": 2.318586826324463,
      "learning_rate": 5.856866219104992e-07,
      "loss": 0.4385,
      "step": 381840
    },
    {
      "epoch": 156.3,
      "grad_norm": 2.2914376258850098,
      "learning_rate": 5.855817216094492e-07,
      "loss": 0.432,
      "step": 381850
    },
    {
      "epoch": 156.31,
      "grad_norm": 2.3731229305267334,
      "learning_rate": 5.854768293622956e-07,
      "loss": 0.4459,
      "step": 381860
    },
    {
      "epoch": 156.31,
      "grad_norm": 1.8569817543029785,
      "learning_rate": 5.853719451695188e-07,
      "loss": 0.4368,
      "step": 381870
    },
    {
      "epoch": 156.32,
      "grad_norm": 2.3512895107269287,
      "learning_rate": 5.852670690316005e-07,
      "loss": 0.4568,
      "step": 381880
    },
    {
      "epoch": 156.32,
      "grad_norm": 2.3185908794403076,
      "learning_rate": 5.851622009490195e-07,
      "loss": 0.4389,
      "step": 381890
    },
    {
      "epoch": 156.32,
      "grad_norm": 2.4195175170898438,
      "learning_rate": 5.850573409222571e-07,
      "loss": 0.4503,
      "step": 381900
    },
    {
      "epoch": 156.33,
      "grad_norm": 2.0536224842071533,
      "learning_rate": 5.84952488951793e-07,
      "loss": 0.4519,
      "step": 381910
    },
    {
      "epoch": 156.33,
      "grad_norm": 2.2385356426239014,
      "learning_rate": 5.84847645038108e-07,
      "loss": 0.4426,
      "step": 381920
    },
    {
      "epoch": 156.34,
      "grad_norm": 2.0081775188446045,
      "learning_rate": 5.847428091816824e-07,
      "loss": 0.4649,
      "step": 381930
    },
    {
      "epoch": 156.34,
      "grad_norm": 2.322146415710449,
      "learning_rate": 5.846379813829963e-07,
      "loss": 0.462,
      "step": 381940
    },
    {
      "epoch": 156.34,
      "grad_norm": 1.8641096353530884,
      "learning_rate": 5.84533161642529e-07,
      "loss": 0.4566,
      "step": 381950
    },
    {
      "epoch": 156.35,
      "grad_norm": 2.0329177379608154,
      "learning_rate": 5.844283499607626e-07,
      "loss": 0.4349,
      "step": 381960
    },
    {
      "epoch": 156.35,
      "grad_norm": 2.203266143798828,
      "learning_rate": 5.843235463381761e-07,
      "loss": 0.4495,
      "step": 381970
    },
    {
      "epoch": 156.36,
      "grad_norm": 2.595855951309204,
      "learning_rate": 5.842187507752498e-07,
      "loss": 0.4348,
      "step": 381980
    },
    {
      "epoch": 156.36,
      "grad_norm": 2.0493688583374023,
      "learning_rate": 5.841139632724635e-07,
      "loss": 0.4523,
      "step": 381990
    },
    {
      "epoch": 156.37,
      "grad_norm": 2.126431941986084,
      "learning_rate": 5.840091838302973e-07,
      "loss": 0.45,
      "step": 382000
    },
    {
      "epoch": 156.37,
      "grad_norm": 1.9756978750228882,
      "learning_rate": 5.839044124492321e-07,
      "loss": 0.4455,
      "step": 382010
    },
    {
      "epoch": 156.37,
      "grad_norm": 1.7344084978103638,
      "learning_rate": 5.837996491297462e-07,
      "loss": 0.4593,
      "step": 382020
    },
    {
      "epoch": 156.38,
      "grad_norm": 2.0518906116485596,
      "learning_rate": 5.836948938723204e-07,
      "loss": 0.4384,
      "step": 382030
    },
    {
      "epoch": 156.38,
      "grad_norm": 2.1221814155578613,
      "learning_rate": 5.835901466774343e-07,
      "loss": 0.4408,
      "step": 382040
    },
    {
      "epoch": 156.39,
      "grad_norm": 2.0740039348602295,
      "learning_rate": 5.834854075455681e-07,
      "loss": 0.4636,
      "step": 382050
    },
    {
      "epoch": 156.39,
      "grad_norm": 1.6905126571655273,
      "learning_rate": 5.833806764772014e-07,
      "loss": 0.4284,
      "step": 382060
    },
    {
      "epoch": 156.39,
      "grad_norm": 1.8172141313552856,
      "learning_rate": 5.832759534728136e-07,
      "loss": 0.4407,
      "step": 382070
    },
    {
      "epoch": 156.4,
      "grad_norm": 2.172018051147461,
      "learning_rate": 5.831712385328851e-07,
      "loss": 0.4366,
      "step": 382080
    },
    {
      "epoch": 156.4,
      "grad_norm": 1.586693286895752,
      "learning_rate": 5.830665316578952e-07,
      "loss": 0.4199,
      "step": 382090
    },
    {
      "epoch": 156.41,
      "grad_norm": 2.72710919380188,
      "learning_rate": 5.829618328483236e-07,
      "loss": 0.4499,
      "step": 382100
    },
    {
      "epoch": 156.41,
      "grad_norm": 2.136892557144165,
      "learning_rate": 5.828571421046497e-07,
      "loss": 0.4489,
      "step": 382110
    },
    {
      "epoch": 156.41,
      "grad_norm": 2.012478828430176,
      "learning_rate": 5.827524594273533e-07,
      "loss": 0.458,
      "step": 382120
    },
    {
      "epoch": 156.42,
      "grad_norm": 1.9191391468048096,
      "learning_rate": 5.826477848169135e-07,
      "loss": 0.456,
      "step": 382130
    },
    {
      "epoch": 156.42,
      "grad_norm": 1.8069919347763062,
      "learning_rate": 5.825431182738107e-07,
      "loss": 0.4367,
      "step": 382140
    },
    {
      "epoch": 156.43,
      "grad_norm": 3.978034257888794,
      "learning_rate": 5.824384597985247e-07,
      "loss": 0.4344,
      "step": 382150
    },
    {
      "epoch": 156.43,
      "grad_norm": 2.0063090324401855,
      "learning_rate": 5.823338093915332e-07,
      "loss": 0.4299,
      "step": 382160
    },
    {
      "epoch": 156.43,
      "grad_norm": 1.9393268823623657,
      "learning_rate": 5.822291670533166e-07,
      "loss": 0.4443,
      "step": 382170
    },
    {
      "epoch": 156.44,
      "grad_norm": 2.207324981689453,
      "learning_rate": 5.82124532784354e-07,
      "loss": 0.4335,
      "step": 382180
    },
    {
      "epoch": 156.44,
      "grad_norm": 2.1119771003723145,
      "learning_rate": 5.82019906585125e-07,
      "loss": 0.4618,
      "step": 382190
    },
    {
      "epoch": 156.45,
      "grad_norm": 1.9538460969924927,
      "learning_rate": 5.819152884561087e-07,
      "loss": 0.4526,
      "step": 382200
    },
    {
      "epoch": 156.45,
      "grad_norm": 2.078139305114746,
      "learning_rate": 5.818106783977843e-07,
      "loss": 0.4483,
      "step": 382210
    },
    {
      "epoch": 156.46,
      "grad_norm": 1.925278902053833,
      "learning_rate": 5.817060764106311e-07,
      "loss": 0.4545,
      "step": 382220
    },
    {
      "epoch": 156.46,
      "grad_norm": 2.0404553413391113,
      "learning_rate": 5.816014824951282e-07,
      "loss": 0.4412,
      "step": 382230
    },
    {
      "epoch": 156.46,
      "grad_norm": 1.7741875648498535,
      "learning_rate": 5.814968966517547e-07,
      "loss": 0.4238,
      "step": 382240
    },
    {
      "epoch": 156.47,
      "grad_norm": 2.033982992172241,
      "learning_rate": 5.813923188809897e-07,
      "loss": 0.4435,
      "step": 382250
    },
    {
      "epoch": 156.47,
      "grad_norm": 1.8579127788543701,
      "learning_rate": 5.812877491833125e-07,
      "loss": 0.4728,
      "step": 382260
    },
    {
      "epoch": 156.48,
      "grad_norm": 1.989048719406128,
      "learning_rate": 5.811831875592019e-07,
      "loss": 0.4489,
      "step": 382270
    },
    {
      "epoch": 156.48,
      "grad_norm": 2.2429144382476807,
      "learning_rate": 5.810786340091368e-07,
      "loss": 0.453,
      "step": 382280
    },
    {
      "epoch": 156.48,
      "grad_norm": 2.1381077766418457,
      "learning_rate": 5.809740885335964e-07,
      "loss": 0.4352,
      "step": 382290
    },
    {
      "epoch": 156.49,
      "grad_norm": 2.3287439346313477,
      "learning_rate": 5.808695511330591e-07,
      "loss": 0.4282,
      "step": 382300
    },
    {
      "epoch": 156.49,
      "grad_norm": 1.8258227109909058,
      "learning_rate": 5.807650218080043e-07,
      "loss": 0.4653,
      "step": 382310
    },
    {
      "epoch": 156.5,
      "grad_norm": 1.71967613697052,
      "learning_rate": 5.806605005589107e-07,
      "loss": 0.4623,
      "step": 382320
    },
    {
      "epoch": 156.5,
      "grad_norm": 2.208325147628784,
      "learning_rate": 5.805559873862571e-07,
      "loss": 0.4443,
      "step": 382330
    },
    {
      "epoch": 156.5,
      "grad_norm": 1.9672911167144775,
      "learning_rate": 5.804514822905221e-07,
      "loss": 0.4536,
      "step": 382340
    },
    {
      "epoch": 156.51,
      "grad_norm": 2.1449923515319824,
      "learning_rate": 5.803469852721848e-07,
      "loss": 0.4321,
      "step": 382350
    },
    {
      "epoch": 156.51,
      "grad_norm": 2.137787103652954,
      "learning_rate": 5.802424963317234e-07,
      "loss": 0.4514,
      "step": 382360
    },
    {
      "epoch": 156.52,
      "grad_norm": 1.9978854656219482,
      "learning_rate": 5.801380154696166e-07,
      "loss": 0.4442,
      "step": 382370
    },
    {
      "epoch": 156.52,
      "grad_norm": 2.124725580215454,
      "learning_rate": 5.800335426863434e-07,
      "loss": 0.4316,
      "step": 382380
    },
    {
      "epoch": 156.52,
      "grad_norm": 1.981078863143921,
      "learning_rate": 5.79929077982382e-07,
      "loss": 0.4509,
      "step": 382390
    },
    {
      "epoch": 156.53,
      "grad_norm": 2.141331911087036,
      "learning_rate": 5.798246213582111e-07,
      "loss": 0.423,
      "step": 382400
    },
    {
      "epoch": 156.53,
      "grad_norm": 1.8790336847305298,
      "learning_rate": 5.797201728143094e-07,
      "loss": 0.4551,
      "step": 382410
    },
    {
      "epoch": 156.54,
      "grad_norm": 2.192267656326294,
      "learning_rate": 5.79615732351155e-07,
      "loss": 0.4617,
      "step": 382420
    },
    {
      "epoch": 156.54,
      "grad_norm": 2.0833945274353027,
      "learning_rate": 5.795112999692264e-07,
      "loss": 0.4348,
      "step": 382430
    },
    {
      "epoch": 156.55,
      "grad_norm": 1.8372132778167725,
      "learning_rate": 5.794068756690022e-07,
      "loss": 0.481,
      "step": 382440
    },
    {
      "epoch": 156.55,
      "grad_norm": 2.028548240661621,
      "learning_rate": 5.793024594509604e-07,
      "loss": 0.4424,
      "step": 382450
    },
    {
      "epoch": 156.55,
      "grad_norm": 1.9943740367889404,
      "learning_rate": 5.791980513155804e-07,
      "loss": 0.4555,
      "step": 382460
    },
    {
      "epoch": 156.56,
      "grad_norm": 1.690572738647461,
      "learning_rate": 5.790936512633389e-07,
      "loss": 0.451,
      "step": 382470
    },
    {
      "epoch": 156.56,
      "grad_norm": 2.2094929218292236,
      "learning_rate": 5.789892592947147e-07,
      "loss": 0.4792,
      "step": 382480
    },
    {
      "epoch": 156.57,
      "grad_norm": 2.2661805152893066,
      "learning_rate": 5.788848754101862e-07,
      "loss": 0.4527,
      "step": 382490
    },
    {
      "epoch": 156.57,
      "grad_norm": 2.3165910243988037,
      "learning_rate": 5.787804996102314e-07,
      "loss": 0.4385,
      "step": 382500
    },
    {
      "epoch": 156.57,
      "grad_norm": 2.1748952865600586,
      "learning_rate": 5.786761318953281e-07,
      "loss": 0.4719,
      "step": 382510
    },
    {
      "epoch": 156.58,
      "grad_norm": 1.9372577667236328,
      "learning_rate": 5.785717722659554e-07,
      "loss": 0.449,
      "step": 382520
    },
    {
      "epoch": 156.58,
      "grad_norm": 2.2403008937835693,
      "learning_rate": 5.784674207225907e-07,
      "loss": 0.4446,
      "step": 382530
    },
    {
      "epoch": 156.59,
      "grad_norm": 2.180959463119507,
      "learning_rate": 5.783630772657123e-07,
      "loss": 0.4462,
      "step": 382540
    },
    {
      "epoch": 156.59,
      "grad_norm": 2.371366500854492,
      "learning_rate": 5.78258741895798e-07,
      "loss": 0.445,
      "step": 382550
    },
    {
      "epoch": 156.59,
      "grad_norm": 2.107926368713379,
      "learning_rate": 5.781544146133256e-07,
      "loss": 0.4582,
      "step": 382560
    },
    {
      "epoch": 156.6,
      "grad_norm": 2.3437438011169434,
      "learning_rate": 5.780500954187731e-07,
      "loss": 0.4447,
      "step": 382570
    },
    {
      "epoch": 156.6,
      "grad_norm": 1.753471851348877,
      "learning_rate": 5.779457843126185e-07,
      "loss": 0.4355,
      "step": 382580
    },
    {
      "epoch": 156.61,
      "grad_norm": 1.801328420639038,
      "learning_rate": 5.778414812953394e-07,
      "loss": 0.4592,
      "step": 382590
    },
    {
      "epoch": 156.61,
      "grad_norm": 2.180832624435425,
      "learning_rate": 5.777371863674147e-07,
      "loss": 0.4442,
      "step": 382600
    },
    {
      "epoch": 156.61,
      "grad_norm": 2.783949136734009,
      "learning_rate": 5.776328995293204e-07,
      "loss": 0.4479,
      "step": 382610
    },
    {
      "epoch": 156.62,
      "grad_norm": 2.631192922592163,
      "learning_rate": 5.77528620781535e-07,
      "loss": 0.4554,
      "step": 382620
    },
    {
      "epoch": 156.62,
      "grad_norm": 1.7276414632797241,
      "learning_rate": 5.774243501245362e-07,
      "loss": 0.4496,
      "step": 382630
    },
    {
      "epoch": 156.63,
      "grad_norm": 2.1898586750030518,
      "learning_rate": 5.773200875588015e-07,
      "loss": 0.4417,
      "step": 382640
    },
    {
      "epoch": 156.63,
      "grad_norm": 2.428436279296875,
      "learning_rate": 5.77215833084809e-07,
      "loss": 0.4589,
      "step": 382650
    },
    {
      "epoch": 156.64,
      "grad_norm": 1.761916160583496,
      "learning_rate": 5.771115867030355e-07,
      "loss": 0.4513,
      "step": 382660
    },
    {
      "epoch": 156.64,
      "grad_norm": 2.203791856765747,
      "learning_rate": 5.770073484139593e-07,
      "loss": 0.4409,
      "step": 382670
    },
    {
      "epoch": 156.64,
      "grad_norm": 1.9806722402572632,
      "learning_rate": 5.769031182180574e-07,
      "loss": 0.4545,
      "step": 382680
    },
    {
      "epoch": 156.65,
      "grad_norm": 2.0352847576141357,
      "learning_rate": 5.767988961158077e-07,
      "loss": 0.4461,
      "step": 382690
    },
    {
      "epoch": 156.65,
      "grad_norm": 2.230278491973877,
      "learning_rate": 5.766946821076864e-07,
      "loss": 0.4494,
      "step": 382700
    },
    {
      "epoch": 156.66,
      "grad_norm": 2.1765034198760986,
      "learning_rate": 5.765904761941727e-07,
      "loss": 0.4409,
      "step": 382710
    },
    {
      "epoch": 156.66,
      "grad_norm": 3.8698387145996094,
      "learning_rate": 5.764862783757429e-07,
      "loss": 0.4324,
      "step": 382720
    },
    {
      "epoch": 156.66,
      "grad_norm": 2.239316940307617,
      "learning_rate": 5.763820886528753e-07,
      "loss": 0.4555,
      "step": 382730
    },
    {
      "epoch": 156.67,
      "grad_norm": 1.9138704538345337,
      "learning_rate": 5.762779070260457e-07,
      "loss": 0.4527,
      "step": 382740
    },
    {
      "epoch": 156.67,
      "grad_norm": 1.8951431512832642,
      "learning_rate": 5.761737334957321e-07,
      "loss": 0.4534,
      "step": 382750
    },
    {
      "epoch": 156.68,
      "grad_norm": 1.7977628707885742,
      "learning_rate": 5.760695680624117e-07,
      "loss": 0.4478,
      "step": 382760
    },
    {
      "epoch": 156.68,
      "grad_norm": 1.5781644582748413,
      "learning_rate": 5.759654107265614e-07,
      "loss": 0.4192,
      "step": 382770
    },
    {
      "epoch": 156.68,
      "grad_norm": 1.7923572063446045,
      "learning_rate": 5.758612614886585e-07,
      "loss": 0.4473,
      "step": 382780
    },
    {
      "epoch": 156.69,
      "grad_norm": 3.0118954181671143,
      "learning_rate": 5.757571203491804e-07,
      "loss": 0.4606,
      "step": 382790
    },
    {
      "epoch": 156.69,
      "grad_norm": 2.6535089015960693,
      "learning_rate": 5.756529873086037e-07,
      "loss": 0.4521,
      "step": 382800
    },
    {
      "epoch": 156.7,
      "grad_norm": 2.0761067867279053,
      "learning_rate": 5.755488623674056e-07,
      "loss": 0.4456,
      "step": 382810
    },
    {
      "epoch": 156.7,
      "grad_norm": 2.107117176055908,
      "learning_rate": 5.754447455260631e-07,
      "loss": 0.4631,
      "step": 382820
    },
    {
      "epoch": 156.7,
      "grad_norm": 2.061868667602539,
      "learning_rate": 5.753406367850529e-07,
      "loss": 0.4537,
      "step": 382830
    },
    {
      "epoch": 156.71,
      "grad_norm": 1.7105741500854492,
      "learning_rate": 5.752365361448525e-07,
      "loss": 0.4499,
      "step": 382840
    },
    {
      "epoch": 156.71,
      "grad_norm": 2.0612244606018066,
      "learning_rate": 5.75132443605938e-07,
      "loss": 0.4559,
      "step": 382850
    },
    {
      "epoch": 156.72,
      "grad_norm": 1.9471222162246704,
      "learning_rate": 5.750283591687868e-07,
      "loss": 0.4489,
      "step": 382860
    },
    {
      "epoch": 156.72,
      "grad_norm": 2.166550874710083,
      "learning_rate": 5.749242828338757e-07,
      "loss": 0.4721,
      "step": 382870
    },
    {
      "epoch": 156.73,
      "grad_norm": 1.9361587762832642,
      "learning_rate": 5.748202146016811e-07,
      "loss": 0.4463,
      "step": 382880
    },
    {
      "epoch": 156.73,
      "grad_norm": 2.1787731647491455,
      "learning_rate": 5.7471615447268e-07,
      "loss": 0.4572,
      "step": 382890
    },
    {
      "epoch": 156.73,
      "grad_norm": 2.0636513233184814,
      "learning_rate": 5.746121024473488e-07,
      "loss": 0.429,
      "step": 382900
    },
    {
      "epoch": 156.74,
      "grad_norm": 2.1644186973571777,
      "learning_rate": 5.745080585261643e-07,
      "loss": 0.4512,
      "step": 382910
    },
    {
      "epoch": 156.74,
      "grad_norm": 2.031944990158081,
      "learning_rate": 5.744040227096033e-07,
      "loss": 0.4628,
      "step": 382920
    },
    {
      "epoch": 156.75,
      "grad_norm": 1.907801866531372,
      "learning_rate": 5.742999949981421e-07,
      "loss": 0.4542,
      "step": 382930
    },
    {
      "epoch": 156.75,
      "grad_norm": 2.1367807388305664,
      "learning_rate": 5.741959753922575e-07,
      "loss": 0.4465,
      "step": 382940
    },
    {
      "epoch": 156.75,
      "grad_norm": 2.17740535736084,
      "learning_rate": 5.740919638924256e-07,
      "loss": 0.4596,
      "step": 382950
    },
    {
      "epoch": 156.76,
      "grad_norm": 2.0558011531829834,
      "learning_rate": 5.739879604991234e-07,
      "loss": 0.4479,
      "step": 382960
    },
    {
      "epoch": 156.76,
      "grad_norm": 2.207810163497925,
      "learning_rate": 5.738839652128265e-07,
      "loss": 0.4436,
      "step": 382970
    },
    {
      "epoch": 156.77,
      "grad_norm": 2.586461305618286,
      "learning_rate": 5.737799780340124e-07,
      "loss": 0.4579,
      "step": 382980
    },
    {
      "epoch": 156.77,
      "grad_norm": 2.0248169898986816,
      "learning_rate": 5.736759989631567e-07,
      "loss": 0.4396,
      "step": 382990
    },
    {
      "epoch": 156.77,
      "grad_norm": 2.1313793659210205,
      "learning_rate": 5.735720280007356e-07,
      "loss": 0.4566,
      "step": 383000
    },
    {
      "epoch": 156.78,
      "grad_norm": 2.8670809268951416,
      "learning_rate": 5.734680651472257e-07,
      "loss": 0.4385,
      "step": 383010
    },
    {
      "epoch": 156.78,
      "grad_norm": 2.155578374862671,
      "learning_rate": 5.733641104031032e-07,
      "loss": 0.4336,
      "step": 383020
    },
    {
      "epoch": 156.79,
      "grad_norm": 1.949585199356079,
      "learning_rate": 5.732601637688446e-07,
      "loss": 0.4484,
      "step": 383030
    },
    {
      "epoch": 156.79,
      "grad_norm": 1.968177318572998,
      "learning_rate": 5.731562252449262e-07,
      "loss": 0.4531,
      "step": 383040
    },
    {
      "epoch": 156.79,
      "grad_norm": 2.377894163131714,
      "learning_rate": 5.73052294831823e-07,
      "loss": 0.441,
      "step": 383050
    },
    {
      "epoch": 156.8,
      "grad_norm": 1.712747573852539,
      "learning_rate": 5.729483725300116e-07,
      "loss": 0.4645,
      "step": 383060
    },
    {
      "epoch": 156.8,
      "grad_norm": 1.7623648643493652,
      "learning_rate": 5.728444583399685e-07,
      "loss": 0.455,
      "step": 383070
    },
    {
      "epoch": 156.81,
      "grad_norm": 1.7983155250549316,
      "learning_rate": 5.727405522621685e-07,
      "loss": 0.4479,
      "step": 383080
    },
    {
      "epoch": 156.81,
      "grad_norm": 2.056093215942383,
      "learning_rate": 5.726366542970893e-07,
      "loss": 0.4259,
      "step": 383090
    },
    {
      "epoch": 156.82,
      "grad_norm": 2.9218294620513916,
      "learning_rate": 5.72532764445206e-07,
      "loss": 0.4574,
      "step": 383100
    },
    {
      "epoch": 156.82,
      "grad_norm": 2.2793524265289307,
      "learning_rate": 5.724288827069945e-07,
      "loss": 0.4697,
      "step": 383110
    },
    {
      "epoch": 156.82,
      "grad_norm": 2.1031789779663086,
      "learning_rate": 5.723250090829309e-07,
      "loss": 0.4517,
      "step": 383120
    },
    {
      "epoch": 156.83,
      "grad_norm": 1.8290247917175293,
      "learning_rate": 5.722211435734908e-07,
      "loss": 0.4441,
      "step": 383130
    },
    {
      "epoch": 156.83,
      "grad_norm": 2.0764243602752686,
      "learning_rate": 5.721172861791499e-07,
      "loss": 0.4512,
      "step": 383140
    },
    {
      "epoch": 156.84,
      "grad_norm": 1.8525617122650146,
      "learning_rate": 5.720134369003843e-07,
      "loss": 0.4498,
      "step": 383150
    },
    {
      "epoch": 156.84,
      "grad_norm": 2.0104517936706543,
      "learning_rate": 5.719095957376693e-07,
      "loss": 0.4493,
      "step": 383160
    },
    {
      "epoch": 156.84,
      "grad_norm": 2.032571792602539,
      "learning_rate": 5.718057626914809e-07,
      "loss": 0.4297,
      "step": 383170
    },
    {
      "epoch": 156.85,
      "grad_norm": 2.0204949378967285,
      "learning_rate": 5.717019377622954e-07,
      "loss": 0.447,
      "step": 383180
    },
    {
      "epoch": 156.85,
      "grad_norm": 2.565720558166504,
      "learning_rate": 5.715981209505866e-07,
      "loss": 0.4424,
      "step": 383190
    },
    {
      "epoch": 156.86,
      "grad_norm": 2.178859233856201,
      "learning_rate": 5.714943122568315e-07,
      "loss": 0.4512,
      "step": 383200
    },
    {
      "epoch": 156.86,
      "grad_norm": 1.8460863828659058,
      "learning_rate": 5.71390511681505e-07,
      "loss": 0.4444,
      "step": 383210
    },
    {
      "epoch": 156.86,
      "grad_norm": 2.2809455394744873,
      "learning_rate": 5.71286719225083e-07,
      "loss": 0.4747,
      "step": 383220
    },
    {
      "epoch": 156.87,
      "grad_norm": 1.767755389213562,
      "learning_rate": 5.711829348880405e-07,
      "loss": 0.443,
      "step": 383230
    },
    {
      "epoch": 156.87,
      "grad_norm": 2.111030101776123,
      "learning_rate": 5.710791586708535e-07,
      "loss": 0.4446,
      "step": 383240
    },
    {
      "epoch": 156.88,
      "grad_norm": 1.7543505430221558,
      "learning_rate": 5.70975390573997e-07,
      "loss": 0.4425,
      "step": 383250
    },
    {
      "epoch": 156.88,
      "grad_norm": 1.834415316581726,
      "learning_rate": 5.708716305979464e-07,
      "loss": 0.4603,
      "step": 383260
    },
    {
      "epoch": 156.88,
      "grad_norm": 2.18925404548645,
      "learning_rate": 5.707678787431764e-07,
      "loss": 0.4403,
      "step": 383270
    },
    {
      "epoch": 156.89,
      "grad_norm": 1.9669013023376465,
      "learning_rate": 5.706641350101636e-07,
      "loss": 0.4531,
      "step": 383280
    },
    {
      "epoch": 156.89,
      "grad_norm": 2.1638240814208984,
      "learning_rate": 5.705603993993825e-07,
      "loss": 0.4472,
      "step": 383290
    },
    {
      "epoch": 156.9,
      "grad_norm": 2.219231367111206,
      "learning_rate": 5.704566719113084e-07,
      "loss": 0.4423,
      "step": 383300
    },
    {
      "epoch": 156.9,
      "grad_norm": 1.743973731994629,
      "learning_rate": 5.703529525464164e-07,
      "loss": 0.4341,
      "step": 383310
    },
    {
      "epoch": 156.91,
      "grad_norm": 1.81306791305542,
      "learning_rate": 5.702492413051823e-07,
      "loss": 0.4543,
      "step": 383320
    },
    {
      "epoch": 156.91,
      "grad_norm": 2.153703212738037,
      "learning_rate": 5.701455381880798e-07,
      "loss": 0.4723,
      "step": 383330
    },
    {
      "epoch": 156.91,
      "grad_norm": 1.779676914215088,
      "learning_rate": 5.700418431955849e-07,
      "loss": 0.4372,
      "step": 383340
    },
    {
      "epoch": 156.92,
      "grad_norm": 2.1880571842193604,
      "learning_rate": 5.69938156328172e-07,
      "loss": 0.4672,
      "step": 383350
    },
    {
      "epoch": 156.92,
      "grad_norm": 2.1876211166381836,
      "learning_rate": 5.698344775863167e-07,
      "loss": 0.4475,
      "step": 383360
    },
    {
      "epoch": 156.93,
      "grad_norm": 1.9400608539581299,
      "learning_rate": 5.697308069704935e-07,
      "loss": 0.4465,
      "step": 383370
    },
    {
      "epoch": 156.93,
      "grad_norm": 1.703096628189087,
      "learning_rate": 5.696271444811775e-07,
      "loss": 0.4434,
      "step": 383380
    },
    {
      "epoch": 156.93,
      "grad_norm": 2.211144208908081,
      "learning_rate": 5.695234901188438e-07,
      "loss": 0.4529,
      "step": 383390
    },
    {
      "epoch": 156.94,
      "grad_norm": 2.2794878482818604,
      "learning_rate": 5.694198438839669e-07,
      "loss": 0.4617,
      "step": 383400
    },
    {
      "epoch": 156.94,
      "grad_norm": 2.026773691177368,
      "learning_rate": 5.693162057770215e-07,
      "loss": 0.4657,
      "step": 383410
    },
    {
      "epoch": 156.95,
      "grad_norm": 2.2371387481689453,
      "learning_rate": 5.692125757984825e-07,
      "loss": 0.449,
      "step": 383420
    },
    {
      "epoch": 156.95,
      "grad_norm": 2.206066608428955,
      "learning_rate": 5.691089539488246e-07,
      "loss": 0.4509,
      "step": 383430
    },
    {
      "epoch": 156.95,
      "grad_norm": 2.203441858291626,
      "learning_rate": 5.690053402285228e-07,
      "loss": 0.4279,
      "step": 383440
    },
    {
      "epoch": 156.96,
      "grad_norm": 1.9801809787750244,
      "learning_rate": 5.689017346380511e-07,
      "loss": 0.4485,
      "step": 383450
    },
    {
      "epoch": 156.96,
      "grad_norm": 2.961909055709839,
      "learning_rate": 5.687981371778844e-07,
      "loss": 0.4438,
      "step": 383460
    },
    {
      "epoch": 156.97,
      "grad_norm": 1.7113491296768188,
      "learning_rate": 5.686945478484976e-07,
      "loss": 0.435,
      "step": 383470
    },
    {
      "epoch": 156.97,
      "grad_norm": 1.915311336517334,
      "learning_rate": 5.685909666503646e-07,
      "loss": 0.4558,
      "step": 383480
    },
    {
      "epoch": 156.98,
      "grad_norm": 1.854203462600708,
      "learning_rate": 5.684873935839604e-07,
      "loss": 0.4473,
      "step": 383490
    },
    {
      "epoch": 156.98,
      "grad_norm": 2.8804779052734375,
      "learning_rate": 5.683838286497593e-07,
      "loss": 0.4505,
      "step": 383500
    },
    {
      "epoch": 156.98,
      "grad_norm": 1.8771679401397705,
      "learning_rate": 5.682802718482356e-07,
      "loss": 0.445,
      "step": 383510
    },
    {
      "epoch": 156.99,
      "grad_norm": 2.7415592670440674,
      "learning_rate": 5.681767231798639e-07,
      "loss": 0.453,
      "step": 383520
    },
    {
      "epoch": 156.99,
      "grad_norm": 2.032501459121704,
      "learning_rate": 5.680731826451182e-07,
      "loss": 0.4534,
      "step": 383530
    },
    {
      "epoch": 157.0,
      "grad_norm": 1.6119139194488525,
      "learning_rate": 5.679696502444731e-07,
      "loss": 0.4551,
      "step": 383540
    },
    {
      "epoch": 157.0,
      "grad_norm": 2.415947198867798,
      "learning_rate": 5.678661259784028e-07,
      "loss": 0.4509,
      "step": 383550
    },
    {
      "epoch": 157.0,
      "eval_loss": 0.4482097625732422,
      "eval_runtime": 52.4117,
      "eval_samples_per_second": 65.806,
      "eval_steps_per_second": 8.242,
      "step": 383551
    },
    {
      "epoch": 157.0,
      "grad_norm": 1.9440172910690308,
      "learning_rate": 5.677626098473814e-07,
      "loss": 0.4243,
      "step": 383560
    },
    {
      "epoch": 157.01,
      "grad_norm": 2.3423502445220947,
      "learning_rate": 5.676591018518831e-07,
      "loss": 0.4617,
      "step": 383570
    },
    {
      "epoch": 157.01,
      "grad_norm": 1.8112481832504272,
      "learning_rate": 5.675556019923824e-07,
      "loss": 0.4436,
      "step": 383580
    },
    {
      "epoch": 157.02,
      "grad_norm": 2.2467286586761475,
      "learning_rate": 5.67452110269353e-07,
      "loss": 0.4504,
      "step": 383590
    },
    {
      "epoch": 157.02,
      "grad_norm": 1.9162062406539917,
      "learning_rate": 5.673486266832693e-07,
      "loss": 0.4567,
      "step": 383600
    },
    {
      "epoch": 157.02,
      "grad_norm": 2.182730197906494,
      "learning_rate": 5.672451512346049e-07,
      "loss": 0.4797,
      "step": 383610
    },
    {
      "epoch": 157.03,
      "grad_norm": 3.540384531021118,
      "learning_rate": 5.67141683923835e-07,
      "loss": 0.4586,
      "step": 383620
    },
    {
      "epoch": 157.03,
      "grad_norm": 2.625145673751831,
      "learning_rate": 5.670382247514318e-07,
      "loss": 0.4366,
      "step": 383630
    },
    {
      "epoch": 157.04,
      "grad_norm": 1.9367786645889282,
      "learning_rate": 5.669347737178696e-07,
      "loss": 0.4457,
      "step": 383640
    },
    {
      "epoch": 157.04,
      "grad_norm": 2.008434534072876,
      "learning_rate": 5.668313308236235e-07,
      "loss": 0.4378,
      "step": 383650
    },
    {
      "epoch": 157.04,
      "grad_norm": 1.9802892208099365,
      "learning_rate": 5.667278960691664e-07,
      "loss": 0.4531,
      "step": 383660
    },
    {
      "epoch": 157.05,
      "grad_norm": 2.3590967655181885,
      "learning_rate": 5.666244694549725e-07,
      "loss": 0.4373,
      "step": 383670
    },
    {
      "epoch": 157.05,
      "grad_norm": 2.0871567726135254,
      "learning_rate": 5.665210509815153e-07,
      "loss": 0.4364,
      "step": 383680
    },
    {
      "epoch": 157.06,
      "grad_norm": 2.1986234188079834,
      "learning_rate": 5.664176406492686e-07,
      "loss": 0.4294,
      "step": 383690
    },
    {
      "epoch": 157.06,
      "grad_norm": 2.5558061599731445,
      "learning_rate": 5.663142384587063e-07,
      "loss": 0.4356,
      "step": 383700
    },
    {
      "epoch": 157.07,
      "grad_norm": 1.8758890628814697,
      "learning_rate": 5.662108444103022e-07,
      "loss": 0.4566,
      "step": 383710
    },
    {
      "epoch": 157.07,
      "grad_norm": 2.628307342529297,
      "learning_rate": 5.661074585045295e-07,
      "loss": 0.4639,
      "step": 383720
    },
    {
      "epoch": 157.07,
      "grad_norm": 2.014892339706421,
      "learning_rate": 5.660040807418618e-07,
      "loss": 0.4539,
      "step": 383730
    },
    {
      "epoch": 157.08,
      "grad_norm": 2.1587071418762207,
      "learning_rate": 5.65900711122773e-07,
      "loss": 0.4515,
      "step": 383740
    },
    {
      "epoch": 157.08,
      "grad_norm": 2.114068031311035,
      "learning_rate": 5.657973496477364e-07,
      "loss": 0.4492,
      "step": 383750
    },
    {
      "epoch": 157.09,
      "grad_norm": 2.1690618991851807,
      "learning_rate": 5.656939963172263e-07,
      "loss": 0.4525,
      "step": 383760
    },
    {
      "epoch": 157.09,
      "grad_norm": 1.9533125162124634,
      "learning_rate": 5.655906511317148e-07,
      "loss": 0.4292,
      "step": 383770
    },
    {
      "epoch": 157.09,
      "grad_norm": 1.9265925884246826,
      "learning_rate": 5.654873140916757e-07,
      "loss": 0.449,
      "step": 383780
    },
    {
      "epoch": 157.1,
      "grad_norm": 2.2621653079986572,
      "learning_rate": 5.653839851975826e-07,
      "loss": 0.4427,
      "step": 383790
    },
    {
      "epoch": 157.1,
      "grad_norm": 2.3277394771575928,
      "learning_rate": 5.652806644499089e-07,
      "loss": 0.4661,
      "step": 383800
    },
    {
      "epoch": 157.11,
      "grad_norm": 3.905146360397339,
      "learning_rate": 5.651773518491276e-07,
      "loss": 0.436,
      "step": 383810
    },
    {
      "epoch": 157.11,
      "grad_norm": 2.031348705291748,
      "learning_rate": 5.650740473957124e-07,
      "loss": 0.4411,
      "step": 383820
    },
    {
      "epoch": 157.11,
      "grad_norm": 1.997446060180664,
      "learning_rate": 5.649707510901355e-07,
      "loss": 0.4597,
      "step": 383830
    },
    {
      "epoch": 157.12,
      "grad_norm": 2.1945784091949463,
      "learning_rate": 5.648674629328714e-07,
      "loss": 0.451,
      "step": 383840
    },
    {
      "epoch": 157.12,
      "grad_norm": 2.054595947265625,
      "learning_rate": 5.647641829243928e-07,
      "loss": 0.4408,
      "step": 383850
    },
    {
      "epoch": 157.13,
      "grad_norm": 2.14275860786438,
      "learning_rate": 5.646609110651729e-07,
      "loss": 0.4408,
      "step": 383860
    },
    {
      "epoch": 157.13,
      "grad_norm": 2.435486316680908,
      "learning_rate": 5.645576473556843e-07,
      "loss": 0.4345,
      "step": 383870
    },
    {
      "epoch": 157.13,
      "grad_norm": 1.9813169240951538,
      "learning_rate": 5.644543917964006e-07,
      "loss": 0.4462,
      "step": 383880
    },
    {
      "epoch": 157.14,
      "grad_norm": 1.8707786798477173,
      "learning_rate": 5.643511443877943e-07,
      "loss": 0.4479,
      "step": 383890
    },
    {
      "epoch": 157.14,
      "grad_norm": 2.0927281379699707,
      "learning_rate": 5.642479051303394e-07,
      "loss": 0.4367,
      "step": 383900
    },
    {
      "epoch": 157.15,
      "grad_norm": 2.12850022315979,
      "learning_rate": 5.641446740245074e-07,
      "loss": 0.4423,
      "step": 383910
    },
    {
      "epoch": 157.15,
      "grad_norm": 1.9893479347229004,
      "learning_rate": 5.640414510707715e-07,
      "loss": 0.4547,
      "step": 383920
    },
    {
      "epoch": 157.16,
      "grad_norm": 1.762170672416687,
      "learning_rate": 5.639382362696052e-07,
      "loss": 0.4672,
      "step": 383930
    },
    {
      "epoch": 157.16,
      "grad_norm": 1.8717765808105469,
      "learning_rate": 5.638350296214806e-07,
      "loss": 0.439,
      "step": 383940
    },
    {
      "epoch": 157.16,
      "grad_norm": 1.7755213975906372,
      "learning_rate": 5.637318311268709e-07,
      "loss": 0.4745,
      "step": 383950
    },
    {
      "epoch": 157.17,
      "grad_norm": 1.9177000522613525,
      "learning_rate": 5.636286407862488e-07,
      "loss": 0.4457,
      "step": 383960
    },
    {
      "epoch": 157.17,
      "grad_norm": 1.6534167528152466,
      "learning_rate": 5.635254586000872e-07,
      "loss": 0.4257,
      "step": 383970
    },
    {
      "epoch": 157.18,
      "grad_norm": 2.098423957824707,
      "learning_rate": 5.634222845688582e-07,
      "loss": 0.4356,
      "step": 383980
    },
    {
      "epoch": 157.18,
      "grad_norm": 2.284032106399536,
      "learning_rate": 5.633191186930348e-07,
      "loss": 0.4424,
      "step": 383990
    },
    {
      "epoch": 157.18,
      "grad_norm": 2.161872148513794,
      "learning_rate": 5.632159609730897e-07,
      "loss": 0.473,
      "step": 384000
    },
    {
      "epoch": 157.19,
      "grad_norm": 1.7416036128997803,
      "learning_rate": 5.631128114094953e-07,
      "loss": 0.4424,
      "step": 384010
    },
    {
      "epoch": 157.19,
      "grad_norm": 2.4104578495025635,
      "learning_rate": 5.630096700027231e-07,
      "loss": 0.4525,
      "step": 384020
    },
    {
      "epoch": 157.2,
      "grad_norm": 2.075985908508301,
      "learning_rate": 5.629065367532482e-07,
      "loss": 0.4483,
      "step": 384030
    },
    {
      "epoch": 157.2,
      "grad_norm": 2.8284552097320557,
      "learning_rate": 5.628034116615407e-07,
      "loss": 0.4784,
      "step": 384040
    },
    {
      "epoch": 157.2,
      "grad_norm": 2.286317825317383,
      "learning_rate": 5.627002947280738e-07,
      "loss": 0.4459,
      "step": 384050
    },
    {
      "epoch": 157.21,
      "grad_norm": 2.0007481575012207,
      "learning_rate": 5.625971859533196e-07,
      "loss": 0.458,
      "step": 384060
    },
    {
      "epoch": 157.21,
      "grad_norm": 2.0285167694091797,
      "learning_rate": 5.624940853377504e-07,
      "loss": 0.4477,
      "step": 384070
    },
    {
      "epoch": 157.22,
      "grad_norm": 1.8379859924316406,
      "learning_rate": 5.623909928818391e-07,
      "loss": 0.4531,
      "step": 384080
    },
    {
      "epoch": 157.22,
      "grad_norm": 2.142324686050415,
      "learning_rate": 5.622879085860573e-07,
      "loss": 0.4493,
      "step": 384090
    },
    {
      "epoch": 157.22,
      "grad_norm": 1.9865013360977173,
      "learning_rate": 5.621848324508775e-07,
      "loss": 0.4409,
      "step": 384100
    },
    {
      "epoch": 157.23,
      "grad_norm": 2.0489413738250732,
      "learning_rate": 5.620817644767718e-07,
      "loss": 0.459,
      "step": 384110
    },
    {
      "epoch": 157.23,
      "grad_norm": 1.8530162572860718,
      "learning_rate": 5.619787046642123e-07,
      "loss": 0.4338,
      "step": 384120
    },
    {
      "epoch": 157.24,
      "grad_norm": 2.8286514282226562,
      "learning_rate": 5.618756530136713e-07,
      "loss": 0.425,
      "step": 384130
    },
    {
      "epoch": 157.24,
      "grad_norm": 1.9995638132095337,
      "learning_rate": 5.617726095256207e-07,
      "loss": 0.4591,
      "step": 384140
    },
    {
      "epoch": 157.25,
      "grad_norm": 3.364000082015991,
      "learning_rate": 5.616695742005323e-07,
      "loss": 0.444,
      "step": 384150
    },
    {
      "epoch": 157.25,
      "grad_norm": 2.5625627040863037,
      "learning_rate": 5.615665470388787e-07,
      "loss": 0.4531,
      "step": 384160
    },
    {
      "epoch": 157.25,
      "grad_norm": 2.000916004180908,
      "learning_rate": 5.614635280411313e-07,
      "loss": 0.4403,
      "step": 384170
    },
    {
      "epoch": 157.26,
      "grad_norm": 1.988614797592163,
      "learning_rate": 5.613605172077622e-07,
      "loss": 0.4406,
      "step": 384180
    },
    {
      "epoch": 157.26,
      "grad_norm": 2.1537582874298096,
      "learning_rate": 5.612575145392435e-07,
      "loss": 0.4451,
      "step": 384190
    },
    {
      "epoch": 157.27,
      "grad_norm": 2.0914902687072754,
      "learning_rate": 5.611545200360467e-07,
      "loss": 0.4705,
      "step": 384200
    },
    {
      "epoch": 157.27,
      "grad_norm": 2.151973247528076,
      "learning_rate": 5.610515336986435e-07,
      "loss": 0.4473,
      "step": 384210
    },
    {
      "epoch": 157.27,
      "grad_norm": 2.163940906524658,
      "learning_rate": 5.60948555527506e-07,
      "loss": 0.4368,
      "step": 384220
    },
    {
      "epoch": 157.28,
      "grad_norm": 2.0457918643951416,
      "learning_rate": 5.608455855231059e-07,
      "loss": 0.4292,
      "step": 384230
    },
    {
      "epoch": 157.28,
      "grad_norm": 1.8571977615356445,
      "learning_rate": 5.607426236859148e-07,
      "loss": 0.44,
      "step": 384240
    },
    {
      "epoch": 157.29,
      "grad_norm": 1.9062750339508057,
      "learning_rate": 5.606396700164044e-07,
      "loss": 0.458,
      "step": 384250
    },
    {
      "epoch": 157.29,
      "grad_norm": 1.7139424085617065,
      "learning_rate": 5.605367245150463e-07,
      "loss": 0.4499,
      "step": 384260
    },
    {
      "epoch": 157.29,
      "grad_norm": 3.0768983364105225,
      "learning_rate": 5.60433787182312e-07,
      "loss": 0.4312,
      "step": 384270
    },
    {
      "epoch": 157.3,
      "grad_norm": 2.07694673538208,
      "learning_rate": 5.603308580186728e-07,
      "loss": 0.4462,
      "step": 384280
    },
    {
      "epoch": 157.3,
      "grad_norm": 1.986839771270752,
      "learning_rate": 5.602279370246011e-07,
      "loss": 0.4299,
      "step": 384290
    },
    {
      "epoch": 157.31,
      "grad_norm": 2.7841975688934326,
      "learning_rate": 5.601250242005673e-07,
      "loss": 0.4584,
      "step": 384300
    },
    {
      "epoch": 157.31,
      "grad_norm": 1.8340637683868408,
      "learning_rate": 5.600221195470434e-07,
      "loss": 0.4394,
      "step": 384310
    },
    {
      "epoch": 157.31,
      "grad_norm": 1.698974847793579,
      "learning_rate": 5.599192230645006e-07,
      "loss": 0.4344,
      "step": 384320
    },
    {
      "epoch": 157.32,
      "grad_norm": 1.9390348196029663,
      "learning_rate": 5.598163347534104e-07,
      "loss": 0.4283,
      "step": 384330
    },
    {
      "epoch": 157.32,
      "grad_norm": 1.8637630939483643,
      "learning_rate": 5.597134546142448e-07,
      "loss": 0.4678,
      "step": 384340
    },
    {
      "epoch": 157.33,
      "grad_norm": 1.6171019077301025,
      "learning_rate": 5.596105826474734e-07,
      "loss": 0.4532,
      "step": 384350
    },
    {
      "epoch": 157.33,
      "grad_norm": 1.9047393798828125,
      "learning_rate": 5.595077188535686e-07,
      "loss": 0.4429,
      "step": 384360
    },
    {
      "epoch": 157.34,
      "grad_norm": 2.1642982959747314,
      "learning_rate": 5.594048632330014e-07,
      "loss": 0.4304,
      "step": 384370
    },
    {
      "epoch": 157.34,
      "grad_norm": 2.013784408569336,
      "learning_rate": 5.593020157862429e-07,
      "loss": 0.4364,
      "step": 384380
    },
    {
      "epoch": 157.34,
      "grad_norm": 2.3125925064086914,
      "learning_rate": 5.591991765137641e-07,
      "loss": 0.4425,
      "step": 384390
    },
    {
      "epoch": 157.35,
      "grad_norm": 1.8107478618621826,
      "learning_rate": 5.590963454160354e-07,
      "loss": 0.451,
      "step": 384400
    },
    {
      "epoch": 157.35,
      "grad_norm": 2.2528140544891357,
      "learning_rate": 5.589935224935296e-07,
      "loss": 0.4596,
      "step": 384410
    },
    {
      "epoch": 157.36,
      "grad_norm": 2.560889482498169,
      "learning_rate": 5.58890707746717e-07,
      "loss": 0.4392,
      "step": 384420
    },
    {
      "epoch": 157.36,
      "grad_norm": 1.9550914764404297,
      "learning_rate": 5.587879011760683e-07,
      "loss": 0.4434,
      "step": 384430
    },
    {
      "epoch": 157.36,
      "grad_norm": 1.573143482208252,
      "learning_rate": 5.586851027820544e-07,
      "loss": 0.4713,
      "step": 384440
    },
    {
      "epoch": 157.37,
      "grad_norm": 1.905814290046692,
      "learning_rate": 5.585823125651465e-07,
      "loss": 0.442,
      "step": 384450
    },
    {
      "epoch": 157.37,
      "grad_norm": 1.7972090244293213,
      "learning_rate": 5.584795305258153e-07,
      "loss": 0.4565,
      "step": 384460
    },
    {
      "epoch": 157.38,
      "grad_norm": 1.9570034742355347,
      "learning_rate": 5.583767566645315e-07,
      "loss": 0.4445,
      "step": 384470
    },
    {
      "epoch": 157.38,
      "grad_norm": 1.9218220710754395,
      "learning_rate": 5.582739909817668e-07,
      "loss": 0.4401,
      "step": 384480
    },
    {
      "epoch": 157.38,
      "grad_norm": 1.8366475105285645,
      "learning_rate": 5.581712334779905e-07,
      "loss": 0.4575,
      "step": 384490
    },
    {
      "epoch": 157.39,
      "grad_norm": 2.205880641937256,
      "learning_rate": 5.580684841536739e-07,
      "loss": 0.4499,
      "step": 384500
    },
    {
      "epoch": 157.39,
      "grad_norm": 2.437729835510254,
      "learning_rate": 5.579657430092881e-07,
      "loss": 0.4429,
      "step": 384510
    },
    {
      "epoch": 157.4,
      "grad_norm": 1.6718133687973022,
      "learning_rate": 5.57863010045303e-07,
      "loss": 0.4516,
      "step": 384520
    },
    {
      "epoch": 157.4,
      "grad_norm": 1.8000775575637817,
      "learning_rate": 5.577602852621899e-07,
      "loss": 0.4394,
      "step": 384530
    },
    {
      "epoch": 157.4,
      "grad_norm": 1.800652265548706,
      "learning_rate": 5.576575686604188e-07,
      "loss": 0.4546,
      "step": 384540
    },
    {
      "epoch": 157.41,
      "grad_norm": 1.7552971839904785,
      "learning_rate": 5.575548602404608e-07,
      "loss": 0.4492,
      "step": 384550
    },
    {
      "epoch": 157.41,
      "grad_norm": 2.0368010997772217,
      "learning_rate": 5.574521600027861e-07,
      "loss": 0.4543,
      "step": 384560
    },
    {
      "epoch": 157.42,
      "grad_norm": 1.8170095682144165,
      "learning_rate": 5.573494679478649e-07,
      "loss": 0.4547,
      "step": 384570
    },
    {
      "epoch": 157.42,
      "grad_norm": 2.1351213455200195,
      "learning_rate": 5.57246784076168e-07,
      "loss": 0.4465,
      "step": 384580
    },
    {
      "epoch": 157.43,
      "grad_norm": 1.9847694635391235,
      "learning_rate": 5.571441083881651e-07,
      "loss": 0.4406,
      "step": 384590
    },
    {
      "epoch": 157.43,
      "grad_norm": 1.8556699752807617,
      "learning_rate": 5.570414408843278e-07,
      "loss": 0.4694,
      "step": 384600
    },
    {
      "epoch": 157.43,
      "grad_norm": 1.6959147453308105,
      "learning_rate": 5.569387815651255e-07,
      "loss": 0.4475,
      "step": 384610
    },
    {
      "epoch": 157.44,
      "grad_norm": 1.9365057945251465,
      "learning_rate": 5.568361304310295e-07,
      "loss": 0.466,
      "step": 384620
    },
    {
      "epoch": 157.44,
      "grad_norm": 2.686375617980957,
      "learning_rate": 5.567334874825083e-07,
      "loss": 0.4291,
      "step": 384630
    },
    {
      "epoch": 157.45,
      "grad_norm": 1.6905921697616577,
      "learning_rate": 5.566308527200333e-07,
      "loss": 0.4518,
      "step": 384640
    },
    {
      "epoch": 157.45,
      "grad_norm": 1.8626008033752441,
      "learning_rate": 5.565282261440743e-07,
      "loss": 0.4575,
      "step": 384650
    },
    {
      "epoch": 157.45,
      "grad_norm": 2.4195544719696045,
      "learning_rate": 5.564256077551013e-07,
      "loss": 0.4657,
      "step": 384660
    },
    {
      "epoch": 157.46,
      "grad_norm": 2.262540578842163,
      "learning_rate": 5.563229975535847e-07,
      "loss": 0.4699,
      "step": 384670
    },
    {
      "epoch": 157.46,
      "grad_norm": 2.143158435821533,
      "learning_rate": 5.562203955399942e-07,
      "loss": 0.4461,
      "step": 384680
    },
    {
      "epoch": 157.47,
      "grad_norm": 2.2984588146209717,
      "learning_rate": 5.561178017148001e-07,
      "loss": 0.459,
      "step": 384690
    },
    {
      "epoch": 157.47,
      "grad_norm": 1.895337462425232,
      "learning_rate": 5.560152160784722e-07,
      "loss": 0.436,
      "step": 384700
    },
    {
      "epoch": 157.47,
      "grad_norm": 1.8928799629211426,
      "learning_rate": 5.559126386314806e-07,
      "loss": 0.4393,
      "step": 384710
    },
    {
      "epoch": 157.48,
      "grad_norm": 2.2772953510284424,
      "learning_rate": 5.55810069374295e-07,
      "loss": 0.4377,
      "step": 384720
    },
    {
      "epoch": 157.48,
      "grad_norm": 1.9538037776947021,
      "learning_rate": 5.557075083073851e-07,
      "loss": 0.464,
      "step": 384730
    },
    {
      "epoch": 157.49,
      "grad_norm": 2.0430428981781006,
      "learning_rate": 5.556049554312212e-07,
      "loss": 0.4703,
      "step": 384740
    },
    {
      "epoch": 157.49,
      "grad_norm": 2.1828174591064453,
      "learning_rate": 5.555024107462727e-07,
      "loss": 0.4399,
      "step": 384750
    },
    {
      "epoch": 157.49,
      "grad_norm": 1.7262506484985352,
      "learning_rate": 5.553998742530096e-07,
      "loss": 0.4362,
      "step": 384760
    },
    {
      "epoch": 157.5,
      "grad_norm": 2.2008001804351807,
      "learning_rate": 5.552973459519012e-07,
      "loss": 0.4376,
      "step": 384770
    },
    {
      "epoch": 157.5,
      "grad_norm": 2.5165770053863525,
      "learning_rate": 5.551948258434178e-07,
      "loss": 0.4506,
      "step": 384780
    },
    {
      "epoch": 157.51,
      "grad_norm": 2.3824007511138916,
      "learning_rate": 5.550923139280286e-07,
      "loss": 0.458,
      "step": 384790
    },
    {
      "epoch": 157.51,
      "grad_norm": 1.9138222932815552,
      "learning_rate": 5.549898102062029e-07,
      "loss": 0.4724,
      "step": 384800
    },
    {
      "epoch": 157.52,
      "grad_norm": 2.4571332931518555,
      "learning_rate": 5.54887314678411e-07,
      "loss": 0.4321,
      "step": 384810
    },
    {
      "epoch": 157.52,
      "grad_norm": 1.7525794506072998,
      "learning_rate": 5.547848273451218e-07,
      "loss": 0.4533,
      "step": 384820
    },
    {
      "epoch": 157.52,
      "grad_norm": 1.7245179414749146,
      "learning_rate": 5.546823482068051e-07,
      "loss": 0.484,
      "step": 384830
    },
    {
      "epoch": 157.53,
      "grad_norm": 1.8496568202972412,
      "learning_rate": 5.545798772639301e-07,
      "loss": 0.4383,
      "step": 384840
    },
    {
      "epoch": 157.53,
      "grad_norm": 1.7544885873794556,
      "learning_rate": 5.544774145169665e-07,
      "loss": 0.4482,
      "step": 384850
    },
    {
      "epoch": 157.54,
      "grad_norm": 1.8145458698272705,
      "learning_rate": 5.543749599663835e-07,
      "loss": 0.4411,
      "step": 384860
    },
    {
      "epoch": 157.54,
      "grad_norm": 1.7868151664733887,
      "learning_rate": 5.542725136126502e-07,
      "loss": 0.441,
      "step": 384870
    },
    {
      "epoch": 157.54,
      "grad_norm": 2.1541779041290283,
      "learning_rate": 5.541700754562365e-07,
      "loss": 0.4547,
      "step": 384880
    },
    {
      "epoch": 157.55,
      "grad_norm": 2.685394287109375,
      "learning_rate": 5.540676454976112e-07,
      "loss": 0.4576,
      "step": 384890
    },
    {
      "epoch": 157.55,
      "grad_norm": 1.9190515279769897,
      "learning_rate": 5.539652237372435e-07,
      "loss": 0.4471,
      "step": 384900
    },
    {
      "epoch": 157.56,
      "grad_norm": 1.769187331199646,
      "learning_rate": 5.538628101756026e-07,
      "loss": 0.426,
      "step": 384910
    },
    {
      "epoch": 157.56,
      "grad_norm": 2.433286190032959,
      "learning_rate": 5.537604048131578e-07,
      "loss": 0.4292,
      "step": 384920
    },
    {
      "epoch": 157.56,
      "grad_norm": 1.7925024032592773,
      "learning_rate": 5.536580076503786e-07,
      "loss": 0.4343,
      "step": 384930
    },
    {
      "epoch": 157.57,
      "grad_norm": 1.8862324953079224,
      "learning_rate": 5.535556186877333e-07,
      "loss": 0.4298,
      "step": 384940
    },
    {
      "epoch": 157.57,
      "grad_norm": 2.179337501525879,
      "learning_rate": 5.534532379256909e-07,
      "loss": 0.459,
      "step": 384950
    },
    {
      "epoch": 157.58,
      "grad_norm": 1.8163655996322632,
      "learning_rate": 5.533508653647201e-07,
      "loss": 0.4476,
      "step": 384960
    },
    {
      "epoch": 157.58,
      "grad_norm": 2.4053916931152344,
      "learning_rate": 5.532485010052914e-07,
      "loss": 0.4295,
      "step": 384970
    },
    {
      "epoch": 157.58,
      "grad_norm": 2.6075439453125,
      "learning_rate": 5.531461448478723e-07,
      "loss": 0.4494,
      "step": 384980
    },
    {
      "epoch": 157.59,
      "grad_norm": 1.8892970085144043,
      "learning_rate": 5.530437968929325e-07,
      "loss": 0.4429,
      "step": 384990
    },
    {
      "epoch": 157.59,
      "grad_norm": 2.0227835178375244,
      "learning_rate": 5.529414571409404e-07,
      "loss": 0.4537,
      "step": 385000
    },
    {
      "epoch": 157.6,
      "grad_norm": 2.29750919342041,
      "learning_rate": 5.528391255923649e-07,
      "loss": 0.4623,
      "step": 385010
    },
    {
      "epoch": 157.6,
      "grad_norm": 2.048013925552368,
      "learning_rate": 5.527368022476747e-07,
      "loss": 0.4379,
      "step": 385020
    },
    {
      "epoch": 157.61,
      "grad_norm": 1.6640980243682861,
      "learning_rate": 5.526344871073385e-07,
      "loss": 0.4533,
      "step": 385030
    },
    {
      "epoch": 157.61,
      "grad_norm": 2.207716703414917,
      "learning_rate": 5.525321801718255e-07,
      "loss": 0.4434,
      "step": 385040
    },
    {
      "epoch": 157.61,
      "grad_norm": 1.9158815145492554,
      "learning_rate": 5.524298814416035e-07,
      "loss": 0.4538,
      "step": 385050
    },
    {
      "epoch": 157.62,
      "grad_norm": 1.857352375984192,
      "learning_rate": 5.523275909171423e-07,
      "loss": 0.4282,
      "step": 385060
    },
    {
      "epoch": 157.62,
      "grad_norm": 2.4298501014709473,
      "learning_rate": 5.522253085989092e-07,
      "loss": 0.4333,
      "step": 385070
    },
    {
      "epoch": 157.63,
      "grad_norm": 1.9233729839324951,
      "learning_rate": 5.52123034487373e-07,
      "loss": 0.4412,
      "step": 385080
    },
    {
      "epoch": 157.63,
      "grad_norm": 2.00295352935791,
      "learning_rate": 5.520207685830027e-07,
      "loss": 0.4582,
      "step": 385090
    },
    {
      "epoch": 157.63,
      "grad_norm": 2.0672402381896973,
      "learning_rate": 5.519185108862664e-07,
      "loss": 0.4626,
      "step": 385100
    },
    {
      "epoch": 157.64,
      "grad_norm": 1.7245285511016846,
      "learning_rate": 5.518162613976328e-07,
      "loss": 0.443,
      "step": 385110
    },
    {
      "epoch": 157.64,
      "grad_norm": 1.9820621013641357,
      "learning_rate": 5.517140201175701e-07,
      "loss": 0.4441,
      "step": 385120
    },
    {
      "epoch": 157.65,
      "grad_norm": 1.8090479373931885,
      "learning_rate": 5.516117870465465e-07,
      "loss": 0.4431,
      "step": 385130
    },
    {
      "epoch": 157.65,
      "grad_norm": 2.241272211074829,
      "learning_rate": 5.515095621850307e-07,
      "loss": 0.4472,
      "step": 385140
    },
    {
      "epoch": 157.65,
      "grad_norm": 2.089949607849121,
      "learning_rate": 5.5140734553349e-07,
      "loss": 0.4393,
      "step": 385150
    },
    {
      "epoch": 157.66,
      "grad_norm": 2.018477439880371,
      "learning_rate": 5.513051370923941e-07,
      "loss": 0.4684,
      "step": 385160
    },
    {
      "epoch": 157.66,
      "grad_norm": 1.955360770225525,
      "learning_rate": 5.512029368622106e-07,
      "loss": 0.4363,
      "step": 385170
    },
    {
      "epoch": 157.67,
      "grad_norm": 1.770324468612671,
      "learning_rate": 5.511007448434075e-07,
      "loss": 0.438,
      "step": 385180
    },
    {
      "epoch": 157.67,
      "grad_norm": 2.5934712886810303,
      "learning_rate": 5.509985610364531e-07,
      "loss": 0.4627,
      "step": 385190
    },
    {
      "epoch": 157.67,
      "grad_norm": 1.8413784503936768,
      "learning_rate": 5.508963854418161e-07,
      "loss": 0.4588,
      "step": 385200
    },
    {
      "epoch": 157.68,
      "grad_norm": 2.2245678901672363,
      "learning_rate": 5.50794218059963e-07,
      "loss": 0.4393,
      "step": 385210
    },
    {
      "epoch": 157.68,
      "grad_norm": 1.6863963603973389,
      "learning_rate": 5.506920588913627e-07,
      "loss": 0.4473,
      "step": 385220
    },
    {
      "epoch": 157.69,
      "grad_norm": 1.799519419670105,
      "learning_rate": 5.505899079364832e-07,
      "loss": 0.4559,
      "step": 385230
    },
    {
      "epoch": 157.69,
      "grad_norm": 2.015160322189331,
      "learning_rate": 5.504877651957923e-07,
      "loss": 0.439,
      "step": 385240
    },
    {
      "epoch": 157.7,
      "grad_norm": 2.0313498973846436,
      "learning_rate": 5.503856306697582e-07,
      "loss": 0.4474,
      "step": 385250
    },
    {
      "epoch": 157.7,
      "grad_norm": 2.580002546310425,
      "learning_rate": 5.502835043588483e-07,
      "loss": 0.4351,
      "step": 385260
    },
    {
      "epoch": 157.7,
      "grad_norm": 1.9993155002593994,
      "learning_rate": 5.501813862635308e-07,
      "loss": 0.4356,
      "step": 385270
    },
    {
      "epoch": 157.71,
      "grad_norm": 2.1238882541656494,
      "learning_rate": 5.500792763842734e-07,
      "loss": 0.4457,
      "step": 385280
    },
    {
      "epoch": 157.71,
      "grad_norm": 1.8533838987350464,
      "learning_rate": 5.499771747215437e-07,
      "loss": 0.4519,
      "step": 385290
    },
    {
      "epoch": 157.72,
      "grad_norm": 1.848105549812317,
      "learning_rate": 5.498750812758095e-07,
      "loss": 0.4586,
      "step": 385300
    },
    {
      "epoch": 157.72,
      "grad_norm": 2.1713287830352783,
      "learning_rate": 5.497729960475386e-07,
      "loss": 0.4376,
      "step": 385310
    },
    {
      "epoch": 157.72,
      "grad_norm": 2.4435856342315674,
      "learning_rate": 5.496709190371983e-07,
      "loss": 0.4385,
      "step": 385320
    },
    {
      "epoch": 157.73,
      "grad_norm": 1.8452274799346924,
      "learning_rate": 5.495688502452565e-07,
      "loss": 0.4582,
      "step": 385330
    },
    {
      "epoch": 157.73,
      "grad_norm": 2.1521553993225098,
      "learning_rate": 5.494667896721808e-07,
      "loss": 0.426,
      "step": 385340
    },
    {
      "epoch": 157.74,
      "grad_norm": 1.9235239028930664,
      "learning_rate": 5.493647373184385e-07,
      "loss": 0.4511,
      "step": 385350
    },
    {
      "epoch": 157.74,
      "grad_norm": 2.4193825721740723,
      "learning_rate": 5.492626931844973e-07,
      "loss": 0.45,
      "step": 385360
    },
    {
      "epoch": 157.74,
      "grad_norm": 1.6850026845932007,
      "learning_rate": 5.491606572708246e-07,
      "loss": 0.4538,
      "step": 385370
    },
    {
      "epoch": 157.75,
      "grad_norm": 2.239535093307495,
      "learning_rate": 5.490586295778874e-07,
      "loss": 0.454,
      "step": 385380
    },
    {
      "epoch": 157.75,
      "grad_norm": 1.9875359535217285,
      "learning_rate": 5.489566101061536e-07,
      "loss": 0.4589,
      "step": 385390
    },
    {
      "epoch": 157.76,
      "grad_norm": 2.12023663520813,
      "learning_rate": 5.488545988560906e-07,
      "loss": 0.4342,
      "step": 385400
    },
    {
      "epoch": 157.76,
      "grad_norm": 2.159914016723633,
      "learning_rate": 5.487525958281651e-07,
      "loss": 0.4663,
      "step": 385410
    },
    {
      "epoch": 157.77,
      "grad_norm": 2.521723747253418,
      "learning_rate": 5.48650601022845e-07,
      "loss": 0.4612,
      "step": 385420
    },
    {
      "epoch": 157.77,
      "grad_norm": 2.1711020469665527,
      "learning_rate": 5.485486144405972e-07,
      "loss": 0.4622,
      "step": 385430
    },
    {
      "epoch": 157.77,
      "grad_norm": 1.7255322933197021,
      "learning_rate": 5.484466360818888e-07,
      "loss": 0.4361,
      "step": 385440
    },
    {
      "epoch": 157.78,
      "grad_norm": 1.9967299699783325,
      "learning_rate": 5.483446659471872e-07,
      "loss": 0.4608,
      "step": 385450
    },
    {
      "epoch": 157.78,
      "grad_norm": 2.172778367996216,
      "learning_rate": 5.482427040369592e-07,
      "loss": 0.4324,
      "step": 385460
    },
    {
      "epoch": 157.79,
      "grad_norm": 1.946829915046692,
      "learning_rate": 5.481407503516721e-07,
      "loss": 0.4367,
      "step": 385470
    },
    {
      "epoch": 157.79,
      "grad_norm": 1.8340431451797485,
      "learning_rate": 5.480388048917931e-07,
      "loss": 0.444,
      "step": 385480
    },
    {
      "epoch": 157.79,
      "grad_norm": 1.6953072547912598,
      "learning_rate": 5.479368676577886e-07,
      "loss": 0.4548,
      "step": 385490
    },
    {
      "epoch": 157.8,
      "grad_norm": 1.8559921979904175,
      "learning_rate": 5.478349386501262e-07,
      "loss": 0.4359,
      "step": 385500
    },
    {
      "epoch": 157.8,
      "grad_norm": 2.208291530609131,
      "learning_rate": 5.477330178692732e-07,
      "loss": 0.4508,
      "step": 385510
    },
    {
      "epoch": 157.81,
      "grad_norm": 2.163097381591797,
      "learning_rate": 5.476311053156952e-07,
      "loss": 0.4377,
      "step": 385520
    },
    {
      "epoch": 157.81,
      "grad_norm": 2.190668821334839,
      "learning_rate": 5.47529200989859e-07,
      "loss": 0.4351,
      "step": 385530
    },
    {
      "epoch": 157.81,
      "grad_norm": 2.0487287044525146,
      "learning_rate": 5.474273048922327e-07,
      "loss": 0.4543,
      "step": 385540
    },
    {
      "epoch": 157.82,
      "grad_norm": 2.309415578842163,
      "learning_rate": 5.473254170232828e-07,
      "loss": 0.4402,
      "step": 385550
    },
    {
      "epoch": 157.82,
      "grad_norm": 2.117366075515747,
      "learning_rate": 5.472235373834753e-07,
      "loss": 0.426,
      "step": 385560
    },
    {
      "epoch": 157.83,
      "grad_norm": 2.0115206241607666,
      "learning_rate": 5.471216659732775e-07,
      "loss": 0.4327,
      "step": 385570
    },
    {
      "epoch": 157.83,
      "grad_norm": 2.494600534439087,
      "learning_rate": 5.470198027931559e-07,
      "loss": 0.4666,
      "step": 385580
    },
    {
      "epoch": 157.83,
      "grad_norm": 2.382664680480957,
      "learning_rate": 5.469179478435769e-07,
      "loss": 0.436,
      "step": 385590
    },
    {
      "epoch": 157.84,
      "grad_norm": 3.4212992191314697,
      "learning_rate": 5.468161011250072e-07,
      "loss": 0.4598,
      "step": 385600
    },
    {
      "epoch": 157.84,
      "grad_norm": 1.9093941450119019,
      "learning_rate": 5.467142626379138e-07,
      "loss": 0.4514,
      "step": 385610
    },
    {
      "epoch": 157.85,
      "grad_norm": 2.188044548034668,
      "learning_rate": 5.466124323827625e-07,
      "loss": 0.4456,
      "step": 385620
    },
    {
      "epoch": 157.85,
      "grad_norm": 2.168574810028076,
      "learning_rate": 5.4651061036002e-07,
      "loss": 0.4521,
      "step": 385630
    },
    {
      "epoch": 157.86,
      "grad_norm": 2.240593194961548,
      "learning_rate": 5.464087965701536e-07,
      "loss": 0.4567,
      "step": 385640
    },
    {
      "epoch": 157.86,
      "grad_norm": 1.741153597831726,
      "learning_rate": 5.463069910136282e-07,
      "loss": 0.4576,
      "step": 385650
    },
    {
      "epoch": 157.86,
      "grad_norm": 1.8814504146575928,
      "learning_rate": 5.462051936909109e-07,
      "loss": 0.4548,
      "step": 385660
    },
    {
      "epoch": 157.87,
      "grad_norm": 2.233799695968628,
      "learning_rate": 5.46103404602468e-07,
      "loss": 0.4608,
      "step": 385670
    },
    {
      "epoch": 157.87,
      "grad_norm": 2.36665940284729,
      "learning_rate": 5.460016237487657e-07,
      "loss": 0.4589,
      "step": 385680
    },
    {
      "epoch": 157.88,
      "grad_norm": 1.9079313278198242,
      "learning_rate": 5.458998511302703e-07,
      "loss": 0.467,
      "step": 385690
    },
    {
      "epoch": 157.88,
      "grad_norm": 1.9190977811813354,
      "learning_rate": 5.45798086747448e-07,
      "loss": 0.4378,
      "step": 385700
    },
    {
      "epoch": 157.88,
      "grad_norm": 1.5566695928573608,
      "learning_rate": 5.45696330600765e-07,
      "loss": 0.4395,
      "step": 385710
    },
    {
      "epoch": 157.89,
      "grad_norm": 1.9478068351745605,
      "learning_rate": 5.455945826906867e-07,
      "loss": 0.4633,
      "step": 385720
    },
    {
      "epoch": 157.89,
      "grad_norm": 1.914294958114624,
      "learning_rate": 5.454928430176806e-07,
      "loss": 0.4303,
      "step": 385730
    },
    {
      "epoch": 157.9,
      "grad_norm": 2.149503469467163,
      "learning_rate": 5.453911115822121e-07,
      "loss": 0.4573,
      "step": 385740
    },
    {
      "epoch": 157.9,
      "grad_norm": 1.9404628276824951,
      "learning_rate": 5.452893883847469e-07,
      "loss": 0.43,
      "step": 385750
    },
    {
      "epoch": 157.9,
      "grad_norm": 2.2117326259613037,
      "learning_rate": 5.451876734257513e-07,
      "loss": 0.4565,
      "step": 385760
    },
    {
      "epoch": 157.91,
      "grad_norm": 1.834531545639038,
      "learning_rate": 5.450859667056913e-07,
      "loss": 0.4282,
      "step": 385770
    },
    {
      "epoch": 157.91,
      "grad_norm": 2.190246820449829,
      "learning_rate": 5.449842682250332e-07,
      "loss": 0.4519,
      "step": 385780
    },
    {
      "epoch": 157.92,
      "grad_norm": 2.052713632583618,
      "learning_rate": 5.44882577984242e-07,
      "loss": 0.4234,
      "step": 385790
    },
    {
      "epoch": 157.92,
      "grad_norm": 1.9406062364578247,
      "learning_rate": 5.447808959837836e-07,
      "loss": 0.441,
      "step": 385800
    },
    {
      "epoch": 157.92,
      "grad_norm": 2.361168384552002,
      "learning_rate": 5.446792222241241e-07,
      "loss": 0.4654,
      "step": 385810
    },
    {
      "epoch": 157.93,
      "grad_norm": 2.0764336585998535,
      "learning_rate": 5.445775567057292e-07,
      "loss": 0.4497,
      "step": 385820
    },
    {
      "epoch": 157.93,
      "grad_norm": 1.7608206272125244,
      "learning_rate": 5.444758994290646e-07,
      "loss": 0.4517,
      "step": 385830
    },
    {
      "epoch": 157.94,
      "grad_norm": 2.3773012161254883,
      "learning_rate": 5.44374250394596e-07,
      "loss": 0.4334,
      "step": 385840
    },
    {
      "epoch": 157.94,
      "grad_norm": 2.355193614959717,
      "learning_rate": 5.442726096027893e-07,
      "loss": 0.4629,
      "step": 385850
    },
    {
      "epoch": 157.95,
      "grad_norm": 2.001490592956543,
      "learning_rate": 5.441709770541095e-07,
      "loss": 0.4579,
      "step": 385860
    },
    {
      "epoch": 157.95,
      "grad_norm": 1.948578953742981,
      "learning_rate": 5.440693527490226e-07,
      "loss": 0.4382,
      "step": 385870
    },
    {
      "epoch": 157.95,
      "grad_norm": 2.506566286087036,
      "learning_rate": 5.439677366879941e-07,
      "loss": 0.4475,
      "step": 385880
    },
    {
      "epoch": 157.96,
      "grad_norm": 1.9103411436080933,
      "learning_rate": 5.438661288714895e-07,
      "loss": 0.4445,
      "step": 385890
    },
    {
      "epoch": 157.96,
      "grad_norm": 1.9665809869766235,
      "learning_rate": 5.43764529299974e-07,
      "loss": 0.4323,
      "step": 385900
    },
    {
      "epoch": 157.97,
      "grad_norm": 2.0301480293273926,
      "learning_rate": 5.436629379739126e-07,
      "loss": 0.4614,
      "step": 385910
    },
    {
      "epoch": 157.97,
      "grad_norm": 2.240415334701538,
      "learning_rate": 5.435613548937724e-07,
      "loss": 0.464,
      "step": 385920
    },
    {
      "epoch": 157.97,
      "grad_norm": 1.9531476497650146,
      "learning_rate": 5.434597800600172e-07,
      "loss": 0.4506,
      "step": 385930
    },
    {
      "epoch": 157.98,
      "grad_norm": 1.7544583082199097,
      "learning_rate": 5.433582134731126e-07,
      "loss": 0.4394,
      "step": 385940
    },
    {
      "epoch": 157.98,
      "grad_norm": 2.2665905952453613,
      "learning_rate": 5.432566551335237e-07,
      "loss": 0.466,
      "step": 385950
    },
    {
      "epoch": 157.99,
      "grad_norm": 2.0664055347442627,
      "learning_rate": 5.431551050417163e-07,
      "loss": 0.4444,
      "step": 385960
    },
    {
      "epoch": 157.99,
      "grad_norm": 14.554312705993652,
      "learning_rate": 5.430535631981551e-07,
      "loss": 0.4452,
      "step": 385970
    },
    {
      "epoch": 157.99,
      "grad_norm": 1.886643648147583,
      "learning_rate": 5.429520296033054e-07,
      "loss": 0.4419,
      "step": 385980
    },
    {
      "epoch": 158.0,
      "grad_norm": 2.2568438053131104,
      "learning_rate": 5.428505042576323e-07,
      "loss": 0.4713,
      "step": 385990
    },
    {
      "epoch": 158.0,
      "eval_loss": 0.44798049330711365,
      "eval_runtime": 53.0383,
      "eval_samples_per_second": 65.028,
      "eval_steps_per_second": 8.145,
      "step": 385994
    },
    {
      "epoch": 158.0,
      "grad_norm": 2.0489754676818848,
      "learning_rate": 5.42748987161601e-07,
      "loss": 0.4504,
      "step": 386000
    },
    {
      "epoch": 158.01,
      "grad_norm": 2.813781499862671,
      "learning_rate": 5.426474783156763e-07,
      "loss": 0.4336,
      "step": 386010
    },
    {
      "epoch": 158.01,
      "grad_norm": 2.29775071144104,
      "learning_rate": 5.425459777203235e-07,
      "loss": 0.461,
      "step": 386020
    },
    {
      "epoch": 158.01,
      "grad_norm": 2.029550313949585,
      "learning_rate": 5.424444853760072e-07,
      "loss": 0.4545,
      "step": 386030
    },
    {
      "epoch": 158.02,
      "grad_norm": 2.315439462661743,
      "learning_rate": 5.423430012831926e-07,
      "loss": 0.4286,
      "step": 386040
    },
    {
      "epoch": 158.02,
      "grad_norm": 1.8685301542282104,
      "learning_rate": 5.422415254423444e-07,
      "loss": 0.4524,
      "step": 386050
    },
    {
      "epoch": 158.03,
      "grad_norm": 2.0553371906280518,
      "learning_rate": 5.421400578539275e-07,
      "loss": 0.4289,
      "step": 386060
    },
    {
      "epoch": 158.03,
      "grad_norm": 1.9360960721969604,
      "learning_rate": 5.420385985184069e-07,
      "loss": 0.4426,
      "step": 386070
    },
    {
      "epoch": 158.04,
      "grad_norm": 1.8073253631591797,
      "learning_rate": 5.419371474362471e-07,
      "loss": 0.4456,
      "step": 386080
    },
    {
      "epoch": 158.04,
      "grad_norm": 1.7955015897750854,
      "learning_rate": 5.418357046079135e-07,
      "loss": 0.4427,
      "step": 386090
    },
    {
      "epoch": 158.04,
      "grad_norm": 2.0968828201293945,
      "learning_rate": 5.417342700338689e-07,
      "loss": 0.4221,
      "step": 386100
    },
    {
      "epoch": 158.05,
      "grad_norm": 2.1390719413757324,
      "learning_rate": 5.4163284371458e-07,
      "loss": 0.4647,
      "step": 386110
    },
    {
      "epoch": 158.05,
      "grad_norm": 2.2438454627990723,
      "learning_rate": 5.415314256505107e-07,
      "loss": 0.4531,
      "step": 386120
    },
    {
      "epoch": 158.06,
      "grad_norm": 1.9995818138122559,
      "learning_rate": 5.414300158421257e-07,
      "loss": 0.4633,
      "step": 386130
    },
    {
      "epoch": 158.06,
      "grad_norm": 2.098794937133789,
      "learning_rate": 5.413286142898892e-07,
      "loss": 0.4326,
      "step": 386140
    },
    {
      "epoch": 158.06,
      "grad_norm": 1.9282355308532715,
      "learning_rate": 5.412272209942662e-07,
      "loss": 0.4434,
      "step": 386150
    },
    {
      "epoch": 158.07,
      "grad_norm": 2.1483352184295654,
      "learning_rate": 5.411258359557207e-07,
      "loss": 0.4474,
      "step": 386160
    },
    {
      "epoch": 158.07,
      "grad_norm": 2.4670071601867676,
      "learning_rate": 5.410244591747173e-07,
      "loss": 0.4526,
      "step": 386170
    },
    {
      "epoch": 158.08,
      "grad_norm": 1.865643858909607,
      "learning_rate": 5.409230906517206e-07,
      "loss": 0.4592,
      "step": 386180
    },
    {
      "epoch": 158.08,
      "grad_norm": 1.89357328414917,
      "learning_rate": 5.408217303871946e-07,
      "loss": 0.4495,
      "step": 386190
    },
    {
      "epoch": 158.08,
      "grad_norm": 1.957267165184021,
      "learning_rate": 5.407203783816039e-07,
      "loss": 0.4353,
      "step": 386200
    },
    {
      "epoch": 158.09,
      "grad_norm": 1.8818694353103638,
      "learning_rate": 5.406190346354125e-07,
      "loss": 0.4537,
      "step": 386210
    },
    {
      "epoch": 158.09,
      "grad_norm": 1.8952406644821167,
      "learning_rate": 5.405176991490849e-07,
      "loss": 0.4275,
      "step": 386220
    },
    {
      "epoch": 158.1,
      "grad_norm": 2.2247090339660645,
      "learning_rate": 5.404163719230857e-07,
      "loss": 0.448,
      "step": 386230
    },
    {
      "epoch": 158.1,
      "grad_norm": 2.0185563564300537,
      "learning_rate": 5.403150529578782e-07,
      "loss": 0.4489,
      "step": 386240
    },
    {
      "epoch": 158.1,
      "grad_norm": 1.8353534936904907,
      "learning_rate": 5.402137422539266e-07,
      "loss": 0.4348,
      "step": 386250
    },
    {
      "epoch": 158.11,
      "grad_norm": 2.017707586288452,
      "learning_rate": 5.401124398116956e-07,
      "loss": 0.4573,
      "step": 386260
    },
    {
      "epoch": 158.11,
      "grad_norm": 2.2593531608581543,
      "learning_rate": 5.400111456316487e-07,
      "loss": 0.4396,
      "step": 386270
    },
    {
      "epoch": 158.12,
      "grad_norm": 1.769289493560791,
      "learning_rate": 5.399098597142496e-07,
      "loss": 0.4377,
      "step": 386280
    },
    {
      "epoch": 158.12,
      "grad_norm": 2.078991174697876,
      "learning_rate": 5.398085820599634e-07,
      "loss": 0.4435,
      "step": 386290
    },
    {
      "epoch": 158.13,
      "grad_norm": 2.147623062133789,
      "learning_rate": 5.397073126692538e-07,
      "loss": 0.4396,
      "step": 386300
    },
    {
      "epoch": 158.13,
      "grad_norm": 2.029395818710327,
      "learning_rate": 5.396060515425838e-07,
      "loss": 0.45,
      "step": 386310
    },
    {
      "epoch": 158.13,
      "grad_norm": 1.8858999013900757,
      "learning_rate": 5.395047986804183e-07,
      "loss": 0.4482,
      "step": 386320
    },
    {
      "epoch": 158.14,
      "grad_norm": 1.9163599014282227,
      "learning_rate": 5.394035540832204e-07,
      "loss": 0.4329,
      "step": 386330
    },
    {
      "epoch": 158.14,
      "grad_norm": 2.347449541091919,
      "learning_rate": 5.393023177514541e-07,
      "loss": 0.4481,
      "step": 386340
    },
    {
      "epoch": 158.15,
      "grad_norm": 2.0080952644348145,
      "learning_rate": 5.392010896855833e-07,
      "loss": 0.4386,
      "step": 386350
    },
    {
      "epoch": 158.15,
      "grad_norm": 1.75718355178833,
      "learning_rate": 5.390998698860723e-07,
      "loss": 0.441,
      "step": 386360
    },
    {
      "epoch": 158.15,
      "grad_norm": 1.9911342859268188,
      "learning_rate": 5.389986583533833e-07,
      "loss": 0.4519,
      "step": 386370
    },
    {
      "epoch": 158.16,
      "grad_norm": 2.155395030975342,
      "learning_rate": 5.388974550879807e-07,
      "loss": 0.441,
      "step": 386380
    },
    {
      "epoch": 158.16,
      "grad_norm": 1.8141164779663086,
      "learning_rate": 5.387962600903278e-07,
      "loss": 0.4364,
      "step": 386390
    },
    {
      "epoch": 158.17,
      "grad_norm": 1.8655472993850708,
      "learning_rate": 5.38695073360889e-07,
      "loss": 0.4429,
      "step": 386400
    },
    {
      "epoch": 158.17,
      "grad_norm": 2.5122201442718506,
      "learning_rate": 5.38593894900127e-07,
      "loss": 0.4433,
      "step": 386410
    },
    {
      "epoch": 158.17,
      "grad_norm": 2.217400074005127,
      "learning_rate": 5.384927247085053e-07,
      "loss": 0.4554,
      "step": 386420
    },
    {
      "epoch": 158.18,
      "grad_norm": 2.239562511444092,
      "learning_rate": 5.38391562786488e-07,
      "loss": 0.453,
      "step": 386430
    },
    {
      "epoch": 158.18,
      "grad_norm": 2.3654537200927734,
      "learning_rate": 5.382904091345381e-07,
      "loss": 0.4373,
      "step": 386440
    },
    {
      "epoch": 158.19,
      "grad_norm": 2.1439220905303955,
      "learning_rate": 5.381892637531187e-07,
      "loss": 0.4458,
      "step": 386450
    },
    {
      "epoch": 158.19,
      "grad_norm": 4.033743858337402,
      "learning_rate": 5.380881266426937e-07,
      "loss": 0.4419,
      "step": 386460
    },
    {
      "epoch": 158.19,
      "grad_norm": 1.82902193069458,
      "learning_rate": 5.379869978037251e-07,
      "loss": 0.4587,
      "step": 386470
    },
    {
      "epoch": 158.2,
      "grad_norm": 2.408444881439209,
      "learning_rate": 5.378858772366779e-07,
      "loss": 0.4408,
      "step": 386480
    },
    {
      "epoch": 158.2,
      "grad_norm": 3.1018261909484863,
      "learning_rate": 5.377847649420149e-07,
      "loss": 0.4449,
      "step": 386490
    },
    {
      "epoch": 158.21,
      "grad_norm": 2.0735092163085938,
      "learning_rate": 5.376836609201993e-07,
      "loss": 0.4508,
      "step": 386500
    },
    {
      "epoch": 158.21,
      "grad_norm": 1.799078345298767,
      "learning_rate": 5.375825651716934e-07,
      "loss": 0.4436,
      "step": 386510
    },
    {
      "epoch": 158.22,
      "grad_norm": 1.9244368076324463,
      "learning_rate": 5.374814776969606e-07,
      "loss": 0.4431,
      "step": 386520
    },
    {
      "epoch": 158.22,
      "grad_norm": 2.0485310554504395,
      "learning_rate": 5.373803984964645e-07,
      "loss": 0.4633,
      "step": 386530
    },
    {
      "epoch": 158.22,
      "grad_norm": 1.753096580505371,
      "learning_rate": 5.372793275706676e-07,
      "loss": 0.4433,
      "step": 386540
    },
    {
      "epoch": 158.23,
      "grad_norm": 2.0883588790893555,
      "learning_rate": 5.371782649200329e-07,
      "loss": 0.4318,
      "step": 386550
    },
    {
      "epoch": 158.23,
      "grad_norm": 1.7122807502746582,
      "learning_rate": 5.370772105450236e-07,
      "loss": 0.4482,
      "step": 386560
    },
    {
      "epoch": 158.24,
      "grad_norm": 1.8858401775360107,
      "learning_rate": 5.369761644461026e-07,
      "loss": 0.4537,
      "step": 386570
    },
    {
      "epoch": 158.24,
      "grad_norm": 1.8741787672042847,
      "learning_rate": 5.36875126623733e-07,
      "loss": 0.4417,
      "step": 386580
    },
    {
      "epoch": 158.24,
      "grad_norm": 2.0697245597839355,
      "learning_rate": 5.367740970783771e-07,
      "loss": 0.4543,
      "step": 386590
    },
    {
      "epoch": 158.25,
      "grad_norm": 1.8511568307876587,
      "learning_rate": 5.36673075810498e-07,
      "loss": 0.4509,
      "step": 386600
    },
    {
      "epoch": 158.25,
      "grad_norm": 2.242331027984619,
      "learning_rate": 5.365720628205586e-07,
      "loss": 0.4547,
      "step": 386610
    },
    {
      "epoch": 158.26,
      "grad_norm": 2.0891404151916504,
      "learning_rate": 5.364710581090212e-07,
      "loss": 0.4398,
      "step": 386620
    },
    {
      "epoch": 158.26,
      "grad_norm": 2.1854686737060547,
      "learning_rate": 5.363700616763487e-07,
      "loss": 0.4666,
      "step": 386630
    },
    {
      "epoch": 158.26,
      "grad_norm": 2.3958239555358887,
      "learning_rate": 5.36269073523004e-07,
      "loss": 0.4469,
      "step": 386640
    },
    {
      "epoch": 158.27,
      "grad_norm": 2.2275795936584473,
      "learning_rate": 5.361680936494493e-07,
      "loss": 0.4375,
      "step": 386650
    },
    {
      "epoch": 158.27,
      "grad_norm": 1.9131947755813599,
      "learning_rate": 5.360671220561475e-07,
      "loss": 0.4504,
      "step": 386660
    },
    {
      "epoch": 158.28,
      "grad_norm": 1.909798264503479,
      "learning_rate": 5.359661587435611e-07,
      "loss": 0.4452,
      "step": 386670
    },
    {
      "epoch": 158.28,
      "grad_norm": 2.1600089073181152,
      "learning_rate": 5.358652037121523e-07,
      "loss": 0.4481,
      "step": 386680
    },
    {
      "epoch": 158.28,
      "grad_norm": 1.9695119857788086,
      "learning_rate": 5.357642569623839e-07,
      "loss": 0.4397,
      "step": 386690
    },
    {
      "epoch": 158.29,
      "grad_norm": 1.8711711168289185,
      "learning_rate": 5.356633184947182e-07,
      "loss": 0.4768,
      "step": 386700
    },
    {
      "epoch": 158.29,
      "grad_norm": 1.9738342761993408,
      "learning_rate": 5.355623883096174e-07,
      "loss": 0.4386,
      "step": 386710
    },
    {
      "epoch": 158.3,
      "grad_norm": 1.92986261844635,
      "learning_rate": 5.354614664075443e-07,
      "loss": 0.4552,
      "step": 386720
    },
    {
      "epoch": 158.3,
      "grad_norm": 1.9127931594848633,
      "learning_rate": 5.353605527889607e-07,
      "loss": 0.4334,
      "step": 386730
    },
    {
      "epoch": 158.31,
      "grad_norm": 1.9712071418762207,
      "learning_rate": 5.352596474543293e-07,
      "loss": 0.4369,
      "step": 386740
    },
    {
      "epoch": 158.31,
      "grad_norm": 2.130265474319458,
      "learning_rate": 5.35158750404112e-07,
      "loss": 0.4672,
      "step": 386750
    },
    {
      "epoch": 158.31,
      "grad_norm": 2.14518404006958,
      "learning_rate": 5.35057861638771e-07,
      "loss": 0.4468,
      "step": 386760
    },
    {
      "epoch": 158.32,
      "grad_norm": 2.01645827293396,
      "learning_rate": 5.349569811587687e-07,
      "loss": 0.4621,
      "step": 386770
    },
    {
      "epoch": 158.32,
      "grad_norm": 1.7242326736450195,
      "learning_rate": 5.34856108964567e-07,
      "loss": 0.4472,
      "step": 386780
    },
    {
      "epoch": 158.33,
      "grad_norm": 2.414696216583252,
      "learning_rate": 5.347552450566281e-07,
      "loss": 0.4565,
      "step": 386790
    },
    {
      "epoch": 158.33,
      "grad_norm": 1.9048192501068115,
      "learning_rate": 5.346543894354142e-07,
      "loss": 0.4295,
      "step": 386800
    },
    {
      "epoch": 158.33,
      "grad_norm": 1.7383053302764893,
      "learning_rate": 5.345535421013876e-07,
      "loss": 0.4641,
      "step": 386810
    },
    {
      "epoch": 158.34,
      "grad_norm": 1.8849411010742188,
      "learning_rate": 5.344527030550093e-07,
      "loss": 0.4671,
      "step": 386820
    },
    {
      "epoch": 158.34,
      "grad_norm": 2.2545437812805176,
      "learning_rate": 5.343518722967414e-07,
      "loss": 0.4427,
      "step": 386830
    },
    {
      "epoch": 158.35,
      "grad_norm": 2.0665807723999023,
      "learning_rate": 5.342510498270463e-07,
      "loss": 0.429,
      "step": 386840
    },
    {
      "epoch": 158.35,
      "grad_norm": 2.0006909370422363,
      "learning_rate": 5.341502356463849e-07,
      "loss": 0.4509,
      "step": 386850
    },
    {
      "epoch": 158.35,
      "grad_norm": 2.075160503387451,
      "learning_rate": 5.340494297552205e-07,
      "loss": 0.4484,
      "step": 386860
    },
    {
      "epoch": 158.36,
      "grad_norm": 2.1278018951416016,
      "learning_rate": 5.339486321540144e-07,
      "loss": 0.4527,
      "step": 386870
    },
    {
      "epoch": 158.36,
      "grad_norm": 1.8250609636306763,
      "learning_rate": 5.338478428432276e-07,
      "loss": 0.4295,
      "step": 386880
    },
    {
      "epoch": 158.37,
      "grad_norm": 1.997890591621399,
      "learning_rate": 5.337470618233225e-07,
      "loss": 0.45,
      "step": 386890
    },
    {
      "epoch": 158.37,
      "grad_norm": 1.9234063625335693,
      "learning_rate": 5.336462890947607e-07,
      "loss": 0.4554,
      "step": 386900
    },
    {
      "epoch": 158.37,
      "grad_norm": 1.8112657070159912,
      "learning_rate": 5.335455246580034e-07,
      "loss": 0.4256,
      "step": 386910
    },
    {
      "epoch": 158.38,
      "grad_norm": 1.8243213891983032,
      "learning_rate": 5.334447685135127e-07,
      "loss": 0.4599,
      "step": 386920
    },
    {
      "epoch": 158.38,
      "grad_norm": 2.5105769634246826,
      "learning_rate": 5.333440206617497e-07,
      "loss": 0.4559,
      "step": 386930
    },
    {
      "epoch": 158.39,
      "grad_norm": 1.783841609954834,
      "learning_rate": 5.332432811031763e-07,
      "loss": 0.4486,
      "step": 386940
    },
    {
      "epoch": 158.39,
      "grad_norm": 2.5264410972595215,
      "learning_rate": 5.331425498382543e-07,
      "loss": 0.4472,
      "step": 386950
    },
    {
      "epoch": 158.4,
      "grad_norm": 1.985224962234497,
      "learning_rate": 5.33041826867444e-07,
      "loss": 0.4533,
      "step": 386960
    },
    {
      "epoch": 158.4,
      "grad_norm": 1.9117399454116821,
      "learning_rate": 5.329411121912075e-07,
      "loss": 0.428,
      "step": 386970
    },
    {
      "epoch": 158.4,
      "grad_norm": 2.117975950241089,
      "learning_rate": 5.328404058100061e-07,
      "loss": 0.4375,
      "step": 386980
    },
    {
      "epoch": 158.41,
      "grad_norm": 1.909820556640625,
      "learning_rate": 5.327397077243008e-07,
      "loss": 0.4403,
      "step": 386990
    },
    {
      "epoch": 158.41,
      "grad_norm": 1.9620373249053955,
      "learning_rate": 5.326390179345534e-07,
      "loss": 0.4445,
      "step": 387000
    },
    {
      "epoch": 158.42,
      "grad_norm": 1.8903043270111084,
      "learning_rate": 5.32538336441225e-07,
      "loss": 0.4524,
      "step": 387010
    },
    {
      "epoch": 158.42,
      "grad_norm": 1.9580050706863403,
      "learning_rate": 5.324376632447765e-07,
      "loss": 0.4578,
      "step": 387020
    },
    {
      "epoch": 158.42,
      "grad_norm": 2.1624200344085693,
      "learning_rate": 5.323369983456694e-07,
      "loss": 0.4428,
      "step": 387030
    },
    {
      "epoch": 158.43,
      "grad_norm": 1.9903955459594727,
      "learning_rate": 5.322363417443642e-07,
      "loss": 0.435,
      "step": 387040
    },
    {
      "epoch": 158.43,
      "grad_norm": 1.7847929000854492,
      "learning_rate": 5.321356934413231e-07,
      "loss": 0.463,
      "step": 387050
    },
    {
      "epoch": 158.44,
      "grad_norm": 2.3223249912261963,
      "learning_rate": 5.320350534370067e-07,
      "loss": 0.4407,
      "step": 387060
    },
    {
      "epoch": 158.44,
      "grad_norm": 2.504542827606201,
      "learning_rate": 5.319344217318755e-07,
      "loss": 0.4519,
      "step": 387070
    },
    {
      "epoch": 158.44,
      "grad_norm": 2.0215771198272705,
      "learning_rate": 5.318337983263919e-07,
      "loss": 0.4425,
      "step": 387080
    },
    {
      "epoch": 158.45,
      "grad_norm": 1.9511739015579224,
      "learning_rate": 5.317331832210149e-07,
      "loss": 0.4534,
      "step": 387090
    },
    {
      "epoch": 158.45,
      "grad_norm": 2.0144190788269043,
      "learning_rate": 5.316325764162065e-07,
      "loss": 0.4533,
      "step": 387100
    },
    {
      "epoch": 158.46,
      "grad_norm": 2.0899994373321533,
      "learning_rate": 5.315319779124275e-07,
      "loss": 0.4461,
      "step": 387110
    },
    {
      "epoch": 158.46,
      "grad_norm": 1.8432459831237793,
      "learning_rate": 5.314313877101382e-07,
      "loss": 0.4424,
      "step": 387120
    },
    {
      "epoch": 158.47,
      "grad_norm": 2.7366437911987305,
      "learning_rate": 5.313308058098002e-07,
      "loss": 0.4355,
      "step": 387130
    },
    {
      "epoch": 158.47,
      "grad_norm": 2.2092063426971436,
      "learning_rate": 5.312302322118735e-07,
      "loss": 0.4597,
      "step": 387140
    },
    {
      "epoch": 158.47,
      "grad_norm": 2.077406883239746,
      "learning_rate": 5.311296669168194e-07,
      "loss": 0.4701,
      "step": 387150
    },
    {
      "epoch": 158.48,
      "grad_norm": 1.9582188129425049,
      "learning_rate": 5.310291099250982e-07,
      "loss": 0.4594,
      "step": 387160
    },
    {
      "epoch": 158.48,
      "grad_norm": 1.9804009199142456,
      "learning_rate": 5.309285612371706e-07,
      "loss": 0.4444,
      "step": 387170
    },
    {
      "epoch": 158.49,
      "grad_norm": 2.350130558013916,
      "learning_rate": 5.308280208534975e-07,
      "loss": 0.4429,
      "step": 387180
    },
    {
      "epoch": 158.49,
      "grad_norm": 2.6309316158294678,
      "learning_rate": 5.30727488774539e-07,
      "loss": 0.4576,
      "step": 387190
    },
    {
      "epoch": 158.49,
      "grad_norm": 1.8284040689468384,
      "learning_rate": 5.306269650007559e-07,
      "loss": 0.4574,
      "step": 387200
    },
    {
      "epoch": 158.5,
      "grad_norm": 1.8802627325057983,
      "learning_rate": 5.305264495326087e-07,
      "loss": 0.4199,
      "step": 387210
    },
    {
      "epoch": 158.5,
      "grad_norm": 1.9995179176330566,
      "learning_rate": 5.304259423705576e-07,
      "loss": 0.4291,
      "step": 387220
    },
    {
      "epoch": 158.51,
      "grad_norm": 2.170079469680786,
      "learning_rate": 5.303254435150633e-07,
      "loss": 0.4523,
      "step": 387230
    },
    {
      "epoch": 158.51,
      "grad_norm": 1.9726847410202026,
      "learning_rate": 5.302249529665859e-07,
      "loss": 0.4191,
      "step": 387240
    },
    {
      "epoch": 158.51,
      "grad_norm": 2.3785316944122314,
      "learning_rate": 5.30124470725586e-07,
      "loss": 0.4318,
      "step": 387250
    },
    {
      "epoch": 158.52,
      "grad_norm": 1.9166202545166016,
      "learning_rate": 5.300239967925238e-07,
      "loss": 0.4515,
      "step": 387260
    },
    {
      "epoch": 158.52,
      "grad_norm": 1.8815248012542725,
      "learning_rate": 5.299235311678595e-07,
      "loss": 0.4568,
      "step": 387270
    },
    {
      "epoch": 158.53,
      "grad_norm": 2.0821635723114014,
      "learning_rate": 5.298230738520536e-07,
      "loss": 0.4377,
      "step": 387280
    },
    {
      "epoch": 158.53,
      "grad_norm": 1.8822717666625977,
      "learning_rate": 5.297226248455656e-07,
      "loss": 0.476,
      "step": 387290
    },
    {
      "epoch": 158.53,
      "grad_norm": 2.2366135120391846,
      "learning_rate": 5.296221841488564e-07,
      "loss": 0.4374,
      "step": 387300
    },
    {
      "epoch": 158.54,
      "grad_norm": 1.857361078262329,
      "learning_rate": 5.295217517623855e-07,
      "loss": 0.4583,
      "step": 387310
    },
    {
      "epoch": 158.54,
      "grad_norm": 2.2826974391937256,
      "learning_rate": 5.294213276866134e-07,
      "loss": 0.4323,
      "step": 387320
    },
    {
      "epoch": 158.55,
      "grad_norm": 1.9109840393066406,
      "learning_rate": 5.293209119219999e-07,
      "loss": 0.4493,
      "step": 387330
    },
    {
      "epoch": 158.55,
      "grad_norm": 2.7062177658081055,
      "learning_rate": 5.292205044690052e-07,
      "loss": 0.4457,
      "step": 387340
    },
    {
      "epoch": 158.56,
      "grad_norm": 2.068444013595581,
      "learning_rate": 5.291201053280888e-07,
      "loss": 0.4375,
      "step": 387350
    },
    {
      "epoch": 158.56,
      "grad_norm": 1.8543673753738403,
      "learning_rate": 5.290197144997112e-07,
      "loss": 0.4264,
      "step": 387360
    },
    {
      "epoch": 158.56,
      "grad_norm": 2.0290069580078125,
      "learning_rate": 5.28919331984332e-07,
      "loss": 0.4253,
      "step": 387370
    },
    {
      "epoch": 158.57,
      "grad_norm": 2.1052427291870117,
      "learning_rate": 5.288189577824108e-07,
      "loss": 0.4386,
      "step": 387380
    },
    {
      "epoch": 158.57,
      "grad_norm": 1.97492516040802,
      "learning_rate": 5.287185918944085e-07,
      "loss": 0.4452,
      "step": 387390
    },
    {
      "epoch": 158.58,
      "grad_norm": 2.1627800464630127,
      "learning_rate": 5.286182343207831e-07,
      "loss": 0.4389,
      "step": 387400
    },
    {
      "epoch": 158.58,
      "grad_norm": 1.8577226400375366,
      "learning_rate": 5.285178850619954e-07,
      "loss": 0.4753,
      "step": 387410
    },
    {
      "epoch": 158.58,
      "grad_norm": 2.081088066101074,
      "learning_rate": 5.284175441185044e-07,
      "loss": 0.4755,
      "step": 387420
    },
    {
      "epoch": 158.59,
      "grad_norm": 3.287607431411743,
      "learning_rate": 5.283172114907704e-07,
      "loss": 0.4516,
      "step": 387430
    },
    {
      "epoch": 158.59,
      "grad_norm": 2.0386440753936768,
      "learning_rate": 5.282168871792533e-07,
      "loss": 0.4541,
      "step": 387440
    },
    {
      "epoch": 158.6,
      "grad_norm": 2.022737979888916,
      "learning_rate": 5.281165711844118e-07,
      "loss": 0.4482,
      "step": 387450
    },
    {
      "epoch": 158.6,
      "grad_norm": 2.683863401412964,
      "learning_rate": 5.280162635067062e-07,
      "loss": 0.4398,
      "step": 387460
    },
    {
      "epoch": 158.6,
      "grad_norm": 1.9786750078201294,
      "learning_rate": 5.279159641465956e-07,
      "loss": 0.4372,
      "step": 387470
    },
    {
      "epoch": 158.61,
      "grad_norm": 2.1041173934936523,
      "learning_rate": 5.278156731045394e-07,
      "loss": 0.4267,
      "step": 387480
    },
    {
      "epoch": 158.61,
      "grad_norm": 1.9380416870117188,
      "learning_rate": 5.277153903809972e-07,
      "loss": 0.4498,
      "step": 387490
    },
    {
      "epoch": 158.62,
      "grad_norm": 2.5203988552093506,
      "learning_rate": 5.276151159764285e-07,
      "loss": 0.4436,
      "step": 387500
    },
    {
      "epoch": 158.62,
      "grad_norm": 2.0154340267181396,
      "learning_rate": 5.27514849891292e-07,
      "loss": 0.4379,
      "step": 387510
    },
    {
      "epoch": 158.62,
      "grad_norm": 1.7086312770843506,
      "learning_rate": 5.274145921260475e-07,
      "loss": 0.4517,
      "step": 387520
    },
    {
      "epoch": 158.63,
      "grad_norm": 2.1188600063323975,
      "learning_rate": 5.273143426811553e-07,
      "loss": 0.4512,
      "step": 387530
    },
    {
      "epoch": 158.63,
      "grad_norm": 1.8561725616455078,
      "learning_rate": 5.272141015570724e-07,
      "loss": 0.4333,
      "step": 387540
    },
    {
      "epoch": 158.64,
      "grad_norm": 1.9144158363342285,
      "learning_rate": 5.271138687542596e-07,
      "loss": 0.4494,
      "step": 387550
    },
    {
      "epoch": 158.64,
      "grad_norm": 2.4683499336242676,
      "learning_rate": 5.27013644273175e-07,
      "loss": 0.4557,
      "step": 387560
    },
    {
      "epoch": 158.65,
      "grad_norm": 2.0890445709228516,
      "learning_rate": 5.269134281142786e-07,
      "loss": 0.4359,
      "step": 387570
    },
    {
      "epoch": 158.65,
      "grad_norm": 1.9417533874511719,
      "learning_rate": 5.268132202780292e-07,
      "loss": 0.4433,
      "step": 387580
    },
    {
      "epoch": 158.65,
      "grad_norm": 1.958634614944458,
      "learning_rate": 5.267130207648858e-07,
      "loss": 0.4634,
      "step": 387590
    },
    {
      "epoch": 158.66,
      "grad_norm": 1.9901025295257568,
      "learning_rate": 5.266128295753066e-07,
      "loss": 0.4438,
      "step": 387600
    },
    {
      "epoch": 158.66,
      "grad_norm": 2.4440009593963623,
      "learning_rate": 5.265126467097522e-07,
      "loss": 0.4596,
      "step": 387610
    },
    {
      "epoch": 158.67,
      "grad_norm": 1.9928269386291504,
      "learning_rate": 5.264124721686806e-07,
      "loss": 0.4583,
      "step": 387620
    },
    {
      "epoch": 158.67,
      "grad_norm": 1.8498787879943848,
      "learning_rate": 5.263123059525507e-07,
      "loss": 0.4558,
      "step": 387630
    },
    {
      "epoch": 158.67,
      "grad_norm": 1.7865835428237915,
      "learning_rate": 5.262121480618214e-07,
      "loss": 0.4528,
      "step": 387640
    },
    {
      "epoch": 158.68,
      "grad_norm": 2.2802212238311768,
      "learning_rate": 5.261119984969516e-07,
      "loss": 0.4453,
      "step": 387650
    },
    {
      "epoch": 158.68,
      "grad_norm": 3.122623920440674,
      "learning_rate": 5.260118572584006e-07,
      "loss": 0.4458,
      "step": 387660
    },
    {
      "epoch": 158.69,
      "grad_norm": 2.3168482780456543,
      "learning_rate": 5.259117243466259e-07,
      "loss": 0.4274,
      "step": 387670
    },
    {
      "epoch": 158.69,
      "grad_norm": 1.8802753686904907,
      "learning_rate": 5.258115997620869e-07,
      "loss": 0.4505,
      "step": 387680
    },
    {
      "epoch": 158.69,
      "grad_norm": 2.261312246322632,
      "learning_rate": 5.25711483505242e-07,
      "loss": 0.4427,
      "step": 387690
    },
    {
      "epoch": 158.7,
      "grad_norm": 4.735170841217041,
      "learning_rate": 5.2561137557655e-07,
      "loss": 0.4409,
      "step": 387700
    },
    {
      "epoch": 158.7,
      "grad_norm": 2.2139368057250977,
      "learning_rate": 5.255112759764696e-07,
      "loss": 0.4494,
      "step": 387710
    },
    {
      "epoch": 158.71,
      "grad_norm": 1.529837965965271,
      "learning_rate": 5.25411184705459e-07,
      "loss": 0.4354,
      "step": 387720
    },
    {
      "epoch": 158.71,
      "grad_norm": 2.3478736877441406,
      "learning_rate": 5.25311101763977e-07,
      "loss": 0.4609,
      "step": 387730
    },
    {
      "epoch": 158.71,
      "grad_norm": 2.050241708755493,
      "learning_rate": 5.252110271524821e-07,
      "loss": 0.452,
      "step": 387740
    },
    {
      "epoch": 158.72,
      "grad_norm": 1.9651212692260742,
      "learning_rate": 5.251109608714325e-07,
      "loss": 0.4556,
      "step": 387750
    },
    {
      "epoch": 158.72,
      "grad_norm": 2.24151611328125,
      "learning_rate": 5.250109029212868e-07,
      "loss": 0.4357,
      "step": 387760
    },
    {
      "epoch": 158.73,
      "grad_norm": 1.7799303531646729,
      "learning_rate": 5.24910853302503e-07,
      "loss": 0.4479,
      "step": 387770
    },
    {
      "epoch": 158.73,
      "grad_norm": 1.7682863473892212,
      "learning_rate": 5.248108120155396e-07,
      "loss": 0.4337,
      "step": 387780
    },
    {
      "epoch": 158.74,
      "grad_norm": 2.005279302597046,
      "learning_rate": 5.247107790608547e-07,
      "loss": 0.447,
      "step": 387790
    },
    {
      "epoch": 158.74,
      "grad_norm": 1.9568380117416382,
      "learning_rate": 5.246107544389079e-07,
      "loss": 0.4482,
      "step": 387800
    },
    {
      "epoch": 158.74,
      "grad_norm": 1.9662768840789795,
      "learning_rate": 5.245107381501552e-07,
      "loss": 0.4165,
      "step": 387810
    },
    {
      "epoch": 158.75,
      "grad_norm": 2.2814040184020996,
      "learning_rate": 5.244107301950563e-07,
      "loss": 0.467,
      "step": 387820
    },
    {
      "epoch": 158.75,
      "grad_norm": 1.7832286357879639,
      "learning_rate": 5.243107305740685e-07,
      "loss": 0.4262,
      "step": 387830
    },
    {
      "epoch": 158.76,
      "grad_norm": 1.931339144706726,
      "learning_rate": 5.242107392876506e-07,
      "loss": 0.4465,
      "step": 387840
    },
    {
      "epoch": 158.76,
      "grad_norm": 1.9989057779312134,
      "learning_rate": 5.2411075633626e-07,
      "loss": 0.4696,
      "step": 387850
    },
    {
      "epoch": 158.76,
      "grad_norm": 2.3810198307037354,
      "learning_rate": 5.24010781720355e-07,
      "loss": 0.4346,
      "step": 387860
    },
    {
      "epoch": 158.77,
      "grad_norm": 1.9774670600891113,
      "learning_rate": 5.239108154403934e-07,
      "loss": 0.4532,
      "step": 387870
    },
    {
      "epoch": 158.77,
      "grad_norm": 1.9815303087234497,
      "learning_rate": 5.238108574968334e-07,
      "loss": 0.4445,
      "step": 387880
    },
    {
      "epoch": 158.78,
      "grad_norm": 2.271548271179199,
      "learning_rate": 5.237109078901329e-07,
      "loss": 0.4331,
      "step": 387890
    },
    {
      "epoch": 158.78,
      "grad_norm": 1.7757744789123535,
      "learning_rate": 5.236109666207494e-07,
      "loss": 0.4458,
      "step": 387900
    },
    {
      "epoch": 158.78,
      "grad_norm": 2.5764434337615967,
      "learning_rate": 5.23511033689141e-07,
      "loss": 0.4518,
      "step": 387910
    },
    {
      "epoch": 158.79,
      "grad_norm": 1.9795769453048706,
      "learning_rate": 5.234111090957654e-07,
      "loss": 0.4457,
      "step": 387920
    },
    {
      "epoch": 158.79,
      "grad_norm": 2.0333540439605713,
      "learning_rate": 5.233111928410804e-07,
      "loss": 0.4293,
      "step": 387930
    },
    {
      "epoch": 158.8,
      "grad_norm": 2.311594009399414,
      "learning_rate": 5.232112849255437e-07,
      "loss": 0.456,
      "step": 387940
    },
    {
      "epoch": 158.8,
      "grad_norm": 2.094144821166992,
      "learning_rate": 5.231113853496126e-07,
      "loss": 0.4375,
      "step": 387950
    },
    {
      "epoch": 158.8,
      "grad_norm": 2.44960355758667,
      "learning_rate": 5.230114941137452e-07,
      "loss": 0.4543,
      "step": 387960
    },
    {
      "epoch": 158.81,
      "grad_norm": 1.7385402917861938,
      "learning_rate": 5.229116112183997e-07,
      "loss": 0.4393,
      "step": 387970
    },
    {
      "epoch": 158.81,
      "grad_norm": 1.8580002784729004,
      "learning_rate": 5.228117366640315e-07,
      "loss": 0.4531,
      "step": 387980
    },
    {
      "epoch": 158.82,
      "grad_norm": 2.292553424835205,
      "learning_rate": 5.227118704511003e-07,
      "loss": 0.4546,
      "step": 387990
    },
    {
      "epoch": 158.82,
      "grad_norm": 2.03913950920105,
      "learning_rate": 5.226120125800623e-07,
      "loss": 0.4468,
      "step": 388000
    },
    {
      "epoch": 158.83,
      "grad_norm": 2.332166910171509,
      "learning_rate": 5.225121630513757e-07,
      "loss": 0.4373,
      "step": 388010
    },
    {
      "epoch": 158.83,
      "grad_norm": 1.8796281814575195,
      "learning_rate": 5.224123218654975e-07,
      "loss": 0.4492,
      "step": 388020
    },
    {
      "epoch": 158.83,
      "grad_norm": 1.9163944721221924,
      "learning_rate": 5.223124890228852e-07,
      "loss": 0.4429,
      "step": 388030
    },
    {
      "epoch": 158.84,
      "grad_norm": 1.9033807516098022,
      "learning_rate": 5.22212664523996e-07,
      "loss": 0.444,
      "step": 388040
    },
    {
      "epoch": 158.84,
      "grad_norm": 1.7173796892166138,
      "learning_rate": 5.221128483692872e-07,
      "loss": 0.4474,
      "step": 388050
    },
    {
      "epoch": 158.85,
      "grad_norm": 2.042442560195923,
      "learning_rate": 5.22013040559216e-07,
      "loss": 0.4603,
      "step": 388060
    },
    {
      "epoch": 158.85,
      "grad_norm": 1.8744959831237793,
      "learning_rate": 5.219132410942399e-07,
      "loss": 0.4458,
      "step": 388070
    },
    {
      "epoch": 158.85,
      "grad_norm": 2.144373655319214,
      "learning_rate": 5.218134499748156e-07,
      "loss": 0.4175,
      "step": 388080
    },
    {
      "epoch": 158.86,
      "grad_norm": 1.8175544738769531,
      "learning_rate": 5.217136672014008e-07,
      "loss": 0.4676,
      "step": 388090
    },
    {
      "epoch": 158.86,
      "grad_norm": 2.111332416534424,
      "learning_rate": 5.21613892774452e-07,
      "loss": 0.4466,
      "step": 388100
    },
    {
      "epoch": 158.87,
      "grad_norm": 1.9334267377853394,
      "learning_rate": 5.215141266944272e-07,
      "loss": 0.457,
      "step": 388110
    },
    {
      "epoch": 158.87,
      "grad_norm": 2.1062629222869873,
      "learning_rate": 5.214143689617821e-07,
      "loss": 0.4408,
      "step": 388120
    },
    {
      "epoch": 158.87,
      "grad_norm": 1.92814040184021,
      "learning_rate": 5.213146195769743e-07,
      "loss": 0.4545,
      "step": 388130
    },
    {
      "epoch": 158.88,
      "grad_norm": 1.9278619289398193,
      "learning_rate": 5.212148785404607e-07,
      "loss": 0.4566,
      "step": 388140
    },
    {
      "epoch": 158.88,
      "grad_norm": 2.112301826477051,
      "learning_rate": 5.211151458526982e-07,
      "loss": 0.4361,
      "step": 388150
    },
    {
      "epoch": 158.89,
      "grad_norm": 1.9285639524459839,
      "learning_rate": 5.210154215141438e-07,
      "loss": 0.4541,
      "step": 388160
    },
    {
      "epoch": 158.89,
      "grad_norm": 1.9215542078018188,
      "learning_rate": 5.209157055252536e-07,
      "loss": 0.4472,
      "step": 388170
    },
    {
      "epoch": 158.89,
      "grad_norm": 1.6102898120880127,
      "learning_rate": 5.208159978864857e-07,
      "loss": 0.4331,
      "step": 388180
    },
    {
      "epoch": 158.9,
      "grad_norm": 1.9173521995544434,
      "learning_rate": 5.207162985982959e-07,
      "loss": 0.462,
      "step": 388190
    },
    {
      "epoch": 158.9,
      "grad_norm": 2.0632643699645996,
      "learning_rate": 5.206166076611413e-07,
      "loss": 0.4264,
      "step": 388200
    },
    {
      "epoch": 158.91,
      "grad_norm": 1.8409101963043213,
      "learning_rate": 5.205169250754785e-07,
      "loss": 0.4523,
      "step": 388210
    },
    {
      "epoch": 158.91,
      "grad_norm": 2.1973583698272705,
      "learning_rate": 5.204172508417641e-07,
      "loss": 0.4588,
      "step": 388220
    },
    {
      "epoch": 158.92,
      "grad_norm": 1.9999183416366577,
      "learning_rate": 5.203175849604544e-07,
      "loss": 0.4479,
      "step": 388230
    },
    {
      "epoch": 158.92,
      "grad_norm": 2.223816156387329,
      "learning_rate": 5.202179274320064e-07,
      "loss": 0.4368,
      "step": 388240
    },
    {
      "epoch": 158.92,
      "grad_norm": 2.241135358810425,
      "learning_rate": 5.20118278256877e-07,
      "loss": 0.4327,
      "step": 388250
    },
    {
      "epoch": 158.93,
      "grad_norm": 1.906752109527588,
      "learning_rate": 5.200186374355214e-07,
      "loss": 0.4609,
      "step": 388260
    },
    {
      "epoch": 158.93,
      "grad_norm": 2.1172025203704834,
      "learning_rate": 5.199190049683967e-07,
      "loss": 0.443,
      "step": 388270
    },
    {
      "epoch": 158.94,
      "grad_norm": 1.9272695779800415,
      "learning_rate": 5.198193808559595e-07,
      "loss": 0.4589,
      "step": 388280
    },
    {
      "epoch": 158.94,
      "grad_norm": 2.0503480434417725,
      "learning_rate": 5.197197650986657e-07,
      "loss": 0.4567,
      "step": 388290
    },
    {
      "epoch": 158.94,
      "grad_norm": 1.9452589750289917,
      "learning_rate": 5.196201576969723e-07,
      "loss": 0.4669,
      "step": 388300
    },
    {
      "epoch": 158.95,
      "grad_norm": 2.1868176460266113,
      "learning_rate": 5.19520558651335e-07,
      "loss": 0.4544,
      "step": 388310
    },
    {
      "epoch": 158.95,
      "grad_norm": 2.2124669551849365,
      "learning_rate": 5.194209679622101e-07,
      "loss": 0.4417,
      "step": 388320
    },
    {
      "epoch": 158.96,
      "grad_norm": 1.6229982376098633,
      "learning_rate": 5.193213856300543e-07,
      "loss": 0.4619,
      "step": 388330
    },
    {
      "epoch": 158.96,
      "grad_norm": 1.9481836557388306,
      "learning_rate": 5.19221811655323e-07,
      "loss": 0.4448,
      "step": 388340
    },
    {
      "epoch": 158.96,
      "grad_norm": 2.0957329273223877,
      "learning_rate": 5.191222460384731e-07,
      "loss": 0.4489,
      "step": 388350
    },
    {
      "epoch": 158.97,
      "grad_norm": 1.8477134704589844,
      "learning_rate": 5.190226887799596e-07,
      "loss": 0.4829,
      "step": 388360
    },
    {
      "epoch": 158.97,
      "grad_norm": 2.6933300495147705,
      "learning_rate": 5.1892313988024e-07,
      "loss": 0.4645,
      "step": 388370
    },
    {
      "epoch": 158.98,
      "grad_norm": 2.4671874046325684,
      "learning_rate": 5.188235993397704e-07,
      "loss": 0.4426,
      "step": 388380
    },
    {
      "epoch": 158.98,
      "grad_norm": 1.6842657327651978,
      "learning_rate": 5.187240671590052e-07,
      "loss": 0.447,
      "step": 388390
    },
    {
      "epoch": 158.98,
      "grad_norm": 3.2050790786743164,
      "learning_rate": 5.186245433384014e-07,
      "loss": 0.4519,
      "step": 388400
    },
    {
      "epoch": 158.99,
      "grad_norm": 2.226837158203125,
      "learning_rate": 5.185250278784143e-07,
      "loss": 0.4555,
      "step": 388410
    },
    {
      "epoch": 158.99,
      "grad_norm": 1.9100826978683472,
      "learning_rate": 5.184255207795004e-07,
      "loss": 0.4493,
      "step": 388420
    },
    {
      "epoch": 159.0,
      "grad_norm": 2.085498571395874,
      "learning_rate": 5.18326022042115e-07,
      "loss": 0.4578,
      "step": 388430
    },
    {
      "epoch": 159.0,
      "eval_loss": 0.44651904702186584,
      "eval_runtime": 52.4936,
      "eval_samples_per_second": 65.703,
      "eval_steps_per_second": 8.23,
      "step": 388437
    },
    {
      "epoch": 159.0,
      "grad_norm": 2.6557397842407227,
      "learning_rate": 5.182265316667143e-07,
      "loss": 0.4323,
      "step": 388440
    },
    {
      "epoch": 159.01,
      "grad_norm": 2.1444103717803955,
      "learning_rate": 5.181270496537539e-07,
      "loss": 0.4429,
      "step": 388450
    },
    {
      "epoch": 159.01,
      "grad_norm": 1.78685462474823,
      "learning_rate": 5.180275760036895e-07,
      "loss": 0.4493,
      "step": 388460
    },
    {
      "epoch": 159.01,
      "grad_norm": 1.8130912780761719,
      "learning_rate": 5.179281107169767e-07,
      "loss": 0.4486,
      "step": 388470
    },
    {
      "epoch": 159.02,
      "grad_norm": 2.1091809272766113,
      "learning_rate": 5.178286537940713e-07,
      "loss": 0.4556,
      "step": 388480
    },
    {
      "epoch": 159.02,
      "grad_norm": 2.111931085586548,
      "learning_rate": 5.177292052354288e-07,
      "loss": 0.4516,
      "step": 388490
    },
    {
      "epoch": 159.03,
      "grad_norm": 1.7347232103347778,
      "learning_rate": 5.176297650415048e-07,
      "loss": 0.44,
      "step": 388500
    },
    {
      "epoch": 159.03,
      "grad_norm": 1.8109408617019653,
      "learning_rate": 5.175303332127548e-07,
      "loss": 0.4367,
      "step": 388510
    },
    {
      "epoch": 159.03,
      "grad_norm": 3.0365872383117676,
      "learning_rate": 5.174309097496343e-07,
      "loss": 0.4484,
      "step": 388520
    },
    {
      "epoch": 159.04,
      "grad_norm": 2.266132116317749,
      "learning_rate": 5.173314946525984e-07,
      "loss": 0.4707,
      "step": 388530
    },
    {
      "epoch": 159.04,
      "grad_norm": 1.8535740375518799,
      "learning_rate": 5.172320879221032e-07,
      "loss": 0.4523,
      "step": 388540
    },
    {
      "epoch": 159.05,
      "grad_norm": 1.9804376363754272,
      "learning_rate": 5.171326895586035e-07,
      "loss": 0.4551,
      "step": 388550
    },
    {
      "epoch": 159.05,
      "grad_norm": 2.061330556869507,
      "learning_rate": 5.170332995625548e-07,
      "loss": 0.445,
      "step": 388560
    },
    {
      "epoch": 159.05,
      "grad_norm": 1.9518340826034546,
      "learning_rate": 5.169339179344125e-07,
      "loss": 0.4406,
      "step": 388570
    },
    {
      "epoch": 159.06,
      "grad_norm": 1.830625295639038,
      "learning_rate": 5.168345446746317e-07,
      "loss": 0.4491,
      "step": 388580
    },
    {
      "epoch": 159.06,
      "grad_norm": 2.205807685852051,
      "learning_rate": 5.167351797836675e-07,
      "loss": 0.4447,
      "step": 388590
    },
    {
      "epoch": 159.07,
      "grad_norm": 4.625211238861084,
      "learning_rate": 5.166358232619754e-07,
      "loss": 0.4496,
      "step": 388600
    },
    {
      "epoch": 159.07,
      "grad_norm": 1.7214187383651733,
      "learning_rate": 5.165364751100104e-07,
      "loss": 0.4459,
      "step": 388610
    },
    {
      "epoch": 159.07,
      "grad_norm": 2.315321922302246,
      "learning_rate": 5.164371353282276e-07,
      "loss": 0.4562,
      "step": 388620
    },
    {
      "epoch": 159.08,
      "grad_norm": 1.839963436126709,
      "learning_rate": 5.16337803917082e-07,
      "loss": 0.4545,
      "step": 388630
    },
    {
      "epoch": 159.08,
      "grad_norm": 1.7842650413513184,
      "learning_rate": 5.162384808770285e-07,
      "loss": 0.426,
      "step": 388640
    },
    {
      "epoch": 159.09,
      "grad_norm": 1.8185631036758423,
      "learning_rate": 5.161391662085225e-07,
      "loss": 0.4568,
      "step": 388650
    },
    {
      "epoch": 159.09,
      "grad_norm": 2.1002516746520996,
      "learning_rate": 5.160398599120185e-07,
      "loss": 0.4481,
      "step": 388660
    },
    {
      "epoch": 159.1,
      "grad_norm": 2.1248018741607666,
      "learning_rate": 5.159405619879716e-07,
      "loss": 0.4392,
      "step": 388670
    },
    {
      "epoch": 159.1,
      "grad_norm": 2.574662923812866,
      "learning_rate": 5.158412724368368e-07,
      "loss": 0.4505,
      "step": 388680
    },
    {
      "epoch": 159.1,
      "grad_norm": 2.319512128829956,
      "learning_rate": 5.157419912590693e-07,
      "loss": 0.4747,
      "step": 388690
    },
    {
      "epoch": 159.11,
      "grad_norm": 1.9952720403671265,
      "learning_rate": 5.15642718455123e-07,
      "loss": 0.4386,
      "step": 388700
    },
    {
      "epoch": 159.11,
      "grad_norm": 2.148245334625244,
      "learning_rate": 5.155434540254528e-07,
      "loss": 0.4432,
      "step": 388710
    },
    {
      "epoch": 159.12,
      "grad_norm": 1.967256784439087,
      "learning_rate": 5.154441979705136e-07,
      "loss": 0.4374,
      "step": 388720
    },
    {
      "epoch": 159.12,
      "grad_norm": 2.28437876701355,
      "learning_rate": 5.153449502907604e-07,
      "loss": 0.4671,
      "step": 388730
    },
    {
      "epoch": 159.12,
      "grad_norm": 1.8386887311935425,
      "learning_rate": 5.152457109866467e-07,
      "loss": 0.4535,
      "step": 388740
    },
    {
      "epoch": 159.13,
      "grad_norm": 2.148068428039551,
      "learning_rate": 5.151464800586286e-07,
      "loss": 0.4561,
      "step": 388750
    },
    {
      "epoch": 159.13,
      "grad_norm": 1.9884108304977417,
      "learning_rate": 5.150472575071603e-07,
      "loss": 0.4454,
      "step": 388760
    },
    {
      "epoch": 159.14,
      "grad_norm": 1.890129566192627,
      "learning_rate": 5.149480433326957e-07,
      "loss": 0.4536,
      "step": 388770
    },
    {
      "epoch": 159.14,
      "grad_norm": 1.9937641620635986,
      "learning_rate": 5.148488375356898e-07,
      "loss": 0.44,
      "step": 388780
    },
    {
      "epoch": 159.14,
      "grad_norm": 3.49127197265625,
      "learning_rate": 5.147496401165968e-07,
      "loss": 0.4555,
      "step": 388790
    },
    {
      "epoch": 159.15,
      "grad_norm": 2.256645679473877,
      "learning_rate": 5.146504510758715e-07,
      "loss": 0.4356,
      "step": 388800
    },
    {
      "epoch": 159.15,
      "grad_norm": 1.8638979196548462,
      "learning_rate": 5.145512704139675e-07,
      "loss": 0.4447,
      "step": 388810
    },
    {
      "epoch": 159.16,
      "grad_norm": 2.421654462814331,
      "learning_rate": 5.144520981313399e-07,
      "loss": 0.4396,
      "step": 388820
    },
    {
      "epoch": 159.16,
      "grad_norm": 1.9115971326828003,
      "learning_rate": 5.143529342284432e-07,
      "loss": 0.4444,
      "step": 388830
    },
    {
      "epoch": 159.16,
      "grad_norm": 2.27180814743042,
      "learning_rate": 5.142537787057306e-07,
      "loss": 0.4515,
      "step": 388840
    },
    {
      "epoch": 159.17,
      "grad_norm": 2.2542548179626465,
      "learning_rate": 5.141546315636569e-07,
      "loss": 0.4563,
      "step": 388850
    },
    {
      "epoch": 159.17,
      "grad_norm": 2.1077895164489746,
      "learning_rate": 5.140554928026762e-07,
      "loss": 0.443,
      "step": 388860
    },
    {
      "epoch": 159.18,
      "grad_norm": 2.0064127445220947,
      "learning_rate": 5.139563624232426e-07,
      "loss": 0.4381,
      "step": 388870
    },
    {
      "epoch": 159.18,
      "grad_norm": 1.9989978075027466,
      "learning_rate": 5.138572404258102e-07,
      "loss": 0.4379,
      "step": 388880
    },
    {
      "epoch": 159.19,
      "grad_norm": 2.241530179977417,
      "learning_rate": 5.137581268108332e-07,
      "loss": 0.4584,
      "step": 388890
    },
    {
      "epoch": 159.19,
      "grad_norm": 1.963381290435791,
      "learning_rate": 5.136590215787657e-07,
      "loss": 0.4362,
      "step": 388900
    },
    {
      "epoch": 159.19,
      "grad_norm": 2.1787900924682617,
      "learning_rate": 5.135599247300614e-07,
      "loss": 0.4473,
      "step": 388910
    },
    {
      "epoch": 159.2,
      "grad_norm": 2.2510414123535156,
      "learning_rate": 5.134608362651738e-07,
      "loss": 0.4477,
      "step": 388920
    },
    {
      "epoch": 159.2,
      "grad_norm": 2.0555739402770996,
      "learning_rate": 5.13361756184558e-07,
      "loss": 0.4603,
      "step": 388930
    },
    {
      "epoch": 159.21,
      "grad_norm": 2.0278377532958984,
      "learning_rate": 5.132626844886675e-07,
      "loss": 0.4489,
      "step": 388940
    },
    {
      "epoch": 159.21,
      "grad_norm": 2.5198397636413574,
      "learning_rate": 5.131636211779557e-07,
      "loss": 0.4709,
      "step": 388950
    },
    {
      "epoch": 159.21,
      "grad_norm": 2.352111339569092,
      "learning_rate": 5.130645662528764e-07,
      "loss": 0.4493,
      "step": 388960
    },
    {
      "epoch": 159.22,
      "grad_norm": 2.939692735671997,
      "learning_rate": 5.129655197138845e-07,
      "loss": 0.4405,
      "step": 388970
    },
    {
      "epoch": 159.22,
      "grad_norm": 1.8885133266448975,
      "learning_rate": 5.12866481561432e-07,
      "loss": 0.4266,
      "step": 388980
    },
    {
      "epoch": 159.23,
      "grad_norm": 1.9148012399673462,
      "learning_rate": 5.127674517959735e-07,
      "loss": 0.4398,
      "step": 388990
    },
    {
      "epoch": 159.23,
      "grad_norm": 2.109243392944336,
      "learning_rate": 5.126684304179622e-07,
      "loss": 0.4641,
      "step": 389000
    },
    {
      "epoch": 159.23,
      "grad_norm": 1.828688144683838,
      "learning_rate": 5.125694174278522e-07,
      "loss": 0.4449,
      "step": 389010
    },
    {
      "epoch": 159.24,
      "grad_norm": 1.9051045179367065,
      "learning_rate": 5.124704128260969e-07,
      "loss": 0.4491,
      "step": 389020
    },
    {
      "epoch": 159.24,
      "grad_norm": 2.1521499156951904,
      "learning_rate": 5.123714166131497e-07,
      "loss": 0.432,
      "step": 389030
    },
    {
      "epoch": 159.25,
      "grad_norm": 2.823880910873413,
      "learning_rate": 5.122724287894641e-07,
      "loss": 0.449,
      "step": 389040
    },
    {
      "epoch": 159.25,
      "grad_norm": 1.8908944129943848,
      "learning_rate": 5.121734493554938e-07,
      "loss": 0.4502,
      "step": 389050
    },
    {
      "epoch": 159.26,
      "grad_norm": 2.1643166542053223,
      "learning_rate": 5.12074478311692e-07,
      "loss": 0.4597,
      "step": 389060
    },
    {
      "epoch": 159.26,
      "grad_norm": 2.0576541423797607,
      "learning_rate": 5.119755156585119e-07,
      "loss": 0.4495,
      "step": 389070
    },
    {
      "epoch": 159.26,
      "grad_norm": 2.0903282165527344,
      "learning_rate": 5.11876561396407e-07,
      "loss": 0.4602,
      "step": 389080
    },
    {
      "epoch": 159.27,
      "grad_norm": 2.2215871810913086,
      "learning_rate": 5.11777615525831e-07,
      "loss": 0.4276,
      "step": 389090
    },
    {
      "epoch": 159.27,
      "grad_norm": 1.753411054611206,
      "learning_rate": 5.116786780472365e-07,
      "loss": 0.4509,
      "step": 389100
    },
    {
      "epoch": 159.28,
      "grad_norm": 2.1613411903381348,
      "learning_rate": 5.11579748961077e-07,
      "loss": 0.428,
      "step": 389110
    },
    {
      "epoch": 159.28,
      "grad_norm": 2.2630503177642822,
      "learning_rate": 5.114808282678057e-07,
      "loss": 0.4596,
      "step": 389120
    },
    {
      "epoch": 159.28,
      "grad_norm": 2.5340020656585693,
      "learning_rate": 5.113819159678757e-07,
      "loss": 0.4474,
      "step": 389130
    },
    {
      "epoch": 159.29,
      "grad_norm": 2.268442153930664,
      "learning_rate": 5.112830120617401e-07,
      "loss": 0.4217,
      "step": 389140
    },
    {
      "epoch": 159.29,
      "grad_norm": 2.0004379749298096,
      "learning_rate": 5.11184116549852e-07,
      "loss": 0.4107,
      "step": 389150
    },
    {
      "epoch": 159.3,
      "grad_norm": 2.1177992820739746,
      "learning_rate": 5.110852294326643e-07,
      "loss": 0.4461,
      "step": 389160
    },
    {
      "epoch": 159.3,
      "grad_norm": 1.7102155685424805,
      "learning_rate": 5.109863507106301e-07,
      "loss": 0.4624,
      "step": 389170
    },
    {
      "epoch": 159.3,
      "grad_norm": 1.9514325857162476,
      "learning_rate": 5.108874803842027e-07,
      "loss": 0.4502,
      "step": 389180
    },
    {
      "epoch": 159.31,
      "grad_norm": 2.063178300857544,
      "learning_rate": 5.107886184538343e-07,
      "loss": 0.4449,
      "step": 389190
    },
    {
      "epoch": 159.31,
      "grad_norm": 2.1054153442382812,
      "learning_rate": 5.106897649199782e-07,
      "loss": 0.4254,
      "step": 389200
    },
    {
      "epoch": 159.32,
      "grad_norm": 2.495266914367676,
      "learning_rate": 5.105909197830871e-07,
      "loss": 0.459,
      "step": 389210
    },
    {
      "epoch": 159.32,
      "grad_norm": 1.8400752544403076,
      "learning_rate": 5.104920830436138e-07,
      "loss": 0.4502,
      "step": 389220
    },
    {
      "epoch": 159.32,
      "grad_norm": 2.079496145248413,
      "learning_rate": 5.103932547020113e-07,
      "loss": 0.4373,
      "step": 389230
    },
    {
      "epoch": 159.33,
      "grad_norm": 2.499108076095581,
      "learning_rate": 5.102944347587319e-07,
      "loss": 0.4476,
      "step": 389240
    },
    {
      "epoch": 159.33,
      "grad_norm": 2.735896110534668,
      "learning_rate": 5.101956232142288e-07,
      "loss": 0.4368,
      "step": 389250
    },
    {
      "epoch": 159.34,
      "grad_norm": 1.7055909633636475,
      "learning_rate": 5.100968200689541e-07,
      "loss": 0.438,
      "step": 389260
    },
    {
      "epoch": 159.34,
      "grad_norm": 1.9390969276428223,
      "learning_rate": 5.099980253233606e-07,
      "loss": 0.4471,
      "step": 389270
    },
    {
      "epoch": 159.35,
      "grad_norm": 2.0931196212768555,
      "learning_rate": 5.098992389779018e-07,
      "loss": 0.4549,
      "step": 389280
    },
    {
      "epoch": 159.35,
      "grad_norm": 1.8654067516326904,
      "learning_rate": 5.098004610330285e-07,
      "loss": 0.4499,
      "step": 389290
    },
    {
      "epoch": 159.35,
      "grad_norm": 1.9996857643127441,
      "learning_rate": 5.097016914891935e-07,
      "loss": 0.4473,
      "step": 389300
    },
    {
      "epoch": 159.36,
      "grad_norm": 1.9897773265838623,
      "learning_rate": 5.096029303468501e-07,
      "loss": 0.4352,
      "step": 389310
    },
    {
      "epoch": 159.36,
      "grad_norm": 1.899812936782837,
      "learning_rate": 5.095041776064507e-07,
      "loss": 0.4549,
      "step": 389320
    },
    {
      "epoch": 159.37,
      "grad_norm": 1.8214763402938843,
      "learning_rate": 5.094054332684474e-07,
      "loss": 0.4493,
      "step": 389330
    },
    {
      "epoch": 159.37,
      "grad_norm": 2.044423818588257,
      "learning_rate": 5.093066973332921e-07,
      "loss": 0.442,
      "step": 389340
    },
    {
      "epoch": 159.37,
      "grad_norm": 1.9581348896026611,
      "learning_rate": 5.092079698014376e-07,
      "loss": 0.4468,
      "step": 389350
    },
    {
      "epoch": 159.38,
      "grad_norm": 2.1646690368652344,
      "learning_rate": 5.09109250673336e-07,
      "loss": 0.4328,
      "step": 389360
    },
    {
      "epoch": 159.38,
      "grad_norm": 1.7449884414672852,
      "learning_rate": 5.090105399494398e-07,
      "loss": 0.4387,
      "step": 389370
    },
    {
      "epoch": 159.39,
      "grad_norm": 2.14113187789917,
      "learning_rate": 5.089118376302008e-07,
      "loss": 0.4452,
      "step": 389380
    },
    {
      "epoch": 159.39,
      "grad_norm": 2.2929720878601074,
      "learning_rate": 5.088131437160711e-07,
      "loss": 0.4406,
      "step": 389390
    },
    {
      "epoch": 159.39,
      "grad_norm": 2.3105196952819824,
      "learning_rate": 5.087144582075033e-07,
      "loss": 0.4692,
      "step": 389400
    },
    {
      "epoch": 159.4,
      "grad_norm": 2.3220057487487793,
      "learning_rate": 5.086157811049493e-07,
      "loss": 0.4437,
      "step": 389410
    },
    {
      "epoch": 159.4,
      "grad_norm": 2.0308804512023926,
      "learning_rate": 5.085171124088608e-07,
      "loss": 0.4513,
      "step": 389420
    },
    {
      "epoch": 159.41,
      "grad_norm": 2.331791400909424,
      "learning_rate": 5.084184521196897e-07,
      "loss": 0.4351,
      "step": 389430
    },
    {
      "epoch": 159.41,
      "grad_norm": 2.156674861907959,
      "learning_rate": 5.08319800237888e-07,
      "loss": 0.4505,
      "step": 389440
    },
    {
      "epoch": 159.41,
      "grad_norm": 2.243729591369629,
      "learning_rate": 5.08221156763908e-07,
      "loss": 0.4402,
      "step": 389450
    },
    {
      "epoch": 159.42,
      "grad_norm": 1.987452507019043,
      "learning_rate": 5.081225216982012e-07,
      "loss": 0.4414,
      "step": 389460
    },
    {
      "epoch": 159.42,
      "grad_norm": 2.04428768157959,
      "learning_rate": 5.080238950412198e-07,
      "loss": 0.4631,
      "step": 389470
    },
    {
      "epoch": 159.43,
      "grad_norm": 1.9333103895187378,
      "learning_rate": 5.079252767934154e-07,
      "loss": 0.4537,
      "step": 389480
    },
    {
      "epoch": 159.43,
      "grad_norm": 2.1624674797058105,
      "learning_rate": 5.078266669552391e-07,
      "loss": 0.4322,
      "step": 389490
    },
    {
      "epoch": 159.44,
      "grad_norm": 2.848846197128296,
      "learning_rate": 5.077280655271438e-07,
      "loss": 0.4379,
      "step": 389500
    },
    {
      "epoch": 159.44,
      "grad_norm": 2.2498838901519775,
      "learning_rate": 5.076294725095807e-07,
      "loss": 0.4285,
      "step": 389510
    },
    {
      "epoch": 159.44,
      "grad_norm": 2.038465976715088,
      "learning_rate": 5.075308879030012e-07,
      "loss": 0.4248,
      "step": 389520
    },
    {
      "epoch": 159.45,
      "grad_norm": 2.455507516860962,
      "learning_rate": 5.074323117078573e-07,
      "loss": 0.4407,
      "step": 389530
    },
    {
      "epoch": 159.45,
      "grad_norm": 2.1717851161956787,
      "learning_rate": 5.073337439246003e-07,
      "loss": 0.4488,
      "step": 389540
    },
    {
      "epoch": 159.46,
      "grad_norm": 2.199582576751709,
      "learning_rate": 5.072351845536823e-07,
      "loss": 0.4676,
      "step": 389550
    },
    {
      "epoch": 159.46,
      "grad_norm": 2.1577975749969482,
      "learning_rate": 5.071366335955537e-07,
      "loss": 0.4439,
      "step": 389560
    },
    {
      "epoch": 159.46,
      "grad_norm": 1.8807659149169922,
      "learning_rate": 5.070380910506664e-07,
      "loss": 0.4712,
      "step": 389570
    },
    {
      "epoch": 159.47,
      "grad_norm": 2.3860433101654053,
      "learning_rate": 5.069395569194719e-07,
      "loss": 0.4584,
      "step": 389580
    },
    {
      "epoch": 159.47,
      "grad_norm": 1.9663962125778198,
      "learning_rate": 5.06841031202422e-07,
      "loss": 0.4466,
      "step": 389590
    },
    {
      "epoch": 159.48,
      "grad_norm": 2.484084129333496,
      "learning_rate": 5.067425138999669e-07,
      "loss": 0.4398,
      "step": 389600
    },
    {
      "epoch": 159.48,
      "grad_norm": 2.3695003986358643,
      "learning_rate": 5.06644005012559e-07,
      "loss": 0.44,
      "step": 389610
    },
    {
      "epoch": 159.48,
      "grad_norm": 1.769565224647522,
      "learning_rate": 5.065455045406493e-07,
      "loss": 0.4528,
      "step": 389620
    },
    {
      "epoch": 159.49,
      "grad_norm": 1.714418649673462,
      "learning_rate": 5.064470124846886e-07,
      "loss": 0.4402,
      "step": 389630
    },
    {
      "epoch": 159.49,
      "grad_norm": 2.0776894092559814,
      "learning_rate": 5.063485288451285e-07,
      "loss": 0.4434,
      "step": 389640
    },
    {
      "epoch": 159.5,
      "grad_norm": 1.837841272354126,
      "learning_rate": 5.062500536224197e-07,
      "loss": 0.4744,
      "step": 389650
    },
    {
      "epoch": 159.5,
      "grad_norm": 2.2151877880096436,
      "learning_rate": 5.061515868170139e-07,
      "loss": 0.4493,
      "step": 389660
    },
    {
      "epoch": 159.5,
      "grad_norm": 1.9549767971038818,
      "learning_rate": 5.060531284293617e-07,
      "loss": 0.4385,
      "step": 389670
    },
    {
      "epoch": 159.51,
      "grad_norm": 2.377835750579834,
      "learning_rate": 5.059546784599144e-07,
      "loss": 0.4557,
      "step": 389680
    },
    {
      "epoch": 159.51,
      "grad_norm": 2.1374316215515137,
      "learning_rate": 5.058562369091226e-07,
      "loss": 0.4401,
      "step": 389690
    },
    {
      "epoch": 159.52,
      "grad_norm": 1.784485936164856,
      "learning_rate": 5.057578037774376e-07,
      "loss": 0.4471,
      "step": 389700
    },
    {
      "epoch": 159.52,
      "grad_norm": 1.7754251956939697,
      "learning_rate": 5.056593790653102e-07,
      "loss": 0.458,
      "step": 389710
    },
    {
      "epoch": 159.53,
      "grad_norm": 2.4030542373657227,
      "learning_rate": 5.055609627731915e-07,
      "loss": 0.4731,
      "step": 389720
    },
    {
      "epoch": 159.53,
      "grad_norm": 2.0938313007354736,
      "learning_rate": 5.054625549015317e-07,
      "loss": 0.4423,
      "step": 389730
    },
    {
      "epoch": 159.53,
      "grad_norm": 1.8130062818527222,
      "learning_rate": 5.053641554507821e-07,
      "loss": 0.4318,
      "step": 389740
    },
    {
      "epoch": 159.54,
      "grad_norm": 2.067873954772949,
      "learning_rate": 5.052657644213937e-07,
      "loss": 0.4477,
      "step": 389750
    },
    {
      "epoch": 159.54,
      "grad_norm": 2.025158643722534,
      "learning_rate": 5.051673818138165e-07,
      "loss": 0.4392,
      "step": 389760
    },
    {
      "epoch": 159.55,
      "grad_norm": 1.768936038017273,
      "learning_rate": 5.050690076285016e-07,
      "loss": 0.4504,
      "step": 389770
    },
    {
      "epoch": 159.55,
      "grad_norm": 2.246969223022461,
      "learning_rate": 5.049706418658997e-07,
      "loss": 0.4434,
      "step": 389780
    },
    {
      "epoch": 159.55,
      "grad_norm": 2.0029661655426025,
      "learning_rate": 5.04872284526461e-07,
      "loss": 0.4309,
      "step": 389790
    },
    {
      "epoch": 159.56,
      "grad_norm": 2.0613131523132324,
      "learning_rate": 5.047739356106364e-07,
      "loss": 0.4283,
      "step": 389800
    },
    {
      "epoch": 159.56,
      "grad_norm": 2.036881923675537,
      "learning_rate": 5.046755951188767e-07,
      "loss": 0.4267,
      "step": 389810
    },
    {
      "epoch": 159.57,
      "grad_norm": 1.897486686706543,
      "learning_rate": 5.045772630516317e-07,
      "loss": 0.4682,
      "step": 389820
    },
    {
      "epoch": 159.57,
      "grad_norm": 2.2065489292144775,
      "learning_rate": 5.044789394093521e-07,
      "loss": 0.435,
      "step": 389830
    },
    {
      "epoch": 159.57,
      "grad_norm": 2.211360454559326,
      "learning_rate": 5.043806241924887e-07,
      "loss": 0.4272,
      "step": 389840
    },
    {
      "epoch": 159.58,
      "grad_norm": 1.8873789310455322,
      "learning_rate": 5.042823174014914e-07,
      "loss": 0.455,
      "step": 389850
    },
    {
      "epoch": 159.58,
      "grad_norm": 2.0597386360168457,
      "learning_rate": 5.041840190368111e-07,
      "loss": 0.4512,
      "step": 389860
    },
    {
      "epoch": 159.59,
      "grad_norm": 1.8323006629943848,
      "learning_rate": 5.040857290988966e-07,
      "loss": 0.454,
      "step": 389870
    },
    {
      "epoch": 159.59,
      "grad_norm": 1.7583484649658203,
      "learning_rate": 5.039874475881999e-07,
      "loss": 0.4523,
      "step": 389880
    },
    {
      "epoch": 159.59,
      "grad_norm": 1.711023211479187,
      "learning_rate": 5.038891745051706e-07,
      "loss": 0.4499,
      "step": 389890
    },
    {
      "epoch": 159.6,
      "grad_norm": 2.3182153701782227,
      "learning_rate": 5.037909098502586e-07,
      "loss": 0.4496,
      "step": 389900
    },
    {
      "epoch": 159.6,
      "grad_norm": 2.4066755771636963,
      "learning_rate": 5.036926536239144e-07,
      "loss": 0.4591,
      "step": 389910
    },
    {
      "epoch": 159.61,
      "grad_norm": 4.285151481628418,
      "learning_rate": 5.035944058265878e-07,
      "loss": 0.4421,
      "step": 389920
    },
    {
      "epoch": 159.61,
      "grad_norm": 2.384202003479004,
      "learning_rate": 5.034961664587289e-07,
      "loss": 0.4729,
      "step": 389930
    },
    {
      "epoch": 159.62,
      "grad_norm": 2.3094682693481445,
      "learning_rate": 5.033979355207881e-07,
      "loss": 0.431,
      "step": 389940
    },
    {
      "epoch": 159.62,
      "grad_norm": 2.015256643295288,
      "learning_rate": 5.032997130132148e-07,
      "loss": 0.4413,
      "step": 389950
    },
    {
      "epoch": 159.62,
      "grad_norm": 2.096540927886963,
      "learning_rate": 5.032014989364597e-07,
      "loss": 0.451,
      "step": 389960
    },
    {
      "epoch": 159.63,
      "grad_norm": 1.75493323802948,
      "learning_rate": 5.031032932909718e-07,
      "loss": 0.4449,
      "step": 389970
    },
    {
      "epoch": 159.63,
      "grad_norm": 1.8869627714157104,
      "learning_rate": 5.030050960772016e-07,
      "loss": 0.4461,
      "step": 389980
    },
    {
      "epoch": 159.64,
      "grad_norm": 1.7147880792617798,
      "learning_rate": 5.02906907295599e-07,
      "loss": 0.436,
      "step": 389990
    },
    {
      "epoch": 159.64,
      "grad_norm": 2.252286911010742,
      "learning_rate": 5.028087269466134e-07,
      "loss": 0.458,
      "step": 390000
    },
    {
      "epoch": 159.64,
      "grad_norm": 1.8882899284362793,
      "learning_rate": 5.027105550306944e-07,
      "loss": 0.4497,
      "step": 390010
    },
    {
      "epoch": 159.65,
      "grad_norm": 1.8674851655960083,
      "learning_rate": 5.026123915482919e-07,
      "loss": 0.4635,
      "step": 390020
    },
    {
      "epoch": 159.65,
      "grad_norm": 2.379998207092285,
      "learning_rate": 5.025142364998559e-07,
      "loss": 0.4474,
      "step": 390030
    },
    {
      "epoch": 159.66,
      "grad_norm": 1.7672125101089478,
      "learning_rate": 5.024160898858355e-07,
      "loss": 0.4442,
      "step": 390040
    },
    {
      "epoch": 159.66,
      "grad_norm": 1.799202799797058,
      "learning_rate": 5.023179517066807e-07,
      "loss": 0.4547,
      "step": 390050
    },
    {
      "epoch": 159.66,
      "grad_norm": 1.8948276042938232,
      "learning_rate": 5.022198219628404e-07,
      "loss": 0.4382,
      "step": 390060
    },
    {
      "epoch": 159.67,
      "grad_norm": 2.044161081314087,
      "learning_rate": 5.021217006547651e-07,
      "loss": 0.4376,
      "step": 390070
    },
    {
      "epoch": 159.67,
      "grad_norm": 2.1103475093841553,
      "learning_rate": 5.020235877829041e-07,
      "loss": 0.4485,
      "step": 390080
    },
    {
      "epoch": 159.68,
      "grad_norm": 4.552765369415283,
      "learning_rate": 5.019254833477064e-07,
      "loss": 0.4464,
      "step": 390090
    },
    {
      "epoch": 159.68,
      "grad_norm": 2.1363415718078613,
      "learning_rate": 5.018273873496216e-07,
      "loss": 0.4268,
      "step": 390100
    },
    {
      "epoch": 159.68,
      "grad_norm": 2.239701271057129,
      "learning_rate": 5.017292997890989e-07,
      "loss": 0.4499,
      "step": 390110
    },
    {
      "epoch": 159.69,
      "grad_norm": 1.922643780708313,
      "learning_rate": 5.016312206665879e-07,
      "loss": 0.4508,
      "step": 390120
    },
    {
      "epoch": 159.69,
      "grad_norm": 2.3244593143463135,
      "learning_rate": 5.015331499825382e-07,
      "loss": 0.4525,
      "step": 390130
    },
    {
      "epoch": 159.7,
      "grad_norm": 2.1031949520111084,
      "learning_rate": 5.014350877373982e-07,
      "loss": 0.4289,
      "step": 390140
    },
    {
      "epoch": 159.7,
      "grad_norm": 2.0426836013793945,
      "learning_rate": 5.013370339316173e-07,
      "loss": 0.4386,
      "step": 390150
    },
    {
      "epoch": 159.71,
      "grad_norm": 2.154392957687378,
      "learning_rate": 5.012389885656449e-07,
      "loss": 0.4435,
      "step": 390160
    },
    {
      "epoch": 159.71,
      "grad_norm": 2.2121715545654297,
      "learning_rate": 5.011409516399302e-07,
      "loss": 0.4315,
      "step": 390170
    },
    {
      "epoch": 159.71,
      "grad_norm": 2.0685226917266846,
      "learning_rate": 5.010429231549221e-07,
      "loss": 0.4397,
      "step": 390180
    },
    {
      "epoch": 159.72,
      "grad_norm": 1.8638306856155396,
      "learning_rate": 5.009449031110697e-07,
      "loss": 0.4292,
      "step": 390190
    },
    {
      "epoch": 159.72,
      "grad_norm": 1.9328786134719849,
      "learning_rate": 5.008468915088222e-07,
      "loss": 0.4424,
      "step": 390200
    },
    {
      "epoch": 159.73,
      "grad_norm": 2.102513074874878,
      "learning_rate": 5.007488883486283e-07,
      "loss": 0.4483,
      "step": 390210
    },
    {
      "epoch": 159.73,
      "grad_norm": 2.407249927520752,
      "learning_rate": 5.00650893630937e-07,
      "loss": 0.4644,
      "step": 390220
    },
    {
      "epoch": 159.73,
      "grad_norm": 1.8782811164855957,
      "learning_rate": 5.005529073561973e-07,
      "loss": 0.4327,
      "step": 390230
    },
    {
      "epoch": 159.74,
      "grad_norm": 1.8273422718048096,
      "learning_rate": 5.004549295248576e-07,
      "loss": 0.438,
      "step": 390240
    },
    {
      "epoch": 159.74,
      "grad_norm": 1.865836262702942,
      "learning_rate": 5.003569601373678e-07,
      "loss": 0.4377,
      "step": 390250
    },
    {
      "epoch": 159.75,
      "grad_norm": 2.3570055961608887,
      "learning_rate": 5.002589991941757e-07,
      "loss": 0.4278,
      "step": 390260
    },
    {
      "epoch": 159.75,
      "grad_norm": 1.8909881114959717,
      "learning_rate": 5.001610466957313e-07,
      "loss": 0.469,
      "step": 390270
    },
    {
      "epoch": 159.75,
      "grad_norm": 2.5642807483673096,
      "learning_rate": 5.000631026424817e-07,
      "loss": 0.4484,
      "step": 390280
    },
    {
      "epoch": 159.76,
      "grad_norm": 3.069027900695801,
      "learning_rate": 4.99965167034876e-07,
      "loss": 0.461,
      "step": 390290
    },
    {
      "epoch": 159.76,
      "grad_norm": 2.0913286209106445,
      "learning_rate": 4.998672398733634e-07,
      "loss": 0.4397,
      "step": 390300
    },
    {
      "epoch": 159.77,
      "grad_norm": 1.861975908279419,
      "learning_rate": 4.997693211583919e-07,
      "loss": 0.4364,
      "step": 390310
    },
    {
      "epoch": 159.77,
      "grad_norm": 1.8201019763946533,
      "learning_rate": 4.996714108904105e-07,
      "loss": 0.4431,
      "step": 390320
    },
    {
      "epoch": 159.77,
      "grad_norm": 1.9331364631652832,
      "learning_rate": 4.995735090698675e-07,
      "loss": 0.4363,
      "step": 390330
    },
    {
      "epoch": 159.78,
      "grad_norm": 1.8691726922988892,
      "learning_rate": 4.994756156972112e-07,
      "loss": 0.4569,
      "step": 390340
    },
    {
      "epoch": 159.78,
      "grad_norm": 1.7706712484359741,
      "learning_rate": 4.993777307728905e-07,
      "loss": 0.439,
      "step": 390350
    },
    {
      "epoch": 159.79,
      "grad_norm": 5.951247692108154,
      "learning_rate": 4.992798542973534e-07,
      "loss": 0.454,
      "step": 390360
    },
    {
      "epoch": 159.79,
      "grad_norm": 2.046628475189209,
      "learning_rate": 4.991819862710484e-07,
      "loss": 0.4458,
      "step": 390370
    },
    {
      "epoch": 159.8,
      "grad_norm": 1.9528664350509644,
      "learning_rate": 4.990841266944237e-07,
      "loss": 0.4519,
      "step": 390380
    },
    {
      "epoch": 159.8,
      "grad_norm": 2.0729713439941406,
      "learning_rate": 4.98986275567928e-07,
      "loss": 0.4471,
      "step": 390390
    },
    {
      "epoch": 159.8,
      "grad_norm": 1.9699070453643799,
      "learning_rate": 4.988884328920089e-07,
      "loss": 0.4514,
      "step": 390400
    },
    {
      "epoch": 159.81,
      "grad_norm": 1.9334343671798706,
      "learning_rate": 4.987905986671151e-07,
      "loss": 0.4369,
      "step": 390410
    },
    {
      "epoch": 159.81,
      "grad_norm": 2.068617343902588,
      "learning_rate": 4.986927728936943e-07,
      "loss": 0.436,
      "step": 390420
    },
    {
      "epoch": 159.82,
      "grad_norm": 1.918609857559204,
      "learning_rate": 4.985949555721951e-07,
      "loss": 0.4603,
      "step": 390430
    },
    {
      "epoch": 159.82,
      "grad_norm": 1.8247801065444946,
      "learning_rate": 4.984971467030657e-07,
      "loss": 0.428,
      "step": 390440
    },
    {
      "epoch": 159.82,
      "grad_norm": 2.1386609077453613,
      "learning_rate": 4.983993462867534e-07,
      "loss": 0.4553,
      "step": 390450
    },
    {
      "epoch": 159.83,
      "grad_norm": 1.7154401540756226,
      "learning_rate": 4.98301554323707e-07,
      "loss": 0.4359,
      "step": 390460
    },
    {
      "epoch": 159.83,
      "grad_norm": 2.269420623779297,
      "learning_rate": 4.982037708143739e-07,
      "loss": 0.4567,
      "step": 390470
    },
    {
      "epoch": 159.84,
      "grad_norm": 2.1690165996551514,
      "learning_rate": 4.981059957592023e-07,
      "loss": 0.4457,
      "step": 390480
    },
    {
      "epoch": 159.84,
      "grad_norm": 1.7004858255386353,
      "learning_rate": 4.980082291586402e-07,
      "loss": 0.4688,
      "step": 390490
    },
    {
      "epoch": 159.84,
      "grad_norm": 3.1323506832122803,
      "learning_rate": 4.979104710131349e-07,
      "loss": 0.4437,
      "step": 390500
    },
    {
      "epoch": 159.85,
      "grad_norm": 2.2074880599975586,
      "learning_rate": 4.978127213231351e-07,
      "loss": 0.4188,
      "step": 390510
    },
    {
      "epoch": 159.85,
      "grad_norm": 1.9759544134140015,
      "learning_rate": 4.977149800890879e-07,
      "loss": 0.4511,
      "step": 390520
    },
    {
      "epoch": 159.86,
      "grad_norm": 1.6480774879455566,
      "learning_rate": 4.976172473114413e-07,
      "loss": 0.4304,
      "step": 390530
    },
    {
      "epoch": 159.86,
      "grad_norm": 1.9173325300216675,
      "learning_rate": 4.975195229906427e-07,
      "loss": 0.4364,
      "step": 390540
    },
    {
      "epoch": 159.86,
      "grad_norm": 1.9639068841934204,
      "learning_rate": 4.974218071271402e-07,
      "loss": 0.4324,
      "step": 390550
    },
    {
      "epoch": 159.87,
      "grad_norm": 1.9440613985061646,
      "learning_rate": 4.973240997213813e-07,
      "loss": 0.4524,
      "step": 390560
    },
    {
      "epoch": 159.87,
      "grad_norm": 2.190605640411377,
      "learning_rate": 4.972264007738132e-07,
      "loss": 0.4521,
      "step": 390570
    },
    {
      "epoch": 159.88,
      "grad_norm": 2.1947364807128906,
      "learning_rate": 4.971287102848845e-07,
      "loss": 0.4449,
      "step": 390580
    },
    {
      "epoch": 159.88,
      "grad_norm": 1.9971810579299927,
      "learning_rate": 4.970310282550413e-07,
      "loss": 0.4556,
      "step": 390590
    },
    {
      "epoch": 159.89,
      "grad_norm": 2.1917412281036377,
      "learning_rate": 4.969333546847319e-07,
      "loss": 0.473,
      "step": 390600
    },
    {
      "epoch": 159.89,
      "grad_norm": 1.7543573379516602,
      "learning_rate": 4.968356895744031e-07,
      "loss": 0.434,
      "step": 390610
    },
    {
      "epoch": 159.89,
      "grad_norm": 2.2590506076812744,
      "learning_rate": 4.967380329245026e-07,
      "loss": 0.4471,
      "step": 390620
    },
    {
      "epoch": 159.9,
      "grad_norm": 1.8796814680099487,
      "learning_rate": 4.966403847354784e-07,
      "loss": 0.435,
      "step": 390630
    },
    {
      "epoch": 159.9,
      "grad_norm": 2.109131097793579,
      "learning_rate": 4.965427450077769e-07,
      "loss": 0.4349,
      "step": 390640
    },
    {
      "epoch": 159.91,
      "grad_norm": 1.8389551639556885,
      "learning_rate": 4.964451137418463e-07,
      "loss": 0.4547,
      "step": 390650
    },
    {
      "epoch": 159.91,
      "grad_norm": 2.504389524459839,
      "learning_rate": 4.963474909381327e-07,
      "loss": 0.4422,
      "step": 390660
    },
    {
      "epoch": 159.91,
      "grad_norm": 1.8330883979797363,
      "learning_rate": 4.962498765970844e-07,
      "loss": 0.441,
      "step": 390670
    },
    {
      "epoch": 159.92,
      "grad_norm": 1.7828489542007446,
      "learning_rate": 4.961522707191479e-07,
      "loss": 0.4576,
      "step": 390680
    },
    {
      "epoch": 159.92,
      "grad_norm": 2.1142892837524414,
      "learning_rate": 4.960546733047703e-07,
      "loss": 0.4658,
      "step": 390690
    },
    {
      "epoch": 159.93,
      "grad_norm": 2.208324909210205,
      "learning_rate": 4.959570843543989e-07,
      "loss": 0.4313,
      "step": 390700
    },
    {
      "epoch": 159.93,
      "grad_norm": 1.888183832168579,
      "learning_rate": 4.958595038684815e-07,
      "loss": 0.4391,
      "step": 390710
    },
    {
      "epoch": 159.93,
      "grad_norm": 2.1124320030212402,
      "learning_rate": 4.957619318474635e-07,
      "loss": 0.4453,
      "step": 390720
    },
    {
      "epoch": 159.94,
      "grad_norm": 2.562553882598877,
      "learning_rate": 4.956643682917927e-07,
      "loss": 0.4363,
      "step": 390730
    },
    {
      "epoch": 159.94,
      "grad_norm": 1.9636293649673462,
      "learning_rate": 4.955668132019158e-07,
      "loss": 0.4539,
      "step": 390740
    },
    {
      "epoch": 159.95,
      "grad_norm": 1.9073659181594849,
      "learning_rate": 4.954692665782799e-07,
      "loss": 0.4473,
      "step": 390750
    },
    {
      "epoch": 159.95,
      "grad_norm": 2.0900566577911377,
      "learning_rate": 4.953717284213319e-07,
      "loss": 0.4379,
      "step": 390760
    },
    {
      "epoch": 159.95,
      "grad_norm": 2.149733781814575,
      "learning_rate": 4.952741987315185e-07,
      "loss": 0.4359,
      "step": 390770
    },
    {
      "epoch": 159.96,
      "grad_norm": 1.612414002418518,
      "learning_rate": 4.951766775092865e-07,
      "loss": 0.4304,
      "step": 390780
    },
    {
      "epoch": 159.96,
      "grad_norm": 1.8558474779129028,
      "learning_rate": 4.950791647550824e-07,
      "loss": 0.4533,
      "step": 390790
    },
    {
      "epoch": 159.97,
      "grad_norm": 2.27402400970459,
      "learning_rate": 4.949816604693533e-07,
      "loss": 0.4556,
      "step": 390800
    },
    {
      "epoch": 159.97,
      "grad_norm": 2.544611692428589,
      "learning_rate": 4.948841646525449e-07,
      "loss": 0.4598,
      "step": 390810
    },
    {
      "epoch": 159.98,
      "grad_norm": 1.9020471572875977,
      "learning_rate": 4.947866773051053e-07,
      "loss": 0.4436,
      "step": 390820
    },
    {
      "epoch": 159.98,
      "grad_norm": 2.266383171081543,
      "learning_rate": 4.946891984274803e-07,
      "loss": 0.4263,
      "step": 390830
    },
    {
      "epoch": 159.98,
      "grad_norm": 2.06337833404541,
      "learning_rate": 4.945917280201165e-07,
      "loss": 0.4564,
      "step": 390840
    },
    {
      "epoch": 159.99,
      "grad_norm": 2.1635448932647705,
      "learning_rate": 4.94494266083461e-07,
      "loss": 0.4377,
      "step": 390850
    },
    {
      "epoch": 159.99,
      "grad_norm": 2.016756296157837,
      "learning_rate": 4.943968126179589e-07,
      "loss": 0.4414,
      "step": 390860
    },
    {
      "epoch": 160.0,
      "grad_norm": 2.3481905460357666,
      "learning_rate": 4.942993676240574e-07,
      "loss": 0.4531,
      "step": 390870
    },
    {
      "epoch": 160.0,
      "grad_norm": 2.321061849594116,
      "learning_rate": 4.942019311022028e-07,
      "loss": 0.4154,
      "step": 390880
    },
    {
      "epoch": 160.0,
      "eval_loss": 0.4464147388935089,
      "eval_runtime": 52.3848,
      "eval_samples_per_second": 65.84,
      "eval_steps_per_second": 8.247,
      "step": 390880
    },
    {
      "epoch": 160.0,
      "grad_norm": 1.9057315587997437,
      "learning_rate": 4.941045030528415e-07,
      "loss": 0.4276,
      "step": 390890
    },
    {
      "epoch": 160.01,
      "grad_norm": 2.1548335552215576,
      "learning_rate": 4.940070834764195e-07,
      "loss": 0.4399,
      "step": 390900
    },
    {
      "epoch": 160.01,
      "grad_norm": 2.2036192417144775,
      "learning_rate": 4.939096723733835e-07,
      "loss": 0.4322,
      "step": 390910
    },
    {
      "epoch": 160.02,
      "grad_norm": 2.0334320068359375,
      "learning_rate": 4.938122697441797e-07,
      "loss": 0.4556,
      "step": 390920
    },
    {
      "epoch": 160.02,
      "grad_norm": 2.8303146362304688,
      "learning_rate": 4.937148755892541e-07,
      "loss": 0.4437,
      "step": 390930
    },
    {
      "epoch": 160.02,
      "grad_norm": 2.204033136367798,
      "learning_rate": 4.936174899090529e-07,
      "loss": 0.4569,
      "step": 390940
    },
    {
      "epoch": 160.03,
      "grad_norm": 2.2314882278442383,
      "learning_rate": 4.935201127040222e-07,
      "loss": 0.4432,
      "step": 390950
    },
    {
      "epoch": 160.03,
      "grad_norm": 2.071871519088745,
      "learning_rate": 4.93422743974608e-07,
      "loss": 0.4419,
      "step": 390960
    },
    {
      "epoch": 160.04,
      "grad_norm": 2.076113700866699,
      "learning_rate": 4.933253837212565e-07,
      "loss": 0.4578,
      "step": 390970
    },
    {
      "epoch": 160.04,
      "grad_norm": 2.211841583251953,
      "learning_rate": 4.932280319444135e-07,
      "loss": 0.4364,
      "step": 390980
    },
    {
      "epoch": 160.05,
      "grad_norm": 2.522467851638794,
      "learning_rate": 4.93130688644525e-07,
      "loss": 0.4579,
      "step": 390990
    },
    {
      "epoch": 160.05,
      "grad_norm": 2.230264186859131,
      "learning_rate": 4.930333538220371e-07,
      "loss": 0.4306,
      "step": 391000
    },
    {
      "epoch": 160.05,
      "grad_norm": 2.498154401779175,
      "learning_rate": 4.929360274773955e-07,
      "loss": 0.4407,
      "step": 391010
    },
    {
      "epoch": 160.06,
      "grad_norm": 2.013859510421753,
      "learning_rate": 4.928387096110462e-07,
      "loss": 0.4286,
      "step": 391020
    },
    {
      "epoch": 160.06,
      "grad_norm": 1.784111738204956,
      "learning_rate": 4.927414002234346e-07,
      "loss": 0.468,
      "step": 391030
    },
    {
      "epoch": 160.07,
      "grad_norm": 2.153777837753296,
      "learning_rate": 4.926440993150067e-07,
      "loss": 0.4403,
      "step": 391040
    },
    {
      "epoch": 160.07,
      "grad_norm": 2.1832592487335205,
      "learning_rate": 4.925468068862084e-07,
      "loss": 0.4375,
      "step": 391050
    },
    {
      "epoch": 160.07,
      "grad_norm": 1.89958655834198,
      "learning_rate": 4.924495229374854e-07,
      "loss": 0.4395,
      "step": 391060
    },
    {
      "epoch": 160.08,
      "grad_norm": 1.8607800006866455,
      "learning_rate": 4.923522474692831e-07,
      "loss": 0.4496,
      "step": 391070
    },
    {
      "epoch": 160.08,
      "grad_norm": 1.9904675483703613,
      "learning_rate": 4.922549804820469e-07,
      "loss": 0.4377,
      "step": 391080
    },
    {
      "epoch": 160.09,
      "grad_norm": 2.192802667617798,
      "learning_rate": 4.921577219762229e-07,
      "loss": 0.4455,
      "step": 391090
    },
    {
      "epoch": 160.09,
      "grad_norm": 1.7828888893127441,
      "learning_rate": 4.920604719522563e-07,
      "loss": 0.4529,
      "step": 391100
    },
    {
      "epoch": 160.09,
      "grad_norm": 1.6807873249053955,
      "learning_rate": 4.919632304105926e-07,
      "loss": 0.4488,
      "step": 391110
    },
    {
      "epoch": 160.1,
      "grad_norm": 1.8869903087615967,
      "learning_rate": 4.918659973516773e-07,
      "loss": 0.4406,
      "step": 391120
    },
    {
      "epoch": 160.1,
      "grad_norm": 2.0244998931884766,
      "learning_rate": 4.91768772775956e-07,
      "loss": 0.4539,
      "step": 391130
    },
    {
      "epoch": 160.11,
      "grad_norm": 2.073402166366577,
      "learning_rate": 4.916715566838736e-07,
      "loss": 0.4489,
      "step": 391140
    },
    {
      "epoch": 160.11,
      "grad_norm": 2.5236048698425293,
      "learning_rate": 4.915743490758759e-07,
      "loss": 0.4398,
      "step": 391150
    },
    {
      "epoch": 160.11,
      "grad_norm": 1.9524701833724976,
      "learning_rate": 4.914771499524087e-07,
      "loss": 0.4376,
      "step": 391160
    },
    {
      "epoch": 160.12,
      "grad_norm": 1.9377522468566895,
      "learning_rate": 4.913799593139158e-07,
      "loss": 0.4527,
      "step": 391170
    },
    {
      "epoch": 160.12,
      "grad_norm": 2.191915512084961,
      "learning_rate": 4.912827771608432e-07,
      "loss": 0.4335,
      "step": 391180
    },
    {
      "epoch": 160.13,
      "grad_norm": 1.8179280757904053,
      "learning_rate": 4.911856034936356e-07,
      "loss": 0.4572,
      "step": 391190
    },
    {
      "epoch": 160.13,
      "grad_norm": 2.094316244125366,
      "learning_rate": 4.910884383127391e-07,
      "loss": 0.4291,
      "step": 391200
    },
    {
      "epoch": 160.14,
      "grad_norm": 1.96163809299469,
      "learning_rate": 4.909912816185982e-07,
      "loss": 0.4275,
      "step": 391210
    },
    {
      "epoch": 160.14,
      "grad_norm": 2.3131940364837646,
      "learning_rate": 4.90894133411658e-07,
      "loss": 0.4705,
      "step": 391220
    },
    {
      "epoch": 160.14,
      "grad_norm": 1.920784831047058,
      "learning_rate": 4.907969936923639e-07,
      "loss": 0.4689,
      "step": 391230
    },
    {
      "epoch": 160.15,
      "grad_norm": 1.7591636180877686,
      "learning_rate": 4.906998624611605e-07,
      "loss": 0.4413,
      "step": 391240
    },
    {
      "epoch": 160.15,
      "grad_norm": 1.9606200456619263,
      "learning_rate": 4.906027397184924e-07,
      "loss": 0.4428,
      "step": 391250
    },
    {
      "epoch": 160.16,
      "grad_norm": 2.0911478996276855,
      "learning_rate": 4.905056254648052e-07,
      "loss": 0.4602,
      "step": 391260
    },
    {
      "epoch": 160.16,
      "grad_norm": 1.8455493450164795,
      "learning_rate": 4.904085197005436e-07,
      "loss": 0.4492,
      "step": 391270
    },
    {
      "epoch": 160.16,
      "grad_norm": 2.1927905082702637,
      "learning_rate": 4.903114224261519e-07,
      "loss": 0.4213,
      "step": 391280
    },
    {
      "epoch": 160.17,
      "grad_norm": 2.283987283706665,
      "learning_rate": 4.902143336420755e-07,
      "loss": 0.4332,
      "step": 391290
    },
    {
      "epoch": 160.17,
      "grad_norm": 2.177520990371704,
      "learning_rate": 4.901172533487597e-07,
      "loss": 0.4314,
      "step": 391300
    },
    {
      "epoch": 160.18,
      "grad_norm": 1.736402988433838,
      "learning_rate": 4.900201815466478e-07,
      "loss": 0.439,
      "step": 391310
    },
    {
      "epoch": 160.18,
      "grad_norm": 1.993673324584961,
      "learning_rate": 4.899231182361849e-07,
      "loss": 0.4272,
      "step": 391320
    },
    {
      "epoch": 160.18,
      "grad_norm": 1.8312132358551025,
      "learning_rate": 4.898260634178159e-07,
      "loss": 0.4478,
      "step": 391330
    },
    {
      "epoch": 160.19,
      "grad_norm": 2.026303768157959,
      "learning_rate": 4.897290170919855e-07,
      "loss": 0.4531,
      "step": 391340
    },
    {
      "epoch": 160.19,
      "grad_norm": 2.0353024005889893,
      "learning_rate": 4.896319792591378e-07,
      "loss": 0.4207,
      "step": 391350
    },
    {
      "epoch": 160.2,
      "grad_norm": 1.8267581462860107,
      "learning_rate": 4.895349499197179e-07,
      "loss": 0.4329,
      "step": 391360
    },
    {
      "epoch": 160.2,
      "grad_norm": 1.9642088413238525,
      "learning_rate": 4.894379290741698e-07,
      "loss": 0.4434,
      "step": 391370
    },
    {
      "epoch": 160.2,
      "grad_norm": 4.179364204406738,
      "learning_rate": 4.893409167229375e-07,
      "loss": 0.4286,
      "step": 391380
    },
    {
      "epoch": 160.21,
      "grad_norm": 2.1673223972320557,
      "learning_rate": 4.892439128664667e-07,
      "loss": 0.4655,
      "step": 391390
    },
    {
      "epoch": 160.21,
      "grad_norm": 2.0113418102264404,
      "learning_rate": 4.891469175052012e-07,
      "loss": 0.4437,
      "step": 391400
    },
    {
      "epoch": 160.22,
      "grad_norm": 1.9066877365112305,
      "learning_rate": 4.89049930639585e-07,
      "loss": 0.4405,
      "step": 391410
    },
    {
      "epoch": 160.22,
      "grad_norm": 1.9766429662704468,
      "learning_rate": 4.889529522700626e-07,
      "loss": 0.4472,
      "step": 391420
    },
    {
      "epoch": 160.23,
      "grad_norm": 2.1467487812042236,
      "learning_rate": 4.88855982397079e-07,
      "loss": 0.449,
      "step": 391430
    },
    {
      "epoch": 160.23,
      "grad_norm": 1.8806954622268677,
      "learning_rate": 4.88759021021077e-07,
      "loss": 0.4493,
      "step": 391440
    },
    {
      "epoch": 160.23,
      "grad_norm": 2.5990242958068848,
      "learning_rate": 4.886620681425012e-07,
      "loss": 0.4694,
      "step": 391450
    },
    {
      "epoch": 160.24,
      "grad_norm": 2.1019070148468018,
      "learning_rate": 4.885651237617962e-07,
      "loss": 0.4655,
      "step": 391460
    },
    {
      "epoch": 160.24,
      "grad_norm": 2.100022554397583,
      "learning_rate": 4.884681878794057e-07,
      "loss": 0.4703,
      "step": 391470
    },
    {
      "epoch": 160.25,
      "grad_norm": 1.9698837995529175,
      "learning_rate": 4.883712604957736e-07,
      "loss": 0.4392,
      "step": 391480
    },
    {
      "epoch": 160.25,
      "grad_norm": 2.044691801071167,
      "learning_rate": 4.882743416113444e-07,
      "loss": 0.4503,
      "step": 391490
    },
    {
      "epoch": 160.25,
      "grad_norm": 2.0672385692596436,
      "learning_rate": 4.881774312265617e-07,
      "loss": 0.4328,
      "step": 391500
    },
    {
      "epoch": 160.26,
      "grad_norm": 2.0305707454681396,
      "learning_rate": 4.880805293418696e-07,
      "loss": 0.4337,
      "step": 391510
    },
    {
      "epoch": 160.26,
      "grad_norm": 1.9291319847106934,
      "learning_rate": 4.879836359577122e-07,
      "loss": 0.4343,
      "step": 391520
    },
    {
      "epoch": 160.27,
      "grad_norm": 1.9389901161193848,
      "learning_rate": 4.87886751074533e-07,
      "loss": 0.446,
      "step": 391530
    },
    {
      "epoch": 160.27,
      "grad_norm": 2.046701192855835,
      "learning_rate": 4.877898746927758e-07,
      "loss": 0.446,
      "step": 391540
    },
    {
      "epoch": 160.27,
      "grad_norm": 1.737349271774292,
      "learning_rate": 4.876930068128844e-07,
      "loss": 0.4425,
      "step": 391550
    },
    {
      "epoch": 160.28,
      "grad_norm": 1.9563604593276978,
      "learning_rate": 4.875961474353023e-07,
      "loss": 0.4604,
      "step": 391560
    },
    {
      "epoch": 160.28,
      "grad_norm": 1.941021203994751,
      "learning_rate": 4.874992965604745e-07,
      "loss": 0.4517,
      "step": 391570
    },
    {
      "epoch": 160.29,
      "grad_norm": 1.9057127237319946,
      "learning_rate": 4.874024541888432e-07,
      "loss": 0.4526,
      "step": 391580
    },
    {
      "epoch": 160.29,
      "grad_norm": 2.1612637042999268,
      "learning_rate": 4.873056203208527e-07,
      "loss": 0.429,
      "step": 391590
    },
    {
      "epoch": 160.29,
      "grad_norm": 1.980133295059204,
      "learning_rate": 4.872087949569462e-07,
      "loss": 0.4489,
      "step": 391600
    },
    {
      "epoch": 160.3,
      "grad_norm": 1.9427623748779297,
      "learning_rate": 4.871119780975674e-07,
      "loss": 0.4597,
      "step": 391610
    },
    {
      "epoch": 160.3,
      "grad_norm": 2.147475481033325,
      "learning_rate": 4.870151697431601e-07,
      "loss": 0.4346,
      "step": 391620
    },
    {
      "epoch": 160.31,
      "grad_norm": 2.138596296310425,
      "learning_rate": 4.869183698941673e-07,
      "loss": 0.4351,
      "step": 391630
    },
    {
      "epoch": 160.31,
      "grad_norm": 2.1594457626342773,
      "learning_rate": 4.868215785510326e-07,
      "loss": 0.4364,
      "step": 391640
    },
    {
      "epoch": 160.32,
      "grad_norm": 2.336188316345215,
      "learning_rate": 4.867247957141995e-07,
      "loss": 0.4451,
      "step": 391650
    },
    {
      "epoch": 160.32,
      "grad_norm": 1.8513050079345703,
      "learning_rate": 4.866280213841113e-07,
      "loss": 0.4441,
      "step": 391660
    },
    {
      "epoch": 160.32,
      "grad_norm": 1.9384503364562988,
      "learning_rate": 4.865312555612111e-07,
      "loss": 0.4408,
      "step": 391670
    },
    {
      "epoch": 160.33,
      "grad_norm": 1.928073525428772,
      "learning_rate": 4.864344982459424e-07,
      "loss": 0.4436,
      "step": 391680
    },
    {
      "epoch": 160.33,
      "grad_norm": 1.9720202684402466,
      "learning_rate": 4.863377494387486e-07,
      "loss": 0.4259,
      "step": 391690
    },
    {
      "epoch": 160.34,
      "grad_norm": 2.513915777206421,
      "learning_rate": 4.862410091400724e-07,
      "loss": 0.4513,
      "step": 391700
    },
    {
      "epoch": 160.34,
      "grad_norm": 2.077655553817749,
      "learning_rate": 4.861442773503574e-07,
      "loss": 0.4416,
      "step": 391710
    },
    {
      "epoch": 160.34,
      "grad_norm": 2.0550756454467773,
      "learning_rate": 4.860475540700462e-07,
      "loss": 0.4597,
      "step": 391720
    },
    {
      "epoch": 160.35,
      "grad_norm": 2.344215154647827,
      "learning_rate": 4.859508392995824e-07,
      "loss": 0.4396,
      "step": 391730
    },
    {
      "epoch": 160.35,
      "grad_norm": 1.80826735496521,
      "learning_rate": 4.858541330394097e-07,
      "loss": 0.4459,
      "step": 391740
    },
    {
      "epoch": 160.36,
      "grad_norm": 1.572346806526184,
      "learning_rate": 4.857574352899687e-07,
      "loss": 0.4404,
      "step": 391750
    },
    {
      "epoch": 160.36,
      "grad_norm": 2.196694850921631,
      "learning_rate": 4.856607460517047e-07,
      "loss": 0.4327,
      "step": 391760
    },
    {
      "epoch": 160.36,
      "grad_norm": 2.0528812408447266,
      "learning_rate": 4.855640653250596e-07,
      "loss": 0.4534,
      "step": 391770
    },
    {
      "epoch": 160.37,
      "grad_norm": 2.052170515060425,
      "learning_rate": 4.854673931104767e-07,
      "loss": 0.4558,
      "step": 391780
    },
    {
      "epoch": 160.37,
      "grad_norm": 2.2609591484069824,
      "learning_rate": 4.853707294083986e-07,
      "loss": 0.4469,
      "step": 391790
    },
    {
      "epoch": 160.38,
      "grad_norm": 2.0374529361724854,
      "learning_rate": 4.85274074219268e-07,
      "loss": 0.4495,
      "step": 391800
    },
    {
      "epoch": 160.38,
      "grad_norm": 2.4630579948425293,
      "learning_rate": 4.851774275435278e-07,
      "loss": 0.4749,
      "step": 391810
    },
    {
      "epoch": 160.38,
      "grad_norm": 1.9404194355010986,
      "learning_rate": 4.850807893816206e-07,
      "loss": 0.4289,
      "step": 391820
    },
    {
      "epoch": 160.39,
      "grad_norm": 1.6501574516296387,
      "learning_rate": 4.849841597339896e-07,
      "loss": 0.4431,
      "step": 391830
    },
    {
      "epoch": 160.39,
      "grad_norm": 1.878912329673767,
      "learning_rate": 4.848875386010768e-07,
      "loss": 0.4462,
      "step": 391840
    },
    {
      "epoch": 160.4,
      "grad_norm": 2.174912214279175,
      "learning_rate": 4.84790925983325e-07,
      "loss": 0.4493,
      "step": 391850
    },
    {
      "epoch": 160.4,
      "grad_norm": 2.2649927139282227,
      "learning_rate": 4.846943218811771e-07,
      "loss": 0.4219,
      "step": 391860
    },
    {
      "epoch": 160.41,
      "grad_norm": 2.0347795486450195,
      "learning_rate": 4.84597726295075e-07,
      "loss": 0.4794,
      "step": 391870
    },
    {
      "epoch": 160.41,
      "grad_norm": 1.8791900873184204,
      "learning_rate": 4.845011392254624e-07,
      "loss": 0.453,
      "step": 391880
    },
    {
      "epoch": 160.41,
      "grad_norm": 1.7876955270767212,
      "learning_rate": 4.844045606727802e-07,
      "loss": 0.452,
      "step": 391890
    },
    {
      "epoch": 160.42,
      "grad_norm": 2.4127542972564697,
      "learning_rate": 4.843079906374713e-07,
      "loss": 0.4547,
      "step": 391900
    },
    {
      "epoch": 160.42,
      "grad_norm": 2.04992413520813,
      "learning_rate": 4.842114291199785e-07,
      "loss": 0.4268,
      "step": 391910
    },
    {
      "epoch": 160.43,
      "grad_norm": 1.7193928956985474,
      "learning_rate": 4.841148761207438e-07,
      "loss": 0.4327,
      "step": 391920
    },
    {
      "epoch": 160.43,
      "grad_norm": 2.175868034362793,
      "learning_rate": 4.840183316402098e-07,
      "loss": 0.4329,
      "step": 391930
    },
    {
      "epoch": 160.43,
      "grad_norm": 2.03116512298584,
      "learning_rate": 4.839217956788179e-07,
      "loss": 0.4445,
      "step": 391940
    },
    {
      "epoch": 160.44,
      "grad_norm": 1.9237349033355713,
      "learning_rate": 4.838252682370114e-07,
      "loss": 0.4466,
      "step": 391950
    },
    {
      "epoch": 160.44,
      "grad_norm": 1.7503139972686768,
      "learning_rate": 4.837287493152321e-07,
      "loss": 0.4409,
      "step": 391960
    },
    {
      "epoch": 160.45,
      "grad_norm": 1.8771462440490723,
      "learning_rate": 4.836322389139221e-07,
      "loss": 0.4341,
      "step": 391970
    },
    {
      "epoch": 160.45,
      "grad_norm": 2.0095901489257812,
      "learning_rate": 4.835357370335235e-07,
      "loss": 0.4654,
      "step": 391980
    },
    {
      "epoch": 160.45,
      "grad_norm": 2.1670408248901367,
      "learning_rate": 4.834392436744785e-07,
      "loss": 0.4555,
      "step": 391990
    },
    {
      "epoch": 160.46,
      "grad_norm": 2.122959613800049,
      "learning_rate": 4.833427588372288e-07,
      "loss": 0.4492,
      "step": 392000
    },
    {
      "epoch": 160.46,
      "grad_norm": 2.4739010334014893,
      "learning_rate": 4.832462825222173e-07,
      "loss": 0.4579,
      "step": 392010
    },
    {
      "epoch": 160.47,
      "grad_norm": 1.8493324518203735,
      "learning_rate": 4.831498147298845e-07,
      "loss": 0.4356,
      "step": 392020
    },
    {
      "epoch": 160.47,
      "grad_norm": 2.045107364654541,
      "learning_rate": 4.830533554606729e-07,
      "loss": 0.4436,
      "step": 392030
    },
    {
      "epoch": 160.47,
      "grad_norm": 2.2634925842285156,
      "learning_rate": 4.829569047150246e-07,
      "loss": 0.4551,
      "step": 392040
    },
    {
      "epoch": 160.48,
      "grad_norm": 1.9701061248779297,
      "learning_rate": 4.828604624933812e-07,
      "loss": 0.4669,
      "step": 392050
    },
    {
      "epoch": 160.48,
      "grad_norm": 1.9871307611465454,
      "learning_rate": 4.827640287961846e-07,
      "loss": 0.452,
      "step": 392060
    },
    {
      "epoch": 160.49,
      "grad_norm": 5.16476583480835,
      "learning_rate": 4.826676036238766e-07,
      "loss": 0.4749,
      "step": 392070
    },
    {
      "epoch": 160.49,
      "grad_norm": 2.2198684215545654,
      "learning_rate": 4.825711869768988e-07,
      "loss": 0.4312,
      "step": 392080
    },
    {
      "epoch": 160.5,
      "grad_norm": 2.0024328231811523,
      "learning_rate": 4.82474778855693e-07,
      "loss": 0.4425,
      "step": 392090
    },
    {
      "epoch": 160.5,
      "grad_norm": 2.3386759757995605,
      "learning_rate": 4.823783792607006e-07,
      "loss": 0.4633,
      "step": 392100
    },
    {
      "epoch": 160.5,
      "grad_norm": 1.8115800619125366,
      "learning_rate": 4.822819881923632e-07,
      "loss": 0.4481,
      "step": 392110
    },
    {
      "epoch": 160.51,
      "grad_norm": 1.9273037910461426,
      "learning_rate": 4.821856056511227e-07,
      "loss": 0.4293,
      "step": 392120
    },
    {
      "epoch": 160.51,
      "grad_norm": 1.9946622848510742,
      "learning_rate": 4.820892316374196e-07,
      "loss": 0.4488,
      "step": 392130
    },
    {
      "epoch": 160.52,
      "grad_norm": 2.178361654281616,
      "learning_rate": 4.81992866151697e-07,
      "loss": 0.4446,
      "step": 392140
    },
    {
      "epoch": 160.52,
      "grad_norm": 2.4157004356384277,
      "learning_rate": 4.818965091943959e-07,
      "loss": 0.4597,
      "step": 392150
    },
    {
      "epoch": 160.52,
      "grad_norm": 2.1063308715820312,
      "learning_rate": 4.818001607659567e-07,
      "loss": 0.4316,
      "step": 392160
    },
    {
      "epoch": 160.53,
      "grad_norm": 1.8644179105758667,
      "learning_rate": 4.817038208668216e-07,
      "loss": 0.4353,
      "step": 392170
    },
    {
      "epoch": 160.53,
      "grad_norm": 2.4077634811401367,
      "learning_rate": 4.816074894974314e-07,
      "loss": 0.4654,
      "step": 392180
    },
    {
      "epoch": 160.54,
      "grad_norm": 2.075782299041748,
      "learning_rate": 4.815111666582278e-07,
      "loss": 0.4474,
      "step": 392190
    },
    {
      "epoch": 160.54,
      "grad_norm": 2.8256747722625732,
      "learning_rate": 4.814148523496519e-07,
      "loss": 0.4602,
      "step": 392200
    },
    {
      "epoch": 160.54,
      "grad_norm": 1.9711947441101074,
      "learning_rate": 4.813185465721447e-07,
      "loss": 0.4563,
      "step": 392210
    },
    {
      "epoch": 160.55,
      "grad_norm": 1.8391032218933105,
      "learning_rate": 4.812222493261477e-07,
      "loss": 0.4511,
      "step": 392220
    },
    {
      "epoch": 160.55,
      "grad_norm": 2.0623128414154053,
      "learning_rate": 4.811259606121019e-07,
      "loss": 0.4378,
      "step": 392230
    },
    {
      "epoch": 160.56,
      "grad_norm": 2.4371747970581055,
      "learning_rate": 4.810296804304482e-07,
      "loss": 0.4336,
      "step": 392240
    },
    {
      "epoch": 160.56,
      "grad_norm": 2.1576831340789795,
      "learning_rate": 4.809334087816282e-07,
      "loss": 0.4347,
      "step": 392250
    },
    {
      "epoch": 160.56,
      "grad_norm": 2.0389974117279053,
      "learning_rate": 4.808371456660822e-07,
      "loss": 0.4439,
      "step": 392260
    },
    {
      "epoch": 160.57,
      "grad_norm": 1.8488706350326538,
      "learning_rate": 4.807408910842514e-07,
      "loss": 0.4359,
      "step": 392270
    },
    {
      "epoch": 160.57,
      "grad_norm": 1.9229017496109009,
      "learning_rate": 4.806446450365771e-07,
      "loss": 0.449,
      "step": 392280
    },
    {
      "epoch": 160.58,
      "grad_norm": 2.889683485031128,
      "learning_rate": 4.805484075234997e-07,
      "loss": 0.4487,
      "step": 392290
    },
    {
      "epoch": 160.58,
      "grad_norm": 1.8124549388885498,
      "learning_rate": 4.8045217854546e-07,
      "loss": 0.4421,
      "step": 392300
    },
    {
      "epoch": 160.59,
      "grad_norm": 2.0600316524505615,
      "learning_rate": 4.803559581028995e-07,
      "loss": 0.4418,
      "step": 392310
    },
    {
      "epoch": 160.59,
      "grad_norm": 1.945264220237732,
      "learning_rate": 4.802597461962583e-07,
      "loss": 0.4506,
      "step": 392320
    },
    {
      "epoch": 160.59,
      "grad_norm": 2.2337088584899902,
      "learning_rate": 4.801635428259774e-07,
      "loss": 0.4464,
      "step": 392330
    },
    {
      "epoch": 160.6,
      "grad_norm": 2.2622694969177246,
      "learning_rate": 4.800673479924974e-07,
      "loss": 0.4565,
      "step": 392340
    },
    {
      "epoch": 160.6,
      "grad_norm": 2.1965906620025635,
      "learning_rate": 4.79971161696259e-07,
      "loss": 0.4396,
      "step": 392350
    },
    {
      "epoch": 160.61,
      "grad_norm": 2.026947259902954,
      "learning_rate": 4.798749839377028e-07,
      "loss": 0.4411,
      "step": 392360
    },
    {
      "epoch": 160.61,
      "grad_norm": 2.153287172317505,
      "learning_rate": 4.797788147172696e-07,
      "loss": 0.4354,
      "step": 392370
    },
    {
      "epoch": 160.61,
      "grad_norm": 2.1723074913024902,
      "learning_rate": 4.796826540353995e-07,
      "loss": 0.4511,
      "step": 392380
    },
    {
      "epoch": 160.62,
      "grad_norm": 2.297290563583374,
      "learning_rate": 4.795865018925333e-07,
      "loss": 0.442,
      "step": 392390
    },
    {
      "epoch": 160.62,
      "grad_norm": 1.8055331707000732,
      "learning_rate": 4.794903582891115e-07,
      "loss": 0.4454,
      "step": 392400
    },
    {
      "epoch": 160.63,
      "grad_norm": 2.3247036933898926,
      "learning_rate": 4.793942232255741e-07,
      "loss": 0.4301,
      "step": 392410
    },
    {
      "epoch": 160.63,
      "grad_norm": 1.9595248699188232,
      "learning_rate": 4.792980967023619e-07,
      "loss": 0.4491,
      "step": 392420
    },
    {
      "epoch": 160.63,
      "grad_norm": 2.267932653427124,
      "learning_rate": 4.792019787199152e-07,
      "loss": 0.4541,
      "step": 392430
    },
    {
      "epoch": 160.64,
      "grad_norm": 2.3001646995544434,
      "learning_rate": 4.791058692786744e-07,
      "loss": 0.4496,
      "step": 392440
    },
    {
      "epoch": 160.64,
      "grad_norm": 1.7513883113861084,
      "learning_rate": 4.790097683790792e-07,
      "loss": 0.441,
      "step": 392450
    },
    {
      "epoch": 160.65,
      "grad_norm": 2.6345455646514893,
      "learning_rate": 4.789136760215711e-07,
      "loss": 0.443,
      "step": 392460
    },
    {
      "epoch": 160.65,
      "grad_norm": 2.084212064743042,
      "learning_rate": 4.788175922065886e-07,
      "loss": 0.4358,
      "step": 392470
    },
    {
      "epoch": 160.65,
      "grad_norm": 2.187192678451538,
      "learning_rate": 4.787215169345729e-07,
      "loss": 0.4551,
      "step": 392480
    },
    {
      "epoch": 160.66,
      "grad_norm": 2.2318286895751953,
      "learning_rate": 4.786254502059636e-07,
      "loss": 0.4339,
      "step": 392490
    },
    {
      "epoch": 160.66,
      "grad_norm": 1.7305341958999634,
      "learning_rate": 4.785293920212013e-07,
      "loss": 0.4579,
      "step": 392500
    },
    {
      "epoch": 160.67,
      "grad_norm": 2.8175249099731445,
      "learning_rate": 4.784333423807247e-07,
      "loss": 0.4472,
      "step": 392510
    },
    {
      "epoch": 160.67,
      "grad_norm": 1.984881043434143,
      "learning_rate": 4.783373012849758e-07,
      "loss": 0.4602,
      "step": 392520
    },
    {
      "epoch": 160.68,
      "grad_norm": 2.4682435989379883,
      "learning_rate": 4.782412687343933e-07,
      "loss": 0.4384,
      "step": 392530
    },
    {
      "epoch": 160.68,
      "grad_norm": 2.2758572101593018,
      "learning_rate": 4.781452447294175e-07,
      "loss": 0.4549,
      "step": 392540
    },
    {
      "epoch": 160.68,
      "grad_norm": 2.41446852684021,
      "learning_rate": 4.780492292704883e-07,
      "loss": 0.4504,
      "step": 392550
    },
    {
      "epoch": 160.69,
      "grad_norm": 2.158836603164673,
      "learning_rate": 4.779532223580451e-07,
      "loss": 0.4625,
      "step": 392560
    },
    {
      "epoch": 160.69,
      "grad_norm": 2.4296932220458984,
      "learning_rate": 4.778572239925283e-07,
      "loss": 0.4503,
      "step": 392570
    },
    {
      "epoch": 160.7,
      "grad_norm": 2.0294086933135986,
      "learning_rate": 4.777612341743772e-07,
      "loss": 0.4272,
      "step": 392580
    },
    {
      "epoch": 160.7,
      "grad_norm": 2.312852382659912,
      "learning_rate": 4.776652529040315e-07,
      "loss": 0.44,
      "step": 392590
    },
    {
      "epoch": 160.7,
      "grad_norm": 1.863576054573059,
      "learning_rate": 4.775692801819318e-07,
      "loss": 0.4309,
      "step": 392600
    },
    {
      "epoch": 160.71,
      "grad_norm": 1.9925775527954102,
      "learning_rate": 4.774733160085166e-07,
      "loss": 0.441,
      "step": 392610
    },
    {
      "epoch": 160.71,
      "grad_norm": 1.820243239402771,
      "learning_rate": 4.773773603842255e-07,
      "loss": 0.4465,
      "step": 392620
    },
    {
      "epoch": 160.72,
      "grad_norm": 1.8510665893554688,
      "learning_rate": 4.772814133094985e-07,
      "loss": 0.4363,
      "step": 392630
    },
    {
      "epoch": 160.72,
      "grad_norm": 2.078005075454712,
      "learning_rate": 4.771854747847751e-07,
      "loss": 0.4623,
      "step": 392640
    },
    {
      "epoch": 160.72,
      "grad_norm": 2.113887071609497,
      "learning_rate": 4.770895448104946e-07,
      "loss": 0.4559,
      "step": 392650
    },
    {
      "epoch": 160.73,
      "grad_norm": 2.30214524269104,
      "learning_rate": 4.769936233870967e-07,
      "loss": 0.4389,
      "step": 392660
    },
    {
      "epoch": 160.73,
      "grad_norm": 1.814645528793335,
      "learning_rate": 4.768977105150204e-07,
      "loss": 0.4618,
      "step": 392670
    },
    {
      "epoch": 160.74,
      "grad_norm": 2.382661819458008,
      "learning_rate": 4.768018061947057e-07,
      "loss": 0.4529,
      "step": 392680
    },
    {
      "epoch": 160.74,
      "grad_norm": 2.115262269973755,
      "learning_rate": 4.767059104265914e-07,
      "loss": 0.444,
      "step": 392690
    },
    {
      "epoch": 160.74,
      "grad_norm": 2.111396312713623,
      "learning_rate": 4.7661002321111614e-07,
      "loss": 0.435,
      "step": 392700
    },
    {
      "epoch": 160.75,
      "grad_norm": 2.0371999740600586,
      "learning_rate": 4.765141445487207e-07,
      "loss": 0.4519,
      "step": 392710
    },
    {
      "epoch": 160.75,
      "grad_norm": 2.1376430988311768,
      "learning_rate": 4.7641827443984336e-07,
      "loss": 0.4504,
      "step": 392720
    },
    {
      "epoch": 160.76,
      "grad_norm": 1.7113463878631592,
      "learning_rate": 4.763224128849243e-07,
      "loss": 0.4451,
      "step": 392730
    },
    {
      "epoch": 160.76,
      "grad_norm": 1.975874662399292,
      "learning_rate": 4.7622655988440127e-07,
      "loss": 0.4578,
      "step": 392740
    },
    {
      "epoch": 160.77,
      "grad_norm": 1.8002392053604126,
      "learning_rate": 4.7613071543871374e-07,
      "loss": 0.464,
      "step": 392750
    },
    {
      "epoch": 160.77,
      "grad_norm": 1.9347401857376099,
      "learning_rate": 4.760348795483009e-07,
      "loss": 0.44,
      "step": 392760
    },
    {
      "epoch": 160.77,
      "grad_norm": 2.406543493270874,
      "learning_rate": 4.7593905221360177e-07,
      "loss": 0.4587,
      "step": 392770
    },
    {
      "epoch": 160.78,
      "grad_norm": 2.405308961868286,
      "learning_rate": 4.7584323343505527e-07,
      "loss": 0.4601,
      "step": 392780
    },
    {
      "epoch": 160.78,
      "grad_norm": 1.5828012228012085,
      "learning_rate": 4.7574742321310044e-07,
      "loss": 0.4418,
      "step": 392790
    },
    {
      "epoch": 160.79,
      "grad_norm": 1.9673529863357544,
      "learning_rate": 4.7565162154817617e-07,
      "loss": 0.4344,
      "step": 392800
    },
    {
      "epoch": 160.79,
      "grad_norm": 2.2861435413360596,
      "learning_rate": 4.7555582844072126e-07,
      "loss": 0.4523,
      "step": 392810
    },
    {
      "epoch": 160.79,
      "grad_norm": 1.897666573524475,
      "learning_rate": 4.7546004389117463e-07,
      "loss": 0.4635,
      "step": 392820
    },
    {
      "epoch": 160.8,
      "grad_norm": 2.7227842807769775,
      "learning_rate": 4.7536426789997474e-07,
      "loss": 0.4386,
      "step": 392830
    },
    {
      "epoch": 160.8,
      "grad_norm": 2.3967394828796387,
      "learning_rate": 4.7526850046756055e-07,
      "loss": 0.4471,
      "step": 392840
    },
    {
      "epoch": 160.81,
      "grad_norm": 2.1017093658447266,
      "learning_rate": 4.7517274159437104e-07,
      "loss": 0.4581,
      "step": 392850
    },
    {
      "epoch": 160.81,
      "grad_norm": 2.1018543243408203,
      "learning_rate": 4.7507699128084413e-07,
      "loss": 0.4356,
      "step": 392860
    },
    {
      "epoch": 160.81,
      "grad_norm": 2.1768689155578613,
      "learning_rate": 4.749812495274191e-07,
      "loss": 0.4626,
      "step": 392870
    },
    {
      "epoch": 160.82,
      "grad_norm": 4.300326347351074,
      "learning_rate": 4.7488551633453436e-07,
      "loss": 0.4523,
      "step": 392880
    },
    {
      "epoch": 160.82,
      "grad_norm": 2.1426448822021484,
      "learning_rate": 4.7478979170262835e-07,
      "loss": 0.4382,
      "step": 392890
    },
    {
      "epoch": 160.83,
      "grad_norm": 2.0848236083984375,
      "learning_rate": 4.7469407563213956e-07,
      "loss": 0.44,
      "step": 392900
    },
    {
      "epoch": 160.83,
      "grad_norm": 2.5337464809417725,
      "learning_rate": 4.745983681235064e-07,
      "loss": 0.437,
      "step": 392910
    },
    {
      "epoch": 160.84,
      "grad_norm": 2.3917994499206543,
      "learning_rate": 4.7450266917716765e-07,
      "loss": 0.4421,
      "step": 392920
    },
    {
      "epoch": 160.84,
      "grad_norm": 1.9062143564224243,
      "learning_rate": 4.7440697879356115e-07,
      "loss": 0.4662,
      "step": 392930
    },
    {
      "epoch": 160.84,
      "grad_norm": 2.490020990371704,
      "learning_rate": 4.7431129697312567e-07,
      "loss": 0.4329,
      "step": 392940
    },
    {
      "epoch": 160.85,
      "grad_norm": 2.1106624603271484,
      "learning_rate": 4.7421562371629937e-07,
      "loss": 0.459,
      "step": 392950
    },
    {
      "epoch": 160.85,
      "grad_norm": 2.3886590003967285,
      "learning_rate": 4.741199590235205e-07,
      "loss": 0.4462,
      "step": 392960
    },
    {
      "epoch": 160.86,
      "grad_norm": 2.022340774536133,
      "learning_rate": 4.7402430289522716e-07,
      "loss": 0.4458,
      "step": 392970
    },
    {
      "epoch": 160.86,
      "grad_norm": 2.019395589828491,
      "learning_rate": 4.739286553318579e-07,
      "loss": 0.4523,
      "step": 392980
    },
    {
      "epoch": 160.86,
      "grad_norm": 2.194704532623291,
      "learning_rate": 4.7383301633385056e-07,
      "loss": 0.4384,
      "step": 392990
    },
    {
      "epoch": 160.87,
      "grad_norm": 1.9558372497558594,
      "learning_rate": 4.737373859016434e-07,
      "loss": 0.4511,
      "step": 393000
    },
    {
      "epoch": 160.87,
      "grad_norm": 1.828081488609314,
      "learning_rate": 4.7364176403567427e-07,
      "loss": 0.4551,
      "step": 393010
    },
    {
      "epoch": 160.88,
      "grad_norm": 1.839842438697815,
      "learning_rate": 4.7354615073638145e-07,
      "loss": 0.4539,
      "step": 393020
    },
    {
      "epoch": 160.88,
      "grad_norm": 2.16139817237854,
      "learning_rate": 4.7345054600420283e-07,
      "loss": 0.4686,
      "step": 393030
    },
    {
      "epoch": 160.88,
      "grad_norm": 2.1289095878601074,
      "learning_rate": 4.733549498395769e-07,
      "loss": 0.4497,
      "step": 393040
    },
    {
      "epoch": 160.89,
      "grad_norm": 2.3315539360046387,
      "learning_rate": 4.732593622429405e-07,
      "loss": 0.4393,
      "step": 393050
    },
    {
      "epoch": 160.89,
      "grad_norm": 1.9280622005462646,
      "learning_rate": 4.731637832147318e-07,
      "loss": 0.4377,
      "step": 393060
    },
    {
      "epoch": 160.9,
      "grad_norm": 2.3797686100006104,
      "learning_rate": 4.7306821275538855e-07,
      "loss": 0.4395,
      "step": 393070
    },
    {
      "epoch": 160.9,
      "grad_norm": 2.3214337825775146,
      "learning_rate": 4.729726508653494e-07,
      "loss": 0.4499,
      "step": 393080
    },
    {
      "epoch": 160.9,
      "grad_norm": 2.0114922523498535,
      "learning_rate": 4.728770975450513e-07,
      "loss": 0.4321,
      "step": 393090
    },
    {
      "epoch": 160.91,
      "grad_norm": 2.158522605895996,
      "learning_rate": 4.727815527949326e-07,
      "loss": 0.4643,
      "step": 393100
    },
    {
      "epoch": 160.91,
      "grad_norm": 2.416703939437866,
      "learning_rate": 4.726860166154303e-07,
      "loss": 0.4514,
      "step": 393110
    },
    {
      "epoch": 160.92,
      "grad_norm": 2.070967674255371,
      "learning_rate": 4.725904890069825e-07,
      "loss": 0.468,
      "step": 393120
    },
    {
      "epoch": 160.92,
      "grad_norm": 1.649631142616272,
      "learning_rate": 4.724949699700267e-07,
      "loss": 0.4483,
      "step": 393130
    },
    {
      "epoch": 160.93,
      "grad_norm": 1.9922279119491577,
      "learning_rate": 4.723994595050002e-07,
      "loss": 0.4442,
      "step": 393140
    },
    {
      "epoch": 160.93,
      "grad_norm": 2.0282320976257324,
      "learning_rate": 4.7230395761234084e-07,
      "loss": 0.4478,
      "step": 393150
    },
    {
      "epoch": 160.93,
      "grad_norm": 2.0120387077331543,
      "learning_rate": 4.7220846429248583e-07,
      "loss": 0.452,
      "step": 393160
    },
    {
      "epoch": 160.94,
      "grad_norm": 2.036102771759033,
      "learning_rate": 4.72112979545873e-07,
      "loss": 0.4494,
      "step": 393170
    },
    {
      "epoch": 160.94,
      "grad_norm": 1.9072048664093018,
      "learning_rate": 4.7201750337293993e-07,
      "loss": 0.4585,
      "step": 393180
    },
    {
      "epoch": 160.95,
      "grad_norm": 2.0717666149139404,
      "learning_rate": 4.71922035774123e-07,
      "loss": 0.4467,
      "step": 393190
    },
    {
      "epoch": 160.95,
      "grad_norm": 1.722475528717041,
      "learning_rate": 4.7182657674985993e-07,
      "loss": 0.4281,
      "step": 393200
    },
    {
      "epoch": 160.95,
      "grad_norm": 1.9157689809799194,
      "learning_rate": 4.7173112630058834e-07,
      "loss": 0.4561,
      "step": 393210
    },
    {
      "epoch": 160.96,
      "grad_norm": 2.6526756286621094,
      "learning_rate": 4.716356844267454e-07,
      "loss": 0.4847,
      "step": 393220
    },
    {
      "epoch": 160.96,
      "grad_norm": 1.866823673248291,
      "learning_rate": 4.7154025112876797e-07,
      "loss": 0.4427,
      "step": 393230
    },
    {
      "epoch": 160.97,
      "grad_norm": 2.2277157306671143,
      "learning_rate": 4.7144482640709354e-07,
      "loss": 0.44,
      "step": 393240
    },
    {
      "epoch": 160.97,
      "grad_norm": 2.04107403755188,
      "learning_rate": 4.7134941026215916e-07,
      "loss": 0.4543,
      "step": 393250
    },
    {
      "epoch": 160.97,
      "grad_norm": 2.4244823455810547,
      "learning_rate": 4.7125400269440126e-07,
      "loss": 0.4373,
      "step": 393260
    },
    {
      "epoch": 160.98,
      "grad_norm": 1.9327927827835083,
      "learning_rate": 4.711586037042581e-07,
      "loss": 0.4503,
      "step": 393270
    },
    {
      "epoch": 160.98,
      "grad_norm": 2.0811569690704346,
      "learning_rate": 4.710632132921661e-07,
      "loss": 0.4363,
      "step": 393280
    },
    {
      "epoch": 160.99,
      "grad_norm": 2.141695499420166,
      "learning_rate": 4.7096783145856236e-07,
      "loss": 0.4489,
      "step": 393290
    },
    {
      "epoch": 160.99,
      "grad_norm": 2.0597944259643555,
      "learning_rate": 4.708724582038837e-07,
      "loss": 0.4503,
      "step": 393300
    },
    {
      "epoch": 160.99,
      "grad_norm": 1.8137762546539307,
      "learning_rate": 4.7077709352856696e-07,
      "loss": 0.4409,
      "step": 393310
    },
    {
      "epoch": 161.0,
      "grad_norm": 1.9322699308395386,
      "learning_rate": 4.706817374330496e-07,
      "loss": 0.4399,
      "step": 393320
    },
    {
      "epoch": 161.0,
      "eval_loss": 0.4488338828086853,
      "eval_runtime": 58.2319,
      "eval_samples_per_second": 59.229,
      "eval_steps_per_second": 7.419,
      "step": 393323
    },
    {
      "epoch": 161.0,
      "grad_norm": 2.2694830894470215,
      "learning_rate": 4.7058638991776725e-07,
      "loss": 0.4714,
      "step": 393330
    },
    {
      "epoch": 161.01,
      "grad_norm": 1.8946645259857178,
      "learning_rate": 4.7049105098315747e-07,
      "loss": 0.4279,
      "step": 393340
    },
    {
      "epoch": 161.01,
      "grad_norm": 1.925947666168213,
      "learning_rate": 4.7039572062965673e-07,
      "loss": 0.4471,
      "step": 393350
    },
    {
      "epoch": 161.02,
      "grad_norm": 1.9375317096710205,
      "learning_rate": 4.7030039885770184e-07,
      "loss": 0.4411,
      "step": 393360
    },
    {
      "epoch": 161.02,
      "grad_norm": 1.9447585344314575,
      "learning_rate": 4.702050856677295e-07,
      "loss": 0.4482,
      "step": 393370
    },
    {
      "epoch": 161.02,
      "grad_norm": 2.061394453048706,
      "learning_rate": 4.7010978106017604e-07,
      "loss": 0.4519,
      "step": 393380
    },
    {
      "epoch": 161.03,
      "grad_norm": 2.140890598297119,
      "learning_rate": 4.700144850354783e-07,
      "loss": 0.4597,
      "step": 393390
    },
    {
      "epoch": 161.03,
      "grad_norm": 2.165382146835327,
      "learning_rate": 4.699191975940728e-07,
      "loss": 0.4464,
      "step": 393400
    },
    {
      "epoch": 161.04,
      "grad_norm": 2.1447067260742188,
      "learning_rate": 4.698239187363961e-07,
      "loss": 0.445,
      "step": 393410
    },
    {
      "epoch": 161.04,
      "grad_norm": 1.802872657775879,
      "learning_rate": 4.697286484628845e-07,
      "loss": 0.4402,
      "step": 393420
    },
    {
      "epoch": 161.04,
      "grad_norm": 2.3245530128479004,
      "learning_rate": 4.6963338677397467e-07,
      "loss": 0.4343,
      "step": 393430
    },
    {
      "epoch": 161.05,
      "grad_norm": 1.7672109603881836,
      "learning_rate": 4.6953813367010244e-07,
      "loss": 0.4366,
      "step": 393440
    },
    {
      "epoch": 161.05,
      "grad_norm": 2.1402246952056885,
      "learning_rate": 4.6944288915170476e-07,
      "loss": 0.4299,
      "step": 393450
    },
    {
      "epoch": 161.06,
      "grad_norm": 1.981934666633606,
      "learning_rate": 4.6934765321921775e-07,
      "loss": 0.4536,
      "step": 393460
    },
    {
      "epoch": 161.06,
      "grad_norm": 1.9857604503631592,
      "learning_rate": 4.692524258730774e-07,
      "loss": 0.4528,
      "step": 393470
    },
    {
      "epoch": 161.06,
      "grad_norm": 2.0672571659088135,
      "learning_rate": 4.6915720711372024e-07,
      "loss": 0.4521,
      "step": 393480
    },
    {
      "epoch": 161.07,
      "grad_norm": 1.9727582931518555,
      "learning_rate": 4.6906199694158224e-07,
      "loss": 0.4424,
      "step": 393490
    },
    {
      "epoch": 161.07,
      "grad_norm": 1.9721177816390991,
      "learning_rate": 4.689667953570996e-07,
      "loss": 0.4567,
      "step": 393500
    },
    {
      "epoch": 161.08,
      "grad_norm": 1.982762336730957,
      "learning_rate": 4.6887160236070864e-07,
      "loss": 0.459,
      "step": 393510
    },
    {
      "epoch": 161.08,
      "grad_norm": 2.1642842292785645,
      "learning_rate": 4.687764179528452e-07,
      "loss": 0.4523,
      "step": 393520
    },
    {
      "epoch": 161.08,
      "grad_norm": 2.094590425491333,
      "learning_rate": 4.6868124213394544e-07,
      "loss": 0.4405,
      "step": 393530
    },
    {
      "epoch": 161.09,
      "grad_norm": 2.108534812927246,
      "learning_rate": 4.685860749044452e-07,
      "loss": 0.4377,
      "step": 393540
    },
    {
      "epoch": 161.09,
      "grad_norm": 1.959328532218933,
      "learning_rate": 4.6849091626478056e-07,
      "loss": 0.4476,
      "step": 393550
    },
    {
      "epoch": 161.1,
      "grad_norm": 2.6677281856536865,
      "learning_rate": 4.683957662153874e-07,
      "loss": 0.4304,
      "step": 393560
    },
    {
      "epoch": 161.1,
      "grad_norm": 2.354451894760132,
      "learning_rate": 4.6830062475670124e-07,
      "loss": 0.4634,
      "step": 393570
    },
    {
      "epoch": 161.11,
      "grad_norm": 2.1164863109588623,
      "learning_rate": 4.6820549188915853e-07,
      "loss": 0.4281,
      "step": 393580
    },
    {
      "epoch": 161.11,
      "grad_norm": 1.8396060466766357,
      "learning_rate": 4.681103676131946e-07,
      "loss": 0.473,
      "step": 393590
    },
    {
      "epoch": 161.11,
      "grad_norm": 1.8216969966888428,
      "learning_rate": 4.680152519292455e-07,
      "loss": 0.4375,
      "step": 393600
    },
    {
      "epoch": 161.12,
      "grad_norm": 1.916247844696045,
      "learning_rate": 4.6792014483774657e-07,
      "loss": 0.4393,
      "step": 393610
    },
    {
      "epoch": 161.12,
      "grad_norm": 2.0036001205444336,
      "learning_rate": 4.6782504633913447e-07,
      "loss": 0.4595,
      "step": 393620
    },
    {
      "epoch": 161.13,
      "grad_norm": 2.177945375442505,
      "learning_rate": 4.677299564338435e-07,
      "loss": 0.44,
      "step": 393630
    },
    {
      "epoch": 161.13,
      "grad_norm": 2.369041681289673,
      "learning_rate": 4.67634875122309e-07,
      "loss": 0.4533,
      "step": 393640
    },
    {
      "epoch": 161.13,
      "grad_norm": 1.974022388458252,
      "learning_rate": 4.6753980240496803e-07,
      "loss": 0.445,
      "step": 393650
    },
    {
      "epoch": 161.14,
      "grad_norm": 1.7630069255828857,
      "learning_rate": 4.6744473828225546e-07,
      "loss": 0.4346,
      "step": 393660
    },
    {
      "epoch": 161.14,
      "grad_norm": 1.9724255800247192,
      "learning_rate": 4.673496827546066e-07,
      "loss": 0.4575,
      "step": 393670
    },
    {
      "epoch": 161.15,
      "grad_norm": 2.168243408203125,
      "learning_rate": 4.6725463582245706e-07,
      "loss": 0.4275,
      "step": 393680
    },
    {
      "epoch": 161.15,
      "grad_norm": 1.6942366361618042,
      "learning_rate": 4.671595974862422e-07,
      "loss": 0.4438,
      "step": 393690
    },
    {
      "epoch": 161.15,
      "grad_norm": 2.1145739555358887,
      "learning_rate": 4.670645677463973e-07,
      "loss": 0.4669,
      "step": 393700
    },
    {
      "epoch": 161.16,
      "grad_norm": 2.0889744758605957,
      "learning_rate": 4.669695466033579e-07,
      "loss": 0.4449,
      "step": 393710
    },
    {
      "epoch": 161.16,
      "grad_norm": 2.0504465103149414,
      "learning_rate": 4.6687453405755894e-07,
      "loss": 0.4451,
      "step": 393720
    },
    {
      "epoch": 161.17,
      "grad_norm": 1.8156769275665283,
      "learning_rate": 4.667795301094359e-07,
      "loss": 0.4427,
      "step": 393730
    },
    {
      "epoch": 161.17,
      "grad_norm": 2.1775529384613037,
      "learning_rate": 4.666845347594236e-07,
      "loss": 0.4441,
      "step": 393740
    },
    {
      "epoch": 161.17,
      "grad_norm": 1.6797887086868286,
      "learning_rate": 4.665895480079577e-07,
      "loss": 0.4478,
      "step": 393750
    },
    {
      "epoch": 161.18,
      "grad_norm": 1.8812578916549683,
      "learning_rate": 4.66494569855474e-07,
      "loss": 0.4659,
      "step": 393760
    },
    {
      "epoch": 161.18,
      "grad_norm": 1.930643081665039,
      "learning_rate": 4.663996003024057e-07,
      "loss": 0.4459,
      "step": 393770
    },
    {
      "epoch": 161.19,
      "grad_norm": 1.8845021724700928,
      "learning_rate": 4.6630463934918905e-07,
      "loss": 0.4373,
      "step": 393780
    },
    {
      "epoch": 161.19,
      "grad_norm": 2.0833535194396973,
      "learning_rate": 4.662096869962587e-07,
      "loss": 0.4606,
      "step": 393790
    },
    {
      "epoch": 161.2,
      "grad_norm": 2.532080888748169,
      "learning_rate": 4.661147432440498e-07,
      "loss": 0.4608,
      "step": 393800
    },
    {
      "epoch": 161.2,
      "grad_norm": 2.7393853664398193,
      "learning_rate": 4.6601980809299716e-07,
      "loss": 0.4556,
      "step": 393810
    },
    {
      "epoch": 161.2,
      "grad_norm": 2.0706980228424072,
      "learning_rate": 4.659248815435358e-07,
      "loss": 0.4541,
      "step": 393820
    },
    {
      "epoch": 161.21,
      "grad_norm": 1.604845643043518,
      "learning_rate": 4.658299635961001e-07,
      "loss": 0.4532,
      "step": 393830
    },
    {
      "epoch": 161.21,
      "grad_norm": 2.2195394039154053,
      "learning_rate": 4.657350542511255e-07,
      "loss": 0.4484,
      "step": 393840
    },
    {
      "epoch": 161.22,
      "grad_norm": 2.0448250770568848,
      "learning_rate": 4.6564015350904674e-07,
      "loss": 0.4519,
      "step": 393850
    },
    {
      "epoch": 161.22,
      "grad_norm": 2.2844650745391846,
      "learning_rate": 4.6554526137029824e-07,
      "loss": 0.4381,
      "step": 393860
    },
    {
      "epoch": 161.22,
      "grad_norm": 1.9159550666809082,
      "learning_rate": 4.654503778353148e-07,
      "loss": 0.4358,
      "step": 393870
    },
    {
      "epoch": 161.23,
      "grad_norm": 7.209243297576904,
      "learning_rate": 4.6535550290453105e-07,
      "loss": 0.4314,
      "step": 393880
    },
    {
      "epoch": 161.23,
      "grad_norm": 2.4756500720977783,
      "learning_rate": 4.652606365783814e-07,
      "loss": 0.4444,
      "step": 393890
    },
    {
      "epoch": 161.24,
      "grad_norm": 1.9084733724594116,
      "learning_rate": 4.651657788573014e-07,
      "loss": 0.4501,
      "step": 393900
    },
    {
      "epoch": 161.24,
      "grad_norm": 4.029448509216309,
      "learning_rate": 4.650709297417241e-07,
      "loss": 0.4394,
      "step": 393910
    },
    {
      "epoch": 161.24,
      "grad_norm": 2.1008150577545166,
      "learning_rate": 4.649760892320849e-07,
      "loss": 0.445,
      "step": 393920
    },
    {
      "epoch": 161.25,
      "grad_norm": 2.2050328254699707,
      "learning_rate": 4.6488125732881784e-07,
      "loss": 0.433,
      "step": 393930
    },
    {
      "epoch": 161.25,
      "grad_norm": 1.9391024112701416,
      "learning_rate": 4.647864340323574e-07,
      "loss": 0.4321,
      "step": 393940
    },
    {
      "epoch": 161.26,
      "grad_norm": 1.7743431329727173,
      "learning_rate": 4.6469161934313825e-07,
      "loss": 0.4551,
      "step": 393950
    },
    {
      "epoch": 161.26,
      "grad_norm": 2.1805694103240967,
      "learning_rate": 4.645968132615943e-07,
      "loss": 0.4397,
      "step": 393960
    },
    {
      "epoch": 161.26,
      "grad_norm": 2.2628185749053955,
      "learning_rate": 4.645020157881602e-07,
      "loss": 0.4484,
      "step": 393970
    },
    {
      "epoch": 161.27,
      "grad_norm": 1.9033174514770508,
      "learning_rate": 4.6440722692327016e-07,
      "loss": 0.4353,
      "step": 393980
    },
    {
      "epoch": 161.27,
      "grad_norm": 1.7336573600769043,
      "learning_rate": 4.6431244666735824e-07,
      "loss": 0.4403,
      "step": 393990
    },
    {
      "epoch": 161.28,
      "grad_norm": 2.277672290802002,
      "learning_rate": 4.642176750208587e-07,
      "loss": 0.4431,
      "step": 394000
    },
    {
      "epoch": 161.28,
      "grad_norm": 2.1086912155151367,
      "learning_rate": 4.641229119842056e-07,
      "loss": 0.4471,
      "step": 394010
    },
    {
      "epoch": 161.29,
      "grad_norm": 2.3535966873168945,
      "learning_rate": 4.6402815755783244e-07,
      "loss": 0.4469,
      "step": 394020
    },
    {
      "epoch": 161.29,
      "grad_norm": 1.9303463697433472,
      "learning_rate": 4.6393341174217534e-07,
      "loss": 0.4685,
      "step": 394030
    },
    {
      "epoch": 161.29,
      "grad_norm": 2.223862409591675,
      "learning_rate": 4.6383867453766624e-07,
      "loss": 0.4332,
      "step": 394040
    },
    {
      "epoch": 161.3,
      "grad_norm": 1.871883511543274,
      "learning_rate": 4.6374394594473964e-07,
      "loss": 0.4302,
      "step": 394050
    },
    {
      "epoch": 161.3,
      "grad_norm": 2.117820978164673,
      "learning_rate": 4.636492259638299e-07,
      "loss": 0.4404,
      "step": 394060
    },
    {
      "epoch": 161.31,
      "grad_norm": 2.1947453022003174,
      "learning_rate": 4.6355451459537026e-07,
      "loss": 0.4436,
      "step": 394070
    },
    {
      "epoch": 161.31,
      "grad_norm": 2.026979446411133,
      "learning_rate": 4.634598118397952e-07,
      "loss": 0.4429,
      "step": 394080
    },
    {
      "epoch": 161.31,
      "grad_norm": 1.7403967380523682,
      "learning_rate": 4.6336511769753827e-07,
      "loss": 0.4503,
      "step": 394090
    },
    {
      "epoch": 161.32,
      "grad_norm": 1.9464399814605713,
      "learning_rate": 4.6327043216903323e-07,
      "loss": 0.4546,
      "step": 394100
    },
    {
      "epoch": 161.32,
      "grad_norm": 2.239006757736206,
      "learning_rate": 4.631757552547139e-07,
      "loss": 0.4381,
      "step": 394110
    },
    {
      "epoch": 161.33,
      "grad_norm": 2.0290634632110596,
      "learning_rate": 4.63081086955014e-07,
      "loss": 0.4515,
      "step": 394120
    },
    {
      "epoch": 161.33,
      "grad_norm": 1.958910584449768,
      "learning_rate": 4.62986427270367e-07,
      "loss": 0.4346,
      "step": 394130
    },
    {
      "epoch": 161.33,
      "grad_norm": 2.04679799079895,
      "learning_rate": 4.628917762012067e-07,
      "loss": 0.4472,
      "step": 394140
    },
    {
      "epoch": 161.34,
      "grad_norm": 1.7976264953613281,
      "learning_rate": 4.627971337479667e-07,
      "loss": 0.4365,
      "step": 394150
    },
    {
      "epoch": 161.34,
      "grad_norm": 1.9837942123413086,
      "learning_rate": 4.627024999110804e-07,
      "loss": 0.4362,
      "step": 394160
    },
    {
      "epoch": 161.35,
      "grad_norm": 3.1614208221435547,
      "learning_rate": 4.6260787469098126e-07,
      "loss": 0.4205,
      "step": 394170
    },
    {
      "epoch": 161.35,
      "grad_norm": 2.1733388900756836,
      "learning_rate": 4.625132580881029e-07,
      "loss": 0.4411,
      "step": 394180
    },
    {
      "epoch": 161.35,
      "grad_norm": 2.108400583267212,
      "learning_rate": 4.6241865010287883e-07,
      "loss": 0.4504,
      "step": 394190
    },
    {
      "epoch": 161.36,
      "grad_norm": 1.9483884572982788,
      "learning_rate": 4.6232405073574245e-07,
      "loss": 0.4383,
      "step": 394200
    },
    {
      "epoch": 161.36,
      "grad_norm": 2.1469149589538574,
      "learning_rate": 4.622294599871268e-07,
      "loss": 0.4514,
      "step": 394210
    },
    {
      "epoch": 161.37,
      "grad_norm": 2.0674216747283936,
      "learning_rate": 4.621348778574652e-07,
      "loss": 0.4433,
      "step": 394220
    },
    {
      "epoch": 161.37,
      "grad_norm": 2.1297919750213623,
      "learning_rate": 4.6204030434719136e-07,
      "loss": 0.4455,
      "step": 394230
    },
    {
      "epoch": 161.38,
      "grad_norm": 2.440995454788208,
      "learning_rate": 4.61945739456738e-07,
      "loss": 0.446,
      "step": 394240
    },
    {
      "epoch": 161.38,
      "grad_norm": 1.9955986738204956,
      "learning_rate": 4.6185118318653855e-07,
      "loss": 0.4663,
      "step": 394250
    },
    {
      "epoch": 161.38,
      "grad_norm": 2.099576711654663,
      "learning_rate": 4.617566355370264e-07,
      "loss": 0.4512,
      "step": 394260
    },
    {
      "epoch": 161.39,
      "grad_norm": 2.345271110534668,
      "learning_rate": 4.6166209650863423e-07,
      "loss": 0.4448,
      "step": 394270
    },
    {
      "epoch": 161.39,
      "grad_norm": 1.878960132598877,
      "learning_rate": 4.6156756610179524e-07,
      "loss": 0.4615,
      "step": 394280
    },
    {
      "epoch": 161.4,
      "grad_norm": 1.8476011753082275,
      "learning_rate": 4.6147304431694246e-07,
      "loss": 0.4458,
      "step": 394290
    },
    {
      "epoch": 161.4,
      "grad_norm": 1.7390186786651611,
      "learning_rate": 4.6137853115450885e-07,
      "loss": 0.4448,
      "step": 394300
    },
    {
      "epoch": 161.4,
      "grad_norm": 2.1500918865203857,
      "learning_rate": 4.612840266149274e-07,
      "loss": 0.4457,
      "step": 394310
    },
    {
      "epoch": 161.41,
      "grad_norm": 1.89137601852417,
      "learning_rate": 4.6118953069863114e-07,
      "loss": 0.4426,
      "step": 394320
    },
    {
      "epoch": 161.41,
      "grad_norm": 2.175501585006714,
      "learning_rate": 4.6109504340605276e-07,
      "loss": 0.4417,
      "step": 394330
    },
    {
      "epoch": 161.42,
      "grad_norm": 1.6793086528778076,
      "learning_rate": 4.6100056473762604e-07,
      "loss": 0.4423,
      "step": 394340
    },
    {
      "epoch": 161.42,
      "grad_norm": 2.3395893573760986,
      "learning_rate": 4.6090609469378185e-07,
      "loss": 0.4369,
      "step": 394350
    },
    {
      "epoch": 161.42,
      "grad_norm": 2.1681621074676514,
      "learning_rate": 4.6081163327495426e-07,
      "loss": 0.4586,
      "step": 394360
    },
    {
      "epoch": 161.43,
      "grad_norm": 2.0141561031341553,
      "learning_rate": 4.607171804815757e-07,
      "loss": 0.4578,
      "step": 394370
    },
    {
      "epoch": 161.43,
      "grad_norm": 1.8594765663146973,
      "learning_rate": 4.6062273631407895e-07,
      "loss": 0.4546,
      "step": 394380
    },
    {
      "epoch": 161.44,
      "grad_norm": 2.2610318660736084,
      "learning_rate": 4.605283007728957e-07,
      "loss": 0.4594,
      "step": 394390
    },
    {
      "epoch": 161.44,
      "grad_norm": 2.3404808044433594,
      "learning_rate": 4.6043387385846013e-07,
      "loss": 0.4469,
      "step": 394400
    },
    {
      "epoch": 161.44,
      "grad_norm": 1.9779199361801147,
      "learning_rate": 4.6033945557120433e-07,
      "loss": 0.4374,
      "step": 394410
    },
    {
      "epoch": 161.45,
      "grad_norm": 1.786916971206665,
      "learning_rate": 4.602450459115601e-07,
      "loss": 0.4513,
      "step": 394420
    },
    {
      "epoch": 161.45,
      "grad_norm": 2.6046102046966553,
      "learning_rate": 4.6015064487996083e-07,
      "loss": 0.4395,
      "step": 394430
    },
    {
      "epoch": 161.46,
      "grad_norm": 1.9962358474731445,
      "learning_rate": 4.600562524768381e-07,
      "loss": 0.4235,
      "step": 394440
    },
    {
      "epoch": 161.46,
      "grad_norm": 1.929555892944336,
      "learning_rate": 4.5996186870262494e-07,
      "loss": 0.437,
      "step": 394450
    },
    {
      "epoch": 161.47,
      "grad_norm": 1.7376749515533447,
      "learning_rate": 4.598674935577533e-07,
      "loss": 0.4257,
      "step": 394460
    },
    {
      "epoch": 161.47,
      "grad_norm": 1.9367395639419556,
      "learning_rate": 4.5977312704265563e-07,
      "loss": 0.4753,
      "step": 394470
    },
    {
      "epoch": 161.47,
      "grad_norm": 1.8685030937194824,
      "learning_rate": 4.596787691577652e-07,
      "loss": 0.4391,
      "step": 394480
    },
    {
      "epoch": 161.48,
      "grad_norm": 2.271519660949707,
      "learning_rate": 4.595844199035124e-07,
      "loss": 0.4381,
      "step": 394490
    },
    {
      "epoch": 161.48,
      "grad_norm": 1.733070969581604,
      "learning_rate": 4.5949007928033044e-07,
      "loss": 0.4539,
      "step": 394500
    },
    {
      "epoch": 161.49,
      "grad_norm": 2.090160369873047,
      "learning_rate": 4.593957472886513e-07,
      "loss": 0.4542,
      "step": 394510
    },
    {
      "epoch": 161.49,
      "grad_norm": 1.6275416612625122,
      "learning_rate": 4.593014239289071e-07,
      "loss": 0.4462,
      "step": 394520
    },
    {
      "epoch": 161.49,
      "grad_norm": 1.8070991039276123,
      "learning_rate": 4.5920710920152993e-07,
      "loss": 0.4529,
      "step": 394530
    },
    {
      "epoch": 161.5,
      "grad_norm": 1.8178296089172363,
      "learning_rate": 4.5911280310695214e-07,
      "loss": 0.4427,
      "step": 394540
    },
    {
      "epoch": 161.5,
      "grad_norm": 1.7160555124282837,
      "learning_rate": 4.590185056456052e-07,
      "loss": 0.4436,
      "step": 394550
    },
    {
      "epoch": 161.51,
      "grad_norm": 2.3742220401763916,
      "learning_rate": 4.589242168179216e-07,
      "loss": 0.4499,
      "step": 394560
    },
    {
      "epoch": 161.51,
      "grad_norm": 2.114208936691284,
      "learning_rate": 4.5882993662433276e-07,
      "loss": 0.4376,
      "step": 394570
    },
    {
      "epoch": 161.51,
      "grad_norm": 2.1342661380767822,
      "learning_rate": 4.587356650652703e-07,
      "loss": 0.4482,
      "step": 394580
    },
    {
      "epoch": 161.52,
      "grad_norm": 1.9078632593154907,
      "learning_rate": 4.586414021411673e-07,
      "loss": 0.4543,
      "step": 394590
    },
    {
      "epoch": 161.52,
      "grad_norm": 1.7271292209625244,
      "learning_rate": 4.585471478524548e-07,
      "loss": 0.4338,
      "step": 394600
    },
    {
      "epoch": 161.53,
      "grad_norm": 1.8794991970062256,
      "learning_rate": 4.5845290219956464e-07,
      "loss": 0.4314,
      "step": 394610
    },
    {
      "epoch": 161.53,
      "grad_norm": 1.9923336505889893,
      "learning_rate": 4.5835866518292895e-07,
      "loss": 0.4566,
      "step": 394620
    },
    {
      "epoch": 161.53,
      "grad_norm": 2.17546010017395,
      "learning_rate": 4.5826443680297837e-07,
      "loss": 0.4474,
      "step": 394630
    },
    {
      "epoch": 161.54,
      "grad_norm": 1.9830970764160156,
      "learning_rate": 4.5817021706014535e-07,
      "loss": 0.4394,
      "step": 394640
    },
    {
      "epoch": 161.54,
      "grad_norm": 2.0926969051361084,
      "learning_rate": 4.580760059548611e-07,
      "loss": 0.4581,
      "step": 394650
    },
    {
      "epoch": 161.55,
      "grad_norm": 1.7850160598754883,
      "learning_rate": 4.579818034875575e-07,
      "loss": 0.4421,
      "step": 394660
    },
    {
      "epoch": 161.55,
      "grad_norm": 1.990077257156372,
      "learning_rate": 4.578876096586661e-07,
      "loss": 0.4577,
      "step": 394670
    },
    {
      "epoch": 161.56,
      "grad_norm": 1.966247320175171,
      "learning_rate": 4.5779342446861797e-07,
      "loss": 0.4685,
      "step": 394680
    },
    {
      "epoch": 161.56,
      "grad_norm": 2.1444036960601807,
      "learning_rate": 4.57699247917845e-07,
      "loss": 0.4504,
      "step": 394690
    },
    {
      "epoch": 161.56,
      "grad_norm": 1.9881795644760132,
      "learning_rate": 4.5760508000677825e-07,
      "loss": 0.4362,
      "step": 394700
    },
    {
      "epoch": 161.57,
      "grad_norm": 2.078873872756958,
      "learning_rate": 4.575109207358495e-07,
      "loss": 0.4488,
      "step": 394710
    },
    {
      "epoch": 161.57,
      "grad_norm": 1.9903322458267212,
      "learning_rate": 4.574167701054898e-07,
      "loss": 0.455,
      "step": 394720
    },
    {
      "epoch": 161.58,
      "grad_norm": 2.5697624683380127,
      "learning_rate": 4.573226281161305e-07,
      "loss": 0.4513,
      "step": 394730
    },
    {
      "epoch": 161.58,
      "grad_norm": 2.678117036819458,
      "learning_rate": 4.5722849476820284e-07,
      "loss": 0.4501,
      "step": 394740
    },
    {
      "epoch": 161.58,
      "grad_norm": 2.3983991146087646,
      "learning_rate": 4.5713437006213785e-07,
      "loss": 0.4439,
      "step": 394750
    },
    {
      "epoch": 161.59,
      "grad_norm": 2.448796510696411,
      "learning_rate": 4.5704025399836707e-07,
      "loss": 0.4385,
      "step": 394760
    },
    {
      "epoch": 161.59,
      "grad_norm": 1.8967034816741943,
      "learning_rate": 4.569461465773213e-07,
      "loss": 0.4297,
      "step": 394770
    },
    {
      "epoch": 161.6,
      "grad_norm": 2.3667140007019043,
      "learning_rate": 4.5685204779943184e-07,
      "loss": 0.4321,
      "step": 394780
    },
    {
      "epoch": 161.6,
      "grad_norm": 1.9607704877853394,
      "learning_rate": 4.567579576651298e-07,
      "loss": 0.4388,
      "step": 394790
    },
    {
      "epoch": 161.6,
      "grad_norm": 1.8075815439224243,
      "learning_rate": 4.566638761748458e-07,
      "loss": 0.4461,
      "step": 394800
    },
    {
      "epoch": 161.61,
      "grad_norm": 1.856167197227478,
      "learning_rate": 4.5656980332901125e-07,
      "loss": 0.4664,
      "step": 394810
    },
    {
      "epoch": 161.61,
      "grad_norm": 2.51357102394104,
      "learning_rate": 4.564757391280571e-07,
      "loss": 0.4404,
      "step": 394820
    },
    {
      "epoch": 161.62,
      "grad_norm": 2.086716890335083,
      "learning_rate": 4.5638168357241373e-07,
      "loss": 0.437,
      "step": 394830
    },
    {
      "epoch": 161.62,
      "grad_norm": 1.9522110223770142,
      "learning_rate": 4.562876366625125e-07,
      "loss": 0.4581,
      "step": 394840
    },
    {
      "epoch": 161.63,
      "grad_norm": 2.250871181488037,
      "learning_rate": 4.5619359839878394e-07,
      "loss": 0.431,
      "step": 394850
    },
    {
      "epoch": 161.63,
      "grad_norm": 1.8981846570968628,
      "learning_rate": 4.5609956878165903e-07,
      "loss": 0.4384,
      "step": 394860
    },
    {
      "epoch": 161.63,
      "grad_norm": 2.170372724533081,
      "learning_rate": 4.560055478115683e-07,
      "loss": 0.4568,
      "step": 394870
    },
    {
      "epoch": 161.64,
      "grad_norm": 1.9515125751495361,
      "learning_rate": 4.5591153548894264e-07,
      "loss": 0.4434,
      "step": 394880
    },
    {
      "epoch": 161.64,
      "grad_norm": 1.9709497690200806,
      "learning_rate": 4.558175318142126e-07,
      "loss": 0.4548,
      "step": 394890
    },
    {
      "epoch": 161.65,
      "grad_norm": 1.7917203903198242,
      "learning_rate": 4.5572353678780896e-07,
      "loss": 0.4693,
      "step": 394900
    },
    {
      "epoch": 161.65,
      "grad_norm": 2.312518358230591,
      "learning_rate": 4.5562955041016196e-07,
      "loss": 0.4475,
      "step": 394910
    },
    {
      "epoch": 161.65,
      "grad_norm": 2.015294075012207,
      "learning_rate": 4.555355726817023e-07,
      "loss": 0.4572,
      "step": 394920
    },
    {
      "epoch": 161.66,
      "grad_norm": 2.14729905128479,
      "learning_rate": 4.5544160360286125e-07,
      "loss": 0.4545,
      "step": 394930
    },
    {
      "epoch": 161.66,
      "grad_norm": 1.7846112251281738,
      "learning_rate": 4.5534764317406774e-07,
      "loss": 0.4425,
      "step": 394940
    },
    {
      "epoch": 161.67,
      "grad_norm": 2.0437777042388916,
      "learning_rate": 4.55253691395753e-07,
      "loss": 0.4302,
      "step": 394950
    },
    {
      "epoch": 161.67,
      "grad_norm": 2.3861405849456787,
      "learning_rate": 4.551597482683469e-07,
      "loss": 0.4585,
      "step": 394960
    },
    {
      "epoch": 161.67,
      "grad_norm": 2.0831668376922607,
      "learning_rate": 4.550658137922809e-07,
      "loss": 0.4578,
      "step": 394970
    },
    {
      "epoch": 161.68,
      "grad_norm": 2.461207628250122,
      "learning_rate": 4.5497188796798456e-07,
      "loss": 0.4617,
      "step": 394980
    },
    {
      "epoch": 161.68,
      "grad_norm": 2.0664560794830322,
      "learning_rate": 4.5487797079588825e-07,
      "loss": 0.4485,
      "step": 394990
    },
    {
      "epoch": 161.69,
      "grad_norm": 2.321972608566284,
      "learning_rate": 4.5478406227642207e-07,
      "loss": 0.437,
      "step": 395000
    },
    {
      "epoch": 161.69,
      "grad_norm": 1.8369641304016113,
      "learning_rate": 4.546901624100165e-07,
      "loss": 0.4309,
      "step": 395010
    },
    {
      "epoch": 161.69,
      "grad_norm": 2.2773141860961914,
      "learning_rate": 4.5459627119710137e-07,
      "loss": 0.4636,
      "step": 395020
    },
    {
      "epoch": 161.7,
      "grad_norm": 1.8486785888671875,
      "learning_rate": 4.54502388638107e-07,
      "loss": 0.4374,
      "step": 395030
    },
    {
      "epoch": 161.7,
      "grad_norm": 2.1581385135650635,
      "learning_rate": 4.5440851473346337e-07,
      "loss": 0.4571,
      "step": 395040
    },
    {
      "epoch": 161.71,
      "grad_norm": 2.182054042816162,
      "learning_rate": 4.543146494836002e-07,
      "loss": 0.4724,
      "step": 395050
    },
    {
      "epoch": 161.71,
      "grad_norm": 2.113206148147583,
      "learning_rate": 4.5422079288894855e-07,
      "loss": 0.4422,
      "step": 395060
    },
    {
      "epoch": 161.72,
      "grad_norm": 2.3081676959991455,
      "learning_rate": 4.541269449499371e-07,
      "loss": 0.4722,
      "step": 395070
    },
    {
      "epoch": 161.72,
      "grad_norm": 2.0785417556762695,
      "learning_rate": 4.540331056669961e-07,
      "loss": 0.4629,
      "step": 395080
    },
    {
      "epoch": 161.72,
      "grad_norm": 2.682199716567993,
      "learning_rate": 4.539392750405556e-07,
      "loss": 0.4463,
      "step": 395090
    },
    {
      "epoch": 161.73,
      "grad_norm": 2.1271615028381348,
      "learning_rate": 4.538454530710452e-07,
      "loss": 0.4516,
      "step": 395100
    },
    {
      "epoch": 161.73,
      "grad_norm": 1.6913177967071533,
      "learning_rate": 4.5375163975889474e-07,
      "loss": 0.4402,
      "step": 395110
    },
    {
      "epoch": 161.74,
      "grad_norm": 2.061570644378662,
      "learning_rate": 4.5365783510453444e-07,
      "loss": 0.4666,
      "step": 395120
    },
    {
      "epoch": 161.74,
      "grad_norm": 2.3149099349975586,
      "learning_rate": 4.5356403910839336e-07,
      "loss": 0.4508,
      "step": 395130
    },
    {
      "epoch": 161.74,
      "grad_norm": 1.8187479972839355,
      "learning_rate": 4.534702517709016e-07,
      "loss": 0.4325,
      "step": 395140
    },
    {
      "epoch": 161.75,
      "grad_norm": 1.9900257587432861,
      "learning_rate": 4.533764730924879e-07,
      "loss": 0.4639,
      "step": 395150
    },
    {
      "epoch": 161.75,
      "grad_norm": 3.234759569168091,
      "learning_rate": 4.532827030735831e-07,
      "loss": 0.4476,
      "step": 395160
    },
    {
      "epoch": 161.76,
      "grad_norm": 1.5109708309173584,
      "learning_rate": 4.5318894171461636e-07,
      "loss": 0.452,
      "step": 395170
    },
    {
      "epoch": 161.76,
      "grad_norm": 1.871781587600708,
      "learning_rate": 4.5309518901601683e-07,
      "loss": 0.4295,
      "step": 395180
    },
    {
      "epoch": 161.76,
      "grad_norm": 2.1887786388397217,
      "learning_rate": 4.530014449782144e-07,
      "loss": 0.471,
      "step": 395190
    },
    {
      "epoch": 161.77,
      "grad_norm": 2.209754467010498,
      "learning_rate": 4.529077096016389e-07,
      "loss": 0.4444,
      "step": 395200
    },
    {
      "epoch": 161.77,
      "grad_norm": 2.166050672531128,
      "learning_rate": 4.5281398288671843e-07,
      "loss": 0.4528,
      "step": 395210
    },
    {
      "epoch": 161.78,
      "grad_norm": 2.556063652038574,
      "learning_rate": 4.52720264833883e-07,
      "loss": 0.4715,
      "step": 395220
    },
    {
      "epoch": 161.78,
      "grad_norm": 1.7751898765563965,
      "learning_rate": 4.526265554435618e-07,
      "loss": 0.4438,
      "step": 395230
    },
    {
      "epoch": 161.78,
      "grad_norm": 1.9533255100250244,
      "learning_rate": 4.5253285471618443e-07,
      "loss": 0.4446,
      "step": 395240
    },
    {
      "epoch": 161.79,
      "grad_norm": 1.612574815750122,
      "learning_rate": 4.524391626521799e-07,
      "loss": 0.4387,
      "step": 395250
    },
    {
      "epoch": 161.79,
      "grad_norm": 1.7536101341247559,
      "learning_rate": 4.523454792519771e-07,
      "loss": 0.453,
      "step": 395260
    },
    {
      "epoch": 161.8,
      "grad_norm": 2.126312732696533,
      "learning_rate": 4.522518045160059e-07,
      "loss": 0.4702,
      "step": 395270
    },
    {
      "epoch": 161.8,
      "grad_norm": 1.8238927125930786,
      "learning_rate": 4.521581384446948e-07,
      "loss": 0.4498,
      "step": 395280
    },
    {
      "epoch": 161.81,
      "grad_norm": 2.152367353439331,
      "learning_rate": 4.520644810384732e-07,
      "loss": 0.4377,
      "step": 395290
    },
    {
      "epoch": 161.81,
      "grad_norm": 2.244044303894043,
      "learning_rate": 4.5197083229776987e-07,
      "loss": 0.4243,
      "step": 395300
    },
    {
      "epoch": 161.81,
      "grad_norm": 1.83716881275177,
      "learning_rate": 4.518771922230137e-07,
      "loss": 0.428,
      "step": 395310
    },
    {
      "epoch": 161.82,
      "grad_norm": 1.8203436136245728,
      "learning_rate": 4.5178356081463424e-07,
      "loss": 0.4583,
      "step": 395320
    },
    {
      "epoch": 161.82,
      "grad_norm": 2.0199668407440186,
      "learning_rate": 4.516899380730598e-07,
      "loss": 0.4484,
      "step": 395330
    },
    {
      "epoch": 161.83,
      "grad_norm": 1.690163016319275,
      "learning_rate": 4.5159632399871976e-07,
      "loss": 0.4503,
      "step": 395340
    },
    {
      "epoch": 161.83,
      "grad_norm": 1.8636043071746826,
      "learning_rate": 4.515027185920424e-07,
      "loss": 0.4611,
      "step": 395350
    },
    {
      "epoch": 161.83,
      "grad_norm": 1.8516429662704468,
      "learning_rate": 4.514091218534568e-07,
      "loss": 0.4531,
      "step": 395360
    },
    {
      "epoch": 161.84,
      "grad_norm": 1.9423558712005615,
      "learning_rate": 4.513155337833918e-07,
      "loss": 0.4535,
      "step": 395370
    },
    {
      "epoch": 161.84,
      "grad_norm": 2.0181422233581543,
      "learning_rate": 4.5122195438227596e-07,
      "loss": 0.4291,
      "step": 395380
    },
    {
      "epoch": 161.85,
      "grad_norm": 2.115105628967285,
      "learning_rate": 4.511283836505381e-07,
      "loss": 0.4394,
      "step": 395390
    },
    {
      "epoch": 161.85,
      "grad_norm": 1.6851897239685059,
      "learning_rate": 4.5103482158860674e-07,
      "loss": 0.446,
      "step": 395400
    },
    {
      "epoch": 161.85,
      "grad_norm": 2.157146453857422,
      "learning_rate": 4.5094126819691047e-07,
      "loss": 0.4372,
      "step": 395410
    },
    {
      "epoch": 161.86,
      "grad_norm": 1.9519602060317993,
      "learning_rate": 4.5084772347587766e-07,
      "loss": 0.4552,
      "step": 395420
    },
    {
      "epoch": 161.86,
      "grad_norm": 2.0577993392944336,
      "learning_rate": 4.507541874259373e-07,
      "loss": 0.451,
      "step": 395430
    },
    {
      "epoch": 161.87,
      "grad_norm": 2.064459800720215,
      "learning_rate": 4.506606600475175e-07,
      "loss": 0.4529,
      "step": 395440
    },
    {
      "epoch": 161.87,
      "grad_norm": 1.8009454011917114,
      "learning_rate": 4.505671413410468e-07,
      "loss": 0.449,
      "step": 395450
    },
    {
      "epoch": 161.87,
      "grad_norm": 2.0083675384521484,
      "learning_rate": 4.504736313069535e-07,
      "loss": 0.4565,
      "step": 395460
    },
    {
      "epoch": 161.88,
      "grad_norm": 1.8121857643127441,
      "learning_rate": 4.5038012994566615e-07,
      "loss": 0.447,
      "step": 395470
    },
    {
      "epoch": 161.88,
      "grad_norm": 2.241095781326294,
      "learning_rate": 4.502866372576128e-07,
      "loss": 0.4685,
      "step": 395480
    },
    {
      "epoch": 161.89,
      "grad_norm": 2.4420244693756104,
      "learning_rate": 4.50193153243222e-07,
      "loss": 0.4337,
      "step": 395490
    },
    {
      "epoch": 161.89,
      "grad_norm": 3.031656503677368,
      "learning_rate": 4.5009967790292186e-07,
      "loss": 0.4253,
      "step": 395500
    },
    {
      "epoch": 161.9,
      "grad_norm": 1.9069575071334839,
      "learning_rate": 4.500062112371411e-07,
      "loss": 0.438,
      "step": 395510
    },
    {
      "epoch": 161.9,
      "grad_norm": 2.076815128326416,
      "learning_rate": 4.4991275324630656e-07,
      "loss": 0.4413,
      "step": 395520
    },
    {
      "epoch": 161.9,
      "grad_norm": 1.954428791999817,
      "learning_rate": 4.4981930393084675e-07,
      "loss": 0.4398,
      "step": 395530
    },
    {
      "epoch": 161.91,
      "grad_norm": 2.4126274585723877,
      "learning_rate": 4.497258632911908e-07,
      "loss": 0.4488,
      "step": 395540
    },
    {
      "epoch": 161.91,
      "grad_norm": 1.8077352046966553,
      "learning_rate": 4.4963243132776614e-07,
      "loss": 0.4476,
      "step": 395550
    },
    {
      "epoch": 161.92,
      "grad_norm": 2.3528668880462646,
      "learning_rate": 4.4953900804100067e-07,
      "loss": 0.4519,
      "step": 395560
    },
    {
      "epoch": 161.92,
      "grad_norm": 2.1725308895111084,
      "learning_rate": 4.494455934313223e-07,
      "loss": 0.4572,
      "step": 395570
    },
    {
      "epoch": 161.92,
      "grad_norm": 1.8520456552505493,
      "learning_rate": 4.493521874991592e-07,
      "loss": 0.4431,
      "step": 395580
    },
    {
      "epoch": 161.93,
      "grad_norm": 2.096071243286133,
      "learning_rate": 4.492587902449391e-07,
      "loss": 0.456,
      "step": 395590
    },
    {
      "epoch": 161.93,
      "grad_norm": 2.4125020503997803,
      "learning_rate": 4.491654016690898e-07,
      "loss": 0.4282,
      "step": 395600
    },
    {
      "epoch": 161.94,
      "grad_norm": 2.178006649017334,
      "learning_rate": 4.49072021772039e-07,
      "loss": 0.447,
      "step": 395610
    },
    {
      "epoch": 161.94,
      "grad_norm": 2.68048095703125,
      "learning_rate": 4.4897865055421457e-07,
      "loss": 0.4626,
      "step": 395620
    },
    {
      "epoch": 161.94,
      "grad_norm": 1.9430959224700928,
      "learning_rate": 4.488852880160443e-07,
      "loss": 0.4489,
      "step": 395630
    },
    {
      "epoch": 161.95,
      "grad_norm": 2.084322690963745,
      "learning_rate": 4.4879193415795646e-07,
      "loss": 0.4396,
      "step": 395640
    },
    {
      "epoch": 161.95,
      "grad_norm": 2.0632965564727783,
      "learning_rate": 4.4869858898037753e-07,
      "loss": 0.4455,
      "step": 395650
    },
    {
      "epoch": 161.96,
      "grad_norm": 1.99631929397583,
      "learning_rate": 4.486052524837355e-07,
      "loss": 0.4529,
      "step": 395660
    },
    {
      "epoch": 161.96,
      "grad_norm": 2.211751699447632,
      "learning_rate": 4.485119246684583e-07,
      "loss": 0.4359,
      "step": 395670
    },
    {
      "epoch": 161.96,
      "grad_norm": 1.7915602922439575,
      "learning_rate": 4.4841860553497284e-07,
      "loss": 0.429,
      "step": 395680
    },
    {
      "epoch": 161.97,
      "grad_norm": 1.8284705877304077,
      "learning_rate": 4.483252950837071e-07,
      "loss": 0.4604,
      "step": 395690
    },
    {
      "epoch": 161.97,
      "grad_norm": 2.3965659141540527,
      "learning_rate": 4.4823199331508844e-07,
      "loss": 0.4441,
      "step": 395700
    },
    {
      "epoch": 161.98,
      "grad_norm": 2.2534615993499756,
      "learning_rate": 4.481387002295436e-07,
      "loss": 0.4483,
      "step": 395710
    },
    {
      "epoch": 161.98,
      "grad_norm": 1.7761037349700928,
      "learning_rate": 4.4804541582750136e-07,
      "loss": 0.4607,
      "step": 395720
    },
    {
      "epoch": 161.99,
      "grad_norm": 2.1611146926879883,
      "learning_rate": 4.479521401093879e-07,
      "loss": 0.4505,
      "step": 395730
    },
    {
      "epoch": 161.99,
      "grad_norm": 2.190965414047241,
      "learning_rate": 4.478588730756311e-07,
      "loss": 0.4524,
      "step": 395740
    },
    {
      "epoch": 161.99,
      "grad_norm": 2.124985456466675,
      "learning_rate": 4.477656147266577e-07,
      "loss": 0.4443,
      "step": 395750
    },
    {
      "epoch": 162.0,
      "grad_norm": 2.1576247215270996,
      "learning_rate": 4.4767236506289515e-07,
      "loss": 0.4547,
      "step": 395760
    },
    {
      "epoch": 162.0,
      "eval_loss": 0.4475673735141754,
      "eval_runtime": 52.8591,
      "eval_samples_per_second": 65.249,
      "eval_steps_per_second": 8.173,
      "step": 395766
    },
    {
      "epoch": 162.0,
      "grad_norm": 2.147355794906616,
      "learning_rate": 4.4757912408477073e-07,
      "loss": 0.4359,
      "step": 395770
    },
    {
      "epoch": 162.01,
      "grad_norm": 1.9991462230682373,
      "learning_rate": 4.47485891792712e-07,
      "loss": 0.4434,
      "step": 395780
    },
    {
      "epoch": 162.01,
      "grad_norm": 1.9322912693023682,
      "learning_rate": 4.473926681871449e-07,
      "loss": 0.4661,
      "step": 395790
    },
    {
      "epoch": 162.01,
      "grad_norm": 2.2404723167419434,
      "learning_rate": 4.4729945326849723e-07,
      "loss": 0.4296,
      "step": 395800
    },
    {
      "epoch": 162.02,
      "grad_norm": 2.048659563064575,
      "learning_rate": 4.472062470371954e-07,
      "loss": 0.4392,
      "step": 395810
    },
    {
      "epoch": 162.02,
      "grad_norm": 2.092052459716797,
      "learning_rate": 4.4711304949366703e-07,
      "loss": 0.4203,
      "step": 395820
    },
    {
      "epoch": 162.03,
      "grad_norm": 2.2600271701812744,
      "learning_rate": 4.470198606383388e-07,
      "loss": 0.4489,
      "step": 395830
    },
    {
      "epoch": 162.03,
      "grad_norm": 2.0113182067871094,
      "learning_rate": 4.469266804716375e-07,
      "loss": 0.4435,
      "step": 395840
    },
    {
      "epoch": 162.03,
      "grad_norm": 1.8738532066345215,
      "learning_rate": 4.468335089939901e-07,
      "loss": 0.4518,
      "step": 395850
    },
    {
      "epoch": 162.04,
      "grad_norm": 1.871521234512329,
      "learning_rate": 4.467403462058236e-07,
      "loss": 0.4424,
      "step": 395860
    },
    {
      "epoch": 162.04,
      "grad_norm": 2.16129469871521,
      "learning_rate": 4.466471921075642e-07,
      "loss": 0.4423,
      "step": 395870
    },
    {
      "epoch": 162.05,
      "grad_norm": 1.8707811832427979,
      "learning_rate": 4.465540466996392e-07,
      "loss": 0.4266,
      "step": 395880
    },
    {
      "epoch": 162.05,
      "grad_norm": 2.2686269283294678,
      "learning_rate": 4.4646090998247484e-07,
      "loss": 0.4441,
      "step": 395890
    },
    {
      "epoch": 162.05,
      "grad_norm": 2.832990884780884,
      "learning_rate": 4.4636778195649755e-07,
      "loss": 0.4587,
      "step": 395900
    },
    {
      "epoch": 162.06,
      "grad_norm": 1.7635308504104614,
      "learning_rate": 4.4627466262213486e-07,
      "loss": 0.435,
      "step": 395910
    },
    {
      "epoch": 162.06,
      "grad_norm": 1.7627155780792236,
      "learning_rate": 4.461815519798133e-07,
      "loss": 0.4461,
      "step": 395920
    },
    {
      "epoch": 162.07,
      "grad_norm": 2.144350051879883,
      "learning_rate": 4.4608845002995853e-07,
      "loss": 0.4198,
      "step": 395930
    },
    {
      "epoch": 162.07,
      "grad_norm": 2.082303524017334,
      "learning_rate": 4.459953567729973e-07,
      "loss": 0.4272,
      "step": 395940
    },
    {
      "epoch": 162.08,
      "grad_norm": 2.1228787899017334,
      "learning_rate": 4.4590227220935634e-07,
      "loss": 0.4421,
      "step": 395950
    },
    {
      "epoch": 162.08,
      "grad_norm": 2.2508816719055176,
      "learning_rate": 4.458091963394616e-07,
      "loss": 0.4292,
      "step": 395960
    },
    {
      "epoch": 162.08,
      "grad_norm": 1.7627127170562744,
      "learning_rate": 4.457161291637402e-07,
      "loss": 0.438,
      "step": 395970
    },
    {
      "epoch": 162.09,
      "grad_norm": 2.25600004196167,
      "learning_rate": 4.4562307068261767e-07,
      "loss": 0.4452,
      "step": 395980
    },
    {
      "epoch": 162.09,
      "grad_norm": 2.38560152053833,
      "learning_rate": 4.455300208965208e-07,
      "loss": 0.4478,
      "step": 395990
    },
    {
      "epoch": 162.1,
      "grad_norm": 1.8379558324813843,
      "learning_rate": 4.4543697980587567e-07,
      "loss": 0.4524,
      "step": 396000
    },
    {
      "epoch": 162.1,
      "grad_norm": 1.7892482280731201,
      "learning_rate": 4.4534394741110835e-07,
      "loss": 0.4302,
      "step": 396010
    },
    {
      "epoch": 162.1,
      "grad_norm": 1.9947470426559448,
      "learning_rate": 4.4525092371264514e-07,
      "loss": 0.4492,
      "step": 396020
    },
    {
      "epoch": 162.11,
      "grad_norm": 1.9670782089233398,
      "learning_rate": 4.451579087109125e-07,
      "loss": 0.4634,
      "step": 396030
    },
    {
      "epoch": 162.11,
      "grad_norm": 1.962606430053711,
      "learning_rate": 4.450649024063359e-07,
      "loss": 0.4657,
      "step": 396040
    },
    {
      "epoch": 162.12,
      "grad_norm": 2.1033687591552734,
      "learning_rate": 4.4497190479934173e-07,
      "loss": 0.446,
      "step": 396050
    },
    {
      "epoch": 162.12,
      "grad_norm": 1.8703745603561401,
      "learning_rate": 4.4487891589035607e-07,
      "loss": 0.4333,
      "step": 396060
    },
    {
      "epoch": 162.12,
      "grad_norm": 1.9088467359542847,
      "learning_rate": 4.447859356798046e-07,
      "loss": 0.4493,
      "step": 396070
    },
    {
      "epoch": 162.13,
      "grad_norm": 2.1102042198181152,
      "learning_rate": 4.4469296416811343e-07,
      "loss": 0.4402,
      "step": 396080
    },
    {
      "epoch": 162.13,
      "grad_norm": 1.9354966878890991,
      "learning_rate": 4.446000013557087e-07,
      "loss": 0.4495,
      "step": 396090
    },
    {
      "epoch": 162.14,
      "grad_norm": 3.6939542293548584,
      "learning_rate": 4.445070472430157e-07,
      "loss": 0.4294,
      "step": 396100
    },
    {
      "epoch": 162.14,
      "grad_norm": 2.0019805431365967,
      "learning_rate": 4.444141018304607e-07,
      "loss": 0.4776,
      "step": 396110
    },
    {
      "epoch": 162.14,
      "grad_norm": 1.6133068799972534,
      "learning_rate": 4.4432116511846914e-07,
      "loss": 0.4575,
      "step": 396120
    },
    {
      "epoch": 162.15,
      "grad_norm": 2.2413575649261475,
      "learning_rate": 4.44228237107467e-07,
      "loss": 0.4289,
      "step": 396130
    },
    {
      "epoch": 162.15,
      "grad_norm": 1.8424861431121826,
      "learning_rate": 4.4413531779788e-07,
      "loss": 0.4471,
      "step": 396140
    },
    {
      "epoch": 162.16,
      "grad_norm": 1.9069336652755737,
      "learning_rate": 4.4404240719013353e-07,
      "loss": 0.4364,
      "step": 396150
    },
    {
      "epoch": 162.16,
      "grad_norm": 1.9565497636795044,
      "learning_rate": 4.439495052846533e-07,
      "loss": 0.4278,
      "step": 396160
    },
    {
      "epoch": 162.17,
      "grad_norm": 1.6596628427505493,
      "learning_rate": 4.4385661208186504e-07,
      "loss": 0.4575,
      "step": 396170
    },
    {
      "epoch": 162.17,
      "grad_norm": 1.8863388299942017,
      "learning_rate": 4.437637275821939e-07,
      "loss": 0.4317,
      "step": 396180
    },
    {
      "epoch": 162.17,
      "grad_norm": 2.309555768966675,
      "learning_rate": 4.4367085178606586e-07,
      "loss": 0.4341,
      "step": 396190
    },
    {
      "epoch": 162.18,
      "grad_norm": 2.6278016567230225,
      "learning_rate": 4.435779846939061e-07,
      "loss": 0.4508,
      "step": 396200
    },
    {
      "epoch": 162.18,
      "grad_norm": 1.9840288162231445,
      "learning_rate": 4.4348512630614006e-07,
      "loss": 0.438,
      "step": 396210
    },
    {
      "epoch": 162.19,
      "grad_norm": 2.018968343734741,
      "learning_rate": 4.4339227662319316e-07,
      "loss": 0.4483,
      "step": 396220
    },
    {
      "epoch": 162.19,
      "grad_norm": 1.909648060798645,
      "learning_rate": 4.432994356454911e-07,
      "loss": 0.4442,
      "step": 396230
    },
    {
      "epoch": 162.19,
      "grad_norm": 2.059896230697632,
      "learning_rate": 4.4320660337345833e-07,
      "loss": 0.4418,
      "step": 396240
    },
    {
      "epoch": 162.2,
      "grad_norm": 2.069906234741211,
      "learning_rate": 4.4311377980752055e-07,
      "loss": 0.4774,
      "step": 396250
    },
    {
      "epoch": 162.2,
      "grad_norm": 2.618441581726074,
      "learning_rate": 4.4302096494810283e-07,
      "loss": 0.4685,
      "step": 396260
    },
    {
      "epoch": 162.21,
      "grad_norm": 1.8746776580810547,
      "learning_rate": 4.429281587956305e-07,
      "loss": 0.4479,
      "step": 396270
    },
    {
      "epoch": 162.21,
      "grad_norm": 2.455723524093628,
      "learning_rate": 4.428353613505279e-07,
      "loss": 0.4639,
      "step": 396280
    },
    {
      "epoch": 162.21,
      "grad_norm": 2.0595996379852295,
      "learning_rate": 4.427425726132217e-07,
      "loss": 0.432,
      "step": 396290
    },
    {
      "epoch": 162.22,
      "grad_norm": 1.8747644424438477,
      "learning_rate": 4.426497925841359e-07,
      "loss": 0.4314,
      "step": 396300
    },
    {
      "epoch": 162.22,
      "grad_norm": 2.251340389251709,
      "learning_rate": 4.425570212636958e-07,
      "loss": 0.4648,
      "step": 396310
    },
    {
      "epoch": 162.23,
      "grad_norm": 2.352576494216919,
      "learning_rate": 4.424642586523265e-07,
      "loss": 0.4518,
      "step": 396320
    },
    {
      "epoch": 162.23,
      "grad_norm": 2.2574663162231445,
      "learning_rate": 4.423715047504525e-07,
      "loss": 0.4413,
      "step": 396330
    },
    {
      "epoch": 162.23,
      "grad_norm": 1.750423550605774,
      "learning_rate": 4.422787595584991e-07,
      "loss": 0.4376,
      "step": 396340
    },
    {
      "epoch": 162.24,
      "grad_norm": 2.168703079223633,
      "learning_rate": 4.4218602307689084e-07,
      "loss": 0.4588,
      "step": 396350
    },
    {
      "epoch": 162.24,
      "grad_norm": 2.134352684020996,
      "learning_rate": 4.420932953060533e-07,
      "loss": 0.4885,
      "step": 396360
    },
    {
      "epoch": 162.25,
      "grad_norm": 2.3030166625976562,
      "learning_rate": 4.4200057624640997e-07,
      "loss": 0.4461,
      "step": 396370
    },
    {
      "epoch": 162.25,
      "grad_norm": 2.008423089981079,
      "learning_rate": 4.4190786589838633e-07,
      "loss": 0.4432,
      "step": 396380
    },
    {
      "epoch": 162.26,
      "grad_norm": 2.0136232376098633,
      "learning_rate": 4.4181516426240684e-07,
      "loss": 0.4283,
      "step": 396390
    },
    {
      "epoch": 162.26,
      "grad_norm": 2.1375913619995117,
      "learning_rate": 4.417224713388964e-07,
      "loss": 0.4321,
      "step": 396400
    },
    {
      "epoch": 162.26,
      "grad_norm": 1.7840644121170044,
      "learning_rate": 4.416297871282793e-07,
      "loss": 0.4491,
      "step": 396410
    },
    {
      "epoch": 162.27,
      "grad_norm": 2.1432173252105713,
      "learning_rate": 4.4153711163098065e-07,
      "loss": 0.4444,
      "step": 396420
    },
    {
      "epoch": 162.27,
      "grad_norm": 1.9146888256072998,
      "learning_rate": 4.414444448474244e-07,
      "loss": 0.4396,
      "step": 396430
    },
    {
      "epoch": 162.28,
      "grad_norm": 1.9127964973449707,
      "learning_rate": 4.413517867780354e-07,
      "loss": 0.4362,
      "step": 396440
    },
    {
      "epoch": 162.28,
      "grad_norm": 2.0341155529022217,
      "learning_rate": 4.4125913742323787e-07,
      "loss": 0.4469,
      "step": 396450
    },
    {
      "epoch": 162.28,
      "grad_norm": 3.579467296600342,
      "learning_rate": 4.411664967834563e-07,
      "loss": 0.4488,
      "step": 396460
    },
    {
      "epoch": 162.29,
      "grad_norm": 1.8894169330596924,
      "learning_rate": 4.4107386485911455e-07,
      "loss": 0.4523,
      "step": 396470
    },
    {
      "epoch": 162.29,
      "grad_norm": 1.8497599363327026,
      "learning_rate": 4.409812416506384e-07,
      "loss": 0.4572,
      "step": 396480
    },
    {
      "epoch": 162.3,
      "grad_norm": 1.918816328048706,
      "learning_rate": 4.4088862715845103e-07,
      "loss": 0.4479,
      "step": 396490
    },
    {
      "epoch": 162.3,
      "grad_norm": 2.19171142578125,
      "learning_rate": 4.407960213829775e-07,
      "loss": 0.449,
      "step": 396500
    },
    {
      "epoch": 162.3,
      "grad_norm": 1.872593641281128,
      "learning_rate": 4.407034243246407e-07,
      "loss": 0.4483,
      "step": 396510
    },
    {
      "epoch": 162.31,
      "grad_norm": 2.3236587047576904,
      "learning_rate": 4.4061083598386565e-07,
      "loss": 0.4257,
      "step": 396520
    },
    {
      "epoch": 162.31,
      "grad_norm": 1.930519938468933,
      "learning_rate": 4.405182563610766e-07,
      "loss": 0.4536,
      "step": 396530
    },
    {
      "epoch": 162.32,
      "grad_norm": 2.2228283882141113,
      "learning_rate": 4.4042568545669713e-07,
      "loss": 0.4418,
      "step": 396540
    },
    {
      "epoch": 162.32,
      "grad_norm": 1.662177324295044,
      "learning_rate": 4.4033312327115163e-07,
      "loss": 0.4469,
      "step": 396550
    },
    {
      "epoch": 162.33,
      "grad_norm": 2.2215614318847656,
      "learning_rate": 4.402405698048643e-07,
      "loss": 0.454,
      "step": 396560
    },
    {
      "epoch": 162.33,
      "grad_norm": 2.443206548690796,
      "learning_rate": 4.401480250582587e-07,
      "loss": 0.4521,
      "step": 396570
    },
    {
      "epoch": 162.33,
      "grad_norm": 2.0987143516540527,
      "learning_rate": 4.400554890317587e-07,
      "loss": 0.4516,
      "step": 396580
    },
    {
      "epoch": 162.34,
      "grad_norm": 1.9357049465179443,
      "learning_rate": 4.3996296172578873e-07,
      "loss": 0.4441,
      "step": 396590
    },
    {
      "epoch": 162.34,
      "grad_norm": 2.233314275741577,
      "learning_rate": 4.3987044314077216e-07,
      "loss": 0.4509,
      "step": 396600
    },
    {
      "epoch": 162.35,
      "grad_norm": 1.9297139644622803,
      "learning_rate": 4.397779332771331e-07,
      "loss": 0.4301,
      "step": 396610
    },
    {
      "epoch": 162.35,
      "grad_norm": 1.9957679510116577,
      "learning_rate": 4.396854321352952e-07,
      "loss": 0.4451,
      "step": 396620
    },
    {
      "epoch": 162.35,
      "grad_norm": 2.9651312828063965,
      "learning_rate": 4.3959293971568233e-07,
      "loss": 0.4524,
      "step": 396630
    },
    {
      "epoch": 162.36,
      "grad_norm": 2.097283124923706,
      "learning_rate": 4.3950045601871784e-07,
      "loss": 0.4715,
      "step": 396640
    },
    {
      "epoch": 162.36,
      "grad_norm": 3.312544822692871,
      "learning_rate": 4.3940798104482564e-07,
      "loss": 0.4502,
      "step": 396650
    },
    {
      "epoch": 162.37,
      "grad_norm": 2.174514055252075,
      "learning_rate": 4.3931551479442926e-07,
      "loss": 0.4454,
      "step": 396660
    },
    {
      "epoch": 162.37,
      "grad_norm": 2.0309972763061523,
      "learning_rate": 4.392230572679525e-07,
      "loss": 0.439,
      "step": 396670
    },
    {
      "epoch": 162.37,
      "grad_norm": 1.8557065725326538,
      "learning_rate": 4.3913060846581875e-07,
      "loss": 0.4693,
      "step": 396680
    },
    {
      "epoch": 162.38,
      "grad_norm": 2.121816635131836,
      "learning_rate": 4.3903816838845123e-07,
      "loss": 0.4297,
      "step": 396690
    },
    {
      "epoch": 162.38,
      "grad_norm": 1.8422720432281494,
      "learning_rate": 4.3894573703627386e-07,
      "loss": 0.4393,
      "step": 396700
    },
    {
      "epoch": 162.39,
      "grad_norm": 1.89821195602417,
      "learning_rate": 4.3885331440970967e-07,
      "loss": 0.4372,
      "step": 396710
    },
    {
      "epoch": 162.39,
      "grad_norm": 2.6215929985046387,
      "learning_rate": 4.38760900509182e-07,
      "loss": 0.4458,
      "step": 396720
    },
    {
      "epoch": 162.39,
      "grad_norm": 2.254023790359497,
      "learning_rate": 4.3866849533511486e-07,
      "loss": 0.4302,
      "step": 396730
    },
    {
      "epoch": 162.4,
      "grad_norm": 2.0864460468292236,
      "learning_rate": 4.385760988879307e-07,
      "loss": 0.4393,
      "step": 396740
    },
    {
      "epoch": 162.4,
      "grad_norm": 2.0591530799865723,
      "learning_rate": 4.3848371116805314e-07,
      "loss": 0.4534,
      "step": 396750
    },
    {
      "epoch": 162.41,
      "grad_norm": 2.538499116897583,
      "learning_rate": 4.3839133217590535e-07,
      "loss": 0.4609,
      "step": 396760
    },
    {
      "epoch": 162.41,
      "grad_norm": 1.9106900691986084,
      "learning_rate": 4.382989619119107e-07,
      "loss": 0.4427,
      "step": 396770
    },
    {
      "epoch": 162.42,
      "grad_norm": 2.00901198387146,
      "learning_rate": 4.382066003764922e-07,
      "loss": 0.4638,
      "step": 396780
    },
    {
      "epoch": 162.42,
      "grad_norm": 1.9542471170425415,
      "learning_rate": 4.381142475700727e-07,
      "loss": 0.4439,
      "step": 396790
    },
    {
      "epoch": 162.42,
      "grad_norm": 2.87211537361145,
      "learning_rate": 4.3802190349307564e-07,
      "loss": 0.4327,
      "step": 396800
    },
    {
      "epoch": 162.43,
      "grad_norm": 2.181828737258911,
      "learning_rate": 4.3792956814592427e-07,
      "loss": 0.4376,
      "step": 396810
    },
    {
      "epoch": 162.43,
      "grad_norm": 1.9621392488479614,
      "learning_rate": 4.378372415290408e-07,
      "loss": 0.4629,
      "step": 396820
    },
    {
      "epoch": 162.44,
      "grad_norm": 1.930749773979187,
      "learning_rate": 4.3774492364284816e-07,
      "loss": 0.4466,
      "step": 396830
    },
    {
      "epoch": 162.44,
      "grad_norm": 1.7831283807754517,
      "learning_rate": 4.3765261448776986e-07,
      "loss": 0.4482,
      "step": 396840
    },
    {
      "epoch": 162.44,
      "grad_norm": 2.278252601623535,
      "learning_rate": 4.3756031406422784e-07,
      "loss": 0.445,
      "step": 396850
    },
    {
      "epoch": 162.45,
      "grad_norm": 2.0637662410736084,
      "learning_rate": 4.3746802237264604e-07,
      "loss": 0.4358,
      "step": 396860
    },
    {
      "epoch": 162.45,
      "grad_norm": 2.4598042964935303,
      "learning_rate": 4.37375739413447e-07,
      "loss": 0.446,
      "step": 396870
    },
    {
      "epoch": 162.46,
      "grad_norm": 1.9209990501403809,
      "learning_rate": 4.3728346518705286e-07,
      "loss": 0.4386,
      "step": 396880
    },
    {
      "epoch": 162.46,
      "grad_norm": 2.0542826652526855,
      "learning_rate": 4.3719119969388687e-07,
      "loss": 0.4481,
      "step": 396890
    },
    {
      "epoch": 162.46,
      "grad_norm": 2.090914011001587,
      "learning_rate": 4.3709894293437146e-07,
      "loss": 0.4559,
      "step": 396900
    },
    {
      "epoch": 162.47,
      "grad_norm": 1.8939483165740967,
      "learning_rate": 4.3700669490892934e-07,
      "loss": 0.4476,
      "step": 396910
    },
    {
      "epoch": 162.47,
      "grad_norm": 2.3547170162200928,
      "learning_rate": 4.3691445561798293e-07,
      "loss": 0.4468,
      "step": 396920
    },
    {
      "epoch": 162.48,
      "grad_norm": 2.1239733695983887,
      "learning_rate": 4.368222250619548e-07,
      "loss": 0.4768,
      "step": 396930
    },
    {
      "epoch": 162.48,
      "grad_norm": 2.02998685836792,
      "learning_rate": 4.3673000324126747e-07,
      "loss": 0.453,
      "step": 396940
    },
    {
      "epoch": 162.48,
      "grad_norm": 1.9330699443817139,
      "learning_rate": 4.366377901563441e-07,
      "loss": 0.4474,
      "step": 396950
    },
    {
      "epoch": 162.49,
      "grad_norm": 1.8306630849838257,
      "learning_rate": 4.365455858076057e-07,
      "loss": 0.442,
      "step": 396960
    },
    {
      "epoch": 162.49,
      "grad_norm": 1.9706140756607056,
      "learning_rate": 4.3645339019547555e-07,
      "loss": 0.4658,
      "step": 396970
    },
    {
      "epoch": 162.5,
      "grad_norm": 2.119743824005127,
      "learning_rate": 4.3636120332037554e-07,
      "loss": 0.4463,
      "step": 396980
    },
    {
      "epoch": 162.5,
      "grad_norm": 1.7495864629745483,
      "learning_rate": 4.362690251827285e-07,
      "loss": 0.4483,
      "step": 396990
    },
    {
      "epoch": 162.51,
      "grad_norm": 1.935522198677063,
      "learning_rate": 4.361768557829562e-07,
      "loss": 0.4452,
      "step": 397000
    },
    {
      "epoch": 162.51,
      "grad_norm": 2.7358925342559814,
      "learning_rate": 4.3608469512148094e-07,
      "loss": 0.4351,
      "step": 397010
    },
    {
      "epoch": 162.51,
      "grad_norm": 2.1897711753845215,
      "learning_rate": 4.359925431987251e-07,
      "loss": 0.4634,
      "step": 397020
    },
    {
      "epoch": 162.52,
      "grad_norm": 2.125457763671875,
      "learning_rate": 4.3590040001511017e-07,
      "loss": 0.4598,
      "step": 397030
    },
    {
      "epoch": 162.52,
      "grad_norm": 2.3241195678710938,
      "learning_rate": 4.358082655710595e-07,
      "loss": 0.4439,
      "step": 397040
    },
    {
      "epoch": 162.53,
      "grad_norm": 2.3238308429718018,
      "learning_rate": 4.3571613986699414e-07,
      "loss": 0.4435,
      "step": 397050
    },
    {
      "epoch": 162.53,
      "grad_norm": 1.656482219696045,
      "learning_rate": 4.3562402290333667e-07,
      "loss": 0.4364,
      "step": 397060
    },
    {
      "epoch": 162.53,
      "grad_norm": 2.000776529312134,
      "learning_rate": 4.355319146805086e-07,
      "loss": 0.4382,
      "step": 397070
    },
    {
      "epoch": 162.54,
      "grad_norm": 1.781844973564148,
      "learning_rate": 4.354398151989328e-07,
      "loss": 0.4358,
      "step": 397080
    },
    {
      "epoch": 162.54,
      "grad_norm": 2.163431406021118,
      "learning_rate": 4.353477244590297e-07,
      "loss": 0.4481,
      "step": 397090
    },
    {
      "epoch": 162.55,
      "grad_norm": 2.2696309089660645,
      "learning_rate": 4.3525564246122195e-07,
      "loss": 0.463,
      "step": 397100
    },
    {
      "epoch": 162.55,
      "grad_norm": 2.7785956859588623,
      "learning_rate": 4.3516356920593135e-07,
      "loss": 0.4559,
      "step": 397110
    },
    {
      "epoch": 162.55,
      "grad_norm": 1.9288122653961182,
      "learning_rate": 4.350715046935794e-07,
      "loss": 0.4574,
      "step": 397120
    },
    {
      "epoch": 162.56,
      "grad_norm": 2.3475606441497803,
      "learning_rate": 4.349794489245882e-07,
      "loss": 0.4378,
      "step": 397130
    },
    {
      "epoch": 162.56,
      "grad_norm": 2.042933225631714,
      "learning_rate": 4.348874018993792e-07,
      "loss": 0.4515,
      "step": 397140
    },
    {
      "epoch": 162.57,
      "grad_norm": 2.1187047958374023,
      "learning_rate": 4.3479536361837404e-07,
      "loss": 0.4533,
      "step": 397150
    },
    {
      "epoch": 162.57,
      "grad_norm": 2.4842522144317627,
      "learning_rate": 4.3470333408199477e-07,
      "loss": 0.4325,
      "step": 397160
    },
    {
      "epoch": 162.57,
      "grad_norm": 2.016399383544922,
      "learning_rate": 4.346113132906623e-07,
      "loss": 0.4365,
      "step": 397170
    },
    {
      "epoch": 162.58,
      "grad_norm": 2.177014112472534,
      "learning_rate": 4.3451930124479856e-07,
      "loss": 0.4459,
      "step": 397180
    },
    {
      "epoch": 162.58,
      "grad_norm": 1.9202749729156494,
      "learning_rate": 4.34427297944825e-07,
      "loss": 0.4349,
      "step": 397190
    },
    {
      "epoch": 162.59,
      "grad_norm": 1.9547847509384155,
      "learning_rate": 4.343353033911632e-07,
      "loss": 0.4327,
      "step": 397200
    },
    {
      "epoch": 162.59,
      "grad_norm": 2.1269216537475586,
      "learning_rate": 4.3424331758423416e-07,
      "loss": 0.4469,
      "step": 397210
    },
    {
      "epoch": 162.6,
      "grad_norm": 3.367029905319214,
      "learning_rate": 4.341513405244597e-07,
      "loss": 0.4183,
      "step": 397220
    },
    {
      "epoch": 162.6,
      "grad_norm": 2.258854389190674,
      "learning_rate": 4.340593722122608e-07,
      "loss": 0.4451,
      "step": 397230
    },
    {
      "epoch": 162.6,
      "grad_norm": 2.111178159713745,
      "learning_rate": 4.3396741264805903e-07,
      "loss": 0.456,
      "step": 397240
    },
    {
      "epoch": 162.61,
      "grad_norm": 2.3570690155029297,
      "learning_rate": 4.338754618322754e-07,
      "loss": 0.4481,
      "step": 397250
    },
    {
      "epoch": 162.61,
      "grad_norm": 2.112368106842041,
      "learning_rate": 4.3378351976533145e-07,
      "loss": 0.4356,
      "step": 397260
    },
    {
      "epoch": 162.62,
      "grad_norm": 1.849556803703308,
      "learning_rate": 4.336915864476479e-07,
      "loss": 0.4361,
      "step": 397270
    },
    {
      "epoch": 162.62,
      "grad_norm": 1.9960345029830933,
      "learning_rate": 4.335996618796463e-07,
      "loss": 0.4405,
      "step": 397280
    },
    {
      "epoch": 162.62,
      "grad_norm": 2.3652219772338867,
      "learning_rate": 4.3350774606174775e-07,
      "loss": 0.4606,
      "step": 397290
    },
    {
      "epoch": 162.63,
      "grad_norm": 2.46950364112854,
      "learning_rate": 4.334158389943729e-07,
      "loss": 0.4448,
      "step": 397300
    },
    {
      "epoch": 162.63,
      "grad_norm": 2.294004440307617,
      "learning_rate": 4.3332394067794285e-07,
      "loss": 0.4465,
      "step": 397310
    },
    {
      "epoch": 162.64,
      "grad_norm": 2.3043551445007324,
      "learning_rate": 4.332320511128788e-07,
      "loss": 0.4425,
      "step": 397320
    },
    {
      "epoch": 162.64,
      "grad_norm": 1.9871912002563477,
      "learning_rate": 4.331401702996018e-07,
      "loss": 0.4568,
      "step": 397330
    },
    {
      "epoch": 162.64,
      "grad_norm": 1.9536722898483276,
      "learning_rate": 4.3304829823853236e-07,
      "loss": 0.4513,
      "step": 397340
    },
    {
      "epoch": 162.65,
      "grad_norm": 1.7854653596878052,
      "learning_rate": 4.3295643493009153e-07,
      "loss": 0.4448,
      "step": 397350
    },
    {
      "epoch": 162.65,
      "grad_norm": 2.3319716453552246,
      "learning_rate": 4.3286458037470023e-07,
      "loss": 0.4538,
      "step": 397360
    },
    {
      "epoch": 162.66,
      "grad_norm": 2.0739622116088867,
      "learning_rate": 4.327727345727788e-07,
      "loss": 0.4714,
      "step": 397370
    },
    {
      "epoch": 162.66,
      "grad_norm": 2.0952136516571045,
      "learning_rate": 4.3268089752474846e-07,
      "loss": 0.4324,
      "step": 397380
    },
    {
      "epoch": 162.66,
      "grad_norm": 1.9458085298538208,
      "learning_rate": 4.325890692310305e-07,
      "loss": 0.4597,
      "step": 397390
    },
    {
      "epoch": 162.67,
      "grad_norm": 2.3596689701080322,
      "learning_rate": 4.324972496920438e-07,
      "loss": 0.458,
      "step": 397400
    },
    {
      "epoch": 162.67,
      "grad_norm": 2.3699893951416016,
      "learning_rate": 4.324054389082095e-07,
      "loss": 0.442,
      "step": 397410
    },
    {
      "epoch": 162.68,
      "grad_norm": 2.2206082344055176,
      "learning_rate": 4.3231363687994933e-07,
      "loss": 0.4389,
      "step": 397420
    },
    {
      "epoch": 162.68,
      "grad_norm": 1.9075438976287842,
      "learning_rate": 4.3222184360768325e-07,
      "loss": 0.4476,
      "step": 397430
    },
    {
      "epoch": 162.69,
      "grad_norm": 1.8794171810150146,
      "learning_rate": 4.3213005909183123e-07,
      "loss": 0.427,
      "step": 397440
    },
    {
      "epoch": 162.69,
      "grad_norm": 2.3819072246551514,
      "learning_rate": 4.3203828333281435e-07,
      "loss": 0.4625,
      "step": 397450
    },
    {
      "epoch": 162.69,
      "grad_norm": 1.8538918495178223,
      "learning_rate": 4.319465163310528e-07,
      "loss": 0.4774,
      "step": 397460
    },
    {
      "epoch": 162.7,
      "grad_norm": 1.9345041513442993,
      "learning_rate": 4.3185475808696677e-07,
      "loss": 0.4542,
      "step": 397470
    },
    {
      "epoch": 162.7,
      "grad_norm": 1.8790652751922607,
      "learning_rate": 4.3176300860097687e-07,
      "loss": 0.4445,
      "step": 397480
    },
    {
      "epoch": 162.71,
      "grad_norm": 1.684876799583435,
      "learning_rate": 4.3167126787350325e-07,
      "loss": 0.4457,
      "step": 397490
    },
    {
      "epoch": 162.71,
      "grad_norm": 1.7781790494918823,
      "learning_rate": 4.315795359049662e-07,
      "loss": 0.422,
      "step": 397500
    },
    {
      "epoch": 162.71,
      "grad_norm": 1.892969012260437,
      "learning_rate": 4.314878126957859e-07,
      "loss": 0.4201,
      "step": 397510
    },
    {
      "epoch": 162.72,
      "grad_norm": 2.2230224609375,
      "learning_rate": 4.313960982463824e-07,
      "loss": 0.4378,
      "step": 397520
    },
    {
      "epoch": 162.72,
      "grad_norm": 2.0844500064849854,
      "learning_rate": 4.3130439255717663e-07,
      "loss": 0.4648,
      "step": 397530
    },
    {
      "epoch": 162.73,
      "grad_norm": 2.2244625091552734,
      "learning_rate": 4.312126956285876e-07,
      "loss": 0.4573,
      "step": 397540
    },
    {
      "epoch": 162.73,
      "grad_norm": 2.1761491298675537,
      "learning_rate": 4.311210074610355e-07,
      "loss": 0.4403,
      "step": 397550
    },
    {
      "epoch": 162.73,
      "grad_norm": 2.3714187145233154,
      "learning_rate": 4.310293280549406e-07,
      "loss": 0.4597,
      "step": 397560
    },
    {
      "epoch": 162.74,
      "grad_norm": 2.007463216781616,
      "learning_rate": 4.3093765741072314e-07,
      "loss": 0.459,
      "step": 397570
    },
    {
      "epoch": 162.74,
      "grad_norm": 2.0662641525268555,
      "learning_rate": 4.308459955288025e-07,
      "loss": 0.4527,
      "step": 397580
    },
    {
      "epoch": 162.75,
      "grad_norm": 3.3404171466827393,
      "learning_rate": 4.30754342409599e-07,
      "loss": 0.4374,
      "step": 397590
    },
    {
      "epoch": 162.75,
      "grad_norm": 1.6982531547546387,
      "learning_rate": 4.3066269805353186e-07,
      "loss": 0.4329,
      "step": 397600
    },
    {
      "epoch": 162.75,
      "grad_norm": 2.258242130279541,
      "learning_rate": 4.3057106246102173e-07,
      "loss": 0.4452,
      "step": 397610
    },
    {
      "epoch": 162.76,
      "grad_norm": 1.893097162246704,
      "learning_rate": 4.304794356324881e-07,
      "loss": 0.4419,
      "step": 397620
    },
    {
      "epoch": 162.76,
      "grad_norm": 2.2121827602386475,
      "learning_rate": 4.303878175683508e-07,
      "loss": 0.4649,
      "step": 397630
    },
    {
      "epoch": 162.77,
      "grad_norm": 2.0201497077941895,
      "learning_rate": 4.3029620826902914e-07,
      "loss": 0.4183,
      "step": 397640
    },
    {
      "epoch": 162.77,
      "grad_norm": 2.002638101577759,
      "learning_rate": 4.302046077349429e-07,
      "loss": 0.455,
      "step": 397650
    },
    {
      "epoch": 162.78,
      "grad_norm": 2.3778557777404785,
      "learning_rate": 4.3011301596651257e-07,
      "loss": 0.4422,
      "step": 397660
    },
    {
      "epoch": 162.78,
      "grad_norm": 1.8929824829101562,
      "learning_rate": 4.300214329641563e-07,
      "loss": 0.457,
      "step": 397670
    },
    {
      "epoch": 162.78,
      "grad_norm": 2.164433240890503,
      "learning_rate": 4.2992985872829407e-07,
      "loss": 0.4549,
      "step": 397680
    },
    {
      "epoch": 162.79,
      "grad_norm": 2.0758535861968994,
      "learning_rate": 4.298382932593456e-07,
      "loss": 0.4512,
      "step": 397690
    },
    {
      "epoch": 162.79,
      "grad_norm": 2.2861571311950684,
      "learning_rate": 4.297467365577301e-07,
      "loss": 0.4391,
      "step": 397700
    },
    {
      "epoch": 162.8,
      "grad_norm": 1.626206874847412,
      "learning_rate": 4.2965518862386725e-07,
      "loss": 0.4291,
      "step": 397710
    },
    {
      "epoch": 162.8,
      "grad_norm": 2.044970750808716,
      "learning_rate": 4.2956364945817626e-07,
      "loss": 0.4516,
      "step": 397720
    },
    {
      "epoch": 162.8,
      "grad_norm": 2.3942654132843018,
      "learning_rate": 4.294721190610766e-07,
      "loss": 0.4324,
      "step": 397730
    },
    {
      "epoch": 162.81,
      "grad_norm": 2.3660728931427,
      "learning_rate": 4.293805974329872e-07,
      "loss": 0.4379,
      "step": 397740
    },
    {
      "epoch": 162.81,
      "grad_norm": 2.357020854949951,
      "learning_rate": 4.292890845743278e-07,
      "loss": 0.4443,
      "step": 397750
    },
    {
      "epoch": 162.82,
      "grad_norm": 2.9176886081695557,
      "learning_rate": 4.2919758048551724e-07,
      "loss": 0.4484,
      "step": 397760
    },
    {
      "epoch": 162.82,
      "grad_norm": 1.9452933073043823,
      "learning_rate": 4.2910608516697486e-07,
      "loss": 0.4645,
      "step": 397770
    },
    {
      "epoch": 162.82,
      "grad_norm": 2.110483169555664,
      "learning_rate": 4.2901459861911973e-07,
      "loss": 0.454,
      "step": 397780
    },
    {
      "epoch": 162.83,
      "grad_norm": 2.0688140392303467,
      "learning_rate": 4.2892312084237034e-07,
      "loss": 0.4455,
      "step": 397790
    },
    {
      "epoch": 162.83,
      "grad_norm": 2.2923171520233154,
      "learning_rate": 4.288316518371474e-07,
      "loss": 0.4708,
      "step": 397800
    },
    {
      "epoch": 162.84,
      "grad_norm": 2.330595016479492,
      "learning_rate": 4.287401916038686e-07,
      "loss": 0.4162,
      "step": 397810
    },
    {
      "epoch": 162.84,
      "grad_norm": 2.2380318641662598,
      "learning_rate": 4.2864874014295285e-07,
      "loss": 0.4417,
      "step": 397820
    },
    {
      "epoch": 162.84,
      "grad_norm": 1.8092390298843384,
      "learning_rate": 4.285572974548196e-07,
      "loss": 0.432,
      "step": 397830
    },
    {
      "epoch": 162.85,
      "grad_norm": 1.8495569229125977,
      "learning_rate": 4.284658635398872e-07,
      "loss": 0.4423,
      "step": 397840
    },
    {
      "epoch": 162.85,
      "grad_norm": 1.8976415395736694,
      "learning_rate": 4.28374438398575e-07,
      "loss": 0.4502,
      "step": 397850
    },
    {
      "epoch": 162.86,
      "grad_norm": 1.8142075538635254,
      "learning_rate": 4.2828302203130154e-07,
      "loss": 0.4392,
      "step": 397860
    },
    {
      "epoch": 162.86,
      "grad_norm": 2.1693077087402344,
      "learning_rate": 4.281916144384858e-07,
      "loss": 0.422,
      "step": 397870
    },
    {
      "epoch": 162.87,
      "grad_norm": 2.236057996749878,
      "learning_rate": 4.281002156205462e-07,
      "loss": 0.4332,
      "step": 397880
    },
    {
      "epoch": 162.87,
      "grad_norm": 1.878570795059204,
      "learning_rate": 4.280088255779017e-07,
      "loss": 0.4496,
      "step": 397890
    },
    {
      "epoch": 162.87,
      "grad_norm": 1.8708292245864868,
      "learning_rate": 4.279174443109709e-07,
      "loss": 0.4313,
      "step": 397900
    },
    {
      "epoch": 162.88,
      "grad_norm": 1.726361870765686,
      "learning_rate": 4.2782607182017234e-07,
      "loss": 0.4373,
      "step": 397910
    },
    {
      "epoch": 162.88,
      "grad_norm": 2.176767349243164,
      "learning_rate": 4.277347081059243e-07,
      "loss": 0.4363,
      "step": 397920
    },
    {
      "epoch": 162.89,
      "grad_norm": 1.9989129304885864,
      "learning_rate": 4.276433531686458e-07,
      "loss": 0.4496,
      "step": 397930
    },
    {
      "epoch": 162.89,
      "grad_norm": 2.0912251472473145,
      "learning_rate": 4.2755200700875495e-07,
      "loss": 0.4399,
      "step": 397940
    },
    {
      "epoch": 162.89,
      "grad_norm": 1.667041301727295,
      "learning_rate": 4.274606696266705e-07,
      "loss": 0.4596,
      "step": 397950
    },
    {
      "epoch": 162.9,
      "grad_norm": 2.006812572479248,
      "learning_rate": 4.2736934102281074e-07,
      "loss": 0.4426,
      "step": 397960
    },
    {
      "epoch": 162.9,
      "grad_norm": 2.0810084342956543,
      "learning_rate": 4.2727802119759444e-07,
      "loss": 0.4455,
      "step": 397970
    },
    {
      "epoch": 162.91,
      "grad_norm": 2.4339394569396973,
      "learning_rate": 4.2718671015143846e-07,
      "loss": 0.4256,
      "step": 397980
    },
    {
      "epoch": 162.91,
      "grad_norm": 2.5279159545898438,
      "learning_rate": 4.2709540788476246e-07,
      "loss": 0.4425,
      "step": 397990
    },
    {
      "epoch": 162.91,
      "grad_norm": 2.6543948650360107,
      "learning_rate": 4.2700411439798435e-07,
      "loss": 0.4454,
      "step": 398000
    },
    {
      "epoch": 162.92,
      "grad_norm": 2.0798022747039795,
      "learning_rate": 4.2691282969152235e-07,
      "loss": 0.4472,
      "step": 398010
    },
    {
      "epoch": 162.92,
      "grad_norm": 1.9326845407485962,
      "learning_rate": 4.268215537657948e-07,
      "loss": 0.4363,
      "step": 398020
    },
    {
      "epoch": 162.93,
      "grad_norm": 1.981114149093628,
      "learning_rate": 4.2673028662121935e-07,
      "loss": 0.4608,
      "step": 398030
    },
    {
      "epoch": 162.93,
      "grad_norm": 1.7953628301620483,
      "learning_rate": 4.2663902825821416e-07,
      "loss": 0.4348,
      "step": 398040
    },
    {
      "epoch": 162.93,
      "grad_norm": 2.6164891719818115,
      "learning_rate": 4.2654777867719766e-07,
      "loss": 0.4604,
      "step": 398050
    },
    {
      "epoch": 162.94,
      "grad_norm": 2.896979808807373,
      "learning_rate": 4.264565378785877e-07,
      "loss": 0.4674,
      "step": 398060
    },
    {
      "epoch": 162.94,
      "grad_norm": 2.143770694732666,
      "learning_rate": 4.26365305862802e-07,
      "loss": 0.453,
      "step": 398070
    },
    {
      "epoch": 162.95,
      "grad_norm": 1.9861713647842407,
      "learning_rate": 4.262740826302586e-07,
      "loss": 0.4464,
      "step": 398080
    },
    {
      "epoch": 162.95,
      "grad_norm": 2.423935651779175,
      "learning_rate": 4.261828681813755e-07,
      "loss": 0.4438,
      "step": 398090
    },
    {
      "epoch": 162.96,
      "grad_norm": 2.1297736167907715,
      "learning_rate": 4.2609166251657053e-07,
      "loss": 0.4332,
      "step": 398100
    },
    {
      "epoch": 162.96,
      "grad_norm": 1.9652247428894043,
      "learning_rate": 4.260004656362619e-07,
      "loss": 0.4453,
      "step": 398110
    },
    {
      "epoch": 162.96,
      "grad_norm": 1.938802719116211,
      "learning_rate": 4.259092775408664e-07,
      "loss": 0.4374,
      "step": 398120
    },
    {
      "epoch": 162.97,
      "grad_norm": 2.9682724475860596,
      "learning_rate": 4.2581809823080194e-07,
      "loss": 0.4489,
      "step": 398130
    },
    {
      "epoch": 162.97,
      "grad_norm": 1.8832635879516602,
      "learning_rate": 4.2572692770648674e-07,
      "loss": 0.4406,
      "step": 398140
    },
    {
      "epoch": 162.98,
      "grad_norm": 2.0181400775909424,
      "learning_rate": 4.2563576596833837e-07,
      "loss": 0.4447,
      "step": 398150
    },
    {
      "epoch": 162.98,
      "grad_norm": 1.9381537437438965,
      "learning_rate": 4.2554461301677394e-07,
      "loss": 0.454,
      "step": 398160
    },
    {
      "epoch": 162.98,
      "grad_norm": 2.408614158630371,
      "learning_rate": 4.254534688522109e-07,
      "loss": 0.4544,
      "step": 398170
    },
    {
      "epoch": 162.99,
      "grad_norm": 1.7069170475006104,
      "learning_rate": 4.253623334750675e-07,
      "loss": 0.4287,
      "step": 398180
    },
    {
      "epoch": 162.99,
      "grad_norm": 1.9819811582565308,
      "learning_rate": 4.2527120688576107e-07,
      "loss": 0.4361,
      "step": 398190
    },
    {
      "epoch": 163.0,
      "grad_norm": 1.9217413663864136,
      "learning_rate": 4.251800890847091e-07,
      "loss": 0.4426,
      "step": 398200
    },
    {
      "epoch": 163.0,
      "eval_loss": 0.44562989473342896,
      "eval_runtime": 52.4145,
      "eval_samples_per_second": 65.802,
      "eval_steps_per_second": 8.242,
      "step": 398209
    },
    {
      "epoch": 163.0,
      "grad_norm": 1.897790789604187,
      "learning_rate": 4.2508898007232847e-07,
      "loss": 0.4554,
      "step": 398210
    },
    {
      "epoch": 163.0,
      "grad_norm": 1.8416985273361206,
      "learning_rate": 4.2499787984903686e-07,
      "loss": 0.4528,
      "step": 398220
    },
    {
      "epoch": 163.01,
      "grad_norm": 2.1459169387817383,
      "learning_rate": 4.249067884152516e-07,
      "loss": 0.4399,
      "step": 398230
    },
    {
      "epoch": 163.01,
      "grad_norm": 2.0755419731140137,
      "learning_rate": 4.248157057713901e-07,
      "loss": 0.439,
      "step": 398240
    },
    {
      "epoch": 163.02,
      "grad_norm": 2.360368013381958,
      "learning_rate": 4.247246319178699e-07,
      "loss": 0.4616,
      "step": 398250
    },
    {
      "epoch": 163.02,
      "grad_norm": 2.1240391731262207,
      "learning_rate": 4.2463356685510695e-07,
      "loss": 0.4465,
      "step": 398260
    },
    {
      "epoch": 163.02,
      "grad_norm": 1.9591566324234009,
      "learning_rate": 4.245425105835193e-07,
      "loss": 0.4539,
      "step": 398270
    },
    {
      "epoch": 163.03,
      "grad_norm": 1.7740423679351807,
      "learning_rate": 4.2445146310352387e-07,
      "loss": 0.4457,
      "step": 398280
    },
    {
      "epoch": 163.03,
      "grad_norm": 2.021536350250244,
      "learning_rate": 4.2436042441553803e-07,
      "loss": 0.4295,
      "step": 398290
    },
    {
      "epoch": 163.04,
      "grad_norm": 1.9997605085372925,
      "learning_rate": 4.2426939451997833e-07,
      "loss": 0.4468,
      "step": 398300
    },
    {
      "epoch": 163.04,
      "grad_norm": 2.174987554550171,
      "learning_rate": 4.2417837341726215e-07,
      "loss": 0.4202,
      "step": 398310
    },
    {
      "epoch": 163.05,
      "grad_norm": 2.1840875148773193,
      "learning_rate": 4.240873611078061e-07,
      "loss": 0.4278,
      "step": 398320
    },
    {
      "epoch": 163.05,
      "grad_norm": 2.369752883911133,
      "learning_rate": 4.239963575920276e-07,
      "loss": 0.432,
      "step": 398330
    },
    {
      "epoch": 163.05,
      "grad_norm": 2.178150177001953,
      "learning_rate": 4.2390536287034313e-07,
      "loss": 0.453,
      "step": 398340
    },
    {
      "epoch": 163.06,
      "grad_norm": 2.0925118923187256,
      "learning_rate": 4.238143769431689e-07,
      "loss": 0.451,
      "step": 398350
    },
    {
      "epoch": 163.06,
      "grad_norm": 2.694643020629883,
      "learning_rate": 4.2372339981092296e-07,
      "loss": 0.4429,
      "step": 398360
    },
    {
      "epoch": 163.07,
      "grad_norm": 2.2157950401306152,
      "learning_rate": 4.236324314740217e-07,
      "loss": 0.4478,
      "step": 398370
    },
    {
      "epoch": 163.07,
      "grad_norm": 2.3235018253326416,
      "learning_rate": 4.2354147193288215e-07,
      "loss": 0.4451,
      "step": 398380
    },
    {
      "epoch": 163.07,
      "grad_norm": 2.2800803184509277,
      "learning_rate": 4.2345052118792e-07,
      "loss": 0.4438,
      "step": 398390
    },
    {
      "epoch": 163.08,
      "grad_norm": 1.9943104982376099,
      "learning_rate": 4.2335957923955226e-07,
      "loss": 0.4383,
      "step": 398400
    },
    {
      "epoch": 163.08,
      "grad_norm": 2.4155006408691406,
      "learning_rate": 4.2326864608819584e-07,
      "loss": 0.4647,
      "step": 398410
    },
    {
      "epoch": 163.09,
      "grad_norm": 1.7610195875167847,
      "learning_rate": 4.2317772173426676e-07,
      "loss": 0.4375,
      "step": 398420
    },
    {
      "epoch": 163.09,
      "grad_norm": 2.056168794631958,
      "learning_rate": 4.230868061781819e-07,
      "loss": 0.4368,
      "step": 398430
    },
    {
      "epoch": 163.09,
      "grad_norm": 1.8283554315567017,
      "learning_rate": 4.229958994203579e-07,
      "loss": 0.4336,
      "step": 398440
    },
    {
      "epoch": 163.1,
      "grad_norm": 2.094242811203003,
      "learning_rate": 4.229050014612108e-07,
      "loss": 0.4449,
      "step": 398450
    },
    {
      "epoch": 163.1,
      "grad_norm": 1.7753574848175049,
      "learning_rate": 4.2281411230115716e-07,
      "loss": 0.4728,
      "step": 398460
    },
    {
      "epoch": 163.11,
      "grad_norm": 2.5951693058013916,
      "learning_rate": 4.227232319406133e-07,
      "loss": 0.4363,
      "step": 398470
    },
    {
      "epoch": 163.11,
      "grad_norm": 1.916979193687439,
      "learning_rate": 4.2263236037999566e-07,
      "loss": 0.4403,
      "step": 398480
    },
    {
      "epoch": 163.12,
      "grad_norm": 2.660588026046753,
      "learning_rate": 4.2254149761972043e-07,
      "loss": 0.4451,
      "step": 398490
    },
    {
      "epoch": 163.12,
      "grad_norm": 1.8644407987594604,
      "learning_rate": 4.224506436602038e-07,
      "loss": 0.4359,
      "step": 398500
    },
    {
      "epoch": 163.12,
      "grad_norm": 2.1324586868286133,
      "learning_rate": 4.223597985018618e-07,
      "loss": 0.4508,
      "step": 398510
    },
    {
      "epoch": 163.13,
      "grad_norm": 1.815865397453308,
      "learning_rate": 4.2226896214511107e-07,
      "loss": 0.4393,
      "step": 398520
    },
    {
      "epoch": 163.13,
      "grad_norm": 1.832877278327942,
      "learning_rate": 4.221781345903671e-07,
      "loss": 0.4508,
      "step": 398530
    },
    {
      "epoch": 163.14,
      "grad_norm": 1.9130332469940186,
      "learning_rate": 4.2208731583804645e-07,
      "loss": 0.4463,
      "step": 398540
    },
    {
      "epoch": 163.14,
      "grad_norm": 1.9305546283721924,
      "learning_rate": 4.219965058885648e-07,
      "loss": 0.4321,
      "step": 398550
    },
    {
      "epoch": 163.14,
      "grad_norm": 2.0733072757720947,
      "learning_rate": 4.2190570474233833e-07,
      "loss": 0.4233,
      "step": 398560
    },
    {
      "epoch": 163.15,
      "grad_norm": 2.400062322616577,
      "learning_rate": 4.21814912399783e-07,
      "loss": 0.4455,
      "step": 398570
    },
    {
      "epoch": 163.15,
      "grad_norm": 1.900636911392212,
      "learning_rate": 4.2172412886131454e-07,
      "loss": 0.4441,
      "step": 398580
    },
    {
      "epoch": 163.16,
      "grad_norm": 2.609137535095215,
      "learning_rate": 4.21633354127349e-07,
      "loss": 0.4479,
      "step": 398590
    },
    {
      "epoch": 163.16,
      "grad_norm": 2.1788864135742188,
      "learning_rate": 4.21542588198302e-07,
      "loss": 0.4728,
      "step": 398600
    },
    {
      "epoch": 163.16,
      "grad_norm": 1.9866735935211182,
      "learning_rate": 4.214518310745896e-07,
      "loss": 0.4412,
      "step": 398610
    },
    {
      "epoch": 163.17,
      "grad_norm": 1.9733645915985107,
      "learning_rate": 4.213610827566276e-07,
      "loss": 0.4493,
      "step": 398620
    },
    {
      "epoch": 163.17,
      "grad_norm": 1.8670439720153809,
      "learning_rate": 4.2127034324483126e-07,
      "loss": 0.4384,
      "step": 398630
    },
    {
      "epoch": 163.18,
      "grad_norm": 1.64609694480896,
      "learning_rate": 4.2117961253961646e-07,
      "loss": 0.4388,
      "step": 398640
    },
    {
      "epoch": 163.18,
      "grad_norm": 2.0967941284179688,
      "learning_rate": 4.21088890641399e-07,
      "loss": 0.4375,
      "step": 398650
    },
    {
      "epoch": 163.18,
      "grad_norm": 1.6333422660827637,
      "learning_rate": 4.209981775505941e-07,
      "loss": 0.4483,
      "step": 398660
    },
    {
      "epoch": 163.19,
      "grad_norm": 2.7033531665802,
      "learning_rate": 4.209074732676179e-07,
      "loss": 0.4535,
      "step": 398670
    },
    {
      "epoch": 163.19,
      "grad_norm": 2.2806012630462646,
      "learning_rate": 4.208167777928852e-07,
      "loss": 0.4519,
      "step": 398680
    },
    {
      "epoch": 163.2,
      "grad_norm": 2.271192789077759,
      "learning_rate": 4.2072609112681247e-07,
      "loss": 0.4426,
      "step": 398690
    },
    {
      "epoch": 163.2,
      "grad_norm": 2.3674063682556152,
      "learning_rate": 4.2063541326981385e-07,
      "loss": 0.4428,
      "step": 398700
    },
    {
      "epoch": 163.21,
      "grad_norm": 1.6992709636688232,
      "learning_rate": 4.205447442223053e-07,
      "loss": 0.4484,
      "step": 398710
    },
    {
      "epoch": 163.21,
      "grad_norm": 2.0263795852661133,
      "learning_rate": 4.2045408398470245e-07,
      "loss": 0.4225,
      "step": 398720
    },
    {
      "epoch": 163.21,
      "grad_norm": 2.183457136154175,
      "learning_rate": 4.203634325574195e-07,
      "loss": 0.4593,
      "step": 398730
    },
    {
      "epoch": 163.22,
      "grad_norm": 2.422254800796509,
      "learning_rate": 4.202727899408731e-07,
      "loss": 0.4424,
      "step": 398740
    },
    {
      "epoch": 163.22,
      "grad_norm": 2.128450870513916,
      "learning_rate": 4.20182156135478e-07,
      "loss": 0.4365,
      "step": 398750
    },
    {
      "epoch": 163.23,
      "grad_norm": 2.1917831897735596,
      "learning_rate": 4.2009153114164915e-07,
      "loss": 0.4308,
      "step": 398760
    },
    {
      "epoch": 163.23,
      "grad_norm": 2.354799270629883,
      "learning_rate": 4.2000091495980197e-07,
      "loss": 0.4458,
      "step": 398770
    },
    {
      "epoch": 163.23,
      "grad_norm": 2.1854708194732666,
      "learning_rate": 4.199103075903514e-07,
      "loss": 0.4574,
      "step": 398780
    },
    {
      "epoch": 163.24,
      "grad_norm": 2.11944317817688,
      "learning_rate": 4.198197090337125e-07,
      "loss": 0.452,
      "step": 398790
    },
    {
      "epoch": 163.24,
      "grad_norm": 1.8578839302062988,
      "learning_rate": 4.1972911929030037e-07,
      "loss": 0.4349,
      "step": 398800
    },
    {
      "epoch": 163.25,
      "grad_norm": 2.2710914611816406,
      "learning_rate": 4.196385383605297e-07,
      "loss": 0.4287,
      "step": 398810
    },
    {
      "epoch": 163.25,
      "grad_norm": 1.9068704843521118,
      "learning_rate": 4.1954796624481576e-07,
      "loss": 0.4496,
      "step": 398820
    },
    {
      "epoch": 163.25,
      "grad_norm": 2.2191760540008545,
      "learning_rate": 4.194574029435739e-07,
      "loss": 0.446,
      "step": 398830
    },
    {
      "epoch": 163.26,
      "grad_norm": 2.0447089672088623,
      "learning_rate": 4.193668484572182e-07,
      "loss": 0.4505,
      "step": 398840
    },
    {
      "epoch": 163.26,
      "grad_norm": 1.860052227973938,
      "learning_rate": 4.192763027861632e-07,
      "loss": 0.4391,
      "step": 398850
    },
    {
      "epoch": 163.27,
      "grad_norm": 1.9411022663116455,
      "learning_rate": 4.1918576593082444e-07,
      "loss": 0.4207,
      "step": 398860
    },
    {
      "epoch": 163.27,
      "grad_norm": 1.819308876991272,
      "learning_rate": 4.1909523789161623e-07,
      "loss": 0.4566,
      "step": 398870
    },
    {
      "epoch": 163.27,
      "grad_norm": 2.011113166809082,
      "learning_rate": 4.190047186689535e-07,
      "loss": 0.434,
      "step": 398880
    },
    {
      "epoch": 163.28,
      "grad_norm": 2.137773036956787,
      "learning_rate": 4.189142082632509e-07,
      "loss": 0.4364,
      "step": 398890
    },
    {
      "epoch": 163.28,
      "grad_norm": 2.159208059310913,
      "learning_rate": 4.188237066749227e-07,
      "loss": 0.4298,
      "step": 398900
    },
    {
      "epoch": 163.29,
      "grad_norm": 1.895747184753418,
      "learning_rate": 4.1873321390438407e-07,
      "loss": 0.4489,
      "step": 398910
    },
    {
      "epoch": 163.29,
      "grad_norm": 2.0038270950317383,
      "learning_rate": 4.186427299520483e-07,
      "loss": 0.4359,
      "step": 398920
    },
    {
      "epoch": 163.3,
      "grad_norm": 2.21037220954895,
      "learning_rate": 4.185522548183316e-07,
      "loss": 0.4378,
      "step": 398930
    },
    {
      "epoch": 163.3,
      "grad_norm": 2.196599006652832,
      "learning_rate": 4.1846178850364757e-07,
      "loss": 0.4357,
      "step": 398940
    },
    {
      "epoch": 163.3,
      "grad_norm": 2.1068711280822754,
      "learning_rate": 4.183713310084106e-07,
      "loss": 0.4692,
      "step": 398950
    },
    {
      "epoch": 163.31,
      "grad_norm": 2.4188272953033447,
      "learning_rate": 4.182808823330351e-07,
      "loss": 0.4428,
      "step": 398960
    },
    {
      "epoch": 163.31,
      "grad_norm": 2.0696046352386475,
      "learning_rate": 4.1819044247793615e-07,
      "loss": 0.4499,
      "step": 398970
    },
    {
      "epoch": 163.32,
      "grad_norm": 2.189241886138916,
      "learning_rate": 4.1810001144352673e-07,
      "loss": 0.4481,
      "step": 398980
    },
    {
      "epoch": 163.32,
      "grad_norm": 2.001957893371582,
      "learning_rate": 4.180095892302215e-07,
      "loss": 0.4645,
      "step": 398990
    },
    {
      "epoch": 163.32,
      "grad_norm": 2.2965846061706543,
      "learning_rate": 4.1791917583843514e-07,
      "loss": 0.4328,
      "step": 399000
    },
    {
      "epoch": 163.33,
      "grad_norm": 2.107048988342285,
      "learning_rate": 4.178287712685812e-07,
      "loss": 0.4565,
      "step": 399010
    },
    {
      "epoch": 163.33,
      "grad_norm": 1.9906578063964844,
      "learning_rate": 4.177383755210743e-07,
      "loss": 0.4687,
      "step": 399020
    },
    {
      "epoch": 163.34,
      "grad_norm": 2.8654706478118896,
      "learning_rate": 4.1764798859632844e-07,
      "loss": 0.4556,
      "step": 399030
    },
    {
      "epoch": 163.34,
      "grad_norm": 2.073930263519287,
      "learning_rate": 4.1755761049475756e-07,
      "loss": 0.4642,
      "step": 399040
    },
    {
      "epoch": 163.34,
      "grad_norm": 1.9365633726119995,
      "learning_rate": 4.174672412167757e-07,
      "loss": 0.4679,
      "step": 399050
    },
    {
      "epoch": 163.35,
      "grad_norm": 2.1554834842681885,
      "learning_rate": 4.173768807627968e-07,
      "loss": 0.4472,
      "step": 399060
    },
    {
      "epoch": 163.35,
      "grad_norm": 1.7924031019210815,
      "learning_rate": 4.172865291332349e-07,
      "loss": 0.4487,
      "step": 399070
    },
    {
      "epoch": 163.36,
      "grad_norm": 1.658362865447998,
      "learning_rate": 4.171961863285037e-07,
      "loss": 0.4386,
      "step": 399080
    },
    {
      "epoch": 163.36,
      "grad_norm": 2.2486937046051025,
      "learning_rate": 4.1710585234901717e-07,
      "loss": 0.4632,
      "step": 399090
    },
    {
      "epoch": 163.36,
      "grad_norm": 2.2694602012634277,
      "learning_rate": 4.170155271951894e-07,
      "loss": 0.4461,
      "step": 399100
    },
    {
      "epoch": 163.37,
      "grad_norm": 1.942482352256775,
      "learning_rate": 4.169252108674335e-07,
      "loss": 0.4498,
      "step": 399110
    },
    {
      "epoch": 163.37,
      "grad_norm": 2.4862546920776367,
      "learning_rate": 4.168349033661637e-07,
      "loss": 0.432,
      "step": 399120
    },
    {
      "epoch": 163.38,
      "grad_norm": 2.3683712482452393,
      "learning_rate": 4.1674460469179334e-07,
      "loss": 0.4408,
      "step": 399130
    },
    {
      "epoch": 163.38,
      "grad_norm": 2.291039228439331,
      "learning_rate": 4.166543148447366e-07,
      "loss": 0.442,
      "step": 399140
    },
    {
      "epoch": 163.39,
      "grad_norm": 2.361647367477417,
      "learning_rate": 4.1656403382540643e-07,
      "loss": 0.443,
      "step": 399150
    },
    {
      "epoch": 163.39,
      "grad_norm": 2.0935099124908447,
      "learning_rate": 4.1647376163421704e-07,
      "loss": 0.4579,
      "step": 399160
    },
    {
      "epoch": 163.39,
      "grad_norm": 2.358751058578491,
      "learning_rate": 4.163834982715812e-07,
      "loss": 0.4433,
      "step": 399170
    },
    {
      "epoch": 163.4,
      "grad_norm": 2.128044843673706,
      "learning_rate": 4.162932437379131e-07,
      "loss": 0.461,
      "step": 399180
    },
    {
      "epoch": 163.4,
      "grad_norm": 2.2277324199676514,
      "learning_rate": 4.16202998033626e-07,
      "loss": 0.4535,
      "step": 399190
    },
    {
      "epoch": 163.41,
      "grad_norm": 2.653959035873413,
      "learning_rate": 4.161127611591328e-07,
      "loss": 0.4691,
      "step": 399200
    },
    {
      "epoch": 163.41,
      "grad_norm": 1.8473857641220093,
      "learning_rate": 4.160225331148476e-07,
      "loss": 0.4395,
      "step": 399210
    },
    {
      "epoch": 163.41,
      "grad_norm": 2.046112537384033,
      "learning_rate": 4.159323139011834e-07,
      "loss": 0.4445,
      "step": 399220
    },
    {
      "epoch": 163.42,
      "grad_norm": 1.8553880453109741,
      "learning_rate": 4.158421035185531e-07,
      "loss": 0.4563,
      "step": 399230
    },
    {
      "epoch": 163.42,
      "grad_norm": 2.0681135654449463,
      "learning_rate": 4.157519019673706e-07,
      "loss": 0.4303,
      "step": 399240
    },
    {
      "epoch": 163.43,
      "grad_norm": 2.058983325958252,
      "learning_rate": 4.1566170924804884e-07,
      "loss": 0.4305,
      "step": 399250
    },
    {
      "epoch": 163.43,
      "grad_norm": 2.3927981853485107,
      "learning_rate": 4.155715253610008e-07,
      "loss": 0.4397,
      "step": 399260
    },
    {
      "epoch": 163.43,
      "grad_norm": 1.9872759580612183,
      "learning_rate": 4.1548135030663973e-07,
      "loss": 0.4411,
      "step": 399270
    },
    {
      "epoch": 163.44,
      "grad_norm": 2.076359987258911,
      "learning_rate": 4.153911840853793e-07,
      "loss": 0.4133,
      "step": 399280
    },
    {
      "epoch": 163.44,
      "grad_norm": 1.887359380722046,
      "learning_rate": 4.1530102669763146e-07,
      "loss": 0.4417,
      "step": 399290
    },
    {
      "epoch": 163.45,
      "grad_norm": 2.1868155002593994,
      "learning_rate": 4.152108781438091e-07,
      "loss": 0.4487,
      "step": 399300
    },
    {
      "epoch": 163.45,
      "grad_norm": 2.002639055252075,
      "learning_rate": 4.1512073842432624e-07,
      "loss": 0.456,
      "step": 399310
    },
    {
      "epoch": 163.45,
      "grad_norm": 1.9654531478881836,
      "learning_rate": 4.1503060753959557e-07,
      "loss": 0.4417,
      "step": 399320
    },
    {
      "epoch": 163.46,
      "grad_norm": 1.7373780012130737,
      "learning_rate": 4.1494048549002957e-07,
      "loss": 0.4438,
      "step": 399330
    },
    {
      "epoch": 163.46,
      "grad_norm": 1.974822998046875,
      "learning_rate": 4.1485037227604143e-07,
      "loss": 0.4307,
      "step": 399340
    },
    {
      "epoch": 163.47,
      "grad_norm": 4.677925109863281,
      "learning_rate": 4.147602678980435e-07,
      "loss": 0.4261,
      "step": 399350
    },
    {
      "epoch": 163.47,
      "grad_norm": 1.8922462463378906,
      "learning_rate": 4.1467017235644886e-07,
      "loss": 0.454,
      "step": 399360
    },
    {
      "epoch": 163.48,
      "grad_norm": 1.748491883277893,
      "learning_rate": 4.1458008565167034e-07,
      "loss": 0.4588,
      "step": 399370
    },
    {
      "epoch": 163.48,
      "grad_norm": 2.2080771923065186,
      "learning_rate": 4.144900077841202e-07,
      "loss": 0.4707,
      "step": 399380
    },
    {
      "epoch": 163.48,
      "grad_norm": 2.004971981048584,
      "learning_rate": 4.143999387542113e-07,
      "loss": 0.4508,
      "step": 399390
    },
    {
      "epoch": 163.49,
      "grad_norm": 1.9485325813293457,
      "learning_rate": 4.1430987856235644e-07,
      "loss": 0.4587,
      "step": 399400
    },
    {
      "epoch": 163.49,
      "grad_norm": 1.9194403886795044,
      "learning_rate": 4.142198272089685e-07,
      "loss": 0.4714,
      "step": 399410
    },
    {
      "epoch": 163.5,
      "grad_norm": 1.881250023841858,
      "learning_rate": 4.141297846944587e-07,
      "loss": 0.4511,
      "step": 399420
    },
    {
      "epoch": 163.5,
      "grad_norm": 2.1253297328948975,
      "learning_rate": 4.140397510192403e-07,
      "loss": 0.4426,
      "step": 399430
    },
    {
      "epoch": 163.5,
      "grad_norm": 2.120709180831909,
      "learning_rate": 4.13949726183726e-07,
      "loss": 0.4371,
      "step": 399440
    },
    {
      "epoch": 163.51,
      "grad_norm": 1.9704879522323608,
      "learning_rate": 4.138597101883275e-07,
      "loss": 0.4886,
      "step": 399450
    },
    {
      "epoch": 163.51,
      "grad_norm": 2.2101988792419434,
      "learning_rate": 4.137697030334578e-07,
      "loss": 0.4483,
      "step": 399460
    },
    {
      "epoch": 163.52,
      "grad_norm": 1.9258416891098022,
      "learning_rate": 4.136797047195287e-07,
      "loss": 0.4298,
      "step": 399470
    },
    {
      "epoch": 163.52,
      "grad_norm": 2.0218474864959717,
      "learning_rate": 4.1358971524695294e-07,
      "loss": 0.436,
      "step": 399480
    },
    {
      "epoch": 163.52,
      "grad_norm": 1.6277637481689453,
      "learning_rate": 4.1349973461614175e-07,
      "loss": 0.4355,
      "step": 399490
    },
    {
      "epoch": 163.53,
      "grad_norm": 1.7648504972457886,
      "learning_rate": 4.1340976282750887e-07,
      "loss": 0.442,
      "step": 399500
    },
    {
      "epoch": 163.53,
      "grad_norm": 1.862008810043335,
      "learning_rate": 4.1331979988146564e-07,
      "loss": 0.4329,
      "step": 399510
    },
    {
      "epoch": 163.54,
      "grad_norm": 1.8382444381713867,
      "learning_rate": 4.1322984577842404e-07,
      "loss": 0.4352,
      "step": 399520
    },
    {
      "epoch": 163.54,
      "grad_norm": 1.9694675207138062,
      "learning_rate": 4.131399005187964e-07,
      "loss": 0.436,
      "step": 399530
    },
    {
      "epoch": 163.54,
      "grad_norm": 2.000345230102539,
      "learning_rate": 4.130499641029945e-07,
      "loss": 0.4321,
      "step": 399540
    },
    {
      "epoch": 163.55,
      "grad_norm": 2.152384042739868,
      "learning_rate": 4.129600365314312e-07,
      "loss": 0.4263,
      "step": 399550
    },
    {
      "epoch": 163.55,
      "grad_norm": 1.950260043144226,
      "learning_rate": 4.1287011780451715e-07,
      "loss": 0.4353,
      "step": 399560
    },
    {
      "epoch": 163.56,
      "grad_norm": 2.054671049118042,
      "learning_rate": 4.127802079226647e-07,
      "loss": 0.4494,
      "step": 399570
    },
    {
      "epoch": 163.56,
      "grad_norm": 1.6199222803115845,
      "learning_rate": 4.1269030688628564e-07,
      "loss": 0.4356,
      "step": 399580
    },
    {
      "epoch": 163.57,
      "grad_norm": 1.9628076553344727,
      "learning_rate": 4.126004146957923e-07,
      "loss": 0.4342,
      "step": 399590
    },
    {
      "epoch": 163.57,
      "grad_norm": 1.8257262706756592,
      "learning_rate": 4.1251053135159607e-07,
      "loss": 0.4341,
      "step": 399600
    },
    {
      "epoch": 163.57,
      "grad_norm": 2.030608892440796,
      "learning_rate": 4.124206568541086e-07,
      "loss": 0.4477,
      "step": 399610
    },
    {
      "epoch": 163.58,
      "grad_norm": 5.141617298126221,
      "learning_rate": 4.1233079120374186e-07,
      "loss": 0.4442,
      "step": 399620
    },
    {
      "epoch": 163.58,
      "grad_norm": 2.2387712001800537,
      "learning_rate": 4.122409344009072e-07,
      "loss": 0.4392,
      "step": 399630
    },
    {
      "epoch": 163.59,
      "grad_norm": 1.9789434671401978,
      "learning_rate": 4.121510864460168e-07,
      "loss": 0.4551,
      "step": 399640
    },
    {
      "epoch": 163.59,
      "grad_norm": 1.9041703939437866,
      "learning_rate": 4.120612473394815e-07,
      "loss": 0.4464,
      "step": 399650
    },
    {
      "epoch": 163.59,
      "grad_norm": 2.010288715362549,
      "learning_rate": 4.1197141708171354e-07,
      "loss": 0.451,
      "step": 399660
    },
    {
      "epoch": 163.6,
      "grad_norm": 2.069690704345703,
      "learning_rate": 4.11881595673124e-07,
      "loss": 0.4337,
      "step": 399670
    },
    {
      "epoch": 163.6,
      "grad_norm": 2.0598156452178955,
      "learning_rate": 4.117917831141244e-07,
      "loss": 0.4326,
      "step": 399680
    },
    {
      "epoch": 163.61,
      "grad_norm": 2.1567189693450928,
      "learning_rate": 4.1170197940512616e-07,
      "loss": 0.4422,
      "step": 399690
    },
    {
      "epoch": 163.61,
      "grad_norm": 2.4070584774017334,
      "learning_rate": 4.116121845465407e-07,
      "loss": 0.4322,
      "step": 399700
    },
    {
      "epoch": 163.61,
      "grad_norm": 1.8858120441436768,
      "learning_rate": 4.115223985387792e-07,
      "loss": 0.4366,
      "step": 399710
    },
    {
      "epoch": 163.62,
      "grad_norm": 2.3229315280914307,
      "learning_rate": 4.1143262138225337e-07,
      "loss": 0.4438,
      "step": 399720
    },
    {
      "epoch": 163.62,
      "grad_norm": 2.120020866394043,
      "learning_rate": 4.1134285307737403e-07,
      "loss": 0.4627,
      "step": 399730
    },
    {
      "epoch": 163.63,
      "grad_norm": 2.1454977989196777,
      "learning_rate": 4.112530936245526e-07,
      "loss": 0.4666,
      "step": 399740
    },
    {
      "epoch": 163.63,
      "grad_norm": 2.1209053993225098,
      "learning_rate": 4.111633430242004e-07,
      "loss": 0.4244,
      "step": 399750
    },
    {
      "epoch": 163.63,
      "grad_norm": 1.9576961994171143,
      "learning_rate": 4.110736012767281e-07,
      "loss": 0.4517,
      "step": 399760
    },
    {
      "epoch": 163.64,
      "grad_norm": 6.971258163452148,
      "learning_rate": 4.1098386838254707e-07,
      "loss": 0.472,
      "step": 399770
    },
    {
      "epoch": 163.64,
      "grad_norm": 1.864751935005188,
      "learning_rate": 4.108941443420685e-07,
      "loss": 0.4512,
      "step": 399780
    },
    {
      "epoch": 163.65,
      "grad_norm": 1.8139102458953857,
      "learning_rate": 4.1080442915570316e-07,
      "loss": 0.442,
      "step": 399790
    },
    {
      "epoch": 163.65,
      "grad_norm": 1.8844902515411377,
      "learning_rate": 4.1071472282386223e-07,
      "loss": 0.4463,
      "step": 399800
    },
    {
      "epoch": 163.66,
      "grad_norm": 2.13332462310791,
      "learning_rate": 4.106250253469566e-07,
      "loss": 0.4592,
      "step": 399810
    },
    {
      "epoch": 163.66,
      "grad_norm": 2.4256858825683594,
      "learning_rate": 4.1053533672539684e-07,
      "loss": 0.4436,
      "step": 399820
    },
    {
      "epoch": 163.66,
      "grad_norm": 4.268185138702393,
      "learning_rate": 4.104456569595943e-07,
      "loss": 0.4407,
      "step": 399830
    },
    {
      "epoch": 163.67,
      "grad_norm": 2.181835412979126,
      "learning_rate": 4.1035598604995946e-07,
      "loss": 0.4592,
      "step": 399840
    },
    {
      "epoch": 163.67,
      "grad_norm": 1.7628192901611328,
      "learning_rate": 4.102663239969031e-07,
      "loss": 0.4338,
      "step": 399850
    },
    {
      "epoch": 163.68,
      "grad_norm": 1.911757230758667,
      "learning_rate": 4.1017667080083606e-07,
      "loss": 0.4352,
      "step": 399860
    },
    {
      "epoch": 163.68,
      "grad_norm": 2.0698070526123047,
      "learning_rate": 4.10087026462169e-07,
      "loss": 0.4318,
      "step": 399870
    },
    {
      "epoch": 163.68,
      "grad_norm": 1.7617589235305786,
      "learning_rate": 4.099973909813128e-07,
      "loss": 0.4509,
      "step": 399880
    },
    {
      "epoch": 163.69,
      "grad_norm": 1.8786661624908447,
      "learning_rate": 4.099077643586775e-07,
      "loss": 0.4526,
      "step": 399890
    },
    {
      "epoch": 163.69,
      "grad_norm": 2.1535024642944336,
      "learning_rate": 4.0981814659467404e-07,
      "loss": 0.428,
      "step": 399900
    },
    {
      "epoch": 163.7,
      "grad_norm": 2.213257312774658,
      "learning_rate": 4.097285376897129e-07,
      "loss": 0.452,
      "step": 399910
    },
    {
      "epoch": 163.7,
      "grad_norm": 2.0253565311431885,
      "learning_rate": 4.096389376442046e-07,
      "loss": 0.4701,
      "step": 399920
    },
    {
      "epoch": 163.7,
      "grad_norm": 1.659183382987976,
      "learning_rate": 4.095493464585597e-07,
      "loss": 0.4635,
      "step": 399930
    },
    {
      "epoch": 163.71,
      "grad_norm": 2.0482239723205566,
      "learning_rate": 4.0945976413318815e-07,
      "loss": 0.4433,
      "step": 399940
    },
    {
      "epoch": 163.71,
      "grad_norm": 1.870144009590149,
      "learning_rate": 4.093701906685007e-07,
      "loss": 0.4234,
      "step": 399950
    },
    {
      "epoch": 163.72,
      "grad_norm": 2.090120315551758,
      "learning_rate": 4.092806260649077e-07,
      "loss": 0.4537,
      "step": 399960
    },
    {
      "epoch": 163.72,
      "grad_norm": 2.464958429336548,
      "learning_rate": 4.091910703228191e-07,
      "loss": 0.4345,
      "step": 399970
    },
    {
      "epoch": 163.72,
      "grad_norm": 2.2381701469421387,
      "learning_rate": 4.091015234426457e-07,
      "loss": 0.4572,
      "step": 399980
    },
    {
      "epoch": 163.73,
      "grad_norm": 2.1837687492370605,
      "learning_rate": 4.090119854247977e-07,
      "loss": 0.429,
      "step": 399990
    },
    {
      "epoch": 163.73,
      "grad_norm": 2.1077218055725098,
      "learning_rate": 4.089224562696843e-07,
      "loss": 0.4506,
      "step": 400000
    },
    {
      "epoch": 163.74,
      "grad_norm": 1.5749081373214722,
      "learning_rate": 4.088329359777164e-07,
      "loss": 0.4502,
      "step": 400010
    },
    {
      "epoch": 163.74,
      "grad_norm": 2.1921846866607666,
      "learning_rate": 4.087434245493038e-07,
      "loss": 0.4641,
      "step": 400020
    },
    {
      "epoch": 163.75,
      "grad_norm": 2.2288379669189453,
      "learning_rate": 4.0865392198485646e-07,
      "loss": 0.4379,
      "step": 400030
    },
    {
      "epoch": 163.75,
      "grad_norm": 2.0993220806121826,
      "learning_rate": 4.0856442828478473e-07,
      "loss": 0.4552,
      "step": 400040
    },
    {
      "epoch": 163.75,
      "grad_norm": 2.599273920059204,
      "learning_rate": 4.08474943449498e-07,
      "loss": 0.4334,
      "step": 400050
    },
    {
      "epoch": 163.76,
      "grad_norm": 2.4473001956939697,
      "learning_rate": 4.083854674794071e-07,
      "loss": 0.4357,
      "step": 400060
    },
    {
      "epoch": 163.76,
      "grad_norm": 2.264195203781128,
      "learning_rate": 4.0829600037492126e-07,
      "loss": 0.4507,
      "step": 400070
    },
    {
      "epoch": 163.77,
      "grad_norm": 1.8712669610977173,
      "learning_rate": 4.0820654213645037e-07,
      "loss": 0.4315,
      "step": 400080
    },
    {
      "epoch": 163.77,
      "grad_norm": 2.0912671089172363,
      "learning_rate": 4.081170927644046e-07,
      "loss": 0.4342,
      "step": 400090
    },
    {
      "epoch": 163.77,
      "grad_norm": 1.9739806652069092,
      "learning_rate": 4.0802765225919313e-07,
      "loss": 0.4245,
      "step": 400100
    },
    {
      "epoch": 163.78,
      "grad_norm": 1.9387234449386597,
      "learning_rate": 4.079382206212261e-07,
      "loss": 0.4437,
      "step": 400110
    },
    {
      "epoch": 163.78,
      "grad_norm": 2.62261700630188,
      "learning_rate": 4.078487978509128e-07,
      "loss": 0.4452,
      "step": 400120
    },
    {
      "epoch": 163.79,
      "grad_norm": 1.8576828241348267,
      "learning_rate": 4.0775938394866386e-07,
      "loss": 0.4556,
      "step": 400130
    },
    {
      "epoch": 163.79,
      "grad_norm": 1.9480897188186646,
      "learning_rate": 4.0766997891488756e-07,
      "loss": 0.4482,
      "step": 400140
    },
    {
      "epoch": 163.79,
      "grad_norm": 2.203892469406128,
      "learning_rate": 4.07580582749994e-07,
      "loss": 0.4338,
      "step": 400150
    },
    {
      "epoch": 163.8,
      "grad_norm": 1.9676038026809692,
      "learning_rate": 4.074911954543927e-07,
      "loss": 0.4555,
      "step": 400160
    },
    {
      "epoch": 163.8,
      "grad_norm": 1.9343675374984741,
      "learning_rate": 4.074018170284932e-07,
      "loss": 0.4589,
      "step": 400170
    },
    {
      "epoch": 163.81,
      "grad_norm": 1.8101305961608887,
      "learning_rate": 4.073124474727046e-07,
      "loss": 0.4284,
      "step": 400180
    },
    {
      "epoch": 163.81,
      "grad_norm": 2.0712199211120605,
      "learning_rate": 4.0722308678743654e-07,
      "loss": 0.4424,
      "step": 400190
    },
    {
      "epoch": 163.81,
      "grad_norm": 2.4997105598449707,
      "learning_rate": 4.0713373497309846e-07,
      "loss": 0.4539,
      "step": 400200
    },
    {
      "epoch": 163.82,
      "grad_norm": 1.840539813041687,
      "learning_rate": 4.0704439203009976e-07,
      "loss": 0.4381,
      "step": 400210
    },
    {
      "epoch": 163.82,
      "grad_norm": 1.8105140924453735,
      "learning_rate": 4.069550579588494e-07,
      "loss": 0.433,
      "step": 400220
    },
    {
      "epoch": 163.83,
      "grad_norm": 1.982182264328003,
      "learning_rate": 4.068657327597565e-07,
      "loss": 0.45,
      "step": 400230
    },
    {
      "epoch": 163.83,
      "grad_norm": 1.9558018445968628,
      "learning_rate": 4.0677641643323005e-07,
      "loss": 0.4574,
      "step": 400240
    },
    {
      "epoch": 163.84,
      "grad_norm": 1.894459843635559,
      "learning_rate": 4.066871089796802e-07,
      "loss": 0.4623,
      "step": 400250
    },
    {
      "epoch": 163.84,
      "grad_norm": 1.9725277423858643,
      "learning_rate": 4.065978103995155e-07,
      "loss": 0.4507,
      "step": 400260
    },
    {
      "epoch": 163.84,
      "grad_norm": 2.196713924407959,
      "learning_rate": 4.0650852069314536e-07,
      "loss": 0.4604,
      "step": 400270
    },
    {
      "epoch": 163.85,
      "grad_norm": 1.9944748878479004,
      "learning_rate": 4.0641923986097797e-07,
      "loss": 0.433,
      "step": 400280
    },
    {
      "epoch": 163.85,
      "grad_norm": 2.615988254547119,
      "learning_rate": 4.063299679034228e-07,
      "loss": 0.4445,
      "step": 400290
    },
    {
      "epoch": 163.86,
      "grad_norm": 2.6050994396209717,
      "learning_rate": 4.0624070482088867e-07,
      "loss": 0.4728,
      "step": 400300
    },
    {
      "epoch": 163.86,
      "grad_norm": 2.0064730644226074,
      "learning_rate": 4.061514506137846e-07,
      "loss": 0.4489,
      "step": 400310
    },
    {
      "epoch": 163.86,
      "grad_norm": 2.0736849308013916,
      "learning_rate": 4.060622052825194e-07,
      "loss": 0.448,
      "step": 400320
    },
    {
      "epoch": 163.87,
      "grad_norm": 1.928384780883789,
      "learning_rate": 4.059729688275019e-07,
      "loss": 0.4406,
      "step": 400330
    },
    {
      "epoch": 163.87,
      "grad_norm": 2.155426025390625,
      "learning_rate": 4.0588374124914103e-07,
      "loss": 0.4416,
      "step": 400340
    },
    {
      "epoch": 163.88,
      "grad_norm": 2.3347697257995605,
      "learning_rate": 4.0579452254784516e-07,
      "loss": 0.4459,
      "step": 400350
    },
    {
      "epoch": 163.88,
      "grad_norm": 2.1340832710266113,
      "learning_rate": 4.057053127240233e-07,
      "loss": 0.4337,
      "step": 400360
    },
    {
      "epoch": 163.88,
      "grad_norm": 2.333461046218872,
      "learning_rate": 4.05616111778084e-07,
      "loss": 0.4662,
      "step": 400370
    },
    {
      "epoch": 163.89,
      "grad_norm": 2.281036376953125,
      "learning_rate": 4.055269197104358e-07,
      "loss": 0.4542,
      "step": 400380
    },
    {
      "epoch": 163.89,
      "grad_norm": 2.238293409347534,
      "learning_rate": 4.0543773652148743e-07,
      "loss": 0.4538,
      "step": 400390
    },
    {
      "epoch": 163.9,
      "grad_norm": 2.1315317153930664,
      "learning_rate": 4.053485622116474e-07,
      "loss": 0.4368,
      "step": 400400
    },
    {
      "epoch": 163.9,
      "grad_norm": 1.962431788444519,
      "learning_rate": 4.052593967813241e-07,
      "loss": 0.4427,
      "step": 400410
    },
    {
      "epoch": 163.91,
      "grad_norm": 1.9096527099609375,
      "learning_rate": 4.0517024023092595e-07,
      "loss": 0.4565,
      "step": 400420
    },
    {
      "epoch": 163.91,
      "grad_norm": 1.7129929065704346,
      "learning_rate": 4.0508109256086145e-07,
      "loss": 0.4537,
      "step": 400430
    },
    {
      "epoch": 163.91,
      "grad_norm": 1.8865395784378052,
      "learning_rate": 4.0499195377153925e-07,
      "loss": 0.4542,
      "step": 400440
    },
    {
      "epoch": 163.92,
      "grad_norm": 2.623133420944214,
      "learning_rate": 4.0490282386336703e-07,
      "loss": 0.4232,
      "step": 400450
    },
    {
      "epoch": 163.92,
      "grad_norm": 2.0102412700653076,
      "learning_rate": 4.0481370283675385e-07,
      "loss": 0.4467,
      "step": 400460
    },
    {
      "epoch": 163.93,
      "grad_norm": 1.8096531629562378,
      "learning_rate": 4.0472459069210714e-07,
      "loss": 0.4605,
      "step": 400470
    },
    {
      "epoch": 163.93,
      "grad_norm": 1.9148951768875122,
      "learning_rate": 4.0463548742983575e-07,
      "loss": 0.4354,
      "step": 400480
    },
    {
      "epoch": 163.93,
      "grad_norm": 1.7836424112319946,
      "learning_rate": 4.045463930503477e-07,
      "loss": 0.4434,
      "step": 400490
    },
    {
      "epoch": 163.94,
      "grad_norm": 1.7744910717010498,
      "learning_rate": 4.0445730755405096e-07,
      "loss": 0.4775,
      "step": 400500
    },
    {
      "epoch": 163.94,
      "grad_norm": 1.9941047430038452,
      "learning_rate": 4.0436823094135377e-07,
      "loss": 0.4539,
      "step": 400510
    },
    {
      "epoch": 163.95,
      "grad_norm": 1.6150718927383423,
      "learning_rate": 4.042791632126642e-07,
      "loss": 0.4464,
      "step": 400520
    },
    {
      "epoch": 163.95,
      "grad_norm": 2.6979868412017822,
      "learning_rate": 4.0419010436838996e-07,
      "loss": 0.4651,
      "step": 400530
    },
    {
      "epoch": 163.95,
      "grad_norm": 1.8957284688949585,
      "learning_rate": 4.0410105440893934e-07,
      "loss": 0.4248,
      "step": 400540
    },
    {
      "epoch": 163.96,
      "grad_norm": 2.8952770233154297,
      "learning_rate": 4.0401201333472e-07,
      "loss": 0.4387,
      "step": 400550
    },
    {
      "epoch": 163.96,
      "grad_norm": 2.0826961994171143,
      "learning_rate": 4.039229811461401e-07,
      "loss": 0.4572,
      "step": 400560
    },
    {
      "epoch": 163.97,
      "grad_norm": 2.021268606185913,
      "learning_rate": 4.038339578436076e-07,
      "loss": 0.4268,
      "step": 400570
    },
    {
      "epoch": 163.97,
      "grad_norm": 1.9922634363174438,
      "learning_rate": 4.0374494342753016e-07,
      "loss": 0.456,
      "step": 400580
    },
    {
      "epoch": 163.97,
      "grad_norm": 2.085236072540283,
      "learning_rate": 4.0365593789831506e-07,
      "loss": 0.4487,
      "step": 400590
    },
    {
      "epoch": 163.98,
      "grad_norm": 1.9773598909378052,
      "learning_rate": 4.0356694125637054e-07,
      "loss": 0.4432,
      "step": 400600
    },
    {
      "epoch": 163.98,
      "grad_norm": 1.914009690284729,
      "learning_rate": 4.0347795350210387e-07,
      "loss": 0.4278,
      "step": 400610
    },
    {
      "epoch": 163.99,
      "grad_norm": 1.917943000793457,
      "learning_rate": 4.0338897463592253e-07,
      "loss": 0.4704,
      "step": 400620
    },
    {
      "epoch": 163.99,
      "grad_norm": 1.8596857786178589,
      "learning_rate": 4.033000046582353e-07,
      "loss": 0.4755,
      "step": 400630
    },
    {
      "epoch": 164.0,
      "grad_norm": 1.8456411361694336,
      "learning_rate": 4.0321104356944865e-07,
      "loss": 0.4451,
      "step": 400640
    },
    {
      "epoch": 164.0,
      "grad_norm": 1.9425384998321533,
      "learning_rate": 4.0312209136997053e-07,
      "loss": 0.4517,
      "step": 400650
    },
    {
      "epoch": 164.0,
      "eval_loss": 0.4483622610569,
      "eval_runtime": 52.5442,
      "eval_samples_per_second": 65.64,
      "eval_steps_per_second": 8.222,
      "step": 400652
    },
    {
      "epoch": 164.0,
      "grad_norm": 2.192408800125122,
      "learning_rate": 4.0303314806020844e-07,
      "loss": 0.4447,
      "step": 400660
    },
    {
      "epoch": 164.01,
      "grad_norm": 2.744377851486206,
      "learning_rate": 4.0294421364056964e-07,
      "loss": 0.451,
      "step": 400670
    },
    {
      "epoch": 164.01,
      "grad_norm": 1.7959364652633667,
      "learning_rate": 4.028552881114616e-07,
      "loss": 0.4447,
      "step": 400680
    },
    {
      "epoch": 164.02,
      "grad_norm": 3.6930229663848877,
      "learning_rate": 4.027663714732916e-07,
      "loss": 0.4389,
      "step": 400690
    },
    {
      "epoch": 164.02,
      "grad_norm": 2.3817782402038574,
      "learning_rate": 4.02677463726467e-07,
      "loss": 0.448,
      "step": 400700
    },
    {
      "epoch": 164.02,
      "grad_norm": 2.2794532775878906,
      "learning_rate": 4.025885648713956e-07,
      "loss": 0.4371,
      "step": 400710
    },
    {
      "epoch": 164.03,
      "grad_norm": 1.8466062545776367,
      "learning_rate": 4.0249967490848365e-07,
      "loss": 0.4395,
      "step": 400720
    },
    {
      "epoch": 164.03,
      "grad_norm": 2.4456522464752197,
      "learning_rate": 4.024107938381386e-07,
      "loss": 0.462,
      "step": 400730
    },
    {
      "epoch": 164.04,
      "grad_norm": 1.8980588912963867,
      "learning_rate": 4.023219216607679e-07,
      "loss": 0.4676,
      "step": 400740
    },
    {
      "epoch": 164.04,
      "grad_norm": 2.063180923461914,
      "learning_rate": 4.0223305837677857e-07,
      "loss": 0.4451,
      "step": 400750
    },
    {
      "epoch": 164.04,
      "grad_norm": 2.449110507965088,
      "learning_rate": 4.021442039865778e-07,
      "loss": 0.4426,
      "step": 400760
    },
    {
      "epoch": 164.05,
      "grad_norm": 1.914093255996704,
      "learning_rate": 4.020553584905723e-07,
      "loss": 0.4703,
      "step": 400770
    },
    {
      "epoch": 164.05,
      "grad_norm": 2.1855175495147705,
      "learning_rate": 4.019665218891693e-07,
      "loss": 0.4554,
      "step": 400780
    },
    {
      "epoch": 164.06,
      "grad_norm": 1.973881721496582,
      "learning_rate": 4.018776941827758e-07,
      "loss": 0.4535,
      "step": 400790
    },
    {
      "epoch": 164.06,
      "grad_norm": 2.020348072052002,
      "learning_rate": 4.0178887537179844e-07,
      "loss": 0.4222,
      "step": 400800
    },
    {
      "epoch": 164.06,
      "grad_norm": 1.907080888748169,
      "learning_rate": 4.017000654566437e-07,
      "loss": 0.4568,
      "step": 400810
    },
    {
      "epoch": 164.07,
      "grad_norm": 2.020012140274048,
      "learning_rate": 4.0161126443771956e-07,
      "loss": 0.462,
      "step": 400820
    },
    {
      "epoch": 164.07,
      "grad_norm": 2.310227632522583,
      "learning_rate": 4.015224723154323e-07,
      "loss": 0.4364,
      "step": 400830
    },
    {
      "epoch": 164.08,
      "grad_norm": 2.007596969604492,
      "learning_rate": 4.0143368909018826e-07,
      "loss": 0.4384,
      "step": 400840
    },
    {
      "epoch": 164.08,
      "grad_norm": 2.10964298248291,
      "learning_rate": 4.0134491476239525e-07,
      "loss": 0.4534,
      "step": 400850
    },
    {
      "epoch": 164.09,
      "grad_norm": 3.0173540115356445,
      "learning_rate": 4.0125614933245835e-07,
      "loss": 0.4488,
      "step": 400860
    },
    {
      "epoch": 164.09,
      "grad_norm": 2.5595829486846924,
      "learning_rate": 4.011673928007851e-07,
      "loss": 0.4691,
      "step": 400870
    },
    {
      "epoch": 164.09,
      "grad_norm": 1.8381173610687256,
      "learning_rate": 4.010786451677819e-07,
      "loss": 0.4499,
      "step": 400880
    },
    {
      "epoch": 164.1,
      "grad_norm": 2.3698818683624268,
      "learning_rate": 4.0098990643385533e-07,
      "loss": 0.4484,
      "step": 400890
    },
    {
      "epoch": 164.1,
      "grad_norm": 1.931098461151123,
      "learning_rate": 4.009011765994119e-07,
      "loss": 0.4252,
      "step": 400900
    },
    {
      "epoch": 164.11,
      "grad_norm": 2.5274174213409424,
      "learning_rate": 4.0081245566485793e-07,
      "loss": 0.4332,
      "step": 400910
    },
    {
      "epoch": 164.11,
      "grad_norm": 2.1273481845855713,
      "learning_rate": 4.0072374363060007e-07,
      "loss": 0.4466,
      "step": 400920
    },
    {
      "epoch": 164.11,
      "grad_norm": 1.9776277542114258,
      "learning_rate": 4.006350404970445e-07,
      "loss": 0.4516,
      "step": 400930
    },
    {
      "epoch": 164.12,
      "grad_norm": 2.4282586574554443,
      "learning_rate": 4.0054634626459774e-07,
      "loss": 0.4323,
      "step": 400940
    },
    {
      "epoch": 164.12,
      "grad_norm": 1.9655648469924927,
      "learning_rate": 4.004576609336659e-07,
      "loss": 0.4553,
      "step": 400950
    },
    {
      "epoch": 164.13,
      "grad_norm": 2.5534281730651855,
      "learning_rate": 4.003689845046552e-07,
      "loss": 0.4431,
      "step": 400960
    },
    {
      "epoch": 164.13,
      "grad_norm": 2.2800965309143066,
      "learning_rate": 4.0028031697797203e-07,
      "loss": 0.4366,
      "step": 400970
    },
    {
      "epoch": 164.13,
      "grad_norm": 1.8036082983016968,
      "learning_rate": 4.001916583540227e-07,
      "loss": 0.4413,
      "step": 400980
    },
    {
      "epoch": 164.14,
      "grad_norm": 1.9314640760421753,
      "learning_rate": 4.00103008633213e-07,
      "loss": 0.4733,
      "step": 400990
    },
    {
      "epoch": 164.14,
      "grad_norm": 2.5480124950408936,
      "learning_rate": 4.0001436781594886e-07,
      "loss": 0.4331,
      "step": 401000
    },
    {
      "epoch": 164.15,
      "grad_norm": 3.154090404510498,
      "learning_rate": 3.9992573590263706e-07,
      "loss": 0.46,
      "step": 401010
    },
    {
      "epoch": 164.15,
      "grad_norm": 2.32399582862854,
      "learning_rate": 3.998371128936828e-07,
      "loss": 0.4534,
      "step": 401020
    },
    {
      "epoch": 164.15,
      "grad_norm": 1.8634153604507446,
      "learning_rate": 3.997484987894926e-07,
      "loss": 0.4392,
      "step": 401030
    },
    {
      "epoch": 164.16,
      "grad_norm": 2.221393585205078,
      "learning_rate": 3.996598935904723e-07,
      "loss": 0.444,
      "step": 401040
    },
    {
      "epoch": 164.16,
      "grad_norm": 2.3744685649871826,
      "learning_rate": 3.9957129729702774e-07,
      "loss": 0.4604,
      "step": 401050
    },
    {
      "epoch": 164.17,
      "grad_norm": 2.1438300609588623,
      "learning_rate": 3.994827099095646e-07,
      "loss": 0.4422,
      "step": 401060
    },
    {
      "epoch": 164.17,
      "grad_norm": 1.6893861293792725,
      "learning_rate": 3.9939413142848875e-07,
      "loss": 0.4717,
      "step": 401070
    },
    {
      "epoch": 164.18,
      "grad_norm": 1.9450528621673584,
      "learning_rate": 3.99305561854206e-07,
      "loss": 0.4524,
      "step": 401080
    },
    {
      "epoch": 164.18,
      "grad_norm": 2.175917387008667,
      "learning_rate": 3.9921700118712215e-07,
      "loss": 0.4553,
      "step": 401090
    },
    {
      "epoch": 164.18,
      "grad_norm": 2.107839822769165,
      "learning_rate": 3.991284494276429e-07,
      "loss": 0.4119,
      "step": 401100
    },
    {
      "epoch": 164.19,
      "grad_norm": 2.008608341217041,
      "learning_rate": 3.9903990657617366e-07,
      "loss": 0.4102,
      "step": 401110
    },
    {
      "epoch": 164.19,
      "grad_norm": 3.073561429977417,
      "learning_rate": 3.9895137263312034e-07,
      "loss": 0.4393,
      "step": 401120
    },
    {
      "epoch": 164.2,
      "grad_norm": 1.6284196376800537,
      "learning_rate": 3.9886284759888836e-07,
      "loss": 0.4229,
      "step": 401130
    },
    {
      "epoch": 164.2,
      "grad_norm": 2.520211696624756,
      "learning_rate": 3.987743314738831e-07,
      "loss": 0.4432,
      "step": 401140
    },
    {
      "epoch": 164.2,
      "grad_norm": 2.3600947856903076,
      "learning_rate": 3.9868582425851003e-07,
      "loss": 0.4586,
      "step": 401150
    },
    {
      "epoch": 164.21,
      "grad_norm": 2.18902587890625,
      "learning_rate": 3.9859732595317557e-07,
      "loss": 0.4832,
      "step": 401160
    },
    {
      "epoch": 164.21,
      "grad_norm": 2.6075282096862793,
      "learning_rate": 3.985088365582835e-07,
      "loss": 0.4288,
      "step": 401170
    },
    {
      "epoch": 164.22,
      "grad_norm": 1.7001299858093262,
      "learning_rate": 3.9842035607423954e-07,
      "loss": 0.4563,
      "step": 401180
    },
    {
      "epoch": 164.22,
      "grad_norm": 1.8170740604400635,
      "learning_rate": 3.9833188450144986e-07,
      "loss": 0.4507,
      "step": 401190
    },
    {
      "epoch": 164.22,
      "grad_norm": 2.208667039871216,
      "learning_rate": 3.982434218403194e-07,
      "loss": 0.4327,
      "step": 401200
    },
    {
      "epoch": 164.23,
      "grad_norm": 2.357396364212036,
      "learning_rate": 3.981549680912529e-07,
      "loss": 0.4433,
      "step": 401210
    },
    {
      "epoch": 164.23,
      "grad_norm": 2.2423789501190186,
      "learning_rate": 3.9806652325465617e-07,
      "loss": 0.4313,
      "step": 401220
    },
    {
      "epoch": 164.24,
      "grad_norm": 2.0729892253875732,
      "learning_rate": 3.9797808733093436e-07,
      "loss": 0.4384,
      "step": 401230
    },
    {
      "epoch": 164.24,
      "grad_norm": 1.9258912801742554,
      "learning_rate": 3.9788966032049197e-07,
      "loss": 0.4376,
      "step": 401240
    },
    {
      "epoch": 164.24,
      "grad_norm": 2.2685015201568604,
      "learning_rate": 3.9780124222373473e-07,
      "loss": 0.4411,
      "step": 401250
    },
    {
      "epoch": 164.25,
      "grad_norm": 2.092686414718628,
      "learning_rate": 3.977128330410673e-07,
      "loss": 0.4533,
      "step": 401260
    },
    {
      "epoch": 164.25,
      "grad_norm": 2.287468671798706,
      "learning_rate": 3.9762443277289456e-07,
      "loss": 0.4671,
      "step": 401270
    },
    {
      "epoch": 164.26,
      "grad_norm": 2.1061885356903076,
      "learning_rate": 3.9753604141962185e-07,
      "loss": 0.4565,
      "step": 401280
    },
    {
      "epoch": 164.26,
      "grad_norm": 1.890285611152649,
      "learning_rate": 3.9744765898165383e-07,
      "loss": 0.4554,
      "step": 401290
    },
    {
      "epoch": 164.27,
      "grad_norm": 1.867671012878418,
      "learning_rate": 3.9735928545939595e-07,
      "loss": 0.4475,
      "step": 401300
    },
    {
      "epoch": 164.27,
      "grad_norm": 2.203667402267456,
      "learning_rate": 3.9727092085325193e-07,
      "loss": 0.4704,
      "step": 401310
    },
    {
      "epoch": 164.27,
      "grad_norm": 1.8707128763198853,
      "learning_rate": 3.9718256516362707e-07,
      "loss": 0.4526,
      "step": 401320
    },
    {
      "epoch": 164.28,
      "grad_norm": 1.8229199647903442,
      "learning_rate": 3.970942183909264e-07,
      "loss": 0.4535,
      "step": 401330
    },
    {
      "epoch": 164.28,
      "grad_norm": 2.015718936920166,
      "learning_rate": 3.9700588053555407e-07,
      "loss": 0.4633,
      "step": 401340
    },
    {
      "epoch": 164.29,
      "grad_norm": 1.6867167949676514,
      "learning_rate": 3.9691755159791526e-07,
      "loss": 0.4548,
      "step": 401350
    },
    {
      "epoch": 164.29,
      "grad_norm": 1.8934478759765625,
      "learning_rate": 3.968292315784145e-07,
      "loss": 0.4611,
      "step": 401360
    },
    {
      "epoch": 164.29,
      "grad_norm": 2.2591638565063477,
      "learning_rate": 3.9674092047745547e-07,
      "loss": 0.4396,
      "step": 401370
    },
    {
      "epoch": 164.3,
      "grad_norm": 2.1240148544311523,
      "learning_rate": 3.96652618295444e-07,
      "loss": 0.4371,
      "step": 401380
    },
    {
      "epoch": 164.3,
      "grad_norm": 2.045671224594116,
      "learning_rate": 3.9656432503278427e-07,
      "loss": 0.4443,
      "step": 401390
    },
    {
      "epoch": 164.31,
      "grad_norm": 2.03753399848938,
      "learning_rate": 3.9647604068988064e-07,
      "loss": 0.454,
      "step": 401400
    },
    {
      "epoch": 164.31,
      "grad_norm": 2.7406790256500244,
      "learning_rate": 3.9638776526713716e-07,
      "loss": 0.4653,
      "step": 401410
    },
    {
      "epoch": 164.31,
      "grad_norm": 2.1772358417510986,
      "learning_rate": 3.962994987649588e-07,
      "loss": 0.454,
      "step": 401420
    },
    {
      "epoch": 164.32,
      "grad_norm": 2.139598846435547,
      "learning_rate": 3.962112411837498e-07,
      "loss": 0.4499,
      "step": 401430
    },
    {
      "epoch": 164.32,
      "grad_norm": 2.133988618850708,
      "learning_rate": 3.961229925239141e-07,
      "loss": 0.4384,
      "step": 401440
    },
    {
      "epoch": 164.33,
      "grad_norm": 1.8781393766403198,
      "learning_rate": 3.960347527858558e-07,
      "loss": 0.4569,
      "step": 401450
    },
    {
      "epoch": 164.33,
      "grad_norm": 2.3345956802368164,
      "learning_rate": 3.959465219699795e-07,
      "loss": 0.4722,
      "step": 401460
    },
    {
      "epoch": 164.33,
      "grad_norm": 2.406731605529785,
      "learning_rate": 3.9585830007668907e-07,
      "loss": 0.459,
      "step": 401470
    },
    {
      "epoch": 164.34,
      "grad_norm": 1.9989296197891235,
      "learning_rate": 3.9577008710638903e-07,
      "loss": 0.4416,
      "step": 401480
    },
    {
      "epoch": 164.34,
      "grad_norm": 2.02213454246521,
      "learning_rate": 3.9568188305948337e-07,
      "loss": 0.4495,
      "step": 401490
    },
    {
      "epoch": 164.35,
      "grad_norm": 1.786301612854004,
      "learning_rate": 3.9559368793637586e-07,
      "loss": 0.4413,
      "step": 401500
    },
    {
      "epoch": 164.35,
      "grad_norm": 5.163729667663574,
      "learning_rate": 3.955055017374709e-07,
      "loss": 0.4426,
      "step": 401510
    },
    {
      "epoch": 164.36,
      "grad_norm": 2.120901584625244,
      "learning_rate": 3.9541732446317224e-07,
      "loss": 0.4521,
      "step": 401520
    },
    {
      "epoch": 164.36,
      "grad_norm": 2.254850149154663,
      "learning_rate": 3.9532915611388383e-07,
      "loss": 0.4617,
      "step": 401530
    },
    {
      "epoch": 164.36,
      "grad_norm": 1.709132432937622,
      "learning_rate": 3.952409966900095e-07,
      "loss": 0.4244,
      "step": 401540
    },
    {
      "epoch": 164.37,
      "grad_norm": 2.3763961791992188,
      "learning_rate": 3.95152846191953e-07,
      "loss": 0.4188,
      "step": 401550
    },
    {
      "epoch": 164.37,
      "grad_norm": 1.9082711935043335,
      "learning_rate": 3.9506470462011775e-07,
      "loss": 0.4244,
      "step": 401560
    },
    {
      "epoch": 164.38,
      "grad_norm": 2.4172658920288086,
      "learning_rate": 3.949765719749094e-07,
      "loss": 0.4372,
      "step": 401570
    },
    {
      "epoch": 164.38,
      "grad_norm": 1.9328887462615967,
      "learning_rate": 3.948884482567297e-07,
      "loss": 0.4456,
      "step": 401580
    },
    {
      "epoch": 164.38,
      "grad_norm": 1.8370361328125,
      "learning_rate": 3.94800333465983e-07,
      "loss": 0.4362,
      "step": 401590
    },
    {
      "epoch": 164.39,
      "grad_norm": 2.308882474899292,
      "learning_rate": 3.9471222760307263e-07,
      "loss": 0.4668,
      "step": 401600
    },
    {
      "epoch": 164.39,
      "grad_norm": 2.1179263591766357,
      "learning_rate": 3.946241306684027e-07,
      "loss": 0.4524,
      "step": 401610
    },
    {
      "epoch": 164.4,
      "grad_norm": 1.9511319398880005,
      "learning_rate": 3.945360426623763e-07,
      "loss": 0.4396,
      "step": 401620
    },
    {
      "epoch": 164.4,
      "grad_norm": 2.079087972640991,
      "learning_rate": 3.9444796358539724e-07,
      "loss": 0.4576,
      "step": 401630
    },
    {
      "epoch": 164.4,
      "grad_norm": 2.218167781829834,
      "learning_rate": 3.943598934378691e-07,
      "loss": 0.4349,
      "step": 401640
    },
    {
      "epoch": 164.41,
      "grad_norm": 2.905977249145508,
      "learning_rate": 3.942718322201949e-07,
      "loss": 0.4347,
      "step": 401650
    },
    {
      "epoch": 164.41,
      "grad_norm": 1.7363464832305908,
      "learning_rate": 3.9418377993277835e-07,
      "loss": 0.4353,
      "step": 401660
    },
    {
      "epoch": 164.42,
      "grad_norm": 2.014481782913208,
      "learning_rate": 3.940957365760229e-07,
      "loss": 0.4616,
      "step": 401670
    },
    {
      "epoch": 164.42,
      "grad_norm": 1.9300637245178223,
      "learning_rate": 3.9400770215033145e-07,
      "loss": 0.4331,
      "step": 401680
    },
    {
      "epoch": 164.42,
      "grad_norm": 2.0341076850891113,
      "learning_rate": 3.9391967665610755e-07,
      "loss": 0.4266,
      "step": 401690
    },
    {
      "epoch": 164.43,
      "grad_norm": 2.0173709392547607,
      "learning_rate": 3.938316600937545e-07,
      "loss": 0.4539,
      "step": 401700
    },
    {
      "epoch": 164.43,
      "grad_norm": 1.9010614156723022,
      "learning_rate": 3.9374365246367535e-07,
      "loss": 0.4356,
      "step": 401710
    },
    {
      "epoch": 164.44,
      "grad_norm": 2.2124013900756836,
      "learning_rate": 3.936556537662732e-07,
      "loss": 0.4144,
      "step": 401720
    },
    {
      "epoch": 164.44,
      "grad_norm": 2.0953071117401123,
      "learning_rate": 3.9356766400195145e-07,
      "loss": 0.4354,
      "step": 401730
    },
    {
      "epoch": 164.45,
      "grad_norm": 3.9581663608551025,
      "learning_rate": 3.9347968317111327e-07,
      "loss": 0.4357,
      "step": 401740
    },
    {
      "epoch": 164.45,
      "grad_norm": 2.373643159866333,
      "learning_rate": 3.9339171127416023e-07,
      "loss": 0.4459,
      "step": 401750
    },
    {
      "epoch": 164.45,
      "grad_norm": 2.178880214691162,
      "learning_rate": 3.9330374831149726e-07,
      "loss": 0.4406,
      "step": 401760
    },
    {
      "epoch": 164.46,
      "grad_norm": 2.2311365604400635,
      "learning_rate": 3.932157942835263e-07,
      "loss": 0.4636,
      "step": 401770
    },
    {
      "epoch": 164.46,
      "grad_norm": 1.7735673189163208,
      "learning_rate": 3.9312784919065045e-07,
      "loss": 0.4344,
      "step": 401780
    },
    {
      "epoch": 164.47,
      "grad_norm": 1.9377516508102417,
      "learning_rate": 3.930399130332728e-07,
      "loss": 0.4517,
      "step": 401790
    },
    {
      "epoch": 164.47,
      "grad_norm": 2.132725238800049,
      "learning_rate": 3.9295198581179583e-07,
      "loss": 0.426,
      "step": 401800
    },
    {
      "epoch": 164.47,
      "grad_norm": 2.111725330352783,
      "learning_rate": 3.9286406752662243e-07,
      "loss": 0.4348,
      "step": 401810
    },
    {
      "epoch": 164.48,
      "grad_norm": 1.6635860204696655,
      "learning_rate": 3.927761581781553e-07,
      "loss": 0.4457,
      "step": 401820
    },
    {
      "epoch": 164.48,
      "grad_norm": 1.9303123950958252,
      "learning_rate": 3.9268825776679736e-07,
      "loss": 0.4396,
      "step": 401830
    },
    {
      "epoch": 164.49,
      "grad_norm": 2.1109840869903564,
      "learning_rate": 3.92600366292951e-07,
      "loss": 0.4331,
      "step": 401840
    },
    {
      "epoch": 164.49,
      "grad_norm": 2.4611241817474365,
      "learning_rate": 3.9251248375701893e-07,
      "loss": 0.4574,
      "step": 401850
    },
    {
      "epoch": 164.49,
      "grad_norm": 2.0261125564575195,
      "learning_rate": 3.924246101594036e-07,
      "loss": 0.4303,
      "step": 401860
    },
    {
      "epoch": 164.5,
      "grad_norm": 1.6307963132858276,
      "learning_rate": 3.9233674550050783e-07,
      "loss": 0.4471,
      "step": 401870
    },
    {
      "epoch": 164.5,
      "grad_norm": 2.07515549659729,
      "learning_rate": 3.9224888978073465e-07,
      "loss": 0.4398,
      "step": 401880
    },
    {
      "epoch": 164.51,
      "grad_norm": 2.6309382915496826,
      "learning_rate": 3.9216104300048504e-07,
      "loss": 0.4416,
      "step": 401890
    },
    {
      "epoch": 164.51,
      "grad_norm": 1.8140647411346436,
      "learning_rate": 3.920732051601624e-07,
      "loss": 0.4342,
      "step": 401900
    },
    {
      "epoch": 164.51,
      "grad_norm": 1.7878402471542358,
      "learning_rate": 3.919853762601687e-07,
      "loss": 0.4455,
      "step": 401910
    },
    {
      "epoch": 164.52,
      "grad_norm": 2.112234354019165,
      "learning_rate": 3.9189755630090653e-07,
      "loss": 0.4357,
      "step": 401920
    },
    {
      "epoch": 164.52,
      "grad_norm": 2.0398221015930176,
      "learning_rate": 3.9180974528277807e-07,
      "loss": 0.4656,
      "step": 401930
    },
    {
      "epoch": 164.53,
      "grad_norm": 2.3508987426757812,
      "learning_rate": 3.9172194320618515e-07,
      "loss": 0.4616,
      "step": 401940
    },
    {
      "epoch": 164.53,
      "grad_norm": 2.003748893737793,
      "learning_rate": 3.916341500715311e-07,
      "loss": 0.4288,
      "step": 401950
    },
    {
      "epoch": 164.54,
      "grad_norm": 2.4452638626098633,
      "learning_rate": 3.915463658792174e-07,
      "loss": 0.4486,
      "step": 401960
    },
    {
      "epoch": 164.54,
      "grad_norm": 1.9092694520950317,
      "learning_rate": 3.9145859062964613e-07,
      "loss": 0.4392,
      "step": 401970
    },
    {
      "epoch": 164.54,
      "grad_norm": 1.9709250926971436,
      "learning_rate": 3.913708243232195e-07,
      "loss": 0.4478,
      "step": 401980
    },
    {
      "epoch": 164.55,
      "grad_norm": 2.211646318435669,
      "learning_rate": 3.912830669603395e-07,
      "loss": 0.4449,
      "step": 401990
    },
    {
      "epoch": 164.55,
      "grad_norm": 2.092451572418213,
      "learning_rate": 3.9119531854140797e-07,
      "loss": 0.4709,
      "step": 402000
    },
    {
      "epoch": 164.56,
      "grad_norm": 1.7496817111968994,
      "learning_rate": 3.9110757906682795e-07,
      "loss": 0.4443,
      "step": 402010
    },
    {
      "epoch": 164.56,
      "grad_norm": 1.9095509052276611,
      "learning_rate": 3.9101984853699967e-07,
      "loss": 0.4738,
      "step": 402020
    },
    {
      "epoch": 164.56,
      "grad_norm": 1.9321759939193726,
      "learning_rate": 3.9093212695232587e-07,
      "loss": 0.4445,
      "step": 402030
    },
    {
      "epoch": 164.57,
      "grad_norm": 2.0223255157470703,
      "learning_rate": 3.908444143132081e-07,
      "loss": 0.4367,
      "step": 402040
    },
    {
      "epoch": 164.57,
      "grad_norm": 2.06888747215271,
      "learning_rate": 3.907567106200486e-07,
      "loss": 0.4509,
      "step": 402050
    },
    {
      "epoch": 164.58,
      "grad_norm": 2.180758476257324,
      "learning_rate": 3.9066901587324895e-07,
      "loss": 0.4243,
      "step": 402060
    },
    {
      "epoch": 164.58,
      "grad_norm": 1.8925342559814453,
      "learning_rate": 3.905813300732106e-07,
      "loss": 0.4467,
      "step": 402070
    },
    {
      "epoch": 164.58,
      "grad_norm": 2.707260847091675,
      "learning_rate": 3.904936532203358e-07,
      "loss": 0.4586,
      "step": 402080
    },
    {
      "epoch": 164.59,
      "grad_norm": 1.8570667505264282,
      "learning_rate": 3.9040598531502554e-07,
      "loss": 0.4367,
      "step": 402090
    },
    {
      "epoch": 164.59,
      "grad_norm": 1.7536758184432983,
      "learning_rate": 3.9031832635768166e-07,
      "loss": 0.4474,
      "step": 402100
    },
    {
      "epoch": 164.6,
      "grad_norm": 1.988343596458435,
      "learning_rate": 3.902306763487058e-07,
      "loss": 0.4573,
      "step": 402110
    },
    {
      "epoch": 164.6,
      "grad_norm": 2.0141665935516357,
      "learning_rate": 3.9014303528849934e-07,
      "loss": 0.4705,
      "step": 402120
    },
    {
      "epoch": 164.6,
      "grad_norm": 2.1690659523010254,
      "learning_rate": 3.9005540317746345e-07,
      "loss": 0.4564,
      "step": 402130
    },
    {
      "epoch": 164.61,
      "grad_norm": 2.2466132640838623,
      "learning_rate": 3.899677800160004e-07,
      "loss": 0.4514,
      "step": 402140
    },
    {
      "epoch": 164.61,
      "grad_norm": 1.91317880153656,
      "learning_rate": 3.898801658045119e-07,
      "loss": 0.4585,
      "step": 402150
    },
    {
      "epoch": 164.62,
      "grad_norm": 2.1438539028167725,
      "learning_rate": 3.8979256054339776e-07,
      "loss": 0.4504,
      "step": 402160
    },
    {
      "epoch": 164.62,
      "grad_norm": 2.9408512115478516,
      "learning_rate": 3.8970496423305996e-07,
      "loss": 0.4462,
      "step": 402170
    },
    {
      "epoch": 164.63,
      "grad_norm": 2.3628485202789307,
      "learning_rate": 3.8961737687389996e-07,
      "loss": 0.4472,
      "step": 402180
    },
    {
      "epoch": 164.63,
      "grad_norm": 2.4651479721069336,
      "learning_rate": 3.895297984663187e-07,
      "loss": 0.4379,
      "step": 402190
    },
    {
      "epoch": 164.63,
      "grad_norm": 2.193458080291748,
      "learning_rate": 3.8944222901071784e-07,
      "loss": 0.4508,
      "step": 402200
    },
    {
      "epoch": 164.64,
      "grad_norm": 1.8182920217514038,
      "learning_rate": 3.8935466850749796e-07,
      "loss": 0.428,
      "step": 402210
    },
    {
      "epoch": 164.64,
      "grad_norm": 1.876000165939331,
      "learning_rate": 3.892671169570606e-07,
      "loss": 0.4477,
      "step": 402220
    },
    {
      "epoch": 164.65,
      "grad_norm": 2.3259036540985107,
      "learning_rate": 3.8917957435980645e-07,
      "loss": 0.4402,
      "step": 402230
    },
    {
      "epoch": 164.65,
      "grad_norm": 1.987339973449707,
      "learning_rate": 3.8909204071613686e-07,
      "loss": 0.4439,
      "step": 402240
    },
    {
      "epoch": 164.65,
      "grad_norm": 1.877065896987915,
      "learning_rate": 3.890045160264527e-07,
      "loss": 0.4525,
      "step": 402250
    },
    {
      "epoch": 164.66,
      "grad_norm": 1.9731546640396118,
      "learning_rate": 3.8891700029115475e-07,
      "loss": 0.4672,
      "step": 402260
    },
    {
      "epoch": 164.66,
      "grad_norm": 2.2431516647338867,
      "learning_rate": 3.888294935106442e-07,
      "loss": 0.4542,
      "step": 402270
    },
    {
      "epoch": 164.67,
      "grad_norm": 1.9034532308578491,
      "learning_rate": 3.8874199568532156e-07,
      "loss": 0.4528,
      "step": 402280
    },
    {
      "epoch": 164.67,
      "grad_norm": 2.253455400466919,
      "learning_rate": 3.886545068155878e-07,
      "loss": 0.463,
      "step": 402290
    },
    {
      "epoch": 164.67,
      "grad_norm": 1.8054715394973755,
      "learning_rate": 3.8856702690184366e-07,
      "loss": 0.4496,
      "step": 402300
    },
    {
      "epoch": 164.68,
      "grad_norm": 1.9828028678894043,
      "learning_rate": 3.8847955594449004e-07,
      "loss": 0.4613,
      "step": 402310
    },
    {
      "epoch": 164.68,
      "grad_norm": 1.9376533031463623,
      "learning_rate": 3.883920939439275e-07,
      "loss": 0.4357,
      "step": 402320
    },
    {
      "epoch": 164.69,
      "grad_norm": 2.1257779598236084,
      "learning_rate": 3.883046409005567e-07,
      "loss": 0.4372,
      "step": 402330
    },
    {
      "epoch": 164.69,
      "grad_norm": 2.037593364715576,
      "learning_rate": 3.882171968147784e-07,
      "loss": 0.4527,
      "step": 402340
    },
    {
      "epoch": 164.7,
      "grad_norm": 1.9968388080596924,
      "learning_rate": 3.881297616869928e-07,
      "loss": 0.4306,
      "step": 402350
    },
    {
      "epoch": 164.7,
      "grad_norm": 1.965275764465332,
      "learning_rate": 3.880423355176008e-07,
      "loss": 0.4438,
      "step": 402360
    },
    {
      "epoch": 164.7,
      "grad_norm": 1.9858378171920776,
      "learning_rate": 3.8795491830700273e-07,
      "loss": 0.4105,
      "step": 402370
    },
    {
      "epoch": 164.71,
      "grad_norm": 2.118748188018799,
      "learning_rate": 3.8786751005559894e-07,
      "loss": 0.4361,
      "step": 402380
    },
    {
      "epoch": 164.71,
      "grad_norm": 1.7659720182418823,
      "learning_rate": 3.877801107637899e-07,
      "loss": 0.4427,
      "step": 402390
    },
    {
      "epoch": 164.72,
      "grad_norm": 2.4308226108551025,
      "learning_rate": 3.876927204319759e-07,
      "loss": 0.454,
      "step": 402400
    },
    {
      "epoch": 164.72,
      "grad_norm": 2.224175453186035,
      "learning_rate": 3.876053390605576e-07,
      "loss": 0.4469,
      "step": 402410
    },
    {
      "epoch": 164.72,
      "grad_norm": 1.639681100845337,
      "learning_rate": 3.8751796664993494e-07,
      "loss": 0.4416,
      "step": 402420
    },
    {
      "epoch": 164.73,
      "grad_norm": 1.8812495470046997,
      "learning_rate": 3.874306032005084e-07,
      "loss": 0.4474,
      "step": 402430
    },
    {
      "epoch": 164.73,
      "grad_norm": 1.9857884645462036,
      "learning_rate": 3.873432487126779e-07,
      "loss": 0.4424,
      "step": 402440
    },
    {
      "epoch": 164.74,
      "grad_norm": 1.834669828414917,
      "learning_rate": 3.8725590318684367e-07,
      "loss": 0.4311,
      "step": 402450
    },
    {
      "epoch": 164.74,
      "grad_norm": 2.038809299468994,
      "learning_rate": 3.8716856662340665e-07,
      "loss": 0.43,
      "step": 402460
    },
    {
      "epoch": 164.74,
      "grad_norm": 2.107311725616455,
      "learning_rate": 3.870812390227655e-07,
      "loss": 0.4578,
      "step": 402470
    },
    {
      "epoch": 164.75,
      "grad_norm": 2.209991455078125,
      "learning_rate": 3.8699392038532093e-07,
      "loss": 0.4288,
      "step": 402480
    },
    {
      "epoch": 164.75,
      "grad_norm": 1.9281173944473267,
      "learning_rate": 3.869066107114729e-07,
      "loss": 0.4473,
      "step": 402490
    },
    {
      "epoch": 164.76,
      "grad_norm": 2.2236669063568115,
      "learning_rate": 3.8681931000162085e-07,
      "loss": 0.4476,
      "step": 402500
    },
    {
      "epoch": 164.76,
      "grad_norm": 2.1833252906799316,
      "learning_rate": 3.8673201825616567e-07,
      "loss": 0.4303,
      "step": 402510
    },
    {
      "epoch": 164.76,
      "grad_norm": 2.1120824813842773,
      "learning_rate": 3.866447354755069e-07,
      "loss": 0.4692,
      "step": 402520
    },
    {
      "epoch": 164.77,
      "grad_norm": 1.9176164865493774,
      "learning_rate": 3.8655746166004443e-07,
      "loss": 0.4377,
      "step": 402530
    },
    {
      "epoch": 164.77,
      "grad_norm": 1.7846182584762573,
      "learning_rate": 3.864701968101776e-07,
      "loss": 0.4541,
      "step": 402540
    },
    {
      "epoch": 164.78,
      "grad_norm": 1.8686413764953613,
      "learning_rate": 3.863829409263065e-07,
      "loss": 0.429,
      "step": 402550
    },
    {
      "epoch": 164.78,
      "grad_norm": 2.1456027030944824,
      "learning_rate": 3.862956940088307e-07,
      "loss": 0.4379,
      "step": 402560
    },
    {
      "epoch": 164.79,
      "grad_norm": 1.9679518938064575,
      "learning_rate": 3.8620845605814995e-07,
      "loss": 0.4752,
      "step": 402570
    },
    {
      "epoch": 164.79,
      "grad_norm": 1.7594026327133179,
      "learning_rate": 3.8612122707466386e-07,
      "loss": 0.442,
      "step": 402580
    },
    {
      "epoch": 164.79,
      "grad_norm": 2.3059167861938477,
      "learning_rate": 3.860340070587721e-07,
      "loss": 0.4554,
      "step": 402590
    },
    {
      "epoch": 164.8,
      "grad_norm": 2.1334400177001953,
      "learning_rate": 3.859467960108745e-07,
      "loss": 0.4645,
      "step": 402600
    },
    {
      "epoch": 164.8,
      "grad_norm": 2.371312141418457,
      "learning_rate": 3.858595939313698e-07,
      "loss": 0.4488,
      "step": 402610
    },
    {
      "epoch": 164.81,
      "grad_norm": 2.1512603759765625,
      "learning_rate": 3.8577240082065773e-07,
      "loss": 0.4506,
      "step": 402620
    },
    {
      "epoch": 164.81,
      "grad_norm": 2.280136823654175,
      "learning_rate": 3.8568521667913775e-07,
      "loss": 0.4298,
      "step": 402630
    },
    {
      "epoch": 164.81,
      "grad_norm": 1.8665492534637451,
      "learning_rate": 3.8559804150720947e-07,
      "loss": 0.434,
      "step": 402640
    },
    {
      "epoch": 164.82,
      "grad_norm": 1.746703863143921,
      "learning_rate": 3.855108753052718e-07,
      "loss": 0.4634,
      "step": 402650
    },
    {
      "epoch": 164.82,
      "grad_norm": 1.874043583869934,
      "learning_rate": 3.854237180737245e-07,
      "loss": 0.459,
      "step": 402660
    },
    {
      "epoch": 164.83,
      "grad_norm": 2.1007418632507324,
      "learning_rate": 3.853365698129667e-07,
      "loss": 0.4433,
      "step": 402670
    },
    {
      "epoch": 164.83,
      "grad_norm": 1.758469581604004,
      "learning_rate": 3.852494305233972e-07,
      "loss": 0.4603,
      "step": 402680
    },
    {
      "epoch": 164.83,
      "grad_norm": 2.043039083480835,
      "learning_rate": 3.8516230020541525e-07,
      "loss": 0.46,
      "step": 402690
    },
    {
      "epoch": 164.84,
      "grad_norm": 1.7783422470092773,
      "learning_rate": 3.850751788594206e-07,
      "loss": 0.4527,
      "step": 402700
    },
    {
      "epoch": 164.84,
      "grad_norm": 2.0881714820861816,
      "learning_rate": 3.8498806648581197e-07,
      "loss": 0.4393,
      "step": 402710
    },
    {
      "epoch": 164.85,
      "grad_norm": 1.8515434265136719,
      "learning_rate": 3.8490096308498865e-07,
      "loss": 0.4501,
      "step": 402720
    },
    {
      "epoch": 164.85,
      "grad_norm": 1.9854106903076172,
      "learning_rate": 3.8481386865734973e-07,
      "loss": 0.4406,
      "step": 402730
    },
    {
      "epoch": 164.85,
      "grad_norm": 1.8631236553192139,
      "learning_rate": 3.8472678320329316e-07,
      "loss": 0.4519,
      "step": 402740
    },
    {
      "epoch": 164.86,
      "grad_norm": 2.0298690795898438,
      "learning_rate": 3.846397067232188e-07,
      "loss": 0.4519,
      "step": 402750
    },
    {
      "epoch": 164.86,
      "grad_norm": 2.338150978088379,
      "learning_rate": 3.845526392175253e-07,
      "loss": 0.4287,
      "step": 402760
    },
    {
      "epoch": 164.87,
      "grad_norm": 2.548140287399292,
      "learning_rate": 3.844655806866113e-07,
      "loss": 0.4344,
      "step": 402770
    },
    {
      "epoch": 164.87,
      "grad_norm": 2.3944852352142334,
      "learning_rate": 3.8437853113087577e-07,
      "loss": 0.447,
      "step": 402780
    },
    {
      "epoch": 164.88,
      "grad_norm": 1.7772125005722046,
      "learning_rate": 3.842914905507178e-07,
      "loss": 0.4499,
      "step": 402790
    },
    {
      "epoch": 164.88,
      "grad_norm": 1.8157914876937866,
      "learning_rate": 3.842044589465355e-07,
      "loss": 0.4369,
      "step": 402800
    },
    {
      "epoch": 164.88,
      "grad_norm": 2.1997365951538086,
      "learning_rate": 3.8411743631872806e-07,
      "loss": 0.4345,
      "step": 402810
    },
    {
      "epoch": 164.89,
      "grad_norm": 1.857382893562317,
      "learning_rate": 3.840304226676938e-07,
      "loss": 0.4436,
      "step": 402820
    },
    {
      "epoch": 164.89,
      "grad_norm": 2.0818943977355957,
      "learning_rate": 3.839434179938315e-07,
      "loss": 0.4504,
      "step": 402830
    },
    {
      "epoch": 164.9,
      "grad_norm": 2.377150535583496,
      "learning_rate": 3.8385642229753937e-07,
      "loss": 0.4582,
      "step": 402840
    },
    {
      "epoch": 164.9,
      "grad_norm": 2.249274730682373,
      "learning_rate": 3.8376943557921645e-07,
      "loss": 0.4268,
      "step": 402850
    },
    {
      "epoch": 164.9,
      "grad_norm": 1.8927571773529053,
      "learning_rate": 3.83682457839261e-07,
      "loss": 0.4706,
      "step": 402860
    },
    {
      "epoch": 164.91,
      "grad_norm": 1.8137280941009521,
      "learning_rate": 3.8359548907807125e-07,
      "loss": 0.4373,
      "step": 402870
    },
    {
      "epoch": 164.91,
      "grad_norm": 2.136181354522705,
      "learning_rate": 3.835085292960458e-07,
      "loss": 0.4356,
      "step": 402880
    },
    {
      "epoch": 164.92,
      "grad_norm": 2.2225258350372314,
      "learning_rate": 3.8342157849358297e-07,
      "loss": 0.4398,
      "step": 402890
    },
    {
      "epoch": 164.92,
      "grad_norm": 1.9219858646392822,
      "learning_rate": 3.833346366710809e-07,
      "loss": 0.4558,
      "step": 402900
    },
    {
      "epoch": 164.92,
      "grad_norm": 1.8588787317276,
      "learning_rate": 3.8324770382893807e-07,
      "loss": 0.4415,
      "step": 402910
    },
    {
      "epoch": 164.93,
      "grad_norm": 2.1630241870880127,
      "learning_rate": 3.831607799675528e-07,
      "loss": 0.4576,
      "step": 402920
    },
    {
      "epoch": 164.93,
      "grad_norm": 2.0315310955047607,
      "learning_rate": 3.830738650873231e-07,
      "loss": 0.4467,
      "step": 402930
    },
    {
      "epoch": 164.94,
      "grad_norm": 1.8384817838668823,
      "learning_rate": 3.82986959188647e-07,
      "loss": 0.4571,
      "step": 402940
    },
    {
      "epoch": 164.94,
      "grad_norm": 6.79097843170166,
      "learning_rate": 3.8290006227192276e-07,
      "loss": 0.4401,
      "step": 402950
    },
    {
      "epoch": 164.94,
      "grad_norm": 2.882514715194702,
      "learning_rate": 3.8281317433754824e-07,
      "loss": 0.427,
      "step": 402960
    },
    {
      "epoch": 164.95,
      "grad_norm": 1.8626258373260498,
      "learning_rate": 3.827262953859219e-07,
      "loss": 0.4387,
      "step": 402970
    },
    {
      "epoch": 164.95,
      "grad_norm": 1.8441014289855957,
      "learning_rate": 3.826394254174413e-07,
      "loss": 0.4459,
      "step": 402980
    },
    {
      "epoch": 164.96,
      "grad_norm": 2.2565369606018066,
      "learning_rate": 3.825525644325046e-07,
      "loss": 0.4402,
      "step": 402990
    },
    {
      "epoch": 164.96,
      "grad_norm": 1.8392332792282104,
      "learning_rate": 3.8246571243150957e-07,
      "loss": 0.4378,
      "step": 403000
    },
    {
      "epoch": 164.97,
      "grad_norm": 2.3171980381011963,
      "learning_rate": 3.823788694148541e-07,
      "loss": 0.4381,
      "step": 403010
    },
    {
      "epoch": 164.97,
      "grad_norm": 2.5197079181671143,
      "learning_rate": 3.822920353829361e-07,
      "loss": 0.4694,
      "step": 403020
    },
    {
      "epoch": 164.97,
      "grad_norm": 2.6674811840057373,
      "learning_rate": 3.822052103361531e-07,
      "loss": 0.4488,
      "step": 403030
    },
    {
      "epoch": 164.98,
      "grad_norm": 2.3245580196380615,
      "learning_rate": 3.821183942749038e-07,
      "loss": 0.4516,
      "step": 403040
    },
    {
      "epoch": 164.98,
      "grad_norm": 2.063718318939209,
      "learning_rate": 3.8203158719958444e-07,
      "loss": 0.4347,
      "step": 403050
    },
    {
      "epoch": 164.99,
      "grad_norm": 2.084639549255371,
      "learning_rate": 3.8194478911059317e-07,
      "loss": 0.4578,
      "step": 403060
    },
    {
      "epoch": 164.99,
      "grad_norm": 1.926239252090454,
      "learning_rate": 3.8185800000832734e-07,
      "loss": 0.4575,
      "step": 403070
    },
    {
      "epoch": 164.99,
      "grad_norm": 1.9634153842926025,
      "learning_rate": 3.817712198931853e-07,
      "loss": 0.4231,
      "step": 403080
    },
    {
      "epoch": 165.0,
      "grad_norm": 1.8665825128555298,
      "learning_rate": 3.816844487655645e-07,
      "loss": 0.4376,
      "step": 403090
    },
    {
      "epoch": 165.0,
      "eval_loss": 0.4454517960548401,
      "eval_runtime": 52.8805,
      "eval_samples_per_second": 65.223,
      "eval_steps_per_second": 8.169,
      "step": 403095
    },
    {
      "epoch": 165.0,
      "grad_norm": 1.8692525625228882,
      "learning_rate": 3.8159768662586206e-07,
      "loss": 0.4583,
      "step": 403100
    },
    {
      "epoch": 165.01,
      "grad_norm": 2.2083089351654053,
      "learning_rate": 3.815109334744752e-07,
      "loss": 0.4301,
      "step": 403110
    },
    {
      "epoch": 165.01,
      "grad_norm": 2.184370517730713,
      "learning_rate": 3.814241893118018e-07,
      "loss": 0.4488,
      "step": 403120
    },
    {
      "epoch": 165.01,
      "grad_norm": 2.186823606491089,
      "learning_rate": 3.8133745413823914e-07,
      "loss": 0.4389,
      "step": 403130
    },
    {
      "epoch": 165.02,
      "grad_norm": 1.8747131824493408,
      "learning_rate": 3.812507279541843e-07,
      "loss": 0.4393,
      "step": 403140
    },
    {
      "epoch": 165.02,
      "grad_norm": 2.7313473224639893,
      "learning_rate": 3.8116401076003493e-07,
      "loss": 0.4607,
      "step": 403150
    },
    {
      "epoch": 165.03,
      "grad_norm": 2.0079989433288574,
      "learning_rate": 3.8107730255618775e-07,
      "loss": 0.4864,
      "step": 403160
    },
    {
      "epoch": 165.03,
      "grad_norm": 2.7724366188049316,
      "learning_rate": 3.8099060334304023e-07,
      "loss": 0.4509,
      "step": 403170
    },
    {
      "epoch": 165.03,
      "grad_norm": 2.224421977996826,
      "learning_rate": 3.809039131209904e-07,
      "loss": 0.446,
      "step": 403180
    },
    {
      "epoch": 165.04,
      "grad_norm": 3.0799028873443604,
      "learning_rate": 3.8081723189043354e-07,
      "loss": 0.4423,
      "step": 403190
    },
    {
      "epoch": 165.04,
      "grad_norm": 2.1439852714538574,
      "learning_rate": 3.80730559651768e-07,
      "loss": 0.4447,
      "step": 403200
    },
    {
      "epoch": 165.05,
      "grad_norm": 2.290236473083496,
      "learning_rate": 3.806438964053904e-07,
      "loss": 0.4435,
      "step": 403210
    },
    {
      "epoch": 165.05,
      "grad_norm": 1.8172900676727295,
      "learning_rate": 3.8055724215169774e-07,
      "loss": 0.4568,
      "step": 403220
    },
    {
      "epoch": 165.06,
      "grad_norm": 1.875073790550232,
      "learning_rate": 3.8047059689108714e-07,
      "loss": 0.4478,
      "step": 403230
    },
    {
      "epoch": 165.06,
      "grad_norm": 2.1854195594787598,
      "learning_rate": 3.803839606239555e-07,
      "loss": 0.4665,
      "step": 403240
    },
    {
      "epoch": 165.06,
      "grad_norm": 2.0986437797546387,
      "learning_rate": 3.802973333506997e-07,
      "loss": 0.4409,
      "step": 403250
    },
    {
      "epoch": 165.07,
      "grad_norm": 1.933170199394226,
      "learning_rate": 3.802107150717157e-07,
      "loss": 0.4473,
      "step": 403260
    },
    {
      "epoch": 165.07,
      "grad_norm": 1.76140296459198,
      "learning_rate": 3.8012410578740183e-07,
      "loss": 0.4652,
      "step": 403270
    },
    {
      "epoch": 165.08,
      "grad_norm": 1.959542989730835,
      "learning_rate": 3.800375054981538e-07,
      "loss": 0.4299,
      "step": 403280
    },
    {
      "epoch": 165.08,
      "grad_norm": 2.121112108230591,
      "learning_rate": 3.7995091420436877e-07,
      "loss": 0.4513,
      "step": 403290
    },
    {
      "epoch": 165.08,
      "grad_norm": 1.5937528610229492,
      "learning_rate": 3.7986433190644326e-07,
      "loss": 0.4361,
      "step": 403300
    },
    {
      "epoch": 165.09,
      "grad_norm": 2.312744140625,
      "learning_rate": 3.797777586047743e-07,
      "loss": 0.4277,
      "step": 403310
    },
    {
      "epoch": 165.09,
      "grad_norm": 2.3035459518432617,
      "learning_rate": 3.7969119429975757e-07,
      "loss": 0.4424,
      "step": 403320
    },
    {
      "epoch": 165.1,
      "grad_norm": 1.9930669069290161,
      "learning_rate": 3.796046389917902e-07,
      "loss": 0.4384,
      "step": 403330
    },
    {
      "epoch": 165.1,
      "grad_norm": 2.1095688343048096,
      "learning_rate": 3.795180926812684e-07,
      "loss": 0.4274,
      "step": 403340
    },
    {
      "epoch": 165.1,
      "grad_norm": 1.5716315507888794,
      "learning_rate": 3.7943155536858864e-07,
      "loss": 0.4517,
      "step": 403350
    },
    {
      "epoch": 165.11,
      "grad_norm": 1.7900872230529785,
      "learning_rate": 3.7934502705414746e-07,
      "loss": 0.4195,
      "step": 403360
    },
    {
      "epoch": 165.11,
      "grad_norm": 1.9333778619766235,
      "learning_rate": 3.792585077383415e-07,
      "loss": 0.4671,
      "step": 403370
    },
    {
      "epoch": 165.12,
      "grad_norm": 2.239849090576172,
      "learning_rate": 3.791719974215664e-07,
      "loss": 0.474,
      "step": 403380
    },
    {
      "epoch": 165.12,
      "grad_norm": 1.9792671203613281,
      "learning_rate": 3.7908549610421915e-07,
      "loss": 0.4492,
      "step": 403390
    },
    {
      "epoch": 165.12,
      "grad_norm": 1.9437347650527954,
      "learning_rate": 3.7899900378669564e-07,
      "loss": 0.4563,
      "step": 403400
    },
    {
      "epoch": 165.13,
      "grad_norm": 2.00042986869812,
      "learning_rate": 3.789125204693923e-07,
      "loss": 0.4551,
      "step": 403410
    },
    {
      "epoch": 165.13,
      "grad_norm": 2.281602382659912,
      "learning_rate": 3.7882604615270507e-07,
      "loss": 0.4774,
      "step": 403420
    },
    {
      "epoch": 165.14,
      "grad_norm": 1.9002766609191895,
      "learning_rate": 3.7873958083703026e-07,
      "loss": 0.4357,
      "step": 403430
    },
    {
      "epoch": 165.14,
      "grad_norm": 2.2055442333221436,
      "learning_rate": 3.7865312452276366e-07,
      "loss": 0.4538,
      "step": 403440
    },
    {
      "epoch": 165.15,
      "grad_norm": 2.021590232849121,
      "learning_rate": 3.785666772103018e-07,
      "loss": 0.4651,
      "step": 403450
    },
    {
      "epoch": 165.15,
      "grad_norm": 2.2611243724823,
      "learning_rate": 3.784802389000403e-07,
      "loss": 0.4321,
      "step": 403460
    },
    {
      "epoch": 165.15,
      "grad_norm": 1.999452829360962,
      "learning_rate": 3.783938095923751e-07,
      "loss": 0.4377,
      "step": 403470
    },
    {
      "epoch": 165.16,
      "grad_norm": 1.936792254447937,
      "learning_rate": 3.7830738928770227e-07,
      "loss": 0.4439,
      "step": 403480
    },
    {
      "epoch": 165.16,
      "grad_norm": 1.9316662549972534,
      "learning_rate": 3.782209779864175e-07,
      "loss": 0.4392,
      "step": 403490
    },
    {
      "epoch": 165.17,
      "grad_norm": 3.9835073947906494,
      "learning_rate": 3.7813457568891703e-07,
      "loss": 0.4372,
      "step": 403500
    },
    {
      "epoch": 165.17,
      "grad_norm": 1.8775821924209595,
      "learning_rate": 3.780481823955964e-07,
      "loss": 0.4702,
      "step": 403510
    },
    {
      "epoch": 165.17,
      "grad_norm": 1.8819268941879272,
      "learning_rate": 3.779617981068511e-07,
      "loss": 0.434,
      "step": 403520
    },
    {
      "epoch": 165.18,
      "grad_norm": 2.393019199371338,
      "learning_rate": 3.778754228230775e-07,
      "loss": 0.4678,
      "step": 403530
    },
    {
      "epoch": 165.18,
      "grad_norm": 2.1225175857543945,
      "learning_rate": 3.7778905654467063e-07,
      "loss": 0.4315,
      "step": 403540
    },
    {
      "epoch": 165.19,
      "grad_norm": 2.6529488563537598,
      "learning_rate": 3.777026992720263e-07,
      "loss": 0.4427,
      "step": 403550
    },
    {
      "epoch": 165.19,
      "grad_norm": 1.8420848846435547,
      "learning_rate": 3.7761635100554035e-07,
      "loss": 0.4368,
      "step": 403560
    },
    {
      "epoch": 165.19,
      "grad_norm": 2.1204235553741455,
      "learning_rate": 3.7753001174560796e-07,
      "loss": 0.4486,
      "step": 403570
    },
    {
      "epoch": 165.2,
      "grad_norm": 1.9499878883361816,
      "learning_rate": 3.774436814926252e-07,
      "loss": 0.4303,
      "step": 403580
    },
    {
      "epoch": 165.2,
      "grad_norm": 2.3302128314971924,
      "learning_rate": 3.7735736024698675e-07,
      "loss": 0.434,
      "step": 403590
    },
    {
      "epoch": 165.21,
      "grad_norm": 2.066377639770508,
      "learning_rate": 3.772710480090887e-07,
      "loss": 0.4613,
      "step": 403600
    },
    {
      "epoch": 165.21,
      "grad_norm": 2.1472268104553223,
      "learning_rate": 3.7718474477932605e-07,
      "loss": 0.4543,
      "step": 403610
    },
    {
      "epoch": 165.21,
      "grad_norm": 2.1804258823394775,
      "learning_rate": 3.770984505580951e-07,
      "loss": 0.4574,
      "step": 403620
    },
    {
      "epoch": 165.22,
      "grad_norm": 2.461280345916748,
      "learning_rate": 3.770121653457895e-07,
      "loss": 0.4328,
      "step": 403630
    },
    {
      "epoch": 165.22,
      "grad_norm": 1.997802495956421,
      "learning_rate": 3.7692588914280477e-07,
      "loss": 0.4376,
      "step": 403640
    },
    {
      "epoch": 165.23,
      "grad_norm": 2.053847074508667,
      "learning_rate": 3.7683962194953725e-07,
      "loss": 0.4509,
      "step": 403650
    },
    {
      "epoch": 165.23,
      "grad_norm": 2.002488851547241,
      "learning_rate": 3.7675336376638165e-07,
      "loss": 0.4477,
      "step": 403660
    },
    {
      "epoch": 165.24,
      "grad_norm": 1.8615710735321045,
      "learning_rate": 3.7666711459373294e-07,
      "loss": 0.4792,
      "step": 403670
    },
    {
      "epoch": 165.24,
      "grad_norm": 2.279693603515625,
      "learning_rate": 3.765808744319865e-07,
      "loss": 0.4343,
      "step": 403680
    },
    {
      "epoch": 165.24,
      "grad_norm": 2.3346474170684814,
      "learning_rate": 3.7649464328153693e-07,
      "loss": 0.4509,
      "step": 403690
    },
    {
      "epoch": 165.25,
      "grad_norm": 1.6469242572784424,
      "learning_rate": 3.764084211427795e-07,
      "loss": 0.4666,
      "step": 403700
    },
    {
      "epoch": 165.25,
      "grad_norm": 2.02167010307312,
      "learning_rate": 3.763222080161093e-07,
      "loss": 0.4442,
      "step": 403710
    },
    {
      "epoch": 165.26,
      "grad_norm": 2.0703847408294678,
      "learning_rate": 3.7623600390192094e-07,
      "loss": 0.4312,
      "step": 403720
    },
    {
      "epoch": 165.26,
      "grad_norm": 2.2011301517486572,
      "learning_rate": 3.7614980880060954e-07,
      "loss": 0.44,
      "step": 403730
    },
    {
      "epoch": 165.26,
      "grad_norm": 1.843377947807312,
      "learning_rate": 3.7606362271257e-07,
      "loss": 0.4212,
      "step": 403740
    },
    {
      "epoch": 165.27,
      "grad_norm": 2.2447919845581055,
      "learning_rate": 3.7597744563819687e-07,
      "loss": 0.4506,
      "step": 403750
    },
    {
      "epoch": 165.27,
      "grad_norm": 1.998767614364624,
      "learning_rate": 3.758912775778859e-07,
      "loss": 0.4624,
      "step": 403760
    },
    {
      "epoch": 165.28,
      "grad_norm": 2.291276693344116,
      "learning_rate": 3.7580511853203024e-07,
      "loss": 0.4831,
      "step": 403770
    },
    {
      "epoch": 165.28,
      "grad_norm": 2.125478982925415,
      "learning_rate": 3.7571896850102517e-07,
      "loss": 0.4582,
      "step": 403780
    },
    {
      "epoch": 165.28,
      "grad_norm": 2.157447576522827,
      "learning_rate": 3.7563282748526567e-07,
      "loss": 0.4349,
      "step": 403790
    },
    {
      "epoch": 165.29,
      "grad_norm": 2.0852441787719727,
      "learning_rate": 3.75546695485146e-07,
      "loss": 0.4498,
      "step": 403800
    },
    {
      "epoch": 165.29,
      "grad_norm": 2.7095446586608887,
      "learning_rate": 3.754605725010608e-07,
      "loss": 0.4292,
      "step": 403810
    },
    {
      "epoch": 165.3,
      "grad_norm": 1.9632681608200073,
      "learning_rate": 3.7537445853340436e-07,
      "loss": 0.4771,
      "step": 403820
    },
    {
      "epoch": 165.3,
      "grad_norm": 2.7384276390075684,
      "learning_rate": 3.752883535825717e-07,
      "loss": 0.4616,
      "step": 403830
    },
    {
      "epoch": 165.3,
      "grad_norm": 2.4758353233337402,
      "learning_rate": 3.7520225764895726e-07,
      "loss": 0.4162,
      "step": 403840
    },
    {
      "epoch": 165.31,
      "grad_norm": 2.618882179260254,
      "learning_rate": 3.75116170732955e-07,
      "loss": 0.454,
      "step": 403850
    },
    {
      "epoch": 165.31,
      "grad_norm": 1.6928856372833252,
      "learning_rate": 3.750300928349594e-07,
      "loss": 0.449,
      "step": 403860
    },
    {
      "epoch": 165.32,
      "grad_norm": 2.1143271923065186,
      "learning_rate": 3.749440239553649e-07,
      "loss": 0.4554,
      "step": 403870
    },
    {
      "epoch": 165.32,
      "grad_norm": 1.756693720817566,
      "learning_rate": 3.748579640945655e-07,
      "loss": 0.4449,
      "step": 403880
    },
    {
      "epoch": 165.33,
      "grad_norm": 2.1666147708892822,
      "learning_rate": 3.7477191325295566e-07,
      "loss": 0.4422,
      "step": 403890
    },
    {
      "epoch": 165.33,
      "grad_norm": 1.8058668375015259,
      "learning_rate": 3.746858714309301e-07,
      "loss": 0.4321,
      "step": 403900
    },
    {
      "epoch": 165.33,
      "grad_norm": 2.1482694149017334,
      "learning_rate": 3.745998386288818e-07,
      "loss": 0.4467,
      "step": 403910
    },
    {
      "epoch": 165.34,
      "grad_norm": 1.8858683109283447,
      "learning_rate": 3.745138148472054e-07,
      "loss": 0.4451,
      "step": 403920
    },
    {
      "epoch": 165.34,
      "grad_norm": 1.8642395734786987,
      "learning_rate": 3.744278000862952e-07,
      "loss": 0.4362,
      "step": 403930
    },
    {
      "epoch": 165.35,
      "grad_norm": 1.9905853271484375,
      "learning_rate": 3.7434179434654484e-07,
      "loss": 0.4348,
      "step": 403940
    },
    {
      "epoch": 165.35,
      "grad_norm": 1.8970320224761963,
      "learning_rate": 3.7425579762834833e-07,
      "loss": 0.4277,
      "step": 403950
    },
    {
      "epoch": 165.35,
      "grad_norm": 2.5214483737945557,
      "learning_rate": 3.7416980993210006e-07,
      "loss": 0.4631,
      "step": 403960
    },
    {
      "epoch": 165.36,
      "grad_norm": 2.096982717514038,
      "learning_rate": 3.7408383125819327e-07,
      "loss": 0.4624,
      "step": 403970
    },
    {
      "epoch": 165.36,
      "grad_norm": 2.467988967895508,
      "learning_rate": 3.7399786160702247e-07,
      "loss": 0.4321,
      "step": 403980
    },
    {
      "epoch": 165.37,
      "grad_norm": 2.038058280944824,
      "learning_rate": 3.73911900978981e-07,
      "loss": 0.4618,
      "step": 403990
    },
    {
      "epoch": 165.37,
      "grad_norm": 1.917786955833435,
      "learning_rate": 3.7382594937446265e-07,
      "loss": 0.413,
      "step": 404000
    },
    {
      "epoch": 165.37,
      "grad_norm": 3.0555431842803955,
      "learning_rate": 3.73740006793861e-07,
      "loss": 0.4444,
      "step": 404010
    },
    {
      "epoch": 165.38,
      "grad_norm": 1.982528567314148,
      "learning_rate": 3.736540732375704e-07,
      "loss": 0.4372,
      "step": 404020
    },
    {
      "epoch": 165.38,
      "grad_norm": 2.105780601501465,
      "learning_rate": 3.7356814870598474e-07,
      "loss": 0.4689,
      "step": 404030
    },
    {
      "epoch": 165.39,
      "grad_norm": 1.975871205329895,
      "learning_rate": 3.734822331994966e-07,
      "loss": 0.4429,
      "step": 404040
    },
    {
      "epoch": 165.39,
      "grad_norm": 2.4173221588134766,
      "learning_rate": 3.7339632671849984e-07,
      "loss": 0.4294,
      "step": 404050
    },
    {
      "epoch": 165.4,
      "grad_norm": 1.7053420543670654,
      "learning_rate": 3.7331042926338804e-07,
      "loss": 0.4507,
      "step": 404060
    },
    {
      "epoch": 165.4,
      "grad_norm": 2.547945976257324,
      "learning_rate": 3.7322454083455494e-07,
      "loss": 0.429,
      "step": 404070
    },
    {
      "epoch": 165.4,
      "grad_norm": 1.9944334030151367,
      "learning_rate": 3.731386614323937e-07,
      "loss": 0.4603,
      "step": 404080
    },
    {
      "epoch": 165.41,
      "grad_norm": 1.74185311794281,
      "learning_rate": 3.7305279105729795e-07,
      "loss": 0.4385,
      "step": 404090
    },
    {
      "epoch": 165.41,
      "grad_norm": 1.9904619455337524,
      "learning_rate": 3.7296692970966073e-07,
      "loss": 0.4254,
      "step": 404100
    },
    {
      "epoch": 165.42,
      "grad_norm": 2.084695339202881,
      "learning_rate": 3.7288107738987564e-07,
      "loss": 0.4419,
      "step": 404110
    },
    {
      "epoch": 165.42,
      "grad_norm": 2.3659679889678955,
      "learning_rate": 3.7279523409833563e-07,
      "loss": 0.4568,
      "step": 404120
    },
    {
      "epoch": 165.42,
      "grad_norm": 2.044861078262329,
      "learning_rate": 3.727093998354344e-07,
      "loss": 0.442,
      "step": 404130
    },
    {
      "epoch": 165.43,
      "grad_norm": 2.7651312351226807,
      "learning_rate": 3.7262357460156487e-07,
      "loss": 0.4636,
      "step": 404140
    },
    {
      "epoch": 165.43,
      "grad_norm": 1.895662546157837,
      "learning_rate": 3.725377583971202e-07,
      "loss": 0.4478,
      "step": 404150
    },
    {
      "epoch": 165.44,
      "grad_norm": 2.041555404663086,
      "learning_rate": 3.7245195122249355e-07,
      "loss": 0.4457,
      "step": 404160
    },
    {
      "epoch": 165.44,
      "grad_norm": 11.177693367004395,
      "learning_rate": 3.7236615307807787e-07,
      "loss": 0.4256,
      "step": 404170
    },
    {
      "epoch": 165.44,
      "grad_norm": 2.177203416824341,
      "learning_rate": 3.722803639642663e-07,
      "loss": 0.4355,
      "step": 404180
    },
    {
      "epoch": 165.45,
      "grad_norm": 2.308681011199951,
      "learning_rate": 3.721945838814521e-07,
      "loss": 0.4375,
      "step": 404190
    },
    {
      "epoch": 165.45,
      "grad_norm": 2.0856094360351562,
      "learning_rate": 3.7210881283002754e-07,
      "loss": 0.4333,
      "step": 404200
    },
    {
      "epoch": 165.46,
      "grad_norm": 2.246968984603882,
      "learning_rate": 3.7202305081038614e-07,
      "loss": 0.4524,
      "step": 404210
    },
    {
      "epoch": 165.46,
      "grad_norm": 1.979291319847107,
      "learning_rate": 3.719372978229205e-07,
      "loss": 0.4397,
      "step": 404220
    },
    {
      "epoch": 165.46,
      "grad_norm": 1.814681887626648,
      "learning_rate": 3.7185155386802333e-07,
      "loss": 0.461,
      "step": 404230
    },
    {
      "epoch": 165.47,
      "grad_norm": 1.8479347229003906,
      "learning_rate": 3.7176581894608776e-07,
      "loss": 0.4404,
      "step": 404240
    },
    {
      "epoch": 165.47,
      "grad_norm": 2.316675901412964,
      "learning_rate": 3.716800930575062e-07,
      "loss": 0.4557,
      "step": 404250
    },
    {
      "epoch": 165.48,
      "grad_norm": 2.1539947986602783,
      "learning_rate": 3.715943762026716e-07,
      "loss": 0.4552,
      "step": 404260
    },
    {
      "epoch": 165.48,
      "grad_norm": 1.9855804443359375,
      "learning_rate": 3.715086683819763e-07,
      "loss": 0.4584,
      "step": 404270
    },
    {
      "epoch": 165.49,
      "grad_norm": 2.0079152584075928,
      "learning_rate": 3.7142296959581324e-07,
      "loss": 0.4507,
      "step": 404280
    },
    {
      "epoch": 165.49,
      "grad_norm": 1.84878408908844,
      "learning_rate": 3.7133727984457477e-07,
      "loss": 0.4363,
      "step": 404290
    },
    {
      "epoch": 165.49,
      "grad_norm": 1.9118010997772217,
      "learning_rate": 3.7125159912865337e-07,
      "loss": 0.4489,
      "step": 404300
    },
    {
      "epoch": 165.5,
      "grad_norm": 2.0827159881591797,
      "learning_rate": 3.711659274484419e-07,
      "loss": 0.4323,
      "step": 404310
    },
    {
      "epoch": 165.5,
      "grad_norm": 2.4107491970062256,
      "learning_rate": 3.7108026480433255e-07,
      "loss": 0.4483,
      "step": 404320
    },
    {
      "epoch": 165.51,
      "grad_norm": 1.9427783489227295,
      "learning_rate": 3.709946111967176e-07,
      "loss": 0.4334,
      "step": 404330
    },
    {
      "epoch": 165.51,
      "grad_norm": 2.5351715087890625,
      "learning_rate": 3.709089666259903e-07,
      "loss": 0.4426,
      "step": 404340
    },
    {
      "epoch": 165.51,
      "grad_norm": 2.1820926666259766,
      "learning_rate": 3.708233310925415e-07,
      "loss": 0.4576,
      "step": 404350
    },
    {
      "epoch": 165.52,
      "grad_norm": 2.138033390045166,
      "learning_rate": 3.707377045967644e-07,
      "loss": 0.4565,
      "step": 404360
    },
    {
      "epoch": 165.52,
      "grad_norm": 2.1311302185058594,
      "learning_rate": 3.7065208713905105e-07,
      "loss": 0.4318,
      "step": 404370
    },
    {
      "epoch": 165.53,
      "grad_norm": 2.130096435546875,
      "learning_rate": 3.7056647871979356e-07,
      "loss": 0.4651,
      "step": 404380
    },
    {
      "epoch": 165.53,
      "grad_norm": 1.7096281051635742,
      "learning_rate": 3.704808793393837e-07,
      "loss": 0.4472,
      "step": 404390
    },
    {
      "epoch": 165.53,
      "grad_norm": 1.8714135885238647,
      "learning_rate": 3.7039528899821476e-07,
      "loss": 0.4295,
      "step": 404400
    },
    {
      "epoch": 165.54,
      "grad_norm": 1.9036200046539307,
      "learning_rate": 3.7030970769667786e-07,
      "loss": 0.4451,
      "step": 404410
    },
    {
      "epoch": 165.54,
      "grad_norm": 2.0894997119903564,
      "learning_rate": 3.7022413543516546e-07,
      "loss": 0.4579,
      "step": 404420
    },
    {
      "epoch": 165.55,
      "grad_norm": 2.011516809463501,
      "learning_rate": 3.7013857221406964e-07,
      "loss": 0.4457,
      "step": 404430
    },
    {
      "epoch": 165.55,
      "grad_norm": 2.1524486541748047,
      "learning_rate": 3.700530180337818e-07,
      "loss": 0.4444,
      "step": 404440
    },
    {
      "epoch": 165.55,
      "grad_norm": 2.077190399169922,
      "learning_rate": 3.6996747289469423e-07,
      "loss": 0.4557,
      "step": 404450
    },
    {
      "epoch": 165.56,
      "grad_norm": 1.654288411140442,
      "learning_rate": 3.6988193679719867e-07,
      "loss": 0.435,
      "step": 404460
    },
    {
      "epoch": 165.56,
      "grad_norm": 1.849990963935852,
      "learning_rate": 3.697964097416872e-07,
      "loss": 0.4393,
      "step": 404470
    },
    {
      "epoch": 165.57,
      "grad_norm": 2.303283214569092,
      "learning_rate": 3.6971089172855193e-07,
      "loss": 0.4398,
      "step": 404480
    },
    {
      "epoch": 165.57,
      "grad_norm": 2.329387664794922,
      "learning_rate": 3.696253827581834e-07,
      "loss": 0.4724,
      "step": 404490
    },
    {
      "epoch": 165.58,
      "grad_norm": 1.8435932397842407,
      "learning_rate": 3.6953988283097416e-07,
      "loss": 0.4475,
      "step": 404500
    },
    {
      "epoch": 165.58,
      "grad_norm": 2.0330610275268555,
      "learning_rate": 3.694543919473157e-07,
      "loss": 0.4262,
      "step": 404510
    },
    {
      "epoch": 165.58,
      "grad_norm": 2.0589425563812256,
      "learning_rate": 3.693689101075996e-07,
      "loss": 0.4361,
      "step": 404520
    },
    {
      "epoch": 165.59,
      "grad_norm": 2.1630637645721436,
      "learning_rate": 3.6928343731221757e-07,
      "loss": 0.4663,
      "step": 404530
    },
    {
      "epoch": 165.59,
      "grad_norm": 2.3400464057922363,
      "learning_rate": 3.6919797356156093e-07,
      "loss": 0.4535,
      "step": 404540
    },
    {
      "epoch": 165.6,
      "grad_norm": 2.1524817943573,
      "learning_rate": 3.691125188560216e-07,
      "loss": 0.447,
      "step": 404550
    },
    {
      "epoch": 165.6,
      "grad_norm": 2.239535093307495,
      "learning_rate": 3.6902707319599055e-07,
      "loss": 0.4614,
      "step": 404560
    },
    {
      "epoch": 165.6,
      "grad_norm": 2.319725513458252,
      "learning_rate": 3.6894163658185946e-07,
      "loss": 0.4469,
      "step": 404570
    },
    {
      "epoch": 165.61,
      "grad_norm": 2.691929578781128,
      "learning_rate": 3.6885620901401924e-07,
      "loss": 0.4627,
      "step": 404580
    },
    {
      "epoch": 165.61,
      "grad_norm": 2.0805604457855225,
      "learning_rate": 3.687707904928619e-07,
      "loss": 0.4476,
      "step": 404590
    },
    {
      "epoch": 165.62,
      "grad_norm": 1.8334168195724487,
      "learning_rate": 3.686853810187786e-07,
      "loss": 0.4337,
      "step": 404600
    },
    {
      "epoch": 165.62,
      "grad_norm": 2.0250935554504395,
      "learning_rate": 3.685999805921604e-07,
      "loss": 0.4571,
      "step": 404610
    },
    {
      "epoch": 165.62,
      "grad_norm": 1.9036515951156616,
      "learning_rate": 3.6851458921339924e-07,
      "loss": 0.4535,
      "step": 404620
    },
    {
      "epoch": 165.63,
      "grad_norm": 1.9445438385009766,
      "learning_rate": 3.6842920688288484e-07,
      "loss": 0.4429,
      "step": 404630
    },
    {
      "epoch": 165.63,
      "grad_norm": 2.223356246948242,
      "learning_rate": 3.6834383360100906e-07,
      "loss": 0.4502,
      "step": 404640
    },
    {
      "epoch": 165.64,
      "grad_norm": 2.2832939624786377,
      "learning_rate": 3.682584693681631e-07,
      "loss": 0.4427,
      "step": 404650
    },
    {
      "epoch": 165.64,
      "grad_norm": 2.178689479827881,
      "learning_rate": 3.6817311418473795e-07,
      "loss": 0.461,
      "step": 404660
    },
    {
      "epoch": 165.64,
      "grad_norm": 1.7540488243103027,
      "learning_rate": 3.680877680511245e-07,
      "loss": 0.4299,
      "step": 404670
    },
    {
      "epoch": 165.65,
      "grad_norm": 2.3026461601257324,
      "learning_rate": 3.6800243096771385e-07,
      "loss": 0.4511,
      "step": 404680
    },
    {
      "epoch": 165.65,
      "grad_norm": 2.305177688598633,
      "learning_rate": 3.6791710293489673e-07,
      "loss": 0.4678,
      "step": 404690
    },
    {
      "epoch": 165.66,
      "grad_norm": 2.538050889968872,
      "learning_rate": 3.6783178395306415e-07,
      "loss": 0.4473,
      "step": 404700
    },
    {
      "epoch": 165.66,
      "grad_norm": 4.197963237762451,
      "learning_rate": 3.677464740226068e-07,
      "loss": 0.4657,
      "step": 404710
    },
    {
      "epoch": 165.67,
      "grad_norm": 2.08126163482666,
      "learning_rate": 3.676611731439158e-07,
      "loss": 0.4257,
      "step": 404720
    },
    {
      "epoch": 165.67,
      "grad_norm": 1.802554726600647,
      "learning_rate": 3.675758813173815e-07,
      "loss": 0.442,
      "step": 404730
    },
    {
      "epoch": 165.67,
      "grad_norm": 1.7884951829910278,
      "learning_rate": 3.6749059854339494e-07,
      "loss": 0.4244,
      "step": 404740
    },
    {
      "epoch": 165.68,
      "grad_norm": 2.1068878173828125,
      "learning_rate": 3.674053248223465e-07,
      "loss": 0.4477,
      "step": 404750
    },
    {
      "epoch": 165.68,
      "grad_norm": 2.1421127319335938,
      "learning_rate": 3.6732006015462697e-07,
      "loss": 0.4483,
      "step": 404760
    },
    {
      "epoch": 165.69,
      "grad_norm": 1.5848714113235474,
      "learning_rate": 3.6723480454062693e-07,
      "loss": 0.4478,
      "step": 404770
    },
    {
      "epoch": 165.69,
      "grad_norm": 1.942206621170044,
      "learning_rate": 3.671495579807367e-07,
      "loss": 0.4316,
      "step": 404780
    },
    {
      "epoch": 165.69,
      "grad_norm": 2.623684883117676,
      "learning_rate": 3.670643204753474e-07,
      "loss": 0.4313,
      "step": 404790
    },
    {
      "epoch": 165.7,
      "grad_norm": 2.642019271850586,
      "learning_rate": 3.6697909202484873e-07,
      "loss": 0.4623,
      "step": 404800
    },
    {
      "epoch": 165.7,
      "grad_norm": 2.094648599624634,
      "learning_rate": 3.668938726296316e-07,
      "loss": 0.4396,
      "step": 404810
    },
    {
      "epoch": 165.71,
      "grad_norm": 2.2491793632507324,
      "learning_rate": 3.668086622900861e-07,
      "loss": 0.4038,
      "step": 404820
    },
    {
      "epoch": 165.71,
      "grad_norm": 2.0561680793762207,
      "learning_rate": 3.6672346100660273e-07,
      "loss": 0.4574,
      "step": 404830
    },
    {
      "epoch": 165.71,
      "grad_norm": 1.853441596031189,
      "learning_rate": 3.6663826877957177e-07,
      "loss": 0.4292,
      "step": 404840
    },
    {
      "epoch": 165.72,
      "grad_norm": 2.189983367919922,
      "learning_rate": 3.6655308560938364e-07,
      "loss": 0.446,
      "step": 404850
    },
    {
      "epoch": 165.72,
      "grad_norm": 2.1823244094848633,
      "learning_rate": 3.664679114964283e-07,
      "loss": 0.455,
      "step": 404860
    },
    {
      "epoch": 165.73,
      "grad_norm": 1.9870374202728271,
      "learning_rate": 3.663827464410959e-07,
      "loss": 0.4517,
      "step": 404870
    },
    {
      "epoch": 165.73,
      "grad_norm": 2.238027334213257,
      "learning_rate": 3.6629759044377664e-07,
      "loss": 0.4417,
      "step": 404880
    },
    {
      "epoch": 165.73,
      "grad_norm": 1.9856014251708984,
      "learning_rate": 3.662124435048606e-07,
      "loss": 0.4665,
      "step": 404890
    },
    {
      "epoch": 165.74,
      "grad_norm": 1.8428527116775513,
      "learning_rate": 3.661273056247379e-07,
      "loss": 0.4344,
      "step": 404900
    },
    {
      "epoch": 165.74,
      "grad_norm": 2.127617835998535,
      "learning_rate": 3.6604217680379834e-07,
      "loss": 0.4389,
      "step": 404910
    },
    {
      "epoch": 165.75,
      "grad_norm": 2.445546865463257,
      "learning_rate": 3.659570570424324e-07,
      "loss": 0.434,
      "step": 404920
    },
    {
      "epoch": 165.75,
      "grad_norm": 1.8691942691802979,
      "learning_rate": 3.658719463410299e-07,
      "loss": 0.4484,
      "step": 404930
    },
    {
      "epoch": 165.76,
      "grad_norm": 2.0173022747039795,
      "learning_rate": 3.6578684469997986e-07,
      "loss": 0.433,
      "step": 404940
    },
    {
      "epoch": 165.76,
      "grad_norm": 2.2864251136779785,
      "learning_rate": 3.657017521196729e-07,
      "loss": 0.4284,
      "step": 404950
    },
    {
      "epoch": 165.76,
      "grad_norm": 2.1861159801483154,
      "learning_rate": 3.6561666860049797e-07,
      "loss": 0.4621,
      "step": 404960
    },
    {
      "epoch": 165.77,
      "grad_norm": 1.9817919731140137,
      "learning_rate": 3.6553159414284594e-07,
      "loss": 0.4669,
      "step": 404970
    },
    {
      "epoch": 165.77,
      "grad_norm": 2.0776805877685547,
      "learning_rate": 3.654465287471063e-07,
      "loss": 0.4437,
      "step": 404980
    },
    {
      "epoch": 165.78,
      "grad_norm": 1.9775913953781128,
      "learning_rate": 3.6536147241366837e-07,
      "loss": 0.4533,
      "step": 404990
    },
    {
      "epoch": 165.78,
      "grad_norm": 2.101463794708252,
      "learning_rate": 3.6527642514292167e-07,
      "loss": 0.4422,
      "step": 405000
    },
    {
      "epoch": 165.78,
      "grad_norm": 2.3321692943573,
      "learning_rate": 3.6519138693525625e-07,
      "loss": 0.4562,
      "step": 405010
    },
    {
      "epoch": 165.79,
      "grad_norm": 2.4709131717681885,
      "learning_rate": 3.6510635779106137e-07,
      "loss": 0.432,
      "step": 405020
    },
    {
      "epoch": 165.79,
      "grad_norm": 1.8281680345535278,
      "learning_rate": 3.6502133771072655e-07,
      "loss": 0.4312,
      "step": 405030
    },
    {
      "epoch": 165.8,
      "grad_norm": 2.113372325897217,
      "learning_rate": 3.6493632669464134e-07,
      "loss": 0.4469,
      "step": 405040
    },
    {
      "epoch": 165.8,
      "grad_norm": 1.9288853406906128,
      "learning_rate": 3.6485132474319505e-07,
      "loss": 0.4435,
      "step": 405050
    },
    {
      "epoch": 165.8,
      "grad_norm": 1.90248703956604,
      "learning_rate": 3.6476633185677794e-07,
      "loss": 0.4508,
      "step": 405060
    },
    {
      "epoch": 165.81,
      "grad_norm": 1.8690578937530518,
      "learning_rate": 3.646813480357777e-07,
      "loss": 0.4459,
      "step": 405070
    },
    {
      "epoch": 165.81,
      "grad_norm": 2.0423896312713623,
      "learning_rate": 3.645963732805845e-07,
      "loss": 0.4514,
      "step": 405080
    },
    {
      "epoch": 165.82,
      "grad_norm": 1.851204514503479,
      "learning_rate": 3.6451140759158745e-07,
      "loss": 0.433,
      "step": 405090
    },
    {
      "epoch": 165.82,
      "grad_norm": 2.1403496265411377,
      "learning_rate": 3.644264509691759e-07,
      "loss": 0.4439,
      "step": 405100
    },
    {
      "epoch": 165.82,
      "grad_norm": 1.9463775157928467,
      "learning_rate": 3.643415034137392e-07,
      "loss": 0.4321,
      "step": 405110
    },
    {
      "epoch": 165.83,
      "grad_norm": 2.4360740184783936,
      "learning_rate": 3.6425656492566623e-07,
      "loss": 0.4501,
      "step": 405120
    },
    {
      "epoch": 165.83,
      "grad_norm": 1.889603614807129,
      "learning_rate": 3.6417163550534606e-07,
      "loss": 0.4435,
      "step": 405130
    },
    {
      "epoch": 165.84,
      "grad_norm": 1.6110804080963135,
      "learning_rate": 3.6408671515316724e-07,
      "loss": 0.4484,
      "step": 405140
    },
    {
      "epoch": 165.84,
      "grad_norm": 1.938653588294983,
      "learning_rate": 3.640018038695197e-07,
      "loss": 0.4459,
      "step": 405150
    },
    {
      "epoch": 165.85,
      "grad_norm": 2.0078842639923096,
      "learning_rate": 3.639169016547923e-07,
      "loss": 0.431,
      "step": 405160
    },
    {
      "epoch": 165.85,
      "grad_norm": 2.5323221683502197,
      "learning_rate": 3.6383200850937367e-07,
      "loss": 0.449,
      "step": 405170
    },
    {
      "epoch": 165.85,
      "grad_norm": 2.1244864463806152,
      "learning_rate": 3.637471244336527e-07,
      "loss": 0.4422,
      "step": 405180
    },
    {
      "epoch": 165.86,
      "grad_norm": 1.9120031595230103,
      "learning_rate": 3.636622494280183e-07,
      "loss": 0.4412,
      "step": 405190
    },
    {
      "epoch": 165.86,
      "grad_norm": 1.9476840496063232,
      "learning_rate": 3.635773834928596e-07,
      "loss": 0.446,
      "step": 405200
    },
    {
      "epoch": 165.87,
      "grad_norm": 2.128375768661499,
      "learning_rate": 3.6349252662856466e-07,
      "loss": 0.4522,
      "step": 405210
    },
    {
      "epoch": 165.87,
      "grad_norm": 2.052743911743164,
      "learning_rate": 3.6340767883552234e-07,
      "loss": 0.4432,
      "step": 405220
    },
    {
      "epoch": 165.87,
      "grad_norm": 2.237928628921509,
      "learning_rate": 3.633228401141217e-07,
      "loss": 0.4311,
      "step": 405230
    },
    {
      "epoch": 165.88,
      "grad_norm": 2.4964590072631836,
      "learning_rate": 3.632380104647511e-07,
      "loss": 0.4539,
      "step": 405240
    },
    {
      "epoch": 165.88,
      "grad_norm": 2.100762128829956,
      "learning_rate": 3.631531898877991e-07,
      "loss": 0.4514,
      "step": 405250
    },
    {
      "epoch": 165.89,
      "grad_norm": 3.5962977409362793,
      "learning_rate": 3.630683783836545e-07,
      "loss": 0.4449,
      "step": 405260
    },
    {
      "epoch": 165.89,
      "grad_norm": 2.0422685146331787,
      "learning_rate": 3.629835759527054e-07,
      "loss": 0.4352,
      "step": 405270
    },
    {
      "epoch": 165.89,
      "grad_norm": 2.0870778560638428,
      "learning_rate": 3.6289878259534074e-07,
      "loss": 0.4557,
      "step": 405280
    },
    {
      "epoch": 165.9,
      "grad_norm": 1.9495375156402588,
      "learning_rate": 3.628139983119488e-07,
      "loss": 0.4544,
      "step": 405290
    },
    {
      "epoch": 165.9,
      "grad_norm": 2.265972137451172,
      "learning_rate": 3.627292231029177e-07,
      "loss": 0.4375,
      "step": 405300
    },
    {
      "epoch": 165.91,
      "grad_norm": 2.0133821964263916,
      "learning_rate": 3.6264445696863616e-07,
      "loss": 0.44,
      "step": 405310
    },
    {
      "epoch": 165.91,
      "grad_norm": 1.7559092044830322,
      "learning_rate": 3.625596999094921e-07,
      "loss": 0.4287,
      "step": 405320
    },
    {
      "epoch": 165.91,
      "grad_norm": 2.12207293510437,
      "learning_rate": 3.6247495192587435e-07,
      "loss": 0.4364,
      "step": 405330
    },
    {
      "epoch": 165.92,
      "grad_norm": 2.1156082153320312,
      "learning_rate": 3.6239021301817027e-07,
      "loss": 0.4692,
      "step": 405340
    },
    {
      "epoch": 165.92,
      "grad_norm": 2.1579813957214355,
      "learning_rate": 3.6230548318676896e-07,
      "loss": 0.443,
      "step": 405350
    },
    {
      "epoch": 165.93,
      "grad_norm": 2.306795358657837,
      "learning_rate": 3.6222076243205783e-07,
      "loss": 0.438,
      "step": 405360
    },
    {
      "epoch": 165.93,
      "grad_norm": 2.2382454872131348,
      "learning_rate": 3.6213605075442536e-07,
      "loss": 0.4567,
      "step": 405370
    },
    {
      "epoch": 165.94,
      "grad_norm": 2.647064447402954,
      "learning_rate": 3.6205134815425953e-07,
      "loss": 0.4435,
      "step": 405380
    },
    {
      "epoch": 165.94,
      "grad_norm": 1.9089300632476807,
      "learning_rate": 3.619666546319483e-07,
      "loss": 0.4538,
      "step": 405390
    },
    {
      "epoch": 165.94,
      "grad_norm": 1.7474983930587769,
      "learning_rate": 3.6188197018787964e-07,
      "loss": 0.4538,
      "step": 405400
    },
    {
      "epoch": 165.95,
      "grad_norm": 2.015493631362915,
      "learning_rate": 3.617972948224413e-07,
      "loss": 0.4443,
      "step": 405410
    },
    {
      "epoch": 165.95,
      "grad_norm": 2.3556807041168213,
      "learning_rate": 3.6171262853602154e-07,
      "loss": 0.4199,
      "step": 405420
    },
    {
      "epoch": 165.96,
      "grad_norm": 2.4656622409820557,
      "learning_rate": 3.6162797132900823e-07,
      "loss": 0.4424,
      "step": 405430
    },
    {
      "epoch": 165.96,
      "grad_norm": 1.9951281547546387,
      "learning_rate": 3.615433232017886e-07,
      "loss": 0.437,
      "step": 405440
    },
    {
      "epoch": 165.96,
      "grad_norm": 2.1743531227111816,
      "learning_rate": 3.6145868415475085e-07,
      "loss": 0.4285,
      "step": 405450
    },
    {
      "epoch": 165.97,
      "grad_norm": 1.8186368942260742,
      "learning_rate": 3.613740541882828e-07,
      "loss": 0.431,
      "step": 405460
    },
    {
      "epoch": 165.97,
      "grad_norm": 1.966171383857727,
      "learning_rate": 3.612894333027718e-07,
      "loss": 0.4529,
      "step": 405470
    },
    {
      "epoch": 165.98,
      "grad_norm": 1.9377145767211914,
      "learning_rate": 3.612048214986059e-07,
      "loss": 0.4305,
      "step": 405480
    },
    {
      "epoch": 165.98,
      "grad_norm": 2.320648431777954,
      "learning_rate": 3.6112021877617217e-07,
      "loss": 0.4399,
      "step": 405490
    },
    {
      "epoch": 165.98,
      "grad_norm": 3.3906610012054443,
      "learning_rate": 3.610356251358584e-07,
      "loss": 0.4384,
      "step": 405500
    },
    {
      "epoch": 165.99,
      "grad_norm": 2.079631805419922,
      "learning_rate": 3.609510405780526e-07,
      "loss": 0.457,
      "step": 405510
    },
    {
      "epoch": 165.99,
      "grad_norm": 2.7214956283569336,
      "learning_rate": 3.6086646510314084e-07,
      "loss": 0.4612,
      "step": 405520
    },
    {
      "epoch": 166.0,
      "grad_norm": 1.767806053161621,
      "learning_rate": 3.607818987115119e-07,
      "loss": 0.4463,
      "step": 405530
    },
    {
      "epoch": 166.0,
      "eval_loss": 0.446331650018692,
      "eval_runtime": 56.6794,
      "eval_samples_per_second": 60.851,
      "eval_steps_per_second": 7.622,
      "step": 405538
    },
    {
      "epoch": 166.0,
      "grad_norm": 2.0962014198303223,
      "learning_rate": 3.60697341403553e-07,
      "loss": 0.4405,
      "step": 405540
    },
    {
      "epoch": 166.0,
      "grad_norm": 1.8718684911727905,
      "learning_rate": 3.60612793179651e-07,
      "loss": 0.4387,
      "step": 405550
    },
    {
      "epoch": 166.01,
      "grad_norm": 1.9659976959228516,
      "learning_rate": 3.6052825404019344e-07,
      "loss": 0.4566,
      "step": 405560
    },
    {
      "epoch": 166.01,
      "grad_norm": 2.3776280879974365,
      "learning_rate": 3.6044372398556743e-07,
      "loss": 0.4364,
      "step": 405570
    },
    {
      "epoch": 166.02,
      "grad_norm": 2.0532236099243164,
      "learning_rate": 3.603592030161605e-07,
      "loss": 0.436,
      "step": 405580
    },
    {
      "epoch": 166.02,
      "grad_norm": 2.132138967514038,
      "learning_rate": 3.602746911323595e-07,
      "loss": 0.4142,
      "step": 405590
    },
    {
      "epoch": 166.03,
      "grad_norm": 1.927698016166687,
      "learning_rate": 3.6019018833455164e-07,
      "loss": 0.4393,
      "step": 405600
    },
    {
      "epoch": 166.03,
      "grad_norm": 2.025761365890503,
      "learning_rate": 3.601056946231244e-07,
      "loss": 0.444,
      "step": 405610
    },
    {
      "epoch": 166.03,
      "grad_norm": 2.7414324283599854,
      "learning_rate": 3.600212099984642e-07,
      "loss": 0.4537,
      "step": 405620
    },
    {
      "epoch": 166.04,
      "grad_norm": 2.053861379623413,
      "learning_rate": 3.599367344609587e-07,
      "loss": 0.43,
      "step": 405630
    },
    {
      "epoch": 166.04,
      "grad_norm": 2.1674818992614746,
      "learning_rate": 3.5985226801099486e-07,
      "loss": 0.4313,
      "step": 405640
    },
    {
      "epoch": 166.05,
      "grad_norm": 1.9953773021697998,
      "learning_rate": 3.597678106489586e-07,
      "loss": 0.4369,
      "step": 405650
    },
    {
      "epoch": 166.05,
      "grad_norm": 2.1499457359313965,
      "learning_rate": 3.5968336237523784e-07,
      "loss": 0.4587,
      "step": 405660
    },
    {
      "epoch": 166.05,
      "grad_norm": 1.9855786561965942,
      "learning_rate": 3.5959892319021875e-07,
      "loss": 0.424,
      "step": 405670
    },
    {
      "epoch": 166.06,
      "grad_norm": 2.3965535163879395,
      "learning_rate": 3.5951449309428876e-07,
      "loss": 0.4632,
      "step": 405680
    },
    {
      "epoch": 166.06,
      "grad_norm": 2.082486391067505,
      "learning_rate": 3.59430072087834e-07,
      "loss": 0.4322,
      "step": 405690
    },
    {
      "epoch": 166.07,
      "grad_norm": 1.9511704444885254,
      "learning_rate": 3.5934566017124177e-07,
      "loss": 0.4415,
      "step": 405700
    },
    {
      "epoch": 166.07,
      "grad_norm": 2.037318706512451,
      "learning_rate": 3.5926125734489785e-07,
      "loss": 0.4307,
      "step": 405710
    },
    {
      "epoch": 166.07,
      "grad_norm": 2.0475802421569824,
      "learning_rate": 3.5917686360919024e-07,
      "loss": 0.4477,
      "step": 405720
    },
    {
      "epoch": 166.08,
      "grad_norm": 2.2604644298553467,
      "learning_rate": 3.5909247896450457e-07,
      "loss": 0.4528,
      "step": 405730
    },
    {
      "epoch": 166.08,
      "grad_norm": 2.0788707733154297,
      "learning_rate": 3.590081034112278e-07,
      "loss": 0.4385,
      "step": 405740
    },
    {
      "epoch": 166.09,
      "grad_norm": 1.8099665641784668,
      "learning_rate": 3.589237369497463e-07,
      "loss": 0.4352,
      "step": 405750
    },
    {
      "epoch": 166.09,
      "grad_norm": 2.046966314315796,
      "learning_rate": 3.5883937958044624e-07,
      "loss": 0.4637,
      "step": 405760
    },
    {
      "epoch": 166.09,
      "grad_norm": 1.9689902067184448,
      "learning_rate": 3.5875503130371456e-07,
      "loss": 0.4598,
      "step": 405770
    },
    {
      "epoch": 166.1,
      "grad_norm": 1.957939624786377,
      "learning_rate": 3.5867069211993816e-07,
      "loss": 0.4583,
      "step": 405780
    },
    {
      "epoch": 166.1,
      "grad_norm": 1.9519267082214355,
      "learning_rate": 3.585863620295019e-07,
      "loss": 0.4458,
      "step": 405790
    },
    {
      "epoch": 166.11,
      "grad_norm": 2.261263132095337,
      "learning_rate": 3.585020410327927e-07,
      "loss": 0.4534,
      "step": 405800
    },
    {
      "epoch": 166.11,
      "grad_norm": 2.2279703617095947,
      "learning_rate": 3.5841772913019703e-07,
      "loss": 0.4393,
      "step": 405810
    },
    {
      "epoch": 166.12,
      "grad_norm": 1.8674213886260986,
      "learning_rate": 3.5833342632210095e-07,
      "loss": 0.4307,
      "step": 405820
    },
    {
      "epoch": 166.12,
      "grad_norm": 2.132366418838501,
      "learning_rate": 3.582491326088909e-07,
      "loss": 0.4541,
      "step": 405830
    },
    {
      "epoch": 166.12,
      "grad_norm": 2.1401641368865967,
      "learning_rate": 3.581648479909528e-07,
      "loss": 0.4283,
      "step": 405840
    },
    {
      "epoch": 166.13,
      "grad_norm": 1.9981634616851807,
      "learning_rate": 3.5808057246867254e-07,
      "loss": 0.4442,
      "step": 405850
    },
    {
      "epoch": 166.13,
      "grad_norm": 2.5444369316101074,
      "learning_rate": 3.5799630604243677e-07,
      "loss": 0.4507,
      "step": 405860
    },
    {
      "epoch": 166.14,
      "grad_norm": 2.154008388519287,
      "learning_rate": 3.5791204871263087e-07,
      "loss": 0.4514,
      "step": 405870
    },
    {
      "epoch": 166.14,
      "grad_norm": 2.1366491317749023,
      "learning_rate": 3.578278004796412e-07,
      "loss": 0.4434,
      "step": 405880
    },
    {
      "epoch": 166.14,
      "grad_norm": 2.351708173751831,
      "learning_rate": 3.577435613438534e-07,
      "loss": 0.4414,
      "step": 405890
    },
    {
      "epoch": 166.15,
      "grad_norm": 1.94428288936615,
      "learning_rate": 3.576593313056532e-07,
      "loss": 0.4696,
      "step": 405900
    },
    {
      "epoch": 166.15,
      "grad_norm": 2.187199831008911,
      "learning_rate": 3.5757511036542717e-07,
      "loss": 0.4474,
      "step": 405910
    },
    {
      "epoch": 166.16,
      "grad_norm": 1.8859351873397827,
      "learning_rate": 3.574908985235612e-07,
      "loss": 0.438,
      "step": 405920
    },
    {
      "epoch": 166.16,
      "grad_norm": 2.241359233856201,
      "learning_rate": 3.574066957804402e-07,
      "loss": 0.4723,
      "step": 405930
    },
    {
      "epoch": 166.16,
      "grad_norm": 1.720744252204895,
      "learning_rate": 3.573225021364503e-07,
      "loss": 0.4378,
      "step": 405940
    },
    {
      "epoch": 166.17,
      "grad_norm": 2.226092576980591,
      "learning_rate": 3.5723831759197706e-07,
      "loss": 0.4334,
      "step": 405950
    },
    {
      "epoch": 166.17,
      "grad_norm": 2.317342519760132,
      "learning_rate": 3.5715414214740595e-07,
      "loss": 0.4357,
      "step": 405960
    },
    {
      "epoch": 166.18,
      "grad_norm": 1.987034559249878,
      "learning_rate": 3.5706997580312305e-07,
      "loss": 0.4457,
      "step": 405970
    },
    {
      "epoch": 166.18,
      "grad_norm": 2.115891933441162,
      "learning_rate": 3.5698581855951375e-07,
      "loss": 0.4587,
      "step": 405980
    },
    {
      "epoch": 166.19,
      "grad_norm": 1.750076413154602,
      "learning_rate": 3.5690167041696313e-07,
      "loss": 0.4352,
      "step": 405990
    },
    {
      "epoch": 166.19,
      "grad_norm": 2.2174742221832275,
      "learning_rate": 3.5681753137585714e-07,
      "loss": 0.4301,
      "step": 406000
    },
    {
      "epoch": 166.19,
      "grad_norm": 1.9973031282424927,
      "learning_rate": 3.5673340143658137e-07,
      "loss": 0.4561,
      "step": 406010
    },
    {
      "epoch": 166.2,
      "grad_norm": 3.965095281600952,
      "learning_rate": 3.566492805995207e-07,
      "loss": 0.4679,
      "step": 406020
    },
    {
      "epoch": 166.2,
      "grad_norm": 2.7585248947143555,
      "learning_rate": 3.5656516886506047e-07,
      "loss": 0.4367,
      "step": 406030
    },
    {
      "epoch": 166.21,
      "grad_norm": 1.8389681577682495,
      "learning_rate": 3.564810662335865e-07,
      "loss": 0.4568,
      "step": 406040
    },
    {
      "epoch": 166.21,
      "grad_norm": 1.99527108669281,
      "learning_rate": 3.5639697270548354e-07,
      "loss": 0.4354,
      "step": 406050
    },
    {
      "epoch": 166.21,
      "grad_norm": 1.956248164176941,
      "learning_rate": 3.563128882811371e-07,
      "loss": 0.4412,
      "step": 406060
    },
    {
      "epoch": 166.22,
      "grad_norm": 1.9860810041427612,
      "learning_rate": 3.5622881296093234e-07,
      "loss": 0.4678,
      "step": 406070
    },
    {
      "epoch": 166.22,
      "grad_norm": 1.944389820098877,
      "learning_rate": 3.561447467452543e-07,
      "loss": 0.4541,
      "step": 406080
    },
    {
      "epoch": 166.23,
      "grad_norm": 2.110952377319336,
      "learning_rate": 3.5606068963448817e-07,
      "loss": 0.4329,
      "step": 406090
    },
    {
      "epoch": 166.23,
      "grad_norm": 1.7905758619308472,
      "learning_rate": 3.5597664162901877e-07,
      "loss": 0.4562,
      "step": 406100
    },
    {
      "epoch": 166.23,
      "grad_norm": 1.8485225439071655,
      "learning_rate": 3.558926027292314e-07,
      "loss": 0.4428,
      "step": 406110
    },
    {
      "epoch": 166.24,
      "grad_norm": 1.9761701822280884,
      "learning_rate": 3.558085729355108e-07,
      "loss": 0.4398,
      "step": 406120
    },
    {
      "epoch": 166.24,
      "grad_norm": 2.022059679031372,
      "learning_rate": 3.5572455224824223e-07,
      "loss": 0.4405,
      "step": 406130
    },
    {
      "epoch": 166.25,
      "grad_norm": 1.9033434391021729,
      "learning_rate": 3.5564054066781e-07,
      "loss": 0.4754,
      "step": 406140
    },
    {
      "epoch": 166.25,
      "grad_norm": 2.192228317260742,
      "learning_rate": 3.5555653819459955e-07,
      "loss": 0.4764,
      "step": 406150
    },
    {
      "epoch": 166.25,
      "grad_norm": 2.1926629543304443,
      "learning_rate": 3.5547254482899544e-07,
      "loss": 0.4452,
      "step": 406160
    },
    {
      "epoch": 166.26,
      "grad_norm": 2.789719343185425,
      "learning_rate": 3.5538856057138225e-07,
      "loss": 0.4447,
      "step": 406170
    },
    {
      "epoch": 166.26,
      "grad_norm": 2.113084554672241,
      "learning_rate": 3.5530458542214504e-07,
      "loss": 0.442,
      "step": 406180
    },
    {
      "epoch": 166.27,
      "grad_norm": 2.295928478240967,
      "learning_rate": 3.552206193816682e-07,
      "loss": 0.4375,
      "step": 406190
    },
    {
      "epoch": 166.27,
      "grad_norm": 1.9297453165054321,
      "learning_rate": 3.5513666245033666e-07,
      "loss": 0.4559,
      "step": 406200
    },
    {
      "epoch": 166.28,
      "grad_norm": 1.811972737312317,
      "learning_rate": 3.5505271462853486e-07,
      "loss": 0.4455,
      "step": 406210
    },
    {
      "epoch": 166.28,
      "grad_norm": 2.291642665863037,
      "learning_rate": 3.5496877591664716e-07,
      "loss": 0.4345,
      "step": 406220
    },
    {
      "epoch": 166.28,
      "grad_norm": 1.8616719245910645,
      "learning_rate": 3.54884846315059e-07,
      "loss": 0.4426,
      "step": 406230
    },
    {
      "epoch": 166.29,
      "grad_norm": 2.093801498413086,
      "learning_rate": 3.548009258241533e-07,
      "loss": 0.4319,
      "step": 406240
    },
    {
      "epoch": 166.29,
      "grad_norm": 2.1216366291046143,
      "learning_rate": 3.5471701444431554e-07,
      "loss": 0.4477,
      "step": 406250
    },
    {
      "epoch": 166.3,
      "grad_norm": 2.1124889850616455,
      "learning_rate": 3.5463311217592975e-07,
      "loss": 0.4536,
      "step": 406260
    },
    {
      "epoch": 166.3,
      "grad_norm": 2.0207605361938477,
      "learning_rate": 3.545492190193802e-07,
      "loss": 0.4426,
      "step": 406270
    },
    {
      "epoch": 166.3,
      "grad_norm": 1.9769861698150635,
      "learning_rate": 3.5446533497505106e-07,
      "loss": 0.4389,
      "step": 406280
    },
    {
      "epoch": 166.31,
      "grad_norm": 2.0199897289276123,
      "learning_rate": 3.5438146004332735e-07,
      "loss": 0.4401,
      "step": 406290
    },
    {
      "epoch": 166.31,
      "grad_norm": 1.8410277366638184,
      "learning_rate": 3.5429759422459264e-07,
      "loss": 0.4619,
      "step": 406300
    },
    {
      "epoch": 166.32,
      "grad_norm": 2.0383801460266113,
      "learning_rate": 3.542137375192315e-07,
      "loss": 0.453,
      "step": 406310
    },
    {
      "epoch": 166.32,
      "grad_norm": 1.9699848890304565,
      "learning_rate": 3.541298899276278e-07,
      "loss": 0.4572,
      "step": 406320
    },
    {
      "epoch": 166.32,
      "grad_norm": 2.4278206825256348,
      "learning_rate": 3.5404605145016587e-07,
      "loss": 0.4508,
      "step": 406330
    },
    {
      "epoch": 166.33,
      "grad_norm": 2.118206262588501,
      "learning_rate": 3.5396222208722914e-07,
      "loss": 0.4494,
      "step": 406340
    },
    {
      "epoch": 166.33,
      "grad_norm": 2.307251214981079,
      "learning_rate": 3.538784018392026e-07,
      "loss": 0.4269,
      "step": 406350
    },
    {
      "epoch": 166.34,
      "grad_norm": 1.9393956661224365,
      "learning_rate": 3.5379459070646997e-07,
      "loss": 0.4533,
      "step": 406360
    },
    {
      "epoch": 166.34,
      "grad_norm": 2.1899094581604004,
      "learning_rate": 3.537107886894143e-07,
      "loss": 0.4478,
      "step": 406370
    },
    {
      "epoch": 166.34,
      "grad_norm": 2.0139451026916504,
      "learning_rate": 3.5362699578841994e-07,
      "loss": 0.4446,
      "step": 406380
    },
    {
      "epoch": 166.35,
      "grad_norm": 2.0641331672668457,
      "learning_rate": 3.535432120038712e-07,
      "loss": 0.4496,
      "step": 406390
    },
    {
      "epoch": 166.35,
      "grad_norm": 2.3525209426879883,
      "learning_rate": 3.5345943733615134e-07,
      "loss": 0.439,
      "step": 406400
    },
    {
      "epoch": 166.36,
      "grad_norm": 3.396454095840454,
      "learning_rate": 3.533756717856442e-07,
      "loss": 0.4345,
      "step": 406410
    },
    {
      "epoch": 166.36,
      "grad_norm": 2.2830395698547363,
      "learning_rate": 3.5329191535273387e-07,
      "loss": 0.4465,
      "step": 406420
    },
    {
      "epoch": 166.37,
      "grad_norm": 2.422581195831299,
      "learning_rate": 3.532081680378036e-07,
      "loss": 0.4502,
      "step": 406430
    },
    {
      "epoch": 166.37,
      "grad_norm": 1.9012930393218994,
      "learning_rate": 3.531244298412372e-07,
      "loss": 0.4601,
      "step": 406440
    },
    {
      "epoch": 166.37,
      "grad_norm": 2.731088161468506,
      "learning_rate": 3.5304070076341824e-07,
      "loss": 0.4461,
      "step": 406450
    },
    {
      "epoch": 166.38,
      "grad_norm": 2.0189318656921387,
      "learning_rate": 3.529569808047302e-07,
      "loss": 0.4178,
      "step": 406460
    },
    {
      "epoch": 166.38,
      "grad_norm": 2.152470350265503,
      "learning_rate": 3.528732699655562e-07,
      "loss": 0.4462,
      "step": 406470
    },
    {
      "epoch": 166.39,
      "grad_norm": 2.0537641048431396,
      "learning_rate": 3.527895682462808e-07,
      "loss": 0.4354,
      "step": 406480
    },
    {
      "epoch": 166.39,
      "grad_norm": 2.134578227996826,
      "learning_rate": 3.527058756472867e-07,
      "loss": 0.4703,
      "step": 406490
    },
    {
      "epoch": 166.39,
      "grad_norm": 1.9137709140777588,
      "learning_rate": 3.526221921689581e-07,
      "loss": 0.4393,
      "step": 406500
    },
    {
      "epoch": 166.4,
      "grad_norm": 2.482114791870117,
      "learning_rate": 3.525385178116768e-07,
      "loss": 0.4619,
      "step": 406510
    },
    {
      "epoch": 166.4,
      "grad_norm": 2.4510676860809326,
      "learning_rate": 3.524548525758273e-07,
      "loss": 0.4622,
      "step": 406520
    },
    {
      "epoch": 166.41,
      "grad_norm": 2.4140474796295166,
      "learning_rate": 3.5237119646179206e-07,
      "loss": 0.4515,
      "step": 406530
    },
    {
      "epoch": 166.41,
      "grad_norm": 1.9783464670181274,
      "learning_rate": 3.522875494699551e-07,
      "loss": 0.4627,
      "step": 406540
    },
    {
      "epoch": 166.41,
      "grad_norm": 2.44088077545166,
      "learning_rate": 3.52203911600699e-07,
      "loss": 0.4605,
      "step": 406550
    },
    {
      "epoch": 166.42,
      "grad_norm": 2.460824489593506,
      "learning_rate": 3.521202828544069e-07,
      "loss": 0.4404,
      "step": 406560
    },
    {
      "epoch": 166.42,
      "grad_norm": 2.003152370452881,
      "learning_rate": 3.5203666323146254e-07,
      "loss": 0.456,
      "step": 406570
    },
    {
      "epoch": 166.43,
      "grad_norm": 1.898287296295166,
      "learning_rate": 3.519530527322483e-07,
      "loss": 0.4587,
      "step": 406580
    },
    {
      "epoch": 166.43,
      "grad_norm": 2.1917779445648193,
      "learning_rate": 3.5186945135714727e-07,
      "loss": 0.4652,
      "step": 406590
    },
    {
      "epoch": 166.43,
      "grad_norm": 1.735732078552246,
      "learning_rate": 3.517858591065427e-07,
      "loss": 0.4434,
      "step": 406600
    },
    {
      "epoch": 166.44,
      "grad_norm": 6.194061279296875,
      "learning_rate": 3.5170227598081735e-07,
      "loss": 0.4348,
      "step": 406610
    },
    {
      "epoch": 166.44,
      "grad_norm": 2.0074965953826904,
      "learning_rate": 3.51618701980354e-07,
      "loss": 0.4386,
      "step": 406620
    },
    {
      "epoch": 166.45,
      "grad_norm": 2.2566444873809814,
      "learning_rate": 3.515351371055357e-07,
      "loss": 0.4538,
      "step": 406630
    },
    {
      "epoch": 166.45,
      "grad_norm": 2.1235861778259277,
      "learning_rate": 3.514515813567452e-07,
      "loss": 0.4668,
      "step": 406640
    },
    {
      "epoch": 166.46,
      "grad_norm": 2.0166733264923096,
      "learning_rate": 3.5136803473436525e-07,
      "loss": 0.4517,
      "step": 406650
    },
    {
      "epoch": 166.46,
      "grad_norm": 2.076932668685913,
      "learning_rate": 3.512844972387784e-07,
      "loss": 0.4469,
      "step": 406660
    },
    {
      "epoch": 166.46,
      "grad_norm": 2.1665115356445312,
      "learning_rate": 3.5120096887036763e-07,
      "loss": 0.4574,
      "step": 406670
    },
    {
      "epoch": 166.47,
      "grad_norm": 2.057340621948242,
      "learning_rate": 3.511174496295152e-07,
      "loss": 0.4342,
      "step": 406680
    },
    {
      "epoch": 166.47,
      "grad_norm": 1.679929494857788,
      "learning_rate": 3.510339395166039e-07,
      "loss": 0.456,
      "step": 406690
    },
    {
      "epoch": 166.48,
      "grad_norm": 1.7867090702056885,
      "learning_rate": 3.509504385320164e-07,
      "loss": 0.4469,
      "step": 406700
    },
    {
      "epoch": 166.48,
      "grad_norm": 1.9931882619857788,
      "learning_rate": 3.508669466761349e-07,
      "loss": 0.4366,
      "step": 406710
    },
    {
      "epoch": 166.48,
      "grad_norm": 2.1217734813690186,
      "learning_rate": 3.507834639493422e-07,
      "loss": 0.4429,
      "step": 406720
    },
    {
      "epoch": 166.49,
      "grad_norm": 2.0113260746002197,
      "learning_rate": 3.506999903520204e-07,
      "loss": 0.4246,
      "step": 406730
    },
    {
      "epoch": 166.49,
      "grad_norm": 2.018669843673706,
      "learning_rate": 3.5061652588455195e-07,
      "loss": 0.4541,
      "step": 406740
    },
    {
      "epoch": 166.5,
      "grad_norm": 2.0112922191619873,
      "learning_rate": 3.505330705473194e-07,
      "loss": 0.441,
      "step": 406750
    },
    {
      "epoch": 166.5,
      "grad_norm": 1.9440395832061768,
      "learning_rate": 3.5044962434070506e-07,
      "loss": 0.4457,
      "step": 406760
    },
    {
      "epoch": 166.5,
      "grad_norm": 2.6563303470611572,
      "learning_rate": 3.503661872650908e-07,
      "loss": 0.4755,
      "step": 406770
    },
    {
      "epoch": 166.51,
      "grad_norm": 1.642783284187317,
      "learning_rate": 3.5028275932085914e-07,
      "loss": 0.4215,
      "step": 406780
    },
    {
      "epoch": 166.51,
      "grad_norm": 1.8568683862686157,
      "learning_rate": 3.501993405083924e-07,
      "loss": 0.4303,
      "step": 406790
    },
    {
      "epoch": 166.52,
      "grad_norm": 1.985227346420288,
      "learning_rate": 3.501159308280722e-07,
      "loss": 0.4379,
      "step": 406800
    },
    {
      "epoch": 166.52,
      "grad_norm": 2.0719354152679443,
      "learning_rate": 3.5003253028028165e-07,
      "loss": 0.4303,
      "step": 406810
    },
    {
      "epoch": 166.52,
      "grad_norm": 2.272777795791626,
      "learning_rate": 3.4994913886540147e-07,
      "loss": 0.445,
      "step": 406820
    },
    {
      "epoch": 166.53,
      "grad_norm": 2.4203855991363525,
      "learning_rate": 3.49865756583814e-07,
      "loss": 0.4661,
      "step": 406830
    },
    {
      "epoch": 166.53,
      "grad_norm": 4.803702354431152,
      "learning_rate": 3.4978238343590117e-07,
      "loss": 0.4423,
      "step": 406840
    },
    {
      "epoch": 166.54,
      "grad_norm": 2.4032976627349854,
      "learning_rate": 3.4969901942204566e-07,
      "loss": 0.4684,
      "step": 406850
    },
    {
      "epoch": 166.54,
      "grad_norm": 1.9947402477264404,
      "learning_rate": 3.4961566454262886e-07,
      "loss": 0.4527,
      "step": 406860
    },
    {
      "epoch": 166.55,
      "grad_norm": 2.028240919113159,
      "learning_rate": 3.4953231879803286e-07,
      "loss": 0.4512,
      "step": 406870
    },
    {
      "epoch": 166.55,
      "grad_norm": 2.121225595474243,
      "learning_rate": 3.4944898218863885e-07,
      "loss": 0.4481,
      "step": 406880
    },
    {
      "epoch": 166.55,
      "grad_norm": 2.227539300918579,
      "learning_rate": 3.493656547148293e-07,
      "loss": 0.4241,
      "step": 406890
    },
    {
      "epoch": 166.56,
      "grad_norm": 2.0745222568511963,
      "learning_rate": 3.492823363769854e-07,
      "loss": 0.4396,
      "step": 406900
    },
    {
      "epoch": 166.56,
      "grad_norm": 2.282080888748169,
      "learning_rate": 3.4919902717548897e-07,
      "loss": 0.4474,
      "step": 406910
    },
    {
      "epoch": 166.57,
      "grad_norm": 1.7148733139038086,
      "learning_rate": 3.491157271107216e-07,
      "loss": 0.4347,
      "step": 406920
    },
    {
      "epoch": 166.57,
      "grad_norm": 2.4254205226898193,
      "learning_rate": 3.490324361830649e-07,
      "loss": 0.4596,
      "step": 406930
    },
    {
      "epoch": 166.57,
      "grad_norm": 2.0255725383758545,
      "learning_rate": 3.489491543929005e-07,
      "loss": 0.4354,
      "step": 406940
    },
    {
      "epoch": 166.58,
      "grad_norm": 2.0639407634735107,
      "learning_rate": 3.4886588174061036e-07,
      "loss": 0.4277,
      "step": 406950
    },
    {
      "epoch": 166.58,
      "grad_norm": 2.5445210933685303,
      "learning_rate": 3.48782618226575e-07,
      "loss": 0.4632,
      "step": 406960
    },
    {
      "epoch": 166.59,
      "grad_norm": 2.014577627182007,
      "learning_rate": 3.486993638511761e-07,
      "loss": 0.4385,
      "step": 406970
    },
    {
      "epoch": 166.59,
      "grad_norm": 2.028458595275879,
      "learning_rate": 3.486161186147954e-07,
      "loss": 0.4547,
      "step": 406980
    },
    {
      "epoch": 166.59,
      "grad_norm": 2.149731397628784,
      "learning_rate": 3.4853288251781383e-07,
      "loss": 0.4672,
      "step": 406990
    },
    {
      "epoch": 166.6,
      "grad_norm": 2.1577868461608887,
      "learning_rate": 3.484496555606129e-07,
      "loss": 0.4569,
      "step": 407000
    },
    {
      "epoch": 166.6,
      "grad_norm": 1.9584869146347046,
      "learning_rate": 3.4836643774357394e-07,
      "loss": 0.4549,
      "step": 407010
    },
    {
      "epoch": 166.61,
      "grad_norm": 1.7909057140350342,
      "learning_rate": 3.4828322906707803e-07,
      "loss": 0.4488,
      "step": 407020
    },
    {
      "epoch": 166.61,
      "grad_norm": 2.144327163696289,
      "learning_rate": 3.4820002953150556e-07,
      "loss": 0.4284,
      "step": 407030
    },
    {
      "epoch": 166.61,
      "grad_norm": 2.210261821746826,
      "learning_rate": 3.4811683913723934e-07,
      "loss": 0.448,
      "step": 407040
    },
    {
      "epoch": 166.62,
      "grad_norm": 1.931422472000122,
      "learning_rate": 3.480336578846594e-07,
      "loss": 0.4387,
      "step": 407050
    },
    {
      "epoch": 166.62,
      "grad_norm": 2.0910751819610596,
      "learning_rate": 3.479504857741471e-07,
      "loss": 0.4573,
      "step": 407060
    },
    {
      "epoch": 166.63,
      "grad_norm": 1.8907952308654785,
      "learning_rate": 3.47867322806083e-07,
      "loss": 0.4348,
      "step": 407070
    },
    {
      "epoch": 166.63,
      "grad_norm": 2.182236671447754,
      "learning_rate": 3.477841689808492e-07,
      "loss": 0.4323,
      "step": 407080
    },
    {
      "epoch": 166.64,
      "grad_norm": 1.9094818830490112,
      "learning_rate": 3.47701024298825e-07,
      "loss": 0.4618,
      "step": 407090
    },
    {
      "epoch": 166.64,
      "grad_norm": 2.64697003364563,
      "learning_rate": 3.476178887603921e-07,
      "loss": 0.4341,
      "step": 407100
    },
    {
      "epoch": 166.64,
      "grad_norm": 2.450732469558716,
      "learning_rate": 3.475347623659313e-07,
      "loss": 0.4355,
      "step": 407110
    },
    {
      "epoch": 166.65,
      "grad_norm": 2.048055648803711,
      "learning_rate": 3.474516451158232e-07,
      "loss": 0.4568,
      "step": 407120
    },
    {
      "epoch": 166.65,
      "grad_norm": 2.0473878383636475,
      "learning_rate": 3.4736853701044885e-07,
      "loss": 0.46,
      "step": 407130
    },
    {
      "epoch": 166.66,
      "grad_norm": 2.299208879470825,
      "learning_rate": 3.4728543805018877e-07,
      "loss": 0.4592,
      "step": 407140
    },
    {
      "epoch": 166.66,
      "grad_norm": 2.220588207244873,
      "learning_rate": 3.4720234823542373e-07,
      "loss": 0.4497,
      "step": 407150
    },
    {
      "epoch": 166.66,
      "grad_norm": 2.3074653148651123,
      "learning_rate": 3.4711926756653416e-07,
      "loss": 0.4234,
      "step": 407160
    },
    {
      "epoch": 166.67,
      "grad_norm": 1.9066917896270752,
      "learning_rate": 3.4703619604390075e-07,
      "loss": 0.4453,
      "step": 407170
    },
    {
      "epoch": 166.67,
      "grad_norm": 1.8707191944122314,
      "learning_rate": 3.4695313366790417e-07,
      "loss": 0.436,
      "step": 407180
    },
    {
      "epoch": 166.68,
      "grad_norm": 1.7410000562667847,
      "learning_rate": 3.4687008043892486e-07,
      "loss": 0.4411,
      "step": 407190
    },
    {
      "epoch": 166.68,
      "grad_norm": 2.1696081161499023,
      "learning_rate": 3.467870363573429e-07,
      "loss": 0.4616,
      "step": 407200
    },
    {
      "epoch": 166.68,
      "grad_norm": 1.8172788619995117,
      "learning_rate": 3.4670400142353914e-07,
      "loss": 0.4459,
      "step": 407210
    },
    {
      "epoch": 166.69,
      "grad_norm": 2.2493298053741455,
      "learning_rate": 3.466209756378939e-07,
      "loss": 0.4555,
      "step": 407220
    },
    {
      "epoch": 166.69,
      "grad_norm": 2.0939559936523438,
      "learning_rate": 3.465379590007873e-07,
      "loss": 0.4426,
      "step": 407230
    },
    {
      "epoch": 166.7,
      "grad_norm": 1.762121558189392,
      "learning_rate": 3.464549515125999e-07,
      "loss": 0.4661,
      "step": 407240
    },
    {
      "epoch": 166.7,
      "grad_norm": 2.176400899887085,
      "learning_rate": 3.4637195317371173e-07,
      "loss": 0.439,
      "step": 407250
    },
    {
      "epoch": 166.7,
      "grad_norm": 1.8985811471939087,
      "learning_rate": 3.46288963984503e-07,
      "loss": 0.4441,
      "step": 407260
    },
    {
      "epoch": 166.71,
      "grad_norm": 1.9364616870880127,
      "learning_rate": 3.4620598394535395e-07,
      "loss": 0.4496,
      "step": 407270
    },
    {
      "epoch": 166.71,
      "grad_norm": 2.2663960456848145,
      "learning_rate": 3.461230130566449e-07,
      "loss": 0.435,
      "step": 407280
    },
    {
      "epoch": 166.72,
      "grad_norm": 2.133953332901001,
      "learning_rate": 3.460400513187555e-07,
      "loss": 0.4388,
      "step": 407290
    },
    {
      "epoch": 166.72,
      "grad_norm": 1.9121208190917969,
      "learning_rate": 3.4595709873206603e-07,
      "loss": 0.4616,
      "step": 407300
    },
    {
      "epoch": 166.73,
      "grad_norm": 2.1200783252716064,
      "learning_rate": 3.4587415529695625e-07,
      "loss": 0.4344,
      "step": 407310
    },
    {
      "epoch": 166.73,
      "grad_norm": 1.7645808458328247,
      "learning_rate": 3.457912210138067e-07,
      "loss": 0.4422,
      "step": 407320
    },
    {
      "epoch": 166.73,
      "grad_norm": 1.8162484169006348,
      "learning_rate": 3.457082958829967e-07,
      "loss": 0.433,
      "step": 407330
    },
    {
      "epoch": 166.74,
      "grad_norm": 1.9874768257141113,
      "learning_rate": 3.4562537990490613e-07,
      "loss": 0.4306,
      "step": 407340
    },
    {
      "epoch": 166.74,
      "grad_norm": 2.085916757583618,
      "learning_rate": 3.455424730799152e-07,
      "loss": 0.4256,
      "step": 407350
    },
    {
      "epoch": 166.75,
      "grad_norm": 2.6603827476501465,
      "learning_rate": 3.454595754084033e-07,
      "loss": 0.4382,
      "step": 407360
    },
    {
      "epoch": 166.75,
      "grad_norm": 1.7786294221878052,
      "learning_rate": 3.453766868907506e-07,
      "loss": 0.4493,
      "step": 407370
    },
    {
      "epoch": 166.75,
      "grad_norm": 2.1658072471618652,
      "learning_rate": 3.4529380752733644e-07,
      "loss": 0.4273,
      "step": 407380
    },
    {
      "epoch": 166.76,
      "grad_norm": 2.168311357498169,
      "learning_rate": 3.452109373185413e-07,
      "loss": 0.4431,
      "step": 407390
    },
    {
      "epoch": 166.76,
      "grad_norm": 1.7641932964324951,
      "learning_rate": 3.451280762647432e-07,
      "loss": 0.4301,
      "step": 407400
    },
    {
      "epoch": 166.77,
      "grad_norm": 1.7936387062072754,
      "learning_rate": 3.4504522436632234e-07,
      "loss": 0.4409,
      "step": 407410
    },
    {
      "epoch": 166.77,
      "grad_norm": 2.330389976501465,
      "learning_rate": 3.449623816236589e-07,
      "loss": 0.4475,
      "step": 407420
    },
    {
      "epoch": 166.77,
      "grad_norm": 2.795459032058716,
      "learning_rate": 3.4487954803713223e-07,
      "loss": 0.4544,
      "step": 407430
    },
    {
      "epoch": 166.78,
      "grad_norm": 1.86549973487854,
      "learning_rate": 3.4479672360712123e-07,
      "loss": 0.4598,
      "step": 407440
    },
    {
      "epoch": 166.78,
      "grad_norm": 2.2954771518707275,
      "learning_rate": 3.447139083340058e-07,
      "loss": 0.432,
      "step": 407450
    },
    {
      "epoch": 166.79,
      "grad_norm": 2.0280611515045166,
      "learning_rate": 3.4463110221816526e-07,
      "loss": 0.4454,
      "step": 407460
    },
    {
      "epoch": 166.79,
      "grad_norm": 2.414138078689575,
      "learning_rate": 3.4454830525997854e-07,
      "loss": 0.4305,
      "step": 407470
    },
    {
      "epoch": 166.79,
      "grad_norm": 2.010114908218384,
      "learning_rate": 3.444655174598253e-07,
      "loss": 0.4315,
      "step": 407480
    },
    {
      "epoch": 166.8,
      "grad_norm": 1.9192872047424316,
      "learning_rate": 3.443827388180846e-07,
      "loss": 0.4451,
      "step": 407490
    },
    {
      "epoch": 166.8,
      "grad_norm": 1.7937374114990234,
      "learning_rate": 3.4429996933513555e-07,
      "loss": 0.4422,
      "step": 407500
    },
    {
      "epoch": 166.81,
      "grad_norm": 2.1686770915985107,
      "learning_rate": 3.4421720901135763e-07,
      "loss": 0.4389,
      "step": 407510
    },
    {
      "epoch": 166.81,
      "grad_norm": 1.9388045072555542,
      "learning_rate": 3.4413445784712963e-07,
      "loss": 0.4417,
      "step": 407520
    },
    {
      "epoch": 166.82,
      "grad_norm": 2.0083954334259033,
      "learning_rate": 3.440517158428315e-07,
      "loss": 0.4409,
      "step": 407530
    },
    {
      "epoch": 166.82,
      "grad_norm": 2.269063949584961,
      "learning_rate": 3.439689829988408e-07,
      "loss": 0.449,
      "step": 407540
    },
    {
      "epoch": 166.82,
      "grad_norm": 2.0395987033843994,
      "learning_rate": 3.438862593155373e-07,
      "loss": 0.4582,
      "step": 407550
    },
    {
      "epoch": 166.83,
      "grad_norm": 2.449120044708252,
      "learning_rate": 3.4380354479329986e-07,
      "loss": 0.4498,
      "step": 407560
    },
    {
      "epoch": 166.83,
      "grad_norm": 2.029366970062256,
      "learning_rate": 3.437208394325074e-07,
      "loss": 0.4299,
      "step": 407570
    },
    {
      "epoch": 166.84,
      "grad_norm": 1.9946095943450928,
      "learning_rate": 3.4363814323353854e-07,
      "loss": 0.4446,
      "step": 407580
    },
    {
      "epoch": 166.84,
      "grad_norm": 2.3844432830810547,
      "learning_rate": 3.4355545619677276e-07,
      "loss": 0.4458,
      "step": 407590
    },
    {
      "epoch": 166.84,
      "grad_norm": 1.8902665376663208,
      "learning_rate": 3.434727783225876e-07,
      "loss": 0.4412,
      "step": 407600
    },
    {
      "epoch": 166.85,
      "grad_norm": 1.9371107816696167,
      "learning_rate": 3.433901096113631e-07,
      "loss": 0.435,
      "step": 407610
    },
    {
      "epoch": 166.85,
      "grad_norm": 2.404229164123535,
      "learning_rate": 3.4330745006347774e-07,
      "loss": 0.4555,
      "step": 407620
    },
    {
      "epoch": 166.86,
      "grad_norm": 2.2940824031829834,
      "learning_rate": 3.432247996793096e-07,
      "loss": 0.4489,
      "step": 407630
    },
    {
      "epoch": 166.86,
      "grad_norm": 2.325895309448242,
      "learning_rate": 3.4314215845923774e-07,
      "loss": 0.4428,
      "step": 407640
    },
    {
      "epoch": 166.86,
      "grad_norm": 2.2409181594848633,
      "learning_rate": 3.4305952640364077e-07,
      "loss": 0.4455,
      "step": 407650
    },
    {
      "epoch": 166.87,
      "grad_norm": 1.7085354328155518,
      "learning_rate": 3.429769035128973e-07,
      "loss": 0.4456,
      "step": 407660
    },
    {
      "epoch": 166.87,
      "grad_norm": 1.9411730766296387,
      "learning_rate": 3.4289428978738496e-07,
      "loss": 0.4222,
      "step": 407670
    },
    {
      "epoch": 166.88,
      "grad_norm": 2.1694259643554688,
      "learning_rate": 3.428116852274828e-07,
      "loss": 0.445,
      "step": 407680
    },
    {
      "epoch": 166.88,
      "grad_norm": 2.174011468887329,
      "learning_rate": 3.427290898335691e-07,
      "loss": 0.4451,
      "step": 407690
    },
    {
      "epoch": 166.88,
      "grad_norm": 6.171507358551025,
      "learning_rate": 3.426465036060224e-07,
      "loss": 0.4363,
      "step": 407700
    },
    {
      "epoch": 166.89,
      "grad_norm": 2.28334379196167,
      "learning_rate": 3.425639265452209e-07,
      "loss": 0.4426,
      "step": 407710
    },
    {
      "epoch": 166.89,
      "grad_norm": 2.0048367977142334,
      "learning_rate": 3.424813586515429e-07,
      "loss": 0.4495,
      "step": 407720
    },
    {
      "epoch": 166.9,
      "grad_norm": 2.3361682891845703,
      "learning_rate": 3.4239879992536664e-07,
      "loss": 0.4611,
      "step": 407730
    },
    {
      "epoch": 166.9,
      "grad_norm": 1.9879509210586548,
      "learning_rate": 3.423162503670703e-07,
      "loss": 0.4408,
      "step": 407740
    },
    {
      "epoch": 166.91,
      "grad_norm": 2.0040836334228516,
      "learning_rate": 3.422337099770319e-07,
      "loss": 0.4517,
      "step": 407750
    },
    {
      "epoch": 166.91,
      "grad_norm": 1.916877269744873,
      "learning_rate": 3.4215117875562953e-07,
      "loss": 0.4419,
      "step": 407760
    },
    {
      "epoch": 166.91,
      "grad_norm": 1.7380338907241821,
      "learning_rate": 3.4206865670324173e-07,
      "loss": 0.4313,
      "step": 407770
    },
    {
      "epoch": 166.92,
      "grad_norm": 1.8600983619689941,
      "learning_rate": 3.41986143820246e-07,
      "loss": 0.4477,
      "step": 407780
    },
    {
      "epoch": 166.92,
      "grad_norm": 2.066643476486206,
      "learning_rate": 3.419036401070199e-07,
      "loss": 0.4389,
      "step": 407790
    },
    {
      "epoch": 166.93,
      "grad_norm": 1.7733992338180542,
      "learning_rate": 3.4182114556394305e-07,
      "loss": 0.4524,
      "step": 407800
    },
    {
      "epoch": 166.93,
      "grad_norm": 1.8768184185028076,
      "learning_rate": 3.417386601913915e-07,
      "loss": 0.4411,
      "step": 407810
    },
    {
      "epoch": 166.93,
      "grad_norm": 1.6300019025802612,
      "learning_rate": 3.4165618398974405e-07,
      "loss": 0.466,
      "step": 407820
    },
    {
      "epoch": 166.94,
      "grad_norm": 2.3022513389587402,
      "learning_rate": 3.4157371695937836e-07,
      "loss": 0.4471,
      "step": 407830
    },
    {
      "epoch": 166.94,
      "grad_norm": 2.167588472366333,
      "learning_rate": 3.414912591006722e-07,
      "loss": 0.4617,
      "step": 407840
    },
    {
      "epoch": 166.95,
      "grad_norm": 2.1272363662719727,
      "learning_rate": 3.4140881041400303e-07,
      "loss": 0.4408,
      "step": 407850
    },
    {
      "epoch": 166.95,
      "grad_norm": 2.260554313659668,
      "learning_rate": 3.4132637089974885e-07,
      "loss": 0.4349,
      "step": 407860
    },
    {
      "epoch": 166.95,
      "grad_norm": 2.003337860107422,
      "learning_rate": 3.4124394055828715e-07,
      "loss": 0.4349,
      "step": 407870
    },
    {
      "epoch": 166.96,
      "grad_norm": 1.9659371376037598,
      "learning_rate": 3.411615193899957e-07,
      "loss": 0.4613,
      "step": 407880
    },
    {
      "epoch": 166.96,
      "grad_norm": 2.2107203006744385,
      "learning_rate": 3.410791073952519e-07,
      "loss": 0.4458,
      "step": 407890
    },
    {
      "epoch": 166.97,
      "grad_norm": 1.6819509267807007,
      "learning_rate": 3.4099670457443324e-07,
      "loss": 0.4354,
      "step": 407900
    },
    {
      "epoch": 166.97,
      "grad_norm": 2.130018711090088,
      "learning_rate": 3.409143109279173e-07,
      "loss": 0.4366,
      "step": 407910
    },
    {
      "epoch": 166.98,
      "grad_norm": 1.884307622909546,
      "learning_rate": 3.408319264560814e-07,
      "loss": 0.4502,
      "step": 407920
    },
    {
      "epoch": 166.98,
      "grad_norm": 1.973192811012268,
      "learning_rate": 3.4074955115930306e-07,
      "loss": 0.4509,
      "step": 407930
    },
    {
      "epoch": 166.98,
      "grad_norm": 2.2712693214416504,
      "learning_rate": 3.406671850379596e-07,
      "loss": 0.4437,
      "step": 407940
    },
    {
      "epoch": 166.99,
      "grad_norm": 2.3578765392303467,
      "learning_rate": 3.405848280924281e-07,
      "loss": 0.4489,
      "step": 407950
    },
    {
      "epoch": 166.99,
      "grad_norm": 2.204688310623169,
      "learning_rate": 3.405024803230864e-07,
      "loss": 0.45,
      "step": 407960
    },
    {
      "epoch": 167.0,
      "grad_norm": 2.2662715911865234,
      "learning_rate": 3.404201417303112e-07,
      "loss": 0.4554,
      "step": 407970
    },
    {
      "epoch": 167.0,
      "grad_norm": 2.055722951889038,
      "learning_rate": 3.403378123144798e-07,
      "loss": 0.4289,
      "step": 407980
    },
    {
      "epoch": 167.0,
      "eval_loss": 0.4465954899787903,
      "eval_runtime": 52.4417,
      "eval_samples_per_second": 65.768,
      "eval_steps_per_second": 8.238,
      "step": 407981
    },
    {
      "epoch": 167.0,
      "grad_norm": 1.8368618488311768,
      "learning_rate": 3.4025549207596934e-07,
      "loss": 0.4512,
      "step": 407990
    },
    {
      "epoch": 167.01,
      "grad_norm": 1.826123595237732,
      "learning_rate": 3.4017318101515676e-07,
      "loss": 0.457,
      "step": 408000
    },
    {
      "epoch": 167.01,
      "grad_norm": 2.096036672592163,
      "learning_rate": 3.4009087913241945e-07,
      "loss": 0.4606,
      "step": 408010
    },
    {
      "epoch": 167.02,
      "grad_norm": 1.8660680055618286,
      "learning_rate": 3.400085864281344e-07,
      "loss": 0.4448,
      "step": 408020
    },
    {
      "epoch": 167.02,
      "grad_norm": 1.9532703161239624,
      "learning_rate": 3.399263029026784e-07,
      "loss": 0.4393,
      "step": 408030
    },
    {
      "epoch": 167.02,
      "grad_norm": 1.9659652709960938,
      "learning_rate": 3.398440285564282e-07,
      "loss": 0.4357,
      "step": 408040
    },
    {
      "epoch": 167.03,
      "grad_norm": 2.5416033267974854,
      "learning_rate": 3.3976176338976116e-07,
      "loss": 0.4482,
      "step": 408050
    },
    {
      "epoch": 167.03,
      "grad_norm": 1.889351487159729,
      "learning_rate": 3.396795074030538e-07,
      "loss": 0.4436,
      "step": 408060
    },
    {
      "epoch": 167.04,
      "grad_norm": 1.7584222555160522,
      "learning_rate": 3.395972605966831e-07,
      "loss": 0.4409,
      "step": 408070
    },
    {
      "epoch": 167.04,
      "grad_norm": 2.2795193195343018,
      "learning_rate": 3.395150229710256e-07,
      "loss": 0.4632,
      "step": 408080
    },
    {
      "epoch": 167.04,
      "grad_norm": 1.98430335521698,
      "learning_rate": 3.394327945264582e-07,
      "loss": 0.4358,
      "step": 408090
    },
    {
      "epoch": 167.05,
      "grad_norm": 2.21553897857666,
      "learning_rate": 3.393505752633572e-07,
      "loss": 0.4379,
      "step": 408100
    },
    {
      "epoch": 167.05,
      "grad_norm": 2.0470612049102783,
      "learning_rate": 3.392683651821004e-07,
      "loss": 0.4486,
      "step": 408110
    },
    {
      "epoch": 167.06,
      "grad_norm": 2.4920108318328857,
      "learning_rate": 3.391861642830631e-07,
      "loss": 0.4451,
      "step": 408120
    },
    {
      "epoch": 167.06,
      "grad_norm": 1.9481585025787354,
      "learning_rate": 3.3910397256662186e-07,
      "loss": 0.4292,
      "step": 408130
    },
    {
      "epoch": 167.07,
      "grad_norm": 2.3230504989624023,
      "learning_rate": 3.390217900331541e-07,
      "loss": 0.4395,
      "step": 408140
    },
    {
      "epoch": 167.07,
      "grad_norm": 1.858993411064148,
      "learning_rate": 3.3893961668303543e-07,
      "loss": 0.4388,
      "step": 408150
    },
    {
      "epoch": 167.07,
      "grad_norm": 1.6702690124511719,
      "learning_rate": 3.3885745251664216e-07,
      "loss": 0.4359,
      "step": 408160
    },
    {
      "epoch": 167.08,
      "grad_norm": 2.25766658782959,
      "learning_rate": 3.3877529753435185e-07,
      "loss": 0.4459,
      "step": 408170
    },
    {
      "epoch": 167.08,
      "grad_norm": 2.2893316745758057,
      "learning_rate": 3.386931517365402e-07,
      "loss": 0.4423,
      "step": 408180
    },
    {
      "epoch": 167.09,
      "grad_norm": 2.4439499378204346,
      "learning_rate": 3.386110151235833e-07,
      "loss": 0.4489,
      "step": 408190
    },
    {
      "epoch": 167.09,
      "grad_norm": 2.107480764389038,
      "learning_rate": 3.385288876958576e-07,
      "loss": 0.4326,
      "step": 408200
    },
    {
      "epoch": 167.09,
      "grad_norm": 2.0443153381347656,
      "learning_rate": 3.3844676945373913e-07,
      "loss": 0.4449,
      "step": 408210
    },
    {
      "epoch": 167.1,
      "grad_norm": 2.5729994773864746,
      "learning_rate": 3.383646603976044e-07,
      "loss": 0.4257,
      "step": 408220
    },
    {
      "epoch": 167.1,
      "grad_norm": 2.0244498252868652,
      "learning_rate": 3.3828256052782917e-07,
      "loss": 0.4288,
      "step": 408230
    },
    {
      "epoch": 167.11,
      "grad_norm": 1.7095903158187866,
      "learning_rate": 3.3820046984478964e-07,
      "loss": 0.454,
      "step": 408240
    },
    {
      "epoch": 167.11,
      "grad_norm": 1.8822134733200073,
      "learning_rate": 3.381183883488627e-07,
      "loss": 0.4642,
      "step": 408250
    },
    {
      "epoch": 167.11,
      "grad_norm": 1.7388888597488403,
      "learning_rate": 3.3803631604042296e-07,
      "loss": 0.4283,
      "step": 408260
    },
    {
      "epoch": 167.12,
      "grad_norm": 2.1649818420410156,
      "learning_rate": 3.3795425291984697e-07,
      "loss": 0.4417,
      "step": 408270
    },
    {
      "epoch": 167.12,
      "grad_norm": 2.007312774658203,
      "learning_rate": 3.378721989875105e-07,
      "loss": 0.4605,
      "step": 408280
    },
    {
      "epoch": 167.13,
      "grad_norm": 1.8390674591064453,
      "learning_rate": 3.377901542437897e-07,
      "loss": 0.441,
      "step": 408290
    },
    {
      "epoch": 167.13,
      "grad_norm": 2.2887704372406006,
      "learning_rate": 3.377081186890602e-07,
      "loss": 0.435,
      "step": 408300
    },
    {
      "epoch": 167.13,
      "grad_norm": 1.801340103149414,
      "learning_rate": 3.3762609232369793e-07,
      "loss": 0.4525,
      "step": 408310
    },
    {
      "epoch": 167.14,
      "grad_norm": 2.248414993286133,
      "learning_rate": 3.3754407514807894e-07,
      "loss": 0.4473,
      "step": 408320
    },
    {
      "epoch": 167.14,
      "grad_norm": 2.6253719329833984,
      "learning_rate": 3.3746206716257806e-07,
      "loss": 0.4423,
      "step": 408330
    },
    {
      "epoch": 167.15,
      "grad_norm": 1.8297916650772095,
      "learning_rate": 3.373800683675717e-07,
      "loss": 0.4368,
      "step": 408340
    },
    {
      "epoch": 167.15,
      "grad_norm": 1.8045060634613037,
      "learning_rate": 3.3729807876343476e-07,
      "loss": 0.4604,
      "step": 408350
    },
    {
      "epoch": 167.16,
      "grad_norm": 1.7855464220046997,
      "learning_rate": 3.37216098350544e-07,
      "loss": 0.4418,
      "step": 408360
    },
    {
      "epoch": 167.16,
      "grad_norm": 2.2981598377227783,
      "learning_rate": 3.3713412712927416e-07,
      "loss": 0.482,
      "step": 408370
    },
    {
      "epoch": 167.16,
      "grad_norm": 1.7940139770507812,
      "learning_rate": 3.3705216510000143e-07,
      "loss": 0.4214,
      "step": 408380
    },
    {
      "epoch": 167.17,
      "grad_norm": 1.8198416233062744,
      "learning_rate": 3.369702122631001e-07,
      "loss": 0.4548,
      "step": 408390
    },
    {
      "epoch": 167.17,
      "grad_norm": 2.0490849018096924,
      "learning_rate": 3.368882686189461e-07,
      "loss": 0.4356,
      "step": 408400
    },
    {
      "epoch": 167.18,
      "grad_norm": 1.7872601747512817,
      "learning_rate": 3.368063341679152e-07,
      "loss": 0.4552,
      "step": 408410
    },
    {
      "epoch": 167.18,
      "grad_norm": 2.235788106918335,
      "learning_rate": 3.3672440891038203e-07,
      "loss": 0.4387,
      "step": 408420
    },
    {
      "epoch": 167.18,
      "grad_norm": 2.158726215362549,
      "learning_rate": 3.366424928467227e-07,
      "loss": 0.4641,
      "step": 408430
    },
    {
      "epoch": 167.19,
      "grad_norm": 2.149895668029785,
      "learning_rate": 3.365605859773116e-07,
      "loss": 0.441,
      "step": 408440
    },
    {
      "epoch": 167.19,
      "grad_norm": 1.8879562616348267,
      "learning_rate": 3.364786883025247e-07,
      "loss": 0.4431,
      "step": 408450
    },
    {
      "epoch": 167.2,
      "grad_norm": 2.0040366649627686,
      "learning_rate": 3.363967998227365e-07,
      "loss": 0.4564,
      "step": 408460
    },
    {
      "epoch": 167.2,
      "grad_norm": 2.002751588821411,
      "learning_rate": 3.3631492053832266e-07,
      "loss": 0.4436,
      "step": 408470
    },
    {
      "epoch": 167.2,
      "grad_norm": 2.169564962387085,
      "learning_rate": 3.3623305044965796e-07,
      "loss": 0.443,
      "step": 408480
    },
    {
      "epoch": 167.21,
      "grad_norm": 1.8469372987747192,
      "learning_rate": 3.3615118955711763e-07,
      "loss": 0.4501,
      "step": 408490
    },
    {
      "epoch": 167.21,
      "grad_norm": 1.7225964069366455,
      "learning_rate": 3.3606933786107627e-07,
      "loss": 0.446,
      "step": 408500
    },
    {
      "epoch": 167.22,
      "grad_norm": 5.675680160522461,
      "learning_rate": 3.3598749536190917e-07,
      "loss": 0.4516,
      "step": 408510
    },
    {
      "epoch": 167.22,
      "grad_norm": 2.2887022495269775,
      "learning_rate": 3.359056620599913e-07,
      "loss": 0.4673,
      "step": 408520
    },
    {
      "epoch": 167.22,
      "grad_norm": 2.171067953109741,
      "learning_rate": 3.3582383795569737e-07,
      "loss": 0.4574,
      "step": 408530
    },
    {
      "epoch": 167.23,
      "grad_norm": 2.315417528152466,
      "learning_rate": 3.357420230494023e-07,
      "loss": 0.4331,
      "step": 408540
    },
    {
      "epoch": 167.23,
      "grad_norm": 2.403801679611206,
      "learning_rate": 3.356602173414805e-07,
      "loss": 0.446,
      "step": 408550
    },
    {
      "epoch": 167.24,
      "grad_norm": 2.307513475418091,
      "learning_rate": 3.3557842083230726e-07,
      "loss": 0.4414,
      "step": 408560
    },
    {
      "epoch": 167.24,
      "grad_norm": 2.381561040878296,
      "learning_rate": 3.35496633522257e-07,
      "loss": 0.4598,
      "step": 408570
    },
    {
      "epoch": 167.25,
      "grad_norm": 1.9753150939941406,
      "learning_rate": 3.354148554117041e-07,
      "loss": 0.4485,
      "step": 408580
    },
    {
      "epoch": 167.25,
      "grad_norm": 2.184013605117798,
      "learning_rate": 3.3533308650102364e-07,
      "loss": 0.4482,
      "step": 408590
    },
    {
      "epoch": 167.25,
      "grad_norm": 2.6683197021484375,
      "learning_rate": 3.3525132679059025e-07,
      "loss": 0.4351,
      "step": 408600
    },
    {
      "epoch": 167.26,
      "grad_norm": 1.8407262563705444,
      "learning_rate": 3.351695762807781e-07,
      "loss": 0.4246,
      "step": 408610
    },
    {
      "epoch": 167.26,
      "grad_norm": 1.9550303220748901,
      "learning_rate": 3.350878349719617e-07,
      "loss": 0.466,
      "step": 408620
    },
    {
      "epoch": 167.27,
      "grad_norm": 2.3713817596435547,
      "learning_rate": 3.350061028645156e-07,
      "loss": 0.4483,
      "step": 408630
    },
    {
      "epoch": 167.27,
      "grad_norm": 2.1690096855163574,
      "learning_rate": 3.349243799588141e-07,
      "loss": 0.4631,
      "step": 408640
    },
    {
      "epoch": 167.27,
      "grad_norm": 1.8886810541152954,
      "learning_rate": 3.348426662552319e-07,
      "loss": 0.4672,
      "step": 408650
    },
    {
      "epoch": 167.28,
      "grad_norm": 1.9731167554855347,
      "learning_rate": 3.347609617541429e-07,
      "loss": 0.4469,
      "step": 408660
    },
    {
      "epoch": 167.28,
      "grad_norm": 2.076613426208496,
      "learning_rate": 3.3467926645592174e-07,
      "loss": 0.4416,
      "step": 408670
    },
    {
      "epoch": 167.29,
      "grad_norm": 2.3774209022521973,
      "learning_rate": 3.3459758036094245e-07,
      "loss": 0.4435,
      "step": 408680
    },
    {
      "epoch": 167.29,
      "grad_norm": 1.922133207321167,
      "learning_rate": 3.345159034695797e-07,
      "loss": 0.4464,
      "step": 408690
    },
    {
      "epoch": 167.29,
      "grad_norm": 1.620200753211975,
      "learning_rate": 3.344342357822065e-07,
      "loss": 0.452,
      "step": 408700
    },
    {
      "epoch": 167.3,
      "grad_norm": 1.9865127801895142,
      "learning_rate": 3.343525772991977e-07,
      "loss": 0.4378,
      "step": 408710
    },
    {
      "epoch": 167.3,
      "grad_norm": 2.2001540660858154,
      "learning_rate": 3.342709280209275e-07,
      "loss": 0.4239,
      "step": 408720
    },
    {
      "epoch": 167.31,
      "grad_norm": 2.1398661136627197,
      "learning_rate": 3.341892879477689e-07,
      "loss": 0.4293,
      "step": 408730
    },
    {
      "epoch": 167.31,
      "grad_norm": 1.7144887447357178,
      "learning_rate": 3.341076570800975e-07,
      "loss": 0.4394,
      "step": 408740
    },
    {
      "epoch": 167.31,
      "grad_norm": 2.0676567554473877,
      "learning_rate": 3.3402603541828636e-07,
      "loss": 0.4609,
      "step": 408750
    },
    {
      "epoch": 167.32,
      "grad_norm": 1.9669321775436401,
      "learning_rate": 3.339444229627094e-07,
      "loss": 0.4513,
      "step": 408760
    },
    {
      "epoch": 167.32,
      "grad_norm": 2.256901741027832,
      "learning_rate": 3.338628197137404e-07,
      "loss": 0.4274,
      "step": 408770
    },
    {
      "epoch": 167.33,
      "grad_norm": 2.210157632827759,
      "learning_rate": 3.337812256717537e-07,
      "loss": 0.4426,
      "step": 408780
    },
    {
      "epoch": 167.33,
      "grad_norm": 1.7740287780761719,
      "learning_rate": 3.336996408371224e-07,
      "loss": 0.4433,
      "step": 408790
    },
    {
      "epoch": 167.34,
      "grad_norm": 1.9718029499053955,
      "learning_rate": 3.3361806521022046e-07,
      "loss": 0.4459,
      "step": 408800
    },
    {
      "epoch": 167.34,
      "grad_norm": 2.1226601600646973,
      "learning_rate": 3.335364987914216e-07,
      "loss": 0.4307,
      "step": 408810
    },
    {
      "epoch": 167.34,
      "grad_norm": 2.3693737983703613,
      "learning_rate": 3.3345494158109964e-07,
      "loss": 0.4492,
      "step": 408820
    },
    {
      "epoch": 167.35,
      "grad_norm": 2.166844367980957,
      "learning_rate": 3.333733935796283e-07,
      "loss": 0.4488,
      "step": 408830
    },
    {
      "epoch": 167.35,
      "grad_norm": 1.9371615648269653,
      "learning_rate": 3.332918547873805e-07,
      "loss": 0.4511,
      "step": 408840
    },
    {
      "epoch": 167.36,
      "grad_norm": 2.7630999088287354,
      "learning_rate": 3.3321032520473005e-07,
      "loss": 0.4633,
      "step": 408850
    },
    {
      "epoch": 167.36,
      "grad_norm": 1.9805291891098022,
      "learning_rate": 3.3312880483205044e-07,
      "loss": 0.4407,
      "step": 408860
    },
    {
      "epoch": 167.36,
      "grad_norm": 3.649980068206787,
      "learning_rate": 3.330472936697153e-07,
      "loss": 0.4465,
      "step": 408870
    },
    {
      "epoch": 167.37,
      "grad_norm": 1.893393635749817,
      "learning_rate": 3.329657917180977e-07,
      "loss": 0.4513,
      "step": 408880
    },
    {
      "epoch": 167.37,
      "grad_norm": 3.5127508640289307,
      "learning_rate": 3.328842989775712e-07,
      "loss": 0.465,
      "step": 408890
    },
    {
      "epoch": 167.38,
      "grad_norm": 1.9853215217590332,
      "learning_rate": 3.3280281544850904e-07,
      "loss": 0.4703,
      "step": 408900
    },
    {
      "epoch": 167.38,
      "grad_norm": 2.612293004989624,
      "learning_rate": 3.3272134113128447e-07,
      "loss": 0.4367,
      "step": 408910
    },
    {
      "epoch": 167.38,
      "grad_norm": 2.2113595008850098,
      "learning_rate": 3.3263987602627027e-07,
      "loss": 0.4368,
      "step": 408920
    },
    {
      "epoch": 167.39,
      "grad_norm": 2.1319386959075928,
      "learning_rate": 3.325584201338407e-07,
      "loss": 0.4255,
      "step": 408930
    },
    {
      "epoch": 167.39,
      "grad_norm": 2.087082624435425,
      "learning_rate": 3.3247697345436806e-07,
      "loss": 0.4372,
      "step": 408940
    },
    {
      "epoch": 167.4,
      "grad_norm": 1.9409693479537964,
      "learning_rate": 3.323955359882258e-07,
      "loss": 0.461,
      "step": 408950
    },
    {
      "epoch": 167.4,
      "grad_norm": 2.314972400665283,
      "learning_rate": 3.3231410773578683e-07,
      "loss": 0.4449,
      "step": 408960
    },
    {
      "epoch": 167.4,
      "grad_norm": 2.0261011123657227,
      "learning_rate": 3.3223268869742466e-07,
      "loss": 0.4334,
      "step": 408970
    },
    {
      "epoch": 167.41,
      "grad_norm": 1.8952559232711792,
      "learning_rate": 3.321512788735113e-07,
      "loss": 0.4423,
      "step": 408980
    },
    {
      "epoch": 167.41,
      "grad_norm": 2.03128719329834,
      "learning_rate": 3.320698782644203e-07,
      "loss": 0.4455,
      "step": 408990
    },
    {
      "epoch": 167.42,
      "grad_norm": 2.0151426792144775,
      "learning_rate": 3.3198848687052404e-07,
      "loss": 0.4482,
      "step": 409000
    },
    {
      "epoch": 167.42,
      "grad_norm": 4.816678047180176,
      "learning_rate": 3.31907104692196e-07,
      "loss": 0.4363,
      "step": 409010
    },
    {
      "epoch": 167.43,
      "grad_norm": 1.8930628299713135,
      "learning_rate": 3.3182573172980867e-07,
      "loss": 0.44,
      "step": 409020
    },
    {
      "epoch": 167.43,
      "grad_norm": 2.23213791847229,
      "learning_rate": 3.3174436798373474e-07,
      "loss": 0.4434,
      "step": 409030
    },
    {
      "epoch": 167.43,
      "grad_norm": 2.1604745388031006,
      "learning_rate": 3.316630134543468e-07,
      "loss": 0.4379,
      "step": 409040
    },
    {
      "epoch": 167.44,
      "grad_norm": 2.344728708267212,
      "learning_rate": 3.3158166814201807e-07,
      "loss": 0.4376,
      "step": 409050
    },
    {
      "epoch": 167.44,
      "grad_norm": 1.9564872980117798,
      "learning_rate": 3.3150033204712077e-07,
      "loss": 0.4393,
      "step": 409060
    },
    {
      "epoch": 167.45,
      "grad_norm": 2.2185304164886475,
      "learning_rate": 3.314190051700274e-07,
      "loss": 0.4453,
      "step": 409070
    },
    {
      "epoch": 167.45,
      "grad_norm": 1.8876566886901855,
      "learning_rate": 3.31337687511111e-07,
      "loss": 0.4449,
      "step": 409080
    },
    {
      "epoch": 167.45,
      "grad_norm": 1.8801095485687256,
      "learning_rate": 3.312563790707434e-07,
      "loss": 0.4561,
      "step": 409090
    },
    {
      "epoch": 167.46,
      "grad_norm": 2.15691876411438,
      "learning_rate": 3.3117507984929756e-07,
      "loss": 0.4546,
      "step": 409100
    },
    {
      "epoch": 167.46,
      "grad_norm": 2.163606882095337,
      "learning_rate": 3.3109378984714583e-07,
      "loss": 0.4381,
      "step": 409110
    },
    {
      "epoch": 167.47,
      "grad_norm": 2.3633453845977783,
      "learning_rate": 3.3101250906466026e-07,
      "loss": 0.4436,
      "step": 409120
    },
    {
      "epoch": 167.47,
      "grad_norm": 1.8971810340881348,
      "learning_rate": 3.3093123750221346e-07,
      "loss": 0.4545,
      "step": 409130
    },
    {
      "epoch": 167.47,
      "grad_norm": 2.4039456844329834,
      "learning_rate": 3.3084997516017784e-07,
      "loss": 0.4374,
      "step": 409140
    },
    {
      "epoch": 167.48,
      "grad_norm": 2.0999717712402344,
      "learning_rate": 3.3076872203892546e-07,
      "loss": 0.4751,
      "step": 409150
    },
    {
      "epoch": 167.48,
      "grad_norm": 1.73282790184021,
      "learning_rate": 3.3068747813882866e-07,
      "loss": 0.4389,
      "step": 409160
    },
    {
      "epoch": 167.49,
      "grad_norm": 1.7960500717163086,
      "learning_rate": 3.3060624346025933e-07,
      "loss": 0.4415,
      "step": 409170
    },
    {
      "epoch": 167.49,
      "grad_norm": 2.097332000732422,
      "learning_rate": 3.3052501800358985e-07,
      "loss": 0.4599,
      "step": 409180
    },
    {
      "epoch": 167.49,
      "grad_norm": 2.431893825531006,
      "learning_rate": 3.304438017691922e-07,
      "loss": 0.439,
      "step": 409190
    },
    {
      "epoch": 167.5,
      "grad_norm": 2.4964699745178223,
      "learning_rate": 3.3036259475743834e-07,
      "loss": 0.4243,
      "step": 409200
    },
    {
      "epoch": 167.5,
      "grad_norm": 2.181854009628296,
      "learning_rate": 3.3028139696870055e-07,
      "loss": 0.4262,
      "step": 409210
    },
    {
      "epoch": 167.51,
      "grad_norm": 2.3210389614105225,
      "learning_rate": 3.3020020840335046e-07,
      "loss": 0.4335,
      "step": 409220
    },
    {
      "epoch": 167.51,
      "grad_norm": 1.9459446668624878,
      "learning_rate": 3.3011902906176016e-07,
      "loss": 0.4446,
      "step": 409230
    },
    {
      "epoch": 167.52,
      "grad_norm": 1.8061805963516235,
      "learning_rate": 3.3003785894430173e-07,
      "loss": 0.4258,
      "step": 409240
    },
    {
      "epoch": 167.52,
      "grad_norm": 2.0273609161376953,
      "learning_rate": 3.2995669805134646e-07,
      "loss": 0.4477,
      "step": 409250
    },
    {
      "epoch": 167.52,
      "grad_norm": 2.0562005043029785,
      "learning_rate": 3.298755463832665e-07,
      "loss": 0.4461,
      "step": 409260
    },
    {
      "epoch": 167.53,
      "grad_norm": 1.658570647239685,
      "learning_rate": 3.297944039404337e-07,
      "loss": 0.4514,
      "step": 409270
    },
    {
      "epoch": 167.53,
      "grad_norm": 2.233964204788208,
      "learning_rate": 3.297132707232198e-07,
      "loss": 0.4446,
      "step": 409280
    },
    {
      "epoch": 167.54,
      "grad_norm": 1.907636284828186,
      "learning_rate": 3.2963214673199547e-07,
      "loss": 0.465,
      "step": 409290
    },
    {
      "epoch": 167.54,
      "grad_norm": 1.9411920309066772,
      "learning_rate": 3.295510319671335e-07,
      "loss": 0.4273,
      "step": 409300
    },
    {
      "epoch": 167.54,
      "grad_norm": 2.1147351264953613,
      "learning_rate": 3.29469926429005e-07,
      "loss": 0.4417,
      "step": 409310
    },
    {
      "epoch": 167.55,
      "grad_norm": 1.9081491231918335,
      "learning_rate": 3.293888301179817e-07,
      "loss": 0.4579,
      "step": 409320
    },
    {
      "epoch": 167.55,
      "grad_norm": 1.816421389579773,
      "learning_rate": 3.29307743034435e-07,
      "loss": 0.4229,
      "step": 409330
    },
    {
      "epoch": 167.56,
      "grad_norm": 1.9345537424087524,
      "learning_rate": 3.2922666517873604e-07,
      "loss": 0.4492,
      "step": 409340
    },
    {
      "epoch": 167.56,
      "grad_norm": 1.9796758890151978,
      "learning_rate": 3.2914559655125674e-07,
      "loss": 0.4457,
      "step": 409350
    },
    {
      "epoch": 167.56,
      "grad_norm": 1.9948757886886597,
      "learning_rate": 3.2906453715236805e-07,
      "loss": 0.4686,
      "step": 409360
    },
    {
      "epoch": 167.57,
      "grad_norm": 1.8909252882003784,
      "learning_rate": 3.289834869824417e-07,
      "loss": 0.4409,
      "step": 409370
    },
    {
      "epoch": 167.57,
      "grad_norm": 2.741520643234253,
      "learning_rate": 3.289024460418486e-07,
      "loss": 0.4542,
      "step": 409380
    },
    {
      "epoch": 167.58,
      "grad_norm": 1.935408592224121,
      "learning_rate": 3.2882141433095996e-07,
      "loss": 0.4504,
      "step": 409390
    },
    {
      "epoch": 167.58,
      "grad_norm": 1.8962754011154175,
      "learning_rate": 3.287403918501471e-07,
      "loss": 0.4295,
      "step": 409400
    },
    {
      "epoch": 167.58,
      "grad_norm": 1.7685494422912598,
      "learning_rate": 3.2865937859978197e-07,
      "loss": 0.4474,
      "step": 409410
    },
    {
      "epoch": 167.59,
      "grad_norm": 1.8370580673217773,
      "learning_rate": 3.285783745802342e-07,
      "loss": 0.4241,
      "step": 409420
    },
    {
      "epoch": 167.59,
      "grad_norm": 2.2085800170898438,
      "learning_rate": 3.284973797918757e-07,
      "loss": 0.4349,
      "step": 409430
    },
    {
      "epoch": 167.6,
      "grad_norm": 1.8295387029647827,
      "learning_rate": 3.2841639423507723e-07,
      "loss": 0.4371,
      "step": 409440
    },
    {
      "epoch": 167.6,
      "grad_norm": 2.0753567218780518,
      "learning_rate": 3.283354179102099e-07,
      "loss": 0.4582,
      "step": 409450
    },
    {
      "epoch": 167.61,
      "grad_norm": 2.682279586791992,
      "learning_rate": 3.2825445081764444e-07,
      "loss": 0.4355,
      "step": 409460
    },
    {
      "epoch": 167.61,
      "grad_norm": 1.9912680387496948,
      "learning_rate": 3.281734929577522e-07,
      "loss": 0.4372,
      "step": 409470
    },
    {
      "epoch": 167.61,
      "grad_norm": 2.511084794998169,
      "learning_rate": 3.280925443309032e-07,
      "loss": 0.4286,
      "step": 409480
    },
    {
      "epoch": 167.62,
      "grad_norm": 3.357051134109497,
      "learning_rate": 3.280116049374695e-07,
      "loss": 0.4266,
      "step": 409490
    },
    {
      "epoch": 167.62,
      "grad_norm": 1.784541368484497,
      "learning_rate": 3.2793067477782126e-07,
      "loss": 0.4579,
      "step": 409500
    },
    {
      "epoch": 167.63,
      "grad_norm": 2.015129327774048,
      "learning_rate": 3.2784975385232883e-07,
      "loss": 0.4676,
      "step": 409510
    },
    {
      "epoch": 167.63,
      "grad_norm": 1.904378890991211,
      "learning_rate": 3.2776884216136333e-07,
      "loss": 0.4191,
      "step": 409520
    },
    {
      "epoch": 167.63,
      "grad_norm": 1.6397382020950317,
      "learning_rate": 3.2768793970529536e-07,
      "loss": 0.4822,
      "step": 409530
    },
    {
      "epoch": 167.64,
      "grad_norm": 1.9129186868667603,
      "learning_rate": 3.2760704648449563e-07,
      "loss": 0.4398,
      "step": 409540
    },
    {
      "epoch": 167.64,
      "grad_norm": 1.8787868022918701,
      "learning_rate": 3.27526162499335e-07,
      "loss": 0.4386,
      "step": 409550
    },
    {
      "epoch": 167.65,
      "grad_norm": 1.9057080745697021,
      "learning_rate": 3.2744528775018266e-07,
      "loss": 0.432,
      "step": 409560
    },
    {
      "epoch": 167.65,
      "grad_norm": 1.7784228324890137,
      "learning_rate": 3.2736442223741025e-07,
      "loss": 0.4464,
      "step": 409570
    },
    {
      "epoch": 167.65,
      "grad_norm": 2.2633705139160156,
      "learning_rate": 3.2728356596138797e-07,
      "loss": 0.4641,
      "step": 409580
    },
    {
      "epoch": 167.66,
      "grad_norm": 1.9560613632202148,
      "learning_rate": 3.272027189224859e-07,
      "loss": 0.451,
      "step": 409590
    },
    {
      "epoch": 167.66,
      "grad_norm": 1.73734712600708,
      "learning_rate": 3.2712188112107476e-07,
      "loss": 0.445,
      "step": 409600
    },
    {
      "epoch": 167.67,
      "grad_norm": 1.9072346687316895,
      "learning_rate": 3.2704105255752466e-07,
      "loss": 0.4498,
      "step": 409610
    },
    {
      "epoch": 167.67,
      "grad_norm": 1.7245901823043823,
      "learning_rate": 3.2696023323220583e-07,
      "loss": 0.446,
      "step": 409620
    },
    {
      "epoch": 167.67,
      "grad_norm": 1.8508483171463013,
      "learning_rate": 3.268794231454888e-07,
      "loss": 0.4233,
      "step": 409630
    },
    {
      "epoch": 167.68,
      "grad_norm": 2.083977460861206,
      "learning_rate": 3.267986222977434e-07,
      "loss": 0.4542,
      "step": 409640
    },
    {
      "epoch": 167.68,
      "grad_norm": 2.06427001953125,
      "learning_rate": 3.2671783068933973e-07,
      "loss": 0.4603,
      "step": 409650
    },
    {
      "epoch": 167.69,
      "grad_norm": 2.1619651317596436,
      "learning_rate": 3.2663704832064824e-07,
      "loss": 0.4399,
      "step": 409660
    },
    {
      "epoch": 167.69,
      "grad_norm": 1.7909486293792725,
      "learning_rate": 3.265562751920384e-07,
      "loss": 0.4504,
      "step": 409670
    },
    {
      "epoch": 167.7,
      "grad_norm": 2.1484599113464355,
      "learning_rate": 3.2647551130388145e-07,
      "loss": 0.4508,
      "step": 409680
    },
    {
      "epoch": 167.7,
      "grad_norm": 1.8535492420196533,
      "learning_rate": 3.2639475665654614e-07,
      "loss": 0.4554,
      "step": 409690
    },
    {
      "epoch": 167.7,
      "grad_norm": 2.218322515487671,
      "learning_rate": 3.2631401125040253e-07,
      "loss": 0.4554,
      "step": 409700
    },
    {
      "epoch": 167.71,
      "grad_norm": 1.9220163822174072,
      "learning_rate": 3.262332750858211e-07,
      "loss": 0.4461,
      "step": 409710
    },
    {
      "epoch": 167.71,
      "grad_norm": 1.9958661794662476,
      "learning_rate": 3.2615254816317113e-07,
      "loss": 0.442,
      "step": 409720
    },
    {
      "epoch": 167.72,
      "grad_norm": 1.8825187683105469,
      "learning_rate": 3.2607183048282257e-07,
      "loss": 0.4392,
      "step": 409730
    },
    {
      "epoch": 167.72,
      "grad_norm": 1.661250352859497,
      "learning_rate": 3.259911220451453e-07,
      "loss": 0.4436,
      "step": 409740
    },
    {
      "epoch": 167.72,
      "grad_norm": 1.8580416440963745,
      "learning_rate": 3.2591042285050894e-07,
      "loss": 0.4394,
      "step": 409750
    },
    {
      "epoch": 167.73,
      "grad_norm": 1.8798232078552246,
      "learning_rate": 3.258297328992833e-07,
      "loss": 0.439,
      "step": 409760
    },
    {
      "epoch": 167.73,
      "grad_norm": 1.9469224214553833,
      "learning_rate": 3.2574905219183763e-07,
      "loss": 0.4293,
      "step": 409770
    },
    {
      "epoch": 167.74,
      "grad_norm": 1.8819442987442017,
      "learning_rate": 3.2566838072854196e-07,
      "loss": 0.428,
      "step": 409780
    },
    {
      "epoch": 167.74,
      "grad_norm": 2.3907666206359863,
      "learning_rate": 3.2558771850976574e-07,
      "loss": 0.4488,
      "step": 409790
    },
    {
      "epoch": 167.74,
      "grad_norm": 2.0011274814605713,
      "learning_rate": 3.255070655358782e-07,
      "loss": 0.4473,
      "step": 409800
    },
    {
      "epoch": 167.75,
      "grad_norm": 1.9752463102340698,
      "learning_rate": 3.2542642180724893e-07,
      "loss": 0.4426,
      "step": 409810
    },
    {
      "epoch": 167.75,
      "grad_norm": 2.0477194786071777,
      "learning_rate": 3.253457873242476e-07,
      "loss": 0.4526,
      "step": 409820
    },
    {
      "epoch": 167.76,
      "grad_norm": 2.076455593109131,
      "learning_rate": 3.252651620872432e-07,
      "loss": 0.4353,
      "step": 409830
    },
    {
      "epoch": 167.76,
      "grad_norm": 1.5715434551239014,
      "learning_rate": 3.2518454609660546e-07,
      "loss": 0.4282,
      "step": 409840
    },
    {
      "epoch": 167.77,
      "grad_norm": 1.9116498231887817,
      "learning_rate": 3.2510393935270316e-07,
      "loss": 0.4429,
      "step": 409850
    },
    {
      "epoch": 167.77,
      "grad_norm": 2.0146210193634033,
      "learning_rate": 3.25023341855906e-07,
      "loss": 0.4447,
      "step": 409860
    },
    {
      "epoch": 167.77,
      "grad_norm": 2.4499247074127197,
      "learning_rate": 3.2494275360658304e-07,
      "loss": 0.4369,
      "step": 409870
    },
    {
      "epoch": 167.78,
      "grad_norm": 1.8352357149124146,
      "learning_rate": 3.248621746051034e-07,
      "loss": 0.4485,
      "step": 409880
    },
    {
      "epoch": 167.78,
      "grad_norm": 2.1068880558013916,
      "learning_rate": 3.247816048518364e-07,
      "loss": 0.4464,
      "step": 409890
    },
    {
      "epoch": 167.79,
      "grad_norm": 2.0262458324432373,
      "learning_rate": 3.2470104434715075e-07,
      "loss": 0.4271,
      "step": 409900
    },
    {
      "epoch": 167.79,
      "grad_norm": 1.8564497232437134,
      "learning_rate": 3.246204930914156e-07,
      "loss": 0.4621,
      "step": 409910
    },
    {
      "epoch": 167.79,
      "grad_norm": 1.7842886447906494,
      "learning_rate": 3.24539951085e-07,
      "loss": 0.4527,
      "step": 409920
    },
    {
      "epoch": 167.8,
      "grad_norm": 2.4223835468292236,
      "learning_rate": 3.244594183282732e-07,
      "loss": 0.4344,
      "step": 409930
    },
    {
      "epoch": 167.8,
      "grad_norm": 2.001984119415283,
      "learning_rate": 3.2437889482160365e-07,
      "loss": 0.4404,
      "step": 409940
    },
    {
      "epoch": 167.81,
      "grad_norm": 2.7453670501708984,
      "learning_rate": 3.242983805653603e-07,
      "loss": 0.4323,
      "step": 409950
    },
    {
      "epoch": 167.81,
      "grad_norm": 2.0278217792510986,
      "learning_rate": 3.242178755599121e-07,
      "loss": 0.4605,
      "step": 409960
    },
    {
      "epoch": 167.81,
      "grad_norm": 2.284691333770752,
      "learning_rate": 3.241373798056281e-07,
      "loss": 0.4365,
      "step": 409970
    },
    {
      "epoch": 167.82,
      "grad_norm": 2.1632442474365234,
      "learning_rate": 3.2405689330287644e-07,
      "loss": 0.4595,
      "step": 409980
    },
    {
      "epoch": 167.82,
      "grad_norm": 1.851858377456665,
      "learning_rate": 3.239764160520266e-07,
      "loss": 0.4446,
      "step": 409990
    },
    {
      "epoch": 167.83,
      "grad_norm": 1.896783471107483,
      "learning_rate": 3.238959480534463e-07,
      "loss": 0.4387,
      "step": 410000
    },
    {
      "epoch": 167.83,
      "grad_norm": 2.0184998512268066,
      "learning_rate": 3.2381548930750477e-07,
      "loss": 0.46,
      "step": 410010
    },
    {
      "epoch": 167.83,
      "grad_norm": 2.268463611602783,
      "learning_rate": 3.2373503981457016e-07,
      "loss": 0.4306,
      "step": 410020
    },
    {
      "epoch": 167.84,
      "grad_norm": 1.8381080627441406,
      "learning_rate": 3.2365459957501125e-07,
      "loss": 0.4472,
      "step": 410030
    },
    {
      "epoch": 167.84,
      "grad_norm": 2.1764585971832275,
      "learning_rate": 3.2357416858919676e-07,
      "loss": 0.4541,
      "step": 410040
    },
    {
      "epoch": 167.85,
      "grad_norm": 2.045123338699341,
      "learning_rate": 3.2349374685749426e-07,
      "loss": 0.447,
      "step": 410050
    },
    {
      "epoch": 167.85,
      "grad_norm": 2.1673502922058105,
      "learning_rate": 3.234133343802731e-07,
      "loss": 0.4575,
      "step": 410060
    },
    {
      "epoch": 167.86,
      "grad_norm": 1.8226045370101929,
      "learning_rate": 3.233329311579017e-07,
      "loss": 0.4426,
      "step": 410070
    },
    {
      "epoch": 167.86,
      "grad_norm": 2.404527187347412,
      "learning_rate": 3.2325253719074767e-07,
      "loss": 0.4227,
      "step": 410080
    },
    {
      "epoch": 167.86,
      "grad_norm": 2.1017253398895264,
      "learning_rate": 3.231721524791798e-07,
      "loss": 0.4382,
      "step": 410090
    },
    {
      "epoch": 167.87,
      "grad_norm": 1.9095313549041748,
      "learning_rate": 3.2309177702356605e-07,
      "loss": 0.4492,
      "step": 410100
    },
    {
      "epoch": 167.87,
      "grad_norm": 2.0472123622894287,
      "learning_rate": 3.230114108242745e-07,
      "loss": 0.4515,
      "step": 410110
    },
    {
      "epoch": 167.88,
      "grad_norm": 2.047168016433716,
      "learning_rate": 3.2293105388167374e-07,
      "loss": 0.4546,
      "step": 410120
    },
    {
      "epoch": 167.88,
      "grad_norm": 2.017179489135742,
      "learning_rate": 3.228507061961319e-07,
      "loss": 0.4438,
      "step": 410130
    },
    {
      "epoch": 167.88,
      "grad_norm": 2.0994110107421875,
      "learning_rate": 3.2277036776801635e-07,
      "loss": 0.4415,
      "step": 410140
    },
    {
      "epoch": 167.89,
      "grad_norm": 2.2218124866485596,
      "learning_rate": 3.2269003859769564e-07,
      "loss": 0.4335,
      "step": 410150
    },
    {
      "epoch": 167.89,
      "grad_norm": 2.247445821762085,
      "learning_rate": 3.2260971868553737e-07,
      "loss": 0.4356,
      "step": 410160
    },
    {
      "epoch": 167.9,
      "grad_norm": 3.3356082439422607,
      "learning_rate": 3.2252940803190973e-07,
      "loss": 0.4371,
      "step": 410170
    },
    {
      "epoch": 167.9,
      "grad_norm": 2.1168558597564697,
      "learning_rate": 3.2244910663718076e-07,
      "loss": 0.4436,
      "step": 410180
    },
    {
      "epoch": 167.9,
      "grad_norm": 2.4165008068084717,
      "learning_rate": 3.2236881450171824e-07,
      "loss": 0.4478,
      "step": 410190
    },
    {
      "epoch": 167.91,
      "grad_norm": 1.930884599685669,
      "learning_rate": 3.222885316258897e-07,
      "loss": 0.4473,
      "step": 410200
    },
    {
      "epoch": 167.91,
      "grad_norm": 1.7686381340026855,
      "learning_rate": 3.222082580100631e-07,
      "loss": 0.4528,
      "step": 410210
    },
    {
      "epoch": 167.92,
      "grad_norm": 3.275197982788086,
      "learning_rate": 3.2212799365460597e-07,
      "loss": 0.4384,
      "step": 410220
    },
    {
      "epoch": 167.92,
      "grad_norm": 2.0765366554260254,
      "learning_rate": 3.2204773855988635e-07,
      "loss": 0.4294,
      "step": 410230
    },
    {
      "epoch": 167.92,
      "grad_norm": 2.396515130996704,
      "learning_rate": 3.219674927262714e-07,
      "loss": 0.4585,
      "step": 410240
    },
    {
      "epoch": 167.93,
      "grad_norm": 2.0395290851593018,
      "learning_rate": 3.2188725615412927e-07,
      "loss": 0.4603,
      "step": 410250
    },
    {
      "epoch": 167.93,
      "grad_norm": 2.370154619216919,
      "learning_rate": 3.218070288438271e-07,
      "loss": 0.4599,
      "step": 410260
    },
    {
      "epoch": 167.94,
      "grad_norm": 2.1861984729766846,
      "learning_rate": 3.2172681079573317e-07,
      "loss": 0.4419,
      "step": 410270
    },
    {
      "epoch": 167.94,
      "grad_norm": 2.04835844039917,
      "learning_rate": 3.2164660201021403e-07,
      "loss": 0.4585,
      "step": 410280
    },
    {
      "epoch": 167.95,
      "grad_norm": 2.2449865341186523,
      "learning_rate": 3.2156640248763714e-07,
      "loss": 0.4375,
      "step": 410290
    },
    {
      "epoch": 167.95,
      "grad_norm": 1.9773776531219482,
      "learning_rate": 3.2148621222837017e-07,
      "loss": 0.4547,
      "step": 410300
    },
    {
      "epoch": 167.95,
      "grad_norm": 2.13763689994812,
      "learning_rate": 3.214060312327805e-07,
      "loss": 0.4519,
      "step": 410310
    },
    {
      "epoch": 167.96,
      "grad_norm": 1.8476935625076294,
      "learning_rate": 3.2132585950123534e-07,
      "loss": 0.432,
      "step": 410320
    },
    {
      "epoch": 167.96,
      "grad_norm": 1.9971927404403687,
      "learning_rate": 3.212456970341018e-07,
      "loss": 0.4401,
      "step": 410330
    },
    {
      "epoch": 167.97,
      "grad_norm": 1.929893970489502,
      "learning_rate": 3.211655438317476e-07,
      "loss": 0.4472,
      "step": 410340
    },
    {
      "epoch": 167.97,
      "grad_norm": 2.101287841796875,
      "learning_rate": 3.210853998945393e-07,
      "loss": 0.4697,
      "step": 410350
    },
    {
      "epoch": 167.97,
      "grad_norm": 3.2800631523132324,
      "learning_rate": 3.210052652228443e-07,
      "loss": 0.4627,
      "step": 410360
    },
    {
      "epoch": 167.98,
      "grad_norm": 1.9041002988815308,
      "learning_rate": 3.2092513981702966e-07,
      "loss": 0.4767,
      "step": 410370
    },
    {
      "epoch": 167.98,
      "grad_norm": 2.2085177898406982,
      "learning_rate": 3.2084502367746237e-07,
      "loss": 0.4273,
      "step": 410380
    },
    {
      "epoch": 167.99,
      "grad_norm": 2.52055025100708,
      "learning_rate": 3.207649168045097e-07,
      "loss": 0.4722,
      "step": 410390
    },
    {
      "epoch": 167.99,
      "grad_norm": 2.16886830329895,
      "learning_rate": 3.206848191985381e-07,
      "loss": 0.459,
      "step": 410400
    },
    {
      "epoch": 167.99,
      "grad_norm": 2.0624401569366455,
      "learning_rate": 3.206047308599152e-07,
      "loss": 0.4428,
      "step": 410410
    },
    {
      "epoch": 168.0,
      "grad_norm": 1.9288008213043213,
      "learning_rate": 3.20524651789007e-07,
      "loss": 0.4291,
      "step": 410420
    },
    {
      "epoch": 168.0,
      "eval_loss": 0.44687068462371826,
      "eval_runtime": 52.309,
      "eval_samples_per_second": 65.935,
      "eval_steps_per_second": 8.259,
      "step": 410424
    },
    {
      "epoch": 168.0,
      "grad_norm": 2.0280511379241943,
      "learning_rate": 3.204445819861812e-07,
      "loss": 0.4503,
      "step": 410430
    },
    {
      "epoch": 168.01,
      "grad_norm": 2.0057618618011475,
      "learning_rate": 3.203645214518039e-07,
      "loss": 0.4557,
      "step": 410440
    },
    {
      "epoch": 168.01,
      "grad_norm": 2.115600109100342,
      "learning_rate": 3.202844701862423e-07,
      "loss": 0.4444,
      "step": 410450
    },
    {
      "epoch": 168.01,
      "grad_norm": 1.831809401512146,
      "learning_rate": 3.202044281898627e-07,
      "loss": 0.4317,
      "step": 410460
    },
    {
      "epoch": 168.02,
      "grad_norm": 1.7793723344802856,
      "learning_rate": 3.2012439546303203e-07,
      "loss": 0.4355,
      "step": 410470
    },
    {
      "epoch": 168.02,
      "grad_norm": 1.7878990173339844,
      "learning_rate": 3.2004437200611693e-07,
      "loss": 0.4379,
      "step": 410480
    },
    {
      "epoch": 168.03,
      "grad_norm": 1.902535319328308,
      "learning_rate": 3.19964357819484e-07,
      "loss": 0.429,
      "step": 410490
    },
    {
      "epoch": 168.03,
      "grad_norm": 2.040428876876831,
      "learning_rate": 3.1988435290349943e-07,
      "loss": 0.4535,
      "step": 410500
    },
    {
      "epoch": 168.04,
      "grad_norm": 2.204397678375244,
      "learning_rate": 3.1980435725853004e-07,
      "loss": 0.4498,
      "step": 410510
    },
    {
      "epoch": 168.04,
      "grad_norm": 1.8121566772460938,
      "learning_rate": 3.197243708849422e-07,
      "loss": 0.4511,
      "step": 410520
    },
    {
      "epoch": 168.04,
      "grad_norm": 2.1592607498168945,
      "learning_rate": 3.1964439378310236e-07,
      "loss": 0.4321,
      "step": 410530
    },
    {
      "epoch": 168.05,
      "grad_norm": 2.0911288261413574,
      "learning_rate": 3.195644259533768e-07,
      "loss": 0.438,
      "step": 410540
    },
    {
      "epoch": 168.05,
      "grad_norm": 2.0626561641693115,
      "learning_rate": 3.19484467396132e-07,
      "loss": 0.4494,
      "step": 410550
    },
    {
      "epoch": 168.06,
      "grad_norm": 1.8931587934494019,
      "learning_rate": 3.19404518111734e-07,
      "loss": 0.4414,
      "step": 410560
    },
    {
      "epoch": 168.06,
      "grad_norm": 1.9898972511291504,
      "learning_rate": 3.193245781005492e-07,
      "loss": 0.4408,
      "step": 410570
    },
    {
      "epoch": 168.06,
      "grad_norm": 3.6655218601226807,
      "learning_rate": 3.1924464736294416e-07,
      "loss": 0.4638,
      "step": 410580
    },
    {
      "epoch": 168.07,
      "grad_norm": 2.334563970565796,
      "learning_rate": 3.191647258992841e-07,
      "loss": 0.4313,
      "step": 410590
    },
    {
      "epoch": 168.07,
      "grad_norm": 2.0577399730682373,
      "learning_rate": 3.190848137099357e-07,
      "loss": 0.4591,
      "step": 410600
    },
    {
      "epoch": 168.08,
      "grad_norm": 1.994919776916504,
      "learning_rate": 3.1900491079526426e-07,
      "loss": 0.4397,
      "step": 410610
    },
    {
      "epoch": 168.08,
      "grad_norm": 2.0676560401916504,
      "learning_rate": 3.189250171556373e-07,
      "loss": 0.4441,
      "step": 410620
    },
    {
      "epoch": 168.08,
      "grad_norm": 2.2697930335998535,
      "learning_rate": 3.188451327914197e-07,
      "loss": 0.461,
      "step": 410630
    },
    {
      "epoch": 168.09,
      "grad_norm": 2.0611929893493652,
      "learning_rate": 3.187652577029779e-07,
      "loss": 0.457,
      "step": 410640
    },
    {
      "epoch": 168.09,
      "grad_norm": 2.542459487915039,
      "learning_rate": 3.186853918906776e-07,
      "loss": 0.4434,
      "step": 410650
    },
    {
      "epoch": 168.1,
      "grad_norm": 2.0892844200134277,
      "learning_rate": 3.186055353548847e-07,
      "loss": 0.4513,
      "step": 410660
    },
    {
      "epoch": 168.1,
      "grad_norm": 1.726556420326233,
      "learning_rate": 3.1852568809596487e-07,
      "loss": 0.4408,
      "step": 410670
    },
    {
      "epoch": 168.1,
      "grad_norm": 1.9905015230178833,
      "learning_rate": 3.184458501142839e-07,
      "loss": 0.4583,
      "step": 410680
    },
    {
      "epoch": 168.11,
      "grad_norm": 2.0736138820648193,
      "learning_rate": 3.1836602141020766e-07,
      "loss": 0.454,
      "step": 410690
    },
    {
      "epoch": 168.11,
      "grad_norm": 2.234376907348633,
      "learning_rate": 3.1828620198410173e-07,
      "loss": 0.4476,
      "step": 410700
    },
    {
      "epoch": 168.12,
      "grad_norm": 2.0469441413879395,
      "learning_rate": 3.1820639183633225e-07,
      "loss": 0.4164,
      "step": 410710
    },
    {
      "epoch": 168.12,
      "grad_norm": 1.7136242389678955,
      "learning_rate": 3.18126590967264e-07,
      "loss": 0.4374,
      "step": 410720
    },
    {
      "epoch": 168.13,
      "grad_norm": 2.5723023414611816,
      "learning_rate": 3.180467993772626e-07,
      "loss": 0.4376,
      "step": 410730
    },
    {
      "epoch": 168.13,
      "grad_norm": 2.2952308654785156,
      "learning_rate": 3.179670170666941e-07,
      "loss": 0.4616,
      "step": 410740
    },
    {
      "epoch": 168.13,
      "grad_norm": 1.8932850360870361,
      "learning_rate": 3.178872440359237e-07,
      "loss": 0.4443,
      "step": 410750
    },
    {
      "epoch": 168.14,
      "grad_norm": 1.7069882154464722,
      "learning_rate": 3.178074802853166e-07,
      "loss": 0.4699,
      "step": 410760
    },
    {
      "epoch": 168.14,
      "grad_norm": 2.129002809524536,
      "learning_rate": 3.177277258152385e-07,
      "loss": 0.4467,
      "step": 410770
    },
    {
      "epoch": 168.15,
      "grad_norm": 1.7738380432128906,
      "learning_rate": 3.176479806260549e-07,
      "loss": 0.4593,
      "step": 410780
    },
    {
      "epoch": 168.15,
      "grad_norm": 1.8513015508651733,
      "learning_rate": 3.1756824471813067e-07,
      "loss": 0.4456,
      "step": 410790
    },
    {
      "epoch": 168.15,
      "grad_norm": 2.1013412475585938,
      "learning_rate": 3.1748851809183086e-07,
      "loss": 0.4656,
      "step": 410800
    },
    {
      "epoch": 168.16,
      "grad_norm": 2.1731927394866943,
      "learning_rate": 3.174088007475216e-07,
      "loss": 0.4511,
      "step": 410810
    },
    {
      "epoch": 168.16,
      "grad_norm": 2.110583543777466,
      "learning_rate": 3.173290926855674e-07,
      "loss": 0.4185,
      "step": 410820
    },
    {
      "epoch": 168.17,
      "grad_norm": 2.0512335300445557,
      "learning_rate": 3.1724939390633365e-07,
      "loss": 0.4507,
      "step": 410830
    },
    {
      "epoch": 168.17,
      "grad_norm": 1.9137191772460938,
      "learning_rate": 3.171697044101854e-07,
      "loss": 0.47,
      "step": 410840
    },
    {
      "epoch": 168.17,
      "grad_norm": 1.8455290794372559,
      "learning_rate": 3.17090024197488e-07,
      "loss": 0.4383,
      "step": 410850
    },
    {
      "epoch": 168.18,
      "grad_norm": 2.1429381370544434,
      "learning_rate": 3.170103532686057e-07,
      "loss": 0.4263,
      "step": 410860
    },
    {
      "epoch": 168.18,
      "grad_norm": 2.0551400184631348,
      "learning_rate": 3.169306916239036e-07,
      "loss": 0.4389,
      "step": 410870
    },
    {
      "epoch": 168.19,
      "grad_norm": 2.356844663619995,
      "learning_rate": 3.1685103926374705e-07,
      "loss": 0.4321,
      "step": 410880
    },
    {
      "epoch": 168.19,
      "grad_norm": 2.12557315826416,
      "learning_rate": 3.167713961885008e-07,
      "loss": 0.4344,
      "step": 410890
    },
    {
      "epoch": 168.19,
      "grad_norm": 2.686432361602783,
      "learning_rate": 3.166917623985295e-07,
      "loss": 0.4459,
      "step": 410900
    },
    {
      "epoch": 168.2,
      "grad_norm": 1.8764398097991943,
      "learning_rate": 3.1661213789419807e-07,
      "loss": 0.4664,
      "step": 410910
    },
    {
      "epoch": 168.2,
      "grad_norm": 2.183104991912842,
      "learning_rate": 3.1653252267587123e-07,
      "loss": 0.434,
      "step": 410920
    },
    {
      "epoch": 168.21,
      "grad_norm": 1.9747759103775024,
      "learning_rate": 3.16452916743914e-07,
      "loss": 0.4449,
      "step": 410930
    },
    {
      "epoch": 168.21,
      "grad_norm": 1.7727924585342407,
      "learning_rate": 3.1637332009869034e-07,
      "loss": 0.457,
      "step": 410940
    },
    {
      "epoch": 168.22,
      "grad_norm": 1.772203803062439,
      "learning_rate": 3.1629373274056537e-07,
      "loss": 0.4345,
      "step": 410950
    },
    {
      "epoch": 168.22,
      "grad_norm": 2.015423059463501,
      "learning_rate": 3.1621415466990367e-07,
      "loss": 0.4434,
      "step": 410960
    },
    {
      "epoch": 168.22,
      "grad_norm": 2.4217846393585205,
      "learning_rate": 3.161345858870695e-07,
      "loss": 0.4276,
      "step": 410970
    },
    {
      "epoch": 168.23,
      "grad_norm": 1.6818147897720337,
      "learning_rate": 3.160550263924277e-07,
      "loss": 0.436,
      "step": 410980
    },
    {
      "epoch": 168.23,
      "grad_norm": 1.983422040939331,
      "learning_rate": 3.1597547618634254e-07,
      "loss": 0.4527,
      "step": 410990
    },
    {
      "epoch": 168.24,
      "grad_norm": 2.560892343521118,
      "learning_rate": 3.1589593526917854e-07,
      "loss": 0.4616,
      "step": 411000
    },
    {
      "epoch": 168.24,
      "grad_norm": 2.0330710411071777,
      "learning_rate": 3.158164036412998e-07,
      "loss": 0.449,
      "step": 411010
    },
    {
      "epoch": 168.24,
      "grad_norm": 2.276918411254883,
      "learning_rate": 3.1573688130307057e-07,
      "loss": 0.4564,
      "step": 411020
    },
    {
      "epoch": 168.25,
      "grad_norm": 1.7080217599868774,
      "learning_rate": 3.156573682548557e-07,
      "loss": 0.4515,
      "step": 411030
    },
    {
      "epoch": 168.25,
      "grad_norm": 2.1061019897460938,
      "learning_rate": 3.1557786449701914e-07,
      "loss": 0.4165,
      "step": 411040
    },
    {
      "epoch": 168.26,
      "grad_norm": 2.130164384841919,
      "learning_rate": 3.154983700299247e-07,
      "loss": 0.4341,
      "step": 411050
    },
    {
      "epoch": 168.26,
      "grad_norm": 2.164193868637085,
      "learning_rate": 3.15418884853937e-07,
      "loss": 0.4493,
      "step": 411060
    },
    {
      "epoch": 168.26,
      "grad_norm": 2.2439422607421875,
      "learning_rate": 3.1533940896942e-07,
      "loss": 0.4585,
      "step": 411070
    },
    {
      "epoch": 168.27,
      "grad_norm": 1.9575051069259644,
      "learning_rate": 3.15259942376738e-07,
      "loss": 0.4453,
      "step": 411080
    },
    {
      "epoch": 168.27,
      "grad_norm": 1.8538110256195068,
      "learning_rate": 3.151804850762545e-07,
      "loss": 0.4483,
      "step": 411090
    },
    {
      "epoch": 168.28,
      "grad_norm": 2.1946463584899902,
      "learning_rate": 3.1510103706833387e-07,
      "loss": 0.4512,
      "step": 411100
    },
    {
      "epoch": 168.28,
      "grad_norm": 2.2066261768341064,
      "learning_rate": 3.1502159835334006e-07,
      "loss": 0.4528,
      "step": 411110
    },
    {
      "epoch": 168.28,
      "grad_norm": 2.0746867656707764,
      "learning_rate": 3.149421689316369e-07,
      "loss": 0.4564,
      "step": 411120
    },
    {
      "epoch": 168.29,
      "grad_norm": 1.8319261074066162,
      "learning_rate": 3.1486274880358806e-07,
      "loss": 0.448,
      "step": 411130
    },
    {
      "epoch": 168.29,
      "grad_norm": 2.297344446182251,
      "learning_rate": 3.147833379695577e-07,
      "loss": 0.4585,
      "step": 411140
    },
    {
      "epoch": 168.3,
      "grad_norm": 2.1774208545684814,
      "learning_rate": 3.147039364299092e-07,
      "loss": 0.4361,
      "step": 411150
    },
    {
      "epoch": 168.3,
      "grad_norm": 2.0210061073303223,
      "learning_rate": 3.14624544185007e-07,
      "loss": 0.4539,
      "step": 411160
    },
    {
      "epoch": 168.31,
      "grad_norm": 2.0497539043426514,
      "learning_rate": 3.1454516123521395e-07,
      "loss": 0.4565,
      "step": 411170
    },
    {
      "epoch": 168.31,
      "grad_norm": 2.4041085243225098,
      "learning_rate": 3.1446578758089313e-07,
      "loss": 0.4476,
      "step": 411180
    },
    {
      "epoch": 168.31,
      "grad_norm": 2.2156429290771484,
      "learning_rate": 3.143864232224099e-07,
      "loss": 0.4567,
      "step": 411190
    },
    {
      "epoch": 168.32,
      "grad_norm": 1.778937816619873,
      "learning_rate": 3.143070681601269e-07,
      "loss": 0.4467,
      "step": 411200
    },
    {
      "epoch": 168.32,
      "grad_norm": 1.8679391145706177,
      "learning_rate": 3.1422772239440736e-07,
      "loss": 0.4285,
      "step": 411210
    },
    {
      "epoch": 168.33,
      "grad_norm": 2.0303332805633545,
      "learning_rate": 3.1414838592561517e-07,
      "loss": 0.4534,
      "step": 411220
    },
    {
      "epoch": 168.33,
      "grad_norm": 3.1400034427642822,
      "learning_rate": 3.140690587541138e-07,
      "loss": 0.4317,
      "step": 411230
    },
    {
      "epoch": 168.33,
      "grad_norm": 2.125126361846924,
      "learning_rate": 3.139897408802664e-07,
      "loss": 0.4465,
      "step": 411240
    },
    {
      "epoch": 168.34,
      "grad_norm": 1.7201331853866577,
      "learning_rate": 3.1391043230443636e-07,
      "loss": 0.4513,
      "step": 411250
    },
    {
      "epoch": 168.34,
      "grad_norm": 1.8666337728500366,
      "learning_rate": 3.1383113302698714e-07,
      "loss": 0.4558,
      "step": 411260
    },
    {
      "epoch": 168.35,
      "grad_norm": 2.0476760864257812,
      "learning_rate": 3.1375184304828197e-07,
      "loss": 0.4661,
      "step": 411270
    },
    {
      "epoch": 168.35,
      "grad_norm": 2.594268321990967,
      "learning_rate": 3.1367256236868364e-07,
      "loss": 0.457,
      "step": 411280
    },
    {
      "epoch": 168.35,
      "grad_norm": 2.441720485687256,
      "learning_rate": 3.1359329098855634e-07,
      "loss": 0.4537,
      "step": 411290
    },
    {
      "epoch": 168.36,
      "grad_norm": 2.0256783962249756,
      "learning_rate": 3.135140289082621e-07,
      "loss": 0.4389,
      "step": 411300
    },
    {
      "epoch": 168.36,
      "grad_norm": 1.8874852657318115,
      "learning_rate": 3.1343477612816443e-07,
      "loss": 0.4507,
      "step": 411310
    },
    {
      "epoch": 168.37,
      "grad_norm": 1.9320348501205444,
      "learning_rate": 3.13355532648626e-07,
      "loss": 0.432,
      "step": 411320
    },
    {
      "epoch": 168.37,
      "grad_norm": 2.1331710815429688,
      "learning_rate": 3.1327629847001067e-07,
      "loss": 0.4226,
      "step": 411330
    },
    {
      "epoch": 168.37,
      "grad_norm": 2.1078479290008545,
      "learning_rate": 3.131970735926806e-07,
      "loss": 0.4422,
      "step": 411340
    },
    {
      "epoch": 168.38,
      "grad_norm": 2.4794487953186035,
      "learning_rate": 3.13117858016999e-07,
      "loss": 0.4296,
      "step": 411350
    },
    {
      "epoch": 168.38,
      "grad_norm": 1.9461430311203003,
      "learning_rate": 3.1303865174332894e-07,
      "loss": 0.4464,
      "step": 411360
    },
    {
      "epoch": 168.39,
      "grad_norm": 1.8842761516571045,
      "learning_rate": 3.1295945477203257e-07,
      "loss": 0.4392,
      "step": 411370
    },
    {
      "epoch": 168.39,
      "grad_norm": 2.356858968734741,
      "learning_rate": 3.1288026710347343e-07,
      "loss": 0.4412,
      "step": 411380
    },
    {
      "epoch": 168.4,
      "grad_norm": 1.9222990274429321,
      "learning_rate": 3.1280108873801427e-07,
      "loss": 0.4601,
      "step": 411390
    },
    {
      "epoch": 168.4,
      "grad_norm": 2.144801378250122,
      "learning_rate": 3.127219196760175e-07,
      "loss": 0.4489,
      "step": 411400
    },
    {
      "epoch": 168.4,
      "grad_norm": 1.87820565700531,
      "learning_rate": 3.1264275991784567e-07,
      "loss": 0.4175,
      "step": 411410
    },
    {
      "epoch": 168.41,
      "grad_norm": 2.001357316970825,
      "learning_rate": 3.125636094638617e-07,
      "loss": 0.448,
      "step": 411420
    },
    {
      "epoch": 168.41,
      "grad_norm": 1.9815154075622559,
      "learning_rate": 3.124844683144284e-07,
      "loss": 0.4809,
      "step": 411430
    },
    {
      "epoch": 168.42,
      "grad_norm": 1.866392970085144,
      "learning_rate": 3.124053364699076e-07,
      "loss": 0.4175,
      "step": 411440
    },
    {
      "epoch": 168.42,
      "grad_norm": 2.4160115718841553,
      "learning_rate": 3.123262139306619e-07,
      "loss": 0.4423,
      "step": 411450
    },
    {
      "epoch": 168.42,
      "grad_norm": 2.184755563735962,
      "learning_rate": 3.122471006970542e-07,
      "loss": 0.4423,
      "step": 411460
    },
    {
      "epoch": 168.43,
      "grad_norm": 2.6137423515319824,
      "learning_rate": 3.121679967694465e-07,
      "loss": 0.442,
      "step": 411470
    },
    {
      "epoch": 168.43,
      "grad_norm": 1.8800249099731445,
      "learning_rate": 3.120889021482012e-07,
      "loss": 0.4446,
      "step": 411480
    },
    {
      "epoch": 168.44,
      "grad_norm": 1.956684947013855,
      "learning_rate": 3.1200981683368113e-07,
      "loss": 0.4371,
      "step": 411490
    },
    {
      "epoch": 168.44,
      "grad_norm": 2.042058229446411,
      "learning_rate": 3.119307408262479e-07,
      "loss": 0.4345,
      "step": 411500
    },
    {
      "epoch": 168.44,
      "grad_norm": 3.1329050064086914,
      "learning_rate": 3.1185167412626404e-07,
      "loss": 0.4521,
      "step": 411510
    },
    {
      "epoch": 168.45,
      "grad_norm": 2.097623586654663,
      "learning_rate": 3.1177261673409175e-07,
      "loss": 0.46,
      "step": 411520
    },
    {
      "epoch": 168.45,
      "grad_norm": 2.114715576171875,
      "learning_rate": 3.1169356865009316e-07,
      "loss": 0.4589,
      "step": 411530
    },
    {
      "epoch": 168.46,
      "grad_norm": 1.914770245552063,
      "learning_rate": 3.116145298746303e-07,
      "loss": 0.4394,
      "step": 411540
    },
    {
      "epoch": 168.46,
      "grad_norm": 2.0784449577331543,
      "learning_rate": 3.115355004080654e-07,
      "loss": 0.4683,
      "step": 411550
    },
    {
      "epoch": 168.47,
      "grad_norm": 2.0852952003479004,
      "learning_rate": 3.1145648025076013e-07,
      "loss": 0.4427,
      "step": 411560
    },
    {
      "epoch": 168.47,
      "grad_norm": 2.3393115997314453,
      "learning_rate": 3.113774694030774e-07,
      "loss": 0.4523,
      "step": 411570
    },
    {
      "epoch": 168.47,
      "grad_norm": 2.076120615005493,
      "learning_rate": 3.1129846786537824e-07,
      "loss": 0.4471,
      "step": 411580
    },
    {
      "epoch": 168.48,
      "grad_norm": 2.3946146965026855,
      "learning_rate": 3.112194756380245e-07,
      "loss": 0.4385,
      "step": 411590
    },
    {
      "epoch": 168.48,
      "grad_norm": 1.990500807762146,
      "learning_rate": 3.1114049272137847e-07,
      "loss": 0.4417,
      "step": 411600
    },
    {
      "epoch": 168.49,
      "grad_norm": 1.5918418169021606,
      "learning_rate": 3.110615191158018e-07,
      "loss": 0.4437,
      "step": 411610
    },
    {
      "epoch": 168.49,
      "grad_norm": 1.9318562746047974,
      "learning_rate": 3.1098255482165615e-07,
      "loss": 0.4455,
      "step": 411620
    },
    {
      "epoch": 168.49,
      "grad_norm": 2.190373182296753,
      "learning_rate": 3.109035998393035e-07,
      "loss": 0.4388,
      "step": 411630
    },
    {
      "epoch": 168.5,
      "grad_norm": 2.373230218887329,
      "learning_rate": 3.1082465416910533e-07,
      "loss": 0.4592,
      "step": 411640
    },
    {
      "epoch": 168.5,
      "grad_norm": 1.8662526607513428,
      "learning_rate": 3.107457178114235e-07,
      "loss": 0.4422,
      "step": 411650
    },
    {
      "epoch": 168.51,
      "grad_norm": 2.3571901321411133,
      "learning_rate": 3.1066679076661926e-07,
      "loss": 0.4598,
      "step": 411660
    },
    {
      "epoch": 168.51,
      "grad_norm": 2.1021342277526855,
      "learning_rate": 3.1058787303505424e-07,
      "loss": 0.433,
      "step": 411670
    },
    {
      "epoch": 168.51,
      "grad_norm": 2.3682432174682617,
      "learning_rate": 3.105089646170901e-07,
      "loss": 0.4288,
      "step": 411680
    },
    {
      "epoch": 168.52,
      "grad_norm": 1.7984154224395752,
      "learning_rate": 3.104300655130884e-07,
      "loss": 0.4615,
      "step": 411690
    },
    {
      "epoch": 168.52,
      "grad_norm": 2.2904155254364014,
      "learning_rate": 3.1035117572341023e-07,
      "loss": 0.4501,
      "step": 411700
    },
    {
      "epoch": 168.53,
      "grad_norm": 2.1441335678100586,
      "learning_rate": 3.1027229524841724e-07,
      "loss": 0.4538,
      "step": 411710
    },
    {
      "epoch": 168.53,
      "grad_norm": 2.416379451751709,
      "learning_rate": 3.1019342408847093e-07,
      "loss": 0.4621,
      "step": 411720
    },
    {
      "epoch": 168.53,
      "grad_norm": 2.1625804901123047,
      "learning_rate": 3.1011456224393214e-07,
      "loss": 0.4513,
      "step": 411730
    },
    {
      "epoch": 168.54,
      "grad_norm": 1.9215095043182373,
      "learning_rate": 3.1003570971516266e-07,
      "loss": 0.4487,
      "step": 411740
    },
    {
      "epoch": 168.54,
      "grad_norm": 2.1641440391540527,
      "learning_rate": 3.099568665025228e-07,
      "loss": 0.4472,
      "step": 411750
    },
    {
      "epoch": 168.55,
      "grad_norm": 2.1797428131103516,
      "learning_rate": 3.098780326063745e-07,
      "loss": 0.4612,
      "step": 411760
    },
    {
      "epoch": 168.55,
      "grad_norm": 1.6724021434783936,
      "learning_rate": 3.097992080270787e-07,
      "loss": 0.451,
      "step": 411770
    },
    {
      "epoch": 168.56,
      "grad_norm": 1.9629831314086914,
      "learning_rate": 3.097203927649966e-07,
      "loss": 0.456,
      "step": 411780
    },
    {
      "epoch": 168.56,
      "grad_norm": 2.701148271560669,
      "learning_rate": 3.0964158682048905e-07,
      "loss": 0.4341,
      "step": 411790
    },
    {
      "epoch": 168.56,
      "grad_norm": 1.9893213510513306,
      "learning_rate": 3.0956279019391727e-07,
      "loss": 0.4372,
      "step": 411800
    },
    {
      "epoch": 168.57,
      "grad_norm": 2.033695697784424,
      "learning_rate": 3.094840028856418e-07,
      "loss": 0.4243,
      "step": 411810
    },
    {
      "epoch": 168.57,
      "grad_norm": 2.1754536628723145,
      "learning_rate": 3.09405224896024e-07,
      "loss": 0.4402,
      "step": 411820
    },
    {
      "epoch": 168.58,
      "grad_norm": 1.8017150163650513,
      "learning_rate": 3.093264562254246e-07,
      "loss": 0.4334,
      "step": 411830
    },
    {
      "epoch": 168.58,
      "grad_norm": 2.0586495399475098,
      "learning_rate": 3.092476968742041e-07,
      "loss": 0.445,
      "step": 411840
    },
    {
      "epoch": 168.58,
      "grad_norm": 1.9725619554519653,
      "learning_rate": 3.0916894684272393e-07,
      "loss": 0.4423,
      "step": 411850
    },
    {
      "epoch": 168.59,
      "grad_norm": 1.700785756111145,
      "learning_rate": 3.090902061313444e-07,
      "loss": 0.4202,
      "step": 411860
    },
    {
      "epoch": 168.59,
      "grad_norm": 2.1101555824279785,
      "learning_rate": 3.090114747404262e-07,
      "loss": 0.4406,
      "step": 411870
    },
    {
      "epoch": 168.6,
      "grad_norm": 2.0667994022369385,
      "learning_rate": 3.089327526703305e-07,
      "loss": 0.4421,
      "step": 411880
    },
    {
      "epoch": 168.6,
      "grad_norm": 2.043804168701172,
      "learning_rate": 3.0885403992141716e-07,
      "loss": 0.4429,
      "step": 411890
    },
    {
      "epoch": 168.6,
      "grad_norm": 2.3872807025909424,
      "learning_rate": 3.0877533649404685e-07,
      "loss": 0.4626,
      "step": 411900
    },
    {
      "epoch": 168.61,
      "grad_norm": 2.584613084793091,
      "learning_rate": 3.086966423885805e-07,
      "loss": 0.4507,
      "step": 411910
    },
    {
      "epoch": 168.61,
      "grad_norm": 1.9370840787887573,
      "learning_rate": 3.086179576053781e-07,
      "loss": 0.4504,
      "step": 411920
    },
    {
      "epoch": 168.62,
      "grad_norm": 2.2004313468933105,
      "learning_rate": 3.0853928214480025e-07,
      "loss": 0.4484,
      "step": 411930
    },
    {
      "epoch": 168.62,
      "grad_norm": 1.8883049488067627,
      "learning_rate": 3.0846061600720794e-07,
      "loss": 0.4496,
      "step": 411940
    },
    {
      "epoch": 168.62,
      "grad_norm": 2.42271089553833,
      "learning_rate": 3.083819591929609e-07,
      "loss": 0.4413,
      "step": 411950
    },
    {
      "epoch": 168.63,
      "grad_norm": 1.9711960554122925,
      "learning_rate": 3.0830331170242e-07,
      "loss": 0.4433,
      "step": 411960
    },
    {
      "epoch": 168.63,
      "grad_norm": 1.727565884590149,
      "learning_rate": 3.082246735359448e-07,
      "loss": 0.45,
      "step": 411970
    },
    {
      "epoch": 168.64,
      "grad_norm": 1.8331984281539917,
      "learning_rate": 3.0814604469389597e-07,
      "loss": 0.4588,
      "step": 411980
    },
    {
      "epoch": 168.64,
      "grad_norm": 2.8217740058898926,
      "learning_rate": 3.080674251766337e-07,
      "loss": 0.4454,
      "step": 411990
    },
    {
      "epoch": 168.65,
      "grad_norm": 1.8467731475830078,
      "learning_rate": 3.079888149845181e-07,
      "loss": 0.4836,
      "step": 412000
    },
    {
      "epoch": 168.65,
      "grad_norm": 2.0877127647399902,
      "learning_rate": 3.0791021411790947e-07,
      "loss": 0.4694,
      "step": 412010
    },
    {
      "epoch": 168.65,
      "grad_norm": 2.176222324371338,
      "learning_rate": 3.078316225771672e-07,
      "loss": 0.4541,
      "step": 412020
    },
    {
      "epoch": 168.66,
      "grad_norm": 2.0478527545928955,
      "learning_rate": 3.0775304036265176e-07,
      "loss": 0.4453,
      "step": 412030
    },
    {
      "epoch": 168.66,
      "grad_norm": 2.001539945602417,
      "learning_rate": 3.076744674747232e-07,
      "loss": 0.4281,
      "step": 412040
    },
    {
      "epoch": 168.67,
      "grad_norm": 2.0078539848327637,
      "learning_rate": 3.075959039137411e-07,
      "loss": 0.436,
      "step": 412050
    },
    {
      "epoch": 168.67,
      "grad_norm": 1.64707612991333,
      "learning_rate": 3.075173496800657e-07,
      "loss": 0.4516,
      "step": 412060
    },
    {
      "epoch": 168.67,
      "grad_norm": 1.9923838376998901,
      "learning_rate": 3.0743880477405683e-07,
      "loss": 0.4466,
      "step": 412070
    },
    {
      "epoch": 168.68,
      "grad_norm": 2.0215113162994385,
      "learning_rate": 3.0736026919607404e-07,
      "loss": 0.4683,
      "step": 412080
    },
    {
      "epoch": 168.68,
      "grad_norm": 1.8244255781173706,
      "learning_rate": 3.072817429464772e-07,
      "loss": 0.423,
      "step": 412090
    },
    {
      "epoch": 168.69,
      "grad_norm": 1.9434082508087158,
      "learning_rate": 3.0720322602562624e-07,
      "loss": 0.4436,
      "step": 412100
    },
    {
      "epoch": 168.69,
      "grad_norm": 2.372495174407959,
      "learning_rate": 3.071247184338804e-07,
      "loss": 0.443,
      "step": 412110
    },
    {
      "epoch": 168.69,
      "grad_norm": 2.152322769165039,
      "learning_rate": 3.070462201715992e-07,
      "loss": 0.439,
      "step": 412120
    },
    {
      "epoch": 168.7,
      "grad_norm": 2.44880747795105,
      "learning_rate": 3.069677312391429e-07,
      "loss": 0.4338,
      "step": 412130
    },
    {
      "epoch": 168.7,
      "grad_norm": 1.920042872428894,
      "learning_rate": 3.0688925163687116e-07,
      "loss": 0.4456,
      "step": 412140
    },
    {
      "epoch": 168.71,
      "grad_norm": 2.603208065032959,
      "learning_rate": 3.068107813651433e-07,
      "loss": 0.4467,
      "step": 412150
    },
    {
      "epoch": 168.71,
      "grad_norm": 2.0869128704071045,
      "learning_rate": 3.0673232042431787e-07,
      "loss": 0.4471,
      "step": 412160
    },
    {
      "epoch": 168.71,
      "grad_norm": 1.8968496322631836,
      "learning_rate": 3.0665386881475534e-07,
      "loss": 0.4752,
      "step": 412170
    },
    {
      "epoch": 168.72,
      "grad_norm": 1.9245322942733765,
      "learning_rate": 3.0657542653681445e-07,
      "loss": 0.4436,
      "step": 412180
    },
    {
      "epoch": 168.72,
      "grad_norm": 2.1810646057128906,
      "learning_rate": 3.064969935908548e-07,
      "loss": 0.4407,
      "step": 412190
    },
    {
      "epoch": 168.73,
      "grad_norm": 2.090939521789551,
      "learning_rate": 3.064185699772356e-07,
      "loss": 0.432,
      "step": 412200
    },
    {
      "epoch": 168.73,
      "grad_norm": 1.9313267469406128,
      "learning_rate": 3.063401556963163e-07,
      "loss": 0.4325,
      "step": 412210
    },
    {
      "epoch": 168.74,
      "grad_norm": 1.9361748695373535,
      "learning_rate": 3.0626175074845614e-07,
      "loss": 0.4651,
      "step": 412220
    },
    {
      "epoch": 168.74,
      "grad_norm": 2.196185827255249,
      "learning_rate": 3.061833551340138e-07,
      "loss": 0.4426,
      "step": 412230
    },
    {
      "epoch": 168.74,
      "grad_norm": 2.052906036376953,
      "learning_rate": 3.0610496885334873e-07,
      "loss": 0.4454,
      "step": 412240
    },
    {
      "epoch": 168.75,
      "grad_norm": 1.8424164056777954,
      "learning_rate": 3.0602659190682e-07,
      "loss": 0.4381,
      "step": 412250
    },
    {
      "epoch": 168.75,
      "grad_norm": 1.9214617013931274,
      "learning_rate": 3.059482242947867e-07,
      "loss": 0.4508,
      "step": 412260
    },
    {
      "epoch": 168.76,
      "grad_norm": 2.0606091022491455,
      "learning_rate": 3.058698660176076e-07,
      "loss": 0.4815,
      "step": 412270
    },
    {
      "epoch": 168.76,
      "grad_norm": 2.303776979446411,
      "learning_rate": 3.0579151707564214e-07,
      "loss": 0.4439,
      "step": 412280
    },
    {
      "epoch": 168.76,
      "grad_norm": 2.091977596282959,
      "learning_rate": 3.0571317746924856e-07,
      "loss": 0.4404,
      "step": 412290
    },
    {
      "epoch": 168.77,
      "grad_norm": 1.8881011009216309,
      "learning_rate": 3.0563484719878595e-07,
      "loss": 0.4508,
      "step": 412300
    },
    {
      "epoch": 168.77,
      "grad_norm": 1.9892741441726685,
      "learning_rate": 3.0555652626461345e-07,
      "loss": 0.4304,
      "step": 412310
    },
    {
      "epoch": 168.78,
      "grad_norm": 2.267575740814209,
      "learning_rate": 3.0547821466708957e-07,
      "loss": 0.437,
      "step": 412320
    },
    {
      "epoch": 168.78,
      "grad_norm": 1.918679118156433,
      "learning_rate": 3.0539991240657287e-07,
      "loss": 0.4408,
      "step": 412330
    },
    {
      "epoch": 168.78,
      "grad_norm": 2.1815848350524902,
      "learning_rate": 3.053216194834225e-07,
      "loss": 0.4636,
      "step": 412340
    },
    {
      "epoch": 168.79,
      "grad_norm": 1.9012473821640015,
      "learning_rate": 3.052433358979967e-07,
      "loss": 0.4485,
      "step": 412350
    },
    {
      "epoch": 168.79,
      "grad_norm": 1.9038246870040894,
      "learning_rate": 3.051650616506544e-07,
      "loss": 0.4684,
      "step": 412360
    },
    {
      "epoch": 168.8,
      "grad_norm": 2.1317169666290283,
      "learning_rate": 3.050867967417538e-07,
      "loss": 0.4349,
      "step": 412370
    },
    {
      "epoch": 168.8,
      "grad_norm": 1.800918698310852,
      "learning_rate": 3.050085411716537e-07,
      "loss": 0.4507,
      "step": 412380
    },
    {
      "epoch": 168.8,
      "grad_norm": 2.262798547744751,
      "learning_rate": 3.0493029494071233e-07,
      "loss": 0.4348,
      "step": 412390
    },
    {
      "epoch": 168.81,
      "grad_norm": 1.8515368700027466,
      "learning_rate": 3.0485205804928835e-07,
      "loss": 0.4367,
      "step": 412400
    },
    {
      "epoch": 168.81,
      "grad_norm": 1.779062032699585,
      "learning_rate": 3.047738304977401e-07,
      "loss": 0.4254,
      "step": 412410
    },
    {
      "epoch": 168.82,
      "grad_norm": 2.053708553314209,
      "learning_rate": 3.046956122864259e-07,
      "loss": 0.4453,
      "step": 412420
    },
    {
      "epoch": 168.82,
      "grad_norm": 2.018644094467163,
      "learning_rate": 3.046174034157044e-07,
      "loss": 0.4299,
      "step": 412430
    },
    {
      "epoch": 168.83,
      "grad_norm": 1.8707101345062256,
      "learning_rate": 3.0453920388593296e-07,
      "loss": 0.4517,
      "step": 412440
    },
    {
      "epoch": 168.83,
      "grad_norm": 2.157636880874634,
      "learning_rate": 3.044610136974708e-07,
      "loss": 0.4508,
      "step": 412450
    },
    {
      "epoch": 168.83,
      "grad_norm": 1.7321090698242188,
      "learning_rate": 3.043828328506759e-07,
      "loss": 0.4505,
      "step": 412460
    },
    {
      "epoch": 168.84,
      "grad_norm": 1.8367923498153687,
      "learning_rate": 3.043046613459058e-07,
      "loss": 0.4287,
      "step": 412470
    },
    {
      "epoch": 168.84,
      "grad_norm": 1.8976691961288452,
      "learning_rate": 3.0422649918351887e-07,
      "loss": 0.4395,
      "step": 412480
    },
    {
      "epoch": 168.85,
      "grad_norm": 1.86589515209198,
      "learning_rate": 3.041483463638733e-07,
      "loss": 0.4398,
      "step": 412490
    },
    {
      "epoch": 168.85,
      "grad_norm": 1.9952017068862915,
      "learning_rate": 3.040702028873265e-07,
      "loss": 0.4424,
      "step": 412500
    },
    {
      "epoch": 168.85,
      "grad_norm": 2.017455577850342,
      "learning_rate": 3.039920687542374e-07,
      "loss": 0.4583,
      "step": 412510
    },
    {
      "epoch": 168.86,
      "grad_norm": 2.151083469390869,
      "learning_rate": 3.0391394396496363e-07,
      "loss": 0.4422,
      "step": 412520
    },
    {
      "epoch": 168.86,
      "grad_norm": 1.7330623865127563,
      "learning_rate": 3.038358285198627e-07,
      "loss": 0.434,
      "step": 412530
    },
    {
      "epoch": 168.87,
      "grad_norm": 2.0894720554351807,
      "learning_rate": 3.037577224192929e-07,
      "loss": 0.4563,
      "step": 412540
    },
    {
      "epoch": 168.87,
      "grad_norm": 1.8774187564849854,
      "learning_rate": 3.0367962566361177e-07,
      "loss": 0.4316,
      "step": 412550
    },
    {
      "epoch": 168.87,
      "grad_norm": 1.8605302572250366,
      "learning_rate": 3.0360153825317684e-07,
      "loss": 0.439,
      "step": 412560
    },
    {
      "epoch": 168.88,
      "grad_norm": 1.759556531906128,
      "learning_rate": 3.035234601883464e-07,
      "loss": 0.47,
      "step": 412570
    },
    {
      "epoch": 168.88,
      "grad_norm": 1.9309698343276978,
      "learning_rate": 3.034453914694775e-07,
      "loss": 0.4442,
      "step": 412580
    },
    {
      "epoch": 168.89,
      "grad_norm": 3.02154541015625,
      "learning_rate": 3.033673320969281e-07,
      "loss": 0.4281,
      "step": 412590
    },
    {
      "epoch": 168.89,
      "grad_norm": 2.2908523082733154,
      "learning_rate": 3.032892820710563e-07,
      "loss": 0.427,
      "step": 412600
    },
    {
      "epoch": 168.89,
      "grad_norm": 2.098461866378784,
      "learning_rate": 3.0321124139221837e-07,
      "loss": 0.4449,
      "step": 412610
    },
    {
      "epoch": 168.9,
      "grad_norm": 2.148787260055542,
      "learning_rate": 3.031332100607728e-07,
      "loss": 0.4461,
      "step": 412620
    },
    {
      "epoch": 168.9,
      "grad_norm": 2.1670982837677,
      "learning_rate": 3.030551880770764e-07,
      "loss": 0.446,
      "step": 412630
    },
    {
      "epoch": 168.91,
      "grad_norm": 1.8265424966812134,
      "learning_rate": 3.0297717544148686e-07,
      "loss": 0.4577,
      "step": 412640
    },
    {
      "epoch": 168.91,
      "grad_norm": 2.154132604598999,
      "learning_rate": 3.028991721543618e-07,
      "loss": 0.4228,
      "step": 412650
    },
    {
      "epoch": 168.92,
      "grad_norm": 2.161252737045288,
      "learning_rate": 3.028211782160582e-07,
      "loss": 0.4428,
      "step": 412660
    },
    {
      "epoch": 168.92,
      "grad_norm": 2.0079264640808105,
      "learning_rate": 3.0274319362693357e-07,
      "loss": 0.4521,
      "step": 412670
    },
    {
      "epoch": 168.92,
      "grad_norm": 2.042578935623169,
      "learning_rate": 3.0266521838734495e-07,
      "loss": 0.457,
      "step": 412680
    },
    {
      "epoch": 168.93,
      "grad_norm": 2.528604030609131,
      "learning_rate": 3.0258725249764906e-07,
      "loss": 0.4248,
      "step": 412690
    },
    {
      "epoch": 168.93,
      "grad_norm": 1.8683834075927734,
      "learning_rate": 3.0250929595820423e-07,
      "loss": 0.4135,
      "step": 412700
    },
    {
      "epoch": 168.94,
      "grad_norm": 1.9424539804458618,
      "learning_rate": 3.0243134876936686e-07,
      "loss": 0.4357,
      "step": 412710
    },
    {
      "epoch": 168.94,
      "grad_norm": 2.0106465816497803,
      "learning_rate": 3.023534109314941e-07,
      "loss": 0.435,
      "step": 412720
    },
    {
      "epoch": 168.94,
      "grad_norm": 2.1263842582702637,
      "learning_rate": 3.022754824449434e-07,
      "loss": 0.4532,
      "step": 412730
    },
    {
      "epoch": 168.95,
      "grad_norm": 2.0619940757751465,
      "learning_rate": 3.021975633100709e-07,
      "loss": 0.4432,
      "step": 412740
    },
    {
      "epoch": 168.95,
      "grad_norm": 2.1601593494415283,
      "learning_rate": 3.0211965352723377e-07,
      "loss": 0.4538,
      "step": 412750
    },
    {
      "epoch": 168.96,
      "grad_norm": 2.133307456970215,
      "learning_rate": 3.0204175309678916e-07,
      "loss": 0.4473,
      "step": 412760
    },
    {
      "epoch": 168.96,
      "grad_norm": 1.9881364107131958,
      "learning_rate": 3.019638620190939e-07,
      "loss": 0.4544,
      "step": 412770
    },
    {
      "epoch": 168.96,
      "grad_norm": 2.035943031311035,
      "learning_rate": 3.0188598029450466e-07,
      "loss": 0.4491,
      "step": 412780
    },
    {
      "epoch": 168.97,
      "grad_norm": 2.289872884750366,
      "learning_rate": 3.018081079233785e-07,
      "loss": 0.4412,
      "step": 412790
    },
    {
      "epoch": 168.97,
      "grad_norm": 8.778199195861816,
      "learning_rate": 3.017302449060716e-07,
      "loss": 0.4445,
      "step": 412800
    },
    {
      "epoch": 168.98,
      "grad_norm": 2.1329610347747803,
      "learning_rate": 3.0165239124294096e-07,
      "loss": 0.434,
      "step": 412810
    },
    {
      "epoch": 168.98,
      "grad_norm": 1.9691938161849976,
      "learning_rate": 3.015745469343434e-07,
      "loss": 0.4654,
      "step": 412820
    },
    {
      "epoch": 168.98,
      "grad_norm": 1.8509358167648315,
      "learning_rate": 3.014967119806353e-07,
      "loss": 0.4405,
      "step": 412830
    },
    {
      "epoch": 168.99,
      "grad_norm": 2.84445858001709,
      "learning_rate": 3.0141888638217326e-07,
      "loss": 0.4583,
      "step": 412840
    },
    {
      "epoch": 168.99,
      "grad_norm": 2.1205968856811523,
      "learning_rate": 3.0134107013931346e-07,
      "loss": 0.4362,
      "step": 412850
    },
    {
      "epoch": 169.0,
      "grad_norm": 2.221339464187622,
      "learning_rate": 3.0126326325241287e-07,
      "loss": 0.4623,
      "step": 412860
    },
    {
      "epoch": 169.0,
      "eval_loss": 0.4455336928367615,
      "eval_runtime": 53.085,
      "eval_samples_per_second": 64.971,
      "eval_steps_per_second": 8.138,
      "step": 412867
    },
    {
      "epoch": 169.0,
      "grad_norm": 2.675137519836426,
      "learning_rate": 3.0118546572182774e-07,
      "loss": 0.4505,
      "step": 412870
    },
    {
      "epoch": 169.01,
      "grad_norm": 2.4088096618652344,
      "learning_rate": 3.011076775479143e-07,
      "loss": 0.4665,
      "step": 412880
    },
    {
      "epoch": 169.01,
      "grad_norm": 2.1781485080718994,
      "learning_rate": 3.010298987310293e-07,
      "loss": 0.4625,
      "step": 412890
    },
    {
      "epoch": 169.01,
      "grad_norm": 2.0933969020843506,
      "learning_rate": 3.0095212927152843e-07,
      "loss": 0.4488,
      "step": 412900
    },
    {
      "epoch": 169.02,
      "grad_norm": 2.3358871936798096,
      "learning_rate": 3.008743691697682e-07,
      "loss": 0.4843,
      "step": 412910
    },
    {
      "epoch": 169.02,
      "grad_norm": 1.7521368265151978,
      "learning_rate": 3.0079661842610507e-07,
      "loss": 0.4346,
      "step": 412920
    },
    {
      "epoch": 169.03,
      "grad_norm": 1.8289772272109985,
      "learning_rate": 3.0071887704089476e-07,
      "loss": 0.4647,
      "step": 412930
    },
    {
      "epoch": 169.03,
      "grad_norm": 1.9900884628295898,
      "learning_rate": 3.0064114501449373e-07,
      "loss": 0.4297,
      "step": 412940
    },
    {
      "epoch": 169.03,
      "grad_norm": 1.6922669410705566,
      "learning_rate": 3.0056342234725795e-07,
      "loss": 0.4226,
      "step": 412950
    },
    {
      "epoch": 169.04,
      "grad_norm": 2.0779173374176025,
      "learning_rate": 3.0048570903954343e-07,
      "loss": 0.4545,
      "step": 412960
    },
    {
      "epoch": 169.04,
      "grad_norm": 2.302283763885498,
      "learning_rate": 3.004080050917064e-07,
      "loss": 0.4382,
      "step": 412970
    },
    {
      "epoch": 169.05,
      "grad_norm": 1.9119274616241455,
      "learning_rate": 3.0033031050410224e-07,
      "loss": 0.4609,
      "step": 412980
    },
    {
      "epoch": 169.05,
      "grad_norm": 2.3791937828063965,
      "learning_rate": 3.0025262527708745e-07,
      "loss": 0.4699,
      "step": 412990
    },
    {
      "epoch": 169.05,
      "grad_norm": 1.7677772045135498,
      "learning_rate": 3.0017494941101773e-07,
      "loss": 0.4476,
      "step": 413000
    },
    {
      "epoch": 169.06,
      "grad_norm": 1.8320865631103516,
      "learning_rate": 3.0009728290624863e-07,
      "loss": 0.4576,
      "step": 413010
    },
    {
      "epoch": 169.06,
      "grad_norm": 2.765807867050171,
      "learning_rate": 3.000196257631362e-07,
      "loss": 0.4476,
      "step": 413020
    },
    {
      "epoch": 169.07,
      "grad_norm": 2.022604465484619,
      "learning_rate": 2.9994197798203607e-07,
      "loss": 0.4456,
      "step": 413030
    },
    {
      "epoch": 169.07,
      "grad_norm": 1.9806855916976929,
      "learning_rate": 2.9986433956330455e-07,
      "loss": 0.4487,
      "step": 413040
    },
    {
      "epoch": 169.07,
      "grad_norm": 1.9460395574569702,
      "learning_rate": 2.997867105072962e-07,
      "loss": 0.4458,
      "step": 413050
    },
    {
      "epoch": 169.08,
      "grad_norm": 2.0110244750976562,
      "learning_rate": 2.9970909081436713e-07,
      "loss": 0.4387,
      "step": 413060
    },
    {
      "epoch": 169.08,
      "grad_norm": 2.4677188396453857,
      "learning_rate": 2.9963148048487227e-07,
      "loss": 0.4337,
      "step": 413070
    },
    {
      "epoch": 169.09,
      "grad_norm": 1.7900723218917847,
      "learning_rate": 2.995538795191682e-07,
      "loss": 0.4366,
      "step": 413080
    },
    {
      "epoch": 169.09,
      "grad_norm": 2.009324073791504,
      "learning_rate": 2.994762879176103e-07,
      "loss": 0.44,
      "step": 413090
    },
    {
      "epoch": 169.1,
      "grad_norm": 1.961635947227478,
      "learning_rate": 2.993987056805535e-07,
      "loss": 0.4474,
      "step": 413100
    },
    {
      "epoch": 169.1,
      "grad_norm": 2.173072099685669,
      "learning_rate": 2.993211328083535e-07,
      "loss": 0.4273,
      "step": 413110
    },
    {
      "epoch": 169.1,
      "grad_norm": 2.389253616333008,
      "learning_rate": 2.992435693013652e-07,
      "loss": 0.4242,
      "step": 413120
    },
    {
      "epoch": 169.11,
      "grad_norm": 1.8600252866744995,
      "learning_rate": 2.991660151599444e-07,
      "loss": 0.433,
      "step": 413130
    },
    {
      "epoch": 169.11,
      "grad_norm": 1.8845272064208984,
      "learning_rate": 2.990884703844464e-07,
      "loss": 0.441,
      "step": 413140
    },
    {
      "epoch": 169.12,
      "grad_norm": 2.0238966941833496,
      "learning_rate": 2.99010934975226e-07,
      "loss": 0.4347,
      "step": 413150
    },
    {
      "epoch": 169.12,
      "grad_norm": 1.917404055595398,
      "learning_rate": 2.989334089326386e-07,
      "loss": 0.444,
      "step": 413160
    },
    {
      "epoch": 169.12,
      "grad_norm": 2.09786057472229,
      "learning_rate": 2.9885589225703934e-07,
      "loss": 0.4527,
      "step": 413170
    },
    {
      "epoch": 169.13,
      "grad_norm": 1.8739349842071533,
      "learning_rate": 2.9877838494878373e-07,
      "loss": 0.4624,
      "step": 413180
    },
    {
      "epoch": 169.13,
      "grad_norm": 2.0068702697753906,
      "learning_rate": 2.9870088700822616e-07,
      "loss": 0.4218,
      "step": 413190
    },
    {
      "epoch": 169.14,
      "grad_norm": 2.085317373275757,
      "learning_rate": 2.986233984357215e-07,
      "loss": 0.43,
      "step": 413200
    },
    {
      "epoch": 169.14,
      "grad_norm": 2.113804817199707,
      "learning_rate": 2.9854591923162524e-07,
      "loss": 0.4472,
      "step": 413210
    },
    {
      "epoch": 169.14,
      "grad_norm": 2.128321647644043,
      "learning_rate": 2.984684493962923e-07,
      "loss": 0.4559,
      "step": 413220
    },
    {
      "epoch": 169.15,
      "grad_norm": 2.640016794204712,
      "learning_rate": 2.9839098893007706e-07,
      "loss": 0.4438,
      "step": 413230
    },
    {
      "epoch": 169.15,
      "grad_norm": 2.2422871589660645,
      "learning_rate": 2.98313537833335e-07,
      "loss": 0.4531,
      "step": 413240
    },
    {
      "epoch": 169.16,
      "grad_norm": 2.2526042461395264,
      "learning_rate": 2.9823609610642024e-07,
      "loss": 0.4371,
      "step": 413250
    },
    {
      "epoch": 169.16,
      "grad_norm": 2.0686018466949463,
      "learning_rate": 2.981586637496878e-07,
      "loss": 0.4462,
      "step": 413260
    },
    {
      "epoch": 169.16,
      "grad_norm": 2.123208522796631,
      "learning_rate": 2.9808124076349266e-07,
      "loss": 0.4527,
      "step": 413270
    },
    {
      "epoch": 169.17,
      "grad_norm": 1.8279169797897339,
      "learning_rate": 2.980038271481894e-07,
      "loss": 0.4375,
      "step": 413280
    },
    {
      "epoch": 169.17,
      "grad_norm": 1.9617807865142822,
      "learning_rate": 2.979264229041326e-07,
      "loss": 0.4311,
      "step": 413290
    },
    {
      "epoch": 169.18,
      "grad_norm": 2.0009963512420654,
      "learning_rate": 2.978490280316767e-07,
      "loss": 0.4455,
      "step": 413300
    },
    {
      "epoch": 169.18,
      "grad_norm": 2.0058321952819824,
      "learning_rate": 2.9777164253117693e-07,
      "loss": 0.449,
      "step": 413310
    },
    {
      "epoch": 169.19,
      "grad_norm": 2.384352207183838,
      "learning_rate": 2.976942664029863e-07,
      "loss": 0.4464,
      "step": 413320
    },
    {
      "epoch": 169.19,
      "grad_norm": 2.066415309906006,
      "learning_rate": 2.976168996474605e-07,
      "loss": 0.4529,
      "step": 413330
    },
    {
      "epoch": 169.19,
      "grad_norm": 2.196018934249878,
      "learning_rate": 2.9753954226495326e-07,
      "loss": 0.4668,
      "step": 413340
    },
    {
      "epoch": 169.2,
      "grad_norm": 2.0882041454315186,
      "learning_rate": 2.9746219425581944e-07,
      "loss": 0.4516,
      "step": 413350
    },
    {
      "epoch": 169.2,
      "grad_norm": 2.4208500385284424,
      "learning_rate": 2.973848556204132e-07,
      "loss": 0.4475,
      "step": 413360
    },
    {
      "epoch": 169.21,
      "grad_norm": 1.9071569442749023,
      "learning_rate": 2.9730752635908843e-07,
      "loss": 0.4444,
      "step": 413370
    },
    {
      "epoch": 169.21,
      "grad_norm": 2.0473110675811768,
      "learning_rate": 2.9723020647220004e-07,
      "loss": 0.4404,
      "step": 413380
    },
    {
      "epoch": 169.21,
      "grad_norm": 1.890225887298584,
      "learning_rate": 2.971528959601017e-07,
      "loss": 0.4487,
      "step": 413390
    },
    {
      "epoch": 169.22,
      "grad_norm": 2.4966325759887695,
      "learning_rate": 2.9707559482314796e-07,
      "loss": 0.4266,
      "step": 413400
    },
    {
      "epoch": 169.22,
      "grad_norm": 2.102544069290161,
      "learning_rate": 2.9699830306169254e-07,
      "loss": 0.4331,
      "step": 413410
    },
    {
      "epoch": 169.23,
      "grad_norm": 2.2023744583129883,
      "learning_rate": 2.9692102067608956e-07,
      "loss": 0.4338,
      "step": 413420
    },
    {
      "epoch": 169.23,
      "grad_norm": 1.849487066268921,
      "learning_rate": 2.968437476666931e-07,
      "loss": 0.4576,
      "step": 413430
    },
    {
      "epoch": 169.23,
      "grad_norm": 2.0326590538024902,
      "learning_rate": 2.9676648403385695e-07,
      "loss": 0.4406,
      "step": 413440
    },
    {
      "epoch": 169.24,
      "grad_norm": 1.8459181785583496,
      "learning_rate": 2.9668922977793613e-07,
      "loss": 0.4461,
      "step": 413450
    },
    {
      "epoch": 169.24,
      "grad_norm": 1.7981600761413574,
      "learning_rate": 2.966119848992833e-07,
      "loss": 0.4468,
      "step": 413460
    },
    {
      "epoch": 169.25,
      "grad_norm": 2.1257057189941406,
      "learning_rate": 2.965347493982526e-07,
      "loss": 0.4618,
      "step": 413470
    },
    {
      "epoch": 169.25,
      "grad_norm": 1.9954277276992798,
      "learning_rate": 2.9645752327519784e-07,
      "loss": 0.4647,
      "step": 413480
    },
    {
      "epoch": 169.26,
      "grad_norm": 2.1585869789123535,
      "learning_rate": 2.9638030653047276e-07,
      "loss": 0.4364,
      "step": 413490
    },
    {
      "epoch": 169.26,
      "grad_norm": 2.127081871032715,
      "learning_rate": 2.9630309916443143e-07,
      "loss": 0.44,
      "step": 413500
    },
    {
      "epoch": 169.26,
      "grad_norm": 1.8425250053405762,
      "learning_rate": 2.96225901177427e-07,
      "loss": 0.4447,
      "step": 413510
    },
    {
      "epoch": 169.27,
      "grad_norm": 1.7482972145080566,
      "learning_rate": 2.961487125698135e-07,
      "loss": 0.4598,
      "step": 413520
    },
    {
      "epoch": 169.27,
      "grad_norm": 2.1712279319763184,
      "learning_rate": 2.960715333419445e-07,
      "loss": 0.4501,
      "step": 413530
    },
    {
      "epoch": 169.28,
      "grad_norm": 1.9493188858032227,
      "learning_rate": 2.959943634941734e-07,
      "loss": 0.4487,
      "step": 413540
    },
    {
      "epoch": 169.28,
      "grad_norm": 2.093845844268799,
      "learning_rate": 2.959172030268534e-07,
      "loss": 0.4344,
      "step": 413550
    },
    {
      "epoch": 169.28,
      "grad_norm": 2.6300485134124756,
      "learning_rate": 2.958400519403387e-07,
      "loss": 0.4605,
      "step": 413560
    },
    {
      "epoch": 169.29,
      "grad_norm": 2.008617639541626,
      "learning_rate": 2.957629102349821e-07,
      "loss": 0.4678,
      "step": 413570
    },
    {
      "epoch": 169.29,
      "grad_norm": 1.9559369087219238,
      "learning_rate": 2.956857779111374e-07,
      "loss": 0.4639,
      "step": 413580
    },
    {
      "epoch": 169.3,
      "grad_norm": 2.148876667022705,
      "learning_rate": 2.956086549691576e-07,
      "loss": 0.4602,
      "step": 413590
    },
    {
      "epoch": 169.3,
      "grad_norm": 2.0948898792266846,
      "learning_rate": 2.9553154140939594e-07,
      "loss": 0.4578,
      "step": 413600
    },
    {
      "epoch": 169.3,
      "grad_norm": 1.6488829851150513,
      "learning_rate": 2.954544372322061e-07,
      "loss": 0.4374,
      "step": 413610
    },
    {
      "epoch": 169.31,
      "grad_norm": 1.8929659128189087,
      "learning_rate": 2.953773424379414e-07,
      "loss": 0.4354,
      "step": 413620
    },
    {
      "epoch": 169.31,
      "grad_norm": 1.765169382095337,
      "learning_rate": 2.953002570269534e-07,
      "loss": 0.4289,
      "step": 413630
    },
    {
      "epoch": 169.32,
      "grad_norm": 2.015657424926758,
      "learning_rate": 2.952231809995973e-07,
      "loss": 0.4506,
      "step": 413640
    },
    {
      "epoch": 169.32,
      "grad_norm": 1.7910765409469604,
      "learning_rate": 2.951461143562252e-07,
      "loss": 0.4436,
      "step": 413650
    },
    {
      "epoch": 169.32,
      "grad_norm": 1.9791728258132935,
      "learning_rate": 2.950690570971901e-07,
      "loss": 0.4583,
      "step": 413660
    },
    {
      "epoch": 169.33,
      "grad_norm": 1.9869208335876465,
      "learning_rate": 2.949920092228452e-07,
      "loss": 0.4361,
      "step": 413670
    },
    {
      "epoch": 169.33,
      "grad_norm": 2.175607442855835,
      "learning_rate": 2.949149707335433e-07,
      "loss": 0.4377,
      "step": 413680
    },
    {
      "epoch": 169.34,
      "grad_norm": 2.230167865753174,
      "learning_rate": 2.948379416296375e-07,
      "loss": 0.4695,
      "step": 413690
    },
    {
      "epoch": 169.34,
      "grad_norm": 2.2435784339904785,
      "learning_rate": 2.947609219114804e-07,
      "loss": 0.4265,
      "step": 413700
    },
    {
      "epoch": 169.35,
      "grad_norm": 2.4837727546691895,
      "learning_rate": 2.9468391157942505e-07,
      "loss": 0.4215,
      "step": 413710
    },
    {
      "epoch": 169.35,
      "grad_norm": 1.9962158203125,
      "learning_rate": 2.946069106338241e-07,
      "loss": 0.4339,
      "step": 413720
    },
    {
      "epoch": 169.35,
      "grad_norm": 2.4908535480499268,
      "learning_rate": 2.945299190750301e-07,
      "loss": 0.4392,
      "step": 413730
    },
    {
      "epoch": 169.36,
      "grad_norm": 1.8803163766860962,
      "learning_rate": 2.9445293690339616e-07,
      "loss": 0.4584,
      "step": 413740
    },
    {
      "epoch": 169.36,
      "grad_norm": 1.531284213066101,
      "learning_rate": 2.943759641192746e-07,
      "loss": 0.4442,
      "step": 413750
    },
    {
      "epoch": 169.37,
      "grad_norm": 2.4007890224456787,
      "learning_rate": 2.9429900072301854e-07,
      "loss": 0.4631,
      "step": 413760
    },
    {
      "epoch": 169.37,
      "grad_norm": 1.937532663345337,
      "learning_rate": 2.942220467149797e-07,
      "loss": 0.4537,
      "step": 413770
    },
    {
      "epoch": 169.37,
      "grad_norm": 2.000843048095703,
      "learning_rate": 2.941451020955111e-07,
      "loss": 0.4422,
      "step": 413780
    },
    {
      "epoch": 169.38,
      "grad_norm": 1.8829576969146729,
      "learning_rate": 2.940681668649649e-07,
      "loss": 0.4316,
      "step": 413790
    },
    {
      "epoch": 169.38,
      "grad_norm": 2.1101202964782715,
      "learning_rate": 2.93991241023694e-07,
      "loss": 0.4593,
      "step": 413800
    },
    {
      "epoch": 169.39,
      "grad_norm": 1.9656351804733276,
      "learning_rate": 2.939143245720502e-07,
      "loss": 0.4362,
      "step": 413810
    },
    {
      "epoch": 169.39,
      "grad_norm": 1.8964787721633911,
      "learning_rate": 2.9383741751038605e-07,
      "loss": 0.4428,
      "step": 413820
    },
    {
      "epoch": 169.39,
      "grad_norm": 2.521383047103882,
      "learning_rate": 2.937605198390542e-07,
      "loss": 0.4559,
      "step": 413830
    },
    {
      "epoch": 169.4,
      "grad_norm": 2.0415000915527344,
      "learning_rate": 2.9368363155840664e-07,
      "loss": 0.429,
      "step": 413840
    },
    {
      "epoch": 169.4,
      "grad_norm": 2.370109796524048,
      "learning_rate": 2.9360675266879574e-07,
      "loss": 0.4619,
      "step": 413850
    },
    {
      "epoch": 169.41,
      "grad_norm": 2.0190937519073486,
      "learning_rate": 2.9352988317057354e-07,
      "loss": 0.4421,
      "step": 413860
    },
    {
      "epoch": 169.41,
      "grad_norm": 2.9221489429473877,
      "learning_rate": 2.9345302306409237e-07,
      "loss": 0.4276,
      "step": 413870
    },
    {
      "epoch": 169.41,
      "grad_norm": 2.164350748062134,
      "learning_rate": 2.9337617234970383e-07,
      "loss": 0.4363,
      "step": 413880
    },
    {
      "epoch": 169.42,
      "grad_norm": 2.5085978507995605,
      "learning_rate": 2.9329933102776036e-07,
      "loss": 0.4325,
      "step": 413890
    },
    {
      "epoch": 169.42,
      "grad_norm": 2.4432051181793213,
      "learning_rate": 2.932224990986144e-07,
      "loss": 0.4506,
      "step": 413900
    },
    {
      "epoch": 169.43,
      "grad_norm": 2.2760367393493652,
      "learning_rate": 2.931456765626167e-07,
      "loss": 0.4369,
      "step": 413910
    },
    {
      "epoch": 169.43,
      "grad_norm": 2.1707229614257812,
      "learning_rate": 2.9306886342012005e-07,
      "loss": 0.4332,
      "step": 413920
    },
    {
      "epoch": 169.44,
      "grad_norm": 2.082202672958374,
      "learning_rate": 2.9299205967147603e-07,
      "loss": 0.4503,
      "step": 413930
    },
    {
      "epoch": 169.44,
      "grad_norm": 2.651031017303467,
      "learning_rate": 2.9291526531703647e-07,
      "loss": 0.4629,
      "step": 413940
    },
    {
      "epoch": 169.44,
      "grad_norm": 1.8610403537750244,
      "learning_rate": 2.9283848035715336e-07,
      "loss": 0.4521,
      "step": 413950
    },
    {
      "epoch": 169.45,
      "grad_norm": 2.1629178524017334,
      "learning_rate": 2.9276170479217806e-07,
      "loss": 0.4441,
      "step": 413960
    },
    {
      "epoch": 169.45,
      "grad_norm": 1.8090587854385376,
      "learning_rate": 2.9268493862246256e-07,
      "loss": 0.4331,
      "step": 413970
    },
    {
      "epoch": 169.46,
      "grad_norm": 2.2280969619750977,
      "learning_rate": 2.9260818184835865e-07,
      "loss": 0.435,
      "step": 413980
    },
    {
      "epoch": 169.46,
      "grad_norm": 2.01395845413208,
      "learning_rate": 2.9253143447021774e-07,
      "loss": 0.4446,
      "step": 413990
    },
    {
      "epoch": 169.46,
      "grad_norm": 1.6927571296691895,
      "learning_rate": 2.924546964883912e-07,
      "loss": 0.4486,
      "step": 414000
    },
    {
      "epoch": 169.47,
      "grad_norm": 2.1578903198242188,
      "learning_rate": 2.923779679032302e-07,
      "loss": 0.4618,
      "step": 414010
    },
    {
      "epoch": 169.47,
      "grad_norm": 1.8962315320968628,
      "learning_rate": 2.923012487150875e-07,
      "loss": 0.4365,
      "step": 414020
    },
    {
      "epoch": 169.48,
      "grad_norm": 2.5293188095092773,
      "learning_rate": 2.922245389243144e-07,
      "loss": 0.4455,
      "step": 414030
    },
    {
      "epoch": 169.48,
      "grad_norm": 2.174882411956787,
      "learning_rate": 2.9214783853126115e-07,
      "loss": 0.4503,
      "step": 414040
    },
    {
      "epoch": 169.48,
      "grad_norm": 1.9762518405914307,
      "learning_rate": 2.920711475362795e-07,
      "loss": 0.4565,
      "step": 414050
    },
    {
      "epoch": 169.49,
      "grad_norm": 2.354735851287842,
      "learning_rate": 2.9199446593972107e-07,
      "loss": 0.4507,
      "step": 414060
    },
    {
      "epoch": 169.49,
      "grad_norm": 2.3615102767944336,
      "learning_rate": 2.919177937419369e-07,
      "loss": 0.4492,
      "step": 414070
    },
    {
      "epoch": 169.5,
      "grad_norm": 2.2044081687927246,
      "learning_rate": 2.918411309432784e-07,
      "loss": 0.4778,
      "step": 414080
    },
    {
      "epoch": 169.5,
      "grad_norm": 1.9467980861663818,
      "learning_rate": 2.9176447754409643e-07,
      "loss": 0.4584,
      "step": 414090
    },
    {
      "epoch": 169.5,
      "grad_norm": 2.483116626739502,
      "learning_rate": 2.916878335447425e-07,
      "loss": 0.4472,
      "step": 414100
    },
    {
      "epoch": 169.51,
      "grad_norm": 1.7741601467132568,
      "learning_rate": 2.9161119894556766e-07,
      "loss": 0.4546,
      "step": 414110
    },
    {
      "epoch": 169.51,
      "grad_norm": 2.3467915058135986,
      "learning_rate": 2.9153457374692295e-07,
      "loss": 0.4205,
      "step": 414120
    },
    {
      "epoch": 169.52,
      "grad_norm": 2.090341091156006,
      "learning_rate": 2.91457957949159e-07,
      "loss": 0.4589,
      "step": 414130
    },
    {
      "epoch": 169.52,
      "grad_norm": 1.967003583908081,
      "learning_rate": 2.9138135155262704e-07,
      "loss": 0.4607,
      "step": 414140
    },
    {
      "epoch": 169.53,
      "grad_norm": 2.1607840061187744,
      "learning_rate": 2.9130475455767814e-07,
      "loss": 0.4414,
      "step": 414150
    },
    {
      "epoch": 169.53,
      "grad_norm": 1.9051750898361206,
      "learning_rate": 2.9122816696466323e-07,
      "loss": 0.4494,
      "step": 414160
    },
    {
      "epoch": 169.53,
      "grad_norm": 2.9422497749328613,
      "learning_rate": 2.911515887739327e-07,
      "loss": 0.4451,
      "step": 414170
    },
    {
      "epoch": 169.54,
      "grad_norm": 2.0636565685272217,
      "learning_rate": 2.9107501998583763e-07,
      "loss": 0.4528,
      "step": 414180
    },
    {
      "epoch": 169.54,
      "grad_norm": 2.1404755115509033,
      "learning_rate": 2.9099846060072876e-07,
      "loss": 0.447,
      "step": 414190
    },
    {
      "epoch": 169.55,
      "grad_norm": 2.1045584678649902,
      "learning_rate": 2.9092191061895684e-07,
      "loss": 0.455,
      "step": 414200
    },
    {
      "epoch": 169.55,
      "grad_norm": 1.895038366317749,
      "learning_rate": 2.908453700408724e-07,
      "loss": 0.4493,
      "step": 414210
    },
    {
      "epoch": 169.55,
      "grad_norm": 2.3371670246124268,
      "learning_rate": 2.9076883886682617e-07,
      "loss": 0.4658,
      "step": 414220
    },
    {
      "epoch": 169.56,
      "grad_norm": 2.032259941101074,
      "learning_rate": 2.906923170971689e-07,
      "loss": 0.4417,
      "step": 414230
    },
    {
      "epoch": 169.56,
      "grad_norm": 2.179401397705078,
      "learning_rate": 2.906158047322507e-07,
      "loss": 0.4299,
      "step": 414240
    },
    {
      "epoch": 169.57,
      "grad_norm": 1.85088050365448,
      "learning_rate": 2.905393017724224e-07,
      "loss": 0.4362,
      "step": 414250
    },
    {
      "epoch": 169.57,
      "grad_norm": 1.9774072170257568,
      "learning_rate": 2.9046280821803405e-07,
      "loss": 0.433,
      "step": 414260
    },
    {
      "epoch": 169.57,
      "grad_norm": 2.165813446044922,
      "learning_rate": 2.903863240694365e-07,
      "loss": 0.4514,
      "step": 414270
    },
    {
      "epoch": 169.58,
      "grad_norm": 2.143338441848755,
      "learning_rate": 2.903098493269801e-07,
      "loss": 0.4323,
      "step": 414280
    },
    {
      "epoch": 169.58,
      "grad_norm": 2.0114924907684326,
      "learning_rate": 2.90233383991015e-07,
      "loss": 0.4062,
      "step": 414290
    },
    {
      "epoch": 169.59,
      "grad_norm": 1.8066014051437378,
      "learning_rate": 2.901569280618913e-07,
      "loss": 0.4166,
      "step": 414300
    },
    {
      "epoch": 169.59,
      "grad_norm": 2.0056560039520264,
      "learning_rate": 2.9008048153995953e-07,
      "loss": 0.4258,
      "step": 414310
    },
    {
      "epoch": 169.59,
      "grad_norm": 2.3967621326446533,
      "learning_rate": 2.9000404442556977e-07,
      "loss": 0.4459,
      "step": 414320
    },
    {
      "epoch": 169.6,
      "grad_norm": 1.8768337965011597,
      "learning_rate": 2.8992761671907214e-07,
      "loss": 0.4401,
      "step": 414330
    },
    {
      "epoch": 169.6,
      "grad_norm": 4.490978717803955,
      "learning_rate": 2.8985119842081716e-07,
      "loss": 0.4415,
      "step": 414340
    },
    {
      "epoch": 169.61,
      "grad_norm": 2.5777344703674316,
      "learning_rate": 2.8977478953115407e-07,
      "loss": 0.4474,
      "step": 414350
    },
    {
      "epoch": 169.61,
      "grad_norm": 1.8063185214996338,
      "learning_rate": 2.896983900504334e-07,
      "loss": 0.4542,
      "step": 414360
    },
    {
      "epoch": 169.62,
      "grad_norm": 1.952684760093689,
      "learning_rate": 2.8962199997900504e-07,
      "loss": 0.4636,
      "step": 414370
    },
    {
      "epoch": 169.62,
      "grad_norm": 2.755967617034912,
      "learning_rate": 2.89545619317219e-07,
      "loss": 0.4447,
      "step": 414380
    },
    {
      "epoch": 169.62,
      "grad_norm": 2.379539728164673,
      "learning_rate": 2.8946924806542454e-07,
      "loss": 0.459,
      "step": 414390
    },
    {
      "epoch": 169.63,
      "grad_norm": 2.59995436668396,
      "learning_rate": 2.8939288622397236e-07,
      "loss": 0.43,
      "step": 414400
    },
    {
      "epoch": 169.63,
      "grad_norm": 1.864492416381836,
      "learning_rate": 2.893165337932122e-07,
      "loss": 0.4354,
      "step": 414410
    },
    {
      "epoch": 169.64,
      "grad_norm": 2.487514019012451,
      "learning_rate": 2.892401907734935e-07,
      "loss": 0.4357,
      "step": 414420
    },
    {
      "epoch": 169.64,
      "grad_norm": 2.0571000576019287,
      "learning_rate": 2.89163857165166e-07,
      "loss": 0.4486,
      "step": 414430
    },
    {
      "epoch": 169.64,
      "grad_norm": 2.036410331726074,
      "learning_rate": 2.890875329685794e-07,
      "loss": 0.4488,
      "step": 414440
    },
    {
      "epoch": 169.65,
      "grad_norm": 1.9303596019744873,
      "learning_rate": 2.890112181840835e-07,
      "loss": 0.4346,
      "step": 414450
    },
    {
      "epoch": 169.65,
      "grad_norm": 2.022799491882324,
      "learning_rate": 2.889349128120277e-07,
      "loss": 0.4531,
      "step": 414460
    },
    {
      "epoch": 169.66,
      "grad_norm": 1.757656455039978,
      "learning_rate": 2.8885861685276166e-07,
      "loss": 0.4566,
      "step": 414470
    },
    {
      "epoch": 169.66,
      "grad_norm": 1.8901009559631348,
      "learning_rate": 2.8878233030663514e-07,
      "loss": 0.4369,
      "step": 414480
    },
    {
      "epoch": 169.66,
      "grad_norm": 2.195840835571289,
      "learning_rate": 2.887060531739969e-07,
      "loss": 0.4294,
      "step": 414490
    },
    {
      "epoch": 169.67,
      "grad_norm": 2.032041549682617,
      "learning_rate": 2.886297854551966e-07,
      "loss": 0.441,
      "step": 414500
    },
    {
      "epoch": 169.67,
      "grad_norm": 1.906268835067749,
      "learning_rate": 2.885535271505839e-07,
      "loss": 0.4398,
      "step": 414510
    },
    {
      "epoch": 169.68,
      "grad_norm": 1.7596865892410278,
      "learning_rate": 2.8847727826050787e-07,
      "loss": 0.4382,
      "step": 414520
    },
    {
      "epoch": 169.68,
      "grad_norm": 1.955135464668274,
      "learning_rate": 2.8840103878531783e-07,
      "loss": 0.4648,
      "step": 414530
    },
    {
      "epoch": 169.68,
      "grad_norm": 2.357891321182251,
      "learning_rate": 2.8832480872536307e-07,
      "loss": 0.4216,
      "step": 414540
    },
    {
      "epoch": 169.69,
      "grad_norm": 2.9748892784118652,
      "learning_rate": 2.882485880809928e-07,
      "loss": 0.4575,
      "step": 414550
    },
    {
      "epoch": 169.69,
      "grad_norm": 2.1675941944122314,
      "learning_rate": 2.8817237685255614e-07,
      "loss": 0.4536,
      "step": 414560
    },
    {
      "epoch": 169.7,
      "grad_norm": 1.854062795639038,
      "learning_rate": 2.8809617504040214e-07,
      "loss": 0.4468,
      "step": 414570
    },
    {
      "epoch": 169.7,
      "grad_norm": 2.056321382522583,
      "learning_rate": 2.8801998264487964e-07,
      "loss": 0.4374,
      "step": 414580
    },
    {
      "epoch": 169.71,
      "grad_norm": 2.114058494567871,
      "learning_rate": 2.8794379966633836e-07,
      "loss": 0.4458,
      "step": 414590
    },
    {
      "epoch": 169.71,
      "grad_norm": 2.4475762844085693,
      "learning_rate": 2.8786762610512707e-07,
      "loss": 0.4437,
      "step": 414600
    },
    {
      "epoch": 169.71,
      "grad_norm": 2.2913267612457275,
      "learning_rate": 2.877914619615944e-07,
      "loss": 0.4622,
      "step": 414610
    },
    {
      "epoch": 169.72,
      "grad_norm": 2.2753653526306152,
      "learning_rate": 2.8771530723609006e-07,
      "loss": 0.4408,
      "step": 414620
    },
    {
      "epoch": 169.72,
      "grad_norm": 1.9658235311508179,
      "learning_rate": 2.876391619289614e-07,
      "loss": 0.4306,
      "step": 414630
    },
    {
      "epoch": 169.73,
      "grad_norm": 2.090611696243286,
      "learning_rate": 2.875630260405582e-07,
      "loss": 0.4471,
      "step": 414640
    },
    {
      "epoch": 169.73,
      "grad_norm": 1.8152165412902832,
      "learning_rate": 2.874868995712291e-07,
      "loss": 0.444,
      "step": 414650
    },
    {
      "epoch": 169.73,
      "grad_norm": 2.109863042831421,
      "learning_rate": 2.874107825213231e-07,
      "loss": 0.4458,
      "step": 414660
    },
    {
      "epoch": 169.74,
      "grad_norm": 2.2552592754364014,
      "learning_rate": 2.873346748911882e-07,
      "loss": 0.437,
      "step": 414670
    },
    {
      "epoch": 169.74,
      "grad_norm": 2.4403679370880127,
      "learning_rate": 2.872585766811736e-07,
      "loss": 0.4293,
      "step": 414680
    },
    {
      "epoch": 169.75,
      "grad_norm": 2.0637948513031006,
      "learning_rate": 2.8718248789162773e-07,
      "loss": 0.4485,
      "step": 414690
    },
    {
      "epoch": 169.75,
      "grad_norm": 2.489572525024414,
      "learning_rate": 2.871064085228989e-07,
      "loss": 0.4239,
      "step": 414700
    },
    {
      "epoch": 169.75,
      "grad_norm": 1.9705510139465332,
      "learning_rate": 2.8703033857533623e-07,
      "loss": 0.4289,
      "step": 414710
    },
    {
      "epoch": 169.76,
      "grad_norm": 2.0863566398620605,
      "learning_rate": 2.8695427804928744e-07,
      "loss": 0.467,
      "step": 414720
    },
    {
      "epoch": 169.76,
      "grad_norm": 1.9690361022949219,
      "learning_rate": 2.8687822694510164e-07,
      "loss": 0.4654,
      "step": 414730
    },
    {
      "epoch": 169.77,
      "grad_norm": 1.97850501537323,
      "learning_rate": 2.8680218526312675e-07,
      "loss": 0.43,
      "step": 414740
    },
    {
      "epoch": 169.77,
      "grad_norm": 2.1318843364715576,
      "learning_rate": 2.867261530037112e-07,
      "loss": 0.4528,
      "step": 414750
    },
    {
      "epoch": 169.77,
      "grad_norm": 2.143136978149414,
      "learning_rate": 2.8665013016720323e-07,
      "loss": 0.4467,
      "step": 414760
    },
    {
      "epoch": 169.78,
      "grad_norm": 1.7858915328979492,
      "learning_rate": 2.865741167539514e-07,
      "loss": 0.4308,
      "step": 414770
    },
    {
      "epoch": 169.78,
      "grad_norm": 1.8700437545776367,
      "learning_rate": 2.864981127643036e-07,
      "loss": 0.444,
      "step": 414780
    },
    {
      "epoch": 169.79,
      "grad_norm": 1.8361093997955322,
      "learning_rate": 2.8642211819860816e-07,
      "loss": 0.4545,
      "step": 414790
    },
    {
      "epoch": 169.79,
      "grad_norm": 1.9512004852294922,
      "learning_rate": 2.8634613305721307e-07,
      "loss": 0.4494,
      "step": 414800
    },
    {
      "epoch": 169.8,
      "grad_norm": 1.9830923080444336,
      "learning_rate": 2.8627015734046624e-07,
      "loss": 0.4366,
      "step": 414810
    },
    {
      "epoch": 169.8,
      "grad_norm": 2.032538652420044,
      "learning_rate": 2.8619419104871615e-07,
      "loss": 0.4361,
      "step": 414820
    },
    {
      "epoch": 169.8,
      "grad_norm": 2.277836799621582,
      "learning_rate": 2.8611823418231074e-07,
      "loss": 0.4379,
      "step": 414830
    },
    {
      "epoch": 169.81,
      "grad_norm": 2.004878520965576,
      "learning_rate": 2.8604228674159754e-07,
      "loss": 0.4547,
      "step": 414840
    },
    {
      "epoch": 169.81,
      "grad_norm": 2.2576260566711426,
      "learning_rate": 2.8596634872692473e-07,
      "loss": 0.4333,
      "step": 414850
    },
    {
      "epoch": 169.82,
      "grad_norm": 1.9187861680984497,
      "learning_rate": 2.8589042013864025e-07,
      "loss": 0.438,
      "step": 414860
    },
    {
      "epoch": 169.82,
      "grad_norm": 1.794217586517334,
      "learning_rate": 2.8581450097709153e-07,
      "loss": 0.4789,
      "step": 414870
    },
    {
      "epoch": 169.82,
      "grad_norm": 2.1476500034332275,
      "learning_rate": 2.8573859124262686e-07,
      "loss": 0.4453,
      "step": 414880
    },
    {
      "epoch": 169.83,
      "grad_norm": 2.168241500854492,
      "learning_rate": 2.856626909355937e-07,
      "loss": 0.4368,
      "step": 414890
    },
    {
      "epoch": 169.83,
      "grad_norm": 2.2345926761627197,
      "learning_rate": 2.855868000563399e-07,
      "loss": 0.4389,
      "step": 414900
    },
    {
      "epoch": 169.84,
      "grad_norm": 1.9684969186782837,
      "learning_rate": 2.855109186052126e-07,
      "loss": 0.4494,
      "step": 414910
    },
    {
      "epoch": 169.84,
      "grad_norm": 2.047325849533081,
      "learning_rate": 2.8543504658256006e-07,
      "loss": 0.4588,
      "step": 414920
    },
    {
      "epoch": 169.84,
      "grad_norm": 2.101850986480713,
      "learning_rate": 2.8535918398873003e-07,
      "loss": 0.4536,
      "step": 414930
    },
    {
      "epoch": 169.85,
      "grad_norm": 1.755854845046997,
      "learning_rate": 2.852833308240691e-07,
      "loss": 0.427,
      "step": 414940
    },
    {
      "epoch": 169.85,
      "grad_norm": 2.090031623840332,
      "learning_rate": 2.852074870889245e-07,
      "loss": 0.4614,
      "step": 414950
    },
    {
      "epoch": 169.86,
      "grad_norm": 1.787840723991394,
      "learning_rate": 2.8513165278364494e-07,
      "loss": 0.4419,
      "step": 414960
    },
    {
      "epoch": 169.86,
      "grad_norm": 1.7935179471969604,
      "learning_rate": 2.850558279085774e-07,
      "loss": 0.4415,
      "step": 414970
    },
    {
      "epoch": 169.86,
      "grad_norm": 2.040858745574951,
      "learning_rate": 2.8498001246406875e-07,
      "loss": 0.4554,
      "step": 414980
    },
    {
      "epoch": 169.87,
      "grad_norm": 2.6492650508880615,
      "learning_rate": 2.8490420645046676e-07,
      "loss": 0.4598,
      "step": 414990
    },
    {
      "epoch": 169.87,
      "grad_norm": 2.411372184753418,
      "learning_rate": 2.8482840986811813e-07,
      "loss": 0.4268,
      "step": 415000
    },
    {
      "epoch": 169.88,
      "grad_norm": 1.602745532989502,
      "learning_rate": 2.847526227173709e-07,
      "loss": 0.4445,
      "step": 415010
    },
    {
      "epoch": 169.88,
      "grad_norm": 2.5189905166625977,
      "learning_rate": 2.8467684499857155e-07,
      "loss": 0.436,
      "step": 415020
    },
    {
      "epoch": 169.89,
      "grad_norm": 1.8056010007858276,
      "learning_rate": 2.8460107671206754e-07,
      "loss": 0.4424,
      "step": 415030
    },
    {
      "epoch": 169.89,
      "grad_norm": 2.160094738006592,
      "learning_rate": 2.8452531785820566e-07,
      "loss": 0.4335,
      "step": 415040
    },
    {
      "epoch": 169.89,
      "grad_norm": 2.3877816200256348,
      "learning_rate": 2.844495684373334e-07,
      "loss": 0.4502,
      "step": 415050
    },
    {
      "epoch": 169.9,
      "grad_norm": 1.8206661939620972,
      "learning_rate": 2.843738284497977e-07,
      "loss": 0.4351,
      "step": 415060
    },
    {
      "epoch": 169.9,
      "grad_norm": 2.049135208129883,
      "learning_rate": 2.842980978959451e-07,
      "loss": 0.4515,
      "step": 415070
    },
    {
      "epoch": 169.91,
      "grad_norm": 1.8483397960662842,
      "learning_rate": 2.8422237677612255e-07,
      "loss": 0.4558,
      "step": 415080
    },
    {
      "epoch": 169.91,
      "grad_norm": 1.7640656232833862,
      "learning_rate": 2.84146665090677e-07,
      "loss": 0.4474,
      "step": 415090
    },
    {
      "epoch": 169.91,
      "grad_norm": 1.9539731740951538,
      "learning_rate": 2.8407096283995553e-07,
      "loss": 0.4271,
      "step": 415100
    },
    {
      "epoch": 169.92,
      "grad_norm": 1.7246309518814087,
      "learning_rate": 2.839952700243048e-07,
      "loss": 0.436,
      "step": 415110
    },
    {
      "epoch": 169.92,
      "grad_norm": 2.2383017539978027,
      "learning_rate": 2.839195866440713e-07,
      "loss": 0.4479,
      "step": 415120
    },
    {
      "epoch": 169.93,
      "grad_norm": 2.4112253189086914,
      "learning_rate": 2.8384391269960203e-07,
      "loss": 0.4386,
      "step": 415130
    },
    {
      "epoch": 169.93,
      "grad_norm": 2.0569205284118652,
      "learning_rate": 2.8376824819124316e-07,
      "loss": 0.4464,
      "step": 415140
    },
    {
      "epoch": 169.93,
      "grad_norm": 1.9096126556396484,
      "learning_rate": 2.8369259311934197e-07,
      "loss": 0.4634,
      "step": 415150
    },
    {
      "epoch": 169.94,
      "grad_norm": 1.8532298803329468,
      "learning_rate": 2.836169474842447e-07,
      "loss": 0.4557,
      "step": 415160
    },
    {
      "epoch": 169.94,
      "grad_norm": 2.2511167526245117,
      "learning_rate": 2.8354131128629793e-07,
      "loss": 0.431,
      "step": 415170
    },
    {
      "epoch": 169.95,
      "grad_norm": 2.553968906402588,
      "learning_rate": 2.8346568452584806e-07,
      "loss": 0.4514,
      "step": 415180
    },
    {
      "epoch": 169.95,
      "grad_norm": 1.8480048179626465,
      "learning_rate": 2.8339006720324167e-07,
      "loss": 0.4659,
      "step": 415190
    },
    {
      "epoch": 169.95,
      "grad_norm": 2.257742404937744,
      "learning_rate": 2.833144593188256e-07,
      "loss": 0.4619,
      "step": 415200
    },
    {
      "epoch": 169.96,
      "grad_norm": 1.6256146430969238,
      "learning_rate": 2.8323886087294494e-07,
      "loss": 0.4346,
      "step": 415210
    },
    {
      "epoch": 169.96,
      "grad_norm": 2.307023763656616,
      "learning_rate": 2.83163271865947e-07,
      "loss": 0.4387,
      "step": 415220
    },
    {
      "epoch": 169.97,
      "grad_norm": 2.3464791774749756,
      "learning_rate": 2.830876922981774e-07,
      "loss": 0.4534,
      "step": 415230
    },
    {
      "epoch": 169.97,
      "grad_norm": 1.8822369575500488,
      "learning_rate": 2.830121221699829e-07,
      "loss": 0.4481,
      "step": 415240
    },
    {
      "epoch": 169.98,
      "grad_norm": 1.9965355396270752,
      "learning_rate": 2.829365614817094e-07,
      "loss": 0.4562,
      "step": 415250
    },
    {
      "epoch": 169.98,
      "grad_norm": 1.8783786296844482,
      "learning_rate": 2.828610102337032e-07,
      "loss": 0.4384,
      "step": 415260
    },
    {
      "epoch": 169.98,
      "grad_norm": 2.546288251876831,
      "learning_rate": 2.827854684263102e-07,
      "loss": 0.4509,
      "step": 415270
    },
    {
      "epoch": 169.99,
      "grad_norm": 2.3162620067596436,
      "learning_rate": 2.8270993605987665e-07,
      "loss": 0.4588,
      "step": 415280
    },
    {
      "epoch": 169.99,
      "grad_norm": 1.950498104095459,
      "learning_rate": 2.826344131347481e-07,
      "loss": 0.4413,
      "step": 415290
    },
    {
      "epoch": 170.0,
      "grad_norm": 2.175699472427368,
      "learning_rate": 2.825588996512712e-07,
      "loss": 0.4413,
      "step": 415300
    },
    {
      "epoch": 170.0,
      "grad_norm": 2.0335848331451416,
      "learning_rate": 2.8248339560979147e-07,
      "loss": 0.4673,
      "step": 415310
    },
    {
      "epoch": 170.0,
      "eval_loss": 0.44553449749946594,
      "eval_runtime": 52.3526,
      "eval_samples_per_second": 65.88,
      "eval_steps_per_second": 8.252,
      "step": 415310
    },
    {
      "epoch": 170.0,
      "grad_norm": 2.01133131980896,
      "learning_rate": 2.8240790101065473e-07,
      "loss": 0.4479,
      "step": 415320
    },
    {
      "epoch": 170.01,
      "grad_norm": 1.8241500854492188,
      "learning_rate": 2.823324158542068e-07,
      "loss": 0.4616,
      "step": 415330
    },
    {
      "epoch": 170.01,
      "grad_norm": 2.1694345474243164,
      "learning_rate": 2.822569401407938e-07,
      "loss": 0.458,
      "step": 415340
    },
    {
      "epoch": 170.02,
      "grad_norm": 1.7894283533096313,
      "learning_rate": 2.821814738707613e-07,
      "loss": 0.4531,
      "step": 415350
    },
    {
      "epoch": 170.02,
      "grad_norm": 2.4193058013916016,
      "learning_rate": 2.8210601704445476e-07,
      "loss": 0.4699,
      "step": 415360
    },
    {
      "epoch": 170.02,
      "grad_norm": 2.185669183731079,
      "learning_rate": 2.8203056966222004e-07,
      "loss": 0.4486,
      "step": 415370
    },
    {
      "epoch": 170.03,
      "grad_norm": 1.8913332223892212,
      "learning_rate": 2.819551317244027e-07,
      "loss": 0.4417,
      "step": 415380
    },
    {
      "epoch": 170.03,
      "grad_norm": 6.734447479248047,
      "learning_rate": 2.8187970323134854e-07,
      "loss": 0.4608,
      "step": 415390
    },
    {
      "epoch": 170.04,
      "grad_norm": 2.0814924240112305,
      "learning_rate": 2.8180428418340267e-07,
      "loss": 0.4411,
      "step": 415400
    },
    {
      "epoch": 170.04,
      "grad_norm": 1.9547001123428345,
      "learning_rate": 2.817288745809108e-07,
      "loss": 0.45,
      "step": 415410
    },
    {
      "epoch": 170.05,
      "grad_norm": 2.150139570236206,
      "learning_rate": 2.816534744242186e-07,
      "loss": 0.4635,
      "step": 415420
    },
    {
      "epoch": 170.05,
      "grad_norm": 1.760870337486267,
      "learning_rate": 2.81578083713671e-07,
      "loss": 0.4376,
      "step": 415430
    },
    {
      "epoch": 170.05,
      "grad_norm": 2.125114917755127,
      "learning_rate": 2.815027024496136e-07,
      "loss": 0.4429,
      "step": 415440
    },
    {
      "epoch": 170.06,
      "grad_norm": 2.0439484119415283,
      "learning_rate": 2.8142733063239173e-07,
      "loss": 0.451,
      "step": 415450
    },
    {
      "epoch": 170.06,
      "grad_norm": 2.1985912322998047,
      "learning_rate": 2.813519682623507e-07,
      "loss": 0.4463,
      "step": 415460
    },
    {
      "epoch": 170.07,
      "grad_norm": 2.1429247856140137,
      "learning_rate": 2.8127661533983543e-07,
      "loss": 0.4389,
      "step": 415470
    },
    {
      "epoch": 170.07,
      "grad_norm": 2.196683645248413,
      "learning_rate": 2.8120127186519154e-07,
      "loss": 0.4648,
      "step": 415480
    },
    {
      "epoch": 170.07,
      "grad_norm": 2.510767698287964,
      "learning_rate": 2.8112593783876385e-07,
      "loss": 0.4309,
      "step": 415490
    },
    {
      "epoch": 170.08,
      "grad_norm": 2.1845247745513916,
      "learning_rate": 2.8105061326089763e-07,
      "loss": 0.4392,
      "step": 415500
    },
    {
      "epoch": 170.08,
      "grad_norm": 1.7963612079620361,
      "learning_rate": 2.809752981319382e-07,
      "loss": 0.4364,
      "step": 415510
    },
    {
      "epoch": 170.09,
      "grad_norm": 2.4263699054718018,
      "learning_rate": 2.808999924522292e-07,
      "loss": 0.4463,
      "step": 415520
    },
    {
      "epoch": 170.09,
      "grad_norm": 2.150101661682129,
      "learning_rate": 2.8082469622211733e-07,
      "loss": 0.4241,
      "step": 415530
    },
    {
      "epoch": 170.09,
      "grad_norm": 1.6229097843170166,
      "learning_rate": 2.8074940944194685e-07,
      "loss": 0.4312,
      "step": 415540
    },
    {
      "epoch": 170.1,
      "grad_norm": 2.3407208919525146,
      "learning_rate": 2.806741321120622e-07,
      "loss": 0.4514,
      "step": 415550
    },
    {
      "epoch": 170.1,
      "grad_norm": 2.1058764457702637,
      "learning_rate": 2.8059886423280893e-07,
      "loss": 0.4494,
      "step": 415560
    },
    {
      "epoch": 170.11,
      "grad_norm": 1.9642962217330933,
      "learning_rate": 2.805236058045314e-07,
      "loss": 0.4468,
      "step": 415570
    },
    {
      "epoch": 170.11,
      "grad_norm": 2.426283597946167,
      "learning_rate": 2.804483568275743e-07,
      "loss": 0.4484,
      "step": 415580
    },
    {
      "epoch": 170.11,
      "grad_norm": 2.2021536827087402,
      "learning_rate": 2.803731173022827e-07,
      "loss": 0.4652,
      "step": 415590
    },
    {
      "epoch": 170.12,
      "grad_norm": 2.2625420093536377,
      "learning_rate": 2.8029788722900086e-07,
      "loss": 0.4358,
      "step": 415600
    },
    {
      "epoch": 170.12,
      "grad_norm": 1.9368551969528198,
      "learning_rate": 2.8022266660807377e-07,
      "loss": 0.4492,
      "step": 415610
    },
    {
      "epoch": 170.13,
      "grad_norm": 1.7663389444351196,
      "learning_rate": 2.8014745543984574e-07,
      "loss": 0.4501,
      "step": 415620
    },
    {
      "epoch": 170.13,
      "grad_norm": 2.0506157875061035,
      "learning_rate": 2.8007225372466126e-07,
      "loss": 0.4497,
      "step": 415630
    },
    {
      "epoch": 170.14,
      "grad_norm": 2.1301538944244385,
      "learning_rate": 2.7999706146286565e-07,
      "loss": 0.4632,
      "step": 415640
    },
    {
      "epoch": 170.14,
      "grad_norm": 2.193988800048828,
      "learning_rate": 2.7992187865480185e-07,
      "loss": 0.4624,
      "step": 415650
    },
    {
      "epoch": 170.14,
      "grad_norm": 2.1471571922302246,
      "learning_rate": 2.798467053008154e-07,
      "loss": 0.4539,
      "step": 415660
    },
    {
      "epoch": 170.15,
      "grad_norm": 1.884702444076538,
      "learning_rate": 2.7977154140125004e-07,
      "loss": 0.4109,
      "step": 415670
    },
    {
      "epoch": 170.15,
      "grad_norm": 2.215390205383301,
      "learning_rate": 2.7969638695645056e-07,
      "loss": 0.4322,
      "step": 415680
    },
    {
      "epoch": 170.16,
      "grad_norm": 2.176178216934204,
      "learning_rate": 2.796212419667607e-07,
      "loss": 0.444,
      "step": 415690
    },
    {
      "epoch": 170.16,
      "grad_norm": 1.9856822490692139,
      "learning_rate": 2.7954610643252525e-07,
      "loss": 0.4652,
      "step": 415700
    },
    {
      "epoch": 170.16,
      "grad_norm": 2.012359619140625,
      "learning_rate": 2.7947098035408763e-07,
      "loss": 0.4399,
      "step": 415710
    },
    {
      "epoch": 170.17,
      "grad_norm": 2.051161050796509,
      "learning_rate": 2.7939586373179316e-07,
      "loss": 0.4407,
      "step": 415720
    },
    {
      "epoch": 170.17,
      "grad_norm": 2.064366102218628,
      "learning_rate": 2.793207565659851e-07,
      "loss": 0.4569,
      "step": 415730
    },
    {
      "epoch": 170.18,
      "grad_norm": 1.8976186513900757,
      "learning_rate": 2.7924565885700765e-07,
      "loss": 0.4346,
      "step": 415740
    },
    {
      "epoch": 170.18,
      "grad_norm": 1.9944325685501099,
      "learning_rate": 2.7917057060520486e-07,
      "loss": 0.4209,
      "step": 415750
    },
    {
      "epoch": 170.18,
      "grad_norm": 1.7944384813308716,
      "learning_rate": 2.790954918109209e-07,
      "loss": 0.4414,
      "step": 415760
    },
    {
      "epoch": 170.19,
      "grad_norm": 1.8877965211868286,
      "learning_rate": 2.7902042247449935e-07,
      "loss": 0.4296,
      "step": 415770
    },
    {
      "epoch": 170.19,
      "grad_norm": 1.9320478439331055,
      "learning_rate": 2.789453625962846e-07,
      "loss": 0.4262,
      "step": 415780
    },
    {
      "epoch": 170.2,
      "grad_norm": 2.4197161197662354,
      "learning_rate": 2.7887031217661974e-07,
      "loss": 0.4398,
      "step": 415790
    },
    {
      "epoch": 170.2,
      "grad_norm": 2.0447585582733154,
      "learning_rate": 2.7879527121584896e-07,
      "loss": 0.4721,
      "step": 415800
    },
    {
      "epoch": 170.2,
      "grad_norm": 2.1787095069885254,
      "learning_rate": 2.78720239714316e-07,
      "loss": 0.4526,
      "step": 415810
    },
    {
      "epoch": 170.21,
      "grad_norm": 1.8447895050048828,
      "learning_rate": 2.786452176723646e-07,
      "loss": 0.4538,
      "step": 415820
    },
    {
      "epoch": 170.21,
      "grad_norm": 2.565500020980835,
      "learning_rate": 2.785702050903382e-07,
      "loss": 0.4527,
      "step": 415830
    },
    {
      "epoch": 170.22,
      "grad_norm": 1.8816086053848267,
      "learning_rate": 2.7849520196858087e-07,
      "loss": 0.4677,
      "step": 415840
    },
    {
      "epoch": 170.22,
      "grad_norm": 1.9527218341827393,
      "learning_rate": 2.7842020830743573e-07,
      "loss": 0.4467,
      "step": 415850
    },
    {
      "epoch": 170.23,
      "grad_norm": 2.161405563354492,
      "learning_rate": 2.783452241072466e-07,
      "loss": 0.4373,
      "step": 415860
    },
    {
      "epoch": 170.23,
      "grad_norm": 2.3441314697265625,
      "learning_rate": 2.7827024936835665e-07,
      "loss": 0.4367,
      "step": 415870
    },
    {
      "epoch": 170.23,
      "grad_norm": 2.2233328819274902,
      "learning_rate": 2.781952840911099e-07,
      "loss": 0.421,
      "step": 415880
    },
    {
      "epoch": 170.24,
      "grad_norm": 1.6596566438674927,
      "learning_rate": 2.781203282758493e-07,
      "loss": 0.4405,
      "step": 415890
    },
    {
      "epoch": 170.24,
      "grad_norm": 2.0242114067077637,
      "learning_rate": 2.780453819229178e-07,
      "loss": 0.4529,
      "step": 415900
    },
    {
      "epoch": 170.25,
      "grad_norm": 1.844277024269104,
      "learning_rate": 2.779704450326596e-07,
      "loss": 0.4358,
      "step": 415910
    },
    {
      "epoch": 170.25,
      "grad_norm": 2.427048444747925,
      "learning_rate": 2.778955176054183e-07,
      "loss": 0.4247,
      "step": 415920
    },
    {
      "epoch": 170.25,
      "grad_norm": 2.0852041244506836,
      "learning_rate": 2.7782059964153596e-07,
      "loss": 0.4153,
      "step": 415930
    },
    {
      "epoch": 170.26,
      "grad_norm": 2.286609172821045,
      "learning_rate": 2.777456911413561e-07,
      "loss": 0.4408,
      "step": 415940
    },
    {
      "epoch": 170.26,
      "grad_norm": 2.352700710296631,
      "learning_rate": 2.776707921052222e-07,
      "loss": 0.4463,
      "step": 415950
    },
    {
      "epoch": 170.27,
      "grad_norm": 2.03670334815979,
      "learning_rate": 2.775959025334769e-07,
      "loss": 0.43,
      "step": 415960
    },
    {
      "epoch": 170.27,
      "grad_norm": 2.0246856212615967,
      "learning_rate": 2.775210224264638e-07,
      "loss": 0.4329,
      "step": 415970
    },
    {
      "epoch": 170.27,
      "grad_norm": 2.2335145473480225,
      "learning_rate": 2.774461517845255e-07,
      "loss": 0.4555,
      "step": 415980
    },
    {
      "epoch": 170.28,
      "grad_norm": 2.595695734024048,
      "learning_rate": 2.773712906080052e-07,
      "loss": 0.4458,
      "step": 415990
    },
    {
      "epoch": 170.28,
      "grad_norm": 2.4032909870147705,
      "learning_rate": 2.7729643889724567e-07,
      "loss": 0.432,
      "step": 416000
    },
    {
      "epoch": 170.29,
      "grad_norm": 1.8791770935058594,
      "learning_rate": 2.7722159665259003e-07,
      "loss": 0.4337,
      "step": 416010
    },
    {
      "epoch": 170.29,
      "grad_norm": 1.6795814037322998,
      "learning_rate": 2.771467638743808e-07,
      "loss": 0.4418,
      "step": 416020
    },
    {
      "epoch": 170.29,
      "grad_norm": 2.3189690113067627,
      "learning_rate": 2.7707194056296115e-07,
      "loss": 0.4685,
      "step": 416030
    },
    {
      "epoch": 170.3,
      "grad_norm": 2.3083407878875732,
      "learning_rate": 2.769971267186736e-07,
      "loss": 0.4433,
      "step": 416040
    },
    {
      "epoch": 170.3,
      "grad_norm": 1.990733027458191,
      "learning_rate": 2.7692232234186067e-07,
      "loss": 0.4203,
      "step": 416050
    },
    {
      "epoch": 170.31,
      "grad_norm": 2.129709482192993,
      "learning_rate": 2.768475274328652e-07,
      "loss": 0.4408,
      "step": 416060
    },
    {
      "epoch": 170.31,
      "grad_norm": 2.304821252822876,
      "learning_rate": 2.7677274199203006e-07,
      "loss": 0.4337,
      "step": 416070
    },
    {
      "epoch": 170.32,
      "grad_norm": 1.9812310934066772,
      "learning_rate": 2.766979660196975e-07,
      "loss": 0.4383,
      "step": 416080
    },
    {
      "epoch": 170.32,
      "grad_norm": 3.700995445251465,
      "learning_rate": 2.766231995162104e-07,
      "loss": 0.4465,
      "step": 416090
    },
    {
      "epoch": 170.32,
      "grad_norm": 2.1250903606414795,
      "learning_rate": 2.765484424819108e-07,
      "loss": 0.4477,
      "step": 416100
    },
    {
      "epoch": 170.33,
      "grad_norm": 2.511646270751953,
      "learning_rate": 2.764736949171415e-07,
      "loss": 0.441,
      "step": 416110
    },
    {
      "epoch": 170.33,
      "grad_norm": 2.0506114959716797,
      "learning_rate": 2.763989568222448e-07,
      "loss": 0.4508,
      "step": 416120
    },
    {
      "epoch": 170.34,
      "grad_norm": 1.8603025674819946,
      "learning_rate": 2.76324228197563e-07,
      "loss": 0.4527,
      "step": 416130
    },
    {
      "epoch": 170.34,
      "grad_norm": 2.3779897689819336,
      "learning_rate": 2.762495090434387e-07,
      "loss": 0.441,
      "step": 416140
    },
    {
      "epoch": 170.34,
      "grad_norm": 1.9148322343826294,
      "learning_rate": 2.7617479936021363e-07,
      "loss": 0.4372,
      "step": 416150
    },
    {
      "epoch": 170.35,
      "grad_norm": 2.170377731323242,
      "learning_rate": 2.7610009914823064e-07,
      "loss": 0.4261,
      "step": 416160
    },
    {
      "epoch": 170.35,
      "grad_norm": 2.4673852920532227,
      "learning_rate": 2.7602540840783126e-07,
      "loss": 0.443,
      "step": 416170
    },
    {
      "epoch": 170.36,
      "grad_norm": 2.049654722213745,
      "learning_rate": 2.7595072713935833e-07,
      "loss": 0.453,
      "step": 416180
    },
    {
      "epoch": 170.36,
      "grad_norm": 1.6749523878097534,
      "learning_rate": 2.7587605534315363e-07,
      "loss": 0.4571,
      "step": 416190
    },
    {
      "epoch": 170.36,
      "grad_norm": 2.1378087997436523,
      "learning_rate": 2.758013930195589e-07,
      "loss": 0.4614,
      "step": 416200
    },
    {
      "epoch": 170.37,
      "grad_norm": 2.231612205505371,
      "learning_rate": 2.7572674016891677e-07,
      "loss": 0.4462,
      "step": 416210
    },
    {
      "epoch": 170.37,
      "grad_norm": 1.8928033113479614,
      "learning_rate": 2.7565209679156896e-07,
      "loss": 0.43,
      "step": 416220
    },
    {
      "epoch": 170.38,
      "grad_norm": 1.8212214708328247,
      "learning_rate": 2.7557746288785754e-07,
      "loss": 0.414,
      "step": 416230
    },
    {
      "epoch": 170.38,
      "grad_norm": 1.6571983098983765,
      "learning_rate": 2.7550283845812404e-07,
      "loss": 0.4445,
      "step": 416240
    },
    {
      "epoch": 170.38,
      "grad_norm": 5.1645827293396,
      "learning_rate": 2.754282235027103e-07,
      "loss": 0.4522,
      "step": 416250
    },
    {
      "epoch": 170.39,
      "grad_norm": 2.09728741645813,
      "learning_rate": 2.7535361802195846e-07,
      "loss": 0.4363,
      "step": 416260
    },
    {
      "epoch": 170.39,
      "grad_norm": 2.7671682834625244,
      "learning_rate": 2.7527902201620947e-07,
      "loss": 0.4556,
      "step": 416270
    },
    {
      "epoch": 170.4,
      "grad_norm": 2.1581568717956543,
      "learning_rate": 2.7520443548580624e-07,
      "loss": 0.438,
      "step": 416280
    },
    {
      "epoch": 170.4,
      "grad_norm": 1.9395920038223267,
      "learning_rate": 2.751298584310899e-07,
      "loss": 0.4548,
      "step": 416290
    },
    {
      "epoch": 170.41,
      "grad_norm": 2.23067569732666,
      "learning_rate": 2.7505529085240186e-07,
      "loss": 0.4411,
      "step": 416300
    },
    {
      "epoch": 170.41,
      "grad_norm": 2.0182723999023438,
      "learning_rate": 2.749807327500843e-07,
      "loss": 0.4316,
      "step": 416310
    },
    {
      "epoch": 170.41,
      "grad_norm": 1.9878355264663696,
      "learning_rate": 2.749061841244781e-07,
      "loss": 0.4313,
      "step": 416320
    },
    {
      "epoch": 170.42,
      "grad_norm": 1.9713952541351318,
      "learning_rate": 2.748316449759251e-07,
      "loss": 0.439,
      "step": 416330
    },
    {
      "epoch": 170.42,
      "grad_norm": 2.0454719066619873,
      "learning_rate": 2.7475711530476655e-07,
      "loss": 0.4613,
      "step": 416340
    },
    {
      "epoch": 170.43,
      "grad_norm": 2.159064769744873,
      "learning_rate": 2.74682595111344e-07,
      "loss": 0.4334,
      "step": 416350
    },
    {
      "epoch": 170.43,
      "grad_norm": 2.200479507446289,
      "learning_rate": 2.746080843959994e-07,
      "loss": 0.4607,
      "step": 416360
    },
    {
      "epoch": 170.43,
      "grad_norm": 1.9970098733901978,
      "learning_rate": 2.745335831590729e-07,
      "loss": 0.4339,
      "step": 416370
    },
    {
      "epoch": 170.44,
      "grad_norm": 1.866626262664795,
      "learning_rate": 2.7445909140090633e-07,
      "loss": 0.417,
      "step": 416380
    },
    {
      "epoch": 170.44,
      "grad_norm": 1.8615633249282837,
      "learning_rate": 2.743846091218411e-07,
      "loss": 0.4187,
      "step": 416390
    },
    {
      "epoch": 170.45,
      "grad_norm": 2.596735715866089,
      "learning_rate": 2.7431013632221815e-07,
      "loss": 0.4574,
      "step": 416400
    },
    {
      "epoch": 170.45,
      "grad_norm": 1.8794662952423096,
      "learning_rate": 2.742356730023788e-07,
      "loss": 0.4517,
      "step": 416410
    },
    {
      "epoch": 170.45,
      "grad_norm": 1.9438369274139404,
      "learning_rate": 2.74161219162664e-07,
      "loss": 0.4525,
      "step": 416420
    },
    {
      "epoch": 170.46,
      "grad_norm": 2.2680468559265137,
      "learning_rate": 2.7408677480341476e-07,
      "loss": 0.4294,
      "step": 416430
    },
    {
      "epoch": 170.46,
      "grad_norm": 2.0962491035461426,
      "learning_rate": 2.740123399249723e-07,
      "loss": 0.4452,
      "step": 416440
    },
    {
      "epoch": 170.47,
      "grad_norm": 1.9778950214385986,
      "learning_rate": 2.7393791452767774e-07,
      "loss": 0.4566,
      "step": 416450
    },
    {
      "epoch": 170.47,
      "grad_norm": 1.9604185819625854,
      "learning_rate": 2.738634986118709e-07,
      "loss": 0.4293,
      "step": 416460
    },
    {
      "epoch": 170.47,
      "grad_norm": 2.202528238296509,
      "learning_rate": 2.7378909217789424e-07,
      "loss": 0.4447,
      "step": 416470
    },
    {
      "epoch": 170.48,
      "grad_norm": 2.088409185409546,
      "learning_rate": 2.7371469522608815e-07,
      "loss": 0.4316,
      "step": 416480
    },
    {
      "epoch": 170.48,
      "grad_norm": 2.1871299743652344,
      "learning_rate": 2.7364030775679287e-07,
      "loss": 0.4467,
      "step": 416490
    },
    {
      "epoch": 170.49,
      "grad_norm": 2.110629081726074,
      "learning_rate": 2.7356592977035035e-07,
      "loss": 0.4401,
      "step": 416500
    },
    {
      "epoch": 170.49,
      "grad_norm": 2.090489387512207,
      "learning_rate": 2.734915612670997e-07,
      "loss": 0.4444,
      "step": 416510
    },
    {
      "epoch": 170.5,
      "grad_norm": 1.8047330379486084,
      "learning_rate": 2.734172022473823e-07,
      "loss": 0.4315,
      "step": 416520
    },
    {
      "epoch": 170.5,
      "grad_norm": 2.38887619972229,
      "learning_rate": 2.7334285271153897e-07,
      "loss": 0.4527,
      "step": 416530
    },
    {
      "epoch": 170.5,
      "grad_norm": 2.2026445865631104,
      "learning_rate": 2.732685126599101e-07,
      "loss": 0.4475,
      "step": 416540
    },
    {
      "epoch": 170.51,
      "grad_norm": 2.1481645107269287,
      "learning_rate": 2.7319418209283603e-07,
      "loss": 0.4515,
      "step": 416550
    },
    {
      "epoch": 170.51,
      "grad_norm": 1.973063349723816,
      "learning_rate": 2.7311986101065766e-07,
      "loss": 0.4524,
      "step": 416560
    },
    {
      "epoch": 170.52,
      "grad_norm": 1.7500183582305908,
      "learning_rate": 2.730455494137153e-07,
      "loss": 0.4414,
      "step": 416570
    },
    {
      "epoch": 170.52,
      "grad_norm": 2.090817928314209,
      "learning_rate": 2.729712473023494e-07,
      "loss": 0.4345,
      "step": 416580
    },
    {
      "epoch": 170.52,
      "grad_norm": 2.0420432090759277,
      "learning_rate": 2.7289695467690043e-07,
      "loss": 0.4324,
      "step": 416590
    },
    {
      "epoch": 170.53,
      "grad_norm": 1.489446997642517,
      "learning_rate": 2.728226715377083e-07,
      "loss": 0.4404,
      "step": 416600
    },
    {
      "epoch": 170.53,
      "grad_norm": 1.9272664785385132,
      "learning_rate": 2.727483978851134e-07,
      "loss": 0.4451,
      "step": 416610
    },
    {
      "epoch": 170.54,
      "grad_norm": 2.155705451965332,
      "learning_rate": 2.726741337194565e-07,
      "loss": 0.4354,
      "step": 416620
    },
    {
      "epoch": 170.54,
      "grad_norm": 1.872678279876709,
      "learning_rate": 2.72599879041077e-07,
      "loss": 0.4473,
      "step": 416630
    },
    {
      "epoch": 170.54,
      "grad_norm": 1.966636300086975,
      "learning_rate": 2.7252563385031574e-07,
      "loss": 0.4493,
      "step": 416640
    },
    {
      "epoch": 170.55,
      "grad_norm": 1.9692912101745605,
      "learning_rate": 2.724513981475123e-07,
      "loss": 0.4442,
      "step": 416650
    },
    {
      "epoch": 170.55,
      "grad_norm": 2.1380562782287598,
      "learning_rate": 2.7237717193300723e-07,
      "loss": 0.4486,
      "step": 416660
    },
    {
      "epoch": 170.56,
      "grad_norm": 1.9706416130065918,
      "learning_rate": 2.7230295520714e-07,
      "loss": 0.4463,
      "step": 416670
    },
    {
      "epoch": 170.56,
      "grad_norm": 1.8219267129898071,
      "learning_rate": 2.7222874797025105e-07,
      "loss": 0.4535,
      "step": 416680
    },
    {
      "epoch": 170.56,
      "grad_norm": 2.019268751144409,
      "learning_rate": 2.721545502226803e-07,
      "loss": 0.4493,
      "step": 416690
    },
    {
      "epoch": 170.57,
      "grad_norm": 2.1692023277282715,
      "learning_rate": 2.7208036196476706e-07,
      "loss": 0.4533,
      "step": 416700
    },
    {
      "epoch": 170.57,
      "grad_norm": 1.9814716577529907,
      "learning_rate": 2.720061831968519e-07,
      "loss": 0.4375,
      "step": 416710
    },
    {
      "epoch": 170.58,
      "grad_norm": 2.254791021347046,
      "learning_rate": 2.719320139192742e-07,
      "loss": 0.4509,
      "step": 416720
    },
    {
      "epoch": 170.58,
      "grad_norm": 1.958322286605835,
      "learning_rate": 2.71857854132374e-07,
      "loss": 0.4482,
      "step": 416730
    },
    {
      "epoch": 170.59,
      "grad_norm": 2.1846346855163574,
      "learning_rate": 2.717837038364906e-07,
      "loss": 0.4392,
      "step": 416740
    },
    {
      "epoch": 170.59,
      "grad_norm": 1.9369758367538452,
      "learning_rate": 2.7170956303196416e-07,
      "loss": 0.4347,
      "step": 416750
    },
    {
      "epoch": 170.59,
      "grad_norm": 2.3470077514648438,
      "learning_rate": 2.7163543171913395e-07,
      "loss": 0.4396,
      "step": 416760
    },
    {
      "epoch": 170.6,
      "grad_norm": 1.8800101280212402,
      "learning_rate": 2.7156130989833955e-07,
      "loss": 0.4629,
      "step": 416770
    },
    {
      "epoch": 170.6,
      "grad_norm": 2.145393133163452,
      "learning_rate": 2.7148719756992077e-07,
      "loss": 0.4451,
      "step": 416780
    },
    {
      "epoch": 170.61,
      "grad_norm": 1.831294298171997,
      "learning_rate": 2.714130947342169e-07,
      "loss": 0.4187,
      "step": 416790
    },
    {
      "epoch": 170.61,
      "grad_norm": 1.8066402673721313,
      "learning_rate": 2.7133900139156763e-07,
      "loss": 0.4364,
      "step": 416800
    },
    {
      "epoch": 170.61,
      "grad_norm": 2.000978469848633,
      "learning_rate": 2.7126491754231237e-07,
      "loss": 0.4675,
      "step": 416810
    },
    {
      "epoch": 170.62,
      "grad_norm": 2.0146284103393555,
      "learning_rate": 2.7119084318679006e-07,
      "loss": 0.4193,
      "step": 416820
    },
    {
      "epoch": 170.62,
      "grad_norm": 2.3343708515167236,
      "learning_rate": 2.7111677832534013e-07,
      "loss": 0.427,
      "step": 416830
    },
    {
      "epoch": 170.63,
      "grad_norm": 2.1453514099121094,
      "learning_rate": 2.710427229583018e-07,
      "loss": 0.4424,
      "step": 416840
    },
    {
      "epoch": 170.63,
      "grad_norm": 2.119873523712158,
      "learning_rate": 2.7096867708601473e-07,
      "loss": 0.4413,
      "step": 416850
    },
    {
      "epoch": 170.63,
      "grad_norm": 2.001612901687622,
      "learning_rate": 2.708946407088178e-07,
      "loss": 0.4441,
      "step": 416860
    },
    {
      "epoch": 170.64,
      "grad_norm": 1.8944746255874634,
      "learning_rate": 2.7082061382705037e-07,
      "loss": 0.4282,
      "step": 416870
    },
    {
      "epoch": 170.64,
      "grad_norm": 1.9174916744232178,
      "learning_rate": 2.707465964410514e-07,
      "loss": 0.4434,
      "step": 416880
    },
    {
      "epoch": 170.65,
      "grad_norm": 2.664391040802002,
      "learning_rate": 2.706725885511602e-07,
      "loss": 0.4609,
      "step": 416890
    },
    {
      "epoch": 170.65,
      "grad_norm": 2.6741435527801514,
      "learning_rate": 2.705985901577154e-07,
      "loss": 0.4445,
      "step": 416900
    },
    {
      "epoch": 170.65,
      "grad_norm": 1.9136768579483032,
      "learning_rate": 2.7052460126105596e-07,
      "loss": 0.432,
      "step": 416910
    },
    {
      "epoch": 170.66,
      "grad_norm": 2.0256412029266357,
      "learning_rate": 2.704506218615212e-07,
      "loss": 0.4556,
      "step": 416920
    },
    {
      "epoch": 170.66,
      "grad_norm": 1.9517632722854614,
      "learning_rate": 2.703766519594497e-07,
      "loss": 0.4518,
      "step": 416930
    },
    {
      "epoch": 170.67,
      "grad_norm": 2.6472177505493164,
      "learning_rate": 2.703026915551805e-07,
      "loss": 0.4456,
      "step": 416940
    },
    {
      "epoch": 170.67,
      "grad_norm": 1.8459746837615967,
      "learning_rate": 2.7022874064905285e-07,
      "loss": 0.4277,
      "step": 416950
    },
    {
      "epoch": 170.68,
      "grad_norm": 2.050348997116089,
      "learning_rate": 2.7015479924140435e-07,
      "loss": 0.4526,
      "step": 416960
    },
    {
      "epoch": 170.68,
      "grad_norm": 2.06882643699646,
      "learning_rate": 2.700808673325742e-07,
      "loss": 0.4417,
      "step": 416970
    },
    {
      "epoch": 170.68,
      "grad_norm": 1.6792354583740234,
      "learning_rate": 2.7000694492290153e-07,
      "loss": 0.4518,
      "step": 416980
    },
    {
      "epoch": 170.69,
      "grad_norm": 1.8777583837509155,
      "learning_rate": 2.699330320127243e-07,
      "loss": 0.4435,
      "step": 416990
    },
    {
      "epoch": 170.69,
      "grad_norm": 1.8428077697753906,
      "learning_rate": 2.698591286023816e-07,
      "loss": 0.429,
      "step": 417000
    },
    {
      "epoch": 170.7,
      "grad_norm": 1.9212040901184082,
      "learning_rate": 2.697852346922119e-07,
      "loss": 0.4344,
      "step": 417010
    },
    {
      "epoch": 170.7,
      "grad_norm": 2.021193504333496,
      "learning_rate": 2.6971135028255354e-07,
      "loss": 0.438,
      "step": 417020
    },
    {
      "epoch": 170.7,
      "grad_norm": 1.7454025745391846,
      "learning_rate": 2.6963747537374426e-07,
      "loss": 0.4434,
      "step": 417030
    },
    {
      "epoch": 170.71,
      "grad_norm": 2.513356924057007,
      "learning_rate": 2.6956360996612414e-07,
      "loss": 0.4452,
      "step": 417040
    },
    {
      "epoch": 170.71,
      "grad_norm": 1.8999266624450684,
      "learning_rate": 2.6948975406003036e-07,
      "loss": 0.4511,
      "step": 417050
    },
    {
      "epoch": 170.72,
      "grad_norm": 2.0754849910736084,
      "learning_rate": 2.6941590765580156e-07,
      "loss": 0.4425,
      "step": 417060
    },
    {
      "epoch": 170.72,
      "grad_norm": 2.2130651473999023,
      "learning_rate": 2.693420707537762e-07,
      "loss": 0.4486,
      "step": 417070
    },
    {
      "epoch": 170.72,
      "grad_norm": 2.1110081672668457,
      "learning_rate": 2.6926824335429255e-07,
      "loss": 0.4446,
      "step": 417080
    },
    {
      "epoch": 170.73,
      "grad_norm": 1.895382285118103,
      "learning_rate": 2.691944254576881e-07,
      "loss": 0.428,
      "step": 417090
    },
    {
      "epoch": 170.73,
      "grad_norm": 2.084282636642456,
      "learning_rate": 2.691206170643017e-07,
      "loss": 0.4474,
      "step": 417100
    },
    {
      "epoch": 170.74,
      "grad_norm": 2.2255804538726807,
      "learning_rate": 2.69046818174471e-07,
      "loss": 0.4396,
      "step": 417110
    },
    {
      "epoch": 170.74,
      "grad_norm": 1.7357226610183716,
      "learning_rate": 2.689730287885341e-07,
      "loss": 0.4533,
      "step": 417120
    },
    {
      "epoch": 170.74,
      "grad_norm": 2.2682430744171143,
      "learning_rate": 2.6889924890682945e-07,
      "loss": 0.4407,
      "step": 417130
    },
    {
      "epoch": 170.75,
      "grad_norm": 2.103682518005371,
      "learning_rate": 2.688254785296946e-07,
      "loss": 0.4497,
      "step": 417140
    },
    {
      "epoch": 170.75,
      "grad_norm": 2.1838130950927734,
      "learning_rate": 2.687517176574679e-07,
      "loss": 0.4339,
      "step": 417150
    },
    {
      "epoch": 170.76,
      "grad_norm": 2.0134103298187256,
      "learning_rate": 2.6867796629048664e-07,
      "loss": 0.4518,
      "step": 417160
    },
    {
      "epoch": 170.76,
      "grad_norm": 2.222109079360962,
      "learning_rate": 2.6860422442908924e-07,
      "loss": 0.4447,
      "step": 417170
    },
    {
      "epoch": 170.77,
      "grad_norm": 1.75339937210083,
      "learning_rate": 2.6853049207361317e-07,
      "loss": 0.45,
      "step": 417180
    },
    {
      "epoch": 170.77,
      "grad_norm": 1.7182400226593018,
      "learning_rate": 2.684567692243961e-07,
      "loss": 0.4786,
      "step": 417190
    },
    {
      "epoch": 170.77,
      "grad_norm": 2.5659995079040527,
      "learning_rate": 2.6838305588177616e-07,
      "loss": 0.4484,
      "step": 417200
    },
    {
      "epoch": 170.78,
      "grad_norm": 1.8788959980010986,
      "learning_rate": 2.683093520460908e-07,
      "loss": 0.4321,
      "step": 417210
    },
    {
      "epoch": 170.78,
      "grad_norm": 1.8441293239593506,
      "learning_rate": 2.6823565771767746e-07,
      "loss": 0.4626,
      "step": 417220
    },
    {
      "epoch": 170.79,
      "grad_norm": 2.2798757553100586,
      "learning_rate": 2.681619728968739e-07,
      "loss": 0.4614,
      "step": 417230
    },
    {
      "epoch": 170.79,
      "grad_norm": 1.845492959022522,
      "learning_rate": 2.680882975840179e-07,
      "loss": 0.4646,
      "step": 417240
    },
    {
      "epoch": 170.79,
      "grad_norm": 2.0290825366973877,
      "learning_rate": 2.6801463177944643e-07,
      "loss": 0.4562,
      "step": 417250
    },
    {
      "epoch": 170.8,
      "grad_norm": 1.9509270191192627,
      "learning_rate": 2.6794097548349725e-07,
      "loss": 0.4242,
      "step": 417260
    },
    {
      "epoch": 170.8,
      "grad_norm": 1.9881510734558105,
      "learning_rate": 2.6786732869650786e-07,
      "loss": 0.4377,
      "step": 417270
    },
    {
      "epoch": 170.81,
      "grad_norm": 2.179229974746704,
      "learning_rate": 2.6779369141881543e-07,
      "loss": 0.4394,
      "step": 417280
    },
    {
      "epoch": 170.81,
      "grad_norm": 2.3897311687469482,
      "learning_rate": 2.6772006365075724e-07,
      "loss": 0.4357,
      "step": 417290
    },
    {
      "epoch": 170.81,
      "grad_norm": 2.093688726425171,
      "learning_rate": 2.676464453926708e-07,
      "loss": 0.4409,
      "step": 417300
    },
    {
      "epoch": 170.82,
      "grad_norm": 2.1429104804992676,
      "learning_rate": 2.675728366448932e-07,
      "loss": 0.4343,
      "step": 417310
    },
    {
      "epoch": 170.82,
      "grad_norm": 1.7572630643844604,
      "learning_rate": 2.674992374077616e-07,
      "loss": 0.4464,
      "step": 417320
    },
    {
      "epoch": 170.83,
      "grad_norm": 2.2683074474334717,
      "learning_rate": 2.674256476816134e-07,
      "loss": 0.4492,
      "step": 417330
    },
    {
      "epoch": 170.83,
      "grad_norm": 2.2214722633361816,
      "learning_rate": 2.673520674667853e-07,
      "loss": 0.4242,
      "step": 417340
    },
    {
      "epoch": 170.84,
      "grad_norm": 2.272942543029785,
      "learning_rate": 2.6727849676361454e-07,
      "loss": 0.4472,
      "step": 417350
    },
    {
      "epoch": 170.84,
      "grad_norm": 2.2819082736968994,
      "learning_rate": 2.672049355724384e-07,
      "loss": 0.4263,
      "step": 417360
    },
    {
      "epoch": 170.84,
      "grad_norm": 1.9998801946640015,
      "learning_rate": 2.6713138389359326e-07,
      "loss": 0.4346,
      "step": 417370
    },
    {
      "epoch": 170.85,
      "grad_norm": 2.187713623046875,
      "learning_rate": 2.670578417274166e-07,
      "loss": 0.4331,
      "step": 417380
    },
    {
      "epoch": 170.85,
      "grad_norm": 1.9790706634521484,
      "learning_rate": 2.669843090742455e-07,
      "loss": 0.4421,
      "step": 417390
    },
    {
      "epoch": 170.86,
      "grad_norm": 2.0500974655151367,
      "learning_rate": 2.669107859344155e-07,
      "loss": 0.4708,
      "step": 417400
    },
    {
      "epoch": 170.86,
      "grad_norm": 2.0389404296875,
      "learning_rate": 2.668372723082647e-07,
      "loss": 0.4578,
      "step": 417410
    },
    {
      "epoch": 170.86,
      "grad_norm": 1.8411858081817627,
      "learning_rate": 2.6676376819612947e-07,
      "loss": 0.4583,
      "step": 417420
    },
    {
      "epoch": 170.87,
      "grad_norm": 1.8571938276290894,
      "learning_rate": 2.666902735983466e-07,
      "loss": 0.4454,
      "step": 417430
    },
    {
      "epoch": 170.87,
      "grad_norm": 2.1504385471343994,
      "learning_rate": 2.6661678851525256e-07,
      "loss": 0.441,
      "step": 417440
    },
    {
      "epoch": 170.88,
      "grad_norm": 2.0306379795074463,
      "learning_rate": 2.6654331294718416e-07,
      "loss": 0.4332,
      "step": 417450
    },
    {
      "epoch": 170.88,
      "grad_norm": 1.9551206827163696,
      "learning_rate": 2.66469846894478e-07,
      "loss": 0.4377,
      "step": 417460
    },
    {
      "epoch": 170.88,
      "grad_norm": 1.9823518991470337,
      "learning_rate": 2.663963903574703e-07,
      "loss": 0.4264,
      "step": 417470
    },
    {
      "epoch": 170.89,
      "grad_norm": 1.7984932661056519,
      "learning_rate": 2.6632294333649787e-07,
      "loss": 0.4632,
      "step": 417480
    },
    {
      "epoch": 170.89,
      "grad_norm": 2.2478225231170654,
      "learning_rate": 2.66249505831897e-07,
      "loss": 0.4642,
      "step": 417490
    },
    {
      "epoch": 170.9,
      "grad_norm": 1.6597907543182373,
      "learning_rate": 2.6617607784400414e-07,
      "loss": 0.4396,
      "step": 417500
    },
    {
      "epoch": 170.9,
      "grad_norm": 2.337890863418579,
      "learning_rate": 2.661026593731557e-07,
      "loss": 0.4344,
      "step": 417510
    },
    {
      "epoch": 170.9,
      "grad_norm": 2.2447474002838135,
      "learning_rate": 2.660292504196882e-07,
      "loss": 0.4328,
      "step": 417520
    },
    {
      "epoch": 170.91,
      "grad_norm": 2.0359816551208496,
      "learning_rate": 2.6595585098393773e-07,
      "loss": 0.4393,
      "step": 417530
    },
    {
      "epoch": 170.91,
      "grad_norm": 1.9562489986419678,
      "learning_rate": 2.658824610662403e-07,
      "loss": 0.4448,
      "step": 417540
    },
    {
      "epoch": 170.92,
      "grad_norm": 2.175319194793701,
      "learning_rate": 2.6580908066693206e-07,
      "loss": 0.4388,
      "step": 417550
    },
    {
      "epoch": 170.92,
      "grad_norm": 2.4680709838867188,
      "learning_rate": 2.657357097863495e-07,
      "loss": 0.451,
      "step": 417560
    },
    {
      "epoch": 170.93,
      "grad_norm": 1.8366425037384033,
      "learning_rate": 2.656623484248285e-07,
      "loss": 0.4577,
      "step": 417570
    },
    {
      "epoch": 170.93,
      "grad_norm": 2.2775607109069824,
      "learning_rate": 2.6558899658270525e-07,
      "loss": 0.4497,
      "step": 417580
    },
    {
      "epoch": 170.93,
      "grad_norm": 2.0870964527130127,
      "learning_rate": 2.655156542603152e-07,
      "loss": 0.4339,
      "step": 417590
    },
    {
      "epoch": 170.94,
      "grad_norm": 2.1394689083099365,
      "learning_rate": 2.6544232145799556e-07,
      "loss": 0.4586,
      "step": 417600
    },
    {
      "epoch": 170.94,
      "grad_norm": 2.2196872234344482,
      "learning_rate": 2.6536899817608123e-07,
      "loss": 0.4354,
      "step": 417610
    },
    {
      "epoch": 170.95,
      "grad_norm": 2.1223649978637695,
      "learning_rate": 2.6529568441490833e-07,
      "loss": 0.4483,
      "step": 417620
    },
    {
      "epoch": 170.95,
      "grad_norm": 2.593162775039673,
      "learning_rate": 2.652223801748129e-07,
      "loss": 0.4333,
      "step": 417630
    },
    {
      "epoch": 170.95,
      "grad_norm": 2.151017427444458,
      "learning_rate": 2.651490854561308e-07,
      "loss": 0.4636,
      "step": 417640
    },
    {
      "epoch": 170.96,
      "grad_norm": 1.9620109796524048,
      "learning_rate": 2.650758002591972e-07,
      "loss": 0.4525,
      "step": 417650
    },
    {
      "epoch": 170.96,
      "grad_norm": 2.092038631439209,
      "learning_rate": 2.650025245843488e-07,
      "loss": 0.4397,
      "step": 417660
    },
    {
      "epoch": 170.97,
      "grad_norm": 1.998793125152588,
      "learning_rate": 2.6492925843192024e-07,
      "loss": 0.4583,
      "step": 417670
    },
    {
      "epoch": 170.97,
      "grad_norm": 1.930747151374817,
      "learning_rate": 2.6485600180224767e-07,
      "loss": 0.4614,
      "step": 417680
    },
    {
      "epoch": 170.97,
      "grad_norm": 2.2627644538879395,
      "learning_rate": 2.647827546956663e-07,
      "loss": 0.45,
      "step": 417690
    },
    {
      "epoch": 170.98,
      "grad_norm": 5.989645957946777,
      "learning_rate": 2.64709517112512e-07,
      "loss": 0.4389,
      "step": 417700
    },
    {
      "epoch": 170.98,
      "grad_norm": 1.9898024797439575,
      "learning_rate": 2.6463628905312025e-07,
      "loss": 0.4401,
      "step": 417710
    },
    {
      "epoch": 170.99,
      "grad_norm": 2.083648443222046,
      "learning_rate": 2.645630705178264e-07,
      "loss": 0.4454,
      "step": 417720
    },
    {
      "epoch": 170.99,
      "grad_norm": 1.6505337953567505,
      "learning_rate": 2.644898615069659e-07,
      "loss": 0.4524,
      "step": 417730
    },
    {
      "epoch": 170.99,
      "grad_norm": 2.1297311782836914,
      "learning_rate": 2.644166620208738e-07,
      "loss": 0.4556,
      "step": 417740
    },
    {
      "epoch": 171.0,
      "grad_norm": 2.0244851112365723,
      "learning_rate": 2.643434720598859e-07,
      "loss": 0.4609,
      "step": 417750
    },
    {
      "epoch": 171.0,
      "eval_loss": 0.4455927312374115,
      "eval_runtime": 57.3531,
      "eval_samples_per_second": 60.136,
      "eval_steps_per_second": 7.532,
      "step": 417753
    },
    {
      "epoch": 171.0,
      "grad_norm": 2.5064871311187744,
      "learning_rate": 2.642702916243373e-07,
      "loss": 0.4312,
      "step": 417760
    },
    {
      "epoch": 171.01,
      "grad_norm": 1.9286441802978516,
      "learning_rate": 2.641971207145631e-07,
      "loss": 0.446,
      "step": 417770
    },
    {
      "epoch": 171.01,
      "grad_norm": 2.114126443862915,
      "learning_rate": 2.6412395933089826e-07,
      "loss": 0.4358,
      "step": 417780
    },
    {
      "epoch": 171.02,
      "grad_norm": 2.249366521835327,
      "learning_rate": 2.6405080747367867e-07,
      "loss": 0.4462,
      "step": 417790
    },
    {
      "epoch": 171.02,
      "grad_norm": 2.184458017349243,
      "learning_rate": 2.639776651432395e-07,
      "loss": 0.4525,
      "step": 417800
    },
    {
      "epoch": 171.02,
      "grad_norm": 1.8335732221603394,
      "learning_rate": 2.6390453233991454e-07,
      "loss": 0.4266,
      "step": 417810
    },
    {
      "epoch": 171.03,
      "grad_norm": 2.326301336288452,
      "learning_rate": 2.638314090640398e-07,
      "loss": 0.4602,
      "step": 417820
    },
    {
      "epoch": 171.03,
      "grad_norm": 2.297546625137329,
      "learning_rate": 2.637582953159501e-07,
      "loss": 0.4308,
      "step": 417830
    },
    {
      "epoch": 171.04,
      "grad_norm": 2.134082555770874,
      "learning_rate": 2.6368519109598014e-07,
      "loss": 0.4466,
      "step": 417840
    },
    {
      "epoch": 171.04,
      "grad_norm": 2.0155084133148193,
      "learning_rate": 2.63612096404465e-07,
      "loss": 0.4681,
      "step": 417850
    },
    {
      "epoch": 171.04,
      "grad_norm": 2.0034422874450684,
      "learning_rate": 2.6353901124173934e-07,
      "loss": 0.4315,
      "step": 417860
    },
    {
      "epoch": 171.05,
      "grad_norm": 1.950432300567627,
      "learning_rate": 2.6346593560813804e-07,
      "loss": 0.4388,
      "step": 417870
    },
    {
      "epoch": 171.05,
      "grad_norm": 2.5315957069396973,
      "learning_rate": 2.63392869503996e-07,
      "loss": 0.4467,
      "step": 417880
    },
    {
      "epoch": 171.06,
      "grad_norm": 1.8793610334396362,
      "learning_rate": 2.6331981292964784e-07,
      "loss": 0.4266,
      "step": 417890
    },
    {
      "epoch": 171.06,
      "grad_norm": 1.9083689451217651,
      "learning_rate": 2.6324676588542794e-07,
      "loss": 0.4672,
      "step": 417900
    },
    {
      "epoch": 171.06,
      "grad_norm": 2.2738196849823,
      "learning_rate": 2.631737283716714e-07,
      "loss": 0.455,
      "step": 417910
    },
    {
      "epoch": 171.07,
      "grad_norm": 2.1661417484283447,
      "learning_rate": 2.6310070038871244e-07,
      "loss": 0.4466,
      "step": 417920
    },
    {
      "epoch": 171.07,
      "grad_norm": 1.9145698547363281,
      "learning_rate": 2.6302768193688557e-07,
      "loss": 0.4604,
      "step": 417930
    },
    {
      "epoch": 171.08,
      "grad_norm": 2.094287395477295,
      "learning_rate": 2.629546730165254e-07,
      "loss": 0.4389,
      "step": 417940
    },
    {
      "epoch": 171.08,
      "grad_norm": 2.051621437072754,
      "learning_rate": 2.6288167362796674e-07,
      "loss": 0.4477,
      "step": 417950
    },
    {
      "epoch": 171.08,
      "grad_norm": 2.1914734840393066,
      "learning_rate": 2.6280868377154325e-07,
      "loss": 0.4348,
      "step": 417960
    },
    {
      "epoch": 171.09,
      "grad_norm": 1.9332128763198853,
      "learning_rate": 2.6273570344758964e-07,
      "loss": 0.447,
      "step": 417970
    },
    {
      "epoch": 171.09,
      "grad_norm": 2.0643601417541504,
      "learning_rate": 2.626627326564405e-07,
      "loss": 0.4406,
      "step": 417980
    },
    {
      "epoch": 171.1,
      "grad_norm": 2.2452564239501953,
      "learning_rate": 2.6258977139842973e-07,
      "loss": 0.4535,
      "step": 417990
    },
    {
      "epoch": 171.1,
      "grad_norm": 2.00982928276062,
      "learning_rate": 2.6251681967389163e-07,
      "loss": 0.4455,
      "step": 418000
    },
    {
      "epoch": 171.11,
      "grad_norm": 2.3726747035980225,
      "learning_rate": 2.624438774831604e-07,
      "loss": 0.4566,
      "step": 418010
    },
    {
      "epoch": 171.11,
      "grad_norm": 2.063774347305298,
      "learning_rate": 2.6237094482657003e-07,
      "loss": 0.4445,
      "step": 418020
    },
    {
      "epoch": 171.11,
      "grad_norm": 2.2513225078582764,
      "learning_rate": 2.62298021704455e-07,
      "loss": 0.434,
      "step": 418030
    },
    {
      "epoch": 171.12,
      "grad_norm": 2.0251882076263428,
      "learning_rate": 2.6222510811714913e-07,
      "loss": 0.4404,
      "step": 418040
    },
    {
      "epoch": 171.12,
      "grad_norm": 2.0112476348876953,
      "learning_rate": 2.621522040649862e-07,
      "loss": 0.439,
      "step": 418050
    },
    {
      "epoch": 171.13,
      "grad_norm": 2.325265884399414,
      "learning_rate": 2.6207930954830064e-07,
      "loss": 0.448,
      "step": 418060
    },
    {
      "epoch": 171.13,
      "grad_norm": 2.0802156925201416,
      "learning_rate": 2.6200642456742605e-07,
      "loss": 0.4434,
      "step": 418070
    },
    {
      "epoch": 171.13,
      "grad_norm": 1.9017815589904785,
      "learning_rate": 2.619335491226962e-07,
      "loss": 0.4322,
      "step": 418080
    },
    {
      "epoch": 171.14,
      "grad_norm": 1.8933207988739014,
      "learning_rate": 2.6186068321444527e-07,
      "loss": 0.4457,
      "step": 418090
    },
    {
      "epoch": 171.14,
      "grad_norm": 1.8784786462783813,
      "learning_rate": 2.6178782684300686e-07,
      "loss": 0.4592,
      "step": 418100
    },
    {
      "epoch": 171.15,
      "grad_norm": 2.19236421585083,
      "learning_rate": 2.6171498000871527e-07,
      "loss": 0.4392,
      "step": 418110
    },
    {
      "epoch": 171.15,
      "grad_norm": 1.8601139783859253,
      "learning_rate": 2.616421427119031e-07,
      "loss": 0.4266,
      "step": 418120
    },
    {
      "epoch": 171.15,
      "grad_norm": 2.0430171489715576,
      "learning_rate": 2.6156931495290463e-07,
      "loss": 0.4528,
      "step": 418130
    },
    {
      "epoch": 171.16,
      "grad_norm": 1.7858328819274902,
      "learning_rate": 2.6149649673205336e-07,
      "loss": 0.4298,
      "step": 418140
    },
    {
      "epoch": 171.16,
      "grad_norm": 2.239847183227539,
      "learning_rate": 2.6142368804968293e-07,
      "loss": 0.4426,
      "step": 418150
    },
    {
      "epoch": 171.17,
      "grad_norm": 10.882783889770508,
      "learning_rate": 2.613508889061266e-07,
      "loss": 0.4645,
      "step": 418160
    },
    {
      "epoch": 171.17,
      "grad_norm": 1.8500854969024658,
      "learning_rate": 2.6127809930171833e-07,
      "loss": 0.474,
      "step": 418170
    },
    {
      "epoch": 171.17,
      "grad_norm": 2.0172712802886963,
      "learning_rate": 2.6120531923679124e-07,
      "loss": 0.444,
      "step": 418180
    },
    {
      "epoch": 171.18,
      "grad_norm": 2.4695210456848145,
      "learning_rate": 2.61132548711679e-07,
      "loss": 0.4506,
      "step": 418190
    },
    {
      "epoch": 171.18,
      "grad_norm": 2.1589481830596924,
      "learning_rate": 2.610597877267145e-07,
      "loss": 0.4286,
      "step": 418200
    },
    {
      "epoch": 171.19,
      "grad_norm": 2.0390779972076416,
      "learning_rate": 2.609870362822315e-07,
      "loss": 0.4235,
      "step": 418210
    },
    {
      "epoch": 171.19,
      "grad_norm": 2.5210041999816895,
      "learning_rate": 2.609142943785632e-07,
      "loss": 0.4343,
      "step": 418220
    },
    {
      "epoch": 171.2,
      "grad_norm": 2.3079309463500977,
      "learning_rate": 2.608415620160425e-07,
      "loss": 0.4484,
      "step": 418230
    },
    {
      "epoch": 171.2,
      "grad_norm": 2.3146820068359375,
      "learning_rate": 2.607688391950029e-07,
      "loss": 0.4259,
      "step": 418240
    },
    {
      "epoch": 171.2,
      "grad_norm": 1.975376009941101,
      "learning_rate": 2.606961259157778e-07,
      "loss": 0.4561,
      "step": 418250
    },
    {
      "epoch": 171.21,
      "grad_norm": 2.3068411350250244,
      "learning_rate": 2.606234221786995e-07,
      "loss": 0.4183,
      "step": 418260
    },
    {
      "epoch": 171.21,
      "grad_norm": 1.8708614110946655,
      "learning_rate": 2.605507279841015e-07,
      "loss": 0.4459,
      "step": 418270
    },
    {
      "epoch": 171.22,
      "grad_norm": 1.951088309288025,
      "learning_rate": 2.604780433323166e-07,
      "loss": 0.4444,
      "step": 418280
    },
    {
      "epoch": 171.22,
      "grad_norm": 1.7731080055236816,
      "learning_rate": 2.60405368223678e-07,
      "loss": 0.4438,
      "step": 418290
    },
    {
      "epoch": 171.22,
      "grad_norm": 2.041952610015869,
      "learning_rate": 2.603327026585184e-07,
      "loss": 0.4486,
      "step": 418300
    },
    {
      "epoch": 171.23,
      "grad_norm": 2.315089464187622,
      "learning_rate": 2.6026004663717094e-07,
      "loss": 0.4324,
      "step": 418310
    },
    {
      "epoch": 171.23,
      "grad_norm": 1.9603066444396973,
      "learning_rate": 2.601874001599681e-07,
      "loss": 0.4285,
      "step": 418320
    },
    {
      "epoch": 171.24,
      "grad_norm": 1.9477598667144775,
      "learning_rate": 2.6011476322724294e-07,
      "loss": 0.4553,
      "step": 418330
    },
    {
      "epoch": 171.24,
      "grad_norm": 2.8927364349365234,
      "learning_rate": 2.60042135839328e-07,
      "loss": 0.4498,
      "step": 418340
    },
    {
      "epoch": 171.24,
      "grad_norm": 2.2118477821350098,
      "learning_rate": 2.599695179965557e-07,
      "loss": 0.4398,
      "step": 418350
    },
    {
      "epoch": 171.25,
      "grad_norm": 2.344442844390869,
      "learning_rate": 2.5989690969925974e-07,
      "loss": 0.4598,
      "step": 418360
    },
    {
      "epoch": 171.25,
      "grad_norm": 1.9002996683120728,
      "learning_rate": 2.598243109477719e-07,
      "loss": 0.4442,
      "step": 418370
    },
    {
      "epoch": 171.26,
      "grad_norm": 1.9592512845993042,
      "learning_rate": 2.5975172174242536e-07,
      "loss": 0.4611,
      "step": 418380
    },
    {
      "epoch": 171.26,
      "grad_norm": 2.567978620529175,
      "learning_rate": 2.596791420835515e-07,
      "loss": 0.4424,
      "step": 418390
    },
    {
      "epoch": 171.26,
      "grad_norm": 1.7491358518600464,
      "learning_rate": 2.5960657197148367e-07,
      "loss": 0.4326,
      "step": 418400
    },
    {
      "epoch": 171.27,
      "grad_norm": 2.439906597137451,
      "learning_rate": 2.595340114065542e-07,
      "loss": 0.4768,
      "step": 418410
    },
    {
      "epoch": 171.27,
      "grad_norm": 2.1839215755462646,
      "learning_rate": 2.5946146038909536e-07,
      "loss": 0.4403,
      "step": 418420
    },
    {
      "epoch": 171.28,
      "grad_norm": 2.0820162296295166,
      "learning_rate": 2.593889189194395e-07,
      "loss": 0.4611,
      "step": 418430
    },
    {
      "epoch": 171.28,
      "grad_norm": 1.9676662683486938,
      "learning_rate": 2.59316386997919e-07,
      "loss": 0.4459,
      "step": 418440
    },
    {
      "epoch": 171.29,
      "grad_norm": 2.0363540649414062,
      "learning_rate": 2.59243864624866e-07,
      "loss": 0.4208,
      "step": 418450
    },
    {
      "epoch": 171.29,
      "grad_norm": 1.8720953464508057,
      "learning_rate": 2.5917135180061283e-07,
      "loss": 0.4389,
      "step": 418460
    },
    {
      "epoch": 171.29,
      "grad_norm": 2.56174898147583,
      "learning_rate": 2.590988485254914e-07,
      "loss": 0.4418,
      "step": 418470
    },
    {
      "epoch": 171.3,
      "grad_norm": 2.0169107913970947,
      "learning_rate": 2.5902635479983433e-07,
      "loss": 0.4344,
      "step": 418480
    },
    {
      "epoch": 171.3,
      "grad_norm": 2.223160743713379,
      "learning_rate": 2.5895387062397337e-07,
      "loss": 0.4467,
      "step": 418490
    },
    {
      "epoch": 171.31,
      "grad_norm": 1.9581146240234375,
      "learning_rate": 2.5888139599824053e-07,
      "loss": 0.4393,
      "step": 418500
    },
    {
      "epoch": 171.31,
      "grad_norm": 1.822302222251892,
      "learning_rate": 2.5880893092296787e-07,
      "loss": 0.4194,
      "step": 418510
    },
    {
      "epoch": 171.31,
      "grad_norm": 2.258781671524048,
      "learning_rate": 2.587364753984874e-07,
      "loss": 0.4692,
      "step": 418520
    },
    {
      "epoch": 171.32,
      "grad_norm": 2.1181023120880127,
      "learning_rate": 2.5866402942513094e-07,
      "loss": 0.4643,
      "step": 418530
    },
    {
      "epoch": 171.32,
      "grad_norm": 2.1564085483551025,
      "learning_rate": 2.5859159300323037e-07,
      "loss": 0.4676,
      "step": 418540
    },
    {
      "epoch": 171.33,
      "grad_norm": 1.797209620475769,
      "learning_rate": 2.585191661331176e-07,
      "loss": 0.4369,
      "step": 418550
    },
    {
      "epoch": 171.33,
      "grad_norm": 1.9164868593215942,
      "learning_rate": 2.584467488151243e-07,
      "loss": 0.4418,
      "step": 418560
    },
    {
      "epoch": 171.33,
      "grad_norm": 1.975294828414917,
      "learning_rate": 2.583743410495823e-07,
      "loss": 0.4349,
      "step": 418570
    },
    {
      "epoch": 171.34,
      "grad_norm": 2.349226474761963,
      "learning_rate": 2.583019428368231e-07,
      "loss": 0.4655,
      "step": 418580
    },
    {
      "epoch": 171.34,
      "grad_norm": 2.068044424057007,
      "learning_rate": 2.582295541771787e-07,
      "loss": 0.4657,
      "step": 418590
    },
    {
      "epoch": 171.35,
      "grad_norm": 1.8846848011016846,
      "learning_rate": 2.5815717507098035e-07,
      "loss": 0.4696,
      "step": 418600
    },
    {
      "epoch": 171.35,
      "grad_norm": 2.027254104614258,
      "learning_rate": 2.5808480551855956e-07,
      "loss": 0.4712,
      "step": 418610
    },
    {
      "epoch": 171.35,
      "grad_norm": 2.168409585952759,
      "learning_rate": 2.5801244552024837e-07,
      "loss": 0.4588,
      "step": 418620
    },
    {
      "epoch": 171.36,
      "grad_norm": 2.209282159805298,
      "learning_rate": 2.579400950763777e-07,
      "loss": 0.4442,
      "step": 418630
    },
    {
      "epoch": 171.36,
      "grad_norm": 1.9659377336502075,
      "learning_rate": 2.5786775418727917e-07,
      "loss": 0.462,
      "step": 418640
    },
    {
      "epoch": 171.37,
      "grad_norm": 2.005955219268799,
      "learning_rate": 2.577954228532842e-07,
      "loss": 0.4491,
      "step": 418650
    },
    {
      "epoch": 171.37,
      "grad_norm": 2.310220241546631,
      "learning_rate": 2.5772310107472406e-07,
      "loss": 0.4463,
      "step": 418660
    },
    {
      "epoch": 171.38,
      "grad_norm": 2.224837064743042,
      "learning_rate": 2.5765078885193026e-07,
      "loss": 0.4545,
      "step": 418670
    },
    {
      "epoch": 171.38,
      "grad_norm": 1.6618928909301758,
      "learning_rate": 2.575784861852338e-07,
      "loss": 0.4456,
      "step": 418680
    },
    {
      "epoch": 171.38,
      "grad_norm": 2.3260204792022705,
      "learning_rate": 2.575061930749661e-07,
      "loss": 0.427,
      "step": 418690
    },
    {
      "epoch": 171.39,
      "grad_norm": 1.8336180448532104,
      "learning_rate": 2.57433909521458e-07,
      "loss": 0.4744,
      "step": 418700
    },
    {
      "epoch": 171.39,
      "grad_norm": 1.8779338598251343,
      "learning_rate": 2.5736163552504097e-07,
      "loss": 0.4407,
      "step": 418710
    },
    {
      "epoch": 171.4,
      "grad_norm": 1.9161781072616577,
      "learning_rate": 2.5728937108604514e-07,
      "loss": 0.4458,
      "step": 418720
    },
    {
      "epoch": 171.4,
      "grad_norm": 1.9827766418457031,
      "learning_rate": 2.572171162048029e-07,
      "loss": 0.4485,
      "step": 418730
    },
    {
      "epoch": 171.4,
      "grad_norm": 1.7920695543289185,
      "learning_rate": 2.571448708816447e-07,
      "loss": 0.4373,
      "step": 418740
    },
    {
      "epoch": 171.41,
      "grad_norm": 1.7706016302108765,
      "learning_rate": 2.5707263511690144e-07,
      "loss": 0.4448,
      "step": 418750
    },
    {
      "epoch": 171.41,
      "grad_norm": 1.9193699359893799,
      "learning_rate": 2.570004089109039e-07,
      "loss": 0.4365,
      "step": 418760
    },
    {
      "epoch": 171.42,
      "grad_norm": 2.1382968425750732,
      "learning_rate": 2.569281922639834e-07,
      "loss": 0.4516,
      "step": 418770
    },
    {
      "epoch": 171.42,
      "grad_norm": 2.370074510574341,
      "learning_rate": 2.5685598517647e-07,
      "loss": 0.4437,
      "step": 418780
    },
    {
      "epoch": 171.42,
      "grad_norm": 1.8875309228897095,
      "learning_rate": 2.56783787648695e-07,
      "loss": 0.4518,
      "step": 418790
    },
    {
      "epoch": 171.43,
      "grad_norm": 11.042028427124023,
      "learning_rate": 2.5671159968098924e-07,
      "loss": 0.4357,
      "step": 418800
    },
    {
      "epoch": 171.43,
      "grad_norm": 2.2971456050872803,
      "learning_rate": 2.5663942127368303e-07,
      "loss": 0.4672,
      "step": 418810
    },
    {
      "epoch": 171.44,
      "grad_norm": 2.0551960468292236,
      "learning_rate": 2.5656725242710716e-07,
      "loss": 0.4494,
      "step": 418820
    },
    {
      "epoch": 171.44,
      "grad_norm": 1.9263995885849,
      "learning_rate": 2.5649509314159256e-07,
      "loss": 0.4373,
      "step": 418830
    },
    {
      "epoch": 171.44,
      "grad_norm": 1.9756819009780884,
      "learning_rate": 2.564229434174693e-07,
      "loss": 0.4371,
      "step": 418840
    },
    {
      "epoch": 171.45,
      "grad_norm": 2.1278464794158936,
      "learning_rate": 2.563508032550677e-07,
      "loss": 0.4358,
      "step": 418850
    },
    {
      "epoch": 171.45,
      "grad_norm": 2.3248677253723145,
      "learning_rate": 2.562786726547186e-07,
      "loss": 0.4543,
      "step": 418860
    },
    {
      "epoch": 171.46,
      "grad_norm": 2.600433111190796,
      "learning_rate": 2.562065516167524e-07,
      "loss": 0.4564,
      "step": 418870
    },
    {
      "epoch": 171.46,
      "grad_norm": 2.1516683101654053,
      "learning_rate": 2.561344401414993e-07,
      "loss": 0.4552,
      "step": 418880
    },
    {
      "epoch": 171.47,
      "grad_norm": 1.897401213645935,
      "learning_rate": 2.5606233822929006e-07,
      "loss": 0.4513,
      "step": 418890
    },
    {
      "epoch": 171.47,
      "grad_norm": 1.915219783782959,
      "learning_rate": 2.5599024588045464e-07,
      "loss": 0.4327,
      "step": 418900
    },
    {
      "epoch": 171.47,
      "grad_norm": 2.1201345920562744,
      "learning_rate": 2.559181630953227e-07,
      "loss": 0.4536,
      "step": 418910
    },
    {
      "epoch": 171.48,
      "grad_norm": 2.215916633605957,
      "learning_rate": 2.558460898742254e-07,
      "loss": 0.4274,
      "step": 418920
    },
    {
      "epoch": 171.48,
      "grad_norm": 2.110839366912842,
      "learning_rate": 2.557740262174928e-07,
      "loss": 0.4511,
      "step": 418930
    },
    {
      "epoch": 171.49,
      "grad_norm": 2.1469004154205322,
      "learning_rate": 2.5570197212545484e-07,
      "loss": 0.446,
      "step": 418940
    },
    {
      "epoch": 171.49,
      "grad_norm": 1.707028865814209,
      "learning_rate": 2.5562992759844145e-07,
      "loss": 0.4388,
      "step": 418950
    },
    {
      "epoch": 171.49,
      "grad_norm": 2.281564235687256,
      "learning_rate": 2.555578926367825e-07,
      "loss": 0.4487,
      "step": 418960
    },
    {
      "epoch": 171.5,
      "grad_norm": 2.1013784408569336,
      "learning_rate": 2.5548586724080883e-07,
      "loss": 0.4592,
      "step": 418970
    },
    {
      "epoch": 171.5,
      "grad_norm": 1.910767674446106,
      "learning_rate": 2.554138514108493e-07,
      "loss": 0.4385,
      "step": 418980
    },
    {
      "epoch": 171.51,
      "grad_norm": 2.6067957878112793,
      "learning_rate": 2.5534184514723426e-07,
      "loss": 0.4621,
      "step": 418990
    },
    {
      "epoch": 171.51,
      "grad_norm": 2.1206796169281006,
      "learning_rate": 2.552698484502936e-07,
      "loss": 0.4535,
      "step": 419000
    },
    {
      "epoch": 171.51,
      "grad_norm": 2.287813186645508,
      "learning_rate": 2.5519786132035684e-07,
      "loss": 0.4502,
      "step": 419010
    },
    {
      "epoch": 171.52,
      "grad_norm": 2.6980226039886475,
      "learning_rate": 2.551258837577543e-07,
      "loss": 0.439,
      "step": 419020
    },
    {
      "epoch": 171.52,
      "grad_norm": 2.108342409133911,
      "learning_rate": 2.55053915762815e-07,
      "loss": 0.4347,
      "step": 419030
    },
    {
      "epoch": 171.53,
      "grad_norm": 1.8517155647277832,
      "learning_rate": 2.549819573358693e-07,
      "loss": 0.4333,
      "step": 419040
    },
    {
      "epoch": 171.53,
      "grad_norm": 2.116461992263794,
      "learning_rate": 2.5491000847724624e-07,
      "loss": 0.4402,
      "step": 419050
    },
    {
      "epoch": 171.53,
      "grad_norm": 1.891728401184082,
      "learning_rate": 2.5483806918727587e-07,
      "loss": 0.4398,
      "step": 419060
    },
    {
      "epoch": 171.54,
      "grad_norm": 2.332491159439087,
      "learning_rate": 2.547661394662876e-07,
      "loss": 0.4503,
      "step": 419070
    },
    {
      "epoch": 171.54,
      "grad_norm": 2.057823657989502,
      "learning_rate": 2.546942193146107e-07,
      "loss": 0.4439,
      "step": 419080
    },
    {
      "epoch": 171.55,
      "grad_norm": 2.534776449203491,
      "learning_rate": 2.5462230873257507e-07,
      "loss": 0.4613,
      "step": 419090
    },
    {
      "epoch": 171.55,
      "grad_norm": 2.2971463203430176,
      "learning_rate": 2.545504077205095e-07,
      "loss": 0.4545,
      "step": 419100
    },
    {
      "epoch": 171.56,
      "grad_norm": 2.0692272186279297,
      "learning_rate": 2.5447851627874393e-07,
      "loss": 0.4401,
      "step": 419110
    },
    {
      "epoch": 171.56,
      "grad_norm": 3.317490816116333,
      "learning_rate": 2.544066344076076e-07,
      "loss": 0.4555,
      "step": 419120
    },
    {
      "epoch": 171.56,
      "grad_norm": 2.3636436462402344,
      "learning_rate": 2.543347621074293e-07,
      "loss": 0.43,
      "step": 419130
    },
    {
      "epoch": 171.57,
      "grad_norm": 2.1476545333862305,
      "learning_rate": 2.542628993785388e-07,
      "loss": 0.4333,
      "step": 419140
    },
    {
      "epoch": 171.57,
      "grad_norm": 1.9414530992507935,
      "learning_rate": 2.5419104622126524e-07,
      "loss": 0.4454,
      "step": 419150
    },
    {
      "epoch": 171.58,
      "grad_norm": 2.128221273422241,
      "learning_rate": 2.541192026359375e-07,
      "loss": 0.4399,
      "step": 419160
    },
    {
      "epoch": 171.58,
      "grad_norm": 2.193817615509033,
      "learning_rate": 2.5404736862288475e-07,
      "loss": 0.4529,
      "step": 419170
    },
    {
      "epoch": 171.58,
      "grad_norm": 2.4330475330352783,
      "learning_rate": 2.539755441824362e-07,
      "loss": 0.447,
      "step": 419180
    },
    {
      "epoch": 171.59,
      "grad_norm": 1.919327974319458,
      "learning_rate": 2.539037293149207e-07,
      "loss": 0.4409,
      "step": 419190
    },
    {
      "epoch": 171.59,
      "grad_norm": 3.4001591205596924,
      "learning_rate": 2.5383192402066743e-07,
      "loss": 0.4507,
      "step": 419200
    },
    {
      "epoch": 171.6,
      "grad_norm": 2.0737383365631104,
      "learning_rate": 2.5376012830000505e-07,
      "loss": 0.4295,
      "step": 419210
    },
    {
      "epoch": 171.6,
      "grad_norm": 3.181746482849121,
      "learning_rate": 2.5368834215326274e-07,
      "loss": 0.4258,
      "step": 419220
    },
    {
      "epoch": 171.6,
      "grad_norm": 2.0560882091522217,
      "learning_rate": 2.5361656558076936e-07,
      "loss": 0.4492,
      "step": 419230
    },
    {
      "epoch": 171.61,
      "grad_norm": 2.135472297668457,
      "learning_rate": 2.535447985828533e-07,
      "loss": 0.4335,
      "step": 419240
    },
    {
      "epoch": 171.61,
      "grad_norm": 2.604480266571045,
      "learning_rate": 2.5347304115984345e-07,
      "loss": 0.4332,
      "step": 419250
    },
    {
      "epoch": 171.62,
      "grad_norm": 1.77279794216156,
      "learning_rate": 2.5340129331206876e-07,
      "loss": 0.4609,
      "step": 419260
    },
    {
      "epoch": 171.62,
      "grad_norm": 2.355445146560669,
      "learning_rate": 2.5332955503985786e-07,
      "loss": 0.4573,
      "step": 419270
    },
    {
      "epoch": 171.63,
      "grad_norm": 1.868756651878357,
      "learning_rate": 2.532578263435396e-07,
      "loss": 0.4188,
      "step": 419280
    },
    {
      "epoch": 171.63,
      "grad_norm": 1.9492295980453491,
      "learning_rate": 2.5318610722344143e-07,
      "loss": 0.4317,
      "step": 419290
    },
    {
      "epoch": 171.63,
      "grad_norm": 2.0383546352386475,
      "learning_rate": 2.531143976798932e-07,
      "loss": 0.4431,
      "step": 419300
    },
    {
      "epoch": 171.64,
      "grad_norm": 2.32913875579834,
      "learning_rate": 2.530426977132228e-07,
      "loss": 0.4401,
      "step": 419310
    },
    {
      "epoch": 171.64,
      "grad_norm": 2.3031764030456543,
      "learning_rate": 2.529710073237586e-07,
      "loss": 0.4272,
      "step": 419320
    },
    {
      "epoch": 171.65,
      "grad_norm": 2.234015703201294,
      "learning_rate": 2.5289932651182954e-07,
      "loss": 0.44,
      "step": 419330
    },
    {
      "epoch": 171.65,
      "grad_norm": 1.8810044527053833,
      "learning_rate": 2.528276552777632e-07,
      "loss": 0.4558,
      "step": 419340
    },
    {
      "epoch": 171.65,
      "grad_norm": 1.6781766414642334,
      "learning_rate": 2.527559936218885e-07,
      "loss": 0.4355,
      "step": 419350
    },
    {
      "epoch": 171.66,
      "grad_norm": 2.217418670654297,
      "learning_rate": 2.526843415445335e-07,
      "loss": 0.4457,
      "step": 419360
    },
    {
      "epoch": 171.66,
      "grad_norm": 2.1075000762939453,
      "learning_rate": 2.5261269904602664e-07,
      "loss": 0.4433,
      "step": 419370
    },
    {
      "epoch": 171.67,
      "grad_norm": 2.427051544189453,
      "learning_rate": 2.5254106612669555e-07,
      "loss": 0.4489,
      "step": 419380
    },
    {
      "epoch": 171.67,
      "grad_norm": 2.140249490737915,
      "learning_rate": 2.524694427868688e-07,
      "loss": 0.4351,
      "step": 419390
    },
    {
      "epoch": 171.67,
      "grad_norm": 2.241185188293457,
      "learning_rate": 2.523978290268746e-07,
      "loss": 0.4306,
      "step": 419400
    },
    {
      "epoch": 171.68,
      "grad_norm": 2.162191867828369,
      "learning_rate": 2.5232622484704126e-07,
      "loss": 0.4346,
      "step": 419410
    },
    {
      "epoch": 171.68,
      "grad_norm": 2.1775002479553223,
      "learning_rate": 2.5225463024769563e-07,
      "loss": 0.4455,
      "step": 419420
    },
    {
      "epoch": 171.69,
      "grad_norm": 2.021017551422119,
      "learning_rate": 2.5218304522916657e-07,
      "loss": 0.4548,
      "step": 419430
    },
    {
      "epoch": 171.69,
      "grad_norm": 2.1181557178497314,
      "learning_rate": 2.5211146979178176e-07,
      "loss": 0.43,
      "step": 419440
    },
    {
      "epoch": 171.69,
      "grad_norm": 1.8954592943191528,
      "learning_rate": 2.52039903935869e-07,
      "loss": 0.4195,
      "step": 419450
    },
    {
      "epoch": 171.7,
      "grad_norm": 2.0199508666992188,
      "learning_rate": 2.5196834766175644e-07,
      "loss": 0.4511,
      "step": 419460
    },
    {
      "epoch": 171.7,
      "grad_norm": 2.5018310546875,
      "learning_rate": 2.5189680096977174e-07,
      "loss": 0.4376,
      "step": 419470
    },
    {
      "epoch": 171.71,
      "grad_norm": 2.1374616622924805,
      "learning_rate": 2.5182526386024186e-07,
      "loss": 0.4159,
      "step": 419480
    },
    {
      "epoch": 171.71,
      "grad_norm": 1.808489441871643,
      "learning_rate": 2.517537363334958e-07,
      "loss": 0.46,
      "step": 419490
    },
    {
      "epoch": 171.72,
      "grad_norm": 2.045073986053467,
      "learning_rate": 2.516822183898608e-07,
      "loss": 0.4478,
      "step": 419500
    },
    {
      "epoch": 171.72,
      "grad_norm": 2.2005326747894287,
      "learning_rate": 2.5161071002966404e-07,
      "loss": 0.4278,
      "step": 419510
    },
    {
      "epoch": 171.72,
      "grad_norm": 2.0861551761627197,
      "learning_rate": 2.515392112532334e-07,
      "loss": 0.4449,
      "step": 419520
    },
    {
      "epoch": 171.73,
      "grad_norm": 2.358886957168579,
      "learning_rate": 2.5146772206089665e-07,
      "loss": 0.4622,
      "step": 419530
    },
    {
      "epoch": 171.73,
      "grad_norm": 1.9792087078094482,
      "learning_rate": 2.513962424529807e-07,
      "loss": 0.4424,
      "step": 419540
    },
    {
      "epoch": 171.74,
      "grad_norm": 2.150702714920044,
      "learning_rate": 2.5132477242981385e-07,
      "loss": 0.4522,
      "step": 419550
    },
    {
      "epoch": 171.74,
      "grad_norm": 2.082601547241211,
      "learning_rate": 2.5125331199172245e-07,
      "loss": 0.4567,
      "step": 419560
    },
    {
      "epoch": 171.74,
      "grad_norm": 2.122366189956665,
      "learning_rate": 2.511818611390341e-07,
      "loss": 0.439,
      "step": 419570
    },
    {
      "epoch": 171.75,
      "grad_norm": 2.0146167278289795,
      "learning_rate": 2.5111041987207666e-07,
      "loss": 0.4565,
      "step": 419580
    },
    {
      "epoch": 171.75,
      "grad_norm": 2.020019292831421,
      "learning_rate": 2.5103898819117664e-07,
      "loss": 0.4365,
      "step": 419590
    },
    {
      "epoch": 171.76,
      "grad_norm": 2.1185214519500732,
      "learning_rate": 2.50967566096662e-07,
      "loss": 0.414,
      "step": 419600
    },
    {
      "epoch": 171.76,
      "grad_norm": 1.8259003162384033,
      "learning_rate": 2.508961535888595e-07,
      "loss": 0.4377,
      "step": 419610
    },
    {
      "epoch": 171.76,
      "grad_norm": 1.8968461751937866,
      "learning_rate": 2.508247506680965e-07,
      "loss": 0.4261,
      "step": 419620
    },
    {
      "epoch": 171.77,
      "grad_norm": 2.305706739425659,
      "learning_rate": 2.5075335733469956e-07,
      "loss": 0.4591,
      "step": 419630
    },
    {
      "epoch": 171.77,
      "grad_norm": 2.2182445526123047,
      "learning_rate": 2.5068197358899656e-07,
      "loss": 0.4355,
      "step": 419640
    },
    {
      "epoch": 171.78,
      "grad_norm": 1.931897521018982,
      "learning_rate": 2.506105994313137e-07,
      "loss": 0.4302,
      "step": 419650
    },
    {
      "epoch": 171.78,
      "grad_norm": 1.663577914237976,
      "learning_rate": 2.5053923486197825e-07,
      "loss": 0.4415,
      "step": 419660
    },
    {
      "epoch": 171.78,
      "grad_norm": 1.9277695417404175,
      "learning_rate": 2.504678798813169e-07,
      "loss": 0.4498,
      "step": 419670
    },
    {
      "epoch": 171.79,
      "grad_norm": 1.987076997756958,
      "learning_rate": 2.503965344896575e-07,
      "loss": 0.4312,
      "step": 419680
    },
    {
      "epoch": 171.79,
      "grad_norm": 2.110649347305298,
      "learning_rate": 2.503251986873259e-07,
      "loss": 0.4526,
      "step": 419690
    },
    {
      "epoch": 171.8,
      "grad_norm": 1.9541234970092773,
      "learning_rate": 2.5025387247464886e-07,
      "loss": 0.4432,
      "step": 419700
    },
    {
      "epoch": 171.8,
      "grad_norm": 1.7892717123031616,
      "learning_rate": 2.501825558519535e-07,
      "loss": 0.4516,
      "step": 419710
    },
    {
      "epoch": 171.81,
      "grad_norm": 2.261251449584961,
      "learning_rate": 2.501112488195664e-07,
      "loss": 0.439,
      "step": 419720
    },
    {
      "epoch": 171.81,
      "grad_norm": 3.3633103370666504,
      "learning_rate": 2.500399513778141e-07,
      "loss": 0.4394,
      "step": 419730
    },
    {
      "epoch": 171.81,
      "grad_norm": 2.0655715465545654,
      "learning_rate": 2.4996866352702313e-07,
      "loss": 0.4522,
      "step": 419740
    },
    {
      "epoch": 171.82,
      "grad_norm": 2.4191172122955322,
      "learning_rate": 2.4989738526752035e-07,
      "loss": 0.4341,
      "step": 419750
    },
    {
      "epoch": 171.82,
      "grad_norm": 2.1503562927246094,
      "learning_rate": 2.4982611659963204e-07,
      "loss": 0.4433,
      "step": 419760
    },
    {
      "epoch": 171.83,
      "grad_norm": 1.7167514562606812,
      "learning_rate": 2.4975485752368507e-07,
      "loss": 0.4571,
      "step": 419770
    },
    {
      "epoch": 171.83,
      "grad_norm": 1.8921375274658203,
      "learning_rate": 2.4968360804000516e-07,
      "loss": 0.4309,
      "step": 419780
    },
    {
      "epoch": 171.83,
      "grad_norm": 2.2383060455322266,
      "learning_rate": 2.496123681489194e-07,
      "loss": 0.4316,
      "step": 419790
    },
    {
      "epoch": 171.84,
      "grad_norm": 1.7739207744598389,
      "learning_rate": 2.4954113785075364e-07,
      "loss": 0.4383,
      "step": 419800
    },
    {
      "epoch": 171.84,
      "grad_norm": 1.977089762687683,
      "learning_rate": 2.4946991714583434e-07,
      "loss": 0.4706,
      "step": 419810
    },
    {
      "epoch": 171.85,
      "grad_norm": 2.041083812713623,
      "learning_rate": 2.4939870603448786e-07,
      "loss": 0.4654,
      "step": 419820
    },
    {
      "epoch": 171.85,
      "grad_norm": 1.9296406507492065,
      "learning_rate": 2.4932750451704026e-07,
      "loss": 0.4346,
      "step": 419830
    },
    {
      "epoch": 171.85,
      "grad_norm": 2.1346256732940674,
      "learning_rate": 2.4925631259381754e-07,
      "loss": 0.4372,
      "step": 419840
    },
    {
      "epoch": 171.86,
      "grad_norm": 2.250065326690674,
      "learning_rate": 2.4918513026514623e-07,
      "loss": 0.4391,
      "step": 419850
    },
    {
      "epoch": 171.86,
      "grad_norm": 2.0305488109588623,
      "learning_rate": 2.4911395753135223e-07,
      "loss": 0.4444,
      "step": 419860
    },
    {
      "epoch": 171.87,
      "grad_norm": 1.842044472694397,
      "learning_rate": 2.490427943927615e-07,
      "loss": 0.4448,
      "step": 419870
    },
    {
      "epoch": 171.87,
      "grad_norm": 2.234663248062134,
      "learning_rate": 2.4897164084970005e-07,
      "loss": 0.45,
      "step": 419880
    },
    {
      "epoch": 171.87,
      "grad_norm": 2.083371162414551,
      "learning_rate": 2.489004969024939e-07,
      "loss": 0.4577,
      "step": 419890
    },
    {
      "epoch": 171.88,
      "grad_norm": 1.9778063297271729,
      "learning_rate": 2.48829362551469e-07,
      "loss": 0.4256,
      "step": 419900
    },
    {
      "epoch": 171.88,
      "grad_norm": 2.018453598022461,
      "learning_rate": 2.487582377969507e-07,
      "loss": 0.4437,
      "step": 419910
    },
    {
      "epoch": 171.89,
      "grad_norm": 2.420924425125122,
      "learning_rate": 2.486871226392657e-07,
      "loss": 0.4538,
      "step": 419920
    },
    {
      "epoch": 171.89,
      "grad_norm": 2.3282880783081055,
      "learning_rate": 2.4861601707873914e-07,
      "loss": 0.4312,
      "step": 419930
    },
    {
      "epoch": 171.9,
      "grad_norm": 2.0701212882995605,
      "learning_rate": 2.485449211156966e-07,
      "loss": 0.4566,
      "step": 419940
    },
    {
      "epoch": 171.9,
      "grad_norm": 1.8876750469207764,
      "learning_rate": 2.484738347504643e-07,
      "loss": 0.4375,
      "step": 419950
    },
    {
      "epoch": 171.9,
      "grad_norm": 1.8190274238586426,
      "learning_rate": 2.4840275798336763e-07,
      "loss": 0.4592,
      "step": 419960
    },
    {
      "epoch": 171.91,
      "grad_norm": 2.056424856185913,
      "learning_rate": 2.48331690814732e-07,
      "loss": 0.4555,
      "step": 419970
    },
    {
      "epoch": 171.91,
      "grad_norm": 1.8201571702957153,
      "learning_rate": 2.482606332448833e-07,
      "loss": 0.4362,
      "step": 419980
    },
    {
      "epoch": 171.92,
      "grad_norm": 2.1638762950897217,
      "learning_rate": 2.481895852741472e-07,
      "loss": 0.4362,
      "step": 419990
    },
    {
      "epoch": 171.92,
      "grad_norm": 2.077523946762085,
      "learning_rate": 2.4811854690284855e-07,
      "loss": 0.4286,
      "step": 420000
    },
    {
      "epoch": 171.92,
      "grad_norm": 2.0354292392730713,
      "learning_rate": 2.4804751813131276e-07,
      "loss": 0.4457,
      "step": 420010
    },
    {
      "epoch": 171.93,
      "grad_norm": 2.1889796257019043,
      "learning_rate": 2.479764989598657e-07,
      "loss": 0.4586,
      "step": 420020
    },
    {
      "epoch": 171.93,
      "grad_norm": 2.1334598064422607,
      "learning_rate": 2.479054893888323e-07,
      "loss": 0.4302,
      "step": 420030
    },
    {
      "epoch": 171.94,
      "grad_norm": 2.8214848041534424,
      "learning_rate": 2.4783448941853814e-07,
      "loss": 0.4505,
      "step": 420040
    },
    {
      "epoch": 171.94,
      "grad_norm": 2.419421911239624,
      "learning_rate": 2.4776349904930794e-07,
      "loss": 0.4469,
      "step": 420050
    },
    {
      "epoch": 171.94,
      "grad_norm": 2.350708484649658,
      "learning_rate": 2.4769251828146766e-07,
      "loss": 0.4457,
      "step": 420060
    },
    {
      "epoch": 171.95,
      "grad_norm": 2.2594356536865234,
      "learning_rate": 2.476215471153422e-07,
      "loss": 0.4646,
      "step": 420070
    },
    {
      "epoch": 171.95,
      "grad_norm": 2.1325902938842773,
      "learning_rate": 2.475505855512564e-07,
      "loss": 0.4569,
      "step": 420080
    },
    {
      "epoch": 171.96,
      "grad_norm": 1.7838504314422607,
      "learning_rate": 2.4747963358953544e-07,
      "loss": 0.4667,
      "step": 420090
    },
    {
      "epoch": 171.96,
      "grad_norm": 2.3357009887695312,
      "learning_rate": 2.474086912305047e-07,
      "loss": 0.4334,
      "step": 420100
    },
    {
      "epoch": 171.96,
      "grad_norm": 2.4768760204315186,
      "learning_rate": 2.473377584744884e-07,
      "loss": 0.4508,
      "step": 420110
    },
    {
      "epoch": 171.97,
      "grad_norm": 2.154531955718994,
      "learning_rate": 2.472668353218121e-07,
      "loss": 0.4237,
      "step": 420120
    },
    {
      "epoch": 171.97,
      "grad_norm": 2.2347002029418945,
      "learning_rate": 2.471959217728011e-07,
      "loss": 0.443,
      "step": 420130
    },
    {
      "epoch": 171.98,
      "grad_norm": 1.9270766973495483,
      "learning_rate": 2.4712501782777903e-07,
      "loss": 0.4445,
      "step": 420140
    },
    {
      "epoch": 171.98,
      "grad_norm": 2.268350124359131,
      "learning_rate": 2.470541234870714e-07,
      "loss": 0.4458,
      "step": 420150
    },
    {
      "epoch": 171.99,
      "grad_norm": 2.0471913814544678,
      "learning_rate": 2.469832387510028e-07,
      "loss": 0.4245,
      "step": 420160
    },
    {
      "epoch": 171.99,
      "grad_norm": 2.0493454933166504,
      "learning_rate": 2.4691236361989786e-07,
      "loss": 0.4287,
      "step": 420170
    },
    {
      "epoch": 171.99,
      "grad_norm": 2.0802628993988037,
      "learning_rate": 2.468414980940817e-07,
      "loss": 0.4553,
      "step": 420180
    },
    {
      "epoch": 172.0,
      "grad_norm": 2.0718657970428467,
      "learning_rate": 2.4677064217387837e-07,
      "loss": 0.4478,
      "step": 420190
    },
    {
      "epoch": 172.0,
      "eval_loss": 0.44684693217277527,
      "eval_runtime": 52.4825,
      "eval_samples_per_second": 65.717,
      "eval_steps_per_second": 8.231,
      "step": 420196
    },
    {
      "epoch": 172.0,
      "grad_norm": 2.3233909606933594,
      "learning_rate": 2.46699795859613e-07,
      "loss": 0.4306,
      "step": 420200
    },
    {
      "epoch": 172.01,
      "grad_norm": 2.8151183128356934,
      "learning_rate": 2.466289591516097e-07,
      "loss": 0.4422,
      "step": 420210
    },
    {
      "epoch": 172.01,
      "grad_norm": 2.0086610317230225,
      "learning_rate": 2.465581320501931e-07,
      "loss": 0.4412,
      "step": 420220
    },
    {
      "epoch": 172.01,
      "grad_norm": 2.1807661056518555,
      "learning_rate": 2.4648731455568705e-07,
      "loss": 0.4452,
      "step": 420230
    },
    {
      "epoch": 172.02,
      "grad_norm": 2.014324903488159,
      "learning_rate": 2.464165066684171e-07,
      "loss": 0.4448,
      "step": 420240
    },
    {
      "epoch": 172.02,
      "grad_norm": 1.7330894470214844,
      "learning_rate": 2.463457083887072e-07,
      "loss": 0.4372,
      "step": 420250
    },
    {
      "epoch": 172.03,
      "grad_norm": 1.9575382471084595,
      "learning_rate": 2.462749197168816e-07,
      "loss": 0.4535,
      "step": 420260
    },
    {
      "epoch": 172.03,
      "grad_norm": 2.259167194366455,
      "learning_rate": 2.4620414065326474e-07,
      "loss": 0.4279,
      "step": 420270
    },
    {
      "epoch": 172.03,
      "grad_norm": 1.8161225318908691,
      "learning_rate": 2.461333711981801e-07,
      "loss": 0.4494,
      "step": 420280
    },
    {
      "epoch": 172.04,
      "grad_norm": 2.342376470565796,
      "learning_rate": 2.4606261135195263e-07,
      "loss": 0.4648,
      "step": 420290
    },
    {
      "epoch": 172.04,
      "grad_norm": 2.301922082901001,
      "learning_rate": 2.459918611149061e-07,
      "loss": 0.4644,
      "step": 420300
    },
    {
      "epoch": 172.05,
      "grad_norm": 2.2211837768554688,
      "learning_rate": 2.4592112048736466e-07,
      "loss": 0.438,
      "step": 420310
    },
    {
      "epoch": 172.05,
      "grad_norm": 2.2353014945983887,
      "learning_rate": 2.458503894696526e-07,
      "loss": 0.4443,
      "step": 420320
    },
    {
      "epoch": 172.05,
      "grad_norm": 2.350881338119507,
      "learning_rate": 2.457796680620938e-07,
      "loss": 0.4556,
      "step": 420330
    },
    {
      "epoch": 172.06,
      "grad_norm": 2.0175909996032715,
      "learning_rate": 2.4570895626501196e-07,
      "loss": 0.4452,
      "step": 420340
    },
    {
      "epoch": 172.06,
      "grad_norm": 2.364880323410034,
      "learning_rate": 2.456382540787311e-07,
      "loss": 0.4412,
      "step": 420350
    },
    {
      "epoch": 172.07,
      "grad_norm": 1.9519611597061157,
      "learning_rate": 2.455675615035755e-07,
      "loss": 0.4398,
      "step": 420360
    },
    {
      "epoch": 172.07,
      "grad_norm": 2.1021602153778076,
      "learning_rate": 2.454968785398687e-07,
      "loss": 0.4339,
      "step": 420370
    },
    {
      "epoch": 172.08,
      "grad_norm": 2.3741095066070557,
      "learning_rate": 2.454262051879345e-07,
      "loss": 0.4681,
      "step": 420380
    },
    {
      "epoch": 172.08,
      "grad_norm": 2.0728485584259033,
      "learning_rate": 2.4535554144809657e-07,
      "loss": 0.4601,
      "step": 420390
    },
    {
      "epoch": 172.08,
      "grad_norm": 1.9500468969345093,
      "learning_rate": 2.452848873206786e-07,
      "loss": 0.4459,
      "step": 420400
    },
    {
      "epoch": 172.09,
      "grad_norm": 2.0838325023651123,
      "learning_rate": 2.452142428060043e-07,
      "loss": 0.4454,
      "step": 420410
    },
    {
      "epoch": 172.09,
      "grad_norm": 2.210552930831909,
      "learning_rate": 2.451436079043974e-07,
      "loss": 0.4514,
      "step": 420420
    },
    {
      "epoch": 172.1,
      "grad_norm": 1.977724313735962,
      "learning_rate": 2.450729826161813e-07,
      "loss": 0.4512,
      "step": 420430
    },
    {
      "epoch": 172.1,
      "grad_norm": 2.4280147552490234,
      "learning_rate": 2.450023669416797e-07,
      "loss": 0.4529,
      "step": 420440
    },
    {
      "epoch": 172.1,
      "grad_norm": 2.2956044673919678,
      "learning_rate": 2.4493176088121606e-07,
      "loss": 0.4523,
      "step": 420450
    },
    {
      "epoch": 172.11,
      "grad_norm": 2.0601422786712646,
      "learning_rate": 2.4486116443511354e-07,
      "loss": 0.4305,
      "step": 420460
    },
    {
      "epoch": 172.11,
      "grad_norm": 1.806828498840332,
      "learning_rate": 2.447905776036959e-07,
      "loss": 0.4631,
      "step": 420470
    },
    {
      "epoch": 172.12,
      "grad_norm": 2.1345608234405518,
      "learning_rate": 2.4472000038728595e-07,
      "loss": 0.4624,
      "step": 420480
    },
    {
      "epoch": 172.12,
      "grad_norm": 2.1534807682037354,
      "learning_rate": 2.4464943278620774e-07,
      "loss": 0.4363,
      "step": 420490
    },
    {
      "epoch": 172.12,
      "grad_norm": 2.1388542652130127,
      "learning_rate": 2.4457887480078387e-07,
      "loss": 0.4439,
      "step": 420500
    },
    {
      "epoch": 172.13,
      "grad_norm": 1.9775397777557373,
      "learning_rate": 2.4450832643133805e-07,
      "loss": 0.4394,
      "step": 420510
    },
    {
      "epoch": 172.13,
      "grad_norm": 2.190761089324951,
      "learning_rate": 2.4443778767819324e-07,
      "loss": 0.4397,
      "step": 420520
    },
    {
      "epoch": 172.14,
      "grad_norm": 1.7027521133422852,
      "learning_rate": 2.4436725854167235e-07,
      "loss": 0.4347,
      "step": 420530
    },
    {
      "epoch": 172.14,
      "grad_norm": 1.9745819568634033,
      "learning_rate": 2.442967390220988e-07,
      "loss": 0.4625,
      "step": 420540
    },
    {
      "epoch": 172.14,
      "grad_norm": 2.715182304382324,
      "learning_rate": 2.442262291197955e-07,
      "loss": 0.44,
      "step": 420550
    },
    {
      "epoch": 172.15,
      "grad_norm": 2.5922622680664062,
      "learning_rate": 2.441557288350854e-07,
      "loss": 0.4432,
      "step": 420560
    },
    {
      "epoch": 172.15,
      "grad_norm": 1.8962160348892212,
      "learning_rate": 2.4408523816829136e-07,
      "loss": 0.441,
      "step": 420570
    },
    {
      "epoch": 172.16,
      "grad_norm": 2.1176276206970215,
      "learning_rate": 2.4401475711973713e-07,
      "loss": 0.4484,
      "step": 420580
    },
    {
      "epoch": 172.16,
      "grad_norm": 2.285862445831299,
      "learning_rate": 2.4394428568974436e-07,
      "loss": 0.4437,
      "step": 420590
    },
    {
      "epoch": 172.17,
      "grad_norm": 3.006256341934204,
      "learning_rate": 2.4387382387863614e-07,
      "loss": 0.4313,
      "step": 420600
    },
    {
      "epoch": 172.17,
      "grad_norm": 1.904906153678894,
      "learning_rate": 2.438033716867352e-07,
      "loss": 0.4571,
      "step": 420610
    },
    {
      "epoch": 172.17,
      "grad_norm": 1.9670345783233643,
      "learning_rate": 2.437329291143649e-07,
      "loss": 0.4436,
      "step": 420620
    },
    {
      "epoch": 172.18,
      "grad_norm": 2.4029486179351807,
      "learning_rate": 2.436624961618477e-07,
      "loss": 0.4474,
      "step": 420630
    },
    {
      "epoch": 172.18,
      "grad_norm": 1.9517402648925781,
      "learning_rate": 2.4359207282950606e-07,
      "loss": 0.4401,
      "step": 420640
    },
    {
      "epoch": 172.19,
      "grad_norm": 2.1097817420959473,
      "learning_rate": 2.435216591176628e-07,
      "loss": 0.4336,
      "step": 420650
    },
    {
      "epoch": 172.19,
      "grad_norm": 2.0781826972961426,
      "learning_rate": 2.4345125502664006e-07,
      "loss": 0.4478,
      "step": 420660
    },
    {
      "epoch": 172.19,
      "grad_norm": 1.918146014213562,
      "learning_rate": 2.433808605567608e-07,
      "loss": 0.4501,
      "step": 420670
    },
    {
      "epoch": 172.2,
      "grad_norm": 2.0347888469696045,
      "learning_rate": 2.4331047570834713e-07,
      "loss": 0.4406,
      "step": 420680
    },
    {
      "epoch": 172.2,
      "grad_norm": 1.922919750213623,
      "learning_rate": 2.4324010048172176e-07,
      "loss": 0.4455,
      "step": 420690
    },
    {
      "epoch": 172.21,
      "grad_norm": 1.8135267496109009,
      "learning_rate": 2.4316973487720676e-07,
      "loss": 0.4678,
      "step": 420700
    },
    {
      "epoch": 172.21,
      "grad_norm": 2.01320743560791,
      "learning_rate": 2.4309937889512506e-07,
      "loss": 0.4422,
      "step": 420710
    },
    {
      "epoch": 172.21,
      "grad_norm": 1.9440888166427612,
      "learning_rate": 2.430290325357983e-07,
      "loss": 0.4591,
      "step": 420720
    },
    {
      "epoch": 172.22,
      "grad_norm": 2.319507360458374,
      "learning_rate": 2.4295869579954895e-07,
      "loss": 0.4688,
      "step": 420730
    },
    {
      "epoch": 172.22,
      "grad_norm": 2.255859851837158,
      "learning_rate": 2.4288836868669896e-07,
      "loss": 0.4446,
      "step": 420740
    },
    {
      "epoch": 172.23,
      "grad_norm": 2.276341199874878,
      "learning_rate": 2.4281805119757094e-07,
      "loss": 0.4356,
      "step": 420750
    },
    {
      "epoch": 172.23,
      "grad_norm": 2.1633498668670654,
      "learning_rate": 2.427477433324869e-07,
      "loss": 0.4362,
      "step": 420760
    },
    {
      "epoch": 172.23,
      "grad_norm": 2.301448345184326,
      "learning_rate": 2.4267744509176845e-07,
      "loss": 0.4309,
      "step": 420770
    },
    {
      "epoch": 172.24,
      "grad_norm": 1.9781947135925293,
      "learning_rate": 2.426071564757383e-07,
      "loss": 0.4515,
      "step": 420780
    },
    {
      "epoch": 172.24,
      "grad_norm": 1.9731390476226807,
      "learning_rate": 2.425368774847181e-07,
      "loss": 0.4357,
      "step": 420790
    },
    {
      "epoch": 172.25,
      "grad_norm": 2.2193310260772705,
      "learning_rate": 2.4246660811902936e-07,
      "loss": 0.454,
      "step": 420800
    },
    {
      "epoch": 172.25,
      "grad_norm": 1.9482054710388184,
      "learning_rate": 2.4239634837899456e-07,
      "loss": 0.444,
      "step": 420810
    },
    {
      "epoch": 172.26,
      "grad_norm": 1.966139316558838,
      "learning_rate": 2.423260982649358e-07,
      "loss": 0.4514,
      "step": 420820
    },
    {
      "epoch": 172.26,
      "grad_norm": 2.2635316848754883,
      "learning_rate": 2.422558577771743e-07,
      "loss": 0.4305,
      "step": 420830
    },
    {
      "epoch": 172.26,
      "grad_norm": 1.6474392414093018,
      "learning_rate": 2.42185626916032e-07,
      "loss": 0.439,
      "step": 420840
    },
    {
      "epoch": 172.27,
      "grad_norm": 1.7539420127868652,
      "learning_rate": 2.421154056818311e-07,
      "loss": 0.462,
      "step": 420850
    },
    {
      "epoch": 172.27,
      "grad_norm": 1.994384527206421,
      "learning_rate": 2.420451940748923e-07,
      "loss": 0.4371,
      "step": 420860
    },
    {
      "epoch": 172.28,
      "grad_norm": 2.7827885150909424,
      "learning_rate": 2.4197499209553784e-07,
      "loss": 0.4448,
      "step": 420870
    },
    {
      "epoch": 172.28,
      "grad_norm": 2.137225389480591,
      "learning_rate": 2.419047997440892e-07,
      "loss": 0.4427,
      "step": 420880
    },
    {
      "epoch": 172.28,
      "grad_norm": 2.439175844192505,
      "learning_rate": 2.418346170208676e-07,
      "loss": 0.4374,
      "step": 420890
    },
    {
      "epoch": 172.29,
      "grad_norm": 2.1888427734375,
      "learning_rate": 2.417644439261953e-07,
      "loss": 0.4462,
      "step": 420900
    },
    {
      "epoch": 172.29,
      "grad_norm": 1.9536060094833374,
      "learning_rate": 2.416942804603931e-07,
      "loss": 0.4392,
      "step": 420910
    },
    {
      "epoch": 172.3,
      "grad_norm": 2.1090645790100098,
      "learning_rate": 2.416241266237824e-07,
      "loss": 0.4548,
      "step": 420920
    },
    {
      "epoch": 172.3,
      "grad_norm": 1.976122498512268,
      "learning_rate": 2.41553982416685e-07,
      "loss": 0.4313,
      "step": 420930
    },
    {
      "epoch": 172.3,
      "grad_norm": 1.746711254119873,
      "learning_rate": 2.41483847839422e-07,
      "loss": 0.4216,
      "step": 420940
    },
    {
      "epoch": 172.31,
      "grad_norm": 2.1170713901519775,
      "learning_rate": 2.414137228923145e-07,
      "loss": 0.4495,
      "step": 420950
    },
    {
      "epoch": 172.31,
      "grad_norm": 2.007030487060547,
      "learning_rate": 2.413436075756842e-07,
      "loss": 0.4383,
      "step": 420960
    },
    {
      "epoch": 172.32,
      "grad_norm": 2.0751380920410156,
      "learning_rate": 2.412735018898516e-07,
      "loss": 0.44,
      "step": 420970
    },
    {
      "epoch": 172.32,
      "grad_norm": 2.068002223968506,
      "learning_rate": 2.412034058351386e-07,
      "loss": 0.4413,
      "step": 420980
    },
    {
      "epoch": 172.33,
      "grad_norm": 2.29111647605896,
      "learning_rate": 2.4113331941186555e-07,
      "loss": 0.4296,
      "step": 420990
    },
    {
      "epoch": 172.33,
      "grad_norm": 2.32671856880188,
      "learning_rate": 2.410632426203543e-07,
      "loss": 0.4557,
      "step": 421000
    },
    {
      "epoch": 172.33,
      "grad_norm": 1.7832380533218384,
      "learning_rate": 2.409931754609251e-07,
      "loss": 0.4439,
      "step": 421010
    },
    {
      "epoch": 172.34,
      "grad_norm": 1.9241411685943604,
      "learning_rate": 2.4092311793389945e-07,
      "loss": 0.4412,
      "step": 421020
    },
    {
      "epoch": 172.34,
      "grad_norm": 1.9436848163604736,
      "learning_rate": 2.4085307003959806e-07,
      "loss": 0.4344,
      "step": 421030
    },
    {
      "epoch": 172.35,
      "grad_norm": 2.143615484237671,
      "learning_rate": 2.407830317783418e-07,
      "loss": 0.4532,
      "step": 421040
    },
    {
      "epoch": 172.35,
      "grad_norm": 2.3538949489593506,
      "learning_rate": 2.4071300315045156e-07,
      "loss": 0.443,
      "step": 421050
    },
    {
      "epoch": 172.35,
      "grad_norm": 1.7952349185943604,
      "learning_rate": 2.406429841562481e-07,
      "loss": 0.4494,
      "step": 421060
    },
    {
      "epoch": 172.36,
      "grad_norm": 2.091610908508301,
      "learning_rate": 2.40572974796052e-07,
      "loss": 0.4478,
      "step": 421070
    },
    {
      "epoch": 172.36,
      "grad_norm": 2.196040391921997,
      "learning_rate": 2.405029750701842e-07,
      "loss": 0.4419,
      "step": 421080
    },
    {
      "epoch": 172.37,
      "grad_norm": 2.085692882537842,
      "learning_rate": 2.4043298497896544e-07,
      "loss": 0.4382,
      "step": 421090
    },
    {
      "epoch": 172.37,
      "grad_norm": 2.1227235794067383,
      "learning_rate": 2.4036300452271606e-07,
      "loss": 0.455,
      "step": 421100
    },
    {
      "epoch": 172.37,
      "grad_norm": 2.34163761138916,
      "learning_rate": 2.4029303370175693e-07,
      "loss": 0.4436,
      "step": 421110
    },
    {
      "epoch": 172.38,
      "grad_norm": 2.258336305618286,
      "learning_rate": 2.402230725164083e-07,
      "loss": 0.4393,
      "step": 421120
    },
    {
      "epoch": 172.38,
      "grad_norm": 1.5093809366226196,
      "learning_rate": 2.401531209669908e-07,
      "loss": 0.437,
      "step": 421130
    },
    {
      "epoch": 172.39,
      "grad_norm": 2.090369939804077,
      "learning_rate": 2.4008317905382477e-07,
      "loss": 0.4656,
      "step": 421140
    },
    {
      "epoch": 172.39,
      "grad_norm": 2.420339822769165,
      "learning_rate": 2.4001324677723046e-07,
      "loss": 0.435,
      "step": 421150
    },
    {
      "epoch": 172.39,
      "grad_norm": 1.9506257772445679,
      "learning_rate": 2.3994332413752925e-07,
      "loss": 0.4654,
      "step": 421160
    },
    {
      "epoch": 172.4,
      "grad_norm": 2.1255836486816406,
      "learning_rate": 2.398734111350399e-07,
      "loss": 0.4324,
      "step": 421170
    },
    {
      "epoch": 172.4,
      "grad_norm": 1.890632152557373,
      "learning_rate": 2.3980350777008295e-07,
      "loss": 0.4559,
      "step": 421180
    },
    {
      "epoch": 172.41,
      "grad_norm": 2.243018388748169,
      "learning_rate": 2.397336140429796e-07,
      "loss": 0.4388,
      "step": 421190
    },
    {
      "epoch": 172.41,
      "grad_norm": 2.003767251968384,
      "learning_rate": 2.396637299540493e-07,
      "loss": 0.4522,
      "step": 421200
    },
    {
      "epoch": 172.42,
      "grad_norm": 2.0145347118377686,
      "learning_rate": 2.395938555036124e-07,
      "loss": 0.4668,
      "step": 421210
    },
    {
      "epoch": 172.42,
      "grad_norm": 2.156209707260132,
      "learning_rate": 2.395239906919889e-07,
      "loss": 0.4309,
      "step": 421220
    },
    {
      "epoch": 172.42,
      "grad_norm": 2.0254523754119873,
      "learning_rate": 2.3945413551949897e-07,
      "loss": 0.4491,
      "step": 421230
    },
    {
      "epoch": 172.43,
      "grad_norm": 2.124061107635498,
      "learning_rate": 2.393842899864626e-07,
      "loss": 0.4668,
      "step": 421240
    },
    {
      "epoch": 172.43,
      "grad_norm": 2.277711868286133,
      "learning_rate": 2.393144540931995e-07,
      "loss": 0.4432,
      "step": 421250
    },
    {
      "epoch": 172.44,
      "grad_norm": 2.0588176250457764,
      "learning_rate": 2.392446278400296e-07,
      "loss": 0.4804,
      "step": 421260
    },
    {
      "epoch": 172.44,
      "grad_norm": 2.30385160446167,
      "learning_rate": 2.3917481122727315e-07,
      "loss": 0.4391,
      "step": 421270
    },
    {
      "epoch": 172.44,
      "grad_norm": 2.360849142074585,
      "learning_rate": 2.3910500425524947e-07,
      "loss": 0.4438,
      "step": 421280
    },
    {
      "epoch": 172.45,
      "grad_norm": 2.4204957485198975,
      "learning_rate": 2.390352069242791e-07,
      "loss": 0.4327,
      "step": 421290
    },
    {
      "epoch": 172.45,
      "grad_norm": 1.90748929977417,
      "learning_rate": 2.3896541923468083e-07,
      "loss": 0.4556,
      "step": 421300
    },
    {
      "epoch": 172.46,
      "grad_norm": 2.4037904739379883,
      "learning_rate": 2.3889564118677465e-07,
      "loss": 0.4633,
      "step": 421310
    },
    {
      "epoch": 172.46,
      "grad_norm": 1.7213696241378784,
      "learning_rate": 2.388258727808804e-07,
      "loss": 0.425,
      "step": 421320
    },
    {
      "epoch": 172.46,
      "grad_norm": 2.181626796722412,
      "learning_rate": 2.3875611401731763e-07,
      "loss": 0.4524,
      "step": 421330
    },
    {
      "epoch": 172.47,
      "grad_norm": 2.218262195587158,
      "learning_rate": 2.386863648964059e-07,
      "loss": 0.4459,
      "step": 421340
    },
    {
      "epoch": 172.47,
      "grad_norm": 2.036400079727173,
      "learning_rate": 2.3861662541846465e-07,
      "loss": 0.4417,
      "step": 421350
    },
    {
      "epoch": 172.48,
      "grad_norm": 2.0116169452667236,
      "learning_rate": 2.385468955838132e-07,
      "loss": 0.4367,
      "step": 421360
    },
    {
      "epoch": 172.48,
      "grad_norm": 2.0238125324249268,
      "learning_rate": 2.384771753927709e-07,
      "loss": 0.4728,
      "step": 421370
    },
    {
      "epoch": 172.48,
      "grad_norm": 2.247340679168701,
      "learning_rate": 2.3840746484565772e-07,
      "loss": 0.4434,
      "step": 421380
    },
    {
      "epoch": 172.49,
      "grad_norm": 1.7527579069137573,
      "learning_rate": 2.3833776394279244e-07,
      "loss": 0.4294,
      "step": 421390
    },
    {
      "epoch": 172.49,
      "grad_norm": 2.058615207672119,
      "learning_rate": 2.3826807268449484e-07,
      "loss": 0.4582,
      "step": 421400
    },
    {
      "epoch": 172.5,
      "grad_norm": 6.81758451461792,
      "learning_rate": 2.381983910710837e-07,
      "loss": 0.4451,
      "step": 421410
    },
    {
      "epoch": 172.5,
      "grad_norm": 2.188539743423462,
      "learning_rate": 2.3812871910287857e-07,
      "loss": 0.4372,
      "step": 421420
    },
    {
      "epoch": 172.51,
      "grad_norm": 2.2214269638061523,
      "learning_rate": 2.3805905678019868e-07,
      "loss": 0.4504,
      "step": 421430
    },
    {
      "epoch": 172.51,
      "grad_norm": 1.7812416553497314,
      "learning_rate": 2.3798940410336254e-07,
      "loss": 0.4594,
      "step": 421440
    },
    {
      "epoch": 172.51,
      "grad_norm": 2.3206207752227783,
      "learning_rate": 2.3791976107268973e-07,
      "loss": 0.4503,
      "step": 421450
    },
    {
      "epoch": 172.52,
      "grad_norm": 1.698170781135559,
      "learning_rate": 2.3785012768849894e-07,
      "loss": 0.453,
      "step": 421460
    },
    {
      "epoch": 172.52,
      "grad_norm": 1.9925122261047363,
      "learning_rate": 2.3778050395110924e-07,
      "loss": 0.4514,
      "step": 421470
    },
    {
      "epoch": 172.53,
      "grad_norm": 2.153045415878296,
      "learning_rate": 2.3771088986083985e-07,
      "loss": 0.4185,
      "step": 421480
    },
    {
      "epoch": 172.53,
      "grad_norm": 1.7579606771469116,
      "learning_rate": 2.3764128541800934e-07,
      "loss": 0.4315,
      "step": 421490
    },
    {
      "epoch": 172.53,
      "grad_norm": 2.0164849758148193,
      "learning_rate": 2.3757169062293664e-07,
      "loss": 0.4421,
      "step": 421500
    },
    {
      "epoch": 172.54,
      "grad_norm": 2.123975992202759,
      "learning_rate": 2.3750210547594058e-07,
      "loss": 0.4487,
      "step": 421510
    },
    {
      "epoch": 172.54,
      "grad_norm": 1.8970636129379272,
      "learning_rate": 2.3743252997734012e-07,
      "loss": 0.4372,
      "step": 421520
    },
    {
      "epoch": 172.55,
      "grad_norm": 2.097682237625122,
      "learning_rate": 2.373629641274538e-07,
      "loss": 0.4326,
      "step": 421530
    },
    {
      "epoch": 172.55,
      "grad_norm": 1.9101563692092896,
      "learning_rate": 2.3729340792660007e-07,
      "loss": 0.4362,
      "step": 421540
    },
    {
      "epoch": 172.55,
      "grad_norm": 2.151165246963501,
      "learning_rate": 2.3722386137509748e-07,
      "loss": 0.4306,
      "step": 421550
    },
    {
      "epoch": 172.56,
      "grad_norm": 1.9752590656280518,
      "learning_rate": 2.3715432447326525e-07,
      "loss": 0.4393,
      "step": 421560
    },
    {
      "epoch": 172.56,
      "grad_norm": 2.001509428024292,
      "learning_rate": 2.3708479722142192e-07,
      "loss": 0.4281,
      "step": 421570
    },
    {
      "epoch": 172.57,
      "grad_norm": 1.9542793035507202,
      "learning_rate": 2.3701527961988517e-07,
      "loss": 0.4232,
      "step": 421580
    },
    {
      "epoch": 172.57,
      "grad_norm": 2.332338333129883,
      "learning_rate": 2.3694577166897405e-07,
      "loss": 0.4461,
      "step": 421590
    },
    {
      "epoch": 172.57,
      "grad_norm": 2.365967273712158,
      "learning_rate": 2.3687627336900677e-07,
      "loss": 0.4351,
      "step": 421600
    },
    {
      "epoch": 172.58,
      "grad_norm": 2.087751865386963,
      "learning_rate": 2.3680678472030183e-07,
      "loss": 0.4498,
      "step": 421610
    },
    {
      "epoch": 172.58,
      "grad_norm": 1.9056850671768188,
      "learning_rate": 2.3673730572317747e-07,
      "loss": 0.4321,
      "step": 421620
    },
    {
      "epoch": 172.59,
      "grad_norm": 1.8358505964279175,
      "learning_rate": 2.366678363779519e-07,
      "loss": 0.4399,
      "step": 421630
    },
    {
      "epoch": 172.59,
      "grad_norm": 1.8165395259857178,
      "learning_rate": 2.365983766849434e-07,
      "loss": 0.4478,
      "step": 421640
    },
    {
      "epoch": 172.6,
      "grad_norm": 2.4612746238708496,
      "learning_rate": 2.365289266444699e-07,
      "loss": 0.4327,
      "step": 421650
    },
    {
      "epoch": 172.6,
      "grad_norm": 2.1366801261901855,
      "learning_rate": 2.364594862568502e-07,
      "loss": 0.4485,
      "step": 421660
    },
    {
      "epoch": 172.6,
      "grad_norm": 2.432421922683716,
      "learning_rate": 2.3639005552240172e-07,
      "loss": 0.4483,
      "step": 421670
    },
    {
      "epoch": 172.61,
      "grad_norm": 3.1072747707366943,
      "learning_rate": 2.3632063444144296e-07,
      "loss": 0.4547,
      "step": 421680
    },
    {
      "epoch": 172.61,
      "grad_norm": 2.554314613342285,
      "learning_rate": 2.3625122301429138e-07,
      "loss": 0.4586,
      "step": 421690
    },
    {
      "epoch": 172.62,
      "grad_norm": 2.091482400894165,
      "learning_rate": 2.3618182124126574e-07,
      "loss": 0.4435,
      "step": 421700
    },
    {
      "epoch": 172.62,
      "grad_norm": 2.3058691024780273,
      "learning_rate": 2.3611242912268323e-07,
      "loss": 0.4326,
      "step": 421710
    },
    {
      "epoch": 172.62,
      "grad_norm": 1.8295676708221436,
      "learning_rate": 2.3604304665886235e-07,
      "loss": 0.4513,
      "step": 421720
    },
    {
      "epoch": 172.63,
      "grad_norm": 1.9645986557006836,
      "learning_rate": 2.359736738501203e-07,
      "loss": 0.4488,
      "step": 421730
    },
    {
      "epoch": 172.63,
      "grad_norm": 1.8267463445663452,
      "learning_rate": 2.359043106967753e-07,
      "loss": 0.4567,
      "step": 421740
    },
    {
      "epoch": 172.64,
      "grad_norm": 2.0331196784973145,
      "learning_rate": 2.3583495719914507e-07,
      "loss": 0.4586,
      "step": 421750
    },
    {
      "epoch": 172.64,
      "grad_norm": 1.9640345573425293,
      "learning_rate": 2.3576561335754705e-07,
      "loss": 0.4466,
      "step": 421760
    },
    {
      "epoch": 172.64,
      "grad_norm": 2.284344434738159,
      "learning_rate": 2.3569627917229897e-07,
      "loss": 0.4312,
      "step": 421770
    },
    {
      "epoch": 172.65,
      "grad_norm": 2.1852173805236816,
      "learning_rate": 2.3562695464371856e-07,
      "loss": 0.4552,
      "step": 421780
    },
    {
      "epoch": 172.65,
      "grad_norm": 2.1615633964538574,
      "learning_rate": 2.3555763977212348e-07,
      "loss": 0.4408,
      "step": 421790
    },
    {
      "epoch": 172.66,
      "grad_norm": 1.8461869955062866,
      "learning_rate": 2.35488334557831e-07,
      "loss": 0.4534,
      "step": 421800
    },
    {
      "epoch": 172.66,
      "grad_norm": 2.112166404724121,
      "learning_rate": 2.3541903900115878e-07,
      "loss": 0.4253,
      "step": 421810
    },
    {
      "epoch": 172.66,
      "grad_norm": 1.803990364074707,
      "learning_rate": 2.3534975310242405e-07,
      "loss": 0.436,
      "step": 421820
    },
    {
      "epoch": 172.67,
      "grad_norm": 1.9436297416687012,
      "learning_rate": 2.3528047686194423e-07,
      "loss": 0.4415,
      "step": 421830
    },
    {
      "epoch": 172.67,
      "grad_norm": 2.336559295654297,
      "learning_rate": 2.3521121028003703e-07,
      "loss": 0.4417,
      "step": 421840
    },
    {
      "epoch": 172.68,
      "grad_norm": 2.5518534183502197,
      "learning_rate": 2.3514195335701943e-07,
      "loss": 0.4558,
      "step": 421850
    },
    {
      "epoch": 172.68,
      "grad_norm": 2.144235610961914,
      "learning_rate": 2.3507270609320859e-07,
      "loss": 0.4232,
      "step": 421860
    },
    {
      "epoch": 172.69,
      "grad_norm": 2.4613919258117676,
      "learning_rate": 2.3500346848892198e-07,
      "loss": 0.4507,
      "step": 421870
    },
    {
      "epoch": 172.69,
      "grad_norm": 2.375788688659668,
      "learning_rate": 2.3493424054447704e-07,
      "loss": 0.4407,
      "step": 421880
    },
    {
      "epoch": 172.69,
      "grad_norm": 1.6926486492156982,
      "learning_rate": 2.3486502226018992e-07,
      "loss": 0.43,
      "step": 421890
    },
    {
      "epoch": 172.7,
      "grad_norm": 2.275167942047119,
      "learning_rate": 2.3479581363637862e-07,
      "loss": 0.4498,
      "step": 421900
    },
    {
      "epoch": 172.7,
      "grad_norm": 2.6165719032287598,
      "learning_rate": 2.3472661467335952e-07,
      "loss": 0.447,
      "step": 421910
    },
    {
      "epoch": 172.71,
      "grad_norm": 2.2088730335235596,
      "learning_rate": 2.3465742537145011e-07,
      "loss": 0.4445,
      "step": 421920
    },
    {
      "epoch": 172.71,
      "grad_norm": 1.8886215686798096,
      "learning_rate": 2.3458824573096654e-07,
      "loss": 0.4329,
      "step": 421930
    },
    {
      "epoch": 172.71,
      "grad_norm": 1.871324062347412,
      "learning_rate": 2.34519075752227e-07,
      "loss": 0.4312,
      "step": 421940
    },
    {
      "epoch": 172.72,
      "grad_norm": 2.596982479095459,
      "learning_rate": 2.3444991543554743e-07,
      "loss": 0.4559,
      "step": 421950
    },
    {
      "epoch": 172.72,
      "grad_norm": 1.9410524368286133,
      "learning_rate": 2.3438076478124501e-07,
      "loss": 0.4475,
      "step": 421960
    },
    {
      "epoch": 172.73,
      "grad_norm": 1.8001961708068848,
      "learning_rate": 2.343116237896364e-07,
      "loss": 0.4337,
      "step": 421970
    },
    {
      "epoch": 172.73,
      "grad_norm": 2.0235376358032227,
      "learning_rate": 2.3424249246103828e-07,
      "loss": 0.4422,
      "step": 421980
    },
    {
      "epoch": 172.73,
      "grad_norm": 2.6120171546936035,
      "learning_rate": 2.3417337079576734e-07,
      "loss": 0.4571,
      "step": 421990
    },
    {
      "epoch": 172.74,
      "grad_norm": 2.188159465789795,
      "learning_rate": 2.3410425879414027e-07,
      "loss": 0.4544,
      "step": 422000
    },
    {
      "epoch": 172.74,
      "grad_norm": 1.8373746871948242,
      "learning_rate": 2.3403515645647394e-07,
      "loss": 0.4505,
      "step": 422010
    },
    {
      "epoch": 172.75,
      "grad_norm": 2.446852922439575,
      "learning_rate": 2.339660637830843e-07,
      "loss": 0.4394,
      "step": 422020
    },
    {
      "epoch": 172.75,
      "grad_norm": 2.1402876377105713,
      "learning_rate": 2.3389698077428798e-07,
      "loss": 0.4449,
      "step": 422030
    },
    {
      "epoch": 172.75,
      "grad_norm": 2.036080837249756,
      "learning_rate": 2.3382790743040171e-07,
      "loss": 0.4475,
      "step": 422040
    },
    {
      "epoch": 172.76,
      "grad_norm": 1.8010506629943848,
      "learning_rate": 2.337588437517416e-07,
      "loss": 0.4501,
      "step": 422050
    },
    {
      "epoch": 172.76,
      "grad_norm": 2.0882465839385986,
      "learning_rate": 2.3368978973862432e-07,
      "loss": 0.4549,
      "step": 422060
    },
    {
      "epoch": 172.77,
      "grad_norm": 1.8769547939300537,
      "learning_rate": 2.3362074539136633e-07,
      "loss": 0.4577,
      "step": 422070
    },
    {
      "epoch": 172.77,
      "grad_norm": 1.912198781967163,
      "learning_rate": 2.3355171071028322e-07,
      "loss": 0.4429,
      "step": 422080
    },
    {
      "epoch": 172.78,
      "grad_norm": 1.826832890510559,
      "learning_rate": 2.3348268569569193e-07,
      "loss": 0.4369,
      "step": 422090
    },
    {
      "epoch": 172.78,
      "grad_norm": 2.018747329711914,
      "learning_rate": 2.3341367034790835e-07,
      "loss": 0.4425,
      "step": 422100
    },
    {
      "epoch": 172.78,
      "grad_norm": 1.7173428535461426,
      "learning_rate": 2.3334466466724864e-07,
      "loss": 0.4269,
      "step": 422110
    },
    {
      "epoch": 172.79,
      "grad_norm": 2.1399788856506348,
      "learning_rate": 2.3327566865402846e-07,
      "loss": 0.426,
      "step": 422120
    },
    {
      "epoch": 172.79,
      "grad_norm": 1.9021663665771484,
      "learning_rate": 2.3320668230856496e-07,
      "loss": 0.4446,
      "step": 422130
    },
    {
      "epoch": 172.8,
      "grad_norm": 2.379563808441162,
      "learning_rate": 2.3313770563117327e-07,
      "loss": 0.4281,
      "step": 422140
    },
    {
      "epoch": 172.8,
      "grad_norm": 1.9806371927261353,
      "learning_rate": 2.3306873862217035e-07,
      "loss": 0.447,
      "step": 422150
    },
    {
      "epoch": 172.8,
      "grad_norm": 2.107079029083252,
      "learning_rate": 2.3299978128187074e-07,
      "loss": 0.4398,
      "step": 422160
    },
    {
      "epoch": 172.81,
      "grad_norm": 1.5895769596099854,
      "learning_rate": 2.3293083361059115e-07,
      "loss": 0.4469,
      "step": 422170
    },
    {
      "epoch": 172.81,
      "grad_norm": 1.9916774034500122,
      "learning_rate": 2.3286189560864722e-07,
      "loss": 0.4494,
      "step": 422180
    },
    {
      "epoch": 172.82,
      "grad_norm": 1.9755265712738037,
      "learning_rate": 2.3279296727635482e-07,
      "loss": 0.4363,
      "step": 422190
    },
    {
      "epoch": 172.82,
      "grad_norm": 1.7792298793792725,
      "learning_rate": 2.3272404861402956e-07,
      "loss": 0.4418,
      "step": 422200
    },
    {
      "epoch": 172.82,
      "grad_norm": 1.7749842405319214,
      "learning_rate": 2.3265513962198713e-07,
      "loss": 0.4344,
      "step": 422210
    },
    {
      "epoch": 172.83,
      "grad_norm": 1.8677339553833008,
      "learning_rate": 2.3258624030054342e-07,
      "loss": 0.4251,
      "step": 422220
    },
    {
      "epoch": 172.83,
      "grad_norm": 2.0584049224853516,
      "learning_rate": 2.32517350650014e-07,
      "loss": 0.4171,
      "step": 422230
    },
    {
      "epoch": 172.84,
      "grad_norm": 1.7208143472671509,
      "learning_rate": 2.3244847067071457e-07,
      "loss": 0.451,
      "step": 422240
    },
    {
      "epoch": 172.84,
      "grad_norm": 2.305143356323242,
      "learning_rate": 2.323796003629602e-07,
      "loss": 0.4441,
      "step": 422250
    },
    {
      "epoch": 172.84,
      "grad_norm": 2.065917730331421,
      "learning_rate": 2.3231073972706653e-07,
      "loss": 0.4325,
      "step": 422260
    },
    {
      "epoch": 172.85,
      "grad_norm": 2.0680038928985596,
      "learning_rate": 2.322418887633495e-07,
      "loss": 0.4461,
      "step": 422270
    },
    {
      "epoch": 172.85,
      "grad_norm": 1.8246309757232666,
      "learning_rate": 2.321730474721239e-07,
      "loss": 0.4417,
      "step": 422280
    },
    {
      "epoch": 172.86,
      "grad_norm": 1.8529301881790161,
      "learning_rate": 2.3210421585370535e-07,
      "loss": 0.4354,
      "step": 422290
    },
    {
      "epoch": 172.86,
      "grad_norm": 2.4338414669036865,
      "learning_rate": 2.3203539390840906e-07,
      "loss": 0.445,
      "step": 422300
    },
    {
      "epoch": 172.87,
      "grad_norm": 2.046898365020752,
      "learning_rate": 2.319665816365503e-07,
      "loss": 0.4342,
      "step": 422310
    },
    {
      "epoch": 172.87,
      "grad_norm": 1.895859956741333,
      "learning_rate": 2.3189777903844425e-07,
      "loss": 0.4385,
      "step": 422320
    },
    {
      "epoch": 172.87,
      "grad_norm": 2.432399272918701,
      "learning_rate": 2.318289861144065e-07,
      "loss": 0.4464,
      "step": 422330
    },
    {
      "epoch": 172.88,
      "grad_norm": 1.9075452089309692,
      "learning_rate": 2.3176020286475166e-07,
      "loss": 0.4814,
      "step": 422340
    },
    {
      "epoch": 172.88,
      "grad_norm": 4.921926498413086,
      "learning_rate": 2.3169142928979485e-07,
      "loss": 0.4544,
      "step": 422350
    },
    {
      "epoch": 172.89,
      "grad_norm": 2.1487772464752197,
      "learning_rate": 2.3162266538985143e-07,
      "loss": 0.4527,
      "step": 422360
    },
    {
      "epoch": 172.89,
      "grad_norm": 1.9292783737182617,
      "learning_rate": 2.3155391116523625e-07,
      "loss": 0.4465,
      "step": 422370
    },
    {
      "epoch": 172.89,
      "grad_norm": 2.1549057960510254,
      "learning_rate": 2.3148516661626393e-07,
      "loss": 0.4554,
      "step": 422380
    },
    {
      "epoch": 172.9,
      "grad_norm": 2.2882769107818604,
      "learning_rate": 2.3141643174325006e-07,
      "loss": 0.4575,
      "step": 422390
    },
    {
      "epoch": 172.9,
      "grad_norm": 1.7754210233688354,
      "learning_rate": 2.3134770654650903e-07,
      "loss": 0.4541,
      "step": 422400
    },
    {
      "epoch": 172.91,
      "grad_norm": 2.0014967918395996,
      "learning_rate": 2.3127899102635563e-07,
      "loss": 0.4276,
      "step": 422410
    },
    {
      "epoch": 172.91,
      "grad_norm": 2.444931745529175,
      "learning_rate": 2.3121028518310498e-07,
      "loss": 0.4248,
      "step": 422420
    },
    {
      "epoch": 172.91,
      "grad_norm": 1.6620455980300903,
      "learning_rate": 2.3114158901707146e-07,
      "loss": 0.4402,
      "step": 422430
    },
    {
      "epoch": 172.92,
      "grad_norm": 2.1232547760009766,
      "learning_rate": 2.3107290252856988e-07,
      "loss": 0.468,
      "step": 422440
    },
    {
      "epoch": 172.92,
      "grad_norm": 2.0479462146759033,
      "learning_rate": 2.3100422571791484e-07,
      "loss": 0.4498,
      "step": 422450
    },
    {
      "epoch": 172.93,
      "grad_norm": 2.158151626586914,
      "learning_rate": 2.3093555858542146e-07,
      "loss": 0.4395,
      "step": 422460
    },
    {
      "epoch": 172.93,
      "grad_norm": 2.202817916870117,
      "learning_rate": 2.3086690113140357e-07,
      "loss": 0.4404,
      "step": 422470
    },
    {
      "epoch": 172.93,
      "grad_norm": 2.0764760971069336,
      "learning_rate": 2.307982533561757e-07,
      "loss": 0.4643,
      "step": 422480
    },
    {
      "epoch": 172.94,
      "grad_norm": 2.2297372817993164,
      "learning_rate": 2.3072961526005278e-07,
      "loss": 0.443,
      "step": 422490
    },
    {
      "epoch": 172.94,
      "grad_norm": 1.99703049659729,
      "learning_rate": 2.306609868433483e-07,
      "loss": 0.4512,
      "step": 422500
    },
    {
      "epoch": 172.95,
      "grad_norm": 2.4657113552093506,
      "learning_rate": 2.3059236810637796e-07,
      "loss": 0.4465,
      "step": 422510
    },
    {
      "epoch": 172.95,
      "grad_norm": 2.2279093265533447,
      "learning_rate": 2.3052375904945548e-07,
      "loss": 0.4281,
      "step": 422520
    },
    {
      "epoch": 172.96,
      "grad_norm": 1.9243781566619873,
      "learning_rate": 2.3045515967289502e-07,
      "loss": 0.4542,
      "step": 422530
    },
    {
      "epoch": 172.96,
      "grad_norm": 2.09751033782959,
      "learning_rate": 2.303865699770111e-07,
      "loss": 0.4549,
      "step": 422540
    },
    {
      "epoch": 172.96,
      "grad_norm": 2.245683431625366,
      "learning_rate": 2.303179899621176e-07,
      "loss": 0.4457,
      "step": 422550
    },
    {
      "epoch": 172.97,
      "grad_norm": 1.977403998374939,
      "learning_rate": 2.3024941962852878e-07,
      "loss": 0.4303,
      "step": 422560
    },
    {
      "epoch": 172.97,
      "grad_norm": 2.0004019737243652,
      "learning_rate": 2.30180858976559e-07,
      "loss": 0.4319,
      "step": 422570
    },
    {
      "epoch": 172.98,
      "grad_norm": 3.873471736907959,
      "learning_rate": 2.3011230800652212e-07,
      "loss": 0.4465,
      "step": 422580
    },
    {
      "epoch": 172.98,
      "grad_norm": 2.0545763969421387,
      "learning_rate": 2.3004376671873188e-07,
      "loss": 0.4526,
      "step": 422590
    },
    {
      "epoch": 172.98,
      "grad_norm": 1.9669650793075562,
      "learning_rate": 2.2997523511350318e-07,
      "loss": 0.4492,
      "step": 422600
    },
    {
      "epoch": 172.99,
      "grad_norm": 1.7862930297851562,
      "learning_rate": 2.2990671319114904e-07,
      "loss": 0.4436,
      "step": 422610
    },
    {
      "epoch": 172.99,
      "grad_norm": 1.9904993772506714,
      "learning_rate": 2.298382009519835e-07,
      "loss": 0.4356,
      "step": 422620
    },
    {
      "epoch": 173.0,
      "grad_norm": 1.828607201576233,
      "learning_rate": 2.2976969839632043e-07,
      "loss": 0.4521,
      "step": 422630
    },
    {
      "epoch": 173.0,
      "eval_loss": 0.44374480843544006,
      "eval_runtime": 52.3061,
      "eval_samples_per_second": 65.939,
      "eval_steps_per_second": 8.259,
      "step": 422639
    },
    {
      "epoch": 173.0,
      "grad_norm": 2.4793343544006348,
      "learning_rate": 2.2970120552447386e-07,
      "loss": 0.4465,
      "step": 422640
    },
    {
      "epoch": 173.0,
      "grad_norm": 1.9158926010131836,
      "learning_rate": 2.2963272233675732e-07,
      "loss": 0.4489,
      "step": 422650
    },
    {
      "epoch": 173.01,
      "grad_norm": 2.0744283199310303,
      "learning_rate": 2.2956424883348468e-07,
      "loss": 0.4425,
      "step": 422660
    },
    {
      "epoch": 173.01,
      "grad_norm": 2.2842516899108887,
      "learning_rate": 2.2949578501496947e-07,
      "loss": 0.4504,
      "step": 422670
    },
    {
      "epoch": 173.02,
      "grad_norm": 1.9184905290603638,
      "learning_rate": 2.2942733088152547e-07,
      "loss": 0.4448,
      "step": 422680
    },
    {
      "epoch": 173.02,
      "grad_norm": 2.186075448989868,
      "learning_rate": 2.2935888643346574e-07,
      "loss": 0.4437,
      "step": 422690
    },
    {
      "epoch": 173.02,
      "grad_norm": 2.0065979957580566,
      "learning_rate": 2.2929045167110463e-07,
      "loss": 0.4323,
      "step": 422700
    },
    {
      "epoch": 173.03,
      "grad_norm": 2.354696035385132,
      "learning_rate": 2.2922202659475512e-07,
      "loss": 0.4324,
      "step": 422710
    },
    {
      "epoch": 173.03,
      "grad_norm": 2.747593641281128,
      "learning_rate": 2.2915361120473107e-07,
      "loss": 0.4382,
      "step": 422720
    },
    {
      "epoch": 173.04,
      "grad_norm": 2.086251735687256,
      "learning_rate": 2.2908520550134574e-07,
      "loss": 0.4377,
      "step": 422730
    },
    {
      "epoch": 173.04,
      "grad_norm": 1.9922820329666138,
      "learning_rate": 2.290168094849122e-07,
      "loss": 0.4458,
      "step": 422740
    },
    {
      "epoch": 173.05,
      "grad_norm": 2.2754151821136475,
      "learning_rate": 2.2894842315574368e-07,
      "loss": 0.4623,
      "step": 422750
    },
    {
      "epoch": 173.05,
      "grad_norm": 2.0022928714752197,
      "learning_rate": 2.2888004651415377e-07,
      "loss": 0.4352,
      "step": 422760
    },
    {
      "epoch": 173.05,
      "grad_norm": 1.9646753072738647,
      "learning_rate": 2.288116795604555e-07,
      "loss": 0.4755,
      "step": 422770
    },
    {
      "epoch": 173.06,
      "grad_norm": 2.059979200363159,
      "learning_rate": 2.2874332229496215e-07,
      "loss": 0.4708,
      "step": 422780
    },
    {
      "epoch": 173.06,
      "grad_norm": 1.6348302364349365,
      "learning_rate": 2.2867497471798675e-07,
      "loss": 0.4434,
      "step": 422790
    },
    {
      "epoch": 173.07,
      "grad_norm": 2.172919273376465,
      "learning_rate": 2.286066368298426e-07,
      "loss": 0.4126,
      "step": 422800
    },
    {
      "epoch": 173.07,
      "grad_norm": 1.818547010421753,
      "learning_rate": 2.2853830863084275e-07,
      "loss": 0.4585,
      "step": 422810
    },
    {
      "epoch": 173.07,
      "grad_norm": 1.9896857738494873,
      "learning_rate": 2.2846999012129993e-07,
      "loss": 0.4438,
      "step": 422820
    },
    {
      "epoch": 173.08,
      "grad_norm": 2.1464717388153076,
      "learning_rate": 2.284016813015272e-07,
      "loss": 0.4285,
      "step": 422830
    },
    {
      "epoch": 173.08,
      "grad_norm": 1.9926207065582275,
      "learning_rate": 2.2833338217183757e-07,
      "loss": 0.4421,
      "step": 422840
    },
    {
      "epoch": 173.09,
      "grad_norm": 2.3059661388397217,
      "learning_rate": 2.2826509273254382e-07,
      "loss": 0.4505,
      "step": 422850
    },
    {
      "epoch": 173.09,
      "grad_norm": 1.7499698400497437,
      "learning_rate": 2.28196812983959e-07,
      "loss": 0.4391,
      "step": 422860
    },
    {
      "epoch": 173.09,
      "grad_norm": 2.084429979324341,
      "learning_rate": 2.2812854292639533e-07,
      "loss": 0.455,
      "step": 422870
    },
    {
      "epoch": 173.1,
      "grad_norm": 4.174048900604248,
      "learning_rate": 2.280602825601661e-07,
      "loss": 0.4508,
      "step": 422880
    },
    {
      "epoch": 173.1,
      "grad_norm": 2.632305145263672,
      "learning_rate": 2.2799203188558385e-07,
      "loss": 0.4286,
      "step": 422890
    },
    {
      "epoch": 173.11,
      "grad_norm": 1.8752928972244263,
      "learning_rate": 2.2792379090296134e-07,
      "loss": 0.4466,
      "step": 422900
    },
    {
      "epoch": 173.11,
      "grad_norm": 2.089838981628418,
      "learning_rate": 2.2785555961261082e-07,
      "loss": 0.4503,
      "step": 422910
    },
    {
      "epoch": 173.12,
      "grad_norm": 1.9670217037200928,
      "learning_rate": 2.2778733801484503e-07,
      "loss": 0.4439,
      "step": 422920
    },
    {
      "epoch": 173.12,
      "grad_norm": 1.68997061252594,
      "learning_rate": 2.2771912610997654e-07,
      "loss": 0.438,
      "step": 422930
    },
    {
      "epoch": 173.12,
      "grad_norm": 2.0659022331237793,
      "learning_rate": 2.276509238983178e-07,
      "loss": 0.4466,
      "step": 422940
    },
    {
      "epoch": 173.13,
      "grad_norm": 2.0787389278411865,
      "learning_rate": 2.275827313801811e-07,
      "loss": 0.4646,
      "step": 422950
    },
    {
      "epoch": 173.13,
      "grad_norm": 2.9254703521728516,
      "learning_rate": 2.2751454855587894e-07,
      "loss": 0.4537,
      "step": 422960
    },
    {
      "epoch": 173.14,
      "grad_norm": 2.353339433670044,
      "learning_rate": 2.274463754257238e-07,
      "loss": 0.4329,
      "step": 422970
    },
    {
      "epoch": 173.14,
      "grad_norm": 4.109028339385986,
      "learning_rate": 2.2737821199002774e-07,
      "loss": 0.4535,
      "step": 422980
    },
    {
      "epoch": 173.14,
      "grad_norm": 1.671997308731079,
      "learning_rate": 2.2731005824910293e-07,
      "loss": 0.454,
      "step": 422990
    },
    {
      "epoch": 173.15,
      "grad_norm": 1.7433276176452637,
      "learning_rate": 2.2724191420326194e-07,
      "loss": 0.4582,
      "step": 423000
    },
    {
      "epoch": 173.15,
      "grad_norm": 2.2655439376831055,
      "learning_rate": 2.271737798528167e-07,
      "loss": 0.4335,
      "step": 423010
    },
    {
      "epoch": 173.16,
      "grad_norm": 1.9041825532913208,
      "learning_rate": 2.271056551980793e-07,
      "loss": 0.4599,
      "step": 423020
    },
    {
      "epoch": 173.16,
      "grad_norm": 2.054675340652466,
      "learning_rate": 2.2703754023936163e-07,
      "loss": 0.456,
      "step": 423030
    },
    {
      "epoch": 173.16,
      "grad_norm": 2.6217007637023926,
      "learning_rate": 2.2696943497697652e-07,
      "loss": 0.4275,
      "step": 423040
    },
    {
      "epoch": 173.17,
      "grad_norm": 2.121772050857544,
      "learning_rate": 2.2690133941123488e-07,
      "loss": 0.4493,
      "step": 423050
    },
    {
      "epoch": 173.17,
      "grad_norm": 2.6806674003601074,
      "learning_rate": 2.268332535424487e-07,
      "loss": 0.4519,
      "step": 423060
    },
    {
      "epoch": 173.18,
      "grad_norm": 2.1363234519958496,
      "learning_rate": 2.267651773709308e-07,
      "loss": 0.4312,
      "step": 423070
    },
    {
      "epoch": 173.18,
      "grad_norm": 2.004647970199585,
      "learning_rate": 2.2669711089699263e-07,
      "loss": 0.4557,
      "step": 423080
    },
    {
      "epoch": 173.18,
      "grad_norm": 1.629813313484192,
      "learning_rate": 2.266290541209456e-07,
      "loss": 0.4524,
      "step": 423090
    },
    {
      "epoch": 173.19,
      "grad_norm": 1.8131011724472046,
      "learning_rate": 2.2656100704310179e-07,
      "loss": 0.4385,
      "step": 423100
    },
    {
      "epoch": 173.19,
      "grad_norm": 2.112384796142578,
      "learning_rate": 2.2649296966377288e-07,
      "loss": 0.4309,
      "step": 423110
    },
    {
      "epoch": 173.2,
      "grad_norm": 2.22895884513855,
      "learning_rate": 2.2642494198327062e-07,
      "loss": 0.4418,
      "step": 423120
    },
    {
      "epoch": 173.2,
      "grad_norm": 1.8843737840652466,
      "learning_rate": 2.2635692400190668e-07,
      "loss": 0.4372,
      "step": 423130
    },
    {
      "epoch": 173.21,
      "grad_norm": 2.072305679321289,
      "learning_rate": 2.262889157199921e-07,
      "loss": 0.4498,
      "step": 423140
    },
    {
      "epoch": 173.21,
      "grad_norm": 1.9357799291610718,
      "learning_rate": 2.2622091713783906e-07,
      "loss": 0.4386,
      "step": 423150
    },
    {
      "epoch": 173.21,
      "grad_norm": 2.0149083137512207,
      "learning_rate": 2.2615292825575882e-07,
      "loss": 0.4387,
      "step": 423160
    },
    {
      "epoch": 173.22,
      "grad_norm": 2.7181856632232666,
      "learning_rate": 2.2608494907406282e-07,
      "loss": 0.4341,
      "step": 423170
    },
    {
      "epoch": 173.22,
      "grad_norm": 2.2752761840820312,
      "learning_rate": 2.2601697959306278e-07,
      "loss": 0.4298,
      "step": 423180
    },
    {
      "epoch": 173.23,
      "grad_norm": 2.3903069496154785,
      "learning_rate": 2.259490198130694e-07,
      "loss": 0.4542,
      "step": 423190
    },
    {
      "epoch": 173.23,
      "grad_norm": 2.301532745361328,
      "learning_rate": 2.2588106973439445e-07,
      "loss": 0.4458,
      "step": 423200
    },
    {
      "epoch": 173.23,
      "grad_norm": 2.1613123416900635,
      "learning_rate": 2.2581312935734883e-07,
      "loss": 0.432,
      "step": 423210
    },
    {
      "epoch": 173.24,
      "grad_norm": 1.897456407546997,
      "learning_rate": 2.2574519868224427e-07,
      "loss": 0.4672,
      "step": 423220
    },
    {
      "epoch": 173.24,
      "grad_norm": 2.004967212677002,
      "learning_rate": 2.2567727770939149e-07,
      "loss": 0.453,
      "step": 423230
    },
    {
      "epoch": 173.25,
      "grad_norm": 1.6572058200836182,
      "learning_rate": 2.2560936643910166e-07,
      "loss": 0.4464,
      "step": 423240
    },
    {
      "epoch": 173.25,
      "grad_norm": 2.113677740097046,
      "learning_rate": 2.2554146487168604e-07,
      "loss": 0.4501,
      "step": 423250
    },
    {
      "epoch": 173.25,
      "grad_norm": 2.3403232097625732,
      "learning_rate": 2.2547357300745578e-07,
      "loss": 0.4549,
      "step": 423260
    },
    {
      "epoch": 173.26,
      "grad_norm": 1.824852466583252,
      "learning_rate": 2.2540569084672186e-07,
      "loss": 0.4367,
      "step": 423270
    },
    {
      "epoch": 173.26,
      "grad_norm": 2.0158803462982178,
      "learning_rate": 2.2533781838979525e-07,
      "loss": 0.4686,
      "step": 423280
    },
    {
      "epoch": 173.27,
      "grad_norm": 1.5740594863891602,
      "learning_rate": 2.2526995563698662e-07,
      "loss": 0.4402,
      "step": 423290
    },
    {
      "epoch": 173.27,
      "grad_norm": 2.0490572452545166,
      "learning_rate": 2.252021025886069e-07,
      "loss": 0.4574,
      "step": 423300
    },
    {
      "epoch": 173.27,
      "grad_norm": 1.9525119066238403,
      "learning_rate": 2.251342592449676e-07,
      "loss": 0.445,
      "step": 423310
    },
    {
      "epoch": 173.28,
      "grad_norm": 1.867278814315796,
      "learning_rate": 2.2506642560637857e-07,
      "loss": 0.4478,
      "step": 423320
    },
    {
      "epoch": 173.28,
      "grad_norm": 1.8662015199661255,
      "learning_rate": 2.2499860167315082e-07,
      "loss": 0.4484,
      "step": 423330
    },
    {
      "epoch": 173.29,
      "grad_norm": 2.222512722015381,
      "learning_rate": 2.24930787445595e-07,
      "loss": 0.4451,
      "step": 423340
    },
    {
      "epoch": 173.29,
      "grad_norm": 1.881211757659912,
      "learning_rate": 2.2486298292402182e-07,
      "loss": 0.4357,
      "step": 423350
    },
    {
      "epoch": 173.3,
      "grad_norm": 2.783740282058716,
      "learning_rate": 2.2479518810874197e-07,
      "loss": 0.4343,
      "step": 423360
    },
    {
      "epoch": 173.3,
      "grad_norm": 2.884033203125,
      "learning_rate": 2.2472740300006586e-07,
      "loss": 0.4738,
      "step": 423370
    },
    {
      "epoch": 173.3,
      "grad_norm": 2.3030123710632324,
      "learning_rate": 2.2465962759830418e-07,
      "loss": 0.4391,
      "step": 423380
    },
    {
      "epoch": 173.31,
      "grad_norm": 1.6138813495635986,
      "learning_rate": 2.2459186190376738e-07,
      "loss": 0.4373,
      "step": 423390
    },
    {
      "epoch": 173.31,
      "grad_norm": 2.0890119075775146,
      "learning_rate": 2.2452410591676588e-07,
      "loss": 0.4325,
      "step": 423400
    },
    {
      "epoch": 173.32,
      "grad_norm": 2.258868932723999,
      "learning_rate": 2.2445635963760983e-07,
      "loss": 0.4557,
      "step": 423410
    },
    {
      "epoch": 173.32,
      "grad_norm": 1.8657019138336182,
      "learning_rate": 2.2438862306660969e-07,
      "loss": 0.4346,
      "step": 423420
    },
    {
      "epoch": 173.32,
      "grad_norm": 1.923192024230957,
      "learning_rate": 2.2432089620407584e-07,
      "loss": 0.4257,
      "step": 423430
    },
    {
      "epoch": 173.33,
      "grad_norm": 2.1007845401763916,
      "learning_rate": 2.2425317905031826e-07,
      "loss": 0.4403,
      "step": 423440
    },
    {
      "epoch": 173.33,
      "grad_norm": 1.8781741857528687,
      "learning_rate": 2.241854716056481e-07,
      "loss": 0.4405,
      "step": 423450
    },
    {
      "epoch": 173.34,
      "grad_norm": 2.259361505508423,
      "learning_rate": 2.2411777387037426e-07,
      "loss": 0.4428,
      "step": 423460
    },
    {
      "epoch": 173.34,
      "grad_norm": 2.036374092102051,
      "learning_rate": 2.240500858448074e-07,
      "loss": 0.443,
      "step": 423470
    },
    {
      "epoch": 173.34,
      "grad_norm": 2.0435054302215576,
      "learning_rate": 2.239824075292577e-07,
      "loss": 0.4342,
      "step": 423480
    },
    {
      "epoch": 173.35,
      "grad_norm": 2.0964298248291016,
      "learning_rate": 2.2391473892403484e-07,
      "loss": 0.4448,
      "step": 423490
    },
    {
      "epoch": 173.35,
      "grad_norm": 1.9351065158843994,
      "learning_rate": 2.2384708002944922e-07,
      "loss": 0.4498,
      "step": 423500
    },
    {
      "epoch": 173.36,
      "grad_norm": 2.2609009742736816,
      "learning_rate": 2.2377943084581047e-07,
      "loss": 0.4315,
      "step": 423510
    },
    {
      "epoch": 173.36,
      "grad_norm": 2.1538381576538086,
      "learning_rate": 2.2371179137342852e-07,
      "loss": 0.458,
      "step": 423520
    },
    {
      "epoch": 173.36,
      "grad_norm": 2.106917381286621,
      "learning_rate": 2.2364416161261355e-07,
      "loss": 0.4501,
      "step": 423530
    },
    {
      "epoch": 173.37,
      "grad_norm": 1.7857204675674438,
      "learning_rate": 2.235765415636747e-07,
      "loss": 0.4356,
      "step": 423540
    },
    {
      "epoch": 173.37,
      "grad_norm": 2.126430034637451,
      "learning_rate": 2.2350893122692233e-07,
      "loss": 0.4347,
      "step": 423550
    },
    {
      "epoch": 173.38,
      "grad_norm": 2.1842470169067383,
      "learning_rate": 2.2344133060266588e-07,
      "loss": 0.43,
      "step": 423560
    },
    {
      "epoch": 173.38,
      "grad_norm": 2.1705281734466553,
      "learning_rate": 2.2337373969121526e-07,
      "loss": 0.4593,
      "step": 423570
    },
    {
      "epoch": 173.39,
      "grad_norm": 1.788565993309021,
      "learning_rate": 2.233061584928796e-07,
      "loss": 0.4374,
      "step": 423580
    },
    {
      "epoch": 173.39,
      "grad_norm": 1.8438440561294556,
      "learning_rate": 2.2323858700796903e-07,
      "loss": 0.4359,
      "step": 423590
    },
    {
      "epoch": 173.39,
      "grad_norm": 1.9009690284729004,
      "learning_rate": 2.231710252367927e-07,
      "loss": 0.4411,
      "step": 423600
    },
    {
      "epoch": 173.4,
      "grad_norm": 4.142370700836182,
      "learning_rate": 2.231034731796603e-07,
      "loss": 0.444,
      "step": 423610
    },
    {
      "epoch": 173.4,
      "grad_norm": 2.0452470779418945,
      "learning_rate": 2.2303593083688168e-07,
      "loss": 0.4363,
      "step": 423620
    },
    {
      "epoch": 173.41,
      "grad_norm": 2.0269086360931396,
      "learning_rate": 2.2296839820876493e-07,
      "loss": 0.4329,
      "step": 423630
    },
    {
      "epoch": 173.41,
      "grad_norm": 1.7512110471725464,
      "learning_rate": 2.2290087529562052e-07,
      "loss": 0.436,
      "step": 423640
    },
    {
      "epoch": 173.41,
      "grad_norm": 2.6476011276245117,
      "learning_rate": 2.2283336209775754e-07,
      "loss": 0.4409,
      "step": 423650
    },
    {
      "epoch": 173.42,
      "grad_norm": 2.144818067550659,
      "learning_rate": 2.227658586154854e-07,
      "loss": 0.4432,
      "step": 423660
    },
    {
      "epoch": 173.42,
      "grad_norm": 2.1944167613983154,
      "learning_rate": 2.2269836484911297e-07,
      "loss": 0.4573,
      "step": 423670
    },
    {
      "epoch": 173.43,
      "grad_norm": 1.8185948133468628,
      "learning_rate": 2.226308807989496e-07,
      "loss": 0.4417,
      "step": 423680
    },
    {
      "epoch": 173.43,
      "grad_norm": 2.1871438026428223,
      "learning_rate": 2.2256340646530447e-07,
      "loss": 0.4648,
      "step": 423690
    },
    {
      "epoch": 173.43,
      "grad_norm": 2.051999092102051,
      "learning_rate": 2.2249594184848667e-07,
      "loss": 0.4611,
      "step": 423700
    },
    {
      "epoch": 173.44,
      "grad_norm": 1.9227941036224365,
      "learning_rate": 2.2242848694880507e-07,
      "loss": 0.4316,
      "step": 423710
    },
    {
      "epoch": 173.44,
      "grad_norm": 2.1203651428222656,
      "learning_rate": 2.2236104176656882e-07,
      "loss": 0.472,
      "step": 423720
    },
    {
      "epoch": 173.45,
      "grad_norm": 1.8613855838775635,
      "learning_rate": 2.2229360630208678e-07,
      "loss": 0.4421,
      "step": 423730
    },
    {
      "epoch": 173.45,
      "grad_norm": 2.1062049865722656,
      "learning_rate": 2.2222618055566808e-07,
      "loss": 0.443,
      "step": 423740
    },
    {
      "epoch": 173.45,
      "grad_norm": 2.3317668437957764,
      "learning_rate": 2.221587645276213e-07,
      "loss": 0.4682,
      "step": 423750
    },
    {
      "epoch": 173.46,
      "grad_norm": 1.918583869934082,
      "learning_rate": 2.2209135821825612e-07,
      "loss": 0.4375,
      "step": 423760
    },
    {
      "epoch": 173.46,
      "grad_norm": 2.0392327308654785,
      "learning_rate": 2.2202396162787984e-07,
      "loss": 0.4369,
      "step": 423770
    },
    {
      "epoch": 173.47,
      "grad_norm": 1.8831143379211426,
      "learning_rate": 2.219565747568021e-07,
      "loss": 0.4316,
      "step": 423780
    },
    {
      "epoch": 173.47,
      "grad_norm": 2.0386667251586914,
      "learning_rate": 2.218891976053315e-07,
      "loss": 0.4387,
      "step": 423790
    },
    {
      "epoch": 173.48,
      "grad_norm": 1.9343420267105103,
      "learning_rate": 2.218218301737764e-07,
      "loss": 0.4347,
      "step": 423800
    },
    {
      "epoch": 173.48,
      "grad_norm": 2.242734909057617,
      "learning_rate": 2.2175447246244566e-07,
      "loss": 0.4509,
      "step": 423810
    },
    {
      "epoch": 173.48,
      "grad_norm": 1.8660751581192017,
      "learning_rate": 2.2168712447164766e-07,
      "loss": 0.4547,
      "step": 423820
    },
    {
      "epoch": 173.49,
      "grad_norm": 2.2176103591918945,
      "learning_rate": 2.216197862016912e-07,
      "loss": 0.4441,
      "step": 423830
    },
    {
      "epoch": 173.49,
      "grad_norm": 1.6454862356185913,
      "learning_rate": 2.215524576528847e-07,
      "loss": 0.4522,
      "step": 423840
    },
    {
      "epoch": 173.5,
      "grad_norm": 2.5011913776397705,
      "learning_rate": 2.2148513882553647e-07,
      "loss": 0.4582,
      "step": 423850
    },
    {
      "epoch": 173.5,
      "grad_norm": 1.6915544271469116,
      "learning_rate": 2.2141782971995484e-07,
      "loss": 0.4511,
      "step": 423860
    },
    {
      "epoch": 173.5,
      "grad_norm": 1.9275474548339844,
      "learning_rate": 2.213505303364482e-07,
      "loss": 0.4529,
      "step": 423870
    },
    {
      "epoch": 173.51,
      "grad_norm": 2.0984320640563965,
      "learning_rate": 2.2128324067532487e-07,
      "loss": 0.4409,
      "step": 423880
    },
    {
      "epoch": 173.51,
      "grad_norm": 1.554153561592102,
      "learning_rate": 2.2121596073689317e-07,
      "loss": 0.4419,
      "step": 423890
    },
    {
      "epoch": 173.52,
      "grad_norm": 1.978529453277588,
      "learning_rate": 2.2114869052146152e-07,
      "loss": 0.4568,
      "step": 423900
    },
    {
      "epoch": 173.52,
      "grad_norm": 1.883412480354309,
      "learning_rate": 2.2108143002933716e-07,
      "loss": 0.4654,
      "step": 423910
    },
    {
      "epoch": 173.52,
      "grad_norm": 3.2740445137023926,
      "learning_rate": 2.21014179260829e-07,
      "loss": 0.4534,
      "step": 423920
    },
    {
      "epoch": 173.53,
      "grad_norm": 2.825824499130249,
      "learning_rate": 2.2094693821624485e-07,
      "loss": 0.4481,
      "step": 423930
    },
    {
      "epoch": 173.53,
      "grad_norm": 2.4387078285217285,
      "learning_rate": 2.208797068958928e-07,
      "loss": 0.4602,
      "step": 423940
    },
    {
      "epoch": 173.54,
      "grad_norm": 2.144286870956421,
      "learning_rate": 2.2081248530008066e-07,
      "loss": 0.4411,
      "step": 423950
    },
    {
      "epoch": 173.54,
      "grad_norm": 1.8289527893066406,
      "learning_rate": 2.2074527342911655e-07,
      "loss": 0.4468,
      "step": 423960
    },
    {
      "epoch": 173.54,
      "grad_norm": 1.9080437421798706,
      "learning_rate": 2.2067807128330825e-07,
      "loss": 0.434,
      "step": 423970
    },
    {
      "epoch": 173.55,
      "grad_norm": 1.7573117017745972,
      "learning_rate": 2.206108788629639e-07,
      "loss": 0.435,
      "step": 423980
    },
    {
      "epoch": 173.55,
      "grad_norm": 2.1242830753326416,
      "learning_rate": 2.2054369616839076e-07,
      "loss": 0.4676,
      "step": 423990
    },
    {
      "epoch": 173.56,
      "grad_norm": 2.257617473602295,
      "learning_rate": 2.2047652319989693e-07,
      "loss": 0.4426,
      "step": 424000
    },
    {
      "epoch": 173.56,
      "grad_norm": 1.8836010694503784,
      "learning_rate": 2.2040935995778974e-07,
      "loss": 0.4394,
      "step": 424010
    },
    {
      "epoch": 173.57,
      "grad_norm": 2.0743823051452637,
      "learning_rate": 2.203422064423778e-07,
      "loss": 0.4318,
      "step": 424020
    },
    {
      "epoch": 173.57,
      "grad_norm": 2.1137821674346924,
      "learning_rate": 2.2027506265396808e-07,
      "loss": 0.4486,
      "step": 424030
    },
    {
      "epoch": 173.57,
      "grad_norm": 1.9266842603683472,
      "learning_rate": 2.2020792859286826e-07,
      "loss": 0.4629,
      "step": 424040
    },
    {
      "epoch": 173.58,
      "grad_norm": 1.971106767654419,
      "learning_rate": 2.201408042593853e-07,
      "loss": 0.4458,
      "step": 424050
    },
    {
      "epoch": 173.58,
      "grad_norm": 2.5673117637634277,
      "learning_rate": 2.2007368965382756e-07,
      "loss": 0.447,
      "step": 424060
    },
    {
      "epoch": 173.59,
      "grad_norm": 1.6012861728668213,
      "learning_rate": 2.2000658477650188e-07,
      "loss": 0.4467,
      "step": 424070
    },
    {
      "epoch": 173.59,
      "grad_norm": 1.7781646251678467,
      "learning_rate": 2.1993948962771602e-07,
      "loss": 0.4351,
      "step": 424080
    },
    {
      "epoch": 173.59,
      "grad_norm": 2.162837266921997,
      "learning_rate": 2.1987240420777708e-07,
      "loss": 0.4239,
      "step": 424090
    },
    {
      "epoch": 173.6,
      "grad_norm": 2.083728551864624,
      "learning_rate": 2.198053285169926e-07,
      "loss": 0.423,
      "step": 424100
    },
    {
      "epoch": 173.6,
      "grad_norm": 2.047780752182007,
      "learning_rate": 2.197382625556696e-07,
      "loss": 0.4247,
      "step": 424110
    },
    {
      "epoch": 173.61,
      "grad_norm": 1.9475668668746948,
      "learning_rate": 2.1967120632411543e-07,
      "loss": 0.4452,
      "step": 424120
    },
    {
      "epoch": 173.61,
      "grad_norm": 1.928043246269226,
      "learning_rate": 2.1960415982263737e-07,
      "loss": 0.4544,
      "step": 424130
    },
    {
      "epoch": 173.61,
      "grad_norm": 2.139082908630371,
      "learning_rate": 2.195371230515425e-07,
      "loss": 0.4292,
      "step": 424140
    },
    {
      "epoch": 173.62,
      "grad_norm": 2.6497318744659424,
      "learning_rate": 2.1947009601113755e-07,
      "loss": 0.4406,
      "step": 424150
    },
    {
      "epoch": 173.62,
      "grad_norm": 2.5730435848236084,
      "learning_rate": 2.1940307870173015e-07,
      "loss": 0.442,
      "step": 424160
    },
    {
      "epoch": 173.63,
      "grad_norm": 1.9930856227874756,
      "learning_rate": 2.1933607112362702e-07,
      "loss": 0.4171,
      "step": 424170
    },
    {
      "epoch": 173.63,
      "grad_norm": 1.9008606672286987,
      "learning_rate": 2.1926907327713476e-07,
      "loss": 0.4615,
      "step": 424180
    },
    {
      "epoch": 173.63,
      "grad_norm": 1.894767165184021,
      "learning_rate": 2.1920208516256087e-07,
      "loss": 0.4089,
      "step": 424190
    },
    {
      "epoch": 173.64,
      "grad_norm": 1.942787528038025,
      "learning_rate": 2.191351067802119e-07,
      "loss": 0.4647,
      "step": 424200
    },
    {
      "epoch": 173.64,
      "grad_norm": 2.5299456119537354,
      "learning_rate": 2.1906813813039465e-07,
      "loss": 0.4459,
      "step": 424210
    },
    {
      "epoch": 173.65,
      "grad_norm": 2.143932580947876,
      "learning_rate": 2.1900117921341615e-07,
      "loss": 0.4553,
      "step": 424220
    },
    {
      "epoch": 173.65,
      "grad_norm": 2.108121156692505,
      "learning_rate": 2.1893423002958266e-07,
      "loss": 0.4459,
      "step": 424230
    },
    {
      "epoch": 173.66,
      "grad_norm": 1.892626166343689,
      "learning_rate": 2.1886729057920125e-07,
      "loss": 0.43,
      "step": 424240
    },
    {
      "epoch": 173.66,
      "grad_norm": 2.235452175140381,
      "learning_rate": 2.1880036086257868e-07,
      "loss": 0.4255,
      "step": 424250
    },
    {
      "epoch": 173.66,
      "grad_norm": 2.42602276802063,
      "learning_rate": 2.1873344088002097e-07,
      "loss": 0.4364,
      "step": 424260
    },
    {
      "epoch": 173.67,
      "grad_norm": 2.105564594268799,
      "learning_rate": 2.1866653063183515e-07,
      "loss": 0.4397,
      "step": 424270
    },
    {
      "epoch": 173.67,
      "grad_norm": 1.777827501296997,
      "learning_rate": 2.1859963011832777e-07,
      "loss": 0.4307,
      "step": 424280
    },
    {
      "epoch": 173.68,
      "grad_norm": 2.202547073364258,
      "learning_rate": 2.1853273933980484e-07,
      "loss": 0.4452,
      "step": 424290
    },
    {
      "epoch": 173.68,
      "grad_norm": 1.8246865272521973,
      "learning_rate": 2.1846585829657335e-07,
      "loss": 0.4496,
      "step": 424300
    },
    {
      "epoch": 173.68,
      "grad_norm": 1.9509069919586182,
      "learning_rate": 2.183989869889391e-07,
      "loss": 0.465,
      "step": 424310
    },
    {
      "epoch": 173.69,
      "grad_norm": 1.8522659540176392,
      "learning_rate": 2.1833212541720889e-07,
      "loss": 0.4452,
      "step": 424320
    },
    {
      "epoch": 173.69,
      "grad_norm": 2.0086722373962402,
      "learning_rate": 2.1826527358168867e-07,
      "loss": 0.4323,
      "step": 424330
    },
    {
      "epoch": 173.7,
      "grad_norm": 2.0933384895324707,
      "learning_rate": 2.1819843148268525e-07,
      "loss": 0.4384,
      "step": 424340
    },
    {
      "epoch": 173.7,
      "grad_norm": 2.018218517303467,
      "learning_rate": 2.1813159912050384e-07,
      "loss": 0.4485,
      "step": 424350
    },
    {
      "epoch": 173.7,
      "grad_norm": 1.9585613012313843,
      "learning_rate": 2.1806477649545127e-07,
      "loss": 0.4388,
      "step": 424360
    },
    {
      "epoch": 173.71,
      "grad_norm": 1.826900601387024,
      "learning_rate": 2.179979636078335e-07,
      "loss": 0.4438,
      "step": 424370
    },
    {
      "epoch": 173.71,
      "grad_norm": 1.9081650972366333,
      "learning_rate": 2.1793116045795606e-07,
      "loss": 0.4515,
      "step": 424380
    },
    {
      "epoch": 173.72,
      "grad_norm": 2.3343303203582764,
      "learning_rate": 2.178643670461262e-07,
      "loss": 0.4464,
      "step": 424390
    },
    {
      "epoch": 173.72,
      "grad_norm": 1.6597791910171509,
      "learning_rate": 2.1779758337264891e-07,
      "loss": 0.4443,
      "step": 424400
    },
    {
      "epoch": 173.72,
      "grad_norm": 1.9571747779846191,
      "learning_rate": 2.1773080943783046e-07,
      "loss": 0.4346,
      "step": 424410
    },
    {
      "epoch": 173.73,
      "grad_norm": 2.0356650352478027,
      "learning_rate": 2.1766404524197687e-07,
      "loss": 0.439,
      "step": 424420
    },
    {
      "epoch": 173.73,
      "grad_norm": 2.042515993118286,
      "learning_rate": 2.175972907853936e-07,
      "loss": 0.4315,
      "step": 424430
    },
    {
      "epoch": 173.74,
      "grad_norm": 1.928871512413025,
      "learning_rate": 2.1753054606838638e-07,
      "loss": 0.4505,
      "step": 424440
    },
    {
      "epoch": 173.74,
      "grad_norm": 2.1027474403381348,
      "learning_rate": 2.1746381109126152e-07,
      "loss": 0.4619,
      "step": 424450
    },
    {
      "epoch": 173.75,
      "grad_norm": 2.191934108734131,
      "learning_rate": 2.1739708585432449e-07,
      "loss": 0.4302,
      "step": 424460
    },
    {
      "epoch": 173.75,
      "grad_norm": 2.124605178833008,
      "learning_rate": 2.173303703578805e-07,
      "loss": 0.4741,
      "step": 424470
    },
    {
      "epoch": 173.75,
      "grad_norm": 2.0427796840667725,
      "learning_rate": 2.172636646022361e-07,
      "loss": 0.4514,
      "step": 424480
    },
    {
      "epoch": 173.76,
      "grad_norm": 2.167484998703003,
      "learning_rate": 2.1719696858769594e-07,
      "loss": 0.4439,
      "step": 424490
    },
    {
      "epoch": 173.76,
      "grad_norm": 1.9825000762939453,
      "learning_rate": 2.171302823145658e-07,
      "loss": 0.4487,
      "step": 424500
    },
    {
      "epoch": 173.77,
      "grad_norm": 2.125932455062866,
      "learning_rate": 2.170636057831512e-07,
      "loss": 0.4425,
      "step": 424510
    },
    {
      "epoch": 173.77,
      "grad_norm": 2.0593693256378174,
      "learning_rate": 2.169969389937576e-07,
      "loss": 0.4251,
      "step": 424520
    },
    {
      "epoch": 173.77,
      "grad_norm": 2.4160494804382324,
      "learning_rate": 2.1693028194669043e-07,
      "loss": 0.4396,
      "step": 424530
    },
    {
      "epoch": 173.78,
      "grad_norm": 1.9578852653503418,
      "learning_rate": 2.1686363464225527e-07,
      "loss": 0.4477,
      "step": 424540
    },
    {
      "epoch": 173.78,
      "grad_norm": 1.7725491523742676,
      "learning_rate": 2.1679699708075675e-07,
      "loss": 0.4429,
      "step": 424550
    },
    {
      "epoch": 173.79,
      "grad_norm": 2.3242409229278564,
      "learning_rate": 2.1673036926250064e-07,
      "loss": 0.4554,
      "step": 424560
    },
    {
      "epoch": 173.79,
      "grad_norm": 2.1701242923736572,
      "learning_rate": 2.166637511877919e-07,
      "loss": 0.4381,
      "step": 424570
    },
    {
      "epoch": 173.79,
      "grad_norm": 2.07259464263916,
      "learning_rate": 2.1659714285693604e-07,
      "loss": 0.4526,
      "step": 424580
    },
    {
      "epoch": 173.8,
      "grad_norm": 3.453763484954834,
      "learning_rate": 2.1653054427023798e-07,
      "loss": 0.4366,
      "step": 424590
    },
    {
      "epoch": 173.8,
      "grad_norm": 1.7460614442825317,
      "learning_rate": 2.1646395542800275e-07,
      "loss": 0.4304,
      "step": 424600
    },
    {
      "epoch": 173.81,
      "grad_norm": 2.524540662765503,
      "learning_rate": 2.1639737633053549e-07,
      "loss": 0.4474,
      "step": 424610
    },
    {
      "epoch": 173.81,
      "grad_norm": 2.0508458614349365,
      "learning_rate": 2.1633080697814175e-07,
      "loss": 0.4426,
      "step": 424620
    },
    {
      "epoch": 173.81,
      "grad_norm": 1.998315453529358,
      "learning_rate": 2.1626424737112516e-07,
      "loss": 0.4347,
      "step": 424630
    },
    {
      "epoch": 173.82,
      "grad_norm": 1.9034616947174072,
      "learning_rate": 2.161976975097915e-07,
      "loss": 0.467,
      "step": 424640
    },
    {
      "epoch": 173.82,
      "grad_norm": 1.9718064069747925,
      "learning_rate": 2.1613115739444543e-07,
      "loss": 0.4447,
      "step": 424650
    },
    {
      "epoch": 173.83,
      "grad_norm": 2.011723756790161,
      "learning_rate": 2.1606462702539169e-07,
      "loss": 0.4704,
      "step": 424660
    },
    {
      "epoch": 173.83,
      "grad_norm": 2.0855212211608887,
      "learning_rate": 2.159981064029352e-07,
      "loss": 0.4421,
      "step": 424670
    },
    {
      "epoch": 173.84,
      "grad_norm": 2.469080686569214,
      "learning_rate": 2.1593159552738044e-07,
      "loss": 0.4369,
      "step": 424680
    },
    {
      "epoch": 173.84,
      "grad_norm": 1.8323386907577515,
      "learning_rate": 2.1586509439903238e-07,
      "loss": 0.4405,
      "step": 424690
    },
    {
      "epoch": 173.84,
      "grad_norm": 2.1438796520233154,
      "learning_rate": 2.1579860301819572e-07,
      "loss": 0.4246,
      "step": 424700
    },
    {
      "epoch": 173.85,
      "grad_norm": 2.1730127334594727,
      "learning_rate": 2.1573212138517463e-07,
      "loss": 0.464,
      "step": 424710
    },
    {
      "epoch": 173.85,
      "grad_norm": 1.6449235677719116,
      "learning_rate": 2.1566564950027382e-07,
      "loss": 0.4463,
      "step": 424720
    },
    {
      "epoch": 173.86,
      "grad_norm": 2.1090505123138428,
      "learning_rate": 2.15599187363798e-07,
      "loss": 0.4256,
      "step": 424730
    },
    {
      "epoch": 173.86,
      "grad_norm": 2.2869765758514404,
      "learning_rate": 2.1553273497605157e-07,
      "loss": 0.4456,
      "step": 424740
    },
    {
      "epoch": 173.86,
      "grad_norm": 1.8869057893753052,
      "learning_rate": 2.1546629233733853e-07,
      "loss": 0.4439,
      "step": 424750
    },
    {
      "epoch": 173.87,
      "grad_norm": 2.0559794902801514,
      "learning_rate": 2.153998594479638e-07,
      "loss": 0.438,
      "step": 424760
    },
    {
      "epoch": 173.87,
      "grad_norm": 2.161733627319336,
      "learning_rate": 2.153334363082313e-07,
      "loss": 0.4473,
      "step": 424770
    },
    {
      "epoch": 173.88,
      "grad_norm": 1.6861077547073364,
      "learning_rate": 2.152670229184455e-07,
      "loss": 0.4592,
      "step": 424780
    },
    {
      "epoch": 173.88,
      "grad_norm": 2.304184675216675,
      "learning_rate": 2.1520061927891055e-07,
      "loss": 0.4607,
      "step": 424790
    },
    {
      "epoch": 173.88,
      "grad_norm": 1.992496132850647,
      "learning_rate": 2.1513422538993063e-07,
      "loss": 0.4423,
      "step": 424800
    },
    {
      "epoch": 173.89,
      "grad_norm": 1.8996636867523193,
      "learning_rate": 2.1506784125181e-07,
      "loss": 0.4404,
      "step": 424810
    },
    {
      "epoch": 173.89,
      "grad_norm": 2.098050832748413,
      "learning_rate": 2.1500146686485275e-07,
      "loss": 0.4441,
      "step": 424820
    },
    {
      "epoch": 173.9,
      "grad_norm": 2.6147377490997314,
      "learning_rate": 2.1493510222936285e-07,
      "loss": 0.4333,
      "step": 424830
    },
    {
      "epoch": 173.9,
      "grad_norm": 2.4249887466430664,
      "learning_rate": 2.148687473456442e-07,
      "loss": 0.452,
      "step": 424840
    },
    {
      "epoch": 173.91,
      "grad_norm": 2.064155101776123,
      "learning_rate": 2.1480240221400075e-07,
      "loss": 0.4526,
      "step": 424850
    },
    {
      "epoch": 173.91,
      "grad_norm": 3.061121940612793,
      "learning_rate": 2.147360668347367e-07,
      "loss": 0.4412,
      "step": 424860
    },
    {
      "epoch": 173.91,
      "grad_norm": 2.0650198459625244,
      "learning_rate": 2.1466974120815566e-07,
      "loss": 0.4503,
      "step": 424870
    },
    {
      "epoch": 173.92,
      "grad_norm": 1.9725710153579712,
      "learning_rate": 2.1460342533456183e-07,
      "loss": 0.4469,
      "step": 424880
    },
    {
      "epoch": 173.92,
      "grad_norm": 2.3814404010772705,
      "learning_rate": 2.1453711921425838e-07,
      "loss": 0.4436,
      "step": 424890
    },
    {
      "epoch": 173.93,
      "grad_norm": 2.323713541030884,
      "learning_rate": 2.1447082284754948e-07,
      "loss": 0.4206,
      "step": 424900
    },
    {
      "epoch": 173.93,
      "grad_norm": 1.9135223627090454,
      "learning_rate": 2.1440453623473877e-07,
      "loss": 0.4277,
      "step": 424910
    },
    {
      "epoch": 173.93,
      "grad_norm": 1.8713456392288208,
      "learning_rate": 2.1433825937612997e-07,
      "loss": 0.4634,
      "step": 424920
    },
    {
      "epoch": 173.94,
      "grad_norm": 2.0665643215179443,
      "learning_rate": 2.1427199227202693e-07,
      "loss": 0.4466,
      "step": 424930
    },
    {
      "epoch": 173.94,
      "grad_norm": 2.2757344245910645,
      "learning_rate": 2.1420573492273233e-07,
      "loss": 0.4345,
      "step": 424940
    },
    {
      "epoch": 173.95,
      "grad_norm": 2.4217567443847656,
      "learning_rate": 2.141394873285498e-07,
      "loss": 0.4406,
      "step": 424950
    },
    {
      "epoch": 173.95,
      "grad_norm": 2.494671583175659,
      "learning_rate": 2.140732494897838e-07,
      "loss": 0.4567,
      "step": 424960
    },
    {
      "epoch": 173.95,
      "grad_norm": 2.8280885219573975,
      "learning_rate": 2.1400702140673695e-07,
      "loss": 0.4398,
      "step": 424970
    },
    {
      "epoch": 173.96,
      "grad_norm": 2.073122501373291,
      "learning_rate": 2.1394080307971292e-07,
      "loss": 0.4342,
      "step": 424980
    },
    {
      "epoch": 173.96,
      "grad_norm": 2.1378047466278076,
      "learning_rate": 2.138745945090151e-07,
      "loss": 0.4356,
      "step": 424990
    },
    {
      "epoch": 173.97,
      "grad_norm": 2.0839626789093018,
      "learning_rate": 2.1380839569494663e-07,
      "loss": 0.459,
      "step": 425000
    },
    {
      "epoch": 173.97,
      "grad_norm": 1.8178586959838867,
      "learning_rate": 2.1374220663781066e-07,
      "loss": 0.4346,
      "step": 425010
    },
    {
      "epoch": 173.97,
      "grad_norm": 2.026585578918457,
      "learning_rate": 2.1367602733791062e-07,
      "loss": 0.4426,
      "step": 425020
    },
    {
      "epoch": 173.98,
      "grad_norm": 1.9923428297042847,
      "learning_rate": 2.136098577955496e-07,
      "loss": 0.4443,
      "step": 425030
    },
    {
      "epoch": 173.98,
      "grad_norm": 2.09285306930542,
      "learning_rate": 2.1354369801103053e-07,
      "loss": 0.4385,
      "step": 425040
    },
    {
      "epoch": 173.99,
      "grad_norm": 2.4988701343536377,
      "learning_rate": 2.134775479846568e-07,
      "loss": 0.4422,
      "step": 425050
    },
    {
      "epoch": 173.99,
      "grad_norm": 1.991753101348877,
      "learning_rate": 2.134114077167315e-07,
      "loss": 0.4283,
      "step": 425060
    },
    {
      "epoch": 174.0,
      "grad_norm": 1.9567663669586182,
      "learning_rate": 2.133452772075571e-07,
      "loss": 0.4536,
      "step": 425070
    },
    {
      "epoch": 174.0,
      "grad_norm": 2.2133302688598633,
      "learning_rate": 2.1327915645743665e-07,
      "loss": 0.4378,
      "step": 425080
    },
    {
      "epoch": 174.0,
      "eval_loss": 0.44596895575523376,
      "eval_runtime": 52.4797,
      "eval_samples_per_second": 65.721,
      "eval_steps_per_second": 8.232,
      "step": 425082
    },
    {
      "epoch": 174.0,
      "grad_norm": 2.142038583755493,
      "learning_rate": 2.1321304546667333e-07,
      "loss": 0.4506,
      "step": 425090
    },
    {
      "epoch": 174.01,
      "grad_norm": 1.8480931520462036,
      "learning_rate": 2.1314694423556974e-07,
      "loss": 0.4464,
      "step": 425100
    },
    {
      "epoch": 174.01,
      "grad_norm": 2.013826608657837,
      "learning_rate": 2.130808527644288e-07,
      "loss": 0.473,
      "step": 425110
    },
    {
      "epoch": 174.02,
      "grad_norm": 1.8823643922805786,
      "learning_rate": 2.1301477105355336e-07,
      "loss": 0.4384,
      "step": 425120
    },
    {
      "epoch": 174.02,
      "grad_norm": 4.100808620452881,
      "learning_rate": 2.129486991032458e-07,
      "loss": 0.4426,
      "step": 425130
    },
    {
      "epoch": 174.02,
      "grad_norm": 2.082300901412964,
      "learning_rate": 2.1288263691380875e-07,
      "loss": 0.4406,
      "step": 425140
    },
    {
      "epoch": 174.03,
      "grad_norm": 1.9688565731048584,
      "learning_rate": 2.128165844855456e-07,
      "loss": 0.4262,
      "step": 425150
    },
    {
      "epoch": 174.03,
      "grad_norm": 1.949966549873352,
      "learning_rate": 2.1275054181875817e-07,
      "loss": 0.4621,
      "step": 425160
    },
    {
      "epoch": 174.04,
      "grad_norm": 1.9099054336547852,
      "learning_rate": 2.1268450891374936e-07,
      "loss": 0.4403,
      "step": 425170
    },
    {
      "epoch": 174.04,
      "grad_norm": 2.283754825592041,
      "learning_rate": 2.1261848577082157e-07,
      "loss": 0.4624,
      "step": 425180
    },
    {
      "epoch": 174.04,
      "grad_norm": 1.908313512802124,
      "learning_rate": 2.125524723902771e-07,
      "loss": 0.4555,
      "step": 425190
    },
    {
      "epoch": 174.05,
      "grad_norm": 1.7649993896484375,
      "learning_rate": 2.124864687724189e-07,
      "loss": 0.4394,
      "step": 425200
    },
    {
      "epoch": 174.05,
      "grad_norm": 2.021742582321167,
      "learning_rate": 2.1242047491754846e-07,
      "loss": 0.4391,
      "step": 425210
    },
    {
      "epoch": 174.06,
      "grad_norm": 1.7756907939910889,
      "learning_rate": 2.1235449082596848e-07,
      "loss": 0.4466,
      "step": 425220
    },
    {
      "epoch": 174.06,
      "grad_norm": 2.378938913345337,
      "learning_rate": 2.1228851649798133e-07,
      "loss": 0.4423,
      "step": 425230
    },
    {
      "epoch": 174.06,
      "grad_norm": 1.7549618482589722,
      "learning_rate": 2.1222255193388905e-07,
      "loss": 0.4304,
      "step": 425240
    },
    {
      "epoch": 174.07,
      "grad_norm": 2.1661064624786377,
      "learning_rate": 2.1215659713399376e-07,
      "loss": 0.4537,
      "step": 425250
    },
    {
      "epoch": 174.07,
      "grad_norm": 2.1823766231536865,
      "learning_rate": 2.1209065209859786e-07,
      "loss": 0.458,
      "step": 425260
    },
    {
      "epoch": 174.08,
      "grad_norm": 2.3840105533599854,
      "learning_rate": 2.1202471682800342e-07,
      "loss": 0.4475,
      "step": 425270
    },
    {
      "epoch": 174.08,
      "grad_norm": 2.2793519496917725,
      "learning_rate": 2.119587913225123e-07,
      "loss": 0.456,
      "step": 425280
    },
    {
      "epoch": 174.09,
      "grad_norm": 2.0743792057037354,
      "learning_rate": 2.1189287558242658e-07,
      "loss": 0.4371,
      "step": 425290
    },
    {
      "epoch": 174.09,
      "grad_norm": 2.249990463256836,
      "learning_rate": 2.118269696080481e-07,
      "loss": 0.4521,
      "step": 425300
    },
    {
      "epoch": 174.09,
      "grad_norm": 2.2924306392669678,
      "learning_rate": 2.1176107339967876e-07,
      "loss": 0.4422,
      "step": 425310
    },
    {
      "epoch": 174.1,
      "grad_norm": 2.0543372631073,
      "learning_rate": 2.1169518695762057e-07,
      "loss": 0.4419,
      "step": 425320
    },
    {
      "epoch": 174.1,
      "grad_norm": 2.344972848892212,
      "learning_rate": 2.1162931028217543e-07,
      "loss": 0.4408,
      "step": 425330
    },
    {
      "epoch": 174.11,
      "grad_norm": 1.905888557434082,
      "learning_rate": 2.1156344337364494e-07,
      "loss": 0.446,
      "step": 425340
    },
    {
      "epoch": 174.11,
      "grad_norm": 2.0523300170898438,
      "learning_rate": 2.1149758623233088e-07,
      "loss": 0.431,
      "step": 425350
    },
    {
      "epoch": 174.11,
      "grad_norm": 2.5442214012145996,
      "learning_rate": 2.1143173885853487e-07,
      "loss": 0.4569,
      "step": 425360
    },
    {
      "epoch": 174.12,
      "grad_norm": 1.9313349723815918,
      "learning_rate": 2.113659012525585e-07,
      "loss": 0.4428,
      "step": 425370
    },
    {
      "epoch": 174.12,
      "grad_norm": 2.019874095916748,
      "learning_rate": 2.1130007341470333e-07,
      "loss": 0.4421,
      "step": 425380
    },
    {
      "epoch": 174.13,
      "grad_norm": 1.9157716035842896,
      "learning_rate": 2.112342553452712e-07,
      "loss": 0.4629,
      "step": 425390
    },
    {
      "epoch": 174.13,
      "grad_norm": 2.2220168113708496,
      "learning_rate": 2.1116844704456344e-07,
      "loss": 0.4449,
      "step": 425400
    },
    {
      "epoch": 174.13,
      "grad_norm": 2.1797430515289307,
      "learning_rate": 2.1110264851288165e-07,
      "loss": 0.4474,
      "step": 425410
    },
    {
      "epoch": 174.14,
      "grad_norm": 2.0205984115600586,
      "learning_rate": 2.1103685975052686e-07,
      "loss": 0.4128,
      "step": 425420
    },
    {
      "epoch": 174.14,
      "grad_norm": 2.1795735359191895,
      "learning_rate": 2.109710807578007e-07,
      "loss": 0.4369,
      "step": 425430
    },
    {
      "epoch": 174.15,
      "grad_norm": 2.397728443145752,
      "learning_rate": 2.1090531153500442e-07,
      "loss": 0.4455,
      "step": 425440
    },
    {
      "epoch": 174.15,
      "grad_norm": 2.3674285411834717,
      "learning_rate": 2.1083955208243964e-07,
      "loss": 0.4417,
      "step": 425450
    },
    {
      "epoch": 174.15,
      "grad_norm": 2.04249906539917,
      "learning_rate": 2.107738024004069e-07,
      "loss": 0.4309,
      "step": 425460
    },
    {
      "epoch": 174.16,
      "grad_norm": 1.8762670755386353,
      "learning_rate": 2.1070806248920806e-07,
      "loss": 0.4449,
      "step": 425470
    },
    {
      "epoch": 174.16,
      "grad_norm": 1.8866080045700073,
      "learning_rate": 2.1064233234914388e-07,
      "loss": 0.4285,
      "step": 425480
    },
    {
      "epoch": 174.17,
      "grad_norm": 2.3692069053649902,
      "learning_rate": 2.1057661198051572e-07,
      "loss": 0.4272,
      "step": 425490
    },
    {
      "epoch": 174.17,
      "grad_norm": 2.288006544113159,
      "learning_rate": 2.1051090138362435e-07,
      "loss": 0.4228,
      "step": 425500
    },
    {
      "epoch": 174.18,
      "grad_norm": 2.198402166366577,
      "learning_rate": 2.1044520055877087e-07,
      "loss": 0.4632,
      "step": 425510
    },
    {
      "epoch": 174.18,
      "grad_norm": 1.907945156097412,
      "learning_rate": 2.103795095062563e-07,
      "loss": 0.4595,
      "step": 425520
    },
    {
      "epoch": 174.18,
      "grad_norm": 2.1616616249084473,
      "learning_rate": 2.1031382822638173e-07,
      "loss": 0.4512,
      "step": 425530
    },
    {
      "epoch": 174.19,
      "grad_norm": 2.2742209434509277,
      "learning_rate": 2.1024815671944768e-07,
      "loss": 0.4417,
      "step": 425540
    },
    {
      "epoch": 174.19,
      "grad_norm": 1.901538372039795,
      "learning_rate": 2.1018249498575523e-07,
      "loss": 0.437,
      "step": 425550
    },
    {
      "epoch": 174.2,
      "grad_norm": 1.7282917499542236,
      "learning_rate": 2.1011684302560516e-07,
      "loss": 0.4489,
      "step": 425560
    },
    {
      "epoch": 174.2,
      "grad_norm": 2.234889030456543,
      "learning_rate": 2.1005120083929802e-07,
      "loss": 0.4311,
      "step": 425570
    },
    {
      "epoch": 174.2,
      "grad_norm": 2.0147864818573,
      "learning_rate": 2.0998556842713463e-07,
      "loss": 0.4385,
      "step": 425580
    },
    {
      "epoch": 174.21,
      "grad_norm": 1.826564908027649,
      "learning_rate": 2.0991994578941576e-07,
      "loss": 0.4549,
      "step": 425590
    },
    {
      "epoch": 174.21,
      "grad_norm": 2.1471660137176514,
      "learning_rate": 2.098543329264417e-07,
      "loss": 0.4569,
      "step": 425600
    },
    {
      "epoch": 174.22,
      "grad_norm": 2.4756219387054443,
      "learning_rate": 2.0978872983851357e-07,
      "loss": 0.4408,
      "step": 425610
    },
    {
      "epoch": 174.22,
      "grad_norm": 1.9004993438720703,
      "learning_rate": 2.0972313652593128e-07,
      "loss": 0.4405,
      "step": 425620
    },
    {
      "epoch": 174.22,
      "grad_norm": 2.6653244495391846,
      "learning_rate": 2.096575529889957e-07,
      "loss": 0.4343,
      "step": 425630
    },
    {
      "epoch": 174.23,
      "grad_norm": 1.783149003982544,
      "learning_rate": 2.0959197922800736e-07,
      "loss": 0.4433,
      "step": 425640
    },
    {
      "epoch": 174.23,
      "grad_norm": 1.7485175132751465,
      "learning_rate": 2.0952641524326603e-07,
      "loss": 0.4415,
      "step": 425650
    },
    {
      "epoch": 174.24,
      "grad_norm": 1.9998600482940674,
      "learning_rate": 2.094608610350725e-07,
      "loss": 0.4424,
      "step": 425660
    },
    {
      "epoch": 174.24,
      "grad_norm": 2.318006992340088,
      "learning_rate": 2.093953166037268e-07,
      "loss": 0.4328,
      "step": 425670
    },
    {
      "epoch": 174.24,
      "grad_norm": 2.384169816970825,
      "learning_rate": 2.0932978194952945e-07,
      "loss": 0.4489,
      "step": 425680
    },
    {
      "epoch": 174.25,
      "grad_norm": 2.834505319595337,
      "learning_rate": 2.0926425707278074e-07,
      "loss": 0.4546,
      "step": 425690
    },
    {
      "epoch": 174.25,
      "grad_norm": 2.1319644451141357,
      "learning_rate": 2.091987419737799e-07,
      "loss": 0.4345,
      "step": 425700
    },
    {
      "epoch": 174.26,
      "grad_norm": 2.073457717895508,
      "learning_rate": 2.0913323665282855e-07,
      "loss": 0.452,
      "step": 425710
    },
    {
      "epoch": 174.26,
      "grad_norm": 1.863924264907837,
      "learning_rate": 2.0906774111022587e-07,
      "loss": 0.4426,
      "step": 425720
    },
    {
      "epoch": 174.27,
      "grad_norm": 2.220703125,
      "learning_rate": 2.0900225534627195e-07,
      "loss": 0.4264,
      "step": 425730
    },
    {
      "epoch": 174.27,
      "grad_norm": 1.806742787361145,
      "learning_rate": 2.0893677936126702e-07,
      "loss": 0.4575,
      "step": 425740
    },
    {
      "epoch": 174.27,
      "grad_norm": 2.1563165187835693,
      "learning_rate": 2.0887131315551058e-07,
      "loss": 0.4381,
      "step": 425750
    },
    {
      "epoch": 174.28,
      "grad_norm": 1.804500699043274,
      "learning_rate": 2.0880585672930294e-07,
      "loss": 0.4429,
      "step": 425760
    },
    {
      "epoch": 174.28,
      "grad_norm": 2.1257336139678955,
      "learning_rate": 2.0874041008294386e-07,
      "loss": 0.4722,
      "step": 425770
    },
    {
      "epoch": 174.29,
      "grad_norm": 2.2811572551727295,
      "learning_rate": 2.0867497321673333e-07,
      "loss": 0.4564,
      "step": 425780
    },
    {
      "epoch": 174.29,
      "grad_norm": 2.323021411895752,
      "learning_rate": 2.0860954613097033e-07,
      "loss": 0.4612,
      "step": 425790
    },
    {
      "epoch": 174.29,
      "grad_norm": 2.259657382965088,
      "learning_rate": 2.0854412882595517e-07,
      "loss": 0.4584,
      "step": 425800
    },
    {
      "epoch": 174.3,
      "grad_norm": 1.8272055387496948,
      "learning_rate": 2.0847872130198762e-07,
      "loss": 0.4522,
      "step": 425810
    },
    {
      "epoch": 174.3,
      "grad_norm": 2.11588191986084,
      "learning_rate": 2.0841332355936688e-07,
      "loss": 0.4499,
      "step": 425820
    },
    {
      "epoch": 174.31,
      "grad_norm": 1.9356937408447266,
      "learning_rate": 2.0834793559839297e-07,
      "loss": 0.4232,
      "step": 425830
    },
    {
      "epoch": 174.31,
      "grad_norm": 1.9789206981658936,
      "learning_rate": 2.082825574193649e-07,
      "loss": 0.4485,
      "step": 425840
    },
    {
      "epoch": 174.31,
      "grad_norm": 2.2396485805511475,
      "learning_rate": 2.0821718902258268e-07,
      "loss": 0.4598,
      "step": 425850
    },
    {
      "epoch": 174.32,
      "grad_norm": 1.809071660041809,
      "learning_rate": 2.0815183040834555e-07,
      "loss": 0.4404,
      "step": 425860
    },
    {
      "epoch": 174.32,
      "grad_norm": 2.0546205043792725,
      "learning_rate": 2.08086481576953e-07,
      "loss": 0.4514,
      "step": 425870
    },
    {
      "epoch": 174.33,
      "grad_norm": 1.8629474639892578,
      "learning_rate": 2.08021142528704e-07,
      "loss": 0.4562,
      "step": 425880
    },
    {
      "epoch": 174.33,
      "grad_norm": 2.0658175945281982,
      "learning_rate": 2.0795581326389782e-07,
      "loss": 0.4606,
      "step": 425890
    },
    {
      "epoch": 174.33,
      "grad_norm": 2.057161569595337,
      "learning_rate": 2.0789049378283447e-07,
      "loss": 0.4488,
      "step": 425900
    },
    {
      "epoch": 174.34,
      "grad_norm": 1.9096782207489014,
      "learning_rate": 2.0782518408581265e-07,
      "loss": 0.4367,
      "step": 425910
    },
    {
      "epoch": 174.34,
      "grad_norm": 2.280864715576172,
      "learning_rate": 2.0775988417313243e-07,
      "loss": 0.4361,
      "step": 425920
    },
    {
      "epoch": 174.35,
      "grad_norm": 2.027280807495117,
      "learning_rate": 2.0769459404509115e-07,
      "loss": 0.4457,
      "step": 425930
    },
    {
      "epoch": 174.35,
      "grad_norm": 1.7607640027999878,
      "learning_rate": 2.0762931370198918e-07,
      "loss": 0.4429,
      "step": 425940
    },
    {
      "epoch": 174.36,
      "grad_norm": 2.0594382286071777,
      "learning_rate": 2.0756404314412542e-07,
      "loss": 0.4299,
      "step": 425950
    },
    {
      "epoch": 174.36,
      "grad_norm": 1.7622355222702026,
      "learning_rate": 2.0749878237179836e-07,
      "loss": 0.4458,
      "step": 425960
    },
    {
      "epoch": 174.36,
      "grad_norm": 2.9971354007720947,
      "learning_rate": 2.0743353138530753e-07,
      "loss": 0.4482,
      "step": 425970
    },
    {
      "epoch": 174.37,
      "grad_norm": 2.266136407852173,
      "learning_rate": 2.0736829018495137e-07,
      "loss": 0.4496,
      "step": 425980
    },
    {
      "epoch": 174.37,
      "grad_norm": 2.1629738807678223,
      "learning_rate": 2.073030587710291e-07,
      "loss": 0.4344,
      "step": 425990
    },
    {
      "epoch": 174.38,
      "grad_norm": 2.4437801837921143,
      "learning_rate": 2.0723783714383944e-07,
      "loss": 0.4404,
      "step": 426000
    },
    {
      "epoch": 174.38,
      "grad_norm": 1.8350428342819214,
      "learning_rate": 2.0717262530368113e-07,
      "loss": 0.4505,
      "step": 426010
    },
    {
      "epoch": 174.38,
      "grad_norm": 2.020714044570923,
      "learning_rate": 2.0710742325085289e-07,
      "loss": 0.4498,
      "step": 426020
    },
    {
      "epoch": 174.39,
      "grad_norm": 2.2680740356445312,
      "learning_rate": 2.0704223098565314e-07,
      "loss": 0.456,
      "step": 426030
    },
    {
      "epoch": 174.39,
      "grad_norm": 2.316119432449341,
      "learning_rate": 2.0697704850838092e-07,
      "loss": 0.4577,
      "step": 426040
    },
    {
      "epoch": 174.4,
      "grad_norm": 2.0550389289855957,
      "learning_rate": 2.0691187581933464e-07,
      "loss": 0.4593,
      "step": 426050
    },
    {
      "epoch": 174.4,
      "grad_norm": 2.1077940464019775,
      "learning_rate": 2.0684671291881305e-07,
      "loss": 0.4431,
      "step": 426060
    },
    {
      "epoch": 174.4,
      "grad_norm": 2.0213873386383057,
      "learning_rate": 2.0678155980711434e-07,
      "loss": 0.4424,
      "step": 426070
    },
    {
      "epoch": 174.41,
      "grad_norm": 2.3063833713531494,
      "learning_rate": 2.0671641648453721e-07,
      "loss": 0.4564,
      "step": 426080
    },
    {
      "epoch": 174.41,
      "grad_norm": 1.9772825241088867,
      "learning_rate": 2.066512829513799e-07,
      "loss": 0.4252,
      "step": 426090
    },
    {
      "epoch": 174.42,
      "grad_norm": 2.458667278289795,
      "learning_rate": 2.0658615920794083e-07,
      "loss": 0.4581,
      "step": 426100
    },
    {
      "epoch": 174.42,
      "grad_norm": 2.1580028533935547,
      "learning_rate": 2.065210452545182e-07,
      "loss": 0.4647,
      "step": 426110
    },
    {
      "epoch": 174.42,
      "grad_norm": 2.0787007808685303,
      "learning_rate": 2.0645594109141053e-07,
      "loss": 0.4475,
      "step": 426120
    },
    {
      "epoch": 174.43,
      "grad_norm": 2.443535804748535,
      "learning_rate": 2.0639084671891594e-07,
      "loss": 0.4358,
      "step": 426130
    },
    {
      "epoch": 174.43,
      "grad_norm": 2.422372341156006,
      "learning_rate": 2.0632576213733266e-07,
      "loss": 0.44,
      "step": 426140
    },
    {
      "epoch": 174.44,
      "grad_norm": 2.0994250774383545,
      "learning_rate": 2.0626068734695888e-07,
      "loss": 0.4266,
      "step": 426150
    },
    {
      "epoch": 174.44,
      "grad_norm": 1.794236183166504,
      "learning_rate": 2.061956223480923e-07,
      "loss": 0.4482,
      "step": 426160
    },
    {
      "epoch": 174.45,
      "grad_norm": 2.1387875080108643,
      "learning_rate": 2.0613056714103162e-07,
      "loss": 0.4258,
      "step": 426170
    },
    {
      "epoch": 174.45,
      "grad_norm": 2.1091718673706055,
      "learning_rate": 2.0606552172607428e-07,
      "loss": 0.4345,
      "step": 426180
    },
    {
      "epoch": 174.45,
      "grad_norm": 1.8978207111358643,
      "learning_rate": 2.0600048610351847e-07,
      "loss": 0.4348,
      "step": 426190
    },
    {
      "epoch": 174.46,
      "grad_norm": 1.787451148033142,
      "learning_rate": 2.0593546027366237e-07,
      "loss": 0.4364,
      "step": 426200
    },
    {
      "epoch": 174.46,
      "grad_norm": 2.340643882751465,
      "learning_rate": 2.058704442368034e-07,
      "loss": 0.444,
      "step": 426210
    },
    {
      "epoch": 174.47,
      "grad_norm": 1.7538145780563354,
      "learning_rate": 2.0580543799323955e-07,
      "loss": 0.4344,
      "step": 426220
    },
    {
      "epoch": 174.47,
      "grad_norm": 2.022773504257202,
      "learning_rate": 2.057404415432692e-07,
      "loss": 0.4456,
      "step": 426230
    },
    {
      "epoch": 174.47,
      "grad_norm": 2.4719603061676025,
      "learning_rate": 2.0567545488718906e-07,
      "loss": 0.4387,
      "step": 426240
    },
    {
      "epoch": 174.48,
      "grad_norm": 1.9558804035186768,
      "learning_rate": 2.0561047802529702e-07,
      "loss": 0.4365,
      "step": 426250
    },
    {
      "epoch": 174.48,
      "grad_norm": 1.6422414779663086,
      "learning_rate": 2.055455109578913e-07,
      "loss": 0.4444,
      "step": 426260
    },
    {
      "epoch": 174.49,
      "grad_norm": 2.095536947250366,
      "learning_rate": 2.0548055368526877e-07,
      "loss": 0.464,
      "step": 426270
    },
    {
      "epoch": 174.49,
      "grad_norm": 1.8582844734191895,
      "learning_rate": 2.0541560620772767e-07,
      "loss": 0.4507,
      "step": 426280
    },
    {
      "epoch": 174.49,
      "grad_norm": 1.847520351409912,
      "learning_rate": 2.0535066852556543e-07,
      "loss": 0.4644,
      "step": 426290
    },
    {
      "epoch": 174.5,
      "grad_norm": 1.8589640855789185,
      "learning_rate": 2.052857406390794e-07,
      "loss": 0.4352,
      "step": 426300
    },
    {
      "epoch": 174.5,
      "grad_norm": 2.1305267810821533,
      "learning_rate": 2.052208225485668e-07,
      "loss": 0.4657,
      "step": 426310
    },
    {
      "epoch": 174.51,
      "grad_norm": 2.1791253089904785,
      "learning_rate": 2.0515591425432526e-07,
      "loss": 0.4418,
      "step": 426320
    },
    {
      "epoch": 174.51,
      "grad_norm": 2.2771289348602295,
      "learning_rate": 2.0509101575665199e-07,
      "loss": 0.4473,
      "step": 426330
    },
    {
      "epoch": 174.51,
      "grad_norm": 2.507080078125,
      "learning_rate": 2.0502612705584412e-07,
      "loss": 0.4404,
      "step": 426340
    },
    {
      "epoch": 174.52,
      "grad_norm": 1.9411109685897827,
      "learning_rate": 2.0496124815219933e-07,
      "loss": 0.4415,
      "step": 426350
    },
    {
      "epoch": 174.52,
      "grad_norm": 2.4149937629699707,
      "learning_rate": 2.0489637904601502e-07,
      "loss": 0.4354,
      "step": 426360
    },
    {
      "epoch": 174.53,
      "grad_norm": 1.8439536094665527,
      "learning_rate": 2.0483151973758734e-07,
      "loss": 0.4439,
      "step": 426370
    },
    {
      "epoch": 174.53,
      "grad_norm": 2.11159610748291,
      "learning_rate": 2.0476667022721422e-07,
      "loss": 0.4496,
      "step": 426380
    },
    {
      "epoch": 174.54,
      "grad_norm": 2.140355110168457,
      "learning_rate": 2.047018305151923e-07,
      "loss": 0.4411,
      "step": 426390
    },
    {
      "epoch": 174.54,
      "grad_norm": 2.601638078689575,
      "learning_rate": 2.0463700060181872e-07,
      "loss": 0.4492,
      "step": 426400
    },
    {
      "epoch": 174.54,
      "grad_norm": 1.8757495880126953,
      "learning_rate": 2.0457218048739039e-07,
      "loss": 0.4403,
      "step": 426410
    },
    {
      "epoch": 174.55,
      "grad_norm": 2.0027217864990234,
      "learning_rate": 2.0450737017220472e-07,
      "loss": 0.4593,
      "step": 426420
    },
    {
      "epoch": 174.55,
      "grad_norm": 2.1649892330169678,
      "learning_rate": 2.0444256965655784e-07,
      "loss": 0.4388,
      "step": 426430
    },
    {
      "epoch": 174.56,
      "grad_norm": 2.1410481929779053,
      "learning_rate": 2.0437777894074716e-07,
      "loss": 0.4475,
      "step": 426440
    },
    {
      "epoch": 174.56,
      "grad_norm": 1.8998398780822754,
      "learning_rate": 2.0431299802506932e-07,
      "loss": 0.4498,
      "step": 426450
    },
    {
      "epoch": 174.56,
      "grad_norm": 2.003277063369751,
      "learning_rate": 2.042482269098207e-07,
      "loss": 0.444,
      "step": 426460
    },
    {
      "epoch": 174.57,
      "grad_norm": 2.0415947437286377,
      "learning_rate": 2.0418346559529871e-07,
      "loss": 0.445,
      "step": 426470
    },
    {
      "epoch": 174.57,
      "grad_norm": 2.3694875240325928,
      "learning_rate": 2.0411871408179977e-07,
      "loss": 0.4266,
      "step": 426480
    },
    {
      "epoch": 174.58,
      "grad_norm": 2.3877620697021484,
      "learning_rate": 2.0405397236962018e-07,
      "loss": 0.4497,
      "step": 426490
    },
    {
      "epoch": 174.58,
      "grad_norm": 1.9077966213226318,
      "learning_rate": 2.0398924045905717e-07,
      "loss": 0.4491,
      "step": 426500
    },
    {
      "epoch": 174.58,
      "grad_norm": 2.54921555519104,
      "learning_rate": 2.0392451835040654e-07,
      "loss": 0.463,
      "step": 426510
    },
    {
      "epoch": 174.59,
      "grad_norm": 2.344135284423828,
      "learning_rate": 2.0385980604396497e-07,
      "loss": 0.4421,
      "step": 426520
    },
    {
      "epoch": 174.59,
      "grad_norm": 1.8099846839904785,
      "learning_rate": 2.0379510354002907e-07,
      "loss": 0.4523,
      "step": 426530
    },
    {
      "epoch": 174.6,
      "grad_norm": 2.151034116744995,
      "learning_rate": 2.0373041083889496e-07,
      "loss": 0.4545,
      "step": 426540
    },
    {
      "epoch": 174.6,
      "grad_norm": 2.537123441696167,
      "learning_rate": 2.0366572794085932e-07,
      "loss": 0.4579,
      "step": 426550
    },
    {
      "epoch": 174.6,
      "grad_norm": 2.1647346019744873,
      "learning_rate": 2.0360105484621822e-07,
      "loss": 0.4493,
      "step": 426560
    },
    {
      "epoch": 174.61,
      "grad_norm": 2.0709564685821533,
      "learning_rate": 2.0353639155526778e-07,
      "loss": 0.445,
      "step": 426570
    },
    {
      "epoch": 174.61,
      "grad_norm": 2.1838550567626953,
      "learning_rate": 2.0347173806830467e-07,
      "loss": 0.4599,
      "step": 426580
    },
    {
      "epoch": 174.62,
      "grad_norm": 2.153705596923828,
      "learning_rate": 2.034070943856247e-07,
      "loss": 0.4415,
      "step": 426590
    },
    {
      "epoch": 174.62,
      "grad_norm": 1.77214777469635,
      "learning_rate": 2.033424605075243e-07,
      "loss": 0.4551,
      "step": 426600
    },
    {
      "epoch": 174.63,
      "grad_norm": 1.879585862159729,
      "learning_rate": 2.032778364342993e-07,
      "loss": 0.4296,
      "step": 426610
    },
    {
      "epoch": 174.63,
      "grad_norm": 2.032594919204712,
      "learning_rate": 2.032132221662458e-07,
      "loss": 0.4388,
      "step": 426620
    },
    {
      "epoch": 174.63,
      "grad_norm": 1.9502005577087402,
      "learning_rate": 2.0314861770365968e-07,
      "loss": 0.438,
      "step": 426630
    },
    {
      "epoch": 174.64,
      "grad_norm": 1.8338419198989868,
      "learning_rate": 2.0308402304683706e-07,
      "loss": 0.4512,
      "step": 426640
    },
    {
      "epoch": 174.64,
      "grad_norm": 1.8970341682434082,
      "learning_rate": 2.0301943819607354e-07,
      "loss": 0.4336,
      "step": 426650
    },
    {
      "epoch": 174.65,
      "grad_norm": 2.0877928733825684,
      "learning_rate": 2.0295486315166547e-07,
      "loss": 0.462,
      "step": 426660
    },
    {
      "epoch": 174.65,
      "grad_norm": 1.8368501663208008,
      "learning_rate": 2.0289029791390821e-07,
      "loss": 0.4311,
      "step": 426670
    },
    {
      "epoch": 174.65,
      "grad_norm": 2.099745273590088,
      "learning_rate": 2.028257424830976e-07,
      "loss": 0.4291,
      "step": 426680
    },
    {
      "epoch": 174.66,
      "grad_norm": 1.9633541107177734,
      "learning_rate": 2.027611968595298e-07,
      "loss": 0.4453,
      "step": 426690
    },
    {
      "epoch": 174.66,
      "grad_norm": 2.2833266258239746,
      "learning_rate": 2.0269666104349982e-07,
      "loss": 0.443,
      "step": 426700
    },
    {
      "epoch": 174.67,
      "grad_norm": 2.4355578422546387,
      "learning_rate": 2.0263213503530383e-07,
      "loss": 0.4543,
      "step": 426710
    },
    {
      "epoch": 174.67,
      "grad_norm": 2.025344133377075,
      "learning_rate": 2.0256761883523688e-07,
      "loss": 0.4431,
      "step": 426720
    },
    {
      "epoch": 174.67,
      "grad_norm": 2.4589638710021973,
      "learning_rate": 2.025031124435951e-07,
      "loss": 0.4294,
      "step": 426730
    },
    {
      "epoch": 174.68,
      "grad_norm": 2.5600876808166504,
      "learning_rate": 2.0243861586067355e-07,
      "loss": 0.4371,
      "step": 426740
    },
    {
      "epoch": 174.68,
      "grad_norm": 2.161746025085449,
      "learning_rate": 2.0237412908676786e-07,
      "loss": 0.4475,
      "step": 426750
    },
    {
      "epoch": 174.69,
      "grad_norm": 1.770696759223938,
      "learning_rate": 2.0230965212217333e-07,
      "loss": 0.4382,
      "step": 426760
    },
    {
      "epoch": 174.69,
      "grad_norm": 1.8397077322006226,
      "learning_rate": 2.0224518496718558e-07,
      "loss": 0.4404,
      "step": 426770
    },
    {
      "epoch": 174.7,
      "grad_norm": 2.3847525119781494,
      "learning_rate": 2.0218072762209969e-07,
      "loss": 0.4469,
      "step": 426780
    },
    {
      "epoch": 174.7,
      "grad_norm": 1.8449876308441162,
      "learning_rate": 2.0211628008721098e-07,
      "loss": 0.4434,
      "step": 426790
    },
    {
      "epoch": 174.7,
      "grad_norm": 2.028693675994873,
      "learning_rate": 2.0205184236281452e-07,
      "loss": 0.4174,
      "step": 426800
    },
    {
      "epoch": 174.71,
      "grad_norm": 1.902061939239502,
      "learning_rate": 2.0198741444920618e-07,
      "loss": 0.4429,
      "step": 426810
    },
    {
      "epoch": 174.71,
      "grad_norm": 1.868886947631836,
      "learning_rate": 2.0192299634668e-07,
      "loss": 0.4471,
      "step": 426820
    },
    {
      "epoch": 174.72,
      "grad_norm": 1.7912577390670776,
      "learning_rate": 2.0185858805553158e-07,
      "loss": 0.4286,
      "step": 426830
    },
    {
      "epoch": 174.72,
      "grad_norm": 1.9364794492721558,
      "learning_rate": 2.017941895760557e-07,
      "loss": 0.4618,
      "step": 426840
    },
    {
      "epoch": 174.72,
      "grad_norm": 2.374741554260254,
      "learning_rate": 2.017298009085483e-07,
      "loss": 0.4625,
      "step": 426850
    },
    {
      "epoch": 174.73,
      "grad_norm": 2.0705604553222656,
      "learning_rate": 2.0166542205330333e-07,
      "loss": 0.44,
      "step": 426860
    },
    {
      "epoch": 174.73,
      "grad_norm": 2.140828847885132,
      "learning_rate": 2.016010530106164e-07,
      "loss": 0.4369,
      "step": 426870
    },
    {
      "epoch": 174.74,
      "grad_norm": 2.1529698371887207,
      "learning_rate": 2.0153669378078185e-07,
      "loss": 0.434,
      "step": 426880
    },
    {
      "epoch": 174.74,
      "grad_norm": 4.152282238006592,
      "learning_rate": 2.014723443640947e-07,
      "loss": 0.4465,
      "step": 426890
    },
    {
      "epoch": 174.74,
      "grad_norm": 1.920588493347168,
      "learning_rate": 2.0140800476084956e-07,
      "loss": 0.4332,
      "step": 426900
    },
    {
      "epoch": 174.75,
      "grad_norm": 1.839089035987854,
      "learning_rate": 2.0134367497134148e-07,
      "loss": 0.4477,
      "step": 426910
    },
    {
      "epoch": 174.75,
      "grad_norm": 1.9224262237548828,
      "learning_rate": 2.0127935499586503e-07,
      "loss": 0.4557,
      "step": 426920
    },
    {
      "epoch": 174.76,
      "grad_norm": 2.045452833175659,
      "learning_rate": 2.0121504483471473e-07,
      "loss": 0.4363,
      "step": 426930
    },
    {
      "epoch": 174.76,
      "grad_norm": 2.1327505111694336,
      "learning_rate": 2.0115074448818542e-07,
      "loss": 0.4329,
      "step": 426940
    },
    {
      "epoch": 174.76,
      "grad_norm": 2.0564939975738525,
      "learning_rate": 2.0108645395657168e-07,
      "loss": 0.4483,
      "step": 426950
    },
    {
      "epoch": 174.77,
      "grad_norm": 1.7427231073379517,
      "learning_rate": 2.0102217324016751e-07,
      "loss": 0.4298,
      "step": 426960
    },
    {
      "epoch": 174.77,
      "grad_norm": 2.075601577758789,
      "learning_rate": 2.0095790233926775e-07,
      "loss": 0.4488,
      "step": 426970
    },
    {
      "epoch": 174.78,
      "grad_norm": 2.052694082260132,
      "learning_rate": 2.0089364125416642e-07,
      "loss": 0.4436,
      "step": 426980
    },
    {
      "epoch": 174.78,
      "grad_norm": 2.0306289196014404,
      "learning_rate": 2.008293899851586e-07,
      "loss": 0.4522,
      "step": 426990
    },
    {
      "epoch": 174.79,
      "grad_norm": 2.3419933319091797,
      "learning_rate": 2.0076514853253779e-07,
      "loss": 0.432,
      "step": 427000
    },
    {
      "epoch": 174.79,
      "grad_norm": 2.29548716545105,
      "learning_rate": 2.007009168965991e-07,
      "loss": 0.4556,
      "step": 427010
    },
    {
      "epoch": 174.79,
      "grad_norm": 1.9792509078979492,
      "learning_rate": 2.0063669507763576e-07,
      "loss": 0.4508,
      "step": 427020
    },
    {
      "epoch": 174.8,
      "grad_norm": 1.928480863571167,
      "learning_rate": 2.0057248307594284e-07,
      "loss": 0.4406,
      "step": 427030
    },
    {
      "epoch": 174.8,
      "grad_norm": 2.372178316116333,
      "learning_rate": 2.005082808918144e-07,
      "loss": 0.423,
      "step": 427040
    },
    {
      "epoch": 174.81,
      "grad_norm": 1.9033269882202148,
      "learning_rate": 2.004440885255445e-07,
      "loss": 0.4542,
      "step": 427050
    },
    {
      "epoch": 174.81,
      "grad_norm": 2.4023232460021973,
      "learning_rate": 2.0037990597742684e-07,
      "loss": 0.4581,
      "step": 427060
    },
    {
      "epoch": 174.81,
      "grad_norm": 1.8710438013076782,
      "learning_rate": 2.0031573324775578e-07,
      "loss": 0.433,
      "step": 427070
    },
    {
      "epoch": 174.82,
      "grad_norm": 1.7127386331558228,
      "learning_rate": 2.002515703368253e-07,
      "loss": 0.4422,
      "step": 427080
    },
    {
      "epoch": 174.82,
      "grad_norm": 1.7763258218765259,
      "learning_rate": 2.0018741724492897e-07,
      "loss": 0.4568,
      "step": 427090
    },
    {
      "epoch": 174.83,
      "grad_norm": 1.8320223093032837,
      "learning_rate": 2.001232739723608e-07,
      "loss": 0.441,
      "step": 427100
    },
    {
      "epoch": 174.83,
      "grad_norm": 2.0417873859405518,
      "learning_rate": 2.000591405194146e-07,
      "loss": 0.4329,
      "step": 427110
    },
    {
      "epoch": 174.83,
      "grad_norm": 1.958196759223938,
      "learning_rate": 1.9999501688638431e-07,
      "loss": 0.4349,
      "step": 427120
    },
    {
      "epoch": 174.84,
      "grad_norm": 2.0876264572143555,
      "learning_rate": 1.9993090307356355e-07,
      "loss": 0.4851,
      "step": 427130
    },
    {
      "epoch": 174.84,
      "grad_norm": 1.6106492280960083,
      "learning_rate": 1.9986679908124606e-07,
      "loss": 0.4436,
      "step": 427140
    },
    {
      "epoch": 174.85,
      "grad_norm": 2.19618821144104,
      "learning_rate": 1.9980270490972558e-07,
      "loss": 0.4665,
      "step": 427150
    },
    {
      "epoch": 174.85,
      "grad_norm": 2.0212795734405518,
      "learning_rate": 1.9973862055929568e-07,
      "loss": 0.4526,
      "step": 427160
    },
    {
      "epoch": 174.85,
      "grad_norm": 2.1180496215820312,
      "learning_rate": 1.9967454603024984e-07,
      "loss": 0.4412,
      "step": 427170
    },
    {
      "epoch": 174.86,
      "grad_norm": 1.9065263271331787,
      "learning_rate": 1.9961048132288133e-07,
      "loss": 0.4248,
      "step": 427180
    },
    {
      "epoch": 174.86,
      "grad_norm": 2.153944253921509,
      "learning_rate": 1.995464264374842e-07,
      "loss": 0.4521,
      "step": 427190
    },
    {
      "epoch": 174.87,
      "grad_norm": 1.8242084980010986,
      "learning_rate": 1.994823813743514e-07,
      "loss": 0.4562,
      "step": 427200
    },
    {
      "epoch": 174.87,
      "grad_norm": 2.0882554054260254,
      "learning_rate": 1.994183461337762e-07,
      "loss": 0.4416,
      "step": 427210
    },
    {
      "epoch": 174.88,
      "grad_norm": 2.1560938358306885,
      "learning_rate": 1.9935432071605293e-07,
      "loss": 0.4567,
      "step": 427220
    },
    {
      "epoch": 174.88,
      "grad_norm": 2.050292730331421,
      "learning_rate": 1.992903051214738e-07,
      "loss": 0.4562,
      "step": 427230
    },
    {
      "epoch": 174.88,
      "grad_norm": 2.1265556812286377,
      "learning_rate": 1.9922629935033252e-07,
      "loss": 0.4254,
      "step": 427240
    },
    {
      "epoch": 174.89,
      "grad_norm": 1.8395538330078125,
      "learning_rate": 1.991623034029219e-07,
      "loss": 0.4658,
      "step": 427250
    },
    {
      "epoch": 174.89,
      "grad_norm": 1.9454835653305054,
      "learning_rate": 1.9909831727953568e-07,
      "loss": 0.4379,
      "step": 427260
    },
    {
      "epoch": 174.9,
      "grad_norm": 2.0787601470947266,
      "learning_rate": 1.990343409804666e-07,
      "loss": 0.4341,
      "step": 427270
    },
    {
      "epoch": 174.9,
      "grad_norm": 2.011323928833008,
      "learning_rate": 1.9897037450600766e-07,
      "loss": 0.4298,
      "step": 427280
    },
    {
      "epoch": 174.9,
      "grad_norm": 2.5615689754486084,
      "learning_rate": 1.989064178564521e-07,
      "loss": 0.4401,
      "step": 427290
    },
    {
      "epoch": 174.91,
      "grad_norm": 2.205698013305664,
      "learning_rate": 1.9884247103209293e-07,
      "loss": 0.4488,
      "step": 427300
    },
    {
      "epoch": 174.91,
      "grad_norm": 1.9688478708267212,
      "learning_rate": 1.987785340332229e-07,
      "loss": 0.4409,
      "step": 427310
    },
    {
      "epoch": 174.92,
      "grad_norm": 2.351702928543091,
      "learning_rate": 1.9871460686013498e-07,
      "loss": 0.4602,
      "step": 427320
    },
    {
      "epoch": 174.92,
      "grad_norm": 2.064685821533203,
      "learning_rate": 1.9865068951312187e-07,
      "loss": 0.4452,
      "step": 427330
    },
    {
      "epoch": 174.92,
      "grad_norm": 1.7486722469329834,
      "learning_rate": 1.9858678199247663e-07,
      "loss": 0.4577,
      "step": 427340
    },
    {
      "epoch": 174.93,
      "grad_norm": 1.8806078433990479,
      "learning_rate": 1.98522884298492e-07,
      "loss": 0.4543,
      "step": 427350
    },
    {
      "epoch": 174.93,
      "grad_norm": 1.8697973489761353,
      "learning_rate": 1.9845899643146037e-07,
      "loss": 0.4377,
      "step": 427360
    },
    {
      "epoch": 174.94,
      "grad_norm": 3.4747989177703857,
      "learning_rate": 1.9839511839167457e-07,
      "loss": 0.4355,
      "step": 427370
    },
    {
      "epoch": 174.94,
      "grad_norm": 2.2125942707061768,
      "learning_rate": 1.9833125017942727e-07,
      "loss": 0.431,
      "step": 427380
    },
    {
      "epoch": 174.94,
      "grad_norm": 1.9290913343429565,
      "learning_rate": 1.982673917950115e-07,
      "loss": 0.4482,
      "step": 427390
    },
    {
      "epoch": 174.95,
      "grad_norm": 2.2806103229522705,
      "learning_rate": 1.9820354323871816e-07,
      "loss": 0.4368,
      "step": 427400
    },
    {
      "epoch": 174.95,
      "grad_norm": 2.3403656482696533,
      "learning_rate": 1.9813970451084153e-07,
      "loss": 0.4624,
      "step": 427410
    },
    {
      "epoch": 174.96,
      "grad_norm": 2.0329701900482178,
      "learning_rate": 1.9807587561167333e-07,
      "loss": 0.4587,
      "step": 427420
    },
    {
      "epoch": 174.96,
      "grad_norm": 1.970302700996399,
      "learning_rate": 1.9801205654150602e-07,
      "loss": 0.4501,
      "step": 427430
    },
    {
      "epoch": 174.97,
      "grad_norm": 1.9705816507339478,
      "learning_rate": 1.9794824730063207e-07,
      "loss": 0.4545,
      "step": 427440
    },
    {
      "epoch": 174.97,
      "grad_norm": 2.055379867553711,
      "learning_rate": 1.9788444788934347e-07,
      "loss": 0.4351,
      "step": 427450
    },
    {
      "epoch": 174.97,
      "grad_norm": 1.999912977218628,
      "learning_rate": 1.978206583079324e-07,
      "loss": 0.444,
      "step": 427460
    },
    {
      "epoch": 174.98,
      "grad_norm": 1.808607578277588,
      "learning_rate": 1.977568785566916e-07,
      "loss": 0.4556,
      "step": 427470
    },
    {
      "epoch": 174.98,
      "grad_norm": 2.0848774909973145,
      "learning_rate": 1.9769310863591302e-07,
      "loss": 0.4664,
      "step": 427480
    },
    {
      "epoch": 174.99,
      "grad_norm": 1.9010579586029053,
      "learning_rate": 1.9762934854588835e-07,
      "loss": 0.4295,
      "step": 427490
    },
    {
      "epoch": 174.99,
      "grad_norm": 5.700091361999512,
      "learning_rate": 1.9756559828691034e-07,
      "loss": 0.4576,
      "step": 427500
    },
    {
      "epoch": 174.99,
      "grad_norm": 2.4624669551849365,
      "learning_rate": 1.9750185785927067e-07,
      "loss": 0.4422,
      "step": 427510
    },
    {
      "epoch": 175.0,
      "grad_norm": 1.838122010231018,
      "learning_rate": 1.9743812726326105e-07,
      "loss": 0.4361,
      "step": 427520
    },
    {
      "epoch": 175.0,
      "eval_loss": 0.4445628821849823,
      "eval_runtime": 52.3798,
      "eval_samples_per_second": 65.846,
      "eval_steps_per_second": 8.247,
      "step": 427525
    },
    {
      "epoch": 175.0,
      "grad_norm": 2.611945152282715,
      "learning_rate": 1.9737440649917448e-07,
      "loss": 0.4571,
      "step": 427530
    },
    {
      "epoch": 175.01,
      "grad_norm": 2.0800883769989014,
      "learning_rate": 1.9731069556730158e-07,
      "loss": 0.4484,
      "step": 427540
    },
    {
      "epoch": 175.01,
      "grad_norm": 2.0579049587249756,
      "learning_rate": 1.972469944679346e-07,
      "loss": 0.4205,
      "step": 427550
    },
    {
      "epoch": 175.01,
      "grad_norm": 1.9187086820602417,
      "learning_rate": 1.9718330320136547e-07,
      "loss": 0.4572,
      "step": 427560
    },
    {
      "epoch": 175.02,
      "grad_norm": 2.0066497325897217,
      "learning_rate": 1.9711962176788614e-07,
      "loss": 0.4562,
      "step": 427570
    },
    {
      "epoch": 175.02,
      "grad_norm": 2.004131317138672,
      "learning_rate": 1.970559501677878e-07,
      "loss": 0.4497,
      "step": 427580
    },
    {
      "epoch": 175.03,
      "grad_norm": 1.9555680751800537,
      "learning_rate": 1.9699228840136241e-07,
      "loss": 0.4364,
      "step": 427590
    },
    {
      "epoch": 175.03,
      "grad_norm": 2.2515876293182373,
      "learning_rate": 1.969286364689019e-07,
      "loss": 0.4595,
      "step": 427600
    },
    {
      "epoch": 175.03,
      "grad_norm": 1.8435373306274414,
      "learning_rate": 1.9686499437069747e-07,
      "loss": 0.4523,
      "step": 427610
    },
    {
      "epoch": 175.04,
      "grad_norm": 4.434436798095703,
      "learning_rate": 1.9680136210704078e-07,
      "loss": 0.4442,
      "step": 427620
    },
    {
      "epoch": 175.04,
      "grad_norm": 1.6706933975219727,
      "learning_rate": 1.967377396782233e-07,
      "loss": 0.4689,
      "step": 427630
    },
    {
      "epoch": 175.05,
      "grad_norm": 2.193427324295044,
      "learning_rate": 1.9667412708453668e-07,
      "loss": 0.4351,
      "step": 427640
    },
    {
      "epoch": 175.05,
      "grad_norm": 1.9276213645935059,
      "learning_rate": 1.9661052432627188e-07,
      "loss": 0.4353,
      "step": 427650
    },
    {
      "epoch": 175.06,
      "grad_norm": 1.949127197265625,
      "learning_rate": 1.965469314037211e-07,
      "loss": 0.4326,
      "step": 427660
    },
    {
      "epoch": 175.06,
      "grad_norm": 1.806534767150879,
      "learning_rate": 1.9648334831717447e-07,
      "loss": 0.4497,
      "step": 427670
    },
    {
      "epoch": 175.06,
      "grad_norm": 1.9013676643371582,
      "learning_rate": 1.9641977506692393e-07,
      "loss": 0.4609,
      "step": 427680
    },
    {
      "epoch": 175.07,
      "grad_norm": 2.1890838146209717,
      "learning_rate": 1.963562116532604e-07,
      "loss": 0.4594,
      "step": 427690
    },
    {
      "epoch": 175.07,
      "grad_norm": 1.9260362386703491,
      "learning_rate": 1.9629265807647532e-07,
      "loss": 0.4423,
      "step": 427700
    },
    {
      "epoch": 175.08,
      "grad_norm": 2.09578800201416,
      "learning_rate": 1.9622911433685986e-07,
      "loss": 0.4687,
      "step": 427710
    },
    {
      "epoch": 175.08,
      "grad_norm": 2.235581874847412,
      "learning_rate": 1.9616558043470493e-07,
      "loss": 0.4459,
      "step": 427720
    },
    {
      "epoch": 175.08,
      "grad_norm": 2.0761334896087646,
      "learning_rate": 1.9610205637030172e-07,
      "loss": 0.442,
      "step": 427730
    },
    {
      "epoch": 175.09,
      "grad_norm": 1.5207676887512207,
      "learning_rate": 1.9603854214394113e-07,
      "loss": 0.4443,
      "step": 427740
    },
    {
      "epoch": 175.09,
      "grad_norm": 2.038457155227661,
      "learning_rate": 1.9597503775591405e-07,
      "loss": 0.4617,
      "step": 427750
    },
    {
      "epoch": 175.1,
      "grad_norm": 2.4210078716278076,
      "learning_rate": 1.9591154320651145e-07,
      "loss": 0.4547,
      "step": 427760
    },
    {
      "epoch": 175.1,
      "grad_norm": 3.024275064468384,
      "learning_rate": 1.9584805849602418e-07,
      "loss": 0.4471,
      "step": 427770
    },
    {
      "epoch": 175.1,
      "grad_norm": 2.4720094203948975,
      "learning_rate": 1.9578458362474267e-07,
      "loss": 0.4708,
      "step": 427780
    },
    {
      "epoch": 175.11,
      "grad_norm": 2.122879981994629,
      "learning_rate": 1.9572111859295836e-07,
      "loss": 0.4497,
      "step": 427790
    },
    {
      "epoch": 175.11,
      "grad_norm": 2.2563023567199707,
      "learning_rate": 1.956576634009624e-07,
      "loss": 0.4561,
      "step": 427800
    },
    {
      "epoch": 175.12,
      "grad_norm": 1.8323307037353516,
      "learning_rate": 1.9559421804904415e-07,
      "loss": 0.4317,
      "step": 427810
    },
    {
      "epoch": 175.12,
      "grad_norm": 2.0393753051757812,
      "learning_rate": 1.9553078253749477e-07,
      "loss": 0.4506,
      "step": 427820
    },
    {
      "epoch": 175.12,
      "grad_norm": 2.1164190769195557,
      "learning_rate": 1.9546735686660495e-07,
      "loss": 0.4374,
      "step": 427830
    },
    {
      "epoch": 175.13,
      "grad_norm": 1.8580032587051392,
      "learning_rate": 1.9540394103666532e-07,
      "loss": 0.4285,
      "step": 427840
    },
    {
      "epoch": 175.13,
      "grad_norm": 1.9302502870559692,
      "learning_rate": 1.9534053504796627e-07,
      "loss": 0.4526,
      "step": 427850
    },
    {
      "epoch": 175.14,
      "grad_norm": 2.031435966491699,
      "learning_rate": 1.9527713890079818e-07,
      "loss": 0.4399,
      "step": 427860
    },
    {
      "epoch": 175.14,
      "grad_norm": 2.1205430030822754,
      "learning_rate": 1.952137525954517e-07,
      "loss": 0.4257,
      "step": 427870
    },
    {
      "epoch": 175.15,
      "grad_norm": 2.099632978439331,
      "learning_rate": 1.9515037613221677e-07,
      "loss": 0.4193,
      "step": 427880
    },
    {
      "epoch": 175.15,
      "grad_norm": 2.2903904914855957,
      "learning_rate": 1.9508700951138424e-07,
      "loss": 0.432,
      "step": 427890
    },
    {
      "epoch": 175.15,
      "grad_norm": 2.10730242729187,
      "learning_rate": 1.95023652733244e-07,
      "loss": 0.4428,
      "step": 427900
    },
    {
      "epoch": 175.16,
      "grad_norm": 1.7582361698150635,
      "learning_rate": 1.9496030579808616e-07,
      "loss": 0.4487,
      "step": 427910
    },
    {
      "epoch": 175.16,
      "grad_norm": 1.8415340185165405,
      "learning_rate": 1.948969687062014e-07,
      "loss": 0.4245,
      "step": 427920
    },
    {
      "epoch": 175.17,
      "grad_norm": 1.6554309129714966,
      "learning_rate": 1.9483364145787957e-07,
      "loss": 0.4287,
      "step": 427930
    },
    {
      "epoch": 175.17,
      "grad_norm": 1.7901047468185425,
      "learning_rate": 1.947703240534106e-07,
      "loss": 0.4597,
      "step": 427940
    },
    {
      "epoch": 175.17,
      "grad_norm": 2.0857272148132324,
      "learning_rate": 1.9470701649308477e-07,
      "loss": 0.4402,
      "step": 427950
    },
    {
      "epoch": 175.18,
      "grad_norm": 2.114187240600586,
      "learning_rate": 1.9464371877719234e-07,
      "loss": 0.4524,
      "step": 427960
    },
    {
      "epoch": 175.18,
      "grad_norm": 1.930547833442688,
      "learning_rate": 1.9458043090602258e-07,
      "loss": 0.4343,
      "step": 427970
    },
    {
      "epoch": 175.19,
      "grad_norm": 2.176588296890259,
      "learning_rate": 1.9451715287986592e-07,
      "loss": 0.4578,
      "step": 427980
    },
    {
      "epoch": 175.19,
      "grad_norm": 2.1807925701141357,
      "learning_rate": 1.9445388469901193e-07,
      "loss": 0.445,
      "step": 427990
    },
    {
      "epoch": 175.19,
      "grad_norm": 2.206238269805908,
      "learning_rate": 1.9439062636375079e-07,
      "loss": 0.4324,
      "step": 428000
    },
    {
      "epoch": 175.2,
      "grad_norm": 1.5697680711746216,
      "learning_rate": 1.9432737787437208e-07,
      "loss": 0.4407,
      "step": 428010
    },
    {
      "epoch": 175.2,
      "grad_norm": 1.8753817081451416,
      "learning_rate": 1.9426413923116546e-07,
      "loss": 0.435,
      "step": 428020
    },
    {
      "epoch": 175.21,
      "grad_norm": 2.3495490550994873,
      "learning_rate": 1.9420091043442049e-07,
      "loss": 0.4537,
      "step": 428030
    },
    {
      "epoch": 175.21,
      "grad_norm": 2.045548439025879,
      "learning_rate": 1.941376914844273e-07,
      "loss": 0.452,
      "step": 428040
    },
    {
      "epoch": 175.21,
      "grad_norm": 2.1201817989349365,
      "learning_rate": 1.94074482381475e-07,
      "loss": 0.4421,
      "step": 428050
    },
    {
      "epoch": 175.22,
      "grad_norm": 2.163370370864868,
      "learning_rate": 1.940112831258532e-07,
      "loss": 0.4361,
      "step": 428060
    },
    {
      "epoch": 175.22,
      "grad_norm": 2.721618175506592,
      "learning_rate": 1.9394809371785182e-07,
      "loss": 0.4539,
      "step": 428070
    },
    {
      "epoch": 175.23,
      "grad_norm": 2.484626054763794,
      "learning_rate": 1.9388491415775988e-07,
      "loss": 0.4433,
      "step": 428080
    },
    {
      "epoch": 175.23,
      "grad_norm": 1.997215747833252,
      "learning_rate": 1.9382174444586702e-07,
      "loss": 0.4362,
      "step": 428090
    },
    {
      "epoch": 175.24,
      "grad_norm": 2.2330305576324463,
      "learning_rate": 1.9375858458246258e-07,
      "loss": 0.4403,
      "step": 428100
    },
    {
      "epoch": 175.24,
      "grad_norm": 2.3622987270355225,
      "learning_rate": 1.936954345678362e-07,
      "loss": 0.4556,
      "step": 428110
    },
    {
      "epoch": 175.24,
      "grad_norm": 1.7368783950805664,
      "learning_rate": 1.936322944022764e-07,
      "loss": 0.4564,
      "step": 428120
    },
    {
      "epoch": 175.25,
      "grad_norm": 2.1870782375335693,
      "learning_rate": 1.9356916408607262e-07,
      "loss": 0.4464,
      "step": 428130
    },
    {
      "epoch": 175.25,
      "grad_norm": 1.9997926950454712,
      "learning_rate": 1.9350604361951464e-07,
      "loss": 0.4292,
      "step": 428140
    },
    {
      "epoch": 175.26,
      "grad_norm": 2.095278739929199,
      "learning_rate": 1.9344293300289083e-07,
      "loss": 0.4296,
      "step": 428150
    },
    {
      "epoch": 175.26,
      "grad_norm": 2.0801033973693848,
      "learning_rate": 1.9337983223649051e-07,
      "loss": 0.4345,
      "step": 428160
    },
    {
      "epoch": 175.26,
      "grad_norm": 2.439133405685425,
      "learning_rate": 1.933167413206033e-07,
      "loss": 0.4354,
      "step": 428170
    },
    {
      "epoch": 175.27,
      "grad_norm": 2.402708053588867,
      "learning_rate": 1.932536602555178e-07,
      "loss": 0.4435,
      "step": 428180
    },
    {
      "epoch": 175.27,
      "grad_norm": 1.8495439291000366,
      "learning_rate": 1.9319058904152276e-07,
      "loss": 0.4548,
      "step": 428190
    },
    {
      "epoch": 175.28,
      "grad_norm": 1.9282876253128052,
      "learning_rate": 1.9312752767890764e-07,
      "loss": 0.4533,
      "step": 428200
    },
    {
      "epoch": 175.28,
      "grad_norm": 1.9925087690353394,
      "learning_rate": 1.9306447616796094e-07,
      "loss": 0.4542,
      "step": 428210
    },
    {
      "epoch": 175.28,
      "grad_norm": 2.13338041305542,
      "learning_rate": 1.9300143450897152e-07,
      "loss": 0.4354,
      "step": 428220
    },
    {
      "epoch": 175.29,
      "grad_norm": 1.7493172883987427,
      "learning_rate": 1.929384027022282e-07,
      "loss": 0.4302,
      "step": 428230
    },
    {
      "epoch": 175.29,
      "grad_norm": 2.015557050704956,
      "learning_rate": 1.9287538074801955e-07,
      "loss": 0.4432,
      "step": 428240
    },
    {
      "epoch": 175.3,
      "grad_norm": 2.078066349029541,
      "learning_rate": 1.928123686466349e-07,
      "loss": 0.4482,
      "step": 428250
    },
    {
      "epoch": 175.3,
      "grad_norm": 3.6110479831695557,
      "learning_rate": 1.9274936639836211e-07,
      "loss": 0.4434,
      "step": 428260
    },
    {
      "epoch": 175.3,
      "grad_norm": 2.1157431602478027,
      "learning_rate": 1.9268637400348995e-07,
      "loss": 0.4218,
      "step": 428270
    },
    {
      "epoch": 175.31,
      "grad_norm": 2.0051374435424805,
      "learning_rate": 1.9262339146230728e-07,
      "loss": 0.4491,
      "step": 428280
    },
    {
      "epoch": 175.31,
      "grad_norm": 1.9953290224075317,
      "learning_rate": 1.9256041877510239e-07,
      "loss": 0.4539,
      "step": 428290
    },
    {
      "epoch": 175.32,
      "grad_norm": 1.8743809461593628,
      "learning_rate": 1.9249745594216384e-07,
      "loss": 0.453,
      "step": 428300
    },
    {
      "epoch": 175.32,
      "grad_norm": 5.99501371383667,
      "learning_rate": 1.9243450296378025e-07,
      "loss": 0.4281,
      "step": 428310
    },
    {
      "epoch": 175.33,
      "grad_norm": 2.197666883468628,
      "learning_rate": 1.9237155984023938e-07,
      "loss": 0.4511,
      "step": 428320
    },
    {
      "epoch": 175.33,
      "grad_norm": 2.073798179626465,
      "learning_rate": 1.9230862657183033e-07,
      "loss": 0.433,
      "step": 428330
    },
    {
      "epoch": 175.33,
      "grad_norm": 2.169748544692993,
      "learning_rate": 1.9224570315884035e-07,
      "loss": 0.4225,
      "step": 428340
    },
    {
      "epoch": 175.34,
      "grad_norm": 2.1138575077056885,
      "learning_rate": 1.921827896015588e-07,
      "loss": 0.4636,
      "step": 428350
    },
    {
      "epoch": 175.34,
      "grad_norm": 2.2750132083892822,
      "learning_rate": 1.921198859002735e-07,
      "loss": 0.4412,
      "step": 428360
    },
    {
      "epoch": 175.35,
      "grad_norm": 2.076366662979126,
      "learning_rate": 1.9205699205527245e-07,
      "loss": 0.4485,
      "step": 428370
    },
    {
      "epoch": 175.35,
      "grad_norm": 2.08992862701416,
      "learning_rate": 1.9199410806684428e-07,
      "loss": 0.4452,
      "step": 428380
    },
    {
      "epoch": 175.35,
      "grad_norm": 2.0530240535736084,
      "learning_rate": 1.919312339352762e-07,
      "loss": 0.425,
      "step": 428390
    },
    {
      "epoch": 175.36,
      "grad_norm": 1.8161354064941406,
      "learning_rate": 1.9186836966085658e-07,
      "loss": 0.4811,
      "step": 428400
    },
    {
      "epoch": 175.36,
      "grad_norm": 2.201889991760254,
      "learning_rate": 1.9180551524387368e-07,
      "loss": 0.4471,
      "step": 428410
    },
    {
      "epoch": 175.37,
      "grad_norm": 2.123518705368042,
      "learning_rate": 1.9174267068461482e-07,
      "loss": 0.4354,
      "step": 428420
    },
    {
      "epoch": 175.37,
      "grad_norm": 2.1556713581085205,
      "learning_rate": 1.9167983598336853e-07,
      "loss": 0.4357,
      "step": 428430
    },
    {
      "epoch": 175.37,
      "grad_norm": 2.3307952880859375,
      "learning_rate": 1.9161701114042234e-07,
      "loss": 0.4298,
      "step": 428440
    },
    {
      "epoch": 175.38,
      "grad_norm": 2.272967576980591,
      "learning_rate": 1.9155419615606407e-07,
      "loss": 0.4515,
      "step": 428450
    },
    {
      "epoch": 175.38,
      "grad_norm": 2.535410165786743,
      "learning_rate": 1.9149139103058148e-07,
      "loss": 0.4323,
      "step": 428460
    },
    {
      "epoch": 175.39,
      "grad_norm": 1.8556342124938965,
      "learning_rate": 1.9142859576426213e-07,
      "loss": 0.4557,
      "step": 428470
    },
    {
      "epoch": 175.39,
      "grad_norm": 2.057003974914551,
      "learning_rate": 1.9136581035739403e-07,
      "loss": 0.45,
      "step": 428480
    },
    {
      "epoch": 175.4,
      "grad_norm": 2.1942601203918457,
      "learning_rate": 1.9130303481026425e-07,
      "loss": 0.4444,
      "step": 428490
    },
    {
      "epoch": 175.4,
      "grad_norm": 2.3443782329559326,
      "learning_rate": 1.912402691231608e-07,
      "loss": 0.45,
      "step": 428500
    },
    {
      "epoch": 175.4,
      "grad_norm": 3.2093522548675537,
      "learning_rate": 1.911775132963712e-07,
      "loss": 0.4505,
      "step": 428510
    },
    {
      "epoch": 175.41,
      "grad_norm": 2.096933603286743,
      "learning_rate": 1.9111476733018276e-07,
      "loss": 0.4245,
      "step": 428520
    },
    {
      "epoch": 175.41,
      "grad_norm": 1.977731704711914,
      "learning_rate": 1.9105203122488297e-07,
      "loss": 0.4607,
      "step": 428530
    },
    {
      "epoch": 175.42,
      "grad_norm": 1.9279587268829346,
      "learning_rate": 1.9098930498075886e-07,
      "loss": 0.4562,
      "step": 428540
    },
    {
      "epoch": 175.42,
      "grad_norm": 2.7265751361846924,
      "learning_rate": 1.9092658859809847e-07,
      "loss": 0.4409,
      "step": 428550
    },
    {
      "epoch": 175.42,
      "grad_norm": 2.0498368740081787,
      "learning_rate": 1.9086388207718858e-07,
      "loss": 0.4504,
      "step": 428560
    },
    {
      "epoch": 175.43,
      "grad_norm": 2.5629067420959473,
      "learning_rate": 1.908011854183167e-07,
      "loss": 0.4373,
      "step": 428570
    },
    {
      "epoch": 175.43,
      "grad_norm": 1.87814462184906,
      "learning_rate": 1.9073849862176983e-07,
      "loss": 0.4595,
      "step": 428580
    },
    {
      "epoch": 175.44,
      "grad_norm": 1.852797508239746,
      "learning_rate": 1.90675821687835e-07,
      "loss": 0.4379,
      "step": 428590
    },
    {
      "epoch": 175.44,
      "grad_norm": 1.7971543073654175,
      "learning_rate": 1.9061315461679972e-07,
      "loss": 0.4482,
      "step": 428600
    },
    {
      "epoch": 175.44,
      "grad_norm": 2.3020060062408447,
      "learning_rate": 1.90550497408951e-07,
      "loss": 0.4289,
      "step": 428610
    },
    {
      "epoch": 175.45,
      "grad_norm": 2.204612970352173,
      "learning_rate": 1.904878500645756e-07,
      "loss": 0.4525,
      "step": 428620
    },
    {
      "epoch": 175.45,
      "grad_norm": 2.15008807182312,
      "learning_rate": 1.9042521258396054e-07,
      "loss": 0.4543,
      "step": 428630
    },
    {
      "epoch": 175.46,
      "grad_norm": 2.220508575439453,
      "learning_rate": 1.9036258496739307e-07,
      "loss": 0.4443,
      "step": 428640
    },
    {
      "epoch": 175.46,
      "grad_norm": 2.104626417160034,
      "learning_rate": 1.902999672151597e-07,
      "loss": 0.4537,
      "step": 428650
    },
    {
      "epoch": 175.46,
      "grad_norm": 2.209774971008301,
      "learning_rate": 1.9023735932754745e-07,
      "loss": 0.4518,
      "step": 428660
    },
    {
      "epoch": 175.47,
      "grad_norm": 2.068150043487549,
      "learning_rate": 1.90174761304843e-07,
      "loss": 0.4282,
      "step": 428670
    },
    {
      "epoch": 175.47,
      "grad_norm": 2.4648120403289795,
      "learning_rate": 1.9011217314733343e-07,
      "loss": 0.442,
      "step": 428680
    },
    {
      "epoch": 175.48,
      "grad_norm": 1.7623919248580933,
      "learning_rate": 1.900495948553055e-07,
      "loss": 0.4334,
      "step": 428690
    },
    {
      "epoch": 175.48,
      "grad_norm": 1.957578182220459,
      "learning_rate": 1.899870264290451e-07,
      "loss": 0.4507,
      "step": 428700
    },
    {
      "epoch": 175.49,
      "grad_norm": 2.4279043674468994,
      "learning_rate": 1.8992446786883957e-07,
      "loss": 0.4375,
      "step": 428710
    },
    {
      "epoch": 175.49,
      "grad_norm": 2.2014479637145996,
      "learning_rate": 1.898619191749746e-07,
      "loss": 0.4442,
      "step": 428720
    },
    {
      "epoch": 175.49,
      "grad_norm": 1.9473680257797241,
      "learning_rate": 1.897993803477377e-07,
      "loss": 0.4253,
      "step": 428730
    },
    {
      "epoch": 175.5,
      "grad_norm": 1.9144911766052246,
      "learning_rate": 1.8973685138741512e-07,
      "loss": 0.4444,
      "step": 428740
    },
    {
      "epoch": 175.5,
      "grad_norm": 2.0332694053649902,
      "learning_rate": 1.8967433229429336e-07,
      "loss": 0.4431,
      "step": 428750
    },
    {
      "epoch": 175.51,
      "grad_norm": 2.004826307296753,
      "learning_rate": 1.8961182306865838e-07,
      "loss": 0.4494,
      "step": 428760
    },
    {
      "epoch": 175.51,
      "grad_norm": 2.0233514308929443,
      "learning_rate": 1.8954932371079691e-07,
      "loss": 0.4349,
      "step": 428770
    },
    {
      "epoch": 175.51,
      "grad_norm": 1.8310874700546265,
      "learning_rate": 1.8948683422099496e-07,
      "loss": 0.4343,
      "step": 428780
    },
    {
      "epoch": 175.52,
      "grad_norm": 2.086674928665161,
      "learning_rate": 1.8942435459953924e-07,
      "loss": 0.4378,
      "step": 428790
    },
    {
      "epoch": 175.52,
      "grad_norm": 2.261833429336548,
      "learning_rate": 1.8936188484671544e-07,
      "loss": 0.4327,
      "step": 428800
    },
    {
      "epoch": 175.53,
      "grad_norm": 2.047008752822876,
      "learning_rate": 1.8929942496280987e-07,
      "loss": 0.4443,
      "step": 428810
    },
    {
      "epoch": 175.53,
      "grad_norm": 2.439070463180542,
      "learning_rate": 1.8923697494810894e-07,
      "loss": 0.4628,
      "step": 428820
    },
    {
      "epoch": 175.53,
      "grad_norm": 1.965994954109192,
      "learning_rate": 1.8917453480289893e-07,
      "loss": 0.4535,
      "step": 428830
    },
    {
      "epoch": 175.54,
      "grad_norm": 2.5467793941497803,
      "learning_rate": 1.8911210452746474e-07,
      "loss": 0.4559,
      "step": 428840
    },
    {
      "epoch": 175.54,
      "grad_norm": 1.940835952758789,
      "learning_rate": 1.890496841220934e-07,
      "loss": 0.4362,
      "step": 428850
    },
    {
      "epoch": 175.55,
      "grad_norm": 1.9706017971038818,
      "learning_rate": 1.8898727358707036e-07,
      "loss": 0.4574,
      "step": 428860
    },
    {
      "epoch": 175.55,
      "grad_norm": 1.9791088104248047,
      "learning_rate": 1.8892487292268156e-07,
      "loss": 0.4313,
      "step": 428870
    },
    {
      "epoch": 175.55,
      "grad_norm": 1.8135019540786743,
      "learning_rate": 1.8886248212921299e-07,
      "loss": 0.4393,
      "step": 428880
    },
    {
      "epoch": 175.56,
      "grad_norm": 1.8015159368515015,
      "learning_rate": 1.888001012069506e-07,
      "loss": 0.4617,
      "step": 428890
    },
    {
      "epoch": 175.56,
      "grad_norm": 1.9301798343658447,
      "learning_rate": 1.8873773015617988e-07,
      "loss": 0.4404,
      "step": 428900
    },
    {
      "epoch": 175.57,
      "grad_norm": 1.807543158531189,
      "learning_rate": 1.8867536897718626e-07,
      "loss": 0.4486,
      "step": 428910
    },
    {
      "epoch": 175.57,
      "grad_norm": 2.0359268188476562,
      "learning_rate": 1.8861301767025593e-07,
      "loss": 0.4512,
      "step": 428920
    },
    {
      "epoch": 175.58,
      "grad_norm": 2.045083999633789,
      "learning_rate": 1.8855067623567466e-07,
      "loss": 0.4597,
      "step": 428930
    },
    {
      "epoch": 175.58,
      "grad_norm": 1.9746860265731812,
      "learning_rate": 1.8848834467372732e-07,
      "loss": 0.4497,
      "step": 428940
    },
    {
      "epoch": 175.58,
      "grad_norm": 2.242105007171631,
      "learning_rate": 1.884260229847002e-07,
      "loss": 0.436,
      "step": 428950
    },
    {
      "epoch": 175.59,
      "grad_norm": 1.660286784172058,
      "learning_rate": 1.8836371116887845e-07,
      "loss": 0.4321,
      "step": 428960
    },
    {
      "epoch": 175.59,
      "grad_norm": 2.2399697303771973,
      "learning_rate": 1.8830140922654753e-07,
      "loss": 0.4695,
      "step": 428970
    },
    {
      "epoch": 175.6,
      "grad_norm": 2.4575936794281006,
      "learning_rate": 1.882391171579926e-07,
      "loss": 0.4604,
      "step": 428980
    },
    {
      "epoch": 175.6,
      "grad_norm": 2.2408535480499268,
      "learning_rate": 1.8817683496349943e-07,
      "loss": 0.4378,
      "step": 428990
    },
    {
      "epoch": 175.6,
      "grad_norm": 1.8542643785476685,
      "learning_rate": 1.8811456264335288e-07,
      "loss": 0.4308,
      "step": 429000
    },
    {
      "epoch": 175.61,
      "grad_norm": 1.6671948432922363,
      "learning_rate": 1.880523001978382e-07,
      "loss": 0.4348,
      "step": 429010
    },
    {
      "epoch": 175.61,
      "grad_norm": 2.119326591491699,
      "learning_rate": 1.8799004762724104e-07,
      "loss": 0.4715,
      "step": 429020
    },
    {
      "epoch": 175.62,
      "grad_norm": 1.8953036069869995,
      "learning_rate": 1.879278049318464e-07,
      "loss": 0.4011,
      "step": 429030
    },
    {
      "epoch": 175.62,
      "grad_norm": 1.899715781211853,
      "learning_rate": 1.878655721119394e-07,
      "loss": 0.4765,
      "step": 429040
    },
    {
      "epoch": 175.62,
      "grad_norm": 2.2423996925354004,
      "learning_rate": 1.87803349167805e-07,
      "loss": 0.4493,
      "step": 429050
    },
    {
      "epoch": 175.63,
      "grad_norm": 2.137127161026001,
      "learning_rate": 1.877411360997284e-07,
      "loss": 0.4249,
      "step": 429060
    },
    {
      "epoch": 175.63,
      "grad_norm": 1.8984698057174683,
      "learning_rate": 1.8767893290799448e-07,
      "loss": 0.4541,
      "step": 429070
    },
    {
      "epoch": 175.64,
      "grad_norm": 2.103912115097046,
      "learning_rate": 1.8761673959288821e-07,
      "loss": 0.4442,
      "step": 429080
    },
    {
      "epoch": 175.64,
      "grad_norm": 1.960844874382019,
      "learning_rate": 1.8755455615469452e-07,
      "loss": 0.4544,
      "step": 429090
    },
    {
      "epoch": 175.64,
      "grad_norm": 2.0322022438049316,
      "learning_rate": 1.8749238259369808e-07,
      "loss": 0.4319,
      "step": 429100
    },
    {
      "epoch": 175.65,
      "grad_norm": 1.7852445840835571,
      "learning_rate": 1.8743021891018378e-07,
      "loss": 0.4495,
      "step": 429110
    },
    {
      "epoch": 175.65,
      "grad_norm": 2.2898826599121094,
      "learning_rate": 1.8736806510443684e-07,
      "loss": 0.4481,
      "step": 429120
    },
    {
      "epoch": 175.66,
      "grad_norm": 2.220834732055664,
      "learning_rate": 1.8730592117674115e-07,
      "loss": 0.4345,
      "step": 429130
    },
    {
      "epoch": 175.66,
      "grad_norm": 1.8131152391433716,
      "learning_rate": 1.8724378712738214e-07,
      "loss": 0.451,
      "step": 429140
    },
    {
      "epoch": 175.67,
      "grad_norm": 1.9700217247009277,
      "learning_rate": 1.8718166295664397e-07,
      "loss": 0.4443,
      "step": 429150
    },
    {
      "epoch": 175.67,
      "grad_norm": 2.199049234390259,
      "learning_rate": 1.871195486648113e-07,
      "loss": 0.4503,
      "step": 429160
    },
    {
      "epoch": 175.67,
      "grad_norm": 2.2531096935272217,
      "learning_rate": 1.8705744425216883e-07,
      "loss": 0.4605,
      "step": 429170
    },
    {
      "epoch": 175.68,
      "grad_norm": 2.113192319869995,
      "learning_rate": 1.869953497190009e-07,
      "loss": 0.4647,
      "step": 429180
    },
    {
      "epoch": 175.68,
      "grad_norm": 2.5559825897216797,
      "learning_rate": 1.8693326506559201e-07,
      "loss": 0.4524,
      "step": 429190
    },
    {
      "epoch": 175.69,
      "grad_norm": 2.014470338821411,
      "learning_rate": 1.8687119029222674e-07,
      "loss": 0.4427,
      "step": 429200
    },
    {
      "epoch": 175.69,
      "grad_norm": 1.9107011556625366,
      "learning_rate": 1.8680912539918904e-07,
      "loss": 0.4409,
      "step": 429210
    },
    {
      "epoch": 175.69,
      "grad_norm": 2.2309248447418213,
      "learning_rate": 1.8674707038676354e-07,
      "loss": 0.4645,
      "step": 429220
    },
    {
      "epoch": 175.7,
      "grad_norm": 2.166898727416992,
      "learning_rate": 1.8668502525523442e-07,
      "loss": 0.4529,
      "step": 429230
    },
    {
      "epoch": 175.7,
      "grad_norm": 2.366434097290039,
      "learning_rate": 1.8662299000488577e-07,
      "loss": 0.4368,
      "step": 429240
    },
    {
      "epoch": 175.71,
      "grad_norm": 2.1878747940063477,
      "learning_rate": 1.8656096463600207e-07,
      "loss": 0.4253,
      "step": 429250
    },
    {
      "epoch": 175.71,
      "grad_norm": 1.9348063468933105,
      "learning_rate": 1.8649894914886692e-07,
      "loss": 0.4639,
      "step": 429260
    },
    {
      "epoch": 175.71,
      "grad_norm": 2.1500418186187744,
      "learning_rate": 1.8643694354376548e-07,
      "loss": 0.467,
      "step": 429270
    },
    {
      "epoch": 175.72,
      "grad_norm": 1.9094994068145752,
      "learning_rate": 1.8637494782098062e-07,
      "loss": 0.4562,
      "step": 429280
    },
    {
      "epoch": 175.72,
      "grad_norm": 2.0250041484832764,
      "learning_rate": 1.8631296198079623e-07,
      "loss": 0.4295,
      "step": 429290
    },
    {
      "epoch": 175.73,
      "grad_norm": 2.3233141899108887,
      "learning_rate": 1.8625098602349724e-07,
      "loss": 0.4418,
      "step": 429300
    },
    {
      "epoch": 175.73,
      "grad_norm": 1.9341260194778442,
      "learning_rate": 1.8618901994936704e-07,
      "loss": 0.4365,
      "step": 429310
    },
    {
      "epoch": 175.73,
      "grad_norm": 2.005037784576416,
      "learning_rate": 1.8612706375868971e-07,
      "loss": 0.4499,
      "step": 429320
    },
    {
      "epoch": 175.74,
      "grad_norm": 1.9045995473861694,
      "learning_rate": 1.8606511745174868e-07,
      "loss": 0.449,
      "step": 429330
    },
    {
      "epoch": 175.74,
      "grad_norm": 2.4550275802612305,
      "learning_rate": 1.8600318102882805e-07,
      "loss": 0.4387,
      "step": 429340
    },
    {
      "epoch": 175.75,
      "grad_norm": 2.134169578552246,
      "learning_rate": 1.859412544902115e-07,
      "loss": 0.4326,
      "step": 429350
    },
    {
      "epoch": 175.75,
      "grad_norm": 2.0678870677948,
      "learning_rate": 1.858793378361826e-07,
      "loss": 0.4413,
      "step": 429360
    },
    {
      "epoch": 175.76,
      "grad_norm": 2.6057496070861816,
      "learning_rate": 1.8581743106702503e-07,
      "loss": 0.4681,
      "step": 429370
    },
    {
      "epoch": 175.76,
      "grad_norm": 2.2963569164276123,
      "learning_rate": 1.8575553418302213e-07,
      "loss": 0.4618,
      "step": 429380
    },
    {
      "epoch": 175.76,
      "grad_norm": 2.149812698364258,
      "learning_rate": 1.8569364718445806e-07,
      "loss": 0.4296,
      "step": 429390
    },
    {
      "epoch": 175.77,
      "grad_norm": 2.152411460876465,
      "learning_rate": 1.8563177007161565e-07,
      "loss": 0.454,
      "step": 429400
    },
    {
      "epoch": 175.77,
      "grad_norm": 2.0783536434173584,
      "learning_rate": 1.8556990284477905e-07,
      "loss": 0.4529,
      "step": 429410
    },
    {
      "epoch": 175.78,
      "grad_norm": 3.3700039386749268,
      "learning_rate": 1.8550804550423085e-07,
      "loss": 0.4722,
      "step": 429420
    },
    {
      "epoch": 175.78,
      "grad_norm": 1.8101495504379272,
      "learning_rate": 1.8544619805025495e-07,
      "loss": 0.4447,
      "step": 429430
    },
    {
      "epoch": 175.78,
      "grad_norm": 2.4706971645355225,
      "learning_rate": 1.8538436048313446e-07,
      "loss": 0.4401,
      "step": 429440
    },
    {
      "epoch": 175.79,
      "grad_norm": 2.2241461277008057,
      "learning_rate": 1.8532253280315245e-07,
      "loss": 0.4493,
      "step": 429450
    },
    {
      "epoch": 175.79,
      "grad_norm": 2.489555597305298,
      "learning_rate": 1.8526071501059258e-07,
      "loss": 0.433,
      "step": 429460
    },
    {
      "epoch": 175.8,
      "grad_norm": 2.835930824279785,
      "learning_rate": 1.8519890710573795e-07,
      "loss": 0.4274,
      "step": 429470
    },
    {
      "epoch": 175.8,
      "grad_norm": 1.8144047260284424,
      "learning_rate": 1.8513710908887088e-07,
      "loss": 0.4336,
      "step": 429480
    },
    {
      "epoch": 175.8,
      "grad_norm": 2.153856039047241,
      "learning_rate": 1.8507532096027578e-07,
      "loss": 0.436,
      "step": 429490
    },
    {
      "epoch": 175.81,
      "grad_norm": 2.231032609939575,
      "learning_rate": 1.8501354272023524e-07,
      "loss": 0.4245,
      "step": 429500
    },
    {
      "epoch": 175.81,
      "grad_norm": 2.411482334136963,
      "learning_rate": 1.8495177436903185e-07,
      "loss": 0.4411,
      "step": 429510
    },
    {
      "epoch": 175.82,
      "grad_norm": 2.1738955974578857,
      "learning_rate": 1.848900159069487e-07,
      "loss": 0.4475,
      "step": 429520
    },
    {
      "epoch": 175.82,
      "grad_norm": 2.272170305252075,
      "learning_rate": 1.8482826733426893e-07,
      "loss": 0.4464,
      "step": 429530
    },
    {
      "epoch": 175.82,
      "grad_norm": 1.9858677387237549,
      "learning_rate": 1.8476652865127536e-07,
      "loss": 0.4313,
      "step": 429540
    },
    {
      "epoch": 175.83,
      "grad_norm": 2.3730316162109375,
      "learning_rate": 1.847047998582511e-07,
      "loss": 0.4367,
      "step": 429550
    },
    {
      "epoch": 175.83,
      "grad_norm": 2.0393269062042236,
      "learning_rate": 1.8464308095547793e-07,
      "loss": 0.4414,
      "step": 429560
    },
    {
      "epoch": 175.84,
      "grad_norm": 2.0713093280792236,
      "learning_rate": 1.845813719432393e-07,
      "loss": 0.446,
      "step": 429570
    },
    {
      "epoch": 175.84,
      "grad_norm": 2.3028347492218018,
      "learning_rate": 1.845196728218177e-07,
      "loss": 0.4129,
      "step": 429580
    },
    {
      "epoch": 175.85,
      "grad_norm": 2.0536439418792725,
      "learning_rate": 1.84457983591496e-07,
      "loss": 0.4424,
      "step": 429590
    },
    {
      "epoch": 175.85,
      "grad_norm": 2.096985340118408,
      "learning_rate": 1.8439630425255658e-07,
      "loss": 0.4481,
      "step": 429600
    },
    {
      "epoch": 175.85,
      "grad_norm": 2.027031183242798,
      "learning_rate": 1.84334634805282e-07,
      "loss": 0.447,
      "step": 429610
    },
    {
      "epoch": 175.86,
      "grad_norm": 1.8193851709365845,
      "learning_rate": 1.8427297524995457e-07,
      "loss": 0.4405,
      "step": 429620
    },
    {
      "epoch": 175.86,
      "grad_norm": 2.140026807785034,
      "learning_rate": 1.842113255868571e-07,
      "loss": 0.4537,
      "step": 429630
    },
    {
      "epoch": 175.87,
      "grad_norm": 2.1852803230285645,
      "learning_rate": 1.84149685816272e-07,
      "loss": 0.4648,
      "step": 429640
    },
    {
      "epoch": 175.87,
      "grad_norm": 1.9188810586929321,
      "learning_rate": 1.840880559384813e-07,
      "loss": 0.4408,
      "step": 429650
    },
    {
      "epoch": 175.87,
      "grad_norm": 2.0980985164642334,
      "learning_rate": 1.8402643595376707e-07,
      "loss": 0.4332,
      "step": 429660
    },
    {
      "epoch": 175.88,
      "grad_norm": 2.0704760551452637,
      "learning_rate": 1.8396482586241265e-07,
      "loss": 0.4636,
      "step": 429670
    },
    {
      "epoch": 175.88,
      "grad_norm": 2.3905422687530518,
      "learning_rate": 1.8390322566469987e-07,
      "loss": 0.4197,
      "step": 429680
    },
    {
      "epoch": 175.89,
      "grad_norm": 2.0369229316711426,
      "learning_rate": 1.8384163536091027e-07,
      "loss": 0.4325,
      "step": 429690
    },
    {
      "epoch": 175.89,
      "grad_norm": 2.103254556655884,
      "learning_rate": 1.8378005495132643e-07,
      "loss": 0.4485,
      "step": 429700
    },
    {
      "epoch": 175.89,
      "grad_norm": 1.6968952417373657,
      "learning_rate": 1.8371848443623016e-07,
      "loss": 0.4387,
      "step": 429710
    },
    {
      "epoch": 175.9,
      "grad_norm": 2.0504870414733887,
      "learning_rate": 1.8365692381590406e-07,
      "loss": 0.4557,
      "step": 429720
    },
    {
      "epoch": 175.9,
      "grad_norm": 2.4289445877075195,
      "learning_rate": 1.8359537309062967e-07,
      "loss": 0.4564,
      "step": 429730
    },
    {
      "epoch": 175.91,
      "grad_norm": 2.453108072280884,
      "learning_rate": 1.8353383226068903e-07,
      "loss": 0.4441,
      "step": 429740
    },
    {
      "epoch": 175.91,
      "grad_norm": 2.3164446353912354,
      "learning_rate": 1.8347230132636425e-07,
      "loss": 0.4443,
      "step": 429750
    },
    {
      "epoch": 175.91,
      "grad_norm": 2.0497660636901855,
      "learning_rate": 1.8341078028793686e-07,
      "loss": 0.4604,
      "step": 429760
    },
    {
      "epoch": 175.92,
      "grad_norm": 1.782863974571228,
      "learning_rate": 1.8334926914568914e-07,
      "loss": 0.4264,
      "step": 429770
    },
    {
      "epoch": 175.92,
      "grad_norm": 1.8509156703948975,
      "learning_rate": 1.832877678999022e-07,
      "loss": 0.4447,
      "step": 429780
    },
    {
      "epoch": 175.93,
      "grad_norm": 2.2645785808563232,
      "learning_rate": 1.8322627655085856e-07,
      "loss": 0.4326,
      "step": 429790
    },
    {
      "epoch": 175.93,
      "grad_norm": 1.9451824426651,
      "learning_rate": 1.8316479509883925e-07,
      "loss": 0.4354,
      "step": 429800
    },
    {
      "epoch": 175.94,
      "grad_norm": 1.9567164182662964,
      "learning_rate": 1.831033235441261e-07,
      "loss": 0.4496,
      "step": 429810
    },
    {
      "epoch": 175.94,
      "grad_norm": 2.167325258255005,
      "learning_rate": 1.8304186188700092e-07,
      "loss": 0.4484,
      "step": 429820
    },
    {
      "epoch": 175.94,
      "grad_norm": 2.19388747215271,
      "learning_rate": 1.8298041012774498e-07,
      "loss": 0.4367,
      "step": 429830
    },
    {
      "epoch": 175.95,
      "grad_norm": 2.108189105987549,
      "learning_rate": 1.8291896826664008e-07,
      "loss": 0.4627,
      "step": 429840
    },
    {
      "epoch": 175.95,
      "grad_norm": 2.33540415763855,
      "learning_rate": 1.8285753630396727e-07,
      "loss": 0.4694,
      "step": 429850
    },
    {
      "epoch": 175.96,
      "grad_norm": 1.8140989542007446,
      "learning_rate": 1.8279611424000833e-07,
      "loss": 0.4397,
      "step": 429860
    },
    {
      "epoch": 175.96,
      "grad_norm": 2.3506104946136475,
      "learning_rate": 1.8273470207504456e-07,
      "loss": 0.4217,
      "step": 429870
    },
    {
      "epoch": 175.96,
      "grad_norm": 1.8233518600463867,
      "learning_rate": 1.8267329980935696e-07,
      "loss": 0.4539,
      "step": 429880
    },
    {
      "epoch": 175.97,
      "grad_norm": 2.3276615142822266,
      "learning_rate": 1.8261190744322712e-07,
      "loss": 0.4591,
      "step": 429890
    },
    {
      "epoch": 175.97,
      "grad_norm": 2.0347063541412354,
      "learning_rate": 1.8255052497693627e-07,
      "loss": 0.4597,
      "step": 429900
    },
    {
      "epoch": 175.98,
      "grad_norm": 1.9392515420913696,
      "learning_rate": 1.8248915241076551e-07,
      "loss": 0.4523,
      "step": 429910
    },
    {
      "epoch": 175.98,
      "grad_norm": 2.049461841583252,
      "learning_rate": 1.8242778974499578e-07,
      "loss": 0.4551,
      "step": 429920
    },
    {
      "epoch": 175.98,
      "grad_norm": 2.105696201324463,
      "learning_rate": 1.8236643697990844e-07,
      "loss": 0.434,
      "step": 429930
    },
    {
      "epoch": 175.99,
      "grad_norm": 2.0507099628448486,
      "learning_rate": 1.8230509411578473e-07,
      "loss": 0.4534,
      "step": 429940
    },
    {
      "epoch": 175.99,
      "grad_norm": 5.440514087677002,
      "learning_rate": 1.8224376115290518e-07,
      "loss": 0.4313,
      "step": 429950
    },
    {
      "epoch": 176.0,
      "grad_norm": 2.0132970809936523,
      "learning_rate": 1.8218243809155077e-07,
      "loss": 0.4321,
      "step": 429960
    },
    {
      "epoch": 176.0,
      "eval_loss": 0.4451003074645996,
      "eval_runtime": 58.0349,
      "eval_samples_per_second": 59.43,
      "eval_steps_per_second": 7.444,
      "step": 429968
    },
    {
      "epoch": 176.0,
      "grad_norm": 1.6876112222671509,
      "learning_rate": 1.8212112493200287e-07,
      "loss": 0.4179,
      "step": 429970
    },
    {
      "epoch": 176.0,
      "grad_norm": 2.0794453620910645,
      "learning_rate": 1.8205982167454214e-07,
      "loss": 0.4421,
      "step": 429980
    },
    {
      "epoch": 176.01,
      "grad_norm": 1.9082974195480347,
      "learning_rate": 1.8199852831944968e-07,
      "loss": 0.441,
      "step": 429990
    },
    {
      "epoch": 176.01,
      "grad_norm": 2.3191912174224854,
      "learning_rate": 1.8193724486700546e-07,
      "loss": 0.446,
      "step": 430000
    },
    {
      "epoch": 176.02,
      "grad_norm": 1.6941804885864258,
      "learning_rate": 1.8187597131749048e-07,
      "loss": 0.4463,
      "step": 430010
    },
    {
      "epoch": 176.02,
      "grad_norm": 1.9328852891921997,
      "learning_rate": 1.818147076711858e-07,
      "loss": 0.4106,
      "step": 430020
    },
    {
      "epoch": 176.03,
      "grad_norm": 2.1612789630889893,
      "learning_rate": 1.817534539283719e-07,
      "loss": 0.4308,
      "step": 430030
    },
    {
      "epoch": 176.03,
      "grad_norm": 1.9842190742492676,
      "learning_rate": 1.8169221008932905e-07,
      "loss": 0.4301,
      "step": 430040
    },
    {
      "epoch": 176.03,
      "grad_norm": 2.0527796745300293,
      "learning_rate": 1.8163097615433823e-07,
      "loss": 0.4256,
      "step": 430050
    },
    {
      "epoch": 176.04,
      "grad_norm": 2.5179004669189453,
      "learning_rate": 1.8156975212368e-07,
      "loss": 0.4411,
      "step": 430060
    },
    {
      "epoch": 176.04,
      "grad_norm": 1.9820665121078491,
      "learning_rate": 1.8150853799763456e-07,
      "loss": 0.4392,
      "step": 430070
    },
    {
      "epoch": 176.05,
      "grad_norm": 1.8873181343078613,
      "learning_rate": 1.8144733377648242e-07,
      "loss": 0.4597,
      "step": 430080
    },
    {
      "epoch": 176.05,
      "grad_norm": 2.280703544616699,
      "learning_rate": 1.813861394605041e-07,
      "loss": 0.4316,
      "step": 430090
    },
    {
      "epoch": 176.05,
      "grad_norm": 1.9915982484817505,
      "learning_rate": 1.8132495504997935e-07,
      "loss": 0.456,
      "step": 430100
    },
    {
      "epoch": 176.06,
      "grad_norm": 2.08697247505188,
      "learning_rate": 1.8126378054518915e-07,
      "loss": 0.4317,
      "step": 430110
    },
    {
      "epoch": 176.06,
      "grad_norm": 2.046967029571533,
      "learning_rate": 1.8120261594641322e-07,
      "loss": 0.4331,
      "step": 430120
    },
    {
      "epoch": 176.07,
      "grad_norm": 1.8404383659362793,
      "learning_rate": 1.8114146125393264e-07,
      "loss": 0.4592,
      "step": 430130
    },
    {
      "epoch": 176.07,
      "grad_norm": 2.147867202758789,
      "learning_rate": 1.810803164680263e-07,
      "loss": 0.4399,
      "step": 430140
    },
    {
      "epoch": 176.07,
      "grad_norm": 2.287917137145996,
      "learning_rate": 1.810191815889747e-07,
      "loss": 0.4783,
      "step": 430150
    },
    {
      "epoch": 176.08,
      "grad_norm": 2.0791265964508057,
      "learning_rate": 1.8095805661705814e-07,
      "loss": 0.4474,
      "step": 430160
    },
    {
      "epoch": 176.08,
      "grad_norm": 2.0520570278167725,
      "learning_rate": 1.8089694155255658e-07,
      "loss": 0.4303,
      "step": 430170
    },
    {
      "epoch": 176.09,
      "grad_norm": 1.725914478302002,
      "learning_rate": 1.808358363957497e-07,
      "loss": 0.4413,
      "step": 430180
    },
    {
      "epoch": 176.09,
      "grad_norm": 1.9383541345596313,
      "learning_rate": 1.8077474114691783e-07,
      "loss": 0.4351,
      "step": 430190
    },
    {
      "epoch": 176.09,
      "grad_norm": 1.9371531009674072,
      "learning_rate": 1.8071365580634062e-07,
      "loss": 0.4388,
      "step": 430200
    },
    {
      "epoch": 176.1,
      "grad_norm": 1.7853822708129883,
      "learning_rate": 1.8065258037429807e-07,
      "loss": 0.4394,
      "step": 430210
    },
    {
      "epoch": 176.1,
      "grad_norm": 2.1782004833221436,
      "learning_rate": 1.8059151485106966e-07,
      "loss": 0.4484,
      "step": 430220
    },
    {
      "epoch": 176.11,
      "grad_norm": 2.2820329666137695,
      "learning_rate": 1.8053045923693458e-07,
      "loss": 0.4363,
      "step": 430230
    },
    {
      "epoch": 176.11,
      "grad_norm": 1.7478357553482056,
      "learning_rate": 1.8046941353217388e-07,
      "loss": 0.429,
      "step": 430240
    },
    {
      "epoch": 176.12,
      "grad_norm": 1.951548457145691,
      "learning_rate": 1.8040837773706674e-07,
      "loss": 0.4299,
      "step": 430250
    },
    {
      "epoch": 176.12,
      "grad_norm": 1.9926457405090332,
      "learning_rate": 1.803473518518921e-07,
      "loss": 0.4256,
      "step": 430260
    },
    {
      "epoch": 176.12,
      "grad_norm": 2.096127510070801,
      "learning_rate": 1.8028633587693049e-07,
      "loss": 0.4464,
      "step": 430270
    },
    {
      "epoch": 176.13,
      "grad_norm": 1.747834324836731,
      "learning_rate": 1.8022532981246057e-07,
      "loss": 0.4289,
      "step": 430280
    },
    {
      "epoch": 176.13,
      "grad_norm": 2.241957902908325,
      "learning_rate": 1.8016433365876208e-07,
      "loss": 0.4555,
      "step": 430290
    },
    {
      "epoch": 176.14,
      "grad_norm": 2.046825885772705,
      "learning_rate": 1.8010334741611446e-07,
      "loss": 0.4528,
      "step": 430300
    },
    {
      "epoch": 176.14,
      "grad_norm": 1.9425729513168335,
      "learning_rate": 1.8004237108479696e-07,
      "loss": 0.4398,
      "step": 430310
    },
    {
      "epoch": 176.14,
      "grad_norm": 3.046144962310791,
      "learning_rate": 1.7998140466508925e-07,
      "loss": 0.4632,
      "step": 430320
    },
    {
      "epoch": 176.15,
      "grad_norm": 2.153212547302246,
      "learning_rate": 1.7992044815727003e-07,
      "loss": 0.4582,
      "step": 430330
    },
    {
      "epoch": 176.15,
      "grad_norm": 2.1146745681762695,
      "learning_rate": 1.798595015616193e-07,
      "loss": 0.4271,
      "step": 430340
    },
    {
      "epoch": 176.16,
      "grad_norm": 2.2543046474456787,
      "learning_rate": 1.7979856487841548e-07,
      "loss": 0.4363,
      "step": 430350
    },
    {
      "epoch": 176.16,
      "grad_norm": 1.8500131368637085,
      "learning_rate": 1.797376381079383e-07,
      "loss": 0.4525,
      "step": 430360
    },
    {
      "epoch": 176.16,
      "grad_norm": 1.7819284200668335,
      "learning_rate": 1.7967672125046642e-07,
      "loss": 0.4546,
      "step": 430370
    },
    {
      "epoch": 176.17,
      "grad_norm": 1.848824143409729,
      "learning_rate": 1.7961581430627907e-07,
      "loss": 0.4465,
      "step": 430380
    },
    {
      "epoch": 176.17,
      "grad_norm": 2.353036403656006,
      "learning_rate": 1.7955491727565542e-07,
      "loss": 0.4589,
      "step": 430390
    },
    {
      "epoch": 176.18,
      "grad_norm": 2.047365427017212,
      "learning_rate": 1.7949403015887417e-07,
      "loss": 0.4253,
      "step": 430400
    },
    {
      "epoch": 176.18,
      "grad_norm": 1.9450324773788452,
      "learning_rate": 1.794331529562143e-07,
      "loss": 0.4334,
      "step": 430410
    },
    {
      "epoch": 176.19,
      "grad_norm": 1.8027042150497437,
      "learning_rate": 1.7937228566795467e-07,
      "loss": 0.4423,
      "step": 430420
    },
    {
      "epoch": 176.19,
      "grad_norm": 2.1293411254882812,
      "learning_rate": 1.793114282943743e-07,
      "loss": 0.4461,
      "step": 430430
    },
    {
      "epoch": 176.19,
      "grad_norm": 2.3171122074127197,
      "learning_rate": 1.7925058083575185e-07,
      "loss": 0.4448,
      "step": 430440
    },
    {
      "epoch": 176.2,
      "grad_norm": 2.0551648139953613,
      "learning_rate": 1.79189743292366e-07,
      "loss": 0.4507,
      "step": 430450
    },
    {
      "epoch": 176.2,
      "grad_norm": 2.4071128368377686,
      "learning_rate": 1.7912891566449544e-07,
      "loss": 0.4546,
      "step": 430460
    },
    {
      "epoch": 176.21,
      "grad_norm": 2.0434963703155518,
      "learning_rate": 1.7906809795241883e-07,
      "loss": 0.4428,
      "step": 430470
    },
    {
      "epoch": 176.21,
      "grad_norm": 2.4172110557556152,
      "learning_rate": 1.7900729015641464e-07,
      "loss": 0.457,
      "step": 430480
    },
    {
      "epoch": 176.21,
      "grad_norm": 2.3305106163024902,
      "learning_rate": 1.7894649227676175e-07,
      "loss": 0.4436,
      "step": 430490
    },
    {
      "epoch": 176.22,
      "grad_norm": 2.0606532096862793,
      "learning_rate": 1.7888570431373837e-07,
      "loss": 0.4399,
      "step": 430500
    },
    {
      "epoch": 176.22,
      "grad_norm": 2.2083983421325684,
      "learning_rate": 1.788249262676229e-07,
      "loss": 0.4342,
      "step": 430510
    },
    {
      "epoch": 176.23,
      "grad_norm": 2.093982696533203,
      "learning_rate": 1.7876415813869406e-07,
      "loss": 0.4292,
      "step": 430520
    },
    {
      "epoch": 176.23,
      "grad_norm": 1.9104465246200562,
      "learning_rate": 1.7870339992723024e-07,
      "loss": 0.4628,
      "step": 430530
    },
    {
      "epoch": 176.23,
      "grad_norm": 2.4752578735351562,
      "learning_rate": 1.7864265163350933e-07,
      "loss": 0.4604,
      "step": 430540
    },
    {
      "epoch": 176.24,
      "grad_norm": 2.6084747314453125,
      "learning_rate": 1.7858191325781005e-07,
      "loss": 0.4555,
      "step": 430550
    },
    {
      "epoch": 176.24,
      "grad_norm": 2.0352516174316406,
      "learning_rate": 1.7852118480041053e-07,
      "loss": 0.4302,
      "step": 430560
    },
    {
      "epoch": 176.25,
      "grad_norm": 2.045219659805298,
      "learning_rate": 1.7846046626158868e-07,
      "loss": 0.4346,
      "step": 430570
    },
    {
      "epoch": 176.25,
      "grad_norm": 1.8482815027236938,
      "learning_rate": 1.7839975764162322e-07,
      "loss": 0.4424,
      "step": 430580
    },
    {
      "epoch": 176.25,
      "grad_norm": 2.1556336879730225,
      "learning_rate": 1.7833905894079173e-07,
      "loss": 0.4485,
      "step": 430590
    },
    {
      "epoch": 176.26,
      "grad_norm": 1.9937164783477783,
      "learning_rate": 1.7827837015937218e-07,
      "loss": 0.4497,
      "step": 430600
    },
    {
      "epoch": 176.26,
      "grad_norm": 2.6312267780303955,
      "learning_rate": 1.7821769129764243e-07,
      "loss": 0.4473,
      "step": 430610
    },
    {
      "epoch": 176.27,
      "grad_norm": 1.8676300048828125,
      "learning_rate": 1.7815702235588116e-07,
      "loss": 0.4528,
      "step": 430620
    },
    {
      "epoch": 176.27,
      "grad_norm": 2.230689287185669,
      "learning_rate": 1.7809636333436605e-07,
      "loss": 0.4422,
      "step": 430630
    },
    {
      "epoch": 176.28,
      "grad_norm": 2.034806251525879,
      "learning_rate": 1.7803571423337496e-07,
      "loss": 0.467,
      "step": 430640
    },
    {
      "epoch": 176.28,
      "grad_norm": 2.287369966506958,
      "learning_rate": 1.7797507505318553e-07,
      "loss": 0.45,
      "step": 430650
    },
    {
      "epoch": 176.28,
      "grad_norm": 1.7667789459228516,
      "learning_rate": 1.7791444579407546e-07,
      "loss": 0.4498,
      "step": 430660
    },
    {
      "epoch": 176.29,
      "grad_norm": 2.0134189128875732,
      "learning_rate": 1.7785382645632286e-07,
      "loss": 0.4474,
      "step": 430670
    },
    {
      "epoch": 176.29,
      "grad_norm": 1.982904076576233,
      "learning_rate": 1.7779321704020484e-07,
      "loss": 0.4426,
      "step": 430680
    },
    {
      "epoch": 176.3,
      "grad_norm": 8.606761932373047,
      "learning_rate": 1.7773261754599988e-07,
      "loss": 0.4528,
      "step": 430690
    },
    {
      "epoch": 176.3,
      "grad_norm": 1.9080835580825806,
      "learning_rate": 1.776720279739848e-07,
      "loss": 0.4467,
      "step": 430700
    },
    {
      "epoch": 176.3,
      "grad_norm": 2.208291530609131,
      "learning_rate": 1.776114483244378e-07,
      "loss": 0.4575,
      "step": 430710
    },
    {
      "epoch": 176.31,
      "grad_norm": 2.489428758621216,
      "learning_rate": 1.7755087859763596e-07,
      "loss": 0.4554,
      "step": 430720
    },
    {
      "epoch": 176.31,
      "grad_norm": 1.936058521270752,
      "learning_rate": 1.7749031879385643e-07,
      "loss": 0.4545,
      "step": 430730
    },
    {
      "epoch": 176.32,
      "grad_norm": 2.1113739013671875,
      "learning_rate": 1.7742976891337736e-07,
      "loss": 0.4418,
      "step": 430740
    },
    {
      "epoch": 176.32,
      "grad_norm": 2.1597418785095215,
      "learning_rate": 1.773692289564756e-07,
      "loss": 0.4511,
      "step": 430750
    },
    {
      "epoch": 176.32,
      "grad_norm": 1.8820902109146118,
      "learning_rate": 1.7730869892342855e-07,
      "loss": 0.4279,
      "step": 430760
    },
    {
      "epoch": 176.33,
      "grad_norm": 1.873879313468933,
      "learning_rate": 1.7724817881451362e-07,
      "loss": 0.4692,
      "step": 430770
    },
    {
      "epoch": 176.33,
      "grad_norm": 4.44749641418457,
      "learning_rate": 1.7718766863000812e-07,
      "loss": 0.4361,
      "step": 430780
    },
    {
      "epoch": 176.34,
      "grad_norm": 2.0932459831237793,
      "learning_rate": 1.7712716837018923e-07,
      "loss": 0.4373,
      "step": 430790
    },
    {
      "epoch": 176.34,
      "grad_norm": 2.058465003967285,
      "learning_rate": 1.7706667803533327e-07,
      "loss": 0.4637,
      "step": 430800
    },
    {
      "epoch": 176.34,
      "grad_norm": 1.7442352771759033,
      "learning_rate": 1.7700619762571869e-07,
      "loss": 0.4366,
      "step": 430810
    },
    {
      "epoch": 176.35,
      "grad_norm": 1.8566762208938599,
      "learning_rate": 1.7694572714162178e-07,
      "loss": 0.4307,
      "step": 430820
    },
    {
      "epoch": 176.35,
      "grad_norm": 2.1992475986480713,
      "learning_rate": 1.768852665833197e-07,
      "loss": 0.4692,
      "step": 430830
    },
    {
      "epoch": 176.36,
      "grad_norm": 1.9899709224700928,
      "learning_rate": 1.7682481595108958e-07,
      "loss": 0.4477,
      "step": 430840
    },
    {
      "epoch": 176.36,
      "grad_norm": 1.7259494066238403,
      "learning_rate": 1.7676437524520825e-07,
      "loss": 0.4482,
      "step": 430850
    },
    {
      "epoch": 176.37,
      "grad_norm": 1.6005496978759766,
      "learning_rate": 1.7670394446595232e-07,
      "loss": 0.4626,
      "step": 430860
    },
    {
      "epoch": 176.37,
      "grad_norm": 2.2512550354003906,
      "learning_rate": 1.7664352361359868e-07,
      "loss": 0.4398,
      "step": 430870
    },
    {
      "epoch": 176.37,
      "grad_norm": 1.7192400693893433,
      "learning_rate": 1.7658311268842414e-07,
      "loss": 0.4321,
      "step": 430880
    },
    {
      "epoch": 176.38,
      "grad_norm": 2.28261399269104,
      "learning_rate": 1.7652271169070535e-07,
      "loss": 0.4487,
      "step": 430890
    },
    {
      "epoch": 176.38,
      "grad_norm": 2.7292349338531494,
      "learning_rate": 1.7646232062071917e-07,
      "loss": 0.4456,
      "step": 430900
    },
    {
      "epoch": 176.39,
      "grad_norm": 2.5191876888275146,
      "learning_rate": 1.7640193947874242e-07,
      "loss": 0.4414,
      "step": 430910
    },
    {
      "epoch": 176.39,
      "grad_norm": 2.015202045440674,
      "learning_rate": 1.7634156826505122e-07,
      "loss": 0.4381,
      "step": 430920
    },
    {
      "epoch": 176.39,
      "grad_norm": 2.0841636657714844,
      "learning_rate": 1.762812069799227e-07,
      "loss": 0.4377,
      "step": 430930
    },
    {
      "epoch": 176.4,
      "grad_norm": 2.1081888675689697,
      "learning_rate": 1.7622085562363288e-07,
      "loss": 0.4527,
      "step": 430940
    },
    {
      "epoch": 176.4,
      "grad_norm": 2.555363655090332,
      "learning_rate": 1.7616051419645844e-07,
      "loss": 0.4275,
      "step": 430950
    },
    {
      "epoch": 176.41,
      "grad_norm": 1.9382086992263794,
      "learning_rate": 1.7610018269867565e-07,
      "loss": 0.4385,
      "step": 430960
    },
    {
      "epoch": 176.41,
      "grad_norm": 1.7637090682983398,
      "learning_rate": 1.7603986113056092e-07,
      "loss": 0.4548,
      "step": 430970
    },
    {
      "epoch": 176.41,
      "grad_norm": 1.9336804151535034,
      "learning_rate": 1.7597954949239052e-07,
      "loss": 0.4441,
      "step": 430980
    },
    {
      "epoch": 176.42,
      "grad_norm": 2.0627152919769287,
      "learning_rate": 1.7591924778444113e-07,
      "loss": 0.4425,
      "step": 430990
    },
    {
      "epoch": 176.42,
      "grad_norm": 1.9422606229782104,
      "learning_rate": 1.7585895600698848e-07,
      "loss": 0.4759,
      "step": 431000
    },
    {
      "epoch": 176.43,
      "grad_norm": 2.3877739906311035,
      "learning_rate": 1.7579867416030925e-07,
      "loss": 0.4606,
      "step": 431010
    },
    {
      "epoch": 176.43,
      "grad_norm": 1.6867796182632446,
      "learning_rate": 1.7573840224467898e-07,
      "loss": 0.4339,
      "step": 431020
    },
    {
      "epoch": 176.43,
      "grad_norm": 2.0647730827331543,
      "learning_rate": 1.7567814026037425e-07,
      "loss": 0.429,
      "step": 431030
    },
    {
      "epoch": 176.44,
      "grad_norm": 2.2855358123779297,
      "learning_rate": 1.756178882076709e-07,
      "loss": 0.4524,
      "step": 431040
    },
    {
      "epoch": 176.44,
      "grad_norm": 1.6928452253341675,
      "learning_rate": 1.7555764608684476e-07,
      "loss": 0.4512,
      "step": 431050
    },
    {
      "epoch": 176.45,
      "grad_norm": 1.927873969078064,
      "learning_rate": 1.754974138981724e-07,
      "loss": 0.4435,
      "step": 431060
    },
    {
      "epoch": 176.45,
      "grad_norm": 2.0590713024139404,
      "learning_rate": 1.7543719164192912e-07,
      "loss": 0.4473,
      "step": 431070
    },
    {
      "epoch": 176.46,
      "grad_norm": 1.8235913515090942,
      "learning_rate": 1.7537697931839103e-07,
      "loss": 0.4348,
      "step": 431080
    },
    {
      "epoch": 176.46,
      "grad_norm": 1.9464744329452515,
      "learning_rate": 1.7531677692783395e-07,
      "loss": 0.4529,
      "step": 431090
    },
    {
      "epoch": 176.46,
      "grad_norm": 2.182864189147949,
      "learning_rate": 1.7525658447053344e-07,
      "loss": 0.4427,
      "step": 431100
    },
    {
      "epoch": 176.47,
      "grad_norm": 1.8973419666290283,
      "learning_rate": 1.751964019467658e-07,
      "loss": 0.4399,
      "step": 431110
    },
    {
      "epoch": 176.47,
      "grad_norm": 1.9550833702087402,
      "learning_rate": 1.7513622935680614e-07,
      "loss": 0.4373,
      "step": 431120
    },
    {
      "epoch": 176.48,
      "grad_norm": 2.2280948162078857,
      "learning_rate": 1.7507606670093022e-07,
      "loss": 0.4395,
      "step": 431130
    },
    {
      "epoch": 176.48,
      "grad_norm": 2.0408263206481934,
      "learning_rate": 1.750159139794139e-07,
      "loss": 0.4428,
      "step": 431140
    },
    {
      "epoch": 176.48,
      "grad_norm": 2.059499979019165,
      "learning_rate": 1.7495577119253245e-07,
      "loss": 0.4584,
      "step": 431150
    },
    {
      "epoch": 176.49,
      "grad_norm": 2.1334228515625,
      "learning_rate": 1.74895638340562e-07,
      "loss": 0.4533,
      "step": 431160
    },
    {
      "epoch": 176.49,
      "grad_norm": 1.8313122987747192,
      "learning_rate": 1.7483551542377673e-07,
      "loss": 0.4385,
      "step": 431170
    },
    {
      "epoch": 176.5,
      "grad_norm": 2.2200846672058105,
      "learning_rate": 1.7477540244245306e-07,
      "loss": 0.4456,
      "step": 431180
    },
    {
      "epoch": 176.5,
      "grad_norm": 1.7377113103866577,
      "learning_rate": 1.7471529939686626e-07,
      "loss": 0.458,
      "step": 431190
    },
    {
      "epoch": 176.5,
      "grad_norm": 2.4912502765655518,
      "learning_rate": 1.7465520628729135e-07,
      "loss": 0.4402,
      "step": 431200
    },
    {
      "epoch": 176.51,
      "grad_norm": 2.0192949771881104,
      "learning_rate": 1.7459512311400365e-07,
      "loss": 0.4616,
      "step": 431210
    },
    {
      "epoch": 176.51,
      "grad_norm": 1.9152048826217651,
      "learning_rate": 1.7453504987727873e-07,
      "loss": 0.4409,
      "step": 431220
    },
    {
      "epoch": 176.52,
      "grad_norm": 2.106250762939453,
      "learning_rate": 1.7447498657739164e-07,
      "loss": 0.4218,
      "step": 431230
    },
    {
      "epoch": 176.52,
      "grad_norm": 1.8155364990234375,
      "learning_rate": 1.7441493321461712e-07,
      "loss": 0.4523,
      "step": 431240
    },
    {
      "epoch": 176.52,
      "grad_norm": 2.3055338859558105,
      "learning_rate": 1.7435488978923073e-07,
      "loss": 0.4624,
      "step": 431250
    },
    {
      "epoch": 176.53,
      "grad_norm": 1.89580500125885,
      "learning_rate": 1.7429485630150757e-07,
      "loss": 0.431,
      "step": 431260
    },
    {
      "epoch": 176.53,
      "grad_norm": 2.5320217609405518,
      "learning_rate": 1.742348327517221e-07,
      "loss": 0.4641,
      "step": 431270
    },
    {
      "epoch": 176.54,
      "grad_norm": 2.270972967147827,
      "learning_rate": 1.741748191401499e-07,
      "loss": 0.4413,
      "step": 431280
    },
    {
      "epoch": 176.54,
      "grad_norm": 1.766717791557312,
      "learning_rate": 1.7411481546706578e-07,
      "loss": 0.4632,
      "step": 431290
    },
    {
      "epoch": 176.55,
      "grad_norm": 2.0301666259765625,
      "learning_rate": 1.740548217327442e-07,
      "loss": 0.4319,
      "step": 431300
    },
    {
      "epoch": 176.55,
      "grad_norm": 2.167294979095459,
      "learning_rate": 1.7399483793746026e-07,
      "loss": 0.4235,
      "step": 431310
    },
    {
      "epoch": 176.55,
      "grad_norm": 2.0859603881835938,
      "learning_rate": 1.7393486408148845e-07,
      "loss": 0.4396,
      "step": 431320
    },
    {
      "epoch": 176.56,
      "grad_norm": 1.795231580734253,
      "learning_rate": 1.7387490016510382e-07,
      "loss": 0.4294,
      "step": 431330
    },
    {
      "epoch": 176.56,
      "grad_norm": 2.00923490524292,
      "learning_rate": 1.7381494618858113e-07,
      "loss": 0.464,
      "step": 431340
    },
    {
      "epoch": 176.57,
      "grad_norm": 1.999538779258728,
      "learning_rate": 1.7375500215219463e-07,
      "loss": 0.4098,
      "step": 431350
    },
    {
      "epoch": 176.57,
      "grad_norm": 1.9422849416732788,
      "learning_rate": 1.736950680562189e-07,
      "loss": 0.4419,
      "step": 431360
    },
    {
      "epoch": 176.57,
      "grad_norm": 1.9739142656326294,
      "learning_rate": 1.736351439009292e-07,
      "loss": 0.4479,
      "step": 431370
    },
    {
      "epoch": 176.58,
      "grad_norm": 1.8928335905075073,
      "learning_rate": 1.7357522968659953e-07,
      "loss": 0.4665,
      "step": 431380
    },
    {
      "epoch": 176.58,
      "grad_norm": 2.1608378887176514,
      "learning_rate": 1.7351532541350416e-07,
      "loss": 0.4503,
      "step": 431390
    },
    {
      "epoch": 176.59,
      "grad_norm": 1.7751226425170898,
      "learning_rate": 1.734554310819181e-07,
      "loss": 0.464,
      "step": 431400
    },
    {
      "epoch": 176.59,
      "grad_norm": 1.537857174873352,
      "learning_rate": 1.7339554669211514e-07,
      "loss": 0.436,
      "step": 431410
    },
    {
      "epoch": 176.59,
      "grad_norm": 1.7873812913894653,
      "learning_rate": 1.7333567224436979e-07,
      "loss": 0.4282,
      "step": 431420
    },
    {
      "epoch": 176.6,
      "grad_norm": 1.7807132005691528,
      "learning_rate": 1.7327580773895677e-07,
      "loss": 0.4529,
      "step": 431430
    },
    {
      "epoch": 176.6,
      "grad_norm": 2.1166582107543945,
      "learning_rate": 1.7321595317614936e-07,
      "loss": 0.4321,
      "step": 431440
    },
    {
      "epoch": 176.61,
      "grad_norm": 1.8554143905639648,
      "learning_rate": 1.731561085562226e-07,
      "loss": 0.4501,
      "step": 431450
    },
    {
      "epoch": 176.61,
      "grad_norm": 2.3811123371124268,
      "learning_rate": 1.7309627387944995e-07,
      "loss": 0.4311,
      "step": 431460
    },
    {
      "epoch": 176.61,
      "grad_norm": 1.936250925064087,
      "learning_rate": 1.7303644914610617e-07,
      "loss": 0.4475,
      "step": 431470
    },
    {
      "epoch": 176.62,
      "grad_norm": 2.519674301147461,
      "learning_rate": 1.7297663435646476e-07,
      "loss": 0.4407,
      "step": 431480
    },
    {
      "epoch": 176.62,
      "grad_norm": 1.909029483795166,
      "learning_rate": 1.7291682951080002e-07,
      "loss": 0.4373,
      "step": 431490
    },
    {
      "epoch": 176.63,
      "grad_norm": 1.7869536876678467,
      "learning_rate": 1.7285703460938588e-07,
      "loss": 0.4316,
      "step": 431500
    },
    {
      "epoch": 176.63,
      "grad_norm": 2.3486204147338867,
      "learning_rate": 1.7279724965249612e-07,
      "loss": 0.436,
      "step": 431510
    },
    {
      "epoch": 176.64,
      "grad_norm": 2.0472097396850586,
      "learning_rate": 1.7273747464040473e-07,
      "loss": 0.4442,
      "step": 431520
    },
    {
      "epoch": 176.64,
      "grad_norm": 2.373978853225708,
      "learning_rate": 1.7267770957338544e-07,
      "loss": 0.4495,
      "step": 431530
    },
    {
      "epoch": 176.64,
      "grad_norm": 2.077688455581665,
      "learning_rate": 1.7261795445171225e-07,
      "loss": 0.4719,
      "step": 431540
    },
    {
      "epoch": 176.65,
      "grad_norm": 2.1338868141174316,
      "learning_rate": 1.7255820927565816e-07,
      "loss": 0.4492,
      "step": 431550
    },
    {
      "epoch": 176.65,
      "grad_norm": 2.367741584777832,
      "learning_rate": 1.724984740454979e-07,
      "loss": 0.4322,
      "step": 431560
    },
    {
      "epoch": 176.66,
      "grad_norm": 1.9038310050964355,
      "learning_rate": 1.7243874876150497e-07,
      "loss": 0.4383,
      "step": 431570
    },
    {
      "epoch": 176.66,
      "grad_norm": 2.066865921020508,
      "learning_rate": 1.7237903342395206e-07,
      "loss": 0.435,
      "step": 431580
    },
    {
      "epoch": 176.66,
      "grad_norm": 2.154045581817627,
      "learning_rate": 1.7231932803311345e-07,
      "loss": 0.4232,
      "step": 431590
    },
    {
      "epoch": 176.67,
      "grad_norm": 2.645585775375366,
      "learning_rate": 1.722596325892626e-07,
      "loss": 0.4549,
      "step": 431600
    },
    {
      "epoch": 176.67,
      "grad_norm": 1.982903003692627,
      "learning_rate": 1.7219994709267272e-07,
      "loss": 0.4555,
      "step": 431610
    },
    {
      "epoch": 176.68,
      "grad_norm": 2.9769856929779053,
      "learning_rate": 1.7214027154361705e-07,
      "loss": 0.4228,
      "step": 431620
    },
    {
      "epoch": 176.68,
      "grad_norm": 2.0255045890808105,
      "learning_rate": 1.720806059423696e-07,
      "loss": 0.4497,
      "step": 431630
    },
    {
      "epoch": 176.68,
      "grad_norm": 2.427042007446289,
      "learning_rate": 1.7202095028920302e-07,
      "loss": 0.4356,
      "step": 431640
    },
    {
      "epoch": 176.69,
      "grad_norm": 1.9900003671646118,
      "learning_rate": 1.719613045843911e-07,
      "loss": 0.4424,
      "step": 431650
    },
    {
      "epoch": 176.69,
      "grad_norm": 2.4772980213165283,
      "learning_rate": 1.7190166882820678e-07,
      "loss": 0.4536,
      "step": 431660
    },
    {
      "epoch": 176.7,
      "grad_norm": 2.9918017387390137,
      "learning_rate": 1.7184204302092326e-07,
      "loss": 0.4449,
      "step": 431670
    },
    {
      "epoch": 176.7,
      "grad_norm": 2.055006742477417,
      "learning_rate": 1.717824271628138e-07,
      "loss": 0.4314,
      "step": 431680
    },
    {
      "epoch": 176.7,
      "grad_norm": 1.8482999801635742,
      "learning_rate": 1.717228212541516e-07,
      "loss": 0.4506,
      "step": 431690
    },
    {
      "epoch": 176.71,
      "grad_norm": 2.0536155700683594,
      "learning_rate": 1.7166322529520931e-07,
      "loss": 0.4561,
      "step": 431700
    },
    {
      "epoch": 176.71,
      "grad_norm": 2.0516207218170166,
      "learning_rate": 1.7160363928625997e-07,
      "loss": 0.4649,
      "step": 431710
    },
    {
      "epoch": 176.72,
      "grad_norm": 1.5935500860214233,
      "learning_rate": 1.71544063227577e-07,
      "loss": 0.4422,
      "step": 431720
    },
    {
      "epoch": 176.72,
      "grad_norm": 1.757527470588684,
      "learning_rate": 1.714844971194329e-07,
      "loss": 0.4398,
      "step": 431730
    },
    {
      "epoch": 176.73,
      "grad_norm": 2.6171274185180664,
      "learning_rate": 1.7142494096210082e-07,
      "loss": 0.4247,
      "step": 431740
    },
    {
      "epoch": 176.73,
      "grad_norm": 2.400813102722168,
      "learning_rate": 1.713653947558532e-07,
      "loss": 0.4331,
      "step": 431750
    },
    {
      "epoch": 176.73,
      "grad_norm": 2.140814781188965,
      "learning_rate": 1.7130585850096308e-07,
      "loss": 0.4572,
      "step": 431760
    },
    {
      "epoch": 176.74,
      "grad_norm": 2.445617437362671,
      "learning_rate": 1.7124633219770309e-07,
      "loss": 0.4548,
      "step": 431770
    },
    {
      "epoch": 176.74,
      "grad_norm": 1.8294545412063599,
      "learning_rate": 1.7118681584634592e-07,
      "loss": 0.456,
      "step": 431780
    },
    {
      "epoch": 176.75,
      "grad_norm": 2.118277072906494,
      "learning_rate": 1.7112730944716431e-07,
      "loss": 0.4402,
      "step": 431790
    },
    {
      "epoch": 176.75,
      "grad_norm": 2.422351121902466,
      "learning_rate": 1.7106781300043094e-07,
      "loss": 0.4494,
      "step": 431800
    },
    {
      "epoch": 176.75,
      "grad_norm": 2.3020315170288086,
      "learning_rate": 1.7100832650641798e-07,
      "loss": 0.4514,
      "step": 431810
    },
    {
      "epoch": 176.76,
      "grad_norm": 2.273653507232666,
      "learning_rate": 1.7094884996539816e-07,
      "loss": 0.434,
      "step": 431820
    },
    {
      "epoch": 176.76,
      "grad_norm": 2.307987928390503,
      "learning_rate": 1.7088938337764387e-07,
      "loss": 0.4377,
      "step": 431830
    },
    {
      "epoch": 176.77,
      "grad_norm": 2.2159793376922607,
      "learning_rate": 1.7082992674342758e-07,
      "loss": 0.4611,
      "step": 431840
    },
    {
      "epoch": 176.77,
      "grad_norm": 1.8632954359054565,
      "learning_rate": 1.7077048006302175e-07,
      "loss": 0.4673,
      "step": 431850
    },
    {
      "epoch": 176.77,
      "grad_norm": 2.004016399383545,
      "learning_rate": 1.7071104333669824e-07,
      "loss": 0.4244,
      "step": 431860
    },
    {
      "epoch": 176.78,
      "grad_norm": 2.7033348083496094,
      "learning_rate": 1.7065161656473004e-07,
      "loss": 0.4428,
      "step": 431870
    },
    {
      "epoch": 176.78,
      "grad_norm": 2.010559320449829,
      "learning_rate": 1.7059219974738905e-07,
      "loss": 0.4349,
      "step": 431880
    },
    {
      "epoch": 176.79,
      "grad_norm": 2.2729544639587402,
      "learning_rate": 1.7053279288494725e-07,
      "loss": 0.4391,
      "step": 431890
    },
    {
      "epoch": 176.79,
      "grad_norm": 1.8222819566726685,
      "learning_rate": 1.7047339597767674e-07,
      "loss": 0.4214,
      "step": 431900
    },
    {
      "epoch": 176.79,
      "grad_norm": 1.8702729940414429,
      "learning_rate": 1.7041400902585e-07,
      "loss": 0.4452,
      "step": 431910
    },
    {
      "epoch": 176.8,
      "grad_norm": 1.8965522050857544,
      "learning_rate": 1.7035463202973867e-07,
      "loss": 0.4211,
      "step": 431920
    },
    {
      "epoch": 176.8,
      "grad_norm": 2.544053554534912,
      "learning_rate": 1.702952649896147e-07,
      "loss": 0.4335,
      "step": 431930
    },
    {
      "epoch": 176.81,
      "grad_norm": 1.9590603113174438,
      "learning_rate": 1.7023590790575049e-07,
      "loss": 0.4404,
      "step": 431940
    },
    {
      "epoch": 176.81,
      "grad_norm": 1.9365919828414917,
      "learning_rate": 1.7017656077841771e-07,
      "loss": 0.4386,
      "step": 431950
    },
    {
      "epoch": 176.82,
      "grad_norm": 1.7392213344573975,
      "learning_rate": 1.7011722360788828e-07,
      "loss": 0.4267,
      "step": 431960
    },
    {
      "epoch": 176.82,
      "grad_norm": 3.1935319900512695,
      "learning_rate": 1.7005789639443412e-07,
      "loss": 0.4512,
      "step": 431970
    },
    {
      "epoch": 176.82,
      "grad_norm": 1.7444974184036255,
      "learning_rate": 1.6999857913832662e-07,
      "loss": 0.4305,
      "step": 431980
    },
    {
      "epoch": 176.83,
      "grad_norm": 1.8407856225967407,
      "learning_rate": 1.6993927183983772e-07,
      "loss": 0.4385,
      "step": 431990
    },
    {
      "epoch": 176.83,
      "grad_norm": 1.976675271987915,
      "learning_rate": 1.698799744992393e-07,
      "loss": 0.4219,
      "step": 432000
    },
    {
      "epoch": 176.84,
      "grad_norm": 2.1767446994781494,
      "learning_rate": 1.6982068711680306e-07,
      "loss": 0.4306,
      "step": 432010
    },
    {
      "epoch": 176.84,
      "grad_norm": 1.8601226806640625,
      "learning_rate": 1.6976140969280011e-07,
      "loss": 0.4601,
      "step": 432020
    },
    {
      "epoch": 176.84,
      "grad_norm": 1.9220404624938965,
      "learning_rate": 1.6970214222750212e-07,
      "loss": 0.4632,
      "step": 432030
    },
    {
      "epoch": 176.85,
      "grad_norm": 1.9158568382263184,
      "learning_rate": 1.6964288472118048e-07,
      "loss": 0.4437,
      "step": 432040
    },
    {
      "epoch": 176.85,
      "grad_norm": 1.9511148929595947,
      "learning_rate": 1.695836371741071e-07,
      "loss": 0.4628,
      "step": 432050
    },
    {
      "epoch": 176.86,
      "grad_norm": 2.085211992263794,
      "learning_rate": 1.6952439958655287e-07,
      "loss": 0.4546,
      "step": 432060
    },
    {
      "epoch": 176.86,
      "grad_norm": 1.7721723318099976,
      "learning_rate": 1.694651719587897e-07,
      "loss": 0.4273,
      "step": 432070
    },
    {
      "epoch": 176.86,
      "grad_norm": 1.8545063734054565,
      "learning_rate": 1.6940595429108824e-07,
      "loss": 0.4299,
      "step": 432080
    },
    {
      "epoch": 176.87,
      "grad_norm": 2.185932159423828,
      "learning_rate": 1.6934674658372033e-07,
      "loss": 0.4375,
      "step": 432090
    },
    {
      "epoch": 176.87,
      "grad_norm": 2.2547645568847656,
      "learning_rate": 1.6928754883695694e-07,
      "loss": 0.446,
      "step": 432100
    },
    {
      "epoch": 176.88,
      "grad_norm": 1.7671706676483154,
      "learning_rate": 1.6922836105106941e-07,
      "loss": 0.4288,
      "step": 432110
    },
    {
      "epoch": 176.88,
      "grad_norm": 2.118147373199463,
      "learning_rate": 1.691691832263281e-07,
      "loss": 0.4448,
      "step": 432120
    },
    {
      "epoch": 176.88,
      "grad_norm": 1.9264196157455444,
      "learning_rate": 1.6911001536300522e-07,
      "loss": 0.4661,
      "step": 432130
    },
    {
      "epoch": 176.89,
      "grad_norm": 1.8414541482925415,
      "learning_rate": 1.6905085746137136e-07,
      "loss": 0.4394,
      "step": 432140
    },
    {
      "epoch": 176.89,
      "grad_norm": 2.220855474472046,
      "learning_rate": 1.689917095216979e-07,
      "loss": 0.4502,
      "step": 432150
    },
    {
      "epoch": 176.9,
      "grad_norm": 2.0836191177368164,
      "learning_rate": 1.6893257154425499e-07,
      "loss": 0.4548,
      "step": 432160
    },
    {
      "epoch": 176.9,
      "grad_norm": 1.916701078414917,
      "learning_rate": 1.6887344352931398e-07,
      "loss": 0.4464,
      "step": 432170
    },
    {
      "epoch": 176.91,
      "grad_norm": 2.2394204139709473,
      "learning_rate": 1.6881432547714552e-07,
      "loss": 0.4342,
      "step": 432180
    },
    {
      "epoch": 176.91,
      "grad_norm": 2.4003679752349854,
      "learning_rate": 1.6875521738802044e-07,
      "loss": 0.4218,
      "step": 432190
    },
    {
      "epoch": 176.91,
      "grad_norm": 1.9317703247070312,
      "learning_rate": 1.686961192622097e-07,
      "loss": 0.4539,
      "step": 432200
    },
    {
      "epoch": 176.92,
      "grad_norm": 2.146588087081909,
      "learning_rate": 1.6863703109998407e-07,
      "loss": 0.4414,
      "step": 432210
    },
    {
      "epoch": 176.92,
      "grad_norm": 2.230536460876465,
      "learning_rate": 1.6857795290161398e-07,
      "loss": 0.4657,
      "step": 432220
    },
    {
      "epoch": 176.93,
      "grad_norm": 2.1395084857940674,
      "learning_rate": 1.685188846673703e-07,
      "loss": 0.4439,
      "step": 432230
    },
    {
      "epoch": 176.93,
      "grad_norm": 2.8629462718963623,
      "learning_rate": 1.6845982639752338e-07,
      "loss": 0.424,
      "step": 432240
    },
    {
      "epoch": 176.93,
      "grad_norm": 2.1754353046417236,
      "learning_rate": 1.684007780923438e-07,
      "loss": 0.4329,
      "step": 432250
    },
    {
      "epoch": 176.94,
      "grad_norm": 2.3305840492248535,
      "learning_rate": 1.68341739752102e-07,
      "loss": 0.4388,
      "step": 432260
    },
    {
      "epoch": 176.94,
      "grad_norm": 2.2450742721557617,
      "learning_rate": 1.6828271137706877e-07,
      "loss": 0.4237,
      "step": 432270
    },
    {
      "epoch": 176.95,
      "grad_norm": 1.8528093099594116,
      "learning_rate": 1.6822369296751423e-07,
      "loss": 0.4461,
      "step": 432280
    },
    {
      "epoch": 176.95,
      "grad_norm": 1.99519681930542,
      "learning_rate": 1.681646845237088e-07,
      "loss": 0.4497,
      "step": 432290
    },
    {
      "epoch": 176.95,
      "grad_norm": 2.065563440322876,
      "learning_rate": 1.6810568604592275e-07,
      "loss": 0.4456,
      "step": 432300
    },
    {
      "epoch": 176.96,
      "grad_norm": 1.9732550382614136,
      "learning_rate": 1.680466975344262e-07,
      "loss": 0.4534,
      "step": 432310
    },
    {
      "epoch": 176.96,
      "grad_norm": 2.2285544872283936,
      "learning_rate": 1.679877189894898e-07,
      "loss": 0.4501,
      "step": 432320
    },
    {
      "epoch": 176.97,
      "grad_norm": 2.6991353034973145,
      "learning_rate": 1.6792875041138338e-07,
      "loss": 0.4648,
      "step": 432330
    },
    {
      "epoch": 176.97,
      "grad_norm": 2.29333758354187,
      "learning_rate": 1.6786979180037699e-07,
      "loss": 0.4283,
      "step": 432340
    },
    {
      "epoch": 176.98,
      "grad_norm": 2.1687841415405273,
      "learning_rate": 1.6781084315674078e-07,
      "loss": 0.4588,
      "step": 432350
    },
    {
      "epoch": 176.98,
      "grad_norm": 1.959145426750183,
      "learning_rate": 1.6775190448074506e-07,
      "loss": 0.464,
      "step": 432360
    },
    {
      "epoch": 176.98,
      "grad_norm": 1.7553465366363525,
      "learning_rate": 1.6769297577265973e-07,
      "loss": 0.4315,
      "step": 432370
    },
    {
      "epoch": 176.99,
      "grad_norm": 1.8961292505264282,
      "learning_rate": 1.6763405703275429e-07,
      "loss": 0.4433,
      "step": 432380
    },
    {
      "epoch": 176.99,
      "grad_norm": 1.8585450649261475,
      "learning_rate": 1.6757514826129915e-07,
      "loss": 0.4444,
      "step": 432390
    },
    {
      "epoch": 177.0,
      "grad_norm": 1.8905690908432007,
      "learning_rate": 1.6751624945856414e-07,
      "loss": 0.4467,
      "step": 432400
    },
    {
      "epoch": 177.0,
      "grad_norm": 2.0829927921295166,
      "learning_rate": 1.6745736062481883e-07,
      "loss": 0.4369,
      "step": 432410
    },
    {
      "epoch": 177.0,
      "eval_loss": 0.4450966417789459,
      "eval_runtime": 52.479,
      "eval_samples_per_second": 65.722,
      "eval_steps_per_second": 8.232,
      "step": 432411
    },
    {
      "epoch": 177.0,
      "grad_norm": 2.280409336090088,
      "learning_rate": 1.6739848176033277e-07,
      "loss": 0.4352,
      "step": 432420
    },
    {
      "epoch": 177.01,
      "grad_norm": 2.3776988983154297,
      "learning_rate": 1.6733961286537636e-07,
      "loss": 0.4442,
      "step": 432430
    },
    {
      "epoch": 177.01,
      "grad_norm": 3.1742677688598633,
      "learning_rate": 1.6728075394021862e-07,
      "loss": 0.4591,
      "step": 432440
    },
    {
      "epoch": 177.02,
      "grad_norm": 1.7829164266586304,
      "learning_rate": 1.672219049851294e-07,
      "loss": 0.4449,
      "step": 432450
    },
    {
      "epoch": 177.02,
      "grad_norm": 1.928145170211792,
      "learning_rate": 1.6716306600037878e-07,
      "loss": 0.4357,
      "step": 432460
    },
    {
      "epoch": 177.02,
      "grad_norm": 2.4988484382629395,
      "learning_rate": 1.671042369862356e-07,
      "loss": 0.4361,
      "step": 432470
    },
    {
      "epoch": 177.03,
      "grad_norm": 1.994888424873352,
      "learning_rate": 1.670454179429694e-07,
      "loss": 0.453,
      "step": 432480
    },
    {
      "epoch": 177.03,
      "grad_norm": 1.9410330057144165,
      "learning_rate": 1.6698660887084926e-07,
      "loss": 0.4354,
      "step": 432490
    },
    {
      "epoch": 177.04,
      "grad_norm": 2.2018935680389404,
      "learning_rate": 1.6692780977014577e-07,
      "loss": 0.4347,
      "step": 432500
    },
    {
      "epoch": 177.04,
      "grad_norm": 2.482907295227051,
      "learning_rate": 1.6686902064112747e-07,
      "loss": 0.4613,
      "step": 432510
    },
    {
      "epoch": 177.04,
      "grad_norm": 2.0741443634033203,
      "learning_rate": 1.6681024148406368e-07,
      "loss": 0.4413,
      "step": 432520
    },
    {
      "epoch": 177.05,
      "grad_norm": 2.331557035446167,
      "learning_rate": 1.66751472299224e-07,
      "loss": 0.4249,
      "step": 432530
    },
    {
      "epoch": 177.05,
      "grad_norm": 1.9960602521896362,
      "learning_rate": 1.6669271308687717e-07,
      "loss": 0.4578,
      "step": 432540
    },
    {
      "epoch": 177.06,
      "grad_norm": 1.914239525794983,
      "learning_rate": 1.6663396384729283e-07,
      "loss": 0.462,
      "step": 432550
    },
    {
      "epoch": 177.06,
      "grad_norm": 1.7694590091705322,
      "learning_rate": 1.6657522458074e-07,
      "loss": 0.4475,
      "step": 432560
    },
    {
      "epoch": 177.07,
      "grad_norm": 1.6897341012954712,
      "learning_rate": 1.6651649528748746e-07,
      "loss": 0.444,
      "step": 432570
    },
    {
      "epoch": 177.07,
      "grad_norm": 2.0819973945617676,
      "learning_rate": 1.664577759678043e-07,
      "loss": 0.431,
      "step": 432580
    },
    {
      "epoch": 177.07,
      "grad_norm": 2.0079360008239746,
      "learning_rate": 1.663990666219598e-07,
      "loss": 0.437,
      "step": 432590
    },
    {
      "epoch": 177.08,
      "grad_norm": 2.164194107055664,
      "learning_rate": 1.663403672502231e-07,
      "loss": 0.4592,
      "step": 432600
    },
    {
      "epoch": 177.08,
      "grad_norm": 2.5619664192199707,
      "learning_rate": 1.662816778528621e-07,
      "loss": 0.4277,
      "step": 432610
    },
    {
      "epoch": 177.09,
      "grad_norm": 1.9530503749847412,
      "learning_rate": 1.6622299843014642e-07,
      "loss": 0.4481,
      "step": 432620
    },
    {
      "epoch": 177.09,
      "grad_norm": 2.2075648307800293,
      "learning_rate": 1.661643289823449e-07,
      "loss": 0.422,
      "step": 432630
    },
    {
      "epoch": 177.09,
      "grad_norm": 2.15531587600708,
      "learning_rate": 1.6610566950972575e-07,
      "loss": 0.4453,
      "step": 432640
    },
    {
      "epoch": 177.1,
      "grad_norm": 1.7209892272949219,
      "learning_rate": 1.6604702001255832e-07,
      "loss": 0.442,
      "step": 432650
    },
    {
      "epoch": 177.1,
      "grad_norm": 2.0266289710998535,
      "learning_rate": 1.6598838049111089e-07,
      "loss": 0.4534,
      "step": 432660
    },
    {
      "epoch": 177.11,
      "grad_norm": 2.4009249210357666,
      "learning_rate": 1.6592975094565198e-07,
      "loss": 0.4668,
      "step": 432670
    },
    {
      "epoch": 177.11,
      "grad_norm": 2.753178119659424,
      "learning_rate": 1.6587113137645015e-07,
      "loss": 0.4485,
      "step": 432680
    },
    {
      "epoch": 177.11,
      "grad_norm": 1.9777050018310547,
      "learning_rate": 1.6581252178377466e-07,
      "loss": 0.437,
      "step": 432690
    },
    {
      "epoch": 177.12,
      "grad_norm": 2.212876081466675,
      "learning_rate": 1.6575392216789307e-07,
      "loss": 0.4628,
      "step": 432700
    },
    {
      "epoch": 177.12,
      "grad_norm": 1.943078637123108,
      "learning_rate": 1.6569533252907466e-07,
      "loss": 0.4373,
      "step": 432710
    },
    {
      "epoch": 177.13,
      "grad_norm": 1.9936342239379883,
      "learning_rate": 1.656367528675869e-07,
      "loss": 0.4433,
      "step": 432720
    },
    {
      "epoch": 177.13,
      "grad_norm": 1.8763576745986938,
      "learning_rate": 1.6557818318369943e-07,
      "loss": 0.4519,
      "step": 432730
    },
    {
      "epoch": 177.13,
      "grad_norm": 2.07991623878479,
      "learning_rate": 1.6551962347767893e-07,
      "loss": 0.4594,
      "step": 432740
    },
    {
      "epoch": 177.14,
      "grad_norm": 2.1199967861175537,
      "learning_rate": 1.654610737497947e-07,
      "loss": 0.4487,
      "step": 432750
    },
    {
      "epoch": 177.14,
      "grad_norm": 2.0024101734161377,
      "learning_rate": 1.6540253400031477e-07,
      "loss": 0.4417,
      "step": 432760
    },
    {
      "epoch": 177.15,
      "grad_norm": 1.8995270729064941,
      "learning_rate": 1.6534400422950693e-07,
      "loss": 0.4369,
      "step": 432770
    },
    {
      "epoch": 177.15,
      "grad_norm": 2.2469167709350586,
      "learning_rate": 1.652854844376399e-07,
      "loss": 0.4448,
      "step": 432780
    },
    {
      "epoch": 177.16,
      "grad_norm": 2.435774087905884,
      "learning_rate": 1.6522697462498123e-07,
      "loss": 0.4455,
      "step": 432790
    },
    {
      "epoch": 177.16,
      "grad_norm": 1.895236849784851,
      "learning_rate": 1.6516847479179919e-07,
      "loss": 0.4312,
      "step": 432800
    },
    {
      "epoch": 177.16,
      "grad_norm": 2.1995773315429688,
      "learning_rate": 1.6510998493836176e-07,
      "loss": 0.4473,
      "step": 432810
    },
    {
      "epoch": 177.17,
      "grad_norm": 2.5969226360321045,
      "learning_rate": 1.6505150506493674e-07,
      "loss": 0.4326,
      "step": 432820
    },
    {
      "epoch": 177.17,
      "grad_norm": 2.116027355194092,
      "learning_rate": 1.6499303517179212e-07,
      "loss": 0.4439,
      "step": 432830
    },
    {
      "epoch": 177.18,
      "grad_norm": 2.3291475772857666,
      "learning_rate": 1.649345752591959e-07,
      "loss": 0.4264,
      "step": 432840
    },
    {
      "epoch": 177.18,
      "grad_norm": 2.4187088012695312,
      "learning_rate": 1.648761253274156e-07,
      "loss": 0.4227,
      "step": 432850
    },
    {
      "epoch": 177.18,
      "grad_norm": 2.5130550861358643,
      "learning_rate": 1.64817685376719e-07,
      "loss": 0.451,
      "step": 432860
    },
    {
      "epoch": 177.19,
      "grad_norm": 1.7845709323883057,
      "learning_rate": 1.6475925540737406e-07,
      "loss": 0.4356,
      "step": 432870
    },
    {
      "epoch": 177.19,
      "grad_norm": 2.2371253967285156,
      "learning_rate": 1.6470083541964802e-07,
      "loss": 0.4502,
      "step": 432880
    },
    {
      "epoch": 177.2,
      "grad_norm": 2.0812556743621826,
      "learning_rate": 1.6464242541380895e-07,
      "loss": 0.4324,
      "step": 432890
    },
    {
      "epoch": 177.2,
      "grad_norm": 2.02429461479187,
      "learning_rate": 1.6458402539012426e-07,
      "loss": 0.4539,
      "step": 432900
    },
    {
      "epoch": 177.2,
      "grad_norm": 2.9633030891418457,
      "learning_rate": 1.6452563534886127e-07,
      "loss": 0.4592,
      "step": 432910
    },
    {
      "epoch": 177.21,
      "grad_norm": 2.1558969020843506,
      "learning_rate": 1.6446725529028768e-07,
      "loss": 0.4798,
      "step": 432920
    },
    {
      "epoch": 177.21,
      "grad_norm": 2.141174554824829,
      "learning_rate": 1.6440888521467074e-07,
      "loss": 0.4352,
      "step": 432930
    },
    {
      "epoch": 177.22,
      "grad_norm": 2.416152238845825,
      "learning_rate": 1.6435052512227793e-07,
      "loss": 0.4418,
      "step": 432940
    },
    {
      "epoch": 177.22,
      "grad_norm": 2.176548719406128,
      "learning_rate": 1.6429217501337653e-07,
      "loss": 0.4378,
      "step": 432950
    },
    {
      "epoch": 177.22,
      "grad_norm": 2.163487434387207,
      "learning_rate": 1.6423383488823397e-07,
      "loss": 0.4369,
      "step": 432960
    },
    {
      "epoch": 177.23,
      "grad_norm": 2.01285982131958,
      "learning_rate": 1.6417550474711752e-07,
      "loss": 0.4461,
      "step": 432970
    },
    {
      "epoch": 177.23,
      "grad_norm": 1.5398262739181519,
      "learning_rate": 1.6411718459029416e-07,
      "loss": 0.4391,
      "step": 432980
    },
    {
      "epoch": 177.24,
      "grad_norm": 1.9989367723464966,
      "learning_rate": 1.6405887441803137e-07,
      "loss": 0.4344,
      "step": 432990
    },
    {
      "epoch": 177.24,
      "grad_norm": 1.9775638580322266,
      "learning_rate": 1.6400057423059586e-07,
      "loss": 0.4327,
      "step": 433000
    },
    {
      "epoch": 177.25,
      "grad_norm": 1.9800095558166504,
      "learning_rate": 1.639422840282549e-07,
      "loss": 0.4496,
      "step": 433010
    },
    {
      "epoch": 177.25,
      "grad_norm": 2.404865026473999,
      "learning_rate": 1.638840038112754e-07,
      "loss": 0.4501,
      "step": 433020
    },
    {
      "epoch": 177.25,
      "grad_norm": 1.7946276664733887,
      "learning_rate": 1.6382573357992464e-07,
      "loss": 0.444,
      "step": 433030
    },
    {
      "epoch": 177.26,
      "grad_norm": 2.3089754581451416,
      "learning_rate": 1.637674733344696e-07,
      "loss": 0.4576,
      "step": 433040
    },
    {
      "epoch": 177.26,
      "grad_norm": 1.7424263954162598,
      "learning_rate": 1.637092230751767e-07,
      "loss": 0.4456,
      "step": 433050
    },
    {
      "epoch": 177.27,
      "grad_norm": 1.827637791633606,
      "learning_rate": 1.6365098280231242e-07,
      "loss": 0.4174,
      "step": 433060
    },
    {
      "epoch": 177.27,
      "grad_norm": 2.149603843688965,
      "learning_rate": 1.635927525161445e-07,
      "loss": 0.4465,
      "step": 433070
    },
    {
      "epoch": 177.27,
      "grad_norm": 2.0479114055633545,
      "learning_rate": 1.6353453221693942e-07,
      "loss": 0.4384,
      "step": 433080
    },
    {
      "epoch": 177.28,
      "grad_norm": 1.944166898727417,
      "learning_rate": 1.634763219049636e-07,
      "loss": 0.436,
      "step": 433090
    },
    {
      "epoch": 177.28,
      "grad_norm": 6.722015857696533,
      "learning_rate": 1.6341812158048402e-07,
      "loss": 0.4512,
      "step": 433100
    },
    {
      "epoch": 177.29,
      "grad_norm": 2.334364652633667,
      "learning_rate": 1.633599312437669e-07,
      "loss": 0.4521,
      "step": 433110
    },
    {
      "epoch": 177.29,
      "grad_norm": 2.3440728187561035,
      "learning_rate": 1.6330175089507915e-07,
      "loss": 0.446,
      "step": 433120
    },
    {
      "epoch": 177.29,
      "grad_norm": 2.1558070182800293,
      "learning_rate": 1.6324358053468705e-07,
      "loss": 0.4586,
      "step": 433130
    },
    {
      "epoch": 177.3,
      "grad_norm": 2.4226126670837402,
      "learning_rate": 1.6318542016285725e-07,
      "loss": 0.4428,
      "step": 433140
    },
    {
      "epoch": 177.3,
      "grad_norm": 1.9974851608276367,
      "learning_rate": 1.6312726977985598e-07,
      "loss": 0.4372,
      "step": 433150
    },
    {
      "epoch": 177.31,
      "grad_norm": 2.305279016494751,
      "learning_rate": 1.630691293859499e-07,
      "loss": 0.449,
      "step": 433160
    },
    {
      "epoch": 177.31,
      "grad_norm": 2.307441234588623,
      "learning_rate": 1.6301099898140503e-07,
      "loss": 0.4484,
      "step": 433170
    },
    {
      "epoch": 177.31,
      "grad_norm": 2.1405506134033203,
      "learning_rate": 1.6295287856648827e-07,
      "loss": 0.4416,
      "step": 433180
    },
    {
      "epoch": 177.32,
      "grad_norm": 2.590602159500122,
      "learning_rate": 1.628947681414648e-07,
      "loss": 0.4447,
      "step": 433190
    },
    {
      "epoch": 177.32,
      "grad_norm": 2.186528444290161,
      "learning_rate": 1.628366677066016e-07,
      "loss": 0.4441,
      "step": 433200
    },
    {
      "epoch": 177.33,
      "grad_norm": 2.0990066528320312,
      "learning_rate": 1.627785772621646e-07,
      "loss": 0.4509,
      "step": 433210
    },
    {
      "epoch": 177.33,
      "grad_norm": 2.2041592597961426,
      "learning_rate": 1.627204968084197e-07,
      "loss": 0.4491,
      "step": 433220
    },
    {
      "epoch": 177.34,
      "grad_norm": 1.8549041748046875,
      "learning_rate": 1.626624263456334e-07,
      "loss": 0.4473,
      "step": 433230
    },
    {
      "epoch": 177.34,
      "grad_norm": 1.6526849269866943,
      "learning_rate": 1.626043658740716e-07,
      "loss": 0.4379,
      "step": 433240
    },
    {
      "epoch": 177.34,
      "grad_norm": 1.9424225091934204,
      "learning_rate": 1.6254631539399948e-07,
      "loss": 0.4445,
      "step": 433250
    },
    {
      "epoch": 177.35,
      "grad_norm": 2.0147881507873535,
      "learning_rate": 1.6248827490568427e-07,
      "loss": 0.4656,
      "step": 433260
    },
    {
      "epoch": 177.35,
      "grad_norm": 2.0164270401000977,
      "learning_rate": 1.624302444093911e-07,
      "loss": 0.4508,
      "step": 433270
    },
    {
      "epoch": 177.36,
      "grad_norm": 2.4386253356933594,
      "learning_rate": 1.6237222390538593e-07,
      "loss": 0.4246,
      "step": 433280
    },
    {
      "epoch": 177.36,
      "grad_norm": 2.227440595626831,
      "learning_rate": 1.623142133939344e-07,
      "loss": 0.4563,
      "step": 433290
    },
    {
      "epoch": 177.36,
      "grad_norm": 2.504455327987671,
      "learning_rate": 1.6225621287530248e-07,
      "loss": 0.4573,
      "step": 433300
    },
    {
      "epoch": 177.37,
      "grad_norm": 2.024062156677246,
      "learning_rate": 1.6219822234975607e-07,
      "loss": 0.4508,
      "step": 433310
    },
    {
      "epoch": 177.37,
      "grad_norm": 2.3010897636413574,
      "learning_rate": 1.6214024181756035e-07,
      "loss": 0.4483,
      "step": 433320
    },
    {
      "epoch": 177.38,
      "grad_norm": 2.189213752746582,
      "learning_rate": 1.6208227127898072e-07,
      "loss": 0.4422,
      "step": 433330
    },
    {
      "epoch": 177.38,
      "grad_norm": 2.3809168338775635,
      "learning_rate": 1.6202431073428337e-07,
      "loss": 0.4495,
      "step": 433340
    },
    {
      "epoch": 177.38,
      "grad_norm": 2.09682559967041,
      "learning_rate": 1.6196636018373346e-07,
      "loss": 0.4384,
      "step": 433350
    },
    {
      "epoch": 177.39,
      "grad_norm": 1.8604339361190796,
      "learning_rate": 1.619084196275964e-07,
      "loss": 0.4504,
      "step": 433360
    },
    {
      "epoch": 177.39,
      "grad_norm": 1.9297242164611816,
      "learning_rate": 1.6185048906613785e-07,
      "loss": 0.4257,
      "step": 433370
    },
    {
      "epoch": 177.4,
      "grad_norm": 1.9924962520599365,
      "learning_rate": 1.6179256849962299e-07,
      "loss": 0.4619,
      "step": 433380
    },
    {
      "epoch": 177.4,
      "grad_norm": 2.220726251602173,
      "learning_rate": 1.6173465792831694e-07,
      "loss": 0.4424,
      "step": 433390
    },
    {
      "epoch": 177.4,
      "grad_norm": 2.058051109313965,
      "learning_rate": 1.6167675735248566e-07,
      "loss": 0.4644,
      "step": 433400
    },
    {
      "epoch": 177.41,
      "grad_norm": 2.1095433235168457,
      "learning_rate": 1.6161886677239378e-07,
      "loss": 0.454,
      "step": 433410
    },
    {
      "epoch": 177.41,
      "grad_norm": 1.9941797256469727,
      "learning_rate": 1.615609861883067e-07,
      "loss": 0.4271,
      "step": 433420
    },
    {
      "epoch": 177.42,
      "grad_norm": 2.077796220779419,
      "learning_rate": 1.615031156004896e-07,
      "loss": 0.4359,
      "step": 433430
    },
    {
      "epoch": 177.42,
      "grad_norm": 2.1798267364501953,
      "learning_rate": 1.6144525500920704e-07,
      "loss": 0.4487,
      "step": 433440
    },
    {
      "epoch": 177.43,
      "grad_norm": 2.051985263824463,
      "learning_rate": 1.613874044147253e-07,
      "loss": 0.4602,
      "step": 433450
    },
    {
      "epoch": 177.43,
      "grad_norm": 2.0140504837036133,
      "learning_rate": 1.6132956381730818e-07,
      "loss": 0.4426,
      "step": 433460
    },
    {
      "epoch": 177.43,
      "grad_norm": 2.095886707305908,
      "learning_rate": 1.6127173321722136e-07,
      "loss": 0.4256,
      "step": 433470
    },
    {
      "epoch": 177.44,
      "grad_norm": 1.6228408813476562,
      "learning_rate": 1.6121391261472924e-07,
      "loss": 0.4322,
      "step": 433480
    },
    {
      "epoch": 177.44,
      "grad_norm": 1.7793442010879517,
      "learning_rate": 1.6115610201009694e-07,
      "loss": 0.4199,
      "step": 433490
    },
    {
      "epoch": 177.45,
      "grad_norm": 2.0820937156677246,
      "learning_rate": 1.6109830140358937e-07,
      "loss": 0.4485,
      "step": 433500
    },
    {
      "epoch": 177.45,
      "grad_norm": 2.11911678314209,
      "learning_rate": 1.6104051079547116e-07,
      "loss": 0.4243,
      "step": 433510
    },
    {
      "epoch": 177.45,
      "grad_norm": 2.222038745880127,
      "learning_rate": 1.6098273018600717e-07,
      "loss": 0.4456,
      "step": 433520
    },
    {
      "epoch": 177.46,
      "grad_norm": 1.9124401807785034,
      "learning_rate": 1.6092495957546183e-07,
      "loss": 0.4348,
      "step": 433530
    },
    {
      "epoch": 177.46,
      "grad_norm": 2.4099960327148438,
      "learning_rate": 1.608671989641e-07,
      "loss": 0.4429,
      "step": 433540
    },
    {
      "epoch": 177.47,
      "grad_norm": 1.8525198698043823,
      "learning_rate": 1.608094483521863e-07,
      "loss": 0.444,
      "step": 433550
    },
    {
      "epoch": 177.47,
      "grad_norm": 2.079927921295166,
      "learning_rate": 1.6075170773998514e-07,
      "loss": 0.4378,
      "step": 433560
    },
    {
      "epoch": 177.47,
      "grad_norm": 2.1393790245056152,
      "learning_rate": 1.606939771277611e-07,
      "loss": 0.445,
      "step": 433570
    },
    {
      "epoch": 177.48,
      "grad_norm": 2.3033971786499023,
      "learning_rate": 1.606362565157786e-07,
      "loss": 0.459,
      "step": 433580
    },
    {
      "epoch": 177.48,
      "grad_norm": 2.4995064735412598,
      "learning_rate": 1.6057854590430197e-07,
      "loss": 0.4562,
      "step": 433590
    },
    {
      "epoch": 177.49,
      "grad_norm": 1.9951645135879517,
      "learning_rate": 1.605208452935959e-07,
      "loss": 0.4219,
      "step": 433600
    },
    {
      "epoch": 177.49,
      "grad_norm": 1.9319711923599243,
      "learning_rate": 1.6046315468392418e-07,
      "loss": 0.4479,
      "step": 433610
    },
    {
      "epoch": 177.49,
      "grad_norm": 2.9609174728393555,
      "learning_rate": 1.60405474075552e-07,
      "loss": 0.4419,
      "step": 433620
    },
    {
      "epoch": 177.5,
      "grad_norm": 2.1492760181427,
      "learning_rate": 1.6034780346874213e-07,
      "loss": 0.45,
      "step": 433630
    },
    {
      "epoch": 177.5,
      "grad_norm": 1.9191633462905884,
      "learning_rate": 1.6029014286376005e-07,
      "loss": 0.4316,
      "step": 433640
    },
    {
      "epoch": 177.51,
      "grad_norm": 1.7149568796157837,
      "learning_rate": 1.602324922608693e-07,
      "loss": 0.4212,
      "step": 433650
    },
    {
      "epoch": 177.51,
      "grad_norm": 2.1230459213256836,
      "learning_rate": 1.60174851660334e-07,
      "loss": 0.4422,
      "step": 433660
    },
    {
      "epoch": 177.52,
      "grad_norm": 1.731863021850586,
      "learning_rate": 1.6011722106241827e-07,
      "loss": 0.4119,
      "step": 433670
    },
    {
      "epoch": 177.52,
      "grad_norm": 1.9460704326629639,
      "learning_rate": 1.6005960046738619e-07,
      "loss": 0.4197,
      "step": 433680
    },
    {
      "epoch": 177.52,
      "grad_norm": 1.9598000049591064,
      "learning_rate": 1.6000198987550193e-07,
      "loss": 0.4583,
      "step": 433690
    },
    {
      "epoch": 177.53,
      "grad_norm": 1.9191036224365234,
      "learning_rate": 1.599443892870288e-07,
      "loss": 0.456,
      "step": 433700
    },
    {
      "epoch": 177.53,
      "grad_norm": 2.098482608795166,
      "learning_rate": 1.5988679870223088e-07,
      "loss": 0.4584,
      "step": 433710
    },
    {
      "epoch": 177.54,
      "grad_norm": 2.146035671234131,
      "learning_rate": 1.598292181213723e-07,
      "loss": 0.4485,
      "step": 433720
    },
    {
      "epoch": 177.54,
      "grad_norm": 2.222968816757202,
      "learning_rate": 1.5977164754471667e-07,
      "loss": 0.4692,
      "step": 433730
    },
    {
      "epoch": 177.54,
      "grad_norm": 2.469499349594116,
      "learning_rate": 1.597140869725273e-07,
      "loss": 0.4231,
      "step": 433740
    },
    {
      "epoch": 177.55,
      "grad_norm": 1.8914129734039307,
      "learning_rate": 1.5965653640506857e-07,
      "loss": 0.4265,
      "step": 433750
    },
    {
      "epoch": 177.55,
      "grad_norm": 2.0656981468200684,
      "learning_rate": 1.595989958426038e-07,
      "loss": 0.4429,
      "step": 433760
    },
    {
      "epoch": 177.56,
      "grad_norm": 1.9777326583862305,
      "learning_rate": 1.5954146528539632e-07,
      "loss": 0.4383,
      "step": 433770
    },
    {
      "epoch": 177.56,
      "grad_norm": 1.7130622863769531,
      "learning_rate": 1.5948394473371e-07,
      "loss": 0.4365,
      "step": 433780
    },
    {
      "epoch": 177.56,
      "grad_norm": 1.8132636547088623,
      "learning_rate": 1.5942643418780793e-07,
      "loss": 0.4435,
      "step": 433790
    },
    {
      "epoch": 177.57,
      "grad_norm": 1.9118313789367676,
      "learning_rate": 1.5936893364795392e-07,
      "loss": 0.437,
      "step": 433800
    },
    {
      "epoch": 177.57,
      "grad_norm": 2.3722522258758545,
      "learning_rate": 1.5931144311441103e-07,
      "loss": 0.4476,
      "step": 433810
    },
    {
      "epoch": 177.58,
      "grad_norm": 1.8541069030761719,
      "learning_rate": 1.5925396258744316e-07,
      "loss": 0.4412,
      "step": 433820
    },
    {
      "epoch": 177.58,
      "grad_norm": 2.5953855514526367,
      "learning_rate": 1.5919649206731334e-07,
      "loss": 0.4317,
      "step": 433830
    },
    {
      "epoch": 177.58,
      "grad_norm": 2.0432305335998535,
      "learning_rate": 1.5913903155428466e-07,
      "loss": 0.4448,
      "step": 433840
    },
    {
      "epoch": 177.59,
      "grad_norm": 1.7156740427017212,
      "learning_rate": 1.5908158104862073e-07,
      "loss": 0.4415,
      "step": 433850
    },
    {
      "epoch": 177.59,
      "grad_norm": 1.860651969909668,
      "learning_rate": 1.5902414055058432e-07,
      "loss": 0.432,
      "step": 433860
    },
    {
      "epoch": 177.6,
      "grad_norm": 2.0690383911132812,
      "learning_rate": 1.589667100604388e-07,
      "loss": 0.4351,
      "step": 433870
    },
    {
      "epoch": 177.6,
      "grad_norm": 1.9590743780136108,
      "learning_rate": 1.589092895784472e-07,
      "loss": 0.432,
      "step": 433880
    },
    {
      "epoch": 177.61,
      "grad_norm": 1.911629557609558,
      "learning_rate": 1.5885187910487261e-07,
      "loss": 0.4537,
      "step": 433890
    },
    {
      "epoch": 177.61,
      "grad_norm": 2.1782004833221436,
      "learning_rate": 1.587944786399781e-07,
      "loss": 0.4597,
      "step": 433900
    },
    {
      "epoch": 177.61,
      "grad_norm": 2.489924430847168,
      "learning_rate": 1.5873708818402625e-07,
      "loss": 0.4621,
      "step": 433910
    },
    {
      "epoch": 177.62,
      "grad_norm": 2.2826764583587646,
      "learning_rate": 1.5867970773728007e-07,
      "loss": 0.4493,
      "step": 433920
    },
    {
      "epoch": 177.62,
      "grad_norm": 2.0697972774505615,
      "learning_rate": 1.5862233730000243e-07,
      "loss": 0.4364,
      "step": 433930
    },
    {
      "epoch": 177.63,
      "grad_norm": 2.3047566413879395,
      "learning_rate": 1.5856497687245636e-07,
      "loss": 0.452,
      "step": 433940
    },
    {
      "epoch": 177.63,
      "grad_norm": 2.081307888031006,
      "learning_rate": 1.5850762645490444e-07,
      "loss": 0.451,
      "step": 433950
    },
    {
      "epoch": 177.63,
      "grad_norm": 2.149423122406006,
      "learning_rate": 1.5845028604760944e-07,
      "loss": 0.4424,
      "step": 433960
    },
    {
      "epoch": 177.64,
      "grad_norm": 2.0859081745147705,
      "learning_rate": 1.583929556508337e-07,
      "loss": 0.442,
      "step": 433970
    },
    {
      "epoch": 177.64,
      "grad_norm": 2.3289377689361572,
      "learning_rate": 1.5833563526484054e-07,
      "loss": 0.4387,
      "step": 433980
    },
    {
      "epoch": 177.65,
      "grad_norm": 2.021662950515747,
      "learning_rate": 1.5827832488989196e-07,
      "loss": 0.4444,
      "step": 433990
    },
    {
      "epoch": 177.65,
      "grad_norm": 2.2333080768585205,
      "learning_rate": 1.5822102452625025e-07,
      "loss": 0.4494,
      "step": 434000
    },
    {
      "epoch": 177.65,
      "grad_norm": 3.1665449142456055,
      "learning_rate": 1.5816373417417854e-07,
      "loss": 0.4354,
      "step": 434010
    },
    {
      "epoch": 177.66,
      "grad_norm": 2.431126594543457,
      "learning_rate": 1.581064538339393e-07,
      "loss": 0.4312,
      "step": 434020
    },
    {
      "epoch": 177.66,
      "grad_norm": 2.1521480083465576,
      "learning_rate": 1.5804918350579486e-07,
      "loss": 0.4392,
      "step": 434030
    },
    {
      "epoch": 177.67,
      "grad_norm": 2.1660773754119873,
      "learning_rate": 1.57991923190007e-07,
      "loss": 0.4415,
      "step": 434040
    },
    {
      "epoch": 177.67,
      "grad_norm": 2.1388731002807617,
      "learning_rate": 1.5793467288683854e-07,
      "loss": 0.4443,
      "step": 434050
    },
    {
      "epoch": 177.67,
      "grad_norm": 2.3376622200012207,
      "learning_rate": 1.5787743259655147e-07,
      "loss": 0.457,
      "step": 434060
    },
    {
      "epoch": 177.68,
      "grad_norm": 2.043886184692383,
      "learning_rate": 1.578202023194081e-07,
      "loss": 0.4406,
      "step": 434070
    },
    {
      "epoch": 177.68,
      "grad_norm": 1.8293023109436035,
      "learning_rate": 1.5776298205567047e-07,
      "loss": 0.4482,
      "step": 434080
    },
    {
      "epoch": 177.69,
      "grad_norm": 2.1883370876312256,
      "learning_rate": 1.577057718056011e-07,
      "loss": 0.4581,
      "step": 434090
    },
    {
      "epoch": 177.69,
      "grad_norm": 2.008697032928467,
      "learning_rate": 1.5764857156946154e-07,
      "loss": 0.4393,
      "step": 434100
    },
    {
      "epoch": 177.7,
      "grad_norm": 1.6644480228424072,
      "learning_rate": 1.575913813475143e-07,
      "loss": 0.4313,
      "step": 434110
    },
    {
      "epoch": 177.7,
      "grad_norm": 2.285688877105713,
      "learning_rate": 1.5753420114002093e-07,
      "loss": 0.4544,
      "step": 434120
    },
    {
      "epoch": 177.7,
      "grad_norm": 2.5457892417907715,
      "learning_rate": 1.5747703094724367e-07,
      "loss": 0.4422,
      "step": 434130
    },
    {
      "epoch": 177.71,
      "grad_norm": 1.9308619499206543,
      "learning_rate": 1.574198707694441e-07,
      "loss": 0.4489,
      "step": 434140
    },
    {
      "epoch": 177.71,
      "grad_norm": 1.8162829875946045,
      "learning_rate": 1.5736272060688444e-07,
      "loss": 0.4354,
      "step": 434150
    },
    {
      "epoch": 177.72,
      "grad_norm": 1.786436915397644,
      "learning_rate": 1.5730558045982624e-07,
      "loss": 0.4289,
      "step": 434160
    },
    {
      "epoch": 177.72,
      "grad_norm": 1.9560306072235107,
      "learning_rate": 1.5724845032853126e-07,
      "loss": 0.4393,
      "step": 434170
    },
    {
      "epoch": 177.72,
      "grad_norm": 1.8884871006011963,
      "learning_rate": 1.571913302132615e-07,
      "loss": 0.428,
      "step": 434180
    },
    {
      "epoch": 177.73,
      "grad_norm": 2.338167190551758,
      "learning_rate": 1.5713422011427802e-07,
      "loss": 0.4433,
      "step": 434190
    },
    {
      "epoch": 177.73,
      "grad_norm": 2.1977832317352295,
      "learning_rate": 1.5707712003184305e-07,
      "loss": 0.4517,
      "step": 434200
    },
    {
      "epoch": 177.74,
      "grad_norm": 1.9937111139297485,
      "learning_rate": 1.5702002996621785e-07,
      "loss": 0.4703,
      "step": 434210
    },
    {
      "epoch": 177.74,
      "grad_norm": 2.468522548675537,
      "learning_rate": 1.5696294991766396e-07,
      "loss": 0.4609,
      "step": 434220
    },
    {
      "epoch": 177.74,
      "grad_norm": 2.0773675441741943,
      "learning_rate": 1.5690587988644285e-07,
      "loss": 0.4447,
      "step": 434230
    },
    {
      "epoch": 177.75,
      "grad_norm": 1.9548214673995972,
      "learning_rate": 1.5684881987281603e-07,
      "loss": 0.4299,
      "step": 434240
    },
    {
      "epoch": 177.75,
      "grad_norm": 1.9216830730438232,
      "learning_rate": 1.5679176987704478e-07,
      "loss": 0.4358,
      "step": 434250
    },
    {
      "epoch": 177.76,
      "grad_norm": 1.9555585384368896,
      "learning_rate": 1.567347298993906e-07,
      "loss": 0.4516,
      "step": 434260
    },
    {
      "epoch": 177.76,
      "grad_norm": 1.9419525861740112,
      "learning_rate": 1.5667769994011447e-07,
      "loss": 0.4447,
      "step": 434270
    },
    {
      "epoch": 177.77,
      "grad_norm": 2.0335302352905273,
      "learning_rate": 1.5662067999947792e-07,
      "loss": 0.4297,
      "step": 434280
    },
    {
      "epoch": 177.77,
      "grad_norm": 2.1047592163085938,
      "learning_rate": 1.5656367007774216e-07,
      "loss": 0.4242,
      "step": 434290
    },
    {
      "epoch": 177.77,
      "grad_norm": 2.394411325454712,
      "learning_rate": 1.5650667017516822e-07,
      "loss": 0.4312,
      "step": 434300
    },
    {
      "epoch": 177.78,
      "grad_norm": 2.2386415004730225,
      "learning_rate": 1.5644968029201705e-07,
      "loss": 0.4361,
      "step": 434310
    },
    {
      "epoch": 177.78,
      "grad_norm": 2.3772382736206055,
      "learning_rate": 1.563927004285502e-07,
      "loss": 0.4793,
      "step": 434320
    },
    {
      "epoch": 177.79,
      "grad_norm": 1.9808694124221802,
      "learning_rate": 1.563357305850281e-07,
      "loss": 0.4442,
      "step": 434330
    },
    {
      "epoch": 177.79,
      "grad_norm": 1.9206775426864624,
      "learning_rate": 1.5627877076171257e-07,
      "loss": 0.4421,
      "step": 434340
    },
    {
      "epoch": 177.79,
      "grad_norm": 2.550771951675415,
      "learning_rate": 1.5622182095886352e-07,
      "loss": 0.4571,
      "step": 434350
    },
    {
      "epoch": 177.8,
      "grad_norm": 2.107882261276245,
      "learning_rate": 1.561648811767422e-07,
      "loss": 0.4569,
      "step": 434360
    },
    {
      "epoch": 177.8,
      "grad_norm": 2.039252519607544,
      "learning_rate": 1.561079514156096e-07,
      "loss": 0.4417,
      "step": 434370
    },
    {
      "epoch": 177.81,
      "grad_norm": 2.258025646209717,
      "learning_rate": 1.560510316757262e-07,
      "loss": 0.44,
      "step": 434380
    },
    {
      "epoch": 177.81,
      "grad_norm": 2.7339134216308594,
      "learning_rate": 1.5599412195735325e-07,
      "loss": 0.444,
      "step": 434390
    },
    {
      "epoch": 177.81,
      "grad_norm": 3.5168561935424805,
      "learning_rate": 1.5593722226075093e-07,
      "loss": 0.4409,
      "step": 434400
    },
    {
      "epoch": 177.82,
      "grad_norm": 2.322761297225952,
      "learning_rate": 1.558803325861803e-07,
      "loss": 0.4776,
      "step": 434410
    },
    {
      "epoch": 177.82,
      "grad_norm": 1.9326772689819336,
      "learning_rate": 1.5582345293390198e-07,
      "loss": 0.4345,
      "step": 434420
    },
    {
      "epoch": 177.83,
      "grad_norm": 1.771489143371582,
      "learning_rate": 1.55766583304176e-07,
      "loss": 0.4355,
      "step": 434430
    },
    {
      "epoch": 177.83,
      "grad_norm": 2.2092947959899902,
      "learning_rate": 1.5570972369726334e-07,
      "loss": 0.4429,
      "step": 434440
    },
    {
      "epoch": 177.83,
      "grad_norm": 2.3159000873565674,
      "learning_rate": 1.5565287411342421e-07,
      "loss": 0.4576,
      "step": 434450
    },
    {
      "epoch": 177.84,
      "grad_norm": 2.0032601356506348,
      "learning_rate": 1.5559603455291931e-07,
      "loss": 0.4489,
      "step": 434460
    },
    {
      "epoch": 177.84,
      "grad_norm": 1.7682145833969116,
      "learning_rate": 1.5553920501600887e-07,
      "loss": 0.4408,
      "step": 434470
    },
    {
      "epoch": 177.85,
      "grad_norm": 2.0793051719665527,
      "learning_rate": 1.5548238550295335e-07,
      "loss": 0.4607,
      "step": 434480
    },
    {
      "epoch": 177.85,
      "grad_norm": 1.5183322429656982,
      "learning_rate": 1.554255760140127e-07,
      "loss": 0.4449,
      "step": 434490
    },
    {
      "epoch": 177.86,
      "grad_norm": 2.1617190837860107,
      "learning_rate": 1.5536877654944738e-07,
      "loss": 0.4388,
      "step": 434500
    },
    {
      "epoch": 177.86,
      "grad_norm": 2.1926534175872803,
      "learning_rate": 1.5531198710951738e-07,
      "loss": 0.4501,
      "step": 434510
    },
    {
      "epoch": 177.86,
      "grad_norm": 2.3558766841888428,
      "learning_rate": 1.5525520769448311e-07,
      "loss": 0.4491,
      "step": 434520
    },
    {
      "epoch": 177.87,
      "grad_norm": 1.9797004461288452,
      "learning_rate": 1.5519843830460428e-07,
      "loss": 0.4381,
      "step": 434530
    },
    {
      "epoch": 177.87,
      "grad_norm": 1.9040230512619019,
      "learning_rate": 1.5514167894014164e-07,
      "loss": 0.431,
      "step": 434540
    },
    {
      "epoch": 177.88,
      "grad_norm": 2.2878036499023438,
      "learning_rate": 1.5508492960135435e-07,
      "loss": 0.4273,
      "step": 434550
    },
    {
      "epoch": 177.88,
      "grad_norm": 2.036027431488037,
      "learning_rate": 1.550281902885031e-07,
      "loss": 0.4622,
      "step": 434560
    },
    {
      "epoch": 177.88,
      "grad_norm": 1.9535319805145264,
      "learning_rate": 1.5497146100184685e-07,
      "loss": 0.4414,
      "step": 434570
    },
    {
      "epoch": 177.89,
      "grad_norm": 2.2932190895080566,
      "learning_rate": 1.5491474174164682e-07,
      "loss": 0.4515,
      "step": 434580
    },
    {
      "epoch": 177.89,
      "grad_norm": 2.394569158554077,
      "learning_rate": 1.5485803250816194e-07,
      "loss": 0.437,
      "step": 434590
    },
    {
      "epoch": 177.9,
      "grad_norm": 2.011392593383789,
      "learning_rate": 1.5480133330165212e-07,
      "loss": 0.4344,
      "step": 434600
    },
    {
      "epoch": 177.9,
      "grad_norm": 1.8705601692199707,
      "learning_rate": 1.547446441223771e-07,
      "loss": 0.4375,
      "step": 434610
    },
    {
      "epoch": 177.9,
      "grad_norm": 2.0575294494628906,
      "learning_rate": 1.5468796497059705e-07,
      "loss": 0.4375,
      "step": 434620
    },
    {
      "epoch": 177.91,
      "grad_norm": 2.324951410293579,
      "learning_rate": 1.5463129584657088e-07,
      "loss": 0.4333,
      "step": 434630
    },
    {
      "epoch": 177.91,
      "grad_norm": 2.078700304031372,
      "learning_rate": 1.5457463675055854e-07,
      "loss": 0.437,
      "step": 434640
    },
    {
      "epoch": 177.92,
      "grad_norm": 2.1359450817108154,
      "learning_rate": 1.545179876828192e-07,
      "loss": 0.436,
      "step": 434650
    },
    {
      "epoch": 177.92,
      "grad_norm": 1.870753526687622,
      "learning_rate": 1.5446134864361284e-07,
      "loss": 0.4349,
      "step": 434660
    },
    {
      "epoch": 177.92,
      "grad_norm": 2.0011789798736572,
      "learning_rate": 1.5440471963319884e-07,
      "loss": 0.4363,
      "step": 434670
    },
    {
      "epoch": 177.93,
      "grad_norm": 1.7788710594177246,
      "learning_rate": 1.543481006518364e-07,
      "loss": 0.4593,
      "step": 434680
    },
    {
      "epoch": 177.93,
      "grad_norm": 2.0155818462371826,
      "learning_rate": 1.5429149169978492e-07,
      "loss": 0.4388,
      "step": 434690
    },
    {
      "epoch": 177.94,
      "grad_norm": 2.234225034713745,
      "learning_rate": 1.5423489277730385e-07,
      "loss": 0.4363,
      "step": 434700
    },
    {
      "epoch": 177.94,
      "grad_norm": 2.390470504760742,
      "learning_rate": 1.5417830388465236e-07,
      "loss": 0.4388,
      "step": 434710
    },
    {
      "epoch": 177.95,
      "grad_norm": 2.1899425983428955,
      "learning_rate": 1.5412172502208987e-07,
      "loss": 0.4431,
      "step": 434720
    },
    {
      "epoch": 177.95,
      "grad_norm": 2.009294033050537,
      "learning_rate": 1.5406515618987528e-07,
      "loss": 0.4589,
      "step": 434730
    },
    {
      "epoch": 177.95,
      "grad_norm": 1.7615795135498047,
      "learning_rate": 1.5400859738826776e-07,
      "loss": 0.4483,
      "step": 434740
    },
    {
      "epoch": 177.96,
      "grad_norm": 2.169914484024048,
      "learning_rate": 1.5395204861752647e-07,
      "loss": 0.4578,
      "step": 434750
    },
    {
      "epoch": 177.96,
      "grad_norm": 2.2902331352233887,
      "learning_rate": 1.5389550987791062e-07,
      "loss": 0.4386,
      "step": 434760
    },
    {
      "epoch": 177.97,
      "grad_norm": 2.1216061115264893,
      "learning_rate": 1.5383898116967906e-07,
      "loss": 0.4294,
      "step": 434770
    },
    {
      "epoch": 177.97,
      "grad_norm": 1.8225767612457275,
      "learning_rate": 1.5378246249309046e-07,
      "loss": 0.4277,
      "step": 434780
    },
    {
      "epoch": 177.97,
      "grad_norm": 2.2808685302734375,
      "learning_rate": 1.537259538484043e-07,
      "loss": 0.4544,
      "step": 434790
    },
    {
      "epoch": 177.98,
      "grad_norm": 2.200719118118286,
      "learning_rate": 1.5366945523587886e-07,
      "loss": 0.4423,
      "step": 434800
    },
    {
      "epoch": 177.98,
      "grad_norm": 2.036102294921875,
      "learning_rate": 1.536129666557734e-07,
      "loss": 0.4524,
      "step": 434810
    },
    {
      "epoch": 177.99,
      "grad_norm": 2.2847843170166016,
      "learning_rate": 1.5355648810834652e-07,
      "loss": 0.4703,
      "step": 434820
    },
    {
      "epoch": 177.99,
      "grad_norm": 2.2608401775360107,
      "learning_rate": 1.535000195938569e-07,
      "loss": 0.4517,
      "step": 434830
    },
    {
      "epoch": 177.99,
      "grad_norm": 2.3628270626068115,
      "learning_rate": 1.5344356111256316e-07,
      "loss": 0.4447,
      "step": 434840
    },
    {
      "epoch": 178.0,
      "grad_norm": 1.9737745523452759,
      "learning_rate": 1.533871126647242e-07,
      "loss": 0.4381,
      "step": 434850
    },
    {
      "epoch": 178.0,
      "eval_loss": 0.44433218240737915,
      "eval_runtime": 52.2406,
      "eval_samples_per_second": 66.021,
      "eval_steps_per_second": 8.269,
      "step": 434854
    },
    {
      "epoch": 178.0,
      "grad_norm": 2.059396505355835,
      "learning_rate": 1.5333067425059819e-07,
      "loss": 0.4489,
      "step": 434860
    },
    {
      "epoch": 178.01,
      "grad_norm": 1.6335546970367432,
      "learning_rate": 1.53274245870444e-07,
      "loss": 0.4436,
      "step": 434870
    },
    {
      "epoch": 178.01,
      "grad_norm": 2.181155204772949,
      "learning_rate": 1.5321782752451975e-07,
      "loss": 0.4335,
      "step": 434880
    },
    {
      "epoch": 178.01,
      "grad_norm": 2.0422556400299072,
      "learning_rate": 1.531614192130844e-07,
      "loss": 0.4321,
      "step": 434890
    },
    {
      "epoch": 178.02,
      "grad_norm": 1.9002926349639893,
      "learning_rate": 1.53105020936396e-07,
      "loss": 0.4273,
      "step": 434900
    },
    {
      "epoch": 178.02,
      "grad_norm": 1.8843156099319458,
      "learning_rate": 1.53048632694713e-07,
      "loss": 0.4371,
      "step": 434910
    },
    {
      "epoch": 178.03,
      "grad_norm": 2.0957438945770264,
      "learning_rate": 1.5299225448829376e-07,
      "loss": 0.4559,
      "step": 434920
    },
    {
      "epoch": 178.03,
      "grad_norm": 2.058506727218628,
      "learning_rate": 1.5293588631739695e-07,
      "loss": 0.4434,
      "step": 434930
    },
    {
      "epoch": 178.04,
      "grad_norm": 2.0221798419952393,
      "learning_rate": 1.5287952818227962e-07,
      "loss": 0.436,
      "step": 434940
    },
    {
      "epoch": 178.04,
      "grad_norm": 2.240020275115967,
      "learning_rate": 1.5282318008320046e-07,
      "loss": 0.4585,
      "step": 434950
    },
    {
      "epoch": 178.04,
      "grad_norm": 2.142230749130249,
      "learning_rate": 1.5276684202041806e-07,
      "loss": 0.4703,
      "step": 434960
    },
    {
      "epoch": 178.05,
      "grad_norm": 2.4291162490844727,
      "learning_rate": 1.5271051399419008e-07,
      "loss": 0.457,
      "step": 434970
    },
    {
      "epoch": 178.05,
      "grad_norm": 2.316227436065674,
      "learning_rate": 1.526541960047749e-07,
      "loss": 0.4356,
      "step": 434980
    },
    {
      "epoch": 178.06,
      "grad_norm": 2.0449109077453613,
      "learning_rate": 1.525978880524301e-07,
      "loss": 0.4277,
      "step": 434990
    },
    {
      "epoch": 178.06,
      "grad_norm": 2.0475871562957764,
      "learning_rate": 1.525415901374138e-07,
      "loss": 0.4282,
      "step": 435000
    },
    {
      "epoch": 178.06,
      "grad_norm": 2.3901865482330322,
      "learning_rate": 1.5248530225998391e-07,
      "loss": 0.4377,
      "step": 435010
    },
    {
      "epoch": 178.07,
      "grad_norm": 2.094890832901001,
      "learning_rate": 1.5242902442039826e-07,
      "loss": 0.4298,
      "step": 435020
    },
    {
      "epoch": 178.07,
      "grad_norm": 1.9769067764282227,
      "learning_rate": 1.5237275661891448e-07,
      "loss": 0.4424,
      "step": 435030
    },
    {
      "epoch": 178.08,
      "grad_norm": 2.2581403255462646,
      "learning_rate": 1.523164988557907e-07,
      "loss": 0.443,
      "step": 435040
    },
    {
      "epoch": 178.08,
      "grad_norm": 2.175837278366089,
      "learning_rate": 1.5226025113128423e-07,
      "loss": 0.4393,
      "step": 435050
    },
    {
      "epoch": 178.08,
      "grad_norm": 2.0582423210144043,
      "learning_rate": 1.5220401344565323e-07,
      "loss": 0.4525,
      "step": 435060
    },
    {
      "epoch": 178.09,
      "grad_norm": 2.2241477966308594,
      "learning_rate": 1.5214778579915476e-07,
      "loss": 0.4587,
      "step": 435070
    },
    {
      "epoch": 178.09,
      "grad_norm": 2.14874267578125,
      "learning_rate": 1.5209156819204673e-07,
      "loss": 0.4431,
      "step": 435080
    },
    {
      "epoch": 178.1,
      "grad_norm": 1.84551203250885,
      "learning_rate": 1.5203536062458646e-07,
      "loss": 0.4682,
      "step": 435090
    },
    {
      "epoch": 178.1,
      "grad_norm": 2.0056610107421875,
      "learning_rate": 1.5197916309703127e-07,
      "loss": 0.4488,
      "step": 435100
    },
    {
      "epoch": 178.1,
      "grad_norm": 2.0054593086242676,
      "learning_rate": 1.5192297560963906e-07,
      "loss": 0.4306,
      "step": 435110
    },
    {
      "epoch": 178.11,
      "grad_norm": 2.611661434173584,
      "learning_rate": 1.5186679816266717e-07,
      "loss": 0.4547,
      "step": 435120
    },
    {
      "epoch": 178.11,
      "grad_norm": 3.3524279594421387,
      "learning_rate": 1.5181063075637217e-07,
      "loss": 0.4498,
      "step": 435130
    },
    {
      "epoch": 178.12,
      "grad_norm": 2.1082069873809814,
      "learning_rate": 1.5175447339101245e-07,
      "loss": 0.4318,
      "step": 435140
    },
    {
      "epoch": 178.12,
      "grad_norm": 1.9155789613723755,
      "learning_rate": 1.5169832606684454e-07,
      "loss": 0.4416,
      "step": 435150
    },
    {
      "epoch": 178.13,
      "grad_norm": 2.690459728240967,
      "learning_rate": 1.5164218878412608e-07,
      "loss": 0.473,
      "step": 435160
    },
    {
      "epoch": 178.13,
      "grad_norm": 2.3355534076690674,
      "learning_rate": 1.515860615431139e-07,
      "loss": 0.4437,
      "step": 435170
    },
    {
      "epoch": 178.13,
      "grad_norm": 2.266284704208374,
      "learning_rate": 1.5152994434406533e-07,
      "loss": 0.4348,
      "step": 435180
    },
    {
      "epoch": 178.14,
      "grad_norm": 2.452040433883667,
      "learning_rate": 1.514738371872372e-07,
      "loss": 0.4389,
      "step": 435190
    },
    {
      "epoch": 178.14,
      "grad_norm": 2.028839349746704,
      "learning_rate": 1.5141774007288713e-07,
      "loss": 0.4483,
      "step": 435200
    },
    {
      "epoch": 178.15,
      "grad_norm": 2.639364004135132,
      "learning_rate": 1.5136165300127112e-07,
      "loss": 0.4469,
      "step": 435210
    },
    {
      "epoch": 178.15,
      "grad_norm": 1.993664026260376,
      "learning_rate": 1.5130557597264683e-07,
      "loss": 0.4369,
      "step": 435220
    },
    {
      "epoch": 178.15,
      "grad_norm": 2.3868966102600098,
      "learning_rate": 1.5124950898727082e-07,
      "loss": 0.438,
      "step": 435230
    },
    {
      "epoch": 178.16,
      "grad_norm": 2.7526423931121826,
      "learning_rate": 1.5119345204539988e-07,
      "loss": 0.4709,
      "step": 435240
    },
    {
      "epoch": 178.16,
      "grad_norm": 2.0048089027404785,
      "learning_rate": 1.5113740514729085e-07,
      "loss": 0.4373,
      "step": 435250
    },
    {
      "epoch": 178.17,
      "grad_norm": 2.5114495754241943,
      "learning_rate": 1.5108136829320084e-07,
      "loss": 0.4545,
      "step": 435260
    },
    {
      "epoch": 178.17,
      "grad_norm": 2.094024419784546,
      "learning_rate": 1.5102534148338612e-07,
      "loss": 0.4411,
      "step": 435270
    },
    {
      "epoch": 178.17,
      "grad_norm": 2.2457661628723145,
      "learning_rate": 1.5096932471810354e-07,
      "loss": 0.4288,
      "step": 435280
    },
    {
      "epoch": 178.18,
      "grad_norm": 1.871052622795105,
      "learning_rate": 1.509133179976094e-07,
      "loss": 0.4329,
      "step": 435290
    },
    {
      "epoch": 178.18,
      "grad_norm": 2.257763385772705,
      "learning_rate": 1.508573213221608e-07,
      "loss": 0.4318,
      "step": 435300
    },
    {
      "epoch": 178.19,
      "grad_norm": 1.9901931285858154,
      "learning_rate": 1.5080133469201374e-07,
      "loss": 0.441,
      "step": 435310
    },
    {
      "epoch": 178.19,
      "grad_norm": 2.3130075931549072,
      "learning_rate": 1.5074535810742458e-07,
      "loss": 0.4405,
      "step": 435320
    },
    {
      "epoch": 178.19,
      "grad_norm": 2.315422773361206,
      "learning_rate": 1.5068939156865088e-07,
      "loss": 0.4783,
      "step": 435330
    },
    {
      "epoch": 178.2,
      "grad_norm": 2.311823606491089,
      "learning_rate": 1.5063343507594766e-07,
      "loss": 0.4362,
      "step": 435340
    },
    {
      "epoch": 178.2,
      "grad_norm": 2.059272289276123,
      "learning_rate": 1.5057748862957178e-07,
      "loss": 0.4473,
      "step": 435350
    },
    {
      "epoch": 178.21,
      "grad_norm": 1.997409462928772,
      "learning_rate": 1.5052155222977976e-07,
      "loss": 0.4481,
      "step": 435360
    },
    {
      "epoch": 178.21,
      "grad_norm": 1.972618579864502,
      "learning_rate": 1.5046562587682742e-07,
      "loss": 0.4435,
      "step": 435370
    },
    {
      "epoch": 178.22,
      "grad_norm": 2.4144365787506104,
      "learning_rate": 1.5040970957097102e-07,
      "loss": 0.4711,
      "step": 435380
    },
    {
      "epoch": 178.22,
      "grad_norm": 2.156418561935425,
      "learning_rate": 1.503538033124669e-07,
      "loss": 0.4394,
      "step": 435390
    },
    {
      "epoch": 178.22,
      "grad_norm": 1.8193981647491455,
      "learning_rate": 1.5029790710157113e-07,
      "loss": 0.4483,
      "step": 435400
    },
    {
      "epoch": 178.23,
      "grad_norm": 1.9639801979064941,
      "learning_rate": 1.5024202093853969e-07,
      "loss": 0.4383,
      "step": 435410
    },
    {
      "epoch": 178.23,
      "grad_norm": 2.1166532039642334,
      "learning_rate": 1.5018614482362867e-07,
      "loss": 0.4403,
      "step": 435420
    },
    {
      "epoch": 178.24,
      "grad_norm": 1.8011986017227173,
      "learning_rate": 1.5013027875709384e-07,
      "loss": 0.4349,
      "step": 435430
    },
    {
      "epoch": 178.24,
      "grad_norm": 1.8654534816741943,
      "learning_rate": 1.5007442273919125e-07,
      "loss": 0.4354,
      "step": 435440
    },
    {
      "epoch": 178.24,
      "grad_norm": 2.339545249938965,
      "learning_rate": 1.5001857677017668e-07,
      "loss": 0.4348,
      "step": 435450
    },
    {
      "epoch": 178.25,
      "grad_norm": 1.9154375791549683,
      "learning_rate": 1.499627408503062e-07,
      "loss": 0.4341,
      "step": 435460
    },
    {
      "epoch": 178.25,
      "grad_norm": 1.9914487600326538,
      "learning_rate": 1.499069149798353e-07,
      "loss": 0.4418,
      "step": 435470
    },
    {
      "epoch": 178.26,
      "grad_norm": 1.9558380842208862,
      "learning_rate": 1.4985109915901977e-07,
      "loss": 0.4229,
      "step": 435480
    },
    {
      "epoch": 178.26,
      "grad_norm": 2.322197675704956,
      "learning_rate": 1.497952933881154e-07,
      "loss": 0.4536,
      "step": 435490
    },
    {
      "epoch": 178.26,
      "grad_norm": 2.0395727157592773,
      "learning_rate": 1.4973949766737773e-07,
      "loss": 0.4419,
      "step": 435500
    },
    {
      "epoch": 178.27,
      "grad_norm": 1.8486928939819336,
      "learning_rate": 1.4968371199706227e-07,
      "loss": 0.436,
      "step": 435510
    },
    {
      "epoch": 178.27,
      "grad_norm": 2.1688055992126465,
      "learning_rate": 1.496279363774248e-07,
      "loss": 0.4512,
      "step": 435520
    },
    {
      "epoch": 178.28,
      "grad_norm": 2.0207693576812744,
      "learning_rate": 1.495721708087206e-07,
      "loss": 0.4527,
      "step": 435530
    },
    {
      "epoch": 178.28,
      "grad_norm": 1.8551805019378662,
      "learning_rate": 1.4951641529120545e-07,
      "loss": 0.4436,
      "step": 435540
    },
    {
      "epoch": 178.28,
      "grad_norm": 2.057081699371338,
      "learning_rate": 1.4946066982513437e-07,
      "loss": 0.4418,
      "step": 435550
    },
    {
      "epoch": 178.29,
      "grad_norm": 2.101189136505127,
      "learning_rate": 1.4940493441076284e-07,
      "loss": 0.4533,
      "step": 435560
    },
    {
      "epoch": 178.29,
      "grad_norm": 2.109614610671997,
      "learning_rate": 1.4934920904834643e-07,
      "loss": 0.4308,
      "step": 435570
    },
    {
      "epoch": 178.3,
      "grad_norm": 1.719394326210022,
      "learning_rate": 1.4929349373813985e-07,
      "loss": 0.4497,
      "step": 435580
    },
    {
      "epoch": 178.3,
      "grad_norm": 1.7192466259002686,
      "learning_rate": 1.4923778848039892e-07,
      "loss": 0.426,
      "step": 435590
    },
    {
      "epoch": 178.31,
      "grad_norm": 2.2716574668884277,
      "learning_rate": 1.491820932753786e-07,
      "loss": 0.4285,
      "step": 435600
    },
    {
      "epoch": 178.31,
      "grad_norm": 1.6944067478179932,
      "learning_rate": 1.4912640812333393e-07,
      "loss": 0.4476,
      "step": 435610
    },
    {
      "epoch": 178.31,
      "grad_norm": 2.114288330078125,
      "learning_rate": 1.4907073302452016e-07,
      "loss": 0.4413,
      "step": 435620
    },
    {
      "epoch": 178.32,
      "grad_norm": 1.7622051239013672,
      "learning_rate": 1.4901506797919205e-07,
      "loss": 0.4546,
      "step": 435630
    },
    {
      "epoch": 178.32,
      "grad_norm": 1.7092878818511963,
      "learning_rate": 1.4895941298760512e-07,
      "loss": 0.437,
      "step": 435640
    },
    {
      "epoch": 178.33,
      "grad_norm": 1.9436310529708862,
      "learning_rate": 1.4890376805001356e-07,
      "loss": 0.4349,
      "step": 435650
    },
    {
      "epoch": 178.33,
      "grad_norm": 1.746925711631775,
      "learning_rate": 1.4884813316667294e-07,
      "loss": 0.4609,
      "step": 435660
    },
    {
      "epoch": 178.33,
      "grad_norm": 1.7815241813659668,
      "learning_rate": 1.487925083378377e-07,
      "loss": 0.4294,
      "step": 435670
    },
    {
      "epoch": 178.34,
      "grad_norm": 2.1922433376312256,
      "learning_rate": 1.4873689356376266e-07,
      "loss": 0.4444,
      "step": 435680
    },
    {
      "epoch": 178.34,
      "grad_norm": 2.042875289916992,
      "learning_rate": 1.48681288844703e-07,
      "loss": 0.4369,
      "step": 435690
    },
    {
      "epoch": 178.35,
      "grad_norm": 2.2167325019836426,
      "learning_rate": 1.4862569418091245e-07,
      "loss": 0.4391,
      "step": 435700
    },
    {
      "epoch": 178.35,
      "grad_norm": 2.2100327014923096,
      "learning_rate": 1.4857010957264706e-07,
      "loss": 0.4375,
      "step": 435710
    },
    {
      "epoch": 178.35,
      "grad_norm": 1.9012131690979004,
      "learning_rate": 1.4851453502016056e-07,
      "loss": 0.4202,
      "step": 435720
    },
    {
      "epoch": 178.36,
      "grad_norm": 2.215216875076294,
      "learning_rate": 1.4845897052370791e-07,
      "loss": 0.4272,
      "step": 435730
    },
    {
      "epoch": 178.36,
      "grad_norm": 1.8804599046707153,
      "learning_rate": 1.4840341608354337e-07,
      "loss": 0.4494,
      "step": 435740
    },
    {
      "epoch": 178.37,
      "grad_norm": 1.8119617700576782,
      "learning_rate": 1.4834787169992168e-07,
      "loss": 0.4262,
      "step": 435750
    },
    {
      "epoch": 178.37,
      "grad_norm": 2.1260783672332764,
      "learning_rate": 1.4829233737309702e-07,
      "loss": 0.4425,
      "step": 435760
    },
    {
      "epoch": 178.37,
      "grad_norm": 2.396368980407715,
      "learning_rate": 1.4823681310332393e-07,
      "loss": 0.4514,
      "step": 435770
    },
    {
      "epoch": 178.38,
      "grad_norm": 2.2438721656799316,
      "learning_rate": 1.481812988908571e-07,
      "loss": 0.4337,
      "step": 435780
    },
    {
      "epoch": 178.38,
      "grad_norm": 1.8846733570098877,
      "learning_rate": 1.4812579473595004e-07,
      "loss": 0.4592,
      "step": 435790
    },
    {
      "epoch": 178.39,
      "grad_norm": 2.1781370639801025,
      "learning_rate": 1.4807030063885743e-07,
      "loss": 0.4644,
      "step": 435800
    },
    {
      "epoch": 178.39,
      "grad_norm": 1.9422519207000732,
      "learning_rate": 1.4801481659983354e-07,
      "loss": 0.4473,
      "step": 435810
    },
    {
      "epoch": 178.4,
      "grad_norm": 1.9352493286132812,
      "learning_rate": 1.4795934261913254e-07,
      "loss": 0.4448,
      "step": 435820
    },
    {
      "epoch": 178.4,
      "grad_norm": 2.146404981613159,
      "learning_rate": 1.4790387869700846e-07,
      "loss": 0.4253,
      "step": 435830
    },
    {
      "epoch": 178.4,
      "grad_norm": 2.0616369247436523,
      "learning_rate": 1.4784842483371522e-07,
      "loss": 0.4485,
      "step": 435840
    },
    {
      "epoch": 178.41,
      "grad_norm": 2.493725299835205,
      "learning_rate": 1.4779298102950733e-07,
      "loss": 0.4453,
      "step": 435850
    },
    {
      "epoch": 178.41,
      "grad_norm": 2.0072104930877686,
      "learning_rate": 1.4773754728463822e-07,
      "loss": 0.4379,
      "step": 435860
    },
    {
      "epoch": 178.42,
      "grad_norm": 1.9585989713668823,
      "learning_rate": 1.4768212359936208e-07,
      "loss": 0.451,
      "step": 435870
    },
    {
      "epoch": 178.42,
      "grad_norm": 2.175558090209961,
      "learning_rate": 1.4762670997393268e-07,
      "loss": 0.4392,
      "step": 435880
    },
    {
      "epoch": 178.42,
      "grad_norm": 2.1380865573883057,
      "learning_rate": 1.4757130640860392e-07,
      "loss": 0.4335,
      "step": 435890
    },
    {
      "epoch": 178.43,
      "grad_norm": 2.297882556915283,
      "learning_rate": 1.4751591290362982e-07,
      "loss": 0.4284,
      "step": 435900
    },
    {
      "epoch": 178.43,
      "grad_norm": 2.2223682403564453,
      "learning_rate": 1.474605294592638e-07,
      "loss": 0.4205,
      "step": 435910
    },
    {
      "epoch": 178.44,
      "grad_norm": 1.9047707319259644,
      "learning_rate": 1.4740515607576032e-07,
      "loss": 0.4468,
      "step": 435920
    },
    {
      "epoch": 178.44,
      "grad_norm": 2.332343101501465,
      "learning_rate": 1.473497927533718e-07,
      "loss": 0.4465,
      "step": 435930
    },
    {
      "epoch": 178.44,
      "grad_norm": 2.219083070755005,
      "learning_rate": 1.4729443949235272e-07,
      "loss": 0.4495,
      "step": 435940
    },
    {
      "epoch": 178.45,
      "grad_norm": 2.235809803009033,
      "learning_rate": 1.472390962929563e-07,
      "loss": 0.442,
      "step": 435950
    },
    {
      "epoch": 178.45,
      "grad_norm": 2.446348190307617,
      "learning_rate": 1.471837631554359e-07,
      "loss": 0.4304,
      "step": 435960
    },
    {
      "epoch": 178.46,
      "grad_norm": 1.8568620681762695,
      "learning_rate": 1.471284400800456e-07,
      "loss": 0.4318,
      "step": 435970
    },
    {
      "epoch": 178.46,
      "grad_norm": 2.1500356197357178,
      "learning_rate": 1.4707312706703847e-07,
      "loss": 0.4514,
      "step": 435980
    },
    {
      "epoch": 178.47,
      "grad_norm": 2.4863104820251465,
      "learning_rate": 1.4701782411666775e-07,
      "loss": 0.4567,
      "step": 435990
    },
    {
      "epoch": 178.47,
      "grad_norm": 1.7502752542495728,
      "learning_rate": 1.469625312291869e-07,
      "loss": 0.4362,
      "step": 436000
    },
    {
      "epoch": 178.47,
      "grad_norm": 2.0125303268432617,
      "learning_rate": 1.469072484048493e-07,
      "loss": 0.4364,
      "step": 436010
    },
    {
      "epoch": 178.48,
      "grad_norm": 1.95015549659729,
      "learning_rate": 1.4685197564390819e-07,
      "loss": 0.4501,
      "step": 436020
    },
    {
      "epoch": 178.48,
      "grad_norm": 2.114575147628784,
      "learning_rate": 1.4679671294661647e-07,
      "loss": 0.4689,
      "step": 436030
    },
    {
      "epoch": 178.49,
      "grad_norm": 2.03932785987854,
      "learning_rate": 1.467414603132276e-07,
      "loss": 0.4554,
      "step": 436040
    },
    {
      "epoch": 178.49,
      "grad_norm": 1.7719167470932007,
      "learning_rate": 1.4668621774399474e-07,
      "loss": 0.4481,
      "step": 436050
    },
    {
      "epoch": 178.49,
      "grad_norm": 1.9356440305709839,
      "learning_rate": 1.4663098523917083e-07,
      "loss": 0.462,
      "step": 436060
    },
    {
      "epoch": 178.5,
      "grad_norm": 2.0884664058685303,
      "learning_rate": 1.4657576279900876e-07,
      "loss": 0.4443,
      "step": 436070
    },
    {
      "epoch": 178.5,
      "grad_norm": 2.391951560974121,
      "learning_rate": 1.4652055042376148e-07,
      "loss": 0.4383,
      "step": 436080
    },
    {
      "epoch": 178.51,
      "grad_norm": 2.069981575012207,
      "learning_rate": 1.464653481136822e-07,
      "loss": 0.4522,
      "step": 436090
    },
    {
      "epoch": 178.51,
      "grad_norm": 2.9575328826904297,
      "learning_rate": 1.4641015586902352e-07,
      "loss": 0.4454,
      "step": 436100
    },
    {
      "epoch": 178.51,
      "grad_norm": 2.493910551071167,
      "learning_rate": 1.463549736900384e-07,
      "loss": 0.4389,
      "step": 436110
    },
    {
      "epoch": 178.52,
      "grad_norm": 2.1179842948913574,
      "learning_rate": 1.4629980157697978e-07,
      "loss": 0.4478,
      "step": 436120
    },
    {
      "epoch": 178.52,
      "grad_norm": 2.6782424449920654,
      "learning_rate": 1.4624463953010002e-07,
      "loss": 0.4461,
      "step": 436130
    },
    {
      "epoch": 178.53,
      "grad_norm": 2.046245574951172,
      "learning_rate": 1.461894875496518e-07,
      "loss": 0.4684,
      "step": 436140
    },
    {
      "epoch": 178.53,
      "grad_norm": 1.9882806539535522,
      "learning_rate": 1.4613434563588832e-07,
      "loss": 0.4549,
      "step": 436150
    },
    {
      "epoch": 178.53,
      "grad_norm": 2.11942195892334,
      "learning_rate": 1.4607921378906143e-07,
      "loss": 0.4531,
      "step": 436160
    },
    {
      "epoch": 178.54,
      "grad_norm": 2.381437301635742,
      "learning_rate": 1.4602409200942435e-07,
      "loss": 0.459,
      "step": 436170
    },
    {
      "epoch": 178.54,
      "grad_norm": 2.389116048812866,
      "learning_rate": 1.459689802972292e-07,
      "loss": 0.4512,
      "step": 436180
    },
    {
      "epoch": 178.55,
      "grad_norm": 2.039318323135376,
      "learning_rate": 1.4591387865272864e-07,
      "loss": 0.4596,
      "step": 436190
    },
    {
      "epoch": 178.55,
      "grad_norm": 1.8654416799545288,
      "learning_rate": 1.4585878707617482e-07,
      "loss": 0.4279,
      "step": 436200
    },
    {
      "epoch": 178.56,
      "grad_norm": 1.9070409536361694,
      "learning_rate": 1.4580370556782038e-07,
      "loss": 0.4588,
      "step": 436210
    },
    {
      "epoch": 178.56,
      "grad_norm": 2.1189944744110107,
      "learning_rate": 1.4574863412791723e-07,
      "loss": 0.4558,
      "step": 436220
    },
    {
      "epoch": 178.56,
      "grad_norm": 2.118072271347046,
      "learning_rate": 1.4569357275671855e-07,
      "loss": 0.4705,
      "step": 436230
    },
    {
      "epoch": 178.57,
      "grad_norm": 1.7076138257980347,
      "learning_rate": 1.456385214544757e-07,
      "loss": 0.4532,
      "step": 436240
    },
    {
      "epoch": 178.57,
      "grad_norm": 2.0987207889556885,
      "learning_rate": 1.455834802214408e-07,
      "loss": 0.4351,
      "step": 436250
    },
    {
      "epoch": 178.58,
      "grad_norm": 2.2629032135009766,
      "learning_rate": 1.4552844905786656e-07,
      "loss": 0.4487,
      "step": 436260
    },
    {
      "epoch": 178.58,
      "grad_norm": 2.452005624771118,
      "learning_rate": 1.454734279640043e-07,
      "loss": 0.4223,
      "step": 436270
    },
    {
      "epoch": 178.58,
      "grad_norm": 1.8496270179748535,
      "learning_rate": 1.4541841694010693e-07,
      "loss": 0.4265,
      "step": 436280
    },
    {
      "epoch": 178.59,
      "grad_norm": 2.166501760482788,
      "learning_rate": 1.4536341598642611e-07,
      "loss": 0.4382,
      "step": 436290
    },
    {
      "epoch": 178.59,
      "grad_norm": 2.240676164627075,
      "learning_rate": 1.4530842510321398e-07,
      "loss": 0.4585,
      "step": 436300
    },
    {
      "epoch": 178.6,
      "grad_norm": 2.0380895137786865,
      "learning_rate": 1.4525344429072186e-07,
      "loss": 0.4328,
      "step": 436310
    },
    {
      "epoch": 178.6,
      "grad_norm": 2.0224976539611816,
      "learning_rate": 1.451984735492022e-07,
      "loss": 0.4403,
      "step": 436320
    },
    {
      "epoch": 178.6,
      "grad_norm": 2.0499913692474365,
      "learning_rate": 1.451435128789066e-07,
      "loss": 0.4541,
      "step": 436330
    },
    {
      "epoch": 178.61,
      "grad_norm": 1.8780282735824585,
      "learning_rate": 1.4508856228008696e-07,
      "loss": 0.4561,
      "step": 436340
    },
    {
      "epoch": 178.61,
      "grad_norm": 1.8485891819000244,
      "learning_rate": 1.450336217529946e-07,
      "loss": 0.4679,
      "step": 436350
    },
    {
      "epoch": 178.62,
      "grad_norm": 1.9589120149612427,
      "learning_rate": 1.44978691297882e-07,
      "loss": 0.4388,
      "step": 436360
    },
    {
      "epoch": 178.62,
      "grad_norm": 2.336141347885132,
      "learning_rate": 1.4492377091499963e-07,
      "loss": 0.4495,
      "step": 436370
    },
    {
      "epoch": 178.62,
      "grad_norm": 2.377962350845337,
      "learning_rate": 1.448688606046e-07,
      "loss": 0.4375,
      "step": 436380
    },
    {
      "epoch": 178.63,
      "grad_norm": 2.1540791988372803,
      "learning_rate": 1.4481396036693412e-07,
      "loss": 0.4449,
      "step": 436390
    },
    {
      "epoch": 178.63,
      "grad_norm": 1.980882167816162,
      "learning_rate": 1.4475907020225368e-07,
      "loss": 0.4403,
      "step": 436400
    },
    {
      "epoch": 178.64,
      "grad_norm": 1.8694202899932861,
      "learning_rate": 1.4470419011081025e-07,
      "loss": 0.4412,
      "step": 436410
    },
    {
      "epoch": 178.64,
      "grad_norm": 2.155280590057373,
      "learning_rate": 1.44649320092855e-07,
      "loss": 0.4637,
      "step": 436420
    },
    {
      "epoch": 178.65,
      "grad_norm": 1.8932024240493774,
      "learning_rate": 1.4459446014863946e-07,
      "loss": 0.4491,
      "step": 436430
    },
    {
      "epoch": 178.65,
      "grad_norm": 1.9862297773361206,
      "learning_rate": 1.4453961027841482e-07,
      "loss": 0.4328,
      "step": 436440
    },
    {
      "epoch": 178.65,
      "grad_norm": 2.0402228832244873,
      "learning_rate": 1.4448477048243211e-07,
      "loss": 0.4523,
      "step": 436450
    },
    {
      "epoch": 178.66,
      "grad_norm": 2.0436148643493652,
      "learning_rate": 1.44429940760943e-07,
      "loss": 0.4299,
      "step": 436460
    },
    {
      "epoch": 178.66,
      "grad_norm": 2.1101183891296387,
      "learning_rate": 1.4437512111419858e-07,
      "loss": 0.4603,
      "step": 436470
    },
    {
      "epoch": 178.67,
      "grad_norm": 2.2223305702209473,
      "learning_rate": 1.4432031154244993e-07,
      "loss": 0.4387,
      "step": 436480
    },
    {
      "epoch": 178.67,
      "grad_norm": 1.996338129043579,
      "learning_rate": 1.4426551204594792e-07,
      "loss": 0.4465,
      "step": 436490
    },
    {
      "epoch": 178.67,
      "grad_norm": 1.9524908065795898,
      "learning_rate": 1.4421072262494416e-07,
      "loss": 0.4401,
      "step": 436500
    },
    {
      "epoch": 178.68,
      "grad_norm": 2.0446715354919434,
      "learning_rate": 1.4415594327968897e-07,
      "loss": 0.436,
      "step": 436510
    },
    {
      "epoch": 178.68,
      "grad_norm": 2.528203010559082,
      "learning_rate": 1.4410117401043343e-07,
      "loss": 0.4387,
      "step": 436520
    },
    {
      "epoch": 178.69,
      "grad_norm": 1.972806692123413,
      "learning_rate": 1.4404641481742842e-07,
      "loss": 0.4332,
      "step": 436530
    },
    {
      "epoch": 178.69,
      "grad_norm": 1.5667451620101929,
      "learning_rate": 1.4399166570092475e-07,
      "loss": 0.4475,
      "step": 436540
    },
    {
      "epoch": 178.69,
      "grad_norm": 1.9036428928375244,
      "learning_rate": 1.4393692666117353e-07,
      "loss": 0.4447,
      "step": 436550
    },
    {
      "epoch": 178.7,
      "grad_norm": 2.3725290298461914,
      "learning_rate": 1.4388219769842538e-07,
      "loss": 0.4454,
      "step": 436560
    },
    {
      "epoch": 178.7,
      "grad_norm": 2.0357484817504883,
      "learning_rate": 1.4382747881293081e-07,
      "loss": 0.4379,
      "step": 436570
    },
    {
      "epoch": 178.71,
      "grad_norm": 4.522687911987305,
      "learning_rate": 1.4377277000494044e-07,
      "loss": 0.453,
      "step": 436580
    },
    {
      "epoch": 178.71,
      "grad_norm": 2.461383819580078,
      "learning_rate": 1.4371807127470537e-07,
      "loss": 0.4455,
      "step": 436590
    },
    {
      "epoch": 178.71,
      "grad_norm": 1.6812655925750732,
      "learning_rate": 1.4366338262247566e-07,
      "loss": 0.4639,
      "step": 436600
    },
    {
      "epoch": 178.72,
      "grad_norm": 2.055824041366577,
      "learning_rate": 1.4360870404850188e-07,
      "loss": 0.4478,
      "step": 436610
    },
    {
      "epoch": 178.72,
      "grad_norm": 2.1470258235931396,
      "learning_rate": 1.435540355530349e-07,
      "loss": 0.4542,
      "step": 436620
    },
    {
      "epoch": 178.73,
      "grad_norm": 1.7560362815856934,
      "learning_rate": 1.4349937713632474e-07,
      "loss": 0.4278,
      "step": 436630
    },
    {
      "epoch": 178.73,
      "grad_norm": 1.825199842453003,
      "learning_rate": 1.43444728798622e-07,
      "loss": 0.426,
      "step": 436640
    },
    {
      "epoch": 178.74,
      "grad_norm": 2.18459153175354,
      "learning_rate": 1.43390090540177e-07,
      "loss": 0.4395,
      "step": 436650
    },
    {
      "epoch": 178.74,
      "grad_norm": 1.9152556657791138,
      "learning_rate": 1.4333546236124006e-07,
      "loss": 0.447,
      "step": 436660
    },
    {
      "epoch": 178.74,
      "grad_norm": 1.9776309728622437,
      "learning_rate": 1.4328084426206126e-07,
      "loss": 0.4285,
      "step": 436670
    },
    {
      "epoch": 178.75,
      "grad_norm": 1.954211711883545,
      "learning_rate": 1.4322623624289064e-07,
      "loss": 0.4376,
      "step": 436680
    },
    {
      "epoch": 178.75,
      "grad_norm": 2.2453718185424805,
      "learning_rate": 1.4317163830397878e-07,
      "loss": 0.4377,
      "step": 436690
    },
    {
      "epoch": 178.76,
      "grad_norm": 1.8169251680374146,
      "learning_rate": 1.4311705044557572e-07,
      "loss": 0.4234,
      "step": 436700
    },
    {
      "epoch": 178.76,
      "grad_norm": 1.9936022758483887,
      "learning_rate": 1.4306247266793104e-07,
      "loss": 0.449,
      "step": 436710
    },
    {
      "epoch": 178.76,
      "grad_norm": 1.9467090368270874,
      "learning_rate": 1.4300790497129557e-07,
      "loss": 0.4398,
      "step": 436720
    },
    {
      "epoch": 178.77,
      "grad_norm": 1.9963772296905518,
      "learning_rate": 1.4295334735591833e-07,
      "loss": 0.454,
      "step": 436730
    },
    {
      "epoch": 178.77,
      "grad_norm": 2.3271992206573486,
      "learning_rate": 1.4289879982205013e-07,
      "loss": 0.4476,
      "step": 436740
    },
    {
      "epoch": 178.78,
      "grad_norm": 2.787182331085205,
      "learning_rate": 1.428442623699403e-07,
      "loss": 0.4728,
      "step": 436750
    },
    {
      "epoch": 178.78,
      "grad_norm": 2.1658847332000732,
      "learning_rate": 1.427897349998386e-07,
      "loss": 0.4713,
      "step": 436760
    },
    {
      "epoch": 178.78,
      "grad_norm": 1.780543565750122,
      "learning_rate": 1.4273521771199512e-07,
      "loss": 0.4286,
      "step": 436770
    },
    {
      "epoch": 178.79,
      "grad_norm": 2.1122395992279053,
      "learning_rate": 1.4268071050665963e-07,
      "loss": 0.4372,
      "step": 436780
    },
    {
      "epoch": 178.79,
      "grad_norm": 1.9445046186447144,
      "learning_rate": 1.4262621338408143e-07,
      "loss": 0.4464,
      "step": 436790
    },
    {
      "epoch": 178.8,
      "grad_norm": 1.9606322050094604,
      "learning_rate": 1.425717263445106e-07,
      "loss": 0.443,
      "step": 436800
    },
    {
      "epoch": 178.8,
      "grad_norm": 2.157958507537842,
      "learning_rate": 1.4251724938819688e-07,
      "loss": 0.4392,
      "step": 436810
    },
    {
      "epoch": 178.8,
      "grad_norm": 1.8758509159088135,
      "learning_rate": 1.4246278251538908e-07,
      "loss": 0.4477,
      "step": 436820
    },
    {
      "epoch": 178.81,
      "grad_norm": 2.2767069339752197,
      "learning_rate": 1.4240832572633697e-07,
      "loss": 0.4584,
      "step": 436830
    },
    {
      "epoch": 178.81,
      "grad_norm": 2.139314889907837,
      "learning_rate": 1.4235387902129014e-07,
      "loss": 0.4266,
      "step": 436840
    },
    {
      "epoch": 178.82,
      "grad_norm": 1.9829574823379517,
      "learning_rate": 1.4229944240049834e-07,
      "loss": 0.4332,
      "step": 436850
    },
    {
      "epoch": 178.82,
      "grad_norm": 1.8552789688110352,
      "learning_rate": 1.4224501586421031e-07,
      "loss": 0.4341,
      "step": 436860
    },
    {
      "epoch": 178.83,
      "grad_norm": 2.196338176727295,
      "learning_rate": 1.4219059941267594e-07,
      "loss": 0.4371,
      "step": 436870
    },
    {
      "epoch": 178.83,
      "grad_norm": 2.218005657196045,
      "learning_rate": 1.4213619304614417e-07,
      "loss": 0.456,
      "step": 436880
    },
    {
      "epoch": 178.83,
      "grad_norm": 7.506904602050781,
      "learning_rate": 1.4208179676486404e-07,
      "loss": 0.4412,
      "step": 436890
    },
    {
      "epoch": 178.84,
      "grad_norm": 1.9073171615600586,
      "learning_rate": 1.4202741056908534e-07,
      "loss": 0.4353,
      "step": 436900
    },
    {
      "epoch": 178.84,
      "grad_norm": 1.8937146663665771,
      "learning_rate": 1.4197303445905661e-07,
      "loss": 0.4548,
      "step": 436910
    },
    {
      "epoch": 178.85,
      "grad_norm": 2.4005329608917236,
      "learning_rate": 1.419186684350271e-07,
      "loss": 0.4275,
      "step": 436920
    },
    {
      "epoch": 178.85,
      "grad_norm": 2.0946390628814697,
      "learning_rate": 1.4186431249724609e-07,
      "loss": 0.448,
      "step": 436930
    },
    {
      "epoch": 178.85,
      "grad_norm": 2.0612223148345947,
      "learning_rate": 1.4180996664596234e-07,
      "loss": 0.4342,
      "step": 436940
    },
    {
      "epoch": 178.86,
      "grad_norm": 2.369407892227173,
      "learning_rate": 1.417556308814254e-07,
      "loss": 0.4375,
      "step": 436950
    },
    {
      "epoch": 178.86,
      "grad_norm": 2.3024215698242188,
      "learning_rate": 1.4170130520388298e-07,
      "loss": 0.4649,
      "step": 436960
    },
    {
      "epoch": 178.87,
      "grad_norm": 2.248394012451172,
      "learning_rate": 1.4164698961358485e-07,
      "loss": 0.4298,
      "step": 436970
    },
    {
      "epoch": 178.87,
      "grad_norm": 2.1214849948883057,
      "learning_rate": 1.4159268411077958e-07,
      "loss": 0.451,
      "step": 436980
    },
    {
      "epoch": 178.87,
      "grad_norm": 1.922329306602478,
      "learning_rate": 1.415383886957156e-07,
      "loss": 0.4294,
      "step": 436990
    },
    {
      "epoch": 178.88,
      "grad_norm": 2.024432897567749,
      "learning_rate": 1.4148410336864224e-07,
      "loss": 0.4324,
      "step": 437000
    },
    {
      "epoch": 178.88,
      "grad_norm": 1.887447714805603,
      "learning_rate": 1.4142982812980799e-07,
      "loss": 0.4169,
      "step": 437010
    },
    {
      "epoch": 178.89,
      "grad_norm": 2.3511507511138916,
      "learning_rate": 1.4137556297946079e-07,
      "loss": 0.4766,
      "step": 437020
    },
    {
      "epoch": 178.89,
      "grad_norm": 1.8820291757583618,
      "learning_rate": 1.4132130791785022e-07,
      "loss": 0.4617,
      "step": 437030
    },
    {
      "epoch": 178.89,
      "grad_norm": 2.2284345626831055,
      "learning_rate": 1.412670629452245e-07,
      "loss": 0.4488,
      "step": 437040
    },
    {
      "epoch": 178.9,
      "grad_norm": 2.156801700592041,
      "learning_rate": 1.4121282806183186e-07,
      "loss": 0.4534,
      "step": 437050
    },
    {
      "epoch": 178.9,
      "grad_norm": 2.605419635772705,
      "learning_rate": 1.4115860326792085e-07,
      "loss": 0.4716,
      "step": 437060
    },
    {
      "epoch": 178.91,
      "grad_norm": 1.8624022006988525,
      "learning_rate": 1.4110438856374018e-07,
      "loss": 0.4338,
      "step": 437070
    },
    {
      "epoch": 178.91,
      "grad_norm": 2.3092243671417236,
      "learning_rate": 1.4105018394953808e-07,
      "loss": 0.4367,
      "step": 437080
    },
    {
      "epoch": 178.92,
      "grad_norm": 2.1635913848876953,
      "learning_rate": 1.409959894255623e-07,
      "loss": 0.4487,
      "step": 437090
    },
    {
      "epoch": 178.92,
      "grad_norm": 2.339057683944702,
      "learning_rate": 1.409418049920616e-07,
      "loss": 0.4436,
      "step": 437100
    },
    {
      "epoch": 178.92,
      "grad_norm": 2.0049872398376465,
      "learning_rate": 1.408876306492842e-07,
      "loss": 0.4147,
      "step": 437110
    },
    {
      "epoch": 178.93,
      "grad_norm": 1.7435652017593384,
      "learning_rate": 1.408334663974781e-07,
      "loss": 0.4642,
      "step": 437120
    },
    {
      "epoch": 178.93,
      "grad_norm": 1.8445650339126587,
      "learning_rate": 1.4077931223689152e-07,
      "loss": 0.45,
      "step": 437130
    },
    {
      "epoch": 178.94,
      "grad_norm": 2.348571538925171,
      "learning_rate": 1.4072516816777246e-07,
      "loss": 0.4761,
      "step": 437140
    },
    {
      "epoch": 178.94,
      "grad_norm": 2.166679859161377,
      "learning_rate": 1.4067103419036888e-07,
      "loss": 0.4368,
      "step": 437150
    },
    {
      "epoch": 178.94,
      "grad_norm": 2.533402681350708,
      "learning_rate": 1.4061691030492875e-07,
      "loss": 0.426,
      "step": 437160
    },
    {
      "epoch": 178.95,
      "grad_norm": 2.0870423316955566,
      "learning_rate": 1.4056279651170032e-07,
      "loss": 0.4577,
      "step": 437170
    },
    {
      "epoch": 178.95,
      "grad_norm": 1.646227478981018,
      "learning_rate": 1.4050869281093132e-07,
      "loss": 0.4314,
      "step": 437180
    },
    {
      "epoch": 178.96,
      "grad_norm": 1.8228148221969604,
      "learning_rate": 1.4045459920286918e-07,
      "loss": 0.4506,
      "step": 437190
    },
    {
      "epoch": 178.96,
      "grad_norm": 2.0111029148101807,
      "learning_rate": 1.4040051568776243e-07,
      "loss": 0.4329,
      "step": 437200
    },
    {
      "epoch": 178.96,
      "grad_norm": 1.6693552732467651,
      "learning_rate": 1.4034644226585799e-07,
      "loss": 0.4447,
      "step": 437210
    },
    {
      "epoch": 178.97,
      "grad_norm": 1.9065850973129272,
      "learning_rate": 1.4029237893740463e-07,
      "loss": 0.4457,
      "step": 437220
    },
    {
      "epoch": 178.97,
      "grad_norm": 2.137728452682495,
      "learning_rate": 1.4023832570264904e-07,
      "loss": 0.4568,
      "step": 437230
    },
    {
      "epoch": 178.98,
      "grad_norm": 2.2150001525878906,
      "learning_rate": 1.4018428256183915e-07,
      "loss": 0.4617,
      "step": 437240
    },
    {
      "epoch": 178.98,
      "grad_norm": 2.4196126461029053,
      "learning_rate": 1.4013024951522248e-07,
      "loss": 0.4449,
      "step": 437250
    },
    {
      "epoch": 178.98,
      "grad_norm": 2.207451581954956,
      "learning_rate": 1.400762265630467e-07,
      "loss": 0.4499,
      "step": 437260
    },
    {
      "epoch": 178.99,
      "grad_norm": 2.1600146293640137,
      "learning_rate": 1.400222137055593e-07,
      "loss": 0.4265,
      "step": 437270
    },
    {
      "epoch": 178.99,
      "grad_norm": 1.913873314857483,
      "learning_rate": 1.3996821094300744e-07,
      "loss": 0.4622,
      "step": 437280
    },
    {
      "epoch": 179.0,
      "grad_norm": 1.9472802877426147,
      "learning_rate": 1.3991421827563888e-07,
      "loss": 0.4408,
      "step": 437290
    },
    {
      "epoch": 179.0,
      "eval_loss": 0.44490739703178406,
      "eval_runtime": 52.99,
      "eval_samples_per_second": 65.088,
      "eval_steps_per_second": 8.152,
      "step": 437297
    },
    {
      "epoch": 179.0,
      "grad_norm": 1.9991168975830078,
      "learning_rate": 1.3986023570370054e-07,
      "loss": 0.4356,
      "step": 437300
    },
    {
      "epoch": 179.01,
      "grad_norm": 2.1263790130615234,
      "learning_rate": 1.398062632274399e-07,
      "loss": 0.4527,
      "step": 437310
    },
    {
      "epoch": 179.01,
      "grad_norm": 1.8581472635269165,
      "learning_rate": 1.397523008471041e-07,
      "loss": 0.4312,
      "step": 437320
    },
    {
      "epoch": 179.01,
      "grad_norm": 2.038836717605591,
      "learning_rate": 1.3969834856294064e-07,
      "loss": 0.4208,
      "step": 437330
    },
    {
      "epoch": 179.02,
      "grad_norm": 2.0441057682037354,
      "learning_rate": 1.3964440637519647e-07,
      "loss": 0.4529,
      "step": 437340
    },
    {
      "epoch": 179.02,
      "grad_norm": 2.1144957542419434,
      "learning_rate": 1.395904742841185e-07,
      "loss": 0.433,
      "step": 437350
    },
    {
      "epoch": 179.03,
      "grad_norm": 2.108757495880127,
      "learning_rate": 1.3953655228995394e-07,
      "loss": 0.4309,
      "step": 437360
    },
    {
      "epoch": 179.03,
      "grad_norm": 2.2006447315216064,
      "learning_rate": 1.3948264039295e-07,
      "loss": 0.4533,
      "step": 437370
    },
    {
      "epoch": 179.03,
      "grad_norm": 2.1335830688476562,
      "learning_rate": 1.3942873859335333e-07,
      "loss": 0.4252,
      "step": 437380
    },
    {
      "epoch": 179.04,
      "grad_norm": 2.3432068824768066,
      "learning_rate": 1.3937484689141142e-07,
      "loss": 0.4517,
      "step": 437390
    },
    {
      "epoch": 179.04,
      "grad_norm": 2.218914747238159,
      "learning_rate": 1.393209652873699e-07,
      "loss": 0.4558,
      "step": 437400
    },
    {
      "epoch": 179.05,
      "grad_norm": 1.9847766160964966,
      "learning_rate": 1.3926709378147696e-07,
      "loss": 0.4596,
      "step": 437410
    },
    {
      "epoch": 179.05,
      "grad_norm": 2.079063892364502,
      "learning_rate": 1.3921323237397856e-07,
      "loss": 0.449,
      "step": 437420
    },
    {
      "epoch": 179.05,
      "grad_norm": 2.1323070526123047,
      "learning_rate": 1.3915938106512163e-07,
      "loss": 0.4442,
      "step": 437430
    },
    {
      "epoch": 179.06,
      "grad_norm": 2.3595378398895264,
      "learning_rate": 1.3910553985515305e-07,
      "loss": 0.4362,
      "step": 437440
    },
    {
      "epoch": 179.06,
      "grad_norm": 2.10186767578125,
      "learning_rate": 1.3905170874431932e-07,
      "loss": 0.4515,
      "step": 437450
    },
    {
      "epoch": 179.07,
      "grad_norm": 1.8501577377319336,
      "learning_rate": 1.3899788773286707e-07,
      "loss": 0.4618,
      "step": 437460
    },
    {
      "epoch": 179.07,
      "grad_norm": 2.0359981060028076,
      "learning_rate": 1.389440768210427e-07,
      "loss": 0.4281,
      "step": 437470
    },
    {
      "epoch": 179.07,
      "grad_norm": 1.7752805948257446,
      "learning_rate": 1.3889027600909268e-07,
      "loss": 0.4373,
      "step": 437480
    },
    {
      "epoch": 179.08,
      "grad_norm": 1.9426101446151733,
      "learning_rate": 1.388364852972639e-07,
      "loss": 0.4652,
      "step": 437490
    },
    {
      "epoch": 179.08,
      "grad_norm": 1.9430190324783325,
      "learning_rate": 1.3878270468580228e-07,
      "loss": 0.439,
      "step": 437500
    },
    {
      "epoch": 179.09,
      "grad_norm": 1.9337748289108276,
      "learning_rate": 1.3872893417495425e-07,
      "loss": 0.4447,
      "step": 437510
    },
    {
      "epoch": 179.09,
      "grad_norm": 2.1252126693725586,
      "learning_rate": 1.3867517376496647e-07,
      "loss": 0.4406,
      "step": 437520
    },
    {
      "epoch": 179.1,
      "grad_norm": 1.9220916032791138,
      "learning_rate": 1.3862142345608508e-07,
      "loss": 0.4363,
      "step": 437530
    },
    {
      "epoch": 179.1,
      "grad_norm": 1.8101989030838013,
      "learning_rate": 1.38567683248556e-07,
      "loss": 0.4351,
      "step": 437540
    },
    {
      "epoch": 179.1,
      "grad_norm": 2.130769729614258,
      "learning_rate": 1.385139531426254e-07,
      "loss": 0.457,
      "step": 437550
    },
    {
      "epoch": 179.11,
      "grad_norm": 2.1737256050109863,
      "learning_rate": 1.3846023313853995e-07,
      "loss": 0.4502,
      "step": 437560
    },
    {
      "epoch": 179.11,
      "grad_norm": 1.6048532724380493,
      "learning_rate": 1.38406523236545e-07,
      "loss": 0.4451,
      "step": 437570
    },
    {
      "epoch": 179.12,
      "grad_norm": 1.5394350290298462,
      "learning_rate": 1.3835282343688725e-07,
      "loss": 0.4451,
      "step": 437580
    },
    {
      "epoch": 179.12,
      "grad_norm": 2.244398832321167,
      "learning_rate": 1.3829913373981207e-07,
      "loss": 0.4574,
      "step": 437590
    },
    {
      "epoch": 179.12,
      "grad_norm": 2.3604941368103027,
      "learning_rate": 1.3824545414556588e-07,
      "loss": 0.4333,
      "step": 437600
    },
    {
      "epoch": 179.13,
      "grad_norm": 2.090479850769043,
      "learning_rate": 1.3819178465439458e-07,
      "loss": 0.4445,
      "step": 437610
    },
    {
      "epoch": 179.13,
      "grad_norm": 2.070552110671997,
      "learning_rate": 1.3813812526654406e-07,
      "loss": 0.4597,
      "step": 437620
    },
    {
      "epoch": 179.14,
      "grad_norm": 1.6633507013320923,
      "learning_rate": 1.380844759822597e-07,
      "loss": 0.4396,
      "step": 437630
    },
    {
      "epoch": 179.14,
      "grad_norm": 2.332977056503296,
      "learning_rate": 1.3803083680178766e-07,
      "loss": 0.4496,
      "step": 437640
    },
    {
      "epoch": 179.14,
      "grad_norm": 2.205339193344116,
      "learning_rate": 1.3797720772537358e-07,
      "loss": 0.4482,
      "step": 437650
    },
    {
      "epoch": 179.15,
      "grad_norm": 2.237678289413452,
      "learning_rate": 1.379235887532631e-07,
      "loss": 0.472,
      "step": 437660
    },
    {
      "epoch": 179.15,
      "grad_norm": 2.272379159927368,
      "learning_rate": 1.3786997988570183e-07,
      "loss": 0.4421,
      "step": 437670
    },
    {
      "epoch": 179.16,
      "grad_norm": 1.9892454147338867,
      "learning_rate": 1.3781638112293492e-07,
      "loss": 0.4325,
      "step": 437680
    },
    {
      "epoch": 179.16,
      "grad_norm": 2.2222554683685303,
      "learning_rate": 1.377627924652085e-07,
      "loss": 0.4561,
      "step": 437690
    },
    {
      "epoch": 179.16,
      "grad_norm": 2.2153613567352295,
      "learning_rate": 1.3770921391276796e-07,
      "loss": 0.431,
      "step": 437700
    },
    {
      "epoch": 179.17,
      "grad_norm": 1.8397443294525146,
      "learning_rate": 1.3765564546585841e-07,
      "loss": 0.4494,
      "step": 437710
    },
    {
      "epoch": 179.17,
      "grad_norm": 2.001661777496338,
      "learning_rate": 1.376020871247255e-07,
      "loss": 0.4461,
      "step": 437720
    },
    {
      "epoch": 179.18,
      "grad_norm": 1.8372570276260376,
      "learning_rate": 1.3754853888961462e-07,
      "loss": 0.4522,
      "step": 437730
    },
    {
      "epoch": 179.18,
      "grad_norm": 2.1921238899230957,
      "learning_rate": 1.3749500076077085e-07,
      "loss": 0.4499,
      "step": 437740
    },
    {
      "epoch": 179.19,
      "grad_norm": 1.8707674741744995,
      "learning_rate": 1.374414727384393e-07,
      "loss": 0.4336,
      "step": 437750
    },
    {
      "epoch": 179.19,
      "grad_norm": 2.2129523754119873,
      "learning_rate": 1.373879548228659e-07,
      "loss": 0.4246,
      "step": 437760
    },
    {
      "epoch": 179.19,
      "grad_norm": 1.921767234802246,
      "learning_rate": 1.3733444701429472e-07,
      "loss": 0.4408,
      "step": 437770
    },
    {
      "epoch": 179.2,
      "grad_norm": 1.990557312965393,
      "learning_rate": 1.3728094931297192e-07,
      "loss": 0.4536,
      "step": 437780
    },
    {
      "epoch": 179.2,
      "grad_norm": 2.0321552753448486,
      "learning_rate": 1.3722746171914208e-07,
      "loss": 0.4447,
      "step": 437790
    },
    {
      "epoch": 179.21,
      "grad_norm": 1.8010077476501465,
      "learning_rate": 1.3717398423305058e-07,
      "loss": 0.4454,
      "step": 437800
    },
    {
      "epoch": 179.21,
      "grad_norm": 2.0268356800079346,
      "learning_rate": 1.3712051685494177e-07,
      "loss": 0.4428,
      "step": 437810
    },
    {
      "epoch": 179.21,
      "grad_norm": 2.3310348987579346,
      "learning_rate": 1.3706705958506074e-07,
      "loss": 0.4158,
      "step": 437820
    },
    {
      "epoch": 179.22,
      "grad_norm": 2.144893169403076,
      "learning_rate": 1.3701361242365264e-07,
      "loss": 0.4492,
      "step": 437830
    },
    {
      "epoch": 179.22,
      "grad_norm": 1.8383382558822632,
      "learning_rate": 1.3696017537096232e-07,
      "loss": 0.4512,
      "step": 437840
    },
    {
      "epoch": 179.23,
      "grad_norm": 2.328108072280884,
      "learning_rate": 1.3690674842723408e-07,
      "loss": 0.4696,
      "step": 437850
    },
    {
      "epoch": 179.23,
      "grad_norm": 2.0283405780792236,
      "learning_rate": 1.3685333159271331e-07,
      "loss": 0.4445,
      "step": 437860
    },
    {
      "epoch": 179.23,
      "grad_norm": 2.5914103984832764,
      "learning_rate": 1.3679992486764438e-07,
      "loss": 0.4745,
      "step": 437870
    },
    {
      "epoch": 179.24,
      "grad_norm": 2.1581897735595703,
      "learning_rate": 1.3674652825227186e-07,
      "loss": 0.4647,
      "step": 437880
    },
    {
      "epoch": 179.24,
      "grad_norm": 1.9598064422607422,
      "learning_rate": 1.3669314174684036e-07,
      "loss": 0.4472,
      "step": 437890
    },
    {
      "epoch": 179.25,
      "grad_norm": 2.50803279876709,
      "learning_rate": 1.366397653515947e-07,
      "loss": 0.4685,
      "step": 437900
    },
    {
      "epoch": 179.25,
      "grad_norm": 2.802123785018921,
      "learning_rate": 1.36586399066779e-07,
      "loss": 0.4434,
      "step": 437910
    },
    {
      "epoch": 179.26,
      "grad_norm": 2.1226234436035156,
      "learning_rate": 1.365330428926381e-07,
      "loss": 0.4287,
      "step": 437920
    },
    {
      "epoch": 179.26,
      "grad_norm": 2.0467727184295654,
      "learning_rate": 1.3647969682941633e-07,
      "loss": 0.4352,
      "step": 437930
    },
    {
      "epoch": 179.26,
      "grad_norm": 2.0049781799316406,
      "learning_rate": 1.3642636087735775e-07,
      "loss": 0.4571,
      "step": 437940
    },
    {
      "epoch": 179.27,
      "grad_norm": 2.228522777557373,
      "learning_rate": 1.36373035036707e-07,
      "loss": 0.4261,
      "step": 437950
    },
    {
      "epoch": 179.27,
      "grad_norm": 1.891447901725769,
      "learning_rate": 1.3631971930770836e-07,
      "loss": 0.4571,
      "step": 437960
    },
    {
      "epoch": 179.28,
      "grad_norm": 2.1135950088500977,
      "learning_rate": 1.3626641369060597e-07,
      "loss": 0.4447,
      "step": 437970
    },
    {
      "epoch": 179.28,
      "grad_norm": 1.804428219795227,
      "learning_rate": 1.3621311818564385e-07,
      "loss": 0.4627,
      "step": 437980
    },
    {
      "epoch": 179.28,
      "grad_norm": 1.934467077255249,
      "learning_rate": 1.3615983279306638e-07,
      "loss": 0.4277,
      "step": 437990
    },
    {
      "epoch": 179.29,
      "grad_norm": 2.207909107208252,
      "learning_rate": 1.361065575131176e-07,
      "loss": 0.4444,
      "step": 438000
    },
    {
      "epoch": 179.29,
      "grad_norm": 3.8285937309265137,
      "learning_rate": 1.3605329234604158e-07,
      "loss": 0.4274,
      "step": 438010
    },
    {
      "epoch": 179.3,
      "grad_norm": 1.86668062210083,
      "learning_rate": 1.360000372920822e-07,
      "loss": 0.4718,
      "step": 438020
    },
    {
      "epoch": 179.3,
      "grad_norm": 1.883146047592163,
      "learning_rate": 1.3594679235148345e-07,
      "loss": 0.4348,
      "step": 438030
    },
    {
      "epoch": 179.3,
      "grad_norm": 1.60296630859375,
      "learning_rate": 1.3589355752448944e-07,
      "loss": 0.4462,
      "step": 438040
    },
    {
      "epoch": 179.31,
      "grad_norm": 2.0993874073028564,
      "learning_rate": 1.3584033281134376e-07,
      "loss": 0.4404,
      "step": 438050
    },
    {
      "epoch": 179.31,
      "grad_norm": 2.6001315116882324,
      "learning_rate": 1.3578711821229044e-07,
      "loss": 0.4322,
      "step": 438060
    },
    {
      "epoch": 179.32,
      "grad_norm": 2.039314031600952,
      "learning_rate": 1.3573391372757279e-07,
      "loss": 0.438,
      "step": 438070
    },
    {
      "epoch": 179.32,
      "grad_norm": 2.061392307281494,
      "learning_rate": 1.3568071935743517e-07,
      "loss": 0.4481,
      "step": 438080
    },
    {
      "epoch": 179.32,
      "grad_norm": 2.2541654109954834,
      "learning_rate": 1.3562753510212108e-07,
      "loss": 0.4519,
      "step": 438090
    },
    {
      "epoch": 179.33,
      "grad_norm": 2.1228790283203125,
      "learning_rate": 1.3557436096187386e-07,
      "loss": 0.4509,
      "step": 438100
    },
    {
      "epoch": 179.33,
      "grad_norm": 2.170469045639038,
      "learning_rate": 1.3552119693693756e-07,
      "loss": 0.4469,
      "step": 438110
    },
    {
      "epoch": 179.34,
      "grad_norm": 2.015873670578003,
      "learning_rate": 1.35468043027555e-07,
      "loss": 0.438,
      "step": 438120
    },
    {
      "epoch": 179.34,
      "grad_norm": 1.9484773874282837,
      "learning_rate": 1.354148992339702e-07,
      "loss": 0.43,
      "step": 438130
    },
    {
      "epoch": 179.35,
      "grad_norm": 1.951233148574829,
      "learning_rate": 1.3536176555642646e-07,
      "loss": 0.45,
      "step": 438140
    },
    {
      "epoch": 179.35,
      "grad_norm": 1.944974422454834,
      "learning_rate": 1.3530864199516687e-07,
      "loss": 0.4336,
      "step": 438150
    },
    {
      "epoch": 179.35,
      "grad_norm": 2.2007243633270264,
      "learning_rate": 1.3525552855043572e-07,
      "loss": 0.4437,
      "step": 438160
    },
    {
      "epoch": 179.36,
      "grad_norm": 1.958988070487976,
      "learning_rate": 1.352024252224755e-07,
      "loss": 0.4462,
      "step": 438170
    },
    {
      "epoch": 179.36,
      "grad_norm": 1.8808047771453857,
      "learning_rate": 1.3514933201152958e-07,
      "loss": 0.4492,
      "step": 438180
    },
    {
      "epoch": 179.37,
      "grad_norm": 1.788543462753296,
      "learning_rate": 1.3509624891784144e-07,
      "loss": 0.4409,
      "step": 438190
    },
    {
      "epoch": 179.37,
      "grad_norm": 2.0724525451660156,
      "learning_rate": 1.350431759416539e-07,
      "loss": 0.4394,
      "step": 438200
    },
    {
      "epoch": 179.37,
      "grad_norm": 1.8342602252960205,
      "learning_rate": 1.349901130832105e-07,
      "loss": 0.443,
      "step": 438210
    },
    {
      "epoch": 179.38,
      "grad_norm": 2.1814253330230713,
      "learning_rate": 1.3493706034275378e-07,
      "loss": 0.4635,
      "step": 438220
    },
    {
      "epoch": 179.38,
      "grad_norm": 2.0591514110565186,
      "learning_rate": 1.3488401772052724e-07,
      "loss": 0.4673,
      "step": 438230
    },
    {
      "epoch": 179.39,
      "grad_norm": 2.656709671020508,
      "learning_rate": 1.348309852167737e-07,
      "loss": 0.4294,
      "step": 438240
    },
    {
      "epoch": 179.39,
      "grad_norm": 1.8629655838012695,
      "learning_rate": 1.3477796283173644e-07,
      "loss": 0.4469,
      "step": 438250
    },
    {
      "epoch": 179.39,
      "grad_norm": 2.328409433364868,
      "learning_rate": 1.3472495056565746e-07,
      "loss": 0.4219,
      "step": 438260
    },
    {
      "epoch": 179.4,
      "grad_norm": 2.5005555152893066,
      "learning_rate": 1.3467194841878032e-07,
      "loss": 0.4588,
      "step": 438270
    },
    {
      "epoch": 179.4,
      "grad_norm": 2.129969596862793,
      "learning_rate": 1.346189563913475e-07,
      "loss": 0.4394,
      "step": 438280
    },
    {
      "epoch": 179.41,
      "grad_norm": 1.8887414932250977,
      "learning_rate": 1.3456597448360208e-07,
      "loss": 0.4329,
      "step": 438290
    },
    {
      "epoch": 179.41,
      "grad_norm": 2.014111280441284,
      "learning_rate": 1.3451300269578628e-07,
      "loss": 0.4256,
      "step": 438300
    },
    {
      "epoch": 179.41,
      "grad_norm": 2.619601011276245,
      "learning_rate": 1.3446004102814316e-07,
      "loss": 0.441,
      "step": 438310
    },
    {
      "epoch": 179.42,
      "grad_norm": 2.3145198822021484,
      "learning_rate": 1.3440708948091495e-07,
      "loss": 0.4518,
      "step": 438320
    },
    {
      "epoch": 179.42,
      "grad_norm": 2.210679531097412,
      "learning_rate": 1.343541480543447e-07,
      "loss": 0.4443,
      "step": 438330
    },
    {
      "epoch": 179.43,
      "grad_norm": 2.240997552871704,
      "learning_rate": 1.3430121674867442e-07,
      "loss": 0.4254,
      "step": 438340
    },
    {
      "epoch": 179.43,
      "grad_norm": 1.9576728343963623,
      "learning_rate": 1.3424829556414685e-07,
      "loss": 0.4466,
      "step": 438350
    },
    {
      "epoch": 179.44,
      "grad_norm": 1.6830699443817139,
      "learning_rate": 1.3419538450100478e-07,
      "loss": 0.4452,
      "step": 438360
    },
    {
      "epoch": 179.44,
      "grad_norm": 2.2066869735717773,
      "learning_rate": 1.3414248355948995e-07,
      "loss": 0.4386,
      "step": 438370
    },
    {
      "epoch": 179.44,
      "grad_norm": 2.1103997230529785,
      "learning_rate": 1.3408959273984538e-07,
      "loss": 0.445,
      "step": 438380
    },
    {
      "epoch": 179.45,
      "grad_norm": 2.116520643234253,
      "learning_rate": 1.3403671204231256e-07,
      "loss": 0.4479,
      "step": 438390
    },
    {
      "epoch": 179.45,
      "grad_norm": 2.4360523223876953,
      "learning_rate": 1.33983841467134e-07,
      "loss": 0.4364,
      "step": 438400
    },
    {
      "epoch": 179.46,
      "grad_norm": 2.4416041374206543,
      "learning_rate": 1.339309810145522e-07,
      "loss": 0.4463,
      "step": 438410
    },
    {
      "epoch": 179.46,
      "grad_norm": 2.045093059539795,
      "learning_rate": 1.3387813068480893e-07,
      "loss": 0.4513,
      "step": 438420
    },
    {
      "epoch": 179.46,
      "grad_norm": 1.708254337310791,
      "learning_rate": 1.3382529047814665e-07,
      "loss": 0.452,
      "step": 438430
    },
    {
      "epoch": 179.47,
      "grad_norm": 2.0783512592315674,
      "learning_rate": 1.3377246039480686e-07,
      "loss": 0.4502,
      "step": 438440
    },
    {
      "epoch": 179.47,
      "grad_norm": 1.9762648344039917,
      "learning_rate": 1.3371964043503236e-07,
      "loss": 0.4387,
      "step": 438450
    },
    {
      "epoch": 179.48,
      "grad_norm": 1.7652688026428223,
      "learning_rate": 1.3366683059906434e-07,
      "loss": 0.4265,
      "step": 438460
    },
    {
      "epoch": 179.48,
      "grad_norm": 1.8145339488983154,
      "learning_rate": 1.336140308871453e-07,
      "loss": 0.4556,
      "step": 438470
    },
    {
      "epoch": 179.48,
      "grad_norm": 2.2371809482574463,
      "learning_rate": 1.3356124129951648e-07,
      "loss": 0.4376,
      "step": 438480
    },
    {
      "epoch": 179.49,
      "grad_norm": 1.9077935218811035,
      "learning_rate": 1.3350846183642038e-07,
      "loss": 0.4454,
      "step": 438490
    },
    {
      "epoch": 179.49,
      "grad_norm": 2.394914150238037,
      "learning_rate": 1.3345569249809846e-07,
      "loss": 0.4434,
      "step": 438500
    },
    {
      "epoch": 179.5,
      "grad_norm": 1.9387603998184204,
      "learning_rate": 1.3340293328479224e-07,
      "loss": 0.4286,
      "step": 438510
    },
    {
      "epoch": 179.5,
      "grad_norm": 2.049386739730835,
      "learning_rate": 1.333501841967439e-07,
      "loss": 0.4425,
      "step": 438520
    },
    {
      "epoch": 179.5,
      "grad_norm": 2.044902801513672,
      "learning_rate": 1.3329744523419448e-07,
      "loss": 0.4635,
      "step": 438530
    },
    {
      "epoch": 179.51,
      "grad_norm": 2.0213701725006104,
      "learning_rate": 1.3324471639738617e-07,
      "loss": 0.4296,
      "step": 438540
    },
    {
      "epoch": 179.51,
      "grad_norm": 2.3742921352386475,
      "learning_rate": 1.3319199768655997e-07,
      "loss": 0.4287,
      "step": 438550
    },
    {
      "epoch": 179.52,
      "grad_norm": 2.336205005645752,
      "learning_rate": 1.331392891019578e-07,
      "loss": 0.4592,
      "step": 438560
    },
    {
      "epoch": 179.52,
      "grad_norm": 1.8777246475219727,
      "learning_rate": 1.3308659064382096e-07,
      "loss": 0.4451,
      "step": 438570
    },
    {
      "epoch": 179.53,
      "grad_norm": 1.716457724571228,
      "learning_rate": 1.3303390231239087e-07,
      "loss": 0.4419,
      "step": 438580
    },
    {
      "epoch": 179.53,
      "grad_norm": 2.156771659851074,
      "learning_rate": 1.3298122410790878e-07,
      "loss": 0.4364,
      "step": 438590
    },
    {
      "epoch": 179.53,
      "grad_norm": 1.9242101907730103,
      "learning_rate": 1.3292855603061584e-07,
      "loss": 0.44,
      "step": 438600
    },
    {
      "epoch": 179.54,
      "grad_norm": 1.8061755895614624,
      "learning_rate": 1.3287589808075385e-07,
      "loss": 0.4416,
      "step": 438610
    },
    {
      "epoch": 179.54,
      "grad_norm": 2.113638401031494,
      "learning_rate": 1.3282325025856375e-07,
      "loss": 0.4393,
      "step": 438620
    },
    {
      "epoch": 179.55,
      "grad_norm": 2.047698736190796,
      "learning_rate": 1.3277061256428645e-07,
      "loss": 0.4333,
      "step": 438630
    },
    {
      "epoch": 179.55,
      "grad_norm": 1.8447449207305908,
      "learning_rate": 1.3271798499816348e-07,
      "loss": 0.4465,
      "step": 438640
    },
    {
      "epoch": 179.55,
      "grad_norm": 2.3558948040008545,
      "learning_rate": 1.3266536756043576e-07,
      "loss": 0.4479,
      "step": 438650
    },
    {
      "epoch": 179.56,
      "grad_norm": 2.133322238922119,
      "learning_rate": 1.3261276025134424e-07,
      "loss": 0.4343,
      "step": 438660
    },
    {
      "epoch": 179.56,
      "grad_norm": 1.6617263555526733,
      "learning_rate": 1.325601630711299e-07,
      "loss": 0.4335,
      "step": 438670
    },
    {
      "epoch": 179.57,
      "grad_norm": 1.9216252565383911,
      "learning_rate": 1.3250757602003395e-07,
      "loss": 0.4322,
      "step": 438680
    },
    {
      "epoch": 179.57,
      "grad_norm": 1.6819634437561035,
      "learning_rate": 1.324549990982973e-07,
      "loss": 0.4368,
      "step": 438690
    },
    {
      "epoch": 179.57,
      "grad_norm": 1.99640691280365,
      "learning_rate": 1.3240243230616017e-07,
      "loss": 0.4511,
      "step": 438700
    },
    {
      "epoch": 179.58,
      "grad_norm": 2.2638981342315674,
      "learning_rate": 1.32349875643864e-07,
      "loss": 0.4395,
      "step": 438710
    },
    {
      "epoch": 179.58,
      "grad_norm": 1.834540843963623,
      "learning_rate": 1.3229732911164875e-07,
      "loss": 0.4501,
      "step": 438720
    },
    {
      "epoch": 179.59,
      "grad_norm": 2.705425500869751,
      "learning_rate": 1.322447927097561e-07,
      "loss": 0.4484,
      "step": 438730
    },
    {
      "epoch": 179.59,
      "grad_norm": 2.270336151123047,
      "learning_rate": 1.321922664384265e-07,
      "loss": 0.4417,
      "step": 438740
    },
    {
      "epoch": 179.59,
      "grad_norm": 2.5320253372192383,
      "learning_rate": 1.3213975029790015e-07,
      "loss": 0.4591,
      "step": 438750
    },
    {
      "epoch": 179.6,
      "grad_norm": 2.1819496154785156,
      "learning_rate": 1.3208724428841797e-07,
      "loss": 0.4635,
      "step": 438760
    },
    {
      "epoch": 179.6,
      "grad_norm": 2.125204086303711,
      "learning_rate": 1.3203474841022014e-07,
      "loss": 0.4395,
      "step": 438770
    },
    {
      "epoch": 179.61,
      "grad_norm": 2.382078170776367,
      "learning_rate": 1.319822626635476e-07,
      "loss": 0.4596,
      "step": 438780
    },
    {
      "epoch": 179.61,
      "grad_norm": 2.0299816131591797,
      "learning_rate": 1.3192978704864056e-07,
      "loss": 0.4536,
      "step": 438790
    },
    {
      "epoch": 179.62,
      "grad_norm": 2.569431781768799,
      "learning_rate": 1.3187732156573917e-07,
      "loss": 0.4528,
      "step": 438800
    },
    {
      "epoch": 179.62,
      "grad_norm": 2.1813807487487793,
      "learning_rate": 1.318248662150841e-07,
      "loss": 0.4414,
      "step": 438810
    },
    {
      "epoch": 179.62,
      "grad_norm": 2.094313144683838,
      "learning_rate": 1.3177242099691533e-07,
      "loss": 0.4204,
      "step": 438820
    },
    {
      "epoch": 179.63,
      "grad_norm": 2.3020706176757812,
      "learning_rate": 1.3171998591147399e-07,
      "loss": 0.4293,
      "step": 438830
    },
    {
      "epoch": 179.63,
      "grad_norm": 2.145719051361084,
      "learning_rate": 1.31667560958999e-07,
      "loss": 0.4271,
      "step": 438840
    },
    {
      "epoch": 179.64,
      "grad_norm": 2.4094414710998535,
      "learning_rate": 1.3161514613973106e-07,
      "loss": 0.4376,
      "step": 438850
    },
    {
      "epoch": 179.64,
      "grad_norm": 2.2733213901519775,
      "learning_rate": 1.3156274145391033e-07,
      "loss": 0.4546,
      "step": 438860
    },
    {
      "epoch": 179.64,
      "grad_norm": 2.2621264457702637,
      "learning_rate": 1.315103469017767e-07,
      "loss": 0.4445,
      "step": 438870
    },
    {
      "epoch": 179.65,
      "grad_norm": 2.0013113021850586,
      "learning_rate": 1.3145796248357037e-07,
      "loss": 0.4557,
      "step": 438880
    },
    {
      "epoch": 179.65,
      "grad_norm": 2.0530078411102295,
      "learning_rate": 1.3140558819953123e-07,
      "loss": 0.4468,
      "step": 438890
    },
    {
      "epoch": 179.66,
      "grad_norm": 1.7837488651275635,
      "learning_rate": 1.3135322404989947e-07,
      "loss": 0.4482,
      "step": 438900
    },
    {
      "epoch": 179.66,
      "grad_norm": 1.8595666885375977,
      "learning_rate": 1.3130087003491425e-07,
      "loss": 0.4457,
      "step": 438910
    },
    {
      "epoch": 179.66,
      "grad_norm": 2.0383219718933105,
      "learning_rate": 1.3124852615481593e-07,
      "loss": 0.4145,
      "step": 438920
    },
    {
      "epoch": 179.67,
      "grad_norm": 1.8651739358901978,
      "learning_rate": 1.3119619240984448e-07,
      "loss": 0.4381,
      "step": 438930
    },
    {
      "epoch": 179.67,
      "grad_norm": 2.0364789962768555,
      "learning_rate": 1.3114386880023927e-07,
      "loss": 0.4352,
      "step": 438940
    },
    {
      "epoch": 179.68,
      "grad_norm": 1.7487016916275024,
      "learning_rate": 1.310915553262402e-07,
      "loss": 0.4361,
      "step": 438950
    },
    {
      "epoch": 179.68,
      "grad_norm": 1.8235501050949097,
      "learning_rate": 1.3103925198808696e-07,
      "loss": 0.44,
      "step": 438960
    },
    {
      "epoch": 179.68,
      "grad_norm": 1.8612154722213745,
      "learning_rate": 1.3098695878601866e-07,
      "loss": 0.444,
      "step": 438970
    },
    {
      "epoch": 179.69,
      "grad_norm": 2.0068962574005127,
      "learning_rate": 1.309346757202752e-07,
      "loss": 0.4284,
      "step": 438980
    },
    {
      "epoch": 179.69,
      "grad_norm": 2.4998533725738525,
      "learning_rate": 1.3088240279109626e-07,
      "loss": 0.438,
      "step": 438990
    },
    {
      "epoch": 179.7,
      "grad_norm": 1.9979826211929321,
      "learning_rate": 1.3083013999872068e-07,
      "loss": 0.4318,
      "step": 439000
    },
    {
      "epoch": 179.7,
      "grad_norm": 2.3562238216400146,
      "learning_rate": 1.3077788734338866e-07,
      "loss": 0.4366,
      "step": 439010
    },
    {
      "epoch": 179.71,
      "grad_norm": 2.1913468837738037,
      "learning_rate": 1.307256448253388e-07,
      "loss": 0.4659,
      "step": 439020
    },
    {
      "epoch": 179.71,
      "grad_norm": 2.110090732574463,
      "learning_rate": 1.30673412444811e-07,
      "loss": 0.4575,
      "step": 439030
    },
    {
      "epoch": 179.71,
      "grad_norm": 2.1931378841400146,
      "learning_rate": 1.3062119020204416e-07,
      "loss": 0.4559,
      "step": 439040
    },
    {
      "epoch": 179.72,
      "grad_norm": 1.9715404510498047,
      "learning_rate": 1.305689780972779e-07,
      "loss": 0.42,
      "step": 439050
    },
    {
      "epoch": 179.72,
      "grad_norm": 2.1299266815185547,
      "learning_rate": 1.3051677613075085e-07,
      "loss": 0.4391,
      "step": 439060
    },
    {
      "epoch": 179.73,
      "grad_norm": 2.6655948162078857,
      "learning_rate": 1.3046458430270265e-07,
      "loss": 0.4347,
      "step": 439070
    },
    {
      "epoch": 179.73,
      "grad_norm": 1.8211755752563477,
      "learning_rate": 1.304124026133722e-07,
      "loss": 0.4693,
      "step": 439080
    },
    {
      "epoch": 179.73,
      "grad_norm": 1.7004956007003784,
      "learning_rate": 1.3036023106299808e-07,
      "loss": 0.4393,
      "step": 439090
    },
    {
      "epoch": 179.74,
      "grad_norm": 1.9652501344680786,
      "learning_rate": 1.3030806965182022e-07,
      "loss": 0.4623,
      "step": 439100
    },
    {
      "epoch": 179.74,
      "grad_norm": 1.9068603515625,
      "learning_rate": 1.3025591838007696e-07,
      "loss": 0.4446,
      "step": 439110
    },
    {
      "epoch": 179.75,
      "grad_norm": 2.1322696208953857,
      "learning_rate": 1.3020377724800717e-07,
      "loss": 0.4584,
      "step": 439120
    },
    {
      "epoch": 179.75,
      "grad_norm": 1.9765684604644775,
      "learning_rate": 1.3015164625584974e-07,
      "loss": 0.4547,
      "step": 439130
    },
    {
      "epoch": 179.75,
      "grad_norm": 1.902835488319397,
      "learning_rate": 1.300995254038438e-07,
      "loss": 0.4422,
      "step": 439140
    },
    {
      "epoch": 179.76,
      "grad_norm": 1.8772343397140503,
      "learning_rate": 1.3004741469222766e-07,
      "loss": 0.443,
      "step": 439150
    },
    {
      "epoch": 179.76,
      "grad_norm": 2.043679714202881,
      "learning_rate": 1.2999531412124024e-07,
      "loss": 0.4316,
      "step": 439160
    },
    {
      "epoch": 179.77,
      "grad_norm": 2.048640489578247,
      "learning_rate": 1.2994322369112012e-07,
      "loss": 0.4529,
      "step": 439170
    },
    {
      "epoch": 179.77,
      "grad_norm": 2.047182559967041,
      "learning_rate": 1.2989114340210618e-07,
      "loss": 0.426,
      "step": 439180
    },
    {
      "epoch": 179.77,
      "grad_norm": 2.0220682621002197,
      "learning_rate": 1.298390732544368e-07,
      "loss": 0.4336,
      "step": 439190
    },
    {
      "epoch": 179.78,
      "grad_norm": 2.4818670749664307,
      "learning_rate": 1.297870132483503e-07,
      "loss": 0.4543,
      "step": 439200
    },
    {
      "epoch": 179.78,
      "grad_norm": 5.6381144523620605,
      "learning_rate": 1.2973496338408552e-07,
      "loss": 0.4477,
      "step": 439210
    },
    {
      "epoch": 179.79,
      "grad_norm": 2.363987684249878,
      "learning_rate": 1.2968292366188088e-07,
      "loss": 0.439,
      "step": 439220
    },
    {
      "epoch": 179.79,
      "grad_norm": 2.2202305793762207,
      "learning_rate": 1.2963089408197444e-07,
      "loss": 0.4376,
      "step": 439230
    },
    {
      "epoch": 179.8,
      "grad_norm": 1.7966384887695312,
      "learning_rate": 1.2957887464460456e-07,
      "loss": 0.4487,
      "step": 439240
    },
    {
      "epoch": 179.8,
      "grad_norm": 1.7358945608139038,
      "learning_rate": 1.2952686535001009e-07,
      "loss": 0.449,
      "step": 439250
    },
    {
      "epoch": 179.8,
      "grad_norm": 1.8810513019561768,
      "learning_rate": 1.294748661984286e-07,
      "loss": 0.4481,
      "step": 439260
    },
    {
      "epoch": 179.81,
      "grad_norm": 1.9275404214859009,
      "learning_rate": 1.2942287719009898e-07,
      "loss": 0.4271,
      "step": 439270
    },
    {
      "epoch": 179.81,
      "grad_norm": 1.6906068325042725,
      "learning_rate": 1.2937089832525827e-07,
      "loss": 0.4263,
      "step": 439280
    },
    {
      "epoch": 179.82,
      "grad_norm": 2.251246690750122,
      "learning_rate": 1.2931892960414562e-07,
      "loss": 0.4579,
      "step": 439290
    },
    {
      "epoch": 179.82,
      "grad_norm": 2.140373945236206,
      "learning_rate": 1.2926697102699882e-07,
      "loss": 0.4633,
      "step": 439300
    },
    {
      "epoch": 179.82,
      "grad_norm": 2.3609371185302734,
      "learning_rate": 1.2921502259405574e-07,
      "loss": 0.4447,
      "step": 439310
    },
    {
      "epoch": 179.83,
      "grad_norm": 2.002366542816162,
      "learning_rate": 1.2916308430555447e-07,
      "loss": 0.4471,
      "step": 439320
    },
    {
      "epoch": 179.83,
      "grad_norm": 2.2654285430908203,
      "learning_rate": 1.2911115616173282e-07,
      "loss": 0.4573,
      "step": 439330
    },
    {
      "epoch": 179.84,
      "grad_norm": 2.2140161991119385,
      "learning_rate": 1.2905923816282888e-07,
      "loss": 0.4368,
      "step": 439340
    },
    {
      "epoch": 179.84,
      "grad_norm": 2.025522232055664,
      "learning_rate": 1.2900733030908025e-07,
      "loss": 0.448,
      "step": 439350
    },
    {
      "epoch": 179.84,
      "grad_norm": 2.393583297729492,
      "learning_rate": 1.28955432600725e-07,
      "loss": 0.4443,
      "step": 439360
    },
    {
      "epoch": 179.85,
      "grad_norm": 2.0159566402435303,
      "learning_rate": 1.289035450380004e-07,
      "loss": 0.4618,
      "step": 439370
    },
    {
      "epoch": 179.85,
      "grad_norm": 2.330430269241333,
      "learning_rate": 1.2885166762114435e-07,
      "loss": 0.4419,
      "step": 439380
    },
    {
      "epoch": 179.86,
      "grad_norm": 1.9020062685012817,
      "learning_rate": 1.2879980035039464e-07,
      "loss": 0.4499,
      "step": 439390
    },
    {
      "epoch": 179.86,
      "grad_norm": 1.985787272453308,
      "learning_rate": 1.2874794322598885e-07,
      "loss": 0.4543,
      "step": 439400
    },
    {
      "epoch": 179.86,
      "grad_norm": 3.214144706726074,
      "learning_rate": 1.2869609624816483e-07,
      "loss": 0.4663,
      "step": 439410
    },
    {
      "epoch": 179.87,
      "grad_norm": 2.024916648864746,
      "learning_rate": 1.2864425941715908e-07,
      "loss": 0.4497,
      "step": 439420
    },
    {
      "epoch": 179.87,
      "grad_norm": 2.0546257495880127,
      "learning_rate": 1.2859243273320995e-07,
      "loss": 0.458,
      "step": 439430
    },
    {
      "epoch": 179.88,
      "grad_norm": 2.031938314437866,
      "learning_rate": 1.2854061619655452e-07,
      "loss": 0.4463,
      "step": 439440
    },
    {
      "epoch": 179.88,
      "grad_norm": 2.415226459503174,
      "learning_rate": 1.2848880980743008e-07,
      "loss": 0.4511,
      "step": 439450
    },
    {
      "epoch": 179.89,
      "grad_norm": 2.3788740634918213,
      "learning_rate": 1.284370135660742e-07,
      "loss": 0.439,
      "step": 439460
    },
    {
      "epoch": 179.89,
      "grad_norm": 2.3377459049224854,
      "learning_rate": 1.2838522747272369e-07,
      "loss": 0.4427,
      "step": 439470
    },
    {
      "epoch": 179.89,
      "grad_norm": 2.2976810932159424,
      "learning_rate": 1.2833345152761633e-07,
      "loss": 0.4212,
      "step": 439480
    },
    {
      "epoch": 179.9,
      "grad_norm": 1.991737961769104,
      "learning_rate": 1.2828168573098923e-07,
      "loss": 0.443,
      "step": 439490
    },
    {
      "epoch": 179.9,
      "grad_norm": 2.4992361068725586,
      "learning_rate": 1.2822993008307912e-07,
      "loss": 0.4439,
      "step": 439500
    },
    {
      "epoch": 179.91,
      "grad_norm": 2.2655091285705566,
      "learning_rate": 1.2817818458412364e-07,
      "loss": 0.4514,
      "step": 439510
    },
    {
      "epoch": 179.91,
      "grad_norm": 2.0811164379119873,
      "learning_rate": 1.2812644923435928e-07,
      "loss": 0.4695,
      "step": 439520
    },
    {
      "epoch": 179.91,
      "grad_norm": 1.839556336402893,
      "learning_rate": 1.2807472403402332e-07,
      "loss": 0.4691,
      "step": 439530
    },
    {
      "epoch": 179.92,
      "grad_norm": 1.9347431659698486,
      "learning_rate": 1.2802300898335288e-07,
      "loss": 0.4641,
      "step": 439540
    },
    {
      "epoch": 179.92,
      "grad_norm": 2.3940021991729736,
      "learning_rate": 1.279713040825847e-07,
      "loss": 0.4255,
      "step": 439550
    },
    {
      "epoch": 179.93,
      "grad_norm": 2.0015921592712402,
      "learning_rate": 1.2791960933195532e-07,
      "loss": 0.4785,
      "step": 439560
    },
    {
      "epoch": 179.93,
      "grad_norm": 1.9875463247299194,
      "learning_rate": 1.278679247317018e-07,
      "loss": 0.4416,
      "step": 439570
    },
    {
      "epoch": 179.93,
      "grad_norm": 2.035717487335205,
      "learning_rate": 1.278162502820609e-07,
      "loss": 0.4388,
      "step": 439580
    },
    {
      "epoch": 179.94,
      "grad_norm": 1.8745797872543335,
      "learning_rate": 1.2776458598326944e-07,
      "loss": 0.4481,
      "step": 439590
    },
    {
      "epoch": 179.94,
      "grad_norm": 1.945651650428772,
      "learning_rate": 1.2771293183556395e-07,
      "loss": 0.442,
      "step": 439600
    },
    {
      "epoch": 179.95,
      "grad_norm": 1.9851799011230469,
      "learning_rate": 1.2766128783918095e-07,
      "loss": 0.4519,
      "step": 439610
    },
    {
      "epoch": 179.95,
      "grad_norm": 2.028441905975342,
      "learning_rate": 1.2760965399435723e-07,
      "loss": 0.4335,
      "step": 439620
    },
    {
      "epoch": 179.95,
      "grad_norm": 1.730022668838501,
      "learning_rate": 1.2755803030132907e-07,
      "loss": 0.4283,
      "step": 439630
    },
    {
      "epoch": 179.96,
      "grad_norm": 2.2491633892059326,
      "learning_rate": 1.2750641676033323e-07,
      "loss": 0.4618,
      "step": 439640
    },
    {
      "epoch": 179.96,
      "grad_norm": 2.23856258392334,
      "learning_rate": 1.2745481337160602e-07,
      "loss": 0.4362,
      "step": 439650
    },
    {
      "epoch": 179.97,
      "grad_norm": 2.2941696643829346,
      "learning_rate": 1.2740322013538342e-07,
      "loss": 0.4515,
      "step": 439660
    },
    {
      "epoch": 179.97,
      "grad_norm": 1.7728385925292969,
      "learning_rate": 1.2735163705190247e-07,
      "loss": 0.433,
      "step": 439670
    },
    {
      "epoch": 179.98,
      "grad_norm": 2.7473950386047363,
      "learning_rate": 1.2730006412139973e-07,
      "loss": 0.449,
      "step": 439680
    },
    {
      "epoch": 179.98,
      "grad_norm": 2.0939629077911377,
      "learning_rate": 1.2724850134411044e-07,
      "loss": 0.4696,
      "step": 439690
    },
    {
      "epoch": 179.98,
      "grad_norm": 1.860910177230835,
      "learning_rate": 1.2719694872027107e-07,
      "loss": 0.4406,
      "step": 439700
    },
    {
      "epoch": 179.99,
      "grad_norm": 1.6881301403045654,
      "learning_rate": 1.271454062501182e-07,
      "loss": 0.4337,
      "step": 439710
    },
    {
      "epoch": 179.99,
      "grad_norm": 2.3571436405181885,
      "learning_rate": 1.2709387393388758e-07,
      "loss": 0.4363,
      "step": 439720
    },
    {
      "epoch": 180.0,
      "grad_norm": 1.9127179384231567,
      "learning_rate": 1.2704235177181543e-07,
      "loss": 0.4355,
      "step": 439730
    },
    {
      "epoch": 180.0,
      "grad_norm": 2.3818461894989014,
      "learning_rate": 1.2699083976413757e-07,
      "loss": 0.4414,
      "step": 439740
    },
    {
      "epoch": 180.0,
      "eval_loss": 0.44480106234550476,
      "eval_runtime": 52.1618,
      "eval_samples_per_second": 66.121,
      "eval_steps_per_second": 8.282,
      "step": 439740
    },
    {
      "epoch": 180.0,
      "grad_norm": 1.8531932830810547,
      "learning_rate": 1.269393379110902e-07,
      "loss": 0.4284,
      "step": 439750
    },
    {
      "epoch": 180.01,
      "grad_norm": 2.3623979091644287,
      "learning_rate": 1.268878462129091e-07,
      "loss": 0.4403,
      "step": 439760
    },
    {
      "epoch": 180.01,
      "grad_norm": 2.2432525157928467,
      "learning_rate": 1.268363646698303e-07,
      "loss": 0.438,
      "step": 439770
    },
    {
      "epoch": 180.02,
      "grad_norm": 2.127708673477173,
      "learning_rate": 1.2678489328208954e-07,
      "loss": 0.4483,
      "step": 439780
    },
    {
      "epoch": 180.02,
      "grad_norm": 2.1705729961395264,
      "learning_rate": 1.2673343204992254e-07,
      "loss": 0.4416,
      "step": 439790
    },
    {
      "epoch": 180.02,
      "grad_norm": 2.0365638732910156,
      "learning_rate": 1.2668198097356506e-07,
      "loss": 0.4504,
      "step": 439800
    },
    {
      "epoch": 180.03,
      "grad_norm": 1.94777250289917,
      "learning_rate": 1.2663054005325287e-07,
      "loss": 0.4424,
      "step": 439810
    },
    {
      "epoch": 180.03,
      "grad_norm": 2.6203112602233887,
      "learning_rate": 1.2657910928922169e-07,
      "loss": 0.4406,
      "step": 439820
    },
    {
      "epoch": 180.04,
      "grad_norm": 1.735758662223816,
      "learning_rate": 1.2652768868170677e-07,
      "loss": 0.4399,
      "step": 439830
    },
    {
      "epoch": 180.04,
      "grad_norm": 1.6174511909484863,
      "learning_rate": 1.2647627823094412e-07,
      "loss": 0.4436,
      "step": 439840
    },
    {
      "epoch": 180.05,
      "grad_norm": 2.238964080810547,
      "learning_rate": 1.264248779371687e-07,
      "loss": 0.4551,
      "step": 439850
    },
    {
      "epoch": 180.05,
      "grad_norm": 1.8918849229812622,
      "learning_rate": 1.263734878006165e-07,
      "loss": 0.4579,
      "step": 439860
    },
    {
      "epoch": 180.05,
      "grad_norm": 1.94590425491333,
      "learning_rate": 1.2632210782152251e-07,
      "loss": 0.447,
      "step": 439870
    },
    {
      "epoch": 180.06,
      "grad_norm": 2.194748640060425,
      "learning_rate": 1.2627073800012248e-07,
      "loss": 0.4272,
      "step": 439880
    },
    {
      "epoch": 180.06,
      "grad_norm": 1.7796741724014282,
      "learning_rate": 1.262193783366514e-07,
      "loss": 0.4317,
      "step": 439890
    },
    {
      "epoch": 180.07,
      "grad_norm": 1.952598214149475,
      "learning_rate": 1.261680288313447e-07,
      "loss": 0.4525,
      "step": 439900
    },
    {
      "epoch": 180.07,
      "grad_norm": 2.1470344066619873,
      "learning_rate": 1.2611668948443765e-07,
      "loss": 0.4362,
      "step": 439910
    },
    {
      "epoch": 180.07,
      "grad_norm": 2.332406759262085,
      "learning_rate": 1.2606536029616522e-07,
      "loss": 0.4528,
      "step": 439920
    },
    {
      "epoch": 180.08,
      "grad_norm": 2.606443405151367,
      "learning_rate": 1.2601404126676265e-07,
      "loss": 0.4535,
      "step": 439930
    },
    {
      "epoch": 180.08,
      "grad_norm": 2.0013771057128906,
      "learning_rate": 1.259627323964651e-07,
      "loss": 0.4255,
      "step": 439940
    },
    {
      "epoch": 180.09,
      "grad_norm": 1.8647325038909912,
      "learning_rate": 1.2591143368550763e-07,
      "loss": 0.45,
      "step": 439950
    },
    {
      "epoch": 180.09,
      "grad_norm": 2.003410577774048,
      "learning_rate": 1.2586014513412514e-07,
      "loss": 0.4448,
      "step": 439960
    },
    {
      "epoch": 180.09,
      "grad_norm": 1.939465880393982,
      "learning_rate": 1.2580886674255263e-07,
      "loss": 0.4473,
      "step": 439970
    },
    {
      "epoch": 180.1,
      "grad_norm": 1.7571932077407837,
      "learning_rate": 1.257575985110248e-07,
      "loss": 0.442,
      "step": 439980
    },
    {
      "epoch": 180.1,
      "grad_norm": 2.1347362995147705,
      "learning_rate": 1.2570634043977714e-07,
      "loss": 0.4531,
      "step": 439990
    },
    {
      "epoch": 180.11,
      "grad_norm": 2.465731620788574,
      "learning_rate": 1.2565509252904356e-07,
      "loss": 0.4749,
      "step": 440000
    },
    {
      "epoch": 180.11,
      "grad_norm": 1.904954433441162,
      "learning_rate": 1.2560385477905934e-07,
      "loss": 0.422,
      "step": 440010
    },
    {
      "epoch": 180.11,
      "grad_norm": 1.9512279033660889,
      "learning_rate": 1.2555262719005885e-07,
      "loss": 0.446,
      "step": 440020
    },
    {
      "epoch": 180.12,
      "grad_norm": 2.471468925476074,
      "learning_rate": 1.2550140976227738e-07,
      "loss": 0.4426,
      "step": 440030
    },
    {
      "epoch": 180.12,
      "grad_norm": 2.04024338722229,
      "learning_rate": 1.254502024959486e-07,
      "loss": 0.4406,
      "step": 440040
    },
    {
      "epoch": 180.13,
      "grad_norm": 1.8915027379989624,
      "learning_rate": 1.2539900539130795e-07,
      "loss": 0.4488,
      "step": 440050
    },
    {
      "epoch": 180.13,
      "grad_norm": 2.3872172832489014,
      "learning_rate": 1.2534781844858967e-07,
      "loss": 0.428,
      "step": 440060
    },
    {
      "epoch": 180.14,
      "grad_norm": 1.914056420326233,
      "learning_rate": 1.2529664166802816e-07,
      "loss": 0.4323,
      "step": 440070
    },
    {
      "epoch": 180.14,
      "grad_norm": 2.1892781257629395,
      "learning_rate": 1.2524547504985817e-07,
      "loss": 0.4531,
      "step": 440080
    },
    {
      "epoch": 180.14,
      "grad_norm": 2.10263729095459,
      "learning_rate": 1.251943185943136e-07,
      "loss": 0.4655,
      "step": 440090
    },
    {
      "epoch": 180.15,
      "grad_norm": 2.300952672958374,
      "learning_rate": 1.251431723016289e-07,
      "loss": 0.4365,
      "step": 440100
    },
    {
      "epoch": 180.15,
      "grad_norm": 2.0752487182617188,
      "learning_rate": 1.250920361720388e-07,
      "loss": 0.4599,
      "step": 440110
    },
    {
      "epoch": 180.16,
      "grad_norm": 2.2420949935913086,
      "learning_rate": 1.2504091020577692e-07,
      "loss": 0.4487,
      "step": 440120
    },
    {
      "epoch": 180.16,
      "grad_norm": 2.038452386856079,
      "learning_rate": 1.2498979440307829e-07,
      "loss": 0.4168,
      "step": 440130
    },
    {
      "epoch": 180.16,
      "grad_norm": 2.1062915325164795,
      "learning_rate": 1.2493868876417627e-07,
      "loss": 0.4148,
      "step": 440140
    },
    {
      "epoch": 180.17,
      "grad_norm": 2.17982816696167,
      "learning_rate": 1.248875932893051e-07,
      "loss": 0.4521,
      "step": 440150
    },
    {
      "epoch": 180.17,
      "grad_norm": 1.5347423553466797,
      "learning_rate": 1.2483650797869914e-07,
      "loss": 0.4464,
      "step": 440160
    },
    {
      "epoch": 180.18,
      "grad_norm": 2.2534689903259277,
      "learning_rate": 1.2478543283259214e-07,
      "loss": 0.4443,
      "step": 440170
    },
    {
      "epoch": 180.18,
      "grad_norm": 7.13697624206543,
      "learning_rate": 1.2473436785121826e-07,
      "loss": 0.4471,
      "step": 440180
    },
    {
      "epoch": 180.18,
      "grad_norm": 1.8812066316604614,
      "learning_rate": 1.2468331303481118e-07,
      "loss": 0.4319,
      "step": 440190
    },
    {
      "epoch": 180.19,
      "grad_norm": 2.002145528793335,
      "learning_rate": 1.2463226838360504e-07,
      "loss": 0.4549,
      "step": 440200
    },
    {
      "epoch": 180.19,
      "grad_norm": 2.2096290588378906,
      "learning_rate": 1.2458123389783356e-07,
      "loss": 0.436,
      "step": 440210
    },
    {
      "epoch": 180.2,
      "grad_norm": 1.8450435400009155,
      "learning_rate": 1.2453020957773037e-07,
      "loss": 0.4418,
      "step": 440220
    },
    {
      "epoch": 180.2,
      "grad_norm": 2.753965139389038,
      "learning_rate": 1.2447919542352916e-07,
      "loss": 0.4652,
      "step": 440230
    },
    {
      "epoch": 180.2,
      "grad_norm": 2.268913984298706,
      "learning_rate": 1.2442819143546412e-07,
      "loss": 0.4384,
      "step": 440240
    },
    {
      "epoch": 180.21,
      "grad_norm": 2.081188678741455,
      "learning_rate": 1.2437719761376865e-07,
      "loss": 0.4459,
      "step": 440250
    },
    {
      "epoch": 180.21,
      "grad_norm": 2.0575101375579834,
      "learning_rate": 1.2432621395867615e-07,
      "loss": 0.4516,
      "step": 440260
    },
    {
      "epoch": 180.22,
      "grad_norm": 2.3530895709991455,
      "learning_rate": 1.2427524047042054e-07,
      "loss": 0.4402,
      "step": 440270
    },
    {
      "epoch": 180.22,
      "grad_norm": 2.122411012649536,
      "learning_rate": 1.2422427714923502e-07,
      "loss": 0.4322,
      "step": 440280
    },
    {
      "epoch": 180.23,
      "grad_norm": 2.2131576538085938,
      "learning_rate": 1.2417332399535293e-07,
      "loss": 0.4345,
      "step": 440290
    },
    {
      "epoch": 180.23,
      "grad_norm": 1.8897684812545776,
      "learning_rate": 1.241223810090077e-07,
      "loss": 0.4268,
      "step": 440300
    },
    {
      "epoch": 180.23,
      "grad_norm": 2.0127429962158203,
      "learning_rate": 1.24071448190433e-07,
      "loss": 0.4472,
      "step": 440310
    },
    {
      "epoch": 180.24,
      "grad_norm": 1.9665379524230957,
      "learning_rate": 1.24020525539862e-07,
      "loss": 0.4384,
      "step": 440320
    },
    {
      "epoch": 180.24,
      "grad_norm": 2.0644302368164062,
      "learning_rate": 1.2396961305752806e-07,
      "loss": 0.4583,
      "step": 440330
    },
    {
      "epoch": 180.25,
      "grad_norm": 2.019785165786743,
      "learning_rate": 1.239187107436641e-07,
      "loss": 0.4528,
      "step": 440340
    },
    {
      "epoch": 180.25,
      "grad_norm": 2.342341184616089,
      "learning_rate": 1.2386781859850354e-07,
      "loss": 0.4444,
      "step": 440350
    },
    {
      "epoch": 180.25,
      "grad_norm": 2.2307093143463135,
      "learning_rate": 1.238169366222795e-07,
      "loss": 0.4325,
      "step": 440360
    },
    {
      "epoch": 180.26,
      "grad_norm": 1.8799903392791748,
      "learning_rate": 1.237660648152251e-07,
      "loss": 0.4332,
      "step": 440370
    },
    {
      "epoch": 180.26,
      "grad_norm": 2.155932903289795,
      "learning_rate": 1.2371520317757303e-07,
      "loss": 0.4477,
      "step": 440380
    },
    {
      "epoch": 180.27,
      "grad_norm": 2.2366206645965576,
      "learning_rate": 1.2366435170955692e-07,
      "loss": 0.4398,
      "step": 440390
    },
    {
      "epoch": 180.27,
      "grad_norm": 1.9843939542770386,
      "learning_rate": 1.2361351041140915e-07,
      "loss": 0.437,
      "step": 440400
    },
    {
      "epoch": 180.27,
      "grad_norm": 2.289137601852417,
      "learning_rate": 1.235626792833626e-07,
      "loss": 0.4537,
      "step": 440410
    },
    {
      "epoch": 180.28,
      "grad_norm": 1.8628147840499878,
      "learning_rate": 1.2351185832565067e-07,
      "loss": 0.4501,
      "step": 440420
    },
    {
      "epoch": 180.28,
      "grad_norm": 2.244730234146118,
      "learning_rate": 1.2346104753850573e-07,
      "loss": 0.4542,
      "step": 440430
    },
    {
      "epoch": 180.29,
      "grad_norm": 1.9046505689620972,
      "learning_rate": 1.234102469221604e-07,
      "loss": 0.4432,
      "step": 440440
    },
    {
      "epoch": 180.29,
      "grad_norm": 1.88018000125885,
      "learning_rate": 1.2335945647684786e-07,
      "loss": 0.4713,
      "step": 440450
    },
    {
      "epoch": 180.29,
      "grad_norm": 2.282294988632202,
      "learning_rate": 1.2330867620280044e-07,
      "loss": 0.4367,
      "step": 440460
    },
    {
      "epoch": 180.3,
      "grad_norm": 2.016279697418213,
      "learning_rate": 1.2325790610025104e-07,
      "loss": 0.4418,
      "step": 440470
    },
    {
      "epoch": 180.3,
      "grad_norm": 2.1894304752349854,
      "learning_rate": 1.2320714616943176e-07,
      "loss": 0.4344,
      "step": 440480
    },
    {
      "epoch": 180.31,
      "grad_norm": 1.9912936687469482,
      "learning_rate": 1.231563964105755e-07,
      "loss": 0.4386,
      "step": 440490
    },
    {
      "epoch": 180.31,
      "grad_norm": 2.072397232055664,
      "learning_rate": 1.231056568239146e-07,
      "loss": 0.431,
      "step": 440500
    },
    {
      "epoch": 180.32,
      "grad_norm": 2.4212729930877686,
      "learning_rate": 1.230549274096817e-07,
      "loss": 0.4208,
      "step": 440510
    },
    {
      "epoch": 180.32,
      "grad_norm": 1.8970155715942383,
      "learning_rate": 1.230042081681087e-07,
      "loss": 0.4571,
      "step": 440520
    },
    {
      "epoch": 180.32,
      "grad_norm": 1.9478451013565063,
      "learning_rate": 1.2295349909942866e-07,
      "loss": 0.4684,
      "step": 440530
    },
    {
      "epoch": 180.33,
      "grad_norm": 3.4532432556152344,
      "learning_rate": 1.229028002038732e-07,
      "loss": 0.4457,
      "step": 440540
    },
    {
      "epoch": 180.33,
      "grad_norm": 2.039104461669922,
      "learning_rate": 1.2285211148167468e-07,
      "loss": 0.4218,
      "step": 440550
    },
    {
      "epoch": 180.34,
      "grad_norm": 1.9847816228866577,
      "learning_rate": 1.2280143293306576e-07,
      "loss": 0.4331,
      "step": 440560
    },
    {
      "epoch": 180.34,
      "grad_norm": 2.4372260570526123,
      "learning_rate": 1.22750764558278e-07,
      "loss": 0.4252,
      "step": 440570
    },
    {
      "epoch": 180.34,
      "grad_norm": 1.962103009223938,
      "learning_rate": 1.2270010635754427e-07,
      "loss": 0.4468,
      "step": 440580
    },
    {
      "epoch": 180.35,
      "grad_norm": 1.719881296157837,
      "learning_rate": 1.2264945833109565e-07,
      "loss": 0.45,
      "step": 440590
    },
    {
      "epoch": 180.35,
      "grad_norm": 1.9602872133255005,
      "learning_rate": 1.2259882047916453e-07,
      "loss": 0.4397,
      "step": 440600
    },
    {
      "epoch": 180.36,
      "grad_norm": 2.1130104064941406,
      "learning_rate": 1.2254819280198297e-07,
      "loss": 0.4369,
      "step": 440610
    },
    {
      "epoch": 180.36,
      "grad_norm": 1.8687095642089844,
      "learning_rate": 1.2249757529978313e-07,
      "loss": 0.4478,
      "step": 440620
    },
    {
      "epoch": 180.36,
      "grad_norm": 1.7013413906097412,
      "learning_rate": 1.224469679727968e-07,
      "loss": 0.4646,
      "step": 440630
    },
    {
      "epoch": 180.37,
      "grad_norm": 2.724642515182495,
      "learning_rate": 1.2239637082125535e-07,
      "loss": 0.4482,
      "step": 440640
    },
    {
      "epoch": 180.37,
      "grad_norm": 1.8836135864257812,
      "learning_rate": 1.2234578384539085e-07,
      "loss": 0.4659,
      "step": 440650
    },
    {
      "epoch": 180.38,
      "grad_norm": 1.6819963455200195,
      "learning_rate": 1.2229520704543517e-07,
      "loss": 0.4522,
      "step": 440660
    },
    {
      "epoch": 180.38,
      "grad_norm": 1.7856913805007935,
      "learning_rate": 1.222446404216199e-07,
      "loss": 0.4328,
      "step": 440670
    },
    {
      "epoch": 180.38,
      "grad_norm": 4.967224597930908,
      "learning_rate": 1.221940839741766e-07,
      "loss": 0.4425,
      "step": 440680
    },
    {
      "epoch": 180.39,
      "grad_norm": 1.854042410850525,
      "learning_rate": 1.2214353770333688e-07,
      "loss": 0.451,
      "step": 440690
    },
    {
      "epoch": 180.39,
      "grad_norm": 1.892478108406067,
      "learning_rate": 1.220930016093323e-07,
      "loss": 0.435,
      "step": 440700
    },
    {
      "epoch": 180.4,
      "grad_norm": 1.8616502285003662,
      "learning_rate": 1.2204247569239473e-07,
      "loss": 0.443,
      "step": 440710
    },
    {
      "epoch": 180.4,
      "grad_norm": 2.1123156547546387,
      "learning_rate": 1.219919599527547e-07,
      "loss": 0.4461,
      "step": 440720
    },
    {
      "epoch": 180.41,
      "grad_norm": 2.196897029876709,
      "learning_rate": 1.2194145439064434e-07,
      "loss": 0.4247,
      "step": 440730
    },
    {
      "epoch": 180.41,
      "grad_norm": 1.7784281969070435,
      "learning_rate": 1.2189095900629496e-07,
      "loss": 0.4588,
      "step": 440740
    },
    {
      "epoch": 180.41,
      "grad_norm": 1.9595835208892822,
      "learning_rate": 1.218404737999376e-07,
      "loss": 0.4553,
      "step": 440750
    },
    {
      "epoch": 180.42,
      "grad_norm": 2.1384599208831787,
      "learning_rate": 1.2178999877180364e-07,
      "loss": 0.4494,
      "step": 440760
    },
    {
      "epoch": 180.42,
      "grad_norm": 2.001743793487549,
      "learning_rate": 1.2173953392212462e-07,
      "loss": 0.4429,
      "step": 440770
    },
    {
      "epoch": 180.43,
      "grad_norm": 2.1781699657440186,
      "learning_rate": 1.2168907925113108e-07,
      "loss": 0.449,
      "step": 440780
    },
    {
      "epoch": 180.43,
      "grad_norm": 2.019973039627075,
      "learning_rate": 1.2163863475905438e-07,
      "loss": 0.417,
      "step": 440790
    },
    {
      "epoch": 180.43,
      "grad_norm": 1.8292640447616577,
      "learning_rate": 1.2158820044612582e-07,
      "loss": 0.4382,
      "step": 440800
    },
    {
      "epoch": 180.44,
      "grad_norm": 2.261416435241699,
      "learning_rate": 1.2153777631257649e-07,
      "loss": 0.4508,
      "step": 440810
    },
    {
      "epoch": 180.44,
      "grad_norm": 2.0212340354919434,
      "learning_rate": 1.2148736235863716e-07,
      "loss": 0.4128,
      "step": 440820
    },
    {
      "epoch": 180.45,
      "grad_norm": 1.8271605968475342,
      "learning_rate": 1.2143695858453893e-07,
      "loss": 0.4346,
      "step": 440830
    },
    {
      "epoch": 180.45,
      "grad_norm": 2.047020196914673,
      "learning_rate": 1.2138656499051232e-07,
      "loss": 0.4485,
      "step": 440840
    },
    {
      "epoch": 180.45,
      "grad_norm": 1.9449666738510132,
      "learning_rate": 1.2133618157678893e-07,
      "loss": 0.4428,
      "step": 440850
    },
    {
      "epoch": 180.46,
      "grad_norm": 1.9696112871170044,
      "learning_rate": 1.212858083435988e-07,
      "loss": 0.4625,
      "step": 440860
    },
    {
      "epoch": 180.46,
      "grad_norm": 1.8089330196380615,
      "learning_rate": 1.2123544529117267e-07,
      "loss": 0.4527,
      "step": 440870
    },
    {
      "epoch": 180.47,
      "grad_norm": 2.280996799468994,
      "learning_rate": 1.2118509241974167e-07,
      "loss": 0.4212,
      "step": 440880
    },
    {
      "epoch": 180.47,
      "grad_norm": 1.8844115734100342,
      "learning_rate": 1.2113474972953634e-07,
      "loss": 0.475,
      "step": 440890
    },
    {
      "epoch": 180.47,
      "grad_norm": 2.0510976314544678,
      "learning_rate": 1.210844172207872e-07,
      "loss": 0.4513,
      "step": 440900
    },
    {
      "epoch": 180.48,
      "grad_norm": 2.2862963676452637,
      "learning_rate": 1.2103409489372479e-07,
      "loss": 0.4505,
      "step": 440910
    },
    {
      "epoch": 180.48,
      "grad_norm": 2.2032458782196045,
      "learning_rate": 1.2098378274857995e-07,
      "loss": 0.4209,
      "step": 440920
    },
    {
      "epoch": 180.49,
      "grad_norm": 1.9906270503997803,
      "learning_rate": 1.2093348078558267e-07,
      "loss": 0.4291,
      "step": 440930
    },
    {
      "epoch": 180.49,
      "grad_norm": 2.2138359546661377,
      "learning_rate": 1.2088318900496378e-07,
      "loss": 0.4473,
      "step": 440940
    },
    {
      "epoch": 180.5,
      "grad_norm": 2.2493879795074463,
      "learning_rate": 1.2083290740695353e-07,
      "loss": 0.4399,
      "step": 440950
    },
    {
      "epoch": 180.5,
      "grad_norm": 2.293252944946289,
      "learning_rate": 1.20782635991782e-07,
      "loss": 0.4555,
      "step": 440960
    },
    {
      "epoch": 180.5,
      "grad_norm": 2.3653311729431152,
      "learning_rate": 1.2073237475967993e-07,
      "loss": 0.4458,
      "step": 440970
    },
    {
      "epoch": 180.51,
      "grad_norm": 1.8916373252868652,
      "learning_rate": 1.206821237108771e-07,
      "loss": 0.4521,
      "step": 440980
    },
    {
      "epoch": 180.51,
      "grad_norm": 2.4073915481567383,
      "learning_rate": 1.2063188284560412e-07,
      "loss": 0.4312,
      "step": 440990
    },
    {
      "epoch": 180.52,
      "grad_norm": 1.9484254121780396,
      "learning_rate": 1.2058165216409092e-07,
      "loss": 0.4469,
      "step": 441000
    },
    {
      "epoch": 180.52,
      "grad_norm": 1.8824024200439453,
      "learning_rate": 1.205314316665676e-07,
      "loss": 0.4494,
      "step": 441010
    },
    {
      "epoch": 180.52,
      "grad_norm": 1.8493831157684326,
      "learning_rate": 1.2048122135326416e-07,
      "loss": 0.4298,
      "step": 441020
    },
    {
      "epoch": 180.53,
      "grad_norm": 1.8898906707763672,
      "learning_rate": 1.2043102122441087e-07,
      "loss": 0.4532,
      "step": 441030
    },
    {
      "epoch": 180.53,
      "grad_norm": 2.0102121829986572,
      "learning_rate": 1.2038083128023722e-07,
      "loss": 0.4272,
      "step": 441040
    },
    {
      "epoch": 180.54,
      "grad_norm": 1.7028597593307495,
      "learning_rate": 1.2033065152097354e-07,
      "loss": 0.4496,
      "step": 441050
    },
    {
      "epoch": 180.54,
      "grad_norm": 1.7596635818481445,
      "learning_rate": 1.2028048194684955e-07,
      "loss": 0.4196,
      "step": 441060
    },
    {
      "epoch": 180.54,
      "grad_norm": 2.2266693115234375,
      "learning_rate": 1.2023032255809507e-07,
      "loss": 0.4501,
      "step": 441070
    },
    {
      "epoch": 180.55,
      "grad_norm": 2.2305567264556885,
      "learning_rate": 1.2018017335493979e-07,
      "loss": 0.4329,
      "step": 441080
    },
    {
      "epoch": 180.55,
      "grad_norm": 2.393939733505249,
      "learning_rate": 1.201300343376138e-07,
      "loss": 0.4442,
      "step": 441090
    },
    {
      "epoch": 180.56,
      "grad_norm": 2.4174652099609375,
      "learning_rate": 1.200799055063463e-07,
      "loss": 0.4424,
      "step": 441100
    },
    {
      "epoch": 180.56,
      "grad_norm": 2.0535879135131836,
      "learning_rate": 1.200297868613671e-07,
      "loss": 0.4327,
      "step": 441110
    },
    {
      "epoch": 180.56,
      "grad_norm": 1.8202602863311768,
      "learning_rate": 1.1997967840290564e-07,
      "loss": 0.4305,
      "step": 441120
    },
    {
      "epoch": 180.57,
      "grad_norm": 2.2140438556671143,
      "learning_rate": 1.1992958013119198e-07,
      "loss": 0.4572,
      "step": 441130
    },
    {
      "epoch": 180.57,
      "grad_norm": 2.584235429763794,
      "learning_rate": 1.1987949204645512e-07,
      "loss": 0.4603,
      "step": 441140
    },
    {
      "epoch": 180.58,
      "grad_norm": 2.447406768798828,
      "learning_rate": 1.1982941414892454e-07,
      "loss": 0.4498,
      "step": 441150
    },
    {
      "epoch": 180.58,
      "grad_norm": 1.8430492877960205,
      "learning_rate": 1.197793464388303e-07,
      "loss": 0.4385,
      "step": 441160
    },
    {
      "epoch": 180.59,
      "grad_norm": 1.611258625984192,
      "learning_rate": 1.1972928891640027e-07,
      "loss": 0.4501,
      "step": 441170
    },
    {
      "epoch": 180.59,
      "grad_norm": 2.178532838821411,
      "learning_rate": 1.196792415818653e-07,
      "loss": 0.4533,
      "step": 441180
    },
    {
      "epoch": 180.59,
      "grad_norm": 2.206239938735962,
      "learning_rate": 1.1962920443545389e-07,
      "loss": 0.4256,
      "step": 441190
    },
    {
      "epoch": 180.6,
      "grad_norm": 2.48005747795105,
      "learning_rate": 1.195791774773955e-07,
      "loss": 0.4407,
      "step": 441200
    },
    {
      "epoch": 180.6,
      "grad_norm": 1.6585274934768677,
      "learning_rate": 1.195291607079191e-07,
      "loss": 0.4624,
      "step": 441210
    },
    {
      "epoch": 180.61,
      "grad_norm": 2.3463661670684814,
      "learning_rate": 1.194791541272537e-07,
      "loss": 0.4559,
      "step": 441220
    },
    {
      "epoch": 180.61,
      "grad_norm": 1.9115300178527832,
      "learning_rate": 1.1942915773562878e-07,
      "loss": 0.4512,
      "step": 441230
    },
    {
      "epoch": 180.61,
      "grad_norm": 1.7782599925994873,
      "learning_rate": 1.1937917153327308e-07,
      "loss": 0.4345,
      "step": 441240
    },
    {
      "epoch": 180.62,
      "grad_norm": 2.248793840408325,
      "learning_rate": 1.1932919552041587e-07,
      "loss": 0.4398,
      "step": 441250
    },
    {
      "epoch": 180.62,
      "grad_norm": 1.972164511680603,
      "learning_rate": 1.1927922969728557e-07,
      "loss": 0.436,
      "step": 441260
    },
    {
      "epoch": 180.63,
      "grad_norm": 2.408827543258667,
      "learning_rate": 1.1922927406411172e-07,
      "loss": 0.4357,
      "step": 441270
    },
    {
      "epoch": 180.63,
      "grad_norm": 2.5130629539489746,
      "learning_rate": 1.1917932862112248e-07,
      "loss": 0.4605,
      "step": 441280
    },
    {
      "epoch": 180.63,
      "grad_norm": 2.5311310291290283,
      "learning_rate": 1.1912939336854739e-07,
      "loss": 0.4322,
      "step": 441290
    },
    {
      "epoch": 180.64,
      "grad_norm": 2.0346317291259766,
      "learning_rate": 1.1907946830661463e-07,
      "loss": 0.4439,
      "step": 441300
    },
    {
      "epoch": 180.64,
      "grad_norm": 1.4749468564987183,
      "learning_rate": 1.1902955343555294e-07,
      "loss": 0.4314,
      "step": 441310
    },
    {
      "epoch": 180.65,
      "grad_norm": 1.7491730451583862,
      "learning_rate": 1.1897964875559078e-07,
      "loss": 0.4575,
      "step": 441320
    },
    {
      "epoch": 180.65,
      "grad_norm": 2.256988763809204,
      "learning_rate": 1.1892975426695739e-07,
      "loss": 0.4378,
      "step": 441330
    },
    {
      "epoch": 180.65,
      "grad_norm": 2.3124871253967285,
      "learning_rate": 1.1887986996988097e-07,
      "loss": 0.4634,
      "step": 441340
    },
    {
      "epoch": 180.66,
      "grad_norm": 1.9798215627670288,
      "learning_rate": 1.1882999586458999e-07,
      "loss": 0.4401,
      "step": 441350
    },
    {
      "epoch": 180.66,
      "grad_norm": 2.4963555335998535,
      "learning_rate": 1.1878013195131264e-07,
      "loss": 0.4449,
      "step": 441360
    },
    {
      "epoch": 180.67,
      "grad_norm": 2.0280110836029053,
      "learning_rate": 1.1873027823027792e-07,
      "loss": 0.4433,
      "step": 441370
    },
    {
      "epoch": 180.67,
      "grad_norm": 2.0727434158325195,
      "learning_rate": 1.1868043470171403e-07,
      "loss": 0.4454,
      "step": 441380
    },
    {
      "epoch": 180.68,
      "grad_norm": 2.2541730403900146,
      "learning_rate": 1.1863060136584915e-07,
      "loss": 0.4687,
      "step": 441390
    },
    {
      "epoch": 180.68,
      "grad_norm": 1.8801227807998657,
      "learning_rate": 1.1858077822291178e-07,
      "loss": 0.4582,
      "step": 441400
    },
    {
      "epoch": 180.68,
      "grad_norm": 2.2737109661102295,
      "learning_rate": 1.1853096527312985e-07,
      "loss": 0.4555,
      "step": 441410
    },
    {
      "epoch": 180.69,
      "grad_norm": 2.053884506225586,
      "learning_rate": 1.1848116251673181e-07,
      "loss": 0.4316,
      "step": 441420
    },
    {
      "epoch": 180.69,
      "grad_norm": 2.2475063800811768,
      "learning_rate": 1.1843136995394612e-07,
      "loss": 0.4529,
      "step": 441430
    },
    {
      "epoch": 180.7,
      "grad_norm": 1.757653832435608,
      "learning_rate": 1.1838158758499996e-07,
      "loss": 0.4508,
      "step": 441440
    },
    {
      "epoch": 180.7,
      "grad_norm": 1.7178419828414917,
      "learning_rate": 1.1833181541012178e-07,
      "loss": 0.4489,
      "step": 441450
    },
    {
      "epoch": 180.7,
      "grad_norm": 2.1366724967956543,
      "learning_rate": 1.182820534295398e-07,
      "loss": 0.4303,
      "step": 441460
    },
    {
      "epoch": 180.71,
      "grad_norm": 2.3357245922088623,
      "learning_rate": 1.1823230164348168e-07,
      "loss": 0.4346,
      "step": 441470
    },
    {
      "epoch": 180.71,
      "grad_norm": 1.720810890197754,
      "learning_rate": 1.1818256005217564e-07,
      "loss": 0.4379,
      "step": 441480
    },
    {
      "epoch": 180.72,
      "grad_norm": 1.8033761978149414,
      "learning_rate": 1.1813282865584935e-07,
      "loss": 0.441,
      "step": 441490
    },
    {
      "epoch": 180.72,
      "grad_norm": 2.153721332550049,
      "learning_rate": 1.1808310745473051e-07,
      "loss": 0.4596,
      "step": 441500
    },
    {
      "epoch": 180.72,
      "grad_norm": 1.6635438203811646,
      "learning_rate": 1.1803339644904704e-07,
      "loss": 0.4361,
      "step": 441510
    },
    {
      "epoch": 180.73,
      "grad_norm": 2.2495546340942383,
      "learning_rate": 1.179836956390269e-07,
      "loss": 0.4572,
      "step": 441520
    },
    {
      "epoch": 180.73,
      "grad_norm": 2.02174711227417,
      "learning_rate": 1.1793400502489724e-07,
      "loss": 0.4391,
      "step": 441530
    },
    {
      "epoch": 180.74,
      "grad_norm": 2.016024351119995,
      "learning_rate": 1.1788432460688603e-07,
      "loss": 0.4566,
      "step": 441540
    },
    {
      "epoch": 180.74,
      "grad_norm": 2.0964303016662598,
      "learning_rate": 1.1783465438522039e-07,
      "loss": 0.4556,
      "step": 441550
    },
    {
      "epoch": 180.74,
      "grad_norm": 2.370720148086548,
      "learning_rate": 1.1778499436012855e-07,
      "loss": 0.437,
      "step": 441560
    },
    {
      "epoch": 180.75,
      "grad_norm": 2.023658037185669,
      "learning_rate": 1.1773534453183792e-07,
      "loss": 0.4387,
      "step": 441570
    },
    {
      "epoch": 180.75,
      "grad_norm": 2.265716791152954,
      "learning_rate": 1.176857049005754e-07,
      "loss": 0.4194,
      "step": 441580
    },
    {
      "epoch": 180.76,
      "grad_norm": 1.9419140815734863,
      "learning_rate": 1.176360754665687e-07,
      "loss": 0.4445,
      "step": 441590
    },
    {
      "epoch": 180.76,
      "grad_norm": 1.9361138343811035,
      "learning_rate": 1.1758645623004522e-07,
      "loss": 0.4459,
      "step": 441600
    },
    {
      "epoch": 180.77,
      "grad_norm": 1.7640836238861084,
      "learning_rate": 1.1753684719123212e-07,
      "loss": 0.4339,
      "step": 441610
    },
    {
      "epoch": 180.77,
      "grad_norm": 2.675189733505249,
      "learning_rate": 1.1748724835035657e-07,
      "loss": 0.4713,
      "step": 441620
    },
    {
      "epoch": 180.77,
      "grad_norm": 1.8219350576400757,
      "learning_rate": 1.1743765970764598e-07,
      "loss": 0.4389,
      "step": 441630
    },
    {
      "epoch": 180.78,
      "grad_norm": 1.9645228385925293,
      "learning_rate": 1.1738808126332752e-07,
      "loss": 0.4419,
      "step": 441640
    },
    {
      "epoch": 180.78,
      "grad_norm": 1.8883765935897827,
      "learning_rate": 1.173385130176281e-07,
      "loss": 0.4362,
      "step": 441650
    },
    {
      "epoch": 180.79,
      "grad_norm": 10.755956649780273,
      "learning_rate": 1.1728895497077487e-07,
      "loss": 0.4405,
      "step": 441660
    },
    {
      "epoch": 180.79,
      "grad_norm": 2.3115031719207764,
      "learning_rate": 1.17239407122995e-07,
      "loss": 0.4402,
      "step": 441670
    },
    {
      "epoch": 180.79,
      "grad_norm": 1.9778631925582886,
      "learning_rate": 1.1718986947451538e-07,
      "loss": 0.4408,
      "step": 441680
    },
    {
      "epoch": 180.8,
      "grad_norm": 2.1465706825256348,
      "learning_rate": 1.1714034202556267e-07,
      "loss": 0.4395,
      "step": 441690
    },
    {
      "epoch": 180.8,
      "grad_norm": 1.854353666305542,
      "learning_rate": 1.1709082477636402e-07,
      "loss": 0.4393,
      "step": 441700
    },
    {
      "epoch": 180.81,
      "grad_norm": 1.989091396331787,
      "learning_rate": 1.1704131772714633e-07,
      "loss": 0.454,
      "step": 441710
    },
    {
      "epoch": 180.81,
      "grad_norm": 1.9143571853637695,
      "learning_rate": 1.1699182087813626e-07,
      "loss": 0.4515,
      "step": 441720
    },
    {
      "epoch": 180.81,
      "grad_norm": 2.1204240322113037,
      "learning_rate": 1.1694233422956068e-07,
      "loss": 0.4478,
      "step": 441730
    },
    {
      "epoch": 180.82,
      "grad_norm": 2.29374361038208,
      "learning_rate": 1.1689285778164575e-07,
      "loss": 0.4238,
      "step": 441740
    },
    {
      "epoch": 180.82,
      "grad_norm": 1.7249131202697754,
      "learning_rate": 1.1684339153461885e-07,
      "loss": 0.4567,
      "step": 441750
    },
    {
      "epoch": 180.83,
      "grad_norm": 2.1421709060668945,
      "learning_rate": 1.1679393548870612e-07,
      "loss": 0.4532,
      "step": 441760
    },
    {
      "epoch": 180.83,
      "grad_norm": 2.1501448154449463,
      "learning_rate": 1.1674448964413422e-07,
      "loss": 0.4206,
      "step": 441770
    },
    {
      "epoch": 180.84,
      "grad_norm": 2.1317636966705322,
      "learning_rate": 1.1669505400112977e-07,
      "loss": 0.4632,
      "step": 441780
    },
    {
      "epoch": 180.84,
      "grad_norm": 1.7616769075393677,
      "learning_rate": 1.1664562855991915e-07,
      "loss": 0.4619,
      "step": 441790
    },
    {
      "epoch": 180.84,
      "grad_norm": 2.2844595909118652,
      "learning_rate": 1.165962133207285e-07,
      "loss": 0.4586,
      "step": 441800
    },
    {
      "epoch": 180.85,
      "grad_norm": 2.086186408996582,
      "learning_rate": 1.165468082837847e-07,
      "loss": 0.436,
      "step": 441810
    },
    {
      "epoch": 180.85,
      "grad_norm": 2.491664171218872,
      "learning_rate": 1.1649741344931362e-07,
      "loss": 0.4368,
      "step": 441820
    },
    {
      "epoch": 180.86,
      "grad_norm": 2.1906392574310303,
      "learning_rate": 1.164480288175419e-07,
      "loss": 0.4327,
      "step": 441830
    },
    {
      "epoch": 180.86,
      "grad_norm": 2.1424787044525146,
      "learning_rate": 1.1639865438869541e-07,
      "loss": 0.4556,
      "step": 441840
    },
    {
      "epoch": 180.86,
      "grad_norm": 2.349876880645752,
      "learning_rate": 1.1634929016300053e-07,
      "loss": 0.4457,
      "step": 441850
    },
    {
      "epoch": 180.87,
      "grad_norm": 2.742291212081909,
      "learning_rate": 1.1629993614068337e-07,
      "loss": 0.441,
      "step": 441860
    },
    {
      "epoch": 180.87,
      "grad_norm": 2.253464937210083,
      "learning_rate": 1.162505923219698e-07,
      "loss": 0.451,
      "step": 441870
    },
    {
      "epoch": 180.88,
      "grad_norm": 2.214193105697632,
      "learning_rate": 1.1620125870708645e-07,
      "loss": 0.4728,
      "step": 441880
    },
    {
      "epoch": 180.88,
      "grad_norm": 2.367168426513672,
      "learning_rate": 1.1615193529625868e-07,
      "loss": 0.4331,
      "step": 441890
    },
    {
      "epoch": 180.88,
      "grad_norm": 2.060382127761841,
      "learning_rate": 1.1610262208971259e-07,
      "loss": 0.4429,
      "step": 441900
    },
    {
      "epoch": 180.89,
      "grad_norm": 1.9963207244873047,
      "learning_rate": 1.1605331908767432e-07,
      "loss": 0.4447,
      "step": 441910
    },
    {
      "epoch": 180.89,
      "grad_norm": 2.0906710624694824,
      "learning_rate": 1.1600402629036894e-07,
      "loss": 0.4476,
      "step": 441920
    },
    {
      "epoch": 180.9,
      "grad_norm": 1.8969184160232544,
      "learning_rate": 1.1595474369802335e-07,
      "loss": 0.4693,
      "step": 441930
    },
    {
      "epoch": 180.9,
      "grad_norm": 1.9023206233978271,
      "learning_rate": 1.159054713108629e-07,
      "loss": 0.4484,
      "step": 441940
    },
    {
      "epoch": 180.9,
      "grad_norm": 1.9371403455734253,
      "learning_rate": 1.1585620912911292e-07,
      "loss": 0.4347,
      "step": 441950
    },
    {
      "epoch": 180.91,
      "grad_norm": 1.885796308517456,
      "learning_rate": 1.1580695715299954e-07,
      "loss": 0.4445,
      "step": 441960
    },
    {
      "epoch": 180.91,
      "grad_norm": 1.8001954555511475,
      "learning_rate": 1.1575771538274835e-07,
      "loss": 0.442,
      "step": 441970
    },
    {
      "epoch": 180.92,
      "grad_norm": 1.9753766059875488,
      "learning_rate": 1.1570848381858444e-07,
      "loss": 0.4372,
      "step": 441980
    },
    {
      "epoch": 180.92,
      "grad_norm": 2.0578620433807373,
      "learning_rate": 1.1565926246073394e-07,
      "loss": 0.4505,
      "step": 441990
    },
    {
      "epoch": 180.93,
      "grad_norm": 2.1056697368621826,
      "learning_rate": 1.156100513094219e-07,
      "loss": 0.4467,
      "step": 442000
    },
    {
      "epoch": 180.93,
      "grad_norm": 3.208831548690796,
      "learning_rate": 1.1556085036487422e-07,
      "loss": 0.457,
      "step": 442010
    },
    {
      "epoch": 180.93,
      "grad_norm": 2.2641866207122803,
      "learning_rate": 1.1551165962731569e-07,
      "loss": 0.4422,
      "step": 442020
    },
    {
      "epoch": 180.94,
      "grad_norm": 2.4434590339660645,
      "learning_rate": 1.1546247909697193e-07,
      "loss": 0.439,
      "step": 442030
    },
    {
      "epoch": 180.94,
      "grad_norm": 2.1027700901031494,
      "learning_rate": 1.15413308774068e-07,
      "loss": 0.4467,
      "step": 442040
    },
    {
      "epoch": 180.95,
      "grad_norm": 2.0148508548736572,
      "learning_rate": 1.1536414865882951e-07,
      "loss": 0.4347,
      "step": 442050
    },
    {
      "epoch": 180.95,
      "grad_norm": 2.21714186668396,
      "learning_rate": 1.1531499875148155e-07,
      "loss": 0.4559,
      "step": 442060
    },
    {
      "epoch": 180.95,
      "grad_norm": 1.8764137029647827,
      "learning_rate": 1.1526585905224919e-07,
      "loss": 0.4364,
      "step": 442070
    },
    {
      "epoch": 180.96,
      "grad_norm": 2.183549404144287,
      "learning_rate": 1.1521672956135751e-07,
      "loss": 0.4555,
      "step": 442080
    },
    {
      "epoch": 180.96,
      "grad_norm": 2.2440438270568848,
      "learning_rate": 1.151676102790316e-07,
      "loss": 0.4452,
      "step": 442090
    },
    {
      "epoch": 180.97,
      "grad_norm": 2.2409586906433105,
      "learning_rate": 1.1511850120549654e-07,
      "loss": 0.4345,
      "step": 442100
    },
    {
      "epoch": 180.97,
      "grad_norm": 2.1475882530212402,
      "learning_rate": 1.1506940234097687e-07,
      "loss": 0.4423,
      "step": 442110
    },
    {
      "epoch": 180.97,
      "grad_norm": 2.9307756423950195,
      "learning_rate": 1.150203136856982e-07,
      "loss": 0.4538,
      "step": 442120
    },
    {
      "epoch": 180.98,
      "grad_norm": 2.1098921298980713,
      "learning_rate": 1.1497123523988511e-07,
      "loss": 0.4392,
      "step": 442130
    },
    {
      "epoch": 180.98,
      "grad_norm": 2.0855636596679688,
      "learning_rate": 1.149221670037624e-07,
      "loss": 0.4331,
      "step": 442140
    },
    {
      "epoch": 180.99,
      "grad_norm": 2.07867693901062,
      "learning_rate": 1.148731089775549e-07,
      "loss": 0.4284,
      "step": 442150
    },
    {
      "epoch": 180.99,
      "grad_norm": 2.1336162090301514,
      "learning_rate": 1.1482406116148715e-07,
      "loss": 0.4476,
      "step": 442160
    },
    {
      "epoch": 180.99,
      "grad_norm": 2.167421340942383,
      "learning_rate": 1.1477502355578374e-07,
      "loss": 0.4379,
      "step": 442170
    },
    {
      "epoch": 181.0,
      "grad_norm": 2.1675751209259033,
      "learning_rate": 1.1472599616066973e-07,
      "loss": 0.4333,
      "step": 442180
    },
    {
      "epoch": 181.0,
      "eval_loss": 0.44383421540260315,
      "eval_runtime": 57.026,
      "eval_samples_per_second": 60.481,
      "eval_steps_per_second": 7.575,
      "step": 442183
    },
    {
      "epoch": 181.0,
      "grad_norm": 2.2288095951080322,
      "learning_rate": 1.1467697897636941e-07,
      "loss": 0.4554,
      "step": 442190
    },
    {
      "epoch": 181.01,
      "grad_norm": 2.2189559936523438,
      "learning_rate": 1.1462797200310738e-07,
      "loss": 0.452,
      "step": 442200
    },
    {
      "epoch": 181.01,
      "grad_norm": 2.3261308670043945,
      "learning_rate": 1.145789752411079e-07,
      "loss": 0.4453,
      "step": 442210
    },
    {
      "epoch": 181.02,
      "grad_norm": 2.0346977710723877,
      "learning_rate": 1.1452998869059606e-07,
      "loss": 0.4211,
      "step": 442220
    },
    {
      "epoch": 181.02,
      "grad_norm": 1.9265398979187012,
      "learning_rate": 1.144810123517954e-07,
      "loss": 0.435,
      "step": 442230
    },
    {
      "epoch": 181.02,
      "grad_norm": 2.0059266090393066,
      "learning_rate": 1.1443204622493099e-07,
      "loss": 0.425,
      "step": 442240
    },
    {
      "epoch": 181.03,
      "grad_norm": 2.2411162853240967,
      "learning_rate": 1.1438309031022684e-07,
      "loss": 0.4327,
      "step": 442250
    },
    {
      "epoch": 181.03,
      "grad_norm": 2.098503828048706,
      "learning_rate": 1.1433414460790729e-07,
      "loss": 0.4336,
      "step": 442260
    },
    {
      "epoch": 181.04,
      "grad_norm": 2.301107406616211,
      "learning_rate": 1.1428520911819662e-07,
      "loss": 0.4364,
      "step": 442270
    },
    {
      "epoch": 181.04,
      "grad_norm": 2.4605681896209717,
      "learning_rate": 1.1423628384131861e-07,
      "loss": 0.4284,
      "step": 442280
    },
    {
      "epoch": 181.04,
      "grad_norm": 1.810559868812561,
      "learning_rate": 1.1418736877749782e-07,
      "loss": 0.4496,
      "step": 442290
    },
    {
      "epoch": 181.05,
      "grad_norm": 2.3488025665283203,
      "learning_rate": 1.1413846392695803e-07,
      "loss": 0.422,
      "step": 442300
    },
    {
      "epoch": 181.05,
      "grad_norm": 1.8040072917938232,
      "learning_rate": 1.1408956928992356e-07,
      "loss": 0.4465,
      "step": 442310
    },
    {
      "epoch": 181.06,
      "grad_norm": 1.6380670070648193,
      "learning_rate": 1.1404068486661815e-07,
      "loss": 0.4575,
      "step": 442320
    },
    {
      "epoch": 181.06,
      "grad_norm": 2.1921699047088623,
      "learning_rate": 1.1399181065726562e-07,
      "loss": 0.4245,
      "step": 442330
    },
    {
      "epoch": 181.06,
      "grad_norm": 2.175229787826538,
      "learning_rate": 1.1394294666209023e-07,
      "loss": 0.4231,
      "step": 442340
    },
    {
      "epoch": 181.07,
      "grad_norm": 1.7552547454833984,
      "learning_rate": 1.1389409288131554e-07,
      "loss": 0.4291,
      "step": 442350
    },
    {
      "epoch": 181.07,
      "grad_norm": 2.4023585319519043,
      "learning_rate": 1.1384524931516556e-07,
      "loss": 0.4307,
      "step": 442360
    },
    {
      "epoch": 181.08,
      "grad_norm": 1.8697259426116943,
      "learning_rate": 1.1379641596386381e-07,
      "loss": 0.4305,
      "step": 442370
    },
    {
      "epoch": 181.08,
      "grad_norm": 1.9996424913406372,
      "learning_rate": 1.1374759282763409e-07,
      "loss": 0.4276,
      "step": 442380
    },
    {
      "epoch": 181.08,
      "grad_norm": 1.894991397857666,
      "learning_rate": 1.136987799066999e-07,
      "loss": 0.4247,
      "step": 442390
    },
    {
      "epoch": 181.09,
      "grad_norm": 1.9488128423690796,
      "learning_rate": 1.1364997720128503e-07,
      "loss": 0.4481,
      "step": 442400
    },
    {
      "epoch": 181.09,
      "grad_norm": 2.009077548980713,
      "learning_rate": 1.1360118471161326e-07,
      "loss": 0.4348,
      "step": 442410
    },
    {
      "epoch": 181.1,
      "grad_norm": 2.028817892074585,
      "learning_rate": 1.1355240243790756e-07,
      "loss": 0.4472,
      "step": 442420
    },
    {
      "epoch": 181.1,
      "grad_norm": 1.9460874795913696,
      "learning_rate": 1.1350363038039175e-07,
      "loss": 0.4465,
      "step": 442430
    },
    {
      "epoch": 181.11,
      "grad_norm": 2.0941262245178223,
      "learning_rate": 1.1345486853928907e-07,
      "loss": 0.4346,
      "step": 442440
    },
    {
      "epoch": 181.11,
      "grad_norm": 2.5459187030792236,
      "learning_rate": 1.1340611691482305e-07,
      "loss": 0.434,
      "step": 442450
    },
    {
      "epoch": 181.11,
      "grad_norm": 2.0162484645843506,
      "learning_rate": 1.133573755072172e-07,
      "loss": 0.4524,
      "step": 442460
    },
    {
      "epoch": 181.12,
      "grad_norm": 1.8684278726577759,
      "learning_rate": 1.1330864431669424e-07,
      "loss": 0.4537,
      "step": 442470
    },
    {
      "epoch": 181.12,
      "grad_norm": 3.227789878845215,
      "learning_rate": 1.1325992334347798e-07,
      "loss": 0.4298,
      "step": 442480
    },
    {
      "epoch": 181.13,
      "grad_norm": 2.1588311195373535,
      "learning_rate": 1.1321121258779063e-07,
      "loss": 0.4385,
      "step": 442490
    },
    {
      "epoch": 181.13,
      "grad_norm": 2.136901378631592,
      "learning_rate": 1.1316251204985649e-07,
      "loss": 0.4453,
      "step": 442500
    },
    {
      "epoch": 181.13,
      "grad_norm": 2.4376580715179443,
      "learning_rate": 1.1311382172989829e-07,
      "loss": 0.4245,
      "step": 442510
    },
    {
      "epoch": 181.14,
      "grad_norm": 2.2555465698242188,
      "learning_rate": 1.1306514162813876e-07,
      "loss": 0.4236,
      "step": 442520
    },
    {
      "epoch": 181.14,
      "grad_norm": 1.7846909761428833,
      "learning_rate": 1.1301647174480118e-07,
      "loss": 0.4361,
      "step": 442530
    },
    {
      "epoch": 181.15,
      "grad_norm": 2.0884735584259033,
      "learning_rate": 1.1296781208010854e-07,
      "loss": 0.4489,
      "step": 442540
    },
    {
      "epoch": 181.15,
      "grad_norm": 1.8392637968063354,
      "learning_rate": 1.1291916263428359e-07,
      "loss": 0.4288,
      "step": 442550
    },
    {
      "epoch": 181.15,
      "grad_norm": 1.7452527284622192,
      "learning_rate": 1.1287052340754905e-07,
      "loss": 0.4386,
      "step": 442560
    },
    {
      "epoch": 181.16,
      "grad_norm": 2.376725196838379,
      "learning_rate": 1.1282189440012792e-07,
      "loss": 0.4533,
      "step": 442570
    },
    {
      "epoch": 181.16,
      "grad_norm": 1.9418810606002808,
      "learning_rate": 1.1277327561224321e-07,
      "loss": 0.4478,
      "step": 442580
    },
    {
      "epoch": 181.17,
      "grad_norm": 2.024235725402832,
      "learning_rate": 1.1272466704411714e-07,
      "loss": 0.4427,
      "step": 442590
    },
    {
      "epoch": 181.17,
      "grad_norm": 1.8703851699829102,
      "learning_rate": 1.1267606869597295e-07,
      "loss": 0.4336,
      "step": 442600
    },
    {
      "epoch": 181.17,
      "grad_norm": 2.1610560417175293,
      "learning_rate": 1.1262748056803262e-07,
      "loss": 0.4555,
      "step": 442610
    },
    {
      "epoch": 181.18,
      "grad_norm": 2.6843526363372803,
      "learning_rate": 1.1257890266051888e-07,
      "loss": 0.4426,
      "step": 442620
    },
    {
      "epoch": 181.18,
      "grad_norm": 1.8293808698654175,
      "learning_rate": 1.1253033497365446e-07,
      "loss": 0.4277,
      "step": 442630
    },
    {
      "epoch": 181.19,
      "grad_norm": 2.1129043102264404,
      "learning_rate": 1.1248177750766184e-07,
      "loss": 0.442,
      "step": 442640
    },
    {
      "epoch": 181.19,
      "grad_norm": 2.17069411277771,
      "learning_rate": 1.1243323026276352e-07,
      "loss": 0.4277,
      "step": 442650
    },
    {
      "epoch": 181.2,
      "grad_norm": 2.141090154647827,
      "learning_rate": 1.1238469323918168e-07,
      "loss": 0.4358,
      "step": 442660
    },
    {
      "epoch": 181.2,
      "grad_norm": 2.7174055576324463,
      "learning_rate": 1.1233616643713858e-07,
      "loss": 0.4391,
      "step": 442670
    },
    {
      "epoch": 181.2,
      "grad_norm": 2.0311591625213623,
      "learning_rate": 1.122876498568564e-07,
      "loss": 0.4463,
      "step": 442680
    },
    {
      "epoch": 181.21,
      "grad_norm": 2.056262969970703,
      "learning_rate": 1.1223914349855817e-07,
      "loss": 0.4453,
      "step": 442690
    },
    {
      "epoch": 181.21,
      "grad_norm": 2.1560351848602295,
      "learning_rate": 1.1219064736246531e-07,
      "loss": 0.4411,
      "step": 442700
    },
    {
      "epoch": 181.22,
      "grad_norm": 1.7350437641143799,
      "learning_rate": 1.1214216144880055e-07,
      "loss": 0.441,
      "step": 442710
    },
    {
      "epoch": 181.22,
      "grad_norm": 1.9728246927261353,
      "learning_rate": 1.1209368575778535e-07,
      "loss": 0.4424,
      "step": 442720
    },
    {
      "epoch": 181.22,
      "grad_norm": 2.0651187896728516,
      "learning_rate": 1.120452202896427e-07,
      "loss": 0.4481,
      "step": 442730
    },
    {
      "epoch": 181.23,
      "grad_norm": 1.8833379745483398,
      "learning_rate": 1.1199676504459352e-07,
      "loss": 0.4418,
      "step": 442740
    },
    {
      "epoch": 181.23,
      "grad_norm": 1.857406497001648,
      "learning_rate": 1.1194832002286055e-07,
      "loss": 0.4408,
      "step": 442750
    },
    {
      "epoch": 181.24,
      "grad_norm": 1.9828155040740967,
      "learning_rate": 1.118998852246652e-07,
      "loss": 0.4542,
      "step": 442760
    },
    {
      "epoch": 181.24,
      "grad_norm": 2.4129414558410645,
      "learning_rate": 1.1185146065022946e-07,
      "loss": 0.4408,
      "step": 442770
    },
    {
      "epoch": 181.24,
      "grad_norm": 2.0316662788391113,
      "learning_rate": 1.1180304629977554e-07,
      "loss": 0.4665,
      "step": 442780
    },
    {
      "epoch": 181.25,
      "grad_norm": 2.128035306930542,
      "learning_rate": 1.1175464217352487e-07,
      "loss": 0.4272,
      "step": 442790
    },
    {
      "epoch": 181.25,
      "grad_norm": 1.8166885375976562,
      "learning_rate": 1.1170624827169942e-07,
      "loss": 0.442,
      "step": 442800
    },
    {
      "epoch": 181.26,
      "grad_norm": 1.767930269241333,
      "learning_rate": 1.1165786459452034e-07,
      "loss": 0.4558,
      "step": 442810
    },
    {
      "epoch": 181.26,
      "grad_norm": 2.3520114421844482,
      "learning_rate": 1.116094911422099e-07,
      "loss": 0.4428,
      "step": 442820
    },
    {
      "epoch": 181.26,
      "grad_norm": 2.2888307571411133,
      "learning_rate": 1.1156112791498947e-07,
      "loss": 0.4536,
      "step": 442830
    },
    {
      "epoch": 181.27,
      "grad_norm": 1.8776756525039673,
      "learning_rate": 1.1151277491308055e-07,
      "loss": 0.4443,
      "step": 442840
    },
    {
      "epoch": 181.27,
      "grad_norm": 2.0027072429656982,
      "learning_rate": 1.1146443213670454e-07,
      "loss": 0.4421,
      "step": 442850
    },
    {
      "epoch": 181.28,
      "grad_norm": 2.130305290222168,
      "learning_rate": 1.1141609958608315e-07,
      "loss": 0.4442,
      "step": 442860
    },
    {
      "epoch": 181.28,
      "grad_norm": 2.0755553245544434,
      "learning_rate": 1.1136777726143756e-07,
      "loss": 0.4567,
      "step": 442870
    },
    {
      "epoch": 181.29,
      "grad_norm": 2.5115020275115967,
      "learning_rate": 1.113194651629892e-07,
      "loss": 0.4323,
      "step": 442880
    },
    {
      "epoch": 181.29,
      "grad_norm": 1.9911646842956543,
      "learning_rate": 1.1127116329095925e-07,
      "loss": 0.4638,
      "step": 442890
    },
    {
      "epoch": 181.29,
      "grad_norm": 2.273685932159424,
      "learning_rate": 1.112228716455694e-07,
      "loss": 0.4323,
      "step": 442900
    },
    {
      "epoch": 181.3,
      "grad_norm": 1.8285467624664307,
      "learning_rate": 1.1117459022704032e-07,
      "loss": 0.4565,
      "step": 442910
    },
    {
      "epoch": 181.3,
      "grad_norm": 1.8176277875900269,
      "learning_rate": 1.111263190355937e-07,
      "loss": 0.4577,
      "step": 442920
    },
    {
      "epoch": 181.31,
      "grad_norm": 1.9640414714813232,
      "learning_rate": 1.110780580714502e-07,
      "loss": 0.465,
      "step": 442930
    },
    {
      "epoch": 181.31,
      "grad_norm": 1.6417495012283325,
      "learning_rate": 1.11029807334831e-07,
      "loss": 0.448,
      "step": 442940
    },
    {
      "epoch": 181.31,
      "grad_norm": 2.0480082035064697,
      "learning_rate": 1.1098156682595727e-07,
      "loss": 0.4452,
      "step": 442950
    },
    {
      "epoch": 181.32,
      "grad_norm": 2.179751396179199,
      "learning_rate": 1.109333365450502e-07,
      "loss": 0.4366,
      "step": 442960
    },
    {
      "epoch": 181.32,
      "grad_norm": 4.9294867515563965,
      "learning_rate": 1.1088511649233016e-07,
      "loss": 0.435,
      "step": 442970
    },
    {
      "epoch": 181.33,
      "grad_norm": 1.878381371498108,
      "learning_rate": 1.1083690666801861e-07,
      "loss": 0.4365,
      "step": 442980
    },
    {
      "epoch": 181.33,
      "grad_norm": 1.8880163431167603,
      "learning_rate": 1.1078870707233595e-07,
      "loss": 0.4398,
      "step": 442990
    },
    {
      "epoch": 181.33,
      "grad_norm": 1.760926604270935,
      "learning_rate": 1.1074051770550308e-07,
      "loss": 0.4355,
      "step": 443000
    },
    {
      "epoch": 181.34,
      "grad_norm": 1.947448968887329,
      "learning_rate": 1.1069233856774119e-07,
      "loss": 0.4321,
      "step": 443010
    },
    {
      "epoch": 181.34,
      "grad_norm": 1.8152438402175903,
      "learning_rate": 1.1064416965927015e-07,
      "loss": 0.479,
      "step": 443020
    },
    {
      "epoch": 181.35,
      "grad_norm": 2.084998607635498,
      "learning_rate": 1.105960109803114e-07,
      "loss": 0.4356,
      "step": 443030
    },
    {
      "epoch": 181.35,
      "grad_norm": 1.9613771438598633,
      "learning_rate": 1.105478625310856e-07,
      "loss": 0.4396,
      "step": 443040
    },
    {
      "epoch": 181.35,
      "grad_norm": 1.952164649963379,
      "learning_rate": 1.1049972431181236e-07,
      "loss": 0.4285,
      "step": 443050
    },
    {
      "epoch": 181.36,
      "grad_norm": 2.4890499114990234,
      "learning_rate": 1.104515963227126e-07,
      "loss": 0.4462,
      "step": 443060
    },
    {
      "epoch": 181.36,
      "grad_norm": 2.3994476795196533,
      "learning_rate": 1.1040347856400722e-07,
      "loss": 0.434,
      "step": 443070
    },
    {
      "epoch": 181.37,
      "grad_norm": 1.9791038036346436,
      "learning_rate": 1.1035537103591639e-07,
      "loss": 0.425,
      "step": 443080
    },
    {
      "epoch": 181.37,
      "grad_norm": 1.810289978981018,
      "learning_rate": 1.1030727373866049e-07,
      "loss": 0.4409,
      "step": 443090
    },
    {
      "epoch": 181.38,
      "grad_norm": 1.8977442979812622,
      "learning_rate": 1.1025918667245965e-07,
      "loss": 0.4465,
      "step": 443100
    },
    {
      "epoch": 181.38,
      "grad_norm": 2.4179089069366455,
      "learning_rate": 1.1021110983753454e-07,
      "loss": 0.4299,
      "step": 443110
    },
    {
      "epoch": 181.38,
      "grad_norm": 2.3041250705718994,
      "learning_rate": 1.1016304323410527e-07,
      "loss": 0.445,
      "step": 443120
    },
    {
      "epoch": 181.39,
      "grad_norm": 2.2267279624938965,
      "learning_rate": 1.1011498686239175e-07,
      "loss": 0.4665,
      "step": 443130
    },
    {
      "epoch": 181.39,
      "grad_norm": 1.7914091348648071,
      "learning_rate": 1.1006694072261437e-07,
      "loss": 0.4396,
      "step": 443140
    },
    {
      "epoch": 181.4,
      "grad_norm": 1.7257694005966187,
      "learning_rate": 1.1001890481499298e-07,
      "loss": 0.4552,
      "step": 443150
    },
    {
      "epoch": 181.4,
      "grad_norm": 1.8275396823883057,
      "learning_rate": 1.09970879139748e-07,
      "loss": 0.4442,
      "step": 443160
    },
    {
      "epoch": 181.4,
      "grad_norm": 2.2980406284332275,
      "learning_rate": 1.099228636970993e-07,
      "loss": 0.452,
      "step": 443170
    },
    {
      "epoch": 181.41,
      "grad_norm": 2.073834180831909,
      "learning_rate": 1.0987485848726676e-07,
      "loss": 0.4501,
      "step": 443180
    },
    {
      "epoch": 181.41,
      "grad_norm": 2.2211098670959473,
      "learning_rate": 1.0982686351047026e-07,
      "loss": 0.4519,
      "step": 443190
    },
    {
      "epoch": 181.42,
      "grad_norm": 2.348928213119507,
      "learning_rate": 1.0977887876692966e-07,
      "loss": 0.456,
      "step": 443200
    },
    {
      "epoch": 181.42,
      "grad_norm": 2.0680060386657715,
      "learning_rate": 1.0973090425686458e-07,
      "loss": 0.4811,
      "step": 443210
    },
    {
      "epoch": 181.42,
      "grad_norm": 2.6251001358032227,
      "learning_rate": 1.0968293998049517e-07,
      "loss": 0.4515,
      "step": 443220
    },
    {
      "epoch": 181.43,
      "grad_norm": 1.4759941101074219,
      "learning_rate": 1.0963498593804104e-07,
      "loss": 0.4533,
      "step": 443230
    },
    {
      "epoch": 181.43,
      "grad_norm": 1.836810827255249,
      "learning_rate": 1.0958704212972128e-07,
      "loss": 0.4553,
      "step": 443240
    },
    {
      "epoch": 181.44,
      "grad_norm": 2.2702903747558594,
      "learning_rate": 1.0953910855575655e-07,
      "loss": 0.4362,
      "step": 443250
    },
    {
      "epoch": 181.44,
      "grad_norm": 2.1009860038757324,
      "learning_rate": 1.0949118521636569e-07,
      "loss": 0.4434,
      "step": 443260
    },
    {
      "epoch": 181.44,
      "grad_norm": 2.1218693256378174,
      "learning_rate": 1.0944327211176857e-07,
      "loss": 0.4403,
      "step": 443270
    },
    {
      "epoch": 181.45,
      "grad_norm": 2.22733998298645,
      "learning_rate": 1.0939536924218454e-07,
      "loss": 0.4394,
      "step": 443280
    },
    {
      "epoch": 181.45,
      "grad_norm": 2.158026933670044,
      "learning_rate": 1.0934747660783297e-07,
      "loss": 0.4348,
      "step": 443290
    },
    {
      "epoch": 181.46,
      "grad_norm": 1.978108525276184,
      "learning_rate": 1.0929959420893346e-07,
      "loss": 0.4561,
      "step": 443300
    },
    {
      "epoch": 181.46,
      "grad_norm": 1.8743244409561157,
      "learning_rate": 1.0925172204570538e-07,
      "loss": 0.4578,
      "step": 443310
    },
    {
      "epoch": 181.47,
      "grad_norm": 1.868588924407959,
      "learning_rate": 1.0920386011836757e-07,
      "loss": 0.4571,
      "step": 443320
    },
    {
      "epoch": 181.47,
      "grad_norm": 1.8671725988388062,
      "learning_rate": 1.0915600842713963e-07,
      "loss": 0.4434,
      "step": 443330
    },
    {
      "epoch": 181.47,
      "grad_norm": 2.3884353637695312,
      "learning_rate": 1.0910816697224066e-07,
      "loss": 0.4418,
      "step": 443340
    },
    {
      "epoch": 181.48,
      "grad_norm": 1.9605472087860107,
      "learning_rate": 1.0906033575388976e-07,
      "loss": 0.4281,
      "step": 443350
    },
    {
      "epoch": 181.48,
      "grad_norm": 2.3967435359954834,
      "learning_rate": 1.0901251477230628e-07,
      "loss": 0.4304,
      "step": 443360
    },
    {
      "epoch": 181.49,
      "grad_norm": 2.0081233978271484,
      "learning_rate": 1.0896470402770906e-07,
      "loss": 0.4574,
      "step": 443370
    },
    {
      "epoch": 181.49,
      "grad_norm": 1.9461145401000977,
      "learning_rate": 1.0891690352031693e-07,
      "loss": 0.4355,
      "step": 443380
    },
    {
      "epoch": 181.49,
      "grad_norm": 2.1823463439941406,
      "learning_rate": 1.0886911325034951e-07,
      "loss": 0.4488,
      "step": 443390
    },
    {
      "epoch": 181.5,
      "grad_norm": 2.032153367996216,
      "learning_rate": 1.0882133321802512e-07,
      "loss": 0.4427,
      "step": 443400
    },
    {
      "epoch": 181.5,
      "grad_norm": 2.0837912559509277,
      "learning_rate": 1.0877356342356286e-07,
      "loss": 0.4181,
      "step": 443410
    },
    {
      "epoch": 181.51,
      "grad_norm": 2.788172960281372,
      "learning_rate": 1.0872580386718155e-07,
      "loss": 0.4422,
      "step": 443420
    },
    {
      "epoch": 181.51,
      "grad_norm": 2.1462137699127197,
      "learning_rate": 1.086780545490995e-07,
      "loss": 0.4431,
      "step": 443430
    },
    {
      "epoch": 181.51,
      "grad_norm": 1.884029507637024,
      "learning_rate": 1.0863031546953635e-07,
      "loss": 0.4357,
      "step": 443440
    },
    {
      "epoch": 181.52,
      "grad_norm": 2.1746068000793457,
      "learning_rate": 1.0858258662871065e-07,
      "loss": 0.4259,
      "step": 443450
    },
    {
      "epoch": 181.52,
      "grad_norm": 2.291602611541748,
      "learning_rate": 1.0853486802684047e-07,
      "loss": 0.4419,
      "step": 443460
    },
    {
      "epoch": 181.53,
      "grad_norm": 2.2324697971343994,
      "learning_rate": 1.0848715966414439e-07,
      "loss": 0.4362,
      "step": 443470
    },
    {
      "epoch": 181.53,
      "grad_norm": 2.001326322555542,
      "learning_rate": 1.084394615408415e-07,
      "loss": 0.4719,
      "step": 443480
    },
    {
      "epoch": 181.53,
      "grad_norm": 2.144982099533081,
      "learning_rate": 1.0839177365714983e-07,
      "loss": 0.434,
      "step": 443490
    },
    {
      "epoch": 181.54,
      "grad_norm": 2.4224460124969482,
      "learning_rate": 1.08344096013288e-07,
      "loss": 0.4695,
      "step": 443500
    },
    {
      "epoch": 181.54,
      "grad_norm": 1.8546198606491089,
      "learning_rate": 1.082964286094743e-07,
      "loss": 0.4346,
      "step": 443510
    },
    {
      "epoch": 181.55,
      "grad_norm": 2.108147144317627,
      "learning_rate": 1.0824877144592729e-07,
      "loss": 0.4459,
      "step": 443520
    },
    {
      "epoch": 181.55,
      "grad_norm": 1.8444079160690308,
      "learning_rate": 1.0820112452286532e-07,
      "loss": 0.4506,
      "step": 443530
    },
    {
      "epoch": 181.56,
      "grad_norm": 1.9472209215164185,
      "learning_rate": 1.0815348784050642e-07,
      "loss": 0.4291,
      "step": 443540
    },
    {
      "epoch": 181.56,
      "grad_norm": 2.0039682388305664,
      "learning_rate": 1.0810586139906917e-07,
      "loss": 0.4549,
      "step": 443550
    },
    {
      "epoch": 181.56,
      "grad_norm": 2.0486273765563965,
      "learning_rate": 1.0805824519877111e-07,
      "loss": 0.4501,
      "step": 443560
    },
    {
      "epoch": 181.57,
      "grad_norm": 2.7676639556884766,
      "learning_rate": 1.0801063923983108e-07,
      "loss": 0.4574,
      "step": 443570
    },
    {
      "epoch": 181.57,
      "grad_norm": 2.560978889465332,
      "learning_rate": 1.079630435224666e-07,
      "loss": 0.457,
      "step": 443580
    },
    {
      "epoch": 181.58,
      "grad_norm": 1.9997934103012085,
      "learning_rate": 1.0791545804689598e-07,
      "loss": 0.4528,
      "step": 443590
    },
    {
      "epoch": 181.58,
      "grad_norm": 1.9903326034545898,
      "learning_rate": 1.0786788281333704e-07,
      "loss": 0.4531,
      "step": 443600
    },
    {
      "epoch": 181.58,
      "grad_norm": 1.8718974590301514,
      "learning_rate": 1.0782031782200782e-07,
      "loss": 0.4619,
      "step": 443610
    },
    {
      "epoch": 181.59,
      "grad_norm": 1.7630943059921265,
      "learning_rate": 1.0777276307312636e-07,
      "loss": 0.4568,
      "step": 443620
    },
    {
      "epoch": 181.59,
      "grad_norm": 2.005397319793701,
      "learning_rate": 1.0772521856691023e-07,
      "loss": 0.442,
      "step": 443630
    },
    {
      "epoch": 181.6,
      "grad_norm": 1.8370146751403809,
      "learning_rate": 1.0767768430357719e-07,
      "loss": 0.4433,
      "step": 443640
    },
    {
      "epoch": 181.6,
      "grad_norm": 2.134824752807617,
      "learning_rate": 1.0763016028334532e-07,
      "loss": 0.4521,
      "step": 443650
    },
    {
      "epoch": 181.6,
      "grad_norm": 2.027211904525757,
      "learning_rate": 1.0758264650643214e-07,
      "loss": 0.428,
      "step": 443660
    },
    {
      "epoch": 181.61,
      "grad_norm": 2.1623456478118896,
      "learning_rate": 1.0753514297305519e-07,
      "loss": 0.4531,
      "step": 443670
    },
    {
      "epoch": 181.61,
      "grad_norm": 2.1611294746398926,
      "learning_rate": 1.0748764968343225e-07,
      "loss": 0.4486,
      "step": 443680
    },
    {
      "epoch": 181.62,
      "grad_norm": 2.0404646396636963,
      "learning_rate": 1.0744016663778088e-07,
      "loss": 0.4405,
      "step": 443690
    },
    {
      "epoch": 181.62,
      "grad_norm": 1.8262561559677124,
      "learning_rate": 1.0739269383631832e-07,
      "loss": 0.4584,
      "step": 443700
    },
    {
      "epoch": 181.63,
      "grad_norm": 2.117152690887451,
      "learning_rate": 1.0734523127926241e-07,
      "loss": 0.4384,
      "step": 443710
    },
    {
      "epoch": 181.63,
      "grad_norm": 2.0881285667419434,
      "learning_rate": 1.0729777896683013e-07,
      "loss": 0.4588,
      "step": 443720
    },
    {
      "epoch": 181.63,
      "grad_norm": 2.3622119426727295,
      "learning_rate": 1.0725033689923928e-07,
      "loss": 0.4549,
      "step": 443730
    },
    {
      "epoch": 181.64,
      "grad_norm": 2.1691224575042725,
      "learning_rate": 1.0720290507670714e-07,
      "loss": 0.4685,
      "step": 443740
    },
    {
      "epoch": 181.64,
      "grad_norm": 2.3205454349517822,
      "learning_rate": 1.0715548349945071e-07,
      "loss": 0.449,
      "step": 443750
    },
    {
      "epoch": 181.65,
      "grad_norm": 2.0515382289886475,
      "learning_rate": 1.0710807216768782e-07,
      "loss": 0.4223,
      "step": 443760
    },
    {
      "epoch": 181.65,
      "grad_norm": 1.7903622388839722,
      "learning_rate": 1.0706067108163492e-07,
      "loss": 0.4495,
      "step": 443770
    },
    {
      "epoch": 181.65,
      "grad_norm": 3.3023312091827393,
      "learning_rate": 1.0701328024150932e-07,
      "loss": 0.4445,
      "step": 443780
    },
    {
      "epoch": 181.66,
      "grad_norm": 2.0359363555908203,
      "learning_rate": 1.0696589964752827e-07,
      "loss": 0.4338,
      "step": 443790
    },
    {
      "epoch": 181.66,
      "grad_norm": 2.202169179916382,
      "learning_rate": 1.0691852929990879e-07,
      "loss": 0.416,
      "step": 443800
    },
    {
      "epoch": 181.67,
      "grad_norm": 2.1006503105163574,
      "learning_rate": 1.0687116919886763e-07,
      "loss": 0.465,
      "step": 443810
    },
    {
      "epoch": 181.67,
      "grad_norm": 1.989016056060791,
      "learning_rate": 1.0682381934462207e-07,
      "loss": 0.4506,
      "step": 443820
    },
    {
      "epoch": 181.67,
      "grad_norm": 2.2936201095581055,
      "learning_rate": 1.0677647973738913e-07,
      "loss": 0.4443,
      "step": 443830
    },
    {
      "epoch": 181.68,
      "grad_norm": 2.12443208694458,
      "learning_rate": 1.067291503773853e-07,
      "loss": 0.4735,
      "step": 443840
    },
    {
      "epoch": 181.68,
      "grad_norm": 2.1234006881713867,
      "learning_rate": 1.0668183126482759e-07,
      "loss": 0.4478,
      "step": 443850
    },
    {
      "epoch": 181.69,
      "grad_norm": 1.7817696332931519,
      "learning_rate": 1.0663452239993275e-07,
      "loss": 0.4461,
      "step": 443860
    },
    {
      "epoch": 181.69,
      "grad_norm": 1.9747308492660522,
      "learning_rate": 1.0658722378291728e-07,
      "loss": 0.4453,
      "step": 443870
    },
    {
      "epoch": 181.69,
      "grad_norm": 2.456059455871582,
      "learning_rate": 1.0653993541399818e-07,
      "loss": 0.4314,
      "step": 443880
    },
    {
      "epoch": 181.7,
      "grad_norm": 2.1931891441345215,
      "learning_rate": 1.0649265729339198e-07,
      "loss": 0.4644,
      "step": 443890
    },
    {
      "epoch": 181.7,
      "grad_norm": 2.0670783519744873,
      "learning_rate": 1.0644538942131513e-07,
      "loss": 0.454,
      "step": 443900
    },
    {
      "epoch": 181.71,
      "grad_norm": 1.7865968942642212,
      "learning_rate": 1.063981317979844e-07,
      "loss": 0.4358,
      "step": 443910
    },
    {
      "epoch": 181.71,
      "grad_norm": 1.8857412338256836,
      "learning_rate": 1.0635088442361577e-07,
      "loss": 0.4267,
      "step": 443920
    },
    {
      "epoch": 181.72,
      "grad_norm": 1.974726676940918,
      "learning_rate": 1.06303647298426e-07,
      "loss": 0.4561,
      "step": 443930
    },
    {
      "epoch": 181.72,
      "grad_norm": 2.051002025604248,
      "learning_rate": 1.0625642042263157e-07,
      "loss": 0.4683,
      "step": 443940
    },
    {
      "epoch": 181.72,
      "grad_norm": 2.129507064819336,
      "learning_rate": 1.0620920379644871e-07,
      "loss": 0.4339,
      "step": 443950
    },
    {
      "epoch": 181.73,
      "grad_norm": 2.1619184017181396,
      "learning_rate": 1.061619974200934e-07,
      "loss": 0.4546,
      "step": 443960
    },
    {
      "epoch": 181.73,
      "grad_norm": 4.868775367736816,
      "learning_rate": 1.0611480129378266e-07,
      "loss": 0.4555,
      "step": 443970
    },
    {
      "epoch": 181.74,
      "grad_norm": 1.6659786701202393,
      "learning_rate": 1.0606761541773193e-07,
      "loss": 0.4224,
      "step": 443980
    },
    {
      "epoch": 181.74,
      "grad_norm": 2.073570728302002,
      "learning_rate": 1.060204397921577e-07,
      "loss": 0.4523,
      "step": 443990
    },
    {
      "epoch": 181.74,
      "grad_norm": 2.0188992023468018,
      "learning_rate": 1.059732744172757e-07,
      "loss": 0.4493,
      "step": 444000
    },
    {
      "epoch": 181.75,
      "grad_norm": 2.5925793647766113,
      "learning_rate": 1.0592611929330268e-07,
      "loss": 0.437,
      "step": 444010
    },
    {
      "epoch": 181.75,
      "grad_norm": 2.178988218307495,
      "learning_rate": 1.0587897442045435e-07,
      "loss": 0.4445,
      "step": 444020
    },
    {
      "epoch": 181.76,
      "grad_norm": 2.135592222213745,
      "learning_rate": 1.0583183979894693e-07,
      "loss": 0.4277,
      "step": 444030
    },
    {
      "epoch": 181.76,
      "grad_norm": 1.9839471578598022,
      "learning_rate": 1.0578471542899564e-07,
      "loss": 0.4491,
      "step": 444040
    },
    {
      "epoch": 181.76,
      "grad_norm": 1.8321855068206787,
      "learning_rate": 1.0573760131081669e-07,
      "loss": 0.4544,
      "step": 444050
    },
    {
      "epoch": 181.77,
      "grad_norm": 2.210029125213623,
      "learning_rate": 1.056904974446258e-07,
      "loss": 0.4315,
      "step": 444060
    },
    {
      "epoch": 181.77,
      "grad_norm": 1.974928855895996,
      "learning_rate": 1.056434038306392e-07,
      "loss": 0.4356,
      "step": 444070
    },
    {
      "epoch": 181.78,
      "grad_norm": 2.5198299884796143,
      "learning_rate": 1.0559632046907209e-07,
      "loss": 0.4341,
      "step": 444080
    },
    {
      "epoch": 181.78,
      "grad_norm": 1.9198269844055176,
      "learning_rate": 1.0554924736014045e-07,
      "loss": 0.4413,
      "step": 444090
    },
    {
      "epoch": 181.78,
      "grad_norm": 2.1742281913757324,
      "learning_rate": 1.055021845040597e-07,
      "loss": 0.4668,
      "step": 444100
    },
    {
      "epoch": 181.79,
      "grad_norm": 2.2094407081604004,
      "learning_rate": 1.0545513190104586e-07,
      "loss": 0.4589,
      "step": 444110
    },
    {
      "epoch": 181.79,
      "grad_norm": 1.7093565464019775,
      "learning_rate": 1.0540808955131382e-07,
      "loss": 0.4292,
      "step": 444120
    },
    {
      "epoch": 181.8,
      "grad_norm": 1.9746575355529785,
      "learning_rate": 1.0536105745507958e-07,
      "loss": 0.4417,
      "step": 444130
    },
    {
      "epoch": 181.8,
      "grad_norm": 2.1543123722076416,
      "learning_rate": 1.0531403561255856e-07,
      "loss": 0.4509,
      "step": 444140
    },
    {
      "epoch": 181.81,
      "grad_norm": 1.8462132215499878,
      "learning_rate": 1.0526702402396574e-07,
      "loss": 0.4328,
      "step": 444150
    },
    {
      "epoch": 181.81,
      "grad_norm": 2.0807971954345703,
      "learning_rate": 1.0522002268951706e-07,
      "loss": 0.4445,
      "step": 444160
    },
    {
      "epoch": 181.81,
      "grad_norm": 2.0216941833496094,
      "learning_rate": 1.051730316094272e-07,
      "loss": 0.4436,
      "step": 444170
    },
    {
      "epoch": 181.82,
      "grad_norm": 1.9034970998764038,
      "learning_rate": 1.0512605078391214e-07,
      "loss": 0.4473,
      "step": 444180
    },
    {
      "epoch": 181.82,
      "grad_norm": 1.9384956359863281,
      "learning_rate": 1.0507908021318628e-07,
      "loss": 0.4308,
      "step": 444190
    },
    {
      "epoch": 181.83,
      "grad_norm": 2.0573534965515137,
      "learning_rate": 1.0503211989746533e-07,
      "loss": 0.4445,
      "step": 444200
    },
    {
      "epoch": 181.83,
      "grad_norm": 1.8607279062271118,
      "learning_rate": 1.049851698369645e-07,
      "loss": 0.4388,
      "step": 444210
    },
    {
      "epoch": 181.83,
      "grad_norm": 2.459304094314575,
      "learning_rate": 1.0493823003189846e-07,
      "loss": 0.4442,
      "step": 444220
    },
    {
      "epoch": 181.84,
      "grad_norm": 2.007687568664551,
      "learning_rate": 1.0489130048248239e-07,
      "loss": 0.4534,
      "step": 444230
    },
    {
      "epoch": 181.84,
      "grad_norm": 2.0418171882629395,
      "learning_rate": 1.0484438118893148e-07,
      "loss": 0.4403,
      "step": 444240
    },
    {
      "epoch": 181.85,
      "grad_norm": 2.0490918159484863,
      "learning_rate": 1.0479747215146016e-07,
      "loss": 0.4423,
      "step": 444250
    },
    {
      "epoch": 181.85,
      "grad_norm": 1.8873069286346436,
      "learning_rate": 1.0475057337028386e-07,
      "loss": 0.4476,
      "step": 444260
    },
    {
      "epoch": 181.85,
      "grad_norm": 1.944751501083374,
      "learning_rate": 1.0470368484561728e-07,
      "loss": 0.4584,
      "step": 444270
    },
    {
      "epoch": 181.86,
      "grad_norm": 2.055910348892212,
      "learning_rate": 1.046568065776748e-07,
      "loss": 0.4373,
      "step": 444280
    },
    {
      "epoch": 181.86,
      "grad_norm": 2.1936612129211426,
      "learning_rate": 1.0460993856667163e-07,
      "loss": 0.4594,
      "step": 444290
    },
    {
      "epoch": 181.87,
      "grad_norm": 5.679248809814453,
      "learning_rate": 1.045630808128222e-07,
      "loss": 0.4579,
      "step": 444300
    },
    {
      "epoch": 181.87,
      "grad_norm": 2.511885643005371,
      "learning_rate": 1.045162333163414e-07,
      "loss": 0.4577,
      "step": 444310
    },
    {
      "epoch": 181.87,
      "grad_norm": 2.177849054336548,
      "learning_rate": 1.0446939607744367e-07,
      "loss": 0.4323,
      "step": 444320
    },
    {
      "epoch": 181.88,
      "grad_norm": 2.3045599460601807,
      "learning_rate": 1.0442256909634368e-07,
      "loss": 0.4337,
      "step": 444330
    },
    {
      "epoch": 181.88,
      "grad_norm": 2.1001839637756348,
      "learning_rate": 1.0437575237325609e-07,
      "loss": 0.4506,
      "step": 444340
    },
    {
      "epoch": 181.89,
      "grad_norm": 2.0646023750305176,
      "learning_rate": 1.043289459083948e-07,
      "loss": 0.4421,
      "step": 444350
    },
    {
      "epoch": 181.89,
      "grad_norm": 2.2713024616241455,
      "learning_rate": 1.042821497019745e-07,
      "loss": 0.4431,
      "step": 444360
    },
    {
      "epoch": 181.9,
      "grad_norm": 2.4076719284057617,
      "learning_rate": 1.0423536375420956e-07,
      "loss": 0.4327,
      "step": 444370
    },
    {
      "epoch": 181.9,
      "grad_norm": 1.853626012802124,
      "learning_rate": 1.0418858806531416e-07,
      "loss": 0.4476,
      "step": 444380
    },
    {
      "epoch": 181.9,
      "grad_norm": 1.9637731313705444,
      "learning_rate": 1.0414182263550298e-07,
      "loss": 0.4448,
      "step": 444390
    },
    {
      "epoch": 181.91,
      "grad_norm": 2.1821982860565186,
      "learning_rate": 1.0409506746498989e-07,
      "loss": 0.44,
      "step": 444400
    },
    {
      "epoch": 181.91,
      "grad_norm": 1.9852060079574585,
      "learning_rate": 1.0404832255398931e-07,
      "loss": 0.4573,
      "step": 444410
    },
    {
      "epoch": 181.92,
      "grad_norm": 1.57022225856781,
      "learning_rate": 1.040015879027154e-07,
      "loss": 0.4383,
      "step": 444420
    },
    {
      "epoch": 181.92,
      "grad_norm": 2.2261645793914795,
      "learning_rate": 1.0395486351138177e-07,
      "loss": 0.4446,
      "step": 444430
    },
    {
      "epoch": 181.92,
      "grad_norm": 1.956685185432434,
      "learning_rate": 1.0390814938020312e-07,
      "loss": 0.4501,
      "step": 444440
    },
    {
      "epoch": 181.93,
      "grad_norm": 1.8308817148208618,
      "learning_rate": 1.038614455093928e-07,
      "loss": 0.4078,
      "step": 444450
    },
    {
      "epoch": 181.93,
      "grad_norm": 2.377753257751465,
      "learning_rate": 1.0381475189916522e-07,
      "loss": 0.4408,
      "step": 444460
    },
    {
      "epoch": 181.94,
      "grad_norm": 2.0447499752044678,
      "learning_rate": 1.0376806854973402e-07,
      "loss": 0.4507,
      "step": 444470
    },
    {
      "epoch": 181.94,
      "grad_norm": 1.8041449785232544,
      "learning_rate": 1.0372139546131362e-07,
      "loss": 0.4512,
      "step": 444480
    },
    {
      "epoch": 181.94,
      "grad_norm": 1.8295170068740845,
      "learning_rate": 1.0367473263411712e-07,
      "loss": 0.4531,
      "step": 444490
    },
    {
      "epoch": 181.95,
      "grad_norm": 2.0560572147369385,
      "learning_rate": 1.0362808006835815e-07,
      "loss": 0.4479,
      "step": 444500
    },
    {
      "epoch": 181.95,
      "grad_norm": 1.8769994974136353,
      "learning_rate": 1.0358143776425112e-07,
      "loss": 0.4514,
      "step": 444510
    },
    {
      "epoch": 181.96,
      "grad_norm": 2.826505661010742,
      "learning_rate": 1.0353480572200916e-07,
      "loss": 0.4399,
      "step": 444520
    },
    {
      "epoch": 181.96,
      "grad_norm": 1.9582898616790771,
      "learning_rate": 1.0348818394184588e-07,
      "loss": 0.4427,
      "step": 444530
    },
    {
      "epoch": 181.96,
      "grad_norm": 2.9659159183502197,
      "learning_rate": 1.0344157242397544e-07,
      "loss": 0.4435,
      "step": 444540
    },
    {
      "epoch": 181.97,
      "grad_norm": 2.0233123302459717,
      "learning_rate": 1.0339497116861068e-07,
      "loss": 0.4547,
      "step": 444550
    },
    {
      "epoch": 181.97,
      "grad_norm": 2.086757183074951,
      "learning_rate": 1.0334838017596499e-07,
      "loss": 0.4405,
      "step": 444560
    },
    {
      "epoch": 181.98,
      "grad_norm": 2.3562185764312744,
      "learning_rate": 1.033017994462525e-07,
      "loss": 0.4483,
      "step": 444570
    },
    {
      "epoch": 181.98,
      "grad_norm": 2.3200864791870117,
      "learning_rate": 1.0325522897968634e-07,
      "loss": 0.4464,
      "step": 444580
    },
    {
      "epoch": 181.99,
      "grad_norm": 2.2263097763061523,
      "learning_rate": 1.0320866877647988e-07,
      "loss": 0.4447,
      "step": 444590
    },
    {
      "epoch": 181.99,
      "grad_norm": 1.9604029655456543,
      "learning_rate": 1.0316211883684596e-07,
      "loss": 0.4351,
      "step": 444600
    },
    {
      "epoch": 181.99,
      "grad_norm": 2.0106441974639893,
      "learning_rate": 1.031155791609982e-07,
      "loss": 0.4371,
      "step": 444610
    },
    {
      "epoch": 182.0,
      "grad_norm": 2.3345837593078613,
      "learning_rate": 1.0306904974915027e-07,
      "loss": 0.4468,
      "step": 444620
    },
    {
      "epoch": 182.0,
      "eval_loss": 0.445212185382843,
      "eval_runtime": 52.4354,
      "eval_samples_per_second": 65.776,
      "eval_steps_per_second": 8.239,
      "step": 444626
    },
    {
      "epoch": 182.0,
      "grad_norm": 1.910001277923584,
      "learning_rate": 1.030225306015142e-07,
      "loss": 0.4289,
      "step": 444630
    },
    {
      "epoch": 182.01,
      "grad_norm": 2.147451162338257,
      "learning_rate": 1.0297602171830365e-07,
      "loss": 0.4542,
      "step": 444640
    },
    {
      "epoch": 182.01,
      "grad_norm": 2.0167531967163086,
      "learning_rate": 1.0292952309973171e-07,
      "loss": 0.4507,
      "step": 444650
    },
    {
      "epoch": 182.01,
      "grad_norm": 1.7737209796905518,
      "learning_rate": 1.028830347460115e-07,
      "loss": 0.4422,
      "step": 444660
    },
    {
      "epoch": 182.02,
      "grad_norm": 1.9089925289154053,
      "learning_rate": 1.028365566573556e-07,
      "loss": 0.4415,
      "step": 444670
    },
    {
      "epoch": 182.02,
      "grad_norm": 2.155569076538086,
      "learning_rate": 1.0279008883397715e-07,
      "loss": 0.4536,
      "step": 444680
    },
    {
      "epoch": 182.03,
      "grad_norm": 1.9964509010314941,
      "learning_rate": 1.0274363127608897e-07,
      "loss": 0.4217,
      "step": 444690
    },
    {
      "epoch": 182.03,
      "grad_norm": 1.985539436340332,
      "learning_rate": 1.0269718398390418e-07,
      "loss": 0.4374,
      "step": 444700
    },
    {
      "epoch": 182.03,
      "grad_norm": 1.9657180309295654,
      "learning_rate": 1.0265074695763485e-07,
      "loss": 0.4394,
      "step": 444710
    },
    {
      "epoch": 182.04,
      "grad_norm": 1.8025957345962524,
      "learning_rate": 1.0260432019749435e-07,
      "loss": 0.4405,
      "step": 444720
    },
    {
      "epoch": 182.04,
      "grad_norm": 1.924653172492981,
      "learning_rate": 1.0255790370369528e-07,
      "loss": 0.4447,
      "step": 444730
    },
    {
      "epoch": 182.05,
      "grad_norm": 2.3913698196411133,
      "learning_rate": 1.0251149747644996e-07,
      "loss": 0.4765,
      "step": 444740
    },
    {
      "epoch": 182.05,
      "grad_norm": 1.9115294218063354,
      "learning_rate": 1.0246510151597099e-07,
      "loss": 0.4284,
      "step": 444750
    },
    {
      "epoch": 182.05,
      "grad_norm": 2.0194549560546875,
      "learning_rate": 1.0241871582247121e-07,
      "loss": 0.4439,
      "step": 444760
    },
    {
      "epoch": 182.06,
      "grad_norm": 2.4043376445770264,
      "learning_rate": 1.0237234039616295e-07,
      "loss": 0.4297,
      "step": 444770
    },
    {
      "epoch": 182.06,
      "grad_norm": 2.015697717666626,
      "learning_rate": 1.0232597523725855e-07,
      "loss": 0.4468,
      "step": 444780
    },
    {
      "epoch": 182.07,
      "grad_norm": 1.8491618633270264,
      "learning_rate": 1.0227962034597033e-07,
      "loss": 0.4307,
      "step": 444790
    },
    {
      "epoch": 182.07,
      "grad_norm": 2.5203094482421875,
      "learning_rate": 1.0223327572251089e-07,
      "loss": 0.4483,
      "step": 444800
    },
    {
      "epoch": 182.08,
      "grad_norm": 2.610734462738037,
      "learning_rate": 1.0218694136709255e-07,
      "loss": 0.4408,
      "step": 444810
    },
    {
      "epoch": 182.08,
      "grad_norm": 1.8870933055877686,
      "learning_rate": 1.0214061727992738e-07,
      "loss": 0.4501,
      "step": 444820
    },
    {
      "epoch": 182.08,
      "grad_norm": 2.1223480701446533,
      "learning_rate": 1.0209430346122747e-07,
      "loss": 0.4405,
      "step": 444830
    },
    {
      "epoch": 182.09,
      "grad_norm": 2.461757183074951,
      "learning_rate": 1.0204799991120538e-07,
      "loss": 0.4706,
      "step": 444840
    },
    {
      "epoch": 182.09,
      "grad_norm": 2.07041597366333,
      "learning_rate": 1.0200170663007294e-07,
      "loss": 0.45,
      "step": 444850
    },
    {
      "epoch": 182.1,
      "grad_norm": 1.982332706451416,
      "learning_rate": 1.0195542361804221e-07,
      "loss": 0.4405,
      "step": 444860
    },
    {
      "epoch": 182.1,
      "grad_norm": 2.0472466945648193,
      "learning_rate": 1.0190915087532527e-07,
      "loss": 0.4408,
      "step": 444870
    },
    {
      "epoch": 182.1,
      "grad_norm": 2.130455493927002,
      "learning_rate": 1.0186288840213393e-07,
      "loss": 0.4598,
      "step": 444880
    },
    {
      "epoch": 182.11,
      "grad_norm": 2.003582715988159,
      "learning_rate": 1.0181663619868024e-07,
      "loss": 0.4438,
      "step": 444890
    },
    {
      "epoch": 182.11,
      "grad_norm": 2.2182958126068115,
      "learning_rate": 1.017703942651763e-07,
      "loss": 0.4246,
      "step": 444900
    },
    {
      "epoch": 182.12,
      "grad_norm": 2.219011068344116,
      "learning_rate": 1.0172416260183363e-07,
      "loss": 0.4398,
      "step": 444910
    },
    {
      "epoch": 182.12,
      "grad_norm": 2.033904552459717,
      "learning_rate": 1.0167794120886406e-07,
      "loss": 0.453,
      "step": 444920
    },
    {
      "epoch": 182.12,
      "grad_norm": 2.0608322620391846,
      "learning_rate": 1.0163173008647966e-07,
      "loss": 0.4317,
      "step": 444930
    },
    {
      "epoch": 182.13,
      "grad_norm": 2.041327953338623,
      "learning_rate": 1.0158552923489119e-07,
      "loss": 0.4327,
      "step": 444940
    },
    {
      "epoch": 182.13,
      "grad_norm": 2.022416591644287,
      "learning_rate": 1.0153933865431124e-07,
      "loss": 0.4418,
      "step": 444950
    },
    {
      "epoch": 182.14,
      "grad_norm": 2.3470191955566406,
      "learning_rate": 1.014931583449511e-07,
      "loss": 0.4315,
      "step": 444960
    },
    {
      "epoch": 182.14,
      "grad_norm": 2.614483594894409,
      "learning_rate": 1.0144698830702233e-07,
      "loss": 0.4414,
      "step": 444970
    },
    {
      "epoch": 182.14,
      "grad_norm": 1.8274685144424438,
      "learning_rate": 1.0140082854073646e-07,
      "loss": 0.4367,
      "step": 444980
    },
    {
      "epoch": 182.15,
      "grad_norm": 1.833772897720337,
      "learning_rate": 1.013546790463048e-07,
      "loss": 0.4352,
      "step": 444990
    },
    {
      "epoch": 182.15,
      "grad_norm": 2.6263267993927,
      "learning_rate": 1.0130853982393861e-07,
      "loss": 0.4443,
      "step": 445000
    },
    {
      "epoch": 182.16,
      "grad_norm": 2.2248103618621826,
      "learning_rate": 1.0126241087384975e-07,
      "loss": 0.4631,
      "step": 445010
    },
    {
      "epoch": 182.16,
      "grad_norm": 1.982689619064331,
      "learning_rate": 1.012162921962492e-07,
      "loss": 0.4384,
      "step": 445020
    },
    {
      "epoch": 182.17,
      "grad_norm": 2.07145357131958,
      "learning_rate": 1.0117018379134801e-07,
      "loss": 0.4409,
      "step": 445030
    },
    {
      "epoch": 182.17,
      "grad_norm": 1.9109001159667969,
      "learning_rate": 1.0112408565935799e-07,
      "loss": 0.4343,
      "step": 445040
    },
    {
      "epoch": 182.17,
      "grad_norm": 2.106107234954834,
      "learning_rate": 1.010779978004899e-07,
      "loss": 0.4658,
      "step": 445050
    },
    {
      "epoch": 182.18,
      "grad_norm": 2.1516969203948975,
      "learning_rate": 1.0103192021495505e-07,
      "loss": 0.4153,
      "step": 445060
    },
    {
      "epoch": 182.18,
      "grad_norm": 1.8867453336715698,
      "learning_rate": 1.0098585290296418e-07,
      "loss": 0.4586,
      "step": 445070
    },
    {
      "epoch": 182.19,
      "grad_norm": 2.2344095706939697,
      "learning_rate": 1.009397958647286e-07,
      "loss": 0.4487,
      "step": 445080
    },
    {
      "epoch": 182.19,
      "grad_norm": 2.1694655418395996,
      "learning_rate": 1.0089374910045908e-07,
      "loss": 0.4242,
      "step": 445090
    },
    {
      "epoch": 182.19,
      "grad_norm": 2.0005476474761963,
      "learning_rate": 1.0084771261036689e-07,
      "loss": 0.4397,
      "step": 445100
    },
    {
      "epoch": 182.2,
      "grad_norm": 2.075880289077759,
      "learning_rate": 1.0080168639466255e-07,
      "loss": 0.4591,
      "step": 445110
    },
    {
      "epoch": 182.2,
      "grad_norm": 1.9219105243682861,
      "learning_rate": 1.007556704535571e-07,
      "loss": 0.4434,
      "step": 445120
    },
    {
      "epoch": 182.21,
      "grad_norm": 2.1031079292297363,
      "learning_rate": 1.0070966478726129e-07,
      "loss": 0.4533,
      "step": 445130
    },
    {
      "epoch": 182.21,
      "grad_norm": 1.9745659828186035,
      "learning_rate": 1.0066366939598592e-07,
      "loss": 0.4439,
      "step": 445140
    },
    {
      "epoch": 182.21,
      "grad_norm": 2.1005942821502686,
      "learning_rate": 1.0061768427994145e-07,
      "loss": 0.437,
      "step": 445150
    },
    {
      "epoch": 182.22,
      "grad_norm": 2.4991281032562256,
      "learning_rate": 1.0057170943933894e-07,
      "loss": 0.4556,
      "step": 445160
    },
    {
      "epoch": 182.22,
      "grad_norm": 2.066882848739624,
      "learning_rate": 1.005257448743889e-07,
      "loss": 0.4424,
      "step": 445170
    },
    {
      "epoch": 182.23,
      "grad_norm": 1.7670156955718994,
      "learning_rate": 1.0047979058530157e-07,
      "loss": 0.428,
      "step": 445180
    },
    {
      "epoch": 182.23,
      "grad_norm": 1.951829195022583,
      "learning_rate": 1.0043384657228773e-07,
      "loss": 0.4325,
      "step": 445190
    },
    {
      "epoch": 182.23,
      "grad_norm": 2.3110296726226807,
      "learning_rate": 1.0038791283555814e-07,
      "loss": 0.4454,
      "step": 445200
    },
    {
      "epoch": 182.24,
      "grad_norm": 1.8867528438568115,
      "learning_rate": 1.0034198937532252e-07,
      "loss": 0.428,
      "step": 445210
    },
    {
      "epoch": 182.24,
      "grad_norm": 2.22428560256958,
      "learning_rate": 1.0029607619179139e-07,
      "loss": 0.429,
      "step": 445220
    },
    {
      "epoch": 182.25,
      "grad_norm": 2.050731658935547,
      "learning_rate": 1.0025017328517551e-07,
      "loss": 0.4379,
      "step": 445230
    },
    {
      "epoch": 182.25,
      "grad_norm": 1.9257383346557617,
      "learning_rate": 1.0020428065568463e-07,
      "loss": 0.4537,
      "step": 445240
    },
    {
      "epoch": 182.26,
      "grad_norm": 2.5649237632751465,
      "learning_rate": 1.0015839830352923e-07,
      "loss": 0.4597,
      "step": 445250
    },
    {
      "epoch": 182.26,
      "grad_norm": 2.2675869464874268,
      "learning_rate": 1.0011252622891983e-07,
      "loss": 0.4395,
      "step": 445260
    },
    {
      "epoch": 182.26,
      "grad_norm": 1.8831557035446167,
      "learning_rate": 1.0006666443206591e-07,
      "loss": 0.4599,
      "step": 445270
    },
    {
      "epoch": 182.27,
      "grad_norm": 2.1559243202209473,
      "learning_rate": 1.0002081291317796e-07,
      "loss": 0.4495,
      "step": 445280
    },
    {
      "epoch": 182.27,
      "grad_norm": 2.1598780155181885,
      "learning_rate": 9.997497167246599e-08,
      "loss": 0.4342,
      "step": 445290
    },
    {
      "epoch": 182.28,
      "grad_norm": 2.6249003410339355,
      "learning_rate": 9.99291407101397e-08,
      "loss": 0.4437,
      "step": 445300
    },
    {
      "epoch": 182.28,
      "grad_norm": 1.8043725490570068,
      "learning_rate": 9.988332002640962e-08,
      "loss": 0.4548,
      "step": 445310
    },
    {
      "epoch": 182.28,
      "grad_norm": 1.7601861953735352,
      "learning_rate": 9.983750962148469e-08,
      "loss": 0.4251,
      "step": 445320
    },
    {
      "epoch": 182.29,
      "grad_norm": 2.6733434200286865,
      "learning_rate": 9.97917094955762e-08,
      "loss": 0.457,
      "step": 445330
    },
    {
      "epoch": 182.29,
      "grad_norm": 1.9971442222595215,
      "learning_rate": 9.974591964889257e-08,
      "loss": 0.4177,
      "step": 445340
    },
    {
      "epoch": 182.3,
      "grad_norm": 2.119767904281616,
      "learning_rate": 9.970014008164433e-08,
      "loss": 0.4258,
      "step": 445350
    },
    {
      "epoch": 182.3,
      "grad_norm": 1.8103386163711548,
      "learning_rate": 9.965437079404093e-08,
      "loss": 0.433,
      "step": 445360
    },
    {
      "epoch": 182.3,
      "grad_norm": 2.251495599746704,
      "learning_rate": 9.960861178629209e-08,
      "loss": 0.4518,
      "step": 445370
    },
    {
      "epoch": 182.31,
      "grad_norm": 1.9474928379058838,
      "learning_rate": 9.95628630586073e-08,
      "loss": 0.4552,
      "step": 445380
    },
    {
      "epoch": 182.31,
      "grad_norm": 9.206365585327148,
      "learning_rate": 9.951712461119628e-08,
      "loss": 0.4549,
      "step": 445390
    },
    {
      "epoch": 182.32,
      "grad_norm": 1.8844447135925293,
      "learning_rate": 9.947139644426849e-08,
      "loss": 0.4406,
      "step": 445400
    },
    {
      "epoch": 182.32,
      "grad_norm": 2.2761197090148926,
      "learning_rate": 9.942567855803314e-08,
      "loss": 0.4567,
      "step": 445410
    },
    {
      "epoch": 182.33,
      "grad_norm": 1.9123725891113281,
      "learning_rate": 9.937997095270022e-08,
      "loss": 0.4357,
      "step": 445420
    },
    {
      "epoch": 182.33,
      "grad_norm": 1.937098503112793,
      "learning_rate": 9.933427362847894e-08,
      "loss": 0.4293,
      "step": 445430
    },
    {
      "epoch": 182.33,
      "grad_norm": 2.3208954334259033,
      "learning_rate": 9.928858658557826e-08,
      "loss": 0.4409,
      "step": 445440
    },
    {
      "epoch": 182.34,
      "grad_norm": 1.7955337762832642,
      "learning_rate": 9.924290982420788e-08,
      "loss": 0.4231,
      "step": 445450
    },
    {
      "epoch": 182.34,
      "grad_norm": 2.3876397609710693,
      "learning_rate": 9.919724334457677e-08,
      "loss": 0.4535,
      "step": 445460
    },
    {
      "epoch": 182.35,
      "grad_norm": 1.7280802726745605,
      "learning_rate": 9.915158714689413e-08,
      "loss": 0.4381,
      "step": 445470
    },
    {
      "epoch": 182.35,
      "grad_norm": 2.0255284309387207,
      "learning_rate": 9.910594123136943e-08,
      "loss": 0.4402,
      "step": 445480
    },
    {
      "epoch": 182.35,
      "grad_norm": 1.8267319202423096,
      "learning_rate": 9.906030559821135e-08,
      "loss": 0.4483,
      "step": 445490
    },
    {
      "epoch": 182.36,
      "grad_norm": 1.8032000064849854,
      "learning_rate": 9.901468024762936e-08,
      "loss": 0.4403,
      "step": 445500
    },
    {
      "epoch": 182.36,
      "grad_norm": 2.4314143657684326,
      "learning_rate": 9.896906517983215e-08,
      "loss": 0.4619,
      "step": 445510
    },
    {
      "epoch": 182.37,
      "grad_norm": 2.305020570755005,
      "learning_rate": 9.892346039502867e-08,
      "loss": 0.4511,
      "step": 445520
    },
    {
      "epoch": 182.37,
      "grad_norm": 2.258849620819092,
      "learning_rate": 9.887786589342786e-08,
      "loss": 0.4468,
      "step": 445530
    },
    {
      "epoch": 182.37,
      "grad_norm": 2.1209428310394287,
      "learning_rate": 9.883228167523869e-08,
      "loss": 0.4616,
      "step": 445540
    },
    {
      "epoch": 182.38,
      "grad_norm": 1.8233755826950073,
      "learning_rate": 9.878670774066981e-08,
      "loss": 0.4408,
      "step": 445550
    },
    {
      "epoch": 182.38,
      "grad_norm": 2.170165538787842,
      "learning_rate": 9.874114408993019e-08,
      "loss": 0.4695,
      "step": 445560
    },
    {
      "epoch": 182.39,
      "grad_norm": 2.2334930896759033,
      "learning_rate": 9.869559072322851e-08,
      "loss": 0.4287,
      "step": 445570
    },
    {
      "epoch": 182.39,
      "grad_norm": 1.989546298980713,
      "learning_rate": 9.865004764077347e-08,
      "loss": 0.4391,
      "step": 445580
    },
    {
      "epoch": 182.39,
      "grad_norm": 2.0494205951690674,
      "learning_rate": 9.86045148427735e-08,
      "loss": 0.4624,
      "step": 445590
    },
    {
      "epoch": 182.4,
      "grad_norm": 2.122865915298462,
      "learning_rate": 9.855899232943725e-08,
      "loss": 0.45,
      "step": 445600
    },
    {
      "epoch": 182.4,
      "grad_norm": 2.669891119003296,
      "learning_rate": 9.851348010097345e-08,
      "loss": 0.4288,
      "step": 445610
    },
    {
      "epoch": 182.41,
      "grad_norm": 1.877832293510437,
      "learning_rate": 9.846797815759025e-08,
      "loss": 0.4364,
      "step": 445620
    },
    {
      "epoch": 182.41,
      "grad_norm": 2.336427927017212,
      "learning_rate": 9.842248649949633e-08,
      "loss": 0.4376,
      "step": 445630
    },
    {
      "epoch": 182.42,
      "grad_norm": 2.374559164047241,
      "learning_rate": 9.837700512690038e-08,
      "loss": 0.442,
      "step": 445640
    },
    {
      "epoch": 182.42,
      "grad_norm": 2.12427020072937,
      "learning_rate": 9.833153404001006e-08,
      "loss": 0.4394,
      "step": 445650
    },
    {
      "epoch": 182.42,
      "grad_norm": 1.7669955492019653,
      "learning_rate": 9.828607323903378e-08,
      "loss": 0.4587,
      "step": 445660
    },
    {
      "epoch": 182.43,
      "grad_norm": 1.8620076179504395,
      "learning_rate": 9.824062272418025e-08,
      "loss": 0.4472,
      "step": 445670
    },
    {
      "epoch": 182.43,
      "grad_norm": 1.736961841583252,
      "learning_rate": 9.819518249565734e-08,
      "loss": 0.4449,
      "step": 445680
    },
    {
      "epoch": 182.44,
      "grad_norm": 1.8467168807983398,
      "learning_rate": 9.814975255367325e-08,
      "loss": 0.434,
      "step": 445690
    },
    {
      "epoch": 182.44,
      "grad_norm": 1.727454423904419,
      "learning_rate": 9.810433289843587e-08,
      "loss": 0.4616,
      "step": 445700
    },
    {
      "epoch": 182.44,
      "grad_norm": 2.3220629692077637,
      "learning_rate": 9.805892353015365e-08,
      "loss": 0.4459,
      "step": 445710
    },
    {
      "epoch": 182.45,
      "grad_norm": 2.307785749435425,
      "learning_rate": 9.801352444903472e-08,
      "loss": 0.4428,
      "step": 445720
    },
    {
      "epoch": 182.45,
      "grad_norm": 2.1546478271484375,
      "learning_rate": 9.796813565528674e-08,
      "loss": 0.4519,
      "step": 445730
    },
    {
      "epoch": 182.46,
      "grad_norm": 2.089479923248291,
      "learning_rate": 9.792275714911739e-08,
      "loss": 0.464,
      "step": 445740
    },
    {
      "epoch": 182.46,
      "grad_norm": 1.9429757595062256,
      "learning_rate": 9.787738893073504e-08,
      "loss": 0.4443,
      "step": 445750
    },
    {
      "epoch": 182.46,
      "grad_norm": 1.9797863960266113,
      "learning_rate": 9.783203100034737e-08,
      "loss": 0.4372,
      "step": 445760
    },
    {
      "epoch": 182.47,
      "grad_norm": 1.7717504501342773,
      "learning_rate": 9.778668335816203e-08,
      "loss": 0.438,
      "step": 445770
    },
    {
      "epoch": 182.47,
      "grad_norm": 2.0558862686157227,
      "learning_rate": 9.774134600438717e-08,
      "loss": 0.4492,
      "step": 445780
    },
    {
      "epoch": 182.48,
      "grad_norm": 2.241844654083252,
      "learning_rate": 9.769601893922967e-08,
      "loss": 0.4352,
      "step": 445790
    },
    {
      "epoch": 182.48,
      "grad_norm": 1.9099560976028442,
      "learning_rate": 9.76507021628977e-08,
      "loss": 0.4433,
      "step": 445800
    },
    {
      "epoch": 182.48,
      "grad_norm": 2.0525944232940674,
      "learning_rate": 9.760539567559887e-08,
      "loss": 0.4385,
      "step": 445810
    },
    {
      "epoch": 182.49,
      "grad_norm": 2.2778468132019043,
      "learning_rate": 9.756009947754035e-08,
      "loss": 0.4509,
      "step": 445820
    },
    {
      "epoch": 182.49,
      "grad_norm": 2.0485575199127197,
      "learning_rate": 9.751481356893003e-08,
      "loss": 0.4223,
      "step": 445830
    },
    {
      "epoch": 182.5,
      "grad_norm": 2.113203525543213,
      "learning_rate": 9.74695379499753e-08,
      "loss": 0.4248,
      "step": 445840
    },
    {
      "epoch": 182.5,
      "grad_norm": 2.370990514755249,
      "learning_rate": 9.742427262088353e-08,
      "loss": 0.4491,
      "step": 445850
    },
    {
      "epoch": 182.51,
      "grad_norm": 2.158123254776001,
      "learning_rate": 9.737901758186187e-08,
      "loss": 0.4369,
      "step": 445860
    },
    {
      "epoch": 182.51,
      "grad_norm": 2.1961593627929688,
      "learning_rate": 9.733377283311769e-08,
      "loss": 0.4559,
      "step": 445870
    },
    {
      "epoch": 182.51,
      "grad_norm": 2.3214645385742188,
      "learning_rate": 9.728853837485837e-08,
      "loss": 0.4362,
      "step": 445880
    },
    {
      "epoch": 182.52,
      "grad_norm": 2.2207608222961426,
      "learning_rate": 9.724331420729106e-08,
      "loss": 0.4584,
      "step": 445890
    },
    {
      "epoch": 182.52,
      "grad_norm": 1.6874699592590332,
      "learning_rate": 9.719810033062312e-08,
      "loss": 0.4158,
      "step": 445900
    },
    {
      "epoch": 182.53,
      "grad_norm": 2.401362657546997,
      "learning_rate": 9.715289674506171e-08,
      "loss": 0.4389,
      "step": 445910
    },
    {
      "epoch": 182.53,
      "grad_norm": 2.4592840671539307,
      "learning_rate": 9.710770345081367e-08,
      "loss": 0.4532,
      "step": 445920
    },
    {
      "epoch": 182.53,
      "grad_norm": 2.2286367416381836,
      "learning_rate": 9.706252044808585e-08,
      "loss": 0.4595,
      "step": 445930
    },
    {
      "epoch": 182.54,
      "grad_norm": 1.9859099388122559,
      "learning_rate": 9.701734773708566e-08,
      "loss": 0.4615,
      "step": 445940
    },
    {
      "epoch": 182.54,
      "grad_norm": 2.020441770553589,
      "learning_rate": 9.697218531801946e-08,
      "loss": 0.4458,
      "step": 445950
    },
    {
      "epoch": 182.55,
      "grad_norm": 1.8383737802505493,
      "learning_rate": 9.692703319109461e-08,
      "loss": 0.4476,
      "step": 445960
    },
    {
      "epoch": 182.55,
      "grad_norm": 2.200706720352173,
      "learning_rate": 9.688189135651798e-08,
      "loss": 0.4747,
      "step": 445970
    },
    {
      "epoch": 182.55,
      "grad_norm": 1.8133442401885986,
      "learning_rate": 9.683675981449592e-08,
      "loss": 0.4307,
      "step": 445980
    },
    {
      "epoch": 182.56,
      "grad_norm": 2.210573434829712,
      "learning_rate": 9.679163856523555e-08,
      "loss": 0.4611,
      "step": 445990
    },
    {
      "epoch": 182.56,
      "grad_norm": 1.873715877532959,
      "learning_rate": 9.674652760894349e-08,
      "loss": 0.4458,
      "step": 446000
    },
    {
      "epoch": 182.57,
      "grad_norm": 1.8691558837890625,
      "learning_rate": 9.670142694582633e-08,
      "loss": 0.4501,
      "step": 446010
    },
    {
      "epoch": 182.57,
      "grad_norm": 1.9008545875549316,
      "learning_rate": 9.665633657609068e-08,
      "loss": 0.4391,
      "step": 446020
    },
    {
      "epoch": 182.57,
      "grad_norm": 2.424670696258545,
      "learning_rate": 9.66112564999429e-08,
      "loss": 0.4502,
      "step": 446030
    },
    {
      "epoch": 182.58,
      "grad_norm": 2.1092050075531006,
      "learning_rate": 9.656618671758983e-08,
      "loss": 0.449,
      "step": 446040
    },
    {
      "epoch": 182.58,
      "grad_norm": 1.9796757698059082,
      "learning_rate": 9.652112722923784e-08,
      "loss": 0.4716,
      "step": 446050
    },
    {
      "epoch": 182.59,
      "grad_norm": 2.1159088611602783,
      "learning_rate": 9.647607803509325e-08,
      "loss": 0.4374,
      "step": 446060
    },
    {
      "epoch": 182.59,
      "grad_norm": 2.1121833324432373,
      "learning_rate": 9.643103913536242e-08,
      "loss": 0.4375,
      "step": 446070
    },
    {
      "epoch": 182.6,
      "grad_norm": 2.5467355251312256,
      "learning_rate": 9.638601053025197e-08,
      "loss": 0.4539,
      "step": 446080
    },
    {
      "epoch": 182.6,
      "grad_norm": 2.09549880027771,
      "learning_rate": 9.634099221996769e-08,
      "loss": 0.4433,
      "step": 446090
    },
    {
      "epoch": 182.6,
      "grad_norm": 1.9997484683990479,
      "learning_rate": 9.629598420471595e-08,
      "loss": 0.4553,
      "step": 446100
    },
    {
      "epoch": 182.61,
      "grad_norm": 2.001988649368286,
      "learning_rate": 9.62509864847031e-08,
      "loss": 0.4693,
      "step": 446110
    },
    {
      "epoch": 182.61,
      "grad_norm": 1.8422046899795532,
      "learning_rate": 9.620599906013521e-08,
      "loss": 0.4649,
      "step": 446120
    },
    {
      "epoch": 182.62,
      "grad_norm": 1.8682984113693237,
      "learning_rate": 9.61610219312181e-08,
      "loss": 0.4365,
      "step": 446130
    },
    {
      "epoch": 182.62,
      "grad_norm": 2.417848825454712,
      "learning_rate": 9.611605509815815e-08,
      "loss": 0.466,
      "step": 446140
    },
    {
      "epoch": 182.62,
      "grad_norm": 2.2285995483398438,
      "learning_rate": 9.607109856116114e-08,
      "loss": 0.4346,
      "step": 446150
    },
    {
      "epoch": 182.63,
      "grad_norm": 2.366806983947754,
      "learning_rate": 9.602615232043318e-08,
      "loss": 0.4542,
      "step": 446160
    },
    {
      "epoch": 182.63,
      "grad_norm": 2.0633020401000977,
      "learning_rate": 9.598121637618008e-08,
      "loss": 0.4523,
      "step": 446170
    },
    {
      "epoch": 182.64,
      "grad_norm": 2.3230345249176025,
      "learning_rate": 9.59362907286077e-08,
      "loss": 0.4282,
      "step": 446180
    },
    {
      "epoch": 182.64,
      "grad_norm": 1.9740073680877686,
      "learning_rate": 9.589137537792157e-08,
      "loss": 0.4393,
      "step": 446190
    },
    {
      "epoch": 182.64,
      "grad_norm": 2.362489938735962,
      "learning_rate": 9.584647032432804e-08,
      "loss": 0.4661,
      "step": 446200
    },
    {
      "epoch": 182.65,
      "grad_norm": 1.8803656101226807,
      "learning_rate": 9.580157556803216e-08,
      "loss": 0.4414,
      "step": 446210
    },
    {
      "epoch": 182.65,
      "grad_norm": 1.8835214376449585,
      "learning_rate": 9.575669110924002e-08,
      "loss": 0.4437,
      "step": 446220
    },
    {
      "epoch": 182.66,
      "grad_norm": 7.1057281494140625,
      "learning_rate": 9.571181694815718e-08,
      "loss": 0.4486,
      "step": 446230
    },
    {
      "epoch": 182.66,
      "grad_norm": 2.163090229034424,
      "learning_rate": 9.566695308498894e-08,
      "loss": 0.4481,
      "step": 446240
    },
    {
      "epoch": 182.66,
      "grad_norm": 2.0848660469055176,
      "learning_rate": 9.562209951994114e-08,
      "loss": 0.4547,
      "step": 446250
    },
    {
      "epoch": 182.67,
      "grad_norm": 1.688430905342102,
      "learning_rate": 9.557725625321854e-08,
      "loss": 0.4436,
      "step": 446260
    },
    {
      "epoch": 182.67,
      "grad_norm": 1.9563270807266235,
      "learning_rate": 9.55324232850275e-08,
      "loss": 0.4483,
      "step": 446270
    },
    {
      "epoch": 182.68,
      "grad_norm": 1.9471060037612915,
      "learning_rate": 9.548760061557307e-08,
      "loss": 0.419,
      "step": 446280
    },
    {
      "epoch": 182.68,
      "grad_norm": 2.1253793239593506,
      "learning_rate": 9.544278824506028e-08,
      "loss": 0.4336,
      "step": 446290
    },
    {
      "epoch": 182.69,
      "grad_norm": 2.0484132766723633,
      "learning_rate": 9.539798617369471e-08,
      "loss": 0.4175,
      "step": 446300
    },
    {
      "epoch": 182.69,
      "grad_norm": 2.115499496459961,
      "learning_rate": 9.535319440168165e-08,
      "loss": 0.4529,
      "step": 446310
    },
    {
      "epoch": 182.69,
      "grad_norm": 1.9981975555419922,
      "learning_rate": 9.530841292922616e-08,
      "loss": 0.4311,
      "step": 446320
    },
    {
      "epoch": 182.7,
      "grad_norm": 4.257490158081055,
      "learning_rate": 9.526364175653327e-08,
      "loss": 0.4532,
      "step": 446330
    },
    {
      "epoch": 182.7,
      "grad_norm": 1.9325889348983765,
      "learning_rate": 9.521888088380829e-08,
      "loss": 0.4119,
      "step": 446340
    },
    {
      "epoch": 182.71,
      "grad_norm": 2.100038528442383,
      "learning_rate": 9.5174130311256e-08,
      "loss": 0.4468,
      "step": 446350
    },
    {
      "epoch": 182.71,
      "grad_norm": 2.3499672412872314,
      "learning_rate": 9.512939003908172e-08,
      "loss": 0.4467,
      "step": 446360
    },
    {
      "epoch": 182.71,
      "grad_norm": 2.095984697341919,
      "learning_rate": 9.508466006749021e-08,
      "loss": 0.4526,
      "step": 446370
    },
    {
      "epoch": 182.72,
      "grad_norm": 1.9334197044372559,
      "learning_rate": 9.503994039668601e-08,
      "loss": 0.4318,
      "step": 446380
    },
    {
      "epoch": 182.72,
      "grad_norm": 2.0676770210266113,
      "learning_rate": 9.499523102687468e-08,
      "loss": 0.4532,
      "step": 446390
    },
    {
      "epoch": 182.73,
      "grad_norm": 2.2017345428466797,
      "learning_rate": 9.495053195826049e-08,
      "loss": 0.4382,
      "step": 446400
    },
    {
      "epoch": 182.73,
      "grad_norm": 2.1660895347595215,
      "learning_rate": 9.490584319104847e-08,
      "loss": 0.445,
      "step": 446410
    },
    {
      "epoch": 182.73,
      "grad_norm": 2.0916974544525146,
      "learning_rate": 9.486116472544316e-08,
      "loss": 0.454,
      "step": 446420
    },
    {
      "epoch": 182.74,
      "grad_norm": 1.7959028482437134,
      "learning_rate": 9.481649656164934e-08,
      "loss": 0.4437,
      "step": 446430
    },
    {
      "epoch": 182.74,
      "grad_norm": 2.1772842407226562,
      "learning_rate": 9.477183869987179e-08,
      "loss": 0.4523,
      "step": 446440
    },
    {
      "epoch": 182.75,
      "grad_norm": 2.0612239837646484,
      "learning_rate": 9.47271911403145e-08,
      "loss": 0.4456,
      "step": 446450
    },
    {
      "epoch": 182.75,
      "grad_norm": 1.9486969709396362,
      "learning_rate": 9.468255388318255e-08,
      "loss": 0.4414,
      "step": 446460
    },
    {
      "epoch": 182.75,
      "grad_norm": 4.871830463409424,
      "learning_rate": 9.463792692868044e-08,
      "loss": 0.4486,
      "step": 446470
    },
    {
      "epoch": 182.76,
      "grad_norm": 1.8857996463775635,
      "learning_rate": 9.459331027701243e-08,
      "loss": 0.4317,
      "step": 446480
    },
    {
      "epoch": 182.76,
      "grad_norm": 3.0866270065307617,
      "learning_rate": 9.454870392838279e-08,
      "loss": 0.4481,
      "step": 446490
    },
    {
      "epoch": 182.77,
      "grad_norm": 1.9197912216186523,
      "learning_rate": 9.450410788299631e-08,
      "loss": 0.4338,
      "step": 446500
    },
    {
      "epoch": 182.77,
      "grad_norm": 2.251833438873291,
      "learning_rate": 9.445952214105646e-08,
      "loss": 0.4428,
      "step": 446510
    },
    {
      "epoch": 182.78,
      "grad_norm": 1.9574763774871826,
      "learning_rate": 9.441494670276803e-08,
      "loss": 0.4368,
      "step": 446520
    },
    {
      "epoch": 182.78,
      "grad_norm": 2.0446503162384033,
      "learning_rate": 9.43703815683353e-08,
      "loss": 0.422,
      "step": 446530
    },
    {
      "epoch": 182.78,
      "grad_norm": 1.989709734916687,
      "learning_rate": 9.432582673796198e-08,
      "loss": 0.435,
      "step": 446540
    },
    {
      "epoch": 182.79,
      "grad_norm": 2.0345215797424316,
      "learning_rate": 9.428128221185263e-08,
      "loss": 0.436,
      "step": 446550
    },
    {
      "epoch": 182.79,
      "grad_norm": 1.9045984745025635,
      "learning_rate": 9.423674799021097e-08,
      "loss": 0.4581,
      "step": 446560
    },
    {
      "epoch": 182.8,
      "grad_norm": 1.8306668996810913,
      "learning_rate": 9.419222407324126e-08,
      "loss": 0.4272,
      "step": 446570
    },
    {
      "epoch": 182.8,
      "grad_norm": 2.470130443572998,
      "learning_rate": 9.414771046114726e-08,
      "loss": 0.4516,
      "step": 446580
    },
    {
      "epoch": 182.8,
      "grad_norm": 2.2629194259643555,
      "learning_rate": 9.410320715413297e-08,
      "loss": 0.4457,
      "step": 446590
    },
    {
      "epoch": 182.81,
      "grad_norm": 2.1575984954833984,
      "learning_rate": 9.405871415240213e-08,
      "loss": 0.4578,
      "step": 446600
    },
    {
      "epoch": 182.81,
      "grad_norm": 2.1302287578582764,
      "learning_rate": 9.401423145615873e-08,
      "loss": 0.4293,
      "step": 446610
    },
    {
      "epoch": 182.82,
      "grad_norm": 1.6967827081680298,
      "learning_rate": 9.396975906560654e-08,
      "loss": 0.4607,
      "step": 446620
    },
    {
      "epoch": 182.82,
      "grad_norm": 1.7480257749557495,
      "learning_rate": 9.392529698094928e-08,
      "loss": 0.4331,
      "step": 446630
    },
    {
      "epoch": 182.82,
      "grad_norm": 2.62111496925354,
      "learning_rate": 9.388084520239045e-08,
      "loss": 0.4564,
      "step": 446640
    },
    {
      "epoch": 182.83,
      "grad_norm": 1.7376883029937744,
      "learning_rate": 9.383640373013378e-08,
      "loss": 0.4461,
      "step": 446650
    },
    {
      "epoch": 182.83,
      "grad_norm": 1.8284319639205933,
      "learning_rate": 9.379197256438277e-08,
      "loss": 0.4372,
      "step": 446660
    },
    {
      "epoch": 182.84,
      "grad_norm": 2.3325788974761963,
      "learning_rate": 9.374755170534114e-08,
      "loss": 0.4527,
      "step": 446670
    },
    {
      "epoch": 182.84,
      "grad_norm": 1.958590030670166,
      "learning_rate": 9.370314115321212e-08,
      "loss": 0.4365,
      "step": 446680
    },
    {
      "epoch": 182.84,
      "grad_norm": 2.471179962158203,
      "learning_rate": 9.365874090819947e-08,
      "loss": 0.4353,
      "step": 446690
    },
    {
      "epoch": 182.85,
      "grad_norm": 2.5197441577911377,
      "learning_rate": 9.36143509705064e-08,
      "loss": 0.4518,
      "step": 446700
    },
    {
      "epoch": 182.85,
      "grad_norm": 1.9627662897109985,
      "learning_rate": 9.356997134033614e-08,
      "loss": 0.4463,
      "step": 446710
    },
    {
      "epoch": 182.86,
      "grad_norm": 2.019256353378296,
      "learning_rate": 9.352560201789242e-08,
      "loss": 0.4415,
      "step": 446720
    },
    {
      "epoch": 182.86,
      "grad_norm": 2.4619288444519043,
      "learning_rate": 9.348124300337795e-08,
      "loss": 0.4483,
      "step": 446730
    },
    {
      "epoch": 182.87,
      "grad_norm": 1.9718462228775024,
      "learning_rate": 9.343689429699595e-08,
      "loss": 0.4412,
      "step": 446740
    },
    {
      "epoch": 182.87,
      "grad_norm": 2.0902345180511475,
      "learning_rate": 9.339255589895017e-08,
      "loss": 0.4514,
      "step": 446750
    },
    {
      "epoch": 182.87,
      "grad_norm": 1.959421157836914,
      "learning_rate": 9.334822780944304e-08,
      "loss": 0.4408,
      "step": 446760
    },
    {
      "epoch": 182.88,
      "grad_norm": 1.8638818264007568,
      "learning_rate": 9.330391002867806e-08,
      "loss": 0.4446,
      "step": 446770
    },
    {
      "epoch": 182.88,
      "grad_norm": 1.7284159660339355,
      "learning_rate": 9.325960255685792e-08,
      "loss": 0.4381,
      "step": 446780
    },
    {
      "epoch": 182.89,
      "grad_norm": 2.646731376647949,
      "learning_rate": 9.321530539418585e-08,
      "loss": 0.4483,
      "step": 446790
    },
    {
      "epoch": 182.89,
      "grad_norm": 2.0299127101898193,
      "learning_rate": 9.317101854086481e-08,
      "loss": 0.4341,
      "step": 446800
    },
    {
      "epoch": 182.89,
      "grad_norm": 1.7544695138931274,
      "learning_rate": 9.312674199709776e-08,
      "loss": 0.4405,
      "step": 446810
    },
    {
      "epoch": 182.9,
      "grad_norm": 2.4322285652160645,
      "learning_rate": 9.308247576308688e-08,
      "loss": 0.4519,
      "step": 446820
    },
    {
      "epoch": 182.9,
      "grad_norm": 1.9660159349441528,
      "learning_rate": 9.303821983903513e-08,
      "loss": 0.4387,
      "step": 446830
    },
    {
      "epoch": 182.91,
      "grad_norm": 1.8449851274490356,
      "learning_rate": 9.299397422514602e-08,
      "loss": 0.4294,
      "step": 446840
    },
    {
      "epoch": 182.91,
      "grad_norm": 2.133923053741455,
      "learning_rate": 9.294973892162144e-08,
      "loss": 0.4474,
      "step": 446850
    },
    {
      "epoch": 182.91,
      "grad_norm": 2.330897331237793,
      "learning_rate": 9.290551392866436e-08,
      "loss": 0.4545,
      "step": 446860
    },
    {
      "epoch": 182.92,
      "grad_norm": 2.2122037410736084,
      "learning_rate": 9.286129924647723e-08,
      "loss": 0.4218,
      "step": 446870
    },
    {
      "epoch": 182.92,
      "grad_norm": 2.347776174545288,
      "learning_rate": 9.281709487526249e-08,
      "loss": 0.4389,
      "step": 446880
    },
    {
      "epoch": 182.93,
      "grad_norm": 2.255281925201416,
      "learning_rate": 9.277290081522309e-08,
      "loss": 0.4494,
      "step": 446890
    },
    {
      "epoch": 182.93,
      "grad_norm": 1.7769075632095337,
      "learning_rate": 9.272871706656098e-08,
      "loss": 0.4523,
      "step": 446900
    },
    {
      "epoch": 182.93,
      "grad_norm": 2.270601987838745,
      "learning_rate": 9.268454362947883e-08,
      "loss": 0.4245,
      "step": 446910
    },
    {
      "epoch": 182.94,
      "grad_norm": 1.9910513162612915,
      "learning_rate": 9.264038050417884e-08,
      "loss": 0.436,
      "step": 446920
    },
    {
      "epoch": 182.94,
      "grad_norm": 1.9113341569900513,
      "learning_rate": 9.259622769086346e-08,
      "loss": 0.4499,
      "step": 446930
    },
    {
      "epoch": 182.95,
      "grad_norm": 1.9401108026504517,
      "learning_rate": 9.255208518973486e-08,
      "loss": 0.4519,
      "step": 446940
    },
    {
      "epoch": 182.95,
      "grad_norm": 2.1438863277435303,
      "learning_rate": 9.250795300099574e-08,
      "loss": 0.4397,
      "step": 446950
    },
    {
      "epoch": 182.96,
      "grad_norm": 1.8924869298934937,
      "learning_rate": 9.246383112484724e-08,
      "loss": 0.4518,
      "step": 446960
    },
    {
      "epoch": 182.96,
      "grad_norm": 2.020026206970215,
      "learning_rate": 9.241971956149207e-08,
      "loss": 0.4401,
      "step": 446970
    },
    {
      "epoch": 182.96,
      "grad_norm": 2.0390985012054443,
      "learning_rate": 9.237561831113215e-08,
      "loss": 0.435,
      "step": 446980
    },
    {
      "epoch": 182.97,
      "grad_norm": 1.8931293487548828,
      "learning_rate": 9.233152737396993e-08,
      "loss": 0.4619,
      "step": 446990
    },
    {
      "epoch": 182.97,
      "grad_norm": 2.278527021408081,
      "learning_rate": 9.22874467502068e-08,
      "loss": 0.4424,
      "step": 447000
    },
    {
      "epoch": 182.98,
      "grad_norm": 1.7831699848175049,
      "learning_rate": 9.224337644004493e-08,
      "loss": 0.442,
      "step": 447010
    },
    {
      "epoch": 182.98,
      "grad_norm": 1.9135819673538208,
      "learning_rate": 9.219931644368602e-08,
      "loss": 0.4407,
      "step": 447020
    },
    {
      "epoch": 182.98,
      "grad_norm": 2.374234914779663,
      "learning_rate": 9.215526676133248e-08,
      "loss": 0.4665,
      "step": 447030
    },
    {
      "epoch": 182.99,
      "grad_norm": 2.0673532485961914,
      "learning_rate": 9.211122739318545e-08,
      "loss": 0.45,
      "step": 447040
    },
    {
      "epoch": 182.99,
      "grad_norm": 2.043626546859741,
      "learning_rate": 9.206719833944713e-08,
      "loss": 0.4538,
      "step": 447050
    },
    {
      "epoch": 183.0,
      "grad_norm": 2.1222238540649414,
      "learning_rate": 9.202317960031891e-08,
      "loss": 0.4394,
      "step": 447060
    },
    {
      "epoch": 183.0,
      "eval_loss": 0.44402626156806946,
      "eval_runtime": 52.9508,
      "eval_samples_per_second": 65.136,
      "eval_steps_per_second": 8.159,
      "step": 447069
    },
    {
      "epoch": 183.0,
      "grad_norm": 1.998214840888977,
      "learning_rate": 9.197917117600272e-08,
      "loss": 0.4428,
      "step": 447070
    },
    {
      "epoch": 183.0,
      "grad_norm": 2.193391799926758,
      "learning_rate": 9.193517306669995e-08,
      "loss": 0.4369,
      "step": 447080
    },
    {
      "epoch": 183.01,
      "grad_norm": 2.296348810195923,
      "learning_rate": 9.189118527261202e-08,
      "loss": 0.4579,
      "step": 447090
    },
    {
      "epoch": 183.01,
      "grad_norm": 1.8185275793075562,
      "learning_rate": 9.184720779394058e-08,
      "loss": 0.4447,
      "step": 447100
    },
    {
      "epoch": 183.02,
      "grad_norm": 1.8552820682525635,
      "learning_rate": 9.180324063088728e-08,
      "loss": 0.4313,
      "step": 447110
    },
    {
      "epoch": 183.02,
      "grad_norm": 2.6969447135925293,
      "learning_rate": 9.175928378365302e-08,
      "loss": 0.4179,
      "step": 447120
    },
    {
      "epoch": 183.02,
      "grad_norm": 1.9344803094863892,
      "learning_rate": 9.171533725243971e-08,
      "loss": 0.4375,
      "step": 447130
    },
    {
      "epoch": 183.03,
      "grad_norm": 2.0893232822418213,
      "learning_rate": 9.167140103744824e-08,
      "loss": 0.4443,
      "step": 447140
    },
    {
      "epoch": 183.03,
      "grad_norm": 1.9106111526489258,
      "learning_rate": 9.162747513887999e-08,
      "loss": 0.432,
      "step": 447150
    },
    {
      "epoch": 183.04,
      "grad_norm": 2.0370492935180664,
      "learning_rate": 9.15835595569364e-08,
      "loss": 0.4349,
      "step": 447160
    },
    {
      "epoch": 183.04,
      "grad_norm": 2.389587879180908,
      "learning_rate": 9.153965429181857e-08,
      "loss": 0.4613,
      "step": 447170
    },
    {
      "epoch": 183.05,
      "grad_norm": 2.024407386779785,
      "learning_rate": 9.149575934372741e-08,
      "loss": 0.4541,
      "step": 447180
    },
    {
      "epoch": 183.05,
      "grad_norm": 1.9016388654708862,
      "learning_rate": 9.145187471286406e-08,
      "loss": 0.4572,
      "step": 447190
    },
    {
      "epoch": 183.05,
      "grad_norm": 1.7869359254837036,
      "learning_rate": 9.140800039942964e-08,
      "loss": 0.4472,
      "step": 447200
    },
    {
      "epoch": 183.06,
      "grad_norm": 2.0499680042266846,
      "learning_rate": 9.136413640362479e-08,
      "loss": 0.4507,
      "step": 447210
    },
    {
      "epoch": 183.06,
      "grad_norm": 2.307791233062744,
      "learning_rate": 9.132028272565142e-08,
      "loss": 0.4383,
      "step": 447220
    },
    {
      "epoch": 183.07,
      "grad_norm": 1.932938575744629,
      "learning_rate": 9.127643936570938e-08,
      "loss": 0.4448,
      "step": 447230
    },
    {
      "epoch": 183.07,
      "grad_norm": 2.0342519283294678,
      "learning_rate": 9.12326063239998e-08,
      "loss": 0.4504,
      "step": 447240
    },
    {
      "epoch": 183.07,
      "grad_norm": 1.9787617921829224,
      "learning_rate": 9.118878360072383e-08,
      "loss": 0.437,
      "step": 447250
    },
    {
      "epoch": 183.08,
      "grad_norm": 2.4975366592407227,
      "learning_rate": 9.114497119608183e-08,
      "loss": 0.4582,
      "step": 447260
    },
    {
      "epoch": 183.08,
      "grad_norm": 1.7634186744689941,
      "learning_rate": 9.110116911027442e-08,
      "loss": 0.4542,
      "step": 447270
    },
    {
      "epoch": 183.09,
      "grad_norm": 1.8376049995422363,
      "learning_rate": 9.105737734350273e-08,
      "loss": 0.4432,
      "step": 447280
    },
    {
      "epoch": 183.09,
      "grad_norm": 1.8939549922943115,
      "learning_rate": 9.101359589596688e-08,
      "loss": 0.4434,
      "step": 447290
    },
    {
      "epoch": 183.09,
      "grad_norm": 2.1204562187194824,
      "learning_rate": 9.096982476786774e-08,
      "loss": 0.4557,
      "step": 447300
    },
    {
      "epoch": 183.1,
      "grad_norm": 2.3677303791046143,
      "learning_rate": 9.092606395940541e-08,
      "loss": 0.4172,
      "step": 447310
    },
    {
      "epoch": 183.1,
      "grad_norm": 1.9922834634780884,
      "learning_rate": 9.088231347078103e-08,
      "loss": 0.4603,
      "step": 447320
    },
    {
      "epoch": 183.11,
      "grad_norm": 2.1032488346099854,
      "learning_rate": 9.083857330219446e-08,
      "loss": 0.4419,
      "step": 447330
    },
    {
      "epoch": 183.11,
      "grad_norm": 2.2452850341796875,
      "learning_rate": 9.079484345384602e-08,
      "loss": 0.4379,
      "step": 447340
    },
    {
      "epoch": 183.12,
      "grad_norm": 2.0477311611175537,
      "learning_rate": 9.075112392593637e-08,
      "loss": 0.4292,
      "step": 447350
    },
    {
      "epoch": 183.12,
      "grad_norm": 1.8673579692840576,
      "learning_rate": 9.070741471866586e-08,
      "loss": 0.4501,
      "step": 447360
    },
    {
      "epoch": 183.12,
      "grad_norm": 2.2255001068115234,
      "learning_rate": 9.066371583223431e-08,
      "loss": 0.4228,
      "step": 447370
    },
    {
      "epoch": 183.13,
      "grad_norm": 1.9335936307907104,
      "learning_rate": 9.062002726684209e-08,
      "loss": 0.4563,
      "step": 447380
    },
    {
      "epoch": 183.13,
      "grad_norm": 1.9536687135696411,
      "learning_rate": 9.057634902268932e-08,
      "loss": 0.4445,
      "step": 447390
    },
    {
      "epoch": 183.14,
      "grad_norm": 2.087278127670288,
      "learning_rate": 9.053268109997635e-08,
      "loss": 0.4511,
      "step": 447400
    },
    {
      "epoch": 183.14,
      "grad_norm": 1.941924810409546,
      "learning_rate": 9.048902349890275e-08,
      "loss": 0.4429,
      "step": 447410
    },
    {
      "epoch": 183.14,
      "grad_norm": 2.168126106262207,
      "learning_rate": 9.044537621966863e-08,
      "loss": 0.4491,
      "step": 447420
    },
    {
      "epoch": 183.15,
      "grad_norm": 2.252568006515503,
      "learning_rate": 9.040173926247408e-08,
      "loss": 0.4506,
      "step": 447430
    },
    {
      "epoch": 183.15,
      "grad_norm": 2.2102484703063965,
      "learning_rate": 9.035811262751895e-08,
      "loss": 0.4477,
      "step": 447440
    },
    {
      "epoch": 183.16,
      "grad_norm": 2.1898670196533203,
      "learning_rate": 9.031449631500308e-08,
      "loss": 0.4105,
      "step": 447450
    },
    {
      "epoch": 183.16,
      "grad_norm": 2.629791259765625,
      "learning_rate": 9.02708903251263e-08,
      "loss": 0.4498,
      "step": 447460
    },
    {
      "epoch": 183.16,
      "grad_norm": 1.822257399559021,
      "learning_rate": 9.022729465808847e-08,
      "loss": 0.4302,
      "step": 447470
    },
    {
      "epoch": 183.17,
      "grad_norm": 2.399387836456299,
      "learning_rate": 9.01837093140889e-08,
      "loss": 0.4356,
      "step": 447480
    },
    {
      "epoch": 183.17,
      "grad_norm": 2.0077126026153564,
      "learning_rate": 9.014013429332768e-08,
      "loss": 0.4485,
      "step": 447490
    },
    {
      "epoch": 183.18,
      "grad_norm": 2.1975598335266113,
      "learning_rate": 9.009656959600413e-08,
      "loss": 0.4382,
      "step": 447500
    },
    {
      "epoch": 183.18,
      "grad_norm": 2.4099912643432617,
      "learning_rate": 9.005301522231785e-08,
      "loss": 0.452,
      "step": 447510
    },
    {
      "epoch": 183.18,
      "grad_norm": 2.2566704750061035,
      "learning_rate": 9.000947117246866e-08,
      "loss": 0.4594,
      "step": 447520
    },
    {
      "epoch": 183.19,
      "grad_norm": 2.0057361125946045,
      "learning_rate": 8.99659374466559e-08,
      "loss": 0.4518,
      "step": 447530
    },
    {
      "epoch": 183.19,
      "grad_norm": 1.8457870483398438,
      "learning_rate": 8.99224140450786e-08,
      "loss": 0.4582,
      "step": 447540
    },
    {
      "epoch": 183.2,
      "grad_norm": 2.0945358276367188,
      "learning_rate": 8.987890096793663e-08,
      "loss": 0.4408,
      "step": 447550
    },
    {
      "epoch": 183.2,
      "grad_norm": 2.161612033843994,
      "learning_rate": 8.983539821542902e-08,
      "loss": 0.4422,
      "step": 447560
    },
    {
      "epoch": 183.21,
      "grad_norm": 1.9555047750473022,
      "learning_rate": 8.979190578775487e-08,
      "loss": 0.4496,
      "step": 447570
    },
    {
      "epoch": 183.21,
      "grad_norm": 2.5112807750701904,
      "learning_rate": 8.974842368511372e-08,
      "loss": 0.427,
      "step": 447580
    },
    {
      "epoch": 183.21,
      "grad_norm": 2.6357645988464355,
      "learning_rate": 8.970495190770491e-08,
      "loss": 0.4497,
      "step": 447590
    },
    {
      "epoch": 183.22,
      "grad_norm": 2.2739858627319336,
      "learning_rate": 8.966149045572723e-08,
      "loss": 0.4476,
      "step": 447600
    },
    {
      "epoch": 183.22,
      "grad_norm": 2.1761415004730225,
      "learning_rate": 8.961803932938001e-08,
      "loss": 0.4532,
      "step": 447610
    },
    {
      "epoch": 183.23,
      "grad_norm": 2.5736684799194336,
      "learning_rate": 8.957459852886231e-08,
      "loss": 0.4422,
      "step": 447620
    },
    {
      "epoch": 183.23,
      "grad_norm": 1.8950035572052002,
      "learning_rate": 8.953116805437291e-08,
      "loss": 0.4497,
      "step": 447630
    },
    {
      "epoch": 183.23,
      "grad_norm": 1.906198501586914,
      "learning_rate": 8.94877479061109e-08,
      "loss": 0.4386,
      "step": 447640
    },
    {
      "epoch": 183.24,
      "grad_norm": 1.6558172702789307,
      "learning_rate": 8.94443380842753e-08,
      "loss": 0.456,
      "step": 447650
    },
    {
      "epoch": 183.24,
      "grad_norm": 1.9837422370910645,
      "learning_rate": 8.940093858906495e-08,
      "loss": 0.4493,
      "step": 447660
    },
    {
      "epoch": 183.25,
      "grad_norm": 2.3710546493530273,
      "learning_rate": 8.935754942067811e-08,
      "loss": 0.4439,
      "step": 447670
    },
    {
      "epoch": 183.25,
      "grad_norm": 2.122401237487793,
      "learning_rate": 8.93141705793141e-08,
      "loss": 0.4609,
      "step": 447680
    },
    {
      "epoch": 183.25,
      "grad_norm": 2.0114519596099854,
      "learning_rate": 8.927080206517145e-08,
      "loss": 0.4213,
      "step": 447690
    },
    {
      "epoch": 183.26,
      "grad_norm": 1.895177960395813,
      "learning_rate": 8.922744387844898e-08,
      "loss": 0.4319,
      "step": 447700
    },
    {
      "epoch": 183.26,
      "grad_norm": 1.9054055213928223,
      "learning_rate": 8.91840960193452e-08,
      "loss": 0.4404,
      "step": 447710
    },
    {
      "epoch": 183.27,
      "grad_norm": 2.168330669403076,
      "learning_rate": 8.914075848805843e-08,
      "loss": 0.441,
      "step": 447720
    },
    {
      "epoch": 183.27,
      "grad_norm": 1.9527201652526855,
      "learning_rate": 8.909743128478744e-08,
      "loss": 0.4281,
      "step": 447730
    },
    {
      "epoch": 183.27,
      "grad_norm": 2.6139557361602783,
      "learning_rate": 8.905411440973078e-08,
      "loss": 0.4443,
      "step": 447740
    },
    {
      "epoch": 183.28,
      "grad_norm": 2.0278706550598145,
      "learning_rate": 8.901080786308696e-08,
      "loss": 0.4222,
      "step": 447750
    },
    {
      "epoch": 183.28,
      "grad_norm": 2.0668842792510986,
      "learning_rate": 8.896751164505404e-08,
      "loss": 0.4376,
      "step": 447760
    },
    {
      "epoch": 183.29,
      "grad_norm": 1.8612957000732422,
      "learning_rate": 8.892422575583003e-08,
      "loss": 0.4513,
      "step": 447770
    },
    {
      "epoch": 183.29,
      "grad_norm": 2.3000035285949707,
      "learning_rate": 8.888095019561423e-08,
      "loss": 0.4418,
      "step": 447780
    },
    {
      "epoch": 183.3,
      "grad_norm": 1.7668884992599487,
      "learning_rate": 8.883768496460416e-08,
      "loss": 0.4563,
      "step": 447790
    },
    {
      "epoch": 183.3,
      "grad_norm": 2.3404345512390137,
      "learning_rate": 8.87944300629986e-08,
      "loss": 0.4373,
      "step": 447800
    },
    {
      "epoch": 183.3,
      "grad_norm": 2.065145492553711,
      "learning_rate": 8.875118549099506e-08,
      "loss": 0.4412,
      "step": 447810
    },
    {
      "epoch": 183.31,
      "grad_norm": 2.2433114051818848,
      "learning_rate": 8.870795124879181e-08,
      "loss": 0.4503,
      "step": 447820
    },
    {
      "epoch": 183.31,
      "grad_norm": 2.135047674179077,
      "learning_rate": 8.86647273365869e-08,
      "loss": 0.4555,
      "step": 447830
    },
    {
      "epoch": 183.32,
      "grad_norm": 2.215052604675293,
      "learning_rate": 8.86215137545783e-08,
      "loss": 0.4182,
      "step": 447840
    },
    {
      "epoch": 183.32,
      "grad_norm": 1.8050447702407837,
      "learning_rate": 8.857831050296405e-08,
      "loss": 0.4448,
      "step": 447850
    },
    {
      "epoch": 183.32,
      "grad_norm": 2.1215240955352783,
      "learning_rate": 8.853511758194219e-08,
      "loss": 0.4296,
      "step": 447860
    },
    {
      "epoch": 183.33,
      "grad_norm": 2.150516986846924,
      "learning_rate": 8.849193499171044e-08,
      "loss": 0.4397,
      "step": 447870
    },
    {
      "epoch": 183.33,
      "grad_norm": 1.9873807430267334,
      "learning_rate": 8.844876273246659e-08,
      "loss": 0.4512,
      "step": 447880
    },
    {
      "epoch": 183.34,
      "grad_norm": 2.6139113903045654,
      "learning_rate": 8.840560080440838e-08,
      "loss": 0.4441,
      "step": 447890
    },
    {
      "epoch": 183.34,
      "grad_norm": 2.2491579055786133,
      "learning_rate": 8.836244920773358e-08,
      "loss": 0.4332,
      "step": 447900
    },
    {
      "epoch": 183.34,
      "grad_norm": 2.032306432723999,
      "learning_rate": 8.831930794263969e-08,
      "loss": 0.448,
      "step": 447910
    },
    {
      "epoch": 183.35,
      "grad_norm": 1.9313992261886597,
      "learning_rate": 8.82761770093247e-08,
      "loss": 0.4517,
      "step": 447920
    },
    {
      "epoch": 183.35,
      "grad_norm": 1.9478479623794556,
      "learning_rate": 8.823305640798615e-08,
      "loss": 0.4564,
      "step": 447930
    },
    {
      "epoch": 183.36,
      "grad_norm": 1.803684949874878,
      "learning_rate": 8.818994613882098e-08,
      "loss": 0.4326,
      "step": 447940
    },
    {
      "epoch": 183.36,
      "grad_norm": 2.1402382850646973,
      "learning_rate": 8.814684620202751e-08,
      "loss": 0.4407,
      "step": 447950
    },
    {
      "epoch": 183.36,
      "grad_norm": 1.9814740419387817,
      "learning_rate": 8.810375659780268e-08,
      "loss": 0.426,
      "step": 447960
    },
    {
      "epoch": 183.37,
      "grad_norm": 2.279066801071167,
      "learning_rate": 8.806067732634375e-08,
      "loss": 0.429,
      "step": 447970
    },
    {
      "epoch": 183.37,
      "grad_norm": 2.489898204803467,
      "learning_rate": 8.801760838784847e-08,
      "loss": 0.4444,
      "step": 447980
    },
    {
      "epoch": 183.38,
      "grad_norm": 1.9032673835754395,
      "learning_rate": 8.797454978251381e-08,
      "loss": 0.4609,
      "step": 447990
    },
    {
      "epoch": 183.38,
      "grad_norm": 1.9632428884506226,
      "learning_rate": 8.793150151053729e-08,
      "loss": 0.4333,
      "step": 448000
    },
    {
      "epoch": 183.39,
      "grad_norm": 2.19711971282959,
      "learning_rate": 8.788846357211587e-08,
      "loss": 0.4415,
      "step": 448010
    },
    {
      "epoch": 183.39,
      "grad_norm": 1.9906715154647827,
      "learning_rate": 8.78454359674468e-08,
      "loss": 0.4703,
      "step": 448020
    },
    {
      "epoch": 183.39,
      "grad_norm": 1.60550856590271,
      "learning_rate": 8.780241869672705e-08,
      "loss": 0.4409,
      "step": 448030
    },
    {
      "epoch": 183.4,
      "grad_norm": 2.1076576709747314,
      "learning_rate": 8.775941176015384e-08,
      "loss": 0.4473,
      "step": 448040
    },
    {
      "epoch": 183.4,
      "grad_norm": 2.1795859336853027,
      "learning_rate": 8.771641515792418e-08,
      "loss": 0.4411,
      "step": 448050
    },
    {
      "epoch": 183.41,
      "grad_norm": 2.3483431339263916,
      "learning_rate": 8.767342889023504e-08,
      "loss": 0.4447,
      "step": 448060
    },
    {
      "epoch": 183.41,
      "grad_norm": 2.111858367919922,
      "learning_rate": 8.763045295728314e-08,
      "loss": 0.4506,
      "step": 448070
    },
    {
      "epoch": 183.41,
      "grad_norm": 2.1571171283721924,
      "learning_rate": 8.758748735926543e-08,
      "loss": 0.4546,
      "step": 448080
    },
    {
      "epoch": 183.42,
      "grad_norm": 2.043551206588745,
      "learning_rate": 8.754453209637892e-08,
      "loss": 0.4589,
      "step": 448090
    },
    {
      "epoch": 183.42,
      "grad_norm": 1.9220359325408936,
      "learning_rate": 8.750158716882032e-08,
      "loss": 0.4411,
      "step": 448100
    },
    {
      "epoch": 183.43,
      "grad_norm": 2.198916435241699,
      "learning_rate": 8.745865257678633e-08,
      "loss": 0.4532,
      "step": 448110
    },
    {
      "epoch": 183.43,
      "grad_norm": 2.1808481216430664,
      "learning_rate": 8.741572832047343e-08,
      "loss": 0.4386,
      "step": 448120
    },
    {
      "epoch": 183.43,
      "grad_norm": 2.1568243503570557,
      "learning_rate": 8.73728144000786e-08,
      "loss": 0.4357,
      "step": 448130
    },
    {
      "epoch": 183.44,
      "grad_norm": 2.4821157455444336,
      "learning_rate": 8.732991081579774e-08,
      "loss": 0.4502,
      "step": 448140
    },
    {
      "epoch": 183.44,
      "grad_norm": 2.1515915393829346,
      "learning_rate": 8.728701756782787e-08,
      "loss": 0.4696,
      "step": 448150
    },
    {
      "epoch": 183.45,
      "grad_norm": 1.9071861505508423,
      "learning_rate": 8.724413465636569e-08,
      "loss": 0.4311,
      "step": 448160
    },
    {
      "epoch": 183.45,
      "grad_norm": 2.55536150932312,
      "learning_rate": 8.72012620816074e-08,
      "loss": 0.4287,
      "step": 448170
    },
    {
      "epoch": 183.45,
      "grad_norm": 2.1970653533935547,
      "learning_rate": 8.715839984374945e-08,
      "loss": 0.4241,
      "step": 448180
    },
    {
      "epoch": 183.46,
      "grad_norm": 1.8481638431549072,
      "learning_rate": 8.71155479429883e-08,
      "loss": 0.4396,
      "step": 448190
    },
    {
      "epoch": 183.46,
      "grad_norm": 1.761932134628296,
      "learning_rate": 8.707270637951991e-08,
      "loss": 0.4157,
      "step": 448200
    },
    {
      "epoch": 183.47,
      "grad_norm": 1.97702956199646,
      "learning_rate": 8.70298751535407e-08,
      "loss": 0.4408,
      "step": 448210
    },
    {
      "epoch": 183.47,
      "grad_norm": 1.8600356578826904,
      "learning_rate": 8.698705426524689e-08,
      "loss": 0.4319,
      "step": 448220
    },
    {
      "epoch": 183.48,
      "grad_norm": 1.84833824634552,
      "learning_rate": 8.694424371483467e-08,
      "loss": 0.4587,
      "step": 448230
    },
    {
      "epoch": 183.48,
      "grad_norm": 2.0937459468841553,
      "learning_rate": 8.690144350249999e-08,
      "loss": 0.4624,
      "step": 448240
    },
    {
      "epoch": 183.48,
      "grad_norm": 1.945263147354126,
      "learning_rate": 8.685865362843955e-08,
      "loss": 0.4415,
      "step": 448250
    },
    {
      "epoch": 183.49,
      "grad_norm": 2.185312032699585,
      "learning_rate": 8.681587409284825e-08,
      "loss": 0.4499,
      "step": 448260
    },
    {
      "epoch": 183.49,
      "grad_norm": 1.694420337677002,
      "learning_rate": 8.677310489592281e-08,
      "loss": 0.4655,
      "step": 448270
    },
    {
      "epoch": 183.5,
      "grad_norm": 2.3292672634124756,
      "learning_rate": 8.673034603785918e-08,
      "loss": 0.4574,
      "step": 448280
    },
    {
      "epoch": 183.5,
      "grad_norm": 1.8981856107711792,
      "learning_rate": 8.668759751885275e-08,
      "loss": 0.4551,
      "step": 448290
    },
    {
      "epoch": 183.5,
      "grad_norm": 1.9604239463806152,
      "learning_rate": 8.664485933909974e-08,
      "loss": 0.4561,
      "step": 448300
    },
    {
      "epoch": 183.51,
      "grad_norm": 2.1746068000793457,
      "learning_rate": 8.660213149879582e-08,
      "loss": 0.4713,
      "step": 448310
    },
    {
      "epoch": 183.51,
      "grad_norm": 2.5454530715942383,
      "learning_rate": 8.655941399813667e-08,
      "loss": 0.457,
      "step": 448320
    },
    {
      "epoch": 183.52,
      "grad_norm": 1.9311389923095703,
      "learning_rate": 8.65167068373182e-08,
      "loss": 0.4537,
      "step": 448330
    },
    {
      "epoch": 183.52,
      "grad_norm": 2.01424503326416,
      "learning_rate": 8.647401001653535e-08,
      "loss": 0.4555,
      "step": 448340
    },
    {
      "epoch": 183.52,
      "grad_norm": 1.9750174283981323,
      "learning_rate": 8.643132353598482e-08,
      "loss": 0.428,
      "step": 448350
    },
    {
      "epoch": 183.53,
      "grad_norm": 2.008300304412842,
      "learning_rate": 8.638864739586122e-08,
      "loss": 0.4375,
      "step": 448360
    },
    {
      "epoch": 183.53,
      "grad_norm": 2.3667914867401123,
      "learning_rate": 8.634598159636052e-08,
      "loss": 0.4368,
      "step": 448370
    },
    {
      "epoch": 183.54,
      "grad_norm": 2.207440137863159,
      "learning_rate": 8.630332613767838e-08,
      "loss": 0.4498,
      "step": 448380
    },
    {
      "epoch": 183.54,
      "grad_norm": 2.1079063415527344,
      "learning_rate": 8.626068102000946e-08,
      "loss": 0.45,
      "step": 448390
    },
    {
      "epoch": 183.54,
      "grad_norm": 2.366297960281372,
      "learning_rate": 8.621804624354967e-08,
      "loss": 0.4601,
      "step": 448400
    },
    {
      "epoch": 183.55,
      "grad_norm": 2.053520917892456,
      "learning_rate": 8.617542180849391e-08,
      "loss": 0.4528,
      "step": 448410
    },
    {
      "epoch": 183.55,
      "grad_norm": 2.795260190963745,
      "learning_rate": 8.613280771503787e-08,
      "loss": 0.4428,
      "step": 448420
    },
    {
      "epoch": 183.56,
      "grad_norm": 2.319915294647217,
      "learning_rate": 8.609020396337642e-08,
      "loss": 0.4441,
      "step": 448430
    },
    {
      "epoch": 183.56,
      "grad_norm": 2.22839093208313,
      "learning_rate": 8.604761055370475e-08,
      "loss": 0.4473,
      "step": 448440
    },
    {
      "epoch": 183.57,
      "grad_norm": 2.0950591564178467,
      "learning_rate": 8.600502748621826e-08,
      "loss": 0.4269,
      "step": 448450
    },
    {
      "epoch": 183.57,
      "grad_norm": 2.262632131576538,
      "learning_rate": 8.596245476111158e-08,
      "loss": 0.4489,
      "step": 448460
    },
    {
      "epoch": 183.57,
      "grad_norm": 2.02181339263916,
      "learning_rate": 8.59198923785804e-08,
      "loss": 0.4568,
      "step": 448470
    },
    {
      "epoch": 183.58,
      "grad_norm": 1.7501107454299927,
      "learning_rate": 8.587734033881884e-08,
      "loss": 0.4373,
      "step": 448480
    },
    {
      "epoch": 183.58,
      "grad_norm": 2.10278582572937,
      "learning_rate": 8.583479864202255e-08,
      "loss": 0.4349,
      "step": 448490
    },
    {
      "epoch": 183.59,
      "grad_norm": 1.687808632850647,
      "learning_rate": 8.579226728838594e-08,
      "loss": 0.4254,
      "step": 448500
    },
    {
      "epoch": 183.59,
      "grad_norm": 2.1529335975646973,
      "learning_rate": 8.574974627810413e-08,
      "loss": 0.4416,
      "step": 448510
    },
    {
      "epoch": 183.59,
      "grad_norm": 2.6299021244049072,
      "learning_rate": 8.570723561137178e-08,
      "loss": 0.4487,
      "step": 448520
    },
    {
      "epoch": 183.6,
      "grad_norm": 2.0197160243988037,
      "learning_rate": 8.566473528838379e-08,
      "loss": 0.4429,
      "step": 448530
    },
    {
      "epoch": 183.6,
      "grad_norm": 1.858047604560852,
      "learning_rate": 8.562224530933451e-08,
      "loss": 0.4475,
      "step": 448540
    },
    {
      "epoch": 183.61,
      "grad_norm": 2.3678457736968994,
      "learning_rate": 8.55797656744186e-08,
      "loss": 0.4378,
      "step": 448550
    },
    {
      "epoch": 183.61,
      "grad_norm": 2.3876092433929443,
      "learning_rate": 8.553729638383095e-08,
      "loss": 0.4552,
      "step": 448560
    },
    {
      "epoch": 183.61,
      "grad_norm": 1.8083245754241943,
      "learning_rate": 8.549483743776594e-08,
      "loss": 0.4419,
      "step": 448570
    },
    {
      "epoch": 183.62,
      "grad_norm": 2.0695981979370117,
      "learning_rate": 8.545238883641819e-08,
      "loss": 0.4334,
      "step": 448580
    },
    {
      "epoch": 183.62,
      "grad_norm": 1.5475399494171143,
      "learning_rate": 8.540995057998209e-08,
      "loss": 0.4267,
      "step": 448590
    },
    {
      "epoch": 183.63,
      "grad_norm": 2.1506361961364746,
      "learning_rate": 8.5367522668652e-08,
      "loss": 0.4354,
      "step": 448600
    },
    {
      "epoch": 183.63,
      "grad_norm": 2.1871933937072754,
      "learning_rate": 8.532510510262207e-08,
      "loss": 0.4496,
      "step": 448610
    },
    {
      "epoch": 183.63,
      "grad_norm": 2.2012765407562256,
      "learning_rate": 8.52826978820869e-08,
      "loss": 0.4563,
      "step": 448620
    },
    {
      "epoch": 183.64,
      "grad_norm": 2.1743955612182617,
      "learning_rate": 8.524030100724088e-08,
      "loss": 0.4434,
      "step": 448630
    },
    {
      "epoch": 183.64,
      "grad_norm": 1.589672327041626,
      "learning_rate": 8.519791447827787e-08,
      "loss": 0.4516,
      "step": 448640
    },
    {
      "epoch": 183.65,
      "grad_norm": 1.934106707572937,
      "learning_rate": 8.515553829539224e-08,
      "loss": 0.4496,
      "step": 448650
    },
    {
      "epoch": 183.65,
      "grad_norm": 2.0573320388793945,
      "learning_rate": 8.511317245877784e-08,
      "loss": 0.4429,
      "step": 448660
    },
    {
      "epoch": 183.66,
      "grad_norm": 2.044282913208008,
      "learning_rate": 8.507081696862931e-08,
      "loss": 0.4598,
      "step": 448670
    },
    {
      "epoch": 183.66,
      "grad_norm": 1.8444206714630127,
      "learning_rate": 8.502847182513999e-08,
      "loss": 0.4336,
      "step": 448680
    },
    {
      "epoch": 183.66,
      "grad_norm": 2.2442243099212646,
      "learning_rate": 8.498613702850451e-08,
      "loss": 0.4438,
      "step": 448690
    },
    {
      "epoch": 183.67,
      "grad_norm": 2.2366130352020264,
      "learning_rate": 8.494381257891648e-08,
      "loss": 0.4464,
      "step": 448700
    },
    {
      "epoch": 183.67,
      "grad_norm": 2.374213933944702,
      "learning_rate": 8.49014984765692e-08,
      "loss": 0.4485,
      "step": 448710
    },
    {
      "epoch": 183.68,
      "grad_norm": 2.1231348514556885,
      "learning_rate": 8.485919472165735e-08,
      "loss": 0.4249,
      "step": 448720
    },
    {
      "epoch": 183.68,
      "grad_norm": 1.9340757131576538,
      "learning_rate": 8.481690131437475e-08,
      "loss": 0.4323,
      "step": 448730
    },
    {
      "epoch": 183.68,
      "grad_norm": 2.1063578128814697,
      "learning_rate": 8.477461825491449e-08,
      "loss": 0.4379,
      "step": 448740
    },
    {
      "epoch": 183.69,
      "grad_norm": 2.157137393951416,
      "learning_rate": 8.473234554347069e-08,
      "loss": 0.4308,
      "step": 448750
    },
    {
      "epoch": 183.69,
      "grad_norm": 2.2538278102874756,
      "learning_rate": 8.469008318023693e-08,
      "loss": 0.4566,
      "step": 448760
    },
    {
      "epoch": 183.7,
      "grad_norm": 1.7780706882476807,
      "learning_rate": 8.464783116540682e-08,
      "loss": 0.454,
      "step": 448770
    },
    {
      "epoch": 183.7,
      "grad_norm": 2.2690341472625732,
      "learning_rate": 8.460558949917396e-08,
      "loss": 0.4371,
      "step": 448780
    },
    {
      "epoch": 183.7,
      "grad_norm": 2.170379161834717,
      "learning_rate": 8.456335818173166e-08,
      "loss": 0.4322,
      "step": 448790
    },
    {
      "epoch": 183.71,
      "grad_norm": 2.061246633529663,
      "learning_rate": 8.452113721327354e-08,
      "loss": 0.4481,
      "step": 448800
    },
    {
      "epoch": 183.71,
      "grad_norm": 2.679029941558838,
      "learning_rate": 8.447892659399291e-08,
      "loss": 0.4463,
      "step": 448810
    },
    {
      "epoch": 183.72,
      "grad_norm": 1.654478907585144,
      "learning_rate": 8.44367263240834e-08,
      "loss": 0.4293,
      "step": 448820
    },
    {
      "epoch": 183.72,
      "grad_norm": 2.2320847511291504,
      "learning_rate": 8.439453640373805e-08,
      "loss": 0.4317,
      "step": 448830
    },
    {
      "epoch": 183.72,
      "grad_norm": 2.597883701324463,
      "learning_rate": 8.435235683314995e-08,
      "loss": 0.444,
      "step": 448840
    },
    {
      "epoch": 183.73,
      "grad_norm": 1.813201904296875,
      "learning_rate": 8.43101876125127e-08,
      "loss": 0.437,
      "step": 448850
    },
    {
      "epoch": 183.73,
      "grad_norm": 2.0802836418151855,
      "learning_rate": 8.426802874201937e-08,
      "loss": 0.4701,
      "step": 448860
    },
    {
      "epoch": 183.74,
      "grad_norm": 1.8306291103363037,
      "learning_rate": 8.422588022186276e-08,
      "loss": 0.4569,
      "step": 448870
    },
    {
      "epoch": 183.74,
      "grad_norm": 2.9127697944641113,
      "learning_rate": 8.41837420522365e-08,
      "loss": 0.4516,
      "step": 448880
    },
    {
      "epoch": 183.75,
      "grad_norm": 1.8698796033859253,
      "learning_rate": 8.414161423333311e-08,
      "loss": 0.4277,
      "step": 448890
    },
    {
      "epoch": 183.75,
      "grad_norm": 1.9987338781356812,
      "learning_rate": 8.409949676534541e-08,
      "loss": 0.4741,
      "step": 448900
    },
    {
      "epoch": 183.75,
      "grad_norm": 2.506730794906616,
      "learning_rate": 8.405738964846727e-08,
      "loss": 0.4288,
      "step": 448910
    },
    {
      "epoch": 183.76,
      "grad_norm": 2.2344746589660645,
      "learning_rate": 8.401529288289098e-08,
      "loss": 0.452,
      "step": 448920
    },
    {
      "epoch": 183.76,
      "grad_norm": 2.016874074935913,
      "learning_rate": 8.397320646880935e-08,
      "loss": 0.4663,
      "step": 448930
    },
    {
      "epoch": 183.77,
      "grad_norm": 1.7479617595672607,
      "learning_rate": 8.393113040641519e-08,
      "loss": 0.4462,
      "step": 448940
    },
    {
      "epoch": 183.77,
      "grad_norm": 2.0653023719787598,
      "learning_rate": 8.388906469590133e-08,
      "loss": 0.4404,
      "step": 448950
    },
    {
      "epoch": 183.77,
      "grad_norm": 2.0977683067321777,
      "learning_rate": 8.384700933746058e-08,
      "loss": 0.4466,
      "step": 448960
    },
    {
      "epoch": 183.78,
      "grad_norm": 1.9706573486328125,
      "learning_rate": 8.380496433128547e-08,
      "loss": 0.4527,
      "step": 448970
    },
    {
      "epoch": 183.78,
      "grad_norm": 1.866013526916504,
      "learning_rate": 8.376292967756832e-08,
      "loss": 0.4427,
      "step": 448980
    },
    {
      "epoch": 183.79,
      "grad_norm": 2.2628586292266846,
      "learning_rate": 8.372090537650195e-08,
      "loss": 0.4519,
      "step": 448990
    },
    {
      "epoch": 183.79,
      "grad_norm": 1.8429768085479736,
      "learning_rate": 8.36788914282789e-08,
      "loss": 0.4435,
      "step": 449000
    },
    {
      "epoch": 183.79,
      "grad_norm": 1.9648176431655884,
      "learning_rate": 8.363688783309171e-08,
      "loss": 0.4488,
      "step": 449010
    },
    {
      "epoch": 183.8,
      "grad_norm": 2.147512197494507,
      "learning_rate": 8.359489459113245e-08,
      "loss": 0.4455,
      "step": 449020
    },
    {
      "epoch": 183.8,
      "grad_norm": 1.7376222610473633,
      "learning_rate": 8.35529117025939e-08,
      "loss": 0.4352,
      "step": 449030
    },
    {
      "epoch": 183.81,
      "grad_norm": 1.9404009580612183,
      "learning_rate": 8.351093916766811e-08,
      "loss": 0.4421,
      "step": 449040
    },
    {
      "epoch": 183.81,
      "grad_norm": 2.1959807872772217,
      "learning_rate": 8.346897698654738e-08,
      "loss": 0.463,
      "step": 449050
    },
    {
      "epoch": 183.81,
      "grad_norm": 2.0626606941223145,
      "learning_rate": 8.3427025159424e-08,
      "loss": 0.4516,
      "step": 449060
    },
    {
      "epoch": 183.82,
      "grad_norm": 2.166903257369995,
      "learning_rate": 8.338508368648999e-08,
      "loss": 0.4231,
      "step": 449070
    },
    {
      "epoch": 183.82,
      "grad_norm": 1.9362401962280273,
      "learning_rate": 8.334315256793766e-08,
      "loss": 0.4501,
      "step": 449080
    },
    {
      "epoch": 183.83,
      "grad_norm": 2.2667086124420166,
      "learning_rate": 8.330123180395878e-08,
      "loss": 0.4407,
      "step": 449090
    },
    {
      "epoch": 183.83,
      "grad_norm": 2.2447400093078613,
      "learning_rate": 8.325932139474616e-08,
      "loss": 0.4435,
      "step": 449100
    },
    {
      "epoch": 183.84,
      "grad_norm": 2.21518874168396,
      "learning_rate": 8.321742134049106e-08,
      "loss": 0.4332,
      "step": 449110
    },
    {
      "epoch": 183.84,
      "grad_norm": 1.975571632385254,
      "learning_rate": 8.317553164138577e-08,
      "loss": 0.4269,
      "step": 449120
    },
    {
      "epoch": 183.84,
      "grad_norm": 2.0462486743927,
      "learning_rate": 8.31336522976218e-08,
      "loss": 0.4395,
      "step": 449130
    },
    {
      "epoch": 183.85,
      "grad_norm": 2.2376832962036133,
      "learning_rate": 8.309178330939119e-08,
      "loss": 0.4558,
      "step": 449140
    },
    {
      "epoch": 183.85,
      "grad_norm": 2.311601400375366,
      "learning_rate": 8.304992467688571e-08,
      "loss": 0.4412,
      "step": 449150
    },
    {
      "epoch": 183.86,
      "grad_norm": 2.0092408657073975,
      "learning_rate": 8.30080764002974e-08,
      "loss": 0.456,
      "step": 449160
    },
    {
      "epoch": 183.86,
      "grad_norm": 2.2021493911743164,
      "learning_rate": 8.296623847981752e-08,
      "loss": 0.4688,
      "step": 449170
    },
    {
      "epoch": 183.86,
      "grad_norm": 1.9683889150619507,
      "learning_rate": 8.292441091563808e-08,
      "loss": 0.4433,
      "step": 449180
    },
    {
      "epoch": 183.87,
      "grad_norm": 2.484851360321045,
      "learning_rate": 8.288259370795034e-08,
      "loss": 0.4419,
      "step": 449190
    },
    {
      "epoch": 183.87,
      "grad_norm": 1.8305113315582275,
      "learning_rate": 8.284078685694634e-08,
      "loss": 0.4272,
      "step": 449200
    },
    {
      "epoch": 183.88,
      "grad_norm": 1.8471699953079224,
      "learning_rate": 8.279899036281708e-08,
      "loss": 0.4698,
      "step": 449210
    },
    {
      "epoch": 183.88,
      "grad_norm": 2.251088857650757,
      "learning_rate": 8.275720422575432e-08,
      "loss": 0.4633,
      "step": 449220
    },
    {
      "epoch": 183.88,
      "grad_norm": 8.000722885131836,
      "learning_rate": 8.271542844594933e-08,
      "loss": 0.4432,
      "step": 449230
    },
    {
      "epoch": 183.89,
      "grad_norm": 2.3258297443389893,
      "learning_rate": 8.267366302359361e-08,
      "loss": 0.4156,
      "step": 449240
    },
    {
      "epoch": 183.89,
      "grad_norm": 2.1505980491638184,
      "learning_rate": 8.263190795887816e-08,
      "loss": 0.4286,
      "step": 449250
    },
    {
      "epoch": 183.9,
      "grad_norm": 1.8191423416137695,
      "learning_rate": 8.259016325199473e-08,
      "loss": 0.4347,
      "step": 449260
    },
    {
      "epoch": 183.9,
      "grad_norm": 1.8330020904541016,
      "learning_rate": 8.254842890313461e-08,
      "loss": 0.4329,
      "step": 449270
    },
    {
      "epoch": 183.91,
      "grad_norm": 2.299041509628296,
      "learning_rate": 8.250670491248825e-08,
      "loss": 0.4353,
      "step": 449280
    },
    {
      "epoch": 183.91,
      "grad_norm": 2.26235294342041,
      "learning_rate": 8.246499128024744e-08,
      "loss": 0.4447,
      "step": 449290
    },
    {
      "epoch": 183.91,
      "grad_norm": 2.1972341537475586,
      "learning_rate": 8.24232880066029e-08,
      "loss": 0.4577,
      "step": 449300
    },
    {
      "epoch": 183.92,
      "grad_norm": 2.0688817501068115,
      "learning_rate": 8.238159509174614e-08,
      "loss": 0.4394,
      "step": 449310
    },
    {
      "epoch": 183.92,
      "grad_norm": 2.066476583480835,
      "learning_rate": 8.233991253586737e-08,
      "loss": 0.4333,
      "step": 449320
    },
    {
      "epoch": 183.93,
      "grad_norm": 2.7071280479431152,
      "learning_rate": 8.229824033915838e-08,
      "loss": 0.4233,
      "step": 449330
    },
    {
      "epoch": 183.93,
      "grad_norm": 2.411710023880005,
      "learning_rate": 8.225657850180938e-08,
      "loss": 0.4339,
      "step": 449340
    },
    {
      "epoch": 183.93,
      "grad_norm": 2.0680227279663086,
      "learning_rate": 8.221492702401187e-08,
      "loss": 0.4268,
      "step": 449350
    },
    {
      "epoch": 183.94,
      "grad_norm": 1.8103643655776978,
      "learning_rate": 8.217328590595582e-08,
      "loss": 0.4536,
      "step": 449360
    },
    {
      "epoch": 183.94,
      "grad_norm": 1.999554991722107,
      "learning_rate": 8.213165514783272e-08,
      "loss": 0.4532,
      "step": 449370
    },
    {
      "epoch": 183.95,
      "grad_norm": 2.537369966506958,
      "learning_rate": 8.209003474983306e-08,
      "loss": 0.431,
      "step": 449380
    },
    {
      "epoch": 183.95,
      "grad_norm": 1.9298700094223022,
      "learning_rate": 8.204842471214732e-08,
      "loss": 0.4487,
      "step": 449390
    },
    {
      "epoch": 183.95,
      "grad_norm": 1.9062936305999756,
      "learning_rate": 8.200682503496621e-08,
      "loss": 0.4493,
      "step": 449400
    },
    {
      "epoch": 183.96,
      "grad_norm": 2.1771740913391113,
      "learning_rate": 8.196523571848074e-08,
      "loss": 0.4369,
      "step": 449410
    },
    {
      "epoch": 183.96,
      "grad_norm": 2.263181447982788,
      "learning_rate": 8.192365676288059e-08,
      "loss": 0.4428,
      "step": 449420
    },
    {
      "epoch": 183.97,
      "grad_norm": 2.2279529571533203,
      "learning_rate": 8.188208816835678e-08,
      "loss": 0.4315,
      "step": 449430
    },
    {
      "epoch": 183.97,
      "grad_norm": 1.695055365562439,
      "learning_rate": 8.184052993509922e-08,
      "loss": 0.4301,
      "step": 449440
    },
    {
      "epoch": 183.97,
      "grad_norm": 2.19903826713562,
      "learning_rate": 8.179898206329894e-08,
      "loss": 0.4492,
      "step": 449450
    },
    {
      "epoch": 183.98,
      "grad_norm": 1.9307386875152588,
      "learning_rate": 8.175744455314586e-08,
      "loss": 0.429,
      "step": 449460
    },
    {
      "epoch": 183.98,
      "grad_norm": 2.2200005054473877,
      "learning_rate": 8.171591740483022e-08,
      "loss": 0.4406,
      "step": 449470
    },
    {
      "epoch": 183.99,
      "grad_norm": 2.3725011348724365,
      "learning_rate": 8.167440061854248e-08,
      "loss": 0.437,
      "step": 449480
    },
    {
      "epoch": 183.99,
      "grad_norm": 1.9611996412277222,
      "learning_rate": 8.163289419447286e-08,
      "loss": 0.4525,
      "step": 449490
    },
    {
      "epoch": 184.0,
      "grad_norm": 2.4262301921844482,
      "learning_rate": 8.15913981328113e-08,
      "loss": 0.4604,
      "step": 449500
    },
    {
      "epoch": 184.0,
      "grad_norm": 2.227750301361084,
      "learning_rate": 8.154991243374802e-08,
      "loss": 0.441,
      "step": 449510
    },
    {
      "epoch": 184.0,
      "eval_loss": 0.4433809518814087,
      "eval_runtime": 52.3723,
      "eval_samples_per_second": 65.855,
      "eval_steps_per_second": 8.249,
      "step": 449512
    },
    {
      "epoch": 184.0,
      "grad_norm": 2.675607919692993,
      "learning_rate": 8.150843709747298e-08,
      "loss": 0.4651,
      "step": 449520
    },
    {
      "epoch": 184.01,
      "grad_norm": 2.316342830657959,
      "learning_rate": 8.146697212417638e-08,
      "loss": 0.4186,
      "step": 449530
    },
    {
      "epoch": 184.01,
      "grad_norm": 2.6303234100341797,
      "learning_rate": 8.142551751404766e-08,
      "loss": 0.4577,
      "step": 449540
    },
    {
      "epoch": 184.02,
      "grad_norm": 2.0557360649108887,
      "learning_rate": 8.138407326727727e-08,
      "loss": 0.4473,
      "step": 449550
    },
    {
      "epoch": 184.02,
      "grad_norm": 2.4561409950256348,
      "learning_rate": 8.134263938405495e-08,
      "loss": 0.4455,
      "step": 449560
    },
    {
      "epoch": 184.02,
      "grad_norm": 2.044391632080078,
      "learning_rate": 8.130121586457008e-08,
      "loss": 0.4499,
      "step": 449570
    },
    {
      "epoch": 184.03,
      "grad_norm": 1.9766590595245361,
      "learning_rate": 8.125980270901264e-08,
      "loss": 0.4285,
      "step": 449580
    },
    {
      "epoch": 184.03,
      "grad_norm": 1.917367696762085,
      "learning_rate": 8.121839991757258e-08,
      "loss": 0.4456,
      "step": 449590
    },
    {
      "epoch": 184.04,
      "grad_norm": 1.732257604598999,
      "learning_rate": 8.117700749043933e-08,
      "loss": 0.4355,
      "step": 449600
    },
    {
      "epoch": 184.04,
      "grad_norm": 2.1331582069396973,
      "learning_rate": 8.113562542780231e-08,
      "loss": 0.442,
      "step": 449610
    },
    {
      "epoch": 184.04,
      "grad_norm": 2.2154343128204346,
      "learning_rate": 8.109425372985175e-08,
      "loss": 0.4418,
      "step": 449620
    },
    {
      "epoch": 184.05,
      "grad_norm": 1.752208948135376,
      "learning_rate": 8.105289239677655e-08,
      "loss": 0.4358,
      "step": 449630
    },
    {
      "epoch": 184.05,
      "grad_norm": 2.0094962120056152,
      "learning_rate": 8.101154142876616e-08,
      "loss": 0.4401,
      "step": 449640
    },
    {
      "epoch": 184.06,
      "grad_norm": 2.0463099479675293,
      "learning_rate": 8.09702008260105e-08,
      "loss": 0.437,
      "step": 449650
    },
    {
      "epoch": 184.06,
      "grad_norm": 2.0539960861206055,
      "learning_rate": 8.092887058869825e-08,
      "loss": 0.4816,
      "step": 449660
    },
    {
      "epoch": 184.06,
      "grad_norm": 2.0574519634246826,
      "learning_rate": 8.088755071701961e-08,
      "loss": 0.4407,
      "step": 449670
    },
    {
      "epoch": 184.07,
      "grad_norm": 2.3754289150238037,
      "learning_rate": 8.08462412111635e-08,
      "loss": 0.4355,
      "step": 449680
    },
    {
      "epoch": 184.07,
      "grad_norm": 1.9452283382415771,
      "learning_rate": 8.080494207131881e-08,
      "loss": 0.4453,
      "step": 449690
    },
    {
      "epoch": 184.08,
      "grad_norm": 1.8528302907943726,
      "learning_rate": 8.076365329767499e-08,
      "loss": 0.4399,
      "step": 449700
    },
    {
      "epoch": 184.08,
      "grad_norm": 2.745513916015625,
      "learning_rate": 8.072237489042121e-08,
      "loss": 0.4336,
      "step": 449710
    },
    {
      "epoch": 184.09,
      "grad_norm": 3.437084197998047,
      "learning_rate": 8.068110684974665e-08,
      "loss": 0.4373,
      "step": 449720
    },
    {
      "epoch": 184.09,
      "grad_norm": 2.287090301513672,
      "learning_rate": 8.063984917583994e-08,
      "loss": 0.4192,
      "step": 449730
    },
    {
      "epoch": 184.09,
      "grad_norm": 2.1596598625183105,
      "learning_rate": 8.059860186889053e-08,
      "loss": 0.4726,
      "step": 449740
    },
    {
      "epoch": 184.1,
      "grad_norm": 1.888553500175476,
      "learning_rate": 8.055736492908705e-08,
      "loss": 0.4454,
      "step": 449750
    },
    {
      "epoch": 184.1,
      "grad_norm": 1.8810501098632812,
      "learning_rate": 8.051613835661844e-08,
      "loss": 0.4362,
      "step": 449760
    },
    {
      "epoch": 184.11,
      "grad_norm": 2.2253293991088867,
      "learning_rate": 8.047492215167384e-08,
      "loss": 0.4388,
      "step": 449770
    },
    {
      "epoch": 184.11,
      "grad_norm": 2.583357572555542,
      "learning_rate": 8.043371631444168e-08,
      "loss": 0.4463,
      "step": 449780
    },
    {
      "epoch": 184.11,
      "grad_norm": 2.035252809524536,
      "learning_rate": 8.03925208451111e-08,
      "loss": 0.4363,
      "step": 449790
    },
    {
      "epoch": 184.12,
      "grad_norm": 1.9125028848648071,
      "learning_rate": 8.035133574387049e-08,
      "loss": 0.4504,
      "step": 449800
    },
    {
      "epoch": 184.12,
      "grad_norm": 1.905135154724121,
      "learning_rate": 8.031016101090852e-08,
      "loss": 0.436,
      "step": 449810
    },
    {
      "epoch": 184.13,
      "grad_norm": 1.7365792989730835,
      "learning_rate": 8.02689966464141e-08,
      "loss": 0.4451,
      "step": 449820
    },
    {
      "epoch": 184.13,
      "grad_norm": 2.223798990249634,
      "learning_rate": 8.02278426505756e-08,
      "loss": 0.4532,
      "step": 449830
    },
    {
      "epoch": 184.13,
      "grad_norm": 1.9816850423812866,
      "learning_rate": 8.018669902358143e-08,
      "loss": 0.4505,
      "step": 449840
    },
    {
      "epoch": 184.14,
      "grad_norm": 2.1985490322113037,
      "learning_rate": 8.014556576562024e-08,
      "loss": 0.4339,
      "step": 449850
    },
    {
      "epoch": 184.14,
      "grad_norm": 1.879905104637146,
      "learning_rate": 8.010444287688042e-08,
      "loss": 0.445,
      "step": 449860
    },
    {
      "epoch": 184.15,
      "grad_norm": 2.1276016235351562,
      "learning_rate": 8.006333035755034e-08,
      "loss": 0.439,
      "step": 449870
    },
    {
      "epoch": 184.15,
      "grad_norm": 2.2943146228790283,
      "learning_rate": 8.002222820781843e-08,
      "loss": 0.4343,
      "step": 449880
    },
    {
      "epoch": 184.15,
      "grad_norm": 2.0104761123657227,
      "learning_rate": 7.998113642787279e-08,
      "loss": 0.4394,
      "step": 449890
    },
    {
      "epoch": 184.16,
      "grad_norm": 2.060579538345337,
      "learning_rate": 7.994005501790155e-08,
      "loss": 0.4328,
      "step": 449900
    },
    {
      "epoch": 184.16,
      "grad_norm": 2.183079719543457,
      "learning_rate": 7.989898397809339e-08,
      "loss": 0.4494,
      "step": 449910
    },
    {
      "epoch": 184.17,
      "grad_norm": 2.081984281539917,
      "learning_rate": 7.98579233086359e-08,
      "loss": 0.4504,
      "step": 449920
    },
    {
      "epoch": 184.17,
      "grad_norm": 2.2291557788848877,
      "learning_rate": 7.981687300971746e-08,
      "loss": 0.4452,
      "step": 449930
    },
    {
      "epoch": 184.18,
      "grad_norm": 2.518493413925171,
      "learning_rate": 7.977583308152597e-08,
      "loss": 0.4367,
      "step": 449940
    },
    {
      "epoch": 184.18,
      "grad_norm": 3.1219139099121094,
      "learning_rate": 7.973480352424953e-08,
      "loss": 0.4457,
      "step": 449950
    },
    {
      "epoch": 184.18,
      "grad_norm": 1.7811466455459595,
      "learning_rate": 7.969378433807628e-08,
      "loss": 0.4475,
      "step": 449960
    },
    {
      "epoch": 184.19,
      "grad_norm": 1.962518572807312,
      "learning_rate": 7.96527755231936e-08,
      "loss": 0.4268,
      "step": 449970
    },
    {
      "epoch": 184.19,
      "grad_norm": 2.652189016342163,
      "learning_rate": 7.961177707978984e-08,
      "loss": 0.4755,
      "step": 449980
    },
    {
      "epoch": 184.2,
      "grad_norm": 2.033369302749634,
      "learning_rate": 7.957078900805288e-08,
      "loss": 0.4368,
      "step": 449990
    },
    {
      "epoch": 184.2,
      "grad_norm": 2.173349142074585,
      "learning_rate": 7.95298113081701e-08,
      "loss": 0.4266,
      "step": 450000
    },
    {
      "epoch": 184.2,
      "grad_norm": 2.0911953449249268,
      "learning_rate": 7.948884398032932e-08,
      "loss": 0.4605,
      "step": 450010
    },
    {
      "epoch": 184.21,
      "grad_norm": 1.786388635635376,
      "learning_rate": 7.944788702471819e-08,
      "loss": 0.4393,
      "step": 450020
    },
    {
      "epoch": 184.21,
      "grad_norm": 1.8785244226455688,
      "learning_rate": 7.940694044152405e-08,
      "loss": 0.4443,
      "step": 450030
    },
    {
      "epoch": 184.22,
      "grad_norm": 2.422086477279663,
      "learning_rate": 7.936600423093502e-08,
      "loss": 0.441,
      "step": 450040
    },
    {
      "epoch": 184.22,
      "grad_norm": 2.008789539337158,
      "learning_rate": 7.932507839313847e-08,
      "loss": 0.4615,
      "step": 450050
    },
    {
      "epoch": 184.22,
      "grad_norm": 1.8456778526306152,
      "learning_rate": 7.928416292832174e-08,
      "loss": 0.4409,
      "step": 450060
    },
    {
      "epoch": 184.23,
      "grad_norm": 1.9530574083328247,
      "learning_rate": 7.924325783667242e-08,
      "loss": 0.4367,
      "step": 450070
    },
    {
      "epoch": 184.23,
      "grad_norm": 2.3241312503814697,
      "learning_rate": 7.920236311837764e-08,
      "loss": 0.4246,
      "step": 450080
    },
    {
      "epoch": 184.24,
      "grad_norm": 2.020935535430908,
      "learning_rate": 7.916147877362499e-08,
      "loss": 0.4279,
      "step": 450090
    },
    {
      "epoch": 184.24,
      "grad_norm": 2.1987156867980957,
      "learning_rate": 7.912060480260155e-08,
      "loss": 0.4195,
      "step": 450100
    },
    {
      "epoch": 184.24,
      "grad_norm": 2.006263256072998,
      "learning_rate": 7.907974120549467e-08,
      "loss": 0.4342,
      "step": 450110
    },
    {
      "epoch": 184.25,
      "grad_norm": 2.0405421257019043,
      "learning_rate": 7.903888798249146e-08,
      "loss": 0.4522,
      "step": 450120
    },
    {
      "epoch": 184.25,
      "grad_norm": 2.1664700508117676,
      "learning_rate": 7.899804513377952e-08,
      "loss": 0.4561,
      "step": 450130
    },
    {
      "epoch": 184.26,
      "grad_norm": 2.0390732288360596,
      "learning_rate": 7.895721265954515e-08,
      "loss": 0.4333,
      "step": 450140
    },
    {
      "epoch": 184.26,
      "grad_norm": 1.643302321434021,
      "learning_rate": 7.891639055997598e-08,
      "loss": 0.4482,
      "step": 450150
    },
    {
      "epoch": 184.27,
      "grad_norm": 2.1995699405670166,
      "learning_rate": 7.887557883525857e-08,
      "loss": 0.4259,
      "step": 450160
    },
    {
      "epoch": 184.27,
      "grad_norm": 2.1785225868225098,
      "learning_rate": 7.883477748558027e-08,
      "loss": 0.4298,
      "step": 450170
    },
    {
      "epoch": 184.27,
      "grad_norm": 3.0870537757873535,
      "learning_rate": 7.879398651112791e-08,
      "loss": 0.4468,
      "step": 450180
    },
    {
      "epoch": 184.28,
      "grad_norm": 2.0285804271698,
      "learning_rate": 7.875320591208805e-08,
      "loss": 0.4566,
      "step": 450190
    },
    {
      "epoch": 184.28,
      "grad_norm": 2.0829756259918213,
      "learning_rate": 7.871243568864779e-08,
      "loss": 0.4657,
      "step": 450200
    },
    {
      "epoch": 184.29,
      "grad_norm": 2.1064634323120117,
      "learning_rate": 7.867167584099396e-08,
      "loss": 0.4472,
      "step": 450210
    },
    {
      "epoch": 184.29,
      "grad_norm": 1.8401683568954468,
      "learning_rate": 7.863092636931287e-08,
      "loss": 0.4197,
      "step": 450220
    },
    {
      "epoch": 184.29,
      "grad_norm": 1.8497012853622437,
      "learning_rate": 7.85901872737916e-08,
      "loss": 0.4481,
      "step": 450230
    },
    {
      "epoch": 184.3,
      "grad_norm": 1.9558532238006592,
      "learning_rate": 7.854945855461673e-08,
      "loss": 0.4575,
      "step": 450240
    },
    {
      "epoch": 184.3,
      "grad_norm": 1.9518189430236816,
      "learning_rate": 7.850874021197483e-08,
      "loss": 0.4502,
      "step": 450250
    },
    {
      "epoch": 184.31,
      "grad_norm": 2.064558744430542,
      "learning_rate": 7.846803224605195e-08,
      "loss": 0.4319,
      "step": 450260
    },
    {
      "epoch": 184.31,
      "grad_norm": 2.04121994972229,
      "learning_rate": 7.842733465703569e-08,
      "loss": 0.4482,
      "step": 450270
    },
    {
      "epoch": 184.31,
      "grad_norm": 2.002554178237915,
      "learning_rate": 7.838664744511105e-08,
      "loss": 0.4315,
      "step": 450280
    },
    {
      "epoch": 184.32,
      "grad_norm": 2.173996925354004,
      "learning_rate": 7.834597061046541e-08,
      "loss": 0.4528,
      "step": 450290
    },
    {
      "epoch": 184.32,
      "grad_norm": 2.3160970211029053,
      "learning_rate": 7.830530415328454e-08,
      "loss": 0.4447,
      "step": 450300
    },
    {
      "epoch": 184.33,
      "grad_norm": 1.9545233249664307,
      "learning_rate": 7.826464807375501e-08,
      "loss": 0.4503,
      "step": 450310
    },
    {
      "epoch": 184.33,
      "grad_norm": 1.7779000997543335,
      "learning_rate": 7.822400237206313e-08,
      "loss": 0.4457,
      "step": 450320
    },
    {
      "epoch": 184.33,
      "grad_norm": 2.4857547283172607,
      "learning_rate": 7.818336704839495e-08,
      "loss": 0.4499,
      "step": 450330
    },
    {
      "epoch": 184.34,
      "grad_norm": 2.3107519149780273,
      "learning_rate": 7.814274210293653e-08,
      "loss": 0.4237,
      "step": 450340
    },
    {
      "epoch": 184.34,
      "grad_norm": 1.9663408994674683,
      "learning_rate": 7.810212753587417e-08,
      "loss": 0.4492,
      "step": 450350
    },
    {
      "epoch": 184.35,
      "grad_norm": 2.057121753692627,
      "learning_rate": 7.80615233473939e-08,
      "loss": 0.4497,
      "step": 450360
    },
    {
      "epoch": 184.35,
      "grad_norm": 1.9829057455062866,
      "learning_rate": 7.80209295376818e-08,
      "loss": 0.4457,
      "step": 450370
    },
    {
      "epoch": 184.36,
      "grad_norm": 1.7711085081100464,
      "learning_rate": 7.798034610692339e-08,
      "loss": 0.4548,
      "step": 450380
    },
    {
      "epoch": 184.36,
      "grad_norm": 2.30544376373291,
      "learning_rate": 7.793977305530497e-08,
      "loss": 0.4475,
      "step": 450390
    },
    {
      "epoch": 184.36,
      "grad_norm": 2.1155178546905518,
      "learning_rate": 7.789921038301233e-08,
      "loss": 0.4424,
      "step": 450400
    },
    {
      "epoch": 184.37,
      "grad_norm": 2.1949875354766846,
      "learning_rate": 7.785865809023127e-08,
      "loss": 0.4538,
      "step": 450410
    },
    {
      "epoch": 184.37,
      "grad_norm": 1.9101976156234741,
      "learning_rate": 7.78181161771473e-08,
      "loss": 0.4355,
      "step": 450420
    },
    {
      "epoch": 184.38,
      "grad_norm": 1.8583552837371826,
      "learning_rate": 7.77775846439465e-08,
      "loss": 0.4553,
      "step": 450430
    },
    {
      "epoch": 184.38,
      "grad_norm": 2.2348275184631348,
      "learning_rate": 7.773706349081462e-08,
      "loss": 0.4296,
      "step": 450440
    },
    {
      "epoch": 184.38,
      "grad_norm": 2.0296847820281982,
      "learning_rate": 7.769655271793671e-08,
      "loss": 0.4546,
      "step": 450450
    },
    {
      "epoch": 184.39,
      "grad_norm": 1.8669583797454834,
      "learning_rate": 7.765605232549905e-08,
      "loss": 0.4265,
      "step": 450460
    },
    {
      "epoch": 184.39,
      "grad_norm": 2.113151788711548,
      "learning_rate": 7.761556231368665e-08,
      "loss": 0.4153,
      "step": 450470
    },
    {
      "epoch": 184.4,
      "grad_norm": 1.9096324443817139,
      "learning_rate": 7.757508268268503e-08,
      "loss": 0.4566,
      "step": 450480
    },
    {
      "epoch": 184.4,
      "grad_norm": 2.230560541152954,
      "learning_rate": 7.753461343267975e-08,
      "loss": 0.4557,
      "step": 450490
    },
    {
      "epoch": 184.4,
      "grad_norm": 2.107788562774658,
      "learning_rate": 7.749415456385633e-08,
      "loss": 0.4225,
      "step": 450500
    },
    {
      "epoch": 184.41,
      "grad_norm": 2.440612554550171,
      "learning_rate": 7.745370607640002e-08,
      "loss": 0.425,
      "step": 450510
    },
    {
      "epoch": 184.41,
      "grad_norm": 1.9008424282073975,
      "learning_rate": 7.741326797049583e-08,
      "loss": 0.4635,
      "step": 450520
    },
    {
      "epoch": 184.42,
      "grad_norm": 2.1911683082580566,
      "learning_rate": 7.737284024632957e-08,
      "loss": 0.45,
      "step": 450530
    },
    {
      "epoch": 184.42,
      "grad_norm": 2.244434356689453,
      "learning_rate": 7.733242290408571e-08,
      "loss": 0.4505,
      "step": 450540
    },
    {
      "epoch": 184.42,
      "grad_norm": 2.3986825942993164,
      "learning_rate": 7.729201594395005e-08,
      "loss": 0.4498,
      "step": 450550
    },
    {
      "epoch": 184.43,
      "grad_norm": 1.8514615297317505,
      "learning_rate": 7.725161936610732e-08,
      "loss": 0.4219,
      "step": 450560
    },
    {
      "epoch": 184.43,
      "grad_norm": 1.9283807277679443,
      "learning_rate": 7.721123317074255e-08,
      "loss": 0.4445,
      "step": 450570
    },
    {
      "epoch": 184.44,
      "grad_norm": 2.0064198970794678,
      "learning_rate": 7.717085735804126e-08,
      "loss": 0.4204,
      "step": 450580
    },
    {
      "epoch": 184.44,
      "grad_norm": 1.9032098054885864,
      "learning_rate": 7.713049192818767e-08,
      "loss": 0.4672,
      "step": 450590
    },
    {
      "epoch": 184.45,
      "grad_norm": 2.5294225215911865,
      "learning_rate": 7.70901368813668e-08,
      "loss": 0.4235,
      "step": 450600
    },
    {
      "epoch": 184.45,
      "grad_norm": 1.9390231370925903,
      "learning_rate": 7.70497922177639e-08,
      "loss": 0.4229,
      "step": 450610
    },
    {
      "epoch": 184.45,
      "grad_norm": 1.9352740049362183,
      "learning_rate": 7.700945793756375e-08,
      "loss": 0.4163,
      "step": 450620
    },
    {
      "epoch": 184.46,
      "grad_norm": 1.9998247623443604,
      "learning_rate": 7.696913404095082e-08,
      "loss": 0.4307,
      "step": 450630
    },
    {
      "epoch": 184.46,
      "grad_norm": 1.9375646114349365,
      "learning_rate": 7.69288205281101e-08,
      "loss": 0.4311,
      "step": 450640
    },
    {
      "epoch": 184.47,
      "grad_norm": 2.021359443664551,
      "learning_rate": 7.688851739922585e-08,
      "loss": 0.4363,
      "step": 450650
    },
    {
      "epoch": 184.47,
      "grad_norm": 2.146200656890869,
      "learning_rate": 7.68482246544833e-08,
      "loss": 0.4263,
      "step": 450660
    },
    {
      "epoch": 184.47,
      "grad_norm": 1.675129771232605,
      "learning_rate": 7.680794229406673e-08,
      "loss": 0.4518,
      "step": 450670
    },
    {
      "epoch": 184.48,
      "grad_norm": 2.031712293624878,
      "learning_rate": 7.67676703181606e-08,
      "loss": 0.4475,
      "step": 450680
    },
    {
      "epoch": 184.48,
      "grad_norm": 1.5850176811218262,
      "learning_rate": 7.672740872694937e-08,
      "loss": 0.4301,
      "step": 450690
    },
    {
      "epoch": 184.49,
      "grad_norm": 5.867962837219238,
      "learning_rate": 7.668715752061732e-08,
      "loss": 0.4258,
      "step": 450700
    },
    {
      "epoch": 184.49,
      "grad_norm": 1.9791752099990845,
      "learning_rate": 7.664691669934968e-08,
      "loss": 0.4364,
      "step": 450710
    },
    {
      "epoch": 184.49,
      "grad_norm": 1.7939095497131348,
      "learning_rate": 7.660668626332964e-08,
      "loss": 0.4704,
      "step": 450720
    },
    {
      "epoch": 184.5,
      "grad_norm": 2.3735005855560303,
      "learning_rate": 7.656646621274223e-08,
      "loss": 0.4464,
      "step": 450730
    },
    {
      "epoch": 184.5,
      "grad_norm": 2.233088493347168,
      "learning_rate": 7.652625654777141e-08,
      "loss": 0.4369,
      "step": 450740
    },
    {
      "epoch": 184.51,
      "grad_norm": 1.7821838855743408,
      "learning_rate": 7.64860572686014e-08,
      "loss": 0.4508,
      "step": 450750
    },
    {
      "epoch": 184.51,
      "grad_norm": 2.3482604026794434,
      "learning_rate": 7.644586837541643e-08,
      "loss": 0.4345,
      "step": 450760
    },
    {
      "epoch": 184.51,
      "grad_norm": 2.508399486541748,
      "learning_rate": 7.640568986840046e-08,
      "loss": 0.4381,
      "step": 450770
    },
    {
      "epoch": 184.52,
      "grad_norm": 1.9582319259643555,
      "learning_rate": 7.636552174773748e-08,
      "loss": 0.4429,
      "step": 450780
    },
    {
      "epoch": 184.52,
      "grad_norm": 1.8608592748641968,
      "learning_rate": 7.63253640136117e-08,
      "loss": 0.4516,
      "step": 450790
    },
    {
      "epoch": 184.53,
      "grad_norm": 2.151155948638916,
      "learning_rate": 7.628521666620708e-08,
      "loss": 0.4506,
      "step": 450800
    },
    {
      "epoch": 184.53,
      "grad_norm": 2.4359121322631836,
      "learning_rate": 7.62450797057076e-08,
      "loss": 0.4394,
      "step": 450810
    },
    {
      "epoch": 184.54,
      "grad_norm": 2.6770429611206055,
      "learning_rate": 7.620495313229698e-08,
      "loss": 0.4518,
      "step": 450820
    },
    {
      "epoch": 184.54,
      "grad_norm": 2.075388193130493,
      "learning_rate": 7.61648369461589e-08,
      "loss": 0.4308,
      "step": 450830
    },
    {
      "epoch": 184.54,
      "grad_norm": 2.201889991760254,
      "learning_rate": 7.612473114747735e-08,
      "loss": 0.4265,
      "step": 450840
    },
    {
      "epoch": 184.55,
      "grad_norm": 2.2547004222869873,
      "learning_rate": 7.608463573643627e-08,
      "loss": 0.4468,
      "step": 450850
    },
    {
      "epoch": 184.55,
      "grad_norm": 2.1071765422821045,
      "learning_rate": 7.604455071321888e-08,
      "loss": 0.4331,
      "step": 450860
    },
    {
      "epoch": 184.56,
      "grad_norm": 2.2609047889709473,
      "learning_rate": 7.600447607800887e-08,
      "loss": 0.4327,
      "step": 450870
    },
    {
      "epoch": 184.56,
      "grad_norm": 2.2523787021636963,
      "learning_rate": 7.596441183098994e-08,
      "loss": 0.4556,
      "step": 450880
    },
    {
      "epoch": 184.56,
      "grad_norm": 1.8367133140563965,
      "learning_rate": 7.592435797234529e-08,
      "loss": 0.4298,
      "step": 450890
    },
    {
      "epoch": 184.57,
      "grad_norm": 2.0424556732177734,
      "learning_rate": 7.588431450225915e-08,
      "loss": 0.4556,
      "step": 450900
    },
    {
      "epoch": 184.57,
      "grad_norm": 1.969618320465088,
      "learning_rate": 7.584428142091416e-08,
      "loss": 0.4508,
      "step": 450910
    },
    {
      "epoch": 184.58,
      "grad_norm": 2.0192782878875732,
      "learning_rate": 7.580425872849407e-08,
      "loss": 0.4295,
      "step": 450920
    },
    {
      "epoch": 184.58,
      "grad_norm": 2.457817792892456,
      "learning_rate": 7.576424642518229e-08,
      "loss": 0.4468,
      "step": 450930
    },
    {
      "epoch": 184.58,
      "grad_norm": 2.215057373046875,
      "learning_rate": 7.572424451116203e-08,
      "loss": 0.4462,
      "step": 450940
    },
    {
      "epoch": 184.59,
      "grad_norm": 2.326810121536255,
      "learning_rate": 7.568425298661619e-08,
      "loss": 0.4649,
      "step": 450950
    },
    {
      "epoch": 184.59,
      "grad_norm": 2.089472532272339,
      "learning_rate": 7.564427185172877e-08,
      "loss": 0.4492,
      "step": 450960
    },
    {
      "epoch": 184.6,
      "grad_norm": 1.983587384223938,
      "learning_rate": 7.560430110668215e-08,
      "loss": 0.4386,
      "step": 450970
    },
    {
      "epoch": 184.6,
      "grad_norm": 2.0367588996887207,
      "learning_rate": 7.55643407516598e-08,
      "loss": 0.4505,
      "step": 450980
    },
    {
      "epoch": 184.6,
      "grad_norm": 1.5831878185272217,
      "learning_rate": 7.552439078684462e-08,
      "loss": 0.4404,
      "step": 450990
    },
    {
      "epoch": 184.61,
      "grad_norm": 2.2397890090942383,
      "learning_rate": 7.548445121241956e-08,
      "loss": 0.4404,
      "step": 451000
    },
    {
      "epoch": 184.61,
      "grad_norm": 1.9752498865127563,
      "learning_rate": 7.54445220285678e-08,
      "loss": 0.4558,
      "step": 451010
    },
    {
      "epoch": 184.62,
      "grad_norm": 1.8563717603683472,
      "learning_rate": 7.540460323547226e-08,
      "loss": 0.4449,
      "step": 451020
    },
    {
      "epoch": 184.62,
      "grad_norm": 1.803507924079895,
      "learning_rate": 7.536469483331561e-08,
      "loss": 0.4244,
      "step": 451030
    },
    {
      "epoch": 184.63,
      "grad_norm": 2.0912790298461914,
      "learning_rate": 7.532479682228051e-08,
      "loss": 0.4504,
      "step": 451040
    },
    {
      "epoch": 184.63,
      "grad_norm": 1.8094089031219482,
      "learning_rate": 7.528490920255016e-08,
      "loss": 0.4359,
      "step": 451050
    },
    {
      "epoch": 184.63,
      "grad_norm": 2.2268149852752686,
      "learning_rate": 7.524503197430722e-08,
      "loss": 0.4381,
      "step": 451060
    },
    {
      "epoch": 184.64,
      "grad_norm": 1.9432623386383057,
      "learning_rate": 7.52051651377341e-08,
      "loss": 0.437,
      "step": 451070
    },
    {
      "epoch": 184.64,
      "grad_norm": 1.5997244119644165,
      "learning_rate": 7.516530869301346e-08,
      "loss": 0.4343,
      "step": 451080
    },
    {
      "epoch": 184.65,
      "grad_norm": 1.7732335329055786,
      "learning_rate": 7.512546264032796e-08,
      "loss": 0.4318,
      "step": 451090
    },
    {
      "epoch": 184.65,
      "grad_norm": 2.116217613220215,
      "learning_rate": 7.50856269798603e-08,
      "loss": 0.4572,
      "step": 451100
    },
    {
      "epoch": 184.65,
      "grad_norm": 1.8391622304916382,
      "learning_rate": 7.504580171179259e-08,
      "loss": 0.4235,
      "step": 451110
    },
    {
      "epoch": 184.66,
      "grad_norm": 1.8639309406280518,
      "learning_rate": 7.500598683630749e-08,
      "loss": 0.4548,
      "step": 451120
    },
    {
      "epoch": 184.66,
      "grad_norm": 2.0952725410461426,
      "learning_rate": 7.496618235358744e-08,
      "loss": 0.4481,
      "step": 451130
    },
    {
      "epoch": 184.67,
      "grad_norm": 2.05861496925354,
      "learning_rate": 7.492638826381456e-08,
      "loss": 0.4601,
      "step": 451140
    },
    {
      "epoch": 184.67,
      "grad_norm": 2.510809898376465,
      "learning_rate": 7.488660456717155e-08,
      "loss": 0.4457,
      "step": 451150
    },
    {
      "epoch": 184.67,
      "grad_norm": 2.045515537261963,
      "learning_rate": 7.484683126384025e-08,
      "loss": 0.4371,
      "step": 451160
    },
    {
      "epoch": 184.68,
      "grad_norm": 2.802628993988037,
      "learning_rate": 7.480706835400283e-08,
      "loss": 0.4309,
      "step": 451170
    },
    {
      "epoch": 184.68,
      "grad_norm": 2.143678665161133,
      "learning_rate": 7.47673158378417e-08,
      "loss": 0.4498,
      "step": 451180
    },
    {
      "epoch": 184.69,
      "grad_norm": 2.665215253829956,
      "learning_rate": 7.472757371553874e-08,
      "loss": 0.4273,
      "step": 451190
    },
    {
      "epoch": 184.69,
      "grad_norm": 1.8601083755493164,
      "learning_rate": 7.468784198727609e-08,
      "loss": 0.4424,
      "step": 451200
    },
    {
      "epoch": 184.7,
      "grad_norm": 2.104888439178467,
      "learning_rate": 7.46481206532359e-08,
      "loss": 0.4376,
      "step": 451210
    },
    {
      "epoch": 184.7,
      "grad_norm": 1.935205340385437,
      "learning_rate": 7.460840971360008e-08,
      "loss": 0.4371,
      "step": 451220
    },
    {
      "epoch": 184.7,
      "grad_norm": 1.9167522192001343,
      "learning_rate": 7.456870916855046e-08,
      "loss": 0.4493,
      "step": 451230
    },
    {
      "epoch": 184.71,
      "grad_norm": 2.2175142765045166,
      "learning_rate": 7.452901901826872e-08,
      "loss": 0.4472,
      "step": 451240
    },
    {
      "epoch": 184.71,
      "grad_norm": 1.9062265157699585,
      "learning_rate": 7.448933926293698e-08,
      "loss": 0.4424,
      "step": 451250
    },
    {
      "epoch": 184.72,
      "grad_norm": 1.7638713121414185,
      "learning_rate": 7.444966990273688e-08,
      "loss": 0.4655,
      "step": 451260
    },
    {
      "epoch": 184.72,
      "grad_norm": 2.0715365409851074,
      "learning_rate": 7.441001093785028e-08,
      "loss": 0.4494,
      "step": 451270
    },
    {
      "epoch": 184.72,
      "grad_norm": 2.168797254562378,
      "learning_rate": 7.437036236845858e-08,
      "loss": 0.4492,
      "step": 451280
    },
    {
      "epoch": 184.73,
      "grad_norm": 2.237947940826416,
      "learning_rate": 7.433072419474363e-08,
      "loss": 0.4492,
      "step": 451290
    },
    {
      "epoch": 184.73,
      "grad_norm": 2.4771816730499268,
      "learning_rate": 7.429109641688682e-08,
      "loss": 0.4425,
      "step": 451300
    },
    {
      "epoch": 184.74,
      "grad_norm": 2.3075509071350098,
      "learning_rate": 7.425147903506977e-08,
      "loss": 0.4381,
      "step": 451310
    },
    {
      "epoch": 184.74,
      "grad_norm": 1.9234713315963745,
      "learning_rate": 7.42118720494741e-08,
      "loss": 0.4422,
      "step": 451320
    },
    {
      "epoch": 184.74,
      "grad_norm": 2.0063462257385254,
      "learning_rate": 7.417227546028092e-08,
      "loss": 0.4654,
      "step": 451330
    },
    {
      "epoch": 184.75,
      "grad_norm": 2.2432010173797607,
      "learning_rate": 7.413268926767185e-08,
      "loss": 0.4407,
      "step": 451340
    },
    {
      "epoch": 184.75,
      "grad_norm": 2.005297899246216,
      "learning_rate": 7.4093113471828e-08,
      "loss": 0.4443,
      "step": 451350
    },
    {
      "epoch": 184.76,
      "grad_norm": 1.879790186882019,
      "learning_rate": 7.4053548072931e-08,
      "loss": 0.4741,
      "step": 451360
    },
    {
      "epoch": 184.76,
      "grad_norm": 2.258087158203125,
      "learning_rate": 7.401399307116222e-08,
      "loss": 0.4458,
      "step": 451370
    },
    {
      "epoch": 184.76,
      "grad_norm": 2.111741542816162,
      "learning_rate": 7.397444846670222e-08,
      "loss": 0.4305,
      "step": 451380
    },
    {
      "epoch": 184.77,
      "grad_norm": 2.263685703277588,
      "learning_rate": 7.393491425973265e-08,
      "loss": 0.4447,
      "step": 451390
    },
    {
      "epoch": 184.77,
      "grad_norm": 2.58477783203125,
      "learning_rate": 7.38953904504346e-08,
      "loss": 0.4365,
      "step": 451400
    },
    {
      "epoch": 184.78,
      "grad_norm": 2.837127447128296,
      "learning_rate": 7.385587703898866e-08,
      "loss": 0.4322,
      "step": 451410
    },
    {
      "epoch": 184.78,
      "grad_norm": 2.2291364669799805,
      "learning_rate": 7.381637402557646e-08,
      "loss": 0.4388,
      "step": 451420
    },
    {
      "epoch": 184.79,
      "grad_norm": 2.40809965133667,
      "learning_rate": 7.377688141037884e-08,
      "loss": 0.4473,
      "step": 451430
    },
    {
      "epoch": 184.79,
      "grad_norm": 2.0262653827667236,
      "learning_rate": 7.373739919357613e-08,
      "loss": 0.4382,
      "step": 451440
    },
    {
      "epoch": 184.79,
      "grad_norm": 2.1643240451812744,
      "learning_rate": 7.369792737534968e-08,
      "loss": 0.4602,
      "step": 451450
    },
    {
      "epoch": 184.8,
      "grad_norm": 1.810963749885559,
      "learning_rate": 7.365846595588035e-08,
      "loss": 0.4441,
      "step": 451460
    },
    {
      "epoch": 184.8,
      "grad_norm": 2.264657497406006,
      "learning_rate": 7.361901493534847e-08,
      "loss": 0.4269,
      "step": 451470
    },
    {
      "epoch": 184.81,
      "grad_norm": 2.1391117572784424,
      "learning_rate": 7.357957431393539e-08,
      "loss": 0.4533,
      "step": 451480
    },
    {
      "epoch": 184.81,
      "grad_norm": 1.973445177078247,
      "learning_rate": 7.354014409182117e-08,
      "loss": 0.4345,
      "step": 451490
    },
    {
      "epoch": 184.81,
      "grad_norm": 2.0302488803863525,
      "learning_rate": 7.350072426918693e-08,
      "loss": 0.4432,
      "step": 451500
    },
    {
      "epoch": 184.82,
      "grad_norm": 2.732210874557495,
      "learning_rate": 7.346131484621299e-08,
      "loss": 0.4465,
      "step": 451510
    },
    {
      "epoch": 184.82,
      "grad_norm": 3.207092761993408,
      "learning_rate": 7.342191582307968e-08,
      "loss": 0.4283,
      "step": 451520
    },
    {
      "epoch": 184.83,
      "grad_norm": 2.163377285003662,
      "learning_rate": 7.338252719996784e-08,
      "loss": 0.4355,
      "step": 451530
    },
    {
      "epoch": 184.83,
      "grad_norm": 2.28857684135437,
      "learning_rate": 7.334314897705779e-08,
      "loss": 0.4092,
      "step": 451540
    },
    {
      "epoch": 184.83,
      "grad_norm": 1.8029581308364868,
      "learning_rate": 7.330378115452985e-08,
      "loss": 0.4488,
      "step": 451550
    },
    {
      "epoch": 184.84,
      "grad_norm": 1.9888814687728882,
      "learning_rate": 7.326442373256462e-08,
      "loss": 0.4569,
      "step": 451560
    },
    {
      "epoch": 184.84,
      "grad_norm": 2.092716932296753,
      "learning_rate": 7.322507671134215e-08,
      "loss": 0.4526,
      "step": 451570
    },
    {
      "epoch": 184.85,
      "grad_norm": 1.770106554031372,
      "learning_rate": 7.318574009104278e-08,
      "loss": 0.428,
      "step": 451580
    },
    {
      "epoch": 184.85,
      "grad_norm": 1.7756009101867676,
      "learning_rate": 7.314641387184657e-08,
      "loss": 0.439,
      "step": 451590
    },
    {
      "epoch": 184.85,
      "grad_norm": 1.8873343467712402,
      "learning_rate": 7.310709805393357e-08,
      "loss": 0.4678,
      "step": 451600
    },
    {
      "epoch": 184.86,
      "grad_norm": 1.8045028448104858,
      "learning_rate": 7.306779263748412e-08,
      "loss": 0.4403,
      "step": 451610
    },
    {
      "epoch": 184.86,
      "grad_norm": 2.085789918899536,
      "learning_rate": 7.302849762267829e-08,
      "loss": 0.451,
      "step": 451620
    },
    {
      "epoch": 184.87,
      "grad_norm": 2.2982101440429688,
      "learning_rate": 7.298921300969612e-08,
      "loss": 0.4638,
      "step": 451630
    },
    {
      "epoch": 184.87,
      "grad_norm": 1.8659015893936157,
      "learning_rate": 7.294993879871716e-08,
      "loss": 0.4255,
      "step": 451640
    },
    {
      "epoch": 184.88,
      "grad_norm": 1.8085886240005493,
      "learning_rate": 7.291067498992175e-08,
      "loss": 0.4368,
      "step": 451650
    },
    {
      "epoch": 184.88,
      "grad_norm": 2.12286376953125,
      "learning_rate": 7.287142158348968e-08,
      "loss": 0.4624,
      "step": 451660
    },
    {
      "epoch": 184.88,
      "grad_norm": 2.3376495838165283,
      "learning_rate": 7.283217857960052e-08,
      "loss": 0.4381,
      "step": 451670
    },
    {
      "epoch": 184.89,
      "grad_norm": 2.3583717346191406,
      "learning_rate": 7.279294597843429e-08,
      "loss": 0.4463,
      "step": 451680
    },
    {
      "epoch": 184.89,
      "grad_norm": 2.119922399520874,
      "learning_rate": 7.275372378017057e-08,
      "loss": 0.4322,
      "step": 451690
    },
    {
      "epoch": 184.9,
      "grad_norm": 2.2049052715301514,
      "learning_rate": 7.271451198498915e-08,
      "loss": 0.4417,
      "step": 451700
    },
    {
      "epoch": 184.9,
      "grad_norm": 1.8098093271255493,
      "learning_rate": 7.267531059306957e-08,
      "loss": 0.4223,
      "step": 451710
    },
    {
      "epoch": 184.9,
      "grad_norm": 2.1923062801361084,
      "learning_rate": 7.263611960459139e-08,
      "loss": 0.4315,
      "step": 451720
    },
    {
      "epoch": 184.91,
      "grad_norm": 1.697883129119873,
      "learning_rate": 7.259693901973414e-08,
      "loss": 0.4523,
      "step": 451730
    },
    {
      "epoch": 184.91,
      "grad_norm": 2.0383355617523193,
      "learning_rate": 7.255776883867735e-08,
      "loss": 0.4494,
      "step": 451740
    },
    {
      "epoch": 184.92,
      "grad_norm": 2.000081777572632,
      "learning_rate": 7.25186090616006e-08,
      "loss": 0.4632,
      "step": 451750
    },
    {
      "epoch": 184.92,
      "grad_norm": 2.6390671730041504,
      "learning_rate": 7.24794596886829e-08,
      "loss": 0.4576,
      "step": 451760
    },
    {
      "epoch": 184.92,
      "grad_norm": 1.748643159866333,
      "learning_rate": 7.244032072010378e-08,
      "loss": 0.4386,
      "step": 451770
    },
    {
      "epoch": 184.93,
      "grad_norm": 2.2413275241851807,
      "learning_rate": 7.240119215604279e-08,
      "loss": 0.4603,
      "step": 451780
    },
    {
      "epoch": 184.93,
      "grad_norm": 1.9733952283859253,
      "learning_rate": 7.236207399667869e-08,
      "loss": 0.452,
      "step": 451790
    },
    {
      "epoch": 184.94,
      "grad_norm": 2.2707202434539795,
      "learning_rate": 7.232296624219103e-08,
      "loss": 0.4362,
      "step": 451800
    },
    {
      "epoch": 184.94,
      "grad_norm": 2.0332682132720947,
      "learning_rate": 7.228386889275884e-08,
      "loss": 0.4358,
      "step": 451810
    },
    {
      "epoch": 184.94,
      "grad_norm": 2.177722930908203,
      "learning_rate": 7.224478194856114e-08,
      "loss": 0.4625,
      "step": 451820
    },
    {
      "epoch": 184.95,
      "grad_norm": 1.94534170627594,
      "learning_rate": 7.220570540977696e-08,
      "loss": 0.4481,
      "step": 451830
    },
    {
      "epoch": 184.95,
      "grad_norm": 2.3205950260162354,
      "learning_rate": 7.216663927658555e-08,
      "loss": 0.4669,
      "step": 451840
    },
    {
      "epoch": 184.96,
      "grad_norm": 2.0839459896087646,
      "learning_rate": 7.21275835491657e-08,
      "loss": 0.4459,
      "step": 451850
    },
    {
      "epoch": 184.96,
      "grad_norm": 2.2496142387390137,
      "learning_rate": 7.208853822769644e-08,
      "loss": 0.4558,
      "step": 451860
    },
    {
      "epoch": 184.97,
      "grad_norm": 1.830744981765747,
      "learning_rate": 7.204950331235627e-08,
      "loss": 0.4344,
      "step": 451870
    },
    {
      "epoch": 184.97,
      "grad_norm": 2.3653578758239746,
      "learning_rate": 7.201047880332472e-08,
      "loss": 0.4582,
      "step": 451880
    },
    {
      "epoch": 184.97,
      "grad_norm": 1.706371784210205,
      "learning_rate": 7.197146470077977e-08,
      "loss": 0.4386,
      "step": 451890
    },
    {
      "epoch": 184.98,
      "grad_norm": 1.9473555088043213,
      "learning_rate": 7.193246100490046e-08,
      "loss": 0.4583,
      "step": 451900
    },
    {
      "epoch": 184.98,
      "grad_norm": 2.227146625518799,
      "learning_rate": 7.189346771586555e-08,
      "loss": 0.4513,
      "step": 451910
    },
    {
      "epoch": 184.99,
      "grad_norm": 2.321105718612671,
      "learning_rate": 7.185448483385328e-08,
      "loss": 0.4456,
      "step": 451920
    },
    {
      "epoch": 184.99,
      "grad_norm": 1.9717754125595093,
      "learning_rate": 7.181551235904267e-08,
      "loss": 0.4352,
      "step": 451930
    },
    {
      "epoch": 184.99,
      "grad_norm": 2.001519203186035,
      "learning_rate": 7.177655029161197e-08,
      "loss": 0.4575,
      "step": 451940
    },
    {
      "epoch": 185.0,
      "grad_norm": 2.4406888484954834,
      "learning_rate": 7.173759863174019e-08,
      "loss": 0.4546,
      "step": 451950
    },
    {
      "epoch": 185.0,
      "eval_loss": 0.44617268443107605,
      "eval_runtime": 52.1849,
      "eval_samples_per_second": 66.092,
      "eval_steps_per_second": 8.278,
      "step": 451955
    },
    {
      "epoch": 185.0,
      "grad_norm": 2.0590267181396484,
      "learning_rate": 7.169865737960507e-08,
      "loss": 0.451,
      "step": 451960
    },
    {
      "epoch": 185.01,
      "grad_norm": 2.1999576091766357,
      "learning_rate": 7.165972653538537e-08,
      "loss": 0.4329,
      "step": 451970
    },
    {
      "epoch": 185.01,
      "grad_norm": 3.0137712955474854,
      "learning_rate": 7.162080609925932e-08,
      "loss": 0.4205,
      "step": 451980
    },
    {
      "epoch": 185.01,
      "grad_norm": 2.210400342941284,
      "learning_rate": 7.158189607140516e-08,
      "loss": 0.4521,
      "step": 451990
    },
    {
      "epoch": 185.02,
      "grad_norm": 2.553109884262085,
      "learning_rate": 7.154299645200117e-08,
      "loss": 0.4512,
      "step": 452000
    },
    {
      "epoch": 185.02,
      "grad_norm": 1.9833893775939941,
      "learning_rate": 7.15041072412258e-08,
      "loss": 0.443,
      "step": 452010
    },
    {
      "epoch": 185.03,
      "grad_norm": 1.5540416240692139,
      "learning_rate": 7.146522843925681e-08,
      "loss": 0.4547,
      "step": 452020
    },
    {
      "epoch": 185.03,
      "grad_norm": 1.737957239151001,
      "learning_rate": 7.142636004627243e-08,
      "loss": 0.4416,
      "step": 452030
    },
    {
      "epoch": 185.03,
      "grad_norm": 1.9765969514846802,
      "learning_rate": 7.13875020624509e-08,
      "loss": 0.4321,
      "step": 452040
    },
    {
      "epoch": 185.04,
      "grad_norm": 2.1957767009735107,
      "learning_rate": 7.134865448796996e-08,
      "loss": 0.4727,
      "step": 452050
    },
    {
      "epoch": 185.04,
      "grad_norm": 2.1097803115844727,
      "learning_rate": 7.130981732300758e-08,
      "loss": 0.4454,
      "step": 452060
    },
    {
      "epoch": 185.05,
      "grad_norm": 2.0407378673553467,
      "learning_rate": 7.127099056774173e-08,
      "loss": 0.4419,
      "step": 452070
    },
    {
      "epoch": 185.05,
      "grad_norm": 2.4002685546875,
      "learning_rate": 7.123217422235015e-08,
      "loss": 0.4414,
      "step": 452080
    },
    {
      "epoch": 185.06,
      "grad_norm": 1.934645414352417,
      "learning_rate": 7.119336828701107e-08,
      "loss": 0.475,
      "step": 452090
    },
    {
      "epoch": 185.06,
      "grad_norm": 2.487833261489868,
      "learning_rate": 7.115457276190197e-08,
      "loss": 0.4508,
      "step": 452100
    },
    {
      "epoch": 185.06,
      "grad_norm": 2.0828964710235596,
      "learning_rate": 7.111578764720004e-08,
      "loss": 0.4388,
      "step": 452110
    },
    {
      "epoch": 185.07,
      "grad_norm": 1.982691764831543,
      "learning_rate": 7.107701294308405e-08,
      "loss": 0.4376,
      "step": 452120
    },
    {
      "epoch": 185.07,
      "grad_norm": 2.4252512454986572,
      "learning_rate": 7.103824864973067e-08,
      "loss": 0.4344,
      "step": 452130
    },
    {
      "epoch": 185.08,
      "grad_norm": 2.324838399887085,
      "learning_rate": 7.099949476731814e-08,
      "loss": 0.4343,
      "step": 452140
    },
    {
      "epoch": 185.08,
      "grad_norm": 2.0356154441833496,
      "learning_rate": 7.096075129602394e-08,
      "loss": 0.4468,
      "step": 452150
    },
    {
      "epoch": 185.08,
      "grad_norm": 2.4515442848205566,
      "learning_rate": 7.092201823602501e-08,
      "loss": 0.4562,
      "step": 452160
    },
    {
      "epoch": 185.09,
      "grad_norm": 1.7315231561660767,
      "learning_rate": 7.088329558749906e-08,
      "loss": 0.4386,
      "step": 452170
    },
    {
      "epoch": 185.09,
      "grad_norm": 1.9944370985031128,
      "learning_rate": 7.08445833506233e-08,
      "loss": 0.4397,
      "step": 452180
    },
    {
      "epoch": 185.1,
      "grad_norm": 1.991165041923523,
      "learning_rate": 7.080588152557544e-08,
      "loss": 0.4313,
      "step": 452190
    },
    {
      "epoch": 185.1,
      "grad_norm": 2.086432695388794,
      "learning_rate": 7.076719011253245e-08,
      "loss": 0.4398,
      "step": 452200
    },
    {
      "epoch": 185.1,
      "grad_norm": 2.2319717407226562,
      "learning_rate": 7.072850911167202e-08,
      "loss": 0.4461,
      "step": 452210
    },
    {
      "epoch": 185.11,
      "grad_norm": 1.7152135372161865,
      "learning_rate": 7.068983852317059e-08,
      "loss": 0.4428,
      "step": 452220
    },
    {
      "epoch": 185.11,
      "grad_norm": 1.9469503164291382,
      "learning_rate": 7.065117834720614e-08,
      "loss": 0.4486,
      "step": 452230
    },
    {
      "epoch": 185.12,
      "grad_norm": 2.2591161727905273,
      "learning_rate": 7.06125285839551e-08,
      "loss": 0.4504,
      "step": 452240
    },
    {
      "epoch": 185.12,
      "grad_norm": 2.251150131225586,
      "learning_rate": 7.05738892335949e-08,
      "loss": 0.4305,
      "step": 452250
    },
    {
      "epoch": 185.12,
      "grad_norm": 1.8061150312423706,
      "learning_rate": 7.053526029630226e-08,
      "loss": 0.4406,
      "step": 452260
    },
    {
      "epoch": 185.13,
      "grad_norm": 1.8522921800613403,
      "learning_rate": 7.049664177225436e-08,
      "loss": 0.4603,
      "step": 452270
    },
    {
      "epoch": 185.13,
      "grad_norm": 1.8946969509124756,
      "learning_rate": 7.045803366162815e-08,
      "loss": 0.45,
      "step": 452280
    },
    {
      "epoch": 185.14,
      "grad_norm": 2.2371304035186768,
      "learning_rate": 7.04194359646003e-08,
      "loss": 0.4469,
      "step": 452290
    },
    {
      "epoch": 185.14,
      "grad_norm": 2.048197031021118,
      "learning_rate": 7.038084868134751e-08,
      "loss": 0.4536,
      "step": 452300
    },
    {
      "epoch": 185.15,
      "grad_norm": 2.012057304382324,
      "learning_rate": 7.034227181204697e-08,
      "loss": 0.4443,
      "step": 452310
    },
    {
      "epoch": 185.15,
      "grad_norm": 1.8386752605438232,
      "learning_rate": 7.03037053568751e-08,
      "loss": 0.452,
      "step": 452320
    },
    {
      "epoch": 185.15,
      "grad_norm": 2.221060037612915,
      "learning_rate": 7.026514931600859e-08,
      "loss": 0.4627,
      "step": 452330
    },
    {
      "epoch": 185.16,
      "grad_norm": 1.747353196144104,
      "learning_rate": 7.022660368962412e-08,
      "loss": 0.4442,
      "step": 452340
    },
    {
      "epoch": 185.16,
      "grad_norm": 2.26131272315979,
      "learning_rate": 7.018806847789811e-08,
      "loss": 0.4424,
      "step": 452350
    },
    {
      "epoch": 185.17,
      "grad_norm": 2.0192582607269287,
      "learning_rate": 7.014954368100724e-08,
      "loss": 0.4428,
      "step": 452360
    },
    {
      "epoch": 185.17,
      "grad_norm": 2.4074032306671143,
      "learning_rate": 7.011102929912817e-08,
      "loss": 0.4707,
      "step": 452370
    },
    {
      "epoch": 185.17,
      "grad_norm": 2.2454090118408203,
      "learning_rate": 7.007252533243685e-08,
      "loss": 0.4449,
      "step": 452380
    },
    {
      "epoch": 185.18,
      "grad_norm": 2.1360437870025635,
      "learning_rate": 7.003403178110992e-08,
      "loss": 0.4494,
      "step": 452390
    },
    {
      "epoch": 185.18,
      "grad_norm": 1.7545230388641357,
      "learning_rate": 6.999554864532382e-08,
      "loss": 0.4321,
      "step": 452400
    },
    {
      "epoch": 185.19,
      "grad_norm": 2.3632373809814453,
      "learning_rate": 6.995707592525471e-08,
      "loss": 0.441,
      "step": 452410
    },
    {
      "epoch": 185.19,
      "grad_norm": 2.2702388763427734,
      "learning_rate": 6.991861362107901e-08,
      "loss": 0.4183,
      "step": 452420
    },
    {
      "epoch": 185.19,
      "grad_norm": 1.9760867357254028,
      "learning_rate": 6.988016173297235e-08,
      "loss": 0.4575,
      "step": 452430
    },
    {
      "epoch": 185.2,
      "grad_norm": 2.017916440963745,
      "learning_rate": 6.984172026111168e-08,
      "loss": 0.4406,
      "step": 452440
    },
    {
      "epoch": 185.2,
      "grad_norm": 2.147184371948242,
      "learning_rate": 6.98032892056724e-08,
      "loss": 0.431,
      "step": 452450
    },
    {
      "epoch": 185.21,
      "grad_norm": 1.9916417598724365,
      "learning_rate": 6.976486856683115e-08,
      "loss": 0.4536,
      "step": 452460
    },
    {
      "epoch": 185.21,
      "grad_norm": 2.0624146461486816,
      "learning_rate": 6.972645834476334e-08,
      "loss": 0.4494,
      "step": 452470
    },
    {
      "epoch": 185.21,
      "grad_norm": 2.52805757522583,
      "learning_rate": 6.968805853964539e-08,
      "loss": 0.4503,
      "step": 452480
    },
    {
      "epoch": 185.22,
      "grad_norm": 2.086299419403076,
      "learning_rate": 6.964966915165267e-08,
      "loss": 0.4439,
      "step": 452490
    },
    {
      "epoch": 185.22,
      "grad_norm": 2.6296043395996094,
      "learning_rate": 6.961129018096161e-08,
      "loss": 0.4456,
      "step": 452500
    },
    {
      "epoch": 185.23,
      "grad_norm": 2.2219042778015137,
      "learning_rate": 6.957292162774758e-08,
      "loss": 0.4385,
      "step": 452510
    },
    {
      "epoch": 185.23,
      "grad_norm": 2.3761675357818604,
      "learning_rate": 6.953456349218677e-08,
      "loss": 0.4368,
      "step": 452520
    },
    {
      "epoch": 185.24,
      "grad_norm": 1.9343708753585815,
      "learning_rate": 6.949621577445452e-08,
      "loss": 0.4586,
      "step": 452530
    },
    {
      "epoch": 185.24,
      "grad_norm": 2.070552110671997,
      "learning_rate": 6.945787847472676e-08,
      "loss": 0.4176,
      "step": 452540
    },
    {
      "epoch": 185.24,
      "grad_norm": 3.3608076572418213,
      "learning_rate": 6.94195515931791e-08,
      "loss": 0.4575,
      "step": 452550
    },
    {
      "epoch": 185.25,
      "grad_norm": 2.030346632003784,
      "learning_rate": 6.93812351299867e-08,
      "loss": 0.4513,
      "step": 452560
    },
    {
      "epoch": 185.25,
      "grad_norm": 1.8689507246017456,
      "learning_rate": 6.93429290853257e-08,
      "loss": 0.4331,
      "step": 452570
    },
    {
      "epoch": 185.26,
      "grad_norm": 2.3772079944610596,
      "learning_rate": 6.930463345937096e-08,
      "loss": 0.4469,
      "step": 452580
    },
    {
      "epoch": 185.26,
      "grad_norm": 1.9391049146652222,
      "learning_rate": 6.926634825229838e-08,
      "loss": 0.4581,
      "step": 452590
    },
    {
      "epoch": 185.26,
      "grad_norm": 3.3705506324768066,
      "learning_rate": 6.922807346428335e-08,
      "loss": 0.4403,
      "step": 452600
    },
    {
      "epoch": 185.27,
      "grad_norm": 2.2652347087860107,
      "learning_rate": 6.918980909550097e-08,
      "loss": 0.4436,
      "step": 452610
    },
    {
      "epoch": 185.27,
      "grad_norm": 1.7478525638580322,
      "learning_rate": 6.915155514612639e-08,
      "loss": 0.4343,
      "step": 452620
    },
    {
      "epoch": 185.28,
      "grad_norm": 1.8712573051452637,
      "learning_rate": 6.911331161633497e-08,
      "loss": 0.4324,
      "step": 452630
    },
    {
      "epoch": 185.28,
      "grad_norm": 2.0724380016326904,
      "learning_rate": 6.90750785063021e-08,
      "loss": 0.4318,
      "step": 452640
    },
    {
      "epoch": 185.28,
      "grad_norm": 2.0963454246520996,
      "learning_rate": 6.903685581620263e-08,
      "loss": 0.4344,
      "step": 452650
    },
    {
      "epoch": 185.29,
      "grad_norm": 2.016500949859619,
      "learning_rate": 6.89986435462117e-08,
      "loss": 0.4726,
      "step": 452660
    },
    {
      "epoch": 185.29,
      "grad_norm": 6.5223069190979,
      "learning_rate": 6.89604416965044e-08,
      "loss": 0.4335,
      "step": 452670
    },
    {
      "epoch": 185.3,
      "grad_norm": 2.041160821914673,
      "learning_rate": 6.892225026725587e-08,
      "loss": 0.448,
      "step": 452680
    },
    {
      "epoch": 185.3,
      "grad_norm": 1.8243328332901,
      "learning_rate": 6.888406925864096e-08,
      "loss": 0.4527,
      "step": 452690
    },
    {
      "epoch": 185.3,
      "grad_norm": 1.960128664970398,
      "learning_rate": 6.88458986708348e-08,
      "loss": 0.4492,
      "step": 452700
    },
    {
      "epoch": 185.31,
      "grad_norm": 1.7137455940246582,
      "learning_rate": 6.880773850401174e-08,
      "loss": 0.4708,
      "step": 452710
    },
    {
      "epoch": 185.31,
      "grad_norm": 3.7240307331085205,
      "learning_rate": 6.876958875834688e-08,
      "loss": 0.4354,
      "step": 452720
    },
    {
      "epoch": 185.32,
      "grad_norm": 2.0886824131011963,
      "learning_rate": 6.873144943401535e-08,
      "loss": 0.4562,
      "step": 452730
    },
    {
      "epoch": 185.32,
      "grad_norm": 2.004746198654175,
      "learning_rate": 6.869332053119122e-08,
      "loss": 0.4461,
      "step": 452740
    },
    {
      "epoch": 185.33,
      "grad_norm": 1.9330174922943115,
      "learning_rate": 6.865520205004936e-08,
      "loss": 0.4271,
      "step": 452750
    },
    {
      "epoch": 185.33,
      "grad_norm": 2.197765827178955,
      "learning_rate": 6.861709399076437e-08,
      "loss": 0.4523,
      "step": 452760
    },
    {
      "epoch": 185.33,
      "grad_norm": 2.166935682296753,
      "learning_rate": 6.85789963535111e-08,
      "loss": 0.4382,
      "step": 452770
    },
    {
      "epoch": 185.34,
      "grad_norm": 1.8762214183807373,
      "learning_rate": 6.854090913846365e-08,
      "loss": 0.445,
      "step": 452780
    },
    {
      "epoch": 185.34,
      "grad_norm": 2.1758272647857666,
      "learning_rate": 6.85028323457966e-08,
      "loss": 0.4396,
      "step": 452790
    },
    {
      "epoch": 185.35,
      "grad_norm": 2.0394809246063232,
      "learning_rate": 6.846476597568455e-08,
      "loss": 0.4482,
      "step": 452800
    },
    {
      "epoch": 185.35,
      "grad_norm": 2.330345392227173,
      "learning_rate": 6.842671002830185e-08,
      "loss": 0.4325,
      "step": 452810
    },
    {
      "epoch": 185.35,
      "grad_norm": 1.8300331830978394,
      "learning_rate": 6.838866450382257e-08,
      "loss": 0.4463,
      "step": 452820
    },
    {
      "epoch": 185.36,
      "grad_norm": 2.36126708984375,
      "learning_rate": 6.835062940242134e-08,
      "loss": 0.4491,
      "step": 452830
    },
    {
      "epoch": 185.36,
      "grad_norm": 1.907192349433899,
      "learning_rate": 6.831260472427219e-08,
      "loss": 0.4211,
      "step": 452840
    },
    {
      "epoch": 185.37,
      "grad_norm": 1.8559415340423584,
      "learning_rate": 6.827459046954949e-08,
      "loss": 0.4328,
      "step": 452850
    },
    {
      "epoch": 185.37,
      "grad_norm": 2.5168516635894775,
      "learning_rate": 6.82365866384268e-08,
      "loss": 0.4386,
      "step": 452860
    },
    {
      "epoch": 185.37,
      "grad_norm": 1.7147468328475952,
      "learning_rate": 6.819859323107922e-08,
      "loss": 0.4348,
      "step": 452870
    },
    {
      "epoch": 185.38,
      "grad_norm": 2.2219746112823486,
      "learning_rate": 6.816061024768008e-08,
      "loss": 0.4261,
      "step": 452880
    },
    {
      "epoch": 185.38,
      "grad_norm": 1.840027928352356,
      "learning_rate": 6.812263768840344e-08,
      "loss": 0.435,
      "step": 452890
    },
    {
      "epoch": 185.39,
      "grad_norm": 1.7853834629058838,
      "learning_rate": 6.808467555342338e-08,
      "loss": 0.433,
      "step": 452900
    },
    {
      "epoch": 185.39,
      "grad_norm": 2.0415425300598145,
      "learning_rate": 6.804672384291346e-08,
      "loss": 0.4567,
      "step": 452910
    },
    {
      "epoch": 185.4,
      "grad_norm": 2.6482887268066406,
      "learning_rate": 6.800878255704802e-08,
      "loss": 0.4319,
      "step": 452920
    },
    {
      "epoch": 185.4,
      "grad_norm": 1.9856438636779785,
      "learning_rate": 6.797085169600062e-08,
      "loss": 0.4167,
      "step": 452930
    },
    {
      "epoch": 185.4,
      "grad_norm": 1.9830094575881958,
      "learning_rate": 6.793293125994509e-08,
      "loss": 0.4518,
      "step": 452940
    },
    {
      "epoch": 185.41,
      "grad_norm": 1.9169822931289673,
      "learning_rate": 6.789502124905523e-08,
      "loss": 0.4437,
      "step": 452950
    },
    {
      "epoch": 185.41,
      "grad_norm": 2.0766377449035645,
      "learning_rate": 6.785712166350433e-08,
      "loss": 0.4331,
      "step": 452960
    },
    {
      "epoch": 185.42,
      "grad_norm": 1.704924464225769,
      "learning_rate": 6.781923250346625e-08,
      "loss": 0.4465,
      "step": 452970
    },
    {
      "epoch": 185.42,
      "grad_norm": 2.2983829975128174,
      "learning_rate": 6.778135376911478e-08,
      "loss": 0.4478,
      "step": 452980
    },
    {
      "epoch": 185.42,
      "grad_norm": 2.1060495376586914,
      "learning_rate": 6.774348546062296e-08,
      "loss": 0.4268,
      "step": 452990
    },
    {
      "epoch": 185.43,
      "grad_norm": 2.3739895820617676,
      "learning_rate": 6.770562757816462e-08,
      "loss": 0.4747,
      "step": 453000
    },
    {
      "epoch": 185.43,
      "grad_norm": 2.170628309249878,
      "learning_rate": 6.766778012191306e-08,
      "loss": 0.4367,
      "step": 453010
    },
    {
      "epoch": 185.44,
      "grad_norm": 2.109173536300659,
      "learning_rate": 6.762994309204156e-08,
      "loss": 0.4632,
      "step": 453020
    },
    {
      "epoch": 185.44,
      "grad_norm": 2.4022645950317383,
      "learning_rate": 6.759211648872344e-08,
      "loss": 0.4635,
      "step": 453030
    },
    {
      "epoch": 185.44,
      "grad_norm": 2.0899035930633545,
      "learning_rate": 6.75543003121325e-08,
      "loss": 0.4315,
      "step": 453040
    },
    {
      "epoch": 185.45,
      "grad_norm": 2.1184165477752686,
      "learning_rate": 6.751649456244101e-08,
      "loss": 0.4564,
      "step": 453050
    },
    {
      "epoch": 185.45,
      "grad_norm": 2.15728759765625,
      "learning_rate": 6.747869923982306e-08,
      "loss": 0.4427,
      "step": 453060
    },
    {
      "epoch": 185.46,
      "grad_norm": 2.6812806129455566,
      "learning_rate": 6.74409143444514e-08,
      "loss": 0.463,
      "step": 453070
    },
    {
      "epoch": 185.46,
      "grad_norm": 2.1885437965393066,
      "learning_rate": 6.740313987649883e-08,
      "loss": 0.4457,
      "step": 453080
    },
    {
      "epoch": 185.46,
      "grad_norm": 2.277693510055542,
      "learning_rate": 6.73653758361389e-08,
      "loss": 0.4204,
      "step": 453090
    },
    {
      "epoch": 185.47,
      "grad_norm": 1.773573637008667,
      "learning_rate": 6.73276222235444e-08,
      "loss": 0.4427,
      "step": 453100
    },
    {
      "epoch": 185.47,
      "grad_norm": 2.103632926940918,
      "learning_rate": 6.728987903888835e-08,
      "loss": 0.4397,
      "step": 453110
    },
    {
      "epoch": 185.48,
      "grad_norm": 2.0600106716156006,
      "learning_rate": 6.725214628234354e-08,
      "loss": 0.4218,
      "step": 453120
    },
    {
      "epoch": 185.48,
      "grad_norm": 6.362812042236328,
      "learning_rate": 6.721442395408277e-08,
      "loss": 0.4367,
      "step": 453130
    },
    {
      "epoch": 185.49,
      "grad_norm": 2.095221996307373,
      "learning_rate": 6.717671205427877e-08,
      "loss": 0.4526,
      "step": 453140
    },
    {
      "epoch": 185.49,
      "grad_norm": 1.9290921688079834,
      "learning_rate": 6.713901058310461e-08,
      "loss": 0.4359,
      "step": 453150
    },
    {
      "epoch": 185.49,
      "grad_norm": 1.8786511421203613,
      "learning_rate": 6.710131954073282e-08,
      "loss": 0.4213,
      "step": 453160
    },
    {
      "epoch": 185.5,
      "grad_norm": 2.0255253314971924,
      "learning_rate": 6.706363892733587e-08,
      "loss": 0.4343,
      "step": 453170
    },
    {
      "epoch": 185.5,
      "grad_norm": 2.177565336227417,
      "learning_rate": 6.702596874308711e-08,
      "loss": 0.4558,
      "step": 453180
    },
    {
      "epoch": 185.51,
      "grad_norm": 2.358450174331665,
      "learning_rate": 6.6988308988158e-08,
      "loss": 0.4498,
      "step": 453190
    },
    {
      "epoch": 185.51,
      "grad_norm": 1.9938000440597534,
      "learning_rate": 6.695065966272158e-08,
      "loss": 0.4343,
      "step": 453200
    },
    {
      "epoch": 185.51,
      "grad_norm": 1.6775729656219482,
      "learning_rate": 6.69130207669506e-08,
      "loss": 0.4545,
      "step": 453210
    },
    {
      "epoch": 185.52,
      "grad_norm": 2.1936585903167725,
      "learning_rate": 6.68753923010171e-08,
      "loss": 0.4499,
      "step": 453220
    },
    {
      "epoch": 185.52,
      "grad_norm": 2.0102765560150146,
      "learning_rate": 6.683777426509359e-08,
      "loss": 0.428,
      "step": 453230
    },
    {
      "epoch": 185.53,
      "grad_norm": 2.0929384231567383,
      "learning_rate": 6.680016665935204e-08,
      "loss": 0.4482,
      "step": 453240
    },
    {
      "epoch": 185.53,
      "grad_norm": 1.8445830345153809,
      "learning_rate": 6.676256948396526e-08,
      "loss": 0.4537,
      "step": 453250
    },
    {
      "epoch": 185.53,
      "grad_norm": 2.089404821395874,
      "learning_rate": 6.672498273910522e-08,
      "loss": 0.4522,
      "step": 453260
    },
    {
      "epoch": 185.54,
      "grad_norm": 2.1140007972717285,
      "learning_rate": 6.668740642494421e-08,
      "loss": 0.4767,
      "step": 453270
    },
    {
      "epoch": 185.54,
      "grad_norm": 2.0539844036102295,
      "learning_rate": 6.664984054165447e-08,
      "loss": 0.4415,
      "step": 453280
    },
    {
      "epoch": 185.55,
      "grad_norm": 4.01939582824707,
      "learning_rate": 6.661228508940773e-08,
      "loss": 0.4496,
      "step": 453290
    },
    {
      "epoch": 185.55,
      "grad_norm": 1.977608561515808,
      "learning_rate": 6.657474006837599e-08,
      "loss": 0.4573,
      "step": 453300
    },
    {
      "epoch": 185.55,
      "grad_norm": 2.013167142868042,
      "learning_rate": 6.653720547873205e-08,
      "loss": 0.4545,
      "step": 453310
    },
    {
      "epoch": 185.56,
      "grad_norm": 1.7797333002090454,
      "learning_rate": 6.649968132064684e-08,
      "loss": 0.4569,
      "step": 453320
    },
    {
      "epoch": 185.56,
      "grad_norm": 2.0141987800598145,
      "learning_rate": 6.646216759429264e-08,
      "loss": 0.4378,
      "step": 453330
    },
    {
      "epoch": 185.57,
      "grad_norm": 2.123525381088257,
      "learning_rate": 6.642466429984143e-08,
      "loss": 0.4346,
      "step": 453340
    },
    {
      "epoch": 185.57,
      "grad_norm": 1.995192289352417,
      "learning_rate": 6.638717143746496e-08,
      "loss": 0.4311,
      "step": 453350
    },
    {
      "epoch": 185.58,
      "grad_norm": 2.439915180206299,
      "learning_rate": 6.63496890073347e-08,
      "loss": 0.4472,
      "step": 453360
    },
    {
      "epoch": 185.58,
      "grad_norm": 2.424370527267456,
      "learning_rate": 6.631221700962265e-08,
      "loss": 0.4434,
      "step": 453370
    },
    {
      "epoch": 185.58,
      "grad_norm": 1.9699163436889648,
      "learning_rate": 6.627475544450055e-08,
      "loss": 0.4349,
      "step": 453380
    },
    {
      "epoch": 185.59,
      "grad_norm": 1.9722917079925537,
      "learning_rate": 6.623730431213961e-08,
      "loss": 0.4598,
      "step": 453390
    },
    {
      "epoch": 185.59,
      "grad_norm": 1.7992080450057983,
      "learning_rate": 6.619986361271158e-08,
      "loss": 0.4304,
      "step": 453400
    },
    {
      "epoch": 185.6,
      "grad_norm": 2.1541519165039062,
      "learning_rate": 6.616243334638818e-08,
      "loss": 0.4429,
      "step": 453410
    },
    {
      "epoch": 185.6,
      "grad_norm": 2.0861458778381348,
      "learning_rate": 6.612501351334065e-08,
      "loss": 0.4542,
      "step": 453420
    },
    {
      "epoch": 185.6,
      "grad_norm": 2.1173276901245117,
      "learning_rate": 6.608760411374019e-08,
      "loss": 0.438,
      "step": 453430
    },
    {
      "epoch": 185.61,
      "grad_norm": 2.0202057361602783,
      "learning_rate": 6.605020514775855e-08,
      "loss": 0.4547,
      "step": 453440
    },
    {
      "epoch": 185.61,
      "grad_norm": 2.2717692852020264,
      "learning_rate": 6.601281661556745e-08,
      "loss": 0.4526,
      "step": 453450
    },
    {
      "epoch": 185.62,
      "grad_norm": 2.755307674407959,
      "learning_rate": 6.597543851733708e-08,
      "loss": 0.452,
      "step": 453460
    },
    {
      "epoch": 185.62,
      "grad_norm": 2.148263454437256,
      "learning_rate": 6.593807085323944e-08,
      "loss": 0.4715,
      "step": 453470
    },
    {
      "epoch": 185.62,
      "grad_norm": 3.259878158569336,
      "learning_rate": 6.590071362344521e-08,
      "loss": 0.4424,
      "step": 453480
    },
    {
      "epoch": 185.63,
      "grad_norm": 2.466325521469116,
      "learning_rate": 6.586336682812614e-08,
      "loss": 0.425,
      "step": 453490
    },
    {
      "epoch": 185.63,
      "grad_norm": 1.9069119691848755,
      "learning_rate": 6.582603046745268e-08,
      "loss": 0.4327,
      "step": 453500
    },
    {
      "epoch": 185.64,
      "grad_norm": 1.8021483421325684,
      "learning_rate": 6.578870454159629e-08,
      "loss": 0.4326,
      "step": 453510
    },
    {
      "epoch": 185.64,
      "grad_norm": 2.1351511478424072,
      "learning_rate": 6.575138905072792e-08,
      "loss": 0.4649,
      "step": 453520
    },
    {
      "epoch": 185.64,
      "grad_norm": 2.2768807411193848,
      "learning_rate": 6.571408399501802e-08,
      "loss": 0.4518,
      "step": 453530
    },
    {
      "epoch": 185.65,
      "grad_norm": 2.1900649070739746,
      "learning_rate": 6.567678937463832e-08,
      "loss": 0.4609,
      "step": 453540
    },
    {
      "epoch": 185.65,
      "grad_norm": 2.052166223526001,
      "learning_rate": 6.563950518975874e-08,
      "loss": 0.4606,
      "step": 453550
    },
    {
      "epoch": 185.66,
      "grad_norm": 2.110578775405884,
      "learning_rate": 6.560223144055077e-08,
      "loss": 0.4466,
      "step": 453560
    },
    {
      "epoch": 185.66,
      "grad_norm": 2.10488224029541,
      "learning_rate": 6.556496812718509e-08,
      "loss": 0.4391,
      "step": 453570
    },
    {
      "epoch": 185.67,
      "grad_norm": 1.979065179824829,
      "learning_rate": 6.552771524983187e-08,
      "loss": 0.4372,
      "step": 453580
    },
    {
      "epoch": 185.67,
      "grad_norm": 2.0859217643737793,
      "learning_rate": 6.549047280866235e-08,
      "loss": 0.4437,
      "step": 453590
    },
    {
      "epoch": 185.67,
      "grad_norm": 2.719680070877075,
      "learning_rate": 6.545324080384669e-08,
      "loss": 0.462,
      "step": 453600
    },
    {
      "epoch": 185.68,
      "grad_norm": 2.0021021366119385,
      "learning_rate": 6.541601923555585e-08,
      "loss": 0.4409,
      "step": 453610
    },
    {
      "epoch": 185.68,
      "grad_norm": 2.2095184326171875,
      "learning_rate": 6.537880810396e-08,
      "loss": 0.441,
      "step": 453620
    },
    {
      "epoch": 185.69,
      "grad_norm": 1.9191405773162842,
      "learning_rate": 6.534160740922987e-08,
      "loss": 0.4413,
      "step": 453630
    },
    {
      "epoch": 185.69,
      "grad_norm": 1.8063247203826904,
      "learning_rate": 6.530441715153559e-08,
      "loss": 0.4537,
      "step": 453640
    },
    {
      "epoch": 185.69,
      "grad_norm": 1.9394367933273315,
      "learning_rate": 6.526723733104762e-08,
      "loss": 0.4319,
      "step": 453650
    },
    {
      "epoch": 185.7,
      "grad_norm": 2.1433098316192627,
      "learning_rate": 6.52300679479364e-08,
      "loss": 0.4456,
      "step": 453660
    },
    {
      "epoch": 185.7,
      "grad_norm": 1.8862606287002563,
      "learning_rate": 6.51929090023721e-08,
      "loss": 0.4472,
      "step": 453670
    },
    {
      "epoch": 185.71,
      "grad_norm": 1.9273258447647095,
      "learning_rate": 6.51557604945249e-08,
      "loss": 0.4388,
      "step": 453680
    },
    {
      "epoch": 185.71,
      "grad_norm": 2.1532857418060303,
      "learning_rate": 6.511862242456521e-08,
      "loss": 0.4514,
      "step": 453690
    },
    {
      "epoch": 185.71,
      "grad_norm": 1.9151413440704346,
      "learning_rate": 6.508149479266273e-08,
      "loss": 0.4344,
      "step": 453700
    },
    {
      "epoch": 185.72,
      "grad_norm": 1.9302113056182861,
      "learning_rate": 6.504437759898787e-08,
      "loss": 0.4353,
      "step": 453710
    },
    {
      "epoch": 185.72,
      "grad_norm": 2.393000602722168,
      "learning_rate": 6.500727084371056e-08,
      "loss": 0.4634,
      "step": 453720
    },
    {
      "epoch": 185.73,
      "grad_norm": 2.6700148582458496,
      "learning_rate": 6.497017452700097e-08,
      "loss": 0.4424,
      "step": 453730
    },
    {
      "epoch": 185.73,
      "grad_norm": 2.157886266708374,
      "learning_rate": 6.493308864902874e-08,
      "loss": 0.4375,
      "step": 453740
    },
    {
      "epoch": 185.73,
      "grad_norm": 1.772763967514038,
      "learning_rate": 6.48960132099638e-08,
      "loss": 0.4438,
      "step": 453750
    },
    {
      "epoch": 185.74,
      "grad_norm": 3.4392435550689697,
      "learning_rate": 6.485894820997635e-08,
      "loss": 0.431,
      "step": 453760
    },
    {
      "epoch": 185.74,
      "grad_norm": 2.077516555786133,
      "learning_rate": 6.482189364923549e-08,
      "loss": 0.4386,
      "step": 453770
    },
    {
      "epoch": 185.75,
      "grad_norm": 2.022991180419922,
      "learning_rate": 6.478484952791141e-08,
      "loss": 0.4326,
      "step": 453780
    },
    {
      "epoch": 185.75,
      "grad_norm": 1.9610426425933838,
      "learning_rate": 6.474781584617376e-08,
      "loss": 0.4626,
      "step": 453790
    },
    {
      "epoch": 185.76,
      "grad_norm": 2.4407830238342285,
      "learning_rate": 6.471079260419219e-08,
      "loss": 0.4219,
      "step": 453800
    },
    {
      "epoch": 185.76,
      "grad_norm": 2.1345746517181396,
      "learning_rate": 6.467377980213613e-08,
      "loss": 0.4217,
      "step": 453810
    },
    {
      "epoch": 185.76,
      "grad_norm": 2.078263282775879,
      "learning_rate": 6.463677744017546e-08,
      "loss": 0.442,
      "step": 453820
    },
    {
      "epoch": 185.77,
      "grad_norm": 2.218111038208008,
      "learning_rate": 6.459978551847933e-08,
      "loss": 0.4326,
      "step": 453830
    },
    {
      "epoch": 185.77,
      "grad_norm": 2.12184476852417,
      "learning_rate": 6.456280403721766e-08,
      "loss": 0.4396,
      "step": 453840
    },
    {
      "epoch": 185.78,
      "grad_norm": 1.9112476110458374,
      "learning_rate": 6.452583299655931e-08,
      "loss": 0.4452,
      "step": 453850
    },
    {
      "epoch": 185.78,
      "grad_norm": 2.272095203399658,
      "learning_rate": 6.448887239667395e-08,
      "loss": 0.4632,
      "step": 453860
    },
    {
      "epoch": 185.78,
      "grad_norm": 2.1270036697387695,
      "learning_rate": 6.445192223773071e-08,
      "loss": 0.4377,
      "step": 453870
    },
    {
      "epoch": 185.79,
      "grad_norm": 1.6680362224578857,
      "learning_rate": 6.441498251989925e-08,
      "loss": 0.4651,
      "step": 453880
    },
    {
      "epoch": 185.79,
      "grad_norm": 2.0197227001190186,
      "learning_rate": 6.437805324334844e-08,
      "loss": 0.4579,
      "step": 453890
    },
    {
      "epoch": 185.8,
      "grad_norm": 1.9692137241363525,
      "learning_rate": 6.434113440824769e-08,
      "loss": 0.4367,
      "step": 453900
    },
    {
      "epoch": 185.8,
      "grad_norm": 2.0447490215301514,
      "learning_rate": 6.430422601476559e-08,
      "loss": 0.4409,
      "step": 453910
    },
    {
      "epoch": 185.8,
      "grad_norm": 2.532392740249634,
      "learning_rate": 6.426732806307156e-08,
      "loss": 0.4355,
      "step": 453920
    },
    {
      "epoch": 185.81,
      "grad_norm": 2.645171642303467,
      "learning_rate": 6.423044055333471e-08,
      "loss": 0.4304,
      "step": 453930
    },
    {
      "epoch": 185.81,
      "grad_norm": 1.8774654865264893,
      "learning_rate": 6.419356348572394e-08,
      "loss": 0.4278,
      "step": 453940
    },
    {
      "epoch": 185.82,
      "grad_norm": 2.3608360290527344,
      "learning_rate": 6.415669686040837e-08,
      "loss": 0.4447,
      "step": 453950
    },
    {
      "epoch": 185.82,
      "grad_norm": 2.3284547328948975,
      "learning_rate": 6.411984067755637e-08,
      "loss": 0.4423,
      "step": 453960
    },
    {
      "epoch": 185.82,
      "grad_norm": 2.179867744445801,
      "learning_rate": 6.408299493733731e-08,
      "loss": 0.4562,
      "step": 453970
    },
    {
      "epoch": 185.83,
      "grad_norm": 2.085365056991577,
      "learning_rate": 6.404615963991957e-08,
      "loss": 0.4501,
      "step": 453980
    },
    {
      "epoch": 185.83,
      "grad_norm": 2.205753803253174,
      "learning_rate": 6.400933478547228e-08,
      "loss": 0.444,
      "step": 453990
    },
    {
      "epoch": 185.84,
      "grad_norm": 1.9140678644180298,
      "learning_rate": 6.397252037416354e-08,
      "loss": 0.4511,
      "step": 454000
    },
    {
      "epoch": 185.84,
      "grad_norm": 1.9126275777816772,
      "learning_rate": 6.393571640616246e-08,
      "loss": 0.4402,
      "step": 454010
    },
    {
      "epoch": 185.85,
      "grad_norm": 2.2210917472839355,
      "learning_rate": 6.389892288163767e-08,
      "loss": 0.4353,
      "step": 454020
    },
    {
      "epoch": 185.85,
      "grad_norm": 2.149758815765381,
      "learning_rate": 6.38621398007578e-08,
      "loss": 0.4302,
      "step": 454030
    },
    {
      "epoch": 185.85,
      "grad_norm": 1.7753922939300537,
      "learning_rate": 6.382536716369093e-08,
      "loss": 0.4515,
      "step": 454040
    },
    {
      "epoch": 185.86,
      "grad_norm": 1.9962857961654663,
      "learning_rate": 6.378860497060568e-08,
      "loss": 0.4353,
      "step": 454050
    },
    {
      "epoch": 185.86,
      "grad_norm": 1.9791181087493896,
      "learning_rate": 6.375185322167065e-08,
      "loss": 0.4292,
      "step": 454060
    },
    {
      "epoch": 185.87,
      "grad_norm": 2.200796365737915,
      "learning_rate": 6.371511191705396e-08,
      "loss": 0.4264,
      "step": 454070
    },
    {
      "epoch": 185.87,
      "grad_norm": 2.24845814704895,
      "learning_rate": 6.367838105692394e-08,
      "loss": 0.4296,
      "step": 454080
    },
    {
      "epoch": 185.87,
      "grad_norm": 2.1197690963745117,
      "learning_rate": 6.36416606414487e-08,
      "loss": 0.4326,
      "step": 454090
    },
    {
      "epoch": 185.88,
      "grad_norm": 1.871915578842163,
      "learning_rate": 6.360495067079686e-08,
      "loss": 0.4251,
      "step": 454100
    },
    {
      "epoch": 185.88,
      "grad_norm": 1.9781770706176758,
      "learning_rate": 6.35682511451365e-08,
      "loss": 0.4202,
      "step": 454110
    },
    {
      "epoch": 185.89,
      "grad_norm": 1.9307292699813843,
      "learning_rate": 6.35315620646352e-08,
      "loss": 0.4234,
      "step": 454120
    },
    {
      "epoch": 185.89,
      "grad_norm": 1.9846521615982056,
      "learning_rate": 6.349488342946184e-08,
      "loss": 0.4292,
      "step": 454130
    },
    {
      "epoch": 185.89,
      "grad_norm": 1.8320984840393066,
      "learning_rate": 6.345821523978372e-08,
      "loss": 0.4638,
      "step": 454140
    },
    {
      "epoch": 185.9,
      "grad_norm": 2.0711569786071777,
      "learning_rate": 6.342155749576921e-08,
      "loss": 0.4578,
      "step": 454150
    },
    {
      "epoch": 185.9,
      "grad_norm": 3.627973794937134,
      "learning_rate": 6.338491019758614e-08,
      "loss": 0.4603,
      "step": 454160
    },
    {
      "epoch": 185.91,
      "grad_norm": 1.9786609411239624,
      "learning_rate": 6.334827334540261e-08,
      "loss": 0.4339,
      "step": 454170
    },
    {
      "epoch": 185.91,
      "grad_norm": 3.8582675457000732,
      "learning_rate": 6.331164693938592e-08,
      "loss": 0.441,
      "step": 454180
    },
    {
      "epoch": 185.91,
      "grad_norm": 2.4365804195404053,
      "learning_rate": 6.327503097970417e-08,
      "loss": 0.4566,
      "step": 454190
    },
    {
      "epoch": 185.92,
      "grad_norm": 2.2709736824035645,
      "learning_rate": 6.323842546652521e-08,
      "loss": 0.4397,
      "step": 454200
    },
    {
      "epoch": 185.92,
      "grad_norm": 2.42903995513916,
      "learning_rate": 6.32018304000166e-08,
      "loss": 0.4503,
      "step": 454210
    },
    {
      "epoch": 185.93,
      "grad_norm": 2.286529779434204,
      "learning_rate": 6.316524578034591e-08,
      "loss": 0.4431,
      "step": 454220
    },
    {
      "epoch": 185.93,
      "grad_norm": 1.7807115316390991,
      "learning_rate": 6.312867160768071e-08,
      "loss": 0.445,
      "step": 454230
    },
    {
      "epoch": 185.94,
      "grad_norm": 2.3593907356262207,
      "learning_rate": 6.309210788218886e-08,
      "loss": 0.4481,
      "step": 454240
    },
    {
      "epoch": 185.94,
      "grad_norm": 1.9001235961914062,
      "learning_rate": 6.30555546040374e-08,
      "loss": 0.4659,
      "step": 454250
    },
    {
      "epoch": 185.94,
      "grad_norm": 2.2055485248565674,
      "learning_rate": 6.301901177339416e-08,
      "loss": 0.4485,
      "step": 454260
    },
    {
      "epoch": 185.95,
      "grad_norm": 2.4293346405029297,
      "learning_rate": 6.29824793904262e-08,
      "loss": 0.4708,
      "step": 454270
    },
    {
      "epoch": 185.95,
      "grad_norm": 2.671538829803467,
      "learning_rate": 6.294595745530107e-08,
      "loss": 0.4267,
      "step": 454280
    },
    {
      "epoch": 185.96,
      "grad_norm": 1.8391506671905518,
      "learning_rate": 6.290944596818613e-08,
      "loss": 0.4533,
      "step": 454290
    },
    {
      "epoch": 185.96,
      "grad_norm": 1.8907428979873657,
      "learning_rate": 6.287294492924839e-08,
      "loss": 0.4439,
      "step": 454300
    },
    {
      "epoch": 185.96,
      "grad_norm": 2.383683443069458,
      "learning_rate": 6.283645433865521e-08,
      "loss": 0.4517,
      "step": 454310
    },
    {
      "epoch": 185.97,
      "grad_norm": 2.1599161624908447,
      "learning_rate": 6.279997419657385e-08,
      "loss": 0.4491,
      "step": 454320
    },
    {
      "epoch": 185.97,
      "grad_norm": 2.1755928993225098,
      "learning_rate": 6.276350450317114e-08,
      "loss": 0.4228,
      "step": 454330
    },
    {
      "epoch": 185.98,
      "grad_norm": 2.1280837059020996,
      "learning_rate": 6.272704525861465e-08,
      "loss": 0.4423,
      "step": 454340
    },
    {
      "epoch": 185.98,
      "grad_norm": 1.949578881263733,
      "learning_rate": 6.26905964630709e-08,
      "loss": 0.4487,
      "step": 454350
    },
    {
      "epoch": 185.98,
      "grad_norm": 2.0383381843566895,
      "learning_rate": 6.26541581167067e-08,
      "loss": 0.4355,
      "step": 454360
    },
    {
      "epoch": 185.99,
      "grad_norm": 1.9472798109054565,
      "learning_rate": 6.261773021968936e-08,
      "loss": 0.4419,
      "step": 454370
    },
    {
      "epoch": 185.99,
      "grad_norm": 2.1767165660858154,
      "learning_rate": 6.258131277218593e-08,
      "loss": 0.4535,
      "step": 454380
    },
    {
      "epoch": 186.0,
      "grad_norm": 2.0890204906463623,
      "learning_rate": 6.254490577436268e-08,
      "loss": 0.4455,
      "step": 454390
    },
    {
      "epoch": 186.0,
      "eval_loss": 0.4458167552947998,
      "eval_runtime": 57.8665,
      "eval_samples_per_second": 59.603,
      "eval_steps_per_second": 7.465,
      "step": 454398
    },
    {
      "epoch": 186.0,
      "grad_norm": 2.34031343460083,
      "learning_rate": 6.250850922638693e-08,
      "loss": 0.448,
      "step": 454400
    },
    {
      "epoch": 186.0,
      "grad_norm": 2.391406536102295,
      "learning_rate": 6.24721231284252e-08,
      "loss": 0.4439,
      "step": 454410
    },
    {
      "epoch": 186.01,
      "grad_norm": 1.9994254112243652,
      "learning_rate": 6.243574748064379e-08,
      "loss": 0.4266,
      "step": 454420
    },
    {
      "epoch": 186.01,
      "grad_norm": 2.440166473388672,
      "learning_rate": 6.239938228320999e-08,
      "loss": 0.4411,
      "step": 454430
    },
    {
      "epoch": 186.02,
      "grad_norm": 2.1048309803009033,
      "learning_rate": 6.236302753628982e-08,
      "loss": 0.4557,
      "step": 454440
    },
    {
      "epoch": 186.02,
      "grad_norm": 1.8050915002822876,
      "learning_rate": 6.232668324005034e-08,
      "loss": 0.4395,
      "step": 454450
    },
    {
      "epoch": 186.03,
      "grad_norm": 2.208928108215332,
      "learning_rate": 6.229034939465756e-08,
      "loss": 0.4658,
      "step": 454460
    },
    {
      "epoch": 186.03,
      "grad_norm": 2.0569427013397217,
      "learning_rate": 6.225402600027801e-08,
      "loss": 0.4379,
      "step": 454470
    },
    {
      "epoch": 186.03,
      "grad_norm": 1.909492015838623,
      "learning_rate": 6.221771305707874e-08,
      "loss": 0.4503,
      "step": 454480
    },
    {
      "epoch": 186.04,
      "grad_norm": 2.2806243896484375,
      "learning_rate": 6.218141056522499e-08,
      "loss": 0.4406,
      "step": 454490
    },
    {
      "epoch": 186.04,
      "grad_norm": 1.987388014793396,
      "learning_rate": 6.214511852488383e-08,
      "loss": 0.4334,
      "step": 454500
    },
    {
      "epoch": 186.05,
      "grad_norm": 1.9556410312652588,
      "learning_rate": 6.210883693622123e-08,
      "loss": 0.4562,
      "step": 454510
    },
    {
      "epoch": 186.05,
      "grad_norm": 1.9312644004821777,
      "learning_rate": 6.207256579940324e-08,
      "loss": 0.4283,
      "step": 454520
    },
    {
      "epoch": 186.05,
      "grad_norm": 2.3991127014160156,
      "learning_rate": 6.203630511459638e-08,
      "loss": 0.4663,
      "step": 454530
    },
    {
      "epoch": 186.06,
      "grad_norm": 1.75284743309021,
      "learning_rate": 6.200005488196666e-08,
      "loss": 0.4435,
      "step": 454540
    },
    {
      "epoch": 186.06,
      "grad_norm": 2.7528748512268066,
      "learning_rate": 6.196381510167985e-08,
      "loss": 0.4536,
      "step": 454550
    },
    {
      "epoch": 186.07,
      "grad_norm": 1.9876338243484497,
      "learning_rate": 6.19275857739022e-08,
      "loss": 0.4548,
      "step": 454560
    },
    {
      "epoch": 186.07,
      "grad_norm": 2.0125792026519775,
      "learning_rate": 6.189136689879974e-08,
      "loss": 0.4499,
      "step": 454570
    },
    {
      "epoch": 186.07,
      "grad_norm": 1.8818351030349731,
      "learning_rate": 6.18551584765382e-08,
      "loss": 0.4389,
      "step": 454580
    },
    {
      "epoch": 186.08,
      "grad_norm": 1.9181909561157227,
      "learning_rate": 6.181896050728363e-08,
      "loss": 0.4298,
      "step": 454590
    },
    {
      "epoch": 186.08,
      "grad_norm": 2.280421495437622,
      "learning_rate": 6.178277299120175e-08,
      "loss": 0.4446,
      "step": 454600
    },
    {
      "epoch": 186.09,
      "grad_norm": 1.7202359437942505,
      "learning_rate": 6.174659592845833e-08,
      "loss": 0.4332,
      "step": 454610
    },
    {
      "epoch": 186.09,
      "grad_norm": 2.0710620880126953,
      "learning_rate": 6.171042931921938e-08,
      "loss": 0.4221,
      "step": 454620
    },
    {
      "epoch": 186.09,
      "grad_norm": 2.209254741668701,
      "learning_rate": 6.167427316364987e-08,
      "loss": 0.4519,
      "step": 454630
    },
    {
      "epoch": 186.1,
      "grad_norm": 2.300328254699707,
      "learning_rate": 6.163812746191607e-08,
      "loss": 0.4452,
      "step": 454640
    },
    {
      "epoch": 186.1,
      "grad_norm": 2.0543150901794434,
      "learning_rate": 6.160199221418322e-08,
      "loss": 0.436,
      "step": 454650
    },
    {
      "epoch": 186.11,
      "grad_norm": 2.7039217948913574,
      "learning_rate": 6.15658674206168e-08,
      "loss": 0.4274,
      "step": 454660
    },
    {
      "epoch": 186.11,
      "grad_norm": 2.1870243549346924,
      "learning_rate": 6.15297530813826e-08,
      "loss": 0.4449,
      "step": 454670
    },
    {
      "epoch": 186.12,
      "grad_norm": 1.8630201816558838,
      "learning_rate": 6.149364919664605e-08,
      "loss": 0.443,
      "step": 454680
    },
    {
      "epoch": 186.12,
      "grad_norm": 2.2213902473449707,
      "learning_rate": 6.145755576657217e-08,
      "loss": 0.4387,
      "step": 454690
    },
    {
      "epoch": 186.12,
      "grad_norm": 2.412329912185669,
      "learning_rate": 6.14214727913267e-08,
      "loss": 0.4485,
      "step": 454700
    },
    {
      "epoch": 186.13,
      "grad_norm": 2.2304182052612305,
      "learning_rate": 6.138540027107462e-08,
      "loss": 0.4324,
      "step": 454710
    },
    {
      "epoch": 186.13,
      "grad_norm": 2.0906119346618652,
      "learning_rate": 6.134933820598139e-08,
      "loss": 0.4518,
      "step": 454720
    },
    {
      "epoch": 186.14,
      "grad_norm": 2.148226499557495,
      "learning_rate": 6.131328659621203e-08,
      "loss": 0.4644,
      "step": 454730
    },
    {
      "epoch": 186.14,
      "grad_norm": 1.7965726852416992,
      "learning_rate": 6.127724544193174e-08,
      "loss": 0.4595,
      "step": 454740
    },
    {
      "epoch": 186.14,
      "grad_norm": 1.6374988555908203,
      "learning_rate": 6.124121474330576e-08,
      "loss": 0.4289,
      "step": 454750
    },
    {
      "epoch": 186.15,
      "grad_norm": 2.122227191925049,
      "learning_rate": 6.120519450049906e-08,
      "loss": 0.442,
      "step": 454760
    },
    {
      "epoch": 186.15,
      "grad_norm": 2.1797139644622803,
      "learning_rate": 6.116918471367663e-08,
      "loss": 0.4471,
      "step": 454770
    },
    {
      "epoch": 186.16,
      "grad_norm": 2.1631534099578857,
      "learning_rate": 6.113318538300314e-08,
      "loss": 0.422,
      "step": 454780
    },
    {
      "epoch": 186.16,
      "grad_norm": 2.1535279750823975,
      "learning_rate": 6.109719650864414e-08,
      "loss": 0.4319,
      "step": 454790
    },
    {
      "epoch": 186.16,
      "grad_norm": 1.9060269594192505,
      "learning_rate": 6.106121809076378e-08,
      "loss": 0.4442,
      "step": 454800
    },
    {
      "epoch": 186.17,
      "grad_norm": 2.1171882152557373,
      "learning_rate": 6.102525012952756e-08,
      "loss": 0.465,
      "step": 454810
    },
    {
      "epoch": 186.17,
      "grad_norm": 1.9961588382720947,
      "learning_rate": 6.098929262509968e-08,
      "loss": 0.4533,
      "step": 454820
    },
    {
      "epoch": 186.18,
      "grad_norm": 2.0928547382354736,
      "learning_rate": 6.09533455776451e-08,
      "loss": 0.4343,
      "step": 454830
    },
    {
      "epoch": 186.18,
      "grad_norm": 1.7457345724105835,
      "learning_rate": 6.091740898732828e-08,
      "loss": 0.4354,
      "step": 454840
    },
    {
      "epoch": 186.19,
      "grad_norm": 1.932791829109192,
      "learning_rate": 6.08814828543142e-08,
      "loss": 0.419,
      "step": 454850
    },
    {
      "epoch": 186.19,
      "grad_norm": 2.1908483505249023,
      "learning_rate": 6.084556717876731e-08,
      "loss": 0.4509,
      "step": 454860
    },
    {
      "epoch": 186.19,
      "grad_norm": 2.0873684883117676,
      "learning_rate": 6.080966196085205e-08,
      "loss": 0.4371,
      "step": 454870
    },
    {
      "epoch": 186.2,
      "grad_norm": 1.9522783756256104,
      "learning_rate": 6.077376720073287e-08,
      "loss": 0.4508,
      "step": 454880
    },
    {
      "epoch": 186.2,
      "grad_norm": 1.9864144325256348,
      "learning_rate": 6.073788289857425e-08,
      "loss": 0.4455,
      "step": 454890
    },
    {
      "epoch": 186.21,
      "grad_norm": 2.9473016262054443,
      "learning_rate": 6.07020090545406e-08,
      "loss": 0.4507,
      "step": 454900
    },
    {
      "epoch": 186.21,
      "grad_norm": 2.14068603515625,
      "learning_rate": 6.066614566879614e-08,
      "loss": 0.4472,
      "step": 454910
    },
    {
      "epoch": 186.21,
      "grad_norm": 1.8671784400939941,
      "learning_rate": 6.06302927415053e-08,
      "loss": 0.4426,
      "step": 454920
    },
    {
      "epoch": 186.22,
      "grad_norm": 2.0263493061065674,
      "learning_rate": 6.059445027283257e-08,
      "loss": 0.4426,
      "step": 454930
    },
    {
      "epoch": 186.22,
      "grad_norm": 2.2416975498199463,
      "learning_rate": 6.055861826294132e-08,
      "loss": 0.4746,
      "step": 454940
    },
    {
      "epoch": 186.23,
      "grad_norm": 2.148496389389038,
      "learning_rate": 6.052279671199629e-08,
      "loss": 0.4298,
      "step": 454950
    },
    {
      "epoch": 186.23,
      "grad_norm": 2.332686185836792,
      "learning_rate": 6.048698562016166e-08,
      "loss": 0.4392,
      "step": 454960
    },
    {
      "epoch": 186.23,
      "grad_norm": 1.8555582761764526,
      "learning_rate": 6.045118498760109e-08,
      "loss": 0.4407,
      "step": 454970
    },
    {
      "epoch": 186.24,
      "grad_norm": 2.046698570251465,
      "learning_rate": 6.041539481447905e-08,
      "loss": 0.4508,
      "step": 454980
    },
    {
      "epoch": 186.24,
      "grad_norm": 2.0723302364349365,
      "learning_rate": 6.037961510095895e-08,
      "loss": 0.4576,
      "step": 454990
    },
    {
      "epoch": 186.25,
      "grad_norm": 2.2783682346343994,
      "learning_rate": 6.034384584720497e-08,
      "loss": 0.4172,
      "step": 455000
    },
    {
      "epoch": 186.25,
      "grad_norm": 1.859879970550537,
      "learning_rate": 6.030808705338106e-08,
      "loss": 0.4516,
      "step": 455010
    },
    {
      "epoch": 186.25,
      "grad_norm": 2.210904836654663,
      "learning_rate": 6.027233871965086e-08,
      "loss": 0.4471,
      "step": 455020
    },
    {
      "epoch": 186.26,
      "grad_norm": 2.1504788398742676,
      "learning_rate": 6.023660084617833e-08,
      "loss": 0.4542,
      "step": 455030
    },
    {
      "epoch": 186.26,
      "grad_norm": 2.932779550552368,
      "learning_rate": 6.020087343312687e-08,
      "loss": 0.4557,
      "step": 455040
    },
    {
      "epoch": 186.27,
      "grad_norm": 2.1600208282470703,
      "learning_rate": 6.016515648066042e-08,
      "loss": 0.4478,
      "step": 455050
    },
    {
      "epoch": 186.27,
      "grad_norm": 2.2027266025543213,
      "learning_rate": 6.012944998894263e-08,
      "loss": 0.4307,
      "step": 455060
    },
    {
      "epoch": 186.28,
      "grad_norm": 1.9633132219314575,
      "learning_rate": 6.009375395813667e-08,
      "loss": 0.4488,
      "step": 455070
    },
    {
      "epoch": 186.28,
      "grad_norm": 2.3152756690979004,
      "learning_rate": 6.005806838840646e-08,
      "loss": 0.4573,
      "step": 455080
    },
    {
      "epoch": 186.28,
      "grad_norm": 2.0883407592773438,
      "learning_rate": 6.002239327991517e-08,
      "loss": 0.4384,
      "step": 455090
    },
    {
      "epoch": 186.29,
      "grad_norm": 2.738581418991089,
      "learning_rate": 5.998672863282644e-08,
      "loss": 0.4346,
      "step": 455100
    },
    {
      "epoch": 186.29,
      "grad_norm": 1.9249979257583618,
      "learning_rate": 5.995107444730346e-08,
      "loss": 0.4505,
      "step": 455110
    },
    {
      "epoch": 186.3,
      "grad_norm": 2.0863170623779297,
      "learning_rate": 5.991543072350986e-08,
      "loss": 0.4577,
      "step": 455120
    },
    {
      "epoch": 186.3,
      "grad_norm": 2.1120262145996094,
      "learning_rate": 5.98797974616083e-08,
      "loss": 0.4605,
      "step": 455130
    },
    {
      "epoch": 186.3,
      "grad_norm": 2.247562885284424,
      "learning_rate": 5.98441746617627e-08,
      "loss": 0.4517,
      "step": 455140
    },
    {
      "epoch": 186.31,
      "grad_norm": 2.056586503982544,
      "learning_rate": 5.98085623241362e-08,
      "loss": 0.4419,
      "step": 455150
    },
    {
      "epoch": 186.31,
      "grad_norm": 1.6403077840805054,
      "learning_rate": 5.977296044889145e-08,
      "loss": 0.4506,
      "step": 455160
    },
    {
      "epoch": 186.32,
      "grad_norm": 2.425349712371826,
      "learning_rate": 5.973736903619187e-08,
      "loss": 0.4559,
      "step": 455170
    },
    {
      "epoch": 186.32,
      "grad_norm": 1.8632347583770752,
      "learning_rate": 5.97017880862003e-08,
      "loss": 0.4312,
      "step": 455180
    },
    {
      "epoch": 186.32,
      "grad_norm": 2.498269557952881,
      "learning_rate": 5.966621759908022e-08,
      "loss": 0.4466,
      "step": 455190
    },
    {
      "epoch": 186.33,
      "grad_norm": 1.7251458168029785,
      "learning_rate": 5.963065757499421e-08,
      "loss": 0.4232,
      "step": 455200
    },
    {
      "epoch": 186.33,
      "grad_norm": 1.7891324758529663,
      "learning_rate": 5.9595108014104905e-08,
      "loss": 0.4565,
      "step": 455210
    },
    {
      "epoch": 186.34,
      "grad_norm": 2.8853085041046143,
      "learning_rate": 5.955956891657548e-08,
      "loss": 0.4318,
      "step": 455220
    },
    {
      "epoch": 186.34,
      "grad_norm": 2.4017045497894287,
      "learning_rate": 5.952404028256854e-08,
      "loss": 0.4666,
      "step": 455230
    },
    {
      "epoch": 186.34,
      "grad_norm": 2.11037015914917,
      "learning_rate": 5.9488522112247255e-08,
      "loss": 0.437,
      "step": 455240
    },
    {
      "epoch": 186.35,
      "grad_norm": 2.0481109619140625,
      "learning_rate": 5.945301440577373e-08,
      "loss": 0.4472,
      "step": 455250
    },
    {
      "epoch": 186.35,
      "grad_norm": 2.3210787773132324,
      "learning_rate": 5.941751716331111e-08,
      "loss": 0.4536,
      "step": 455260
    },
    {
      "epoch": 186.36,
      "grad_norm": 2.220214366912842,
      "learning_rate": 5.938203038502176e-08,
      "loss": 0.4468,
      "step": 455270
    },
    {
      "epoch": 186.36,
      "grad_norm": 2.527306079864502,
      "learning_rate": 5.9346554071068326e-08,
      "loss": 0.4554,
      "step": 455280
    },
    {
      "epoch": 186.37,
      "grad_norm": 3.18328857421875,
      "learning_rate": 5.931108822161317e-08,
      "loss": 0.4392,
      "step": 455290
    },
    {
      "epoch": 186.37,
      "grad_norm": 4.880648136138916,
      "learning_rate": 5.927563283681892e-08,
      "loss": 0.4326,
      "step": 455300
    },
    {
      "epoch": 186.37,
      "grad_norm": 2.3790433406829834,
      "learning_rate": 5.924018791684821e-08,
      "loss": 0.4562,
      "step": 455310
    },
    {
      "epoch": 186.38,
      "grad_norm": 4.419362545013428,
      "learning_rate": 5.920475346186263e-08,
      "loss": 0.4389,
      "step": 455320
    },
    {
      "epoch": 186.38,
      "grad_norm": 2.214459180831909,
      "learning_rate": 5.916932947202559e-08,
      "loss": 0.4452,
      "step": 455330
    },
    {
      "epoch": 186.39,
      "grad_norm": 1.9679073095321655,
      "learning_rate": 5.9133915947498415e-08,
      "loss": 0.4512,
      "step": 455340
    },
    {
      "epoch": 186.39,
      "grad_norm": 1.9917948246002197,
      "learning_rate": 5.9098512888444e-08,
      "loss": 0.4515,
      "step": 455350
    },
    {
      "epoch": 186.39,
      "grad_norm": 2.044152021408081,
      "learning_rate": 5.9063120295023934e-08,
      "loss": 0.4514,
      "step": 455360
    },
    {
      "epoch": 186.4,
      "grad_norm": 2.1224803924560547,
      "learning_rate": 5.902773816740085e-08,
      "loss": 0.4468,
      "step": 455370
    },
    {
      "epoch": 186.4,
      "grad_norm": 2.1797866821289062,
      "learning_rate": 5.899236650573633e-08,
      "loss": 0.4587,
      "step": 455380
    },
    {
      "epoch": 186.41,
      "grad_norm": 2.357520818710327,
      "learning_rate": 5.895700531019301e-08,
      "loss": 0.4397,
      "step": 455390
    },
    {
      "epoch": 186.41,
      "grad_norm": 1.7548465728759766,
      "learning_rate": 5.892165458093248e-08,
      "loss": 0.4341,
      "step": 455400
    },
    {
      "epoch": 186.41,
      "grad_norm": 1.81541907787323,
      "learning_rate": 5.888631431811659e-08,
      "loss": 0.4377,
      "step": 455410
    },
    {
      "epoch": 186.42,
      "grad_norm": 2.053677797317505,
      "learning_rate": 5.885098452190745e-08,
      "loss": 0.4476,
      "step": 455420
    },
    {
      "epoch": 186.42,
      "grad_norm": 1.9374979734420776,
      "learning_rate": 5.88156651924669e-08,
      "loss": 0.4272,
      "step": 455430
    },
    {
      "epoch": 186.43,
      "grad_norm": 2.335923910140991,
      "learning_rate": 5.8780356329956796e-08,
      "loss": 0.4501,
      "step": 455440
    },
    {
      "epoch": 186.43,
      "grad_norm": 2.113593578338623,
      "learning_rate": 5.8745057934538474e-08,
      "loss": 0.4527,
      "step": 455450
    },
    {
      "epoch": 186.43,
      "grad_norm": 2.031712293624878,
      "learning_rate": 5.870977000637404e-08,
      "loss": 0.4132,
      "step": 455460
    },
    {
      "epoch": 186.44,
      "grad_norm": 1.980558156967163,
      "learning_rate": 5.867449254562508e-08,
      "loss": 0.4304,
      "step": 455470
    },
    {
      "epoch": 186.44,
      "grad_norm": 1.9323465824127197,
      "learning_rate": 5.863922555245319e-08,
      "loss": 0.453,
      "step": 455480
    },
    {
      "epoch": 186.45,
      "grad_norm": 1.636699914932251,
      "learning_rate": 5.860396902701968e-08,
      "loss": 0.4326,
      "step": 455490
    },
    {
      "epoch": 186.45,
      "grad_norm": 1.93903386592865,
      "learning_rate": 5.856872296948617e-08,
      "loss": 0.444,
      "step": 455500
    },
    {
      "epoch": 186.46,
      "grad_norm": 2.1150519847869873,
      "learning_rate": 5.853348738001449e-08,
      "loss": 0.4395,
      "step": 455510
    },
    {
      "epoch": 186.46,
      "grad_norm": 1.6779483556747437,
      "learning_rate": 5.849826225876545e-08,
      "loss": 0.4241,
      "step": 455520
    },
    {
      "epoch": 186.46,
      "grad_norm": 2.29707670211792,
      "learning_rate": 5.84630476059009e-08,
      "loss": 0.4501,
      "step": 455530
    },
    {
      "epoch": 186.47,
      "grad_norm": 2.1168558597564697,
      "learning_rate": 5.842784342158192e-08,
      "loss": 0.4352,
      "step": 455540
    },
    {
      "epoch": 186.47,
      "grad_norm": 2.555586814880371,
      "learning_rate": 5.839264970596983e-08,
      "loss": 0.4657,
      "step": 455550
    },
    {
      "epoch": 186.48,
      "grad_norm": 2.2115278244018555,
      "learning_rate": 5.8357466459225696e-08,
      "loss": 0.457,
      "step": 455560
    },
    {
      "epoch": 186.48,
      "grad_norm": 2.162696361541748,
      "learning_rate": 5.832229368151085e-08,
      "loss": 0.4424,
      "step": 455570
    },
    {
      "epoch": 186.48,
      "grad_norm": 1.9352680444717407,
      "learning_rate": 5.828713137298636e-08,
      "loss": 0.4282,
      "step": 455580
    },
    {
      "epoch": 186.49,
      "grad_norm": 1.7440950870513916,
      "learning_rate": 5.8251979533813565e-08,
      "loss": 0.4675,
      "step": 455590
    },
    {
      "epoch": 186.49,
      "grad_norm": 2.0933351516723633,
      "learning_rate": 5.8216838164153006e-08,
      "loss": 0.4393,
      "step": 455600
    },
    {
      "epoch": 186.5,
      "grad_norm": 1.8463243246078491,
      "learning_rate": 5.818170726416601e-08,
      "loss": 0.4261,
      "step": 455610
    },
    {
      "epoch": 186.5,
      "grad_norm": 2.277294635772705,
      "learning_rate": 5.814658683401339e-08,
      "loss": 0.4553,
      "step": 455620
    },
    {
      "epoch": 186.5,
      "grad_norm": 1.8126113414764404,
      "learning_rate": 5.811147687385595e-08,
      "loss": 0.4391,
      "step": 455630
    },
    {
      "epoch": 186.51,
      "grad_norm": 1.8804278373718262,
      "learning_rate": 5.8076377383855024e-08,
      "loss": 0.4379,
      "step": 455640
    },
    {
      "epoch": 186.51,
      "grad_norm": 2.5509166717529297,
      "learning_rate": 5.804128836417064e-08,
      "loss": 0.4537,
      "step": 455650
    },
    {
      "epoch": 186.52,
      "grad_norm": 2.5084738731384277,
      "learning_rate": 5.8006209814963863e-08,
      "loss": 0.4509,
      "step": 455660
    },
    {
      "epoch": 186.52,
      "grad_norm": 2.172193765640259,
      "learning_rate": 5.797114173639525e-08,
      "loss": 0.4455,
      "step": 455670
    },
    {
      "epoch": 186.52,
      "grad_norm": 1.918947696685791,
      "learning_rate": 5.793608412862586e-08,
      "loss": 0.4403,
      "step": 455680
    },
    {
      "epoch": 186.53,
      "grad_norm": 2.228466510772705,
      "learning_rate": 5.790103699181573e-08,
      "loss": 0.4174,
      "step": 455690
    },
    {
      "epoch": 186.53,
      "grad_norm": 1.9688446521759033,
      "learning_rate": 5.786600032612592e-08,
      "loss": 0.4405,
      "step": 455700
    },
    {
      "epoch": 186.54,
      "grad_norm": 1.8476011753082275,
      "learning_rate": 5.7830974131716464e-08,
      "loss": 0.4802,
      "step": 455710
    },
    {
      "epoch": 186.54,
      "grad_norm": 2.415529251098633,
      "learning_rate": 5.779595840874817e-08,
      "loss": 0.4409,
      "step": 455720
    },
    {
      "epoch": 186.55,
      "grad_norm": 2.0108580589294434,
      "learning_rate": 5.776095315738159e-08,
      "loss": 0.4386,
      "step": 455730
    },
    {
      "epoch": 186.55,
      "grad_norm": 2.2321648597717285,
      "learning_rate": 5.772595837777649e-08,
      "loss": 0.4382,
      "step": 455740
    },
    {
      "epoch": 186.55,
      "grad_norm": 2.198727607727051,
      "learning_rate": 5.769097407009367e-08,
      "loss": 0.4565,
      "step": 455750
    },
    {
      "epoch": 186.56,
      "grad_norm": 2.1479504108428955,
      "learning_rate": 5.7656000234492916e-08,
      "loss": 0.4383,
      "step": 455760
    },
    {
      "epoch": 186.56,
      "grad_norm": 1.9057972431182861,
      "learning_rate": 5.762103687113502e-08,
      "loss": 0.4439,
      "step": 455770
    },
    {
      "epoch": 186.57,
      "grad_norm": 2.1522269248962402,
      "learning_rate": 5.758608398018002e-08,
      "loss": 0.4535,
      "step": 455780
    },
    {
      "epoch": 186.57,
      "grad_norm": 2.035540819168091,
      "learning_rate": 5.755114156178768e-08,
      "loss": 0.4449,
      "step": 455790
    },
    {
      "epoch": 186.57,
      "grad_norm": 1.8139671087265015,
      "learning_rate": 5.751620961611802e-08,
      "loss": 0.4301,
      "step": 455800
    },
    {
      "epoch": 186.58,
      "grad_norm": 2.2027480602264404,
      "learning_rate": 5.748128814333161e-08,
      "loss": 0.4414,
      "step": 455810
    },
    {
      "epoch": 186.58,
      "grad_norm": 2.2854511737823486,
      "learning_rate": 5.744637714358794e-08,
      "loss": 0.4287,
      "step": 455820
    },
    {
      "epoch": 186.59,
      "grad_norm": 1.594376802444458,
      "learning_rate": 5.741147661704705e-08,
      "loss": 0.449,
      "step": 455830
    },
    {
      "epoch": 186.59,
      "grad_norm": 2.1557202339172363,
      "learning_rate": 5.7376586563868693e-08,
      "loss": 0.4356,
      "step": 455840
    },
    {
      "epoch": 186.59,
      "grad_norm": 1.8773729801177979,
      "learning_rate": 5.734170698421317e-08,
      "loss": 0.4407,
      "step": 455850
    },
    {
      "epoch": 186.6,
      "grad_norm": 2.3837945461273193,
      "learning_rate": 5.730683787823973e-08,
      "loss": 0.4174,
      "step": 455860
    },
    {
      "epoch": 186.6,
      "grad_norm": 2.0911643505096436,
      "learning_rate": 5.727197924610839e-08,
      "loss": 0.4448,
      "step": 455870
    },
    {
      "epoch": 186.61,
      "grad_norm": 2.113445520401001,
      "learning_rate": 5.723713108797866e-08,
      "loss": 0.4488,
      "step": 455880
    },
    {
      "epoch": 186.61,
      "grad_norm": 2.0082695484161377,
      "learning_rate": 5.720229340401032e-08,
      "loss": 0.4322,
      "step": 455890
    },
    {
      "epoch": 186.61,
      "grad_norm": 1.9110026359558105,
      "learning_rate": 5.716746619436286e-08,
      "loss": 0.4404,
      "step": 455900
    },
    {
      "epoch": 186.62,
      "grad_norm": 1.9169986248016357,
      "learning_rate": 5.713264945919607e-08,
      "loss": 0.4361,
      "step": 455910
    },
    {
      "epoch": 186.62,
      "grad_norm": 1.8015373945236206,
      "learning_rate": 5.709784319866918e-08,
      "loss": 0.4346,
      "step": 455920
    },
    {
      "epoch": 186.63,
      "grad_norm": 2.130396604537964,
      "learning_rate": 5.70630474129417e-08,
      "loss": 0.4317,
      "step": 455930
    },
    {
      "epoch": 186.63,
      "grad_norm": 2.001899003982544,
      "learning_rate": 5.702826210217289e-08,
      "loss": 0.4401,
      "step": 455940
    },
    {
      "epoch": 186.64,
      "grad_norm": 1.894518256187439,
      "learning_rate": 5.6993487266522496e-08,
      "loss": 0.4514,
      "step": 455950
    },
    {
      "epoch": 186.64,
      "grad_norm": 2.315485715866089,
      "learning_rate": 5.695872290614927e-08,
      "loss": 0.4354,
      "step": 455960
    },
    {
      "epoch": 186.64,
      "grad_norm": 2.077392816543579,
      "learning_rate": 5.6923969021212703e-08,
      "loss": 0.4491,
      "step": 455970
    },
    {
      "epoch": 186.65,
      "grad_norm": 1.846231460571289,
      "learning_rate": 5.688922561187205e-08,
      "loss": 0.4493,
      "step": 455980
    },
    {
      "epoch": 186.65,
      "grad_norm": 2.1738739013671875,
      "learning_rate": 5.685449267828655e-08,
      "loss": 0.4488,
      "step": 455990
    },
    {
      "epoch": 186.66,
      "grad_norm": 2.304295301437378,
      "learning_rate": 5.68197702206152e-08,
      "loss": 0.4381,
      "step": 456000
    },
    {
      "epoch": 186.66,
      "grad_norm": 1.7901132106781006,
      "learning_rate": 5.678505823901698e-08,
      "loss": 0.4463,
      "step": 456010
    },
    {
      "epoch": 186.66,
      "grad_norm": 2.2896058559417725,
      "learning_rate": 5.6750356733650893e-08,
      "loss": 0.4438,
      "step": 456020
    },
    {
      "epoch": 186.67,
      "grad_norm": 1.978683590888977,
      "learning_rate": 5.671566570467616e-08,
      "loss": 0.4475,
      "step": 456030
    },
    {
      "epoch": 186.67,
      "grad_norm": 2.3263192176818848,
      "learning_rate": 5.6680985152251534e-08,
      "loss": 0.4448,
      "step": 456040
    },
    {
      "epoch": 186.68,
      "grad_norm": 1.944924235343933,
      "learning_rate": 5.664631507653573e-08,
      "loss": 0.4439,
      "step": 456050
    },
    {
      "epoch": 186.68,
      "grad_norm": 1.972046136856079,
      "learning_rate": 5.6611655477687735e-08,
      "loss": 0.4268,
      "step": 456060
    },
    {
      "epoch": 186.68,
      "grad_norm": 2.241687297821045,
      "learning_rate": 5.657700635586628e-08,
      "loss": 0.4431,
      "step": 456070
    },
    {
      "epoch": 186.69,
      "grad_norm": 2.039198875427246,
      "learning_rate": 5.6542367711230086e-08,
      "loss": 0.4416,
      "step": 456080
    },
    {
      "epoch": 186.69,
      "grad_norm": 1.9851362705230713,
      "learning_rate": 5.650773954393789e-08,
      "loss": 0.4312,
      "step": 456090
    },
    {
      "epoch": 186.7,
      "grad_norm": 2.1966655254364014,
      "learning_rate": 5.647312185414841e-08,
      "loss": 0.4338,
      "step": 456100
    },
    {
      "epoch": 186.7,
      "grad_norm": 2.5989603996276855,
      "learning_rate": 5.6438514642019856e-08,
      "loss": 0.4602,
      "step": 456110
    },
    {
      "epoch": 186.7,
      "grad_norm": 2.1728994846343994,
      "learning_rate": 5.6403917907710956e-08,
      "loss": 0.4365,
      "step": 456120
    },
    {
      "epoch": 186.71,
      "grad_norm": 2.1116783618927,
      "learning_rate": 5.6369331651380436e-08,
      "loss": 0.4521,
      "step": 456130
    },
    {
      "epoch": 186.71,
      "grad_norm": 2.0759785175323486,
      "learning_rate": 5.633475587318625e-08,
      "loss": 0.4365,
      "step": 456140
    },
    {
      "epoch": 186.72,
      "grad_norm": 1.9390959739685059,
      "learning_rate": 5.630019057328711e-08,
      "loss": 0.4505,
      "step": 456150
    },
    {
      "epoch": 186.72,
      "grad_norm": 2.0550127029418945,
      "learning_rate": 5.6265635751841494e-08,
      "loss": 0.4469,
      "step": 456160
    },
    {
      "epoch": 186.73,
      "grad_norm": 2.36738920211792,
      "learning_rate": 5.6231091409007343e-08,
      "loss": 0.4371,
      "step": 456170
    },
    {
      "epoch": 186.73,
      "grad_norm": 2.1660501956939697,
      "learning_rate": 5.619655754494313e-08,
      "loss": 0.4611,
      "step": 456180
    },
    {
      "epoch": 186.73,
      "grad_norm": 2.3022356033325195,
      "learning_rate": 5.61620341598068e-08,
      "loss": 0.4425,
      "step": 456190
    },
    {
      "epoch": 186.74,
      "grad_norm": 2.1241137981414795,
      "learning_rate": 5.612752125375682e-08,
      "loss": 0.4369,
      "step": 456200
    },
    {
      "epoch": 186.74,
      "grad_norm": 2.2139008045196533,
      "learning_rate": 5.609301882695087e-08,
      "loss": 0.4357,
      "step": 456210
    },
    {
      "epoch": 186.75,
      "grad_norm": 1.9816657304763794,
      "learning_rate": 5.605852687954769e-08,
      "loss": 0.4411,
      "step": 456220
    },
    {
      "epoch": 186.75,
      "grad_norm": 1.8559845685958862,
      "learning_rate": 5.602404541170469e-08,
      "loss": 0.4557,
      "step": 456230
    },
    {
      "epoch": 186.75,
      "grad_norm": 2.3174612522125244,
      "learning_rate": 5.598957442358009e-08,
      "loss": 0.4455,
      "step": 456240
    },
    {
      "epoch": 186.76,
      "grad_norm": 1.938205361366272,
      "learning_rate": 5.595511391533131e-08,
      "loss": 0.4312,
      "step": 456250
    },
    {
      "epoch": 186.76,
      "grad_norm": 1.888850450515747,
      "learning_rate": 5.592066388711682e-08,
      "loss": 0.4698,
      "step": 456260
    },
    {
      "epoch": 186.77,
      "grad_norm": 2.170841932296753,
      "learning_rate": 5.5886224339094043e-08,
      "loss": 0.4464,
      "step": 456270
    },
    {
      "epoch": 186.77,
      "grad_norm": 1.911054253578186,
      "learning_rate": 5.5851795271421186e-08,
      "loss": 0.4439,
      "step": 456280
    },
    {
      "epoch": 186.77,
      "grad_norm": 2.1116702556610107,
      "learning_rate": 5.581737668425542e-08,
      "loss": 0.4303,
      "step": 456290
    },
    {
      "epoch": 186.78,
      "grad_norm": 2.020498514175415,
      "learning_rate": 5.578296857775494e-08,
      "loss": 0.4305,
      "step": 456300
    },
    {
      "epoch": 186.78,
      "grad_norm": 1.9350842237472534,
      "learning_rate": 5.5748570952076925e-08,
      "loss": 0.4546,
      "step": 456310
    },
    {
      "epoch": 186.79,
      "grad_norm": 1.993107557296753,
      "learning_rate": 5.571418380737906e-08,
      "loss": 0.4652,
      "step": 456320
    },
    {
      "epoch": 186.79,
      "grad_norm": 2.3837954998016357,
      "learning_rate": 5.567980714381902e-08,
      "loss": 0.4297,
      "step": 456330
    },
    {
      "epoch": 186.79,
      "grad_norm": 2.046384572982788,
      "learning_rate": 5.5645440961554245e-08,
      "loss": 0.4382,
      "step": 456340
    },
    {
      "epoch": 186.8,
      "grad_norm": 1.938098430633545,
      "learning_rate": 5.561108526074189e-08,
      "loss": 0.4354,
      "step": 456350
    },
    {
      "epoch": 186.8,
      "grad_norm": 1.821979284286499,
      "learning_rate": 5.557674004153991e-08,
      "loss": 0.4252,
      "step": 456360
    },
    {
      "epoch": 186.81,
      "grad_norm": 2.393677234649658,
      "learning_rate": 5.554240530410495e-08,
      "loss": 0.4431,
      "step": 456370
    },
    {
      "epoch": 186.81,
      "grad_norm": 2.2671732902526855,
      "learning_rate": 5.550808104859469e-08,
      "loss": 0.4478,
      "step": 456380
    },
    {
      "epoch": 186.82,
      "grad_norm": 1.8961701393127441,
      "learning_rate": 5.5473767275166296e-08,
      "loss": 0.433,
      "step": 456390
    },
    {
      "epoch": 186.82,
      "grad_norm": 2.0565261840820312,
      "learning_rate": 5.5439463983976936e-08,
      "loss": 0.4529,
      "step": 456400
    },
    {
      "epoch": 186.82,
      "grad_norm": 1.8148969411849976,
      "learning_rate": 5.540517117518378e-08,
      "loss": 0.4496,
      "step": 456410
    },
    {
      "epoch": 186.83,
      "grad_norm": 2.4918501377105713,
      "learning_rate": 5.537088884894373e-08,
      "loss": 0.4335,
      "step": 456420
    },
    {
      "epoch": 186.83,
      "grad_norm": 2.406207323074341,
      "learning_rate": 5.5336617005414216e-08,
      "loss": 0.445,
      "step": 456430
    },
    {
      "epoch": 186.84,
      "grad_norm": 1.769470453262329,
      "learning_rate": 5.530235564475188e-08,
      "loss": 0.444,
      "step": 456440
    },
    {
      "epoch": 186.84,
      "grad_norm": 2.1736648082733154,
      "learning_rate": 5.5268104767113624e-08,
      "loss": 0.4432,
      "step": 456450
    },
    {
      "epoch": 186.84,
      "grad_norm": 1.9625827074050903,
      "learning_rate": 5.5233864372656625e-08,
      "loss": 0.4294,
      "step": 456460
    },
    {
      "epoch": 186.85,
      "grad_norm": 2.0276126861572266,
      "learning_rate": 5.519963446153778e-08,
      "loss": 0.4441,
      "step": 456470
    },
    {
      "epoch": 186.85,
      "grad_norm": 1.8822956085205078,
      "learning_rate": 5.516541503391373e-08,
      "loss": 0.44,
      "step": 456480
    },
    {
      "epoch": 186.86,
      "grad_norm": 2.125004291534424,
      "learning_rate": 5.5131206089941133e-08,
      "loss": 0.4384,
      "step": 456490
    },
    {
      "epoch": 186.86,
      "grad_norm": 1.9741848707199097,
      "learning_rate": 5.5097007629776883e-08,
      "loss": 0.4415,
      "step": 456500
    },
    {
      "epoch": 186.86,
      "grad_norm": 1.6215320825576782,
      "learning_rate": 5.5062819653577637e-08,
      "loss": 0.4755,
      "step": 456510
    },
    {
      "epoch": 186.87,
      "grad_norm": 1.9545795917510986,
      "learning_rate": 5.502864216149951e-08,
      "loss": 0.4696,
      "step": 456520
    },
    {
      "epoch": 186.87,
      "grad_norm": 1.8528021574020386,
      "learning_rate": 5.499447515369994e-08,
      "loss": 0.4501,
      "step": 456530
    },
    {
      "epoch": 186.88,
      "grad_norm": 2.3216655254364014,
      "learning_rate": 5.4960318630334516e-08,
      "loss": 0.4579,
      "step": 456540
    },
    {
      "epoch": 186.88,
      "grad_norm": 2.167081117630005,
      "learning_rate": 5.492617259156042e-08,
      "loss": 0.4582,
      "step": 456550
    },
    {
      "epoch": 186.88,
      "grad_norm": 2.054116725921631,
      "learning_rate": 5.489203703753376e-08,
      "loss": 0.4316,
      "step": 456560
    },
    {
      "epoch": 186.89,
      "grad_norm": 2.3703701496124268,
      "learning_rate": 5.485791196841094e-08,
      "loss": 0.4199,
      "step": 456570
    },
    {
      "epoch": 186.89,
      "grad_norm": 2.1668336391448975,
      "learning_rate": 5.482379738434833e-08,
      "loss": 0.4412,
      "step": 456580
    },
    {
      "epoch": 186.9,
      "grad_norm": 1.9911085367202759,
      "learning_rate": 5.4789693285502316e-08,
      "loss": 0.4361,
      "step": 456590
    },
    {
      "epoch": 186.9,
      "grad_norm": 2.049077272415161,
      "learning_rate": 5.475559967202878e-08,
      "loss": 0.4148,
      "step": 456600
    },
    {
      "epoch": 186.91,
      "grad_norm": 2.053891181945801,
      "learning_rate": 5.4721516544084086e-08,
      "loss": 0.4392,
      "step": 456610
    },
    {
      "epoch": 186.91,
      "grad_norm": 1.876854419708252,
      "learning_rate": 5.4687443901824376e-08,
      "loss": 0.4439,
      "step": 456620
    },
    {
      "epoch": 186.91,
      "grad_norm": 2.0406198501586914,
      "learning_rate": 5.465338174540576e-08,
      "loss": 0.4491,
      "step": 456630
    },
    {
      "epoch": 186.92,
      "grad_norm": 2.4450299739837646,
      "learning_rate": 5.461933007498437e-08,
      "loss": 0.454,
      "step": 456640
    },
    {
      "epoch": 186.92,
      "grad_norm": 4.574876308441162,
      "learning_rate": 5.4585288890716085e-08,
      "loss": 0.4382,
      "step": 456650
    },
    {
      "epoch": 186.93,
      "grad_norm": 2.045074462890625,
      "learning_rate": 5.455125819275649e-08,
      "loss": 0.43,
      "step": 456660
    },
    {
      "epoch": 186.93,
      "grad_norm": 2.0086543560028076,
      "learning_rate": 5.451723798126197e-08,
      "loss": 0.4593,
      "step": 456670
    },
    {
      "epoch": 186.93,
      "grad_norm": 2.3543736934661865,
      "learning_rate": 5.44832282563884e-08,
      "loss": 0.4447,
      "step": 456680
    },
    {
      "epoch": 186.94,
      "grad_norm": 2.0932886600494385,
      "learning_rate": 5.4449229018291124e-08,
      "loss": 0.441,
      "step": 456690
    },
    {
      "epoch": 186.94,
      "grad_norm": 2.2694671154022217,
      "learning_rate": 5.441524026712626e-08,
      "loss": 0.4602,
      "step": 456700
    },
    {
      "epoch": 186.95,
      "grad_norm": 1.8165432214736938,
      "learning_rate": 5.438126200304942e-08,
      "loss": 0.4559,
      "step": 456710
    },
    {
      "epoch": 186.95,
      "grad_norm": 2.094085693359375,
      "learning_rate": 5.434729422621594e-08,
      "loss": 0.4451,
      "step": 456720
    },
    {
      "epoch": 186.95,
      "grad_norm": 1.7613261938095093,
      "learning_rate": 5.431333693678196e-08,
      "loss": 0.4506,
      "step": 456730
    },
    {
      "epoch": 186.96,
      "grad_norm": 2.335557222366333,
      "learning_rate": 5.427939013490281e-08,
      "loss": 0.4516,
      "step": 456740
    },
    {
      "epoch": 186.96,
      "grad_norm": 2.301133394241333,
      "learning_rate": 5.424545382073384e-08,
      "loss": 0.446,
      "step": 456750
    },
    {
      "epoch": 186.97,
      "grad_norm": 1.9867017269134521,
      "learning_rate": 5.421152799443065e-08,
      "loss": 0.4369,
      "step": 456760
    },
    {
      "epoch": 186.97,
      "grad_norm": 2.5358474254608154,
      "learning_rate": 5.41776126561486e-08,
      "loss": 0.4425,
      "step": 456770
    },
    {
      "epoch": 186.98,
      "grad_norm": 1.7854187488555908,
      "learning_rate": 5.414370780604301e-08,
      "loss": 0.4372,
      "step": 456780
    },
    {
      "epoch": 186.98,
      "grad_norm": 1.9127510786056519,
      "learning_rate": 5.410981344426925e-08,
      "loss": 0.4413,
      "step": 456790
    },
    {
      "epoch": 186.98,
      "grad_norm": 2.198866605758667,
      "learning_rate": 5.407592957098264e-08,
      "loss": 0.4468,
      "step": 456800
    },
    {
      "epoch": 186.99,
      "grad_norm": 1.838700294494629,
      "learning_rate": 5.404205618633854e-08,
      "loss": 0.4388,
      "step": 456810
    },
    {
      "epoch": 186.99,
      "grad_norm": 2.3319272994995117,
      "learning_rate": 5.400819329049178e-08,
      "loss": 0.4429,
      "step": 456820
    },
    {
      "epoch": 187.0,
      "grad_norm": 1.7733969688415527,
      "learning_rate": 5.397434088359742e-08,
      "loss": 0.4282,
      "step": 456830
    },
    {
      "epoch": 187.0,
      "grad_norm": 2.0704855918884277,
      "learning_rate": 5.394049896581083e-08,
      "loss": 0.4431,
      "step": 456840
    },
    {
      "epoch": 187.0,
      "eval_loss": 0.44262972474098206,
      "eval_runtime": 43.2477,
      "eval_samples_per_second": 79.75,
      "eval_steps_per_second": 9.989,
      "step": 456841
    },
    {
      "epoch": 187.0,
      "grad_norm": 2.0132391452789307,
      "learning_rate": 5.390666753728708e-08,
      "loss": 0.4412,
      "step": 456850
    },
    {
      "epoch": 187.01,
      "grad_norm": 1.9923040866851807,
      "learning_rate": 5.3872846598181e-08,
      "loss": 0.4505,
      "step": 456860
    },
    {
      "epoch": 187.01,
      "grad_norm": 2.0419416427612305,
      "learning_rate": 5.383903614864741e-08,
      "loss": 0.4332,
      "step": 456870
    },
    {
      "epoch": 187.02,
      "grad_norm": 1.9687880277633667,
      "learning_rate": 5.3805236188841397e-08,
      "loss": 0.4443,
      "step": 456880
    },
    {
      "epoch": 187.02,
      "grad_norm": 2.0448479652404785,
      "learning_rate": 5.377144671891778e-08,
      "loss": 0.4486,
      "step": 456890
    },
    {
      "epoch": 187.02,
      "grad_norm": 2.141786575317383,
      "learning_rate": 5.3737667739031135e-08,
      "loss": 0.4649,
      "step": 456900
    },
    {
      "epoch": 187.03,
      "grad_norm": 2.38903546333313,
      "learning_rate": 5.370389924933627e-08,
      "loss": 0.4466,
      "step": 456910
    },
    {
      "epoch": 187.03,
      "grad_norm": 1.9109693765640259,
      "learning_rate": 5.367014124998802e-08,
      "loss": 0.4591,
      "step": 456920
    },
    {
      "epoch": 187.04,
      "grad_norm": 1.9225218296051025,
      "learning_rate": 5.363639374114095e-08,
      "loss": 0.435,
      "step": 456930
    },
    {
      "epoch": 187.04,
      "grad_norm": 2.0460455417633057,
      "learning_rate": 5.3602656722949876e-08,
      "loss": 0.4443,
      "step": 456940
    },
    {
      "epoch": 187.04,
      "grad_norm": 2.194833517074585,
      "learning_rate": 5.356893019556911e-08,
      "loss": 0.4572,
      "step": 456950
    },
    {
      "epoch": 187.05,
      "grad_norm": 1.8840223550796509,
      "learning_rate": 5.3535214159152685e-08,
      "loss": 0.4485,
      "step": 456960
    },
    {
      "epoch": 187.05,
      "grad_norm": 2.067323923110962,
      "learning_rate": 5.350150861385596e-08,
      "loss": 0.4476,
      "step": 456970
    },
    {
      "epoch": 187.06,
      "grad_norm": 2.141353130340576,
      "learning_rate": 5.346781355983244e-08,
      "loss": 0.4509,
      "step": 456980
    },
    {
      "epoch": 187.06,
      "grad_norm": 2.0150768756866455,
      "learning_rate": 5.343412899723723e-08,
      "loss": 0.4385,
      "step": 456990
    },
    {
      "epoch": 187.07,
      "grad_norm": 2.1091768741607666,
      "learning_rate": 5.3400454926224355e-08,
      "loss": 0.433,
      "step": 457000
    },
    {
      "epoch": 187.07,
      "grad_norm": 2.1623177528381348,
      "learning_rate": 5.336679134694761e-08,
      "loss": 0.4579,
      "step": 457010
    },
    {
      "epoch": 187.07,
      "grad_norm": 1.7549914121627808,
      "learning_rate": 5.3333138259562066e-08,
      "loss": 0.4412,
      "step": 457020
    },
    {
      "epoch": 187.08,
      "grad_norm": 2.3489534854888916,
      "learning_rate": 5.329949566422126e-08,
      "loss": 0.4492,
      "step": 457030
    },
    {
      "epoch": 187.08,
      "grad_norm": 2.417947769165039,
      "learning_rate": 5.3265863561079496e-08,
      "loss": 0.4515,
      "step": 457040
    },
    {
      "epoch": 187.09,
      "grad_norm": 1.9118930101394653,
      "learning_rate": 5.323224195029081e-08,
      "loss": 0.4414,
      "step": 457050
    },
    {
      "epoch": 187.09,
      "grad_norm": 2.2211601734161377,
      "learning_rate": 5.31986308320095e-08,
      "loss": 0.428,
      "step": 457060
    },
    {
      "epoch": 187.09,
      "grad_norm": 2.070833444595337,
      "learning_rate": 5.31650302063891e-08,
      "loss": 0.4393,
      "step": 457070
    },
    {
      "epoch": 187.1,
      "grad_norm": 2.0175931453704834,
      "learning_rate": 5.313144007358392e-08,
      "loss": 0.4501,
      "step": 457080
    },
    {
      "epoch": 187.1,
      "grad_norm": 2.751713991165161,
      "learning_rate": 5.3097860433747465e-08,
      "loss": 0.4616,
      "step": 457090
    },
    {
      "epoch": 187.11,
      "grad_norm": 2.1077358722686768,
      "learning_rate": 5.306429128703353e-08,
      "loss": 0.4354,
      "step": 457100
    },
    {
      "epoch": 187.11,
      "grad_norm": 2.4378128051757812,
      "learning_rate": 5.303073263359615e-08,
      "loss": 0.4268,
      "step": 457110
    },
    {
      "epoch": 187.11,
      "grad_norm": 1.9452214241027832,
      "learning_rate": 5.299718447358912e-08,
      "loss": 0.4478,
      "step": 457120
    },
    {
      "epoch": 187.12,
      "grad_norm": 2.2813308238983154,
      "learning_rate": 5.296364680716595e-08,
      "loss": 0.4728,
      "step": 457130
    },
    {
      "epoch": 187.12,
      "grad_norm": 1.96356999874115,
      "learning_rate": 5.2930119634480173e-08,
      "loss": 0.4542,
      "step": 457140
    },
    {
      "epoch": 187.13,
      "grad_norm": 2.132148504257202,
      "learning_rate": 5.289660295568557e-08,
      "loss": 0.4439,
      "step": 457150
    },
    {
      "epoch": 187.13,
      "grad_norm": 2.0554869174957275,
      "learning_rate": 5.28630967709354e-08,
      "loss": 0.4321,
      "step": 457160
    },
    {
      "epoch": 187.13,
      "grad_norm": 2.162386417388916,
      "learning_rate": 5.282960108038345e-08,
      "loss": 0.45,
      "step": 457170
    },
    {
      "epoch": 187.14,
      "grad_norm": 1.9713406562805176,
      "learning_rate": 5.279611588418324e-08,
      "loss": 0.4537,
      "step": 457180
    },
    {
      "epoch": 187.14,
      "grad_norm": 2.1111397743225098,
      "learning_rate": 5.276264118248778e-08,
      "loss": 0.4515,
      "step": 457190
    },
    {
      "epoch": 187.15,
      "grad_norm": 2.162804365158081,
      "learning_rate": 5.272917697545032e-08,
      "loss": 0.4678,
      "step": 457200
    },
    {
      "epoch": 187.15,
      "grad_norm": 2.729948043823242,
      "learning_rate": 5.269572326322491e-08,
      "loss": 0.4417,
      "step": 457210
    },
    {
      "epoch": 187.16,
      "grad_norm": 2.273477792739868,
      "learning_rate": 5.26622800459643e-08,
      "loss": 0.435,
      "step": 457220
    },
    {
      "epoch": 187.16,
      "grad_norm": 1.940871238708496,
      "learning_rate": 5.262884732382174e-08,
      "loss": 0.424,
      "step": 457230
    },
    {
      "epoch": 187.16,
      "grad_norm": 2.0238864421844482,
      "learning_rate": 5.259542509695024e-08,
      "loss": 0.4429,
      "step": 457240
    },
    {
      "epoch": 187.17,
      "grad_norm": 2.247722625732422,
      "learning_rate": 5.2562013365503067e-08,
      "loss": 0.4629,
      "step": 457250
    },
    {
      "epoch": 187.17,
      "grad_norm": 2.195740222930908,
      "learning_rate": 5.252861212963295e-08,
      "loss": 0.4635,
      "step": 457260
    },
    {
      "epoch": 187.18,
      "grad_norm": 1.897636890411377,
      "learning_rate": 5.2495221389493425e-08,
      "loss": 0.4496,
      "step": 457270
    },
    {
      "epoch": 187.18,
      "grad_norm": 1.9300737380981445,
      "learning_rate": 5.246184114523723e-08,
      "loss": 0.4267,
      "step": 457280
    },
    {
      "epoch": 187.18,
      "grad_norm": 1.8978315591812134,
      "learning_rate": 5.24284713970171e-08,
      "loss": 0.4528,
      "step": 457290
    },
    {
      "epoch": 187.19,
      "grad_norm": 2.0446555614471436,
      "learning_rate": 5.239511214498605e-08,
      "loss": 0.4414,
      "step": 457300
    },
    {
      "epoch": 187.19,
      "grad_norm": 2.2486462593078613,
      "learning_rate": 5.2361763389297074e-08,
      "loss": 0.452,
      "step": 457310
    },
    {
      "epoch": 187.2,
      "grad_norm": 1.9731738567352295,
      "learning_rate": 5.2328425130102656e-08,
      "loss": 0.4392,
      "step": 457320
    },
    {
      "epoch": 187.2,
      "grad_norm": 1.931825041770935,
      "learning_rate": 5.229509736755554e-08,
      "loss": 0.4336,
      "step": 457330
    },
    {
      "epoch": 187.2,
      "grad_norm": 2.2941248416900635,
      "learning_rate": 5.226178010180847e-08,
      "loss": 0.4439,
      "step": 457340
    },
    {
      "epoch": 187.21,
      "grad_norm": 2.0784912109375,
      "learning_rate": 5.222847333301392e-08,
      "loss": 0.4289,
      "step": 457350
    },
    {
      "epoch": 187.21,
      "grad_norm": 1.8201066255569458,
      "learning_rate": 5.2195177061324906e-08,
      "loss": 0.4198,
      "step": 457360
    },
    {
      "epoch": 187.22,
      "grad_norm": 2.2070960998535156,
      "learning_rate": 5.216189128689338e-08,
      "loss": 0.4453,
      "step": 457370
    },
    {
      "epoch": 187.22,
      "grad_norm": 2.126946449279785,
      "learning_rate": 5.212861600987208e-08,
      "loss": 0.4221,
      "step": 457380
    },
    {
      "epoch": 187.22,
      "grad_norm": 1.9710829257965088,
      "learning_rate": 5.20953512304135e-08,
      "loss": 0.4492,
      "step": 457390
    },
    {
      "epoch": 187.23,
      "grad_norm": 1.717462420463562,
      "learning_rate": 5.206209694867012e-08,
      "loss": 0.428,
      "step": 457400
    },
    {
      "epoch": 187.23,
      "grad_norm": 2.0390000343322754,
      "learning_rate": 5.20288531647939e-08,
      "loss": 0.4621,
      "step": 457410
    },
    {
      "epoch": 187.24,
      "grad_norm": 2.931379795074463,
      "learning_rate": 5.1995619878937316e-08,
      "loss": 0.44,
      "step": 457420
    },
    {
      "epoch": 187.24,
      "grad_norm": 1.894526720046997,
      "learning_rate": 5.196239709125286e-08,
      "loss": 0.4541,
      "step": 457430
    },
    {
      "epoch": 187.25,
      "grad_norm": 1.8566277027130127,
      "learning_rate": 5.1929184801892226e-08,
      "loss": 0.448,
      "step": 457440
    },
    {
      "epoch": 187.25,
      "grad_norm": 2.2867887020111084,
      "learning_rate": 5.1895983011007896e-08,
      "loss": 0.4438,
      "step": 457450
    },
    {
      "epoch": 187.25,
      "grad_norm": 2.422372817993164,
      "learning_rate": 5.18627917187521e-08,
      "loss": 0.4579,
      "step": 457460
    },
    {
      "epoch": 187.26,
      "grad_norm": 2.072197914123535,
      "learning_rate": 5.1829610925276533e-08,
      "loss": 0.4284,
      "step": 457470
    },
    {
      "epoch": 187.26,
      "grad_norm": 2.050022840499878,
      "learning_rate": 5.1796440630733155e-08,
      "loss": 0.4285,
      "step": 457480
    },
    {
      "epoch": 187.27,
      "grad_norm": 2.0357823371887207,
      "learning_rate": 5.17632808352742e-08,
      "loss": 0.4449,
      "step": 457490
    },
    {
      "epoch": 187.27,
      "grad_norm": 2.3333611488342285,
      "learning_rate": 5.173013153905135e-08,
      "loss": 0.4577,
      "step": 457500
    },
    {
      "epoch": 187.27,
      "grad_norm": 2.068669080734253,
      "learning_rate": 5.1696992742216584e-08,
      "loss": 0.4469,
      "step": 457510
    },
    {
      "epoch": 187.28,
      "grad_norm": 2.3200302124023438,
      "learning_rate": 5.166386444492186e-08,
      "loss": 0.4414,
      "step": 457520
    },
    {
      "epoch": 187.28,
      "grad_norm": 1.9747493267059326,
      "learning_rate": 5.163074664731887e-08,
      "loss": 0.4563,
      "step": 457530
    },
    {
      "epoch": 187.29,
      "grad_norm": 2.1269490718841553,
      "learning_rate": 5.159763934955881e-08,
      "loss": 0.4592,
      "step": 457540
    },
    {
      "epoch": 187.29,
      "grad_norm": 2.0459606647491455,
      "learning_rate": 5.156454255179389e-08,
      "loss": 0.4504,
      "step": 457550
    },
    {
      "epoch": 187.29,
      "grad_norm": 1.764477014541626,
      "learning_rate": 5.1531456254175555e-08,
      "loss": 0.4392,
      "step": 457560
    },
    {
      "epoch": 187.3,
      "grad_norm": 1.9151231050491333,
      "learning_rate": 5.149838045685551e-08,
      "loss": 0.4446,
      "step": 457570
    },
    {
      "epoch": 187.3,
      "grad_norm": 4.053582191467285,
      "learning_rate": 5.146531515998467e-08,
      "loss": 0.4544,
      "step": 457580
    },
    {
      "epoch": 187.31,
      "grad_norm": 2.2883410453796387,
      "learning_rate": 5.1432260363715256e-08,
      "loss": 0.4478,
      "step": 457590
    },
    {
      "epoch": 187.31,
      "grad_norm": 2.2389657497406006,
      "learning_rate": 5.139921606819846e-08,
      "loss": 0.4334,
      "step": 457600
    },
    {
      "epoch": 187.31,
      "grad_norm": 1.8781534433364868,
      "learning_rate": 5.136618227358546e-08,
      "loss": 0.4396,
      "step": 457610
    },
    {
      "epoch": 187.32,
      "grad_norm": 2.9630908966064453,
      "learning_rate": 5.1333158980027944e-08,
      "loss": 0.4334,
      "step": 457620
    },
    {
      "epoch": 187.32,
      "grad_norm": 1.8422939777374268,
      "learning_rate": 5.1300146187676584e-08,
      "loss": 0.4636,
      "step": 457630
    },
    {
      "epoch": 187.33,
      "grad_norm": 1.9092214107513428,
      "learning_rate": 5.1267143896683346e-08,
      "loss": 0.443,
      "step": 457640
    },
    {
      "epoch": 187.33,
      "grad_norm": 2.0634071826934814,
      "learning_rate": 5.1234152107198885e-08,
      "loss": 0.4371,
      "step": 457650
    },
    {
      "epoch": 187.34,
      "grad_norm": 2.012277126312256,
      "learning_rate": 5.120117081937464e-08,
      "loss": 0.4577,
      "step": 457660
    },
    {
      "epoch": 187.34,
      "grad_norm": 2.3907182216644287,
      "learning_rate": 5.116820003336127e-08,
      "loss": 0.4478,
      "step": 457670
    },
    {
      "epoch": 187.34,
      "grad_norm": 2.277904987335205,
      "learning_rate": 5.113523974931023e-08,
      "loss": 0.4465,
      "step": 457680
    },
    {
      "epoch": 187.35,
      "grad_norm": 1.9114729166030884,
      "learning_rate": 5.110228996737242e-08,
      "loss": 0.4331,
      "step": 457690
    },
    {
      "epoch": 187.35,
      "grad_norm": 2.1112499237060547,
      "learning_rate": 5.1069350687698524e-08,
      "loss": 0.425,
      "step": 457700
    },
    {
      "epoch": 187.36,
      "grad_norm": 2.0856943130493164,
      "learning_rate": 5.10364219104397e-08,
      "loss": 0.4479,
      "step": 457710
    },
    {
      "epoch": 187.36,
      "grad_norm": 2.106046199798584,
      "learning_rate": 5.100350363574663e-08,
      "loss": 0.4542,
      "step": 457720
    },
    {
      "epoch": 187.36,
      "grad_norm": 2.3903772830963135,
      "learning_rate": 5.097059586377022e-08,
      "loss": 0.4196,
      "step": 457730
    },
    {
      "epoch": 187.37,
      "grad_norm": 2.301774263381958,
      "learning_rate": 5.09376985946614e-08,
      "loss": 0.447,
      "step": 457740
    },
    {
      "epoch": 187.37,
      "grad_norm": 2.052128791809082,
      "learning_rate": 5.09048118285703e-08,
      "loss": 0.46,
      "step": 457750
    },
    {
      "epoch": 187.38,
      "grad_norm": 2.1417715549468994,
      "learning_rate": 5.0871935565648115e-08,
      "loss": 0.4513,
      "step": 457760
    },
    {
      "epoch": 187.38,
      "grad_norm": 1.8313770294189453,
      "learning_rate": 5.083906980604497e-08,
      "loss": 0.4392,
      "step": 457770
    },
    {
      "epoch": 187.38,
      "grad_norm": 2.2141430377960205,
      "learning_rate": 5.08062145499118e-08,
      "loss": 0.4329,
      "step": 457780
    },
    {
      "epoch": 187.39,
      "grad_norm": 4.346005916595459,
      "learning_rate": 5.077336979739926e-08,
      "loss": 0.4356,
      "step": 457790
    },
    {
      "epoch": 187.39,
      "grad_norm": 1.8006231784820557,
      "learning_rate": 5.0740535548657483e-08,
      "loss": 0.4346,
      "step": 457800
    },
    {
      "epoch": 187.4,
      "grad_norm": 2.2038097381591797,
      "learning_rate": 5.0707711803836885e-08,
      "loss": 0.4422,
      "step": 457810
    },
    {
      "epoch": 187.4,
      "grad_norm": 2.4962968826293945,
      "learning_rate": 5.067489856308811e-08,
      "loss": 0.4299,
      "step": 457820
    },
    {
      "epoch": 187.4,
      "grad_norm": 1.9788711071014404,
      "learning_rate": 5.0642095826560794e-08,
      "loss": 0.4532,
      "step": 457830
    },
    {
      "epoch": 187.41,
      "grad_norm": 2.116630792617798,
      "learning_rate": 5.0609303594406106e-08,
      "loss": 0.4342,
      "step": 457840
    },
    {
      "epoch": 187.41,
      "grad_norm": 2.251163959503174,
      "learning_rate": 5.057652186677341e-08,
      "loss": 0.451,
      "step": 457850
    },
    {
      "epoch": 187.42,
      "grad_norm": 2.114201307296753,
      "learning_rate": 5.054375064381362e-08,
      "loss": 0.4545,
      "step": 457860
    },
    {
      "epoch": 187.42,
      "grad_norm": 1.8426296710968018,
      "learning_rate": 5.051098992567611e-08,
      "loss": 0.4237,
      "step": 457870
    },
    {
      "epoch": 187.43,
      "grad_norm": 2.215555429458618,
      "learning_rate": 5.04782397125118e-08,
      "loss": 0.4435,
      "step": 457880
    },
    {
      "epoch": 187.43,
      "grad_norm": 1.9552276134490967,
      "learning_rate": 5.044550000447003e-08,
      "loss": 0.4192,
      "step": 457890
    },
    {
      "epoch": 187.43,
      "grad_norm": 2.2328526973724365,
      "learning_rate": 5.0412770801700964e-08,
      "loss": 0.4604,
      "step": 457900
    },
    {
      "epoch": 187.44,
      "grad_norm": 1.934743881225586,
      "learning_rate": 5.0380052104354726e-08,
      "loss": 0.4326,
      "step": 457910
    },
    {
      "epoch": 187.44,
      "grad_norm": 2.02697491645813,
      "learning_rate": 5.034734391258094e-08,
      "loss": 0.4501,
      "step": 457920
    },
    {
      "epoch": 187.45,
      "grad_norm": 1.6952890157699585,
      "learning_rate": 5.031464622652949e-08,
      "loss": 0.4467,
      "step": 457930
    },
    {
      "epoch": 187.45,
      "grad_norm": 2.0546114444732666,
      "learning_rate": 5.0281959046350255e-08,
      "loss": 0.4518,
      "step": 457940
    },
    {
      "epoch": 187.45,
      "grad_norm": 1.9804233312606812,
      "learning_rate": 5.0249282372192854e-08,
      "loss": 0.4298,
      "step": 457950
    },
    {
      "epoch": 187.46,
      "grad_norm": 1.9301037788391113,
      "learning_rate": 5.0216616204207165e-08,
      "loss": 0.4322,
      "step": 457960
    },
    {
      "epoch": 187.46,
      "grad_norm": 2.172677993774414,
      "learning_rate": 5.018396054254281e-08,
      "loss": 0.4369,
      "step": 457970
    },
    {
      "epoch": 187.47,
      "grad_norm": 2.071786880493164,
      "learning_rate": 5.015131538734915e-08,
      "loss": 0.4461,
      "step": 457980
    },
    {
      "epoch": 187.47,
      "grad_norm": 2.69382643699646,
      "learning_rate": 5.01186807387758e-08,
      "loss": 0.4439,
      "step": 457990
    },
    {
      "epoch": 187.47,
      "grad_norm": 6.467480182647705,
      "learning_rate": 5.0086056596972384e-08,
      "loss": 0.4504,
      "step": 458000
    },
    {
      "epoch": 187.48,
      "grad_norm": 2.223231315612793,
      "learning_rate": 5.0053442962088265e-08,
      "loss": 0.46,
      "step": 458010
    },
    {
      "epoch": 187.48,
      "grad_norm": 2.9827163219451904,
      "learning_rate": 5.002083983427279e-08,
      "loss": 0.4467,
      "step": 458020
    },
    {
      "epoch": 187.49,
      "grad_norm": 6.588624477386475,
      "learning_rate": 4.9988247213675595e-08,
      "loss": 0.4496,
      "step": 458030
    },
    {
      "epoch": 187.49,
      "grad_norm": 1.6595836877822876,
      "learning_rate": 4.995566510044551e-08,
      "loss": 0.4306,
      "step": 458040
    },
    {
      "epoch": 187.49,
      "grad_norm": 1.9821109771728516,
      "learning_rate": 4.992309349473216e-08,
      "loss": 0.4254,
      "step": 458050
    },
    {
      "epoch": 187.5,
      "grad_norm": 1.9734985828399658,
      "learning_rate": 4.9890532396684637e-08,
      "loss": 0.4576,
      "step": 458060
    },
    {
      "epoch": 187.5,
      "grad_norm": 1.9848414659500122,
      "learning_rate": 4.985798180645205e-08,
      "loss": 0.4329,
      "step": 458070
    },
    {
      "epoch": 187.51,
      "grad_norm": 2.2442054748535156,
      "learning_rate": 4.982544172418349e-08,
      "loss": 0.4595,
      "step": 458080
    },
    {
      "epoch": 187.51,
      "grad_norm": 2.24373197555542,
      "learning_rate": 4.979291215002805e-08,
      "loss": 0.429,
      "step": 458090
    },
    {
      "epoch": 187.52,
      "grad_norm": 2.3447928428649902,
      "learning_rate": 4.976039308413485e-08,
      "loss": 0.4457,
      "step": 458100
    },
    {
      "epoch": 187.52,
      "grad_norm": 2.13582181930542,
      "learning_rate": 4.9727884526652975e-08,
      "loss": 0.4514,
      "step": 458110
    },
    {
      "epoch": 187.52,
      "grad_norm": 2.387478828430176,
      "learning_rate": 4.969538647773075e-08,
      "loss": 0.4499,
      "step": 458120
    },
    {
      "epoch": 187.53,
      "grad_norm": 2.1751811504364014,
      "learning_rate": 4.966289893751779e-08,
      "loss": 0.4477,
      "step": 458130
    },
    {
      "epoch": 187.53,
      "grad_norm": 2.1283748149871826,
      "learning_rate": 4.963042190616216e-08,
      "loss": 0.4473,
      "step": 458140
    },
    {
      "epoch": 187.54,
      "grad_norm": 1.9040029048919678,
      "learning_rate": 4.9597955383813214e-08,
      "loss": 0.4719,
      "step": 458150
    },
    {
      "epoch": 187.54,
      "grad_norm": 2.0805115699768066,
      "learning_rate": 4.956549937061953e-08,
      "loss": 0.4322,
      "step": 458160
    },
    {
      "epoch": 187.54,
      "grad_norm": 2.261885404586792,
      "learning_rate": 4.953305386672969e-08,
      "loss": 0.4411,
      "step": 458170
    },
    {
      "epoch": 187.55,
      "grad_norm": 2.2303481101989746,
      "learning_rate": 4.9500618872292275e-08,
      "loss": 0.4396,
      "step": 458180
    },
    {
      "epoch": 187.55,
      "grad_norm": 2.2948803901672363,
      "learning_rate": 4.946819438745612e-08,
      "loss": 0.4363,
      "step": 458190
    },
    {
      "epoch": 187.56,
      "grad_norm": 2.237414836883545,
      "learning_rate": 4.943578041236955e-08,
      "loss": 0.4452,
      "step": 458200
    },
    {
      "epoch": 187.56,
      "grad_norm": 1.7917709350585938,
      "learning_rate": 4.9403376947181137e-08,
      "loss": 0.4314,
      "step": 458210
    },
    {
      "epoch": 187.56,
      "grad_norm": 2.3085691928863525,
      "learning_rate": 4.93709839920392e-08,
      "loss": 0.4343,
      "step": 458220
    },
    {
      "epoch": 187.57,
      "grad_norm": 2.04138445854187,
      "learning_rate": 4.9338601547092325e-08,
      "loss": 0.4436,
      "step": 458230
    },
    {
      "epoch": 187.57,
      "grad_norm": 1.6200860738754272,
      "learning_rate": 4.9306229612488566e-08,
      "loss": 0.4419,
      "step": 458240
    },
    {
      "epoch": 187.58,
      "grad_norm": 2.270292282104492,
      "learning_rate": 4.9273868188376765e-08,
      "loss": 0.446,
      "step": 458250
    },
    {
      "epoch": 187.58,
      "grad_norm": 2.075226306915283,
      "learning_rate": 4.924151727490446e-08,
      "loss": 0.4383,
      "step": 458260
    },
    {
      "epoch": 187.58,
      "grad_norm": 1.8273091316223145,
      "learning_rate": 4.9209176872220224e-08,
      "loss": 0.4259,
      "step": 458270
    },
    {
      "epoch": 187.59,
      "grad_norm": 1.864064335823059,
      "learning_rate": 4.917684698047212e-08,
      "loss": 0.456,
      "step": 458280
    },
    {
      "epoch": 187.59,
      "grad_norm": 2.183582067489624,
      "learning_rate": 4.914452759980847e-08,
      "loss": 0.4603,
      "step": 458290
    },
    {
      "epoch": 187.6,
      "grad_norm": 2.2847001552581787,
      "learning_rate": 4.9112218730376804e-08,
      "loss": 0.4453,
      "step": 458300
    },
    {
      "epoch": 187.6,
      "grad_norm": 2.1936378479003906,
      "learning_rate": 4.907992037232571e-08,
      "loss": 0.4461,
      "step": 458310
    },
    {
      "epoch": 187.61,
      "grad_norm": 1.9334253072738647,
      "learning_rate": 4.9047632525802715e-08,
      "loss": 0.4315,
      "step": 458320
    },
    {
      "epoch": 187.61,
      "grad_norm": 2.0620269775390625,
      "learning_rate": 4.9015355190955634e-08,
      "loss": 0.4256,
      "step": 458330
    },
    {
      "epoch": 187.61,
      "grad_norm": 1.8606077432632446,
      "learning_rate": 4.898308836793277e-08,
      "loss": 0.4407,
      "step": 458340
    },
    {
      "epoch": 187.62,
      "grad_norm": 1.777143955230713,
      "learning_rate": 4.8950832056881934e-08,
      "loss": 0.435,
      "step": 458350
    },
    {
      "epoch": 187.62,
      "grad_norm": 2.0663325786590576,
      "learning_rate": 4.891858625795065e-08,
      "loss": 0.4479,
      "step": 458360
    },
    {
      "epoch": 187.63,
      "grad_norm": 2.1753060817718506,
      "learning_rate": 4.888635097128647e-08,
      "loss": 0.4633,
      "step": 458370
    },
    {
      "epoch": 187.63,
      "grad_norm": 2.0121817588806152,
      "learning_rate": 4.8854126197037707e-08,
      "loss": 0.4396,
      "step": 458380
    },
    {
      "epoch": 187.63,
      "grad_norm": 1.9861640930175781,
      "learning_rate": 4.882191193535112e-08,
      "loss": 0.4432,
      "step": 458390
    },
    {
      "epoch": 187.64,
      "grad_norm": 2.102137327194214,
      "learning_rate": 4.8789708186374765e-08,
      "loss": 0.4593,
      "step": 458400
    },
    {
      "epoch": 187.64,
      "grad_norm": 1.9610377550125122,
      "learning_rate": 4.8757514950256183e-08,
      "loss": 0.4364,
      "step": 458410
    },
    {
      "epoch": 187.65,
      "grad_norm": 2.039658308029175,
      "learning_rate": 4.872533222714265e-08,
      "loss": 0.4404,
      "step": 458420
    },
    {
      "epoch": 187.65,
      "grad_norm": 1.8736172914505005,
      "learning_rate": 4.869316001718171e-08,
      "loss": 0.4511,
      "step": 458430
    },
    {
      "epoch": 187.65,
      "grad_norm": 1.9521517753601074,
      "learning_rate": 4.866099832052089e-08,
      "loss": 0.4329,
      "step": 458440
    },
    {
      "epoch": 187.66,
      "grad_norm": 1.8750278949737549,
      "learning_rate": 4.862884713730721e-08,
      "loss": 0.4132,
      "step": 458450
    },
    {
      "epoch": 187.66,
      "grad_norm": 1.8293380737304688,
      "learning_rate": 4.859670646768796e-08,
      "loss": 0.4399,
      "step": 458460
    },
    {
      "epoch": 187.67,
      "grad_norm": 1.8600990772247314,
      "learning_rate": 4.856457631181067e-08,
      "loss": 0.4666,
      "step": 458470
    },
    {
      "epoch": 187.67,
      "grad_norm": 2.048630475997925,
      "learning_rate": 4.853245666982235e-08,
      "loss": 0.4465,
      "step": 458480
    },
    {
      "epoch": 187.67,
      "grad_norm": 1.9827136993408203,
      "learning_rate": 4.850034754187004e-08,
      "loss": 0.4408,
      "step": 458490
    },
    {
      "epoch": 187.68,
      "grad_norm": 1.765856146812439,
      "learning_rate": 4.8468248928100996e-08,
      "loss": 0.4306,
      "step": 458500
    },
    {
      "epoch": 187.68,
      "grad_norm": 2.0212879180908203,
      "learning_rate": 4.843616082866198e-08,
      "loss": 0.4404,
      "step": 458510
    },
    {
      "epoch": 187.69,
      "grad_norm": 1.7933142185211182,
      "learning_rate": 4.840408324370054e-08,
      "loss": 0.4571,
      "step": 458520
    },
    {
      "epoch": 187.69,
      "grad_norm": 1.7848248481750488,
      "learning_rate": 4.8372016173362905e-08,
      "loss": 0.4401,
      "step": 458530
    },
    {
      "epoch": 187.7,
      "grad_norm": 2.056065082550049,
      "learning_rate": 4.8339959617796614e-08,
      "loss": 0.4403,
      "step": 458540
    },
    {
      "epoch": 187.7,
      "grad_norm": 1.827712893486023,
      "learning_rate": 4.8307913577148167e-08,
      "loss": 0.4281,
      "step": 458550
    },
    {
      "epoch": 187.7,
      "grad_norm": 2.1802878379821777,
      "learning_rate": 4.827587805156432e-08,
      "loss": 0.428,
      "step": 458560
    },
    {
      "epoch": 187.71,
      "grad_norm": 2.2178688049316406,
      "learning_rate": 4.824385304119183e-08,
      "loss": 0.4347,
      "step": 458570
    },
    {
      "epoch": 187.71,
      "grad_norm": 2.0788967609405518,
      "learning_rate": 4.821183854617772e-08,
      "loss": 0.4391,
      "step": 458580
    },
    {
      "epoch": 187.72,
      "grad_norm": 1.8873183727264404,
      "learning_rate": 4.8179834566668475e-08,
      "loss": 0.426,
      "step": 458590
    },
    {
      "epoch": 187.72,
      "grad_norm": 2.2573065757751465,
      "learning_rate": 4.8147841102810346e-08,
      "loss": 0.4367,
      "step": 458600
    },
    {
      "epoch": 187.72,
      "grad_norm": 2.0000386238098145,
      "learning_rate": 4.8115858154750345e-08,
      "loss": 0.4529,
      "step": 458610
    },
    {
      "epoch": 187.73,
      "grad_norm": 1.9282236099243164,
      "learning_rate": 4.808388572263497e-08,
      "loss": 0.4458,
      "step": 458620
    },
    {
      "epoch": 187.73,
      "grad_norm": 2.5053513050079346,
      "learning_rate": 4.8051923806610455e-08,
      "loss": 0.4434,
      "step": 458630
    },
    {
      "epoch": 187.74,
      "grad_norm": 2.079937696456909,
      "learning_rate": 4.801997240682304e-08,
      "loss": 0.455,
      "step": 458640
    },
    {
      "epoch": 187.74,
      "grad_norm": 1.918238639831543,
      "learning_rate": 4.798803152341975e-08,
      "loss": 0.4409,
      "step": 458650
    },
    {
      "epoch": 187.74,
      "grad_norm": 1.7254053354263306,
      "learning_rate": 4.7956101156546285e-08,
      "loss": 0.4418,
      "step": 458660
    },
    {
      "epoch": 187.75,
      "grad_norm": 2.104004383087158,
      "learning_rate": 4.792418130634915e-08,
      "loss": 0.452,
      "step": 458670
    },
    {
      "epoch": 187.75,
      "grad_norm": 1.7386151552200317,
      "learning_rate": 4.789227197297433e-08,
      "loss": 0.4333,
      "step": 458680
    },
    {
      "epoch": 187.76,
      "grad_norm": 2.2204713821411133,
      "learning_rate": 4.786037315656858e-08,
      "loss": 0.4406,
      "step": 458690
    },
    {
      "epoch": 187.76,
      "grad_norm": 2.4470911026000977,
      "learning_rate": 4.7828484857277604e-08,
      "loss": 0.4468,
      "step": 458700
    },
    {
      "epoch": 187.77,
      "grad_norm": 2.2708323001861572,
      "learning_rate": 4.7796607075247136e-08,
      "loss": 0.4355,
      "step": 458710
    },
    {
      "epoch": 187.77,
      "grad_norm": 2.2827417850494385,
      "learning_rate": 4.776473981062392e-08,
      "loss": 0.4591,
      "step": 458720
    },
    {
      "epoch": 187.77,
      "grad_norm": 1.8077229261398315,
      "learning_rate": 4.7732883063553425e-08,
      "loss": 0.4378,
      "step": 458730
    },
    {
      "epoch": 187.78,
      "grad_norm": 1.9473497867584229,
      "learning_rate": 4.770103683418187e-08,
      "loss": 0.435,
      "step": 458740
    },
    {
      "epoch": 187.78,
      "grad_norm": 3.6182491779327393,
      "learning_rate": 4.766920112265499e-08,
      "loss": 0.4324,
      "step": 458750
    },
    {
      "epoch": 187.79,
      "grad_norm": 1.9566227197647095,
      "learning_rate": 4.76373759291185e-08,
      "loss": 0.4269,
      "step": 458760
    },
    {
      "epoch": 187.79,
      "grad_norm": 2.1119847297668457,
      "learning_rate": 4.7605561253718626e-08,
      "loss": 0.4481,
      "step": 458770
    },
    {
      "epoch": 187.79,
      "grad_norm": 2.367344617843628,
      "learning_rate": 4.757375709660057e-08,
      "loss": 0.45,
      "step": 458780
    },
    {
      "epoch": 187.8,
      "grad_norm": 2.106966972351074,
      "learning_rate": 4.754196345791057e-08,
      "loss": 0.4618,
      "step": 458790
    },
    {
      "epoch": 187.8,
      "grad_norm": 2.36194109916687,
      "learning_rate": 4.7510180337793825e-08,
      "loss": 0.4456,
      "step": 458800
    },
    {
      "epoch": 187.81,
      "grad_norm": 1.9781347513198853,
      "learning_rate": 4.747840773639605e-08,
      "loss": 0.4623,
      "step": 458810
    },
    {
      "epoch": 187.81,
      "grad_norm": 1.7464524507522583,
      "learning_rate": 4.7446645653862694e-08,
      "loss": 0.4411,
      "step": 458820
    },
    {
      "epoch": 187.81,
      "grad_norm": 1.9770710468292236,
      "learning_rate": 4.741489409033975e-08,
      "loss": 0.4622,
      "step": 458830
    },
    {
      "epoch": 187.82,
      "grad_norm": 1.9663077592849731,
      "learning_rate": 4.738315304597213e-08,
      "loss": 0.4183,
      "step": 458840
    },
    {
      "epoch": 187.82,
      "grad_norm": 2.530324935913086,
      "learning_rate": 4.735142252090532e-08,
      "loss": 0.4354,
      "step": 458850
    },
    {
      "epoch": 187.83,
      "grad_norm": 1.831682801246643,
      "learning_rate": 4.731970251528475e-08,
      "loss": 0.4257,
      "step": 458860
    },
    {
      "epoch": 187.83,
      "grad_norm": 2.1199445724487305,
      "learning_rate": 4.728799302925589e-08,
      "loss": 0.4597,
      "step": 458870
    },
    {
      "epoch": 187.83,
      "grad_norm": 2.1662254333496094,
      "learning_rate": 4.725629406296368e-08,
      "loss": 0.4518,
      "step": 458880
    },
    {
      "epoch": 187.84,
      "grad_norm": 2.046722412109375,
      "learning_rate": 4.722460561655331e-08,
      "loss": 0.4552,
      "step": 458890
    },
    {
      "epoch": 187.84,
      "grad_norm": 2.1914119720458984,
      "learning_rate": 4.719292769017023e-08,
      "loss": 0.4356,
      "step": 458900
    },
    {
      "epoch": 187.85,
      "grad_norm": 1.6184566020965576,
      "learning_rate": 4.716126028395939e-08,
      "loss": 0.4643,
      "step": 458910
    },
    {
      "epoch": 187.85,
      "grad_norm": 1.9513574838638306,
      "learning_rate": 4.712960339806598e-08,
      "loss": 0.4458,
      "step": 458920
    },
    {
      "epoch": 187.86,
      "grad_norm": 2.2305526733398438,
      "learning_rate": 4.7097957032634934e-08,
      "loss": 0.4486,
      "step": 458930
    },
    {
      "epoch": 187.86,
      "grad_norm": 1.7964773178100586,
      "learning_rate": 4.7066321187811185e-08,
      "loss": 0.4513,
      "step": 458940
    },
    {
      "epoch": 187.86,
      "grad_norm": 2.21662974357605,
      "learning_rate": 4.703469586373968e-08,
      "loss": 0.4449,
      "step": 458950
    },
    {
      "epoch": 187.87,
      "grad_norm": 1.9374819993972778,
      "learning_rate": 4.700308106056561e-08,
      "loss": 0.4374,
      "step": 458960
    },
    {
      "epoch": 187.87,
      "grad_norm": 2.2884209156036377,
      "learning_rate": 4.697147677843313e-08,
      "loss": 0.4466,
      "step": 458970
    },
    {
      "epoch": 187.88,
      "grad_norm": 1.5935648679733276,
      "learning_rate": 4.693988301748743e-08,
      "loss": 0.4345,
      "step": 458980
    },
    {
      "epoch": 187.88,
      "grad_norm": 1.794024109840393,
      "learning_rate": 4.69082997778732e-08,
      "loss": 0.4433,
      "step": 458990
    },
    {
      "epoch": 187.88,
      "grad_norm": 2.437835693359375,
      "learning_rate": 4.6876727059735105e-08,
      "loss": 0.4335,
      "step": 459000
    },
    {
      "epoch": 187.89,
      "grad_norm": 2.390458345413208,
      "learning_rate": 4.684516486321756e-08,
      "loss": 0.4567,
      "step": 459010
    },
    {
      "epoch": 187.89,
      "grad_norm": 2.014312982559204,
      "learning_rate": 4.6813613188465516e-08,
      "loss": 0.4431,
      "step": 459020
    },
    {
      "epoch": 187.9,
      "grad_norm": 2.282059669494629,
      "learning_rate": 4.678207203562337e-08,
      "loss": 0.4383,
      "step": 459030
    },
    {
      "epoch": 187.9,
      "grad_norm": 2.0985801219940186,
      "learning_rate": 4.675054140483555e-08,
      "loss": 0.4311,
      "step": 459040
    },
    {
      "epoch": 187.9,
      "grad_norm": 1.6656250953674316,
      "learning_rate": 4.671902129624646e-08,
      "loss": 0.4435,
      "step": 459050
    },
    {
      "epoch": 187.91,
      "grad_norm": 2.314411163330078,
      "learning_rate": 4.6687511710000527e-08,
      "loss": 0.4367,
      "step": 459060
    },
    {
      "epoch": 187.91,
      "grad_norm": 1.830002784729004,
      "learning_rate": 4.665601264624216e-08,
      "loss": 0.4325,
      "step": 459070
    },
    {
      "epoch": 187.92,
      "grad_norm": 1.9083186388015747,
      "learning_rate": 4.662452410511552e-08,
      "loss": 0.466,
      "step": 459080
    },
    {
      "epoch": 187.92,
      "grad_norm": 2.4685311317443848,
      "learning_rate": 4.659304608676476e-08,
      "loss": 0.4423,
      "step": 459090
    },
    {
      "epoch": 187.92,
      "grad_norm": 1.885233998298645,
      "learning_rate": 4.656157859133482e-08,
      "loss": 0.446,
      "step": 459100
    },
    {
      "epoch": 187.93,
      "grad_norm": 1.881644606590271,
      "learning_rate": 4.6530121618968806e-08,
      "loss": 0.4563,
      "step": 459110
    },
    {
      "epoch": 187.93,
      "grad_norm": 2.6670615673065186,
      "learning_rate": 4.64986751698114e-08,
      "loss": 0.4382,
      "step": 459120
    },
    {
      "epoch": 187.94,
      "grad_norm": 2.2082278728485107,
      "learning_rate": 4.6467239244006755e-08,
      "loss": 0.4674,
      "step": 459130
    },
    {
      "epoch": 187.94,
      "grad_norm": 2.1389033794403076,
      "learning_rate": 4.643581384169825e-08,
      "loss": 0.4486,
      "step": 459140
    },
    {
      "epoch": 187.95,
      "grad_norm": 2.6004631519317627,
      "learning_rate": 4.6404398963030553e-08,
      "loss": 0.4634,
      "step": 459150
    },
    {
      "epoch": 187.95,
      "grad_norm": 2.320800542831421,
      "learning_rate": 4.637299460814731e-08,
      "loss": 0.4477,
      "step": 459160
    },
    {
      "epoch": 187.95,
      "grad_norm": 1.9122360944747925,
      "learning_rate": 4.6341600777192146e-08,
      "loss": 0.4523,
      "step": 459170
    },
    {
      "epoch": 187.96,
      "grad_norm": 2.2650856971740723,
      "learning_rate": 4.631021747030896e-08,
      "loss": 0.4422,
      "step": 459180
    },
    {
      "epoch": 187.96,
      "grad_norm": 2.321627140045166,
      "learning_rate": 4.627884468764191e-08,
      "loss": 0.4365,
      "step": 459190
    },
    {
      "epoch": 187.97,
      "grad_norm": 2.156198024749756,
      "learning_rate": 4.6247482429334106e-08,
      "loss": 0.4382,
      "step": 459200
    },
    {
      "epoch": 187.97,
      "grad_norm": 3.552997589111328,
      "learning_rate": 4.6216130695529706e-08,
      "loss": 0.4406,
      "step": 459210
    },
    {
      "epoch": 187.97,
      "grad_norm": 1.8090505599975586,
      "learning_rate": 4.618478948637209e-08,
      "loss": 0.4434,
      "step": 459220
    },
    {
      "epoch": 187.98,
      "grad_norm": 2.1290199756622314,
      "learning_rate": 4.6153458802004884e-08,
      "loss": 0.4623,
      "step": 459230
    },
    {
      "epoch": 187.98,
      "grad_norm": 2.4624717235565186,
      "learning_rate": 4.6122138642571726e-08,
      "loss": 0.4349,
      "step": 459240
    },
    {
      "epoch": 187.99,
      "grad_norm": 2.0582361221313477,
      "learning_rate": 4.6090829008215995e-08,
      "loss": 0.4207,
      "step": 459250
    },
    {
      "epoch": 187.99,
      "grad_norm": 2.0072600841522217,
      "learning_rate": 4.60595298990808e-08,
      "loss": 0.4537,
      "step": 459260
    },
    {
      "epoch": 187.99,
      "grad_norm": 2.478506326675415,
      "learning_rate": 4.6028241315310305e-08,
      "loss": 0.4466,
      "step": 459270
    },
    {
      "epoch": 188.0,
      "grad_norm": 2.120976448059082,
      "learning_rate": 4.5996963257046835e-08,
      "loss": 0.4489,
      "step": 459280
    },
    {
      "epoch": 188.0,
      "eval_loss": 0.44328808784484863,
      "eval_runtime": 43.3292,
      "eval_samples_per_second": 79.6,
      "eval_steps_per_second": 9.97,
      "step": 459284
    },
    {
      "epoch": 188.0,
      "grad_norm": 1.8585647344589233,
      "learning_rate": 4.5965695724434293e-08,
      "loss": 0.4414,
      "step": 459290
    },
    {
      "epoch": 188.01,
      "grad_norm": 2.1745574474334717,
      "learning_rate": 4.593443871761578e-08,
      "loss": 0.4386,
      "step": 459300
    },
    {
      "epoch": 188.01,
      "grad_norm": 2.1153194904327393,
      "learning_rate": 4.590319223673443e-08,
      "loss": 0.4443,
      "step": 459310
    },
    {
      "epoch": 188.01,
      "grad_norm": 2.042726755142212,
      "learning_rate": 4.58719562819336e-08,
      "loss": 0.45,
      "step": 459320
    },
    {
      "epoch": 188.02,
      "grad_norm": 2.1017580032348633,
      "learning_rate": 4.584073085335616e-08,
      "loss": 0.4735,
      "step": 459330
    },
    {
      "epoch": 188.02,
      "grad_norm": 2.0797126293182373,
      "learning_rate": 4.580951595114547e-08,
      "loss": 0.4313,
      "step": 459340
    },
    {
      "epoch": 188.03,
      "grad_norm": 1.887120008468628,
      "learning_rate": 4.577831157544388e-08,
      "loss": 0.4401,
      "step": 459350
    },
    {
      "epoch": 188.03,
      "grad_norm": 1.6238983869552612,
      "learning_rate": 4.574711772639475e-08,
      "loss": 0.4439,
      "step": 459360
    },
    {
      "epoch": 188.04,
      "grad_norm": 2.288351058959961,
      "learning_rate": 4.57159344041412e-08,
      "loss": 0.4718,
      "step": 459370
    },
    {
      "epoch": 188.04,
      "grad_norm": 1.945987582206726,
      "learning_rate": 4.568476160882557e-08,
      "loss": 0.4426,
      "step": 459380
    },
    {
      "epoch": 188.04,
      "grad_norm": 2.2905991077423096,
      "learning_rate": 4.565359934059097e-08,
      "loss": 0.4438,
      "step": 459390
    },
    {
      "epoch": 188.05,
      "grad_norm": 1.859388828277588,
      "learning_rate": 4.562244759957999e-08,
      "loss": 0.4397,
      "step": 459400
    },
    {
      "epoch": 188.05,
      "grad_norm": 2.1948482990264893,
      "learning_rate": 4.559130638593575e-08,
      "loss": 0.4509,
      "step": 459410
    },
    {
      "epoch": 188.06,
      "grad_norm": 1.7225736379623413,
      "learning_rate": 4.556017569980032e-08,
      "loss": 0.4475,
      "step": 459420
    },
    {
      "epoch": 188.06,
      "grad_norm": 1.7001748085021973,
      "learning_rate": 4.552905554131656e-08,
      "loss": 0.4258,
      "step": 459430
    },
    {
      "epoch": 188.06,
      "grad_norm": 2.0920488834381104,
      "learning_rate": 4.549794591062706e-08,
      "loss": 0.4235,
      "step": 459440
    },
    {
      "epoch": 188.07,
      "grad_norm": 2.055577278137207,
      "learning_rate": 4.546684680787416e-08,
      "loss": 0.4311,
      "step": 459450
    },
    {
      "epoch": 188.07,
      "grad_norm": 2.1737611293792725,
      "learning_rate": 4.5435758233200445e-08,
      "loss": 0.4328,
      "step": 459460
    },
    {
      "epoch": 188.08,
      "grad_norm": 2.076143980026245,
      "learning_rate": 4.540468018674826e-08,
      "loss": 0.4568,
      "step": 459470
    },
    {
      "epoch": 188.08,
      "grad_norm": 1.7295646667480469,
      "learning_rate": 4.537361266866045e-08,
      "loss": 0.4473,
      "step": 459480
    },
    {
      "epoch": 188.08,
      "grad_norm": 2.0106401443481445,
      "learning_rate": 4.534255567907857e-08,
      "loss": 0.4529,
      "step": 459490
    },
    {
      "epoch": 188.09,
      "grad_norm": 2.34489369392395,
      "learning_rate": 4.5311509218145475e-08,
      "loss": 0.4431,
      "step": 459500
    },
    {
      "epoch": 188.09,
      "grad_norm": 2.2260310649871826,
      "learning_rate": 4.528047328600298e-08,
      "loss": 0.4308,
      "step": 459510
    },
    {
      "epoch": 188.1,
      "grad_norm": 1.9649492502212524,
      "learning_rate": 4.524944788279342e-08,
      "loss": 0.4396,
      "step": 459520
    },
    {
      "epoch": 188.1,
      "grad_norm": 1.9002885818481445,
      "learning_rate": 4.5218433008658864e-08,
      "loss": 0.4526,
      "step": 459530
    },
    {
      "epoch": 188.1,
      "grad_norm": 2.4437806606292725,
      "learning_rate": 4.518742866374166e-08,
      "loss": 0.4308,
      "step": 459540
    },
    {
      "epoch": 188.11,
      "grad_norm": 2.018095016479492,
      "learning_rate": 4.51564348481836e-08,
      "loss": 0.4262,
      "step": 459550
    },
    {
      "epoch": 188.11,
      "grad_norm": 1.8956632614135742,
      "learning_rate": 4.5125451562126775e-08,
      "loss": 0.4535,
      "step": 459560
    },
    {
      "epoch": 188.12,
      "grad_norm": 2.0029866695404053,
      "learning_rate": 4.5094478805712735e-08,
      "loss": 0.4407,
      "step": 459570
    },
    {
      "epoch": 188.12,
      "grad_norm": 1.6720640659332275,
      "learning_rate": 4.5063516579083806e-08,
      "loss": 0.437,
      "step": 459580
    },
    {
      "epoch": 188.13,
      "grad_norm": 2.060311794281006,
      "learning_rate": 4.503256488238155e-08,
      "loss": 0.4491,
      "step": 459590
    },
    {
      "epoch": 188.13,
      "grad_norm": 1.7631111145019531,
      "learning_rate": 4.500162371574778e-08,
      "loss": 0.4385,
      "step": 459600
    },
    {
      "epoch": 188.13,
      "grad_norm": 2.119713544845581,
      "learning_rate": 4.497069307932456e-08,
      "loss": 0.451,
      "step": 459610
    },
    {
      "epoch": 188.14,
      "grad_norm": 2.1982264518737793,
      "learning_rate": 4.49397729732532e-08,
      "loss": 0.4403,
      "step": 459620
    },
    {
      "epoch": 188.14,
      "grad_norm": 1.920556664466858,
      "learning_rate": 4.49088633976755e-08,
      "loss": 0.4319,
      "step": 459630
    },
    {
      "epoch": 188.15,
      "grad_norm": 1.7715201377868652,
      "learning_rate": 4.4877964352733016e-08,
      "loss": 0.4668,
      "step": 459640
    },
    {
      "epoch": 188.15,
      "grad_norm": 2.0013654232025146,
      "learning_rate": 4.484707583856705e-08,
      "loss": 0.4314,
      "step": 459650
    },
    {
      "epoch": 188.15,
      "grad_norm": 1.8962291479110718,
      "learning_rate": 4.4816197855319665e-08,
      "loss": 0.4511,
      "step": 459660
    },
    {
      "epoch": 188.16,
      "grad_norm": 2.173539161682129,
      "learning_rate": 4.47853304031319e-08,
      "loss": 0.44,
      "step": 459670
    },
    {
      "epoch": 188.16,
      "grad_norm": 1.7014408111572266,
      "learning_rate": 4.475447348214556e-08,
      "loss": 0.4375,
      "step": 459680
    },
    {
      "epoch": 188.17,
      "grad_norm": 2.3210864067077637,
      "learning_rate": 4.472362709250143e-08,
      "loss": 0.4399,
      "step": 459690
    },
    {
      "epoch": 188.17,
      "grad_norm": 1.9550113677978516,
      "learning_rate": 4.469279123434106e-08,
      "loss": 0.4444,
      "step": 459700
    },
    {
      "epoch": 188.17,
      "grad_norm": 2.114994525909424,
      "learning_rate": 4.4661965907805475e-08,
      "loss": 0.4365,
      "step": 459710
    },
    {
      "epoch": 188.18,
      "grad_norm": 2.1286916732788086,
      "learning_rate": 4.4631151113036503e-08,
      "loss": 0.4286,
      "step": 459720
    },
    {
      "epoch": 188.18,
      "grad_norm": 2.0595784187316895,
      "learning_rate": 4.460034685017464e-08,
      "loss": 0.4494,
      "step": 459730
    },
    {
      "epoch": 188.19,
      "grad_norm": 2.4512908458709717,
      "learning_rate": 4.456955311936145e-08,
      "loss": 0.4453,
      "step": 459740
    },
    {
      "epoch": 188.19,
      "grad_norm": 2.1474742889404297,
      "learning_rate": 4.4538769920737706e-08,
      "loss": 0.4295,
      "step": 459750
    },
    {
      "epoch": 188.19,
      "grad_norm": 1.8697961568832397,
      "learning_rate": 4.4507997254444435e-08,
      "loss": 0.4306,
      "step": 459760
    },
    {
      "epoch": 188.2,
      "grad_norm": 2.7226808071136475,
      "learning_rate": 4.447723512062294e-08,
      "loss": 0.4269,
      "step": 459770
    },
    {
      "epoch": 188.2,
      "grad_norm": 2.302724838256836,
      "learning_rate": 4.4446483519413724e-08,
      "loss": 0.4384,
      "step": 459780
    },
    {
      "epoch": 188.21,
      "grad_norm": 3.967996835708618,
      "learning_rate": 4.441574245095782e-08,
      "loss": 0.4434,
      "step": 459790
    },
    {
      "epoch": 188.21,
      "grad_norm": 2.26162052154541,
      "learning_rate": 4.4385011915396265e-08,
      "loss": 0.4363,
      "step": 459800
    },
    {
      "epoch": 188.22,
      "grad_norm": 2.1812808513641357,
      "learning_rate": 4.4354291912869316e-08,
      "loss": 0.449,
      "step": 459810
    },
    {
      "epoch": 188.22,
      "grad_norm": 2.209652900695801,
      "learning_rate": 4.4323582443518256e-08,
      "loss": 0.448,
      "step": 459820
    },
    {
      "epoch": 188.22,
      "grad_norm": 2.7305171489715576,
      "learning_rate": 4.4292883507483343e-08,
      "loss": 0.4436,
      "step": 459830
    },
    {
      "epoch": 188.23,
      "grad_norm": 2.3528573513031006,
      "learning_rate": 4.426219510490535e-08,
      "loss": 0.4451,
      "step": 459840
    },
    {
      "epoch": 188.23,
      "grad_norm": 2.482452392578125,
      "learning_rate": 4.423151723592505e-08,
      "loss": 0.4609,
      "step": 459850
    },
    {
      "epoch": 188.24,
      "grad_norm": 4.880699157714844,
      "learning_rate": 4.420084990068269e-08,
      "loss": 0.44,
      "step": 459860
    },
    {
      "epoch": 188.24,
      "grad_norm": 1.9187384843826294,
      "learning_rate": 4.41701930993188e-08,
      "loss": 0.4409,
      "step": 459870
    },
    {
      "epoch": 188.24,
      "grad_norm": 2.0435781478881836,
      "learning_rate": 4.4139546831973864e-08,
      "loss": 0.4487,
      "step": 459880
    },
    {
      "epoch": 188.25,
      "grad_norm": 1.8799954652786255,
      "learning_rate": 4.4108911098788417e-08,
      "loss": 0.4246,
      "step": 459890
    },
    {
      "epoch": 188.25,
      "grad_norm": 2.1305599212646484,
      "learning_rate": 4.407828589990245e-08,
      "loss": 0.4246,
      "step": 459900
    },
    {
      "epoch": 188.26,
      "grad_norm": 2.2059130668640137,
      "learning_rate": 4.404767123545672e-08,
      "loss": 0.4552,
      "step": 459910
    },
    {
      "epoch": 188.26,
      "grad_norm": 2.1100783348083496,
      "learning_rate": 4.4017067105590976e-08,
      "loss": 0.4347,
      "step": 459920
    },
    {
      "epoch": 188.26,
      "grad_norm": 1.9941906929016113,
      "learning_rate": 4.398647351044572e-08,
      "loss": 0.449,
      "step": 459930
    },
    {
      "epoch": 188.27,
      "grad_norm": 2.260267496109009,
      "learning_rate": 4.3955890450160945e-08,
      "loss": 0.4413,
      "step": 459940
    },
    {
      "epoch": 188.27,
      "grad_norm": 1.9076311588287354,
      "learning_rate": 4.392531792487691e-08,
      "loss": 0.4302,
      "step": 459950
    },
    {
      "epoch": 188.28,
      "grad_norm": 1.7361798286437988,
      "learning_rate": 4.389475593473361e-08,
      "loss": 0.4476,
      "step": 459960
    },
    {
      "epoch": 188.28,
      "grad_norm": 1.7618408203125,
      "learning_rate": 4.386420447987102e-08,
      "loss": 0.4597,
      "step": 459970
    },
    {
      "epoch": 188.28,
      "grad_norm": 2.425575017929077,
      "learning_rate": 4.383366356042889e-08,
      "loss": 0.4522,
      "step": 459980
    },
    {
      "epoch": 188.29,
      "grad_norm": 1.8010934591293335,
      "learning_rate": 4.3803133176547724e-08,
      "loss": 0.4443,
      "step": 459990
    },
    {
      "epoch": 188.29,
      "grad_norm": 1.9084700345993042,
      "learning_rate": 4.3772613328366473e-08,
      "loss": 0.4267,
      "step": 460000
    },
    {
      "epoch": 188.3,
      "grad_norm": 2.157623052597046,
      "learning_rate": 4.374210401602565e-08,
      "loss": 0.4535,
      "step": 460010
    },
    {
      "epoch": 188.3,
      "grad_norm": 2.227293014526367,
      "learning_rate": 4.371160523966499e-08,
      "loss": 0.4512,
      "step": 460020
    },
    {
      "epoch": 188.31,
      "grad_norm": 2.1740589141845703,
      "learning_rate": 4.368111699942344e-08,
      "loss": 0.4504,
      "step": 460030
    },
    {
      "epoch": 188.31,
      "grad_norm": 1.9861105680465698,
      "learning_rate": 4.365063929544152e-08,
      "loss": 0.4285,
      "step": 460040
    },
    {
      "epoch": 188.31,
      "grad_norm": 2.303190231323242,
      "learning_rate": 4.3620172127858695e-08,
      "loss": 0.4687,
      "step": 460050
    },
    {
      "epoch": 188.32,
      "grad_norm": 2.138676881790161,
      "learning_rate": 4.358971549681444e-08,
      "loss": 0.439,
      "step": 460060
    },
    {
      "epoch": 188.32,
      "grad_norm": 2.170292615890503,
      "learning_rate": 4.355926940244796e-08,
      "loss": 0.4342,
      "step": 460070
    },
    {
      "epoch": 188.33,
      "grad_norm": 2.1383962631225586,
      "learning_rate": 4.352883384489926e-08,
      "loss": 0.4228,
      "step": 460080
    },
    {
      "epoch": 188.33,
      "grad_norm": 2.076808452606201,
      "learning_rate": 4.349840882430728e-08,
      "loss": 0.4389,
      "step": 460090
    },
    {
      "epoch": 188.33,
      "grad_norm": 3.7914505004882812,
      "learning_rate": 4.3467994340811754e-08,
      "loss": 0.4547,
      "step": 460100
    },
    {
      "epoch": 188.34,
      "grad_norm": 1.950577735900879,
      "learning_rate": 4.3437590394551636e-08,
      "loss": 0.4465,
      "step": 460110
    },
    {
      "epoch": 188.34,
      "grad_norm": 2.3938755989074707,
      "learning_rate": 4.340719698566666e-08,
      "loss": 0.4698,
      "step": 460120
    },
    {
      "epoch": 188.35,
      "grad_norm": 1.8176299333572388,
      "learning_rate": 4.337681411429577e-08,
      "loss": 0.4409,
      "step": 460130
    },
    {
      "epoch": 188.35,
      "grad_norm": 1.9867136478424072,
      "learning_rate": 4.3346441780578185e-08,
      "loss": 0.4444,
      "step": 460140
    },
    {
      "epoch": 188.35,
      "grad_norm": 2.1647839546203613,
      "learning_rate": 4.331607998465285e-08,
      "loss": 0.4559,
      "step": 460150
    },
    {
      "epoch": 188.36,
      "grad_norm": 2.4028480052948,
      "learning_rate": 4.328572872665924e-08,
      "loss": 0.4736,
      "step": 460160
    },
    {
      "epoch": 188.36,
      "grad_norm": 2.1435964107513428,
      "learning_rate": 4.325538800673605e-08,
      "loss": 0.4341,
      "step": 460170
    },
    {
      "epoch": 188.37,
      "grad_norm": 1.8924611806869507,
      "learning_rate": 4.322505782502249e-08,
      "loss": 0.4428,
      "step": 460180
    },
    {
      "epoch": 188.37,
      "grad_norm": 2.040598154067993,
      "learning_rate": 4.3194738181657505e-08,
      "loss": 0.4293,
      "step": 460190
    },
    {
      "epoch": 188.37,
      "grad_norm": 2.0126492977142334,
      "learning_rate": 4.3164429076779794e-08,
      "loss": 0.446,
      "step": 460200
    },
    {
      "epoch": 188.38,
      "grad_norm": 1.9521044492721558,
      "learning_rate": 4.313413051052804e-08,
      "loss": 0.4457,
      "step": 460210
    },
    {
      "epoch": 188.38,
      "grad_norm": 1.8792556524276733,
      "learning_rate": 4.3103842483041465e-08,
      "loss": 0.439,
      "step": 460220
    },
    {
      "epoch": 188.39,
      "grad_norm": 2.144888162612915,
      "learning_rate": 4.3073564994458746e-08,
      "loss": 0.4511,
      "step": 460230
    },
    {
      "epoch": 188.39,
      "grad_norm": 1.78317391872406,
      "learning_rate": 4.304329804491833e-08,
      "loss": 0.4654,
      "step": 460240
    },
    {
      "epoch": 188.4,
      "grad_norm": 1.8749396800994873,
      "learning_rate": 4.301304163455916e-08,
      "loss": 0.4375,
      "step": 460250
    },
    {
      "epoch": 188.4,
      "grad_norm": 2.161904811859131,
      "learning_rate": 4.29827957635194e-08,
      "loss": 0.4321,
      "step": 460260
    },
    {
      "epoch": 188.4,
      "grad_norm": 1.7034069299697876,
      "learning_rate": 4.295256043193828e-08,
      "loss": 0.4514,
      "step": 460270
    },
    {
      "epoch": 188.41,
      "grad_norm": 2.3954079151153564,
      "learning_rate": 4.292233563995395e-08,
      "loss": 0.4248,
      "step": 460280
    },
    {
      "epoch": 188.41,
      "grad_norm": 2.040088415145874,
      "learning_rate": 4.289212138770459e-08,
      "loss": 0.4535,
      "step": 460290
    },
    {
      "epoch": 188.42,
      "grad_norm": 2.473018169403076,
      "learning_rate": 4.2861917675328636e-08,
      "loss": 0.4383,
      "step": 460300
    },
    {
      "epoch": 188.42,
      "grad_norm": 2.142218589782715,
      "learning_rate": 4.283172450296503e-08,
      "loss": 0.4357,
      "step": 460310
    },
    {
      "epoch": 188.42,
      "grad_norm": 1.8830300569534302,
      "learning_rate": 4.280154187075143e-08,
      "loss": 0.4355,
      "step": 460320
    },
    {
      "epoch": 188.43,
      "grad_norm": 2.1149022579193115,
      "learning_rate": 4.277136977882652e-08,
      "loss": 0.4427,
      "step": 460330
    },
    {
      "epoch": 188.43,
      "grad_norm": 2.413391590118408,
      "learning_rate": 4.2741208227328217e-08,
      "loss": 0.4361,
      "step": 460340
    },
    {
      "epoch": 188.44,
      "grad_norm": 2.273021697998047,
      "learning_rate": 4.271105721639495e-08,
      "loss": 0.4371,
      "step": 460350
    },
    {
      "epoch": 188.44,
      "grad_norm": 2.2111165523529053,
      "learning_rate": 4.268091674616463e-08,
      "loss": 0.4529,
      "step": 460360
    },
    {
      "epoch": 188.44,
      "grad_norm": 1.7245839834213257,
      "learning_rate": 4.265078681677542e-08,
      "loss": 0.4441,
      "step": 460370
    },
    {
      "epoch": 188.45,
      "grad_norm": 1.9496163129806519,
      "learning_rate": 4.2620667428365504e-08,
      "loss": 0.4482,
      "step": 460380
    },
    {
      "epoch": 188.45,
      "grad_norm": 1.8299601078033447,
      "learning_rate": 4.259055858107253e-08,
      "loss": 0.4455,
      "step": 460390
    },
    {
      "epoch": 188.46,
      "grad_norm": 1.8817431926727295,
      "learning_rate": 4.2560460275034655e-08,
      "loss": 0.4348,
      "step": 460400
    },
    {
      "epoch": 188.46,
      "grad_norm": 2.2227230072021484,
      "learning_rate": 4.2530372510389546e-08,
      "loss": 0.4314,
      "step": 460410
    },
    {
      "epoch": 188.47,
      "grad_norm": 2.0480306148529053,
      "learning_rate": 4.250029528727536e-08,
      "loss": 0.4516,
      "step": 460420
    },
    {
      "epoch": 188.47,
      "grad_norm": 1.6913400888442993,
      "learning_rate": 4.24702286058295e-08,
      "loss": 0.4461,
      "step": 460430
    },
    {
      "epoch": 188.47,
      "grad_norm": 1.61143159866333,
      "learning_rate": 4.244017246619012e-08,
      "loss": 0.4575,
      "step": 460440
    },
    {
      "epoch": 188.48,
      "grad_norm": 2.024794101715088,
      "learning_rate": 4.241012686849437e-08,
      "loss": 0.4314,
      "step": 460450
    },
    {
      "epoch": 188.48,
      "grad_norm": 1.9625917673110962,
      "learning_rate": 4.2380091812880406e-08,
      "loss": 0.4445,
      "step": 460460
    },
    {
      "epoch": 188.49,
      "grad_norm": 2.3762741088867188,
      "learning_rate": 4.235006729948536e-08,
      "loss": 0.4403,
      "step": 460470
    },
    {
      "epoch": 188.49,
      "grad_norm": 2.364255666732788,
      "learning_rate": 4.2320053328447155e-08,
      "loss": 0.4507,
      "step": 460480
    },
    {
      "epoch": 188.49,
      "grad_norm": 2.122331142425537,
      "learning_rate": 4.229004989990291e-08,
      "loss": 0.433,
      "step": 460490
    },
    {
      "epoch": 188.5,
      "grad_norm": 2.3423328399658203,
      "learning_rate": 4.226005701399053e-08,
      "loss": 0.4626,
      "step": 460500
    },
    {
      "epoch": 188.5,
      "grad_norm": 2.140244483947754,
      "learning_rate": 4.223007467084715e-08,
      "loss": 0.4543,
      "step": 460510
    },
    {
      "epoch": 188.51,
      "grad_norm": 1.906706690788269,
      "learning_rate": 4.2200102870609904e-08,
      "loss": 0.4376,
      "step": 460520
    },
    {
      "epoch": 188.51,
      "grad_norm": 1.8200773000717163,
      "learning_rate": 4.2170141613416445e-08,
      "loss": 0.4422,
      "step": 460530
    },
    {
      "epoch": 188.51,
      "grad_norm": 2.0837550163269043,
      "learning_rate": 4.214019089940363e-08,
      "loss": 0.4592,
      "step": 460540
    },
    {
      "epoch": 188.52,
      "grad_norm": 1.8993605375289917,
      "learning_rate": 4.211025072870912e-08,
      "loss": 0.4392,
      "step": 460550
    },
    {
      "epoch": 188.52,
      "grad_norm": 2.155376672744751,
      "learning_rate": 4.2080321101469786e-08,
      "loss": 0.4685,
      "step": 460560
    },
    {
      "epoch": 188.53,
      "grad_norm": 2.1199119091033936,
      "learning_rate": 4.205040201782301e-08,
      "loss": 0.4221,
      "step": 460570
    },
    {
      "epoch": 188.53,
      "grad_norm": 1.9447928667068481,
      "learning_rate": 4.202049347790566e-08,
      "loss": 0.4593,
      "step": 460580
    },
    {
      "epoch": 188.53,
      "grad_norm": 1.7513819932937622,
      "learning_rate": 4.1990595481854354e-08,
      "loss": 0.465,
      "step": 460590
    },
    {
      "epoch": 188.54,
      "grad_norm": 2.139253616333008,
      "learning_rate": 4.196070802980648e-08,
      "loss": 0.4456,
      "step": 460600
    },
    {
      "epoch": 188.54,
      "grad_norm": 2.4310617446899414,
      "learning_rate": 4.193083112189891e-08,
      "loss": 0.448,
      "step": 460610
    },
    {
      "epoch": 188.55,
      "grad_norm": 2.061997652053833,
      "learning_rate": 4.190096475826851e-08,
      "loss": 0.4418,
      "step": 460620
    },
    {
      "epoch": 188.55,
      "grad_norm": 2.298769235610962,
      "learning_rate": 4.187110893905188e-08,
      "loss": 0.4378,
      "step": 460630
    },
    {
      "epoch": 188.56,
      "grad_norm": 2.299232006072998,
      "learning_rate": 4.1841263664386166e-08,
      "loss": 0.4485,
      "step": 460640
    },
    {
      "epoch": 188.56,
      "grad_norm": 1.9560565948486328,
      "learning_rate": 4.181142893440797e-08,
      "loss": 0.4482,
      "step": 460650
    },
    {
      "epoch": 188.56,
      "grad_norm": 2.1606340408325195,
      "learning_rate": 4.1781604749253644e-08,
      "loss": 0.4466,
      "step": 460660
    },
    {
      "epoch": 188.57,
      "grad_norm": 2.176842451095581,
      "learning_rate": 4.1751791109060056e-08,
      "loss": 0.4405,
      "step": 460670
    },
    {
      "epoch": 188.57,
      "grad_norm": 2.1373462677001953,
      "learning_rate": 4.172198801396381e-08,
      "loss": 0.4374,
      "step": 460680
    },
    {
      "epoch": 188.58,
      "grad_norm": 1.7246861457824707,
      "learning_rate": 4.169219546410153e-08,
      "loss": 0.4401,
      "step": 460690
    },
    {
      "epoch": 188.58,
      "grad_norm": 1.9102368354797363,
      "learning_rate": 4.166241345960929e-08,
      "loss": 0.4291,
      "step": 460700
    },
    {
      "epoch": 188.58,
      "grad_norm": 2.00758695602417,
      "learning_rate": 4.163264200062423e-08,
      "loss": 0.4287,
      "step": 460710
    },
    {
      "epoch": 188.59,
      "grad_norm": 2.407837152481079,
      "learning_rate": 4.160288108728165e-08,
      "loss": 0.4553,
      "step": 460720
    },
    {
      "epoch": 188.59,
      "grad_norm": 2.4421629905700684,
      "learning_rate": 4.1573130719718945e-08,
      "loss": 0.4469,
      "step": 460730
    },
    {
      "epoch": 188.6,
      "grad_norm": 2.4211440086364746,
      "learning_rate": 4.154339089807168e-08,
      "loss": 0.4416,
      "step": 460740
    },
    {
      "epoch": 188.6,
      "grad_norm": 2.0988240242004395,
      "learning_rate": 4.151366162247621e-08,
      "loss": 0.4644,
      "step": 460750
    },
    {
      "epoch": 188.6,
      "grad_norm": 2.140965700149536,
      "learning_rate": 4.1483942893069146e-08,
      "loss": 0.4409,
      "step": 460760
    },
    {
      "epoch": 188.61,
      "grad_norm": 2.0131452083587646,
      "learning_rate": 4.145423470998631e-08,
      "loss": 0.4377,
      "step": 460770
    },
    {
      "epoch": 188.61,
      "grad_norm": 2.173936367034912,
      "learning_rate": 4.142453707336353e-08,
      "loss": 0.444,
      "step": 460780
    },
    {
      "epoch": 188.62,
      "grad_norm": 2.0400514602661133,
      "learning_rate": 4.1394849983337426e-08,
      "loss": 0.4612,
      "step": 460790
    },
    {
      "epoch": 188.62,
      "grad_norm": 1.937798261642456,
      "learning_rate": 4.1365173440043555e-08,
      "loss": 0.4446,
      "step": 460800
    },
    {
      "epoch": 188.62,
      "grad_norm": 2.2153260707855225,
      "learning_rate": 4.133550744361801e-08,
      "loss": 0.4594,
      "step": 460810
    },
    {
      "epoch": 188.63,
      "grad_norm": 3.5293288230895996,
      "learning_rate": 4.1305851994196877e-08,
      "loss": 0.4685,
      "step": 460820
    },
    {
      "epoch": 188.63,
      "grad_norm": 2.1245217323303223,
      "learning_rate": 4.127620709191547e-08,
      "loss": 0.4462,
      "step": 460830
    },
    {
      "epoch": 188.64,
      "grad_norm": 1.7383896112442017,
      "learning_rate": 4.124657273691014e-08,
      "loss": 0.444,
      "step": 460840
    },
    {
      "epoch": 188.64,
      "grad_norm": 2.0880279541015625,
      "learning_rate": 4.121694892931671e-08,
      "loss": 0.4553,
      "step": 460850
    },
    {
      "epoch": 188.65,
      "grad_norm": 1.9281845092773438,
      "learning_rate": 4.118733566927023e-08,
      "loss": 0.4363,
      "step": 460860
    },
    {
      "epoch": 188.65,
      "grad_norm": 1.906343936920166,
      "learning_rate": 4.1157732956906796e-08,
      "loss": 0.4427,
      "step": 460870
    },
    {
      "epoch": 188.65,
      "grad_norm": 1.8448848724365234,
      "learning_rate": 4.112814079236197e-08,
      "loss": 0.4339,
      "step": 460880
    },
    {
      "epoch": 188.66,
      "grad_norm": 2.430025339126587,
      "learning_rate": 4.109855917577131e-08,
      "loss": 0.4429,
      "step": 460890
    },
    {
      "epoch": 188.66,
      "grad_norm": 1.9272713661193848,
      "learning_rate": 4.106898810727014e-08,
      "loss": 0.4632,
      "step": 460900
    },
    {
      "epoch": 188.67,
      "grad_norm": 1.9859788417816162,
      "learning_rate": 4.1039427586994284e-08,
      "loss": 0.4413,
      "step": 460910
    },
    {
      "epoch": 188.67,
      "grad_norm": 2.137576103210449,
      "learning_rate": 4.1009877615078796e-08,
      "loss": 0.447,
      "step": 460920
    },
    {
      "epoch": 188.67,
      "grad_norm": 2.469837188720703,
      "learning_rate": 4.0980338191658974e-08,
      "loss": 0.4417,
      "step": 460930
    },
    {
      "epoch": 188.68,
      "grad_norm": 2.2223174571990967,
      "learning_rate": 4.095080931687065e-08,
      "loss": 0.4317,
      "step": 460940
    },
    {
      "epoch": 188.68,
      "grad_norm": 1.6540247201919556,
      "learning_rate": 4.092129099084861e-08,
      "loss": 0.4553,
      "step": 460950
    },
    {
      "epoch": 188.69,
      "grad_norm": 2.056622266769409,
      "learning_rate": 4.0891783213728436e-08,
      "loss": 0.4558,
      "step": 460960
    },
    {
      "epoch": 188.69,
      "grad_norm": 2.539860248565674,
      "learning_rate": 4.08622859856449e-08,
      "loss": 0.4537,
      "step": 460970
    },
    {
      "epoch": 188.69,
      "grad_norm": 2.340320587158203,
      "learning_rate": 4.083279930673358e-08,
      "loss": 0.4447,
      "step": 460980
    },
    {
      "epoch": 188.7,
      "grad_norm": 2.009251356124878,
      "learning_rate": 4.080332317712926e-08,
      "loss": 0.4454,
      "step": 460990
    },
    {
      "epoch": 188.7,
      "grad_norm": 2.0698888301849365,
      "learning_rate": 4.077385759696673e-08,
      "loss": 0.4325,
      "step": 461000
    },
    {
      "epoch": 188.71,
      "grad_norm": 2.0137407779693604,
      "learning_rate": 4.074440256638155e-08,
      "loss": 0.4443,
      "step": 461010
    },
    {
      "epoch": 188.71,
      "grad_norm": 1.9820860624313354,
      "learning_rate": 4.0714958085508265e-08,
      "loss": 0.443,
      "step": 461020
    },
    {
      "epoch": 188.71,
      "grad_norm": 2.1797235012054443,
      "learning_rate": 4.068552415448165e-08,
      "loss": 0.4577,
      "step": 461030
    },
    {
      "epoch": 188.72,
      "grad_norm": 1.756517767906189,
      "learning_rate": 4.065610077343701e-08,
      "loss": 0.4602,
      "step": 461040
    },
    {
      "epoch": 188.72,
      "grad_norm": 1.7861216068267822,
      "learning_rate": 4.062668794250863e-08,
      "loss": 0.4214,
      "step": 461050
    },
    {
      "epoch": 188.73,
      "grad_norm": 1.7337100505828857,
      "learning_rate": 4.059728566183154e-08,
      "loss": 0.4586,
      "step": 461060
    },
    {
      "epoch": 188.73,
      "grad_norm": 1.830955982208252,
      "learning_rate": 4.056789393154028e-08,
      "loss": 0.4498,
      "step": 461070
    },
    {
      "epoch": 188.74,
      "grad_norm": 2.4234962463378906,
      "learning_rate": 4.053851275176963e-08,
      "loss": 0.4463,
      "step": 461080
    },
    {
      "epoch": 188.74,
      "grad_norm": 1.8259719610214233,
      "learning_rate": 4.0509142122654124e-08,
      "loss": 0.4458,
      "step": 461090
    },
    {
      "epoch": 188.74,
      "grad_norm": 2.297560214996338,
      "learning_rate": 4.0479782044328284e-08,
      "loss": 0.4222,
      "step": 461100
    },
    {
      "epoch": 188.75,
      "grad_norm": 2.0911896228790283,
      "learning_rate": 4.0450432516926645e-08,
      "loss": 0.4516,
      "step": 461110
    },
    {
      "epoch": 188.75,
      "grad_norm": 2.1286003589630127,
      "learning_rate": 4.042109354058347e-08,
      "loss": 0.4522,
      "step": 461120
    },
    {
      "epoch": 188.76,
      "grad_norm": 2.2611334323883057,
      "learning_rate": 4.039176511543355e-08,
      "loss": 0.4483,
      "step": 461130
    },
    {
      "epoch": 188.76,
      "grad_norm": 1.967610239982605,
      "learning_rate": 4.036244724161062e-08,
      "loss": 0.4474,
      "step": 461140
    },
    {
      "epoch": 188.76,
      "grad_norm": 2.8631093502044678,
      "learning_rate": 4.033313991924974e-08,
      "loss": 0.4304,
      "step": 461150
    },
    {
      "epoch": 188.77,
      "grad_norm": 1.9710111618041992,
      "learning_rate": 4.03038431484844e-08,
      "loss": 0.4209,
      "step": 461160
    },
    {
      "epoch": 188.77,
      "grad_norm": 1.8550854921340942,
      "learning_rate": 4.0274556929449384e-08,
      "loss": 0.4256,
      "step": 461170
    },
    {
      "epoch": 188.78,
      "grad_norm": 1.8477675914764404,
      "learning_rate": 4.024528126227869e-08,
      "loss": 0.4277,
      "step": 461180
    },
    {
      "epoch": 188.78,
      "grad_norm": 2.193072557449341,
      "learning_rate": 4.0216016147106084e-08,
      "loss": 0.442,
      "step": 461190
    },
    {
      "epoch": 188.78,
      "grad_norm": 2.238405227661133,
      "learning_rate": 4.0186761584066335e-08,
      "loss": 0.4337,
      "step": 461200
    },
    {
      "epoch": 188.79,
      "grad_norm": 2.5776121616363525,
      "learning_rate": 4.015751757329268e-08,
      "loss": 0.4425,
      "step": 461210
    },
    {
      "epoch": 188.79,
      "grad_norm": 1.77360999584198,
      "learning_rate": 4.0128284114919645e-08,
      "loss": 0.4326,
      "step": 461220
    },
    {
      "epoch": 188.8,
      "grad_norm": 1.8549989461898804,
      "learning_rate": 4.009906120908072e-08,
      "loss": 0.4413,
      "step": 461230
    },
    {
      "epoch": 188.8,
      "grad_norm": 3.1979641914367676,
      "learning_rate": 4.006984885591017e-08,
      "loss": 0.4646,
      "step": 461240
    },
    {
      "epoch": 188.8,
      "grad_norm": 1.891726016998291,
      "learning_rate": 4.0040647055541477e-08,
      "loss": 0.4294,
      "step": 461250
    },
    {
      "epoch": 188.81,
      "grad_norm": 2.263016700744629,
      "learning_rate": 4.0011455808108656e-08,
      "loss": 0.4421,
      "step": 461260
    },
    {
      "epoch": 188.81,
      "grad_norm": 2.132131338119507,
      "learning_rate": 3.998227511374519e-08,
      "loss": 0.4467,
      "step": 461270
    },
    {
      "epoch": 188.82,
      "grad_norm": 2.106952667236328,
      "learning_rate": 3.995310497258509e-08,
      "loss": 0.4424,
      "step": 461280
    },
    {
      "epoch": 188.82,
      "grad_norm": 2.942243814468384,
      "learning_rate": 3.992394538476183e-08,
      "loss": 0.4384,
      "step": 461290
    },
    {
      "epoch": 188.83,
      "grad_norm": 1.8208271265029907,
      "learning_rate": 3.989479635040866e-08,
      "loss": 0.4599,
      "step": 461300
    },
    {
      "epoch": 188.83,
      "grad_norm": 1.7936216592788696,
      "learning_rate": 3.986565786965957e-08,
      "loss": 0.449,
      "step": 461310
    },
    {
      "epoch": 188.83,
      "grad_norm": 2.388723611831665,
      "learning_rate": 3.9836529942647787e-08,
      "loss": 0.463,
      "step": 461320
    },
    {
      "epoch": 188.84,
      "grad_norm": 1.7552911043167114,
      "learning_rate": 3.9807412569506804e-08,
      "loss": 0.4466,
      "step": 461330
    },
    {
      "epoch": 188.84,
      "grad_norm": 1.9757182598114014,
      "learning_rate": 3.9778305750370105e-08,
      "loss": 0.4519,
      "step": 461340
    },
    {
      "epoch": 188.85,
      "grad_norm": 2.2385447025299072,
      "learning_rate": 3.9749209485370404e-08,
      "loss": 0.4375,
      "step": 461350
    },
    {
      "epoch": 188.85,
      "grad_norm": 2.022149085998535,
      "learning_rate": 3.9720123774641955e-08,
      "loss": 0.4625,
      "step": 461360
    },
    {
      "epoch": 188.85,
      "grad_norm": 1.6571736335754395,
      "learning_rate": 3.969104861831747e-08,
      "loss": 0.4364,
      "step": 461370
    },
    {
      "epoch": 188.86,
      "grad_norm": 2.2489829063415527,
      "learning_rate": 3.9661984016530446e-08,
      "loss": 0.4304,
      "step": 461380
    },
    {
      "epoch": 188.86,
      "grad_norm": 2.0528833866119385,
      "learning_rate": 3.963292996941357e-08,
      "loss": 0.4536,
      "step": 461390
    },
    {
      "epoch": 188.87,
      "grad_norm": 1.6810814142227173,
      "learning_rate": 3.960388647710009e-08,
      "loss": 0.446,
      "step": 461400
    },
    {
      "epoch": 188.87,
      "grad_norm": 1.8828365802764893,
      "learning_rate": 3.957485353972322e-08,
      "loss": 0.447,
      "step": 461410
    },
    {
      "epoch": 188.87,
      "grad_norm": 2.177182912826538,
      "learning_rate": 3.954583115741592e-08,
      "loss": 0.4366,
      "step": 461420
    },
    {
      "epoch": 188.88,
      "grad_norm": 2.271191120147705,
      "learning_rate": 3.951681933031117e-08,
      "loss": 0.4537,
      "step": 461430
    },
    {
      "epoch": 188.88,
      "grad_norm": 1.9371403455734253,
      "learning_rate": 3.948781805854142e-08,
      "loss": 0.4433,
      "step": 461440
    },
    {
      "epoch": 188.89,
      "grad_norm": 2.277268648147583,
      "learning_rate": 3.945882734224014e-08,
      "loss": 0.4535,
      "step": 461450
    },
    {
      "epoch": 188.89,
      "grad_norm": 2.299935817718506,
      "learning_rate": 3.9429847181539784e-08,
      "loss": 0.4315,
      "step": 461460
    },
    {
      "epoch": 188.89,
      "grad_norm": 1.936933159828186,
      "learning_rate": 3.9400877576573064e-08,
      "loss": 0.4496,
      "step": 461470
    },
    {
      "epoch": 188.9,
      "grad_norm": 1.8465955257415771,
      "learning_rate": 3.937191852747294e-08,
      "loss": 0.4338,
      "step": 461480
    },
    {
      "epoch": 188.9,
      "grad_norm": 1.9418703317642212,
      "learning_rate": 3.9342970034371857e-08,
      "loss": 0.4516,
      "step": 461490
    },
    {
      "epoch": 188.91,
      "grad_norm": 2.1238696575164795,
      "learning_rate": 3.931403209740252e-08,
      "loss": 0.4261,
      "step": 461500
    },
    {
      "epoch": 188.91,
      "grad_norm": 2.3206849098205566,
      "learning_rate": 3.928510471669739e-08,
      "loss": 0.4372,
      "step": 461510
    },
    {
      "epoch": 188.92,
      "grad_norm": 1.781968593597412,
      "learning_rate": 3.9256187892389154e-08,
      "loss": 0.4445,
      "step": 461520
    },
    {
      "epoch": 188.92,
      "grad_norm": 2.0352275371551514,
      "learning_rate": 3.9227281624610014e-08,
      "loss": 0.438,
      "step": 461530
    },
    {
      "epoch": 188.92,
      "grad_norm": 2.0282726287841797,
      "learning_rate": 3.91983859134924e-08,
      "loss": 0.4469,
      "step": 461540
    },
    {
      "epoch": 188.93,
      "grad_norm": 1.8994262218475342,
      "learning_rate": 3.9169500759169034e-08,
      "loss": 0.4393,
      "step": 461550
    },
    {
      "epoch": 188.93,
      "grad_norm": 2.3763952255249023,
      "learning_rate": 3.914062616177209e-08,
      "loss": 0.4304,
      "step": 461560
    },
    {
      "epoch": 188.94,
      "grad_norm": 2.0746872425079346,
      "learning_rate": 3.911176212143402e-08,
      "loss": 0.4365,
      "step": 461570
    },
    {
      "epoch": 188.94,
      "grad_norm": 1.9042243957519531,
      "learning_rate": 3.908290863828649e-08,
      "loss": 0.4393,
      "step": 461580
    },
    {
      "epoch": 188.94,
      "grad_norm": 1.9164730310440063,
      "learning_rate": 3.90540657124622e-08,
      "loss": 0.4167,
      "step": 461590
    },
    {
      "epoch": 188.95,
      "grad_norm": 2.350764274597168,
      "learning_rate": 3.9025233344092826e-08,
      "loss": 0.4386,
      "step": 461600
    },
    {
      "epoch": 188.95,
      "grad_norm": 2.124431848526001,
      "learning_rate": 3.8996411533310806e-08,
      "loss": 0.4328,
      "step": 461610
    },
    {
      "epoch": 188.96,
      "grad_norm": 2.2932286262512207,
      "learning_rate": 3.896760028024781e-08,
      "loss": 0.4281,
      "step": 461620
    },
    {
      "epoch": 188.96,
      "grad_norm": 2.0221385955810547,
      "learning_rate": 3.893879958503628e-08,
      "loss": 0.4362,
      "step": 461630
    },
    {
      "epoch": 188.96,
      "grad_norm": 2.0057458877563477,
      "learning_rate": 3.8910009447807885e-08,
      "loss": 0.4406,
      "step": 461640
    },
    {
      "epoch": 188.97,
      "grad_norm": 2.2588038444519043,
      "learning_rate": 3.8881229868694544e-08,
      "loss": 0.4538,
      "step": 461650
    },
    {
      "epoch": 188.97,
      "grad_norm": 2.172015428543091,
      "learning_rate": 3.885246084782794e-08,
      "loss": 0.4615,
      "step": 461660
    },
    {
      "epoch": 188.98,
      "grad_norm": 1.967675805091858,
      "learning_rate": 3.882370238534023e-08,
      "loss": 0.4498,
      "step": 461670
    },
    {
      "epoch": 188.98,
      "grad_norm": 2.022573471069336,
      "learning_rate": 3.8794954481362587e-08,
      "loss": 0.4606,
      "step": 461680
    },
    {
      "epoch": 188.98,
      "grad_norm": 2.272550344467163,
      "learning_rate": 3.876621713602745e-08,
      "loss": 0.4493,
      "step": 461690
    },
    {
      "epoch": 188.99,
      "grad_norm": 1.9142688512802124,
      "learning_rate": 3.8737490349465686e-08,
      "loss": 0.4476,
      "step": 461700
    },
    {
      "epoch": 188.99,
      "grad_norm": 2.118692398071289,
      "learning_rate": 3.8708774121809505e-08,
      "loss": 0.4501,
      "step": 461710
    },
    {
      "epoch": 189.0,
      "grad_norm": 2.265516519546509,
      "learning_rate": 3.868006845319004e-08,
      "loss": 0.4485,
      "step": 461720
    },
    {
      "epoch": 189.0,
      "eval_loss": 0.44351306557655334,
      "eval_runtime": 43.4508,
      "eval_samples_per_second": 79.377,
      "eval_steps_per_second": 9.942,
      "step": 461727
    },
    {
      "epoch": 189.0,
      "grad_norm": 1.8356919288635254,
      "learning_rate": 3.8651373343738956e-08,
      "loss": 0.4316,
      "step": 461730
    },
    {
      "epoch": 189.01,
      "grad_norm": 1.8054072856903076,
      "learning_rate": 3.862268879358792e-08,
      "loss": 0.4586,
      "step": 461740
    },
    {
      "epoch": 189.01,
      "grad_norm": 1.7749017477035522,
      "learning_rate": 3.859401480286782e-08,
      "loss": 0.4431,
      "step": 461750
    },
    {
      "epoch": 189.01,
      "grad_norm": 1.925776720046997,
      "learning_rate": 3.8565351371710586e-08,
      "loss": 0.4398,
      "step": 461760
    },
    {
      "epoch": 189.02,
      "grad_norm": 2.0068960189819336,
      "learning_rate": 3.8536698500246825e-08,
      "loss": 0.4292,
      "step": 461770
    },
    {
      "epoch": 189.02,
      "grad_norm": 2.1367979049682617,
      "learning_rate": 3.8508056188608483e-08,
      "loss": 0.435,
      "step": 461780
    },
    {
      "epoch": 189.03,
      "grad_norm": 1.8707135915756226,
      "learning_rate": 3.8479424436926434e-08,
      "loss": 0.4303,
      "step": 461790
    },
    {
      "epoch": 189.03,
      "grad_norm": 2.397535800933838,
      "learning_rate": 3.845080324533182e-08,
      "loss": 0.4279,
      "step": 461800
    },
    {
      "epoch": 189.03,
      "grad_norm": 1.5126616954803467,
      "learning_rate": 3.842219261395579e-08,
      "loss": 0.4324,
      "step": 461810
    },
    {
      "epoch": 189.04,
      "grad_norm": 2.1022539138793945,
      "learning_rate": 3.83935925429295e-08,
      "loss": 0.4472,
      "step": 461820
    },
    {
      "epoch": 189.04,
      "grad_norm": 2.0188486576080322,
      "learning_rate": 3.8365003032383814e-08,
      "loss": 0.4393,
      "step": 461830
    },
    {
      "epoch": 189.05,
      "grad_norm": 2.0256941318511963,
      "learning_rate": 3.833642408244989e-08,
      "loss": 0.4315,
      "step": 461840
    },
    {
      "epoch": 189.05,
      "grad_norm": 2.1978225708007812,
      "learning_rate": 3.8307855693258356e-08,
      "loss": 0.4322,
      "step": 461850
    },
    {
      "epoch": 189.05,
      "grad_norm": 1.8440362215042114,
      "learning_rate": 3.8279297864940356e-08,
      "loss": 0.4273,
      "step": 461860
    },
    {
      "epoch": 189.06,
      "grad_norm": 2.1228225231170654,
      "learning_rate": 3.82507505976265e-08,
      "loss": 0.4407,
      "step": 461870
    },
    {
      "epoch": 189.06,
      "grad_norm": 1.8395923376083374,
      "learning_rate": 3.822221389144795e-08,
      "loss": 0.4351,
      "step": 461880
    },
    {
      "epoch": 189.07,
      "grad_norm": 2.3553223609924316,
      "learning_rate": 3.81936877465348e-08,
      "loss": 0.4219,
      "step": 461890
    },
    {
      "epoch": 189.07,
      "grad_norm": 2.4591431617736816,
      "learning_rate": 3.81651721630182e-08,
      "loss": 0.4482,
      "step": 461900
    },
    {
      "epoch": 189.07,
      "grad_norm": 1.9960036277770996,
      "learning_rate": 3.813666714102851e-08,
      "loss": 0.4363,
      "step": 461910
    },
    {
      "epoch": 189.08,
      "grad_norm": 2.0114660263061523,
      "learning_rate": 3.8108172680696364e-08,
      "loss": 0.4135,
      "step": 461920
    },
    {
      "epoch": 189.08,
      "grad_norm": 1.9078540802001953,
      "learning_rate": 3.807968878215264e-08,
      "loss": 0.4187,
      "step": 461930
    },
    {
      "epoch": 189.09,
      "grad_norm": 2.6746480464935303,
      "learning_rate": 3.805121544552719e-08,
      "loss": 0.4406,
      "step": 461940
    },
    {
      "epoch": 189.09,
      "grad_norm": 1.999087929725647,
      "learning_rate": 3.802275267095114e-08,
      "loss": 0.4252,
      "step": 461950
    },
    {
      "epoch": 189.1,
      "grad_norm": 1.9766820669174194,
      "learning_rate": 3.799430045855435e-08,
      "loss": 0.4582,
      "step": 461960
    },
    {
      "epoch": 189.1,
      "grad_norm": 2.195500612258911,
      "learning_rate": 3.796585880846718e-08,
      "loss": 0.4322,
      "step": 461970
    },
    {
      "epoch": 189.1,
      "grad_norm": 2.4430103302001953,
      "learning_rate": 3.793742772082026e-08,
      "loss": 0.4336,
      "step": 461980
    },
    {
      "epoch": 189.11,
      "grad_norm": 2.5243184566497803,
      "learning_rate": 3.790900719574368e-08,
      "loss": 0.4275,
      "step": 461990
    },
    {
      "epoch": 189.11,
      "grad_norm": 1.9461474418640137,
      "learning_rate": 3.7880597233367563e-08,
      "loss": 0.4456,
      "step": 462000
    },
    {
      "epoch": 189.12,
      "grad_norm": 1.9460328817367554,
      "learning_rate": 3.7852197833822e-08,
      "loss": 0.4386,
      "step": 462010
    },
    {
      "epoch": 189.12,
      "grad_norm": 1.9068714380264282,
      "learning_rate": 3.782380899723736e-08,
      "loss": 0.4319,
      "step": 462020
    },
    {
      "epoch": 189.12,
      "grad_norm": 2.274883508682251,
      "learning_rate": 3.779543072374323e-08,
      "loss": 0.4249,
      "step": 462030
    },
    {
      "epoch": 189.13,
      "grad_norm": 1.9437192678451538,
      "learning_rate": 3.776706301346997e-08,
      "loss": 0.4247,
      "step": 462040
    },
    {
      "epoch": 189.13,
      "grad_norm": 2.0200750827789307,
      "learning_rate": 3.773870586654742e-08,
      "loss": 0.4372,
      "step": 462050
    },
    {
      "epoch": 189.14,
      "grad_norm": 2.0915579795837402,
      "learning_rate": 3.7710359283105175e-08,
      "loss": 0.4359,
      "step": 462060
    },
    {
      "epoch": 189.14,
      "grad_norm": 1.97407066822052,
      "learning_rate": 3.768202326327385e-08,
      "loss": 0.4601,
      "step": 462070
    },
    {
      "epoch": 189.14,
      "grad_norm": 4.297144412994385,
      "learning_rate": 3.765369780718251e-08,
      "loss": 0.4548,
      "step": 462080
    },
    {
      "epoch": 189.15,
      "grad_norm": 1.9560588598251343,
      "learning_rate": 3.762538291496127e-08,
      "loss": 0.4442,
      "step": 462090
    },
    {
      "epoch": 189.15,
      "grad_norm": 1.9342975616455078,
      "learning_rate": 3.75970785867397e-08,
      "loss": 0.4555,
      "step": 462100
    },
    {
      "epoch": 189.16,
      "grad_norm": 1.829522728919983,
      "learning_rate": 3.756878482264738e-08,
      "loss": 0.4306,
      "step": 462110
    },
    {
      "epoch": 189.16,
      "grad_norm": 2.635573148727417,
      "learning_rate": 3.754050162281417e-08,
      "loss": 0.4424,
      "step": 462120
    },
    {
      "epoch": 189.16,
      "grad_norm": 1.9288853406906128,
      "learning_rate": 3.751222898736965e-08,
      "loss": 0.4405,
      "step": 462130
    },
    {
      "epoch": 189.17,
      "grad_norm": 3.214250326156616,
      "learning_rate": 3.748396691644314e-08,
      "loss": 0.4546,
      "step": 462140
    },
    {
      "epoch": 189.17,
      "grad_norm": 2.000783920288086,
      "learning_rate": 3.7455715410164217e-08,
      "loss": 0.4759,
      "step": 462150
    },
    {
      "epoch": 189.18,
      "grad_norm": 1.9395123720169067,
      "learning_rate": 3.742747446866222e-08,
      "loss": 0.4497,
      "step": 462160
    },
    {
      "epoch": 189.18,
      "grad_norm": 1.9035390615463257,
      "learning_rate": 3.739924409206646e-08,
      "loss": 0.4637,
      "step": 462170
    },
    {
      "epoch": 189.19,
      "grad_norm": 2.2369155883789062,
      "learning_rate": 3.737102428050627e-08,
      "loss": 0.4588,
      "step": 462180
    },
    {
      "epoch": 189.19,
      "grad_norm": 1.9235002994537354,
      "learning_rate": 3.734281503411096e-08,
      "loss": 0.4659,
      "step": 462190
    },
    {
      "epoch": 189.19,
      "grad_norm": 2.4553062915802,
      "learning_rate": 3.731461635300987e-08,
      "loss": 0.4357,
      "step": 462200
    },
    {
      "epoch": 189.2,
      "grad_norm": 2.274568796157837,
      "learning_rate": 3.728642823733205e-08,
      "loss": 0.4516,
      "step": 462210
    },
    {
      "epoch": 189.2,
      "grad_norm": 2.1204028129577637,
      "learning_rate": 3.7258250687206574e-08,
      "loss": 0.4379,
      "step": 462220
    },
    {
      "epoch": 189.21,
      "grad_norm": 2.3570263385772705,
      "learning_rate": 3.723008370276276e-08,
      "loss": 0.4298,
      "step": 462230
    },
    {
      "epoch": 189.21,
      "grad_norm": 1.8572452068328857,
      "learning_rate": 3.720192728412941e-08,
      "loss": 0.4256,
      "step": 462240
    },
    {
      "epoch": 189.21,
      "grad_norm": 2.7266054153442383,
      "learning_rate": 3.717378143143534e-08,
      "loss": 0.4372,
      "step": 462250
    },
    {
      "epoch": 189.22,
      "grad_norm": 2.1302738189697266,
      "learning_rate": 3.714564614481011e-08,
      "loss": 0.4633,
      "step": 462260
    },
    {
      "epoch": 189.22,
      "grad_norm": 2.0475027561187744,
      "learning_rate": 3.711752142438202e-08,
      "loss": 0.443,
      "step": 462270
    },
    {
      "epoch": 189.23,
      "grad_norm": 2.3000266551971436,
      "learning_rate": 3.708940727027987e-08,
      "loss": 0.4651,
      "step": 462280
    },
    {
      "epoch": 189.23,
      "grad_norm": 2.4195516109466553,
      "learning_rate": 3.706130368263298e-08,
      "loss": 0.4426,
      "step": 462290
    },
    {
      "epoch": 189.23,
      "grad_norm": 2.215552806854248,
      "learning_rate": 3.703321066156964e-08,
      "loss": 0.4672,
      "step": 462300
    },
    {
      "epoch": 189.24,
      "grad_norm": 1.8729876279830933,
      "learning_rate": 3.7005128207218644e-08,
      "loss": 0.4457,
      "step": 462310
    },
    {
      "epoch": 189.24,
      "grad_norm": 1.8425066471099854,
      "learning_rate": 3.6977056319708806e-08,
      "loss": 0.4527,
      "step": 462320
    },
    {
      "epoch": 189.25,
      "grad_norm": 2.379093647003174,
      "learning_rate": 3.69489949991684e-08,
      "loss": 0.4636,
      "step": 462330
    },
    {
      "epoch": 189.25,
      "grad_norm": 2.520768165588379,
      "learning_rate": 3.692094424572623e-08,
      "loss": 0.4373,
      "step": 462340
    },
    {
      "epoch": 189.26,
      "grad_norm": 1.6596862077713013,
      "learning_rate": 3.6892904059510583e-08,
      "loss": 0.4308,
      "step": 462350
    },
    {
      "epoch": 189.26,
      "grad_norm": 2.2151131629943848,
      "learning_rate": 3.686487444065026e-08,
      "loss": 0.4464,
      "step": 462360
    },
    {
      "epoch": 189.26,
      "grad_norm": 1.9801870584487915,
      "learning_rate": 3.6836855389273285e-08,
      "loss": 0.4237,
      "step": 462370
    },
    {
      "epoch": 189.27,
      "grad_norm": 2.212111234664917,
      "learning_rate": 3.68088469055082e-08,
      "loss": 0.4435,
      "step": 462380
    },
    {
      "epoch": 189.27,
      "grad_norm": 2.0038018226623535,
      "learning_rate": 3.6780848989483284e-08,
      "loss": 0.4536,
      "step": 462390
    },
    {
      "epoch": 189.28,
      "grad_norm": 1.9293936491012573,
      "learning_rate": 3.6752861641326824e-08,
      "loss": 0.4322,
      "step": 462400
    },
    {
      "epoch": 189.28,
      "grad_norm": 1.4974614381790161,
      "learning_rate": 3.6724884861166846e-08,
      "loss": 0.4568,
      "step": 462410
    },
    {
      "epoch": 189.28,
      "grad_norm": 2.3099544048309326,
      "learning_rate": 3.669691864913163e-08,
      "loss": 0.4496,
      "step": 462420
    },
    {
      "epoch": 189.29,
      "grad_norm": 1.9602792263031006,
      "learning_rate": 3.666896300534919e-08,
      "loss": 0.4546,
      "step": 462430
    },
    {
      "epoch": 189.29,
      "grad_norm": 2.3035497665405273,
      "learning_rate": 3.664101792994782e-08,
      "loss": 0.418,
      "step": 462440
    },
    {
      "epoch": 189.3,
      "grad_norm": 2.268040180206299,
      "learning_rate": 3.661308342305554e-08,
      "loss": 0.4308,
      "step": 462450
    },
    {
      "epoch": 189.3,
      "grad_norm": 1.867861270904541,
      "learning_rate": 3.658515948480011e-08,
      "loss": 0.4412,
      "step": 462460
    },
    {
      "epoch": 189.3,
      "grad_norm": 2.0939853191375732,
      "learning_rate": 3.6557246115309547e-08,
      "loss": 0.4427,
      "step": 462470
    },
    {
      "epoch": 189.31,
      "grad_norm": 1.9695792198181152,
      "learning_rate": 3.652934331471137e-08,
      "loss": 0.4517,
      "step": 462480
    },
    {
      "epoch": 189.31,
      "grad_norm": 1.9508410692214966,
      "learning_rate": 3.650145108313385e-08,
      "loss": 0.4446,
      "step": 462490
    },
    {
      "epoch": 189.32,
      "grad_norm": 1.676628828048706,
      "learning_rate": 3.647356942070448e-08,
      "loss": 0.4349,
      "step": 462500
    },
    {
      "epoch": 189.32,
      "grad_norm": 2.7428622245788574,
      "learning_rate": 3.6445698327551304e-08,
      "loss": 0.4494,
      "step": 462510
    },
    {
      "epoch": 189.32,
      "grad_norm": 1.6501401662826538,
      "learning_rate": 3.6417837803801815e-08,
      "loss": 0.4439,
      "step": 462520
    },
    {
      "epoch": 189.33,
      "grad_norm": 2.848750591278076,
      "learning_rate": 3.6389987849583505e-08,
      "loss": 0.4493,
      "step": 462530
    },
    {
      "epoch": 189.33,
      "grad_norm": 1.7309027910232544,
      "learning_rate": 3.636214846502415e-08,
      "loss": 0.439,
      "step": 462540
    },
    {
      "epoch": 189.34,
      "grad_norm": 2.0879933834075928,
      "learning_rate": 3.633431965025125e-08,
      "loss": 0.4346,
      "step": 462550
    },
    {
      "epoch": 189.34,
      "grad_norm": 2.0290966033935547,
      "learning_rate": 3.63065014053923e-08,
      "loss": 0.4384,
      "step": 462560
    },
    {
      "epoch": 189.35,
      "grad_norm": 2.0457029342651367,
      "learning_rate": 3.627869373057455e-08,
      "loss": 0.4232,
      "step": 462570
    },
    {
      "epoch": 189.35,
      "grad_norm": 1.9858317375183105,
      "learning_rate": 3.6250896625925496e-08,
      "loss": 0.4439,
      "step": 462580
    },
    {
      "epoch": 189.35,
      "grad_norm": 2.1179075241088867,
      "learning_rate": 3.622311009157238e-08,
      "loss": 0.4456,
      "step": 462590
    },
    {
      "epoch": 189.36,
      "grad_norm": 1.8868870735168457,
      "learning_rate": 3.6195334127642976e-08,
      "loss": 0.4294,
      "step": 462600
    },
    {
      "epoch": 189.36,
      "grad_norm": 9.724771499633789,
      "learning_rate": 3.616756873426373e-08,
      "loss": 0.4304,
      "step": 462610
    },
    {
      "epoch": 189.37,
      "grad_norm": 1.9651246070861816,
      "learning_rate": 3.6139813911562155e-08,
      "loss": 0.4321,
      "step": 462620
    },
    {
      "epoch": 189.37,
      "grad_norm": 2.3143398761749268,
      "learning_rate": 3.6112069659665744e-08,
      "loss": 0.4294,
      "step": 462630
    },
    {
      "epoch": 189.37,
      "grad_norm": 2.2348434925079346,
      "learning_rate": 3.6084335978701226e-08,
      "loss": 0.4428,
      "step": 462640
    },
    {
      "epoch": 189.38,
      "grad_norm": 2.25100040435791,
      "learning_rate": 3.605661286879558e-08,
      "loss": 0.4455,
      "step": 462650
    },
    {
      "epoch": 189.38,
      "grad_norm": 2.0916852951049805,
      "learning_rate": 3.602890033007604e-08,
      "loss": 0.4184,
      "step": 462660
    },
    {
      "epoch": 189.39,
      "grad_norm": 1.907963514328003,
      "learning_rate": 3.600119836266934e-08,
      "loss": 0.4422,
      "step": 462670
    },
    {
      "epoch": 189.39,
      "grad_norm": 2.0804989337921143,
      "learning_rate": 3.5973506966702725e-08,
      "loss": 0.4572,
      "step": 462680
    },
    {
      "epoch": 189.39,
      "grad_norm": 2.055762767791748,
      "learning_rate": 3.594582614230264e-08,
      "loss": 0.4417,
      "step": 462690
    },
    {
      "epoch": 189.4,
      "grad_norm": 1.7408479452133179,
      "learning_rate": 3.5918155889596344e-08,
      "loss": 0.4523,
      "step": 462700
    },
    {
      "epoch": 189.4,
      "grad_norm": 1.9127713441848755,
      "learning_rate": 3.589049620871002e-08,
      "loss": 0.4507,
      "step": 462710
    },
    {
      "epoch": 189.41,
      "grad_norm": 2.0930914878845215,
      "learning_rate": 3.586284709977093e-08,
      "loss": 0.471,
      "step": 462720
    },
    {
      "epoch": 189.41,
      "grad_norm": 2.000731945037842,
      "learning_rate": 3.5835208562905515e-08,
      "loss": 0.4517,
      "step": 462730
    },
    {
      "epoch": 189.41,
      "grad_norm": 2.1148908138275146,
      "learning_rate": 3.580758059823999e-08,
      "loss": 0.4271,
      "step": 462740
    },
    {
      "epoch": 189.42,
      "grad_norm": 2.0475621223449707,
      "learning_rate": 3.577996320590134e-08,
      "loss": 0.4503,
      "step": 462750
    },
    {
      "epoch": 189.42,
      "grad_norm": 1.751592993736267,
      "learning_rate": 3.5752356386016276e-08,
      "loss": 0.438,
      "step": 462760
    },
    {
      "epoch": 189.43,
      "grad_norm": 1.6527823209762573,
      "learning_rate": 3.572476013871074e-08,
      "loss": 0.4282,
      "step": 462770
    },
    {
      "epoch": 189.43,
      "grad_norm": 1.8712011575698853,
      "learning_rate": 3.56971744641112e-08,
      "loss": 0.4423,
      "step": 462780
    },
    {
      "epoch": 189.44,
      "grad_norm": 1.9102996587753296,
      "learning_rate": 3.5669599362344623e-08,
      "loss": 0.4312,
      "step": 462790
    },
    {
      "epoch": 189.44,
      "grad_norm": 2.356757640838623,
      "learning_rate": 3.5642034833536715e-08,
      "loss": 0.443,
      "step": 462800
    },
    {
      "epoch": 189.44,
      "grad_norm": 2.0348198413848877,
      "learning_rate": 3.561448087781391e-08,
      "loss": 0.4393,
      "step": 462810
    },
    {
      "epoch": 189.45,
      "grad_norm": 1.958892822265625,
      "learning_rate": 3.558693749530243e-08,
      "loss": 0.4427,
      "step": 462820
    },
    {
      "epoch": 189.45,
      "grad_norm": 1.821932315826416,
      "learning_rate": 3.555940468612847e-08,
      "loss": 0.4451,
      "step": 462830
    },
    {
      "epoch": 189.46,
      "grad_norm": 2.0031514167785645,
      "learning_rate": 3.553188245041822e-08,
      "loss": 0.4423,
      "step": 462840
    },
    {
      "epoch": 189.46,
      "grad_norm": 2.078745126724243,
      "learning_rate": 3.5504370788297635e-08,
      "loss": 0.4623,
      "step": 462850
    },
    {
      "epoch": 189.46,
      "grad_norm": 1.9247444868087769,
      "learning_rate": 3.547686969989265e-08,
      "loss": 0.4384,
      "step": 462860
    },
    {
      "epoch": 189.47,
      "grad_norm": 1.8572641611099243,
      "learning_rate": 3.544937918532973e-08,
      "loss": 0.4329,
      "step": 462870
    },
    {
      "epoch": 189.47,
      "grad_norm": 2.4855895042419434,
      "learning_rate": 3.5421899244734555e-08,
      "loss": 0.4318,
      "step": 462880
    },
    {
      "epoch": 189.48,
      "grad_norm": 2.2765347957611084,
      "learning_rate": 3.539442987823254e-08,
      "loss": 0.4372,
      "step": 462890
    },
    {
      "epoch": 189.48,
      "grad_norm": 1.8616424798965454,
      "learning_rate": 3.536697108595015e-08,
      "loss": 0.4451,
      "step": 462900
    },
    {
      "epoch": 189.48,
      "grad_norm": 2.521287679672241,
      "learning_rate": 3.5339522868012806e-08,
      "loss": 0.4336,
      "step": 462910
    },
    {
      "epoch": 189.49,
      "grad_norm": 6.910282135009766,
      "learning_rate": 3.5312085224546445e-08,
      "loss": 0.4462,
      "step": 462920
    },
    {
      "epoch": 189.49,
      "grad_norm": 2.5476362705230713,
      "learning_rate": 3.5284658155676495e-08,
      "loss": 0.4536,
      "step": 462930
    },
    {
      "epoch": 189.5,
      "grad_norm": 2.6580100059509277,
      "learning_rate": 3.525724166152889e-08,
      "loss": 0.4615,
      "step": 462940
    },
    {
      "epoch": 189.5,
      "grad_norm": 1.8227970600128174,
      "learning_rate": 3.522983574222905e-08,
      "loss": 0.473,
      "step": 462950
    },
    {
      "epoch": 189.5,
      "grad_norm": 2.0456371307373047,
      "learning_rate": 3.520244039790267e-08,
      "loss": 0.4403,
      "step": 462960
    },
    {
      "epoch": 189.51,
      "grad_norm": 2.633277177810669,
      "learning_rate": 3.517505562867489e-08,
      "loss": 0.4526,
      "step": 462970
    },
    {
      "epoch": 189.51,
      "grad_norm": 2.521495819091797,
      "learning_rate": 3.5147681434671665e-08,
      "loss": 0.4439,
      "step": 462980
    },
    {
      "epoch": 189.52,
      "grad_norm": 2.141003370285034,
      "learning_rate": 3.512031781601789e-08,
      "loss": 0.4465,
      "step": 462990
    },
    {
      "epoch": 189.52,
      "grad_norm": 2.094731569290161,
      "learning_rate": 3.5092964772839244e-08,
      "loss": 0.431,
      "step": 463000
    },
    {
      "epoch": 189.53,
      "grad_norm": 2.520381212234497,
      "learning_rate": 3.506562230526115e-08,
      "loss": 0.4405,
      "step": 463010
    },
    {
      "epoch": 189.53,
      "grad_norm": 2.046616315841675,
      "learning_rate": 3.5038290413408504e-08,
      "loss": 0.4359,
      "step": 463020
    },
    {
      "epoch": 189.53,
      "grad_norm": 1.815657615661621,
      "learning_rate": 3.5010969097406474e-08,
      "loss": 0.4473,
      "step": 463030
    },
    {
      "epoch": 189.54,
      "grad_norm": 2.051565647125244,
      "learning_rate": 3.498365835738074e-08,
      "loss": 0.4249,
      "step": 463040
    },
    {
      "epoch": 189.54,
      "grad_norm": 2.169534206390381,
      "learning_rate": 3.4956358193455675e-08,
      "loss": 0.4642,
      "step": 463050
    },
    {
      "epoch": 189.55,
      "grad_norm": 2.027738571166992,
      "learning_rate": 3.492906860575696e-08,
      "loss": 0.4392,
      "step": 463060
    },
    {
      "epoch": 189.55,
      "grad_norm": 1.9166524410247803,
      "learning_rate": 3.490178959440924e-08,
      "loss": 0.4301,
      "step": 463070
    },
    {
      "epoch": 189.55,
      "grad_norm": 2.521597385406494,
      "learning_rate": 3.487452115953767e-08,
      "loss": 0.4494,
      "step": 463080
    },
    {
      "epoch": 189.56,
      "grad_norm": 1.9628790616989136,
      "learning_rate": 3.484726330126716e-08,
      "loss": 0.4646,
      "step": 463090
    },
    {
      "epoch": 189.56,
      "grad_norm": 1.9305232763290405,
      "learning_rate": 3.4820016019722344e-08,
      "loss": 0.4533,
      "step": 463100
    },
    {
      "epoch": 189.57,
      "grad_norm": 2.4736709594726562,
      "learning_rate": 3.479277931502837e-08,
      "loss": 0.4379,
      "step": 463110
    },
    {
      "epoch": 189.57,
      "grad_norm": 1.970604419708252,
      "learning_rate": 3.47655531873099e-08,
      "loss": 0.4536,
      "step": 463120
    },
    {
      "epoch": 189.57,
      "grad_norm": 2.143298864364624,
      "learning_rate": 3.47383376366913e-08,
      "loss": 0.4255,
      "step": 463130
    },
    {
      "epoch": 189.58,
      "grad_norm": 1.9947795867919922,
      "learning_rate": 3.471113266329774e-08,
      "loss": 0.4399,
      "step": 463140
    },
    {
      "epoch": 189.58,
      "grad_norm": 1.9368246793746948,
      "learning_rate": 3.468393826725358e-08,
      "loss": 0.4457,
      "step": 463150
    },
    {
      "epoch": 189.59,
      "grad_norm": 2.798463821411133,
      "learning_rate": 3.465675444868348e-08,
      "loss": 0.44,
      "step": 463160
    },
    {
      "epoch": 189.59,
      "grad_norm": 1.8519915342330933,
      "learning_rate": 3.462958120771181e-08,
      "loss": 0.462,
      "step": 463170
    },
    {
      "epoch": 189.59,
      "grad_norm": 2.2670741081237793,
      "learning_rate": 3.4602418544463476e-08,
      "loss": 0.4377,
      "step": 463180
    },
    {
      "epoch": 189.6,
      "grad_norm": 2.2386014461517334,
      "learning_rate": 3.457526645906233e-08,
      "loss": 0.4486,
      "step": 463190
    },
    {
      "epoch": 189.6,
      "grad_norm": 2.6313915252685547,
      "learning_rate": 3.4548124951633025e-08,
      "loss": 0.4403,
      "step": 463200
    },
    {
      "epoch": 189.61,
      "grad_norm": 1.881654977798462,
      "learning_rate": 3.452099402229992e-08,
      "loss": 0.4699,
      "step": 463210
    },
    {
      "epoch": 189.61,
      "grad_norm": 2.41980242729187,
      "learning_rate": 3.449387367118741e-08,
      "loss": 0.434,
      "step": 463220
    },
    {
      "epoch": 189.62,
      "grad_norm": 1.712565541267395,
      "learning_rate": 3.446676389841935e-08,
      "loss": 0.4529,
      "step": 463230
    },
    {
      "epoch": 189.62,
      "grad_norm": 2.1729564666748047,
      "learning_rate": 3.443966470412013e-08,
      "loss": 0.4486,
      "step": 463240
    },
    {
      "epoch": 189.62,
      "grad_norm": 2.038386106491089,
      "learning_rate": 3.441257608841411e-08,
      "loss": 0.4602,
      "step": 463250
    },
    {
      "epoch": 189.63,
      "grad_norm": 1.8489826917648315,
      "learning_rate": 3.43854980514249e-08,
      "loss": 0.4374,
      "step": 463260
    },
    {
      "epoch": 189.63,
      "grad_norm": 2.162468433380127,
      "learning_rate": 3.4358430593277137e-08,
      "loss": 0.451,
      "step": 463270
    },
    {
      "epoch": 189.64,
      "grad_norm": 1.9124678373336792,
      "learning_rate": 3.433137371409442e-08,
      "loss": 0.4416,
      "step": 463280
    },
    {
      "epoch": 189.64,
      "grad_norm": 2.3999674320220947,
      "learning_rate": 3.4304327414000617e-08,
      "loss": 0.4323,
      "step": 463290
    },
    {
      "epoch": 189.64,
      "grad_norm": 2.22483491897583,
      "learning_rate": 3.4277291693119834e-08,
      "loss": 0.4509,
      "step": 463300
    },
    {
      "epoch": 189.65,
      "grad_norm": 1.6525545120239258,
      "learning_rate": 3.42502665515762e-08,
      "loss": 0.4678,
      "step": 463310
    },
    {
      "epoch": 189.65,
      "grad_norm": 5.885268211364746,
      "learning_rate": 3.422325198949279e-08,
      "loss": 0.4482,
      "step": 463320
    },
    {
      "epoch": 189.66,
      "grad_norm": 2.123009443283081,
      "learning_rate": 3.419624800699398e-08,
      "loss": 0.4404,
      "step": 463330
    },
    {
      "epoch": 189.66,
      "grad_norm": 2.1888060569763184,
      "learning_rate": 3.416925460420312e-08,
      "loss": 0.4215,
      "step": 463340
    },
    {
      "epoch": 189.66,
      "grad_norm": 2.1150753498077393,
      "learning_rate": 3.41422717812438e-08,
      "loss": 0.4438,
      "step": 463350
    },
    {
      "epoch": 189.67,
      "grad_norm": 2.193417549133301,
      "learning_rate": 3.4115299538239884e-08,
      "loss": 0.4361,
      "step": 463360
    },
    {
      "epoch": 189.67,
      "grad_norm": 1.9731959104537964,
      "learning_rate": 3.4088337875314966e-08,
      "loss": 0.4493,
      "step": 463370
    },
    {
      "epoch": 189.68,
      "grad_norm": 1.8610881567001343,
      "learning_rate": 3.4061386792592386e-08,
      "loss": 0.4595,
      "step": 463380
    },
    {
      "epoch": 189.68,
      "grad_norm": 2.174314022064209,
      "learning_rate": 3.403444629019574e-08,
      "loss": 0.4649,
      "step": 463390
    },
    {
      "epoch": 189.68,
      "grad_norm": 1.827915906906128,
      "learning_rate": 3.400751636824812e-08,
      "loss": 0.4408,
      "step": 463400
    },
    {
      "epoch": 189.69,
      "grad_norm": 1.9085021018981934,
      "learning_rate": 3.398059702687311e-08,
      "loss": 0.4376,
      "step": 463410
    },
    {
      "epoch": 189.69,
      "grad_norm": 1.9724210500717163,
      "learning_rate": 3.3953688266194054e-08,
      "loss": 0.4482,
      "step": 463420
    },
    {
      "epoch": 189.7,
      "grad_norm": 2.033172130584717,
      "learning_rate": 3.39267900863343e-08,
      "loss": 0.4249,
      "step": 463430
    },
    {
      "epoch": 189.7,
      "grad_norm": 2.2024848461151123,
      "learning_rate": 3.389990248741665e-08,
      "loss": 0.4578,
      "step": 463440
    },
    {
      "epoch": 189.71,
      "grad_norm": 1.7592791318893433,
      "learning_rate": 3.3873025469565234e-08,
      "loss": 0.43,
      "step": 463450
    },
    {
      "epoch": 189.71,
      "grad_norm": 2.0109477043151855,
      "learning_rate": 3.384615903290209e-08,
      "loss": 0.4426,
      "step": 463460
    },
    {
      "epoch": 189.71,
      "grad_norm": 2.0007522106170654,
      "learning_rate": 3.381930317755055e-08,
      "loss": 0.4535,
      "step": 463470
    },
    {
      "epoch": 189.72,
      "grad_norm": 2.5478129386901855,
      "learning_rate": 3.379245790363395e-08,
      "loss": 0.4482,
      "step": 463480
    },
    {
      "epoch": 189.72,
      "grad_norm": 2.0123486518859863,
      "learning_rate": 3.3765623211275125e-08,
      "loss": 0.417,
      "step": 463490
    },
    {
      "epoch": 189.73,
      "grad_norm": 2.0044445991516113,
      "learning_rate": 3.373879910059688e-08,
      "loss": 0.4423,
      "step": 463500
    },
    {
      "epoch": 189.73,
      "grad_norm": 2.122661828994751,
      "learning_rate": 3.371198557172231e-08,
      "loss": 0.4365,
      "step": 463510
    },
    {
      "epoch": 189.73,
      "grad_norm": 2.2134313583374023,
      "learning_rate": 3.368518262477421e-08,
      "loss": 0.4615,
      "step": 463520
    },
    {
      "epoch": 189.74,
      "grad_norm": 1.909777045249939,
      "learning_rate": 3.365839025987515e-08,
      "loss": 0.4553,
      "step": 463530
    },
    {
      "epoch": 189.74,
      "grad_norm": 2.0186522006988525,
      "learning_rate": 3.3631608477147947e-08,
      "loss": 0.459,
      "step": 463540
    },
    {
      "epoch": 189.75,
      "grad_norm": 2.0674524307250977,
      "learning_rate": 3.360483727671542e-08,
      "loss": 0.4429,
      "step": 463550
    },
    {
      "epoch": 189.75,
      "grad_norm": 2.188399314880371,
      "learning_rate": 3.3578076658700124e-08,
      "loss": 0.4293,
      "step": 463560
    },
    {
      "epoch": 189.75,
      "grad_norm": 2.1280808448791504,
      "learning_rate": 3.3551326623224626e-08,
      "loss": 0.4453,
      "step": 463570
    },
    {
      "epoch": 189.76,
      "grad_norm": 2.169374465942383,
      "learning_rate": 3.3524587170411476e-08,
      "loss": 0.4413,
      "step": 463580
    },
    {
      "epoch": 189.76,
      "grad_norm": 2.3831424713134766,
      "learning_rate": 3.349785830038298e-08,
      "loss": 0.4698,
      "step": 463590
    },
    {
      "epoch": 189.77,
      "grad_norm": 2.312084197998047,
      "learning_rate": 3.347114001326194e-08,
      "loss": 0.4428,
      "step": 463600
    },
    {
      "epoch": 189.77,
      "grad_norm": 1.9354370832443237,
      "learning_rate": 3.344443230917067e-08,
      "loss": 0.4493,
      "step": 463610
    },
    {
      "epoch": 189.77,
      "grad_norm": 2.199239730834961,
      "learning_rate": 3.341773518823121e-08,
      "loss": 0.4485,
      "step": 463620
    },
    {
      "epoch": 189.78,
      "grad_norm": 2.0917303562164307,
      "learning_rate": 3.3391048650566103e-08,
      "loss": 0.4309,
      "step": 463630
    },
    {
      "epoch": 189.78,
      "grad_norm": 1.8686772584915161,
      "learning_rate": 3.336437269629766e-08,
      "loss": 0.4485,
      "step": 463640
    },
    {
      "epoch": 189.79,
      "grad_norm": 2.2790815830230713,
      "learning_rate": 3.3337707325547643e-08,
      "loss": 0.4379,
      "step": 463650
    },
    {
      "epoch": 189.79,
      "grad_norm": 2.1994526386260986,
      "learning_rate": 3.331105253843889e-08,
      "loss": 0.4335,
      "step": 463660
    },
    {
      "epoch": 189.8,
      "grad_norm": 1.924852728843689,
      "learning_rate": 3.328440833509291e-08,
      "loss": 0.4462,
      "step": 463670
    },
    {
      "epoch": 189.8,
      "grad_norm": 2.2488608360290527,
      "learning_rate": 3.3257774715632e-08,
      "loss": 0.4569,
      "step": 463680
    },
    {
      "epoch": 189.8,
      "grad_norm": 2.152420997619629,
      "learning_rate": 3.3231151680178455e-08,
      "loss": 0.4498,
      "step": 463690
    },
    {
      "epoch": 189.81,
      "grad_norm": 1.7937461137771606,
      "learning_rate": 3.320453922885354e-08,
      "loss": 0.4264,
      "step": 463700
    },
    {
      "epoch": 189.81,
      "grad_norm": 1.9399209022521973,
      "learning_rate": 3.3177937361779806e-08,
      "loss": 0.4431,
      "step": 463710
    },
    {
      "epoch": 189.82,
      "grad_norm": 2.106201171875,
      "learning_rate": 3.315134607907877e-08,
      "loss": 0.4319,
      "step": 463720
    },
    {
      "epoch": 189.82,
      "grad_norm": 3.2013912200927734,
      "learning_rate": 3.312476538087222e-08,
      "loss": 0.4488,
      "step": 463730
    },
    {
      "epoch": 189.82,
      "grad_norm": 2.211851119995117,
      "learning_rate": 3.309819526728218e-08,
      "loss": 0.4501,
      "step": 463740
    },
    {
      "epoch": 189.83,
      "grad_norm": 2.38069748878479,
      "learning_rate": 3.307163573843018e-08,
      "loss": 0.4434,
      "step": 463750
    },
    {
      "epoch": 189.83,
      "grad_norm": 2.4210174083709717,
      "learning_rate": 3.304508679443799e-08,
      "loss": 0.4339,
      "step": 463760
    },
    {
      "epoch": 189.84,
      "grad_norm": 2.0430572032928467,
      "learning_rate": 3.301854843542687e-08,
      "loss": 0.4417,
      "step": 463770
    },
    {
      "epoch": 189.84,
      "grad_norm": 2.0924789905548096,
      "learning_rate": 3.299202066151885e-08,
      "loss": 0.4338,
      "step": 463780
    },
    {
      "epoch": 189.84,
      "grad_norm": 2.4245097637176514,
      "learning_rate": 3.296550347283546e-08,
      "loss": 0.4362,
      "step": 463790
    },
    {
      "epoch": 189.85,
      "grad_norm": 2.1480844020843506,
      "learning_rate": 3.2938996869497426e-08,
      "loss": 0.4286,
      "step": 463800
    },
    {
      "epoch": 189.85,
      "grad_norm": 2.4145259857177734,
      "learning_rate": 3.291250085162705e-08,
      "loss": 0.428,
      "step": 463810
    },
    {
      "epoch": 189.86,
      "grad_norm": 2.236790180206299,
      "learning_rate": 3.2886015419345335e-08,
      "loss": 0.4415,
      "step": 463820
    },
    {
      "epoch": 189.86,
      "grad_norm": 2.150416612625122,
      "learning_rate": 3.2859540572773794e-08,
      "loss": 0.4538,
      "step": 463830
    },
    {
      "epoch": 189.86,
      "grad_norm": 1.9837415218353271,
      "learning_rate": 3.283307631203342e-08,
      "loss": 0.4467,
      "step": 463840
    },
    {
      "epoch": 189.87,
      "grad_norm": 2.170154333114624,
      "learning_rate": 3.280662263724573e-08,
      "loss": 0.4514,
      "step": 463850
    },
    {
      "epoch": 189.87,
      "grad_norm": 1.8745365142822266,
      "learning_rate": 3.278017954853173e-08,
      "loss": 0.4546,
      "step": 463860
    },
    {
      "epoch": 189.88,
      "grad_norm": 1.8572890758514404,
      "learning_rate": 3.275374704601241e-08,
      "loss": 0.4658,
      "step": 463870
    },
    {
      "epoch": 189.88,
      "grad_norm": 2.0234193801879883,
      "learning_rate": 3.2727325129809024e-08,
      "loss": 0.4368,
      "step": 463880
    },
    {
      "epoch": 189.89,
      "grad_norm": 2.318974733352661,
      "learning_rate": 3.270091380004284e-08,
      "loss": 0.4292,
      "step": 463890
    },
    {
      "epoch": 189.89,
      "grad_norm": 1.6625478267669678,
      "learning_rate": 3.267451305683459e-08,
      "loss": 0.4541,
      "step": 463900
    },
    {
      "epoch": 189.89,
      "grad_norm": 2.3225040435791016,
      "learning_rate": 3.264812290030526e-08,
      "loss": 0.4428,
      "step": 463910
    },
    {
      "epoch": 189.9,
      "grad_norm": 5.431838512420654,
      "learning_rate": 3.262174333057534e-08,
      "loss": 0.4504,
      "step": 463920
    },
    {
      "epoch": 189.9,
      "grad_norm": 2.3402578830718994,
      "learning_rate": 3.2595374347766345e-08,
      "loss": 0.4565,
      "step": 463930
    },
    {
      "epoch": 189.91,
      "grad_norm": 2.3246009349823,
      "learning_rate": 3.2569015951998746e-08,
      "loss": 0.4433,
      "step": 463940
    },
    {
      "epoch": 189.91,
      "grad_norm": 2.432328701019287,
      "learning_rate": 3.254266814339329e-08,
      "loss": 0.464,
      "step": 463950
    },
    {
      "epoch": 189.91,
      "grad_norm": 2.5224390029907227,
      "learning_rate": 3.25163309220707e-08,
      "loss": 0.4512,
      "step": 463960
    },
    {
      "epoch": 189.92,
      "grad_norm": 2.254426956176758,
      "learning_rate": 3.2490004288151466e-08,
      "loss": 0.4564,
      "step": 463970
    },
    {
      "epoch": 189.92,
      "grad_norm": 2.074631929397583,
      "learning_rate": 3.246368824175632e-08,
      "loss": 0.4398,
      "step": 463980
    },
    {
      "epoch": 189.93,
      "grad_norm": 1.9947419166564941,
      "learning_rate": 3.243738278300573e-08,
      "loss": 0.4134,
      "step": 463990
    },
    {
      "epoch": 189.93,
      "grad_norm": 2.055860757827759,
      "learning_rate": 3.241108791202044e-08,
      "loss": 0.4469,
      "step": 464000
    },
    {
      "epoch": 189.93,
      "grad_norm": 2.135702133178711,
      "learning_rate": 3.238480362892093e-08,
      "loss": 0.4597,
      "step": 464010
    },
    {
      "epoch": 189.94,
      "grad_norm": 2.36175537109375,
      "learning_rate": 3.2358529933827154e-08,
      "loss": 0.4408,
      "step": 464020
    },
    {
      "epoch": 189.94,
      "grad_norm": 2.0838513374328613,
      "learning_rate": 3.23322668268601e-08,
      "loss": 0.4331,
      "step": 464030
    },
    {
      "epoch": 189.95,
      "grad_norm": 1.914725422859192,
      "learning_rate": 3.230601430813948e-08,
      "loss": 0.4307,
      "step": 464040
    },
    {
      "epoch": 189.95,
      "grad_norm": 2.1782472133636475,
      "learning_rate": 3.227977237778576e-08,
      "loss": 0.4491,
      "step": 464050
    },
    {
      "epoch": 189.95,
      "grad_norm": 1.7917729616165161,
      "learning_rate": 3.2253541035919155e-08,
      "loss": 0.4518,
      "step": 464060
    },
    {
      "epoch": 189.96,
      "grad_norm": 2.073430061340332,
      "learning_rate": 3.222732028265989e-08,
      "loss": 0.4269,
      "step": 464070
    },
    {
      "epoch": 189.96,
      "grad_norm": 1.852164387702942,
      "learning_rate": 3.220111011812791e-08,
      "loss": 0.4287,
      "step": 464080
    },
    {
      "epoch": 189.97,
      "grad_norm": 2.558037281036377,
      "learning_rate": 3.2174910542443704e-08,
      "loss": 0.4516,
      "step": 464090
    },
    {
      "epoch": 189.97,
      "grad_norm": 2.5692222118377686,
      "learning_rate": 3.2148721555726696e-08,
      "loss": 0.4406,
      "step": 464100
    },
    {
      "epoch": 189.98,
      "grad_norm": 1.9380733966827393,
      "learning_rate": 3.212254315809737e-08,
      "loss": 0.4418,
      "step": 464110
    },
    {
      "epoch": 189.98,
      "grad_norm": 2.078920364379883,
      "learning_rate": 3.2096375349675154e-08,
      "loss": 0.4514,
      "step": 464120
    },
    {
      "epoch": 189.98,
      "grad_norm": 1.8951122760772705,
      "learning_rate": 3.207021813058054e-08,
      "loss": 0.4412,
      "step": 464130
    },
    {
      "epoch": 189.99,
      "grad_norm": 2.2531375885009766,
      "learning_rate": 3.204407150093268e-08,
      "loss": 0.4401,
      "step": 464140
    },
    {
      "epoch": 189.99,
      "grad_norm": 1.944289207458496,
      "learning_rate": 3.201793546085182e-08,
      "loss": 0.4443,
      "step": 464150
    },
    {
      "epoch": 190.0,
      "grad_norm": 2.106724977493286,
      "learning_rate": 3.1991810010457626e-08,
      "loss": 0.4515,
      "step": 464160
    },
    {
      "epoch": 190.0,
      "grad_norm": 2.1810412406921387,
      "learning_rate": 3.196569514986955e-08,
      "loss": 0.4449,
      "step": 464170
    },
    {
      "epoch": 190.0,
      "eval_loss": 0.4432806968688965,
      "eval_runtime": 43.283,
      "eval_samples_per_second": 79.685,
      "eval_steps_per_second": 9.981,
      "step": 464170
    },
    {
      "epoch": 190.0,
      "grad_norm": 1.95705246925354,
      "learning_rate": 3.193959087920729e-08,
      "loss": 0.4259,
      "step": 464180
    },
    {
      "epoch": 190.01,
      "grad_norm": 2.140611171722412,
      "learning_rate": 3.191349719859053e-08,
      "loss": 0.462,
      "step": 464190
    },
    {
      "epoch": 190.01,
      "grad_norm": 2.136227607727051,
      "learning_rate": 3.1887414108138724e-08,
      "loss": 0.4252,
      "step": 464200
    },
    {
      "epoch": 190.02,
      "grad_norm": 2.020390272140503,
      "learning_rate": 3.186134160797155e-08,
      "loss": 0.444,
      "step": 464210
    },
    {
      "epoch": 190.02,
      "grad_norm": 2.5353140830993652,
      "learning_rate": 3.183527969820819e-08,
      "loss": 0.4248,
      "step": 464220
    },
    {
      "epoch": 190.02,
      "grad_norm": 1.8224024772644043,
      "learning_rate": 3.180922837896808e-08,
      "loss": 0.4519,
      "step": 464230
    },
    {
      "epoch": 190.03,
      "grad_norm": 2.1366612911224365,
      "learning_rate": 3.178318765037066e-08,
      "loss": 0.4275,
      "step": 464240
    },
    {
      "epoch": 190.03,
      "grad_norm": 2.061155319213867,
      "learning_rate": 3.1757157512535095e-08,
      "loss": 0.4491,
      "step": 464250
    },
    {
      "epoch": 190.04,
      "grad_norm": 2.3004794120788574,
      "learning_rate": 3.173113796558057e-08,
      "loss": 0.4227,
      "step": 464260
    },
    {
      "epoch": 190.04,
      "grad_norm": 1.9668687582015991,
      "learning_rate": 3.170512900962652e-08,
      "loss": 0.4481,
      "step": 464270
    },
    {
      "epoch": 190.05,
      "grad_norm": 2.079838752746582,
      "learning_rate": 3.1679130644792115e-08,
      "loss": 0.4557,
      "step": 464280
    },
    {
      "epoch": 190.05,
      "grad_norm": 4.6652913093566895,
      "learning_rate": 3.165314287119601e-08,
      "loss": 0.4368,
      "step": 464290
    },
    {
      "epoch": 190.05,
      "grad_norm": 2.147416830062866,
      "learning_rate": 3.1627165688957646e-08,
      "loss": 0.4597,
      "step": 464300
    },
    {
      "epoch": 190.06,
      "grad_norm": 1.8970451354980469,
      "learning_rate": 3.160119909819594e-08,
      "loss": 0.4587,
      "step": 464310
    },
    {
      "epoch": 190.06,
      "grad_norm": 2.353450298309326,
      "learning_rate": 3.1575243099029796e-08,
      "loss": 0.4334,
      "step": 464320
    },
    {
      "epoch": 190.07,
      "grad_norm": 2.0715391635894775,
      "learning_rate": 3.154929769157814e-08,
      "loss": 0.4345,
      "step": 464330
    },
    {
      "epoch": 190.07,
      "grad_norm": 2.0695688724517822,
      "learning_rate": 3.152336287596015e-08,
      "loss": 0.4266,
      "step": 464340
    },
    {
      "epoch": 190.07,
      "grad_norm": 2.2150495052337646,
      "learning_rate": 3.1497438652293946e-08,
      "loss": 0.4407,
      "step": 464350
    },
    {
      "epoch": 190.08,
      "grad_norm": 1.8704525232315063,
      "learning_rate": 3.147152502069846e-08,
      "loss": 0.4432,
      "step": 464360
    },
    {
      "epoch": 190.08,
      "grad_norm": 2.000626802444458,
      "learning_rate": 3.144562198129259e-08,
      "loss": 0.4358,
      "step": 464370
    },
    {
      "epoch": 190.09,
      "grad_norm": 2.049856185913086,
      "learning_rate": 3.141972953419527e-08,
      "loss": 0.4368,
      "step": 464380
    },
    {
      "epoch": 190.09,
      "grad_norm": 2.1584043502807617,
      "learning_rate": 3.139384767952462e-08,
      "loss": 0.4384,
      "step": 464390
    },
    {
      "epoch": 190.09,
      "grad_norm": 1.9552557468414307,
      "learning_rate": 3.13679764173993e-08,
      "loss": 0.442,
      "step": 464400
    },
    {
      "epoch": 190.1,
      "grad_norm": 2.4086296558380127,
      "learning_rate": 3.134211574793797e-08,
      "loss": 0.4442,
      "step": 464410
    },
    {
      "epoch": 190.1,
      "grad_norm": 1.8367633819580078,
      "learning_rate": 3.131626567125928e-08,
      "loss": 0.454,
      "step": 464420
    },
    {
      "epoch": 190.11,
      "grad_norm": 2.200045108795166,
      "learning_rate": 3.12904261874811e-08,
      "loss": 0.4627,
      "step": 464430
    },
    {
      "epoch": 190.11,
      "grad_norm": 2.3056881427764893,
      "learning_rate": 3.1264597296722354e-08,
      "loss": 0.4325,
      "step": 464440
    },
    {
      "epoch": 190.11,
      "grad_norm": 1.8282748460769653,
      "learning_rate": 3.123877899910091e-08,
      "loss": 0.432,
      "step": 464450
    },
    {
      "epoch": 190.12,
      "grad_norm": 1.866855502128601,
      "learning_rate": 3.121297129473517e-08,
      "loss": 0.4377,
      "step": 464460
    },
    {
      "epoch": 190.12,
      "grad_norm": 1.9082778692245483,
      "learning_rate": 3.1187174183743525e-08,
      "loss": 0.4621,
      "step": 464470
    },
    {
      "epoch": 190.13,
      "grad_norm": 1.8885581493377686,
      "learning_rate": 3.1161387666244364e-08,
      "loss": 0.4515,
      "step": 464480
    },
    {
      "epoch": 190.13,
      "grad_norm": 2.003809928894043,
      "learning_rate": 3.113561174235505e-08,
      "loss": 0.4255,
      "step": 464490
    },
    {
      "epoch": 190.14,
      "grad_norm": 5.829644680023193,
      "learning_rate": 3.110984641219423e-08,
      "loss": 0.4267,
      "step": 464500
    },
    {
      "epoch": 190.14,
      "grad_norm": 2.1774792671203613,
      "learning_rate": 3.108409167587979e-08,
      "loss": 0.4565,
      "step": 464510
    },
    {
      "epoch": 190.14,
      "grad_norm": 2.28715181350708,
      "learning_rate": 3.1058347533529855e-08,
      "loss": 0.4464,
      "step": 464520
    },
    {
      "epoch": 190.15,
      "grad_norm": 1.883569598197937,
      "learning_rate": 3.1032613985262044e-08,
      "loss": 0.4523,
      "step": 464530
    },
    {
      "epoch": 190.15,
      "grad_norm": 2.1479921340942383,
      "learning_rate": 3.1006891031194486e-08,
      "loss": 0.4426,
      "step": 464540
    },
    {
      "epoch": 190.16,
      "grad_norm": 1.727513074874878,
      "learning_rate": 3.09811786714448e-08,
      "loss": 0.4235,
      "step": 464550
    },
    {
      "epoch": 190.16,
      "grad_norm": 2.091904878616333,
      "learning_rate": 3.095547690613086e-08,
      "loss": 0.4375,
      "step": 464560
    },
    {
      "epoch": 190.16,
      "grad_norm": 2.4541282653808594,
      "learning_rate": 3.092978573537054e-08,
      "loss": 0.4648,
      "step": 464570
    },
    {
      "epoch": 190.17,
      "grad_norm": 2.3230385780334473,
      "learning_rate": 3.0904105159281456e-08,
      "loss": 0.4667,
      "step": 464580
    },
    {
      "epoch": 190.17,
      "grad_norm": 2.211076498031616,
      "learning_rate": 3.087843517798122e-08,
      "loss": 0.4337,
      "step": 464590
    },
    {
      "epoch": 190.18,
      "grad_norm": 2.222916603088379,
      "learning_rate": 3.0852775791587444e-08,
      "loss": 0.4499,
      "step": 464600
    },
    {
      "epoch": 190.18,
      "grad_norm": 2.0298709869384766,
      "learning_rate": 3.082712700021774e-08,
      "loss": 0.4553,
      "step": 464610
    },
    {
      "epoch": 190.18,
      "grad_norm": 2.2985074520111084,
      "learning_rate": 3.080148880398973e-08,
      "loss": 0.4437,
      "step": 464620
    },
    {
      "epoch": 190.19,
      "grad_norm": 2.009685754776001,
      "learning_rate": 3.0775861203020254e-08,
      "loss": 0.4615,
      "step": 464630
    },
    {
      "epoch": 190.19,
      "grad_norm": 2.281127691268921,
      "learning_rate": 3.0750244197427166e-08,
      "loss": 0.4492,
      "step": 464640
    },
    {
      "epoch": 190.2,
      "grad_norm": 1.9641231298446655,
      "learning_rate": 3.072463778732784e-08,
      "loss": 0.432,
      "step": 464650
    },
    {
      "epoch": 190.2,
      "grad_norm": 2.0143415927886963,
      "learning_rate": 3.0699041972839355e-08,
      "loss": 0.4373,
      "step": 464660
    },
    {
      "epoch": 190.2,
      "grad_norm": 2.038905620574951,
      "learning_rate": 3.067345675407933e-08,
      "loss": 0.4329,
      "step": 464670
    },
    {
      "epoch": 190.21,
      "grad_norm": 2.18990421295166,
      "learning_rate": 3.064788213116435e-08,
      "loss": 0.4301,
      "step": 464680
    },
    {
      "epoch": 190.21,
      "grad_norm": 7.309529781341553,
      "learning_rate": 3.062231810421227e-08,
      "loss": 0.4271,
      "step": 464690
    },
    {
      "epoch": 190.22,
      "grad_norm": 2.192899465560913,
      "learning_rate": 3.059676467333967e-08,
      "loss": 0.4306,
      "step": 464700
    },
    {
      "epoch": 190.22,
      "grad_norm": 2.10418438911438,
      "learning_rate": 3.0571221838663915e-08,
      "loss": 0.4589,
      "step": 464710
    },
    {
      "epoch": 190.23,
      "grad_norm": 2.0808069705963135,
      "learning_rate": 3.054568960030182e-08,
      "loss": 0.456,
      "step": 464720
    },
    {
      "epoch": 190.23,
      "grad_norm": 2.1708054542541504,
      "learning_rate": 3.05201679583705e-08,
      "loss": 0.4678,
      "step": 464730
    },
    {
      "epoch": 190.23,
      "grad_norm": 2.0609002113342285,
      "learning_rate": 3.049465691298677e-08,
      "loss": 0.4395,
      "step": 464740
    },
    {
      "epoch": 190.24,
      "grad_norm": 1.9295017719268799,
      "learning_rate": 3.046915646426748e-08,
      "loss": 0.4365,
      "step": 464750
    },
    {
      "epoch": 190.24,
      "grad_norm": 2.04065203666687,
      "learning_rate": 3.0443666612329704e-08,
      "loss": 0.4574,
      "step": 464760
    },
    {
      "epoch": 190.25,
      "grad_norm": 1.9001072645187378,
      "learning_rate": 3.041818735728977e-08,
      "loss": 0.4446,
      "step": 464770
    },
    {
      "epoch": 190.25,
      "grad_norm": 1.9001719951629639,
      "learning_rate": 3.039271869926451e-08,
      "loss": 0.4408,
      "step": 464780
    },
    {
      "epoch": 190.25,
      "grad_norm": 1.8531489372253418,
      "learning_rate": 3.036726063837076e-08,
      "loss": 0.446,
      "step": 464790
    },
    {
      "epoch": 190.26,
      "grad_norm": 2.1108126640319824,
      "learning_rate": 3.034181317472509e-08,
      "loss": 0.4248,
      "step": 464800
    },
    {
      "epoch": 190.26,
      "grad_norm": 2.315753221511841,
      "learning_rate": 3.031637630844406e-08,
      "loss": 0.4535,
      "step": 464810
    },
    {
      "epoch": 190.27,
      "grad_norm": 2.1141700744628906,
      "learning_rate": 3.0290950039644266e-08,
      "loss": 0.4618,
      "step": 464820
    },
    {
      "epoch": 190.27,
      "grad_norm": 1.9795513153076172,
      "learning_rate": 3.0265534368442006e-08,
      "loss": 0.4456,
      "step": 464830
    },
    {
      "epoch": 190.27,
      "grad_norm": 1.8410475254058838,
      "learning_rate": 3.024012929495386e-08,
      "loss": 0.4233,
      "step": 464840
    },
    {
      "epoch": 190.28,
      "grad_norm": 2.0706849098205566,
      "learning_rate": 3.0214734819295885e-08,
      "loss": 0.456,
      "step": 464850
    },
    {
      "epoch": 190.28,
      "grad_norm": 2.208798408508301,
      "learning_rate": 3.01893509415849e-08,
      "loss": 0.4221,
      "step": 464860
    },
    {
      "epoch": 190.29,
      "grad_norm": 2.3441991806030273,
      "learning_rate": 3.016397766193697e-08,
      "loss": 0.4378,
      "step": 464870
    },
    {
      "epoch": 190.29,
      "grad_norm": 1.7612627744674683,
      "learning_rate": 3.013861498046815e-08,
      "loss": 0.433,
      "step": 464880
    },
    {
      "epoch": 190.29,
      "grad_norm": 1.8855396509170532,
      "learning_rate": 3.0113262897295006e-08,
      "loss": 0.4402,
      "step": 464890
    },
    {
      "epoch": 190.3,
      "grad_norm": 2.2342045307159424,
      "learning_rate": 3.008792141253333e-08,
      "loss": 0.4521,
      "step": 464900
    },
    {
      "epoch": 190.3,
      "grad_norm": 2.457826852798462,
      "learning_rate": 3.006259052629918e-08,
      "loss": 0.442,
      "step": 464910
    },
    {
      "epoch": 190.31,
      "grad_norm": 1.9466019868850708,
      "learning_rate": 3.003727023870887e-08,
      "loss": 0.4523,
      "step": 464920
    },
    {
      "epoch": 190.31,
      "grad_norm": 1.9277856349945068,
      "learning_rate": 3.001196054987819e-08,
      "loss": 0.4388,
      "step": 464930
    },
    {
      "epoch": 190.32,
      "grad_norm": 2.4261929988861084,
      "learning_rate": 2.998666145992319e-08,
      "loss": 0.4506,
      "step": 464940
    },
    {
      "epoch": 190.32,
      "grad_norm": 2.0186116695404053,
      "learning_rate": 2.996137296895967e-08,
      "loss": 0.4398,
      "step": 464950
    },
    {
      "epoch": 190.32,
      "grad_norm": 1.7561333179473877,
      "learning_rate": 2.993609507710368e-08,
      "loss": 0.4587,
      "step": 464960
    },
    {
      "epoch": 190.33,
      "grad_norm": 1.4498789310455322,
      "learning_rate": 2.9910827784470746e-08,
      "loss": 0.4452,
      "step": 464970
    },
    {
      "epoch": 190.33,
      "grad_norm": 1.9408416748046875,
      "learning_rate": 2.988557109117667e-08,
      "loss": 0.4543,
      "step": 464980
    },
    {
      "epoch": 190.34,
      "grad_norm": 2.2243852615356445,
      "learning_rate": 2.9860324997337244e-08,
      "loss": 0.4577,
      "step": 464990
    },
    {
      "epoch": 190.34,
      "grad_norm": 2.009516477584839,
      "learning_rate": 2.9835089503067996e-08,
      "loss": 0.4475,
      "step": 465000
    },
    {
      "epoch": 190.34,
      "grad_norm": 1.9515595436096191,
      "learning_rate": 2.980986460848446e-08,
      "loss": 0.4731,
      "step": 465010
    },
    {
      "epoch": 190.35,
      "grad_norm": 1.9472761154174805,
      "learning_rate": 2.9784650313702685e-08,
      "loss": 0.4352,
      "step": 465020
    },
    {
      "epoch": 190.35,
      "grad_norm": 2.060382127761841,
      "learning_rate": 2.975944661883769e-08,
      "loss": 0.4382,
      "step": 465030
    },
    {
      "epoch": 190.36,
      "grad_norm": 1.8255541324615479,
      "learning_rate": 2.973425352400501e-08,
      "loss": 0.4277,
      "step": 465040
    },
    {
      "epoch": 190.36,
      "grad_norm": 2.1378815174102783,
      "learning_rate": 2.9709071029319908e-08,
      "loss": 0.4332,
      "step": 465050
    },
    {
      "epoch": 190.36,
      "grad_norm": 1.921256422996521,
      "learning_rate": 2.9683899134898183e-08,
      "loss": 0.4376,
      "step": 465060
    },
    {
      "epoch": 190.37,
      "grad_norm": 1.9969305992126465,
      "learning_rate": 2.9658737840854844e-08,
      "loss": 0.4465,
      "step": 465070
    },
    {
      "epoch": 190.37,
      "grad_norm": 2.040304660797119,
      "learning_rate": 2.9633587147305167e-08,
      "loss": 0.451,
      "step": 465080
    },
    {
      "epoch": 190.38,
      "grad_norm": 1.92618989944458,
      "learning_rate": 2.9608447054364424e-08,
      "loss": 0.4331,
      "step": 465090
    },
    {
      "epoch": 190.38,
      "grad_norm": 2.2514781951904297,
      "learning_rate": 2.9583317562147623e-08,
      "loss": 0.4304,
      "step": 465100
    },
    {
      "epoch": 190.38,
      "grad_norm": 2.26352596282959,
      "learning_rate": 2.9558198670770297e-08,
      "loss": 0.4533,
      "step": 465110
    },
    {
      "epoch": 190.39,
      "grad_norm": 2.2336301803588867,
      "learning_rate": 2.9533090380346684e-08,
      "loss": 0.4423,
      "step": 465120
    },
    {
      "epoch": 190.39,
      "grad_norm": 2.137009859085083,
      "learning_rate": 2.950799269099257e-08,
      "loss": 0.4412,
      "step": 465130
    },
    {
      "epoch": 190.4,
      "grad_norm": 2.0527775287628174,
      "learning_rate": 2.948290560282297e-08,
      "loss": 0.4359,
      "step": 465140
    },
    {
      "epoch": 190.4,
      "grad_norm": 2.2867112159729004,
      "learning_rate": 2.9457829115952114e-08,
      "loss": 0.4591,
      "step": 465150
    },
    {
      "epoch": 190.41,
      "grad_norm": 2.2518975734710693,
      "learning_rate": 2.943276323049554e-08,
      "loss": 0.4556,
      "step": 465160
    },
    {
      "epoch": 190.41,
      "grad_norm": 2.1132307052612305,
      "learning_rate": 2.940770794656773e-08,
      "loss": 0.4586,
      "step": 465170
    },
    {
      "epoch": 190.41,
      "grad_norm": 2.5052332878112793,
      "learning_rate": 2.9382663264283704e-08,
      "loss": 0.4391,
      "step": 465180
    },
    {
      "epoch": 190.42,
      "grad_norm": 2.3613669872283936,
      "learning_rate": 2.935762918375769e-08,
      "loss": 0.4521,
      "step": 465190
    },
    {
      "epoch": 190.42,
      "grad_norm": 2.3565168380737305,
      "learning_rate": 2.9332605705105223e-08,
      "loss": 0.4392,
      "step": 465200
    },
    {
      "epoch": 190.43,
      "grad_norm": 2.165243625640869,
      "learning_rate": 2.930759282844001e-08,
      "loss": 0.4418,
      "step": 465210
    },
    {
      "epoch": 190.43,
      "grad_norm": 2.116248607635498,
      "learning_rate": 2.928259055387707e-08,
      "loss": 0.4487,
      "step": 465220
    },
    {
      "epoch": 190.43,
      "grad_norm": 2.0158438682556152,
      "learning_rate": 2.925759888153089e-08,
      "loss": 0.4523,
      "step": 465230
    },
    {
      "epoch": 190.44,
      "grad_norm": 2.8569414615631104,
      "learning_rate": 2.923261781151596e-08,
      "loss": 0.45,
      "step": 465240
    },
    {
      "epoch": 190.44,
      "grad_norm": 2.0704762935638428,
      "learning_rate": 2.9207647343946777e-08,
      "loss": 0.4487,
      "step": 465250
    },
    {
      "epoch": 190.45,
      "grad_norm": 2.1363706588745117,
      "learning_rate": 2.918268747893757e-08,
      "loss": 0.4343,
      "step": 465260
    },
    {
      "epoch": 190.45,
      "grad_norm": 1.9392789602279663,
      "learning_rate": 2.915773821660283e-08,
      "loss": 0.4407,
      "step": 465270
    },
    {
      "epoch": 190.45,
      "grad_norm": 1.8624776601791382,
      "learning_rate": 2.913279955705679e-08,
      "loss": 0.4473,
      "step": 465280
    },
    {
      "epoch": 190.46,
      "grad_norm": 1.7082175016403198,
      "learning_rate": 2.910787150041368e-08,
      "loss": 0.4369,
      "step": 465290
    },
    {
      "epoch": 190.46,
      "grad_norm": 2.783663034439087,
      "learning_rate": 2.9082954046787468e-08,
      "loss": 0.4439,
      "step": 465300
    },
    {
      "epoch": 190.47,
      "grad_norm": 1.974030613899231,
      "learning_rate": 2.9058047196292652e-08,
      "loss": 0.4554,
      "step": 465310
    },
    {
      "epoch": 190.47,
      "grad_norm": 2.0025107860565186,
      "learning_rate": 2.9033150949043204e-08,
      "loss": 0.4356,
      "step": 465320
    },
    {
      "epoch": 190.47,
      "grad_norm": 2.002723455429077,
      "learning_rate": 2.900826530515335e-08,
      "loss": 0.4462,
      "step": 465330
    },
    {
      "epoch": 190.48,
      "grad_norm": 2.453195810317993,
      "learning_rate": 2.8983390264736803e-08,
      "loss": 0.4324,
      "step": 465340
    },
    {
      "epoch": 190.48,
      "grad_norm": 2.1157779693603516,
      "learning_rate": 2.8958525827907538e-08,
      "loss": 0.4224,
      "step": 465350
    },
    {
      "epoch": 190.49,
      "grad_norm": 2.21329402923584,
      "learning_rate": 2.8933671994779522e-08,
      "loss": 0.4331,
      "step": 465360
    },
    {
      "epoch": 190.49,
      "grad_norm": 1.9836184978485107,
      "learning_rate": 2.8908828765466727e-08,
      "loss": 0.4529,
      "step": 465370
    },
    {
      "epoch": 190.5,
      "grad_norm": 1.6814968585968018,
      "learning_rate": 2.8883996140082866e-08,
      "loss": 0.4453,
      "step": 465380
    },
    {
      "epoch": 190.5,
      "grad_norm": 2.4293646812438965,
      "learning_rate": 2.8859174118741386e-08,
      "loss": 0.4627,
      "step": 465390
    },
    {
      "epoch": 190.5,
      "grad_norm": 2.054830551147461,
      "learning_rate": 2.8834362701556526e-08,
      "loss": 0.4477,
      "step": 465400
    },
    {
      "epoch": 190.51,
      "grad_norm": 2.4861655235290527,
      "learning_rate": 2.8809561888641727e-08,
      "loss": 0.4559,
      "step": 465410
    },
    {
      "epoch": 190.51,
      "grad_norm": 2.200479507446289,
      "learning_rate": 2.878477168011045e-08,
      "loss": 0.4519,
      "step": 465420
    },
    {
      "epoch": 190.52,
      "grad_norm": 2.167182683944702,
      "learning_rate": 2.8759992076076398e-08,
      "loss": 0.462,
      "step": 465430
    },
    {
      "epoch": 190.52,
      "grad_norm": 2.518277883529663,
      "learning_rate": 2.8735223076652767e-08,
      "loss": 0.4363,
      "step": 465440
    },
    {
      "epoch": 190.52,
      "grad_norm": 2.136828660964966,
      "learning_rate": 2.8710464681953786e-08,
      "loss": 0.4384,
      "step": 465450
    },
    {
      "epoch": 190.53,
      "grad_norm": 1.928483486175537,
      "learning_rate": 2.8685716892092124e-08,
      "loss": 0.4447,
      "step": 465460
    },
    {
      "epoch": 190.53,
      "grad_norm": 2.240971326828003,
      "learning_rate": 2.8660979707181234e-08,
      "loss": 0.4523,
      "step": 465470
    },
    {
      "epoch": 190.54,
      "grad_norm": 2.1879820823669434,
      "learning_rate": 2.8636253127334827e-08,
      "loss": 0.4517,
      "step": 465480
    },
    {
      "epoch": 190.54,
      "grad_norm": 2.025933027267456,
      "learning_rate": 2.8611537152666095e-08,
      "loss": 0.4224,
      "step": 465490
    },
    {
      "epoch": 190.54,
      "grad_norm": 2.1446692943573,
      "learning_rate": 2.8586831783287966e-08,
      "loss": 0.4476,
      "step": 465500
    },
    {
      "epoch": 190.55,
      "grad_norm": 2.817864418029785,
      "learning_rate": 2.8562137019313634e-08,
      "loss": 0.4357,
      "step": 465510
    },
    {
      "epoch": 190.55,
      "grad_norm": 3.875474214553833,
      "learning_rate": 2.8537452860856546e-08,
      "loss": 0.4325,
      "step": 465520
    },
    {
      "epoch": 190.56,
      "grad_norm": 2.058520793914795,
      "learning_rate": 2.8512779308029634e-08,
      "loss": 0.4551,
      "step": 465530
    },
    {
      "epoch": 190.56,
      "grad_norm": 1.860203742980957,
      "learning_rate": 2.8488116360945568e-08,
      "loss": 0.4487,
      "step": 465540
    },
    {
      "epoch": 190.56,
      "grad_norm": 2.0729572772979736,
      "learning_rate": 2.84634640197178e-08,
      "loss": 0.4485,
      "step": 465550
    },
    {
      "epoch": 190.57,
      "grad_norm": 2.2557783126831055,
      "learning_rate": 2.8438822284459257e-08,
      "loss": 0.4422,
      "step": 465560
    },
    {
      "epoch": 190.57,
      "grad_norm": 2.130668878555298,
      "learning_rate": 2.8414191155282355e-08,
      "loss": 0.4362,
      "step": 465570
    },
    {
      "epoch": 190.58,
      "grad_norm": 2.160327911376953,
      "learning_rate": 2.838957063230028e-08,
      "loss": 0.4791,
      "step": 465580
    },
    {
      "epoch": 190.58,
      "grad_norm": 1.842992901802063,
      "learning_rate": 2.8364960715625966e-08,
      "loss": 0.4438,
      "step": 465590
    },
    {
      "epoch": 190.59,
      "grad_norm": 2.09488582611084,
      "learning_rate": 2.834036140537182e-08,
      "loss": 0.4528,
      "step": 465600
    },
    {
      "epoch": 190.59,
      "grad_norm": 4.327830791473389,
      "learning_rate": 2.8315772701650515e-08,
      "loss": 0.4394,
      "step": 465610
    },
    {
      "epoch": 190.59,
      "grad_norm": 1.9283772706985474,
      "learning_rate": 2.829119460457498e-08,
      "loss": 0.4498,
      "step": 465620
    },
    {
      "epoch": 190.6,
      "grad_norm": 1.808840274810791,
      "learning_rate": 2.8266627114257625e-08,
      "loss": 0.4291,
      "step": 465630
    },
    {
      "epoch": 190.6,
      "grad_norm": 2.1502294540405273,
      "learning_rate": 2.824207023081112e-08,
      "loss": 0.4511,
      "step": 465640
    },
    {
      "epoch": 190.61,
      "grad_norm": 2.5107932090759277,
      "learning_rate": 2.8217523954347613e-08,
      "loss": 0.4405,
      "step": 465650
    },
    {
      "epoch": 190.61,
      "grad_norm": 1.8715764284133911,
      "learning_rate": 2.8192988284979777e-08,
      "loss": 0.4356,
      "step": 465660
    },
    {
      "epoch": 190.61,
      "grad_norm": 2.0012454986572266,
      "learning_rate": 2.8168463222820024e-08,
      "loss": 0.4626,
      "step": 465670
    },
    {
      "epoch": 190.62,
      "grad_norm": 2.187739610671997,
      "learning_rate": 2.8143948767980762e-08,
      "loss": 0.4584,
      "step": 465680
    },
    {
      "epoch": 190.62,
      "grad_norm": 1.9706190824508667,
      "learning_rate": 2.8119444920573877e-08,
      "loss": 0.4558,
      "step": 465690
    },
    {
      "epoch": 190.63,
      "grad_norm": 2.1777920722961426,
      "learning_rate": 2.8094951680712306e-08,
      "loss": 0.4293,
      "step": 465700
    },
    {
      "epoch": 190.63,
      "grad_norm": 2.4304237365722656,
      "learning_rate": 2.8070469048507674e-08,
      "loss": 0.45,
      "step": 465710
    },
    {
      "epoch": 190.63,
      "grad_norm": 2.1295742988586426,
      "learning_rate": 2.8045997024072654e-08,
      "loss": 0.4261,
      "step": 465720
    },
    {
      "epoch": 190.64,
      "grad_norm": 1.9266918897628784,
      "learning_rate": 2.8021535607518872e-08,
      "loss": 0.4506,
      "step": 465730
    },
    {
      "epoch": 190.64,
      "grad_norm": 2.5874319076538086,
      "learning_rate": 2.7997084798958483e-08,
      "loss": 0.4476,
      "step": 465740
    },
    {
      "epoch": 190.65,
      "grad_norm": 2.535912275314331,
      "learning_rate": 2.797264459850363e-08,
      "loss": 0.4408,
      "step": 465750
    },
    {
      "epoch": 190.65,
      "grad_norm": 2.1555564403533936,
      "learning_rate": 2.794821500626621e-08,
      "loss": 0.4326,
      "step": 465760
    },
    {
      "epoch": 190.65,
      "grad_norm": 1.6843231916427612,
      "learning_rate": 2.7923796022358107e-08,
      "loss": 0.4314,
      "step": 465770
    },
    {
      "epoch": 190.66,
      "grad_norm": 1.732014775276184,
      "learning_rate": 2.7899387646891213e-08,
      "loss": 0.4384,
      "step": 465780
    },
    {
      "epoch": 190.66,
      "grad_norm": 2.1357815265655518,
      "learning_rate": 2.7874989879977417e-08,
      "loss": 0.4553,
      "step": 465790
    },
    {
      "epoch": 190.67,
      "grad_norm": 1.7657690048217773,
      "learning_rate": 2.785060272172809e-08,
      "loss": 0.4503,
      "step": 465800
    },
    {
      "epoch": 190.67,
      "grad_norm": 2.014554500579834,
      "learning_rate": 2.782622617225538e-08,
      "loss": 0.4534,
      "step": 465810
    },
    {
      "epoch": 190.68,
      "grad_norm": 2.133197546005249,
      "learning_rate": 2.7801860231670917e-08,
      "loss": 0.4361,
      "step": 465820
    },
    {
      "epoch": 190.68,
      "grad_norm": 1.8637166023254395,
      "learning_rate": 2.777750490008607e-08,
      "loss": 0.4634,
      "step": 465830
    },
    {
      "epoch": 190.68,
      "grad_norm": 1.977378487586975,
      "learning_rate": 2.775316017761273e-08,
      "loss": 0.4375,
      "step": 465840
    },
    {
      "epoch": 190.69,
      "grad_norm": 1.971225619316101,
      "learning_rate": 2.7728826064362004e-08,
      "loss": 0.4397,
      "step": 465850
    },
    {
      "epoch": 190.69,
      "grad_norm": 2.4168972969055176,
      "learning_rate": 2.7704502560445784e-08,
      "loss": 0.4361,
      "step": 465860
    },
    {
      "epoch": 190.7,
      "grad_norm": 1.6794545650482178,
      "learning_rate": 2.7680189665975176e-08,
      "loss": 0.4236,
      "step": 465870
    },
    {
      "epoch": 190.7,
      "grad_norm": 2.4174718856811523,
      "learning_rate": 2.765588738106155e-08,
      "loss": 0.4518,
      "step": 465880
    },
    {
      "epoch": 190.7,
      "grad_norm": 1.8983925580978394,
      "learning_rate": 2.763159570581654e-08,
      "loss": 0.4445,
      "step": 465890
    },
    {
      "epoch": 190.71,
      "grad_norm": 1.9454542398452759,
      "learning_rate": 2.7607314640351248e-08,
      "loss": 0.4422,
      "step": 465900
    },
    {
      "epoch": 190.71,
      "grad_norm": 2.214944362640381,
      "learning_rate": 2.758304418477679e-08,
      "loss": 0.4479,
      "step": 465910
    },
    {
      "epoch": 190.72,
      "grad_norm": 1.9607969522476196,
      "learning_rate": 2.755878433920479e-08,
      "loss": 0.4351,
      "step": 465920
    },
    {
      "epoch": 190.72,
      "grad_norm": 2.0492546558380127,
      "learning_rate": 2.7534535103745838e-08,
      "loss": 0.455,
      "step": 465930
    },
    {
      "epoch": 190.72,
      "grad_norm": 2.247509241104126,
      "learning_rate": 2.7510296478511045e-08,
      "loss": 0.4551,
      "step": 465940
    },
    {
      "epoch": 190.73,
      "grad_norm": 2.337350368499756,
      "learning_rate": 2.7486068463611778e-08,
      "loss": 0.4471,
      "step": 465950
    },
    {
      "epoch": 190.73,
      "grad_norm": 1.7328134775161743,
      "learning_rate": 2.7461851059158886e-08,
      "loss": 0.4241,
      "step": 465960
    },
    {
      "epoch": 190.74,
      "grad_norm": 1.918838620185852,
      "learning_rate": 2.743764426526322e-08,
      "loss": 0.4428,
      "step": 465970
    },
    {
      "epoch": 190.74,
      "grad_norm": 1.8589330911636353,
      "learning_rate": 2.7413448082035885e-08,
      "loss": 0.4555,
      "step": 465980
    },
    {
      "epoch": 190.74,
      "grad_norm": 1.924157738685608,
      "learning_rate": 2.7389262509587474e-08,
      "loss": 0.4488,
      "step": 465990
    },
    {
      "epoch": 190.75,
      "grad_norm": 2.09753680229187,
      "learning_rate": 2.7365087548028836e-08,
      "loss": 0.4557,
      "step": 466000
    },
    {
      "epoch": 190.75,
      "grad_norm": 1.6016939878463745,
      "learning_rate": 2.7340923197470553e-08,
      "loss": 0.439,
      "step": 466010
    },
    {
      "epoch": 190.76,
      "grad_norm": 1.9395567178726196,
      "learning_rate": 2.7316769458023744e-08,
      "loss": 0.4339,
      "step": 466020
    },
    {
      "epoch": 190.76,
      "grad_norm": 2.4730019569396973,
      "learning_rate": 2.7292626329798728e-08,
      "loss": 0.4501,
      "step": 466030
    },
    {
      "epoch": 190.77,
      "grad_norm": 2.008678913116455,
      "learning_rate": 2.726849381290636e-08,
      "loss": 0.4158,
      "step": 466040
    },
    {
      "epoch": 190.77,
      "grad_norm": 1.7899973392486572,
      "learning_rate": 2.7244371907456706e-08,
      "loss": 0.4318,
      "step": 466050
    },
    {
      "epoch": 190.77,
      "grad_norm": 2.1144893169403076,
      "learning_rate": 2.7220260613560874e-08,
      "loss": 0.4342,
      "step": 466060
    },
    {
      "epoch": 190.78,
      "grad_norm": 2.3288838863372803,
      "learning_rate": 2.7196159931328933e-08,
      "loss": 0.4365,
      "step": 466070
    },
    {
      "epoch": 190.78,
      "grad_norm": 2.162139892578125,
      "learning_rate": 2.717206986087121e-08,
      "loss": 0.4323,
      "step": 466080
    },
    {
      "epoch": 190.79,
      "grad_norm": 2.709409475326538,
      "learning_rate": 2.71479904022983e-08,
      "loss": 0.4689,
      "step": 466090
    },
    {
      "epoch": 190.79,
      "grad_norm": 2.10862135887146,
      "learning_rate": 2.7123921555720526e-08,
      "loss": 0.4436,
      "step": 466100
    },
    {
      "epoch": 190.79,
      "grad_norm": 2.494183301925659,
      "learning_rate": 2.7099863321247956e-08,
      "loss": 0.434,
      "step": 466110
    },
    {
      "epoch": 190.8,
      "grad_norm": 2.4056146144866943,
      "learning_rate": 2.707581569899092e-08,
      "loss": 0.4513,
      "step": 466120
    },
    {
      "epoch": 190.8,
      "grad_norm": 2.039468765258789,
      "learning_rate": 2.7051778689059743e-08,
      "loss": 0.4428,
      "step": 466130
    },
    {
      "epoch": 190.81,
      "grad_norm": 1.8312294483184814,
      "learning_rate": 2.7027752291563976e-08,
      "loss": 0.4624,
      "step": 466140
    },
    {
      "epoch": 190.81,
      "grad_norm": 1.8879718780517578,
      "learning_rate": 2.700373650661421e-08,
      "loss": 0.4671,
      "step": 466150
    },
    {
      "epoch": 190.81,
      "grad_norm": 2.186950445175171,
      "learning_rate": 2.697973133432025e-08,
      "loss": 0.4187,
      "step": 466160
    },
    {
      "epoch": 190.82,
      "grad_norm": 2.7795088291168213,
      "learning_rate": 2.6955736774792163e-08,
      "loss": 0.4455,
      "step": 466170
    },
    {
      "epoch": 190.82,
      "grad_norm": 2.31318736076355,
      "learning_rate": 2.6931752828139758e-08,
      "loss": 0.4463,
      "step": 466180
    },
    {
      "epoch": 190.83,
      "grad_norm": 2.0717952251434326,
      "learning_rate": 2.6907779494472845e-08,
      "loss": 0.4321,
      "step": 466190
    },
    {
      "epoch": 190.83,
      "grad_norm": 2.183838367462158,
      "learning_rate": 2.688381677390149e-08,
      "loss": 0.4467,
      "step": 466200
    },
    {
      "epoch": 190.84,
      "grad_norm": 2.0690386295318604,
      "learning_rate": 2.6859864666535242e-08,
      "loss": 0.4516,
      "step": 466210
    },
    {
      "epoch": 190.84,
      "grad_norm": 2.0686891078948975,
      "learning_rate": 2.6835923172483908e-08,
      "loss": 0.4446,
      "step": 466220
    },
    {
      "epoch": 190.84,
      "grad_norm": 2.0656821727752686,
      "learning_rate": 2.6811992291857295e-08,
      "loss": 0.4589,
      "step": 466230
    },
    {
      "epoch": 190.85,
      "grad_norm": 2.1351943016052246,
      "learning_rate": 2.6788072024764435e-08,
      "loss": 0.4203,
      "step": 466240
    },
    {
      "epoch": 190.85,
      "grad_norm": 2.158536434173584,
      "learning_rate": 2.676416237131565e-08,
      "loss": 0.4568,
      "step": 466250
    },
    {
      "epoch": 190.86,
      "grad_norm": 2.18749737739563,
      "learning_rate": 2.674026333161997e-08,
      "loss": 0.4469,
      "step": 466260
    },
    {
      "epoch": 190.86,
      "grad_norm": 2.1789135932922363,
      "learning_rate": 2.6716374905787208e-08,
      "loss": 0.4486,
      "step": 466270
    },
    {
      "epoch": 190.86,
      "grad_norm": 1.8428775072097778,
      "learning_rate": 2.6692497093926644e-08,
      "loss": 0.453,
      "step": 466280
    },
    {
      "epoch": 190.87,
      "grad_norm": 2.003715753555298,
      "learning_rate": 2.666862989614757e-08,
      "loss": 0.449,
      "step": 466290
    },
    {
      "epoch": 190.87,
      "grad_norm": 1.8622008562088013,
      "learning_rate": 2.664477331255953e-08,
      "loss": 0.4328,
      "step": 466300
    },
    {
      "epoch": 190.88,
      "grad_norm": 2.0206596851348877,
      "learning_rate": 2.6620927343271557e-08,
      "loss": 0.4363,
      "step": 466310
    },
    {
      "epoch": 190.88,
      "grad_norm": 2.400193214416504,
      "learning_rate": 2.6597091988393193e-08,
      "loss": 0.4279,
      "step": 466320
    },
    {
      "epoch": 190.88,
      "grad_norm": 2.2005438804626465,
      "learning_rate": 2.6573267248033212e-08,
      "loss": 0.4307,
      "step": 466330
    },
    {
      "epoch": 190.89,
      "grad_norm": 2.3226375579833984,
      "learning_rate": 2.6549453122301152e-08,
      "loss": 0.4293,
      "step": 466340
    },
    {
      "epoch": 190.89,
      "grad_norm": 2.376361131668091,
      "learning_rate": 2.652564961130605e-08,
      "loss": 0.4285,
      "step": 466350
    },
    {
      "epoch": 190.9,
      "grad_norm": 2.2679247856140137,
      "learning_rate": 2.650185671515667e-08,
      "loss": 0.4488,
      "step": 466360
    },
    {
      "epoch": 190.9,
      "grad_norm": 2.1025145053863525,
      "learning_rate": 2.6478074433962297e-08,
      "loss": 0.443,
      "step": 466370
    },
    {
      "epoch": 190.9,
      "grad_norm": 1.973328948020935,
      "learning_rate": 2.6454302767831436e-08,
      "loss": 0.4498,
      "step": 466380
    },
    {
      "epoch": 190.91,
      "grad_norm": 2.1622021198272705,
      "learning_rate": 2.6430541716873642e-08,
      "loss": 0.4245,
      "step": 466390
    },
    {
      "epoch": 190.91,
      "grad_norm": 1.8972564935684204,
      "learning_rate": 2.640679128119716e-08,
      "loss": 0.44,
      "step": 466400
    },
    {
      "epoch": 190.92,
      "grad_norm": 1.9693301916122437,
      "learning_rate": 2.6383051460911015e-08,
      "loss": 0.4708,
      "step": 466410
    },
    {
      "epoch": 190.92,
      "grad_norm": 1.8700655698776245,
      "learning_rate": 2.635932225612397e-08,
      "loss": 0.433,
      "step": 466420
    },
    {
      "epoch": 190.93,
      "grad_norm": 1.9171313047409058,
      "learning_rate": 2.6335603666944803e-08,
      "loss": 0.4387,
      "step": 466430
    },
    {
      "epoch": 190.93,
      "grad_norm": 2.252800226211548,
      "learning_rate": 2.631189569348201e-08,
      "loss": 0.4318,
      "step": 466440
    },
    {
      "epoch": 190.93,
      "grad_norm": 2.5651214122772217,
      "learning_rate": 2.6288198335844365e-08,
      "loss": 0.4421,
      "step": 466450
    },
    {
      "epoch": 190.94,
      "grad_norm": 1.9669270515441895,
      "learning_rate": 2.6264511594140375e-08,
      "loss": 0.4438,
      "step": 466460
    },
    {
      "epoch": 190.94,
      "grad_norm": 2.242366075515747,
      "learning_rate": 2.6240835468478283e-08,
      "loss": 0.451,
      "step": 466470
    },
    {
      "epoch": 190.95,
      "grad_norm": 2.013822317123413,
      "learning_rate": 2.6217169958967117e-08,
      "loss": 0.4297,
      "step": 466480
    },
    {
      "epoch": 190.95,
      "grad_norm": 2.771034002304077,
      "learning_rate": 2.6193515065714606e-08,
      "loss": 0.4388,
      "step": 466490
    },
    {
      "epoch": 190.95,
      "grad_norm": 1.6277925968170166,
      "learning_rate": 2.6169870788829773e-08,
      "loss": 0.4598,
      "step": 466500
    },
    {
      "epoch": 190.96,
      "grad_norm": 1.8031624555587769,
      "learning_rate": 2.614623712842035e-08,
      "loss": 0.4659,
      "step": 466510
    },
    {
      "epoch": 190.96,
      "grad_norm": 1.962024450302124,
      "learning_rate": 2.6122614084594843e-08,
      "loss": 0.4465,
      "step": 466520
    },
    {
      "epoch": 190.97,
      "grad_norm": 1.848768949508667,
      "learning_rate": 2.6099001657461492e-08,
      "loss": 0.4393,
      "step": 466530
    },
    {
      "epoch": 190.97,
      "grad_norm": 2.1769776344299316,
      "learning_rate": 2.607539984712855e-08,
      "loss": 0.4402,
      "step": 466540
    },
    {
      "epoch": 190.97,
      "grad_norm": 2.0392327308654785,
      "learning_rate": 2.6051808653703744e-08,
      "loss": 0.44,
      "step": 466550
    },
    {
      "epoch": 190.98,
      "grad_norm": 10.921940803527832,
      "learning_rate": 2.6028228077295578e-08,
      "loss": 0.4585,
      "step": 466560
    },
    {
      "epoch": 190.98,
      "grad_norm": 2.064716100692749,
      "learning_rate": 2.600465811801204e-08,
      "loss": 0.4404,
      "step": 466570
    },
    {
      "epoch": 190.99,
      "grad_norm": 2.1678812503814697,
      "learning_rate": 2.5981098775960594e-08,
      "loss": 0.43,
      "step": 466580
    },
    {
      "epoch": 190.99,
      "grad_norm": 2.113286256790161,
      "learning_rate": 2.5957550051249747e-08,
      "loss": 0.4562,
      "step": 466590
    },
    {
      "epoch": 190.99,
      "grad_norm": 2.4217183589935303,
      "learning_rate": 2.593401194398723e-08,
      "loss": 0.4655,
      "step": 466600
    },
    {
      "epoch": 191.0,
      "grad_norm": 2.095350980758667,
      "learning_rate": 2.5910484454280764e-08,
      "loss": 0.4482,
      "step": 466610
    },
    {
      "epoch": 191.0,
      "eval_loss": 0.44494199752807617,
      "eval_runtime": 43.2251,
      "eval_samples_per_second": 79.792,
      "eval_steps_per_second": 9.994,
      "step": 466613
    },
    {
      "epoch": 191.0,
      "grad_norm": 2.019941568374634,
      "learning_rate": 2.5886967582238078e-08,
      "loss": 0.4474,
      "step": 466620
    },
    {
      "epoch": 191.01,
      "grad_norm": 2.310634136199951,
      "learning_rate": 2.5863461327966898e-08,
      "loss": 0.4493,
      "step": 466630
    },
    {
      "epoch": 191.01,
      "grad_norm": 2.042239189147949,
      "learning_rate": 2.583996569157521e-08,
      "loss": 0.4338,
      "step": 466640
    },
    {
      "epoch": 191.02,
      "grad_norm": 2.2069318294525146,
      "learning_rate": 2.5816480673170222e-08,
      "loss": 0.4588,
      "step": 466650
    },
    {
      "epoch": 191.02,
      "grad_norm": 2.156663417816162,
      "learning_rate": 2.5793006272859918e-08,
      "loss": 0.4373,
      "step": 466660
    },
    {
      "epoch": 191.02,
      "grad_norm": 2.071225643157959,
      "learning_rate": 2.5769542490751502e-08,
      "loss": 0.443,
      "step": 466670
    },
    {
      "epoch": 191.03,
      "grad_norm": 2.0583431720733643,
      "learning_rate": 2.5746089326952444e-08,
      "loss": 0.4412,
      "step": 466680
    },
    {
      "epoch": 191.03,
      "grad_norm": 1.812294840812683,
      "learning_rate": 2.572264678157047e-08,
      "loss": 0.4375,
      "step": 466690
    },
    {
      "epoch": 191.04,
      "grad_norm": 1.7599705457687378,
      "learning_rate": 2.5699214854712786e-08,
      "loss": 0.4463,
      "step": 466700
    },
    {
      "epoch": 191.04,
      "grad_norm": 2.0255353450775146,
      "learning_rate": 2.5675793546486855e-08,
      "loss": 0.4299,
      "step": 466710
    },
    {
      "epoch": 191.04,
      "grad_norm": 2.1953206062316895,
      "learning_rate": 2.5652382856999626e-08,
      "loss": 0.4574,
      "step": 466720
    },
    {
      "epoch": 191.05,
      "grad_norm": 1.9369655847549438,
      "learning_rate": 2.5628982786358824e-08,
      "loss": 0.4419,
      "step": 466730
    },
    {
      "epoch": 191.05,
      "grad_norm": 2.1538922786712646,
      "learning_rate": 2.5605593334671135e-08,
      "loss": 0.453,
      "step": 466740
    },
    {
      "epoch": 191.06,
      "grad_norm": 2.025970697402954,
      "learning_rate": 2.5582214502044285e-08,
      "loss": 0.4429,
      "step": 466750
    },
    {
      "epoch": 191.06,
      "grad_norm": 2.502436637878418,
      "learning_rate": 2.55588462885847e-08,
      "loss": 0.4438,
      "step": 466760
    },
    {
      "epoch": 191.06,
      "grad_norm": 2.0021612644195557,
      "learning_rate": 2.5535488694400102e-08,
      "loss": 0.4488,
      "step": 466770
    },
    {
      "epoch": 191.07,
      "grad_norm": 2.6465933322906494,
      "learning_rate": 2.5512141719596922e-08,
      "loss": 0.4136,
      "step": 466780
    },
    {
      "epoch": 191.07,
      "grad_norm": 2.2130959033966064,
      "learning_rate": 2.5488805364282366e-08,
      "loss": 0.4344,
      "step": 466790
    },
    {
      "epoch": 191.08,
      "grad_norm": 2.0451478958129883,
      "learning_rate": 2.5465479628563374e-08,
      "loss": 0.4436,
      "step": 466800
    },
    {
      "epoch": 191.08,
      "grad_norm": 2.1058971881866455,
      "learning_rate": 2.54421645125469e-08,
      "loss": 0.4522,
      "step": 466810
    },
    {
      "epoch": 191.08,
      "grad_norm": 2.319612503051758,
      "learning_rate": 2.541886001633936e-08,
      "loss": 0.4547,
      "step": 466820
    },
    {
      "epoch": 191.09,
      "grad_norm": 1.731553077697754,
      "learning_rate": 2.5395566140047705e-08,
      "loss": 0.4322,
      "step": 466830
    },
    {
      "epoch": 191.09,
      "grad_norm": 2.3607094287872314,
      "learning_rate": 2.5372282883778882e-08,
      "loss": 0.4548,
      "step": 466840
    },
    {
      "epoch": 191.1,
      "grad_norm": 1.839798092842102,
      "learning_rate": 2.534901024763905e-08,
      "loss": 0.4431,
      "step": 466850
    },
    {
      "epoch": 191.1,
      "grad_norm": 2.0108654499053955,
      "learning_rate": 2.5325748231735423e-08,
      "loss": 0.4549,
      "step": 466860
    },
    {
      "epoch": 191.11,
      "grad_norm": 1.9912763833999634,
      "learning_rate": 2.53024968361739e-08,
      "loss": 0.4446,
      "step": 466870
    },
    {
      "epoch": 191.11,
      "grad_norm": 3.7154622077941895,
      "learning_rate": 2.527925606106169e-08,
      "loss": 0.4536,
      "step": 466880
    },
    {
      "epoch": 191.11,
      "grad_norm": 1.8740195035934448,
      "learning_rate": 2.5256025906504695e-08,
      "loss": 0.4542,
      "step": 466890
    },
    {
      "epoch": 191.12,
      "grad_norm": 2.404608964920044,
      "learning_rate": 2.52328063726096e-08,
      "loss": 0.4627,
      "step": 466900
    },
    {
      "epoch": 191.12,
      "grad_norm": 2.0991599559783936,
      "learning_rate": 2.5209597459482574e-08,
      "loss": 0.453,
      "step": 466910
    },
    {
      "epoch": 191.13,
      "grad_norm": 1.921108365058899,
      "learning_rate": 2.5186399167230042e-08,
      "loss": 0.4219,
      "step": 466920
    },
    {
      "epoch": 191.13,
      "grad_norm": 2.455063581466675,
      "learning_rate": 2.5163211495958427e-08,
      "loss": 0.4559,
      "step": 466930
    },
    {
      "epoch": 191.13,
      "grad_norm": 2.259061336517334,
      "learning_rate": 2.5140034445773892e-08,
      "loss": 0.4457,
      "step": 466940
    },
    {
      "epoch": 191.14,
      "grad_norm": 2.105271100997925,
      "learning_rate": 2.5116868016782608e-08,
      "loss": 0.4507,
      "step": 466950
    },
    {
      "epoch": 191.14,
      "grad_norm": 2.2459139823913574,
      "learning_rate": 2.5093712209090216e-08,
      "loss": 0.4378,
      "step": 466960
    },
    {
      "epoch": 191.15,
      "grad_norm": 2.016042470932007,
      "learning_rate": 2.5070567022803402e-08,
      "loss": 0.446,
      "step": 466970
    },
    {
      "epoch": 191.15,
      "grad_norm": 1.9152501821517944,
      "learning_rate": 2.5047432458028068e-08,
      "loss": 0.4296,
      "step": 466980
    },
    {
      "epoch": 191.15,
      "grad_norm": 1.737215518951416,
      "learning_rate": 2.5024308514870127e-08,
      "loss": 0.4533,
      "step": 466990
    },
    {
      "epoch": 191.16,
      "grad_norm": 2.0937812328338623,
      "learning_rate": 2.5001195193435213e-08,
      "loss": 0.4172,
      "step": 467000
    },
    {
      "epoch": 191.16,
      "grad_norm": 2.2735631465911865,
      "learning_rate": 2.49780924938295e-08,
      "loss": 0.4594,
      "step": 467010
    },
    {
      "epoch": 191.17,
      "grad_norm": 2.097752332687378,
      "learning_rate": 2.4955000416158888e-08,
      "loss": 0.4504,
      "step": 467020
    },
    {
      "epoch": 191.17,
      "grad_norm": 2.0040714740753174,
      "learning_rate": 2.4931918960529024e-08,
      "loss": 0.4538,
      "step": 467030
    },
    {
      "epoch": 191.17,
      "grad_norm": 2.280149221420288,
      "learning_rate": 2.490884812704581e-08,
      "loss": 0.4613,
      "step": 467040
    },
    {
      "epoch": 191.18,
      "grad_norm": 2.1824254989624023,
      "learning_rate": 2.4885787915814633e-08,
      "loss": 0.4299,
      "step": 467050
    },
    {
      "epoch": 191.18,
      "grad_norm": 1.9037668704986572,
      "learning_rate": 2.4862738326941398e-08,
      "loss": 0.443,
      "step": 467060
    },
    {
      "epoch": 191.19,
      "grad_norm": 1.8632209300994873,
      "learning_rate": 2.4839699360531487e-08,
      "loss": 0.4412,
      "step": 467070
    },
    {
      "epoch": 191.19,
      "grad_norm": 1.8234566450119019,
      "learning_rate": 2.4816671016690808e-08,
      "loss": 0.4464,
      "step": 467080
    },
    {
      "epoch": 191.2,
      "grad_norm": 2.3230724334716797,
      "learning_rate": 2.4793653295524484e-08,
      "loss": 0.4478,
      "step": 467090
    },
    {
      "epoch": 191.2,
      "grad_norm": 2.173825263977051,
      "learning_rate": 2.47706461971379e-08,
      "loss": 0.4427,
      "step": 467100
    },
    {
      "epoch": 191.2,
      "grad_norm": 2.0979058742523193,
      "learning_rate": 2.4747649721636434e-08,
      "loss": 0.4448,
      "step": 467110
    },
    {
      "epoch": 191.21,
      "grad_norm": 1.8906373977661133,
      "learning_rate": 2.4724663869126e-08,
      "loss": 0.43,
      "step": 467120
    },
    {
      "epoch": 191.21,
      "grad_norm": 1.7664364576339722,
      "learning_rate": 2.4701688639711203e-08,
      "loss": 0.4455,
      "step": 467130
    },
    {
      "epoch": 191.22,
      "grad_norm": 1.8727282285690308,
      "learning_rate": 2.4678724033497678e-08,
      "loss": 0.4575,
      "step": 467140
    },
    {
      "epoch": 191.22,
      "grad_norm": 1.8983982801437378,
      "learning_rate": 2.4655770050590295e-08,
      "loss": 0.4352,
      "step": 467150
    },
    {
      "epoch": 191.22,
      "grad_norm": 1.8371156454086304,
      "learning_rate": 2.4632826691094698e-08,
      "loss": 0.4644,
      "step": 467160
    },
    {
      "epoch": 191.23,
      "grad_norm": 2.1636500358581543,
      "learning_rate": 2.460989395511575e-08,
      "loss": 0.4264,
      "step": 467170
    },
    {
      "epoch": 191.23,
      "grad_norm": 2.0344948768615723,
      "learning_rate": 2.4586971842758314e-08,
      "loss": 0.4514,
      "step": 467180
    },
    {
      "epoch": 191.24,
      "grad_norm": 2.20381760597229,
      "learning_rate": 2.4564060354127518e-08,
      "loss": 0.4342,
      "step": 467190
    },
    {
      "epoch": 191.24,
      "grad_norm": 1.963287353515625,
      "learning_rate": 2.4541159489328226e-08,
      "loss": 0.435,
      "step": 467200
    },
    {
      "epoch": 191.24,
      "grad_norm": 2.2292537689208984,
      "learning_rate": 2.4518269248465556e-08,
      "loss": 0.4454,
      "step": 467210
    },
    {
      "epoch": 191.25,
      "grad_norm": 1.9702264070510864,
      "learning_rate": 2.449538963164438e-08,
      "loss": 0.4443,
      "step": 467220
    },
    {
      "epoch": 191.25,
      "grad_norm": 2.1832845211029053,
      "learning_rate": 2.4472520638969557e-08,
      "loss": 0.4515,
      "step": 467230
    },
    {
      "epoch": 191.26,
      "grad_norm": 2.145566940307617,
      "learning_rate": 2.4449662270545432e-08,
      "loss": 0.4532,
      "step": 467240
    },
    {
      "epoch": 191.26,
      "grad_norm": 1.8756283521652222,
      "learning_rate": 2.442681452647687e-08,
      "loss": 0.4555,
      "step": 467250
    },
    {
      "epoch": 191.26,
      "grad_norm": 2.092761993408203,
      "learning_rate": 2.440397740686873e-08,
      "loss": 0.4514,
      "step": 467260
    },
    {
      "epoch": 191.27,
      "grad_norm": 1.8417096138000488,
      "learning_rate": 2.4381150911825624e-08,
      "loss": 0.4281,
      "step": 467270
    },
    {
      "epoch": 191.27,
      "grad_norm": 1.7736246585845947,
      "learning_rate": 2.435833504145189e-08,
      "loss": 0.4225,
      "step": 467280
    },
    {
      "epoch": 191.28,
      "grad_norm": 1.9819433689117432,
      "learning_rate": 2.4335529795852135e-08,
      "loss": 0.4535,
      "step": 467290
    },
    {
      "epoch": 191.28,
      "grad_norm": 1.9974522590637207,
      "learning_rate": 2.431273517513096e-08,
      "loss": 0.4606,
      "step": 467300
    },
    {
      "epoch": 191.29,
      "grad_norm": 1.9697041511535645,
      "learning_rate": 2.4289951179392706e-08,
      "loss": 0.4534,
      "step": 467310
    },
    {
      "epoch": 191.29,
      "grad_norm": 2.0741398334503174,
      "learning_rate": 2.4267177808741726e-08,
      "loss": 0.4342,
      "step": 467320
    },
    {
      "epoch": 191.29,
      "grad_norm": 2.1732053756713867,
      "learning_rate": 2.4244415063282356e-08,
      "loss": 0.4361,
      "step": 467330
    },
    {
      "epoch": 191.3,
      "grad_norm": 2.139068841934204,
      "learning_rate": 2.422166294311868e-08,
      "loss": 0.456,
      "step": 467340
    },
    {
      "epoch": 191.3,
      "grad_norm": 2.0219693183898926,
      "learning_rate": 2.4198921448355303e-08,
      "loss": 0.4291,
      "step": 467350
    },
    {
      "epoch": 191.31,
      "grad_norm": 1.794211506843567,
      "learning_rate": 2.417619057909605e-08,
      "loss": 0.4395,
      "step": 467360
    },
    {
      "epoch": 191.31,
      "grad_norm": 2.0017712116241455,
      "learning_rate": 2.4153470335445262e-08,
      "loss": 0.4288,
      "step": 467370
    },
    {
      "epoch": 191.31,
      "grad_norm": 2.0515103340148926,
      "learning_rate": 2.4130760717506766e-08,
      "loss": 0.4475,
      "step": 467380
    },
    {
      "epoch": 191.32,
      "grad_norm": 2.1133174896240234,
      "learning_rate": 2.41080617253849e-08,
      "loss": 0.4569,
      "step": 467390
    },
    {
      "epoch": 191.32,
      "grad_norm": 2.0081229209899902,
      "learning_rate": 2.408537335918349e-08,
      "loss": 0.4399,
      "step": 467400
    },
    {
      "epoch": 191.33,
      "grad_norm": 2.478332757949829,
      "learning_rate": 2.406269561900662e-08,
      "loss": 0.4516,
      "step": 467410
    },
    {
      "epoch": 191.33,
      "grad_norm": 2.3603694438934326,
      "learning_rate": 2.4040028504957852e-08,
      "loss": 0.4344,
      "step": 467420
    },
    {
      "epoch": 191.33,
      "grad_norm": 2.241856098175049,
      "learning_rate": 2.401737201714153e-08,
      "loss": 0.4438,
      "step": 467430
    },
    {
      "epoch": 191.34,
      "grad_norm": 2.034099578857422,
      "learning_rate": 2.39947261556607e-08,
      "loss": 0.4474,
      "step": 467440
    },
    {
      "epoch": 191.34,
      "grad_norm": 2.1471214294433594,
      "learning_rate": 2.39720909206197e-08,
      "loss": 0.4703,
      "step": 467450
    },
    {
      "epoch": 191.35,
      "grad_norm": 2.219017744064331,
      "learning_rate": 2.3949466312122093e-08,
      "loss": 0.4294,
      "step": 467460
    },
    {
      "epoch": 191.35,
      "grad_norm": 1.8226240873336792,
      "learning_rate": 2.3926852330271446e-08,
      "loss": 0.4297,
      "step": 467470
    },
    {
      "epoch": 191.35,
      "grad_norm": 1.9016315937042236,
      "learning_rate": 2.390424897517132e-08,
      "loss": 0.4587,
      "step": 467480
    },
    {
      "epoch": 191.36,
      "grad_norm": 1.9482933282852173,
      "learning_rate": 2.388165624692528e-08,
      "loss": 0.451,
      "step": 467490
    },
    {
      "epoch": 191.36,
      "grad_norm": 2.1920125484466553,
      "learning_rate": 2.3859074145636882e-08,
      "loss": 0.4312,
      "step": 467500
    },
    {
      "epoch": 191.37,
      "grad_norm": 2.143944501876831,
      "learning_rate": 2.38365026714097e-08,
      "loss": 0.4656,
      "step": 467510
    },
    {
      "epoch": 191.37,
      "grad_norm": 2.3051326274871826,
      "learning_rate": 2.3813941824346766e-08,
      "loss": 0.4276,
      "step": 467520
    },
    {
      "epoch": 191.38,
      "grad_norm": 1.8698053359985352,
      "learning_rate": 2.3791391604551912e-08,
      "loss": 0.4306,
      "step": 467530
    },
    {
      "epoch": 191.38,
      "grad_norm": 1.738635778427124,
      "learning_rate": 2.3768852012127915e-08,
      "loss": 0.4479,
      "step": 467540
    },
    {
      "epoch": 191.38,
      "grad_norm": 2.2724740505218506,
      "learning_rate": 2.3746323047178078e-08,
      "loss": 0.4466,
      "step": 467550
    },
    {
      "epoch": 191.39,
      "grad_norm": 2.346696138381958,
      "learning_rate": 2.3723804709805965e-08,
      "loss": 0.4404,
      "step": 467560
    },
    {
      "epoch": 191.39,
      "grad_norm": 2.196319103240967,
      "learning_rate": 2.370129700011436e-08,
      "loss": 0.4105,
      "step": 467570
    },
    {
      "epoch": 191.4,
      "grad_norm": 2.644550323486328,
      "learning_rate": 2.3678799918206825e-08,
      "loss": 0.4465,
      "step": 467580
    },
    {
      "epoch": 191.4,
      "grad_norm": 1.9734441041946411,
      "learning_rate": 2.365631346418614e-08,
      "loss": 0.4226,
      "step": 467590
    },
    {
      "epoch": 191.4,
      "grad_norm": 2.704505681991577,
      "learning_rate": 2.363383763815535e-08,
      "loss": 0.4206,
      "step": 467600
    },
    {
      "epoch": 191.41,
      "grad_norm": 1.8657552003860474,
      "learning_rate": 2.361137244021724e-08,
      "loss": 0.4184,
      "step": 467610
    },
    {
      "epoch": 191.41,
      "grad_norm": 2.036935329437256,
      "learning_rate": 2.3588917870474842e-08,
      "loss": 0.44,
      "step": 467620
    },
    {
      "epoch": 191.42,
      "grad_norm": 2.2205917835235596,
      "learning_rate": 2.3566473929031212e-08,
      "loss": 0.4388,
      "step": 467630
    },
    {
      "epoch": 191.42,
      "grad_norm": 1.959058165550232,
      "learning_rate": 2.3544040615988864e-08,
      "loss": 0.4343,
      "step": 467640
    },
    {
      "epoch": 191.42,
      "grad_norm": 1.9604620933532715,
      "learning_rate": 2.3521617931450584e-08,
      "loss": 0.424,
      "step": 467650
    },
    {
      "epoch": 191.43,
      "grad_norm": 4.533605098724365,
      "learning_rate": 2.349920587551967e-08,
      "loss": 0.4536,
      "step": 467660
    },
    {
      "epoch": 191.43,
      "grad_norm": 2.2944371700286865,
      "learning_rate": 2.3476804448297868e-08,
      "loss": 0.4417,
      "step": 467670
    },
    {
      "epoch": 191.44,
      "grad_norm": 1.9228664636611938,
      "learning_rate": 2.345441364988822e-08,
      "loss": 0.4612,
      "step": 467680
    },
    {
      "epoch": 191.44,
      "grad_norm": 1.836862325668335,
      "learning_rate": 2.343203348039351e-08,
      "loss": 0.4393,
      "step": 467690
    },
    {
      "epoch": 191.44,
      "grad_norm": 2.343230724334717,
      "learning_rate": 2.3409663939915994e-08,
      "loss": 0.4508,
      "step": 467700
    },
    {
      "epoch": 191.45,
      "grad_norm": 3.1811532974243164,
      "learning_rate": 2.33873050285582e-08,
      "loss": 0.4325,
      "step": 467710
    },
    {
      "epoch": 191.45,
      "grad_norm": 2.3310534954071045,
      "learning_rate": 2.3364956746422387e-08,
      "loss": 0.4357,
      "step": 467720
    },
    {
      "epoch": 191.46,
      "grad_norm": 2.1382181644439697,
      "learning_rate": 2.334261909361134e-08,
      "loss": 0.4299,
      "step": 467730
    },
    {
      "epoch": 191.46,
      "grad_norm": 1.866146206855774,
      "learning_rate": 2.3320292070227055e-08,
      "loss": 0.4523,
      "step": 467740
    },
    {
      "epoch": 191.47,
      "grad_norm": 2.241278886795044,
      "learning_rate": 2.3297975676372057e-08,
      "loss": 0.4185,
      "step": 467750
    },
    {
      "epoch": 191.47,
      "grad_norm": 2.057586431503296,
      "learning_rate": 2.3275669912148092e-08,
      "loss": 0.4487,
      "step": 467760
    },
    {
      "epoch": 191.47,
      "grad_norm": 2.1183090209960938,
      "learning_rate": 2.325337477765768e-08,
      "loss": 0.4354,
      "step": 467770
    },
    {
      "epoch": 191.48,
      "grad_norm": 1.8444396257400513,
      "learning_rate": 2.323109027300308e-08,
      "loss": 0.458,
      "step": 467780
    },
    {
      "epoch": 191.48,
      "grad_norm": 2.417357921600342,
      "learning_rate": 2.3208816398286294e-08,
      "loss": 0.4435,
      "step": 467790
    },
    {
      "epoch": 191.49,
      "grad_norm": 3.1050498485565186,
      "learning_rate": 2.3186553153609327e-08,
      "loss": 0.4286,
      "step": 467800
    },
    {
      "epoch": 191.49,
      "grad_norm": 2.079728126525879,
      "learning_rate": 2.316430053907418e-08,
      "loss": 0.4549,
      "step": 467810
    },
    {
      "epoch": 191.49,
      "grad_norm": 2.1494967937469482,
      "learning_rate": 2.314205855478233e-08,
      "loss": 0.4388,
      "step": 467820
    },
    {
      "epoch": 191.5,
      "grad_norm": 1.9479269981384277,
      "learning_rate": 2.3119827200836563e-08,
      "loss": 0.443,
      "step": 467830
    },
    {
      "epoch": 191.5,
      "grad_norm": 1.7971428632736206,
      "learning_rate": 2.309760647733784e-08,
      "loss": 0.4388,
      "step": 467840
    },
    {
      "epoch": 191.51,
      "grad_norm": 1.967678189277649,
      "learning_rate": 2.3075396384388422e-08,
      "loss": 0.4348,
      "step": 467850
    },
    {
      "epoch": 191.51,
      "grad_norm": 2.2195706367492676,
      "learning_rate": 2.3053196922090053e-08,
      "loss": 0.4541,
      "step": 467860
    },
    {
      "epoch": 191.51,
      "grad_norm": 1.8390681743621826,
      "learning_rate": 2.303100809054421e-08,
      "loss": 0.4357,
      "step": 467870
    },
    {
      "epoch": 191.52,
      "grad_norm": 2.051712989807129,
      "learning_rate": 2.30088298898529e-08,
      "loss": 0.4537,
      "step": 467880
    },
    {
      "epoch": 191.52,
      "grad_norm": 2.0332283973693848,
      "learning_rate": 2.298666232011734e-08,
      "loss": 0.4392,
      "step": 467890
    },
    {
      "epoch": 191.53,
      "grad_norm": 2.1973226070404053,
      "learning_rate": 2.296450538143901e-08,
      "loss": 0.4462,
      "step": 467900
    },
    {
      "epoch": 191.53,
      "grad_norm": 2.071650743484497,
      "learning_rate": 2.2942359073919917e-08,
      "loss": 0.4529,
      "step": 467910
    },
    {
      "epoch": 191.53,
      "grad_norm": 2.2411351203918457,
      "learning_rate": 2.2920223397661016e-08,
      "loss": 0.4608,
      "step": 467920
    },
    {
      "epoch": 191.54,
      "grad_norm": 2.297607660293579,
      "learning_rate": 2.2898098352764055e-08,
      "loss": 0.4472,
      "step": 467930
    },
    {
      "epoch": 191.54,
      "grad_norm": 2.533571720123291,
      "learning_rate": 2.287598393932999e-08,
      "loss": 0.448,
      "step": 467940
    },
    {
      "epoch": 191.55,
      "grad_norm": 2.208813190460205,
      "learning_rate": 2.2853880157460565e-08,
      "loss": 0.4508,
      "step": 467950
    },
    {
      "epoch": 191.55,
      "grad_norm": 2.1812009811401367,
      "learning_rate": 2.2831787007256737e-08,
      "loss": 0.4543,
      "step": 467960
    },
    {
      "epoch": 191.56,
      "grad_norm": 2.1178436279296875,
      "learning_rate": 2.2809704488819993e-08,
      "loss": 0.4454,
      "step": 467970
    },
    {
      "epoch": 191.56,
      "grad_norm": 2.6023874282836914,
      "learning_rate": 2.278763260225103e-08,
      "loss": 0.4577,
      "step": 467980
    },
    {
      "epoch": 191.56,
      "grad_norm": 2.834317445755005,
      "learning_rate": 2.276557134765133e-08,
      "loss": 0.4359,
      "step": 467990
    },
    {
      "epoch": 191.57,
      "grad_norm": 1.9364689588546753,
      "learning_rate": 2.2743520725121855e-08,
      "loss": 0.4395,
      "step": 468000
    },
    {
      "epoch": 191.57,
      "grad_norm": 1.8877956867218018,
      "learning_rate": 2.2721480734763567e-08,
      "loss": 0.4547,
      "step": 468010
    },
    {
      "epoch": 191.58,
      "grad_norm": 2.045745849609375,
      "learning_rate": 2.269945137667768e-08,
      "loss": 0.4343,
      "step": 468020
    },
    {
      "epoch": 191.58,
      "grad_norm": 1.8422836065292358,
      "learning_rate": 2.2677432650964646e-08,
      "loss": 0.4568,
      "step": 468030
    },
    {
      "epoch": 191.58,
      "grad_norm": 1.9058098793029785,
      "learning_rate": 2.2655424557725677e-08,
      "loss": 0.4472,
      "step": 468040
    },
    {
      "epoch": 191.59,
      "grad_norm": 2.061309814453125,
      "learning_rate": 2.2633427097061478e-08,
      "loss": 0.4358,
      "step": 468050
    },
    {
      "epoch": 191.59,
      "grad_norm": 2.205690860748291,
      "learning_rate": 2.2611440269072745e-08,
      "loss": 0.453,
      "step": 468060
    },
    {
      "epoch": 191.6,
      "grad_norm": 2.43829083442688,
      "learning_rate": 2.2589464073860443e-08,
      "loss": 0.4518,
      "step": 468070
    },
    {
      "epoch": 191.6,
      "grad_norm": 2.0117998123168945,
      "learning_rate": 2.2567498511525013e-08,
      "loss": 0.4605,
      "step": 468080
    },
    {
      "epoch": 191.6,
      "grad_norm": 1.9385120868682861,
      "learning_rate": 2.2545543582167153e-08,
      "loss": 0.4253,
      "step": 468090
    },
    {
      "epoch": 191.61,
      "grad_norm": 2.023444175720215,
      "learning_rate": 2.2523599285887306e-08,
      "loss": 0.4424,
      "step": 468100
    },
    {
      "epoch": 191.61,
      "grad_norm": 2.1961138248443604,
      "learning_rate": 2.250166562278643e-08,
      "loss": 0.4516,
      "step": 468110
    },
    {
      "epoch": 191.62,
      "grad_norm": 2.007333517074585,
      "learning_rate": 2.247974259296445e-08,
      "loss": 0.4364,
      "step": 468120
    },
    {
      "epoch": 191.62,
      "grad_norm": 2.2142930030822754,
      "learning_rate": 2.2457830196522063e-08,
      "loss": 0.4546,
      "step": 468130
    },
    {
      "epoch": 191.63,
      "grad_norm": 2.1288506984710693,
      "learning_rate": 2.243592843355945e-08,
      "loss": 0.4471,
      "step": 468140
    },
    {
      "epoch": 191.63,
      "grad_norm": 1.8580459356307983,
      "learning_rate": 2.241403730417705e-08,
      "loss": 0.4409,
      "step": 468150
    },
    {
      "epoch": 191.63,
      "grad_norm": 1.8250317573547363,
      "learning_rate": 2.2392156808475564e-08,
      "loss": 0.452,
      "step": 468160
    },
    {
      "epoch": 191.64,
      "grad_norm": 2.3585290908813477,
      "learning_rate": 2.2370286946554653e-08,
      "loss": 0.4368,
      "step": 468170
    },
    {
      "epoch": 191.64,
      "grad_norm": 2.080921173095703,
      "learning_rate": 2.2348427718514497e-08,
      "loss": 0.4485,
      "step": 468180
    },
    {
      "epoch": 191.65,
      "grad_norm": 1.7885931730270386,
      "learning_rate": 2.23265791244558e-08,
      "loss": 0.4505,
      "step": 468190
    },
    {
      "epoch": 191.65,
      "grad_norm": 1.936889410018921,
      "learning_rate": 2.2304741164477953e-08,
      "loss": 0.413,
      "step": 468200
    },
    {
      "epoch": 191.65,
      "grad_norm": 1.7770507335662842,
      "learning_rate": 2.2282913838681405e-08,
      "loss": 0.4458,
      "step": 468210
    },
    {
      "epoch": 191.66,
      "grad_norm": 1.7172480821609497,
      "learning_rate": 2.2261097147166334e-08,
      "loss": 0.4406,
      "step": 468220
    },
    {
      "epoch": 191.66,
      "grad_norm": 2.1809113025665283,
      "learning_rate": 2.2239291090032136e-08,
      "loss": 0.45,
      "step": 468230
    },
    {
      "epoch": 191.67,
      "grad_norm": 1.995606541633606,
      "learning_rate": 2.2217495667378998e-08,
      "loss": 0.4597,
      "step": 468240
    },
    {
      "epoch": 191.67,
      "grad_norm": 2.4317901134490967,
      "learning_rate": 2.2195710879306834e-08,
      "loss": 0.4411,
      "step": 468250
    },
    {
      "epoch": 191.67,
      "grad_norm": 2.1626131534576416,
      "learning_rate": 2.2173936725915306e-08,
      "loss": 0.4305,
      "step": 468260
    },
    {
      "epoch": 191.68,
      "grad_norm": 1.967155933380127,
      "learning_rate": 2.2152173207304075e-08,
      "loss": 0.4583,
      "step": 468270
    },
    {
      "epoch": 191.68,
      "grad_norm": 1.9174730777740479,
      "learning_rate": 2.213042032357306e-08,
      "loss": 0.4546,
      "step": 468280
    },
    {
      "epoch": 191.69,
      "grad_norm": 2.2661495208740234,
      "learning_rate": 2.210867807482166e-08,
      "loss": 0.4434,
      "step": 468290
    },
    {
      "epoch": 191.69,
      "grad_norm": 1.725956678390503,
      "learning_rate": 2.2086946461149536e-08,
      "loss": 0.4523,
      "step": 468300
    },
    {
      "epoch": 191.69,
      "grad_norm": 2.2571985721588135,
      "learning_rate": 2.2065225482656607e-08,
      "loss": 0.4453,
      "step": 468310
    },
    {
      "epoch": 191.7,
      "grad_norm": 2.067126512527466,
      "learning_rate": 2.2043515139441753e-08,
      "loss": 0.4309,
      "step": 468320
    },
    {
      "epoch": 191.7,
      "grad_norm": 2.057990550994873,
      "learning_rate": 2.2021815431604895e-08,
      "loss": 0.4462,
      "step": 468330
    },
    {
      "epoch": 191.71,
      "grad_norm": 2.256474733352661,
      "learning_rate": 2.200012635924517e-08,
      "loss": 0.4386,
      "step": 468340
    },
    {
      "epoch": 191.71,
      "grad_norm": 2.1704165935516357,
      "learning_rate": 2.197844792246224e-08,
      "loss": 0.4382,
      "step": 468350
    },
    {
      "epoch": 191.72,
      "grad_norm": 2.2047548294067383,
      "learning_rate": 2.1956780121354985e-08,
      "loss": 0.4365,
      "step": 468360
    },
    {
      "epoch": 191.72,
      "grad_norm": 1.9976677894592285,
      "learning_rate": 2.1935122956023062e-08,
      "loss": 0.4348,
      "step": 468370
    },
    {
      "epoch": 191.72,
      "grad_norm": 1.9748681783676147,
      "learning_rate": 2.1913476426565612e-08,
      "loss": 0.4387,
      "step": 468380
    },
    {
      "epoch": 191.73,
      "grad_norm": 1.9828730821609497,
      "learning_rate": 2.1891840533081774e-08,
      "loss": 0.4342,
      "step": 468390
    },
    {
      "epoch": 191.73,
      "grad_norm": 1.5603911876678467,
      "learning_rate": 2.1870215275670167e-08,
      "loss": 0.427,
      "step": 468400
    },
    {
      "epoch": 191.74,
      "grad_norm": 2.123055934906006,
      "learning_rate": 2.1848600654430712e-08,
      "loss": 0.4542,
      "step": 468410
    },
    {
      "epoch": 191.74,
      "grad_norm": 2.052886486053467,
      "learning_rate": 2.1826996669461506e-08,
      "loss": 0.4585,
      "step": 468420
    },
    {
      "epoch": 191.74,
      "grad_norm": 1.9315686225891113,
      "learning_rate": 2.180540332086221e-08,
      "loss": 0.4454,
      "step": 468430
    },
    {
      "epoch": 191.75,
      "grad_norm": 1.9000107049942017,
      "learning_rate": 2.1783820608731703e-08,
      "loss": 0.4424,
      "step": 468440
    },
    {
      "epoch": 191.75,
      "grad_norm": 1.9030104875564575,
      "learning_rate": 2.1762248533168346e-08,
      "loss": 0.4447,
      "step": 468450
    },
    {
      "epoch": 191.76,
      "grad_norm": 2.0448663234710693,
      "learning_rate": 2.1740687094271533e-08,
      "loss": 0.4518,
      "step": 468460
    },
    {
      "epoch": 191.76,
      "grad_norm": 2.0919711589813232,
      "learning_rate": 2.171913629213963e-08,
      "loss": 0.4392,
      "step": 468470
    },
    {
      "epoch": 191.76,
      "grad_norm": 2.1466193199157715,
      "learning_rate": 2.169759612687151e-08,
      "loss": 0.4505,
      "step": 468480
    },
    {
      "epoch": 191.77,
      "grad_norm": 2.1952712535858154,
      "learning_rate": 2.1676066598565798e-08,
      "loss": 0.4387,
      "step": 468490
    },
    {
      "epoch": 191.77,
      "grad_norm": 1.773675799369812,
      "learning_rate": 2.1654547707321366e-08,
      "loss": 0.4524,
      "step": 468500
    },
    {
      "epoch": 191.78,
      "grad_norm": 2.5131163597106934,
      "learning_rate": 2.1633039453236317e-08,
      "loss": 0.4345,
      "step": 468510
    },
    {
      "epoch": 191.78,
      "grad_norm": 1.9037693738937378,
      "learning_rate": 2.161154183640953e-08,
      "loss": 0.4423,
      "step": 468520
    },
    {
      "epoch": 191.78,
      "grad_norm": 2.108088731765747,
      "learning_rate": 2.1590054856939366e-08,
      "loss": 0.4333,
      "step": 468530
    },
    {
      "epoch": 191.79,
      "grad_norm": 2.589571475982666,
      "learning_rate": 2.156857851492444e-08,
      "loss": 0.4435,
      "step": 468540
    },
    {
      "epoch": 191.79,
      "grad_norm": 1.8988778591156006,
      "learning_rate": 2.1547112810462853e-08,
      "loss": 0.4643,
      "step": 468550
    },
    {
      "epoch": 191.8,
      "grad_norm": 1.7316112518310547,
      "learning_rate": 2.152565774365296e-08,
      "loss": 0.4359,
      "step": 468560
    },
    {
      "epoch": 191.8,
      "grad_norm": 1.8137812614440918,
      "learning_rate": 2.1504213314593126e-08,
      "loss": 0.4361,
      "step": 468570
    },
    {
      "epoch": 191.81,
      "grad_norm": 1.7920233011245728,
      "learning_rate": 2.1482779523381705e-08,
      "loss": 0.4467,
      "step": 468580
    },
    {
      "epoch": 191.81,
      "grad_norm": 2.0407068729400635,
      "learning_rate": 2.1461356370116797e-08,
      "loss": 0.4675,
      "step": 468590
    },
    {
      "epoch": 191.81,
      "grad_norm": 2.3121285438537598,
      "learning_rate": 2.14399438548965e-08,
      "loss": 0.4393,
      "step": 468600
    },
    {
      "epoch": 191.82,
      "grad_norm": 1.717414379119873,
      "learning_rate": 2.1418541977818918e-08,
      "loss": 0.4269,
      "step": 468610
    },
    {
      "epoch": 191.82,
      "grad_norm": 2.1359944343566895,
      "learning_rate": 2.139715073898214e-08,
      "loss": 0.4562,
      "step": 468620
    },
    {
      "epoch": 191.83,
      "grad_norm": 1.9379956722259521,
      "learning_rate": 2.1375770138484012e-08,
      "loss": 0.4595,
      "step": 468630
    },
    {
      "epoch": 191.83,
      "grad_norm": 2.2842466831207275,
      "learning_rate": 2.1354400176422632e-08,
      "loss": 0.4198,
      "step": 468640
    },
    {
      "epoch": 191.83,
      "grad_norm": 2.4287426471710205,
      "learning_rate": 2.1333040852895834e-08,
      "loss": 0.4455,
      "step": 468650
    },
    {
      "epoch": 191.84,
      "grad_norm": 2.0683341026306152,
      "learning_rate": 2.1311692168001197e-08,
      "loss": 0.472,
      "step": 468660
    },
    {
      "epoch": 191.84,
      "grad_norm": 1.982649803161621,
      "learning_rate": 2.1290354121837085e-08,
      "loss": 0.427,
      "step": 468670
    },
    {
      "epoch": 191.85,
      "grad_norm": 1.9290080070495605,
      "learning_rate": 2.126902671450081e-08,
      "loss": 0.4391,
      "step": 468680
    },
    {
      "epoch": 191.85,
      "grad_norm": 2.518439292907715,
      "learning_rate": 2.1247709946090214e-08,
      "loss": 0.4593,
      "step": 468690
    },
    {
      "epoch": 191.85,
      "grad_norm": 1.9799749851226807,
      "learning_rate": 2.1226403816703136e-08,
      "loss": 0.4467,
      "step": 468700
    },
    {
      "epoch": 191.86,
      "grad_norm": 1.9689162969589233,
      "learning_rate": 2.120510832643663e-08,
      "loss": 0.4375,
      "step": 468710
    },
    {
      "epoch": 191.86,
      "grad_norm": 2.0106489658355713,
      "learning_rate": 2.1183823475388536e-08,
      "loss": 0.4493,
      "step": 468720
    },
    {
      "epoch": 191.87,
      "grad_norm": 2.074697971343994,
      "learning_rate": 2.1162549263656694e-08,
      "loss": 0.4388,
      "step": 468730
    },
    {
      "epoch": 191.87,
      "grad_norm": 3.043307304382324,
      "learning_rate": 2.1141285691338162e-08,
      "loss": 0.4464,
      "step": 468740
    },
    {
      "epoch": 191.87,
      "grad_norm": 2.5937774181365967,
      "learning_rate": 2.1120032758530253e-08,
      "loss": 0.4241,
      "step": 468750
    },
    {
      "epoch": 191.88,
      "grad_norm": 1.8174231052398682,
      "learning_rate": 2.1098790465330813e-08,
      "loss": 0.4557,
      "step": 468760
    },
    {
      "epoch": 191.88,
      "grad_norm": 1.8791446685791016,
      "learning_rate": 2.1077558811836634e-08,
      "loss": 0.4546,
      "step": 468770
    },
    {
      "epoch": 191.89,
      "grad_norm": 2.2113354206085205,
      "learning_rate": 2.1056337798145298e-08,
      "loss": 0.4426,
      "step": 468780
    },
    {
      "epoch": 191.89,
      "grad_norm": 2.026951313018799,
      "learning_rate": 2.1035127424353855e-08,
      "loss": 0.4568,
      "step": 468790
    },
    {
      "epoch": 191.9,
      "grad_norm": 2.051129102706909,
      "learning_rate": 2.1013927690559633e-08,
      "loss": 0.4756,
      "step": 468800
    },
    {
      "epoch": 191.9,
      "grad_norm": 2.3365206718444824,
      "learning_rate": 2.0992738596859686e-08,
      "loss": 0.4336,
      "step": 468810
    },
    {
      "epoch": 191.9,
      "grad_norm": 1.9421287775039673,
      "learning_rate": 2.097156014335081e-08,
      "loss": 0.4423,
      "step": 468820
    },
    {
      "epoch": 191.91,
      "grad_norm": 2.0379765033721924,
      "learning_rate": 2.0950392330130583e-08,
      "loss": 0.4515,
      "step": 468830
    },
    {
      "epoch": 191.91,
      "grad_norm": 2.391035556793213,
      "learning_rate": 2.0929235157295283e-08,
      "loss": 0.4449,
      "step": 468840
    },
    {
      "epoch": 191.92,
      "grad_norm": 2.2061269283294678,
      "learning_rate": 2.090808862494223e-08,
      "loss": 0.4396,
      "step": 468850
    },
    {
      "epoch": 191.92,
      "grad_norm": 2.149014711380005,
      "learning_rate": 2.088695273316848e-08,
      "loss": 0.4503,
      "step": 468860
    },
    {
      "epoch": 191.92,
      "grad_norm": 1.7885375022888184,
      "learning_rate": 2.086582748207031e-08,
      "loss": 0.4653,
      "step": 468870
    },
    {
      "epoch": 191.93,
      "grad_norm": 2.1718807220458984,
      "learning_rate": 2.0844712871745037e-08,
      "loss": 0.4619,
      "step": 468880
    },
    {
      "epoch": 191.93,
      "grad_norm": 2.1060094833374023,
      "learning_rate": 2.082360890228894e-08,
      "loss": 0.4378,
      "step": 468890
    },
    {
      "epoch": 191.94,
      "grad_norm": 2.2820961475372314,
      "learning_rate": 2.0802515573798815e-08,
      "loss": 0.4433,
      "step": 468900
    },
    {
      "epoch": 191.94,
      "grad_norm": 1.9853585958480835,
      "learning_rate": 2.078143288637146e-08,
      "loss": 0.4628,
      "step": 468910
    },
    {
      "epoch": 191.94,
      "grad_norm": 2.2860255241394043,
      "learning_rate": 2.0760360840103418e-08,
      "loss": 0.4388,
      "step": 468920
    },
    {
      "epoch": 191.95,
      "grad_norm": 1.8362247943878174,
      "learning_rate": 2.0739299435091216e-08,
      "loss": 0.4551,
      "step": 468930
    },
    {
      "epoch": 191.95,
      "grad_norm": 2.1024982929229736,
      "learning_rate": 2.071824867143114e-08,
      "loss": 0.4505,
      "step": 468940
    },
    {
      "epoch": 191.96,
      "grad_norm": 2.0310721397399902,
      "learning_rate": 2.069720854921972e-08,
      "loss": 0.4487,
      "step": 468950
    },
    {
      "epoch": 191.96,
      "grad_norm": 2.298105478286743,
      "learning_rate": 2.06761790685535e-08,
      "loss": 0.445,
      "step": 468960
    },
    {
      "epoch": 191.96,
      "grad_norm": 2.0757877826690674,
      "learning_rate": 2.0655160229528753e-08,
      "loss": 0.4515,
      "step": 468970
    },
    {
      "epoch": 191.97,
      "grad_norm": 1.9892735481262207,
      "learning_rate": 2.06341520322415e-08,
      "loss": 0.4585,
      "step": 468980
    },
    {
      "epoch": 191.97,
      "grad_norm": 1.7197239398956299,
      "learning_rate": 2.0613154476788272e-08,
      "loss": 0.4331,
      "step": 468990
    },
    {
      "epoch": 191.98,
      "grad_norm": 2.136842966079712,
      "learning_rate": 2.059216756326509e-08,
      "loss": 0.4446,
      "step": 469000
    },
    {
      "epoch": 191.98,
      "grad_norm": 1.9467312097549438,
      "learning_rate": 2.0571191291768232e-08,
      "loss": 0.431,
      "step": 469010
    },
    {
      "epoch": 191.99,
      "grad_norm": 2.306328773498535,
      "learning_rate": 2.0550225662393712e-08,
      "loss": 0.4319,
      "step": 469020
    },
    {
      "epoch": 191.99,
      "grad_norm": 1.690578579902649,
      "learning_rate": 2.052927067523729e-08,
      "loss": 0.4423,
      "step": 469030
    },
    {
      "epoch": 191.99,
      "grad_norm": 2.0765395164489746,
      "learning_rate": 2.05083263303955e-08,
      "loss": 0.4295,
      "step": 469040
    },
    {
      "epoch": 192.0,
      "grad_norm": 1.813637137413025,
      "learning_rate": 2.0487392627964104e-08,
      "loss": 0.4395,
      "step": 469050
    },
    {
      "epoch": 192.0,
      "eval_loss": 0.4432755708694458,
      "eval_runtime": 43.5458,
      "eval_samples_per_second": 79.204,
      "eval_steps_per_second": 9.921,
      "step": 469056
    },
    {
      "epoch": 192.0,
      "grad_norm": 2.133248805999756,
      "learning_rate": 2.0466469568038592e-08,
      "loss": 0.4472,
      "step": 469060
    },
    {
      "epoch": 192.01,
      "grad_norm": 2.2726171016693115,
      "learning_rate": 2.0445557150715243e-08,
      "loss": 0.4267,
      "step": 469070
    },
    {
      "epoch": 192.01,
      "grad_norm": 1.8986127376556396,
      "learning_rate": 2.0424655376089817e-08,
      "loss": 0.4361,
      "step": 469080
    },
    {
      "epoch": 192.01,
      "grad_norm": 1.961848258972168,
      "learning_rate": 2.040376424425781e-08,
      "loss": 0.4409,
      "step": 469090
    },
    {
      "epoch": 192.02,
      "grad_norm": 2.075535774230957,
      "learning_rate": 2.0382883755315492e-08,
      "loss": 0.4562,
      "step": 469100
    },
    {
      "epoch": 192.02,
      "grad_norm": 1.8474504947662354,
      "learning_rate": 2.0362013909357848e-08,
      "loss": 0.4212,
      "step": 469110
    },
    {
      "epoch": 192.03,
      "grad_norm": 2.9221198558807373,
      "learning_rate": 2.0341154706480634e-08,
      "loss": 0.4518,
      "step": 469120
    },
    {
      "epoch": 192.03,
      "grad_norm": 2.3707151412963867,
      "learning_rate": 2.0320306146779862e-08,
      "loss": 0.4492,
      "step": 469130
    },
    {
      "epoch": 192.03,
      "grad_norm": 2.075336456298828,
      "learning_rate": 2.0299468230350248e-08,
      "loss": 0.4336,
      "step": 469140
    },
    {
      "epoch": 192.04,
      "grad_norm": 2.325139284133911,
      "learning_rate": 2.0278640957288073e-08,
      "loss": 0.4564,
      "step": 469150
    },
    {
      "epoch": 192.04,
      "grad_norm": 1.7946325540542603,
      "learning_rate": 2.025782432768805e-08,
      "loss": 0.4436,
      "step": 469160
    },
    {
      "epoch": 192.05,
      "grad_norm": 2.0356392860412598,
      "learning_rate": 2.02370183416462e-08,
      "loss": 0.4276,
      "step": 469170
    },
    {
      "epoch": 192.05,
      "grad_norm": 1.9024147987365723,
      "learning_rate": 2.0216222999257237e-08,
      "loss": 0.4486,
      "step": 469180
    },
    {
      "epoch": 192.05,
      "grad_norm": 1.9336819648742676,
      "learning_rate": 2.0195438300616655e-08,
      "loss": 0.4483,
      "step": 469190
    },
    {
      "epoch": 192.06,
      "grad_norm": 2.3658246994018555,
      "learning_rate": 2.017466424581969e-08,
      "loss": 0.4435,
      "step": 469200
    },
    {
      "epoch": 192.06,
      "grad_norm": 1.8120287656784058,
      "learning_rate": 2.0153900834961584e-08,
      "loss": 0.4438,
      "step": 469210
    },
    {
      "epoch": 192.07,
      "grad_norm": 1.938591718673706,
      "learning_rate": 2.0133148068137305e-08,
      "loss": 0.4474,
      "step": 469220
    },
    {
      "epoch": 192.07,
      "grad_norm": 2.136212110519409,
      "learning_rate": 2.011240594544184e-08,
      "loss": 0.4412,
      "step": 469230
    },
    {
      "epoch": 192.08,
      "grad_norm": 1.994956612586975,
      "learning_rate": 2.0091674466970413e-08,
      "loss": 0.4481,
      "step": 469240
    },
    {
      "epoch": 192.08,
      "grad_norm": 2.3129384517669678,
      "learning_rate": 2.0070953632818007e-08,
      "loss": 0.4349,
      "step": 469250
    },
    {
      "epoch": 192.08,
      "grad_norm": 2.233869791030884,
      "learning_rate": 2.0050243443079337e-08,
      "loss": 0.442,
      "step": 469260
    },
    {
      "epoch": 192.09,
      "grad_norm": 2.3144516944885254,
      "learning_rate": 2.002954389784938e-08,
      "loss": 0.4489,
      "step": 469270
    },
    {
      "epoch": 192.09,
      "grad_norm": 2.0935187339782715,
      "learning_rate": 2.0008854997223112e-08,
      "loss": 0.4266,
      "step": 469280
    },
    {
      "epoch": 192.1,
      "grad_norm": 2.583165407180786,
      "learning_rate": 1.9988176741294985e-08,
      "loss": 0.4343,
      "step": 469290
    },
    {
      "epoch": 192.1,
      "grad_norm": 2.091217517852783,
      "learning_rate": 1.996750913015998e-08,
      "loss": 0.4611,
      "step": 469300
    },
    {
      "epoch": 192.1,
      "grad_norm": 2.0907390117645264,
      "learning_rate": 1.994685216391255e-08,
      "loss": 0.4545,
      "step": 469310
    },
    {
      "epoch": 192.11,
      "grad_norm": 1.7312856912612915,
      "learning_rate": 1.992620584264767e-08,
      "loss": 0.4376,
      "step": 469320
    },
    {
      "epoch": 192.11,
      "grad_norm": 2.2313485145568848,
      "learning_rate": 1.990557016645954e-08,
      "loss": 0.4473,
      "step": 469330
    },
    {
      "epoch": 192.12,
      "grad_norm": 2.156440019607544,
      "learning_rate": 1.9884945135442873e-08,
      "loss": 0.4521,
      "step": 469340
    },
    {
      "epoch": 192.12,
      "grad_norm": 1.9809136390686035,
      "learning_rate": 1.986433074969212e-08,
      "loss": 0.4338,
      "step": 469350
    },
    {
      "epoch": 192.12,
      "grad_norm": 2.167018413543701,
      "learning_rate": 1.9843727009301746e-08,
      "loss": 0.4452,
      "step": 469360
    },
    {
      "epoch": 192.13,
      "grad_norm": 1.912735939025879,
      "learning_rate": 1.98231339143662e-08,
      "loss": 0.4535,
      "step": 469370
    },
    {
      "epoch": 192.13,
      "grad_norm": 1.8815522193908691,
      "learning_rate": 1.9802551464979422e-08,
      "loss": 0.4448,
      "step": 469380
    },
    {
      "epoch": 192.14,
      "grad_norm": 2.0600180625915527,
      "learning_rate": 1.9781979661236384e-08,
      "loss": 0.444,
      "step": 469390
    },
    {
      "epoch": 192.14,
      "grad_norm": 2.0446226596832275,
      "learning_rate": 1.9761418503230762e-08,
      "loss": 0.427,
      "step": 469400
    },
    {
      "epoch": 192.14,
      "grad_norm": 1.9634215831756592,
      "learning_rate": 1.9740867991057015e-08,
      "loss": 0.4405,
      "step": 469410
    },
    {
      "epoch": 192.15,
      "grad_norm": 2.4134936332702637,
      "learning_rate": 1.9720328124809073e-08,
      "loss": 0.4402,
      "step": 469420
    },
    {
      "epoch": 192.15,
      "grad_norm": 2.1709887981414795,
      "learning_rate": 1.9699798904581136e-08,
      "loss": 0.4418,
      "step": 469430
    },
    {
      "epoch": 192.16,
      "grad_norm": 2.0001845359802246,
      "learning_rate": 1.9679280330467135e-08,
      "loss": 0.4392,
      "step": 469440
    },
    {
      "epoch": 192.16,
      "grad_norm": 2.537003517150879,
      "learning_rate": 1.965877240256127e-08,
      "loss": 0.4398,
      "step": 469450
    },
    {
      "epoch": 192.17,
      "grad_norm": 1.749159574508667,
      "learning_rate": 1.963827512095747e-08,
      "loss": 0.4217,
      "step": 469460
    },
    {
      "epoch": 192.17,
      "grad_norm": 2.1397438049316406,
      "learning_rate": 1.961778848574942e-08,
      "loss": 0.4257,
      "step": 469470
    },
    {
      "epoch": 192.17,
      "grad_norm": 2.053912878036499,
      "learning_rate": 1.9597312497031046e-08,
      "loss": 0.4575,
      "step": 469480
    },
    {
      "epoch": 192.18,
      "grad_norm": 1.9885975122451782,
      "learning_rate": 1.9576847154896288e-08,
      "loss": 0.4599,
      "step": 469490
    },
    {
      "epoch": 192.18,
      "grad_norm": 2.0968739986419678,
      "learning_rate": 1.955639245943882e-08,
      "loss": 0.4449,
      "step": 469500
    },
    {
      "epoch": 192.19,
      "grad_norm": 1.9772515296936035,
      "learning_rate": 1.9535948410752313e-08,
      "loss": 0.4574,
      "step": 469510
    },
    {
      "epoch": 192.19,
      "grad_norm": 2.4370734691619873,
      "learning_rate": 1.951551500893045e-08,
      "loss": 0.4412,
      "step": 469520
    },
    {
      "epoch": 192.19,
      "grad_norm": 2.360489845275879,
      "learning_rate": 1.94950922540669e-08,
      "loss": 0.444,
      "step": 469530
    },
    {
      "epoch": 192.2,
      "grad_norm": 1.9303209781646729,
      "learning_rate": 1.947468014625508e-08,
      "loss": 0.4361,
      "step": 469540
    },
    {
      "epoch": 192.2,
      "grad_norm": 2.3749539852142334,
      "learning_rate": 1.9454278685588407e-08,
      "loss": 0.4601,
      "step": 469550
    },
    {
      "epoch": 192.21,
      "grad_norm": 1.9030349254608154,
      "learning_rate": 1.943388787216055e-08,
      "loss": 0.4336,
      "step": 469560
    },
    {
      "epoch": 192.21,
      "grad_norm": 2.2470500469207764,
      "learning_rate": 1.941350770606493e-08,
      "loss": 0.4375,
      "step": 469570
    },
    {
      "epoch": 192.21,
      "grad_norm": 2.0517406463623047,
      "learning_rate": 1.93931381873947e-08,
      "loss": 0.4288,
      "step": 469580
    },
    {
      "epoch": 192.22,
      "grad_norm": 2.161409854888916,
      "learning_rate": 1.937277931624327e-08,
      "loss": 0.4719,
      "step": 469590
    },
    {
      "epoch": 192.22,
      "grad_norm": 2.1561639308929443,
      "learning_rate": 1.9352431092704065e-08,
      "loss": 0.4597,
      "step": 469600
    },
    {
      "epoch": 192.23,
      "grad_norm": 2.238288402557373,
      "learning_rate": 1.933209351686997e-08,
      "loss": 0.4366,
      "step": 469610
    },
    {
      "epoch": 192.23,
      "grad_norm": 2.024280071258545,
      "learning_rate": 1.9311766588834405e-08,
      "loss": 0.4417,
      "step": 469620
    },
    {
      "epoch": 192.23,
      "grad_norm": 1.8646305799484253,
      "learning_rate": 1.929145030869026e-08,
      "loss": 0.4521,
      "step": 469630
    },
    {
      "epoch": 192.24,
      "grad_norm": 1.9710257053375244,
      "learning_rate": 1.9271144676530956e-08,
      "loss": 0.4427,
      "step": 469640
    },
    {
      "epoch": 192.24,
      "grad_norm": 2.165050506591797,
      "learning_rate": 1.925084969244886e-08,
      "loss": 0.442,
      "step": 469650
    },
    {
      "epoch": 192.25,
      "grad_norm": 2.0900235176086426,
      "learning_rate": 1.9230565356537915e-08,
      "loss": 0.4398,
      "step": 469660
    },
    {
      "epoch": 192.25,
      "grad_norm": 2.4638702869415283,
      "learning_rate": 1.9210291668890227e-08,
      "loss": 0.4368,
      "step": 469670
    },
    {
      "epoch": 192.26,
      "grad_norm": 2.256060838699341,
      "learning_rate": 1.9190028629599217e-08,
      "loss": 0.441,
      "step": 469680
    },
    {
      "epoch": 192.26,
      "grad_norm": 2.165076971054077,
      "learning_rate": 1.9169776238757254e-08,
      "loss": 0.4492,
      "step": 469690
    },
    {
      "epoch": 192.26,
      "grad_norm": 1.8367034196853638,
      "learning_rate": 1.9149534496457237e-08,
      "loss": 0.4482,
      "step": 469700
    },
    {
      "epoch": 192.27,
      "grad_norm": 1.970171570777893,
      "learning_rate": 1.9129303402791797e-08,
      "loss": 0.4428,
      "step": 469710
    },
    {
      "epoch": 192.27,
      "grad_norm": 2.3300838470458984,
      "learning_rate": 1.910908295785409e-08,
      "loss": 0.4575,
      "step": 469720
    },
    {
      "epoch": 192.28,
      "grad_norm": 2.2350921630859375,
      "learning_rate": 1.9088873161736232e-08,
      "loss": 0.4302,
      "step": 469730
    },
    {
      "epoch": 192.28,
      "grad_norm": 1.958794116973877,
      "learning_rate": 1.9068674014530856e-08,
      "loss": 0.4751,
      "step": 469740
    },
    {
      "epoch": 192.28,
      "grad_norm": 2.0148937702178955,
      "learning_rate": 1.9048485516330852e-08,
      "loss": 0.4527,
      "step": 469750
    },
    {
      "epoch": 192.29,
      "grad_norm": 2.08792781829834,
      "learning_rate": 1.902830766722834e-08,
      "loss": 0.451,
      "step": 469760
    },
    {
      "epoch": 192.29,
      "grad_norm": 1.903591513633728,
      "learning_rate": 1.9008140467315947e-08,
      "loss": 0.4504,
      "step": 469770
    },
    {
      "epoch": 192.3,
      "grad_norm": 2.071223258972168,
      "learning_rate": 1.8987983916685796e-08,
      "loss": 0.4206,
      "step": 469780
    },
    {
      "epoch": 192.3,
      "grad_norm": 2.08364200592041,
      "learning_rate": 1.8967838015430514e-08,
      "loss": 0.4291,
      "step": 469790
    },
    {
      "epoch": 192.3,
      "grad_norm": 2.145061731338501,
      "learning_rate": 1.8947702763642477e-08,
      "loss": 0.444,
      "step": 469800
    },
    {
      "epoch": 192.31,
      "grad_norm": 2.144421339035034,
      "learning_rate": 1.892757816141354e-08,
      "loss": 0.4346,
      "step": 469810
    },
    {
      "epoch": 192.31,
      "grad_norm": 1.9507008790969849,
      "learning_rate": 1.8907464208836077e-08,
      "loss": 0.4378,
      "step": 469820
    },
    {
      "epoch": 192.32,
      "grad_norm": 2.1635425090789795,
      "learning_rate": 1.88873609060022e-08,
      "loss": 0.4399,
      "step": 469830
    },
    {
      "epoch": 192.32,
      "grad_norm": 2.251152992248535,
      "learning_rate": 1.8867268253004025e-08,
      "loss": 0.4156,
      "step": 469840
    },
    {
      "epoch": 192.33,
      "grad_norm": 2.074939727783203,
      "learning_rate": 1.8847186249933664e-08,
      "loss": 0.4351,
      "step": 469850
    },
    {
      "epoch": 192.33,
      "grad_norm": 1.8310410976409912,
      "learning_rate": 1.882711489688297e-08,
      "loss": 0.4416,
      "step": 469860
    },
    {
      "epoch": 192.33,
      "grad_norm": 1.9966458082199097,
      "learning_rate": 1.8807054193943802e-08,
      "loss": 0.4433,
      "step": 469870
    },
    {
      "epoch": 192.34,
      "grad_norm": 1.915509819984436,
      "learning_rate": 1.8787004141208525e-08,
      "loss": 0.4377,
      "step": 469880
    },
    {
      "epoch": 192.34,
      "grad_norm": 1.8652334213256836,
      "learning_rate": 1.876696473876848e-08,
      "loss": 0.4527,
      "step": 469890
    },
    {
      "epoch": 192.35,
      "grad_norm": 2.2796261310577393,
      "learning_rate": 1.8746935986715518e-08,
      "loss": 0.4297,
      "step": 469900
    },
    {
      "epoch": 192.35,
      "grad_norm": 2.0146501064300537,
      "learning_rate": 1.872691788514175e-08,
      "loss": 0.4373,
      "step": 469910
    },
    {
      "epoch": 192.35,
      "grad_norm": 2.0320167541503906,
      "learning_rate": 1.8706910434138513e-08,
      "loss": 0.4411,
      "step": 469920
    },
    {
      "epoch": 192.36,
      "grad_norm": 2.231407403945923,
      "learning_rate": 1.868691363379766e-08,
      "loss": 0.4438,
      "step": 469930
    },
    {
      "epoch": 192.36,
      "grad_norm": 2.0064451694488525,
      "learning_rate": 1.866692748421078e-08,
      "loss": 0.45,
      "step": 469940
    },
    {
      "epoch": 192.37,
      "grad_norm": 2.114020347595215,
      "learning_rate": 1.8646951985469212e-08,
      "loss": 0.4579,
      "step": 469950
    },
    {
      "epoch": 192.37,
      "grad_norm": 1.9736950397491455,
      "learning_rate": 1.862698713766481e-08,
      "loss": 0.4416,
      "step": 469960
    },
    {
      "epoch": 192.37,
      "grad_norm": 1.7072668075561523,
      "learning_rate": 1.8607032940888902e-08,
      "loss": 0.4489,
      "step": 469970
    },
    {
      "epoch": 192.38,
      "grad_norm": 2.229616165161133,
      "learning_rate": 1.8587089395232566e-08,
      "loss": 0.4519,
      "step": 469980
    },
    {
      "epoch": 192.38,
      "grad_norm": 2.1690542697906494,
      "learning_rate": 1.856715650078765e-08,
      "loss": 0.4595,
      "step": 469990
    },
    {
      "epoch": 192.39,
      "grad_norm": 1.9898676872253418,
      "learning_rate": 1.8547234257645233e-08,
      "loss": 0.425,
      "step": 470000
    },
    {
      "epoch": 192.39,
      "grad_norm": 2.4058849811553955,
      "learning_rate": 1.8527322665896385e-08,
      "loss": 0.455,
      "step": 470010
    },
    {
      "epoch": 192.39,
      "grad_norm": 2.124800682067871,
      "learning_rate": 1.8507421725632698e-08,
      "loss": 0.449,
      "step": 470020
    },
    {
      "epoch": 192.4,
      "grad_norm": 1.9737547636032104,
      "learning_rate": 1.8487531436944728e-08,
      "loss": 0.4469,
      "step": 470030
    },
    {
      "epoch": 192.4,
      "grad_norm": 2.025707244873047,
      "learning_rate": 1.8467651799924586e-08,
      "loss": 0.4637,
      "step": 470040
    },
    {
      "epoch": 192.41,
      "grad_norm": 2.0307130813598633,
      "learning_rate": 1.8447782814662304e-08,
      "loss": 0.4373,
      "step": 470050
    },
    {
      "epoch": 192.41,
      "grad_norm": 2.121948480606079,
      "learning_rate": 1.8427924481249737e-08,
      "loss": 0.4428,
      "step": 470060
    },
    {
      "epoch": 192.42,
      "grad_norm": 1.830664038658142,
      "learning_rate": 1.840807679977718e-08,
      "loss": 0.4452,
      "step": 470070
    },
    {
      "epoch": 192.42,
      "grad_norm": 2.074579954147339,
      "learning_rate": 1.838823977033622e-08,
      "loss": 0.4288,
      "step": 470080
    },
    {
      "epoch": 192.42,
      "grad_norm": 2.485240936279297,
      "learning_rate": 1.8368413393017155e-08,
      "loss": 0.4526,
      "step": 470090
    },
    {
      "epoch": 192.43,
      "grad_norm": 1.8432691097259521,
      "learning_rate": 1.8348597667910796e-08,
      "loss": 0.4695,
      "step": 470100
    },
    {
      "epoch": 192.43,
      "grad_norm": 2.10094952583313,
      "learning_rate": 1.8328792595108476e-08,
      "loss": 0.4378,
      "step": 470110
    },
    {
      "epoch": 192.44,
      "grad_norm": 1.8299610614776611,
      "learning_rate": 1.8308998174700226e-08,
      "loss": 0.4633,
      "step": 470120
    },
    {
      "epoch": 192.44,
      "grad_norm": 2.174492359161377,
      "learning_rate": 1.828921440677738e-08,
      "loss": 0.4345,
      "step": 470130
    },
    {
      "epoch": 192.44,
      "grad_norm": 2.0557162761688232,
      "learning_rate": 1.826944129143023e-08,
      "loss": 0.4825,
      "step": 470140
    },
    {
      "epoch": 192.45,
      "grad_norm": 1.9928492307662964,
      "learning_rate": 1.824967882874933e-08,
      "loss": 0.4442,
      "step": 470150
    },
    {
      "epoch": 192.45,
      "grad_norm": 2.0786051750183105,
      "learning_rate": 1.822992701882497e-08,
      "loss": 0.4254,
      "step": 470160
    },
    {
      "epoch": 192.46,
      "grad_norm": 1.7227253913879395,
      "learning_rate": 1.8210185861748227e-08,
      "loss": 0.4267,
      "step": 470170
    },
    {
      "epoch": 192.46,
      "grad_norm": 1.6782795190811157,
      "learning_rate": 1.819045535760913e-08,
      "loss": 0.4606,
      "step": 470180
    },
    {
      "epoch": 192.46,
      "grad_norm": 1.8155933618545532,
      "learning_rate": 1.817073550649823e-08,
      "loss": 0.4398,
      "step": 470190
    },
    {
      "epoch": 192.47,
      "grad_norm": 2.1657416820526123,
      "learning_rate": 1.8151026308505562e-08,
      "loss": 0.4343,
      "step": 470200
    },
    {
      "epoch": 192.47,
      "grad_norm": 2.1372056007385254,
      "learning_rate": 1.8131327763721682e-08,
      "loss": 0.452,
      "step": 470210
    },
    {
      "epoch": 192.48,
      "grad_norm": 2.1855733394622803,
      "learning_rate": 1.8111639872236874e-08,
      "loss": 0.4317,
      "step": 470220
    },
    {
      "epoch": 192.48,
      "grad_norm": 2.094284772872925,
      "learning_rate": 1.8091962634141174e-08,
      "loss": 0.4412,
      "step": 470230
    },
    {
      "epoch": 192.48,
      "grad_norm": 2.476317882537842,
      "learning_rate": 1.8072296049524878e-08,
      "loss": 0.4506,
      "step": 470240
    },
    {
      "epoch": 192.49,
      "grad_norm": 2.064997434616089,
      "learning_rate": 1.805264011847775e-08,
      "loss": 0.4704,
      "step": 470250
    },
    {
      "epoch": 192.49,
      "grad_norm": 1.8783293962478638,
      "learning_rate": 1.8032994841090093e-08,
      "loss": 0.4528,
      "step": 470260
    },
    {
      "epoch": 192.5,
      "grad_norm": 2.148850917816162,
      "learning_rate": 1.801336021745219e-08,
      "loss": 0.4528,
      "step": 470270
    },
    {
      "epoch": 192.5,
      "grad_norm": 1.9590250253677368,
      "learning_rate": 1.7993736247653297e-08,
      "loss": 0.4301,
      "step": 470280
    },
    {
      "epoch": 192.51,
      "grad_norm": 2.1985490322113037,
      "learning_rate": 1.7974122931783704e-08,
      "loss": 0.4397,
      "step": 470290
    },
    {
      "epoch": 192.51,
      "grad_norm": 2.0024306774139404,
      "learning_rate": 1.7954520269933442e-08,
      "loss": 0.4333,
      "step": 470300
    },
    {
      "epoch": 192.51,
      "grad_norm": 1.8050198554992676,
      "learning_rate": 1.7934928262191768e-08,
      "loss": 0.4457,
      "step": 470310
    },
    {
      "epoch": 192.52,
      "grad_norm": 2.013362169265747,
      "learning_rate": 1.7915346908648972e-08,
      "loss": 0.4736,
      "step": 470320
    },
    {
      "epoch": 192.52,
      "grad_norm": 1.79959237575531,
      "learning_rate": 1.7895776209394305e-08,
      "loss": 0.4376,
      "step": 470330
    },
    {
      "epoch": 192.53,
      "grad_norm": 2.340972900390625,
      "learning_rate": 1.7876216164517797e-08,
      "loss": 0.4401,
      "step": 470340
    },
    {
      "epoch": 192.53,
      "grad_norm": 1.8732540607452393,
      "learning_rate": 1.7856666774108964e-08,
      "loss": 0.4423,
      "step": 470350
    },
    {
      "epoch": 192.53,
      "grad_norm": 1.5475846529006958,
      "learning_rate": 1.7837128038257054e-08,
      "loss": 0.4366,
      "step": 470360
    },
    {
      "epoch": 192.54,
      "grad_norm": 1.9426382780075073,
      "learning_rate": 1.781759995705184e-08,
      "loss": 0.4436,
      "step": 470370
    },
    {
      "epoch": 192.54,
      "grad_norm": 2.03948974609375,
      "learning_rate": 1.7798082530582832e-08,
      "loss": 0.4269,
      "step": 470380
    },
    {
      "epoch": 192.55,
      "grad_norm": 2.0527772903442383,
      "learning_rate": 1.7778575758939287e-08,
      "loss": 0.4391,
      "step": 470390
    },
    {
      "epoch": 192.55,
      "grad_norm": 1.9024865627288818,
      "learning_rate": 1.775907964221071e-08,
      "loss": 0.4474,
      "step": 470400
    },
    {
      "epoch": 192.55,
      "grad_norm": 2.3176045417785645,
      "learning_rate": 1.7739594180486096e-08,
      "loss": 0.4516,
      "step": 470410
    },
    {
      "epoch": 192.56,
      "grad_norm": 2.279677391052246,
      "learning_rate": 1.772011937385522e-08,
      "loss": 0.4476,
      "step": 470420
    },
    {
      "epoch": 192.56,
      "grad_norm": 1.8363392353057861,
      "learning_rate": 1.7700655222406807e-08,
      "loss": 0.4429,
      "step": 470430
    },
    {
      "epoch": 192.57,
      "grad_norm": 2.057710886001587,
      "learning_rate": 1.768120172623011e-08,
      "loss": 0.4291,
      "step": 470440
    },
    {
      "epoch": 192.57,
      "grad_norm": 1.9151493310928345,
      "learning_rate": 1.7661758885414642e-08,
      "loss": 0.4294,
      "step": 470450
    },
    {
      "epoch": 192.57,
      "grad_norm": 1.777237892150879,
      "learning_rate": 1.7642326700048876e-08,
      "loss": 0.4422,
      "step": 470460
    },
    {
      "epoch": 192.58,
      "grad_norm": 2.4402801990509033,
      "learning_rate": 1.762290517022232e-08,
      "loss": 0.4337,
      "step": 470470
    },
    {
      "epoch": 192.58,
      "grad_norm": 1.9644883871078491,
      "learning_rate": 1.7603494296023706e-08,
      "loss": 0.4363,
      "step": 470480
    },
    {
      "epoch": 192.59,
      "grad_norm": 2.154773712158203,
      "learning_rate": 1.7584094077541764e-08,
      "loss": 0.4582,
      "step": 470490
    },
    {
      "epoch": 192.59,
      "grad_norm": 1.955637812614441,
      "learning_rate": 1.7564704514865748e-08,
      "loss": 0.4556,
      "step": 470500
    },
    {
      "epoch": 192.6,
      "grad_norm": 1.5636957883834839,
      "learning_rate": 1.754532560808439e-08,
      "loss": 0.4434,
      "step": 470510
    },
    {
      "epoch": 192.6,
      "grad_norm": 2.8317782878875732,
      "learning_rate": 1.752595735728642e-08,
      "loss": 0.4506,
      "step": 470520
    },
    {
      "epoch": 192.6,
      "grad_norm": 1.7500079870224,
      "learning_rate": 1.7506599762560307e-08,
      "loss": 0.4314,
      "step": 470530
    },
    {
      "epoch": 192.61,
      "grad_norm": 1.85174560546875,
      "learning_rate": 1.748725282399504e-08,
      "loss": 0.434,
      "step": 470540
    },
    {
      "epoch": 192.61,
      "grad_norm": 1.9594624042510986,
      "learning_rate": 1.746791654167936e-08,
      "loss": 0.4434,
      "step": 470550
    },
    {
      "epoch": 192.62,
      "grad_norm": 1.7281609773635864,
      "learning_rate": 1.7448590915701208e-08,
      "loss": 0.4425,
      "step": 470560
    },
    {
      "epoch": 192.62,
      "grad_norm": 2.206444025039673,
      "learning_rate": 1.742927594614984e-08,
      "loss": 0.4427,
      "step": 470570
    },
    {
      "epoch": 192.62,
      "grad_norm": 2.4103970527648926,
      "learning_rate": 1.7409971633113462e-08,
      "loss": 0.4439,
      "step": 470580
    },
    {
      "epoch": 192.63,
      "grad_norm": 2.202198028564453,
      "learning_rate": 1.739067797668029e-08,
      "loss": 0.4353,
      "step": 470590
    },
    {
      "epoch": 192.63,
      "grad_norm": 2.5001883506774902,
      "learning_rate": 1.7371394976939056e-08,
      "loss": 0.4394,
      "step": 470600
    },
    {
      "epoch": 192.64,
      "grad_norm": 1.7520215511322021,
      "learning_rate": 1.7352122633977964e-08,
      "loss": 0.4297,
      "step": 470610
    },
    {
      "epoch": 192.64,
      "grad_norm": 1.8604835271835327,
      "learning_rate": 1.733286094788497e-08,
      "loss": 0.4265,
      "step": 470620
    },
    {
      "epoch": 192.64,
      "grad_norm": 1.8180322647094727,
      "learning_rate": 1.73136099187488e-08,
      "loss": 0.4694,
      "step": 470630
    },
    {
      "epoch": 192.65,
      "grad_norm": 1.7284334897994995,
      "learning_rate": 1.7294369546657413e-08,
      "loss": 0.4442,
      "step": 470640
    },
    {
      "epoch": 192.65,
      "grad_norm": 1.8684817552566528,
      "learning_rate": 1.7275139831699013e-08,
      "loss": 0.4473,
      "step": 470650
    },
    {
      "epoch": 192.66,
      "grad_norm": 2.1568515300750732,
      "learning_rate": 1.725592077396155e-08,
      "loss": 0.4483,
      "step": 470660
    },
    {
      "epoch": 192.66,
      "grad_norm": 1.9321227073669434,
      "learning_rate": 1.723671237353324e-08,
      "loss": 0.4506,
      "step": 470670
    },
    {
      "epoch": 192.66,
      "grad_norm": 1.8726040124893188,
      "learning_rate": 1.7217514630502027e-08,
      "loss": 0.4546,
      "step": 470680
    },
    {
      "epoch": 192.67,
      "grad_norm": 1.8339155912399292,
      "learning_rate": 1.7198327544955866e-08,
      "loss": 0.4202,
      "step": 470690
    },
    {
      "epoch": 192.67,
      "grad_norm": 2.026524066925049,
      "learning_rate": 1.7179151116982446e-08,
      "loss": 0.4652,
      "step": 470700
    },
    {
      "epoch": 192.68,
      "grad_norm": 2.4830315113067627,
      "learning_rate": 1.715998534666998e-08,
      "loss": 0.4432,
      "step": 470710
    },
    {
      "epoch": 192.68,
      "grad_norm": 2.0918142795562744,
      "learning_rate": 1.714083023410589e-08,
      "loss": 0.4249,
      "step": 470720
    },
    {
      "epoch": 192.69,
      "grad_norm": 2.219944477081299,
      "learning_rate": 1.7121685779378138e-08,
      "loss": 0.4266,
      "step": 470730
    },
    {
      "epoch": 192.69,
      "grad_norm": 2.9532315731048584,
      "learning_rate": 1.7102551982574405e-08,
      "loss": 0.442,
      "step": 470740
    },
    {
      "epoch": 192.69,
      "grad_norm": 2.1185739040374756,
      "learning_rate": 1.7083428843782126e-08,
      "loss": 0.4397,
      "step": 470750
    },
    {
      "epoch": 192.7,
      "grad_norm": 2.302842855453491,
      "learning_rate": 1.706431636308925e-08,
      "loss": 0.4396,
      "step": 470760
    },
    {
      "epoch": 192.7,
      "grad_norm": 2.2905261516571045,
      "learning_rate": 1.704521454058295e-08,
      "loss": 0.4436,
      "step": 470770
    },
    {
      "epoch": 192.71,
      "grad_norm": 2.1774308681488037,
      "learning_rate": 1.7026123376351167e-08,
      "loss": 0.4285,
      "step": 470780
    },
    {
      "epoch": 192.71,
      "grad_norm": 1.9969861507415771,
      "learning_rate": 1.7007042870480823e-08,
      "loss": 0.4558,
      "step": 470790
    },
    {
      "epoch": 192.71,
      "grad_norm": 1.8660223484039307,
      "learning_rate": 1.6987973023059858e-08,
      "loss": 0.4348,
      "step": 470800
    },
    {
      "epoch": 192.72,
      "grad_norm": 2.113480567932129,
      "learning_rate": 1.6968913834175447e-08,
      "loss": 0.4186,
      "step": 470810
    },
    {
      "epoch": 192.72,
      "grad_norm": 2.302851438522339,
      "learning_rate": 1.6949865303914763e-08,
      "loss": 0.4548,
      "step": 470820
    },
    {
      "epoch": 192.73,
      "grad_norm": 2.1936116218566895,
      "learning_rate": 1.6930827432365227e-08,
      "loss": 0.4261,
      "step": 470830
    },
    {
      "epoch": 192.73,
      "grad_norm": 2.26833176612854,
      "learning_rate": 1.6911800219614016e-08,
      "loss": 0.4436,
      "step": 470840
    },
    {
      "epoch": 192.73,
      "grad_norm": 2.1574931144714355,
      "learning_rate": 1.689278366574804e-08,
      "loss": 0.4317,
      "step": 470850
    },
    {
      "epoch": 192.74,
      "grad_norm": 1.8393316268920898,
      "learning_rate": 1.6873777770854726e-08,
      "loss": 0.4485,
      "step": 470860
    },
    {
      "epoch": 192.74,
      "grad_norm": 1.8651131391525269,
      "learning_rate": 1.6854782535020984e-08,
      "loss": 0.4352,
      "step": 470870
    },
    {
      "epoch": 192.75,
      "grad_norm": 1.9561773538589478,
      "learning_rate": 1.6835797958333984e-08,
      "loss": 0.4379,
      "step": 470880
    },
    {
      "epoch": 192.75,
      "grad_norm": 2.356968402862549,
      "learning_rate": 1.6816824040880375e-08,
      "loss": 0.4291,
      "step": 470890
    },
    {
      "epoch": 192.75,
      "grad_norm": 2.1094157695770264,
      "learning_rate": 1.679786078274733e-08,
      "loss": 0.4277,
      "step": 470900
    },
    {
      "epoch": 192.76,
      "grad_norm": 1.949410080909729,
      "learning_rate": 1.6778908184021754e-08,
      "loss": 0.4416,
      "step": 470910
    },
    {
      "epoch": 192.76,
      "grad_norm": 1.9273858070373535,
      "learning_rate": 1.67599662447903e-08,
      "loss": 0.4495,
      "step": 470920
    },
    {
      "epoch": 192.77,
      "grad_norm": 2.0165746212005615,
      "learning_rate": 1.6741034965139875e-08,
      "loss": 0.4378,
      "step": 470930
    },
    {
      "epoch": 192.77,
      "grad_norm": 2.1279547214508057,
      "learning_rate": 1.672211434515713e-08,
      "loss": 0.4306,
      "step": 470940
    },
    {
      "epoch": 192.78,
      "grad_norm": 1.7258338928222656,
      "learning_rate": 1.6703204384928716e-08,
      "loss": 0.4449,
      "step": 470950
    },
    {
      "epoch": 192.78,
      "grad_norm": 2.120943069458008,
      "learning_rate": 1.668430508454128e-08,
      "loss": 0.45,
      "step": 470960
    },
    {
      "epoch": 192.78,
      "grad_norm": 2.4917349815368652,
      "learning_rate": 1.666541644408121e-08,
      "loss": 0.4388,
      "step": 470970
    },
    {
      "epoch": 192.79,
      "grad_norm": 2.0805749893188477,
      "learning_rate": 1.6646538463635678e-08,
      "loss": 0.4499,
      "step": 470980
    },
    {
      "epoch": 192.79,
      "grad_norm": 2.3001017570495605,
      "learning_rate": 1.6627671143290293e-08,
      "loss": 0.4549,
      "step": 470990
    },
    {
      "epoch": 192.8,
      "grad_norm": 2.00148344039917,
      "learning_rate": 1.6608814483131966e-08,
      "loss": 0.4505,
      "step": 471000
    },
    {
      "epoch": 192.8,
      "grad_norm": 1.9624316692352295,
      "learning_rate": 1.6589968483247084e-08,
      "loss": 0.4663,
      "step": 471010
    },
    {
      "epoch": 192.8,
      "grad_norm": 1.9387277364730835,
      "learning_rate": 1.6571133143722035e-08,
      "loss": 0.4551,
      "step": 471020
    },
    {
      "epoch": 192.81,
      "grad_norm": 4.078879356384277,
      "learning_rate": 1.655230846464269e-08,
      "loss": 0.4361,
      "step": 471030
    },
    {
      "epoch": 192.81,
      "grad_norm": 1.8010921478271484,
      "learning_rate": 1.6533494446095697e-08,
      "loss": 0.4236,
      "step": 471040
    },
    {
      "epoch": 192.82,
      "grad_norm": 2.3996336460113525,
      "learning_rate": 1.6514691088167186e-08,
      "loss": 0.4579,
      "step": 471050
    },
    {
      "epoch": 192.82,
      "grad_norm": 2.2810423374176025,
      "learning_rate": 1.6495898390943284e-08,
      "loss": 0.4411,
      "step": 471060
    },
    {
      "epoch": 192.82,
      "grad_norm": 2.0263397693634033,
      "learning_rate": 1.6477116354509606e-08,
      "loss": 0.4324,
      "step": 471070
    },
    {
      "epoch": 192.83,
      "grad_norm": 2.4773552417755127,
      "learning_rate": 1.6458344978953053e-08,
      "loss": 0.465,
      "step": 471080
    },
    {
      "epoch": 192.83,
      "grad_norm": 2.2103278636932373,
      "learning_rate": 1.6439584264358718e-08,
      "loss": 0.4349,
      "step": 471090
    },
    {
      "epoch": 192.84,
      "grad_norm": 2.45493745803833,
      "learning_rate": 1.642083421081325e-08,
      "loss": 0.4421,
      "step": 471100
    },
    {
      "epoch": 192.84,
      "grad_norm": 2.4755308628082275,
      "learning_rate": 1.6402094818402258e-08,
      "loss": 0.4298,
      "step": 471110
    },
    {
      "epoch": 192.84,
      "grad_norm": 2.1860344409942627,
      "learning_rate": 1.638336608721135e-08,
      "loss": 0.4481,
      "step": 471120
    },
    {
      "epoch": 192.85,
      "grad_norm": 1.8336706161499023,
      "learning_rate": 1.6364648017326653e-08,
      "loss": 0.4393,
      "step": 471130
    },
    {
      "epoch": 192.85,
      "grad_norm": 1.8617916107177734,
      "learning_rate": 1.6345940608834038e-08,
      "loss": 0.4618,
      "step": 471140
    },
    {
      "epoch": 192.86,
      "grad_norm": 1.9078150987625122,
      "learning_rate": 1.632724386181859e-08,
      "loss": 0.4504,
      "step": 471150
    },
    {
      "epoch": 192.86,
      "grad_norm": 2.2194979190826416,
      "learning_rate": 1.6308557776366707e-08,
      "loss": 0.4484,
      "step": 471160
    },
    {
      "epoch": 192.87,
      "grad_norm": 1.9583884477615356,
      "learning_rate": 1.628988235256347e-08,
      "loss": 0.4478,
      "step": 471170
    },
    {
      "epoch": 192.87,
      "grad_norm": 2.125293254852295,
      "learning_rate": 1.6271217590494488e-08,
      "loss": 0.431,
      "step": 471180
    },
    {
      "epoch": 192.87,
      "grad_norm": 2.334643602371216,
      "learning_rate": 1.6252563490245368e-08,
      "loss": 0.4478,
      "step": 471190
    },
    {
      "epoch": 192.88,
      "grad_norm": 2.1477408409118652,
      "learning_rate": 1.6233920051901725e-08,
      "loss": 0.4481,
      "step": 471200
    },
    {
      "epoch": 192.88,
      "grad_norm": 2.368583917617798,
      "learning_rate": 1.621528727554864e-08,
      "loss": 0.4148,
      "step": 471210
    },
    {
      "epoch": 192.89,
      "grad_norm": 2.44258189201355,
      "learning_rate": 1.6196665161271727e-08,
      "loss": 0.4327,
      "step": 471220
    },
    {
      "epoch": 192.89,
      "grad_norm": 2.030874252319336,
      "learning_rate": 1.6178053709156072e-08,
      "loss": 0.4342,
      "step": 471230
    },
    {
      "epoch": 192.89,
      "grad_norm": 1.784454345703125,
      "learning_rate": 1.6159452919287023e-08,
      "loss": 0.436,
      "step": 471240
    },
    {
      "epoch": 192.9,
      "grad_norm": 2.0726308822631836,
      "learning_rate": 1.6140862791749666e-08,
      "loss": 0.4506,
      "step": 471250
    },
    {
      "epoch": 192.9,
      "grad_norm": 2.00514817237854,
      "learning_rate": 1.6122283326629617e-08,
      "loss": 0.4579,
      "step": 471260
    },
    {
      "epoch": 192.91,
      "grad_norm": 1.935181975364685,
      "learning_rate": 1.6103714524011435e-08,
      "loss": 0.4259,
      "step": 471270
    },
    {
      "epoch": 192.91,
      "grad_norm": 2.4436914920806885,
      "learning_rate": 1.6085156383980474e-08,
      "loss": 0.4714,
      "step": 471280
    },
    {
      "epoch": 192.91,
      "grad_norm": 2.141045093536377,
      "learning_rate": 1.606660890662182e-08,
      "loss": 0.4433,
      "step": 471290
    },
    {
      "epoch": 192.92,
      "grad_norm": 1.7207839488983154,
      "learning_rate": 1.6048072092020044e-08,
      "loss": 0.4419,
      "step": 471300
    },
    {
      "epoch": 192.92,
      "grad_norm": 1.8840012550354004,
      "learning_rate": 1.602954594026049e-08,
      "loss": 0.4339,
      "step": 471310
    },
    {
      "epoch": 192.93,
      "grad_norm": 1.85657799243927,
      "learning_rate": 1.6011030451427725e-08,
      "loss": 0.4217,
      "step": 471320
    },
    {
      "epoch": 192.93,
      "grad_norm": 2.329318046569824,
      "learning_rate": 1.5992525625606584e-08,
      "loss": 0.4336,
      "step": 471330
    },
    {
      "epoch": 192.93,
      "grad_norm": 2.007192611694336,
      "learning_rate": 1.597403146288215e-08,
      "loss": 0.4398,
      "step": 471340
    },
    {
      "epoch": 192.94,
      "grad_norm": 1.9801591634750366,
      "learning_rate": 1.595554796333873e-08,
      "loss": 0.4718,
      "step": 471350
    },
    {
      "epoch": 192.94,
      "grad_norm": 1.9360398054122925,
      "learning_rate": 1.593707512706141e-08,
      "loss": 0.4473,
      "step": 471360
    },
    {
      "epoch": 192.95,
      "grad_norm": 2.142504930496216,
      "learning_rate": 1.5918612954134503e-08,
      "loss": 0.4547,
      "step": 471370
    },
    {
      "epoch": 192.95,
      "grad_norm": 2.05367112159729,
      "learning_rate": 1.590016144464258e-08,
      "loss": 0.4552,
      "step": 471380
    },
    {
      "epoch": 192.96,
      "grad_norm": 2.0767581462860107,
      "learning_rate": 1.5881720598670198e-08,
      "loss": 0.4336,
      "step": 471390
    },
    {
      "epoch": 192.96,
      "grad_norm": 2.7021560668945312,
      "learning_rate": 1.5863290416302192e-08,
      "loss": 0.4421,
      "step": 471400
    },
    {
      "epoch": 192.96,
      "grad_norm": 1.9089455604553223,
      "learning_rate": 1.5844870897622346e-08,
      "loss": 0.4521,
      "step": 471410
    },
    {
      "epoch": 192.97,
      "grad_norm": 2.654949903488159,
      "learning_rate": 1.582646204271549e-08,
      "loss": 0.4574,
      "step": 471420
    },
    {
      "epoch": 192.97,
      "grad_norm": 2.3370773792266846,
      "learning_rate": 1.5808063851666196e-08,
      "loss": 0.4404,
      "step": 471430
    },
    {
      "epoch": 192.98,
      "grad_norm": 2.6287150382995605,
      "learning_rate": 1.5789676324557982e-08,
      "loss": 0.4389,
      "step": 471440
    },
    {
      "epoch": 192.98,
      "grad_norm": 2.4489552974700928,
      "learning_rate": 1.577129946147568e-08,
      "loss": 0.4233,
      "step": 471450
    },
    {
      "epoch": 192.98,
      "grad_norm": 1.8228195905685425,
      "learning_rate": 1.575293326250308e-08,
      "loss": 0.4389,
      "step": 471460
    },
    {
      "epoch": 192.99,
      "grad_norm": 1.7386093139648438,
      "learning_rate": 1.5734577727724488e-08,
      "loss": 0.4385,
      "step": 471470
    },
    {
      "epoch": 192.99,
      "grad_norm": 1.65558922290802,
      "learning_rate": 1.5716232857224207e-08,
      "loss": 0.4445,
      "step": 471480
    },
    {
      "epoch": 193.0,
      "grad_norm": 2.057535409927368,
      "learning_rate": 1.569789865108603e-08,
      "loss": 0.4557,
      "step": 471490
    },
    {
      "epoch": 193.0,
      "eval_loss": 0.4436228275299072,
      "eval_runtime": 43.1453,
      "eval_samples_per_second": 79.939,
      "eval_steps_per_second": 10.013,
      "step": 471499
    },
    {
      "epoch": 193.0,
      "grad_norm": 2.055454730987549,
      "learning_rate": 1.567957510939374e-08,
      "loss": 0.4393,
      "step": 471500
    },
    {
      "epoch": 193.0,
      "grad_norm": 2.358619213104248,
      "learning_rate": 1.5661262232231908e-08,
      "loss": 0.4349,
      "step": 471510
    },
    {
      "epoch": 193.01,
      "grad_norm": 3.955352306365967,
      "learning_rate": 1.56429600196838e-08,
      "loss": 0.4414,
      "step": 471520
    },
    {
      "epoch": 193.01,
      "grad_norm": 1.9592301845550537,
      "learning_rate": 1.5624668471833462e-08,
      "loss": 0.4421,
      "step": 471530
    },
    {
      "epoch": 193.02,
      "grad_norm": 1.6706053018569946,
      "learning_rate": 1.5606387588764682e-08,
      "loss": 0.436,
      "step": 471540
    },
    {
      "epoch": 193.02,
      "grad_norm": 2.1802971363067627,
      "learning_rate": 1.5588117370561246e-08,
      "loss": 0.4517,
      "step": 471550
    },
    {
      "epoch": 193.02,
      "grad_norm": 1.9444133043289185,
      "learning_rate": 1.5569857817306942e-08,
      "loss": 0.4373,
      "step": 471560
    },
    {
      "epoch": 193.03,
      "grad_norm": 1.9212584495544434,
      "learning_rate": 1.5551608929085297e-08,
      "loss": 0.4587,
      "step": 471570
    },
    {
      "epoch": 193.03,
      "grad_norm": 1.953393578529358,
      "learning_rate": 1.55333707059801e-08,
      "loss": 0.4308,
      "step": 471580
    },
    {
      "epoch": 193.04,
      "grad_norm": 2.026636838912964,
      "learning_rate": 1.5515143148074355e-08,
      "loss": 0.4566,
      "step": 471590
    },
    {
      "epoch": 193.04,
      "grad_norm": 2.2677087783813477,
      "learning_rate": 1.549692625545211e-08,
      "loss": 0.4455,
      "step": 471600
    },
    {
      "epoch": 193.05,
      "grad_norm": 2.0853023529052734,
      "learning_rate": 1.5478720028196634e-08,
      "loss": 0.4462,
      "step": 471610
    },
    {
      "epoch": 193.05,
      "grad_norm": 2.1737334728240967,
      "learning_rate": 1.546052446639119e-08,
      "loss": 0.4371,
      "step": 471620
    },
    {
      "epoch": 193.05,
      "grad_norm": 1.5944850444793701,
      "learning_rate": 1.5442339570119304e-08,
      "loss": 0.4288,
      "step": 471630
    },
    {
      "epoch": 193.06,
      "grad_norm": 2.399214744567871,
      "learning_rate": 1.5424165339464245e-08,
      "loss": 0.4453,
      "step": 471640
    },
    {
      "epoch": 193.06,
      "grad_norm": 1.7931314706802368,
      "learning_rate": 1.5406001774509284e-08,
      "loss": 0.4507,
      "step": 471650
    },
    {
      "epoch": 193.07,
      "grad_norm": 2.25285267829895,
      "learning_rate": 1.5387848875337423e-08,
      "loss": 0.4265,
      "step": 471660
    },
    {
      "epoch": 193.07,
      "grad_norm": 1.9950993061065674,
      "learning_rate": 1.5369706642032187e-08,
      "loss": 0.4579,
      "step": 471670
    },
    {
      "epoch": 193.07,
      "grad_norm": 2.330089569091797,
      "learning_rate": 1.5351575074676325e-08,
      "loss": 0.4468,
      "step": 471680
    },
    {
      "epoch": 193.08,
      "grad_norm": 3.3640806674957275,
      "learning_rate": 1.5333454173353364e-08,
      "loss": 0.4385,
      "step": 471690
    },
    {
      "epoch": 193.08,
      "grad_norm": 1.9662425518035889,
      "learning_rate": 1.531534393814553e-08,
      "loss": 0.4497,
      "step": 471700
    },
    {
      "epoch": 193.09,
      "grad_norm": 2.1003901958465576,
      "learning_rate": 1.529724436913661e-08,
      "loss": 0.4405,
      "step": 471710
    },
    {
      "epoch": 193.09,
      "grad_norm": 1.8433493375778198,
      "learning_rate": 1.527915546640909e-08,
      "loss": 0.4496,
      "step": 471720
    },
    {
      "epoch": 193.09,
      "grad_norm": 2.3903896808624268,
      "learning_rate": 1.5261077230045716e-08,
      "loss": 0.4403,
      "step": 471730
    },
    {
      "epoch": 193.1,
      "grad_norm": 2.6374642848968506,
      "learning_rate": 1.5243009660129753e-08,
      "loss": 0.466,
      "step": 471740
    },
    {
      "epoch": 193.1,
      "grad_norm": 1.88126540184021,
      "learning_rate": 1.522495275674343e-08,
      "loss": 0.4578,
      "step": 471750
    },
    {
      "epoch": 193.11,
      "grad_norm": 7.239246368408203,
      "learning_rate": 1.520690651996975e-08,
      "loss": 0.4317,
      "step": 471760
    },
    {
      "epoch": 193.11,
      "grad_norm": 1.887335181236267,
      "learning_rate": 1.5188870949891466e-08,
      "loss": 0.4557,
      "step": 471770
    },
    {
      "epoch": 193.12,
      "grad_norm": 5.7130608558654785,
      "learning_rate": 1.5170846046591316e-08,
      "loss": 0.4421,
      "step": 471780
    },
    {
      "epoch": 193.12,
      "grad_norm": 2.0805139541625977,
      "learning_rate": 1.5152831810151272e-08,
      "loss": 0.4325,
      "step": 471790
    },
    {
      "epoch": 193.12,
      "grad_norm": 1.9149562120437622,
      "learning_rate": 1.513482824065434e-08,
      "loss": 0.4476,
      "step": 471800
    },
    {
      "epoch": 193.13,
      "grad_norm": 2.363349676132202,
      "learning_rate": 1.5116835338183e-08,
      "loss": 0.4606,
      "step": 471810
    },
    {
      "epoch": 193.13,
      "grad_norm": 2.205537796020508,
      "learning_rate": 1.5098853102819484e-08,
      "loss": 0.4446,
      "step": 471820
    },
    {
      "epoch": 193.14,
      "grad_norm": 2.0908355712890625,
      "learning_rate": 1.5080881534646276e-08,
      "loss": 0.42,
      "step": 471830
    },
    {
      "epoch": 193.14,
      "grad_norm": 2.063568592071533,
      "learning_rate": 1.50629206337456e-08,
      "loss": 0.4364,
      "step": 471840
    },
    {
      "epoch": 193.14,
      "grad_norm": 1.9852910041809082,
      "learning_rate": 1.5044970400199685e-08,
      "loss": 0.4612,
      "step": 471850
    },
    {
      "epoch": 193.15,
      "grad_norm": 1.910294532775879,
      "learning_rate": 1.5027030834091017e-08,
      "loss": 0.4504,
      "step": 471860
    },
    {
      "epoch": 193.15,
      "grad_norm": 2.871124029159546,
      "learning_rate": 1.500910193550156e-08,
      "loss": 0.4378,
      "step": 471870
    },
    {
      "epoch": 193.16,
      "grad_norm": 2.5240871906280518,
      "learning_rate": 1.4991183704513543e-08,
      "loss": 0.442,
      "step": 471880
    },
    {
      "epoch": 193.16,
      "grad_norm": 1.9282341003417969,
      "learning_rate": 1.4973276141209187e-08,
      "loss": 0.4291,
      "step": 471890
    },
    {
      "epoch": 193.16,
      "grad_norm": 2.004404067993164,
      "learning_rate": 1.4955379245669938e-08,
      "loss": 0.4594,
      "step": 471900
    },
    {
      "epoch": 193.17,
      "grad_norm": 1.8009381294250488,
      "learning_rate": 1.4937493017978286e-08,
      "loss": 0.4431,
      "step": 471910
    },
    {
      "epoch": 193.17,
      "grad_norm": 2.154759168624878,
      "learning_rate": 1.4919617458216195e-08,
      "loss": 0.4463,
      "step": 471920
    },
    {
      "epoch": 193.18,
      "grad_norm": 1.901341199874878,
      "learning_rate": 1.4901752566465632e-08,
      "loss": 0.4419,
      "step": 471930
    },
    {
      "epoch": 193.18,
      "grad_norm": 1.760987401008606,
      "learning_rate": 1.4883898342807776e-08,
      "loss": 0.4219,
      "step": 471940
    },
    {
      "epoch": 193.18,
      "grad_norm": 2.0375730991363525,
      "learning_rate": 1.4866054787325121e-08,
      "loss": 0.4345,
      "step": 471950
    },
    {
      "epoch": 193.19,
      "grad_norm": 1.7579987049102783,
      "learning_rate": 1.4848221900099107e-08,
      "loss": 0.4362,
      "step": 471960
    },
    {
      "epoch": 193.19,
      "grad_norm": 1.9324243068695068,
      "learning_rate": 1.483039968121144e-08,
      "loss": 0.4372,
      "step": 471970
    },
    {
      "epoch": 193.2,
      "grad_norm": 2.84153413772583,
      "learning_rate": 1.481258813074383e-08,
      "loss": 0.4435,
      "step": 471980
    },
    {
      "epoch": 193.2,
      "grad_norm": 2.366455316543579,
      "learning_rate": 1.4794787248777977e-08,
      "loss": 0.4511,
      "step": 471990
    },
    {
      "epoch": 193.21,
      "grad_norm": 1.7166138887405396,
      "learning_rate": 1.4776997035395068e-08,
      "loss": 0.4428,
      "step": 472000
    },
    {
      "epoch": 193.21,
      "grad_norm": 1.9001188278198242,
      "learning_rate": 1.4759217490676809e-08,
      "loss": 0.4346,
      "step": 472010
    },
    {
      "epoch": 193.21,
      "grad_norm": 2.0876624584198,
      "learning_rate": 1.4741448614704644e-08,
      "loss": 0.4233,
      "step": 472020
    },
    {
      "epoch": 193.22,
      "grad_norm": 2.020998239517212,
      "learning_rate": 1.472369040756002e-08,
      "loss": 0.4219,
      "step": 472030
    },
    {
      "epoch": 193.22,
      "grad_norm": 4.524340629577637,
      "learning_rate": 1.4705942869324122e-08,
      "loss": 0.4216,
      "step": 472040
    },
    {
      "epoch": 193.23,
      "grad_norm": 2.213146924972534,
      "learning_rate": 1.4688206000078394e-08,
      "loss": 0.4674,
      "step": 472050
    },
    {
      "epoch": 193.23,
      "grad_norm": 1.9543883800506592,
      "learning_rate": 1.467047979990402e-08,
      "loss": 0.4305,
      "step": 472060
    },
    {
      "epoch": 193.23,
      "grad_norm": 2.0276200771331787,
      "learning_rate": 1.4652764268882448e-08,
      "loss": 0.4357,
      "step": 472070
    },
    {
      "epoch": 193.24,
      "grad_norm": 2.4077236652374268,
      "learning_rate": 1.4635059407094342e-08,
      "loss": 0.4451,
      "step": 472080
    },
    {
      "epoch": 193.24,
      "grad_norm": 2.324709177017212,
      "learning_rate": 1.4617365214621147e-08,
      "loss": 0.4515,
      "step": 472090
    },
    {
      "epoch": 193.25,
      "grad_norm": 1.9157354831695557,
      "learning_rate": 1.4599681691543787e-08,
      "loss": 0.4438,
      "step": 472100
    },
    {
      "epoch": 193.25,
      "grad_norm": 1.9874902963638306,
      "learning_rate": 1.4582008837943186e-08,
      "loss": 0.4658,
      "step": 472110
    },
    {
      "epoch": 193.25,
      "grad_norm": 3.524466037750244,
      "learning_rate": 1.4564346653900793e-08,
      "loss": 0.4439,
      "step": 472120
    },
    {
      "epoch": 193.26,
      "grad_norm": 2.253922700881958,
      "learning_rate": 1.4546695139497009e-08,
      "loss": 0.4414,
      "step": 472130
    },
    {
      "epoch": 193.26,
      "grad_norm": 2.0329954624176025,
      "learning_rate": 1.452905429481276e-08,
      "loss": 0.456,
      "step": 472140
    },
    {
      "epoch": 193.27,
      "grad_norm": 2.5470523834228516,
      "learning_rate": 1.4511424119928971e-08,
      "loss": 0.4472,
      "step": 472150
    },
    {
      "epoch": 193.27,
      "grad_norm": 1.7505124807357788,
      "learning_rate": 1.4493804614926566e-08,
      "loss": 0.4441,
      "step": 472160
    },
    {
      "epoch": 193.27,
      "grad_norm": 1.6891744136810303,
      "learning_rate": 1.447619577988569e-08,
      "loss": 0.4628,
      "step": 472170
    },
    {
      "epoch": 193.28,
      "grad_norm": 2.130542039871216,
      "learning_rate": 1.4458597614887788e-08,
      "loss": 0.4489,
      "step": 472180
    },
    {
      "epoch": 193.28,
      "grad_norm": 2.400508165359497,
      "learning_rate": 1.4441010120012745e-08,
      "loss": 0.4509,
      "step": 472190
    },
    {
      "epoch": 193.29,
      "grad_norm": 2.0776243209838867,
      "learning_rate": 1.4423433295341486e-08,
      "loss": 0.4487,
      "step": 472200
    },
    {
      "epoch": 193.29,
      "grad_norm": 2.2219321727752686,
      "learning_rate": 1.4405867140954672e-08,
      "loss": 0.4518,
      "step": 472210
    },
    {
      "epoch": 193.3,
      "grad_norm": 1.9823404550552368,
      "learning_rate": 1.4388311656932452e-08,
      "loss": 0.4528,
      "step": 472220
    },
    {
      "epoch": 193.3,
      "grad_norm": 2.0328521728515625,
      "learning_rate": 1.4370766843355489e-08,
      "loss": 0.4622,
      "step": 472230
    },
    {
      "epoch": 193.3,
      "grad_norm": 2.2573633193969727,
      "learning_rate": 1.4353232700303925e-08,
      "loss": 0.4497,
      "step": 472240
    },
    {
      "epoch": 193.31,
      "grad_norm": 1.666372537612915,
      "learning_rate": 1.4335709227858168e-08,
      "loss": 0.4428,
      "step": 472250
    },
    {
      "epoch": 193.31,
      "grad_norm": 2.050511598587036,
      "learning_rate": 1.4318196426098618e-08,
      "loss": 0.442,
      "step": 472260
    },
    {
      "epoch": 193.32,
      "grad_norm": 2.0803890228271484,
      "learning_rate": 1.4300694295105424e-08,
      "loss": 0.4298,
      "step": 472270
    },
    {
      "epoch": 193.32,
      "grad_norm": 1.945885181427002,
      "learning_rate": 1.4283202834958467e-08,
      "loss": 0.4403,
      "step": 472280
    },
    {
      "epoch": 193.32,
      "grad_norm": 2.0514328479766846,
      "learning_rate": 1.4265722045738414e-08,
      "loss": 0.4285,
      "step": 472290
    },
    {
      "epoch": 193.33,
      "grad_norm": 2.5047895908355713,
      "learning_rate": 1.4248251927525145e-08,
      "loss": 0.4302,
      "step": 472300
    },
    {
      "epoch": 193.33,
      "grad_norm": 2.2026729583740234,
      "learning_rate": 1.4230792480398288e-08,
      "loss": 0.4389,
      "step": 472310
    },
    {
      "epoch": 193.34,
      "grad_norm": 2.42642879486084,
      "learning_rate": 1.4213343704438506e-08,
      "loss": 0.4212,
      "step": 472320
    },
    {
      "epoch": 193.34,
      "grad_norm": 4.510265827178955,
      "learning_rate": 1.4195905599725163e-08,
      "loss": 0.4392,
      "step": 472330
    },
    {
      "epoch": 193.34,
      "grad_norm": 3.099546432495117,
      "learning_rate": 1.4178478166338401e-08,
      "loss": 0.4271,
      "step": 472340
    },
    {
      "epoch": 193.35,
      "grad_norm": 1.991537094116211,
      "learning_rate": 1.4161061404358107e-08,
      "loss": 0.4704,
      "step": 472350
    },
    {
      "epoch": 193.35,
      "grad_norm": 2.1213667392730713,
      "learning_rate": 1.4143655313863643e-08,
      "loss": 0.4193,
      "step": 472360
    },
    {
      "epoch": 193.36,
      "grad_norm": 1.8947646617889404,
      "learning_rate": 1.4126259894935415e-08,
      "loss": 0.436,
      "step": 472370
    },
    {
      "epoch": 193.36,
      "grad_norm": 1.8678126335144043,
      "learning_rate": 1.4108875147652524e-08,
      "loss": 0.4566,
      "step": 472380
    },
    {
      "epoch": 193.36,
      "grad_norm": 1.8890326023101807,
      "learning_rate": 1.4091501072094856e-08,
      "loss": 0.4367,
      "step": 472390
    },
    {
      "epoch": 193.37,
      "grad_norm": 1.7709397077560425,
      "learning_rate": 1.4074137668342034e-08,
      "loss": 0.429,
      "step": 472400
    },
    {
      "epoch": 193.37,
      "grad_norm": 2.224515676498413,
      "learning_rate": 1.4056784936473681e-08,
      "loss": 0.4232,
      "step": 472410
    },
    {
      "epoch": 193.38,
      "grad_norm": 2.06349515914917,
      "learning_rate": 1.4039442876568903e-08,
      "loss": 0.4526,
      "step": 472420
    },
    {
      "epoch": 193.38,
      "grad_norm": 2.252706289291382,
      "learning_rate": 1.402211148870758e-08,
      "loss": 0.4324,
      "step": 472430
    },
    {
      "epoch": 193.39,
      "grad_norm": 2.126950740814209,
      "learning_rate": 1.400479077296882e-08,
      "loss": 0.442,
      "step": 472440
    },
    {
      "epoch": 193.39,
      "grad_norm": 1.9895261526107788,
      "learning_rate": 1.3987480729431982e-08,
      "loss": 0.4517,
      "step": 472450
    },
    {
      "epoch": 193.39,
      "grad_norm": 2.4364888668060303,
      "learning_rate": 1.3970181358176693e-08,
      "loss": 0.4445,
      "step": 472460
    },
    {
      "epoch": 193.4,
      "grad_norm": 3.2492644786834717,
      "learning_rate": 1.3952892659281796e-08,
      "loss": 0.436,
      "step": 472470
    },
    {
      "epoch": 193.4,
      "grad_norm": 2.0264365673065186,
      "learning_rate": 1.3935614632826392e-08,
      "loss": 0.4375,
      "step": 472480
    },
    {
      "epoch": 193.41,
      "grad_norm": 2.084528923034668,
      "learning_rate": 1.3918347278890108e-08,
      "loss": 0.4262,
      "step": 472490
    },
    {
      "epoch": 193.41,
      "grad_norm": 2.3032400608062744,
      "learning_rate": 1.3901090597551786e-08,
      "loss": 0.4384,
      "step": 472500
    },
    {
      "epoch": 193.41,
      "grad_norm": 2.3086507320404053,
      "learning_rate": 1.3883844588890528e-08,
      "loss": 0.4333,
      "step": 472510
    },
    {
      "epoch": 193.42,
      "grad_norm": 1.7759954929351807,
      "learning_rate": 1.3866609252985179e-08,
      "loss": 0.4373,
      "step": 472520
    },
    {
      "epoch": 193.42,
      "grad_norm": 1.8920317888259888,
      "learning_rate": 1.38493845899151e-08,
      "loss": 0.4374,
      "step": 472530
    },
    {
      "epoch": 193.43,
      "grad_norm": 1.9081133604049683,
      "learning_rate": 1.3832170599758618e-08,
      "loss": 0.4603,
      "step": 472540
    },
    {
      "epoch": 193.43,
      "grad_norm": 2.09588623046875,
      "learning_rate": 1.3814967282594832e-08,
      "loss": 0.4473,
      "step": 472550
    },
    {
      "epoch": 193.43,
      "grad_norm": 1.9354392290115356,
      "learning_rate": 1.3797774638502849e-08,
      "loss": 0.4227,
      "step": 472560
    },
    {
      "epoch": 193.44,
      "grad_norm": 2.5281143188476562,
      "learning_rate": 1.378059266756099e-08,
      "loss": 0.4438,
      "step": 472570
    },
    {
      "epoch": 193.44,
      "grad_norm": 1.9759280681610107,
      "learning_rate": 1.3763421369848358e-08,
      "loss": 0.4406,
      "step": 472580
    },
    {
      "epoch": 193.45,
      "grad_norm": 2.1435463428497314,
      "learning_rate": 1.3746260745443274e-08,
      "loss": 0.4346,
      "step": 472590
    },
    {
      "epoch": 193.45,
      "grad_norm": 2.0635581016540527,
      "learning_rate": 1.3729110794424327e-08,
      "loss": 0.4451,
      "step": 472600
    },
    {
      "epoch": 193.45,
      "grad_norm": 1.9177566766738892,
      "learning_rate": 1.3711971516870355e-08,
      "loss": 0.452,
      "step": 472610
    },
    {
      "epoch": 193.46,
      "grad_norm": 2.2150683403015137,
      "learning_rate": 1.3694842912859941e-08,
      "loss": 0.4404,
      "step": 472620
    },
    {
      "epoch": 193.46,
      "grad_norm": 2.4557230472564697,
      "learning_rate": 1.3677724982471151e-08,
      "loss": 0.458,
      "step": 472630
    },
    {
      "epoch": 193.47,
      "grad_norm": 2.1298065185546875,
      "learning_rate": 1.3660617725782564e-08,
      "loss": 0.4468,
      "step": 472640
    },
    {
      "epoch": 193.47,
      "grad_norm": 2.1013715267181396,
      "learning_rate": 1.3643521142872507e-08,
      "loss": 0.4564,
      "step": 472650
    },
    {
      "epoch": 193.48,
      "grad_norm": 2.1168601512908936,
      "learning_rate": 1.362643523381956e-08,
      "loss": 0.4575,
      "step": 472660
    },
    {
      "epoch": 193.48,
      "grad_norm": 2.114511489868164,
      "learning_rate": 1.3609359998701525e-08,
      "loss": 0.4299,
      "step": 472670
    },
    {
      "epoch": 193.48,
      "grad_norm": 1.9804985523223877,
      "learning_rate": 1.3592295437596987e-08,
      "loss": 0.4506,
      "step": 472680
    },
    {
      "epoch": 193.49,
      "grad_norm": 2.3482868671417236,
      "learning_rate": 1.3575241550584267e-08,
      "loss": 0.4459,
      "step": 472690
    },
    {
      "epoch": 193.49,
      "grad_norm": 2.1673343181610107,
      "learning_rate": 1.3558198337740908e-08,
      "loss": 0.4335,
      "step": 472700
    },
    {
      "epoch": 193.5,
      "grad_norm": 2.2387773990631104,
      "learning_rate": 1.3541165799145493e-08,
      "loss": 0.4395,
      "step": 472710
    },
    {
      "epoch": 193.5,
      "grad_norm": 1.7152434587478638,
      "learning_rate": 1.3524143934875824e-08,
      "loss": 0.448,
      "step": 472720
    },
    {
      "epoch": 193.5,
      "grad_norm": 2.0958964824676514,
      "learning_rate": 1.3507132745010225e-08,
      "loss": 0.422,
      "step": 472730
    },
    {
      "epoch": 193.51,
      "grad_norm": 2.001821517944336,
      "learning_rate": 1.3490132229625974e-08,
      "loss": 0.4494,
      "step": 472740
    },
    {
      "epoch": 193.51,
      "grad_norm": 1.9096044301986694,
      "learning_rate": 1.34731423888014e-08,
      "loss": 0.458,
      "step": 472750
    },
    {
      "epoch": 193.52,
      "grad_norm": 1.9053512811660767,
      "learning_rate": 1.3456163222614042e-08,
      "loss": 0.4297,
      "step": 472760
    },
    {
      "epoch": 193.52,
      "grad_norm": 1.8135039806365967,
      "learning_rate": 1.343919473114222e-08,
      "loss": 0.4435,
      "step": 472770
    },
    {
      "epoch": 193.52,
      "grad_norm": 2.6594066619873047,
      "learning_rate": 1.3422236914462959e-08,
      "loss": 0.4456,
      "step": 472780
    },
    {
      "epoch": 193.53,
      "grad_norm": 2.538639783859253,
      "learning_rate": 1.3405289772654322e-08,
      "loss": 0.4488,
      "step": 472790
    },
    {
      "epoch": 193.53,
      "grad_norm": 2.4628922939300537,
      "learning_rate": 1.338835330579411e-08,
      "loss": 0.4613,
      "step": 472800
    },
    {
      "epoch": 193.54,
      "grad_norm": 2.259328603744507,
      "learning_rate": 1.3371427513959604e-08,
      "loss": 0.4471,
      "step": 472810
    },
    {
      "epoch": 193.54,
      "grad_norm": 1.6958023309707642,
      "learning_rate": 1.3354512397228349e-08,
      "loss": 0.4291,
      "step": 472820
    },
    {
      "epoch": 193.54,
      "grad_norm": 2.2621493339538574,
      "learning_rate": 1.3337607955678145e-08,
      "loss": 0.438,
      "step": 472830
    },
    {
      "epoch": 193.55,
      "grad_norm": 1.9982085227966309,
      "learning_rate": 1.3320714189386014e-08,
      "loss": 0.4411,
      "step": 472840
    },
    {
      "epoch": 193.55,
      "grad_norm": 2.1262996196746826,
      "learning_rate": 1.3303831098429499e-08,
      "loss": 0.4419,
      "step": 472850
    },
    {
      "epoch": 193.56,
      "grad_norm": 2.071176052093506,
      "learning_rate": 1.328695868288588e-08,
      "loss": 0.4488,
      "step": 472860
    },
    {
      "epoch": 193.56,
      "grad_norm": 2.505812644958496,
      "learning_rate": 1.3270096942832701e-08,
      "loss": 0.477,
      "step": 472870
    },
    {
      "epoch": 193.57,
      "grad_norm": 1.9927787780761719,
      "learning_rate": 1.3253245878346984e-08,
      "loss": 0.4453,
      "step": 472880
    },
    {
      "epoch": 193.57,
      "grad_norm": 1.9089645147323608,
      "learning_rate": 1.323640548950601e-08,
      "loss": 0.4344,
      "step": 472890
    },
    {
      "epoch": 193.57,
      "grad_norm": 1.6206753253936768,
      "learning_rate": 1.3219575776387061e-08,
      "loss": 0.464,
      "step": 472900
    },
    {
      "epoch": 193.58,
      "grad_norm": 1.9433114528656006,
      "learning_rate": 1.32027567390669e-08,
      "loss": 0.4175,
      "step": 472910
    },
    {
      "epoch": 193.58,
      "grad_norm": 1.8021363019943237,
      "learning_rate": 1.3185948377622546e-08,
      "loss": 0.4536,
      "step": 472920
    },
    {
      "epoch": 193.59,
      "grad_norm": 1.9498258829116821,
      "learning_rate": 1.3169150692131283e-08,
      "loss": 0.4398,
      "step": 472930
    },
    {
      "epoch": 193.59,
      "grad_norm": 2.2464003562927246,
      "learning_rate": 1.3152363682670132e-08,
      "loss": 0.4459,
      "step": 472940
    },
    {
      "epoch": 193.59,
      "grad_norm": 2.3671329021453857,
      "learning_rate": 1.3135587349315593e-08,
      "loss": 0.4684,
      "step": 472950
    },
    {
      "epoch": 193.6,
      "grad_norm": 2.1717135906219482,
      "learning_rate": 1.311882169214495e-08,
      "loss": 0.4511,
      "step": 472960
    },
    {
      "epoch": 193.6,
      "grad_norm": 2.4164535999298096,
      "learning_rate": 1.3102066711234704e-08,
      "loss": 0.4472,
      "step": 472970
    },
    {
      "epoch": 193.61,
      "grad_norm": 2.317121982574463,
      "learning_rate": 1.3085322406661616e-08,
      "loss": 0.4353,
      "step": 472980
    },
    {
      "epoch": 193.61,
      "grad_norm": 1.8360813856124878,
      "learning_rate": 1.3068588778502448e-08,
      "loss": 0.4408,
      "step": 472990
    },
    {
      "epoch": 193.61,
      "grad_norm": 2.3635590076446533,
      "learning_rate": 1.3051865826833961e-08,
      "loss": 0.4486,
      "step": 473000
    },
    {
      "epoch": 193.62,
      "grad_norm": 2.223416805267334,
      "learning_rate": 1.3035153551732656e-08,
      "loss": 0.4463,
      "step": 473010
    },
    {
      "epoch": 193.62,
      "grad_norm": 2.0927765369415283,
      "learning_rate": 1.3018451953275295e-08,
      "loss": 0.4326,
      "step": 473020
    },
    {
      "epoch": 193.63,
      "grad_norm": 2.177794933319092,
      "learning_rate": 1.3001761031537858e-08,
      "loss": 0.4302,
      "step": 473030
    },
    {
      "epoch": 193.63,
      "grad_norm": 1.8494261503219604,
      "learning_rate": 1.2985080786597369e-08,
      "loss": 0.455,
      "step": 473040
    },
    {
      "epoch": 193.63,
      "grad_norm": 2.1681153774261475,
      "learning_rate": 1.2968411218529806e-08,
      "loss": 0.429,
      "step": 473050
    },
    {
      "epoch": 193.64,
      "grad_norm": 1.933140754699707,
      "learning_rate": 1.2951752327411933e-08,
      "loss": 0.4282,
      "step": 473060
    },
    {
      "epoch": 193.64,
      "grad_norm": 2.03722882270813,
      "learning_rate": 1.293510411331973e-08,
      "loss": 0.4267,
      "step": 473070
    },
    {
      "epoch": 193.65,
      "grad_norm": 1.8725781440734863,
      "learning_rate": 1.2918466576329697e-08,
      "loss": 0.4324,
      "step": 473080
    },
    {
      "epoch": 193.65,
      "grad_norm": 1.8734337091445923,
      "learning_rate": 1.2901839716517818e-08,
      "loss": 0.4388,
      "step": 473090
    },
    {
      "epoch": 193.66,
      "grad_norm": 2.6781532764434814,
      "learning_rate": 1.2885223533960591e-08,
      "loss": 0.4517,
      "step": 473100
    },
    {
      "epoch": 193.66,
      "grad_norm": 1.860116958618164,
      "learning_rate": 1.2868618028733738e-08,
      "loss": 0.4368,
      "step": 473110
    },
    {
      "epoch": 193.66,
      "grad_norm": 1.90703547000885,
      "learning_rate": 1.285202320091376e-08,
      "loss": 0.4454,
      "step": 473120
    },
    {
      "epoch": 193.67,
      "grad_norm": 1.8340392112731934,
      "learning_rate": 1.2835439050576119e-08,
      "loss": 0.4538,
      "step": 473130
    },
    {
      "epoch": 193.67,
      "grad_norm": 2.0269806385040283,
      "learning_rate": 1.2818865577797315e-08,
      "loss": 0.4324,
      "step": 473140
    },
    {
      "epoch": 193.68,
      "grad_norm": 1.9909414052963257,
      "learning_rate": 1.2802302782652808e-08,
      "loss": 0.4472,
      "step": 473150
    },
    {
      "epoch": 193.68,
      "grad_norm": 2.0975699424743652,
      "learning_rate": 1.278575066521884e-08,
      "loss": 0.4369,
      "step": 473160
    },
    {
      "epoch": 193.68,
      "grad_norm": 1.7069827318191528,
      "learning_rate": 1.276920922557113e-08,
      "loss": 0.4417,
      "step": 473170
    },
    {
      "epoch": 193.69,
      "grad_norm": 2.324690580368042,
      "learning_rate": 1.275267846378566e-08,
      "loss": 0.4554,
      "step": 473180
    },
    {
      "epoch": 193.69,
      "grad_norm": 2.3488845825195312,
      "learning_rate": 1.2736158379937632e-08,
      "loss": 0.4535,
      "step": 473190
    },
    {
      "epoch": 193.7,
      "grad_norm": 1.9529242515563965,
      "learning_rate": 1.2719648974103026e-08,
      "loss": 0.4391,
      "step": 473200
    },
    {
      "epoch": 193.7,
      "grad_norm": 1.9988700151443481,
      "learning_rate": 1.2703150246357562e-08,
      "loss": 0.4488,
      "step": 473210
    },
    {
      "epoch": 193.7,
      "grad_norm": 2.374936580657959,
      "learning_rate": 1.2686662196776439e-08,
      "loss": 0.4266,
      "step": 473220
    },
    {
      "epoch": 193.71,
      "grad_norm": 2.4330594539642334,
      "learning_rate": 1.267018482543564e-08,
      "loss": 0.4344,
      "step": 473230
    },
    {
      "epoch": 193.71,
      "grad_norm": 2.4590327739715576,
      "learning_rate": 1.2653718132410364e-08,
      "loss": 0.423,
      "step": 473240
    },
    {
      "epoch": 193.72,
      "grad_norm": 1.8405357599258423,
      "learning_rate": 1.2637262117776333e-08,
      "loss": 0.4402,
      "step": 473250
    },
    {
      "epoch": 193.72,
      "grad_norm": 2.275503158569336,
      "learning_rate": 1.2620816781608745e-08,
      "loss": 0.453,
      "step": 473260
    },
    {
      "epoch": 193.72,
      "grad_norm": 2.2423391342163086,
      "learning_rate": 1.2604382123982802e-08,
      "loss": 0.4269,
      "step": 473270
    },
    {
      "epoch": 193.73,
      "grad_norm": 2.103492021560669,
      "learning_rate": 1.2587958144973706e-08,
      "loss": 0.4787,
      "step": 473280
    },
    {
      "epoch": 193.73,
      "grad_norm": 2.6660892963409424,
      "learning_rate": 1.2571544844657174e-08,
      "loss": 0.4355,
      "step": 473290
    },
    {
      "epoch": 193.74,
      "grad_norm": 1.769880771636963,
      "learning_rate": 1.2555142223108147e-08,
      "loss": 0.4543,
      "step": 473300
    },
    {
      "epoch": 193.74,
      "grad_norm": 2.067237138748169,
      "learning_rate": 1.2538750280401827e-08,
      "loss": 0.4393,
      "step": 473310
    },
    {
      "epoch": 193.75,
      "grad_norm": 2.1426920890808105,
      "learning_rate": 1.2522369016612892e-08,
      "loss": 0.4523,
      "step": 473320
    },
    {
      "epoch": 193.75,
      "grad_norm": 2.367128610610962,
      "learning_rate": 1.2505998431816805e-08,
      "loss": 0.4347,
      "step": 473330
    },
    {
      "epoch": 193.75,
      "grad_norm": 2.0654122829437256,
      "learning_rate": 1.2489638526088503e-08,
      "loss": 0.4415,
      "step": 473340
    },
    {
      "epoch": 193.76,
      "grad_norm": 2.0451321601867676,
      "learning_rate": 1.2473289299502927e-08,
      "loss": 0.4419,
      "step": 473350
    },
    {
      "epoch": 193.76,
      "grad_norm": 13.204495429992676,
      "learning_rate": 1.2456950752134758e-08,
      "loss": 0.4404,
      "step": 473360
    },
    {
      "epoch": 193.77,
      "grad_norm": 2.2024543285369873,
      "learning_rate": 1.2440622884059195e-08,
      "loss": 0.4401,
      "step": 473370
    },
    {
      "epoch": 193.77,
      "grad_norm": 2.297536849975586,
      "learning_rate": 1.2424305695350658e-08,
      "loss": 0.4437,
      "step": 473380
    },
    {
      "epoch": 193.77,
      "grad_norm": 2.060312271118164,
      "learning_rate": 1.2407999186084088e-08,
      "loss": 0.4486,
      "step": 473390
    },
    {
      "epoch": 193.78,
      "grad_norm": 2.2245032787323,
      "learning_rate": 1.2391703356334423e-08,
      "loss": 0.4713,
      "step": 473400
    },
    {
      "epoch": 193.78,
      "grad_norm": 2.4350929260253906,
      "learning_rate": 1.2375418206175824e-08,
      "loss": 0.4332,
      "step": 473410
    },
    {
      "epoch": 193.79,
      "grad_norm": 1.9337952136993408,
      "learning_rate": 1.2359143735682971e-08,
      "loss": 0.4469,
      "step": 473420
    },
    {
      "epoch": 193.79,
      "grad_norm": 1.8910248279571533,
      "learning_rate": 1.2342879944930803e-08,
      "loss": 0.4457,
      "step": 473430
    },
    {
      "epoch": 193.79,
      "grad_norm": 2.01911997795105,
      "learning_rate": 1.232662683399348e-08,
      "loss": 0.4348,
      "step": 473440
    },
    {
      "epoch": 193.8,
      "grad_norm": 2.169569492340088,
      "learning_rate": 1.2310384402945943e-08,
      "loss": 0.4738,
      "step": 473450
    },
    {
      "epoch": 193.8,
      "grad_norm": 2.1714494228363037,
      "learning_rate": 1.229415265186183e-08,
      "loss": 0.4518,
      "step": 473460
    },
    {
      "epoch": 193.81,
      "grad_norm": 2.192577600479126,
      "learning_rate": 1.227793158081608e-08,
      "loss": 0.4637,
      "step": 473470
    },
    {
      "epoch": 193.81,
      "grad_norm": 2.671269655227661,
      "learning_rate": 1.2261721189882595e-08,
      "loss": 0.465,
      "step": 473480
    },
    {
      "epoch": 193.81,
      "grad_norm": 1.9094160795211792,
      "learning_rate": 1.224552147913579e-08,
      "loss": 0.4483,
      "step": 473490
    },
    {
      "epoch": 193.82,
      "grad_norm": 1.9347772598266602,
      "learning_rate": 1.222933244865009e-08,
      "loss": 0.4598,
      "step": 473500
    },
    {
      "epoch": 193.82,
      "grad_norm": 2.255117416381836,
      "learning_rate": 1.2213154098499653e-08,
      "loss": 0.4429,
      "step": 473510
    },
    {
      "epoch": 193.83,
      "grad_norm": 1.9806232452392578,
      "learning_rate": 1.2196986428758114e-08,
      "loss": 0.4462,
      "step": 473520
    },
    {
      "epoch": 193.83,
      "grad_norm": 2.1754305362701416,
      "learning_rate": 1.2180829439499898e-08,
      "loss": 0.4447,
      "step": 473530
    },
    {
      "epoch": 193.84,
      "grad_norm": 2.186717987060547,
      "learning_rate": 1.2164683130799162e-08,
      "loss": 0.4692,
      "step": 473540
    },
    {
      "epoch": 193.84,
      "grad_norm": 2.1220903396606445,
      "learning_rate": 1.2148547502729544e-08,
      "loss": 0.4384,
      "step": 473550
    },
    {
      "epoch": 193.84,
      "grad_norm": 1.838843584060669,
      "learning_rate": 1.2132422555364943e-08,
      "loss": 0.425,
      "step": 473560
    },
    {
      "epoch": 193.85,
      "grad_norm": 1.812619686126709,
      "learning_rate": 1.2116308288779523e-08,
      "loss": 0.4513,
      "step": 473570
    },
    {
      "epoch": 193.85,
      "grad_norm": 2.0826449394226074,
      "learning_rate": 1.2100204703046918e-08,
      "loss": 0.452,
      "step": 473580
    },
    {
      "epoch": 193.86,
      "grad_norm": 2.1872594356536865,
      "learning_rate": 1.2084111798241029e-08,
      "loss": 0.4283,
      "step": 473590
    },
    {
      "epoch": 193.86,
      "grad_norm": 2.170285224914551,
      "learning_rate": 1.2068029574435236e-08,
      "loss": 0.4463,
      "step": 473600
    },
    {
      "epoch": 193.86,
      "grad_norm": 1.9974725246429443,
      "learning_rate": 1.2051958031703696e-08,
      "loss": 0.441,
      "step": 473610
    },
    {
      "epoch": 193.87,
      "grad_norm": 1.9854211807250977,
      "learning_rate": 1.203589717011953e-08,
      "loss": 0.4587,
      "step": 473620
    },
    {
      "epoch": 193.87,
      "grad_norm": 2.086876630783081,
      "learning_rate": 1.2019846989756634e-08,
      "loss": 0.4268,
      "step": 473630
    },
    {
      "epoch": 193.88,
      "grad_norm": 2.207390308380127,
      "learning_rate": 1.200380749068865e-08,
      "loss": 0.4454,
      "step": 473640
    },
    {
      "epoch": 193.88,
      "grad_norm": 2.381014823913574,
      "learning_rate": 1.1987778672988695e-08,
      "loss": 0.4381,
      "step": 473650
    },
    {
      "epoch": 193.88,
      "grad_norm": 1.9400380849838257,
      "learning_rate": 1.197176053673041e-08,
      "loss": 0.4401,
      "step": 473660
    },
    {
      "epoch": 193.89,
      "grad_norm": 2.1779448986053467,
      "learning_rate": 1.1955753081987173e-08,
      "loss": 0.4359,
      "step": 473670
    },
    {
      "epoch": 193.89,
      "grad_norm": 2.705082416534424,
      "learning_rate": 1.19397563088321e-08,
      "loss": 0.4212,
      "step": 473680
    },
    {
      "epoch": 193.9,
      "grad_norm": 1.939012885093689,
      "learning_rate": 1.1923770217338573e-08,
      "loss": 0.4407,
      "step": 473690
    },
    {
      "epoch": 193.9,
      "grad_norm": 2.1098122596740723,
      "learning_rate": 1.190779480757997e-08,
      "loss": 0.4667,
      "step": 473700
    },
    {
      "epoch": 193.91,
      "grad_norm": 2.3865807056427,
      "learning_rate": 1.1891830079629409e-08,
      "loss": 0.4406,
      "step": 473710
    },
    {
      "epoch": 193.91,
      "grad_norm": 2.4896271228790283,
      "learning_rate": 1.1875876033560008e-08,
      "loss": 0.4496,
      "step": 473720
    },
    {
      "epoch": 193.91,
      "grad_norm": 1.7606805562973022,
      "learning_rate": 1.1859932669444888e-08,
      "loss": 0.4439,
      "step": 473730
    },
    {
      "epoch": 193.92,
      "grad_norm": 2.0512640476226807,
      "learning_rate": 1.1843999987356905e-08,
      "loss": 0.4404,
      "step": 473740
    },
    {
      "epoch": 193.92,
      "grad_norm": 1.9962892532348633,
      "learning_rate": 1.1828077987369177e-08,
      "loss": 0.4444,
      "step": 473750
    },
    {
      "epoch": 193.93,
      "grad_norm": 1.8698506355285645,
      "learning_rate": 1.1812166669554825e-08,
      "loss": 0.4339,
      "step": 473760
    },
    {
      "epoch": 193.93,
      "grad_norm": 1.7888139486312866,
      "learning_rate": 1.1796266033986448e-08,
      "loss": 0.4523,
      "step": 473770
    },
    {
      "epoch": 193.93,
      "grad_norm": 2.2639546394348145,
      "learning_rate": 1.17803760807369e-08,
      "loss": 0.4451,
      "step": 473780
    },
    {
      "epoch": 193.94,
      "grad_norm": 1.8798861503601074,
      "learning_rate": 1.1764496809879044e-08,
      "loss": 0.42,
      "step": 473790
    },
    {
      "epoch": 193.94,
      "grad_norm": 1.9815257787704468,
      "learning_rate": 1.1748628221485476e-08,
      "loss": 0.4454,
      "step": 473800
    },
    {
      "epoch": 193.95,
      "grad_norm": 1.920305609703064,
      "learning_rate": 1.1732770315629314e-08,
      "loss": 0.4452,
      "step": 473810
    },
    {
      "epoch": 193.95,
      "grad_norm": 2.009124994277954,
      "learning_rate": 1.1716923092382896e-08,
      "loss": 0.4626,
      "step": 473820
    },
    {
      "epoch": 193.95,
      "grad_norm": 2.2387683391571045,
      "learning_rate": 1.1701086551818562e-08,
      "loss": 0.4696,
      "step": 473830
    },
    {
      "epoch": 193.96,
      "grad_norm": 1.9016057252883911,
      "learning_rate": 1.1685260694009428e-08,
      "loss": 0.4308,
      "step": 473840
    },
    {
      "epoch": 193.96,
      "grad_norm": 2.038705825805664,
      "learning_rate": 1.1669445519027574e-08,
      "loss": 0.4546,
      "step": 473850
    },
    {
      "epoch": 193.97,
      "grad_norm": 1.9807486534118652,
      "learning_rate": 1.1653641026945596e-08,
      "loss": 0.4454,
      "step": 473860
    },
    {
      "epoch": 193.97,
      "grad_norm": 1.9148271083831787,
      "learning_rate": 1.1637847217835834e-08,
      "loss": 0.4445,
      "step": 473870
    },
    {
      "epoch": 193.97,
      "grad_norm": 1.8141562938690186,
      "learning_rate": 1.1622064091770887e-08,
      "loss": 0.4404,
      "step": 473880
    },
    {
      "epoch": 193.98,
      "grad_norm": 2.0414319038391113,
      "learning_rate": 1.1606291648822569e-08,
      "loss": 0.4427,
      "step": 473890
    },
    {
      "epoch": 193.98,
      "grad_norm": 2.0585668087005615,
      "learning_rate": 1.159052988906348e-08,
      "loss": 0.4565,
      "step": 473900
    },
    {
      "epoch": 193.99,
      "grad_norm": 2.278822660446167,
      "learning_rate": 1.1574778812565959e-08,
      "loss": 0.4385,
      "step": 473910
    },
    {
      "epoch": 193.99,
      "grad_norm": 1.8501248359680176,
      "learning_rate": 1.1559038419401561e-08,
      "loss": 0.4575,
      "step": 473920
    },
    {
      "epoch": 194.0,
      "grad_norm": 2.058539628982544,
      "learning_rate": 1.1543308709643148e-08,
      "loss": 0.4584,
      "step": 473930
    },
    {
      "epoch": 194.0,
      "grad_norm": 2.4968621730804443,
      "learning_rate": 1.1527589683362014e-08,
      "loss": 0.4208,
      "step": 473940
    },
    {
      "epoch": 194.0,
      "eval_loss": 0.44495490193367004,
      "eval_runtime": 43.3618,
      "eval_samples_per_second": 79.54,
      "eval_steps_per_second": 9.963,
      "step": 473942
    },
    {
      "epoch": 194.0,
      "grad_norm": 1.9281388521194458,
      "learning_rate": 1.1511881340630757e-08,
      "loss": 0.4355,
      "step": 473950
    },
    {
      "epoch": 194.01,
      "grad_norm": 2.025378704071045,
      "learning_rate": 1.1496183681521197e-08,
      "loss": 0.4387,
      "step": 473960
    },
    {
      "epoch": 194.01,
      "grad_norm": 2.2821812629699707,
      "learning_rate": 1.148049670610515e-08,
      "loss": 0.4326,
      "step": 473970
    },
    {
      "epoch": 194.02,
      "grad_norm": 2.0186212062835693,
      "learning_rate": 1.1464820414454431e-08,
      "loss": 0.4387,
      "step": 473980
    },
    {
      "epoch": 194.02,
      "grad_norm": 2.30830717086792,
      "learning_rate": 1.1449154806640864e-08,
      "loss": 0.4406,
      "step": 473990
    },
    {
      "epoch": 194.02,
      "grad_norm": 3.372868537902832,
      "learning_rate": 1.1433499882736262e-08,
      "loss": 0.4344,
      "step": 474000
    },
    {
      "epoch": 194.03,
      "grad_norm": 2.3672029972076416,
      "learning_rate": 1.1417855642812442e-08,
      "loss": 0.4429,
      "step": 474010
    },
    {
      "epoch": 194.03,
      "grad_norm": 2.2401912212371826,
      "learning_rate": 1.1402222086940964e-08,
      "loss": 0.4508,
      "step": 474020
    },
    {
      "epoch": 194.04,
      "grad_norm": 2.300785779953003,
      "learning_rate": 1.1386599215193383e-08,
      "loss": 0.4509,
      "step": 474030
    },
    {
      "epoch": 194.04,
      "grad_norm": 3.153221845626831,
      "learning_rate": 1.137098702764152e-08,
      "loss": 0.4469,
      "step": 474040
    },
    {
      "epoch": 194.04,
      "grad_norm": 1.6925150156021118,
      "learning_rate": 1.1355385524356408e-08,
      "loss": 0.4513,
      "step": 474050
    },
    {
      "epoch": 194.05,
      "grad_norm": 2.063511371612549,
      "learning_rate": 1.1339794705410125e-08,
      "loss": 0.4513,
      "step": 474060
    },
    {
      "epoch": 194.05,
      "grad_norm": 2.6301310062408447,
      "learning_rate": 1.132421457087345e-08,
      "loss": 0.4315,
      "step": 474070
    },
    {
      "epoch": 194.06,
      "grad_norm": 2.4570870399475098,
      "learning_rate": 1.13086451208182e-08,
      "loss": 0.4359,
      "step": 474080
    },
    {
      "epoch": 194.06,
      "grad_norm": 2.0014681816101074,
      "learning_rate": 1.1293086355315672e-08,
      "loss": 0.4307,
      "step": 474090
    },
    {
      "epoch": 194.06,
      "grad_norm": 3.2521238327026367,
      "learning_rate": 1.1277538274436903e-08,
      "loss": 0.4524,
      "step": 474100
    },
    {
      "epoch": 194.07,
      "grad_norm": 2.509404182434082,
      "learning_rate": 1.1262000878253449e-08,
      "loss": 0.4493,
      "step": 474110
    },
    {
      "epoch": 194.07,
      "grad_norm": 1.963625431060791,
      "learning_rate": 1.1246474166836088e-08,
      "loss": 0.4252,
      "step": 474120
    },
    {
      "epoch": 194.08,
      "grad_norm": 1.9679396152496338,
      "learning_rate": 1.1230958140256377e-08,
      "loss": 0.4335,
      "step": 474130
    },
    {
      "epoch": 194.08,
      "grad_norm": 2.1486268043518066,
      "learning_rate": 1.1215452798584833e-08,
      "loss": 0.4397,
      "step": 474140
    },
    {
      "epoch": 194.09,
      "grad_norm": 2.3807523250579834,
      "learning_rate": 1.1199958141893013e-08,
      "loss": 0.4526,
      "step": 474150
    },
    {
      "epoch": 194.09,
      "grad_norm": 2.6137988567352295,
      "learning_rate": 1.1184474170251693e-08,
      "loss": 0.4233,
      "step": 474160
    },
    {
      "epoch": 194.09,
      "grad_norm": 1.780057430267334,
      "learning_rate": 1.1169000883731651e-08,
      "loss": 0.439,
      "step": 474170
    },
    {
      "epoch": 194.1,
      "grad_norm": 1.832170009613037,
      "learning_rate": 1.1153538282404183e-08,
      "loss": 0.4268,
      "step": 474180
    },
    {
      "epoch": 194.1,
      "grad_norm": 2.061065196990967,
      "learning_rate": 1.1138086366339547e-08,
      "loss": 0.4571,
      "step": 474190
    },
    {
      "epoch": 194.11,
      "grad_norm": 2.2879035472869873,
      "learning_rate": 1.1122645135609038e-08,
      "loss": 0.4411,
      "step": 474200
    },
    {
      "epoch": 194.11,
      "grad_norm": 2.0281546115875244,
      "learning_rate": 1.1107214590283173e-08,
      "loss": 0.4299,
      "step": 474210
    },
    {
      "epoch": 194.11,
      "grad_norm": 2.083664655685425,
      "learning_rate": 1.109179473043273e-08,
      "loss": 0.4397,
      "step": 474220
    },
    {
      "epoch": 194.12,
      "grad_norm": 2.402672290802002,
      "learning_rate": 1.1076385556128223e-08,
      "loss": 0.4507,
      "step": 474230
    },
    {
      "epoch": 194.12,
      "grad_norm": 1.947627067565918,
      "learning_rate": 1.1060987067440172e-08,
      "loss": 0.4352,
      "step": 474240
    },
    {
      "epoch": 194.13,
      "grad_norm": 2.1394002437591553,
      "learning_rate": 1.1045599264439351e-08,
      "loss": 0.4357,
      "step": 474250
    },
    {
      "epoch": 194.13,
      "grad_norm": 2.0562074184417725,
      "learning_rate": 1.1030222147196017e-08,
      "loss": 0.4442,
      "step": 474260
    },
    {
      "epoch": 194.13,
      "grad_norm": 1.9395416975021362,
      "learning_rate": 1.1014855715780687e-08,
      "loss": 0.4582,
      "step": 474270
    },
    {
      "epoch": 194.14,
      "grad_norm": 2.1445364952087402,
      "learning_rate": 1.0999499970263878e-08,
      "loss": 0.4373,
      "step": 474280
    },
    {
      "epoch": 194.14,
      "grad_norm": 2.213181495666504,
      "learning_rate": 1.0984154910715844e-08,
      "loss": 0.4553,
      "step": 474290
    },
    {
      "epoch": 194.15,
      "grad_norm": 2.202697277069092,
      "learning_rate": 1.0968820537206845e-08,
      "loss": 0.4465,
      "step": 474300
    },
    {
      "epoch": 194.15,
      "grad_norm": 2.3353261947631836,
      "learning_rate": 1.0953496849807134e-08,
      "loss": 0.4493,
      "step": 474310
    },
    {
      "epoch": 194.15,
      "grad_norm": 2.119021415710449,
      "learning_rate": 1.0938183848586967e-08,
      "loss": 0.4298,
      "step": 474320
    },
    {
      "epoch": 194.16,
      "grad_norm": 2.0036208629608154,
      "learning_rate": 1.0922881533616344e-08,
      "loss": 0.4447,
      "step": 474330
    },
    {
      "epoch": 194.16,
      "grad_norm": 2.0324435234069824,
      "learning_rate": 1.0907589904965778e-08,
      "loss": 0.4215,
      "step": 474340
    },
    {
      "epoch": 194.17,
      "grad_norm": 1.8863118886947632,
      "learning_rate": 1.0892308962704744e-08,
      "loss": 0.4578,
      "step": 474350
    },
    {
      "epoch": 194.17,
      "grad_norm": 2.1038169860839844,
      "learning_rate": 1.0877038706903242e-08,
      "loss": 0.4472,
      "step": 474360
    },
    {
      "epoch": 194.18,
      "grad_norm": 3.9272055625915527,
      "learning_rate": 1.0861779137631785e-08,
      "loss": 0.4496,
      "step": 474370
    },
    {
      "epoch": 194.18,
      "grad_norm": 2.7484614849090576,
      "learning_rate": 1.0846530254960112e-08,
      "loss": 0.4568,
      "step": 474380
    },
    {
      "epoch": 194.18,
      "grad_norm": 2.1660969257354736,
      "learning_rate": 1.0831292058957695e-08,
      "loss": 0.4528,
      "step": 474390
    },
    {
      "epoch": 194.19,
      "grad_norm": 1.9916454553604126,
      "learning_rate": 1.0816064549694792e-08,
      "loss": 0.4421,
      "step": 474400
    },
    {
      "epoch": 194.19,
      "grad_norm": 2.0065817832946777,
      "learning_rate": 1.080084772724088e-08,
      "loss": 0.46,
      "step": 474410
    },
    {
      "epoch": 194.2,
      "grad_norm": 2.1023924350738525,
      "learning_rate": 1.0785641591665693e-08,
      "loss": 0.4325,
      "step": 474420
    },
    {
      "epoch": 194.2,
      "grad_norm": 2.0394697189331055,
      "learning_rate": 1.0770446143038968e-08,
      "loss": 0.4432,
      "step": 474430
    },
    {
      "epoch": 194.2,
      "grad_norm": 1.9819210767745972,
      "learning_rate": 1.075526138143018e-08,
      "loss": 0.4591,
      "step": 474440
    },
    {
      "epoch": 194.21,
      "grad_norm": 1.9991704225540161,
      "learning_rate": 1.0740087306909066e-08,
      "loss": 0.4283,
      "step": 474450
    },
    {
      "epoch": 194.21,
      "grad_norm": 1.9893794059753418,
      "learning_rate": 1.072492391954484e-08,
      "loss": 0.4451,
      "step": 474460
    },
    {
      "epoch": 194.22,
      "grad_norm": 2.0034091472625732,
      "learning_rate": 1.0709771219407498e-08,
      "loss": 0.4393,
      "step": 474470
    },
    {
      "epoch": 194.22,
      "grad_norm": 2.0174601078033447,
      "learning_rate": 1.0694629206565996e-08,
      "loss": 0.4349,
      "step": 474480
    },
    {
      "epoch": 194.22,
      "grad_norm": 2.1322309970855713,
      "learning_rate": 1.0679497881089812e-08,
      "loss": 0.4447,
      "step": 474490
    },
    {
      "epoch": 194.23,
      "grad_norm": 1.8435211181640625,
      "learning_rate": 1.0664377243048156e-08,
      "loss": 0.4358,
      "step": 474500
    },
    {
      "epoch": 194.23,
      "grad_norm": 2.959712266921997,
      "learning_rate": 1.0649267292510509e-08,
      "loss": 0.45,
      "step": 474510
    },
    {
      "epoch": 194.24,
      "grad_norm": 2.37099289894104,
      "learning_rate": 1.0634168029546083e-08,
      "loss": 0.4447,
      "step": 474520
    },
    {
      "epoch": 194.24,
      "grad_norm": 2.103961706161499,
      "learning_rate": 1.0619079454223834e-08,
      "loss": 0.4285,
      "step": 474530
    },
    {
      "epoch": 194.24,
      "grad_norm": 1.8918232917785645,
      "learning_rate": 1.060400156661324e-08,
      "loss": 0.4352,
      "step": 474540
    },
    {
      "epoch": 194.25,
      "grad_norm": 2.128434896469116,
      "learning_rate": 1.0588934366782733e-08,
      "loss": 0.4406,
      "step": 474550
    },
    {
      "epoch": 194.25,
      "grad_norm": 2.1798810958862305,
      "learning_rate": 1.057387785480205e-08,
      "loss": 0.4457,
      "step": 474560
    },
    {
      "epoch": 194.26,
      "grad_norm": 2.401165723800659,
      "learning_rate": 1.0558832030739626e-08,
      "loss": 0.4401,
      "step": 474570
    },
    {
      "epoch": 194.26,
      "grad_norm": 1.8307174444198608,
      "learning_rate": 1.0543796894664677e-08,
      "loss": 0.4569,
      "step": 474580
    },
    {
      "epoch": 194.27,
      "grad_norm": 2.2353880405426025,
      "learning_rate": 1.0528772446646155e-08,
      "loss": 0.4641,
      "step": 474590
    },
    {
      "epoch": 194.27,
      "grad_norm": 2.151844024658203,
      "learning_rate": 1.0513758686752499e-08,
      "loss": 0.4474,
      "step": 474600
    },
    {
      "epoch": 194.27,
      "grad_norm": 2.023606538772583,
      "learning_rate": 1.0498755615052922e-08,
      "loss": 0.4289,
      "step": 474610
    },
    {
      "epoch": 194.28,
      "grad_norm": 2.1157572269439697,
      "learning_rate": 1.0483763231615861e-08,
      "loss": 0.4393,
      "step": 474620
    },
    {
      "epoch": 194.28,
      "grad_norm": 2.2652435302734375,
      "learning_rate": 1.046878153651001e-08,
      "loss": 0.4443,
      "step": 474630
    },
    {
      "epoch": 194.29,
      "grad_norm": 2.4292707443237305,
      "learning_rate": 1.0453810529804321e-08,
      "loss": 0.4507,
      "step": 474640
    },
    {
      "epoch": 194.29,
      "grad_norm": 1.9693267345428467,
      "learning_rate": 1.0438850211566715e-08,
      "loss": 0.4395,
      "step": 474650
    },
    {
      "epoch": 194.29,
      "grad_norm": 1.949804425239563,
      "learning_rate": 1.0423900581866402e-08,
      "loss": 0.4556,
      "step": 474660
    },
    {
      "epoch": 194.3,
      "grad_norm": 2.0090768337249756,
      "learning_rate": 1.0408961640771557e-08,
      "loss": 0.4397,
      "step": 474670
    },
    {
      "epoch": 194.3,
      "grad_norm": 1.8895695209503174,
      "learning_rate": 1.0394033388350618e-08,
      "loss": 0.454,
      "step": 474680
    },
    {
      "epoch": 194.31,
      "grad_norm": 2.0803751945495605,
      "learning_rate": 1.0379115824672019e-08,
      "loss": 0.4534,
      "step": 474690
    },
    {
      "epoch": 194.31,
      "grad_norm": 1.7879754304885864,
      "learning_rate": 1.0364208949803932e-08,
      "loss": 0.4518,
      "step": 474700
    },
    {
      "epoch": 194.31,
      "grad_norm": 2.4551618099212646,
      "learning_rate": 1.0349312763815055e-08,
      "loss": 0.4374,
      "step": 474710
    },
    {
      "epoch": 194.32,
      "grad_norm": 1.9353302717208862,
      "learning_rate": 1.033442726677304e-08,
      "loss": 0.4448,
      "step": 474720
    },
    {
      "epoch": 194.32,
      "grad_norm": 1.9373928308486938,
      "learning_rate": 1.0319552458746584e-08,
      "loss": 0.4655,
      "step": 474730
    },
    {
      "epoch": 194.33,
      "grad_norm": 2.022920608520508,
      "learning_rate": 1.03046883398036e-08,
      "loss": 0.4342,
      "step": 474740
    },
    {
      "epoch": 194.33,
      "grad_norm": 2.2826087474823,
      "learning_rate": 1.0289834910012264e-08,
      "loss": 0.4591,
      "step": 474750
    },
    {
      "epoch": 194.33,
      "grad_norm": 2.446577787399292,
      "learning_rate": 1.0274992169440489e-08,
      "loss": 0.4297,
      "step": 474760
    },
    {
      "epoch": 194.34,
      "grad_norm": 1.794519066810608,
      "learning_rate": 1.026016011815645e-08,
      "loss": 0.4226,
      "step": 474770
    },
    {
      "epoch": 194.34,
      "grad_norm": 2.0693821907043457,
      "learning_rate": 1.0245338756227801e-08,
      "loss": 0.4384,
      "step": 474780
    },
    {
      "epoch": 194.35,
      "grad_norm": 1.8838818073272705,
      "learning_rate": 1.0230528083722718e-08,
      "loss": 0.4503,
      "step": 474790
    },
    {
      "epoch": 194.35,
      "grad_norm": 2.433645248413086,
      "learning_rate": 1.0215728100708854e-08,
      "loss": 0.4281,
      "step": 474800
    },
    {
      "epoch": 194.36,
      "grad_norm": 1.8570376634597778,
      "learning_rate": 1.0200938807254384e-08,
      "loss": 0.4509,
      "step": 474810
    },
    {
      "epoch": 194.36,
      "grad_norm": 2.057410955429077,
      "learning_rate": 1.0186160203426441e-08,
      "loss": 0.4501,
      "step": 474820
    },
    {
      "epoch": 194.36,
      "grad_norm": 2.130741596221924,
      "learning_rate": 1.01713922892932e-08,
      "loss": 0.4482,
      "step": 474830
    },
    {
      "epoch": 194.37,
      "grad_norm": 1.8891618251800537,
      "learning_rate": 1.0156635064922057e-08,
      "loss": 0.4352,
      "step": 474840
    },
    {
      "epoch": 194.37,
      "grad_norm": 1.7163385152816772,
      "learning_rate": 1.0141888530380924e-08,
      "loss": 0.441,
      "step": 474850
    },
    {
      "epoch": 194.38,
      "grad_norm": 2.2549350261688232,
      "learning_rate": 1.0127152685736936e-08,
      "loss": 0.4468,
      "step": 474860
    },
    {
      "epoch": 194.38,
      "grad_norm": 2.0329060554504395,
      "learning_rate": 1.0112427531057746e-08,
      "loss": 0.425,
      "step": 474870
    },
    {
      "epoch": 194.38,
      "grad_norm": 2.2246432304382324,
      "learning_rate": 1.009771306641101e-08,
      "loss": 0.4402,
      "step": 474880
    },
    {
      "epoch": 194.39,
      "grad_norm": 1.9315992593765259,
      "learning_rate": 1.008300929186412e-08,
      "loss": 0.4512,
      "step": 474890
    },
    {
      "epoch": 194.39,
      "grad_norm": 2.423154592514038,
      "learning_rate": 1.006831620748395e-08,
      "loss": 0.4199,
      "step": 474900
    },
    {
      "epoch": 194.4,
      "grad_norm": 1.8827977180480957,
      "learning_rate": 1.0053633813338416e-08,
      "loss": 0.4577,
      "step": 474910
    },
    {
      "epoch": 194.4,
      "grad_norm": 1.9274053573608398,
      "learning_rate": 1.0038962109494651e-08,
      "loss": 0.4355,
      "step": 474920
    },
    {
      "epoch": 194.4,
      "grad_norm": 2.6595842838287354,
      "learning_rate": 1.0024301096019527e-08,
      "loss": 0.4395,
      "step": 474930
    },
    {
      "epoch": 194.41,
      "grad_norm": 2.1521739959716797,
      "learning_rate": 1.0009650772980181e-08,
      "loss": 0.4547,
      "step": 474940
    },
    {
      "epoch": 194.41,
      "grad_norm": 2.224809408187866,
      "learning_rate": 9.995011140444265e-09,
      "loss": 0.4755,
      "step": 474950
    },
    {
      "epoch": 194.42,
      "grad_norm": 1.8931885957717896,
      "learning_rate": 9.980382198478391e-09,
      "loss": 0.429,
      "step": 474960
    },
    {
      "epoch": 194.42,
      "grad_norm": 2.2433571815490723,
      "learning_rate": 9.965763947149957e-09,
      "loss": 0.4235,
      "step": 474970
    },
    {
      "epoch": 194.42,
      "grad_norm": 2.1223301887512207,
      "learning_rate": 9.951156386525313e-09,
      "loss": 0.4384,
      "step": 474980
    },
    {
      "epoch": 194.43,
      "grad_norm": 2.2087674140930176,
      "learning_rate": 9.936559516672115e-09,
      "loss": 0.4448,
      "step": 474990
    },
    {
      "epoch": 194.43,
      "grad_norm": 1.7559915781021118,
      "learning_rate": 9.921973337656455e-09,
      "loss": 0.4247,
      "step": 475000
    },
    {
      "epoch": 194.44,
      "grad_norm": 2.0818850994110107,
      "learning_rate": 9.907397849545727e-09,
      "loss": 0.4687,
      "step": 475010
    },
    {
      "epoch": 194.44,
      "grad_norm": 2.411578893661499,
      "learning_rate": 9.892833052406543e-09,
      "loss": 0.4343,
      "step": 475020
    },
    {
      "epoch": 194.45,
      "grad_norm": 2.131997585296631,
      "learning_rate": 9.87827894630552e-09,
      "loss": 0.4261,
      "step": 475030
    },
    {
      "epoch": 194.45,
      "grad_norm": 2.1137161254882812,
      "learning_rate": 9.863735531309529e-09,
      "loss": 0.4438,
      "step": 475040
    },
    {
      "epoch": 194.45,
      "grad_norm": 2.4322075843811035,
      "learning_rate": 9.849202807485183e-09,
      "loss": 0.4243,
      "step": 475050
    },
    {
      "epoch": 194.46,
      "grad_norm": 2.108083963394165,
      "learning_rate": 9.834680774898838e-09,
      "loss": 0.4575,
      "step": 475060
    },
    {
      "epoch": 194.46,
      "grad_norm": 2.0704362392425537,
      "learning_rate": 9.820169433616842e-09,
      "loss": 0.4373,
      "step": 475070
    },
    {
      "epoch": 194.47,
      "grad_norm": 1.9585599899291992,
      "learning_rate": 9.805668783706336e-09,
      "loss": 0.4515,
      "step": 475080
    },
    {
      "epoch": 194.47,
      "grad_norm": 2.045506000518799,
      "learning_rate": 9.791178825233145e-09,
      "loss": 0.4462,
      "step": 475090
    },
    {
      "epoch": 194.47,
      "grad_norm": 1.682202935218811,
      "learning_rate": 9.776699558263888e-09,
      "loss": 0.4403,
      "step": 475100
    },
    {
      "epoch": 194.48,
      "grad_norm": 3.4208574295043945,
      "learning_rate": 9.762230982864657e-09,
      "loss": 0.4556,
      "step": 475110
    },
    {
      "epoch": 194.48,
      "grad_norm": 2.1074368953704834,
      "learning_rate": 9.747773099102063e-09,
      "loss": 0.4347,
      "step": 475120
    },
    {
      "epoch": 194.49,
      "grad_norm": 2.066279649734497,
      "learning_rate": 9.733325907042202e-09,
      "loss": 0.4385,
      "step": 475130
    },
    {
      "epoch": 194.49,
      "grad_norm": 2.088488817214966,
      "learning_rate": 9.718889406751425e-09,
      "loss": 0.4242,
      "step": 475140
    },
    {
      "epoch": 194.49,
      "grad_norm": 1.764164924621582,
      "learning_rate": 9.704463598295566e-09,
      "loss": 0.4431,
      "step": 475150
    },
    {
      "epoch": 194.5,
      "grad_norm": 2.1711392402648926,
      "learning_rate": 9.690048481740976e-09,
      "loss": 0.4541,
      "step": 475160
    },
    {
      "epoch": 194.5,
      "grad_norm": 2.131103515625,
      "learning_rate": 9.67564405715323e-09,
      "loss": 0.4426,
      "step": 475170
    },
    {
      "epoch": 194.51,
      "grad_norm": 1.96232008934021,
      "learning_rate": 9.66125032459894e-09,
      "loss": 0.4419,
      "step": 475180
    },
    {
      "epoch": 194.51,
      "grad_norm": 2.597113847732544,
      "learning_rate": 9.64686728414368e-09,
      "loss": 0.4529,
      "step": 475190
    },
    {
      "epoch": 194.51,
      "grad_norm": 2.247073173522949,
      "learning_rate": 9.63249493585328e-09,
      "loss": 0.4184,
      "step": 475200
    },
    {
      "epoch": 194.52,
      "grad_norm": 2.4982776641845703,
      "learning_rate": 9.618133279793837e-09,
      "loss": 0.439,
      "step": 475210
    },
    {
      "epoch": 194.52,
      "grad_norm": 1.9996285438537598,
      "learning_rate": 9.60378231603092e-09,
      "loss": 0.4289,
      "step": 475220
    },
    {
      "epoch": 194.53,
      "grad_norm": 2.0305845737457275,
      "learning_rate": 9.589442044630364e-09,
      "loss": 0.4634,
      "step": 475230
    },
    {
      "epoch": 194.53,
      "grad_norm": 1.814094066619873,
      "learning_rate": 9.575112465657742e-09,
      "loss": 0.4494,
      "step": 475240
    },
    {
      "epoch": 194.54,
      "grad_norm": 2.010871648788452,
      "learning_rate": 9.560793579178882e-09,
      "loss": 0.4578,
      "step": 475250
    },
    {
      "epoch": 194.54,
      "grad_norm": 2.3728532791137695,
      "learning_rate": 9.546485385259101e-09,
      "loss": 0.453,
      "step": 475260
    },
    {
      "epoch": 194.54,
      "grad_norm": 1.8739691972732544,
      "learning_rate": 9.532187883964231e-09,
      "loss": 0.4461,
      "step": 475270
    },
    {
      "epoch": 194.55,
      "grad_norm": 1.9061391353607178,
      "learning_rate": 9.517901075359844e-09,
      "loss": 0.4397,
      "step": 475280
    },
    {
      "epoch": 194.55,
      "grad_norm": 2.364302158355713,
      "learning_rate": 9.503624959510993e-09,
      "loss": 0.4324,
      "step": 475290
    },
    {
      "epoch": 194.56,
      "grad_norm": 1.7923740148544312,
      "learning_rate": 9.489359536483248e-09,
      "loss": 0.4344,
      "step": 475300
    },
    {
      "epoch": 194.56,
      "grad_norm": 2.188681125640869,
      "learning_rate": 9.475104806341925e-09,
      "loss": 0.438,
      "step": 475310
    },
    {
      "epoch": 194.56,
      "grad_norm": 2.146390914916992,
      "learning_rate": 9.460860769152593e-09,
      "loss": 0.475,
      "step": 475320
    },
    {
      "epoch": 194.57,
      "grad_norm": 2.1713342666625977,
      "learning_rate": 9.446627424980306e-09,
      "loss": 0.4252,
      "step": 475330
    },
    {
      "epoch": 194.57,
      "grad_norm": 2.1408705711364746,
      "learning_rate": 9.432404773890376e-09,
      "loss": 0.4322,
      "step": 475340
    },
    {
      "epoch": 194.58,
      "grad_norm": 1.9279401302337646,
      "learning_rate": 9.418192815947596e-09,
      "loss": 0.4406,
      "step": 475350
    },
    {
      "epoch": 194.58,
      "grad_norm": 2.2158100605010986,
      "learning_rate": 9.403991551217276e-09,
      "loss": 0.4439,
      "step": 475360
    },
    {
      "epoch": 194.58,
      "grad_norm": 2.182025671005249,
      "learning_rate": 9.38980097976447e-09,
      "loss": 0.4038,
      "step": 475370
    },
    {
      "epoch": 194.59,
      "grad_norm": 2.048368453979492,
      "learning_rate": 9.375621101654491e-09,
      "loss": 0.4276,
      "step": 475380
    },
    {
      "epoch": 194.59,
      "grad_norm": 6.562432765960693,
      "learning_rate": 9.361451916951868e-09,
      "loss": 0.4391,
      "step": 475390
    },
    {
      "epoch": 194.6,
      "grad_norm": 2.194888114929199,
      "learning_rate": 9.347293425721654e-09,
      "loss": 0.4513,
      "step": 475400
    },
    {
      "epoch": 194.6,
      "grad_norm": 2.0860042572021484,
      "learning_rate": 9.333145628028643e-09,
      "loss": 0.4501,
      "step": 475410
    },
    {
      "epoch": 194.6,
      "grad_norm": 1.8280034065246582,
      "learning_rate": 9.319008523937886e-09,
      "loss": 0.4439,
      "step": 475420
    },
    {
      "epoch": 194.61,
      "grad_norm": 1.9268739223480225,
      "learning_rate": 9.304882113513653e-09,
      "loss": 0.4436,
      "step": 475430
    },
    {
      "epoch": 194.61,
      "grad_norm": 2.1319615840911865,
      "learning_rate": 9.290766396821258e-09,
      "loss": 0.4575,
      "step": 475440
    },
    {
      "epoch": 194.62,
      "grad_norm": 2.215083599090576,
      "learning_rate": 9.276661373924971e-09,
      "loss": 0.4372,
      "step": 475450
    },
    {
      "epoch": 194.62,
      "grad_norm": 2.8291802406311035,
      "learning_rate": 9.262567044889587e-09,
      "loss": 0.4587,
      "step": 475460
    },
    {
      "epoch": 194.63,
      "grad_norm": 2.1914896965026855,
      "learning_rate": 9.248483409779635e-09,
      "loss": 0.4448,
      "step": 475470
    },
    {
      "epoch": 194.63,
      "grad_norm": 2.1559641361236572,
      "learning_rate": 9.234410468659388e-09,
      "loss": 0.4616,
      "step": 475480
    },
    {
      "epoch": 194.63,
      "grad_norm": 2.0161492824554443,
      "learning_rate": 9.220348221593635e-09,
      "loss": 0.4576,
      "step": 475490
    },
    {
      "epoch": 194.64,
      "grad_norm": 2.115851879119873,
      "learning_rate": 9.20629666864665e-09,
      "loss": 0.4414,
      "step": 475500
    },
    {
      "epoch": 194.64,
      "grad_norm": 2.079747200012207,
      "learning_rate": 9.192255809882705e-09,
      "loss": 0.4297,
      "step": 475510
    },
    {
      "epoch": 194.65,
      "grad_norm": 2.0006260871887207,
      "learning_rate": 9.17822564536633e-09,
      "loss": 0.4411,
      "step": 475520
    },
    {
      "epoch": 194.65,
      "grad_norm": 2.0193259716033936,
      "learning_rate": 9.164206175161799e-09,
      "loss": 0.4606,
      "step": 475530
    },
    {
      "epoch": 194.65,
      "grad_norm": 2.06913423538208,
      "learning_rate": 9.15019739933312e-09,
      "loss": 0.4425,
      "step": 475540
    },
    {
      "epoch": 194.66,
      "grad_norm": 2.0551187992095947,
      "learning_rate": 9.136199317944567e-09,
      "loss": 0.4395,
      "step": 475550
    },
    {
      "epoch": 194.66,
      "grad_norm": 3.3767619132995605,
      "learning_rate": 9.12221193106041e-09,
      "loss": 0.4235,
      "step": 475560
    },
    {
      "epoch": 194.67,
      "grad_norm": 2.182403326034546,
      "learning_rate": 9.1082352387444e-09,
      "loss": 0.4202,
      "step": 475570
    },
    {
      "epoch": 194.67,
      "grad_norm": 1.940056324005127,
      "learning_rate": 9.094269241060812e-09,
      "loss": 0.435,
      "step": 475580
    },
    {
      "epoch": 194.67,
      "grad_norm": 2.287482976913452,
      "learning_rate": 9.080313938073653e-09,
      "loss": 0.4366,
      "step": 475590
    },
    {
      "epoch": 194.68,
      "grad_norm": 2.1691136360168457,
      "learning_rate": 9.066369329846675e-09,
      "loss": 0.4415,
      "step": 475600
    },
    {
      "epoch": 194.68,
      "grad_norm": 1.9073184728622437,
      "learning_rate": 9.05243541644389e-09,
      "loss": 0.4507,
      "step": 475610
    },
    {
      "epoch": 194.69,
      "grad_norm": 1.7042992115020752,
      "learning_rate": 9.03851219792905e-09,
      "loss": 0.4474,
      "step": 475620
    },
    {
      "epoch": 194.69,
      "grad_norm": 2.3345556259155273,
      "learning_rate": 9.024599674365904e-09,
      "loss": 0.4441,
      "step": 475630
    },
    {
      "epoch": 194.7,
      "grad_norm": 2.007666826248169,
      "learning_rate": 9.010697845818464e-09,
      "loss": 0.4526,
      "step": 475640
    },
    {
      "epoch": 194.7,
      "grad_norm": 2.0781803131103516,
      "learning_rate": 8.996806712349963e-09,
      "loss": 0.4594,
      "step": 475650
    },
    {
      "epoch": 194.7,
      "grad_norm": 2.395692825317383,
      "learning_rate": 8.98292627402441e-09,
      "loss": 0.4383,
      "step": 475660
    },
    {
      "epoch": 194.71,
      "grad_norm": 2.1453521251678467,
      "learning_rate": 8.969056530905033e-09,
      "loss": 0.4508,
      "step": 475670
    },
    {
      "epoch": 194.71,
      "grad_norm": 2.539745330810547,
      "learning_rate": 8.955197483055589e-09,
      "loss": 0.4331,
      "step": 475680
    },
    {
      "epoch": 194.72,
      "grad_norm": 2.289264678955078,
      "learning_rate": 8.941349130539565e-09,
      "loss": 0.4324,
      "step": 475690
    },
    {
      "epoch": 194.72,
      "grad_norm": 2.0417864322662354,
      "learning_rate": 8.927511473420454e-09,
      "loss": 0.4575,
      "step": 475700
    },
    {
      "epoch": 194.72,
      "grad_norm": 2.588778018951416,
      "learning_rate": 8.913684511761484e-09,
      "loss": 0.4447,
      "step": 475710
    },
    {
      "epoch": 194.73,
      "grad_norm": 2.0121326446533203,
      "learning_rate": 8.899868245626409e-09,
      "loss": 0.4301,
      "step": 475720
    },
    {
      "epoch": 194.73,
      "grad_norm": 2.1939470767974854,
      "learning_rate": 8.886062675077938e-09,
      "loss": 0.4591,
      "step": 475730
    },
    {
      "epoch": 194.74,
      "grad_norm": 2.5031704902648926,
      "learning_rate": 8.87226780017956e-09,
      "loss": 0.4244,
      "step": 475740
    },
    {
      "epoch": 194.74,
      "grad_norm": 1.9936435222625732,
      "learning_rate": 8.858483620994249e-09,
      "loss": 0.4202,
      "step": 475750
    },
    {
      "epoch": 194.74,
      "grad_norm": 1.9676402807235718,
      "learning_rate": 8.844710137585753e-09,
      "loss": 0.4344,
      "step": 475760
    },
    {
      "epoch": 194.75,
      "grad_norm": 2.2061681747436523,
      "learning_rate": 8.830947350016524e-09,
      "loss": 0.441,
      "step": 475770
    },
    {
      "epoch": 194.75,
      "grad_norm": 2.672844409942627,
      "learning_rate": 8.817195258350051e-09,
      "loss": 0.4652,
      "step": 475780
    },
    {
      "epoch": 194.76,
      "grad_norm": 2.1908316612243652,
      "learning_rate": 8.803453862649048e-09,
      "loss": 0.446,
      "step": 475790
    },
    {
      "epoch": 194.76,
      "grad_norm": 1.9795621633529663,
      "learning_rate": 8.789723162976742e-09,
      "loss": 0.4371,
      "step": 475800
    },
    {
      "epoch": 194.76,
      "grad_norm": 2.973466396331787,
      "learning_rate": 8.776003159395584e-09,
      "loss": 0.4516,
      "step": 475810
    },
    {
      "epoch": 194.77,
      "grad_norm": 2.2259905338287354,
      "learning_rate": 8.762293851968805e-09,
      "loss": 0.4505,
      "step": 475820
    },
    {
      "epoch": 194.77,
      "grad_norm": 2.461500644683838,
      "learning_rate": 8.748595240759114e-09,
      "loss": 0.4369,
      "step": 475830
    },
    {
      "epoch": 194.78,
      "grad_norm": 2.0932838916778564,
      "learning_rate": 8.734907325829225e-09,
      "loss": 0.4456,
      "step": 475840
    },
    {
      "epoch": 194.78,
      "grad_norm": 2.5819265842437744,
      "learning_rate": 8.721230107241844e-09,
      "loss": 0.4482,
      "step": 475850
    },
    {
      "epoch": 194.79,
      "grad_norm": 2.0421464443206787,
      "learning_rate": 8.707563585059681e-09,
      "loss": 0.451,
      "step": 475860
    },
    {
      "epoch": 194.79,
      "grad_norm": 1.8488653898239136,
      "learning_rate": 8.69390775934493e-09,
      "loss": 0.4256,
      "step": 475870
    },
    {
      "epoch": 194.79,
      "grad_norm": 1.752866268157959,
      "learning_rate": 8.68026263016082e-09,
      "loss": 0.4444,
      "step": 475880
    },
    {
      "epoch": 194.8,
      "grad_norm": 1.9011926651000977,
      "learning_rate": 8.66662819756954e-09,
      "loss": 0.4362,
      "step": 475890
    },
    {
      "epoch": 194.8,
      "grad_norm": 2.339305877685547,
      "learning_rate": 8.653004461633538e-09,
      "loss": 0.4328,
      "step": 475900
    },
    {
      "epoch": 194.81,
      "grad_norm": 1.9544049501419067,
      "learning_rate": 8.639391422415269e-09,
      "loss": 0.4494,
      "step": 475910
    },
    {
      "epoch": 194.81,
      "grad_norm": 2.34635853767395,
      "learning_rate": 8.625789079977178e-09,
      "loss": 0.4451,
      "step": 475920
    },
    {
      "epoch": 194.81,
      "grad_norm": 1.9090372323989868,
      "learning_rate": 8.612197434381459e-09,
      "loss": 0.4378,
      "step": 475930
    },
    {
      "epoch": 194.82,
      "grad_norm": 2.4288899898529053,
      "learning_rate": 8.59861648569004e-09,
      "loss": 0.4443,
      "step": 475940
    },
    {
      "epoch": 194.82,
      "grad_norm": 1.6573652029037476,
      "learning_rate": 8.58504623396589e-09,
      "loss": 0.4509,
      "step": 475950
    },
    {
      "epoch": 194.83,
      "grad_norm": 1.98775315284729,
      "learning_rate": 8.57148667927042e-09,
      "loss": 0.4442,
      "step": 475960
    },
    {
      "epoch": 194.83,
      "grad_norm": 2.1549389362335205,
      "learning_rate": 8.55793782166608e-09,
      "loss": 0.4402,
      "step": 475970
    },
    {
      "epoch": 194.83,
      "grad_norm": 2.0938994884490967,
      "learning_rate": 8.544399661215058e-09,
      "loss": 0.4501,
      "step": 475980
    },
    {
      "epoch": 194.84,
      "grad_norm": 1.7761296033859253,
      "learning_rate": 8.530872197979286e-09,
      "loss": 0.4417,
      "step": 475990
    },
    {
      "epoch": 194.84,
      "grad_norm": 1.9604007005691528,
      "learning_rate": 8.517355432020433e-09,
      "loss": 0.4402,
      "step": 476000
    },
    {
      "epoch": 194.85,
      "grad_norm": 1.6439354419708252,
      "learning_rate": 8.503849363400948e-09,
      "loss": 0.4337,
      "step": 476010
    },
    {
      "epoch": 194.85,
      "grad_norm": 1.9083483219146729,
      "learning_rate": 8.49035399218224e-09,
      "loss": 0.4375,
      "step": 476020
    },
    {
      "epoch": 194.85,
      "grad_norm": 1.812812328338623,
      "learning_rate": 8.4768693184265e-09,
      "loss": 0.4343,
      "step": 476030
    },
    {
      "epoch": 194.86,
      "grad_norm": 2.321188449859619,
      "learning_rate": 8.463395342195138e-09,
      "loss": 0.4541,
      "step": 476040
    },
    {
      "epoch": 194.86,
      "grad_norm": 2.014294147491455,
      "learning_rate": 8.44993206355008e-09,
      "loss": 0.4208,
      "step": 476050
    },
    {
      "epoch": 194.87,
      "grad_norm": 1.9198241233825684,
      "learning_rate": 8.436479482553e-09,
      "loss": 0.4431,
      "step": 476060
    },
    {
      "epoch": 194.87,
      "grad_norm": 2.1101114749908447,
      "learning_rate": 8.423037599265304e-09,
      "loss": 0.4539,
      "step": 476070
    },
    {
      "epoch": 194.88,
      "grad_norm": 2.140026092529297,
      "learning_rate": 8.409606413748924e-09,
      "loss": 0.4301,
      "step": 476080
    },
    {
      "epoch": 194.88,
      "grad_norm": 2.4881346225738525,
      "learning_rate": 8.396185926065266e-09,
      "loss": 0.4365,
      "step": 476090
    },
    {
      "epoch": 194.88,
      "grad_norm": 6.354722499847412,
      "learning_rate": 8.382776136275742e-09,
      "loss": 0.4422,
      "step": 476100
    },
    {
      "epoch": 194.89,
      "grad_norm": 1.7444984912872314,
      "learning_rate": 8.369377044441759e-09,
      "loss": 0.4405,
      "step": 476110
    },
    {
      "epoch": 194.89,
      "grad_norm": 2.405712604522705,
      "learning_rate": 8.35598865062473e-09,
      "loss": 0.4467,
      "step": 476120
    },
    {
      "epoch": 194.9,
      "grad_norm": 1.9116038084030151,
      "learning_rate": 8.342610954885798e-09,
      "loss": 0.462,
      "step": 476130
    },
    {
      "epoch": 194.9,
      "grad_norm": 1.9737951755523682,
      "learning_rate": 8.329243957286638e-09,
      "loss": 0.4594,
      "step": 476140
    },
    {
      "epoch": 194.9,
      "grad_norm": 2.534086227416992,
      "learning_rate": 8.315887657888398e-09,
      "loss": 0.4357,
      "step": 476150
    },
    {
      "epoch": 194.91,
      "grad_norm": 1.8799010515213013,
      "learning_rate": 8.302542056751963e-09,
      "loss": 0.4385,
      "step": 476160
    },
    {
      "epoch": 194.91,
      "grad_norm": 2.5161526203155518,
      "learning_rate": 8.289207153938486e-09,
      "loss": 0.4495,
      "step": 476170
    },
    {
      "epoch": 194.92,
      "grad_norm": 1.6977274417877197,
      "learning_rate": 8.275882949509375e-09,
      "loss": 0.4407,
      "step": 476180
    },
    {
      "epoch": 194.92,
      "grad_norm": 2.2533390522003174,
      "learning_rate": 8.262569443525519e-09,
      "loss": 0.4547,
      "step": 476190
    },
    {
      "epoch": 194.92,
      "grad_norm": 2.080294609069824,
      "learning_rate": 8.249266636048065e-09,
      "loss": 0.4362,
      "step": 476200
    },
    {
      "epoch": 194.93,
      "grad_norm": 1.6460825204849243,
      "learning_rate": 8.235974527137645e-09,
      "loss": 0.4351,
      "step": 476210
    },
    {
      "epoch": 194.93,
      "grad_norm": 1.899373173713684,
      "learning_rate": 8.222693116855147e-09,
      "loss": 0.4433,
      "step": 476220
    },
    {
      "epoch": 194.94,
      "grad_norm": 1.9304481744766235,
      "learning_rate": 8.209422405261717e-09,
      "loss": 0.4647,
      "step": 476230
    },
    {
      "epoch": 194.94,
      "grad_norm": 1.8292338848114014,
      "learning_rate": 8.196162392417727e-09,
      "loss": 0.4398,
      "step": 476240
    },
    {
      "epoch": 194.94,
      "grad_norm": 1.845803141593933,
      "learning_rate": 8.182913078384323e-09,
      "loss": 0.4723,
      "step": 476250
    },
    {
      "epoch": 194.95,
      "grad_norm": 1.904463291168213,
      "learning_rate": 8.169674463221876e-09,
      "loss": 0.4387,
      "step": 476260
    },
    {
      "epoch": 194.95,
      "grad_norm": 2.1202940940856934,
      "learning_rate": 8.156446546991532e-09,
      "loss": 0.4238,
      "step": 476270
    },
    {
      "epoch": 194.96,
      "grad_norm": 2.1318702697753906,
      "learning_rate": 8.143229329753143e-09,
      "loss": 0.4508,
      "step": 476280
    },
    {
      "epoch": 194.96,
      "grad_norm": 2.4584949016571045,
      "learning_rate": 8.130022811567854e-09,
      "loss": 0.4357,
      "step": 476290
    },
    {
      "epoch": 194.97,
      "grad_norm": 2.060544013977051,
      "learning_rate": 8.116826992496036e-09,
      "loss": 0.4388,
      "step": 476300
    },
    {
      "epoch": 194.97,
      "grad_norm": 1.7327697277069092,
      "learning_rate": 8.103641872598057e-09,
      "loss": 0.4394,
      "step": 476310
    },
    {
      "epoch": 194.97,
      "grad_norm": 2.828157901763916,
      "learning_rate": 8.090467451934285e-09,
      "loss": 0.4546,
      "step": 476320
    },
    {
      "epoch": 194.98,
      "grad_norm": 1.7938165664672852,
      "learning_rate": 8.077303730565087e-09,
      "loss": 0.4376,
      "step": 476330
    },
    {
      "epoch": 194.98,
      "grad_norm": 2.1187331676483154,
      "learning_rate": 8.064150708550575e-09,
      "loss": 0.4477,
      "step": 476340
    },
    {
      "epoch": 194.99,
      "grad_norm": 2.1732370853424072,
      "learning_rate": 8.051008385951632e-09,
      "loss": 0.4362,
      "step": 476350
    },
    {
      "epoch": 194.99,
      "grad_norm": 1.74446702003479,
      "learning_rate": 8.037876762827851e-09,
      "loss": 0.4286,
      "step": 476360
    },
    {
      "epoch": 194.99,
      "grad_norm": 2.2747933864593506,
      "learning_rate": 8.0247558392396e-09,
      "loss": 0.4332,
      "step": 476370
    },
    {
      "epoch": 195.0,
      "grad_norm": 1.9373271465301514,
      "learning_rate": 8.011645615246722e-09,
      "loss": 0.4274,
      "step": 476380
    },
    {
      "epoch": 195.0,
      "eval_loss": 0.44291582703590393,
      "eval_runtime": 43.2193,
      "eval_samples_per_second": 79.802,
      "eval_steps_per_second": 9.996,
      "step": 476385
    },
    {
      "epoch": 195.0,
      "grad_norm": 2.4935245513916016,
      "learning_rate": 7.998546090909851e-09,
      "loss": 0.4507,
      "step": 476390
    },
    {
      "epoch": 195.01,
      "grad_norm": 2.032109498977661,
      "learning_rate": 7.985457266288572e-09,
      "loss": 0.4344,
      "step": 476400
    },
    {
      "epoch": 195.01,
      "grad_norm": 2.1879022121429443,
      "learning_rate": 7.972379141442733e-09,
      "loss": 0.4239,
      "step": 476410
    },
    {
      "epoch": 195.01,
      "grad_norm": 2.183328628540039,
      "learning_rate": 7.959311716432445e-09,
      "loss": 0.4402,
      "step": 476420
    },
    {
      "epoch": 195.02,
      "grad_norm": 1.8261659145355225,
      "learning_rate": 7.946254991317812e-09,
      "loss": 0.4355,
      "step": 476430
    },
    {
      "epoch": 195.02,
      "grad_norm": 1.918377161026001,
      "learning_rate": 7.933208966158165e-09,
      "loss": 0.4558,
      "step": 476440
    },
    {
      "epoch": 195.03,
      "grad_norm": 1.7967673540115356,
      "learning_rate": 7.920173641013608e-09,
      "loss": 0.4325,
      "step": 476450
    },
    {
      "epoch": 195.03,
      "grad_norm": 1.6687787771224976,
      "learning_rate": 7.907149015943735e-09,
      "loss": 0.4487,
      "step": 476460
    },
    {
      "epoch": 195.03,
      "grad_norm": 2.109926223754883,
      "learning_rate": 7.894135091008128e-09,
      "loss": 0.4376,
      "step": 476470
    },
    {
      "epoch": 195.04,
      "grad_norm": 1.9709727764129639,
      "learning_rate": 7.881131866266377e-09,
      "loss": 0.4575,
      "step": 476480
    },
    {
      "epoch": 195.04,
      "grad_norm": 2.0200841426849365,
      "learning_rate": 7.86813934177833e-09,
      "loss": 0.4558,
      "step": 476490
    },
    {
      "epoch": 195.05,
      "grad_norm": 2.1673460006713867,
      "learning_rate": 7.855157517603575e-09,
      "loss": 0.442,
      "step": 476500
    },
    {
      "epoch": 195.05,
      "grad_norm": 1.9884991645812988,
      "learning_rate": 7.84218639380092e-09,
      "loss": 0.4491,
      "step": 476510
    },
    {
      "epoch": 195.06,
      "grad_norm": 1.9094326496124268,
      "learning_rate": 7.82922597043047e-09,
      "loss": 0.4733,
      "step": 476520
    },
    {
      "epoch": 195.06,
      "grad_norm": 1.9787696599960327,
      "learning_rate": 7.816276247551295e-09,
      "loss": 0.447,
      "step": 476530
    },
    {
      "epoch": 195.06,
      "grad_norm": 2.2957141399383545,
      "learning_rate": 7.803337225222463e-09,
      "loss": 0.455,
      "step": 476540
    },
    {
      "epoch": 195.07,
      "grad_norm": 1.8725939989089966,
      "learning_rate": 7.790408903503818e-09,
      "loss": 0.433,
      "step": 476550
    },
    {
      "epoch": 195.07,
      "grad_norm": 1.8841451406478882,
      "learning_rate": 7.777491282454171e-09,
      "loss": 0.4408,
      "step": 476560
    },
    {
      "epoch": 195.08,
      "grad_norm": 2.1826326847076416,
      "learning_rate": 7.764584362132848e-09,
      "loss": 0.4468,
      "step": 476570
    },
    {
      "epoch": 195.08,
      "grad_norm": 3.0844533443450928,
      "learning_rate": 7.751688142598918e-09,
      "loss": 0.4509,
      "step": 476580
    },
    {
      "epoch": 195.08,
      "grad_norm": 2.104365587234497,
      "learning_rate": 7.738802623911446e-09,
      "loss": 0.4498,
      "step": 476590
    },
    {
      "epoch": 195.09,
      "grad_norm": 1.991690754890442,
      "learning_rate": 7.7259278061295e-09,
      "loss": 0.4442,
      "step": 476600
    },
    {
      "epoch": 195.09,
      "grad_norm": 2.6004559993743896,
      "learning_rate": 7.713063689312148e-09,
      "loss": 0.4748,
      "step": 476610
    },
    {
      "epoch": 195.1,
      "grad_norm": 2.0565226078033447,
      "learning_rate": 7.700210273518195e-09,
      "loss": 0.4462,
      "step": 476620
    },
    {
      "epoch": 195.1,
      "grad_norm": 2.133748769760132,
      "learning_rate": 7.687367558806453e-09,
      "loss": 0.4487,
      "step": 476630
    },
    {
      "epoch": 195.1,
      "grad_norm": 2.044625759124756,
      "learning_rate": 7.674535545235983e-09,
      "loss": 0.4441,
      "step": 476640
    },
    {
      "epoch": 195.11,
      "grad_norm": 1.8885117769241333,
      "learning_rate": 7.661714232865338e-09,
      "loss": 0.4369,
      "step": 476650
    },
    {
      "epoch": 195.11,
      "grad_norm": 2.21589994430542,
      "learning_rate": 7.64890362175332e-09,
      "loss": 0.4434,
      "step": 476660
    },
    {
      "epoch": 195.12,
      "grad_norm": 2.335103988647461,
      "learning_rate": 7.63610371195874e-09,
      "loss": 0.4574,
      "step": 476670
    },
    {
      "epoch": 195.12,
      "grad_norm": 2.4628076553344727,
      "learning_rate": 7.623314503539882e-09,
      "loss": 0.4448,
      "step": 476680
    },
    {
      "epoch": 195.12,
      "grad_norm": 2.28902006149292,
      "learning_rate": 7.610535996555814e-09,
      "loss": 0.4372,
      "step": 476690
    },
    {
      "epoch": 195.13,
      "grad_norm": 1.9993641376495361,
      "learning_rate": 7.597768191064825e-09,
      "loss": 0.4341,
      "step": 476700
    },
    {
      "epoch": 195.13,
      "grad_norm": 2.169053077697754,
      "learning_rate": 7.585011087125198e-09,
      "loss": 0.445,
      "step": 476710
    },
    {
      "epoch": 195.14,
      "grad_norm": 1.9912089109420776,
      "learning_rate": 7.572264684795742e-09,
      "loss": 0.4337,
      "step": 476720
    },
    {
      "epoch": 195.14,
      "grad_norm": 1.7733471393585205,
      "learning_rate": 7.559528984134745e-09,
      "loss": 0.4264,
      "step": 476730
    },
    {
      "epoch": 195.15,
      "grad_norm": 1.8677515983581543,
      "learning_rate": 7.546803985200492e-09,
      "loss": 0.4277,
      "step": 476740
    },
    {
      "epoch": 195.15,
      "grad_norm": 1.8760727643966675,
      "learning_rate": 7.534089688051269e-09,
      "loss": 0.464,
      "step": 476750
    },
    {
      "epoch": 195.15,
      "grad_norm": 2.1878154277801514,
      "learning_rate": 7.521386092745104e-09,
      "loss": 0.4516,
      "step": 476760
    },
    {
      "epoch": 195.16,
      "grad_norm": 1.9405319690704346,
      "learning_rate": 7.508693199340543e-09,
      "loss": 0.4645,
      "step": 476770
    },
    {
      "epoch": 195.16,
      "grad_norm": 1.8956358432769775,
      "learning_rate": 7.496011007895874e-09,
      "loss": 0.4433,
      "step": 476780
    },
    {
      "epoch": 195.17,
      "grad_norm": 2.0678179264068604,
      "learning_rate": 7.483339518468603e-09,
      "loss": 0.4539,
      "step": 476790
    },
    {
      "epoch": 195.17,
      "grad_norm": 2.118968963623047,
      "learning_rate": 7.470678731117016e-09,
      "loss": 0.4461,
      "step": 476800
    },
    {
      "epoch": 195.17,
      "grad_norm": 2.2591795921325684,
      "learning_rate": 7.458028645899397e-09,
      "loss": 0.45,
      "step": 476810
    },
    {
      "epoch": 195.18,
      "grad_norm": 1.8678791522979736,
      "learning_rate": 7.445389262873257e-09,
      "loss": 0.4324,
      "step": 476820
    },
    {
      "epoch": 195.18,
      "grad_norm": 1.8833577632904053,
      "learning_rate": 7.432760582096881e-09,
      "loss": 0.4329,
      "step": 476830
    },
    {
      "epoch": 195.19,
      "grad_norm": 2.3876352310180664,
      "learning_rate": 7.420142603628034e-09,
      "loss": 0.41,
      "step": 476840
    },
    {
      "epoch": 195.19,
      "grad_norm": 1.978416919708252,
      "learning_rate": 7.407535327524222e-09,
      "loss": 0.4344,
      "step": 476850
    },
    {
      "epoch": 195.19,
      "grad_norm": 2.005026340484619,
      "learning_rate": 7.3949387538437335e-09,
      "loss": 0.4579,
      "step": 476860
    },
    {
      "epoch": 195.2,
      "grad_norm": 1.8762762546539307,
      "learning_rate": 7.382352882643553e-09,
      "loss": 0.4299,
      "step": 476870
    },
    {
      "epoch": 195.2,
      "grad_norm": 2.1138112545013428,
      "learning_rate": 7.369777713981967e-09,
      "loss": 0.4569,
      "step": 476880
    },
    {
      "epoch": 195.21,
      "grad_norm": 1.9586976766586304,
      "learning_rate": 7.357213247916223e-09,
      "loss": 0.4434,
      "step": 476890
    },
    {
      "epoch": 195.21,
      "grad_norm": 1.8801060914993286,
      "learning_rate": 7.344659484504085e-09,
      "loss": 0.448,
      "step": 476900
    },
    {
      "epoch": 195.21,
      "grad_norm": 1.7752327919006348,
      "learning_rate": 7.332116423803061e-09,
      "loss": 0.4581,
      "step": 476910
    },
    {
      "epoch": 195.22,
      "grad_norm": 1.9791475534439087,
      "learning_rate": 7.319584065870396e-09,
      "loss": 0.4467,
      "step": 476920
    },
    {
      "epoch": 195.22,
      "grad_norm": 2.7714314460754395,
      "learning_rate": 7.307062410763596e-09,
      "loss": 0.4295,
      "step": 476930
    },
    {
      "epoch": 195.23,
      "grad_norm": 1.8671857118606567,
      "learning_rate": 7.294551458540167e-09,
      "loss": 0.4512,
      "step": 476940
    },
    {
      "epoch": 195.23,
      "grad_norm": 2.6571848392486572,
      "learning_rate": 7.282051209257355e-09,
      "loss": 0.4489,
      "step": 476950
    },
    {
      "epoch": 195.24,
      "grad_norm": 2.4729583263397217,
      "learning_rate": 7.269561662972146e-09,
      "loss": 0.4421,
      "step": 476960
    },
    {
      "epoch": 195.24,
      "grad_norm": 1.6981505155563354,
      "learning_rate": 7.257082819742046e-09,
      "loss": 0.4419,
      "step": 476970
    },
    {
      "epoch": 195.24,
      "grad_norm": 2.0316779613494873,
      "learning_rate": 7.244614679624301e-09,
      "loss": 0.4295,
      "step": 476980
    },
    {
      "epoch": 195.25,
      "grad_norm": 2.0397284030914307,
      "learning_rate": 7.232157242675635e-09,
      "loss": 0.4445,
      "step": 476990
    },
    {
      "epoch": 195.25,
      "grad_norm": 2.0085530281066895,
      "learning_rate": 7.219710508953557e-09,
      "loss": 0.4622,
      "step": 477000
    },
    {
      "epoch": 195.26,
      "grad_norm": 2.2016592025756836,
      "learning_rate": 7.2072744785147895e-09,
      "loss": 0.4419,
      "step": 477010
    },
    {
      "epoch": 195.26,
      "grad_norm": 1.9965194463729858,
      "learning_rate": 7.194849151416319e-09,
      "loss": 0.4467,
      "step": 477020
    },
    {
      "epoch": 195.26,
      "grad_norm": 2.237401247024536,
      "learning_rate": 7.182434527715133e-09,
      "loss": 0.4364,
      "step": 477030
    },
    {
      "epoch": 195.27,
      "grad_norm": 1.9103732109069824,
      "learning_rate": 7.170030607468215e-09,
      "loss": 0.4184,
      "step": 477040
    },
    {
      "epoch": 195.27,
      "grad_norm": 1.861664891242981,
      "learning_rate": 7.157637390732291e-09,
      "loss": 0.4461,
      "step": 477050
    },
    {
      "epoch": 195.28,
      "grad_norm": 2.405048370361328,
      "learning_rate": 7.145254877564087e-09,
      "loss": 0.4582,
      "step": 477060
    },
    {
      "epoch": 195.28,
      "grad_norm": 1.7698129415512085,
      "learning_rate": 7.132883068020328e-09,
      "loss": 0.4488,
      "step": 477070
    },
    {
      "epoch": 195.28,
      "grad_norm": 2.307039499282837,
      "learning_rate": 7.12052196215774e-09,
      "loss": 0.4394,
      "step": 477080
    },
    {
      "epoch": 195.29,
      "grad_norm": 1.5115001201629639,
      "learning_rate": 7.108171560032787e-09,
      "loss": 0.4541,
      "step": 477090
    },
    {
      "epoch": 195.29,
      "grad_norm": 2.0810322761535645,
      "learning_rate": 7.0958318617024555e-09,
      "loss": 0.4323,
      "step": 477100
    },
    {
      "epoch": 195.3,
      "grad_norm": 1.9966397285461426,
      "learning_rate": 7.083502867222691e-09,
      "loss": 0.4399,
      "step": 477110
    },
    {
      "epoch": 195.3,
      "grad_norm": 1.8559879064559937,
      "learning_rate": 7.071184576650478e-09,
      "loss": 0.441,
      "step": 477120
    },
    {
      "epoch": 195.3,
      "grad_norm": 1.876571536064148,
      "learning_rate": 7.058876990041762e-09,
      "loss": 0.4329,
      "step": 477130
    },
    {
      "epoch": 195.31,
      "grad_norm": 1.9340018033981323,
      "learning_rate": 7.046580107453528e-09,
      "loss": 0.4388,
      "step": 477140
    },
    {
      "epoch": 195.31,
      "grad_norm": 1.9282788038253784,
      "learning_rate": 7.034293928941462e-09,
      "loss": 0.4319,
      "step": 477150
    },
    {
      "epoch": 195.32,
      "grad_norm": 2.186288833618164,
      "learning_rate": 7.022018454562548e-09,
      "loss": 0.4326,
      "step": 477160
    },
    {
      "epoch": 195.32,
      "grad_norm": 2.1476190090179443,
      "learning_rate": 7.009753684372212e-09,
      "loss": 0.4645,
      "step": 477170
    },
    {
      "epoch": 195.33,
      "grad_norm": 2.061861753463745,
      "learning_rate": 6.997499618427438e-09,
      "loss": 0.4445,
      "step": 477180
    },
    {
      "epoch": 195.33,
      "grad_norm": 2.080505847930908,
      "learning_rate": 6.985256256783651e-09,
      "loss": 0.4361,
      "step": 477190
    },
    {
      "epoch": 195.33,
      "grad_norm": 1.9530751705169678,
      "learning_rate": 6.973023599497317e-09,
      "loss": 0.4268,
      "step": 477200
    },
    {
      "epoch": 195.34,
      "grad_norm": 1.914882779121399,
      "learning_rate": 6.9608016466246414e-09,
      "loss": 0.4463,
      "step": 477210
    },
    {
      "epoch": 195.34,
      "grad_norm": 1.9020531177520752,
      "learning_rate": 6.948590398221307e-09,
      "loss": 0.4585,
      "step": 477220
    },
    {
      "epoch": 195.35,
      "grad_norm": 1.9826492071151733,
      "learning_rate": 6.93638985434326e-09,
      "loss": 0.4292,
      "step": 477230
    },
    {
      "epoch": 195.35,
      "grad_norm": 1.7018098831176758,
      "learning_rate": 6.924200015046445e-09,
      "loss": 0.4504,
      "step": 477240
    },
    {
      "epoch": 195.35,
      "grad_norm": 2.393026828765869,
      "learning_rate": 6.912020880386806e-09,
      "loss": 0.4551,
      "step": 477250
    },
    {
      "epoch": 195.36,
      "grad_norm": 2.2646000385284424,
      "learning_rate": 6.899852450419768e-09,
      "loss": 0.4278,
      "step": 477260
    },
    {
      "epoch": 195.36,
      "grad_norm": 2.345226764678955,
      "learning_rate": 6.887694725201797e-09,
      "loss": 0.4367,
      "step": 477270
    },
    {
      "epoch": 195.37,
      "grad_norm": 1.9912686347961426,
      "learning_rate": 6.875547704787796e-09,
      "loss": 0.446,
      "step": 477280
    },
    {
      "epoch": 195.37,
      "grad_norm": 1.8218451738357544,
      "learning_rate": 6.863411389233971e-09,
      "loss": 0.4433,
      "step": 477290
    },
    {
      "epoch": 195.37,
      "grad_norm": 1.7747756242752075,
      "learning_rate": 6.851285778595485e-09,
      "loss": 0.4302,
      "step": 477300
    },
    {
      "epoch": 195.38,
      "grad_norm": 2.5164403915405273,
      "learning_rate": 6.8391708729282835e-09,
      "loss": 0.4584,
      "step": 477310
    },
    {
      "epoch": 195.38,
      "grad_norm": 2.0140247344970703,
      "learning_rate": 6.827066672287531e-09,
      "loss": 0.4421,
      "step": 477320
    },
    {
      "epoch": 195.39,
      "grad_norm": 2.254030466079712,
      "learning_rate": 6.81497317672891e-09,
      "loss": 0.4512,
      "step": 477330
    },
    {
      "epoch": 195.39,
      "grad_norm": 2.014279365539551,
      "learning_rate": 6.802890386307589e-09,
      "loss": 0.4519,
      "step": 477340
    },
    {
      "epoch": 195.4,
      "grad_norm": 1.9062469005584717,
      "learning_rate": 6.790818301079248e-09,
      "loss": 0.4508,
      "step": 477350
    },
    {
      "epoch": 195.4,
      "grad_norm": 2.0708439350128174,
      "learning_rate": 6.778756921098794e-09,
      "loss": 0.434,
      "step": 477360
    },
    {
      "epoch": 195.4,
      "grad_norm": 2.4521310329437256,
      "learning_rate": 6.766706246421651e-09,
      "loss": 0.4484,
      "step": 477370
    },
    {
      "epoch": 195.41,
      "grad_norm": 2.392517566680908,
      "learning_rate": 6.754666277103242e-09,
      "loss": 0.438,
      "step": 477380
    },
    {
      "epoch": 195.41,
      "grad_norm": 2.294386386871338,
      "learning_rate": 6.742637013198472e-09,
      "loss": 0.4594,
      "step": 477390
    },
    {
      "epoch": 195.42,
      "grad_norm": 2.0082836151123047,
      "learning_rate": 6.730618454762506e-09,
      "loss": 0.4498,
      "step": 477400
    },
    {
      "epoch": 195.42,
      "grad_norm": 2.157944679260254,
      "learning_rate": 6.718610601850506e-09,
      "loss": 0.4436,
      "step": 477410
    },
    {
      "epoch": 195.42,
      "grad_norm": 2.00350022315979,
      "learning_rate": 6.706613454517118e-09,
      "loss": 0.4395,
      "step": 477420
    },
    {
      "epoch": 195.43,
      "grad_norm": 1.8204951286315918,
      "learning_rate": 6.694627012817765e-09,
      "loss": 0.4519,
      "step": 477430
    },
    {
      "epoch": 195.43,
      "grad_norm": 2.0762977600097656,
      "learning_rate": 6.682651276807092e-09,
      "loss": 0.45,
      "step": 477440
    },
    {
      "epoch": 195.44,
      "grad_norm": 2.1716411113739014,
      "learning_rate": 6.670686246539741e-09,
      "loss": 0.4378,
      "step": 477450
    },
    {
      "epoch": 195.44,
      "grad_norm": 2.195284128189087,
      "learning_rate": 6.658731922071138e-09,
      "loss": 0.4428,
      "step": 477460
    },
    {
      "epoch": 195.44,
      "grad_norm": 2.3247368335723877,
      "learning_rate": 6.646788303455407e-09,
      "loss": 0.4468,
      "step": 477470
    },
    {
      "epoch": 195.45,
      "grad_norm": 2.378272533416748,
      "learning_rate": 6.63485539074771e-09,
      "loss": 0.4579,
      "step": 477480
    },
    {
      "epoch": 195.45,
      "grad_norm": 2.127286195755005,
      "learning_rate": 6.622933184002692e-09,
      "loss": 0.438,
      "step": 477490
    },
    {
      "epoch": 195.46,
      "grad_norm": 2.322967529296875,
      "learning_rate": 6.6110216832744766e-09,
      "loss": 0.4552,
      "step": 477500
    },
    {
      "epoch": 195.46,
      "grad_norm": 1.9449535608291626,
      "learning_rate": 6.599120888618227e-09,
      "loss": 0.448,
      "step": 477510
    },
    {
      "epoch": 195.46,
      "grad_norm": 2.2924578189849854,
      "learning_rate": 6.587230800088068e-09,
      "loss": 0.45,
      "step": 477520
    },
    {
      "epoch": 195.47,
      "grad_norm": 2.035569667816162,
      "learning_rate": 6.575351417738642e-09,
      "loss": 0.4481,
      "step": 477530
    },
    {
      "epoch": 195.47,
      "grad_norm": 1.9384511709213257,
      "learning_rate": 6.563482741624333e-09,
      "loss": 0.4441,
      "step": 477540
    },
    {
      "epoch": 195.48,
      "grad_norm": 1.8575485944747925,
      "learning_rate": 6.551624771799525e-09,
      "loss": 0.4303,
      "step": 477550
    },
    {
      "epoch": 195.48,
      "grad_norm": 2.0780396461486816,
      "learning_rate": 6.539777508318602e-09,
      "loss": 0.4265,
      "step": 477560
    },
    {
      "epoch": 195.49,
      "grad_norm": 2.1542046070098877,
      "learning_rate": 6.5279409512354265e-09,
      "loss": 0.4385,
      "step": 477570
    },
    {
      "epoch": 195.49,
      "grad_norm": 2.128723382949829,
      "learning_rate": 6.516115100604902e-09,
      "loss": 0.4617,
      "step": 477580
    },
    {
      "epoch": 195.49,
      "grad_norm": 2.0176303386688232,
      "learning_rate": 6.504299956480632e-09,
      "loss": 0.4487,
      "step": 477590
    },
    {
      "epoch": 195.5,
      "grad_norm": 1.9914242029190063,
      "learning_rate": 6.492495518917e-09,
      "loss": 0.4621,
      "step": 477600
    },
    {
      "epoch": 195.5,
      "grad_norm": 1.5544369220733643,
      "learning_rate": 6.480701787967869e-09,
      "loss": 0.4482,
      "step": 477610
    },
    {
      "epoch": 195.51,
      "grad_norm": 2.3656656742095947,
      "learning_rate": 6.4689187636876236e-09,
      "loss": 0.4513,
      "step": 477620
    },
    {
      "epoch": 195.51,
      "grad_norm": 2.1477444171905518,
      "learning_rate": 6.457146446129866e-09,
      "loss": 0.4576,
      "step": 477630
    },
    {
      "epoch": 195.51,
      "grad_norm": 1.7628682851791382,
      "learning_rate": 6.445384835348979e-09,
      "loss": 0.4406,
      "step": 477640
    },
    {
      "epoch": 195.52,
      "grad_norm": 1.7722078561782837,
      "learning_rate": 6.433633931398306e-09,
      "loss": 0.4255,
      "step": 477650
    },
    {
      "epoch": 195.52,
      "grad_norm": 1.965741753578186,
      "learning_rate": 6.421893734332231e-09,
      "loss": 0.4398,
      "step": 477660
    },
    {
      "epoch": 195.53,
      "grad_norm": 1.6794023513793945,
      "learning_rate": 6.4101642442038365e-09,
      "loss": 0.4385,
      "step": 477670
    },
    {
      "epoch": 195.53,
      "grad_norm": 1.9513176679611206,
      "learning_rate": 6.3984454610675055e-09,
      "loss": 0.448,
      "step": 477680
    },
    {
      "epoch": 195.53,
      "grad_norm": 1.9940778017044067,
      "learning_rate": 6.386737384976581e-09,
      "loss": 0.4635,
      "step": 477690
    },
    {
      "epoch": 195.54,
      "grad_norm": 2.4438276290893555,
      "learning_rate": 6.375040015984666e-09,
      "loss": 0.4562,
      "step": 477700
    },
    {
      "epoch": 195.54,
      "grad_norm": 1.9647436141967773,
      "learning_rate": 6.363353354145624e-09,
      "loss": 0.4368,
      "step": 477710
    },
    {
      "epoch": 195.55,
      "grad_norm": 1.9893405437469482,
      "learning_rate": 6.351677399512797e-09,
      "loss": 0.4483,
      "step": 477720
    },
    {
      "epoch": 195.55,
      "grad_norm": 1.8346495628356934,
      "learning_rate": 6.340012152139529e-09,
      "loss": 0.4492,
      "step": 477730
    },
    {
      "epoch": 195.55,
      "grad_norm": 1.6522653102874756,
      "learning_rate": 6.328357612079681e-09,
      "loss": 0.4398,
      "step": 477740
    },
    {
      "epoch": 195.56,
      "grad_norm": 1.8597804307937622,
      "learning_rate": 6.316713779386078e-09,
      "loss": 0.4463,
      "step": 477750
    },
    {
      "epoch": 195.56,
      "grad_norm": 2.1939797401428223,
      "learning_rate": 6.305080654112581e-09,
      "loss": 0.4483,
      "step": 477760
    },
    {
      "epoch": 195.57,
      "grad_norm": 2.238311529159546,
      "learning_rate": 6.293458236312273e-09,
      "loss": 0.4387,
      "step": 477770
    },
    {
      "epoch": 195.57,
      "grad_norm": 2.0521974563598633,
      "learning_rate": 6.281846526038237e-09,
      "loss": 0.4322,
      "step": 477780
    },
    {
      "epoch": 195.58,
      "grad_norm": 2.2402396202087402,
      "learning_rate": 6.270245523343817e-09,
      "loss": 0.4573,
      "step": 477790
    },
    {
      "epoch": 195.58,
      "grad_norm": 1.8980071544647217,
      "learning_rate": 6.2586552282823525e-09,
      "loss": 0.4479,
      "step": 477800
    },
    {
      "epoch": 195.58,
      "grad_norm": 2.018904447555542,
      "learning_rate": 6.2470756409066684e-09,
      "loss": 0.4698,
      "step": 477810
    },
    {
      "epoch": 195.59,
      "grad_norm": 1.937920093536377,
      "learning_rate": 6.235506761269847e-09,
      "loss": 0.4152,
      "step": 477820
    },
    {
      "epoch": 195.59,
      "grad_norm": 2.2414307594299316,
      "learning_rate": 6.2239485894247095e-09,
      "loss": 0.4393,
      "step": 477830
    },
    {
      "epoch": 195.6,
      "grad_norm": 1.8837571144104004,
      "learning_rate": 6.2124011254246e-09,
      "loss": 0.439,
      "step": 477840
    },
    {
      "epoch": 195.6,
      "grad_norm": 2.1969962120056152,
      "learning_rate": 6.20086436932208e-09,
      "loss": 0.4499,
      "step": 477850
    },
    {
      "epoch": 195.6,
      "grad_norm": 1.6552895307540894,
      "learning_rate": 6.1893383211702315e-09,
      "loss": 0.4584,
      "step": 477860
    },
    {
      "epoch": 195.61,
      "grad_norm": 2.090275287628174,
      "learning_rate": 6.1778229810218775e-09,
      "loss": 0.457,
      "step": 477870
    },
    {
      "epoch": 195.61,
      "grad_norm": 1.9633402824401855,
      "learning_rate": 6.16631834892958e-09,
      "loss": 0.4416,
      "step": 477880
    },
    {
      "epoch": 195.62,
      "grad_norm": 1.792526364326477,
      "learning_rate": 6.154824424945901e-09,
      "loss": 0.446,
      "step": 477890
    },
    {
      "epoch": 195.62,
      "grad_norm": 2.4493885040283203,
      "learning_rate": 6.1433412091239225e-09,
      "loss": 0.4561,
      "step": 477900
    },
    {
      "epoch": 195.62,
      "grad_norm": 1.910370111465454,
      "learning_rate": 6.131868701515948e-09,
      "loss": 0.4593,
      "step": 477910
    },
    {
      "epoch": 195.63,
      "grad_norm": 2.3099913597106934,
      "learning_rate": 6.120406902174538e-09,
      "loss": 0.4231,
      "step": 477920
    },
    {
      "epoch": 195.63,
      "grad_norm": 1.8262323141098022,
      "learning_rate": 6.108955811152515e-09,
      "loss": 0.4531,
      "step": 477930
    },
    {
      "epoch": 195.64,
      "grad_norm": 1.8118257522583008,
      "learning_rate": 6.097515428501922e-09,
      "loss": 0.4221,
      "step": 477940
    },
    {
      "epoch": 195.64,
      "grad_norm": 2.379741668701172,
      "learning_rate": 6.08608575427532e-09,
      "loss": 0.4467,
      "step": 477950
    },
    {
      "epoch": 195.64,
      "grad_norm": 2.20566725730896,
      "learning_rate": 6.074666788525012e-09,
      "loss": 0.4503,
      "step": 477960
    },
    {
      "epoch": 195.65,
      "grad_norm": 2.2226476669311523,
      "learning_rate": 6.063258531303298e-09,
      "loss": 0.4378,
      "step": 477970
    },
    {
      "epoch": 195.65,
      "grad_norm": 2.2946479320526123,
      "learning_rate": 6.051860982662742e-09,
      "loss": 0.4493,
      "step": 477980
    },
    {
      "epoch": 195.66,
      "grad_norm": 1.945503830909729,
      "learning_rate": 6.040474142655124e-09,
      "loss": 0.4426,
      "step": 477990
    },
    {
      "epoch": 195.66,
      "grad_norm": 2.611598253250122,
      "learning_rate": 6.029098011333007e-09,
      "loss": 0.4568,
      "step": 478000
    },
    {
      "epoch": 195.67,
      "grad_norm": 2.677536964416504,
      "learning_rate": 6.017732588748173e-09,
      "loss": 0.44,
      "step": 478010
    },
    {
      "epoch": 195.67,
      "grad_norm": 2.7704362869262695,
      "learning_rate": 6.006377874952922e-09,
      "loss": 0.4474,
      "step": 478020
    },
    {
      "epoch": 195.67,
      "grad_norm": 2.311964750289917,
      "learning_rate": 5.995033869999038e-09,
      "loss": 0.4278,
      "step": 478030
    },
    {
      "epoch": 195.68,
      "grad_norm": 1.8320143222808838,
      "learning_rate": 5.983700573938561e-09,
      "loss": 0.4396,
      "step": 478040
    },
    {
      "epoch": 195.68,
      "grad_norm": 2.056649923324585,
      "learning_rate": 5.972377986823533e-09,
      "loss": 0.4422,
      "step": 478050
    },
    {
      "epoch": 195.69,
      "grad_norm": 2.115626811981201,
      "learning_rate": 5.961066108705996e-09,
      "loss": 0.4439,
      "step": 478060
    },
    {
      "epoch": 195.69,
      "grad_norm": 2.1320087909698486,
      "learning_rate": 5.949764939637211e-09,
      "loss": 0.4202,
      "step": 478070
    },
    {
      "epoch": 195.69,
      "grad_norm": 2.347585916519165,
      "learning_rate": 5.93847447966922e-09,
      "loss": 0.4574,
      "step": 478080
    },
    {
      "epoch": 195.7,
      "grad_norm": 1.6979146003723145,
      "learning_rate": 5.927194728853803e-09,
      "loss": 0.4333,
      "step": 478090
    },
    {
      "epoch": 195.7,
      "grad_norm": 1.9049696922302246,
      "learning_rate": 5.9159256872427435e-09,
      "loss": 0.4439,
      "step": 478100
    },
    {
      "epoch": 195.71,
      "grad_norm": 2.8059046268463135,
      "learning_rate": 5.9046673548875625e-09,
      "loss": 0.4421,
      "step": 478110
    },
    {
      "epoch": 195.71,
      "grad_norm": 1.9342586994171143,
      "learning_rate": 5.893419731839519e-09,
      "loss": 0.4305,
      "step": 478120
    },
    {
      "epoch": 195.71,
      "grad_norm": 2.244403839111328,
      "learning_rate": 5.882182818150658e-09,
      "loss": 0.4468,
      "step": 478130
    },
    {
      "epoch": 195.72,
      "grad_norm": 2.1757664680480957,
      "learning_rate": 5.870956613871977e-09,
      "loss": 0.4567,
      "step": 478140
    },
    {
      "epoch": 195.72,
      "grad_norm": 2.314885139465332,
      "learning_rate": 5.8597411190552595e-09,
      "loss": 0.4361,
      "step": 478150
    },
    {
      "epoch": 195.73,
      "grad_norm": 1.9166498184204102,
      "learning_rate": 5.848536333751767e-09,
      "loss": 0.4274,
      "step": 478160
    },
    {
      "epoch": 195.73,
      "grad_norm": 2.050816535949707,
      "learning_rate": 5.837342258012759e-09,
      "loss": 0.4534,
      "step": 478170
    },
    {
      "epoch": 195.73,
      "grad_norm": 2.0068578720092773,
      "learning_rate": 5.826158891889497e-09,
      "loss": 0.4572,
      "step": 478180
    },
    {
      "epoch": 195.74,
      "grad_norm": 1.9767473936080933,
      "learning_rate": 5.814986235433243e-09,
      "loss": 0.4301,
      "step": 478190
    },
    {
      "epoch": 195.74,
      "grad_norm": 2.2832748889923096,
      "learning_rate": 5.803824288695518e-09,
      "loss": 0.4465,
      "step": 478200
    },
    {
      "epoch": 195.75,
      "grad_norm": 2.156947612762451,
      "learning_rate": 5.792673051726801e-09,
      "loss": 0.452,
      "step": 478210
    },
    {
      "epoch": 195.75,
      "grad_norm": 1.8242206573486328,
      "learning_rate": 5.781532524578875e-09,
      "loss": 0.4164,
      "step": 478220
    },
    {
      "epoch": 195.76,
      "grad_norm": 1.8378651142120361,
      "learning_rate": 5.7704027073022205e-09,
      "loss": 0.4489,
      "step": 478230
    },
    {
      "epoch": 195.76,
      "grad_norm": 2.2129077911376953,
      "learning_rate": 5.759283599947838e-09,
      "loss": 0.419,
      "step": 478240
    },
    {
      "epoch": 195.76,
      "grad_norm": 2.13272762298584,
      "learning_rate": 5.7481752025672485e-09,
      "loss": 0.4655,
      "step": 478250
    },
    {
      "epoch": 195.77,
      "grad_norm": 2.164445638656616,
      "learning_rate": 5.737077515210672e-09,
      "loss": 0.4497,
      "step": 478260
    },
    {
      "epoch": 195.77,
      "grad_norm": 2.499025344848633,
      "learning_rate": 5.725990537929111e-09,
      "loss": 0.4649,
      "step": 478270
    },
    {
      "epoch": 195.78,
      "grad_norm": 1.9793148040771484,
      "learning_rate": 5.714914270773564e-09,
      "loss": 0.4452,
      "step": 478280
    },
    {
      "epoch": 195.78,
      "grad_norm": 2.5070996284484863,
      "learning_rate": 5.7038487137947735e-09,
      "loss": 0.4349,
      "step": 478290
    },
    {
      "epoch": 195.78,
      "grad_norm": 2.3930559158325195,
      "learning_rate": 5.692793867043219e-09,
      "loss": 0.4468,
      "step": 478300
    },
    {
      "epoch": 195.79,
      "grad_norm": 2.0982186794281006,
      "learning_rate": 5.681749730569642e-09,
      "loss": 0.4733,
      "step": 478310
    },
    {
      "epoch": 195.79,
      "grad_norm": 1.8494114875793457,
      "learning_rate": 5.670716304424523e-09,
      "loss": 0.45,
      "step": 478320
    },
    {
      "epoch": 195.8,
      "grad_norm": 1.777076005935669,
      "learning_rate": 5.659693588658601e-09,
      "loss": 0.4317,
      "step": 478330
    },
    {
      "epoch": 195.8,
      "grad_norm": 2.2561047077178955,
      "learning_rate": 5.6486815833220975e-09,
      "loss": 0.4382,
      "step": 478340
    },
    {
      "epoch": 195.8,
      "grad_norm": 2.5571060180664062,
      "learning_rate": 5.637680288466013e-09,
      "loss": 0.4318,
      "step": 478350
    },
    {
      "epoch": 195.81,
      "grad_norm": 2.019535541534424,
      "learning_rate": 5.626689704140049e-09,
      "loss": 0.4359,
      "step": 478360
    },
    {
      "epoch": 195.81,
      "grad_norm": 4.289093971252441,
      "learning_rate": 5.615709830394943e-09,
      "loss": 0.4379,
      "step": 478370
    },
    {
      "epoch": 195.82,
      "grad_norm": 2.075867176055908,
      "learning_rate": 5.604740667280918e-09,
      "loss": 0.4346,
      "step": 478380
    },
    {
      "epoch": 195.82,
      "grad_norm": 1.9577016830444336,
      "learning_rate": 5.593782214848453e-09,
      "loss": 0.4491,
      "step": 478390
    },
    {
      "epoch": 195.82,
      "grad_norm": 2.257336378097534,
      "learning_rate": 5.582834473147249e-09,
      "loss": 0.4453,
      "step": 478400
    },
    {
      "epoch": 195.83,
      "grad_norm": 2.630309581756592,
      "learning_rate": 5.571897442227784e-09,
      "loss": 0.4599,
      "step": 478410
    },
    {
      "epoch": 195.83,
      "grad_norm": 1.8643065690994263,
      "learning_rate": 5.560971122140021e-09,
      "loss": 0.4325,
      "step": 478420
    },
    {
      "epoch": 195.84,
      "grad_norm": 2.2463271617889404,
      "learning_rate": 5.550055512934179e-09,
      "loss": 0.4516,
      "step": 478430
    },
    {
      "epoch": 195.84,
      "grad_norm": 2.114291191101074,
      "learning_rate": 5.5391506146599574e-09,
      "loss": 0.4353,
      "step": 478440
    },
    {
      "epoch": 195.85,
      "grad_norm": 2.1918416023254395,
      "learning_rate": 5.528256427367837e-09,
      "loss": 0.4327,
      "step": 478450
    },
    {
      "epoch": 195.85,
      "grad_norm": 2.4034714698791504,
      "learning_rate": 5.517372951107258e-09,
      "loss": 0.4484,
      "step": 478460
    },
    {
      "epoch": 195.85,
      "grad_norm": 3.1397459506988525,
      "learning_rate": 5.506500185928179e-09,
      "loss": 0.4439,
      "step": 478470
    },
    {
      "epoch": 195.86,
      "grad_norm": 2.231269121170044,
      "learning_rate": 5.495638131880562e-09,
      "loss": 0.434,
      "step": 478480
    },
    {
      "epoch": 195.86,
      "grad_norm": 2.4768528938293457,
      "learning_rate": 5.484786789013845e-09,
      "loss": 0.4258,
      "step": 478490
    },
    {
      "epoch": 195.87,
      "grad_norm": 2.6113929748535156,
      "learning_rate": 5.473946157378249e-09,
      "loss": 0.4674,
      "step": 478500
    },
    {
      "epoch": 195.87,
      "grad_norm": 1.955709457397461,
      "learning_rate": 5.463116237022693e-09,
      "loss": 0.4776,
      "step": 478510
    },
    {
      "epoch": 195.87,
      "grad_norm": 1.9413361549377441,
      "learning_rate": 5.452297027997658e-09,
      "loss": 0.4221,
      "step": 478520
    },
    {
      "epoch": 195.88,
      "grad_norm": 1.9567071199417114,
      "learning_rate": 5.4414885303520616e-09,
      "loss": 0.4386,
      "step": 478530
    },
    {
      "epoch": 195.88,
      "grad_norm": 2.200005054473877,
      "learning_rate": 5.430690744135605e-09,
      "loss": 0.4662,
      "step": 478540
    },
    {
      "epoch": 195.89,
      "grad_norm": 1.936868667602539,
      "learning_rate": 5.419903669397988e-09,
      "loss": 0.4411,
      "step": 478550
    },
    {
      "epoch": 195.89,
      "grad_norm": 1.9148898124694824,
      "learning_rate": 5.40912730618813e-09,
      "loss": 0.4259,
      "step": 478560
    },
    {
      "epoch": 195.89,
      "grad_norm": 2.195241689682007,
      "learning_rate": 5.3983616545557296e-09,
      "loss": 0.455,
      "step": 478570
    },
    {
      "epoch": 195.9,
      "grad_norm": 1.5415867567062378,
      "learning_rate": 5.387606714550227e-09,
      "loss": 0.4337,
      "step": 478580
    },
    {
      "epoch": 195.9,
      "grad_norm": 1.9690159559249878,
      "learning_rate": 5.376862486220542e-09,
      "loss": 0.4447,
      "step": 478590
    },
    {
      "epoch": 195.91,
      "grad_norm": 1.7888214588165283,
      "learning_rate": 5.366128969616114e-09,
      "loss": 0.4609,
      "step": 478600
    },
    {
      "epoch": 195.91,
      "grad_norm": 2.5395703315734863,
      "learning_rate": 5.355406164786122e-09,
      "loss": 0.4401,
      "step": 478610
    },
    {
      "epoch": 195.91,
      "grad_norm": 2.136235475540161,
      "learning_rate": 5.344694071779485e-09,
      "loss": 0.4462,
      "step": 478620
    },
    {
      "epoch": 195.92,
      "grad_norm": 1.7859036922454834,
      "learning_rate": 5.333992690645383e-09,
      "loss": 0.4166,
      "step": 478630
    },
    {
      "epoch": 195.92,
      "grad_norm": 3.359510898590088,
      "learning_rate": 5.323302021432996e-09,
      "loss": 0.4572,
      "step": 478640
    },
    {
      "epoch": 195.93,
      "grad_norm": 1.6650755405426025,
      "learning_rate": 5.312622064190981e-09,
      "loss": 0.4465,
      "step": 478650
    },
    {
      "epoch": 195.93,
      "grad_norm": 2.1326699256896973,
      "learning_rate": 5.30195281896878e-09,
      "loss": 0.4364,
      "step": 478660
    },
    {
      "epoch": 195.94,
      "grad_norm": 1.6775208711624146,
      "learning_rate": 5.2912942858145295e-09,
      "loss": 0.4486,
      "step": 478670
    },
    {
      "epoch": 195.94,
      "grad_norm": 2.418158769607544,
      "learning_rate": 5.280646464777671e-09,
      "loss": 0.4323,
      "step": 478680
    },
    {
      "epoch": 195.94,
      "grad_norm": 1.8462252616882324,
      "learning_rate": 5.270009355906862e-09,
      "loss": 0.4281,
      "step": 478690
    },
    {
      "epoch": 195.95,
      "grad_norm": 2.5263962745666504,
      "learning_rate": 5.259382959250501e-09,
      "loss": 0.4234,
      "step": 478700
    },
    {
      "epoch": 195.95,
      "grad_norm": 1.9248043298721313,
      "learning_rate": 5.248767274857509e-09,
      "loss": 0.441,
      "step": 478710
    },
    {
      "epoch": 195.96,
      "grad_norm": 1.9672400951385498,
      "learning_rate": 5.238162302776544e-09,
      "loss": 0.4533,
      "step": 478720
    },
    {
      "epoch": 195.96,
      "grad_norm": 2.208754777908325,
      "learning_rate": 5.227568043056265e-09,
      "loss": 0.4582,
      "step": 478730
    },
    {
      "epoch": 195.96,
      "grad_norm": 1.9836407899856567,
      "learning_rate": 5.21698449574507e-09,
      "loss": 0.4637,
      "step": 478740
    },
    {
      "epoch": 195.97,
      "grad_norm": 2.147036075592041,
      "learning_rate": 5.20641166089136e-09,
      "loss": 0.4586,
      "step": 478750
    },
    {
      "epoch": 195.97,
      "grad_norm": 1.7765817642211914,
      "learning_rate": 5.195849538543531e-09,
      "loss": 0.4468,
      "step": 478760
    },
    {
      "epoch": 195.98,
      "grad_norm": 2.0105960369110107,
      "learning_rate": 5.185298128750244e-09,
      "loss": 0.4558,
      "step": 478770
    },
    {
      "epoch": 195.98,
      "grad_norm": 2.049842357635498,
      "learning_rate": 5.174757431559897e-09,
      "loss": 0.4301,
      "step": 478780
    },
    {
      "epoch": 195.98,
      "grad_norm": 2.2300925254821777,
      "learning_rate": 5.1642274470203685e-09,
      "loss": 0.4557,
      "step": 478790
    },
    {
      "epoch": 195.99,
      "grad_norm": 2.084766387939453,
      "learning_rate": 5.153708175180058e-09,
      "loss": 0.4547,
      "step": 478800
    },
    {
      "epoch": 195.99,
      "grad_norm": 2.1370866298675537,
      "learning_rate": 5.143199616087102e-09,
      "loss": 0.4593,
      "step": 478810
    },
    {
      "epoch": 196.0,
      "grad_norm": 2.000521421432495,
      "learning_rate": 5.132701769789901e-09,
      "loss": 0.4423,
      "step": 478820
    },
    {
      "epoch": 196.0,
      "eval_loss": 0.44342944025993347,
      "eval_runtime": 43.2861,
      "eval_samples_per_second": 79.679,
      "eval_steps_per_second": 9.98,
      "step": 478828
    },
    {
      "epoch": 196.0,
      "grad_norm": 2.194579839706421,
      "learning_rate": 5.122214636336334e-09,
      "loss": 0.4496,
      "step": 478830
    },
    {
      "epoch": 196.0,
      "grad_norm": 2.008692741394043,
      "learning_rate": 5.111738215774278e-09,
      "loss": 0.4463,
      "step": 478840
    },
    {
      "epoch": 196.01,
      "grad_norm": 1.832016944885254,
      "learning_rate": 5.101272508152132e-09,
      "loss": 0.4419,
      "step": 478850
    },
    {
      "epoch": 196.01,
      "grad_norm": 2.2772741317749023,
      "learning_rate": 5.090817513517514e-09,
      "loss": 0.4417,
      "step": 478860
    },
    {
      "epoch": 196.02,
      "grad_norm": 1.84866201877594,
      "learning_rate": 5.080373231918303e-09,
      "loss": 0.4519,
      "step": 478870
    },
    {
      "epoch": 196.02,
      "grad_norm": 2.4122040271759033,
      "learning_rate": 5.069939663402637e-09,
      "loss": 0.46,
      "step": 478880
    },
    {
      "epoch": 196.03,
      "grad_norm": 2.109985828399658,
      "learning_rate": 5.059516808018135e-09,
      "loss": 0.4759,
      "step": 478890
    },
    {
      "epoch": 196.03,
      "grad_norm": 2.482599973678589,
      "learning_rate": 5.049104665812414e-09,
      "loss": 0.4672,
      "step": 478900
    },
    {
      "epoch": 196.03,
      "grad_norm": 1.78165864944458,
      "learning_rate": 5.038703236833613e-09,
      "loss": 0.451,
      "step": 478910
    },
    {
      "epoch": 196.04,
      "grad_norm": 1.944416880607605,
      "learning_rate": 5.028312521128831e-09,
      "loss": 0.4476,
      "step": 478920
    },
    {
      "epoch": 196.04,
      "grad_norm": 1.8550245761871338,
      "learning_rate": 5.017932518745944e-09,
      "loss": 0.4239,
      "step": 478930
    },
    {
      "epoch": 196.05,
      "grad_norm": 1.9835740327835083,
      "learning_rate": 5.007563229732312e-09,
      "loss": 0.4291,
      "step": 478940
    },
    {
      "epoch": 196.05,
      "grad_norm": 1.8865379095077515,
      "learning_rate": 4.997204654135813e-09,
      "loss": 0.4368,
      "step": 478950
    },
    {
      "epoch": 196.05,
      "grad_norm": 1.9959005117416382,
      "learning_rate": 4.986856792003543e-09,
      "loss": 0.4468,
      "step": 478960
    },
    {
      "epoch": 196.06,
      "grad_norm": 4.689440727233887,
      "learning_rate": 4.976519643383123e-09,
      "loss": 0.4312,
      "step": 478970
    },
    {
      "epoch": 196.06,
      "grad_norm": 2.2296221256256104,
      "learning_rate": 4.966193208321649e-09,
      "loss": 0.445,
      "step": 478980
    },
    {
      "epoch": 196.07,
      "grad_norm": 1.8731224536895752,
      "learning_rate": 4.955877486866738e-09,
      "loss": 0.4463,
      "step": 478990
    },
    {
      "epoch": 196.07,
      "grad_norm": 1.9646973609924316,
      "learning_rate": 4.945572479065491e-09,
      "loss": 0.4366,
      "step": 479000
    },
    {
      "epoch": 196.07,
      "grad_norm": 2.0557634830474854,
      "learning_rate": 4.9352781849650024e-09,
      "loss": 0.4384,
      "step": 479010
    },
    {
      "epoch": 196.08,
      "grad_norm": 1.7643675804138184,
      "learning_rate": 4.924994604612632e-09,
      "loss": 0.4513,
      "step": 479020
    },
    {
      "epoch": 196.08,
      "grad_norm": 2.123037576675415,
      "learning_rate": 4.9147217380554775e-09,
      "loss": 0.4443,
      "step": 479030
    },
    {
      "epoch": 196.09,
      "grad_norm": 1.7584211826324463,
      "learning_rate": 4.9044595853403765e-09,
      "loss": 0.4461,
      "step": 479040
    },
    {
      "epoch": 196.09,
      "grad_norm": 1.7526073455810547,
      "learning_rate": 4.8942081465144255e-09,
      "loss": 0.4338,
      "step": 479050
    },
    {
      "epoch": 196.09,
      "grad_norm": 1.7135937213897705,
      "learning_rate": 4.8839674216247235e-09,
      "loss": 0.4529,
      "step": 479060
    },
    {
      "epoch": 196.1,
      "grad_norm": 2.2700631618499756,
      "learning_rate": 4.873737410718108e-09,
      "loss": 0.4412,
      "step": 479070
    },
    {
      "epoch": 196.1,
      "grad_norm": 1.989006757736206,
      "learning_rate": 4.863518113841416e-09,
      "loss": 0.4157,
      "step": 479080
    },
    {
      "epoch": 196.11,
      "grad_norm": 2.0426712036132812,
      "learning_rate": 4.853309531041485e-09,
      "loss": 0.4358,
      "step": 479090
    },
    {
      "epoch": 196.11,
      "grad_norm": 1.8667442798614502,
      "learning_rate": 4.843111662365152e-09,
      "loss": 0.4434,
      "step": 479100
    },
    {
      "epoch": 196.12,
      "grad_norm": 1.8177943229675293,
      "learning_rate": 4.832924507858996e-09,
      "loss": 0.4213,
      "step": 479110
    },
    {
      "epoch": 196.12,
      "grad_norm": 2.0431265830993652,
      "learning_rate": 4.822748067569594e-09,
      "loss": 0.447,
      "step": 479120
    },
    {
      "epoch": 196.12,
      "grad_norm": 1.999587059020996,
      "learning_rate": 4.812582341544042e-09,
      "loss": 0.4366,
      "step": 479130
    },
    {
      "epoch": 196.13,
      "grad_norm": 2.5107309818267822,
      "learning_rate": 4.802427329828138e-09,
      "loss": 0.4529,
      "step": 479140
    },
    {
      "epoch": 196.13,
      "grad_norm": 2.429072380065918,
      "learning_rate": 4.79228303246924e-09,
      "loss": 0.4299,
      "step": 479150
    },
    {
      "epoch": 196.14,
      "grad_norm": 1.7693097591400146,
      "learning_rate": 4.7821494495131446e-09,
      "loss": 0.4412,
      "step": 479160
    },
    {
      "epoch": 196.14,
      "grad_norm": 2.065180540084839,
      "learning_rate": 4.772026581006689e-09,
      "loss": 0.4465,
      "step": 479170
    },
    {
      "epoch": 196.14,
      "grad_norm": 2.653383731842041,
      "learning_rate": 4.76191442699593e-09,
      "loss": 0.4614,
      "step": 479180
    },
    {
      "epoch": 196.15,
      "grad_norm": 2.064167022705078,
      "learning_rate": 4.751812987527446e-09,
      "loss": 0.4396,
      "step": 479190
    },
    {
      "epoch": 196.15,
      "grad_norm": 2.1065595149993896,
      "learning_rate": 4.7417222626472925e-09,
      "loss": 0.449,
      "step": 479200
    },
    {
      "epoch": 196.16,
      "grad_norm": 2.2057626247406006,
      "learning_rate": 4.731642252402048e-09,
      "loss": 0.4334,
      "step": 479210
    },
    {
      "epoch": 196.16,
      "grad_norm": 2.1095597743988037,
      "learning_rate": 4.7215729568375084e-09,
      "loss": 0.4443,
      "step": 479220
    },
    {
      "epoch": 196.16,
      "grad_norm": 2.331007719039917,
      "learning_rate": 4.711514375999991e-09,
      "loss": 0.4387,
      "step": 479230
    },
    {
      "epoch": 196.17,
      "grad_norm": 2.5841877460479736,
      "learning_rate": 4.701466509935553e-09,
      "loss": 0.446,
      "step": 479240
    },
    {
      "epoch": 196.17,
      "grad_norm": 1.9338370561599731,
      "learning_rate": 4.691429358690251e-09,
      "loss": 0.4245,
      "step": 479250
    },
    {
      "epoch": 196.18,
      "grad_norm": 1.8586782217025757,
      "learning_rate": 4.681402922309882e-09,
      "loss": 0.4486,
      "step": 479260
    },
    {
      "epoch": 196.18,
      "grad_norm": 1.7231498956680298,
      "learning_rate": 4.671387200840503e-09,
      "loss": 0.4448,
      "step": 479270
    },
    {
      "epoch": 196.19,
      "grad_norm": 2.2844643592834473,
      "learning_rate": 4.66138219432791e-09,
      "loss": 0.4459,
      "step": 479280
    },
    {
      "epoch": 196.19,
      "grad_norm": 1.9108383655548096,
      "learning_rate": 4.65138790281816e-09,
      "loss": 0.4397,
      "step": 479290
    },
    {
      "epoch": 196.19,
      "grad_norm": 2.1443569660186768,
      "learning_rate": 4.6414043263570504e-09,
      "loss": 0.4129,
      "step": 479300
    },
    {
      "epoch": 196.2,
      "grad_norm": 2.2551660537719727,
      "learning_rate": 4.631431464989857e-09,
      "loss": 0.4403,
      "step": 479310
    },
    {
      "epoch": 196.2,
      "grad_norm": 2.241032838821411,
      "learning_rate": 4.621469318762897e-09,
      "loss": 0.4535,
      "step": 479320
    },
    {
      "epoch": 196.21,
      "grad_norm": 2.074453353881836,
      "learning_rate": 4.611517887721187e-09,
      "loss": 0.4367,
      "step": 479330
    },
    {
      "epoch": 196.21,
      "grad_norm": 2.023238182067871,
      "learning_rate": 4.601577171910783e-09,
      "loss": 0.4203,
      "step": 479340
    },
    {
      "epoch": 196.21,
      "grad_norm": 1.8788725137710571,
      "learning_rate": 4.591647171376962e-09,
      "loss": 0.4612,
      "step": 479350
    },
    {
      "epoch": 196.22,
      "grad_norm": 2.2303547859191895,
      "learning_rate": 4.58172788616526e-09,
      "loss": 0.4435,
      "step": 479360
    },
    {
      "epoch": 196.22,
      "grad_norm": 2.4256222248077393,
      "learning_rate": 4.571819316321213e-09,
      "loss": 0.4533,
      "step": 479370
    },
    {
      "epoch": 196.23,
      "grad_norm": 1.858974575996399,
      "learning_rate": 4.561921461890099e-09,
      "loss": 0.432,
      "step": 479380
    },
    {
      "epoch": 196.23,
      "grad_norm": 1.987871527671814,
      "learning_rate": 4.552034322917453e-09,
      "loss": 0.4613,
      "step": 479390
    },
    {
      "epoch": 196.23,
      "grad_norm": 2.1171367168426514,
      "learning_rate": 4.542157899448291e-09,
      "loss": 0.4561,
      "step": 479400
    },
    {
      "epoch": 196.24,
      "grad_norm": 2.0661354064941406,
      "learning_rate": 4.532292191527891e-09,
      "loss": 0.4482,
      "step": 479410
    },
    {
      "epoch": 196.24,
      "grad_norm": 2.0750701427459717,
      "learning_rate": 4.5224371992017875e-09,
      "loss": 0.4342,
      "step": 479420
    },
    {
      "epoch": 196.25,
      "grad_norm": 2.176860809326172,
      "learning_rate": 4.512592922514737e-09,
      "loss": 0.4365,
      "step": 479430
    },
    {
      "epoch": 196.25,
      "grad_norm": 2.0336029529571533,
      "learning_rate": 4.502759361511756e-09,
      "loss": 0.4216,
      "step": 479440
    },
    {
      "epoch": 196.25,
      "grad_norm": 1.8606663942337036,
      "learning_rate": 4.492936516238381e-09,
      "loss": 0.4395,
      "step": 479450
    },
    {
      "epoch": 196.26,
      "grad_norm": 2.1128673553466797,
      "learning_rate": 4.483124386739107e-09,
      "loss": 0.4298,
      "step": 479460
    },
    {
      "epoch": 196.26,
      "grad_norm": 2.082125186920166,
      "learning_rate": 4.473322973059211e-09,
      "loss": 0.447,
      "step": 479470
    },
    {
      "epoch": 196.27,
      "grad_norm": 2.110673189163208,
      "learning_rate": 4.463532275243709e-09,
      "loss": 0.4493,
      "step": 479480
    },
    {
      "epoch": 196.27,
      "grad_norm": 2.0179526805877686,
      "learning_rate": 4.453752293337096e-09,
      "loss": 0.4303,
      "step": 479490
    },
    {
      "epoch": 196.28,
      "grad_norm": 2.1287853717803955,
      "learning_rate": 4.443983027384128e-09,
      "loss": 0.4501,
      "step": 479500
    },
    {
      "epoch": 196.28,
      "grad_norm": 2.066659450531006,
      "learning_rate": 4.434224477429822e-09,
      "loss": 0.4371,
      "step": 479510
    },
    {
      "epoch": 196.28,
      "grad_norm": 2.0540826320648193,
      "learning_rate": 4.4244766435189325e-09,
      "loss": 0.4651,
      "step": 479520
    },
    {
      "epoch": 196.29,
      "grad_norm": 2.1623005867004395,
      "learning_rate": 4.414739525695956e-09,
      "loss": 0.4382,
      "step": 479530
    },
    {
      "epoch": 196.29,
      "grad_norm": 2.3532350063323975,
      "learning_rate": 4.405013124005388e-09,
      "loss": 0.4256,
      "step": 479540
    },
    {
      "epoch": 196.3,
      "grad_norm": 2.268099308013916,
      "learning_rate": 4.395297438491985e-09,
      "loss": 0.431,
      "step": 479550
    },
    {
      "epoch": 196.3,
      "grad_norm": 1.9754587411880493,
      "learning_rate": 4.385592469200241e-09,
      "loss": 0.45,
      "step": 479560
    },
    {
      "epoch": 196.3,
      "grad_norm": 2.2099461555480957,
      "learning_rate": 4.375898216174393e-09,
      "loss": 0.4516,
      "step": 479570
    },
    {
      "epoch": 196.31,
      "grad_norm": 1.878753662109375,
      "learning_rate": 4.366214679459197e-09,
      "loss": 0.4516,
      "step": 479580
    },
    {
      "epoch": 196.31,
      "grad_norm": 1.7642154693603516,
      "learning_rate": 4.356541859098627e-09,
      "loss": 0.4337,
      "step": 479590
    },
    {
      "epoch": 196.32,
      "grad_norm": 1.898701548576355,
      "learning_rate": 4.34687975513744e-09,
      "loss": 0.46,
      "step": 479600
    },
    {
      "epoch": 196.32,
      "grad_norm": 1.9395726919174194,
      "learning_rate": 4.33722836761935e-09,
      "loss": 0.4202,
      "step": 479610
    },
    {
      "epoch": 196.32,
      "grad_norm": 1.902884840965271,
      "learning_rate": 4.327587696589113e-09,
      "loss": 0.462,
      "step": 479620
    },
    {
      "epoch": 196.33,
      "grad_norm": 2.140011787414551,
      "learning_rate": 4.317957742090445e-09,
      "loss": 0.4445,
      "step": 479630
    },
    {
      "epoch": 196.33,
      "grad_norm": 1.622885823249817,
      "learning_rate": 4.3083385041678405e-09,
      "loss": 0.4406,
      "step": 479640
    },
    {
      "epoch": 196.34,
      "grad_norm": 2.1977672576904297,
      "learning_rate": 4.298729982865016e-09,
      "loss": 0.4322,
      "step": 479650
    },
    {
      "epoch": 196.34,
      "grad_norm": 2.110962152481079,
      "learning_rate": 4.289132178226205e-09,
      "loss": 0.4394,
      "step": 479660
    },
    {
      "epoch": 196.34,
      "grad_norm": 1.9365224838256836,
      "learning_rate": 4.279545090295124e-09,
      "loss": 0.4266,
      "step": 479670
    },
    {
      "epoch": 196.35,
      "grad_norm": 2.000408172607422,
      "learning_rate": 4.269968719116268e-09,
      "loss": 0.4467,
      "step": 479680
    },
    {
      "epoch": 196.35,
      "grad_norm": 2.1980576515197754,
      "learning_rate": 4.260403064732833e-09,
      "loss": 0.4502,
      "step": 479690
    },
    {
      "epoch": 196.36,
      "grad_norm": 1.8496592044830322,
      "learning_rate": 4.250848127189052e-09,
      "loss": 0.4511,
      "step": 479700
    },
    {
      "epoch": 196.36,
      "grad_norm": 1.7481768131256104,
      "learning_rate": 4.241303906528642e-09,
      "loss": 0.4194,
      "step": 479710
    },
    {
      "epoch": 196.37,
      "grad_norm": 1.8483341932296753,
      "learning_rate": 4.231770402795056e-09,
      "loss": 0.4522,
      "step": 479720
    },
    {
      "epoch": 196.37,
      "grad_norm": 2.119077682495117,
      "learning_rate": 4.2222476160322716e-09,
      "loss": 0.4563,
      "step": 479730
    },
    {
      "epoch": 196.37,
      "grad_norm": 1.9867663383483887,
      "learning_rate": 4.2127355462837425e-09,
      "loss": 0.4476,
      "step": 479740
    },
    {
      "epoch": 196.38,
      "grad_norm": 1.7961007356643677,
      "learning_rate": 4.203234193593183e-09,
      "loss": 0.4269,
      "step": 479750
    },
    {
      "epoch": 196.38,
      "grad_norm": 1.8592932224273682,
      "learning_rate": 4.193743558004049e-09,
      "loss": 0.4659,
      "step": 479760
    },
    {
      "epoch": 196.39,
      "grad_norm": 2.2746999263763428,
      "learning_rate": 4.184263639559795e-09,
      "loss": 0.4294,
      "step": 479770
    },
    {
      "epoch": 196.39,
      "grad_norm": 1.7507010698318481,
      "learning_rate": 4.1747944383038746e-09,
      "loss": 0.4355,
      "step": 479780
    },
    {
      "epoch": 196.39,
      "grad_norm": 2.1961662769317627,
      "learning_rate": 4.165335954279484e-09,
      "loss": 0.4521,
      "step": 479790
    },
    {
      "epoch": 196.4,
      "grad_norm": 2.4590156078338623,
      "learning_rate": 4.155888187530078e-09,
      "loss": 0.4507,
      "step": 479800
    },
    {
      "epoch": 196.4,
      "grad_norm": 1.960934042930603,
      "learning_rate": 4.14645113809911e-09,
      "loss": 0.45,
      "step": 479810
    },
    {
      "epoch": 196.41,
      "grad_norm": 2.1390013694763184,
      "learning_rate": 4.137024806029777e-09,
      "loss": 0.438,
      "step": 479820
    },
    {
      "epoch": 196.41,
      "grad_norm": 2.386078119277954,
      "learning_rate": 4.127609191365011e-09,
      "loss": 0.4351,
      "step": 479830
    },
    {
      "epoch": 196.41,
      "grad_norm": 2.2168567180633545,
      "learning_rate": 4.118204294148008e-09,
      "loss": 0.4451,
      "step": 479840
    },
    {
      "epoch": 196.42,
      "grad_norm": 2.3040554523468018,
      "learning_rate": 4.108810114421962e-09,
      "loss": 0.4249,
      "step": 479850
    },
    {
      "epoch": 196.42,
      "grad_norm": 1.9790689945220947,
      "learning_rate": 4.099426652230068e-09,
      "loss": 0.4334,
      "step": 479860
    },
    {
      "epoch": 196.43,
      "grad_norm": 1.8798458576202393,
      "learning_rate": 4.09005390761474e-09,
      "loss": 0.4395,
      "step": 479870
    },
    {
      "epoch": 196.43,
      "grad_norm": 1.9806751012802124,
      "learning_rate": 4.080691880619692e-09,
      "loss": 0.4387,
      "step": 479880
    },
    {
      "epoch": 196.43,
      "grad_norm": 2.149467706680298,
      "learning_rate": 4.071340571287079e-09,
      "loss": 0.4504,
      "step": 479890
    },
    {
      "epoch": 196.44,
      "grad_norm": 2.231388807296753,
      "learning_rate": 4.0619999796600955e-09,
      "loss": 0.4508,
      "step": 479900
    },
    {
      "epoch": 196.44,
      "grad_norm": 2.1324410438537598,
      "learning_rate": 4.052670105781676e-09,
      "loss": 0.4457,
      "step": 479910
    },
    {
      "epoch": 196.45,
      "grad_norm": 2.0915136337280273,
      "learning_rate": 4.043350949694234e-09,
      "loss": 0.4342,
      "step": 479920
    },
    {
      "epoch": 196.45,
      "grad_norm": 2.10290789604187,
      "learning_rate": 4.034042511440704e-09,
      "loss": 0.4458,
      "step": 479930
    },
    {
      "epoch": 196.46,
      "grad_norm": 2.3121819496154785,
      "learning_rate": 4.024744791063499e-09,
      "loss": 0.4403,
      "step": 479940
    },
    {
      "epoch": 196.46,
      "grad_norm": 2.2825021743774414,
      "learning_rate": 4.015457788605556e-09,
      "loss": 0.4666,
      "step": 479950
    },
    {
      "epoch": 196.46,
      "grad_norm": 2.2016446590423584,
      "learning_rate": 4.006181504109026e-09,
      "loss": 0.4536,
      "step": 479960
    },
    {
      "epoch": 196.47,
      "grad_norm": 1.9440885782241821,
      "learning_rate": 3.996915937616585e-09,
      "loss": 0.4541,
      "step": 479970
    },
    {
      "epoch": 196.47,
      "grad_norm": 2.056734561920166,
      "learning_rate": 3.987661089170646e-09,
      "loss": 0.4514,
      "step": 479980
    },
    {
      "epoch": 196.48,
      "grad_norm": 1.694320797920227,
      "learning_rate": 3.978416958813624e-09,
      "loss": 0.4465,
      "step": 479990
    },
    {
      "epoch": 196.48,
      "grad_norm": 1.6511088609695435,
      "learning_rate": 3.969183546587932e-09,
      "loss": 0.4423,
      "step": 480000
    },
    {
      "epoch": 196.48,
      "grad_norm": 2.314872980117798,
      "learning_rate": 3.959960852535724e-09,
      "loss": 0.4497,
      "step": 480010
    },
    {
      "epoch": 196.49,
      "grad_norm": 1.7017699480056763,
      "learning_rate": 3.950748876699414e-09,
      "loss": 0.4372,
      "step": 480020
    },
    {
      "epoch": 196.49,
      "grad_norm": 1.9063678979873657,
      "learning_rate": 3.9415476191208965e-09,
      "loss": 0.452,
      "step": 480030
    },
    {
      "epoch": 196.5,
      "grad_norm": 2.0378401279449463,
      "learning_rate": 3.932357079842845e-09,
      "loss": 0.4557,
      "step": 480040
    },
    {
      "epoch": 196.5,
      "grad_norm": 1.9681304693222046,
      "learning_rate": 3.923177258906892e-09,
      "loss": 0.4322,
      "step": 480050
    },
    {
      "epoch": 196.5,
      "grad_norm": 2.0124871730804443,
      "learning_rate": 3.914008156355193e-09,
      "loss": 0.4425,
      "step": 480060
    },
    {
      "epoch": 196.51,
      "grad_norm": 2.2356278896331787,
      "learning_rate": 3.904849772229899e-09,
      "loss": 0.4302,
      "step": 480070
    },
    {
      "epoch": 196.51,
      "grad_norm": 2.092442512512207,
      "learning_rate": 3.895702106572647e-09,
      "loss": 0.4578,
      "step": 480080
    },
    {
      "epoch": 196.52,
      "grad_norm": 1.86335027217865,
      "learning_rate": 3.886565159425849e-09,
      "loss": 0.4381,
      "step": 480090
    },
    {
      "epoch": 196.52,
      "grad_norm": 2.1683902740478516,
      "learning_rate": 3.8774389308308775e-09,
      "loss": 0.4492,
      "step": 480100
    },
    {
      "epoch": 196.52,
      "grad_norm": 2.1091463565826416,
      "learning_rate": 3.868323420829888e-09,
      "loss": 0.4329,
      "step": 480110
    },
    {
      "epoch": 196.53,
      "grad_norm": 2.193084955215454,
      "learning_rate": 3.859218629464513e-09,
      "loss": 0.4393,
      "step": 480120
    },
    {
      "epoch": 196.53,
      "grad_norm": 1.9500733613967896,
      "learning_rate": 3.850124556776385e-09,
      "loss": 0.4428,
      "step": 480130
    },
    {
      "epoch": 196.54,
      "grad_norm": 1.7780464887619019,
      "learning_rate": 3.8410412028071385e-09,
      "loss": 0.464,
      "step": 480140
    },
    {
      "epoch": 196.54,
      "grad_norm": 1.9468209743499756,
      "learning_rate": 3.831968567598407e-09,
      "loss": 0.4386,
      "step": 480150
    },
    {
      "epoch": 196.55,
      "grad_norm": 1.8179407119750977,
      "learning_rate": 3.822906651191823e-09,
      "loss": 0.4621,
      "step": 480160
    },
    {
      "epoch": 196.55,
      "grad_norm": 1.8855394124984741,
      "learning_rate": 3.8138554536290205e-09,
      "loss": 0.4536,
      "step": 480170
    },
    {
      "epoch": 196.55,
      "grad_norm": 2.1282901763916016,
      "learning_rate": 3.804814974951113e-09,
      "loss": 0.4289,
      "step": 480180
    },
    {
      "epoch": 196.56,
      "grad_norm": 2.3423409461975098,
      "learning_rate": 3.795785215199732e-09,
      "loss": 0.4557,
      "step": 480190
    },
    {
      "epoch": 196.56,
      "grad_norm": 1.8927048444747925,
      "learning_rate": 3.786766174416253e-09,
      "loss": 0.4547,
      "step": 480200
    },
    {
      "epoch": 196.57,
      "grad_norm": 1.9143898487091064,
      "learning_rate": 3.777757852642047e-09,
      "loss": 0.4394,
      "step": 480210
    },
    {
      "epoch": 196.57,
      "grad_norm": 2.259906768798828,
      "learning_rate": 3.768760249918228e-09,
      "loss": 0.4458,
      "step": 480220
    },
    {
      "epoch": 196.57,
      "grad_norm": 2.3486669063568115,
      "learning_rate": 3.759773366286169e-09,
      "loss": 0.4412,
      "step": 480230
    },
    {
      "epoch": 196.58,
      "grad_norm": 2.0443170070648193,
      "learning_rate": 3.750797201786722e-09,
      "loss": 0.4443,
      "step": 480240
    },
    {
      "epoch": 196.58,
      "grad_norm": 2.374342679977417,
      "learning_rate": 3.741831756461523e-09,
      "loss": 0.4446,
      "step": 480250
    },
    {
      "epoch": 196.59,
      "grad_norm": 1.834717035293579,
      "learning_rate": 3.7328770303511616e-09,
      "loss": 0.4611,
      "step": 480260
    },
    {
      "epoch": 196.59,
      "grad_norm": 1.9320029020309448,
      "learning_rate": 3.7239330234970117e-09,
      "loss": 0.4629,
      "step": 480270
    },
    {
      "epoch": 196.59,
      "grad_norm": 2.2610971927642822,
      "learning_rate": 3.7149997359396667e-09,
      "loss": 0.4465,
      "step": 480280
    },
    {
      "epoch": 196.6,
      "grad_norm": 2.6943066120147705,
      "learning_rate": 3.7060771677202386e-09,
      "loss": 0.4445,
      "step": 480290
    },
    {
      "epoch": 196.6,
      "grad_norm": 1.9293816089630127,
      "learning_rate": 3.6971653188798413e-09,
      "loss": 0.4629,
      "step": 480300
    },
    {
      "epoch": 196.61,
      "grad_norm": 2.017634868621826,
      "learning_rate": 3.688264189458806e-09,
      "loss": 0.4359,
      "step": 480310
    },
    {
      "epoch": 196.61,
      "grad_norm": 2.660207509994507,
      "learning_rate": 3.6793737794982465e-09,
      "loss": 0.4545,
      "step": 480320
    },
    {
      "epoch": 196.61,
      "grad_norm": 2.129223585128784,
      "learning_rate": 3.670494089039015e-09,
      "loss": 0.44,
      "step": 480330
    },
    {
      "epoch": 196.62,
      "grad_norm": 1.7242140769958496,
      "learning_rate": 3.6616251181214444e-09,
      "loss": 0.4244,
      "step": 480340
    },
    {
      "epoch": 196.62,
      "grad_norm": 2.1151866912841797,
      "learning_rate": 3.652766866786387e-09,
      "loss": 0.4528,
      "step": 480350
    },
    {
      "epoch": 196.63,
      "grad_norm": 2.063014507293701,
      "learning_rate": 3.6439193350741754e-09,
      "loss": 0.4248,
      "step": 480360
    },
    {
      "epoch": 196.63,
      "grad_norm": 1.910444736480713,
      "learning_rate": 3.635082523025662e-09,
      "loss": 0.4378,
      "step": 480370
    },
    {
      "epoch": 196.64,
      "grad_norm": 1.8774415254592896,
      "learning_rate": 3.6262564306811794e-09,
      "loss": 0.4366,
      "step": 480380
    },
    {
      "epoch": 196.64,
      "grad_norm": 1.8935697078704834,
      "learning_rate": 3.6174410580810593e-09,
      "loss": 0.4598,
      "step": 480390
    },
    {
      "epoch": 196.64,
      "grad_norm": 2.0562076568603516,
      "learning_rate": 3.6086364052658952e-09,
      "loss": 0.4484,
      "step": 480400
    },
    {
      "epoch": 196.65,
      "grad_norm": 1.9631340503692627,
      "learning_rate": 3.5998424722757586e-09,
      "loss": 0.4458,
      "step": 480410
    },
    {
      "epoch": 196.65,
      "grad_norm": 2.5464963912963867,
      "learning_rate": 3.5910592591512426e-09,
      "loss": 0.4469,
      "step": 480420
    },
    {
      "epoch": 196.66,
      "grad_norm": 2.097975492477417,
      "learning_rate": 3.5822867659324183e-09,
      "loss": 0.4441,
      "step": 480430
    },
    {
      "epoch": 196.66,
      "grad_norm": 2.022052526473999,
      "learning_rate": 3.5735249926593588e-09,
      "loss": 0.4413,
      "step": 480440
    },
    {
      "epoch": 196.66,
      "grad_norm": 2.5386788845062256,
      "learning_rate": 3.5647739393721356e-09,
      "loss": 0.4416,
      "step": 480450
    },
    {
      "epoch": 196.67,
      "grad_norm": 1.9897570610046387,
      "learning_rate": 3.5560336061113416e-09,
      "loss": 0.4359,
      "step": 480460
    },
    {
      "epoch": 196.67,
      "grad_norm": 2.1519837379455566,
      "learning_rate": 3.547303992916528e-09,
      "loss": 0.4539,
      "step": 480470
    },
    {
      "epoch": 196.68,
      "grad_norm": 2.264025926589966,
      "learning_rate": 3.5385850998277677e-09,
      "loss": 0.4403,
      "step": 480480
    },
    {
      "epoch": 196.68,
      "grad_norm": 2.1823623180389404,
      "learning_rate": 3.5298769268851323e-09,
      "loss": 0.437,
      "step": 480490
    },
    {
      "epoch": 196.68,
      "grad_norm": 2.1593074798583984,
      "learning_rate": 3.5211794741284336e-09,
      "loss": 0.4363,
      "step": 480500
    },
    {
      "epoch": 196.69,
      "grad_norm": 2.1446263790130615,
      "learning_rate": 3.5124927415977435e-09,
      "loss": 0.4365,
      "step": 480510
    },
    {
      "epoch": 196.69,
      "grad_norm": 2.0813398361206055,
      "learning_rate": 3.5038167293323545e-09,
      "loss": 0.4669,
      "step": 480520
    },
    {
      "epoch": 196.7,
      "grad_norm": 2.1460320949554443,
      "learning_rate": 3.495151437372338e-09,
      "loss": 0.4464,
      "step": 480530
    },
    {
      "epoch": 196.7,
      "grad_norm": 1.846854329109192,
      "learning_rate": 3.486496865757506e-09,
      "loss": 0.4265,
      "step": 480540
    },
    {
      "epoch": 196.7,
      "grad_norm": 2.3175413608551025,
      "learning_rate": 3.4778530145274096e-09,
      "loss": 0.4601,
      "step": 480550
    },
    {
      "epoch": 196.71,
      "grad_norm": 2.0866284370422363,
      "learning_rate": 3.4692198837213413e-09,
      "loss": 0.4568,
      "step": 480560
    },
    {
      "epoch": 196.71,
      "grad_norm": 1.7445305585861206,
      "learning_rate": 3.4605974733791125e-09,
      "loss": 0.4409,
      "step": 480570
    },
    {
      "epoch": 196.72,
      "grad_norm": 2.019619941711426,
      "learning_rate": 3.451985783540275e-09,
      "loss": 0.4183,
      "step": 480580
    },
    {
      "epoch": 196.72,
      "grad_norm": 2.120579242706299,
      "learning_rate": 3.4433848142443806e-09,
      "loss": 0.4452,
      "step": 480590
    },
    {
      "epoch": 196.73,
      "grad_norm": 1.7586212158203125,
      "learning_rate": 3.4347945655304604e-09,
      "loss": 0.4214,
      "step": 480600
    },
    {
      "epoch": 196.73,
      "grad_norm": 3.1720030307769775,
      "learning_rate": 3.426215037437806e-09,
      "loss": 0.4564,
      "step": 480610
    },
    {
      "epoch": 196.73,
      "grad_norm": 1.72977614402771,
      "learning_rate": 3.417646230006229e-09,
      "loss": 0.4434,
      "step": 480620
    },
    {
      "epoch": 196.74,
      "grad_norm": 1.8274192810058594,
      "learning_rate": 3.4090881432747607e-09,
      "loss": 0.4514,
      "step": 480630
    },
    {
      "epoch": 196.74,
      "grad_norm": 1.8420101404190063,
      "learning_rate": 3.400540777282693e-09,
      "loss": 0.4458,
      "step": 480640
    },
    {
      "epoch": 196.75,
      "grad_norm": 1.84889554977417,
      "learning_rate": 3.392004132068797e-09,
      "loss": 0.4337,
      "step": 480650
    },
    {
      "epoch": 196.75,
      "grad_norm": 2.197718381881714,
      "learning_rate": 3.383478207672364e-09,
      "loss": 0.4724,
      "step": 480660
    },
    {
      "epoch": 196.75,
      "grad_norm": 1.9780741930007935,
      "learning_rate": 3.374963004132685e-09,
      "loss": 0.4327,
      "step": 480670
    },
    {
      "epoch": 196.76,
      "grad_norm": 2.126774549484253,
      "learning_rate": 3.3664585214885318e-09,
      "loss": 0.4352,
      "step": 480680
    },
    {
      "epoch": 196.76,
      "grad_norm": 2.261976957321167,
      "learning_rate": 3.3579647597789355e-09,
      "loss": 0.4486,
      "step": 480690
    },
    {
      "epoch": 196.77,
      "grad_norm": 1.7480195760726929,
      "learning_rate": 3.349481719042927e-09,
      "loss": 0.425,
      "step": 480700
    },
    {
      "epoch": 196.77,
      "grad_norm": 2.3016042709350586,
      "learning_rate": 3.3410093993192775e-09,
      "loss": 0.4519,
      "step": 480710
    },
    {
      "epoch": 196.77,
      "grad_norm": 2.363558053970337,
      "learning_rate": 3.3325478006464985e-09,
      "loss": 0.4532,
      "step": 480720
    },
    {
      "epoch": 196.78,
      "grad_norm": 1.8414860963821411,
      "learning_rate": 3.3240969230636202e-09,
      "loss": 0.4338,
      "step": 480730
    },
    {
      "epoch": 196.78,
      "grad_norm": 2.4069721698760986,
      "learning_rate": 3.3156567666094148e-09,
      "loss": 0.4498,
      "step": 480740
    },
    {
      "epoch": 196.79,
      "grad_norm": 2.39604115486145,
      "learning_rate": 3.3072273313226526e-09,
      "loss": 0.4404,
      "step": 480750
    },
    {
      "epoch": 196.79,
      "grad_norm": 2.0850000381469727,
      "learning_rate": 3.2988086172415844e-09,
      "loss": 0.4238,
      "step": 480760
    },
    {
      "epoch": 196.79,
      "grad_norm": 2.004157781600952,
      "learning_rate": 3.290400624404981e-09,
      "loss": 0.4299,
      "step": 480770
    },
    {
      "epoch": 196.8,
      "grad_norm": 2.2166523933410645,
      "learning_rate": 3.2820033528513543e-09,
      "loss": 0.4447,
      "step": 480780
    },
    {
      "epoch": 196.8,
      "grad_norm": 2.3159310817718506,
      "learning_rate": 3.273616802618954e-09,
      "loss": 0.4579,
      "step": 480790
    },
    {
      "epoch": 196.81,
      "grad_norm": 1.8414360284805298,
      "learning_rate": 3.265240973746811e-09,
      "loss": 0.4473,
      "step": 480800
    },
    {
      "epoch": 196.81,
      "grad_norm": 1.8349071741104126,
      "learning_rate": 3.2568758662726568e-09,
      "loss": 0.4215,
      "step": 480810
    },
    {
      "epoch": 196.82,
      "grad_norm": 1.8288272619247437,
      "learning_rate": 3.2485214802350014e-09,
      "loss": 0.4371,
      "step": 480820
    },
    {
      "epoch": 196.82,
      "grad_norm": 2.0399234294891357,
      "learning_rate": 3.240177815672096e-09,
      "loss": 0.443,
      "step": 480830
    },
    {
      "epoch": 196.82,
      "grad_norm": 2.1987311840057373,
      "learning_rate": 3.2318448726221905e-09,
      "loss": 0.4477,
      "step": 480840
    },
    {
      "epoch": 196.83,
      "grad_norm": 2.006671905517578,
      "learning_rate": 3.2235226511237965e-09,
      "loss": 0.4354,
      "step": 480850
    },
    {
      "epoch": 196.83,
      "grad_norm": 2.4741671085357666,
      "learning_rate": 3.2152111512143835e-09,
      "loss": 0.4239,
      "step": 480860
    },
    {
      "epoch": 196.84,
      "grad_norm": 2.0239861011505127,
      "learning_rate": 3.2069103729324626e-09,
      "loss": 0.4571,
      "step": 480870
    },
    {
      "epoch": 196.84,
      "grad_norm": 2.164066791534424,
      "learning_rate": 3.1986203163160242e-09,
      "loss": 0.4554,
      "step": 480880
    },
    {
      "epoch": 196.84,
      "grad_norm": 1.8147550821304321,
      "learning_rate": 3.190340981403059e-09,
      "loss": 0.4459,
      "step": 480890
    },
    {
      "epoch": 196.85,
      "grad_norm": 2.2343599796295166,
      "learning_rate": 3.1820723682312967e-09,
      "loss": 0.4338,
      "step": 480900
    },
    {
      "epoch": 196.85,
      "grad_norm": 2.152397632598877,
      "learning_rate": 3.173814476838468e-09,
      "loss": 0.443,
      "step": 480910
    },
    {
      "epoch": 196.86,
      "grad_norm": 2.4882993698120117,
      "learning_rate": 3.165567307263084e-09,
      "loss": 0.4565,
      "step": 480920
    },
    {
      "epoch": 196.86,
      "grad_norm": 2.1768667697906494,
      "learning_rate": 3.1573308595423536e-09,
      "loss": 0.4275,
      "step": 480930
    },
    {
      "epoch": 196.86,
      "grad_norm": 1.8677183389663696,
      "learning_rate": 3.1491051337140076e-09,
      "loss": 0.4448,
      "step": 480940
    },
    {
      "epoch": 196.87,
      "grad_norm": 2.250032424926758,
      "learning_rate": 3.1408901298160365e-09,
      "loss": 0.4506,
      "step": 480950
    },
    {
      "epoch": 196.87,
      "grad_norm": 2.254890203475952,
      "learning_rate": 3.13268584788591e-09,
      "loss": 0.4551,
      "step": 480960
    },
    {
      "epoch": 196.88,
      "grad_norm": 2.4891369342803955,
      "learning_rate": 3.1244922879610985e-09,
      "loss": 0.4228,
      "step": 480970
    },
    {
      "epoch": 196.88,
      "grad_norm": 2.4479963779449463,
      "learning_rate": 3.116309450079332e-09,
      "loss": 0.4473,
      "step": 480980
    },
    {
      "epoch": 196.88,
      "grad_norm": 2.5194051265716553,
      "learning_rate": 3.1081373342778207e-09,
      "loss": 0.459,
      "step": 480990
    },
    {
      "epoch": 196.89,
      "grad_norm": 2.0147976875305176,
      "learning_rate": 3.0999759405945547e-09,
      "loss": 0.4428,
      "step": 481000
    },
    {
      "epoch": 196.89,
      "grad_norm": 2.0577361583709717,
      "learning_rate": 3.0918252690662234e-09,
      "loss": 0.4572,
      "step": 481010
    },
    {
      "epoch": 196.9,
      "grad_norm": 2.4771852493286133,
      "learning_rate": 3.083685319730557e-09,
      "loss": 0.4422,
      "step": 481020
    },
    {
      "epoch": 196.9,
      "grad_norm": 2.144690752029419,
      "learning_rate": 3.0755560926250257e-09,
      "loss": 0.4356,
      "step": 481030
    },
    {
      "epoch": 196.91,
      "grad_norm": 2.318871259689331,
      "learning_rate": 3.0674375877863184e-09,
      "loss": 0.4451,
      "step": 481040
    },
    {
      "epoch": 196.91,
      "grad_norm": 1.7351200580596924,
      "learning_rate": 3.0593298052521663e-09,
      "loss": 0.4284,
      "step": 481050
    },
    {
      "epoch": 196.91,
      "grad_norm": 1.5738691091537476,
      "learning_rate": 3.051232745059258e-09,
      "loss": 0.4602,
      "step": 481060
    },
    {
      "epoch": 196.92,
      "grad_norm": 1.8816370964050293,
      "learning_rate": 3.043146407245064e-09,
      "loss": 0.4444,
      "step": 481070
    },
    {
      "epoch": 196.92,
      "grad_norm": 1.837425708770752,
      "learning_rate": 3.0350707918462733e-09,
      "loss": 0.467,
      "step": 481080
    },
    {
      "epoch": 196.93,
      "grad_norm": 2.2106070518493652,
      "learning_rate": 3.027005898900356e-09,
      "loss": 0.4324,
      "step": 481090
    },
    {
      "epoch": 196.93,
      "grad_norm": 1.853401780128479,
      "learning_rate": 3.0189517284437414e-09,
      "loss": 0.4579,
      "step": 481100
    },
    {
      "epoch": 196.93,
      "grad_norm": 2.760331153869629,
      "learning_rate": 3.0109082805136395e-09,
      "loss": 0.429,
      "step": 481110
    },
    {
      "epoch": 196.94,
      "grad_norm": 2.1393139362335205,
      "learning_rate": 3.0028755551467394e-09,
      "loss": 0.4555,
      "step": 481120
    },
    {
      "epoch": 196.94,
      "grad_norm": 2.174914836883545,
      "learning_rate": 2.994853552379991e-09,
      "loss": 0.4456,
      "step": 481130
    },
    {
      "epoch": 196.95,
      "grad_norm": 2.1662392616271973,
      "learning_rate": 2.9868422722500834e-09,
      "loss": 0.4384,
      "step": 481140
    },
    {
      "epoch": 196.95,
      "grad_norm": 1.7979000806808472,
      "learning_rate": 2.978841714793446e-09,
      "loss": 0.4353,
      "step": 481150
    },
    {
      "epoch": 196.95,
      "grad_norm": 3.1934614181518555,
      "learning_rate": 2.9708518800472883e-09,
      "loss": 0.4593,
      "step": 481160
    },
    {
      "epoch": 196.96,
      "grad_norm": 1.9096969366073608,
      "learning_rate": 2.9628727680477796e-09,
      "loss": 0.4274,
      "step": 481170
    },
    {
      "epoch": 196.96,
      "grad_norm": 2.2808241844177246,
      "learning_rate": 2.954904378831349e-09,
      "loss": 0.4524,
      "step": 481180
    },
    {
      "epoch": 196.97,
      "grad_norm": 2.19053316116333,
      "learning_rate": 2.9469467124349464e-09,
      "loss": 0.4536,
      "step": 481190
    },
    {
      "epoch": 196.97,
      "grad_norm": 1.941028118133545,
      "learning_rate": 2.9389997688950007e-09,
      "loss": 0.4326,
      "step": 481200
    },
    {
      "epoch": 196.98,
      "grad_norm": 1.8036190271377563,
      "learning_rate": 2.9310635482474205e-09,
      "loss": 0.4624,
      "step": 481210
    },
    {
      "epoch": 196.98,
      "grad_norm": 2.032902956008911,
      "learning_rate": 2.9231380505288957e-09,
      "loss": 0.4415,
      "step": 481220
    },
    {
      "epoch": 196.98,
      "grad_norm": 1.7815666198730469,
      "learning_rate": 2.915223275775855e-09,
      "loss": 0.4355,
      "step": 481230
    },
    {
      "epoch": 196.99,
      "grad_norm": 1.935282826423645,
      "learning_rate": 2.9073192240244674e-09,
      "loss": 0.4355,
      "step": 481240
    },
    {
      "epoch": 196.99,
      "grad_norm": 2.061774969100952,
      "learning_rate": 2.899425895310642e-09,
      "loss": 0.4595,
      "step": 481250
    },
    {
      "epoch": 197.0,
      "grad_norm": 1.984893560409546,
      "learning_rate": 2.891543289670808e-09,
      "loss": 0.4506,
      "step": 481260
    },
    {
      "epoch": 197.0,
      "grad_norm": 1.6130762100219727,
      "learning_rate": 2.883671407141134e-09,
      "loss": 0.4331,
      "step": 481270
    },
    {
      "epoch": 197.0,
      "eval_loss": 0.44527602195739746,
      "eval_runtime": 43.2315,
      "eval_samples_per_second": 79.78,
      "eval_steps_per_second": 9.993,
      "step": 481271
    },
    {
      "epoch": 197.0,
      "grad_norm": 2.1920008659362793,
      "learning_rate": 2.875810247757529e-09,
      "loss": 0.4527,
      "step": 481280
    },
    {
      "epoch": 197.01,
      "grad_norm": 1.8506228923797607,
      "learning_rate": 2.867959811556162e-09,
      "loss": 0.4264,
      "step": 481290
    },
    {
      "epoch": 197.01,
      "grad_norm": 2.6664927005767822,
      "learning_rate": 2.860120098572942e-09,
      "loss": 0.4579,
      "step": 481300
    },
    {
      "epoch": 197.02,
      "grad_norm": 2.1826119422912598,
      "learning_rate": 2.852291108843517e-09,
      "loss": 0.4513,
      "step": 481310
    },
    {
      "epoch": 197.02,
      "grad_norm": 2.2895116806030273,
      "learning_rate": 2.8444728424040566e-09,
      "loss": 0.4264,
      "step": 481320
    },
    {
      "epoch": 197.02,
      "grad_norm": 2.269070863723755,
      "learning_rate": 2.836665299290209e-09,
      "loss": 0.4455,
      "step": 481330
    },
    {
      "epoch": 197.03,
      "grad_norm": 2.0440123081207275,
      "learning_rate": 2.828868479537883e-09,
      "loss": 0.4625,
      "step": 481340
    },
    {
      "epoch": 197.03,
      "grad_norm": 2.1871190071105957,
      "learning_rate": 2.8210823831827273e-09,
      "loss": 0.4673,
      "step": 481350
    },
    {
      "epoch": 197.04,
      "grad_norm": 2.1483845710754395,
      "learning_rate": 2.8133070102603905e-09,
      "loss": 0.4347,
      "step": 481360
    },
    {
      "epoch": 197.04,
      "grad_norm": 2.274148941040039,
      "learning_rate": 2.8055423608062608e-09,
      "loss": 0.4241,
      "step": 481370
    },
    {
      "epoch": 197.04,
      "grad_norm": 1.857280969619751,
      "learning_rate": 2.797788434856507e-09,
      "loss": 0.4581,
      "step": 481380
    },
    {
      "epoch": 197.05,
      "grad_norm": 1.8964951038360596,
      "learning_rate": 2.790045232446258e-09,
      "loss": 0.4338,
      "step": 481390
    },
    {
      "epoch": 197.05,
      "grad_norm": 3.837796211242676,
      "learning_rate": 2.782312753610901e-09,
      "loss": 0.4383,
      "step": 481400
    },
    {
      "epoch": 197.06,
      "grad_norm": 2.110368013381958,
      "learning_rate": 2.7745909983858256e-09,
      "loss": 0.461,
      "step": 481410
    },
    {
      "epoch": 197.06,
      "grad_norm": 2.154695987701416,
      "learning_rate": 2.7668799668069397e-09,
      "loss": 0.4363,
      "step": 481420
    },
    {
      "epoch": 197.07,
      "grad_norm": 1.8745083808898926,
      "learning_rate": 2.7591796589088513e-09,
      "loss": 0.4455,
      "step": 481430
    },
    {
      "epoch": 197.07,
      "grad_norm": 2.227658987045288,
      "learning_rate": 2.7514900747274692e-09,
      "loss": 0.4398,
      "step": 481440
    },
    {
      "epoch": 197.07,
      "grad_norm": 2.216655969619751,
      "learning_rate": 2.7438112142974015e-09,
      "loss": 0.439,
      "step": 481450
    },
    {
      "epoch": 197.08,
      "grad_norm": 1.9335169792175293,
      "learning_rate": 2.7361430776542958e-09,
      "loss": 0.4256,
      "step": 481460
    },
    {
      "epoch": 197.08,
      "grad_norm": 1.8359589576721191,
      "learning_rate": 2.728485664833021e-09,
      "loss": 0.4328,
      "step": 481470
    },
    {
      "epoch": 197.09,
      "grad_norm": 2.2975637912750244,
      "learning_rate": 2.7208389758687046e-09,
      "loss": 0.4428,
      "step": 481480
    },
    {
      "epoch": 197.09,
      "grad_norm": 1.6277341842651367,
      "learning_rate": 2.7132030107967355e-09,
      "loss": 0.4601,
      "step": 481490
    },
    {
      "epoch": 197.09,
      "grad_norm": 2.1353728771209717,
      "learning_rate": 2.705577769651461e-09,
      "loss": 0.4508,
      "step": 481500
    },
    {
      "epoch": 197.1,
      "grad_norm": 2.2610299587249756,
      "learning_rate": 2.6979632524682696e-09,
      "loss": 0.4516,
      "step": 481510
    },
    {
      "epoch": 197.1,
      "grad_norm": 2.015902042388916,
      "learning_rate": 2.6903594592817685e-09,
      "loss": 0.4167,
      "step": 481520
    },
    {
      "epoch": 197.11,
      "grad_norm": 1.8733575344085693,
      "learning_rate": 2.6827663901270865e-09,
      "loss": 0.4561,
      "step": 481530
    },
    {
      "epoch": 197.11,
      "grad_norm": 2.1925103664398193,
      "learning_rate": 2.6751840450388307e-09,
      "loss": 0.4521,
      "step": 481540
    },
    {
      "epoch": 197.11,
      "grad_norm": 2.185858726501465,
      "learning_rate": 2.6676124240516093e-09,
      "loss": 0.4304,
      "step": 481550
    },
    {
      "epoch": 197.12,
      "grad_norm": 1.9399876594543457,
      "learning_rate": 2.6600515272002903e-09,
      "loss": 0.4216,
      "step": 481560
    },
    {
      "epoch": 197.12,
      "grad_norm": 1.8898952007293701,
      "learning_rate": 2.652501354519481e-09,
      "loss": 0.4428,
      "step": 481570
    },
    {
      "epoch": 197.13,
      "grad_norm": 2.4345743656158447,
      "learning_rate": 2.6449619060437895e-09,
      "loss": 0.4567,
      "step": 481580
    },
    {
      "epoch": 197.13,
      "grad_norm": 2.6260032653808594,
      "learning_rate": 2.6374331818078237e-09,
      "loss": 0.4412,
      "step": 481590
    },
    {
      "epoch": 197.13,
      "grad_norm": 2.0973806381225586,
      "learning_rate": 2.6299151818456703e-09,
      "loss": 0.4377,
      "step": 481600
    },
    {
      "epoch": 197.14,
      "grad_norm": 2.068561553955078,
      "learning_rate": 2.6224079061921975e-09,
      "loss": 0.4583,
      "step": 481610
    },
    {
      "epoch": 197.14,
      "grad_norm": 2.0983595848083496,
      "learning_rate": 2.6149113548817532e-09,
      "loss": 0.4425,
      "step": 481620
    },
    {
      "epoch": 197.15,
      "grad_norm": 2.2011542320251465,
      "learning_rate": 2.6074255279484243e-09,
      "loss": 0.4411,
      "step": 481630
    },
    {
      "epoch": 197.15,
      "grad_norm": 1.9900459051132202,
      "learning_rate": 2.5999504254268186e-09,
      "loss": 0.4616,
      "step": 481640
    },
    {
      "epoch": 197.16,
      "grad_norm": 1.8978759050369263,
      "learning_rate": 2.5924860473510236e-09,
      "loss": 0.4474,
      "step": 481650
    },
    {
      "epoch": 197.16,
      "grad_norm": 2.0511114597320557,
      "learning_rate": 2.5850323937551264e-09,
      "loss": 0.4448,
      "step": 481660
    },
    {
      "epoch": 197.16,
      "grad_norm": 3.4059882164001465,
      "learning_rate": 2.5775894646732144e-09,
      "loss": 0.4471,
      "step": 481670
    },
    {
      "epoch": 197.17,
      "grad_norm": 2.233938217163086,
      "learning_rate": 2.5701572601396355e-09,
      "loss": 0.4286,
      "step": 481680
    },
    {
      "epoch": 197.17,
      "grad_norm": 1.8926993608474731,
      "learning_rate": 2.562735780188476e-09,
      "loss": 0.4366,
      "step": 481690
    },
    {
      "epoch": 197.18,
      "grad_norm": 2.349437713623047,
      "learning_rate": 2.555325024853564e-09,
      "loss": 0.4362,
      "step": 481700
    },
    {
      "epoch": 197.18,
      "grad_norm": 2.4831409454345703,
      "learning_rate": 2.547924994168726e-09,
      "loss": 0.448,
      "step": 481710
    },
    {
      "epoch": 197.18,
      "grad_norm": 1.8031344413757324,
      "learning_rate": 2.5405356881683103e-09,
      "loss": 0.4463,
      "step": 481720
    },
    {
      "epoch": 197.19,
      "grad_norm": 1.9135098457336426,
      "learning_rate": 2.533157106885623e-09,
      "loss": 0.4578,
      "step": 481730
    },
    {
      "epoch": 197.19,
      "grad_norm": 1.7473503351211548,
      "learning_rate": 2.5257892503547513e-09,
      "loss": 0.4376,
      "step": 481740
    },
    {
      "epoch": 197.2,
      "grad_norm": 1.9657047986984253,
      "learning_rate": 2.518432118609262e-09,
      "loss": 0.4375,
      "step": 481750
    },
    {
      "epoch": 197.2,
      "grad_norm": 2.0644352436065674,
      "learning_rate": 2.511085711683243e-09,
      "loss": 0.433,
      "step": 481760
    },
    {
      "epoch": 197.2,
      "grad_norm": 1.7961255311965942,
      "learning_rate": 2.5037500296100006e-09,
      "loss": 0.4499,
      "step": 481770
    },
    {
      "epoch": 197.21,
      "grad_norm": 2.1549084186553955,
      "learning_rate": 2.496425072423362e-09,
      "loss": 0.4269,
      "step": 481780
    },
    {
      "epoch": 197.21,
      "grad_norm": 1.8896832466125488,
      "learning_rate": 2.4891108401566344e-09,
      "loss": 0.4447,
      "step": 481790
    },
    {
      "epoch": 197.22,
      "grad_norm": 2.3173351287841797,
      "learning_rate": 2.4818073328433838e-09,
      "loss": 0.4488,
      "step": 481800
    },
    {
      "epoch": 197.22,
      "grad_norm": 2.1197993755340576,
      "learning_rate": 2.4745145505171775e-09,
      "loss": 0.4591,
      "step": 481810
    },
    {
      "epoch": 197.22,
      "grad_norm": 2.1170265674591064,
      "learning_rate": 2.467232493211582e-09,
      "loss": 0.4487,
      "step": 481820
    },
    {
      "epoch": 197.23,
      "grad_norm": 2.436352252960205,
      "learning_rate": 2.459961160959645e-09,
      "loss": 0.4424,
      "step": 481830
    },
    {
      "epoch": 197.23,
      "grad_norm": 2.373950719833374,
      "learning_rate": 2.452700553794672e-09,
      "loss": 0.4602,
      "step": 481840
    },
    {
      "epoch": 197.24,
      "grad_norm": 2.3910577297210693,
      "learning_rate": 2.4454506717499696e-09,
      "loss": 0.4556,
      "step": 481850
    },
    {
      "epoch": 197.24,
      "grad_norm": 2.0472660064697266,
      "learning_rate": 2.4382115148591057e-09,
      "loss": 0.4489,
      "step": 481860
    },
    {
      "epoch": 197.25,
      "grad_norm": 2.1989996433258057,
      "learning_rate": 2.4309830831546058e-09,
      "loss": 0.4388,
      "step": 481870
    },
    {
      "epoch": 197.25,
      "grad_norm": 1.802024006843567,
      "learning_rate": 2.423765376670297e-09,
      "loss": 0.4511,
      "step": 481880
    },
    {
      "epoch": 197.25,
      "grad_norm": 2.2781050205230713,
      "learning_rate": 2.4165583954384448e-09,
      "loss": 0.4461,
      "step": 481890
    },
    {
      "epoch": 197.26,
      "grad_norm": 1.8789976835250854,
      "learning_rate": 2.409362139492877e-09,
      "loss": 0.4358,
      "step": 481900
    },
    {
      "epoch": 197.26,
      "grad_norm": 2.1023073196411133,
      "learning_rate": 2.4021766088658593e-09,
      "loss": 0.4321,
      "step": 481910
    },
    {
      "epoch": 197.27,
      "grad_norm": 1.935013771057129,
      "learning_rate": 2.395001803590698e-09,
      "loss": 0.4495,
      "step": 481920
    },
    {
      "epoch": 197.27,
      "grad_norm": 2.045546054840088,
      "learning_rate": 2.38783772370044e-09,
      "loss": 0.4397,
      "step": 481930
    },
    {
      "epoch": 197.27,
      "grad_norm": 2.21187424659729,
      "learning_rate": 2.3806843692273508e-09,
      "loss": 0.4441,
      "step": 481940
    },
    {
      "epoch": 197.28,
      "grad_norm": 2.192444086074829,
      "learning_rate": 2.3735417402044773e-09,
      "loss": 0.4632,
      "step": 481950
    },
    {
      "epoch": 197.28,
      "grad_norm": 1.6635483503341675,
      "learning_rate": 2.366409836664605e-09,
      "loss": 0.4419,
      "step": 481960
    },
    {
      "epoch": 197.29,
      "grad_norm": 2.207198143005371,
      "learning_rate": 2.3592886586405215e-09,
      "loss": 0.4371,
      "step": 481970
    },
    {
      "epoch": 197.29,
      "grad_norm": 2.1304097175598145,
      "learning_rate": 2.3521782061644914e-09,
      "loss": 0.444,
      "step": 481980
    },
    {
      "epoch": 197.29,
      "grad_norm": 2.2730536460876465,
      "learning_rate": 2.3450784792693015e-09,
      "loss": 0.4513,
      "step": 481990
    },
    {
      "epoch": 197.3,
      "grad_norm": 1.9494017362594604,
      "learning_rate": 2.3379894779874777e-09,
      "loss": 0.4508,
      "step": 482000
    },
    {
      "epoch": 197.3,
      "grad_norm": 2.344820976257324,
      "learning_rate": 2.3309112023515464e-09,
      "loss": 0.425,
      "step": 482010
    },
    {
      "epoch": 197.31,
      "grad_norm": 1.6420159339904785,
      "learning_rate": 2.3238436523937735e-09,
      "loss": 0.4292,
      "step": 482020
    },
    {
      "epoch": 197.31,
      "grad_norm": 1.9507802724838257,
      "learning_rate": 2.316786828146684e-09,
      "loss": 0.4457,
      "step": 482030
    },
    {
      "epoch": 197.31,
      "grad_norm": 2.1285817623138428,
      "learning_rate": 2.309740729642285e-09,
      "loss": 0.4468,
      "step": 482040
    },
    {
      "epoch": 197.32,
      "grad_norm": 2.114206075668335,
      "learning_rate": 2.3027053569133617e-09,
      "loss": 0.4437,
      "step": 482050
    },
    {
      "epoch": 197.32,
      "grad_norm": 2.7952699661254883,
      "learning_rate": 2.2956807099916605e-09,
      "loss": 0.4364,
      "step": 482060
    },
    {
      "epoch": 197.33,
      "grad_norm": 2.1015260219573975,
      "learning_rate": 2.2886667889097066e-09,
      "loss": 0.4294,
      "step": 482070
    },
    {
      "epoch": 197.33,
      "grad_norm": 2.578268051147461,
      "learning_rate": 2.281663593699506e-09,
      "loss": 0.4311,
      "step": 482080
    },
    {
      "epoch": 197.34,
      "grad_norm": 1.9381004571914673,
      "learning_rate": 2.274671124393064e-09,
      "loss": 0.4456,
      "step": 482090
    },
    {
      "epoch": 197.34,
      "grad_norm": 2.054051637649536,
      "learning_rate": 2.2676893810223867e-09,
      "loss": 0.4459,
      "step": 482100
    },
    {
      "epoch": 197.34,
      "grad_norm": 2.187638759613037,
      "learning_rate": 2.26071836361974e-09,
      "loss": 0.4355,
      "step": 482110
    },
    {
      "epoch": 197.35,
      "grad_norm": 2.085932970046997,
      "learning_rate": 2.2537580722168694e-09,
      "loss": 0.4532,
      "step": 482120
    },
    {
      "epoch": 197.35,
      "grad_norm": 1.6811659336090088,
      "learning_rate": 2.2468085068455196e-09,
      "loss": 0.4619,
      "step": 482130
    },
    {
      "epoch": 197.36,
      "grad_norm": 2.0181398391723633,
      "learning_rate": 2.239869667537697e-09,
      "loss": 0.4465,
      "step": 482140
    },
    {
      "epoch": 197.36,
      "grad_norm": 2.006258010864258,
      "learning_rate": 2.232941554325407e-09,
      "loss": 0.4274,
      "step": 482150
    },
    {
      "epoch": 197.36,
      "grad_norm": 1.9890996217727661,
      "learning_rate": 2.2260241672398748e-09,
      "loss": 0.4374,
      "step": 482160
    },
    {
      "epoch": 197.37,
      "grad_norm": 2.435312509536743,
      "learning_rate": 2.2191175063131054e-09,
      "loss": 0.4499,
      "step": 482170
    },
    {
      "epoch": 197.37,
      "grad_norm": 1.5030381679534912,
      "learning_rate": 2.212221571576585e-09,
      "loss": 0.4233,
      "step": 482180
    },
    {
      "epoch": 197.38,
      "grad_norm": 2.1875193119049072,
      "learning_rate": 2.205336363062058e-09,
      "loss": 0.4346,
      "step": 482190
    },
    {
      "epoch": 197.38,
      "grad_norm": 2.136981964111328,
      "learning_rate": 2.1984618808010107e-09,
      "loss": 0.4457,
      "step": 482200
    },
    {
      "epoch": 197.38,
      "grad_norm": 1.9860434532165527,
      "learning_rate": 2.1915981248249276e-09,
      "loss": 0.4311,
      "step": 482210
    },
    {
      "epoch": 197.39,
      "grad_norm": 3.431370973587036,
      "learning_rate": 2.184745095165294e-09,
      "loss": 0.4494,
      "step": 482220
    },
    {
      "epoch": 197.39,
      "grad_norm": 2.450894594192505,
      "learning_rate": 2.1779027918533353e-09,
      "loss": 0.4409,
      "step": 482230
    },
    {
      "epoch": 197.4,
      "grad_norm": 1.75382399559021,
      "learning_rate": 2.1710712149205362e-09,
      "loss": 0.4835,
      "step": 482240
    },
    {
      "epoch": 197.4,
      "grad_norm": 2.4019722938537598,
      "learning_rate": 2.164250364398382e-09,
      "loss": 0.4427,
      "step": 482250
    },
    {
      "epoch": 197.4,
      "grad_norm": 1.9707331657409668,
      "learning_rate": 2.1574402403175784e-09,
      "loss": 0.4284,
      "step": 482260
    },
    {
      "epoch": 197.41,
      "grad_norm": 1.903574824333191,
      "learning_rate": 2.1506408427098693e-09,
      "loss": 0.4617,
      "step": 482270
    },
    {
      "epoch": 197.41,
      "grad_norm": 1.8277885913848877,
      "learning_rate": 2.1438521716062203e-09,
      "loss": 0.4305,
      "step": 482280
    },
    {
      "epoch": 197.42,
      "grad_norm": 1.9732025861740112,
      "learning_rate": 2.137074227037596e-09,
      "loss": 0.4412,
      "step": 482290
    },
    {
      "epoch": 197.42,
      "grad_norm": 2.2172162532806396,
      "learning_rate": 2.1303070090349614e-09,
      "loss": 0.4411,
      "step": 482300
    },
    {
      "epoch": 197.43,
      "grad_norm": 2.2244253158569336,
      "learning_rate": 2.1235505176298016e-09,
      "loss": 0.4269,
      "step": 482310
    },
    {
      "epoch": 197.43,
      "grad_norm": 1.9202159643173218,
      "learning_rate": 2.116804752852561e-09,
      "loss": 0.4653,
      "step": 482320
    },
    {
      "epoch": 197.43,
      "grad_norm": 2.0005102157592773,
      "learning_rate": 2.110069714734465e-09,
      "loss": 0.4382,
      "step": 482330
    },
    {
      "epoch": 197.44,
      "grad_norm": 2.277801752090454,
      "learning_rate": 2.1033454033062177e-09,
      "loss": 0.4333,
      "step": 482340
    },
    {
      "epoch": 197.44,
      "grad_norm": 1.9736982583999634,
      "learning_rate": 2.096631818598524e-09,
      "loss": 0.4331,
      "step": 482350
    },
    {
      "epoch": 197.45,
      "grad_norm": 1.7412396669387817,
      "learning_rate": 2.089928960642348e-09,
      "loss": 0.4325,
      "step": 482360
    },
    {
      "epoch": 197.45,
      "grad_norm": 2.5312185287475586,
      "learning_rate": 2.083236829468395e-09,
      "loss": 0.4371,
      "step": 482370
    },
    {
      "epoch": 197.45,
      "grad_norm": 2.7963480949401855,
      "learning_rate": 2.07655542510711e-09,
      "loss": 0.4498,
      "step": 482380
    },
    {
      "epoch": 197.46,
      "grad_norm": 1.675098180770874,
      "learning_rate": 2.0698847475891963e-09,
      "loss": 0.4421,
      "step": 482390
    },
    {
      "epoch": 197.46,
      "grad_norm": 2.217376708984375,
      "learning_rate": 2.063224796945359e-09,
      "loss": 0.4417,
      "step": 482400
    },
    {
      "epoch": 197.47,
      "grad_norm": 1.9434797763824463,
      "learning_rate": 2.056575573206043e-09,
      "loss": 0.4356,
      "step": 482410
    },
    {
      "epoch": 197.47,
      "grad_norm": 1.8449400663375854,
      "learning_rate": 2.0499370764014317e-09,
      "loss": 0.4373,
      "step": 482420
    },
    {
      "epoch": 197.47,
      "grad_norm": 2.0915563106536865,
      "learning_rate": 2.0433093065622304e-09,
      "loss": 0.4354,
      "step": 482430
    },
    {
      "epoch": 197.48,
      "grad_norm": 2.7981555461883545,
      "learning_rate": 2.036692263718883e-09,
      "loss": 0.4423,
      "step": 482440
    },
    {
      "epoch": 197.48,
      "grad_norm": 2.00180721282959,
      "learning_rate": 2.0300859479013137e-09,
      "loss": 0.444,
      "step": 482450
    },
    {
      "epoch": 197.49,
      "grad_norm": 2.182436943054199,
      "learning_rate": 2.023490359139967e-09,
      "loss": 0.4366,
      "step": 482460
    },
    {
      "epoch": 197.49,
      "grad_norm": 1.7087513208389282,
      "learning_rate": 2.016905497465287e-09,
      "loss": 0.4227,
      "step": 482470
    },
    {
      "epoch": 197.49,
      "grad_norm": 2.0594842433929443,
      "learning_rate": 2.0103313629071977e-09,
      "loss": 0.4456,
      "step": 482480
    },
    {
      "epoch": 197.5,
      "grad_norm": 2.1228575706481934,
      "learning_rate": 2.0037679554958837e-09,
      "loss": 0.462,
      "step": 482490
    },
    {
      "epoch": 197.5,
      "grad_norm": 2.3023924827575684,
      "learning_rate": 1.9972152752612687e-09,
      "loss": 0.4253,
      "step": 482500
    },
    {
      "epoch": 197.51,
      "grad_norm": 2.0653295516967773,
      "learning_rate": 1.9906733222335366e-09,
      "loss": 0.4602,
      "step": 482510
    },
    {
      "epoch": 197.51,
      "grad_norm": 1.8522697687149048,
      "learning_rate": 1.9841420964428723e-09,
      "loss": 0.4297,
      "step": 482520
    },
    {
      "epoch": 197.52,
      "grad_norm": 2.2543587684631348,
      "learning_rate": 1.9776215979186785e-09,
      "loss": 0.432,
      "step": 482530
    },
    {
      "epoch": 197.52,
      "grad_norm": 2.114093065261841,
      "learning_rate": 1.97111182669114e-09,
      "loss": 0.438,
      "step": 482540
    },
    {
      "epoch": 197.52,
      "grad_norm": 1.983607292175293,
      "learning_rate": 1.9646127827899204e-09,
      "loss": 0.444,
      "step": 482550
    },
    {
      "epoch": 197.53,
      "grad_norm": 2.416229009628296,
      "learning_rate": 1.958124466244944e-09,
      "loss": 0.4416,
      "step": 482560
    },
    {
      "epoch": 197.53,
      "grad_norm": 2.0619001388549805,
      "learning_rate": 1.951646877085874e-09,
      "loss": 0.4566,
      "step": 482570
    },
    {
      "epoch": 197.54,
      "grad_norm": 2.2366981506347656,
      "learning_rate": 1.9451800153426346e-09,
      "loss": 0.4571,
      "step": 482580
    },
    {
      "epoch": 197.54,
      "grad_norm": 2.0825984477996826,
      "learning_rate": 1.938723881044369e-09,
      "loss": 0.4488,
      "step": 482590
    },
    {
      "epoch": 197.54,
      "grad_norm": 1.835930585861206,
      "learning_rate": 1.9322784742207415e-09,
      "loss": 0.4457,
      "step": 482600
    },
    {
      "epoch": 197.55,
      "grad_norm": 2.051891326904297,
      "learning_rate": 1.9258437949016756e-09,
      "loss": 0.4516,
      "step": 482610
    },
    {
      "epoch": 197.55,
      "grad_norm": 1.7940086126327515,
      "learning_rate": 1.919419843116315e-09,
      "loss": 0.437,
      "step": 482620
    },
    {
      "epoch": 197.56,
      "grad_norm": 1.958666205406189,
      "learning_rate": 1.9130066188940624e-09,
      "loss": 0.4528,
      "step": 482630
    },
    {
      "epoch": 197.56,
      "grad_norm": 1.7248846292495728,
      "learning_rate": 1.906604122264583e-09,
      "loss": 0.4358,
      "step": 482640
    },
    {
      "epoch": 197.56,
      "grad_norm": 1.9737306833267212,
      "learning_rate": 1.900212353256759e-09,
      "loss": 0.455,
      "step": 482650
    },
    {
      "epoch": 197.57,
      "grad_norm": 2.266988754272461,
      "learning_rate": 1.893831311900254e-09,
      "loss": 0.4437,
      "step": 482660
    },
    {
      "epoch": 197.57,
      "grad_norm": 1.6454161405563354,
      "learning_rate": 1.887460998224212e-09,
      "loss": 0.4365,
      "step": 482670
    },
    {
      "epoch": 197.58,
      "grad_norm": 2.071261167526245,
      "learning_rate": 1.881101412257776e-09,
      "loss": 0.46,
      "step": 482680
    },
    {
      "epoch": 197.58,
      "grad_norm": 1.9019849300384521,
      "learning_rate": 1.8747525540300892e-09,
      "loss": 0.4276,
      "step": 482690
    },
    {
      "epoch": 197.58,
      "grad_norm": 1.7513617277145386,
      "learning_rate": 1.8684144235702956e-09,
      "loss": 0.4533,
      "step": 482700
    },
    {
      "epoch": 197.59,
      "grad_norm": 1.824881911277771,
      "learning_rate": 1.8620870209072776e-09,
      "loss": 0.4724,
      "step": 482710
    },
    {
      "epoch": 197.59,
      "grad_norm": 2.3201215267181396,
      "learning_rate": 1.8557703460701789e-09,
      "loss": 0.4233,
      "step": 482720
    },
    {
      "epoch": 197.6,
      "grad_norm": 2.2148115634918213,
      "learning_rate": 1.8494643990878827e-09,
      "loss": 0.4496,
      "step": 482730
    },
    {
      "epoch": 197.6,
      "grad_norm": 1.7328715324401855,
      "learning_rate": 1.8431691799890118e-09,
      "loss": 0.4562,
      "step": 482740
    },
    {
      "epoch": 197.61,
      "grad_norm": 1.8876674175262451,
      "learning_rate": 1.8368846888029697e-09,
      "loss": 0.442,
      "step": 482750
    },
    {
      "epoch": 197.61,
      "grad_norm": 2.4654381275177,
      "learning_rate": 1.8306109255578593e-09,
      "loss": 0.4388,
      "step": 482760
    },
    {
      "epoch": 197.61,
      "grad_norm": 1.9423487186431885,
      "learning_rate": 1.8243478902830837e-09,
      "loss": 0.4408,
      "step": 482770
    },
    {
      "epoch": 197.62,
      "grad_norm": 2.004995584487915,
      "learning_rate": 1.8180955830067456e-09,
      "loss": 0.4391,
      "step": 482780
    },
    {
      "epoch": 197.62,
      "grad_norm": 2.1379384994506836,
      "learning_rate": 1.8118540037579884e-09,
      "loss": 0.4426,
      "step": 482790
    },
    {
      "epoch": 197.63,
      "grad_norm": 1.8665134906768799,
      "learning_rate": 1.805623152565175e-09,
      "loss": 0.4496,
      "step": 482800
    },
    {
      "epoch": 197.63,
      "grad_norm": 2.3456828594207764,
      "learning_rate": 1.7994030294566675e-09,
      "loss": 0.438,
      "step": 482810
    },
    {
      "epoch": 197.63,
      "grad_norm": 1.9148006439208984,
      "learning_rate": 1.7931936344613498e-09,
      "loss": 0.4617,
      "step": 482820
    },
    {
      "epoch": 197.64,
      "grad_norm": 2.410244941711426,
      "learning_rate": 1.786994967607324e-09,
      "loss": 0.4186,
      "step": 482830
    },
    {
      "epoch": 197.64,
      "grad_norm": 1.9997929334640503,
      "learning_rate": 1.7808070289232133e-09,
      "loss": 0.4534,
      "step": 482840
    },
    {
      "epoch": 197.65,
      "grad_norm": 2.1567909717559814,
      "learning_rate": 1.77462981843738e-09,
      "loss": 0.4433,
      "step": 482850
    },
    {
      "epoch": 197.65,
      "grad_norm": 2.0231411457061768,
      "learning_rate": 1.7684633361779269e-09,
      "loss": 0.4448,
      "step": 482860
    },
    {
      "epoch": 197.65,
      "grad_norm": 2.162048578262329,
      "learning_rate": 1.7623075821729566e-09,
      "loss": 0.4548,
      "step": 482870
    },
    {
      "epoch": 197.66,
      "grad_norm": 2.133885622024536,
      "learning_rate": 1.756162556451092e-09,
      "loss": 0.4481,
      "step": 482880
    },
    {
      "epoch": 197.66,
      "grad_norm": 2.096083879470825,
      "learning_rate": 1.7500282590404354e-09,
      "loss": 0.4238,
      "step": 482890
    },
    {
      "epoch": 197.67,
      "grad_norm": 2.215536117553711,
      "learning_rate": 1.7439046899688295e-09,
      "loss": 0.4553,
      "step": 482900
    },
    {
      "epoch": 197.67,
      "grad_norm": 1.9374603033065796,
      "learning_rate": 1.7377918492643763e-09,
      "loss": 0.4373,
      "step": 482910
    },
    {
      "epoch": 197.67,
      "grad_norm": 2.4209578037261963,
      "learning_rate": 1.731689736955179e-09,
      "loss": 0.4223,
      "step": 482920
    },
    {
      "epoch": 197.68,
      "grad_norm": 2.438924789428711,
      "learning_rate": 1.7255983530693393e-09,
      "loss": 0.4575,
      "step": 482930
    },
    {
      "epoch": 197.68,
      "grad_norm": 2.2897837162017822,
      "learning_rate": 1.7195176976341797e-09,
      "loss": 0.4584,
      "step": 482940
    },
    {
      "epoch": 197.69,
      "grad_norm": 2.146959066390991,
      "learning_rate": 1.713447770678323e-09,
      "loss": 0.4368,
      "step": 482950
    },
    {
      "epoch": 197.69,
      "grad_norm": 1.7955330610275269,
      "learning_rate": 1.707388572229091e-09,
      "loss": 0.4306,
      "step": 482960
    },
    {
      "epoch": 197.7,
      "grad_norm": 1.669766902923584,
      "learning_rate": 1.7013401023143261e-09,
      "loss": 0.4412,
      "step": 482970
    },
    {
      "epoch": 197.7,
      "grad_norm": 1.763484001159668,
      "learning_rate": 1.6953023609616104e-09,
      "loss": 0.4548,
      "step": 482980
    },
    {
      "epoch": 197.7,
      "grad_norm": 2.2689719200134277,
      "learning_rate": 1.6892753481990464e-09,
      "loss": 0.449,
      "step": 482990
    },
    {
      "epoch": 197.71,
      "grad_norm": 2.011657476425171,
      "learning_rate": 1.6832590640536958e-09,
      "loss": 0.4376,
      "step": 483000
    },
    {
      "epoch": 197.71,
      "grad_norm": 2.0530929565429688,
      "learning_rate": 1.6772535085534008e-09,
      "loss": 0.4374,
      "step": 483010
    },
    {
      "epoch": 197.72,
      "grad_norm": 2.320255994796753,
      "learning_rate": 1.6712586817257436e-09,
      "loss": 0.4292,
      "step": 483020
    },
    {
      "epoch": 197.72,
      "grad_norm": 1.5360301733016968,
      "learning_rate": 1.6652745835980461e-09,
      "loss": 0.4659,
      "step": 483030
    },
    {
      "epoch": 197.72,
      "grad_norm": 1.9049127101898193,
      "learning_rate": 1.6593012141978904e-09,
      "loss": 0.4532,
      "step": 483040
    },
    {
      "epoch": 197.73,
      "grad_norm": 2.157968521118164,
      "learning_rate": 1.6533385735523381e-09,
      "loss": 0.4526,
      "step": 483050
    },
    {
      "epoch": 197.73,
      "grad_norm": 2.150747537612915,
      "learning_rate": 1.6473866616887114e-09,
      "loss": 0.433,
      "step": 483060
    },
    {
      "epoch": 197.74,
      "grad_norm": 1.9502017498016357,
      "learning_rate": 1.6414454786348522e-09,
      "loss": 0.4316,
      "step": 483070
    },
    {
      "epoch": 197.74,
      "grad_norm": 1.8218908309936523,
      "learning_rate": 1.635515024417302e-09,
      "loss": 0.4411,
      "step": 483080
    },
    {
      "epoch": 197.74,
      "grad_norm": 2.38911509513855,
      "learning_rate": 1.6295952990633826e-09,
      "loss": 0.4436,
      "step": 483090
    },
    {
      "epoch": 197.75,
      "grad_norm": 1.9810911417007446,
      "learning_rate": 1.623686302600416e-09,
      "loss": 0.447,
      "step": 483100
    },
    {
      "epoch": 197.75,
      "grad_norm": 2.2042858600616455,
      "learning_rate": 1.6177880350554638e-09,
      "loss": 0.4375,
      "step": 483110
    },
    {
      "epoch": 197.76,
      "grad_norm": 2.155620813369751,
      "learning_rate": 1.6119004964555877e-09,
      "loss": 0.4316,
      "step": 483120
    },
    {
      "epoch": 197.76,
      "grad_norm": 2.684797763824463,
      "learning_rate": 1.606023686827329e-09,
      "loss": 0.4381,
      "step": 483130
    },
    {
      "epoch": 197.77,
      "grad_norm": 1.943386435508728,
      "learning_rate": 1.6001576061980096e-09,
      "loss": 0.4427,
      "step": 483140
    },
    {
      "epoch": 197.77,
      "grad_norm": 2.0017387866973877,
      "learning_rate": 1.594302254594431e-09,
      "loss": 0.4438,
      "step": 483150
    },
    {
      "epoch": 197.77,
      "grad_norm": 2.155108690261841,
      "learning_rate": 1.5884576320431344e-09,
      "loss": 0.4461,
      "step": 483160
    },
    {
      "epoch": 197.78,
      "grad_norm": 2.2821993827819824,
      "learning_rate": 1.5826237385714417e-09,
      "loss": 0.4658,
      "step": 483170
    },
    {
      "epoch": 197.78,
      "grad_norm": 1.7060357332229614,
      "learning_rate": 1.576800574205634e-09,
      "loss": 0.4416,
      "step": 483180
    },
    {
      "epoch": 197.79,
      "grad_norm": 1.870288610458374,
      "learning_rate": 1.5709881389725128e-09,
      "loss": 0.433,
      "step": 483190
    },
    {
      "epoch": 197.79,
      "grad_norm": 2.2631571292877197,
      "learning_rate": 1.5651864328988796e-09,
      "loss": 0.4337,
      "step": 483200
    },
    {
      "epoch": 197.79,
      "grad_norm": 1.9015917778015137,
      "learning_rate": 1.5593954560110152e-09,
      "loss": 0.4513,
      "step": 483210
    },
    {
      "epoch": 197.8,
      "grad_norm": 1.7958303689956665,
      "learning_rate": 1.5536152083354612e-09,
      "loss": 0.4396,
      "step": 483220
    },
    {
      "epoch": 197.8,
      "grad_norm": 1.9278005361557007,
      "learning_rate": 1.5478456898990188e-09,
      "loss": 0.4583,
      "step": 483230
    },
    {
      "epoch": 197.81,
      "grad_norm": 2.3903017044067383,
      "learning_rate": 1.542086900727709e-09,
      "loss": 0.4434,
      "step": 483240
    },
    {
      "epoch": 197.81,
      "grad_norm": 2.104468822479248,
      "learning_rate": 1.5363388408480732e-09,
      "loss": 0.4429,
      "step": 483250
    },
    {
      "epoch": 197.81,
      "grad_norm": 2.1047275066375732,
      "learning_rate": 1.5306015102866526e-09,
      "loss": 0.4351,
      "step": 483260
    },
    {
      "epoch": 197.82,
      "grad_norm": 2.0081610679626465,
      "learning_rate": 1.5248749090697282e-09,
      "loss": 0.4503,
      "step": 483270
    },
    {
      "epoch": 197.82,
      "grad_norm": 1.7820905447006226,
      "learning_rate": 1.5191590372230604e-09,
      "loss": 0.4297,
      "step": 483280
    },
    {
      "epoch": 197.83,
      "grad_norm": 1.9306013584136963,
      "learning_rate": 1.5134538947731909e-09,
      "loss": 0.4554,
      "step": 483290
    },
    {
      "epoch": 197.83,
      "grad_norm": 2.3935599327087402,
      "learning_rate": 1.5077594817461402e-09,
      "loss": 0.447,
      "step": 483300
    },
    {
      "epoch": 197.83,
      "grad_norm": 1.9244917631149292,
      "learning_rate": 1.5020757981681897e-09,
      "loss": 0.4383,
      "step": 483310
    },
    {
      "epoch": 197.84,
      "grad_norm": 1.9845725297927856,
      "learning_rate": 1.49640284406536e-09,
      "loss": 0.4306,
      "step": 483320
    },
    {
      "epoch": 197.84,
      "grad_norm": 1.9474656581878662,
      "learning_rate": 1.490740619463412e-09,
      "loss": 0.4484,
      "step": 483330
    },
    {
      "epoch": 197.85,
      "grad_norm": 1.941420316696167,
      "learning_rate": 1.485089124388366e-09,
      "loss": 0.4446,
      "step": 483340
    },
    {
      "epoch": 197.85,
      "grad_norm": 1.9796680212020874,
      "learning_rate": 1.4794483588659832e-09,
      "loss": 0.4336,
      "step": 483350
    },
    {
      "epoch": 197.86,
      "grad_norm": 2.1973788738250732,
      "learning_rate": 1.4738183229222843e-09,
      "loss": 0.4466,
      "step": 483360
    },
    {
      "epoch": 197.86,
      "grad_norm": 1.9566094875335693,
      "learning_rate": 1.46819901658329e-09,
      "loss": 0.4397,
      "step": 483370
    },
    {
      "epoch": 197.86,
      "grad_norm": 2.676663875579834,
      "learning_rate": 1.4625904398742407e-09,
      "loss": 0.436,
      "step": 483380
    },
    {
      "epoch": 197.87,
      "grad_norm": 2.1917479038238525,
      "learning_rate": 1.4569925928211573e-09,
      "loss": 0.4398,
      "step": 483390
    },
    {
      "epoch": 197.87,
      "grad_norm": 2.360661745071411,
      "learning_rate": 1.45140547544954e-09,
      "loss": 0.4496,
      "step": 483400
    },
    {
      "epoch": 197.88,
      "grad_norm": 3.2289533615112305,
      "learning_rate": 1.4458290877848896e-09,
      "loss": 0.4535,
      "step": 483410
    },
    {
      "epoch": 197.88,
      "grad_norm": 2.082826614379883,
      "learning_rate": 1.4402634298529663e-09,
      "loss": 0.4603,
      "step": 483420
    },
    {
      "epoch": 197.88,
      "grad_norm": 2.0680365562438965,
      "learning_rate": 1.4347085016792708e-09,
      "loss": 0.4415,
      "step": 483430
    },
    {
      "epoch": 197.89,
      "grad_norm": 2.2041285037994385,
      "learning_rate": 1.4291643032893033e-09,
      "loss": 0.4563,
      "step": 483440
    },
    {
      "epoch": 197.89,
      "grad_norm": 2.1948375701904297,
      "learning_rate": 1.4236308347083046e-09,
      "loss": 0.445,
      "step": 483450
    },
    {
      "epoch": 197.9,
      "grad_norm": 2.5878875255584717,
      "learning_rate": 1.4181080959615142e-09,
      "loss": 0.4352,
      "step": 483460
    },
    {
      "epoch": 197.9,
      "grad_norm": 2.022029161453247,
      "learning_rate": 1.412596087074433e-09,
      "loss": 0.442,
      "step": 483470
    },
    {
      "epoch": 197.9,
      "grad_norm": 1.7321257591247559,
      "learning_rate": 1.4070948080720408e-09,
      "loss": 0.4539,
      "step": 483480
    },
    {
      "epoch": 197.91,
      "grad_norm": 2.384082555770874,
      "learning_rate": 1.4016042589800985e-09,
      "loss": 0.4428,
      "step": 483490
    },
    {
      "epoch": 197.91,
      "grad_norm": 1.658305287361145,
      "learning_rate": 1.3961244398230653e-09,
      "loss": 0.4263,
      "step": 483500
    },
    {
      "epoch": 197.92,
      "grad_norm": 2.0312745571136475,
      "learning_rate": 1.390655350626442e-09,
      "loss": 0.4409,
      "step": 483510
    },
    {
      "epoch": 197.92,
      "grad_norm": 2.0519795417785645,
      "learning_rate": 1.3851969914154687e-09,
      "loss": 0.4488,
      "step": 483520
    },
    {
      "epoch": 197.92,
      "grad_norm": 2.0085978507995605,
      "learning_rate": 1.379749362214605e-09,
      "loss": 0.4652,
      "step": 483530
    },
    {
      "epoch": 197.93,
      "grad_norm": 2.026198148727417,
      "learning_rate": 1.3743124630493513e-09,
      "loss": 0.4603,
      "step": 483540
    },
    {
      "epoch": 197.93,
      "grad_norm": 2.08512806892395,
      "learning_rate": 1.3688862939441672e-09,
      "loss": 0.4287,
      "step": 483550
    },
    {
      "epoch": 197.94,
      "grad_norm": 1.582051157951355,
      "learning_rate": 1.3634708549240328e-09,
      "loss": 0.4368,
      "step": 483560
    },
    {
      "epoch": 197.94,
      "grad_norm": 2.6561789512634277,
      "learning_rate": 1.3580661460141883e-09,
      "loss": 0.4388,
      "step": 483570
    },
    {
      "epoch": 197.95,
      "grad_norm": 2.0658788681030273,
      "learning_rate": 1.3526721672385727e-09,
      "loss": 0.4186,
      "step": 483580
    },
    {
      "epoch": 197.95,
      "grad_norm": 2.2811825275421143,
      "learning_rate": 1.3472889186226867e-09,
      "loss": 0.4147,
      "step": 483590
    },
    {
      "epoch": 197.95,
      "grad_norm": 2.1874022483825684,
      "learning_rate": 1.3419164001907296e-09,
      "loss": 0.4335,
      "step": 483600
    },
    {
      "epoch": 197.96,
      "grad_norm": 1.882328987121582,
      "learning_rate": 1.3365546119674213e-09,
      "loss": 0.4423,
      "step": 483610
    },
    {
      "epoch": 197.96,
      "grad_norm": 2.112658977508545,
      "learning_rate": 1.3312035539774814e-09,
      "loss": 0.4256,
      "step": 483620
    },
    {
      "epoch": 197.97,
      "grad_norm": 1.8469510078430176,
      "learning_rate": 1.3258632262451093e-09,
      "loss": 0.454,
      "step": 483630
    },
    {
      "epoch": 197.97,
      "grad_norm": 1.7284349203109741,
      "learning_rate": 1.3205336287950251e-09,
      "loss": 0.4278,
      "step": 483640
    },
    {
      "epoch": 197.97,
      "grad_norm": 1.9146552085876465,
      "learning_rate": 1.3152147616514278e-09,
      "loss": 0.4471,
      "step": 483650
    },
    {
      "epoch": 197.98,
      "grad_norm": 2.2997238636016846,
      "learning_rate": 1.3099066248387774e-09,
      "loss": 0.4406,
      "step": 483660
    },
    {
      "epoch": 197.98,
      "grad_norm": 1.9912580251693726,
      "learning_rate": 1.3046092183815332e-09,
      "loss": 0.4603,
      "step": 483670
    },
    {
      "epoch": 197.99,
      "grad_norm": 2.093039035797119,
      "learning_rate": 1.2993225423038946e-09,
      "loss": 0.4472,
      "step": 483680
    },
    {
      "epoch": 197.99,
      "grad_norm": 2.3070452213287354,
      "learning_rate": 1.2940465966300613e-09,
      "loss": 0.4359,
      "step": 483690
    },
    {
      "epoch": 197.99,
      "grad_norm": 2.4011900424957275,
      "learning_rate": 1.2887813813842323e-09,
      "loss": 0.4395,
      "step": 483700
    },
    {
      "epoch": 198.0,
      "grad_norm": 2.6764209270477295,
      "learning_rate": 1.283526896590347e-09,
      "loss": 0.43,
      "step": 483710
    },
    {
      "epoch": 198.0,
      "eval_loss": 0.4448104202747345,
      "eval_runtime": 43.359,
      "eval_samples_per_second": 79.545,
      "eval_steps_per_second": 9.963,
      "step": 483714
    },
    {
      "epoch": 198.0,
      "grad_norm": 1.866934895515442,
      "learning_rate": 1.2782831422726049e-09,
      "loss": 0.4488,
      "step": 483720
    },
    {
      "epoch": 198.01,
      "grad_norm": 1.7986252307891846,
      "learning_rate": 1.2730501184549449e-09,
      "loss": 0.4585,
      "step": 483730
    },
    {
      "epoch": 198.01,
      "grad_norm": 2.3009939193725586,
      "learning_rate": 1.2678278251615667e-09,
      "loss": 0.4634,
      "step": 483740
    },
    {
      "epoch": 198.01,
      "grad_norm": 1.9941760301589966,
      "learning_rate": 1.2626162624161489e-09,
      "loss": 0.4488,
      "step": 483750
    },
    {
      "epoch": 198.02,
      "grad_norm": 2.2098710536956787,
      "learning_rate": 1.257415430242631e-09,
      "loss": 0.4568,
      "step": 483760
    },
    {
      "epoch": 198.02,
      "grad_norm": 1.9909262657165527,
      "learning_rate": 1.252225328664952e-09,
      "loss": 0.4484,
      "step": 483770
    },
    {
      "epoch": 198.03,
      "grad_norm": 2.2212696075439453,
      "learning_rate": 1.2470459577065306e-09,
      "loss": 0.4615,
      "step": 483780
    },
    {
      "epoch": 198.03,
      "grad_norm": 1.8734196424484253,
      "learning_rate": 1.2418773173915667e-09,
      "loss": 0.4545,
      "step": 483790
    },
    {
      "epoch": 198.04,
      "grad_norm": 1.71030855178833,
      "learning_rate": 1.2367194077437386e-09,
      "loss": 0.4469,
      "step": 483800
    },
    {
      "epoch": 198.04,
      "grad_norm": 2.554654598236084,
      "learning_rate": 1.2315722287862054e-09,
      "loss": 0.4554,
      "step": 483810
    },
    {
      "epoch": 198.04,
      "grad_norm": 1.8734384775161743,
      "learning_rate": 1.226435780542906e-09,
      "loss": 0.4615,
      "step": 483820
    },
    {
      "epoch": 198.05,
      "grad_norm": 2.5133025646209717,
      "learning_rate": 1.221310063037259e-09,
      "loss": 0.4379,
      "step": 483830
    },
    {
      "epoch": 198.05,
      "grad_norm": 2.1165332794189453,
      "learning_rate": 1.2161950762926836e-09,
      "loss": 0.4508,
      "step": 483840
    },
    {
      "epoch": 198.06,
      "grad_norm": 2.1590981483459473,
      "learning_rate": 1.2110908203325983e-09,
      "loss": 0.4487,
      "step": 483850
    },
    {
      "epoch": 198.06,
      "grad_norm": 2.179182529449463,
      "learning_rate": 1.2059972951806821e-09,
      "loss": 0.4358,
      "step": 483860
    },
    {
      "epoch": 198.06,
      "grad_norm": 2.2301130294799805,
      "learning_rate": 1.2009145008598335e-09,
      "loss": 0.4324,
      "step": 483870
    },
    {
      "epoch": 198.07,
      "grad_norm": 3.8349852561950684,
      "learning_rate": 1.1958424373937313e-09,
      "loss": 0.4322,
      "step": 483880
    },
    {
      "epoch": 198.07,
      "grad_norm": 2.1713788509368896,
      "learning_rate": 1.1907811048052738e-09,
      "loss": 0.4601,
      "step": 483890
    },
    {
      "epoch": 198.08,
      "grad_norm": 3.2811293601989746,
      "learning_rate": 1.1857305031181402e-09,
      "loss": 0.4443,
      "step": 483900
    },
    {
      "epoch": 198.08,
      "grad_norm": 2.2360849380493164,
      "learning_rate": 1.1806906323549684e-09,
      "loss": 0.4314,
      "step": 483910
    },
    {
      "epoch": 198.08,
      "grad_norm": 2.1289169788360596,
      "learning_rate": 1.1756614925389172e-09,
      "loss": 0.4745,
      "step": 483920
    },
    {
      "epoch": 198.09,
      "grad_norm": 2.2060866355895996,
      "learning_rate": 1.170643083693145e-09,
      "loss": 0.4447,
      "step": 483930
    },
    {
      "epoch": 198.09,
      "grad_norm": 1.9886912107467651,
      "learning_rate": 1.1656354058405501e-09,
      "loss": 0.442,
      "step": 483940
    },
    {
      "epoch": 198.1,
      "grad_norm": 1.7670587301254272,
      "learning_rate": 1.1606384590042913e-09,
      "loss": 0.4388,
      "step": 483950
    },
    {
      "epoch": 198.1,
      "grad_norm": 2.0773327350616455,
      "learning_rate": 1.1556522432070066e-09,
      "loss": 0.4485,
      "step": 483960
    },
    {
      "epoch": 198.1,
      "grad_norm": 1.8057423830032349,
      "learning_rate": 1.1506767584715942e-09,
      "loss": 0.4333,
      "step": 483970
    },
    {
      "epoch": 198.11,
      "grad_norm": 1.9931873083114624,
      "learning_rate": 1.1457120048209527e-09,
      "loss": 0.438,
      "step": 483980
    },
    {
      "epoch": 198.11,
      "grad_norm": 2.0176618099212646,
      "learning_rate": 1.14075798227772e-09,
      "loss": 0.4441,
      "step": 483990
    },
    {
      "epoch": 198.12,
      "grad_norm": 2.1814112663269043,
      "learning_rate": 1.1358146908645345e-09,
      "loss": 0.4216,
      "step": 484000
    },
    {
      "epoch": 198.12,
      "grad_norm": 1.7709556818008423,
      "learning_rate": 1.1308821306042944e-09,
      "loss": 0.4393,
      "step": 484010
    },
    {
      "epoch": 198.13,
      "grad_norm": 2.1078741550445557,
      "learning_rate": 1.1259603015193776e-09,
      "loss": 0.4361,
      "step": 484020
    },
    {
      "epoch": 198.13,
      "grad_norm": 2.496427297592163,
      "learning_rate": 1.1210492036324223e-09,
      "loss": 0.4581,
      "step": 484030
    },
    {
      "epoch": 198.13,
      "grad_norm": 2.244811534881592,
      "learning_rate": 1.1161488369658065e-09,
      "loss": 0.4391,
      "step": 484040
    },
    {
      "epoch": 198.14,
      "grad_norm": 1.8753526210784912,
      "learning_rate": 1.1112592015421681e-09,
      "loss": 0.435,
      "step": 484050
    },
    {
      "epoch": 198.14,
      "grad_norm": 2.1705703735351562,
      "learning_rate": 1.1063802973836252e-09,
      "loss": 0.472,
      "step": 484060
    },
    {
      "epoch": 198.15,
      "grad_norm": 2.2107956409454346,
      "learning_rate": 1.1015121245128155e-09,
      "loss": 0.4516,
      "step": 484070
    },
    {
      "epoch": 198.15,
      "grad_norm": 2.279146671295166,
      "learning_rate": 1.0966546829521173e-09,
      "loss": 0.438,
      "step": 484080
    },
    {
      "epoch": 198.15,
      "grad_norm": 2.4850242137908936,
      "learning_rate": 1.091807972723388e-09,
      "loss": 0.4475,
      "step": 484090
    },
    {
      "epoch": 198.16,
      "grad_norm": 1.932492971420288,
      "learning_rate": 1.0869719938490056e-09,
      "loss": 0.4516,
      "step": 484100
    },
    {
      "epoch": 198.16,
      "grad_norm": 2.2961342334747314,
      "learning_rate": 1.0821467463510876e-09,
      "loss": 0.426,
      "step": 484110
    },
    {
      "epoch": 198.17,
      "grad_norm": 2.041384696960449,
      "learning_rate": 1.077332230252012e-09,
      "loss": 0.4237,
      "step": 484120
    },
    {
      "epoch": 198.17,
      "grad_norm": 2.1849918365478516,
      "learning_rate": 1.0725284455733764e-09,
      "loss": 0.4435,
      "step": 484130
    },
    {
      "epoch": 198.17,
      "grad_norm": 2.651289939880371,
      "learning_rate": 1.0677353923375585e-09,
      "loss": 0.4691,
      "step": 484140
    },
    {
      "epoch": 198.18,
      "grad_norm": 1.945177435874939,
      "learning_rate": 1.0629530705664156e-09,
      "loss": 0.4539,
      "step": 484150
    },
    {
      "epoch": 198.18,
      "grad_norm": 2.444638729095459,
      "learning_rate": 1.0581814802818057e-09,
      "loss": 0.4436,
      "step": 484160
    },
    {
      "epoch": 198.19,
      "grad_norm": 2.4350249767303467,
      "learning_rate": 1.053420621505586e-09,
      "loss": 0.4614,
      "step": 484170
    },
    {
      "epoch": 198.19,
      "grad_norm": 1.934563159942627,
      "learning_rate": 1.0486704942596139e-09,
      "loss": 0.446,
      "step": 484180
    },
    {
      "epoch": 198.19,
      "grad_norm": 2.2114274501800537,
      "learning_rate": 1.043931098565487e-09,
      "loss": 0.4445,
      "step": 484190
    },
    {
      "epoch": 198.2,
      "grad_norm": 1.740790605545044,
      "learning_rate": 1.0392024344450626e-09,
      "loss": 0.4269,
      "step": 484200
    },
    {
      "epoch": 198.2,
      "grad_norm": 2.2042198181152344,
      "learning_rate": 1.0344845019201984e-09,
      "loss": 0.4514,
      "step": 484210
    },
    {
      "epoch": 198.21,
      "grad_norm": 1.7139761447906494,
      "learning_rate": 1.0297773010122316e-09,
      "loss": 0.4327,
      "step": 484220
    },
    {
      "epoch": 198.21,
      "grad_norm": 1.7806708812713623,
      "learning_rate": 1.025080831742759e-09,
      "loss": 0.424,
      "step": 484230
    },
    {
      "epoch": 198.22,
      "grad_norm": 1.9794524908065796,
      "learning_rate": 1.0203950941331181e-09,
      "loss": 0.4681,
      "step": 484240
    },
    {
      "epoch": 198.22,
      "grad_norm": 1.81545090675354,
      "learning_rate": 1.0157200882051664e-09,
      "loss": 0.4315,
      "step": 484250
    },
    {
      "epoch": 198.22,
      "grad_norm": 1.7924765348434448,
      "learning_rate": 1.011055813980241e-09,
      "loss": 0.437,
      "step": 484260
    },
    {
      "epoch": 198.23,
      "grad_norm": 1.9182710647583008,
      "learning_rate": 1.0064022714794187e-09,
      "loss": 0.4482,
      "step": 484270
    },
    {
      "epoch": 198.23,
      "grad_norm": 2.278701066970825,
      "learning_rate": 1.0017594607242969e-09,
      "loss": 0.4518,
      "step": 484280
    },
    {
      "epoch": 198.24,
      "grad_norm": 2.42193865776062,
      "learning_rate": 9.971273817359522e-10,
      "loss": 0.4481,
      "step": 484290
    },
    {
      "epoch": 198.24,
      "grad_norm": 1.7701966762542725,
      "learning_rate": 9.92506034535722e-10,
      "loss": 0.4433,
      "step": 484300
    },
    {
      "epoch": 198.24,
      "grad_norm": 1.8301023244857788,
      "learning_rate": 9.878954191446832e-10,
      "loss": 0.4418,
      "step": 484310
    },
    {
      "epoch": 198.25,
      "grad_norm": 1.9429186582565308,
      "learning_rate": 9.832955355841727e-10,
      "loss": 0.4542,
      "step": 484320
    },
    {
      "epoch": 198.25,
      "grad_norm": 2.127073287963867,
      "learning_rate": 9.787063838750074e-10,
      "loss": 0.4653,
      "step": 484330
    },
    {
      "epoch": 198.26,
      "grad_norm": 2.0179085731506348,
      "learning_rate": 9.741279640385243e-10,
      "loss": 0.4434,
      "step": 484340
    },
    {
      "epoch": 198.26,
      "grad_norm": 2.0592198371887207,
      "learning_rate": 9.695602760952798e-10,
      "loss": 0.462,
      "step": 484350
    },
    {
      "epoch": 198.26,
      "grad_norm": 1.875564694404602,
      "learning_rate": 9.650033200663508e-10,
      "loss": 0.4398,
      "step": 484360
    },
    {
      "epoch": 198.27,
      "grad_norm": 2.0865564346313477,
      "learning_rate": 9.604570959728142e-10,
      "loss": 0.4589,
      "step": 484370
    },
    {
      "epoch": 198.27,
      "grad_norm": 2.195943593978882,
      "learning_rate": 9.559216038352267e-10,
      "loss": 0.4497,
      "step": 484380
    },
    {
      "epoch": 198.28,
      "grad_norm": 1.9560773372650146,
      "learning_rate": 9.51396843674665e-10,
      "loss": 0.4343,
      "step": 484390
    },
    {
      "epoch": 198.28,
      "grad_norm": 1.709317922592163,
      "learning_rate": 9.468828155116854e-10,
      "loss": 0.4376,
      "step": 484400
    },
    {
      "epoch": 198.28,
      "grad_norm": 2.313563823699951,
      "learning_rate": 9.423795193668447e-10,
      "loss": 0.4369,
      "step": 484410
    },
    {
      "epoch": 198.29,
      "grad_norm": 1.5080647468566895,
      "learning_rate": 9.378869552606991e-10,
      "loss": 0.4444,
      "step": 484420
    },
    {
      "epoch": 198.29,
      "grad_norm": 1.8097294569015503,
      "learning_rate": 9.334051232143258e-10,
      "loss": 0.4279,
      "step": 484430
    },
    {
      "epoch": 198.3,
      "grad_norm": 2.0482959747314453,
      "learning_rate": 9.289340232477606e-10,
      "loss": 0.444,
      "step": 484440
    },
    {
      "epoch": 198.3,
      "grad_norm": 2.1079416275024414,
      "learning_rate": 9.2447365538156e-10,
      "loss": 0.4463,
      "step": 484450
    },
    {
      "epoch": 198.31,
      "grad_norm": 2.1481335163116455,
      "learning_rate": 9.200240196365408e-10,
      "loss": 0.4326,
      "step": 484460
    },
    {
      "epoch": 198.31,
      "grad_norm": 1.9687654972076416,
      "learning_rate": 9.155851160327389e-10,
      "loss": 0.4463,
      "step": 484470
    },
    {
      "epoch": 198.31,
      "grad_norm": 2.105384588241577,
      "learning_rate": 9.111569445904507e-10,
      "loss": 0.4359,
      "step": 484480
    },
    {
      "epoch": 198.32,
      "grad_norm": 1.8030235767364502,
      "learning_rate": 9.067395053302326e-10,
      "loss": 0.4293,
      "step": 484490
    },
    {
      "epoch": 198.32,
      "grad_norm": 2.209336757659912,
      "learning_rate": 9.023327982718605e-10,
      "loss": 0.4369,
      "step": 484500
    },
    {
      "epoch": 198.33,
      "grad_norm": 1.929114818572998,
      "learning_rate": 8.97936823436151e-10,
      "loss": 0.4548,
      "step": 484510
    },
    {
      "epoch": 198.33,
      "grad_norm": 2.176734209060669,
      "learning_rate": 8.935515808426198e-10,
      "loss": 0.4342,
      "step": 484520
    },
    {
      "epoch": 198.33,
      "grad_norm": 2.1049625873565674,
      "learning_rate": 8.891770705118234e-10,
      "loss": 0.4454,
      "step": 484530
    },
    {
      "epoch": 198.34,
      "grad_norm": 2.005261182785034,
      "learning_rate": 8.848132924635376e-10,
      "loss": 0.4438,
      "step": 484540
    },
    {
      "epoch": 198.34,
      "grad_norm": 2.757352113723755,
      "learning_rate": 8.804602467180587e-10,
      "loss": 0.4471,
      "step": 484550
    },
    {
      "epoch": 198.35,
      "grad_norm": 1.9567246437072754,
      "learning_rate": 8.761179332949023e-10,
      "loss": 0.4276,
      "step": 484560
    },
    {
      "epoch": 198.35,
      "grad_norm": 2.137430191040039,
      "learning_rate": 8.717863522141044e-10,
      "loss": 0.4394,
      "step": 484570
    },
    {
      "epoch": 198.35,
      "grad_norm": 2.3191370964050293,
      "learning_rate": 8.674655034957012e-10,
      "loss": 0.4489,
      "step": 484580
    },
    {
      "epoch": 198.36,
      "grad_norm": 2.064418077468872,
      "learning_rate": 8.631553871592082e-10,
      "loss": 0.4408,
      "step": 484590
    },
    {
      "epoch": 198.36,
      "grad_norm": 2.221503496170044,
      "learning_rate": 8.588560032246616e-10,
      "loss": 0.4631,
      "step": 484600
    },
    {
      "epoch": 198.37,
      "grad_norm": 2.1007559299468994,
      "learning_rate": 8.545673517113168e-10,
      "loss": 0.4371,
      "step": 484610
    },
    {
      "epoch": 198.37,
      "grad_norm": 2.081801652908325,
      "learning_rate": 8.5028943263947e-10,
      "loss": 0.441,
      "step": 484620
    },
    {
      "epoch": 198.37,
      "grad_norm": 2.1062726974487305,
      "learning_rate": 8.460222460281163e-10,
      "loss": 0.4543,
      "step": 484630
    },
    {
      "epoch": 198.38,
      "grad_norm": 1.8564260005950928,
      "learning_rate": 8.417657918970318e-10,
      "loss": 0.4339,
      "step": 484640
    },
    {
      "epoch": 198.38,
      "grad_norm": 2.064578056335449,
      "learning_rate": 8.375200702659921e-10,
      "loss": 0.4623,
      "step": 484650
    },
    {
      "epoch": 198.39,
      "grad_norm": 2.1907777786254883,
      "learning_rate": 8.332850811539928e-10,
      "loss": 0.4227,
      "step": 484660
    },
    {
      "epoch": 198.39,
      "grad_norm": 2.0596041679382324,
      "learning_rate": 8.290608245805491e-10,
      "loss": 0.4331,
      "step": 484670
    },
    {
      "epoch": 198.4,
      "grad_norm": 2.2070839405059814,
      "learning_rate": 8.248473005651769e-10,
      "loss": 0.4438,
      "step": 484680
    },
    {
      "epoch": 198.4,
      "grad_norm": 2.349792718887329,
      "learning_rate": 8.206445091271316e-10,
      "loss": 0.4497,
      "step": 484690
    },
    {
      "epoch": 198.4,
      "grad_norm": 1.996528148651123,
      "learning_rate": 8.164524502854083e-10,
      "loss": 0.4326,
      "step": 484700
    },
    {
      "epoch": 198.41,
      "grad_norm": 2.21614670753479,
      "learning_rate": 8.122711240595228e-10,
      "loss": 0.4379,
      "step": 484710
    },
    {
      "epoch": 198.41,
      "grad_norm": 2.091613531112671,
      "learning_rate": 8.081005304687305e-10,
      "loss": 0.4473,
      "step": 484720
    },
    {
      "epoch": 198.42,
      "grad_norm": 2.118759870529175,
      "learning_rate": 8.039406695317662e-10,
      "loss": 0.4519,
      "step": 484730
    },
    {
      "epoch": 198.42,
      "grad_norm": 2.1572909355163574,
      "learning_rate": 7.997915412676255e-10,
      "loss": 0.464,
      "step": 484740
    },
    {
      "epoch": 198.42,
      "grad_norm": 1.8446303606033325,
      "learning_rate": 7.956531456958237e-10,
      "loss": 0.4751,
      "step": 484750
    },
    {
      "epoch": 198.43,
      "grad_norm": 2.072227716445923,
      "learning_rate": 7.915254828348358e-10,
      "loss": 0.4349,
      "step": 484760
    },
    {
      "epoch": 198.43,
      "grad_norm": 2.201467275619507,
      "learning_rate": 7.874085527039172e-10,
      "loss": 0.4407,
      "step": 484770
    },
    {
      "epoch": 198.44,
      "grad_norm": 2.3700270652770996,
      "learning_rate": 7.833023553218027e-10,
      "loss": 0.4355,
      "step": 484780
    },
    {
      "epoch": 198.44,
      "grad_norm": 2.2199037075042725,
      "learning_rate": 7.792068907072277e-10,
      "loss": 0.4335,
      "step": 484790
    },
    {
      "epoch": 198.44,
      "grad_norm": 6.627198696136475,
      "learning_rate": 7.751221588789269e-10,
      "loss": 0.4232,
      "step": 484800
    },
    {
      "epoch": 198.45,
      "grad_norm": 2.390455484390259,
      "learning_rate": 7.710481598558956e-10,
      "loss": 0.4408,
      "step": 484810
    },
    {
      "epoch": 198.45,
      "grad_norm": 2.1755881309509277,
      "learning_rate": 7.669848936563486e-10,
      "loss": 0.4375,
      "step": 484820
    },
    {
      "epoch": 198.46,
      "grad_norm": 2.2386252880096436,
      "learning_rate": 7.629323602992808e-10,
      "loss": 0.4512,
      "step": 484830
    },
    {
      "epoch": 198.46,
      "grad_norm": 1.8117449283599854,
      "learning_rate": 7.588905598031672e-10,
      "loss": 0.4359,
      "step": 484840
    },
    {
      "epoch": 198.47,
      "grad_norm": 1.9058244228363037,
      "learning_rate": 7.548594921862224e-10,
      "loss": 0.4487,
      "step": 484850
    },
    {
      "epoch": 198.47,
      "grad_norm": 2.4215664863586426,
      "learning_rate": 7.508391574671813e-10,
      "loss": 0.4381,
      "step": 484860
    },
    {
      "epoch": 198.47,
      "grad_norm": 1.921356201171875,
      "learning_rate": 7.46829555664779e-10,
      "loss": 0.4411,
      "step": 484870
    },
    {
      "epoch": 198.48,
      "grad_norm": 1.8325440883636475,
      "learning_rate": 7.428306867967096e-10,
      "loss": 0.4608,
      "step": 484880
    },
    {
      "epoch": 198.48,
      "grad_norm": 2.1419260501861572,
      "learning_rate": 7.388425508817082e-10,
      "loss": 0.4525,
      "step": 484890
    },
    {
      "epoch": 198.49,
      "grad_norm": 2.1072068214416504,
      "learning_rate": 7.348651479379894e-10,
      "loss": 0.4409,
      "step": 484900
    },
    {
      "epoch": 198.49,
      "grad_norm": 1.991872787475586,
      "learning_rate": 7.308984779837677e-10,
      "loss": 0.4046,
      "step": 484910
    },
    {
      "epoch": 198.49,
      "grad_norm": 2.3926656246185303,
      "learning_rate": 7.269425410372578e-10,
      "loss": 0.4455,
      "step": 484920
    },
    {
      "epoch": 198.5,
      "grad_norm": 3.8419835567474365,
      "learning_rate": 7.229973371161539e-10,
      "loss": 0.4566,
      "step": 484930
    },
    {
      "epoch": 198.5,
      "grad_norm": 2.084595203399658,
      "learning_rate": 7.190628662391909e-10,
      "loss": 0.4445,
      "step": 484940
    },
    {
      "epoch": 198.51,
      "grad_norm": 2.0364303588867188,
      "learning_rate": 7.151391284240631e-10,
      "loss": 0.4343,
      "step": 484950
    },
    {
      "epoch": 198.51,
      "grad_norm": 1.946005940437317,
      "learning_rate": 7.112261236887247e-10,
      "loss": 0.4527,
      "step": 484960
    },
    {
      "epoch": 198.51,
      "grad_norm": 2.415827512741089,
      "learning_rate": 7.073238520511303e-10,
      "loss": 0.4479,
      "step": 484970
    },
    {
      "epoch": 198.52,
      "grad_norm": 3.3815648555755615,
      "learning_rate": 7.034323135292341e-10,
      "loss": 0.448,
      "step": 484980
    },
    {
      "epoch": 198.52,
      "grad_norm": 2.272582530975342,
      "learning_rate": 6.995515081407305e-10,
      "loss": 0.4541,
      "step": 484990
    },
    {
      "epoch": 198.53,
      "grad_norm": 1.9216029644012451,
      "learning_rate": 6.956814359035736e-10,
      "loss": 0.45,
      "step": 485000
    },
    {
      "epoch": 198.53,
      "grad_norm": 1.8481093645095825,
      "learning_rate": 6.918220968351977e-10,
      "loss": 0.4242,
      "step": 485010
    },
    {
      "epoch": 198.53,
      "grad_norm": 1.990736961364746,
      "learning_rate": 6.879734909535569e-10,
      "loss": 0.4544,
      "step": 485020
    },
    {
      "epoch": 198.54,
      "grad_norm": 2.248225688934326,
      "learning_rate": 6.841356182763457e-10,
      "loss": 0.4352,
      "step": 485030
    },
    {
      "epoch": 198.54,
      "grad_norm": 1.8724017143249512,
      "learning_rate": 6.803084788207375e-10,
      "loss": 0.4411,
      "step": 485040
    },
    {
      "epoch": 198.55,
      "grad_norm": 1.8996342420578003,
      "learning_rate": 6.764920726046868e-10,
      "loss": 0.4542,
      "step": 485050
    },
    {
      "epoch": 198.55,
      "grad_norm": 2.1034693717956543,
      "learning_rate": 6.726863996453676e-10,
      "loss": 0.4431,
      "step": 485060
    },
    {
      "epoch": 198.56,
      "grad_norm": 1.9797669649124146,
      "learning_rate": 6.688914599604737e-10,
      "loss": 0.4278,
      "step": 485070
    },
    {
      "epoch": 198.56,
      "grad_norm": 2.0169312953948975,
      "learning_rate": 6.651072535671792e-10,
      "loss": 0.4526,
      "step": 485080
    },
    {
      "epoch": 198.56,
      "grad_norm": 2.0757460594177246,
      "learning_rate": 6.613337804831781e-10,
      "loss": 0.4334,
      "step": 485090
    },
    {
      "epoch": 198.57,
      "grad_norm": 2.1552484035491943,
      "learning_rate": 6.575710407251238e-10,
      "loss": 0.4456,
      "step": 485100
    },
    {
      "epoch": 198.57,
      "grad_norm": 2.2392191886901855,
      "learning_rate": 6.538190343107104e-10,
      "loss": 0.4537,
      "step": 485110
    },
    {
      "epoch": 198.58,
      "grad_norm": 1.867903709411621,
      "learning_rate": 6.500777612571117e-10,
      "loss": 0.4614,
      "step": 485120
    },
    {
      "epoch": 198.58,
      "grad_norm": 1.8079255819320679,
      "learning_rate": 6.463472215812414e-10,
      "loss": 0.4398,
      "step": 485130
    },
    {
      "epoch": 198.58,
      "grad_norm": 3.2250709533691406,
      "learning_rate": 6.426274153005333e-10,
      "loss": 0.433,
      "step": 485140
    },
    {
      "epoch": 198.59,
      "grad_norm": 1.992783784866333,
      "learning_rate": 6.389183424316408e-10,
      "loss": 0.4424,
      "step": 485150
    },
    {
      "epoch": 198.59,
      "grad_norm": 1.8897746801376343,
      "learning_rate": 6.352200029917375e-10,
      "loss": 0.4591,
      "step": 485160
    },
    {
      "epoch": 198.6,
      "grad_norm": 2.0298407077789307,
      "learning_rate": 6.315323969979975e-10,
      "loss": 0.4513,
      "step": 485170
    },
    {
      "epoch": 198.6,
      "grad_norm": 1.946493148803711,
      "learning_rate": 6.278555244668137e-10,
      "loss": 0.4462,
      "step": 485180
    },
    {
      "epoch": 198.6,
      "grad_norm": 2.141009569168091,
      "learning_rate": 6.2418938541536e-10,
      "loss": 0.4346,
      "step": 485190
    },
    {
      "epoch": 198.61,
      "grad_norm": 1.8245677947998047,
      "learning_rate": 6.205339798605498e-10,
      "loss": 0.4385,
      "step": 485200
    },
    {
      "epoch": 198.61,
      "grad_norm": 1.9368531703948975,
      "learning_rate": 6.168893078187763e-10,
      "loss": 0.4384,
      "step": 485210
    },
    {
      "epoch": 198.62,
      "grad_norm": 1.7244857549667358,
      "learning_rate": 6.132553693069531e-10,
      "loss": 0.4345,
      "step": 485220
    },
    {
      "epoch": 198.62,
      "grad_norm": 2.0955450534820557,
      "learning_rate": 6.096321643414733e-10,
      "loss": 0.4506,
      "step": 485230
    },
    {
      "epoch": 198.62,
      "grad_norm": 1.7163546085357666,
      "learning_rate": 6.060196929395106e-10,
      "loss": 0.4581,
      "step": 485240
    },
    {
      "epoch": 198.63,
      "grad_norm": 1.9282176494598389,
      "learning_rate": 6.024179551169378e-10,
      "loss": 0.4408,
      "step": 485250
    },
    {
      "epoch": 198.63,
      "grad_norm": 1.733066201210022,
      "learning_rate": 5.988269508906684e-10,
      "loss": 0.4524,
      "step": 485260
    },
    {
      "epoch": 198.64,
      "grad_norm": 1.9644132852554321,
      "learning_rate": 5.952466802770957e-10,
      "loss": 0.44,
      "step": 485270
    },
    {
      "epoch": 198.64,
      "grad_norm": 1.9414031505584717,
      "learning_rate": 5.916771432923524e-10,
      "loss": 0.4517,
      "step": 485280
    },
    {
      "epoch": 198.65,
      "grad_norm": 2.057298421859741,
      "learning_rate": 5.881183399533522e-10,
      "loss": 0.4356,
      "step": 485290
    },
    {
      "epoch": 198.65,
      "grad_norm": 2.028224468231201,
      "learning_rate": 5.845702702757075e-10,
      "loss": 0.4554,
      "step": 485300
    },
    {
      "epoch": 198.65,
      "grad_norm": 1.9653990268707275,
      "learning_rate": 5.810329342760718e-10,
      "loss": 0.444,
      "step": 485310
    },
    {
      "epoch": 198.66,
      "grad_norm": 2.0719003677368164,
      "learning_rate": 5.775063319708381e-10,
      "loss": 0.4343,
      "step": 485320
    },
    {
      "epoch": 198.66,
      "grad_norm": 1.893890380859375,
      "learning_rate": 5.73990463375619e-10,
      "loss": 0.4424,
      "step": 485330
    },
    {
      "epoch": 198.67,
      "grad_norm": 1.8579593896865845,
      "learning_rate": 5.704853285070678e-10,
      "loss": 0.4445,
      "step": 485340
    },
    {
      "epoch": 198.67,
      "grad_norm": 1.8294771909713745,
      "learning_rate": 5.669909273807969e-10,
      "loss": 0.4433,
      "step": 485350
    },
    {
      "epoch": 198.67,
      "grad_norm": 1.9530612230300903,
      "learning_rate": 5.635072600129395e-10,
      "loss": 0.4222,
      "step": 485360
    },
    {
      "epoch": 198.68,
      "grad_norm": 2.1765480041503906,
      "learning_rate": 5.600343264198885e-10,
      "loss": 0.4665,
      "step": 485370
    },
    {
      "epoch": 198.68,
      "grad_norm": 2.4073596000671387,
      "learning_rate": 5.565721266167361e-10,
      "loss": 0.4774,
      "step": 485380
    },
    {
      "epoch": 198.69,
      "grad_norm": 4.273149013519287,
      "learning_rate": 5.531206606201356e-10,
      "loss": 0.4324,
      "step": 485390
    },
    {
      "epoch": 198.69,
      "grad_norm": 1.9490578174591064,
      "learning_rate": 5.496799284454394e-10,
      "loss": 0.4226,
      "step": 485400
    },
    {
      "epoch": 198.69,
      "grad_norm": 1.9339548349380493,
      "learning_rate": 5.462499301082599e-10,
      "loss": 0.4378,
      "step": 485410
    },
    {
      "epoch": 198.7,
      "grad_norm": 2.4735453128814697,
      "learning_rate": 5.428306656249902e-10,
      "loss": 0.4138,
      "step": 485420
    },
    {
      "epoch": 198.7,
      "grad_norm": 2.189457416534424,
      "learning_rate": 5.394221350104623e-10,
      "loss": 0.4396,
      "step": 485430
    },
    {
      "epoch": 198.71,
      "grad_norm": 2.515819787979126,
      "learning_rate": 5.360243382810693e-10,
      "loss": 0.4486,
      "step": 485440
    },
    {
      "epoch": 198.71,
      "grad_norm": 2.532461166381836,
      "learning_rate": 5.326372754516431e-10,
      "loss": 0.4232,
      "step": 485450
    },
    {
      "epoch": 198.71,
      "grad_norm": 1.7826443910598755,
      "learning_rate": 5.292609465383165e-10,
      "loss": 0.4384,
      "step": 485460
    },
    {
      "epoch": 198.72,
      "grad_norm": 1.8937057256698608,
      "learning_rate": 5.25895351556442e-10,
      "loss": 0.4579,
      "step": 485470
    },
    {
      "epoch": 198.72,
      "grad_norm": 1.8941705226898193,
      "learning_rate": 5.225404905208515e-10,
      "loss": 0.4338,
      "step": 485480
    },
    {
      "epoch": 198.73,
      "grad_norm": 2.475270986557007,
      "learning_rate": 5.191963634476777e-10,
      "loss": 0.4603,
      "step": 485490
    },
    {
      "epoch": 198.73,
      "grad_norm": 1.8819174766540527,
      "learning_rate": 5.158629703520128e-10,
      "loss": 0.4475,
      "step": 485500
    },
    {
      "epoch": 198.74,
      "grad_norm": 2.3708348274230957,
      "learning_rate": 5.125403112486888e-10,
      "loss": 0.4334,
      "step": 485510
    },
    {
      "epoch": 198.74,
      "grad_norm": 2.101828098297119,
      "learning_rate": 5.092283861535783e-10,
      "loss": 0.4262,
      "step": 485520
    },
    {
      "epoch": 198.74,
      "grad_norm": 1.83229660987854,
      "learning_rate": 5.05927195081253e-10,
      "loss": 0.4576,
      "step": 485530
    },
    {
      "epoch": 198.75,
      "grad_norm": 1.9029558897018433,
      "learning_rate": 5.026367380473254e-10,
      "loss": 0.4162,
      "step": 485540
    },
    {
      "epoch": 198.75,
      "grad_norm": 1.9060091972351074,
      "learning_rate": 4.993570150666274e-10,
      "loss": 0.4552,
      "step": 485550
    },
    {
      "epoch": 198.76,
      "grad_norm": 2.648379325866699,
      "learning_rate": 4.96088026154251e-10,
      "loss": 0.428,
      "step": 485560
    },
    {
      "epoch": 198.76,
      "grad_norm": 2.091053009033203,
      "learning_rate": 4.928297713252885e-10,
      "loss": 0.4413,
      "step": 485570
    },
    {
      "epoch": 198.76,
      "grad_norm": 1.9743179082870483,
      "learning_rate": 4.895822505943115e-10,
      "loss": 0.4247,
      "step": 485580
    },
    {
      "epoch": 198.77,
      "grad_norm": 2.378258228302002,
      "learning_rate": 4.863454639764119e-10,
      "loss": 0.4314,
      "step": 485590
    },
    {
      "epoch": 198.77,
      "grad_norm": 2.153113842010498,
      "learning_rate": 4.831194114866821e-10,
      "loss": 0.446,
      "step": 485600
    },
    {
      "epoch": 198.78,
      "grad_norm": 2.4088497161865234,
      "learning_rate": 4.799040931394334e-10,
      "loss": 0.4365,
      "step": 485610
    },
    {
      "epoch": 198.78,
      "grad_norm": 1.947129487991333,
      "learning_rate": 4.766995089494977e-10,
      "loss": 0.4368,
      "step": 485620
    },
    {
      "epoch": 198.78,
      "grad_norm": 2.060518980026245,
      "learning_rate": 4.735056589317068e-10,
      "loss": 0.4449,
      "step": 485630
    },
    {
      "epoch": 198.79,
      "grad_norm": 2.1377813816070557,
      "learning_rate": 4.703225431006326e-10,
      "loss": 0.4478,
      "step": 485640
    },
    {
      "epoch": 198.79,
      "grad_norm": 1.7467172145843506,
      "learning_rate": 4.671501614708466e-10,
      "loss": 0.4422,
      "step": 485650
    },
    {
      "epoch": 198.8,
      "grad_norm": 2.1262779235839844,
      "learning_rate": 4.639885140569205e-10,
      "loss": 0.4265,
      "step": 485660
    },
    {
      "epoch": 198.8,
      "grad_norm": 2.1724791526794434,
      "learning_rate": 4.6083760087342605e-10,
      "loss": 0.4525,
      "step": 485670
    },
    {
      "epoch": 198.8,
      "grad_norm": 2.263655424118042,
      "learning_rate": 4.576974219346746e-10,
      "loss": 0.439,
      "step": 485680
    },
    {
      "epoch": 198.81,
      "grad_norm": 2.1668944358825684,
      "learning_rate": 4.5456797725471753e-10,
      "loss": 0.4506,
      "step": 485690
    },
    {
      "epoch": 198.81,
      "grad_norm": 1.8545058965682983,
      "learning_rate": 4.514492668486469e-10,
      "loss": 0.4632,
      "step": 485700
    },
    {
      "epoch": 198.82,
      "grad_norm": 1.696295142173767,
      "learning_rate": 4.4834129072999354e-10,
      "loss": 0.4413,
      "step": 485710
    },
    {
      "epoch": 198.82,
      "grad_norm": 2.208881378173828,
      "learning_rate": 4.452440489135893e-10,
      "loss": 0.4471,
      "step": 485720
    },
    {
      "epoch": 198.83,
      "grad_norm": 2.178767442703247,
      "learning_rate": 4.421575414132253e-10,
      "loss": 0.438,
      "step": 485730
    },
    {
      "epoch": 198.83,
      "grad_norm": 1.9691524505615234,
      "learning_rate": 4.3908176824321295e-10,
      "loss": 0.4546,
      "step": 485740
    },
    {
      "epoch": 198.83,
      "grad_norm": 1.7489519119262695,
      "learning_rate": 4.3601672941760356e-10,
      "loss": 0.4404,
      "step": 485750
    },
    {
      "epoch": 198.84,
      "grad_norm": 1.9308465719223022,
      "learning_rate": 4.3296242495044833e-10,
      "loss": 0.4463,
      "step": 485760
    },
    {
      "epoch": 198.84,
      "grad_norm": 2.3413913249969482,
      "learning_rate": 4.2991885485579855e-10,
      "loss": 0.4384,
      "step": 485770
    },
    {
      "epoch": 198.85,
      "grad_norm": 1.82936429977417,
      "learning_rate": 4.2688601914770553e-10,
      "loss": 0.4508,
      "step": 485780
    },
    {
      "epoch": 198.85,
      "grad_norm": 1.888498306274414,
      "learning_rate": 4.2386391783943987e-10,
      "loss": 0.4515,
      "step": 485790
    },
    {
      "epoch": 198.85,
      "grad_norm": 2.0341408252716064,
      "learning_rate": 4.208525509455732e-10,
      "loss": 0.4533,
      "step": 485800
    },
    {
      "epoch": 198.86,
      "grad_norm": 1.8933926820755005,
      "learning_rate": 4.178519184796364e-10,
      "loss": 0.4225,
      "step": 485810
    },
    {
      "epoch": 198.86,
      "grad_norm": 1.9239414930343628,
      "learning_rate": 4.148620204551603e-10,
      "loss": 0.4486,
      "step": 485820
    },
    {
      "epoch": 198.87,
      "grad_norm": 1.9877384901046753,
      "learning_rate": 4.118828568861962e-10,
      "loss": 0.4519,
      "step": 485830
    },
    {
      "epoch": 198.87,
      "grad_norm": 1.9502733945846558,
      "learning_rate": 4.089144277862749e-10,
      "loss": 0.4424,
      "step": 485840
    },
    {
      "epoch": 198.87,
      "grad_norm": 1.7944426536560059,
      "learning_rate": 4.059567331689273e-10,
      "loss": 0.4342,
      "step": 485850
    },
    {
      "epoch": 198.88,
      "grad_norm": 2.131998300552368,
      "learning_rate": 4.0300977304768415e-10,
      "loss": 0.437,
      "step": 485860
    },
    {
      "epoch": 198.88,
      "grad_norm": 1.9821701049804688,
      "learning_rate": 4.000735474360764e-10,
      "loss": 0.4551,
      "step": 485870
    },
    {
      "epoch": 198.89,
      "grad_norm": 1.8996951580047607,
      "learning_rate": 3.971480563476348e-10,
      "loss": 0.4536,
      "step": 485880
    },
    {
      "epoch": 198.89,
      "grad_norm": 2.3237552642822266,
      "learning_rate": 3.9423329979563005e-10,
      "loss": 0.4446,
      "step": 485890
    },
    {
      "epoch": 198.89,
      "grad_norm": 1.8696016073226929,
      "learning_rate": 3.9132927779359294e-10,
      "loss": 0.4228,
      "step": 485900
    },
    {
      "epoch": 198.9,
      "grad_norm": 2.001507520675659,
      "learning_rate": 3.8843599035453393e-10,
      "loss": 0.4432,
      "step": 485910
    },
    {
      "epoch": 198.9,
      "grad_norm": 2.082460403442383,
      "learning_rate": 3.855534374919839e-10,
      "loss": 0.4524,
      "step": 485920
    },
    {
      "epoch": 198.91,
      "grad_norm": 2.311711072921753,
      "learning_rate": 3.8268161921895316e-10,
      "loss": 0.4306,
      "step": 485930
    },
    {
      "epoch": 198.91,
      "grad_norm": 1.9893438816070557,
      "learning_rate": 3.798205355487124e-10,
      "loss": 0.4465,
      "step": 485940
    },
    {
      "epoch": 198.92,
      "grad_norm": 1.8256210088729858,
      "learning_rate": 3.7697018649453234e-10,
      "loss": 0.459,
      "step": 485950
    },
    {
      "epoch": 198.92,
      "grad_norm": 2.113344669342041,
      "learning_rate": 3.7413057206916313e-10,
      "loss": 0.4401,
      "step": 485960
    },
    {
      "epoch": 198.92,
      "grad_norm": 2.005622148513794,
      "learning_rate": 3.713016922856152e-10,
      "loss": 0.4699,
      "step": 485970
    },
    {
      "epoch": 198.93,
      "grad_norm": 1.794519066810608,
      "learning_rate": 3.6848354715715916e-10,
      "loss": 0.4592,
      "step": 485980
    },
    {
      "epoch": 198.93,
      "grad_norm": 1.9768208265304565,
      "learning_rate": 3.6567613669654524e-10,
      "loss": 0.4323,
      "step": 485990
    },
    {
      "epoch": 198.94,
      "grad_norm": 1.9211113452911377,
      "learning_rate": 3.6287946091626343e-10,
      "loss": 0.4447,
      "step": 486000
    },
    {
      "epoch": 198.94,
      "grad_norm": 1.9172515869140625,
      "learning_rate": 3.600935198298446e-10,
      "loss": 0.4331,
      "step": 486010
    },
    {
      "epoch": 198.94,
      "grad_norm": 1.9880890846252441,
      "learning_rate": 3.573183134495186e-10,
      "loss": 0.4421,
      "step": 486020
    },
    {
      "epoch": 198.95,
      "grad_norm": 2.356177806854248,
      "learning_rate": 3.545538417880356e-10,
      "loss": 0.4589,
      "step": 486030
    },
    {
      "epoch": 198.95,
      "grad_norm": 2.033003091812134,
      "learning_rate": 3.51800104858406e-10,
      "loss": 0.4408,
      "step": 486040
    },
    {
      "epoch": 198.96,
      "grad_norm": 2.341278076171875,
      "learning_rate": 3.490571026728596e-10,
      "loss": 0.4405,
      "step": 486050
    },
    {
      "epoch": 198.96,
      "grad_norm": 1.7075799703598022,
      "learning_rate": 3.463248352441467e-10,
      "loss": 0.4415,
      "step": 486060
    },
    {
      "epoch": 198.96,
      "grad_norm": 2.5940463542938232,
      "learning_rate": 3.436033025847572e-10,
      "loss": 0.4476,
      "step": 486070
    },
    {
      "epoch": 198.97,
      "grad_norm": 1.909129023551941,
      "learning_rate": 3.4089250470692096e-10,
      "loss": 0.4127,
      "step": 486080
    },
    {
      "epoch": 198.97,
      "grad_norm": 2.2577366828918457,
      "learning_rate": 3.381924416236484e-10,
      "loss": 0.4353,
      "step": 486090
    },
    {
      "epoch": 198.98,
      "grad_norm": 1.8468468189239502,
      "learning_rate": 3.355031133466489e-10,
      "loss": 0.4421,
      "step": 486100
    },
    {
      "epoch": 198.98,
      "grad_norm": 1.9326318502426147,
      "learning_rate": 3.328245198886727e-10,
      "loss": 0.4496,
      "step": 486110
    },
    {
      "epoch": 198.98,
      "grad_norm": 2.153134346008301,
      "learning_rate": 3.301566612619495e-10,
      "loss": 0.4258,
      "step": 486120
    },
    {
      "epoch": 198.99,
      "grad_norm": 2.4977362155914307,
      "learning_rate": 3.2749953747844906e-10,
      "loss": 0.4479,
      "step": 486130
    },
    {
      "epoch": 198.99,
      "grad_norm": 1.9712201356887817,
      "learning_rate": 3.2485314855040107e-10,
      "loss": 0.4285,
      "step": 486140
    },
    {
      "epoch": 199.0,
      "grad_norm": 1.959460973739624,
      "learning_rate": 3.2221749449029555e-10,
      "loss": 0.4308,
      "step": 486150
    },
    {
      "epoch": 199.0,
      "eval_loss": 0.44599923491477966,
      "eval_runtime": 43.2956,
      "eval_samples_per_second": 79.662,
      "eval_steps_per_second": 9.978,
      "step": 486157
    },
    {
      "epoch": 199.0,
      "grad_norm": 2.7633798122406006,
      "learning_rate": 3.195925753098419e-10,
      "loss": 0.4549,
      "step": 486160
    },
    {
      "epoch": 199.01,
      "grad_norm": 2.1988985538482666,
      "learning_rate": 3.169783910210097e-10,
      "loss": 0.4466,
      "step": 486170
    },
    {
      "epoch": 199.01,
      "grad_norm": 2.116224527359009,
      "learning_rate": 3.1437494163602874e-10,
      "loss": 0.4562,
      "step": 486180
    },
    {
      "epoch": 199.01,
      "grad_norm": 2.1450698375701904,
      "learning_rate": 3.1178222716686865e-10,
      "loss": 0.4391,
      "step": 486190
    },
    {
      "epoch": 199.02,
      "grad_norm": 2.046663761138916,
      "learning_rate": 3.092002476252388e-10,
      "loss": 0.4488,
      "step": 486200
    },
    {
      "epoch": 199.02,
      "grad_norm": 1.7371224164962769,
      "learning_rate": 3.066290030228485e-10,
      "loss": 0.4406,
      "step": 486210
    },
    {
      "epoch": 199.03,
      "grad_norm": 2.501904249191284,
      "learning_rate": 3.040684933714073e-10,
      "loss": 0.4482,
      "step": 486220
    },
    {
      "epoch": 199.03,
      "grad_norm": 2.1287834644317627,
      "learning_rate": 3.0151871868314485e-10,
      "loss": 0.4641,
      "step": 486230
    },
    {
      "epoch": 199.03,
      "grad_norm": 1.8376028537750244,
      "learning_rate": 2.9897967896951034e-10,
      "loss": 0.4457,
      "step": 486240
    },
    {
      "epoch": 199.04,
      "grad_norm": 1.9984815120697021,
      "learning_rate": 2.9645137424195304e-10,
      "loss": 0.4416,
      "step": 486250
    },
    {
      "epoch": 199.04,
      "grad_norm": 2.111064910888672,
      "learning_rate": 2.9393380451218226e-10,
      "loss": 0.4459,
      "step": 486260
    },
    {
      "epoch": 199.05,
      "grad_norm": 1.9782867431640625,
      "learning_rate": 2.9142696979164715e-10,
      "loss": 0.4569,
      "step": 486270
    },
    {
      "epoch": 199.05,
      "grad_norm": 1.8582261800765991,
      "learning_rate": 2.8893087009179694e-10,
      "loss": 0.4432,
      "step": 486280
    },
    {
      "epoch": 199.05,
      "grad_norm": 1.8440662622451782,
      "learning_rate": 2.86445505424341e-10,
      "loss": 0.4317,
      "step": 486290
    },
    {
      "epoch": 199.06,
      "grad_norm": 1.9293911457061768,
      "learning_rate": 2.839708758004683e-10,
      "loss": 0.4343,
      "step": 486300
    },
    {
      "epoch": 199.06,
      "grad_norm": 2.381657600402832,
      "learning_rate": 2.8150698123162797e-10,
      "loss": 0.4399,
      "step": 486310
    },
    {
      "epoch": 199.07,
      "grad_norm": 2.033802032470703,
      "learning_rate": 2.7905382172900904e-10,
      "loss": 0.4403,
      "step": 486320
    },
    {
      "epoch": 199.07,
      "grad_norm": 2.3556530475616455,
      "learning_rate": 2.7661139730380047e-10,
      "loss": 0.4463,
      "step": 486330
    },
    {
      "epoch": 199.07,
      "grad_norm": 2.270521640777588,
      "learning_rate": 2.7417970796719116e-10,
      "loss": 0.4497,
      "step": 486340
    },
    {
      "epoch": 199.08,
      "grad_norm": 1.7457910776138306,
      "learning_rate": 2.717587537306304e-10,
      "loss": 0.4195,
      "step": 486350
    },
    {
      "epoch": 199.08,
      "grad_norm": 2.1754696369171143,
      "learning_rate": 2.693485346047866e-10,
      "loss": 0.437,
      "step": 486360
    },
    {
      "epoch": 199.09,
      "grad_norm": 2.009281635284424,
      "learning_rate": 2.6694905060084885e-10,
      "loss": 0.4596,
      "step": 486370
    },
    {
      "epoch": 199.09,
      "grad_norm": 1.9031611680984497,
      "learning_rate": 2.645603017297458e-10,
      "loss": 0.4563,
      "step": 486380
    },
    {
      "epoch": 199.1,
      "grad_norm": 2.005253553390503,
      "learning_rate": 2.621822880026665e-10,
      "loss": 0.4393,
      "step": 486390
    },
    {
      "epoch": 199.1,
      "grad_norm": 2.3929226398468018,
      "learning_rate": 2.598150094302795e-10,
      "loss": 0.4552,
      "step": 486400
    },
    {
      "epoch": 199.1,
      "grad_norm": 1.9052104949951172,
      "learning_rate": 2.574584660235135e-10,
      "loss": 0.4358,
      "step": 486410
    },
    {
      "epoch": 199.11,
      "grad_norm": 2.326777696609497,
      "learning_rate": 2.5511265779329727e-10,
      "loss": 0.4361,
      "step": 486420
    },
    {
      "epoch": 199.11,
      "grad_norm": 2.1932995319366455,
      "learning_rate": 2.5277758475003923e-10,
      "loss": 0.4424,
      "step": 486430
    },
    {
      "epoch": 199.12,
      "grad_norm": 2.038655996322632,
      "learning_rate": 2.50453246904668e-10,
      "loss": 0.4588,
      "step": 486440
    },
    {
      "epoch": 199.12,
      "grad_norm": 2.1161954402923584,
      "learning_rate": 2.48139644267592e-10,
      "loss": 0.4378,
      "step": 486450
    },
    {
      "epoch": 199.12,
      "grad_norm": 1.8552695512771606,
      "learning_rate": 2.4583677684974005e-10,
      "loss": 0.4322,
      "step": 486460
    },
    {
      "epoch": 199.13,
      "grad_norm": 2.010681390762329,
      "learning_rate": 2.4354464466178056e-10,
      "loss": 0.4491,
      "step": 486470
    },
    {
      "epoch": 199.13,
      "grad_norm": 1.983208179473877,
      "learning_rate": 2.412632477136016e-10,
      "loss": 0.4318,
      "step": 486480
    },
    {
      "epoch": 199.14,
      "grad_norm": 2.102219581604004,
      "learning_rate": 2.389925860161318e-10,
      "loss": 0.4464,
      "step": 486490
    },
    {
      "epoch": 199.14,
      "grad_norm": 2.089486598968506,
      "learning_rate": 2.3673265957977956e-10,
      "loss": 0.4454,
      "step": 486500
    },
    {
      "epoch": 199.14,
      "grad_norm": 2.16579270362854,
      "learning_rate": 2.3448346841443285e-10,
      "loss": 0.4365,
      "step": 486510
    },
    {
      "epoch": 199.15,
      "grad_norm": 2.128183603286743,
      "learning_rate": 2.3224501253102032e-10,
      "loss": 0.436,
      "step": 486520
    },
    {
      "epoch": 199.15,
      "grad_norm": 1.9849278926849365,
      "learning_rate": 2.3001729193942997e-10,
      "loss": 0.4322,
      "step": 486530
    },
    {
      "epoch": 199.16,
      "grad_norm": 2.0434279441833496,
      "learning_rate": 2.2780030665007012e-10,
      "loss": 0.4425,
      "step": 486540
    },
    {
      "epoch": 199.16,
      "grad_norm": 1.9284536838531494,
      "learning_rate": 2.255940566728287e-10,
      "loss": 0.4635,
      "step": 486550
    },
    {
      "epoch": 199.16,
      "grad_norm": 2.005556344985962,
      "learning_rate": 2.2339854201785384e-10,
      "loss": 0.4336,
      "step": 486560
    },
    {
      "epoch": 199.17,
      "grad_norm": 1.6945886611938477,
      "learning_rate": 2.2121376269529366e-10,
      "loss": 0.4338,
      "step": 486570
    },
    {
      "epoch": 199.17,
      "grad_norm": 2.095383882522583,
      "learning_rate": 2.1903971871529632e-10,
      "loss": 0.4651,
      "step": 486580
    },
    {
      "epoch": 199.18,
      "grad_norm": 1.822257399559021,
      "learning_rate": 2.1687641008774974e-10,
      "loss": 0.4553,
      "step": 486590
    },
    {
      "epoch": 199.18,
      "grad_norm": 1.9762521982192993,
      "learning_rate": 2.1472383682228158e-10,
      "loss": 0.4435,
      "step": 486600
    },
    {
      "epoch": 199.19,
      "grad_norm": 2.056405544281006,
      "learning_rate": 2.1258199892930025e-10,
      "loss": 0.4478,
      "step": 486610
    },
    {
      "epoch": 199.19,
      "grad_norm": 1.9059407711029053,
      "learning_rate": 2.10450896417913e-10,
      "loss": 0.4381,
      "step": 486620
    },
    {
      "epoch": 199.19,
      "grad_norm": 2.0242292881011963,
      "learning_rate": 2.0833052929852822e-10,
      "loss": 0.424,
      "step": 486630
    },
    {
      "epoch": 199.2,
      "grad_norm": 2.0043609142303467,
      "learning_rate": 2.0622089758051337e-10,
      "loss": 0.4528,
      "step": 486640
    },
    {
      "epoch": 199.2,
      "grad_norm": 2.2746388912200928,
      "learning_rate": 2.041220012734962e-10,
      "loss": 0.4337,
      "step": 486650
    },
    {
      "epoch": 199.21,
      "grad_norm": 1.934775948524475,
      "learning_rate": 2.0203384038736462e-10,
      "loss": 0.4402,
      "step": 486660
    },
    {
      "epoch": 199.21,
      "grad_norm": 1.8275845050811768,
      "learning_rate": 1.999564149314861e-10,
      "loss": 0.4449,
      "step": 486670
    },
    {
      "epoch": 199.21,
      "grad_norm": 1.9833667278289795,
      "learning_rate": 1.9788972491522822e-10,
      "loss": 0.4379,
      "step": 486680
    },
    {
      "epoch": 199.22,
      "grad_norm": 1.9278768301010132,
      "learning_rate": 1.9583377034847886e-10,
      "loss": 0.4598,
      "step": 486690
    },
    {
      "epoch": 199.22,
      "grad_norm": 2.2280449867248535,
      "learning_rate": 1.9378855124034532e-10,
      "loss": 0.4345,
      "step": 486700
    },
    {
      "epoch": 199.23,
      "grad_norm": 1.9547333717346191,
      "learning_rate": 1.917540676001951e-10,
      "loss": 0.4559,
      "step": 486710
    },
    {
      "epoch": 199.23,
      "grad_norm": 1.6457289457321167,
      "learning_rate": 1.8973031943739572e-10,
      "loss": 0.4867,
      "step": 486720
    },
    {
      "epoch": 199.23,
      "grad_norm": 2.6464710235595703,
      "learning_rate": 1.8771730676131468e-10,
      "loss": 0.4467,
      "step": 486730
    },
    {
      "epoch": 199.24,
      "grad_norm": 2.212686061859131,
      "learning_rate": 1.857150295813195e-10,
      "loss": 0.4332,
      "step": 486740
    },
    {
      "epoch": 199.24,
      "grad_norm": 2.3643579483032227,
      "learning_rate": 1.8372348790599706e-10,
      "loss": 0.4456,
      "step": 486750
    },
    {
      "epoch": 199.25,
      "grad_norm": 1.887522578239441,
      "learning_rate": 1.8174268174523527e-10,
      "loss": 0.4334,
      "step": 486760
    },
    {
      "epoch": 199.25,
      "grad_norm": 2.2515275478363037,
      "learning_rate": 1.797726111073608e-10,
      "loss": 0.4385,
      "step": 486770
    },
    {
      "epoch": 199.26,
      "grad_norm": 2.181877613067627,
      "learning_rate": 1.7781327600200137e-10,
      "loss": 0.4296,
      "step": 486780
    },
    {
      "epoch": 199.26,
      "grad_norm": 1.91645348072052,
      "learning_rate": 1.758646764377439e-10,
      "loss": 0.4461,
      "step": 486790
    },
    {
      "epoch": 199.26,
      "grad_norm": 1.977379322052002,
      "learning_rate": 1.739268124236956e-10,
      "loss": 0.4389,
      "step": 486800
    },
    {
      "epoch": 199.27,
      "grad_norm": 2.224874973297119,
      "learning_rate": 1.7199968396844345e-10,
      "loss": 0.4429,
      "step": 486810
    },
    {
      "epoch": 199.27,
      "grad_norm": 1.9672999382019043,
      "learning_rate": 1.7008329108135488e-10,
      "loss": 0.4414,
      "step": 486820
    },
    {
      "epoch": 199.28,
      "grad_norm": 1.9140933752059937,
      "learning_rate": 1.681776337710168e-10,
      "loss": 0.4262,
      "step": 486830
    },
    {
      "epoch": 199.28,
      "grad_norm": 2.8447372913360596,
      "learning_rate": 1.6628271204575592e-10,
      "loss": 0.4505,
      "step": 486840
    },
    {
      "epoch": 199.28,
      "grad_norm": 2.1767423152923584,
      "learning_rate": 1.6439852591467948e-10,
      "loss": 0.4458,
      "step": 486850
    },
    {
      "epoch": 199.29,
      "grad_norm": 2.319446086883545,
      "learning_rate": 1.6252507538637437e-10,
      "loss": 0.4457,
      "step": 486860
    },
    {
      "epoch": 199.29,
      "grad_norm": 2.4376745223999023,
      "learning_rate": 1.6066236046916731e-10,
      "loss": 0.4308,
      "step": 486870
    },
    {
      "epoch": 199.3,
      "grad_norm": 1.9431259632110596,
      "learning_rate": 1.5881038117190535e-10,
      "loss": 0.4298,
      "step": 486880
    },
    {
      "epoch": 199.3,
      "grad_norm": 2.1916565895080566,
      "learning_rate": 1.5696913750291515e-10,
      "loss": 0.4405,
      "step": 486890
    },
    {
      "epoch": 199.3,
      "grad_norm": 2.2672924995422363,
      "learning_rate": 1.5513862947052341e-10,
      "loss": 0.4265,
      "step": 486900
    },
    {
      "epoch": 199.31,
      "grad_norm": 2.370957851409912,
      "learning_rate": 1.5331885708331702e-10,
      "loss": 0.4398,
      "step": 486910
    },
    {
      "epoch": 199.31,
      "grad_norm": 2.045487403869629,
      "learning_rate": 1.5150982034962262e-10,
      "loss": 0.4517,
      "step": 486920
    },
    {
      "epoch": 199.32,
      "grad_norm": 2.1683645248413086,
      "learning_rate": 1.4971151927776689e-10,
      "loss": 0.4501,
      "step": 486930
    },
    {
      "epoch": 199.32,
      "grad_norm": 2.0397145748138428,
      "learning_rate": 1.4792395387555612e-10,
      "loss": 0.4363,
      "step": 486940
    },
    {
      "epoch": 199.32,
      "grad_norm": 1.9860862493515015,
      "learning_rate": 1.4614712415183735e-10,
      "loss": 0.4515,
      "step": 486950
    },
    {
      "epoch": 199.33,
      "grad_norm": 2.0494580268859863,
      "learning_rate": 1.4438103011415667e-10,
      "loss": 0.4359,
      "step": 486960
    },
    {
      "epoch": 199.33,
      "grad_norm": 2.66611385345459,
      "learning_rate": 1.4262567177110094e-10,
      "loss": 0.4452,
      "step": 486970
    },
    {
      "epoch": 199.34,
      "grad_norm": 2.0648083686828613,
      "learning_rate": 1.408810491302162e-10,
      "loss": 0.4454,
      "step": 486980
    },
    {
      "epoch": 199.34,
      "grad_norm": 2.28576397895813,
      "learning_rate": 1.3914716219982914e-10,
      "loss": 0.4325,
      "step": 486990
    },
    {
      "epoch": 199.35,
      "grad_norm": 1.9445457458496094,
      "learning_rate": 1.37424010987746e-10,
      "loss": 0.4273,
      "step": 487000
    },
    {
      "epoch": 199.35,
      "grad_norm": 2.1545333862304688,
      "learning_rate": 1.3571159550203324e-10,
      "loss": 0.4291,
      "step": 487010
    },
    {
      "epoch": 199.35,
      "grad_norm": 2.194664716720581,
      "learning_rate": 1.3400991575049716e-10,
      "loss": 0.4469,
      "step": 487020
    },
    {
      "epoch": 199.36,
      "grad_norm": 2.014575719833374,
      "learning_rate": 1.3231897174068375e-10,
      "loss": 0.4589,
      "step": 487030
    },
    {
      "epoch": 199.36,
      "grad_norm": 2.0211358070373535,
      "learning_rate": 1.3063876348065952e-10,
      "loss": 0.465,
      "step": 487040
    },
    {
      "epoch": 199.37,
      "grad_norm": 1.8991422653198242,
      "learning_rate": 1.2896929097797048e-10,
      "loss": 0.4394,
      "step": 487050
    },
    {
      "epoch": 199.37,
      "grad_norm": 2.1272385120391846,
      "learning_rate": 1.273105542401627e-10,
      "loss": 0.4604,
      "step": 487060
    },
    {
      "epoch": 199.37,
      "grad_norm": 1.8771896362304688,
      "learning_rate": 1.2566255327504244e-10,
      "loss": 0.4447,
      "step": 487070
    },
    {
      "epoch": 199.38,
      "grad_norm": 2.1188690662384033,
      "learning_rate": 1.240252880898955e-10,
      "loss": 0.4383,
      "step": 487080
    },
    {
      "epoch": 199.38,
      "grad_norm": 2.331423759460449,
      "learning_rate": 1.2239875869252818e-10,
      "loss": 0.4735,
      "step": 487090
    },
    {
      "epoch": 199.39,
      "grad_norm": 1.9378727674484253,
      "learning_rate": 1.207829650902263e-10,
      "loss": 0.4473,
      "step": 487100
    },
    {
      "epoch": 199.39,
      "grad_norm": 1.7757856845855713,
      "learning_rate": 1.191779072902757e-10,
      "loss": 0.4442,
      "step": 487110
    },
    {
      "epoch": 199.39,
      "grad_norm": 1.9073855876922607,
      "learning_rate": 1.1758358530048265e-10,
      "loss": 0.4419,
      "step": 487120
    },
    {
      "epoch": 199.4,
      "grad_norm": 1.5683854818344116,
      "learning_rate": 1.1599999912761255e-10,
      "loss": 0.4514,
      "step": 487130
    },
    {
      "epoch": 199.4,
      "grad_norm": 2.3102452754974365,
      "learning_rate": 1.1442714877921145e-10,
      "loss": 0.4401,
      "step": 487140
    },
    {
      "epoch": 199.41,
      "grad_norm": 1.479303002357483,
      "learning_rate": 1.1286503426230498e-10,
      "loss": 0.4411,
      "step": 487150
    },
    {
      "epoch": 199.41,
      "grad_norm": 2.119997262954712,
      "learning_rate": 1.1131365558417899e-10,
      "loss": 0.4438,
      "step": 487160
    },
    {
      "epoch": 199.41,
      "grad_norm": 2.652599573135376,
      "learning_rate": 1.0977301275185909e-10,
      "loss": 0.4358,
      "step": 487170
    },
    {
      "epoch": 199.42,
      "grad_norm": 2.0619871616363525,
      "learning_rate": 1.0824310577263115e-10,
      "loss": 0.4526,
      "step": 487180
    },
    {
      "epoch": 199.42,
      "grad_norm": 1.9239963293075562,
      "learning_rate": 1.0672393465326057e-10,
      "loss": 0.4329,
      "step": 487190
    },
    {
      "epoch": 199.43,
      "grad_norm": 2.3517253398895264,
      "learning_rate": 1.0521549940077298e-10,
      "loss": 0.4485,
      "step": 487200
    },
    {
      "epoch": 199.43,
      "grad_norm": 2.1709487438201904,
      "learning_rate": 1.0371780002193381e-10,
      "loss": 0.4355,
      "step": 487210
    },
    {
      "epoch": 199.44,
      "grad_norm": 2.070063829421997,
      "learning_rate": 1.0223083652402889e-10,
      "loss": 0.4526,
      "step": 487220
    },
    {
      "epoch": 199.44,
      "grad_norm": 2.1088573932647705,
      "learning_rate": 1.0075460891330323e-10,
      "loss": 0.4323,
      "step": 487230
    },
    {
      "epoch": 199.44,
      "grad_norm": 1.9091531038284302,
      "learning_rate": 9.928911719704266e-11,
      "loss": 0.4353,
      "step": 487240
    },
    {
      "epoch": 199.45,
      "grad_norm": 1.8687238693237305,
      "learning_rate": 9.783436138175241e-11,
      "loss": 0.4382,
      "step": 487250
    },
    {
      "epoch": 199.45,
      "grad_norm": 2.4918642044067383,
      "learning_rate": 9.639034147393768e-11,
      "loss": 0.4419,
      "step": 487260
    },
    {
      "epoch": 199.46,
      "grad_norm": 2.0385477542877197,
      "learning_rate": 9.49570574803639e-11,
      "loss": 0.4382,
      "step": 487270
    },
    {
      "epoch": 199.46,
      "grad_norm": 2.1032156944274902,
      "learning_rate": 9.353450940753626e-11,
      "loss": 0.4371,
      "step": 487280
    },
    {
      "epoch": 199.46,
      "grad_norm": 2.588564157485962,
      "learning_rate": 9.212269726222021e-11,
      "loss": 0.4521,
      "step": 487290
    },
    {
      "epoch": 199.47,
      "grad_norm": 1.9727600812911987,
      "learning_rate": 9.072162105040054e-11,
      "loss": 0.4233,
      "step": 487300
    },
    {
      "epoch": 199.47,
      "grad_norm": 2.914921760559082,
      "learning_rate": 8.933128077884266e-11,
      "loss": 0.4272,
      "step": 487310
    },
    {
      "epoch": 199.48,
      "grad_norm": 2.2305450439453125,
      "learning_rate": 8.79516764540518e-11,
      "loss": 0.454,
      "step": 487320
    },
    {
      "epoch": 199.48,
      "grad_norm": 1.9531461000442505,
      "learning_rate": 8.658280808201274e-11,
      "loss": 0.4404,
      "step": 487330
    },
    {
      "epoch": 199.48,
      "grad_norm": 1.9710739850997925,
      "learning_rate": 8.522467566897048e-11,
      "loss": 0.4201,
      "step": 487340
    },
    {
      "epoch": 199.49,
      "grad_norm": 1.8712294101715088,
      "learning_rate": 8.387727922169046e-11,
      "loss": 0.4539,
      "step": 487350
    },
    {
      "epoch": 199.49,
      "grad_norm": 2.622986316680908,
      "learning_rate": 8.254061874563706e-11,
      "loss": 0.4365,
      "step": 487360
    },
    {
      "epoch": 199.5,
      "grad_norm": 2.2041335105895996,
      "learning_rate": 8.121469424731547e-11,
      "loss": 0.4431,
      "step": 487370
    },
    {
      "epoch": 199.5,
      "grad_norm": 2.1781556606292725,
      "learning_rate": 7.989950573271051e-11,
      "loss": 0.4626,
      "step": 487380
    },
    {
      "epoch": 199.5,
      "grad_norm": 1.8400336503982544,
      "learning_rate": 7.859505320806716e-11,
      "loss": 0.441,
      "step": 487390
    },
    {
      "epoch": 199.51,
      "grad_norm": 2.1399333477020264,
      "learning_rate": 7.730133667884983e-11,
      "loss": 0.4686,
      "step": 487400
    },
    {
      "epoch": 199.51,
      "grad_norm": 1.7683095932006836,
      "learning_rate": 7.60183561515637e-11,
      "loss": 0.4672,
      "step": 487410
    },
    {
      "epoch": 199.52,
      "grad_norm": 2.024677276611328,
      "learning_rate": 7.474611163167319e-11,
      "loss": 0.4549,
      "step": 487420
    },
    {
      "epoch": 199.52,
      "grad_norm": 1.782149076461792,
      "learning_rate": 7.348460312516305e-11,
      "loss": 0.4373,
      "step": 487430
    },
    {
      "epoch": 199.53,
      "grad_norm": 1.7535377740859985,
      "learning_rate": 7.22338306380181e-11,
      "loss": 0.4652,
      "step": 487440
    },
    {
      "epoch": 199.53,
      "grad_norm": 2.2281126976013184,
      "learning_rate": 7.099379417544251e-11,
      "loss": 0.4419,
      "step": 487450
    },
    {
      "epoch": 199.53,
      "grad_norm": 2.3414242267608643,
      "learning_rate": 6.976449374368128e-11,
      "loss": 0.4483,
      "step": 487460
    },
    {
      "epoch": 199.54,
      "grad_norm": 2.056467294692993,
      "learning_rate": 6.854592934793859e-11,
      "loss": 0.453,
      "step": 487470
    },
    {
      "epoch": 199.54,
      "grad_norm": 2.251821994781494,
      "learning_rate": 6.733810099419923e-11,
      "loss": 0.4403,
      "step": 487480
    },
    {
      "epoch": 199.55,
      "grad_norm": 2.2813878059387207,
      "learning_rate": 6.614100868740715e-11,
      "loss": 0.4456,
      "step": 487490
    },
    {
      "epoch": 199.55,
      "grad_norm": 1.95235013961792,
      "learning_rate": 6.495465243380738e-11,
      "loss": 0.4315,
      "step": 487500
    },
    {
      "epoch": 199.55,
      "grad_norm": 2.0600574016571045,
      "learning_rate": 6.377903223808364e-11,
      "loss": 0.4361,
      "step": 487510
    },
    {
      "epoch": 199.56,
      "grad_norm": 1.9098412990570068,
      "learning_rate": 6.261414810622075e-11,
      "loss": 0.4402,
      "step": 487520
    },
    {
      "epoch": 199.56,
      "grad_norm": 1.669885516166687,
      "learning_rate": 6.146000004316267e-11,
      "loss": 0.4275,
      "step": 487530
    },
    {
      "epoch": 199.57,
      "grad_norm": 2.120866060256958,
      "learning_rate": 6.031658805437379e-11,
      "loss": 0.4373,
      "step": 487540
    },
    {
      "epoch": 199.57,
      "grad_norm": 2.078727960586548,
      "learning_rate": 5.918391214505825e-11,
      "loss": 0.432,
      "step": 487550
    },
    {
      "epoch": 199.57,
      "grad_norm": 2.1089632511138916,
      "learning_rate": 5.806197232042023e-11,
      "loss": 0.4406,
      "step": 487560
    },
    {
      "epoch": 199.58,
      "grad_norm": 1.806905746459961,
      "learning_rate": 5.695076858566391e-11,
      "loss": 0.4467,
      "step": 487570
    },
    {
      "epoch": 199.58,
      "grad_norm": 2.4942588806152344,
      "learning_rate": 5.5850300945733254e-11,
      "loss": 0.4432,
      "step": 487580
    },
    {
      "epoch": 199.59,
      "grad_norm": 2.0817270278930664,
      "learning_rate": 5.476056940583243e-11,
      "loss": 0.4493,
      "step": 487590
    },
    {
      "epoch": 199.59,
      "grad_norm": 2.280264139175415,
      "learning_rate": 5.368157397064518e-11,
      "loss": 0.4781,
      "step": 487600
    },
    {
      "epoch": 199.59,
      "grad_norm": 1.8081214427947998,
      "learning_rate": 5.2613314645375684e-11,
      "loss": 0.4435,
      "step": 487610
    },
    {
      "epoch": 199.6,
      "grad_norm": 2.3714444637298584,
      "learning_rate": 5.1555791434967905e-11,
      "loss": 0.4256,
      "step": 487620
    },
    {
      "epoch": 199.6,
      "grad_norm": 2.1137590408325195,
      "learning_rate": 5.05090043441056e-11,
      "loss": 0.4451,
      "step": 487630
    },
    {
      "epoch": 199.61,
      "grad_norm": 2.141472101211548,
      "learning_rate": 4.9472953377732715e-11,
      "loss": 0.457,
      "step": 487640
    },
    {
      "epoch": 199.61,
      "grad_norm": 1.9761319160461426,
      "learning_rate": 4.8447638540533015e-11,
      "loss": 0.4423,
      "step": 487650
    },
    {
      "epoch": 199.62,
      "grad_norm": 1.8186960220336914,
      "learning_rate": 4.743305983693005e-11,
      "loss": 0.4234,
      "step": 487660
    },
    {
      "epoch": 199.62,
      "grad_norm": 2.048097610473633,
      "learning_rate": 4.642921727212798e-11,
      "loss": 0.4372,
      "step": 487670
    },
    {
      "epoch": 199.62,
      "grad_norm": 1.9873778820037842,
      "learning_rate": 4.543611085029014e-11,
      "loss": 0.4355,
      "step": 487680
    },
    {
      "epoch": 199.63,
      "grad_norm": 2.111924648284912,
      "learning_rate": 4.4453740575840083e-11,
      "loss": 0.4568,
      "step": 487690
    },
    {
      "epoch": 199.63,
      "grad_norm": 1.8537074327468872,
      "learning_rate": 4.348210645398198e-11,
      "loss": 0.4404,
      "step": 487700
    },
    {
      "epoch": 199.64,
      "grad_norm": 2.735666513442993,
      "learning_rate": 4.2521208488358747e-11,
      "loss": 0.4427,
      "step": 487710
    },
    {
      "epoch": 199.64,
      "grad_norm": 1.939996600151062,
      "learning_rate": 4.157104668391434e-11,
      "loss": 0.4507,
      "step": 487720
    },
    {
      "epoch": 199.64,
      "grad_norm": 2.145467519760132,
      "learning_rate": 4.06316210448121e-11,
      "loss": 0.4433,
      "step": 487730
    },
    {
      "epoch": 199.65,
      "grad_norm": 2.197192430496216,
      "learning_rate": 3.970293157521537e-11,
      "loss": 0.4431,
      "step": 487740
    },
    {
      "epoch": 199.65,
      "grad_norm": 2.4376707077026367,
      "learning_rate": 3.8784978279547684e-11,
      "loss": 0.453,
      "step": 487750
    },
    {
      "epoch": 199.66,
      "grad_norm": 2.3086659908294678,
      "learning_rate": 3.7877761161972384e-11,
      "loss": 0.4406,
      "step": 487760
    },
    {
      "epoch": 199.66,
      "grad_norm": 1.9351012706756592,
      "learning_rate": 3.698128022691302e-11,
      "loss": 0.4246,
      "step": 487770
    },
    {
      "epoch": 199.66,
      "grad_norm": 6.485424041748047,
      "learning_rate": 3.60955354780125e-11,
      "loss": 0.4432,
      "step": 487780
    },
    {
      "epoch": 199.67,
      "grad_norm": 2.0431058406829834,
      "learning_rate": 3.522052691943417e-11,
      "loss": 0.4409,
      "step": 487790
    },
    {
      "epoch": 199.67,
      "grad_norm": 2.4851088523864746,
      "learning_rate": 3.4356254555341353e-11,
      "loss": 0.4488,
      "step": 487800
    },
    {
      "epoch": 199.68,
      "grad_norm": 2.1066248416900635,
      "learning_rate": 3.3502718389897404e-11,
      "loss": 0.4401,
      "step": 487810
    },
    {
      "epoch": 199.68,
      "grad_norm": 2.058173179626465,
      "learning_rate": 3.265991842648502e-11,
      "loss": 0.4463,
      "step": 487820
    },
    {
      "epoch": 199.68,
      "grad_norm": 2.3089511394500732,
      "learning_rate": 3.1827854669267546e-11,
      "loss": 0.4539,
      "step": 487830
    },
    {
      "epoch": 199.69,
      "grad_norm": 1.8623311519622803,
      "learning_rate": 3.1006527122148103e-11,
      "loss": 0.4508,
      "step": 487840
    },
    {
      "epoch": 199.69,
      "grad_norm": 2.449361562728882,
      "learning_rate": 3.019593578876961e-11,
      "loss": 0.4522,
      "step": 487850
    },
    {
      "epoch": 199.7,
      "grad_norm": 1.823448896408081,
      "learning_rate": 2.939608067277499e-11,
      "loss": 0.4364,
      "step": 487860
    },
    {
      "epoch": 199.7,
      "grad_norm": 1.9546849727630615,
      "learning_rate": 2.8606961777807164e-11,
      "loss": 0.4542,
      "step": 487870
    },
    {
      "epoch": 199.71,
      "grad_norm": 2.088235378265381,
      "learning_rate": 2.7828579107769255e-11,
      "loss": 0.4566,
      "step": 487880
    },
    {
      "epoch": 199.71,
      "grad_norm": 2.0133352279663086,
      "learning_rate": 2.7060932665783768e-11,
      "loss": 0.4303,
      "step": 487890
    },
    {
      "epoch": 199.71,
      "grad_norm": 2.3388218879699707,
      "learning_rate": 2.6304022455753832e-11,
      "loss": 0.427,
      "step": 487900
    },
    {
      "epoch": 199.72,
      "grad_norm": 2.7157936096191406,
      "learning_rate": 2.5557848481062154e-11,
      "loss": 0.4689,
      "step": 487910
    },
    {
      "epoch": 199.72,
      "grad_norm": 2.0545952320098877,
      "learning_rate": 2.4822410745091448e-11,
      "loss": 0.4579,
      "step": 487920
    },
    {
      "epoch": 199.73,
      "grad_norm": 1.9919077157974243,
      "learning_rate": 2.4097709250964215e-11,
      "loss": 0.4574,
      "step": 487930
    },
    {
      "epoch": 199.73,
      "grad_norm": 1.6864122152328491,
      "learning_rate": 2.3383744002323373e-11,
      "loss": 0.4403,
      "step": 487940
    },
    {
      "epoch": 199.73,
      "grad_norm": 2.0108299255371094,
      "learning_rate": 2.2680515002551635e-11,
      "loss": 0.4496,
      "step": 487950
    },
    {
      "epoch": 199.74,
      "grad_norm": 2.571153402328491,
      "learning_rate": 2.1988022254511295e-11,
      "loss": 0.4282,
      "step": 487960
    },
    {
      "epoch": 199.74,
      "grad_norm": 2.2403581142425537,
      "learning_rate": 2.1306265761324852e-11,
      "loss": 0.4566,
      "step": 487970
    },
    {
      "epoch": 199.75,
      "grad_norm": 1.8254796266555786,
      "learning_rate": 2.063524552663523e-11,
      "loss": 0.4347,
      "step": 487980
    },
    {
      "epoch": 199.75,
      "grad_norm": 1.7896604537963867,
      "learning_rate": 1.9974961553044512e-11,
      "loss": 0.4519,
      "step": 487990
    },
    {
      "epoch": 199.75,
      "grad_norm": 2.1654043197631836,
      "learning_rate": 1.9325413843675196e-11,
      "loss": 0.4173,
      "step": 488000
    },
    {
      "epoch": 199.76,
      "grad_norm": 1.9074783325195312,
      "learning_rate": 1.868660240164979e-11,
      "loss": 0.4256,
      "step": 488010
    },
    {
      "epoch": 199.76,
      "grad_norm": 2.3064119815826416,
      "learning_rate": 1.805852722957038e-11,
      "loss": 0.4512,
      "step": 488020
    },
    {
      "epoch": 199.77,
      "grad_norm": 2.264338254928589,
      "learning_rate": 1.744118833081967e-11,
      "loss": 0.456,
      "step": 488030
    },
    {
      "epoch": 199.77,
      "grad_norm": 2.3765804767608643,
      "learning_rate": 1.6834585707999753e-11,
      "loss": 0.4668,
      "step": 488040
    },
    {
      "epoch": 199.77,
      "grad_norm": 2.2728805541992188,
      "learning_rate": 1.6238719363452502e-11,
      "loss": 0.4518,
      "step": 488050
    },
    {
      "epoch": 199.78,
      "grad_norm": 1.8816723823547363,
      "learning_rate": 1.5653589300560625e-11,
      "loss": 0.4644,
      "step": 488060
    },
    {
      "epoch": 199.78,
      "grad_norm": 2.083212375640869,
      "learning_rate": 1.5079195521926214e-11,
      "loss": 0.4339,
      "step": 488070
    },
    {
      "epoch": 199.79,
      "grad_norm": 2.7033228874206543,
      "learning_rate": 1.451553802963093e-11,
      "loss": 0.4524,
      "step": 488080
    },
    {
      "epoch": 199.79,
      "grad_norm": 2.1253364086151123,
      "learning_rate": 1.396261682679728e-11,
      "loss": 0.4216,
      "step": 488090
    },
    {
      "epoch": 199.8,
      "grad_norm": 1.950002670288086,
      "learning_rate": 1.3420431915767138e-11,
      "loss": 0.4416,
      "step": 488100
    },
    {
      "epoch": 199.8,
      "grad_norm": 1.7528687715530396,
      "learning_rate": 1.2888983298882383e-11,
      "loss": 0.4636,
      "step": 488110
    },
    {
      "epoch": 199.8,
      "grad_norm": 2.1263487339019775,
      "learning_rate": 1.2368270979005307e-11,
      "loss": 0.4439,
      "step": 488120
    },
    {
      "epoch": 199.81,
      "grad_norm": 2.100292921066284,
      "learning_rate": 1.185829495795737e-11,
      "loss": 0.4638,
      "step": 488130
    },
    {
      "epoch": 199.81,
      "grad_norm": 1.9992505311965942,
      "learning_rate": 1.1359055238600867e-11,
      "loss": 0.4355,
      "step": 488140
    },
    {
      "epoch": 199.82,
      "grad_norm": 2.9250149726867676,
      "learning_rate": 1.0870551822757256e-11,
      "loss": 0.4437,
      "step": 488150
    },
    {
      "epoch": 199.82,
      "grad_norm": 1.9811673164367676,
      "learning_rate": 1.0392784713028624e-11,
      "loss": 0.4446,
      "step": 488160
    },
    {
      "epoch": 199.82,
      "grad_norm": 2.099233388900757,
      "learning_rate": 9.925753911496637e-12,
      "loss": 0.4134,
      "step": 488170
    },
    {
      "epoch": 199.83,
      "grad_norm": 1.853216528892517,
      "learning_rate": 9.469459420242965e-12,
      "loss": 0.4302,
      "step": 488180
    },
    {
      "epoch": 199.83,
      "grad_norm": 2.052605152130127,
      "learning_rate": 9.023901241349274e-12,
      "loss": 0.44,
      "step": 488190
    },
    {
      "epoch": 199.84,
      "grad_norm": 2.1237337589263916,
      "learning_rate": 8.589079376897234e-12,
      "loss": 0.4371,
      "step": 488200
    },
    {
      "epoch": 199.84,
      "grad_norm": 2.1465747356414795,
      "learning_rate": 8.164993828968514e-12,
      "loss": 0.4541,
      "step": 488210
    },
    {
      "epoch": 199.84,
      "grad_norm": 2.0443389415740967,
      "learning_rate": 7.751644599384571e-12,
      "loss": 0.4617,
      "step": 488220
    },
    {
      "epoch": 199.85,
      "grad_norm": 2.0592362880706787,
      "learning_rate": 7.349031689966867e-12,
      "loss": 0.4476,
      "step": 488230
    },
    {
      "epoch": 199.85,
      "grad_norm": 1.9780315160751343,
      "learning_rate": 6.957155102797069e-12,
      "loss": 0.4486,
      "step": 488240
    },
    {
      "epoch": 199.86,
      "grad_norm": 2.138235092163086,
      "learning_rate": 6.576014839436428e-12,
      "loss": 0.4401,
      "step": 488250
    },
    {
      "epoch": 199.86,
      "grad_norm": 2.4009785652160645,
      "learning_rate": 6.205610901706404e-12,
      "loss": 0.4272,
      "step": 488260
    },
    {
      "epoch": 199.86,
      "grad_norm": 1.9694232940673828,
      "learning_rate": 5.845943291168248e-12,
      "loss": 0.4471,
      "step": 488270
    },
    {
      "epoch": 199.87,
      "grad_norm": 2.150320053100586,
      "learning_rate": 5.4970120099036285e-12,
      "loss": 0.4411,
      "step": 488280
    },
    {
      "epoch": 199.87,
      "grad_norm": 2.3837945461273193,
      "learning_rate": 5.158817059213588e-12,
      "loss": 0.4515,
      "step": 488290
    },
    {
      "epoch": 199.88,
      "grad_norm": 1.9928960800170898,
      "learning_rate": 4.831358440659377e-12,
      "loss": 0.4205,
      "step": 488300
    },
    {
      "epoch": 199.88,
      "grad_norm": 1.9837597608566284,
      "learning_rate": 4.51463615554204e-12,
      "loss": 0.4587,
      "step": 488310
    },
    {
      "epoch": 199.89,
      "grad_norm": 2.261925220489502,
      "learning_rate": 4.208650205683034e-12,
      "loss": 0.4411,
      "step": 488320
    },
    {
      "epoch": 199.89,
      "grad_norm": 1.9421992301940918,
      "learning_rate": 3.9134005923834035e-12,
      "loss": 0.4522,
      "step": 488330
    },
    {
      "epoch": 199.89,
      "grad_norm": 2.0075523853302,
      "learning_rate": 3.6288873166839817e-12,
      "loss": 0.4284,
      "step": 488340
    },
    {
      "epoch": 199.9,
      "grad_norm": 2.244292974472046,
      "learning_rate": 3.3551103804062284e-12,
      "loss": 0.444,
      "step": 488350
    },
    {
      "epoch": 199.9,
      "grad_norm": 2.1788642406463623,
      "learning_rate": 3.0920697843307694e-12,
      "loss": 0.4345,
      "step": 488360
    },
    {
      "epoch": 199.91,
      "grad_norm": 1.9149953126907349,
      "learning_rate": 2.839765530018856e-12,
      "loss": 0.4232,
      "step": 488370
    },
    {
      "epoch": 199.91,
      "grad_norm": 2.469050168991089,
      "learning_rate": 2.5981976185113217e-12,
      "loss": 0.4407,
      "step": 488380
    },
    {
      "epoch": 199.91,
      "grad_norm": 1.7127951383590698,
      "learning_rate": 2.3673660505887926e-12,
      "loss": 0.4518,
      "step": 488390
    },
    {
      "epoch": 199.92,
      "grad_norm": 2.272862195968628,
      "learning_rate": 2.1472708278125196e-12,
      "loss": 0.4745,
      "step": 488400
    },
    {
      "epoch": 199.92,
      "grad_norm": 1.8739975690841675,
      "learning_rate": 1.937911950963128e-12,
      "loss": 0.4271,
      "step": 488410
    },
    {
      "epoch": 199.93,
      "grad_norm": 2.327120065689087,
      "learning_rate": 1.7392894210814524e-12,
      "loss": 0.431,
      "step": 488420
    },
    {
      "epoch": 199.93,
      "grad_norm": 2.255388021469116,
      "learning_rate": 1.5514032389481179e-12,
      "loss": 0.4402,
      "step": 488430
    },
    {
      "epoch": 199.93,
      "grad_norm": 1.8585106134414673,
      "learning_rate": 1.3742534053437504e-12,
      "loss": 0.4606,
      "step": 488440
    },
    {
      "epoch": 199.94,
      "grad_norm": 2.3823070526123047,
      "learning_rate": 1.207839921309184e-12,
      "loss": 0.4483,
      "step": 488450
    },
    {
      "epoch": 199.94,
      "grad_norm": 2.1989269256591797,
      "learning_rate": 1.0521627876250438e-12,
      "loss": 0.4349,
      "step": 488460
    },
    {
      "epoch": 199.95,
      "grad_norm": 2.5578091144561768,
      "learning_rate": 9.072220048117474e-13,
      "loss": 0.4435,
      "step": 488470
    },
    {
      "epoch": 199.95,
      "grad_norm": 3.0409860610961914,
      "learning_rate": 7.730175736499201e-13,
      "loss": 0.4454,
      "step": 488480
    },
    {
      "epoch": 199.95,
      "grad_norm": 1.9321659803390503,
      "learning_rate": 6.495494946599789e-13,
      "loss": 0.4352,
      "step": 488490
    },
    {
      "epoch": 199.96,
      "grad_norm": 2.364468574523926,
      "learning_rate": 5.368177683623409e-13,
      "loss": 0.4684,
      "step": 488500
    },
    {
      "epoch": 199.96,
      "grad_norm": 2.2490856647491455,
      "learning_rate": 4.3482239553763174e-13,
      "loss": 0.4397,
      "step": 488510
    },
    {
      "epoch": 199.97,
      "grad_norm": 2.1553852558135986,
      "learning_rate": 3.435633764460599e-13,
      "loss": 0.44,
      "step": 488520
    },
    {
      "epoch": 199.97,
      "grad_norm": 2.1025569438934326,
      "learning_rate": 2.630407116080424e-13,
      "loss": 0.4456,
      "step": 488530
    },
    {
      "epoch": 199.98,
      "grad_norm": 2.2531886100769043,
      "learning_rate": 1.9325440128378777e-13,
      "loss": 0.44,
      "step": 488540
    },
    {
      "epoch": 199.98,
      "grad_norm": 2.048909902572632,
      "learning_rate": 1.3420444599371306e-13,
      "loss": 0.4354,
      "step": 488550
    },
    {
      "epoch": 199.98,
      "grad_norm": 2.352147340774536,
      "learning_rate": 8.589084573781824e-14,
      "loss": 0.4326,
      "step": 488560
    },
    {
      "epoch": 199.99,
      "grad_norm": 1.9424015283584595,
      "learning_rate": 4.831360077631186e-14,
      "loss": 0.4442,
      "step": 488570
    },
    {
      "epoch": 199.99,
      "grad_norm": 2.139491558074951,
      "learning_rate": 2.147271162961095e-14,
      "loss": 0.4372,
      "step": 488580
    },
    {
      "epoch": 200.0,
      "grad_norm": 1.9118189811706543,
      "learning_rate": 5.368177777298477e-15,
      "loss": 0.4376,
      "step": 488590
    },
    {
      "epoch": 200.0,
      "grad_norm": 2.2365453243255615,
      "learning_rate": 0.0,
      "loss": 0.4373,
      "step": 488600
    },
    {
      "epoch": 200.0,
      "eval_loss": 0.4430118799209595,
      "eval_runtime": 43.5753,
      "eval_samples_per_second": 79.15,
      "eval_steps_per_second": 9.914,
      "step": 488600
    },
    {
      "epoch": 200.0,
      "step": 488600,
      "total_flos": 3.950483777330872e+20,
      "train_loss": 0.513909718911276,
      "train_runtime": 133465.2589,
      "train_samples_per_second": 29.286,
      "train_steps_per_second": 3.661
    }
  ],
  "logging_steps": 10,
  "max_steps": 488600,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 200,
  "save_steps": 500,
  "total_flos": 3.950483777330872e+20,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}